Merge branch 'master' into player-12.5.2

mark branch player-12.5.2 as retired
misc: warn if branch is marked as retired
2018-04-02 15:37:19 +02:00 · 2018-04-02 15:33:39 +02:00 · 2018-04-02 15:31:34 +02:00 · 2018-02-04 21:48:58 +01:00 · 2018-02-04 21:48:58 +01:00 · 2018-02-02 18:09:09 +01:00
22 changed files with 714 additions and 330 deletions
@@ -3,6 +3,8 @@
 *.ko
 *.ko.cmd
 *.mod.c
+*.tar
 .tmp_versions
+.cache.mk
 Module.symvers
 modules.order
@@ -0,0 +1,197 @@
+This document explains how to use the repository to retrieve the module
+source, build modules and install them. There are two basic methods: we can
+either build the modules from source and install them ourselves or replace
+VMware provided source tarballs with patched ones and let it use its own
+vmware-modconfig tool.
+
+
+0. Quick guide for impatient
+----------------------------
+
+First method (build and install):
+
+  wget https://github.com/mkubecek/vmware-host-modules/archive/workstation-14.1.1.tar.gz
+  tar -xzf workstation-14.1.1.tar.gz
+  cd vmware-host-modules-workstation-14.1.1
+  make
+  make install
+
+Last command must be executed with root privileges (first four shouldn't).
+Based on your VMware product, replace "14.1.1" with your installed version
+and/or "workstation" with "player".
+
+Second method (replace original tarballs):
+
+  wget https://github.com/mkubecek/vmware-host-modules/archive/workstation-14.1.1.tar.gz
+  tar -xzf workstation-14.1.1.tar.gz
+  cd vmware-host-modules-workstation-14.1.1
+  tar -cf vmmon.tar vmmon-only
+  tar -cf vmnet.tar vmnet-only
+  cp -v vmmon.tar vmnet.tar /usr/lib/vmware/modules/source/
+  vmware-modconfig --console --install-all
+
+In this case, last two commands require root privileges.
+
+
+1a. Retrieve the source - with git
+----------------------------------
+
+Using git is preferrable as it allows an easy switch to sources for
+different product/version or update for newer kernel version. Without git,
+you can only download one particular source snapshot.
+
+First, clone the repository from GitHub either using HTTPS
+
+  git clone https://github.com/mkubecek/vmware-host-modules.git
+
+or using SSH
+
+  git clone git@github.com:mkubecek/vmware-host-modules.git
+
+Using SSH is preferrable if you already have an account on GitHub, use
+HTTPS if you don't.
+
+Branch "master" cannot be used to build modules, it contains only common
+files so that changes in them can be merged into all other branches easily.
+To get actual sources, checkout a "real" branch, e.g.
+
+  git checkout workstation-14.1.1
+
+Branch name consists of "workstation" (for Workstation) or "player" (for
+Player), a dash and version number ("14.1.1" here). Always use correct
+product and version, the modules check if internal source version matches
+installed product and refuse to load otherwise. Actually, in all recent
+versions, "workstation" and "player" sources have been identical but there
+is no guarantee it will always be the case.
+
+To pull out updated branch (if there are new commits), run "git pull", to
+switch to a new branch run "git fetch" and "git checkout ...".
+
+
+1b. Retrieve the source - without git
+-------------------------------------
+
+GitHub allows to download a particular snapshot (branch or tag) directly
+without git:
+
+  wget https://github.com/mkubecek/vmware-host-modules/archive/workstation-14.1.1.tar.gz
+
+where "workstation" and "14.1.1" have the same meaning as in section 1a.
+Any other tool can be used to retrieve the URL, of course. It's also
+possible to download directly from the web interface, switch to the branch
+you are interested in, click the "Clone or download" button and then
+"Download ZIP".
+
+Unpack the downloaded archive
+
+  tar -xzf workstation-14.1.1.tar.gz
+
+or
+
+  unzip vmware-host-modules-workstation-14.1.1.zip
+
+depending on archive type. Then change to the resulting directory.
+
+
+2a. Build and install modules - directly
+----------------------------------------
+
+To build modules against currently running kernel, just run "make" in top
+level directory. It creates two module files vmmon-only/vmmon.ko and
+vmnet-only/vmnet.ko (unstripped). To build against a different kernel, set
+variable VM_UNAME, e.g.
+
+  make VM_UNAME='4.14.15-5-default'
+
+As name suggests, this variable should be set to what "uname -r" would show
+when that kernel is booted (it's the same as name of the subdirectory under
+/lib/modules with modules of that kernel).
+
+To install built modules for currently running kernel, run "make install".
+This command needs to be run as user with permissions to write into the
+directory with kernel modules (under normal circumstances, this should be
+only root). The VM_UNAME variable can be used to install for a different
+kernel. Unless you know well what are you doing and why, the same value
+should be always used for "make" and "make install". That's why makefile
+checks if vermagic tag of the moudules match VM_UNAME (or current kernel if
+VM_UNAME isn't set).
+
+If DESTDIR is not set (or is empty), "make install" also runs depmod to
+update module dependency cache.
+
+Note that "make install" replaces existing module files but it does not
+replace modules in running kernel if they are already loaded. Reloading the
+modules can be enforced e.g. by
+
+  /etc/init.d/vmware restart
+
+All running VMs should be turned off before this.
+
+
+2b. Build and install modules - replacing the original source tarballs
+----------------------------------------------------------------------
+
+Manual build and installation of host modules may be natural for developers
+and power users but those who prefer to "just use" would appreciate if they
+could rely on standard VMware tools. To this goal, we need to replace the
+module source tarballs with patched versions.
+
+In local checked out repository, the replacement tarballs can be created
+with
+
+  make tarballs
+
+This command creates files vmmon.tar and vmnet.tar which can be used to
+replace the original tarballs provided with the VMware product. Note that
+this command creates the tarball from current HEAD (in git language), i.e.
+they do not contain any uncommited local changes (e.g. files created by
+test build).
+
+When using a downloaded tarball, simply run
+
+  tar -cf vmmon.tar vmmon-only
+  tar -cf vmnet.tar vmmon-only
+
+to create the tarballs. In this case, there is no protection against
+unwanted local changes.
+
+Whatever way you used to create the tarballs, replace the original ones
+provided by VMware
+
+  /usr/lib/vmware/modules/source/vmmon.tar
+  /usr/lib/vmware/modules/source/vmnet.tar
+
+by patched versions. It is highly recommended to backup the original
+tarballs before replacing them.
+
+Once patched tarballs are installed, you can rebuild the modules as usual:
+
+  vmware-modconfig --console --install-all
+
+or let the GUI run the command for you.
+
+Replacing the original tarballs prevents having to rebuild and install the
+modules manually with every new kernel version. However, once the VMware
+product is upgraded or reinstalled, tarballs are replaced so that if the
+new VMware version doesn't build/work with a recent kernel, you still need
+to repeat the process. The same also holds if there is a new kernel which
+requires updated version of the patches.
+
+
+3. Final notes
+--------------
+
+Always make sure that you are using a branch matching the Workstation or
+Player version you are running. VMware software may refuse to start if
+module intended for different version is loaded. And unless you have
+a reason not to, use current head of that branch; even if the module
+builds, there may still be later fixes which may be important.
+
+In some web forums, one can comments suggesting to rip out one patched file
+and use it as replacement for the corresponding file with VMware tarballs.
+Please don't do this - and if you do, don't complain if something doesn't
+work. A commit fixing some problem is meant to be used as a whole; picking
+just changes in one file can lead to rather unpredictable results. While
+picking just one commit (or some of the commits) in a branch is perfectly
+fine (you should still know what are you doing and why), picking just one
+patched file rarely makes sense.
@@ -0,0 +1,48 @@
+MODULES = vmmon vmnet
+SUBDIRS = $(MODULES:%=%-only)
+TARBALLS = $(MODULES:%=%.tar)
+MODFILES = $(foreach mod,$(MODULES),$(mod)-only/$(mod).ko)
+VM_UNAME = $(shell uname -r)
+MODDIR = /lib/modules/$(VM_UNAME)/misc
+
+MODINFO = /sbin/modinfo
+DEPMOD = /sbin/depmod
+
+%.tar: FORCE gitcleancheck
+	git archive -o $@ --format=tar HEAD $(@:.tar=-only)
+
+.PHONY: FORCE subdirs $(SUBDIRS) clean tarballs
+
+subdirs: retiredcheck $(SUBDIRS)
+
+FORCE:
+
+$(SUBDIRS):
+	$(MAKE) -C $@ $(MAKECMDGOALS)
+
+gitcheck:
+	@git status >/dev/null 2>&1 \
+	     || ( echo "This only works in a git repository."; exit 1 )
+
+gitcleancheck: gitcheck
+	@git diff --exit-code HEAD >/dev/null 2>&1 \
+	     || echo "Warning: tarballs will reflect current HEAD (no uncommited changes)"
+
+retiredcheck:
+	@test -f RETIRED && cat RETIRED || true
+
+install: retiredcheck $(MODFILES)
+	@for f in $(MODFILES); do \
+	    mver=$$($(MODINFO) -F vermagic $$f);\
+	    mver=$${mver%% *};\
+	    test "$${mver}" = "$(VM_UNAME)" \
+	        || ( echo "Version mismatch: module $$f $${mver}, kernel $(VM_UNAME)" ; exit 1 );\
+	done
+	install -D -t $(DESTDIR)$(MODDIR) $(MODFILES)
+	strip --strip-debug $(MODULES:%=$(DESTDIR)$(MODDIR)/%.ko)
+	if test -z "$(DESTDIR)"; then $(DEPMOD) -a $(VM_UNAME); fi
+
+clean: $(SUBDIRS)
+
+tarballs: $(TARBALLS)
+
@@ -12,12 +12,17 @@ a particular version of Player or Workstation.

 From these tags, branches "player-${version}" and "workstation-${version}"
 are forked. These branches track changes needed to build these modules
-against recent kernel versions. Tag "p${ver}-k${ver}" marks the set of
-changes needed to build modules from Player ${ver} against kernel ${ver},
+against recent kernel versions. Tag "p${ver}-k${kver}" marks the set of
+changes needed to build modules from Player ${ver} against kernel ${kver},
 e.g. p12.5.5-k4.11 is for Player 12.5.5 and kernel 4.11; similar naming
 scheme "w${ver}-k${kver}" is used for Workstation modules. In general,
 later points in the branch should also work with older kernels (e.g.
-p12.5.5-k4.11 with kernel 4.10) but it's not completely reliable.
+p12.5.5-k4.11 with kernel 4.10) but it's not guaranteed; it may happen that
+a fix for new kernel version may be too complicated to make also work with
+older versions.
+
+At the moment, changes are tested to build against all (vanilla) kernel
+releases starting with 4.9.

 This repository is provided "as is" with no guarantees. Use the contents on
 your own risk.
@@ -0,0 +1,7 @@
+*-----------------------------------------------------------------------------
+* Branch player-12.5.2 has been retired on 2017-05-31 and has not been
+* receiving any updates since. Please switch to latest 12.5.x version
+* available.
+*
+* Latest kernel version this branch is intended to work with is 4.10.
+*-----------------------------------------------------------------------------
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998-2017 VMware, Inc. All rights reserved.
+ * Copyright (C) 1998-2014 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
@@ -122,10 +122,14 @@ EXTERN MPN HostIF_GetNextAnonPage(VMDriver *vm, MPN mpn);
 EXTERN int HostIF_GetLockedPageList(VMDriver *vm, VA64 uAddr,
                                    unsigned int numPages);

-EXTERN int HostIF_ReadPage(VMDriver *vm, MPN mpn, VA64 addr, Bool kernelBuffer);
-EXTERN int HostIF_WritePage(VMDriver *vm, MPN mpn, VA64 addr,
-                            Bool kernelBuffer);
+EXTERN int HostIF_ReadPage(MPN mpn, VA64 addr, Bool kernelBuffer);
+EXTERN int HostIF_WritePage(MPN mpn, VA64 addr, Bool kernelBuffer);
+#ifdef _WIN32
+/* Add a HostIF_ReadMachinePage() if/when needed */
 EXTERN int HostIF_WriteMachinePage(MPN mpn, VA64 addr);
+#else
+#define HostIF_WriteMachinePage(_a, _b) HostIF_WritePage((_a), (_b), TRUE)
+#endif
 #if defined __APPLE__
 // There is no need for a fast clock lock on Mac OS.
 #define HostIF_FastClockLock(_callerID) do {} while (0)
@@ -141,8 +145,4 @@ EXTERN void HostIF_FreeMachinePage(MPN mpn);

 EXTERN int HostIF_SafeRDMSR(uint32 msr, uint64 *val);

-#if defined __APPLE__
-EXTERN void HostIF_PageUnitTest(void);
-#endif
-
 #endif // ifdef _HOSTIF_H_
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998-2017 VMware, Inc. All rights reserved.
+ * 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
@@ -88,7 +88,6 @@

 #include "vmware.h"
 #include "hostif.h"
-#include "vmx86.h"

 #include "memtrack.h"

@@ -147,11 +146,12 @@ typedef struct MemTrackHT {
 typedef uint64 MemTrackHTKey;

 typedef struct MemTrack {
-   VMDriver         *vm;            /* The VM instance. */
   unsigned          numPages;      /* Number of pages tracked. */
   MemTrackDir1      dir1;          /* First level directory. */
   MemTrackHT        vpnHashTable;  /* VPN to entry hashtable. */
+#if defined(MEMTRACK_MPN_LOOKUP)
   MemTrackHT        mpnHashTable;  /* MPN to entry hashtable. */
+#endif
 } MemTrack;

 /*
@@ -304,9 +304,11 @@ MemTrackCleanup(MemTrack *mt)    // IN
      if (mt->vpnHashTable.pages[idx] != NULL) {
         HostIF_FreePage(mt->vpnHashTable.pages[idx]);
      }
+#if defined(MEMTRACK_MPN_LOOKUP)
      if (mt->mpnHashTable.pages[idx] != NULL) {
         HostIF_FreePage(mt->mpnHashTable.pages[idx]);
      }
+#endif
   }

   HostIF_FreeKernelMem(mt);
@@ -330,7 +332,7 @@ MemTrackCleanup(MemTrack *mt)    // IN
 */

 MemTrack *
-MemTrack_Init(VMDriver *vm) // IN:
+MemTrack_Init(void)
 {
   MemTrack *mt;
   unsigned idx;
@@ -347,7 +349,6 @@ MemTrack_Init(VMDriver *vm) // IN:
      goto error;
   }
   memset(mt, 0, sizeof *mt);
-   mt->vm = vm;

   for (idx = 0; idx < MEMTRACK_HT_PAGES; idx++) {
      MemTrackHTPage *htPage = MemTrackAllocPage();
@@ -359,6 +360,7 @@ MemTrack_Init(VMDriver *vm) // IN:
      mt->vpnHashTable.pages[idx] = htPage;
   }

+#if defined(MEMTRACK_MPN_LOOKUP)
   for (idx = 0; idx < MEMTRACK_HT_PAGES; idx++) {
      MemTrackHTPage *htPage = MemTrackAllocPage();

@@ -368,6 +370,7 @@ MemTrack_Init(VMDriver *vm) // IN:
      }
      mt->mpnHashTable.pages[idx] = htPage;
   }
+#endif

   return mt;

@@ -406,8 +409,6 @@ MemTrack_Add(MemTrack *mt,    // IN
   MemTrackDir3 *dir3;
   MEMTRACK_IDX2DIR(idx, p1, p2, p3);

-   ASSERT(HostIF_VMLockIsHeld(mt->vm));
-
   if (p1 >= MEMTRACK_DIR1_ENTRIES ||
       p2 >= MEMTRACK_DIR2_ENTRIES ||
       p3 >= MEMTRACK_DIR3_ENTRIES) {
@@ -429,7 +430,9 @@ MemTrack_Add(MemTrack *mt,    // IN
   ent->mpn = mpn;

   MemTrackHTInsert(&mt->vpnHashTable, ent, &ent->vpnChain, ent->vpn);
+#if defined(MEMTRACK_MPN_LOOKUP)
   MemTrackHTInsert(&mt->mpnHashTable, ent, &ent->mpnChain, ent->mpn);
+#endif

   mt->numPages++;

@@ -458,7 +461,6 @@ MemTrack_LookupVPN(MemTrack *mt, // IN
                   VPN64 vpn)    // IN
 {
   MemTrackEntry *next = *MemTrackHTLookup(&mt->vpnHashTable, vpn);
-   ASSERT(HostIF_VMLockIsHeld(mt->vm));

   while (next != NULL) {
      if (next->vpn == vpn) {
@@ -471,6 +473,7 @@ MemTrack_LookupVPN(MemTrack *mt, // IN
 }


+#if defined(MEMTRACK_MPN_LOOKUP)
 /*
 *----------------------------------------------------------------------
 *
@@ -490,9 +493,7 @@ MemTrackEntry *
 MemTrack_LookupMPN(MemTrack *mt, // IN
                   MPN mpn)      // IN
 {
-   MemTrackEntry *next;
-   ASSERT(HostIF_VMLockIsHeld(mt->vm));
-   next = *MemTrackHTLookup(&mt->mpnHashTable, mpn);
+   MemTrackEntry *next = *MemTrackHTLookup(&mt->mpnHashTable, mpn);

   while (next != NULL) {
      if (next->mpn == mpn) {
@@ -503,6 +504,7 @@ MemTrack_LookupMPN(MemTrack *mt, // IN

   return NULL;
 }
+#endif


 /*
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998-2017 VMware, Inc. All rights reserved.
+ * 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
@@ -31,22 +31,30 @@
 #define INCLUDE_ALLOW_VMCORE
 #include "includeCheck.h"

+#if defined(VMX86_DEBUG)
+#define MEMTRACK_MPN_LOOKUP
+#endif
+
 struct MemTrack;

 typedef struct MemTrackEntry {
   VPN64                   vpn;
   MPN                     mpn;
   struct MemTrackEntry   *vpnChain;
+#if defined(MEMTRACK_MPN_LOOKUP)
   struct MemTrackEntry   *mpnChain;
+#endif
 } MemTrackEntry;

 typedef void (MemTrackCleanupCb)(void *cData, MemTrackEntry *entry);

-extern struct MemTrack *MemTrack_Init(VMDriver *vm);
+extern struct MemTrack *MemTrack_Init(void);
 extern unsigned MemTrack_Cleanup(struct MemTrack *mt, MemTrackCleanupCb *cb,
                                 void *cbData);
 extern MemTrackEntry *MemTrack_Add(struct MemTrack *mt, VPN64 vpn, MPN mpn);
 extern MemTrackEntry *MemTrack_LookupVPN(struct MemTrack *mt, VPN64 vpn);
+#if defined(MEMTRACK_MPN_LOOKUP)
 extern MemTrackEntry *MemTrack_LookupMPN(struct MemTrack *mt, MPN mpn);
+#endif

 #endif // _MEMTRACK_H_
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998-2017 VMware, Inc. All rights reserved.
+ * 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
@@ -39,9 +39,6 @@
 #   include <linux/string.h> /* memset() in the kernel */

 #   define EXPORT_SYMTAB
-#   if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 12, 0)
-#      define LINUX_GDT_IS_RO
-#   endif
 #else
 #   include <string.h>
 #endif
@@ -63,13 +60,6 @@
 #include "apic.h"
 #include "x86perfctr.h"

-#ifdef LINUX_GDT_IS_RO
-#   include <asm/desc.h>
-#   define default_rw_gdt get_current_gdt_rw()
-#else
-#   define default_rw_gdt NULL
-#endif
-
 #if defined(_WIN64)
 #   include "x86.h"
 #   include "vmmon-asm-x86-64.h"
@@ -701,28 +691,11 @@ TaskRestoreHostGDTTRLDT(Descriptor *tempGDTBase,
       */

      desc = (Descriptor *)((VA)HOST_KERNEL_LA_2_VA(hostGDT64.offset + tr));
-#ifdef LINUX_GDT_IS_RO
-      /*
-       * If GDT is read-only, we must always load TR from alternative gdt,
-       * otherwise CPU gets page fault when marking TR busy.
-       */
-      {
-         DTR64 rwGDT64;
-
-         rwGDT64.offset = (unsigned long)tempGDTBase;
-         rwGDT64.limit = hostGDT64.limit;
-         Desc_SetType((Descriptor *)((unsigned long)tempGDTBase + tr), TASK_DESC);
-         _Set_GDT((DTR *)&rwGDT64);
-         SET_TR(tr);
-         _Set_GDT((DTR *)&hostGDT64);
-      }
-#else
      if (Desc_Type(desc) == TASK_DESC_BUSY) {
         Desc_SetType(desc, TASK_DESC);
      }
      _Set_GDT((DTR *)&hostGDT64);
      SET_TR(tr);
-#endif
      SET_LDT(ldt);
   }
 }
@@ -1774,8 +1747,7 @@ Task_Switch(VMDriver *vm,  // IN
   ASSERT(pCPU < ARRAYSIZE(hvRootPage) && pCPU < ARRAYSIZE(tmpGDT));

   hvRootMPN = Atomic_Read64(&hvRootPage[pCPU]);
-   tempGDTBase = USE_TEMPORARY_GDT ? Atomic_ReadPtr(&tmpGDT[pCPU])
-                                   : default_rw_gdt;
+   tempGDTBase = USE_TEMPORARY_GDT ? Atomic_ReadPtr(&tmpGDT[pCPU]) : NULL;

   /*
    * We can't allocate memory with interrupts disabled on all hosts
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998-2017 VMware, Inc. All rights reserved.
+ * 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
@@ -720,35 +720,6 @@ cleanup:
 }


-/*
- *----------------------------------------------------------------------
- *
- * Vmx86_LookupUserMPN --
- *
- *      Look up the MPN of a locked user page by user VA under the VM lock.
- *
- * Results:
- *      A status code and the MPN on success.
- *
- * Side effects:
- *      None
- *
- *----------------------------------------------------------------------
- */
-
-int
-Vmx86_LookupUserMPN(VMDriver *vm, // IN: VMDriver
-                    VA64 uAddr,   // IN: user VA of the page
-                    MPN *mpn)     // OUT
-{
-   int ret;
-   HostIF_VMLock(vm, 38);
-   ret = HostIF_LookupUserMPN(vm, uAddr, mpn);
-   HostIF_VMUnlock(vm, 38);
-   return ret;
-}
-
-
 /*
 *----------------------------------------------------------------------
 *
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998-2017 VMware, Inc. All rights reserved.
+ * Copyright (C) 1998-2013,2015-2016 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
@@ -106,7 +106,6 @@ extern PseudoTSC pseudoTSC;
 #define MAX_LOCKED_PAGES (-1)

 extern VMDriver *Vmx86_CreateVM(void);
-extern int Vmx86_LookupUserMPN(VMDriver *vm, VA64 uAddr, MPN *mpn);
 extern int Vmx86_ReleaseVM(VMDriver *vm);
 extern int Vmx86_InitVM(VMDriver *vm, InitBlock *initParams);
 extern int Vmx86_LateInitVM(VMDriver *vm);
@@ -24,11 +24,7 @@
 * 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)
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 2002-2017 VMware, Inc. All rights reserved.
+ * 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
@@ -30,32 +30,80 @@
 #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;
+/* pte_page() API modified in 2.3.23 to return a struct page * --hpreg */
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 3, 23)
+#   define compat_pte_page pte_page
 #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;
+#   include "compat_page.h"
+
+#   define compat_pte_page(_pte) virt_to_page(pte_page(_pte))
 #endif
-/* p[gu]d_large did not exist before 2.6.25 */
-#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 25)
-#   define pud_large(pud) 0
-#   define pgd_large(pgd) 0
+
+
+/* Appeared in 2.5.5 --hpreg */
+#ifndef pte_offset_map
+/*  Appeared in SuSE 8.0's 2.4.18 --hpreg */
+#   ifdef pte_offset_atomic
+#      define pte_offset_map pte_offset_atomic
+#      define pte_unmap pte_kunmap
+#   else
+#      define pte_offset_map pte_offset
+#      define pte_unmap(_pte)
+#   endif
 #endif
-/* pud_pfn did not exist before 3.8. */
-#if LINUX_VERSION_CODE < KERNEL_VERSION(3, 8, 0)
-#   define pud_pfn(pud)  INVALID_MPN
+
+
+/* Appeared in 2.5.74-mmX --petr */
+#ifndef pmd_offset_map
+#   define pmd_offset_map(pgd, address) pmd_offset(pgd, address)
+#   define pmd_unmap(pmd)
+#endif
+
+
+/*
+ * Appeared in 2.6.10-rc2-mm1.  Older kernels did L4 page tables as 
+ * part of pgd_offset, or they did not have L4 page tables at all.
+ * In 2.6.11 pml4 -> pgd -> pmd -> pte hierarchy was replaced by
+ * pgd -> pud -> pmd -> pte hierarchy.
+ */
+#ifdef PUD_MASK
+#   define compat_pgd_offset(mm, address)   pgd_offset(mm, address)
+#   define compat_pgd_present(pgd)          pgd_present(pgd)
+#   define compat_pud_offset(pgd, address)  pud_offset(pgd, address)
+#   define compat_pud_present(pud)          pud_present(pud)
+typedef pgd_t  compat_pgd_t;
+typedef pud_t  compat_pud_t;
+#elif defined(pml4_offset)
+#   define compat_pgd_offset(mm, address)   pml4_offset(mm, address)
+#   define compat_pgd_present(pml4)         pml4_present(pml4)
+#   define compat_pud_offset(pml4, address) pml4_pgd_offset(pml4, address)
+#   define compat_pud_present(pgd)          pgd_present(pgd)
+typedef pml4_t compat_pgd_t;
+typedef pgd_t  compat_pud_t;
+#else
+#   define compat_pgd_offset(mm, address)   pgd_offset(mm, address)
+#   define compat_pgd_present(pgd)          pgd_present(pgd)
+#   define compat_pud_offset(pgd, address)  (pgd)
+#   define compat_pud_present(pud)          (1)
+typedef pgd_t  compat_pgd_t;
+typedef pgd_t  compat_pud_t;
+#endif
+
+
+#define compat_pgd_offset_k(mm, address) pgd_offset_k(address)
+
+
+/* Introduced somewhere in 2.6.0, + backported to some 2.4 RedHat kernels */
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 0) && !defined(pte_pfn)
+#   define pte_pfn(pte) page_to_pfn(compat_pte_page(pte))
+#endif
+
+
+/* A page_table_lock field is added to struct mm_struct in 2.3.10 --hpreg */
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(2, 3, 10)
+#   define compat_get_page_table_lock(_mm) (&(_mm)->page_table_lock)
+#else
+#   define compat_get_page_table_lock(_mm) NULL
 #endif


@@ -80,8 +128,12 @@ typedef pgd_t compat_p4d_t;
 #define VM_PAGE_KERNEL_EXEC PAGE_KERNEL
 #endif
 #else
+#ifdef PAGE_KERNEL_EXECUTABLE
+#define VM_PAGE_KERNEL_EXEC PAGE_KERNEL_EXECUTABLE
+#else
 #define VM_PAGE_KERNEL_EXEC PAGE_KERNEL_EXEC
 #endif
+#endif


 #endif /* __COMPAT_PGTABLE_H__ */
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 2002,2014-2017 VMware, Inc. All rights reserved.
+ * Copyright (C) 2002,2014 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
@@ -26,14 +26,154 @@
 #include "compat_spinlock.h"
 #include "compat_page.h"

+/*
+ *-----------------------------------------------------------------------------
+ *
+ * PgtblPte2MPN --
+ *
+ *    Returns the page structure associated to a Page Table Entry.
+ *
+ *    This function is not allowed to schedule() because it can be called while
+ *    holding a spinlock --hpreg
+ *
+ * Results:
+ *    INVALID_MPN on failure
+ *    mpn         on success
+ *
+ * Side effects:
+ *    None
+ *
+ *-----------------------------------------------------------------------------
+ */
+
+static INLINE MPN
+PgtblPte2MPN(pte_t *pte)   // IN
+{
+   MPN mpn;
+   if (pte_present(*pte) == 0) {
+      return INVALID_MPN;
+   }
+   mpn = pte_pfn(*pte);
+   if (mpn >= INVALID_MPN) {
+      return INVALID_MPN;
+   }
+   return mpn;
+}
+
+
+/*
+ *-----------------------------------------------------------------------------
+ *
+ * PgtblPte2Page --
+ *
+ *    Returns the page structure associated to a Page Table Entry.
+ *
+ *    This function is not allowed to schedule() because it can be called while
+ *    holding a spinlock --hpreg
+ *
+ * Results:
+ *    The page structure if the page table entry points to a physical page
+ *    NULL if the page table entry does not point to a physical page
+ *
+ * Side effects:
+ *    None
+ *
+ *-----------------------------------------------------------------------------
+ */
+
+static INLINE struct page *
+PgtblPte2Page(pte_t *pte) // IN
+{
+   if (pte_present(*pte) == 0) {
+      return NULL;
+   }
+
+   return compat_pte_page(*pte);
+}
+
+
+/*
+ *-----------------------------------------------------------------------------
+ *
+ * PgtblPGD2PTELocked --
+ *
+ *    Walks through the hardware page tables to try to find the pte
+ *    associated to a virtual address.
+ *
+ * Results:
+ *    pte. Caller must call pte_unmap if valid pte returned.
+ *
+ * Side effects:
+ *    None
+ *
+ *-----------------------------------------------------------------------------
+ */
+
+static INLINE pte_t *
+PgtblPGD2PTELocked(compat_pgd_t *pgd,    // IN: PGD to start with
+                   VA addr)              // IN: Address in the virtual address
+                                         //     space of that process
+{
+   compat_pud_t *pud;
+   pmd_t *pmd;
+   pte_t *pte;
+
+   if (compat_pgd_present(*pgd) == 0) {
+      return NULL;
+   }
+
+   pud = compat_pud_offset(pgd, addr);
+   if (compat_pud_present(*pud) == 0) {
+      return NULL;
+   }
+
+   pmd = pmd_offset_map(pud, addr);
+   if (pmd_present(*pmd) == 0) {
+      pmd_unmap(pmd);
+      return NULL;
+   }
+
+   pte = pte_offset_map(pmd, addr);
+   pmd_unmap(pmd);
+   return pte;
+}
+
+
+/*
+ *-----------------------------------------------------------------------------
+ *
+ * PgtblVa2PTELocked --
+ *
+ *    Walks through the hardware page tables to try to find the pte
+ *    associated to a virtual address.
+ *
+ * Results:
+ *    pte. Caller must call pte_unmap if valid pte returned.
+ *
+ * Side effects:
+ *    None
+ *
+ *-----------------------------------------------------------------------------
+ */
+
+static INLINE pte_t *
+PgtblVa2PTELocked(struct mm_struct *mm, // IN: Mm structure of a process
+                  VA addr)              // IN: Address in the virtual address
+                                        //     space of that process
+{
+   return PgtblPGD2PTELocked(compat_pgd_offset(mm, addr), addr);
+}
+

 /*
 *-----------------------------------------------------------------------------
 *
 * PgtblVa2MPNLocked --
 *
- *    Walks through the hardware page tables to try to find the pte
- *    associated to a virtual address.  Then maps PTE to MPN.
+ *    Retrieve MPN for a given va.
+ *
+ *    Caller must call pte_unmap if valid pte returned. The mm->page_table_lock
+ *    must be held, so this function is not allowed to schedule() --hpreg
 *
 * Results:
 *    INVALID_MPN on failure
@@ -48,66 +188,91 @@
 static INLINE MPN
 PgtblVa2MPNLocked(struct mm_struct *mm, // IN: Mm structure of a process
                  VA addr)              // IN: Address in the virtual address
-                                        //     space of that process
 {
-   pgd_t *pgd;
-   compat_p4d_t *p4d;
-   MPN mpn;
+   pte_t *pte;

-   pgd = pgd_offset(mm, addr);
-   if (pgd_present(*pgd) == 0) {
-      return INVALID_MPN;
+   pte = PgtblVa2PTELocked(mm, addr);
+   if (pte != NULL) {
+      MPN mpn = PgtblPte2MPN(pte);
+      pte_unmap(pte);
+      return mpn;
   }
-   if (pgd_large(*pgd)) {
-      /* Linux kernel does not support PGD huge pages. */
-      /* return pgd_pfn(*pgd) + ((addr & PGD_MASK) >> PAGE_SHIFT); */
-      return INVALID_MPN;
-   }
-
-   p4d = compat_p4d_offset(pgd, addr);
-   if (compat_p4d_present(*p4d) == 0) {
-      return INVALID_MPN;
-   }
-   if (compat_p4d_large(*p4d)) {
-      mpn = compat_p4d_pfn(*p4d) + ((addr & ~COMPAT_P4D_MASK) >> PAGE_SHIFT);
-   } else {
-      pud_t *pud;
-
-      pud = pud_offset(p4d, addr);
-      if (pud_present(*pud) == 0) {
-         return INVALID_MPN;
-      }
-      if (pud_large(*pud)) {
-         mpn = pud_pfn(*pud) + ((addr & ~PUD_MASK) >> PAGE_SHIFT);
-      } else {
-         pmd_t *pmd;
-
-         pmd = pmd_offset(pud, addr);
-         if (pmd_present(*pmd) == 0) {
-            return INVALID_MPN;
-         }
-         if (pmd_large(*pmd)) {
-            mpn = pmd_pfn(*pmd) + ((addr & ~PMD_MASK) >> PAGE_SHIFT);
-         } else {
-            pte_t *pte;
-
-            pte = pte_offset_map(pmd, addr);
-            if (pte_present(*pte) == 0) {
-               pte_unmap(pte);
-               return INVALID_MPN;
-            }
-            mpn = pte_pfn(*pte);
-            pte_unmap(pte);
-         }
-      }
-   }
-   if (mpn >= INVALID_MPN) {
-      mpn = INVALID_MPN;
-   }
-   return mpn;
+   return INVALID_MPN;
 }


+#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 0)
+/*
+ *-----------------------------------------------------------------------------
+ *
+ * PgtblKVa2MPNLocked --
+ *
+ *    Retrieve MPN for a given kernel va.
+ *
+ *    Caller must call pte_unmap if valid pte returned. The mm->page_table_lock
+ *    must be held, so this function is not allowed to schedule() --hpreg
+ *
+ * Results:
+ *    INVALID_MPN on failure
+ *    mpn         on success
+ *
+ * Side effects:
+ *    None
+ *
+ *-----------------------------------------------------------------------------
+ */
+
+static INLINE MPN
+PgtblKVa2MPNLocked(struct mm_struct *mm, // IN: Mm structure of a caller
+                   VA addr)              // IN: Address in the virtual address
+{
+   pte_t *pte;
+
+   pte = PgtblPGD2PTELocked(compat_pgd_offset_k(mm, addr), addr);
+   if (pte != NULL) {
+      MPN mpn = PgtblPte2MPN(pte);
+      pte_unmap(pte);
+      return mpn;
+   }
+   return INVALID_MPN;
+}
+#endif
+
+
+/*
+ *-----------------------------------------------------------------------------
+ *
+ * PgtblVa2PageLocked --
+ *
+ *    Return the "page" struct for a given va.
+ *
+ * Results:
+ *    struct page or NULL.  The mm->page_table_lock must be held, so this 
+ *    function is not allowed to schedule() --hpreg
+ *
+ * Side effects:
+ *    None
+ *
+ *-----------------------------------------------------------------------------
+ */
+
+static INLINE struct page *
+PgtblVa2PageLocked(struct mm_struct *mm, // IN: Mm structure of a process
+                   VA addr)              // IN: Address in the virtual address
+{
+   pte_t *pte;
+
+   pte = PgtblVa2PTELocked(mm, addr);
+   if (pte != NULL) {
+      struct page *page = PgtblPte2Page(pte);
+      pte_unmap(pte);
+      return page;
+   } else {
+      return NULL;
+   }
+} 
+
+
 /*
 *-----------------------------------------------------------------------------
 *
@@ -133,10 +298,85 @@ PgtblVa2MPN(VA addr)  // IN

   /* current->mm is NULL for kernel threads, so use active_mm. */
   mm = current->active_mm;
-   spin_lock(&mm->page_table_lock);
+   if (compat_get_page_table_lock(mm)) {
+      spin_lock(compat_get_page_table_lock(mm));
+   }
   mpn = PgtblVa2MPNLocked(mm, addr);
-   spin_unlock(&mm->page_table_lock);
+   if (compat_get_page_table_lock(mm)) {
+      spin_unlock(compat_get_page_table_lock(mm));
+   }
   return mpn;
 }

+
+#if LINUX_VERSION_CODE < KERNEL_VERSION(2, 6, 0)
+/*
+ *-----------------------------------------------------------------------------
+ *
+ * PgtblKVa2MPN --
+ *
+ *    Walks through the hardware page tables of the current process to try to
+ *    find the page structure associated to a virtual address.
+ *
+ * Results:
+ *    Same as PgtblVa2MPNLocked()
+ *
+ * Side effects:
+ *    None
+ *
+ *-----------------------------------------------------------------------------
+ */
+
+static INLINE MPN
+PgtblKVa2MPN(VA addr)  // IN
+{
+   struct mm_struct *mm = current->active_mm;
+   MPN mpn;
+
+   if (compat_get_page_table_lock(mm)) {
+      spin_lock(compat_get_page_table_lock(mm));
+   }
+   mpn = PgtblKVa2MPNLocked(mm, addr);
+   if (compat_get_page_table_lock(mm)) {
+      spin_unlock(compat_get_page_table_lock(mm));
+   }
+   return mpn;
+}
+#endif
+
+
+/*
+ *-----------------------------------------------------------------------------
+ *
+ * PgtblVa2Page --
+ *
+ *    Walks through the hardware page tables of the current process to try to
+ *    find the page structure associated to a virtual address.
+ *
+ * Results:
+ *    Same as PgtblVa2PageLocked()
+ *
+ * Side effects:
+ *    None
+ *
+ *-----------------------------------------------------------------------------
+ */
+
+static INLINE struct page *
+PgtblVa2Page(VA addr) // IN
+{
+   struct mm_struct *mm = current->active_mm;
+   struct page *page;
+
+   if (compat_get_page_table_lock(mm)) {
+      spin_lock(compat_get_page_table_lock(mm));
+   }
+   page = PgtblVa2PageLocked(mm, addr);
+   if (compat_get_page_table_lock(mm)) {
+      spin_unlock(compat_get_page_table_lock(mm));
+   }
+   return page;
+}
+
+
 #endif /* __PGTBL_H__ */
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998-2017 VMware, Inc. All rights reserved.
+ * 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
@@ -104,9 +104,7 @@ long LinuxDriver_Ioctl(struct file *filp, u_int iocmd,

 static int LinuxDriver_Close(struct inode *inode, struct file *filp);
 static unsigned int LinuxDriverPoll(struct file *file, poll_table *wait);
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 11, 0)
-static int LinuxDriverFault(struct vm_fault *fault);
-#elif defined(VMW_NOPAGE_2624)
+#if defined(VMW_NOPAGE_2624)
 static int LinuxDriverFault(struct vm_area_struct *vma, struct vm_fault *fault);
 #else
 static struct page *LinuxDriverNoPage(struct vm_area_struct *vma,
@@ -883,10 +881,7 @@ LinuxDriverPollTimeout(unsigned long clientData)  // IN:
 *-----------------------------------------------------------------------------
 */

-#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 11, 0)
-static int
-LinuxDriverFault(struct vm_fault *fault)     //IN/OUT
-#elif defined(VMW_NOPAGE_2624)
+#if defined(VMW_NOPAGE_2624)
 static int LinuxDriverFault(struct vm_area_struct *vma, //IN
                            struct vm_fault *fault)     //IN/OUT
 #else
@@ -895,9 +890,6 @@ static struct page *LinuxDriverNoPage(struct vm_area_struct *vma, //IN
                                      int *type)                  //OUT: Fault type
 #endif
 {
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 11, 0)
-   struct vm_area_struct *vma = fault->vma;
-#endif
   VMLinux *vmLinux = (VMLinux *) vma->vm_file->private_data;
   unsigned long pg;
   struct page* page;
@@ -1358,6 +1350,7 @@ LinuxDriver_Ioctl(struct file *filp,    // IN:
   case IOCTL_VMX86_CREATE_VM:
   case IOCTL_VMX86_INIT_CROSSGDT:
   case IOCTL_VMX86_SET_UID:
+   case IOCTL_VMX86_LOOK_UP_MPN:
   case IOCTL_VMX86_GET_NUM_VMS:
   case IOCTL_VMX86_GET_TOTAL_MEM_USAGE:
   case IOCTL_VMX86_SET_HARD_LIMIT:
@@ -1367,6 +1360,8 @@ LinuxDriver_Ioctl(struct file *filp,    // IN:
   case IOCTL_VMX86_GET_KHZ_ESTIMATE:
   case IOCTL_VMX86_GET_ALL_CPUID:
   case IOCTL_VMX86_GET_ALL_MSRS:
+   case IOCTL_VMX86_READ_PAGE:
+   case IOCTL_VMX86_WRITE_PAGE:
   case IOCTL_VMX86_SET_POLL_TIMEOUT_PTR:
   case IOCTL_VMX86_GET_KERNEL_CLOCK_RATE:
   case IOCTL_VMX86_GET_REFERENCE_CLOCK_HZ:
@@ -1527,7 +1522,7 @@ LinuxDriver_Ioctl(struct file *filp,    // IN:
      if (retval) {
         break;
      }
-      args.ret.status = Vmx86_LookupUserMPN(vm, args.uAddr, &args.ret.mpn);
+      args.ret.status = HostIF_LookupUserMPN(vm, args.uAddr, &args.ret.mpn);
      retval = HostIF_CopyToUser((void *)ioarg, &args, sizeof args);
      break;
   }
@@ -1846,7 +1841,7 @@ LinuxDriver_Ioctl(struct file *filp,    // IN:
         if (retval) {
            break;
         }
-         retval = HostIF_ReadPage(vm, req.mpn, req.uAddr, FALSE);
+         retval = HostIF_ReadPage(req.mpn, req.uAddr, FALSE);
         break;
      }

@@ -1857,7 +1852,7 @@ LinuxDriver_Ioctl(struct file *filp,    // IN:
         if (retval) {
            break;
         }
-         retval = HostIF_WritePage(vm, req.mpn, req.uAddr, FALSE);
+         retval = HostIF_WritePage(req.mpn, req.uAddr, FALSE);
         break;
      }

@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998-2017 VMware, Inc. All rights reserved.
+ * Copyright (C) 1998-2014 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
@@ -73,22 +73,19 @@
 #include <linux/capability.h>
 #include <linux/kthread.h>
 #include <linux/wait.h>
-#include <linux/signal.h>
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 11, 0)
-#include <linux/taskstats_kern.h> // For linux/sched/signal.h without version check
-#endif
+

 #include "vmware.h"
 #include "x86apic.h"
 #include "vm_asm.h"
 #include "modulecall.h"
-#include "driver.h"
 #include "memtrack.h"
 #include "phystrack.h"
 #include "cpuid.h"
 #include "cpuid_info.h"
 #include "hostif.h"
 #include "hostif_priv.h"
+#include "driver.h"
 #include "vmhost.h"
 #include "x86msr.h"
 #include "apic.h"
@@ -1009,7 +1006,7 @@ HostIF_FreeLockedPages(VMDriver *vm,	     // IN: VM instance pointer
 int
 HostIF_Init(VMDriver *vm)  // IN:
 {
-   vm->memtracker = MemTrack_Init(vm);
+   vm->memtracker = MemTrack_Init();
   if (vm->memtracker == NULL) {
      return -1;
   }
@@ -2001,15 +1998,15 @@ HostIF_MapCrossPage(VMDriver *vm, // IN
      return NULL;
   }
   vPgAddr = (VA) MapCrossPage(page);
-   HostIF_VMLock(vm, 27);
+   HostIF_GlobalLock(16);
   if (vm->vmhost->crosspagePagesCount >= MAX_INITBLOCK_CPUS) {
-      HostIF_VMUnlock(vm, 27);
+      HostIF_GlobalUnlock(16);
      UnmapCrossPage(page, (void*)vPgAddr);

      return NULL;
   }
   vm->vmhost->crosspagePages[vm->vmhost->crosspagePagesCount++] = page;
-   HostIF_VMUnlock(vm, 27);
+   HostIF_GlobalUnlock(16);

   ret = vPgAddr | (((VA)p) & (PAGE_SIZE - 1));

@@ -2817,75 +2814,13 @@ HostIF_CallOnEachCPU(void (*func)(void*), // IN: function to call
 }


-/*
- *-----------------------------------------------------------------------------
- *
- * HostIFCheckTrackedMPN --
- *
- *      Check if a given MPN is tracked for the specified VM.
- *
- * Result:
- *      TRUE if the MPN is tracked in one of the trackers for the specified VM,
- *      FALSE otherwise.
- *
- * Side effects:
- *      None
- *
- *-----------------------------------------------------------------------------
- */
-
-Bool
-HostIFCheckTrackedMPN(VMDriver *vm, // IN: The VM instance
-                      MPN mpn)      // IN: The MPN
-{
-   VMHost * const vmh = vm->vmhost;
-
-   if (vmh == NULL) {
-      return FALSE;
-   }
-
-   HostIF_VMLock(vm, 32); // Debug version of PhysTrack wants VM's lock.
-   if (vmh->lockedPages) {
-      if (PhysTrack_Test(vmh->lockedPages, mpn)) {
-         HostIF_VMUnlock(vm, 32);
-         return TRUE;
-      }
-   }
-
-   if (vmh->AWEPages) {
-      if (PhysTrack_Test(vmh->AWEPages, mpn)) {
-         HostIF_VMUnlock(vm, 32);
-         return TRUE;
-      }
-   }
-
-   if (vm->memtracker) {
-      if (MemTrack_LookupMPN(vm->memtracker, mpn) != NULL) {
-         HostIF_VMUnlock(vm, 32);
-         return TRUE;
-      }
-   }
-   HostIF_VMUnlock(vm, 32);
-
-   if (vmx86_debug) {
-      /*
-       * The monitor may have old KSeg mappings to pages which it no longer
-       * owns.  Minimize customer noise by only logging this for developers.
-       */
-      Log("%s: MPN %" FMT64 "x not owned by this VM\n", __FUNCTION__, mpn);
-   }
-   return FALSE;
-}
-
-
 /*
 *----------------------------------------------------------------------
 *
 * HostIF_ReadPage --
 *
- *      Reads one page of data from a machine page and returns it in the
- *      specified kernel or user buffer.  The machine page must be owned by
- *      the specified VM.
+ *      puts the content of a machine page into a kernel or user mode 
+ *      buffer. 
 *
 * Results:
 *      0 on success
@@ -2898,8 +2833,7 @@ HostIFCheckTrackedMPN(VMDriver *vm, // IN: The VM instance
 */

 int
-HostIF_ReadPage(VMDriver *vm,        // IN: The VM instance
-                MPN mpn,             // MPN of the page
+HostIF_ReadPage(MPN mpn,             // MPN of the page
                VA64 addr,           // buffer for data
                Bool kernelBuffer)   // is the buffer in kernel space?
 {
@@ -2911,9 +2845,6 @@ HostIF_ReadPage(VMDriver *vm,        // IN: The VM instance
   if (mpn == INVALID_MPN) {
      return -EFAULT;
   }
-   if (HostIFCheckTrackedMPN(vm, mpn) == FALSE) {
-      return -EFAULT;
-   }

   page = pfn_to_page(mpn);
   ptr = kmap(page);
@@ -2937,8 +2868,8 @@ HostIF_ReadPage(VMDriver *vm,        // IN: The VM instance
 *
 * HostIF_WritePage --
 *
- *      Writes one page of data from a kernel or user buffer onto the specified
- *      machine page.  The machine page must be owned by the specified VM.
+ *      Put the content of a kernel or user mode buffer into a machine 
+ *      page.
 *
 * Results:
 *      0 on success
@@ -2951,9 +2882,9 @@ HostIF_ReadPage(VMDriver *vm,        // IN: The VM instance
 */

 int
-HostIFWritePageWork(MPN mpn,              // MPN of the page
-                    VA64 addr,            // data to write to the page
-                    Bool kernelBuffer)    // is the buffer in kernel space?
+HostIF_WritePage(MPN mpn,              // MPN of the page
+                 VA64 addr,            // data to write to the page
+                 Bool kernelBuffer)    // is the buffer in kernel space?
 {
   void const *buf = VA64ToPtr(addr);
   int ret = 0;
@@ -2980,45 +2911,6 @@ HostIFWritePageWork(MPN mpn,              // MPN of the page
   return ret;
 }

-int
-HostIF_WritePage(VMDriver *vm,      // IN: The VM instance
-                 MPN mpn,              // MPN of the page
-                 VA64 addr,            // data to write to the page
-                 Bool kernelBuffer)    // is the buffer in kernel space?
-{
-   if (HostIFCheckTrackedMPN(vm, mpn) == FALSE) {
-      return -EFAULT;
-   }
-   return HostIFWritePageWork(mpn, addr, kernelBuffer);
-}
-
-
-/*
- *-----------------------------------------------------------------------------
- *
- * HostIF_WriteMachinePage --
- *
- *      Puts the content of a machine page into a kernel or user mode
- *      buffer.  This should only be used for host-global pages, not any
- *      VM-owned pages.
- *
- * Results:
- *      On success: 0
- *      On failure: a negative error code
- *
- * Side effects:
- *      None
- *
- *-----------------------------------------------------------------------------
- */
-
-int
-HostIF_WriteMachinePage(MPN mpn,   // IN: MPN of the page
-                        VA64 addr) // IN: data to write to the page
-{
-   return HostIFWritePageWork(mpn, addr, TRUE);
-}
-

 /*
 *----------------------------------------------------------------------
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998-2017 VMware, Inc. All rights reserved.
+ * Copyright (C) 1998-2015 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
@@ -205,13 +205,11 @@ skipTaskSwitch:;
         uint32 nPages = (uint32)crosspage->args[1];
         VA64   uAddr  = (VA64)VPN_2_VA(vpn);
         ASSERT(nPages <= MODULECALL_NUM_ARGS);
-         HostIF_VMLock(vm, 38);
         for (i = 0; i < nPages; i++) {
            MPN mpn;
            HostIF_LookupUserMPN(vm, uAddr + i * PAGE_SIZE, &mpn);
            crosspage->args[i] = mpn;
         }
-         HostIF_VMUnlock(vm, 38);
         break;
      }

@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998-2013, 2017 VMware, Inc. All rights reserved.
+ * Copyright (C) 1998-2013 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
@@ -1104,7 +1104,7 @@ VNetBridgeNotifyLogBridgeUpError(int errno,        // IN: the error number
      case -ENODEV:
         LOG(0, (KERN_WARNING "bridge-%s: interface %s not found or not "
                 "up\n", bridgeName, devName));
-         break;
+	 break;
      case -EINVAL:
         LOG(0, (KERN_WARNING "bridge-%s: interface %s is not a valid "
                 "Ethernet interface\n", bridgeName, devName));
@@ -1143,16 +1143,11 @@ VNetBridgeNotifyLogBridgeUpError(int errno,        // IN: the error number
 static int
 VNetBridgeNotify(struct notifier_block *this, // IN: callback data (bridge)
                 u_long msg,                  // IN: type of event
-                 void *data)                  // IN: net_device or notifier info
+                 void *data)                  // IN: device pertaining to event
 {
   VNetBridge *bridge = list_entry(this, VNetBridge, notifier);
-   struct net_device *dev;
+   struct net_device *dev = (struct net_device *) data;

-#if LINUX_VERSION_CODE >= KERNEL_VERSION(3, 11, 0)
-   dev = netdev_notifier_info_to_dev(data);
-#else
-   dev = (struct net_device *)data;
-#endif
   switch (msg) {
   case NETDEV_UNREGISTER:
      LOG(2, (KERN_DEBUG "bridge-%s: interface %s is unregistering\n",
@@ -1184,7 +1179,7 @@ VNetBridgeNotify(struct notifier_block *this, // IN: callback data (bridge)

         LOG(1, (KERN_DEBUG "bridge-%s: enabling the bridge on dev up\n",
                 bridge->name));
-         errno = VNetBridgeUp(bridge, FALSE);
+	 errno = VNetBridgeUp(bridge, FALSE);
         if (errno != 0) {
            VNetBridgeNotifyLogBridgeUpError(errno, bridge->name, dev->name);
         }
@@ -337,8 +337,7 @@ typedef netdev_features_t compat_netdev_features_t;
 typedef u32 compat_netdev_features_t;
 #endif

-#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 7, 0) || \
-    (defined(RHEL_RELEASE_CODE) && RHEL_RELEASE_CODE >= 0x0704)
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 7, 0)
 #define compat_netif_trans_update(d) netif_trans_update(d)
 #else
 #define compat_netif_trans_update(d) do { (d)->trans_start = jiffies; } while (0)
@@ -36,9 +36,6 @@
 #include <linux/slab.h>
 #include <linux/version.h>
 #include <linux/wait.h>
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 11, 0)
-#include <linux/taskstats_kern.h>  // For <linux/sched/signal.h> without version dependency
-#endif

 #include <net/checksum.h>
 #include <net/sock.h>
@@ -35,6 +35,8 @@
 #endif
 #endif

+#include <linux/pci_ids.h>
+
 /* LSILogic 53C1030 Parallel SCSI controller
 * LSILogic SAS1068 SAS controller
 */
@@ -53,7 +55,10 @@
 *    VMware HD Audio codec
 *    VMware HD Audio controller
 */
+#ifndef PCI_VENDOR_ID_VMWARE
 #define PCI_VENDOR_ID_VMWARE                    0x15AD
+#endif
+
 #define PCI_DEVICE_ID_VMWARE_SVGA2              0x0405
 #define PCI_DEVICE_ID_VMWARE_SVGA               0x0710
 #define PCI_DEVICE_ID_VMWARE_VGA                0x0711
@@ -70,7 +75,11 @@
 #define PCI_DEVICE_ID_VMWARE_1394               0x0780
 #define PCI_DEVICE_ID_VMWARE_BRIDGE             0x0790
 #define PCI_DEVICE_ID_VMWARE_ROOTPORT           0x07A0
+
+#ifndef PCI_DEVICE_ID_VMWARE_VMXNET3
 #define PCI_DEVICE_ID_VMWARE_VMXNET3            0x07B0
+#endif
+
 #define PCI_DEVICE_ID_VMWARE_PVSCSI             0x07C0
 #define PCI_DEVICE_ID_VMWARE_82574              0x07D0
 #define PCI_DEVICE_ID_VMWARE_AHCI               0x07E0
@@ -1,5 +1,5 @@
 /*********************************************************
- * Copyright (C) 1998, 2017 VMware, Inc. All rights reserved.
+ * 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
@@ -77,7 +77,7 @@


 extern struct proto vmnet_proto;
-#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 2, 0) || defined(sk_net_refcnt)
+#if LINUX_VERSION_CODE >= KERNEL_VERSION(4, 2, 0)
 #   define compat_sk_alloc(_bri, _pri) sk_alloc(&init_net, \
                                                PF_NETLINK, _pri, &vmnet_proto, 1)
 #elif defined(VMW_NETDEV_HAS_NET)
Author	SHA1	Message	Date
Michal Kubecek	0dee99d40a	Merge branch 'master' into player-12.5.2	2018-04-02 15:37:19 +02:00
Michal Kubecek	343e2bbbc9	mark branch player-12.5.2 as retired	2018-04-02 15:33:39 +02:00
Michal Kubecek	77a6966b97	misc: warn if branch is marked as retired When "make" or "make install is issued on branch marked as retired, warn user about the fact.	2018-04-02 15:31:34 +02:00
Michal Kubecek	625084e96a	misc: add INSTALL document Add file INSTALL explaining how to use the repository (get patched sources, build the modules and install them).	2018-02-04 21:48:58 +01:00
Michal Kubecek	1d91cc5b6a	misc: add top level makefile Add a top level makefile to make common tasks easier. Supported targets: make - build modules make install - install modules (honors DESTDIR) make clean - clean up make tarballs - create source tarballs from current HEAD Note: "make tarballs" uses "git archive" so that it only works in a local copy of the git repository, not in a directory created by unpacking a tarball. Add *.tar to .gitignore to hide tarballs created by "make tarballs".	2018-02-04 21:48:58 +01:00
Michal Kubecek	6fd497d954	misc: update README Fix a type and clarify the part about building against older kernels.	2018-02-02 18:09:09 +01:00
Michal Kubecek	3fc524fd27	misc: ignore .cache.mk Add .cache.mk files to .gitignore.	2018-02-02 18:08:41 +01:00
Michal Kubecek	50b2ddbda3	vmnet: use standard definition of PCI_VENDOR_ID_VMWARE if available The PCI_DEVICE_ID_VMWARE_VMXNET3 macro is defined in mainline pci_ids.h since commit b1226c7db1d9 ("vmxnet3: Move PCI Id to pci_ids.h") in v4.10-rc1.	2017-05-31 09:21:31 +02:00
Michal Kubecek	bab1fc0501	all: replace get_user_pages() write/force parameters with flags Since commit 3b913179c3fa ("mm: replace get_user_pages_locked() write/force parameters with gup_flags") in v4.9-rc2, get_user_pages() has one flags argument rather than separate write and force flags.	2017-05-31 09:21:31 +02:00
Michal Kubecek	04b5b8ff63	vmnet: use standard definition of PCI_VENDOR_ID_VMWARE if available The PCI_VENDOR_ID_VMWARE macro is defined in mainline pci_ids.h since commit 94e57fea6202 ("PCI: Move PCI_VENDOR_ID_VMWARE to pci_ids.h") in v3.18-rc1.	2017-05-31 09:21:31 +02:00