Tryag File Manager
Home
-
Turbo Force
Current Path :
/
proc
/
self
/
root
/
usr
/
lib
/
vmware-tools
/
modules
/
source
/
Upload File :
New :
File
Dir
//proc/self/root/usr/lib/vmware-tools/modules/source/vsock.tar
vsock-only/ 0000755 0000000 0000000 00000000000 13432726375 011674 5 ustar root root vsock-only/Makefile 0000644 0000000 0000000 00000007404 13432725347 013337 0 ustar root root #!/usr/bin/make -f ########################################################## # Copyright (C) 1998-2016 VMware, Inc. All rights reserved. # # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation version 2 and no later version. # # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY # or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License # for more details. # # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA # ########################################################## #### #### VMware kernel module Makefile to be distributed externally #### #### #### SRCROOT _must_ be a relative path. #### SRCROOT = . # # open-vm-tools doesn't replicate shared source files for different modules; # instead, files are kept in shared locations. So define a few useful macros # to be able to handle both cases cleanly. # INCLUDE := ifdef OVT_SOURCE_DIR AUTOCONF_DIR := $(OVT_SOURCE_DIR)/modules/linux/shared/autoconf VMLIB_PATH = $(OVT_SOURCE_DIR)/lib/$(1) INCLUDE += -I$(OVT_SOURCE_DIR)/modules/linux/shared INCLUDE += -I$(OVT_SOURCE_DIR)/lib/include else AUTOCONF_DIR := $(SRCROOT)/shared/autoconf INCLUDE += -I$(SRCROOT)/shared endif VM_UNAME = $(shell uname -r) # Header directory for the running kernel ifdef LINUXINCLUDE HEADER_DIR = $(LINUXINCLUDE) else HEADER_DIR = /lib/modules/$(VM_UNAME)/build/include endif BUILD_DIR = $(HEADER_DIR)/.. DRIVER := vsock PRODUCT := drivers-for-linux-guest # Grep program GREP = /bin/grep vm_check_gcc = $(shell if $(CC) $(1) -S -o /dev/null -xc /dev/null \ > /dev/null 2>&1; then echo "$(1)"; else echo "$(2)"; fi) vm_check_file = $(shell if test -f $(1); then echo "yes"; else echo "no"; fi) ifndef VM_KBUILD VM_KBUILD := no ifeq ($(call vm_check_file,$(BUILD_DIR)/Makefile), yes) VM_KBUILD := yes endif export VM_KBUILD endif ifndef VM_KBUILD_SHOWN ifeq ($(VM_KBUILD), no) VM_DUMMY := $(shell echo >&2 "Using standalone build system.") else VM_DUMMY := $(shell echo >&2 "Using kernel build system.") endif VM_KBUILD_SHOWN := yes export VM_KBUILD_SHOWN endif ifneq ($(VM_KBUILD), no) # If there is no version defined, we are in toplevel pass, not yet in kernel makefiles... ifeq ($(VERSION),) DRIVER_KO := $(DRIVER).ko .PHONY: $(DRIVER_KO) auto-build: $(DRIVER_KO) cp -f $< $(SRCROOT)/../$(DRIVER).o # $(DRIVER_KO) is a phony target, so compare file times explicitly $(DRIVER): $(DRIVER_KO) if [ $< -nt $@ ] || [ ! -e $@ ] ; then cp -f $< $@; fi # # Define a setup target that gets built before the actual driver. # This target may not be used at all, but if it is then it will be defined # in Makefile.kernel # prebuild:: ; postbuild:: ; $(DRIVER_KO): prebuild $(MAKE) -C $(BUILD_DIR) SUBDIRS=$$PWD SRCROOT=$$PWD/$(SRCROOT) \ MODULEBUILDDIR=$(MODULEBUILDDIR) modules $(MAKE) -C $$PWD SRCROOT=$$PWD/$(SRCROOT) \ MODULEBUILDDIR=$(MODULEBUILDDIR) postbuild endif vm_check_build = $(shell if $(CC) $(KBUILD_CPPFLAGS) $(KBUILD_CFLAGS) \ $(CPPFLAGS) $(CFLAGS) $(CFLAGS_KERNEL) $(LINUXINCLUDE) \ $(EXTRA_CFLAGS) -Iinclude2/asm/mach-default \ -DKBUILD_BASENAME=\"$(DRIVER)\" \ -Werror -S -o /dev/null -xc $(1) \ > /dev/null 2>&1; then echo "$(2)"; else echo "$(3)"; fi) CC_WARNINGS := -Wall -Wstrict-prototypes CC_OPTS := $(GLOBAL_DEFS) $(CC_WARNINGS) -DVMW_USING_KBUILD ifdef VMX86_DEVEL CC_OPTS += -DVMX86_DEVEL endif ifdef VMX86_DEBUG CC_OPTS += -DVMX86_DEBUG endif include $(SRCROOT)/Makefile.kernel else include $(SRCROOT)/Makefile.normal endif #.SILENT: vsock-only/shared/ 0000755 0000000 0000000 00000000000 13432726375 013142 5 ustar root root vsock-only/shared/compat_statfs.h 0000444 0000000 0000000 00000002306 13432725347 016157 0 ustar root root /********************************************************* * Copyright (C) 2006 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_STATFS_H__ # define __COMPAT_STATFS_H__ /* vfs.h simply include statfs.h, but it knows what directory statfs.h is in. */ #include <linux/vfs.h> /* 2.5.74 renamed struct statfs to kstatfs. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 5, 74) #define compat_kstatfs kstatfs #else #define compat_kstatfs statfs #endif #endif /* __COMPAT_STATFS_H__ */ vsock-only/shared/vmciKernelAPI.h 0000444 0000000 0000000 00000002451 13432725350 015734 0 ustar root root /********************************************************* * Copyright (C) 2010,2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmciKernelAPI.h -- * * Kernel API (current) exported from the VMCI host and guest drivers. */ #ifndef __VMCI_KERNELAPI_H__ #define __VMCI_KERNELAPI_H__ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" /* With this file you always get the latest version. */ #include "vmciKernelAPI1.h" #include "vmciKernelAPI2.h" #include "vmciKernelAPI3.h" #endif /* !__VMCI_KERNELAPI_H__ */ vsock-only/shared/community_source.h 0000444 0000000 0000000 00000003712 13432725350 016710 0 ustar root root /********************************************************* * Copyright (C) 2009-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * community_source.h -- * * Macros for excluding source code from community. */ #ifndef _COMMUNITY_SOURCE_H_ #define _COMMUNITY_SOURCE_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_VMKDRIVERS #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" /* * Convenience macro for COMMUNITY_SOURCE */ #undef EXCLUDE_COMMUNITY_SOURCE #ifdef COMMUNITY_SOURCE #define EXCLUDE_COMMUNITY_SOURCE(x) #else #define EXCLUDE_COMMUNITY_SOURCE(x) x #endif #undef COMMUNITY_SOURCE_AMD_SECRET #if !defined(COMMUNITY_SOURCE) || defined(AMD_SOURCE) /* * It's ok to include AMD_SECRET source code for non-Community Source, * or for drops directed at AMD. */ #define COMMUNITY_SOURCE_AMD_SECRET #endif #undef COMMUNITY_SOURCE_INTEL_SECRET #if !defined(COMMUNITY_SOURCE) || defined(INTEL_SOURCE) /* * It's ok to include INTEL_SECRET source code for non-Community Source, * or for drops directed at Intel. */ #define COMMUNITY_SOURCE_INTEL_SECRET #endif #endif vsock-only/shared/vmci_iocontrols.h 0000444 0000000 0000000 00000062156 13432725350 016524 0 ustar root root /********************************************************* * Copyright (C) 2007-2014,2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmci_iocontrols.h * * The VMCI driver io controls. */ #ifndef _VMCI_IOCONTROLS_H_ #define _VMCI_IOCONTROLS_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" #include "vm_assert.h" #include "vmci_defs.h" #if defined(_WIN32) && defined(WINNT_DDK) /* We need to expose the API through an IOCTL on Windows. Use latest API. */ #include "vmciKernelAPI.h" #endif // _WIN32 && WINNT_DDK #if defined __cplusplus extern "C" { #endif /* *----------------------------------------------------------------------------- * * VMCIVA64ToPtr -- * * Convert a VA64 to a pointer. * * Results: * Virtual address. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void * VMCIVA64ToPtr(VA64 va64) // IN { #ifdef VM_64BIT ASSERT_ON_COMPILE(sizeof (void *) == 8); #else ASSERT_ON_COMPILE(sizeof (void *) == 4); // Check that nothing of value will be lost. ASSERT(!(va64 >> 32)); #endif return (void *)(uintptr_t)va64; } /* *----------------------------------------------------------------------------- * * VMCIPtrToVA64 -- * * Convert a pointer to a VA64. * * Results: * Virtual address. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE VA64 VMCIPtrToVA64(void const *ptr) // IN { ASSERT_ON_COMPILE(sizeof ptr <= sizeof (VA64)); return (VA64)(uintptr_t)ptr; } /* * Driver version. * * Increment major version when you make an incompatible change. * Compatibility goes both ways (old driver with new executable * as well as new driver with old executable). */ #define VMCI_VERSION_SHIFT_WIDTH 16 /* Never change this. */ #define VMCI_MAKE_VERSION(_major, _minor) ((_major) << \ VMCI_VERSION_SHIFT_WIDTH | \ (uint16) (_minor)) #define VMCI_VERSION_MAJOR(v) ((uint32) (v) >> VMCI_VERSION_SHIFT_WIDTH) #define VMCI_VERSION_MINOR(v) ((uint16) (v)) /* * VMCI_VERSION is always the current version. Subsequently listed * versions are ways of detecting previous versions of the connecting * application (i.e., VMX). * * VMCI_VERSION_NOVMVM: This version removed support for VM to VM * communication. * * VMCI_VERSION_NOTIFY: This version introduced doorbell notification * support. * * VMCI_VERSION_HOSTQP: This version introduced host end point support * for hosted products. * * VMCI_VERSION_PREHOSTQP: This is the version prior to the adoption of * support for host end-points. * * VMCI_VERSION_PREVERS2: This fictional version number is intended to * represent the version of a VMX which doesn't call into the driver * with ioctl VERSION2 and thus doesn't establish its version with the * driver. */ #define VMCI_VERSION VMCI_VERSION_NOVMVM #define VMCI_VERSION_NOVMVM VMCI_MAKE_VERSION(11, 0) #define VMCI_VERSION_NOTIFY VMCI_MAKE_VERSION(10, 0) #define VMCI_VERSION_HOSTQP VMCI_MAKE_VERSION(9, 0) #define VMCI_VERSION_PREHOSTQP VMCI_MAKE_VERSION(8, 0) #define VMCI_VERSION_PREVERS2 VMCI_MAKE_VERSION(1, 0) /* * VMCISockets driver version. The version is platform-dependent and is * embedded in vsock_version.h for each platform. It can be obtained via * VMCISock_Version() (which uses IOCTL_VMCI_SOCKETS_VERSION). The * following is simply for constructing an unsigned integer value from the * comma-separated version in the header. This must match the macros defined * in vmci_sockets.h. An example of using this is: * uint16 parts[4] = { VSOCK_DRIVER_VERSION_COMMAS }; * uint32 version = VMCI_SOCKETS_MAKE_VERSION(parts); */ #define VMCI_SOCKETS_MAKE_VERSION(_p) \ ((((_p)[0] & 0xFF) << 24) | (((_p)[1] & 0xFF) << 16) | ((_p)[2])) #if defined(__linux__) || defined(VMKERNEL) /* * Linux defines _IO* macros, but the core kernel code ignore the encoded * ioctl value. It is up to individual drivers to decode the value (for * example to look at the size of a structure to determine which version * of a specific command should be used) or not (which is what we * currently do, so right now the ioctl value for a given command is the * command itself). * * Hence, we just define the IOCTL_VMCI_foo values directly, with no * intermediate IOCTLCMD_ representation. */ # define IOCTLCMD(_cmd) IOCTL_VMCI_ ## _cmd #elif defined (__APPLE__) #include <sys/ioccom.h> #define IOCTLCMD(_cmd) IOCTL_VMCI_ ## _cmd #define IOCTLCMD_I(_cmd, _type) \ IOCTL_VMCI_MACOS_ ## _cmd = _IOW('V', IOCTL_VMCI_ ## _cmd, _type) #define IOCTLCMD_O(_cmd, _type) \ IOCTL_VMCI_MACOS_ ## _cmd = _IOR('V', IOCTL_VMCI_ ## _cmd, _type) #define IOCTLCMD_IO(_cmd, _type) \ IOCTL_VMCI_MACOS_ ## _cmd = _IOWR('V', IOCTL_VMCI_ ## _cmd, _type) #else // if defined(__linux__) /* * On platforms other than Linux, IOCTLCMD_foo values are just numbers, and * we build the IOCTL_VMCI_foo values around these using platform-specific * format for encoding arguments and sizes. */ # define IOCTLCMD(_cmd) IOCTLCMD_VMCI_ ## _cmd #endif enum IOCTLCmd_VMCI { /* * We need to bracket the range of values used for ioctls, because x86_64 * Linux forces us to explicitly register ioctl handlers by value for * handling 32 bit ioctl syscalls. Hence FIRST and LAST. Pick something * for FIRST that doesn't collide with vmmon (2001+). */ #if defined(__linux__) IOCTLCMD(FIRST) = 1951, #else /* Start at 0. */ IOCTLCMD(FIRST), #endif IOCTLCMD(VERSION) = IOCTLCMD(FIRST), /* BEGIN VMCI */ IOCTLCMD(INIT_CONTEXT), /* * The following two were used for process and datagram process creation. * They are not used anymore and reserved for future use. * They will fail if issued. */ IOCTLCMD(RESERVED1), IOCTLCMD(RESERVED2), /* * The following used to be for shared memory. It is now unused and and is * reserved for future use. It will fail if issued. */ IOCTLCMD(RESERVED3), /* * The follwoing three were also used to be for shared memory. An * old WS6 user-mode client might try to use them with the new * driver, but since we ensure that only contexts created by VMX'en * of the appropriate version (VMCI_VERSION_NOTIFY or * VMCI_VERSION_NEWQP) or higher use these ioctl, everything is * fine. */ IOCTLCMD(QUEUEPAIR_SETVA), IOCTLCMD(NOTIFY_RESOURCE), IOCTLCMD(NOTIFICATIONS_RECEIVE), IOCTLCMD(VERSION2), IOCTLCMD(QUEUEPAIR_ALLOC), IOCTLCMD(QUEUEPAIR_SETPAGEFILE), IOCTLCMD(QUEUEPAIR_DETACH), IOCTLCMD(DATAGRAM_SEND), IOCTLCMD(DATAGRAM_RECEIVE), IOCTLCMD(DATAGRAM_REQUEST_MAP), IOCTLCMD(DATAGRAM_REMOVE_MAP), IOCTLCMD(CTX_ADD_NOTIFICATION), IOCTLCMD(CTX_REMOVE_NOTIFICATION), IOCTLCMD(CTX_GET_CPT_STATE), IOCTLCMD(CTX_SET_CPT_STATE), IOCTLCMD(GET_CONTEXT_ID), /* END VMCI */ /* * BEGIN VMCI SOCKETS * * We mark the end of the vmci commands and the start of the vmci sockets * commands since they are used in separate modules on Linux. * */ IOCTLCMD(LAST), IOCTLCMD(SOCKETS_FIRST) = IOCTLCMD(LAST), /* * This used to be for accept() on Windows and Mac OS, which is now * redundant (since we now use real handles). It is used instead for * getting the version. This value is now public, so it cannot change. */ IOCTLCMD(SOCKETS_VERSION) = IOCTLCMD(SOCKETS_FIRST), IOCTLCMD(SOCKETS_BIND), /* * This used to be for close() on Windows and Mac OS, but is no longer * used for the same reason as accept() above. It is used instead for * sending private symbols to the Mac OS driver. */ IOCTLCMD(SOCKETS_SET_SYMBOLS), IOCTLCMD(SOCKETS_CONNECT), /* * The next two values are public (vmci_sockets.h) and cannot be changed. * That means the number of values above these cannot be changed either * unless the base index (specified below) is updated accordingly. */ IOCTLCMD(SOCKETS_GET_AF_VALUE), IOCTLCMD(SOCKETS_GET_LOCAL_CID), IOCTLCMD(SOCKETS_GET_SOCK_NAME), IOCTLCMD(SOCKETS_GET_SOCK_OPT), IOCTLCMD(SOCKETS_GET_VM_BY_NAME), IOCTLCMD(SOCKETS_IOCTL), IOCTLCMD(SOCKETS_LISTEN), IOCTLCMD(SOCKETS_RECV), IOCTLCMD(SOCKETS_RECV_FROM), IOCTLCMD(SOCKETS_SELECT), IOCTLCMD(SOCKETS_SEND), IOCTLCMD(SOCKETS_SEND_TO), IOCTLCMD(SOCKETS_SET_SOCK_OPT), IOCTLCMD(SOCKETS_SHUTDOWN), IOCTLCMD(SOCKETS_SOCKET), IOCTLCMD(SOCKETS_UUID_2_CID), /* 1991 on Linux. */ /* END VMCI SOCKETS */ /* * We reserve a range of 3 ioctls for VMCI Sockets to grow. We cannot * reserve many ioctls here since we are close to overlapping with vmmon * ioctls. Define a meta-ioctl if running out of this binary space. */ // Must be last. IOCTLCMD(SOCKETS_LAST) = IOCTLCMD(SOCKETS_UUID_2_CID) + 3, /* 1994 on Linux. */ /* * The VSockets ioctls occupy the block above. We define a new range of * VMCI ioctls to maintain binary compatibility between the user land and * the kernel driver. Careful, vmmon ioctls start from 2001, so this means * we can add only 4 new VMCI ioctls. Define a meta-ioctl if running out of * this binary space. */ IOCTLCMD(FIRST2), IOCTLCMD(SET_NOTIFY) = IOCTLCMD(FIRST2), /* 1995 on Linux. */ IOCTLCMD(LAST2), }; #if defined (__APPLE__) /* * The size of this must match the size of VSockIoctlPrivSyms in * modules/vsock/common/vsockIoctl.h. */ #pragma pack(push, 1) struct IOCTLCmd_VMCIMacOS_PrivSyms { char data[344]; }; #pragma pack(pop) enum IOCTLCmd_VMCIMacOS { IOCTLCMD_I(SOCKETS_SET_SYMBOLS, struct IOCTLCmd_VMCIMacOS_PrivSyms), IOCTLCMD_O(SOCKETS_VERSION, unsigned int), IOCTLCMD_O(SOCKETS_GET_AF_VALUE, int), IOCTLCMD_O(SOCKETS_GET_LOCAL_CID, unsigned int), }; #endif // __APPLE__ #if defined _WIN32 /* * Windows VMCI ioctl definitions. */ /* PUBLIC: For VMCISockets user-mode clients that use CreateFile(). */ #define VMCI_INTERFACE_VSOCK_PUBLIC_NAME TEXT("\\\\.\\VMCI") /* PUBLIC: For VMCISockets user-mode clients that use NtCreateFile(). */ #define VMCI_INTERFACE_VSOCK_PUBLIC_NAME_NT L"\\??\\VMCI" /* PUBLIC: For the VMX, which uses CreateFile(). */ #define VMCI_INTERFACE_VMX_PUBLIC_NAME TEXT("\\\\.\\VMCIDev\\VMX") /* PRIVATE NAMES */ #define VMCI_DEVICE_VMCI_LINK_PATH L"\\DosDevices\\VMCIDev" #define VMCI_DEVICE_VSOCK_LINK_PATH L"\\DosDevices\\vmci" #define VMCI_DEVICE_HOST_NAME_PATH L"\\Device\\VMCIHostDev" #define VMCI_DEVICE_GUEST_NAME_PATH L"\\Device\\VMCIGuestDev" /* PRIVATE NAMES */ /* These values cannot be changed since some of the ioctl values are public. */ #define FILE_DEVICE_VMCI 0x8103 #define VMCI_IOCTL_BASE_INDEX 0x801 #define VMCIIOCTL_BUFFERED(name) \ CTL_CODE(FILE_DEVICE_VMCI, \ VMCI_IOCTL_BASE_INDEX + IOCTLCMD_VMCI_ ## name, \ METHOD_BUFFERED, \ FILE_ANY_ACCESS) #define VMCIIOCTL_NEITHER(name) \ CTL_CODE(FILE_DEVICE_VMCI, \ VMCI_IOCTL_BASE_INDEX + IOCTLCMD_VMCI_ ## name, \ METHOD_NEITHER, \ FILE_ANY_ACCESS) enum IOCTLCmd_VMCIWin32 { IOCTLCMD(DEVICE_GET) = IOCTLCMD(LAST2) + 1, IOCTLCMD(SOCKETS_SERVICE_GET), IOCTLCMD(SOCKETS_STOP), }; #define IOCTL_VMCI_VERSION VMCIIOCTL_BUFFERED(VERSION) /* BEGIN VMCI */ #define IOCTL_VMCI_INIT_CONTEXT \ VMCIIOCTL_BUFFERED(INIT_CONTEXT) #define IOCTL_VMCI_HYPERCALL \ VMCIIOCTL_BUFFERED(HYPERCALL) #define IOCTL_VMCI_CREATE_DATAGRAM_HANDLE \ VMCIIOCTL_BUFFERED(CREATE_DATAGRAM_HANDLE) #define IOCTL_VMCI_DESTROY_DATAGRAM_HANDLE \ VMCIIOCTL_BUFFERED(DESTROY_DATAGRAM_HANDLE) #define IOCTL_VMCI_NOTIFY_RESOURCE \ VMCIIOCTL_BUFFERED(NOTIFY_RESOURCE) #define IOCTL_VMCI_NOTIFICATIONS_RECEIVE \ VMCIIOCTL_BUFFERED(NOTIFICATIONS_RECEIVE) #define IOCTL_VMCI_VERSION2 \ VMCIIOCTL_BUFFERED(VERSION2) #define IOCTL_VMCI_QUEUEPAIR_ALLOC \ VMCIIOCTL_BUFFERED(QUEUEPAIR_ALLOC) #define IOCTL_VMCI_QUEUEPAIR_SETVA \ VMCIIOCTL_BUFFERED(QUEUEPAIR_SETVA) #define IOCTL_VMCI_QUEUEPAIR_SETPAGEFILE \ VMCIIOCTL_BUFFERED(QUEUEPAIR_SETPAGEFILE) #define IOCTL_VMCI_QUEUEPAIR_DETACH \ VMCIIOCTL_BUFFERED(QUEUEPAIR_DETACH) #define IOCTL_VMCI_DATAGRAM_SEND \ VMCIIOCTL_BUFFERED(DATAGRAM_SEND) #define IOCTL_VMCI_DATAGRAM_RECEIVE \ VMCIIOCTL_NEITHER(DATAGRAM_RECEIVE) #define IOCTL_VMCI_DATAGRAM_REQUEST_MAP \ VMCIIOCTL_BUFFERED(DATAGRAM_REQUEST_MAP) #define IOCTL_VMCI_DATAGRAM_REMOVE_MAP \ VMCIIOCTL_BUFFERED(DATAGRAM_REMOVE_MAP) #define IOCTL_VMCI_CTX_ADD_NOTIFICATION \ VMCIIOCTL_BUFFERED(CTX_ADD_NOTIFICATION) #define IOCTL_VMCI_CTX_REMOVE_NOTIFICATION \ VMCIIOCTL_BUFFERED(CTX_REMOVE_NOTIFICATION) #define IOCTL_VMCI_CTX_GET_CPT_STATE \ VMCIIOCTL_BUFFERED(CTX_GET_CPT_STATE) #define IOCTL_VMCI_CTX_SET_CPT_STATE \ VMCIIOCTL_BUFFERED(CTX_SET_CPT_STATE) #define IOCTL_VMCI_GET_CONTEXT_ID \ VMCIIOCTL_BUFFERED(GET_CONTEXT_ID) #define IOCTL_VMCI_DEVICE_GET \ VMCIIOCTL_BUFFERED(DEVICE_GET) /* END VMCI */ /* BEGIN VMCI SOCKETS */ #define IOCTL_VMCI_SOCKETS_VERSION \ VMCIIOCTL_BUFFERED(SOCKETS_VERSION) #define IOCTL_VMCI_SOCKETS_BIND \ VMCIIOCTL_BUFFERED(SOCKETS_BIND) #define IOCTL_VMCI_SOCKETS_CONNECT \ VMCIIOCTL_BUFFERED(SOCKETS_CONNECT) #define IOCTL_VMCI_SOCKETS_GET_AF_VALUE \ VMCIIOCTL_BUFFERED(SOCKETS_GET_AF_VALUE) #define IOCTL_VMCI_SOCKETS_GET_LOCAL_CID \ VMCIIOCTL_BUFFERED(SOCKETS_GET_LOCAL_CID) #define IOCTL_VMCI_SOCKETS_GET_SOCK_NAME \ VMCIIOCTL_BUFFERED(SOCKETS_GET_SOCK_NAME) #define IOCTL_VMCI_SOCKETS_GET_SOCK_OPT \ VMCIIOCTL_BUFFERED(SOCKETS_GET_SOCK_OPT) #define IOCTL_VMCI_SOCKETS_GET_VM_BY_NAME \ VMCIIOCTL_BUFFERED(SOCKETS_GET_VM_BY_NAME) #define IOCTL_VMCI_SOCKETS_IOCTL \ VMCIIOCTL_BUFFERED(SOCKETS_IOCTL) #define IOCTL_VMCI_SOCKETS_LISTEN \ VMCIIOCTL_BUFFERED(SOCKETS_LISTEN) #define IOCTL_VMCI_SOCKETS_RECV_FROM \ VMCIIOCTL_BUFFERED(SOCKETS_RECV_FROM) #define IOCTL_VMCI_SOCKETS_SELECT \ VMCIIOCTL_BUFFERED(SOCKETS_SELECT) #define IOCTL_VMCI_SOCKETS_SEND_TO \ VMCIIOCTL_BUFFERED(SOCKETS_SEND_TO) #define IOCTL_VMCI_SOCKETS_SET_SOCK_OPT \ VMCIIOCTL_BUFFERED(SOCKETS_SET_SOCK_OPT) #define IOCTL_VMCI_SOCKETS_SHUTDOWN \ VMCIIOCTL_BUFFERED(SOCKETS_SHUTDOWN) #define IOCTL_VMCI_SOCKETS_SERVICE_GET \ VMCIIOCTL_BUFFERED(SOCKETS_SERVICE_GET) #define IOCTL_VMCI_SOCKETS_STOP \ VMCIIOCTL_NEITHER(SOCKETS_STOP) /* END VMCI SOCKETS */ #endif // _WIN32 /* * VMCI driver initialization. This block can also be used to * pass initial group membership etc. */ typedef struct VMCIInitBlock { VMCIId cid; VMCIPrivilegeFlags flags; #ifdef _WIN32 uint64 event; /* Handle for signalling vmci calls on windows. */ #endif // _WIN32 } VMCIInitBlock; typedef struct VMCISharedMemInfo { VMCIHandle handle; uint32 size; uint32 result; VA64 va; /* Currently only used in the guest. */ char pageFileName[VMCI_PATH_MAX]; } VMCISharedMemInfo; typedef struct VMCIQueuePairAllocInfo_VMToVM { VMCIHandle handle; VMCIId peer; uint32 flags; uint64 produceSize; uint64 consumeSize; #if !defined(VMX86_SERVER) && !defined(VMKERNEL) VA64 producePageFile; /* User VA. */ VA64 consumePageFile; /* User VA. */ uint64 producePageFileSize; /* Size of the file name array. */ uint64 consumePageFileSize; /* Size of the file name array. */ #else PPN64 * PPNs; uint64 numPPNs; #endif int32 result; uint32 _pad; } VMCIQueuePairAllocInfo_VMToVM; typedef struct VMCIQueuePairAllocInfo { VMCIHandle handle; VMCIId peer; uint32 flags; uint64 produceSize; uint64 consumeSize; #if !defined(VMX86_SERVER) && !defined(VMKERNEL) VA64 ppnVA; /* Start VA of queue pair PPNs. */ #else PPN64 * PPNs; #endif uint64 numPPNs; int32 result; uint32 version; } VMCIQueuePairAllocInfo; typedef struct VMCIQueuePairSetVAInfo { VMCIHandle handle; VA64 va; /* Start VA of queue pair PPNs. */ uint64 numPPNs; uint32 version; int32 result; } VMCIQueuePairSetVAInfo; /* * For backwards compatibility, here is a version of the * VMCIQueuePairPageFileInfo before host support end-points was added. * Note that the current version of that structure requires VMX to * pass down the VA of the mapped file. Before host support was added * there was nothing of the sort. So, when the driver sees the ioctl * with a parameter that is the sizeof * VMCIQueuePairPageFileInfo_NoHostQP then it can infer that the version * of VMX running can't attach to host end points because it doesn't * provide the VA of the mapped files. * * The Linux driver doesn't get an indication of the size of the * structure passed down from user space. So, to fix a long standing * but unfiled bug, the _pad field has been renamed to version. * Existing versions of VMX always initialize the PageFileInfo * structure so that _pad, er, version is set to 0. * * A version value of 1 indicates that the size of the structure has * been increased to include two UVA's: produceUVA and consumeUVA. * These UVA's are of the mmap()'d queue contents backing files. * * In addition, if when VMX is sending down the * VMCIQueuePairPageFileInfo structure it gets an error then it will * try again with the _NoHostQP version of the file to see if an older * VMCI kernel module is running. */ typedef struct VMCIQueuePairPageFileInfo_NoHostQP { VMCIHandle handle; VA64 producePageFile; /* User VA. */ VA64 consumePageFile; /* User VA. */ uint64 producePageFileSize; /* Size of the file name array. */ uint64 consumePageFileSize; /* Size of the file name array. */ int32 result; uint32 version; /* Was _pad. Must be 0. */ } VMCIQueuePairPageFileInfo_NoHostQP; typedef struct VMCIQueuePairPageFileInfo { VMCIHandle handle; #if !defined(VMX86_SERVER) && !defined(VMKERNEL) VA64 producePageFile; /* User VA. */ VA64 consumePageFile; /* User VA. */ uint64 producePageFileSize; /* Size of the file name array. */ uint64 consumePageFileSize; /* Size of the file name array. */ #endif int32 result; uint32 version; /* Was _pad. */ VA64 produceVA; /* User VA of the mapped file. */ VA64 consumeVA; /* User VA of the mapped file. */ } VMCIQueuePairPageFileInfo; typedef struct VMCIQueuePairDetachInfo { VMCIHandle handle; int32 result; uint32 _pad; } VMCIQueuePairDetachInfo; typedef struct VMCIDatagramSendRecvInfo { VA64 addr; uint32 len; int32 result; } VMCIDatagramSendRecvInfo; /* Used to add/remove well-known datagram mappings. */ typedef struct VMCIDatagramMapInfo { VMCIId wellKnownID; int result; } VMCIDatagramMapInfo; /* Used to add/remove remote context notifications. */ typedef struct VMCINotifyAddRemoveInfo { VMCIId remoteCID; int result; } VMCINotifyAddRemoveInfo; /* Used to set/get current context's checkpoint state. */ typedef struct VMCICptBufInfo { VA64 cptBuf; uint32 cptType; uint32 bufSize; int32 result; uint32 _pad; } VMCICptBufInfo; /* Used to pass notify flag's address to the host driver. */ typedef struct VMCISetNotifyInfo { VA64 notifyUVA; int32 result; uint32 _pad; } VMCISetNotifyInfo; #define VMCI_NOTIFY_RESOURCE_QUEUE_PAIR 0 #define VMCI_NOTIFY_RESOURCE_DOOR_BELL 1 #define VMCI_NOTIFY_RESOURCE_ACTION_NOTIFY 0 #define VMCI_NOTIFY_RESOURCE_ACTION_CREATE 1 #define VMCI_NOTIFY_RESOURCE_ACTION_DESTROY 2 /* * Used to create and destroy doorbells, and generate a notification * for a doorbell or queue pair. */ typedef struct VMCINotifyResourceInfo { VMCIHandle handle; uint16 resource; uint16 action; int32 result; } VMCINotifyResourceInfo; /* * Used to recieve pending notifications for doorbells and queue * pairs. */ typedef struct VMCINotificationReceiveInfo { VA64 dbHandleBufUVA; uint64 dbHandleBufSize; VA64 qpHandleBufUVA; uint64 qpHandleBufSize; int32 result; uint32 _pad; } VMCINotificationReceiveInfo; #if defined(_WIN32) && defined(WINNT_DDK) /* * Used on Windows to expose the API calls that are no longer exported. This * is kernel-mode only, and both sides will have the same bitness, so we can * use pointers directly. */ /* Version 1. */ typedef struct VMCIDeviceGetInfoVer1 { VMCI_DeviceReleaseFct *deviceRelease; VMCIDatagram_CreateHndFct *dgramCreateHnd; VMCIDatagram_CreateHndPrivFct *dgramCreateHndPriv; VMCIDatagram_DestroyHndFct *dgramDestroyHnd; VMCIDatagram_SendFct *dgramSend; VMCI_GetContextIDFct *getContextId; VMCI_VersionFct *version; VMCIEvent_SubscribeFct *eventSubscribe; VMCIEvent_UnsubscribeFct *eventUnsubscribe; VMCIQPair_AllocFct *qpairAlloc; VMCIQPair_DetachFct *qpairDetach; VMCIQPair_GetProduceIndexesFct *qpairGetProduceIndexes; VMCIQPair_GetConsumeIndexesFct *qpairGetConsumeIndexes; VMCIQPair_ProduceFreeSpaceFct *qpairProduceFreeSpace; VMCIQPair_ProduceBufReadyFct *qpairProduceBufReady; VMCIQPair_ConsumeFreeSpaceFct *qpairConsumeFreeSpace; VMCIQPair_ConsumeBufReadyFct *qpairConsumeBufReady; VMCIQPair_EnqueueFct *qpairEnqueue; VMCIQPair_DequeueFct *qpairDequeue; VMCIQPair_PeekFct *qpairPeek; VMCIQPair_EnqueueVFct *qpairEnqueueV; VMCIQPair_DequeueVFct *qpairDequeueV; VMCIQPair_PeekVFct *qpairPeekV; VMCI_ContextID2HostVmIDFct *contextID2HostVmID; VMCI_IsContextOwnerFct *isContextOwner; VMCIContext_GetPrivFlagsFct *contextGetPrivFlags; } VMCIDeviceGetInfoVer1; /* Version 2. */ typedef struct VMCIDeviceGetInfoVer2 { VMCIDoorbell_CreateFct *doorbellCreate; VMCIDoorbell_DestroyFct *doorbellDestroy; VMCIDoorbell_NotifyFct *doorbellNotify; } VMCIDeviceGetInfoVer2; typedef struct VMCIDeviceGetInfoHdr { /* Requested API version on input, supported version on output. */ uint32 apiVersion; VMCI_DeviceShutdownFn *deviceShutdownCB; void *userData; void *deviceRegistration; } VMCIDeviceGetInfoHdr; /* Combination of all versions. */ typedef struct VMCIDeviceGetInfo { VMCIDeviceGetInfoHdr hdr; VMCIDeviceGetInfoVer1 ver1; VMCIDeviceGetInfoVer2 ver2; } VMCIDeviceGetInfo; #endif // _WIN32 && WINNT_DDK #ifdef __APPLE__ /* * Mac OS ioctl definitions. * * Mac OS defines _IO* macros, and the core kernel code uses the size encoded * in the ioctl value to copy the memory back and forth (depending on the * direction encoded in the ioctl value) between the user and kernel address * spaces. * See iocontrolsMacOS.h for details on how this is done. We use sockets only * for vmci. */ #include <sys/ioccom.h> enum VMCrossTalkSockOpt { VMCI_SO_VERSION = 0, VMCI_SO_CONTEXT = IOCTL_VMCI_INIT_CONTEXT, VMCI_SO_NOTIFY_RESOURCE = IOCTL_VMCI_NOTIFY_RESOURCE, VMCI_SO_NOTIFICATIONS_RECEIVE = IOCTL_VMCI_NOTIFICATIONS_RECEIVE, VMCI_SO_VERSION2 = IOCTL_VMCI_VERSION2, VMCI_SO_QUEUEPAIR_ALLOC = IOCTL_VMCI_QUEUEPAIR_ALLOC, VMCI_SO_QUEUEPAIR_SETVA = IOCTL_VMCI_QUEUEPAIR_SETVA, VMCI_SO_QUEUEPAIR_SETPAGEFILE = IOCTL_VMCI_QUEUEPAIR_SETPAGEFILE, VMCI_SO_QUEUEPAIR_DETACH = IOCTL_VMCI_QUEUEPAIR_DETACH, VMCI_SO_DATAGRAM_SEND = IOCTL_VMCI_DATAGRAM_SEND, VMCI_SO_DATAGRAM_RECEIVE = IOCTL_VMCI_DATAGRAM_RECEIVE, VMCI_SO_DATAGRAM_REQUEST_MAP = IOCTL_VMCI_DATAGRAM_REQUEST_MAP, VMCI_SO_DATAGRAM_REMOVE_MAP = IOCTL_VMCI_DATAGRAM_REMOVE_MAP, VMCI_SO_CTX_ADD_NOTIFICATION = IOCTL_VMCI_CTX_ADD_NOTIFICATION, VMCI_SO_CTX_REMOVE_NOTIFICATION = IOCTL_VMCI_CTX_REMOVE_NOTIFICATION, VMCI_SO_CTX_GET_CPT_STATE = IOCTL_VMCI_CTX_GET_CPT_STATE, VMCI_SO_CTX_SET_CPT_STATE = IOCTL_VMCI_CTX_SET_CPT_STATE, VMCI_SO_GET_CONTEXT_ID = IOCTL_VMCI_GET_CONTEXT_ID, VMCI_SO_USERFD, }; #define VMCI_MACOS_HOST_DEVICE "com.vmware.kext.vmci" #endif /* Clean up helper macros */ #undef IOCTLCMD #if defined __cplusplus } // extern "C" #endif #endif // ifndef _VMCI_IOCONTROLS_H_ vsock-only/shared/vm_assert.h 0000444 0000000 0000000 00000024303 13432725350 015306 0 ustar root root /********************************************************* * Copyright (C) 1998-2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vm_assert.h -- * * The basic assertion facility for all VMware code. * * For proper use, see bora/doc/assert and * http://vmweb.vmware.com/~mts/WebSite/guide/programming/asserts.html. */ #ifndef _VM_ASSERT_H_ #define _VM_ASSERT_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_VMKDRIVERS #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" // XXX not necessary except some places include vm_assert.h improperly #include "vm_basic_types.h" #ifdef __cplusplus extern "C" { #endif /* * Some bits of vmcore are used in VMKernel code and cannot have * the VMKERNEL define due to other header dependencies. */ #if defined(VMKERNEL) && !defined(VMKPANIC) #define VMKPANIC 1 #endif /* * Internal macros, functions, and strings * * The monitor wants to save space at call sites, so it has specialized * functions for each situation. User level wants to save on implementation * so it uses generic functions. */ #if !defined VMM || defined MONITOR_APP // { #if defined (VMKPANIC) #include "vmk_assert.h" #else /* !VMKPANIC */ #define _ASSERT_PANIC(name) \ Panic(_##name##Fmt "\n", __FILE__, __LINE__) #define _ASSERT_PANIC_BUG(bug, name) \ Panic(_##name##Fmt " bugNr=%d\n", __FILE__, __LINE__, bug) #define _ASSERT_PANIC_NORETURN(name) \ Panic(_##name##Fmt "\n", __FILE__, __LINE__) #define _ASSERT_PANIC_BUG_NORETURN(bug, name) \ Panic(_##name##Fmt " bugNr=%d\n", __FILE__, __LINE__, bug) #endif /* VMKPANIC */ #endif // } // These strings don't have newline so that a bug can be tacked on. #define _AssertPanicFmt "PANIC %s:%d" #define _AssertAssertFmt "ASSERT %s:%d" #define _AssertVerifyFmt "VERIFY %s:%d" #define _AssertNotImplementedFmt "NOT_IMPLEMENTED %s:%d" #define _AssertNotReachedFmt "NOT_REACHED %s:%d" #define _AssertMemAllocFmt "MEM_ALLOC %s:%d" #define _AssertNotTestedFmt "NOT_TESTED %s:%d" /* * Panic and log functions */ void Log(const char *fmt, ...) PRINTF_DECL(1, 2); void Warning(const char *fmt, ...) PRINTF_DECL(1, 2); #if defined VMKPANIC void Panic_SaveRegs(void); NORETURN void Panic_NoSave(const char *fmt, ...) PRINTF_DECL(1, 2); #define Panic(fmt...) do { \ Panic_SaveRegs(); \ Panic_NoSave(fmt); \ } while(0) #else NORETURN void Panic(const char *fmt, ...) PRINTF_DECL(1, 2); #endif void LogThrottled(uint32 *count, const char *fmt, ...) PRINTF_DECL(2, 3); void WarningThrottled(uint32 *count, const char *fmt, ...) PRINTF_DECL(2, 3); #ifndef ASSERT_IFNOT /* * PR 271512: When compiling with gcc, catch assignments inside an ASSERT. * * 'UNLIKELY' is defined with __builtin_expect, which does not warn when * passed an assignment (gcc bug 36050). To get around this, we put 'cond' * in an 'if' statement and make sure it never gets executed by putting * that inside of 'if (0)'. We use gcc's statement expression syntax to * make ASSERT an expression because some code uses it that way. * * Since statement expression syntax is a gcc extension and since it's * not clear if this is a problem with other compilers, the ASSERT * definition was not changed for them. Using a bare 'cond' with the * ternary operator may provide a solution. */ #ifdef __GNUC__ #define ASSERT_IFNOT(cond, panic) \ ({if (UNLIKELY(!(cond))) { panic; if (0) { if (cond) {;}}} (void)0;}) #else #define ASSERT_IFNOT(cond, panic) \ (UNLIKELY(!(cond)) ? (panic) : (void)0) #endif #endif /* * Assert, panic, and log macros * * Some of these are redefined below undef !VMX86_DEBUG. * ASSERT() is special cased because of interaction with Windows DDK. */ #if defined VMX86_DEBUG #undef ASSERT #define ASSERT(cond) ASSERT_IFNOT(cond, _ASSERT_PANIC(AssertAssert)) #define ASSERT_BUG(bug, cond) \ ASSERT_IFNOT(cond, _ASSERT_PANIC_BUG(bug, AssertAssert)) #endif #undef VERIFY #define VERIFY(cond) \ ASSERT_IFNOT(cond, _ASSERT_PANIC_NORETURN(AssertVerify)) #define VERIFY_BUG(bug, cond) \ ASSERT_IFNOT(cond, _ASSERT_PANIC_BUG_NORETURN(bug, AssertVerify)) #define PANIC() _ASSERT_PANIC(AssertPanic) #define PANIC_BUG(bug) _ASSERT_PANIC_BUG(bug, AssertPanic) #define ASSERT_NOT_IMPLEMENTED(cond) \ ASSERT_IFNOT(cond, NOT_IMPLEMENTED()) #if defined VMKPANIC || defined VMM #define NOT_IMPLEMENTED() _ASSERT_PANIC_NORETURN(AssertNotImplemented) #else #define NOT_IMPLEMENTED() _ASSERT_PANIC(AssertNotImplemented) #endif #if defined VMM #define NOT_IMPLEMENTED_BUG(bug) \ _ASSERT_PANIC_BUG_NORETURN(bug, AssertNotImplemented) #else #define NOT_IMPLEMENTED_BUG(bug) _ASSERT_PANIC_BUG(bug, AssertNotImplemented) #endif #if defined VMKPANIC || defined VMM #define NOT_REACHED() _ASSERT_PANIC_NORETURN(AssertNotReached) #else #define NOT_REACHED() _ASSERT_PANIC(AssertNotReached) #endif #define ASSERT_MEM_ALLOC(cond) \ ASSERT_IFNOT(cond, _ASSERT_PANIC(AssertMemAlloc)) #ifdef VMX86_DEVEL #define NOT_TESTED() Warning(_AssertNotTestedFmt "\n", __FILE__, __LINE__) #else #define NOT_TESTED() Log(_AssertNotTestedFmt "\n", __FILE__, __LINE__) #endif #define ASSERT_NO_INTERRUPTS() ASSERT(!INTERRUPTS_ENABLED()) #define ASSERT_HAS_INTERRUPTS() ASSERT(INTERRUPTS_ENABLED()) #define ASSERT_NOT_TESTED(cond) (UNLIKELY(!(cond)) ? NOT_TESTED() : (void)0) #define NOT_TESTED_ONCE() DO_ONCE(NOT_TESTED()) #define NOT_TESTED_1024() \ do { \ static uint16 count = 0; \ if (UNLIKELY(count == 0)) { NOT_TESTED(); } \ count = (count + 1) & 1023; \ } while (0) #define LOG_ONCE(_s) DO_ONCE(Log _s) /* * Redefine macros that are only in debug versions */ #if !defined VMX86_DEBUG // { #undef ASSERT #define ASSERT(cond) ((void)0) #define ASSERT_BUG(bug, cond) ((void)0) /* * Expand NOT_REACHED() as appropriate for each situation. * * Mainly, we want the compiler to infer the same control-flow * information as it would from Panic(). Otherwise, different * compilation options will lead to different control-flow-derived * errors, causing some make targets to fail while others succeed. * * VC++ has the __assume() built-in function which we don't trust * (see bug 43485); gcc has no such construct; we just panic in * userlevel code. The monitor doesn't want to pay the size penalty * (measured at 212 bytes for the release vmm for a minimal infinite * loop; panic would cost even more) so it does without and lives * with the inconsistency. */ #if defined VMKPANIC || defined VMM #undef NOT_REACHED #if defined __GNUC__ && (__GNUC__ > 4 || __GNUC__ == 4 && __GNUC_MINOR__ >= 5) #define NOT_REACHED() (__builtin_unreachable()) #else #define NOT_REACHED() ((void)0) #endif #else // keep debug definition #endif #undef LOG_UNEXPECTED #define LOG_UNEXPECTED(bug) ((void)0) #undef ASSERT_NOT_TESTED #define ASSERT_NOT_TESTED(cond) ((void)0) #undef NOT_TESTED #define NOT_TESTED() ((void)0) #undef NOT_TESTED_ONCE #define NOT_TESTED_ONCE() ((void)0) #undef NOT_TESTED_1024 #define NOT_TESTED_1024() ((void)0) #endif // !VMX86_DEBUG } /* * Compile-time assertions. * * ASSERT_ON_COMPILE does not use the common * switch (0) { case 0: case (e): ; } trick because some compilers (e.g. MSVC) * generate code for it. * * The implementation uses both enum and typedef because the typedef alone is * insufficient; gcc allows arrays to be declared with non-constant expressions * (even in typedefs, where it makes no sense). * * NOTE: if GCC ever changes so that it ignores unused types altogether, this * assert might not fire! We explicitly mark it as unused because GCC 4.8+ * uses -Wunused-local-typedefs as part of -Wall, which means the typedef will * generate a warning. */ #if defined(_Static_assert) || defined(__cplusplus) || \ !defined(__GNUC__) || __GNUC__ < 4 || (__GNUC__ == 4 && __GNUC_MINOR__ < 6) #define ASSERT_ON_COMPILE(e) \ do { \ enum { AssertOnCompileMisused = ((e) ? 1 : -1) }; \ UNUSED_TYPE(typedef char AssertOnCompileFailed[AssertOnCompileMisused]); \ } while (0) #else #define ASSERT_ON_COMPILE(e) \ do { \ _Static_assert(e, #e); \ } while (0) #endif /* * To put an ASSERT_ON_COMPILE() outside a function, wrap it * in MY_ASSERTS(). The first parameter must be unique in * each .c file where it appears. For example, * * MY_ASSERTS(FS3_INT, * ASSERT_ON_COMPILE(sizeof(FS3_DiskLock) == 128); * ASSERT_ON_COMPILE(sizeof(FS3_DiskLockReserved) == DISK_BLOCK_SIZE); * ASSERT_ON_COMPILE(sizeof(FS3_DiskBlock) == DISK_BLOCK_SIZE); * ASSERT_ON_COMPILE(sizeof(Hardware_DMIUUID) == 16); * ) * * Caution: ASSERT() within MY_ASSERTS() is silently ignored. * The same goes for anything else not evaluated at compile time. */ #define MY_ASSERTS(name, assertions) \ static INLINE void name(void) { \ assertions \ } #ifdef __cplusplus } /* extern "C" */ #endif #endif /* ifndef _VM_ASSERT_H_ */ vsock-only/shared/compat_sock.h 0000444 0000000 0000000 00000006002 13432725347 015607 0 ustar root root /********************************************************* * Copyright (C) 2003 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_SOCK_H__ # define __COMPAT_SOCK_H__ #include <linux/stddef.h> /* for NULL */ #include <net/sock.h> #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 35) static inline wait_queue_head_t *sk_sleep(struct sock *sk) { return sk->sk_sleep; } #endif /* * Prior to 2.6.24, there was no sock network namespace member. In 2.6.26, it * was hidden behind accessor functions so that its behavior could vary * depending on the value of CONFIG_NET_NS. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 26) # define compat_sock_net(sk) sock_net(sk) #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 24) # define compat_sock_net(sk) sk->sk_net #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 16) #ifndef CONFIG_FILTER # define sk_filter(sk, skb, needlock) 0 #endif /* Taken from 2.6.16's sock.h and modified for macro. */ # define compat_sk_receive_skb(sk, skb, nested) \ ({ \ int rc = NET_RX_SUCCESS; \ \ if (sk_filter(sk, skb, 0)) { \ kfree_skb(skb); \ } else { \ skb->dev = NULL; \ bh_lock_sock(sk); \ if (!sock_owned_by_user(sk)) { \ rc = (sk)->sk_backlog_rcv(sk, skb); \ } else { \ sk_add_backlog(sk, skb); \ } \ bh_unlock_sock(sk); \ } \ \ sock_put(sk); \ rc; \ }) #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 20) # define compat_sk_receive_skb(sk, skb, nested) sk_receive_skb(sk, skb) #else # define compat_sk_receive_skb(sk, skb, nested) sk_receive_skb(sk, skb, nested) #endif #endif /* __COMPAT_SOCK_H__ */ vsock-only/shared/compat_highmem.h 0000444 0000000 0000000 00000002423 13432725347 016271 0 ustar root root /********************************************************* * Copyright (C) 2012 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_HIGHMEM_H__ # define __COMPAT_HIGHMEM_H__ #include <linux/highmem.h> #if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 5, 0) # define compat_kmap_atomic(_page) kmap_atomic(_page) # define compat_kunmap_atomic(_page) kunmap_atomic(_page) #else # define compat_kmap_atomic(_page) kmap_atomic((_page), KM_USER0) # define compat_kunmap_atomic(_page) kunmap_atomic((_page), KM_USER0) #endif #endif /* __COMPAT_HIGHMEM_H__ */ vsock-only/shared/compat_namei.h 0000444 0000000 0000000 00000003416 13432725347 015747 0 ustar root root /********************************************************* * Copyright (C) 2006 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_NAMEI_H__ # define __COMPAT_NAMEI_H__ #include <linux/namei.h> /* * In 2.6.25-rc2, dentry and mount objects were removed from the nameidata * struct. They were both replaced with a struct path. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 25) #define compat_vmw_nd_to_dentry(nd) (nd).path.dentry #else #define compat_vmw_nd_to_dentry(nd) (nd).dentry #endif /* In 2.6.25-rc2, path_release(&nd) was replaced with path_put(&nd.path). */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 25) #define compat_path_release(nd) path_put(&(nd)->path) #else #define compat_path_release(nd) path_release(nd) #endif /* path_lookup was removed in 2.6.39 merge window VFS merge */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 38) #define compat_path_lookup(name, flags, nd) kern_path(name, flags, &((nd)->path)) #else #define compat_path_lookup(name, flags, nd) path_lookup(name, flags, nd) #endif #endif /* __COMPAT_NAMEI_H__ */ vsock-only/shared/compat_kernel.h 0000444 0000000 0000000 00000002735 13432725347 016141 0 ustar root root /********************************************************* * Copyright (C) 2004 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_KERNEL_H__ # define __COMPAT_KERNEL_H__ #include <asm/unistd.h> #include <linux/kernel.h> /* * container_of was introduced in 2.5.28 but it's easier to check like this. */ #ifndef container_of #define container_of(ptr, type, member) ({ \ const typeof( ((type *)0)->member ) *__mptr = (ptr); \ (type *)( (char *)__mptr - offsetof(type,member) );}) #endif /* * vsnprintf became available in 2.4.10. For older kernels, just fall back on * vsprintf. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 10) #define vsnprintf(str, size, fmt, args) vsprintf(str, fmt, args) #endif #endif /* __COMPAT_KERNEL_H__ */ vsock-only/shared/circList.h 0000444 0000000 0000000 00000025262 13432725346 015071 0 ustar root root /********************************************************* * Copyright (C) 1998-2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * circList.h -- * * macros, prototypes and struct definitions for double-linked * circular lists. */ #ifndef _CIRCLIST_H_ #define _CIRCLIST_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" #include "vmware.h" #if defined(__cplusplus) extern "C" { #endif typedef struct ListItem { struct ListItem *prev; struct ListItem *next; } ListItem; /* *---------------------------------------------------------------------- * * CircList_IsEmpty -- * * A NULL list is an empty list. * * Result: * TRUE if list is empty, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------- */ static INLINE Bool CircList_IsEmpty(const ListItem *item) // IN { return item == NULL; } /* *---------------------------------------------------------------------- * * CircList_InitItem -- * * Initialize item as a single-element circular list. * * Result: * None. * * Side effects: * None. * *---------------------------------------------------------------------- */ static INLINE void CircList_InitItem(ListItem *item) // OUT { item->prev = item->next = item; } /* *---------------------------------------------------------------------- * * CircList_First -- * * Return first item in the list. * * Result: * First item. * * Side effects: * None. * *---------------------------------------------------------------------- */ static INLINE ListItem * CircList_First(ListItem *item) // IN { return item; } /* *---------------------------------------------------------------------- * * CircList_Last -- * * Return last item in the list. * * Result: * Last item. * * Side effects: * None. * *---------------------------------------------------------------------- */ static INLINE ListItem * CircList_Last(ListItem *item) { return item->prev; } /* * CIRC_LIST_CONTAINER - get the struct for this entry (like list_entry) * @ptr: the &struct ListItem pointer. * @type: the type of the struct this is embedded in. * @member: the name of the list struct within the struct. */ #define CIRC_LIST_CONTAINER(ptr, type, member) \ VMW_CONTAINER_OF(ptr, type, member) /* * Historical name, left here to reduce churn. * TODO: remove, all LIST_CONTAINER uses should be * VMW_CONTAINER_OF and stop depending on circList.h * to provide the definition. */ #define LIST_CONTAINER(ptr, type, member) VMW_CONTAINER_OF(ptr, type, member) /* * LIST_SCAN_FROM scans the list from "from" up until "until". * The loop variable p should not be destroyed in the process. * "from" is an element in the list where to start scanning. * "until" is the element where search should stop. * member is the field to use for the search - either "next" or "prev". */ #define CIRC_LIST_SCAN_FROM(p, from, until, member) \ for (p = (from); (p) != NULL; \ (p) = (((p)->member == (until)) ? NULL : (p)->member)) /* scan the entire list (non-destructively) */ #define CIRC_LIST_SCAN(p, l) \ CIRC_LIST_SCAN_FROM(p, CircList_First(l), CircList_First(l), next) /* scan the entire list where loop element may be destroyed */ #define CIRC_LIST_SCAN_SAFE(p, pn, l) \ if (!CircList_IsEmpty(l)) \ for (p = (l), (pn) = CircList_Next(p, l); (p) != NULL; \ (p) = (pn), (pn) = CircList_Next(p, l)) /* scan the entire list backwards where loop element may be destroyed */ #define CIRC_LIST_SCAN_BACK_SAFE(p, pn, l) \ if (!CircList_IsEmpty(l)) \ for (p = CircList_Last(l), (pn) = CircList_Prev(p, l); (p) != NULL; \ (p) = (pn), (pn) = CircList_Prev(p, l)) /* *---------------------------------------------------------------------- * * CircList_Next -- * * Returns the next member of a doubly linked list, or NULL if last. * Assumes: p is member of the list headed by head. * * Result: * If head or p is NULL, return NULL. Otherwise, * next list member (or null if last). * * Side effects: * None. * *---------------------------------------------------------------------- */ static INLINE ListItem * CircList_Next(ListItem *p, // IN ListItem *head) // IN { if (head == NULL || p == NULL) { return NULL; } /* both p and head are non-null */ p = p->next; return p == head ? NULL : p; } /* *---------------------------------------------------------------------- * * CircList_Prev -- * * Returns the prev member of a doubly linked list, or NULL if first. * Assumes: p is member of the list headed by head. * * Result: * If head or prev is NULL, return NULL. Otherwise, * prev list member (or null if first). * * Side effects: * None. * *---------------------------------------------------------------------- */ static INLINE ListItem * CircList_Prev(ListItem *p, // IN ListItem *head) // IN { if (head == NULL || p == NULL) { return NULL; } /* both p and head are non-null */ return p == head ? NULL : p->prev; } /* *---------------------------------------------------------------------- * * CircList_DeleteItem -- * * Deletes a member of a doubly linked list, possibly modifies the * list header itself. * Assumes neither p nor headp is null and p is a member of *headp. * * Result: * None * * Side effects: * Modifies *headp. * *---------------------------------------------------------------------- */ static INLINE void CircList_DeleteItem(ListItem *p, // IN ListItem **headp) // IN/OUT { ListItem *next; ASSERT(p != NULL); ASSERT(headp != NULL); next = p->next; if (p == next) { *headp = NULL; } else { next->prev = p->prev; p->prev->next = next; if (*headp == p) { *headp = next; } } } /* *---------------------------------------------------------------------- * * CircList_Queue -- * * Adds a new member to the back of a doubly linked list (queue) * Assumes neither p nor headp is null and p is not a member of *headp. * * Result: * None * * Side effects: * Modifies *headp. * *---------------------------------------------------------------------- */ static INLINE void CircList_Queue(ListItem *p, // IN ListItem **headp) // IN/OUT { ListItem *head; head = *headp; if (CircList_IsEmpty(head)) { CircList_InitItem(p); *headp = p; } else { p->prev = head->prev; p->next = head; p->prev->next = p; head->prev = p; } } /* *---------------------------------------------------------------------- * * CircList_Push -- * * Adds a new member to the front of a doubly linked list (stack) * Assumes neither p nor headp is null and p is not a member of *headp. * * Result: * None * * Side effects: * Modifies *headp. * *---------------------------------------------------------------------- */ static INLINE void CircList_Push(ListItem *p, // IN ListItem **headp) // IN/OUT { CircList_Queue(p, headp); *headp = p; } /* *---------------------------------------------------------------------- * * CircList_Splice -- * * Make a single list {l1 l2} from {l1} and {l2} and return it. * It is okay for one or both lists to be NULL. * No checking is done. It is assumed that l1 and l2 are two * distinct lists. * * Result: * A list { l1 l2 }. * * Side effects: * Modifies l1 and l2 list pointers. * *---------------------------------------------------------------------- */ static INLINE ListItem * CircList_Splice(ListItem *l1, // IN ListItem *l2) // IN { ListItem *l1Last, *l2Last; if (CircList_IsEmpty(l1)) { return l2; } if (CircList_IsEmpty(l2)) { return l1; } l1Last = l1->prev; /* last elem of l1 */ l2Last = l2->prev; /* last elem of l2 */ /* * l1 -> ... -> l1Last l2 -> ... l2Last */ l1Last->next = l2; l2->prev = l1Last; l1->prev = l2Last; l2Last->next = l1; return l1; } #if 0 /* Presently unused, enable if a use is found */ /* *---------------------------------------------------------------------- * * CircList_Split -- * * Make a list l = {l1 l2} into two separate lists {l1} and {l2}, where: * l = { ... x -> p -> ... } split into: * l1 = { ... -> x } * l2 = { p -> ... } * Assumes neither p nor l is null and p is a member of l. * If p is the first element of l, then l1 will be NULL. * * Result: * None. * * Side effects: * Sets *l1p and *l2p to the resulting two lists. * Modifies l's pointers. * *---------------------------------------------------------------------- */ static INLINE void CircList_Split(ListItem *p, // IN ListItem *l, // IN ListItem **l1p, // OUT ListItem **l2p) // OUT { ListItem *last; if (p == CircList_First(l)) { /* first element */ *l1p = NULL; *l2p = l; return; } last = l->prev; *l1p = l; p->prev->next = l; l->prev = p->prev; *l2p = p; p->prev = last; last->next = p; } #endif /* *---------------------------------------------------------------------- * * CircList_Size -- * * Return the number of items in the list. * * Result: * The number of items in the list. * * Side effects: * None. * *---------------------------------------------------------------------- */ static INLINE int CircList_Size(ListItem *head) // IN { ListItem *li; int ret = 0; CIRC_LIST_SCAN(li, head) { ret++; } return ret; } #if defined(__cplusplus) } // extern "C" #endif #endif /* _CIRCLIST_H_ */ vsock-only/shared/vmware.h 0000444 0000000 0000000 00000003507 13432725350 014607 0 ustar root root /********************************************************* * Copyright (C) 2003-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmware.h -- * * Standard include file for VMware source code. */ #ifndef _VMWARE_H_ #define _VMWARE_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_DISTRIBUTE #include "includeCheck.h" #include "vm_basic_types.h" #include "vm_basic_defs.h" #include "vm_assert.h" /* * Global error codes. Currently used internally, but may be exported * to customers one day, like VM_E_XXX in vmcontrol_constants.h */ typedef enum VMwareStatus { VMWARE_STATUS_SUCCESS, /* success */ VMWARE_STATUS_ERROR, /* generic error */ VMWARE_STATUS_NOMEM, /* generic memory allocation error */ VMWARE_STATUS_INSUFFICIENT_RESOURCES, /* internal or system resource limit exceeded */ VMWARE_STATUS_INVALID_ARGS /* invalid arguments */ } VMwareStatus; #define VMWARE_SUCCESS(s) ((s) == VMWARE_STATUS_SUCCESS) #endif // ifndef _VMWARE_H_ vsock-only/shared/compat_netdevice.h 0000444 0000000 0000000 00000024324 13432725347 016625 0 ustar root root /********************************************************* * Copyright (C) 2002 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_NETDEVICE_H__ # define __COMPAT_NETDEVICE_H__ #include <linux/skbuff.h> #include <linux/rtnetlink.h> #include <linux/netdevice.h> #include <linux/etherdevice.h> #include <linux/pci.h> /* * The enet_statistics structure moved from linux/if_ether.h to * linux/netdevice.h and is renamed net_device_stats in 2.1.25 --hpreg */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 1, 25) # include <linux/if_ether.h> # define net_device_stats enet_statistics #endif /* The netif_rx_ni() API appeared in 2.4.8 --hpreg */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 8) # define netif_rx_ni netif_rx #endif /* The device struct was renamed net_device in 2.3.14 --hpreg */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 3, 14) # define net_device device #endif /* * SET_MODULE_OWNER appeared sometime during 2.3.x. It was setting * dev->owner = THIS_MODULE until 2.5.70, where netdevice refcounting * was completely changed. SET_MODULE_OWNER was nop for whole * 2.6.x series, and finally disappeared in 2.6.24. * * MOD_xxx_USE_COUNT wrappers are here, as they must be mutually * exclusive with SET_MODULE_OWNER call. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 0) # define COMPAT_SET_MODULE_OWNER(dev) do {} while (0) # define COMPAT_NETDEV_MOD_INC_USE_COUNT MOD_INC_USE_COUNT # define COMPAT_NETDEV_MOD_DEC_USE_COUNT MOD_DEC_USE_COUNT #else # if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 0) # define COMPAT_SET_MODULE_OWNER(dev) SET_MODULE_OWNER(dev) # else # define COMPAT_SET_MODULE_OWNER(dev) do {} while (0) # endif # define COMPAT_NETDEV_MOD_INC_USE_COUNT do {} while (0) # define COMPAT_NETDEV_MOD_DEC_USE_COUNT do {} while (0) #endif /* * SET_NETDEV_DEV appeared sometime during 2.5.x, and later was * crossported to various 2.4.x kernels (as dummy macro). */ #ifdef SET_NETDEV_DEV # define COMPAT_SET_NETDEV_DEV(dev, pdev) SET_NETDEV_DEV(dev, pdev) #else # define COMPAT_SET_NETDEV_DEV(dev, pdev) do {} while (0) #endif /* * Build alloc_etherdev API on the top of init_etherdev. For 2.0.x kernels * we must provide dummy init method, otherwise register_netdev does * nothing. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 3) #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 2, 0) int vmware_dummy_init(struct net_device *dev) { return 0; } #endif static inline struct net_device* compat_alloc_etherdev(int priv_size) { struct net_device* dev; int size = sizeof *dev + priv_size; /* * The name is dynamically allocated before 2.4.0, but * is an embedded array in later kernels. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 0) size += sizeof("ethXXXXXXX"); #endif dev = kmalloc(size, GFP_KERNEL); if (dev) { memset(dev, 0, size); if (priv_size) { dev->priv = dev + 1; } #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 0) dev->name = (char *)(dev + 1) + priv_size; #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 2, 0) dev->init = vmware_dummy_init; #endif if (init_etherdev(dev, 0) != dev) { kfree(dev); dev = NULL; } } return dev; } #else #define compat_alloc_etherdev(sz) alloc_etherdev(sz) #endif /* * alloc_netdev and free_netdev are there since 2.4.23. Their use is mandatory * since 2.6.24. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 23) static inline struct net_device * compat_alloc_netdev(int priv_size, const char *mask, void (*setup)(struct net_device *)) { struct net_device *dev; int netdev_size = sizeof *dev; int alloc_size; # if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 0) netdev_size += IFNAMSIZ; # endif alloc_size = netdev_size + priv_size; dev = kmalloc(alloc_size, GFP_KERNEL); if (dev) { memset(dev, 0, alloc_size); dev->priv = (char*)dev + netdev_size; setup(dev); # if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 0) dev->name = (char*)(dev + 1); # endif strcpy(dev->name, mask); } return dev; } # define compat_free_netdev(dev) kfree(dev) #else # define compat_alloc_netdev(size, mask, setup) alloc_netdev(size, mask, setup) # define compat_free_netdev(dev) free_netdev(dev) #endif /* netdev_priv() appeared in 2.6.3 */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 3) # define compat_netdev_priv(netdev) (netdev)->priv #else # define compat_netdev_priv(netdev) netdev_priv(netdev) #endif /* * In 3.1 merge window feature maros were removed from mainline, * so let's add back ones we care about. */ #if !defined(HAVE_NET_DEVICE_OPS) && \ LINUX_VERSION_CODE >= KERNEL_VERSION(3, 0, 0) # define HAVE_NET_DEVICE_OPS 1 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 9) # define COMPAT_NETDEV_TX_OK NETDEV_TX_OK # define COMPAT_NETDEV_TX_BUSY NETDEV_TX_BUSY #else # define COMPAT_NETDEV_TX_OK 0 # define COMPAT_NETDEV_TX_BUSY 1 #endif /* unregister_netdevice_notifier was not safe prior to 2.6.17 */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 17) && \ !defined(ATOMIC_NOTIFIER_INIT) /* pre 2.6.17 and not patched */ static inline int compat_unregister_netdevice_notifier(struct notifier_block *nb) { int err; rtnl_lock(); err = unregister_netdevice_notifier(nb); rtnl_unlock(); return err; } #else /* post 2.6.17 or patched */ #define compat_unregister_netdevice_notifier(_nb) \ unregister_netdevice_notifier(_nb); #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 24) || defined(__VMKLNX__) # define compat_netif_napi_add(dev, napi, poll, quota) \ netif_napi_add(dev, napi, poll, quota) # if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 30) || \ defined VMW_NETIF_SINGLE_NAPI_PARM # define compat_napi_complete(dev, napi) napi_complete(napi) # define compat_napi_schedule(dev, napi) napi_schedule(napi) # else # define compat_napi_complete(dev, napi) netif_rx_complete(dev, napi) # define compat_napi_schedule(dev, napi) netif_rx_schedule(dev, napi) # endif # define compat_napi_enable(dev, napi) napi_enable(napi) # define compat_napi_disable(dev, napi) napi_disable(napi) #else # define compat_napi_complete(dev, napi) netif_rx_complete(dev) # define compat_napi_schedule(dev, napi) netif_rx_schedule(dev) # define compat_napi_enable(dev, napi) netif_poll_enable(dev) # define compat_napi_disable(dev, napi) netif_poll_disable(dev) /* RedHat ported GRO to 2.6.18 bringing new napi_struct with it */ # if defined NETIF_F_GRO # define compat_netif_napi_add(netdev, napi, pollcb, quota) \ do { \ (netdev)->poll = (pollcb); \ (netdev)->weight = (quota);\ (napi)->dev = (netdev); \ } while (0) # else struct napi_struct { int dummy; }; # define compat_netif_napi_add(dev, napi, pollcb, quota) \ do { \ (dev)->poll = (pollcb); \ (dev)->weight = (quota);\ } while (0) # endif #endif #ifdef NETIF_F_TSO6 # define COMPAT_NETIF_F_TSO (NETIF_F_TSO6 | NETIF_F_TSO) #else # define COMPAT_NETIF_F_TSO (NETIF_F_TSO) #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 18) # define compat_netif_tx_lock(dev) netif_tx_lock(dev) # define compat_netif_tx_unlock(dev) netif_tx_unlock(dev) #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 16) # define compat_netif_tx_lock(dev) spin_lock(&dev->xmit_lock) # define compat_netif_tx_unlock(dev) spin_unlock(&dev->xmit_lock) #else /* Vendor backporting (SLES 10) has muddled the tx_lock situation. Pick whichever * of the above works for you. */ # define compat_netif_tx_lock(dev) do {} while (0) # define compat_netif_tx_unlock(dev) do {} while (0) #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 37) # define COMPAT_VLAN_GROUP_ARRAY_LEN VLAN_N_VID # define compat_flush_scheduled_work(work) cancel_work_sync(work) #else # define COMPAT_VLAN_GROUP_ARRAY_LEN VLAN_GROUP_ARRAY_LEN # define compat_flush_scheduled_work(work) flush_scheduled_work() #endif /* * For kernel versions older than 2.6.29, where pci_msi_enabled is not * available, check if * 1. CONFIG_PCI_MSI is present * 2. kernel version is newer than 2.6.25 (because multiqueue is not * supporter) in kernels older than that) * 3. msi can be enabled. If it fails it means that MSI is not available. * When all the above are true, return non-zero so that multiple queues will be * allowed in the driver. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 29) # define compat_multiqueue_allowed(dev) pci_msi_enabled() #else # if defined CONFIG_PCI_MSI && LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 25) static inline int compat_multiqueue_allowed(struct pci_dev *dev) { int ret; if (!pci_enable_msi(dev)) ret = 1; else ret = 0; pci_disable_msi(dev); return ret; } # else # define compat_multiqueue_allowed(dev) (0) # endif #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 37) # define compat_vlan_get_protocol(skb) vlan_get_protocol(skb) #else # define compat_vlan_get_protocol(skb) (skb->protocol) #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 3, 0) typedef netdev_features_t compat_netdev_features_t; #else typedef u32 compat_netdev_features_t; #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 7, 0) || defined(VMW_NETIF_TRANS_UPDATE) #define compat_netif_trans_update(d) netif_trans_update(d) #else #define compat_netif_trans_update(d) do { (d)->trans_start = jiffies; } while (0) #endif #endif /* __COMPAT_NETDEVICE_H__ */ vsock-only/shared/compat_dcache.h 0000444 0000000 0000000 00000004003 13432725347 016056 0 ustar root root /********************************************************* * Copyright (C) 2013 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_DCACHE_H__ # define __COMPAT_DCACHE_H__ #include <linux/dcache.h> /* * per-dentry locking was born in 2.5.62. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 5, 62) #define compat_lock_dentry(dentry) spin_lock(&dentry->d_lock) #define compat_unlock_dentry(dentry) spin_unlock(&dentry->d_lock) #else #define compat_lock_dentry(dentry) do {} while (0) #define compat_unlock_dentry(dentry) do {} while (0) #endif /* * d_alloc_name was born in 2.6.10. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 10) #define compat_d_alloc_name(parent, s) d_alloc_name(parent, s) #else #define compat_d_alloc_name(parent, s) \ ({ \ struct qstr q; \ q.name = s; \ q.len = strlen(s); \ q.hash = full_name_hash(q.name, q.len); \ d_alloc(parent, &q); \ }) #endif #endif /* __COMPAT_DCACHE_H__ */ vsock-only/shared/compat_pci_mapping.h 0000444 0000000 0000000 00000004741 13432725347 017146 0 ustar root root /********************************************************* * Copyright (C) 2008 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_PCI_MAPPING_H__ #define __COMPAT_PCI_MAPPING_H__ #include <asm/types.h> #include <asm/io.h> #include <linux/pci.h> #if LINUX_VERSION_CODE < KERNEL_VERSION(2,3,41) typedef u32 dma_addr_t; static __inline__ int get_order(unsigned long size) { int order; size = (size - 1) >> (PAGE_SHIFT - 1); order = -1; do { size >>= 1; order++; } while (size); return order; } static inline void * compat_pci_alloc_consistent(struct pci_dev *hwdev, size_t size, dma_addr_t *dma_handle) { void *ptr = (void *)__get_free_pages(GFP_ATOMIC, get_order(size)); if (ptr) { memset(ptr, 0, size); *dma_handle = virt_to_phys(ptr); } return ptr; } static inline void compat_pci_free_consistent(struct pci_dev *hwdev, size_t size, void *vaddr, dma_addr_t dma_handle) { free_pages((unsigned long)vaddr, get_order(size)); } static inline dma_addr_t compat_pci_map_single(struct pci_dev *hwdev, void *ptr, size_t size, int direction) { return virt_to_phys(ptr); } static inline void compat_pci_unmap_single(struct pci_dev *hwdev, dma_addr_t dma_addr, size_t size, int direction) { } #else #define compat_pci_alloc_consistent(hwdev, size, dma_handle) \ pci_alloc_consistent(hwdev, size, dma_handle) #define compat_pci_free_consistent(hwdev, size, vaddr, dma_handle) \ pci_free_consistent(hwdev, size, vaddr, dma_handle) #define compat_pci_map_single(hwdev, ptr, size, direction) \ pci_map_single(hwdev, ptr, size, direction) #define compat_pci_unmap_single(hwdev, dma_addr, size, direction) \ pci_unmap_single(hwdev, dma_addr, size, direction) #endif #endif vsock-only/shared/vm_basic_asm_x86.h 0000444 0000000 0000000 00000042724 13432725350 016442 0 ustar root root /********************************************************* * Copyright (C) 1998-2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vm_basic_asm_x86.h * * Basic IA32 asm macros */ #ifndef _VM_BASIC_ASM_X86_H_ #define _VM_BASIC_ASM_X86_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" #if defined __cplusplus extern "C" { #endif #ifdef VM_X86_64 /* * The gcc inline asm uses the "A" constraint which differs in 32 & 64 * bit mode. 32 bit means eax and edx, 64 means rax or rdx. */ #error "x86-64 not supported" #endif /* * XTEST * Return TRUE if processor is in transaction region. * */ #if defined(__GNUC__) && (defined(VMM) || defined(VMKERNEL) || defined(FROBOS)) static INLINE Bool xtest(void) { uint8 al; __asm__ __volatile__(".byte 0x0f, 0x01, 0xd6 # xtest \n" "setnz %%al\n" : "=a"(al) : : "cc"); return al; } #endif /* __GNUC__ */ /* * FXSAVE/FXRSTOR * save/restore SIMD/MMX fpu state * * The pointer passed in must be 16-byte aligned. * * Intel and AMD processors behave differently w.r.t. fxsave/fxrstor. Intel * processors unconditionally save the exception pointer state (instruction * ptr., data ptr., and error instruction opcode). FXSAVE_ES1 and FXRSTOR_ES1 * work correctly for Intel processors. * * AMD processors only save the exception pointer state if ES=1. This leads to a * security hole whereby one process/VM can inspect the state of another process * VM. The AMD recommended workaround involves clobbering the exception pointer * state unconditionally, and this is implemented in FXRSTOR_AMD_ES0. Note that * FXSAVE_ES1 will only save the exception pointer state for AMD processors if * ES=1. * * The workaround (FXRSTOR_AMD_ES0) only costs 1 cycle more than just doing an * fxrstor, on both AMD Opteron and Intel Core CPUs. */ #if defined(__GNUC__) static INLINE void FXSAVE_ES1(void *save) { __asm__ __volatile__ ("fxsave %0\n" : "=m" (*(uint8 *)save) : : "memory"); } static INLINE void FXRSTOR_ES1(const void *load) { __asm__ __volatile__ ("fxrstor %0\n" : : "m" (*(const uint8 *)load) : "memory"); } static INLINE void FXRSTOR_AMD_ES0(const void *load) { uint64 dummy = 0; __asm__ __volatile__ ("fnstsw %%ax \n" // Grab x87 ES bit "bt $7,%%ax \n" // Test ES bit "jnc 1f \n" // Jump if ES=0 "fnclex \n" // ES=1. Clear it so fild doesn't trap "1: \n" "ffree %%st(7) \n" // Clear tag bit - avoid poss. stack overflow "fildl %0 \n" // Dummy Load from "safe address" changes all // x87 exception pointers. "fxrstor %1 \n" : : "m" (dummy), "m" (*(const uint8 *)load) : "ax", "memory"); } #endif /* __GNUC__ */ /* * XSAVE/XRSTOR * save/restore GSSE/SIMD/MMX fpu state * * The pointer passed in must be 64-byte aligned. * See above comment for more information. */ #if defined(__GNUC__) && (defined(VMM) || defined(VMKERNEL) || defined(FROBOS)) static INLINE void XSAVE_ES1(void *save, uint64 mask) { #if __GNUC__ < 4 || __GNUC__ == 4 && __GNUC_MINOR__ == 1 __asm__ __volatile__ ( ".byte 0x0f, 0xae, 0x21 \n" : : "c" ((uint8 *)save), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #else __asm__ __volatile__ ( "xsave %0 \n" : "=m" (*(uint8 *)save) : "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #endif } static INLINE void XSAVEOPT_ES1(void *save, uint64 mask) { __asm__ __volatile__ ( ".byte 0x0f, 0xae, 0x31 \n" : : "c" ((uint8 *)save), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); } static INLINE void XRSTOR_ES1(const void *load, uint64 mask) { #if __GNUC__ < 4 || __GNUC__ == 4 && __GNUC_MINOR__ == 1 __asm__ __volatile__ ( ".byte 0x0f, 0xae, 0x29 \n" : : "c" ((const uint8 *)load), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #else __asm__ __volatile__ ( "xrstor %0 \n" : : "m" (*(const uint8 *)load), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #endif } static INLINE void XRSTOR_AMD_ES0(const void *load, uint64 mask) { uint64 dummy = 0; __asm__ __volatile__ ("fnstsw %%ax \n" // Grab x87 ES bit "bt $7,%%ax \n" // Test ES bit "jnc 1f \n" // Jump if ES=0 "fnclex \n" // ES=1. Clear it so fild doesn't trap "1: \n" "ffree %%st(7) \n" // Clear tag bit - avoid poss. stack overflow "fildl %0 \n" // Dummy Load from "safe address" changes all // x87 exception pointers. "mov %%ebx, %%eax \n" #if __GNUC__ < 4 || __GNUC__ == 4 && __GNUC_MINOR__ == 1 ".byte 0x0f, 0xae, 0x29 \n" : : "m" (dummy), "c" ((const uint8 *)load), "b" ((uint32)mask), "d" ((uint32)(mask >> 32)) #else "xrstor %1 \n" : : "m" (dummy), "m" (*(const uint8 *)load), "b" ((uint32)mask), "d" ((uint32)(mask >> 32)) #endif : "eax", "memory"); } #endif /* __GNUC__ */ /* *----------------------------------------------------------------------------- * * Div643232 -- * * Unsigned integer division: * The dividend is 64-bit wide * The divisor is 32-bit wide * The quotient is 32-bit wide * * Use this function if you are certain that: * o Either the quotient will fit in 32 bits, * o Or your code is ready to handle a #DE exception indicating overflow. * If that is not the case, then use Div643264(). * * Results: * Quotient and remainder * * Side effects: * None * *----------------------------------------------------------------------------- */ #if defined(__GNUC__) static INLINE void Div643232(uint64 dividend, // IN uint32 divisor, // IN uint32 *quotient, // OUT uint32 *remainder) // OUT { __asm__( "divl %4" : "=a" (*quotient), "=d" (*remainder) : "0" ((uint32)dividend), "1" ((uint32)(dividend >> 32)), "rm" (divisor) : "cc" ); } #elif defined _MSC_VER static INLINE void Div643232(uint64 dividend, // IN uint32 divisor, // IN uint32 *quotient, // OUT uint32 *remainder) // OUT { __asm { mov eax, DWORD PTR [dividend] mov edx, DWORD PTR [dividend+4] div DWORD PTR [divisor] mov edi, DWORD PTR [quotient] mov [edi], eax mov edi, DWORD PTR [remainder] mov [edi], edx } } #else #error No compiler defined for Div643232 #endif #if defined(__GNUC__) /* *----------------------------------------------------------------------------- * * Div643264 -- * * Unsigned integer division: * The dividend is 64-bit wide * The divisor is 32-bit wide * The quotient is 64-bit wide * * Results: * Quotient and remainder * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Div643264(uint64 dividend, // IN uint32 divisor, // IN uint64 *quotient, // OUT uint32 *remainder) // OUT { uint32 hQuotient; uint32 lQuotient; __asm__( "divl %5" "\n\t" "movl %%eax, %0" "\n\t" "movl %4, %%eax" "\n\t" "divl %5" : "=&rm" (hQuotient), "=a" (lQuotient), "=d" (*remainder) : "1" ((uint32)(dividend >> 32)), "g" ((uint32)dividend), "rm" (divisor), "2" (0) : "cc" ); *quotient = (uint64)hQuotient << 32 | lQuotient; } #endif /* *----------------------------------------------------------------------------- * * Mul64x3264 -- * * Unsigned integer by fixed point multiplication, with rounding: * result = floor(multiplicand * multiplier * 2**(-shift) + 0.5) * * Unsigned 64-bit integer multiplicand. * Unsigned 32-bit fixed point multiplier, represented as * (multiplier, shift), where shift < 64. * * Result: * Unsigned 64-bit integer product. * *----------------------------------------------------------------------------- */ #if defined(__GNUC__) && \ (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 4)) && \ !defined(MUL64_NO_ASM) static INLINE uint64 Mul64x3264(uint64 multiplicand, uint32 multiplier, uint32 shift) { uint64 result; uint32 tmp1, tmp2; // ASSERT(shift >= 0 && shift < 64); __asm__("mov %%eax, %2\n\t" // Save lo(multiplicand) "mov %%edx, %%eax\n\t" // Get hi(multiplicand) "mull %4\n\t" // p2 = hi(multiplicand) * multiplier "xchg %%eax, %2\n\t" // Save lo(p2), get lo(multiplicand) "mov %%edx, %1\n\t" // Save hi(p2) "mull %4\n\t" // p1 = lo(multiplicand) * multiplier "addl %2, %%edx\n\t" // hi(p1) += lo(p2) "adcl $0, %1\n\t" // hi(p2) += carry from previous step "cmpl $32, %%ecx\n\t" // shift < 32? "jl 2f\n\t" // Go if so "shll $1, %%eax\n\t" // Save lo(p1) bit 31 in CF in case shift=32 "mov %%edx, %%eax\n\t" // result = hi(p2):hi(p1) >> (shift & 31) "mov %1, %%edx\n\t" "shrdl %%edx, %%eax\n\t" "mov $0, %2\n\t" "adcl $0, %2\n\t" // Get highest order bit shifted out, from CF "shrl %%cl, %%edx\n\t" "jmp 3f\n" "2:\n\t" "xor %2, %2\n\t" "shrdl %%edx, %%eax\n\t" // result = hi(p2):hi(p1):lo(p1) >> shift "adcl $0, %2\n\t" // Get highest order bit shifted out, from CF "shrdl %1, %%edx\n" "3:\n\t" "addl %2, %%eax\n\t" // result += highest order bit shifted out "adcl $0, %%edx" : "=A" (result), "=&r" (tmp1), "=&r" (tmp2) : "0" (multiplicand), "rm" (multiplier), "c" (shift) : "cc"); return result; } #elif defined _MSC_VER #pragma warning(disable: 4035) static INLINE uint64 Mul64x3264(uint64 multiplicand, uint32 multiplier, uint32 shift) { // ASSERT(shift >= 0 && shift < 64); __asm { mov eax, DWORD PTR [multiplicand+4] // Get hi(multiplicand) mul DWORD PTR [multiplier] // p2 = hi(multiplicand) * multiplier mov ecx, eax // Save lo(p2) mov ebx, edx // Save hi(p2) mov eax, DWORD PTR [multiplicand] // Get lo(multiplicand) mul DWORD PTR [multiplier+0] // p1 = lo(multiplicand) * multiplier add edx, ecx // hi(p1) += lo(p2) adc ebx, 0 // hi(p2) += carry from previous step mov ecx, DWORD PTR [shift] // Get shift cmp ecx, 32 // shift < 32? jl SHORT l2 // Go if so shl eax, 1 // Save lo(p1) bit 31 in CF in case shift=32 mov eax, edx // result = hi(p2):hi(p1) >> (shift & 31) mov edx, ebx shrd eax, edx, cl mov esi, 0 adc esi, 0 // Get highest order bit shifted out, from CF shr edx, cl jmp SHORT l3 l2: xor esi, esi shrd eax, edx, cl // result = hi(p2):hi(p1):lo(p1) >> shift adc esi, 0 // Get highest order bit shifted out, from CF shrd edx, ebx, cl l3: add eax, esi // result += highest order bit shifted out adc edx, 0 } // return with result in edx:eax } #pragma warning(default: 4035) #else #define MUL64_NO_ASM 1 #include "mul64.h" #endif /* *----------------------------------------------------------------------------- * * Muls64x32s64 -- * * Signed integer by fixed point multiplication, with rounding: * result = floor(multiplicand * multiplier * 2**(-shift) + 0.5) * * Signed 64-bit integer multiplicand. * Unsigned 32-bit fixed point multiplier, represented as * (multiplier, shift), where shift < 64. * * Result: * Signed 64-bit integer product. * *----------------------------------------------------------------------------- */ #if defined(__GNUC__) && \ (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 4)) && \ !defined(MUL64_NO_ASM) static INLINE int64 Muls64x32s64(int64 multiplicand, uint32 multiplier, uint32 shift) { int64 result; uint32 tmp1, tmp2; // ASSERT(shift >= 0 && shift < 64); __asm__("mov %%eax, %2\n\t" // Save lo(multiplicand) "mov %%edx, %%eax\n\t" // Get hi(multiplicand) "test %%eax, %%eax\n\t" // Check sign of multiplicand "jl 0f\n\t" // Go if negative "mull %4\n\t" // p2 = hi(multiplicand) * multiplier "jmp 1f\n" "0:\n\t" "mull %4\n\t" // p2 = hi(multiplicand) * multiplier "sub %4, %%edx\n" // hi(p2) += -1 * multiplier "1:\n\t" "xchg %%eax, %2\n\t" // Save lo(p2), get lo(multiplicand) "mov %%edx, %1\n\t" // Save hi(p2) "mull %4\n\t" // p1 = lo(multiplicand) * multiplier "addl %2, %%edx\n\t" // hi(p1) += lo(p2) "adcl $0, %1\n\t" // hi(p2) += carry from previous step "cmpl $32, %%ecx\n\t" // shift < 32? "jl 2f\n\t" // Go if so "shll $1, %%eax\n\t" // Save lo(p1) bit 31 in CF in case shift=32 "mov %%edx, %%eax\n\t" // result = hi(p2):hi(p1) >> (shift & 31) "mov %1, %%edx\n\t" "shrdl %%edx, %%eax\n\t" "mov $0, %2\n\t" "adcl $0, %2\n\t" // Get highest order bit shifted out from CF "sarl %%cl, %%edx\n\t" "jmp 3f\n" "2:\n\t" "xor %2, %2\n\t" "shrdl %%edx, %%eax\n\t" // result = hi(p2):hi(p1):lo(p1) >> shift "adcl $0, %2\n\t" // Get highest order bit shifted out from CF "shrdl %1, %%edx\n" "3:\n\t" "addl %2, %%eax\n\t" // result += highest order bit shifted out "adcl $0, %%edx" : "=A" (result), "=&r" (tmp1), "=&rm" (tmp2) : "0" (multiplicand), "rm" (multiplier), "c" (shift) : "cc"); return result; } #elif defined(_MSC_VER) #pragma warning(disable: 4035) static INLINE int64 Muls64x32s64(int64 multiplicand, uint32 multiplier, uint32 shift) { //ASSERT(shift >= 0 && shift < 64); __asm { mov eax, DWORD PTR [multiplicand+4] // Get hi(multiplicand) test eax, eax // Check sign of multiplicand jl SHORT l0 // Go if negative mul DWORD PTR [multiplier] // p2 = hi(multiplicand) * multiplier jmp SHORT l1 l0: mul DWORD PTR [multiplier] // p2 = hi(multiplicand) * multiplier sub edx, DWORD PTR [multiplier] // hi(p2) += -1 * multiplier l1: mov ecx, eax // Save lo(p2) mov ebx, edx // Save hi(p2) mov eax, DWORD PTR [multiplicand] // Get lo(multiplicand) mul DWORD PTR [multiplier] // p1 = lo(multiplicand) * multiplier add edx, ecx // hi(p1) += lo(p2) adc ebx, 0 // hi(p2) += carry from previous step mov ecx, DWORD PTR [shift] // Get shift cmp ecx, 32 // shift < 32? jl SHORT l2 // Go if so shl eax, 1 // Save lo(p1) bit 31 in CF in case shift=32 mov eax, edx // result = hi(p2):hi(p1) >> (shift & 31) mov edx, ebx shrd eax, edx, cl mov esi, 0 adc esi, 0 // Get highest order bit shifted out, from CF sar edx, cl jmp SHORT l3 l2: xor esi, esi shrd eax, edx, cl // result = hi(p2):hi(p1):lo(p1) << shift adc esi, 0 // Get highest order bit shifted out, from CF shrd edx, ebx, cl l3: add eax, esi // result += highest order bit shifted out adc edx, 0 } // return with result in edx:eax } #pragma warning(default: 4035) #endif #if defined __cplusplus } // extern "C" #endif #endif // _VM_BASIC_ASM_X86_H_ vsock-only/shared/compat_module.h 0000444 0000000 0000000 00000005127 13432725347 016144 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * compat_module.h -- */ #ifndef __COMPAT_MODULE_H__ # define __COMPAT_MODULE_H__ #include <linux/module.h> /* * Modules wishing to use the GPL license are required to include a * MODULE_LICENSE definition in their module source as of 2.4.10. */ #ifndef MODULE_LICENSE #define MODULE_LICENSE(license) #endif /* * To make use of our own home-brewed MODULE_INFO, we need macros to * concatenate two expressions to "__mod_", and and to convert an * expression into a string. I'm sure we've got these in our codebase, * but I'd rather not introduce such a dependency in a compat header. */ #ifndef __module_cat #define __module_cat_1(a, b) __mod_ ## a ## b #define __module_cat(a, b) __module_cat_1(a, b) #endif #ifndef __stringify #define __stringify_1(x) #x #define __stringify(x) __stringify_1(x) #endif /* * MODULE_INFO was born in 2.5.69. */ #ifndef MODULE_INFO #define MODULE_INFO(tag, info) \ static const char __module_cat(tag, __LINE__)[] \ __attribute__((section(".modinfo"), unused)) = __stringify(tag) "=" info #endif /* * MODULE_VERSION was born in 2.6.4. The earlier form appends a long "\0xxx" * string to the module's version, but that was removed in 2.6.10, so we'll * ignore it in our wrapper. */ #ifndef MODULE_VERSION #define MODULE_VERSION(_version) MODULE_INFO(version, _version) #endif /* * Linux kernel < 2.6.31 takes 'int' for 'bool' module parameters. * Linux kernel >= 3.3.0 takes 'bool' for 'bool' module parameters. * Kernels between the two take either. So flip switch at 3.0.0. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 0, 0) typedef bool compat_mod_param_bool; #else typedef int compat_mod_param_bool; #endif #endif /* __COMPAT_MODULE_H__ */ vsock-only/shared/vm_basic_asm_x86_64.h 0000444 0000000 0000000 00000042355 13432725350 016753 0 ustar root root /********************************************************* * Copyright (C) 1998-2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vm_basic_asm_x86_64.h * * Basic x86_64 asm macros. */ #ifndef _VM_BASIC_ASM_X86_64_H_ #define _VM_BASIC_ASM_X86_64_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" #ifndef VM_X86_64 #error "This file is x86-64 only!" #endif #if defined(_MSC_VER) && !defined(BORA_NO_WIN32_INTRINS) #ifdef __cplusplus extern "C" { #endif uint64 _umul128(uint64 multiplier, uint64 multiplicand, uint64 *highProduct); int64 _mul128(int64 multiplier, int64 multiplicand, int64 *highProduct); uint64 __shiftright128(uint64 lowPart, uint64 highPart, uint8 shift); #ifdef __cplusplus } #endif #pragma intrinsic(_umul128, _mul128, __shiftright128) #endif // _MSC_VER #if defined(__GNUC__) /* * GET_CURRENT_PC * * Returns the current program counter (i.e. instruction pointer i.e. rip * register on x86_64). In the example below: * * foo.c * L123: Foo(GET_CURRENT_PC()) * * the return value from GET_CURRENT_PC will point a debugger to L123. */ #define GET_CURRENT_PC() ({ \ void *__rip; \ asm("lea 0(%%rip), %0;\n\t" \ : "=r" (__rip)); \ __rip; \ }) /* * GET_CURRENT_LOCATION * * Updates the arguments with the values of the %rip, %rbp, and %rsp * registers at the current code location where the macro is invoked, * and the return address. */ #define GET_CURRENT_LOCATION(rip, rbp, rsp, retAddr) do { \ asm("lea 0(%%rip), %0\n" \ "mov %%rbp, %1\n" \ "mov %%rsp, %2\n" \ : "=r" (rip), "=r" (rbp), "=r" (rsp)); \ retAddr = (uint64) GetReturnAddress(); \ } while (0) #endif /* * FXSAVE/FXRSTOR * save/restore SIMD/MMX fpu state * * The pointer passed in must be 16-byte aligned. * * Intel and AMD processors behave differently w.r.t. fxsave/fxrstor. Intel * processors unconditionally save the exception pointer state (instruction * ptr., data ptr., and error instruction opcode). FXSAVE_ES1 and FXRSTOR_ES1 * work correctly for Intel processors. * * AMD processors only save the exception pointer state if ES=1. This leads to a * security hole whereby one process/VM can inspect the state of another process * VM. The AMD recommended workaround involves clobbering the exception pointer * state unconditionally, and this is implemented in FXRSTOR_AMD_ES0. Note that * FXSAVE_ES1 will only save the exception pointer state for AMD processors if * ES=1. * * The workaround (FXRSTOR_AMD_ES0) only costs 1 cycle more than just doing an * fxrstor, on both AMD Opteron and Intel Core CPUs. */ #if defined(__GNUC__) static INLINE void FXSAVE_ES1(void *save) { __asm__ __volatile__ ("fxsaveq %0 \n" : "=m" (*(uint8 *)save) : : "memory"); } static INLINE void FXSAVE_COMPAT_ES1(void *save) { __asm__ __volatile__ ("fxsave %0 \n" : "=m" (*(uint8 *)save) : : "memory"); } static INLINE void FXRSTOR_ES1(const void *load) { __asm__ __volatile__ ("fxrstorq %0 \n" : : "m" (*(const uint8 *)load) : "memory"); } static INLINE void FXRSTOR_COMPAT_ES1(const void *load) { __asm__ __volatile__ ("fxrstor %0 \n" : : "m" (*(const uint8 *)load) : "memory"); } static INLINE void FXRSTOR_AMD_ES0(const void *load) { uint64 dummy = 0; __asm__ __volatile__ ("fnstsw %%ax \n" // Grab x87 ES bit "bt $7,%%ax \n" // Test ES bit "jnc 1f \n" // Jump if ES=0 "fnclex \n" // ES=1. Clear it so fild doesn't trap "1: \n" "ffree %%st(7) \n" // Clear tag bit - avoid poss. stack overflow "fildl %0 \n" // Dummy Load from "safe address" changes all // x87 exception pointers. "fxrstorq %1 \n" : : "m" (dummy), "m" (*(const uint8 *)load) : "ax", "memory"); } #endif /* __GNUC__ */ /* * XSAVE/XRSTOR * save/restore GSSE/SIMD/MMX fpu state * * The pointer passed in must be 64-byte aligned. * See above comment for more information. */ #if defined(__GNUC__) && (defined(VMM) || defined(VMKERNEL) || defined(FROBOS)) static INLINE void XSAVE_ES1(void *save, uint64 mask) { #if __GNUC__ < 4 || __GNUC__ == 4 && __GNUC_MINOR__ == 1 __asm__ __volatile__ ( ".byte 0x48, 0x0f, 0xae, 0x21 \n" : : "c" ((uint8 *)save), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #else __asm__ __volatile__ ( "xsaveq %0 \n" : "=m" (*(uint8 *)save) : "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #endif } static INLINE void XSAVE_COMPAT_ES1(void *save, uint64 mask) { #if __GNUC__ < 4 || __GNUC__ == 4 && __GNUC_MINOR__ == 1 __asm__ __volatile__ ( ".byte 0x0f, 0xae, 0x21 \n" : : "c" ((uint8 *)save), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #else __asm__ __volatile__ ( "xsave %0 \n" : "=m" (*(uint8 *)save) : "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #endif } static INLINE void XSAVEOPT_ES1(void *save, uint64 mask) { __asm__ __volatile__ ( ".byte 0x48, 0x0f, 0xae, 0x31 \n" : : "c" ((uint8 *)save), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); } static INLINE void XRSTOR_ES1(const void *load, uint64 mask) { #if __GNUC__ < 4 || __GNUC__ == 4 && __GNUC_MINOR__ == 1 __asm__ __volatile__ ( ".byte 0x48, 0x0f, 0xae, 0x29 \n" : : "c" ((const uint8 *)load), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #else __asm__ __volatile__ ( "xrstorq %0 \n" : : "m" (*(const uint8 *)load), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #endif } static INLINE void XRSTOR_COMPAT_ES1(const void *load, uint64 mask) { #if __GNUC__ < 4 || __GNUC__ == 4 && __GNUC_MINOR__ == 1 __asm__ __volatile__ ( ".byte 0x0f, 0xae, 0x29 \n" : : "c" ((const uint8 *)load), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #else __asm__ __volatile__ ( "xrstor %0 \n" : : "m" (*(const uint8 *)load), "a" ((uint32)mask), "d" ((uint32)(mask >> 32)) : "memory"); #endif } static INLINE void XRSTOR_AMD_ES0(const void *load, uint64 mask) { uint64 dummy = 0; __asm__ __volatile__ ("fnstsw %%ax \n" // Grab x87 ES bit "bt $7,%%ax \n" // Test ES bit "jnc 1f \n" // Jump if ES=0 "fnclex \n" // ES=1. Clear it so fild doesn't trap "1: \n" "ffree %%st(7) \n" // Clear tag bit - avoid poss. stack overflow "fildl %0 \n" // Dummy Load from "safe address" changes all // x87 exception pointers. "mov %%ebx, %%eax \n" #if __GNUC__ < 4 || __GNUC__ == 4 && __GNUC_MINOR__ == 1 ".byte 0x48, 0x0f, 0xae, 0x29 \n" : : "m" (dummy), "c" ((const uint8 *)load), "b" ((uint32)mask), "d" ((uint32)(mask >> 32)) #else "xrstorq %1 \n" : : "m" (dummy), "m" (*(const uint8 *)load), "b" ((uint32)mask), "d" ((uint32)(mask >> 32)) #endif : "eax", "memory"); } #endif /* __GNUC__ */ /* * XTEST * Return TRUE if processor is in transaction region. * */ #if defined(__GNUC__) && (defined(VMM) || defined(VMKERNEL) || defined(FROBOS)) static INLINE Bool xtest(void) { uint8 al; __asm__ __volatile__(".byte 0x0f, 0x01, 0xd6 # xtest \n" "setnz %%al\n" : "=a"(al) : : "cc"); return al; } #endif /* __GNUC__ */ /* *----------------------------------------------------------------------------- * * Mul64x6464 -- * * Unsigned integer by fixed point multiplication, with rounding: * result = floor(multiplicand * multiplier * 2**(-shift) + 0.5) * * Unsigned 64-bit integer multiplicand. * Unsigned 64-bit fixed point multiplier, represented as * (multiplier, shift), where shift < 64. * * Result: * Unsigned 64-bit integer product. * *----------------------------------------------------------------------------- */ #if defined(__GNUC__) && !defined(MUL64_NO_ASM) static INLINE uint64 Mul64x6464(uint64 multiplicand, uint64 multiplier, uint32 shift) { /* * Implementation: * Multiply 64x64 bits to yield a full 128-bit product. * Clear the carry bit (needed for the shift == 0 case). * Shift result in RDX:RAX right by "shift". * Add the carry bit. (If shift > 0, this is the highest order bit * that was discarded by the shift; else it is 0.) * Return the low-order 64 bits of the above. * */ uint64 result, dummy; __asm__("mulq %3 \n\t" "clc \n\t" "shrdq %b4, %1, %0 \n\t" "adc $0, %0 \n\t" : "=a" (result), "=d" (dummy) : "0" (multiplier), "rm" (multiplicand), "c" (shift) : "cc"); return result; } #elif defined(_MSC_VER) && !defined(MUL64_NO_ASM) static INLINE uint64 Mul64x6464(uint64 multiplicand, uint64 multiplier, uint32 shift) { /* * Unfortunately, MSVC intrinsics don't give us access to the carry * flag after a 128-bit shift, so the implementation is more * awkward: * Multiply 64x64 bits to yield a full 128-bit product. * Shift result right by "shift". * If shift != 0, extract and add in highest order bit that was * discarded by the shift. * Return the low-order 64 bits of the above. */ uint64 tmplo, tmphi; tmplo = _umul128(multiplicand, multiplier, &tmphi); if (shift == 0) { return tmplo; } else { return __shiftright128(tmplo, tmphi, (uint8) shift) + ((tmplo >> (shift - 1)) & 1); } } #else #define MUL64_NO_ASM 1 #include "mul64.h" #endif /* *----------------------------------------------------------------------------- * * Muls64x64s64 -- * * Signed integer by fixed point multiplication, with rounding: * result = floor(multiplicand * multiplier * 2**(-shift) + 0.5) * * Signed 64-bit integer multiplicand. * Unsigned 64-bit fixed point multiplier, represented as * (multiplier, shift), where shift < 64. * * Result: * Signed 64-bit integer product. * *----------------------------------------------------------------------------- */ #if defined(__GNUC__) && !defined(MUL64_NO_ASM) static inline int64 Muls64x64s64(int64 multiplicand, int64 multiplier, uint32 shift) { int64 result, dummy; /* Implementation: * Multiply 64x64 bits to yield a full 128-bit product. * Clear the carry bit (needed for the shift == 0 case). * Shift result in RDX:RAX right by "shift". * Add the carry bit. (If shift > 0, this is the highest order bit * that was discarded by the shift; else it is 0.) * Return the low-order 64 bits of the above. * * Note: using the unsigned shrd instruction is correct because * shift < 64 and we return only the low 64 bits of the shifted * result. */ __asm__("imulq %3 \n\t" "clc \n\t" "shrdq %b4, %1, %0 \n\t" "adc $0, %0 \n\t" : "=a" (result), "=d" (dummy) : "0" (multiplier), "rm" (multiplicand), "c" (shift) : "cc"); return result; } #elif defined(_MSC_VER) && !defined(MUL64_NO_ASM) static INLINE int64 Muls64x64s64(int64 multiplicand, int64 multiplier, uint32 shift) { /* * Unfortunately, MSVC intrinsics don't give us access to the carry * flag after a 128-bit shift, so the implementation is more * awkward: * Multiply 64x64 bits to yield a full 128-bit product. * Shift result right by "shift". * If shift != 0, extract and add in highest order bit that was * discarded by the shift. * Return the low-order 64 bits of the above. * * Note: using an unsigned shift is correct because shift < 64 and * we return only the low 64 bits of the shifted result. */ int64 tmplo, tmphi; tmplo = _mul128(multiplicand, multiplier, &tmphi); if (shift == 0) { return tmplo; } else { return __shiftright128(tmplo, tmphi, (uint8) shift) + ((tmplo >> (shift - 1)) & 1); } } #endif #ifndef MUL64_NO_ASM /* *----------------------------------------------------------------------------- * * Mul64x3264 -- * * Unsigned integer by fixed point multiplication, with rounding: * result = floor(multiplicand * multiplier * 2**(-shift) + 0.5) * * Unsigned 64-bit integer multiplicand. * Unsigned 32-bit fixed point multiplier, represented as * (multiplier, shift), where shift < 64. * * Result: * Unsigned 64-bit integer product. * *----------------------------------------------------------------------------- */ static INLINE uint64 Mul64x3264(uint64 multiplicand, uint32 multiplier, uint32 shift) { return Mul64x6464(multiplicand, multiplier, shift); } /* *----------------------------------------------------------------------------- * * Muls64x32s64 -- * * Signed integer by fixed point multiplication, with rounding: * result = floor(multiplicand * multiplier * 2**(-shift) + 0.5) * * Signed 64-bit integer multiplicand. * Unsigned 32-bit fixed point multiplier, represented as * (multiplier, shift), where shift < 64. * * Result: * Signed 64-bit integer product. * *----------------------------------------------------------------------------- */ static INLINE int64 Muls64x32s64(int64 multiplicand, uint32 multiplier, uint32 shift) { return Muls64x64s64(multiplicand, multiplier, shift); } #endif #if defined(__GNUC__) static INLINE void * uint64set(void *dst, uint64 val, uint64 count) { int dummy0; int dummy1; __asm__ __volatile__("\t" "cld" "\n\t" "rep ; stosq" "\n" : "=c" (dummy0), "=D" (dummy1) : "0" (count), "1" (dst), "a" (val) : "memory", "cc"); return dst; } #endif /* *----------------------------------------------------------------------------- * * Div643232 -- * * Unsigned integer division: * The dividend is 64-bit wide * The divisor is 32-bit wide * The quotient is 32-bit wide * * Use this function if you are certain that the quotient will fit in 32 bits, * If that is not the case, a #DE exception was generated in 32-bit version, * but not in this 64-bit version. So please be careful. * * Results: * Quotient and remainder * * Side effects: * None * *----------------------------------------------------------------------------- */ #if defined(__GNUC__) || defined(_MSC_VER) static INLINE void Div643232(uint64 dividend, // IN uint32 divisor, // IN uint32 *quotient, // OUT uint32 *remainder) // OUT { *quotient = (uint32)(dividend / divisor); *remainder = (uint32)(dividend % divisor); } #endif /* *----------------------------------------------------------------------------- * * Div643264 -- * * Unsigned integer division: * The dividend is 64-bit wide * The divisor is 32-bit wide * The quotient is 64-bit wide * * Results: * Quotient and remainder * * Side effects: * None * *----------------------------------------------------------------------------- */ #if defined(__GNUC__) static INLINE void Div643264(uint64 dividend, // IN uint32 divisor, // IN uint64 *quotient, // OUT uint32 *remainder) // OUT { *quotient = dividend / divisor; *remainder = dividend % divisor; } #endif #endif // _VM_BASIC_ASM_X86_64_H_ vsock-only/shared/compat_timer.h 0000444 0000000 0000000 00000006551 13432725347 016001 0 ustar root root /********************************************************* * Copyright (C) 2002 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_TIMER_H__ # define __COMPAT_TIMER_H__ /* * The del_timer_sync() API appeared in 2.3.43 * It became reliable in 2.4.0-test3 * * --hpreg */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 4, 0) # define compat_del_timer_sync(timer) del_timer_sync(timer) #else # if LINUX_VERSION_CODE < KERNEL_VERSION(2, 3, 43) /* 2.3.43 removed asm/softirq.h's reference to bh_base. */ # include <linux/interrupt.h> # endif # include <asm/softirq.h> static inline int compat_del_timer_sync(struct timer_list *timer) // IN { int wasPending; start_bh_atomic(); wasPending = del_timer(timer); end_bh_atomic(); return wasPending; } #endif /* * The msleep_interruptible() API appeared in 2.6.9. * It is based on the msleep() API, which appeared in 2.4.29. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 9) # include <linux/delay.h> # define compat_msleep_interruptible(msecs) msleep_interruptible(msecs) # define compat_msleep(msecs) msleep(msecs) #else # include <linux/sched.h> /* * msecs_to_jiffies appeared in 2.6.7. For earlier kernels, * fall back to slow-case code (we don't use this operation * enough to need the performance). */ # if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 7) # define msecs_to_jiffies(msecs) (((msecs) * HZ + 999) / 1000) # endif /* * set_current_state appeared in 2.2.18. */ # if LINUX_VERSION_CODE < KERNEL_VERSION(2, 2, 18) # define set_current_state(a) do { current->state = (a); } while(0) # endif static inline void compat_msleep_interruptible(unsigned long msecs) // IN { set_current_state(TASK_INTERRUPTIBLE); schedule_timeout(msecs_to_jiffies(msecs) + 1); } static inline void compat_msleep(unsigned long msecs) // IN { set_current_state(TASK_UNINTERRUPTIBLE); schedule_timeout(msecs_to_jiffies(msecs) + 1); } #endif /* * There is init_timer_deferrable() since 2.6.22. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 22) # define compat_init_timer_deferrable(timer) init_timer_deferrable(timer) #else # define compat_init_timer_deferrable(timer) init_timer(timer) #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,15) static inline void compat_setup_timer(struct timer_list * timer, void (*function)(unsigned long), unsigned long data) { timer->function = function; timer->data = data; init_timer(timer); } #else # define compat_setup_timer(timer, function, data) \ setup_timer(timer, function, data) #endif #endif /* __COMPAT_TIMER_H__ */ vsock-only/shared/vmciKernelAPI1.h 0000444 0000000 0000000 00000020110 13432725350 016005 0 ustar root root /********************************************************* * Copyright (C) 2010,2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmciKernelAPI1.h -- * * Kernel API (v1) exported from the VMCI host and guest drivers. */ #ifndef __VMCI_KERNELAPI_1_H__ #define __VMCI_KERNELAPI_1_H__ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" #include "vmci_defs.h" #include "vmci_call_defs.h" #if defined __cplusplus extern "C" { #endif /* VMCI module namespace on vmkernel. */ #define MOD_VMCI_NAMESPACE "com.vmware.vmci" /* Define version 1. */ #undef VMCI_KERNEL_API_VERSION #define VMCI_KERNEL_API_VERSION_1 1 #define VMCI_KERNEL_API_VERSION VMCI_KERNEL_API_VERSION_1 /* Macros to operate on the driver version number. */ #define VMCI_MAJOR_VERSION(v) (((v) >> 16) & 0xffff) #define VMCI_MINOR_VERSION(v) ((v) & 0xffff) #if defined(_WIN32) /* Path to callback object in object manager, for Windows only. */ #define VMCI_CALLBACK_OBJECT_PATH L"\\Callback\\VMCIDetachCB" #endif // _WIN32 /* VMCI Device Usage API. */ #if defined(__linux__) && !defined(VMKERNEL) #define vmci_device_get(_a, _b, _c, _d) 1 #define vmci_device_release(_x) #else // !linux typedef void (VMCI_DeviceShutdownFn)(void *deviceRegistration, void *userData); Bool vmci_device_get(uint32 *apiVersion, VMCI_DeviceShutdownFn *deviceShutdownCB, void *userData, void **deviceRegistration); void vmci_device_release(void *deviceRegistration); #endif // !linux #if defined(_WIN32) /* Called when the client is unloading, for Windows only. */ void vmci_exit(void); #endif // _WIN32 /* VMCI Datagram API. */ int vmci_datagram_create_handle(uint32 resourceId, uint32 flags, VMCIDatagramRecvCB recvCB, void *clientData, VMCIHandle *outHandle); int vmci_datagram_create_handle_priv(uint32 resourceID, uint32 flags, VMCIPrivilegeFlags privFlags, VMCIDatagramRecvCB recvCB, void *clientData, VMCIHandle *outHandle); int vmci_datagram_destroy_handle(VMCIHandle handle); int vmci_datagram_send(VMCIDatagram *msg); /* VMCI Utility API. */ VMCIId vmci_get_context_id(void); #if defined(__linux__) && !defined(VMKERNEL) /* Returned value is a bool, 0 for false, 1 for true. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 5, 0) int vmci_is_context_owner(VMCIId contextID, kuid_t uid); #else int vmci_is_context_owner(VMCIId contextID, uid_t uid); #endif #else // !linux || VMKERNEL /* Returned value is a VMCI error code. */ int vmci_is_context_owner(VMCIId contextID, void *hostUser); #endif // !linux || VMKERNEL uint32 vmci_version(void); int vmci_cid_2_host_vm_id(VMCIId contextID, void *hostVmID, size_t hostVmIDLen); /* VMCI Event API. */ typedef void (*VMCI_EventCB)(VMCIId subID, VMCI_EventData *ed, void *clientData); int vmci_event_subscribe(VMCI_Event event, #if !defined(__linux__) && !defined(__FreeBSD__) || defined(VMKERNEL) uint32 flags, #endif // !linux && !FreeBSD || VMKERNEL VMCI_EventCB callback, void *callbackData, VMCIId *subID); int vmci_event_unsubscribe(VMCIId subID); /* VMCI Context API */ VMCIPrivilegeFlags vmci_context_get_priv_flags(VMCIId contextID); /* VMCI Queue Pair API. */ typedef struct VMCIQPair VMCIQPair; int vmci_qpair_alloc(VMCIQPair **qpair, VMCIHandle *handle, uint64 produceQSize, uint64 consumeQSize, VMCIId peer, uint32 flags, VMCIPrivilegeFlags privFlags); int vmci_qpair_detach(VMCIQPair **qpair); int vmci_qpair_get_produce_indexes(const VMCIQPair *qpair, uint64 *producerTail, uint64 *consumerHead); int vmci_qpair_get_consume_indexes(const VMCIQPair *qpair, uint64 *consumerTail, uint64 *producerHead); int64 vmci_qpair_produce_free_space(const VMCIQPair *qpair); int64 vmci_qpair_produce_buf_ready(const VMCIQPair *qpair); int64 vmci_qpair_consume_free_space(const VMCIQPair *qpair); int64 vmci_qpair_consume_buf_ready(const VMCIQPair *qpair); ssize_t vmci_qpair_enqueue(VMCIQPair *qpair, const void *buf, size_t bufSize, int mode); ssize_t vmci_qpair_dequeue(VMCIQPair *qpair, void *buf, size_t bufSize, int mode); ssize_t vmci_qpair_peek(VMCIQPair *qpair, void *buf, size_t bufSize, int mode); #if (defined(__APPLE__) && !defined (VMX86_TOOLS)) || \ (defined(__linux__) && defined(__KERNEL__)) || \ (defined(_WIN32) && defined(WINNT_DDK)) /* * Environments that support struct iovec */ ssize_t vmci_qpair_enquev(VMCIQPair *qpair, void *iov, size_t iovSize, int mode); ssize_t vmci_qpair_dequev(VMCIQPair *qpair, void *iov, size_t iovSize, int mode); ssize_t vmci_qpair_peekv(VMCIQPair *qpair, void *iov, size_t iovSize, int mode); #endif /* Systems that support struct iovec */ /* Typedefs for all of the above, used by the IOCTLs and the kernel library. */ typedef void (VMCI_DeviceReleaseFct)(void *); typedef int (VMCIDatagram_CreateHndFct)(VMCIId, uint32, VMCIDatagramRecvCB, void *, VMCIHandle *); typedef int (VMCIDatagram_CreateHndPrivFct)(VMCIId, uint32, VMCIPrivilegeFlags, VMCIDatagramRecvCB, void *, VMCIHandle *); typedef int (VMCIDatagram_DestroyHndFct)(VMCIHandle); typedef int (VMCIDatagram_SendFct)(VMCIDatagram *); typedef VMCIId (VMCI_GetContextIDFct)(void); typedef uint32 (VMCI_VersionFct)(void); typedef int (VMCI_ContextID2HostVmIDFct)(VMCIId, void *, size_t); typedef int (VMCI_IsContextOwnerFct)(VMCIId, void *); typedef int (VMCIEvent_SubscribeFct)(VMCI_Event, uint32, VMCI_EventCB, void *, VMCIId *); typedef int (VMCIEvent_UnsubscribeFct)(VMCIId); typedef VMCIPrivilegeFlags (VMCIContext_GetPrivFlagsFct)(VMCIId); typedef int (VMCIQPair_AllocFct)(VMCIQPair **, VMCIHandle *, uint64, uint64, VMCIId, uint32, VMCIPrivilegeFlags); typedef int (VMCIQPair_DetachFct)(VMCIQPair **); typedef int (VMCIQPair_GetProduceIndexesFct)(const VMCIQPair *, uint64 *, uint64 *); typedef int (VMCIQPair_GetConsumeIndexesFct)(const VMCIQPair *, uint64 *, uint64 *); typedef int64 (VMCIQPair_ProduceFreeSpaceFct)(const VMCIQPair *); typedef int64 (VMCIQPair_ProduceBufReadyFct)(const VMCIQPair *); typedef int64 (VMCIQPair_ConsumeFreeSpaceFct)(const VMCIQPair *); typedef int64 (VMCIQPair_ConsumeBufReadyFct)(const VMCIQPair *); typedef ssize_t (VMCIQPair_EnqueueFct)(VMCIQPair *, const void *, size_t, int); typedef ssize_t (VMCIQPair_DequeueFct)(VMCIQPair *, void *, size_t, int); typedef ssize_t (VMCIQPair_PeekFct)(VMCIQPair *, void *, size_t, int); typedef ssize_t (VMCIQPair_EnqueueVFct)(VMCIQPair *qpair, void *, size_t, int); typedef ssize_t (VMCIQPair_DequeueVFct)(VMCIQPair *qpair, void *, size_t, int); typedef ssize_t (VMCIQPair_PeekVFct)(VMCIQPair *qpair, void *, size_t, int); #if defined __cplusplus } // extern "C" #endif #endif /* !__VMCI_KERNELAPI_1_H__ */ vsock-only/shared/vmci_infrastructure.h 0000444 0000000 0000000 00000010362 13432725350 017401 0 ustar root root /********************************************************* * Copyright (C) 2006,2014 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmci_infrastructure.h -- * * This file implements the VMCI infrastructure. */ #ifndef _VMCI_INFRASTRUCTURE_H_ #define _VMCI_INFRASTRUCTURE_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_DISTRIBUTE #include "includeCheck.h" #include "vmware.h" #include "vmci_defs.h" #if defined __cplusplus extern "C" { #endif typedef enum { VMCIOBJ_VMX_VM = 10, VMCIOBJ_CONTEXT, VMCIOBJ_SOCKET, VMCIOBJ_NOT_SET, } VMCIObjType; /* For storing VMCI structures in file handles. */ typedef struct VMCIObj { void *ptr; VMCIObjType type; } VMCIObj; /* Guestcalls currently support a maximum of 8 uint64 arguments. */ #define VMCI_GUESTCALL_MAX_ARGS_SIZE 64 /* * Structure used for checkpointing the doorbell mappings. It is * written to the checkpoint as is, so changing this structure will * break checkpoint compatibility. */ typedef struct VMCIDoorbellCptState { VMCIHandle handle; uint64 bitmapIdx; } VMCIDoorbellCptState; /* Used to determine what checkpoint state to get and set. */ #define VMCI_NOTIFICATION_CPT_STATE 0x1 #define VMCI_WELLKNOWN_CPT_STATE 0x2 #define VMCI_DG_OUT_STATE 0x3 #define VMCI_DG_IN_STATE 0x4 #define VMCI_DG_IN_SIZE_STATE 0x5 #define VMCI_DOORBELL_CPT_STATE 0x6 #define VMCI_DG_HYPERVISOR_SAVE_STATE_SIZE 0x7 #define VMCI_DG_HYPERVISOR_SAVE_STATE 0x8 /* Used to control the VMCI device in the vmkernel */ #define VMCI_DEV_RESET 0x01 #define VMCI_DEV_QP_RESET 0x02 // DEPRECATED #define VMCI_DEV_QUIESCE 0x03 #define VMCI_DEV_UNQUIESCE 0x04 #define VMCI_DEV_QP_BREAK_SHARING 0x05 // DEPRECATED #define VMCI_DEV_RESTORE_SYNC 0x06 #define VMCI_DEV_BMASTER_OFF 0x07 #define VMCI_DEV_BMASTER_ON 0x08 /* *------------------------------------------------------------------------- * * VMCI_Hash -- * * Hash function used by the Simple Datagram API. Based on the djb2 * hash function by Dan Bernstein. * * Result: * Returns guest call size. * * Side effects: * None. * *------------------------------------------------------------------------- */ static INLINE int VMCI_Hash(VMCIHandle handle, // IN unsigned size) // IN { unsigned i; int hash = 5381; const uint64 handleValue = QWORD(handle.resource, handle.context); for (i = 0; i < sizeof handle; i++) { hash = ((hash << 5) + hash) + (uint8)(handleValue >> (i * 8)); } return hash & (size - 1); } /* *------------------------------------------------------------------------- * * VMCI_HashId -- * * Hash function used by the Simple Datagram API. Hashes only a VMCI id * (not the full VMCI handle) Based on the djb2 * hash function by Dan Bernstein. * * Result: * Returns guest call size. * * Side effects: * None. * *------------------------------------------------------------------------- */ static INLINE int VMCI_HashId(VMCIId id, // IN unsigned size) // IN { unsigned i; int hash = 5381; for (i = 0; i < sizeof id; i++) { hash = ((hash << 5) + hash) + (uint8)(id >> (i * 8)); } return hash & (size - 1); } #if defined __cplusplus } // extern "C" #endif #endif // _VMCI_INFRASTRUCTURE_H_ vsock-only/shared/autoconf/ 0000755 0000000 0000000 00000000000 13432726375 014760 5 ustar root root vsock-only/shared/autoconf/file_operations_fsync.c 0000444 0000000 0000000 00000002746 13432725347 021515 0 ustar root root /********************************************************* * Copyright (C) 2011 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * Linux v3.1 added 2 params to fsync for fine-grained locking control. * But SLES11 SP2 has backported the change to its 3.0 kernel, * so we can't rely solely on kernel version to determine number of * arguments. */ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(3, 0, 0) # error This compile test intentionally fails. #else #include <linux/fs.h> #include <linux/types.h> /* loff_t */ static int TestFsync(struct file *file, loff_t start, loff_t end, int datasync) { return 0; } struct file_operations testFO = { .fsync = TestFsync, }; #endif vsock-only/shared/autoconf/getsb1.c 0000444 0000000 0000000 00000003076 13432725347 016313 0 ustar root root /********************************************************* * Copyright (C) 2006 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 19) #include <linux/fs.h> /* * Around 2.6.18, a pointer to a vfsmount was added to get_sb. Red Hat * backported this behavior into a 2.6.17 kernel. * * This test will fail on a kernel with such a patch. */ static struct super_block * LinuxDriverGetSb(struct file_system_type *fs_type, int flags, const char *dev_name, void *rawData) { return 0; } struct file_system_type fs_type = { .get_sb = LinuxDriverGetSb }; #else #error "This test intentionally fails on 2.6.19 or newer kernels." #endif vsock-only/shared/autoconf/file_operations_flush.c 0000444 0000000 0000000 00000002664 13432725347 021513 0 ustar root root /********************************************************* * Copyright (C) 2013-2014 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * Linux v2.6.18 added an owner parameter to flush. * But SLES10 has backported the change to its 2.6.16.60 kernel, * so we can't rely solely on kernel version to determine number of * arguments. * * This test will fail on a kernel with such a patch. */ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 18) #error This compile test intentionally fails on 2.6.18 and newer kernels. #else #include <linux/fs.h> static int TestFlush(struct file *file); { return 0; } struct file_operations testFO = { .flush = TestFlush, }; #endif vsock-only/shared/autoconf/netif_num_params.c 0000444 0000000 0000000 00000003352 13432725347 020452 0 ustar root root /********************************************************* * Copyright (C) 2009 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * Detect whether netif_rx_complete (and netif_rx_schedule) take a single * napi_struct argument. The foundation was laid whith introducing Generic * Receive Offload infrastructure but dropping unneeded net_device argument * did not happen till few commits later so we can't simply test for presence * of NETIF_F_GRO. * * Test succeeds if netif_rx_complete takes dev & napi arguments, or if it * takes dev argument only (kernels before 2.6.24). Test fails if netif_rx_complete * takes only single napi argument. */ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 30) # error This compile test intentionally fails. #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 29) #include <linux/netdevice.h> #ifdef NETIF_F_GRO void test_netif_rx_complete(struct net_device *dev, struct napi_struct *napi) { netif_rx_complete(dev, napi); } #endif #endif vsock-only/shared/autoconf/cachector.c 0000444 0000000 0000000 00000003270 13432725347 017055 0 ustar root root /********************************************************* * Copyright (C) 2006 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" /* * Between 2.6.23 and 2.6.24-rc1 ctor prototype was changed from * ctor(ptr, cache, flags) to ctor(cache, ptr). Unfortunately there * is no typedef for ctor, so we have to redefine kmem_cache_create * to find out ctor prototype. This assumes that kmem_cache_create * takes 5 arguments and not 6 - that change occured between * 2.6.22 and 2.6.23-rc1. If prototype matches, then this is old * kernel. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 24) #error "This test intentionally fails on 2.6.24 and newer kernels." #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 23) #include <linux/slab.h> struct kmem_cache *kmem_cache_create(const char *, size_t, size_t, unsigned long, void (*)(void *, struct kmem_cache *, unsigned long)); #endif vsock-only/shared/autoconf/geninclude.c 0000444 0000000 0000000 00000002321 13432725347 017233 0 ustar root root /********************************************************* * Copyright (C) 2003 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" #ifdef CONFIG_X86_VOYAGER APATH/mach-voyager #endif #ifdef CONFIG_X86_VISWS APATH/mach-visws #endif #ifdef CONFIG_X86_NUMAQ APATH/mach-numaq #endif #ifdef CONFIG_X86_BIGSMP APATH/mach-bigsmp #endif #ifdef CONFIG_X86_SUMMIT APATH/mach-summit #endif #ifdef CONFIG_X86_GENERICARCH APATH/mach-generic #endif APATH/mach-default vsock-only/shared/autoconf/cachector1.c 0000444 0000000 0000000 00000003075 13432725347 017141 0 ustar root root /********************************************************* * Copyright (C) 2008 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" /* * Between 2.6.27-rc1 and 2.6.27-rc2 ctor prototype was changed from * ctor(cache, ptr) to ctor(ptr). Unfortunately there * is no typedef for ctor, so we have to redefine kmem_cache_create * to find out ctor prototype. If prototype matches, then this is old * kernel. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 28) #error "This test intentionally fails on 2.6.28 and newer kernels." #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 26) #include <linux/slab.h> struct kmem_cache *kmem_cache_create(const char *, size_t, size_t, unsigned long, void (*)(struct kmem_cache *, void *)); #endif vsock-only/shared/autoconf/wait_on_bit.c 0000444 0000000 0000000 00000003306 13432725347 017420 0 ustar root root /********************************************************* * Copyright (C) 2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(3, 17, 0) && \ LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 13) #include <linux/fs.h> #include <linux/wait.h> #include <linux/sched.h> unsigned long test_bits; /* * After 3.17.0, wait_on_bit changed its interface to remove the action * callback argument and this was backported to some Linux kernel versions * such as 3.10 for the RHEL 7.3 version. * * This test will fail on a kernel with such a patch. */ int test(void) { return wait_on_bit(&test_bits, 0, NULL, TASK_UNINTERRUPTIBLE); } #elif LINUX_VERSION_CODE >= KERNEL_VERSION(3, 17, 0) #error "This test intentionally fails on 3.17.0 and newer kernels." #else /* * It must be older than 2.6.13 in which case we don't use the function. */ #endif vsock-only/shared/autoconf/cachecreate.c 0000444 0000000 0000000 00000003210 13432725347 017343 0 ustar root root /********************************************************* * Copyright (C) 2006 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" /* * All kernels before 2.6.22 take 6 arguments. All kernels since * 2.6.23-rc1 take 5 arguments. Only kernels between 2.6.22 and * 2.6.23-rc1 are questionable - we could ignore them if we wanted, * nobody cares about them even now. But unfortunately RedHat is * re-releasing 2.6.X-rc kernels under 2.6.(X-1) name, so they * are releasing 2.6.23-rc1 as 2.6.22-5055-something, so we have * to do autodetection for them. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 22) /* Success... */ #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 23) #error "This test intentionally fails on 2.6.23 and newer kernels." #else #include <linux/slab.h> struct kmem_cache *kmemtest(void) { return kmem_cache_create("test", 12, 0, 0, NULL, NULL); } #endif vsock-only/shared/autoconf/skblin.c 0000444 0000000 0000000 00000002572 13432725347 016410 0 ustar root root /********************************************************* * Copyright (C) 2006 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * Detect whether skb_linearize takes one or two arguments. */ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE <= KERNEL_VERSION(2, 6, 17) /* * Since 2.6.18 all kernels have single-argument skb_linearize. For * older kernels use autodetection. Not using autodetection on newer * kernels saves us from compile failure on some post 2.6.18 kernels * which do not have selfcontained skbuff.h. */ #include <linux/skbuff.h> int test_skb_linearize(struct sk_buff *skb) { return skb_linearize(skb); } #endif vsock-only/shared/autoconf/dalias.c 0000444 0000000 0000000 00000003217 13432725347 016360 0 ustar root root /********************************************************* * Copyright (C) 2015-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(3, 19, 0) #if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 6, 0) #include <linux/dcache.h> #include <linux/list.h> /* * After 3.19.0, the dentry d_alias field was moved. Fedora * backported this behavior into earlier kernel versions. * The type of the d_alias field changed from 3.6 onwards * which was a list head to being a list node. The check * for earlier than 3.6 is done separately. * * This test will fail on a kernel with such a patch. */ void test(void) { struct dentry aliasDentry; INIT_HLIST_NODE(&aliasDentry.d_alias); } #else /* Intentionally passes for earlier than 3.6.0 kernels as a separate test is done. */ #endif #else #error "This test intentionally fails on 3.19.0 or newer kernels." #endif vsock-only/shared/autoconf/truncate_pagecache.c 0000444 0000000 0000000 00000003315 13432725347 020727 0 ustar root root /********************************************************* * Copyright (C) 2015-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(3, 12, 0) && \ LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 32) #include <linux/fs.h> #include <linux/mm.h> #include <linux/types.h> /* loff_t */ /* * After 3.12.0, truncate_pagecache changed its interface to just use * the new file size only. Red Hat backported this behavior into a 3.10.0 * kernel. * * This test will fail on a kernel with such a patch. */ void test(void) { struct inode inode; loff_t oldSize = 0; loff_t newSize = 4096; truncate_pagecache(&inode, oldSize, newSize); } #elif LINUX_VERSION_CODE >= KERNEL_VERSION(3, 12, 0) #error "This test intentionally fails on 3.12.0 and newer kernels." #else /* * It must be older than 2.6.32 in which case we assume success. * So not 3.12 compatible. There is no function for these versions. */ #endif vsock-only/shared/autoconf/netcreate_num_params.c 0000444 0000000 0000000 00000003160 13432725347 021314 0 ustar root root /********************************************************* * Copyright (C) 2010 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * During 2.6.33 merge window net_proto_ops->create() method was changed - * a new 'kern' field, signalling whether socket is being created by kernel * or userspace application, was added to it. Unfortunately, some * distributions, such as RHEL 6, have backported the change to earlier * kernels, so we can't rely solely on kernel version to determine number of * arguments. */ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 32) # error This compile test intentionally fails. #else #include <linux/net.h> static int TestCreate(struct net *net, struct socket *sock, int protocol, int kern) { return 0; } struct net_proto_family testFamily = { .create = TestCreate, }; #endif vsock-only/shared/autoconf/filldir1.c 0000444 0000000 0000000 00000003260 13432725347 016627 0 ustar root root /********************************************************* * Copyright (C) 2006 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 20) #include <linux/fs.h> #include <linux/types.h> /* loff_t */ #include <linux/stddef.h> /* NULL */ /* * After 2.6.18, filldir and statfs were changed to send 64-bit inode * numbers to user space. Red Hat backported this behavior into a 2.6.17 * kernel. * * This test will fail on a kernel with such a patch. */ static int LinuxDriverFilldir(void *buf, const char *name, int namelen, loff_t offset, ino_t ino, unsigned int d_type) { return 0; } void test(void) { vfs_readdir(NULL, LinuxDriverFilldir, NULL); } #else #error "This test intentionally fails on 2.6.20 and newer kernels." #endif vsock-only/shared/autoconf/statfs1.c 0000444 0000000 0000000 00000002671 13432725347 016513 0 ustar root root /********************************************************* * Copyright (C) 2006 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 19) #include <linux/fs.h> /* * Around 2.6.18, the super_block pointer in statfs was changed to a dentry * pointer. Red Hat backported this behavior into a 2.6.17 kernel. * * This test will fail on a kernel with such a patch. */ static int LinuxDriverStatFs(struct super_block *sb, struct kstatfs *stat) { return 0; } struct super_operations super_ops = { .statfs = LinuxDriverStatFs }; #else #error "This test intentionally fails on 2.6.19 and newer kernels." #endif vsock-only/shared/autoconf/dcount.c 0000444 0000000 0000000 00000002603 13432725347 016415 0 ustar root root /********************************************************* * Copyright (C) 2014 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(3, 11, 0) #include <linux/dcache.h> /* * After 3.11.0, the dentry d_count field was removed. Red Hat * backported this behavior into a 3.10.0 kernel. * * This test will fail on a kernel with such a patch. */ void test(void) { struct dentry dentry; #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 38) dentry.d_count = 1; #else atomic_set(&dentry.d_count, 1); #endif } #else #error "This test intentionally fails on 3.11.0 or newer kernels." #endif vsock-only/shared/autoconf/dalias1.c 0000444 0000000 0000000 00000003325 13432725347 016441 0 ustar root root /********************************************************* * Copyright (C) 2015-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(3, 6, 0) && \ LINUX_VERSION_CODE >= KERNEL_VERSION(3, 2, 0) #include <linux/dcache.h> #include <linux/list.h> /* * After 3.19.0, the dentry d_alias field was moved. Fedora * backported this behavior into earlier kernels. * The type of the d_alias field changed from 3.6 onwards * which was a list head to being a list node. The check * for 3.6 onwards is done separately. * * This test will fail on a kernel with such a patch. */ void test(void) { struct dentry aliasDentry; INIT_LIST_HEAD(&aliasDentry.d_alias); } #else /* * Intentionally passes for earlier than 3.2.0 kernels as d_alias is valid. * * Intentionally passes for 3.6.0 or later kernels as d_alias is a different type. * A separate test with the different type is run for those kernel versions. */ #endif vsock-only/shared/autoconf/inode1.c 0000444 0000000 0000000 00000002703 13432725347 016301 0 ustar root root /********************************************************* * Copyright (C) 2006 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #include "compat_version.h" #include "compat_autoconf.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 20) #include <linux/fs.h> #include <linux/stddef.h> /* NULL */ /* * After 2.6.18, inodes were "slimmed". This involved removing the union * that encapsulates inode private data (and using i_private instead), as well * as removing i_blksize. Red Hat backported this behavior into a 2.6.17 * kernel. * * This test will fail on a kernel with such a patch. */ void test(void) { struct inode inode; inode.u.generic_ip = NULL; } #else #error "This test intentionally fails on 2.6.20 and newer kernels." #endif vsock-only/shared/compat_pagemap.h 0000444 0000000 0000000 00000002535 13432725347 016271 0 ustar root root /********************************************************* * Copyright (C) 2009 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_PAGEMAP_H__ # define __COMPAT_PAGEMAP_H__ #include <linux/pagemap.h> /* * AOP_FLAG_NOFS was defined in the same changeset that * grab_cache_page_write_begin() was introduced. */ #ifdef AOP_FLAG_NOFS #define compat_grab_cache_page_write_begin(mapping, index, flags) \ grab_cache_page_write_begin((mapping), (index), (flags)) #else #define compat_grab_cache_page_write_begin(mapping, index, flags) \ __grab_cache_page((mapping), (index)); #endif #endif /* __COMPAT_PAGEMAP_H__ */ vsock-only/shared/compat_mutex.h 0000444 0000000 0000000 00000003475 13432725347 016025 0 ustar root root /********************************************************* * Copyright (C) 2009 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_MUTEX_H__ # define __COMPAT_MUTEX_H__ /* Blocking mutexes were introduced in 2.6.16. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 16) #include "compat_semaphore.h" typedef struct semaphore compat_mutex_t; # define compat_define_mutex(_mx) DECLARE_MUTEX(_mx) # define compat_mutex_init(_mx) init_MUTEX(_mx) # define compat_mutex_lock(_mx) down(_mx) # define compat_mutex_lock_interruptible(_mx) down_interruptible(_mx) # define compat_mutex_unlock(_mx) up(_mx) #else #include <linux/mutex.h> typedef struct mutex compat_mutex_t; # define compat_define_mutex(_mx) DEFINE_MUTEX(_mx) # define compat_mutex_init(_mx) mutex_init(_mx) # define compat_mutex_lock(_mx) mutex_lock(_mx) # define compat_mutex_lock_interruptible(_mx) mutex_lock_interruptible(_mx) # define compat_mutex_unlock(_mx) mutex_unlock(_mx) #endif #endif /* __COMPAT_MUTEX_H__ */ vsock-only/shared/compat_string.h 0000444 0000000 0000000 00000003563 13432725347 016167 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_STRING_H__ # define __COMPAT_STRING_H__ #include <linux/string.h> /* * kstrdup was born in 2.6.13. This implementation is almost identical to the * one found there. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 13) #define compat_kstrdup(s, gfp) kstrdup(s, gfp) #else #define compat_kstrdup(s, gfp) \ ({ \ size_t len; \ char *buf; \ len = strlen(s) + 1; \ buf = kmalloc(len, gfp); \ memcpy(buf, s, len); \ buf; \ }) #endif #endif /* __COMPAT_STRING_H__ */ vsock-only/shared/vmciKernelAPI2.h 0000444 0000000 0000000 00000004240 13432725350 016014 0 ustar root root /********************************************************* * Copyright (C) 2010 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmciKernelAPI2.h -- * * Kernel API (v2) exported from the VMCI host and guest drivers. */ #ifndef __VMCI_KERNELAPI_2_H__ #define __VMCI_KERNELAPI_2_H__ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" #include "vmciKernelAPI1.h" #if defined __cplusplus extern "C" { #endif /* Define version 2. */ #undef VMCI_KERNEL_API_VERSION #define VMCI_KERNEL_API_VERSION_2 2 #define VMCI_KERNEL_API_VERSION VMCI_KERNEL_API_VERSION_2 /* VMCI Doorbell API. */ #define VMCI_FLAG_DELAYED_CB 0x01 typedef void (*VMCICallback)(void *clientData); int vmci_doorbell_create(VMCIHandle *handle, uint32 flags, VMCIPrivilegeFlags privFlags, VMCICallback notifyCB, void *clientData); int vmci_doorbell_destroy(VMCIHandle handle); int vmci_doorbell_notify(VMCIHandle handle, VMCIPrivilegeFlags privFlags); /* Typedefs for all of the above, used by the IOCTLs and the kernel library. */ typedef int (VMCIDoorbell_CreateFct)(VMCIHandle *, uint32, VMCIPrivilegeFlags, VMCICallback, void *); typedef int (VMCIDoorbell_DestroyFct)(VMCIHandle); typedef int (VMCIDoorbell_NotifyFct)(VMCIHandle, VMCIPrivilegeFlags); #if defined __cplusplus } // extern "C" #endif #endif /* !__VMCI_KERNELAPI_2_H__ */ vsock-only/shared/vmci_kernel_if.h 0000444 0000000 0000000 00000042051 13432725350 016257 0 ustar root root /********************************************************* * Copyright (C) 2006-2016,2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmci_kernel_if.h -- * * This file defines helper functions for VMCI host _and_ guest * kernel code. It must work for Windows, Mac OS, vmkernel, Linux and * Solaris kernels, i.e. using defines where necessary. */ #ifndef _VMCI_KERNEL_IF_H_ #define _VMCI_KERNEL_IF_H_ #if !defined(__linux__) && !defined(_WIN32) && !defined(__APPLE__) && \ !defined(VMKERNEL) # error "Platform not supported." #endif #if defined(_WIN32) # include <ntddk.h> #else #define UNREFERENCED_PARAMETER(P) #endif #if defined(__linux__) && !defined(VMKERNEL) # include "driver-config.h" # include "compat_cred.h" # include "compat_module.h" # include "compat_semaphore.h" # include "compat_spinlock.h" # include "compat_version.h" # include <linux/wait.h> #endif // linux #ifdef __APPLE__ # include <IOKit/IOLib.h> # include <mach/task.h> # include <mach/semaphore.h> # include <sys/kauth.h> #endif #ifdef VMKERNEL # include "splock.h" # include "splock_customRanks.h" # include "semaphore_ext.h" # include "vmkapi.h" # include "world_dist.h" #endif #include "vm_basic_types.h" #include "vmci_defs.h" #if defined(VMKERNEL) # include "list.h" #else # include "dbllnklst.h" #endif #if defined __cplusplus extern "C" { #endif /* Flags for specifying memory type. */ #define VMCI_MEMORY_NORMAL 0x0 #define VMCI_MEMORY_ATOMIC 0x1 #define VMCI_MEMORY_NONPAGED 0x2 /* Platform specific type definitions. */ #if defined(VMKERNEL) # define VMCI_EXPORT_SYMBOL(_SYMBOL) VMK_MODULE_EXPORT_SYMBOL(_SYMBOL); #elif defined(__linux__) # define VMCI_EXPORT_SYMBOL(_symbol) EXPORT_SYMBOL(_symbol); #elif defined(__APPLE__) # define VMCI_EXPORT_SYMBOL(_symbol) __attribute__((visibility("default"))) #else # define VMCI_EXPORT_SYMBOL(_symbol) #endif #if defined(VMKERNEL) typedef MCSLock VMCILock; typedef SP_IRQL VMCILockFlags; typedef Semaphore VMCIEvent; typedef Semaphore VMCIMutex; typedef World_ID VMCIHostVmID; typedef uint32 VMCIHostUser; typedef PPN64 *VMCIQPGuestMem; #elif defined(__linux__) typedef spinlock_t VMCILock; typedef unsigned long VMCILockFlags; typedef wait_queue_head_t VMCIEvent; typedef struct semaphore VMCIMutex; typedef PPN *VMCIPpnList; /* List of PPNs in produce/consume queue. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 5, 0) typedef kuid_t VMCIHostUser; #else typedef uid_t VMCIHostUser; #endif typedef VA64 VMCIQPGuestMem; #elif defined(__APPLE__) typedef IOLock *VMCILock; typedef unsigned long VMCILockFlags; typedef struct { IOLock *lock; DblLnkLst_Links waiters; int buffered; } VMCIEvent; typedef IOLock *VMCIMutex; typedef void *VMCIPpnList; /* Actually a pointer to the C++ Object IOMemoryDescriptor */ typedef uid_t VMCIHostUser; typedef VA64 *VMCIQPGuestMem; #elif defined(_WIN32) typedef KSPIN_LOCK VMCILock; typedef KIRQL VMCILockFlags; typedef KEVENT VMCIEvent; typedef FAST_MUTEX VMCIMutex; typedef PMDL VMCIPpnList; /* MDL to map the produce/consume queue. */ typedef PSID VMCIHostUser; typedef VA64 *VMCIQPGuestMem; #endif // VMKERNEL /* Callback needed for correctly waiting on events. */ typedef int (*VMCIEventReleaseCB)(void *clientData); /* * Internal locking dependencies within VMCI: * * CONTEXTFIRE < CONTEXT, CONTEXTLIST, EVENT, HASHTABLE * * DOORBELL < HASHTABLE * * QPHIBERNATE < EVENT */ #ifdef VMKERNEL typedef Lock_Rank VMCILockRank; typedef SemaRank VMCISemaRank; #define VMCI_SEMA_RANK_QPHEADER (SEMA_RANK_FS - 1) #define VMCI_LOCK_RANK_MAX_NONBLOCK (MIN(SP_RANK_WAIT, \ SP_RANK_HEAPLOCK_DYNAMIC) - 1) #define VMCI_LOCK_RANK_MAX (SP_RANK_BLOCKABLE_HIGHEST_MAJOR - 2) /* * Determines whether VMCI locks will be blockable or not. If blockable, * all locks will be at or below VMCI_LOCK_RANK_MAX. If not, locks will * instead use VMCI_LOCK_RANK_MAX_NONBLOCK as the maximum. The other * VMCI_LOCK_RANK_XXX values will be rebased to be non-blocking as well * in that case. */ extern Bool vmciBlockableLock; #else typedef unsigned long VMCILockRank; typedef unsigned long VMCISemaRank; #define VMCI_LOCK_RANK_MAX 0x0fff #define VMCI_SEMA_RANK_QPHEADER 0x0fff #endif // VMKERNEL #define VMCI_LOCK_RANK_CONTEXT VMCI_LOCK_RANK_MAX #define VMCI_LOCK_RANK_CONTEXTLIST VMCI_LOCK_RANK_MAX #define VMCI_LOCK_RANK_DATAGRAMVMK VMCI_LOCK_RANK_MAX #define VMCI_LOCK_RANK_EVENT VMCI_LOCK_RANK_MAX #define VMCI_LOCK_RANK_HASHTABLE VMCI_LOCK_RANK_MAX #define VMCI_LOCK_RANK_RESOURCE VMCI_LOCK_RANK_MAX #define VMCI_LOCK_RANK_QPHEADER VMCI_LOCK_RANK_MAX #define VMCI_LOCK_RANK_DOORBELL (VMCI_LOCK_RANK_HASHTABLE - 1) #define VMCI_LOCK_RANK_CONTEXTFIRE (MIN(VMCI_LOCK_RANK_CONTEXT, \ MIN(VMCI_LOCK_RANK_CONTEXTLIST, \ MIN(VMCI_LOCK_RANK_EVENT, \ VMCI_LOCK_RANK_HASHTABLE))) - 1) #define VMCI_LOCK_RANK_QPHIBERNATE (VMCI_LOCK_RANK_EVENT - 1) #define VMCI_LOCK_RANK_PACKET_QP (VMCI_LOCK_RANK_QPHEADER - 1) //#define VMCI_LOCK_RANK_PACKET_QP 0xffd /* For vVol */ #define VMCI_SEMA_RANK_QUEUEPAIRLIST (VMCI_SEMA_RANK_QPHEADER - 1) #define VMCI_SEMA_RANK_GUESTMEM (VMCI_SEMA_RANK_QUEUEPAIRLIST - 1) /* * Host specific struct used for signalling. */ typedef struct VMCIHost { #if defined(VMKERNEL) World_ID vmmWorldID[2]; /* * First one is the active one and the second * one is shadow world during FSR. */ #elif defined(__linux__) wait_queue_head_t waitQueue; #elif defined(__APPLE__) struct Socket *socket; /* vmci Socket object on Mac OS. */ #elif defined(_WIN32) KEVENT *callEvent; /* Ptr to userlevel event used when signalling * new pending guestcalls in kernel. */ #endif } VMCIHost; /* * Guest device port I/O. */ #if defined(__linux__) typedef unsigned short int VMCIIoPort; typedef int VMCIIoHandle; #elif defined(_WIN32) typedef PUCHAR VMCIIoPort; typedef int VMCIIoHandle; #elif defined(__APPLE__) typedef unsigned short int VMCIIoPort; typedef void *VMCIIoHandle; #endif // __APPLE__ void VMCI_ReadPortBytes(VMCIIoHandle handle, VMCIIoPort port, uint8 *buffer, size_t bufferLength); int VMCI_InitLock(VMCILock *lock, char *name, VMCILockRank rank); void VMCI_CleanupLock(VMCILock *lock); void VMCI_GrabLock(VMCILock *lock, VMCILockFlags *flags); void VMCI_ReleaseLock(VMCILock *lock, VMCILockFlags flags); void VMCI_GrabLock_BH(VMCILock *lock, VMCILockFlags *flags); void VMCI_ReleaseLock_BH(VMCILock *lock, VMCILockFlags flags); void VMCIHost_InitContext(VMCIHost *hostContext, uintptr_t eventHnd); void VMCIHost_ReleaseContext(VMCIHost *hostContext); void VMCIHost_SignalCall(VMCIHost *hostContext); void VMCIHost_ClearCall(VMCIHost *hostContext); Bool VMCIHost_WaitForCallLocked(VMCIHost *hostContext, VMCILock *lock, VMCILockFlags *flags, Bool useBH); #ifdef VMKERNEL int VMCIHost_ContextToHostVmID(VMCIHost *hostContext, VMCIHostVmID *hostVmID); int VMCIHost_ContextHasUuid(VMCIHost *hostContext, const char *uuid); void VMCIHost_SetActiveHnd(VMCIHost *hostContext, uintptr_t eventHnd); Bool VMCIHost_RemoveHnd(VMCIHost *hostContext, uintptr_t eventHnd); Bool VMCIHost_IsActiveHnd(VMCIHost *hostContext, uintptr_t eventHnd); void VMCIHost_SetInactiveHnd(VMCIHost *hostContext, uintptr_t eventHnd); uint32 VMCIHost_NumHnds(VMCIHost *hostContext); uintptr_t VMCIHost_GetActiveHnd(VMCIHost *hostContext); void VMCIHost_SignalBitmap(VMCIHost *hostContext); void VMCIHost_SignalBitmapAlways(VMCIHost *hostContext); void VMCIHost_SignalCallAlways(VMCIHost *hostContext); #endif #if defined(_WIN32) /* * On Windows, Driver Verifier will panic() if we leak memory when we are * unloaded. It dumps the leaked blocks for us along with callsites, which * it handily tracks, but if we embed ExAllocate() inside a function, then * the callsite is useless. So make this a macro on this platform only. */ # define VMCI_AllocKernelMem(_sz, _f) \ ExAllocatePoolWithTag((((_f) & VMCI_MEMORY_NONPAGED) ? \ NonPagedPool : PagedPool), \ (_sz), 'MMTC') #else // _WIN32 void *VMCI_AllocKernelMem(size_t size, int flags); #endif // _WIN32 void VMCI_FreeKernelMem(void *ptr, size_t size); int VMCI_CopyToUser(VA64 dst, const void *src, size_t len); Bool VMCIWellKnownID_AllowMap(VMCIId wellKnownID, VMCIPrivilegeFlags privFlags); int VMCIHost_CompareUser(VMCIHostUser *user1, VMCIHostUser *user2); void VMCI_CreateEvent(VMCIEvent *event); void VMCI_DestroyEvent(VMCIEvent *event); void VMCI_SignalEvent(VMCIEvent *event); void VMCI_WaitOnEvent(VMCIEvent *event, VMCIEventReleaseCB releaseCB, void *clientData); #if (defined(__APPLE__) || defined(__linux__) || defined(_WIN32)) && !defined(VMKERNEL) Bool VMCI_WaitOnEventInterruptible(VMCIEvent *event, VMCIEventReleaseCB releaseCB, void *clientData); #endif #if !defined(VMKERNEL) && (defined(__linux__) || defined(_WIN32) || \ defined(__APPLE__)) int VMCI_CopyFromUser(void *dst, VA64 src, size_t len); #endif typedef void (VMCIWorkFn)(void *data); Bool VMCI_CanScheduleDelayedWork(void); int VMCI_ScheduleDelayedWork(VMCIWorkFn *workFn, void *data); int VMCIMutex_Init(VMCIMutex *mutex, char *name, VMCILockRank rank); void VMCIMutex_Destroy(VMCIMutex *mutex); void VMCIMutex_Acquire(VMCIMutex *mutex); void VMCIMutex_Release(VMCIMutex *mutex); #if defined(_WIN32) || defined(__APPLE__) int VMCIKernelIf_Init(void); void VMCIKernelIf_Exit(void); #if defined(_WIN32) void VMCIKernelIf_DrainDelayedWork(void); #endif // _WIN32 #endif // _WIN32 || __APPLE__ #if !defined(VMKERNEL) && \ (defined(__linux__) || defined(_WIN32) || defined(__APPLE__)) void *VMCI_AllocQueue(uint64 size, uint32 flags); void VMCI_FreeQueue(void *q, uint64 size); typedef struct PPNSet { uint64 numProducePages; uint64 numConsumePages; VMCIPpnList producePPNs; VMCIPpnList consumePPNs; Bool initialized; } PPNSet; int VMCI_AllocPPNSet(void *produceQ, uint64 numProducePages, void *consumeQ, uint64 numConsumePages, PPNSet *ppnSet); void VMCI_FreePPNSet(PPNSet *ppnSet); int VMCI_PopulatePPNList(uint8 *callBuf, const PPNSet *ppnSet); #endif struct VMCIQueue; struct PageStoreAttachInfo; struct VMCIQueue *VMCIHost_AllocQueue(uint64 queueSize); void VMCIHost_FreeQueue(struct VMCIQueue *queue, uint64 queueSize); #if defined(VMKERNEL) typedef World_Handle *VMCIGuestMemID; #define INVALID_VMCI_GUEST_MEM_ID NULL #else typedef uint32 VMCIGuestMemID; #define INVALID_VMCI_GUEST_MEM_ID 0 #endif #if defined(VMKERNEL) || defined(__linux__) || defined(_WIN32) || \ defined(__APPLE__) struct QueuePairPageStore; int VMCIHost_RegisterUserMemory(struct QueuePairPageStore *pageStore, struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ); void VMCIHost_UnregisterUserMemory(struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ); int VMCIHost_MapQueues(struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ, uint32 flags); int VMCIHost_UnmapQueues(VMCIGuestMemID gid, struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ); void VMCI_InitQueueMutex(struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ); void VMCI_CleanupQueueMutex(struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ); int VMCI_AcquireQueueMutex(struct VMCIQueue *queue, Bool canBlock); void VMCI_ReleaseQueueMutex(struct VMCIQueue *queue); #else // Below are the guest OS'es without host side support. # define VMCI_InitQueueMutex(_pq, _cq) # define VMCI_CleanupQueueMutex(_pq, _cq) do { } while (0) # define VMCI_AcquireQueueMutex(_q, _cb) VMCI_SUCCESS # define VMCI_ReleaseQueueMutex(_q) do { } while (0) # define VMCIHost_RegisterUserMemory(_ps, _pq, _cq) VMCI_ERROR_UNAVAILABLE # define VMCIHost_UnregisterUserMemory(_pq, _cq) do { } while (0) # define VMCIHost_MapQueues(_pq, _cq, _f) VMCI_SUCCESS # define VMCIHost_UnmapQueues(_gid, _pq, _cq) VMCI_SUCCESS #endif #if defined(VMKERNEL) void VMCIHost_MarkQueuesAvailable(struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ); void VMCIHost_MarkQueuesUnavailable(struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ); int VMCIHost_RevalidateQueues(struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ); #else # define VMCIHost_MarkQueuesAvailable(_q, _p) do { } while (0) # define VMCIHost_MarkQueuesUnavailable(_q, _p) do { } while(0) #endif #if defined(VMKERNEL) || defined(__linux__) void VMCI_LockQueueHeader(struct VMCIQueue *queue); void VMCI_UnlockQueueHeader(struct VMCIQueue *queue); #else # define VMCI_LockQueueHeader(_q) NOT_IMPLEMENTED() # define VMCI_UnlockQueueHeader(_q) NOT_IMPLEMENTED() #endif #if defined(VMKERNEL) int VMCI_QueueHeaderUpdated(struct VMCIQueue *produceQ); #else # define VMCI_QueueHeaderUpdated(_q) VMCI_SUCCESS #endif #if (!defined(VMKERNEL) && defined(__linux__)) || defined(_WIN32) || \ defined(__APPLE__) int VMCIHost_GetUserMemory(VA64 produceUVA, VA64 consumeUVA, struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ); void VMCIHost_ReleaseUserMemory(struct VMCIQueue *produceQ, struct VMCIQueue *consumeQ); #else # define VMCIHost_GetUserMemory(_puva, _cuva, _pq, _cq) VMCI_ERROR_UNAVAILABLE # define VMCIHost_ReleaseUserMemory(_pq, _cq) NOT_IMPLEMENTED() #endif #if defined(_WIN32) Bool VMCI_EnqueueToDevNull(struct VMCIQueue *queue); int VMCI_ConvertToLocalQueue(struct VMCIQueue *queueInfo, struct VMCIQueue *otherQueueInfo, uint64 size, Bool keepContent, void **oldQueue); void VMCI_RevertToNonLocalQueue(struct VMCIQueue *queueInfo, void *nonLocalQueue, uint64 size); void VMCI_FreeQueueBuffer(void *queue, uint64 size); Bool VMCI_CanCreate(void); #else // _WIN32 # define VMCI_EnqueueToDevNull(_q) FALSE # define VMCI_ConvertToLocalQueue(_pq, _cq, _s, _oq, _kc) VMCI_ERROR_UNAVAILABLE # define VMCI_RevertToNonLocalQueue(_q, _nlq, _s) # define VMCI_FreeQueueBuffer(_q, _s) # define VMCI_CanCreate() TRUE #endif // !_WIN32 Bool VMCI_GuestPersonalityActive(void); Bool VMCI_HostPersonalityActive(void); #if defined(VMKERNEL) typedef List_Links VMCIListItem; typedef List_Links VMCIList; # define VMCIList_Init(_l) List_Init(_l) # define VMCIList_InitEntry(_e) List_InitElement(_e) # define VMCIList_Empty(_l) List_IsEmpty(_l) # define VMCIList_Insert(_e, _l) List_Insert(_e, LIST_ATREAR(_l)) # define VMCIList_Remove(_e) List_Remove(_e) # define VMCIList_Scan(_cur, _l) LIST_FORALL(_l, _cur) # define VMCIList_ScanSafe(_cur, _next, _l) LIST_FORALL_SAFE(_l, _cur, _next) # define VMCIList_Entry(_elem, _type, _field) List_Entry(_elem, _type, _field) # define VMCIList_First(_l) (VMCIList_Empty(_l)?NULL:List_First(_l)) #else typedef DblLnkLst_Links VMCIListItem; typedef DblLnkLst_Links VMCIList; # define VMCIList_Init(_l) DblLnkLst_Init(_l) # define VMCIList_InitEntry(_e) DblLnkLst_Init(_e) # define VMCIList_Empty(_l) (!DblLnkLst_IsLinked(_l)) # define VMCIList_Insert(_e, _l) DblLnkLst_LinkLast(_l, _e) # define VMCIList_Remove(_e) DblLnkLst_Unlink1(_e) # define VMCIList_Scan(_cur, _l) DblLnkLst_ForEach(_cur, _l) # define VMCIList_ScanSafe(_cur, _next, _l) DblLnkLst_ForEachSafe(_cur, _next, _l) # define VMCIList_Entry(_elem, _type, _field) DblLnkLst_Container(_elem, _type, _field) # define VMCIList_First(_l) (VMCIList_Empty(_l)?NULL:(_l)->next) #endif #if defined __cplusplus } // extern "C" #endif #endif // _VMCI_KERNEL_IF_H_ vsock-only/shared/vmware_pack_init.h 0000444 0000000 0000000 00000003651 13432725350 016630 0 ustar root root /********************************************************* * Copyright (C) 2002-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __VMWARE_PACK_INIT_H__ # define __VMWARE_PACK_INIT_H__ /* * vmware_pack_init.h -- * * Platform-independent code to make the compiler pack (i.e. have them * occupy the smallest possible space) structure definitions. The following * constructs are known to work --hpreg * * #include "vmware_pack_begin.h" * struct foo { * ... * } * #include "vmware_pack_end.h" * ; * * typedef * #include "vmware_pack_begin.h" * struct foo { * ... * } * #include "vmware_pack_end.h" * foo; */ #ifdef _MSC_VER /* * MSVC 6.0 emits warning 4103 when the pack push and pop pragma pairing is * not balanced within 1 included file. That is annoying because our scheme * is based on the pairing being balanced between 2 included files. * * So we disable this warning, but this is safe because the compiler will also * emit warning 4161 when there is more pops than pushes within 1 main * file --hpreg */ # pragma warning(disable:4103) #elif __GNUC__ #else # error Compiler packing... #endif #endif /* __VMWARE_PACK_INIT_H__ */ vsock-only/shared/vm_basic_defs.h 0000444 0000000 0000000 00000044517 13432725350 016100 0 ustar root root /********************************************************* * Copyright (C) 2003-2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vm_basic_defs.h -- * * Standard macros for VMware source code. */ #ifndef _VM_BASIC_DEFS_H_ #define _VM_BASIC_DEFS_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_VMKDRIVERS #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" #include "vm_basic_types.h" // For INLINE. /* Checks for FreeBSD, filtering out VMKERNEL. */ #if !defined(VMKERNEL) && defined(__FreeBSD__) #define __IS_FREEBSD__ 1 #else #define __IS_FREEBSD__ 0 #endif #define __IS_FREEBSD_VER__(ver) (__IS_FREEBSD__ && __FreeBSD_version >= (ver)) #if defined _WIN32 && defined USERLEVEL #include <stddef.h> /* * We redefine offsetof macro from stddef; make * sure that it's already defined before we do that. */ #include <windows.h> // for Sleep() and LOWORD() etc. #undef GetFreeSpace // Unpollute preprocessor namespace. #endif /* * Simple macros */ #ifndef vmw_offsetof #define vmw_offsetof(TYPE, MEMBER) ((size_t) &((TYPE *)0)->MEMBER) #endif #if (defined __APPLE__ || defined __FreeBSD__) && \ (!defined KERNEL && !defined _KERNEL && !defined VMKERNEL && !defined __KERNEL__) # include <stddef.h> #else #ifndef offsetof #define VMW_DEFINED_OFFSETOF /* * XXX While the _WIN32 implementation appears to be identical to vmw_offsetof * in terms of behavior, they need to be separate to match verbatim the * definition used by the respective compilers, to avoid a redefinition warning. * * This is necessary until we eliminate the inclusion of <windows.h> above. */ #ifdef _WIN32 #define offsetof(s,m) (size_t)&(((s *)0)->m) /* * We use the builtin offset for gcc/clang, except when we're running under the * vmkernel's GDB macro preprocessor, since gdb doesn't understand * __builtin_offsetof. */ #elif defined __GNUC__ && !defined VMKERNEL_GDB_MACRO_BUILDER #define offsetof __builtin_offsetof #else #define offsetof vmw_offsetof #endif #endif // offsetof #endif // __APPLE__ #define VMW_CONTAINER_OF(ptr, type, member) \ ((type *)((char *)(ptr) - vmw_offsetof(type, member))) #ifndef ARRAYSIZE #define ARRAYSIZE(a) (sizeof (a) / sizeof *(a)) #endif #ifndef MIN #define MIN(_a, _b) (((_a) < (_b)) ? (_a) : (_b)) #endif /* The Solaris 9 cross-compiler complains about these not being used */ #ifndef sun static INLINE int Min(int a, int b) { return a < b ? a : b; } #endif #ifndef MAX #define MAX(_a, _b) (((_a) > (_b)) ? (_a) : (_b)) #endif #ifndef sun static INLINE int Max(int a, int b) { return a > b ? a : b; } #endif #define VMW_CLAMP(x, min, max) \ ((x) < (min) ? (min) : ((x) > (max) ? (max) : (x))) #define ROUNDUP(x,y) (((x) + (y) - 1) / (y) * (y)) #define ROUNDDOWN(x,y) ((x) / (y) * (y)) #define ROUNDUPBITS(x, bits) (((uintptr_t) (x) + MASK(bits)) & ~MASK(bits)) #define ROUNDDOWNBITS(x, bits) ((uintptr_t) (x) & ~MASK(bits)) #define CEILING(x, y) (((x) + (y) - 1) / (y)) #if defined __APPLE__ #include <machine/param.h> #undef MASK #endif /* * The MASK macro behaves badly when given negative numbers or numbers larger * than the highest order bit number (e.g. 32 on a 32-bit machine) as an * argument. The range 0..31 is safe. */ #define MASK(n) ((1 << (n)) - 1) /* make an n-bit mask */ #define MASK64(n) ((CONST64U(1) << (n)) - 1) /* make an n-bit mask */ /* * MASKRANGE64 makes a bit vector starting at bit lo and ending at bit hi. No * checking for lo < hi is done. */ #define MASKRANGE64(hi, lo) (MASK64((hi) - (lo) + 1) << (lo)) /* SIGNEXT64 sign extends a n-bit value to 64-bits. */ #define SIGNEXT64(val, n) (((int64)(val) << (64 - (n))) >> (64 - (n))) #define DWORD_ALIGN(x) ((((x) + 3) >> 2) << 2) #define QWORD_ALIGN(x) ((((x) + 7) >> 3) << 3) #define IMPLIES(a,b) (!(a) || (b)) /* * Not everybody (e.g., the monitor) has NULL */ #ifndef NULL #ifdef __cplusplus #define NULL 0 #else #define NULL ((void *)0) #endif #endif /* * Token concatenation * * The C preprocessor doesn't prescan arguments when they are * concatenated or stringified. So we need extra levels of * indirection to convince the preprocessor to expand its * arguments. */ #define CONC(x, y) x##y #define XCONC(x, y) CONC(x, y) #define XXCONC(x, y) XCONC(x, y) #define MAKESTR(x) #x #define XSTR(x) MAKESTR(x) /* * Wide versions of string constants. */ #ifndef WSTR #define WSTR_(X) L ## X #define WSTR(X) WSTR_(X) #endif /* * Page operations * * It has been suggested that these definitions belong elsewhere * (like x86types.h). However, I deem them common enough * (since even regular user-level programs may want to do * page-based memory manipulation) to be here. * -- edward */ #ifndef PAGE_SHIFT // { #if defined VM_I386 #define PAGE_SHIFT 12 #elif defined __APPLE__ #define PAGE_SHIFT 12 #elif defined VM_ARM_64 #define PAGE_SHIFT 12 #elif defined __arm__ #define PAGE_SHIFT 12 #else #error #endif #endif // } #ifndef PAGE_SIZE #define PAGE_SIZE (1 << PAGE_SHIFT) #endif #ifndef PAGE_MASK #define PAGE_MASK (PAGE_SIZE - 1) #endif #ifndef PAGE_OFFSET #define PAGE_OFFSET(_addr) ((uintptr_t)(_addr) & (PAGE_SIZE - 1)) #endif #ifndef PAGE_NUMBER #define PAGE_NUMBER(_addr) ((uintptr_t)(_addr) / PAGE_SIZE) #endif #ifndef VM_PAGE_BASE #define VM_PAGE_BASE(_addr) ((_addr) & ~(PAGE_SIZE - 1)) #endif #ifndef VM_PAGES_SPANNED #define VM_PAGES_SPANNED(_addr, _size) \ ((((_addr) & (PAGE_SIZE - 1)) + (_size) + (PAGE_SIZE - 1)) >> PAGE_SHIFT) #endif #ifndef BYTES_2_PAGES #define BYTES_2_PAGES(_nbytes) ((_nbytes) >> PAGE_SHIFT) #endif #ifndef PAGES_2_BYTES #define PAGES_2_BYTES(_npages) (((uint64)(_npages)) << PAGE_SHIFT) #endif #ifndef MBYTES_SHIFT #define MBYTES_SHIFT 20 #endif #ifndef MBYTES_2_PAGES #define MBYTES_2_PAGES(_nbytes) \ ((uint64)(_nbytes) << (MBYTES_SHIFT - PAGE_SHIFT)) #endif #ifndef PAGES_2_MBYTES #define PAGES_2_MBYTES(_npages) ((_npages) >> (MBYTES_SHIFT - PAGE_SHIFT)) #endif #ifndef ROUNDUP_PAGES_2_MBYTES #define ROUNDUP_PAGES_2_MBYTES(_npages) \ (((_npages) + MASK(MBYTES_SHIFT - PAGE_SHIFT)) >> (MBYTES_SHIFT - PAGE_SHIFT)) #endif #ifndef ROUNDDOWN_PAGES_2_MBYTES #define ROUNDDOWN_PAGES_2_MBYTES(_npages) \ ((_npages) >> (MBYTES_SHIFT - PAGE_SHIFT)) #endif #ifndef GBYTES_2_PAGES #define GBYTES_2_PAGES(_nbytes) ((uint64)(_nbytes) << (30 - PAGE_SHIFT)) #endif #ifndef PAGES_2_GBYTES #define PAGES_2_GBYTES(_npages) ((_npages) >> (30 - PAGE_SHIFT)) #endif #ifndef BYTES_2_MBYTES #define BYTES_2_MBYTES(_nbytes) ((_nbytes) >> MBYTES_SHIFT) #endif #ifndef MBYTES_2_BYTES #define MBYTES_2_BYTES(_nbytes) ((uint64)(_nbytes) << MBYTES_SHIFT) #endif #ifndef BYTES_2_GBYTES #define BYTES_2_GBYTES(_nbytes) ((_nbytes) >> 30) #endif #ifndef GBYTES_2_BYTES #define GBYTES_2_BYTES(_nbytes) ((uint64)(_nbytes) << 30) #endif #ifndef VM_PAE_LARGE_PAGE_SHIFT #define VM_PAE_LARGE_PAGE_SHIFT 21 #endif #ifndef VM_PAE_LARGE_PAGE_SIZE #define VM_PAE_LARGE_PAGE_SIZE (1 << VM_PAE_LARGE_PAGE_SHIFT) #endif #ifndef VM_PAE_LARGE_PAGE_MASK #define VM_PAE_LARGE_PAGE_MASK (VM_PAE_LARGE_PAGE_SIZE - 1) #endif #ifndef VM_PAE_LARGE_2_SMALL_PAGES #define VM_PAE_LARGE_2_SMALL_PAGES (BYTES_2_PAGES(VM_PAE_LARGE_PAGE_SIZE)) #endif #ifndef VM_PAE_LARGE_2_BYTES #define VM_PAE_LARGE_2_BYTES(_2mbytes) ((_2mbytes) << VM_PAE_LARGE_PAGE_SHIFT) #endif #ifndef VM_1GB_PAGE_SHIFT #define VM_1GB_PAGE_SHIFT 30 #endif #ifndef VM_1GB_PAGE_SIZE #define VM_1GB_PAGE_SIZE (1 << VM_1GB_PAGE_SHIFT) #endif #ifndef VM_1GB_2_PAGES #define VM_1GB_2_PAGES (BYTES_2_PAGES(VM_1GB_PAGE_SIZE)) #endif #ifndef VM_1GB_2_PDIRS #define VM_1GB_2_PDIRS (VM_1GB_PAGE_SIZE / VM_PAE_LARGE_PAGE_SIZE) #endif /* * Word operations */ #ifndef LOWORD #define LOWORD(_dw) ((_dw) & 0xffff) #endif #ifndef HIWORD #define HIWORD(_dw) (((_dw) >> 16) & 0xffff) #endif #ifndef LOBYTE #define LOBYTE(_w) ((_w) & 0xff) #endif #ifndef HIBYTE #define HIBYTE(_w) (((_w) >> 8) & 0xff) #endif #ifndef HIDWORD #define HIDWORD(_qw) ((uint32)((_qw) >> 32)) #endif #ifndef LODWORD #define LODWORD(_qw) ((uint32)(_qw)) #endif #define QWORD(_hi, _lo) ((((uint64)(_hi)) << 32) | ((uint32)(_lo))) /* * Deposit a field _src at _pos bits from the right, * with a length of _len, into the integer _target. */ #define DEPOSIT_BITS(_src,_pos,_len,_target) { \ unsigned mask = ((1 << _len) - 1); \ unsigned shiftedmask = ((1 << _len) - 1) << _pos; \ _target = (_target & ~shiftedmask) | ((_src & mask) << _pos); \ } /* * Get return address. */ #ifdef _MSC_VER #ifdef __cplusplus extern "C" #endif void *_ReturnAddress(void); #pragma intrinsic(_ReturnAddress) #define GetReturnAddress() _ReturnAddress() #elif __GNUC__ #define GetReturnAddress() __builtin_return_address(0) #endif #ifdef __GNUC__ #ifndef sun /* * A bug in __builtin_frame_address was discovered in gcc 4.1.1, and * fixed in 4.2.0; assume it originated in 4.0. PR 147638 and 554369. */ #if !(__GNUC__ == 4 && (__GNUC_MINOR__ == 0 || __GNUC_MINOR__ == 1)) #define GetFrameAddr() __builtin_frame_address(0) #endif #endif // sun #endif // __GNUC__ /* * Data prefetch was added in gcc 3.1.1 * http://www.gnu.org/software/gcc/gcc-3.1/changes.html */ #ifdef __GNUC__ # if ((__GNUC__ > 3) || (__GNUC__ == 3 && __GNUC_MINOR__ > 1) || \ (__GNUC__ == 3 && __GNUC_MINOR__ == 1 && __GNUC_PATCHLEVEL__ >= 1)) # define PREFETCH_R(var) __builtin_prefetch((var), 0 /* read */, \ 3 /* high temporal locality */) # define PREFETCH_W(var) __builtin_prefetch((var), 1 /* write */, \ 3 /* high temporal locality */) # else # define PREFETCH_R(var) ((void)(var)) # define PREFETCH_W(var) ((void)(var)) # endif #endif /* __GNUC__ */ #ifdef USERLEVEL // { /* * Standardize some Posix names on Windows. */ #ifdef _WIN32 // { /* Conflict with definition of Visual Studio 2015 */ #if (_MSC_VER < 1900) #define snprintf _snprintf #endif #define strtok_r strtok_s #if (_MSC_VER < 1500) #define vsnprintf _vsnprintf #endif typedef int uid_t; typedef int gid_t; static INLINE void sleep(unsigned int sec) { Sleep(sec * 1000); } static INLINE int usleep(unsigned long usec) { Sleep(CEILING(usec, 1000)); return 0; } typedef int pid_t; #define F_OK 0 #define X_OK 1 #define W_OK 2 #define R_OK 4 #endif // } /* * Macro for username comparison. */ #ifdef _WIN32 // { #define USERCMP(x,y) Str_Strcasecmp(x,y) #else #define USERCMP(x,y) strcmp(x,y) #endif // } #endif // } #ifndef va_copy #ifdef _WIN32 /* * Windows needs va_copy. This works for both 32 and 64-bit Windows * based on inspection of how varags.h from the Visual C CRTL is * implemented. (Future versions of the RTL may break this). */ #define va_copy(dest, src) ((dest) = (src)) #elif defined(__APPLE__) && defined(KERNEL) // The macOS kernel SDK defines va_copy in stdarg.h. #include <stdarg.h> #elif defined(__GNUC__) && (__GNUC__ < 3) /* * Old versions of gcc recognize __va_copy, but not va_copy. */ #define va_copy(dest, src) __va_copy(dest, src) #endif // _WIN32 #endif // va_copy /* * This one is outside USERLEVEL because it's used by * files compiled into the Windows hgfs driver or the display * driver. */ #if defined(_WIN32) && defined(_MSC_VER) #define PATH_MAX 256 #ifndef strcasecmp #define strcasecmp(_s1,_s2) _stricmp((_s1),(_s2)) #endif #ifndef strncasecmp #define strncasecmp(_s1,_s2,_n) _strnicmp((_s1),(_s2),(_n)) #endif #endif /* * Convenience definitions of unicode characters. */ #ifndef UTF8_ELLIPSIS #define UTF8_ELLIPSIS "\xe2\x80\xa6" #endif /* * Convenience macros and definitions. Can often be used instead of #ifdef. */ #undef ARM64_ONLY #ifdef VM_ARM_64 #define ARM64_ONLY(x) x #else #define ARM64_ONLY(x) #endif #undef X86_ONLY #ifdef VM_X86_ANY #define X86_ONLY(x) x #else #define X86_ONLY(x) #endif #undef DEBUG_ONLY #ifdef VMX86_DEBUG #define vmx86_debug 1 #define DEBUG_ONLY(x) x #else #define vmx86_debug 0 #define DEBUG_ONLY(x) #endif #ifdef VMX86_STATS #define vmx86_stats 1 #define STATS_ONLY(x) x #else #define vmx86_stats 0 #define STATS_ONLY(x) #endif #ifdef VMX86_DEVEL #define vmx86_devel 1 #define DEVEL_ONLY(x) x #else #define vmx86_devel 0 #define DEVEL_ONLY(x) #endif #ifdef VMX86_LOG #define vmx86_log 1 #define LOG_ONLY(x) x #else #define vmx86_log 0 #define LOG_ONLY(x) #endif #ifdef VMX86_BETA #define vmx86_beta 1 #define BETA_ONLY(x) x #else #define vmx86_beta 0 #define BETA_ONLY(x) #endif #ifdef VMX86_RELEASE #define vmx86_release 1 #define RELEASE_ONLY(x) x #else #define vmx86_release 0 #define RELEASE_ONLY(x) #endif #ifdef VMX86_SERVER #define vmx86_server 1 #define SERVER_ONLY(x) x #define HOSTED_ONLY(x) #else #define vmx86_server 0 #define SERVER_ONLY(x) #define HOSTED_ONLY(x) x #endif #ifdef VMKERNEL #define vmkernel 1 #define VMKERNEL_ONLY(x) x #else #define vmkernel 0 #define VMKERNEL_ONLY(x) #endif #ifdef _WIN32 #define WIN32_ONLY(x) x #define POSIX_ONLY(x) #define vmx86_win32 1 #else #define WIN32_ONLY(x) #define POSIX_ONLY(x) x #define vmx86_win32 0 #endif #ifdef __linux__ #define vmx86_linux 1 #define LINUX_ONLY(x) x #else #define vmx86_linux 0 #define LINUX_ONLY(x) #endif #ifdef __APPLE__ #define vmx86_apple 1 #define APPLE_ONLY(x) x #else #define vmx86_apple 0 #define APPLE_ONLY(x) #endif #if defined(__APPLE__) && defined(VMW_APPLE_SANDBOX) #define vmw_apple_sandbox 1 #else #define vmw_apple_sandbox 0 #endif #if defined(__APPLE__) && defined(VMW_APPLE_APP_STORE) #define vmw_apple_app_store 1 #else #define vmw_apple_app_store 0 #endif #ifdef VMM #define VMM_ONLY(x) x #else #define VMM_ONLY(x) #endif #if defined(VMM) || defined(VMKERNEL) #define USER_ONLY(x) #else #define USER_ONLY(x) x #endif #ifdef _WIN32 #define VMW_INVALID_HANDLE INVALID_HANDLE_VALUE #else #define VMW_INVALID_HANDLE (-1LL) #endif #ifdef _WIN32 #define fsync(fd) _commit(fd) #define fileno(f) _fileno(f) #else #endif /* * Debug output macros for Windows drivers (the Eng variant is for * display/printer drivers only. */ #ifdef _WIN32 #ifndef USES_OLD_WINDDK #if defined(VMX86_LOG) #ifdef _WIN64 #define WinDrvPrint(arg, ...) DbgPrintEx(DPFLTR_IHVDRIVER_ID, (ULONG)~0, arg, __VA_ARGS__) #else #define WinDrvPrint(arg, ...) DbgPrint(arg, __VA_ARGS__) #endif #define WinDrvEngPrint(arg, ...) EngDbgPrint(arg, __VA_ARGS__) #else #define WinDrvPrint(arg, ...) #define WinDrvEngPrint(arg, ...) #endif #endif #endif // _WIN32 #ifdef HOSTED_LG_PG #define hosted_lg_pg 1 #else #define hosted_lg_pg 0 #endif /* * Use to initialize cbSize for this structure to preserve < Vista * compatibility. */ #define NONCLIENTMETRICSINFO_V1_SIZE CCSIZEOF_STRUCT(NONCLIENTMETRICS, \ lfMessageFont) /* This is not intended to be thread-safe. */ #define DO_ONCE(code) \ do { \ static Bool _doOnceDone = FALSE; \ if (UNLIKELY(!_doOnceDone)) { \ _doOnceDone = TRUE; \ code; \ } \ } while (0) /* * Bug 827422 and 838523. */ #if defined __GNUC__ && __GNUC__ >= 4 #define VISIBILITY_HIDDEN __attribute__((visibility("hidden"))) #else #define VISIBILITY_HIDDEN /* nothing */ #endif /* * Bitfield extraction. */ #define EXTRACT_BITSLICE32(_val , _lsb, _msb) \ (((uint32)(_val) << (31 - (_msb))) >> ((31 - (_msb)) + (_lsb))) #define EXTRACT_BITFIELD32(_val, _pos, _len) \ EXTRACT_BITSLICE32((_val), (_pos), ((_pos) + (_len) - 1)) #define EXTRACT_BITSLICE64(_val, _lsb, _msb) \ (((uint64)(_val) << (63 - (_msb))) >> ((63 - (_msb)) + (_lsb))) #define EXTRACT_BITFIELD64(_val, _pos, _len) \ EXTRACT_BITSLICE64((_val), (_pos), ((_pos) + (_len) - 1)) /* * Typical cache line size. Use this for aligning structures to cache * lines for performance, but do not rely on it for correctness. * * On x86, all current processors newer than P4 have 64-byte lines, * but P4 had 128. * * On ARM, the line size can vary between cores. 64-byte lines are * common, but either larger or smaller powers of two are possible. */ #define CACHELINE_SIZE 64 #define CACHELINE_SHIFT 6 #define CACHELINE_ALIGNMENT_MASK (CACHELINE_SIZE - 1) /* * Bits to bytes sizes. */ #define SIZE_8BIT 1 #define SIZE_16BIT 2 #define SIZE_24BIT 3 #define SIZE_32BIT 4 #define SIZE_48BIT 6 #define SIZE_64BIT 8 #define SIZE_80BIT 10 #define SIZE_128BIT 16 #define SIZE_256BIT 32 #define SIZE_512BIT 64 /* * Allocate a variable of type _type, aligned to _align bytes, returning a * pointer to the variable in _var. Potentially _align - 1 bytes may be * wasted. On x86, GCC 6.3.0 behaves sub-optimally when variables are declared * on the stack using the aligned attribute, so this pattern is preferred. * See PRs 1795155, 1819963. */ #define WITH_PTR_TO_ALIGNED_VAR(_type, _align, _var) \ do { \ uint8 _buf_##_var[sizeof(_type) + (_align) - 1]; \ _type *_var = (_type *) ((uintptr_t)(_buf_##_var + (_align) - 1) & \ ~((uintptr_t) ((_align) - 1))); #define END_PTR_TO_ALIGNED_VAR \ } while (0) #endif // ifndef _VM_BASIC_DEFS_H_ vsock-only/shared/backdoor_def.h 0000444 0000000 0000000 00000033730 13432725350 015711 0 ustar root root /********************************************************* * Copyright (C) 1998-2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * backdoor_def.h -- * * This contains backdoor defines that can be included from * an assembly language file. */ #ifndef _BACKDOOR_DEF_H_ #define _BACKDOOR_DEF_H_ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" #if defined __cplusplus extern "C" { #endif /* * If you want to add a new low-level backdoor call for a guest userland * application, please consider using the GuestRpc mechanism instead. --hpreg */ #define BDOOR_MAGIC 0x564D5868 /* Low-bandwidth backdoor port. --hpreg */ #define BDOOR_PORT 0x5658 #define BDOOR_CMD_GETMHZ 1 /* * BDOOR_CMD_APMFUNCTION is used by: * * o The FrobOS code, which instead should either program the virtual chipset * (like the new BIOS code does, matthias offered to implement that), or not * use any VM-specific code (which requires that we correctly implement * "power off on CLI HLT" for SMP VMs, boris offered to implement that) * * o The old BIOS code, which will soon be jettisoned * * --hpreg */ #define BDOOR_CMD_APMFUNCTION 2 /* CPL0 only. */ #define BDOOR_CMD_GETDISKGEO 3 #define BDOOR_CMD_GETPTRLOCATION 4 #define BDOOR_CMD_SETPTRLOCATION 5 #define BDOOR_CMD_GETSELLENGTH 6 #define BDOOR_CMD_GETNEXTPIECE 7 #define BDOOR_CMD_SETSELLENGTH 8 #define BDOOR_CMD_SETNEXTPIECE 9 #define BDOOR_CMD_GETVERSION 10 #define BDOOR_CMD_GETDEVICELISTELEMENT 11 #define BDOOR_CMD_TOGGLEDEVICE 12 #define BDOOR_CMD_GETGUIOPTIONS 13 #define BDOOR_CMD_SETGUIOPTIONS 14 #define BDOOR_CMD_GETSCREENSIZE 15 #define BDOOR_CMD_MONITOR_CONTROL 16 /* Disabled by default. */ #define BDOOR_CMD_GETHWVERSION 17 #define BDOOR_CMD_OSNOTFOUND 18 /* CPL0 only. */ #define BDOOR_CMD_GETUUID 19 #define BDOOR_CMD_GETMEMSIZE 20 //#define BDOOR_CMD_HOSTCOPY 21 /* Not in use. Was devel only. */ //#define BDOOR_CMD_SERVICE_VM 22 /* Not in use. Never shipped. */ #define BDOOR_CMD_GETTIME 23 /* Deprecated -> GETTIMEFULL. */ #define BDOOR_CMD_STOPCATCHUP 24 #define BDOOR_CMD_PUTCHR 25 /* Disabled by default. */ #define BDOOR_CMD_ENABLE_MSG 26 /* Devel only. */ //#define BDOOR_CMD_GOTO_TCL 27 /* Not in use. Was devel only */ #define BDOOR_CMD_INITPCIOPROM 28 /* CPL 0 only. */ //#define BDOOR_CMD_INT13 29 /* Not in use. */ #define BDOOR_CMD_MESSAGE 30 #define BDOOR_CMD_SIDT 31 #define BDOOR_CMD_SGDT 32 #define BDOOR_CMD_SLDT_STR 33 #define BDOOR_CMD_ISACPIDISABLED 34 //#define BDOOR_CMD_TOE 35 /* Not in use. */ #define BDOOR_CMD_ISMOUSEABSOLUTE 36 #define BDOOR_CMD_PATCH_SMBIOS_STRUCTS 37 /* CPL 0 only. */ #define BDOOR_CMD_MAPMEM 38 /* Devel only */ #define BDOOR_CMD_ABSPOINTER_DATA 39 #define BDOOR_CMD_ABSPOINTER_STATUS 40 #define BDOOR_CMD_ABSPOINTER_COMMAND 41 //#define BDOOR_CMD_TIMER_SPONGE 42 /* Not in use. */ #define BDOOR_CMD_PATCH_ACPI_TABLES 43 /* CPL 0 only. */ //#define BDOOR_CMD_DEVEL_FAKEHARDWARE 44 /* Not in use. */ #define BDOOR_CMD_GETHZ 45 #define BDOOR_CMD_GETTIMEFULL 46 //#define BDOOR_CMD_STATELOGGER 47 /* Not in use. */ #define BDOOR_CMD_CHECKFORCEBIOSSETUP 48 /* CPL 0 only. */ #define BDOOR_CMD_LAZYTIMEREMULATION 49 /* CPL 0 only. */ #define BDOOR_CMD_BIOSBBS 50 /* CPL 0 only. */ //#define BDOOR_CMD_VASSERT 51 /* Not in use. */ #define BDOOR_CMD_ISGOSDARWIN 52 #define BDOOR_CMD_DEBUGEVENT 53 #define BDOOR_CMD_OSNOTMACOSXSERVER 54 /* CPL 0 only. */ #define BDOOR_CMD_GETTIMEFULL_WITH_LAG 55 #define BDOOR_CMD_ACPI_HOTPLUG_DEVICE 56 /* Devel only. */ #define BDOOR_CMD_ACPI_HOTPLUG_MEMORY 57 /* Devel only. */ #define BDOOR_CMD_ACPI_HOTPLUG_CBRET 58 /* Devel only. */ //#define BDOOR_CMD_GET_HOST_VIDEO_MODES 59 /* Not in use. */ #define BDOOR_CMD_ACPI_HOTPLUG_CPU 60 /* Devel only. */ //#define BDOOR_CMD_USB_HOTPLUG_MOUSE 61 /* Not in use. Never shipped. */ #define BDOOR_CMD_XPMODE 62 /* CPL 0 only. */ #define BDOOR_CMD_NESTING_CONTROL 63 #define BDOOR_CMD_FIRMWARE_INIT 64 /* CPL 0 only. */ #define BDOOR_CMD_FIRMWARE_ACPI_SERVICES 65 /* CPL 0 only. */ # define BDOOR_CMD_FAS_GET_TABLE_SIZE 0 # define BDOOR_CMD_FAS_GET_TABLE_DATA 1 # define BDOOR_CMD_FAS_GET_PLATFORM_NAME 2 # define BDOOR_CMD_FAS_GET_PCIE_OSC_MASK 3 # define BDOOR_CMD_FAS_GET_APIC_ROUTING 4 # define BDOOR_CMD_FAS_GET_TABLE_SKIP 5 # define BDOOR_CMD_FAS_GET_SLEEP_ENABLES 6 # define BDOOR_CMD_FAS_GET_HARD_RESET_ENABLE 7 # define BDOOR_CMD_FAS_GET_MOUSE_HID 8 # define BDOOR_CMD_FAS_GET_SMBIOS_VERSION 9 # define BDOOR_CMD_FAS_GET_64BIT_PCI_HOLE_SIZE 10 //#define BDOOR_CMD_FAS_GET_NVDIMM_FMT_CODE 11 /* Not in use. Never shipped. */ # define BDOOR_CMD_FAS_SRP_ENABLED 12 # define BDOOR_CMD_FAS_EXIT_BOOT_SERVICES 13 #define BDOOR_CMD_SENDPSHAREHINTS 66 /* Not in use. Deprecated. */ #define BDOOR_CMD_ENABLE_USB_MOUSE 67 #define BDOOR_CMD_GET_VCPU_INFO 68 # define BDOOR_CMD_VCPU_SLC64 0 # define BDOOR_CMD_VCPU_SYNC_VTSCS 1 # define BDOOR_CMD_VCPU_HV_REPLAY_OK 2 # define BDOOR_CMD_VCPU_LEGACY_X2APIC_OK 3 # define BDOOR_CMD_VCPU_MMIO_HONORS_PAT 4 # define BDOOR_CMD_VCPU_RESERVED 31 #define BDOOR_CMD_EFI_SERIALCON_CONFIG 69 /* CPL 0 only. */ #define BDOOR_CMD_BUG328986 70 /* CPL 0 only. */ #define BDOOR_CMD_FIRMWARE_ERROR 71 /* CPL 0 only. */ # define BDOOR_CMD_FE_INSUFFICIENT_MEM 0 # define BDOOR_CMD_FE_EXCEPTION 1 # define BDOOR_CMD_FE_SGX 2 # define BDOOR_CMD_FE_PCI_MMIO 3 #define BDOOR_CMD_VMK_INFO 72 #define BDOOR_CMD_EFI_BOOT_CONFIG 73 /* CPL 0 only. */ # define BDOOR_CMD_EBC_LEGACYBOOT_ENABLED 0 # define BDOOR_CMD_EBC_GET_ORDER 1 # define BDOOR_CMD_EBC_SHELL_ACTIVE 2 # define BDOOR_CMD_EBC_GET_NETWORK_BOOT_PROTOCOL 3 # define BDOOR_CMD_EBC_QUICKBOOT_ENABLED 4 # define BDOOR_CMD_EBC_GET_PXE_ARCH 5 #define BDOOR_CMD_GET_HW_MODEL 74 /* CPL 0 only. */ #define BDOOR_CMD_GET_SVGA_CAPABILITIES 75 /* CPL 0 only. */ #define BDOOR_CMD_GET_FORCE_X2APIC 76 /* CPL 0 only */ #define BDOOR_CMD_SET_PCI_HOLE 77 /* CPL 0 only */ #define BDOOR_CMD_GET_PCI_HOLE 78 /* CPL 0 only */ #define BDOOR_CMD_GET_PCI_BAR 79 /* CPL 0 only */ #define BDOOR_CMD_SHOULD_GENERATE_SYSTEMID 80 /* CPL 0 only */ #define BDOOR_CMD_READ_DEBUG_FILE 81 /* Devel only. */ #define BDOOR_CMD_SCREENSHOT 82 /* Devel only. */ #define BDOOR_CMD_INJECT_KEY 83 /* Devel only. */ #define BDOOR_CMD_INJECT_MOUSE 84 /* Devel only. */ #define BDOOR_CMD_MKS_GUEST_STATS 85 /* CPL 0 only. */ # define BDOOR_CMD_MKSGS_RESET 0 # define BDOOR_CMD_MKSGS_ADD_PPN 1 # define BDOOR_CMD_MKSGS_REMOVE_PPN 2 #define BDOOR_CMD_ABSPOINTER_RESTRICT 86 #define BDOOR_CMD_GUEST_INTEGRITY 87 # define BDOOR_CMD_GI_GET_CAPABILITIES 0 # define BDOOR_CMD_GI_SETUP_ENTRY_POINT 1 # define BDOOR_CMD_GI_SETUP_ALERTS 2 # define BDOOR_CMD_GI_SETUP_STORE 3 # define BDOOR_CMD_GI_SETUP_EVENT_RING 4 # define BDOOR_CMD_GI_SETUP_NON_FAULT_READ 5 # define BDOOR_CMD_GI_ENTER_INTEGRITY_MODE 6 # define BDOOR_CMD_GI_EXIT_INTEGRITY_MODE 7 # define BDOOR_CMD_GI_RESET_INTEGRITY_MODE 8 # define BDOOR_CMD_GI_GET_EVENT_RING_STATE 9 # define BDOOR_CMD_GI_CONSUME_RING_EVENTS 10 # define BDOOR_CMD_GI_WATCH_MAPPINGS_START 11 # define BDOOR_CMD_GI_WATCH_MAPPINGS_STOP 12 # define BDOOR_CMD_GI_CHECK_MAPPINGS_NOW 13 # define BDOOR_CMD_GI_WATCH_PPNS_START 14 # define BDOOR_CMD_GI_WATCH_PPNS_STOP 15 # define BDOOR_CMD_GI_SEND_MSG 16 # define BDOOR_CMD_GI_TEST_READ_MOB 128 # define BDOOR_CMD_GI_TEST_ADD_EVENT 129 # define BDOOR_CMD_GI_TEST_MAPPING 130 # define BDOOR_CMD_GI_TEST_PPN 131 # define BDOOR_CMD_GI_MAX 131 #define BDOOR_CMD_MKSSTATS_SNAPSHOT 88 /* Devel only. */ # define BDOOR_CMD_MKSSTATS_START 0 # define BDOOR_CMD_MKSSTATS_STOP 1 #define BDOOR_CMD_SECUREBOOT 89 #define BDOOR_CMD_COPY_PHYSMEM 90 /* Devel only. */ #define BDOOR_CMD_STEALCLOCK 91 /* CPL 0 only. */ # define BDOOR_STEALCLOCK_STATUS_DISABLED 0 # define BDOOR_STEALCLOCK_STATUS_ENABLED 1 #define BDOOR_CMD_GUEST_PAGE_HINTS 92 /* CPL 0 only */ #define BDOOR_CMD_FIRMWARE_UPDATE 93 /* CPL 0 only. */ # define BDOOR_CMD_FU_GET_HOST_VERSION 0 # define BDOOR_CMD_FU_UPDATE_FROM_HOST 1 # define BDOOR_CMD_FU_LOCK 2 #define BDOOR_CMD_FUZZER_HELPER 94 /* Devel only. */ # define BDOOR_CMD_FUZZER_INIT 0 # define BDOOR_CMD_FUZZER_NEXT 1 #define BDOOR_CMD_PUTCHR12 95 #define BDOOR_CMD_MAX 96 /* * IMPORTANT NOTE: When modifying the behavior of an existing backdoor command, * you must adhere to the semantics expected by the oldest Tools who use that * command. Specifically, do not alter the way in which the command modifies * the registers. Otherwise backwards compatibility will suffer. */ /* Nesting control operations */ #define NESTING_CONTROL_RESTRICT_BACKDOOR 0 #define NESTING_CONTROL_OPEN_BACKDOOR 1 #define NESTING_CONTROL_QUERY 2 #define NESTING_CONTROL_MAX 2 /* EFI Boot Order options, nibble-sized. */ #define EFI_BOOT_ORDER_TYPE_EFI 0x0 #define EFI_BOOT_ORDER_TYPE_LEGACY 0x1 #define EFI_BOOT_ORDER_TYPE_NONE 0xf #define BDOOR_NETWORK_BOOT_PROTOCOL_NONE 0x0 #define BDOOR_NETWORK_BOOT_PROTOCOL_IPV4 0x1 #define BDOOR_NETWORK_BOOT_PROTOCOL_IPV6 0x2 #define BDOOR_SECUREBOOT_STATUS_DISABLED 0xFFFFFFFFUL #define BDOOR_SECUREBOOT_STATUS_APPROVED 1 #define BDOOR_SECUREBOOT_STATUS_DENIED 2 /* High-bandwidth backdoor port. --hpreg */ #define BDOORHB_PORT 0x5659 #define BDOORHB_CMD_MESSAGE 0 #define BDOORHB_CMD_VASSERT 1 #define BDOORHB_CMD_MAX 2 /* * There is another backdoor which allows access to certain TSC-related * values using otherwise illegal PMC indices when the pseudo_perfctr * control flag is set. */ #define BDOOR_PMC_HW_TSC 0x10000 #define BDOOR_PMC_REAL_NS 0x10001 #define BDOOR_PMC_APPARENT_NS 0x10002 #define BDOOR_PMC_PSEUDO_TSC 0x10003 #define IS_BDOOR_PMC(index) (((index) | 3) == 0x10003) #define BDOOR_CMD(ecx) ((ecx) & 0xffff) /* Sub commands for BDOOR_CMD_VMK_INFO */ #define BDOOR_CMD_VMK_INFO_ENTRY 1 /* * Current format for the guest page hints is: * * Arg0: BDOOR_MAGIC, Arg3: BDOOR_PORT * * Arg1: (rbx on x86) * * 0 64 * | PPN | * * Arg2: (rcx on x86) * * 0 16 32 64 * | Command | Type | Reserved | * * Arg4: (rsi on x86) * * 0 16 64 * | numPages | Reserved | * */ #define BDOOR_GUEST_PAGE_HINTS_NOT_SUPPORTED ((unsigned)-1) #define BDOOR_GUEST_PAGE_HINTS_MAX_PAGES (0xffff) #define BDOOR_GUEST_PAGE_HINTS_TYPE_PSHARE (0) #define BDOOR_GUEST_PAGE_HINTS_TYPE(reg) (((reg) >> 16) & 0xffff) #ifdef VMM /* *---------------------------------------------------------------------- * * Backdoor_CmdRequiresFullyValidVCPU -- * * A few backdoor commands require the full VCPU to be valid * (including GDTR, IDTR, TR and LDTR). The rest get read/write * access to GPRs and read access to Segment registers (selectors). * * Result: * True iff VECX contains a command that require the full VCPU to * be valid. * *---------------------------------------------------------------------- */ static INLINE Bool Backdoor_CmdRequiresFullyValidVCPU(unsigned cmd) { return cmd == BDOOR_CMD_SIDT || cmd == BDOOR_CMD_SGDT || cmd == BDOOR_CMD_SLDT_STR; } #endif #ifdef VM_ARM_64 #define BDOOR_ARM64_LB_PORT (BDOOR_PORT) #define BDOOR_ARM64_HB_PORT_IN (BDOORHB_PORT) #define BDOOR_ARM64_HB_PORT_OUT (BDOORHB_PORT +1) #define BDOOR_ARG0 REG_X0 #define BDOOR_ARG1 REG_X1 #define BDOOR_ARG2 REG_X2 #define BDOOR_ARG3 REG_X3 #define BDOOR_ARG4 REG_X4 #define BDOOR_ARG5 REG_X5 #define BDOOR_ARG6 REG_X6 #else #define BDOOR_ARG0 REG_RAX #define BDOOR_ARG1 REG_RBX #define BDOOR_ARG2 REG_RCX #define BDOOR_ARG3 REG_RDX #define BDOOR_ARG4 REG_RSI #define BDOOR_ARG5 REG_RDI #define BDOOR_ARG6 REG_RBP #endif #if defined __cplusplus } #endif #endif // _BACKDOOR_DEF_H_ vsock-only/shared/compat_ethtool.h 0000444 0000000 0000000 00000003662 13432725347 016337 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef _COMPAT_ETHTOOL_H #define _COMPAT_ETHTOOL_H /* * ethtool is a userspace utility for getting and setting ethernet device * settings. Kernel support for it was first published in 2.4.0-test11, but * only in 2.4.15 were the ethtool_value struct and the ETHTOOL_GLINK ioctl * added to ethtool.h (together, because the ETHTOOL_GLINK ioctl expects a * single value response). * * Likewise, ioctls for getting and setting TSO were published in 2.4.22. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 4, 0) # include <linux/ethtool.h> # ifndef ETHTOOL_GLINK # define ETHTOOL_GLINK 0x0a typedef struct { __u32 cmd; __u32 data; } compat_ethtool_value; # else typedef struct ethtool_value compat_ethtool_value; # endif # ifndef ETHTOOL_GTSO # define ETHTOOL_GTSO 0x1E # define ETHTOOL_STSO 0x1F # endif #endif #if COMPAT_LINUX_VERSION_CHECK_LT(3, 3, 0) # define compat_ethtool_rxfh_indir_default(i, num_queues) (i % num_queues) #else # define compat_ethtool_rxfh_indir_default(i, num_queues) ethtool_rxfh_indir_default(i, num_queues) #endif #endif /* _COMPAT_ETHTOOL_H */ vsock-only/shared/vm_basic_asm_x86_common.h 0000444 0000000 0000000 00000033175 13432725350 020012 0 ustar root root /********************************************************* * Copyright (C) 2013-2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vm_basic_asm_x86_common.h -- * * Basic assembler macros common to 32-bit and 64-bit x86 ISA. */ #ifndef _VM_BASIC_ASM_X86_COMMON_H_ #define _VM_BASIC_ASM_X86_COMMON_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" #ifndef VM_X86_ANY #error "Should be included only in x86 builds" #endif /* * x86-64 windows doesn't support inline asm so we have to use these * intrinsic functions defined in the compiler. Not all of these are well * documented. There is an array in the compiler dll (c1.dll) which has * an array of the names of all the intrinsics minus the leading * underscore. Searching around in the ntddk.h file can also be helpful. * * The declarations for the intrinsic functions were taken from the DDK. * Our declarations must match the ddk's otherwise the 64-bit c++ compiler * will complain about second linkage of the intrinsic functions. * We define the intrinsic using the basic types corresponding to the * Windows typedefs. This avoids having to include windows header files * to get to the windows types. */ #if defined(_MSC_VER) && !defined(BORA_NO_WIN32_INTRINS) #ifdef __cplusplus extern "C" { #endif /* * It seems x86 & x86-64 windows still implements these intrinsic * functions. The documentation for the x86-64 suggest the * __inbyte/__outbyte intrinsics even though the _in/_out work fine and * __inbyte/__outbyte aren't supported on x86. */ int _inp(unsigned short); unsigned short _inpw(unsigned short); unsigned long _inpd(unsigned short); int _outp(unsigned short, int); unsigned short _outpw(unsigned short, unsigned short); unsigned long _outpd(uint16, unsigned long); #pragma intrinsic(_inp, _inpw, _inpd, _outp, _outpw, _outpw, _outpd) /* * Prevents compiler from re-ordering reads, writes and reads&writes. * These functions do not add any instructions thus only affect * the compiler ordering. * * See: * `Lockless Programming Considerations for Xbox 360 and Microsoft Windows' * http://msdn.microsoft.com/en-us/library/bb310595(VS.85).aspx */ void _ReadBarrier(void); void _WriteBarrier(void); void _ReadWriteBarrier(void); #pragma intrinsic(_ReadBarrier, _WriteBarrier, _ReadWriteBarrier) void _mm_mfence(void); void _mm_lfence(void); #pragma intrinsic(_mm_mfence, _mm_lfence) long _InterlockedXor(long volatile *, long); #pragma intrinsic(_InterlockedXor) unsigned int __getcallerseflags(void); #pragma intrinsic(__getcallerseflags) #ifdef VM_X86_64 /* * intrinsic functions only supported by x86-64 windows as of 2k3sp1 */ unsigned __int64 __rdtsc(void); void __stosw(unsigned short *, unsigned short, size_t); void __stosd(unsigned long *, unsigned long, size_t); void _mm_pause(void); #pragma intrinsic(__rdtsc, __stosw, __stosd, _mm_pause) unsigned char _BitScanForward64(unsigned long *, unsigned __int64); unsigned char _BitScanReverse64(unsigned long *, unsigned __int64); #pragma intrinsic(_BitScanForward64, _BitScanReverse64) #endif /* VM_X86_64 */ unsigned char _BitScanForward(unsigned long *, unsigned long); unsigned char _BitScanReverse(unsigned long *, unsigned long); #pragma intrinsic(_BitScanForward, _BitScanReverse) unsigned char _bittest(const long *, long); unsigned char _bittestandset(long *, long); unsigned char _bittestandreset(long *, long); unsigned char _bittestandcomplement(long *, long); #pragma intrinsic(_bittest, _bittestandset, _bittestandreset, _bittestandcomplement) #ifdef VM_X86_64 unsigned char _bittestandset64(__int64 *, __int64); unsigned char _bittestandreset64(__int64 *, __int64); #pragma intrinsic(_bittestandset64, _bittestandreset64) #endif // VM_X86_64 #ifdef __cplusplus } #endif #endif // _MSC_VER #ifdef __GNUC__ /* * Checked against the Intel manual and GCC --hpreg * * volatile because reading from port can modify the state of the underlying * hardware. * * Note: The undocumented %z construct doesn't work (internal compiler error) * with gcc-2.95.1 */ #define __GCC_IN(s, type, name) \ static INLINE type \ name(uint16 port) \ { \ type val; \ \ __asm__ __volatile__( \ "in" #s " %w1, %0" \ : "=a" (val) \ : "Nd" (port) \ ); \ \ return val; \ } __GCC_IN(b, uint8, INB) __GCC_IN(w, uint16, INW) __GCC_IN(l, uint32, IN32) /* * Checked against the Intel manual and GCC --hpreg * * Note: The undocumented %z construct doesn't work (internal compiler error) * with gcc-2.95.1 */ #define __GCC_OUT(s, s2, port, val) do { \ __asm__( \ "out" #s " %" #s2 "1, %w0" \ : \ : "Nd" (port), "a" (val) \ ); \ } while (0) #define OUTB(port, val) __GCC_OUT(b, b, port, val) #define OUTW(port, val) __GCC_OUT(w, w, port, val) #define OUT32(port, val) __GCC_OUT(l, , port, val) #define GET_CURRENT_EIP(_eip) \ __asm__ __volatile("call 0\n\tpopl %0" : "=r" (_eip): ); static INLINE unsigned int GetCallerEFlags(void) { unsigned long flags; asm volatile("pushf; pop %0" : "=r"(flags)); return flags; } #elif defined(_MSC_VER) static INLINE uint8 INB(uint16 port) { return (uint8)_inp(port); } static INLINE void OUTB(uint16 port, uint8 value) { _outp(port, value); } static INLINE uint16 INW(uint16 port) { return _inpw(port); } static INLINE void OUTW(uint16 port, uint16 value) { _outpw(port, value); } static INLINE uint32 IN32(uint16 port) { return _inpd(port); } static INLINE void OUT32(uint16 port, uint32 value) { _outpd(port, value); } #ifndef VM_X86_64 #ifdef NEAR #undef NEAR #endif #define GET_CURRENT_EIP(_eip) do { \ __asm call NEAR PTR $+5 \ __asm pop eax \ __asm mov _eip, eax \ } while (0) #endif // VM_X86_64 static INLINE unsigned int GetCallerEFlags(void) { return __getcallerseflags(); } #endif // __GNUC__ /* Sequence recommended by Intel for the Pentium 4. */ #define INTEL_MICROCODE_VERSION() ( \ __SET_MSR(MSR_BIOS_SIGN_ID, 0), \ __GET_EAX_FROM_CPUID(1), \ __GET_MSR(MSR_BIOS_SIGN_ID)) /* *----------------------------------------------------------------------------- * * RDTSC_BARRIER -- * * Implements an RDTSC fence. Instructions executed prior to the * fence will have completed before the fence and all stores to * memory are flushed from the store buffer. * * On AMD, MFENCE is sufficient. On Intel, only LFENCE is * documented to fence RDTSC, but LFENCE won't drain the store * buffer. So, use MFENCE;LFENCE, which will work on both AMD and * Intel. * * It is the callers' responsibility to check for SSE2 before * calling this function. * * Results: * None. * * Side effects: * Cause loads and stores prior to this to be globally visible, and * RDTSC will not pass. * *----------------------------------------------------------------------------- */ static INLINE void RDTSC_BARRIER(void) { #ifdef __GNUC__ __asm__ __volatile__( "mfence \n\t" "lfence \n\t" ::: "memory" ); #elif defined _MSC_VER /* Prevent compiler from moving code across mfence/lfence. */ _ReadWriteBarrier(); _mm_mfence(); _mm_lfence(); _ReadWriteBarrier(); #else #error No compiler defined for RDTSC_BARRIER #endif } /* * Memory Barriers * =============== * * Terminology * ----------- * * A compiler memory barrier prevents the compiler from re-ordering memory * accesses accross the barrier. It is not a CPU instruction, it is a compiler * directive (i.e. it does not emit any code). * * A CPU memory barrier prevents the CPU from re-ordering memory accesses * accross the barrier. It is a CPU instruction. * * A memory barrier is the union of a compiler memory barrier and a CPU memory * barrier. A compiler memory barrier is a useless construct by itself. It is * only useful when combined with a CPU memory barrier, to implement a memory * barrier. * * Semantics * --------- * * At the time COMPILER_*_BARRIER were created (and references to them were * added to the code), the code was only targetting x86. The intent of the code * was really to use a memory barrier, but because x86 uses a strongly ordered * memory model, the CPU would not re-order memory accesses, and the code could * get away with using just a compiler memory barrier. So COMPILER_*_BARRIER * were born and were implemented as compiler memory barriers _on x86_. But * make no mistake, _the semantics that the code expects from * COMPILER_*_BARRIER are that of a memory barrier_! * * DO NOT USE! * ----------- * * On at least one non-x86 architecture, COMPILER_*_BARRIER are * 1) Misnomers * 2) Not fine-grained enough to provide the best performance. * For the above two reasons, usage of COMPILER_*_BARRIER is now deprecated. * _Do not add new references to COMPILER_*_BARRIER._ Instead, precisely * document the intent of your code by using * <mem_type/purpose>_<before_access_type>_BARRIER_<after_access_type>. * Existing references to COMPILER_*_BARRIER are being slowly but surely * converted, and when no references are left, COMPILER_*_BARRIER will be * retired. * * Thanks for pasting this whole comment into every architecture header. */ #if defined __GNUC__ # define COMPILER_READ_BARRIER() COMPILER_MEM_BARRIER() # define COMPILER_WRITE_BARRIER() COMPILER_MEM_BARRIER() # define COMPILER_MEM_BARRIER() __asm__ __volatile__("" ::: "memory") #elif defined _MSC_VER # define COMPILER_READ_BARRIER() _ReadBarrier() # define COMPILER_WRITE_BARRIER() _WriteBarrier() # define COMPILER_MEM_BARRIER() _ReadWriteBarrier() #endif /* * Memory barriers. These take the form of * * <mem_type/purpose>_<before_access_type>_BARRIER_<after_access_type> * * where: * <mem_type/purpose> is either SMP, DMA, or MMIO. * <*_access type> is either R(load), W(store) or RW(any). * * Above every use of these memory barriers in the code, there _must_ be a * comment to justify the use, i.e. a comment which: * * 1) Precisely identifies which memory accesses must not be re-ordered across * the memory barrier. * 2) Explains why it is important that the memory accesses not be re-ordered. * * Thanks for pasting this whole comment into every architecture header. * * On x86, we only need to care specifically about store-load reordering on * normal memory types. In other cases, only a compiler barrier is needed. * SMP_W_BARRIER_R is implemented with a locked xor operation (instead of the * mfence instruction) for performance reasons. See PR 1674199 for more * details. * * On x64, special instructions are only provided for load-load (lfence) and * store-store (sfence) ordering, and they don't apply to normal memory. */ static INLINE void SMP_W_BARRIER_R(void) { volatile long temp; #if defined __GNUC__ __asm__ __volatile__ ( "lock xorl $1, %0" : "+m" (temp) : /* no additional inputs */ : "cc", "memory"); #elif defined _MSC_VER _InterlockedXor(&temp, 1); #else #error SMP_W_BARRIER_R not defined for this compiler #endif } #define SMP_R_BARRIER_R() COMPILER_READ_BARRIER() #define SMP_R_BARRIER_W() COMPILER_MEM_BARRIER() #define SMP_R_BARRIER_RW() COMPILER_MEM_BARRIER() #define SMP_W_BARRIER_W() COMPILER_WRITE_BARRIER() #define SMP_W_BARRIER_RW() SMP_W_BARRIER_R() #define SMP_RW_BARRIER_R() SMP_W_BARRIER_R() #define SMP_RW_BARRIER_W() COMPILER_MEM_BARRIER() #define SMP_RW_BARRIER_RW() SMP_W_BARRIER_R() /* * Like the above, only for use with observers other than CPUs, * i.e. DMA masters. */ #define DMA_R_BARRIER_R() SMP_R_BARRIER_R() #define DMA_R_BARRIER_W() SMP_R_BARRIER_W() #define DMA_R_BARRIER_RW() SMP_R_BARRIER_RW() #define DMA_W_BARRIER_R() SMP_W_BARRIER_R() #define DMA_W_BARRIER_W() SMP_W_BARRIER_W() #define DMA_W_BARRIER_RW() SMP_W_BARRIER_RW() #define DMA_RW_BARRIER_R() SMP_RW_BARRIER_R() #define DMA_RW_BARRIER_W() SMP_RW_BARRIER_W() #define DMA_RW_BARRIER_RW() SMP_RW_BARRIER_RW() /* * And finally a set for use with MMIO accesses. */ #define MMIO_R_BARRIER_R() SMP_R_BARRIER_R() #define MMIO_R_BARRIER_W() SMP_R_BARRIER_W() #define MMIO_R_BARRIER_RW() SMP_R_BARRIER_RW() #define MMIO_W_BARRIER_R() SMP_W_BARRIER_R() #define MMIO_W_BARRIER_W() SMP_W_BARRIER_W() #define MMIO_W_BARRIER_RW() SMP_W_BARRIER_RW() #define MMIO_RW_BARRIER_R() SMP_RW_BARRIER_R() #define MMIO_RW_BARRIER_W() SMP_RW_BARRIER_W() #define MMIO_RW_BARRIER_RW() SMP_RW_BARRIER_RW() #endif // _VM_BASIC_ASM_X86_COMMON_H_ vsock-only/shared/vm_device_version.h 0000444 0000000 0000000 00000034705 13432725350 017020 0 ustar root root /********************************************************* * Copyright (C) 1998,2005-2012,2014-2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef VM_DEVICE_VERSION_H #define VM_DEVICE_VERSION_H #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMKDRIVERS #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" #ifdef _WIN32 #ifdef __MINGW32__ #include "initguid.h" #else #include "guiddef.h" #endif #endif /* LSILogic 53C1030 Parallel SCSI controller * LSILogic SAS1068 SAS controller */ #define PCI_VENDOR_ID_LSILOGIC 0x1000 #define PCI_DEVICE_ID_LSI53C1030 0x0030 #define PCI_DEVICE_ID_LSISAS1068 0x0054 /* Our own PCI IDs * VMware SVGA II (Unified VGA) * VMware SVGA (PCI Accelerator) * VMware vmxnet (Idealized NIC) * VMware vmxscsi (Abortive idealized SCSI controller) * VMware chipset (Subsystem ID for our motherboards) * VMware e1000 (Subsystem ID) * VMware vmxnet3 (Uniform Pass Through NIC) * VMware HD Audio codec * VMware HD Audio controller */ #define PCI_VENDOR_ID_VMWARE 0x15AD #define PCI_DEVICE_ID_VMWARE_SVGA2 0x0405 #define PCI_DEVICE_ID_VMWARE_SVGA 0x0710 #define PCI_DEVICE_ID_VMWARE_VGA 0x0711 #define PCI_DEVICE_ID_VMWARE_NET 0x0720 #define PCI_DEVICE_ID_VMWARE_SCSI 0x0730 #define PCI_DEVICE_ID_VMWARE_VMCI 0x0740 #define PCI_DEVICE_ID_VMWARE_CHIPSET 0x1976 #define PCI_DEVICE_ID_VMWARE_82545EM 0x0750 /* single port */ #define PCI_DEVICE_ID_VMWARE_82546EB 0x0760 /* dual port */ #define PCI_DEVICE_ID_VMWARE_EHCI 0x0770 #define PCI_DEVICE_ID_VMWARE_UHCI 0x0774 #define PCI_DEVICE_ID_VMWARE_XHCI_0096 0x0778 #define PCI_DEVICE_ID_VMWARE_XHCI_0100 0x0779 #define PCI_DEVICE_ID_VMWARE_1394 0x0780 #define PCI_DEVICE_ID_VMWARE_BRIDGE 0x0790 #define PCI_DEVICE_ID_VMWARE_ROOTPORT 0x07A0 #define PCI_DEVICE_ID_VMWARE_VMXNET3 0x07B0 #define PCI_DEVICE_ID_VMWARE_PVSCSI 0x07C0 #define PCI_DEVICE_ID_VMWARE_82574 0x07D0 #define PCI_DEVICE_ID_VMWARE_AHCI 0x07E0 #define PCI_DEVICE_ID_VMWARE_NVME 0x07F0 #define PCI_DEVICE_ID_VMWARE_HDAUDIO_CODEC 0x1975 #define PCI_DEVICE_ID_VMWARE_HDAUDIO_CONTROLLER 0x1977 /* * TXT vendor, device and revision ID. We are keeping vendor * as Intel since tboot code does not like anything other * than Intel in the SINIT ACM header. */ #define TXT_VENDOR_ID 0x8086 #define TXT_DEVICE_ID 0xB002 #define TXT_REVISION_ID 0x01 /* The hypervisor device might grow. Please leave room * for 7 more subfunctions. */ #define PCI_DEVICE_ID_VMWARE_HYPER 0x0800 #define PCI_DEVICE_ID_VMWARE_VMI 0x0801 #define PCI_DEVICE_VMI_CLASS 0x05 #define PCI_DEVICE_VMI_SUBCLASS 0x80 #define PCI_DEVICE_VMI_INTERFACE 0x00 #define PCI_DEVICE_VMI_REVISION 0x01 /* * Device IDs for the PCI passthru test device: * * 0x0809 is for old fashioned PCI with MSI. * 0x080A is for PCI express with MSI-X. * 0x080B is for PCI express with configurable BARs. */ #define PCI_DEVICE_ID_VMWARE_PCI_TEST 0x0809 #define PCI_DEVICE_ID_VMWARE_PCIE_TEST1 0x080A #define PCI_DEVICE_ID_VMWARE_PCIE_TEST2 0x080B #define PCI_DEVICE_ID_VMWARE_VRDMA 0x0820 #define PCI_DEVICE_ID_VMWARE_VTPM 0x0830 /* * VMware Virtual Device Test Infrastructure (VDTI) devices */ #define PCI_DEVICE_ID_VMWARE_VDTI 0x7E57 /* stands for "TEST" */ /* From linux/pci_ids.h: * AMD Lance Ethernet controller * BusLogic SCSI controller * Ensoniq ES1371 sound controller */ #define PCI_VENDOR_ID_AMD 0x1022 #define PCI_DEVICE_ID_AMD_VLANCE 0x2000 #define PCI_DEVICE_ID_AMD_IOMMU 0x1577 #define PCI_VENDOR_ID_BUSLOGIC 0x104B #define PCI_DEVICE_ID_BUSLOGIC_MULTIMASTER_NC 0x0140 #define PCI_DEVICE_ID_BUSLOGIC_MULTIMASTER 0x1040 #define PCI_VENDOR_ID_ENSONIQ 0x1274 #define PCI_DEVICE_ID_ENSONIQ_ES1371 0x1371 /* From linux/pci_ids.h: * Intel 82439TX (430 HX North Bridge) * Intel 82371AB (PIIX4 South Bridge) * Intel 82443BX (440 BX North Bridge and AGP Bridge) * Intel 82545EM (e1000, server adapter, single port) * Intel 82546EB (e1000, server adapter, dual port) * Intel HECI (as embedded in ich9m) * Intel XHCI (Panther Point / Intel 7 Series) */ #define PCI_VENDOR_ID_INTEL 0x8086 #define PCI_DEVICE_ID_INTEL_82439TX 0x7100 #define PCI_DEVICE_ID_INTEL_82371AB_0 0x7110 #define PCI_DEVICE_ID_INTEL_82371AB_2 0x7112 #define PCI_DEVICE_ID_INTEL_82371AB_3 0x7113 #define PCI_DEVICE_ID_INTEL_82371AB 0x7111 #define PCI_DEVICE_ID_INTEL_82443BX 0x7190 #define PCI_DEVICE_ID_INTEL_82443BX_1 0x7191 #define PCI_DEVICE_ID_INTEL_82443BX_2 0x7192 /* Used when no AGP support */ #define PCI_DEVICE_ID_INTEL_82545EM 0x100f #define PCI_DEVICE_ID_INTEL_82546EB 0x1010 #define PCI_DEVICE_ID_INTEL_82574 0x10d3 #define PCI_DEVICE_ID_INTEL_82574_APPLE 0x10f6 #define PCI_DEVICE_ID_INTEL_HECI 0x2a74 #define PCI_DEVICE_ID_INTEL_PANTHERPOINT_XHCI 0x1e31 /* * From drivers/usb/host/xhci-pci.c: * Intel XHCI (Lynx Point / Intel 8 Series) */ #define PCI_DEVICE_ID_INTEL_LYNXPOINT_XHCI 0x8c31 /* * Intel Volume Management Device (VMD) */ #define PCI_DEVICE_ID_INTEL_VMD_V1 0x201d /* * Intel Quickassist (QAT) devices. */ #define PCI_DEVICE_ID_INTEL_QAT_DH895XCC 0x0435 #define PCI_DEVICE_ID_INTEL_QAT_DH895XCC_VF 0x0443 #define PCI_DEVICE_ID_INTEL_QAT_C62X 0x37c8 #define PCI_DEVICE_ID_INTEL_QAT_C62X_VF 0x37c9 /* * Intel FPGAs */ #define PCI_DEVICE_ID_INTEL_FPGA_SKL_PF 0xbcc0 #define PCI_DEVICE_ID_INTEL_FPGA_SKL_VF 0xbcc1 #define E1000E_PCI_DEVICE_ID_CONFIG_STR "e1000e.pci.deviceID" #define E1000E_PCI_SUB_VENDOR_ID_CONFIG_STR "e1000e.pci.subVendorID" #define E1000E_PCI_SUB_DEVICE_ID_CONFIG_STR "e1000e.pci.subDeviceID" /* * Intel HD Audio controller and Realtek ALC885 codec. */ #define PCI_DEVICE_ID_INTEL_631XESB_632XESB 0x269a #define PCI_VENDOR_ID_REALTEK 0x10ec #define PCI_DEVICE_ID_REALTEK_ALC885 0x0885 /* * Fresco Logic xHCI (USB 3.0) Controller */ #define PCI_VENDOR_ID_FRESCO 0x1B73 #define PCI_DEVICE_ID_FRESCO_FL1000 0x1000 // Original 1-port chip #define PCI_DEVICE_ID_FRESCO_FL1009 0x1009 // New 2-port chip (Driver 3.0.98+) #define PCI_DEVICE_ID_FRESCO_FL1400 0x1400 // Unknown (4-port? Dev hardware?) /* * NEC/Renesas xHCI (USB 3.0) Controller */ #define PCI_VENDOR_ID_NEC 0x1033 #define PCI_DEVICE_ID_NEC_UPD720200 0x0194 #define PCI_REVISION_NEC_UPD720200 0x03 #define PCI_FIRMWARE_NEC_UPD720200 0x3015 #define SATA_ID_SERIAL_STR "00000000000000000001" /* Must be 20 Bytes */ #define SATA_ID_FIRMWARE_STR "00000001" /* Must be 8 Bytes */ #define AHCI_ATA_MODEL_STR PRODUCT_GENERIC_NAME " Virtual SATA Hard Drive" #define AHCI_ATAPI_MODEL_STR PRODUCT_GENERIC_NAME " Virtual SATA CDRW Drive" /************* Strings for IDE Identity Fields **************************/ #define VIDE_ID_SERIAL_STR "00000000000000000001" /* Must be 20 Bytes */ #define VIDE_ID_FIRMWARE_STR "00000001" /* Must be 8 Bytes */ /* No longer than 40 Bytes */ #define VIDE_ATA_MODEL_STR PRODUCT_GENERIC_NAME " Virtual IDE Hard Drive" #define VIDE_ATAPI_MODEL_STR PRODUCT_GENERIC_NAME " Virtual IDE CDROM Drive" #define ATAPI_VENDOR_ID "NECVMWar" /* Must be 8 Bytes */ #define ATAPI_PRODUCT_ID PRODUCT_GENERIC_NAME " IDE CDROM" /* Must be 16 Bytes */ #define ATAPI_REV_LEVEL "1.00" /* Must be 4 Bytes */ #define IDE_NUM_INTERFACES 2 /* support for two interfaces */ #define IDE_DRIVES_PER_IF 2 /************* Strings for SCSI Identity Fields **************************/ #define SCSI_DISK_MODEL_STR PRODUCT_GENERIC_NAME " Virtual SCSI Hard Drive" #define SCSI_DISK_VENDOR_NAME COMPANY_NAME #define SCSI_DISK_REV_LEVEL "1.0" #define SCSI_CDROM_MODEL_STR PRODUCT_GENERIC_NAME " Virtual SCSI CDROM Drive" #define SCSI_CDROM_VENDOR_NAME COMPANY_NAME #define SCSI_CDROM_REV_LEVEL "1.0" /************* NVME implementation limits ********************************/ #define NVME_MAX_CONTROLLERS 4 #define NVME_MIN_NAMESPACES 1 #define NVME_MAX_NAMESPACES 15 /* We support only 15 namespaces same * as SCSI devices. */ /************* SCSI implementation limits ********************************/ #define SCSI_MAX_CONTROLLERS 4 // Need more than 1 for MSCS clustering #define SCSI_MAX_DEVICES 16 // BT-958 emulates only 16 #define PVSCSI_HWV14_MAX_DEVICES 65 /* HWv14 And Later Supports 64 * + controller at ID 7 */ #define PVSCSI_MAX_DEVICES 255 // 255 (including the controller) #define PVSCSI_MAX_NUM_DISKS (PVSCSI_HWV14_MAX_DEVICES - 1) /************* SATA implementation limits ********************************/ #define SATA_MAX_CONTROLLERS 4 #define SATA_MAX_DEVICES 30 #define AHCI_MIN_PORTS 1 #define AHCI_MAX_PORTS SATA_MAX_DEVICES /* * Publicly supported maximum number of disks per VM. */ #define MAX_NUM_DISKS \ ((SATA_MAX_CONTROLLERS * SATA_MAX_DEVICES) + \ (SCSI_MAX_CONTROLLERS * SCSI_MAX_DEVICES) + \ (NVME_MAX_CONTROLLERS * NVME_MAX_NAMESPACES) + \ (IDE_NUM_INTERFACES * IDE_DRIVES_PER_IF)) /* * Maximum number of supported disks in a VM from HWV14 or later, using PVSCSI updated max * devices. The note above still holds true, but instead of publicly supporting * all devices, HWv14 simply extends the maximum support to 256 devices, * instead ~244 calculated above. * * PVSCSI_HW_MAX_DEVICES is 65 - allowing 64 disks + controller (at ID 7) * 4 * 64 = 256 devices. * */ #define MAX_NUM_DISKS_HWV14 MAX(MAX_NUM_DISKS, \ (SCSI_MAX_CONTROLLERS * PVSCSI_MAX_NUM_DISKS)) /* * VSCSI_BV_INTS is the number of uint32's needed for a bit vector * to cover all scsi devices per target. */ #define VSCSI_BV_INTS CEILING(PVSCSI_MAX_DEVICES, 8 * sizeof (uint32)) #define SCSI_IDE_CHANNEL SCSI_MAX_CONTROLLERS #define SCSI_IDE_HOSTED_CHANNEL (SCSI_MAX_CONTROLLERS + 1) #define SCSI_SATA_CHANNEL_FIRST (SCSI_IDE_HOSTED_CHANNEL + 1) #define SCSI_NVME_CHANNEL_FIRST (SCSI_SATA_CHANNEL_FIRST + \ SATA_MAX_CONTROLLERS) #define SCSI_MAX_CHANNELS (SCSI_NVME_CHANNEL_FIRST + \ NVME_MAX_CONTROLLERS) /************* SCSI-NVME channel IDs *******************************/ #define NVME_ID_TO_SCSI_ID(nvmeId) \ (SCSI_NVME_CHANNEL_FIRST + (nvmeId)) #define SCSI_ID_TO_NVME_ID(scsiId) \ ((scsiId) - SCSI_NVME_CHANNEL_FIRST) /************* SCSI-SATA channel IDs********************************/ #define SATA_ID_TO_SCSI_ID(sataId) \ (SCSI_SATA_CHANNEL_FIRST + (sataId)) #define SCSI_ID_TO_SATA_ID(scsiId) \ ((scsiId) - SCSI_SATA_CHANNEL_FIRST) /************* Strings for the VESA BIOS Identity Fields *****************/ #define VBE_OEM_STRING COMPANY_NAME " SVGA" #define VBE_VENDOR_NAME COMPANY_NAME #define VBE_PRODUCT_NAME PRODUCT_GENERIC_NAME /************* PCI implementation limits ********************************/ #define PCI_MAX_BRIDGES 15 /************* Ethernet implementation limits ***************************/ #define MAX_ETHERNET_CARDS 10 /********************** Floppy limits ***********************************/ #define MAX_FLOPPY_DRIVES 2 /************* PCI Passthrough implementation limits ********************/ #define MAX_PCI_PASSTHRU_DEVICES 16 /************* Test device implementation limits ********************/ #define MAX_PCI_TEST_DEVICES 16 /************* VDTI PCI Device implementation limits ********************/ #define MAX_VDTI_PCI_DEVICES 16 /************* USB implementation limits ********************************/ #define MAX_USB_DEVICES_PER_HOST_CONTROLLER 127 /************* NVDIMM implementation limits ********************************/ #define NVDIMM_MAX_CONTROLLERS 1 #define MAX_NVDIMM 64 /************* vRDMA implementation limits ******************************/ #define MAX_VRDMA_DEVICES 1 /************* QAT implementation limits ********************/ #define MAX_QAT_PCI_DEVICES 4 /************* Strings for Host USB Driver *******************************/ #ifdef _WIN32 /* * Globally unique ID for the VMware device interface. Define INITGUID before including * this header file to instantiate the variable. */ DEFINE_GUID(GUID_DEVICE_INTERFACE_VMWARE_USB_DEVICES, 0x2da1fe75, 0xaab3, 0x4d2c, 0xac, 0xdf, 0x39, 0x8, 0x8c, 0xad, 0xa6, 0x65); /* * Globally unique ID for the VMware device setup class. */ DEFINE_GUID(GUID_CLASS_VMWARE_USB_DEVICES, 0x3b3e62a5, 0x3556, 0x4d7e, 0xad, 0xad, 0xf5, 0xfa, 0x3a, 0x71, 0x2b, 0x56); /* * This string defines the device ID string of a VMware USB device. * The format is USB\Vid_XXXX&Pid_YYYY, where XXXX and YYYY are the * hexadecimal representations of the vendor and product ids, respectively. * * The official vendor ID for VMware, Inc. is 0x0E0F. * The product id for USB generic devices is 0x0001. */ #define USB_VMWARE_DEVICE_ID_WIDE L"USB\\Vid_0E0F&Pid_0001" #define USB_DEVICE_ID_LENGTH (sizeof(USB_VMWARE_DEVICE_ID_WIDE) / sizeof(WCHAR)) #ifdef UNICODE #define USB_PNP_SETUP_CLASS_NAME L"VMwareUSBDevices" #define USB_PNP_DRIVER_NAME L"vmusb" #else #define USB_PNP_SETUP_CLASS_NAME "VMwareUSBDevices" #define USB_PNP_DRIVER_NAME "vmusb" #endif #endif /* * Our JEDEC 2 Manufacturer ID number is 2 in bank 10. Our number is nine * bytes of continuation code (with an odd parity bit in bit 7) followed by the * number itself. * */ #define JEDEC_VENDOR_ID_VMWARE 0x289 #define JEDEC_DEVICE_ID_VMWARE_NVDIMM 0x1 #endif /* VM_DEVICE_VERSION_H */ vsock-only/shared/vmci_defs.h 0000444 0000000 0000000 00000066156 13432725350 015256 0 ustar root root /********************************************************* * Copyright (C) 2005-2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef _VMCI_DEF_H_ #define _VMCI_DEF_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_VMMEXT #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_DISTRIBUTE #include "includeCheck.h" #include "vm_basic_types.h" #include "vm_basic_defs.h" #include "vm_atomic.h" #include "vm_assert.h" #if defined __cplusplus extern "C" { #endif /* Register offsets. */ #define VMCI_STATUS_ADDR 0x00 #define VMCI_CONTROL_ADDR 0x04 #define VMCI_ICR_ADDR 0x08 #define VMCI_IMR_ADDR 0x0c #define VMCI_DATA_OUT_ADDR 0x10 #define VMCI_DATA_IN_ADDR 0x14 #define VMCI_CAPS_ADDR 0x18 #define VMCI_RESULT_LOW_ADDR 0x1c #define VMCI_RESULT_HIGH_ADDR 0x20 /* Max number of devices. */ #define VMCI_MAX_DEVICES 1 /* Status register bits. */ #define VMCI_STATUS_INT_ON 0x1 /* Control register bits. */ #define VMCI_CONTROL_RESET 0x1 #define VMCI_CONTROL_INT_ENABLE 0x2 #define VMCI_CONTROL_INT_DISABLE 0x4 /* Capabilities register bits. */ #define VMCI_CAPS_HYPERCALL 0x1 #define VMCI_CAPS_GUESTCALL 0x2 #define VMCI_CAPS_DATAGRAM 0x4 #define VMCI_CAPS_NOTIFICATIONS 0x8 /* Interrupt Cause register bits. */ #define VMCI_ICR_DATAGRAM 0x1 #define VMCI_ICR_NOTIFICATION 0x2 /* Interrupt Mask register bits. */ #define VMCI_IMR_DATAGRAM 0x1 #define VMCI_IMR_NOTIFICATION 0x2 /* Interrupt type. */ typedef enum VMCIIntrType { VMCI_INTR_TYPE_INTX = 0, VMCI_INTR_TYPE_MSI = 1, VMCI_INTR_TYPE_MSIX = 2 } VMCIIntrType; /* * Maximum MSI/MSI-X interrupt vectors in the device. */ #define VMCI_MAX_INTRS 2 /* * Supported interrupt vectors. There is one for each ICR value above, * but here they indicate the position in the vector array/message ID. */ #define VMCI_INTR_DATAGRAM 0 #define VMCI_INTR_NOTIFICATION 1 /* * A single VMCI device has an upper limit of 128 MiB on the amount of * memory that can be used for queue pairs. */ #define VMCI_MAX_GUEST_QP_MEMORY (128 * 1024 * 1024) /* * We have a fixed set of resource IDs available in the VMX. * This allows us to have a very simple implementation since we statically * know how many will create datagram handles. If a new caller arrives and * we have run out of slots we can manually increment the maximum size of * available resource IDs. */ typedef uint32 VMCI_Resource; /* VMCI reserved hypervisor datagram resource IDs. */ #define VMCI_RESOURCES_QUERY 0 #define VMCI_GET_CONTEXT_ID 1 #define VMCI_SET_NOTIFY_BITMAP 2 #define VMCI_DOORBELL_LINK 3 #define VMCI_DOORBELL_UNLINK 4 #define VMCI_DOORBELL_NOTIFY 5 /* * VMCI_DATAGRAM_REQUEST_MAP and VMCI_DATAGRAM_REMOVE_MAP are * obsoleted by the removal of VM to VM communication. */ #define VMCI_DATAGRAM_REQUEST_MAP 6 #define VMCI_DATAGRAM_REMOVE_MAP 7 #define VMCI_EVENT_SUBSCRIBE 8 #define VMCI_EVENT_UNSUBSCRIBE 9 #define VMCI_QUEUEPAIR_ALLOC 10 #define VMCI_QUEUEPAIR_DETACH 11 /* * VMCI_VSOCK_VMX_LOOKUP was assigned to 12 for Fusion 3.0/3.1, * WS 7.0/7.1 and ESX 4.1 */ #define VMCI_HGFS_TRANSPORT 13 #define VMCI_UNITY_PBRPC_REGISTER 14 /* * This resource is used for VMCI socket control packets sent to the * hypervisor (CID 0) because RID 1 is already reserved. */ #define VSOCK_PACKET_HYPERVISOR_RID 15 #define VMCI_RESOURCE_MAX 16 /* * The core VMCI device functionality only requires the resource IDs of * VMCI_QUEUEPAIR_DETACH and below. */ #define VMCI_CORE_DEVICE_RESOURCE_MAX VMCI_QUEUEPAIR_DETACH /* * VMCI reserved host datagram resource IDs. * vsock control channel has resource id 1. */ #define VMCI_DVFILTER_DATA_PATH_DATAGRAM 2 /* VMCI Ids. */ typedef uint32 VMCIId; typedef struct VMCIIdRange { int8 action; // VMCI_FA_X, for use in filters. VMCIId begin; // Beginning of range VMCIId end; // End of range } VMCIIdRange; typedef struct VMCIHandle { VMCIId context; VMCIId resource; } VMCIHandle; static INLINE VMCIHandle VMCI_MAKE_HANDLE(VMCIId cid, // IN: VMCIId rid) // IN: { VMCIHandle h; h.context = cid; h.resource = rid; return h; } /* *---------------------------------------------------------------------- * * VMCI_HANDLE_TO_UINT64 -- * * Helper for VMCI handle to uint64 conversion. * * Results: * The uint64 value. * * Side effects: * None. * *---------------------------------------------------------------------- */ static INLINE uint64 VMCI_HANDLE_TO_UINT64(VMCIHandle handle) // IN: { uint64 handle64; handle64 = handle.context; handle64 <<= 32; handle64 |= handle.resource; return handle64; } /* *---------------------------------------------------------------------- * * VMCI_UINT64_TO_HANDLE -- * * Helper for uint64 to VMCI handle conversion. * * Results: * The VMCI handle value. * * Side effects: * None. * *---------------------------------------------------------------------- */ static INLINE VMCIHandle VMCI_UINT64_TO_HANDLE(uint64 handle64) // IN: { VMCIId context = (VMCIId)(handle64 >> 32); VMCIId resource = (VMCIId)handle64; return VMCI_MAKE_HANDLE(context, resource); } #define VMCI_HANDLE_TO_CONTEXT_ID(_handle) ((_handle).context) #define VMCI_HANDLE_TO_RESOURCE_ID(_handle) ((_handle).resource) #define VMCI_HANDLE_EQUAL(_h1, _h2) ((_h1).context == (_h2).context && \ (_h1).resource == (_h2).resource) #define VMCI_INVALID_ID 0xFFFFFFFF static const VMCIHandle VMCI_INVALID_HANDLE = {VMCI_INVALID_ID, VMCI_INVALID_ID}; #define VMCI_HANDLE_INVALID(_handle) \ VMCI_HANDLE_EQUAL((_handle), VMCI_INVALID_HANDLE) /* * The below defines can be used to send anonymous requests. * This also indicates that no response is expected. */ #define VMCI_ANON_SRC_CONTEXT_ID VMCI_INVALID_ID #define VMCI_ANON_SRC_RESOURCE_ID VMCI_INVALID_ID #define VMCI_ANON_SRC_HANDLE VMCI_MAKE_HANDLE(VMCI_ANON_SRC_CONTEXT_ID, \ VMCI_ANON_SRC_RESOURCE_ID) /* The lowest 16 context ids are reserved for internal use. */ #define VMCI_RESERVED_CID_LIMIT 16 /* * Hypervisor context id, used for calling into hypervisor * supplied services from the VM. */ #define VMCI_HYPERVISOR_CONTEXT_ID 0 /* * Well-known context id, a logical context that contains a set of * well-known services. This context ID is now obsolete. */ #define VMCI_WELL_KNOWN_CONTEXT_ID 1 /* * Context ID used by host endpoints. */ #define VMCI_HOST_CONTEXT_ID 2 #define VMCI_HOST_CONTEXT_INVALID_EVENT ((uintptr_t)~0) #define VMCI_CONTEXT_IS_VM(_cid) (VMCI_INVALID_ID != _cid && \ _cid > VMCI_HOST_CONTEXT_ID) /* * The VMCI_CONTEXT_RESOURCE_ID is used together with VMCI_MAKE_HANDLE to make * handles that refer to a specific context. */ #define VMCI_CONTEXT_RESOURCE_ID 0 /* *----------------------------------------------------------------------------- * * VMCI error codes. * *----------------------------------------------------------------------------- */ #define VMCI_SUCCESS_QUEUEPAIR_ATTACH 5 #define VMCI_SUCCESS_QUEUEPAIR_CREATE 4 #define VMCI_SUCCESS_LAST_DETACH 3 #define VMCI_SUCCESS_ACCESS_GRANTED 2 #define VMCI_SUCCESS_ENTRY_DEAD 1 #define VMCI_SUCCESS 0LL #define VMCI_ERROR_INVALID_RESOURCE (-1) #define VMCI_ERROR_INVALID_ARGS (-2) #define VMCI_ERROR_NO_MEM (-3) #define VMCI_ERROR_DATAGRAM_FAILED (-4) #define VMCI_ERROR_MORE_DATA (-5) #define VMCI_ERROR_NO_MORE_DATAGRAMS (-6) #define VMCI_ERROR_NO_ACCESS (-7) #define VMCI_ERROR_NO_HANDLE (-8) #define VMCI_ERROR_DUPLICATE_ENTRY (-9) #define VMCI_ERROR_DST_UNREACHABLE (-10) #define VMCI_ERROR_PAYLOAD_TOO_LARGE (-11) #define VMCI_ERROR_INVALID_PRIV (-12) #define VMCI_ERROR_GENERIC (-13) #define VMCI_ERROR_PAGE_ALREADY_SHARED (-14) #define VMCI_ERROR_CANNOT_SHARE_PAGE (-15) #define VMCI_ERROR_CANNOT_UNSHARE_PAGE (-16) #define VMCI_ERROR_NO_PROCESS (-17) #define VMCI_ERROR_NO_DATAGRAM (-18) #define VMCI_ERROR_NO_RESOURCES (-19) #define VMCI_ERROR_UNAVAILABLE (-20) #define VMCI_ERROR_NOT_FOUND (-21) #define VMCI_ERROR_ALREADY_EXISTS (-22) #define VMCI_ERROR_NOT_PAGE_ALIGNED (-23) #define VMCI_ERROR_INVALID_SIZE (-24) #define VMCI_ERROR_REGION_ALREADY_SHARED (-25) #define VMCI_ERROR_TIMEOUT (-26) #define VMCI_ERROR_DATAGRAM_INCOMPLETE (-27) #define VMCI_ERROR_INCORRECT_IRQL (-28) #define VMCI_ERROR_EVENT_UNKNOWN (-29) #define VMCI_ERROR_OBSOLETE (-30) #define VMCI_ERROR_QUEUEPAIR_MISMATCH (-31) #define VMCI_ERROR_QUEUEPAIR_NOTSET (-32) #define VMCI_ERROR_QUEUEPAIR_NOTOWNER (-33) #define VMCI_ERROR_QUEUEPAIR_NOTATTACHED (-34) #define VMCI_ERROR_QUEUEPAIR_NOSPACE (-35) #define VMCI_ERROR_QUEUEPAIR_NODATA (-36) #define VMCI_ERROR_BUSMEM_INVALIDATION (-37) #define VMCI_ERROR_MODULE_NOT_LOADED (-38) #define VMCI_ERROR_DEVICE_NOT_FOUND (-39) #define VMCI_ERROR_QUEUEPAIR_NOT_READY (-40) #define VMCI_ERROR_WOULD_BLOCK (-41) /* VMCI clients should return error code withing this range */ #define VMCI_ERROR_CLIENT_MIN (-500) #define VMCI_ERROR_CLIENT_MAX (-550) /* Internal error codes. */ #define VMCI_SHAREDMEM_ERROR_BAD_CONTEXT (-1000) #define VMCI_PATH_MAX 256 /* VMCI reserved events. */ typedef uint32 VMCI_Event; #define VMCI_EVENT_CTX_ID_UPDATE 0 // Only applicable to guest endpoints #define VMCI_EVENT_CTX_REMOVED 1 // Applicable to guest and host #define VMCI_EVENT_QP_RESUMED 2 // Only applicable to guest endpoints #define VMCI_EVENT_QP_PEER_ATTACH 3 // Applicable to guest, host and VMX #define VMCI_EVENT_QP_PEER_DETACH 4 // Applicable to guest, host and VMX #define VMCI_EVENT_MEM_ACCESS_ON 5 // Applicable to VMX and vmk. On vmk, // this event has the Context payload type. #define VMCI_EVENT_MEM_ACCESS_OFF 6 // Applicable to VMX and vmk. Same as // above for the payload type. #define VMCI_EVENT_GUEST_PAUSED 7 // Applicable to vmk. This event has the // Context payload type. #define VMCI_EVENT_GUEST_UNPAUSED 8 // Applicable to vmk. Same as above for // the payload type. #define VMCI_EVENT_MAX 9 /* * Of the above events, a few are reserved for use in the VMX, and * other endpoints (guest and host kernel) should not use them. For * the rest of the events, we allow both host and guest endpoints to * subscribe to them, to maintain the same API for host and guest * endpoints. */ #define VMCI_EVENT_VALID_VMX(_event) (_event == VMCI_EVENT_QP_PEER_ATTACH || \ _event == VMCI_EVENT_QP_PEER_DETACH || \ _event == VMCI_EVENT_MEM_ACCESS_ON || \ _event == VMCI_EVENT_MEM_ACCESS_OFF) #if defined(VMX86_SERVER) #define VMCI_EVENT_VALID(_event) (_event < VMCI_EVENT_MAX) #else // VMX86_SERVER #define VMCI_EVENT_VALID(_event) (_event < VMCI_EVENT_MAX && \ _event != VMCI_EVENT_MEM_ACCESS_ON && \ _event != VMCI_EVENT_MEM_ACCESS_OFF && \ _event != VMCI_EVENT_GUEST_PAUSED && \ _event != VMCI_EVENT_GUEST_UNPAUSED) #endif // VMX86_SERVER /* Reserved guest datagram resource ids. */ #define VMCI_EVENT_HANDLER 0 /* VMCI privileges. */ typedef enum VMCIResourcePrivilegeType { VMCI_PRIV_CH_PRIV, VMCI_PRIV_DESTROY_RESOURCE, VMCI_PRIV_ASSIGN_CLIENT, VMCI_PRIV_DG_CREATE, VMCI_PRIV_DG_SEND, VMCI_PRIV_NOTIFY, VMCI_NUM_PRIVILEGES, } VMCIResourcePrivilegeType; /* * VMCI coarse-grained privileges (per context or host * process/endpoint. An entity with the restricted flag is only * allowed to interact with the hypervisor and trusted entities. */ typedef uint32 VMCIPrivilegeFlags; #define VMCI_PRIVILEGE_FLAG_RESTRICTED 0x01 #define VMCI_PRIVILEGE_FLAG_TRUSTED 0x02 #define VMCI_PRIVILEGE_ALL_FLAGS (VMCI_PRIVILEGE_FLAG_RESTRICTED | \ VMCI_PRIVILEGE_FLAG_TRUSTED) #define VMCI_NO_PRIVILEGE_FLAGS 0x00 #define VMCI_DEFAULT_PROC_PRIVILEGE_FLAGS VMCI_NO_PRIVILEGE_FLAGS #define VMCI_LEAST_PRIVILEGE_FLAGS VMCI_PRIVILEGE_FLAG_RESTRICTED #define VMCI_MAX_PRIVILEGE_FLAGS VMCI_PRIVILEGE_FLAG_TRUSTED #define VMCI_PUBLIC_GROUP_NAME "vmci public group" /* 0 through VMCI_RESERVED_RESOURCE_ID_MAX are reserved. */ #define VMCI_RESERVED_RESOURCE_ID_MAX 1023 #define VMCI_DOMAIN_NAME_MAXLEN 32 #define VMCI_LGPFX "VMCI: " /* * VMCIQueueHeader * * A Queue cannot stand by itself as designed. Each Queue's header * contains a pointer into itself (the producerTail) and into its peer * (consumerHead). The reason for the separation is one of * accessibility: Each end-point can modify two things: where the next * location to enqueue is within its produceQ (producerTail); and * where the next dequeue location is in its consumeQ (consumerHead). * * An end-point cannot modify the pointers of its peer (guest to * guest; NOTE that in the host both queue headers are mapped r/w). * But, each end-point needs read access to both Queue header * structures in order to determine how much space is used (or left) * in the Queue. This is because for an end-point to know how full * its produceQ is, it needs to use the consumerHead that points into * the produceQ but -that- consumerHead is in the Queue header for * that end-points consumeQ. * * Thoroughly confused? Sorry. * * producerTail: the point to enqueue new entrants. When you approach * a line in a store, for example, you walk up to the tail. * * consumerHead: the point in the queue from which the next element is * dequeued. In other words, who is next in line is he who is at the * head of the line. * * Also, producerTail points to an empty byte in the Queue, whereas * consumerHead points to a valid byte of data (unless producerTail == * consumerHead in which case consumerHead does not point to a valid * byte of data). * * For a queue of buffer 'size' bytes, the tail and head pointers will be in * the range [0, size-1]. * * If produceQHeader->producerTail == consumeQHeader->consumerHead * then the produceQ is empty. */ typedef struct VMCIQueueHeader { /* All fields are 64bit and aligned. */ VMCIHandle handle; /* Identifier. */ Atomic_uint64 producerTail; /* Offset in this queue. */ Atomic_uint64 consumerHead; /* Offset in peer queue. */ } VMCIQueueHeader; /* * If one client of a QueuePair is a 32bit entity, we restrict the QueuePair * size to be less than 4GB, and use 32bit atomic operations on the head and * tail pointers. 64bit atomic read on a 32bit entity involves cmpxchg8b which * is an atomic read-modify-write. This will cause traces to fire when a 32bit * consumer tries to read the producer's tail pointer, for example, because the * consumer has read-only access to the producer's tail pointer. * * We provide the following macros to invoke 32bit or 64bit atomic operations * based on the architecture the code is being compiled on. */ /* Architecture independent maximum queue size. */ #define QP_MAX_QUEUE_SIZE_ARCH_ANY CONST64U(0xffffffff) #ifdef __x86_64__ # define QP_MAX_QUEUE_SIZE_ARCH CONST64U(0xffffffffffffffff) # define QPAtomic_ReadOffset(x) Atomic_Read64(x) # define QPAtomic_WriteOffset(x, y) Atomic_Write64(x, y) #else /* * Wrappers below are being used to call Atomic_Read32 because of the * 'type punned' compilation warning received when Atomic_Read32 is * called with a Atomic_uint64 pointer typecasted to Atomic_uint32 * pointer from QPAtomic_ReadOffset. Ditto with QPAtomic_WriteOffset. */ static INLINE uint32 TypeSafe_Atomic_Read32(void *var) // IN: { return Atomic_Read32((Atomic_uint32 *)(var)); } static INLINE void TypeSafe_Atomic_Write32(void *var, uint32 val) // IN: { Atomic_Write32((Atomic_uint32 *)(var), (uint32)(val)); } # define QP_MAX_QUEUE_SIZE_ARCH CONST64U(0xffffffff) # define QPAtomic_ReadOffset(x) TypeSafe_Atomic_Read32((void *)(x)) # define QPAtomic_WriteOffset(x, y) \ TypeSafe_Atomic_Write32((void *)(x), (uint32)(y)) #endif /* __x86_64__ */ /* *----------------------------------------------------------------------------- * * QPAddPointer -- * * Helper to add a given offset to a head or tail pointer. Wraps the value * of the pointer around the max size of the queue. * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE void QPAddPointer(Atomic_uint64 *var, // IN: size_t add, // IN: uint64 size) // IN: { uint64 newVal = QPAtomic_ReadOffset(var); if (newVal >= size - add) { newVal -= size; } newVal += add; QPAtomic_WriteOffset(var, newVal); } /* *----------------------------------------------------------------------------- * * VMCIQueueHeader_ProducerTail() -- * * Helper routine to get the Producer Tail from the supplied queue. * * Results: * The contents of the queue's producer tail. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE uint64 VMCIQueueHeader_ProducerTail(const VMCIQueueHeader *qHeader) // IN: { VMCIQueueHeader *qh = (VMCIQueueHeader *)qHeader; return QPAtomic_ReadOffset(&qh->producerTail); } /* *----------------------------------------------------------------------------- * * VMCIQueueHeader_ConsumerHead() -- * * Helper routine to get the Consumer Head from the supplied queue. * * Results: * The contents of the queue's consumer tail. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE uint64 VMCIQueueHeader_ConsumerHead(const VMCIQueueHeader *qHeader) // IN: { VMCIQueueHeader *qh = (VMCIQueueHeader *)qHeader; return QPAtomic_ReadOffset(&qh->consumerHead); } /* *----------------------------------------------------------------------------- * * VMCIQueueHeader_AddProducerTail() -- * * Helper routine to increment the Producer Tail. Fundamentally, * QPAddPointer() is used to manipulate the tail itself. * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE void VMCIQueueHeader_AddProducerTail(VMCIQueueHeader *qHeader, // IN/OUT: size_t add, // IN: uint64 queueSize) // IN: { QPAddPointer(&qHeader->producerTail, add, queueSize); } /* *----------------------------------------------------------------------------- * * VMCIQueueHeader_AddConsumerHead() -- * * Helper routine to increment the Consumer Head. Fundamentally, * QPAddPointer() is used to manipulate the head itself. * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE void VMCIQueueHeader_AddConsumerHead(VMCIQueueHeader *qHeader, // IN/OUT: size_t add, // IN: uint64 queueSize) // IN: { QPAddPointer(&qHeader->consumerHead, add, queueSize); } /* *----------------------------------------------------------------------------- * * VMCIQueueHeader_CheckAlignment -- * * Checks if the given queue is aligned to page boundary. Returns TRUE if * the alignment is good. * * Results: * TRUE or FALSE. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE Bool VMCIQueueHeader_CheckAlignment(const VMCIQueueHeader *qHeader) // IN: { uintptr_t hdr, offset; hdr = (uintptr_t) qHeader; offset = hdr & (PAGE_SIZE -1); return offset == 0; } /* *----------------------------------------------------------------------------- * * VMCIQueueHeader_GetPointers -- * * Helper routine for getting the head and the tail pointer for a queue. * Both the VMCIQueues are needed to get both the pointers for one queue. * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE void VMCIQueueHeader_GetPointers(const VMCIQueueHeader *produceQHeader, // IN: const VMCIQueueHeader *consumeQHeader, // IN: uint64 *producerTail, // OUT: uint64 *consumerHead) // OUT: { if (producerTail) { *producerTail = VMCIQueueHeader_ProducerTail(produceQHeader); } if (consumerHead) { *consumerHead = VMCIQueueHeader_ConsumerHead(consumeQHeader); } } /* *----------------------------------------------------------------------------- * * VMCIQueueHeader_ResetPointers -- * * Reset the tail pointer (of "this" queue) and the head pointer (of * "peer" queue). * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE void VMCIQueueHeader_ResetPointers(VMCIQueueHeader *qHeader) // IN/OUT: { QPAtomic_WriteOffset(&qHeader->producerTail, CONST64U(0)); QPAtomic_WriteOffset(&qHeader->consumerHead, CONST64U(0)); } /* *----------------------------------------------------------------------------- * * VMCIQueueHeader_Init -- * * Initializes a queue's state (head & tail pointers). * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE void VMCIQueueHeader_Init(VMCIQueueHeader *qHeader, // IN/OUT: const VMCIHandle handle) // IN: { qHeader->handle = handle; VMCIQueueHeader_ResetPointers(qHeader); } /* *----------------------------------------------------------------------------- * * VMCIQueueHeader_FreeSpace -- * * Finds available free space in a produce queue to enqueue more * data or reports an error if queue pair corruption is detected. * * Results: * Free space size in bytes or an error code. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE int64 VMCIQueueHeader_FreeSpace(const VMCIQueueHeader *produceQHeader, // IN: const VMCIQueueHeader *consumeQHeader, // IN: const uint64 produceQSize) // IN: { uint64 tail; uint64 head; uint64 freeSpace; tail = VMCIQueueHeader_ProducerTail(produceQHeader); head = VMCIQueueHeader_ConsumerHead(consumeQHeader); if (tail >= produceQSize || head >= produceQSize) { return VMCI_ERROR_INVALID_SIZE; } /* * Deduct 1 to avoid tail becoming equal to head which causes ambiguity. If * head and tail are equal it means that the queue is empty. */ if (tail >= head) { freeSpace = produceQSize - (tail - head) - 1; } else { freeSpace = head - tail - 1; } return freeSpace; } /* *----------------------------------------------------------------------------- * * VMCIQueueHeader_BufReady -- * * VMCIQueueHeader_FreeSpace() does all the heavy lifting of * determing the number of free bytes in a Queue. This routine, * then subtracts that size from the full size of the Queue so * the caller knows how many bytes are ready to be dequeued. * * Results: * On success, available data size in bytes (up to MAX_INT64). * On failure, appropriate error code. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE int64 VMCIQueueHeader_BufReady(const VMCIQueueHeader *consumeQHeader, // IN: const VMCIQueueHeader *produceQHeader, // IN: const uint64 consumeQSize) // IN: { int64 freeSpace; freeSpace = VMCIQueueHeader_FreeSpace(consumeQHeader, produceQHeader, consumeQSize); if (freeSpace < VMCI_SUCCESS) { return freeSpace; } else { return consumeQSize - freeSpace - 1; } } /* * Defines for the VMCI traffic filter: * - VMCI_FA_<name> defines the filter action values * - VMCI_FP_<name> defines the filter protocol values * - VMCI_FD_<name> defines the direction values (guest or host) * - VMCI_FT_<name> are the type values (allow or deny) */ #define VMCI_FA_INVALID -1 #define VMCI_FA_ALLOW 0 #define VMCI_FA_DENY (VMCI_FA_ALLOW + 1) #define VMCI_FA_MAX (VMCI_FA_DENY + 1) #define VMCI_FP_INVALID -1 #define VMCI_FP_HYPERVISOR 0 #define VMCI_FP_QUEUEPAIR (VMCI_FP_HYPERVISOR + 1) #define VMCI_FP_DOORBELL (VMCI_FP_QUEUEPAIR + 1) #define VMCI_FP_DATAGRAM (VMCI_FP_DOORBELL + 1) #define VMCI_FP_STREAMSOCK (VMCI_FP_DATAGRAM + 1) #define VMCI_FP_ANY (VMCI_FP_STREAMSOCK + 1) #define VMCI_FP_MAX (VMCI_FP_ANY + 1) #define VMCI_FD_INVALID -1 #define VMCI_FD_GUEST 0 #define VMCI_FD_HOST (VMCI_FD_GUEST + 1) #define VMCI_FD_ANY (VMCI_FD_HOST + 1) #define VMCI_FD_MAX (VMCI_FD_ANY + 1) /* * The filter list tracks VMCI Id ranges for a given filter. */ typedef struct { uint32 len; VMCIIdRange *list; } VMCIFilterList; /* * The filter info is used to communicate the filter configuration * from the VMX to the host kernel. */ typedef struct { VA64 list; // List of VMCIIdRange uint32 len; // Length of list uint8 dir; // VMCI_FD_X uint8 proto; // VMCI_FP_X } VMCIFilterInfo; /* * In the host kernel, the ingoing and outgoing filters are * separated. The VMCIProtoFilters type captures all filters in one * direction. The VMCIFilters type captures all filters. */ typedef VMCIFilterList VMCIProtoFilters[VMCI_FP_MAX]; typedef VMCIProtoFilters VMCIFilters[VMCI_FD_MAX]; #if defined __cplusplus } // extern "C" #endif #endif // _VMCI_DEF_H_ vsock-only/shared/compat_version.h 0000444 0000000 0000000 00000007363 13432725347 016350 0 ustar root root /********************************************************* * Copyright (C) 1998 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_VERSION_H__ # define __COMPAT_VERSION_H__ #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMKDRIVERS #include "includeCheck.h" #ifndef __linux__ # error "linux-version.h" #endif #include <linux/version.h> #ifndef KERNEL_VERSION # error KERNEL_VERSION macro is not defined, environment is busted #endif /* * Distinguish relevant classes of Linux kernels. * * The convention is that version X defines all * the KERNEL_Y symbols where Y <= X. * * XXX Do not add more definitions here. This way of doing things does not * scale, and we are going to phase it out soon --hpreg */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 1, 0) # define KERNEL_2_1 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 2, 0) # define KERNEL_2_2 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 3, 1) # define KERNEL_2_3_1 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 3, 15) /* new networking */ # define KERNEL_2_3_15 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 3, 25) /* new procfs */ # define KERNEL_2_3_25 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 3, 29) /* even newer procfs */ # define KERNEL_2_3_29 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 3, 43) /* softnet changes */ # define KERNEL_2_3_43 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 3, 47) /* more softnet changes */ # define KERNEL_2_3_47 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 3, 99) /* name in netdevice struct is array and not pointer */ # define KERNEL_2_3_99 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 4, 0) /* New 'owner' member at the beginning of struct file_operations */ # define KERNEL_2_4_0 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 4, 8) /* New netif_rx_ni() --hpreg */ # define KERNEL_2_4_8 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 5, 2) /* New kdev_t, major()/minor() API --hpreg */ # define KERNEL_2_5_2 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 5, 5) /* New sk_alloc(), pte_offset_map()/pte_unmap() --hpreg */ # define KERNEL_2_5_5 #endif /* Linux kernel 3.0 can be called 2.6.40, and 3.1 can be 2.6.41... * Use COMPAT_LINUX_VERSION_CHECK_LT iff you need to compare running kernel to * versions 3.0 and above. * */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 0, 0) /* Straight forward comparison if kernel version is 3.0.0 and beyond */ # define COMPAT_LINUX_VERSION_CHECK_LT(a, b, c) LINUX_VERSION_CODE < KERNEL_VERSION (a, b, c) #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 40) /* Use b of the check to calculate corresponding c of kernel * version to compare */ # define COMPAT_LINUX_VERSION_CHECK_LT(a, b, c) LINUX_VERSION_CODE < KERNEL_VERSION (2, 6, (b + 40)) #else /* This is anyways lesser than any 3.x versions */ # define COMPAT_LINUX_VERSION_CHECK_LT(a, b, c) 1 #endif #endif /* __COMPAT_VERSION_H__ */ vsock-only/shared/compat_pgtable.h 0000444 0000000 0000000 00000005563 13432725347 016301 0 ustar root root /********************************************************* * Copyright (C) 2002-2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_PGTABLE_H__ # define __COMPAT_PGTABLE_H__ #if defined(CONFIG_PARAVIRT) && defined(CONFIG_HIGHPTE) # if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 21) # include <asm/paravirt.h> # undef paravirt_map_pt_hook # define paravirt_map_pt_hook(type, va, pfn) do {} while (0) # endif #endif #include <asm/pgtable.h> /* * p4d level appeared in 4.12. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 12, 0) # define compat_p4d_offset(pgd, address) p4d_offset(pgd, address) # define compat_p4d_present(p4d) p4d_present(p4d) # define compat_p4d_large(p4d) p4d_large(p4d) # define compat_p4d_pfn(p4d) p4d_pfn(p4d) # define COMPAT_P4D_MASK P4D_MASK typedef p4d_t compat_p4d_t; #else # define compat_p4d_offset(pgd, address) (pgd) # define compat_p4d_present(p4d) (1) # define compat_p4d_large(p4d) (0) # define compat_p4d_pfn(p4d) INVALID_MPN /* Not used */ # define COMPAT_P4D_MASK 0 /* Not used */ typedef pgd_t compat_p4d_t; #endif /* pud_pfn did not exist before 3.8. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(3, 8, 0) # define pud_pfn(pud) INVALID_MPN #endif /* * Define VM_PAGE_KERNEL_EXEC for vmapping executable pages. * * On ia32 PAGE_KERNEL_EXEC was introduced in 2.6.8.1. Unfortunately it accesses * __PAGE_KERNEL_EXEC which is not exported for modules. So we use * __PAGE_KERNEL and just cut _PAGE_NX bit from it. * * For ia32 kernels before 2.6.8.1 we use PAGE_KERNEL directly, these kernels * do not have noexec support. * * On x86-64 situation is a bit better: they always supported noexec, but * before 2.6.8.1 flag was named PAGE_KERNEL_EXECUTABLE, and it was renamed * to PAGE_KERNEL_EXEC when ia32 got noexec too (see above). */ #ifdef CONFIG_X86 #ifdef _PAGE_NX #define VM_PAGE_KERNEL_EXEC __pgprot(__PAGE_KERNEL & ~_PAGE_NX) #else #define VM_PAGE_KERNEL_EXEC PAGE_KERNEL #endif #else #define VM_PAGE_KERNEL_EXEC PAGE_KERNEL_EXEC #endif #endif /* __COMPAT_PGTABLE_H__ */ vsock-only/shared/compat_page.h 0000444 0000000 0000000 00000004663 13432725347 015577 0 ustar root root /********************************************************* * Copyright (C) 2002 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_PAGE_H__ # define __COMPAT_PAGE_H__ #include <linux/mm.h> #include <asm/page.h> /* The pfn_to_page() API appeared in 2.5.14 and changed to function during 2.6.x */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 0) && !defined(pfn_to_page) # define pfn_to_page(_pfn) (mem_map + (_pfn)) # define page_to_pfn(_page) ((_page) - mem_map) #endif /* The virt_to_page() API appeared in 2.4.0 --hpreg */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 0) && !defined(virt_to_page) # define virt_to_page(_kvAddr) pfn_to_page(MAP_NR(_kvAddr)) #endif /* * The get_order() API appeared at some point in 2.3.x, and was then backported * in 2.2.17-21mdk and in the stock 2.2.18. Because we can only detect its * definition through makefile tricks, we provide our own for now --hpreg */ static inline int compat_get_order(unsigned long size) // IN { int order; size = (size - 1) >> (PAGE_SHIFT - 1); order = -1; do { size >>= 1; order++; } while (size); return order; } /* * BUG() was added to <asm/page.h> in 2.2.18, and was moved to <asm/bug.h> * in 2.5.58. * * XXX: Technically, this belongs in some sort of "compat_asm_page.h" file, but * since our compatibility wrappers don't distinguish between <asm/xxx.h> and * <linux/xxx.h>, putting it here is reasonable. */ #ifndef BUG #define BUG() do { \ printk("kernel BUG at %s:%d!\n", __FILE__, __LINE__); \ __asm__ __volatile__(".byte 0x0f,0x0b"); \ } while (0) #endif #endif /* __COMPAT_PAGE_H__ */ vsock-only/shared/includeCheck.h 0000444 0000000 0000000 00000007463 13432725350 015674 0 ustar root root /********************************************************* * Copyright (C) 1998-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * includeCheck.h -- * * Restrict include file use. * * In every .h file, define one or more of these * * INCLUDE_ALLOW_VMX * INCLUDE_ALLOW_USERLEVEL * INCLUDE_ALLOW_VMCORE * INCLUDE_ALLOW_MODULE * INCLUDE_ALLOW_VMKERNEL * INCLUDE_ALLOW_DISTRIBUTE * INCLUDE_ALLOW_VMK_MODULE * INCLUDE_ALLOW_VMKDRIVERS * INCLUDE_ALLOW_MKS * * Then include this file. * * Any file that has INCLUDE_ALLOW_DISTRIBUTE defined will potentially * be distributed in source form along with GPLed code. Ensure * that this is acceptable. */ /* * Declare a VMCORE-only variable to help classify object * files. The variable goes in the common block and does * not create multiple definition link-time conflicts. */ #if defined VMCORE && defined VMX86_DEVEL && defined VMX86_DEBUG && \ defined linux && !defined MODULE && \ !defined COMPILED_WITH_VMCORE #define COMPILED_WITH_VMCORE compiled_with_vmcore #ifdef ASM .comm compiled_with_vmcore, 0 #else asm(".comm compiled_with_vmcore, 0"); #endif /* ASM */ #endif #if defined VMCORE && \ !(defined VMX86_VMX || defined VMM || \ defined MONITOR_APP || defined VMMON) #error "Makefile problem: VMCORE without VMX86_VMX or \ VMM or MONITOR_APP or MODULE." #endif #if defined VMCORE && !defined INCLUDE_ALLOW_VMCORE #error "The surrounding include file is not allowed in vmcore." #endif #undef INCLUDE_ALLOW_VMCORE #if defined VMX86_VMX && !defined VMCORE && \ !defined INCLUDE_ALLOW_VMX && !defined INCLUDE_ALLOW_USERLEVEL && \ !defined INCLUDE_ALLOW_MKS #error "The surrounding include file is not allowed in the VMX." #endif #undef INCLUDE_ALLOW_VMX #if defined USERLEVEL && !defined VMX86_VMX && !defined VMCORE && \ !defined INCLUDE_ALLOW_USERLEVEL && !defined INCLUDE_ALLOW_MKS #error "The surrounding include file is not allowed at userlevel." #endif #undef INCLUDE_ALLOW_USERLEVEL #if defined MODULE && !defined VMKERNEL_MODULE && \ !defined VMMON && !defined INCLUDE_ALLOW_MODULE #error "The surrounding include file is not allowed in driver modules." #endif #undef INCLUDE_ALLOW_MODULE #if defined VMMON && !defined INCLUDE_ALLOW_VMMON #error "The surrounding include file is not allowed in vmmon." #endif #undef INCLUDE_ALLOW_VMMON #if defined VMKERNEL && !defined INCLUDE_ALLOW_VMKERNEL #error "The surrounding include file is not allowed in the vmkernel." #endif #undef INCLUDE_ALLOW_VMKERNEL #if defined GPLED_CODE && !defined INCLUDE_ALLOW_DISTRIBUTE #error "The surrounding include file is not allowed in GPL code." #endif #undef INCLUDE_ALLOW_DISTRIBUTE #if defined VMKERNEL_MODULE && !defined VMKERNEL && \ !defined INCLUDE_ALLOW_VMK_MODULE && !defined INCLUDE_ALLOW_VMKDRIVERS #error "The surrounding include file is not allowed in vmkernel modules." #endif #undef INCLUDE_ALLOW_VMK_MODULE #undef INCLUDE_ALLOW_VMKDRIVERS #if defined INCLUDE_ALLOW_MKS && !(defined COREMKS) #error "The surrounding include file is not allowed outside of the MKS." #endif #undef INCLUDE_ALLOW_MKS vsock-only/shared/compat_pci.h 0000444 0000000 0000000 00000005152 13432725347 015430 0 ustar root root /********************************************************* * Copyright (C) 1999 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * compat_pci.h: PCI compatibility wrappers. */ #ifndef __COMPAT_PCI_H__ #define __COMPAT_PCI_H__ #include "compat_ioport.h" #include <linux/pci.h> #ifndef DMA_BIT_MASK # define DMA_BIT_MASK(n) DMA_##n##BIT_MASK #endif /* * Power Management related compat wrappers. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 10) # define compat_pci_save_state(pdev) pci_save_state((pdev), NULL) # define compat_pci_restore_state(pdev) pci_restore_state((pdev), NULL) #else # define compat_pci_save_state(pdev) pci_save_state((pdev)) # define compat_pci_restore_state(pdev) pci_restore_state((pdev)) #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 11) # define pm_message_t u32 # define compat_pci_choose_state(pdev, state) (state) # define PCI_D0 0 # define PCI_D3hot 3 #else # define compat_pci_choose_state(pdev, state) pci_choose_state((pdev), (state)) #endif /* 2.6.14 changed the PCI shutdown callback */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 14) # define COMPAT_PCI_SHUTDOWN(func) .driver = { .shutdown = (func), } # define COMPAT_PCI_DECLARE_SHUTDOWN(func, var) (func)(struct device *(var)) # define COMPAT_PCI_TO_DEV(dev) (to_pci_dev(dev)) #else # define COMPAT_PCI_SHUTDOWN(func) .shutdown = (func) # define COMPAT_PCI_DECLARE_SHUTDOWN(func, var) (func)(struct pci_dev *(var)) # define COMPAT_PCI_TO_DEV(dev) (dev) #endif /* 2.6.26 introduced the device_set_wakeup_enable() function */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 26) # define compat_device_set_wakeup_enable(dev, val) do {} while(0) #else # define compat_device_set_wakeup_enable(dev, val) \ device_set_wakeup_enable(dev, val) #endif #endif /* __COMPAT_PCI_H__ */ vsock-only/shared/compat_log2.h 0000444 0000000 0000000 00000003672 13432725347 015525 0 ustar root root /********************************************************* * Copyright (C) 2011 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_LOG2_H__ # define __COMPAT_LOG2_H__ #ifndef LINUX_VERSION_CODE # error "Include compat_version.h before compat_log2.h" #endif /* linux/log2.h was introduced in 2.6.20. */ #if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 19) # include <linux/log2.h> #endif /* * is_power_of_2 was introduced in 2.6.21. This implementation is almost * identical to the one found there. */ #if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 20) #define compat_is_power_of_2(n) is_power_of_2(n) #else static inline __attribute__((const)) int compat_is_power_of_2(unsigned long n) { return (n != 0 && ((n && (n - 1)) == 0)); } #endif /* * rounddown_power_of_two was introduced in 2.6.24. This implementation is * similar to the one in log2.h but with input of int instead of long to * avoid more version related checks for fls_long(). */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 24) #define compat_rounddown_pow_of_two(n) rounddown_pow_of_two(n) #else static inline __attribute__((const)) unsigned int compat_rounddown_pow_of_two(unsigned int n) { return 1U << (fls(n) -1); } #endif #endif /* __COMPAT_LOG2_H__ */ vsock-only/shared/compat_cred.h 0000444 0000000 0000000 00000003244 13432725347 015572 0 ustar root root /********************************************************* * Copyright (C) 2002 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_CRED_H__ # define __COMPAT_CRED_H__ /* * Include linux/cred.h via linux/sched.h - it is not nice, but * as cpp does not have #ifexist... */ #if LINUX_VERSION_CODE < KERNEL_VERSION(4, 11, 0) #include <linux/sched.h> #else #include <linux/cred.h> #endif #if !defined(current_fsuid) && LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 29) #define current_uid() (current->uid) #define current_euid() (current->euid) #define current_fsuid() (current->fsuid) #define current_gid() (current->gid) #define current_egid() (current->egid) #define current_fsgid() (current->fsgid) #endif #if !defined(cap_set_full) /* cap_set_full was removed in kernel version 3.0-rc4. */ #define cap_set_full(_c) do { (_c) = CAP_FULL_SET; } while (0) #endif #if !defined(GLOBAL_ROOT_UID) #define GLOBAL_ROOT_UID (0) #endif #endif /* __COMPAT_CRED_H__ */ vsock-only/shared/compat_ioport.h 0000444 0000000 0000000 00000004041 13432725347 016165 0 ustar root root /********************************************************* * Copyright (C) 2003 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_IOPORT_H__ # define __COMPAT_IOPORT_H__ #include <linux/ioport.h> #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 0) static inline void * compat_request_region(unsigned long start, unsigned long len, const char *name) { if (check_region(start, len)) { return NULL; } request_region(start, len, name); return (void*)1; } #else #define compat_request_region(start, len, name) request_region(start, len, name) #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 3, 7) /* mmap io support starts from 2.3.7, fail the call for kernel prior to that */ static inline void * compat_request_mem_region(unsigned long start, unsigned long len, const char *name) { return NULL; } static inline void compat_release_mem_region(unsigned long start, unsigned long len) { return; } #else #define compat_request_mem_region(start, len, name) request_mem_region(start, len, name) #define compat_release_mem_region(start, len) release_mem_region(start, len) #endif /* these two macro defs are needed by compat_pci_request_region */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 3, 15) # define IORESOURCE_IO 0x00000100 # define IORESOURCE_MEM 0x00000200 #endif #endif /* __COMPAT_IOPORT_H__ */ vsock-only/shared/vm_atomic.h 0000444 0000000 0000000 00000305435 13432725350 015271 0 ustar root root /********************************************************* * Copyright (C) 1998-2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vm_atomic.h -- * * Atomic power * * Note: Only partially tested on ARM processors: Works for View Open * Client, which shouldn't have threads, and ARMv8 processors. * * In ARM, GCC intrinsics (__sync*) compile but might not * work, while MS intrinsics (_Interlocked*) do not compile. */ #ifndef _ATOMIC_H_ #define _ATOMIC_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMKDRIVERS #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" #include "vm_basic_types.h" #include "vm_assert.h" #if defined __cplusplus extern "C" { #endif /* * There are two concepts involved when dealing with atomic accesses: * 1. Atomicity of the access itself * 2. Ordering of the access with respect to other reads&writes (from the view * of other processors/devices). * * Two examples help to clarify #2: * a. Inc: A caller implementing a simple independent global event counter * might not care if the compiler or processor visibly reorders the * increment around other memory accesses. * b. Dec: A caller implementing a reference count absolutely *doesn't* want * the compiler or processor to visibly reordering writes after that * decrement: if that happened, the program could then end up writing * to memory that was freed by another processor. * * C11 has standardized a good model for expressing these orderings when doing * atomics. It defines three *tiers* of ordering: * 1. Sequential Consistency (every processor sees the same total order of * events) * * 2. Acquire/Release ordering (roughly, everybody can agree previous events * have completed, but they might disagree on the ordering of previous * independent events). * * The relative ordering provided by this tier is sufficient for common * locking and initialization activities, but is insufficient for unusual * synchronization schemes (e.g. IRIW aka Independent Read Independent * Write designs such Dekker's algorithm, Peterson's algorithm, etc.) * * In other words, this tier is close in behavior to Sequential Consistency * in much the same way a General-Relativity universe is close to a * Newtonian universe. * 3. Relaxed (i.e unordered/unfenced) * * In C11 standard's terminology for atomic memory ordering, * - in case (a) we want "relaxed" ordering for perf and, * - in case (b) we want "sequentially consistent" ordering (or perhaps the * only slightly weaker "release" ordering) for correctness. * * There are standardized mappings of operations to orderings for every * processor architecture. See * - https://www.cl.cam.ac.uk/~pes20/cpp/cpp0xmappings.html * - http://preshing.com/20120913/acquire-and-release-semantics/ * * In this file: * 1. all RMW (Read/Modify/Write) operations are sequentially consistent. * This includes operations like Atomic_IncN, Atomic_ReadIfEqualWriteN, * Atomic_ReadWriteN, etc. * 2. all R and W operations are relaxed. This includes operations like * Atomic_WriteN, Atomic_ReadN, Atomic_TestBitN, etc. * * The below routines of course ensure both the CPU and compiler honor the * ordering constraint. * * Notes: * 1. Since R-only and W-only operations do not provide ordering, callers * using them for synchronizing operations like double-checked * initialization or releasing spinlocks must provide extra barriers. * 2. This implementation of Atomic operations is suboptimal. On x86,simple * reads and writes have acquire/release semantics at the hardware level. * On arm64, we have separate instructions for sequentially consistent * reads and writes (the same instructions are used for acquire/release). * Neither of these are exposed for R-only or W-only callers. * * For further details on x86 and ARM memory ordering see * https://wiki.eng.vmware.com/ARM/MemoryOrdering. */ #ifdef VM_ARM_64 # include "vm_atomic_arm64_begin.h" #endif /* Basic atomic types: 8, 16, 32, 64 and 128 bits */ typedef struct Atomic_uint8 { volatile uint8 value; } Atomic_uint8 ALIGNED(1); typedef struct Atomic_uint16 { volatile uint16 value; } Atomic_uint16 ALIGNED(2); typedef struct Atomic_uint32 { volatile uint32 value; } Atomic_uint32 ALIGNED(4); typedef struct Atomic_uint64 { volatile uint64 value; } Atomic_uint64 ALIGNED(8); #if defined __GNUC__ && defined VM_64BIT && \ (defined __GCC_HAVE_SYNC_COMPARE_AND_SWAP_16 || defined VM_ARM_64) typedef struct Atomic_uint128 { volatile uint128 value; } Atomic_uint128 ALIGNED(16); #endif /* * Prototypes for msft atomics. These are defined & inlined by the * compiler so no function definition is needed. The prototypes are * needed for C++. * * The declarations for the intrinsic functions were taken from ntddk.h * in the DDK. The declarations must match otherwise the 64-bit C++ * compiler will complain about second linkage of the intrinsic functions. * We define the intrinsic using the basic types corresponding to the * Windows typedefs. This avoids having to include windows header files * to get to the windows types. */ #if defined _MSC_VER && !defined BORA_NO_WIN32_INTRINS #ifdef __cplusplus extern "C" { #endif long _InterlockedExchange(long volatile*, long); long _InterlockedCompareExchange(long volatile*, long, long); long _InterlockedExchangeAdd(long volatile*, long); long _InterlockedDecrement(long volatile*); long _InterlockedIncrement(long volatile*); __int64 _InterlockedCompareExchange64(__int64 volatile*, __int64, __int64); #pragma intrinsic(_InterlockedExchange, _InterlockedCompareExchange) #pragma intrinsic(_InterlockedExchangeAdd, _InterlockedDecrement) #pragma intrinsic(_InterlockedIncrement) #pragma intrinsic(_InterlockedCompareExchange64) # if _MSC_VER >= 1600 char _InterlockedExchange8(char volatile *, char); char _InterlockedCompareExchange8(char volatile *, char, char); #pragma intrinsic(_InterlockedCompareExchange8, _InterlockedCompareExchange8) #endif #if defined VM_X86_64 long _InterlockedAnd(long volatile*, long); __int64 _InterlockedAnd64(__int64 volatile*, __int64); long _InterlockedOr(long volatile*, long); __int64 _InterlockedOr64(__int64 volatile*, __int64); long _InterlockedXor(long volatile*, long); __int64 _InterlockedXor64(__int64 volatile*, __int64); __int64 _InterlockedExchangeAdd64(__int64 volatile*, __int64); __int64 _InterlockedIncrement64(__int64 volatile*); __int64 _InterlockedDecrement64(__int64 volatile*); __int64 _InterlockedExchange64(__int64 volatile*, __int64); #if !defined _WIN64 #pragma intrinsic(_InterlockedAnd, _InterlockedAnd64) #pragma intrinsic(_InterlockedOr, _InterlockedOr64) #pragma intrinsic(_InterlockedXor, _InterlockedXor64) #pragma intrinsic(_InterlockedExchangeAdd64, _InterlockedIncrement64) #pragma intrinsic(_InterlockedDecrement64, _InterlockedExchange64) #endif /* !_WIN64 */ #endif /* __x86_64__ */ #ifdef __cplusplus } #endif #endif /* _MSC_VER */ #if defined __arm__ /* * LDREX without STREX or CLREX may cause problems in environments where the * context switch may not clear the reference monitor - according ARM manual * the reference monitor should be cleared after a context switch, but some * may not like Linux kernel's non-preemptive context switch path. So use of * ARM routines in kernel code may not be safe. */ # if defined __ARM_ARCH_7__ || defined __ARM_ARCH_7A__ || \ defined __ARM_ARCH_7R__|| defined __ARM_ARCH_7M__ # define VM_ARM_V7 # ifdef __KERNEL__ # warning LDREX/STREX may not be safe in linux kernel, since it \ does not issue CLREX on context switch (as of 2011-09-29). # endif # else # error Only ARMv7 extends the synchronization primitives ldrex/strex. \ For the lower ARM version, please implement the atomic functions \ by kernel APIs. # endif #endif /* Data Memory Barrier */ #ifdef VM_ARM_V7 #define dmb() __asm__ __volatile__("dmb" : : : "memory") #endif /* Convert a volatile uint32 to Atomic_uint32. */ static INLINE Atomic_uint32 * Atomic_VolatileToAtomic32(volatile uint32 *var) // IN: { return (Atomic_uint32 *)var; } #define Atomic_VolatileToAtomic Atomic_VolatileToAtomic32 /* Convert a volatile uint64 to Atomic_uint64. */ static INLINE Atomic_uint64 * Atomic_VolatileToAtomic64(volatile uint64 *var) // IN: { return (Atomic_uint64 *)var; } /* * The Read/Modify/Write operations on x86/x64 are all written using the * "memory" constraint. This is to ensure the compiler treats the operation as * a full barrier, flushing any pending/cached state currently residing in * registers. */ #if defined _MSC_VER && _MSC_VER < 1600 && defined __x86_64__ uint8 VMWInterlockedExchange8(uint8 volatile *ptr, uint8 val); uint8 VMWInterlockedCompareExchange8(uint8 volatile *ptr, uint8 newVal, uint8 oldVal); #endif #if defined __GNUC__ && defined VM_ARM_32 /* Force the link step to fail for unimplemented functions. */ extern int AtomicUndefined(void const *); #endif /* *----------------------------------------------------------------------------- * * Atomic_ReadIfEqualWrite128 -- * * Compare and exchange a 16 byte tuple. * * Results: * old value * * Side effects: * None * *----------------------------------------------------------------------------- */ #if defined __GNUC__ && defined VM_64BIT && \ (defined __GCC_HAVE_SYNC_COMPARE_AND_SWAP_16 || defined VM_ARM_64) static INLINE uint128 Atomic_ReadIfEqualWrite128(Atomic_uint128 *ptr, // IN/OUT uint128 oldVal, // IN uint128 newVal) // IN { #ifdef __GCC_HAVE_SYNC_COMPARE_AND_SWAP_16 return __sync_val_compare_and_swap(&ptr->value, oldVal, newVal); #elif defined VM_ARM_64 union { uint128 raw; struct { uint64 lo; uint64 hi; }; } res, _old = { oldVal }, _new = { newVal }; uint32 failed; SMP_RW_BARRIER_RW(); __asm__ __volatile__( "1: ldxp %x0, %x1, %3 \n\t" " cmp %x0, %x4 \n\t" " ccmp %x1, %x5, #0, eq \n\t" " b.ne 2f \n\t" " stxp %w2, %x6, %x7, %3 \n\t" " cbnz %w2, 1b \n\t" "2: \n\t" : "=&r" (res.lo), "=&r" (res.hi), "=&r" (failed), "+Q" (ptr->value) : "r" (_old.lo), "r" (_old.hi), "r" (_new.lo), "r" (_new.hi) : "cc" ); SMP_RW_BARRIER_RW(); return res.raw; #endif } #endif /* *----------------------------------------------------------------------------- * * Atomic_Read8 -- * * Read the value of the specified object atomically. * * Results: * The value of the atomic variable. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE uint8 Atomic_Read8(Atomic_uint8 const *var) // IN: { uint8 val; #if defined __GNUC__ && defined VM_ARM_32 val = AtomicUndefined(var); #elif defined VM_ARM_64 val = _VMATOM_X(R, 8, &var->value); #elif defined __GNUC__ && (defined __i386__ || defined __x86_64__) __asm__ __volatile__( "movb %1, %0" : "=q" (val) : "m" (var->value) ); #elif defined _MSC_VER val = var->value; #else #error No compiler defined for Atomic_Read8 #endif return val; } /* *----------------------------------------------------------------------------- * * Atomic_ReadWrite8 -- * * Read followed by write. * * Results: * The value of the atomic variable before the write. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint8 Atomic_ReadWrite8(Atomic_uint8 *var, // IN/OUT: uint8 val) // IN: { #if defined __GNUC__ && defined VM_ARM_32 return AtomicUndefined(var + val); #elif defined VM_ARM_64 return _VMATOM_X(RW, 8, TRUE, &var->value, val); #elif defined __GNUC__ && (defined __i386__ || defined __x86_64__) __asm__ __volatile__( "xchgb %0, %1" : "=q" (val), "+m" (var->value) : "0" (val) : "memory" ); return val; #elif defined _MSC_VER && _MSC_VER >= 1600 return _InterlockedExchange8((volatile char *)&var->value, val); #elif defined _MSC_VER && defined __i386__ #pragma warning(push) #pragma warning(disable : 4035) // disable no-return warning { __asm movzx eax, val __asm mov ebx, var __asm xchg [ebx]Atomic_uint8.value, al } #pragma warning(pop) #elif defined _MSC_VER && defined __x86_64__ return VMWInterlockedExchange8(&var->value, val); #else #error No compiler defined for Atomic_ReadWrite8 #endif } /* *----------------------------------------------------------------------------- * * Atomic_Write8 -- * * Write the specified value to the specified object atomically. * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Write8(Atomic_uint8 *var, // IN/OUT: uint8 val) // IN: { #if defined __GNUC__ && defined VM_ARM_32 AtomicUndefined(var + val); #elif defined VM_ARM_64 _VMATOM_X(W, 8, &var->value, val); #elif defined __GNUC__ && (defined __i386__ || defined __x86_64__) __asm__ __volatile__( "movb %1, %0" : "=m" (var->value) : "qn" (val) ); #elif defined _MSC_VER var->value = val; #else #error No compiler defined for Atomic_Write8 #endif } /* *----------------------------------------------------------------------------- * * Atomic_ReadIfEqualWrite8 -- * * Compare exchange: Read variable, if equal to oldVal, write newVal. * * Results: * The value of the atomic variable before the write. * * Side effects: * The variable may be modified. * *----------------------------------------------------------------------------- */ static INLINE uint8 Atomic_ReadIfEqualWrite8(Atomic_uint8 *var, // IN/OUT: uint8 oldVal, // IN: uint8 newVal) // IN: { #if defined __GNUC__ && defined VM_ARM_32 return AtomicUndefined(var + oldVal + newVal); #elif defined VM_ARM_64 return _VMATOM_X(RIFEQW, 8, TRUE, &var->value, oldVal, newVal); #elif defined __GNUC__ && (defined __i386__ || defined __x86_64__) uint8 val; __asm__ __volatile__( "lock; cmpxchgb %2, %1" : "=a" (val), "+m" (var->value) : "q" (newVal), "0" (oldVal) : "cc", "memory" ); return val; #elif defined _MSC_VER && _MSC_VER >= 1600 return _InterlockedCompareExchange8((volatile char *)&var->value, newVal, oldVal); #elif defined _MSC_VER && defined __i386__ #pragma warning(push) #pragma warning(disable : 4035) // disable no-return warning { __asm mov al, oldVal __asm mov ebx, var __asm mov cl, newVal __asm lock cmpxchg [ebx]Atomic_uint8.value, cl __asm movzx eax, al // eax is the return value, this is documented to work - edward } #pragma warning(pop) #elif defined _MSC_VER && defined __x86_64__ return VMWInterlockedCompareExchange8(&var->value, newVal, oldVal); #else #error No compiler defined for Atomic_ReadIfEqualWrite8 #endif } /* *----------------------------------------------------------------------------- * * Atomic_ReadAnd8 -- * * Atomic read (returned), bitwise AND with a value, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint8 Atomic_ReadAnd8(Atomic_uint8 *var, // IN/OUT uint8 val) // IN { uint8 res; #if defined VM_ARM_64 res = _VMATOM_X(ROP, 8, TRUE, &var->value, and, val); #else do { res = Atomic_Read8(var); } while (res != Atomic_ReadIfEqualWrite8(var, res, res & val)); #endif return res; } /* *----------------------------------------------------------------------------- * * Atomic_And8 -- * * Atomic read, bitwise AND with a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_And8(Atomic_uint8 *var, // IN/OUT uint8 val) // IN { #if defined VM_ARM_64 _VMATOM_X(OP, 8, TRUE, &var->value, and, val); #else (void)Atomic_ReadAnd8(var, val); #endif } /* *----------------------------------------------------------------------------- * * Atomic_ReadOr8 -- * * Atomic read (returned), bitwise OR with a value, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint8 Atomic_ReadOr8(Atomic_uint8 *var, // IN/OUT uint8 val) // IN { uint8 res; #if defined VM_ARM_64 res = _VMATOM_X(ROP, 8, TRUE, &var->value, orr, val); #else do { res = Atomic_Read8(var); } while (res != Atomic_ReadIfEqualWrite8(var, res, res | val)); #endif return res; } /* *----------------------------------------------------------------------------- * * Atomic_Or8 -- * * Atomic read, bitwise OR with a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Or8(Atomic_uint8 *var, // IN/OUT uint8 val) // IN { #if defined VM_ARM_64 _VMATOM_X(OP, 8, TRUE, &var->value, orr, val); #else (void)Atomic_ReadOr8(var, val); #endif } /* *----------------------------------------------------------------------------- * * Atomic_ReadXor8 -- * * Atomic read (returned), bitwise XOR with a value, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint8 Atomic_ReadXor8(Atomic_uint8 *var, // IN/OUT uint8 val) // IN { uint8 res; #if defined VM_ARM_64 res = _VMATOM_X(ROP, 8, TRUE, &var->value, eor, val); #else do { res = Atomic_Read8(var); } while (res != Atomic_ReadIfEqualWrite8(var, res, res ^ val)); #endif return res; } /* *----------------------------------------------------------------------------- * * Atomic_Xor8 -- * * Atomic read, bitwise XOR with a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Xor8(Atomic_uint8 *var, // IN/OUT uint8 val) // IN { #if defined VM_ARM_64 _VMATOM_X(OP, 8, TRUE, &var->value, eor, val); #else (void)Atomic_ReadXor8(var, val); #endif } /* *----------------------------------------------------------------------------- * * Atomic_ReadAdd8 -- * * Atomic read (returned), add a value, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint8 Atomic_ReadAdd8(Atomic_uint8 *var, // IN/OUT uint8 val) // IN { uint8 res; #if defined VM_ARM_64 res = _VMATOM_X(ROP, 8, TRUE, &var->value, add, val); #else do { res = Atomic_Read8(var); } while (res != Atomic_ReadIfEqualWrite8(var, res, res + val)); #endif return res; } /* *----------------------------------------------------------------------------- * * Atomic_Add8 -- * * Atomic read, add a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Add8(Atomic_uint8 *var, // IN/OUT uint8 val) // IN { #if defined VM_ARM_64 _VMATOM_X(OP, 8, TRUE, &var->value, add, val); #else (void)Atomic_ReadAdd8(var, val); #endif } /* *----------------------------------------------------------------------------- * * Atomic_Sub8 -- * * Atomic read, subtract a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Sub8(Atomic_uint8 *var, // IN/OUT uint8 val) // IN { #if defined VM_ARM_64 _VMATOM_X(OP, 8, TRUE, &var->value, sub, val); #else Atomic_Add8(var, -val); #endif } /* *----------------------------------------------------------------------------- * * Atomic_Inc8 -- * * Atomic read, increment, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Inc8(Atomic_uint8 *var) // IN/OUT { Atomic_Add8(var, 1); } /* *----------------------------------------------------------------------------- * * Atomic_Dec8 -- * * Atomic read, decrement, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Dec8(Atomic_uint8 *var) // IN/OUT { Atomic_Sub8(var, 1); } /* *----------------------------------------------------------------------------- * * Atomic_ReadInc8 -- * * Atomic read (returned), increment, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint8 Atomic_ReadInc8(Atomic_uint8 *var) // IN/OUT { return Atomic_ReadAdd8(var, 1); } /* *----------------------------------------------------------------------------- * * Atomic_ReadDec8 -- * * Atomic read (returned), decrement, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint8 Atomic_ReadDec8(Atomic_uint8 *var) // IN/OUT { return Atomic_ReadAdd8(var, (uint8)-1); } /* *----------------------------------------------------------------------------- * * Atomic_Read32 -- * * Read * * Results: * The value of the atomic variable. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE uint32 Atomic_Read32(Atomic_uint32 const *var) // IN { uint32 value; #if defined VMM || defined VM_ARM_64 || defined VMKERNEL || defined VMKERNEL_MODULE ASSERT(((uintptr_t)var % 4) == 0); #endif #if defined __GNUC__ /* * Use inline assembler to force using a single load instruction to * ensure that the compiler doesn't split a transfer operation into multiple * instructions. */ #if defined VM_ARM_32 __asm__ __volatile__( "ldr %0, [%1]" : "=r" (value) : "r" (&var->value) ); #elif defined VM_ARM_64 value = _VMATOM_X(R, 32, &var->value); #else __asm__ __volatile__( "mov %1, %0" : "=r" (value) : "m" (var->value) ); #endif #elif defined _MSC_VER /* * Microsoft docs guarantee simple reads and writes to properly * aligned 32-bit variables use only a single instruction. * http://msdn.microsoft.com/en-us/library/ms684122%28VS.85%29.aspx */ value = var->value; #else #error No compiler defined for Atomic_Read #endif return value; } #define Atomic_Read Atomic_Read32 /* *----------------------------------------------------------------------------- * * Atomic_ReadWrite32 -- * * Read followed by write * * Results: * The value of the atomic variable before the write. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint32 Atomic_ReadWrite32(Atomic_uint32 *var, // IN/OUT uint32 val) // IN { #if defined __GNUC__ #ifdef VM_ARM_V7 register volatile uint32 retVal; register volatile uint32 res; dmb(); __asm__ __volatile__( "1: ldrex %[retVal], [%[var]] \n\t" "strex %[res], %[val], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [retVal] "=&r" (retVal), [res] "=&r" (res) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); return retVal; #elif defined VM_ARM_64 return _VMATOM_X(RW, 32, TRUE, &var->value, val); #else /* VM_X86_ANY */ /* Checked against the Intel manual and GCC --walken */ __asm__ __volatile__( "xchgl %0, %1" : "=r" (val), "+m" (var->value) : "0" (val) : "memory" ); return val; #endif /* VM_X86_ANY */ #elif defined _MSC_VER return _InterlockedExchange((long *)&var->value, (long)val); #else #error No compiler defined for Atomic_ReadWrite #endif // __GNUC__ } #define Atomic_ReadWrite Atomic_ReadWrite32 /* *----------------------------------------------------------------------------- * * Atomic_Write32 -- * * Write * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Write32(Atomic_uint32 *var, // OUT uint32 val) // IN { #if defined VMM || defined VM_ARM_64 || defined VMKERNEL || defined VMKERNEL_MODULE ASSERT(((uintptr_t)var % 4) == 0); #endif #if defined __GNUC__ #if defined VM_ARM_64 _VMATOM_X(W, 32, &var->value, val); #elif defined VM_ARM_32 /* * Best left this way due to the intricacies of exclusive load/store * operations on legacy (32-bit) ARM. * * A3.4.1 ARM DDI 0406C: * * When a processor writes using any instruction other than a * Store-Exclusive: * * - if the write is to a physical address that is not covered by its local * monitor the write does not affect the state of the local monitor * - if the write is to a physical address that is covered by its local * monitor it is IMPLEMENTATION DEFINED whether the write affects the * state of the local monitor. * * A3.4.5 ARM DDI 0406C: * * If two STREX instructions are executed without an intervening LDREX the * second STREX returns a status value of 1. This means that: * * - ARM recommends that, in a given thread of execution, every STREX has a * preceding LDREX associated with it * - it is not necessary for every LDREX to have a subsequent STREX. */ Atomic_ReadWrite32(var, val); #else /* * Use inline assembler to force using a single store instruction to * ensure that the compiler doesn't split a transfer operation into multiple * instructions. */ __asm__ __volatile__( "mov %1, %0" : "=m" (var->value) : "r" (val) ); #endif #elif defined _MSC_VER /* * Microsoft docs guarantee simple reads and writes to properly * aligned 32-bit variables use only a single instruction. * http://msdn.microsoft.com/en-us/library/ms684122%28VS.85%29.aspx */ var->value = val; #else #error No compiler defined for Atomic_Write #endif } #define Atomic_Write Atomic_Write32 /* *----------------------------------------------------------------------------- * * Atomic_ReadIfEqualWrite32 -- * * Compare exchange: Read variable, if equal to oldVal, write newVal * * Results: * The value of the atomic variable before the write. * * Side effects: * The variable may be modified. * *----------------------------------------------------------------------------- */ static INLINE uint32 Atomic_ReadIfEqualWrite32(Atomic_uint32 *var, // IN/OUT uint32 oldVal, // IN uint32 newVal) // IN { #if defined __GNUC__ #ifdef VM_ARM_V7 register uint32 retVal; register uint32 res; dmb(); __asm__ __volatile__( "1: ldrex %[retVal], [%[var]] \n\t" "mov %[res], #0 \n\t" "teq %[retVal], %[oldVal] \n\t" "strexeq %[res], %[newVal], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [retVal] "=&r" (retVal), [res] "=&r" (res) : [var] "r" (&var->value), [oldVal] "r" (oldVal), [newVal] "r" (newVal) : "cc" ); dmb(); return retVal; #elif defined VM_ARM_64 return _VMATOM_X(RIFEQW, 32, TRUE, &var->value, oldVal, newVal); #else /* VM_X86_ANY */ uint32 val; /* Checked against the Intel manual and GCC --walken */ __asm__ __volatile__( "lock; cmpxchgl %2, %1" : "=a" (val), "+m" (var->value) : "r" (newVal), "0" (oldVal) : "cc", "memory" ); return val; #endif /* VM_X86_ANY */ #elif defined _MSC_VER return _InterlockedCompareExchange((long *)&var->value, (long)newVal, (long)oldVal); #else #error No compiler defined for Atomic_ReadIfEqualWrite #endif } #define Atomic_ReadIfEqualWrite Atomic_ReadIfEqualWrite32 #if defined VM_64BIT || defined VM_ARM_V7 /* *----------------------------------------------------------------------------- * * Atomic_ReadIfEqualWrite64 -- * * Compare exchange: Read variable, if equal to oldVal, write newVal * * Results: * The value of the atomic variable before the write. * * Side effects: * The variable may be modified. * *----------------------------------------------------------------------------- */ static INLINE uint64 Atomic_ReadIfEqualWrite64(Atomic_uint64 *var, // IN/OUT uint64 oldVal, // IN uint64 newVal) // IN { #if defined __GNUC__ #ifdef VM_ARM_V7 register uint64 retVal; register uint32 res; dmb(); /* * Under Apple LLVM version 5.0 (clang-500.2.76) (based on LLVM 3.3svn) * There will be a warning: * "value size does not match register size specified by the constraint * and modifier [-Wasm-operand-widths]" * on the lines: * : [var] "r" (&var->value), [oldVal] "r" (oldVal), [newVal] "r" (newVal) * ^ * : [var] "r" (&var->value), [oldVal] "r" (oldVal), [newVal] "r" (newVal) * ^ * * Furthermore, using a 32-bits register to store a * 64-bits value of an variable looks risky. */ #if defined __APPLE__ && __clang__ == 1 && __clang_major__ >= 5 #pragma clang diagnostic push #pragma clang diagnostic ignored "-Wasm-operand-widths" #endif __asm__ __volatile__( "1: ldrexd %[retVal], %H[retVal], [%[var]] \n\t" "mov %[res], #0 \n\t" "teq %[retVal], %[oldVal] \n\t" "teqeq %H[retVal], %H[oldVal] \n\t" "strexdeq %[res], %[newVal], %H[newVal], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [retVal] "=&r" (retVal), [res] "=&r" (res) : [var] "r" (&var->value), [oldVal] "r" (oldVal), [newVal] "r" (newVal) : "cc" ); #if defined __APPLE__ && __clang__ == 1 && __clang_major__ >= 5 #pragma clang diagnostic pop #endif // defined __APPLE__ && __clang__ == 1 && __clang_major__ >= 5 dmb(); return retVal; #elif defined VM_ARM_64 return _VMATOM_X(RIFEQW, 64, TRUE, &var->value, oldVal, newVal); #else /* VM_X86_64 */ uint64 val; /* Checked against the AMD manual and GCC --hpreg */ __asm__ __volatile__( "lock; cmpxchgq %2, %1" : "=a" (val), "+m" (var->value) : "r" (newVal), "0" (oldVal) : "cc", "memory" ); return val; #endif //VM_ARM_V7 #elif defined _MSC_VER return _InterlockedCompareExchange64((__int64 *)&var->value, (__int64)newVal, (__int64)oldVal); #else #error No compiler defined for Atomic_ReadIfEqualWrite64 #endif } #endif /* *----------------------------------------------------------------------------- * * Atomic_And32 -- * * Atomic read, bitwise AND with a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_And32(Atomic_uint32 *var, // IN/OUT uint32 val) // IN { #if defined __GNUC__ #ifdef VM_ARM_V7 register volatile uint32 res; register volatile uint32 tmp; dmb(); __asm__ __volatile__( "1: ldrex %[tmp], [%[var]] \n\t" "and %[tmp], %[tmp], %[val] \n\t" "strex %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [res] "=&r" (res), [tmp] "=&r" (tmp) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); #elif defined VM_ARM_64 _VMATOM_X(OP, 32, TRUE, &var->value, and, val); #else /* VM_X86_ANY */ /* Checked against the Intel manual and GCC --walken */ __asm__ __volatile__( "lock; andl %1, %0" : "+m" (var->value) : "ri" (val) : "cc", "memory" ); #endif /* VM_X86_ANY */ #elif defined _MSC_VER #if defined __x86_64__ || defined VM_ARM_32 _InterlockedAnd((long *)&var->value, (long)val); #else __asm mov eax, val __asm mov ebx, var __asm lock and [ebx]Atomic_uint32.value, eax #endif #else #error No compiler defined for Atomic_And #endif } #define Atomic_And Atomic_And32 /* *----------------------------------------------------------------------------- * * Atomic_Or32 -- * * Atomic read, bitwise OR with a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Or32(Atomic_uint32 *var, // IN/OUT uint32 val) // IN { #if defined __GNUC__ #ifdef VM_ARM_V7 register volatile uint32 res; register volatile uint32 tmp; dmb(); __asm__ __volatile__( "1: ldrex %[tmp], [%[var]] \n\t" "orr %[tmp], %[tmp], %[val] \n\t" "strex %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [res] "=&r" (res), [tmp] "=&r" (tmp) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); #elif defined VM_ARM_64 _VMATOM_X(OP, 32, TRUE, &var->value, orr, val); #else /* VM_X86_ANY */ /* Checked against the Intel manual and GCC --walken */ __asm__ __volatile__( "lock; orl %1, %0" : "+m" (var->value) : "ri" (val) : "cc", "memory" ); #endif /* VM_X86_ANY */ #elif defined _MSC_VER #if defined __x86_64__ || defined VM_ARM_32 _InterlockedOr((long *)&var->value, (long)val); #else __asm mov eax, val __asm mov ebx, var __asm lock or [ebx]Atomic_uint32.value, eax #endif #else #error No compiler defined for Atomic_Or #endif } #define Atomic_Or Atomic_Or32 /* *----------------------------------------------------------------------------- * * Atomic_Xor32 -- * * Atomic read, bitwise XOR with a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Xor32(Atomic_uint32 *var, // IN/OUT uint32 val) // IN { #if defined __GNUC__ #ifdef VM_ARM_V7 register volatile uint32 res; register volatile uint32 tmp; dmb(); __asm__ __volatile__( "1: ldrex %[tmp], [%[var]] \n\t" "eor %[tmp], %[tmp], %[val] \n\t" "strex %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [res] "=&r" (res), [tmp] "=&r" (tmp) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); #elif defined VM_ARM_64 _VMATOM_X(OP, 32, TRUE, &var->value, eor, val); #else /* VM_X86_ANY */ /* Checked against the Intel manual and GCC --walken */ __asm__ __volatile__( "lock; xorl %1, %0" : "+m" (var->value) : "ri" (val) : "cc", "memory" ); #endif /* VM_X86_ANY */ #elif defined _MSC_VER #if defined __x86_64__ || defined VM_ARM_32 _InterlockedXor((long *)&var->value, (long)val); #else __asm mov eax, val __asm mov ebx, var __asm lock xor [ebx]Atomic_uint32.value, eax #endif #else #error No compiler defined for Atomic_Xor #endif } #define Atomic_Xor Atomic_Xor32 #if defined VM_64BIT /* *----------------------------------------------------------------------------- * * Atomic_Xor64 -- * * Atomic read, bitwise XOR with a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Xor64(Atomic_uint64 *var, // IN/OUT uint64 val) // IN { #if defined __GNUC__ #if defined VM_ARM_64 _VMATOM_X(OP, 64, TRUE, &var->value, eor, val); #else /* VM_X86_64 */ /* Checked against the AMD manual and GCC --hpreg */ __asm__ __volatile__( "lock; xorq %1, %0" : "+m" (var->value) : "re" (val) : "cc", "memory" ); #endif #elif defined _MSC_VER _InterlockedXor64((__int64 *)&var->value, (__int64)val); #else #error No compiler defined for Atomic_Xor64 #endif } #endif /* *----------------------------------------------------------------------------- * * Atomic_Add32 -- * * Atomic read, add a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Add32(Atomic_uint32 *var, // IN/OUT uint32 val) // IN { #if defined __GNUC__ #ifdef VM_ARM_V7 register volatile uint32 res; register volatile uint32 tmp; dmb(); __asm__ __volatile__( "1: ldrex %[tmp], [%[var]] \n\t" "add %[tmp], %[tmp], %[val] \n\t" "strex %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [res] "=&r" (res), [tmp] "=&r" (tmp) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); #elif defined VM_ARM_64 _VMATOM_X(OP, 32, TRUE, &var->value, add, val); #else /* VM_X86_ANY */ /* Checked against the Intel manual and GCC --walken */ __asm__ __volatile__( "lock; addl %1, %0" : "+m" (var->value) : "ri" (val) : "cc", "memory" ); #endif /* VM_X86_ANY */ #elif defined _MSC_VER _InterlockedExchangeAdd((long *)&var->value, (long)val); #else #error No compiler defined for Atomic_Add #endif } #define Atomic_Add Atomic_Add32 /* *----------------------------------------------------------------------------- * * Atomic_Sub32 -- * * Atomic read, subtract a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Sub32(Atomic_uint32 *var, // IN/OUT uint32 val) // IN { #if defined __GNUC__ #ifdef VM_ARM_V7 register volatile uint32 res; register volatile uint32 tmp; dmb(); __asm__ __volatile__( "1: ldrex %[tmp], [%[var]] \n\t" "sub %[tmp], %[tmp], %[val] \n\t" "strex %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [res] "=&r" (res), [tmp] "=&r" (tmp) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); #elif defined VM_ARM_64 _VMATOM_X(OP, 32, TRUE, &var->value, sub, val); #else /* VM_X86_ANY */ /* Checked against the Intel manual and GCC --walken */ __asm__ __volatile__( "lock; subl %1, %0" : "+m" (var->value) : "ri" (val) : "cc", "memory" ); #endif /* VM_X86_ANY */ #elif defined _MSC_VER _InterlockedExchangeAdd((long *)&var->value, -(long)val); #else #error No compiler defined for Atomic_Sub #endif } #define Atomic_Sub Atomic_Sub32 /* *----------------------------------------------------------------------------- * * Atomic_Inc32 -- * * Atomic read, increment, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Inc32(Atomic_uint32 *var) // IN/OUT { #ifdef __GNUC__ #if defined VM_ARM_ANY Atomic_Add32(var, 1); #else /* VM_X86_ANY */ /* Checked against the Intel manual and GCC --walken */ __asm__ __volatile__( "lock; incl %0" : "+m" (var->value) : : "cc", "memory" ); #endif /* VM_X86_ANY */ #elif defined _MSC_VER _InterlockedIncrement((long *)&var->value); #else #error No compiler defined for Atomic_Inc #endif } #define Atomic_Inc Atomic_Inc32 /* *----------------------------------------------------------------------------- * * Atomic_Dec32 -- * * Atomic read, decrement, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Dec32(Atomic_uint32 *var) // IN/OUT { #ifdef __GNUC__ #if defined VM_ARM_ANY Atomic_Sub32(var, 1); #else /* VM_X86_ANY */ /* Checked against the Intel manual and GCC --walken */ __asm__ __volatile__( "lock; decl %0" : "+m" (var->value) : : "cc", "memory" ); #endif /* VM_X86_ANY */ #elif defined _MSC_VER _InterlockedDecrement((long *)&var->value); #else #error No compiler defined for Atomic_Dec #endif } #define Atomic_Dec Atomic_Dec32 /* * Note that the technique below can be used to implement ReadX(), where X is * an arbitrary mathematical function. */ /* *----------------------------------------------------------------------------- * * Atomic_ReadOr32 -- * * Atomic read (returned), bitwise OR with a value, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint32 Atomic_ReadOr32(Atomic_uint32 *var, // IN/OUT uint32 val) // IN { uint32 res; #if defined VM_ARM_64 res = _VMATOM_X(ROP, 32, TRUE, &var->value, orr, val); #else do { res = Atomic_Read32(var); } while (res != Atomic_ReadIfEqualWrite32(var, res, res | val)); #endif return res; } /* *----------------------------------------------------------------------------- * * Atomic_ReadAnd32 -- * * Atomic read (returned), bitwise And with a value, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint32 Atomic_ReadAnd32(Atomic_uint32 *var, // IN/OUT uint32 val) // IN { uint32 res; #if defined VM_ARM_64 res = _VMATOM_X(ROP, 32, TRUE, &var->value, and, val); #else do { res = Atomic_Read32(var); } while (res != Atomic_ReadIfEqualWrite32(var, res, res & val)); #endif return res; } #if defined VM_64BIT /* *----------------------------------------------------------------------------- * * Atomic_ReadOr64 -- * * Atomic read (returned), bitwise OR with a value, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint64 Atomic_ReadOr64(Atomic_uint64 *var, // IN/OUT uint64 val) // IN { uint64 res; #if defined VM_ARM_64 res = _VMATOM_X(ROP, 64, TRUE, &var->value, orr, val); #else do { res = var->value; } while (res != Atomic_ReadIfEqualWrite64(var, res, res | val)); #endif return res; } /* *----------------------------------------------------------------------------- * * Atomic_ReadAnd64 -- * * Atomic read (returned), bitwise AND with a value, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint64 Atomic_ReadAnd64(Atomic_uint64 *var, // IN/OUT uint64 val) // IN { uint64 res; #if defined VM_ARM_64 res = _VMATOM_X(ROP, 64, TRUE, &var->value, and, val); #else do { res = var->value; } while (res != Atomic_ReadIfEqualWrite64(var, res, res & val)); #endif return res; } #endif /* defined VM_64BIT */ /* *----------------------------------------------------------------------------- * * Atomic_ReadAdd32 -- * * Atomic read (returned), add a value, write. * * If you have to implement ReadAdd32() on an architecture other than * x86 or x86-64, you might want to consider doing something similar to * Atomic_ReadOr32(). * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint32 Atomic_ReadAdd32(Atomic_uint32 *var, // IN/OUT uint32 val) // IN { #if defined __GNUC__ #ifdef VM_ARM_V7 register volatile uint32 res; register volatile uint32 retVal; register volatile uint32 tmp; dmb(); __asm__ __volatile__( "1: ldrex %[retVal], [%[var]] \n\t" "add %[tmp], %[val], %[retVal] \n\t" "strex %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [tmp] "=&r" (tmp), [res] "=&r" (res), [retVal] "=&r" (retVal) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); return retVal; #elif defined VM_ARM_64 return _VMATOM_X(ROP, 32, TRUE, &var->value, add, val); #else /* VM_X86_ANY */ /* Checked against the Intel manual and GCC --walken */ __asm__ __volatile__( "lock; xaddl %0, %1" : "=r" (val), "+m" (var->value) : "0" (val) : "cc", "memory" ); return val; #endif /* VM_X86_ANY */ #elif defined _MSC_VER return _InterlockedExchangeAdd((long *)&var->value, (long)val); #else #error No compiler defined for Atomic_ReadAdd32 #endif } /* *----------------------------------------------------------------------------- * * Atomic_ReadInc32 -- * * Atomic read (returned), increment, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint32 Atomic_ReadInc32(Atomic_uint32 *var) // IN/OUT { return Atomic_ReadAdd32(var, 1); } /* *----------------------------------------------------------------------------- * * Atomic_ReadDec32 -- * * Atomic read (returned), decrement, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint32 Atomic_ReadDec32(Atomic_uint32 *var) // IN/OUT { return Atomic_ReadAdd32(var, (uint32)-1); } /* *----------------------------------------------------------------------------- * * Atomic_CMPXCHG64 -- * * Compare exchange: Read variable, if equal to oldVal, write newVal * * Results: * TRUE if equal, FALSE if not equal * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE Bool Atomic_CMPXCHG64(Atomic_uint64 *var, // IN/OUT uint64 oldVal, // IN uint64 newVal) // IN { #if defined __GNUC__ #if defined VM_ARM_ANY return Atomic_ReadIfEqualWrite64(var, oldVal, newVal) == oldVal; #else /* VM_X86_ANY */ Bool equal; /* Checked against the Intel manual and GCC --walken */ #if defined __x86_64__ uint64 dummy; __asm__ __volatile__( "lock; cmpxchgq %3, %0" "\n\t" "sete %1" : "+m" (*var), "=qm" (equal), "=a" (dummy) : "r" (newVal), "2" (oldVal) : "cc", "memory" ); #else /* 32-bit version for non-ARM */ typedef struct { uint32 lowValue; uint32 highValue; } S_uint64; int dummy1, dummy2; # if defined __PIC__ /* * Rules for __asm__ statements in __PIC__ code * -------------------------------------------- * * The compiler uses %ebx for __PIC__ code, so an __asm__ statement cannot * clobber %ebx. The __asm__ statement can temporarily modify %ebx, but _for * each parameter that is used while %ebx is temporarily modified_: * * 1) The constraint cannot be "m", because the memory location the compiler * chooses could then be relative to %ebx. * * 2) The constraint cannot be a register class which contains %ebx (such as * "r" or "q"), because the register the compiler chooses could then be * %ebx. (This happens when compiling the Fusion UI with gcc 4.2.1, Apple * build 5577.) * * 3) Using register classes even for other values is problematic, as gcc * can decide e.g. %ecx == %edi == 0 (as compile-time constants) and * ends up using one register for two things. Which breaks xchg's ability * to temporarily put the PIC pointer somewhere else. PR772455 * * For that reason alone, the __asm__ statement should keep the regions * where it temporarily modifies %ebx as small as possible, and should * prefer specific register assignments. */ __asm__ __volatile__( "xchgl %%ebx, %6" "\n\t" "lock; cmpxchg8b (%3)" "\n\t" "xchgl %%ebx, %6" "\n\t" "sete %0" : "=qm" (equal), "=a" (dummy1), "=d" (dummy2) : /* * See the "Rules for __asm__ statements in __PIC__ code" above: %3 * must use a register class which does not contain %ebx. * "a"/"c"/"d" are already used, so we are left with either "S" or "D". * * Note that this assembly uses ALL GP registers (with %esp reserved for * stack, %ebp reserved for frame, %ebx reserved for PIC). */ "S" (var), "1" (((S_uint64 *)&oldVal)->lowValue), "2" (((S_uint64 *)&oldVal)->highValue), "D" (((S_uint64 *)&newVal)->lowValue), "c" (((S_uint64 *)&newVal)->highValue) : "cc", "memory" ); # else __asm__ __volatile__( "lock; cmpxchg8b %0" "\n\t" "sete %1" : "+m" (*var), "=qm" (equal), "=a" (dummy1), "=d" (dummy2) : "2" (((S_uint64 *)&oldVal)->lowValue), "3" (((S_uint64 *)&oldVal)->highValue), "b" (((S_uint64 *)&newVal)->lowValue), "c" (((S_uint64 *)&newVal)->highValue) : "cc", "memory" ); # endif #endif return equal; #endif //VM_ARM_V7 #elif defined _MSC_VER return (__int64)oldVal == _InterlockedCompareExchange64((__int64 *)&var->value, (__int64)newVal, (__int64)oldVal); #else #error No compiler defined for Atomic_CMPXCHG64 #endif // !GNUC } /* *----------------------------------------------------------------------------- * * Atomic_CMPXCHG32 -- * * Compare exchange: Read variable, if equal to oldVal, write newVal * * Results: * TRUE if equal, FALSE if not equal * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE Bool Atomic_CMPXCHG32(Atomic_uint32 *var, // IN/OUT uint32 oldVal, // IN uint32 newVal) // IN { #if defined __GNUC__ #if defined VM_ARM_ANY return Atomic_ReadIfEqualWrite32(var, oldVal, newVal) == oldVal; #else /* VM_X86_ANY */ Bool equal; uint32 dummy; __asm__ __volatile__( "lock; cmpxchgl %3, %0" "\n\t" "sete %1" : "+m" (*var), "=qm" (equal), "=a" (dummy) : "r" (newVal), "2" (oldVal) : "cc", "memory" ); return equal; #endif /* VM_X86_ANY */ #else // defined __GNUC__ return Atomic_ReadIfEqualWrite32(var, oldVal, newVal) == oldVal; #endif // !defined __GNUC__ } /* *----------------------------------------------------------------------------- * * Atomic_Read64 -- * * Read and return. * * Results: * The value of the atomic variable. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE uint64 Atomic_Read64(Atomic_uint64 const *var) // IN { #if defined __GNUC__ uint64 value; #endif #if defined VMM || defined VM_ARM_64 || defined VMKERNEL || defined VMKERNEL_MODULE ASSERT((uintptr_t)var % 8 == 0); #endif #if defined __GNUC__ && defined __x86_64__ /* * Use asm to ensure we emit a single load. */ __asm__ __volatile__( "movq %1, %0" : "=r" (value) : "m" (var->value) ); #elif defined __GNUC__ && defined __i386__ /* * Since cmpxchg8b will replace the contents of EDX:EAX with the * value in memory if there is no match, we need only execute the * instruction once in order to atomically read 64 bits from * memory. The only constraint is that ECX:EBX must have the same * value as EDX:EAX so that if the comparison succeeds. We * intentionally don't tell gcc that we are using ebx and ecx as we * don't modify them and do not care what value they store. */ __asm__ __volatile__( "mov %%ebx, %%eax" "\n\t" "mov %%ecx, %%edx" "\n\t" "lock; cmpxchg8b %1" : "=&A" (value) : "m" (*var) : "cc" ); #elif defined _MSC_VER && defined __x86_64__ /* * Microsoft docs guarantee "Simple reads and writes to properly * aligned 64-bit variables are atomic on 64-bit Windows." * http://msdn.microsoft.com/en-us/library/ms684122%28VS.85%29.aspx * * XXX Verify that value is properly aligned. Bug 61315. */ return var->value; #elif defined _MSC_VER && defined VM_ARM_32 return _InterlockedAdd64((__int64 *)&var->value, 0); #elif defined _MSC_VER && defined __i386__ # pragma warning(push) # pragma warning(disable : 4035) // disable no-return warning { __asm mov ecx, var __asm mov edx, ecx __asm mov eax, ebx __asm lock cmpxchg8b [ecx] // edx:eax is the return value; this is documented to work. --mann } # pragma warning(pop) #elif defined __GNUC__ && defined VM_ARM_V7 __asm__ __volatile__( "ldrexd %[value], %H[value], [%[var]] \n\t" : [value] "=&r" (value) : [var] "r" (&var->value) ); #elif defined VM_ARM_64 value = _VMATOM_X(R, 64, &var->value); #endif #if defined __GNUC__ return value; #endif } /* *---------------------------------------------------------------------- * * Atomic_ReadUnaligned64 -- * * Atomically read a 64 bit integer, possibly misaligned. * This function can be *very* expensive, costing over 50 kcycles * on Nehalem. * * Note that "var" needs to be writable, even though it will not * be modified. * * Results: * The value of the atomic variable. * * Side effects: * None * *---------------------------------------------------------------------- */ #if defined VM_64BIT static INLINE uint64 Atomic_ReadUnaligned64(Atomic_uint64 const *var) // IN: { return Atomic_ReadIfEqualWrite64((Atomic_uint64*)var, 0, 0); } #endif /* *---------------------------------------------------------------------- * * Atomic_ReadAdd64 -- * * Atomically adds a 64-bit integer to another * * Results: * Returns the old value just prior to the addition * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE uint64 Atomic_ReadAdd64(Atomic_uint64 *var, // IN/OUT uint64 val) // IN { #if defined VM_ARM_64 return _VMATOM_X(ROP, 64, TRUE, &var->value, add, val); #elif defined __x86_64__ #if defined __GNUC__ __asm__ __volatile__( "lock; xaddq %0, %1" : "=r" (val), "+m" (var->value) : "0" (val) : "cc", "memory" ); return val; #elif defined _MSC_VER return _InterlockedExchangeAdd64((__int64 *)&var->value, (__int64)val); #else #error No compiler defined for Atomic_ReadAdd64 #endif #else uint64 oldVal; uint64 newVal; do { oldVal = var->value; newVal = oldVal + val; } while (!Atomic_CMPXCHG64(var, oldVal, newVal)); return oldVal; #endif } /* *---------------------------------------------------------------------- * * Atomic_ReadSub64 -- * * Atomically subtracts a 64-bit integer to another * * Results: * Returns the old value just prior to the subtraction * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE uint64 Atomic_ReadSub64(Atomic_uint64 *var, // IN/OUT uint64 val) // IN { #if defined VM_ARM_64 return _VMATOM_X(ROP, 64, TRUE, &var->value, sub, val); #else return Atomic_ReadAdd64(var, -(int64)val); #endif } /* *---------------------------------------------------------------------- * * Atomic_ReadInc64 -- * * Atomically increments a 64-bit integer * * Results: * Returns the old value just prior to incrementing * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE uint64 Atomic_ReadInc64(Atomic_uint64 *var) // IN/OUT { return Atomic_ReadAdd64(var, 1); } /* *---------------------------------------------------------------------- * * Atomic_ReadDec64 -- * * Atomically decrements a 64-bit integer * * Results: * Returns the old value just prior to decrementing * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE uint64 Atomic_ReadDec64(Atomic_uint64 *var) // IN/OUT { return Atomic_ReadAdd64(var, (uint64)CONST64(-1)); } /* *----------------------------------------------------------------------------- * * Atomic_Add64 -- * * Atomic read, add a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Add64(Atomic_uint64 *var, // IN/OUT uint64 val) // IN { #if !defined VM_64BIT Atomic_ReadAdd64(var, val); /* Return value is unused. */ #elif defined __GNUC__ #if defined VM_ARM_64 _VMATOM_X(OP, 64, TRUE, &var->value, add, val); #else /* defined VM_X86_64 */ /* Checked against the AMD manual and GCC --hpreg */ __asm__ __volatile__( "lock; addq %1, %0" : "+m" (var->value) : "re" (val) : "cc", "memory" ); #endif #elif defined _MSC_VER _InterlockedExchangeAdd64((__int64 *)&var->value, (__int64)val); #else #error No compiler defined for Atomic_Add64 #endif } /* *----------------------------------------------------------------------------- * * Atomic_Sub64 -- * * Atomic read, subtract a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Sub64(Atomic_uint64 *var, // IN/OUT uint64 val) // IN { #if !defined VM_64BIT Atomic_ReadSub64(var, val); /* Return value is unused. */ #elif defined __GNUC__ #if defined VM_ARM_64 _VMATOM_X(OP, 64, TRUE, &var->value, sub, val); #else /* VM_X86_64 */ /* Checked against the AMD manual and GCC --hpreg */ __asm__ __volatile__( "lock; subq %1, %0" : "+m" (var->value) : "re" (val) : "cc", "memory" ); #endif #elif defined _MSC_VER _InterlockedExchangeAdd64((__int64 *)&var->value, (__int64)-val); #else #error No compiler defined for Atomic_Sub64 #endif } /* *----------------------------------------------------------------------------- * * Atomic_Inc64 -- * * Atomic read, increment, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Inc64(Atomic_uint64 *var) // IN/OUT { #if defined VM_ARM_64 Atomic_Add64(var, 1); #elif !defined __x86_64__ Atomic_ReadInc64(var); /* Return value is unused. */ #elif defined __GNUC__ /* Checked against the AMD manual and GCC --hpreg */ __asm__ __volatile__( "lock; incq %0" : "+m" (var->value) : : "cc", "memory" ); #elif defined _MSC_VER _InterlockedIncrement64((__int64 *)&var->value); #else #error No compiler defined for Atomic_Inc64 #endif } /* *----------------------------------------------------------------------------- * * Atomic_Dec64 -- * * Atomic read, decrement, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Dec64(Atomic_uint64 *var) // IN/OUT { #if defined VM_ARM_64 Atomic_Sub64(var, 1); #elif !defined __x86_64__ Atomic_ReadDec64(var); /* Return value is unused. */ #elif defined __GNUC__ /* Checked against the AMD manual and GCC --hpreg */ __asm__ __volatile__( "lock; decq %0" : "+m" (var->value) : : "cc", "memory" ); #elif defined _MSC_VER _InterlockedDecrement64((__int64 *)&var->value); #else #error No compiler defined for Atomic_Dec64 #endif } /* *----------------------------------------------------------------------------- * * Atomic_ReadWrite64 -- * * Read followed by write * * Results: * The value of the atomic variable before the write. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint64 Atomic_ReadWrite64(Atomic_uint64 *var, // IN/OUT uint64 val) // IN { #if defined __x86_64__ #if defined __GNUC__ /* Checked against the AMD manual and GCC --hpreg */ __asm__ __volatile__( "xchgq %0, %1" : "=r" (val), "+m" (var->value) : "0" (val) : "memory" ); return val; #elif defined _MSC_VER return _InterlockedExchange64((__int64 *)&var->value, (__int64)val); #else #error No compiler defined for Atomic_ReadWrite64 #endif #elif defined VM_ARM_64 return _VMATOM_X(RW, 64, TRUE, &var->value, val); #else uint64 oldVal; do { oldVal = var->value; } while (!Atomic_CMPXCHG64(var, oldVal, val)); return oldVal; #endif } /* *----------------------------------------------------------------------------- * * Atomic_Write64 -- * * Write * * Results: * None. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Write64(Atomic_uint64 *var, // OUT uint64 val) // IN { #if defined VMM || defined VM_ARM_64 || defined VMKERNEL || defined VMKERNEL_MODULE ASSERT((uintptr_t)var % 8 == 0); #endif #if defined __x86_64__ #if defined __GNUC__ /* * There is no move instruction for 64-bit immediate to memory, so unless * the immediate value fits in 32-bit (i.e. can be sign-extended), GCC * breaks the assignment into two movl instructions. The code below forces * GCC to load the immediate value into a register first. */ __asm__ __volatile__( "movq %1, %0" : "=m" (var->value) : "r" (val) ); #elif defined _MSC_VER /* * Microsoft docs guarantee "Simple reads and writes to properly aligned * 64-bit variables are atomic on 64-bit Windows." * http://msdn.microsoft.com/en-us/library/ms684122%28VS.85%29.aspx * * XXX Verify that value is properly aligned. Bug 61315. */ var->value = val; #else #error No compiler defined for Atomic_Write64 #endif #elif defined VM_ARM_64 _VMATOM_X(W, 64, &var->value, val); #else (void)Atomic_ReadWrite64(var, val); #endif } /* *----------------------------------------------------------------------------- * * Atomic_Or64 -- * * Atomic read, bitwise OR with a 64-bit value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Or64(Atomic_uint64 *var, // IN/OUT uint64 val) // IN { #if defined __x86_64__ #if defined __GNUC__ /* Checked against the AMD manual and GCC --hpreg */ __asm__ __volatile__( "lock; orq %1, %0" : "+m" (var->value) : "re" (val) : "cc", "memory" ); #elif defined _MSC_VER _InterlockedOr64((__int64 *)&var->value, (__int64)val); #else #error No compiler defined for Atomic_Or64 #endif #elif defined VM_ARM_64 _VMATOM_X(OP, 64, TRUE, &var->value, orr, val); #else // __x86_64__ uint64 oldVal; uint64 newVal; do { oldVal = var->value; newVal = oldVal | val; } while (!Atomic_CMPXCHG64(var, oldVal, newVal)); #endif } /* *----------------------------------------------------------------------------- * * Atomic_And64 -- * * Atomic read, bitwise AND with a 64-bit value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_And64(Atomic_uint64 *var, // IN/OUT uint64 val) // IN { #if defined __x86_64__ #if defined __GNUC__ /* Checked against the AMD manual and GCC --hpreg */ __asm__ __volatile__( "lock; andq %1, %0" : "+m" (var->value) : "re" (val) : "cc", "memory" ); #elif defined _MSC_VER _InterlockedAnd64((__int64 *)&var->value, (__int64)val); #else #error No compiler defined for Atomic_And64 #endif #elif defined VM_ARM_64 _VMATOM_X(OP, 64, TRUE, &var->value, and, val); #else // __x86_64__ uint64 oldVal; uint64 newVal; do { oldVal = var->value; newVal = oldVal & val; } while (!Atomic_CMPXCHG64(var, oldVal, newVal)); #endif } /* *----------------------------------------------------------------------------- * * Atomic_SetBit64 -- * * Atomically set the bit 'bit' in var. Bit must be between 0 and 63. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_SetBit64(Atomic_uint64 *var, // IN/OUT unsigned bit) // IN { #if defined __x86_64__ && defined __GNUC__ ASSERT(bit <= 63); __asm__ __volatile__( "lock; btsq %1, %0" : "+m" (var->value) : "ri" ((uint64)bit) : "cc", "memory" ); #else uint64 oldVal; uint64 newVal; ASSERT(bit <= 63); do { oldVal = var->value; newVal = oldVal | (CONST64U(1) << bit); } while (!Atomic_CMPXCHG64(var, oldVal, newVal)); #endif } /* *----------------------------------------------------------------------------- * * Atomic_ClearBit64 -- * * Atomically clear the bit 'bit' in var. Bit must be between 0 and 63. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_ClearBit64(Atomic_uint64 *var, // IN/OUT unsigned bit) // IN { #if defined __x86_64__ && defined __GNUC__ ASSERT(bit <= 63); __asm__ __volatile__( "lock; btrq %1, %0" : "+m" (var->value) : "ri" ((uint64)bit) : "cc", "memory" ); #else uint64 oldVal; uint64 newVal; ASSERT(bit <= 63); do { oldVal = var->value; newVal = oldVal & ~(CONST64U(1) << bit); } while (!Atomic_CMPXCHG64(var, oldVal, newVal)); #endif } /* *----------------------------------------------------------------------------- * * Atomic_TestBit64 -- * * Read the bit 'bit' in var. Bit must be between 0 and 63. * * Results: * TRUE if the tested bit was set; else FALSE. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE Bool Atomic_TestBit64(Atomic_uint64 *var, // IN unsigned bit) // IN { Bool out; ASSERT(bit <= 63); #if defined __x86_64__ && defined __GNUC__ __asm__ __volatile__( "btq %2, %1; setc %0" : "=rm"(out) : "m" (var->value), "rJ" ((uint64)bit) : "cc" ); #else out = (var->value & (CONST64U(1) << bit)) != 0; #endif return out; } /* *----------------------------------------------------------------------------- * * Atomic_TestSetBit64 -- * * Atomically test and set the bit 'bit' in var. * Bit must be between 0 and 63. * * Results: * TRUE if the tested bit was set; else FALSE. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE Bool Atomic_TestSetBit64(Atomic_uint64 *var, // IN/OUT unsigned bit) // IN { #if defined __x86_64__ && defined __GNUC__ Bool out; ASSERT(bit <= 63); __asm__ __volatile__( "lock; btsq %2, %1; setc %0" : "=rm" (out), "+m" (var->value) : "rJ" ((uint64)bit) : "cc", "memory" ); return out; #else uint64 oldVal; uint64 mask; ASSERT(bit <= 63); mask = CONST64U(1) << bit; do { oldVal = var->value; } while (!Atomic_CMPXCHG64(var, oldVal, oldVal | mask)); return (oldVal & mask) != 0; #endif } #if defined __GNUC__ /* *----------------------------------------------------------------------------- * * Atomic_Read16 -- * * Read and return. * * Results: * The value of the atomic variable. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE uint16 Atomic_Read16(Atomic_uint16 const *var) // IN { uint16 value; #if defined VMM || defined VM_ARM_64 || defined VMKERNEL || defined VMKERNEL_MODULE ASSERT((uintptr_t)var % 2 == 0); #endif #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "movw %1, %0" : "=r" (value) : "m" (var->value) ); #elif defined VM_ARM_V7 NOT_TESTED(); __asm__ __volatile__( "ldrh %0, [%1]" : "=r" (value) : "r" (&var->value) ); #elif defined VM_ARM_64 value = _VMATOM_X(R, 16, &var->value); #else #error No 16-bits atomics. #endif #endif return value; } /* *----------------------------------------------------------------------------- * * Atomic_ReadWrite16 -- * * Read followed by write * * Results: * The value of the atomic variable before the write. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint16 Atomic_ReadWrite16(Atomic_uint16 *var, // IN/OUT: uint16 val) // IN: { #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "xchgw %0, %1" : "=r" (val), "+m" (var->value) : "0" (val) : "memory" ); return val; #elif defined VM_ARM_V7 register volatile uint16 retVal; register volatile uint16 res; NOT_TESTED(); dmb(); __asm__ __volatile__( "1: ldrexh %[retVal], [%[var]] \n\t" "strexh %[res], %[val], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [retVal] "=&r" (retVal), [res] "=&r" (res) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); return retVal; #elif defined VM_ARM_64 return _VMATOM_X(RW, 16, TRUE, &var->value, val); #else #error No 16-bits atomics. #endif #endif } /* *----------------------------------------------------------------------------- * * Atomic_Write16 -- * * Write * * Results: * None. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Write16(Atomic_uint16 *var, // OUT: uint16 val) // IN: { #if defined VMM || defined VM_ARM_64 || defined VMKERNEL || defined VMKERNEL_MODULE ASSERT((uintptr_t)var % 2 == 0); #endif #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "movw %1, %0" : "=m" (var->value) : "r" (val) ); #elif defined VM_ARM_64 _VMATOM_X(W, 16, &var->value, val); #elif defined VM_ARM_32 /* * Best left this way due to the intricacies of exclusive load/store * operations on legacy (32-bit) ARM. */ Atomic_ReadWrite16(var, val); #else #error No 16-bits atomics. #endif #endif } /* *----------------------------------------------------------------------------- * * Atomic_ReadIfEqualWrite16 -- * * Compare exchange: Read variable, if equal to oldVal, write newVal * * Results: * The value of the atomic variable before the write. * * Side effects: * The variable may be modified. * *----------------------------------------------------------------------------- */ static INLINE uint16 Atomic_ReadIfEqualWrite16(Atomic_uint16 *var, // IN/OUT uint16 oldVal, // IN uint16 newVal) // IN { #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ uint16 val; __asm__ __volatile__( "lock; cmpxchgw %2, %1" : "=a" (val), "+m" (var->value) : "r" (newVal), "0" (oldVal) : "cc", "memory" ); return val; #elif defined VM_ARM_V7 register uint16 retVal; register uint16 res; NOT_TESTED(); dmb(); __asm__ __volatile__( "1: ldrexh %[retVal], [%[var]] \n\t" "mov %[res], #0 \n\t" "teq %[retVal], %[oldVal] \n\t" "strexheq %[res], %[newVal], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [retVal] "=&r" (retVal), [res] "=&r" (res) : [var] "r" (&var->value), [oldVal] "r" (oldVal), [newVal] "r" (newVal) : "cc" ); dmb(); return retVal; #elif defined VM_ARM_64 return _VMATOM_X(RIFEQW, 16, TRUE, &var->value, oldVal, newVal); #else #error No 16-bits atomics. #endif #endif } /* *----------------------------------------------------------------------------- * * Atomic_And16 -- * * Atomic read, bitwise AND with a 16-bit value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_And16(Atomic_uint16 *var, // IN/OUT uint16 val) // IN { #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "lock; andw %1, %0" : "+m" (var->value) : "re" (val) : "cc", "memory" ); #elif defined VM_ARM_V7 register volatile uint16 res; register volatile uint16 tmp; NOT_TESTED(); dmb(); __asm__ __volatile__( "1: ldrexh %[tmp], [%[var]] \n\t" "and %[tmp], %[tmp], %[val] \n\t" "strexh %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [res] "=&r" (res), [tmp] "=&r" (tmp) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); #elif defined VM_ARM_64 _VMATOM_X(OP, 16, TRUE, &var->value, and, val); #else #error No 16-bits atomics. #endif #endif } /* *----------------------------------------------------------------------------- * * Atomic_Or16 -- * * Atomic read, bitwise OR with a 16-bit value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Or16(Atomic_uint16 *var, // IN/OUT uint16 val) // IN { #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "lock; orw %1, %0" : "+m" (var->value) : "re" (val) : "cc", "memory" ); #elif defined VM_ARM_V7 register volatile uint16 res; register volatile uint16 tmp; NOT_TESTED(); dmb(); __asm__ __volatile__( "1: ldrexh %[tmp], [%[var]] \n\t" "orr %[tmp], %[tmp], %[val] \n\t" "strexh %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [res] "=&r" (res), [tmp] "=&r" (tmp) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); #elif defined VM_ARM_64 _VMATOM_X(OP, 16, TRUE, &var->value, orr, val); #else #error No 16-bits atomics. #endif #endif } /* *----------------------------------------------------------------------------- * * Atomic_Xor16 -- * * Atomic read, bitwise XOR with a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Xor16(Atomic_uint16 *var, // IN/OUT uint16 val) // IN { #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "lock; xorw %1, %0" : "+m" (var->value) : "re" (val) : "cc", "memory" ); #elif defined VM_ARM_V7 register volatile uint16 res; register volatile uint16 tmp; NOT_TESTED(); dmb(); __asm__ __volatile__( "1: ldrexh %[tmp], [%[var]] \n\t" "eor %[tmp], %[tmp], %[val] \n\t" "strexh %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [res] "=&r" (res), [tmp] "=&r" (tmp) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); #elif defined VM_ARM_64 _VMATOM_X(OP, 16, TRUE, &var->value, eor, val); #else #error No 16-bits atomics. #endif #endif } /* *----------------------------------------------------------------------------- * * Atomic_Add16 -- * * Atomic read, add a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Add16(Atomic_uint16 *var, // IN/OUT uint16 val) // IN { #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "lock; addw %1, %0" : "+m" (var->value) : "re" (val) : "cc", "memory" ); #elif defined VM_ARM_V7 register volatile uint16 res; register volatile uint16 tmp; NOT_TESTED(); dmb(); __asm__ __volatile__( "1: ldrexh %[tmp], [%[var]] \n\t" "add %[tmp], %[tmp], %[val] \n\t" "strexh %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [res] "=&r" (res), [tmp] "=&r" (tmp) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); #elif defined VM_ARM_64 _VMATOM_X(OP, 16, TRUE, &var->value, add, val); #else #error No 16-bits atomics. #endif #endif } /* *----------------------------------------------------------------------------- * * Atomic_Sub16 -- * * Atomic read, subtract a value, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Sub16(Atomic_uint16 *var, // IN/OUT uint16 val) // IN { #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "lock; subw %1, %0" : "+m" (var->value) : "re" (val) : "cc", "memory" ); #elif defined VM_ARM_V7 register volatile uint16 res; register volatile uint16 tmp; NOT_TESTED(); dmb(); __asm__ __volatile__( "1: ldrexh %[tmp], [%[var]] \n\t" "sub %[tmp], %[tmp], %[val] \n\t" "strexh %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [res] "=&r" (res), [tmp] "=&r" (tmp) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); #elif defined VM_ARM_64 _VMATOM_X(OP, 16, TRUE, &var->value, sub, val); #else #error No 16-bits atomics. #endif #endif } /* *----------------------------------------------------------------------------- * * Atomic_Inc16 -- * * Atomic read, increment, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Inc16(Atomic_uint16 *var) // IN/OUT { #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "lock; incw %0" : "+m" (var->value) : : "cc", "memory" ); #elif defined VM_ARM_ANY Atomic_Add16(var, 1); #else #error No 16-bits atomics. #endif #endif } /* *----------------------------------------------------------------------------- * * Atomic_Dec16 -- * * Atomic read, decrement, write. * * Results: * None * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void Atomic_Dec16(Atomic_uint16 *var) // IN/OUT { #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "lock; decw %0" : "+m" (var->value) : : "cc", "memory" ); #elif defined VM_ARM_ANY Atomic_Sub16(var, 1); #else #error No 16-bits atomics. #endif #endif } /* *----------------------------------------------------------------------------- * * Atomic_ReadOr16 -- * * Atomic read (returned), bitwise OR with a value, write. * * Results: * The value of the variable before the operation. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE uint16 Atomic_ReadOr16(Atomic_uint16 *var, // IN/OUT uint16 val) // IN { uint16 res; #if defined VM_ARM_64 res = _VMATOM_X(ROP, 16, TRUE, &var->value, orr, val); #else do { res = var->value; } while (res != Atomic_ReadIfEqualWrite16(var, res, res | val)); #endif return res; } /* *---------------------------------------------------------------------- * * Atomic_ReadAdd16 -- * * Atomically adds a 16-bit integer to another * * Results: * Returns the old value just prior to the addition * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE uint16 Atomic_ReadAdd16(Atomic_uint16 *var, // IN/OUT uint16 val) // IN: { #if defined __GNUC__ #if defined __x86_64__ || defined __i386__ __asm__ __volatile__( "lock; xaddw %0, %1" : "=r" (val), "+m" (var->value) : "0" (val) : "cc", "memory" ); return val; #elif defined VM_ARM_V7 register volatile uint16 res; register volatile uint16 retVal; register volatile uint16 tmp; NOT_TESTED(); dmb(); __asm__ __volatile__( "1: ldrexh %[retVal], [%[var]] \n\t" "add %[tmp], %[val], %[retVal] \n\t" "strexh %[res], %[tmp], [%[var]] \n\t" "teq %[res], #0 \n\t" "bne 1b" : [tmp] "=&r" (tmp), [res] "=&r" (res), [retVal] "=&r" (retVal) : [var] "r" (&var->value), [val] "r" (val) : "cc" ); dmb(); return retVal; #elif defined VM_ARM_64 return _VMATOM_X(ROP, 16, TRUE, &var->value, add, val); #else #error No 16-bits atomics. #endif #endif } /* *---------------------------------------------------------------------- * * Atomic_ReadInc16 -- * * Atomically increments a 64-bit integer * * Results: * Returns the old value just prior to incrementing * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE uint16 Atomic_ReadInc16(Atomic_uint16 *var) // IN/OUT { return Atomic_ReadAdd16(var, 1); } /* *---------------------------------------------------------------------- * * Atomic_ReadDec16 -- * * Atomically decrements a 64-bit integer * * Results: * Returns the old value just prior to decrementing * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE uint16 Atomic_ReadDec16(Atomic_uint16 *var) // IN/OUT { return Atomic_ReadAdd16(var, -1); } #endif /* * Template code for the Atomic_<name> type and its operators. * * The cast argument is an intermediate type cast to make some * compilers stop complaining about casting uint32 <-> void *, * even though we only do it in the 32-bit case so they are always * the same size. So for val of type uint32, instead of * (void *)val, we have (void *)(uintptr_t)val. * The specific problem case is the Windows ddk compiler * (as used by the SVGA driver). -- edward * * NOTE: See the comment in vm_assert.h for why we need UNUSED_TYPE in * AtomicAssertOnCompile(), and why we need to be very careful doing so. */ #define MAKE_ATOMIC_TYPE(name, size, in, out, cast) \ typedef Atomic_uint ## size Atomic_ ## name; \ \ \ static INLINE void \ AtomicAssertOnCompile ## name(void) \ { \ enum { AssertOnCompileMisused = 8 * sizeof (in) == size \ && 8 * sizeof (out) == size \ && 8 * sizeof (cast) == size \ ? 1 : -1 }; \ UNUSED_TYPE(typedef char AssertOnCompileFailed[AssertOnCompileMisused]);\ } \ \ \ static INLINE out \ Atomic_Read ## name(Atomic_ ## name const *var) \ { \ return (out)(cast)Atomic_Read ## size(var); \ } \ \ \ static INLINE void \ Atomic_Write ## name(Atomic_ ## name *var, \ in val) \ { \ Atomic_Write ## size(var, (uint ## size)(cast)val); \ } \ \ \ static INLINE out \ Atomic_ReadWrite ## name(Atomic_ ## name *var, \ in val) \ { \ return (out)(cast)Atomic_ReadWrite ## size(var, \ (uint ## size)(cast)val); \ } \ \ \ static INLINE out \ Atomic_ReadIfEqualWrite ## name(Atomic_ ## name *var, \ in oldVal, \ in newVal) \ { \ return (out)(cast)Atomic_ReadIfEqualWrite ## size(var, \ (uint ## size)(cast)oldVal, (uint ## size)(cast)newVal); \ } \ \ \ static INLINE void \ Atomic_And ## name(Atomic_ ## name *var, \ in val) \ { \ Atomic_And ## size(var, (uint ## size)(cast)val); \ } \ \ \ static INLINE void \ Atomic_Or ## name(Atomic_ ## name *var, \ in val) \ { \ Atomic_Or ## size(var, (uint ## size)(cast)val); \ } \ \ \ static INLINE void \ Atomic_Xor ## name(Atomic_ ## name *var, \ in val) \ { \ Atomic_Xor ## size(var, (uint ## size)(cast)val); \ } \ \ \ static INLINE void \ Atomic_Add ## name(Atomic_ ## name *var, \ in val) \ { \ Atomic_Add ## size(var, (uint ## size)(cast)val); \ } \ \ \ static INLINE void \ Atomic_Sub ## name(Atomic_ ## name *var, \ in val) \ { \ Atomic_Sub ## size(var, (uint ## size)(cast)val); \ } \ \ \ static INLINE void \ Atomic_Inc ## name(Atomic_ ## name *var) \ { \ Atomic_Inc ## size(var); \ } \ \ \ static INLINE void \ Atomic_Dec ## name(Atomic_ ## name *var) \ { \ Atomic_Dec ## size(var); \ } \ \ \ static INLINE out \ Atomic_ReadOr ## name(Atomic_ ## name *var, \ in val) \ { \ return (out)(cast)Atomic_ReadOr ## size(var, (uint ## size)(cast)val); \ } \ \ \ static INLINE out \ Atomic_ReadAdd ## name(Atomic_ ## name *var, \ in val) \ { \ return (out)(cast)Atomic_ReadAdd ## size(var, (uint ## size)(cast)val); \ } \ \ \ static INLINE out \ Atomic_ReadInc ## name(Atomic_ ## name *var) \ { \ return (out)(cast)Atomic_ReadInc ## size(var); \ } \ \ \ static INLINE out \ Atomic_ReadDec ## name(Atomic_ ## name *var) \ { \ return (out)(cast)Atomic_ReadDec ## size(var); \ } /* * Since we use a macro to generate these definitions, it is hard to look for * them. So DO NOT REMOVE THIS COMMENT and keep it up-to-date. --hpreg * * Atomic_Ptr * Atomic_ReadPtr -- * Atomic_WritePtr -- * Atomic_ReadWritePtr -- * Atomic_ReadIfEqualWritePtr -- * Atomic_AndPtr -- * Atomic_OrPtr -- * Atomic_XorPtr -- * Atomic_AddPtr -- * Atomic_SubPtr -- * Atomic_IncPtr -- * Atomic_DecPtr -- * Atomic_ReadOrPtr -- * Atomic_ReadAddPtr -- * Atomic_ReadIncPtr -- * Atomic_ReadDecPtr -- * * Atomic_Int * Atomic_ReadInt -- * Atomic_WriteInt -- * Atomic_ReadWriteInt -- * Atomic_ReadIfEqualWriteInt -- * Atomic_AndInt -- * Atomic_OrInt -- * Atomic_XorInt -- * Atomic_AddInt -- * Atomic_SubInt -- * Atomic_IncInt -- * Atomic_DecInt -- * Atomic_ReadOrInt -- * Atomic_ReadAddInt -- * Atomic_ReadIncInt -- * Atomic_ReadDecInt -- * * Atomic_Bool * Atomic_ReadBool -- * Atomic_WriteBool -- * Atomic_ReadWriteBool -- * Atomic_ReadIfEqualWriteBool -- * Atomic_AndBool -- * Atomic_OrBool -- * Atomic_XorBool -- * Atomic_AddBool -- * Atomic_SubBool -- * Atomic_IncBool -- * Atomic_DecBool -- * Atomic_ReadOrBool -- * Atomic_ReadAddBool -- * Atomic_ReadIncBool -- * Atomic_ReadDecBool -- */ #if defined VM_64BIT MAKE_ATOMIC_TYPE(Ptr, 64, void const *, void *, uintptr_t) #else MAKE_ATOMIC_TYPE(Ptr, 32, void const *, void *, uintptr_t) #endif MAKE_ATOMIC_TYPE(Int, 32, int, int, int) MAKE_ATOMIC_TYPE(Bool, 8, Bool, Bool, Bool) /* * Define arbitrary sized bit vector to be used by * Atomic_TestSetBitVector and Atomic_TestClearBitVector. */ #define ATOMIC_BITVECTOR(varName, capacity) \ Atomic_uint8 varName[CEILING(capacity, 8)] /* *----------------------------------------------------------------------------- * * Atomic_TestSetBitVector -- * * Atomically test and set the bit 'index' in bit vector var. * * The index input value specifies which bit to modify and is 0-based. * * Results: * Returns the value of the bit before modification. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE Bool Atomic_TestSetBitVector(Atomic_uint8 *var, // IN/OUT unsigned index) // IN { #if defined __x86_64__ && defined __GNUC__ Bool bit; __asm__ __volatile__( "lock; bts %2, %1;" "setc %0" : "=qQm" (bit), "+m" (var->value) : "rI" (index) : "cc", "memory" ); return bit; #else uint8 bit = 1 << index % 8; return (Atomic_ReadOr8(var + index / 8, bit) & bit) != 0; #endif } /* *----------------------------------------------------------------------------- * * Atomic_TestClearBitVector -- * * Atomically test and clear the bit 'index' in bit vector var. * * The index input value specifies which bit to modify and is 0-based. * * Results: * Returns the value of the bit before modification. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE Bool Atomic_TestClearBitVector(Atomic_uint8 *var, // IN/OUT unsigned index) // IN { #if defined __x86_64__ && defined __GNUC__ Bool bit; __asm__ __volatile__( "lock; btr %2, %1;" "setc %0" : "=qQm" (bit), "+m" (var->value) : "rI" (index) : "cc", "memory" ); return bit; #else uint8 bit = 1 << index % 8; return (Atomic_ReadAnd8(var + index / 8, ~bit) & bit) != 0; #endif } /* *----------------------------------------------------------------------------- * * Atomic_TestBitVector -- * * Test the bit 'index' (zero-based) in bit vector var. *----------------------------------------------------------------------------- */ static INLINE Bool Atomic_TestBitVector(const Atomic_uint8 *var, // IN unsigned index) // IN { uint8 bit = 1 << index % 8; return (Atomic_Read8(var + index / 8) & bit) != 0; } #ifdef VM_ARM_64 # include "vm_atomic_arm64_end.h" #endif #if defined __cplusplus } // extern "C" #endif #endif // ifndef _ATOMIC_H_ vsock-only/shared/vm_basic_types.h 0000444 0000000 0000000 00000070343 13432725350 016317 0 ustar root root /********************************************************* * Copyright (C) 1998-2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * * vm_basic_types.h -- * * basic data types. */ #ifndef _VM_BASIC_TYPES_H_ #define _VM_BASIC_TYPES_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_VMKDRIVERS #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" /* * Macros __i386__ and __ia64 are intrinsically defined by GCC */ #if defined _MSC_VER && defined _M_X64 # define __x86_64__ #elif defined _MSC_VER && defined _M_IX86 # define __i386__ #endif /* * Setup a bunch of defines for instruction set architecture (ISA) related * properties. * * For compiler types/size: * * - VM_32BIT for a 32-bit ISA (with the same C compiler types/sizes as 32-bit * x86/ARM). * - VM_64BIT for a 64-bit ISA (with the same C compiler types/sizes as 64-bit * x86/ARM). * * For a given <arch> in {X86, ARM}: * * - VM_<arch>_32 for the 32-bit variant. * - VM_<arch>_64 for the 64-bit variant. * - VM_<arch>_ANY for any variant of <arch>. * * VM_X86_ANY is synonymous with the confusing and deprecated VM_I386 (which * should really be VM_X86_32). */ #ifdef __i386__ /* * VM_I386 is historically synonymous with VM_X86_ANY in bora, but misleading, * since it is confused with the __i386__ gcc but defined for both 32- and * 64-bit x86. We retain it here for legacy compatibility. */ #define VM_I386 #define VM_X86_32 #define VM_X86_ANY #define VM_32BIT #endif #ifdef __x86_64__ #define VM_X86_64 #define vm_x86_64 1 #define VM_I386 #define VM_X86_ANY #define VM_64BIT #else #define vm_x86_64 0 #endif #ifdef __arm__ #define VM_ARM_32 #define VM_ARM_ANY #define VM_32BIT #endif #ifdef __aarch64__ #define VM_ARM_64 #define vm_arm_64 1 #define VM_ARM_ANY #define VM_64BIT #else #define vm_arm_64 0 #endif #define vm_64bit (sizeof (void *) == 8) #ifdef _MSC_VER #pragma warning (3 :4505) // unreferenced local function #pragma warning (disable :4018) // signed/unsigned mismatch #pragma warning (disable :4761) // integral size mismatch in argument; conversion supplied #pragma warning (disable :4305) // truncation from 'const int' to 'short' #pragma warning (disable :4244) // conversion from 'unsigned short' to 'unsigned char' #pragma warning (disable :4267) // truncation of 'size_t' #pragma warning (disable :4146) // unary minus operator applied to unsigned type, result still unsigned #pragma warning (disable :4142) // benign redefinition of type #endif #if defined(__cplusplus) && __cplusplus >= 201103L || \ defined(__STDC_VERSION__) && __STDC_VERSION__ >= 199901L || \ defined(__APPLE__) || defined(HAVE_STDINT_H) /* * We're using <stdint.h> instead of <cstdint> below because some C++ code * deliberately compiles without C++ include paths. */ #include <stdint.h> typedef uint64_t uint64; typedef int64_t int64; typedef uint32_t uint32; typedef int32_t int32; typedef uint16_t uint16; typedef int16_t int16; typedef uint8_t uint8; typedef int8_t int8; #else /* !HAVE_STDINT_H */ /* Pre-c99 or pre-c++11; use compiler extension to get 64-bit types */ #ifdef _MSC_VER typedef unsigned __int64 uint64; typedef signed __int64 int64; #elif __GNUC__ # if defined(VM_X86_64) || defined(VM_ARM_64) typedef unsigned long uint64; typedef long int64; # else /* * Only strict c90 (without extensions) lacks a 'long long' type. * If this declaration fails ... use -std=c99 or -std=gnu90. */ typedef unsigned long long uint64; typedef long long int64; # endif #else # error - Need compiler define for int64/uint64 #endif /* _MSC_VER */ typedef unsigned int uint32; typedef unsigned short uint16; typedef unsigned char uint8; typedef int int32; typedef short int16; typedef signed char int8; #endif /* HAVE_STDINT_H */ /* * The _XTYPEDEF_BOOL guard prevents colliding with: * <X11/Xlib.h> #define Bool int * <X11/Xdefs.h> typedef int Bool; * If using this header AND X11 headers, be sure to #undef Bool and * be careful about the different size. */ #if !defined(_XTYPEDEF_BOOL) #define _XTYPEDEF_BOOL /* * C does not specify whether char is signed or unsigned, and * both gcc and msvc implement it as a non-signed, non-unsigned type. * Thus, (uint8_t *)&Bool and (int8_t *)&Bool are possible compile errors. * This is intentional. */ typedef char Bool; #endif #ifndef FALSE #define FALSE 0 #endif #ifndef TRUE #define TRUE 1 #endif #define IS_BOOL(x) (((x) & ~1) == 0) /* * FreeBSD (for the tools build) unconditionally defines these in * sys/inttypes.h so don't redefine them if this file has already * been included. [greg] * * This applies to Solaris as well. */ /* * Before trying to do the includes based on OS defines, see if we can use * feature-based defines to get as much functionality as possible */ #ifdef HAVE_INTTYPES_H #include <inttypes.h> #endif #ifdef HAVE_SYS_TYPES_H #include <sys/types.h> #endif #ifdef HAVE_SYS_INTTYPES_H #include <sys/inttypes.h> #endif #ifdef HAVE_STDLIB_H #include <stdlib.h> #endif #ifdef __FreeBSD__ #include <sys/param.h> /* For __FreeBSD_version */ #endif #if !defined(USING_AUTOCONF) # if defined(__FreeBSD__) || defined(sun) # ifdef KLD_MODULE # include <sys/types.h> # else # if __FreeBSD_version >= 500043 # if !defined(VMKERNEL) # include <inttypes.h> # endif # include <sys/types.h> # else # include <sys/inttypes.h> # endif # endif # elif defined __APPLE__ # if KERNEL # include <sys/unistd.h> # include <sys/types.h> /* mostly for size_t */ # include <stdint.h> # else # include <unistd.h> # include <inttypes.h> # include <stdlib.h> # include <stdint.h> # endif # elif defined __ANDROID__ # include <stdint.h> # else # if !defined(__intptr_t_defined) && !defined(intptr_t) # ifdef VM_I386 # define __intptr_t_defined # if defined(VM_X86_64) typedef int64 intptr_t; # else typedef int32 intptr_t; # endif # elif defined(VM_ARM_64) # define __intptr_t_defined typedef int64 intptr_t; # elif defined(__arm__) # define __intptr_t_defined typedef int32 intptr_t; # endif # endif # ifndef _STDINT_H # ifdef VM_I386 # if defined(VM_X86_64) typedef uint64 uintptr_t; # else typedef uint32 uintptr_t; # endif # elif defined(VM_ARM_64) typedef uint64 uintptr_t; # elif defined(__arm__) typedef uint32 uintptr_t; # endif # endif # endif #endif #if defined(__GNUC__) && defined(__SIZEOF_INT128__) typedef unsigned __int128 uint128; typedef __int128 int128; #define MIN_INT128 ((int128)1 << 127) #define MAX_INT128 (~MIN_INT128) #define MIN_UINT128 ((uint128)0) #define MAX_UINT128 (~MIN_UINT128) #endif /* * Time * XXX These should be cleaned up. -- edward */ typedef int64 VmTimeType; /* Time in microseconds */ typedef int64 VmTimeRealClock; /* Real clock kept in microseconds */ typedef int64 VmTimeVirtualClock; /* Virtual Clock kept in CPU cycles */ /* * Printf format specifiers for size_t and 64-bit number. * Use them like this: * printf("%" FMT64 "d\n", big); * The spaces are important for C++11 compatibility. * * FMTH is for handles/fds. */ #ifdef _MSC_VER /* MSVC added C99-compatible formatting in vs2015. */ #define FMT64 "I64" #ifdef VM_X86_64 #define FMTSZ "I64" #define FMTPD "I64" #define FMTH "I64" #else #define FMTSZ "I" #define FMTPD "I" #define FMTH "I" #endif #elif defined __APPLE__ /* macOS hosts use the same formatters for 32- and 64-bit. */ #define FMT64 "ll" #if KERNEL /* macOS osfmk/kern added 'z' length specifier in 10.13 */ #define FMTSZ "l" #else #define FMTSZ "z" #endif #define FMTPD "l" #define FMTH "" #elif defined __GNUC__ /* * Every POSIX system we target has C99-compatible printf * (supports 'z' for size_t and 'll' for long long). */ #define FMTH "" #define FMTSZ "z" #if defined(VM_X86_64) || defined(VM_ARM_64) #define FMT64 "l" #define FMTPD "l" #else #define FMT64 "ll" #define FMTPD "" #endif #else #error - Need compiler define for FMT64 and FMTSZ #endif /* * Suffix for 64-bit constants. Use it like this: * CONST64(0x7fffffffffffffff) for signed or * CONST64U(0x7fffffffffffffff) for unsigned. * * 2004.08.30(thutt): * The vmcore/asm64/gen* programs are compiled as 32-bit * applications, but must handle 64 bit constants. If the * 64-bit-constant defining macros are already defined, the * definition will not be overwritten. */ #if !defined(CONST64) || !defined(CONST64U) #ifdef _MSC_VER #define CONST64(c) c##I64 #define CONST64U(c) c##uI64 #elif defined __APPLE__ #define CONST64(c) c##LL #define CONST64U(c) c##uLL #elif __GNUC__ #if defined(VM_X86_64) || defined(VM_ARM_64) #define CONST64(c) c##L #define CONST64U(c) c##uL #else #define CONST64(c) c##LL #define CONST64U(c) c##uLL #endif #else #error - Need compiler define for CONST64 #endif #endif /* * Use CONST3264/CONST3264U if you want a constant to be * treated as a 32-bit number on 32-bit compiles and * a 64-bit number on 64-bit compiles. Useful in the case * of shifts, like (CONST3264U(1) << x), where x could be * more than 31 on a 64-bit compile. */ #if defined(VM_X86_64) || defined(VM_ARM_64) #define CONST3264(a) CONST64(a) #define CONST3264U(a) CONST64U(a) #else #define CONST3264(a) (a) #define CONST3264U(a) (a) #endif #define MIN_INT8 ((int8)0x80) #define MAX_INT8 ((int8)0x7f) #define MIN_UINT8 ((uint8)0) #define MAX_UINT8 ((uint8)0xff) #define MIN_INT16 ((int16)0x8000) #define MAX_INT16 ((int16)0x7fff) #define MIN_UINT16 ((uint16)0) #define MAX_UINT16 ((uint16)0xffff) #define MIN_INT32 ((int32)0x80000000) #define MAX_INT32 ((int32)0x7fffffff) #define MIN_UINT32 ((uint32)0) #define MAX_UINT32 ((uint32)0xffffffff) #define MIN_INT64 (CONST64(0x8000000000000000)) #define MAX_INT64 (CONST64(0x7fffffffffffffff)) #define MIN_UINT64 (CONST64U(0)) #define MAX_UINT64 (CONST64U(0xffffffffffffffff)) typedef uint8 *TCA; /* Pointer into TC (usually). */ /* * Type big enough to hold an integer between 0..100 */ typedef uint8 Percent; #define AsPercent(v) ((Percent)(v)) typedef uintptr_t VA; typedef uintptr_t VPN; typedef uint64 PA; typedef uint32 PPN; typedef uint64 TPA; typedef uint32 TPPN; typedef uint64 PhysMemOff; typedef uint64 PhysMemSize; typedef uint64 BA; #ifdef VMKERNEL typedef void *BPN; #else typedef uint64 BPN; #endif #define UINT64_2_BPN(u) ((BPN)(u)) #define BPN_2_UINT64(b) ((uint64)(b)) typedef uint32 PageCnt; typedef uint64 PgCnt64; typedef uint32 PageNum; typedef uint64 PgNum64; typedef unsigned MemHandle; typedef unsigned int IoHandle; typedef int32 World_ID; /* !! do not alter the definition of INVALID_WORLD_ID without ensuring * that the values defined in both bora/public/vm_basic_types.h and * lib/vprobe/vm_basic_types.h are the same. Additionally, the definition * of VMK_INVALID_WORLD_ID in vmkapi_world.h also must be defined with * the same value */ #define INVALID_WORLD_ID ((World_ID)0) typedef World_ID User_CartelID; #define INVALID_CARTEL_ID INVALID_WORLD_ID typedef User_CartelID User_SessionID; #define INVALID_SESSION_ID INVALID_CARTEL_ID typedef User_CartelID User_CartelGroupID; #define INVALID_CARTELGROUP_ID INVALID_CARTEL_ID typedef uint32 Worldlet_ID; #define INVALID_WORLDLET_ID ((Worldlet_ID)-1) typedef int8 Reg8; typedef int16 Reg16; typedef int32 Reg32; typedef int64 Reg64; typedef uint8 UReg8; typedef uint16 UReg16; typedef uint32 UReg32; typedef uint64 UReg64; #if defined(__GNUC__) && defined(__SIZEOF_INT128__) typedef int128 Reg128; typedef uint128 UReg128; #endif #if defined(VMM) || defined(COREQUERY) || defined(EXTDECODER) || \ defined (VMKERNEL) || defined (VMKBOOT) typedef Reg64 Reg; typedef UReg64 UReg; #endif typedef uint64 MA; typedef uint32 MPN32; /* * This type should be used for variables that contain sector * position/quantity. */ typedef uint64 SectorType; /* * Linear address */ typedef uintptr_t LA; typedef uintptr_t LPN; #define LA_2_LPN(_la) ((_la) >> PAGE_SHIFT) #define LPN_2_LA(_lpn) ((_lpn) << PAGE_SHIFT) #define LAST_LPN ((((LA) 1) << (8 * sizeof(LA) - PAGE_SHIFT)) - 1) #define LAST_LPN32 ((((LA32)1) << (8 * sizeof(LA32) - PAGE_SHIFT)) - 1) #define LAST_LPN64 ((((LA64)1) << (8 * sizeof(LA64) - PAGE_SHIFT)) - 1) /* Valid bits in a LPN. */ #define LPN_MASK LAST_LPN #define LPN_MASK32 LAST_LPN32 #define LPN_MASK64 LAST_LPN64 /* * On 64 bit platform, address and page number types default * to 64 bit. When we need to represent a 32 bit address, we use * types defined below. * * On 32 bit platform, the following types are the same as the * default types. */ typedef uint32 VA32; typedef uint32 VPN32; typedef uint32 LA32; typedef uint32 LPN32; typedef uint32 PA32; typedef uint32 PPN32; /* * On 64 bit platform, the following types are the same as the * default types. */ typedef uint64 VA64; typedef uint64 VPN64; typedef uint64 LA64; typedef uint64 LPN64; typedef uint64 PA64; typedef uint64 PPN64; typedef uint64 TPPN64; typedef uint64 MA64; typedef uint64 MPN; /* * Remove after PPN->PPN64 conversion is finished. */ #define PPN64_2_PPN(_ppn) ((PPN)_ppn) #define FMTPPN "" /* * IO device DMA virtual address and page number (translated by IOMMU to * MA/MPN). IOPN can be in the inclusive range 0 -> MAX_IOPN. */ typedef uint64 IOA; typedef uint64 IOPN; /* * VA typedefs for user world apps. */ typedef VA32 UserVA32; typedef VA64 UserVA64; typedef UserVA64 UserVAConst; /* Userspace ptr to data that we may only read. */ typedef UserVA32 UserVA32Const; /* Userspace ptr to data that we may only read. */ typedef UserVA64 UserVA64Const; /* Used by 64-bit syscalls until conversion is finished. */ #ifdef VMKERNEL typedef UserVA64 UserVA; #else typedef void * UserVA; #endif #define MAX_PPN_BITS 31 #define MAX_PPN (((PPN)1 << MAX_PPN_BITS) - 1) /* Maximal observable PPN value. */ #define INVALID_PPN ((PPN)0xffffffff) #define INVALID_PPN32 ((PPN32)0xffffffff) #define INVALID_PPN64 ((PPN64)0xffffffffffffffffull) #define APIC_INVALID_PPN ((PPN)0xfffffffe) #define INVALID_BPN ((BPN)0x000000ffffffffffull) #define MPN38_MASK ((1ull << 38) - 1) #define RESERVED_MPN ((MPN)0) #define INVALID_MPN ((MPN)MPN38_MASK) #define MEMREF_MPN ((MPN)MPN38_MASK - 1) #define RELEASED_MPN ((MPN)MPN38_MASK - 2) /* account for special MPNs defined above */ #define MAX_MPN ((MPN)MPN38_MASK - 3) /* 50 bits of address space */ #define INVALID_IOPN ((IOPN)-1) #define MAX_IOPN (INVALID_IOPN - 1) #define INVALID_LPN ((LPN)-1) #define INVALID_VPN ((VPN)-1) #define INVALID_LPN64 ((LPN64)-1) #define INVALID_PAGENUM ((PageNum)-1) /* * Format modifier for printing VA, LA, and VPN. * Use them like this: Log("%#" FMTLA "x\n", laddr) */ #if defined(VMM) || defined(FROBOS64) || vm_x86_64 || vm_arm_64 || defined __APPLE__ # define FMTLA "l" # define FMTVA "l" # define FMTVPN "l" #else # define FMTLA "" # define FMTVA "" # define FMTVPN "" #endif #ifndef EXTERN #define EXTERN extern #endif #define CONST const #ifndef INLINE # ifdef _MSC_VER /* * On UWP(Universal Windows Platform), * Only X86 32bit support '__inline' */ # if defined(VM_WIN_UWP) && !defined(VM_X86_32) # define INLINE # else # define INLINE __inline # endif # else # define INLINE inline # endif #endif /* * Annotation for data that may be exported into a DLL and used by other * apps that load that DLL and import the data. */ #if defined(_WIN32) && defined(VMX86_IMPORT_DLLDATA) # define VMX86_EXTERN_DATA extern __declspec(dllimport) #else // !_WIN32 # define VMX86_EXTERN_DATA extern #endif #ifdef _WIN32 /* under windows, __declspec(thread) is supported since VS 2003 */ #define __thread __declspec(thread) #else /* * under other platforms instead, __thread is supported by gcc since * version 3.3.1 and by clang since version 3.x */ #endif /* * Due to the wonderful "registry redirection" feature introduced in * 64-bit Windows, if you access any key under HKLM\Software in 64-bit * code, you need to open/create/delete that key with * VMKEY_WOW64_32KEY if you want a consistent view with 32-bit code. */ #ifdef _WIN32 #ifdef _WIN64 #define VMW_KEY_WOW64_32KEY KEY_WOW64_32KEY #else #define VMW_KEY_WOW64_32KEY 0x0 #endif #endif /* * At present, we effectively require a compiler that is at least * gcc-3.3 (circa 2003). Enforce this here, various things below * this line depend upon it. * * In practice, most things presently compile with gcc-4.1 or gcc-4.4. * The various linux kernel modules may use older (gcc-3.3) compilers. */ #if defined __GNUC__ && (__GNUC__ < 3 || (__GNUC__ == 3 && __GNUC_MINOR__ < 3)) #error "gcc version is too old to compile assembly, need gcc-3.3 or better" #endif /* * Similarly, we require a compiler that is at least vc80 (vs2005). * Enforce this here. */ #if defined _MSC_VER && _MSC_VER < 1400 #error "cl.exe version is too old, need vc80 or better" #endif /* * Consider the following reasons functions are inlined: * * 1) inlined for performance reasons * 2) inlined because it's a single-use function * * Functions which meet only condition 2 should be marked with this * inline macro; It is not critical to be inlined (but there is a * code-space & runtime savings by doing so), so when other callers * are added the inline-ness should be removed. */ #if defined __GNUC__ /* * Starting at version 3.3, gcc does not always inline functions marked * 'inline' (it depends on their size and other factors). To force gcc * to inline a function, one must use the __always_inline__ attribute. * This attribute should be used sparingly and with care. It is usually * preferable to let gcc make its own inlining decisions */ # define INLINE_ALWAYS INLINE __attribute__((__always_inline__)) #else # define INLINE_ALWAYS INLINE #endif #define INLINE_SINGLE_CALLER INLINE_ALWAYS /* * Used when a hard guarantee of no inlining is needed. Very few * instances need this since the absence of INLINE is a good hint * that gcc will not do inlining. */ #if defined(__GNUC__) #define ABSOLUTELY_NOINLINE __attribute__((__noinline__)) #elif defined(_MSC_VER) #define ABSOLUTELY_NOINLINE __declspec(noinline) #endif /* * Used when a function has no effects except the return value and the * return value depends only on the parameters and/or global variables * Such a function can be subject to common subexpression elimination * and loop optimization just as an arithmetic operator would be. */ #if defined(__GNUC__) && (defined(VMM) || defined (VMKERNEL)) #define SIDE_EFFECT_FREE __attribute__((__pure__)) #else #define SIDE_EFFECT_FREE #endif /* * Used when a function exmaines no input other than its arguments and * has no side effects other than its return value. Stronger than * SIDE_EFFECT_FREE as the function is not allowed to read from global * memory. */ #if defined(__GNUC__) && (defined(VMM) || defined (VMKERNEL)) #define CONST_FUNCTION __attribute__((__const__)) #else #define CONST_FUNCTION #endif /* * Attributes placed on function declarations to tell the compiler * that the function never returns. */ #ifdef _MSC_VER #define NORETURN __declspec(noreturn) #elif defined __GNUC__ #define NORETURN __attribute__((__noreturn__)) #else #define NORETURN #endif /* * Static profiling hints for functions. * A function can be either hot, cold, or neither. * It is an error to specify both hot and cold for the same function. * Note that there is no annotation for "neither." */ #if defined __GNUC__ && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3)) #define HOT __attribute__((hot)) #define COLD __attribute__((cold)) #else #define HOT #define COLD #endif /* * Branch prediction hints: * LIKELY(exp) - Expression exp is likely TRUE. * UNLIKELY(exp) - Expression exp is likely FALSE. * Usage example: * if (LIKELY(excCode == EXC_NONE)) { * or * if (UNLIKELY(REAL_MODE(vc))) { * * We know how to predict branches on gcc3 and later (hopefully), * all others we don't so we do nothing. */ #if defined __GNUC__ /* * gcc3 uses __builtin_expect() to inform the compiler of an expected value. * We use this to inform the static branch predictor. The '!!' in LIKELY * will convert any !=0 to a 1. */ #define LIKELY(_exp) __builtin_expect(!!(_exp), 1) #define UNLIKELY(_exp) __builtin_expect((_exp), 0) #else #define LIKELY(_exp) (_exp) #define UNLIKELY(_exp) (_exp) #endif /* * GCC's argument checking for printf-like functions * This is conditional until we have replaced all `"%x", void *' * with `"0x%08x", (uint32) void *'. Note that %p prints different things * on different platforms. Argument checking is enabled for the * vmkernel, which has already been cleansed. * * fmtPos is the position of the format string argument, beginning at 1 * varPos is the position of the variable argument, beginning at 1 */ #if defined(__GNUC__) # define PRINTF_DECL(fmtPos, varPos) __attribute__((__format__(__printf__, fmtPos, varPos))) #else # define PRINTF_DECL(fmtPos, varPos) #endif #if defined(__GNUC__) # define SCANF_DECL(fmtPos, varPos) __attribute__((__format__(__scanf__, fmtPos, varPos))) #else # define SCANF_DECL(fmtPos, varPos) #endif /* * UNUSED_PARAM should surround the parameter name and type declaration, * e.g. "int MyFunction(int var1, UNUSED_PARAM(int var2))" * */ #ifndef UNUSED_PARAM # if defined(__GNUC__) # define UNUSED_PARAM(_parm) _parm __attribute__((__unused__)) # elif defined _MSC_VER # define UNUSED_PARAM(_parm) __pragma(warning(suppress:4100)) _parm # else # define UNUSED_PARAM(_parm) _parm # endif #endif #ifndef UNUSED_TYPE // XXX _Pragma would better but doesn't always work right now. # define UNUSED_TYPE(_parm) UNUSED_PARAM(_parm) #endif #ifndef UNUSED_VARIABLE // XXX is there a better way? # define UNUSED_VARIABLE(_var) (void)_var #endif /* * gcc can warn us if we're ignoring returns */ #if defined(__GNUC__) # define MUST_CHECK_RETURN __attribute__((warn_unused_result)) #else # define MUST_CHECK_RETURN #endif /* * ALIGNED specifies minimum alignment in "n" bytes. */ #ifdef __GNUC__ #define ALIGNED(n) __attribute__((__aligned__(n))) #else #define ALIGNED(n) #endif /* * Encapsulate the syntactic differences between gcc and msvc alignment control. * BOUNDARY must match in the prefix and suffix. */ #ifdef _WIN32 #define ALIGN_PREFIX(BOUNDRY) __declspec(align(BOUNDRY)) #define ALIGN_SUFFIX(BOUNDRY) #else #define ALIGN_PREFIX(BOUNDRY) #define ALIGN_SUFFIX(BOUNDRY) __attribute__((__aligned__(BOUNDRY))) #endif /* * Once upon a time, this was used to silence compiler warnings that * get generated when the compiler thinks that a function returns * when it is marked noreturn. Don't do it. Use NOT_REACHED(). */ #define INFINITE_LOOP() do { } while (1) /* * On FreeBSD (for the tools build), size_t is typedef'd if _BSD_SIZE_T_ * is defined. Use the same logic here so we don't define it twice. [greg] */ #ifdef __FreeBSD__ # ifdef _BSD_SIZE_T_ # undef _BSD_SIZE_T_ # ifdef VM_I386 # ifdef VM_X86_64 typedef uint64 size_t; # else typedef uint32 size_t; # endif # endif /* VM_I386 */ # endif # ifdef _BSD_SSIZE_T_ # undef _BSD_SSIZE_T_ # ifdef VM_I386 # ifdef VM_X86_64 typedef int64 ssize_t; # else typedef int32 ssize_t; # endif # endif /* VM_I386 */ # endif #else # if !defined(_SIZE_T) && !defined(_SIZE_T_DEFINED) # ifdef VM_I386 # define _SIZE_T # ifdef VM_X86_64 typedef uint64 size_t; # else typedef uint32 size_t; # endif # elif defined(VM_ARM_64) # define _SIZE_T typedef uint64 size_t; # elif defined(__arm__) # define _SIZE_T typedef uint32 size_t; # endif # endif # if !defined(FROBOS) && !defined(_SSIZE_T) && !defined(_SSIZE_T_) && \ !defined(ssize_t) && !defined(__ssize_t_defined) && \ !defined(_SSIZE_T_DECLARED) && !defined(_SSIZE_T_DEFINED) && \ !defined(_SSIZE_T_DEFINED_) # ifdef VM_I386 # define _SSIZE_T # define __ssize_t_defined # define _SSIZE_T_DECLARED # define _SSIZE_T_DEFINED_ # ifdef VM_X86_64 typedef int64 ssize_t; # else typedef int32 ssize_t; # endif # elif defined(VM_ARM_64) # define _SSIZE_T # define __ssize_t_defined # define _SSIZE_T_DECLARED # define _SSIZE_T_DEFINED_ typedef int64 ssize_t; # elif defined(__arm__) # define _SSIZE_T # define __ssize_t_defined # define _SSIZE_T_DECLARED # define _SSIZE_T_DEFINED_ typedef int32 ssize_t; # endif # endif #endif /* * Format modifier for printing pid_t. On sun the pid_t is a ulong, but on * Linux it's an int. * Use this like this: printf("The pid is %" FMTPID ".\n", pid); */ #ifdef sun # ifdef VM_X86_64 # define FMTPID "d" # else # define FMTPID "lu" # endif #else # define FMTPID "d" #endif /* * Format modifier for printing uid_t. On Solaris 10 and earlier, uid_t * is a ulong, but on other platforms it's an unsigned int. * Use this like this: printf("The uid is %" FMTUID ".\n", uid); */ #if defined(sun) && !defined(SOL11) # ifdef VM_X86_64 # define FMTUID "u" # else # define FMTUID "lu" # endif #else # define FMTUID "u" #endif /* * Format modifier for printing mode_t. On sun the mode_t is a ulong, but on * Linux it's an int. * Use this like this: printf("The mode is %" FMTMODE ".\n", mode); */ #ifdef sun # ifdef VM_X86_64 # define FMTMODE "o" # else # define FMTMODE "lo" # endif #else # define FMTMODE "o" #endif #ifdef __APPLE__ /* * Format specifier for all these annoying types such as {S,U}Int32 * which are 'long' in 32-bit builds * and 'int' in 64-bit builds. */ # ifdef __LP64__ # define FMTLI "" # else # define FMTLI "l" # endif /* * Format specifier for all these annoying types such as NS[U]Integer * which are 'int' in 32-bit builds * and 'long' in 64-bit builds. */ # ifdef __LP64__ # define FMTIL "l" # else # define FMTIL "" # endif #endif /* * Define type for poll device handles. */ typedef int64 PollDevHandle; /* * Define the utf16_t type. */ #if defined(_WIN32) && defined(_NATIVE_WCHAR_T_DEFINED) typedef wchar_t utf16_t; #else typedef uint16 utf16_t; #endif /* * Define for point and rectangle types. Defined here so they * can be used by other externally facing headers in bora/public. */ typedef struct VMPoint { int x, y; } VMPoint; #if defined _WIN32 && defined USERLEVEL struct tagRECT; typedef struct tagRECT VMRect; #else typedef struct VMRect { int left; int top; int right; int bottom; } VMRect; #endif /* * ranked locks "everywhere" */ typedef uint32 MX_Rank; #endif /* _VM_BASIC_TYPES_H_ */ vsock-only/shared/vmware_pack_begin.h 0000444 0000000 0000000 00000002451 13432725350 016746 0 ustar root root /********************************************************* * Copyright (C) 2002-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmware_pack_begin.h -- * * Begin of structure packing. See vmware_pack_init.h for details. * * Note that we do not use the following construct in this include file, * because we want to emit the code every time the file is included --hpreg * * #ifndef foo * # define foo * ... * #endif * */ #include "vmware_pack_init.h" #ifdef _MSC_VER # pragma pack(push, 1) #elif __GNUC__ #else # error Compiler packing... #endif vsock-only/shared/backdoor_types.h 0000444 0000000 0000000 00000006773 13432725350 016326 0 ustar root root /********************************************************* * Copyright (C) 1999-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * backdoor_types.h -- * * Type definitions for backdoor interaction code. */ #ifndef _BACKDOOR_TYPES_H_ #define _BACKDOOR_TYPES_H_ #ifndef VM_I386 #error The backdoor protocol is only supported on x86 architectures. #endif /* * These #defines are intended for defining register structs as part of * existing named unions. If the union should encapsulate the register * (and nothing else), use DECLARE_REG_NAMED_STRUCT defined below. */ #define DECLARE_REG32_STRUCT \ struct { \ uint16 low; \ uint16 high; \ } halfs; \ uint32 word #define DECLARE_REG64_STRUCT \ DECLARE_REG32_STRUCT; \ struct { \ uint32 low; \ uint32 high; \ } words; \ uint64 quad #if defined (VM_X86_64) || defined (VM_ARM_64) #define DECLARE_REG_STRUCT DECLARE_REG64_STRUCT #else #define DECLARE_REG_STRUCT DECLARE_REG32_STRUCT #endif #define DECLARE_REG_NAMED_STRUCT(_r) \ union { DECLARE_REG_STRUCT; } _r /* * Some of the registers are expressed by semantic name, because if they were * expressed as register structs declared above, we could only address them * by fixed size (half-word, word, quad, etc.) instead of by varying size * (size_t, uintptr_t). * * To be cleaner, these registers are expressed ONLY by semantic name, * rather than by a union of the semantic name and a register struct. */ typedef union { struct { DECLARE_REG_NAMED_STRUCT(ax); size_t size; /* Register bx. */ DECLARE_REG_NAMED_STRUCT(cx); DECLARE_REG_NAMED_STRUCT(dx); DECLARE_REG_NAMED_STRUCT(si); DECLARE_REG_NAMED_STRUCT(di); } in; struct { DECLARE_REG_NAMED_STRUCT(ax); DECLARE_REG_NAMED_STRUCT(bx); DECLARE_REG_NAMED_STRUCT(cx); DECLARE_REG_NAMED_STRUCT(dx); DECLARE_REG_NAMED_STRUCT(si); DECLARE_REG_NAMED_STRUCT(di); } out; } Backdoor_proto; typedef union { struct { DECLARE_REG_NAMED_STRUCT(ax); DECLARE_REG_NAMED_STRUCT(bx); size_t size; /* Register cx. */ DECLARE_REG_NAMED_STRUCT(dx); uintptr_t srcAddr; /* Register si. */ uintptr_t dstAddr; /* Register di. */ DECLARE_REG_NAMED_STRUCT(bp); } in; struct { DECLARE_REG_NAMED_STRUCT(ax); DECLARE_REG_NAMED_STRUCT(bx); DECLARE_REG_NAMED_STRUCT(cx); DECLARE_REG_NAMED_STRUCT(dx); DECLARE_REG_NAMED_STRUCT(si); DECLARE_REG_NAMED_STRUCT(di); DECLARE_REG_NAMED_STRUCT(bp); } out; } Backdoor_proto_hb; MY_ASSERTS(BACKDOOR_STRUCT_SIZES, ASSERT_ON_COMPILE(sizeof(Backdoor_proto) == 6 * sizeof(uintptr_t)); ASSERT_ON_COMPILE(sizeof(Backdoor_proto_hb) == 7 * sizeof(uintptr_t)); ) #undef DECLARE_REG_STRUCT #endif /* _BACKDOOR_TYPES_H_ */ vsock-only/shared/compat_workqueue.h 0000444 0000000 0000000 00000014361 13432725347 016706 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_WORKQUEUE_H__ # define __COMPAT_WORKQUEUE_H__ #include <linux/kernel.h> #if LINUX_VERSION_CODE > KERNEL_VERSION(2, 5, 41) # include <linux/workqueue.h> #endif /* * * Work queues and delayed work queues. * * Prior to 2.5.41, the notion of work queues did not exist. Taskqueues are * used for work queues and timers are used for delayed work queues. * * After 2.6.20, normal work structs ("work_struct") and delayed work * ("delayed_work") structs were separated so that the work_struct could be * slimmed down. The interface was also changed such that the address of the * work_struct itself is passed in as the argument to the work function. This * requires that one embed the work struct in the larger struct containing the * information necessary to complete the work and use container_of() to obtain * the address of the containing structure. * * Users of these macros should embed a compat_work or compat_delayed_work in * a larger structure, then specify the larger structure as the _data argument * for the initialization functions, specify the work function to take * a compat_work_arg or compat_delayed_work_arg, then use the appropriate * _GET_DATA macro to obtain the reference to the structure passed in as _data. * An example is below. * * * typedef struct WorkData { * int data; * compat_work work; * } WorkData; * * * void * WorkFunc(compat_work_arg data) * { * WorkData *workData = COMPAT_WORK_GET_DATA(data, WorkData, work); * * ... * } * * * { * WorkData *workData = kmalloc(sizeof *workData, GFP_EXAMPLE); * if (!workData) { * return -ENOMEM; * } * * COMPAT_INIT_WORK(&workData->work, WorkFunc, workData); * compat_schedule_work(&workData->work); * } */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 41) /* { */ typedef struct tq_struct compat_work; typedef struct compat_delayed_work { struct tq_struct work; struct timer_list timer; } compat_delayed_work; typedef void * compat_work_arg; typedef void * compat_delayed_work_arg; /* * Delayed work queues need to run at some point in the future in process * context, but task queues don't support delaying the task one is scheduling. * Timers allow us to delay the execution of our work queue until the future, * but timer handlers run in bottom-half context. As such, we use both a timer * and task queue and use the timer handler below to schedule the task in * process context immediately. The timer lets us delay execution, and the * task queue lets us run in process context. * * Note that this is similar to how delayed_work is implemented with work * queues in later kernel versions. */ static inline void __compat_delayed_work_timer(unsigned long arg) { compat_delayed_work *dwork = (compat_delayed_work *)arg; if (dwork) { schedule_task(&dwork->work); } } # define COMPAT_INIT_WORK(_work, _func, _data) \ INIT_LIST_HEAD(&(_work)->list); \ (_work)->sync = 0; \ (_work)->routine = _func; \ (_work)->data = _data # define COMPAT_INIT_DELAYED_WORK(_work, _func, _data) \ COMPAT_INIT_WORK(&(_work)->work, _func, _data); \ init_timer(&(_work)->timer); \ (_work)->timer.expires = 0; \ (_work)->timer.function = __compat_delayed_work_timer; \ (_work)->timer.data = (unsigned long)_work # define compat_schedule_work(_work) \ schedule_task(_work) # define compat_schedule_delayed_work(_work, _delay) \ (_work)->timer.expires = jiffies + _delay; \ add_timer(&(_work)->timer) # define COMPAT_WORK_GET_DATA(_p, _type, _member) \ (_type *)(_p) # define COMPAT_DELAYED_WORK_GET_DATA(_p, _type, _member) \ (_type *)(_p) #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 20) \ && !defined(__VMKLNX__) /* } { */ typedef struct work_struct compat_work; typedef struct work_struct compat_delayed_work; typedef void * compat_work_arg; typedef void * compat_delayed_work_arg; # define COMPAT_INIT_WORK(_work, _func, _data) \ INIT_WORK(_work, _func, _data) # define COMPAT_INIT_DELAYED_WORK(_work, _func, _data) \ INIT_WORK(_work, _func, _data) # define compat_schedule_work(_work) \ schedule_work(_work) # define compat_schedule_delayed_work(_work, _delay) \ schedule_delayed_work(_work, _delay) # define COMPAT_WORK_GET_DATA(_p, _type, _member) \ (_type *)(_p) # define COMPAT_DELAYED_WORK_GET_DATA(_p, _type, _member) \ (_type *)(_p) #else /* } Linux >= 2.6.20 { */ typedef struct work_struct compat_work; typedef struct delayed_work compat_delayed_work; typedef struct work_struct * compat_work_arg; typedef struct work_struct * compat_delayed_work_arg; # define COMPAT_INIT_WORK(_work, _func, _data) \ INIT_WORK(_work, _func) # define COMPAT_INIT_DELAYED_WORK(_work, _func, _data) \ INIT_DELAYED_WORK(_work, _func) # define compat_schedule_work(_work) \ schedule_work(_work) # define compat_schedule_delayed_work(_work, _delay) \ schedule_delayed_work(_work, _delay) # define COMPAT_WORK_GET_DATA(_p, _type, _member) \ container_of(_p, _type, _member) # define COMPAT_DELAYED_WORK_GET_DATA(_p, _type, _member) \ container_of(_p, _type, _member.work) #endif /* } */ #endif /* __COMPAT_WORKQUEUE_H__ */ vsock-only/shared/mul64.h 0000444 0000000 0000000 00000007233 13432725350 014255 0 ustar root root /********************************************************* * Copyright (C) 2003-2014,2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * mul64.h * * Integer by fixed point multiplication, with rounding. * * These routines are implemented in assembly language for most * supported platforms. This file has plain C fallback versions. */ #ifndef _MUL64_H_ #define _MUL64_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" #include "vm_basic_asm.h" #if defined __cplusplus extern "C" { #endif #ifdef MUL64_NO_ASM /* *----------------------------------------------------------------------------- * * Mul64x3264 -- * * Unsigned integer by fixed point multiplication, with rounding: * result = floor(multiplicand * multiplier * 2**(-shift) + 0.5) * * Unsigned 64-bit integer multiplicand. * Unsigned 32-bit fixed point multiplier, represented as * (multiplier, shift), where shift < 64. * * Result: * Unsigned 64-bit integer product. * *----------------------------------------------------------------------------- */ static INLINE uint64 Mul64x3264(uint64 multiplicand, uint32 multiplier, uint32 shift) { uint64 lo, hi, lo2, hi2; unsigned carry; // ASSERT(shift >= 0 && shift < 64); lo = (multiplicand & 0xffffffff) * multiplier; hi = (multiplicand >> 32) * multiplier; lo2 = lo + (hi << 32); carry = lo2 < lo; hi2 = (hi >> 32) + carry; if (shift == 0) { return lo2; } else { return (lo2 >> shift) + (hi2 << (64 - shift)) + ((lo2 >> (shift - 1)) & 1); } } /* *----------------------------------------------------------------------------- * * Muls64x32s64 -- * * Signed integer by fixed point multiplication, with rounding: * result = floor(multiplicand * multiplier * 2**(-shift) + 0.5) * * Signed 64-bit integer multiplicand. * Unsigned 32-bit fixed point multiplier, represented as * (multiplier, shift), where shift < 64. * * Result: * Signed 64-bit integer product. * *----------------------------------------------------------------------------- */ static INLINE int64 Muls64x32s64(int64 multiplicand, uint32 multiplier, uint32 shift) { uint64 lo, hi, lo2, hi2; unsigned carry; // ASSERT(shift >= 0 && shift < 64); hi = ((uint64)multiplicand >> 32) * multiplier; if (multiplicand < 0) { hi -= (uint64)multiplier << 32; } lo = ((uint64)multiplicand & 0xffffffff) * multiplier; lo2 = lo + (hi << 32); carry = lo2 < lo; hi2 = (((int64)hi >> 32) + carry); if (shift == 0) { return lo2; } else { return (lo2 >> shift) + (hi2 << (64 - shift)) + ((lo2 >> (shift - 1)) & 1); } } #endif #if defined __cplusplus } // extern "C" #endif #endif // _MUL64_NOASM_H_ vsock-only/shared/compat_sched.h 0000444 0000000 0000000 00000024236 13432725347 015747 0 ustar root root /********************************************************* * Copyright (C) 2002 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_SCHED_H__ # define __COMPAT_SCHED_H__ #include <linux/sched.h> /* CLONE_KERNEL available in 2.5.35 and higher. */ #ifndef CLONE_KERNEL #define CLONE_KERNEL CLONE_FILES | CLONE_FS | CLONE_SIGHAND #endif /* TASK_COMM_LEN become available in 2.6.11. */ #ifndef TASK_COMM_LEN #define TASK_COMM_LEN 16 #endif /* The capable() API appeared in 2.1.92 --hpreg */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 1, 92) # define capable(_capability) suser() #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 2, 0) # define need_resched() need_resched #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 3) # define need_resched() (current->need_resched) #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 3) # define cond_resched() (need_resched() ? schedule() : (void) 0) #endif /* Oh well. We need yield... Happy us! */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 20) # ifdef __x86_64__ # define compat_yield() there_is_nothing_like_yield() # else # include <linux/unistd.h> # include <linux/kernel.h> /* * Used by _syscallX macros. Note that this is global variable, so * do not rely on its contents too much. As exit() is only function * we use, and we never check return value from exit(), we have * no problem... */ extern int errno; /* * compat_exit() provides an access to the exit() function. It must * be named compat_exit(), as exit() (with different signature) is * provided by x86-64, arm and other (but not by i386). */ # define __NR_compat_yield __NR_sched_yield static inline _syscall0(int, compat_yield); # endif #else # define compat_yield() yield() #endif /* * Since 2.5.34 there are two methods to enumerate tasks: * for_each_process(p) { ... } which enumerates only tasks and * do_each_thread(g,t) { ... } while_each_thread(g,t) which enumerates * also threads even if they share same pid. */ #ifndef for_each_process # define for_each_process(p) for_each_task(p) #endif #ifndef do_each_thread # define do_each_thread(g, t) for_each_task(g) { t = g; do # define while_each_thread(g, t) while (0) } #endif /* * Lock for signal mask is moving target... */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 40) && defined(CLONE_PID) /* 2.4.x without NPTL patches or early 2.5.x */ #define compat_sigmask_lock sigmask_lock #define compat_dequeue_signal_current(siginfo_ptr) \ dequeue_signal(¤t->blocked, (siginfo_ptr)) #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 60) && !defined(INIT_SIGHAND) /* RedHat's 2.4.x with first version of NPTL support, or 2.5.40 to 2.5.59 */ #define compat_sigmask_lock sig->siglock #define compat_dequeue_signal_current(siginfo_ptr) \ dequeue_signal(¤t->blocked, (siginfo_ptr)) #else /* RedHat's 2.4.x with second version of NPTL support, or 2.5.60+. */ #define compat_sigmask_lock sighand->siglock #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 0) #define compat_dequeue_signal_current(siginfo_ptr) \ dequeue_signal(¤t->blocked, (siginfo_ptr)) #else #define compat_dequeue_signal_current(siginfo_ptr) \ dequeue_signal(current, ¤t->blocked, (siginfo_ptr)) #endif #endif /* * recalc_sigpending() had task argument in the past */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 29) && defined(CLONE_PID) /* 2.4.x without NPTL patches or early 2.5.x */ #define compat_recalc_sigpending() recalc_sigpending(current) #else /* RedHat's 2.4.x with NPTL support, or 2.5.29+ */ #define compat_recalc_sigpending() recalc_sigpending() #endif /* * reparent_to_init() was introduced in 2.4.8. In 2.5.38 (or possibly * earlier, but later than 2.5.31) a call to it was added into * daemonize(), so compat_daemonize no longer needs to call it. * * In 2.4.x kernels reparent_to_init() forgets to do correct refcounting * on current->user. It is better to count one too many than one too few... */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 4, 8) && LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 38) #define compat_reparent_to_init() do { \ reparent_to_init(); \ atomic_inc(¤t->user->__count); \ } while (0) #else #define compat_reparent_to_init() do {} while (0) #endif /* * daemonize appeared in 2.2.18. Except 2.2.17-4-RH7.0, which has it too. * Fortunately 2.2.17-4-RH7.0 uses versioned symbols, so we can check * its existence with defined(). */ #if (LINUX_VERSION_CODE < KERNEL_VERSION(2, 2, 18)) && !defined(daemonize) static inline void daemonize(void) { struct fs_struct *fs; exit_mm(current); current->session = 1; current->pgrp = 1; exit_fs(current); fs = init_task.fs; current->fs = fs; atomic_inc(&fs->count); } #endif /* * flush_signals acquires sighand->siglock since 2.5.61... Verify RH's kernels! */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 61) #define compat_flush_signals(task) do { \ spin_lock_irq(&task->compat_sigmask_lock); \ flush_signals(task); \ spin_unlock_irq(&task->compat_sigmask_lock); \ } while (0) #else #define compat_flush_signals(task) flush_signals(task) #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 61) #define compat_allow_signal(signr) do { \ spin_lock_irq(¤t->compat_sigmask_lock); \ sigdelset(¤t->blocked, signr); \ compat_recalc_sigpending(); \ spin_unlock_irq(¤t->compat_sigmask_lock); \ } while (0) #else #define compat_allow_signal(signr) allow_signal(signr) #endif /* * daemonize can set process name since 2.5.61. Prior to 2.5.61, daemonize * didn't block signals on our behalf. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 61) #define compat_daemonize(x...) \ ({ \ /* Beware! No snprintf here, so verify arguments! */ \ sprintf(current->comm, x); \ \ /* Block all signals. */ \ spin_lock_irq(¤t->compat_sigmask_lock); \ sigfillset(¤t->blocked); \ compat_recalc_sigpending(); \ spin_unlock_irq(¤t->compat_sigmask_lock); \ compat_flush_signals(current); \ \ daemonize(); \ compat_reparent_to_init(); \ }) #else #define compat_daemonize(x...) daemonize(x) #endif /* * try to freeze a process. For kernels 2.6.11 or newer, we know how to choose * the interface. The problem is that the oldest interface, introduced in * 2.5.18, was backported to 2.4.x kernels. So if we're older than 2.6.11, * we'll decide what to do based on whether or not swsusp was configured * for the kernel. For kernels 2.6.20 and newer, we'll also need to include * freezer.h since the try_to_freeze definition was pulled out of sched.h. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 20) #include <linux/freezer.h> #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 13) || defined(VMW_TL10S64_WORKAROUND) #define compat_try_to_freeze() try_to_freeze() #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 11) #define compat_try_to_freeze() try_to_freeze(PF_FREEZE) #elif defined(CONFIG_SOFTWARE_SUSPEND) || defined(CONFIG_SOFTWARE_SUSPEND2) #include "compat_mm.h" #include <linux/errno.h> #include <linux/suspend.h> static inline int compat_try_to_freeze(void) { if (current->flags & PF_FREEZE) { refrigerator(PF_FREEZE); return 1; } else { return 0; } } #else static inline int compat_try_to_freeze(void) { return 0; } #endif /* * As of 2.6.23-rc1, kernel threads are no longer freezable by * default. Instead, kernel threads that need to be frozen must opt-in * by calling set_freezable() as soon as the thread is created. */ #if LINUX_VERSION_CODE > KERNEL_VERSION(2, 6, 22) #define compat_set_freezable() do { set_freezable(); } while (0) #else #define compat_set_freezable() do {} while (0) #endif /* * Around 2.6.27 kernel stopped sending signals to kernel * threads being frozen, instead threads have to check * freezing() or use wait_event_freezable(). Unfortunately * wait_event_freezable() completely hides the fact that * thread was frozen from calling code and sometimes we do * want to know that. */ #ifdef PF_FREEZER_NOSIG #define compat_wait_check_freezing() freezing(current) #else #define compat_wait_check_freezing() (0) #endif /* * Since 2.6.27-rc2 kill_proc() is gone... Replacement (GPL-only!) * API is available since 2.6.19. Use them from 2.6.27-rc1 up. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 27) typedef int compat_pid; #define compat_find_get_pid(pid) (pid) #define compat_put_pid(pid) do { } while (0) #define compat_kill_pid(pid, sig, flag) kill_proc(pid, sig, flag) #else typedef struct pid * compat_pid; #define compat_find_get_pid(pid) find_get_pid(pid) #define compat_put_pid(pid) put_pid(pid) #define compat_kill_pid(pid, sig, flag) kill_pid(pid, sig, flag) #endif #endif /* __COMPAT_SCHED_H__ */ vsock-only/shared/compat_page-flags.h 0000444 0000000 0000000 00000005037 13432725347 016665 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_PAGE_FLAGS_H__ # define __COMPAT_PAGE_FLAGS_H__ /* No page-flags.h prior to 2.5.12. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 5, 12) # include <linux/page-flags.h> #endif /* * The pgoff_t type was introduced in 2.5.20, but we'll look for it by * definition since it's more convenient. Note that we want to avoid a * situation where, in the future, a #define is changed to a typedef, * so if pgoff_t is not defined in some future kernel, we won't define it. */ #if !defined(pgoff_t) && LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 19) #define pgoff_t unsigned long #endif /* * set_page_writeback() was introduced in 2.6.6. Prior to that, callers were * using the SetPageWriteback() macro directly, so that's what we'll use. * Prior to 2.5.12, the writeback bit didn't exist, so we don't need to do * anything. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 12) #define compat_set_page_writeback(page) #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 6) #define compat_set_page_writeback(page) SetPageWriteback(page) #else #define compat_set_page_writeback(page) set_page_writeback(page) #endif /* * end_page_writeback() was introduced in 2.5.12. Prior to that, it looks like * there was no page writeback bit, and everything the function accomplished * was done by unlock_page(), so we'll define it out. * * Note that we could just #define end_page_writeback to nothing and avoid * needing the compat_ prefix, but this is more complete with respect to * compat_set_page_writeback. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 12) #define compat_end_page_writeback(page) #else #define compat_end_page_writeback(page) end_page_writeback(page) #endif #endif /* __COMPAT_PAGE_FLAGS_H__ */ vsock-only/shared/compat_fs.h 0000444 0000000 0000000 00000024277 13432725347 015276 0 ustar root root /********************************************************* * Copyright (C) 2006 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_FS_H__ # define __COMPAT_FS_H__ #include <linux/fs.h> /* * 2.6.5+ kernels define FS_BINARY_MOUNTDATA. Since it didn't exist and * wasn't used prior, it's safe to define it to zero. */ #ifndef FS_BINARY_MOUNTDATA #define FS_BINARY_MOUNTDATA 0 #endif /* * MAX_LFS_FILESIZE wasn't defined until 2.5.4. */ #ifndef MAX_LFS_FILESIZE # include <linux/pagemap.h> # if BITS_PER_LONG == 32 # define MAX_LFS_FILESIZE (((u64)PAGE_CACHE_SIZE << (BITS_PER_LONG - 1)) - 1) # elif BITS_PER_LONG == 64 # define MAX_LFS_FILESIZE 0x7fffffffffffffffUL # endif #endif /* * sendfile as a VFS op was born in 2.5.30. Unfortunately, it also changed * signatures, first in 2.5.47, then again in 2.5.70, then again in 2.6.8. * Luckily, the 2.6.8+ signature is the same as the 2.5.47 signature. And * as of 2.6.23-rc1 sendfile is gone, replaced by splice_read... * * Let's not support sendfile from 2.5.30 to 2.5.47, because the 2.5.30 * signature is much different and file_send_actor isn't externed. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 23) #define VMW_SENDFILE_NONE #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 8) #define VMW_SENDFILE_NEW #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 5, 70) #define VMW_SENDFILE_OLD #elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 5, 47) #define VMW_SENDFILE_NEW #else #define VMW_SENDFILE_NONE #endif /* * splice_read is there since 2.6.17, but let's avoid 2.6.17-rcX kernels... * After all nobody is using splice system call until 2.6.23 using it to * implement sendfile. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 18) #define VMW_SPLICE_READ 1 #endif /* * Filesystems wishing to use generic page cache read/write routines are * supposed to implement aio_read and aio_write (calling into * generic_file_aio_read() and generic_file_aio_write() if necessary). * * The VFS exports do_sync_read() and do_sync_write() as the "new" * generic_file_read() and generic_file_write(), but filesystems need not * actually implement read and write- the VFS will automatically call * do_sync_write() and do_sync_read() when applications invoke the standard * read() and write() system calls. * * In 2.6.19, generic_file_read() and generic_file_write() were removed, * necessitating this change. AIO dates as far back as 2.5.42, but the API has * changed over time, so for simplicity, we'll only enable it from 2.6.19 and * on. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 19) # define VMW_USE_AIO #endif /* * The alloc_inode and destroy_inode VFS ops didn't exist prior to 2.4.21. * Without these functions, file systems can't embed inodes. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 4, 21) # define VMW_EMBED_INODE #endif /* * iget() was removed from the VFS as of 2.6.25-rc1. The replacement for iget() * is iget_locked() which was added in 2.5.17. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 5, 17) # define VMW_USE_IGET_LOCKED #endif /* * parent_ino was born in 2.5.5. For older kernels, let's use 2.5.5 * implementation. It uses the dcache lock which is OK because per-dentry * locking appeared after 2.5.5. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 5, 5) #define compat_parent_ino(dentry) parent_ino(dentry) #else #define compat_parent_ino(dentry) \ ({ \ ino_t res; \ spin_lock(&dcache_lock); \ res = dentry->d_parent->d_inode->i_ino; \ spin_unlock(&dcache_lock); \ res; \ }) #endif /* * putname changed to __putname in 2.6.6. */ #define compat___getname() __getname() #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 6) #define compat___putname(name) putname(name) #else #define compat___putname(name) __putname(name) #endif /* * inc_nlink, drop_nlink, and clear_nlink were added in 2.6.19. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 19) #define compat_inc_nlink(inode) ((inode)->i_nlink++) #define compat_drop_nlink(inode) ((inode)->i_nlink--) #define compat_clear_nlink(inode) ((inode)->i_nlink = 0) #else #define compat_inc_nlink(inode) inc_nlink(inode) #define compat_drop_nlink(inode) drop_nlink(inode) #define compat_clear_nlink(inode) clear_nlink(inode) #endif /* * i_size_write and i_size_read were introduced in 2.6.0-test1 * (though we'll look for them as of 2.6.1). They employ slightly different * locking in order to guarantee atomicity, depending on the length of a long, * whether the kernel is SMP, or whether the kernel is preemptible. Prior to * i_size_write and i_size_read, there was no such locking, so that's the * behavior we'll emulate. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 1) #define compat_i_size_read(inode) ((inode)->i_size) #define compat_i_size_write(inode, size) ((inode)->i_size = size) #else #define compat_i_size_read(inode) i_size_read(inode) #define compat_i_size_write(inode, size) i_size_write(inode, size) #endif /* * filemap_fdatawrite was introduced in 2.5.12. Prior to that, modules used * filemap_fdatasync instead. In 2.4.18, both filemap_fdatawrite and * filemap_fdatawait began returning status codes. Prior to that, they were * void functions, so we'll just have them return 0. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 18) #define compat_filemap_fdatawrite(mapping) \ ({ \ int result = 0; \ filemap_fdatasync(mapping); \ result; \ }) #define compat_filemap_fdatawait(mapping) \ ({ \ int result = 0; \ filemap_fdatawait(mapping); \ result; \ }) #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 12) #define compat_filemap_fdatawrite(mapping) filemap_fdatasync(mapping) #define compat_filemap_fdatawait(mapping) filemap_fdatawait(mapping) #else #define compat_filemap_fdatawrite(mapping) filemap_fdatawrite(mapping) #define compat_filemap_fdatawait(mapping) filemap_fdatawait(mapping) #endif /* * filemap_write_and_wait was introduced in 2.6.6 and exported for module use * in 2.6.16. It's really just a simple wrapper around filemap_fdatawrite and * and filemap_fdatawait, which initiates a flush of all dirty pages, then * waits for the pages to flush. The implementation here is a simplified form * of the one found in 2.6.20-rc3. * * Unfortunately, it just isn't possible to implement this prior to 2.4.5, when * neither filemap_fdatawait nor filemap_fdatasync were exported for module * use. So we'll define it out and hope for the best. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 5) #define compat_filemap_write_and_wait(mapping) #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 16) #define compat_filemap_write_and_wait(mapping) \ ({ \ int result = 0; \ if (mapping->nrpages) { \ result = compat_filemap_fdatawrite(mapping); \ if (result != -EIO) { \ int result2 = compat_filemap_fdatawait(mapping); \ if (!result) { \ result = result2; \ } \ } \ } \ result; \ }) #else #define compat_filemap_write_and_wait(mapping) filemap_write_and_wait(mapping) #endif /* * invalidate_remote_inode was introduced in 2.6.0-test5. Prior to that, * filesystems wishing to invalidate pages belonging to an inode called * invalidate_inode_pages. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 0) #define compat_invalidate_remote_inode(inode) invalidate_inode_pages(inode) #else #define compat_invalidate_remote_inode(inode) invalidate_remote_inode(inode) #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 35) #define VMW_FSYNC_OLD #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 3, 0) typedef umode_t compat_umode_t; #else typedef int compat_umode_t; #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(3, 4, 0) #define d_make_root(inode) ({ \ struct dentry * ____res = d_alloc_root(inode); \ if (!____res) { \ iput(inode); \ } \ ____res; \ }) #endif #endif /* __COMPAT_FS_H__ */ vsock-only/shared/driverLog.c 0000444 0000000 0000000 00000011117 13432725347 015240 0 ustar root root /********************************************************* * Copyright (C) 2007-2014 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * driverLog.c -- * * Common logging functions for Linux kernel modules. */ #include "driver-config.h" #include "compat_kernel.h" #include "compat_sched.h" #include <asm/current.h> #include "driverLog.h" #define LINUXLOG_BUFFER_SIZE 1024 static const char *driverLogPrefix = ""; /* * vsnprintf was born in 2.4.10. Fall back on vsprintf if we're * an older kernel. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 10) # define vsnprintf(str, size, fmt, args) vsprintf(str, fmt, args) #endif /* *---------------------------------------------------------------------------- * * DriverLog_Init -- * * Initializes the Linux logging. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ void DriverLog_Init(const char *prefix) // IN { driverLogPrefix = prefix ? prefix : ""; } /* *---------------------------------------------------------------------- * * DriverLogPrint -- * * Log error message from a Linux module. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------- */ static void DriverLogPrint(const char *level, // IN: KERN_* constant const char *fmt, // IN: error format string va_list args) // IN: arguments for format string { static char staticBuf[LINUXLOG_BUFFER_SIZE]; char stackBuf[128]; va_list args2; const char *buf; /* * By default, use a small buffer on the stack (thread safe). If it is too * small, fall back to a larger static buffer (not thread safe). */ va_copy(args2, args); if (vsnprintf(stackBuf, sizeof stackBuf, fmt, args2) < sizeof stackBuf) { buf = stackBuf; } else { vsnprintf(staticBuf, sizeof staticBuf, fmt, args); buf = staticBuf; } va_end(args2); printk("%s%s[%d]: %s", level, driverLogPrefix, current->pid, buf); } /* *---------------------------------------------------------------------- * * Warning -- * * Warning messages from kernel module: logged into kernel log * as warnings. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------- */ void Warning(const char *fmt, ...) // IN: warning format string { va_list args; va_start(args, fmt); DriverLogPrint(KERN_WARNING, fmt, args); va_end(args); } /* *---------------------------------------------------------------------- * * Log -- * * Log messages from kernel module: logged into kernel log * as debug information. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------- */ void Log(const char *fmt, ...) // IN: log format string { va_list args; /* * Use the kernel log with at least a KERN_DEBUG level * so it doesn't garbage the screen at (re)boot time on RedHat 6.0. */ va_start(args, fmt); DriverLogPrint(KERN_DEBUG, fmt, args); va_end(args); } /* *---------------------------------------------------------------------- * * Panic -- * * ASSERTION failures and Panics from kernel module get here. * Message is logged to the kernel log and on console. * * Results: * None. * * Side effects: * Never returns * *---------------------------------------------------------------------- */ void Panic(const char *fmt, ...) // IN: panic format string { va_list args; va_start(args, fmt); DriverLogPrint(KERN_EMERG, fmt, args); va_end(args); #ifdef BUG BUG(); #else /* Should die with %cs unwritable, or at least with page fault. */ asm volatile("movb $0, %cs:(0)"); #endif while (1); } vsock-only/shared/compat_autoconf.h 0000444 0000000 0000000 00000002641 13432725347 016473 0 ustar root root /********************************************************* * Copyright (C) 2009 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_AUTOCONF_H__ # define __COMPAT_AUTOCONF_H__ #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMKDRIVERS #include "includeCheck.h" #ifndef LINUX_VERSION_CODE # error "Include compat_version.h before compat_autoconf.h" #endif /* autoconf.h moved from linux/autoconf.h to generated/autoconf.h in 2.6.33-rc1. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 33) # include <linux/autoconf.h> #else # include <generated/autoconf.h> #endif #endif /* __COMPAT_AUTOCONF_H__ */ vsock-only/shared/compat_skbuff.h 0000444 0000000 0000000 00000016131 13432725347 016134 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_SKBUFF_H__ # define __COMPAT_SKBUFF_H__ #include <linux/skbuff.h> /* * When transition from mac/nh/h to skb_* accessors was made, also SKB_WITH_OVERHEAD * was introduced. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 22) || \ (LINUX_VERSION_CODE == KERNEL_VERSION(2, 6, 21) && defined(SKB_WITH_OVERHEAD)) #define compat_skb_mac_header(skb) skb_mac_header(skb) #define compat_skb_network_header(skb) skb_network_header(skb) #define compat_skb_network_offset(skb) skb_network_offset(skb) #define compat_skb_transport_header(skb) skb_transport_header(skb) #define compat_skb_transport_offset(skb) skb_transport_offset(skb) #define compat_skb_network_header_len(skb) skb_network_header_len(skb) #define compat_skb_tail_pointer(skb) skb_tail_pointer(skb) #define compat_skb_end_pointer(skb) skb_end_pointer(skb) #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 22) # define compat_skb_ip_header(skb) ip_hdr(skb) # define compat_skb_ipv6_header(skb) ipv6_hdr(skb) # define compat_skb_tcp_header(skb) tcp_hdr(skb) #else # define compat_skb_ip_header(skb) ((struct iphdr *)skb_network_header(skb)) # define compat_skb_ipv6_header(skb) ((struct ipv6hdr *)skb_network_header(skb)) # define compat_skb_tcp_header(skb) ((struct tcphdr *)skb_transport_header(skb)) #endif #define compat_skb_reset_mac_header(skb) skb_reset_mac_header(skb) #define compat_skb_reset_network_header(skb) skb_reset_network_header(skb) #define compat_skb_reset_transport_header(skb) skb_reset_transport_header(skb) #define compat_skb_set_network_header(skb, off) skb_set_network_header(skb, off) #define compat_skb_set_transport_header(skb, off) skb_set_transport_header(skb, off) #else #define compat_skb_mac_header(skb) (skb)->mac.raw #define compat_skb_network_header(skb) (skb)->nh.raw #define compat_skb_network_offset(skb) ((skb)->nh.raw - (skb)->data) #define compat_skb_transport_header(skb) (skb)->h.raw #define compat_skb_transport_offset(skb) ((skb)->h.raw - (skb)->data) #define compat_skb_network_header_len(skb) ((skb)->h.raw - (skb)->nh.raw) #define compat_skb_tail_pointer(skb) (skb)->tail #define compat_skb_end_pointer(skb) (skb)->end #define compat_skb_ip_header(skb) (skb)->nh.iph #define compat_skb_ipv6_header(skb) (skb)->nh.ipv6h #define compat_skb_tcp_header(skb) (skb)->h.th #define compat_skb_reset_mac_header(skb) ((skb)->mac.raw = (skb)->data) #define compat_skb_reset_network_header(skb) ((skb)->nh.raw = (skb)->data) #define compat_skb_reset_transport_header(skb) ((skb)->h.raw = (skb)->data) #define compat_skb_set_network_header(skb, off) ((skb)->nh.raw = (skb)->data + (off)) #define compat_skb_set_transport_header(skb, off) ((skb)->h.raw = (skb)->data + (off)) #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 18) || defined(VMW_SKB_LINEARIZE_2618) # define compat_skb_linearize(skb) skb_linearize((skb)) #else # if LINUX_VERSION_CODE == KERNEL_VERSION(2, 6, 0) # define compat_skb_linearize(skb) __skb_linearize((skb), GFP_ATOMIC) # elif LINUX_VERSION_CODE >= KERNEL_VERSION(2, 4, 4) # define compat_skb_linearize(skb) skb_linearize((skb), GFP_ATOMIC) # else static inline int compat_skb_linearize(struct sk_buff *skb) { return 0; } # endif #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 20) #define compat_skb_csum_offset(skb) (skb)->csum_offset #else #define compat_skb_csum_offset(skb) (skb)->csum #endif /* * Note that compat_skb_csum_start() has semantic different from kernel's csum_start: * kernel's skb->csum_start is offset between start of checksummed area and start of * complete skb buffer, while our compat_skb_csum_start(skb) is offset from start * of packet itself. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 22) #define compat_skb_csum_start(skb) ((skb)->csum_start - skb_headroom(skb)) #else #define compat_skb_csum_start(skb) compat_skb_transport_offset(skb) #endif #if defined(NETIF_F_GSO) /* 2.6.18 and upwards */ #define compat_skb_mss(skb) (skb_shinfo(skb)->gso_size) #else #define compat_skb_mss(skb) (skb_shinfo(skb)->tso_size) #endif /* used by both received pkts and outgoing ones */ #define VM_CHECKSUM_UNNECESSARY CHECKSUM_UNNECESSARY /* csum status of received pkts */ #if defined(CHECKSUM_COMPLETE) # define VM_RX_CHECKSUM_PARTIAL CHECKSUM_COMPLETE #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 19) && defined(CHECKSUM_HW) # define VM_RX_CHECKSUM_PARTIAL CHECKSUM_HW #else # define VM_RX_CHECKSUM_PARTIAL CHECKSUM_PARTIAL #endif /* csum status of outgoing pkts */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 19) && defined(CHECKSUM_HW) # define VM_TX_CHECKSUM_PARTIAL CHECKSUM_HW #else # define VM_TX_CHECKSUM_PARTIAL CHECKSUM_PARTIAL #endif #if (LINUX_VERSION_CODE < KERNEL_VERSION(2,1,0)) # define compat_kfree_skb(skb, type) kfree_skb(skb, type) # define compat_dev_kfree_skb(skb, type) dev_kfree_skb(skb, type) # define compat_dev_kfree_skb_any(skb, type) dev_kfree_skb(skb, type) # define compat_dev_kfree_skb_irq(skb, type) dev_kfree_skb(skb, type) #else # define compat_kfree_skb(skb, type) kfree_skb(skb) # define compat_dev_kfree_skb(skb, type) dev_kfree_skb(skb) # if (LINUX_VERSION_CODE < KERNEL_VERSION(2,3,43)) # define compat_dev_kfree_skb_any(skb, type) dev_kfree_skb(skb) # define compat_dev_kfree_skb_irq(skb, type) dev_kfree_skb(skb) # else # define compat_dev_kfree_skb_any(skb, type) dev_kfree_skb_any(skb) # define compat_dev_kfree_skb_irq(skb, type) dev_kfree_skb_irq(skb) # endif #endif #ifndef NET_IP_ALIGN # define COMPAT_NET_IP_ALIGN 2 #else # define COMPAT_NET_IP_ALIGN NET_IP_ALIGN #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 4, 4) # define compat_skb_headlen(skb) skb_headlen(skb) # define compat_pskb_may_pull(skb, len) pskb_may_pull(skb, len) # define compat_skb_is_nonlinear(skb) skb_is_nonlinear(skb) #else # define compat_skb_headlen(skb) (skb)->len # define compat_pskb_may_pull(skb, len) 1 # define compat_skb_is_nonlinear(skb) 0 #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 12) # define compat_skb_header_cloned(skb) skb_header_cloned(skb) #else # define compat_skb_header_cloned(skb) 0 #endif #endif /* __COMPAT_SKBUFF_H__ */ vsock-only/shared/compat_semaphore.h 0000444 0000000 0000000 00000003142 13432725347 016635 0 ustar root root /********************************************************* * Copyright (C) 2002 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_SEMAPHORE_H__ # define __COMPAT_SEMAPHORE_H__ /* <= 2.6.25 have asm only, 2.6.26 has both, and 2.6.27-rc2+ has linux only. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 27) # include <asm/semaphore.h> #else # include <linux/semaphore.h> #endif /* * The init_MUTEX_LOCKED() API appeared in 2.2.18, and is also in * 2.2.17-21mdk --hpreg */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 2, 18) #ifndef init_MUTEX_LOCKED #define init_MUTEX_LOCKED(_sem) *(_sem) = MUTEX_LOCKED #endif #ifndef DECLARE_MUTEX #define DECLARE_MUTEX(name) struct semaphore name = MUTEX #endif #ifndef DECLARE_MUTEX_LOCKED #define DECLARE_MUTEX_LOCKED(name) struct semaphore name = MUTEX_LOCKED #endif #endif #endif /* __COMPAT_SEMAPHORE_H__ */ vsock-only/shared/vmware_pack_end.h 0000444 0000000 0000000 00000002475 13432725350 016436 0 ustar root root /********************************************************* * Copyright (C) 2002-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmware_pack_end.h -- * * End of structure packing. See vmware_pack_init.h for details. * * Note that we do not use the following construct in this include file, * because we want to emit the code every time the file is included --hpreg * * #ifndef foo * # define foo * ... * #endif * */ #include "vmware_pack_init.h" #ifdef _MSC_VER # pragma pack(pop) #elif __GNUC__ __attribute__((__packed__)) #else # error Compiler packing... #endif vsock-only/shared/compat_uaccess.h 0000444 0000000 0000000 00000006062 13432725347 016304 0 ustar root root /********************************************************* * Copyright (C) 2002 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_UACCESS_H__ # define __COMPAT_UACCESS_H__ /* User space access functions moved in 2.1.7 to asm/uaccess.h --hpreg */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 1, 7) # include <asm/uaccess.h> #else # include <asm/segment.h> #endif /* get_user() API modified in 2.1.4 to take 2 arguments --hpreg */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 1, 4) # define compat_get_user get_user #else /* * We assign 0 to the variable in case of failure to prevent "`_var' might be * used uninitialized in this function" compiler warnings. I think it is OK, * because the hardware-based version in newer kernels probably has the same * semantics and does not guarantee that the value of _var will not be * modified, should the access fail --hpreg */ # define compat_get_user(_var, _uvAddr) ({ \ int _status; \ \ _status = verify_area(VERIFY_READ, _uvAddr, sizeof(*(_uvAddr))); \ if (_status == 0) { \ (_var) = get_user(_uvAddr); \ } else { \ (_var) = 0; \ } \ _status; \ }) #endif /* * The copy_from_user() API appeared in 2.1.4 * * The emulation is not perfect here, but it is conservative: on failure, we * always return the total size, instead of the potentially smaller faulty * size --hpreg * * Since 2.5.55 copy_from_user() is no longer macro. */ #if !defined(copy_from_user) && LINUX_VERSION_CODE < KERNEL_VERSION(2, 2, 0) # define copy_from_user(_to, _from, _size) ( \ verify_area(VERIFY_READ, _from, _size) \ ? (_size) \ : (memcpy_fromfs(_to, _from, _size), 0) \ ) # define copy_to_user(_to, _from, _size) ( \ verify_area(VERIFY_WRITE, _to, _size) \ ? (_size) \ : (memcpy_tofs(_to, _from, _size), 0) \ ) #endif #endif /* __COMPAT_UACCESS_H__ */ vsock-only/shared/compat_spinlock.h 0000444 0000000 0000000 00000003377 13432725347 016506 0 ustar root root /********************************************************* * Copyright (C) 2005 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_SPINLOCK_H__ # define __COMPAT_SPINLOCK_H__ #include <linux/spinlock.h> /* * Preempt support was added during 2.5.x development cycle, and later * it was backported to 2.4.x. In 2.4.x backport these definitions * live in linux/spinlock.h, that's why we put them here (in 2.6.x they * are defined in linux/preempt.h which is included by linux/spinlock.h). */ #ifdef CONFIG_PREEMPT #define compat_preempt_disable() preempt_disable() #define compat_preempt_enable() preempt_enable() #else #define compat_preempt_disable() do { } while (0) #define compat_preempt_enable() do { } while (0) #endif /* Some older kernels - 2.6.10 and earlier - lack DEFINE_SPINLOCK */ #ifndef DEFINE_SPINLOCK #define DEFINE_SPINLOCK(x) spinlock_t x = SPIN_LOCK_UNLOCKED #endif /* Same goes for DEFINE_RWLOCK */ #ifndef DEFINE_RWLOCK #define DEFINE_RWLOCK(x) rwlock_t x = RW_LOCK_UNLOCKED #endif #endif /* __COMPAT_SPINLOCK_H__ */ vsock-only/shared/driverLog.h 0000444 0000000 0000000 00000002237 13432725347 015250 0 ustar root root /********************************************************* * Copyright (C) 2007-2014 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * driverLog.h -- * * Logging functions for Linux kernel modules. */ #ifndef __DRIVERLOG_H__ #define __DRIVERLOG_H__ /* * The definitions of Warning(), Log(), and Panic() come from vm_assert.h for * consistency. */ #include "vm_assert.h" void DriverLog_Init(const char *prefix); #endif /* __DRIVERLOG_H__ */ vsock-only/shared/x86cpuid.h 0000644 0000000 0000000 00000327332 13432725350 014767 0 ustar root root /********************************************************* * Copyright (C) 1998-2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef _X86CPUID_H_ #define _X86CPUID_H_ /* http://www.sandpile.org/ia32/cpuid.htm */ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_VMX #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_VMMON #include "includeCheck.h" #include "vm_basic_types.h" #include "community_source.h" #include "x86vendor.h" #include "vm_assert.h" #if defined __cplusplus extern "C" { #endif /* * The linux kernel's ptrace.h stupidly defines the bare * EAX/EBX/ECX/EDX, which wrecks havoc with our preprocessor tricks. */ #undef EAX #undef EBX #undef ECX #undef EDX typedef struct CPUIDRegs { uint32 eax, ebx, ecx, edx; } CPUIDRegs; typedef union CPUIDRegsUnion { uint32 array[4]; CPUIDRegs regs; } CPUIDRegsUnion; /* * Results of calling cpuid(eax, ecx) on all host logical CPU. */ #ifdef _MSC_VER // TODO: Move this under the push #pragma warning (disable :4200) // non-std extension: zero-sized array in struct #pragma warning (push) #pragma warning (disable :4100) // unreferenced parameters #endif typedef #include "vmware_pack_begin.h" struct CPUIDReply { /* * Unique host logical CPU identifier. It does not change across queries, so * we use it to correlate the replies of multiple queries. */ uint64 tag; // OUT CPUIDRegs regs; // OUT } #include "vmware_pack_end.h" CPUIDReply; typedef #include "vmware_pack_begin.h" struct CPUIDQuery { uint32 eax; // IN uint32 ecx; // IN uint32 numLogicalCPUs; // IN/OUT CPUIDReply logicalCPUs[0]; // OUT } #include "vmware_pack_end.h" CPUIDQuery; /* * CPUID levels the monitor caches. * * The first parameter defines whether the level has its default masks * generated from the values in this file. Any level which is marked as FALSE * here *must* have all monitor support types set to NA. A static assert in * lib/cpuidcompat/cpuidcompat.c will check this. * * The second parameter is the "short name" of the level. It's mainly used for * token concatenation in various macros. * * The third parameter is the actual numeric value of that level (the EAX input * value). * * The fourth parameter is a "subleaf count", where 0 means that ecx is * ignored, otherwise is the count of sub-leaves. * * The fifth parameter is the first hardware version that is *aware* of the * CPUID level (0 = existed since dawn of time), even though we may not expose * this level or parts of it to guest. */ #define CPUID_CACHED_LEVELS \ CPUIDLEVEL(TRUE, 0, 0, 0, 0) \ CPUIDLEVEL(TRUE, 1, 1, 0, 0) \ CPUIDLEVEL(FALSE, 2, 2, 0, 0) \ CPUIDLEVEL(FALSE, 4, 4, 7, 0) \ CPUIDLEVEL(FALSE, 5, 5, 0, 0) \ CPUIDLEVEL(TRUE, 6, 6, 0, 0) \ CPUIDLEVEL(TRUE, 7, 7, 1, 0) \ CPUIDLEVEL(FALSE, A, 0xA, 0, 0) \ CPUIDLEVEL(FALSE, B, 0xB, 2, 0) \ CPUIDLEVEL(TRUE, D, 0xD, 10, 0) \ CPUIDLEVEL(TRUE, F, 0xF, 2, 13) \ CPUIDLEVEL(TRUE, 10, 0x10, 2, 13) \ CPUIDLEVEL(TRUE, 12, 0x12, 4, 13) \ CPUIDLEVEL(TRUE, 14, 0x14, 2, 13) \ CPUIDLEVEL(TRUE, 15, 0x15, 0, 13) \ CPUIDLEVEL(TRUE, 16, 0x16, 0, 13) \ CPUIDLEVEL(TRUE, 17, 0x17, 4, 14) \ CPUIDLEVEL(FALSE, 400, 0x40000000, 0, 0) \ CPUIDLEVEL(FALSE, 401, 0x40000001, 0, 0) \ CPUIDLEVEL(FALSE, 402, 0x40000002, 0, 0) \ CPUIDLEVEL(FALSE, 403, 0x40000003, 0, 0) \ CPUIDLEVEL(FALSE, 404, 0x40000004, 0, 0) \ CPUIDLEVEL(FALSE, 405, 0x40000005, 0, 0) \ CPUIDLEVEL(FALSE, 406, 0x40000006, 0, 0) \ CPUIDLEVEL(FALSE, 410, 0x40000010, 0, 0) \ CPUIDLEVEL(FALSE, 80, 0x80000000, 0, 0) \ CPUIDLEVEL(TRUE, 81, 0x80000001, 0, 0) \ CPUIDLEVEL(FALSE, 82, 0x80000002, 0, 0) \ CPUIDLEVEL(FALSE, 83, 0x80000003, 0, 0) \ CPUIDLEVEL(FALSE, 84, 0x80000004, 0, 0) \ CPUIDLEVEL(FALSE, 85, 0x80000005, 0, 0) \ CPUIDLEVEL(FALSE, 86, 0x80000006, 0, 0) \ CPUIDLEVEL(FALSE, 87, 0x80000007, 0, 0) \ CPUIDLEVEL(TRUE, 88, 0x80000008, 0, 0) \ CPUIDLEVEL(TRUE, 8A, 0x8000000A, 0, 0) \ CPUIDLEVEL(FALSE, 819, 0x80000019, 0, 0) \ CPUIDLEVEL(FALSE, 81A, 0x8000001A, 0, 0) \ CPUIDLEVEL(FALSE, 81B, 0x8000001B, 0, 0) \ CPUIDLEVEL(FALSE, 81C, 0x8000001C, 0, 0) \ CPUIDLEVEL(FALSE, 81D, 0x8000001D, 5, 0) \ CPUIDLEVEL(FALSE, 81E, 0x8000001E, 0, 0) \ CPUIDLEVEL(TRUE, 81F, 0x8000001F, 0, 14) #define CPUID_ALL_LEVELS CPUID_CACHED_LEVELS /* Define cached CPUID levels in the form: CPUID_LEVEL_<ShortName> */ typedef enum { #define CPUIDLEVEL(t, s, v, c, h) CPUID_LEVEL_##s, CPUID_CACHED_LEVELS #undef CPUIDLEVEL CPUID_NUM_CACHED_LEVELS } CpuidCachedLevel; /* Enum to translate between shorthand name and actual CPUID level value. */ enum { #define CPUIDLEVEL(t, s, v, c, h) CPUID_LEVEL_VAL_##s = v, CPUID_ALL_LEVELS #undef CPUIDLEVEL }; /* Named feature leaves */ #define CPUID_FEATURE_INFORMATION 0x01 #define CPUID_PROCESSOR_TOPOLOGY 4 #define CPUID_MWAIT_FEATURES 5 #define CPUID_XSAVE_FEATURES 0xd #define CPUID_SGX_FEATURES 0x12 #define CPUID_PT_FEATURES 0x14 #define CPUID_HYPERVISOR_LEVEL_0 0x40000000 #define CPUID_SVM_FEATURES 0x8000000a /* * CPUID result registers */ #define CPUID_REGS \ CPUIDREG(EAX, eax) \ CPUIDREG(EBX, ebx) \ CPUIDREG(ECX, ecx) \ CPUIDREG(EDX, edx) typedef enum { #define CPUIDREG(uc, lc) CPUID_REG_##uc, CPUID_REGS #undef CPUIDREG CPUID_NUM_REGS } CpuidReg; #define CPUID_INTEL_VENDOR_STRING "GenuntelineI" #define CPUID_AMD_VENDOR_STRING "AuthcAMDenti" #define CPUID_CYRIX_VENDOR_STRING "CyriteadxIns" #define CPUID_VIA_VENDOR_STRING "CentaulsaurH" #define CPUID_HYGON_VENDOR_STRING "HygouinenGen" #define CPUID_HYPERV_HYPERVISOR_VENDOR_STRING "Microsoft Hv" #define CPUID_KVM_HYPERVISOR_VENDOR_STRING "KVMKVMKVM\0\0\0" #define CPUID_VMWARE_HYPERVISOR_VENDOR_STRING "VMwareVMware" #define CPUID_XEN_HYPERVISOR_VENDOR_STRING "XenVMMXenVMM" #define CPUID_INTEL_VENDOR_STRING_FIXED "GenuineIntel" #define CPUID_AMD_VENDOR_STRING_FIXED "AuthenticAMD" #define CPUID_CYRIX_VENDOR_STRING_FIXED "CyrixInstead" #define CPUID_VIA_VENDOR_STRING_FIXED "CentaurHauls" #define CPUID_HYGON_VENDOR_STRING_FIXED "HygonGenuine" /* * FIELD can be defined to process the CPUID information provided in the * following CPUID_FIELD_DATA macro. * * The first parameter is the CPUID level of the feature (must be defined in * CPUID_ALL_LEVELS, above). * * The second parameter is the CPUID sub-level (subleaf) of the feature. Please * make sure here the number is consistent with the "subleaf count" in * CPUIDLEVEL macro. I.e., if a feature is being added to a _new_ subleaf, * update the subleaf count above as well. * * The third parameter is the result register. * * The fourth and fifth parameters are the bit position of the field and the * width, respectively. * * The sixth is the name of the field. * * The seventh parameter specifies the monitor support characteristics for * this field. The value must be a valid CpuidFieldSupported value (omitting * CPUID_FIELD_SUPPORT_ for convenience). The meaning of those values are * described below. * * The eighth parameter specifies the first virtual hardware version that * implements the field (if 7th field is YES or ANY), or 0 (if 7th field is * NO or NA). The field's hardware version must match the version in * defaultMasks (cpuidcompat.c) if defined there, and must be less than or * equal to the version of the cpuid leaf it's in. * * The ninth parameter describes whether the feature is capable of being used * by usermode code (TRUE), or just CPL0 kernel code (FALSE). * * FLAG is defined identically to FIELD, but its accessors are more appropriate * for 1-bit flags, and compile-time asserts enforce that the size is 1 bit * wide. */ /* * CpuidFieldSupported is made up of the following values: * * NO: A feature/field that IS NOT SUPPORTED by the monitor. Even * if the host supports this feature, we will never expose it to * the guest. * * YES: A feature/field that IS SUPPORTED by the monitor. If the * host supports this feature, we will expose it to the guest. If * not, then we will not set the feature. * * ANY: A feature/field that IS ALWAYS SUPPORTED by the monitor. * Even if the host does not support the feature, the monitor can * expose the feature to the guest. As with "YES", the guest cpuid * value defaults to the host/evc cpuid value. But usually the * guest cpuid value is recomputed at power on, ignoring the default * value. * * * NA: Only legal for levels not masked/tested by default (see * above for this definition). Such fields must always be marked * as NA. * * These distinctions, when combined with the feature's CPL3 * properties can be translated into a common CPUID mask string as * follows: * * NO + CPL3 --> "R" (Reserved). We don't support the feature, * but we can't properly hide this from applications when using * direct execution or HV with apps that do try/catch/fail, so we * must still perform compatibility checks. * * NO + !CPL3 --> "0" (Masked). We can hide this from the guest. * * YES --> "H" (Host). We support the feature, so show it to the * guest if the host has the feature. * * ANY/NA --> "X" (Ignore). By default, don't perform checks for * this feature bit. Per-GOS masks may choose to set this bit in * the guest. (e.g. the APIC feature bit is always set to 1.) * * See lib/cpuidcompat/cpuidcompat.c for any possible overrides to * these defaults. */ typedef enum { CPUID_FIELD_SUPPORTED_NO, CPUID_FIELD_SUPPORTED_YES, CPUID_FIELD_SUPPORTED_ANY, CPUID_FIELD_SUPPORTED_NA, CPUID_NUM_FIELD_SUPPORTEDS } CpuidFieldSupported; /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_0 \ FIELD( 0, 0, EAX, 0, 32, NUMLEVELS, ANY, 4, FALSE) \ FIELD( 0, 0, EBX, 0, 32, VENDOR1, YES, 4, TRUE) \ FIELD( 0, 0, ECX, 0, 32, VENDOR3, YES, 4, TRUE) \ FIELD( 0, 0, EDX, 0, 32, VENDOR2, YES, 4, TRUE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_1 \ FIELD( 1, 0, EAX, 0, 4, STEPPING, ANY, 4, FALSE) \ FIELD( 1, 0, EAX, 4, 4, MODEL, ANY, 4, FALSE) \ FIELD( 1, 0, EAX, 8, 4, FAMILY, YES, 4, FALSE) \ FIELD( 1, 0, EAX, 12, 2, TYPE, ANY, 4, FALSE) \ FIELD( 1, 0, EAX, 16, 4, EXTENDED_MODEL, ANY, 4, FALSE) \ FIELD( 1, 0, EAX, 20, 8, EXTENDED_FAMILY, YES, 4, FALSE) \ FIELD( 1, 0, EBX, 0, 8, BRAND_ID, ANY, 4, FALSE) \ FIELD( 1, 0, EBX, 8, 8, CLFL_SIZE, ANY, 4, FALSE) \ FIELD( 1, 0, EBX, 16, 8, LCPU_COUNT, ANY, 4, FALSE) \ FIELD( 1, 0, EBX, 24, 8, APICID, ANY, 4, FALSE) \ FLAG( 1, 0, ECX, 0, 1, SSE3, YES, 4, TRUE) \ FLAG( 1, 0, ECX, 1, 1, PCLMULQDQ, YES, 7, TRUE) \ FLAG( 1, 0, ECX, 2, 1, DTES64, NO, 0, FALSE) \ FLAG( 1, 0, ECX, 3, 1, MWAIT, YES, 4, FALSE) \ FLAG( 1, 0, ECX, 4, 1, DSCPL, NO, 0, FALSE) \ FLAG( 1, 0, ECX, 5, 1, VMX, YES, 4, FALSE) \ FLAG( 1, 0, ECX, 6, 1, SMX, YES, 15, FALSE) \ FLAG( 1, 0, ECX, 7, 1, EIST, NO, 0, FALSE) \ FLAG( 1, 0, ECX, 8, 1, TM2, NO, 0, FALSE) \ FLAG( 1, 0, ECX, 9, 1, SSSE3, YES, 4, TRUE) \ FLAG( 1, 0, ECX, 10, 1, CNXTID, NO, 0, FALSE) \ FLAG( 1, 0, ECX, 11, 1, SDBG, NO, 0, FALSE) \ FLAG( 1, 0, ECX, 12, 1, FMA, YES, 8, TRUE) \ FLAG( 1, 0, ECX, 13, 1, CMPXCHG16B, YES, 4, TRUE) \ FLAG( 1, 0, ECX, 14, 1, xTPR, NO, 0, FALSE) \ FLAG( 1, 0, ECX, 15, 1, PDCM, NO, 0, FALSE) \ FLAG( 1, 0, ECX, 17, 1, PCID, YES, 8, FALSE) \ FLAG( 1, 0, ECX, 18, 1, DCA, NO, 0, FALSE) \ FLAG( 1, 0, ECX, 19, 1, SSE41, YES, 4, TRUE) \ FLAG( 1, 0, ECX, 20, 1, SSE42, YES, 4, TRUE) \ FLAG( 1, 0, ECX, 21, 1, x2APIC, ANY, 9, FALSE) \ FLAG( 1, 0, ECX, 22, 1, MOVBE, YES, 7, TRUE) \ FLAG( 1, 0, ECX, 23, 1, POPCNT, YES, 4, TRUE) \ FLAG( 1, 0, ECX, 24, 1, TSC_DEADLINE, ANY, 11, FALSE) \ FLAG( 1, 0, ECX, 25, 1, AES, YES, 7, TRUE) \ FLAG( 1, 0, ECX, 26, 1, XSAVE, YES, 8, FALSE) \ FLAG( 1, 0, ECX, 27, 1, OSXSAVE, ANY, 8, FALSE) \ FLAG( 1, 0, ECX, 28, 1, AVX, YES, 8, FALSE) \ FLAG( 1, 0, ECX, 29, 1, F16C, YES, 9, TRUE) \ FLAG( 1, 0, ECX, 30, 1, RDRAND, YES, 9, TRUE) \ FLAG( 1, 0, ECX, 31, 1, HYPERVISOR, ANY, 4, TRUE) \ FLAG( 1, 0, EDX, 0, 1, FPU, YES, 4, TRUE) \ FLAG( 1, 0, EDX, 1, 1, VME, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 2, 1, DE, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 3, 1, PSE, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 4, 1, TSC, YES, 4, TRUE) \ FLAG( 1, 0, EDX, 5, 1, MSR, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 6, 1, PAE, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 7, 1, MCE, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 8, 1, CX8, YES, 4, TRUE) \ FLAG( 1, 0, EDX, 9, 1, APIC, ANY, 4, FALSE) \ FLAG( 1, 0, EDX, 11, 1, SEP, YES, 4, TRUE) \ FLAG( 1, 0, EDX, 12, 1, MTRR, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 13, 1, PGE, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 14, 1, MCA, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 15, 1, CMOV, YES, 4, TRUE) \ FLAG( 1, 0, EDX, 16, 1, PAT, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 17, 1, PSE36, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 18, 1, PSN, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 19, 1, CLFSH, YES, 4, TRUE) \ FLAG( 1, 0, EDX, 21, 1, DS, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 22, 1, ACPI, ANY, 4, FALSE) \ FLAG( 1, 0, EDX, 23, 1, MMX, YES, 4, TRUE) \ FLAG( 1, 0, EDX, 24, 1, FXSR, YES, 4, TRUE) \ FLAG( 1, 0, EDX, 25, 1, SSE, YES, 4, TRUE) \ FLAG( 1, 0, EDX, 26, 1, SSE2, YES, 4, TRUE) \ FLAG( 1, 0, EDX, 27, 1, SS, YES, 4, FALSE) \ FLAG( 1, 0, EDX, 28, 1, HTT, ANY, 7, FALSE) \ FLAG( 1, 0, EDX, 29, 1, TM, NO, 0, FALSE) \ FLAG( 1, 0, EDX, 30, 1, IA64, NO, 0, FALSE) \ FLAG( 1, 0, EDX, 31, 1, PBE, NO, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_2 \ FIELD( 2, 0, EAX, 0, 8, LEAF2_COUNT, NA, 0, FALSE) \ FIELD( 2, 0, EAX, 8, 8, LEAF2_CACHE1, NA, 0, FALSE) \ FIELD( 2, 0, EAX, 16, 8, LEAF2_CACHE2, NA, 0, FALSE) \ FIELD( 2, 0, EAX, 24, 8, LEAF2_CACHE3, NA, 0, FALSE) \ FIELD( 2, 0, EBX, 0, 8, LEAF2_CACHE4, NA, 0, FALSE) \ FIELD( 2, 0, EBX, 8, 8, LEAF2_CACHE5, NA, 0, FALSE) \ FIELD( 2, 0, EBX, 16, 8, LEAF2_CACHE6, NA, 0, FALSE) \ FIELD( 2, 0, EBX, 24, 8, LEAF2_CACHE7, NA, 0, FALSE) \ FIELD( 2, 0, ECX, 0, 8, LEAF2_CACHE8, NA, 0, FALSE) \ FIELD( 2, 0, ECX, 8, 8, LEAF2_CACHE9, NA, 0, FALSE) \ FIELD( 2, 0, ECX, 16, 8, LEAF2_CACHE10, NA, 0, FALSE) \ FIELD( 2, 0, ECX, 24, 8, LEAF2_CACHE11, NA, 0, FALSE) \ FIELD( 2, 0, EDX, 0, 8, LEAF2_CACHE12, NA, 0, FALSE) \ FIELD( 2, 0, EDX, 8, 8, LEAF2_CACHE13, NA, 0, FALSE) \ FIELD( 2, 0, EDX, 16, 8, LEAF2_CACHE14, NA, 0, FALSE) \ FIELD( 2, 0, EDX, 24, 8, LEAF2_CACHE15, NA, 0, FALSE) \ /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_4 \ FIELD( 4, 0, EAX, 0, 5, LEAF4_CACHE_TYPE, NA, 0, FALSE) \ FIELD( 4, 0, EAX, 5, 3, LEAF4_CACHE_LEVEL, NA, 0, FALSE) \ FLAG( 4, 0, EAX, 8, 1, LEAF4_CACHE_SELF_INIT, NA, 0, FALSE) \ FLAG( 4, 0, EAX, 9, 1, LEAF4_CACHE_FULLY_ASSOC, NA, 0, FALSE) \ FIELD( 4, 0, EAX, 14, 12, LEAF4_CACHE_NUMHT_SHARING, NA, 0, FALSE) \ FIELD( 4, 0, EAX, 26, 6, LEAF4_CORE_COUNT, NA, 0, FALSE) \ FIELD( 4, 0, EBX, 0, 12, LEAF4_CACHE_LINE, NA, 0, FALSE) \ FIELD( 4, 0, EBX, 12, 10, LEAF4_CACHE_PART, NA, 0, FALSE) \ FIELD( 4, 0, EBX, 22, 10, LEAF4_CACHE_WAYS, NA, 0, FALSE) \ FIELD( 4, 0, ECX, 0, 32, LEAF4_CACHE_SETS, NA, 0, FALSE) \ FLAG( 4, 0, EDX, 0, 1, LEAF4_CACHE_WBINVD_NOT_GUARANTEED, NA, 0, FALSE) \ FLAG( 4, 0, EDX, 1, 1, LEAF4_CACHE_IS_INCLUSIVE, NA, 0, FALSE) \ FLAG( 4, 0, EDX, 2, 1, LEAF4_CACHE_COMPLEX_INDEXING, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_5 \ FIELD( 5, 0, EAX, 0, 16, MWAIT_MIN_SIZE, NA, 0, FALSE) \ FIELD( 5, 0, EBX, 0, 16, MWAIT_MAX_SIZE, NA, 0, FALSE) \ FLAG( 5, 0, ECX, 0, 1, MWAIT_EXTENSIONS, NA, 0, FALSE) \ FLAG( 5, 0, ECX, 1, 1, MWAIT_INTR_BREAK, NA, 0, FALSE) \ FIELD( 5, 0, EDX, 0, 4, MWAIT_C0_SUBSTATE, NA, 0, FALSE) \ FIELD( 5, 0, EDX, 4, 4, MWAIT_C1_SUBSTATE, NA, 0, FALSE) \ FIELD( 5, 0, EDX, 8, 4, MWAIT_C2_SUBSTATE, NA, 0, FALSE) \ FIELD( 5, 0, EDX, 12, 4, MWAIT_C3_SUBSTATE, NA, 0, FALSE) \ FIELD( 5, 0, EDX, 16, 4, MWAIT_C4_SUBSTATE, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_6 \ FLAG( 6, 0, EAX, 0, 1, THERMAL_SENSOR, NO, 0, FALSE) \ FLAG( 6, 0, EAX, 1, 1, TURBO_MODE, NO, 0, FALSE) \ FLAG( 6, 0, EAX, 2, 1, APIC_INVARIANT, ANY, 13, FALSE) \ FLAG( 6, 0, EAX, 4, 1, PLN, NO, 0, FALSE) \ FLAG( 6, 0, EAX, 5, 1, ECMD, NO, 0, FALSE) \ FLAG( 6, 0, EAX, 6, 1, PTM, NO, 0, FALSE) \ FLAG( 6, 0, EAX, 7, 1, HWP, NO, 0, FALSE) \ FLAG( 6, 0, EAX, 8, 1, HWP_NOTIFICATION, NO, 0, FALSE) \ FLAG( 6, 0, EAX, 9, 1, HWP_ACTIVITY_WINDOW, NO, 0, FALSE) \ FLAG( 6, 0, EAX, 10, 1, HWP_ENERGY_PERFORMANCE_PREFERENCE, NO, 0, FALSE) \ FLAG( 6, 0, EAX, 11, 1, HWP_PACKAGE_LEVEL_REQUEST, NO, 0, FALSE) \ FLAG( 6, 0, EAX, 13, 1, HDC, NO, 0, FALSE) \ FIELD( 6, 0, EBX, 0, 4, NUM_INTR_THRESHOLDS, NO, 0, FALSE) \ FLAG( 6, 0, ECX, 0, 1, HW_COORD_FEEDBACK, NO, 0, FALSE) \ FLAG( 6, 0, ECX, 1, 1, ACNT2, ANY, 13, FALSE) \ FLAG( 6, 0, ECX, 3, 1, ENERGY_PERF_BIAS, NO, 0, FALSE) #define CPUID_7_EDX_28 \ FLAG( 7, 0, EDX, 28, 1, LEVEL7EDX_RSVD1, NO, 0, FALSE) #define CPUID_7_EDX_31 \ FLAG( 7, 0, EDX, 31, 1, LEVEL7EDX_RSVD2, NO, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_7 \ FLAG( 7, 0, EBX, 0, 1, FSGSBASE, YES, 9, FALSE) \ FLAG( 7, 0, EBX, 1, 1, TSC_ADJUST, ANY, 11, FALSE) \ FLAG( 7, 0, EBX, 2, 1, SGX, YES, 16, FALSE) \ FLAG( 7, 0, EBX, 3, 1, BMI1, YES, 9, TRUE) \ FLAG( 7, 0, EBX, 4, 1, HLE, YES, 11, TRUE) \ FLAG( 7, 0, EBX, 5, 1, AVX2, YES, 11, TRUE) \ FLAG( 7, 0, EBX, 6, 1, FDP_EXCPTN_ONLY, ANY, 13, TRUE) \ FLAG( 7, 0, EBX, 7, 1, SMEP, YES, 9, FALSE) \ FLAG( 7, 0, EBX, 8, 1, BMI2, YES, 11, TRUE) \ FLAG( 7, 0, EBX, 9, 1, ENFSTRG, YES, 9, FALSE) \ FLAG( 7, 0, EBX, 10, 1, INVPCID, YES, 11, FALSE) \ FLAG( 7, 0, EBX, 11, 1, RTM, YES, 11, TRUE) \ FLAG( 7, 0, EBX, 12, 1, PQM, NO, 0, FALSE) \ FLAG( 7, 0, EBX, 13, 1, FP_SEGMENT_ZERO, ANY, 11, TRUE) \ FLAG( 7, 0, EBX, 14, 1, MPX, YES, 13, TRUE) \ FLAG( 7, 0, EBX, 15, 1, PQE, NO, 0, FALSE) \ FLAG( 7, 0, EBX, 16, 1, AVX512F, YES, 13, TRUE) \ FLAG( 7, 0, EBX, 17, 1, AVX512DQ, YES, 13, TRUE) \ FLAG( 7, 0, EBX, 18, 1, RDSEED, YES, 11, TRUE) \ FLAG( 7, 0, EBX, 19, 1, ADX, YES, 11, TRUE) \ FLAG( 7, 0, EBX, 20, 1, SMAP, YES, 11, FALSE) \ FLAG( 7, 0, EBX, 21, 1, AVX512IFMA, YES, 16, TRUE) \ FLAG( 7, 0, EBX, 23, 1, CLFLUSHOPT, YES, 13, TRUE) \ FLAG( 7, 0, EBX, 24, 1, CLWB, YES, 13, TRUE) \ FLAG( 7, 0, EBX, 25, 1, PT, YES, 16, FALSE) \ FLAG( 7, 0, EBX, 26, 1, AVX512PF, YES, 13, TRUE) \ FLAG( 7, 0, EBX, 27, 1, AVX512ER, YES, 13, TRUE) \ FLAG( 7, 0, EBX, 28, 1, AVX512CD, YES, 13, TRUE) \ FLAG( 7, 0, EBX, 29, 1, SHA, YES, 14, TRUE) \ FLAG( 7, 0, EBX, 30, 1, AVX512BW, YES, 13, TRUE) \ FLAG( 7, 0, EBX, 31, 1, AVX512VL, YES, 13, TRUE) \ FLAG( 7, 0, ECX, 0, 1, PREFETCHWT1, YES, 13, TRUE) \ FLAG( 7, 0, ECX, 1, 1, AVX512VBMI, YES, 16, TRUE) \ FLAG( 7, 0, ECX, 2, 1, UMIP, NO, 0, FALSE) \ FLAG( 7, 0, ECX, 3, 1, PKU, YES, 13, TRUE) \ FLAG( 7, 0, ECX, 4, 1, OSPKE, ANY, 13, TRUE) \ FLAG( 7, 0, ECX, 6, 1, AVX512VBMI2, NO, 0, TRUE) \ FLAG( 7, 0, ECX, 8, 1, GFNI, NO, 0, TRUE) \ FLAG( 7, 0, ECX, 9, 1, VAES, NO, 0, TRUE) \ FLAG( 7, 0, ECX, 10, 1, VPCLMULQDQ, NO, 0, TRUE) \ FLAG( 7, 0, ECX, 11, 1, AVX512VNNI, NO, 0, TRUE) \ FLAG( 7, 0, ECX, 12, 1, AVX512BITALG, NO, 0, TRUE) \ FLAG( 7, 0, ECX, 14, 1, AVX512VPOPCNTDQ, YES, 15, TRUE) \ FLAG( 7, 0, ECX, 16, 1, VA57, NO, 0, TRUE) \ FIELD( 7, 0, ECX, 17, 5, MAWA, NO, 0, TRUE) \ FLAG( 7, 0, ECX, 22, 1, RDPID, NO, 0, TRUE) \ FLAG( 7, 0, ECX, 30, 1, SGX_LC, YES, 16, FALSE) \ FLAG( 7, 0, EDX, 2, 1, AVX512QVNNIW, YES, 15, TRUE) \ FLAG( 7, 0, EDX, 3, 1, AVX512QFMAPS, YES, 15, TRUE) \ FLAG( 7, 0, EDX, 26, 1, IBRSIBPB, ANY, 9, FALSE) \ FLAG( 7, 0, EDX, 27, 1, STIBP, YES, 9, FALSE) \ CPUID_7_EDX_28 \ FLAG( 7, 0, EDX, 29, 1, ARCH_CAPABILITIES, ANY, 9, FALSE) \ CPUID_7_EDX_31 /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_A \ FIELD( A, 0, EAX, 0, 8, PMC_VERSION, NA, 0, FALSE) \ FIELD( A, 0, EAX, 8, 8, PMC_NUM_GEN, NA, 0, FALSE) \ FIELD( A, 0, EAX, 16, 8, PMC_WIDTH_GEN, NA, 0, FALSE) \ FIELD( A, 0, EAX, 24, 8, PMC_EBX_LENGTH, NA, 0, FALSE) \ FLAG( A, 0, EBX, 0, 1, PMC_CORE_CYCLES, NA, 0, FALSE) \ FLAG( A, 0, EBX, 1, 1, PMC_INSTR_RETIRED, NA, 0, FALSE) \ FLAG( A, 0, EBX, 2, 1, PMC_REF_CYCLES, NA, 0, FALSE) \ FLAG( A, 0, EBX, 3, 1, PMC_LAST_LVL_CREF, NA, 0, FALSE) \ FLAG( A, 0, EBX, 4, 1, PMC_LAST_LVL_CMISS, NA, 0, FALSE) \ FLAG( A, 0, EBX, 5, 1, PMC_BR_INST_RETIRED, NA, 0, FALSE) \ FLAG( A, 0, EBX, 6, 1, PMC_BR_MISS_RETIRED, NA, 0, FALSE) \ FIELD( A, 0, EDX, 0, 5, PMC_NUM_FIXED, NA, 0, FALSE) \ FIELD( A, 0, EDX, 5, 8, PMC_WIDTH_FIXED, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_B \ FIELD( B, 0, EAX, 0, 5, TOPOLOGY_MASK_WIDTH, NA, 0, FALSE) \ FIELD( B, 0, EBX, 0, 16, TOPOLOGY_CPUS_SHARING_LEVEL, NA, 0, FALSE) \ FIELD( B, 0, ECX, 0, 8, TOPOLOGY_LEVEL_NUMBER, NA, 0, FALSE) \ FIELD( B, 0, ECX, 8, 8, TOPOLOGY_LEVEL_TYPE, NA, 0, FALSE) \ FIELD( B, 0, EDX, 0, 32, TOPOLOGY_X2APIC_ID, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_D \ FLAG( D, 0, EAX, 0, 1, XCR0_MASTER_LEGACY_FP, YES, 8, FALSE) \ FLAG( D, 0, EAX, 1, 1, XCR0_MASTER_SSE, YES, 8, FALSE) \ FLAG( D, 0, EAX, 2, 1, XCR0_MASTER_YMM_H, YES, 8, FALSE) \ FLAG( D, 0, EAX, 3, 1, XCR0_MASTER_BNDREGS, YES, 13, FALSE) \ FLAG( D, 0, EAX, 4, 1, XCR0_MASTER_BNDCSR, YES, 13, FALSE) \ FLAG( D, 0, EAX, 5, 1, XCR0_MASTER_OPMASK, YES, 13, FALSE) \ FLAG( D, 0, EAX, 6, 1, XCR0_MASTER_ZMM_H, YES, 13, FALSE) \ FLAG( D, 0, EAX, 7, 1, XCR0_MASTER_HI16_ZMM, YES, 13, FALSE) \ FLAG( D, 0, EAX, 8, 1, XCR0_MASTER_XSS, NO, 0, FALSE) \ FLAG( D, 0, EAX, 9, 1, XCR0_MASTER_PKRU, YES, 13, FALSE) \ FIELD( D, 0, EAX, 10,22, XCR0_MASTER_LOWER, NO, 0, FALSE) \ FIELD( D, 0, EBX, 0, 32, XSAVE_ENABLED_SIZE, ANY, 8, FALSE) \ FIELD( D, 0, ECX, 0, 32, XSAVE_MAX_SIZE, YES, 8, FALSE) \ FIELD( D, 0, EDX, 0, 29, XCR0_MASTER_UPPER, NO, 0, FALSE) \ FLAG( D, 0, EDX, 30, 1, XCR0_MASTER_LWP, NO, 0, FALSE) \ FLAG( D, 0, EDX, 31, 1, XCR0_MASTER_EXTENDED_XSAVE, NO, 0, FALSE) \ FLAG( D, 1, EAX, 0, 1, XSAVEOPT, YES, 11, FALSE) \ FLAG( D, 1, EAX, 1, 1, XSAVEC, YES, 13, FALSE) \ FLAG( D, 1, EAX, 2, 1, XGETBV_ECX1, NO, 0, FALSE) \ FLAG( D, 1, EAX, 3, 1, XSAVES, YES, 13, FALSE) \ FIELD( D, 1, EBX, 0, 32, XSAVES_ENABLED_SIZE, ANY, 13, FALSE) \ FIELD( D, 1, ECX, 0, 7, XSS_XCR0_USED0, NO, 0, FALSE) \ FLAG( D, 1, ECX, 8, 1, XSS_PT, NO, 0, FALSE) \ FIELD( D, 1, ECX, 9, 1, XSS_XCR0_USED1, NO, 0, FALSE) \ FIELD( D, 1, ECX, 10,22, XSS_RSVD0, NO, 0, FALSE) \ FIELD( D, 1, EDX, 0, 32, XSS_RSVD1, NO, 0, FALSE) \ FIELD( D, 2, EAX, 0, 32, XSAVE_YMM_SIZE, YES, 9, FALSE) \ FIELD( D, 2, EBX, 0, 32, XSAVE_YMM_OFFSET, YES, 9, FALSE) \ FLAG( D, 2, ECX, 0, 1, XSAVE_YMM_SUP_BY_XSS, NO, 0, FALSE) \ FLAG( D, 2, ECX, 1, 1, XSAVE_YMM_ALIGN, YES, 13, FALSE) \ FIELD( D, 2, ECX, 2, 30, XSAVE_YMM_RSVD1, NO, 0, FALSE) \ FIELD( D, 2, EDX, 0, 32, XSAVE_YMM_RSVD2, NO, 0, FALSE) \ FIELD( D, 3, EAX, 0, 32, XSAVE_BNDREGS_SIZE, YES, 13, FALSE) \ FIELD( D, 3, EBX, 0, 32, XSAVE_BNDREGS_OFFSET, YES, 13, FALSE) \ FLAG( D, 3, ECX, 0, 1, XSAVE_BNDREGS_SUP_BY_XSS, NO, 0, FALSE) \ FLAG( D, 3, ECX, 1, 1, XSAVE_BNDREGS_ALIGN, YES, 13, FALSE) \ FIELD( D, 3, ECX, 2, 30, XSAVE_BNDREGS_RSVD1, NO, 0, FALSE) \ FIELD( D, 3, EDX, 0, 32, XSAVE_BNDREGS_RSVD2, NO, 0, FALSE) \ FIELD( D, 4, EAX, 0, 32, XSAVE_BNDCSR_SIZE, YES, 13, FALSE) \ FIELD( D, 4, EBX, 0, 32, XSAVE_BNDCSR_OFFSET, YES, 13, FALSE) \ FLAG( D, 4, ECX, 0, 1, XSAVE_BNDCSR_SUP_BY_XSS, NO, 0, FALSE) \ FLAG( D, 4, ECX, 1, 1, XSAVE_BNDCSR_ALIGN, YES, 13, FALSE) \ FIELD( D, 4, ECX, 2, 30, XSAVE_BNDCSR_RSVD1, NO, 0, FALSE) \ FIELD( D, 4, EDX, 0, 32, XSAVE_BNDCSR_RSVD2, NO, 0, FALSE) \ FIELD( D, 5, EAX, 0, 32, XSAVE_OPMASK_SIZE, YES, 13, FALSE) \ FIELD( D, 5, EBX, 0, 32, XSAVE_OPMASK_OFFSET, YES, 13, FALSE) \ FLAG( D, 5, ECX, 0, 1, XSAVE_OPMASK_SUP_BY_XSS, NO, 0, FALSE) \ FLAG( D, 5, ECX, 1, 1, XSAVE_OPMASK_ALIGN, YES, 13, FALSE) \ FIELD( D, 5, ECX, 2, 30, XSAVE_OPMASK_RSVD1, NO, 0, FALSE) \ FIELD( D, 5, EDX, 0, 32, XSAVE_OPMASK_RSVD2, NO, 0, FALSE) \ FIELD( D, 6, EAX, 0, 32, XSAVE_ZMM_H_SIZE, YES, 13, FALSE) \ FIELD( D, 6, EBX, 0, 32, XSAVE_ZMM_H_OFFSET, YES, 13, FALSE) \ FLAG( D, 6, ECX, 0, 1, XSAVE_ZMM_H_SUP_BY_XSS, NO, 0, FALSE) \ FLAG( D, 6, ECX, 1, 1, XSAVE_ZMM_H_ALIGN, YES, 13, FALSE) \ FIELD( D, 6, ECX, 2, 30, XSAVE_ZMM_H_RSVD1, NO, 0, FALSE) \ FIELD( D, 6, EDX, 0, 32, XSAVE_ZMM_H_RSVD2, NO, 0, FALSE) \ FIELD( D, 7, EAX, 0, 32, XSAVE_HI16_ZMM_SIZE, YES, 13, FALSE) \ FIELD( D, 7, EBX, 0, 32, XSAVE_HI16_ZMM_OFFSET, YES, 13, FALSE) \ FLAG( D, 7, ECX, 0, 1, XSAVE_HI16_ZMM_SUP_BY_XSS, NO, 0, FALSE) \ FLAG( D, 7, ECX, 1, 1, XSAVE_HI16_ZMM_ALIGN, YES, 13, FALSE) \ FIELD( D, 7, ECX, 2, 30, XSAVE_HI16_ZMM_RSVD1, NO, 0, FALSE) \ FIELD( D, 7, EDX, 0, 32, XSAVE_HI16_ZMM_RSVD2, NO, 0, FALSE) \ FIELD( D, 8, EAX, 0, 32, XSAVE_PT_STATE_SIZE, NO, 0, FALSE) \ FIELD( D, 8, EBX, 0, 32, XSAVE_PT_STATE_OFFSET, NO, 0, FALSE) \ FLAG( D, 8, ECX, 0, 1, XSAVE_PT_STATE_SUP_BY_XSS, NO, 0, FALSE) \ FLAG( D, 8, ECX, 1, 1, XSAVE_PT_STATE_ALIGN, NO, 0, FALSE) \ FIELD( D, 8, ECX, 2, 30, XSAVE_PT_STATE_RSVD1, NO, 0, FALSE) \ FIELD( D, 8, EDX, 0, 32, XSAVE_PT_STATE_RSVD2, NO, 0, FALSE) \ FIELD( D, 9, EAX, 0, 32, XSAVE_PKRU_SIZE, YES, 13, FALSE) \ FIELD( D, 9, EBX, 0, 32, XSAVE_PKRU_OFFSET, YES, 13, FALSE) \ FLAG( D, 9, ECX, 0, 1, XSAVE_PKRU_SUP_BY_XSS, NO, 0, FALSE) \ FLAG( D, 9, ECX, 1, 1, XSAVE_PKRU_ALIGN, YES, 13, FALSE) \ FIELD( D, 9, ECX, 2, 30, XSAVE_PKRU_RSVD1, NO, 0, FALSE) \ FIELD( D, 9, EDX, 0, 32, XSAVE_PKRU_RSVD2, NO, 0, FALSE) \ FIELD( D, 62, EAX, 0, 32, XSAVE_LWP_SIZE, NO, 0, FALSE) \ FIELD( D, 62, EBX, 0, 32, XSAVE_LWP_OFFSET, NO, 0, FALSE) \ FIELD( D, 62, ECX, 0, 32, XSAVE_LWP_RSVD1, NO, 0, FALSE) \ FIELD( D, 62, EDX, 0, 32, XSAVE_LWP_RSVD2, NO, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_F \ FIELD( F, 0, EBX, 0, 32, PQM_MAX_RMID, NO, 0, FALSE) \ FLAG( F, 0, EDX, 1, 1, PQM_CMT_SUPPORT, NO, 0, FALSE) \ FIELD( F, 1, EBX, 0, 32, PQM_CMT_CONV, NO, 0, FALSE) \ FIELD( F, 1, ECX, 0, 32, PQM_CMT_NUM_RMID, NO, 0, FALSE) \ FLAG( F, 1, EDX, 0, 1, PQM_CMT_OCCUPANCY, NO, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_10 \ FLAG( 10, 0, EBX, 1, 1, L3_QOS_ENFORCEMENT, NO, 0, FALSE) \ FIELD( 10, 1, EAX, 0, 4, RESID_CAPACITY_MASK_LENGTH, NO, 0, FALSE) \ FIELD( 10, 1, EBX, 0, 32, ISOLATION_UNIT_MAP, NO, 0, FALSE) \ FLAG( 10, 1, ECX, 1, 1, INFREQUENT_COS_UPDATE, NO, 0, FALSE) \ FLAG( 10, 1, ECX, 2, 1, CODE_AND_DATA_PRIORITIZATION, NO, 0, FALSE) \ FIELD( 10, 1, EDX, 0, 16, MAX_COS_NUMBER, NO, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_12 \ FLAG( 12, 0, EAX, 0, 1, SGX1, YES, 16, FALSE) \ FLAG( 12, 0, EAX, 1, 1, SGX2, NO, 0, FALSE) \ FLAG( 12, 0, EBX, 0, 1, SGX_MISCSELECT_EXINFO, YES, 16, FALSE) \ FIELD( 12, 0, EBX, 1, 31, SGX_MISCSELECT_RSVD, NO, 0, FALSE) \ FIELD( 12, 0, EDX, 0, 8, MAX_ENCLAVE_SIZE_NOT64, YES, 16, FALSE) \ FIELD( 12, 0, EDX, 8, 8, MAX_ENCLAVE_SIZE_64, YES, 16, FALSE) \ FIELD( 12, 1, EAX, 0, 32, SECS_ATTRIBUTES0, YES, 16, FALSE) \ FIELD( 12, 1, EBX, 0, 32, SECS_ATTRIBUTES1, YES, 16, FALSE) \ FIELD( 12, 1, ECX, 0, 32, SECS_ATTRIBUTES2, YES, 16, FALSE) \ FIELD( 12, 1, EDX, 0, 32, SECS_ATTRIBUTES3, YES, 16, FALSE) \ FIELD( 12, 2, EAX, 0, 15, EPC00_VALID, YES, 16, FALSE) \ FIELD( 12, 2, EAX, 12, 20, EPC00_BASE_LOW, YES, 16, FALSE) \ FIELD( 12, 2, EBX, 0, 20, EPC00_BASE_HIGH, YES, 16, FALSE) \ FIELD( 12, 2, ECX, 0, 15, EPC00_PROTECTED, YES, 16, FALSE) \ FIELD( 12, 2, ECX, 12, 20, EPC00_SIZE_LOW, YES, 16, FALSE) \ FIELD( 12, 2, EDX, 0, 20, EPC00_SIZE_HIGH, YES, 16, FALSE) \ FIELD( 12, 3, EAX, 0, 4, EPC01_VALID, NO, 0, FALSE) \ FIELD( 12, 3, EAX, 12, 20, EPC01_BASE_LOW, NO, 0, FALSE) \ FIELD( 12, 3, EBX, 0, 20, EPC01_BASE_HIGH, NO, 0, FALSE) \ FIELD( 12, 3, ECX, 0, 4, EPC01_PROTECTED, NO, 0, FALSE) \ FIELD( 12, 3, ECX, 12, 20, EPC01_SIZE_LOW, NO, 0, FALSE) \ FIELD( 12, 3, EDX, 0, 20, EPC01_SIZE_HIGH, NO, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_14 \ FIELD( 14, 0, EAX, 0, 32, MAX_PT_SUB_LEAF, YES, 16, FALSE) \ FLAG( 14, 0, EBX, 0, 1, CR3FTR_AND_MATCHMSR_AVAILABLE, YES, 16, FALSE) \ FLAG( 14, 0, EBX, 1, 1, PSB_AND_CYCLE_ACCURATE_MODE, YES, 16, FALSE) \ FLAG( 14, 0, EBX, 2, 1, IP_TRACESTOP_FTR_PTMSR_PERSIST, YES, 16, FALSE) \ FLAG( 14, 0, EBX, 3, 1, MTC_PKT_GENERATION_SUPPORTED, YES, 16, FALSE) \ FLAG( 14, 0, ECX, 0, 1, TOPA_OUTPUT_SUPPORTED, YES, 16, FALSE) \ FLAG( 14, 0, ECX, 1, 1, TOPA_ALLOW_MULTIPLE_ENTRIES, YES, 16, FALSE) \ FLAG( 14, 0, ECX, 2, 1, SINGLE_RANGE_OUTPUT_SCHEME, YES, 16, FALSE) \ FLAG( 14, 0, ECX, 3, 1, TRACE_TRANSPORT_SUBSYSTEM, NO, 0, FALSE) \ FLAG( 14, 0, ECX, 31, 1, LIP_PRESENT_FOR_IP_PAYLOADS, YES, 16, FALSE) \ FIELD( 14, 1, EAX, 0, 2, NUM_ADDR_RANGE_FOR_FILTERING, YES, 16, FALSE) \ FIELD( 14, 1, EAX, 16, 16, SUPPORTED_MTC_ENCODINGS, YES, 16, FALSE) \ FIELD( 14, 1, EBX, 0, 16, SUPPORTED_CYCLE_THRESHOLD_ENCODINGS, YES,16,FALSE) \ FIELD( 14, 1, EBX, 16, 16, SUPPORTED_PSB_FREQ_ENCODINGS, YES, 16, FALSE) \ /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_15 \ FIELD( 15, 0, EAX, 0, 32, DENOM_TSC_TO_CORE_CRYSTAL_CLK, NO, 0, FALSE) \ FIELD( 15, 0, EBX, 0, 32, NUMER_TSC_TO_CORE_CRYSTAL_CLK, NO, 0, FALSE) \ /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_16 \ FIELD( 16, 0, EAX, 0, 16, PROC_BASE_FREQ, NO, 0, FALSE) \ FIELD( 16, 0, EBX, 0, 16, PROC_MIN_FREQ, NO, 0, FALSE) \ FIELD( 16, 0, ECX, 0, 16, BUS_FREQ, NO, 0, FALSE) \ /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_17 \ FIELD( 17, 0, EAX, 0, 31, MAX_SOCID_INDEX, NO, 0, FALSE) \ FIELD( 17, 0, EBX, 0, 16, SOC_VENDOR_ID, NO, 0, FALSE) \ FIELD( 17, 0, EBX, 16, 1, SOC_INDUSTRY_STD, NO, 0, FALSE) \ FIELD( 17, 0, ECX, 0, 31, SOC_PROJECT_ID, NO, 0, FALSE) \ FIELD( 17, 0, EDX, 0, 31, SOC_STEPPING_ID, NO, 0, FALSE) \ FIELD( 17, 1, EAX, 0, 32, SOC_VENDOR_BRAND_STRING_1_0, NO, 0, FALSE) \ FIELD( 17, 1, EBX, 0, 32, SOC_VENDOR_BRAND_STRING_1_1, NO, 0, FALSE) \ FIELD( 17, 1, ECX, 0, 32, SOC_VENDOR_BRAND_STRING_1_2, NO, 0, FALSE) \ FIELD( 17, 1, EDX, 0, 32, SOC_VENDOR_BRAND_STRING_1_3, NO, 0, FALSE) \ FIELD( 17, 2, EAX, 0, 32, SOC_VENDOR_BRAND_STRING_2_0, NO, 0, FALSE) \ FIELD( 17, 2, EBX, 0, 32, SOC_VENDOR_BRAND_STRING_2_1, NO, 0, FALSE) \ FIELD( 17, 2, ECX, 0, 32, SOC_VENDOR_BRAND_STRING_2_2, NO, 0, FALSE) \ FIELD( 17, 2, EDX, 0, 32, SOC_VENDOR_BRAND_STRING_2_3, NO, 0, FALSE) \ FIELD( 17, 3, EAX, 0, 32, SOC_VENDOR_BRAND_STRING_3_0, NO, 0, FALSE) \ FIELD( 17, 3, EBX, 0, 32, SOC_VENDOR_BRAND_STRING_3_1, NO, 0, FALSE) \ FIELD( 17, 3, ECX, 0, 32, SOC_VENDOR_BRAND_STRING_3_2, NO, 0, FALSE) \ FIELD( 17, 3, EDX, 0, 32, SOC_VENDOR_BRAND_STRING_3_3, NO, 0, FALSE) \ /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_400 \ FIELD(400, 0, EAX, 0, 32, MAX_HYP_LEVEL, NA, 0, FALSE) \ FIELD(400, 0, EBX, 0, 32, HYPERVISOR_VENDOR0, NA, 0, FALSE) \ FIELD(400, 0, ECX, 0, 32, HYPERVISOR_VENDOR1, NA, 0, FALSE) \ FIELD(400, 0, EDX, 0, 32, HYPERVISOR_VENDOR2, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_401 \ FIELD(401, 0, EAX, 0, 32, HV_INTERFACE_SIGNATURE, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_402 \ FIELD(402, 0, EAX, 0, 32, BUILD_NUMBER, NA, 0, FALSE) \ FIELD(402, 0, EBX, 0, 16, MINOR_VERSION, NA, 0, FALSE) \ FIELD(402, 0, EBX, 16, 16, MAJOR_VERSION, NA, 0, FALSE) \ FIELD(402, 0, ECX, 0, 32, SERVICE_PACK, NA, 0, FALSE) \ FIELD(402, 0, EDX, 0, 24, SERVICE_NUMBER, NA, 0, FALSE) \ FIELD(402, 0, EDX, 24, 8, SERVICE_BRANCH, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_403 \ FLAG( 403, 0, EAX, 0, 1, VP_RUNTIME_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 1, 1, REF_COUNTER_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 2, 1, BASIC_SYNIC_MSRS_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 3, 1, SYNTH_TIMER_MSRS_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 4, 1, APIC_ACCESS_MSRS_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 5, 1, HYPERCALL_MSRS_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 6, 1, VP_INDEX_MSR_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 7, 1, VIRT_RESET_MSR_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 8, 1, STATS_PAGES_MSRS_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 9, 1, REF_TSC_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 10, 1, GUEST_IDLE_MSR_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 11, 1, FREQUENCY_MSRS_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EAX, 12, 1, SYNTH_DEBUG_MSRS_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 0, 1, CREATE_PARTITIONS_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 1, 1, ACCESS_PARTITION_ID_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 2, 1, ACCESS_MEMORY_POOL_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 3, 1, ADJUST_MESSAGE_BUFFERS_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 4, 1, POST_MESSAGES_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 5, 1, SIGNAL_EVENTS_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 6, 1, CREATE_PORT_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 7, 1, CONNECT_PORT_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 8, 1, ACCESS_STATS_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 11, 1, DEBUGGING_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 12, 1, CPU_MANAGEMENT_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 13, 1, CONFIGURE_PROFILER_FLAG, NA, 0, FALSE) \ FLAG( 403, 0, EBX, 14, 1, ENABLE_EXPANDED_STACKWALKING_FLAG, NA, 0, FALSE) \ FIELD(403, 0, ECX, 0, 4, MAX_POWER_STATE, NA, 0, FALSE) \ FLAG( 403, 0, ECX, 4, 1, HPET_NEEDED_FOR_C3, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 0, 1, MWAIT_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 1, 1, GUEST_DEBUGGING_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 2, 1, PERFORMANCE_MONITOR_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 3, 1, CPU_DYN_PARTITIONING_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 4, 1, XMM_REGISTERS_FOR_HYPERCALL_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 5, 1, GUEST_IDLE_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 6, 1, HYPERVISOR_SLEEP_STATE_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 7, 1, NUMA_DISTANCE_QUERY_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 8, 1, TIMER_FREQUENCY_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 9, 1, SYNTH_MACHINE_CHECK_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 10, 1, GUEST_CRASH_MSRS_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 11, 1, DEBUG_MSRS_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 12, 1, NPIEP1_AVAIL, NA, 0, FALSE) \ FLAG( 403, 0, EDX, 13, 1, DISABLE_HYPERVISOR_AVAIL, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_404 \ FLAG( 404, 0, EAX, 0, 1, USE_HYPERCALL_TO_SWITCH_ADDR_SPACE,NA, 0, FALSE) \ FLAG( 404, 0, EAX, 1, 1, USE_HYPERCALL_TO_FLUSH_TLB, NA, 0, FALSE) \ FLAG( 404, 0, EAX, 2, 1, USE_HYPERCALL_FOR_TLB_SHOOTDOWN, NA, 0, FALSE) \ FLAG( 404, 0, EAX, 3, 1, USE_MSRS_FOR_EOI_ICR_TPR, NA, 0, FALSE) \ FLAG( 404, 0, EAX, 4, 1, USE_MSR_FOR_RESET, NA, 0, FALSE) \ FLAG( 404, 0, EAX, 5, 1, USE_RELAXED_TIMING, NA, 0, FALSE) \ FLAG( 404, 0, EAX, 6, 1, USE_DMA_REMAPPING, NA, 0, FALSE) \ FLAG( 404, 0, EAX, 7, 1, USE_INTERRUPT_REMAPPING, NA, 0, FALSE) \ FLAG( 404, 0, EAX, 8, 1, USE_X2APIC, NA, 0, FALSE) \ FLAG( 404, 0, EAX, 9, 1, DEPRECATE_AUTOEOI, NA, 0, FALSE) \ FIELD(404, 0, EBX, 0, 32, SPINLOCK_RETRIES, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_405 \ FIELD(405, 0, EAX, 0, 32, MAX_VCPU, NA, 0, FALSE) \ FIELD(405, 0, EBX, 0, 32, MAX_LCPU, NA, 0, FALSE) \ FIELD(405, 0, ECX, 0, 32, MAX_REMAPPABLE_VECTORS, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_406 \ FLAG( 406, 0, EAX, 0, 1, APIC_OVERLAY_ASSIST, NA, 0, FALSE) \ FLAG( 406, 0, EAX, 1, 1, MSR_BITMAPS, NA, 0, FALSE) \ FLAG( 406, 0, EAX, 2, 1, ARCH_PMCS, NA, 0, FALSE) \ FLAG( 406, 0, EAX, 3, 1, SLAT, NA, 0, FALSE) \ FLAG( 406, 0, EAX, 4, 1, DMA_REMAPPING, NA, 0, FALSE) \ FLAG( 406, 0, EAX, 5, 1, INTERRUPT_REMAPPING, NA, 0, FALSE) \ FLAG( 406, 0, EAX, 6, 1, MEMORY_PATROL_SCRUBBER, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_410 \ FIELD(410, 0, EAX, 0, 32, TSC_HZ, NA, 0, FALSE) \ FIELD(410, 0, EBX, 0, 32, APICBUS_HZ, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_80 \ FIELD( 80, 0, EAX, 0, 32, NUM_EXT_LEVELS, NA, 0, FALSE) \ FIELD( 80, 0, EBX, 0, 32, LEAF80_VENDOR1, NA, 0, FALSE) \ FIELD( 80, 0, ECX, 0, 32, LEAF80_VENDOR3, NA, 0, FALSE) \ FIELD( 80, 0, EDX, 0, 32, LEAF80_VENDOR2, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_81 \ FIELD( 81, 0, EAX, 0, 32, UNKNOWN81EAX, ANY, 4, FALSE) \ FIELD( 81, 0, EAX, 0, 4, LEAF81_STEPPING, ANY, 4, FALSE) \ FIELD( 81, 0, EAX, 4, 4, LEAF81_MODEL, ANY, 4, FALSE) \ FIELD( 81, 0, EAX, 8, 4, LEAF81_FAMILY, ANY, 4, FALSE) \ FIELD( 81, 0, EAX, 12, 2, LEAF81_TYPE, ANY, 4, FALSE) \ FIELD( 81, 0, EAX, 16, 4, LEAF81_EXTENDED_MODEL, ANY, 4, FALSE) \ FIELD( 81, 0, EAX, 20, 8, LEAF81_EXTENDED_FAMILY, ANY, 4, FALSE) \ FIELD( 81, 0, EBX, 0, 32, UNKNOWN81EBX, ANY, 4, FALSE) \ FIELD( 81, 0, EBX, 0, 16, LEAF81_BRAND_ID, ANY, 4, FALSE) \ FIELD( 81, 0, EBX, 16, 16, UNDEF, ANY, 4, FALSE) \ FLAG( 81, 0, ECX, 0, 1, LAHF64, YES, 4, TRUE) \ FLAG( 81, 0, ECX, 1, 1, CMPLEGACY, ANY, 9, FALSE) \ FLAG( 81, 0, ECX, 2, 1, SVM, YES, 7, FALSE) \ FLAG( 81, 0, ECX, 3, 1, EXTAPICSPC, YES, 4, FALSE) \ FLAG( 81, 0, ECX, 4, 1, CR8AVAIL, YES, 4, FALSE) \ FLAG( 81, 0, ECX, 5, 1, ABM, YES, 7, TRUE) \ FLAG( 81, 0, ECX, 6, 1, SSE4A, YES, 4, TRUE) \ FLAG( 81, 0, ECX, 7, 1, MISALIGNED_SSE, YES, 4, TRUE) \ FLAG( 81, 0, ECX, 8, 1, 3DNPREFETCH, YES, 4, TRUE) \ FLAG( 81, 0, ECX, 9, 1, OSVW, ANY, 8, FALSE) \ FLAG( 81, 0, ECX, 10, 1, IBS, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 11, 1, XOP, YES, 8, TRUE) \ FLAG( 81, 0, ECX, 12, 1, SKINIT, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 13, 1, WATCHDOG, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 15, 1, LWP, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 16, 1, FMA4, YES, 8, TRUE) \ FLAG( 81, 0, ECX, 17, 1, TCE, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 19, 1, NODEID_MSR, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 21, 1, TBM, YES, 9, TRUE) \ FLAG( 81, 0, ECX, 22, 1, TOPOLOGY, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 23, 1, PERFCORE, ANY, 4, TRUE) \ FLAG( 81, 0, ECX, 24, 1, PERFNB, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 26, 1, DATABK, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 27, 1, PERFTSC, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 28, 1, PERFL3, NO, 0, FALSE) \ FLAG( 81, 0, ECX, 29, 1, MWAITX, NO, 0, FALSE) \ FLAG( 81, 0, EDX, 0, 1, LEAF81_FPU, YES, 4, TRUE) \ FLAG( 81, 0, EDX, 1, 1, LEAF81_VME, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 2, 1, LEAF81_DE, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 3, 1, LEAF81_PSE, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 4, 1, LEAF81_TSC, YES, 4, TRUE) \ FLAG( 81, 0, EDX, 5, 1, LEAF81_MSR, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 6, 1, LEAF81_PAE, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 7, 1, LEAF81_MCE, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 8, 1, LEAF81_CX8, YES, 4, TRUE) \ FLAG( 81, 0, EDX, 9, 1, LEAF81_APIC, ANY, 4, FALSE) \ FLAG( 81, 0, EDX, 11, 1, SYSC, ANY, 4, TRUE) \ FLAG( 81, 0, EDX, 12, 1, LEAF81_MTRR, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 13, 1, LEAF81_PGE, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 14, 1, LEAF81_MCA, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 15, 1, LEAF81_CMOV, YES, 4, TRUE) \ FLAG( 81, 0, EDX, 16, 1, LEAF81_PAT, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 17, 1, LEAF81_PSE36, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 20, 1, NX, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 22, 1, MMXEXT, YES, 4, TRUE) \ FLAG( 81, 0, EDX, 23, 1, LEAF81_MMX, YES, 4, TRUE) \ FLAG( 81, 0, EDX, 24, 1, LEAF81_FXSR, YES, 4, TRUE) \ FLAG( 81, 0, EDX, 25, 1, FFXSR, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 26, 1, PDPE1GB, YES, 7, FALSE) \ FLAG( 81, 0, EDX, 27, 1, RDTSCP, YES, 4, TRUE) \ FLAG( 81, 0, EDX, 29, 1, LM, YES, 4, FALSE) \ FLAG( 81, 0, EDX, 30, 1, 3DNOWPLUS, YES, 4, TRUE) \ FLAG( 81, 0, EDX, 31, 1, 3DNOW, YES, 4, TRUE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_82 \ FIELD( 82, 0, EAX, 0, 32, LEAF82_BRAND_STRING_EAX, NA, 0, FALSE) \ FIELD( 82, 0, EBX, 0, 32, LEAF82_BRAND_STRING_EBX, NA, 0, FALSE) \ FIELD( 82, 0, ECX, 0, 32, LEAF82_BRAND_STRING_ECX, NA, 0, FALSE) \ FIELD( 82, 0, EDX, 0, 32, LEAF82_BRAND_STRING_EDX, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_83 \ FIELD( 83, 0, EAX, 0, 32, LEAF83_BRAND_STRING_EAX, NA, 0, FALSE) \ FIELD( 83, 0, EBX, 0, 32, LEAF83_BRAND_STRING_EBX, NA, 0, FALSE) \ FIELD( 83, 0, ECX, 0, 32, LEAF83_BRAND_STRING_ECX, NA, 0, FALSE) \ FIELD( 83, 0, EDX, 0, 32, LEAF83_BRAND_STRING_EDX, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_84 \ FIELD( 84, 0, EAX, 0, 32, LEAF84_BRAND_STRING_EAX, NA, 0, FALSE) \ FIELD( 84, 0, EBX, 0, 32, LEAF84_BRAND_STRING_EBX, NA, 0, FALSE) \ FIELD( 84, 0, ECX, 0, 32, LEAF84_BRAND_STRING_ECX, NA, 0, FALSE) \ FIELD( 84, 0, EDX, 0, 32, LEAF84_BRAND_STRING_EDX, NA, 0, FALSE) /* LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_85 \ FIELD( 85, 0, EAX, 0, 8, ITLB_ENTRIES_2M4M_PGS, NA, 0, FALSE) \ FIELD( 85, 0, EAX, 8, 8, ITLB_ASSOC_2M4M_PGS, NA, 0, FALSE) \ FIELD( 85, 0, EAX, 16, 8, DTLB_ENTRIES_2M4M_PGS, NA, 0, FALSE) \ FIELD( 85, 0, EAX, 24, 8, DTLB_ASSOC_2M4M_PGS, NA, 0, FALSE) \ FIELD( 85, 0, EBX, 0, 8, ITLB_ENTRIES_4K_PGS, NA, 0, FALSE) \ FIELD( 85, 0, EBX, 8, 8, ITLB_ASSOC_4K_PGS, NA, 0, FALSE) \ FIELD( 85, 0, EBX, 16, 8, DTLB_ENTRIES_4K_PGS, NA, 0, FALSE) \ FIELD( 85, 0, EBX, 24, 8, DTLB_ASSOC_4K_PGS, NA, 0, FALSE) \ FIELD( 85, 0, ECX, 0, 8, L1_DCACHE_LINE_SIZE, NA, 0, FALSE) \ FIELD( 85, 0, ECX, 8, 8, L1_DCACHE_LINES_PER_TAG, NA, 0, FALSE) \ FIELD( 85, 0, ECX, 16, 8, L1_DCACHE_ASSOC, NA, 0, FALSE) \ FIELD( 85, 0, ECX, 24, 8, L1_DCACHE_SIZE, NA, 0, FALSE) \ FIELD( 85, 0, EDX, 0, 8, L1_ICACHE_LINE_SIZE, NA, 0, FALSE) \ FIELD( 85, 0, EDX, 8, 8, L1_ICACHE_LINES_PER_TAG, NA, 0, FALSE) \ FIELD( 85, 0, EDX, 16, 8, L1_ICACHE_ASSOC, NA, 0, FALSE) \ FIELD( 85, 0, EDX, 24, 8, L1_ICACHE_SIZE, NA, 0, FALSE) /* LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_86 \ FIELD( 86, 0, EAX, 0, 12, L2_ITLB_ENTRIES_2M4M_PGS, NA, 0, FALSE) \ FIELD( 86, 0, EAX, 12, 4, L2_ITLB_ASSOC_2M4M_PGS, NA, 0, FALSE) \ FIELD( 86, 0, EAX, 16, 12, L2_DTLB_ENTRIES_2M4M_PGS, NA, 0, FALSE) \ FIELD( 86, 0, EAX, 28, 4, L2_DTLB_ASSOC_2M4M_PGS, NA, 0, FALSE) \ FIELD( 86, 0, EBX, 0, 12, L2_ITLB_ENTRIES_4K_PGS, NA, 0, FALSE) \ FIELD( 86, 0, EBX, 12, 4, L2_ITLB_ASSOC_4K_PGS, NA, 0, FALSE) \ FIELD( 86, 0, EBX, 16, 12, L2_DTLB_ENTRIES_4K_PGS, NA, 0, FALSE) \ FIELD( 86, 0, EBX, 28, 4, L2_DTLB_ASSOC_4K_PGS, NA, 0, FALSE) \ FIELD( 86, 0, ECX, 0, 8, L2CACHE_LINE, NA, 0, FALSE) \ FIELD( 86, 0, ECX, 8, 4, L2CACHE_LINE_PER_TAG, NA, 0, FALSE) \ FIELD( 86, 0, ECX, 12, 4, L2CACHE_WAYS, NA, 0, FALSE) \ FIELD( 86, 0, ECX, 16, 16, L2CACHE_SIZE, NA, 0, FALSE) \ FIELD( 86, 0, EDX, 0, 8, L3CACHE_LINE, NA, 0, FALSE) \ FIELD( 86, 0, EDX, 8, 4, L3CACHE_LINE_PER_TAG, NA, 0, FALSE) \ FIELD( 86, 0, EDX, 12, 4, L3CACHE_WAYS, NA, 0, FALSE) \ FIELD( 86, 0, EDX, 18, 14, L3CACHE_SIZE, NA, 0, FALSE) /* LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_87 \ FLAG( 87, 0, EBX, 0, 1, MCA_OVERFLOW_RECOV, NA, 0, FALSE) \ FLAG( 87, 0, EBX, 1, 1, SUCCOR, NA, 0, FALSE) \ FLAG( 87, 0, EBX, 2, 1, HWA, NA, 0, FALSE) \ FLAG( 87, 0, EBX, 3, 1, SCALABLE_MCA, NA, 0, FALSE) \ FLAG( 87, 0, EBX, 4, 1, PFEH_SUPPORT_PRESENT, NA, 0, FALSE) \ FLAG( 87, 0, EDX, 0, 1, TS, NA, 0, FALSE) \ FLAG( 87, 0, EDX, 1, 1, FID, NA, 0, FALSE) \ FLAG( 87, 0, EDX, 2, 1, VID, NA, 0, FALSE) \ FLAG( 87, 0, EDX, 3, 1, TTP, NA, 0, FALSE) \ FLAG( 87, 0, EDX, 4, 1, LEAF87_TM, NA, 0, FALSE) \ FLAG( 87, 0, EDX, 5, 1, STC, NA, 0, FALSE) \ FLAG( 87, 0, EDX, 6, 1, 100MHZSTEPS, NA, 0, FALSE) \ FLAG( 87, 0, EDX, 7, 1, HWPSTATE, NA, 0, FALSE) \ FLAG( 87, 0, EDX, 8, 1, TSC_INVARIANT, NA, 0, FALSE) \ FLAG( 87, 0, EDX, 9, 1, CORE_PERF_BOOST, NA, 0, FALSE) #define CPUID_88_EBX_14 \ FLAG( 88, 0, EBX, 14, 1, LEAF88_RSVD1, NO, 0, FALSE) #define CPUID_88_EBX_15 \ FLAG( 88, 0, EBX, 15, 1, LEAF88_RSVD2, NO, 0, FALSE) #define CPUID_88_EBX_16 \ FLAG( 88, 0, EBX, 16, 1, LEAF88_RSVD3, NO, 0, FALSE) #define CPUID_88_EBX_17 \ FLAG( 88, 0, EBX, 17, 1, LEAF88_RSVD4, NO, 0, FALSE) #define CPUID_88_EBX_18 \ FLAG( 88, 0, EBX, 18, 1, LEAF88_RSVD5, NO, 0, FALSE) /* LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_88 \ FIELD( 88, 0, EAX, 0, 8, PHYS_BITS, YES, 4, FALSE) \ FIELD( 88, 0, EAX, 8, 8, VIRT_BITS, YES, 4, FALSE) \ FIELD( 88, 0, EAX, 16, 8, GUEST_PHYS_ADDR_SZ, YES, 8, FALSE) \ FLAG( 88, 0, EBX, 0, 1, CLZERO, YES, 14, TRUE) \ FLAG( 88, 0, EBX, 1, 1, IRPERF, NO, 0, FALSE) \ FLAG( 88, 0, EBX, 2, 1, XSAVE_ERR_PTR, NO, 0, FALSE) \ FLAG( 88, 0, EBX, 12, 1, LEAF88_IBPB, ANY, 9, FALSE) \ CPUID_88_EBX_14 \ CPUID_88_EBX_15 \ CPUID_88_EBX_16 \ CPUID_88_EBX_17 \ CPUID_88_EBX_18 \ FIELD( 88, 0, ECX, 0, 8, LEAF88_CORE_COUNT, YES, 4, FALSE) \ FIELD( 88, 0, ECX, 12, 4, APICID_COREID_SIZE, YES, 7, FALSE) \ FIELD( 88, 0, ECX, 16, 2, PERFTSC_SIZE, NO, 0, FALSE) #define CPUID_8A_EDX_11 \ FLAG( 8A, 0, EDX, 11, 1, SVMEDX_RSVD1, NO, 0, FALSE) #define CPUID_8A_EDX_14 \ FLAG( 8A, 0, EDX, 14, 1, SVMEDX_RSVD2, NO, 0, FALSE) /* LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_8A \ FIELD( 8A, 0, EAX, 0, 8, SVM_REVISION, YES, 4, FALSE) \ FLAG( 8A, 0, EAX, 8, 1, SVM_HYPERVISOR, NO, 0, FALSE) \ FIELD( 8A, 0, EAX, 9, 23, SVMEAX_RSVD, NO, 0, FALSE) \ FIELD( 8A, 0, EBX, 0, 32, SVM_NUM_ASIDS, YES, 7, FALSE) \ FIELD( 8A, 0, ECX, 0, 32, SVMECX_RSVD, NO, 0, FALSE) \ FLAG( 8A, 0, EDX, 0, 1, SVM_NPT, YES, 7, FALSE) \ FLAG( 8A, 0, EDX, 1, 1, SVM_LBR, NO, 0, FALSE) \ FLAG( 8A, 0, EDX, 2, 1, SVM_LOCK, ANY, 7, FALSE) \ FLAG( 8A, 0, EDX, 3, 1, SVM_NRIP, YES, 7, FALSE) \ FLAG( 8A, 0, EDX, 4, 1, SVM_TSC_RATE_MSR, NO, 0, FALSE) \ FLAG( 8A, 0, EDX, 5, 1, SVM_VMCB_CLEAN, YES, 7, FALSE) \ FLAG( 8A, 0, EDX, 6, 1, SVM_FLUSH_BY_ASID, YES, 7, FALSE) \ FLAG( 8A, 0, EDX, 7, 1, SVM_DECODE_ASSISTS, YES, 7, FALSE) \ FIELD( 8A, 0, EDX, 8, 2, SVMEDX_RSVD0, NO, 0, FALSE) \ FLAG( 8A, 0, EDX, 10, 1, SVM_PAUSE_FILTER, NO, 0, FALSE) \ CPUID_8A_EDX_11 \ FLAG( 8A, 0, EDX, 12, 1, SVM_PAUSE_THRESHOLD, NO, 0, FALSE) \ FLAG( 8A, 0, EDX, 13, 1, SVM_AVIC, NO, 0, FALSE) \ CPUID_8A_EDX_14 \ FLAG( 8A, 0, EDX, 15, 1, SVM_V_VMSAVE_VMLOAD, NO, 0, FALSE) \ FLAG( 8A, 0, EDX, 16, 1, SVM_VGIF, NO, 0, FALSE) \ FIELD( 8A, 0, EDX, 17, 15, SVMEDX_RSVD, NO, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_819 \ FIELD(819, 0, EAX, 0, 12, L1_ITLB_ENTRIES_1G_PGS, NA, 0, FALSE) \ FIELD(819, 0, EAX, 12, 4, L1_ITLB_ASSOC_1G_PGS, NA, 0, FALSE) \ FIELD(819, 0, EAX, 16, 12, L1_DTLB_ENTRIES_1G_PGS, NA, 0, FALSE) \ FIELD(819, 0, EAX, 28, 4, L1_DTLB_ASSOC_1G_PGS, NA, 0, FALSE) \ FIELD(819, 0, EBX, 0, 12, L2_ITLB_ENTRIES_1G_PGS, NA, 0, FALSE) \ FIELD(819, 0, EBX, 12, 4, L2_ITLB_ASSOC_1G_PGS, NA, 0, FALSE) \ FIELD(819, 0, EBX, 16, 12, L2_DTLB_ENTRIES_1G_PGS, NA, 0, FALSE) \ FIELD(819, 0, EBX, 28, 4, L2_DTLB_ASSOC_1G_PGS, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_81A \ FLAG( 81A, 0, EAX, 0, 1, FP128, NA, 0, FALSE) \ FLAG( 81A, 0, EAX, 1, 1, MOVU, NA, 0, FALSE) \ FLAG( 81A, 0, EAX, 2, 1, FP256, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_81B \ FLAG( 81B, 0, EAX, 0, 1, IBS_FFV, NA, 0, FALSE) \ FLAG( 81B, 0, EAX, 1, 1, IBS_FETCHSAM, NA, 0, FALSE) \ FLAG( 81B, 0, EAX, 2, 1, IBS_OPSAM, NA, 0, FALSE) \ FLAG( 81B, 0, EAX, 3, 1, RW_OPCOUNT, NA, 0, FALSE) \ FLAG( 81B, 0, EAX, 4, 1, OPCOUNT, NA, 0, FALSE) \ FLAG( 81B, 0, EAX, 5, 1, BRANCH_TARGET_ADDR, NA, 0, FALSE) \ FLAG( 81B, 0, EAX, 6, 1, OPCOUNT_EXT, NA, 0, FALSE) \ FLAG( 81B, 0, EAX, 7, 1, RIP_INVALID_CHECK, NA, 0, FALSE) \ FLAG( 81B, 0, EAX, 8, 1, OP_BRN_FUSE, NA, 0, FALSE) \ FLAG( 81B, 0, EAX, 9, 1, IBS_FETCH_CTL_EXTD, NA, 0, FALSE) \ FLAG( 81B, 0, EAX, 10, 1, IBS_OP_DATA4, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_81C \ FLAG( 81C, 0, EAX, 0, 1, LWP_AVAIL, NA, 0, FALSE) \ FLAG( 81C, 0, EAX, 1, 1, LWP_VAL_AVAIL, NA, 0, FALSE) \ FLAG( 81C, 0, EAX, 2, 1, LWP_IRE_AVAIL, NA, 0, FALSE) \ FLAG( 81C, 0, EAX, 3, 1, LWP_BRE_AVAIL, NA, 0, FALSE) \ FLAG( 81C, 0, EAX, 4, 1, LWP_DME_AVAIL, NA, 0, FALSE) \ FLAG( 81C, 0, EAX, 5, 1, LWP_CNH_AVAIL, NA, 0, FALSE) \ FLAG( 81C, 0, EAX, 6, 1, LWP_RNH_AVAIL, NA, 0, FALSE) \ FLAG( 81C, 0, EAX, 29, 1, LWP_CONT_AVAIL, NA, 0, FALSE) \ FLAG( 81C, 0, EAX, 30, 1, LWP_PTSC_AVAIL, NA, 0, FALSE) \ FLAG( 81C, 0, EAX, 31, 1, LWP_INT_AVAIL, NA, 0, FALSE) \ FIELD(81C, 0, EBX, 0, 8, LWP_CB_SIZE, NA, 0, FALSE) \ FIELD(81C, 0, EBX, 8, 8, LWP_EVENT_SIZE, NA, 0, FALSE) \ FIELD(81C, 0, EBX, 16, 8, LWP_MAX_EVENTS, NA, 0, FALSE) \ FIELD(81C, 0, EBX, 24, 8, LWP_EVENT_OFFSET, NA, 0, FALSE) \ FIELD(81C, 0, ECX, 0, 4, LWP_LATENCY_MAX, NA, 0, FALSE) \ FLAG( 81C, 0, ECX, 5, 1, LWP_DATA_ADDR_VALID, NA, 0, FALSE) \ FIELD(81C, 0, ECX, 6, 3, LWP_LATENCY_ROUND, NA, 0, FALSE) \ FIELD(81C, 0, ECX, 9, 7, LWP_VERSION, NA, 0, FALSE) \ FIELD(81C, 0, ECX, 16, 8, LWP_MIN_BUF_SIZE, NA, 0, FALSE) \ FLAG( 81C, 0, ECX, 28, 1, LWP_BRANCH_PRED, NA, 0, FALSE) \ FLAG( 81C, 0, ECX, 29, 1, LWP_IP_FILTERING, NA, 0, FALSE) \ FLAG( 81C, 0, ECX, 30, 1, LWP_CACHE_LEVEL, NA, 0, FALSE) \ FLAG( 81C, 0, ECX, 31, 1, LWP_CACHE_LATENCY, NA, 0, FALSE) \ FLAG( 81C, 0, EDX, 0, 1, LWP_SUPPORTED, NA, 0, FALSE) \ FLAG( 81C, 0, EDX, 1, 1, LWP_VAL_SUPPORTED, NA, 0, FALSE) \ FLAG( 81C, 0, EDX, 2, 1, LWP_IRE_SUPPORTED, NA, 0, FALSE) \ FLAG( 81C, 0, EDX, 3, 1, LWP_BRE_SUPPORTED, NA, 0, FALSE) \ FLAG( 81C, 0, EDX, 4, 1, LWP_DME_SUPPORTED, NA, 0, FALSE) \ FLAG( 81C, 0, EDX, 5, 1, LWP_CNH_SUPPORTED, NA, 0, FALSE) \ FLAG( 81C, 0, EDX, 6, 1, LWP_RNH_SUPPORTED, NA, 0, FALSE) \ FLAG( 81C, 0, EDX, 29, 1, LWP_CONT_SUPPORTED, NA, 0, FALSE) \ FLAG( 81C, 0, EDX, 30, 1, LWP_PTSC_SUPPORTED, NA, 0, FALSE) \ FLAG( 81C, 0, EDX, 31, 1, LWP_INT_SUPPORTED, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_81D \ FIELD(81D, 0, EAX, 0, 5, LEAF81D_CACHE_TYPE, NA, 0, FALSE) \ FIELD(81D, 0, EAX, 5, 3, LEAF81D_CACHE_LEVEL, NA, 0, FALSE) \ FLAG( 81D, 0, EAX, 8, 1, LEAF81D_CACHE_SELF_INIT, NA, 0, FALSE) \ FLAG( 81D, 0, EAX, 9, 1, LEAF81D_CACHE_FULLY_ASSOC, NA, 0, FALSE) \ FIELD(81D, 0, EAX, 14, 12, LEAF81D_NUM_SHARING_CACHE, NA, 0, FALSE) \ FIELD(81D, 0, EBX, 0, 12, LEAF81D_CACHE_LINE_SIZE, NA, 0, FALSE) \ FIELD(81D, 0, EBX, 12, 10, LEAF81D_CACHE_PHYS_PARTITIONS, NA, 0, FALSE) \ FIELD(81D, 0, EBX, 22, 10, LEAF81D_CACHE_WAYS, NA, 0, FALSE) \ FIELD(81D, 0, ECX, 0, 32, LEAF81D_CACHE_NUM_SETS, NA, 0, FALSE) \ FLAG( 81D, 0, EDX, 0, 1, LEAF81D_CACHE_WBINVD, NA, 0, FALSE) \ FLAG( 81D, 0, EDX, 1, 1, LEAF81D_CACHE_INCLUSIVE, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_81E \ FIELD(81E, 0, EAX, 0, 32, EXTENDED_APICID, NA, 0, FALSE) \ FIELD(81E, 0, EBX, 0, 8, COMPUTE_UNIT_ID, NA, 0, FALSE) \ FIELD(81E, 0, EBX, 8, 2, CORES_PER_COMPUTE_UNIT, NA, 0, FALSE) \ FIELD(81E, 0, ECX, 0, 8, NODEID_VAL, NA, 0, FALSE) \ FIELD(81E, 0, ECX, 8, 3, NODES_PER_PKG, NA, 0, FALSE) /* LEVEL, SUB-LEVEL, REG, POS, SIZE, NAME, MON SUPP, HWV, CPL3 */ #define CPUID_FIELD_DATA_LEVEL_81F \ FLAG( 81F, 0, EAX, 0, 1, SME, NO, 0, FALSE) \ FLAG( 81F, 0, EAX, 1, 1, SEV, NO, 0, FALSE) \ FLAG( 81F, 0, EAX, 2, 1, PAGE_FLUSH_MSR, NO, 0, FALSE) \ FLAG( 81F, 0, EAX, 3, 1, SEV_ES, NO, 0, FALSE) \ FIELD(81F, 0, EBX, 0, 5, SME_PAGE_TABLE_BIT_NUM, NO, 0, FALSE) \ FIELD(81F, 0, EBX, 6, 6, SME_PHYS_ADDR_SPACE_REDUCTION, NO, 0, FALSE) \ FIELD(81F, 0, ECX, 0, 32, NUM_ENCRYPTED_GUESTS, NO, 0, FALSE) \ FIELD(81F, 0, EDX, 0, 32, SEV_MIN_ASID, NO, 0, FALSE) #define INTEL_CPUID_FIELD_DATA #define AMD_CPUID_FIELD_DATA #define CPUID_FIELD_DATA \ CPUID_FIELD_DATA_LEVEL_0 \ CPUID_FIELD_DATA_LEVEL_1 \ CPUID_FIELD_DATA_LEVEL_2 \ CPUID_FIELD_DATA_LEVEL_4 \ CPUID_FIELD_DATA_LEVEL_5 \ CPUID_FIELD_DATA_LEVEL_6 \ CPUID_FIELD_DATA_LEVEL_7 \ CPUID_FIELD_DATA_LEVEL_A \ CPUID_FIELD_DATA_LEVEL_B \ CPUID_FIELD_DATA_LEVEL_D \ CPUID_FIELD_DATA_LEVEL_F \ CPUID_FIELD_DATA_LEVEL_10 \ CPUID_FIELD_DATA_LEVEL_12 \ CPUID_FIELD_DATA_LEVEL_14 \ CPUID_FIELD_DATA_LEVEL_15 \ CPUID_FIELD_DATA_LEVEL_16 \ CPUID_FIELD_DATA_LEVEL_17 \ CPUID_FIELD_DATA_LEVEL_400 \ CPUID_FIELD_DATA_LEVEL_401 \ CPUID_FIELD_DATA_LEVEL_402 \ CPUID_FIELD_DATA_LEVEL_403 \ CPUID_FIELD_DATA_LEVEL_404 \ CPUID_FIELD_DATA_LEVEL_405 \ CPUID_FIELD_DATA_LEVEL_406 \ CPUID_FIELD_DATA_LEVEL_410 \ CPUID_FIELD_DATA_LEVEL_80 \ CPUID_FIELD_DATA_LEVEL_81 \ CPUID_FIELD_DATA_LEVEL_82 \ CPUID_FIELD_DATA_LEVEL_83 \ CPUID_FIELD_DATA_LEVEL_84 \ CPUID_FIELD_DATA_LEVEL_85 \ CPUID_FIELD_DATA_LEVEL_86 \ CPUID_FIELD_DATA_LEVEL_87 \ CPUID_FIELD_DATA_LEVEL_88 \ CPUID_FIELD_DATA_LEVEL_8A \ CPUID_FIELD_DATA_LEVEL_819 \ CPUID_FIELD_DATA_LEVEL_81A \ CPUID_FIELD_DATA_LEVEL_81B \ CPUID_FIELD_DATA_LEVEL_81C \ CPUID_FIELD_DATA_LEVEL_81D \ CPUID_FIELD_DATA_LEVEL_81E \ CPUID_FIELD_DATA_LEVEL_81F \ INTEL_CPUID_FIELD_DATA \ AMD_CPUID_FIELD_DATA /* * Define all field and flag values as an enum. The result is a full * set of values taken from the table above in the form: * * CPUID_<name>_MASK == mask for feature/field * CPUID_<name>_SHIFT == offset of field * * e.g. - CPUID_VIRT_BITS_MASK = 0xff00 * - CPUID_VIRT_BITS_SHIFT = 8 */ #define VMW_BIT_MASK(shift) (0xffffffffu >> (32 - shift)) #define FIELD(lvl, ecxIn, reg, bitpos, size, name, s, hwv, c3) \ CPUID_##name##_SHIFT = bitpos, \ CPUID_##name##_MASK = VMW_BIT_MASK(size) << bitpos, \ CPUID_INTERNAL_SHIFT_##name = bitpos, \ CPUID_INTERNAL_MASK_##name = VMW_BIT_MASK(size) << bitpos, \ CPUID_INTERNAL_REG_##name = CPUID_REG_##reg, \ CPUID_INTERNAL_EAXIN_##name = CPUID_LEVEL_VAL_##lvl, \ CPUID_INTERNAL_ECXIN_##name = ecxIn, \ CPUID_INTERNAL_HWV_##name = hwv, #define FLAG FIELD enum { /* Define data for every CPUID field we have */ CPUID_FIELD_DATA }; #undef VMW_BIT_MASK #undef FIELD #undef FLAG /* * CPUID_MASK -- * CPUID_SHIFT -- * CPUID_ISSET -- * CPUID_GET -- * CPUID_SET -- * CPUID_CLEAR -- * CPUID_SETTO -- * * Accessor macros for all CPUID consts/fields/flags. Level and reg are not * required, but are used to force compile-time asserts which help verify that * the flag is being used on the right CPUID input and result register. * * Note: ASSERT_ON_COMPILE is duplicated rather than factored into its own * macro, because token concatenation does not work as expected if an input is * #defined (e.g. APIC) when macros are nested. Also, compound statements * within parenthes is a GCC extension, so we must use runtime asserts with * other compilers. */ #if defined(__GNUC__) && !defined(__clang__) #define CPUID_MASK(eaxIn, reg, flag) \ ({ \ ASSERT_ON_COMPILE(eaxIn == CPUID_INTERNAL_EAXIN_##flag && \ CPUID_REG_##reg == (CpuidReg)CPUID_INTERNAL_REG_##flag); \ CPUID_INTERNAL_MASK_##flag; \ }) #define CPUID_SHIFT(eaxIn, reg, flag) \ ({ \ ASSERT_ON_COMPILE(eaxIn == CPUID_INTERNAL_EAXIN_##flag && \ CPUID_REG_##reg == (CpuidReg)CPUID_INTERNAL_REG_##flag); \ CPUID_INTERNAL_SHIFT_##flag; \ }) #define CPUID_ISSET(eaxIn, reg, flag, data) \ ({ \ ASSERT_ON_COMPILE(eaxIn == CPUID_INTERNAL_EAXIN_##flag && \ CPUID_REG_##reg == (CpuidReg)CPUID_INTERNAL_REG_##flag); \ (((data) & CPUID_INTERNAL_MASK_##flag) != 0); \ }) #define CPUID_GET(eaxIn, reg, field, data) \ ({ \ ASSERT_ON_COMPILE(eaxIn == CPUID_INTERNAL_EAXIN_##field && \ CPUID_REG_##reg == (CpuidReg)CPUID_INTERNAL_REG_##field); \ (((uint32)(data) & CPUID_INTERNAL_MASK_##field) >> \ CPUID_INTERNAL_SHIFT_##field); \ }) #else /* * CPUIDCheck -- * * Return val after verifying parameters. */ static INLINE uint32 CPUIDCheck(int32 eaxIn, int32 eaxInCheck, CpuidReg reg, CpuidReg regCheck, uint32 val) { ASSERT(eaxIn == eaxInCheck && reg == regCheck); return val; } #define CPUID_MASK(eaxIn, reg, flag) \ CPUIDCheck(eaxIn, CPUID_INTERNAL_EAXIN_##flag, \ CPUID_REG_##reg, (CpuidReg)CPUID_INTERNAL_REG_##flag, \ CPUID_INTERNAL_MASK_##flag) #define CPUID_SHIFT(eaxIn, reg, flag) \ CPUIDCheck(eaxIn, CPUID_INTERNAL_EAXIN_##flag, \ CPUID_REG_##reg, (CpuidReg)CPUID_INTERNAL_REG_##flag, \ CPUID_INTERNAL_SHIFT_##flag) #define CPUID_ISSET(eaxIn, reg, flag, data) \ (CPUIDCheck(eaxIn, CPUID_INTERNAL_EAXIN_##flag, \ CPUID_REG_##reg, (CpuidReg)CPUID_INTERNAL_REG_##flag, \ CPUID_INTERNAL_MASK_##flag & (data)) != 0) #define CPUID_GET(eaxIn, reg, field, data) \ CPUIDCheck(eaxIn, CPUID_INTERNAL_EAXIN_##field, \ CPUID_REG_##reg, (CpuidReg)CPUID_INTERNAL_REG_##field, \ ((uint32)(data) & CPUID_INTERNAL_MASK_##field) >> \ CPUID_INTERNAL_SHIFT_##field) #endif #define CPUID_SET(eaxIn, reg, flag, dataPtr) \ do { \ ASSERT_ON_COMPILE( \ (uint32)eaxIn == (uint32)CPUID_INTERNAL_EAXIN_##flag && \ CPUID_REG_##reg == (CpuidReg)CPUID_INTERNAL_REG_##flag); \ *(dataPtr) |= CPUID_INTERNAL_MASK_##flag; \ } while (0) #define CPUID_CLEAR(eaxIn, reg, flag, dataPtr) \ do { \ ASSERT_ON_COMPILE( \ (uint32)eaxIn == (uint32)CPUID_INTERNAL_EAXIN_##flag && \ CPUID_REG_##reg == (CpuidReg)CPUID_INTERNAL_REG_##flag); \ *(dataPtr) &= ~CPUID_INTERNAL_MASK_##flag; \ } while (0) #define CPUID_SETTO(eaxIn, reg, field, dataPtr, val) \ do { \ uint32 _v = val; \ uint32 *_d = dataPtr; \ ASSERT_ON_COMPILE( \ (uint32)eaxIn == (uint32)CPUID_INTERNAL_EAXIN_##field && \ CPUID_REG_##reg == (CpuidReg)CPUID_INTERNAL_REG_##field); \ *_d = (*_d & ~CPUID_INTERNAL_MASK_##field) | \ (_v << CPUID_INTERNAL_SHIFT_##field); \ ASSERT(_v == (*_d & CPUID_INTERNAL_MASK_##field) >> \ CPUID_INTERNAL_SHIFT_##field); \ } while (0) #define CPUID_SETTO_SAFE(eaxIn, reg, field, dataPtr, val) \ do { \ uint32 _v = val & \ (CPUID_INTERNAL_MASK_##field >> CPUID_INTERNAL_SHIFT_##field); \ uint32 *_d = dataPtr; \ ASSERT_ON_COMPILE( \ (uint32)eaxIn == (uint32)CPUID_INTERNAL_EAXIN_##field && \ CPUID_REG_##reg == (CpuidReg)CPUID_INTERNAL_REG_##field); \ *_d = (*_d & ~CPUID_INTERNAL_MASK_##field) | \ (_v << CPUID_INTERNAL_SHIFT_##field); \ } while (0) /* * Definitions of various fields' values and more complicated * macros/functions for reading cpuid fields. */ #define CPUID_FAMILY_EXTENDED 15 /* Effective Intel CPU Families */ #define CPUID_FAMILY_486 4 #define CPUID_FAMILY_P5 5 #define CPUID_FAMILY_P6 6 #define CPUID_FAMILY_P4 15 /* Effective AMD CPU Families */ #define CPUID_FAMILY_5x86 0x4 #define CPUID_FAMILY_K5 0x5 #define CPUID_FAMILY_K6 0x5 #define CPUID_FAMILY_K7 0x6 #define CPUID_FAMILY_K8 0xf #define CPUID_FAMILY_K8L 0x10 #define CPUID_FAMILY_K8MOBILE 0x11 #define CPUID_FAMILY_LLANO 0x12 #define CPUID_FAMILY_BOBCAT 0x14 #define CPUID_FAMILY_BULLDOZER 0x15 // BD PD SR EX #define CPUID_FAMILY_KYOTO 0x16 // Note: Jaguar microarch #define CPUID_FAMILY_ZEN 0x17 /* Effective VIA CPU Families */ #define CPUID_FAMILY_C7 6 /* Effective Hygon CPU Families. */ #define CPUID_FAMILY_DHYANA 0x18 /* Intel model information */ #define CPUID_MODEL_PPRO 1 #define CPUID_MODEL_PII_03 3 #define CPUID_MODEL_PII_05 5 #define CPUID_MODEL_CELERON_06 6 #define CPUID_MODEL_PM_09 9 #define CPUID_MODEL_PM_0D 13 #define CPUID_MODEL_PM_0E 14 // Yonah / Sossaman #define CPUID_MODEL_CORE_0F 15 // Conroe / Merom #define CPUID_MODEL_CORE_17 0x17 // Penryn #define CPUID_MODEL_NEHALEM_1A 0x1a // Nehalem / Gainestown #define CPUID_MODEL_ATOM_1C 0x1c // Silverthorne / Diamondville #define CPUID_MODEL_CORE_1D 0x1d // Dunnington #define CPUID_MODEL_NEHALEM_1E 0x1e // Lynnfield #define CPUID_MODEL_NEHALEM_1F 0x1f // Havendale #define CPUID_MODEL_NEHALEM_25 0x25 // Westmere / Clarkdale #define CPUID_MODEL_ATOM_26 0x26 // Lincroft #define CPUID_MODEL_ATOM_27 0x27 // Saltwell #define CPUID_MODEL_SANDYBRIDGE_2A 0x2a // Sandybridge (desktop/mobile) #define CPUID_MODEL_NEHALEM_2C 0x2c // Westmere-EP #define CPUID_MODEL_SANDYBRIDGE_2D 0x2d // Sandybridge-EP #define CPUID_MODEL_NEHALEM_2E 0x2e // Nehalem-EX #define CPUID_MODEL_NEHALEM_2F 0x2f // Westmere-EX #define CPUID_MODEL_ATOM_35 0x35 // Cloverview #define CPUID_MODEL_ATOM_36 0x36 // Cedarview #define CPUID_MODEL_ATOM_37 0x37 // Bay Trail #define CPUID_MODEL_SANDYBRIDGE_3A 0x3a // Ivy Bridge #define CPUID_MODEL_HASWELL_3C 0x3c // Haswell DT #define CPUID_MODEL_BROADWELL_3D 0x3d // Broadwell-Ult #define CPUID_MODEL_SANDYBRIDGE_3E 0x3e // Ivy Bridge-EP #define CPUID_MODEL_HASWELL_3F 0x3f // Haswell EP/EN/EX #define CPUID_MODEL_HASWELL_45 0x45 // Haswell Ultrathin #define CPUID_MODEL_HASWELL_46 0x46 // Haswell (Crystal Well) #define CPUID_MODEL_BROADWELL_47 0x47 // Broadwell (Denlow) #define CPUID_MODEL_ATOM_4A 0x4a // Future Silvermont #define CPUID_MODEL_ATOM_4C 0x4c // Airmont #define CPUID_MODEL_ATOM_4D 0x4d // Avoton #define CPUID_MODEL_SKYLAKE_4E 0x4e // Skylake-Y / Kabylake U/Y ES #define CPUID_MODEL_BROADWELL_4F 0x4f // Broadwell EP/EN/EX #define CPUID_MODEL_SKYLAKE_55 0x55 // Skylake EP/EN/EX #define CPUID_MODEL_BROADWELL_56 0x56 // Broadwell DE #define CPUID_MODEL_KNL_57 0x57 // Knights Landing #define CPUID_MODEL_ATOM_5A 0x5a // Future Silvermont #define CPUID_MODEL_ATOM_5D 0x5d // Future Silvermont #define CPUID_MODEL_SKYLAKE_5E 0x5e // Skylake-S / Kabylake S/H ES #define CPUID_MODEL_ATOM_5F 0x5f // Denverton #define CPUID_MODEL_KNM_85 0x85 // Knights Mill #define CPUID_MODEL_KABYLAKE_8E 0x8e // Kabylake U/Y QS #define CPUID_MODEL_KABYLAKE_9E 0x9e // Kabylake S/H QS /* Intel stepping information */ #define CPUID_STEPPING_KABYLAKE_ES 0x8 // Kabylake S/H/U/Y ES #define CPUID_STEPPING_COFFEELAKE_A 0xA // Coffeelake U/S/H #define CPUID_STEPPING_COFFEELAKE_B 0xB // Coffeelake S/H #define CPUID_MODEL_PIII_07 7 #define CPUID_MODEL_PIII_08 8 #define CPUID_MODEL_PIII_0A 10 /* AMD model information */ #define CPUID_MODEL_BARCELONA_02 0x02 // Barcelona (Opteron & Phenom) #define CPUID_MODEL_SHANGHAI_04 0x04 // Shanghai RB #define CPUID_MODEL_SHANGHAI_05 0x05 // Shanghai BL #define CPUID_MODEL_SHANGHAI_06 0x06 // Shanghai DA #define CPUID_MODEL_ISTANBUL_MAGNY_08 0x08 // Istanbul (6 core) & Magny-cours (12) HY #define CPUID_MODEL_ISTANBUL_MAGNY_09 0x09 // HY - G34 package #define CPUID_MODEL_PHAROAH_HOUND_0A 0x0A // Pharoah Hound #define CPUID_MODEL_PILEDRIVER_1F 0x1F // Max piledriver model defined in BKDG #define CPUID_MODEL_PILEDRIVER_10 0x10 // family == CPUID_FAMILY_BULLDOZER #define CPUID_MODEL_PILEDRIVER_02 0x02 // family == CPUID_FAMILY_BULLDOZER #define CPUID_MODEL_OPTERON_REVF_41 0x41 // family == CPUID_FAMILY_K8 #define CPUID_MODEL_KYOTO_00 0x00 // family == CPUID_FAMILY_KYOTO #define CPUID_MODEL_STEAMROLLER_3F 0x3F // Max Steamroller model defined in BKDG #define CPUID_MODEL_STEAMROLLER_30 0x30 // family == CPUID_FAMILY_BULLDOZER #define CPUID_MODEL_EXCAVATOR_60 0x60 // family == CPUID_FAMILY_BULLDOZER #define CPUID_MODEL_EXCAVATOR_6F 0x6F // Max Excavator model defined in BKDG #define CPUID_MODEL_ZEN_00 0x00 // family == CPUID_FAMILY_ZEN #define CPUID_MODEL_ZEN_1F 0x1F // Max Zen model defined in BKDG /* VIA model information */ #define CPUID_MODEL_NANO 15 // Isaiah /* Hygon model information. */ #define CPUID_MODEL_DHYANA_A 0 // Dhyana A /* *---------------------------------------------------------------------- * * CPUID_IsVendor{AMD,Intel,VIA,Hygon} -- * * Determines if the vendor string in cpuid id0 is from * {AMD,Intel,VIA,Hygon}. * * Results: * True iff vendor string is CPUID_{AMD,INTEL,VIA,HYGON}_VENDOR_STRING * * Side effects: * None. * *---------------------------------------------------------------------- */ static INLINE Bool CPUID_IsRawVendor(CPUIDRegs *id0, const char* vendor) { // hard to get strcmp() in some environments, so do it in the raw return (id0->ebx == *(const uint32 *) (vendor + 0) && id0->ecx == *(const uint32 *) (vendor + 4) && id0->edx == *(const uint32 *) (vendor + 8)); } static INLINE Bool CPUID_IsVendorAMD(CPUIDRegs *id0) { return CPUID_IsRawVendor(id0, CPUID_AMD_VENDOR_STRING); } static INLINE Bool CPUID_IsVendorIntel(CPUIDRegs *id0) { return CPUID_IsRawVendor(id0, CPUID_INTEL_VENDOR_STRING); } static INLINE Bool CPUID_IsVendorVIA(CPUIDRegs *id0) { return CPUID_IsRawVendor(id0, CPUID_VIA_VENDOR_STRING); } static INLINE Bool CPUID_IsVendorHygon(CPUIDRegs *id0) { return CPUID_IsRawVendor(id0, CPUID_HYGON_VENDOR_STRING); } static INLINE uint32 CPUID_EFFECTIVE_FAMILY(uint32 v) /* %eax from CPUID with %eax=1. */ { uint32 f = CPUID_GET(1, EAX, FAMILY, v); return f != CPUID_FAMILY_EXTENDED ? f : f + CPUID_GET(1, EAX, EXTENDED_FAMILY, v); } /* Normally only used when FAMILY==CPUID_FAMILY_EXTENDED, but Intel is * now using the extended model field for FAMILY==CPUID_FAMILY_P6 to * refer to the newer Core2 CPUs */ static INLINE uint32 CPUID_EFFECTIVE_MODEL(uint32 v) /* %eax from CPUID with %eax=1. */ { uint32 m = CPUID_GET(1, EAX, MODEL, v); uint32 em = CPUID_GET(1, EAX, EXTENDED_MODEL, v); return m + (em << 4); } static INLINE uint32 CPUID_EFFECTIVE_STEPPING(uint32 v) /* %eax from CPUID with %eax=1. */ { return CPUID_GET(1, EAX, STEPPING, v); } /* * Notice that CPUID families for Intel and AMD overlap. The following macros * should only be used AFTER the manufacturer has been established (through * the use of CPUID standard function 0). */ static INLINE Bool CPUID_FAMILY_IS_486(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_486; } static INLINE Bool CPUID_FAMILY_IS_P5(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_P5; } static INLINE Bool CPUID_FAMILY_IS_P6(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_P6; } static INLINE Bool CPUID_FAMILY_IS_PENTIUM4(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_P4; } /* * Intel Pentium M processors are Yonah/Sossaman or an older P-M */ static INLINE Bool CPUID_UARCH_IS_PENTIUM_M(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && (CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_PM_09 || CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_PM_0D || CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_PM_0E); } /* * Intel Core processors are Merom, Conroe, Woodcrest, Clovertown, * Penryn, Dunnington, Kentsfield, Yorktown, Harpertown, ........ */ static INLINE Bool CPUID_UARCH_IS_CORE(uint32 v) // IN: %eax from CPUID with %eax=1. { uint32 model = CPUID_EFFECTIVE_MODEL(v); /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && model >= CPUID_MODEL_CORE_0F && (model < CPUID_MODEL_NEHALEM_1A || model == CPUID_MODEL_CORE_1D); } /* * Intel Nehalem processors are: Nehalem, Gainestown, Lynnfield, Clarkdale. */ static INLINE Bool CPUID_UARCH_IS_NEHALEM(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ uint32 effectiveModel = CPUID_EFFECTIVE_MODEL(v); return CPUID_FAMILY_IS_P6(v) && (effectiveModel == CPUID_MODEL_NEHALEM_1A || effectiveModel == CPUID_MODEL_NEHALEM_1E || effectiveModel == CPUID_MODEL_NEHALEM_1F || effectiveModel == CPUID_MODEL_NEHALEM_25 || effectiveModel == CPUID_MODEL_NEHALEM_2C || effectiveModel == CPUID_MODEL_NEHALEM_2E || effectiveModel == CPUID_MODEL_NEHALEM_2F); } static INLINE Bool CPUID_UARCH_IS_SANDYBRIDGE(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ uint32 effectiveModel = CPUID_EFFECTIVE_MODEL(v); return CPUID_FAMILY_IS_P6(v) && (effectiveModel == CPUID_MODEL_SANDYBRIDGE_2A || effectiveModel == CPUID_MODEL_SANDYBRIDGE_2D || effectiveModel == CPUID_MODEL_SANDYBRIDGE_3E || effectiveModel == CPUID_MODEL_SANDYBRIDGE_3A); } static INLINE Bool CPUID_MODEL_IS_BROADWELL(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ uint32 effectiveModel = CPUID_EFFECTIVE_MODEL(v); return CPUID_FAMILY_IS_P6(v) && (effectiveModel == CPUID_MODEL_BROADWELL_3D || effectiveModel == CPUID_MODEL_BROADWELL_47 || effectiveModel == CPUID_MODEL_BROADWELL_4F || effectiveModel == CPUID_MODEL_BROADWELL_56); } static INLINE Bool CPUID_MODEL_IS_HASWELL(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ uint32 effectiveModel = CPUID_EFFECTIVE_MODEL(v); return CPUID_FAMILY_IS_P6(v) && (effectiveModel == CPUID_MODEL_HASWELL_3C || effectiveModel == CPUID_MODEL_HASWELL_3F || effectiveModel == CPUID_MODEL_HASWELL_45 || effectiveModel == CPUID_MODEL_HASWELL_46); } static INLINE Bool CPUID_MODEL_IS_SKYLAKE(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && ((CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_SKYLAKE_5E && CPUID_EFFECTIVE_STEPPING(v) != CPUID_STEPPING_KABYLAKE_ES) || CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_SKYLAKE_55 || (CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_SKYLAKE_4E && CPUID_EFFECTIVE_STEPPING(v) != CPUID_STEPPING_KABYLAKE_ES)); } static INLINE Bool CPUID_MODEL_IS_COFFEELAKE(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && ((CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_KABYLAKE_9E && (CPUID_EFFECTIVE_STEPPING(v) == CPUID_STEPPING_COFFEELAKE_A || CPUID_EFFECTIVE_STEPPING(v) == CPUID_STEPPING_COFFEELAKE_B)) || (CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_KABYLAKE_8E && CPUID_EFFECTIVE_STEPPING(v) == CPUID_STEPPING_COFFEELAKE_A)); } static INLINE Bool CPUID_MODEL_IS_KABYLAKE(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && !CPUID_MODEL_IS_COFFEELAKE(v) && (CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_KABYLAKE_9E || CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_KABYLAKE_8E || (CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_SKYLAKE_5E && CPUID_EFFECTIVE_STEPPING(v) == CPUID_STEPPING_KABYLAKE_ES) || (CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_SKYLAKE_4E && CPUID_EFFECTIVE_STEPPING(v) == CPUID_STEPPING_KABYLAKE_ES)); } static INLINE Bool CPUID_UARCH_IS_SKYLAKE(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && (CPUID_MODEL_IS_COFFEELAKE(v) || CPUID_MODEL_IS_KABYLAKE(v) || CPUID_MODEL_IS_SKYLAKE(v)); } static INLINE Bool CPUID_UARCH_IS_HASWELL(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && (CPUID_MODEL_IS_BROADWELL(v) || CPUID_MODEL_IS_HASWELL(v)); } static INLINE Bool CPUID_MODEL_IS_CENTERTON(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_ATOM_1C; } static INLINE Bool CPUID_MODEL_IS_AVOTON(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_ATOM_4D; } static INLINE Bool CPUID_MODEL_IS_BAYTRAIL(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_ATOM_37; } static INLINE Bool CPUID_UARCH_IS_SILVERMONT(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && (CPUID_MODEL_IS_AVOTON(v) || CPUID_MODEL_IS_BAYTRAIL(v)); } static INLINE Bool CPUID_MODEL_IS_DENVERTON(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_ATOM_5F; } static INLINE Bool CPUID_MODEL_IS_WESTMERE(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ uint32 effectiveModel = CPUID_EFFECTIVE_MODEL(v); return CPUID_FAMILY_IS_P6(v) && (effectiveModel == CPUID_MODEL_NEHALEM_25 || // Clarkdale effectiveModel == CPUID_MODEL_NEHALEM_2C || // Westmere-EP effectiveModel == CPUID_MODEL_NEHALEM_2F); // Westmere-EX } static INLINE Bool CPUID_MODEL_IS_SANDYBRIDGE(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ uint32 effectiveModel = CPUID_EFFECTIVE_MODEL(v); return CPUID_FAMILY_IS_P6(v) && (effectiveModel == CPUID_MODEL_SANDYBRIDGE_2A || effectiveModel == CPUID_MODEL_SANDYBRIDGE_2D); } static INLINE Bool CPUID_MODEL_IS_IVYBRIDGE(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ uint32 effectiveModel = CPUID_EFFECTIVE_MODEL(v); return CPUID_FAMILY_IS_P6(v) && ( effectiveModel == CPUID_MODEL_SANDYBRIDGE_3E || effectiveModel == CPUID_MODEL_SANDYBRIDGE_3A); } static INLINE Bool CPUID_MODEL_IS_KNIGHTS_LANDING(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_KNL_57; } static INLINE Bool CPUID_MODEL_IS_KNIGHTS_MILL(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is Intel. */ return CPUID_FAMILY_IS_P6(v) && CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_KNM_85; } static INLINE Bool CPUID_FAMILY_IS_K7(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_K7; } static INLINE Bool CPUID_FAMILY_IS_K8(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_K8; } /* *---------------------------------------------------------------------- * * CPUID_FAMILY_IS_K8EXT -- * * Return TRUE for family K8 with effective model >= 0x10. * *---------------------------------------------------------------------- */ static INLINE Bool CPUID_FAMILY_IS_K8EXT(uint32 eax) { return CPUID_FAMILY_IS_K8(eax) && CPUID_GET(1, EAX, EXTENDED_MODEL, eax) != 0; } static INLINE Bool CPUID_FAMILY_IS_K8L(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_K8L || CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_LLANO; } static INLINE Bool CPUID_FAMILY_IS_LLANO(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_LLANO; } static INLINE Bool CPUID_FAMILY_IS_K8MOBILE(uint32 eax) { /* Essentially a K8 (not K8L) part, but with mobile features. */ return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_K8MOBILE; } static INLINE Bool CPUID_FAMILY_IS_K8STAR(uint32 eax) { /* * Read function name as "K8*", as in wildcard. * Matches K8 or K8L or K8MOBILE */ return CPUID_FAMILY_IS_K8(eax) || CPUID_FAMILY_IS_K8L(eax) || CPUID_FAMILY_IS_K8MOBILE(eax); } static INLINE Bool CPUID_FAMILY_IS_BOBCAT(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_BOBCAT; } static INLINE Bool CPUID_FAMILY_IS_BULLDOZER(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_BULLDOZER; } static INLINE Bool CPUID_FAMILY_IS_KYOTO(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_KYOTO; } static INLINE Bool CPUID_FAMILY_IS_ZEN(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_ZEN; } /* * AMD Barcelona (of either Opteron or Phenom kind). */ static INLINE Bool CPUID_MODEL_IS_BARCELONA(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is AMD. */ return CPUID_EFFECTIVE_FAMILY(v) == CPUID_FAMILY_K8L && CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_BARCELONA_02; } static INLINE Bool CPUID_MODEL_IS_SHANGHAI(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is AMD. */ return CPUID_EFFECTIVE_FAMILY(v) == CPUID_FAMILY_K8L && (CPUID_MODEL_SHANGHAI_04 <= CPUID_EFFECTIVE_MODEL(v) && CPUID_EFFECTIVE_MODEL(v) <= CPUID_MODEL_SHANGHAI_06); } static INLINE Bool CPUID_MODEL_IS_ISTANBUL_MAGNY(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is AMD. */ return CPUID_EFFECTIVE_FAMILY(v) == CPUID_FAMILY_K8L && (CPUID_MODEL_ISTANBUL_MAGNY_08 <= CPUID_EFFECTIVE_MODEL(v) && CPUID_EFFECTIVE_MODEL(v) <= CPUID_MODEL_ISTANBUL_MAGNY_09); } static INLINE Bool CPUID_MODEL_IS_PHAROAH_HOUND(uint32 v) // IN: %eax from CPUID with %eax=1. { /* Assumes the CPU manufacturer is AMD. */ return CPUID_EFFECTIVE_FAMILY(v) == CPUID_FAMILY_K8L && CPUID_EFFECTIVE_MODEL(v) == CPUID_MODEL_PHAROAH_HOUND_0A; } static INLINE Bool CPUID_MODEL_IS_BULLDOZER(uint32 eax) { /* * Bulldozer is models of family 0x15 that are below 10 excluding * Piledriver 02. */ return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_BULLDOZER && CPUID_EFFECTIVE_MODEL(eax) < CPUID_MODEL_PILEDRIVER_10 && CPUID_EFFECTIVE_MODEL(eax) != CPUID_MODEL_PILEDRIVER_02; } static INLINE Bool CPUID_MODEL_IS_PILEDRIVER(uint32 eax) { /* Piledriver is models 0x02 & 0x10 of family 0x15 (so far). */ return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_BULLDOZER && ((CPUID_EFFECTIVE_MODEL(eax) >= CPUID_MODEL_PILEDRIVER_10 && CPUID_EFFECTIVE_MODEL(eax) <= CPUID_MODEL_PILEDRIVER_1F) || CPUID_EFFECTIVE_MODEL(eax) == CPUID_MODEL_PILEDRIVER_02); } static INLINE Bool CPUID_MODEL_IS_STEAMROLLER(uint32 eax) { /* Steamroller is model 0x30 of family 0x15 (so far). */ return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_BULLDOZER && (CPUID_EFFECTIVE_MODEL(eax) >= CPUID_MODEL_STEAMROLLER_30 && CPUID_EFFECTIVE_MODEL(eax) <= CPUID_MODEL_STEAMROLLER_3F); } static INLINE Bool CPUID_MODEL_IS_EXCAVATOR(uint32 eax) { /* Excavator is model 0x60 of family 0x15 (so far). */ return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_BULLDOZER && (CPUID_EFFECTIVE_MODEL(eax) >= CPUID_MODEL_EXCAVATOR_60 && CPUID_EFFECTIVE_MODEL(eax) <= CPUID_MODEL_EXCAVATOR_6F); } static INLINE Bool CPUID_MODEL_IS_KYOTO(uint32 eax) { /* Kyoto is models 0x00 of family 0x16 (so far). */ return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_KYOTO && CPUID_EFFECTIVE_MODEL(eax) == CPUID_MODEL_KYOTO_00; } static INLINE Bool CPUID_MODEL_IS_ZEN(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_ZEN && CPUID_EFFECTIVE_MODEL(eax) <= CPUID_MODEL_ZEN_1F; } static INLINE Bool CPUID_FAMILY_IS_DHYANA(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_DHYANA; } static INLINE Bool CPUID_MODEL_IS_DHYANA_A(uint32 eax) { return CPUID_EFFECTIVE_FAMILY(eax) == CPUID_FAMILY_DHYANA && CPUID_EFFECTIVE_MODEL(eax) == CPUID_MODEL_DHYANA_A; } #define CPUID_TYPE_PRIMARY 0 #define CPUID_TYPE_OVERDRIVE 1 #define CPUID_TYPE_SECONDARY 2 #define CPUID_INTEL_ID4EAX_LEAF4_CACHE_TYPE_NULL 0 #define CPUID_INTEL_ID4EAX_LEAF4_CACHE_TYPE_DATA 1 #define CPUID_INTEL_ID4EAX_LEAF4_CACHE_TYPE_INST 2 #define CPUID_INTEL_ID4EAX_LEAF4_CACHE_TYPE_UNIF 3 #define CPUID_LEAF4_CACHE_TYPE_NULL 0 #define CPUID_LEAF4_CACHE_TYPE_DATA 1 #define CPUID_LEAF4_CACHE_TYPE_INST 2 #define CPUID_LEAF4_CACHE_TYPE_UNIF 3 #define CPUID_LEAF4_CACHE_INDEXING_DIRECT 0 #define CPUID_LEAF4_CACHE_INDEXING_COMPLEX 1 #define CPUID_INTEL_ID4EAX_LEAF4_CACHE_SELF_INIT 0x00000100 #define CPUID_INTEL_ID4EAX_LEAF4_CACHE_FULLY_ASSOC 0x00000200 #define CPUID_LEAF4_CACHE_SELF_INIT 0x00000100 #define CPUID_LEAF4_CACHE_FULLY_ASSOC 0x00000200 #define CPUID_INTEL_IDBECX_LEVEL_TYPE_INVALID 0 #define CPUID_INTEL_IDBECX_LEVEL_TYPE_SMT 1 #define CPUID_INTEL_IDBECX_LEVEL_TYPE_CORE 2 #define CPUID_TOPOLOGY_LEVEL_TYPE_INVALID 0 #define CPUID_TOPOLOGY_LEVEL_TYPE_SMT 1 #define CPUID_TOPOLOGY_LEVEL_TYPE_CORE 2 /* * For certain AMD processors, an lfence instruction is necessary at various * places to ensure ordering. */ static INLINE Bool CPUID_VendorRequiresFence(CpuidVendor vendor) { return vendor == CPUID_VENDOR_AMD; } static INLINE Bool CPUID_VersionRequiresFence(uint32 version) { return CPUID_EFFECTIVE_FAMILY(version) == CPUID_FAMILY_K8 && CPUID_EFFECTIVE_MODEL(version) < 0x40; } static INLINE Bool CPUID_ID0RequiresFence(CPUIDRegs *id0) { if (id0->eax == 0) { return FALSE; } return CPUID_IsVendorAMD(id0); } static INLINE Bool CPUID_ID1RequiresFence(CPUIDRegs *id1) { return CPUID_VersionRequiresFence(id1->eax); } static INLINE Bool CPUID_RequiresFence(CpuidVendor vendor, // IN uint32 version) // IN: %eax from CPUID with %eax=1. { return CPUID_VendorRequiresFence(vendor) && CPUID_VersionRequiresFence(version); } /* * The following low-level functions compute the number of * cores per cpu. They should be used cautiously because * they do not necessarily work on all types of CPUs. * High-level functions that are correct for all CPUs are * available elsewhere: see lib/cpuidInfo/cpuidInfo.c. */ static INLINE uint32 CPUID_IntelCoresPerPackage(uint32 v) /* %eax from CPUID with %eax=4 and %ecx=0. */ { // Note: This is not guaranteed to work on older Intel CPUs. return 1 + CPUID_GET(4, EAX, LEAF4_CORE_COUNT, v); } static INLINE uint32 CPUID_AMDCoresPerPackage(uint32 v) /* %ecx from CPUID with %eax=0x80000008. */ { // Note: This is not guaranteed to work on older AMD CPUs. return 1 + CPUID_GET(0x80000008, ECX, LEAF88_CORE_COUNT, v); } /* * Hypervisor CPUID space is 0x400000XX. */ static INLINE Bool CPUID_IsHypervisorLevel(uint32 level) { return (level & 0xffffff00) == 0x40000000; } /* *---------------------------------------------------------------------- * * CPUID_LevelUsesEcx -- * * Returns TRUE for leaves that support input ECX != 0 (subleaves). * *---------------------------------------------------------------------- */ static INLINE Bool CPUID_LevelUsesEcx(uint32 level) { switch (level) { #define CPUIDLEVEL(t, s, v, c, h) \ case v: \ return c != 0; CPUID_ALL_LEVELS #undef CPUIDLEVEL default: return FALSE; } } /* *---------------------------------------------------------------------- * * CPUID_IsValid*Subleaf -- * * Functions to determine the last subleaf for the level specified * *---------------------------------------------------------------------- */ static INLINE Bool CPUID_IsValidBSubleaf(uint32 ebx) // IN: %ebx = cpuid.b.sublevel.ebx { return ebx != 0; } static INLINE Bool CPUID_IsValid4Subleaf(uint32 eax) // IN: %eax = cpuid.4.sublevel.eax { return eax != 0; } static INLINE Bool CPUID_IsValid7Subleaf(uint32 eax, uint32 subleaf) // IN: %eax = cpuid.7.0.eax { /* * cpuid.7.0.eax is the max ecx (subleaf) index */ return subleaf <= eax; } /* *---------------------------------------------------------------------- * * CPUID_IsValidDSubleaf -- * * It is the caller's repsonsibility to determine if the processor * supports XSAVE and therefore has D sub-leaves. * *---------------------------------------------------------------------- */ static INLINE Bool CPUID_IsValidDSubleaf(uint32 subleaf) // IN: subleaf to check { return subleaf <= 63; } /* *---------------------------------------------------------------------- * * CPUID_SupportsMsrPlatformInfo -- * * Uses vendor and cpuid.1.0.eax to determine if the processor * supports MSR_PLATFORM_INFO. * *---------------------------------------------------------------------- */ static INLINE Bool CPUID_SupportsMsrPlatformInfo(CpuidVendor vendor, uint32 version) { return vendor == CPUID_VENDOR_INTEL && (CPUID_UARCH_IS_NEHALEM(version) || CPUID_UARCH_IS_SANDYBRIDGE(version) || CPUID_UARCH_IS_HASWELL(version) || CPUID_UARCH_IS_SKYLAKE(version) || CPUID_MODEL_IS_KNIGHTS_LANDING(version) || CPUID_MODEL_IS_DENVERTON(version) || CPUID_UARCH_IS_SILVERMONT(version)); } #ifdef _MSC_VER #pragma warning (pop) #endif #if defined __cplusplus } // extern "C" #endif #endif // _X86CPUID_H_ vsock-only/shared/vm_basic_asm.h 0000444 0000000 0000000 00000071321 13432725350 015730 0 ustar root root /********************************************************* * Copyright (C) 2003-2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vm_basic_asm.h * * Basic asm macros */ #ifndef _VM_BASIC_ASM_H_ #define _VM_BASIC_ASM_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" #include "vm_basic_types.h" #if defined VM_X86_64 #include "vm_basic_asm_x86_common.h" #include "vm_basic_asm_x86_64.h" #elif defined VM_X86_32 #include "vm_basic_asm_x86_common.h" #include "vm_basic_asm_x86.h" #elif defined VM_ARM_32 #include "vm_basic_asm_arm32.h" #define MUL64_NO_ASM 1 #include "mul64.h" #elif defined VM_ARM_64 #include "arm64_basic_defs.h" #include "vm_basic_asm_arm64.h" #else #define MUL64_NO_ASM 1 #include "mul64.h" #endif #if defined __cplusplus extern "C" { #endif /* * Locate most and least significant bit set functions. Use our own name * space to avoid namespace collisions. The new names follow a pattern, * <prefix><size><option>, where: * * <prefix> is [lm]ssb (least/most significant bit set) * <size> is size of the argument: 32 (32-bit), 64 (64-bit) or Ptr (pointer) * <option> is for alternative versions of the functions * * NAME FUNCTION BITS FUNC(0) *----- -------- ---- ------- * lssb32_0 LSB set (uint32) 0..31 -1 * mssb32_0 MSB set (uint32) 0..31 -1 * lssb64_0 LSB set (uint64) 0..63 -1 * mssb64_0 MSB set (uint64) 0..63 -1 * lssbPtr_0 LSB set (uintptr_t;32-bit) 0..31 -1 * lssbPtr_0 LSB set (uintptr_t;64-bit) 0..63 -1 * mssbPtr_0 MSB set (uintptr_t;32-bit) 0..31 -1 * mssbPtr_0 MSB set (uintptr_t;64-bit) 0..63 -1 * lssbPtr LSB set (uintptr_t;32-bit) 1..32 0 * lssbPtr LSB set (uintptr_t;64-bit) 1..64 0 * mssbPtr MSB set (uintptr_t;32-bit) 1..32 0 * mssbPtr MSB set (uintptr_t;64-bit) 1..64 0 * lssb32 LSB set (uint32) 1..32 0 * mssb32 MSB set (uint32) 1..32 0 * lssb64 LSB set (uint64) 1..64 0 * mssb64 MSB set (uint64) 1..64 0 */ #ifdef _MSC_VER static INLINE int lssb32_0(const uint32 value) { unsigned long idx; unsigned char ret; if (UNLIKELY(value == 0)) { return -1; } ret = _BitScanForward(&idx, (unsigned long)value); #ifdef __analysis_assume __analysis_assume(ret != 0); #endif #pragma warning(suppress: 6001 6102) // Suppress bogus complaint that idx may be uninitialized in error case return idx; } static INLINE int mssb32_0(const uint32 value) { unsigned long idx; unsigned char ret; if (UNLIKELY(value == 0)) { return -1; } ret = _BitScanReverse(&idx, (unsigned long)value); #ifdef __analysis_assume __analysis_assume(ret != 0); #endif #pragma warning(suppress: 6001 6102) // Suppress bogus complaint that idx may be uninitialized in error case return idx; } static INLINE int lssb64_0(const uint64 value) { if (UNLIKELY(value == 0)) { return -1; } else { #ifdef VM_X86_64 unsigned long idx; unsigned char ret; ret = _BitScanForward64(&idx, (unsigned __int64)value); #ifdef __analysis_assume __analysis_assume(ret != 0); #endif #pragma warning(suppress: 6001 6102) // Suppress bogus complaint that idx may be uninitialized in error case return idx; #else /* The coding was chosen to minimize conditionals and operations */ int lowFirstBit = lssb32_0((uint32) value); if (lowFirstBit == -1) { lowFirstBit = lssb32_0((uint32) (value >> 32)); if (lowFirstBit != -1) { return lowFirstBit + 32; } } return lowFirstBit; #endif } } static INLINE int mssb64_0(const uint64 value) { if (UNLIKELY(value == 0)) { return -1; } else { #ifdef VM_X86_64 unsigned long idx; unsigned char ret; ret = _BitScanReverse64(&idx, (unsigned __int64)value); #ifdef __analysis_assume __analysis_assume(ret != 0); #endif #pragma warning(suppress: 6001 6102) // Suppress bogus complaint that idx may be uninitialized in error case return idx; #else /* The coding was chosen to minimize conditionals and operations */ if (value > 0xFFFFFFFFULL) { return 32 + mssb32_0((uint32) (value >> 32)); } return mssb32_0((uint32) value); #endif } } #endif #ifdef __GNUC__ #ifdef VM_X86_ANY #define USE_ARCH_X86_CUSTOM #endif /* ********************************************************** * GCC's intrinsics for the lssb and mssb family produce sub-optimal code, * so we use inline assembly to improve matters. However, GCC cannot * propagate constants through inline assembly, so we help GCC out by * allowing it to use its intrinsics for compile-time constant values. * Some day, GCC will make better code and these can collapse to intrinsics. * * For example, in Decoder_AddressSize, inlined into VVT_GetVTInstrInfo: * __builtin_ffs(a) compiles to: * mov $0xffffffff, %esi * bsf %eax, %eax * cmovz %esi, %eax * sub $0x1, %eax * and $0x7, %eax * * While the code below compiles to: * bsf %eax, %eax * sub $0x1, %eax * * Ideally, GCC should have recognized non-zero input in the first case. * Other instances of the intrinsic produce code like * sub $1, %eax; add $1, %eax; clts * ********************************************************** */ #if __GNUC__ < 4 #define FEWER_BUILTINS #endif static INLINE int lssb32_0(uint32 value) { #ifdef USE_ARCH_X86_CUSTOM if (!__builtin_constant_p(value)) { if (UNLIKELY(value == 0)) { return -1; } else { int pos; __asm__ ("bsfl %1, %0\n" : "=r" (pos) : "rm" (value) : "cc"); return pos; } } #endif return __builtin_ffs(value) - 1; } #ifndef FEWER_BUILTINS static INLINE int mssb32_0(uint32 value) { /* * We must keep the UNLIKELY(...) outside the #if defined ... * because __builtin_clz(0) is undefined according to gcc's * documentation. */ if (UNLIKELY(value == 0)) { return -1; } else { int pos; #ifdef USE_ARCH_X86_CUSTOM if (!__builtin_constant_p(value)) { __asm__ ("bsrl %1, %0\n" : "=r" (pos) : "rm" (value) : "cc"); return pos; } #endif pos = 32 - __builtin_clz(value) - 1; return pos; } } static INLINE int lssb64_0(const uint64 value) { #ifdef USE_ARCH_X86_CUSTOM if (!__builtin_constant_p(value)) { if (UNLIKELY(value == 0)) { return -1; } else { intptr_t pos; #ifdef VM_X86_64 __asm__ ("bsf %1, %0\n" : "=r" (pos) : "rm" (value) : "cc"); #else /* The coding was chosen to minimize conditionals and operations */ pos = lssb32_0((uint32) value); if (pos == -1) { pos = lssb32_0((uint32) (value >> 32)); if (pos != -1) { return pos + 32; } } #endif return pos; } } #endif return __builtin_ffsll(value) - 1; } #endif /* !FEWER_BUILTINS */ #ifdef FEWER_BUILTINS /* GCC 3.3.x does not like __bulitin_clz or __builtin_ffsll. */ static INLINE int mssb32_0(uint32 value) { if (UNLIKELY(value == 0)) { return -1; } else { int pos; __asm__ __volatile__("bsrl %1, %0\n" : "=r" (pos) : "rm" (value) : "cc"); return pos; } } static INLINE int lssb64_0(const uint64 value) { if (UNLIKELY(value == 0)) { return -1; } else { intptr_t pos; #ifdef VM_X86_64 __asm__ __volatile__("bsf %1, %0\n" : "=r" (pos) : "rm" (value) : "cc"); #else /* The coding was chosen to minimize conditionals and operations */ pos = lssb32_0((uint32) value); if (pos == -1) { pos = lssb32_0((uint32) (value >> 32)); if (pos != -1) { return pos + 32; } } #endif /* VM_X86_64 */ return pos; } } #endif /* FEWER_BUILTINS */ static INLINE int mssb64_0(const uint64 value) { if (UNLIKELY(value == 0)) { return -1; } else { intptr_t pos; #ifdef USE_ARCH_X86_CUSTOM #ifdef VM_X86_64 __asm__ ("bsr %1, %0\n" : "=r" (pos) : "rm" (value) : "cc"); #else /* The coding was chosen to minimize conditionals and operations */ if (value > 0xFFFFFFFFULL) { pos = 32 + mssb32_0((uint32) (value >> 32)); } else { pos = mssb32_0((uint32) value); } #endif #else pos = 64 - __builtin_clzll(value) - 1; #endif return pos; } } #ifdef USE_ARCH_X86_CUSTOM #undef USE_ARCH_X86_CUSTOM #endif #endif // __GNUC__ static INLINE int lssbPtr_0(const uintptr_t value) { #ifdef VM_64BIT return lssb64_0((uint64) value); #else return lssb32_0((uint32) value); #endif } static INLINE int lssbPtr(const uintptr_t value) { return lssbPtr_0(value) + 1; } static INLINE int mssbPtr_0(const uintptr_t value) { #ifdef VM_64BIT return mssb64_0((uint64) value); #else return mssb32_0((uint32) value); #endif } static INLINE int mssbPtr(const uintptr_t value) { return mssbPtr_0(value) + 1; } static INLINE int lssb32(const uint32 value) { return lssb32_0(value) + 1; } static INLINE int mssb32(const uint32 value) { return mssb32_0(value) + 1; } static INLINE int lssb64(const uint64 value) { return lssb64_0(value) + 1; } static INLINE int mssb64(const uint64 value) { return mssb64_0(value) + 1; } #ifdef __GNUC__ #if defined(VM_X86_ANY) || defined(VM_ARM_ANY) /* *---------------------------------------------------------------------- * * uint16set -- * * memset a given address with an uint16 value, count times. * * Results: * Pointer to filled memory range. * * Side effects: * As with memset. * *---------------------------------------------------------------------- */ static INLINE void * uint16set(void *dst, uint16 val, size_t count) { #ifdef VM_ARM_32 void *tmpDst = dst; __asm__ __volatile__ ( "cmp %1, #0\n\t" "beq 2f\n" "1:\n\t" "strh %2, [%0], #2\n\t" "subs %1, %1, #1\n\t" "bne 1b\n" "2:" : "+r" (tmpDst), "+r" (count) : "r" (val) : "cc", "memory"); #elif defined(VM_ARM_64) void *tmpDst = dst; uint64 tmpVal = 0; if (count == 0) { return dst; } __asm__ __volatile__ ( "cbz %3, 1f\n\t" // Copy 16 bits twice... "bfm %2, %3, #0, #15\n\t" "lsl %2, %2, #16\n\t" "bfm %2, %3, #0, #15\n\t" // Copy 32 bits from the bottom of the reg. to the top... "lsl %2, %2, #32\n\t" "bfm %2, %2, #32, #63\n" // Copy into dst 8 bytes (4 uint16s) at a time "1:\t" "cmp %1, #4\n\t" "b.lo 2f\n\t" "str %2, [%0], #8\n\t" "sub %1, %1, #4\n\t" "b 1b\n" // Copy into dst 4 bytes at a time "2:\t" "cmp %1, #2\n\t" "b.lo 3f\n\t" "str %w2, [%0], #4\n\t" "sub %1, %1, #2\n\t" "b 2b\n" // We have 1 or zero items left... "3:\t" "cbz %1, 4f\n\t" "strh %w2, [%0]\n" "4:" : "+r" (tmpDst), "+r" (count), "+r" (tmpVal) : "r" (val) : "cc", "memory"); #else size_t dummy0; void *dummy1; __asm__ __volatile__("\t" "cld" "\n\t" "rep ; stosw" "\n" : "=c" (dummy0), "=D" (dummy1) : "0" (count), "1" (dst), "a" (val) : "memory", "cc" ); #endif return dst; } /* *---------------------------------------------------------------------- * * uint32set -- * * memset a given address with an uint32 value, count times. * * Results: * Pointer to filled memory range. * * Side effects: * As with memset. * *---------------------------------------------------------------------- */ static INLINE void * uint32set(void *dst, uint32 val, size_t count) { #ifdef VM_ARM_32 void *tmpDst = dst; __asm__ __volatile__ ( "cmp %1, #0\n\t" "beq 2f\n" "1:\n\t" "str %2, [%0], #4\n\t" "subs %1, %1, #1\n\t" "bne 1b\n" "2:" : "+r" (tmpDst), "+r" (count) : "r" (val) : "cc", "memory"); #elif defined(VM_ARM_64) void *tmpDst = dst; if (count == 0) { return dst; } __asm__ __volatile__ ( "cbz %2, 1f\n\t" // Drop our value in the top 32 bits, then copy from there to the bottom "lsl %2, %2, #32\n\t" "bfm %2, %2, #32, #63\n" // Copy four at a time "1:\t" "cmp %1, #16\n\t" "b.lo 2f\n\t" "stp %2, %2, [%0], #16\n\t" "stp %2, %2, [%0], #16\n\t" "stp %2, %2, [%0], #16\n\t" "stp %2, %2, [%0], #16\n\t" "sub %1, %1, #16\n\t" "b 1b\n" // Copy remaining pairs of data "2:\t" "cmp %1, #2\n\t" "b.lo 3f\n\t" "str %2, [%0], #8\n\t" "sub %1, %1, #2\n\t" "b 2b\n" // One or zero values left to copy "3:\t" "cbz %1, 4f\n\t" "str %w2, [%0]\n\t" // No incr "4:" : "+r" (tmpDst), "+r" (count), "+r" (val) : : "cc", "memory"); #else size_t dummy0; void *dummy1; __asm__ __volatile__("\t" "cld" "\n\t" "rep ; stosl" "\n" : "=c" (dummy0), "=D" (dummy1) : "0" (count), "1" (dst), "a" (val) : "memory", "cc" ); #endif return dst; } #else /* unknown system: rely on C to write */ static INLINE void * uint16set(void *dst, uint16 val, size_t count) { size_t i; for (i = 0; i < count; i++) { ((uint16 *) dst)[i] = val; } return dst; } static INLINE void * uint32set(void *dst, uint32 val, size_t count) { size_t i; for (i = 0; i < count; i++) { ((uint32 *) dst)[i] = val; } return dst; } #endif // defined(VM_X86_ANY) || defined(VM_ARM_ANY) #elif defined(_MSC_VER) static INLINE void * uint16set(void *dst, uint16 val, size_t count) { #ifdef VM_X86_64 __stosw((uint16*)dst, val, count); #elif defined(VM_ARM_32) size_t i; for (i = 0; i < count; i++) { ((uint16 *)dst)[i] = val; } #else __asm { pushf; mov ax, val; mov ecx, count; mov edi, dst; cld; rep stosw; popf; } #endif return dst; } static INLINE void * uint32set(void *dst, uint32 val, size_t count) { #ifdef VM_X86_64 __stosd((unsigned long*)dst, (unsigned long)val, count); #elif defined(VM_ARM_32) size_t i; for (i = 0; i < count; i++) { ((uint32 *)dst)[i] = val; } #else __asm { pushf; mov eax, val; mov ecx, count; mov edi, dst; cld; rep stosd; popf; } #endif return dst; } #else #error "No compiler defined for uint*set" #endif /* *----------------------------------------------------------------------------- * * Bswap16 -- * * Swap the 2 bytes of "v" as follows: 32 -> 23. * *----------------------------------------------------------------------------- */ static INLINE uint16 Bswap16(uint16 v) { #if defined(VM_ARM_64) __asm__("rev16 %w0, %w0" : "+r"(v)); return v; #else return ((v >> 8) & 0x00ff) | ((v << 8) & 0xff00); #endif } /* *----------------------------------------------------------------------------- * * Bswap32 -- * * Swap the 4 bytes of "v" as follows: 3210 -> 0123. * *----------------------------------------------------------------------------- */ static INLINE uint32 Bswap32(uint32 v) // IN { #if defined(__GNUC__) && defined(VM_X86_ANY) /* Checked against the Intel manual and GCC. --hpreg */ __asm__( "bswap %0" : "=r" (v) : "0" (v) ); return v; #elif defined(VM_ARM_32) && !defined(__ANDROID__) && !defined(_MSC_VER) __asm__("rev %0, %0" : "+r"(v)); return v; #elif defined(VM_ARM_64) __asm__("rev32 %x0, %x0" : "+r"(v)); return v; #else return (v >> 24) | ((v >> 8) & 0xFF00) | ((v & 0xFF00) << 8) | (v << 24) ; #endif } #define Bswap Bswap32 /* *----------------------------------------------------------------------------- * * Bswap64 -- * * Swap the 8 bytes of "v" as follows: 76543210 -> 01234567. * *----------------------------------------------------------------------------- */ static INLINE uint64 Bswap64(uint64 v) // IN { #if defined(VM_ARM_64) __asm__("rev %0, %0" : "+r"(v)); return v; #else return ((uint64)Bswap((uint32)v) << 32) | Bswap((uint32)(v >> 32)); #endif } /* * PAUSE is a P4 instruction that improves spinlock power+performance; * on non-P4 IA32 systems, the encoding is interpreted as a REPZ-NOP. * Use volatile to avoid NOP removal. */ static INLINE void PAUSE(void) #if defined(__GNUC__) || defined(VM_ARM_32) { #ifdef VM_ARM_ANY /* * ARM has no instruction to execute "spin-wait loop", just leave it * empty. */ #else __asm__ __volatile__( "pause" :); #endif } #elif defined(_MSC_VER) #ifdef VM_X86_64 { _mm_pause(); } #else /* VM_X86_64 */ #pragma warning( disable : 4035) { __asm _emit 0xf3 __asm _emit 0x90 } #pragma warning (default: 4035) #endif /* VM_X86_64 */ #else /* __GNUC__ */ #error No compiler defined for PAUSE #endif /* * Checked against the Intel manual and GCC --hpreg * * volatile because the tsc always changes without the compiler knowing it. */ static INLINE uint64 RDTSC(void) #ifdef __GNUC__ { #ifdef VM_X86_64 uint64 tscLow; uint64 tscHigh; __asm__ __volatile__( "rdtsc" : "=a" (tscLow), "=d" (tscHigh) ); return tscHigh << 32 | tscLow; #elif defined(VM_X86_32) uint64 tim; __asm__ __volatile__( "rdtsc" : "=A" (tim) ); return tim; #elif defined(VM_ARM_64) #if (defined(VMKERNEL) || defined(VMM)) && !defined(VMK_ARM_EL1) return MRS(CNTPCT_EL0); #else return MRS(CNTVCT_EL0); #endif #else /* * For platform without cheap timer, just return 0. */ return 0; #endif } #elif defined(_MSC_VER) #ifdef VM_X86_64 { return __rdtsc(); } #elif defined(VM_ARM_32) { /* * We need to do more inverstagetion here to find * a microsoft equivalent of that code */ NOT_IMPLEMENTED(); return 0; } #else #pragma warning( disable : 4035) { __asm _emit 0x0f __asm _emit 0x31 } #pragma warning (default: 4035) #endif /* VM_X86_64 */ #else /* __GNUC__ */ #error No compiler defined for RDTSC #endif /* __GNUC__ */ /* *----------------------------------------------------------------------------- * * DEBUGBREAK -- * * Does an int3 for MSVC / GCC, bkpt/brk for ARM. This is a macro to make * sure int3 is always inlined. * *----------------------------------------------------------------------------- */ #ifdef VM_ARM_32 #define DEBUGBREAK() __asm__("bkpt") #elif defined(VM_ARM_64) #define DEBUGBREAK() __asm__("brk #0") #elif defined(_MSC_VER) #define DEBUGBREAK() __debugbreak() #else #define DEBUGBREAK() __asm__("int $3") #endif /* *----------------------------------------------------------------------------- * * {Clear,Set,Test}Bit{32,64} -- * * Sets or clears a specified single bit in the provided variable. * * The index input value specifies which bit to modify and is 0-based. * Index is truncated by hardware to a 5-bit or 6-bit offset for the * 32 and 64-bit flavors, respectively, but input values are not validated * with asserts to avoid include dependencies. * * 64-bit flavors are not handcrafted for 32-bit builds because they may * defeat compiler optimizations. * *----------------------------------------------------------------------------- */ static INLINE void SetBit32(uint32 *var, uint32 index) { #if defined(__GNUC__) && defined(VM_X86_ANY) __asm__ ( "bts %1, %0" : "+mr" (*var) : "rI" (index) : "cc" ); #elif defined(_MSC_VER) _bittestandset((long *)var, index); #else *var |= (1 << index); #endif } static INLINE void ClearBit32(uint32 *var, uint32 index) { #if defined(__GNUC__) && defined(VM_X86_ANY) __asm__ ( "btr %1, %0" : "+mr" (*var) : "rI" (index) : "cc" ); #elif defined(_MSC_VER) _bittestandreset((long *)var, index); #else *var &= ~(1 << index); #endif } static INLINE void SetBit64(uint64 *var, uint64 index) { #if defined(VM_64BIT) && !defined(VM_ARM_64) #ifdef __GNUC__ __asm__ ( "bts %1, %0" : "+mr" (*var) : "rJ" (index) : "cc" ); #elif defined(_MSC_VER) _bittestandset64((__int64 *)var, index); #endif #else *var |= ((uint64)1 << index); #endif } static INLINE void ClearBit64(uint64 *var, uint64 index) { #if defined(VM_64BIT) && !defined(VM_ARM_64) #ifdef __GNUC__ __asm__ ( "btrq %1, %0" : "+mr" (*var) : "rJ" (index) : "cc" ); #elif defined(_MSC_VER) _bittestandreset64((__int64 *)var, index); #endif #else *var &= ~((uint64)1 << index); #endif } static INLINE Bool TestBit32(const uint32 *var, uint32 index) { #if defined(__GNUC__) && defined(VM_X86_ANY) Bool bit; __asm__ ( "bt %[index], %[var] \n" "setc %[bit]" : [bit] "=qQm" (bit) : [index] "rI" (index), [var] "r" (*var) : "cc" ); return bit; #else return (*var & (1 << index)) != 0; #endif } static INLINE Bool TestBit64(const uint64 *var, uint64 index) { #if defined __GNUC__ && defined VM_X86_64 Bool bit; __asm__ ( "bt %[index], %[var] \n" "setc %[bit]" : [bit] "=qQm" (bit) : [index] "rJ" (index), [var] "r" (*var) : "cc" ); return bit; #else return (*var & (CONST64U(1) << index)) != 0; #endif } /* *----------------------------------------------------------------------------- * * {Clear,Set,Complement,Test}BitVector -- * * Sets, clears, complements, or tests a specified single bit in the * provided array. The index input value specifies which bit to modify * and is 0-based. Bit number can be +-2Gb (+-128MB) relative from 'var' * variable. * * All functions return value of the bit before modification was performed. * *----------------------------------------------------------------------------- */ static INLINE Bool SetBitVector(void *var, int32 index) { #if defined(__GNUC__) && defined(VM_X86_ANY) Bool bit; __asm__ ( "bts %2, %1;" "setc %0" : "=qQm" (bit), "+m" (*(uint32 *)var) : "rI" (index) : "memory", "cc" ); return bit; #elif defined(_MSC_VER) return _bittestandset((long *)var, index) != 0; #else Bool retVal = (((uint8 *)var)[index / 8] & (1 << (index % 8))) != 0; ((uint8 *)var)[index / 8] |= 1 << (index % 8); return retVal; #endif } static INLINE Bool ClearBitVector(void *var, int32 index) { #if defined(__GNUC__) && defined(VM_X86_ANY) Bool bit; __asm__ ( "btr %2, %1;" "setc %0" : "=qQm" (bit), "+m" (*(uint32 *)var) : "rI" (index) : "memory", "cc" ); return bit; #elif defined(_MSC_VER) return _bittestandreset((long *)var, index) != 0; #else Bool retVal = (((uint8 *)var)[index / 8] & (1 << (index % 8))) != 0; ((uint8 *)var)[index / 8] &= ~(1 << (index % 8)); return retVal; #endif } static INLINE Bool ComplementBitVector(void *var, int32 index) { #if defined(__GNUC__) && defined(VM_X86_ANY) Bool bit; __asm__ ( "btc %2, %1;" "setc %0" : "=qQm" (bit), "+m" (*(uint32 *)var) : "rI" (index) : "memory", "cc" ); return bit; #elif defined(_MSC_VER) return _bittestandcomplement((long *)var, index) != 0; #else Bool retVal = (((uint8 *)var)[index / 8] & (1 << (index % 8))) != 0; ((uint8 *)var)[index / 8] ^= ~(1 << (index % 8)); return retVal; #endif } static INLINE Bool TestBitVector(const void *var, int32 index) { #if defined(__GNUC__) && defined(VM_X86_ANY) Bool bit; __asm__ ( "bt %2, %1;" "setc %0" : "=qQm" (bit) : "m" (*(const uint32 *)var), "rI" (index) : "cc" ); return bit; #elif defined _MSC_VER return _bittest((long *)var, index) != 0; #else return (((const uint8 *)var)[index / 8] & (1 << (index % 8))) != 0; #endif } /* *----------------------------------------------------------------------------- * RoundUpPow2_{64,32} -- * * Rounds a value up to the next higher power of 2. Returns the original * value if it is a power of 2. The next power of 2 for inputs {0, 1} is 1. * The result is undefined for inputs above {2^63, 2^31} (but equal to 1 * in this implementation). *----------------------------------------------------------------------------- */ static INLINE uint64 RoundUpPow2C64(uint64 value) { if (value <= 1 || value > (CONST64U(1) << 63)) { return 1; // Match the assembly's undefined value for large inputs. } else { return (CONST64U(2) << mssb64_0(value - 1)); } } #if defined(__GNUC__) && defined(VM_X86_64) static INLINE uint64 RoundUpPow2Asm64(uint64 value) { uint64 out = 2; __asm__("lea -1(%[in]), %%rcx;" // rcx = value - 1. Preserve original. "bsr %%rcx, %%rcx;" // rcx = log2(value - 1) if value != 1 // if value == 0, then rcx = 63 // if value == 1 then zf = 1, else zf = 0. "rol %%cl, %[out];" // out = 2 << rcx (if rcx != -1) // = 2^(log2(value - 1) + 1) // if rcx == -1 (value == 0), out = 1 // zf is always unmodified. "cmovz %[in], %[out]" // if value == 1 (zf == 1), write 1 to out. : [out]"+r"(out) : [in]"r"(value) : "%rcx", "cc"); return out; } #endif static INLINE uint64 RoundUpPow2_64(uint64 value) { #if defined(__GNUC__) && defined(VM_X86_64) if (__builtin_constant_p(value)) { return RoundUpPow2C64(value); } else { return RoundUpPow2Asm64(value); } #else return RoundUpPow2C64(value); #endif } static INLINE uint32 RoundUpPow2C32(uint32 value) { if (value <= 1 || value > (1U << 31)) { return 1; // Match the assembly's undefined value for large inputs. } else { return (2 << mssb32_0(value - 1)); } } #ifdef __GNUC__ static INLINE uint32 RoundUpPow2Asm32(uint32 value) { #ifdef VM_ARM_32 uint32 out = 1; // Note: None Thumb only! // The value of the argument "value" // will be affected! __asm__("sub %[in], %[in], #1;" // r1 = value - 1 . if value == 0 then r1 = 0xFFFFFFFF "clz %[in], %[in];" // r1 = log2(value - 1) if value != 1 // if value == 0 then r1 = 0 // if value == 1 then r1 = 32 "mov %[out], %[out], ror %[in]" // out = 2^(32 - r1) // if out == 2^32 then out = 1 as it is right rotate : [in]"+r"(value),[out]"+r"(out)); return out; #elif defined(VM_ARM_64) return RoundUpPow2C32(value); #else uint32 out = 2; __asm__("lea -1(%[in]), %%ecx;" // ecx = value - 1. Preserve original. "bsr %%ecx, %%ecx;" // ecx = log2(value - 1) if value != 1 // if value == 0, then ecx = 31 // if value == 1 then zf = 1, else zf = 0. "rol %%cl, %[out];" // out = 2 << ecx (if ecx != -1) // = 2^(log2(value - 1) + 1). // if ecx == -1 (value == 0), out = 1 // zf is always unmodified "cmovz %[in], %[out]" // if value == 1 (zf == 1), write 1 to out. : [out]"+r"(out) : [in]"r"(value) : "%ecx", "cc"); return out; #endif } #endif // __GNUC__ static INLINE uint32 RoundUpPow2_32(uint32 value) { #ifdef __GNUC__ if (__builtin_constant_p(value)) { return RoundUpPow2C32(value); } else { return RoundUpPow2Asm32(value); } #else return RoundUpPow2C32(value); #endif } #if defined __cplusplus } // extern "C" #endif #endif // _VM_BASIC_ASM_H_ vsock-only/shared/vmciKernelAPI3.h 0000444 0000000 0000000 00000003150 13432725350 016014 0 ustar root root /********************************************************* * Copyright (C) 2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmciKernelAPI3.h -- * * Kernel API (v3) exported from the VMCI host and guest drivers. */ #ifndef __VMCI_KERNELAPI_3_H__ #define __VMCI_KERNELAPI_3_H__ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" #include "vmciKernelAPI2.h" #if defined __cplusplus extern "C" { #endif /* Define version 3. */ #undef VMCI_KERNEL_API_VERSION #define VMCI_KERNEL_API_VERSION_3 3 #define VMCI_KERNEL_API_VERSION VMCI_KERNEL_API_VERSION_3 /* VMCI Detach Cause API (only available in vmkernel). */ #define VMCI_DETACH_REGULAR 0 #define VMCI_DETACH_VMOTION 1 int vmci_qpair_get_detach_cause(VMCIQPair *qpair, uint8 *cause); #if defined __cplusplus } // extern "C" #endif #endif /* !__VMCI_KERNELAPI_3_H__ */ vsock-only/shared/guest_msg_def.h 0000444 0000000 0000000 00000005647 13432725350 016130 0 ustar root root /********************************************************* * Copyright (C) 1998-2016 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * guest_msg_def.h -- * * Second layer of the internal communication channel between guest * applications and vmware * */ #ifndef _GUEST_MSG_DEF_H_ #define _GUEST_MSG_DEF_H_ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_USERLEVEL #include "includeCheck.h" /* Basic request types */ typedef enum { MESSAGE_TYPE_OPEN, MESSAGE_TYPE_SENDSIZE, MESSAGE_TYPE_SENDPAYLOAD, MESSAGE_TYPE_RECVSIZE, MESSAGE_TYPE_RECVPAYLOAD, MESSAGE_TYPE_RECVSTATUS, MESSAGE_TYPE_CLOSE, } MessageType; /* Reply statuses */ /* The basic request succeeded */ #define MESSAGE_STATUS_SUCCESS 0x0001 /* vmware has a message available for its party */ #define MESSAGE_STATUS_DORECV 0x0002 /* The channel has been closed */ #define MESSAGE_STATUS_CLOSED 0x0004 /* vmware removed the message before the party fetched it */ #define MESSAGE_STATUS_UNSENT 0x0008 /* A checkpoint occurred */ #define MESSAGE_STATUS_CPT 0x0010 /* An underlying device is powering off */ #define MESSAGE_STATUS_POWEROFF 0x0020 /* vmware has detected a timeout on the channel */ #define MESSAGE_STATUS_TIMEOUT 0x0040 /* vmware supports high-bandwidth for sending and receiving the payload */ #define MESSAGE_STATUS_HB 0x0080 /* * This mask defines the status bits that the guest is allowed to set; * we use this to mask out all other bits when receiving the status * from the guest. Otherwise, the guest can manipulate VMX state by * setting status bits that are only supposed to be changed by the * VMX. See bug 45385. */ #define MESSAGE_STATUS_GUEST_MASK MESSAGE_STATUS_SUCCESS /* * Max number of channels. * Unfortunately this has to be public because the monitor part * of the backdoor needs it for its trivial-case optimization. [greg] */ #define GUESTMSG_MAX_CHANNEL 8 /* Flags to open a channel. --hpreg */ #define GUESTMSG_FLAG_COOKIE 0x80000000 #define GUESTMSG_FLAG_ALL GUESTMSG_FLAG_COOKIE /* * Maximum size of incoming message. This is to prevent denial of host service * attacks from guest applications. */ #define GUESTMSG_MAX_IN_SIZE (64 * 1024) #endif /* _GUEST_MSG_DEF_H_ */ vsock-only/shared/dbllnklst.h 0000444 0000000 0000000 00000015155 13432725346 015306 0 ustar root root /********************************************************* * Copyright (C) 1998-2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * dbllnklst.h -- * * Double linked lists */ #ifndef _DBLLNKLST_H_ #define _DBLLNKLST_H_ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_USERLEVEL #include "includeCheck.h" #include "vm_basic_types.h" #if defined(__cplusplus) extern "C" { #endif #define DblLnkLst_OffsetOf(type, field) ((intptr_t)&((type *)0)->field) #define DblLnkLst_Container(addr, type, field) \ ((type *)((char *)(addr) - DblLnkLst_OffsetOf(type, field))) #define DblLnkLst_ForEach(curr, head) \ for (curr = (head)->next; curr != (head); curr = (curr)->next) /* Safe from list element removal within loop body. */ #define DblLnkLst_ForEachSafe(curr, nextElem, head) \ for (curr = (head)->next, nextElem = (curr)->next; \ curr != (head); \ curr = nextElem, nextElem = (curr)->next) typedef struct DblLnkLst_Links { struct DblLnkLst_Links *prev; struct DblLnkLst_Links *next; } DblLnkLst_Links; /* * Functions * * DblLnkLst_LinkFirst, DblLnkLst_LinkLast, and DblLnkLst_Swap are specific * to anchored lists. The rest are for both circular and anchored lists. */ /* *---------------------------------------------------------------------- * * DblLnkLst_Init -- * * Initialize a member of a doubly linked list * * Result * None * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE void DblLnkLst_Init(DblLnkLst_Links *l) // OUT { l->prev = l->next = l; } /* *---------------------------------------------------------------------- * * DblLnkLst_Link -- * * Merge two doubly linked lists into one * * The operation is commutative * The operation is inversible (its inverse is DblLnkLst_Unlink) * * Result * None * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE void DblLnkLst_Link(DblLnkLst_Links *l1, // IN/OUT DblLnkLst_Links *l2) // IN/OUT { DblLnkLst_Links *tmp; (tmp = l1->prev)->next = l2; (l1->prev = l2->prev)->next = l1; l2->prev = tmp ; } /* *---------------------------------------------------------------------- * * DblLnkLst_Unlink -- * * Split one doubly linked list into two * * No check is performed: the caller must ensure that both members * belong to the same doubly linked list * * The operation is commutative * The operation is inversible (its inverse is DblLnkLst_Link) * * Result * None * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE void DblLnkLst_Unlink(DblLnkLst_Links *l1, // IN/OUT DblLnkLst_Links *l2) // IN/OUT { DblLnkLst_Links *tmp; tmp = l1->prev ; (l1->prev = l2->prev)->next = l1; (l2->prev = tmp )->next = l2; } /* *---------------------------------------------------------------------- * * DblLnkLst_Unlink1 -- * * Unlink an element from its list. * * Result * None * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE void DblLnkLst_Unlink1(DblLnkLst_Links *l) // IN/OUT { DblLnkLst_Unlink(l, l->next); } /* *---------------------------------------------------------------------------- * * DblLnkLst_IsLinked -- * * Determines whether an element is linked with any other elements. * * Results: * TRUE if link is linked, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static INLINE Bool DblLnkLst_IsLinked(DblLnkLst_Links const *l) // IN { /* * A DblLnkLst_Links is either linked to itself (not linked) or linked to * other elements in a list (linked). */ return l->prev != l; } /* *---------------------------------------------------------------------- * * DblLnkLst_LinkFirst -- * * Insert 'l' at the beginning of the list anchored at 'head' * * Result * None * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE void DblLnkLst_LinkFirst(DblLnkLst_Links *head, // IN/OUT DblLnkLst_Links *l) // IN/OUT { DblLnkLst_Link(head->next, l); } /* *---------------------------------------------------------------------- * * DblLnkLst_LinkLast -- * * Insert 'l' at the end of the list anchored at 'head' * * Result * None * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE void DblLnkLst_LinkLast(DblLnkLst_Links *head, // IN/OUT DblLnkLst_Links *l) // IN/OUT { DblLnkLst_Link(head, l); } /* *---------------------------------------------------------------------- * * DblLnkLst_Swap -- * * Swap all entries between the list anchored at 'head1' and the list * anchored at 'head2'. * * The operation is commutative * The operation is inversible (its inverse is itself) * * Result * None * * Side effects: * None * *---------------------------------------------------------------------- */ static INLINE void DblLnkLst_Swap(DblLnkLst_Links *head1, // IN/OUT DblLnkLst_Links *head2) // IN/OUT { DblLnkLst_Links const tmp = *head1; if (DblLnkLst_IsLinked(head2)) { (head1->prev = head2->prev)->next = head1; (head1->next = head2->next)->prev = head1; } else { DblLnkLst_Init(head1); } if (tmp.prev != head1) { (head2->prev = tmp.prev)->next = head2; (head2->next = tmp.next)->prev = head2; } else { DblLnkLst_Init(head2); } } #if defined(__cplusplus) } // extern "C" #endif #endif /* _DBLLNKLST_H_ */ vsock-only/shared/compat_scsi.h 0000444 0000000 0000000 00000003024 13432725347 015612 0 ustar root root /********************************************************* * Copyright (C) 2002 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_SCSI_H__ # define __COMPAT_SCSI_H__ /* The scsi_bufflen() API appeared somewhere in time --hpreg */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 23) # define scsi_bufflen(cmd) ((cmd)->request_bufflen) # define scsi_sg_count(cmd) ((cmd)->use_sg) # define scsi_sglist(cmd) ((struct scatterlist *)(cmd)->request_buffer) # define scsi_set_resid(cmd, _resid) ((cmd)->resid = _resid) #endif /* * Using scsi_sglist to access the request buffer looks strange * so instead we define this macro. What happened is later kernel * put all SCSI data in sglists, since it simplifies passing buffers */ #define scsi_request_buffer(cmd) scsi_sglist(cmd) #endif /* __COMPAT_SCSI_H__ */ vsock-only/shared/vm_basic_math.h 0000444 0000000 0000000 00000010721 13432725350 016076 0 ustar root root /********************************************************* * Copyright (C) 2008-2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vm_basic_math.h -- * * Standard mathematical macros for VMware source code. */ #ifndef _VM_BASIC_MATH_H_ #define _VM_BASIC_MATH_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_VMKDRIVERS #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" #include "vm_basic_types.h" // For INLINE. #include "vm_basic_asm.h" // For Div64... #if defined __cplusplus extern "C" { #endif static INLINE uint32 RatioOf(uint32 numer1, uint32 numer2, uint32 denom) { uint64 numer = (uint64)numer1 * numer2; /* Calculate "(numer1 * numer2) / denom" avoiding round-off errors. */ #if defined(VMM) || !(defined(__i386__) || defined(__x86_64__)) return numer / denom; #else uint32 ratio; uint32 unused; Div643232(numer, denom, &ratio, &unused); return ratio; #endif } static INLINE uint32 ExponentialAvg(uint32 avg, uint32 value, uint32 gainNumer, uint32 gainDenom) { uint32 term1 = gainNumer * avg; uint32 term2 = (gainDenom - gainNumer) * value; return (term1 + term2) / gainDenom; } /* *----------------------------------------------------------------------------- * * IsZeroOrPowerOfTwo -- * IsZeroOrPowerOfTwo64 -- * * Results: * TRUE iff the value is 0 or a power of two. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE Bool IsZeroOrPowerOfTwo64(uint64 x) { return !(x & (x - 1)); } static INLINE Bool IsZeroOrPowerOfTwo(uint32 x) // IN { return !(x & (x - 1)); } static INLINE uint32 GetPowerOfTwo(uint32 x) { /* Returns next-greatest power-of-two. */ uint32 power2 = 1; while (x > power2) { power2 = power2 << 1; } return power2; } #if !defined(_WIN32) && !defined(_WIN64) /* *----------------------------------------------------------------------------- * * RotateLeft32 -- * * Results: * Value rotated to the left by 'shift' bits. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE uint32 RotateLeft32(uint32 value, uint8 shift) { return ((value << shift) | (value >> (32 - shift))); } /* *----------------------------------------------------------------------------- * * RotateRight32 -- * * Results: * Value rotated to the right by 'shift' bits. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE uint32 RotateRight32(uint32 value, uint8 shift) { return ((value >> shift) | (value << (32 - shift))); } /* *----------------------------------------------------------------------------- * * RotateLeft64 -- * * Results: * Value rotated to the left by 'shift' bits. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE uint64 RotateLeft64(uint64 value, uint8 shift) { return ((value << shift) | (value >> (64 - shift))); } /* *----------------------------------------------------------------------------- * * RotateRight64 -- * * Results: * Value rotated to the right by 'shift' bits. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE uint64 RotateRight64(uint64 value, uint8 shift) { return ((value >> shift) | (value << (64 - shift))); } #endif // if !defined(_WIN32) && !defined(_WIN64) #if defined __cplusplus } // extern "C" #endif #endif // ifndef _VM_BASIC_MATH_H_ vsock-only/shared/compat_slab.h 0000444 0000000 0000000 00000006653 13432725347 015605 0 ustar root root /********************************************************* * Copyright (C) 2005 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_SLAB_H__ # define __COMPAT_SLAB_H__ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 2, 0) # include <linux/slab.h> #else # include <linux/malloc.h> #endif /* * Before 2.6.20, kmem_cache_t was the accepted way to refer to a kmem_cache * structure. Prior to 2.6.15, this structure was called kmem_cache_s, and * afterwards it was renamed to kmem_cache. Here we keep things simple and use * the accepted typedef until it became deprecated, at which point we switch * over to the kmem_cache name. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 20) # define compat_kmem_cache struct kmem_cache #else # define compat_kmem_cache kmem_cache_t #endif /* * Up to 2.6.22 kmem_cache_create has 6 arguments - name, size, alignment, flags, * constructor, and destructor. Then for some time kernel was asserting that * destructor is NULL, and since 2.6.23-pre1 kmem_cache_create takes only 5 * arguments - destructor is gone. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 22) || defined(VMW_KMEMCR_HAS_DTOR) #define compat_kmem_cache_create(name, size, align, flags, ctor) \ kmem_cache_create(name, size, align, flags, ctor, NULL) #else #define compat_kmem_cache_create(name, size, align, flags, ctor) \ kmem_cache_create(name, size, align, flags, ctor) #endif /* * Up to 2.6.23 kmem_cache constructor has three arguments - pointer to block to * prepare (aka "this"), from which cache it came, and some unused flags. After * 2.6.23 flags were removed, and order of "this" and cache parameters was swapped... * Since 2.6.27-rc2 everything is different again, and ctor has only one argument. * * HAS_3_ARGS has precedence over HAS_2_ARGS if both are defined. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 23) && !defined(VMW_KMEMCR_CTOR_HAS_3_ARGS) # define VMW_KMEMCR_CTOR_HAS_3_ARGS #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 26) && !defined(VMW_KMEMCR_CTOR_HAS_2_ARGS) # define VMW_KMEMCR_CTOR_HAS_2_ARGS #endif #if defined(VMW_KMEMCR_CTOR_HAS_3_ARGS) typedef void compat_kmem_cache_ctor(void *, compat_kmem_cache *, unsigned long); #define COMPAT_KMEM_CACHE_CTOR_ARGS(arg) void *arg, \ compat_kmem_cache *cache, \ unsigned long flags #elif defined(VMW_KMEMCR_CTOR_HAS_2_ARGS) typedef void compat_kmem_cache_ctor(compat_kmem_cache *, void *); #define COMPAT_KMEM_CACHE_CTOR_ARGS(arg) compat_kmem_cache *cache, \ void *arg #else typedef void compat_kmem_cache_ctor(void *); #define COMPAT_KMEM_CACHE_CTOR_ARGS(arg) void *arg #endif #endif /* __COMPAT_SLAB_H__ */ vsock-only/shared/compat_mm.h 0000444 0000000 0000000 00000003002 13432725347 015256 0 ustar root root /********************************************************* * Copyright (C) 2002 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_MM_H__ # define __COMPAT_MM_H__ #include <linux/mm.h> /* 2.2.x uses 0 instead of some define */ #ifndef NOPAGE_SIGBUS #define NOPAGE_SIGBUS (0) #endif /* 2.2.x does not have HIGHMEM support */ #ifndef GFP_HIGHUSER #define GFP_HIGHUSER (GFP_USER) #endif /* * In 2.4.14, the logic behind the UnlockPage macro was moved to the * unlock_page() function. Later (in 2.5.12), the UnlockPage macro was removed * altogether, and nowadays everyone uses unlock_page(). */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 4, 14) #define compat_unlock_page(page) UnlockPage(page) #else #define compat_unlock_page(page) unlock_page(page) #endif #endif /* __COMPAT_MM_H__ */ vsock-only/shared/driver-config.h 0000444 0000000 0000000 00000004314 13432725347 016047 0 ustar root root /********************************************************* * Copyright (C) 1998 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * Sets the proper defines from the Linux header files * * This file must be included before the inclusion of any kernel header file, * with the exception of linux/autoconf.h and linux/version.h --hpreg */ #ifndef __VMX_CONFIG_H__ #define __VMX_CONFIG_H__ #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMKDRIVERS #include "includeCheck.h" #include "compat_version.h" #include "compat_autoconf.h" /* * We rely on Kernel Module support. Check here. */ #ifndef CONFIG_MODULES # error "No Module support in this kernel. Please configure with CONFIG_MODULES" #endif /* * 2.2 kernels still use __SMP__ (derived from CONFIG_SMP * in the main Makefile), so we do it here. */ #ifdef CONFIG_SMP # define __SMP__ 1 #endif #if defined(CONFIG_MODVERSIONS) && defined(KERNEL_2_1) # if LINUX_VERSION_CODE < KERNEL_VERSION(2,5,60) /* * MODVERSIONS might be already defined when using kernel's Makefiles. */ # ifndef MODVERSIONS # define MODVERSIONS # endif # include <linux/modversions.h> # endif #endif #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 24) /* * Force the uintptr_t definition to come from linux/types.h instead of vm_basic_types.h. */ # include <linux/types.h> # define _STDINT_H 1 #endif #ifndef __KERNEL__ # define __KERNEL__ #endif #endif vsock-only/shared/x86cpuid_asm.h 0000444 0000000 0000000 00000023356 13432725350 015624 0 ustar root root /********************************************************* * Copyright (C) 2003-2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * x86cpuid_asm.h * * CPUID-related assembly functions. */ #ifndef _X86CPUID_ASM_H_ #define _X86CPUID_ASM_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_DISTRIBUTE #define INCLUDE_ALLOW_VMCORE #include "includeCheck.h" #include "vm_basic_asm.h" #include "x86cpuid.h" /* * x86-64 windows doesn't support inline asm so we have to use these * intrinsic functions defined in the compiler. Not all of these are well * documented. There is an array in the compiler dll (c1.dll) which has * an array of the names of all the intrinsics minus the leading * underscore. Searching around in the ntddk.h file can also be helpful. * * The declarations for the intrinsic functions were taken from the DDK. * Our declarations must match the ddk's otherwise the 64-bit c++ compiler * will complain about second linkage of the intrinsic functions. * We define the intrinsic using the basic types corresponding to the * Windows typedefs. This avoids having to include windows header files * to get to the windows types. */ #ifdef _MSC_VER #ifdef __cplusplus extern "C" { #endif #ifdef VM_X86_64 /* * intrinsic functions only supported by x86-64 windows as of 2k3sp1 */ void __cpuid(int regs[4], int eax); #pragma intrinsic(__cpuid) /* * __cpuidex has been supported since VS2008 */ #if _MSC_VER >= 1500 void __cpuidex(int regs[4], int eax, int ecx); #pragma intrinsic(__cpuidex) #endif /* _MSC_VER >= 1500 */ #endif /* VM_X86_64 */ #ifdef __cplusplus } #endif #endif /* _MSC_VER */ #ifdef __GNUC__ // { /* * Checked against the Intel manual and GCC --hpreg * * Need __volatile__ and "memory" since CPUID has a synchronizing effect. * The CPUID may also change at runtime (APIC flag, etc). * */ /* * %ebx is reserved on i386 PIC. Apple's gcc-5493 (gcc 4.0) compiling * for x86_64 incorrectly errors out saying %ebx is reserved. This is * Apple bug 7304232. */ #if vm_x86_64 ? (defined __APPLE_CC__ && __APPLE_CC__ == 5493) : defined __PIC__ #if vm_x86_64 /* * Note that this generates movq %rbx,%rbx; cpuid; xchgq %rbx,%rbx ... * Unfortunately Apple's assembler does not have .ifnes, and I cannot * figure out how to do that with .if. If we ever enable this code * on other 64bit systems, both movq & xchgq should be surrounded by * .ifnes \"%%rbx\", \"%q1\" & .endif */ #define VM_CPUID_BLOCK "movq %%rbx, %q1\n\t" \ "cpuid\n\t" \ "xchgq %%rbx, %q1\n\t" #define VM_EBX_OUT(reg) "=&r"(reg) #else #define VM_CPUID_BLOCK "movl %%ebx, %1\n\t" \ "cpuid\n\t" \ "xchgl %%ebx, %1\n\t" #define VM_EBX_OUT(reg) "=&rm"(reg) #endif #else #define VM_CPUID_BLOCK "cpuid" #define VM_EBX_OUT(reg) "=b"(reg) #endif static INLINE void __GET_CPUID(int eax, // IN CPUIDRegs *regs) // OUT { __asm__ __volatile__( VM_CPUID_BLOCK : "=a" (regs->eax), VM_EBX_OUT(regs->ebx), "=c" (regs->ecx), "=d" (regs->edx) : "a" (eax) : "memory" ); } static INLINE void __GET_CPUID2(int eax, // IN int ecx, // IN CPUIDRegs *regs) // OUT { __asm__ __volatile__( VM_CPUID_BLOCK : "=a" (regs->eax), VM_EBX_OUT(regs->ebx), "=c" (regs->ecx), "=d" (regs->edx) : "a" (eax), "c" (ecx) : "memory" ); } static INLINE uint32 __GET_EAX_FROM_CPUID(int eax) // IN { uint32 ebx; __asm__ __volatile__( VM_CPUID_BLOCK : "=a" (eax), VM_EBX_OUT(ebx) : "a" (eax) : "memory", "%ecx", "%edx" ); return eax; } static INLINE uint32 __GET_EBX_FROM_CPUID(int eax) // IN { uint32 ebx; __asm__ __volatile__( VM_CPUID_BLOCK : "=a" (eax), VM_EBX_OUT(ebx) : "a" (eax) : "memory", "%ecx", "%edx" ); return ebx; } static INLINE uint32 __GET_ECX_FROM_CPUID(int eax) // IN { uint32 ecx; uint32 ebx; __asm__ __volatile__( VM_CPUID_BLOCK : "=a" (eax), VM_EBX_OUT(ebx), "=c" (ecx) : "a" (eax) : "memory", "%edx" ); return ecx; } static INLINE uint32 __GET_EDX_FROM_CPUID(int eax) // IN { uint32 edx; uint32 ebx; __asm__ __volatile__( VM_CPUID_BLOCK : "=a" (eax), VM_EBX_OUT(ebx), "=d" (edx) : "a" (eax) : "memory", "%ecx" ); return edx; } static INLINE uint32 __GET_EAX_FROM_CPUID4(int ecx) // IN { uint32 eax; uint32 ebx; __asm__ __volatile__( VM_CPUID_BLOCK : "=a" (eax), VM_EBX_OUT(ebx), "=c" (ecx) : "a" (4), "c" (ecx) : "memory", "%edx" ); return eax; } #undef VM_CPUID_BLOCK #undef VM_EBX_OUT #elif defined(_MSC_VER) // } { static INLINE void __GET_CPUID(int input, CPUIDRegs *regs) { #ifdef VM_X86_64 __cpuid((int *)regs, input); #else __asm push esi __asm push ebx __asm push ecx __asm push edx __asm mov eax, input __asm mov esi, regs __asm _emit 0x0f __asm _emit 0xa2 __asm mov 0x0[esi], eax __asm mov 0x4[esi], ebx __asm mov 0x8[esi], ecx __asm mov 0xC[esi], edx __asm pop edx __asm pop ecx __asm pop ebx __asm pop esi #endif } #ifdef VM_X86_64 #if _MSC_VER >= 1500 /* * __cpuidex has been supported since VS2008 */ static INLINE void __GET_CPUID2(int inputEax, int inputEcx, CPUIDRegs *regs) { __cpuidex((int *)regs, inputEax, inputEcx); } #else // _MSC_VER >= 1500 /* * No inline assembly in Win64. Implemented in bora/lib/misc in * cpuidMasm64.asm. */ extern void __GET_CPUID2(int inputEax, int inputEcx, CPUIDRegs *regs); #endif // _MSC_VER >= 1500 #else // VM_X86_64 static INLINE void __GET_CPUID2(int inputEax, int inputEcx, CPUIDRegs *regs) { __asm push esi __asm push ebx __asm push ecx __asm push edx __asm mov eax, inputEax __asm mov ecx, inputEcx __asm mov esi, regs __asm _emit 0x0f __asm _emit 0xa2 __asm mov 0x0[esi], eax __asm mov 0x4[esi], ebx __asm mov 0x8[esi], ecx __asm mov 0xC[esi], edx __asm pop edx __asm pop ecx __asm pop ebx __asm pop esi } #endif static INLINE uint32 __GET_EAX_FROM_CPUID(int input) { #ifdef VM_X86_64 CPUIDRegs regs; __cpuid((int *)®s, input); return regs.eax; #else uint32 output; //NOT_TESTED(); __asm push ebx __asm push ecx __asm push edx __asm mov eax, input __asm _emit 0x0f __asm _emit 0xa2 __asm mov output, eax __asm pop edx __asm pop ecx __asm pop ebx return output; #endif } static INLINE uint32 __GET_EBX_FROM_CPUID(int input) { #ifdef VM_X86_64 CPUIDRegs regs; __cpuid((int *)®s, input); return regs.ebx; #else uint32 output; //NOT_TESTED(); __asm push ebx __asm push ecx __asm push edx __asm mov eax, input __asm _emit 0x0f __asm _emit 0xa2 __asm mov output, ebx __asm pop edx __asm pop ecx __asm pop ebx return output; #endif } static INLINE uint32 __GET_ECX_FROM_CPUID(int input) { #ifdef VM_X86_64 CPUIDRegs regs; __cpuid((int *)®s, input); return regs.ecx; #else uint32 output; //NOT_TESTED(); __asm push ebx __asm push ecx __asm push edx __asm mov eax, input __asm _emit 0x0f __asm _emit 0xa2 __asm mov output, ecx __asm pop edx __asm pop ecx __asm pop ebx return output; #endif } static INLINE uint32 __GET_EDX_FROM_CPUID(int input) { #ifdef VM_X86_64 CPUIDRegs regs; __cpuid((int *)®s, input); return regs.edx; #else uint32 output; //NOT_TESTED(); __asm push ebx __asm push ecx __asm push edx __asm mov eax, input __asm _emit 0x0f __asm _emit 0xa2 __asm mov output, edx __asm pop edx __asm pop ecx __asm pop ebx return output; #endif } #ifdef VM_X86_64 /* * No inline assembly in Win64. Implemented in bora/lib/misc in * cpuidMasm64.asm. */ extern uint32 __GET_EAX_FROM_CPUID4(int inputEcx); #else // VM_X86_64 static INLINE uint32 __GET_EAX_FROM_CPUID4(int inputEcx) { uint32 output; //NOT_TESTED(); __asm push ebx __asm push ecx __asm push edx __asm mov eax, 4 __asm mov ecx, inputEcx __asm _emit 0x0f __asm _emit 0xa2 __asm mov output, eax __asm pop edx __asm pop ecx __asm pop ebx return output; } #endif // VM_X86_64 #else // } #error #endif #define CPUID_FOR_SIDE_EFFECTS() ((void)__GET_EAX_FROM_CPUID(0)) /* The first parameter is used as an rvalue and then as an lvalue. */ #define GET_CPUID(_ax, _bx, _cx, _dx) { \ CPUIDRegs regs; \ __GET_CPUID(_ax, ®s); \ _ax = regs.eax; \ _bx = regs.ebx; \ _cx = regs.ecx; \ _dx = regs.edx; \ } #define GET_CPUID2(_ax, _bx, _cx, _dx) {\ CPUIDRegs regs; \ __GET_CPUID2(_ax, _cx, ®s); \ _ax = regs.eax; \ _bx = regs.ebx; \ _cx = regs.ecx; \ _dx = regs.edx; \ } #endif vsock-only/shared/vmci_call_defs.h 0000444 0000000 0000000 00000023713 13432725350 016241 0 ustar root root /********************************************************* * Copyright (C) 2006-2016,2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef _VMCI_CALL_DEFS_H_ #define _VMCI_CALL_DEFS_H_ #define INCLUDE_ALLOW_USERLEVEL #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMMON #define INCLUDE_ALLOW_VMCORE #define INCLUDE_ALLOW_VMKMOD #define INCLUDE_ALLOW_VMKERNEL #define INCLUDE_ALLOW_DISTRIBUTE #include "includeCheck.h" #include "vm_basic_types.h" #include "vmci_defs.h" #if defined __cplusplus extern "C" { #endif /* * All structs here are an integral size of their largest member, ie. a struct * with at least one 8-byte member will have a size that is an integral of 8. * A struct which has a largest member of size 4 will have a size that is an * integral of 4. This is because Windows CL enforces this rule. 32 bit gcc * doesn't e.g. 32 bit gcc can misalign an 8 byte member if it is preceeded by * a 4 byte member. */ /* * Base struct for vmci datagrams. */ typedef struct VMCIDatagram { VMCIHandle dst; VMCIHandle src; uint64 payloadSize; } VMCIDatagram; /* * Second flag is for creating a well-known handle instead of a per context * handle. Next flag is for deferring datagram delivery, so that the * datagram callback is invoked in a delayed context (not interrupt context). */ #define VMCI_FLAG_DG_NONE 0 #define VMCI_FLAG_WELLKNOWN_DG_HND 0x1 #define VMCI_FLAG_ANYCID_DG_HND 0x2 #define VMCI_FLAG_DG_DELAYED_CB 0x4 /* Event callback should fire in a delayed context (not interrupt context.) */ #define VMCI_FLAG_EVENT_NONE 0 #define VMCI_FLAG_EVENT_DELAYED_CB 0x1 /* * Maximum supported size of a VMCI datagram for routable datagrams. * Datagrams going to the hypervisor are allowed to be larger. */ #define VMCI_MAX_DG_SIZE (17 * 4096) #define VMCI_MAX_DG_PAYLOAD_SIZE (VMCI_MAX_DG_SIZE - sizeof(VMCIDatagram)) #define VMCI_DG_PAYLOAD(_dg) (void *)((char *)(_dg) + sizeof(VMCIDatagram)) #define VMCI_DG_HEADERSIZE sizeof(VMCIDatagram) #define VMCI_DG_SIZE(_dg) (VMCI_DG_HEADERSIZE + (size_t)(_dg)->payloadSize) #define VMCI_DG_SIZE_ALIGNED(_dg) ((VMCI_DG_SIZE(_dg) + 7) & (size_t)~7) #define VMCI_MAX_DATAGRAM_QUEUE_SIZE (VMCI_MAX_DG_SIZE * 2) /* * We allow at least 1024 more event datagrams from the hypervisor past the * normally allowed datagrams pending for a given context. We define this * limit on event datagrams from the hypervisor to guard against DoS attack * from a malicious VM which could repeatedly attach to and detach from a queue * pair, causing events to be queued at the destination VM. However, the rate * at which such events can be generated is small since it requires a VM exit * and handling of queue pair attach/detach call at the hypervisor. Event * datagrams may be queued up at the destination VM if it has interrupts * disabled or if it is not draining events for some other reason. 1024 * datagrams is a grossly conservative estimate of the time for which * interrupts may be disabled in the destination VM, but at the same time does * not exacerbate the memory pressure problem on the host by much (size of each * event datagram is small). */ #define VMCI_MAX_DATAGRAM_AND_EVENT_QUEUE_SIZE \ (VMCI_MAX_DATAGRAM_QUEUE_SIZE + \ 1024 * (sizeof(VMCIDatagram) + sizeof(VMCIEventData_Max))) /* * Struct for sending VMCI_DATAGRAM_REQUEST_MAP and * VMCI_DATAGRAM_REMOVE_MAP datagrams. Struct size is 32 bytes. All * fields in struct are aligned to their natural alignment. These * datagrams are obsoleted by the removal of VM to VM communication. */ typedef struct VMCIDatagramWellKnownMapMsg { VMCIDatagram hdr; VMCIId wellKnownID; uint32 _pad; } VMCIDatagramWellKnownMapMsg; /* * Struct used for querying, via VMCI_RESOURCES_QUERY, the availability of * hypervisor resources. * Struct size is 16 bytes. All fields in struct are aligned to their natural * alignment. */ typedef struct VMCIResourcesQueryHdr { VMCIDatagram hdr; uint32 numResources; uint32 _padding; } VMCIResourcesQueryHdr; /* * Convenience struct for negotiating vectors. Must match layout of * VMCIResourceQueryHdr minus the VMCIDatagram header. */ typedef struct VMCIResourcesQueryMsg { uint32 numResources; uint32 _padding; VMCI_Resource resources[1]; } VMCIResourcesQueryMsg; /* * The maximum number of resources that can be queried using * VMCI_RESOURCE_QUERY is 31, as the result is encoded in the lower 31 * bits of a positive return value. Negative values are reserved for * errors. */ #define VMCI_RESOURCE_QUERY_MAX_NUM 31 /* Maximum size for the VMCI_RESOURCE_QUERY request. */ #define VMCI_RESOURCE_QUERY_MAX_SIZE sizeof(VMCIResourcesQueryHdr) \ + VMCI_RESOURCE_QUERY_MAX_NUM * sizeof(VMCI_Resource) /* * Struct used for setting the notification bitmap. All fields in * struct are aligned to their natural alignment. */ typedef struct VMCINotifyBitmapSetMsg { VMCIDatagram hdr; PPN32 bitmapPPN; uint32 _pad; } VMCINotifyBitmapSetMsg; /* * Struct used for linking a doorbell handle with an index in the * notify bitmap. All fields in struct are aligned to their natural * alignment. */ typedef struct VMCIDoorbellLinkMsg { VMCIDatagram hdr; VMCIHandle handle; uint64 notifyIdx; } VMCIDoorbellLinkMsg; /* * Struct used for unlinking a doorbell handle from an index in the * notify bitmap. All fields in struct are aligned to their natural * alignment. */ typedef struct VMCIDoorbellUnlinkMsg { VMCIDatagram hdr; VMCIHandle handle; } VMCIDoorbellUnlinkMsg; /* * Struct used for generating a notification on a doorbell handle. All * fields in struct are aligned to their natural alignment. */ typedef struct VMCIDoorbellNotifyMsg { VMCIDatagram hdr; VMCIHandle handle; } VMCIDoorbellNotifyMsg; /* * This struct is used to contain data for events. Size of this struct is a * multiple of 8 bytes, and all fields are aligned to their natural alignment. */ typedef struct VMCI_EventData { VMCI_Event event; /* 4 bytes. */ uint32 _pad; /* * Event payload is put here. */ } VMCI_EventData; /* Callback needed for correctly waiting on events. */ typedef int (*VMCIDatagramRecvCB)(void *clientData, // IN: client data for handler VMCIDatagram *msg); // IN: /* * We use the following inline function to access the payload data associated * with an event data. */ static INLINE void * VMCIEventDataPayload(VMCI_EventData *evData) // IN: { return (void *)((char *)evData + sizeof *evData); } /* * Define the different VMCI_EVENT payload data types here. All structs must * be a multiple of 8 bytes, and fields must be aligned to their natural * alignment. */ typedef struct VMCIEventPayload_Context { VMCIId contextID; /* 4 bytes. */ uint32 _pad; } VMCIEventPayload_Context; typedef struct VMCIEventPayload_QP { VMCIHandle handle; /* QueuePair handle. */ VMCIId peerId; /* Context id of attaching/detaching VM. */ uint32 _pad; } VMCIEventPayload_QP; /* * We define the following struct to get the size of the maximum event data * the hypervisor may send to the guest. If adding a new event payload type * above, add it to the following struct too (inside the union). */ typedef struct VMCIEventData_Max { VMCI_EventData eventData; union { VMCIEventPayload_Context contextPayload; VMCIEventPayload_QP qpPayload; } evDataPayload; } VMCIEventData_Max; /* * Struct used for VMCI_EVENT_SUBSCRIBE/UNSUBSCRIBE and VMCI_EVENT_HANDLER * messages. Struct size is 32 bytes. All fields in struct are aligned to * their natural alignment. */ typedef struct VMCIEventMsg { VMCIDatagram hdr; VMCI_EventData eventData; /* Has event type and payload. */ /* * Payload gets put here. */ } VMCIEventMsg; /* * We use the following inline function to access the payload data associated * with an event message. */ static INLINE void * VMCIEventMsgPayload(VMCIEventMsg *eMsg) // IN: { return VMCIEventDataPayload(&eMsg->eventData); } /* Flags for VMCI QueuePair API. */ #define VMCI_QPFLAG_ATTACH_ONLY 0x1 /* Fail alloc if QP not created by peer. */ #define VMCI_QPFLAG_LOCAL 0x2 /* Only allow attaches from local context. */ #define VMCI_QPFLAG_NONBLOCK 0x4 /* Host won't block when guest is quiesced. */ /* For asymmetric queuepairs, update as new flags are added. */ #define VMCI_QP_ASYMM VMCI_QPFLAG_NONBLOCK #define VMCI_QP_ASYMM_PEER (VMCI_QPFLAG_ATTACH_ONLY | VMCI_QP_ASYMM) /* Update the following (bitwise OR flags) while adding new flags. */ #define VMCI_QP_ALL_FLAGS (VMCI_QPFLAG_ATTACH_ONLY | VMCI_QPFLAG_LOCAL | \ VMCI_QPFLAG_NONBLOCK) /* * Structs used for QueuePair alloc and detach messages. We align fields of * these structs to 64bit boundaries. */ typedef struct VMCIQueuePairAllocMsg { VMCIDatagram hdr; VMCIHandle handle; VMCIId peer; /* 32bit field. */ uint32 flags; uint64 produceSize; uint64 consumeSize; uint64 numPPNs; /* List of PPNs placed here. */ } VMCIQueuePairAllocMsg; typedef struct VMCIQueuePairDetachMsg { VMCIDatagram hdr; VMCIHandle handle; } VMCIQueuePairDetachMsg; #if defined __cplusplus } // extern "C" #endif #endif // _VMCI_CALL_DEFS_H_ vsock-only/shared/compat_interrupt.h 0000444 0000000 0000000 00000003573 13432725347 016716 0 ustar root root /********************************************************* * Copyright (C) 2003 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ #ifndef __COMPAT_INTERRUPT_H__ # define __COMPAT_INTERRUPT_H__ #include <linux/interrupt.h> #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 5, 69) /* * We cannot just define irqreturn_t, as some 2.4.x kernels have * typedef void irqreturn_t; for "increasing" backward compatibility. */ typedef void compat_irqreturn_t; #define COMPAT_IRQ_NONE #define COMPAT_IRQ_HANDLED #define COMPAT_IRQ_RETVAL(x) #else typedef irqreturn_t compat_irqreturn_t; #define COMPAT_IRQ_NONE IRQ_NONE #define COMPAT_IRQ_HANDLED IRQ_HANDLED #define COMPAT_IRQ_RETVAL(x) IRQ_RETVAL(x) #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 18) #define COMPAT_IRQF_DISABLED SA_INTERRUPT #define COMPAT_IRQF_SHARED SA_SHIRQ #else #define COMPAT_IRQF_DISABLED IRQF_DISABLED #define COMPAT_IRQF_SHARED IRQF_SHARED #endif #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 20) #define COMPAT_IRQ_HANDLER_ARGS(irq, devp) (int irq, void *devp, struct pt_regs *regs) #else #define COMPAT_IRQ_HANDLER_ARGS(irq, devp) (int irq, void *devp) #endif #endif /* __COMPAT_INTERRUPT_H__ */ vsock-only/COPYING 0000444 0000000 0000000 00000043103 13432725341 012716 0 ustar root root GNU GENERAL PUBLIC LICENSE Version 2, June 1991 Copyright (C) 1989, 1991 Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA Everyone is permitted to copy and distribute verbatim copies of this license document, but changing it is not allowed. Preamble The licenses for most software are designed to take away your freedom to share and change it. By contrast, the GNU General Public License is intended to guarantee your freedom to share and change free software--to make sure the software is free for all its users. This General Public License applies to most of the Free Software Foundation's software and to any other program whose authors commit to using it. (Some other Free Software Foundation software is covered by the GNU Lesser General Public License instead.) You can apply it to your programs, too. When we speak of free software, we are referring to freedom, not price. Our General Public Licenses are designed to make sure that you have the freedom to distribute copies of free software (and charge for this service if you wish), that you receive source code or can get it if you want it, that you can change the software or use pieces of it in new free programs; and that you know you can do these things. To protect your rights, we need to make restrictions that forbid anyone to deny you these rights or to ask you to surrender the rights. These restrictions translate to certain responsibilities for you if you distribute copies of the software, or if you modify it. For example, if you distribute copies of such a program, whether gratis or for a fee, you must give the recipients all the rights that you have. You must make sure that they, too, receive or can get the source code. And you must show them these terms so they know their rights. We protect your rights with two steps: (1) copyright the software, and (2) offer you this license which gives you legal permission to copy, distribute and/or modify the software. Also, for each author's protection and ours, we want to make certain that everyone understands that there is no warranty for this free software. If the software is modified by someone else and passed on, we want its recipients to know that what they have is not the original, so that any problems introduced by others will not reflect on the original authors' reputations. Finally, any free program is threatened constantly by software patents. We wish to avoid the danger that redistributors of a free program will individually obtain patent licenses, in effect making the program proprietary. To prevent this, we have made it clear that any patent must be licensed for everyone's free use or not licensed at all. The precise terms and conditions for copying, distribution and modification follow. GNU GENERAL PUBLIC LICENSE TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION 0. This License applies to any program or other work which contains a notice placed by the copyright holder saying it may be distributed under the terms of this General Public License. The "Program", below, refers to any such program or work, and a "work based on the Program" means either the Program or any derivative work under copyright law: that is to say, a work containing the Program or a portion of it, either verbatim or with modifications and/or translated into another language. (Hereinafter, translation is included without limitation in the term "modification".) Each licensee is addressed as "you". Activities other than copying, distribution and modification are not covered by this License; they are outside its scope. The act of running the Program is not restricted, and the output from the Program is covered only if its contents constitute a work based on the Program (independent of having been made by running the Program). Whether that is true depends on what the Program does. 1. You may copy and distribute verbatim copies of the Program's source code as you receive it, in any medium, provided that you conspicuously and appropriately publish on each copy an appropriate copyright notice and disclaimer of warranty; keep intact all the notices that refer to this License and to the absence of any warranty; and give any other recipients of the Program a copy of this License along with the Program. You may charge a fee for the physical act of transferring a copy, and you may at your option offer warranty protection in exchange for a fee. 2. You may modify your copy or copies of the Program or any portion of it, thus forming a work based on the Program, and copy and distribute such modifications or work under the terms of Section 1 above, provided that you also meet all of these conditions: a) You must cause the modified files to carry prominent notices stating that you changed the files and the date of any change. b) You must cause any work that you distribute or publish, that in whole or in part contains or is derived from the Program or any part thereof, to be licensed as a whole at no charge to all third parties under the terms of this License. c) If the modified program normally reads commands interactively when run, you must cause it, when started running for such interactive use in the most ordinary way, to print or display an announcement including an appropriate copyright notice and a notice that there is no warranty (or else, saying that you provide a warranty) and that users may redistribute the program under these conditions, and telling the user how to view a copy of this License. (Exception: if the Program itself is interactive but does not normally print such an announcement, your work based on the Program is not required to print an announcement.) These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it. Thus, it is not the intent of this section to claim rights or contest your rights to work written entirely by you; rather, the intent is to exercise the right to control the distribution of derivative or collective works based on the Program. In addition, mere aggregation of another work not based on the Program with the Program (or with a work based on the Program) on a volume of a storage or distribution medium does not bring the other work under the scope of this License. 3. You may copy and distribute the Program (or a work based on it, under Section 2) in object code or executable form under the terms of Sections 1 and 2 above provided that you also do one of the following: a) Accompany it with the complete corresponding machine-readable source code, which must be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, b) Accompany it with a written offer, valid for at least three years, to give any third party, for a charge no more than your cost of physically performing source distribution, a complete machine-readable copy of the corresponding source code, to be distributed under the terms of Sections 1 and 2 above on a medium customarily used for software interchange; or, c) Accompany it with the information you received as to the offer to distribute corresponding source code. (This alternative is allowed only for noncommercial distribution and only if you received the program in object code or executable form with such an offer, in accord with Subsection b above.) The source code for a work means the preferred form of the work for making modifications to it. For an executable work, complete source code means all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable. However, as a special exception, the source code distributed need not include anything that is normally distributed (in either source or binary form) with the major components (compiler, kernel, and so on) of the operating system on which the executable runs, unless that component itself accompanies the executable. If distribution of executable or object code is made by offering access to copy from a designated place, then offering equivalent access to copy the source code from the same place counts as distribution of the source code, even though third parties are not compelled to copy the source along with the object code. 4. You may not copy, modify, sublicense, or distribute the Program except as expressly provided under this License. Any attempt otherwise to copy, modify, sublicense or distribute the Program is void, and will automatically terminate your rights under this License. However, parties who have received copies, or rights, from you under this License will not have their licenses terminated so long as such parties remain in full compliance. 5. You are not required to accept this License, since you have not signed it. However, nothing else grants you permission to modify or distribute the Program or its derivative works. These actions are prohibited by law if you do not accept this License. Therefore, by modifying or distributing the Program (or any work based on the Program), you indicate your acceptance of this License to do so, and all its terms and conditions for copying, distributing or modifying the Program or works based on it. 6. Each time you redistribute the Program (or any work based on the Program), the recipient automatically receives a license from the original licensor to copy, distribute or modify the Program subject to these terms and conditions. You may not impose any further restrictions on the recipients' exercise of the rights granted herein. You are not responsible for enforcing compliance by third parties to this License. 7. If, as a consequence of a court judgment or allegation of patent infringement or for any other reason (not limited to patent issues), conditions are imposed on you (whether by court order, agreement or otherwise) that contradict the conditions of this License, they do not excuse you from the conditions of this License. If you cannot distribute so as to satisfy simultaneously your obligations under this License and any other pertinent obligations, then as a consequence you may not distribute the Program at all. For example, if a patent license would not permit royalty-free redistribution of the Program by all those who receive copies directly or indirectly through you, then the only way you could satisfy both it and this License would be to refrain entirely from distribution of the Program. If any portion of this section is held invalid or unenforceable under any particular circumstance, the balance of the section is intended to apply and the section as a whole is intended to apply in other circumstances. It is not the purpose of this section to induce you to infringe any patents or other property right claims or to contest validity of any such claims; this section has the sole purpose of protecting the integrity of the free software distribution system, which is implemented by public license practices. Many people have made generous contributions to the wide range of software distributed through that system in reliance on consistent application of that system; it is up to the author/donor to decide if he or she is willing to distribute software through any other system and a licensee cannot impose that choice. This section is intended to make thoroughly clear what is believed to be a consequence of the rest of this License. 8. If the distribution and/or use of the Program is restricted in certain countries either by patents or by copyrighted interfaces, the original copyright holder who places the Program under this License may add an explicit geographical distribution limitation excluding those countries, so that distribution is permitted only in or among countries not thus excluded. In such case, this License incorporates the limitation as if written in the body of this License. 9. The Free Software Foundation may publish revised and/or new versions of the General Public License from time to time. Such new versions will be similar in spirit to the present version, but may differ in detail to address new problems or concerns. Each version is given a distinguishing version number. If the Program specifies a version number of this License which applies to it and "any later version", you have the option of following the terms and conditions either of that version or of any later version published by the Free Software Foundation. If the Program does not specify a version number of this License, you may choose any version ever published by the Free Software Foundation. 10. If you wish to incorporate parts of the Program into other free programs whose distribution conditions are different, write to the author to ask for permission. For software which is copyrighted by the Free Software Foundation, write to the Free Software Foundation; we sometimes make exceptions for this. Our decision will be guided by the two goals of preserving the free status of all derivatives of our free software and of promoting the sharing and reuse of software generally. NO WARRANTY 11. BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES PROVIDE THE PROGRAM "AS IS" WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING, REPAIR OR CORRECTION. 12. IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES, INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES. END OF TERMS AND CONDITIONS How to Apply These Terms to Your New Programs If you develop a new program, and you want it to be of the greatest possible use to the public, the best way to achieve this is to make it free software which everyone can redistribute and change under these terms. To do so, attach the following notices to the program. It is safest to attach them to the start of each source file to most effectively convey the exclusion of warranty; and each file should have at least the "copyright" line and a pointer to where the full notice is found. <one line to give the program's name and a brief idea of what it does.> Copyright (C) <year> <name of author> This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation; either version 2 of the License, or (at your option) any later version. This program is distributed in the hope that it will be useful, but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details. You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA. Also add information on how to contact you by electronic and paper mail. If the program is interactive, make it output a short notice like this when it starts in an interactive mode: Gnomovision version 69, Copyright (C) year name of author Gnomovision comes with ABSOLUTELY NO WARRANTY; for details type `show w'. This is free software, and you are welcome to redistribute it under certain conditions; type `show c' for details. The hypothetical commands `show w' and `show c' should show the appropriate parts of the General Public License. Of course, the commands you use may be called something other than `show w' and `show c'; they could even be mouse-clicks or menu items--whatever suits your program. You should also get your employer (if you work as a programmer) or your school, if any, to sign a "copyright disclaimer" for the program, if necessary. Here is a sample; alter the names: Yoyodyne, Inc., hereby disclaims all copyright interest in the program `Gnomovision' (which makes passes at compilers) written by James Hacker. <signature of Ty Coon>, 1 April 1989 Ty Coon, President of Vice This General Public License does not permit incorporating your program into proprietary programs. If your program is a subroutine library, you may consider it more useful to permit linking proprietary applications with the library. If this is what you want to do, use the GNU Lesser General Public License instead of this License. vsock-only/README 0000444 0000000 0000000 00000000755 13432725350 012551 0 ustar root root This files in this directory and its subdirectories are the kernel module for the VMware VSockets module. In order to build, make certain the Makefile is correct and then just type make from this directory. A copy of the module will be left in driver-<kernel version>/vsock-<kernel-version> (e.g. driver-up-2.4.20/vsock-up-2.4.20) for 2.4 series kernels and in ../vsock.o for 2.6 series kernels. If you have any problems or questions, send mail to support@vmware.com vsock-only/linux/ 0000755 0000000 0000000 00000000000 13432726375 013033 5 ustar root root vsock-only/linux/vmciKernelAPI.h 0000444 0000000 0000000 00000002451 13432725350 015625 0 ustar root root /********************************************************* * Copyright (C) 2010,2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmciKernelAPI.h -- * * Kernel API (current) exported from the VMCI host and guest drivers. */ #ifndef __VMCI_KERNELAPI_H__ #define __VMCI_KERNELAPI_H__ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" /* With this file you always get the latest version. */ #include "vmciKernelAPI1.h" #include "vmciKernelAPI2.h" #include "vmciKernelAPI3.h" #endif /* !__VMCI_KERNELAPI_H__ */ vsock-only/linux/af_vsock.h 0000444 0000000 0000000 00000020745 13432725350 014775 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * af_vsock.h -- * * Definitions for Linux VSockets module. */ #ifndef __AF_VSOCK_H__ #define __AF_VSOCK_H__ #include "vsockCommon.h" #include "vsockPacket.h" #include "compat_workqueue.h" #include "vmciKernelAPI.h" #include "notify.h" #ifdef VMX86_DEVEL extern int LOGLEVEL_THRESHOLD; #define LOG(level, args) ((void) (LOGLEVEL_THRESHOLD >= (level) ? (Log args) : 0)) #else #define LOG(level, args) #endif # define vsock_sk(__sk) ((VSockVmciSock *)__sk) # define sk_vsock(__vsk) (&(__vsk)->sk) typedef struct VSockVmciSock { /* sk must be the first member. */ struct sock sk; struct sockaddr_vm localAddr; struct sockaddr_vm remoteAddr; /* Links for the global tables of bound and connected sockets. */ struct list_head boundTable; struct list_head connectedTable; /* * Accessed without the socket lock held. This means it can never be * modified outsided of socket create or destruct. */ Bool trusted; Bool cachedPeerAllowDgram; /* Dgram communication allowed to cached peer? */ VMCIId cachedPeer; /* Context ID of last dgram destination check. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 5, 0) kuid_t owner; #else uid_t owner; #endif VMCIHandle dgHandle; /* For SOCK_DGRAM only. */ /* Rest are SOCK_STREAM only. */ VMCIHandle qpHandle; VMCIQPair *qpair; uint64 produceSize; uint64 consumeSize; uint64 queuePairSize; uint64 queuePairMinSize; uint64 queuePairMaxSize; long connectTimeout; VSockVmciNotify notify; VSockVmciNotifyOps *notifyOps; VMCIId attachSubId; VMCIId detachSubId; /* Listening socket that this came from. */ struct sock *listener; /* * Used for pending list and accept queue during connection handshake. The * listening socket is the head for both lists. Sockets created for * connection requests are placed in the pending list until they are * connected, at which point they are put in the accept queue list so they * can be accepted in accept(). If accept() cannot accept the connection, * it is marked as rejected so the cleanup function knows to clean up the * socket. */ struct list_head pendingLinks; struct list_head acceptQueue; Bool rejected; compat_delayed_work dwork; uint32 peerShutdown; Bool sentRequest; Bool ignoreConnectingRst; } VSockVmciSock; int VSockVmciSendControlPktBH(struct sockaddr_vm *src, struct sockaddr_vm *dst, VSockPacketType type, uint64 size, uint64 mode, VSockWaitingInfo *wait, VMCIHandle handle); int VSockVmciReplyControlPktFast(VSockPacket *pkt, VSockPacketType type, uint64 size, uint64 mode, VSockWaitingInfo *wait, VMCIHandle handle); int VSockVmciSendControlPkt(struct sock *sk, VSockPacketType type, uint64 size, uint64 mode, VSockWaitingInfo *wait, VSockProtoVersion version, VMCIHandle handle); int64 VSockVmciStreamHasData(VSockVmciSock *vsk); int64 VSockVmciStreamHasSpace(VSockVmciSock *vsk); #define VSOCK_SEND_RESET_BH(_dst, _src, _pkt) \ ((_pkt)->type == VSOCK_PACKET_TYPE_RST) ? \ 0 : \ VSockVmciSendControlPktBH(_dst, _src, VSOCK_PACKET_TYPE_RST, 0, \ 0, NULL, VMCI_INVALID_HANDLE) #define VSOCK_SEND_INVALID_BH(_dst, _src) \ VSockVmciSendControlPktBH(_dst, _src, VSOCK_PACKET_TYPE_INVALID, 0, \ 0, NULL, VMCI_INVALID_HANDLE) #define VSOCK_SEND_WROTE_BH(_dst, _src) \ VSockVmciSendControlPktBH(_dst, _src, VSOCK_PACKET_TYPE_WROTE, 0, \ 0, NULL, VMCI_INVALID_HANDLE) #define VSOCK_SEND_READ_BH(_dst, _src) \ VSockVmciSendControlPktBH(_dst, _src, VSOCK_PACKET_TYPE_READ, 0, \ 0, NULL, VMCI_INVALID_HANDLE) #define VSOCK_SEND_RESET(_sk, _pkt) \ ((_pkt)->type == VSOCK_PACKET_TYPE_RST) ? \ 0 : \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_RST, \ 0, 0, NULL, VSOCK_PROTO_INVALID, \ VMCI_INVALID_HANDLE) #define VSOCK_SEND_NEGOTIATE(_sk, _size) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_NEGOTIATE, \ _size, 0, NULL, VSOCK_PROTO_INVALID, \ VMCI_INVALID_HANDLE) #define VSOCK_SEND_NEGOTIATE2(_sk, _size, signalProto) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_NEGOTIATE2, \ _size, 0, NULL, signalProto, \ VMCI_INVALID_HANDLE) #define VSOCK_SEND_QP_OFFER(_sk, _handle) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_OFFER, \ 0, 0, NULL, VSOCK_PROTO_INVALID, _handle) #define VSOCK_SEND_CONN_REQUEST(_sk, _size) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_REQUEST, \ _size, 0, NULL, VSOCK_PROTO_INVALID, \ VMCI_INVALID_HANDLE) #define VSOCK_SEND_CONN_REQUEST2(_sk, _size, signalProto) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_REQUEST2, \ _size, 0, NULL, signalProto, \ VMCI_INVALID_HANDLE) #define VSOCK_SEND_ATTACH(_sk, _handle) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_ATTACH, \ 0, 0, NULL, VSOCK_PROTO_INVALID, _handle) #define VSOCK_SEND_WROTE(_sk) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_WROTE, \ 0, 0, NULL, VSOCK_PROTO_INVALID, \ VMCI_INVALID_HANDLE) #define VSOCK_SEND_READ(_sk) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_READ, \ 0, 0, NULL, VSOCK_PROTO_INVALID, \ VMCI_INVALID_HANDLE) #define VSOCK_SEND_SHUTDOWN(_sk, _mode) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_SHUTDOWN, \ 0, _mode, NULL, VSOCK_PROTO_INVALID, \ VMCI_INVALID_HANDLE) #define VSOCK_SEND_WAITING_WRITE(_sk, _waitInfo) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_WAITING_WRITE, \ 0, 0, _waitInfo, VSOCK_PROTO_INVALID, \ VMCI_INVALID_HANDLE) #define VSOCK_SEND_WAITING_READ(_sk, _waitInfo) \ VSockVmciSendControlPkt(_sk, VSOCK_PACKET_TYPE_WAITING_READ, \ 0, 0, _waitInfo, VSOCK_PROTO_INVALID, \ VMCI_INVALID_HANDLE) #define VSOCK_REPLY_RESET(_pkt) \ VSockVmciReplyControlPktFast(_pkt, VSOCK_PACKET_TYPE_RST, \ 0, 0, NULL, VMCI_INVALID_HANDLE) #endif /* __AF_VSOCK_H__ */ vsock-only/linux/vsockAddr.c 0000444 0000000 0000000 00000024230 13432725346 015113 0 ustar root root /********************************************************* * Copyright (C) 2007-2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vsockAddr.c -- * * VSockets address implementation. */ /* * These includes come before vsockCommon.h to ensure that VMware's ASSERT * macro is used instead of Linux's irda.h definition. */ #if defined(__linux__) && !defined(VMKERNEL) # if defined(__KERNEL__) # include "driver-config.h" # include <linux/socket.h> # include "compat_sock.h" # else # include <string.h> # include <errno.h> # endif #elif defined(VMKERNEL) # include "vm_libc.h" # include "return_status.h" #elif defined(__APPLE__) # include <sys/errno.h> #endif #include "vsockCommon.h" /* *----------------------------------------------------------------------------- * * VSockAddr_Init -- * * Initialize the given address with the given context id and port. This * will clear the address, set the correct family, and add the given * values. * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ void VSockAddr_Init(struct sockaddr_vm *addr, // OUT uint32 cid, // IN uint32 port) // IN { ASSERT(addr); VSockAddr_InitNoFamily(addr, cid, port); addr->svm_family = VMCISockGetAFValueInt(); VSOCK_ADDR_ASSERT(addr); } /* *----------------------------------------------------------------------------- * * VSockAddr_InitNoFamily -- * * Initialize the given address with the given context id and port. This * will clear the address and add the given values, but not set the * family. Note that this is needed because in some places we don't want * to re-register the address family in the Linux kernel and all we need * is to check the context id and port. * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ void VSockAddr_InitNoFamily(struct sockaddr_vm *addr, // OUT uint32 cid, // IN uint32 port) // IN { ASSERT(addr); memset(addr, 0, sizeof *addr); #if defined(__APPLE__) addr->svm_len = sizeof *addr; #endif addr->svm_cid = cid; addr->svm_port = port; VSOCK_ADDR_NOFAMILY_ASSERT(addr); } /* *----------------------------------------------------------------------------- * * VSockAddr_Validate -- * * Try to validate the given address. The address must not be null and * must have the correct address family. Any reserved fields must be * zero. * * Results: * 0 on success, VSOCK_EFAULT if the address is null, VSOCK_EAFNOSUPPORT if the * address is of the wrong family, and VSOCK_EINVAL if the reserved fields are * not zero. * * Side effects: * None. * *----------------------------------------------------------------------------- */ int32 VSockAddr_Validate(const struct sockaddr_vm *addr) // IN { int32 err; if (NULL == addr) { err = VSOCK_EFAULT; goto exit; } if (VMCISockGetAFValueInt() != addr->svm_family) { err = VSOCK_EAFNOSUPPORT; goto exit; } if (0 != addr->svm_zero[0]) { err = VSOCK_EINVAL; goto exit; } err = 0; exit: return sockerr2err(err); } /* *----------------------------------------------------------------------------- * * VSockAddr_ValidateNoFamily -- * * Try to validate the given address. The address must not be null and * any reserved fields must be zero, but the address family is not * checked. Note that this is needed because in some places we don't want * to re-register the address family with the Linux kernel. * * Also note that we duplicate the code from _Validate() since we want to * retain the ordering or the error return values. * * Results: * 0 on success, VSOCK_EFAULT if the address is null and VSOCK_EINVAL if the reserved * fields are not zero. * * Side effects: * None. * *----------------------------------------------------------------------------- */ int32 VSockAddr_ValidateNoFamily(const struct sockaddr_vm *addr) // IN { int32 err; if (NULL == addr) { err = VSOCK_EFAULT; goto exit; } if (0 != addr->svm_zero[0]) { err = VSOCK_EINVAL; goto exit; } err = 0; exit: return sockerr2err(err); } /* *---------------------------------------------------------------------------- * * VSockAddr_Bound -- * * Determines whether the provided address is bound. * * Results: * TRUE if the address structure is bound, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool VSockAddr_Bound(struct sockaddr_vm *addr) // IN: socket address to check { ASSERT(addr); return addr->svm_port != VMADDR_PORT_ANY; } /* *---------------------------------------------------------------------------- * * VSockAddr_Unbind -- * * Unbind the given addresss. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ void VSockAddr_Unbind(struct sockaddr_vm *addr) // IN { VSockAddr_Init(addr, VMADDR_CID_ANY, VMADDR_PORT_ANY); } /* *---------------------------------------------------------------------------- * * VSockAddr_EqualsAddr -- * * Determine if the given addresses are equal. * * Results: * TRUE if the addresses are equal, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool VSockAddr_EqualsAddr(struct sockaddr_vm *addr, // IN struct sockaddr_vm *other) // IN { /* * XXX We don't ASSERT on the family here since this is used on the receive * path in Linux and we don't want to re-register the address family * unnecessarily. */ VSOCK_ADDR_NOFAMILY_ASSERT(addr); VSOCK_ADDR_NOFAMILY_ASSERT(other); return (addr->svm_cid == other->svm_cid && addr->svm_port == other->svm_port); } /* *---------------------------------------------------------------------------- * * VSockAddr_EqualsHandlePort -- * * Determines if the given address matches the given handle and port. * * Results: * TRUE if the address matches the handle and port, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool VSockAddr_EqualsHandlePort(struct sockaddr_vm *addr, // IN VMCIHandle handle, // IN uint32 port) // IN { VSOCK_ADDR_ASSERT(addr); return (addr->svm_cid == VMCI_HANDLE_TO_CONTEXT_ID(handle) && addr->svm_port == port); } /* *----------------------------------------------------------------------------- * * VSockAddr_Cast -- * * Try to cast the given generic address to a VM address. The given * length must match that of a VM address and the address must be valid. * The "outAddr" parameter contains the address if successful. * * Results: * 0 on success, VSOCK_EFAULT if the length is too small. See * VSockAddr_Validate() for other possible return codes. * * Side effects: * None. * *----------------------------------------------------------------------------- */ int32 VSockAddr_Cast(const struct sockaddr *addr, // IN int32 len, // IN struct sockaddr_vm **outAddr) // OUT { int32 err; ASSERT(outAddr); if (len < sizeof **outAddr) { err = VSOCK_EFAULT; goto exit; } *outAddr = (struct sockaddr_vm *) addr; err = VSockAddr_Validate(*outAddr); exit: return sockerr2err(err); } /* *---------------------------------------------------------------------------- * * VSockAddr_SocketContextStream -- * * Determines whether the provided context id represents a context that * contains a stream socket endpoints. * * Results: * TRUE if the context does have socket endpoints, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool VSockAddr_SocketContextStream(uint32 cid) // IN { uint32 i; VMCIId nonSocketContexts[] = { VMCI_WELL_KNOWN_CONTEXT_ID, }; ASSERT_ON_COMPILE(sizeof cid == sizeof *nonSocketContexts); for (i = 0; i < ARRAYSIZE(nonSocketContexts); i++) { if (cid == nonSocketContexts[i]) { return FALSE; } } return TRUE; } /* *---------------------------------------------------------------------------- * * VSockAddr_SocketContextDgram -- * * Determines whether the provided <context id, resource id> represent * a protected datagram endpoint. * * Results: * TRUE if the context does have socket endpoints, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool VSockAddr_SocketContextDgram(uint32 cid, // IN uint32 rid) // IN { if (cid == VMCI_HYPERVISOR_CONTEXT_ID) { /* * Registrations of PBRPC Servers do not modify VMX/Hypervisor state and * are allowed. */ if (rid == VMCI_UNITY_PBRPC_REGISTER) { return TRUE; } else { return FALSE; } } return TRUE; } vsock-only/linux/vmci_sockets_packet.h 0000444 0000000 0000000 00000011112 13432725350 017206 0 ustar root root /********************************************************* * Copyright (C) 2012,2014 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmci_sockets_packet.h -- * * Definition of vSockets packet format, constants, and types. */ #ifndef _VMCI_SOCKETS_PACKET_H_ #define _VMCI_SOCKETS_PACKET_H_ #include "vmci_defs.h" #include "vmci_call_defs.h" /* * STREAM control packets. */ /* If the packet format changes in a release then this should change too. */ #define VSOCK_PACKET_VERSION 1 /* The resource ID on which control packets are sent. */ #define VSOCK_PACKET_RID 1 /* * Assert that the given packet is valid. * We check that the two original reserved fields equal zero because the * version of the common code that shipped with ESX 4.0 and WS 6.5 did so and * will return a RST packet if they aren't set that way. For newer packet * types added after that release we don't do this. */ #define VSOCK_PACKET_ASSERT(_p) \ do { \ ASSERT((_p)); \ ASSERT((_p)->type < VSOCK_PACKET_TYPE_MAX); \ if ((_p)->type < VSOCK_PACKET_TYPE_REQUEST2) { \ ASSERT(0 == (_p)->proto); \ ASSERT(0 == (_p)->_reserved2); \ } \ } while(0) typedef enum VSockPacketType { VSOCK_PACKET_TYPE_INVALID = 0, // Invalid type. VSOCK_PACKET_TYPE_REQUEST, // Connection request (WR/WW/READ/WRITE) VSOCK_PACKET_TYPE_NEGOTIATE, // Connection negotiate. VSOCK_PACKET_TYPE_OFFER, // Connection offer queue pair. VSOCK_PACKET_TYPE_ATTACH, // Connection attach. VSOCK_PACKET_TYPE_WROTE, // Wrote data to queue pair. VSOCK_PACKET_TYPE_READ, // Read data from queue pair. VSOCK_PACKET_TYPE_RST, // Reset. VSOCK_PACKET_TYPE_SHUTDOWN, // Shutdown the connection. VSOCK_PACKET_TYPE_WAITING_WRITE, // Notify peer we are waiting to write. VSOCK_PACKET_TYPE_WAITING_READ, // Notify peer we are waiting to read. VSOCK_PACKET_TYPE_REQUEST2, // Connection request (new proto flags) VSOCK_PACKET_TYPE_NEGOTIATE2, // Connection request (new proto flags) VSOCK_PACKET_TYPE_MAX // Last message. } VSockPacketType; typedef uint16 VSockProtoVersion; #define VSOCK_PROTO_INVALID 0 // Invalid protocol version. #define VSOCK_PROTO_PKT_ON_NOTIFY (1 << 0) // Queuepair inspection proto. #define VSOCK_PROTO_ALL_SUPPORTED (VSOCK_PROTO_PKT_ON_NOTIFY) typedef struct VSockWaitingInfo { uint64 generation; // Generation of the queue. uint64 offset; // Offset within the queue. } VSockWaitingInfo; /* * Control packet type for STREAM sockets. DGRAMs have no control packets * nor special packet header for data packets, they are just raw VMCI DGRAM * messages. For STREAMs, control packets are sent over the control channel * while data is written and read directly from queue pairs with no packet * format. */ typedef struct VSockPacket { VMCIDatagram dg; // Datagram header. uint8 version; // Version. uint8 type; // Type of message. VSockProtoVersion proto; // Supported proto versions in CONNECT2 and // NEGOTIATE2. 0 otherwise. uint32 srcPort; // Source port. uint32 dstPort; // Destination port. uint32 _reserved2; // Reserved. union { uint64 size; // Size of queue pair for request/negotiation. uint64 mode; // Mode of shutdown for shutdown. VMCIHandle handle; // Queue pair handle once size negotiated. VSockWaitingInfo wait; // Information provided for wait notifications. } u; } VSockPacket; /* * Size assertions. */ MY_ASSERTS(VSockSeqPacketAsserts, ASSERT_ON_COMPILE(sizeof (VSockPacket) == 56); ) #endif // _VMCI_SOCKETS_PACKET_H_ vsock-only/linux/vmciKernelAPI1.h 0000444 0000000 0000000 00000020110 13432725350 015676 0 ustar root root /********************************************************* * Copyright (C) 2010,2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmciKernelAPI1.h -- * * Kernel API (v1) exported from the VMCI host and guest drivers. */ #ifndef __VMCI_KERNELAPI_1_H__ #define __VMCI_KERNELAPI_1_H__ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" #include "vmci_defs.h" #include "vmci_call_defs.h" #if defined __cplusplus extern "C" { #endif /* VMCI module namespace on vmkernel. */ #define MOD_VMCI_NAMESPACE "com.vmware.vmci" /* Define version 1. */ #undef VMCI_KERNEL_API_VERSION #define VMCI_KERNEL_API_VERSION_1 1 #define VMCI_KERNEL_API_VERSION VMCI_KERNEL_API_VERSION_1 /* Macros to operate on the driver version number. */ #define VMCI_MAJOR_VERSION(v) (((v) >> 16) & 0xffff) #define VMCI_MINOR_VERSION(v) ((v) & 0xffff) #if defined(_WIN32) /* Path to callback object in object manager, for Windows only. */ #define VMCI_CALLBACK_OBJECT_PATH L"\\Callback\\VMCIDetachCB" #endif // _WIN32 /* VMCI Device Usage API. */ #if defined(__linux__) && !defined(VMKERNEL) #define vmci_device_get(_a, _b, _c, _d) 1 #define vmci_device_release(_x) #else // !linux typedef void (VMCI_DeviceShutdownFn)(void *deviceRegistration, void *userData); Bool vmci_device_get(uint32 *apiVersion, VMCI_DeviceShutdownFn *deviceShutdownCB, void *userData, void **deviceRegistration); void vmci_device_release(void *deviceRegistration); #endif // !linux #if defined(_WIN32) /* Called when the client is unloading, for Windows only. */ void vmci_exit(void); #endif // _WIN32 /* VMCI Datagram API. */ int vmci_datagram_create_handle(uint32 resourceId, uint32 flags, VMCIDatagramRecvCB recvCB, void *clientData, VMCIHandle *outHandle); int vmci_datagram_create_handle_priv(uint32 resourceID, uint32 flags, VMCIPrivilegeFlags privFlags, VMCIDatagramRecvCB recvCB, void *clientData, VMCIHandle *outHandle); int vmci_datagram_destroy_handle(VMCIHandle handle); int vmci_datagram_send(VMCIDatagram *msg); /* VMCI Utility API. */ VMCIId vmci_get_context_id(void); #if defined(__linux__) && !defined(VMKERNEL) /* Returned value is a bool, 0 for false, 1 for true. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 5, 0) int vmci_is_context_owner(VMCIId contextID, kuid_t uid); #else int vmci_is_context_owner(VMCIId contextID, uid_t uid); #endif #else // !linux || VMKERNEL /* Returned value is a VMCI error code. */ int vmci_is_context_owner(VMCIId contextID, void *hostUser); #endif // !linux || VMKERNEL uint32 vmci_version(void); int vmci_cid_2_host_vm_id(VMCIId contextID, void *hostVmID, size_t hostVmIDLen); /* VMCI Event API. */ typedef void (*VMCI_EventCB)(VMCIId subID, VMCI_EventData *ed, void *clientData); int vmci_event_subscribe(VMCI_Event event, #if !defined(__linux__) && !defined(__FreeBSD__) || defined(VMKERNEL) uint32 flags, #endif // !linux && !FreeBSD || VMKERNEL VMCI_EventCB callback, void *callbackData, VMCIId *subID); int vmci_event_unsubscribe(VMCIId subID); /* VMCI Context API */ VMCIPrivilegeFlags vmci_context_get_priv_flags(VMCIId contextID); /* VMCI Queue Pair API. */ typedef struct VMCIQPair VMCIQPair; int vmci_qpair_alloc(VMCIQPair **qpair, VMCIHandle *handle, uint64 produceQSize, uint64 consumeQSize, VMCIId peer, uint32 flags, VMCIPrivilegeFlags privFlags); int vmci_qpair_detach(VMCIQPair **qpair); int vmci_qpair_get_produce_indexes(const VMCIQPair *qpair, uint64 *producerTail, uint64 *consumerHead); int vmci_qpair_get_consume_indexes(const VMCIQPair *qpair, uint64 *consumerTail, uint64 *producerHead); int64 vmci_qpair_produce_free_space(const VMCIQPair *qpair); int64 vmci_qpair_produce_buf_ready(const VMCIQPair *qpair); int64 vmci_qpair_consume_free_space(const VMCIQPair *qpair); int64 vmci_qpair_consume_buf_ready(const VMCIQPair *qpair); ssize_t vmci_qpair_enqueue(VMCIQPair *qpair, const void *buf, size_t bufSize, int mode); ssize_t vmci_qpair_dequeue(VMCIQPair *qpair, void *buf, size_t bufSize, int mode); ssize_t vmci_qpair_peek(VMCIQPair *qpair, void *buf, size_t bufSize, int mode); #if (defined(__APPLE__) && !defined (VMX86_TOOLS)) || \ (defined(__linux__) && defined(__KERNEL__)) || \ (defined(_WIN32) && defined(WINNT_DDK)) /* * Environments that support struct iovec */ ssize_t vmci_qpair_enquev(VMCIQPair *qpair, void *iov, size_t iovSize, int mode); ssize_t vmci_qpair_dequev(VMCIQPair *qpair, void *iov, size_t iovSize, int mode); ssize_t vmci_qpair_peekv(VMCIQPair *qpair, void *iov, size_t iovSize, int mode); #endif /* Systems that support struct iovec */ /* Typedefs for all of the above, used by the IOCTLs and the kernel library. */ typedef void (VMCI_DeviceReleaseFct)(void *); typedef int (VMCIDatagram_CreateHndFct)(VMCIId, uint32, VMCIDatagramRecvCB, void *, VMCIHandle *); typedef int (VMCIDatagram_CreateHndPrivFct)(VMCIId, uint32, VMCIPrivilegeFlags, VMCIDatagramRecvCB, void *, VMCIHandle *); typedef int (VMCIDatagram_DestroyHndFct)(VMCIHandle); typedef int (VMCIDatagram_SendFct)(VMCIDatagram *); typedef VMCIId (VMCI_GetContextIDFct)(void); typedef uint32 (VMCI_VersionFct)(void); typedef int (VMCI_ContextID2HostVmIDFct)(VMCIId, void *, size_t); typedef int (VMCI_IsContextOwnerFct)(VMCIId, void *); typedef int (VMCIEvent_SubscribeFct)(VMCI_Event, uint32, VMCI_EventCB, void *, VMCIId *); typedef int (VMCIEvent_UnsubscribeFct)(VMCIId); typedef VMCIPrivilegeFlags (VMCIContext_GetPrivFlagsFct)(VMCIId); typedef int (VMCIQPair_AllocFct)(VMCIQPair **, VMCIHandle *, uint64, uint64, VMCIId, uint32, VMCIPrivilegeFlags); typedef int (VMCIQPair_DetachFct)(VMCIQPair **); typedef int (VMCIQPair_GetProduceIndexesFct)(const VMCIQPair *, uint64 *, uint64 *); typedef int (VMCIQPair_GetConsumeIndexesFct)(const VMCIQPair *, uint64 *, uint64 *); typedef int64 (VMCIQPair_ProduceFreeSpaceFct)(const VMCIQPair *); typedef int64 (VMCIQPair_ProduceBufReadyFct)(const VMCIQPair *); typedef int64 (VMCIQPair_ConsumeFreeSpaceFct)(const VMCIQPair *); typedef int64 (VMCIQPair_ConsumeBufReadyFct)(const VMCIQPair *); typedef ssize_t (VMCIQPair_EnqueueFct)(VMCIQPair *, const void *, size_t, int); typedef ssize_t (VMCIQPair_DequeueFct)(VMCIQPair *, void *, size_t, int); typedef ssize_t (VMCIQPair_PeekFct)(VMCIQPair *, void *, size_t, int); typedef ssize_t (VMCIQPair_EnqueueVFct)(VMCIQPair *qpair, void *, size_t, int); typedef ssize_t (VMCIQPair_DequeueVFct)(VMCIQPair *qpair, void *, size_t, int); typedef ssize_t (VMCIQPair_PeekVFct)(VMCIQPair *qpair, void *, size_t, int); #if defined __cplusplus } // extern "C" #endif #endif /* !__VMCI_KERNELAPI_1_H__ */ vsock-only/linux/vsockVmci.h 0000444 0000000 0000000 00000007451 13432725350 015145 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vsockVmci.h -- * * VSockets VMCI constants, types and functions. */ #ifndef _VSOCK_VMCI_H_ #define _VSOCK_VMCI_H_ extern VMCIId VMCI_GetContextID(void); /* *----------------------------------------------------------------------------- * * VSockVmci_IsLocal -- * * Determine if the given handle points to the local context. * * Results: * TRUE if the given handle is for the local context, FALSE otherwise. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE Bool VSockVmci_IsLocal(VMCIHandle handle) // IN { return VMCI_GetContextID() == VMCI_HANDLE_TO_CONTEXT_ID(handle); } /* *---------------------------------------------------------------------------- * * VSockVmci_ErrorToVSockError -- * * Converts from a VMCI error code to a VSock error code. * * Results: * Appropriate error code. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static INLINE int32 VSockVmci_ErrorToVSockError(int32 vmciError) // IN { int32 err; switch (vmciError) { case VMCI_ERROR_NO_MEM: #if defined(_WIN32) err = ENOBUFS; #else // _WIN32 err = ENOMEM; #endif // _WIN32 break; case VMCI_ERROR_DUPLICATE_ENTRY: err = EADDRINUSE; break; case VMCI_ERROR_NO_ACCESS: err = EPERM; break; case VMCI_ERROR_NO_RESOURCES: err = ENOBUFS; break; case VMCI_ERROR_INVALID_RESOURCE: err = EHOSTUNREACH; break; case VMCI_ERROR_MODULE_NOT_LOADED: err = ESYSNOTREADY; break; case VMCI_ERROR_NO_HANDLE: err = ENETUNREACH; break; case VMCI_ERROR_INVALID_ARGS: default: err = EINVAL; } return sockerr2err(err); } /* *---------------------------------------------------------------------------- * * VSockVmci_GetVmciObjSocket -- * * Get a socket from a VMCI object, but only if the object is of the * appropriate type. * * Results: * A socket if the object is of the correct type, NULL otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static INLINE void * VSockVmci_GetVmciObjSocket(VMCIObj *obj) // IN { ASSERT(obj); if (NULL != obj->ptr && VMCIOBJ_SOCKET == obj->type) { return obj->ptr; } return NULL; } /* *---------------------------------------------------------------------------- * * VSockVmci_SetVmciObjSocket -- * * Set the socket in a VMCI object. This will also set the type * accordingly. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static INLINE void VSockVmci_SetVmciObjSocket(VMCIObj *obj, // OUT void *s) // IN { ASSERT(obj); ASSERT(s); obj->ptr = s; obj->type = VMCIOBJ_SOCKET; } #endif // _VSOCK_VMCI_H_ vsock-only/linux/util.c 0000444 0000000 0000000 00000046575 13432725350 014163 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * util.c -- * * Utility functions for Linux VSocket module. */ #include "driver-config.h" #include <linux/list.h> #include <linux/socket.h> #include "compat_sock.h" #include "af_vsock.h" #include "util.h" struct list_head vsockBindTable[VSOCK_HASH_SIZE + 1]; struct list_head vsockConnectedTable[VSOCK_HASH_SIZE]; DEFINE_SPINLOCK(vsockTableLock); /* *---------------------------------------------------------------------------- * * VSockVmciLogPkt -- * * Logs the provided packet. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ void VSockVmciLogPkt(char const *function, // IN uint32 line, // IN VSockPacket *pkt) // IN { char buf[256]; char *cur = buf; int left = sizeof buf; int written = 0; char *typeStrings[] = { [VSOCK_PACKET_TYPE_INVALID] = "INVALID", [VSOCK_PACKET_TYPE_REQUEST] = "REQUEST", [VSOCK_PACKET_TYPE_NEGOTIATE] = "NEGOTIATE", [VSOCK_PACKET_TYPE_OFFER] = "OFFER", [VSOCK_PACKET_TYPE_ATTACH] = "ATTACH", [VSOCK_PACKET_TYPE_WROTE] = "WROTE", [VSOCK_PACKET_TYPE_READ] = "READ", [VSOCK_PACKET_TYPE_RST] = "RST", [VSOCK_PACKET_TYPE_SHUTDOWN] = "SHUTDOWN", [VSOCK_PACKET_TYPE_WAITING_WRITE] = "WAITING_WRITE", [VSOCK_PACKET_TYPE_WAITING_READ] = "WAITING_READ", [VSOCK_PACKET_TYPE_REQUEST2] = "REQUEST2", [VSOCK_PACKET_TYPE_NEGOTIATE2] = "NEGOTIATE2", }; written = snprintf(cur, left, "PKT: %u:%u -> %u:%u", VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.src), pkt->srcPort, VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.dst), pkt->dstPort); if (written >= left) { goto error; } left -= written; cur += written; switch (pkt->type) { case VSOCK_PACKET_TYPE_REQUEST: case VSOCK_PACKET_TYPE_NEGOTIATE: written = snprintf(cur, left, ", %s, size = %"FMT64"u", typeStrings[pkt->type], pkt->u.size); break; case VSOCK_PACKET_TYPE_OFFER: case VSOCK_PACKET_TYPE_ATTACH: written = snprintf(cur, left, ", %s, handle = %u:%u", typeStrings[pkt->type], VMCI_HANDLE_TO_CONTEXT_ID(pkt->u.handle), VMCI_HANDLE_TO_RESOURCE_ID(pkt->u.handle)); break; case VSOCK_PACKET_TYPE_WROTE: case VSOCK_PACKET_TYPE_READ: case VSOCK_PACKET_TYPE_RST: written = snprintf(cur, left, ", %s", typeStrings[pkt->type]); break; case VSOCK_PACKET_TYPE_SHUTDOWN: { Bool recv; Bool send; recv = pkt->u.mode & RCV_SHUTDOWN; send = pkt->u.mode & SEND_SHUTDOWN; written = snprintf(cur, left, ", %s, mode = %c%c", typeStrings[pkt->type], recv ? 'R' : ' ', send ? 'S' : ' '); } break; case VSOCK_PACKET_TYPE_WAITING_WRITE: case VSOCK_PACKET_TYPE_WAITING_READ: written = snprintf(cur, left, ", %s, generation = %"FMT64"u, " "offset = %"FMT64"u", typeStrings[pkt->type], pkt->u.wait.generation, pkt->u.wait.offset); break; case VSOCK_PACKET_TYPE_REQUEST2: case VSOCK_PACKET_TYPE_NEGOTIATE2: written = snprintf(cur, left, ", %s, size = %"FMT64"u, " "proto = %u", typeStrings[pkt->type], pkt->u.size, pkt->proto); break; default: written = snprintf(cur, left, ", unrecognized type"); } if (written >= left) { goto error; } left -= written; cur += written; written = snprintf(cur, left, " [%s:%u]\n", function, line); if (written >= left) { goto error; } LOG(8, ("%s", buf)); return; error: LOG(8, ("could not log packet\n")); } /* *---------------------------------------------------------------------------- * * VSockVmciInitTables -- * * Initializes the tables used for socket lookup. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ void VSockVmciInitTables(void) { uint32 i; for (i = 0; i < ARRAYSIZE(vsockBindTable); i++) { INIT_LIST_HEAD(&vsockBindTable[i]); } for (i = 0; i < ARRAYSIZE(vsockConnectedTable); i++) { INIT_LIST_HEAD(&vsockConnectedTable[i]); } } /* *---------------------------------------------------------------------------- * * __VSockVmciInsertBound -- * * Inserts socket into the bound table. * * Note that this assumes any necessary locks are held. * * Results: * None. * * Side effects: * The reference count for sk is incremented. * *---------------------------------------------------------------------------- */ void __VSockVmciInsertBound(struct list_head *list, // IN struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(list); ASSERT(sk); vsk = vsock_sk(sk); sock_hold(sk); list_add(&vsk->boundTable, list); } /* *---------------------------------------------------------------------------- * * __VSockVmciInsertConnected -- * * Inserts socket into the connected table. * * Note that this assumes any necessary locks are held. * * Results: * None. * * Side effects: * The reference count for sk is incremented. * *---------------------------------------------------------------------------- */ void __VSockVmciInsertConnected(struct list_head *list, // IN struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(list); ASSERT(sk); vsk = vsock_sk(sk); sock_hold(sk); list_add(&vsk->connectedTable, list); } /* *---------------------------------------------------------------------------- * * __VSockVmciRemoveBound -- * * Removes socket from the bound table. * * Note that this assumes any necessary locks are held. * * Results: * None. * * Side effects: * The reference count for sk is decremented. * *---------------------------------------------------------------------------- */ void __VSockVmciRemoveBound(struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(sk); ASSERT(__VSockVmciInBoundTable(sk)); vsk = vsock_sk(sk); list_del_init(&vsk->boundTable); sock_put(sk); } /* *---------------------------------------------------------------------------- * * __VSockVmciRemoveConnected -- * * Removes socket from the connected table. * * Note that this assumes any necessary locks are held. * * Results: * None. * * Side effects: * The reference count for sk is decremented. * *---------------------------------------------------------------------------- */ void __VSockVmciRemoveConnected(struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(sk); ASSERT(__VSockVmciInConnectedTable(sk)); vsk = vsock_sk(sk); list_del_init(&vsk->connectedTable); sock_put(sk); } /* *---------------------------------------------------------------------------- * * __VSockVmciFindBoundSocket -- * * Finds the socket corresponding to the provided address in the bound * sockets hash table. * * Note that this assumes any necessary locks are held. * * Results: * The sock structure if found, NULL if not found. * * Side effects: * None. * *---------------------------------------------------------------------------- */ struct sock * __VSockVmciFindBoundSocket(struct sockaddr_vm *addr) // IN { VSockVmciSock *vsk; struct sock *sk; ASSERT(addr); list_for_each_entry(vsk, vsockBoundSockets(addr), boundTable) { if (addr->svm_port == vsk->localAddr.svm_port) { sk = sk_vsock(vsk); /* We only store stream sockets in the bound table. */ ASSERT(sk->sk_socket ? sk->sk_socket->type == SOCK_STREAM : 1); goto found; } } sk = NULL; found: return sk; } /* *---------------------------------------------------------------------------- * * __VSockVmciFindConnectedSocket -- * * Finds the socket corresponding to the provided addresses in the connected * sockets hash table. * * Note that this assumes any necessary locks are held. * * Results: * The sock structure if found, NULL if not found. * * Side effects: * None. * *---------------------------------------------------------------------------- */ struct sock * __VSockVmciFindConnectedSocket(struct sockaddr_vm *src, // IN struct sockaddr_vm *dst) // IN { VSockVmciSock *vsk; struct sock *sk; ASSERT(src); ASSERT(dst); list_for_each_entry(vsk, vsockConnectedSockets(src, dst), connectedTable) { if (VSockAddr_EqualsAddr(src, &vsk->remoteAddr) && dst->svm_port == vsk->localAddr.svm_port) { sk = sk_vsock(vsk); goto found; } } sk = NULL; found: return sk; } /* *---------------------------------------------------------------------------- * * __VSockVmciInBoundTable -- * * Determines whether the provided socket is in the bound table. * * Results: * TRUE is socket is in bound table, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool __VSockVmciInBoundTable(struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(sk); vsk = vsock_sk(sk); return !list_empty(&vsk->boundTable); } /* *---------------------------------------------------------------------------- * * __VSockVmciInConnectedTable -- * * Determines whether the provided socket is in the connected table. * * Results: * TRUE is socket is in connected table, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool __VSockVmciInConnectedTable(struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(sk); vsk = vsock_sk(sk); return !list_empty(&vsk->connectedTable); } /* *---------------------------------------------------------------------------- * * VSockVmciGetPending -- * * Retrieves a pending connection that matches the addresses specified in * the provided packet. * * Assumes the socket lock is held for listener. * * Results: * Socket of the pending connection on success, NULL if not found. * * Side effects: * A reference is held on the socket until the release function is called. * *---------------------------------------------------------------------------- */ struct sock * VSockVmciGetPending(struct sock *listener, // IN: listening socket VSockPacket *pkt) // IN: incoming packet { VSockVmciSock *vlistener; VSockVmciSock *vpending; struct sock *pending; struct sockaddr_vm src; ASSERT(listener); ASSERT(pkt); VSockAddr_Init(&src, VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.src), pkt->srcPort); vlistener = vsock_sk(listener); list_for_each_entry(vpending, &vlistener->pendingLinks, pendingLinks) { if (VSockAddr_EqualsAddr(&src, &vpending->remoteAddr) && pkt->dstPort == vpending->localAddr.svm_port) { pending = sk_vsock(vpending); sock_hold(pending); goto found; } } pending = NULL; found: return pending; } /* *---------------------------------------------------------------------------- * * VSockVmciReleasePending -- * * Releases the reference on a socket previously obtained by a call to * VSockVmciGetPending(). * * Results: * None. * * Side effects: * The socket may be freed if this was the last reference. * *---------------------------------------------------------------------------- */ void VSockVmciReleasePending(struct sock *pending) // IN: pending connection { ASSERT(pending); sock_put(pending); } /* *---------------------------------------------------------------------------- * * VSockVmciAddPending -- * * Adds a pending connection on listener's pending list. * * Assumes the socket lock is held for listener. * Assumes the socket lock is held for pending. * * Results: * None. * * Side effects: * The reference count of the sockets is incremented. * *---------------------------------------------------------------------------- */ void VSockVmciAddPending(struct sock *listener, // IN: listening socket struct sock *pending) // IN: pending connection { VSockVmciSock *vlistener; VSockVmciSock *vpending; ASSERT(listener); ASSERT(pending); vlistener = vsock_sk(listener); vpending = vsock_sk(pending); sock_hold(pending); sock_hold(listener); list_add_tail(&vpending->pendingLinks, &vlistener->pendingLinks); } /* *---------------------------------------------------------------------------- * * VSockVmciRemovePending -- * * Removes a pending connection from the listener's pending list. * * Assumes the socket lock is held for listener. * Assumes the socket lock is held for pending. * * Results: * None. * * Side effects: * The reference count of the sockets is decremented. * *---------------------------------------------------------------------------- */ void VSockVmciRemovePending(struct sock *listener, // IN: listening socket struct sock *pending) // IN: pending connection { VSockVmciSock *vpending; ASSERT(listener); ASSERT(pending); vpending = vsock_sk(pending); list_del_init(&vpending->pendingLinks); sock_put(listener); sock_put(pending); } /* *---------------------------------------------------------------------------- * * VSockVmciEnqueueAccept -- * * Enqueues the connected socket on the listening socket's accepting * queue. * * Assumes the socket lock is held for listener. * Assumes the socket lock is held for connected. * * Results: * None. * * Side effects: * The sockets' reference counts are incremented. * *---------------------------------------------------------------------------- */ void VSockVmciEnqueueAccept(struct sock *listener, // IN: listening socket struct sock *connected) // IN: connected socket { VSockVmciSock *vlistener; VSockVmciSock *vconnected; ASSERT(listener); ASSERT(connected); vlistener = vsock_sk(listener); vconnected = vsock_sk(connected); sock_hold(connected); sock_hold(listener); list_add_tail(&vconnected->acceptQueue, &vlistener->acceptQueue); } /* *---------------------------------------------------------------------------- * * VSockVmciDequeueAccept -- * * Dequeues the next connected socket from the listening socket's accept * queue. * * Assumes the socket lock is held for listener. * * Note that the caller must call sock_put() on the returned socket once it * is done with the socket. * * Results: * The next socket from the queue, or NULL if the queue is empty. * * Side effects: * The reference count of the listener is decremented. * *---------------------------------------------------------------------------- */ struct sock * VSockVmciDequeueAccept(struct sock *listener) // IN: listening socket { VSockVmciSock *vlistener; VSockVmciSock *vconnected; ASSERT(listener); vlistener = vsock_sk(listener); if (list_empty(&vlistener->acceptQueue)) { return NULL; } vconnected = list_entry(vlistener->acceptQueue.next, VSockVmciSock, acceptQueue); ASSERT(vconnected); list_del_init(&vconnected->acceptQueue); sock_put(listener); /* * The caller will need a reference on the connected socket so we let it * call sock_put(). */ ASSERT(sk_vsock(vconnected)); return sk_vsock(vconnected); } /* *---------------------------------------------------------------------------- * * VSockVmciRemoveAccept -- * * Removes a socket from the accept queue of a listening socket. * * Assumes the socket lock is held for listener. * Assumes the socket lock is held for connected. * * Results: * None. * * Side effects: * The sockets' reference counts are decremented. * *---------------------------------------------------------------------------- */ void VSockVmciRemoveAccept(struct sock *listener, // IN: listening socket struct sock *connected) // IN: connected socket { VSockVmciSock *vconnected; ASSERT(listener); ASSERT(connected); if (!VSockVmciInAcceptQueue(connected)) { return; } vconnected = vsock_sk(connected); ASSERT(vconnected->listener == listener); list_del_init(&vconnected->acceptQueue); sock_put(listener); sock_put(connected); } /* *---------------------------------------------------------------------------- * * VSockVmciInAcceptQueue -- * * Determines whether a socket is on an accept queue. * * Assumes the socket lock is held for sk. * * Results: * TRUE if the socket is in an accept queue, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool VSockVmciInAcceptQueue(struct sock *sk) // IN: socket { ASSERT(sk); /* * If our accept queue isn't empty, it means we're linked into some listener * socket's accept queue. */ return !VSockVmciIsAcceptQueueEmpty(sk); } /* *---------------------------------------------------------------------------- * * VSockVmciIsAcceptQueueEmpty -- * * Determines whether the provided socket's accept queue is empty. * * Assumes the socket lock is held for sk. * * Results: * TRUE if the socket's accept queue is empty, FALSE otherwsise. * * Side effects: * None. * * *---------------------------------------------------------------------------- */ Bool VSockVmciIsAcceptQueueEmpty(struct sock *sk) // IN: socket { VSockVmciSock *vsk; ASSERT(sk); vsk = vsock_sk(sk); return list_empty(&vsk->acceptQueue); } /* *---------------------------------------------------------------------------- * * VSockVmciIsPending -- * * Determines whether a socket is pending. * * Assumes the socket lock is held for sk. * * Results: * TRUE if the socket is pending, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool VSockVmciIsPending(struct sock *sk) // IN: socket { VSockVmciSock *vsk; ASSERT(sk); vsk = vsock_sk(sk); return !list_empty(&vsk->pendingLinks); } vsock-only/linux/vmciKernelAPI2.h 0000444 0000000 0000000 00000004240 13432725350 015705 0 ustar root root /********************************************************* * Copyright (C) 2010 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmciKernelAPI2.h -- * * Kernel API (v2) exported from the VMCI host and guest drivers. */ #ifndef __VMCI_KERNELAPI_2_H__ #define __VMCI_KERNELAPI_2_H__ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" #include "vmciKernelAPI1.h" #if defined __cplusplus extern "C" { #endif /* Define version 2. */ #undef VMCI_KERNEL_API_VERSION #define VMCI_KERNEL_API_VERSION_2 2 #define VMCI_KERNEL_API_VERSION VMCI_KERNEL_API_VERSION_2 /* VMCI Doorbell API. */ #define VMCI_FLAG_DELAYED_CB 0x01 typedef void (*VMCICallback)(void *clientData); int vmci_doorbell_create(VMCIHandle *handle, uint32 flags, VMCIPrivilegeFlags privFlags, VMCICallback notifyCB, void *clientData); int vmci_doorbell_destroy(VMCIHandle handle); int vmci_doorbell_notify(VMCIHandle handle, VMCIPrivilegeFlags privFlags); /* Typedefs for all of the above, used by the IOCTLs and the kernel library. */ typedef int (VMCIDoorbell_CreateFct)(VMCIHandle *, uint32, VMCIPrivilegeFlags, VMCICallback, void *); typedef int (VMCIDoorbell_DestroyFct)(VMCIHandle); typedef int (VMCIDoorbell_NotifyFct)(VMCIHandle, VMCIPrivilegeFlags); #if defined __cplusplus } // extern "C" #endif #endif /* !__VMCI_KERNELAPI_2_H__ */ vsock-only/linux/vsock_version.h 0000444 0000000 0000000 00000002215 13432725350 016064 0 ustar root root /********************************************************* * Copyright (C) 2011-2015 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vsock_version.h -- * * Version definitions for the Linux vsock driver. */ #ifndef _VSOCK_VERSION_H_ #define _VSOCK_VERSION_H_ #define VSOCK_DRIVER_VERSION 9.8.1.0 #define VSOCK_DRIVER_VERSION_COMMAS 9,8,1,0 #define VSOCK_DRIVER_VERSION_STRING "9.8.1.0" #endif /* _VSOCK_VERSION_H_ */ vsock-only/linux/notify.h 0000444 0000000 0000000 00000011627 13432725350 014511 0 ustar root root /********************************************************* * Copyright (C) 2009-2014 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * notify.h -- * * Notify functions for Linux VSocket module. */ #ifndef __NOTIFY_H__ #define __NOTIFY_H__ #include "driver-config.h" #include "vsockCommon.h" #include "vsockPacket.h" /* Comment this out to compare with old protocol. */ #define VSOCK_OPTIMIZATION_WAITING_NOTIFY 1 #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) /* Comment this out to remove flow control for "new" protocol */ # define VSOCK_OPTIMIZATION_FLOW_CONTROL 1 #endif #define VSOCK_MAX_DGRAM_RESENDS 10 #define NOTIFYCALLRET(vsk, rv, mod_fn, args...) \ do { \ if (vsk->notifyOps && \ vsk->notifyOps->mod_fn != NULL) { \ rv = (vsk->notifyOps->mod_fn)(args); \ } else { \ rv = 0; \ } \ } while (0) #define NOTIFYCALL(vsk, mod_fn, args...) \ do { \ if (vsk->notifyOps && \ vsk->notifyOps->mod_fn != NULL) { \ (vsk->notifyOps->mod_fn)(args); \ } \ } while (0) typedef struct VSockVmciNotifyPkt { uint64 writeNotifyWindow; uint64 writeNotifyMinWindow; Bool peerWaitingRead; Bool peerWaitingWrite; Bool peerWaitingWriteDetected; Bool sentWaitingRead; Bool sentWaitingWrite; VSockWaitingInfo peerWaitingReadInfo; VSockWaitingInfo peerWaitingWriteInfo; uint64 produceQGeneration; uint64 consumeQGeneration; } VSockVmciNotifyPkt; typedef struct VSockVmciNotifyPktQState { uint64 writeNotifyWindow; uint64 writeNotifyMinWindow; Bool peerWaitingWrite; Bool peerWaitingWriteDetected; } VSockVmciNotifyPktQState; typedef union VSockVmciNotify { VSockVmciNotifyPkt pkt; VSockVmciNotifyPktQState pktQState; } VSockVmciNotify; typedef struct VSockVmciRecvNotifyData { uint64 consumeHead; uint64 produceTail; Bool notifyOnBlock; } VSockVmciRecvNotifyData; typedef struct VSockVmciSendNotifyData { uint64 consumeHead; uint64 produceTail; } VSockVmciSendNotifyData; /* Socket notification callbacks. */ typedef struct VSockVmciNotifyOps { void (*socketInit)(struct sock *sk); void (*socketDestruct)(struct sock *sk); int32 (*pollIn)(struct sock *sk, size_t target, Bool *dataReadyNow); int32 (*pollOut)(struct sock *sk, size_t target, Bool *spaceAvailNow); void (*handleNotifyPkt)(struct sock *sk, VSockPacket *pkt, Bool bottomHalf, struct sockaddr_vm *dst, struct sockaddr_vm *src, Bool *pktProcessed); int32 (*recvInit)(struct sock *sk, size_t target, VSockVmciRecvNotifyData *data); int32 (*recvPreBlock)(struct sock *sk, size_t target, VSockVmciRecvNotifyData *data); int32 (*recvPreDequeue)(struct sock *sk, size_t target, VSockVmciRecvNotifyData *data); int32 (*recvPostDequeue)(struct sock *sk, size_t target, ssize_t copied, Bool dataRead, VSockVmciRecvNotifyData *data); int32 (*sendInit)(struct sock *sk, VSockVmciSendNotifyData *data); int32 (*sendPreBlock)(struct sock *sk, VSockVmciSendNotifyData *data); int32 (*sendPreEnqueue)(struct sock *sk, VSockVmciSendNotifyData *data); int32 (*sendPostEnqueue)(struct sock *sk, ssize_t written, VSockVmciSendNotifyData *data); void (*processRequest)(struct sock *sk); void (*processNegotiate)(struct sock *sk); } VSockVmciNotifyOps; extern VSockVmciNotifyOps vSockVmciNotifyPktOps; extern VSockVmciNotifyOps vSockVmciNotifyPktQStateOps; #endif /* __NOTIFY_H__ */ vsock-only/linux/stats.c 0000444 0000000 0000000 00000002513 13432725350 014324 0 ustar root root /********************************************************* * Copyright (C) 2009 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * stats.c -- * * Linux stats for the VMCI Stream Sockets protocol. */ #include "driver-config.h" #include <linux/socket.h> #include "compat_sock.h" #include "af_vsock.h" #include "stats.h" #ifdef VSOCK_GATHER_STATISTICS uint64 vSockStatsCtlPktCount[VSOCK_PACKET_TYPE_MAX]; uint64 vSockStatsConsumeQueueHist[VSOCK_NUM_QUEUE_LEVEL_BUCKETS]; uint64 vSockStatsProduceQueueHist[VSOCK_NUM_QUEUE_LEVEL_BUCKETS]; Atomic_uint64 vSockStatsConsumeTotal; Atomic_uint64 vSockStatsProduceTotal; #endif vsock-only/linux/stats.h 0000444 0000000 0000000 00000017576 13432725350 014350 0 ustar root root /********************************************************* * Copyright (C) 2009 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * stats.h -- * * Stats functions for Linux vsock module. */ #ifndef __STATS_H__ #define __STATS_H__ #include "driver-config.h" #include "vm_basic_math.h" #include "vsockCommon.h" #include "vsockPacket.h" /* * Define VSOCK_GATHER_STATISTICS to turn on statistics gathering. * Currently this consists of 3 types of stats: * 1. The number of control datagram messages sent. * 2. The level of queuepair fullness (in 10% buckets) whenever data is * about to be enqueued or dequeued from the queuepair. * 3. The total number of bytes enqueued/dequeued. */ //#define VSOCK_GATHER_STATISTICS 1 #ifdef VSOCK_GATHER_STATISTICS #define VSOCK_NUM_QUEUE_LEVEL_BUCKETS 10 extern uint64 vSockStatsCtlPktCount[VSOCK_PACKET_TYPE_MAX]; extern uint64 vSockStatsConsumeQueueHist[VSOCK_NUM_QUEUE_LEVEL_BUCKETS]; extern uint64 vSockStatsProduceQueueHist[VSOCK_NUM_QUEUE_LEVEL_BUCKETS]; extern Atomic_uint64 vSockStatsConsumeTotal; extern Atomic_uint64 vSockStatsProduceTotal; #define VSOCK_STATS_STREAM_CONSUME_HIST(vsk) \ VSockVmciStatsUpdateQueueBucketCount((vsk)->qpair, \ (vsk)->consumeSize, \ VMCIQPair_ConsumeBufReady((vsk)->qpair), \ vSockStatsConsumeQueueHist) #define VSOCK_STATS_STREAM_PRODUCE_HIST(vsk) \ VSockVmciStatsUpdateQueueBucketCount((vsk)->qpair, \ (vsk)->produceSize, \ VMCIQPair_ProduceBufReady((vsk)->qpair), \ vSockStatsProduceQueueHist) #define VSOCK_STATS_CTLPKT_LOG(pktType) \ do { \ ++vSockStatsCtlPktCount[pktType]; \ } while (0) #define VSOCK_STATS_STREAM_CONSUME(bytes) \ Atomic_ReadAdd64(&vSockStatsConsumeTotal, bytes) #define VSOCK_STATS_STREAM_PRODUCE(bytes) \ Atomic_ReadAdd64(&vSockStatsProduceTotal, bytes) #define VSOCK_STATS_CTLPKT_DUMP_ALL() VSockVmciStatsCtlPktDumpAll() #define VSOCK_STATS_HIST_DUMP_ALL() VSockVmciStatsHistDumpAll() #define VSOCK_STATS_TOTALS_DUMP_ALL() VSockVmciStatsTotalsDumpAll() #define VSOCK_STATS_RESET() VSockVmciStatsReset() /* *---------------------------------------------------------------------------- * * VSockVmciStatsUpdateQueueBucketCount -- * * Given a queue, determine how much data is enqueued and add that to * the specified queue level statistic bucket. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static INLINE void VSockVmciStatsUpdateQueueBucketCount(VMCIQPair *qpair, // IN uint64 queueSize, // IN uint64 dataReady, // IN uint64 queueHist[]) // IN/OUT { uint64 bucket = 0; uint32 remainder = 0; ASSERT(qpair); ASSERT(queueHist); /* * We can't do 64 / 64 = 64 bit divides on linux because it requires a * libgcc which is not linked into the kernel module. Since this code is * only used by developers we just limit the queueSize to be less than * MAX_UINT for now. */ ASSERT(queueSize <= MAX_UINT32); Div643264(dataReady * 10, queueSize, &bucket, &remainder); ASSERT(bucket < VSOCK_NUM_QUEUE_LEVEL_BUCKETS); ++queueHist[bucket]; } /* *---------------------------------------------------------------------------- * * VSockVmciStatsCtlPktDumpAll -- * * Prints all stream control packet counts out to the console using * the appropriate platform logging. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static INLINE void VSockVmciStatsCtlPktDumpAll(void) { uint32 index; ASSERT_ON_COMPILE(VSOCK_PACKET_TYPE_MAX == ARRAYSIZE(vSockStatsCtlPktCount)); for (index = 0; index < ARRAYSIZE(vSockStatsCtlPktCount); index++) { Warning("Control packet count: Type = %u, Count = %"FMT64"u\n", index, vSockStatsCtlPktCount[index]); } } /* *---------------------------------------------------------------------------- * * VSockVmciStatsHistDumpAll -- * * Prints the produce and consume queue histograms to the console. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static INLINE void VSockVmciStatsHistDumpAll(void) { uint32 index; #define VSOCK_DUMP_HIST(strname, name) do { \ for (index = 0; index < ARRAYSIZE(name); index++) { \ Warning(strname " Bucket count %u = %"FMT64"u\n", \ index, name[index]); \ } \ } while (0) VSOCK_DUMP_HIST("Produce Queue", vSockStatsProduceQueueHist); VSOCK_DUMP_HIST("Consume Queue", vSockStatsConsumeQueueHist); #undef VSOCK_DUMP_HIST } /* *---------------------------------------------------------------------------- * * VSockVmciStatsTotalsDumpAll -- * * Prints the produce and consume totals. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static INLINE void VSockVmciStatsTotalsDumpAll(void) { Warning("Produced %"FMT64"u total bytes\n", Atomic_Read64(&vSockStatsProduceTotal)); Warning("Consumed %"FMT64"u total bytes\n", Atomic_Read64(&vSockStatsConsumeTotal)); } /* *---------------------------------------------------------------------------- * * VSockVmciStatsReset -- * * Reset all VSock statistics. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static INLINE void VSockVmciStatsReset(void) { uint32 index; #define VSOCK_RESET_ARRAY(name) do { \ for (index = 0; index < ARRAYSIZE(name); index++) { \ name[index] = 0; \ } \ } while (0) VSOCK_RESET_ARRAY(vSockStatsCtlPktCount); VSOCK_RESET_ARRAY(vSockStatsProduceQueueHist); VSOCK_RESET_ARRAY(vSockStatsConsumeQueueHist); #undef VSOCK_RESET_ARRAY Atomic_Write64(&vSockStatsConsumeTotal, 0); Atomic_Write64(&vSockStatsProduceTotal, 0); } #else #define VSOCK_STATS_STREAM_CONSUME_HIST(vsk) #define VSOCK_STATS_STREAM_PRODUCE_HIST(vsk) #define VSOCK_STATS_STREAM_PRODUCE(bytes) #define VSOCK_STATS_STREAM_CONSUME(bytes) #define VSOCK_STATS_CTLPKT_LOG(pktType) #define VSOCK_STATS_CTLPKT_DUMP_ALL() #define VSOCK_STATS_HIST_DUMP_ALL() #define VSOCK_STATS_TOTALS_DUMP_ALL() #define VSOCK_STATS_RESET() #endif // VSOCK_GATHER_STATISTICS #endif // __STATS_H__ vsock-only/linux/util.h 0000444 0000000 0000000 00000025527 13432725350 014162 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * util.h -- * * Utility functions for Linux VSocket module. */ #ifndef __UTIL_H__ #define __UTIL_H__ #include "driver-config.h" #include "compat_sock.h" #include "compat_spinlock.h" #include "vsockCommon.h" #include "vsockPacket.h" /* * Each bound VSocket is stored in the bind hash table and each connected * VSocket is stored in the connected hash table. * * Unbound sockets are all put on the same list attached to the end of the hash * table (vsockUnboundSockets). Bound sockets are added to the hash table in * the bucket that their local address hashes to (vsockBoundSockets(addr) * represents the list that addr hashes to). * * Specifically, we initialize the vsockBindTable array to a size of * VSOCK_HASH_SIZE + 1 so that vsockBindTable[0] through * vsockBindTable[VSOCK_HASH_SIZE - 1] are for bound sockets and * vsockBindTable[VSOCK_HASH_SIZE] is for unbound sockets. The hash function * mods with VSOCK_HASH_SIZE - 1 to ensure this. */ #define VSOCK_HASH_SIZE 251 #define LAST_RESERVED_PORT 1023 #define MAX_PORT_RETRIES 24 extern struct list_head vsockBindTable[VSOCK_HASH_SIZE + 1]; extern struct list_head vsockConnectedTable[VSOCK_HASH_SIZE]; extern spinlock_t vsockTableLock; #define VSOCK_HASH(addr) ((addr)->svm_port % (VSOCK_HASH_SIZE - 1)) #define vsockBoundSockets(addr) (&vsockBindTable[VSOCK_HASH(addr)]) #define vsockUnboundSockets (&vsockBindTable[VSOCK_HASH_SIZE]) /* XXX This can probably be implemented in a better way. */ #define VSOCK_CONN_HASH(src, dst) \ (((src)->svm_cid ^ (dst)->svm_port) % (VSOCK_HASH_SIZE - 1)) #define vsockConnectedSockets(src, dst) \ (&vsockConnectedTable[VSOCK_CONN_HASH(src, dst)]) #define vsockConnectedSocketsVsk(vsk) \ vsockConnectedSockets(&(vsk)->remoteAddr, &(vsk)->localAddr) /* * Prototypes. */ void VSockVmciLogPkt(char const *function, uint32 line, VSockPacket *pkt); void VSockVmciInitTables(void); void __VSockVmciInsertBound(struct list_head *list, struct sock *sk); void __VSockVmciInsertConnected(struct list_head *list, struct sock *sk); void __VSockVmciRemoveBound(struct sock *sk); void __VSockVmciRemoveConnected(struct sock *sk); struct sock *__VSockVmciFindBoundSocket(struct sockaddr_vm *addr); struct sock *__VSockVmciFindConnectedSocket(struct sockaddr_vm *src, struct sockaddr_vm *dst); Bool __VSockVmciInBoundTable(struct sock *sk); Bool __VSockVmciInConnectedTable(struct sock *sk); struct sock *VSockVmciGetPending(struct sock *listener, VSockPacket *pkt); void VSockVmciReleasePending(struct sock *pending); void VSockVmciAddPending(struct sock *listener, struct sock *pending); void VSockVmciRemovePending(struct sock *listener, struct sock *pending); void VSockVmciEnqueueAccept(struct sock *listener, struct sock *connected); struct sock *VSockVmciDequeueAccept(struct sock *listener); void VSockVmciRemoveAccept(struct sock *listener, struct sock *connected); Bool VSockVmciInAcceptQueue(struct sock *sk); Bool VSockVmciIsAcceptQueueEmpty(struct sock *sk); Bool VSockVmciIsPending(struct sock *sk); static INLINE void VSockVmciInsertBound(struct list_head *list, struct sock *sk); static INLINE void VSockVmciInsertConnected(struct list_head *list, struct sock *sk); static INLINE void VSockVmciRemoveBound(struct sock *sk); static INLINE void VSockVmciRemoveConnected(struct sock *sk); static INLINE struct sock *VSockVmciFindBoundSocket(struct sockaddr_vm *addr); static INLINE struct sock *VSockVmciFindConnectedSocket(struct sockaddr_vm *src, struct sockaddr_vm *dst); static INLINE Bool VSockVmciInBoundTable(struct sock *sk); static INLINE Bool VSockVmciInConnectedTable(struct sock *sk); /* *---------------------------------------------------------------------------- * * VSockVmciInsertBound -- * * Inserts socket into the bound table. * * Note that it is important to invoke the bottom-half versions of the * spinlock functions since these may be called from tasklets. * * Results: * None. * * Side effects: * vsockTableLock is acquired and released. * *---------------------------------------------------------------------------- */ static INLINE void VSockVmciInsertBound(struct list_head *list, // IN struct sock *sk) // IN { ASSERT(list); ASSERT(sk); spin_lock_bh(&vsockTableLock); __VSockVmciInsertBound(list, sk); spin_unlock_bh(&vsockTableLock); } /* *---------------------------------------------------------------------------- * * VSockVmciInsertConnected -- * * Inserts socket into the connected table. * * Note that it is important to invoke the bottom-half versions of the * spinlock functions since these may be called from tasklets. * * Results: * None. * * Side effects: * vsockTableLock is acquired and released. * *---------------------------------------------------------------------------- */ static INLINE void VSockVmciInsertConnected(struct list_head *list, // IN struct sock *sk) // IN { ASSERT(list); ASSERT(sk); spin_lock_bh(&vsockTableLock); __VSockVmciInsertConnected(list, sk); spin_unlock_bh(&vsockTableLock); } /* *---------------------------------------------------------------------------- * * VSockVmciRemoveBound -- * * Removes socket from the bound list. * * Note that it is important to invoke the bottom-half versions of the * spinlock functions since these may be called from tasklets. * * Results: * None. * * Side effects: * vsockTableLock is acquired and released. * *---------------------------------------------------------------------------- */ static INLINE void VSockVmciRemoveBound(struct sock *sk) // IN { ASSERT(sk); spin_lock_bh(&vsockTableLock); __VSockVmciRemoveBound(sk); spin_unlock_bh(&vsockTableLock); } /* *---------------------------------------------------------------------------- * * VSockVmciRemoveConnected -- * * Removes socket from the connected list. * * Note that it is important to invoke the bottom-half versions of the * spinlock functions since these may be called from tasklets. * * Results: * None. * * Side effects: * vsockTableLock is acquired and released. * *---------------------------------------------------------------------------- */ static INLINE void VSockVmciRemoveConnected(struct sock *sk) // IN { ASSERT(sk); spin_lock_bh(&vsockTableLock); __VSockVmciRemoveConnected(sk); spin_unlock_bh(&vsockTableLock); } /* *---------------------------------------------------------------------------- * * VSockVmciFindBoundSocket -- * * Finds the socket corresponding to the provided address in the bound * sockets hash table. * * Note that it is important to invoke the bottom-half versions of the * spinlock functions since these are called from tasklets. * * Results: * The sock structure if found, NULL on failure. * * Side effects: * vsockTableLock is acquired and released. * The socket's reference count is increased. * *---------------------------------------------------------------------------- */ static INLINE struct sock * VSockVmciFindBoundSocket(struct sockaddr_vm *addr) // IN { struct sock *sk; ASSERT(addr); spin_lock_bh(&vsockTableLock); sk = __VSockVmciFindBoundSocket(addr); if (sk) { sock_hold(sk); } spin_unlock_bh(&vsockTableLock); return sk; } /* *---------------------------------------------------------------------------- * * VSockVmciFindConnectedSocket -- * * Finds the socket corresponding to the provided address in the connected * sockets hash table. * * Note that it is important to invoke the bottom-half versions of the * spinlock functions since these are called from tasklets. * * Results: * The sock structure if found, NULL on failure. * * Side effects: * vsockTableLock is acquired and released. * The socket's reference count is increased. * *---------------------------------------------------------------------------- */ static INLINE struct sock * VSockVmciFindConnectedSocket(struct sockaddr_vm *src, // IN struct sockaddr_vm *dst) // IN { struct sock *sk; ASSERT(src); ASSERT(dst); spin_lock_bh(&vsockTableLock); sk = __VSockVmciFindConnectedSocket(src, dst); if (sk) { sock_hold(sk); } spin_unlock_bh(&vsockTableLock); return sk; } /* *---------------------------------------------------------------------------- * * VSockVmciInBoundTable -- * * Determines whether the provided socket is in the bound table. * * Note that it is important to invoke the bottom-half versions of the * spinlock functions since these may be called from tasklets. * * Results: * TRUE is socket is in bound table, FALSE otherwise. * * Side effects: * vsockTableLock is acquired and released. * *---------------------------------------------------------------------------- */ static INLINE Bool VSockVmciInBoundTable(struct sock *sk) // IN { Bool ret; ASSERT(sk); spin_lock_bh(&vsockTableLock); ret = __VSockVmciInBoundTable(sk); spin_unlock_bh(&vsockTableLock); return ret; } /* *---------------------------------------------------------------------------- * * VSockVmciInConnectedTable -- * * Determines whether the provided socket is in the connected table. * * Note that it is important to invoke the bottom-half versions of the * spinlock functions since these may be called from tasklets. * * Results: * TRUE is socket is in connected table, FALSE otherwise. * * Side effects: * vsockTableLock is acquired and released. * *---------------------------------------------------------------------------- */ static INLINE Bool VSockVmciInConnectedTable(struct sock *sk) // IN { Bool ret; ASSERT(sk); spin_lock_bh(&vsockTableLock); ret = __VSockVmciInConnectedTable(sk); spin_unlock_bh(&vsockTableLock); return ret; } #endif /* __UTIL_H__ */ vsock-only/linux/vmciKernelAPI3.h 0000444 0000000 0000000 00000003150 13432725350 015705 0 ustar root root /********************************************************* * Copyright (C) 2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmciKernelAPI3.h -- * * Kernel API (v3) exported from the VMCI host and guest drivers. */ #ifndef __VMCI_KERNELAPI_3_H__ #define __VMCI_KERNELAPI_3_H__ #define INCLUDE_ALLOW_MODULE #define INCLUDE_ALLOW_VMK_MODULE #define INCLUDE_ALLOW_VMKERNEL #include "includeCheck.h" #include "vmciKernelAPI2.h" #if defined __cplusplus extern "C" { #endif /* Define version 3. */ #undef VMCI_KERNEL_API_VERSION #define VMCI_KERNEL_API_VERSION_3 3 #define VMCI_KERNEL_API_VERSION VMCI_KERNEL_API_VERSION_3 /* VMCI Detach Cause API (only available in vmkernel). */ #define VMCI_DETACH_REGULAR 0 #define VMCI_DETACH_VMOTION 1 int vmci_qpair_get_detach_cause(VMCIQPair *qpair, uint8 *cause); #if defined __cplusplus } // extern "C" #endif #endif /* !__VMCI_KERNELAPI_3_H__ */ vsock-only/linux/notifyQState.c 0000444 0000000 0000000 00000046501 13432725350 015625 0 ustar root root /********************************************************* * Copyright (C) 2009-2014,2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * notifyQState.c -- * * Linux control notifications based on Queuepair state for the VMCI * Stream Sockets protocol. */ #include "driver-config.h" #include <linux/socket.h> #include "compat_sock.h" #include "notify.h" #include "af_vsock.h" #define PKT_FIELD(vsk, fieldName) \ (vsk)->notify.pktQState.fieldName /* *---------------------------------------------------------------------------- * * VSockVmciNotifyWaitingWrite -- * * Determines if the conditions have been met to notify a waiting writer. * * Results: * TRUE if a notification should be sent, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static Bool VSockVmciNotifyWaitingWrite(VSockVmciSock *vsk) // IN { Bool retval; uint64 notifyLimit; if (!PKT_FIELD(vsk, peerWaitingWrite)) { return FALSE; } /* * When the sender blocks, we take that as a sign that the sender * is faster than the receiver. To reduce the transmit rate of the * sender, we delay the sending of the read notification by * decreasing the writeNotifyWindow. The notification is delayed * until the number of bytes used in the queue drops below the * writeNotifyWindow. */ if (!PKT_FIELD(vsk, peerWaitingWriteDetected)) { PKT_FIELD(vsk, peerWaitingWriteDetected) = TRUE; if (PKT_FIELD(vsk, writeNotifyWindow) < PAGE_SIZE) { PKT_FIELD(vsk, writeNotifyWindow) = PKT_FIELD(vsk, writeNotifyMinWindow); } else { PKT_FIELD(vsk, writeNotifyWindow) -= PAGE_SIZE; if (PKT_FIELD(vsk, writeNotifyWindow) < PKT_FIELD(vsk, writeNotifyMinWindow)) { PKT_FIELD(vsk, writeNotifyWindow) = PKT_FIELD(vsk, writeNotifyMinWindow); } } } notifyLimit = vsk->consumeSize - PKT_FIELD(vsk, writeNotifyWindow); /* * The notifyLimit is used to delay notifications in the case where * flow control is enabled. Below the test is expressed in terms of * free space in the queue: * if freeSpace > ConsumeSize - writeNotifyWindow then notify * An alternate way of expressing this is to rewrite the expression * to use the data ready in the receive queue: * if writeNotifyWindow > bufferReady then notify * as freeSpace == ConsumeSize - bufferReady. */ retval = vmci_qpair_consume_free_space(vsk->qpair) > notifyLimit; if (retval) { /* * Once we notify the peer, we reset the detected flag so the * next wait will again cause a decrease in the window size. */ PKT_FIELD(vsk, peerWaitingWriteDetected) = FALSE; } return retval; } /* *---------------------------------------------------------------------------- * * VSockVmciHandleRead -- * * Handles an incoming read message. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciHandleRead(struct sock *sk, // IN VSockPacket *pkt, // IN: unused Bool bottomHalf, // IN: unused struct sockaddr_vm *dst, // IN: unused struct sockaddr_vm *src) // IN: unused { sk->sk_write_space(sk); } /* *---------------------------------------------------------------------------- * * VSockVmciHandleWrote -- * * Handles an incoming wrote message. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciHandleWrote(struct sock *sk, // IN VSockPacket *pkt, // IN: unused Bool bottomHalf, // IN: unused struct sockaddr_vm *dst, // IN: unused struct sockaddr_vm *src) // IN: unused { sk->sk_data_ready(sk, 0); } /* *---------------------------------------------------------------------------- * * VSockVmciBlockUpdateWriteWindow -- * * Updates the write window when we are blocking for data. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciBlockUpdateWriteWindow(struct sock *sk) // IN { VSockVmciSock *vsk; vsk = vsock_sk(sk); if (PKT_FIELD(vsk, writeNotifyWindow) < vsk->consumeSize) { PKT_FIELD(vsk, writeNotifyWindow) = MIN(PKT_FIELD(vsk, writeNotifyWindow) + PAGE_SIZE, vsk->consumeSize); } } /* *---------------------------------------------------------------------------- * * VSockVmciSendReadNotification -- * * Sends a read notification to this socket's peer. * * Results: * >= 0 if the datagram is sent successfully, negative error value * otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciSendReadNotification(struct sock *sk) // IN { VSockVmciSock *vsk; Bool sentRead; unsigned int retries; int err; ASSERT(sk); vsk = vsock_sk(sk); sentRead = FALSE; retries = 0; err = 0; if (VSockVmciNotifyWaitingWrite(vsk)) { /* * Notify the peer that we have read, retrying the send on failure up to our * maximum value. XXX For now we just log the failure, but later we should * schedule a work item to handle the resend until it succeeds. That would * require keeping track of work items in the vsk and cleaning them up upon * socket close. */ while (!(vsk->peerShutdown & RCV_SHUTDOWN) && !sentRead && retries < VSOCK_MAX_DGRAM_RESENDS) { err = VSOCK_SEND_READ(sk); if (err >= 0) { sentRead = TRUE; } retries++; } if (retries >= VSOCK_MAX_DGRAM_RESENDS && !sentRead) { Warning("unable to send read notification to peer for socket %p.\n", sk); } else { PKT_FIELD(vsk, peerWaitingWrite) = FALSE; } } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktSocketInit -- * * Function that is called after a socket is created and before any * notify ops are used. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciNotifyPktSocketInit(struct sock *sk) // IN { VSockVmciSock *vsk; vsk = vsock_sk(sk); PKT_FIELD(vsk, writeNotifyWindow) = PAGE_SIZE; PKT_FIELD(vsk, writeNotifyMinWindow) = PAGE_SIZE; PKT_FIELD(vsk, peerWaitingWrite) = FALSE; PKT_FIELD(vsk, peerWaitingWriteDetected) = FALSE; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktSocketDestruct -- * * Function that is called when the socket is being released. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciNotifyPktSocketDestruct(struct sock *sk) // IN { VSockVmciSock *vsk; vsk = vsock_sk(sk); PKT_FIELD(vsk, writeNotifyWindow) = PAGE_SIZE; PKT_FIELD(vsk, writeNotifyMinWindow) = PAGE_SIZE; PKT_FIELD(vsk, peerWaitingWrite) = FALSE; PKT_FIELD(vsk, peerWaitingWriteDetected) = FALSE; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktPollIn -- * * Called by the poll function to figure out if there is data to read * and to setup future notifications if needed. Only called on sockets * that aren't shutdown for recv. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktPollIn(struct sock *sk, // IN size_t target, // IN Bool *dataReadyNow) // IN { VSockVmciSock *vsk; ASSERT(sk); ASSERT(dataReadyNow); vsk = vsock_sk(sk); if (VSockVmciStreamHasData(vsk)) { *dataReadyNow = TRUE; } else { /* * We can't read right now because there is nothing in the queue. * Ask for notifications when there is something to read. */ if (sk->sk_state == SS_CONNECTED) { VSockVmciBlockUpdateWriteWindow(sk); } *dataReadyNow = FALSE; } return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktPollOut * * Called by the poll function to figure out if there is space to write * and to setup future notifications if needed. Only called on a * connected socket that isn't shutdown for send. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktPollOut(struct sock *sk, // IN size_t target, // IN Bool *spaceAvailNow) // IN { int64 produceQFreeSpace; VSockVmciSock *vsk; ASSERT(sk); ASSERT(spaceAvailNow); vsk = vsock_sk(sk); produceQFreeSpace = VSockVmciStreamHasSpace(vsk); if (produceQFreeSpace > 0) { *spaceAvailNow = TRUE; return 0; } else if (produceQFreeSpace == 0) { /* * This is a connected socket but we can't currently send data. Nothing * else to do. */ *spaceAvailNow = FALSE; } return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktRecvInit -- * * Called at the start of a stream recv call with the socket lock held. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktRecvInit(struct sock *sk, // IN size_t target, // IN VSockVmciRecvNotifyData *data) // IN { VSockVmciSock *vsk; ASSERT(sk); ASSERT(data); vsk = vsock_sk(sk); data->consumeHead = 0; data->produceTail = 0; data->notifyOnBlock = FALSE; if (PKT_FIELD(vsk, writeNotifyMinWindow) < target + 1) { ASSERT(target < vsk->consumeSize); PKT_FIELD(vsk, writeNotifyMinWindow) = target + 1; if (PKT_FIELD(vsk, writeNotifyWindow) < PKT_FIELD(vsk, writeNotifyMinWindow)) { /* * If the current window is smaller than the new minimal * window size, we need to reevaluate whether we need to * notify the sender. If the number of ready bytes are * smaller than the new window, we need to send a * notification to the sender before we block. */ PKT_FIELD(vsk, writeNotifyWindow) = PKT_FIELD(vsk, writeNotifyMinWindow); data->notifyOnBlock = TRUE; } } return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktRecvPreBlock -- * * Called right before a socket is about to block with the socket lock * held. The socket lock may have been released between the entry * function and the preblock call. * * Note: This function may be called multiple times before the post * block function is called. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktRecvPreBlock(struct sock *sk, // IN size_t target, // IN VSockVmciRecvNotifyData *data) // IN { int err; ASSERT(sk); ASSERT(data); err = 0; VSockVmciBlockUpdateWriteWindow(sk); if (data->notifyOnBlock) { err = VSockVmciSendReadNotification(sk); if (err < 0) { return err; } data->notifyOnBlock = FALSE; } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktRecvPostDequeue -- * * Called right after we dequeue / peek data from a socket. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktRecvPostDequeue(struct sock *sk, // IN size_t target, // IN ssize_t copied, // IN Bool dataRead, // IN VSockVmciRecvNotifyData *data) // IN { VSockVmciSock *vsk; int err; Bool wasFull = FALSE; uint64 freeSpace; ASSERT(sk); ASSERT(data); vsk = vsock_sk(sk); err = 0; if (dataRead) { SMP_RW_BARRIER_RW(); freeSpace = vmci_qpair_consume_free_space(vsk->qpair); wasFull = freeSpace == copied; if (wasFull) { PKT_FIELD(vsk, peerWaitingWrite) = TRUE; } err = VSockVmciSendReadNotification(sk); if (err < 0) { return err; } /* See the comment in VSockVmciNotifyPktSendPostEnqueue */ sk->sk_data_ready(sk, 0); } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktSendInit -- * * Called at the start of a stream send call with the socket lock held. * * Results: * 0 on success. A negative error code on failure. * * Side effects: * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktSendInit(struct sock *sk, // IN VSockVmciSendNotifyData *data) // IN { ASSERT(sk); ASSERT(data); data->consumeHead = 0; data->produceTail = 0; return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifySendPostEnqueue -- * * Called right after we enqueue data to a socket. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktSendPostEnqueue(struct sock *sk, // IN ssize_t written, // IN VSockVmciSendNotifyData *data) // IN { int err = 0; VSockVmciSock *vsk; Bool sentWrote = FALSE; Bool wasEmpty; int retries = 0; ASSERT(sk); ASSERT(data); vsk = vsock_sk(sk); SMP_RW_BARRIER_RW(); wasEmpty = (vmci_qpair_produce_buf_ready(vsk->qpair) == written); if (wasEmpty) { while (!(vsk->peerShutdown & RCV_SHUTDOWN) && !sentWrote && retries < VSOCK_MAX_DGRAM_RESENDS) { err = VSOCK_SEND_WROTE(sk); if (err >= 0) { sentWrote = TRUE; } retries++; } } if (retries >= VSOCK_MAX_DGRAM_RESENDS && !sentWrote) { Warning("unable to send wrote notification to peer for socket %p.\n", sk); return err; } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktHandlePkt * * Called when a notify packet is recieved for a socket in the connected * state. Note this might be called from a bottom half. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciNotifyPktHandlePkt(struct sock *sk, // IN VSockPacket *pkt, // IN Bool bottomHalf, // IN struct sockaddr_vm *dst, // IN struct sockaddr_vm *src, // IN Bool *pktProcessed) // In { Bool processed = FALSE; ASSERT(sk); ASSERT(pkt); switch (pkt->type) { case VSOCK_PACKET_TYPE_WROTE: VSockVmciHandleWrote(sk, pkt, bottomHalf, dst, src); processed = TRUE; break; case VSOCK_PACKET_TYPE_READ: VSockVmciHandleRead(sk, pkt, bottomHalf, dst, src); processed = TRUE; break; } if (pktProcessed) { *pktProcessed = processed; } } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktProcessRequest * * Called near the end of process request. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciNotifyPktProcessRequest(struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(sk); vsk = vsock_sk(sk); PKT_FIELD(vsk, writeNotifyWindow) = vsk->consumeSize; if (vsk->consumeSize < PKT_FIELD(vsk, writeNotifyMinWindow)) { PKT_FIELD(vsk, writeNotifyMinWindow) = vsk->consumeSize; } } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktProcessNegotiate * * Called near the end of process negotiate. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciNotifyPktProcessNegotiate(struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(sk); vsk = vsock_sk(sk); PKT_FIELD(vsk, writeNotifyWindow) = vsk->consumeSize; if (vsk->consumeSize < PKT_FIELD(vsk, writeNotifyMinWindow)) { PKT_FIELD(vsk, writeNotifyMinWindow) = vsk->consumeSize; } } /* Socket always on control packet based operations. */ VSockVmciNotifyOps vSockVmciNotifyPktQStateOps = { VSockVmciNotifyPktSocketInit, VSockVmciNotifyPktSocketDestruct, VSockVmciNotifyPktPollIn, VSockVmciNotifyPktPollOut, VSockVmciNotifyPktHandlePkt, VSockVmciNotifyPktRecvInit, VSockVmciNotifyPktRecvPreBlock, NULL, /* recvPreDequeue */ VSockVmciNotifyPktRecvPostDequeue, VSockVmciNotifyPktSendInit, NULL, /* sendPreBlock */ NULL, /* sendPreEnqueue */ VSockVmciNotifyPktSendPostEnqueue, VSockVmciNotifyPktProcessRequest, VSockVmciNotifyPktProcessNegotiate, }; vsock-only/linux/vmci_sockets.h 0000444 0000000 0000000 00000063565 13432725350 015702 0 ustar root root /********************************************************* * Copyright (C) 2007-2017 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmci_sockets.h -- * * vSockets public constants and types. */ #ifndef _VMCI_SOCKETS_H_ #define _VMCI_SOCKETS_H_ #if defined(_WIN32) # if !defined(NT_INCLUDED) # include <winsock2.h> # endif // !NT_INCLUDED #else // _WIN32 #if defined(__linux__) && !defined(VMKERNEL) # if !defined(__KERNEL__) # include <sys/socket.h> # endif // __KERNEL__ #else // linux && !VMKERNEL # if defined(__APPLE__) # include <sys/socket.h> # include <string.h> # elif defined(__FreeBSD__) # include <sys/socket.h> # endif // __FreeBSD__ #endif // linux && !VMKERNEL #endif #if defined __cplusplus extern "C" { #endif /** * \brief Option name for STREAM socket buffer size. * * Use as the option name in \c setsockopt(3) or \c getsockopt(3) to set * or get an \c unsigned \c long \c long that specifies the size of the * buffer underlying a vSockets STREAM socket. * * \note Value is clamped to the MIN and MAX. * * \see VMCISock_GetAFValueFd() * \see SO_VMCI_BUFFER_MIN_SIZE * \see SO_VMCI_BUFFER_MAX_SIZE * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * unsigned long long val = 0x1000; * int fd = socket(af, SOCK_STREAM, 0); * setsockopt(fd, af, SO_VMCI_BUFFER_SIZE, &val, sizeof val); * ... * close(fd); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode */ #define SO_VMCI_BUFFER_SIZE 0 /** * \brief Option name for STREAM socket minimum buffer size. * * Use as the option name in \c setsockopt(3) or \c getsockopt(3) to set * or get an \c unsigned \c long \c long that specifies the minimum size * allowed for the buffer underlying a vSockets STREAM socket. * * \see VMCISock_GetAFValueFd() * \see SO_VMCI_BUFFER_SIZE * \see SO_VMCI_BUFFER_MAX_SIZE * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * unsigned long long val = 0x500; * int fd = socket(af, SOCK_STREAM, 0); * setsockopt(fd, af, SO_VMCI_BUFFER_MIN_SIZE, &val, sizeof val); * ... * close(fd); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode */ #define SO_VMCI_BUFFER_MIN_SIZE 1 /** * \brief Option name for STREAM socket maximum buffer size. * * Use as the option name in \c setsockopt(3) or \c getsockopt(3) to set or * get an unsigned long long that specifies the maximum size allowed for the * buffer underlying a vSockets STREAM socket. * * \see VMCISock_GetAFValueFd() * \see SO_VMCI_BUFFER_SIZE * \see SO_VMCI_BUFFER_MIN_SIZE * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * unsigned long long val = 0x4000; * int fd = socket(af, SOCK_STREAM, 0); * setsockopt(fd, af, SO_VMCI_BUFFER_MAX_SIZE, &val, sizeof val); * ... * close(fd); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode */ #define SO_VMCI_BUFFER_MAX_SIZE 2 /** * \brief Option name for socket peer's host-specific VM ID. * * Use as the option name in \c getsockopt(3) to get a host-specific identifier * for the peer endpoint's VM. The identifier is a signed integer. * * \note Only available for ESX (VMKernel/userworld) endpoints. * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * int id; * socklen_t len = sizeof id; * int fd = socket(af, SOCK_DGRAM, 0); * getsockopt(fd, af, SO_VMCI_PEER_HOST_VM_ID, &id, &len); * ... * close(fd); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode */ #define SO_VMCI_PEER_HOST_VM_ID 3 /** * \brief Option name for socket's service label. * * Use as the option name in \c setsockopt(3) or \c getsockopt(3) to set or * get the service label for a socket. The service label is a C-style * NUL-terminated string. * * \note Only available for ESX (VMkernel/userworld) endpoints. */ #define SO_VMCI_SERVICE_LABEL 4 /** * \brief Option name for determining if a socket is trusted. * * Use as the option name in \c getsockopt(3) to determine if a socket is * trusted. The value is a signed integer. * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * int trusted; * socklen_t len = sizeof trusted; * int fd = socket(af, SOCK_DGRAM, 0); * getsockopt(fd, af, SO_VMCI_TRUSTED, &trusted, &len); * ... * close(fd); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode */ #define SO_VMCI_TRUSTED 5 /** * \brief Option name for STREAM socket connection timeout. * * Use as the option name in \c setsockopt(3) or \c getsockopt(3) to set or * get the connection timeout for a STREAM socket. The value is platform * dependent. On ESX, Linux and Mac OS, it is a \c struct \c timeval. * On Windows, it is a \c DWORD. * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * struct timeval t = { 5, 100000 }; // 5.1 seconds * int fd = socket(af, SOCK_STREAM, 0); * setsockopt(fd, af, SO_VMCI_CONNECT_TIMEOUT, &t, sizeof t); * ... * close(fd); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode */ #define SO_VMCI_CONNECT_TIMEOUT 6 /** * \brief Option name for using non-blocking send/receive. * * Use as the option name for \c setsockopt(3) or \c getsockopt(3) to set or * get the non-blocking transmit/receive flag for a STREAM socket. This flag * determines whether \c send() and \c recv() can be called in non-blocking * contexts for the given socket. The value is a signed integer. * * This option is only relevant to kernel endpoints, where descheduling * the thread of execution is not allowed, for example, while holding a * spinlock. It is not to be confused with conventional non-blocking socket * operations. * * \note Only available for VMKernel endpoints. * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * int nonblock; * socklen_t len = sizeof nonblock; * int fd = socket(af, SOCK_STREAM, 0); * getsockopt(fd, af, SO_VMCI_NONBLOCK_TXRX, &nonblock, &len); * ... * close(fd); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode */ #define SO_VMCI_NONBLOCK_TXRX 7 /** * \brief Option name for STREAM socket connection disconect cause * * Use as the option name in \c getsockopt(3) to get the cause of the * peer disconnect for a stream socket. * * \note Only available for ESX (VMkernel/userworld) endpoints. * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * int32 cause; * socklen_t len = sizeof cause; * int fd = socket(af, SOCK_DGRAM, 0); * ... * if (recv(fd, buf, buflen, 0) == 0) { * getsockopt(fd, af, SO_VMCI_DISCONNECT_CAUSE, &cause, &len); * } * close(fd); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode */ #define SO_VMCI_DISCONNECT_CAUSE 8 #define VMCI_SOCKETS_DISCONNECT_REGULAR 0 #define VMCI_SOCKETS_DISCONNECT_VMOTION 1 /** * \brief The vSocket equivalent of INADDR_ANY. * * This works for the \c svm_cid field of sockaddr_vm and indicates the * context ID of the current endpoint. * * \see sockaddr_vm * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * struct sockaddr_vm addr; * int fd = socket(af, SOCK_DGRAM, 0); * addr.svm_family = af; * addr.svm_cid = VMADDR_CID_ANY; * addr.svm_port = 2000; * bind(fd, &addr, sizeof addr); * ... * close(fd); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode */ #define VMADDR_CID_ANY ((unsigned int)-1) /** * \brief Bind to any available port. * * Works for the \c svm_port field of sockaddr_vm. * * \see sockaddr_vm * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * struct sockaddr_vm addr; * int fd = socket(af, SOCK_DGRAM, 0); * addr.svm_family = af; * addr.svm_cid = VMADDR_CID_ANY; * addr.svm_port = VMADDR_PORT_ANY; * bind(fd, &addr, sizeof addr); * ... * close(fd); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode */ #define VMADDR_PORT_ANY ((unsigned int)-1) /** * \brief Invalid vSockets version. * * \see VMCISock_Version() */ #define VMCI_SOCKETS_INVALID_VERSION ((unsigned int)-1) /** * \brief The epoch (first) component of the vSockets version. * * A single byte representing the epoch component of the vSockets version. * * \see VMCISock_Version() * * An example is given below. * * \code * unsigned int ver = VMCISock_Version(); * unsigned char epoch = VMCI_SOCKETS_VERSION_EPOCH(ver); * \endcode */ #define VMCI_SOCKETS_VERSION_EPOCH(_v) (((_v) & 0xFF000000) >> 24) /** * \brief The major (second) component of the vSockets version. * * A single byte representing the major component of the vSockets version. * Typically changes for every major release of a product. * * \see VMCISock_Version() * * An example is given below. * * \code * unsigned int ver = VMCISock_Version(); * unsigned char major = VMCI_SOCKETS_VERSION_MAJOR(ver); * \endcode */ #define VMCI_SOCKETS_VERSION_MAJOR(_v) (((_v) & 0x00FF0000) >> 16) /** * \brief The minor (third) component of the vSockets version. * * Two bytes representing the minor component of the vSockets version. * * \see VMCISock_Version() * * An example is given below. * * \code * unsigned int ver = VMCISock_Version(); * unsigned short minor = VMCI_SOCKETS_VERSION_MINOR(ver); * \endcode */ #define VMCI_SOCKETS_VERSION_MINOR(_v) (((_v) & 0x0000FFFF)) /** \cond PRIVATE */ #if defined(_WIN32) || defined(VMKERNEL) typedef unsigned short sa_family_t; #endif // _WIN32 #if defined(VMKERNEL) struct sockaddr { sa_family_t sa_family; char sa_data[14]; }; #endif /** \endcond */ /** * \brief Address structure for vSockets. * * The address family should be set to whatever VMCISock_GetAFValueFd() * returns. The structure members should all align on their natural * boundaries without resorting to compiler packing directives. The total * size of this structure should be exactly the same as that of \c struct * \c sockaddr. * * \see VMCISock_GetAFValueFd() */ struct sockaddr_vm { #if defined(__APPLE__) || defined(__FreeBSD__) unsigned char svm_len; #endif // __APPLE__ || __FreeBSD__ /** \brief Address family. \see VMCISock_GetAFValueFd() */ sa_family_t svm_family; /** \cond PRIVATE */ unsigned short svm_reserved1; /** \endcond */ /** \brief Port. \see VMADDR_PORT_ANY */ unsigned int svm_port; /** \brief Context ID. \see VMADDR_CID_ANY */ unsigned int svm_cid; /** \cond PRIVATE */ unsigned char svm_zero[sizeof(struct sockaddr) - #if defined(__APPLE__) sizeof(unsigned char) - #endif // __APPLE__ sizeof(sa_family_t) - sizeof(unsigned short) - sizeof(unsigned int) - sizeof(unsigned int)]; /** \endcond */ }; /** \cond PRIVATE */ struct uuid_2_cid { unsigned int u2c_context_id; unsigned int u2c_pad; char u2c_uuid_string[128]; }; /** \endcond */ #if defined(_WIN32) # if !defined(NT_INCLUDED) # include <winioctl.h> # define VMCI_SOCKETS_DEVICE L"\\\\.\\VMCI" # define VMCI_SOCKETS_VERSION 0x81032058 # define VMCI_SOCKETS_GET_AF_VALUE 0x81032068 # define VMCI_SOCKETS_GET_LOCAL_CID 0x8103206c # define VMCI_SOCKETS_UUID_2_CID 0x810320a4 static __inline unsigned int __VMCISock_DeviceIoControl(DWORD cmd) { unsigned int val = (unsigned int)-1; HANDLE device = CreateFileW(VMCI_SOCKETS_DEVICE, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_FLAG_OVERLAPPED, NULL); if (INVALID_HANDLE_VALUE != device) { DWORD ioReturn; DeviceIoControl(device, cmd, &val, sizeof val, &val, sizeof val, &ioReturn, NULL); CloseHandle(device); device = INVALID_HANDLE_VALUE; } return val; } static __inline unsigned int VMCISock_Version(void) { return __VMCISock_DeviceIoControl(VMCI_SOCKETS_VERSION); } static __inline int VMCISock_GetAFValue(void) { return (int)__VMCISock_DeviceIoControl(VMCI_SOCKETS_GET_AF_VALUE); } static __inline int VMCISock_GetAFValueFd(int *outFd) { (void)outFd; /* Unused parameter. */ return VMCISock_GetAFValue(); } static __inline void VMCISock_ReleaseAFValueFd(int fd) { (void)fd; /* Unused parameter. */ } static __inline unsigned int VMCISock_GetLocalCID(void) { return __VMCISock_DeviceIoControl(VMCI_SOCKETS_GET_LOCAL_CID); } static __inline unsigned int VMCISock_Uuid2ContextId(const char *uuidString) { struct uuid_2_cid io; HANDLE device = CreateFileW(VMCI_SOCKETS_DEVICE, GENERIC_READ, 0, NULL, OPEN_EXISTING, FILE_FLAG_OVERLAPPED, NULL); io.u2c_context_id = VMADDR_CID_ANY; if (INVALID_HANDLE_VALUE != device) { DWORD ioReturn; strncpy_s(io.u2c_uuid_string, sizeof io.u2c_uuid_string, uuidString, _TRUNCATE); DeviceIoControl(device, VMCI_SOCKETS_UUID_2_CID, &io, sizeof io, &io, sizeof io, &ioReturn, NULL); CloseHandle(device); device = INVALID_HANDLE_VALUE; } return io.u2c_context_id; } # endif // !NT_INCLUDED #else // _WIN32 #if (defined(__linux__) && !defined(VMKERNEL)) || (defined(__APPLE__)) # if defined(__linux__) && defined(__KERNEL__) void VMCISock_KernelRegister(void); void VMCISock_KernelDeregister(void); int VMCISock_GetAFValue(void); int VMCISock_GetLocalCID(void); # elif defined(__APPLE__) && (KERNEL) /* Nothing to define here. */ # else // __KERNEL__ # include <fcntl.h> # include <stdio.h> # include <string.h> # include <sys/ioctl.h> # include <sys/stat.h> # include <sys/types.h> # include <unistd.h> /** \cond PRIVATE */ # define VMCI_SOCKETS_DEFAULT_DEVICE "/dev/vsock" # define VMCI_SOCKETS_CLASSIC_ESX_DEVICE "/vmfs/devices/char/vsock/vsock" # if defined(__linux__) # define VMCI_SOCKETS_VERSION 1972 # define VMCI_SOCKETS_GET_AF_VALUE 1976 # define VMCI_SOCKETS_GET_LOCAL_CID 1977 # define VMCI_SOCKETS_UUID_2_CID 1991 # elif defined(__APPLE__) # include <sys/ioccom.h> # define VMCI_SOCKETS_VERSION _IOR( 'V', 21, unsigned) # define VMCI_SOCKETS_GET_AF_VALUE _IOR( 'V', 25, int) # define VMCI_SOCKETS_GET_LOCAL_CID _IOR( 'V', 26, unsigned) # define VMCI_SOCKETS_UUID_2_CID _IOWR('V', 40, struct uuid_2_cid) #endif /** \endcond */ /* *********************************************************************** * VMCISock_Version */ /** * * \brief Retrieve the vSockets version. * * Returns the current version of vSockets. The version is a 32-bit * unsigned integer that consist of three components: the epoch, the * major version, and the minor version. Use the \c VMCI_SOCKETS_VERSION * macros to extract the components. * * \see VMCI_SOCKETS_VERSION_EPOCH() * \see VMCI_SOCKETS_VERSION_MAJOR() * \see VMCI_SOCKETS_VERSION_MINOR() * * \retval VMCI_SOCKETS_INVALID_VERSION Not available. * \retval other The current version. * * An example is given below. * * \code * unsigned int ver = VMCISock_Version(); * if (ver != VMCI_SOCKETS_INVALID_VERSION) { * printf("vSockets version=%d.%d.%d\n", * VMCI_SOCKETS_VERSION_EPOCH(ver), * VMCI_SOCKETS_VERSION_MAJOR(ver), * VMCI_SOCKETS_VERSION_MINOR(ver)); * } * \endcode * *********************************************************************** */ static inline unsigned int VMCISock_Version(void) { int fd; unsigned int version; fd = open(VMCI_SOCKETS_DEFAULT_DEVICE, O_RDONLY); if (fd < 0) { fd = open(VMCI_SOCKETS_CLASSIC_ESX_DEVICE, O_RDONLY); if (fd < 0) { return VMCI_SOCKETS_INVALID_VERSION; } } if (ioctl(fd, VMCI_SOCKETS_VERSION, &version) < 0) { version = VMCI_SOCKETS_INVALID_VERSION; } close(fd); return version; } /* *********************************************************************** * VMCISock_GetAFValueFd */ /** * * \brief Retrieve the address family value for vSockets. * * Returns the value to be used for the vSockets address family. * This value should be used as the domain argument to \c socket(2) (when * you might otherwise use \c AF_INET). For vSocket-specific options, * this value should also be used for the level argument to * \c setsockopt(2) (when you might otherwise use \c SOL_TCP). * * \see VMCISock_ReleaseAFValueFd() * \see sockaddr_vm * * \param[out] outFd File descriptor to the VMCI device. The * address family value is valid until this * descriptor is closed. This parameter is * not necessarily valid, but it is set if * the return value is not -1. * Call VMCISock_ReleaseAFValueFd() to close * this descriptor. * * \retval -1 Not available. * \retval other The address family value. * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * if (af != -1) { * int fd = socket(af, SOCK_STREAM, 0); * ... * close(fd); * close(vmciFd); * } * \endcode * *********************************************************************** */ static inline int VMCISock_GetAFValueFd(int *outFd) { int fd; int family = -1; #if defined(__linux__) /* * vSockets is now in mainline kernel with address family 40. As part * of upstreaming, we removed the IOCTL we use below to determine the * address family. So to handle both a new and old kernel we do this: * 1. Check if our family already exists by making a socket with it. * Some weird kernel might claim this too, but it's very unlikely * (Linus' tree has us at 40, and that's what we care about). * 2. If that fails, try the normal IOCTL path, since it's probably an * older kernel with vSockets from Tools. * 3. If that fails, then vSockets really isn't available. */ #define AF_VSOCK_LOCAL 40 { int s = socket(AF_VSOCK_LOCAL, SOCK_DGRAM, 0); if (s != -1) { close(s); if (outFd) { *outFd = -1; } return AF_VSOCK_LOCAL; } } #undef AF_VSOCK_LOCAL #endif // linux fd = open(VMCI_SOCKETS_DEFAULT_DEVICE, O_RDONLY); if (fd < 0) { fd = open(VMCI_SOCKETS_CLASSIC_ESX_DEVICE, O_RDONLY); if (fd < 0) { return -1; } } if (ioctl(fd, VMCI_SOCKETS_GET_AF_VALUE, &family) < 0) { family = -1; } if (family < 0) { close(fd); } else if (outFd) { *outFd = fd; } return family; } /** \cond PRIVATE */ /* *********************************************************************** * VMCISock_GetAFValue */ /** * * \brief Retrieve the address family value for vSockets. * * Returns the value to be used for the vSockets address family. * This value should be used as the domain argument to \c socket(2) (when * you might otherwise use \c AF_INET). For vSocket-specific options, * this value should also be used for the level argument to * \c setsockopt(2) (when you might otherwise use \c SOL_TCP). * * \note This function leaves its descriptor to the vsock device open so * that the socket implementation knows that the socket family is still in * use. This is done because the address family is registered with the * kernel on-demand and a notification is needed to unregister the address * family. Use of this function is thus discouraged; please use * VMCISock_GetAFValueFd() instead. * * \see VMCISock_GetAFValueFd() * \see sockaddr_vm * * \retval -1 Not available. * \retval other The address family value. * * An example is given below. * * \code * int af = VMCISock_GetAFValue(); * if (af != -1) { * int fd = socket(af, SOCK_STREAM, 0); * ... * close(fd); * } * \endcode * *********************************************************************** */ static inline int VMCISock_GetAFValue(void) { return VMCISock_GetAFValueFd(NULL); } /** \endcond PRIVATE */ /* *********************************************************************** * VMCISock_ReleaseAFValueFd */ /** * * \brief Release the file descriptor obtained when retrieving the * address family value. * * Use this to release the file descriptor obtained by calling * VMCISock_GetAFValueFd(). * * \see VMCISock_GetAFValueFd() * * \param[in] fd File descriptor to the VMCI device. * *********************************************************************** */ static inline void VMCISock_ReleaseAFValueFd(int fd) { if (fd >= 0) { close(fd); } } /* *********************************************************************** * VMCISock_GetLocalCID */ /** * * \brief Retrieve the current context ID. * * \see VMADDR_CID_ANY * * \retval VMADDR_CID_ANY Not available. * \retval other The current context ID. * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * struct sockaddr_vm addr; * addr.svm_family = af; * addr.svm_cid = VMCISock_GetLocalCID(); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode * *********************************************************************** */ static inline unsigned int VMCISock_GetLocalCID(void) { int fd; unsigned int contextId; fd = open(VMCI_SOCKETS_DEFAULT_DEVICE, O_RDONLY); if (fd < 0) { fd = open(VMCI_SOCKETS_CLASSIC_ESX_DEVICE, O_RDONLY); if (fd < 0) { return VMADDR_CID_ANY; } } if (ioctl(fd, VMCI_SOCKETS_GET_LOCAL_CID, &contextId) < 0) { contextId = VMADDR_CID_ANY; } close(fd); return contextId; } /* *********************************************************************** * VMCISock_Uuid2ContextId */ /** * * \brief Retrieve the context ID of a running VM, given a VM's UUID. * * Retrieves the context ID of a running virtual machine given that virtual * machines's unique identifier. The identifier is local to the host and * its meaning is platform-specific. On ESX, which is currently the only * supported platform, it is the "bios.uuid" field as specified in the VM's * VMX file. * * \see VMADDR_CID_ANY * * \retval VMADDR_CID_ANY Not available. * \retval other The VM's context ID. * * \note Only available for ESX (userworld) endpoints. * * An example is given below. * * \code * int vmciFd; * int af = VMCISock_GetAFValueFd(&vmciFd); * unsigned int cid = VMCISock_Uuid2ContextId( * "56 4d 07 d8 cc d5 c4 0d-98 44 dc 1e 8f e0 da f3"); * VMCISock_ReleaseAFValueFd(vmciFd); * \endcode * *********************************************************************** */ static inline unsigned int VMCISock_Uuid2ContextId(const char *uuidString) { int fd; struct uuid_2_cid io; fd = open(VMCI_SOCKETS_DEFAULT_DEVICE, O_RDONLY); if (fd < 0) { fd = open(VMCI_SOCKETS_CLASSIC_ESX_DEVICE, O_RDONLY); if (fd < 0) { return VMADDR_CID_ANY; } } strncpy(io.u2c_uuid_string, uuidString, sizeof io.u2c_uuid_string); if (ioctl(fd, VMCI_SOCKETS_UUID_2_CID, &io) < 0) { io.u2c_context_id = VMADDR_CID_ANY; } close(fd); return io.u2c_context_id; } # endif // __KERNEL__ #elif defined(__FreeBSD__) /* * No FreeBSD support yet, but it might appear in the future. Just define * some stubs that return errors - that way a client doesn't have to ifdef * the calls (assuming it can handle the failures). */ # define VMCISock_Version() VMCI_SOCKETS_INVALID_VERSION # define VMCISock_GetAFValueFd(outFd) (-1) # define VMCISock_GetAFValue() VMCISock_GetAFValueFd(NULL) # define VMCISock_ReleaseAFValueFd(fd) do { } while (0) # define VMCISock_GetLocalCID() VMADDR_CID_ANY # define VMCISock_Uuid2ContextId(uuidString) VMADDR_CID_ANY #endif // __FreeBSD__ #endif // _WIN32 #if defined __cplusplus } // extern "C" #endif #endif // _VMCI_SOCKETS_H_ vsock-only/linux/notify.c 0000444 0000000 0000000 00000075222 13432725350 014505 0 ustar root root /********************************************************* * Copyright (C) 2009-2014 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * notify.c -- * * Linux control notifications for the VMCI Stream Sockets protocol. */ #include "driver-config.h" #include <linux/socket.h> #include "compat_sock.h" #include "notify.h" #include "af_vsock.h" #define PKT_FIELD(vsk, fieldName) \ (vsk)->notify.pkt.fieldName #define VSOCK_MAX_DGRAM_RESENDS 10 /* *---------------------------------------------------------------------------- * * VSockVmciNotifyWaitingWrite -- * * Determines if the conditions have been met to notify a waiting writer. * * Results: * TRUE if a notification should be sent, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static Bool VSockVmciNotifyWaitingWrite(VSockVmciSock *vsk) // IN { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) Bool retval; uint64 notifyLimit; if (!PKT_FIELD(vsk, peerWaitingWrite)) { return FALSE; } #ifdef VSOCK_OPTIMIZATION_FLOW_CONTROL /* * When the sender blocks, we take that as a sign that the sender * is faster than the receiver. To reduce the transmit rate of the * sender, we delay the sending of the read notification by * decreasing the writeNotifyWindow. The notification is delayed * until the number of bytes used in the queue drops below the * writeNotifyWindow. */ if (!PKT_FIELD(vsk, peerWaitingWriteDetected)) { PKT_FIELD(vsk, peerWaitingWriteDetected) = TRUE; if (PKT_FIELD(vsk, writeNotifyWindow) < PAGE_SIZE) { PKT_FIELD(vsk, writeNotifyWindow) = PKT_FIELD(vsk, writeNotifyMinWindow); } else { PKT_FIELD(vsk, writeNotifyWindow) -= PAGE_SIZE; if (PKT_FIELD(vsk, writeNotifyWindow) < PKT_FIELD(vsk, writeNotifyMinWindow)) { PKT_FIELD(vsk, writeNotifyWindow) = PKT_FIELD(vsk, writeNotifyMinWindow); } } } notifyLimit = vsk->consumeSize - PKT_FIELD(vsk, writeNotifyWindow); #else notifyLimit = 0; #endif // VSOCK_OPTIMIZATION_FLOW_CONTROL /* * For now we ignore the wait information and just see if the free * space exceeds the notify limit. Note that improving this * function to be more intelligent will not require a protocol * change and will retain compatibility between endpoints with * mixed versions of this function. * * The notifyLimit is used to delay notifications in the case where * flow control is enabled. Below the test is expressed in terms of * free space in the queue: * if freeSpace > ConsumeSize - writeNotifyWindow then notify * An alternate way of expressing this is to rewrite the expression * to use the data ready in the receive queue: * if writeNotifyWindow > bufferReady then notify * as freeSpace == ConsumeSize - bufferReady. */ retval = vmci_qpair_consume_free_space(vsk->qpair) > notifyLimit; #ifdef VSOCK_OPTIMIZATION_FLOW_CONTROL if (retval) { /* * Once we notify the peer, we reset the detected flag so the * next wait will again cause a decrease in the window size. */ PKT_FIELD(vsk, peerWaitingWriteDetected) = FALSE; } #endif // VSOCK_OPTIMIZATION_FLOW_CONTROL return retval; #else return TRUE; #endif } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyWaitingRead -- * v * Determines if the conditions have been met to notify a waiting reader. * * Results: * TRUE if a notification should be sent, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static Bool VSockVmciNotifyWaitingRead(VSockVmciSock *vsk) // IN { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) if (!PKT_FIELD(vsk, peerWaitingRead)) { return FALSE; } /* * For now we ignore the wait information and just see if there is any data * for our peer to read. Note that improving this function to be more intelligent will * not require a protocol change and will retain compatibility between * endpoints with mixed versions of this function. */ return vmci_qpair_produce_buf_ready(vsk->qpair) > 0; #else return TRUE; #endif } /* *---------------------------------------------------------------------------- * * VSockVmciHandleWaitingRead -- * * Handles an incoming waiting read message. * * Results: * None. * * Side effects: * May send a notification to the peer, may update socket's wait info * structure. * *---------------------------------------------------------------------------- */ static void VSockVmciHandleWaitingRead(struct sock *sk, // IN VSockPacket *pkt, // IN Bool bottomHalf, // IN struct sockaddr_vm *dst, // IN struct sockaddr_vm *src) // IN { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) VSockVmciSock *vsk; vsk = vsock_sk(sk); PKT_FIELD(vsk, peerWaitingRead) = TRUE; memcpy(&PKT_FIELD(vsk, peerWaitingReadInfo), &pkt->u.wait, sizeof PKT_FIELD(vsk, peerWaitingReadInfo)); if (VSockVmciNotifyWaitingRead(vsk)) { Bool sent; if (bottomHalf) { sent = VSOCK_SEND_WROTE_BH(dst, src) > 0; } else { sent = VSOCK_SEND_WROTE(sk) > 0; } if (sent) { PKT_FIELD(vsk, peerWaitingRead) = FALSE; } } #endif } /* *---------------------------------------------------------------------------- * * VSockVmciHandleWaitingWrite -- * * Handles an incoming waiting write message. * * Results: * None. * * Side effects: * May send a notification to the peer, may update socket's wait info * structure. * *---------------------------------------------------------------------------- */ static void VSockVmciHandleWaitingWrite(struct sock *sk, // IN VSockPacket *pkt, // IN Bool bottomHalf, // IN struct sockaddr_vm *dst, // IN struct sockaddr_vm *src) // IN { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) VSockVmciSock *vsk; vsk = vsock_sk(sk); PKT_FIELD(vsk, peerWaitingWrite) = TRUE; memcpy(&PKT_FIELD(vsk, peerWaitingWriteInfo), &pkt->u.wait, sizeof PKT_FIELD(vsk,peerWaitingWriteInfo)); if (VSockVmciNotifyWaitingWrite(vsk)) { Bool sent; if (bottomHalf) { sent = VSOCK_SEND_READ_BH(dst, src) > 0; } else { sent = VSOCK_SEND_READ(sk) > 0; } if (sent) { PKT_FIELD(vsk, peerWaitingWrite) = FALSE; } } #endif } /* *---------------------------------------------------------------------------- * * VSockVmciHandleRead -- * * Handles an incoming read message. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciHandleRead(struct sock *sk, // IN VSockPacket *pkt, // IN: unused Bool bottomHalf, // IN: unused struct sockaddr_vm *dst, // IN: unused struct sockaddr_vm *src) // IN: unused { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) VSockVmciSock *vsk; vsk = vsock_sk(sk); PKT_FIELD(vsk, sentWaitingWrite) = FALSE; #endif sk->sk_write_space(sk); } /* *---------------------------------------------------------------------------- * * VSockVmciSendWaitingRead -- * * Sends a waiting read notification to this socket's peer. * * Results: * TRUE if the datagram is sent successfully, FALSE otherwise. * * Side effects: * Our peer will notify us when there is data to read from our consume * queue. * *---------------------------------------------------------------------------- */ static Bool VSockVmciSendWaitingRead(struct sock *sk, // IN uint64 roomNeeded) // IN { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) VSockVmciSock *vsk; VSockWaitingInfo waitingInfo; uint64 tail; uint64 head; uint64 roomLeft; Bool ret; ASSERT(sk); vsk = vsock_sk(sk); if (PKT_FIELD(vsk, sentWaitingRead)) { return TRUE; } if (PKT_FIELD(vsk, writeNotifyWindow) < vsk->consumeSize) { PKT_FIELD(vsk, writeNotifyWindow) = MIN(PKT_FIELD(vsk, writeNotifyWindow) + PAGE_SIZE, vsk->consumeSize); } vmci_qpair_get_consume_indexes(vsk->qpair, &tail, &head); roomLeft = vsk->consumeSize - head; if (roomNeeded >= roomLeft) { waitingInfo.offset = roomNeeded - roomLeft; waitingInfo.generation = PKT_FIELD(vsk, consumeQGeneration) + 1; } else { waitingInfo.offset = head + roomNeeded; waitingInfo.generation = PKT_FIELD(vsk, consumeQGeneration); } ret = VSOCK_SEND_WAITING_READ(sk, &waitingInfo) > 0; if (ret) { PKT_FIELD(vsk, sentWaitingRead) = TRUE; } return ret; #else return TRUE; #endif } /* *---------------------------------------------------------------------------- * * VSockVmciSendWaitingWrite -- * * Sends a waiting write notification to this socket's peer. * * Results: * TRUE if the datagram is sent successfully or does not need to be sent. * FALSE otherwise. * * Side effects: * Our peer will notify us when there is room to write in to our produce * queue. * *---------------------------------------------------------------------------- */ static Bool VSockVmciSendWaitingWrite(struct sock *sk, // IN uint64 roomNeeded) // IN { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) VSockVmciSock *vsk; VSockWaitingInfo waitingInfo; uint64 tail; uint64 head; uint64 roomLeft; Bool ret; ASSERT(sk); vsk = vsock_sk(sk); if (PKT_FIELD(vsk, sentWaitingWrite)) { return TRUE; } vmci_qpair_get_produce_indexes(vsk->qpair, &tail, &head); roomLeft = vsk->produceSize - tail; if (roomNeeded + 1 >= roomLeft) { /* Wraps around to current generation. */ waitingInfo.offset = roomNeeded + 1 - roomLeft; waitingInfo.generation = PKT_FIELD(vsk, produceQGeneration); } else { waitingInfo.offset = tail + roomNeeded + 1; waitingInfo.generation = PKT_FIELD(vsk, produceQGeneration) - 1; } ret = VSOCK_SEND_WAITING_WRITE(sk, &waitingInfo) > 0; if (ret) { PKT_FIELD(vsk, sentWaitingWrite) = TRUE; } return ret; #else return TRUE; #endif } /* *---------------------------------------------------------------------------- * * VSockVmciSendReadNotification -- * * Sends a read notification to this socket's peer. * * Results: * >= 0 if the datagram is sent successfully, negative error value * otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciSendReadNotification(struct sock *sk) // IN { VSockVmciSock *vsk; Bool sentRead; unsigned int retries; int err; ASSERT(sk); vsk = vsock_sk(sk); sentRead = FALSE; retries = 0; err = 0; if (VSockVmciNotifyWaitingWrite(vsk)) { /* * Notify the peer that we have read, retrying the send on failure up to our * maximum value. XXX For now we just log the failure, but later we should * schedule a work item to handle the resend until it succeeds. That would * require keeping track of work items in the vsk and cleaning them up upon * socket close. */ while (!(vsk->peerShutdown & RCV_SHUTDOWN) && !sentRead && retries < VSOCK_MAX_DGRAM_RESENDS) { err = VSOCK_SEND_READ(sk); if (err >= 0) { sentRead = TRUE; } retries++; } if (retries >= VSOCK_MAX_DGRAM_RESENDS) { Warning("unable to send read notification to peer for socket %p.\n", sk); } else { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) PKT_FIELD(vsk, peerWaitingWrite) = FALSE; #endif } } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciHandleWrote -- * * Handles an incoming wrote message. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciHandleWrote(struct sock *sk, // IN VSockPacket *pkt, // IN: unused Bool bottomHalf, // IN: unused struct sockaddr_vm *dst, // IN: unused struct sockaddr_vm *src) // IN: unused { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) VSockVmciSock *vsk; vsk = vsock_sk(sk); PKT_FIELD(vsk, sentWaitingRead) = FALSE; #endif sk->sk_data_ready(sk, 0); } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktSocketInit -- * * Function that is called after a socket is created and before any * notify ops are used. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciNotifyPktSocketInit(struct sock *sk) // IN { VSockVmciSock *vsk; vsk = vsock_sk(sk); PKT_FIELD(vsk, writeNotifyWindow) = PAGE_SIZE; PKT_FIELD(vsk, writeNotifyMinWindow) = PAGE_SIZE; PKT_FIELD(vsk, peerWaitingRead) = FALSE; PKT_FIELD(vsk, peerWaitingWrite) = FALSE; PKT_FIELD(vsk, peerWaitingWriteDetected) = FALSE; PKT_FIELD(vsk, sentWaitingRead) = FALSE; PKT_FIELD(vsk, sentWaitingWrite) = FALSE; PKT_FIELD(vsk, produceQGeneration) = 0; PKT_FIELD(vsk, consumeQGeneration) = 0; memset(&PKT_FIELD(vsk, peerWaitingReadInfo), 0, sizeof PKT_FIELD(vsk, peerWaitingReadInfo)); memset(&PKT_FIELD(vsk, peerWaitingWriteInfo), 0, sizeof PKT_FIELD(vsk, peerWaitingWriteInfo)); } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktSocketDestruct -- * * Function that is called when the socket is being released. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciNotifyPktSocketDestruct(struct sock *sk) // IN { return; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktPollIn -- * * Called by the poll function to figure out if there is data to read * and to setup future notifications if needed. Only called on sockets * that aren't shutdown for recv. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktPollIn(struct sock *sk, // IN size_t target, // IN Bool *dataReadyNow) // IN { VSockVmciSock *vsk; ASSERT(sk); ASSERT(dataReadyNow); vsk = vsock_sk(sk); if (VSockVmciStreamHasData(vsk)) { *dataReadyNow = TRUE; } else { /* * We can't read right now because there is nothing in the queue. * Ask for notifications when there is something to read. */ if (sk->sk_state == SS_CONNECTED) { if (!VSockVmciSendWaitingRead(sk, 1)) { return -1; } } *dataReadyNow = FALSE; } return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktPollOut * * Called by the poll function to figure out if there is space to write * and to setup future notifications if needed. Only called on a * connected socket that isn't shutdown for send. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktPollOut(struct sock *sk, // IN size_t target, // IN Bool *spaceAvailNow) // IN { int64 produceQFreeSpace; VSockVmciSock *vsk; ASSERT(sk); ASSERT(spaceAvailNow); vsk = vsock_sk(sk); produceQFreeSpace = VSockVmciStreamHasSpace(vsk); if (produceQFreeSpace > 0) { *spaceAvailNow = TRUE; return 0; } else if (produceQFreeSpace == 0) { /* * This is a connected socket but we can't currently send data. Notify * the peer that we are waiting if the queue is full. * We only send a waiting write if the queue is full because otherwise * we end up in an infinite WAITING_WRITE, READ, WAITING_WRITE, READ, etc. * loop. Treat failing to send the notification as a socket error, passing * that back through the mask. */ if (!VSockVmciSendWaitingWrite(sk, 1)) { return -1; } *spaceAvailNow = FALSE; } return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktRecvInit -- * * Called at the start of a stream recv call with the socket lock held. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktRecvInit(struct sock *sk, // IN size_t target, // IN VSockVmciRecvNotifyData *data) // IN { VSockVmciSock *vsk; ASSERT(sk); ASSERT(data); vsk = vsock_sk(sk); #ifdef VSOCK_OPTIMIZATION_WAITING_NOTIFY data->consumeHead = 0; data->produceTail = 0; #ifdef VSOCK_OPTIMIZATION_FLOW_CONTROL data->notifyOnBlock = FALSE; if (PKT_FIELD(vsk, writeNotifyMinWindow) < target + 1) { ASSERT(target < vsk->consumeSize); PKT_FIELD(vsk, writeNotifyMinWindow) = target + 1; if (PKT_FIELD(vsk, writeNotifyWindow) < PKT_FIELD(vsk, writeNotifyMinWindow)) { /* * If the current window is smaller than the new minimal * window size, we need to reevaluate whether we need to * notify the sender. If the number of ready bytes are * smaller than the new window, we need to send a * notification to the sender before we block. */ PKT_FIELD(vsk, writeNotifyWindow) = PKT_FIELD(vsk, writeNotifyMinWindow); data->notifyOnBlock = TRUE; } } #endif #endif return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktRecvPreBlock -- * * Called right before a socket is about to block with the socket lock * held. The socket lock may have been released between the entry * function and the preblock call. * * Note: This function may be called multiple times before the post * block function is called. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktRecvPreBlock(struct sock *sk, // IN size_t target, // IN VSockVmciRecvNotifyData *data) // IN { int err; ASSERT(sk); ASSERT(data); err = 0; /* Notify our peer that we are waiting for data to read. */ if (!VSockVmciSendWaitingRead(sk, target)) { err = -EHOSTUNREACH; return err; } #ifdef VSOCK_OPTIMIZATION_FLOW_CONTROL if (data->notifyOnBlock) { err = VSockVmciSendReadNotification(sk); if (err < 0) { return err; } data->notifyOnBlock = FALSE; } #endif return err; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktRecvPreDequeue -- * * Called right before we dequeue / peek data from a socket. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktRecvPreDequeue(struct sock *sk, // IN size_t target, // IN VSockVmciRecvNotifyData *data) // IN { VSockVmciSock *vsk; ASSERT(sk); ASSERT(data); vsk = vsock_sk(sk); /* * Now consume up to len bytes from the queue. Note that since we have the * socket locked we should copy at least ready bytes. */ #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) vmci_qpair_get_consume_indexes(vsk->qpair, &data->produceTail, &data->consumeHead); #endif return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktRecvPostDequeue -- * * Called right after we dequeue / peek data from a socket. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktRecvPostDequeue(struct sock *sk, // IN size_t target, // IN ssize_t copied, // IN Bool dataRead, // IN VSockVmciRecvNotifyData *data) // IN { VSockVmciSock *vsk; int err; ASSERT(sk); ASSERT(data); vsk = vsock_sk(sk); err = 0; if (dataRead) { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) /* * Detect a wrap-around to maintain queue generation. Note that this is * safe since we hold the socket lock across the two queue pair * operations. */ if (copied >= vsk->consumeSize - data->consumeHead) { PKT_FIELD(vsk, consumeQGeneration)++; } #endif err = VSockVmciSendReadNotification(sk); if (err < 0) { return err; } } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktSendInit -- * * Called at the start of a stream send call with the socket lock held. * * Results: * 0 on success. A negative error code on failure. * * Side effects: * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktSendInit(struct sock *sk, // IN VSockVmciSendNotifyData *data) // IN { ASSERT(sk); ASSERT(data); #ifdef VSOCK_OPTIMIZATION_WAITING_NOTIFY data->consumeHead = 0; data->produceTail = 0; #endif return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktSendPreBlock -- * * Called right before a socket is about to block with the socket lock * held. The socket lock may have been released between the entry * function and the preblock call. * * Note: This function may be called multiple times before the post * block function is called. * * Results. * 0 on success. A negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktSendPreBlock(struct sock *sk, // IN VSockVmciSendNotifyData *data) // IN { ASSERT(sk); ASSERT(data); /* Notify our peer that we are waiting for room to write. */ if (!VSockVmciSendWaitingWrite(sk, 1)) { return -EHOSTUNREACH; } return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifySendPreEnqueue -- * * Called right before we Enqueue to a socket. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktSendPreEnqueue(struct sock *sk, // IN VSockVmciSendNotifyData *data) // IN { VSockVmciSock *vsk; ASSERT(sk); ASSERT(data); vsk = vsock_sk(sk); #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) vmci_qpair_get_produce_indexes(vsk->qpair, &data->produceTail, &data->consumeHead); #endif return 0;; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifySendPostEnqueue -- * * Called right after we enqueue data to a socket. * * Results: * 0 on success. Negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int32 VSockVmciNotifyPktSendPostEnqueue(struct sock *sk, // IN ssize_t written, // IN VSockVmciSendNotifyData *data) // IN { int err = 0; VSockVmciSock *vsk; Bool sentWrote = FALSE; int retries = 0; ASSERT(sk); ASSERT(data); vsk = vsock_sk(sk); #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) /* * Detect a wrap-around to maintain queue generation. Note that this is * safe since we hold the socket lock across the two queue pair * operations. */ if (written >= vsk->produceSize - data->produceTail) { PKT_FIELD(vsk, produceQGeneration)++; } #endif if (VSockVmciNotifyWaitingRead(vsk)) { /* * Notify the peer that we have written, retrying the send on failure up to * our maximum value. See the XXX comment for the corresponding piece of * code in StreamRecvmsg() for potential improvements. */ while (!(vsk->peerShutdown & RCV_SHUTDOWN) && !sentWrote && retries < VSOCK_MAX_DGRAM_RESENDS) { err = VSOCK_SEND_WROTE(sk); if (err >= 0) { sentWrote = TRUE; } retries++; } if (retries >= VSOCK_MAX_DGRAM_RESENDS) { Warning("unable to send wrote notification to peer for socket %p.\n", sk); return err; } else { #if defined(VSOCK_OPTIMIZATION_WAITING_NOTIFY) PKT_FIELD(vsk, peerWaitingRead) = FALSE; #endif } } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktHandlePkt * * Called when a notify packet is recieved for a socket in the connected * state. Note this might be called from a bottom half. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciNotifyPktHandlePkt(struct sock *sk, // IN VSockPacket *pkt, // IN Bool bottomHalf, // IN struct sockaddr_vm *dst, // IN struct sockaddr_vm *src, // IN Bool *pktProcessed) // In { Bool processed = FALSE; ASSERT(sk); ASSERT(pkt); switch (pkt->type) { case VSOCK_PACKET_TYPE_WROTE: VSockVmciHandleWrote(sk, pkt, bottomHalf, dst, src); processed = TRUE; break; case VSOCK_PACKET_TYPE_READ: VSockVmciHandleRead(sk, pkt, bottomHalf, dst, src); processed = TRUE; break; case VSOCK_PACKET_TYPE_WAITING_WRITE: VSockVmciHandleWaitingWrite(sk, pkt, bottomHalf, dst, src); processed = TRUE; break; case VSOCK_PACKET_TYPE_WAITING_READ: VSockVmciHandleWaitingRead(sk, pkt, bottomHalf, dst, src); processed = TRUE; break; } if (pktProcessed) { *pktProcessed = processed; } } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktProcessRequest * * Called near the end of process request. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciNotifyPktProcessRequest(struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(sk); vsk = vsock_sk(sk); PKT_FIELD(vsk, writeNotifyWindow) = vsk->consumeSize; if (vsk->consumeSize < PKT_FIELD(vsk, writeNotifyMinWindow)) { PKT_FIELD(vsk, writeNotifyMinWindow) = vsk->consumeSize; } } /* *---------------------------------------------------------------------------- * * VSockVmciNotifyPktProcessNegotiate * * Called near the end of process negotiate. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciNotifyPktProcessNegotiate(struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(sk); vsk = vsock_sk(sk); PKT_FIELD(vsk, writeNotifyWindow) = vsk->consumeSize; if (vsk->consumeSize < PKT_FIELD(vsk, writeNotifyMinWindow)) { PKT_FIELD(vsk, writeNotifyMinWindow) = vsk->consumeSize; } } /* Socket control packet based operations. */ VSockVmciNotifyOps vSockVmciNotifyPktOps = { VSockVmciNotifyPktSocketInit, VSockVmciNotifyPktSocketDestruct, VSockVmciNotifyPktPollIn, VSockVmciNotifyPktPollOut, VSockVmciNotifyPktHandlePkt, VSockVmciNotifyPktRecvInit, VSockVmciNotifyPktRecvPreBlock, VSockVmciNotifyPktRecvPreDequeue, VSockVmciNotifyPktRecvPostDequeue, VSockVmciNotifyPktSendInit, VSockVmciNotifyPktSendPreBlock, VSockVmciNotifyPktSendPreEnqueue, VSockVmciNotifyPktSendPostEnqueue, VSockVmciNotifyPktProcessRequest, VSockVmciNotifyPktProcessNegotiate, }; vsock-only/linux/vmci_sockets_int.h 0000444 0000000 0000000 00000004757 13432725350 016552 0 ustar root root /********************************************************* * Copyright (C) 2009,2014 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vmci_sockets_int.h -- * * vSockets private constants and types. * * This file is internal only, we do not ship the kernel interface yet. * You need to include this file *before* vmci_sockets.h in your kernel * module. */ #ifndef _VMCI_SOCKETS_INT_H_ #define _VMCI_SOCKETS_INT_H_ #if defined __cplusplus extern "C" { #endif #if defined(_WIN32) # if defined(NT_INCLUDED) # if (_WIN32_WINNT < 0x0600) /* * WinSockKernel is targetted at Vista and later. We want to allow * drivers built from W2K onwards to work with the interface, so we * need to define some missing types before we bring in the WSK header. */ typedef unsigned short u_short; # include <windef.h> # include <ws2def.h> typedef WSACMSGHDR CMSGHDR, *PCMSGHDR; # endif // (_WIN32_WINNT < 0x0600) # include <wsk.h> NTSTATUS VMCISock_WskRegister(PWSK_CLIENT_NPI wskClientNpi, PWSK_REGISTRATION wskRegistration); NTSTATUS VMCISock_WskDeregister(PWSK_REGISTRATION wskRegistration); NTSTATUS VMCISock_WskCaptureProviderNPI(PWSK_REGISTRATION wskRegistration, ULONG waitTimeout, PWSK_PROVIDER_NPI wskProviderNpi); NTSTATUS VMCISock_WskReleaseProviderNPI(PWSK_REGISTRATION wskRegistration); NTSTATUS VMCISock_WskGetAFValue(PWSK_CLIENT wskClient, PIRP irp); NTSTATUS VMCISock_WskGetLocalCID(PWSK_CLIENT wskClient, PIRP irp); # endif // NT_INCLUDED #endif // _WIN32 #if defined __cplusplus } // extern "C" #endif #endif // _VMCI_SOCKETS_INT_H_ vsock-only/linux/vsockPacket.h 0000444 0000000 0000000 00000014345 13432725350 015456 0 ustar root root /********************************************************* * Copyright (C) 2007-2012 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vsockPacket.h -- * * Packet constants, types and functions. */ #ifndef _VSOCK_PACKET_H_ #define _VSOCK_PACKET_H_ #include "vmci_sockets_packet.h" #if defined(_WIN32) || defined(VMKERNEL) || defined(__APPLE__) || defined(VMX86_VMX) # include "vsockOSInt.h" #else # define VSockOS_ClearMemory(_dst, _sz) memset(_dst, 0, _sz) # define VSockOS_Memcpy(_dst, _src, _sz) memcpy(_dst, _src, _sz) #endif #include "vsockCommon.h" /* *----------------------------------------------------------------------------- * * VSockPacket_Init -- * * Initialize the given packet. The packet version is set and the fields * are filled out. Reserved fields are cleared. * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE void VSockPacket_Init(VSockPacket *pkt, // OUT struct sockaddr_vm *src, // IN struct sockaddr_vm *dst, // IN uint8 type, // IN uint64 size, // IN uint64 mode, // IN VSockWaitingInfo *wait, // IN VSockProtoVersion proto, // IN VMCIHandle handle) // IN { ASSERT(pkt); VSOCK_ADDR_NOFAMILY_ASSERT(src); VSOCK_ADDR_NOFAMILY_ASSERT(dst); /* * We register the stream control handler as an any cid handle so we * must always send from a source address of VMADDR_CID_ANY */ pkt->dg.src = VMCI_MAKE_HANDLE(VMADDR_CID_ANY, VSOCK_PACKET_LOCAL_RID); pkt->dg.dst = VMCI_MAKE_HANDLE(dst->svm_cid, dst->svm_cid == VMCI_HYPERVISOR_CONTEXT_ID ? VSOCK_PACKET_HYPERVISOR_RID : VSOCK_PACKET_RID); pkt->dg.payloadSize = sizeof *pkt - sizeof pkt->dg; pkt->version = VSOCK_PACKET_VERSION; pkt->type = type; pkt->srcPort = src->svm_port; pkt->dstPort = dst->svm_port; VSockOS_ClearMemory(&pkt->proto, sizeof pkt->proto); VSockOS_ClearMemory(&pkt->_reserved2, sizeof pkt->_reserved2); switch (pkt->type) { case VSOCK_PACKET_TYPE_INVALID: pkt->u.size = 0; break; case VSOCK_PACKET_TYPE_REQUEST: case VSOCK_PACKET_TYPE_NEGOTIATE: pkt->u.size = size; break; case VSOCK_PACKET_TYPE_OFFER: case VSOCK_PACKET_TYPE_ATTACH: pkt->u.handle = handle; break; case VSOCK_PACKET_TYPE_WROTE: case VSOCK_PACKET_TYPE_READ: case VSOCK_PACKET_TYPE_RST: pkt->u.size = 0; break; case VSOCK_PACKET_TYPE_SHUTDOWN: pkt->u.mode = mode; break; case VSOCK_PACKET_TYPE_WAITING_READ: case VSOCK_PACKET_TYPE_WAITING_WRITE: ASSERT(wait); VSockOS_Memcpy(&pkt->u.wait, wait, sizeof pkt->u.wait); break; case VSOCK_PACKET_TYPE_REQUEST2: case VSOCK_PACKET_TYPE_NEGOTIATE2: pkt->u.size = size; pkt->proto = proto; break; } VSOCK_PACKET_ASSERT(pkt); } /* *----------------------------------------------------------------------------- * * VSockPacket_Validate -- * * Validate the given packet. * * Results: * 0 on success, EFAULT if the address is invalid, EINVAL if the packet * fields are invalid. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE int32 VSockPacket_Validate(VSockPacket *pkt) { int32 err = EINVAL; if (NULL == pkt) { err = EFAULT; goto exit; } if (VMCI_HANDLE_INVALID(pkt->dg.src)) { goto exit; } if (VMCI_HANDLE_INVALID(pkt->dg.dst)) { goto exit; } if (VMCI_INVALID_ID == pkt->dstPort || VMCI_INVALID_ID == pkt->srcPort) { goto exit; } if (VSOCK_PACKET_VERSION != pkt->version) { goto exit; } /* See the comment above VSOCK_PACKET_ASSERT. */ if (pkt->type < VSOCK_PACKET_TYPE_REQUEST2) { if (0 != pkt->proto || 0 != pkt->_reserved2) { goto exit; } } switch (pkt->type) { case VSOCK_PACKET_TYPE_INVALID: if (0 != pkt->u.size) { goto exit; } break; case VSOCK_PACKET_TYPE_REQUEST: case VSOCK_PACKET_TYPE_NEGOTIATE: if (0 == pkt->u.size) { goto exit; } break; case VSOCK_PACKET_TYPE_OFFER: case VSOCK_PACKET_TYPE_ATTACH: if (VMCI_HANDLE_INVALID(pkt->u.handle)) { goto exit; } break; case VSOCK_PACKET_TYPE_WROTE: case VSOCK_PACKET_TYPE_READ: case VSOCK_PACKET_TYPE_RST: if (0 != pkt->u.size) { goto exit; } break; } err = 0; exit: return sockerr2err(err); } /* *----------------------------------------------------------------------------- * * VSockPacket_GetAddresses -- * * Get the local and remote addresses from the given packet. * * Results: * None. * * Side effects: * None. * *----------------------------------------------------------------------------- */ static INLINE void VSockPacket_GetAddresses(VSockPacket *pkt, // IN struct sockaddr_vm *local, // OUT struct sockaddr_vm *remote) // OUT { VSOCK_PACKET_ASSERT(pkt); VSockAddr_Init(local, VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.dst), pkt->dstPort); VSockAddr_Init(remote, VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.src), pkt->srcPort); } #endif // _VSOCK_PACKET_H_ vsock-only/linux/vsockSocketWrapper.h 0000444 0000000 0000000 00000041160 13432725346 017040 0 ustar root root /********************************************************* * Copyright (C) 2007-2018 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vsockSocketWrapper.h -- * * Socket wrapper constants, types and functions. */ #ifndef _VSOCK_SOCKET_WRAPPER_H_ #define _VSOCK_SOCKET_WRAPPER_H_ /* * Socket states and flags. Note that MSG_WAITALL is only defined on 2K3, * XP-SP2 and above. Since we currently build for 2K to maintain backwards * compatibility, we pull the value from the newer header. Same for the * POLLXXX flags, which are not defined before Vista. */ #if defined(_WIN32) # define MSG_DONTWAIT 0 # define MSG_NOSIGNAL 0 # if (_WIN32_WINNT < 0x0502) # define MSG_WAITALL 0x8 # endif # if (_WIN32_WINNT < 0x0600) # define POLLRDNORM 0x0100 # define POLLRDBAND 0x0200 # define POLLIN (POLLRDNORM | POLLRDBAND) # define POLLPRI 0x0400 # define POLLWRNORM 0x0010 # define POLLOUT (POLLWRNORM) # define POLLWRBAND 0x0020 # define POLLERR 0x0001 # define POLLHUP 0x0002 # define POLLNVAL 0x0004 # endif #endif #if defined __APPLE__ # define MSG_NOSIGNAL 0 /* * Custom options for setting socket behavious in kVsockSetOptions. * These values fall after the common Mac OS X Socket options * in /usr/inclue/sys/socket.h */ #define SO_NONBLOCKING 0x1200 #endif // __APPLE__ #if defined(_WIN32) || defined(VMKERNEL) || defined(__APPLE__) || defined(VMX86_VMX) # define SS_FREE 0 # define SS_UNCONNECTED 1 # define SS_CONNECTING 2 # define SS_CONNECTED 3 # define SS_DISCONNECTING 4 # define SS_DISCONNECTED 5 # define RCV_SHUTDOWN 1 # define SEND_SHUTDOWN 2 # define SHUTDOWN_MASK 3 #endif // _WIN32 || VMKERNEL /* * For signalling sockets. These are defined as standard on Windows. We do * not use them on Linux. So define them here only for VMKernel. */ #if defined(_WIN32) # define SOCKET_EVENT_READ FD_READ # define SOCKET_EVENT_WRITE FD_WRITE # define SOCKET_EVENT_ACCEPT FD_ACCEPT # define SOCKET_EVENT_CONNECT FD_CONNECT # define SOCKET_EVENT_CLOSE FD_CLOSE #else #if defined(VMKERNEL) || defined(__APPLE__) || defined(VMX86_VMX) # define SOCKET_EVENT_READ 0x1 # define SOCKET_EVENT_WRITE 0x2 # define SOCKET_EVENT_ACCEPT 0x8 # define SOCKET_EVENT_CONNECT 0x10 # define SOCKET_EVENT_CLOSE 0x20 #endif // VMKERNEL #endif // _WIN32 /* * Custom socket control option values. These are internal. The public ones * are in vmci_sockets.h. As with the public options, use the address family * as the option level. */ #define SO_VMCI_EVENT_ENUMERATE_SELECT 1000 /* * Error codes. * These defines are needed for crossplatform support. */ #if defined(_WIN32) # define VSOCK_EINTR WSAEINTR # define VSOCK_EACCES WSAEACCES # define VSOCK_EFAULT WSAEFAULT # define VSOCK_EINVAL WSAEINVAL # define VSOCK_EPERM WSAEACCES /* WSA doesn't have VSOCK_EPERM */ # define VSOCK_ENOSYS WSAEOPNOTSUPP # define VSOCK_EAGAIN WSAEWOULDBLOCK # define VSOCK_EWOULDBLOCK WSAEWOULDBLOCK # define VSOCK_EINPROGRESS WSAEINPROGRESS # define VSOCK_EALREADY WSAEALREADY # define VSOCK_ENOTSOCK WSAENOTSOCK # define VSOCK_EDESTADDRREQ WSAEDESTADDRREQ # define VSOCK_EMSGSIZE WSAEMSGSIZE # define VSOCK_EPROTOTYPE WSAEPROTOTYPE # define VSOCK_ENOPROTOOPT WSAENOPROTOOPT # define VSOCK_EPROTONOSUPPORT WSAEPROTONOSUPPORT # define VSOCK_EOPNOTSUPP WSAEOPNOTSUPP # define VSOCK_EAFNOSUPPORT WSAEAFNOSUPPORT # define VSOCK_EADDRINUSE WSAEADDRINUSE # define VSOCK_EADDRNOTAVAIL WSAEADDRNOTAVAIL # define VSOCK_ENETDOWN WSAENETDOWN # define VSOCK_ENETUNREACH WSAENETUNREACH # define VSOCK_ENETRESET WSAENETRESET # define VSOCK_ECONNABORTED WSAECONNABORTED # define VSOCK_ECONNRESET WSAECONNRESET # define VSOCK_ENOBUFS WSAENOBUFS //# define VSOCK_ENOMEM # define VSOCK_EISCONN WSAEISCONN # define VSOCK_ENOTCONN WSAENOTCONN # define VSOCK_ETIMEDOUT WSAETIMEDOUT # define VSOCK_ECONNREFUSED WSAECONNREFUSED # define VSOCK_EHOSTUNREACH WSAEHOSTUNREACH # define VSOCK_ESOCKTNOSUPPORT WSAESOCKTNOSUPPORT # define VSOCK_EPFNOSUPPORT WSAEPFNOSUPPORT # define VSOCK_ESHUTDOWN WSAESHUTDOWN # define VSOCK_EHOSTDOWN WSAEHOSTDOWN // # define VSOCK_EPIPE # define VSOCK__ELOCALSHUTDOWN VSOCK_ESHUTDOWN # define VSOCK__ELOCALRCVSHUTDOWN VSOCK__ELOCALSHUTDOWN # define VSOCK__EPEERSHUTDOWN VSOCK_ECONNABORTED # define VSOCK__ECONNINPROGRESS VSOCK_EWOULDBLOCK # define VSOCK__ESNDRCVTIMEDOUT VSOCK_ETIMEDOUT # define VSOCK_ESYSNOTREADY WSASYSNOTREADY #elif defined(VMKERNEL) # define VSOCK_EINTR VMK_WAIT_INTERRUPTED # define VSOCK_EACCES VMK_NO_ACCESS # define VSOCK_EFAULT VMK_INVALID_ADDRESS # define VSOCK_EINVAL VMK_FAILURE # define VSOCK_EPERM VMK_ACCESS_DENIED // ENOSYS # define VSOCK_EAGAIN VMK_RETRY # define VSOCK_EWOULDBLOCK VMK_WOULD_BLOCK # define VSOCK_EINPROGRESS VMK_EINPROGRESS # define VSOCK_EALREADY VMK_EALREADY # define VSOCK_ENOTSOCK VMK_NOT_A_SOCKET # define VSOCK_EDESTADDRREQ VMK_EDESTADDRREQ /* * Do not change VSOCK_EMSGSIZE definition without changing uses of * VMK_LIMIT_EXCEEDED in userSocketVmci.c's implementation of recvmsg(). */ # define VSOCK_EMSGSIZE VMK_LIMIT_EXCEEDED # define VSOCK_EPROTOTYPE VMK_NOT_SUPPORTED # define VSOCK_ENOPROTOOPT VMK_NOT_SUPPORTED # define VSOCK_EPROTONOSUPPORT VMK_EPROTONOSUPPORT # define VSOCK_EOPNOTSUPP VMK_EOPNOTSUPP # define VSOCK_EAFNOSUPPORT VMK_ADDRFAM_UNSUPP # define VSOCK_EADDRINUSE VMK_EADDRINUSE # define VSOCK_EADDRNOTAVAIL VMK_EADDRNOTAVAIL # define VSOCK_ENETDOWN VMK_ENETDOWN # define VSOCK_ENETUNREACH VMK_ENETUNREACH # define VSOCK_ENETRESET VMK_ENETRESET # define VSOCK_ECONNABORTED VMK_ECONNABORTED # define VSOCK_ECONNRESET VMK_ECONNRESET # define VSOCK_ENOBUFS VMK_NO_MEMORY # define VSOCK_ENOMEM VMK_NO_MEMORY # define VSOCK_EISCONN VMK_ALREADY_CONNECTED # define VSOCK_ENOTCONN VMK_ENOTCONN # define VSOCK_ETIMEDOUT VMK_TIMEOUT # define VSOCK_ECONNREFUSED VMK_ECONNREFUSED # define VSOCK_EHOSTUNREACH VMK_EHOSTUNREACH # define VSOCK_ESOCKTNOSUPPORT VMK_NOT_SUPPORTED # define VSOCK_EPFNOSUPPORT VMK_ADDRFAM_UNSUPP # define VSOCK_ESHUTDOWN VMK_ESHUTDOWN # define VSOCK_EHOSTDOWN VMK_EHOSTDOWN # define VSOCK_EPIPE VMK_BROKEN_PIPE # define VSOCK__ELOCALSHUTDOWN VSOCK_EPIPE # define VSOCK__ELOCALRCVSHUTDOWN 0 # define VSOCK__EPEERSHUTDOWN VSOCK_EPIPE # define VSOCK__ECONNINPROGRESS VSOCK_EINPROGRESS # define VSOCK__ESNDRCVTIMEDOUT VMK_WOULD_BLOCK # define VSOCK_ESYSNOTREADY VMK_NOT_SUPPORTED #else // Linux or Apple # define VSOCK_EINTR EINTR # define VSOCK_EACCES EACCES # define VSOCK_EFAULT EFAULT # define VSOCK_EINVAL EINVAL # define VSOCK_EPERM EPERM # define VSOCK_ENOSYS ENOSYS # define VSOCK_EAGAIN EAGAIN # define VSOCK_EWOULDBLOCK EWOULDBLOCK # define VSOCK_EINPROGRESS EINPROGRESS # define VSOCK_EALREADY EALREADY # define VSOCK_ENOTSOCK ENOTSOCK # define VSOCK_EDESTADDRREQ EDESTADDRREQ # define VSOCK_EMSGSIZE EMSGSIZE # define VSOCK_EPROTOTYPE EPROTOTYPE # define VSOCK_ENOPROTOOPT ENOPROTOOPT # define VSOCK_EPROTONOSUPPORT EPROTONOSUPPORT # define VSOCK_EOPNOTSUPP EOPNOTSUPP # define VSOCK_EAFNOSUPPORT EAFNOSUPPORT # define VSOCK_EADDRINUSE EADDRINUSE # define VSOCK_EADDRNOTAVAIL EADDRNOTAVAIL # define VSOCK_ENETDOWN ENETDOWN # define VSOCK_ENETUNREACH ENETUNREACH # define VSOCK_ENETRESET ENETRESET # define VSOCK_ECONNABORTED ECONNABORTED # define VSOCK_ECONNRESET ECONNRESET # define VSOCK_ENOBUFS ENOBUFS # define VSOCK_ENOMEM ENOMEM # define VSOCK_EISCONN EISCONN # define VSOCK_ENOTCONN ENOTCONN # define VSOCK_ETIMEDOUT ETIMEDOUT # define VSOCK_ECONNREFUSED ECONNREFUSED # define VSOCK_EHOSTUNREACH EHOSTUNREACH # define VSOCK_ESOCKTNOSUPPORT ESOCKTNOSUPPORT # define VSOCK_EPFNOSUPPORT EPFNOSUPPORT # define VSOCK_ESHUTDOWN ESHUTDOWN # define VSOCK_EHOSTDOWN EHOSTDOWN # define VSOCK_EPIPE EPIPE #endif #if defined(__APPLE__) # define VSOCK__ELOCALSHUTDOWN ESHUTDOWN # define VSOCK__ELOCALRCVSHUTDOWN 0 # define VSOCK__EPEERSHUTDOWN ECONNABORTED # define VSOCK__ECONNINPROGRESS EINPROGRESS # define VSOCK__ESNDRCVTIMEDOUT EAGAIN # define VSOCK_ESYSNOTREADY EOPNOTSUPP #elif defined(__linux__) && !defined(VMKERNEL) # define VSOCK__ELOCALSHUTDOWN EPIPE # define VSOCK__ELOCALRCVSHUTDOWN 0 # define VSOCK__EPEERSHUTDOWN EPIPE # define VSOCK__ECONNINPROGRESS EINPROGRESS # define VSOCK__ESNDRCVTIMEDOUT EAGAIN # define VSOCK_ESYSNOTREADY EOPNOTSUPP #endif // _WIN32 /* * XXX - This is used by the vsocktest and common vsock driver code * but should be converted over to use the above crossplatform values * above. When that is completed this section can be removed! */ #if defined(_WIN32) // Some defines are needed for the older SDK. # if !defined(EINTR) # define EINTR WSAEINTR # endif # if !defined(EACCES) # define EACCES WSAEACCES # endif # if !defined(EFAULT) # define EFAULT WSAEFAULT # endif # if !defined(EINVAL) # define EINVAL WSAEINVAL # endif # if !defined(EPERM) # define EPERM WSAEACCES /* WSA doesn't have EPERM */ # endif # if !defined(ENOSYS) # define ENOSYS WSAEOPNOTSUPP # endif # if !defined(EAGAIN) # define EAGAIN WSAEWOULDBLOCK # endif # if !defined(EWOULDBLOCK) # define EWOULDBLOCK WSAEWOULDBLOCK # define EINPROGRESS WSAEINPROGRESS # define EALREADY WSAEALREADY # define ENOTSOCK WSAENOTSOCK # define EDESTADDRREQ WSAEDESTADDRREQ # define EMSGSIZE WSAEMSGSIZE # define EPROTOTYPE WSAEPROTOTYPE # define ENOPROTOOPT WSAENOPROTOOPT # define EPROTONOSUPPORT WSAEPROTONOSUPPORT # define EOPNOTSUPP WSAEOPNOTSUPP # define EAFNOSUPPORT WSAEAFNOSUPPORT # define EADDRINUSE WSAEADDRINUSE # define EADDRNOTAVAIL WSAEADDRNOTAVAIL # define ENETDOWN WSAENETDOWN # define ENETUNREACH WSAENETUNREACH # define ENETRESET WSAENETRESET # define ECONNABORTED WSAECONNABORTED # define ECONNRESET WSAECONNRESET # define ENOBUFS WSAENOBUFS # define EISCONN WSAEISCONN # define ENOTCONN WSAENOTCONN # define ETIMEDOUT WSAETIMEDOUT # define ECONNREFUSED WSAECONNREFUSED # define EHOSTUNREACH WSAEHOSTUNREACH # endif # define ESOCKTNOSUPPORT WSAESOCKTNOSUPPORT # define EPFNOSUPPORT WSAEPFNOSUPPORT # define ESHUTDOWN WSAESHUTDOWN # define EHOSTDOWN WSAEHOSTDOWN # define __ELOCALSHUTDOWN ESHUTDOWN # define __ELOCALRCVSHUTDOWN __ELOCALSHUTDOWN # define __EPEERSHUTDOWN ECONNABORTED # define __ECONNINPROGRESS EWOULDBLOCK # define __ESNDRCVTIMEDOUT ETIMEDOUT # define ESYSNOTREADY WSASYSNOTREADY #elif defined(VMKERNEL) # define EINTR VMK_WAIT_INTERRUPTED # define EPERM VMK_ACCESS_DENIED # define EACCES VMK_NO_ACCESS # define EFAULT VMK_INVALID_ADDRESS # define EINVAL VMK_FAILURE # define EWOULDBLOCK VMK_WOULD_BLOCK # define EINPROGRESS VMK_EINPROGRESS # define EALREADY VMK_EALREADY # define ENOTSOCK VMK_NOT_A_SOCKET # define EDESTADDRREQ VMK_EDESTADDRREQ /* * Do not change EMSGSIZE definition without changing uses of * VMK_LIMIT_EXCEEDED in userSocketVmci.c's implementation of recvmsg(). */ # define EMSGSIZE VMK_LIMIT_EXCEEDED # define EPROTOTYPE VMK_NOT_SUPPORTED # define ENOPROTOOPT VMK_NOT_SUPPORTED # define EPROTONOSUPPORT VMK_EPROTONOSUPPORT # define ESOCKTNOSUPPORT VMK_NOT_SUPPORTED # define EOPNOTSUPP VMK_EOPNOTSUPP # define EPFNOSUPPORT VMK_ADDRFAM_UNSUPP # define EAFNOSUPPORT VMK_ADDRFAM_UNSUPP # define EADDRINUSE VMK_EADDRINUSE # define EADDRNOTAVAIL VMK_EADDRNOTAVAIL # define ENETDOWN VMK_ENETDOWN # define ENETUNREACH VMK_ENETUNREACH # define ENETRESET VMK_ENETRESET # define ECONNABORTED VMK_ECONNABORTED # define ECONNRESET VMK_ECONNRESET # define ENOBUFS VMK_NO_MEMORY # define ENOMEM VMK_NO_MEMORY # define EISCONN VMK_ALREADY_CONNECTED # define ENOTCONN VMK_ENOTCONN # define ESHUTDOWN VMK_ESHUTDOWN # define ETIMEDOUT VMK_TIMEOUT # define ECONNREFUSED VMK_ECONNREFUSED # define EHOSTDOWN VMK_EHOSTDOWN # define EHOSTUNREACH VMK_EHOSTUNREACH # define EPIPE VMK_BROKEN_PIPE # define __ELOCALSHUTDOWN EPIPE # define __ELOCALRCVSHUTDOWN 0 # define __EPEERSHUTDOWN EPIPE # define __ECONNINPROGRESS EINPROGRESS # define __ESNDRCVTIMEDOUT VMK_WOULD_BLOCK # define ESYSNOTREADY VMK_NOT_SUPPORTED # define EAGAIN VMK_RETRY #elif defined(__APPLE__) # define __ELOCALSHUTDOWN ESHUTDOWN # define __ELOCALRCVSHUTDOWN 0 # define __EPEERSHUTDOWN ECONNABORTED # define __ECONNINPROGRESS EINPROGRESS # define __ESNDRCVTIMEDOUT EAGAIN # define ESYSNOTREADY EOPNOTSUPP #elif defined(__linux__) # define ESYSNOTREADY EOPNOTSUPP # define __ELOCALSHUTDOWN EPIPE # define __ELOCALRCVSHUTDOWN 0 # define __EPEERSHUTDOWN EPIPE # define __ECONNINPROGRESS EINPROGRESS # define __ESNDRCVTIMEDOUT EAGAIN #endif // _WIN32 #if defined(_WIN32) # define sockerr() WSAGetLastError() # define sockerr2err(_e) (((_e) < 0) ? -(_e) : (_e)) # define sockcleanup() WSACleanup() typedef uint32 socklen_t; typedef uint32 in_addr_t; #else // _WIN32 #if defined(VMKERNEL) # define SOCKET_ERROR (-1) # define INVALID_SOCKET ((SOCKET) -1) # define sockerr() errno # define sockerr2err(_e) (_e) # define sockcleanup() do {} while (0) # define closesocket(_s) close((_s)) typedef int32 SOCKET; #else #if defined(__linux__) || defined(__APPLE__) # define SOCKET_ERROR (-1) # define INVALID_SOCKET ((SOCKET) -1) # define sockerr() errno # define sockcleanup() do {} while (0) #if defined(__linux__) # define sockerr2err(_e) (((_e) > 0) ? -(_e) : (_e)) # define closesocket(_s) close((_s)) typedef int32 SOCKET; #else # define sockerr2err(_e) (_e) # define closesocket(_s) VMCISock_close(_s) typedef int32 SOCKET; #endif #endif // linux #endif // VMKERNEL #endif // _WIN32 /* * There is no SS_XXX state equivalent to TCP_LISTEN. Linux does have a flag * __SO_ACCEPTCON which some of the socket implementations use, but it does * not fit in the state field (although it is sometimes incorrectly used that * way). So we define our own listen state here for all platforms. */ #define SS_LISTEN 255 /* * Initialize sockets. This is really for platforms that do not have * on-by-default socket implementations like Windows. */ int sockinit(void); #endif // _VSOCK_SOCKET_WRAPPER_H_ vsock-only/linux/vsockCommon.h 0000444 0000000 0000000 00000017636 13432725346 015512 0 ustar root root /********************************************************* * Copyright (C) 2007,2014 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vsockCommon.h -- * * VSockets common constants, types and functions. */ #ifndef _VSOCK_COMMON_H_ #define _VSOCK_COMMON_H_ /* * VMCISockGetAFValueInt is defined separately from VMCISock_GetAFValue because * it is used in several different contexts. In particular it is called from * vsockAddr.c which gets compiled into both our kernel modules as well as * the user level vsock library. In the linux kernel we need different behavior * than external kernel modules using vSockets API inside the kernel. */ #if defined VMX86_VMX /* * The VMX is a very special case because hypervisor sockets do not go * through the host kernel and thus do not require an address family. */ # define VMCI_SOCKETS_AF_VALUE 0 # define VMCISockGetAFValueInt() VMCI_SOCKETS_AF_VALUE #elif defined _WIN32 # define VMCI_SOCKETS_AF_VALUE 28 # if defined WINNT_DDK # define _WIN2K_COMPAT_SLIST_USAGE # include <ntddk.h> # include <windef.h> # define _INC_WINDOWS /* In the kernel we can't call into the provider. */ # define VMCISockGetAFValueInt() VMCI_SOCKETS_AF_VALUE # else // WINNT_DDK /* In userland, just use the normal exported userlevel API. */ # define VMCISockGetAFValueInt() VMCISock_GetAFValue() # include <windows.h> # endif // WINNT_DDK #elif defined VMKERNEL # include "uwvmkAPI.h" # define VMCI_SOCKETS_AF_VALUE AF_VMCI /* Defined in uwvmkAPI.h. */ /* The address family is fixed in the vmkernel. */ # define VMCISockGetAFValueInt() VMCI_SOCKETS_AF_VALUE #elif defined linux # if defined __KERNEL__ /* Include compat_page.h now so PAGE_SIZE and friends don't get redefined. */ # include "driver-config.h" # include "compat_page.h" /* * In the kernel we call back into af_vsock.c to get the address family * being used. Otherwise an ioctl(2) is performed (see vmci_sockets.h). */ extern int VSockVmci_GetAFValue(void); # define VMCISockGetAFValueInt() VSockVmci_GetAFValue() # else // __KERNEL__ /* In userland, just use the normal exported userlevel API. */ # define VMCISockGetAFValueInt() VMCISock_GetAFValue() # endif #elif defined __APPLE__ # if defined KERNEL # include <IOKit/system.h> # define VMCI_SOCKETS_AF_VALUE 14 # define VMCISockGetAFValueInt() VMCI_SOCKETS_AF_VALUE # else // KERNEL # define VMCISockGetAFValueInt() VMCISock_GetAFValue() # endif // KERNEL #endif // __APPLE__ #include "vmware.h" #include "vm_basic_asm.h" #include "vmci_defs.h" #include "vmci_call_defs.h" #include "vmci_infrastructure.h" #include "vmci_sockets_int.h" #include "vmci_sockets.h" #if defined WINNT_DDK # include <winsock2.h> #endif // WINNT_DDK #include "vsockAddr.h" #include "vsockSocketWrapper.h" /* * Local VSocket control packet resource ID. * * Stream sockets to the hypervisor were added later so VSOCK_PACKET_RID was * already assigned to another application. VSOCK_PACKET_HYPERVISOR_RID is * used instead. */ #if defined VMX86_VMX # define VSOCK_PACKET_LOCAL_RID VSOCK_PACKET_HYPERVISOR_RID #else # define VSOCK_PACKET_LOCAL_RID VSOCK_PACKET_RID #endif /* Memory allocation flags. */ #define VSOCK_MEMORY_NORMAL 0 #define VSOCK_MEMORY_ATOMIC (1 << 0) #define VSOCK_MEMORY_NONPAGED (1 << 1) /* *----------------------------------------------------------------------------- * * VSockVA64ToPtr -- * * Convert a VA64 to a pointer. * * Results: * Virtual address. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE void * VSockVA64ToPtr(VA64 va64) // IN { #ifdef VM_64BIT ASSERT_ON_COMPILE(sizeof (void *) == 8); #else ASSERT_ON_COMPILE(sizeof (void *) == 4); // Check that nothing of value will be lost. ASSERT(!(va64 >> 32)); #endif return (void *)(uintptr_t)va64; } /* *----------------------------------------------------------------------------- * * VSockPtrToVA64 -- * * Convert a pointer to a VA64. * * Results: * Virtual address. * * Side effects: * None * *----------------------------------------------------------------------------- */ static INLINE VA64 VSockPtrToVA64(void const *ptr) // IN { ASSERT_ON_COMPILE(sizeof ptr <= sizeof (VA64)); return (VA64)(uintptr_t)ptr; } #if defined(_WIN32) && !defined(WINNT_DDK) /* VSock transport provider structures */ __declspec(selectany) extern const WSAPROTOCOL_INFOW vsockProtocolInfos[] = { { (XP1_CONNECTIONLESS | /* No connection required. */ XP1_MESSAGE_ORIENTED), /* Message. */ 0, /* Reserved. */ 0, /* Reserved. */ 0, /* Reserved. */ PFL_MATCHES_PROTOCOL_ZERO, /* Always protocol 0. */ { 0 }, /* None. */ 0, /* Assigned by Winsock. */ { 1, 0 }, /* Base provider. */ 0, /* Version 0. */ VMCI_SOCKETS_AF_VALUE, /* vSockets protocol. */ 16, /* Maximum address length in bytes. */ 16, /* Minimum address length in bytes. */ SOCK_DGRAM, /* STREAM. */ 0, /* Protocol. */ 0, /* Protocol max offset. */ BIGENDIAN, /* Network byte order. */ SECURITY_PROTOCOL_NONE, /* No security. */ 0, /* Message size unimportant. */ 0, /* None. */ L"vSockets DGRAM" /* Protocol name. */ }, { (XP1_GUARANTEED_DELIVERY | /* Guaranteed delivery. */ XP1_GUARANTEED_ORDER | /* Guaranteed order. */ XP1_GRACEFUL_CLOSE), /* Graceful close. */ 0, /* Reserved. */ 0, /* Reserved. */ 0, /* Reserved. */ PFL_MATCHES_PROTOCOL_ZERO, /* Always protocol 0. */ { 0 }, /* None. */ 0, /* Assigned by Winsock. */ { 1, 0 }, /* Base provider. */ 0, /* Version 0. */ VMCI_SOCKETS_AF_VALUE, /* vSockets protocol. */ 16, /* Maximum address length in bytes. */ 16, /* Minimum address length in bytes. */ SOCK_STREAM, /* STREAM. */ 0, /* Protocol. */ 0, /* Protocol max offset. */ BIGENDIAN, /* Network byte order. */ SECURITY_PROTOCOL_NONE, /* No security. */ 0, /* Message size unimportant. */ 0, /* None. */ L"vSockets STREAM" /* Protocol name. */ }, }; __declspec(selectany) extern const size_t numVSockProtocolInfos = ARRAYSIZE(vsockProtocolInfos); /* {570ADC4B-67B2-42ce-92B2-ACD33D88D842} */ __declspec(selectany) extern const GUID vsockProviderID = { 0x570adc4b, 0x67b2, 0x42ce, { 0x92, 0xb2, 0xac, 0xd3, 0x3d, 0x88, 0xd8, 0x42 } }; #endif // _WIN32 && !WINNT_DDK #endif // _VSOCK_COMMON_H_ vsock-only/linux/af_vsock.c 0000644 0000000 0000000 00000446464 13432725350 015004 0 ustar root root /********************************************************* * Copyright (C) 2007-2015 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * af_vsock.c -- * * Linux socket module for the vSockets protocol family. */ /* * Implementation notes: * * - There are two kinds of sockets: those created by user action (such as * calling socket(2)) and those created by incoming connection request * packets. * * - There are two "global" tables, one for bound sockets (sockets that have * specified an address that they are responsible for) and one for connected * sockets (sockets that have established a connection with another socket). * These tables are "global" in that all sockets on the system are placed * within them. * - Note, though, that the bound table contains an extra entry for a list of * unbound sockets and SOCK_DGRAM sockets will always remain in that list. * The bound table is used solely for lookup of sockets when packets are * received and that's not necessary for SOCK_DGRAM sockets since we create * a datagram handle for each and need not perform a lookup. Keeping * SOCK_DGRAM sockets out of the bound hash buckets will reduce the chance * of collisions when looking for SOCK_STREAM sockets and prevents us from * having to check the socket type in the hash table lookups. * * - Sockets created by user action will either be "client" sockets that * initiate a connection or "server" sockets that listen for connections; we * do not support simultaneous connects (two "client" sockets connecting). * * - "Server" sockets are referred to as listener sockets throughout this * implementation because they are in the SS_LISTEN state. When a connection * request is received (the second kind of socket mentioned above), we create * a new socket and refer to it as a pending socket. These pending sockets * are placed on the pending connection list of the listener socket. When * future packets are received for the address the listener socket is bound * to, we check if the source of the packet is from one that has an existing * pending connection. If it does, we process the packet for the pending * socket. When that socket reaches the connected state, it is removed from * the listener socket's pending list and enqueued in the listener socket's * accept queue. Callers of accept(2) will accept connected sockets from the * listener socket's accept queue. If the socket cannot be accepted for some * reason then it is marked rejected. Once the connection is accepted, it is * owned by the user process and the responsibility for cleanup falls with * that user process. * * - It is possible that these pending sockets will never reach the connected * state; in fact, we may never receive another packet after the connection * request. Because of this, we must schedule a cleanup function to run in * the future, after some amount of time passes where a connection should * have been established. This function ensures that the socket is off all * lists so it cannot be retrieved, then drops all references to the socket * so it is cleaned up (sock_put() -> sk_free() -> our sk_destruct * implementation). Note this function will also cleanup rejected sockets, * those that reach the connected state but leave it before they have been * accepted. * * - Sockets created by user action will be cleaned up when the user * process calls close(2), causing our release implementation to be called. * Our release implementation will perform some cleanup then drop the * last reference so our sk_destruct implementation is invoked. Our * sk_destruct implementation will perform additional cleanup that's common * for both types of sockets. * * - A socket's reference count is what ensures that the structure won't be * freed. Each entry in a list (such as the "global" bound and connected * tables and the listener socket's pending list and connected queue) ensures * a reference. When we defer work until process context and pass a socket * as our argument, we must ensure the reference count is increased to ensure * the socket isn't freed before the function is run; the deferred function * will then drop the reference. * */ #include "driver-config.h" #define EXPORT_SYMTAB #include <linux/kmod.h> #include <linux/socket.h> #include <linux/net.h> #include <linux/skbuff.h> #include <linux/miscdevice.h> #include <linux/poll.h> #include <linux/smp.h> #include <linux/bitops.h> #include <linux/list.h> #include <linux/wait.h> #include <linux/init.h> #include <asm/io.h> #if defined(__x86_64__) && LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 12) # include <linux/ioctl32.h> /* Use weak: not all kernels export sys_ioctl for use by modules */ asmlinkage __attribute__((weak)) long sys_ioctl(unsigned int fd, unsigned int cmd, unsigned long arg); #endif #include "compat_cred.h" #include "compat_module.h" #include "compat_kernel.h" #include "compat_sock.h" #include "compat_version.h" #include "compat_workqueue.h" #include "compat_mutex.h" #include "vmware.h" #include "vsockCommon.h" #include "vsockPacket.h" #include "vsockVmci.h" #include "vmci_iocontrols.h" #include "af_vsock.h" #include "stats.h" #include "util.h" #include "vsock_version.h" #include "driverLog.h" #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 9) # error "Linux kernels before 2.6.9 are not supported." #endif /* * All kernels above 2.6.33 have the kern parameter for the create * call in struct net_proto_family. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 33) && \ !defined(VMW_NETCREATE_KERNARG) # define VMW_NETCREATE_KERNARG #endif #define VSOCK_INVALID_FAMILY NPROTO #define VSOCK_AF_IS_REGISTERED(val) ((val) >= 0 && (val) < NPROTO) /* Some kernel versions don't define __user. Define it ourself if so. */ #ifndef __user #define __user #endif /* * Prototypes */ int VSockVmci_GetAFValue(void); /* Internal functions. */ static Bool VSockVmciProtoToNotifyStruct(struct sock *sk, VSockProtoVersion *proto, Bool oldPktProto); static int VSockVmciGetAFValue(void); static int VSockVmciRecvDgramCB(void *data, VMCIDatagram *dg); static int VSockVmciRecvStreamCB(void *data, VMCIDatagram *dg); static void VSockVmciPeerAttachCB(VMCIId subId, VMCI_EventData *ed, void *clientData); static void VSockVmciPeerDetachCB(VMCIId subId, VMCI_EventData *ed, void *clientData); static void VSockVmciRecvPktWork(compat_work_arg work); static void VSockVmciDelayedSockPut(compat_work_arg work); static int VSockVmciRecvListen(struct sock *sk, VSockPacket *pkt); static int VSockVmciRecvConnectingServer(struct sock *sk, struct sock *pending, VSockPacket *pkt); static int VSockVmciRecvConnectingClient(struct sock *sk, VSockPacket *pkt); static int VSockVmciRecvConnectingClientNegotiate(struct sock *sk, VSockPacket *pkt); static int VSockVmciRecvConnectingClientInvalid(struct sock *sk, VSockPacket *pkt); static int VSockVmciRecvConnected(struct sock *sk, VSockPacket *pkt); static int __VSockVmciBind(struct sock *sk, struct sockaddr_vm *addr); #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 14) static struct sock *__VSockVmciCreate(struct socket *sock, struct sock *parent, unsigned int priority, unsigned short type); #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 24) static struct sock *__VSockVmciCreate(struct socket *sock, struct sock *parent, gfp_t priority, unsigned short type); #else static struct sock *__VSockVmciCreate(struct net *net, struct socket *sock, struct sock *parent, gfp_t priority, unsigned short type); #endif static void VSockVmciTestUnregister(void); static int VSockVmciRegisterWithVmci(void); static void VSockVmciUnregisterWithVmci(void); static int VSockVmciRegisterAddressFamily(void); static void VSockVmciUnregisterAddressFamily(void); /* Socket operations. */ static void VSockVmciSkDestruct(struct sock *sk); static int VSockVmciQueueRcvSkb(struct sock *sk, struct sk_buff *skb); static int VSockVmciRelease(struct socket *sock); static int VSockVmciBind(struct socket *sock, struct sockaddr *addr, int addrLen); static int VSockVmciDgramConnect(struct socket *sock, struct sockaddr *addr, int addrLen, int flags); static int VSockVmciStreamConnect(struct socket *sock, struct sockaddr *addr, int addrLen, int flags); static int VSockVmciAccept(struct socket *sock, struct socket *newsock, int flags); static int VSockVmciGetname(struct socket *sock, struct sockaddr *addr, int *addrLen, int peer); static unsigned int VSockVmciPoll(struct file *file, struct socket *sock, poll_table *wait); static int VSockVmciListen(struct socket *sock, int backlog); static int VSockVmciShutdown(struct socket *sock, int mode); #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 32) typedef int VSockSetsockoptLenType; #else typedef unsigned int VSockSetsockoptLenType; #endif static int VSockVmciStreamSetsockopt(struct socket *sock, int level, int optname, char __user *optval, VSockSetsockoptLenType optlen); static int VSockVmciStreamGetsockopt(struct socket *sock, int level, int optname, char __user *optval, int __user * optlen); static int VSockVmciDgramSendmsg(struct kiocb *kiocb, struct socket *sock, struct msghdr *msg, size_t len); static int VSockVmciDgramRecvmsg(struct kiocb *kiocb, struct socket *sock, struct msghdr *msg, size_t len, int flags); static int VSockVmciStreamSendmsg(struct kiocb *kiocb, struct socket *sock, struct msghdr *msg, size_t len); static int VSockVmciStreamRecvmsg(struct kiocb *kiocb, struct socket *sock, struct msghdr *msg, size_t len, int flags); static int VSockVmciCreate( #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 24) struct net *net, #endif struct socket *sock, int protocol #ifdef VMW_NETCREATE_KERNARG , int kern #endif ); /* * Device operations. */ int VSockVmciDevOpen(struct inode *inode, struct file *file); int VSockVmciDevRelease(struct inode *inode, struct file *file); static int VSockVmciDevIoctl(struct inode *inode, struct file *filp, u_int iocmd, unsigned long ioarg); #if defined(HAVE_COMPAT_IOCTL) || defined(HAVE_UNLOCKED_IOCTL) static long VSockVmciDevUnlockedIoctl(struct file *filp, u_int iocmd, unsigned long ioarg); #endif /* * Variables. */ /* Protocol family. */ static struct proto vsockVmciProto = { .name = "AF_VMCI", #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 10) /* Added in 2.6.10. */ .owner = THIS_MODULE, #endif /* * From 2.6.9 until 2.6.11, these address families called sk_alloc_slab() * and the allocated slab was assigned to the slab variable in the proto * struct and was created of size slab_obj_size. * As of 2.6.12 and later, this slab allocation was moved into * proto_register() and only done if you specified a non-zero value for * the second argument (alloc_slab); the size of the slab element was * changed to obj_size. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 12) .slab_obj_size = sizeof (VSockVmciSock), #else .obj_size = sizeof (VSockVmciSock), #endif }; static struct net_proto_family vsockVmciFamilyOps = { .family = VSOCK_INVALID_FAMILY, .create = VSockVmciCreate, .owner = THIS_MODULE, }; /* Socket operations, split for DGRAM and STREAM sockets. */ static struct proto_ops vsockVmciDgramOps = { .family = VSOCK_INVALID_FAMILY, .owner = THIS_MODULE, .release = VSockVmciRelease, .bind = VSockVmciBind, .connect = VSockVmciDgramConnect, .socketpair = sock_no_socketpair, .accept = sock_no_accept, .getname = VSockVmciGetname, .poll = VSockVmciPoll, .ioctl = sock_no_ioctl, .listen = sock_no_listen, .shutdown = VSockVmciShutdown, .setsockopt = sock_no_setsockopt, .getsockopt = sock_no_getsockopt, .sendmsg = VSockVmciDgramSendmsg, .recvmsg = VSockVmciDgramRecvmsg, .mmap = sock_no_mmap, .sendpage = sock_no_sendpage, }; static struct proto_ops vsockVmciStreamOps = { .family = VSOCK_INVALID_FAMILY, .owner = THIS_MODULE, .release = VSockVmciRelease, .bind = VSockVmciBind, .connect = VSockVmciStreamConnect, .socketpair = sock_no_socketpair, .accept = VSockVmciAccept, .getname = VSockVmciGetname, .poll = VSockVmciPoll, .ioctl = sock_no_ioctl, .listen = VSockVmciListen, .shutdown = VSockVmciShutdown, .setsockopt = VSockVmciStreamSetsockopt, .getsockopt = VSockVmciStreamGetsockopt, .sendmsg = VSockVmciStreamSendmsg, .recvmsg = VSockVmciStreamRecvmsg, .mmap = sock_no_mmap, .sendpage = sock_no_sendpage, }; static struct file_operations vsockVmciDeviceOps = { .owner = THIS_MODULE, #ifdef HAVE_UNLOCKED_IOCTL .unlocked_ioctl = VSockVmciDevUnlockedIoctl, #else .ioctl = VSockVmciDevIoctl, #endif #ifdef HAVE_COMPAT_IOCTL .compat_ioctl = VSockVmciDevUnlockedIoctl, #endif .open = VSockVmciDevOpen, .release = VSockVmciDevRelease, }; static struct miscdevice vsockVmciDevice = { .name = "vsock", .minor = MISC_DYNAMIC_MINOR, .fops = &vsockVmciDeviceOps, }; typedef struct VSockRecvPktInfo { compat_work work; struct sock *sk; VSockPacket pkt; } VSockRecvPktInfo; typedef struct VSockDelayedSockPut { compat_work work; struct sock *sk; } VSockDelayedSockPut; static compat_define_mutex(registrationMutex); static int devOpenCount = 0; static int vsockVmciSocketCount = 0; static int vsockVmciKernClientCount = 0; static Bool vmciDevicePresent = FALSE; static VMCIHandle vmciStreamHandle = { VMCI_INVALID_ID, VMCI_INVALID_ID }; static VMCIId qpResumedSubId = VMCI_INVALID_ID; static int PROTOCOL_OVERRIDE = -1; /* * Netperf benchmarks have shown significant throughput improvements when the * QP size is bumped from 64k to 256k. These measurements were taken during the * K/L.next timeframe. Give users better performance by default. */ #define VSOCK_DEFAULT_QP_SIZE_MIN 128 #define VSOCK_DEFAULT_QP_SIZE 262144 #define VSOCK_DEFAULT_QP_SIZE_MAX 262144 /* * The default peer timeout indicates how long we will wait for a peer * response to a control message. */ #define VSOCK_DEFAULT_CONNECT_TIMEOUT (2 * HZ) #ifdef VMX86_DEVEL # define LOG_PACKET(_pkt) VSockVmciLogPkt(__FUNCTION__, __LINE__, _pkt) #else # define LOG_PACKET(_pkt) #endif /* *---------------------------------------------------------------------------- * * VSockVmciOldProtoOverride -- * * Check to see if the user has asked us to override all sockets to use * the vsock notify protocol. * * Results: * TRUE if there is a protocol override in effect. * - oldPktProto is TRUE the original protocol should be used. * FALSE if there is no override in effect. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static Bool VSockVmciOldProtoOverride(Bool *oldPktProto) // IN { ASSERT(oldPktProto); if (PROTOCOL_OVERRIDE != -1) { if (PROTOCOL_OVERRIDE == 0) { *oldPktProto = TRUE; } else { *oldPktProto = FALSE; } Warning("Proto override in use.\n"); return TRUE; } return FALSE; } /* *---------------------------------------------------------------------------- * * VSockVmciProtoToNotifyStruct -- * * Given a particular notify protocol version, setup the socket's notify * struct correctly. * * Results: * TRUE on success. FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static Bool VSockVmciProtoToNotifyStruct(struct sock *sk, // IN VSockProtoVersion *proto, // IN Bool oldPktProto) // IN { VSockVmciSock *vsk; ASSERT(sk); ASSERT(proto); vsk = vsock_sk(sk); if (oldPktProto) { if (*proto != VSOCK_PROTO_INVALID) { Warning("Can't set both an old and new protocol\n"); return FALSE; } vsk->notifyOps = &vSockVmciNotifyPktOps; goto exit; } switch(*proto) { case VSOCK_PROTO_PKT_ON_NOTIFY: vsk->notifyOps= &vSockVmciNotifyPktQStateOps; break; default: Warning("Unknown notify protocol version\n"); return FALSE; } exit: NOTIFYCALL(vsk, socketInit, sk); return TRUE; } /* *---------------------------------------------------------------------------- * * VSockVmciNewProtoSupportedVersions * * Gets the supported REQUEST2/NEGOTIATE2 vsock protocol versions. * * Results: * Either 1 specific protocol version (override mode) or * VSOCK_PROTO_ALL_SUPPORTED. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static VSockProtoVersion VSockVmciNewProtoSupportedVersions(void) // IN { if (PROTOCOL_OVERRIDE != -1) { return PROTOCOL_OVERRIDE; } return VSOCK_PROTO_ALL_SUPPORTED; } /* *---------------------------------------------------------------------------- * * VSockSocket_Trusted -- * * We allow two kinds of sockets to communicate with a restricted VM: * 1) trusted sockets * 2) sockets from applications running as the same user as the VM (this * is only true for the host side and only when using hosted products) * * Results: * TRUE if trusted communication is allowed to peerCid, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool VSockVmciTrusted(VSockVmciSock *vsock, // IN: Local socket VMCIId peerCid) // IN: Context ID of peer { return vsock->trusted || vmci_is_context_owner(peerCid, vsock->owner); } /* *---------------------------------------------------------------------------- * * VSockSocket_AllowDgram -- * * We allow sending datagrams to and receiving datagrams from a * restricted VM only if it is trusted as described in * VSockVmciTrusted. * * Results: * TRUE if datagram communication is allowed to peerCid, FALSE otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ Bool VSockVmciAllowDgram(VSockVmciSock *vsock, // IN: Local socket VMCIId peerCid) // IN: Context ID of peer { if (peerCid == VMCI_HYPERVISOR_CONTEXT_ID) { return TRUE; } if (vsock->cachedPeer != peerCid) { vsock->cachedPeer = peerCid; if (!VSockVmciTrusted(vsock, peerCid) && (vmci_context_get_priv_flags(peerCid) & VMCI_PRIVILEGE_FLAG_RESTRICTED)) { vsock->cachedPeerAllowDgram = FALSE; } else { vsock->cachedPeerAllowDgram = TRUE; } } return vsock->cachedPeerAllowDgram; } /* *---------------------------------------------------------------------------- * * VMCISock_GetAFValue -- * * Kernel interface that allows external kernel modules to get the current * vSockets address family. * This version of the function is exported to kernel clients and should not * change. * * Results: * The address family on success, a negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int VMCISock_GetAFValue(void) { int afvalue; compat_mutex_lock(®istrationMutex); /* * Kernel clients are required to explicitly register themselves before they * can use vSockets. */ if (vsockVmciKernClientCount <= 0) { afvalue = -1; goto exit; } afvalue = VSockVmciGetAFValue(); exit: compat_mutex_unlock(®istrationMutex); return afvalue; } EXPORT_SYMBOL(VMCISock_GetAFValue); /* *---------------------------------------------------------------------------- * * VMCISock_GetLocalCID -- * * Kernel interface that allows external kernel modules to get the current * VMCI context id. * This version of the function is exported to kernel clients and should not * change. * * Results: * The context id on success, a negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int VMCISock_GetLocalCID(void) { int cid; compat_mutex_lock(®istrationMutex); /* * Kernel clients are required to explicitly register themselves before they * can use vSockets. */ if (vsockVmciKernClientCount <= 0) { cid = -1; goto exit; } cid = vmci_get_context_id(); exit: compat_mutex_unlock(®istrationMutex); return cid; } EXPORT_SYMBOL(VMCISock_GetLocalCID); /* *---------------------------------------------------------------------------- * * VMCISock_KernelRegister -- * * Allows a kernel client to register with vSockets. Must be called * before VMCISock_GetAFValue within a kernel module. Note that we don't * actually register the address family until the first time the module * needs to use it. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ void VMCISock_KernelRegister(void) { compat_mutex_lock(®istrationMutex); vsockVmciKernClientCount++; compat_mutex_unlock(®istrationMutex); } EXPORT_SYMBOL(VMCISock_KernelRegister); /* *---------------------------------------------------------------------------- * * VMCISock_KernelDeregister -- * * Allows a kernel client to unregister with vSockets. Every call * to VMCISock_KernRegister must be matched with a call to * VMCISock_KernUnregister. * * Results: None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ void VMCISock_KernelDeregister(void) { compat_mutex_lock(®istrationMutex); vsockVmciKernClientCount--; VSockVmciTestUnregister(); compat_mutex_unlock(®istrationMutex); } EXPORT_SYMBOL(VMCISock_KernelDeregister); /* *---------------------------------------------------------------------------- * * VSockVmciGetAFValue -- * * Returns the address family value being used. * Note: The registration mutex must be held when calling this function. * * Results: * The address family on success, a negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciGetAFValue(void) { int afvalue; afvalue = vsockVmciFamilyOps.family; if (!VSOCK_AF_IS_REGISTERED(afvalue)) { afvalue = VSockVmciRegisterAddressFamily(); } return afvalue; } /* *---------------------------------------------------------------------------- * * VSockVmci_GetAFValue -- * * Returns the address family value being used. * * Results: * The address family on success, a negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int VSockVmci_GetAFValue(void) { int afvalue; compat_mutex_lock(®istrationMutex); afvalue = VSockVmciGetAFValue(); compat_mutex_unlock(®istrationMutex); return afvalue; } /* * Helper functions. */ /* *---------------------------------------------------------------------------- * * VSockVmciQueuePairAlloc -- * * Allocates or attaches to a queue pair. Tries to register with trusted * status if requested but does not fail if the queuepair could not be * allocate as trusted (running in the guest) * * Results: * 0 on success. A VSock error on error. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciQueuePairAlloc(VMCIQPair **qpair, // OUT VMCIHandle *handle, // IN/OUT uint64 produceSize, // IN uint64 consumeSize, // IN VMCIId peer, // IN uint32 flags, // IN Bool trusted) // IN { int err = 0; if (trusted) { /* * Try to allocate our queue pair as trusted. This will only work * if vsock is running in the host. */ err = vmci_qpair_alloc(qpair, handle, produceSize, consumeSize, peer, flags, VMCI_PRIVILEGE_FLAG_TRUSTED); if (err != VMCI_ERROR_NO_ACCESS) { goto out; } } err = vmci_qpair_alloc(qpair, handle, produceSize, consumeSize, peer, flags, VMCI_NO_PRIVILEGE_FLAGS); out: if (err < 0) { Log("Could not attach to queue pair with %d\n", err); err = VSockVmci_ErrorToVSockError(err); } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciDatagramCreateHnd -- * * Creates a datagram handle. Tries to register with trusted * status but does not fail if the handler could not be allocated * as trusted (running in the guest). * * Results: * 0 on success. A VMCI error on error. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciDatagramCreateHnd(VMCIId resourceID, // IN uint32 flags, // IN VMCIDatagramRecvCB recvCB, // IN void *clientData, // IN VMCIHandle *outHandle) // OUT { int err = 0; /* * Try to allocate our datagram handler as trusted. This will only work * if vsock is running in the host. */ err = vmci_datagram_create_handle_priv(resourceID, flags, VMCI_PRIVILEGE_FLAG_TRUSTED, recvCB, clientData, outHandle); if (err == VMCI_ERROR_NO_ACCESS) { err = vmci_datagram_create_handle(resourceID, flags, recvCB, clientData, outHandle); } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciTestUnregister -- * * Tests if it's necessary to unregister the socket family, and does so. * * Note that this assumes the registration lock is held. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciTestUnregister(void) { if (devOpenCount <= 0 && vsockVmciSocketCount <= 0 && vsockVmciKernClientCount <= 0) { if (VSOCK_AF_IS_REGISTERED(vsockVmciFamilyOps.family)) { VSockVmciUnregisterAddressFamily(); } } } /* *---------------------------------------------------------------------------- * * VSockVmciRecvDgramCB -- * * VMCI Datagram receive callback. This function is used specifically for * SOCK_DGRAM sockets. * * This is invoked as part of a tasklet that's scheduled when the VMCI * interrupt fires. This is run in bottom-half context and if it ever needs * to sleep it should defer that work to a work queue. * * Results: * Zero on success, negative error code on failure. * * Side effects: * An sk_buff is created and queued with this socket. * *---------------------------------------------------------------------------- */ static int VSockVmciRecvDgramCB(void *data, // IN VMCIDatagram *dg) // IN { struct sock *sk; size_t size; struct sk_buff *skb; VSockVmciSock *vsk; ASSERT(dg); ASSERT(dg->payloadSize <= VMCI_MAX_DG_PAYLOAD_SIZE); sk = (struct sock *)data; ASSERT(sk); /* XXX Figure out why sk->sk_socket can be NULL. */ ASSERT(sk->sk_socket ? sk->sk_socket->type == SOCK_DGRAM : 1); /* * This handler is privileged when this module is running on the * host. We will get datagrams from all endpoints (even VMs that * are in a restricted context). If we get one from a restricted * context then the destination socket must be trusted. * * NOTE: We access the socket struct without holding the lock here. This * is ok because the field we are interested is never modified outside * of the create and destruct socket functions. */ vsk = vsock_sk(sk); if (!VSockVmciAllowDgram(vsk, VMCI_HANDLE_TO_CONTEXT_ID(dg->src))) { return VMCI_ERROR_NO_ACCESS; } size = VMCI_DG_SIZE(dg); /* * Attach the packet to the socket's receive queue as an sk_buff. */ skb = alloc_skb(size, GFP_ATOMIC); if (skb) { /* compat_sk_receive_skb() will do a sock_put(), so hold here. */ sock_hold(sk); skb_put(skb, size); memcpy(skb->data, dg, size); compat_sk_receive_skb(sk, skb, 0); } return VMCI_SUCCESS; } /* *---------------------------------------------------------------------------- * * VSockVmciRecvStreamCB -- * * VMCI stream receive callback for control datagrams. This function is * used specifically for SOCK_STREAM sockets. * * This is invoked as part of a tasklet that's scheduled when the VMCI * interrupt fires. This is run in bottom-half context but it defers most * of its work to the packet handling work queue. * * Results: * Zero on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciRecvStreamCB(void *data, // IN VMCIDatagram *dg) // IN { struct sock *sk; struct sockaddr_vm dst; struct sockaddr_vm src; VSockPacket *pkt; VSockVmciSock *vsk; VMCIId expectedSrcRid; Bool bhProcessPkt; int err; VSockDelayedSockPut *delayedSockPut; ASSERT(dg); ASSERT(dg->payloadSize <= VMCI_MAX_DG_PAYLOAD_SIZE); sk = NULL; err = VMCI_SUCCESS; bhProcessPkt = FALSE; /* * Ignore incoming packets from contexts without sockets, or resources that * aren't vsock implementations. */ expectedSrcRid = VMCI_HYPERVISOR_CONTEXT_ID == VMCI_HANDLE_TO_CONTEXT_ID(dg->src) ? VSOCK_PACKET_HYPERVISOR_RID : VSOCK_PACKET_RID; if (!VSockAddr_SocketContextStream(VMCI_HANDLE_TO_CONTEXT_ID(dg->src)) || expectedSrcRid != VMCI_HANDLE_TO_RESOURCE_ID(dg->src)) { return VMCI_ERROR_NO_ACCESS; } if (VMCI_DG_SIZE(dg) < sizeof *pkt) { /* Drop datagrams that do not contain full VSock packets. */ return VMCI_ERROR_INVALID_ARGS; } /* * We need to preallocate this since we otherwise may end up in a situation * where we can't put the socket due to out of memory. */ delayedSockPut = kmalloc(sizeof *delayedSockPut, GFP_ATOMIC); if (!delayedSockPut) { return VMCI_ERROR_NO_MEM; } pkt = (VSockPacket *)dg; LOG_PACKET(pkt); /* * Find the socket that should handle this packet. First we look for * a connected socket and if there is none we look for a socket bound to * the destintation address. * * Note that we don't initialize the family member of the src and dst * sockaddr_vm since we don't want to call VMCISock_GetAFValue() and * possibly register the address family. */ VSockAddr_InitNoFamily(&src, VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.src), pkt->srcPort); VSockAddr_InitNoFamily(&dst, VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.dst), pkt->dstPort); sk = VSockVmciFindConnectedSocket(&src, &dst); if (!sk) { sk = VSockVmciFindBoundSocket(&dst); if (!sk) { /* * We could not find a socket for this specified address. If this * packet is a RST, we just drop it. If it is another packet, we send * a RST. Note that we do not send a RST reply to RSTs so that we do * not continually send RSTs between two endpoints. * * Note that since this is a reply, dst is src and src is dst. */ if (VSOCK_SEND_RESET_BH(&dst, &src, pkt) < 0) { Log("unable to send reset.\n"); } err = VMCI_ERROR_NOT_FOUND; goto out; } } /* * If the received packet type is beyond all types known to this * implementation, reply with an invalid message. Hopefully this will help * when implementing backwards compatibility in the future. */ if (pkt->type >= VSOCK_PACKET_TYPE_MAX) { VSOCK_SEND_INVALID_BH(&dst, &src); err = VMCI_ERROR_INVALID_ARGS; goto out; } /* * This handler is privileged when this module is running on the host. * We will get datagram connect requests from all endpoints (even VMs that * are in a restricted context). If we get one from a restricted context * then the destination socket must be trusted. * * NOTE: We access the socket struct without holding the lock here. This * is ok because the field we are interested is never modified outside * of the create and destruct socket functions. */ vsk = vsock_sk(sk); if (!VSockVmciAllowDgram(vsk, VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.src))) { err = VMCI_ERROR_NO_ACCESS; goto out; } /* * We do most everything in a work queue, but let's fast path the * notification of reads and writes to help data transfer performance. We * can only do this if there is no process context code executing for this * socket since that may change the state. */ bh_lock_sock(sk); if (!sock_owned_by_user(sk)) { /* The local context ID may be out of date. */ vsk->localAddr.svm_cid = dst.svm_cid; if (sk->sk_state == SS_CONNECTED) { NOTIFYCALL(vsk, handleNotifyPkt, sk, pkt, TRUE, &dst, &src, &bhProcessPkt); } } bh_unlock_sock(sk); if (!bhProcessPkt) { VSockRecvPktInfo *recvPktInfo; recvPktInfo = kmalloc(sizeof *recvPktInfo, GFP_ATOMIC); if (!recvPktInfo) { if (VSOCK_SEND_RESET_BH(&dst, &src, pkt) < 0) { Warning("unable to send reset\n"); } err = VMCI_ERROR_NO_MEM; goto out; } recvPktInfo->sk = sk; memcpy(&recvPktInfo->pkt, pkt, sizeof recvPktInfo->pkt); COMPAT_INIT_WORK(&recvPktInfo->work, VSockVmciRecvPktWork, recvPktInfo); compat_schedule_work(&recvPktInfo->work); /* * Clear sk so that the reference count incremented by one of the Find * functions above is not decremented below. We need that reference * count for the packet handler we've scheduled to run. */ sk = NULL; } out: if (sk) { delayedSockPut->sk = sk; COMPAT_INIT_WORK(&delayedSockPut->work, VSockVmciDelayedSockPut, delayedSockPut); compat_schedule_work(&delayedSockPut->work); } else { kfree(delayedSockPut); } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciPeerAttachCB -- * * Invoked when a peer attaches to a queue pair. * * Right now this does not do anything. * * Results: * None. * * Side effects: * May modify socket state and signal socket. * *---------------------------------------------------------------------------- */ static void VSockVmciPeerAttachCB(VMCIId subId, // IN VMCI_EventData *eData, // IN void *clientData) // IN { struct sock *sk; VMCIEventPayload_QP *ePayload; VSockVmciSock *vsk; ASSERT(eData); ASSERT(clientData); sk = (struct sock *)clientData; ePayload = VMCIEventDataPayload(eData); vsk = vsock_sk(sk); /* * We don't ask for delayed CBs when we subscribe to this event (we pass 0 * as flags to VMCIEvent_Subscribe()). VMCI makes no guarantees in that * case about what context we might be running in, so it could be BH or * process, blockable or non-blockable. And bh_lock_sock() is very * particular about how it gets called (it's *not* the same as * spin_lock_bh(), it expands directly into a spin_lock()). So we need to * account for all possible contexts here. */ local_bh_disable(); bh_lock_sock(sk); /* * XXX This is lame, we should provide a way to lookup sockets by qpHandle. */ if (VMCI_HANDLE_EQUAL(vsk->qpHandle, ePayload->handle)) { /* * XXX This doesn't do anything, but in the future we may want to set * a flag here to verify the attach really did occur and we weren't just * sent a datagram claiming it was. */ goto out; } out: bh_unlock_sock(sk); local_bh_enable(); } /* *---------------------------------------------------------------------------- * * VSockVmciHandleDetach -- * * Perform the work necessary when the peer has detached. * * Note that this assumes the socket lock is held. * * Results: * None. * * Side effects: * The socket's and its peer's shutdown mask will be set appropriately, * and any callers waiting on this socket will be awoken. * *---------------------------------------------------------------------------- */ static void VSockVmciHandleDetach(struct sock *sk) // IN { VSockVmciSock *vsk; ASSERT(sk); vsk = vsock_sk(sk); if (!VMCI_HANDLE_INVALID(vsk->qpHandle)) { ASSERT(vsk->qpair); sock_set_flag(sk, SOCK_DONE); /* On a detach the peer will not be sending or receiving anymore. */ vsk->peerShutdown = SHUTDOWN_MASK; /* * We should not be sending anymore since the peer won't be there to * receive, but we can still receive if there is data left in our consume * queue. */ if (VSockVmciStreamHasData(vsk) <= 0) { if (sk->sk_state == SS_CONNECTING) { /* * The peer may detach from a queue pair while we are * still in the connecting state, i.e., if the peer VM is * killed after attaching to a queue pair, but before we * complete the handshake. In that case, we treat the * detach event like a reset. */ sk->sk_state = SS_UNCONNECTED; sk->sk_err = ECONNRESET; sk->sk_error_report(sk); return; } sk->sk_state = SS_UNCONNECTED; } sk->sk_state_change(sk); } } /* *---------------------------------------------------------------------------- * * VSockVmciPeerDetachCB -- * * Invoked when a peer detaches from a queue pair. * * Results: * None. * * Side effects: * May modify socket state and signal socket. * *---------------------------------------------------------------------------- */ static void VSockVmciPeerDetachCB(VMCIId subId, // IN VMCI_EventData *eData, // IN void *clientData) // IN { struct sock *sk; VMCIEventPayload_QP *ePayload; VSockVmciSock *vsk; ASSERT(eData); ASSERT(clientData); sk = (struct sock *)clientData; ePayload = VMCIEventDataPayload(eData); vsk = vsock_sk(sk); if (VMCI_HANDLE_INVALID(ePayload->handle)) { return; } /* Same rules for locking as for PeerAttachCB(). */ local_bh_disable(); bh_lock_sock(sk); /* * XXX This is lame, we should provide a way to lookup sockets by qpHandle. */ if (VMCI_HANDLE_EQUAL(vsk->qpHandle, ePayload->handle)) { VSockVmciHandleDetach(sk); } bh_unlock_sock(sk); local_bh_enable(); } /* *---------------------------------------------------------------------------- * * VSockVmciQPResumedCB -- * * Invoked when a VM is resumed. We must mark all connected stream sockets * as detached. * * Results: * None. * * Side effects: * May modify socket state and signal socket. * *---------------------------------------------------------------------------- */ static void VSockVmciQPResumedCB(VMCIId subId, // IN VMCI_EventData *eData, // IN void *clientData) // IN { uint32 i; spin_lock_bh(&vsockTableLock); /* * XXX This loop should probably be provided by util.{h,c}, but that's for * another day. */ for (i = 0; i < ARRAYSIZE(vsockConnectedTable); i++) { VSockVmciSock *vsk; list_for_each_entry(vsk, &vsockConnectedTable[i], connectedTable) { struct sock *sk = sk_vsock(vsk); /* * XXX Technically this is racy but the resulting outcome from such * a race is relatively harmless. My next change will be a fix to * this. */ VSockVmciHandleDetach(sk); } } spin_unlock_bh(&vsockTableLock); } /* *---------------------------------------------------------------------------- * * VSockVmciPendingWork -- * * Releases the resources for a pending socket if it has not reached the * connected state and been accepted by a user process. * * Results: * None. * * Side effects: * The socket may be removed from the connected list and all its resources * freed. * *---------------------------------------------------------------------------- */ static void VSockVmciPendingWork(compat_delayed_work_arg work) // IN { struct sock *sk; struct sock *listener; VSockVmciSock *vsk; Bool cleanup; vsk = COMPAT_DELAYED_WORK_GET_DATA(work, VSockVmciSock, dwork); ASSERT(vsk); sk = sk_vsock(vsk); listener = vsk->listener; cleanup = TRUE; ASSERT(listener); lock_sock(listener); lock_sock(sk); /* * The socket should be on the pending list or the accept queue, but not * both. It's also possible that the socket isn't on either. */ ASSERT( ( VSockVmciIsPending(sk) && !VSockVmciInAcceptQueue(sk)) || (!VSockVmciIsPending(sk) && VSockVmciInAcceptQueue(sk)) || (!VSockVmciIsPending(sk) && !VSockVmciInAcceptQueue(sk))); if (VSockVmciIsPending(sk)) { VSockVmciRemovePending(listener, sk); } else if (!vsk->rejected) { /* * We are not on the pending list and accept() did not reject us, so we * must have been accepted by our user process. We just need to drop our * references to the sockets and be on our way. */ cleanup = FALSE; goto out; } listener->sk_ack_backlog--; /* * We need to remove ourself from the global connected sockets list so * incoming packets can't find this socket, and to reduce the reference * count. */ if (VSockVmciInConnectedTable(sk)) { VSockVmciRemoveConnected(sk); } sk->sk_state = SS_FREE; out: release_sock(sk); release_sock(listener); if (cleanup) { sock_put(sk); } sock_put(sk); sock_put(listener); } /* *---------------------------------------------------------------------------- * * VSockVmciDelayedSocketPut -- * * Drops a reference to the given socket. * * Results: * None. * * Side effects: * Socket may be freed. * *---------------------------------------------------------------------------- */ static void VSockVmciDelayedSockPut(compat_work_arg work) // IN { VSockDelayedSockPut *delayedSockPut; delayedSockPut = COMPAT_WORK_GET_DATA(work, VSockDelayedSockPut, work); ASSERT(delayedSockPut); sock_put(delayedSockPut->sk); kfree(delayedSockPut); } /* *---------------------------------------------------------------------------- * * VSockVmciRecvPktWork -- * * Handles an incoming control packet for the provided socket. This is the * state machine for our stream sockets. * * Results: * None. * * Side effects: * May set state and wakeup threads waiting for socket state to change. * *---------------------------------------------------------------------------- */ static void VSockVmciRecvPktWork(compat_work_arg work) // IN { VSockRecvPktInfo *recvPktInfo; VSockPacket *pkt; struct sock *sk; recvPktInfo = COMPAT_WORK_GET_DATA(work, VSockRecvPktInfo, work); ASSERT(recvPktInfo); sk = recvPktInfo->sk; pkt = &recvPktInfo->pkt; ASSERT(pkt); ASSERT(pkt->type < VSOCK_PACKET_TYPE_MAX); lock_sock(sk); /* The local context ID may be out of date. */ vsock_sk(sk)->localAddr.svm_cid = VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.dst); switch (sk->sk_state) { case SS_LISTEN: VSockVmciRecvListen(sk, pkt); break; case SS_CONNECTING: /* * Processing of pending connections for servers goes through the * listening socket, so see VSockVmciRecvListen() for that path. */ VSockVmciRecvConnectingClient(sk, pkt); break; case SS_CONNECTED: VSockVmciRecvConnected(sk, pkt); break; default: /* * Because this function does not run in the same context as * VSockVmciRecvStreamCB it is possible that the socket * has closed. We need to let the other side know or it could * be sitting in a connect and hang forever. Send a reset to prevent * that. */ VSOCK_SEND_RESET(sk, pkt); goto out; } out: release_sock(sk); kfree(recvPktInfo); /* * Release reference obtained in the stream callback when we fetched this * socket out of the bound or connected list. */ sock_put(sk); } /* *---------------------------------------------------------------------------- * * VSockVmciRecvListen -- * * Receives packets for sockets in the listen state. * * Note that this assumes the socket lock is held. * * Results: * Zero on success, negative error code on failure. * * Side effects: * A new socket may be created and a negotiate control packet is sent. * *---------------------------------------------------------------------------- */ static int VSockVmciRecvListen(struct sock *sk, // IN VSockPacket *pkt) // IN { struct sock *pending; VSockVmciSock *vpending; int err; uint64 qpSize; Bool oldRequest = FALSE; Bool oldPktProto = FALSE; ASSERT(sk); ASSERT(pkt); ASSERT(sk->sk_state == SS_LISTEN); err = 0; /* * Because we are in the listen state, we could be receiving a packet for * ourself or any previous connection requests that we received. If it's * the latter, we try to find a socket in our list of pending connections * and, if we do, call the appropriate handler for the state that that * socket is in. Otherwise we try to service the connection request. */ pending = VSockVmciGetPending(sk, pkt); if (pending) { lock_sock(pending); /* The local context ID may be out of date. */ vsock_sk(pending)->localAddr.svm_cid = VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.dst); switch (pending->sk_state) { case SS_CONNECTING: err = VSockVmciRecvConnectingServer(sk, pending, pkt); break; default: VSOCK_SEND_RESET(pending, pkt); err = -EINVAL; } if (err < 0) { VSockVmciRemovePending(sk, pending); } release_sock(pending); VSockVmciReleasePending(pending); return err; } /* * The listen state only accepts connection requests. Reply with a reset * unless we received a reset. */ if (!(pkt->type == VSOCK_PACKET_TYPE_REQUEST || pkt->type == VSOCK_PACKET_TYPE_REQUEST2)) { VSOCK_REPLY_RESET(pkt); return -EINVAL; } if (pkt->u.size == 0) { VSOCK_REPLY_RESET(pkt); return -EINVAL; } /* * If this socket can't accommodate this connection request, we send * a reset. Otherwise we create and initialize a child socket and reply * with a connection negotiation. */ if (sk->sk_ack_backlog >= sk->sk_max_ack_backlog) { VSOCK_REPLY_RESET(pkt); return -ECONNREFUSED; } #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 24) pending = __VSockVmciCreate(NULL, sk, GFP_KERNEL, sk->sk_type); #else pending = __VSockVmciCreate(compat_sock_net(sk), NULL, sk, GFP_KERNEL, sk->sk_type); #endif if (!pending) { VSOCK_SEND_RESET(sk, pkt); return -ENOMEM; } vpending = vsock_sk(pending); ASSERT(vpending); ASSERT(vsock_sk(sk)->localAddr.svm_port == pkt->dstPort); VSockAddr_Init(&vpending->localAddr, VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.dst), pkt->dstPort); VSockAddr_Init(&vpending->remoteAddr, VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.src), pkt->srcPort); /* * If the proposed size fits within our min/max, accept * it. Otherwise propose our own size. */ if (pkt->u.size >= vpending->queuePairMinSize && pkt->u.size <= vpending->queuePairMaxSize) { qpSize = pkt->u.size; } else { qpSize = vpending->queuePairSize; } /* * Figure out if we are using old or new requests based on the overrides * pkt types sent by our peer. */ if (VSockVmciOldProtoOverride(&oldPktProto)) { oldRequest = oldPktProto; } else { if (pkt->type == VSOCK_PACKET_TYPE_REQUEST) { oldRequest = TRUE; } else if (pkt->type == VSOCK_PACKET_TYPE_REQUEST2) { oldRequest = FALSE; } } if (oldRequest) { /* Handle a REQUEST (or override) */ VSockProtoVersion version = VSOCK_PROTO_INVALID; if (VSockVmciProtoToNotifyStruct(pending, &version, TRUE)) { err = VSOCK_SEND_NEGOTIATE(pending, qpSize); } else { err = -EINVAL; } } else { /* Handle a REQUEST2 (or override) */ int protoInt = pkt->proto; int pos; uint16 activeProtoVersion = 0; /* * The list of possible protocols is the intersection of all protocols * the client supports ... plus all the protocols we support. */ protoInt &= VSockVmciNewProtoSupportedVersions(); /* We choose the highest possible protocol version and use that one. */ pos = mssb32(protoInt); if (pos) { activeProtoVersion = (1 << (pos - 1)); if (VSockVmciProtoToNotifyStruct(pending, &activeProtoVersion, FALSE)) { err = VSOCK_SEND_NEGOTIATE2(pending, qpSize, activeProtoVersion); } else { err = -EINVAL; } } else { err = -EINVAL; } } if (err < 0) { VSOCK_SEND_RESET(sk, pkt); sock_put(pending); err = VSockVmci_ErrorToVSockError(err); goto out; } VSockVmciAddPending(sk, pending); sk->sk_ack_backlog++; pending->sk_state = SS_CONNECTING; vpending->produceSize = vpending->consumeSize = qpSize; vpending->queuePairSize = qpSize; NOTIFYCALL(vpending, processRequest, pending); /* * We might never receive another message for this socket and it's not * connected to any process, so we have to ensure it gets cleaned up * ourself. Our delayed work function will take care of that. Note that we * do not ever cancel this function since we have few guarantees about its * state when calling cancel_delayed_work(). Instead we hold a reference on * the socket for that function and make it capable of handling cases where * it needs to do nothing but release that reference. */ vpending->listener = sk; sock_hold(sk); sock_hold(pending); COMPAT_INIT_DELAYED_WORK(&vpending->dwork, VSockVmciPendingWork, vpending); compat_schedule_delayed_work(&vpending->dwork, HZ); out: return err; } /* *---------------------------------------------------------------------------- * * VSockVmciRecvConnectingServer -- * * Receives packets for sockets in the connecting state on the server side. * * Connecting sockets on the server side can only receive queue pair offer * packets. All others should be treated as cause for closing the * connection. * * Note that this assumes the socket lock is held for both sk and pending. * * Results: * Zero on success, negative error code on failure. * * Side effects: * A queue pair may be created, an attach control packet may be sent, the * socket may transition to the connected state, and a pending caller in * accept() may be woken up. * *---------------------------------------------------------------------------- */ static int VSockVmciRecvConnectingServer(struct sock *listener, // IN: the listening socket struct sock *pending, // IN: the pending connection VSockPacket *pkt) // IN: current packet { VSockVmciSock *vpending; VMCIHandle handle; VMCIQPair *qpair; Bool isLocal; uint32 flags; VMCIId detachSubId; int err; int skerr; ASSERT(listener); ASSERT(pkt); ASSERT(listener->sk_state == SS_LISTEN); ASSERT(pending->sk_state == SS_CONNECTING); vpending = vsock_sk(pending); detachSubId = VMCI_INVALID_ID; switch (pkt->type) { case VSOCK_PACKET_TYPE_OFFER: if (VMCI_HANDLE_INVALID(pkt->u.handle)) { VSOCK_SEND_RESET(pending, pkt); skerr = EPROTO; err = -EINVAL; goto destroy; } break; default: /* Close and cleanup the connection. */ VSOCK_SEND_RESET(pending, pkt); skerr = EPROTO; err = pkt->type == VSOCK_PACKET_TYPE_RST ? 0 : -EINVAL; goto destroy; } ASSERT(pkt->type == VSOCK_PACKET_TYPE_OFFER); /* * In order to complete the connection we need to attach to the offered * queue pair and send an attach notification. We also subscribe to the * detach event so we know when our peer goes away, and we do that before * attaching so we don't miss an event. If all this succeeds, we update our * state and wakeup anything waiting in accept() for a connection. */ /* * We don't care about attach since we ensure the other side has attached by * specifying the ATTACH_ONLY flag below. */ err = vmci_event_subscribe(VMCI_EVENT_QP_PEER_DETACH, VSockVmciPeerDetachCB, pending, &detachSubId); if (err < VMCI_SUCCESS) { VSOCK_SEND_RESET(pending, pkt); err = VSockVmci_ErrorToVSockError(err); skerr = -err; goto destroy; } vpending->detachSubId = detachSubId; /* Now attach to the queue pair the client created. */ handle = pkt->u.handle; /* * vpending->localAddr always has a context id so we do not * need to worry about VMADDR_CID_ANY in this case. */ isLocal = vpending->remoteAddr.svm_cid == vpending->localAddr.svm_cid; flags = VMCI_QPFLAG_ATTACH_ONLY; flags |= isLocal ? VMCI_QPFLAG_LOCAL : 0; err = VSockVmciQueuePairAlloc(&qpair, &handle, vpending->produceSize, vpending->consumeSize, VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.src), flags, VSockVmciTrusted(vpending, vpending->remoteAddr.svm_cid)); if (err < 0) { VSOCK_SEND_RESET(pending, pkt); skerr = -err; goto destroy; } ASSERT(VMCI_HANDLE_EQUAL(handle, pkt->u.handle)); vpending->qpHandle = handle; vpending->qpair = qpair; /* * When we send the attach message, we must be ready to handle * incoming control messages on the newly connected socket. So we * move the pending socket to the connected state before sending * the attach message. Otherwise, an incoming packet triggered by * the attach being received by the peer may be processed * concurrently with what happens below after sending the attach * message, and that incoming packet will find the listening socket * instead of the (currently) pending socket. Note that enqueueing * the socket increments the reference count, so even if a reset * comes before the connection is accepted, the socket will be * valid until it is removed from the queue. * * If we fail sending the attach below, we remove the socket from * the connected list and move the socket to SS_UNCONNECTED before * releasing the lock, so a pending slow path processing of an * incoming packet will not see the socket in the connected state * in that case. */ pending->sk_state = SS_CONNECTED; VSockVmciInsertConnected(vsockConnectedSocketsVsk(vpending), pending); /* Notify our peer of our attach. */ err = VSOCK_SEND_ATTACH(pending, handle); if (err < 0) { VSockVmciRemoveConnected(pending); Log("Could not send attach\n"); VSOCK_SEND_RESET(pending, pkt); err = VSockVmci_ErrorToVSockError(err); skerr = -err; goto destroy; } /* * We have a connection. Move the now connected socket from the * listener's pending list to the accept queue so callers of * accept() can find it. */ VSockVmciRemovePending(listener, pending); VSockVmciEnqueueAccept(listener, pending); /* * Callers of accept() will be be waiting on the listening socket, not the * pending socket. */ listener->sk_state_change(listener); return 0; destroy: pending->sk_err = skerr; pending->sk_state = SS_UNCONNECTED; /* * As long as we drop our reference, all necessary cleanup will handle when * the cleanup function drops its reference and our destruct implementation * is called. Note that since the listen handler will remove pending from * the pending list upon our failure, the cleanup function won't drop the * additional reference, which is why we do it here. */ sock_put(pending); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciRecvConnectingClient -- * * Receives packets for sockets in the connecting state on the client side. * * Connecting sockets on the client side should only receive attach packets. * All others should be treated as cause for closing the connection. * * Note that this assumes the socket lock is held for both sk and pending. * * Results: * Zero on success, negative error code on failure. * * Side effects: * The socket may transition to the connected state and wakeup the pending * caller of connect(). * *---------------------------------------------------------------------------- */ static int VSockVmciRecvConnectingClient(struct sock *sk, // IN: socket VSockPacket *pkt) // IN: current packet { VSockVmciSock *vsk; int err; int skerr; ASSERT(sk); ASSERT(pkt); ASSERT(sk->sk_state == SS_CONNECTING); vsk = vsock_sk(sk); switch (pkt->type) { case VSOCK_PACKET_TYPE_ATTACH: if (VMCI_HANDLE_INVALID(pkt->u.handle) || !VMCI_HANDLE_EQUAL(pkt->u.handle, vsk->qpHandle)) { skerr = EPROTO; err = -EINVAL; goto destroy; } /* * Signify the socket is connected and wakeup the waiter in connect(). * Also place the socket in the connected table for accounting (it can * already be found since it's in the bound table). */ sk->sk_state = SS_CONNECTED; sk->sk_socket->state = SS_CONNECTED; VSockVmciInsertConnected(vsockConnectedSocketsVsk(vsk), sk); sk->sk_state_change(sk); break; case VSOCK_PACKET_TYPE_NEGOTIATE: case VSOCK_PACKET_TYPE_NEGOTIATE2: if (pkt->u.size == 0 || VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.src) != vsk->remoteAddr.svm_cid || pkt->srcPort != vsk->remoteAddr.svm_port || !VMCI_HANDLE_INVALID(vsk->qpHandle) || vsk->qpair || vsk->produceSize != 0 || vsk->consumeSize != 0 || vsk->attachSubId != VMCI_INVALID_ID || vsk->detachSubId != VMCI_INVALID_ID) { skerr = EPROTO; err = -EINVAL; goto destroy; } err = VSockVmciRecvConnectingClientNegotiate(sk, pkt); if (err) { skerr = -err; goto destroy; } break; case VSOCK_PACKET_TYPE_INVALID: err = VSockVmciRecvConnectingClientInvalid(sk, pkt); if (err) { skerr = -err; goto destroy; } break; case VSOCK_PACKET_TYPE_RST: /* * Older versions of the linux code (WS 6.5 / ESX 4.0) used to continue * processing here after they sent an INVALID packet. This meant that we * got a RST after the INVALID. We ignore a RST after an INVALID. The * common code doesn't send the RST ... so we can hang if an old version * of the common code fails between getting a REQUEST and sending an * OFFER back. Not much we can do about it... except hope that it * doesn't happen. */ if (vsk->ignoreConnectingRst) { vsk->ignoreConnectingRst = FALSE; } else { skerr = ECONNRESET; err = 0; goto destroy; } break; default: /* Close and cleanup the connection. */ skerr = EPROTO; err = -EINVAL; goto destroy; } ASSERT(pkt->type == VSOCK_PACKET_TYPE_ATTACH || pkt->type == VSOCK_PACKET_TYPE_NEGOTIATE || pkt->type == VSOCK_PACKET_TYPE_NEGOTIATE2 || pkt->type == VSOCK_PACKET_TYPE_INVALID || pkt->type == VSOCK_PACKET_TYPE_RST); return 0; destroy: VSOCK_SEND_RESET(sk, pkt); sk->sk_state = SS_UNCONNECTED; sk->sk_err = skerr; sk->sk_error_report(sk); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciRecvConnectingClientNegotiate -- * * Handles a negotiate packet for a client in the connecting state. * * Note that this assumes the socket lock is held for both sk and pending. * * Results: * Zero on success, negative error code on failure. * * Side effects: * The socket may transition to the connected state and wakeup the pending * caller of connect(). * *---------------------------------------------------------------------------- */ static int VSockVmciRecvConnectingClientNegotiate(struct sock *sk, // IN: socket VSockPacket *pkt) // IN: current packet { int err; VSockVmciSock *vsk; VMCIHandle handle; VMCIQPair *qpair; VMCIId attachSubId; VMCIId detachSubId; Bool isLocal; uint32 flags; Bool oldProto = TRUE; Bool oldPktProto; VSockProtoVersion version; vsk = vsock_sk(sk); handle = VMCI_INVALID_HANDLE; attachSubId = VMCI_INVALID_ID; detachSubId = VMCI_INVALID_ID; ASSERT(sk); ASSERT(pkt); ASSERT(pkt->u.size > 0); ASSERT(vsk->remoteAddr.svm_cid == VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.src)); ASSERT(vsk->remoteAddr.svm_port == pkt->srcPort); ASSERT(VMCI_HANDLE_INVALID(vsk->qpHandle)); ASSERT(vsk->qpair == NULL); ASSERT(vsk->produceSize == 0); ASSERT(vsk->consumeSize == 0); ASSERT(vsk->attachSubId == VMCI_INVALID_ID); ASSERT(vsk->detachSubId == VMCI_INVALID_ID); /* * If we have gotten here then we should be past the point where old linux * vsock could have sent the bogus rst. */ vsk->sentRequest = FALSE; vsk->ignoreConnectingRst = FALSE; /* Verify that we're OK with the proposed queue pair size */ if (pkt->u.size < vsk->queuePairMinSize || pkt->u.size > vsk->queuePairMaxSize) { err = -EINVAL; goto destroy; } /* * At this point we know the CID the peer is using to talk to us. */ if (vsk->localAddr.svm_cid == VMADDR_CID_ANY) { vsk->localAddr.svm_cid = VMCI_HANDLE_TO_CONTEXT_ID(pkt->dg.dst); } /* * Setup the notify ops to be the highest supported version that both the * server and the client support. */ if (VSockVmciOldProtoOverride(&oldPktProto)) { oldProto = oldPktProto; } else { if (pkt->type == VSOCK_PACKET_TYPE_NEGOTIATE) { oldProto = TRUE; } else if (pkt->type == VSOCK_PACKET_TYPE_NEGOTIATE2) { oldProto = FALSE; } } if (oldProto) { version = VSOCK_PROTO_INVALID; } else { version = pkt->proto; } if (!VSockVmciProtoToNotifyStruct(sk, &version, oldProto)) { err = -EINVAL; goto destroy; } /* * Subscribe to attach and detach events first. * * XXX We attach once for each queue pair created for now so it is easy * to find the socket (it's provided), but later we should only subscribe * once and add a way to lookup sockets by queue pair handle. */ err = vmci_event_subscribe(VMCI_EVENT_QP_PEER_ATTACH, VSockVmciPeerAttachCB, sk, &attachSubId); if (err < VMCI_SUCCESS) { err = VSockVmci_ErrorToVSockError(err); goto destroy; } err = vmci_event_subscribe(VMCI_EVENT_QP_PEER_DETACH, VSockVmciPeerDetachCB, sk, &detachSubId); if (err < VMCI_SUCCESS) { err = VSockVmci_ErrorToVSockError(err); goto destroy; } /* Make VMCI select the handle for us. */ handle = VMCI_INVALID_HANDLE; isLocal = vsk->remoteAddr.svm_cid == vsk->localAddr.svm_cid; flags = isLocal ? VMCI_QPFLAG_LOCAL : 0; err = VSockVmciQueuePairAlloc(&qpair, &handle, pkt->u.size, pkt->u.size, vsk->remoteAddr.svm_cid, flags, VSockVmciTrusted(vsk, vsk->remoteAddr.svm_cid)); if (err < 0) { goto destroy; } err = VSOCK_SEND_QP_OFFER(sk, handle); if (err < 0) { err = VSockVmci_ErrorToVSockError(err); goto destroy; } vsk->qpHandle = handle; vsk->qpair = qpair; vsk->produceSize = vsk->consumeSize = pkt->u.size; vsk->attachSubId = attachSubId; vsk->detachSubId = detachSubId; NOTIFYCALL(vsk, processNegotiate, sk); return 0; destroy: if (attachSubId != VMCI_INVALID_ID) { vmci_event_unsubscribe(attachSubId); ASSERT(vsk->attachSubId == VMCI_INVALID_ID); } if (detachSubId != VMCI_INVALID_ID) { vmci_event_unsubscribe(detachSubId); ASSERT(vsk->detachSubId == VMCI_INVALID_ID); } if (!VMCI_HANDLE_INVALID(handle)) { ASSERT(vsk->qpair); vmci_qpair_detach(&qpair); ASSERT(VMCI_HANDLE_INVALID(vsk->qpHandle)); } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciRecvConnectingClientInvalid -- * * Handles an invalid packet for a client in the connecting state. * * Note that this assumes the socket lock is held for both sk and pending. * * Results: * Zero on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciRecvConnectingClientInvalid(struct sock *sk, // IN: socket VSockPacket *pkt) // IN: current packet { int err = 0; VSockVmciSock *vsk; ASSERT(sk); ASSERT(pkt); vsk = vsock_sk(sk); if (vsk->sentRequest) { vsk->sentRequest = FALSE; vsk->ignoreConnectingRst = TRUE; err = VSOCK_SEND_CONN_REQUEST(sk, vsk->queuePairSize); if (err < 0) { err = VSockVmci_ErrorToVSockError(err); } else { err = 0; } } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciRecvConnected -- * * Receives packets for sockets in the connected state. * * Connected sockets should only ever receive detach, wrote, read, or reset * control messages. Others are treated as errors that are ignored. * * Wrote and read signify that the peer has produced or consumed, * respectively. * * Detach messages signify that the connection is being closed cleanly and * reset messages signify that the connection is being closed in error. * * Note that this assumes the socket lock is held. * * Results: * Zero on success, negative error code on failure. * * Side effects: * A queue pair may be created, an offer control packet sent, and the socket * may transition to the connecting state. * * *---------------------------------------------------------------------------- */ static int VSockVmciRecvConnected(struct sock *sk, // IN VSockPacket *pkt) // IN { VSockVmciSock *vsk; Bool pktProcessed = FALSE; ASSERT(sk); ASSERT(pkt); ASSERT(sk->sk_state == SS_CONNECTED); /* * In cases where we are closing the connection, it's sufficient to mark * the state change (and maybe error) and wake up any waiting threads. * Since this is a connected socket, it's owned by a user process and will * be cleaned up when the failure is passed back on the current or next * system call. Our system call implementations must therefore check for * error and state changes on entry and when being awoken. */ switch (pkt->type) { case VSOCK_PACKET_TYPE_SHUTDOWN: if (pkt->u.mode) { vsk = vsock_sk(sk); vsk->peerShutdown |= pkt->u.mode; sk->sk_state_change(sk); } break; case VSOCK_PACKET_TYPE_RST: vsk = vsock_sk(sk); /* * It is possible that we sent our peer a message (e.g * a WAITING_READ) right before we got notified that the peer * had detached. If that happens then we can get a RST pkt back * from our peer even though there is data available for us * to read. In that case, don't shutdown the socket completely * but instead allow the local client to finish reading data * off the queuepair. Always treat a RST pkt in connected mode * like a clean shutdown. */ sock_set_flag(sk, SOCK_DONE); vsk->peerShutdown = SHUTDOWN_MASK; if (VSockVmciStreamHasData(vsk) <= 0) { sk->sk_state = SS_DISCONNECTING; } sk->sk_state_change(sk); break; default: vsk = vsock_sk(sk); NOTIFYCALL(vsk, handleNotifyPkt, sk, pkt, FALSE, NULL, NULL, &pktProcessed); if (!pktProcessed) { return -EINVAL; } break; } return 0; } /* *---------------------------------------------------------------------------- * * __VSockVmciSendControlPkt -- * * Common code to send a control packet. * * Results: * Size of datagram sent on success, negative error code otherwise. * If convertError is TRUE, error code is a vsock error, otherwise, * result is a VMCI error code. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int __VSockVmciSendControlPkt(VSockPacket *pkt, // IN struct sockaddr_vm *src, // IN struct sockaddr_vm *dst, // IN VSockPacketType type, // IN uint64 size, // IN uint64 mode, // IN VSockWaitingInfo *wait, // IN VSockProtoVersion proto, // IN VMCIHandle handle, // IN Bool convertError) // IN { int err; ASSERT(pkt); /* * This function can be called in different contexts, so family value is not * necessarily consistent. */ VSOCK_ADDR_NOFAMILY_ASSERT(src); VSOCK_ADDR_NOFAMILY_ASSERT(dst); VSockPacket_Init(pkt, src, dst, type, size, mode, wait, proto, handle); LOG_PACKET(pkt); VSOCK_STATS_CTLPKT_LOG(pkt->type); err = vmci_datagram_send(&pkt->dg); if (convertError && (err < 0)) { return VSockVmci_ErrorToVSockError(err); } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciReplyControlPktFast -- * * Sends a control packet back to the source of an incoming packet. * The control packet is allocated in the stack. * * Results: * Size of datagram sent on success, negative error code otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int VSockVmciReplyControlPktFast(VSockPacket *pkt, // IN VSockPacketType type, // IN uint64 size, // IN uint64 mode, // IN VSockWaitingInfo *wait, // IN VMCIHandle handle) // IN { VSockPacket reply; struct sockaddr_vm src, dst; ASSERT(pkt); if (pkt->type == VSOCK_PACKET_TYPE_RST) { return 0; } else { VSockPacket_GetAddresses(pkt, &src, &dst); return __VSockVmciSendControlPkt(&reply, &src, &dst, type, size, mode, wait, VSOCK_PROTO_INVALID, handle, TRUE); } } /* *---------------------------------------------------------------------------- * * VSockVmciSendControlPktBH -- * * Sends a control packet from bottom-half context. The control packet is * static data to minimize the resource cost. * * Results: * Size of datagram sent on success, negative error code otherwise. Note * that we return a VMCI error message since that's what callers will need * to provide. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int VSockVmciSendControlPktBH(struct sockaddr_vm *src, // IN struct sockaddr_vm *dst, // IN VSockPacketType type, // IN uint64 size, // IN uint64 mode, // IN VSockWaitingInfo *wait, // IN VMCIHandle handle) // IN { /* * Note that it is safe to use a single packet across all CPUs since two * tasklets of the same type are guaranteed to not ever run simultaneously. * If that ever changes, or VMCI stops using tasklets, we can use per-cpu * packets. */ static VSockPacket pkt; return __VSockVmciSendControlPkt(&pkt, src, dst, type, size, mode, wait, VSOCK_PROTO_INVALID, handle, FALSE); } /* *---------------------------------------------------------------------------- * * VSockVmciSendControlPkt -- * * Sends a control packet. * * Results: * Size of datagram sent on success, negative error on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int VSockVmciSendControlPkt(struct sock *sk, // IN VSockPacketType type, // IN uint64 size, // IN uint64 mode, // IN VSockWaitingInfo *wait, // IN VSockProtoVersion proto, // IN VMCIHandle handle) // IN { VSockPacket *pkt; VSockVmciSock *vsk; int err; ASSERT(sk); /* * New sockets for connection establishment won't have socket structures * yet; if one exists, ensure it is of the proper type. */ ASSERT(sk->sk_socket ? sk->sk_socket->type == SOCK_STREAM : 1); vsk = vsock_sk(sk); if (!VSockAddr_Bound(&vsk->localAddr)) { return -EINVAL; } if (!VSockAddr_Bound(&vsk->remoteAddr)) { return -EINVAL; } pkt = kmalloc(sizeof *pkt, GFP_KERNEL); if (!pkt) { return -ENOMEM; } err = __VSockVmciSendControlPkt(pkt, &vsk->localAddr, &vsk->remoteAddr, type, size, mode, wait, proto, handle, TRUE); kfree(pkt); return err; } /* *---------------------------------------------------------------------------- * * __VSockVmciBind -- * * Common functionality needed to bind the specified address to the * VSocket. If VMADDR_CID_ANY or VMADDR_PORT_ANY are specified, the context * ID or port are selected automatically. * * Results: * Zero on success, negative error code on failure. * * Side effects: * On success, a new datagram handle is created. * *---------------------------------------------------------------------------- */ static int __VSockVmciBind(struct sock *sk, // IN/OUT struct sockaddr_vm *addr) // IN { static unsigned int port = LAST_RESERVED_PORT + 1; struct sockaddr_vm newAddr; VSockVmciSock *vsk; VMCIId cid; int err; ASSERT(sk); ASSERT(sk->sk_socket); ASSERT(addr); vsk = vsock_sk(sk); /* First ensure this socket isn't already bound. */ if (VSockAddr_Bound(&vsk->localAddr)) { return -EINVAL; } /* * Now bind to the provided address or select appropriate values if none are * provided (VMADDR_CID_ANY and VMADDR_PORT_ANY). Note that like AF_INET * prevents binding to a non-local IP address (in most cases), we only allow * binding to the local CID. */ VSockAddr_Init(&newAddr, VMADDR_CID_ANY, VMADDR_PORT_ANY); cid = vmci_get_context_id(); if (addr->svm_cid != cid && addr->svm_cid != VMADDR_CID_ANY) { return -EADDRNOTAVAIL; } newAddr.svm_cid = addr->svm_cid; switch (sk->sk_socket->type) { case SOCK_STREAM: { spin_lock_bh(&vsockTableLock); if (addr->svm_port == VMADDR_PORT_ANY) { Bool found = FALSE; unsigned int i; for (i = 0; i < MAX_PORT_RETRIES; i++) { if (port <= LAST_RESERVED_PORT) { port = LAST_RESERVED_PORT + 1; } newAddr.svm_port = port++; if (!__VSockVmciFindBoundSocket(&newAddr)) { found = TRUE; break; } } if (!found) { err = -EADDRNOTAVAIL; goto out; } } else { /* If port is in reserved range, ensure caller has necessary privileges. */ if (addr->svm_port <= LAST_RESERVED_PORT && !capable(CAP_NET_BIND_SERVICE)) { err = -EACCES; goto out; } newAddr.svm_port = addr->svm_port; if (__VSockVmciFindBoundSocket(&newAddr)) { err = -EADDRINUSE; goto out; } } break; } case SOCK_DGRAM: { uint32 flags = 0; /* VMCI will select a resource ID for us if we provide VMCI_INVALID_ID. */ newAddr.svm_port = addr->svm_port == VMADDR_PORT_ANY ? VMCI_INVALID_ID : addr->svm_port; if (newAddr.svm_port <= LAST_RESERVED_PORT && !capable(CAP_NET_BIND_SERVICE)) { err = -EACCES; goto out; } if (newAddr.svm_cid == VMADDR_CID_ANY) { flags = VMCI_FLAG_ANYCID_DG_HND; } err = VSockVmciDatagramCreateHnd(newAddr.svm_port, flags, VSockVmciRecvDgramCB, sk, &vsk->dgHandle); if (err < VMCI_SUCCESS) { err = VSockVmci_ErrorToVSockError(err); goto out; } newAddr.svm_port = VMCI_HANDLE_TO_RESOURCE_ID(vsk->dgHandle); break; } default: { err = -EINVAL; goto out; } } /* * VSockVmci_GetAFValue() acquires a mutex and may sleep, so fill the * field after unlocking socket tables. */ VSockAddr_InitNoFamily(&vsk->localAddr, newAddr.svm_cid, newAddr.svm_port); /* * Remove stream sockets from the unbound list and add them to the hash * table for easy lookup by its address. The unbound list is simply an * extra entry at the end of the hash table, a trick used by AF_UNIX. */ if (sk->sk_socket->type == SOCK_STREAM) { __VSockVmciRemoveBound(sk); __VSockVmciInsertBound(vsockBoundSockets(&vsk->localAddr), sk); spin_unlock_bh(&vsockTableLock); } vsk->localAddr.svm_family = VSockVmci_GetAFValue(); VSOCK_ADDR_ASSERT(&vsk->localAddr); return 0; out: if (sk->sk_socket->type == SOCK_STREAM) { spin_unlock_bh(&vsockTableLock); } return err; } /* *---------------------------------------------------------------------------- * * __VSockVmciCreate -- * * Does the work to create the sock structure. * Note: If sock is NULL then the type field must be non-zero. * Otherwise, sock is non-NULL and the type of sock is used in the * newly created socket. * * Results: * sock structure on success, NULL on failure. * * Side effects: * Allocated sk is added to the unbound sockets list iff it is owned by * a struct socket. * *---------------------------------------------------------------------------- */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 14) static struct sock * __VSockVmciCreate(struct socket *sock, // IN: Owning socket, may be NULL struct sock *parent, // IN: Parent socket, may be NULL unsigned int priority, // IN: Allocation flags unsigned short type) // IN: Socket type if sock is NULL #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 24) static struct sock * __VSockVmciCreate(struct socket *sock, // IN: Owning socket, may be NULL struct sock *parent, // IN: Parent socket, may be NULL gfp_t priority, // IN: Allocation flags unsigned short type) // IN: Socket type if sock is NULL #else static struct sock * __VSockVmciCreate(struct net *net, // IN: Network namespace struct socket *sock, // IN: Owning socket, may be NULL struct sock *parent, // IN: Parent socket, may be NULL gfp_t priority, // IN: Allocation flags unsigned short type) // IN: Socket type if sock is NULL #endif { struct sock *sk; VSockVmciSock *psk; VSockVmciSock *vsk; ASSERT((sock && !type) || (!sock && type)); vsk = NULL; /* * From 2.6.9 to until 2.6.12 sk_alloc() used a cache in * the protocol structure, but you still had to specify the size and cache * yourself. * Most recently (in 2.6.24), sk_alloc() was changed to expect the * network namespace, and the option to zero the sock was dropped. * */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 12) sk = sk_alloc(vsockVmciFamilyOps.family, priority, vsockVmciProto.slab_obj_size, vsockVmciProto.slab); #elif LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 24) sk = sk_alloc(vsockVmciFamilyOps.family, priority, &vsockVmciProto, 1); #else sk = sk_alloc(net, vsockVmciFamilyOps.family, priority, &vsockVmciProto); #endif if (!sk) { return NULL; } /* * If we go this far, we know the socket family is registered, so there's no * need to register it now. */ compat_mutex_lock(®istrationMutex); vsockVmciSocketCount++; compat_mutex_unlock(®istrationMutex); sock_init_data(sock, sk); /* * sk->sk_type is normally set in sock_init_data, but only if * sock is non-NULL. We make sure that our sockets always have a type * by setting it here if needed. */ if (!sock) { sk->sk_type = type; } vsk = vsock_sk(sk); VSockAddr_Init(&vsk->localAddr, VMADDR_CID_ANY, VMADDR_PORT_ANY); VSockAddr_Init(&vsk->remoteAddr, VMADDR_CID_ANY, VMADDR_PORT_ANY); sk->sk_destruct = VSockVmciSkDestruct; sk->sk_backlog_rcv = VSockVmciQueueRcvSkb; sk->sk_state = 0; sock_reset_flag(sk, SOCK_DONE); INIT_LIST_HEAD(&vsk->boundTable); INIT_LIST_HEAD(&vsk->connectedTable); vsk->dgHandle = VMCI_INVALID_HANDLE; vsk->qpHandle = VMCI_INVALID_HANDLE; vsk->qpair = NULL; vsk->produceSize = vsk->consumeSize = 0; vsk->listener = NULL; INIT_LIST_HEAD(&vsk->pendingLinks); INIT_LIST_HEAD(&vsk->acceptQueue); vsk->rejected = FALSE; vsk->sentRequest = FALSE; vsk->ignoreConnectingRst = FALSE; vsk->attachSubId = vsk->detachSubId = VMCI_INVALID_ID; vsk->peerShutdown = 0; if (parent) { psk = vsock_sk(parent); vsk->trusted = psk->trusted; vsk->owner = psk->owner; vsk->queuePairSize = psk->queuePairSize; vsk->queuePairMinSize = psk->queuePairMinSize; vsk->queuePairMaxSize = psk->queuePairMaxSize; vsk->connectTimeout = psk->connectTimeout; } else { vsk->trusted = capable(CAP_NET_ADMIN); vsk->owner = current_uid(); vsk->queuePairSize = VSOCK_DEFAULT_QP_SIZE; vsk->queuePairMinSize = VSOCK_DEFAULT_QP_SIZE_MIN; vsk->queuePairMaxSize = VSOCK_DEFAULT_QP_SIZE_MAX; vsk->connectTimeout = VSOCK_DEFAULT_CONNECT_TIMEOUT; } vsk->notifyOps = NULL; if (sock) { VSockVmciInsertBound(vsockUnboundSockets, sk); } return sk; } /* *---------------------------------------------------------------------------- * * __VSockVmciRelease -- * * Releases the provided socket. * * Results: * None. * * Side effects: * Any pending sockets are also released. * *---------------------------------------------------------------------------- */ static void __VSockVmciRelease(struct sock *sk) // IN { if (sk) { struct sk_buff *skb; struct sock *pending; struct VSockVmciSock *vsk; vsk = vsock_sk(sk); pending = NULL; /* Compiler warning. */ if (VSockVmciInBoundTable(sk)) { VSockVmciRemoveBound(sk); } if (VSockVmciInConnectedTable(sk)) { VSockVmciRemoveConnected(sk); } if (!VMCI_HANDLE_INVALID(vsk->dgHandle)) { vmci_datagram_destroy_handle(vsk->dgHandle); vsk->dgHandle = VMCI_INVALID_HANDLE; } lock_sock(sk); sock_orphan(sk); sk->sk_shutdown = SHUTDOWN_MASK; while ((skb = skb_dequeue(&sk->sk_receive_queue))) { kfree_skb(skb); } /* Clean up any sockets that never were accepted. */ while ((pending = VSockVmciDequeueAccept(sk)) != NULL) { __VSockVmciRelease(pending); sock_put(pending); } release_sock(sk); sock_put(sk); } } /* * Sock operations. */ /* *---------------------------------------------------------------------------- * * VSockVmciSkDestruct -- * * Destroys the provided socket. This is called by sk_free(), which is * invoked when the reference count of the socket drops to zero. * * Results: * None. * * Side effects: * Socket count is decremented. * *---------------------------------------------------------------------------- */ static void VSockVmciSkDestruct(struct sock *sk) // IN { VSockVmciSock *vsk; vsk = vsock_sk(sk); if (vsk->attachSubId != VMCI_INVALID_ID) { vmci_event_unsubscribe(vsk->attachSubId); vsk->attachSubId = VMCI_INVALID_ID; } if (vsk->detachSubId != VMCI_INVALID_ID) { vmci_event_unsubscribe(vsk->detachSubId); vsk->detachSubId = VMCI_INVALID_ID; } if (!VMCI_HANDLE_INVALID(vsk->qpHandle)) { ASSERT(vsk->qpair); vmci_qpair_detach(&vsk->qpair); vsk->qpHandle = VMCI_INVALID_HANDLE; ASSERT(vsk->qpair == NULL); vsk->produceSize = vsk->consumeSize = 0; } /* * Each list entry holds a reference on the socket, so we should not even be * here if the socket is in one of our lists. If we are we have a stray * sock_put() that needs to go away. */ ASSERT(!VSockVmciInBoundTable(sk)); ASSERT(!VSockVmciInConnectedTable(sk)); ASSERT(!VSockVmciIsPending(sk)); ASSERT(!VSockVmciInAcceptQueue(sk)); /* * When clearing these addresses, there's no need to set the family and * possibly register the address family with the kernel. */ VSockAddr_InitNoFamily(&vsk->localAddr, VMADDR_CID_ANY, VMADDR_PORT_ANY); VSockAddr_InitNoFamily(&vsk->remoteAddr, VMADDR_CID_ANY, VMADDR_PORT_ANY); NOTIFYCALL(vsk, socketDestruct, sk); compat_mutex_lock(®istrationMutex); vsockVmciSocketCount--; VSockVmciTestUnregister(); compat_mutex_unlock(®istrationMutex); VSOCK_STATS_CTLPKT_DUMP_ALL(); VSOCK_STATS_HIST_DUMP_ALL(); VSOCK_STATS_TOTALS_DUMP_ALL(); } /* *---------------------------------------------------------------------------- * * VSockVmciQueueRcvSkb -- * * Receives skb on the socket's receive queue. * * Results: * Zero on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciQueueRcvSkb(struct sock *sk, // IN struct sk_buff *skb) // IN { int err; err = sock_queue_rcv_skb(sk, skb); if (err) { kfree_skb(skb); } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciRegisterProto -- * * Registers the vSockets protocol family. * * Results: * Zero on success, error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciRegisterProto(void) { int err = 0; /* * From 2.6.9 until 2.6.11, these address families called sk_alloc_slab() * and the allocated slab was assigned to the slab variable in the proto * struct and was created of size slab_obj_size. As of 2.6.12 and later, * this slab allocation was moved * into proto_register() and only done if you specified a non-zero value * for the second argument (alloc_slab); the size of the slab element was * changed to obj_size. */ #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 12) err = sk_alloc_slab(&vsockVmciProto, "vsock"); if (err != 0) { sk_alloc_slab_error(&vsockVmciProto); } #else /* Specify 1 as the second argument so the slab is created for us. */ err = proto_register(&vsockVmciProto, 1); #endif return err; } /* *---------------------------------------------------------------------------- * * VSockVmciUnregisterProto -- * * Unregisters the vSockets protocol family. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void VSockVmciUnregisterProto(void) { #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 12) sk_free_slab(&vsockVmciProto); #else proto_unregister(&vsockVmciProto); #endif VSOCK_STATS_RESET(); } /* *---------------------------------------------------------------------------- * * VSockVmciRegisterAddressFamily -- * * Registers our socket address family with the kernel. * * Note that this assumes the registration lock is held. * * Results: * The address family value on success, negative error code on failure. * * Side effects: * Callers of socket operations with the returned value, on success, will * be able to use our socket implementation. * *---------------------------------------------------------------------------- */ static int VSockVmciRegisterAddressFamily(void) { int err = 0; int i; /* * Linux will not allocate an address family to code that is not part of the * kernel proper, so until that time comes we need a workaround. Here we * loop through the allowed values and claim the first one that's not * currently used. Users will then make an ioctl(2) into our module to * retrieve this value before calling socket(2). * * This is undesirable, but it's better than having users' programs break * when a hard-coded, currently-available value gets assigned to someone * else in the future. */ for (i = NPROTO - 1; i >= 0; i--) { vsockVmciFamilyOps.family = i; err = sock_register(&vsockVmciFamilyOps); if (err) { vsockVmciFamilyOps.family = VSOCK_INVALID_FAMILY; } else { vsockVmciDgramOps.family = i; vsockVmciStreamOps.family = i; err = i; break; } } if (VSOCK_INVALID_FAMILY == vsockVmciFamilyOps.family) { Warning("Could not register address family.\n"); } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciUnregisterAddressFamily -- * * Unregisters the address family with the kernel. * * Note that this assumes the registration lock is held. * * Results: * None. * * Side effects: * Our socket implementation is no longer accessible. * *---------------------------------------------------------------------------- */ static void VSockVmciUnregisterAddressFamily(void) { if (vsockVmciFamilyOps.family != VSOCK_INVALID_FAMILY) { sock_unregister(vsockVmciFamilyOps.family); } vsockVmciDgramOps.family = vsockVmciFamilyOps.family = VSOCK_INVALID_FAMILY; vsockVmciStreamOps.family = vsockVmciFamilyOps.family; } /* *---------------------------------------------------------------------------- * * VSockVmciRegisterWithVmci -- * * Registers with the VMCI device, and creates control message * and event handlers. * * Results: * Zero on success, error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciRegisterWithVmci(void) { int err = 0; uint32 apiVersion; /* * We don't call into the vmci module if the vmci device isn't * present. */ apiVersion = VMCI_KERNEL_API_VERSION_1; vmciDevicePresent = vmci_device_get(&apiVersion, NULL, NULL, NULL); if (!vmciDevicePresent) { Warning("VMCI device not present.\n"); return -1; } /* * Create the datagram handle that we will use to send and receive all * VSocket control messages for this context. */ err = VSockVmciDatagramCreateHnd(VSOCK_PACKET_RID, VMCI_FLAG_ANYCID_DG_HND, VSockVmciRecvStreamCB, NULL, &vmciStreamHandle); if (err < VMCI_SUCCESS) { Warning("Unable to create datagram handle. (%d)\n", err); err = VSockVmci_ErrorToVSockError(err); goto out; } err = vmci_event_subscribe(VMCI_EVENT_QP_RESUMED, VSockVmciQPResumedCB, NULL, &qpResumedSubId); if (err < VMCI_SUCCESS) { Warning("Unable to subscribe to QP resumed event. (%d)\n", err); err = VSockVmci_ErrorToVSockError(err); qpResumedSubId = VMCI_INVALID_ID; goto out; } out: if (err != 0) { VSockVmciUnregisterWithVmci(); } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciUnregisterWithVmci -- * * Destroys control message and event handlers, and unregisters * with the VMCI device * * Results: * None. * * Side effects: * Our socket implementation is no longer accessible. * *---------------------------------------------------------------------------- */ static void VSockVmciUnregisterWithVmci(void) { if (!vmciDevicePresent) { /* Nothing was registered. */ return; } if (!VMCI_HANDLE_INVALID(vmciStreamHandle)) { if (vmci_datagram_destroy_handle(vmciStreamHandle) != VMCI_SUCCESS) { Warning("Could not destroy VMCI datagram handle.\n"); } vmciStreamHandle = VMCI_INVALID_HANDLE; } if (qpResumedSubId != VMCI_INVALID_ID) { vmci_event_unsubscribe(qpResumedSubId); qpResumedSubId = VMCI_INVALID_ID; } vmci_device_release(NULL); vmciDevicePresent = FALSE; } /* *---------------------------------------------------------------------------- * * VSockVmciStreamHasData -- * * Gets the amount of data available for a given stream socket's consume * queue. * * Note that this assumes the socket lock is held. * * Results: * The amount of data available or a VMCI error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int64 VSockVmciStreamHasData(VSockVmciSock *vsk) // IN { ASSERT(vsk); return vmci_qpair_consume_buf_ready(vsk->qpair); } /* *---------------------------------------------------------------------------- * * VSockVmciStreamHasSpace -- * * Gets the amount of space available for a give stream socket's produce * queue. * * Note that this assumes the socket lock is held. * * Results: * The amount of space available or a VMCI error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int64 VSockVmciStreamHasSpace(VSockVmciSock *vsk) // IN { ASSERT(vsk); return vmci_qpair_produce_free_space(vsk->qpair); } /* * Socket operations. */ /* *---------------------------------------------------------------------------- * * VSockVmciRelease -- * * Releases the provided socket by freeing the contents of its queue. This * is called when a user process calls close(2) on the socket. * * Results: * Zero on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciRelease(struct socket *sock) // IN { __VSockVmciRelease(sock->sk); sock->sk = NULL; sock->state = SS_FREE; return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciBind -- * * Binds the provided address to the provided socket. * * Results: * Zero on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciBind(struct socket *sock, // IN struct sockaddr *addr, // IN int addrLen) // IN { int err; struct sock *sk; struct sockaddr_vm *vmciAddr; sk = sock->sk; if (VSockAddr_Cast(addr, addrLen, &vmciAddr) != 0) { return -EINVAL; } lock_sock(sk); err = __VSockVmciBind(sk, vmciAddr); release_sock(sk); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciDgramConnect -- * * Connects a datagram socket. This can be called multiple times to change * the socket's association and can be called with a sockaddr whose family * is set to AF_UNSPEC to dissolve any existing association. * * Results: * Zero on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciDgramConnect(struct socket *sock, // IN struct sockaddr *addr, // IN int addrLen, // IN int flags) // IN { int err; struct sock *sk; VSockVmciSock *vsk; struct sockaddr_vm *remoteAddr; sk = sock->sk; vsk = vsock_sk(sk); err = VSockAddr_Cast(addr, addrLen, &remoteAddr); if (err == -EAFNOSUPPORT && remoteAddr->svm_family == AF_UNSPEC) { lock_sock(sk); VSockAddr_Init(&vsk->remoteAddr, VMADDR_CID_ANY, VMADDR_PORT_ANY); sock->state = SS_UNCONNECTED; release_sock(sk); return 0; } else if (err != 0) { return -EINVAL; } lock_sock(sk); if (!VSockAddr_Bound(&vsk->localAddr)) { struct sockaddr_vm localAddr; VSockAddr_Init(&localAddr, VMADDR_CID_ANY, VMADDR_PORT_ANY); if ((err = __VSockVmciBind(sk, &localAddr))) { goto out; } } if (!VSockAddr_SocketContextDgram(remoteAddr->svm_cid, remoteAddr->svm_port)) { err = -EINVAL; goto out; } memcpy(&vsk->remoteAddr, remoteAddr, sizeof vsk->remoteAddr); sock->state = SS_CONNECTED; out: release_sock(sk); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciConnectTimeout -- * * Asynchronous connection attempts schedule this timeout function * to notify the connector of an unsuccessfull connection * attempt. If the socket is still in the connecting state and * hasn't been closed, we mark the socket as timed out. Otherwise, * we do nothing. * * Results: * None. * * Side effects: * May destroy the socket. * *---------------------------------------------------------------------------- */ static void VSockVmciConnectTimeout(compat_delayed_work_arg work) // IN { struct sock *sk; VSockVmciSock *vsk; vsk = COMPAT_DELAYED_WORK_GET_DATA(work, VSockVmciSock, dwork); ASSERT(vsk); sk = sk_vsock(vsk); lock_sock(sk); if (sk->sk_state == SS_CONNECTING && (sk->sk_shutdown != SHUTDOWN_MASK)) { sk->sk_state = SS_UNCONNECTED; sk->sk_err = ETIMEDOUT; sk->sk_error_report(sk); } release_sock(sk); sock_put(sk); } /* *---------------------------------------------------------------------------- * * VSockVmciStreamConnect -- * * Connects a stream socket. * * Results: * Zero on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciStreamConnect(struct socket *sock, // IN struct sockaddr *addr, // IN int addrLen, // IN int flags) // IN { int err; struct sock *sk; VSockVmciSock *vsk; struct sockaddr_vm *remoteAddr; long timeout; Bool oldPktProto = FALSE; DEFINE_WAIT(wait); err = 0; sk = sock->sk; vsk = vsock_sk(sk); lock_sock(sk); /* XXX AF_UNSPEC should make us disconnect like AF_INET. */ switch (sock->state) { case SS_CONNECTED: err = -EISCONN; goto out; case SS_DISCONNECTING: err = -EINVAL; goto out; case SS_CONNECTING: /* * This continues on so we can move sock into the SS_CONNECTED state once * the connection has completed (at which point err will be set to zero * also). Otherwise, we will either wait for the connection or return * -EALREADY should this be a non-blocking call. */ err = -EALREADY; break; default: ASSERT(sk->sk_state == SS_FREE || sk->sk_state == SS_UNCONNECTED || sk->sk_state == SS_LISTEN); if ((sk->sk_state == SS_LISTEN) || VSockAddr_Cast(addr, addrLen, &remoteAddr) != 0) { err = -EINVAL; goto out; } /* The hypervisor and well-known contexts do not have socket endpoints. */ if (!VSockAddr_SocketContextStream(remoteAddr->svm_cid)) { err = -ENETUNREACH; goto out; } /* Set the remote address that we are connecting to. */ memcpy(&vsk->remoteAddr, remoteAddr, sizeof vsk->remoteAddr); /* Autobind this socket to the local address if necessary. */ if (!VSockAddr_Bound(&vsk->localAddr)) { struct sockaddr_vm localAddr; VSockAddr_Init(&localAddr, VMADDR_CID_ANY, VMADDR_PORT_ANY); if ((err = __VSockVmciBind(sk, &localAddr))) { goto out; } } sk->sk_state = SS_CONNECTING; if (VSockVmciOldProtoOverride(&oldPktProto) && oldPktProto) { err = VSOCK_SEND_CONN_REQUEST(sk, vsk->queuePairSize); if (err < 0) { sk->sk_state = SS_UNCONNECTED; goto out; } } else { int supportedProtoVersions = VSockVmciNewProtoSupportedVersions(); err = VSOCK_SEND_CONN_REQUEST2(sk, vsk->queuePairSize, supportedProtoVersions); if (err < 0) { sk->sk_state = SS_UNCONNECTED; goto out; } vsk->sentRequest = TRUE; } /* * Mark sock as connecting and set the error code to in progress in case * this is a non-blocking connect. */ sock->state = SS_CONNECTING; err = -EINPROGRESS; } /* * The receive path will handle all communication until we are able to enter * the connected state. Here we wait for the connection to be completed or * a notification of an error. */ timeout = vsk->connectTimeout; prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); while (sk->sk_state != SS_CONNECTED && sk->sk_err == 0) { if (flags & O_NONBLOCK) { /* * If we're not going to block, we schedule a timeout * function to generate a timeout on the connection attempt, * in case the peer doesn't respond in a timely manner. We * hold on to the socket until the timeout fires. */ sock_hold(sk); COMPAT_INIT_DELAYED_WORK(&vsk->dwork, VSockVmciConnectTimeout, vsk); compat_schedule_delayed_work(&vsk->dwork, timeout); /* Skip ahead to preserve error code set above. */ goto outWait; } release_sock(sk); timeout = schedule_timeout(timeout); lock_sock(sk); if (signal_pending(current)) { err = sock_intr_errno(timeout); goto outWaitError; } else if (timeout == 0) { err = -ETIMEDOUT; goto outWaitError; } prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); } if (sk->sk_err) { err = -sk->sk_err; goto outWaitError; } else { ASSERT(sk->sk_state == SS_CONNECTED); err = 0; } outWait: finish_wait(sk_sleep(sk), &wait); out: release_sock(sk); return err; outWaitError: sk->sk_state = SS_UNCONNECTED; sock->state = SS_UNCONNECTED; goto outWait; } /* *---------------------------------------------------------------------------- * * VSockVmciAccept -- * * Accepts next available connection request for this socket. * * Results: * Zero on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciAccept(struct socket *sock, // IN struct socket *newsock, // IN/OUT int flags) // IN { struct sock *listener; int err; struct sock *connected; VSockVmciSock *vconnected; long timeout; DEFINE_WAIT(wait); err = 0; listener = sock->sk; lock_sock(listener); if (sock->type != SOCK_STREAM) { err = -EOPNOTSUPP; goto out; } if (listener->sk_state != SS_LISTEN) { err = -EINVAL; goto out; } /* * Wait for children sockets to appear; these are the new sockets created * upon connection establishment. */ timeout = sock_sndtimeo(listener, flags & O_NONBLOCK); prepare_to_wait(sk_sleep(listener), &wait, TASK_INTERRUPTIBLE); while ((connected = VSockVmciDequeueAccept(listener)) == NULL && listener->sk_err == 0) { release_sock(listener); timeout = schedule_timeout(timeout); lock_sock(listener); if (signal_pending(current)) { err = sock_intr_errno(timeout); goto outWait; } else if (timeout == 0) { err = -EAGAIN; goto outWait; } prepare_to_wait(sk_sleep(listener), &wait, TASK_INTERRUPTIBLE); } if (listener->sk_err) { err = -listener->sk_err; } if (connected) { listener->sk_ack_backlog--; lock_sock(connected); vconnected = vsock_sk(connected); /* * If the listener socket has received an error, then we should reject * this socket and return. Note that we simply mark the socket rejected, * drop our reference, and let the cleanup function handle the cleanup; * the fact that we found it in the listener's accept queue guarantees * that the cleanup function hasn't run yet. */ if (err) { vconnected->rejected = TRUE; release_sock(connected); sock_put(connected); goto outWait; } newsock->state = SS_CONNECTED; sock_graft(connected, newsock); release_sock(connected); sock_put(connected); } outWait: finish_wait(sk_sleep(listener), &wait); out: release_sock(listener); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciGetname -- * * Provides the local or remote address for the socket. * * Results: * Zero on success, negative error code otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciGetname(struct socket *sock, // IN struct sockaddr *addr, // OUT int *addrLen, // OUT int peer) // IN { int err; struct sock *sk; VSockVmciSock *vsk; struct sockaddr_vm *vmciAddr; sk = sock->sk; vsk = vsock_sk(sk); err = 0; lock_sock(sk); if (peer) { if (sock->state != SS_CONNECTED) { err = -ENOTCONN; goto out; } vmciAddr = &vsk->remoteAddr; } else { vmciAddr = &vsk->localAddr; } if (!vmciAddr) { err = -EINVAL; goto out; } /* * sys_getsockname() and sys_getpeername() pass us a MAX_SOCK_ADDR-sized * buffer and don't set addrLen. Unfortunately that macro is defined in * socket.c instead of .h, so we hardcode its value here. */ ASSERT_ON_COMPILE(sizeof *vmciAddr <= 128); memcpy(addr, vmciAddr, sizeof *vmciAddr); *addrLen = sizeof *vmciAddr; out: release_sock(sk); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciPoll -- * * Waits on file for activity then provides mask indicating state of socket. * * Results: * Mask of flags containing socket state. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static unsigned int VSockVmciPoll(struct file *file, // IN struct socket *sock, // IN poll_table *wait) // IN { struct sock *sk; unsigned int mask; VSockVmciSock *vsk; sk = sock->sk; vsk = vsock_sk(sk); poll_wait(file, sk_sleep(sk), wait); mask = 0; if (sk->sk_err) { /* Signify that there has been an error on this socket. */ mask |= POLLERR; } /* * INET sockets treat local write shutdown and peer write shutdown * as a case of POLLHUP set. */ if ((sk->sk_shutdown == SHUTDOWN_MASK) || ((sk->sk_shutdown & SEND_SHUTDOWN) && (vsk->peerShutdown & SEND_SHUTDOWN))) { mask |= POLLHUP; } /* POLLRDHUP wasn't added until 2.6.17. */ #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 17) if (sk->sk_shutdown & RCV_SHUTDOWN || vsk->peerShutdown & SEND_SHUTDOWN) { mask |= POLLRDHUP; } #endif if (sock->type == SOCK_DGRAM) { /* * For datagram sockets we can read if there is something in the queue * and write as long as the socket isn't shutdown for sending. */ if (!skb_queue_empty(&sk->sk_receive_queue) || (sk->sk_shutdown & RCV_SHUTDOWN)) { mask |= POLLIN | POLLRDNORM; } if (!(sk->sk_shutdown & SEND_SHUTDOWN)) { mask |= POLLOUT | POLLWRNORM | POLLWRBAND; } } else if (sock->type == SOCK_STREAM) { lock_sock(sk); /* * Listening sockets that have connections in their accept queue can be read. */ if (sk->sk_state == SS_LISTEN && !VSockVmciIsAcceptQueueEmpty(sk)) { mask |= POLLIN | POLLRDNORM; } /* * If there is something in the queue then we can read. */ if (!VMCI_HANDLE_INVALID(vsk->qpHandle) && !(sk->sk_shutdown & RCV_SHUTDOWN)) { Bool dataReadyNow = FALSE; int32 ret = 0; NOTIFYCALLRET(vsk, ret, pollIn, sk, 1, &dataReadyNow); if (ret < 0) { mask |= POLLERR; } else { if (dataReadyNow) { mask |= POLLIN | POLLRDNORM; } } } /* * Sockets whose connections have been closed, reset, or terminated * should also be considered read, and we check the shutdown flag for * that. */ if (sk->sk_shutdown & RCV_SHUTDOWN || vsk->peerShutdown & SEND_SHUTDOWN) { mask |= POLLIN | POLLRDNORM; } /* * Connected sockets that can produce data can be written. */ if (sk->sk_state == SS_CONNECTED) { if (!(sk->sk_shutdown & SEND_SHUTDOWN)) { Bool spaceAvailNow = FALSE; int32 ret = 0; NOTIFYCALLRET(vsk, ret, pollOut, sk, 1, &spaceAvailNow); if (ret < 0) { mask |= POLLERR; } else { if (spaceAvailNow) { /* Remove POLLWRBAND since INET sockets are not setting it.*/ mask |= POLLOUT | POLLWRNORM; } } } } /* * Simulate INET socket poll behaviors, which sets POLLOUT|POLLWRNORM when * peer is closed and nothing to read, but local send is not shutdown. */ if (sk->sk_state == SS_UNCONNECTED) { if (!(sk->sk_shutdown & SEND_SHUTDOWN)) { mask |= POLLOUT | POLLWRNORM; } } release_sock(sk); } return mask; } /* *---------------------------------------------------------------------------- * * VSockVmciListen -- * * Signify that this socket is listening for connection requests. * * Results: * Zero on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciListen(struct socket *sock, // IN int backlog) // IN { int err; struct sock *sk; VSockVmciSock *vsk; sk = sock->sk; lock_sock(sk); if (sock->type != SOCK_STREAM) { err = -EOPNOTSUPP; goto out; } if (sock->state != SS_UNCONNECTED) { err = -EINVAL; goto out; } vsk = vsock_sk(sk); if (!VSockAddr_Bound(&vsk->localAddr)) { err = -EINVAL; goto out; } sk->sk_max_ack_backlog = backlog; sk->sk_state = SS_LISTEN; err = 0; out: release_sock(sk); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciShutdown -- * * Shuts down the provided socket in the provided method. * * Results: * Zero on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciShutdown(struct socket *sock, // IN int mode) // IN { int32 err; struct sock *sk; /* * User level uses SHUT_RD (0) and SHUT_WR (1), but the kernel uses * RCV_SHUTDOWN (1) and SEND_SHUTDOWN (2), so we must increment mode here * like the other address families do. Note also that the increment makes * SHUT_RDWR (2) into RCV_SHUTDOWN | SEND_SHUTDOWN (3), which is what we * want. */ mode++; if ((mode & ~SHUTDOWN_MASK) || !mode) { return -EINVAL; } /* * If this is a STREAM socket and it is not connected then bail out * immediately. If it is a DGRAM socket then we must first kick the socket * so that it wakes up from any sleeping calls, for example recv(), and then * afterwards return the error. */ sk = sock->sk; if (sock->state == SS_UNCONNECTED) { err = -ENOTCONN; if (sk->sk_type == SOCK_STREAM) { return err; } } else { sock->state = SS_DISCONNECTING; err = 0; } /* Receive and send shutdowns are treated alike. */ mode = mode & (RCV_SHUTDOWN | SEND_SHUTDOWN); if (mode) { lock_sock(sk); sk->sk_shutdown |= mode; sk->sk_state_change(sk); release_sock(sk); if (sk->sk_type == SOCK_STREAM) { sock_reset_flag(sk, SOCK_DONE); VSOCK_SEND_SHUTDOWN(sk, mode); } } return err; } /* *---------------------------------------------------------------------------- * * VSockVmciDgramSendmsg -- * * Sends a datagram. * * Results: * Number of bytes sent on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciDgramSendmsg(struct kiocb *kiocb, // UNUSED struct socket *sock, // IN: socket to send on struct msghdr *msg, // IN: message to send size_t len) // IN: length of message { int err; struct sock *sk; VSockVmciSock *vsk; struct sockaddr_vm *remoteAddr; VMCIDatagram *dg; if (msg->msg_flags & MSG_OOB) { return -EOPNOTSUPP; } if (len > VMCI_MAX_DG_PAYLOAD_SIZE) { return -EMSGSIZE; } /* For now, MSG_DONTWAIT is always assumed... */ err = 0; sk = sock->sk; vsk = vsock_sk(sk); lock_sock(sk); if (!VSockAddr_Bound(&vsk->localAddr)) { struct sockaddr_vm localAddr; VSockAddr_Init(&localAddr, VMADDR_CID_ANY, VMADDR_PORT_ANY); if ((err = __VSockVmciBind(sk, &localAddr))) { goto out; } } /* * If the provided message contains an address, use that. Otherwise fall * back on the socket's remote handle (if it has been connected). */ if (msg->msg_name && VSockAddr_Cast(msg->msg_name, msg->msg_namelen, &remoteAddr) == 0) { /* Ensure this address is of the right type and is a valid destination. */ // XXXAB Temporary to handle test program if (remoteAddr->svm_cid == VMADDR_CID_ANY) { remoteAddr->svm_cid = vmci_get_context_id(); } if (!VSockAddr_Bound(remoteAddr)) { err = -EINVAL; goto out; } } else if (sock->state == SS_CONNECTED) { remoteAddr = &vsk->remoteAddr; // XXXAB Temporary to handle test program if (remoteAddr->svm_cid == VMADDR_CID_ANY) { remoteAddr->svm_cid = vmci_get_context_id(); } /* XXX Should connect() or this function ensure remoteAddr is bound? */ if (!VSockAddr_Bound(&vsk->remoteAddr)) { err = -EINVAL; goto out; } } else { err = -EINVAL; goto out; } /* * Make sure that we don't allow a userlevel app to send datagrams * to the hypervisor that modify VMCI device state. */ if (!VSockAddr_SocketContextDgram(remoteAddr->svm_cid, remoteAddr->svm_port)) { err = -EINVAL; goto out; } if (!VSockVmciAllowDgram(vsk, remoteAddr->svm_cid)) { err = -EPERM; goto out; } /* * Allocate a buffer for the user's message and our packet header. */ dg = kmalloc(len + sizeof *dg, GFP_KERNEL); if (!dg) { err = -ENOMEM; goto out; } memcpy_fromiovec(VMCI_DG_PAYLOAD(dg), msg->msg_iov, len); dg->dst = VMCI_MAKE_HANDLE(remoteAddr->svm_cid, remoteAddr->svm_port); dg->src = VMCI_MAKE_HANDLE(vsk->localAddr.svm_cid, vsk->localAddr.svm_port); dg->payloadSize = len; err = vmci_datagram_send(dg); kfree(dg); if (err < 0) { err = VSockVmci_ErrorToVSockError(err); goto out; } err -= sizeof *dg; out: release_sock(sk); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciStreamSetsockopt -- * * Set a socket option on a stream socket * * Results: * 0 on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int VSockVmciStreamSetsockopt(struct socket *sock, // IN/OUT int level, // IN int optname, // IN char __user *optval, // IN VSockSetsockoptLenType optlen) // IN { int err; struct sock *sk; VSockVmciSock *vsk; uint64 val; if (level != VSockVmci_GetAFValue()) { return -ENOPROTOOPT; } # define COPY_IN(_v) \ do { \ if (optlen < sizeof _v) { \ err = -EINVAL; \ goto exit; \ } \ if (copy_from_user(&_v, optval, sizeof _v) != 0) { \ err = -EFAULT; \ goto exit; \ } \ } while (0) err = 0; sk = sock->sk; vsk = vsock_sk(sk); ASSERT(vsk->queuePairMinSize <= vsk->queuePairSize && vsk->queuePairSize <= vsk->queuePairMaxSize); lock_sock(sk); switch (optname) { case SO_VMCI_BUFFER_SIZE: COPY_IN(val); if (val < vsk->queuePairMinSize) { vsk->queuePairMinSize = val; } if (val > vsk->queuePairMaxSize) { vsk->queuePairMaxSize = val; } vsk->queuePairSize = val; break; case SO_VMCI_BUFFER_MAX_SIZE: COPY_IN(val); if (val < vsk->queuePairSize) { vsk->queuePairSize = val; } vsk->queuePairMaxSize = val; break; case SO_VMCI_BUFFER_MIN_SIZE: COPY_IN(val); if (val > vsk->queuePairSize) { vsk->queuePairSize = val; } vsk->queuePairMinSize = val; break; case SO_VMCI_CONNECT_TIMEOUT: { struct timeval tv; COPY_IN(tv); if (tv.tv_sec >= 0 && tv.tv_usec < USEC_PER_SEC && tv.tv_sec < (MAX_SCHEDULE_TIMEOUT/HZ - 1)) { vsk->connectTimeout = tv.tv_sec * HZ + CEILING(tv.tv_usec, (1000000 / HZ)); if (vsk->connectTimeout == 0) { vsk->connectTimeout = VSOCK_DEFAULT_CONNECT_TIMEOUT; } } else { err = -ERANGE; } break; } default: err = -ENOPROTOOPT; break; } # undef COPY_IN ASSERT(vsk->queuePairMinSize <= vsk->queuePairSize && vsk->queuePairSize <= vsk->queuePairMaxSize); exit: release_sock(sk); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciStreamGetsockopt -- * * Get a socket option for a stream socket * * Results: * 0 on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int VSockVmciStreamGetsockopt(struct socket *sock, // IN int level, // IN int optname, // IN char __user *optval, // OUT int __user * optlen) // IN/OUT { int err; int len; struct sock *sk; VSockVmciSock *vsk; if (level != VSockVmci_GetAFValue()) { return -ENOPROTOOPT; } if ((err = get_user(len, optlen)) != 0) { return err; } # define COPY_OUT(_v) \ do { \ if (len < sizeof _v) { \ return -EINVAL; \ } \ len = sizeof _v; \ if (copy_to_user(optval, &_v, len) != 0) { \ return -EFAULT; \ } \ } while (0) err = 0; sk = sock->sk; vsk = vsock_sk(sk); switch (optname) { case SO_VMCI_BUFFER_SIZE: COPY_OUT(vsk->queuePairSize); break; case SO_VMCI_BUFFER_MAX_SIZE: COPY_OUT(vsk->queuePairMaxSize); break; case SO_VMCI_BUFFER_MIN_SIZE: COPY_OUT(vsk->queuePairMinSize); break; case SO_VMCI_CONNECT_TIMEOUT: { struct timeval tv; tv.tv_sec = vsk->connectTimeout / HZ; tv.tv_usec = (vsk->connectTimeout - tv.tv_sec * HZ) * (1000000 / HZ); COPY_OUT(tv); break; } default: return -ENOPROTOOPT; } if ((err = put_user(len, optlen)) != 0) { return -EFAULT; } # undef COPY_OUT return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciStreamSendmsg -- * * Sends a message on the socket. * * Results: * Number of bytes sent on success, negative error code on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciStreamSendmsg(struct kiocb *kiocb, // UNUSED struct socket *sock, // IN: socket to send on struct msghdr *msg, // IN: message to send size_t len) // IN: length of message { struct sock *sk; VSockVmciSock *vsk; ssize_t totalWritten; long timeout; int err; VSockVmciSendNotifyData sendData; DEFINE_WAIT(wait); sk = sock->sk; vsk = vsock_sk(sk); totalWritten = 0; err = 0; if (msg->msg_flags & MSG_OOB) { return -EOPNOTSUPP; } lock_sock(sk); /* Callers should not provide a destination with stream sockets. */ if (msg->msg_namelen) { err = sk->sk_state == SS_CONNECTED ? -EISCONN : -EOPNOTSUPP; goto out; } /* Send data only if both sides are not shutdown in the direction. */ if (sk->sk_shutdown & SEND_SHUTDOWN || vsk->peerShutdown & RCV_SHUTDOWN) { err = -EPIPE; goto out; } if (sk->sk_state != SS_CONNECTED || !VSockAddr_Bound(&vsk->localAddr)) { err = -ENOTCONN; goto out; } if (!VSockAddr_Bound(&vsk->remoteAddr)) { err = -EDESTADDRREQ; goto out; } /* * Wait for room in the produce queue to enqueue our user's data. */ timeout = sock_sndtimeo(sk, msg->msg_flags & MSG_DONTWAIT); NOTIFYCALLRET(vsk, err, sendInit, sk, &sendData); if (err < 0) { goto out; } prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); while (totalWritten < len) { ssize_t written; while (VSockVmciStreamHasSpace(vsk) == 0 && sk->sk_err == 0 && !(sk->sk_shutdown & SEND_SHUTDOWN) && !(vsk->peerShutdown & RCV_SHUTDOWN)) { /* Don't wait for non-blocking sockets. */ if (timeout == 0) { err = -EAGAIN; goto outWait; } NOTIFYCALLRET(vsk, err, sendPreBlock, sk, &sendData); if (err < 0) { goto outWait; } release_sock(sk); timeout = schedule_timeout(timeout); lock_sock(sk); if (signal_pending(current)) { err = sock_intr_errno(timeout); goto outWait; } else if (timeout == 0) { err = -EAGAIN; goto outWait; } prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); } /* * These checks occur both as part of and after the loop conditional * since we need to check before and after sleeping. */ if (sk->sk_err) { err = -sk->sk_err; goto outWait; } else if ((sk->sk_shutdown & SEND_SHUTDOWN) || (vsk->peerShutdown & RCV_SHUTDOWN)) { err = -EPIPE; goto outWait; } VSOCK_STATS_STREAM_PRODUCE_HIST(vsk); NOTIFYCALLRET(vsk, err, sendPreEnqueue, sk, &sendData); if (err < 0) { goto outWait; } /* * Note that enqueue will only write as many bytes as are free in the * produce queue, so we don't need to ensure len is smaller than the queue * size. It is the caller's responsibility to check how many bytes we were * able to send. */ written = vmci_qpair_enquev(vsk->qpair, msg->msg_iov, len - totalWritten, 0); if (written < 0) { err = -ENOMEM; goto outWait; } totalWritten += written; NOTIFYCALLRET(vsk, err, sendPostEnqueue, sk, written, &sendData); if (err < 0) { goto outWait; } } ASSERT(totalWritten <= INT_MAX); outWait: if (totalWritten > 0) { VSOCK_STATS_STREAM_PRODUCE(totalWritten); err = totalWritten; } finish_wait(sk_sleep(sk), &wait); out: release_sock(sk); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciDgramRecvmsg -- * * Receives a datagram and places it in the caller's msg. * * Results: * The size of the payload on success, negative value on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciDgramRecvmsg(struct kiocb *kiocb, // UNUSED struct socket *sock, // IN: socket to receive from struct msghdr *msg, // IN/OUT: message to receive into size_t len, // IN: length of receive buffer int flags) // IN: receive flags { int err; int noblock; struct sock *sk; VMCIDatagram *dg; size_t payloadLen; struct sk_buff *skb; sk = sock->sk; noblock = flags & MSG_DONTWAIT; if (flags & MSG_OOB || flags & MSG_ERRQUEUE) { return -EOPNOTSUPP; } /* Retrieve the head sk_buff from the socket's receive queue. */ err = 0; skb = skb_recv_datagram(sk, flags, noblock, &err); if (err) { return err; } if (!skb) { return -EAGAIN; } dg = (VMCIDatagram *)skb->data; if (!dg) { /* err is 0, meaning we read zero bytes. */ goto out; } payloadLen = dg->payloadSize; /* Ensure the sk_buff matches the payload size claimed in the packet. */ if (payloadLen != skb->len - sizeof *dg) { err = -EINVAL; goto out; } if (payloadLen > len) { payloadLen = len; msg->msg_flags |= MSG_TRUNC; } /* Place the datagram payload in the user's iovec. */ err = skb_copy_datagram_iovec(skb, sizeof *dg, msg->msg_iov, payloadLen); if (err) { goto out; } msg->msg_namelen = 0; if (msg->msg_name) { struct sockaddr_vm *vmciAddr; /* Provide the address of the sender. */ vmciAddr = (struct sockaddr_vm *)msg->msg_name; VSockAddr_Init(vmciAddr, VMCI_HANDLE_TO_CONTEXT_ID(dg->src), VMCI_HANDLE_TO_RESOURCE_ID(dg->src)); msg->msg_namelen = sizeof *vmciAddr; } err = payloadLen; out: skb_free_datagram(sk, skb); return err; } /* *---------------------------------------------------------------------------- * * VSockVmciStreamRecvmsg -- * * Receives a datagram and places it in the caller's msg. * * Results: * The size of the payload on success, negative value on failure. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciStreamRecvmsg(struct kiocb *kiocb, // UNUSED struct socket *sock, // IN: socket to receive from struct msghdr *msg, // IN/OUT: message to receive into size_t len, // IN: length of receive buffer int flags) // IN: receive flags { struct sock *sk; VSockVmciSock *vsk; int err; size_t target; ssize_t copied; long timeout; VSockVmciRecvNotifyData recvData; DEFINE_WAIT(wait); sk = sock->sk; vsk = vsock_sk(sk); err = 0; lock_sock(sk); if (sk->sk_state != SS_CONNECTED) { /* * Recvmsg is supposed to return 0 if a peer performs an orderly shutdown. * Differentiate between that case and when a peer has not connected or a * local shutdown occured with the SOCK_DONE flag. */ if (sock_flag(sk, SOCK_DONE)) { err = 0; } else { err = -ENOTCONN; } goto out; } if (flags & MSG_OOB) { err = -EOPNOTSUPP; goto out; } /* * We don't check peerShutdown flag here since peer may actually shut down, * but there can be data in the VMCI queue that local socket can receive. */ if (sk->sk_shutdown & RCV_SHUTDOWN) { err = 0; goto out; } /* * It is valid on Linux to pass in a zero-length receive buffer. This * is not an error. We may as well bail out now. Note that if we don't, * we will fail "ASSERT(copied >= target)" after we dequeue, because the * minimum target is always 1 byte. */ if (!len) { err = 0; goto out; } /* * We must not copy less than target bytes into the user's buffer before * returning successfully, so we wait for the consume queue to have that * much data to consume before dequeueing. Note that this makes it * impossible to handle cases where target is greater than the queue size. */ target = sock_rcvlowat(sk, flags & MSG_WAITALL, len); if (target >= vsk->consumeSize) { err = -ENOMEM; goto out; } timeout = sock_rcvtimeo(sk, flags & MSG_DONTWAIT); copied = 0; NOTIFYCALLRET(vsk, err, recvInit, sk, target, &recvData); if (err < 0) { goto out; } prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); while (1) { int64 ready = VSockVmciStreamHasData(vsk); if (ready < 0) { /* * Invalid queue pair content. XXX This should be changed to * a connection reset in a later change. */ err = -ENOMEM; goto outWait; } else if (ready > 0) { ssize_t read; VSOCK_STATS_STREAM_CONSUME_HIST(vsk); NOTIFYCALLRET(vsk, err, recvPreDequeue, sk, target, &recvData); if (err < 0) { break; } if (flags & MSG_PEEK) { read = vmci_qpair_peekv(vsk->qpair, msg->msg_iov, len - copied, 0); } else { read = vmci_qpair_dequev(vsk->qpair, msg->msg_iov, len - copied, 0); } if (read < 0) { err = -ENOMEM; break; } ASSERT(read <= INT_MAX); copied += read; NOTIFYCALLRET(vsk, err, recvPostDequeue, sk, target, read, !(flags & MSG_PEEK), &recvData); if (err < 0) { goto outWait; } if (read >= target || flags & MSG_PEEK) { break; } target -= read; } else { if (sk->sk_err != 0 || (sk->sk_shutdown & RCV_SHUTDOWN) || (vsk->peerShutdown & SEND_SHUTDOWN)) { break; } /* Don't wait for non-blocking sockets. */ if (timeout == 0) { err = -EAGAIN; break; } NOTIFYCALLRET(vsk, err, recvPreBlock, sk, target, &recvData); if (err < 0) { break; } release_sock(sk); timeout = schedule_timeout(timeout); lock_sock(sk); if (signal_pending(current)) { err = sock_intr_errno(timeout); break; } else if (timeout == 0) { err = -EAGAIN; break; } prepare_to_wait(sk_sleep(sk), &wait, TASK_INTERRUPTIBLE); } } if (sk->sk_err) { err = -sk->sk_err; } else if (sk->sk_shutdown & RCV_SHUTDOWN) { err = 0; } if (copied > 0) { /* * We only do these additional bookkeeping/notification steps if we * actually copied something out of the queue pair instead of just peeking * ahead. */ if (!(flags & MSG_PEEK)) { VSOCK_STATS_STREAM_CONSUME(copied); /* * If the other side has shutdown for sending and there is nothing more * to read, then modify the socket state. */ if (vsk->peerShutdown & SEND_SHUTDOWN) { if (VSockVmciStreamHasData(vsk) <= 0) { sk->sk_state = SS_UNCONNECTED; sock_set_flag(sk, SOCK_DONE); sk->sk_state_change(sk); } } } err = copied; } outWait: finish_wait(sk_sleep(sk), &wait); out: release_sock(sk); return err; } /* * Protocol operation. */ /* *---------------------------------------------------------------------------- * * VSockVmciCreate -- * * Creates a VSocket socket. * * Results: * Zero on success, negative error code on failure. * * Side effects: * Socket count is incremented. * *---------------------------------------------------------------------------- */ static int VSockVmciCreate( #if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 6, 24) struct net *net, // IN #endif struct socket *sock, // IN int protocol // IN #ifdef VMW_NETCREATE_KERNARG , int kern // IN #endif ) { if (!sock) { return -EINVAL; } if (protocol) { return -EPROTONOSUPPORT; } switch (sock->type) { case SOCK_DGRAM: sock->ops = &vsockVmciDgramOps; break; case SOCK_STREAM: sock->ops = &vsockVmciStreamOps; break; default: return -ESOCKTNOSUPPORT; } sock->state = SS_UNCONNECTED; #if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 24) return __VSockVmciCreate(sock, NULL, GFP_KERNEL, 0) ? 0 : -ENOMEM; #else return __VSockVmciCreate(net, sock, NULL, GFP_KERNEL, 0) ? 0 : -ENOMEM; #endif } /* *---------------------------------------------------------------------------- * * VSockVmciIoctl32Handler -- * * Handler for 32-bit ioctl(2) on 64-bit. * * Results: * Same as VsockVmciDevIoctl(). * * Side effects: * None. * *---------------------------------------------------------------------------- */ #ifdef VM_X86_64 #ifndef HAVE_COMPAT_IOCTL static int VSockVmciIoctl32Handler(unsigned int fd, // IN unsigned int iocmd, // IN unsigned long ioarg, // IN/OUT struct file * filp) // IN { int ret; ret = -ENOTTY; if (filp && filp->f_op && filp->f_op->ioctl == VSockVmciDevIoctl) { ret = VSockVmciDevIoctl(filp->f_dentry->d_inode, filp, iocmd, ioarg); } return ret; } #endif /* !HAVE_COMPAT_IOCTL */ /* *---------------------------------------------------------------------------- * * register_ioctl32_handlers -- * * Registers the ioctl conversion handler. * * Results: * Zero on success, error code otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int register_ioctl32_handlers(void) { #ifndef HAVE_COMPAT_IOCTL { int i; for (i = IOCTL_VMCI_SOCKETS_FIRST; i < IOCTL_VMCI_SOCKETS_LAST; i++) { int retval = register_ioctl32_conversion(i, VSockVmciIoctl32Handler); if (retval) { Warning("Fail to register ioctl32 conversion for cmd %d\n", i); return retval; } } } #endif /* !HAVE_COMPAT_IOCTL */ return 0; } /* *---------------------------------------------------------------------------- * * unregister_ioctl32_handlers -- * * Unregisters the ioctl converstion handler. * * Results: * None. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static void unregister_ioctl32_handlers(void) { #ifndef HAVE_COMPAT_IOCTL { int i; for (i = IOCTL_VMCI_SOCKETS_FIRST; i < IOCTL_VMCI_SOCKETS_LAST; i++) { int retval = unregister_ioctl32_conversion(i); if (retval) { Warning("Fail to unregister ioctl32 conversion for cmd %d\n", i); } } } #endif /* !HAVE_COMPAT_IOCTL */ } #else /* VM_X86_64 */ #define register_ioctl32_handlers() (0) #define unregister_ioctl32_handlers() do { } while (0) #endif /* VM_X86_64 */ /* * Device operations. */ /* *---------------------------------------------------------------------------- * * VSockVmciDevOpen -- * * Invoked when the device is opened. Simply maintains a count of open * instances. * * Results: * Zero on success, negative value otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int VSockVmciDevOpen(struct inode *inode, // IN struct file *file) // IN { compat_mutex_lock(®istrationMutex); devOpenCount++; compat_mutex_unlock(®istrationMutex); return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciDevRelease -- * * Invoked when the device is closed. Updates the open instance count and * unregisters the socket family if this is the last user. * * Results: * Zero on success, negative value otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ int VSockVmciDevRelease(struct inode *inode, // IN struct file *file) // IN { compat_mutex_lock(®istrationMutex); devOpenCount--; VSockVmciTestUnregister(); compat_mutex_unlock(®istrationMutex); return 0; } /* *---------------------------------------------------------------------------- * * VSockVmciDevIoctl -- * * ioctl(2) handler. * * Results: * Zero on success, negative error code otherwise. * * Side effects: * None. * *---------------------------------------------------------------------------- */ static int VSockVmciDevIoctl(struct inode *inode, // IN struct file *filp, // IN u_int iocmd, // IN unsigned long ioarg) // IN/OUT { int retval; retval = 0; switch (iocmd) { case IOCTL_VMCI_SOCKETS_VERSION: { uint16 parts[4] = { VSOCK_DRIVER_VERSION_COMMAS }; uint32 version = VMCI_SOCKETS_MAKE_VERSION(parts); if (copy_to_user((void*)ioarg, &version, sizeof version) != 0) { retval = -EFAULT; } break; } case IOCTL_VMCI_SOCKETS_GET_AF_VALUE: { int family; family = VSockVmci_GetAFValue(); if (family < 0) { Warning("AF_VSOCK is not registered\n"); } if (copy_to_user((void *)ioarg, &family, sizeof family) != 0) { retval = -EFAULT; } break; } case IOCTL_VMCI_SOCKETS_GET_LOCAL_CID: { VMCIId cid = vmci_get_context_id(); if (copy_to_user((void *)ioarg, &cid, sizeof cid) != 0) { retval = -EFAULT; } break; } default: Warning("Unknown ioctl %d\n", iocmd); retval = -EINVAL; } return retval; } #if defined(HAVE_COMPAT_IOCTL) || defined(HAVE_UNLOCKED_IOCTL) /* *----------------------------------------------------------------------------- * * VSockVmciDevUnlockedIoctl -- * * Wrapper for VSockVmciDevIoctl() supporting the compat_ioctl and * unlocked_ioctl methods that have signatures different from the * old ioctl. Used as compat_ioctl method for 32bit apps running * on 64bit kernel and for unlocked_ioctl on systems supporting * those. VSockVmciDevIoctl() may safely be called without holding * the BKL. * * Results: * Same as VSockVmciDevIoctl(). * * Side effects: * None. * *----------------------------------------------------------------------------- */ static long VSockVmciDevUnlockedIoctl(struct file *filp, // IN u_int iocmd, // IN unsigned long ioarg) // IN/OUT { return VSockVmciDevIoctl(NULL, filp, iocmd, ioarg); } #endif /* * Module operations. */ /* *---------------------------------------------------------------------------- * * VSockVmciInit -- * * Initialization routine for the VSockets module. * * Results: * Zero on success, error code on failure. * * Side effects: * The VSocket protocol family and socket operations are registered. * *---------------------------------------------------------------------------- */ static int __init VSockVmciInit(void) { int err; DriverLog_Init("VSock"); request_module("vmci"); err = misc_register(&vsockVmciDevice); if (err) { return -ENOENT; } err = register_ioctl32_handlers(); if (err) { misc_deregister(&vsockVmciDevice); return err; } err = VSockVmciRegisterWithVmci(); if (err) { Warning("Cannot register with VMCI device.\n"); unregister_ioctl32_handlers(); misc_deregister(&vsockVmciDevice); return err; } err = VSockVmciRegisterProto(); if (err) { Warning("Cannot register vsock protocol.\n"); VSockVmciUnregisterWithVmci(); unregister_ioctl32_handlers(); misc_deregister(&vsockVmciDevice); return err; } VSockVmciInitTables(); return 0; } /* *---------------------------------------------------------------------------- * * VSocketVmciExit -- * * VSockets module exit routine. * * Results: * None. * * Side effects: * Unregisters VSocket protocol family and socket operations. * *---------------------------------------------------------------------------- */ static void __exit VSockVmciExit(void) { unregister_ioctl32_handlers(); misc_deregister(&vsockVmciDevice); compat_mutex_lock(®istrationMutex); VSockVmciUnregisterAddressFamily(); compat_mutex_unlock(®istrationMutex); VSockVmciUnregisterProto(); VSockVmciUnregisterWithVmci(); } module_init(VSockVmciInit); module_exit(VSockVmciExit); MODULE_AUTHOR("VMware, Inc."); MODULE_DESCRIPTION("VMware Virtual Socket Family"); MODULE_VERSION(VSOCK_DRIVER_VERSION_STRING); MODULE_LICENSE("GPL v2"); MODULE_ALIAS("vmware_vsock"); /* * Starting with SLE10sp2, Novell requires that IHVs sign a support agreement * with them and mark their kernel modules as externally supported via a * change to the module header. If this isn't done, the module will not load * by default (i.e., neither mkinitrd nor modprobe will accept it). */ MODULE_INFO(supported, "external"); #ifdef VMX86_DEVEL /* We only support protocol negotiation overrides on devel builds. */ module_param(PROTOCOL_OVERRIDE, int, 0444); MODULE_PARM_DESC(PROTOCOL_OVERRIDE, "Specify a vsock protocol (auto negotiated by default"); int LOGLEVEL_THRESHOLD = 4; module_param(LOGLEVEL_THRESHOLD, int, 0444); MODULE_PARM_DESC(LOGLEVEL_THRESHOLD, "Set verbosity (0 means no log, 10 means very verbose, 4 is default)"); #endif vsock-only/linux/vsockAddr.h 0000444 0000000 0000000 00000003742 13432725346 015125 0 ustar root root /********************************************************* * Copyright (C) 2007 VMware, Inc. All rights reserved. * * This program is free software; you can redistribute it and/or modify it * under the terms of the GNU General Public License as published by the * Free Software Foundation version 2 and no later version. * * This program is distributed in the hope that it will be useful, but * WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY * or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License * for more details. * * You should have received a copy of the GNU General Public License along * with this program; if not, write to the Free Software Foundation, Inc., * 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA * *********************************************************/ /* * vsockAddr.h -- * * VSockets address constants, types and functions. */ #ifndef _VSOCK_ADDR_H_ #define _VSOCK_ADDR_H_ /* Assert that the given address is valid. */ #define VSOCK_ADDR_ASSERT(_a) \ ASSERT(0 == VSockAddr_Validate((_a))) #define VSOCK_ADDR_NOFAMILY_ASSERT(_a) \ ASSERT(0 == VSockAddr_ValidateNoFamily((_a))) void VSockAddr_Init(struct sockaddr_vm *addr, uint32 cid, uint32 port); void VSockAddr_InitNoFamily(struct sockaddr_vm *addr, uint32 cid, uint32 port); int32 VSockAddr_Validate(const struct sockaddr_vm *addr); int32 VSockAddr_ValidateNoFamily(const struct sockaddr_vm *addr); Bool VSockAddr_Bound(struct sockaddr_vm *addr); void VSockAddr_Unbind(struct sockaddr_vm *addr); Bool VSockAddr_EqualsAddr(struct sockaddr_vm *addr, struct sockaddr_vm *other); Bool VSockAddr_EqualsHandlePort(struct sockaddr_vm *addr, VMCIHandle handle, uint32 port); int32 VSockAddr_Cast(const struct sockaddr *addr, int32 len, struct sockaddr_vm **outAddr); Bool VSockAddr_SocketContextStream(uint32 cid); Bool VSockAddr_SocketContextDgram(uint32 cid, uint32 rid); #endif // _VSOCK_ADDR_H_ vsock-only/Makefile.kernel 0000444 0000000 0000000 00000005023 13432725350 014601 0 ustar root root #!/usr/bin/make -f ########################################################## # Copyright (C) 2007,2014 VMware, Inc. All rights reserved. # # This program is free software; you can redistribute it and/or modify it # under the terms of the GNU General Public License as published by the # Free Software Foundation version 2 and no later version. # # This program is distributed in the hope that it will be useful, but # WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY # or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License # for more details. # # You should have received a copy of the GNU General Public License along # with this program; if not, write to the Free Software Foundation, Inc., # 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA # ########################################################## vm_product_defines = $(if $(findstring tools,$(1)), -DVMX86_TOOLS,) CC_OPTS += $(call vm_product_defines, $(PRODUCT)) INCLUDE += -I. INCLUDE += -I$(SRCROOT)/include INCLUDE += -I$(SRCROOT)/linux INCLUDE += -I$(SRCROOT)/common EXTRA_CFLAGS := $(CC_OPTS) $(INCLUDE) EXTRA_CFLAGS += $(call vm_check_build, $(AUTOCONF_DIR)/netcreate_num_params.c, -DVMW_NETCREATE_KERNARG, ) MODPOST_VMCI_SYMVERS := $(wildcard $(MODULEBUILDDIR)/VMwareVMCIModule.symvers) obj-m += $(DRIVER).o $(DRIVER)-y := $(subst $(SRCROOT)/, , $(patsubst %.c, %.o, $(wildcard $(SRCROOT)/linux/*.c))) # # In open-vm-tools, need to compile the common sources from the shared directory. # DRIVERLOG := driverLog.o $(DRIVER)-y += $(DRIVERLOG) VSOCK_PATH := $(shell cd $(SRCROOT) && pwd) ifdef OVT_SOURCE_DIR DRIVERLOG_PATH := $(OVT_SOURCE_DIR)/modules/linux/shared else DRIVERLOG_PATH := $(VSOCK_PATH)/shared endif $(addprefix $(VSOCK_PATH)/,$(DRIVERLOG)): $(VSOCK_PATH)/%.o: $(DRIVERLOG_PATH)/%.c $(Q)$(rule_cc_o_c) clean: rm -rf $(wildcard $(DRIVER).mod.c $(DRIVER).ko .tmp_versions \ Module.symvers Module.markers modules.order \ $(foreach dir,./ linux/ \ ,$(addprefix $(dir),.*.cmd .*.o.flags *.o))) # # The VSock kernel module uses symbols from the VMCI kernel module. Copy the # Module.symvers file here so that the Vsock module knows about the VMCI version. This is not done for tar builds because the tools install takes care of it. # prebuild:: ifneq ($(MODULEBUILDDIR),) ifeq ($(MODPOST_VMCI_SYMVERS),) $(shell echo >&2 "Building vSockets without VMCI module symbols.") else $(shell echo >&2 "Building vSockets with VMCI module symbols.") cp -f $(MODPOST_VMCI_SYMVERS) $(SRCROOT)/Module.symvers endif endif