[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
/*
|
|
|
|
* Kernel-based Virtual Machine driver for Linux
|
|
|
|
*
|
|
|
|
* This module enables machines with Intel VT-x extensions to run virtual
|
|
|
|
* machines without emulation or binary translation.
|
|
|
|
*
|
|
|
|
* MMU support
|
|
|
|
*
|
|
|
|
* Copyright (C) 2006 Qumranet, Inc.
|
2010-10-06 12:23:22 +00:00
|
|
|
* Copyright 2010 Red Hat, Inc. and/or its affiliates.
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
*
|
|
|
|
* Authors:
|
|
|
|
* Yaniv Kamay <yaniv@qumranet.com>
|
|
|
|
* Avi Kivity <avi@qumranet.com>
|
|
|
|
*
|
|
|
|
* This work is licensed under the terms of the GNU GPL, version 2. See
|
|
|
|
* the COPYING file in the top-level directory.
|
|
|
|
*
|
|
|
|
*/
|
2007-06-28 18:15:57 +00:00
|
|
|
|
2010-10-14 09:22:46 +00:00
|
|
|
#include "irq.h"
|
2007-12-14 01:35:10 +00:00
|
|
|
#include "mmu.h"
|
2010-01-21 13:31:49 +00:00
|
|
|
#include "x86.h"
|
2009-05-31 19:58:47 +00:00
|
|
|
#include "kvm_cache_regs.h"
|
2014-05-07 12:32:50 +00:00
|
|
|
#include "cpuid.h"
|
2007-06-28 18:15:57 +00:00
|
|
|
|
2007-12-16 09:02:48 +00:00
|
|
|
#include <linux/kvm_host.h>
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
#include <linux/types.h>
|
|
|
|
#include <linux/string.h>
|
|
|
|
#include <linux/mm.h>
|
|
|
|
#include <linux/highmem.h>
|
2016-07-14 00:19:00 +00:00
|
|
|
#include <linux/moduleparam.h>
|
|
|
|
#include <linux/export.h>
|
2007-11-26 12:08:14 +00:00
|
|
|
#include <linux/swap.h>
|
2008-02-23 14:44:30 +00:00
|
|
|
#include <linux/hugetlb.h>
|
2008-02-22 17:21:37 +00:00
|
|
|
#include <linux/compiler.h>
|
2009-12-23 16:35:21 +00:00
|
|
|
#include <linux/srcu.h>
|
include cleanup: Update gfp.h and slab.h includes to prepare for breaking implicit slab.h inclusion from percpu.h
percpu.h is included by sched.h and module.h and thus ends up being
included when building most .c files. percpu.h includes slab.h which
in turn includes gfp.h making everything defined by the two files
universally available and complicating inclusion dependencies.
percpu.h -> slab.h dependency is about to be removed. Prepare for
this change by updating users of gfp and slab facilities include those
headers directly instead of assuming availability. As this conversion
needs to touch large number of source files, the following script is
used as the basis of conversion.
http://userweb.kernel.org/~tj/misc/slabh-sweep.py
The script does the followings.
* Scan files for gfp and slab usages and update includes such that
only the necessary includes are there. ie. if only gfp is used,
gfp.h, if slab is used, slab.h.
* When the script inserts a new include, it looks at the include
blocks and try to put the new include such that its order conforms
to its surrounding. It's put in the include block which contains
core kernel includes, in the same order that the rest are ordered -
alphabetical, Christmas tree, rev-Xmas-tree or at the end if there
doesn't seem to be any matching order.
* If the script can't find a place to put a new include (mostly
because the file doesn't have fitting include block), it prints out
an error message indicating which .h file needs to be added to the
file.
The conversion was done in the following steps.
1. The initial automatic conversion of all .c files updated slightly
over 4000 files, deleting around 700 includes and adding ~480 gfp.h
and ~3000 slab.h inclusions. The script emitted errors for ~400
files.
2. Each error was manually checked. Some didn't need the inclusion,
some needed manual addition while adding it to implementation .h or
embedding .c file was more appropriate for others. This step added
inclusions to around 150 files.
3. The script was run again and the output was compared to the edits
from #2 to make sure no file was left behind.
4. Several build tests were done and a couple of problems were fixed.
e.g. lib/decompress_*.c used malloc/free() wrappers around slab
APIs requiring slab.h to be added manually.
5. The script was run on all .h files but without automatically
editing them as sprinkling gfp.h and slab.h inclusions around .h
files could easily lead to inclusion dependency hell. Most gfp.h
inclusion directives were ignored as stuff from gfp.h was usually
wildly available and often used in preprocessor macros. Each
slab.h inclusion directive was examined and added manually as
necessary.
6. percpu.h was updated not to include slab.h.
7. Build test were done on the following configurations and failures
were fixed. CONFIG_GCOV_KERNEL was turned off for all tests (as my
distributed build env didn't work with gcov compiles) and a few
more options had to be turned off depending on archs to make things
build (like ipr on powerpc/64 which failed due to missing writeq).
* x86 and x86_64 UP and SMP allmodconfig and a custom test config.
* powerpc and powerpc64 SMP allmodconfig
* sparc and sparc64 SMP allmodconfig
* ia64 SMP allmodconfig
* s390 SMP allmodconfig
* alpha SMP allmodconfig
* um on x86_64 SMP allmodconfig
8. percpu.h modifications were reverted so that it could be applied as
a separate patch and serve as bisection point.
Given the fact that I had only a couple of failures from tests on step
6, I'm fairly confident about the coverage of this conversion patch.
If there is a breakage, it's likely to be something in one of the arch
headers which should be easily discoverable easily on most builds of
the specific arch.
Signed-off-by: Tejun Heo <tj@kernel.org>
Guess-its-ok-by: Christoph Lameter <cl@linux-foundation.org>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Lee Schermerhorn <Lee.Schermerhorn@hp.com>
2010-03-24 08:04:11 +00:00
|
|
|
#include <linux/slab.h>
|
2017-02-08 17:51:30 +00:00
|
|
|
#include <linux/sched/signal.h>
|
2010-05-31 06:28:19 +00:00
|
|
|
#include <linux/uaccess.h>
|
kvm: x86: reduce collisions in mmu_page_hash
When using two-dimensional paging, the mmu_page_hash (which provides
lookups for existing kvm_mmu_page structs), becomes imbalanced; with
too many collisions in buckets 0 and 512. This has been seen to cause
mmu_lock to be held for multiple milliseconds in kvm_mmu_get_page on
VMs with a large amount of RAM mapped with 4K pages.
The current hash function uses the lower 10 bits of gfn to index into
mmu_page_hash. When doing shadow paging, gfn is the address of the
guest page table being shadow. These tables are 4K-aligned, which
makes the low bits of gfn a good hash. However, with two-dimensional
paging, no guest page tables are being shadowed, so gfn is the base
address that is mapped by the table. Thus page tables (level=1) have
a 2MB aligned gfn, page directories (level=2) have a 1GB aligned gfn,
etc. This means hashes will only differ in their 10th bit.
hash_64() provides a better hash. For example, on a VM with ~200G
(99458 direct=1 kvm_mmu_page structs):
hash max_mmu_page_hash_collisions
--------------------------------------------
low 10 bits 49847
hash_64 105
perfect 97
While we're changing the hash, increase the table size by 4x to better
support large VMs (further reduces number of collisions in 200G VM to
29).
Note that hash_64() does not provide a good distribution prior to commit
ef703f49a6c5 ("Eliminate bad hash multipliers from hash_32() and
hash_64()").
Signed-off-by: David Matlack <dmatlack@google.com>
Change-Id: I5aa6b13c834722813c6cca46b8b1ed6f53368ade
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2016-12-19 21:58:25 +00:00
|
|
|
#include <linux/hash.h>
|
2016-12-07 00:46:16 +00:00
|
|
|
#include <linux/kern_levels.h>
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2007-06-28 18:15:57 +00:00
|
|
|
#include <asm/page.h>
|
2017-12-20 07:29:29 +00:00
|
|
|
#include <asm/pat.h>
|
2007-06-28 18:15:57 +00:00
|
|
|
#include <asm/cmpxchg.h>
|
2007-11-21 12:08:40 +00:00
|
|
|
#include <asm/io.h>
|
2008-11-17 21:03:13 +00:00
|
|
|
#include <asm/vmx.h>
|
2016-02-24 09:51:11 +00:00
|
|
|
#include <asm/kvm_page_track.h>
|
2017-07-14 01:30:40 +00:00
|
|
|
#include "trace.h"
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2008-02-07 12:47:41 +00:00
|
|
|
/*
|
|
|
|
* When setting this variable to true it enables Two-Dimensional-Paging
|
|
|
|
* where the hardware walks 2 page tables:
|
|
|
|
* 1. the guest-virtual to guest-physical
|
|
|
|
* 2. while doing 1. it walks guest-physical to host-physical
|
|
|
|
* If the hardware supports that we don't need to do shadow paging.
|
|
|
|
*/
|
2008-02-22 17:21:37 +00:00
|
|
|
bool tdp_enabled = false;
|
2008-02-07 12:47:41 +00:00
|
|
|
|
2010-08-30 10:22:53 +00:00
|
|
|
enum {
|
|
|
|
AUDIT_PRE_PAGE_FAULT,
|
|
|
|
AUDIT_POST_PAGE_FAULT,
|
|
|
|
AUDIT_PRE_PTE_WRITE,
|
2010-09-27 10:09:29 +00:00
|
|
|
AUDIT_POST_PTE_WRITE,
|
|
|
|
AUDIT_PRE_SYNC,
|
|
|
|
AUDIT_POST_SYNC
|
2010-08-30 10:22:53 +00:00
|
|
|
};
|
2007-01-06 00:36:56 +00:00
|
|
|
|
2010-08-30 10:22:53 +00:00
|
|
|
#undef MMU_DEBUG
|
2007-01-06 00:36:56 +00:00
|
|
|
|
|
|
|
#ifdef MMU_DEBUG
|
2013-10-02 14:56:16 +00:00
|
|
|
static bool dbg = 0;
|
|
|
|
module_param(dbg, bool, 0644);
|
2007-01-06 00:36:56 +00:00
|
|
|
|
|
|
|
#define pgprintk(x...) do { if (dbg) printk(x); } while (0)
|
|
|
|
#define rmap_printk(x...) do { if (dbg) printk(x); } while (0)
|
2013-10-02 14:56:16 +00:00
|
|
|
#define MMU_WARN_ON(x) WARN_ON(x)
|
2007-01-06 00:36:56 +00:00
|
|
|
#else
|
|
|
|
#define pgprintk(x...) do { } while (0)
|
|
|
|
#define rmap_printk(x...) do { } while (0)
|
2013-10-02 14:56:16 +00:00
|
|
|
#define MMU_WARN_ON(x) do { } while (0)
|
2007-04-25 06:17:25 +00:00
|
|
|
#endif
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2010-08-22 11:12:48 +00:00
|
|
|
#define PTE_PREFETCH_NUM 8
|
|
|
|
|
2012-06-07 10:26:07 +00:00
|
|
|
#define PT_FIRST_AVAIL_BITS_SHIFT 10
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
#define PT64_SECOND_AVAIL_BITS_SHIFT 52
|
|
|
|
|
|
|
|
#define PT64_LEVEL_BITS 9
|
|
|
|
|
|
|
|
#define PT64_LEVEL_SHIFT(level) \
|
2007-10-08 13:02:08 +00:00
|
|
|
(PAGE_SHIFT + (level - 1) * PT64_LEVEL_BITS)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
|
|
|
#define PT64_INDEX(address, level)\
|
|
|
|
(((address) >> PT64_LEVEL_SHIFT(level)) & ((1 << PT64_LEVEL_BITS) - 1))
|
|
|
|
|
|
|
|
|
|
|
|
#define PT32_LEVEL_BITS 10
|
|
|
|
|
|
|
|
#define PT32_LEVEL_SHIFT(level) \
|
2007-10-08 13:02:08 +00:00
|
|
|
(PAGE_SHIFT + (level - 1) * PT32_LEVEL_BITS)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2009-07-27 14:30:45 +00:00
|
|
|
#define PT32_LVL_OFFSET_MASK(level) \
|
|
|
|
(PT32_BASE_ADDR_MASK & ((1ULL << (PAGE_SHIFT + (((level) - 1) \
|
|
|
|
* PT32_LEVEL_BITS))) - 1))
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
|
|
|
#define PT32_INDEX(address, level)\
|
|
|
|
(((address) >> PT32_LEVEL_SHIFT(level)) & ((1 << PT32_LEVEL_BITS) - 1))
|
|
|
|
|
|
|
|
|
2019-01-15 04:28:40 +00:00
|
|
|
#ifdef CONFIG_DYNAMIC_PHYSICAL_MASK
|
|
|
|
#define PT64_BASE_ADDR_MASK (physical_mask & ~(u64)(PAGE_SIZE-1))
|
|
|
|
#else
|
|
|
|
#define PT64_BASE_ADDR_MASK (((1ULL << 52) - 1) & ~(u64)(PAGE_SIZE-1))
|
|
|
|
#endif
|
2009-07-27 14:30:45 +00:00
|
|
|
#define PT64_LVL_ADDR_MASK(level) \
|
|
|
|
(PT64_BASE_ADDR_MASK & ~((1ULL << (PAGE_SHIFT + (((level) - 1) \
|
|
|
|
* PT64_LEVEL_BITS))) - 1))
|
|
|
|
#define PT64_LVL_OFFSET_MASK(level) \
|
|
|
|
(PT64_BASE_ADDR_MASK & ((1ULL << (PAGE_SHIFT + (((level) - 1) \
|
|
|
|
* PT64_LEVEL_BITS))) - 1))
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
|
|
|
#define PT32_BASE_ADDR_MASK PAGE_MASK
|
|
|
|
#define PT32_DIR_BASE_ADDR_MASK \
|
|
|
|
(PAGE_MASK & ~((1ULL << (PAGE_SHIFT + PT32_LEVEL_BITS)) - 1))
|
2009-07-27 14:30:45 +00:00
|
|
|
#define PT32_LVL_ADDR_MASK(level) \
|
|
|
|
(PAGE_MASK & ~((1ULL << (PAGE_SHIFT + (((level) - 1) \
|
|
|
|
* PT32_LEVEL_BITS))) - 1))
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2013-08-05 08:07:14 +00:00
|
|
|
#define PT64_PERM_MASK (PT_PRESENT_MASK | PT_WRITABLE_MASK | shadow_user_mask \
|
2017-07-17 21:10:27 +00:00
|
|
|
| shadow_x_mask | shadow_nx_mask | shadow_me_mask)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2007-12-09 14:15:46 +00:00
|
|
|
#define ACC_EXEC_MASK 1
|
|
|
|
#define ACC_WRITE_MASK PT_WRITABLE_MASK
|
|
|
|
#define ACC_USER_MASK PT_USER_MASK
|
|
|
|
#define ACC_ALL (ACC_EXEC_MASK | ACC_WRITE_MASK | ACC_USER_MASK)
|
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
/* The mask for the R/X bits in EPT PTEs */
|
|
|
|
#define PT64_EPT_READABLE_MASK 0x1ull
|
|
|
|
#define PT64_EPT_EXECUTABLE_MASK 0x4ull
|
|
|
|
|
2009-12-31 10:10:16 +00:00
|
|
|
#include <trace/events/kvm.h>
|
|
|
|
|
2009-07-06 09:21:32 +00:00
|
|
|
#define CREATE_TRACE_POINTS
|
|
|
|
#include "mmutrace.h"
|
|
|
|
|
2012-06-20 07:58:58 +00:00
|
|
|
#define SPTE_HOST_WRITEABLE (1ULL << PT_FIRST_AVAIL_BITS_SHIFT)
|
|
|
|
#define SPTE_MMU_WRITEABLE (1ULL << (PT_FIRST_AVAIL_BITS_SHIFT + 1))
|
2009-09-23 18:47:17 +00:00
|
|
|
|
2008-08-21 14:49:56 +00:00
|
|
|
#define SHADOW_PT_INDEX(addr, level) PT64_INDEX(addr, level)
|
|
|
|
|
2012-03-21 14:49:39 +00:00
|
|
|
/* make pte_list_desc fit well in cache line */
|
|
|
|
#define PTE_LIST_EXT 3
|
|
|
|
|
2017-08-17 13:03:32 +00:00
|
|
|
/*
|
|
|
|
* Return values of handle_mmio_page_fault and mmu.page_fault:
|
|
|
|
* RET_PF_RETRY: let CPU fault again on the address.
|
|
|
|
* RET_PF_EMULATE: mmio page fault, emulate the instruction directly.
|
|
|
|
*
|
|
|
|
* For handle_mmio_page_fault only:
|
|
|
|
* RET_PF_INVALID: the spte is invalid, let the real page fault path update it.
|
|
|
|
*/
|
|
|
|
enum {
|
|
|
|
RET_PF_RETRY = 0,
|
|
|
|
RET_PF_EMULATE = 1,
|
|
|
|
RET_PF_INVALID = 2,
|
|
|
|
};
|
|
|
|
|
2011-05-15 15:26:20 +00:00
|
|
|
struct pte_list_desc {
|
|
|
|
u64 *sptes[PTE_LIST_EXT];
|
|
|
|
struct pte_list_desc *more;
|
2007-01-06 00:36:38 +00:00
|
|
|
};
|
|
|
|
|
2008-12-25 12:39:47 +00:00
|
|
|
struct kvm_shadow_walk_iterator {
|
|
|
|
u64 addr;
|
|
|
|
hpa_t shadow_addr;
|
|
|
|
u64 *sptep;
|
2011-07-11 19:32:54 +00:00
|
|
|
int level;
|
2008-12-25 12:39:47 +00:00
|
|
|
unsigned index;
|
|
|
|
};
|
|
|
|
|
2018-06-27 21:59:07 +00:00
|
|
|
static const union kvm_mmu_page_role mmu_base_role_mask = {
|
|
|
|
.cr0_wp = 1,
|
|
|
|
.cr4_pae = 1,
|
|
|
|
.nxe = 1,
|
|
|
|
.smep_andnot_wp = 1,
|
|
|
|
.smap_andnot_wp = 1,
|
|
|
|
.smm = 1,
|
|
|
|
.guest_mode = 1,
|
|
|
|
.ad_disabled = 1,
|
|
|
|
};
|
|
|
|
|
2018-06-27 21:59:16 +00:00
|
|
|
#define for_each_shadow_entry_using_root(_vcpu, _root, _addr, _walker) \
|
|
|
|
for (shadow_walk_init_using_root(&(_walker), (_vcpu), \
|
|
|
|
(_root), (_addr)); \
|
|
|
|
shadow_walk_okay(&(_walker)); \
|
|
|
|
shadow_walk_next(&(_walker)))
|
|
|
|
|
|
|
|
#define for_each_shadow_entry(_vcpu, _addr, _walker) \
|
2008-12-25 12:39:47 +00:00
|
|
|
for (shadow_walk_init(&(_walker), _vcpu, _addr); \
|
|
|
|
shadow_walk_okay(&(_walker)); \
|
|
|
|
shadow_walk_next(&(_walker)))
|
|
|
|
|
2011-07-11 19:32:13 +00:00
|
|
|
#define for_each_shadow_entry_lockless(_vcpu, _addr, _walker, spte) \
|
|
|
|
for (shadow_walk_init(&(_walker), _vcpu, _addr); \
|
|
|
|
shadow_walk_okay(&(_walker)) && \
|
|
|
|
({ spte = mmu_spte_get_lockless(_walker.sptep); 1; }); \
|
|
|
|
__shadow_walk_next(&(_walker), spte))
|
|
|
|
|
2011-05-15 15:26:20 +00:00
|
|
|
static struct kmem_cache *pte_list_desc_cache;
|
2007-05-30 09:34:53 +00:00
|
|
|
static struct kmem_cache *mmu_page_header_cache;
|
KVM: create aggregate kvm_total_used_mmu_pages value
Of slab shrinkers, the VM code says:
* Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
* querying the cache size, so a fastpath for that case is appropriate.
and it *means* it. Look at how it calls the shrinkers:
nr_before = (*shrinker->shrink)(0, gfp_mask);
shrink_ret = (*shrinker->shrink)(this_scan, gfp_mask);
So, if you do anything stupid in your shrinker, the VM will doubly
punish you.
The mmu_shrink() function takes the global kvm_lock, then acquires
every VM's kvm->mmu_lock in sequence. If we have 100 VMs, then
we're going to take 101 locks. We do it twice, so each call takes
202 locks. If we're under memory pressure, we can have each cpu
trying to do this. It can get really hairy, and we've seen lock
spinning in mmu_shrink() be the dominant entry in profiles.
This is guaranteed to optimize at least half of those lock
aquisitions away. It removes the need to take any of the locks
when simply trying to count objects.
A 'percpu_counter' can be a large object, but we only have one
of these for the entire system. There are not any better
alternatives at the moment, especially ones that handle CPU
hotplug.
Signed-off-by: Dave Hansen <dave@linux.vnet.ibm.com>
Signed-off-by: Tim Pepper <lnxninja@linux.vnet.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2010-08-20 01:11:37 +00:00
|
|
|
static struct percpu_counter kvm_total_used_mmu_pages;
|
2007-04-15 13:31:09 +00:00
|
|
|
|
2008-04-25 13:13:50 +00:00
|
|
|
static u64 __read_mostly shadow_nx_mask;
|
|
|
|
static u64 __read_mostly shadow_x_mask; /* mutual exclusive with nx_mask */
|
|
|
|
static u64 __read_mostly shadow_user_mask;
|
|
|
|
static u64 __read_mostly shadow_accessed_mask;
|
|
|
|
static u64 __read_mostly shadow_dirty_mask;
|
2011-07-11 19:33:44 +00:00
|
|
|
static u64 __read_mostly shadow_mmio_mask;
|
2017-07-01 00:26:30 +00:00
|
|
|
static u64 __read_mostly shadow_mmio_value;
|
2016-07-12 22:18:49 +00:00
|
|
|
static u64 __read_mostly shadow_present_mask;
|
2017-07-17 21:10:27 +00:00
|
|
|
static u64 __read_mostly shadow_me_mask;
|
2011-07-11 19:33:44 +00:00
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
/*
|
2017-07-01 00:26:31 +00:00
|
|
|
* SPTEs used by MMUs without A/D bits are marked with shadow_acc_track_value.
|
|
|
|
* Non-present SPTEs with shadow_acc_track_value set are in place for access
|
|
|
|
* tracking.
|
2016-12-07 00:46:16 +00:00
|
|
|
*/
|
|
|
|
static u64 __read_mostly shadow_acc_track_mask;
|
|
|
|
static const u64 shadow_acc_track_value = SPTE_SPECIAL_MASK;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The mask/shift to use for saving the original R/X bits when marking the PTE
|
|
|
|
* as not-present for access tracking purposes. We do not save the W bit as the
|
|
|
|
* PTEs being access tracked also need to be dirty tracked, so the W bit will be
|
|
|
|
* restored only when a write is attempted to the page.
|
|
|
|
*/
|
|
|
|
static const u64 shadow_acc_track_saved_bits_mask = PT64_EPT_READABLE_MASK |
|
|
|
|
PT64_EPT_EXECUTABLE_MASK;
|
|
|
|
static const u64 shadow_acc_track_saved_bits_shift = PT64_SECOND_AVAIL_BITS_SHIFT;
|
|
|
|
|
2018-08-14 17:15:34 +00:00
|
|
|
/*
|
|
|
|
* This mask must be set on all non-zero Non-Present or Reserved SPTEs in order
|
|
|
|
* to guard against L1TF attacks.
|
|
|
|
*/
|
|
|
|
static u64 __read_mostly shadow_nonpresent_or_rsvd_mask;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The number of high-order 1 bits to use in the mask above.
|
|
|
|
*/
|
|
|
|
static const u64 shadow_nonpresent_or_rsvd_mask_len = 5;
|
|
|
|
|
KVM: x86: fix L1TF's MMIO GFN calculation
One defense against L1TF in KVM is to always set the upper five bits
of the *legal* physical address in the SPTEs for non-present and
reserved SPTEs, e.g. MMIO SPTEs. In the MMIO case, the GFN of the
MMIO SPTE may overlap with the upper five bits that are being usurped
to defend against L1TF. To preserve the GFN, the bits of the GFN that
overlap with the repurposed bits are shifted left into the reserved
bits, i.e. the GFN in the SPTE will be split into high and low parts.
When retrieving the GFN from the MMIO SPTE, e.g. to check for an MMIO
access, get_mmio_spte_gfn() unshifts the affected bits and restores
the original GFN for comparison. Unfortunately, get_mmio_spte_gfn()
neglects to mask off the reserved bits in the SPTE that were used to
store the upper chunk of the GFN. As a result, KVM fails to detect
MMIO accesses whose GPA overlaps the repurprosed bits, which in turn
causes guest panics and hangs.
Fix the bug by generating a mask that covers the lower chunk of the
GFN, i.e. the bits that aren't shifted by the L1TF mitigation. The
alternative approach would be to explicitly zero the five reserved
bits that are used to store the upper chunk of the GFN, but that
requires additional run-time computation and makes an already-ugly
bit of code even more inscrutable.
I considered adding a WARN_ON_ONCE(low_phys_bits-1 <= PAGE_SHIFT) to
warn if GENMASK_ULL() generated a nonsensical value, but that seemed
silly since that would mean a system that supports VMX has less than
18 bits of physical address space...
Reported-by: Sakari Ailus <sakari.ailus@iki.fi>
Fixes: d9b47449c1a1 ("kvm: x86: Set highest physical address bits in non-present/reserved SPTEs")
Cc: Junaid Shahid <junaids@google.com>
Cc: Jim Mattson <jmattson@google.com>
Cc: stable@vger.kernel.org
Reviewed-by: Junaid Shahid <junaids@google.com>
Tested-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-09-25 20:20:00 +00:00
|
|
|
/*
|
|
|
|
* In some cases, we need to preserve the GFN of a non-present or reserved
|
|
|
|
* SPTE when we usurp the upper five bits of the physical address space to
|
|
|
|
* defend against L1TF, e.g. for MMIO SPTEs. To preserve the GFN, we'll
|
|
|
|
* shift bits of the GFN that overlap with shadow_nonpresent_or_rsvd_mask
|
|
|
|
* left into the reserved bits, i.e. the GFN in the SPTE will be split into
|
|
|
|
* high and low parts. This mask covers the lower bits of the GFN.
|
|
|
|
*/
|
|
|
|
static u64 __read_mostly shadow_nonpresent_or_rsvd_lower_gfn_mask;
|
|
|
|
|
|
|
|
|
2011-07-11 19:33:44 +00:00
|
|
|
static void mmu_spte_set(u64 *sptep, u64 spte);
|
2018-06-27 21:59:07 +00:00
|
|
|
static union kvm_mmu_page_role
|
|
|
|
kvm_mmu_calc_root_page_role(struct kvm_vcpu *vcpu);
|
2011-07-11 19:33:44 +00:00
|
|
|
|
2018-12-06 13:21:08 +00:00
|
|
|
|
|
|
|
static inline bool kvm_available_flush_tlb_with_range(void)
|
|
|
|
{
|
|
|
|
return kvm_x86_ops->tlb_remote_flush_with_range;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void kvm_flush_remote_tlbs_with_range(struct kvm *kvm,
|
|
|
|
struct kvm_tlb_range *range)
|
|
|
|
{
|
|
|
|
int ret = -ENOTSUPP;
|
|
|
|
|
|
|
|
if (range && kvm_x86_ops->tlb_remote_flush_with_range)
|
|
|
|
ret = kvm_x86_ops->tlb_remote_flush_with_range(kvm, range);
|
|
|
|
|
|
|
|
if (ret)
|
|
|
|
kvm_flush_remote_tlbs(kvm);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void kvm_flush_remote_tlbs_with_address(struct kvm *kvm,
|
|
|
|
u64 start_gfn, u64 pages)
|
|
|
|
{
|
|
|
|
struct kvm_tlb_range range;
|
|
|
|
|
|
|
|
range.start_gfn = start_gfn;
|
|
|
|
range.pages = pages;
|
|
|
|
|
|
|
|
kvm_flush_remote_tlbs_with_range(kvm, &range);
|
|
|
|
}
|
|
|
|
|
2017-07-01 00:26:30 +00:00
|
|
|
void kvm_mmu_set_mmio_spte_mask(u64 mmio_mask, u64 mmio_value)
|
2011-07-11 19:33:44 +00:00
|
|
|
{
|
2017-07-01 00:26:30 +00:00
|
|
|
BUG_ON((mmio_mask & mmio_value) != mmio_value);
|
|
|
|
shadow_mmio_value = mmio_value | SPTE_SPECIAL_MASK;
|
2016-12-22 04:29:29 +00:00
|
|
|
shadow_mmio_mask = mmio_mask | SPTE_SPECIAL_MASK;
|
2011-07-11 19:33:44 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_set_mmio_spte_mask);
|
|
|
|
|
2017-07-01 00:26:31 +00:00
|
|
|
static inline bool sp_ad_disabled(struct kvm_mmu_page *sp)
|
|
|
|
{
|
|
|
|
return sp->role.ad_disabled;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline bool spte_ad_enabled(u64 spte)
|
|
|
|
{
|
|
|
|
MMU_WARN_ON((spte & shadow_mmio_mask) == shadow_mmio_value);
|
|
|
|
return !(spte & shadow_acc_track_value);
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline u64 spte_shadow_accessed_mask(u64 spte)
|
|
|
|
{
|
|
|
|
MMU_WARN_ON((spte & shadow_mmio_mask) == shadow_mmio_value);
|
|
|
|
return spte_ad_enabled(spte) ? shadow_accessed_mask : 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline u64 spte_shadow_dirty_mask(u64 spte)
|
|
|
|
{
|
|
|
|
MMU_WARN_ON((spte & shadow_mmio_mask) == shadow_mmio_value);
|
|
|
|
return spte_ad_enabled(spte) ? shadow_dirty_mask : 0;
|
|
|
|
}
|
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
static inline bool is_access_track_spte(u64 spte)
|
|
|
|
{
|
2017-07-01 00:26:31 +00:00
|
|
|
return !spte_ad_enabled(spte) && (spte & shadow_acc_track_mask) == 0;
|
2016-12-07 00:46:16 +00:00
|
|
|
}
|
|
|
|
|
2013-06-07 08:51:24 +00:00
|
|
|
/*
|
2019-02-05 21:01:16 +00:00
|
|
|
* Due to limited space in PTEs, the MMIO generation is a 19 bit subset of
|
|
|
|
* the memslots generation and is derived as follows:
|
2014-08-18 22:46:06 +00:00
|
|
|
*
|
2019-02-05 21:01:18 +00:00
|
|
|
* Bits 0-8 of the MMIO generation are propagated to spte bits 3-11
|
|
|
|
* Bits 9-18 of the MMIO generation are propagated to spte bits 52-61
|
2019-02-05 21:01:16 +00:00
|
|
|
*
|
2019-02-05 21:01:18 +00:00
|
|
|
* The KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS flag is intentionally not included in
|
|
|
|
* the MMIO generation number, as doing so would require stealing a bit from
|
|
|
|
* the "real" generation number and thus effectively halve the maximum number
|
|
|
|
* of MMIO generations that can be handled before encountering a wrap (which
|
|
|
|
* requires a full MMU zap). The flag is instead explicitly queried when
|
|
|
|
* checking for MMIO spte cache hits.
|
2013-06-07 08:51:24 +00:00
|
|
|
*/
|
2019-02-05 21:01:18 +00:00
|
|
|
#define MMIO_SPTE_GEN_MASK GENMASK_ULL(18, 0)
|
2019-02-05 21:01:16 +00:00
|
|
|
|
|
|
|
#define MMIO_SPTE_GEN_LOW_START 3
|
|
|
|
#define MMIO_SPTE_GEN_LOW_END 11
|
|
|
|
#define MMIO_SPTE_GEN_LOW_MASK GENMASK_ULL(MMIO_SPTE_GEN_LOW_END, \
|
|
|
|
MMIO_SPTE_GEN_LOW_START)
|
|
|
|
|
|
|
|
#define MMIO_SPTE_GEN_HIGH_START 52
|
|
|
|
#define MMIO_SPTE_GEN_HIGH_END 61
|
|
|
|
#define MMIO_SPTE_GEN_HIGH_MASK GENMASK_ULL(MMIO_SPTE_GEN_HIGH_END, \
|
|
|
|
MMIO_SPTE_GEN_HIGH_START)
|
2019-02-05 21:01:15 +00:00
|
|
|
static u64 generation_mmio_spte_mask(u64 gen)
|
2013-06-07 08:51:24 +00:00
|
|
|
{
|
|
|
|
u64 mask;
|
|
|
|
|
2019-02-05 21:01:16 +00:00
|
|
|
WARN_ON(gen & ~MMIO_SPTE_GEN_MASK);
|
2013-06-07 08:51:24 +00:00
|
|
|
|
2019-02-05 21:01:16 +00:00
|
|
|
mask = (gen << MMIO_SPTE_GEN_LOW_START) & MMIO_SPTE_GEN_LOW_MASK;
|
|
|
|
mask |= (gen << MMIO_SPTE_GEN_HIGH_START) & MMIO_SPTE_GEN_HIGH_MASK;
|
2013-06-07 08:51:24 +00:00
|
|
|
return mask;
|
|
|
|
}
|
|
|
|
|
2019-02-05 21:01:15 +00:00
|
|
|
static u64 get_mmio_spte_generation(u64 spte)
|
2013-06-07 08:51:24 +00:00
|
|
|
{
|
2019-02-05 21:01:15 +00:00
|
|
|
u64 gen;
|
2013-06-07 08:51:24 +00:00
|
|
|
|
|
|
|
spte &= ~shadow_mmio_mask;
|
|
|
|
|
2019-02-05 21:01:16 +00:00
|
|
|
gen = (spte & MMIO_SPTE_GEN_LOW_MASK) >> MMIO_SPTE_GEN_LOW_START;
|
|
|
|
gen |= (spte & MMIO_SPTE_GEN_HIGH_MASK) >> MMIO_SPTE_GEN_HIGH_START;
|
2019-02-05 21:01:18 +00:00
|
|
|
return gen;
|
2013-06-07 08:51:26 +00:00
|
|
|
}
|
|
|
|
|
2015-04-08 13:39:23 +00:00
|
|
|
static void mark_mmio_spte(struct kvm_vcpu *vcpu, u64 *sptep, u64 gfn,
|
2013-06-07 08:51:24 +00:00
|
|
|
unsigned access)
|
2011-07-11 19:33:44 +00:00
|
|
|
{
|
2019-02-05 21:01:16 +00:00
|
|
|
u64 gen = kvm_vcpu_memslots(vcpu)->generation & MMIO_SPTE_GEN_MASK;
|
2013-06-07 08:51:26 +00:00
|
|
|
u64 mask = generation_mmio_spte_mask(gen);
|
2018-08-14 17:15:34 +00:00
|
|
|
u64 gpa = gfn << PAGE_SHIFT;
|
2013-03-12 08:44:40 +00:00
|
|
|
|
2011-07-11 19:33:44 +00:00
|
|
|
access &= ACC_WRITE_MASK | ACC_USER_MASK;
|
2018-08-14 17:15:34 +00:00
|
|
|
mask |= shadow_mmio_value | access;
|
|
|
|
mask |= gpa | shadow_nonpresent_or_rsvd_mask;
|
|
|
|
mask |= (gpa & shadow_nonpresent_or_rsvd_mask)
|
|
|
|
<< shadow_nonpresent_or_rsvd_mask_len;
|
2013-06-07 08:51:24 +00:00
|
|
|
|
2019-02-05 21:01:23 +00:00
|
|
|
page_header(__pa(sptep))->mmio_cached = true;
|
|
|
|
|
2013-06-07 08:51:26 +00:00
|
|
|
trace_mark_mmio_spte(sptep, gfn, access, gen);
|
2013-06-07 08:51:24 +00:00
|
|
|
mmu_spte_set(sptep, mask);
|
2011-07-11 19:33:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static bool is_mmio_spte(u64 spte)
|
|
|
|
{
|
2017-07-01 00:26:30 +00:00
|
|
|
return (spte & shadow_mmio_mask) == shadow_mmio_value;
|
2011-07-11 19:33:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static gfn_t get_mmio_spte_gfn(u64 spte)
|
|
|
|
{
|
KVM: x86: fix L1TF's MMIO GFN calculation
One defense against L1TF in KVM is to always set the upper five bits
of the *legal* physical address in the SPTEs for non-present and
reserved SPTEs, e.g. MMIO SPTEs. In the MMIO case, the GFN of the
MMIO SPTE may overlap with the upper five bits that are being usurped
to defend against L1TF. To preserve the GFN, the bits of the GFN that
overlap with the repurposed bits are shifted left into the reserved
bits, i.e. the GFN in the SPTE will be split into high and low parts.
When retrieving the GFN from the MMIO SPTE, e.g. to check for an MMIO
access, get_mmio_spte_gfn() unshifts the affected bits and restores
the original GFN for comparison. Unfortunately, get_mmio_spte_gfn()
neglects to mask off the reserved bits in the SPTE that were used to
store the upper chunk of the GFN. As a result, KVM fails to detect
MMIO accesses whose GPA overlaps the repurprosed bits, which in turn
causes guest panics and hangs.
Fix the bug by generating a mask that covers the lower chunk of the
GFN, i.e. the bits that aren't shifted by the L1TF mitigation. The
alternative approach would be to explicitly zero the five reserved
bits that are used to store the upper chunk of the GFN, but that
requires additional run-time computation and makes an already-ugly
bit of code even more inscrutable.
I considered adding a WARN_ON_ONCE(low_phys_bits-1 <= PAGE_SHIFT) to
warn if GENMASK_ULL() generated a nonsensical value, but that seemed
silly since that would mean a system that supports VMX has less than
18 bits of physical address space...
Reported-by: Sakari Ailus <sakari.ailus@iki.fi>
Fixes: d9b47449c1a1 ("kvm: x86: Set highest physical address bits in non-present/reserved SPTEs")
Cc: Junaid Shahid <junaids@google.com>
Cc: Jim Mattson <jmattson@google.com>
Cc: stable@vger.kernel.org
Reviewed-by: Junaid Shahid <junaids@google.com>
Tested-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-09-25 20:20:00 +00:00
|
|
|
u64 gpa = spte & shadow_nonpresent_or_rsvd_lower_gfn_mask;
|
2018-08-14 17:15:34 +00:00
|
|
|
|
|
|
|
gpa |= (spte >> shadow_nonpresent_or_rsvd_mask_len)
|
|
|
|
& shadow_nonpresent_or_rsvd_mask;
|
|
|
|
|
|
|
|
return gpa >> PAGE_SHIFT;
|
2011-07-11 19:33:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static unsigned get_mmio_spte_access(u64 spte)
|
|
|
|
{
|
2019-02-05 21:01:16 +00:00
|
|
|
u64 mask = generation_mmio_spte_mask(MMIO_SPTE_GEN_MASK) | shadow_mmio_mask;
|
2013-06-07 08:51:24 +00:00
|
|
|
return (spte & ~mask) & ~PAGE_MASK;
|
2011-07-11 19:33:44 +00:00
|
|
|
}
|
|
|
|
|
2015-04-08 13:39:23 +00:00
|
|
|
static bool set_mmio_spte(struct kvm_vcpu *vcpu, u64 *sptep, gfn_t gfn,
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
kvm_pfn_t pfn, unsigned access)
|
2011-07-11 19:33:44 +00:00
|
|
|
{
|
|
|
|
if (unlikely(is_noslot_pfn(pfn))) {
|
2015-04-08 13:39:23 +00:00
|
|
|
mark_mmio_spte(vcpu, sptep, gfn, access);
|
2011-07-11 19:33:44 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
KVM: Allow not-present guest page faults to bypass kvm
There are two classes of page faults trapped by kvm:
- host page faults, where the fault is needed to allow kvm to install
the shadow pte or update the guest accessed and dirty bits
- guest page faults, where the guest has faulted and kvm simply injects
the fault back into the guest to handle
The second class, guest page faults, is pure overhead. We can eliminate
some of it on vmx using the following evil trick:
- when we set up a shadow page table entry, if the corresponding guest pte
is not present, set up the shadow pte as not present
- if the guest pte _is_ present, mark the shadow pte as present but also
set one of the reserved bits in the shadow pte
- tell the vmx hardware not to trap faults which have the present bit clear
With this, normal page-not-present faults go directly to the guest,
bypassing kvm entirely.
Unfortunately, this trick only works on Intel hardware, as AMD lacks a
way to discriminate among page faults based on error code. It is also
a little risky since it uses reserved bits which might become unreserved
in the future, so a module parameter is provided to disable it.
Signed-off-by: Avi Kivity <avi@qumranet.com>
2007-09-16 16:58:32 +00:00
|
|
|
|
2015-04-08 13:39:23 +00:00
|
|
|
static bool check_mmio_spte(struct kvm_vcpu *vcpu, u64 spte)
|
2013-06-07 08:51:26 +00:00
|
|
|
{
|
2019-02-05 21:01:16 +00:00
|
|
|
u64 kvm_gen, spte_gen, gen;
|
|
|
|
|
|
|
|
gen = kvm_vcpu_memslots(vcpu)->generation;
|
|
|
|
if (unlikely(gen & KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS))
|
|
|
|
return false;
|
2013-06-07 08:51:27 +00:00
|
|
|
|
2019-02-05 21:01:16 +00:00
|
|
|
kvm_gen = gen & MMIO_SPTE_GEN_MASK;
|
2013-06-07 08:51:27 +00:00
|
|
|
spte_gen = get_mmio_spte_generation(spte);
|
|
|
|
|
|
|
|
trace_check_mmio_spte(spte, kvm_gen, spte_gen);
|
|
|
|
return likely(kvm_gen == spte_gen);
|
2013-06-07 08:51:26 +00:00
|
|
|
}
|
|
|
|
|
2017-07-01 00:26:29 +00:00
|
|
|
/*
|
|
|
|
* Sets the shadow PTE masks used by the MMU.
|
|
|
|
*
|
|
|
|
* Assumptions:
|
|
|
|
* - Setting either @accessed_mask or @dirty_mask requires setting both
|
|
|
|
* - At least one of @accessed_mask or @acc_track_mask must be set
|
|
|
|
*/
|
2008-04-25 13:13:50 +00:00
|
|
|
void kvm_mmu_set_mask_ptes(u64 user_mask, u64 accessed_mask,
|
2016-12-07 00:46:16 +00:00
|
|
|
u64 dirty_mask, u64 nx_mask, u64 x_mask, u64 p_mask,
|
2017-07-17 21:10:27 +00:00
|
|
|
u64 acc_track_mask, u64 me_mask)
|
2008-04-25 13:13:50 +00:00
|
|
|
{
|
2017-07-01 00:26:29 +00:00
|
|
|
BUG_ON(!dirty_mask != !accessed_mask);
|
|
|
|
BUG_ON(!accessed_mask && !acc_track_mask);
|
2017-07-01 00:26:31 +00:00
|
|
|
BUG_ON(acc_track_mask & shadow_acc_track_value);
|
2016-12-22 04:29:29 +00:00
|
|
|
|
2008-04-25 13:13:50 +00:00
|
|
|
shadow_user_mask = user_mask;
|
|
|
|
shadow_accessed_mask = accessed_mask;
|
|
|
|
shadow_dirty_mask = dirty_mask;
|
|
|
|
shadow_nx_mask = nx_mask;
|
|
|
|
shadow_x_mask = x_mask;
|
2016-07-12 22:18:49 +00:00
|
|
|
shadow_present_mask = p_mask;
|
2016-12-07 00:46:16 +00:00
|
|
|
shadow_acc_track_mask = acc_track_mask;
|
2017-07-17 21:10:27 +00:00
|
|
|
shadow_me_mask = me_mask;
|
2008-04-25 13:13:50 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_set_mask_ptes);
|
|
|
|
|
2018-08-14 17:15:34 +00:00
|
|
|
static void kvm_mmu_reset_all_pte_masks(void)
|
2016-12-07 00:46:16 +00:00
|
|
|
{
|
KVM: x86: fix L1TF's MMIO GFN calculation
One defense against L1TF in KVM is to always set the upper five bits
of the *legal* physical address in the SPTEs for non-present and
reserved SPTEs, e.g. MMIO SPTEs. In the MMIO case, the GFN of the
MMIO SPTE may overlap with the upper five bits that are being usurped
to defend against L1TF. To preserve the GFN, the bits of the GFN that
overlap with the repurposed bits are shifted left into the reserved
bits, i.e. the GFN in the SPTE will be split into high and low parts.
When retrieving the GFN from the MMIO SPTE, e.g. to check for an MMIO
access, get_mmio_spte_gfn() unshifts the affected bits and restores
the original GFN for comparison. Unfortunately, get_mmio_spte_gfn()
neglects to mask off the reserved bits in the SPTE that were used to
store the upper chunk of the GFN. As a result, KVM fails to detect
MMIO accesses whose GPA overlaps the repurprosed bits, which in turn
causes guest panics and hangs.
Fix the bug by generating a mask that covers the lower chunk of the
GFN, i.e. the bits that aren't shifted by the L1TF mitigation. The
alternative approach would be to explicitly zero the five reserved
bits that are used to store the upper chunk of the GFN, but that
requires additional run-time computation and makes an already-ugly
bit of code even more inscrutable.
I considered adding a WARN_ON_ONCE(low_phys_bits-1 <= PAGE_SHIFT) to
warn if GENMASK_ULL() generated a nonsensical value, but that seemed
silly since that would mean a system that supports VMX has less than
18 bits of physical address space...
Reported-by: Sakari Ailus <sakari.ailus@iki.fi>
Fixes: d9b47449c1a1 ("kvm: x86: Set highest physical address bits in non-present/reserved SPTEs")
Cc: Junaid Shahid <junaids@google.com>
Cc: Jim Mattson <jmattson@google.com>
Cc: stable@vger.kernel.org
Reviewed-by: Junaid Shahid <junaids@google.com>
Tested-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-09-25 20:20:00 +00:00
|
|
|
u8 low_phys_bits;
|
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
shadow_user_mask = 0;
|
|
|
|
shadow_accessed_mask = 0;
|
|
|
|
shadow_dirty_mask = 0;
|
|
|
|
shadow_nx_mask = 0;
|
|
|
|
shadow_x_mask = 0;
|
|
|
|
shadow_mmio_mask = 0;
|
|
|
|
shadow_present_mask = 0;
|
|
|
|
shadow_acc_track_mask = 0;
|
2018-08-14 17:15:34 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If the CPU has 46 or less physical address bits, then set an
|
|
|
|
* appropriate mask to guard against L1TF attacks. Otherwise, it is
|
|
|
|
* assumed that the CPU is not vulnerable to L1TF.
|
|
|
|
*/
|
KVM: x86: fix L1TF's MMIO GFN calculation
One defense against L1TF in KVM is to always set the upper five bits
of the *legal* physical address in the SPTEs for non-present and
reserved SPTEs, e.g. MMIO SPTEs. In the MMIO case, the GFN of the
MMIO SPTE may overlap with the upper five bits that are being usurped
to defend against L1TF. To preserve the GFN, the bits of the GFN that
overlap with the repurposed bits are shifted left into the reserved
bits, i.e. the GFN in the SPTE will be split into high and low parts.
When retrieving the GFN from the MMIO SPTE, e.g. to check for an MMIO
access, get_mmio_spte_gfn() unshifts the affected bits and restores
the original GFN for comparison. Unfortunately, get_mmio_spte_gfn()
neglects to mask off the reserved bits in the SPTE that were used to
store the upper chunk of the GFN. As a result, KVM fails to detect
MMIO accesses whose GPA overlaps the repurprosed bits, which in turn
causes guest panics and hangs.
Fix the bug by generating a mask that covers the lower chunk of the
GFN, i.e. the bits that aren't shifted by the L1TF mitigation. The
alternative approach would be to explicitly zero the five reserved
bits that are used to store the upper chunk of the GFN, but that
requires additional run-time computation and makes an already-ugly
bit of code even more inscrutable.
I considered adding a WARN_ON_ONCE(low_phys_bits-1 <= PAGE_SHIFT) to
warn if GENMASK_ULL() generated a nonsensical value, but that seemed
silly since that would mean a system that supports VMX has less than
18 bits of physical address space...
Reported-by: Sakari Ailus <sakari.ailus@iki.fi>
Fixes: d9b47449c1a1 ("kvm: x86: Set highest physical address bits in non-present/reserved SPTEs")
Cc: Junaid Shahid <junaids@google.com>
Cc: Jim Mattson <jmattson@google.com>
Cc: stable@vger.kernel.org
Reviewed-by: Junaid Shahid <junaids@google.com>
Tested-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-09-25 20:20:00 +00:00
|
|
|
low_phys_bits = boot_cpu_data.x86_phys_bits;
|
2018-08-14 17:15:34 +00:00
|
|
|
if (boot_cpu_data.x86_phys_bits <
|
KVM: x86: fix L1TF's MMIO GFN calculation
One defense against L1TF in KVM is to always set the upper five bits
of the *legal* physical address in the SPTEs for non-present and
reserved SPTEs, e.g. MMIO SPTEs. In the MMIO case, the GFN of the
MMIO SPTE may overlap with the upper five bits that are being usurped
to defend against L1TF. To preserve the GFN, the bits of the GFN that
overlap with the repurposed bits are shifted left into the reserved
bits, i.e. the GFN in the SPTE will be split into high and low parts.
When retrieving the GFN from the MMIO SPTE, e.g. to check for an MMIO
access, get_mmio_spte_gfn() unshifts the affected bits and restores
the original GFN for comparison. Unfortunately, get_mmio_spte_gfn()
neglects to mask off the reserved bits in the SPTE that were used to
store the upper chunk of the GFN. As a result, KVM fails to detect
MMIO accesses whose GPA overlaps the repurprosed bits, which in turn
causes guest panics and hangs.
Fix the bug by generating a mask that covers the lower chunk of the
GFN, i.e. the bits that aren't shifted by the L1TF mitigation. The
alternative approach would be to explicitly zero the five reserved
bits that are used to store the upper chunk of the GFN, but that
requires additional run-time computation and makes an already-ugly
bit of code even more inscrutable.
I considered adding a WARN_ON_ONCE(low_phys_bits-1 <= PAGE_SHIFT) to
warn if GENMASK_ULL() generated a nonsensical value, but that seemed
silly since that would mean a system that supports VMX has less than
18 bits of physical address space...
Reported-by: Sakari Ailus <sakari.ailus@iki.fi>
Fixes: d9b47449c1a1 ("kvm: x86: Set highest physical address bits in non-present/reserved SPTEs")
Cc: Junaid Shahid <junaids@google.com>
Cc: Jim Mattson <jmattson@google.com>
Cc: stable@vger.kernel.org
Reviewed-by: Junaid Shahid <junaids@google.com>
Tested-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-09-25 20:20:00 +00:00
|
|
|
52 - shadow_nonpresent_or_rsvd_mask_len) {
|
2018-08-14 17:15:34 +00:00
|
|
|
shadow_nonpresent_or_rsvd_mask =
|
|
|
|
rsvd_bits(boot_cpu_data.x86_phys_bits -
|
|
|
|
shadow_nonpresent_or_rsvd_mask_len,
|
|
|
|
boot_cpu_data.x86_phys_bits - 1);
|
KVM: x86: fix L1TF's MMIO GFN calculation
One defense against L1TF in KVM is to always set the upper five bits
of the *legal* physical address in the SPTEs for non-present and
reserved SPTEs, e.g. MMIO SPTEs. In the MMIO case, the GFN of the
MMIO SPTE may overlap with the upper five bits that are being usurped
to defend against L1TF. To preserve the GFN, the bits of the GFN that
overlap with the repurposed bits are shifted left into the reserved
bits, i.e. the GFN in the SPTE will be split into high and low parts.
When retrieving the GFN from the MMIO SPTE, e.g. to check for an MMIO
access, get_mmio_spte_gfn() unshifts the affected bits and restores
the original GFN for comparison. Unfortunately, get_mmio_spte_gfn()
neglects to mask off the reserved bits in the SPTE that were used to
store the upper chunk of the GFN. As a result, KVM fails to detect
MMIO accesses whose GPA overlaps the repurprosed bits, which in turn
causes guest panics and hangs.
Fix the bug by generating a mask that covers the lower chunk of the
GFN, i.e. the bits that aren't shifted by the L1TF mitigation. The
alternative approach would be to explicitly zero the five reserved
bits that are used to store the upper chunk of the GFN, but that
requires additional run-time computation and makes an already-ugly
bit of code even more inscrutable.
I considered adding a WARN_ON_ONCE(low_phys_bits-1 <= PAGE_SHIFT) to
warn if GENMASK_ULL() generated a nonsensical value, but that seemed
silly since that would mean a system that supports VMX has less than
18 bits of physical address space...
Reported-by: Sakari Ailus <sakari.ailus@iki.fi>
Fixes: d9b47449c1a1 ("kvm: x86: Set highest physical address bits in non-present/reserved SPTEs")
Cc: Junaid Shahid <junaids@google.com>
Cc: Jim Mattson <jmattson@google.com>
Cc: stable@vger.kernel.org
Reviewed-by: Junaid Shahid <junaids@google.com>
Tested-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-09-25 20:20:00 +00:00
|
|
|
low_phys_bits -= shadow_nonpresent_or_rsvd_mask_len;
|
|
|
|
}
|
|
|
|
shadow_nonpresent_or_rsvd_lower_gfn_mask =
|
|
|
|
GENMASK_ULL(low_phys_bits - 1, PAGE_SHIFT);
|
2016-12-07 00:46:16 +00:00
|
|
|
}
|
|
|
|
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
static int is_cpuid_PSE36(void)
|
|
|
|
{
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2007-01-26 08:56:41 +00:00
|
|
|
static int is_nx(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2010-01-21 13:31:50 +00:00
|
|
|
return vcpu->arch.efer & EFER_NX;
|
2007-01-26 08:56:41 +00:00
|
|
|
}
|
|
|
|
|
KVM: Allow not-present guest page faults to bypass kvm
There are two classes of page faults trapped by kvm:
- host page faults, where the fault is needed to allow kvm to install
the shadow pte or update the guest accessed and dirty bits
- guest page faults, where the guest has faulted and kvm simply injects
the fault back into the guest to handle
The second class, guest page faults, is pure overhead. We can eliminate
some of it on vmx using the following evil trick:
- when we set up a shadow page table entry, if the corresponding guest pte
is not present, set up the shadow pte as not present
- if the guest pte _is_ present, mark the shadow pte as present but also
set one of the reserved bits in the shadow pte
- tell the vmx hardware not to trap faults which have the present bit clear
With this, normal page-not-present faults go directly to the guest,
bypassing kvm entirely.
Unfortunately, this trick only works on Intel hardware, as AMD lacks a
way to discriminate among page faults based on error code. It is also
a little risky since it uses reserved bits which might become unreserved
in the future, so a module parameter is provided to disable it.
Signed-off-by: Avi Kivity <avi@qumranet.com>
2007-09-16 16:58:32 +00:00
|
|
|
static int is_shadow_present_pte(u64 pte)
|
|
|
|
{
|
2016-12-07 00:46:16 +00:00
|
|
|
return (pte != 0) && !is_mmio_spte(pte);
|
KVM: Allow not-present guest page faults to bypass kvm
There are two classes of page faults trapped by kvm:
- host page faults, where the fault is needed to allow kvm to install
the shadow pte or update the guest accessed and dirty bits
- guest page faults, where the guest has faulted and kvm simply injects
the fault back into the guest to handle
The second class, guest page faults, is pure overhead. We can eliminate
some of it on vmx using the following evil trick:
- when we set up a shadow page table entry, if the corresponding guest pte
is not present, set up the shadow pte as not present
- if the guest pte _is_ present, mark the shadow pte as present but also
set one of the reserved bits in the shadow pte
- tell the vmx hardware not to trap faults which have the present bit clear
With this, normal page-not-present faults go directly to the guest,
bypassing kvm entirely.
Unfortunately, this trick only works on Intel hardware, as AMD lacks a
way to discriminate among page faults based on error code. It is also
a little risky since it uses reserved bits which might become unreserved
in the future, so a module parameter is provided to disable it.
Signed-off-by: Avi Kivity <avi@qumranet.com>
2007-09-16 16:58:32 +00:00
|
|
|
}
|
|
|
|
|
2008-02-23 14:44:30 +00:00
|
|
|
static int is_large_pte(u64 pte)
|
|
|
|
{
|
|
|
|
return pte & PT_PAGE_SIZE_MASK;
|
|
|
|
}
|
|
|
|
|
2009-06-10 15:27:03 +00:00
|
|
|
static int is_last_spte(u64 pte, int level)
|
|
|
|
{
|
|
|
|
if (level == PT_PAGE_TABLE_LEVEL)
|
|
|
|
return 1;
|
2009-07-27 14:30:44 +00:00
|
|
|
if (is_large_pte(pte))
|
2009-06-10 15:27:03 +00:00
|
|
|
return 1;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-12-22 04:29:32 +00:00
|
|
|
static bool is_executable_pte(u64 spte)
|
|
|
|
{
|
|
|
|
return (spte & (shadow_x_mask | shadow_nx_mask)) == shadow_x_mask;
|
|
|
|
}
|
|
|
|
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
static kvm_pfn_t spte_to_pfn(u64 pte)
|
2008-03-23 13:06:23 +00:00
|
|
|
{
|
2008-04-02 19:46:56 +00:00
|
|
|
return (pte & PT64_BASE_ADDR_MASK) >> PAGE_SHIFT;
|
2008-03-23 13:06:23 +00:00
|
|
|
}
|
|
|
|
|
2007-11-21 11:54:47 +00:00
|
|
|
static gfn_t pse36_gfn_delta(u32 gpte)
|
|
|
|
{
|
|
|
|
int shift = 32 - PT32_DIR_PSE36_SHIFT - PAGE_SHIFT;
|
|
|
|
|
|
|
|
return (gpte & PT32_DIR_PSE36_MASK) << shift;
|
|
|
|
}
|
|
|
|
|
2011-07-11 19:31:28 +00:00
|
|
|
#ifdef CONFIG_X86_64
|
2009-06-10 11:24:23 +00:00
|
|
|
static void __set_spte(u64 *sptep, u64 spte)
|
2007-05-31 12:46:04 +00:00
|
|
|
{
|
2016-05-11 15:04:29 +00:00
|
|
|
WRITE_ONCE(*sptep, spte);
|
2007-05-31 12:46:04 +00:00
|
|
|
}
|
|
|
|
|
2011-07-11 19:31:28 +00:00
|
|
|
static void __update_clear_spte_fast(u64 *sptep, u64 spte)
|
2010-06-06 11:48:06 +00:00
|
|
|
{
|
2016-05-11 15:04:29 +00:00
|
|
|
WRITE_ONCE(*sptep, spte);
|
2011-07-11 19:31:28 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static u64 __update_clear_spte_slow(u64 *sptep, u64 spte)
|
|
|
|
{
|
|
|
|
return xchg(sptep, spte);
|
|
|
|
}
|
2011-07-11 19:32:13 +00:00
|
|
|
|
|
|
|
static u64 __get_spte_lockless(u64 *sptep)
|
|
|
|
{
|
locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE()
Please do not apply this to mainline directly, instead please re-run the
coccinelle script shown below and apply its output.
For several reasons, it is desirable to use {READ,WRITE}_ONCE() in
preference to ACCESS_ONCE(), and new code is expected to use one of the
former. So far, there's been no reason to change most existing uses of
ACCESS_ONCE(), as these aren't harmful, and changing them results in
churn.
However, for some features, the read/write distinction is critical to
correct operation. To distinguish these cases, separate read/write
accessors must be used. This patch migrates (most) remaining
ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following
coccinelle script:
----
// Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and
// WRITE_ONCE()
// $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch
virtual patch
@ depends on patch @
expression E1, E2;
@@
- ACCESS_ONCE(E1) = E2
+ WRITE_ONCE(E1, E2)
@ depends on patch @
expression E;
@@
- ACCESS_ONCE(E)
+ READ_ONCE(E)
----
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: davem@davemloft.net
Cc: linux-arch@vger.kernel.org
Cc: mpe@ellerman.id.au
Cc: shuah@kernel.org
Cc: snitzer@redhat.com
Cc: thor.thayer@linux.intel.com
Cc: tj@kernel.org
Cc: viro@zeniv.linux.org.uk
Cc: will.deacon@arm.com
Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-10-23 21:07:29 +00:00
|
|
|
return READ_ONCE(*sptep);
|
2011-07-11 19:32:13 +00:00
|
|
|
}
|
2010-06-06 11:48:06 +00:00
|
|
|
#else
|
2011-07-11 19:31:28 +00:00
|
|
|
union split_spte {
|
|
|
|
struct {
|
|
|
|
u32 spte_low;
|
|
|
|
u32 spte_high;
|
|
|
|
};
|
|
|
|
u64 spte;
|
|
|
|
};
|
2010-06-06 11:48:06 +00:00
|
|
|
|
2011-07-11 19:32:13 +00:00
|
|
|
static void count_spte_clear(u64 *sptep, u64 spte)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp = page_header(__pa(sptep));
|
|
|
|
|
|
|
|
if (is_shadow_present_pte(spte))
|
|
|
|
return;
|
|
|
|
|
|
|
|
/* Ensure the spte is completely set before we increase the count */
|
|
|
|
smp_wmb();
|
|
|
|
sp->clear_spte_count++;
|
|
|
|
}
|
|
|
|
|
2011-07-11 19:31:28 +00:00
|
|
|
static void __set_spte(u64 *sptep, u64 spte)
|
|
|
|
{
|
|
|
|
union split_spte *ssptep, sspte;
|
2010-06-06 11:48:06 +00:00
|
|
|
|
2011-07-11 19:31:28 +00:00
|
|
|
ssptep = (union split_spte *)sptep;
|
|
|
|
sspte = (union split_spte)spte;
|
|
|
|
|
|
|
|
ssptep->spte_high = sspte.spte_high;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If we map the spte from nonpresent to present, We should store
|
|
|
|
* the high bits firstly, then set present bit, so cpu can not
|
|
|
|
* fetch this spte while we are setting the spte.
|
|
|
|
*/
|
|
|
|
smp_wmb();
|
|
|
|
|
2016-05-11 15:04:29 +00:00
|
|
|
WRITE_ONCE(ssptep->spte_low, sspte.spte_low);
|
2010-06-06 11:48:06 +00:00
|
|
|
}
|
|
|
|
|
2011-07-11 19:31:28 +00:00
|
|
|
static void __update_clear_spte_fast(u64 *sptep, u64 spte)
|
|
|
|
{
|
|
|
|
union split_spte *ssptep, sspte;
|
|
|
|
|
|
|
|
ssptep = (union split_spte *)sptep;
|
|
|
|
sspte = (union split_spte)spte;
|
|
|
|
|
2016-05-11 15:04:29 +00:00
|
|
|
WRITE_ONCE(ssptep->spte_low, sspte.spte_low);
|
2011-07-11 19:31:28 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If we map the spte from present to nonpresent, we should clear
|
|
|
|
* present bit firstly to avoid vcpu fetch the old high bits.
|
|
|
|
*/
|
|
|
|
smp_wmb();
|
|
|
|
|
|
|
|
ssptep->spte_high = sspte.spte_high;
|
2011-07-11 19:32:13 +00:00
|
|
|
count_spte_clear(sptep, spte);
|
2011-07-11 19:31:28 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static u64 __update_clear_spte_slow(u64 *sptep, u64 spte)
|
|
|
|
{
|
|
|
|
union split_spte *ssptep, sspte, orig;
|
|
|
|
|
|
|
|
ssptep = (union split_spte *)sptep;
|
|
|
|
sspte = (union split_spte)spte;
|
|
|
|
|
|
|
|
/* xchg acts as a barrier before the setting of the high bits */
|
|
|
|
orig.spte_low = xchg(&ssptep->spte_low, sspte.spte_low);
|
2011-09-19 04:19:51 +00:00
|
|
|
orig.spte_high = ssptep->spte_high;
|
|
|
|
ssptep->spte_high = sspte.spte_high;
|
2011-07-11 19:32:13 +00:00
|
|
|
count_spte_clear(sptep, spte);
|
2011-07-11 19:31:28 +00:00
|
|
|
|
|
|
|
return orig.spte;
|
|
|
|
}
|
2011-07-11 19:32:13 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The idea using the light way get the spte on x86_32 guest is from
|
|
|
|
* gup_get_pte(arch/x86/mm/gup.c).
|
2013-06-19 09:09:20 +00:00
|
|
|
*
|
|
|
|
* An spte tlb flush may be pending, because kvm_set_pte_rmapp
|
|
|
|
* coalesces them and we are running out of the MMU lock. Therefore
|
|
|
|
* we need to protect against in-progress updates of the spte.
|
|
|
|
*
|
|
|
|
* Reading the spte while an update is in progress may get the old value
|
|
|
|
* for the high part of the spte. The race is fine for a present->non-present
|
|
|
|
* change (because the high part of the spte is ignored for non-present spte),
|
|
|
|
* but for a present->present change we must reread the spte.
|
|
|
|
*
|
|
|
|
* All such changes are done in two steps (present->non-present and
|
|
|
|
* non-present->present), hence it is enough to count the number of
|
|
|
|
* present->non-present updates: if it changed while reading the spte,
|
|
|
|
* we might have hit the race. This is done using clear_spte_count.
|
2011-07-11 19:32:13 +00:00
|
|
|
*/
|
|
|
|
static u64 __get_spte_lockless(u64 *sptep)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp = page_header(__pa(sptep));
|
|
|
|
union split_spte spte, *orig = (union split_spte *)sptep;
|
|
|
|
int count;
|
|
|
|
|
|
|
|
retry:
|
|
|
|
count = sp->clear_spte_count;
|
|
|
|
smp_rmb();
|
|
|
|
|
|
|
|
spte.spte_low = orig->spte_low;
|
|
|
|
smp_rmb();
|
|
|
|
|
|
|
|
spte.spte_high = orig->spte_high;
|
|
|
|
smp_rmb();
|
|
|
|
|
|
|
|
if (unlikely(spte.spte_low != orig->spte_low ||
|
|
|
|
count != sp->clear_spte_count))
|
|
|
|
goto retry;
|
|
|
|
|
|
|
|
return spte.spte;
|
|
|
|
}
|
2011-07-11 19:31:28 +00:00
|
|
|
#endif
|
|
|
|
|
2016-12-07 00:46:11 +00:00
|
|
|
static bool spte_can_locklessly_be_made_writable(u64 spte)
|
2012-06-20 07:59:18 +00:00
|
|
|
{
|
2013-01-30 14:45:00 +00:00
|
|
|
return (spte & (SPTE_HOST_WRITEABLE | SPTE_MMU_WRITEABLE)) ==
|
|
|
|
(SPTE_HOST_WRITEABLE | SPTE_MMU_WRITEABLE);
|
2012-06-20 07:59:18 +00:00
|
|
|
}
|
|
|
|
|
2010-08-02 08:14:04 +00:00
|
|
|
static bool spte_has_volatile_bits(u64 spte)
|
|
|
|
{
|
2016-12-07 00:46:16 +00:00
|
|
|
if (!is_shadow_present_pte(spte))
|
|
|
|
return false;
|
|
|
|
|
2012-06-20 07:59:18 +00:00
|
|
|
/*
|
2016-02-23 23:34:30 +00:00
|
|
|
* Always atomically update spte if it can be updated
|
2012-06-20 07:59:18 +00:00
|
|
|
* out of mmu-lock, it can ensure dirty bit is not lost,
|
|
|
|
* also, it can help us to get a stable is_writable_pte()
|
|
|
|
* to ensure tlb flush is not missed.
|
|
|
|
*/
|
2016-12-07 00:46:16 +00:00
|
|
|
if (spte_can_locklessly_be_made_writable(spte) ||
|
|
|
|
is_access_track_spte(spte))
|
2012-06-20 07:59:18 +00:00
|
|
|
return true;
|
|
|
|
|
2017-07-01 00:26:31 +00:00
|
|
|
if (spte_ad_enabled(spte)) {
|
2016-12-07 00:46:16 +00:00
|
|
|
if ((spte & shadow_accessed_mask) == 0 ||
|
|
|
|
(is_writable_pte(spte) && (spte & shadow_dirty_mask) == 0))
|
|
|
|
return true;
|
|
|
|
}
|
2010-08-02 08:14:04 +00:00
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
return false;
|
2010-08-02 08:14:04 +00:00
|
|
|
}
|
|
|
|
|
2016-12-07 00:46:13 +00:00
|
|
|
static bool is_accessed_spte(u64 spte)
|
2010-08-02 08:15:08 +00:00
|
|
|
{
|
2017-07-01 00:26:31 +00:00
|
|
|
u64 accessed_mask = spte_shadow_accessed_mask(spte);
|
|
|
|
|
|
|
|
return accessed_mask ? spte & accessed_mask
|
|
|
|
: !is_access_track_spte(spte);
|
2010-08-02 08:15:08 +00:00
|
|
|
}
|
|
|
|
|
2016-12-07 00:46:13 +00:00
|
|
|
static bool is_dirty_spte(u64 spte)
|
2015-01-09 08:44:30 +00:00
|
|
|
{
|
2017-07-01 00:26:31 +00:00
|
|
|
u64 dirty_mask = spte_shadow_dirty_mask(spte);
|
|
|
|
|
|
|
|
return dirty_mask ? spte & dirty_mask : spte & PT_WRITABLE_MASK;
|
2015-01-09 08:44:30 +00:00
|
|
|
}
|
|
|
|
|
2011-07-11 19:30:35 +00:00
|
|
|
/* Rules for using mmu_spte_set:
|
|
|
|
* Set the sptep from nonpresent to present.
|
|
|
|
* Note: the sptep being assigned *must* be either not present
|
|
|
|
* or in a state where the hardware will not attempt to update
|
|
|
|
* the spte.
|
|
|
|
*/
|
|
|
|
static void mmu_spte_set(u64 *sptep, u64 new_spte)
|
|
|
|
{
|
|
|
|
WARN_ON(is_shadow_present_pte(*sptep));
|
|
|
|
__set_spte(sptep, new_spte);
|
|
|
|
}
|
|
|
|
|
2016-12-07 00:46:14 +00:00
|
|
|
/*
|
|
|
|
* Update the SPTE (excluding the PFN), but do not track changes in its
|
|
|
|
* accessed/dirty status.
|
2011-07-11 19:30:35 +00:00
|
|
|
*/
|
2016-12-07 00:46:14 +00:00
|
|
|
static u64 mmu_spte_update_no_track(u64 *sptep, u64 new_spte)
|
2010-06-06 12:46:44 +00:00
|
|
|
{
|
2012-06-20 07:59:18 +00:00
|
|
|
u64 old_spte = *sptep;
|
2010-08-02 08:15:08 +00:00
|
|
|
|
2015-11-20 08:44:55 +00:00
|
|
|
WARN_ON(!is_shadow_present_pte(new_spte));
|
2010-06-06 12:46:44 +00:00
|
|
|
|
2012-06-20 07:58:33 +00:00
|
|
|
if (!is_shadow_present_pte(old_spte)) {
|
|
|
|
mmu_spte_set(sptep, new_spte);
|
2016-12-07 00:46:14 +00:00
|
|
|
return old_spte;
|
2012-06-20 07:58:33 +00:00
|
|
|
}
|
2010-08-02 08:15:08 +00:00
|
|
|
|
2012-06-20 07:59:18 +00:00
|
|
|
if (!spte_has_volatile_bits(old_spte))
|
2011-07-11 19:31:28 +00:00
|
|
|
__update_clear_spte_fast(sptep, new_spte);
|
2010-08-02 08:15:08 +00:00
|
|
|
else
|
2011-07-11 19:31:28 +00:00
|
|
|
old_spte = __update_clear_spte_slow(sptep, new_spte);
|
2010-08-02 08:15:08 +00:00
|
|
|
|
2016-12-07 00:46:13 +00:00
|
|
|
WARN_ON(spte_to_pfn(old_spte) != spte_to_pfn(new_spte));
|
|
|
|
|
2016-12-07 00:46:14 +00:00
|
|
|
return old_spte;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Rules for using mmu_spte_update:
|
|
|
|
* Update the state bits, it means the mapped pfn is not changed.
|
|
|
|
*
|
|
|
|
* Whenever we overwrite a writable spte with a read-only one we
|
|
|
|
* should flush remote TLBs. Otherwise rmap_write_protect
|
|
|
|
* will find a read-only spte, even though the writable spte
|
|
|
|
* might be cached on a CPU's TLB, the return value indicates this
|
|
|
|
* case.
|
|
|
|
*
|
|
|
|
* Returns true if the TLB needs to be flushed
|
|
|
|
*/
|
|
|
|
static bool mmu_spte_update(u64 *sptep, u64 new_spte)
|
|
|
|
{
|
|
|
|
bool flush = false;
|
|
|
|
u64 old_spte = mmu_spte_update_no_track(sptep, new_spte);
|
|
|
|
|
|
|
|
if (!is_shadow_present_pte(old_spte))
|
|
|
|
return false;
|
|
|
|
|
2012-06-20 07:59:18 +00:00
|
|
|
/*
|
|
|
|
* For the spte updated out of mmu-lock is safe, since
|
2016-02-23 23:34:30 +00:00
|
|
|
* we always atomically update it, see the comments in
|
2012-06-20 07:59:18 +00:00
|
|
|
* spte_has_volatile_bits().
|
|
|
|
*/
|
2016-12-07 00:46:11 +00:00
|
|
|
if (spte_can_locklessly_be_made_writable(old_spte) &&
|
2014-04-17 09:06:15 +00:00
|
|
|
!is_writable_pte(new_spte))
|
2016-12-07 00:46:13 +00:00
|
|
|
flush = true;
|
2010-08-02 08:15:08 +00:00
|
|
|
|
2015-01-09 08:44:30 +00:00
|
|
|
/*
|
2016-12-07 00:46:13 +00:00
|
|
|
* Flush TLB when accessed/dirty states are changed in the page tables,
|
2015-01-09 08:44:30 +00:00
|
|
|
* to guarantee consistency between TLB and page tables.
|
|
|
|
*/
|
|
|
|
|
2016-12-07 00:46:13 +00:00
|
|
|
if (is_accessed_spte(old_spte) && !is_accessed_spte(new_spte)) {
|
|
|
|
flush = true;
|
2010-08-02 08:15:08 +00:00
|
|
|
kvm_set_pfn_accessed(spte_to_pfn(old_spte));
|
2016-12-07 00:46:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
if (is_dirty_spte(old_spte) && !is_dirty_spte(new_spte)) {
|
|
|
|
flush = true;
|
2010-08-02 08:15:08 +00:00
|
|
|
kvm_set_pfn_dirty(spte_to_pfn(old_spte));
|
2016-12-07 00:46:13 +00:00
|
|
|
}
|
2012-06-20 07:58:33 +00:00
|
|
|
|
2016-12-07 00:46:13 +00:00
|
|
|
return flush;
|
2010-06-06 12:46:44 +00:00
|
|
|
}
|
|
|
|
|
2011-07-11 19:30:35 +00:00
|
|
|
/*
|
|
|
|
* Rules for using mmu_spte_clear_track_bits:
|
|
|
|
* It sets the sptep from present to nonpresent, and track the
|
|
|
|
* state bits, it is used to clear the last level sptep.
|
2016-12-07 00:46:13 +00:00
|
|
|
* Returns non-zero if the PTE was previously valid.
|
2011-07-11 19:30:35 +00:00
|
|
|
*/
|
|
|
|
static int mmu_spte_clear_track_bits(u64 *sptep)
|
|
|
|
{
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
kvm_pfn_t pfn;
|
2011-07-11 19:30:35 +00:00
|
|
|
u64 old_spte = *sptep;
|
|
|
|
|
|
|
|
if (!spte_has_volatile_bits(old_spte))
|
2011-07-11 19:31:28 +00:00
|
|
|
__update_clear_spte_fast(sptep, 0ull);
|
2011-07-11 19:30:35 +00:00
|
|
|
else
|
2011-07-11 19:31:28 +00:00
|
|
|
old_spte = __update_clear_spte_slow(sptep, 0ull);
|
2011-07-11 19:30:35 +00:00
|
|
|
|
2015-11-20 08:44:55 +00:00
|
|
|
if (!is_shadow_present_pte(old_spte))
|
2011-07-11 19:30:35 +00:00
|
|
|
return 0;
|
|
|
|
|
|
|
|
pfn = spte_to_pfn(old_spte);
|
2012-07-17 13:52:52 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* KVM does not hold the refcount of the page used by
|
|
|
|
* kvm mmu, before reclaiming the page, we should
|
|
|
|
* unmap it from mmu first.
|
|
|
|
*/
|
2014-11-10 08:33:56 +00:00
|
|
|
WARN_ON(!kvm_is_reserved_pfn(pfn) && !page_count(pfn_to_page(pfn)));
|
2012-07-17 13:52:52 +00:00
|
|
|
|
2016-12-07 00:46:13 +00:00
|
|
|
if (is_accessed_spte(old_spte))
|
2011-07-11 19:30:35 +00:00
|
|
|
kvm_set_pfn_accessed(pfn);
|
2016-12-07 00:46:13 +00:00
|
|
|
|
|
|
|
if (is_dirty_spte(old_spte))
|
2011-07-11 19:30:35 +00:00
|
|
|
kvm_set_pfn_dirty(pfn);
|
2016-12-07 00:46:13 +00:00
|
|
|
|
2011-07-11 19:30:35 +00:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Rules for using mmu_spte_clear_no_track:
|
|
|
|
* Directly clear spte without caring the state bits of sptep,
|
|
|
|
* it is used to set the upper level spte.
|
|
|
|
*/
|
|
|
|
static void mmu_spte_clear_no_track(u64 *sptep)
|
|
|
|
{
|
2011-07-11 19:31:28 +00:00
|
|
|
__update_clear_spte_fast(sptep, 0ull);
|
2011-07-11 19:30:35 +00:00
|
|
|
}
|
|
|
|
|
2011-07-11 19:32:13 +00:00
|
|
|
static u64 mmu_spte_get_lockless(u64 *sptep)
|
|
|
|
{
|
|
|
|
return __get_spte_lockless(sptep);
|
|
|
|
}
|
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
static u64 mark_spte_for_access_track(u64 spte)
|
|
|
|
{
|
2017-07-01 00:26:31 +00:00
|
|
|
if (spte_ad_enabled(spte))
|
2016-12-07 00:46:16 +00:00
|
|
|
return spte & ~shadow_accessed_mask;
|
|
|
|
|
2017-07-01 00:26:31 +00:00
|
|
|
if (is_access_track_spte(spte))
|
2016-12-07 00:46:16 +00:00
|
|
|
return spte;
|
|
|
|
|
|
|
|
/*
|
2016-12-22 04:29:31 +00:00
|
|
|
* Making an Access Tracking PTE will result in removal of write access
|
|
|
|
* from the PTE. So, verify that we will be able to restore the write
|
|
|
|
* access in the fast page fault path later on.
|
2016-12-07 00:46:16 +00:00
|
|
|
*/
|
|
|
|
WARN_ONCE((spte & PT_WRITABLE_MASK) &&
|
|
|
|
!spte_can_locklessly_be_made_writable(spte),
|
|
|
|
"kvm: Writable SPTE is not locklessly dirty-trackable\n");
|
|
|
|
|
|
|
|
WARN_ONCE(spte & (shadow_acc_track_saved_bits_mask <<
|
|
|
|
shadow_acc_track_saved_bits_shift),
|
|
|
|
"kvm: Access Tracking saved bit locations are not zero\n");
|
|
|
|
|
|
|
|
spte |= (spte & shadow_acc_track_saved_bits_mask) <<
|
|
|
|
shadow_acc_track_saved_bits_shift;
|
|
|
|
spte &= ~shadow_acc_track_mask;
|
|
|
|
|
|
|
|
return spte;
|
|
|
|
}
|
|
|
|
|
2016-12-22 04:29:32 +00:00
|
|
|
/* Restore an acc-track PTE back to a regular PTE */
|
|
|
|
static u64 restore_acc_track_spte(u64 spte)
|
|
|
|
{
|
|
|
|
u64 new_spte = spte;
|
|
|
|
u64 saved_bits = (spte >> shadow_acc_track_saved_bits_shift)
|
|
|
|
& shadow_acc_track_saved_bits_mask;
|
|
|
|
|
2017-07-01 00:26:31 +00:00
|
|
|
WARN_ON_ONCE(spte_ad_enabled(spte));
|
2016-12-22 04:29:32 +00:00
|
|
|
WARN_ON_ONCE(!is_access_track_spte(spte));
|
|
|
|
|
|
|
|
new_spte &= ~shadow_acc_track_mask;
|
|
|
|
new_spte &= ~(shadow_acc_track_saved_bits_mask <<
|
|
|
|
shadow_acc_track_saved_bits_shift);
|
|
|
|
new_spte |= saved_bits;
|
|
|
|
|
|
|
|
return new_spte;
|
|
|
|
}
|
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
/* Returns the Accessed status of the PTE and resets it at the same time. */
|
|
|
|
static bool mmu_spte_age(u64 *sptep)
|
|
|
|
{
|
|
|
|
u64 spte = mmu_spte_get_lockless(sptep);
|
|
|
|
|
|
|
|
if (!is_accessed_spte(spte))
|
|
|
|
return false;
|
|
|
|
|
2017-07-01 00:26:31 +00:00
|
|
|
if (spte_ad_enabled(spte)) {
|
2016-12-07 00:46:16 +00:00
|
|
|
clear_bit((ffs(shadow_accessed_mask) - 1),
|
|
|
|
(unsigned long *)sptep);
|
|
|
|
} else {
|
|
|
|
/*
|
|
|
|
* Capture the dirty status of the page, so that it doesn't get
|
|
|
|
* lost when the SPTE is marked for access tracking.
|
|
|
|
*/
|
|
|
|
if (is_writable_pte(spte))
|
|
|
|
kvm_set_pfn_dirty(spte_to_pfn(spte));
|
|
|
|
|
|
|
|
spte = mark_spte_for_access_track(spte);
|
|
|
|
mmu_spte_update_no_track(sptep, spte);
|
|
|
|
}
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2011-07-11 19:32:13 +00:00
|
|
|
static void walk_shadow_page_lockless_begin(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2012-05-14 12:44:06 +00:00
|
|
|
/*
|
|
|
|
* Prevent page table teardown by making any free-er wait during
|
|
|
|
* kvm_flush_remote_tlbs() IPI to all active vcpus.
|
|
|
|
*/
|
|
|
|
local_irq_disable();
|
2016-03-13 03:10:25 +00:00
|
|
|
|
2012-05-14 12:44:06 +00:00
|
|
|
/*
|
|
|
|
* Make sure a following spte read is not reordered ahead of the write
|
|
|
|
* to vcpu->mode.
|
|
|
|
*/
|
2016-03-13 03:10:25 +00:00
|
|
|
smp_store_mb(vcpu->mode, READING_SHADOW_PAGE_TABLES);
|
2011-07-11 19:32:13 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void walk_shadow_page_lockless_end(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2012-05-14 12:44:06 +00:00
|
|
|
/*
|
|
|
|
* Make sure the write to vcpu->mode is not reordered in front of
|
2018-09-07 05:45:02 +00:00
|
|
|
* reads to sptes. If it does, kvm_mmu_commit_zap_page() can see us
|
2012-05-14 12:44:06 +00:00
|
|
|
* OUTSIDE_GUEST_MODE and proceed to free the shadow page table.
|
|
|
|
*/
|
2016-03-13 03:10:25 +00:00
|
|
|
smp_store_release(&vcpu->mode, OUTSIDE_GUEST_MODE);
|
2012-05-14 12:44:06 +00:00
|
|
|
local_irq_enable();
|
2011-07-11 19:32:13 +00:00
|
|
|
}
|
|
|
|
|
2007-01-06 00:36:54 +00:00
|
|
|
static int mmu_topup_memory_cache(struct kvm_mmu_memory_cache *cache,
|
2007-09-10 08:28:17 +00:00
|
|
|
struct kmem_cache *base_cache, int min)
|
2007-01-06 00:36:53 +00:00
|
|
|
{
|
|
|
|
void *obj;
|
|
|
|
|
|
|
|
if (cache->nobjs >= min)
|
2007-01-06 00:36:54 +00:00
|
|
|
return 0;
|
2007-01-06 00:36:53 +00:00
|
|
|
while (cache->nobjs < ARRAY_SIZE(cache->objects)) {
|
2019-02-11 19:02:50 +00:00
|
|
|
obj = kmem_cache_zalloc(base_cache, GFP_KERNEL_ACCOUNT);
|
2007-01-06 00:36:53 +00:00
|
|
|
if (!obj)
|
2018-09-04 15:57:32 +00:00
|
|
|
return cache->nobjs >= min ? 0 : -ENOMEM;
|
2007-01-06 00:36:53 +00:00
|
|
|
cache->objects[cache->nobjs++] = obj;
|
|
|
|
}
|
2007-01-06 00:36:54 +00:00
|
|
|
return 0;
|
2007-01-06 00:36:53 +00:00
|
|
|
}
|
|
|
|
|
2011-09-22 08:53:17 +00:00
|
|
|
static int mmu_memory_cache_free_objects(struct kvm_mmu_memory_cache *cache)
|
|
|
|
{
|
|
|
|
return cache->nobjs;
|
|
|
|
}
|
|
|
|
|
2010-05-13 02:06:02 +00:00
|
|
|
static void mmu_free_memory_cache(struct kvm_mmu_memory_cache *mc,
|
|
|
|
struct kmem_cache *cache)
|
2007-01-06 00:36:53 +00:00
|
|
|
{
|
|
|
|
while (mc->nobjs)
|
2010-05-13 02:06:02 +00:00
|
|
|
kmem_cache_free(cache, mc->objects[--mc->nobjs]);
|
2007-01-06 00:36:53 +00:00
|
|
|
}
|
|
|
|
|
2007-07-20 05:18:27 +00:00
|
|
|
static int mmu_topup_memory_cache_page(struct kvm_mmu_memory_cache *cache,
|
2007-09-10 08:28:17 +00:00
|
|
|
int min)
|
2007-07-20 05:18:27 +00:00
|
|
|
{
|
2011-03-04 11:01:10 +00:00
|
|
|
void *page;
|
2007-07-20 05:18:27 +00:00
|
|
|
|
|
|
|
if (cache->nobjs >= min)
|
|
|
|
return 0;
|
|
|
|
while (cache->nobjs < ARRAY_SIZE(cache->objects)) {
|
2018-07-26 23:37:45 +00:00
|
|
|
page = (void *)__get_free_page(GFP_KERNEL_ACCOUNT);
|
2007-07-20 05:18:27 +00:00
|
|
|
if (!page)
|
2018-09-04 15:57:32 +00:00
|
|
|
return cache->nobjs >= min ? 0 : -ENOMEM;
|
2011-03-04 11:01:10 +00:00
|
|
|
cache->objects[cache->nobjs++] = page;
|
2007-07-20 05:18:27 +00:00
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void mmu_free_memory_cache_page(struct kvm_mmu_memory_cache *mc)
|
|
|
|
{
|
|
|
|
while (mc->nobjs)
|
2007-07-21 06:06:46 +00:00
|
|
|
free_page((unsigned long)mc->objects[--mc->nobjs]);
|
2007-07-20 05:18:27 +00:00
|
|
|
}
|
|
|
|
|
2007-09-10 08:28:17 +00:00
|
|
|
static int mmu_topup_memory_caches(struct kvm_vcpu *vcpu)
|
2007-01-06 00:36:53 +00:00
|
|
|
{
|
2007-01-06 00:36:54 +00:00
|
|
|
int r;
|
|
|
|
|
2011-05-15 15:26:20 +00:00
|
|
|
r = mmu_topup_memory_cache(&vcpu->arch.mmu_pte_list_desc_cache,
|
2011-05-15 15:27:08 +00:00
|
|
|
pte_list_desc_cache, 8 + PTE_PREFETCH_NUM);
|
2007-05-30 09:34:53 +00:00
|
|
|
if (r)
|
|
|
|
goto out;
|
2007-12-13 15:50:52 +00:00
|
|
|
r = mmu_topup_memory_cache_page(&vcpu->arch.mmu_page_cache, 8);
|
2007-05-30 09:34:53 +00:00
|
|
|
if (r)
|
|
|
|
goto out;
|
2007-12-13 15:50:52 +00:00
|
|
|
r = mmu_topup_memory_cache(&vcpu->arch.mmu_page_header_cache,
|
2007-09-10 08:28:17 +00:00
|
|
|
mmu_page_header_cache, 4);
|
2007-01-06 00:36:54 +00:00
|
|
|
out:
|
|
|
|
return r;
|
2007-01-06 00:36:53 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void mmu_free_memory_caches(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2011-05-15 15:26:20 +00:00
|
|
|
mmu_free_memory_cache(&vcpu->arch.mmu_pte_list_desc_cache,
|
|
|
|
pte_list_desc_cache);
|
2007-12-13 15:50:52 +00:00
|
|
|
mmu_free_memory_cache_page(&vcpu->arch.mmu_page_cache);
|
2010-05-13 02:06:02 +00:00
|
|
|
mmu_free_memory_cache(&vcpu->arch.mmu_page_header_cache,
|
|
|
|
mmu_page_header_cache);
|
2007-01-06 00:36:53 +00:00
|
|
|
}
|
|
|
|
|
2012-05-29 14:54:26 +00:00
|
|
|
static void *mmu_memory_cache_alloc(struct kvm_mmu_memory_cache *mc)
|
2007-01-06 00:36:53 +00:00
|
|
|
{
|
|
|
|
void *p;
|
|
|
|
|
|
|
|
BUG_ON(!mc->nobjs);
|
|
|
|
p = mc->objects[--mc->nobjs];
|
|
|
|
return p;
|
|
|
|
}
|
|
|
|
|
2011-05-15 15:26:20 +00:00
|
|
|
static struct pte_list_desc *mmu_alloc_pte_list_desc(struct kvm_vcpu *vcpu)
|
2007-01-06 00:36:53 +00:00
|
|
|
{
|
2012-05-29 14:54:26 +00:00
|
|
|
return mmu_memory_cache_alloc(&vcpu->arch.mmu_pte_list_desc_cache);
|
2007-01-06 00:36:53 +00:00
|
|
|
}
|
|
|
|
|
2011-05-15 15:26:20 +00:00
|
|
|
static void mmu_free_pte_list_desc(struct pte_list_desc *pte_list_desc)
|
2007-01-06 00:36:53 +00:00
|
|
|
{
|
2011-05-15 15:26:20 +00:00
|
|
|
kmem_cache_free(pte_list_desc_cache, pte_list_desc);
|
2007-01-06 00:36:53 +00:00
|
|
|
}
|
|
|
|
|
2010-05-26 08:49:59 +00:00
|
|
|
static gfn_t kvm_mmu_page_get_gfn(struct kvm_mmu_page *sp, int index)
|
|
|
|
{
|
|
|
|
if (!sp->role.direct)
|
|
|
|
return sp->gfns[index];
|
|
|
|
|
|
|
|
return sp->gfn + (index << ((sp->role.level - 1) * PT64_LEVEL_BITS));
|
|
|
|
}
|
|
|
|
|
|
|
|
static void kvm_mmu_page_set_gfn(struct kvm_mmu_page *sp, int index, gfn_t gfn)
|
|
|
|
{
|
|
|
|
if (sp->role.direct)
|
|
|
|
BUG_ON(gfn != kvm_mmu_page_get_gfn(sp, index));
|
|
|
|
else
|
|
|
|
sp->gfns[index] = gfn;
|
|
|
|
}
|
|
|
|
|
2008-02-23 14:44:30 +00:00
|
|
|
/*
|
2010-12-07 03:59:07 +00:00
|
|
|
* Return the pointer to the large page information for a given gfn,
|
|
|
|
* handling slots that are not large page aligned.
|
2008-02-23 14:44:30 +00:00
|
|
|
*/
|
2010-12-07 03:59:07 +00:00
|
|
|
static struct kvm_lpage_info *lpage_info_slot(gfn_t gfn,
|
|
|
|
struct kvm_memory_slot *slot,
|
|
|
|
int level)
|
2008-02-23 14:44:30 +00:00
|
|
|
{
|
|
|
|
unsigned long idx;
|
|
|
|
|
2012-02-08 03:59:10 +00:00
|
|
|
idx = gfn_to_index(gfn, slot->base_gfn, level);
|
2012-02-08 04:02:18 +00:00
|
|
|
return &slot->arch.lpage_info[level - 2][idx];
|
2008-02-23 14:44:30 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:07 +00:00
|
|
|
static void update_gfn_disallow_lpage_count(struct kvm_memory_slot *slot,
|
|
|
|
gfn_t gfn, int count)
|
|
|
|
{
|
|
|
|
struct kvm_lpage_info *linfo;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = PT_DIRECTORY_LEVEL; i <= PT_MAX_HUGEPAGE_LEVEL; ++i) {
|
|
|
|
linfo = lpage_info_slot(gfn, slot, i);
|
|
|
|
linfo->disallow_lpage += count;
|
|
|
|
WARN_ON(linfo->disallow_lpage < 0);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void kvm_mmu_gfn_disallow_lpage(struct kvm_memory_slot *slot, gfn_t gfn)
|
|
|
|
{
|
|
|
|
update_gfn_disallow_lpage_count(slot, gfn, 1);
|
|
|
|
}
|
|
|
|
|
|
|
|
void kvm_mmu_gfn_allow_lpage(struct kvm_memory_slot *slot, gfn_t gfn)
|
|
|
|
{
|
|
|
|
update_gfn_disallow_lpage_count(slot, gfn, -1);
|
|
|
|
}
|
|
|
|
|
2015-05-19 14:29:22 +00:00
|
|
|
static void account_shadowed(struct kvm *kvm, struct kvm_mmu_page *sp)
|
2008-02-23 14:44:30 +00:00
|
|
|
{
|
2015-05-18 13:03:39 +00:00
|
|
|
struct kvm_memslots *slots;
|
2009-07-27 14:30:43 +00:00
|
|
|
struct kvm_memory_slot *slot;
|
2015-05-19 14:29:22 +00:00
|
|
|
gfn_t gfn;
|
2008-02-23 14:44:30 +00:00
|
|
|
|
2016-02-24 09:51:14 +00:00
|
|
|
kvm->arch.indirect_shadow_pages++;
|
2015-05-19 14:29:22 +00:00
|
|
|
gfn = sp->gfn;
|
2015-05-18 13:03:39 +00:00
|
|
|
slots = kvm_memslots_for_spte_role(kvm, sp->role);
|
|
|
|
slot = __gfn_to_memslot(slots, gfn);
|
2016-02-24 09:51:14 +00:00
|
|
|
|
|
|
|
/* the non-leaf shadow pages are keeping readonly. */
|
|
|
|
if (sp->role.level > PT_PAGE_TABLE_LEVEL)
|
|
|
|
return kvm_slot_page_track_add_page(kvm, slot, gfn,
|
|
|
|
KVM_PAGE_TRACK_WRITE);
|
|
|
|
|
2016-02-24 09:51:07 +00:00
|
|
|
kvm_mmu_gfn_disallow_lpage(slot, gfn);
|
2008-02-23 14:44:30 +00:00
|
|
|
}
|
|
|
|
|
2015-05-19 14:29:22 +00:00
|
|
|
static void unaccount_shadowed(struct kvm *kvm, struct kvm_mmu_page *sp)
|
2008-02-23 14:44:30 +00:00
|
|
|
{
|
2015-05-18 13:03:39 +00:00
|
|
|
struct kvm_memslots *slots;
|
2009-07-27 14:30:43 +00:00
|
|
|
struct kvm_memory_slot *slot;
|
2015-05-19 14:29:22 +00:00
|
|
|
gfn_t gfn;
|
2008-02-23 14:44:30 +00:00
|
|
|
|
2016-02-24 09:51:14 +00:00
|
|
|
kvm->arch.indirect_shadow_pages--;
|
2015-05-19 14:29:22 +00:00
|
|
|
gfn = sp->gfn;
|
2015-05-18 13:03:39 +00:00
|
|
|
slots = kvm_memslots_for_spte_role(kvm, sp->role);
|
|
|
|
slot = __gfn_to_memslot(slots, gfn);
|
2016-02-24 09:51:14 +00:00
|
|
|
if (sp->role.level > PT_PAGE_TABLE_LEVEL)
|
|
|
|
return kvm_slot_page_track_remove_page(kvm, slot, gfn,
|
|
|
|
KVM_PAGE_TRACK_WRITE);
|
|
|
|
|
2016-02-24 09:51:07 +00:00
|
|
|
kvm_mmu_gfn_allow_lpage(slot, gfn);
|
2008-02-23 14:44:30 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:06 +00:00
|
|
|
static bool __mmu_gfn_lpage_is_disallowed(gfn_t gfn, int level,
|
|
|
|
struct kvm_memory_slot *slot)
|
2008-02-23 14:44:30 +00:00
|
|
|
{
|
2010-12-07 03:59:07 +00:00
|
|
|
struct kvm_lpage_info *linfo;
|
2008-02-23 14:44:30 +00:00
|
|
|
|
|
|
|
if (slot) {
|
2010-12-07 03:59:07 +00:00
|
|
|
linfo = lpage_info_slot(gfn, slot, level);
|
2016-02-24 09:51:06 +00:00
|
|
|
return !!linfo->disallow_lpage;
|
2008-02-23 14:44:30 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:06 +00:00
|
|
|
return true;
|
2008-02-23 14:44:30 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:06 +00:00
|
|
|
static bool mmu_gfn_lpage_is_disallowed(struct kvm_vcpu *vcpu, gfn_t gfn,
|
|
|
|
int level)
|
2015-10-16 08:08:03 +00:00
|
|
|
{
|
|
|
|
struct kvm_memory_slot *slot;
|
|
|
|
|
|
|
|
slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);
|
2016-02-24 09:51:06 +00:00
|
|
|
return __mmu_gfn_lpage_is_disallowed(gfn, level, slot);
|
2015-10-16 08:08:03 +00:00
|
|
|
}
|
|
|
|
|
2009-07-27 14:30:43 +00:00
|
|
|
static int host_mapping_level(struct kvm *kvm, gfn_t gfn)
|
2008-02-23 14:44:30 +00:00
|
|
|
{
|
2010-01-28 11:37:56 +00:00
|
|
|
unsigned long page_size;
|
2009-07-27 14:30:43 +00:00
|
|
|
int i, ret = 0;
|
2008-02-23 14:44:30 +00:00
|
|
|
|
2010-01-28 11:37:56 +00:00
|
|
|
page_size = kvm_host_page_size(kvm, gfn);
|
2008-02-23 14:44:30 +00:00
|
|
|
|
2015-05-13 06:42:21 +00:00
|
|
|
for (i = PT_PAGE_TABLE_LEVEL; i <= PT_MAX_HUGEPAGE_LEVEL; ++i) {
|
2009-07-27 14:30:43 +00:00
|
|
|
if (page_size >= KVM_HPAGE_SIZE(i))
|
|
|
|
ret = i;
|
|
|
|
else
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2008-09-16 23:54:47 +00:00
|
|
|
return ret;
|
2008-02-23 14:44:30 +00:00
|
|
|
}
|
|
|
|
|
2015-10-16 08:07:01 +00:00
|
|
|
static inline bool memslot_valid_for_gpte(struct kvm_memory_slot *slot,
|
|
|
|
bool no_dirty_log)
|
|
|
|
{
|
|
|
|
if (!slot || slot->flags & KVM_MEMSLOT_INVALID)
|
|
|
|
return false;
|
|
|
|
if (no_dirty_log && slot->dirty_bitmap)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2011-03-09 07:43:00 +00:00
|
|
|
static struct kvm_memory_slot *
|
|
|
|
gfn_to_memslot_dirty_bitmap(struct kvm_vcpu *vcpu, gfn_t gfn,
|
|
|
|
bool no_dirty_log)
|
2008-02-23 14:44:30 +00:00
|
|
|
{
|
|
|
|
struct kvm_memory_slot *slot;
|
2011-03-09 07:43:00 +00:00
|
|
|
|
2015-04-08 13:39:23 +00:00
|
|
|
slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);
|
2015-10-16 08:07:01 +00:00
|
|
|
if (!memslot_valid_for_gpte(slot, no_dirty_log))
|
2011-03-09 07:43:00 +00:00
|
|
|
slot = NULL;
|
|
|
|
|
|
|
|
return slot;
|
|
|
|
}
|
|
|
|
|
2015-10-16 08:06:02 +00:00
|
|
|
static int mapping_level(struct kvm_vcpu *vcpu, gfn_t large_gfn,
|
|
|
|
bool *force_pt_level)
|
2011-01-13 23:46:48 +00:00
|
|
|
{
|
|
|
|
int host_level, level, max_level;
|
2015-10-16 08:07:01 +00:00
|
|
|
struct kvm_memory_slot *slot;
|
|
|
|
|
2015-10-19 06:13:29 +00:00
|
|
|
if (unlikely(*force_pt_level))
|
|
|
|
return PT_PAGE_TABLE_LEVEL;
|
2008-02-23 14:44:30 +00:00
|
|
|
|
2015-10-19 06:13:29 +00:00
|
|
|
slot = kvm_vcpu_gfn_to_memslot(vcpu, large_gfn);
|
|
|
|
*force_pt_level = !memslot_valid_for_gpte(slot, true);
|
2015-10-16 08:06:02 +00:00
|
|
|
if (unlikely(*force_pt_level))
|
|
|
|
return PT_PAGE_TABLE_LEVEL;
|
|
|
|
|
2009-07-27 14:30:43 +00:00
|
|
|
host_level = host_mapping_level(vcpu->kvm, large_gfn);
|
|
|
|
|
|
|
|
if (host_level == PT_PAGE_TABLE_LEVEL)
|
|
|
|
return host_level;
|
|
|
|
|
2013-02-05 07:26:54 +00:00
|
|
|
max_level = min(kvm_x86_ops->get_lpage_level(), host_level);
|
2010-01-05 11:02:29 +00:00
|
|
|
|
|
|
|
for (level = PT_DIRECTORY_LEVEL; level <= max_level; ++level)
|
2016-02-24 09:51:06 +00:00
|
|
|
if (__mmu_gfn_lpage_is_disallowed(large_gfn, level, slot))
|
2009-07-27 14:30:43 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
return level - 1;
|
2008-02-23 14:44:30 +00:00
|
|
|
}
|
|
|
|
|
2007-09-27 12:11:22 +00:00
|
|
|
/*
|
2015-11-20 08:41:28 +00:00
|
|
|
* About rmap_head encoding:
|
2007-01-06 00:36:38 +00:00
|
|
|
*
|
2015-11-20 08:41:28 +00:00
|
|
|
* If the bit zero of rmap_head->val is clear, then it points to the only spte
|
|
|
|
* in this rmap chain. Otherwise, (rmap_head->val & ~1) points to a struct
|
2011-05-15 15:26:20 +00:00
|
|
|
* pte_list_desc containing more mappings.
|
2015-11-20 08:41:28 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Returns the number of pointers in the rmap chain, not counting the new one.
|
2007-01-06 00:36:38 +00:00
|
|
|
*/
|
2011-05-15 15:26:20 +00:00
|
|
|
static int pte_list_add(struct kvm_vcpu *vcpu, u64 *spte,
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *rmap_head)
|
2007-01-06 00:36:38 +00:00
|
|
|
{
|
2011-05-15 15:26:20 +00:00
|
|
|
struct pte_list_desc *desc;
|
2009-08-05 18:43:58 +00:00
|
|
|
int i, count = 0;
|
2007-01-06 00:36:38 +00:00
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
if (!rmap_head->val) {
|
2011-05-15 15:26:20 +00:00
|
|
|
rmap_printk("pte_list_add: %p %llx 0->1\n", spte, *spte);
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head->val = (unsigned long)spte;
|
|
|
|
} else if (!(rmap_head->val & 1)) {
|
2011-05-15 15:26:20 +00:00
|
|
|
rmap_printk("pte_list_add: %p %llx 1->many\n", spte, *spte);
|
|
|
|
desc = mmu_alloc_pte_list_desc(vcpu);
|
2015-11-20 08:41:28 +00:00
|
|
|
desc->sptes[0] = (u64 *)rmap_head->val;
|
2009-06-10 11:24:23 +00:00
|
|
|
desc->sptes[1] = spte;
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head->val = (unsigned long)desc | 1;
|
2010-09-18 00:41:02 +00:00
|
|
|
++count;
|
2007-01-06 00:36:38 +00:00
|
|
|
} else {
|
2011-05-15 15:26:20 +00:00
|
|
|
rmap_printk("pte_list_add: %p %llx many->many\n", spte, *spte);
|
2015-11-20 08:41:28 +00:00
|
|
|
desc = (struct pte_list_desc *)(rmap_head->val & ~1ul);
|
2011-05-15 15:26:20 +00:00
|
|
|
while (desc->sptes[PTE_LIST_EXT-1] && desc->more) {
|
2007-01-06 00:36:38 +00:00
|
|
|
desc = desc->more;
|
2011-05-15 15:26:20 +00:00
|
|
|
count += PTE_LIST_EXT;
|
2009-08-05 18:43:58 +00:00
|
|
|
}
|
2011-05-15 15:26:20 +00:00
|
|
|
if (desc->sptes[PTE_LIST_EXT-1]) {
|
|
|
|
desc->more = mmu_alloc_pte_list_desc(vcpu);
|
2007-01-06 00:36:38 +00:00
|
|
|
desc = desc->more;
|
|
|
|
}
|
2009-06-10 11:24:23 +00:00
|
|
|
for (i = 0; desc->sptes[i]; ++i)
|
2010-09-18 00:41:02 +00:00
|
|
|
++count;
|
2009-06-10 11:24:23 +00:00
|
|
|
desc->sptes[i] = spte;
|
2007-01-06 00:36:38 +00:00
|
|
|
}
|
2009-08-05 18:43:58 +00:00
|
|
|
return count;
|
2007-01-06 00:36:38 +00:00
|
|
|
}
|
|
|
|
|
2011-05-15 15:26:20 +00:00
|
|
|
static void
|
2015-11-20 08:41:28 +00:00
|
|
|
pte_list_desc_remove_entry(struct kvm_rmap_head *rmap_head,
|
|
|
|
struct pte_list_desc *desc, int i,
|
|
|
|
struct pte_list_desc *prev_desc)
|
2007-01-06 00:36:38 +00:00
|
|
|
{
|
|
|
|
int j;
|
|
|
|
|
2011-05-15 15:26:20 +00:00
|
|
|
for (j = PTE_LIST_EXT - 1; !desc->sptes[j] && j > i; --j)
|
2007-01-06 00:36:38 +00:00
|
|
|
;
|
2009-06-10 11:24:23 +00:00
|
|
|
desc->sptes[i] = desc->sptes[j];
|
|
|
|
desc->sptes[j] = NULL;
|
2007-01-06 00:36:38 +00:00
|
|
|
if (j != 0)
|
|
|
|
return;
|
|
|
|
if (!prev_desc && !desc->more)
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head->val = (unsigned long)desc->sptes[0];
|
2007-01-06 00:36:38 +00:00
|
|
|
else
|
|
|
|
if (prev_desc)
|
|
|
|
prev_desc->more = desc->more;
|
|
|
|
else
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head->val = (unsigned long)desc->more | 1;
|
2011-05-15 15:26:20 +00:00
|
|
|
mmu_free_pte_list_desc(desc);
|
2007-01-06 00:36:38 +00:00
|
|
|
}
|
|
|
|
|
2018-10-04 02:04:22 +00:00
|
|
|
static void __pte_list_remove(u64 *spte, struct kvm_rmap_head *rmap_head)
|
2007-01-06 00:36:38 +00:00
|
|
|
{
|
2011-05-15 15:26:20 +00:00
|
|
|
struct pte_list_desc *desc;
|
|
|
|
struct pte_list_desc *prev_desc;
|
2007-01-06 00:36:38 +00:00
|
|
|
int i;
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
if (!rmap_head->val) {
|
2018-10-04 02:04:22 +00:00
|
|
|
pr_err("%s: %p 0->BUG\n", __func__, spte);
|
2007-01-06 00:36:38 +00:00
|
|
|
BUG();
|
2015-11-20 08:41:28 +00:00
|
|
|
} else if (!(rmap_head->val & 1)) {
|
2018-10-04 02:04:22 +00:00
|
|
|
rmap_printk("%s: %p 1->0\n", __func__, spte);
|
2015-11-20 08:41:28 +00:00
|
|
|
if ((u64 *)rmap_head->val != spte) {
|
2018-10-04 02:04:22 +00:00
|
|
|
pr_err("%s: %p 1->BUG\n", __func__, spte);
|
2007-01-06 00:36:38 +00:00
|
|
|
BUG();
|
|
|
|
}
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head->val = 0;
|
2007-01-06 00:36:38 +00:00
|
|
|
} else {
|
2018-10-04 02:04:22 +00:00
|
|
|
rmap_printk("%s: %p many->many\n", __func__, spte);
|
2015-11-20 08:41:28 +00:00
|
|
|
desc = (struct pte_list_desc *)(rmap_head->val & ~1ul);
|
2007-01-06 00:36:38 +00:00
|
|
|
prev_desc = NULL;
|
|
|
|
while (desc) {
|
2015-11-20 08:41:28 +00:00
|
|
|
for (i = 0; i < PTE_LIST_EXT && desc->sptes[i]; ++i) {
|
2009-06-10 11:24:23 +00:00
|
|
|
if (desc->sptes[i] == spte) {
|
2015-11-20 08:41:28 +00:00
|
|
|
pte_list_desc_remove_entry(rmap_head,
|
|
|
|
desc, i, prev_desc);
|
2007-01-06 00:36:38 +00:00
|
|
|
return;
|
|
|
|
}
|
2015-11-20 08:41:28 +00:00
|
|
|
}
|
2007-01-06 00:36:38 +00:00
|
|
|
prev_desc = desc;
|
|
|
|
desc = desc->more;
|
|
|
|
}
|
2018-10-04 02:04:22 +00:00
|
|
|
pr_err("%s: %p many->many\n", __func__, spte);
|
2007-01-06 00:36:38 +00:00
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-10-04 02:04:23 +00:00
|
|
|
static void pte_list_remove(struct kvm_rmap_head *rmap_head, u64 *sptep)
|
|
|
|
{
|
|
|
|
mmu_spte_clear_track_bits(sptep);
|
|
|
|
__pte_list_remove(sptep, rmap_head);
|
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
static struct kvm_rmap_head *__gfn_to_rmap(gfn_t gfn, int level,
|
|
|
|
struct kvm_memory_slot *slot)
|
2011-05-15 15:26:20 +00:00
|
|
|
{
|
2012-07-02 08:57:17 +00:00
|
|
|
unsigned long idx;
|
2011-05-15 15:26:20 +00:00
|
|
|
|
2012-07-02 08:57:17 +00:00
|
|
|
idx = gfn_to_index(gfn, slot->base_gfn, level);
|
2012-08-01 09:03:28 +00:00
|
|
|
return &slot->arch.rmap[level - PT_PAGE_TABLE_LEVEL][idx];
|
2011-05-15 15:26:20 +00:00
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
static struct kvm_rmap_head *gfn_to_rmap(struct kvm *kvm, gfn_t gfn,
|
|
|
|
struct kvm_mmu_page *sp)
|
2011-11-14 09:22:28 +00:00
|
|
|
{
|
2015-05-18 13:03:39 +00:00
|
|
|
struct kvm_memslots *slots;
|
2011-11-14 09:22:28 +00:00
|
|
|
struct kvm_memory_slot *slot;
|
|
|
|
|
2015-05-18 13:03:39 +00:00
|
|
|
slots = kvm_memslots_for_spte_role(kvm, sp->role);
|
|
|
|
slot = __gfn_to_memslot(slots, gfn);
|
2015-05-18 13:11:46 +00:00
|
|
|
return __gfn_to_rmap(gfn, sp->role.level, slot);
|
2011-11-14 09:22:28 +00:00
|
|
|
}
|
|
|
|
|
2011-09-22 08:53:17 +00:00
|
|
|
static bool rmap_can_add(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_memory_cache *cache;
|
|
|
|
|
|
|
|
cache = &vcpu->arch.mmu_pte_list_desc_cache;
|
|
|
|
return mmu_memory_cache_free_objects(cache);
|
|
|
|
}
|
|
|
|
|
2011-05-15 15:26:20 +00:00
|
|
|
static int rmap_add(struct kvm_vcpu *vcpu, u64 *spte, gfn_t gfn)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp;
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *rmap_head;
|
2011-05-15 15:26:20 +00:00
|
|
|
|
|
|
|
sp = page_header(__pa(spte));
|
|
|
|
kvm_mmu_page_set_gfn(sp, spte - sp->spt, gfn);
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head = gfn_to_rmap(vcpu->kvm, gfn, sp);
|
|
|
|
return pte_list_add(vcpu, spte, rmap_head);
|
2011-05-15 15:26:20 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void rmap_remove(struct kvm *kvm, u64 *spte)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
gfn_t gfn;
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *rmap_head;
|
2011-05-15 15:26:20 +00:00
|
|
|
|
|
|
|
sp = page_header(__pa(spte));
|
|
|
|
gfn = kvm_mmu_page_get_gfn(sp, spte - sp->spt);
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head = gfn_to_rmap(kvm, gfn, sp);
|
2018-10-04 02:04:22 +00:00
|
|
|
__pte_list_remove(spte, rmap_head);
|
2011-05-15 15:26:20 +00:00
|
|
|
}
|
|
|
|
|
2012-03-21 14:50:34 +00:00
|
|
|
/*
|
|
|
|
* Used by the following functions to iterate through the sptes linked by a
|
|
|
|
* rmap. All fields are private and not assumed to be used outside.
|
|
|
|
*/
|
|
|
|
struct rmap_iterator {
|
|
|
|
/* private fields */
|
|
|
|
struct pte_list_desc *desc; /* holds the sptep if not NULL */
|
|
|
|
int pos; /* index of the sptep */
|
|
|
|
};
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Iteration must be started by this function. This should also be used after
|
|
|
|
* removing/dropping sptes from the rmap link because in such cases the
|
|
|
|
* information in the itererator may not be valid.
|
|
|
|
*
|
|
|
|
* Returns sptep if found, NULL otherwise.
|
|
|
|
*/
|
2015-11-20 08:41:28 +00:00
|
|
|
static u64 *rmap_get_first(struct kvm_rmap_head *rmap_head,
|
|
|
|
struct rmap_iterator *iter)
|
2012-03-21 14:50:34 +00:00
|
|
|
{
|
2015-11-20 08:45:44 +00:00
|
|
|
u64 *sptep;
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
if (!rmap_head->val)
|
2012-03-21 14:50:34 +00:00
|
|
|
return NULL;
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
if (!(rmap_head->val & 1)) {
|
2012-03-21 14:50:34 +00:00
|
|
|
iter->desc = NULL;
|
2015-11-20 08:45:44 +00:00
|
|
|
sptep = (u64 *)rmap_head->val;
|
|
|
|
goto out;
|
2012-03-21 14:50:34 +00:00
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
iter->desc = (struct pte_list_desc *)(rmap_head->val & ~1ul);
|
2012-03-21 14:50:34 +00:00
|
|
|
iter->pos = 0;
|
2015-11-20 08:45:44 +00:00
|
|
|
sptep = iter->desc->sptes[iter->pos];
|
|
|
|
out:
|
|
|
|
BUG_ON(!is_shadow_present_pte(*sptep));
|
|
|
|
return sptep;
|
2012-03-21 14:50:34 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Must be used with a valid iterator: e.g. after rmap_get_first().
|
|
|
|
*
|
|
|
|
* Returns sptep if found, NULL otherwise.
|
|
|
|
*/
|
|
|
|
static u64 *rmap_get_next(struct rmap_iterator *iter)
|
|
|
|
{
|
2015-11-20 08:45:44 +00:00
|
|
|
u64 *sptep;
|
|
|
|
|
2012-03-21 14:50:34 +00:00
|
|
|
if (iter->desc) {
|
|
|
|
if (iter->pos < PTE_LIST_EXT - 1) {
|
|
|
|
++iter->pos;
|
|
|
|
sptep = iter->desc->sptes[iter->pos];
|
|
|
|
if (sptep)
|
2015-11-20 08:45:44 +00:00
|
|
|
goto out;
|
2012-03-21 14:50:34 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
iter->desc = iter->desc->more;
|
|
|
|
|
|
|
|
if (iter->desc) {
|
|
|
|
iter->pos = 0;
|
|
|
|
/* desc->sptes[0] cannot be NULL */
|
2015-11-20 08:45:44 +00:00
|
|
|
sptep = iter->desc->sptes[iter->pos];
|
|
|
|
goto out;
|
2012-03-21 14:50:34 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return NULL;
|
2015-11-20 08:45:44 +00:00
|
|
|
out:
|
|
|
|
BUG_ON(!is_shadow_present_pte(*sptep));
|
|
|
|
return sptep;
|
2012-03-21 14:50:34 +00:00
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
#define for_each_rmap_spte(_rmap_head_, _iter_, _spte_) \
|
|
|
|
for (_spte_ = rmap_get_first(_rmap_head_, _iter_); \
|
2015-11-20 08:45:44 +00:00
|
|
|
_spte_; _spte_ = rmap_get_next(_iter_))
|
2015-05-13 06:42:20 +00:00
|
|
|
|
2011-07-11 19:28:04 +00:00
|
|
|
static void drop_spte(struct kvm *kvm, u64 *sptep)
|
2010-07-16 03:28:09 +00:00
|
|
|
{
|
2011-07-11 19:30:35 +00:00
|
|
|
if (mmu_spte_clear_track_bits(sptep))
|
2010-10-25 13:58:22 +00:00
|
|
|
rmap_remove(kvm, sptep);
|
2010-06-06 11:31:27 +00:00
|
|
|
}
|
|
|
|
|
2012-06-20 07:57:39 +00:00
|
|
|
|
|
|
|
static bool __drop_large_spte(struct kvm *kvm, u64 *sptep)
|
|
|
|
{
|
|
|
|
if (is_large_pte(*sptep)) {
|
|
|
|
WARN_ON(page_header(__pa(sptep))->role.level ==
|
|
|
|
PT_PAGE_TABLE_LEVEL);
|
|
|
|
drop_spte(kvm, sptep);
|
|
|
|
--kvm->stat.lpages;
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void drop_large_spte(struct kvm_vcpu *vcpu, u64 *sptep)
|
|
|
|
{
|
2018-12-06 13:21:09 +00:00
|
|
|
if (__drop_large_spte(vcpu->kvm, sptep)) {
|
|
|
|
struct kvm_mmu_page *sp = page_header(__pa(sptep));
|
|
|
|
|
|
|
|
kvm_flush_remote_tlbs_with_address(vcpu->kvm, sp->gfn,
|
|
|
|
KVM_PAGES_PER_HPAGE(sp->role.level));
|
|
|
|
}
|
2012-06-20 07:57:39 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2012-06-20 07:58:58 +00:00
|
|
|
* Write-protect on the specified @sptep, @pt_protect indicates whether
|
2014-04-17 09:06:14 +00:00
|
|
|
* spte write-protection is caused by protecting shadow page table.
|
2012-06-20 07:58:58 +00:00
|
|
|
*
|
2014-09-22 02:31:38 +00:00
|
|
|
* Note: write protection is difference between dirty logging and spte
|
2012-06-20 07:58:58 +00:00
|
|
|
* protection:
|
|
|
|
* - for dirty logging, the spte can be set to writable at anytime if
|
|
|
|
* its dirty bitmap is properly set.
|
|
|
|
* - for spte protection, the spte can be writable only after unsync-ing
|
|
|
|
* shadow page.
|
2012-06-20 07:57:39 +00:00
|
|
|
*
|
2014-04-17 09:06:14 +00:00
|
|
|
* Return true if tlb need be flushed.
|
2012-06-20 07:57:39 +00:00
|
|
|
*/
|
2016-08-02 20:32:37 +00:00
|
|
|
static bool spte_write_protect(u64 *sptep, bool pt_protect)
|
2012-06-20 07:57:15 +00:00
|
|
|
{
|
|
|
|
u64 spte = *sptep;
|
|
|
|
|
2012-06-20 07:58:58 +00:00
|
|
|
if (!is_writable_pte(spte) &&
|
2016-12-07 00:46:11 +00:00
|
|
|
!(pt_protect && spte_can_locklessly_be_made_writable(spte)))
|
2012-06-20 07:57:15 +00:00
|
|
|
return false;
|
|
|
|
|
|
|
|
rmap_printk("rmap_write_protect: spte %p %llx\n", sptep, *sptep);
|
|
|
|
|
2012-06-20 07:58:58 +00:00
|
|
|
if (pt_protect)
|
|
|
|
spte &= ~SPTE_MMU_WRITEABLE;
|
2012-06-20 07:57:15 +00:00
|
|
|
spte = spte & ~PT_WRITABLE_MASK;
|
2012-06-20 07:58:58 +00:00
|
|
|
|
2014-04-17 09:06:14 +00:00
|
|
|
return mmu_spte_update(sptep, spte);
|
2012-06-20 07:57:15 +00:00
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
static bool __rmap_write_protect(struct kvm *kvm,
|
|
|
|
struct kvm_rmap_head *rmap_head,
|
2013-01-08 10:44:09 +00:00
|
|
|
bool pt_protect)
|
2007-10-16 12:42:30 +00:00
|
|
|
{
|
2012-03-21 14:50:34 +00:00
|
|
|
u64 *sptep;
|
|
|
|
struct rmap_iterator iter;
|
2012-06-20 07:57:15 +00:00
|
|
|
bool flush = false;
|
2007-01-06 00:36:43 +00:00
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
for_each_rmap_spte(rmap_head, &iter, sptep)
|
2016-08-02 20:32:37 +00:00
|
|
|
flush |= spte_write_protect(sptep, pt_protect);
|
2008-03-20 16:17:24 +00:00
|
|
|
|
2012-06-20 07:57:15 +00:00
|
|
|
return flush;
|
2012-03-01 10:31:22 +00:00
|
|
|
}
|
|
|
|
|
2016-08-02 20:32:37 +00:00
|
|
|
static bool spte_clear_dirty(u64 *sptep)
|
2015-01-28 02:54:24 +00:00
|
|
|
{
|
|
|
|
u64 spte = *sptep;
|
|
|
|
|
|
|
|
rmap_printk("rmap_clear_dirty: spte %p %llx\n", sptep, *sptep);
|
|
|
|
|
|
|
|
spte &= ~shadow_dirty_mask;
|
|
|
|
|
|
|
|
return mmu_spte_update(sptep, spte);
|
|
|
|
}
|
|
|
|
|
2017-07-01 00:26:31 +00:00
|
|
|
static bool wrprot_ad_disabled_spte(u64 *sptep)
|
|
|
|
{
|
|
|
|
bool was_writable = test_and_clear_bit(PT_WRITABLE_SHIFT,
|
|
|
|
(unsigned long *)sptep);
|
|
|
|
if (was_writable)
|
|
|
|
kvm_set_pfn_dirty(spte_to_pfn(*sptep));
|
|
|
|
|
|
|
|
return was_writable;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Gets the GFN ready for another round of dirty logging by clearing the
|
|
|
|
* - D bit on ad-enabled SPTEs, and
|
|
|
|
* - W bit on ad-disabled SPTEs.
|
|
|
|
* Returns true iff any D or W bits were cleared.
|
|
|
|
*/
|
2015-11-20 08:41:28 +00:00
|
|
|
static bool __rmap_clear_dirty(struct kvm *kvm, struct kvm_rmap_head *rmap_head)
|
2015-01-28 02:54:24 +00:00
|
|
|
{
|
|
|
|
u64 *sptep;
|
|
|
|
struct rmap_iterator iter;
|
|
|
|
bool flush = false;
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
for_each_rmap_spte(rmap_head, &iter, sptep)
|
2017-07-01 00:26:31 +00:00
|
|
|
if (spte_ad_enabled(*sptep))
|
|
|
|
flush |= spte_clear_dirty(sptep);
|
|
|
|
else
|
|
|
|
flush |= wrprot_ad_disabled_spte(sptep);
|
2015-01-28 02:54:24 +00:00
|
|
|
|
|
|
|
return flush;
|
|
|
|
}
|
|
|
|
|
2016-08-02 20:32:37 +00:00
|
|
|
static bool spte_set_dirty(u64 *sptep)
|
2015-01-28 02:54:24 +00:00
|
|
|
{
|
|
|
|
u64 spte = *sptep;
|
|
|
|
|
|
|
|
rmap_printk("rmap_set_dirty: spte %p %llx\n", sptep, *sptep);
|
|
|
|
|
|
|
|
spte |= shadow_dirty_mask;
|
|
|
|
|
|
|
|
return mmu_spte_update(sptep, spte);
|
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
static bool __rmap_set_dirty(struct kvm *kvm, struct kvm_rmap_head *rmap_head)
|
2015-01-28 02:54:24 +00:00
|
|
|
{
|
|
|
|
u64 *sptep;
|
|
|
|
struct rmap_iterator iter;
|
|
|
|
bool flush = false;
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
for_each_rmap_spte(rmap_head, &iter, sptep)
|
2017-07-01 00:26:31 +00:00
|
|
|
if (spte_ad_enabled(*sptep))
|
|
|
|
flush |= spte_set_dirty(sptep);
|
2015-01-28 02:54:24 +00:00
|
|
|
|
|
|
|
return flush;
|
|
|
|
}
|
|
|
|
|
2012-03-01 10:32:16 +00:00
|
|
|
/**
|
2015-01-28 02:54:23 +00:00
|
|
|
* kvm_mmu_write_protect_pt_masked - write protect selected PT level pages
|
2012-03-01 10:32:16 +00:00
|
|
|
* @kvm: kvm instance
|
|
|
|
* @slot: slot to protect
|
|
|
|
* @gfn_offset: start of the BITS_PER_LONG pages we care about
|
|
|
|
* @mask: indicates which pages we should protect
|
|
|
|
*
|
|
|
|
* Used when we do not need to care about huge page mappings: e.g. during dirty
|
|
|
|
* logging we do not have any such mappings.
|
|
|
|
*/
|
2015-01-28 02:54:23 +00:00
|
|
|
static void kvm_mmu_write_protect_pt_masked(struct kvm *kvm,
|
2012-03-01 10:32:16 +00:00
|
|
|
struct kvm_memory_slot *slot,
|
|
|
|
gfn_t gfn_offset, unsigned long mask)
|
2012-03-01 10:31:22 +00:00
|
|
|
{
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *rmap_head;
|
2012-03-01 10:31:22 +00:00
|
|
|
|
2012-03-01 10:32:16 +00:00
|
|
|
while (mask) {
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head = __gfn_to_rmap(slot->base_gfn + gfn_offset + __ffs(mask),
|
|
|
|
PT_PAGE_TABLE_LEVEL, slot);
|
|
|
|
__rmap_write_protect(kvm, rmap_head, false);
|
2008-02-23 14:44:30 +00:00
|
|
|
|
2012-03-01 10:32:16 +00:00
|
|
|
/* clear the first set bit */
|
|
|
|
mask &= mask - 1;
|
|
|
|
}
|
2007-01-06 00:36:43 +00:00
|
|
|
}
|
|
|
|
|
2015-01-28 02:54:24 +00:00
|
|
|
/**
|
2017-07-01 00:26:31 +00:00
|
|
|
* kvm_mmu_clear_dirty_pt_masked - clear MMU D-bit for PT level pages, or write
|
|
|
|
* protect the page if the D-bit isn't supported.
|
2015-01-28 02:54:24 +00:00
|
|
|
* @kvm: kvm instance
|
|
|
|
* @slot: slot to clear D-bit
|
|
|
|
* @gfn_offset: start of the BITS_PER_LONG pages we care about
|
|
|
|
* @mask: indicates which pages we should clear D-bit
|
|
|
|
*
|
|
|
|
* Used for PML to re-log the dirty GPAs after userspace querying dirty_bitmap.
|
|
|
|
*/
|
|
|
|
void kvm_mmu_clear_dirty_pt_masked(struct kvm *kvm,
|
|
|
|
struct kvm_memory_slot *slot,
|
|
|
|
gfn_t gfn_offset, unsigned long mask)
|
|
|
|
{
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *rmap_head;
|
2015-01-28 02:54:24 +00:00
|
|
|
|
|
|
|
while (mask) {
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head = __gfn_to_rmap(slot->base_gfn + gfn_offset + __ffs(mask),
|
|
|
|
PT_PAGE_TABLE_LEVEL, slot);
|
|
|
|
__rmap_clear_dirty(kvm, rmap_head);
|
2015-01-28 02:54:24 +00:00
|
|
|
|
|
|
|
/* clear the first set bit */
|
|
|
|
mask &= mask - 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_clear_dirty_pt_masked);
|
|
|
|
|
2015-01-28 02:54:23 +00:00
|
|
|
/**
|
|
|
|
* kvm_arch_mmu_enable_log_dirty_pt_masked - enable dirty logging for selected
|
|
|
|
* PT level pages.
|
|
|
|
*
|
|
|
|
* It calls kvm_mmu_write_protect_pt_masked to write protect selected pages to
|
|
|
|
* enable dirty logging for them.
|
|
|
|
*
|
|
|
|
* Used when we do not need to care about huge page mappings: e.g. during dirty
|
|
|
|
* logging we do not have any such mappings.
|
|
|
|
*/
|
|
|
|
void kvm_arch_mmu_enable_log_dirty_pt_masked(struct kvm *kvm,
|
|
|
|
struct kvm_memory_slot *slot,
|
|
|
|
gfn_t gfn_offset, unsigned long mask)
|
|
|
|
{
|
2015-01-28 02:54:27 +00:00
|
|
|
if (kvm_x86_ops->enable_log_dirty_pt_masked)
|
|
|
|
kvm_x86_ops->enable_log_dirty_pt_masked(kvm, slot, gfn_offset,
|
|
|
|
mask);
|
|
|
|
else
|
|
|
|
kvm_mmu_write_protect_pt_masked(kvm, slot, gfn_offset, mask);
|
2015-01-28 02:54:23 +00:00
|
|
|
}
|
|
|
|
|
2017-05-05 19:25:13 +00:00
|
|
|
/**
|
|
|
|
* kvm_arch_write_log_dirty - emulate dirty page logging
|
|
|
|
* @vcpu: Guest mode vcpu
|
|
|
|
*
|
|
|
|
* Emulate arch specific page modification logging for the
|
|
|
|
* nested hypervisor
|
|
|
|
*/
|
|
|
|
int kvm_arch_write_log_dirty(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
|
|
|
if (kvm_x86_ops->write_log_dirty)
|
|
|
|
return kvm_x86_ops->write_log_dirty(vcpu);
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:08 +00:00
|
|
|
bool kvm_mmu_slot_gfn_write_protect(struct kvm *kvm,
|
|
|
|
struct kvm_memory_slot *slot, u64 gfn)
|
2011-11-14 09:24:50 +00:00
|
|
|
{
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *rmap_head;
|
2012-03-01 10:32:16 +00:00
|
|
|
int i;
|
2012-06-20 07:56:53 +00:00
|
|
|
bool write_protected = false;
|
2011-11-14 09:24:50 +00:00
|
|
|
|
2015-05-13 06:42:21 +00:00
|
|
|
for (i = PT_PAGE_TABLE_LEVEL; i <= PT_MAX_HUGEPAGE_LEVEL; ++i) {
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head = __gfn_to_rmap(gfn, i, slot);
|
2016-02-24 09:51:08 +00:00
|
|
|
write_protected |= __rmap_write_protect(kvm, rmap_head, true);
|
2012-03-01 10:32:16 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return write_protected;
|
2011-11-14 09:24:50 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:08 +00:00
|
|
|
static bool rmap_write_protect(struct kvm_vcpu *vcpu, u64 gfn)
|
|
|
|
{
|
|
|
|
struct kvm_memory_slot *slot;
|
|
|
|
|
|
|
|
slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);
|
|
|
|
return kvm_mmu_slot_gfn_write_protect(vcpu->kvm, slot, gfn);
|
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
static bool kvm_zap_rmapp(struct kvm *kvm, struct kvm_rmap_head *rmap_head)
|
2008-07-25 14:24:52 +00:00
|
|
|
{
|
2012-03-21 14:50:34 +00:00
|
|
|
u64 *sptep;
|
|
|
|
struct rmap_iterator iter;
|
2015-05-13 06:42:25 +00:00
|
|
|
bool flush = false;
|
2008-07-25 14:24:52 +00:00
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
while ((sptep = rmap_get_first(rmap_head, &iter))) {
|
2015-05-13 06:42:25 +00:00
|
|
|
rmap_printk("%s: spte %p %llx.\n", __func__, sptep, *sptep);
|
2012-03-21 14:50:34 +00:00
|
|
|
|
2018-10-04 02:04:23 +00:00
|
|
|
pte_list_remove(rmap_head, sptep);
|
2015-05-13 06:42:25 +00:00
|
|
|
flush = true;
|
2008-07-25 14:24:52 +00:00
|
|
|
}
|
2012-03-21 14:50:34 +00:00
|
|
|
|
2015-05-13 06:42:25 +00:00
|
|
|
return flush;
|
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
static int kvm_unmap_rmapp(struct kvm *kvm, struct kvm_rmap_head *rmap_head,
|
2015-05-13 06:42:25 +00:00
|
|
|
struct kvm_memory_slot *slot, gfn_t gfn, int level,
|
|
|
|
unsigned long data)
|
|
|
|
{
|
2015-11-20 08:41:28 +00:00
|
|
|
return kvm_zap_rmapp(kvm, rmap_head);
|
2008-07-25 14:24:52 +00:00
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
static int kvm_set_pte_rmapp(struct kvm *kvm, struct kvm_rmap_head *rmap_head,
|
2014-09-23 19:34:54 +00:00
|
|
|
struct kvm_memory_slot *slot, gfn_t gfn, int level,
|
|
|
|
unsigned long data)
|
2009-09-23 18:47:18 +00:00
|
|
|
{
|
2012-03-21 14:50:34 +00:00
|
|
|
u64 *sptep;
|
|
|
|
struct rmap_iterator iter;
|
2009-09-23 18:47:18 +00:00
|
|
|
int need_flush = 0;
|
2012-03-21 14:50:34 +00:00
|
|
|
u64 new_spte;
|
2009-09-23 18:47:18 +00:00
|
|
|
pte_t *ptep = (pte_t *)data;
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
kvm_pfn_t new_pfn;
|
2009-09-23 18:47:18 +00:00
|
|
|
|
|
|
|
WARN_ON(pte_huge(*ptep));
|
|
|
|
new_pfn = pte_pfn(*ptep);
|
2012-03-21 14:50:34 +00:00
|
|
|
|
2015-05-13 06:42:20 +00:00
|
|
|
restart:
|
2015-11-20 08:41:28 +00:00
|
|
|
for_each_rmap_spte(rmap_head, &iter, sptep) {
|
2014-09-23 19:34:54 +00:00
|
|
|
rmap_printk("kvm_set_pte_rmapp: spte %p %llx gfn %llx (%d)\n",
|
2016-12-07 00:46:16 +00:00
|
|
|
sptep, *sptep, gfn, level);
|
2012-03-21 14:50:34 +00:00
|
|
|
|
2009-09-23 18:47:18 +00:00
|
|
|
need_flush = 1;
|
2012-03-21 14:50:34 +00:00
|
|
|
|
2009-09-23 18:47:18 +00:00
|
|
|
if (pte_write(*ptep)) {
|
2018-10-04 02:04:23 +00:00
|
|
|
pte_list_remove(rmap_head, sptep);
|
2015-05-13 06:42:20 +00:00
|
|
|
goto restart;
|
2009-09-23 18:47:18 +00:00
|
|
|
} else {
|
2012-03-21 14:50:34 +00:00
|
|
|
new_spte = *sptep & ~PT64_BASE_ADDR_MASK;
|
2009-09-23 18:47:18 +00:00
|
|
|
new_spte |= (u64)new_pfn << PAGE_SHIFT;
|
|
|
|
|
|
|
|
new_spte &= ~PT_WRITABLE_MASK;
|
|
|
|
new_spte &= ~SPTE_HOST_WRITEABLE;
|
2016-12-07 00:46:16 +00:00
|
|
|
|
|
|
|
new_spte = mark_spte_for_access_track(new_spte);
|
2012-03-21 14:50:34 +00:00
|
|
|
|
|
|
|
mmu_spte_clear_track_bits(sptep);
|
|
|
|
mmu_spte_set(sptep, new_spte);
|
2009-09-23 18:47:18 +00:00
|
|
|
}
|
|
|
|
}
|
2012-03-21 14:50:34 +00:00
|
|
|
|
2018-12-06 13:21:12 +00:00
|
|
|
if (need_flush && kvm_available_flush_tlb_with_range()) {
|
|
|
|
kvm_flush_remote_tlbs_with_address(kvm, gfn, 1);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2018-12-06 13:21:11 +00:00
|
|
|
return need_flush;
|
2009-09-23 18:47:18 +00:00
|
|
|
}
|
|
|
|
|
2015-05-13 06:42:22 +00:00
|
|
|
struct slot_rmap_walk_iterator {
|
|
|
|
/* input fields. */
|
|
|
|
struct kvm_memory_slot *slot;
|
|
|
|
gfn_t start_gfn;
|
|
|
|
gfn_t end_gfn;
|
|
|
|
int start_level;
|
|
|
|
int end_level;
|
|
|
|
|
|
|
|
/* output fields. */
|
|
|
|
gfn_t gfn;
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *rmap;
|
2015-05-13 06:42:22 +00:00
|
|
|
int level;
|
|
|
|
|
|
|
|
/* private field. */
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *end_rmap;
|
2015-05-13 06:42:22 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
static void
|
|
|
|
rmap_walk_init_level(struct slot_rmap_walk_iterator *iterator, int level)
|
|
|
|
{
|
|
|
|
iterator->level = level;
|
|
|
|
iterator->gfn = iterator->start_gfn;
|
|
|
|
iterator->rmap = __gfn_to_rmap(iterator->gfn, level, iterator->slot);
|
|
|
|
iterator->end_rmap = __gfn_to_rmap(iterator->end_gfn, level,
|
|
|
|
iterator->slot);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
slot_rmap_walk_init(struct slot_rmap_walk_iterator *iterator,
|
|
|
|
struct kvm_memory_slot *slot, int start_level,
|
|
|
|
int end_level, gfn_t start_gfn, gfn_t end_gfn)
|
|
|
|
{
|
|
|
|
iterator->slot = slot;
|
|
|
|
iterator->start_level = start_level;
|
|
|
|
iterator->end_level = end_level;
|
|
|
|
iterator->start_gfn = start_gfn;
|
|
|
|
iterator->end_gfn = end_gfn;
|
|
|
|
|
|
|
|
rmap_walk_init_level(iterator, iterator->start_level);
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool slot_rmap_walk_okay(struct slot_rmap_walk_iterator *iterator)
|
|
|
|
{
|
|
|
|
return !!iterator->rmap;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void slot_rmap_walk_next(struct slot_rmap_walk_iterator *iterator)
|
|
|
|
{
|
|
|
|
if (++iterator->rmap <= iterator->end_rmap) {
|
|
|
|
iterator->gfn += (1UL << KVM_HPAGE_GFN_SHIFT(iterator->level));
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (++iterator->level > iterator->end_level) {
|
|
|
|
iterator->rmap = NULL;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
rmap_walk_init_level(iterator, iterator->level);
|
|
|
|
}
|
|
|
|
|
|
|
|
#define for_each_slot_rmap_range(_slot_, _start_level_, _end_level_, \
|
|
|
|
_start_gfn, _end_gfn, _iter_) \
|
|
|
|
for (slot_rmap_walk_init(_iter_, _slot_, _start_level_, \
|
|
|
|
_end_level_, _start_gfn, _end_gfn); \
|
|
|
|
slot_rmap_walk_okay(_iter_); \
|
|
|
|
slot_rmap_walk_next(_iter_))
|
|
|
|
|
KVM: MMU: Make kvm_handle_hva() handle range of addresses
When guest's memory is backed by THP pages, MMU notifier needs to call
kvm_unmap_hva(), which in turn leads to kvm_handle_hva(), in a loop to
invalidate a range of pages which constitute one huge page:
for each page
for each memslot
if page is in memslot
unmap using rmap
This means although every page in that range is expected to be found in
the same memslot, we are forced to check unrelated memslots many times.
If the guest has more memslots, the situation will become worse.
Furthermore, if the range does not include any pages in the guest's
memory, the loop over the pages will just consume extra time.
This patch, together with the following patches, solves this problem by
introducing kvm_handle_hva_range() which makes the loop look like this:
for each memslot
for each page in memslot
unmap using rmap
In this new processing, the actual work is converted to a loop over rmap
which is much more cache friendly than before.
Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp>
Cc: Alexander Graf <agraf@suse.de>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
2012-07-02 08:55:48 +00:00
|
|
|
static int kvm_handle_hva_range(struct kvm *kvm,
|
|
|
|
unsigned long start,
|
|
|
|
unsigned long end,
|
|
|
|
unsigned long data,
|
|
|
|
int (*handler)(struct kvm *kvm,
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *rmap_head,
|
2012-07-02 08:57:59 +00:00
|
|
|
struct kvm_memory_slot *slot,
|
2014-09-23 19:34:54 +00:00
|
|
|
gfn_t gfn,
|
|
|
|
int level,
|
KVM: MMU: Make kvm_handle_hva() handle range of addresses
When guest's memory is backed by THP pages, MMU notifier needs to call
kvm_unmap_hva(), which in turn leads to kvm_handle_hva(), in a loop to
invalidate a range of pages which constitute one huge page:
for each page
for each memslot
if page is in memslot
unmap using rmap
This means although every page in that range is expected to be found in
the same memslot, we are forced to check unrelated memslots many times.
If the guest has more memslots, the situation will become worse.
Furthermore, if the range does not include any pages in the guest's
memory, the loop over the pages will just consume extra time.
This patch, together with the following patches, solves this problem by
introducing kvm_handle_hva_range() which makes the loop look like this:
for each memslot
for each page in memslot
unmap using rmap
In this new processing, the actual work is converted to a loop over rmap
which is much more cache friendly than before.
Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp>
Cc: Alexander Graf <agraf@suse.de>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
2012-07-02 08:55:48 +00:00
|
|
|
unsigned long data))
|
2008-07-25 14:24:52 +00:00
|
|
|
{
|
2009-12-23 16:35:21 +00:00
|
|
|
struct kvm_memslots *slots;
|
2011-11-24 09:39:18 +00:00
|
|
|
struct kvm_memory_slot *memslot;
|
2015-05-13 06:42:22 +00:00
|
|
|
struct slot_rmap_walk_iterator iterator;
|
|
|
|
int ret = 0;
|
2015-05-18 11:33:16 +00:00
|
|
|
int i;
|
2009-12-23 16:35:21 +00:00
|
|
|
|
2015-05-18 11:33:16 +00:00
|
|
|
for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++) {
|
|
|
|
slots = __kvm_memslots(kvm, i);
|
|
|
|
kvm_for_each_memslot(memslot, slots) {
|
|
|
|
unsigned long hva_start, hva_end;
|
|
|
|
gfn_t gfn_start, gfn_end;
|
2008-07-25 14:24:52 +00:00
|
|
|
|
2015-05-18 11:33:16 +00:00
|
|
|
hva_start = max(start, memslot->userspace_addr);
|
|
|
|
hva_end = min(end, memslot->userspace_addr +
|
|
|
|
(memslot->npages << PAGE_SHIFT));
|
|
|
|
if (hva_start >= hva_end)
|
|
|
|
continue;
|
|
|
|
/*
|
|
|
|
* {gfn(page) | page intersects with [hva_start, hva_end)} =
|
|
|
|
* {gfn_start, gfn_start+1, ..., gfn_end-1}.
|
|
|
|
*/
|
|
|
|
gfn_start = hva_to_gfn_memslot(hva_start, memslot);
|
|
|
|
gfn_end = hva_to_gfn_memslot(hva_end + PAGE_SIZE - 1, memslot);
|
|
|
|
|
|
|
|
for_each_slot_rmap_range(memslot, PT_PAGE_TABLE_LEVEL,
|
|
|
|
PT_MAX_HUGEPAGE_LEVEL,
|
|
|
|
gfn_start, gfn_end - 1,
|
|
|
|
&iterator)
|
|
|
|
ret |= handler(kvm, iterator.rmap, memslot,
|
|
|
|
iterator.gfn, iterator.level, data);
|
|
|
|
}
|
2008-07-25 14:24:52 +00:00
|
|
|
}
|
|
|
|
|
2012-07-02 08:58:48 +00:00
|
|
|
return ret;
|
2008-07-25 14:24:52 +00:00
|
|
|
}
|
|
|
|
|
KVM: MMU: Make kvm_handle_hva() handle range of addresses
When guest's memory is backed by THP pages, MMU notifier needs to call
kvm_unmap_hva(), which in turn leads to kvm_handle_hva(), in a loop to
invalidate a range of pages which constitute one huge page:
for each page
for each memslot
if page is in memslot
unmap using rmap
This means although every page in that range is expected to be found in
the same memslot, we are forced to check unrelated memslots many times.
If the guest has more memslots, the situation will become worse.
Furthermore, if the range does not include any pages in the guest's
memory, the loop over the pages will just consume extra time.
This patch, together with the following patches, solves this problem by
introducing kvm_handle_hva_range() which makes the loop look like this:
for each memslot
for each page in memslot
unmap using rmap
In this new processing, the actual work is converted to a loop over rmap
which is much more cache friendly than before.
Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp>
Cc: Alexander Graf <agraf@suse.de>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
2012-07-02 08:55:48 +00:00
|
|
|
static int kvm_handle_hva(struct kvm *kvm, unsigned long hva,
|
|
|
|
unsigned long data,
|
2015-11-20 08:41:28 +00:00
|
|
|
int (*handler)(struct kvm *kvm,
|
|
|
|
struct kvm_rmap_head *rmap_head,
|
2012-07-02 08:57:59 +00:00
|
|
|
struct kvm_memory_slot *slot,
|
2014-09-23 19:34:54 +00:00
|
|
|
gfn_t gfn, int level,
|
KVM: MMU: Make kvm_handle_hva() handle range of addresses
When guest's memory is backed by THP pages, MMU notifier needs to call
kvm_unmap_hva(), which in turn leads to kvm_handle_hva(), in a loop to
invalidate a range of pages which constitute one huge page:
for each page
for each memslot
if page is in memslot
unmap using rmap
This means although every page in that range is expected to be found in
the same memslot, we are forced to check unrelated memslots many times.
If the guest has more memslots, the situation will become worse.
Furthermore, if the range does not include any pages in the guest's
memory, the loop over the pages will just consume extra time.
This patch, together with the following patches, solves this problem by
introducing kvm_handle_hva_range() which makes the loop look like this:
for each memslot
for each page in memslot
unmap using rmap
In this new processing, the actual work is converted to a loop over rmap
which is much more cache friendly than before.
Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp>
Cc: Alexander Graf <agraf@suse.de>
Cc: Paul Mackerras <paulus@samba.org>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
2012-07-02 08:55:48 +00:00
|
|
|
unsigned long data))
|
|
|
|
{
|
|
|
|
return kvm_handle_hva_range(kvm, hva, hva + 1, data, handler);
|
2008-07-25 14:24:52 +00:00
|
|
|
}
|
|
|
|
|
2012-07-02 08:56:33 +00:00
|
|
|
int kvm_unmap_hva_range(struct kvm *kvm, unsigned long start, unsigned long end)
|
|
|
|
{
|
|
|
|
return kvm_handle_hva_range(kvm, start, end, 0, kvm_unmap_rmapp);
|
|
|
|
}
|
|
|
|
|
2018-12-06 13:21:10 +00:00
|
|
|
int kvm_set_spte_hva(struct kvm *kvm, unsigned long hva, pte_t pte)
|
2009-09-23 18:47:18 +00:00
|
|
|
{
|
2018-12-06 13:21:11 +00:00
|
|
|
return kvm_handle_hva(kvm, hva, (unsigned long)&pte, kvm_set_pte_rmapp);
|
2008-07-25 14:24:52 +00:00
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
static int kvm_age_rmapp(struct kvm *kvm, struct kvm_rmap_head *rmap_head,
|
2014-09-23 19:34:54 +00:00
|
|
|
struct kvm_memory_slot *slot, gfn_t gfn, int level,
|
|
|
|
unsigned long data)
|
2008-07-25 14:24:52 +00:00
|
|
|
{
|
2012-03-21 14:50:34 +00:00
|
|
|
u64 *sptep;
|
2012-06-03 08:34:08 +00:00
|
|
|
struct rmap_iterator uninitialized_var(iter);
|
2008-07-25 14:24:52 +00:00
|
|
|
int young = 0;
|
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
for_each_rmap_spte(rmap_head, &iter, sptep)
|
|
|
|
young |= mmu_spte_age(sptep);
|
2015-05-13 06:42:20 +00:00
|
|
|
|
2014-09-23 19:34:54 +00:00
|
|
|
trace_kvm_age_page(gfn, level, slot, young);
|
2008-07-25 14:24:52 +00:00
|
|
|
return young;
|
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
static int kvm_test_age_rmapp(struct kvm *kvm, struct kvm_rmap_head *rmap_head,
|
2014-09-23 19:34:54 +00:00
|
|
|
struct kvm_memory_slot *slot, gfn_t gfn,
|
|
|
|
int level, unsigned long data)
|
2011-01-13 23:47:10 +00:00
|
|
|
{
|
2012-03-21 14:50:34 +00:00
|
|
|
u64 *sptep;
|
|
|
|
struct rmap_iterator iter;
|
2011-01-13 23:47:10 +00:00
|
|
|
|
2016-12-07 00:46:13 +00:00
|
|
|
for_each_rmap_spte(rmap_head, &iter, sptep)
|
|
|
|
if (is_accessed_spte(*sptep))
|
|
|
|
return 1;
|
|
|
|
return 0;
|
2011-01-13 23:47:10 +00:00
|
|
|
}
|
|
|
|
|
2009-08-05 18:43:58 +00:00
|
|
|
#define RMAP_RECYCLE_THRESHOLD 1000
|
|
|
|
|
2009-07-27 14:30:44 +00:00
|
|
|
static void rmap_recycle(struct kvm_vcpu *vcpu, u64 *spte, gfn_t gfn)
|
2009-08-05 18:43:58 +00:00
|
|
|
{
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *rmap_head;
|
2009-07-27 14:30:44 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
|
|
|
|
sp = page_header(__pa(spte));
|
2009-08-05 18:43:58 +00:00
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
rmap_head = gfn_to_rmap(vcpu->kvm, gfn, sp);
|
2009-08-05 18:43:58 +00:00
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
kvm_unmap_rmapp(vcpu->kvm, rmap_head, NULL, gfn, sp->role.level, 0);
|
2018-12-06 13:21:09 +00:00
|
|
|
kvm_flush_remote_tlbs_with_address(vcpu->kvm, sp->gfn,
|
|
|
|
KVM_PAGES_PER_HPAGE(sp->role.level));
|
2009-08-05 18:43:58 +00:00
|
|
|
}
|
|
|
|
|
2014-09-22 21:54:42 +00:00
|
|
|
int kvm_age_hva(struct kvm *kvm, unsigned long start, unsigned long end)
|
2008-07-25 14:24:52 +00:00
|
|
|
{
|
2014-09-22 21:54:42 +00:00
|
|
|
return kvm_handle_hva_range(kvm, start, end, 0, kvm_age_rmapp);
|
2008-07-25 14:24:52 +00:00
|
|
|
}
|
|
|
|
|
2011-01-13 23:47:10 +00:00
|
|
|
int kvm_test_age_hva(struct kvm *kvm, unsigned long hva)
|
|
|
|
{
|
|
|
|
return kvm_handle_hva(kvm, hva, 0, kvm_test_age_rmapp);
|
|
|
|
}
|
|
|
|
|
2007-04-25 06:17:25 +00:00
|
|
|
#ifdef MMU_DEBUG
|
2007-05-06 12:50:58 +00:00
|
|
|
static int is_empty_shadow_page(u64 *spt)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2007-01-06 00:36:50 +00:00
|
|
|
u64 *pos;
|
|
|
|
u64 *end;
|
|
|
|
|
2007-05-06 12:50:58 +00:00
|
|
|
for (pos = spt, end = pos + PAGE_SIZE / sizeof(u64); pos != end; pos++)
|
2008-05-20 13:21:13 +00:00
|
|
|
if (is_shadow_present_pte(*pos)) {
|
2008-03-03 20:59:56 +00:00
|
|
|
printk(KERN_ERR "%s: %p %llx\n", __func__,
|
2007-01-06 00:36:50 +00:00
|
|
|
pos, *pos);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
return 0;
|
2007-01-06 00:36:50 +00:00
|
|
|
}
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
return 1;
|
|
|
|
}
|
2007-04-25 06:17:25 +00:00
|
|
|
#endif
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
KVM: create aggregate kvm_total_used_mmu_pages value
Of slab shrinkers, the VM code says:
* Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
* querying the cache size, so a fastpath for that case is appropriate.
and it *means* it. Look at how it calls the shrinkers:
nr_before = (*shrinker->shrink)(0, gfp_mask);
shrink_ret = (*shrinker->shrink)(this_scan, gfp_mask);
So, if you do anything stupid in your shrinker, the VM will doubly
punish you.
The mmu_shrink() function takes the global kvm_lock, then acquires
every VM's kvm->mmu_lock in sequence. If we have 100 VMs, then
we're going to take 101 locks. We do it twice, so each call takes
202 locks. If we're under memory pressure, we can have each cpu
trying to do this. It can get really hairy, and we've seen lock
spinning in mmu_shrink() be the dominant entry in profiles.
This is guaranteed to optimize at least half of those lock
aquisitions away. It removes the need to take any of the locks
when simply trying to count objects.
A 'percpu_counter' can be a large object, but we only have one
of these for the entire system. There are not any better
alternatives at the moment, especially ones that handle CPU
hotplug.
Signed-off-by: Dave Hansen <dave@linux.vnet.ibm.com>
Signed-off-by: Tim Pepper <lnxninja@linux.vnet.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2010-08-20 01:11:37 +00:00
|
|
|
/*
|
|
|
|
* This value is the sum of all of the kvm instances's
|
|
|
|
* kvm->arch.n_used_mmu_pages values. We need a global,
|
|
|
|
* aggregate version in order to make the slab shrinker
|
|
|
|
* faster
|
|
|
|
*/
|
|
|
|
static inline void kvm_mod_used_mmu_pages(struct kvm *kvm, int nr)
|
|
|
|
{
|
|
|
|
kvm->arch.n_used_mmu_pages += nr;
|
|
|
|
percpu_counter_add(&kvm_total_used_mmu_pages, nr);
|
|
|
|
}
|
|
|
|
|
2013-01-30 14:45:05 +00:00
|
|
|
static void kvm_mmu_free_page(struct kvm_mmu_page *sp)
|
2007-01-06 00:36:49 +00:00
|
|
|
{
|
2013-10-02 14:56:16 +00:00
|
|
|
MMU_WARN_ON(!is_empty_shadow_page(sp->spt));
|
2010-06-04 13:53:54 +00:00
|
|
|
hlist_del(&sp->hash_link);
|
2011-07-11 19:27:14 +00:00
|
|
|
list_del(&sp->link);
|
|
|
|
free_page((unsigned long)sp->spt);
|
2013-01-30 14:45:05 +00:00
|
|
|
if (!sp->role.direct)
|
|
|
|
free_page((unsigned long)sp->gfns);
|
2010-05-13 02:06:02 +00:00
|
|
|
kmem_cache_free(mmu_page_header_cache, sp);
|
2007-01-06 00:36:49 +00:00
|
|
|
}
|
|
|
|
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
static unsigned kvm_page_table_hashfn(gfn_t gfn)
|
|
|
|
{
|
kvm: x86: reduce collisions in mmu_page_hash
When using two-dimensional paging, the mmu_page_hash (which provides
lookups for existing kvm_mmu_page structs), becomes imbalanced; with
too many collisions in buckets 0 and 512. This has been seen to cause
mmu_lock to be held for multiple milliseconds in kvm_mmu_get_page on
VMs with a large amount of RAM mapped with 4K pages.
The current hash function uses the lower 10 bits of gfn to index into
mmu_page_hash. When doing shadow paging, gfn is the address of the
guest page table being shadow. These tables are 4K-aligned, which
makes the low bits of gfn a good hash. However, with two-dimensional
paging, no guest page tables are being shadowed, so gfn is the base
address that is mapped by the table. Thus page tables (level=1) have
a 2MB aligned gfn, page directories (level=2) have a 1GB aligned gfn,
etc. This means hashes will only differ in their 10th bit.
hash_64() provides a better hash. For example, on a VM with ~200G
(99458 direct=1 kvm_mmu_page structs):
hash max_mmu_page_hash_collisions
--------------------------------------------
low 10 bits 49847
hash_64 105
perfect 97
While we're changing the hash, increase the table size by 4x to better
support large VMs (further reduces number of collisions in 200G VM to
29).
Note that hash_64() does not provide a good distribution prior to commit
ef703f49a6c5 ("Eliminate bad hash multipliers from hash_32() and
hash_64()").
Signed-off-by: David Matlack <dmatlack@google.com>
Change-Id: I5aa6b13c834722813c6cca46b8b1ed6f53368ade
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2016-12-19 21:58:25 +00:00
|
|
|
return hash_64(gfn, KVM_MMU_HASH_SHIFT);
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
}
|
|
|
|
|
2007-01-06 00:36:53 +00:00
|
|
|
static void mmu_page_add_parent_pte(struct kvm_vcpu *vcpu,
|
2007-11-21 13:28:32 +00:00
|
|
|
struct kvm_mmu_page *sp, u64 *parent_pte)
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
{
|
|
|
|
if (!parent_pte)
|
|
|
|
return;
|
|
|
|
|
2011-05-15 15:27:08 +00:00
|
|
|
pte_list_add(vcpu, parent_pte, &sp->parent_ptes);
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
}
|
|
|
|
|
2007-11-21 13:28:32 +00:00
|
|
|
static void mmu_page_remove_parent_pte(struct kvm_mmu_page *sp,
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
u64 *parent_pte)
|
|
|
|
{
|
2018-10-04 02:04:22 +00:00
|
|
|
__pte_list_remove(parent_pte, &sp->parent_ptes);
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
}
|
|
|
|
|
2011-05-15 15:28:29 +00:00
|
|
|
static void drop_parent_pte(struct kvm_mmu_page *sp,
|
|
|
|
u64 *parent_pte)
|
|
|
|
{
|
|
|
|
mmu_page_remove_parent_pte(sp, parent_pte);
|
2011-07-11 19:30:35 +00:00
|
|
|
mmu_spte_clear_no_track(parent_pte);
|
2011-05-15 15:28:29 +00:00
|
|
|
}
|
|
|
|
|
2015-11-20 08:46:29 +00:00
|
|
|
static struct kvm_mmu_page *kvm_mmu_alloc_page(struct kvm_vcpu *vcpu, int direct)
|
2008-09-23 16:18:36 +00:00
|
|
|
{
|
2011-05-15 15:27:08 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
2013-03-21 10:33:43 +00:00
|
|
|
|
2012-05-29 14:54:26 +00:00
|
|
|
sp = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_header_cache);
|
|
|
|
sp->spt = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_cache);
|
2011-05-15 15:27:08 +00:00
|
|
|
if (!direct)
|
2012-05-29 14:54:26 +00:00
|
|
|
sp->gfns = mmu_memory_cache_alloc(&vcpu->arch.mmu_page_cache);
|
2011-05-15 15:27:08 +00:00
|
|
|
set_page_private(virt_to_page(sp->spt), (unsigned long)sp);
|
2013-05-31 00:36:22 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* The active_mmu_pages list is the FIFO list, do not move the
|
|
|
|
* page until it is zapped. kvm_zap_obsolete_pages depends on
|
|
|
|
* this feature. See the comments in kvm_zap_obsolete_pages().
|
|
|
|
*/
|
2011-05-15 15:27:08 +00:00
|
|
|
list_add(&sp->link, &vcpu->kvm->arch.active_mmu_pages);
|
|
|
|
kvm_mod_used_mmu_pages(vcpu->kvm, +1);
|
|
|
|
return sp;
|
2008-09-23 16:18:36 +00:00
|
|
|
}
|
|
|
|
|
2011-05-15 15:27:08 +00:00
|
|
|
static void mark_unsync(u64 *spte);
|
2010-06-11 13:35:15 +00:00
|
|
|
static void kvm_mmu_mark_parents_unsync(struct kvm_mmu_page *sp)
|
2008-09-23 16:18:40 +00:00
|
|
|
{
|
2015-11-26 12:15:38 +00:00
|
|
|
u64 *sptep;
|
|
|
|
struct rmap_iterator iter;
|
|
|
|
|
|
|
|
for_each_rmap_spte(&sp->parent_ptes, &iter, sptep) {
|
|
|
|
mark_unsync(sptep);
|
|
|
|
}
|
2008-09-23 16:18:40 +00:00
|
|
|
}
|
|
|
|
|
2011-05-15 15:27:08 +00:00
|
|
|
static void mark_unsync(u64 *spte)
|
2008-09-23 16:18:40 +00:00
|
|
|
{
|
2011-05-15 15:27:08 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
2010-06-11 13:35:15 +00:00
|
|
|
unsigned int index;
|
2008-09-23 16:18:40 +00:00
|
|
|
|
2011-05-15 15:27:08 +00:00
|
|
|
sp = page_header(__pa(spte));
|
2010-06-11 13:35:15 +00:00
|
|
|
index = spte - sp->spt;
|
|
|
|
if (__test_and_set_bit(index, sp->unsync_child_bitmap))
|
2008-09-23 16:18:40 +00:00
|
|
|
return;
|
2010-06-11 13:35:15 +00:00
|
|
|
if (sp->unsync_children++)
|
2008-09-23 16:18:40 +00:00
|
|
|
return;
|
2010-06-11 13:35:15 +00:00
|
|
|
kvm_mmu_mark_parents_unsync(sp);
|
2008-09-23 16:18:40 +00:00
|
|
|
}
|
|
|
|
|
2008-09-23 16:18:33 +00:00
|
|
|
static int nonpaging_sync_page(struct kvm_vcpu *vcpu,
|
2010-11-19 09:04:03 +00:00
|
|
|
struct kvm_mmu_page *sp)
|
2008-09-23 16:18:33 +00:00
|
|
|
{
|
2016-02-24 10:07:14 +00:00
|
|
|
return 0;
|
2008-09-23 16:18:33 +00:00
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:16 +00:00
|
|
|
static void nonpaging_invlpg(struct kvm_vcpu *vcpu, gva_t gva, hpa_t root)
|
2008-09-23 16:18:35 +00:00
|
|
|
{
|
|
|
|
}
|
|
|
|
|
2011-03-09 07:43:51 +00:00
|
|
|
static void nonpaging_update_pte(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu_page *sp, u64 *spte,
|
2011-03-28 02:29:27 +00:00
|
|
|
const void *pte)
|
2011-03-09 07:43:51 +00:00
|
|
|
{
|
|
|
|
WARN_ON(1);
|
|
|
|
}
|
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
#define KVM_PAGE_ARRAY_NR 16
|
|
|
|
|
|
|
|
struct kvm_mmu_pages {
|
|
|
|
struct mmu_page_and_offset {
|
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
unsigned int idx;
|
|
|
|
} page[KVM_PAGE_ARRAY_NR];
|
|
|
|
unsigned int nr;
|
|
|
|
};
|
|
|
|
|
2009-02-21 01:19:13 +00:00
|
|
|
static int mmu_pages_add(struct kvm_mmu_pages *pvec, struct kvm_mmu_page *sp,
|
|
|
|
int idx)
|
2008-09-23 16:18:39 +00:00
|
|
|
{
|
2008-12-02 00:32:02 +00:00
|
|
|
int i;
|
2008-09-23 16:18:39 +00:00
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
if (sp->unsync)
|
|
|
|
for (i=0; i < pvec->nr; i++)
|
|
|
|
if (pvec->page[i].sp == sp)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
pvec->page[pvec->nr].sp = sp;
|
|
|
|
pvec->page[pvec->nr].idx = idx;
|
|
|
|
pvec->nr++;
|
|
|
|
return (pvec->nr == KVM_PAGE_ARRAY_NR);
|
|
|
|
}
|
|
|
|
|
2015-11-20 08:43:13 +00:00
|
|
|
static inline void clear_unsync_child_bit(struct kvm_mmu_page *sp, int idx)
|
|
|
|
{
|
|
|
|
--sp->unsync_children;
|
|
|
|
WARN_ON((int)sp->unsync_children < 0);
|
|
|
|
__clear_bit(idx, sp->unsync_child_bitmap);
|
|
|
|
}
|
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
static int __mmu_unsync_walk(struct kvm_mmu_page *sp,
|
|
|
|
struct kvm_mmu_pages *pvec)
|
|
|
|
{
|
|
|
|
int i, ret, nr_unsync_leaf = 0;
|
2008-09-23 16:18:39 +00:00
|
|
|
|
2011-11-29 05:02:45 +00:00
|
|
|
for_each_set_bit(i, sp->unsync_child_bitmap, 512) {
|
2010-06-11 13:34:04 +00:00
|
|
|
struct kvm_mmu_page *child;
|
2008-09-23 16:18:39 +00:00
|
|
|
u64 ent = sp->spt[i];
|
|
|
|
|
2015-11-20 08:43:13 +00:00
|
|
|
if (!is_shadow_present_pte(ent) || is_large_pte(ent)) {
|
|
|
|
clear_unsync_child_bit(sp, i);
|
|
|
|
continue;
|
|
|
|
}
|
2010-06-11 13:34:04 +00:00
|
|
|
|
|
|
|
child = page_header(ent & PT64_BASE_ADDR_MASK);
|
|
|
|
|
|
|
|
if (child->unsync_children) {
|
|
|
|
if (mmu_pages_add(pvec, child, i))
|
|
|
|
return -ENOSPC;
|
|
|
|
|
|
|
|
ret = __mmu_unsync_walk(child, pvec);
|
2015-11-20 08:43:13 +00:00
|
|
|
if (!ret) {
|
|
|
|
clear_unsync_child_bit(sp, i);
|
|
|
|
continue;
|
|
|
|
} else if (ret > 0) {
|
2010-06-11 13:34:04 +00:00
|
|
|
nr_unsync_leaf += ret;
|
2015-11-20 08:43:13 +00:00
|
|
|
} else
|
2010-06-11 13:34:04 +00:00
|
|
|
return ret;
|
|
|
|
} else if (child->unsync) {
|
|
|
|
nr_unsync_leaf++;
|
|
|
|
if (mmu_pages_add(pvec, child, i))
|
|
|
|
return -ENOSPC;
|
|
|
|
} else
|
2015-11-20 08:43:13 +00:00
|
|
|
clear_unsync_child_bit(sp, i);
|
2008-09-23 16:18:39 +00:00
|
|
|
}
|
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
return nr_unsync_leaf;
|
|
|
|
}
|
|
|
|
|
2016-02-24 08:46:06 +00:00
|
|
|
#define INVALID_INDEX (-1)
|
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
static int mmu_unsync_walk(struct kvm_mmu_page *sp,
|
|
|
|
struct kvm_mmu_pages *pvec)
|
|
|
|
{
|
2016-02-23 12:54:25 +00:00
|
|
|
pvec->nr = 0;
|
2008-12-02 00:32:02 +00:00
|
|
|
if (!sp->unsync_children)
|
|
|
|
return 0;
|
|
|
|
|
2016-02-24 08:46:06 +00:00
|
|
|
mmu_pages_add(pvec, sp, INVALID_INDEX);
|
2008-12-02 00:32:02 +00:00
|
|
|
return __mmu_unsync_walk(sp, pvec);
|
2008-09-23 16:18:39 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void kvm_unlink_unsync_page(struct kvm *kvm, struct kvm_mmu_page *sp)
|
|
|
|
{
|
|
|
|
WARN_ON(!sp->unsync);
|
2010-04-28 03:55:06 +00:00
|
|
|
trace_kvm_mmu_sync_page(sp);
|
2008-09-23 16:18:39 +00:00
|
|
|
sp->unsync = 0;
|
|
|
|
--kvm->stat.mmu_unsync;
|
|
|
|
}
|
|
|
|
|
2010-06-04 13:53:54 +00:00
|
|
|
static int kvm_mmu_prepare_zap_page(struct kvm *kvm, struct kvm_mmu_page *sp,
|
|
|
|
struct list_head *invalid_list);
|
|
|
|
static void kvm_mmu_commit_zap_page(struct kvm *kvm,
|
|
|
|
struct list_head *invalid_list);
|
2008-09-23 16:18:39 +00:00
|
|
|
|
2013-05-31 00:36:28 +00:00
|
|
|
/*
|
|
|
|
* NOTE: we should pay more attention on the zapped-obsolete page
|
|
|
|
* (is_obsolete_sp(sp) && sp->role.invalid) when you do hash list walk
|
|
|
|
* since it has been deleted from active_mmu_pages but still can be found
|
|
|
|
* at hast list.
|
|
|
|
*
|
2016-12-20 23:25:57 +00:00
|
|
|
* for_each_valid_sp() has skipped that kind of pages.
|
2013-05-31 00:36:28 +00:00
|
|
|
*/
|
2016-12-20 23:25:57 +00:00
|
|
|
#define for_each_valid_sp(_kvm, _sp, _gfn) \
|
2013-03-06 07:05:07 +00:00
|
|
|
hlist_for_each_entry(_sp, \
|
|
|
|
&(_kvm)->arch.mmu_page_hash[kvm_page_table_hashfn(_gfn)], hash_link) \
|
2016-12-20 23:25:57 +00:00
|
|
|
if (is_obsolete_sp((_kvm), (_sp)) || (_sp)->role.invalid) { \
|
|
|
|
} else
|
2013-03-06 07:05:07 +00:00
|
|
|
|
|
|
|
#define for_each_gfn_indirect_valid_sp(_kvm, _sp, _gfn) \
|
2016-12-20 23:25:57 +00:00
|
|
|
for_each_valid_sp(_kvm, _sp, _gfn) \
|
|
|
|
if ((_sp)->gfn != (_gfn) || (_sp)->role.direct) {} else
|
2010-06-04 13:53:07 +00:00
|
|
|
|
2010-06-11 13:30:36 +00:00
|
|
|
/* @sp->gfn should be write-protected at the call site */
|
2016-02-24 10:07:14 +00:00
|
|
|
static bool __kvm_sync_page(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
|
|
|
|
struct list_head *invalid_list)
|
2008-09-23 16:18:39 +00:00
|
|
|
{
|
2018-07-18 06:12:04 +00:00
|
|
|
if (sp->role.cr4_pae != !!is_pae(vcpu)
|
2018-10-08 19:28:05 +00:00
|
|
|
|| vcpu->arch.mmu->sync_page(vcpu, sp) == 0) {
|
2010-06-04 13:55:29 +00:00
|
|
|
kvm_mmu_prepare_zap_page(vcpu->kvm, sp, invalid_list);
|
2016-02-24 10:07:14 +00:00
|
|
|
return false;
|
2008-09-23 16:18:39 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 10:07:14 +00:00
|
|
|
return true;
|
2008-09-23 16:18:39 +00:00
|
|
|
}
|
|
|
|
|
2019-02-05 21:01:20 +00:00
|
|
|
static bool kvm_mmu_remote_flush_or_zap(struct kvm *kvm,
|
|
|
|
struct list_head *invalid_list,
|
|
|
|
bool remote_flush)
|
|
|
|
{
|
|
|
|
if (!remote_flush && !list_empty(invalid_list))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
if (!list_empty(invalid_list))
|
|
|
|
kvm_mmu_commit_zap_page(kvm, invalid_list);
|
|
|
|
else
|
|
|
|
kvm_flush_remote_tlbs(kvm);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2016-02-24 09:03:27 +00:00
|
|
|
static void kvm_mmu_flush_or_zap(struct kvm_vcpu *vcpu,
|
|
|
|
struct list_head *invalid_list,
|
|
|
|
bool remote_flush, bool local_flush)
|
2010-05-15 10:51:24 +00:00
|
|
|
{
|
2019-02-05 21:01:20 +00:00
|
|
|
if (kvm_mmu_remote_flush_or_zap(vcpu->kvm, invalid_list, remote_flush))
|
2016-02-24 09:03:27 +00:00
|
|
|
return;
|
2010-06-04 13:55:29 +00:00
|
|
|
|
2019-02-05 21:01:20 +00:00
|
|
|
if (local_flush)
|
2016-02-24 09:03:27 +00:00
|
|
|
kvm_make_request(KVM_REQ_TLB_FLUSH, vcpu);
|
2010-05-15 10:51:24 +00:00
|
|
|
}
|
|
|
|
|
2011-11-30 09:43:24 +00:00
|
|
|
#ifdef CONFIG_KVM_MMU_AUDIT
|
|
|
|
#include "mmu_audit.c"
|
|
|
|
#else
|
|
|
|
static void kvm_mmu_audit(struct kvm_vcpu *vcpu, int point) { }
|
|
|
|
static void mmu_audit_disable(void) { }
|
|
|
|
#endif
|
|
|
|
|
2016-03-25 13:19:38 +00:00
|
|
|
static bool is_obsolete_sp(struct kvm *kvm, struct kvm_mmu_page *sp)
|
|
|
|
{
|
|
|
|
return unlikely(sp->mmu_valid_gen != kvm->arch.mmu_valid_gen);
|
|
|
|
}
|
|
|
|
|
2016-02-24 10:07:14 +00:00
|
|
|
static bool kvm_sync_page(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
|
2010-06-04 13:55:29 +00:00
|
|
|
struct list_head *invalid_list)
|
2010-05-15 10:51:24 +00:00
|
|
|
{
|
2016-02-24 09:28:01 +00:00
|
|
|
kvm_unlink_unsync_page(vcpu->kvm, sp);
|
|
|
|
return __kvm_sync_page(vcpu, sp, invalid_list);
|
2010-05-15 10:51:24 +00:00
|
|
|
}
|
|
|
|
|
2010-05-24 07:41:33 +00:00
|
|
|
/* @gfn should be write-protected at the call site */
|
2016-02-24 10:26:10 +00:00
|
|
|
static bool kvm_sync_pages(struct kvm_vcpu *vcpu, gfn_t gfn,
|
|
|
|
struct list_head *invalid_list)
|
2010-05-24 07:41:33 +00:00
|
|
|
{
|
|
|
|
struct kvm_mmu_page *s;
|
2016-02-24 10:26:10 +00:00
|
|
|
bool ret = false;
|
2010-05-24 07:41:33 +00:00
|
|
|
|
hlist: drop the node parameter from iterators
I'm not sure why, but the hlist for each entry iterators were conceived
list_for_each_entry(pos, head, member)
The hlist ones were greedy and wanted an extra parameter:
hlist_for_each_entry(tpos, pos, head, member)
Why did they need an extra pos parameter? I'm not quite sure. Not only
they don't really need it, it also prevents the iterator from looking
exactly like the list iterator, which is unfortunate.
Besides the semantic patch, there was some manual work required:
- Fix up the actual hlist iterators in linux/list.h
- Fix up the declaration of other iterators based on the hlist ones.
- A very small amount of places were using the 'node' parameter, this
was modified to use 'obj->member' instead.
- Coccinelle didn't handle the hlist_for_each_entry_safe iterator
properly, so those had to be fixed up manually.
The semantic patch which is mostly the work of Peter Senna Tschudin is here:
@@
iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host;
type T;
expression a,c,d,e;
identifier b;
statement S;
@@
-T b;
<+... when != b
(
hlist_for_each_entry(a,
- b,
c, d) S
|
hlist_for_each_entry_continue(a,
- b,
c) S
|
hlist_for_each_entry_from(a,
- b,
c) S
|
hlist_for_each_entry_rcu(a,
- b,
c, d) S
|
hlist_for_each_entry_rcu_bh(a,
- b,
c, d) S
|
hlist_for_each_entry_continue_rcu_bh(a,
- b,
c) S
|
for_each_busy_worker(a, c,
- b,
d) S
|
ax25_uid_for_each(a,
- b,
c) S
|
ax25_for_each(a,
- b,
c) S
|
inet_bind_bucket_for_each(a,
- b,
c) S
|
sctp_for_each_hentry(a,
- b,
c) S
|
sk_for_each(a,
- b,
c) S
|
sk_for_each_rcu(a,
- b,
c) S
|
sk_for_each_from
-(a, b)
+(a)
S
+ sk_for_each_from(a) S
|
sk_for_each_safe(a,
- b,
c, d) S
|
sk_for_each_bound(a,
- b,
c) S
|
hlist_for_each_entry_safe(a,
- b,
c, d, e) S
|
hlist_for_each_entry_continue_rcu(a,
- b,
c) S
|
nr_neigh_for_each(a,
- b,
c) S
|
nr_neigh_for_each_safe(a,
- b,
c, d) S
|
nr_node_for_each(a,
- b,
c) S
|
nr_node_for_each_safe(a,
- b,
c, d) S
|
- for_each_gfn_sp(a, c, d, b) S
+ for_each_gfn_sp(a, c, d) S
|
- for_each_gfn_indirect_valid_sp(a, c, d, b) S
+ for_each_gfn_indirect_valid_sp(a, c, d) S
|
for_each_host(a,
- b,
c) S
|
for_each_host_safe(a,
- b,
c, d) S
|
for_each_mesh_entry(a,
- b,
c, d) S
)
...+>
[akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c]
[akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c]
[akpm@linux-foundation.org: checkpatch fixes]
[akpm@linux-foundation.org: fix warnings]
[akpm@linux-foudnation.org: redo intrusive kvm changes]
Tested-by: Peter Senna Tschudin <peter.senna@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:06:00 +00:00
|
|
|
for_each_gfn_indirect_valid_sp(vcpu->kvm, s, gfn) {
|
2010-06-04 13:53:07 +00:00
|
|
|
if (!s->unsync)
|
2010-05-24 07:41:33 +00:00
|
|
|
continue;
|
|
|
|
|
|
|
|
WARN_ON(s->role.level != PT_PAGE_TABLE_LEVEL);
|
2016-02-24 10:26:10 +00:00
|
|
|
ret |= kvm_sync_page(vcpu, s, invalid_list);
|
2010-05-24 07:41:33 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 10:26:10 +00:00
|
|
|
return ret;
|
2010-05-24 07:41:33 +00:00
|
|
|
}
|
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
struct mmu_page_path {
|
2017-08-24 12:27:54 +00:00
|
|
|
struct kvm_mmu_page *parent[PT64_ROOT_MAX_LEVEL];
|
|
|
|
unsigned int idx[PT64_ROOT_MAX_LEVEL];
|
2008-09-23 16:18:39 +00:00
|
|
|
};
|
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
#define for_each_sp(pvec, sp, parents, i) \
|
2016-02-23 12:54:25 +00:00
|
|
|
for (i = mmu_pages_first(&pvec, &parents); \
|
2008-12-02 00:32:02 +00:00
|
|
|
i < pvec.nr && ({ sp = pvec.page[i].sp; 1;}); \
|
|
|
|
i = mmu_pages_next(&pvec, &parents, i))
|
|
|
|
|
2009-02-21 01:19:13 +00:00
|
|
|
static int mmu_pages_next(struct kvm_mmu_pages *pvec,
|
|
|
|
struct mmu_page_path *parents,
|
|
|
|
int i)
|
2008-12-02 00:32:02 +00:00
|
|
|
{
|
|
|
|
int n;
|
|
|
|
|
|
|
|
for (n = i+1; n < pvec->nr; n++) {
|
|
|
|
struct kvm_mmu_page *sp = pvec->page[n].sp;
|
2016-02-23 12:54:25 +00:00
|
|
|
unsigned idx = pvec->page[n].idx;
|
|
|
|
int level = sp->role.level;
|
2008-12-02 00:32:02 +00:00
|
|
|
|
2016-02-23 12:54:25 +00:00
|
|
|
parents->idx[level-1] = idx;
|
|
|
|
if (level == PT_PAGE_TABLE_LEVEL)
|
|
|
|
break;
|
2008-12-02 00:32:02 +00:00
|
|
|
|
2016-02-23 12:54:25 +00:00
|
|
|
parents->parent[level-2] = sp;
|
2008-12-02 00:32:02 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return n;
|
|
|
|
}
|
|
|
|
|
2016-02-23 12:54:25 +00:00
|
|
|
static int mmu_pages_first(struct kvm_mmu_pages *pvec,
|
|
|
|
struct mmu_page_path *parents)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
int level;
|
|
|
|
|
|
|
|
if (pvec->nr == 0)
|
|
|
|
return 0;
|
|
|
|
|
2016-02-24 08:46:06 +00:00
|
|
|
WARN_ON(pvec->page[0].idx != INVALID_INDEX);
|
|
|
|
|
2016-02-23 12:54:25 +00:00
|
|
|
sp = pvec->page[0].sp;
|
|
|
|
level = sp->role.level;
|
|
|
|
WARN_ON(level == PT_PAGE_TABLE_LEVEL);
|
|
|
|
|
|
|
|
parents->parent[level-2] = sp;
|
|
|
|
|
|
|
|
/* Also set up a sentinel. Further entries in pvec are all
|
|
|
|
* children of sp, so this element is never overwritten.
|
|
|
|
*/
|
|
|
|
parents->parent[level-1] = NULL;
|
|
|
|
return mmu_pages_next(pvec, parents, 0);
|
|
|
|
}
|
|
|
|
|
2009-02-21 01:19:13 +00:00
|
|
|
static void mmu_pages_clear_parents(struct mmu_page_path *parents)
|
2008-09-23 16:18:39 +00:00
|
|
|
{
|
2008-12-02 00:32:02 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
unsigned int level = 0;
|
|
|
|
|
|
|
|
do {
|
|
|
|
unsigned int idx = parents->idx[level];
|
|
|
|
sp = parents->parent[level];
|
|
|
|
if (!sp)
|
|
|
|
return;
|
|
|
|
|
2016-02-24 08:46:06 +00:00
|
|
|
WARN_ON(idx == INVALID_INDEX);
|
2015-11-20 08:43:13 +00:00
|
|
|
clear_unsync_child_bit(sp, idx);
|
2008-12-02 00:32:02 +00:00
|
|
|
level++;
|
2016-02-23 12:54:25 +00:00
|
|
|
} while (!sp->unsync_children);
|
2008-12-02 00:32:02 +00:00
|
|
|
}
|
2008-09-23 16:18:39 +00:00
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
static void mmu_sync_children(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu_page *parent)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
struct mmu_page_path parents;
|
|
|
|
struct kvm_mmu_pages pages;
|
2010-06-04 13:55:29 +00:00
|
|
|
LIST_HEAD(invalid_list);
|
2016-02-25 09:47:38 +00:00
|
|
|
bool flush = false;
|
2008-12-02 00:32:02 +00:00
|
|
|
|
|
|
|
while (mmu_unsync_walk(parent, &pages)) {
|
2012-06-20 07:56:53 +00:00
|
|
|
bool protected = false;
|
2008-12-02 00:32:03 +00:00
|
|
|
|
|
|
|
for_each_sp(pages, sp, parents, i)
|
2015-04-08 13:39:23 +00:00
|
|
|
protected |= rmap_write_protect(vcpu, sp->gfn);
|
2008-12-02 00:32:03 +00:00
|
|
|
|
2016-02-25 09:47:38 +00:00
|
|
|
if (protected) {
|
2008-12-02 00:32:03 +00:00
|
|
|
kvm_flush_remote_tlbs(vcpu->kvm);
|
2016-02-25 09:47:38 +00:00
|
|
|
flush = false;
|
|
|
|
}
|
2008-12-02 00:32:03 +00:00
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
for_each_sp(pages, sp, parents, i) {
|
2016-02-24 10:07:14 +00:00
|
|
|
flush |= kvm_sync_page(vcpu, sp, &invalid_list);
|
2008-12-02 00:32:02 +00:00
|
|
|
mmu_pages_clear_parents(&parents);
|
|
|
|
}
|
2016-02-25 09:47:38 +00:00
|
|
|
if (need_resched() || spin_needbreak(&vcpu->kvm->mmu_lock)) {
|
|
|
|
kvm_mmu_flush_or_zap(vcpu, &invalid_list, false, flush);
|
|
|
|
cond_resched_lock(&vcpu->kvm->mmu_lock);
|
|
|
|
flush = false;
|
|
|
|
}
|
2008-12-02 00:32:02 +00:00
|
|
|
}
|
2016-02-25 09:47:38 +00:00
|
|
|
|
|
|
|
kvm_mmu_flush_or_zap(vcpu, &invalid_list, false, flush);
|
2008-09-23 16:18:39 +00:00
|
|
|
}
|
|
|
|
|
KVM: MMU: improve write flooding detected
Detecting write-flooding does not work well, when we handle page written, if
the last speculative spte is not accessed, we treat the page is
write-flooding, however, we can speculative spte on many path, such as pte
prefetch, page synced, that means the last speculative spte may be not point
to the written page and the written page can be accessed via other sptes, so
depends on the Accessed bit of the last speculative spte is not enough
Instead of detected page accessed, we can detect whether the spte is accessed
after it is written, if the spte is not accessed but it is written frequently,
we treat is not a page table or it not used for a long time
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2011-09-22 08:58:36 +00:00
|
|
|
static void __clear_sp_write_flooding_count(struct kvm_mmu_page *sp)
|
|
|
|
{
|
2016-02-24 09:51:12 +00:00
|
|
|
atomic_set(&sp->write_flooding_count, 0);
|
KVM: MMU: improve write flooding detected
Detecting write-flooding does not work well, when we handle page written, if
the last speculative spte is not accessed, we treat the page is
write-flooding, however, we can speculative spte on many path, such as pte
prefetch, page synced, that means the last speculative spte may be not point
to the written page and the written page can be accessed via other sptes, so
depends on the Accessed bit of the last speculative spte is not enough
Instead of detected page accessed, we can detect whether the spte is accessed
after it is written, if the spte is not accessed but it is written frequently,
we treat is not a page table or it not used for a long time
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2011-09-22 08:58:36 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void clear_sp_write_flooding_count(u64 *spte)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp = page_header(__pa(spte));
|
|
|
|
|
|
|
|
__clear_sp_write_flooding_count(sp);
|
|
|
|
}
|
|
|
|
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
static struct kvm_mmu_page *kvm_mmu_get_page(struct kvm_vcpu *vcpu,
|
|
|
|
gfn_t gfn,
|
|
|
|
gva_t gaddr,
|
|
|
|
unsigned level,
|
2009-01-11 11:02:10 +00:00
|
|
|
int direct,
|
2015-11-26 12:16:35 +00:00
|
|
|
unsigned access)
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
{
|
|
|
|
union kvm_mmu_page_role role;
|
|
|
|
unsigned quadrant;
|
2010-05-24 07:41:33 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
bool need_sync = false;
|
2016-02-24 10:26:10 +00:00
|
|
|
bool flush = false;
|
2016-12-20 23:25:57 +00:00
|
|
|
int collisions = 0;
|
2016-02-24 10:26:10 +00:00
|
|
|
LIST_HEAD(invalid_list);
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
|
2018-10-08 19:28:10 +00:00
|
|
|
role = vcpu->arch.mmu->mmu_role.base;
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
role.level = level;
|
2009-01-11 11:02:10 +00:00
|
|
|
role.direct = direct;
|
2010-03-14 08:16:40 +00:00
|
|
|
if (role.direct)
|
2010-04-14 16:20:03 +00:00
|
|
|
role.cr4_pae = 0;
|
2007-12-09 15:00:02 +00:00
|
|
|
role.access = access;
|
2018-10-08 19:28:05 +00:00
|
|
|
if (!vcpu->arch.mmu->direct_map
|
|
|
|
&& vcpu->arch.mmu->root_level <= PT32_ROOT_LEVEL) {
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
quadrant = gaddr >> (PAGE_SHIFT + (PT64_PT_BITS * level));
|
|
|
|
quadrant &= (1 << ((PT32_PT_BITS - PT64_PT_BITS) * level)) - 1;
|
|
|
|
role.quadrant = quadrant;
|
|
|
|
}
|
2016-12-20 23:25:57 +00:00
|
|
|
for_each_valid_sp(vcpu->kvm, sp, gfn) {
|
|
|
|
if (sp->gfn != gfn) {
|
|
|
|
collisions++;
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2010-06-04 13:53:07 +00:00
|
|
|
if (!need_sync && sp->unsync)
|
|
|
|
need_sync = true;
|
2008-09-23 16:18:39 +00:00
|
|
|
|
2010-06-04 13:53:07 +00:00
|
|
|
if (sp->role.word != role.word)
|
|
|
|
continue;
|
2008-09-23 16:18:39 +00:00
|
|
|
|
2016-02-24 10:26:10 +00:00
|
|
|
if (sp->unsync) {
|
|
|
|
/* The page is good, but __kvm_sync_page might still end
|
|
|
|
* up zapping it. If so, break in order to rebuild it.
|
|
|
|
*/
|
|
|
|
if (!__kvm_sync_page(vcpu, sp, &invalid_list))
|
|
|
|
break;
|
|
|
|
|
|
|
|
WARN_ON(!list_empty(&invalid_list));
|
|
|
|
kvm_make_request(KVM_REQ_TLB_FLUSH, vcpu);
|
|
|
|
}
|
KVM: MMU: don't write-protect if have new mapping to unsync page
Two cases maybe happen in kvm_mmu_get_page() function:
- one case is, the goal sp is already in cache, if the sp is unsync,
we only need update it to assure this mapping is valid, but not
mark it sync and not write-protect sp->gfn since it not broke unsync
rule(one shadow page for a gfn)
- another case is, the goal sp not existed, we need create a new sp
for gfn, i.e, gfn (may)has another shadow page, to keep unsync rule,
we should sync(mark sync and write-protect) gfn's unsync shadow page.
After enabling multiple unsync shadows, we sync those shadow pages
only when the new sp not allow to become unsync(also for the unsyc
rule, the new rule is: allow all pte page become unsync)
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2010-05-15 10:52:34 +00:00
|
|
|
|
2015-11-26 12:14:34 +00:00
|
|
|
if (sp->unsync_children)
|
2010-05-10 09:34:53 +00:00
|
|
|
kvm_make_request(KVM_REQ_MMU_SYNC, vcpu);
|
KVM: MMU: don't write-protect if have new mapping to unsync page
Two cases maybe happen in kvm_mmu_get_page() function:
- one case is, the goal sp is already in cache, if the sp is unsync,
we only need update it to assure this mapping is valid, but not
mark it sync and not write-protect sp->gfn since it not broke unsync
rule(one shadow page for a gfn)
- another case is, the goal sp not existed, we need create a new sp
for gfn, i.e, gfn (may)has another shadow page, to keep unsync rule,
we should sync(mark sync and write-protect) gfn's unsync shadow page.
After enabling multiple unsync shadows, we sync those shadow pages
only when the new sp not allow to become unsync(also for the unsyc
rule, the new rule is: allow all pte page become unsync)
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2010-05-15 10:52:34 +00:00
|
|
|
|
KVM: MMU: improve write flooding detected
Detecting write-flooding does not work well, when we handle page written, if
the last speculative spte is not accessed, we treat the page is
write-flooding, however, we can speculative spte on many path, such as pte
prefetch, page synced, that means the last speculative spte may be not point
to the written page and the written page can be accessed via other sptes, so
depends on the Accessed bit of the last speculative spte is not enough
Instead of detected page accessed, we can detect whether the spte is accessed
after it is written, if the spte is not accessed but it is written frequently,
we treat is not a page table or it not used for a long time
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2011-09-22 08:58:36 +00:00
|
|
|
__clear_sp_write_flooding_count(sp);
|
2010-06-04 13:53:07 +00:00
|
|
|
trace_kvm_mmu_get_page(sp, false);
|
2016-12-20 23:25:57 +00:00
|
|
|
goto out;
|
2010-06-04 13:53:07 +00:00
|
|
|
}
|
2015-11-20 08:46:29 +00:00
|
|
|
|
2007-12-18 17:47:18 +00:00
|
|
|
++vcpu->kvm->stat.mmu_cache_miss;
|
2015-11-20 08:46:29 +00:00
|
|
|
|
|
|
|
sp = kvm_mmu_alloc_page(vcpu, direct);
|
|
|
|
|
2007-11-21 13:28:32 +00:00
|
|
|
sp->gfn = gfn;
|
|
|
|
sp->role = role;
|
2010-06-04 13:53:07 +00:00
|
|
|
hlist_add_head(&sp->hash_link,
|
|
|
|
&vcpu->kvm->arch.mmu_page_hash[kvm_page_table_hashfn(gfn)]);
|
2009-01-11 11:02:10 +00:00
|
|
|
if (!direct) {
|
2016-02-24 09:51:14 +00:00
|
|
|
/*
|
|
|
|
* we should do write protection before syncing pages
|
|
|
|
* otherwise the content of the synced shadow page may
|
|
|
|
* be inconsistent with guest page table.
|
|
|
|
*/
|
|
|
|
account_shadowed(vcpu->kvm, sp);
|
|
|
|
if (level == PT_PAGE_TABLE_LEVEL &&
|
|
|
|
rmap_write_protect(vcpu, gfn))
|
2018-12-06 13:21:09 +00:00
|
|
|
kvm_flush_remote_tlbs_with_address(vcpu->kvm, gfn, 1);
|
2010-05-24 07:41:33 +00:00
|
|
|
|
|
|
|
if (level > PT_PAGE_TABLE_LEVEL && need_sync)
|
2016-02-24 10:26:10 +00:00
|
|
|
flush |= kvm_sync_pages(vcpu, gfn, &invalid_list);
|
2008-09-23 16:18:39 +00:00
|
|
|
}
|
2013-05-31 00:36:22 +00:00
|
|
|
sp->mmu_valid_gen = vcpu->kvm->arch.mmu_valid_gen;
|
2015-12-18 09:54:49 +00:00
|
|
|
clear_page(sp->spt);
|
2009-07-06 12:58:14 +00:00
|
|
|
trace_kvm_mmu_get_page(sp, true);
|
2016-02-24 10:26:10 +00:00
|
|
|
|
|
|
|
kvm_mmu_flush_or_zap(vcpu, &invalid_list, false, flush);
|
2016-12-20 23:25:57 +00:00
|
|
|
out:
|
|
|
|
if (collisions > vcpu->kvm->stat.max_mmu_page_hash_collisions)
|
|
|
|
vcpu->kvm->stat.max_mmu_page_hash_collisions = collisions;
|
2007-11-21 13:28:32 +00:00
|
|
|
return sp;
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:16 +00:00
|
|
|
static void shadow_walk_init_using_root(struct kvm_shadow_walk_iterator *iterator,
|
|
|
|
struct kvm_vcpu *vcpu, hpa_t root,
|
|
|
|
u64 addr)
|
2008-12-25 12:39:47 +00:00
|
|
|
{
|
|
|
|
iterator->addr = addr;
|
2018-06-27 21:59:16 +00:00
|
|
|
iterator->shadow_addr = root;
|
2018-10-08 19:28:05 +00:00
|
|
|
iterator->level = vcpu->arch.mmu->shadow_root_level;
|
2010-09-10 15:31:00 +00:00
|
|
|
|
2017-08-24 12:27:54 +00:00
|
|
|
if (iterator->level == PT64_ROOT_4LEVEL &&
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->root_level < PT64_ROOT_4LEVEL &&
|
|
|
|
!vcpu->arch.mmu->direct_map)
|
2010-09-10 15:31:00 +00:00
|
|
|
--iterator->level;
|
|
|
|
|
2008-12-25 12:39:47 +00:00
|
|
|
if (iterator->level == PT32E_ROOT_LEVEL) {
|
2018-06-27 21:59:16 +00:00
|
|
|
/*
|
|
|
|
* prev_root is currently only used for 64-bit hosts. So only
|
|
|
|
* the active root_hpa is valid here.
|
|
|
|
*/
|
2018-10-08 19:28:05 +00:00
|
|
|
BUG_ON(root != vcpu->arch.mmu->root_hpa);
|
2018-06-27 21:59:16 +00:00
|
|
|
|
2008-12-25 12:39:47 +00:00
|
|
|
iterator->shadow_addr
|
2018-10-08 19:28:05 +00:00
|
|
|
= vcpu->arch.mmu->pae_root[(addr >> 30) & 3];
|
2008-12-25 12:39:47 +00:00
|
|
|
iterator->shadow_addr &= PT64_BASE_ADDR_MASK;
|
|
|
|
--iterator->level;
|
|
|
|
if (!iterator->shadow_addr)
|
|
|
|
iterator->level = 0;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:16 +00:00
|
|
|
static void shadow_walk_init(struct kvm_shadow_walk_iterator *iterator,
|
|
|
|
struct kvm_vcpu *vcpu, u64 addr)
|
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
shadow_walk_init_using_root(iterator, vcpu, vcpu->arch.mmu->root_hpa,
|
2018-06-27 21:59:16 +00:00
|
|
|
addr);
|
|
|
|
}
|
|
|
|
|
2008-12-25 12:39:47 +00:00
|
|
|
static bool shadow_walk_okay(struct kvm_shadow_walk_iterator *iterator)
|
|
|
|
{
|
|
|
|
if (iterator->level < PT_PAGE_TABLE_LEVEL)
|
|
|
|
return false;
|
2009-06-11 15:07:41 +00:00
|
|
|
|
2008-12-25 12:39:47 +00:00
|
|
|
iterator->index = SHADOW_PT_INDEX(iterator->addr, iterator->level);
|
|
|
|
iterator->sptep = ((u64 *)__va(iterator->shadow_addr)) + iterator->index;
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2011-07-11 19:32:13 +00:00
|
|
|
static void __shadow_walk_next(struct kvm_shadow_walk_iterator *iterator,
|
|
|
|
u64 spte)
|
2008-12-25 12:39:47 +00:00
|
|
|
{
|
2011-07-11 19:32:13 +00:00
|
|
|
if (is_last_spte(spte, iterator->level)) {
|
2011-07-11 19:21:17 +00:00
|
|
|
iterator->level = 0;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2011-07-11 19:32:13 +00:00
|
|
|
iterator->shadow_addr = spte & PT64_BASE_ADDR_MASK;
|
2008-12-25 12:39:47 +00:00
|
|
|
--iterator->level;
|
|
|
|
}
|
|
|
|
|
2011-07-11 19:32:13 +00:00
|
|
|
static void shadow_walk_next(struct kvm_shadow_walk_iterator *iterator)
|
|
|
|
{
|
2017-08-24 18:51:23 +00:00
|
|
|
__shadow_walk_next(iterator, *iterator->sptep);
|
2011-07-11 19:32:13 +00:00
|
|
|
}
|
|
|
|
|
2015-11-26 12:14:34 +00:00
|
|
|
static void link_shadow_page(struct kvm_vcpu *vcpu, u64 *sptep,
|
|
|
|
struct kvm_mmu_page *sp)
|
2010-07-13 11:27:04 +00:00
|
|
|
{
|
|
|
|
u64 spte;
|
|
|
|
|
2016-07-12 22:18:49 +00:00
|
|
|
BUILD_BUG_ON(VMX_EPT_WRITABLE_MASK != PT_WRITABLE_MASK);
|
2013-08-05 08:07:13 +00:00
|
|
|
|
2016-07-12 22:18:49 +00:00
|
|
|
spte = __pa(sp->spt) | shadow_present_mask | PT_WRITABLE_MASK |
|
2017-07-17 21:10:27 +00:00
|
|
|
shadow_user_mask | shadow_x_mask | shadow_me_mask;
|
2017-07-01 00:26:31 +00:00
|
|
|
|
|
|
|
if (sp_ad_disabled(sp))
|
|
|
|
spte |= shadow_acc_track_value;
|
|
|
|
else
|
|
|
|
spte |= shadow_accessed_mask;
|
2013-02-05 07:28:02 +00:00
|
|
|
|
2011-07-11 19:30:35 +00:00
|
|
|
mmu_spte_set(sptep, spte);
|
2015-11-26 12:14:34 +00:00
|
|
|
|
|
|
|
mmu_page_add_parent_pte(vcpu, sp, sptep);
|
|
|
|
|
|
|
|
if (sp->unsync_children || sp->unsync)
|
|
|
|
mark_unsync(sptep);
|
2010-07-13 11:27:04 +00:00
|
|
|
}
|
|
|
|
|
2010-07-13 11:27:07 +00:00
|
|
|
static void validate_direct_spte(struct kvm_vcpu *vcpu, u64 *sptep,
|
|
|
|
unsigned direct_access)
|
|
|
|
{
|
|
|
|
if (is_shadow_present_pte(*sptep) && !is_large_pte(*sptep)) {
|
|
|
|
struct kvm_mmu_page *child;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* For the direct sp, if the guest pte's dirty bit
|
|
|
|
* changed form clean to dirty, it will corrupt the
|
|
|
|
* sp's access: allow writable in the read-only sp,
|
|
|
|
* so we should update the spte at this point to get
|
|
|
|
* a new sp with the correct access.
|
|
|
|
*/
|
|
|
|
child = page_header(*sptep & PT64_BASE_ADDR_MASK);
|
|
|
|
if (child->role.access == direct_access)
|
|
|
|
return;
|
|
|
|
|
2011-05-15 15:28:29 +00:00
|
|
|
drop_parent_pte(child, sptep);
|
2018-12-06 13:21:09 +00:00
|
|
|
kvm_flush_remote_tlbs_with_address(vcpu->kvm, child->gfn, 1);
|
2010-07-13 11:27:07 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-09-22 08:56:06 +00:00
|
|
|
static bool mmu_page_zap_pte(struct kvm *kvm, struct kvm_mmu_page *sp,
|
2011-05-15 15:27:52 +00:00
|
|
|
u64 *spte)
|
|
|
|
{
|
|
|
|
u64 pte;
|
|
|
|
struct kvm_mmu_page *child;
|
|
|
|
|
|
|
|
pte = *spte;
|
|
|
|
if (is_shadow_present_pte(pte)) {
|
2011-09-22 08:56:06 +00:00
|
|
|
if (is_last_spte(pte, sp->role.level)) {
|
2011-07-11 19:28:04 +00:00
|
|
|
drop_spte(kvm, spte);
|
2011-09-22 08:56:06 +00:00
|
|
|
if (is_large_pte(pte))
|
|
|
|
--kvm->stat.lpages;
|
|
|
|
} else {
|
2011-05-15 15:27:52 +00:00
|
|
|
child = page_header(pte & PT64_BASE_ADDR_MASK);
|
2011-05-15 15:28:29 +00:00
|
|
|
drop_parent_pte(child, spte);
|
2011-05-15 15:27:52 +00:00
|
|
|
}
|
2011-09-22 08:56:06 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (is_mmio_spte(pte))
|
2011-07-11 19:33:44 +00:00
|
|
|
mmu_spte_clear_no_track(spte);
|
2011-07-11 19:28:04 +00:00
|
|
|
|
2011-09-22 08:56:06 +00:00
|
|
|
return false;
|
2011-05-15 15:27:52 +00:00
|
|
|
}
|
|
|
|
|
2007-07-17 10:04:56 +00:00
|
|
|
static void kvm_mmu_page_unlink_children(struct kvm *kvm,
|
2007-11-21 13:28:32 +00:00
|
|
|
struct kvm_mmu_page *sp)
|
2007-01-06 00:36:45 +00:00
|
|
|
{
|
2007-01-06 00:36:46 +00:00
|
|
|
unsigned i;
|
|
|
|
|
2011-05-15 15:27:52 +00:00
|
|
|
for (i = 0; i < PT64_ENT_PER_PAGE; ++i)
|
|
|
|
mmu_page_zap_pte(kvm, sp, sp->spt + i);
|
2007-01-06 00:36:45 +00:00
|
|
|
}
|
|
|
|
|
2008-07-11 14:59:46 +00:00
|
|
|
static void kvm_mmu_unlink_parents(struct kvm *kvm, struct kvm_mmu_page *sp)
|
2007-01-06 00:36:45 +00:00
|
|
|
{
|
2012-03-21 14:50:34 +00:00
|
|
|
u64 *sptep;
|
|
|
|
struct rmap_iterator iter;
|
2007-01-06 00:36:45 +00:00
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
while ((sptep = rmap_get_first(&sp->parent_ptes, &iter)))
|
2012-03-21 14:50:34 +00:00
|
|
|
drop_parent_pte(sp, sptep);
|
2008-07-11 14:59:46 +00:00
|
|
|
}
|
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
static int mmu_zap_unsync_children(struct kvm *kvm,
|
2010-06-04 13:53:54 +00:00
|
|
|
struct kvm_mmu_page *parent,
|
|
|
|
struct list_head *invalid_list)
|
2008-09-23 16:18:39 +00:00
|
|
|
{
|
2008-12-02 00:32:02 +00:00
|
|
|
int i, zapped = 0;
|
|
|
|
struct mmu_page_path parents;
|
|
|
|
struct kvm_mmu_pages pages;
|
2008-09-23 16:18:39 +00:00
|
|
|
|
2008-12-02 00:32:02 +00:00
|
|
|
if (parent->role.level == PT_PAGE_TABLE_LEVEL)
|
2008-09-23 16:18:39 +00:00
|
|
|
return 0;
|
2008-12-02 00:32:02 +00:00
|
|
|
|
|
|
|
while (mmu_unsync_walk(parent, &pages)) {
|
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
|
|
|
|
for_each_sp(pages, sp, parents, i) {
|
2010-06-04 13:53:54 +00:00
|
|
|
kvm_mmu_prepare_zap_page(kvm, sp, invalid_list);
|
2008-12-02 00:32:02 +00:00
|
|
|
mmu_pages_clear_parents(&parents);
|
2010-04-16 08:34:42 +00:00
|
|
|
zapped++;
|
2008-12-02 00:32:02 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return zapped;
|
2008-09-23 16:18:39 +00:00
|
|
|
}
|
|
|
|
|
2010-06-04 13:53:54 +00:00
|
|
|
static int kvm_mmu_prepare_zap_page(struct kvm *kvm, struct kvm_mmu_page *sp,
|
|
|
|
struct list_head *invalid_list)
|
2008-07-11 14:59:46 +00:00
|
|
|
{
|
2008-09-23 16:18:39 +00:00
|
|
|
int ret;
|
2009-07-06 12:58:14 +00:00
|
|
|
|
2010-06-04 13:53:54 +00:00
|
|
|
trace_kvm_mmu_prepare_zap_page(sp);
|
2008-07-11 14:59:46 +00:00
|
|
|
++kvm->stat.mmu_shadow_zapped;
|
2010-06-04 13:53:54 +00:00
|
|
|
ret = mmu_zap_unsync_children(kvm, sp, invalid_list);
|
2007-11-21 13:28:32 +00:00
|
|
|
kvm_mmu_page_unlink_children(kvm, sp);
|
2008-07-11 14:59:46 +00:00
|
|
|
kvm_mmu_unlink_parents(kvm, sp);
|
2013-05-31 00:36:22 +00:00
|
|
|
|
2009-01-11 11:02:10 +00:00
|
|
|
if (!sp->role.invalid && !sp->role.direct)
|
2015-05-19 14:29:22 +00:00
|
|
|
unaccount_shadowed(kvm, sp);
|
2013-05-31 00:36:22 +00:00
|
|
|
|
2008-09-23 16:18:39 +00:00
|
|
|
if (sp->unsync)
|
|
|
|
kvm_unlink_unsync_page(kvm, sp);
|
2007-11-21 13:28:32 +00:00
|
|
|
if (!sp->root_count) {
|
2010-05-05 01:03:49 +00:00
|
|
|
/* Count self */
|
|
|
|
ret++;
|
2010-06-04 13:53:54 +00:00
|
|
|
list_move(&sp->link, invalid_list);
|
2011-07-11 19:26:40 +00:00
|
|
|
kvm_mod_used_mmu_pages(kvm, -1);
|
2008-02-20 19:47:24 +00:00
|
|
|
} else {
|
2008-07-11 15:07:26 +00:00
|
|
|
list_move(&sp->link, &kvm->arch.active_mmu_pages);
|
2013-05-31 00:36:30 +00:00
|
|
|
|
2019-02-05 21:01:25 +00:00
|
|
|
if (!sp->role.invalid)
|
2013-05-31 00:36:30 +00:00
|
|
|
kvm_reload_remote_mmus(kvm);
|
2008-02-20 19:47:24 +00:00
|
|
|
}
|
2010-06-04 13:53:54 +00:00
|
|
|
|
|
|
|
sp->role.invalid = 1;
|
2008-09-23 16:18:39 +00:00
|
|
|
return ret;
|
2007-01-06 00:36:45 +00:00
|
|
|
}
|
|
|
|
|
2010-06-04 13:53:54 +00:00
|
|
|
static void kvm_mmu_commit_zap_page(struct kvm *kvm,
|
|
|
|
struct list_head *invalid_list)
|
|
|
|
{
|
2013-03-06 07:05:52 +00:00
|
|
|
struct kvm_mmu_page *sp, *nsp;
|
2010-06-04 13:53:54 +00:00
|
|
|
|
|
|
|
if (list_empty(invalid_list))
|
|
|
|
return;
|
|
|
|
|
2012-05-14 12:44:06 +00:00
|
|
|
/*
|
2016-03-13 03:10:24 +00:00
|
|
|
* We need to make sure everyone sees our modifications to
|
|
|
|
* the page tables and see changes to vcpu->mode here. The barrier
|
|
|
|
* in the kvm_flush_remote_tlbs() achieves this. This pairs
|
|
|
|
* with vcpu_enter_guest and walk_shadow_page_lockless_begin/end.
|
|
|
|
*
|
|
|
|
* In addition, kvm_flush_remote_tlbs waits for all vcpus to exit
|
|
|
|
* guest mode and/or lockless shadow page table walks.
|
2012-05-14 12:44:06 +00:00
|
|
|
*/
|
|
|
|
kvm_flush_remote_tlbs(kvm);
|
2011-07-11 19:32:13 +00:00
|
|
|
|
2013-03-06 07:05:52 +00:00
|
|
|
list_for_each_entry_safe(sp, nsp, invalid_list, link) {
|
2010-06-04 13:53:54 +00:00
|
|
|
WARN_ON(!sp->role.invalid || sp->root_count);
|
2011-07-11 19:26:40 +00:00
|
|
|
kvm_mmu_free_page(sp);
|
2013-03-06 07:05:52 +00:00
|
|
|
}
|
2010-06-04 13:53:54 +00:00
|
|
|
}
|
|
|
|
|
2013-03-06 07:06:58 +00:00
|
|
|
static bool prepare_zap_oldest_mmu_page(struct kvm *kvm,
|
|
|
|
struct list_head *invalid_list)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
|
|
|
|
if (list_empty(&kvm->arch.active_mmu_pages))
|
|
|
|
return false;
|
|
|
|
|
2016-01-01 11:47:14 +00:00
|
|
|
sp = list_last_entry(&kvm->arch.active_mmu_pages,
|
|
|
|
struct kvm_mmu_page, link);
|
2017-08-10 13:55:51 +00:00
|
|
|
return kvm_mmu_prepare_zap_page(kvm, sp, invalid_list);
|
2013-03-06 07:06:58 +00:00
|
|
|
}
|
|
|
|
|
2007-10-02 16:52:55 +00:00
|
|
|
/*
|
|
|
|
* Changing the number of mmu pages allocated to the vm
|
2010-08-20 01:11:28 +00:00
|
|
|
* Note: if goal_nr_mmu_pages is too small, you will get dead lock
|
2007-10-02 16:52:55 +00:00
|
|
|
*/
|
2010-08-20 01:11:28 +00:00
|
|
|
void kvm_mmu_change_mmu_pages(struct kvm *kvm, unsigned int goal_nr_mmu_pages)
|
2007-10-02 16:52:55 +00:00
|
|
|
{
|
2010-06-04 13:55:29 +00:00
|
|
|
LIST_HEAD(invalid_list);
|
2007-10-02 16:52:55 +00:00
|
|
|
|
2013-01-08 10:46:07 +00:00
|
|
|
spin_lock(&kvm->mmu_lock);
|
|
|
|
|
2010-08-20 01:11:28 +00:00
|
|
|
if (kvm->arch.n_used_mmu_pages > goal_nr_mmu_pages) {
|
2013-03-06 07:06:58 +00:00
|
|
|
/* Need to free some mmu pages to achieve the goal. */
|
|
|
|
while (kvm->arch.n_used_mmu_pages > goal_nr_mmu_pages)
|
|
|
|
if (!prepare_zap_oldest_mmu_page(kvm, &invalid_list))
|
|
|
|
break;
|
2007-10-02 16:52:55 +00:00
|
|
|
|
2011-07-11 19:26:40 +00:00
|
|
|
kvm_mmu_commit_zap_page(kvm, &invalid_list);
|
2010-08-20 01:11:28 +00:00
|
|
|
goal_nr_mmu_pages = kvm->arch.n_used_mmu_pages;
|
2007-10-02 16:52:55 +00:00
|
|
|
}
|
|
|
|
|
2010-08-20 01:11:28 +00:00
|
|
|
kvm->arch.n_max_mmu_pages = goal_nr_mmu_pages;
|
2013-01-08 10:46:07 +00:00
|
|
|
|
|
|
|
spin_unlock(&kvm->mmu_lock);
|
2007-10-02 16:52:55 +00:00
|
|
|
}
|
|
|
|
|
2011-09-22 09:02:48 +00:00
|
|
|
int kvm_mmu_unprotect_page(struct kvm *kvm, gfn_t gfn)
|
2007-01-06 00:36:45 +00:00
|
|
|
{
|
2007-11-21 13:28:32 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
2010-06-04 13:55:29 +00:00
|
|
|
LIST_HEAD(invalid_list);
|
2007-01-06 00:36:45 +00:00
|
|
|
int r;
|
|
|
|
|
2010-08-28 11:19:42 +00:00
|
|
|
pgprintk("%s: looking for gfn %llx\n", __func__, gfn);
|
2007-01-06 00:36:45 +00:00
|
|
|
r = 0;
|
2011-09-22 09:02:48 +00:00
|
|
|
spin_lock(&kvm->mmu_lock);
|
hlist: drop the node parameter from iterators
I'm not sure why, but the hlist for each entry iterators were conceived
list_for_each_entry(pos, head, member)
The hlist ones were greedy and wanted an extra parameter:
hlist_for_each_entry(tpos, pos, head, member)
Why did they need an extra pos parameter? I'm not quite sure. Not only
they don't really need it, it also prevents the iterator from looking
exactly like the list iterator, which is unfortunate.
Besides the semantic patch, there was some manual work required:
- Fix up the actual hlist iterators in linux/list.h
- Fix up the declaration of other iterators based on the hlist ones.
- A very small amount of places were using the 'node' parameter, this
was modified to use 'obj->member' instead.
- Coccinelle didn't handle the hlist_for_each_entry_safe iterator
properly, so those had to be fixed up manually.
The semantic patch which is mostly the work of Peter Senna Tschudin is here:
@@
iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host;
type T;
expression a,c,d,e;
identifier b;
statement S;
@@
-T b;
<+... when != b
(
hlist_for_each_entry(a,
- b,
c, d) S
|
hlist_for_each_entry_continue(a,
- b,
c) S
|
hlist_for_each_entry_from(a,
- b,
c) S
|
hlist_for_each_entry_rcu(a,
- b,
c, d) S
|
hlist_for_each_entry_rcu_bh(a,
- b,
c, d) S
|
hlist_for_each_entry_continue_rcu_bh(a,
- b,
c) S
|
for_each_busy_worker(a, c,
- b,
d) S
|
ax25_uid_for_each(a,
- b,
c) S
|
ax25_for_each(a,
- b,
c) S
|
inet_bind_bucket_for_each(a,
- b,
c) S
|
sctp_for_each_hentry(a,
- b,
c) S
|
sk_for_each(a,
- b,
c) S
|
sk_for_each_rcu(a,
- b,
c) S
|
sk_for_each_from
-(a, b)
+(a)
S
+ sk_for_each_from(a) S
|
sk_for_each_safe(a,
- b,
c, d) S
|
sk_for_each_bound(a,
- b,
c) S
|
hlist_for_each_entry_safe(a,
- b,
c, d, e) S
|
hlist_for_each_entry_continue_rcu(a,
- b,
c) S
|
nr_neigh_for_each(a,
- b,
c) S
|
nr_neigh_for_each_safe(a,
- b,
c, d) S
|
nr_node_for_each(a,
- b,
c) S
|
nr_node_for_each_safe(a,
- b,
c, d) S
|
- for_each_gfn_sp(a, c, d, b) S
+ for_each_gfn_sp(a, c, d) S
|
- for_each_gfn_indirect_valid_sp(a, c, d, b) S
+ for_each_gfn_indirect_valid_sp(a, c, d) S
|
for_each_host(a,
- b,
c) S
|
for_each_host_safe(a,
- b,
c, d) S
|
for_each_mesh_entry(a,
- b,
c, d) S
)
...+>
[akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c]
[akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c]
[akpm@linux-foundation.org: checkpatch fixes]
[akpm@linux-foundation.org: fix warnings]
[akpm@linux-foudnation.org: redo intrusive kvm changes]
Tested-by: Peter Senna Tschudin <peter.senna@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:06:00 +00:00
|
|
|
for_each_gfn_indirect_valid_sp(kvm, sp, gfn) {
|
2010-08-28 11:19:42 +00:00
|
|
|
pgprintk("%s: gfn %llx role %x\n", __func__, gfn,
|
2010-06-04 13:53:07 +00:00
|
|
|
sp->role.word);
|
|
|
|
r = 1;
|
2010-06-04 13:56:11 +00:00
|
|
|
kvm_mmu_prepare_zap_page(kvm, sp, &invalid_list);
|
2010-06-04 13:53:07 +00:00
|
|
|
}
|
2010-06-04 13:55:29 +00:00
|
|
|
kvm_mmu_commit_zap_page(kvm, &invalid_list);
|
2011-09-22 09:02:48 +00:00
|
|
|
spin_unlock(&kvm->mmu_lock);
|
|
|
|
|
2007-01-06 00:36:45 +00:00
|
|
|
return r;
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
}
|
2011-09-22 09:02:48 +00:00
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_unprotect_page);
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
|
2016-02-24 09:51:15 +00:00
|
|
|
static void kvm_unsync_page(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp)
|
2010-05-24 07:40:07 +00:00
|
|
|
{
|
|
|
|
trace_kvm_mmu_unsync_page(sp);
|
|
|
|
++vcpu->kvm->stat.mmu_unsync;
|
|
|
|
sp->unsync = 1;
|
|
|
|
|
|
|
|
kvm_mmu_mark_parents_unsync(sp);
|
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:11 +00:00
|
|
|
static bool mmu_need_write_protect(struct kvm_vcpu *vcpu, gfn_t gfn,
|
|
|
|
bool can_unsync)
|
2008-09-23 16:18:39 +00:00
|
|
|
{
|
2016-02-24 09:51:15 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
2008-09-23 16:18:39 +00:00
|
|
|
|
2016-02-24 09:51:11 +00:00
|
|
|
if (kvm_page_track_is_active(vcpu, gfn, KVM_PAGE_TRACK_WRITE))
|
|
|
|
return true;
|
2010-05-24 07:40:07 +00:00
|
|
|
|
2016-02-24 09:51:15 +00:00
|
|
|
for_each_gfn_indirect_valid_sp(vcpu->kvm, sp, gfn) {
|
KVM: MMU: fix writable sync sp mapping
While we sync many unsync sp at one time(in mmu_sync_children()),
we may mapping the spte writable, it's dangerous, if one unsync
sp's mapping gfn is another unsync page's gfn.
For example:
SP1.pte[0] = P
SP2.gfn's pfn = P
[SP1.pte[0] = SP2.gfn's pfn]
First, we write protected SP1 and SP2, but SP1 and SP2 are still the
unsync sp.
Then, sync SP1 first, it will detect SP1.pte[0].gfn only has one unsync-sp,
that is SP2, so it will mapping it writable, but we plan to sync SP2 soon,
at this point, the SP2->unsync is not reliable since later we sync SP2 but
SP2->gfn is already writable.
So the final result is: SP2 is the sync page but SP2.gfn is writable.
This bug will corrupt guest's page table, fixed by mark read-only mapping
if the mapped gfn has shadow pages.
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
2010-06-30 08:02:02 +00:00
|
|
|
if (!can_unsync)
|
2016-02-24 09:51:11 +00:00
|
|
|
return true;
|
KVM: MMU: fix writable sync sp mapping
While we sync many unsync sp at one time(in mmu_sync_children()),
we may mapping the spte writable, it's dangerous, if one unsync
sp's mapping gfn is another unsync page's gfn.
For example:
SP1.pte[0] = P
SP2.gfn's pfn = P
[SP1.pte[0] = SP2.gfn's pfn]
First, we write protected SP1 and SP2, but SP1 and SP2 are still the
unsync sp.
Then, sync SP1 first, it will detect SP1.pte[0].gfn only has one unsync-sp,
that is SP2, so it will mapping it writable, but we plan to sync SP2 soon,
at this point, the SP2->unsync is not reliable since later we sync SP2 but
SP2->gfn is already writable.
So the final result is: SP2 is the sync page but SP2.gfn is writable.
This bug will corrupt guest's page table, fixed by mark read-only mapping
if the mapped gfn has shadow pages.
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
2010-06-30 08:02:02 +00:00
|
|
|
|
2016-02-24 09:51:15 +00:00
|
|
|
if (sp->unsync)
|
|
|
|
continue;
|
2010-05-24 07:40:07 +00:00
|
|
|
|
2016-02-24 09:51:15 +00:00
|
|
|
WARN_ON(sp->role.level != PT_PAGE_TABLE_LEVEL);
|
|
|
|
kvm_unsync_page(vcpu, sp);
|
2008-09-23 16:18:39 +00:00
|
|
|
}
|
2016-02-24 09:51:11 +00:00
|
|
|
|
2018-06-27 21:59:05 +00:00
|
|
|
/*
|
|
|
|
* We need to ensure that the marking of unsync pages is visible
|
|
|
|
* before the SPTE is updated to allow writes because
|
|
|
|
* kvm_mmu_sync_roots() checks the unsync flags without holding
|
|
|
|
* the MMU lock and so can race with this. If the SPTE was updated
|
|
|
|
* before the page had been marked as unsync-ed, something like the
|
|
|
|
* following could happen:
|
|
|
|
*
|
|
|
|
* CPU 1 CPU 2
|
|
|
|
* ---------------------------------------------------------------------
|
|
|
|
* 1.2 Host updates SPTE
|
|
|
|
* to be writable
|
|
|
|
* 2.1 Guest writes a GPTE for GVA X.
|
|
|
|
* (GPTE being in the guest page table shadowed
|
|
|
|
* by the SP from CPU 1.)
|
|
|
|
* This reads SPTE during the page table walk.
|
|
|
|
* Since SPTE.W is read as 1, there is no
|
|
|
|
* fault.
|
|
|
|
*
|
|
|
|
* 2.2 Guest issues TLB flush.
|
|
|
|
* That causes a VM Exit.
|
|
|
|
*
|
|
|
|
* 2.3 kvm_mmu_sync_pages() reads sp->unsync.
|
|
|
|
* Since it is false, so it just returns.
|
|
|
|
*
|
|
|
|
* 2.4 Guest accesses GVA X.
|
|
|
|
* Since the mapping in the SP was not updated,
|
|
|
|
* so the old mapping for GVA X incorrectly
|
|
|
|
* gets used.
|
|
|
|
* 1.1 Host marks SP
|
|
|
|
* as unsync
|
|
|
|
* (sp->unsync = true)
|
|
|
|
*
|
|
|
|
* The write barrier below ensures that 1.1 happens before 1.2 and thus
|
|
|
|
* the situation in 2.4 does not arise. The implicit barrier in 2.2
|
|
|
|
* pairs with this write barrier.
|
|
|
|
*/
|
|
|
|
smp_wmb();
|
|
|
|
|
2016-02-24 09:51:11 +00:00
|
|
|
return false;
|
2008-09-23 16:18:39 +00:00
|
|
|
}
|
|
|
|
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
static bool kvm_is_mmio_pfn(kvm_pfn_t pfn)
|
2015-07-07 13:03:18 +00:00
|
|
|
{
|
|
|
|
if (pfn_valid(pfn))
|
2017-12-20 07:29:29 +00:00
|
|
|
return !is_zero_pfn(pfn) && PageReserved(pfn_to_page(pfn)) &&
|
|
|
|
/*
|
|
|
|
* Some reserved pages, such as those from NVDIMM
|
|
|
|
* DAX devices, are not for MMIO, and can be mapped
|
|
|
|
* with cached memory type for better performance.
|
|
|
|
* However, the above check misconceives those pages
|
|
|
|
* as MMIO, and results in KVM mapping them with UC
|
|
|
|
* memory type, which would hurt the performance.
|
|
|
|
* Therefore, we check the host memory type in addition
|
|
|
|
* and only treat UC/UC-/WC pages as MMIO.
|
|
|
|
*/
|
|
|
|
(!pat_enabled() || pat_pfn_immune_to_uc_mtrr(pfn));
|
2015-07-07 13:03:18 +00:00
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:04 +00:00
|
|
|
/* Bits which may be returned by set_spte() */
|
|
|
|
#define SET_SPTE_WRITE_PROTECTED_PT BIT(0)
|
|
|
|
#define SET_SPTE_NEED_REMOTE_TLB_FLUSH BIT(1)
|
|
|
|
|
2009-06-10 11:24:23 +00:00
|
|
|
static int set_spte(struct kvm_vcpu *vcpu, u64 *sptep,
|
2013-01-08 06:36:04 +00:00
|
|
|
unsigned pte_access, int level,
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
gfn_t gfn, kvm_pfn_t pfn, bool speculative,
|
2010-11-19 09:03:22 +00:00
|
|
|
bool can_unsync, bool host_writable)
|
2007-12-09 15:40:31 +00:00
|
|
|
{
|
2016-07-12 22:18:49 +00:00
|
|
|
u64 spte = 0;
|
2008-09-23 16:18:30 +00:00
|
|
|
int ret = 0;
|
2017-07-01 00:26:31 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
2008-10-09 08:01:57 +00:00
|
|
|
|
2015-04-08 13:39:23 +00:00
|
|
|
if (set_mmio_spte(vcpu, sptep, gfn, pfn, pte_access))
|
2011-07-11 19:33:44 +00:00
|
|
|
return 0;
|
|
|
|
|
2017-07-01 00:26:31 +00:00
|
|
|
sp = page_header(__pa(sptep));
|
|
|
|
if (sp_ad_disabled(sp))
|
|
|
|
spte |= shadow_acc_track_value;
|
|
|
|
|
2016-07-12 22:18:51 +00:00
|
|
|
/*
|
|
|
|
* For the EPT case, shadow_present_mask is 0 if hardware
|
|
|
|
* supports exec-only page table entries. In that case,
|
|
|
|
* ACC_USER_MASK and shadow_user_mask are used to represent
|
|
|
|
* read access. See FNAME(gpte_access) in paging_tmpl.h.
|
|
|
|
*/
|
2016-07-12 22:18:49 +00:00
|
|
|
spte |= shadow_present_mask;
|
2008-03-18 09:05:52 +00:00
|
|
|
if (!speculative)
|
2017-07-01 00:26:31 +00:00
|
|
|
spte |= spte_shadow_accessed_mask(spte);
|
2011-07-11 19:24:39 +00:00
|
|
|
|
2008-04-25 13:13:50 +00:00
|
|
|
if (pte_access & ACC_EXEC_MASK)
|
|
|
|
spte |= shadow_x_mask;
|
|
|
|
else
|
|
|
|
spte |= shadow_nx_mask;
|
2012-06-20 07:58:58 +00:00
|
|
|
|
2007-12-09 15:40:31 +00:00
|
|
|
if (pte_access & ACC_USER_MASK)
|
2008-04-25 13:13:50 +00:00
|
|
|
spte |= shadow_user_mask;
|
2012-06-20 07:58:58 +00:00
|
|
|
|
2009-07-27 14:30:44 +00:00
|
|
|
if (level > PT_PAGE_TABLE_LEVEL)
|
2008-02-23 14:44:30 +00:00
|
|
|
spte |= PT_PAGE_SIZE_MASK;
|
2010-09-13 14:45:28 +00:00
|
|
|
if (tdp_enabled)
|
2009-04-27 12:35:42 +00:00
|
|
|
spte |= kvm_x86_ops->get_mt_mask(vcpu, gfn,
|
2015-07-07 13:03:18 +00:00
|
|
|
kvm_is_mmio_pfn(pfn));
|
2007-12-09 15:40:31 +00:00
|
|
|
|
2010-11-19 09:03:22 +00:00
|
|
|
if (host_writable)
|
2009-09-23 18:47:17 +00:00
|
|
|
spte |= SPTE_HOST_WRITEABLE;
|
2010-12-23 08:09:29 +00:00
|
|
|
else
|
|
|
|
pte_access &= ~ACC_WRITE_MASK;
|
2009-09-23 18:47:17 +00:00
|
|
|
|
2018-03-08 23:17:31 +00:00
|
|
|
if (!kvm_is_mmio_pfn(pfn))
|
|
|
|
spte |= shadow_me_mask;
|
|
|
|
|
2008-04-02 19:46:56 +00:00
|
|
|
spte |= (u64)pfn << PAGE_SHIFT;
|
2007-12-09 15:40:31 +00:00
|
|
|
|
2013-01-08 06:36:04 +00:00
|
|
|
if (pte_access & ACC_WRITE_MASK) {
|
2007-12-09 15:40:31 +00:00
|
|
|
|
KVM: MMU: optimize for set_spte
There are two cases we need to adjust page size in set_spte:
1): the one is other vcpu creates new sp in the window between mapping_level()
and acquiring mmu-lock.
2): the another case is the new sp is created by itself (page-fault path) when
guest uses the target gfn as its page table.
In current code, set_spte drop the spte and emulate the access for these case,
it works not good:
- for the case 1, it may destroy the mapping established by other vcpu, and
do expensive instruction emulation.
- for the case 2, it may emulate the access even if the guest is accessing
the page which not used as page table. There is a example, 0~2M is used as
huge page in guest, in this huge page, only page 3 used as page table, then
guest read/writes on other pages can cause instruction emulation.
Both of these cases can be fixed by allowing guest to retry the access, it
will refault, then we can establish the mapping by using small page
Signed-off-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Acked-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Gleb Natapov <gleb@redhat.com>
2012-12-03 23:17:11 +00:00
|
|
|
/*
|
2013-01-08 06:36:51 +00:00
|
|
|
* Other vcpu creates new sp in the window between
|
|
|
|
* mapping_level() and acquiring mmu-lock. We can
|
|
|
|
* allow guest to retry the access, the mapping can
|
|
|
|
* be fixed if guest refault.
|
KVM: MMU: optimize for set_spte
There are two cases we need to adjust page size in set_spte:
1): the one is other vcpu creates new sp in the window between mapping_level()
and acquiring mmu-lock.
2): the another case is the new sp is created by itself (page-fault path) when
guest uses the target gfn as its page table.
In current code, set_spte drop the spte and emulate the access for these case,
it works not good:
- for the case 1, it may destroy the mapping established by other vcpu, and
do expensive instruction emulation.
- for the case 2, it may emulate the access even if the guest is accessing
the page which not used as page table. There is a example, 0~2M is used as
huge page in guest, in this huge page, only page 3 used as page table, then
guest read/writes on other pages can cause instruction emulation.
Both of these cases can be fixed by allowing guest to retry the access, it
will refault, then we can establish the mapping by using small page
Signed-off-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Acked-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Gleb Natapov <gleb@redhat.com>
2012-12-03 23:17:11 +00:00
|
|
|
*/
|
2009-07-27 14:30:44 +00:00
|
|
|
if (level > PT_PAGE_TABLE_LEVEL &&
|
2016-02-24 09:51:06 +00:00
|
|
|
mmu_gfn_lpage_is_disallowed(vcpu, gfn, level))
|
2010-06-06 11:31:27 +00:00
|
|
|
goto done;
|
2008-09-23 16:18:32 +00:00
|
|
|
|
2012-06-20 07:58:58 +00:00
|
|
|
spte |= PT_WRITABLE_MASK | SPTE_MMU_WRITEABLE;
|
2007-12-09 15:40:31 +00:00
|
|
|
|
2008-11-25 14:58:07 +00:00
|
|
|
/*
|
|
|
|
* Optimization: for pte sync, if spte was writable the hash
|
|
|
|
* lookup is unnecessary (and expensive). Write protection
|
|
|
|
* is responsibility of mmu_get_page / kvm_sync_page.
|
|
|
|
* Same reasoning can be applied to dirty page accounting.
|
|
|
|
*/
|
2010-01-18 09:45:10 +00:00
|
|
|
if (!can_unsync && is_writable_pte(*sptep))
|
2008-11-25 14:58:07 +00:00
|
|
|
goto set_pte;
|
|
|
|
|
2008-09-23 16:18:39 +00:00
|
|
|
if (mmu_need_write_protect(vcpu, gfn, can_unsync)) {
|
2010-08-28 11:19:42 +00:00
|
|
|
pgprintk("%s: found shadow page for %llx, marking ro\n",
|
2008-03-03 20:59:56 +00:00
|
|
|
__func__, gfn);
|
2018-06-27 21:59:04 +00:00
|
|
|
ret |= SET_SPTE_WRITE_PROTECTED_PT;
|
2007-12-09 15:40:31 +00:00
|
|
|
pte_access &= ~ACC_WRITE_MASK;
|
2012-06-20 07:58:58 +00:00
|
|
|
spte &= ~(PT_WRITABLE_MASK | SPTE_MMU_WRITEABLE);
|
2007-12-09 15:40:31 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-01-28 02:54:25 +00:00
|
|
|
if (pte_access & ACC_WRITE_MASK) {
|
2015-04-08 13:39:23 +00:00
|
|
|
kvm_vcpu_mark_page_dirty(vcpu, gfn);
|
2017-07-01 00:26:31 +00:00
|
|
|
spte |= spte_shadow_dirty_mask(spte);
|
2015-01-28 02:54:25 +00:00
|
|
|
}
|
2007-12-09 15:40:31 +00:00
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
if (speculative)
|
|
|
|
spte = mark_spte_for_access_track(spte);
|
|
|
|
|
2008-09-23 16:18:32 +00:00
|
|
|
set_pte:
|
2012-06-20 07:58:33 +00:00
|
|
|
if (mmu_spte_update(sptep, spte))
|
2018-06-27 21:59:04 +00:00
|
|
|
ret |= SET_SPTE_NEED_REMOTE_TLB_FLUSH;
|
2010-06-06 11:31:27 +00:00
|
|
|
done:
|
2008-09-23 16:18:30 +00:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2017-08-17 13:03:32 +00:00
|
|
|
static int mmu_set_spte(struct kvm_vcpu *vcpu, u64 *sptep, unsigned pte_access,
|
|
|
|
int write_fault, int level, gfn_t gfn, kvm_pfn_t pfn,
|
|
|
|
bool speculative, bool host_writable)
|
2008-09-23 16:18:30 +00:00
|
|
|
{
|
|
|
|
int was_rmapped = 0;
|
2009-08-05 18:43:58 +00:00
|
|
|
int rmap_count;
|
2018-06-27 21:59:04 +00:00
|
|
|
int set_spte_ret;
|
2017-08-17 13:03:32 +00:00
|
|
|
int ret = RET_PF_RETRY;
|
2018-07-24 08:17:07 +00:00
|
|
|
bool flush = false;
|
2008-09-23 16:18:30 +00:00
|
|
|
|
2013-02-05 07:27:27 +00:00
|
|
|
pgprintk("%s: spte %llx write_fault %d gfn %llx\n", __func__,
|
|
|
|
*sptep, write_fault, gfn);
|
2008-09-23 16:18:30 +00:00
|
|
|
|
2015-11-20 08:44:55 +00:00
|
|
|
if (is_shadow_present_pte(*sptep)) {
|
2008-09-23 16:18:30 +00:00
|
|
|
/*
|
|
|
|
* If we overwrite a PTE page pointer with a 2MB PMD, unlink
|
|
|
|
* the parent of the now unreachable PTE.
|
|
|
|
*/
|
2009-07-27 14:30:44 +00:00
|
|
|
if (level > PT_PAGE_TABLE_LEVEL &&
|
|
|
|
!is_large_pte(*sptep)) {
|
2008-09-23 16:18:30 +00:00
|
|
|
struct kvm_mmu_page *child;
|
2009-06-10 11:24:23 +00:00
|
|
|
u64 pte = *sptep;
|
2008-09-23 16:18:30 +00:00
|
|
|
|
|
|
|
child = page_header(pte & PT64_BASE_ADDR_MASK);
|
2011-05-15 15:28:29 +00:00
|
|
|
drop_parent_pte(child, sptep);
|
2018-07-24 08:17:07 +00:00
|
|
|
flush = true;
|
2009-06-10 11:24:23 +00:00
|
|
|
} else if (pfn != spte_to_pfn(*sptep)) {
|
2010-08-28 11:19:42 +00:00
|
|
|
pgprintk("hfn old %llx new %llx\n",
|
2009-06-10 11:24:23 +00:00
|
|
|
spte_to_pfn(*sptep), pfn);
|
2011-07-11 19:28:04 +00:00
|
|
|
drop_spte(vcpu->kvm, sptep);
|
2018-07-24 08:17:07 +00:00
|
|
|
flush = true;
|
2009-02-18 13:08:59 +00:00
|
|
|
} else
|
|
|
|
was_rmapped = 1;
|
2008-09-23 16:18:30 +00:00
|
|
|
}
|
2009-07-27 14:30:44 +00:00
|
|
|
|
2018-06-27 21:59:04 +00:00
|
|
|
set_spte_ret = set_spte(vcpu, sptep, pte_access, level, gfn, pfn,
|
|
|
|
speculative, true, host_writable);
|
|
|
|
if (set_spte_ret & SET_SPTE_WRITE_PROTECTED_PT) {
|
2008-09-23 16:18:30 +00:00
|
|
|
if (write_fault)
|
2017-08-17 13:03:32 +00:00
|
|
|
ret = RET_PF_EMULATE;
|
2014-09-18 16:38:37 +00:00
|
|
|
kvm_make_request(KVM_REQ_TLB_FLUSH, vcpu);
|
2008-09-23 16:18:31 +00:00
|
|
|
}
|
2018-12-06 13:21:09 +00:00
|
|
|
|
2018-07-24 08:17:07 +00:00
|
|
|
if (set_spte_ret & SET_SPTE_NEED_REMOTE_TLB_FLUSH || flush)
|
2018-12-06 13:21:09 +00:00
|
|
|
kvm_flush_remote_tlbs_with_address(vcpu->kvm, gfn,
|
|
|
|
KVM_PAGES_PER_HPAGE(level));
|
2008-09-23 16:18:30 +00:00
|
|
|
|
2015-11-20 08:44:05 +00:00
|
|
|
if (unlikely(is_mmio_spte(*sptep)))
|
2017-08-17 13:03:32 +00:00
|
|
|
ret = RET_PF_EMULATE;
|
2011-07-11 19:33:44 +00:00
|
|
|
|
2009-06-10 11:24:23 +00:00
|
|
|
pgprintk("%s: setting spte %llx\n", __func__, *sptep);
|
2010-08-28 11:19:42 +00:00
|
|
|
pgprintk("instantiating %s PTE (%s) at %llx (%llx) addr %p\n",
|
2009-06-10 11:24:23 +00:00
|
|
|
is_large_pte(*sptep)? "2MB" : "4kB",
|
2016-12-07 00:46:16 +00:00
|
|
|
*sptep & PT_WRITABLE_MASK ? "RW" : "R", gfn,
|
2009-07-09 14:36:01 +00:00
|
|
|
*sptep, sptep);
|
2009-06-10 11:24:23 +00:00
|
|
|
if (!was_rmapped && is_large_pte(*sptep))
|
2008-02-23 14:44:30 +00:00
|
|
|
++vcpu->kvm->stat.lpages;
|
|
|
|
|
2011-07-11 19:22:01 +00:00
|
|
|
if (is_shadow_present_pte(*sptep)) {
|
|
|
|
if (!was_rmapped) {
|
|
|
|
rmap_count = rmap_add(vcpu, sptep, gfn);
|
|
|
|
if (rmap_count > RMAP_RECYCLE_THRESHOLD)
|
|
|
|
rmap_recycle(vcpu, sptep, gfn);
|
|
|
|
}
|
2007-12-09 15:40:31 +00:00
|
|
|
}
|
2012-08-03 07:42:10 +00:00
|
|
|
|
2012-10-16 12:07:03 +00:00
|
|
|
kvm_release_pfn_clean(pfn);
|
2015-11-20 08:44:05 +00:00
|
|
|
|
2017-08-17 13:03:32 +00:00
|
|
|
return ret;
|
2007-12-09 15:40:31 +00:00
|
|
|
}
|
|
|
|
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
static kvm_pfn_t pte_prefetch_gfn_to_pfn(struct kvm_vcpu *vcpu, gfn_t gfn,
|
2010-08-22 11:12:48 +00:00
|
|
|
bool no_dirty_log)
|
|
|
|
{
|
|
|
|
struct kvm_memory_slot *slot;
|
|
|
|
|
2011-03-09 07:43:00 +00:00
|
|
|
slot = gfn_to_memslot_dirty_bitmap(vcpu, gfn, no_dirty_log);
|
2012-07-17 13:54:11 +00:00
|
|
|
if (!slot)
|
2012-08-03 07:37:54 +00:00
|
|
|
return KVM_PFN_ERR_FAULT;
|
2010-08-22 11:12:48 +00:00
|
|
|
|
2012-08-21 02:59:12 +00:00
|
|
|
return gfn_to_pfn_memslot_atomic(slot, gfn);
|
2010-08-22 11:12:48 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int direct_pte_prefetch_many(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu_page *sp,
|
|
|
|
u64 *start, u64 *end)
|
|
|
|
{
|
|
|
|
struct page *pages[PTE_PREFETCH_NUM];
|
2015-05-19 14:01:50 +00:00
|
|
|
struct kvm_memory_slot *slot;
|
2010-08-22 11:12:48 +00:00
|
|
|
unsigned access = sp->role.access;
|
|
|
|
int i, ret;
|
|
|
|
gfn_t gfn;
|
|
|
|
|
|
|
|
gfn = kvm_mmu_page_get_gfn(sp, start - sp->spt);
|
2015-05-19 14:01:50 +00:00
|
|
|
slot = gfn_to_memslot_dirty_bitmap(vcpu, gfn, access & ACC_WRITE_MASK);
|
|
|
|
if (!slot)
|
2010-08-22 11:12:48 +00:00
|
|
|
return -1;
|
|
|
|
|
2015-05-19 14:01:50 +00:00
|
|
|
ret = gfn_to_page_many_atomic(slot, gfn, pages, end - start);
|
2010-08-22 11:12:48 +00:00
|
|
|
if (ret <= 0)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
for (i = 0; i < ret; i++, gfn++, start++)
|
2015-11-20 08:44:05 +00:00
|
|
|
mmu_set_spte(vcpu, start, access, 0, sp->role.level, gfn,
|
|
|
|
page_to_pfn(pages[i]), true, true);
|
2010-08-22 11:12:48 +00:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void __direct_pte_prefetch(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu_page *sp, u64 *sptep)
|
|
|
|
{
|
|
|
|
u64 *spte, *start = NULL;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
WARN_ON(!sp->role.direct);
|
|
|
|
|
|
|
|
i = (sptep - sp->spt) & ~(PTE_PREFETCH_NUM - 1);
|
|
|
|
spte = sp->spt + i;
|
|
|
|
|
|
|
|
for (i = 0; i < PTE_PREFETCH_NUM; i++, spte++) {
|
2011-07-11 19:28:04 +00:00
|
|
|
if (is_shadow_present_pte(*spte) || spte == sptep) {
|
2010-08-22 11:12:48 +00:00
|
|
|
if (!start)
|
|
|
|
continue;
|
|
|
|
if (direct_pte_prefetch_many(vcpu, sp, start, spte) < 0)
|
|
|
|
break;
|
|
|
|
start = NULL;
|
|
|
|
} else if (!start)
|
|
|
|
start = spte;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void direct_pte_prefetch(struct kvm_vcpu *vcpu, u64 *sptep)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
|
2017-07-01 00:26:31 +00:00
|
|
|
sp = page_header(__pa(sptep));
|
|
|
|
|
2010-08-22 11:12:48 +00:00
|
|
|
/*
|
2017-07-01 00:26:31 +00:00
|
|
|
* Without accessed bits, there's no way to distinguish between
|
|
|
|
* actually accessed translations and prefetched, so disable pte
|
|
|
|
* prefetch if accessed bits aren't available.
|
2010-08-22 11:12:48 +00:00
|
|
|
*/
|
2017-07-01 00:26:31 +00:00
|
|
|
if (sp_ad_disabled(sp))
|
2010-08-22 11:12:48 +00:00
|
|
|
return;
|
|
|
|
|
|
|
|
if (sp->role.level > PT_PAGE_TABLE_LEVEL)
|
|
|
|
return;
|
|
|
|
|
|
|
|
__direct_pte_prefetch(vcpu, sp, sptep);
|
|
|
|
}
|
|
|
|
|
2015-11-20 08:42:23 +00:00
|
|
|
static int __direct_map(struct kvm_vcpu *vcpu, int write, int map_writable,
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
int level, gfn_t gfn, kvm_pfn_t pfn, bool prefault)
|
2008-08-22 16:28:04 +00:00
|
|
|
{
|
2008-12-25 12:54:25 +00:00
|
|
|
struct kvm_shadow_walk_iterator iterator;
|
2008-08-22 16:28:04 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
2011-07-11 19:25:56 +00:00
|
|
|
int emulate = 0;
|
2008-08-22 16:28:04 +00:00
|
|
|
gfn_t pseudo_gfn;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
if (!VALID_PAGE(vcpu->arch.mmu->root_hpa))
|
2013-12-19 17:28:51 +00:00
|
|
|
return 0;
|
|
|
|
|
2008-12-25 12:54:25 +00:00
|
|
|
for_each_shadow_entry(vcpu, (u64)gfn << PAGE_SHIFT, iterator) {
|
2009-07-27 14:30:44 +00:00
|
|
|
if (iterator.level == level) {
|
2015-11-20 08:44:05 +00:00
|
|
|
emulate = mmu_set_spte(vcpu, iterator.sptep, ACC_ALL,
|
|
|
|
write, level, gfn, pfn, prefault,
|
|
|
|
map_writable);
|
2010-08-22 11:12:48 +00:00
|
|
|
direct_pte_prefetch(vcpu, iterator.sptep);
|
2008-12-25 12:54:25 +00:00
|
|
|
++vcpu->stat.pf_fixed;
|
|
|
|
break;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2014-02-24 16:59:32 +00:00
|
|
|
drop_large_spte(vcpu, iterator.sptep);
|
2011-07-11 19:28:04 +00:00
|
|
|
if (!is_shadow_present_pte(*iterator.sptep)) {
|
2010-05-26 08:48:25 +00:00
|
|
|
u64 base_addr = iterator.addr;
|
|
|
|
|
|
|
|
base_addr &= PT64_LVL_ADDR_MASK(iterator.level);
|
|
|
|
pseudo_gfn = base_addr >> PAGE_SHIFT;
|
2008-12-25 12:54:25 +00:00
|
|
|
sp = kvm_mmu_get_page(vcpu, pseudo_gfn, iterator.addr,
|
2015-11-26 12:16:35 +00:00
|
|
|
iterator.level - 1, 1, ACC_ALL);
|
2008-08-22 16:28:04 +00:00
|
|
|
|
2015-11-26 12:14:34 +00:00
|
|
|
link_shadow_page(vcpu, iterator.sptep, sp);
|
2008-12-25 12:54:25 +00:00
|
|
|
}
|
|
|
|
}
|
2011-07-11 19:25:56 +00:00
|
|
|
return emulate;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2010-10-08 08:24:15 +00:00
|
|
|
static void kvm_send_hwpoison_signal(unsigned long address, struct task_struct *tsk)
|
2010-05-31 06:28:19 +00:00
|
|
|
{
|
2018-04-16 19:23:27 +00:00
|
|
|
send_sig_mceerr(BUS_MCEERR_AR, (void __user *)address, PAGE_SHIFT, tsk);
|
2010-05-31 06:28:19 +00:00
|
|
|
}
|
|
|
|
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
static int kvm_handle_bad_page(struct kvm_vcpu *vcpu, gfn_t gfn, kvm_pfn_t pfn)
|
2010-05-31 06:28:19 +00:00
|
|
|
{
|
2012-08-21 03:02:51 +00:00
|
|
|
/*
|
|
|
|
* Do not cache the mmio info caused by writing the readonly gfn
|
|
|
|
* into the spte otherwise read access on readonly gfn also can
|
|
|
|
* caused mmio page fault and treat it as mmio access.
|
|
|
|
*/
|
|
|
|
if (pfn == KVM_PFN_ERR_RO_FAULT)
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_EMULATE;
|
2012-08-21 03:02:51 +00:00
|
|
|
|
2012-08-03 07:38:36 +00:00
|
|
|
if (pfn == KVM_PFN_ERR_HWPOISON) {
|
2015-04-08 13:39:23 +00:00
|
|
|
kvm_send_hwpoison_signal(kvm_vcpu_gfn_to_hva(vcpu, gfn), current);
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_RETRY;
|
2011-07-11 19:29:38 +00:00
|
|
|
}
|
2010-07-07 17:16:45 +00:00
|
|
|
|
Revert "KVM: X86: Fix SMRAM accessing even if VM is shutdown"
The bug that led to commit 95e057e25892eaa48cad1e2d637b80d0f1a4fac5
was a benign warning (no adverse affects other than the warning
itself) that was detected by syzkaller. Further inspection shows
that the WARN_ON in question, in handle_ept_misconfig(), is
unnecessary and flawed (this was also briefly discussed in the
original patch: https://patchwork.kernel.org/patch/10204649).
* The WARN_ON is unnecessary as kvm_mmu_page_fault() will WARN
if reserved bits are set in the SPTEs, i.e. it covers the case
where an EPT misconfig occurred because of a KVM bug.
* The WARN_ON is flawed because it will fire on any system error
code that is hit while handling the fault, e.g. -ENOMEM can be
returned by mmu_topup_memory_caches() while handling a legitmate
MMIO EPT misconfig.
The original behavior of returning -EFAULT when userspace munmaps
an HVA without first removing the memslot is correct and desirable,
i.e. KVM is letting userspace know it has generated a bad address.
Returning RET_PF_EMULATE masks the WARN_ON in the EPT misconfig path,
but does not fix the underlying bug, i.e. the WARN_ON is bogus.
Furthermore, returning RET_PF_EMULATE has the unwanted side effect of
causing KVM to attempt to emulate an instruction on any page fault
with an invalid HVA translation, e.g. a not-present EPT violation
on a VM_PFNMAP VMA whose fault handler failed to insert a PFN.
* There is no guarantee that the fault is directly related to the
instruction, i.e. the fault could have been triggered by a side
effect memory access in the guest, e.g. while vectoring a #DB or
writing a tracing record. This could cause KVM to effectively
mask the fault if KVM doesn't model the behavior leading to the
fault, i.e. emulation could succeed and resume the guest.
* If emulation does fail, KVM will return EMULATION_FAILED instead
of -EFAULT, which is a red herring as the user will either debug
a bogus emulation attempt or scratch their head wondering why we
were attempting emulation in the first place.
TL;DR: revert to returning -EFAULT and remove the bogus WARN_ON in
handle_ept_misconfig in a future patch.
This reverts commit 95e057e25892eaa48cad1e2d637b80d0f1a4fac5.
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
2018-03-29 21:48:30 +00:00
|
|
|
return -EFAULT;
|
2010-05-31 06:28:19 +00:00
|
|
|
}
|
|
|
|
|
2011-01-13 23:46:48 +00:00
|
|
|
static void transparent_hugepage_adjust(struct kvm_vcpu *vcpu,
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
gfn_t *gfnp, kvm_pfn_t *pfnp,
|
|
|
|
int *levelp)
|
2011-01-13 23:46:48 +00:00
|
|
|
{
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
kvm_pfn_t pfn = *pfnp;
|
2011-01-13 23:46:48 +00:00
|
|
|
gfn_t gfn = *gfnp;
|
|
|
|
int level = *levelp;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Check if it's a transparent hugepage. If this would be an
|
|
|
|
* hugetlbfs page, level wouldn't be set to
|
|
|
|
* PT_PAGE_TABLE_LEVEL and there would be no adjustment done
|
|
|
|
* here.
|
|
|
|
*/
|
2014-11-10 08:33:56 +00:00
|
|
|
if (!is_error_noslot_pfn(pfn) && !kvm_is_reserved_pfn(pfn) &&
|
2011-01-13 23:46:48 +00:00
|
|
|
level == PT_PAGE_TABLE_LEVEL &&
|
mm: thp: kvm: fix memory corruption in KVM with THP enabled
After the THP refcounting change, obtaining a compound pages from
get_user_pages() no longer allows us to assume the entire compound page
is immediately mappable from a secondary MMU.
A secondary MMU doesn't want to call get_user_pages() more than once for
each compound page, in order to know if it can map the whole compound
page. So a secondary MMU needs to know from a single get_user_pages()
invocation when it can map immediately the entire compound page to avoid
a flood of unnecessary secondary MMU faults and spurious
atomic_inc()/atomic_dec() (pages don't have to be pinned by MMU notifier
users).
Ideally instead of the page->_mapcount < 1 check, get_user_pages()
should return the granularity of the "page" mapping in the "mm" passed
to get_user_pages(). However it's non trivial change to pass the "pmd"
status belonging to the "mm" walked by get_user_pages up the stack (up
to the caller of get_user_pages). So the fix just checks if there is
not a single pte mapping on the page returned by get_user_pages, and in
turn if the caller can assume that the whole compound page is mapped in
the current "mm" (in a pmd_trans_huge()). In such case the entire
compound page is safe to map into the secondary MMU without additional
get_user_pages() calls on the surrounding tail/head pages. In addition
of being faster, not having to run other get_user_pages() calls also
reduces the memory footprint of the secondary MMU fault in case the pmd
split happened as result of memory pressure.
Without this fix after a MADV_DONTNEED (like invoked by QEMU during
postcopy live migration or balloning) or after generic swapping (with a
failure in split_huge_page() that would only result in pmd splitting and
not a physical page split), KVM would map the whole compound page into
the shadow pagetables, despite regular faults or userfaults (like
UFFDIO_COPY) may map regular pages into the primary MMU as result of the
pte faults, leading to the guest mode and userland mode going out of
sync and not working on the same memory at all times.
Any other secondary MMU notifier manager (KVM is just one of the many
MMU notifier users) will need the same information if it doesn't want to
run a flood of get_user_pages_fast and it can support multiple
granularity in the secondary MMU mappings, so I think it is justified to
be exposed not just to KVM.
The other option would be to move transparent_hugepage_adjust to
mm/huge_memory.c but that currently has all kind of KVM data structures
in it, so it's definitely not a cut-and-paste work, so I couldn't do a
fix as cleaner as this one for 4.6.
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: "Li, Liang Z" <liang.z.li@intel.com>
Cc: Amit Shah <amit.shah@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-05 23:22:20 +00:00
|
|
|
PageTransCompoundMap(pfn_to_page(pfn)) &&
|
2016-02-24 09:51:06 +00:00
|
|
|
!mmu_gfn_lpage_is_disallowed(vcpu, gfn, PT_DIRECTORY_LEVEL)) {
|
2011-01-13 23:46:48 +00:00
|
|
|
unsigned long mask;
|
|
|
|
/*
|
|
|
|
* mmu_notifier_retry was successful and we hold the
|
|
|
|
* mmu_lock here, so the pmd can't become splitting
|
|
|
|
* from under us, and in turn
|
|
|
|
* __split_huge_page_refcount() can't run from under
|
|
|
|
* us and we can safely transfer the refcount from
|
|
|
|
* PG_tail to PG_head as we switch the pfn to tail to
|
|
|
|
* head.
|
|
|
|
*/
|
|
|
|
*levelp = level = PT_DIRECTORY_LEVEL;
|
|
|
|
mask = KVM_PAGES_PER_HPAGE(level) - 1;
|
|
|
|
VM_BUG_ON((gfn & mask) != (pfn & mask));
|
|
|
|
if (pfn & mask) {
|
|
|
|
gfn &= ~mask;
|
|
|
|
*gfnp = gfn;
|
|
|
|
kvm_release_pfn_clean(pfn);
|
|
|
|
pfn &= ~mask;
|
2012-05-28 06:10:43 +00:00
|
|
|
kvm_get_pfn(pfn);
|
2011-01-13 23:46:48 +00:00
|
|
|
*pfnp = pfn;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2011-07-11 19:29:38 +00:00
|
|
|
static bool handle_abnormal_pfn(struct kvm_vcpu *vcpu, gva_t gva, gfn_t gfn,
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
kvm_pfn_t pfn, unsigned access, int *ret_val)
|
2011-07-11 19:29:38 +00:00
|
|
|
{
|
|
|
|
/* The pfn is invalid, report the error! */
|
2012-10-16 12:10:59 +00:00
|
|
|
if (unlikely(is_error_pfn(pfn))) {
|
2011-07-11 19:29:38 +00:00
|
|
|
*ret_val = kvm_handle_bad_page(vcpu, gfn, pfn);
|
2016-02-23 14:28:51 +00:00
|
|
|
return true;
|
2011-07-11 19:29:38 +00:00
|
|
|
}
|
|
|
|
|
2011-07-11 19:33:44 +00:00
|
|
|
if (unlikely(is_noslot_pfn(pfn)))
|
2011-07-11 19:29:38 +00:00
|
|
|
vcpu_cache_mmio_info(vcpu, gva, gfn, access);
|
|
|
|
|
2016-02-23 14:28:51 +00:00
|
|
|
return false;
|
2011-07-11 19:29:38 +00:00
|
|
|
}
|
|
|
|
|
2013-07-30 13:01:59 +00:00
|
|
|
static bool page_fault_can_be_fast(u32 error_code)
|
2012-06-20 07:59:18 +00:00
|
|
|
{
|
2013-07-18 04:52:37 +00:00
|
|
|
/*
|
|
|
|
* Do not fix the mmio spte with invalid generation number which
|
|
|
|
* need to be updated by slow page fault path.
|
|
|
|
*/
|
|
|
|
if (unlikely(error_code & PFERR_RSVD_MASK))
|
|
|
|
return false;
|
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
/* See if the page fault is due to an NX violation */
|
|
|
|
if (unlikely(((error_code & (PFERR_FETCH_MASK | PFERR_PRESENT_MASK))
|
|
|
|
== (PFERR_FETCH_MASK | PFERR_PRESENT_MASK))))
|
|
|
|
return false;
|
|
|
|
|
2012-06-20 07:59:18 +00:00
|
|
|
/*
|
2016-12-07 00:46:16 +00:00
|
|
|
* #PF can be fast if:
|
|
|
|
* 1. The shadow page table entry is not present, which could mean that
|
|
|
|
* the fault is potentially caused by access tracking (if enabled).
|
|
|
|
* 2. The shadow page table entry is present and the fault
|
|
|
|
* is caused by write-protect, that means we just need change the W
|
|
|
|
* bit of the spte which can be done out of mmu-lock.
|
|
|
|
*
|
|
|
|
* However, if access tracking is disabled we know that a non-present
|
|
|
|
* page must be a genuine page fault where we have to create a new SPTE.
|
|
|
|
* So, if access tracking is disabled, we return true only for write
|
|
|
|
* accesses to a present page.
|
2012-06-20 07:59:18 +00:00
|
|
|
*/
|
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
return shadow_acc_track_mask != 0 ||
|
|
|
|
((error_code & (PFERR_WRITE_MASK | PFERR_PRESENT_MASK))
|
|
|
|
== (PFERR_WRITE_MASK | PFERR_PRESENT_MASK));
|
2012-06-20 07:59:18 +00:00
|
|
|
}
|
|
|
|
|
2016-12-07 00:46:12 +00:00
|
|
|
/*
|
|
|
|
* Returns true if the SPTE was fixed successfully. Otherwise,
|
|
|
|
* someone else modified the SPTE from its original value.
|
|
|
|
*/
|
2012-06-20 07:59:18 +00:00
|
|
|
static bool
|
2014-04-17 09:06:13 +00:00
|
|
|
fast_pf_fix_direct_spte(struct kvm_vcpu *vcpu, struct kvm_mmu_page *sp,
|
2016-12-22 04:29:32 +00:00
|
|
|
u64 *sptep, u64 old_spte, u64 new_spte)
|
2012-06-20 07:59:18 +00:00
|
|
|
{
|
|
|
|
gfn_t gfn;
|
|
|
|
|
|
|
|
WARN_ON(!sp->role.direct);
|
|
|
|
|
2015-01-28 02:54:25 +00:00
|
|
|
/*
|
|
|
|
* Theoretically we could also set dirty bit (and flush TLB) here in
|
|
|
|
* order to eliminate unnecessary PML logging. See comments in
|
|
|
|
* set_spte. But fast_page_fault is very unlikely to happen with PML
|
|
|
|
* enabled, so we do not do this. This might result in the same GPA
|
|
|
|
* to be logged in PML buffer again when the write really happens, and
|
|
|
|
* eventually to be called by mark_page_dirty twice. But it's also no
|
|
|
|
* harm. This also avoids the TLB flush needed after setting dirty bit
|
|
|
|
* so non-PML cases won't be impacted.
|
|
|
|
*
|
|
|
|
* Compare with set_spte where instead shadow_dirty_mask is set.
|
|
|
|
*/
|
2016-12-07 00:46:16 +00:00
|
|
|
if (cmpxchg64(sptep, old_spte, new_spte) != old_spte)
|
2016-12-07 00:46:12 +00:00
|
|
|
return false;
|
|
|
|
|
2016-12-22 04:29:32 +00:00
|
|
|
if (is_writable_pte(new_spte) && !is_writable_pte(old_spte)) {
|
2016-12-07 00:46:16 +00:00
|
|
|
/*
|
|
|
|
* The gfn of direct spte is stable since it is
|
|
|
|
* calculated by sp->gfn.
|
|
|
|
*/
|
|
|
|
gfn = kvm_mmu_page_get_gfn(sp, sptep - sp->spt);
|
|
|
|
kvm_vcpu_mark_page_dirty(vcpu, gfn);
|
|
|
|
}
|
2012-06-20 07:59:18 +00:00
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2016-12-22 04:29:32 +00:00
|
|
|
static bool is_access_allowed(u32 fault_err_code, u64 spte)
|
|
|
|
{
|
|
|
|
if (fault_err_code & PFERR_FETCH_MASK)
|
|
|
|
return is_executable_pte(spte);
|
|
|
|
|
|
|
|
if (fault_err_code & PFERR_WRITE_MASK)
|
|
|
|
return is_writable_pte(spte);
|
|
|
|
|
|
|
|
/* Fault was on Read access */
|
|
|
|
return spte & PT_PRESENT_MASK;
|
|
|
|
}
|
|
|
|
|
2012-06-20 07:59:18 +00:00
|
|
|
/*
|
|
|
|
* Return value:
|
|
|
|
* - true: let the vcpu to access on the same address again.
|
|
|
|
* - false: let the real page fault path to fix it.
|
|
|
|
*/
|
|
|
|
static bool fast_page_fault(struct kvm_vcpu *vcpu, gva_t gva, int level,
|
|
|
|
u32 error_code)
|
|
|
|
{
|
|
|
|
struct kvm_shadow_walk_iterator iterator;
|
2014-04-17 09:06:13 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
2016-12-07 00:46:12 +00:00
|
|
|
bool fault_handled = false;
|
2012-06-20 07:59:18 +00:00
|
|
|
u64 spte = 0ull;
|
2016-12-07 00:46:12 +00:00
|
|
|
uint retry_count = 0;
|
2012-06-20 07:59:18 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
if (!VALID_PAGE(vcpu->arch.mmu->root_hpa))
|
2014-01-03 19:09:32 +00:00
|
|
|
return false;
|
|
|
|
|
2013-07-30 13:01:59 +00:00
|
|
|
if (!page_fault_can_be_fast(error_code))
|
2012-06-20 07:59:18 +00:00
|
|
|
return false;
|
|
|
|
|
|
|
|
walk_shadow_page_lockless_begin(vcpu);
|
|
|
|
|
2016-12-07 00:46:12 +00:00
|
|
|
do {
|
2016-12-22 04:29:32 +00:00
|
|
|
u64 new_spte;
|
2012-06-20 07:59:18 +00:00
|
|
|
|
2016-12-22 04:29:30 +00:00
|
|
|
for_each_shadow_entry_lockless(vcpu, gva, iterator, spte)
|
|
|
|
if (!is_shadow_present_pte(spte) ||
|
|
|
|
iterator.level < level)
|
|
|
|
break;
|
|
|
|
|
2016-12-07 00:46:12 +00:00
|
|
|
sp = page_header(__pa(iterator.sptep));
|
|
|
|
if (!is_last_spte(spte, sp->role.level))
|
|
|
|
break;
|
2012-06-20 07:59:18 +00:00
|
|
|
|
2016-12-07 00:46:12 +00:00
|
|
|
/*
|
2016-12-07 00:46:16 +00:00
|
|
|
* Check whether the memory access that caused the fault would
|
|
|
|
* still cause it if it were to be performed right now. If not,
|
|
|
|
* then this is a spurious fault caused by TLB lazily flushed,
|
|
|
|
* or some other CPU has already fixed the PTE after the
|
|
|
|
* current CPU took the fault.
|
2016-12-07 00:46:12 +00:00
|
|
|
*
|
|
|
|
* Need not check the access of upper level table entries since
|
|
|
|
* they are always ACC_ALL.
|
|
|
|
*/
|
2016-12-22 04:29:32 +00:00
|
|
|
if (is_access_allowed(error_code, spte)) {
|
|
|
|
fault_handled = true;
|
|
|
|
break;
|
|
|
|
}
|
2016-12-07 00:46:16 +00:00
|
|
|
|
2016-12-22 04:29:32 +00:00
|
|
|
new_spte = spte;
|
|
|
|
|
|
|
|
if (is_access_track_spte(spte))
|
|
|
|
new_spte = restore_acc_track_spte(new_spte);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Currently, to simplify the code, write-protection can
|
|
|
|
* be removed in the fast path only if the SPTE was
|
|
|
|
* write-protected for dirty-logging or access tracking.
|
|
|
|
*/
|
|
|
|
if ((error_code & PFERR_WRITE_MASK) &&
|
|
|
|
spte_can_locklessly_be_made_writable(spte))
|
|
|
|
{
|
|
|
|
new_spte |= PT_WRITABLE_MASK;
|
2016-12-07 00:46:16 +00:00
|
|
|
|
|
|
|
/*
|
2016-12-22 04:29:32 +00:00
|
|
|
* Do not fix write-permission on the large spte. Since
|
|
|
|
* we only dirty the first page into the dirty-bitmap in
|
|
|
|
* fast_pf_fix_direct_spte(), other pages are missed
|
|
|
|
* if its slot has dirty logging enabled.
|
|
|
|
*
|
|
|
|
* Instead, we let the slow page fault path create a
|
|
|
|
* normal spte to fix the access.
|
|
|
|
*
|
|
|
|
* See the comments in kvm_arch_commit_memory_region().
|
2016-12-07 00:46:16 +00:00
|
|
|
*/
|
2016-12-22 04:29:32 +00:00
|
|
|
if (sp->role.level > PT_PAGE_TABLE_LEVEL)
|
2016-12-07 00:46:16 +00:00
|
|
|
break;
|
2016-12-07 00:46:12 +00:00
|
|
|
}
|
2012-06-20 07:59:18 +00:00
|
|
|
|
2016-12-07 00:46:16 +00:00
|
|
|
/* Verify that the fault can be handled in the fast path */
|
2016-12-22 04:29:32 +00:00
|
|
|
if (new_spte == spte ||
|
|
|
|
!is_access_allowed(error_code, new_spte))
|
2016-12-07 00:46:12 +00:00
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Currently, fast page fault only works for direct mapping
|
|
|
|
* since the gfn is not stable for indirect shadow page. See
|
|
|
|
* Documentation/virtual/kvm/locking.txt to get more detail.
|
|
|
|
*/
|
|
|
|
fault_handled = fast_pf_fix_direct_spte(vcpu, sp,
|
2016-12-07 00:46:16 +00:00
|
|
|
iterator.sptep, spte,
|
2016-12-22 04:29:32 +00:00
|
|
|
new_spte);
|
2016-12-07 00:46:12 +00:00
|
|
|
if (fault_handled)
|
|
|
|
break;
|
|
|
|
|
|
|
|
if (++retry_count > 4) {
|
|
|
|
printk_once(KERN_WARNING
|
|
|
|
"kvm: Fast #PF retrying more than 4 times.\n");
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
} while (true);
|
2014-04-17 09:06:14 +00:00
|
|
|
|
2012-06-20 07:59:41 +00:00
|
|
|
trace_fast_page_fault(vcpu, gva, error_code, iterator.sptep,
|
2016-12-07 00:46:12 +00:00
|
|
|
spte, fault_handled);
|
2012-06-20 07:59:18 +00:00
|
|
|
walk_shadow_page_lockless_end(vcpu);
|
|
|
|
|
2016-12-07 00:46:12 +00:00
|
|
|
return fault_handled;
|
2012-06-20 07:59:18 +00:00
|
|
|
}
|
|
|
|
|
2010-12-07 02:48:06 +00:00
|
|
|
static bool try_async_pf(struct kvm_vcpu *vcpu, bool prefault, gfn_t gfn,
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
gva_t gva, kvm_pfn_t *pfn, bool write, bool *writable);
|
2017-08-10 23:28:02 +00:00
|
|
|
static int make_mmu_pages_available(struct kvm_vcpu *vcpu);
|
2010-11-12 06:49:11 +00:00
|
|
|
|
2012-06-20 07:59:18 +00:00
|
|
|
static int nonpaging_map(struct kvm_vcpu *vcpu, gva_t v, u32 error_code,
|
|
|
|
gfn_t gfn, bool prefault)
|
2007-12-21 00:18:22 +00:00
|
|
|
{
|
|
|
|
int r;
|
2009-07-27 14:30:44 +00:00
|
|
|
int level;
|
2015-10-16 08:06:02 +00:00
|
|
|
bool force_pt_level = false;
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
kvm_pfn_t pfn;
|
2008-07-25 14:24:52 +00:00
|
|
|
unsigned long mmu_seq;
|
2012-06-20 07:59:18 +00:00
|
|
|
bool map_writable, write = error_code & PFERR_WRITE_MASK;
|
2007-12-21 00:18:26 +00:00
|
|
|
|
2015-10-16 08:06:02 +00:00
|
|
|
level = mapping_level(vcpu, gfn, &force_pt_level);
|
2011-01-13 23:46:48 +00:00
|
|
|
if (likely(!force_pt_level)) {
|
|
|
|
/*
|
|
|
|
* This path builds a PAE pagetable - so we can map
|
|
|
|
* 2mb pages at maximum. Therefore check if the level
|
|
|
|
* is larger than that.
|
|
|
|
*/
|
|
|
|
if (level > PT_DIRECTORY_LEVEL)
|
|
|
|
level = PT_DIRECTORY_LEVEL;
|
2009-07-27 14:30:44 +00:00
|
|
|
|
2011-01-13 23:46:48 +00:00
|
|
|
gfn &= ~(KVM_PAGES_PER_HPAGE(level) - 1);
|
2015-10-16 08:06:02 +00:00
|
|
|
}
|
2008-02-23 14:44:30 +00:00
|
|
|
|
2012-06-20 07:59:18 +00:00
|
|
|
if (fast_page_fault(vcpu, v, level, error_code))
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_RETRY;
|
2012-06-20 07:59:18 +00:00
|
|
|
|
2008-07-25 14:24:52 +00:00
|
|
|
mmu_seq = vcpu->kvm->mmu_notifier_seq;
|
2008-09-16 23:54:47 +00:00
|
|
|
smp_rmb();
|
2010-11-12 06:49:11 +00:00
|
|
|
|
2010-12-07 02:48:06 +00:00
|
|
|
if (try_async_pf(vcpu, prefault, gfn, v, &pfn, write, &map_writable))
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_RETRY;
|
2007-12-21 00:18:26 +00:00
|
|
|
|
2011-07-11 19:29:38 +00:00
|
|
|
if (handle_abnormal_pfn(vcpu, v, gfn, pfn, ACC_ALL, &r))
|
|
|
|
return r;
|
2008-01-24 09:44:11 +00:00
|
|
|
|
2007-12-21 00:18:26 +00:00
|
|
|
spin_lock(&vcpu->kvm->mmu_lock);
|
2012-10-15 03:10:18 +00:00
|
|
|
if (mmu_notifier_retry(vcpu->kvm, mmu_seq))
|
2008-07-25 14:24:52 +00:00
|
|
|
goto out_unlock;
|
2017-08-10 23:28:02 +00:00
|
|
|
if (make_mmu_pages_available(vcpu) < 0)
|
|
|
|
goto out_unlock;
|
2011-01-13 23:46:48 +00:00
|
|
|
if (likely(!force_pt_level))
|
|
|
|
transparent_hugepage_adjust(vcpu, &gfn, &pfn, &level);
|
2015-11-20 08:42:23 +00:00
|
|
|
r = __direct_map(vcpu, write, map_writable, level, gfn, pfn, prefault);
|
2007-12-21 00:18:26 +00:00
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
|
|
|
|
2007-12-21 00:18:22 +00:00
|
|
|
return r;
|
2008-07-25 14:24:52 +00:00
|
|
|
|
|
|
|
out_unlock:
|
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
|
|
|
kvm_release_pfn_clean(pfn);
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_RETRY;
|
2007-12-21 00:18:22 +00:00
|
|
|
}
|
|
|
|
|
2018-05-04 18:37:11 +00:00
|
|
|
static void mmu_free_root_page(struct kvm *kvm, hpa_t *root_hpa,
|
|
|
|
struct list_head *invalid_list)
|
2007-01-06 00:36:40 +00:00
|
|
|
{
|
2007-11-21 13:28:32 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
2007-01-06 00:36:40 +00:00
|
|
|
|
2018-05-04 18:37:11 +00:00
|
|
|
if (!VALID_PAGE(*root_hpa))
|
2007-06-05 09:17:03 +00:00
|
|
|
return;
|
2013-05-16 08:55:51 +00:00
|
|
|
|
2018-05-04 18:37:11 +00:00
|
|
|
sp = page_header(*root_hpa & PT64_BASE_ADDR_MASK);
|
|
|
|
--sp->root_count;
|
|
|
|
if (!sp->root_count && sp->role.invalid)
|
|
|
|
kvm_mmu_prepare_zap_page(kvm, sp, invalid_list);
|
2007-01-06 00:36:40 +00:00
|
|
|
|
2018-05-04 18:37:11 +00:00
|
|
|
*root_hpa = INVALID_PAGE;
|
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:17 +00:00
|
|
|
/* roots_to_free must be some combination of the KVM_MMU_ROOT_* flags */
|
2018-10-08 19:28:07 +00:00
|
|
|
void kvm_mmu_free_roots(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
|
|
|
|
ulong roots_to_free)
|
2018-05-04 18:37:11 +00:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
LIST_HEAD(invalid_list);
|
2018-06-27 21:59:17 +00:00
|
|
|
bool free_active_root = roots_to_free & KVM_MMU_ROOT_CURRENT;
|
2018-05-04 18:37:11 +00:00
|
|
|
|
2018-06-27 21:59:20 +00:00
|
|
|
BUILD_BUG_ON(KVM_MMU_NUM_PREV_ROOTS >= BITS_PER_LONG);
|
2018-05-04 18:37:11 +00:00
|
|
|
|
2018-06-27 21:59:17 +00:00
|
|
|
/* Before acquiring the MMU lock, see if we need to do any real work. */
|
2018-06-27 21:59:20 +00:00
|
|
|
if (!(free_active_root && VALID_PAGE(mmu->root_hpa))) {
|
|
|
|
for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++)
|
|
|
|
if ((roots_to_free & KVM_MMU_ROOT_PREVIOUS(i)) &&
|
|
|
|
VALID_PAGE(mmu->prev_roots[i].hpa))
|
|
|
|
break;
|
|
|
|
|
|
|
|
if (i == KVM_MMU_NUM_PREV_ROOTS)
|
|
|
|
return;
|
|
|
|
}
|
2013-05-16 08:55:51 +00:00
|
|
|
|
|
|
|
spin_lock(&vcpu->kvm->mmu_lock);
|
2007-01-06 00:36:40 +00:00
|
|
|
|
2018-06-27 21:59:20 +00:00
|
|
|
for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++)
|
|
|
|
if (roots_to_free & KVM_MMU_ROOT_PREVIOUS(i))
|
|
|
|
mmu_free_root_page(vcpu->kvm, &mmu->prev_roots[i].hpa,
|
|
|
|
&invalid_list);
|
2018-06-27 21:59:06 +00:00
|
|
|
|
2018-06-27 21:59:17 +00:00
|
|
|
if (free_active_root) {
|
|
|
|
if (mmu->shadow_root_level >= PT64_ROOT_4LEVEL &&
|
|
|
|
(mmu->root_level >= PT64_ROOT_4LEVEL || mmu->direct_map)) {
|
|
|
|
mmu_free_root_page(vcpu->kvm, &mmu->root_hpa,
|
|
|
|
&invalid_list);
|
|
|
|
} else {
|
|
|
|
for (i = 0; i < 4; ++i)
|
|
|
|
if (mmu->pae_root[i] != 0)
|
|
|
|
mmu_free_root_page(vcpu->kvm,
|
|
|
|
&mmu->pae_root[i],
|
|
|
|
&invalid_list);
|
|
|
|
mmu->root_hpa = INVALID_PAGE;
|
|
|
|
}
|
2007-01-06 00:36:40 +00:00
|
|
|
}
|
2018-05-04 18:37:11 +00:00
|
|
|
|
2010-06-04 13:55:29 +00:00
|
|
|
kvm_mmu_commit_zap_page(vcpu->kvm, &invalid_list);
|
2007-12-21 00:18:26 +00:00
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2007-01-06 00:36:40 +00:00
|
|
|
}
|
2018-05-04 18:37:11 +00:00
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_free_roots);
|
2007-01-06 00:36:40 +00:00
|
|
|
|
2009-05-12 21:55:45 +00:00
|
|
|
static int mmu_check_root(struct kvm_vcpu *vcpu, gfn_t root_gfn)
|
|
|
|
{
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
if (!kvm_is_visible_gfn(vcpu->kvm, root_gfn)) {
|
2010-05-10 09:34:53 +00:00
|
|
|
kvm_make_request(KVM_REQ_TRIPLE_FAULT, vcpu);
|
2009-05-12 21:55:45 +00:00
|
|
|
ret = 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2010-09-10 15:30:59 +00:00
|
|
|
static int mmu_alloc_direct_roots(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp;
|
2010-10-03 16:51:39 +00:00
|
|
|
unsigned i;
|
2010-09-10 15:30:59 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->shadow_root_level >= PT64_ROOT_4LEVEL) {
|
2010-09-10 15:30:59 +00:00
|
|
|
spin_lock(&vcpu->kvm->mmu_lock);
|
2017-08-10 23:28:02 +00:00
|
|
|
if(make_mmu_pages_available(vcpu) < 0) {
|
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2017-12-05 06:21:30 +00:00
|
|
|
return -ENOSPC;
|
2017-08-10 23:28:02 +00:00
|
|
|
}
|
2017-08-24 12:27:55 +00:00
|
|
|
sp = kvm_mmu_get_page(vcpu, 0, 0,
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->shadow_root_level, 1, ACC_ALL);
|
2010-09-10 15:30:59 +00:00
|
|
|
++sp->root_count;
|
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->root_hpa = __pa(sp->spt);
|
|
|
|
} else if (vcpu->arch.mmu->shadow_root_level == PT32E_ROOT_LEVEL) {
|
2010-09-10 15:30:59 +00:00
|
|
|
for (i = 0; i < 4; ++i) {
|
2018-10-08 19:28:05 +00:00
|
|
|
hpa_t root = vcpu->arch.mmu->pae_root[i];
|
2010-09-10 15:30:59 +00:00
|
|
|
|
2013-10-02 14:56:16 +00:00
|
|
|
MMU_WARN_ON(VALID_PAGE(root));
|
2010-09-10 15:30:59 +00:00
|
|
|
spin_lock(&vcpu->kvm->mmu_lock);
|
2017-08-10 23:28:02 +00:00
|
|
|
if (make_mmu_pages_available(vcpu) < 0) {
|
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2017-12-05 06:21:30 +00:00
|
|
|
return -ENOSPC;
|
2017-08-10 23:28:02 +00:00
|
|
|
}
|
2010-12-28 10:09:07 +00:00
|
|
|
sp = kvm_mmu_get_page(vcpu, i << (30 - PAGE_SHIFT),
|
2015-11-26 12:16:35 +00:00
|
|
|
i << 30, PT32_ROOT_LEVEL, 1, ACC_ALL);
|
2010-09-10 15:30:59 +00:00
|
|
|
root = __pa(sp->spt);
|
|
|
|
++sp->root_count;
|
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->pae_root[i] = root | PT_PRESENT_MASK;
|
2010-09-10 15:30:59 +00:00
|
|
|
}
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->root_hpa = __pa(vcpu->arch.mmu->pae_root);
|
2010-09-10 15:30:59 +00:00
|
|
|
} else
|
|
|
|
BUG();
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int mmu_alloc_shadow_roots(struct kvm_vcpu *vcpu)
|
2007-01-06 00:36:40 +00:00
|
|
|
{
|
2007-11-21 13:28:32 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
2010-09-10 15:31:00 +00:00
|
|
|
u64 pdptr, pm_mask;
|
|
|
|
gfn_t root_gfn;
|
|
|
|
int i;
|
2007-01-06 00:36:51 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
root_gfn = vcpu->arch.mmu->get_cr3(vcpu) >> PAGE_SHIFT;
|
2007-01-06 00:36:40 +00:00
|
|
|
|
2010-09-10 15:30:59 +00:00
|
|
|
if (mmu_check_root(vcpu, root_gfn))
|
|
|
|
return 1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Do we shadow a long mode page table? If so we need to
|
|
|
|
* write-protect the guests page table root.
|
|
|
|
*/
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->root_level >= PT64_ROOT_4LEVEL) {
|
|
|
|
hpa_t root = vcpu->arch.mmu->root_hpa;
|
2007-01-06 00:36:40 +00:00
|
|
|
|
2013-10-02 14:56:16 +00:00
|
|
|
MMU_WARN_ON(VALID_PAGE(root));
|
2010-09-10 15:30:59 +00:00
|
|
|
|
2010-05-04 09:58:32 +00:00
|
|
|
spin_lock(&vcpu->kvm->mmu_lock);
|
2017-08-10 23:28:02 +00:00
|
|
|
if (make_mmu_pages_available(vcpu) < 0) {
|
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2017-12-05 06:21:30 +00:00
|
|
|
return -ENOSPC;
|
2017-08-10 23:28:02 +00:00
|
|
|
}
|
2017-08-24 12:27:55 +00:00
|
|
|
sp = kvm_mmu_get_page(vcpu, root_gfn, 0,
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->shadow_root_level, 0, ACC_ALL);
|
2007-11-21 13:28:32 +00:00
|
|
|
root = __pa(sp->spt);
|
|
|
|
++sp->root_count;
|
2010-05-04 09:58:32 +00:00
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->root_hpa = root;
|
2009-05-12 21:55:45 +00:00
|
|
|
return 0;
|
2007-01-06 00:36:40 +00:00
|
|
|
}
|
2010-09-02 15:29:45 +00:00
|
|
|
|
2010-09-10 15:30:59 +00:00
|
|
|
/*
|
|
|
|
* We shadow a 32 bit page table. This may be a legacy 2-level
|
2010-09-10 15:31:00 +00:00
|
|
|
* or a PAE 3-level page table. In either case we need to be aware that
|
|
|
|
* the shadow page table may be a PAE or a long mode page table.
|
2010-09-10 15:30:59 +00:00
|
|
|
*/
|
2010-09-10 15:31:00 +00:00
|
|
|
pm_mask = PT_PRESENT_MASK;
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->shadow_root_level == PT64_ROOT_4LEVEL)
|
2010-09-10 15:31:00 +00:00
|
|
|
pm_mask |= PT_ACCESSED_MASK | PT_WRITABLE_MASK | PT_USER_MASK;
|
|
|
|
|
2007-01-06 00:36:40 +00:00
|
|
|
for (i = 0; i < 4; ++i) {
|
2018-10-08 19:28:05 +00:00
|
|
|
hpa_t root = vcpu->arch.mmu->pae_root[i];
|
2007-01-06 00:36:40 +00:00
|
|
|
|
2013-10-02 14:56:16 +00:00
|
|
|
MMU_WARN_ON(VALID_PAGE(root));
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->root_level == PT32E_ROOT_LEVEL) {
|
|
|
|
pdptr = vcpu->arch.mmu->get_pdptr(vcpu, i);
|
2016-07-12 22:18:50 +00:00
|
|
|
if (!(pdptr & PT_PRESENT_MASK)) {
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->pae_root[i] = 0;
|
2007-04-12 14:35:58 +00:00
|
|
|
continue;
|
|
|
|
}
|
2009-05-31 19:58:47 +00:00
|
|
|
root_gfn = pdptr >> PAGE_SHIFT;
|
2010-09-02 15:29:45 +00:00
|
|
|
if (mmu_check_root(vcpu, root_gfn))
|
|
|
|
return 1;
|
2010-04-27 00:00:05 +00:00
|
|
|
}
|
2010-05-04 09:58:32 +00:00
|
|
|
spin_lock(&vcpu->kvm->mmu_lock);
|
2017-08-10 23:28:02 +00:00
|
|
|
if (make_mmu_pages_available(vcpu) < 0) {
|
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2017-12-05 06:21:30 +00:00
|
|
|
return -ENOSPC;
|
2017-08-10 23:28:02 +00:00
|
|
|
}
|
2015-11-26 12:16:35 +00:00
|
|
|
sp = kvm_mmu_get_page(vcpu, root_gfn, i << 30, PT32_ROOT_LEVEL,
|
|
|
|
0, ACC_ALL);
|
2007-11-21 13:28:32 +00:00
|
|
|
root = __pa(sp->spt);
|
|
|
|
++sp->root_count;
|
2010-05-04 09:58:32 +00:00
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->pae_root[i] = root | pm_mask;
|
2007-01-06 00:36:40 +00:00
|
|
|
}
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->root_hpa = __pa(vcpu->arch.mmu->pae_root);
|
2010-09-10 15:31:00 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If we shadow a 32 bit page table with a long mode page
|
|
|
|
* table we enter this path.
|
|
|
|
*/
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->shadow_root_level == PT64_ROOT_4LEVEL) {
|
|
|
|
if (vcpu->arch.mmu->lm_root == NULL) {
|
2010-09-10 15:31:00 +00:00
|
|
|
/*
|
|
|
|
* The additional page necessary for this is only
|
|
|
|
* allocated on demand.
|
|
|
|
*/
|
|
|
|
|
|
|
|
u64 *lm_root;
|
|
|
|
|
2019-02-11 19:02:50 +00:00
|
|
|
lm_root = (void*)get_zeroed_page(GFP_KERNEL_ACCOUNT);
|
2010-09-10 15:31:00 +00:00
|
|
|
if (lm_root == NULL)
|
|
|
|
return 1;
|
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
lm_root[0] = __pa(vcpu->arch.mmu->pae_root) | pm_mask;
|
2010-09-10 15:31:00 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->lm_root = lm_root;
|
2010-09-10 15:31:00 +00:00
|
|
|
}
|
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->root_hpa = __pa(vcpu->arch.mmu->lm_root);
|
2010-09-10 15:31:00 +00:00
|
|
|
}
|
|
|
|
|
2009-05-12 21:55:45 +00:00
|
|
|
return 0;
|
2007-01-06 00:36:40 +00:00
|
|
|
}
|
|
|
|
|
2010-09-10 15:30:59 +00:00
|
|
|
static int mmu_alloc_roots(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->direct_map)
|
2010-09-10 15:30:59 +00:00
|
|
|
return mmu_alloc_direct_roots(vcpu);
|
|
|
|
else
|
|
|
|
return mmu_alloc_shadow_roots(vcpu);
|
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:05 +00:00
|
|
|
void kvm_mmu_sync_roots(struct kvm_vcpu *vcpu)
|
2008-09-23 16:18:34 +00:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->direct_map)
|
2010-09-10 15:31:00 +00:00
|
|
|
return;
|
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
if (!VALID_PAGE(vcpu->arch.mmu->root_hpa))
|
2008-09-23 16:18:34 +00:00
|
|
|
return;
|
2010-09-27 10:09:29 +00:00
|
|
|
|
2014-08-18 22:46:07 +00:00
|
|
|
vcpu_clear_mmio_info(vcpu, MMIO_GVA_ANY);
|
2018-06-27 21:59:05 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->root_level >= PT64_ROOT_4LEVEL) {
|
|
|
|
hpa_t root = vcpu->arch.mmu->root_hpa;
|
2008-09-23 16:18:34 +00:00
|
|
|
sp = page_header(root);
|
2018-06-27 21:59:05 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Even if another CPU was marking the SP as unsync-ed
|
|
|
|
* simultaneously, any guest page table changes are not
|
|
|
|
* guaranteed to be visible anyway until this VCPU issues a TLB
|
|
|
|
* flush strictly after those changes are made. We only need to
|
|
|
|
* ensure that the other CPU sets these flags before any actual
|
|
|
|
* changes to the page tables are made. The comments in
|
|
|
|
* mmu_need_write_protect() describe what could go wrong if this
|
|
|
|
* requirement isn't satisfied.
|
|
|
|
*/
|
|
|
|
if (!smp_load_acquire(&sp->unsync) &&
|
|
|
|
!smp_load_acquire(&sp->unsync_children))
|
|
|
|
return;
|
|
|
|
|
|
|
|
spin_lock(&vcpu->kvm->mmu_lock);
|
|
|
|
kvm_mmu_audit(vcpu, AUDIT_PRE_SYNC);
|
|
|
|
|
2008-09-23 16:18:34 +00:00
|
|
|
mmu_sync_children(vcpu, sp);
|
2018-06-27 21:59:05 +00:00
|
|
|
|
2011-11-28 12:41:00 +00:00
|
|
|
kvm_mmu_audit(vcpu, AUDIT_POST_SYNC);
|
2018-06-27 21:59:05 +00:00
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2008-09-23 16:18:34 +00:00
|
|
|
return;
|
|
|
|
}
|
2018-06-27 21:59:05 +00:00
|
|
|
|
|
|
|
spin_lock(&vcpu->kvm->mmu_lock);
|
|
|
|
kvm_mmu_audit(vcpu, AUDIT_PRE_SYNC);
|
|
|
|
|
2008-09-23 16:18:34 +00:00
|
|
|
for (i = 0; i < 4; ++i) {
|
2018-10-08 19:28:05 +00:00
|
|
|
hpa_t root = vcpu->arch.mmu->pae_root[i];
|
2008-09-23 16:18:34 +00:00
|
|
|
|
2009-05-12 21:55:45 +00:00
|
|
|
if (root && VALID_PAGE(root)) {
|
2008-09-23 16:18:34 +00:00
|
|
|
root &= PT64_BASE_ADDR_MASK;
|
|
|
|
sp = page_header(root);
|
|
|
|
mmu_sync_children(vcpu, sp);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:05 +00:00
|
|
|
kvm_mmu_audit(vcpu, AUDIT_POST_SYNC);
|
2008-12-02 00:32:04 +00:00
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2008-09-23 16:18:34 +00:00
|
|
|
}
|
2013-08-05 08:07:17 +00:00
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_sync_roots);
|
2008-09-23 16:18:34 +00:00
|
|
|
|
2010-02-10 12:21:32 +00:00
|
|
|
static gpa_t nonpaging_gva_to_gpa(struct kvm_vcpu *vcpu, gva_t vaddr,
|
2010-11-22 15:53:26 +00:00
|
|
|
u32 access, struct x86_exception *exception)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2010-11-22 15:53:26 +00:00
|
|
|
if (exception)
|
|
|
|
exception->error_code = 0;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
return vaddr;
|
|
|
|
}
|
|
|
|
|
2010-09-10 15:30:50 +00:00
|
|
|
static gpa_t nonpaging_gva_to_gpa_nested(struct kvm_vcpu *vcpu, gva_t vaddr,
|
2010-11-22 15:53:26 +00:00
|
|
|
u32 access,
|
|
|
|
struct x86_exception *exception)
|
2010-09-10 15:30:50 +00:00
|
|
|
{
|
2010-11-22 15:53:26 +00:00
|
|
|
if (exception)
|
|
|
|
exception->error_code = 0;
|
2014-09-02 11:23:06 +00:00
|
|
|
return vcpu->arch.nested_mmu.translate_gpa(vcpu, vaddr, access, exception);
|
2010-09-10 15:30:50 +00:00
|
|
|
}
|
|
|
|
|
2015-08-05 04:04:25 +00:00
|
|
|
static bool
|
|
|
|
__is_rsvd_bits_set(struct rsvd_bits_validate *rsvd_check, u64 pte, int level)
|
|
|
|
{
|
|
|
|
int bit7 = (pte >> 7) & 1, low6 = pte & 0x3f;
|
|
|
|
|
|
|
|
return (pte & rsvd_check->rsvd_bits_mask[bit7][level-1]) |
|
|
|
|
((rsvd_check->bad_mt_xwr & (1ull << low6)) != 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool is_rsvd_bits_set(struct kvm_mmu *mmu, u64 gpte, int level)
|
|
|
|
{
|
|
|
|
return __is_rsvd_bits_set(&mmu->guest_rsvd_check, gpte, level);
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool is_shadow_zero_bits_set(struct kvm_mmu *mmu, u64 spte, int level)
|
|
|
|
{
|
|
|
|
return __is_rsvd_bits_set(&mmu->shadow_zero_check, spte, level);
|
|
|
|
}
|
|
|
|
|
2016-02-22 08:23:40 +00:00
|
|
|
static bool mmio_info_in_cache(struct kvm_vcpu *vcpu, u64 addr, bool direct)
|
2011-07-11 19:33:44 +00:00
|
|
|
{
|
2017-08-17 16:36:58 +00:00
|
|
|
/*
|
|
|
|
* A nested guest cannot use the MMIO cache if it is using nested
|
|
|
|
* page tables, because cr2 is a nGPA while the cache stores GPAs.
|
|
|
|
*/
|
|
|
|
if (mmu_is_nested(vcpu))
|
|
|
|
return false;
|
|
|
|
|
2011-07-11 19:33:44 +00:00
|
|
|
if (direct)
|
|
|
|
return vcpu_match_mmio_gpa(vcpu, addr);
|
|
|
|
|
|
|
|
return vcpu_match_mmio_gva(vcpu, addr);
|
|
|
|
}
|
|
|
|
|
2015-08-05 04:04:26 +00:00
|
|
|
/* return true if reserved bit is detected on spte. */
|
|
|
|
static bool
|
|
|
|
walk_shadow_page_get_mmio_spte(struct kvm_vcpu *vcpu, u64 addr, u64 *sptep)
|
2011-07-11 19:33:44 +00:00
|
|
|
{
|
|
|
|
struct kvm_shadow_walk_iterator iterator;
|
2017-08-24 12:27:54 +00:00
|
|
|
u64 sptes[PT64_ROOT_MAX_LEVEL], spte = 0ull;
|
2015-08-05 04:04:26 +00:00
|
|
|
int root, leaf;
|
|
|
|
bool reserved = false;
|
2011-07-11 19:33:44 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
if (!VALID_PAGE(vcpu->arch.mmu->root_hpa))
|
2015-08-05 04:04:26 +00:00
|
|
|
goto exit;
|
2014-01-03 19:09:32 +00:00
|
|
|
|
2011-07-11 19:33:44 +00:00
|
|
|
walk_shadow_page_lockless_begin(vcpu);
|
2015-08-05 04:04:26 +00:00
|
|
|
|
2015-09-06 14:24:50 +00:00
|
|
|
for (shadow_walk_init(&iterator, vcpu, addr),
|
|
|
|
leaf = root = iterator.level;
|
2015-08-05 04:04:26 +00:00
|
|
|
shadow_walk_okay(&iterator);
|
|
|
|
__shadow_walk_next(&iterator, spte)) {
|
|
|
|
spte = mmu_spte_get_lockless(iterator.sptep);
|
|
|
|
|
|
|
|
sptes[leaf - 1] = spte;
|
2015-09-06 14:24:50 +00:00
|
|
|
leaf--;
|
2015-08-05 04:04:26 +00:00
|
|
|
|
2011-07-11 19:33:44 +00:00
|
|
|
if (!is_shadow_present_pte(spte))
|
|
|
|
break;
|
2015-08-05 04:04:26 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
reserved |= is_shadow_zero_bits_set(vcpu->arch.mmu, spte,
|
2015-09-22 08:15:59 +00:00
|
|
|
iterator.level);
|
2015-08-05 04:04:26 +00:00
|
|
|
}
|
|
|
|
|
2011-07-11 19:33:44 +00:00
|
|
|
walk_shadow_page_lockless_end(vcpu);
|
|
|
|
|
2015-08-05 04:04:26 +00:00
|
|
|
if (reserved) {
|
|
|
|
pr_err("%s: detect reserved bits on spte, addr 0x%llx, dump hierarchy:\n",
|
|
|
|
__func__, addr);
|
2015-09-06 14:24:50 +00:00
|
|
|
while (root > leaf) {
|
2015-08-05 04:04:26 +00:00
|
|
|
pr_err("------ spte 0x%llx level %d.\n",
|
|
|
|
sptes[root - 1], root);
|
|
|
|
root--;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
exit:
|
|
|
|
*sptep = spte;
|
|
|
|
return reserved;
|
2011-07-11 19:33:44 +00:00
|
|
|
}
|
|
|
|
|
2017-08-17 16:36:56 +00:00
|
|
|
static int handle_mmio_page_fault(struct kvm_vcpu *vcpu, u64 addr, bool direct)
|
2011-07-11 19:33:44 +00:00
|
|
|
{
|
|
|
|
u64 spte;
|
2015-08-05 04:04:26 +00:00
|
|
|
bool reserved;
|
2011-07-11 19:33:44 +00:00
|
|
|
|
2016-02-22 08:23:40 +00:00
|
|
|
if (mmio_info_in_cache(vcpu, addr, direct))
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_EMULATE;
|
2011-07-11 19:33:44 +00:00
|
|
|
|
2015-08-05 04:04:26 +00:00
|
|
|
reserved = walk_shadow_page_get_mmio_spte(vcpu, addr, &spte);
|
2015-11-04 12:41:21 +00:00
|
|
|
if (WARN_ON(reserved))
|
2017-08-17 13:03:32 +00:00
|
|
|
return -EINVAL;
|
2011-07-11 19:33:44 +00:00
|
|
|
|
|
|
|
if (is_mmio_spte(spte)) {
|
|
|
|
gfn_t gfn = get_mmio_spte_gfn(spte);
|
|
|
|
unsigned access = get_mmio_spte_access(spte);
|
|
|
|
|
2015-04-08 13:39:23 +00:00
|
|
|
if (!check_mmio_spte(vcpu, spte))
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_INVALID;
|
2013-06-07 08:51:26 +00:00
|
|
|
|
2011-07-11 19:33:44 +00:00
|
|
|
if (direct)
|
|
|
|
addr = 0;
|
2011-07-11 19:34:24 +00:00
|
|
|
|
|
|
|
trace_handle_mmio_page_fault(addr, gfn, access);
|
2011-07-11 19:33:44 +00:00
|
|
|
vcpu_cache_mmio_info(vcpu, addr, gfn, access);
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_EMULATE;
|
2011-07-11 19:33:44 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If the page table is zapped by other cpus, let CPU fault again on
|
|
|
|
* the address.
|
|
|
|
*/
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_RETRY;
|
2011-07-11 19:33:44 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:11 +00:00
|
|
|
static bool page_fault_handle_page_track(struct kvm_vcpu *vcpu,
|
|
|
|
u32 error_code, gfn_t gfn)
|
|
|
|
{
|
|
|
|
if (unlikely(error_code & PFERR_RSVD_MASK))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
if (!(error_code & PFERR_PRESENT_MASK) ||
|
|
|
|
!(error_code & PFERR_WRITE_MASK))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* guest is writing the page which is write tracked which can
|
|
|
|
* not be fixed by page fault handler.
|
|
|
|
*/
|
|
|
|
if (kvm_page_track_is_active(vcpu, gfn, KVM_PAGE_TRACK_WRITE))
|
|
|
|
return true;
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:12 +00:00
|
|
|
static void shadow_page_table_clear_flood(struct kvm_vcpu *vcpu, gva_t addr)
|
|
|
|
{
|
|
|
|
struct kvm_shadow_walk_iterator iterator;
|
|
|
|
u64 spte;
|
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
if (!VALID_PAGE(vcpu->arch.mmu->root_hpa))
|
2016-02-24 09:51:12 +00:00
|
|
|
return;
|
|
|
|
|
|
|
|
walk_shadow_page_lockless_begin(vcpu);
|
|
|
|
for_each_shadow_entry_lockless(vcpu, addr, iterator, spte) {
|
|
|
|
clear_sp_write_flooding_count(iterator.sptep);
|
|
|
|
if (!is_shadow_present_pte(spte))
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
walk_shadow_page_lockless_end(vcpu);
|
|
|
|
}
|
|
|
|
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
static int nonpaging_page_fault(struct kvm_vcpu *vcpu, gva_t gva,
|
2010-12-07 02:48:06 +00:00
|
|
|
u32 error_code, bool prefault)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2016-02-24 09:51:11 +00:00
|
|
|
gfn_t gfn = gva >> PAGE_SHIFT;
|
2007-01-06 00:36:54 +00:00
|
|
|
int r;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2008-03-03 20:59:56 +00:00
|
|
|
pgprintk("%s: gva %lx error %x\n", __func__, gva, error_code);
|
2011-07-11 19:33:44 +00:00
|
|
|
|
2016-02-24 09:51:11 +00:00
|
|
|
if (page_fault_handle_page_track(vcpu, error_code, gfn))
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_EMULATE;
|
2011-07-11 19:33:44 +00:00
|
|
|
|
2007-01-06 00:36:54 +00:00
|
|
|
r = mmu_topup_memory_caches(vcpu);
|
|
|
|
if (r)
|
|
|
|
return r;
|
2007-01-06 00:36:53 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
MMU_WARN_ON(!VALID_PAGE(vcpu->arch.mmu->root_hpa));
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
|
|
|
|
2007-12-09 16:43:00 +00:00
|
|
|
return nonpaging_map(vcpu, gva & PAGE_MASK,
|
2012-06-20 07:59:18 +00:00
|
|
|
error_code, gfn, prefault);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2010-10-20 13:18:02 +00:00
|
|
|
static int kvm_arch_setup_async_pf(struct kvm_vcpu *vcpu, gva_t gva, gfn_t gfn)
|
2010-10-14 09:22:46 +00:00
|
|
|
{
|
|
|
|
struct kvm_arch_async_pf arch;
|
2010-12-07 02:35:25 +00:00
|
|
|
|
2010-10-14 09:22:53 +00:00
|
|
|
arch.token = (vcpu->arch.apf.id++ << 12) | vcpu->vcpu_id;
|
2010-10-14 09:22:46 +00:00
|
|
|
arch.gfn = gfn;
|
2018-10-08 19:28:05 +00:00
|
|
|
arch.direct_map = vcpu->arch.mmu->direct_map;
|
|
|
|
arch.cr3 = vcpu->arch.mmu->get_cr3(vcpu);
|
2010-10-14 09:22:46 +00:00
|
|
|
|
2015-04-08 13:39:23 +00:00
|
|
|
return kvm_setup_async_pf(vcpu, gva, kvm_vcpu_gfn_to_hva(vcpu, gfn), &arch);
|
2010-10-14 09:22:46 +00:00
|
|
|
}
|
|
|
|
|
2017-06-09 03:13:40 +00:00
|
|
|
bool kvm_can_do_async_pf(struct kvm_vcpu *vcpu)
|
2010-10-14 09:22:46 +00:00
|
|
|
{
|
2015-07-29 10:05:37 +00:00
|
|
|
if (unlikely(!lapic_in_kernel(vcpu) ||
|
2018-01-10 13:44:42 +00:00
|
|
|
kvm_event_needs_reinjection(vcpu) ||
|
|
|
|
vcpu->arch.exception.pending))
|
2010-10-14 09:22:46 +00:00
|
|
|
return false;
|
|
|
|
|
2017-07-14 01:30:42 +00:00
|
|
|
if (!vcpu->arch.apf.delivery_as_pf_vmexit && is_guest_mode(vcpu))
|
2017-06-09 03:13:40 +00:00
|
|
|
return false;
|
|
|
|
|
2010-10-14 09:22:46 +00:00
|
|
|
return kvm_x86_ops->interrupt_allowed(vcpu);
|
|
|
|
}
|
|
|
|
|
2010-12-07 02:48:06 +00:00
|
|
|
static bool try_async_pf(struct kvm_vcpu *vcpu, bool prefault, gfn_t gfn,
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
gva_t gva, kvm_pfn_t *pfn, bool write, bool *writable)
|
2010-10-14 09:22:46 +00:00
|
|
|
{
|
2015-04-02 09:20:48 +00:00
|
|
|
struct kvm_memory_slot *slot;
|
2010-10-14 09:22:46 +00:00
|
|
|
bool async;
|
|
|
|
|
2018-05-09 21:02:05 +00:00
|
|
|
/*
|
|
|
|
* Don't expose private memslots to L2.
|
|
|
|
*/
|
|
|
|
if (is_guest_mode(vcpu) && !kvm_is_visible_gfn(vcpu->kvm, gfn)) {
|
|
|
|
*pfn = KVM_PFN_NOSLOT;
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2015-04-08 13:39:23 +00:00
|
|
|
slot = kvm_vcpu_gfn_to_memslot(vcpu, gfn);
|
2015-04-02 09:20:48 +00:00
|
|
|
async = false;
|
|
|
|
*pfn = __gfn_to_pfn_memslot(slot, gfn, false, &async, write, writable);
|
2010-10-14 09:22:46 +00:00
|
|
|
if (!async)
|
|
|
|
return false; /* *pfn has correct page already */
|
|
|
|
|
2017-06-09 03:13:40 +00:00
|
|
|
if (!prefault && kvm_can_do_async_pf(vcpu)) {
|
2010-11-01 08:58:43 +00:00
|
|
|
trace_kvm_try_async_get_page(gva, gfn);
|
2010-10-14 09:22:46 +00:00
|
|
|
if (kvm_find_async_pf_gfn(vcpu, gfn)) {
|
|
|
|
trace_kvm_async_pf_doublefault(gva, gfn);
|
|
|
|
kvm_make_request(KVM_REQ_APF_HALT, vcpu);
|
|
|
|
return true;
|
|
|
|
} else if (kvm_arch_setup_async_pf(vcpu, gva, gfn))
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2015-04-02 09:20:48 +00:00
|
|
|
*pfn = __gfn_to_pfn_memslot(slot, gfn, false, NULL, write, writable);
|
2010-10-14 09:22:46 +00:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2017-07-14 01:30:40 +00:00
|
|
|
int kvm_handle_page_fault(struct kvm_vcpu *vcpu, u64 error_code,
|
2017-08-11 16:36:43 +00:00
|
|
|
u64 fault_address, char *insn, int insn_len)
|
2017-07-14 01:30:40 +00:00
|
|
|
{
|
|
|
|
int r = 1;
|
|
|
|
|
x86/KVM/VMX: Add L1D flush logic
Add the logic for flushing L1D on VMENTER. The flush depends on the static
key being enabled and the new l1tf_flush_l1d flag being set.
The flags is set:
- Always, if the flush module parameter is 'always'
- Conditionally at:
- Entry to vcpu_run(), i.e. after executing user space
- From the sched_in notifier, i.e. when switching to a vCPU thread.
- From vmexit handlers which are considered unsafe, i.e. where
sensitive data can be brought into L1D:
- The emulator, which could be a good target for other speculative
execution-based threats,
- The MMU, which can bring host page tables in the L1 cache.
- External interrupts
- Nested operations that require the MMU (see above). That is
vmptrld, vmptrst, vmclear,vmwrite,vmread.
- When handling invept,invvpid
[ tglx: Split out from combo patch and reduced to a single flag ]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
2018-07-02 11:07:14 +00:00
|
|
|
vcpu->arch.l1tf_flush_l1d = true;
|
2017-07-14 01:30:40 +00:00
|
|
|
switch (vcpu->arch.apf.host_apf_reason) {
|
|
|
|
default:
|
|
|
|
trace_kvm_page_fault(fault_address, error_code);
|
|
|
|
|
2017-08-11 16:36:43 +00:00
|
|
|
if (kvm_event_needs_reinjection(vcpu))
|
2017-07-14 01:30:40 +00:00
|
|
|
kvm_mmu_unprotect_page_virt(vcpu, fault_address);
|
|
|
|
r = kvm_mmu_page_fault(vcpu, fault_address, error_code, insn,
|
|
|
|
insn_len);
|
|
|
|
break;
|
|
|
|
case KVM_PV_REASON_PAGE_NOT_PRESENT:
|
|
|
|
vcpu->arch.apf.host_apf_reason = 0;
|
|
|
|
local_irq_disable();
|
kvm/x86: Avoid async PF preempting the kernel incorrectly
Currently, in PREEMPT_COUNT=n kernel, kvm_async_pf_task_wait() could call
schedule() to reschedule in some cases. This could result in
accidentally ending the current RCU read-side critical section early,
causing random memory corruption in the guest, or otherwise preempting
the currently running task inside between preempt_disable and
preempt_enable.
The difficulty to handle this well is because we don't know whether an
async PF delivered in a preemptible section or RCU read-side critical section
for PREEMPT_COUNT=n, since preempt_disable()/enable() and rcu_read_lock/unlock()
are both no-ops in that case.
To cure this, we treat any async PF interrupting a kernel context as one
that cannot be preempted, preventing kvm_async_pf_task_wait() from choosing
the schedule() path in that case.
To do so, a second parameter for kvm_async_pf_task_wait() is introduced,
so that we know whether it's called from a context interrupting the
kernel, and the parameter is set properly in all the callsites.
Cc: "Paul E. McKenney" <paulmck@linux.vnet.ibm.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Wanpeng Li <wanpeng.li@hotmail.com>
Cc: stable@vger.kernel.org
Signed-off-by: Boqun Feng <boqun.feng@gmail.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
2017-10-03 13:36:51 +00:00
|
|
|
kvm_async_pf_task_wait(fault_address, 0);
|
2017-07-14 01:30:40 +00:00
|
|
|
local_irq_enable();
|
|
|
|
break;
|
|
|
|
case KVM_PV_REASON_PAGE_READY:
|
|
|
|
vcpu->arch.apf.host_apf_reason = 0;
|
|
|
|
local_irq_disable();
|
|
|
|
kvm_async_pf_task_wake(fault_address);
|
|
|
|
local_irq_enable();
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
return r;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_handle_page_fault);
|
|
|
|
|
2015-06-15 08:55:35 +00:00
|
|
|
static bool
|
|
|
|
check_hugepage_cache_consistency(struct kvm_vcpu *vcpu, gfn_t gfn, int level)
|
|
|
|
{
|
|
|
|
int page_num = KVM_PAGES_PER_HPAGE(level);
|
|
|
|
|
|
|
|
gfn &= ~(page_num - 1);
|
|
|
|
|
|
|
|
return kvm_mtrr_check_gfn_range_consistency(vcpu, gfn, page_num);
|
|
|
|
}
|
|
|
|
|
2010-10-17 16:13:42 +00:00
|
|
|
static int tdp_page_fault(struct kvm_vcpu *vcpu, gva_t gpa, u32 error_code,
|
2010-12-07 02:48:06 +00:00
|
|
|
bool prefault)
|
2008-02-07 12:47:44 +00:00
|
|
|
{
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
kvm_pfn_t pfn;
|
2008-02-07 12:47:44 +00:00
|
|
|
int r;
|
2009-07-27 14:30:44 +00:00
|
|
|
int level;
|
2015-10-16 08:04:13 +00:00
|
|
|
bool force_pt_level;
|
2008-02-23 14:44:30 +00:00
|
|
|
gfn_t gfn = gpa >> PAGE_SHIFT;
|
2008-07-25 14:24:52 +00:00
|
|
|
unsigned long mmu_seq;
|
2010-10-22 16:18:18 +00:00
|
|
|
int write = error_code & PFERR_WRITE_MASK;
|
|
|
|
bool map_writable;
|
2008-02-07 12:47:44 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
MMU_WARN_ON(!VALID_PAGE(vcpu->arch.mmu->root_hpa));
|
2008-02-07 12:47:44 +00:00
|
|
|
|
2016-02-24 09:51:11 +00:00
|
|
|
if (page_fault_handle_page_track(vcpu, error_code, gfn))
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_EMULATE;
|
2011-07-11 19:33:44 +00:00
|
|
|
|
2008-02-07 12:47:44 +00:00
|
|
|
r = mmu_topup_memory_caches(vcpu);
|
|
|
|
if (r)
|
|
|
|
return r;
|
|
|
|
|
2015-10-16 08:06:02 +00:00
|
|
|
force_pt_level = !check_hugepage_cache_consistency(vcpu, gfn,
|
|
|
|
PT_DIRECTORY_LEVEL);
|
|
|
|
level = mapping_level(vcpu, gfn, &force_pt_level);
|
2011-01-13 23:46:48 +00:00
|
|
|
if (likely(!force_pt_level)) {
|
2015-06-15 08:55:35 +00:00
|
|
|
if (level > PT_DIRECTORY_LEVEL &&
|
|
|
|
!check_hugepage_cache_consistency(vcpu, gfn, level))
|
|
|
|
level = PT_DIRECTORY_LEVEL;
|
2011-01-13 23:46:48 +00:00
|
|
|
gfn &= ~(KVM_PAGES_PER_HPAGE(level) - 1);
|
2015-10-16 08:06:02 +00:00
|
|
|
}
|
2009-07-27 14:30:44 +00:00
|
|
|
|
2012-06-20 07:59:18 +00:00
|
|
|
if (fast_page_fault(vcpu, gpa, level, error_code))
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_RETRY;
|
2012-06-20 07:59:18 +00:00
|
|
|
|
2008-07-25 14:24:52 +00:00
|
|
|
mmu_seq = vcpu->kvm->mmu_notifier_seq;
|
2008-09-16 23:54:47 +00:00
|
|
|
smp_rmb();
|
2010-10-14 09:22:46 +00:00
|
|
|
|
2010-12-07 02:48:06 +00:00
|
|
|
if (try_async_pf(vcpu, prefault, gfn, gpa, &pfn, write, &map_writable))
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_RETRY;
|
2010-10-14 09:22:46 +00:00
|
|
|
|
2011-07-11 19:29:38 +00:00
|
|
|
if (handle_abnormal_pfn(vcpu, 0, gfn, pfn, ACC_ALL, &r))
|
|
|
|
return r;
|
|
|
|
|
2008-02-07 12:47:44 +00:00
|
|
|
spin_lock(&vcpu->kvm->mmu_lock);
|
2012-10-15 03:10:18 +00:00
|
|
|
if (mmu_notifier_retry(vcpu->kvm, mmu_seq))
|
2008-07-25 14:24:52 +00:00
|
|
|
goto out_unlock;
|
2017-08-10 23:28:02 +00:00
|
|
|
if (make_mmu_pages_available(vcpu) < 0)
|
|
|
|
goto out_unlock;
|
2011-01-13 23:46:48 +00:00
|
|
|
if (likely(!force_pt_level))
|
|
|
|
transparent_hugepage_adjust(vcpu, &gfn, &pfn, &level);
|
2015-11-20 08:42:23 +00:00
|
|
|
r = __direct_map(vcpu, write, map_writable, level, gfn, pfn, prefault);
|
2008-02-07 12:47:44 +00:00
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
|
|
|
|
|
|
|
return r;
|
2008-07-25 14:24:52 +00:00
|
|
|
|
|
|
|
out_unlock:
|
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
|
|
|
kvm_release_pfn_clean(pfn);
|
2017-08-17 13:03:32 +00:00
|
|
|
return RET_PF_RETRY;
|
2008-02-07 12:47:44 +00:00
|
|
|
}
|
|
|
|
|
2013-10-02 14:56:13 +00:00
|
|
|
static void nonpaging_init_context(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu *context)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
|
|
|
context->page_fault = nonpaging_page_fault;
|
|
|
|
context->gva_to_gpa = nonpaging_gva_to_gpa;
|
2008-09-23 16:18:33 +00:00
|
|
|
context->sync_page = nonpaging_sync_page;
|
2008-09-23 16:18:35 +00:00
|
|
|
context->invlpg = nonpaging_invlpg;
|
2011-03-09 07:43:51 +00:00
|
|
|
context->update_pte = nonpaging_update_pte;
|
[PATCH] KVM: MMU: Shadow page table caching
Define a hashtable for caching shadow page tables. Look up the cache on
context switch (cr3 change) or during page faults.
The key to the cache is a combination of
- the guest page table frame number
- the number of paging levels in the guest
* we can cache real mode, 32-bit mode, pae, and long mode page
tables simultaneously. this is useful for smp bootup.
- the guest page table table
* some kernels use a page as both a page table and a page directory. this
allows multiple shadow pages to exist for that page, one per level
- the "quadrant"
* 32-bit mode page tables span 4MB, whereas a shadow page table spans
2MB. similarly, a 32-bit page directory spans 4GB, while a shadow
page directory spans 1GB. the quadrant allows caching up to 4 shadow page
tables for one guest page in one level.
- a "metaphysical" bit
* for real mode, and for pse pages, there is no guest page table, so set
the bit to avoid write protecting the page.
Signed-off-by: Avi Kivity <avi@qumranet.com>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2007-01-06 00:36:43 +00:00
|
|
|
context->root_level = 0;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
context->shadow_root_level = PT32E_ROOT_LEVEL;
|
2010-09-10 15:30:39 +00:00
|
|
|
context->direct_map = true;
|
2010-09-10 15:31:01 +00:00
|
|
|
context->nx = false;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:20 +00:00
|
|
|
/*
|
|
|
|
* Find out if a previously cached root matching the new CR3/role is available.
|
|
|
|
* The current root is also inserted into the cache.
|
|
|
|
* If a matching root was found, it is assigned to kvm_mmu->root_hpa and true is
|
|
|
|
* returned.
|
|
|
|
* Otherwise, the LRU root from the cache is assigned to kvm_mmu->root_hpa and
|
|
|
|
* false is returned. This root should now be freed by the caller.
|
|
|
|
*/
|
|
|
|
static bool cached_root_available(struct kvm_vcpu *vcpu, gpa_t new_cr3,
|
|
|
|
union kvm_mmu_page_role new_role)
|
|
|
|
{
|
|
|
|
uint i;
|
|
|
|
struct kvm_mmu_root_info root;
|
2018-10-08 19:28:05 +00:00
|
|
|
struct kvm_mmu *mmu = vcpu->arch.mmu;
|
2018-06-27 21:59:20 +00:00
|
|
|
|
|
|
|
root.cr3 = mmu->get_cr3(vcpu);
|
|
|
|
root.hpa = mmu->root_hpa;
|
|
|
|
|
|
|
|
for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++) {
|
|
|
|
swap(root, mmu->prev_roots[i]);
|
|
|
|
|
|
|
|
if (new_cr3 == root.cr3 && VALID_PAGE(root.hpa) &&
|
|
|
|
page_header(root.hpa) != NULL &&
|
|
|
|
new_role.word == page_header(root.hpa)->role.word)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
mmu->root_hpa = root.hpa;
|
|
|
|
|
|
|
|
return i < KVM_MMU_NUM_PREV_ROOTS;
|
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:09 +00:00
|
|
|
static bool fast_cr3_switch(struct kvm_vcpu *vcpu, gpa_t new_cr3,
|
2018-06-27 21:59:15 +00:00
|
|
|
union kvm_mmu_page_role new_role,
|
|
|
|
bool skip_tlb_flush)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
struct kvm_mmu *mmu = vcpu->arch.mmu;
|
2018-06-27 21:59:06 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* For now, limit the fast switch to 64-bit hosts+VMs in order to avoid
|
|
|
|
* having to deal with PDPTEs. We may add support for 32-bit hosts/VMs
|
|
|
|
* later if necessary.
|
|
|
|
*/
|
|
|
|
if (mmu->shadow_root_level >= PT64_ROOT_4LEVEL &&
|
|
|
|
mmu->root_level >= PT64_ROOT_4LEVEL) {
|
|
|
|
if (mmu_check_root(vcpu, new_cr3 >> PAGE_SHIFT))
|
|
|
|
return false;
|
|
|
|
|
2018-06-27 21:59:20 +00:00
|
|
|
if (cached_root_available(vcpu, new_cr3, new_role)) {
|
2018-06-27 21:59:06 +00:00
|
|
|
/*
|
|
|
|
* It is possible that the cached previous root page is
|
|
|
|
* obsolete because of a change in the MMU
|
|
|
|
* generation number. However, that is accompanied by
|
|
|
|
* KVM_REQ_MMU_RELOAD, which will free the root that we
|
|
|
|
* have set here and allocate a new one.
|
|
|
|
*/
|
|
|
|
|
2018-06-27 21:59:09 +00:00
|
|
|
kvm_make_request(KVM_REQ_LOAD_CR3, vcpu);
|
2018-06-27 21:59:18 +00:00
|
|
|
if (!skip_tlb_flush) {
|
|
|
|
kvm_make_request(KVM_REQ_MMU_SYNC, vcpu);
|
2018-06-27 21:59:15 +00:00
|
|
|
kvm_x86_ops->tlb_flush(vcpu, true);
|
2018-06-27 21:59:18 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The last MMIO access's GVA and GPA are cached in the
|
|
|
|
* VCPU. When switching to a new CR3, that GVA->GPA
|
|
|
|
* mapping may no longer be valid. So clear any cached
|
|
|
|
* MMIO info even when we don't need to sync the shadow
|
|
|
|
* page tables.
|
|
|
|
*/
|
|
|
|
vcpu_clear_mmio_info(vcpu, MMIO_GVA_ANY);
|
2018-06-27 21:59:15 +00:00
|
|
|
|
2018-06-27 21:59:06 +00:00
|
|
|
__clear_sp_write_flooding_count(
|
|
|
|
page_header(mmu->root_hpa));
|
|
|
|
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:09 +00:00
|
|
|
static void __kvm_mmu_new_cr3(struct kvm_vcpu *vcpu, gpa_t new_cr3,
|
2018-06-27 21:59:15 +00:00
|
|
|
union kvm_mmu_page_role new_role,
|
|
|
|
bool skip_tlb_flush)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2018-06-27 21:59:15 +00:00
|
|
|
if (!fast_cr3_switch(vcpu, new_cr3, new_role, skip_tlb_flush))
|
2018-10-08 19:28:07 +00:00
|
|
|
kvm_mmu_free_roots(vcpu, vcpu->arch.mmu,
|
|
|
|
KVM_MMU_ROOT_CURRENT);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:15 +00:00
|
|
|
void kvm_mmu_new_cr3(struct kvm_vcpu *vcpu, gpa_t new_cr3, bool skip_tlb_flush)
|
2018-06-27 21:59:09 +00:00
|
|
|
{
|
2018-06-27 21:59:15 +00:00
|
|
|
__kvm_mmu_new_cr3(vcpu, new_cr3, kvm_mmu_calc_root_page_role(vcpu),
|
|
|
|
skip_tlb_flush);
|
2018-06-27 21:59:09 +00:00
|
|
|
}
|
2018-06-27 21:59:11 +00:00
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_new_cr3);
|
2018-06-27 21:59:09 +00:00
|
|
|
|
2010-09-10 15:30:42 +00:00
|
|
|
static unsigned long get_cr3(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2010-12-05 15:30:00 +00:00
|
|
|
return kvm_read_cr3(vcpu);
|
2010-09-10 15:30:42 +00:00
|
|
|
}
|
|
|
|
|
2010-11-29 14:12:30 +00:00
|
|
|
static void inject_page_fault(struct kvm_vcpu *vcpu,
|
|
|
|
struct x86_exception *fault)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->inject_page_fault(vcpu, fault);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2015-04-08 13:39:23 +00:00
|
|
|
static bool sync_mmio_spte(struct kvm_vcpu *vcpu, u64 *sptep, gfn_t gfn,
|
2013-06-07 08:51:24 +00:00
|
|
|
unsigned access, int *nr_present)
|
2011-07-11 19:33:44 +00:00
|
|
|
{
|
|
|
|
if (unlikely(is_mmio_spte(*sptep))) {
|
|
|
|
if (gfn != get_mmio_spte_gfn(*sptep)) {
|
|
|
|
mmu_spte_clear_no_track(sptep);
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
(*nr_present)++;
|
2015-04-08 13:39:23 +00:00
|
|
|
mark_mmio_spte(vcpu, sptep, gfn, access);
|
2011-07-11 19:33:44 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
2016-02-23 11:51:19 +00:00
|
|
|
static inline bool is_last_gpte(struct kvm_mmu *mmu,
|
|
|
|
unsigned level, unsigned gpte)
|
2012-09-12 17:46:56 +00:00
|
|
|
{
|
2016-02-23 11:51:19 +00:00
|
|
|
/*
|
|
|
|
* The RHS has bit 7 set iff level < mmu->last_nonleaf_level.
|
|
|
|
* If it is clear, there are no large pages at this level, so clear
|
|
|
|
* PT_PAGE_SIZE_MASK in gpte if that is the case.
|
|
|
|
*/
|
|
|
|
gpte &= level - mmu->last_nonleaf_level;
|
|
|
|
|
2017-10-05 09:10:23 +00:00
|
|
|
/*
|
|
|
|
* PT_PAGE_TABLE_LEVEL always terminates. The RHS has bit 7 set
|
|
|
|
* iff level <= PT_PAGE_TABLE_LEVEL, which for our purpose means
|
|
|
|
* level == PT_PAGE_TABLE_LEVEL; set PT_PAGE_SIZE_MASK in gpte then.
|
|
|
|
*/
|
|
|
|
gpte |= level - PT_PAGE_TABLE_LEVEL - 1;
|
|
|
|
|
2016-02-23 11:51:19 +00:00
|
|
|
return gpte & PT_PAGE_SIZE_MASK;
|
2012-09-12 17:46:56 +00:00
|
|
|
}
|
|
|
|
|
2013-08-05 08:07:12 +00:00
|
|
|
#define PTTYPE_EPT 18 /* arbitrary */
|
|
|
|
#define PTTYPE PTTYPE_EPT
|
|
|
|
#include "paging_tmpl.h"
|
|
|
|
#undef PTTYPE
|
|
|
|
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
#define PTTYPE 64
|
|
|
|
#include "paging_tmpl.h"
|
|
|
|
#undef PTTYPE
|
|
|
|
|
|
|
|
#define PTTYPE 32
|
|
|
|
#include "paging_tmpl.h"
|
|
|
|
#undef PTTYPE
|
|
|
|
|
2015-08-05 04:04:22 +00:00
|
|
|
static void
|
|
|
|
__reset_rsvds_bits_mask(struct kvm_vcpu *vcpu,
|
|
|
|
struct rsvd_bits_validate *rsvd_check,
|
|
|
|
int maxphyaddr, int level, bool nx, bool gbpages,
|
2015-09-22 21:02:14 +00:00
|
|
|
bool pse, bool amd)
|
2009-03-30 08:21:08 +00:00
|
|
|
{
|
|
|
|
u64 exb_bit_rsvd = 0;
|
2014-05-07 12:32:50 +00:00
|
|
|
u64 gbpages_bit_rsvd = 0;
|
2014-09-02 11:24:12 +00:00
|
|
|
u64 nonleaf_bit8_rsvd = 0;
|
2009-03-30 08:21:08 +00:00
|
|
|
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->bad_mt_xwr = 0;
|
2013-08-06 09:00:32 +00:00
|
|
|
|
2015-08-05 04:04:22 +00:00
|
|
|
if (!nx)
|
2009-03-30 08:21:08 +00:00
|
|
|
exb_bit_rsvd = rsvd_bits(63, 63);
|
2015-08-05 04:04:22 +00:00
|
|
|
if (!gbpages)
|
2014-05-07 12:32:50 +00:00
|
|
|
gbpages_bit_rsvd = rsvd_bits(7, 7);
|
2014-09-02 11:24:12 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Non-leaf PML4Es and PDPEs reserve bit 8 (which would be the G bit for
|
|
|
|
* leaf entries) on AMD CPUs only.
|
|
|
|
*/
|
2015-09-22 21:02:14 +00:00
|
|
|
if (amd)
|
2014-09-02 11:24:12 +00:00
|
|
|
nonleaf_bit8_rsvd = rsvd_bits(8, 8);
|
|
|
|
|
2015-08-05 04:04:22 +00:00
|
|
|
switch (level) {
|
2009-03-30 08:21:08 +00:00
|
|
|
case PT32_ROOT_LEVEL:
|
|
|
|
/* no rsvd bits for 2 level 4K page table entries */
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][1] = 0;
|
|
|
|
rsvd_check->rsvd_bits_mask[0][0] = 0;
|
|
|
|
rsvd_check->rsvd_bits_mask[1][0] =
|
|
|
|
rsvd_check->rsvd_bits_mask[0][0];
|
2010-03-19 09:58:53 +00:00
|
|
|
|
2015-08-05 04:04:22 +00:00
|
|
|
if (!pse) {
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][1] = 0;
|
2010-03-19 09:58:53 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2009-03-30 08:21:08 +00:00
|
|
|
if (is_cpuid_PSE36())
|
|
|
|
/* 36bits PSE 4MB page */
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][1] = rsvd_bits(17, 21);
|
2009-03-30 08:21:08 +00:00
|
|
|
else
|
|
|
|
/* 32 bits PSE 4MB page */
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][1] = rsvd_bits(13, 21);
|
2009-03-30 08:21:08 +00:00
|
|
|
break;
|
|
|
|
case PT32E_ROOT_LEVEL:
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][2] =
|
2009-03-31 15:03:45 +00:00
|
|
|
rsvd_bits(maxphyaddr, 63) |
|
2014-04-04 03:31:04 +00:00
|
|
|
rsvd_bits(5, 8) | rsvd_bits(1, 2); /* PDPTE */
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][1] = exb_bit_rsvd |
|
2009-04-02 02:28:37 +00:00
|
|
|
rsvd_bits(maxphyaddr, 62); /* PDE */
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][0] = exb_bit_rsvd |
|
2009-03-30 08:21:08 +00:00
|
|
|
rsvd_bits(maxphyaddr, 62); /* PTE */
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][1] = exb_bit_rsvd |
|
2009-03-30 08:21:08 +00:00
|
|
|
rsvd_bits(maxphyaddr, 62) |
|
|
|
|
rsvd_bits(13, 20); /* large page */
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][0] =
|
|
|
|
rsvd_check->rsvd_bits_mask[0][0];
|
2009-03-30 08:21:08 +00:00
|
|
|
break;
|
2017-08-24 12:27:55 +00:00
|
|
|
case PT64_ROOT_5LEVEL:
|
|
|
|
rsvd_check->rsvd_bits_mask[0][4] = exb_bit_rsvd |
|
|
|
|
nonleaf_bit8_rsvd | rsvd_bits(7, 7) |
|
|
|
|
rsvd_bits(maxphyaddr, 51);
|
|
|
|
rsvd_check->rsvd_bits_mask[1][4] =
|
|
|
|
rsvd_check->rsvd_bits_mask[0][4];
|
2019-01-25 18:23:17 +00:00
|
|
|
/* fall through */
|
2017-08-24 12:27:54 +00:00
|
|
|
case PT64_ROOT_4LEVEL:
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][3] = exb_bit_rsvd |
|
|
|
|
nonleaf_bit8_rsvd | rsvd_bits(7, 7) |
|
2009-04-02 02:28:37 +00:00
|
|
|
rsvd_bits(maxphyaddr, 51);
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][2] = exb_bit_rsvd |
|
|
|
|
nonleaf_bit8_rsvd | gbpages_bit_rsvd |
|
2009-03-30 08:21:08 +00:00
|
|
|
rsvd_bits(maxphyaddr, 51);
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][1] = exb_bit_rsvd |
|
|
|
|
rsvd_bits(maxphyaddr, 51);
|
|
|
|
rsvd_check->rsvd_bits_mask[0][0] = exb_bit_rsvd |
|
|
|
|
rsvd_bits(maxphyaddr, 51);
|
|
|
|
rsvd_check->rsvd_bits_mask[1][3] =
|
|
|
|
rsvd_check->rsvd_bits_mask[0][3];
|
|
|
|
rsvd_check->rsvd_bits_mask[1][2] = exb_bit_rsvd |
|
2014-05-07 12:32:50 +00:00
|
|
|
gbpages_bit_rsvd | rsvd_bits(maxphyaddr, 51) |
|
2009-07-27 14:30:45 +00:00
|
|
|
rsvd_bits(13, 29);
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][1] = exb_bit_rsvd |
|
2009-04-02 02:28:37 +00:00
|
|
|
rsvd_bits(maxphyaddr, 51) |
|
|
|
|
rsvd_bits(13, 20); /* large page */
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][0] =
|
|
|
|
rsvd_check->rsvd_bits_mask[0][0];
|
2009-03-30 08:21:08 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2015-08-05 04:04:22 +00:00
|
|
|
static void reset_rsvds_bits_mask(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu *context)
|
|
|
|
{
|
|
|
|
__reset_rsvds_bits_mask(vcpu, &context->guest_rsvd_check,
|
|
|
|
cpuid_maxphyaddr(vcpu), context->root_level,
|
2017-08-04 22:12:49 +00:00
|
|
|
context->nx,
|
|
|
|
guest_cpuid_has(vcpu, X86_FEATURE_GBPAGES),
|
2015-09-22 21:02:14 +00:00
|
|
|
is_pse(vcpu), guest_cpuid_is_amd(vcpu));
|
2015-08-05 04:04:22 +00:00
|
|
|
}
|
|
|
|
|
2015-08-05 04:04:23 +00:00
|
|
|
static void
|
|
|
|
__reset_rsvds_bits_mask_ept(struct rsvd_bits_validate *rsvd_check,
|
|
|
|
int maxphyaddr, bool execonly)
|
2013-08-06 09:00:32 +00:00
|
|
|
{
|
2015-09-23 08:34:26 +00:00
|
|
|
u64 bad_mt_xwr;
|
2013-08-06 09:00:32 +00:00
|
|
|
|
2017-08-24 12:27:55 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][4] =
|
|
|
|
rsvd_bits(maxphyaddr, 51) | rsvd_bits(3, 7);
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][3] =
|
2013-08-06 09:00:32 +00:00
|
|
|
rsvd_bits(maxphyaddr, 51) | rsvd_bits(3, 7);
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][2] =
|
2013-08-06 09:00:32 +00:00
|
|
|
rsvd_bits(maxphyaddr, 51) | rsvd_bits(3, 6);
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][1] =
|
2013-08-06 09:00:32 +00:00
|
|
|
rsvd_bits(maxphyaddr, 51) | rsvd_bits(3, 6);
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[0][0] = rsvd_bits(maxphyaddr, 51);
|
2013-08-06 09:00:32 +00:00
|
|
|
|
|
|
|
/* large page */
|
2017-08-24 12:27:55 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][4] = rsvd_check->rsvd_bits_mask[0][4];
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][3] = rsvd_check->rsvd_bits_mask[0][3];
|
|
|
|
rsvd_check->rsvd_bits_mask[1][2] =
|
2013-08-06 09:00:32 +00:00
|
|
|
rsvd_bits(maxphyaddr, 51) | rsvd_bits(12, 29);
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][1] =
|
2013-08-06 09:00:32 +00:00
|
|
|
rsvd_bits(maxphyaddr, 51) | rsvd_bits(12, 20);
|
2015-08-05 04:04:21 +00:00
|
|
|
rsvd_check->rsvd_bits_mask[1][0] = rsvd_check->rsvd_bits_mask[0][0];
|
2013-08-06 09:00:32 +00:00
|
|
|
|
2015-09-23 08:34:26 +00:00
|
|
|
bad_mt_xwr = 0xFFull << (2 * 8); /* bits 3..5 must not be 2 */
|
|
|
|
bad_mt_xwr |= 0xFFull << (3 * 8); /* bits 3..5 must not be 3 */
|
|
|
|
bad_mt_xwr |= 0xFFull << (7 * 8); /* bits 3..5 must not be 7 */
|
|
|
|
bad_mt_xwr |= REPEAT_BYTE(1ull << 2); /* bits 0..2 must not be 010 */
|
|
|
|
bad_mt_xwr |= REPEAT_BYTE(1ull << 6); /* bits 0..2 must not be 110 */
|
|
|
|
if (!execonly) {
|
|
|
|
/* bits 0..2 must not be 100 unless VMX capabilities allow it */
|
|
|
|
bad_mt_xwr |= REPEAT_BYTE(1ull << 4);
|
2013-08-06 09:00:32 +00:00
|
|
|
}
|
2015-09-23 08:34:26 +00:00
|
|
|
rsvd_check->bad_mt_xwr = bad_mt_xwr;
|
2013-08-06 09:00:32 +00:00
|
|
|
}
|
|
|
|
|
2015-08-05 04:04:23 +00:00
|
|
|
static void reset_rsvds_bits_mask_ept(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu *context, bool execonly)
|
|
|
|
{
|
|
|
|
__reset_rsvds_bits_mask_ept(&context->guest_rsvd_check,
|
|
|
|
cpuid_maxphyaddr(vcpu), execonly);
|
|
|
|
}
|
|
|
|
|
2015-08-05 04:04:24 +00:00
|
|
|
/*
|
|
|
|
* the page table on host is the shadow page table for the page
|
|
|
|
* table in guest or amd nested guest, its mmu features completely
|
|
|
|
* follow the features in guest.
|
|
|
|
*/
|
|
|
|
void
|
|
|
|
reset_shadow_zero_bits_mask(struct kvm_vcpu *vcpu, struct kvm_mmu *context)
|
|
|
|
{
|
2018-10-08 19:28:10 +00:00
|
|
|
bool uses_nx = context->nx ||
|
|
|
|
context->mmu_role.base.smep_andnot_wp;
|
2017-08-25 20:55:40 +00:00
|
|
|
struct rsvd_bits_validate *shadow_zero_check;
|
|
|
|
int i;
|
2016-03-09 13:28:02 +00:00
|
|
|
|
2015-09-22 21:02:14 +00:00
|
|
|
/*
|
|
|
|
* Passing "true" to the last argument is okay; it adds a check
|
|
|
|
* on bit 8 of the SPTEs which KVM doesn't use anyway.
|
|
|
|
*/
|
2017-08-25 20:55:40 +00:00
|
|
|
shadow_zero_check = &context->shadow_zero_check;
|
|
|
|
__reset_rsvds_bits_mask(vcpu, shadow_zero_check,
|
2015-08-05 04:04:24 +00:00
|
|
|
boot_cpu_data.x86_phys_bits,
|
2016-03-09 13:28:02 +00:00
|
|
|
context->shadow_root_level, uses_nx,
|
2017-08-04 22:12:49 +00:00
|
|
|
guest_cpuid_has(vcpu, X86_FEATURE_GBPAGES),
|
|
|
|
is_pse(vcpu), true);
|
2017-08-25 20:55:40 +00:00
|
|
|
|
|
|
|
if (!shadow_me_mask)
|
|
|
|
return;
|
|
|
|
|
|
|
|
for (i = context->shadow_root_level; --i >= 0;) {
|
|
|
|
shadow_zero_check->rsvd_bits_mask[0][i] &= ~shadow_me_mask;
|
|
|
|
shadow_zero_check->rsvd_bits_mask[1][i] &= ~shadow_me_mask;
|
|
|
|
}
|
|
|
|
|
2015-08-05 04:04:24 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(reset_shadow_zero_bits_mask);
|
|
|
|
|
2015-09-22 21:02:14 +00:00
|
|
|
static inline bool boot_cpu_is_amd(void)
|
|
|
|
{
|
|
|
|
WARN_ON_ONCE(!tdp_enabled);
|
|
|
|
return shadow_x_mask == 0;
|
|
|
|
}
|
|
|
|
|
2015-08-05 04:04:24 +00:00
|
|
|
/*
|
|
|
|
* the direct page table on host, use as much mmu features as
|
|
|
|
* possible, however, kvm currently does not do execution-protection.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
reset_tdp_shadow_zero_bits_mask(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu *context)
|
|
|
|
{
|
2017-08-25 20:55:40 +00:00
|
|
|
struct rsvd_bits_validate *shadow_zero_check;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
shadow_zero_check = &context->shadow_zero_check;
|
|
|
|
|
2015-09-22 21:02:14 +00:00
|
|
|
if (boot_cpu_is_amd())
|
2017-08-25 20:55:40 +00:00
|
|
|
__reset_rsvds_bits_mask(vcpu, shadow_zero_check,
|
2015-08-05 04:04:24 +00:00
|
|
|
boot_cpu_data.x86_phys_bits,
|
|
|
|
context->shadow_root_level, false,
|
2016-03-29 15:41:58 +00:00
|
|
|
boot_cpu_has(X86_FEATURE_GBPAGES),
|
|
|
|
true, true);
|
2015-08-05 04:04:24 +00:00
|
|
|
else
|
2017-08-25 20:55:40 +00:00
|
|
|
__reset_rsvds_bits_mask_ept(shadow_zero_check,
|
2015-08-05 04:04:24 +00:00
|
|
|
boot_cpu_data.x86_phys_bits,
|
|
|
|
false);
|
|
|
|
|
2017-08-25 20:55:40 +00:00
|
|
|
if (!shadow_me_mask)
|
|
|
|
return;
|
|
|
|
|
|
|
|
for (i = context->shadow_root_level; --i >= 0;) {
|
|
|
|
shadow_zero_check->rsvd_bits_mask[0][i] &= ~shadow_me_mask;
|
|
|
|
shadow_zero_check->rsvd_bits_mask[1][i] &= ~shadow_me_mask;
|
|
|
|
}
|
2015-08-05 04:04:24 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* as the comments in reset_shadow_zero_bits_mask() except it
|
|
|
|
* is the shadow page table for intel nested guest.
|
|
|
|
*/
|
|
|
|
static void
|
|
|
|
reset_ept_shadow_zero_bits_mask(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu *context, bool execonly)
|
|
|
|
{
|
|
|
|
__reset_rsvds_bits_mask_ept(&context->shadow_zero_check,
|
|
|
|
boot_cpu_data.x86_phys_bits, execonly);
|
|
|
|
}
|
|
|
|
|
2017-08-24 15:37:25 +00:00
|
|
|
#define BYTE_MASK(access) \
|
|
|
|
((1 & (access) ? 2 : 0) | \
|
|
|
|
(2 & (access) ? 4 : 0) | \
|
|
|
|
(3 & (access) ? 8 : 0) | \
|
|
|
|
(4 & (access) ? 16 : 0) | \
|
|
|
|
(5 & (access) ? 32 : 0) | \
|
|
|
|
(6 & (access) ? 64 : 0) | \
|
|
|
|
(7 & (access) ? 128 : 0))
|
|
|
|
|
|
|
|
|
2015-05-11 14:55:21 +00:00
|
|
|
static void update_permission_bitmask(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu *mmu, bool ept)
|
KVM: MMU: Optimize pte permission checks
walk_addr_generic() permission checks are a maze of branchy code, which is
performed four times per lookup. It depends on the type of access, efer.nxe,
cr0.wp, cr4.smep, and in the near future, cr4.smap.
Optimize this away by precalculating all variants and storing them in a
bitmap. The bitmap is recalculated when rarely-changing variables change
(cr0, cr4) and is indexed by the often-changing variables (page fault error
code, pte access permissions).
The permission check is moved to the end of the loop, otherwise an SMEP
fault could be reported as a false positive, when PDE.U=1 but PTE.U=0.
Noted by Xiao Guangrong.
The result is short, branch-free code.
Reviewed-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2012-09-12 11:52:00 +00:00
|
|
|
{
|
2017-08-24 15:37:25 +00:00
|
|
|
unsigned byte;
|
|
|
|
|
|
|
|
const u8 x = BYTE_MASK(ACC_EXEC_MASK);
|
|
|
|
const u8 w = BYTE_MASK(ACC_WRITE_MASK);
|
|
|
|
const u8 u = BYTE_MASK(ACC_USER_MASK);
|
|
|
|
|
|
|
|
bool cr4_smep = kvm_read_cr4_bits(vcpu, X86_CR4_SMEP) != 0;
|
|
|
|
bool cr4_smap = kvm_read_cr4_bits(vcpu, X86_CR4_SMAP) != 0;
|
|
|
|
bool cr0_wp = is_write_protection(vcpu);
|
KVM: MMU: Optimize pte permission checks
walk_addr_generic() permission checks are a maze of branchy code, which is
performed four times per lookup. It depends on the type of access, efer.nxe,
cr0.wp, cr4.smep, and in the near future, cr4.smap.
Optimize this away by precalculating all variants and storing them in a
bitmap. The bitmap is recalculated when rarely-changing variables change
(cr0, cr4) and is indexed by the often-changing variables (page fault error
code, pte access permissions).
The permission check is moved to the end of the loop, otherwise an SMEP
fault could be reported as a false positive, when PDE.U=1 but PTE.U=0.
Noted by Xiao Guangrong.
The result is short, branch-free code.
Reviewed-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2012-09-12 11:52:00 +00:00
|
|
|
|
|
|
|
for (byte = 0; byte < ARRAY_SIZE(mmu->permissions); ++byte) {
|
2017-08-24 15:37:25 +00:00
|
|
|
unsigned pfec = byte << 1;
|
|
|
|
|
2014-04-01 09:46:34 +00:00
|
|
|
/*
|
2017-08-24 15:37:25 +00:00
|
|
|
* Each "*f" variable has a 1 bit for each UWX value
|
|
|
|
* that causes a fault with the given PFEC.
|
2014-04-01 09:46:34 +00:00
|
|
|
*/
|
KVM: MMU: Optimize pte permission checks
walk_addr_generic() permission checks are a maze of branchy code, which is
performed four times per lookup. It depends on the type of access, efer.nxe,
cr0.wp, cr4.smep, and in the near future, cr4.smap.
Optimize this away by precalculating all variants and storing them in a
bitmap. The bitmap is recalculated when rarely-changing variables change
(cr0, cr4) and is indexed by the often-changing variables (page fault error
code, pte access permissions).
The permission check is moved to the end of the loop, otherwise an SMEP
fault could be reported as a false positive, when PDE.U=1 but PTE.U=0.
Noted by Xiao Guangrong.
The result is short, branch-free code.
Reviewed-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2012-09-12 11:52:00 +00:00
|
|
|
|
2017-08-24 15:37:25 +00:00
|
|
|
/* Faults from writes to non-writable pages */
|
|
|
|
u8 wf = (pfec & PFERR_WRITE_MASK) ? ~w : 0;
|
|
|
|
/* Faults from user mode accesses to supervisor pages */
|
|
|
|
u8 uf = (pfec & PFERR_USER_MASK) ? ~u : 0;
|
|
|
|
/* Faults from fetches of non-executable pages*/
|
|
|
|
u8 ff = (pfec & PFERR_FETCH_MASK) ? ~x : 0;
|
|
|
|
/* Faults from kernel mode fetches of user pages */
|
|
|
|
u8 smepf = 0;
|
|
|
|
/* Faults from kernel mode accesses of user pages */
|
|
|
|
u8 smapf = 0;
|
|
|
|
|
|
|
|
if (!ept) {
|
|
|
|
/* Faults from kernel mode accesses to user pages */
|
|
|
|
u8 kf = (pfec & PFERR_USER_MASK) ? 0 : u;
|
|
|
|
|
|
|
|
/* Not really needed: !nx will cause pte.nx to fault */
|
|
|
|
if (!mmu->nx)
|
|
|
|
ff = 0;
|
|
|
|
|
|
|
|
/* Allow supervisor writes if !cr0.wp */
|
|
|
|
if (!cr0_wp)
|
|
|
|
wf = (pfec & PFERR_USER_MASK) ? wf : 0;
|
|
|
|
|
|
|
|
/* Disallow supervisor fetches of user code if cr4.smep */
|
|
|
|
if (cr4_smep)
|
|
|
|
smepf = (pfec & PFERR_FETCH_MASK) ? kf : 0;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* SMAP:kernel-mode data accesses from user-mode
|
|
|
|
* mappings should fault. A fault is considered
|
|
|
|
* as a SMAP violation if all of the following
|
2018-10-04 15:45:00 +00:00
|
|
|
* conditions are true:
|
2017-08-24 15:37:25 +00:00
|
|
|
* - X86_CR4_SMAP is set in CR4
|
|
|
|
* - A user page is accessed
|
|
|
|
* - The access is not a fetch
|
|
|
|
* - Page fault in kernel mode
|
|
|
|
* - if CPL = 3 or X86_EFLAGS_AC is clear
|
|
|
|
*
|
|
|
|
* Here, we cover the first three conditions.
|
|
|
|
* The fourth is computed dynamically in permission_fault();
|
|
|
|
* PFERR_RSVD_MASK bit will be set in PFEC if the access is
|
|
|
|
* *not* subject to SMAP restrictions.
|
|
|
|
*/
|
|
|
|
if (cr4_smap)
|
|
|
|
smapf = (pfec & (PFERR_RSVD_MASK|PFERR_FETCH_MASK)) ? 0 : kf;
|
KVM: MMU: Optimize pte permission checks
walk_addr_generic() permission checks are a maze of branchy code, which is
performed four times per lookup. It depends on the type of access, efer.nxe,
cr0.wp, cr4.smep, and in the near future, cr4.smap.
Optimize this away by precalculating all variants and storing them in a
bitmap. The bitmap is recalculated when rarely-changing variables change
(cr0, cr4) and is indexed by the often-changing variables (page fault error
code, pte access permissions).
The permission check is moved to the end of the loop, otherwise an SMEP
fault could be reported as a false positive, when PDE.U=1 but PTE.U=0.
Noted by Xiao Guangrong.
The result is short, branch-free code.
Reviewed-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2012-09-12 11:52:00 +00:00
|
|
|
}
|
2017-08-24 15:37:25 +00:00
|
|
|
|
|
|
|
mmu->permissions[byte] = ff | uf | wf | smepf | smapf;
|
KVM: MMU: Optimize pte permission checks
walk_addr_generic() permission checks are a maze of branchy code, which is
performed four times per lookup. It depends on the type of access, efer.nxe,
cr0.wp, cr4.smep, and in the near future, cr4.smap.
Optimize this away by precalculating all variants and storing them in a
bitmap. The bitmap is recalculated when rarely-changing variables change
(cr0, cr4) and is indexed by the often-changing variables (page fault error
code, pte access permissions).
The permission check is moved to the end of the loop, otherwise an SMEP
fault could be reported as a false positive, when PDE.U=1 but PTE.U=0.
Noted by Xiao Guangrong.
The result is short, branch-free code.
Reviewed-by: Xiao Guangrong <xiaoguangrong@linux.vnet.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2012-09-12 11:52:00 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-03-22 08:51:19 +00:00
|
|
|
/*
|
|
|
|
* PKU is an additional mechanism by which the paging controls access to
|
|
|
|
* user-mode addresses based on the value in the PKRU register. Protection
|
|
|
|
* key violations are reported through a bit in the page fault error code.
|
|
|
|
* Unlike other bits of the error code, the PK bit is not known at the
|
|
|
|
* call site of e.g. gva_to_gpa; it must be computed directly in
|
|
|
|
* permission_fault based on two bits of PKRU, on some machine state (CR4,
|
|
|
|
* CR0, EFER, CPL), and on other bits of the error code and the page tables.
|
|
|
|
*
|
|
|
|
* In particular the following conditions come from the error code, the
|
|
|
|
* page tables and the machine state:
|
|
|
|
* - PK is always zero unless CR4.PKE=1 and EFER.LMA=1
|
|
|
|
* - PK is always zero if RSVD=1 (reserved bit set) or F=1 (instruction fetch)
|
|
|
|
* - PK is always zero if U=0 in the page tables
|
|
|
|
* - PKRU.WD is ignored if CR0.WP=0 and the access is a supervisor access.
|
|
|
|
*
|
|
|
|
* The PKRU bitmask caches the result of these four conditions. The error
|
|
|
|
* code (minus the P bit) and the page table's U bit form an index into the
|
|
|
|
* PKRU bitmask. Two bits of the PKRU bitmask are then extracted and ANDed
|
|
|
|
* with the two bits of the PKRU register corresponding to the protection key.
|
|
|
|
* For the first three conditions above the bits will be 00, thus masking
|
|
|
|
* away both AD and WD. For all reads or if the last condition holds, WD
|
|
|
|
* only will be masked away.
|
|
|
|
*/
|
|
|
|
static void update_pkru_bitmask(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu,
|
|
|
|
bool ept)
|
|
|
|
{
|
|
|
|
unsigned bit;
|
|
|
|
bool wp;
|
|
|
|
|
|
|
|
if (ept) {
|
|
|
|
mmu->pkru_mask = 0;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* PKEY is enabled only if CR4.PKE and EFER.LMA are both set. */
|
|
|
|
if (!kvm_read_cr4_bits(vcpu, X86_CR4_PKE) || !is_long_mode(vcpu)) {
|
|
|
|
mmu->pkru_mask = 0;
|
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
|
|
|
wp = is_write_protection(vcpu);
|
|
|
|
|
|
|
|
for (bit = 0; bit < ARRAY_SIZE(mmu->permissions); ++bit) {
|
|
|
|
unsigned pfec, pkey_bits;
|
|
|
|
bool check_pkey, check_write, ff, uf, wf, pte_user;
|
|
|
|
|
|
|
|
pfec = bit << 1;
|
|
|
|
ff = pfec & PFERR_FETCH_MASK;
|
|
|
|
uf = pfec & PFERR_USER_MASK;
|
|
|
|
wf = pfec & PFERR_WRITE_MASK;
|
|
|
|
|
|
|
|
/* PFEC.RSVD is replaced by ACC_USER_MASK. */
|
|
|
|
pte_user = pfec & PFERR_RSVD_MASK;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Only need to check the access which is not an
|
|
|
|
* instruction fetch and is to a user page.
|
|
|
|
*/
|
|
|
|
check_pkey = (!ff && pte_user);
|
|
|
|
/*
|
|
|
|
* write access is controlled by PKRU if it is a
|
|
|
|
* user access or CR0.WP = 1.
|
|
|
|
*/
|
|
|
|
check_write = check_pkey && wf && (uf || wp);
|
|
|
|
|
|
|
|
/* PKRU.AD stops both read and write access. */
|
|
|
|
pkey_bits = !!check_pkey;
|
|
|
|
/* PKRU.WD stops write access. */
|
|
|
|
pkey_bits |= (!!check_write) << 1;
|
|
|
|
|
|
|
|
mmu->pkru_mask |= (pkey_bits & 3) << pfec;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-02-23 11:51:19 +00:00
|
|
|
static void update_last_nonleaf_level(struct kvm_vcpu *vcpu, struct kvm_mmu *mmu)
|
2012-09-12 17:46:56 +00:00
|
|
|
{
|
2016-02-23 11:51:19 +00:00
|
|
|
unsigned root_level = mmu->root_level;
|
|
|
|
|
|
|
|
mmu->last_nonleaf_level = root_level;
|
|
|
|
if (root_level == PT32_ROOT_LEVEL && is_pse(vcpu))
|
|
|
|
mmu->last_nonleaf_level++;
|
2012-09-12 17:46:56 +00:00
|
|
|
}
|
|
|
|
|
2013-10-02 14:56:13 +00:00
|
|
|
static void paging64_init_context_common(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu *context,
|
|
|
|
int level)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2010-09-10 15:31:01 +00:00
|
|
|
context->nx = is_nx(vcpu);
|
2012-03-05 15:53:06 +00:00
|
|
|
context->root_level = level;
|
2010-09-10 15:31:01 +00:00
|
|
|
|
2012-03-05 15:53:06 +00:00
|
|
|
reset_rsvds_bits_mask(vcpu, context);
|
2013-08-06 09:00:32 +00:00
|
|
|
update_permission_bitmask(vcpu, context, false);
|
2016-03-22 08:51:19 +00:00
|
|
|
update_pkru_bitmask(vcpu, context, false);
|
2016-02-23 11:51:19 +00:00
|
|
|
update_last_nonleaf_level(vcpu, context);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2013-10-02 14:56:16 +00:00
|
|
|
MMU_WARN_ON(!is_pae(vcpu));
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
context->page_fault = paging64_page_fault;
|
|
|
|
context->gva_to_gpa = paging64_gva_to_gpa;
|
2008-09-23 16:18:33 +00:00
|
|
|
context->sync_page = paging64_sync_page;
|
2008-09-23 16:18:35 +00:00
|
|
|
context->invlpg = paging64_invlpg;
|
2011-03-09 07:43:51 +00:00
|
|
|
context->update_pte = paging64_update_pte;
|
2007-01-06 00:36:40 +00:00
|
|
|
context->shadow_root_level = level;
|
2010-09-10 15:30:39 +00:00
|
|
|
context->direct_map = false;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2013-10-02 14:56:13 +00:00
|
|
|
static void paging64_init_context(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu *context)
|
2007-01-06 00:36:40 +00:00
|
|
|
{
|
2017-08-24 12:27:55 +00:00
|
|
|
int root_level = is_la57_mode(vcpu) ?
|
|
|
|
PT64_ROOT_5LEVEL : PT64_ROOT_4LEVEL;
|
|
|
|
|
|
|
|
paging64_init_context_common(vcpu, context, root_level);
|
2007-01-06 00:36:40 +00:00
|
|
|
}
|
|
|
|
|
2013-10-02 14:56:13 +00:00
|
|
|
static void paging32_init_context(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu *context)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2010-09-10 15:31:01 +00:00
|
|
|
context->nx = false;
|
2012-03-05 15:53:06 +00:00
|
|
|
context->root_level = PT32_ROOT_LEVEL;
|
2010-09-10 15:31:01 +00:00
|
|
|
|
2012-03-05 15:53:06 +00:00
|
|
|
reset_rsvds_bits_mask(vcpu, context);
|
2013-08-06 09:00:32 +00:00
|
|
|
update_permission_bitmask(vcpu, context, false);
|
2016-03-22 08:51:19 +00:00
|
|
|
update_pkru_bitmask(vcpu, context, false);
|
2016-02-23 11:51:19 +00:00
|
|
|
update_last_nonleaf_level(vcpu, context);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
|
|
|
context->page_fault = paging32_page_fault;
|
|
|
|
context->gva_to_gpa = paging32_gva_to_gpa;
|
2008-09-23 16:18:33 +00:00
|
|
|
context->sync_page = paging32_sync_page;
|
2008-09-23 16:18:35 +00:00
|
|
|
context->invlpg = paging32_invlpg;
|
2011-03-09 07:43:51 +00:00
|
|
|
context->update_pte = paging32_update_pte;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
context->shadow_root_level = PT32E_ROOT_LEVEL;
|
2010-09-10 15:30:39 +00:00
|
|
|
context->direct_map = false;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2013-10-02 14:56:13 +00:00
|
|
|
static void paging32E_init_context(struct kvm_vcpu *vcpu,
|
|
|
|
struct kvm_mmu *context)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2013-10-02 14:56:13 +00:00
|
|
|
paging64_init_context_common(vcpu, context, PT32E_ROOT_LEVEL);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2018-10-08 19:28:11 +00:00
|
|
|
static union kvm_mmu_extended_role kvm_calc_mmu_role_ext(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
|
|
|
union kvm_mmu_extended_role ext = {0};
|
|
|
|
|
2018-10-08 19:28:12 +00:00
|
|
|
ext.cr0_pg = !!is_paging(vcpu);
|
2018-10-08 19:28:11 +00:00
|
|
|
ext.cr4_smep = !!kvm_read_cr4_bits(vcpu, X86_CR4_SMEP);
|
|
|
|
ext.cr4_smap = !!kvm_read_cr4_bits(vcpu, X86_CR4_SMAP);
|
|
|
|
ext.cr4_pse = !!is_pse(vcpu);
|
|
|
|
ext.cr4_pke = !!kvm_read_cr4_bits(vcpu, X86_CR4_PKE);
|
2018-10-08 19:28:12 +00:00
|
|
|
ext.cr4_la57 = !!kvm_read_cr4_bits(vcpu, X86_CR4_LA57);
|
2018-10-08 19:28:11 +00:00
|
|
|
|
|
|
|
ext.valid = 1;
|
|
|
|
|
|
|
|
return ext;
|
|
|
|
}
|
|
|
|
|
2018-10-08 19:28:12 +00:00
|
|
|
static union kvm_mmu_role kvm_calc_mmu_role_common(struct kvm_vcpu *vcpu,
|
|
|
|
bool base_only)
|
|
|
|
{
|
|
|
|
union kvm_mmu_role role = {0};
|
|
|
|
|
|
|
|
role.base.access = ACC_ALL;
|
|
|
|
role.base.nxe = !!is_nx(vcpu);
|
|
|
|
role.base.cr4_pae = !!is_pae(vcpu);
|
|
|
|
role.base.cr0_wp = is_write_protection(vcpu);
|
|
|
|
role.base.smm = is_smm(vcpu);
|
|
|
|
role.base.guest_mode = is_guest_mode(vcpu);
|
|
|
|
|
|
|
|
if (base_only)
|
|
|
|
return role;
|
|
|
|
|
|
|
|
role.ext = kvm_calc_mmu_role_ext(vcpu);
|
|
|
|
|
|
|
|
return role;
|
|
|
|
}
|
|
|
|
|
|
|
|
static union kvm_mmu_role
|
|
|
|
kvm_calc_tdp_mmu_root_page_role(struct kvm_vcpu *vcpu, bool base_only)
|
2018-06-27 21:59:07 +00:00
|
|
|
{
|
2018-10-08 19:28:12 +00:00
|
|
|
union kvm_mmu_role role = kvm_calc_mmu_role_common(vcpu, base_only);
|
2018-06-27 21:59:07 +00:00
|
|
|
|
2018-10-08 19:28:12 +00:00
|
|
|
role.base.ad_disabled = (shadow_accessed_mask == 0);
|
|
|
|
role.base.level = kvm_x86_ops->get_tdp_level(vcpu);
|
|
|
|
role.base.direct = true;
|
2018-06-27 21:59:07 +00:00
|
|
|
|
|
|
|
return role;
|
|
|
|
}
|
|
|
|
|
2013-10-02 14:56:13 +00:00
|
|
|
static void init_kvm_tdp_mmu(struct kvm_vcpu *vcpu)
|
2008-02-07 12:47:44 +00:00
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
struct kvm_mmu *context = vcpu->arch.mmu;
|
2018-10-08 19:28:12 +00:00
|
|
|
union kvm_mmu_role new_role =
|
|
|
|
kvm_calc_tdp_mmu_root_page_role(vcpu, false);
|
2008-02-07 12:47:44 +00:00
|
|
|
|
2018-10-08 19:28:12 +00:00
|
|
|
new_role.base.word &= mmu_base_role_mask.word;
|
|
|
|
if (new_role.as_u64 == context->mmu_role.as_u64)
|
|
|
|
return;
|
|
|
|
|
|
|
|
context->mmu_role.as_u64 = new_role.as_u64;
|
2008-02-07 12:47:44 +00:00
|
|
|
context->page_fault = tdp_page_fault;
|
2008-09-23 16:18:33 +00:00
|
|
|
context->sync_page = nonpaging_sync_page;
|
2008-09-23 16:18:35 +00:00
|
|
|
context->invlpg = nonpaging_invlpg;
|
2011-03-09 07:43:51 +00:00
|
|
|
context->update_pte = nonpaging_update_pte;
|
2017-08-24 12:27:55 +00:00
|
|
|
context->shadow_root_level = kvm_x86_ops->get_tdp_level(vcpu);
|
2010-09-10 15:30:39 +00:00
|
|
|
context->direct_map = true;
|
2010-09-10 15:30:41 +00:00
|
|
|
context->set_cr3 = kvm_x86_ops->set_tdp_cr3;
|
2010-09-10 15:30:42 +00:00
|
|
|
context->get_cr3 = get_cr3;
|
2011-07-28 08:36:17 +00:00
|
|
|
context->get_pdptr = kvm_pdptr_read;
|
2010-09-10 15:30:43 +00:00
|
|
|
context->inject_page_fault = kvm_inject_page_fault;
|
2008-02-07 12:47:44 +00:00
|
|
|
|
|
|
|
if (!is_paging(vcpu)) {
|
2010-09-10 15:31:01 +00:00
|
|
|
context->nx = false;
|
2008-02-07 12:47:44 +00:00
|
|
|
context->gva_to_gpa = nonpaging_gva_to_gpa;
|
|
|
|
context->root_level = 0;
|
|
|
|
} else if (is_long_mode(vcpu)) {
|
2010-09-10 15:31:01 +00:00
|
|
|
context->nx = is_nx(vcpu);
|
2017-08-24 12:27:55 +00:00
|
|
|
context->root_level = is_la57_mode(vcpu) ?
|
|
|
|
PT64_ROOT_5LEVEL : PT64_ROOT_4LEVEL;
|
2012-03-05 15:53:06 +00:00
|
|
|
reset_rsvds_bits_mask(vcpu, context);
|
|
|
|
context->gva_to_gpa = paging64_gva_to_gpa;
|
2008-02-07 12:47:44 +00:00
|
|
|
} else if (is_pae(vcpu)) {
|
2010-09-10 15:31:01 +00:00
|
|
|
context->nx = is_nx(vcpu);
|
2008-02-07 12:47:44 +00:00
|
|
|
context->root_level = PT32E_ROOT_LEVEL;
|
2012-03-05 15:53:06 +00:00
|
|
|
reset_rsvds_bits_mask(vcpu, context);
|
|
|
|
context->gva_to_gpa = paging64_gva_to_gpa;
|
2008-02-07 12:47:44 +00:00
|
|
|
} else {
|
2010-09-10 15:31:01 +00:00
|
|
|
context->nx = false;
|
2008-02-07 12:47:44 +00:00
|
|
|
context->root_level = PT32_ROOT_LEVEL;
|
2012-03-05 15:53:06 +00:00
|
|
|
reset_rsvds_bits_mask(vcpu, context);
|
|
|
|
context->gva_to_gpa = paging32_gva_to_gpa;
|
2008-02-07 12:47:44 +00:00
|
|
|
}
|
|
|
|
|
2013-08-06 09:00:32 +00:00
|
|
|
update_permission_bitmask(vcpu, context, false);
|
2016-03-22 08:51:19 +00:00
|
|
|
update_pkru_bitmask(vcpu, context, false);
|
2016-02-23 11:51:19 +00:00
|
|
|
update_last_nonleaf_level(vcpu, context);
|
2015-08-05 04:04:24 +00:00
|
|
|
reset_tdp_shadow_zero_bits_mask(vcpu, context);
|
2008-02-07 12:47:44 +00:00
|
|
|
}
|
|
|
|
|
2018-10-08 19:28:12 +00:00
|
|
|
static union kvm_mmu_role
|
|
|
|
kvm_calc_shadow_mmu_root_page_role(struct kvm_vcpu *vcpu, bool base_only)
|
|
|
|
{
|
|
|
|
union kvm_mmu_role role = kvm_calc_mmu_role_common(vcpu, base_only);
|
|
|
|
|
|
|
|
role.base.smep_andnot_wp = role.ext.cr4_smep &&
|
|
|
|
!is_write_protection(vcpu);
|
|
|
|
role.base.smap_andnot_wp = role.ext.cr4_smap &&
|
|
|
|
!is_write_protection(vcpu);
|
|
|
|
role.base.direct = !is_paging(vcpu);
|
2018-06-27 21:59:07 +00:00
|
|
|
|
|
|
|
if (!is_long_mode(vcpu))
|
2018-10-08 19:28:12 +00:00
|
|
|
role.base.level = PT32E_ROOT_LEVEL;
|
2018-06-27 21:59:07 +00:00
|
|
|
else if (is_la57_mode(vcpu))
|
2018-10-08 19:28:12 +00:00
|
|
|
role.base.level = PT64_ROOT_5LEVEL;
|
2018-06-27 21:59:07 +00:00
|
|
|
else
|
2018-10-08 19:28:12 +00:00
|
|
|
role.base.level = PT64_ROOT_4LEVEL;
|
2018-06-27 21:59:07 +00:00
|
|
|
|
|
|
|
return role;
|
|
|
|
}
|
|
|
|
|
|
|
|
void kvm_init_shadow_mmu(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
struct kvm_mmu *context = vcpu->arch.mmu;
|
2018-10-08 19:28:12 +00:00
|
|
|
union kvm_mmu_role new_role =
|
|
|
|
kvm_calc_shadow_mmu_root_page_role(vcpu, false);
|
|
|
|
|
|
|
|
new_role.base.word &= mmu_base_role_mask.word;
|
|
|
|
if (new_role.as_u64 == context->mmu_role.as_u64)
|
|
|
|
return;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
|
|
|
if (!is_paging(vcpu))
|
2013-10-02 14:56:13 +00:00
|
|
|
nonpaging_init_context(vcpu, context);
|
2006-12-30 00:49:37 +00:00
|
|
|
else if (is_long_mode(vcpu))
|
2013-10-02 14:56:13 +00:00
|
|
|
paging64_init_context(vcpu, context);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
else if (is_pae(vcpu))
|
2013-10-02 14:56:13 +00:00
|
|
|
paging32E_init_context(vcpu, context);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
else
|
2013-10-02 14:56:13 +00:00
|
|
|
paging32_init_context(vcpu, context);
|
2008-12-21 17:20:09 +00:00
|
|
|
|
2018-10-08 19:28:12 +00:00
|
|
|
context->mmu_role.as_u64 = new_role.as_u64;
|
2015-08-05 04:04:24 +00:00
|
|
|
reset_shadow_zero_bits_mask(vcpu, context);
|
2010-09-10 15:30:44 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_init_shadow_mmu);
|
|
|
|
|
2018-10-08 19:28:11 +00:00
|
|
|
static union kvm_mmu_role
|
|
|
|
kvm_calc_shadow_ept_root_page_role(struct kvm_vcpu *vcpu, bool accessed_dirty,
|
|
|
|
bool execonly)
|
2018-06-27 21:59:07 +00:00
|
|
|
{
|
2018-10-08 19:28:11 +00:00
|
|
|
union kvm_mmu_role role;
|
2018-10-08 19:28:08 +00:00
|
|
|
|
2018-10-08 19:28:11 +00:00
|
|
|
/* Base role is inherited from root_mmu */
|
|
|
|
role.base.word = vcpu->arch.root_mmu.mmu_role.base.word;
|
|
|
|
role.ext = kvm_calc_mmu_role_ext(vcpu);
|
2018-06-27 21:59:07 +00:00
|
|
|
|
2018-10-08 19:28:11 +00:00
|
|
|
role.base.level = PT64_ROOT_4LEVEL;
|
|
|
|
role.base.direct = false;
|
|
|
|
role.base.ad_disabled = !accessed_dirty;
|
|
|
|
role.base.guest_mode = true;
|
|
|
|
role.base.access = ACC_ALL;
|
2018-06-27 21:59:07 +00:00
|
|
|
|
2018-10-08 19:28:11 +00:00
|
|
|
role.ext.execonly = execonly;
|
2018-06-27 21:59:07 +00:00
|
|
|
|
|
|
|
return role;
|
|
|
|
}
|
|
|
|
|
2017-03-30 09:55:30 +00:00
|
|
|
void kvm_init_shadow_ept_mmu(struct kvm_vcpu *vcpu, bool execonly,
|
2018-06-27 21:59:11 +00:00
|
|
|
bool accessed_dirty, gpa_t new_eptp)
|
2013-08-05 08:07:16 +00:00
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
struct kvm_mmu *context = vcpu->arch.mmu;
|
2018-10-08 19:28:11 +00:00
|
|
|
union kvm_mmu_role new_role =
|
|
|
|
kvm_calc_shadow_ept_root_page_role(vcpu, accessed_dirty,
|
|
|
|
execonly);
|
|
|
|
|
|
|
|
__kvm_mmu_new_cr3(vcpu, new_eptp, new_role.base, false);
|
|
|
|
|
|
|
|
new_role.base.word &= mmu_base_role_mask.word;
|
|
|
|
if (new_role.as_u64 == context->mmu_role.as_u64)
|
|
|
|
return;
|
2013-10-02 14:56:14 +00:00
|
|
|
|
2017-08-24 12:27:55 +00:00
|
|
|
context->shadow_root_level = PT64_ROOT_4LEVEL;
|
2013-08-05 08:07:16 +00:00
|
|
|
|
|
|
|
context->nx = true;
|
2017-03-30 09:55:30 +00:00
|
|
|
context->ept_ad = accessed_dirty;
|
2013-08-05 08:07:16 +00:00
|
|
|
context->page_fault = ept_page_fault;
|
|
|
|
context->gva_to_gpa = ept_gva_to_gpa;
|
|
|
|
context->sync_page = ept_sync_page;
|
|
|
|
context->invlpg = ept_invlpg;
|
|
|
|
context->update_pte = ept_update_pte;
|
2017-08-24 12:27:55 +00:00
|
|
|
context->root_level = PT64_ROOT_4LEVEL;
|
2013-08-05 08:07:16 +00:00
|
|
|
context->direct_map = false;
|
2018-10-08 19:28:11 +00:00
|
|
|
context->mmu_role.as_u64 = new_role.as_u64;
|
2018-10-08 19:28:06 +00:00
|
|
|
|
2013-08-05 08:07:16 +00:00
|
|
|
update_permission_bitmask(vcpu, context, true);
|
2016-03-22 08:51:19 +00:00
|
|
|
update_pkru_bitmask(vcpu, context, true);
|
2017-10-05 09:10:22 +00:00
|
|
|
update_last_nonleaf_level(vcpu, context);
|
2013-08-05 08:07:16 +00:00
|
|
|
reset_rsvds_bits_mask_ept(vcpu, context, execonly);
|
2015-08-05 04:04:24 +00:00
|
|
|
reset_ept_shadow_zero_bits_mask(vcpu, context, execonly);
|
2013-08-05 08:07:16 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_init_shadow_ept_mmu);
|
|
|
|
|
2013-10-02 14:56:13 +00:00
|
|
|
static void init_kvm_softmmu(struct kvm_vcpu *vcpu)
|
2010-09-10 15:30:44 +00:00
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
struct kvm_mmu *context = vcpu->arch.mmu;
|
2013-10-02 14:56:14 +00:00
|
|
|
|
|
|
|
kvm_init_shadow_mmu(vcpu);
|
|
|
|
context->set_cr3 = kvm_x86_ops->set_cr3;
|
|
|
|
context->get_cr3 = get_cr3;
|
|
|
|
context->get_pdptr = kvm_pdptr_read;
|
|
|
|
context->inject_page_fault = kvm_inject_page_fault;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2013-10-02 14:56:13 +00:00
|
|
|
static void init_kvm_nested_mmu(struct kvm_vcpu *vcpu)
|
2010-09-10 15:30:54 +00:00
|
|
|
{
|
2018-10-08 19:28:13 +00:00
|
|
|
union kvm_mmu_role new_role = kvm_calc_mmu_role_common(vcpu, false);
|
2010-09-10 15:30:54 +00:00
|
|
|
struct kvm_mmu *g_context = &vcpu->arch.nested_mmu;
|
|
|
|
|
2018-10-08 19:28:13 +00:00
|
|
|
new_role.base.word &= mmu_base_role_mask.word;
|
|
|
|
if (new_role.as_u64 == g_context->mmu_role.as_u64)
|
|
|
|
return;
|
|
|
|
|
|
|
|
g_context->mmu_role.as_u64 = new_role.as_u64;
|
2010-09-10 15:30:54 +00:00
|
|
|
g_context->get_cr3 = get_cr3;
|
2011-07-28 08:36:17 +00:00
|
|
|
g_context->get_pdptr = kvm_pdptr_read;
|
2010-09-10 15:30:54 +00:00
|
|
|
g_context->inject_page_fault = kvm_inject_page_fault;
|
|
|
|
|
|
|
|
/*
|
2018-10-08 19:28:05 +00:00
|
|
|
* Note that arch.mmu->gva_to_gpa translates l2_gpa to l1_gpa using
|
2015-12-30 16:26:17 +00:00
|
|
|
* L1's nested page tables (e.g. EPT12). The nested translation
|
|
|
|
* of l2_gva to l1_gpa is done by arch.nested_mmu.gva_to_gpa using
|
|
|
|
* L2's page tables as the first level of translation and L1's
|
|
|
|
* nested page tables as the second level of translation. Basically
|
|
|
|
* the gva_to_gpa functions between mmu and nested_mmu are swapped.
|
2010-09-10 15:30:54 +00:00
|
|
|
*/
|
|
|
|
if (!is_paging(vcpu)) {
|
2010-09-10 15:31:01 +00:00
|
|
|
g_context->nx = false;
|
2010-09-10 15:30:54 +00:00
|
|
|
g_context->root_level = 0;
|
|
|
|
g_context->gva_to_gpa = nonpaging_gva_to_gpa_nested;
|
|
|
|
} else if (is_long_mode(vcpu)) {
|
2010-09-10 15:31:01 +00:00
|
|
|
g_context->nx = is_nx(vcpu);
|
2017-08-24 12:27:55 +00:00
|
|
|
g_context->root_level = is_la57_mode(vcpu) ?
|
|
|
|
PT64_ROOT_5LEVEL : PT64_ROOT_4LEVEL;
|
2012-03-05 15:53:06 +00:00
|
|
|
reset_rsvds_bits_mask(vcpu, g_context);
|
2010-09-10 15:30:54 +00:00
|
|
|
g_context->gva_to_gpa = paging64_gva_to_gpa_nested;
|
|
|
|
} else if (is_pae(vcpu)) {
|
2010-09-10 15:31:01 +00:00
|
|
|
g_context->nx = is_nx(vcpu);
|
2010-09-10 15:30:54 +00:00
|
|
|
g_context->root_level = PT32E_ROOT_LEVEL;
|
2012-03-05 15:53:06 +00:00
|
|
|
reset_rsvds_bits_mask(vcpu, g_context);
|
2010-09-10 15:30:54 +00:00
|
|
|
g_context->gva_to_gpa = paging64_gva_to_gpa_nested;
|
|
|
|
} else {
|
2010-09-10 15:31:01 +00:00
|
|
|
g_context->nx = false;
|
2010-09-10 15:30:54 +00:00
|
|
|
g_context->root_level = PT32_ROOT_LEVEL;
|
2012-03-05 15:53:06 +00:00
|
|
|
reset_rsvds_bits_mask(vcpu, g_context);
|
2010-09-10 15:30:54 +00:00
|
|
|
g_context->gva_to_gpa = paging32_gva_to_gpa_nested;
|
|
|
|
}
|
|
|
|
|
2013-08-06 09:00:32 +00:00
|
|
|
update_permission_bitmask(vcpu, g_context, false);
|
2016-03-22 08:51:19 +00:00
|
|
|
update_pkru_bitmask(vcpu, g_context, false);
|
2016-02-23 11:51:19 +00:00
|
|
|
update_last_nonleaf_level(vcpu, g_context);
|
2010-09-10 15:30:54 +00:00
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:10 +00:00
|
|
|
void kvm_init_mmu(struct kvm_vcpu *vcpu, bool reset_roots)
|
2008-02-07 12:47:44 +00:00
|
|
|
{
|
2018-06-27 21:59:10 +00:00
|
|
|
if (reset_roots) {
|
2018-06-27 21:59:20 +00:00
|
|
|
uint i;
|
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->root_hpa = INVALID_PAGE;
|
2018-06-27 21:59:20 +00:00
|
|
|
|
|
|
|
for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++)
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->prev_roots[i] = KVM_MMU_ROOT_INFO_INVALID;
|
2018-06-27 21:59:10 +00:00
|
|
|
}
|
|
|
|
|
2010-09-10 15:30:54 +00:00
|
|
|
if (mmu_is_nested(vcpu))
|
2014-12-23 12:39:46 +00:00
|
|
|
init_kvm_nested_mmu(vcpu);
|
2010-09-10 15:30:54 +00:00
|
|
|
else if (tdp_enabled)
|
2014-12-23 12:39:46 +00:00
|
|
|
init_kvm_tdp_mmu(vcpu);
|
2008-02-07 12:47:44 +00:00
|
|
|
else
|
2014-12-23 12:39:46 +00:00
|
|
|
init_kvm_softmmu(vcpu);
|
2008-02-07 12:47:44 +00:00
|
|
|
}
|
2018-06-27 21:59:10 +00:00
|
|
|
EXPORT_SYMBOL_GPL(kvm_init_mmu);
|
2008-02-07 12:47:44 +00:00
|
|
|
|
2018-06-27 21:59:07 +00:00
|
|
|
static union kvm_mmu_page_role
|
|
|
|
kvm_mmu_calc_root_page_role(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2018-10-08 19:28:12 +00:00
|
|
|
union kvm_mmu_role role;
|
|
|
|
|
2018-06-27 21:59:07 +00:00
|
|
|
if (tdp_enabled)
|
2018-10-08 19:28:12 +00:00
|
|
|
role = kvm_calc_tdp_mmu_root_page_role(vcpu, true);
|
2018-06-27 21:59:07 +00:00
|
|
|
else
|
2018-10-08 19:28:12 +00:00
|
|
|
role = kvm_calc_shadow_mmu_root_page_role(vcpu, true);
|
|
|
|
|
|
|
|
return role.base;
|
2018-06-27 21:59:07 +00:00
|
|
|
}
|
2008-02-07 12:47:44 +00:00
|
|
|
|
2013-10-02 14:56:13 +00:00
|
|
|
void kvm_mmu_reset_context(struct kvm_vcpu *vcpu)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2013-10-02 14:56:12 +00:00
|
|
|
kvm_mmu_unload(vcpu);
|
2018-06-27 21:59:10 +00:00
|
|
|
kvm_init_mmu(vcpu, true);
|
2007-06-04 12:58:30 +00:00
|
|
|
}
|
2007-10-10 06:26:45 +00:00
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_reset_context);
|
2007-06-04 12:58:30 +00:00
|
|
|
|
|
|
|
int kvm_mmu_load(struct kvm_vcpu *vcpu)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2007-01-06 00:36:53 +00:00
|
|
|
int r;
|
|
|
|
|
2007-01-06 00:36:54 +00:00
|
|
|
r = mmu_topup_memory_caches(vcpu);
|
2007-06-04 12:58:30 +00:00
|
|
|
if (r)
|
|
|
|
goto out;
|
2009-05-12 21:55:45 +00:00
|
|
|
r = mmu_alloc_roots(vcpu);
|
2013-05-09 06:45:12 +00:00
|
|
|
kvm_mmu_sync_roots(vcpu);
|
2009-05-12 21:55:45 +00:00
|
|
|
if (r)
|
|
|
|
goto out;
|
2018-06-27 21:59:08 +00:00
|
|
|
kvm_mmu_load_cr3(vcpu);
|
2018-06-27 21:59:12 +00:00
|
|
|
kvm_x86_ops->tlb_flush(vcpu, true);
|
2007-01-06 00:36:53 +00:00
|
|
|
out:
|
|
|
|
return r;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
2007-06-04 12:58:30 +00:00
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_load);
|
|
|
|
|
|
|
|
void kvm_mmu_unload(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2018-10-08 19:28:08 +00:00
|
|
|
kvm_mmu_free_roots(vcpu, &vcpu->arch.root_mmu, KVM_MMU_ROOTS_ALL);
|
|
|
|
WARN_ON(VALID_PAGE(vcpu->arch.root_mmu.root_hpa));
|
|
|
|
kvm_mmu_free_roots(vcpu, &vcpu->arch.guest_mmu, KVM_MMU_ROOTS_ALL);
|
|
|
|
WARN_ON(VALID_PAGE(vcpu->arch.guest_mmu.root_hpa));
|
2007-06-04 12:58:30 +00:00
|
|
|
}
|
2010-09-10 15:31:03 +00:00
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_unload);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2007-05-01 13:53:31 +00:00
|
|
|
static void mmu_pte_write_new_pte(struct kvm_vcpu *vcpu,
|
2011-03-28 02:29:27 +00:00
|
|
|
struct kvm_mmu_page *sp, u64 *spte,
|
|
|
|
const void *new)
|
2007-05-01 13:53:31 +00:00
|
|
|
{
|
2008-06-11 23:32:40 +00:00
|
|
|
if (sp->role.level != PT_PAGE_TABLE_LEVEL) {
|
2009-07-27 14:30:46 +00:00
|
|
|
++vcpu->kvm->stat.mmu_pde_zapped;
|
|
|
|
return;
|
2008-06-11 23:32:40 +00:00
|
|
|
}
|
2007-05-01 13:53:31 +00:00
|
|
|
|
2007-11-18 14:37:07 +00:00
|
|
|
++vcpu->kvm->stat.mmu_pte_updated;
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->update_pte(vcpu, sp, spte, new);
|
2007-05-01 13:53:31 +00:00
|
|
|
}
|
|
|
|
|
2007-11-21 00:06:21 +00:00
|
|
|
static bool need_remote_flush(u64 old, u64 new)
|
|
|
|
{
|
|
|
|
if (!is_shadow_present_pte(old))
|
|
|
|
return false;
|
|
|
|
if (!is_shadow_present_pte(new))
|
|
|
|
return true;
|
|
|
|
if ((old ^ new) & PT64_BASE_ADDR_MASK)
|
|
|
|
return true;
|
2013-08-05 08:07:14 +00:00
|
|
|
old ^= shadow_nx_mask;
|
|
|
|
new ^= shadow_nx_mask;
|
2007-11-21 00:06:21 +00:00
|
|
|
return (old & ~new & PT64_PERM_MASK) != 0;
|
|
|
|
}
|
|
|
|
|
2011-09-22 08:57:23 +00:00
|
|
|
static u64 mmu_pte_write_fetch_gpte(struct kvm_vcpu *vcpu, gpa_t *gpa,
|
2018-10-31 21:53:57 +00:00
|
|
|
int *bytes)
|
2007-01-06 00:36:44 +00:00
|
|
|
{
|
2018-10-31 21:53:57 +00:00
|
|
|
u64 gentry = 0;
|
2011-09-22 08:57:23 +00:00
|
|
|
int r;
|
2010-03-15 11:59:53 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Assume that the pte write on a page table of the same type
|
2011-03-04 11:00:00 +00:00
|
|
|
* as the current vcpu paging mode since we update the sptes only
|
|
|
|
* when they have the same mode.
|
2010-03-15 11:59:53 +00:00
|
|
|
*/
|
2011-09-22 08:57:23 +00:00
|
|
|
if (is_pae(vcpu) && *bytes == 4) {
|
2010-03-15 11:59:53 +00:00
|
|
|
/* Handle a 32-bit guest writing two halves of a 64-bit gpte */
|
2011-09-22 08:57:23 +00:00
|
|
|
*gpa &= ~(gpa_t)7;
|
|
|
|
*bytes = 8;
|
2010-03-15 11:59:57 +00:00
|
|
|
}
|
|
|
|
|
2018-10-31 21:53:57 +00:00
|
|
|
if (*bytes == 4 || *bytes == 8) {
|
|
|
|
r = kvm_vcpu_read_guest_atomic(vcpu, *gpa, &gentry, *bytes);
|
|
|
|
if (r)
|
|
|
|
gentry = 0;
|
2010-03-15 11:59:53 +00:00
|
|
|
}
|
|
|
|
|
2011-09-22 08:57:23 +00:00
|
|
|
return gentry;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* If we're seeing too many writes to a page, it may no longer be a page table,
|
|
|
|
* or we may be forking, in which case it is better to unmap the page.
|
|
|
|
*/
|
2011-12-16 10:18:10 +00:00
|
|
|
static bool detect_write_flooding(struct kvm_mmu_page *sp)
|
2011-09-22 08:57:23 +00:00
|
|
|
{
|
KVM: MMU: improve write flooding detected
Detecting write-flooding does not work well, when we handle page written, if
the last speculative spte is not accessed, we treat the page is
write-flooding, however, we can speculative spte on many path, such as pte
prefetch, page synced, that means the last speculative spte may be not point
to the written page and the written page can be accessed via other sptes, so
depends on the Accessed bit of the last speculative spte is not enough
Instead of detected page accessed, we can detect whether the spte is accessed
after it is written, if the spte is not accessed but it is written frequently,
we treat is not a page table or it not used for a long time
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2011-09-22 08:58:36 +00:00
|
|
|
/*
|
|
|
|
* Skip write-flooding detected for the sp whose level is 1, because
|
|
|
|
* it can become unsync, then the guest page is not write-protected.
|
|
|
|
*/
|
2012-04-18 10:24:29 +00:00
|
|
|
if (sp->role.level == PT_PAGE_TABLE_LEVEL)
|
KVM: MMU: improve write flooding detected
Detecting write-flooding does not work well, when we handle page written, if
the last speculative spte is not accessed, we treat the page is
write-flooding, however, we can speculative spte on many path, such as pte
prefetch, page synced, that means the last speculative spte may be not point
to the written page and the written page can be accessed via other sptes, so
depends on the Accessed bit of the last speculative spte is not enough
Instead of detected page accessed, we can detect whether the spte is accessed
after it is written, if the spte is not accessed but it is written frequently,
we treat is not a page table or it not used for a long time
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2011-09-22 08:58:36 +00:00
|
|
|
return false;
|
2010-04-16 08:35:54 +00:00
|
|
|
|
2016-02-24 09:51:12 +00:00
|
|
|
atomic_inc(&sp->write_flooding_count);
|
|
|
|
return atomic_read(&sp->write_flooding_count) >= 3;
|
2011-09-22 08:57:23 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Misaligned accesses are too much trouble to fix up; also, they usually
|
|
|
|
* indicate a page is not used as a page table.
|
|
|
|
*/
|
|
|
|
static bool detect_write_misaligned(struct kvm_mmu_page *sp, gpa_t gpa,
|
|
|
|
int bytes)
|
|
|
|
{
|
|
|
|
unsigned offset, pte_size, misaligned;
|
|
|
|
|
|
|
|
pgprintk("misaligned: gpa %llx bytes %d role %x\n",
|
|
|
|
gpa, bytes, sp->role.word);
|
|
|
|
|
|
|
|
offset = offset_in_page(gpa);
|
|
|
|
pte_size = sp->role.cr4_pae ? 8 : 4;
|
2011-09-22 08:57:55 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Sometimes, the OS only writes the last one bytes to update status
|
|
|
|
* bits, for example, in linux, andb instruction is used in clear_bit().
|
|
|
|
*/
|
|
|
|
if (!(offset & (pte_size - 1)) && bytes == 1)
|
|
|
|
return false;
|
|
|
|
|
2011-09-22 08:57:23 +00:00
|
|
|
misaligned = (offset ^ (offset + bytes - 1)) & ~(pte_size - 1);
|
|
|
|
misaligned |= bytes < 4;
|
|
|
|
|
|
|
|
return misaligned;
|
|
|
|
}
|
|
|
|
|
|
|
|
static u64 *get_written_sptes(struct kvm_mmu_page *sp, gpa_t gpa, int *nspte)
|
|
|
|
{
|
|
|
|
unsigned page_offset, quadrant;
|
|
|
|
u64 *spte;
|
|
|
|
int level;
|
|
|
|
|
|
|
|
page_offset = offset_in_page(gpa);
|
|
|
|
level = sp->role.level;
|
|
|
|
*nspte = 1;
|
|
|
|
if (!sp->role.cr4_pae) {
|
|
|
|
page_offset <<= 1; /* 32->64 */
|
|
|
|
/*
|
|
|
|
* A 32-bit pde maps 4MB while the shadow pdes map
|
|
|
|
* only 2MB. So we need to double the offset again
|
|
|
|
* and zap two pdes instead of one.
|
|
|
|
*/
|
|
|
|
if (level == PT32_ROOT_LEVEL) {
|
|
|
|
page_offset &= ~7; /* kill rounding error */
|
|
|
|
page_offset <<= 1;
|
|
|
|
*nspte = 2;
|
|
|
|
}
|
|
|
|
quadrant = page_offset >> PAGE_SHIFT;
|
|
|
|
page_offset &= ~PAGE_MASK;
|
|
|
|
if (quadrant != sp->role.quadrant)
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
spte = &sp->spt[page_offset / sizeof(*spte)];
|
|
|
|
return spte;
|
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:16 +00:00
|
|
|
static void kvm_mmu_pte_write(struct kvm_vcpu *vcpu, gpa_t gpa,
|
2016-10-25 07:50:42 +00:00
|
|
|
const u8 *new, int bytes,
|
|
|
|
struct kvm_page_track_notifier_node *node)
|
2011-09-22 08:57:23 +00:00
|
|
|
{
|
|
|
|
gfn_t gfn = gpa >> PAGE_SHIFT;
|
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
LIST_HEAD(invalid_list);
|
|
|
|
u64 entry, gentry, *spte;
|
|
|
|
int npte;
|
2016-02-24 10:21:55 +00:00
|
|
|
bool remote_flush, local_flush;
|
2011-09-22 08:57:23 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If we don't have indirect shadow pages, it means no page is
|
|
|
|
* write-protected, so we can exit simply.
|
|
|
|
*/
|
locking/atomics: COCCINELLE/treewide: Convert trivial ACCESS_ONCE() patterns to READ_ONCE()/WRITE_ONCE()
Please do not apply this to mainline directly, instead please re-run the
coccinelle script shown below and apply its output.
For several reasons, it is desirable to use {READ,WRITE}_ONCE() in
preference to ACCESS_ONCE(), and new code is expected to use one of the
former. So far, there's been no reason to change most existing uses of
ACCESS_ONCE(), as these aren't harmful, and changing them results in
churn.
However, for some features, the read/write distinction is critical to
correct operation. To distinguish these cases, separate read/write
accessors must be used. This patch migrates (most) remaining
ACCESS_ONCE() instances to {READ,WRITE}_ONCE(), using the following
coccinelle script:
----
// Convert trivial ACCESS_ONCE() uses to equivalent READ_ONCE() and
// WRITE_ONCE()
// $ make coccicheck COCCI=/home/mark/once.cocci SPFLAGS="--include-headers" MODE=patch
virtual patch
@ depends on patch @
expression E1, E2;
@@
- ACCESS_ONCE(E1) = E2
+ WRITE_ONCE(E1, E2)
@ depends on patch @
expression E;
@@
- ACCESS_ONCE(E)
+ READ_ONCE(E)
----
Signed-off-by: Mark Rutland <mark.rutland@arm.com>
Signed-off-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: davem@davemloft.net
Cc: linux-arch@vger.kernel.org
Cc: mpe@ellerman.id.au
Cc: shuah@kernel.org
Cc: snitzer@redhat.com
Cc: thor.thayer@linux.intel.com
Cc: tj@kernel.org
Cc: viro@zeniv.linux.org.uk
Cc: will.deacon@arm.com
Link: http://lkml.kernel.org/r/1508792849-3115-19-git-send-email-paulmck@linux.vnet.ibm.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2017-10-23 21:07:29 +00:00
|
|
|
if (!READ_ONCE(vcpu->kvm->arch.indirect_shadow_pages))
|
2011-09-22 08:57:23 +00:00
|
|
|
return;
|
|
|
|
|
2016-02-24 10:21:55 +00:00
|
|
|
remote_flush = local_flush = false;
|
2011-09-22 08:57:23 +00:00
|
|
|
|
|
|
|
pgprintk("%s: gpa %llx bytes %d\n", __func__, gpa, bytes);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* No need to care whether allocation memory is successful
|
|
|
|
* or not since pte prefetch is skiped if it does not have
|
|
|
|
* enough objects in the cache.
|
|
|
|
*/
|
|
|
|
mmu_topup_memory_caches(vcpu);
|
|
|
|
|
|
|
|
spin_lock(&vcpu->kvm->mmu_lock);
|
2018-10-31 21:53:57 +00:00
|
|
|
|
|
|
|
gentry = mmu_pte_write_fetch_gpte(vcpu, &gpa, &bytes);
|
|
|
|
|
2011-09-22 08:57:23 +00:00
|
|
|
++vcpu->kvm->stat.mmu_pte_write;
|
2011-11-28 12:41:00 +00:00
|
|
|
kvm_mmu_audit(vcpu, AUDIT_PRE_PTE_WRITE);
|
2011-09-22 08:57:23 +00:00
|
|
|
|
hlist: drop the node parameter from iterators
I'm not sure why, but the hlist for each entry iterators were conceived
list_for_each_entry(pos, head, member)
The hlist ones were greedy and wanted an extra parameter:
hlist_for_each_entry(tpos, pos, head, member)
Why did they need an extra pos parameter? I'm not quite sure. Not only
they don't really need it, it also prevents the iterator from looking
exactly like the list iterator, which is unfortunate.
Besides the semantic patch, there was some manual work required:
- Fix up the actual hlist iterators in linux/list.h
- Fix up the declaration of other iterators based on the hlist ones.
- A very small amount of places were using the 'node' parameter, this
was modified to use 'obj->member' instead.
- Coccinelle didn't handle the hlist_for_each_entry_safe iterator
properly, so those had to be fixed up manually.
The semantic patch which is mostly the work of Peter Senna Tschudin is here:
@@
iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host;
type T;
expression a,c,d,e;
identifier b;
statement S;
@@
-T b;
<+... when != b
(
hlist_for_each_entry(a,
- b,
c, d) S
|
hlist_for_each_entry_continue(a,
- b,
c) S
|
hlist_for_each_entry_from(a,
- b,
c) S
|
hlist_for_each_entry_rcu(a,
- b,
c, d) S
|
hlist_for_each_entry_rcu_bh(a,
- b,
c, d) S
|
hlist_for_each_entry_continue_rcu_bh(a,
- b,
c) S
|
for_each_busy_worker(a, c,
- b,
d) S
|
ax25_uid_for_each(a,
- b,
c) S
|
ax25_for_each(a,
- b,
c) S
|
inet_bind_bucket_for_each(a,
- b,
c) S
|
sctp_for_each_hentry(a,
- b,
c) S
|
sk_for_each(a,
- b,
c) S
|
sk_for_each_rcu(a,
- b,
c) S
|
sk_for_each_from
-(a, b)
+(a)
S
+ sk_for_each_from(a) S
|
sk_for_each_safe(a,
- b,
c, d) S
|
sk_for_each_bound(a,
- b,
c) S
|
hlist_for_each_entry_safe(a,
- b,
c, d, e) S
|
hlist_for_each_entry_continue_rcu(a,
- b,
c) S
|
nr_neigh_for_each(a,
- b,
c) S
|
nr_neigh_for_each_safe(a,
- b,
c, d) S
|
nr_node_for_each(a,
- b,
c) S
|
nr_node_for_each_safe(a,
- b,
c, d) S
|
- for_each_gfn_sp(a, c, d, b) S
+ for_each_gfn_sp(a, c, d) S
|
- for_each_gfn_indirect_valid_sp(a, c, d, b) S
+ for_each_gfn_indirect_valid_sp(a, c, d) S
|
for_each_host(a,
- b,
c) S
|
for_each_host_safe(a,
- b,
c, d) S
|
for_each_mesh_entry(a,
- b,
c, d) S
)
...+>
[akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c]
[akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c]
[akpm@linux-foundation.org: checkpatch fixes]
[akpm@linux-foundation.org: fix warnings]
[akpm@linux-foudnation.org: redo intrusive kvm changes]
Tested-by: Peter Senna Tschudin <peter.senna@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2013-02-28 01:06:00 +00:00
|
|
|
for_each_gfn_indirect_valid_sp(vcpu->kvm, sp, gfn) {
|
KVM: MMU: improve write flooding detected
Detecting write-flooding does not work well, when we handle page written, if
the last speculative spte is not accessed, we treat the page is
write-flooding, however, we can speculative spte on many path, such as pte
prefetch, page synced, that means the last speculative spte may be not point
to the written page and the written page can be accessed via other sptes, so
depends on the Accessed bit of the last speculative spte is not enough
Instead of detected page accessed, we can detect whether the spte is accessed
after it is written, if the spte is not accessed but it is written frequently,
we treat is not a page table or it not used for a long time
Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2011-09-22 08:58:36 +00:00
|
|
|
if (detect_write_misaligned(sp, gpa, bytes) ||
|
2011-12-16 10:18:10 +00:00
|
|
|
detect_write_flooding(sp)) {
|
2016-02-24 10:21:55 +00:00
|
|
|
kvm_mmu_prepare_zap_page(vcpu->kvm, sp, &invalid_list);
|
2007-11-18 14:37:07 +00:00
|
|
|
++vcpu->kvm->stat.mmu_flooded;
|
2007-01-06 00:36:48 +00:00
|
|
|
continue;
|
|
|
|
}
|
2011-09-22 08:57:23 +00:00
|
|
|
|
|
|
|
spte = get_written_sptes(sp, gpa, &npte);
|
|
|
|
if (!spte)
|
|
|
|
continue;
|
|
|
|
|
2010-06-04 13:56:59 +00:00
|
|
|
local_flush = true;
|
2007-03-08 15:13:32 +00:00
|
|
|
while (npte--) {
|
2018-10-08 19:28:10 +00:00
|
|
|
u32 base_role = vcpu->arch.mmu->mmu_role.base.word;
|
|
|
|
|
2007-11-21 00:06:21 +00:00
|
|
|
entry = *spte;
|
2011-05-15 15:27:52 +00:00
|
|
|
mmu_page_zap_pte(vcpu->kvm, sp, spte);
|
2010-07-16 03:19:51 +00:00
|
|
|
if (gentry &&
|
2018-10-08 19:28:10 +00:00
|
|
|
!((sp->role.word ^ base_role)
|
2018-06-27 21:59:07 +00:00
|
|
|
& mmu_base_role_mask.word) && rmap_can_add(vcpu))
|
2011-03-28 02:29:27 +00:00
|
|
|
mmu_pte_write_new_pte(vcpu, sp, spte, &gentry);
|
2013-01-30 14:45:01 +00:00
|
|
|
if (need_remote_flush(entry, *spte))
|
2010-06-04 13:56:59 +00:00
|
|
|
remote_flush = true;
|
2007-03-08 15:13:32 +00:00
|
|
|
++spte;
|
2007-01-06 00:36:45 +00:00
|
|
|
}
|
|
|
|
}
|
2016-02-24 10:21:55 +00:00
|
|
|
kvm_mmu_flush_or_zap(vcpu, &invalid_list, remote_flush, local_flush);
|
2011-11-28 12:41:00 +00:00
|
|
|
kvm_mmu_audit(vcpu, AUDIT_POST_PTE_WRITE);
|
2007-12-21 00:18:26 +00:00
|
|
|
spin_unlock(&vcpu->kvm->mmu_lock);
|
2007-01-06 00:36:44 +00:00
|
|
|
}
|
|
|
|
|
2007-01-06 00:36:45 +00:00
|
|
|
int kvm_mmu_unprotect_page_virt(struct kvm_vcpu *vcpu, gva_t gva)
|
|
|
|
{
|
2007-12-21 00:18:22 +00:00
|
|
|
gpa_t gpa;
|
|
|
|
int r;
|
2007-01-06 00:36:45 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->direct_map)
|
2009-08-27 10:37:06 +00:00
|
|
|
return 0;
|
|
|
|
|
2010-02-10 12:21:32 +00:00
|
|
|
gpa = kvm_mmu_gva_to_gpa_read(vcpu, gva, NULL);
|
2007-12-21 00:18:22 +00:00
|
|
|
|
|
|
|
r = kvm_mmu_unprotect_page(vcpu->kvm, gpa >> PAGE_SHIFT);
|
2011-09-22 09:02:48 +00:00
|
|
|
|
2007-12-21 00:18:22 +00:00
|
|
|
return r;
|
2007-01-06 00:36:45 +00:00
|
|
|
}
|
2008-07-19 05:57:05 +00:00
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_unprotect_page_virt);
|
2007-01-06 00:36:45 +00:00
|
|
|
|
2017-08-10 23:28:02 +00:00
|
|
|
static int make_mmu_pages_available(struct kvm_vcpu *vcpu)
|
2007-01-06 00:36:47 +00:00
|
|
|
{
|
2010-06-04 13:55:29 +00:00
|
|
|
LIST_HEAD(invalid_list);
|
2010-06-04 13:54:38 +00:00
|
|
|
|
2013-03-21 10:34:27 +00:00
|
|
|
if (likely(kvm_mmu_available_pages(vcpu->kvm) >= KVM_MIN_FREE_MMU_PAGES))
|
2017-08-10 23:28:02 +00:00
|
|
|
return 0;
|
2013-03-21 10:34:27 +00:00
|
|
|
|
2013-03-06 07:06:58 +00:00
|
|
|
while (kvm_mmu_available_pages(vcpu->kvm) < KVM_REFILL_PAGES) {
|
|
|
|
if (!prepare_zap_oldest_mmu_page(vcpu->kvm, &invalid_list))
|
|
|
|
break;
|
2007-01-06 00:36:47 +00:00
|
|
|
|
2007-11-18 14:37:07 +00:00
|
|
|
++vcpu->kvm->stat.mmu_recycled;
|
2007-01-06 00:36:47 +00:00
|
|
|
}
|
2011-07-11 19:26:40 +00:00
|
|
|
kvm_mmu_commit_zap_page(vcpu->kvm, &invalid_list);
|
2017-08-10 23:28:02 +00:00
|
|
|
|
|
|
|
if (!kvm_mmu_available_pages(vcpu->kvm))
|
|
|
|
return -ENOSPC;
|
|
|
|
return 0;
|
2007-01-06 00:36:47 +00:00
|
|
|
}
|
|
|
|
|
2016-11-23 17:01:38 +00:00
|
|
|
int kvm_mmu_page_fault(struct kvm_vcpu *vcpu, gva_t cr2, u64 error_code,
|
2010-12-21 10:12:07 +00:00
|
|
|
void *insn, int insn_len)
|
2007-10-28 16:48:59 +00:00
|
|
|
{
|
2018-08-23 20:56:50 +00:00
|
|
|
int r, emulation_type = 0;
|
2007-10-28 16:48:59 +00:00
|
|
|
enum emulation_result er;
|
2018-10-08 19:28:05 +00:00
|
|
|
bool direct = vcpu->arch.mmu->direct_map;
|
2007-10-28 16:48:59 +00:00
|
|
|
|
2017-08-17 16:36:57 +00:00
|
|
|
/* With shadow page tables, fault_address contains a GVA or nGPA. */
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->direct_map) {
|
2017-08-17 16:36:57 +00:00
|
|
|
vcpu->arch.gpa_available = true;
|
|
|
|
vcpu->arch.gpa_val = cr2;
|
|
|
|
}
|
2007-10-28 16:48:59 +00:00
|
|
|
|
2017-08-17 13:03:32 +00:00
|
|
|
r = RET_PF_INVALID;
|
2016-02-22 08:23:41 +00:00
|
|
|
if (unlikely(error_code & PFERR_RSVD_MASK)) {
|
|
|
|
r = handle_mmio_page_fault(vcpu, cr2, direct);
|
2018-08-23 20:56:50 +00:00
|
|
|
if (r == RET_PF_EMULATE)
|
2016-02-22 08:23:41 +00:00
|
|
|
goto emulate;
|
|
|
|
}
|
2007-10-28 16:48:59 +00:00
|
|
|
|
2017-08-17 13:03:32 +00:00
|
|
|
if (r == RET_PF_INVALID) {
|
2018-10-08 19:28:05 +00:00
|
|
|
r = vcpu->arch.mmu->page_fault(vcpu, cr2,
|
|
|
|
lower_32_bits(error_code),
|
|
|
|
false);
|
2017-08-17 13:03:32 +00:00
|
|
|
WARN_ON(r == RET_PF_INVALID);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (r == RET_PF_RETRY)
|
|
|
|
return 1;
|
2007-10-28 16:48:59 +00:00
|
|
|
if (r < 0)
|
2016-02-22 08:23:41 +00:00
|
|
|
return r;
|
2007-10-28 16:48:59 +00:00
|
|
|
|
2016-11-23 17:01:38 +00:00
|
|
|
/*
|
|
|
|
* Before emulating the instruction, check if the error code
|
|
|
|
* was due to a RO violation while translating the guest page.
|
|
|
|
* This can occur when using nested virtualization with nested
|
|
|
|
* paging in both guests. If true, we simply unprotect the page
|
|
|
|
* and resume the guest.
|
|
|
|
*/
|
2018-10-08 19:28:05 +00:00
|
|
|
if (vcpu->arch.mmu->direct_map &&
|
2016-11-28 13:39:58 +00:00
|
|
|
(error_code & PFERR_NESTED_GUEST_PAGE) == PFERR_NESTED_GUEST_PAGE) {
|
2016-11-23 17:01:38 +00:00
|
|
|
kvm_mmu_unprotect_page(vcpu->kvm, gpa_to_gfn(cr2));
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2018-08-23 20:56:50 +00:00
|
|
|
/*
|
|
|
|
* vcpu->arch.mmu.page_fault returned RET_PF_EMULATE, but we can still
|
|
|
|
* optimistically try to just unprotect the page and let the processor
|
|
|
|
* re-execute the instruction that caused the page fault. Do not allow
|
|
|
|
* retrying MMIO emulation, as it's not only pointless but could also
|
|
|
|
* cause us to enter an infinite loop because the processor will keep
|
KVM: x86: Do not re-{try,execute} after failed emulation in L2
Commit a6f177efaa58 ("KVM: Reenter guest after emulation failure if
due to access to non-mmio address") added reexecute_instruction() to
handle the scenario where two (or more) vCPUS race to write a shadowed
page, i.e. reexecute_instruction() is intended to return true if and
only if the instruction being emulated was accessing a shadowed page.
As L0 is only explicitly shadowing L1 tables, an emulation failure of
a nested VM instruction cannot be due to a race to write a shadowed
page and so should never be re-executed.
This fixes an issue where an "MMIO" emulation failure[1] in L2 is all
but guaranteed to result in an infinite loop when TDP is enabled.
Because "cr2" is actually an L2 GPA when TDP is enabled, calling
kvm_mmu_gva_to_gpa_write() to translate cr2 in the non-direct mapped
case (L2 is never direct mapped) will almost always yield UNMAPPED_GVA
and cause reexecute_instruction() to immediately return true. The
!mmio_info_in_cache() check in kvm_mmu_page_fault() doesn't catch this
case because mmio_info_in_cache() returns false for a nested MMU (the
MMIO caching currently handles L1 only, e.g. to cache nested guests'
GPAs we'd have to manually flush the cache when switching between
VMs and when L1 updated its page tables controlling the nested guest).
Way back when, commit 68be0803456b ("KVM: x86: never re-execute
instruction with enabled tdp") changed reexecute_instruction() to
always return false when using TDP under the assumption that KVM would
only get into the emulator for MMIO. Commit 95b3cf69bdf8 ("KVM: x86:
let reexecute_instruction work for tdp") effectively reverted that
behavior in order to handle the scenario where emulation failed due to
an access from L1 to the shadow page tables for L2, but it didn't
account for the case where emulation failed in L2 with TDP enabled.
All of the above logic also applies to retry_instruction(), added by
commit 1cb3f3ae5a38 ("KVM: x86: retry non-page-table writing
instructions"). An indefinite loop in retry_instruction() should be
impossible as it protects against retrying the same instruction over
and over, but it's still correct to not retry an L2 instruction in
the first place.
Fix the immediate issue by adding a check for a nested guest when
determining whether or not to allow retry in kvm_mmu_page_fault().
In addition to fixing the immediate bug, add WARN_ON_ONCE in the
retry functions since they are not designed to handle nested cases,
i.e. they need to be modified even if there is some scenario in the
future where we want to allow retrying a nested guest.
[1] This issue was encountered after commit 3a2936dedd20 ("kvm: mmu:
Don't expose private memslots to L2") changed the page fault path
to return KVM_PFN_NOSLOT when translating an L2 access to a
prive memslot. Returning KVM_PFN_NOSLOT is semantically correct
when we want to hide a memslot from L2, i.e. there effectively is
no defined memory region for L2, but it has the unfortunate side
effect of making KVM think the GFN is a MMIO page, thus triggering
emulation. The failure occurred with in-development code that
deliberately exposed a private memslot to L2, which L2 accessed
with an instruction that is not emulated by KVM.
Fixes: 95b3cf69bdf8 ("KVM: x86: let reexecute_instruction work for tdp")
Fixes: 1cb3f3ae5a38 ("KVM: x86: retry non-page-table writing instructions")
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Cc: Jim Mattson <jmattson@google.com>
Cc: Krish Sadhukhan <krish.sadhukhan@oracle.com>
Cc: Xiao Guangrong <xiaoguangrong@tencent.com>
Cc: stable@vger.kernel.org
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
2018-08-23 20:56:51 +00:00
|
|
|
* faulting on the non-existent MMIO address. Retrying an instruction
|
|
|
|
* from a nested guest is also pointless and dangerous as we are only
|
|
|
|
* explicitly shadowing L1's page tables, i.e. unprotecting something
|
|
|
|
* for L1 isn't going to magically fix whatever issue cause L2 to fail.
|
2018-08-23 20:56:50 +00:00
|
|
|
*/
|
KVM: x86: Do not re-{try,execute} after failed emulation in L2
Commit a6f177efaa58 ("KVM: Reenter guest after emulation failure if
due to access to non-mmio address") added reexecute_instruction() to
handle the scenario where two (or more) vCPUS race to write a shadowed
page, i.e. reexecute_instruction() is intended to return true if and
only if the instruction being emulated was accessing a shadowed page.
As L0 is only explicitly shadowing L1 tables, an emulation failure of
a nested VM instruction cannot be due to a race to write a shadowed
page and so should never be re-executed.
This fixes an issue where an "MMIO" emulation failure[1] in L2 is all
but guaranteed to result in an infinite loop when TDP is enabled.
Because "cr2" is actually an L2 GPA when TDP is enabled, calling
kvm_mmu_gva_to_gpa_write() to translate cr2 in the non-direct mapped
case (L2 is never direct mapped) will almost always yield UNMAPPED_GVA
and cause reexecute_instruction() to immediately return true. The
!mmio_info_in_cache() check in kvm_mmu_page_fault() doesn't catch this
case because mmio_info_in_cache() returns false for a nested MMU (the
MMIO caching currently handles L1 only, e.g. to cache nested guests'
GPAs we'd have to manually flush the cache when switching between
VMs and when L1 updated its page tables controlling the nested guest).
Way back when, commit 68be0803456b ("KVM: x86: never re-execute
instruction with enabled tdp") changed reexecute_instruction() to
always return false when using TDP under the assumption that KVM would
only get into the emulator for MMIO. Commit 95b3cf69bdf8 ("KVM: x86:
let reexecute_instruction work for tdp") effectively reverted that
behavior in order to handle the scenario where emulation failed due to
an access from L1 to the shadow page tables for L2, but it didn't
account for the case where emulation failed in L2 with TDP enabled.
All of the above logic also applies to retry_instruction(), added by
commit 1cb3f3ae5a38 ("KVM: x86: retry non-page-table writing
instructions"). An indefinite loop in retry_instruction() should be
impossible as it protects against retrying the same instruction over
and over, but it's still correct to not retry an L2 instruction in
the first place.
Fix the immediate issue by adding a check for a nested guest when
determining whether or not to allow retry in kvm_mmu_page_fault().
In addition to fixing the immediate bug, add WARN_ON_ONCE in the
retry functions since they are not designed to handle nested cases,
i.e. they need to be modified even if there is some scenario in the
future where we want to allow retrying a nested guest.
[1] This issue was encountered after commit 3a2936dedd20 ("kvm: mmu:
Don't expose private memslots to L2") changed the page fault path
to return KVM_PFN_NOSLOT when translating an L2 access to a
prive memslot. Returning KVM_PFN_NOSLOT is semantically correct
when we want to hide a memslot from L2, i.e. there effectively is
no defined memory region for L2, but it has the unfortunate side
effect of making KVM think the GFN is a MMIO page, thus triggering
emulation. The failure occurred with in-development code that
deliberately exposed a private memslot to L2, which L2 accessed
with an instruction that is not emulated by KVM.
Fixes: 95b3cf69bdf8 ("KVM: x86: let reexecute_instruction work for tdp")
Fixes: 1cb3f3ae5a38 ("KVM: x86: retry non-page-table writing instructions")
Signed-off-by: Sean Christopherson <sean.j.christopherson@intel.com>
Cc: Jim Mattson <jmattson@google.com>
Cc: Krish Sadhukhan <krish.sadhukhan@oracle.com>
Cc: Xiao Guangrong <xiaoguangrong@tencent.com>
Cc: stable@vger.kernel.org
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
2018-08-23 20:56:51 +00:00
|
|
|
if (!mmio_info_in_cache(vcpu, cr2, direct) && !is_guest_mode(vcpu))
|
2018-08-23 20:56:50 +00:00
|
|
|
emulation_type = EMULTYPE_ALLOW_RETRY;
|
2016-02-22 08:23:41 +00:00
|
|
|
emulate:
|
2017-12-04 16:57:40 +00:00
|
|
|
/*
|
|
|
|
* On AMD platforms, under certain conditions insn_len may be zero on #NPF.
|
|
|
|
* This can happen if a guest gets a page-fault on data access but the HW
|
|
|
|
* table walker is not able to read the instruction page (e.g instruction
|
|
|
|
* page is not present in memory). In those cases we simply restart the
|
|
|
|
* guest.
|
|
|
|
*/
|
|
|
|
if (unlikely(insn && !insn_len))
|
|
|
|
return 1;
|
|
|
|
|
2011-09-22 09:02:48 +00:00
|
|
|
er = x86_emulate_instruction(vcpu, cr2, emulation_type, insn, insn_len);
|
2007-10-28 16:48:59 +00:00
|
|
|
|
|
|
|
switch (er) {
|
|
|
|
case EMULATE_DONE:
|
|
|
|
return 1;
|
2013-06-25 16:24:41 +00:00
|
|
|
case EMULATE_USER_EXIT:
|
2007-10-28 16:48:59 +00:00
|
|
|
++vcpu->stat.mmio_exits;
|
2010-05-10 08:16:56 +00:00
|
|
|
/* fall through */
|
2007-10-28 16:48:59 +00:00
|
|
|
case EMULATE_FAIL:
|
2009-06-11 12:43:28 +00:00
|
|
|
return 0;
|
2007-10-28 16:48:59 +00:00
|
|
|
default:
|
|
|
|
BUG();
|
|
|
|
}
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_page_fault);
|
|
|
|
|
2008-09-23 16:18:35 +00:00
|
|
|
void kvm_mmu_invlpg(struct kvm_vcpu *vcpu, gva_t gva)
|
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
struct kvm_mmu *mmu = vcpu->arch.mmu;
|
2018-06-27 21:59:20 +00:00
|
|
|
int i;
|
2018-06-27 21:59:16 +00:00
|
|
|
|
2018-06-29 20:10:05 +00:00
|
|
|
/* INVLPG on a * non-canonical address is a NOP according to the SDM. */
|
|
|
|
if (is_noncanonical_address(gva, vcpu))
|
|
|
|
return;
|
|
|
|
|
2018-06-27 21:59:16 +00:00
|
|
|
mmu->invlpg(vcpu, gva, mmu->root_hpa);
|
2018-06-27 21:59:18 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* INVLPG is required to invalidate any global mappings for the VA,
|
|
|
|
* irrespective of PCID. Since it would take us roughly similar amount
|
2018-06-27 21:59:20 +00:00
|
|
|
* of work to determine whether any of the prev_root mappings of the VA
|
|
|
|
* is marked global, or to just sync it blindly, so we might as well
|
|
|
|
* just always sync it.
|
2018-06-27 21:59:18 +00:00
|
|
|
*
|
2018-06-27 21:59:20 +00:00
|
|
|
* Mappings not reachable via the current cr3 or the prev_roots will be
|
|
|
|
* synced when switching to that cr3, so nothing needs to be done here
|
|
|
|
* for them.
|
2018-06-27 21:59:18 +00:00
|
|
|
*/
|
2018-06-27 21:59:20 +00:00
|
|
|
for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++)
|
|
|
|
if (VALID_PAGE(mmu->prev_roots[i].hpa))
|
|
|
|
mmu->invlpg(vcpu, gva, mmu->prev_roots[i].hpa);
|
2018-06-27 21:59:18 +00:00
|
|
|
|
2018-06-29 20:10:05 +00:00
|
|
|
kvm_x86_ops->tlb_flush_gva(vcpu, gva);
|
2008-09-23 16:18:35 +00:00
|
|
|
++vcpu->stat.invlpg;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_invlpg);
|
|
|
|
|
2018-06-27 21:59:14 +00:00
|
|
|
void kvm_mmu_invpcid_gva(struct kvm_vcpu *vcpu, gva_t gva, unsigned long pcid)
|
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
struct kvm_mmu *mmu = vcpu->arch.mmu;
|
2018-06-29 20:10:05 +00:00
|
|
|
bool tlb_flush = false;
|
2018-06-27 21:59:20 +00:00
|
|
|
uint i;
|
2018-06-27 21:59:14 +00:00
|
|
|
|
|
|
|
if (pcid == kvm_get_active_pcid(vcpu)) {
|
2018-06-27 21:59:16 +00:00
|
|
|
mmu->invlpg(vcpu, gva, mmu->root_hpa);
|
2018-06-29 20:10:05 +00:00
|
|
|
tlb_flush = true;
|
2018-06-27 21:59:14 +00:00
|
|
|
}
|
|
|
|
|
2018-06-27 21:59:20 +00:00
|
|
|
for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++) {
|
|
|
|
if (VALID_PAGE(mmu->prev_roots[i].hpa) &&
|
|
|
|
pcid == kvm_get_pcid(vcpu, mmu->prev_roots[i].cr3)) {
|
|
|
|
mmu->invlpg(vcpu, gva, mmu->prev_roots[i].hpa);
|
|
|
|
tlb_flush = true;
|
|
|
|
}
|
2018-06-27 21:59:18 +00:00
|
|
|
}
|
2018-06-27 21:59:15 +00:00
|
|
|
|
2018-06-29 20:10:05 +00:00
|
|
|
if (tlb_flush)
|
|
|
|
kvm_x86_ops->tlb_flush_gva(vcpu, gva);
|
|
|
|
|
2018-06-27 21:59:14 +00:00
|
|
|
++vcpu->stat.invlpg;
|
|
|
|
|
|
|
|
/*
|
2018-06-27 21:59:20 +00:00
|
|
|
* Mappings not reachable via the current cr3 or the prev_roots will be
|
|
|
|
* synced when switching to that cr3, so nothing needs to be done here
|
|
|
|
* for them.
|
2018-06-27 21:59:14 +00:00
|
|
|
*/
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_invpcid_gva);
|
|
|
|
|
2008-02-07 12:47:41 +00:00
|
|
|
void kvm_enable_tdp(void)
|
|
|
|
{
|
|
|
|
tdp_enabled = true;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_enable_tdp);
|
|
|
|
|
2008-07-14 18:36:36 +00:00
|
|
|
void kvm_disable_tdp(void)
|
|
|
|
{
|
|
|
|
tdp_enabled = false;
|
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_disable_tdp);
|
|
|
|
|
2019-02-05 21:01:19 +00:00
|
|
|
|
|
|
|
/* The return value indicates if tlb flush on all vcpus is needed. */
|
|
|
|
typedef bool (*slot_level_handler) (struct kvm *kvm, struct kvm_rmap_head *rmap_head);
|
|
|
|
|
|
|
|
/* The caller should hold mmu-lock before calling this function. */
|
|
|
|
static __always_inline bool
|
|
|
|
slot_handle_level_range(struct kvm *kvm, struct kvm_memory_slot *memslot,
|
|
|
|
slot_level_handler fn, int start_level, int end_level,
|
|
|
|
gfn_t start_gfn, gfn_t end_gfn, bool lock_flush_tlb)
|
|
|
|
{
|
|
|
|
struct slot_rmap_walk_iterator iterator;
|
|
|
|
bool flush = false;
|
|
|
|
|
|
|
|
for_each_slot_rmap_range(memslot, start_level, end_level, start_gfn,
|
|
|
|
end_gfn, &iterator) {
|
|
|
|
if (iterator.rmap)
|
|
|
|
flush |= fn(kvm, iterator.rmap);
|
|
|
|
|
|
|
|
if (need_resched() || spin_needbreak(&kvm->mmu_lock)) {
|
|
|
|
if (flush && lock_flush_tlb) {
|
|
|
|
kvm_flush_remote_tlbs(kvm);
|
|
|
|
flush = false;
|
|
|
|
}
|
|
|
|
cond_resched_lock(&kvm->mmu_lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (flush && lock_flush_tlb) {
|
|
|
|
kvm_flush_remote_tlbs(kvm);
|
|
|
|
flush = false;
|
|
|
|
}
|
|
|
|
|
|
|
|
return flush;
|
|
|
|
}
|
|
|
|
|
|
|
|
static __always_inline bool
|
|
|
|
slot_handle_level(struct kvm *kvm, struct kvm_memory_slot *memslot,
|
|
|
|
slot_level_handler fn, int start_level, int end_level,
|
|
|
|
bool lock_flush_tlb)
|
|
|
|
{
|
|
|
|
return slot_handle_level_range(kvm, memslot, fn, start_level,
|
|
|
|
end_level, memslot->base_gfn,
|
|
|
|
memslot->base_gfn + memslot->npages - 1,
|
|
|
|
lock_flush_tlb);
|
|
|
|
}
|
|
|
|
|
|
|
|
static __always_inline bool
|
|
|
|
slot_handle_all_level(struct kvm *kvm, struct kvm_memory_slot *memslot,
|
|
|
|
slot_level_handler fn, bool lock_flush_tlb)
|
|
|
|
{
|
|
|
|
return slot_handle_level(kvm, memslot, fn, PT_PAGE_TABLE_LEVEL,
|
|
|
|
PT_MAX_HUGEPAGE_LEVEL, lock_flush_tlb);
|
|
|
|
}
|
|
|
|
|
|
|
|
static __always_inline bool
|
|
|
|
slot_handle_large_level(struct kvm *kvm, struct kvm_memory_slot *memslot,
|
|
|
|
slot_level_handler fn, bool lock_flush_tlb)
|
|
|
|
{
|
|
|
|
return slot_handle_level(kvm, memslot, fn, PT_PAGE_TABLE_LEVEL + 1,
|
|
|
|
PT_MAX_HUGEPAGE_LEVEL, lock_flush_tlb);
|
|
|
|
}
|
|
|
|
|
|
|
|
static __always_inline bool
|
|
|
|
slot_handle_leaf(struct kvm *kvm, struct kvm_memory_slot *memslot,
|
|
|
|
slot_level_handler fn, bool lock_flush_tlb)
|
|
|
|
{
|
|
|
|
return slot_handle_level(kvm, memslot, fn, PT_PAGE_TABLE_LEVEL,
|
|
|
|
PT_PAGE_TABLE_LEVEL, lock_flush_tlb);
|
|
|
|
}
|
|
|
|
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
static void free_mmu_pages(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2018-10-08 19:28:05 +00:00
|
|
|
free_page((unsigned long)vcpu->arch.mmu->pae_root);
|
|
|
|
free_page((unsigned long)vcpu->arch.mmu->lm_root);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int alloc_mmu_pages(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2007-01-06 00:36:40 +00:00
|
|
|
struct page *page;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
int i;
|
|
|
|
|
2018-07-25 08:32:14 +00:00
|
|
|
if (tdp_enabled)
|
|
|
|
return 0;
|
|
|
|
|
2007-01-06 00:36:40 +00:00
|
|
|
/*
|
|
|
|
* When emulating 32-bit mode, cr3 is only 32 bits even on x86_64.
|
|
|
|
* Therefore we need to allocate shadow page tables in the first
|
|
|
|
* 4GB of memory, which happens to fit the DMA32 zone.
|
|
|
|
*/
|
2019-02-11 19:02:50 +00:00
|
|
|
page = alloc_page(GFP_KERNEL_ACCOUNT | __GFP_DMA32);
|
2007-01-06 00:36:40 +00:00
|
|
|
if (!page)
|
2010-01-22 08:55:05 +00:00
|
|
|
return -ENOMEM;
|
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->pae_root = page_address(page);
|
2007-01-06 00:36:40 +00:00
|
|
|
for (i = 0; i < 4; ++i)
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu->pae_root[i] = INVALID_PAGE;
|
2007-01-06 00:36:40 +00:00
|
|
|
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2006-12-30 00:50:01 +00:00
|
|
|
int kvm_mmu_create(struct kvm_vcpu *vcpu)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2018-06-27 21:59:20 +00:00
|
|
|
uint i;
|
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.mmu = &vcpu->arch.root_mmu;
|
|
|
|
vcpu->arch.walk_mmu = &vcpu->arch.root_mmu;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.root_mmu.root_hpa = INVALID_PAGE;
|
|
|
|
vcpu->arch.root_mmu.translate_gpa = translate_gpa;
|
2018-06-27 21:59:20 +00:00
|
|
|
for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++)
|
2018-10-08 19:28:05 +00:00
|
|
|
vcpu->arch.root_mmu.prev_roots[i] = KVM_MMU_ROOT_INFO_INVALID;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2018-10-08 19:28:08 +00:00
|
|
|
vcpu->arch.guest_mmu.root_hpa = INVALID_PAGE;
|
|
|
|
vcpu->arch.guest_mmu.translate_gpa = translate_gpa;
|
|
|
|
for (i = 0; i < KVM_MMU_NUM_PREV_ROOTS; i++)
|
|
|
|
vcpu->arch.guest_mmu.prev_roots[i] = KVM_MMU_ROOT_INFO_INVALID;
|
2006-12-22 09:05:28 +00:00
|
|
|
|
2018-10-08 19:28:08 +00:00
|
|
|
vcpu->arch.nested_mmu.translate_gpa = translate_nested_gpa;
|
2006-12-30 00:50:01 +00:00
|
|
|
return alloc_mmu_pages(vcpu);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
|
|
|
|
2016-10-09 07:41:44 +00:00
|
|
|
static void kvm_mmu_invalidate_zap_pages_in_memslot(struct kvm *kvm,
|
2016-10-25 07:50:42 +00:00
|
|
|
struct kvm_memory_slot *slot,
|
|
|
|
struct kvm_page_track_notifier_node *node)
|
2016-10-09 07:41:44 +00:00
|
|
|
{
|
2019-02-05 21:01:21 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
LIST_HEAD(invalid_list);
|
|
|
|
unsigned long i;
|
|
|
|
bool flush;
|
|
|
|
gfn_t gfn;
|
|
|
|
|
|
|
|
spin_lock(&kvm->mmu_lock);
|
|
|
|
|
|
|
|
if (list_empty(&kvm->arch.active_mmu_pages))
|
|
|
|
goto out_unlock;
|
|
|
|
|
|
|
|
flush = slot_handle_all_level(kvm, slot, kvm_zap_rmapp, false);
|
|
|
|
|
|
|
|
for (i = 0; i < slot->npages; i++) {
|
|
|
|
gfn = slot->base_gfn + i;
|
|
|
|
|
|
|
|
for_each_valid_sp(kvm, sp, gfn) {
|
|
|
|
if (sp->gfn != gfn)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
kvm_mmu_prepare_zap_page(kvm, sp, &invalid_list);
|
|
|
|
}
|
|
|
|
if (need_resched() || spin_needbreak(&kvm->mmu_lock)) {
|
|
|
|
kvm_mmu_remote_flush_or_zap(kvm, &invalid_list, flush);
|
|
|
|
flush = false;
|
|
|
|
cond_resched_lock(&kvm->mmu_lock);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
kvm_mmu_remote_flush_or_zap(kvm, &invalid_list, flush);
|
|
|
|
|
|
|
|
out_unlock:
|
|
|
|
spin_unlock(&kvm->mmu_lock);
|
2016-10-09 07:41:44 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 09:51:16 +00:00
|
|
|
void kvm_mmu_init_vm(struct kvm *kvm)
|
|
|
|
{
|
|
|
|
struct kvm_page_track_notifier_node *node = &kvm->arch.mmu_sp_tracker;
|
|
|
|
|
|
|
|
node->track_write = kvm_mmu_pte_write;
|
2016-10-09 07:41:44 +00:00
|
|
|
node->track_flush_slot = kvm_mmu_invalidate_zap_pages_in_memslot;
|
2016-02-24 09:51:16 +00:00
|
|
|
kvm_page_track_register_notifier(kvm, node);
|
|
|
|
}
|
|
|
|
|
|
|
|
void kvm_mmu_uninit_vm(struct kvm *kvm)
|
|
|
|
{
|
|
|
|
struct kvm_page_track_notifier_node *node = &kvm->arch.mmu_sp_tracker;
|
|
|
|
|
|
|
|
kvm_page_track_unregister_notifier(kvm, node);
|
|
|
|
}
|
|
|
|
|
2015-05-13 06:42:27 +00:00
|
|
|
void kvm_zap_gfn_range(struct kvm *kvm, gfn_t gfn_start, gfn_t gfn_end)
|
|
|
|
{
|
|
|
|
struct kvm_memslots *slots;
|
|
|
|
struct kvm_memory_slot *memslot;
|
2018-12-06 13:21:13 +00:00
|
|
|
bool flush_tlb = true;
|
|
|
|
bool flush = false;
|
2015-05-18 11:33:16 +00:00
|
|
|
int i;
|
2015-05-13 06:42:27 +00:00
|
|
|
|
2018-12-06 13:21:13 +00:00
|
|
|
if (kvm_available_flush_tlb_with_range())
|
|
|
|
flush_tlb = false;
|
|
|
|
|
2015-05-13 06:42:27 +00:00
|
|
|
spin_lock(&kvm->mmu_lock);
|
2015-05-18 11:33:16 +00:00
|
|
|
for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++) {
|
|
|
|
slots = __kvm_memslots(kvm, i);
|
|
|
|
kvm_for_each_memslot(memslot, slots) {
|
|
|
|
gfn_t start, end;
|
|
|
|
|
|
|
|
start = max(gfn_start, memslot->base_gfn);
|
|
|
|
end = min(gfn_end, memslot->base_gfn + memslot->npages);
|
|
|
|
if (start >= end)
|
|
|
|
continue;
|
2015-05-13 06:42:27 +00:00
|
|
|
|
2018-12-06 13:21:13 +00:00
|
|
|
flush |= slot_handle_level_range(kvm, memslot,
|
|
|
|
kvm_zap_rmapp, PT_PAGE_TABLE_LEVEL,
|
|
|
|
PT_MAX_HUGEPAGE_LEVEL, start,
|
|
|
|
end - 1, flush_tlb);
|
2015-05-18 11:33:16 +00:00
|
|
|
}
|
2015-05-13 06:42:27 +00:00
|
|
|
}
|
|
|
|
|
2018-12-06 13:21:13 +00:00
|
|
|
if (flush)
|
|
|
|
kvm_flush_remote_tlbs_with_address(kvm, gfn_start,
|
|
|
|
gfn_end - gfn_start + 1);
|
|
|
|
|
2015-05-13 06:42:27 +00:00
|
|
|
spin_unlock(&kvm->mmu_lock);
|
|
|
|
}
|
|
|
|
|
2015-11-20 08:41:28 +00:00
|
|
|
static bool slot_rmap_write_protect(struct kvm *kvm,
|
|
|
|
struct kvm_rmap_head *rmap_head)
|
2015-05-13 06:42:24 +00:00
|
|
|
{
|
2015-11-20 08:41:28 +00:00
|
|
|
return __rmap_write_protect(kvm, rmap_head, false);
|
2015-05-13 06:42:24 +00:00
|
|
|
}
|
|
|
|
|
2015-01-28 02:54:26 +00:00
|
|
|
void kvm_mmu_slot_remove_write_access(struct kvm *kvm,
|
|
|
|
struct kvm_memory_slot *memslot)
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
{
|
2015-05-13 06:42:24 +00:00
|
|
|
bool flush;
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
|
2013-01-08 10:46:48 +00:00
|
|
|
spin_lock(&kvm->mmu_lock);
|
2015-05-13 06:42:24 +00:00
|
|
|
flush = slot_handle_all_level(kvm, memslot, slot_rmap_write_protect,
|
|
|
|
false);
|
2013-01-08 10:46:48 +00:00
|
|
|
spin_unlock(&kvm->mmu_lock);
|
2014-04-17 09:06:16 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* kvm_mmu_slot_remove_write_access() and kvm_vm_ioctl_get_dirty_log()
|
|
|
|
* which do tlb flush out of mmu-lock should be serialized by
|
|
|
|
* kvm->slots_lock otherwise tlb flush would be missed.
|
|
|
|
*/
|
|
|
|
lockdep_assert_held(&kvm->slots_lock);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We can flush all the TLBs out of the mmu lock without TLB
|
|
|
|
* corruption since we just change the spte from writable to
|
|
|
|
* readonly so that we only need to care the case of changing
|
|
|
|
* spte from present to present (changing the spte from present
|
|
|
|
* to nonpresent will flush all the TLBs immediately), in other
|
|
|
|
* words, the only case we care is mmu_spte_update() where we
|
2018-11-05 06:45:03 +00:00
|
|
|
* have checked SPTE_HOST_WRITEABLE | SPTE_MMU_WRITEABLE
|
2014-04-17 09:06:16 +00:00
|
|
|
* instead of PT_WRITABLE_MASK, that means it does not depend
|
|
|
|
* on PT_WRITABLE_MASK anymore.
|
|
|
|
*/
|
2015-01-12 07:28:54 +00:00
|
|
|
if (flush)
|
2018-12-06 13:21:09 +00:00
|
|
|
kvm_flush_remote_tlbs_with_address(kvm, memslot->base_gfn,
|
|
|
|
memslot->npages);
|
[PATCH] kvm: userspace interface
web site: http://kvm.sourceforge.net
mailing list: kvm-devel@lists.sourceforge.net
(http://lists.sourceforge.net/lists/listinfo/kvm-devel)
The following patchset adds a driver for Intel's hardware virtualization
extensions to the x86 architecture. The driver adds a character device
(/dev/kvm) that exposes the virtualization capabilities to userspace. Using
this driver, a process can run a virtual machine (a "guest") in a fully
virtualized PC containing its own virtual hard disks, network adapters, and
display.
Using this driver, one can start multiple virtual machines on a host.
Each virtual machine is a process on the host; a virtual cpu is a thread in
that process. kill(1), nice(1), top(1) work as expected. In effect, the
driver adds a third execution mode to the existing two: we now have kernel
mode, user mode, and guest mode. Guest mode has its own address space mapping
guest physical memory (which is accessible to user mode by mmap()ing
/dev/kvm). Guest mode has no access to any I/O devices; any such access is
intercepted and directed to user mode for emulation.
The driver supports i386 and x86_64 hosts and guests. All combinations are
allowed except x86_64 guest on i386 host. For i386 guests and hosts, both pae
and non-pae paging modes are supported.
SMP hosts and UP guests are supported. At the moment only Intel
hardware is supported, but AMD virtualization support is being worked on.
Performance currently is non-stellar due to the naive implementation of the
mmu virtualization, which throws away most of the shadow page table entries
every context switch. We plan to address this in two ways:
- cache shadow page tables across tlb flushes
- wait until AMD and Intel release processors with nested page tables
Currently a virtual desktop is responsive but consumes a lot of CPU. Under
Windows I tried playing pinball and watching a few flash movies; with a recent
CPU one can hardly feel the virtualization. Linux/X is slower, probably due
to X being in a separate process.
In addition to the driver, you need a slightly modified qemu to provide I/O
device emulation and the BIOS.
Caveats (akpm: might no longer be true):
- The Windows install currently bluescreens due to a problem with the
virtual APIC. We are working on a fix. A temporary workaround is to
use an existing image or install through qemu
- Windows 64-bit does not work. That's also true for qemu, so it's
probably a problem with the device model.
[bero@arklinux.org: build fix]
[simon.kagstrom@bth.se: build fix, other fixes]
[uril@qumranet.com: KVM: Expose interrupt bitmap]
[akpm@osdl.org: i386 build fix]
[mingo@elte.hu: i386 fixes]
[rdreier@cisco.com: add log levels to all printks]
[randy.dunlap@oracle.com: Fix sparse NULL and C99 struct init warnings]
[anthony@codemonkey.ws: KVM: AMD SVM: 32-bit host support]
Signed-off-by: Yaniv Kamay <yaniv@qumranet.com>
Signed-off-by: Avi Kivity <avi@qumranet.com>
Cc: Simon Kagstrom <simon.kagstrom@bth.se>
Cc: Bernhard Rosenkraenzer <bero@arklinux.org>
Signed-off-by: Uri Lublin <uril@qumranet.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Roland Dreier <rolandd@cisco.com>
Signed-off-by: Randy Dunlap <randy.dunlap@oracle.com>
Signed-off-by: Anthony Liguori <anthony@codemonkey.ws>
Signed-off-by: Andrew Morton <akpm@osdl.org>
Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-12-10 10:21:36 +00:00
|
|
|
}
|
2007-01-06 00:36:56 +00:00
|
|
|
|
2015-04-03 07:40:25 +00:00
|
|
|
static bool kvm_mmu_zap_collapsible_spte(struct kvm *kvm,
|
2015-11-20 08:41:28 +00:00
|
|
|
struct kvm_rmap_head *rmap_head)
|
2015-04-03 07:40:25 +00:00
|
|
|
{
|
|
|
|
u64 *sptep;
|
|
|
|
struct rmap_iterator iter;
|
|
|
|
int need_tlb_flush = 0;
|
kvm: rename pfn_t to kvm_pfn_t
To date, we have implemented two I/O usage models for persistent memory,
PMEM (a persistent "ram disk") and DAX (mmap persistent memory into
userspace). This series adds a third, DAX-GUP, that allows DAX mappings
to be the target of direct-i/o. It allows userspace to coordinate
DMA/RDMA from/to persistent memory.
The implementation leverages the ZONE_DEVICE mm-zone that went into
4.3-rc1 (also discussed at kernel summit) to flag pages that are owned
and dynamically mapped by a device driver. The pmem driver, after
mapping a persistent memory range into the system memmap via
devm_memremap_pages(), arranges for DAX to distinguish pfn-only versus
page-backed pmem-pfns via flags in the new pfn_t type.
The DAX code, upon seeing a PFN_DEV+PFN_MAP flagged pfn, flags the
resulting pte(s) inserted into the process page tables with a new
_PAGE_DEVMAP flag. Later, when get_user_pages() is walking ptes it keys
off _PAGE_DEVMAP to pin the device hosting the page range active.
Finally, get_page() and put_page() are modified to take references
against the device driver established page mapping.
Finally, this need for "struct page" for persistent memory requires
memory capacity to store the memmap array. Given the memmap array for a
large pool of persistent may exhaust available DRAM introduce a
mechanism to allocate the memmap from persistent memory. The new
"struct vmem_altmap *" parameter to devm_memremap_pages() enables
arch_add_memory() to use reserved pmem capacity rather than the page
allocator.
This patch (of 18):
The core has developed a need for a "pfn_t" type [1]. Move the existing
pfn_t in KVM to kvm_pfn_t [2].
[1]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002199.html
[2]: https://lists.01.org/pipermail/linux-nvdimm/2015-September/002218.html
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Acked-by: Christoffer Dall <christoffer.dall@linaro.org>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-01-16 00:56:11 +00:00
|
|
|
kvm_pfn_t pfn;
|
2015-04-03 07:40:25 +00:00
|
|
|
struct kvm_mmu_page *sp;
|
|
|
|
|
2015-05-13 06:42:20 +00:00
|
|
|
restart:
|
2015-11-20 08:41:28 +00:00
|
|
|
for_each_rmap_spte(rmap_head, &iter, sptep) {
|
2015-04-03 07:40:25 +00:00
|
|
|
sp = page_header(__pa(sptep));
|
|
|
|
pfn = spte_to_pfn(*sptep);
|
|
|
|
|
|
|
|
/*
|
2015-04-14 04:04:10 +00:00
|
|
|
* We cannot do huge page mapping for indirect shadow pages,
|
|
|
|
* which are found on the last rmap (level = 1) when not using
|
|
|
|
* tdp; such shadow pages are synced with the page table in
|
|
|
|
* the guest, and the guest page table is using 4K page size
|
|
|
|
* mapping if the indirect sp has level = 1.
|
2015-04-03 07:40:25 +00:00
|
|
|
*/
|
|
|
|
if (sp->role.direct &&
|
|
|
|
!kvm_is_reserved_pfn(pfn) &&
|
mm: thp: kvm: fix memory corruption in KVM with THP enabled
After the THP refcounting change, obtaining a compound pages from
get_user_pages() no longer allows us to assume the entire compound page
is immediately mappable from a secondary MMU.
A secondary MMU doesn't want to call get_user_pages() more than once for
each compound page, in order to know if it can map the whole compound
page. So a secondary MMU needs to know from a single get_user_pages()
invocation when it can map immediately the entire compound page to avoid
a flood of unnecessary secondary MMU faults and spurious
atomic_inc()/atomic_dec() (pages don't have to be pinned by MMU notifier
users).
Ideally instead of the page->_mapcount < 1 check, get_user_pages()
should return the granularity of the "page" mapping in the "mm" passed
to get_user_pages(). However it's non trivial change to pass the "pmd"
status belonging to the "mm" walked by get_user_pages up the stack (up
to the caller of get_user_pages). So the fix just checks if there is
not a single pte mapping on the page returned by get_user_pages, and in
turn if the caller can assume that the whole compound page is mapped in
the current "mm" (in a pmd_trans_huge()). In such case the entire
compound page is safe to map into the secondary MMU without additional
get_user_pages() calls on the surrounding tail/head pages. In addition
of being faster, not having to run other get_user_pages() calls also
reduces the memory footprint of the secondary MMU fault in case the pmd
split happened as result of memory pressure.
Without this fix after a MADV_DONTNEED (like invoked by QEMU during
postcopy live migration or balloning) or after generic swapping (with a
failure in split_huge_page() that would only result in pmd splitting and
not a physical page split), KVM would map the whole compound page into
the shadow pagetables, despite regular faults or userfaults (like
UFFDIO_COPY) may map regular pages into the primary MMU as result of the
pte faults, leading to the guest mode and userland mode going out of
sync and not working on the same memory at all times.
Any other secondary MMU notifier manager (KVM is just one of the many
MMU notifier users) will need the same information if it doesn't want to
run a flood of get_user_pages_fast and it can support multiple
granularity in the secondary MMU mappings, so I think it is justified to
be exposed not just to KVM.
The other option would be to move transparent_hugepage_adjust to
mm/huge_memory.c but that currently has all kind of KVM data structures
in it, so it's definitely not a cut-and-paste work, so I couldn't do a
fix as cleaner as this one for 4.6.
Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Cc: "Dr. David Alan Gilbert" <dgilbert@redhat.com>
Cc: "Kirill A. Shutemov" <kirill@shutemov.name>
Cc: "Li, Liang Z" <liang.z.li@intel.com>
Cc: Amit Shah <amit.shah@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
2016-05-05 23:22:20 +00:00
|
|
|
PageTransCompoundMap(pfn_to_page(pfn))) {
|
2018-10-04 02:04:23 +00:00
|
|
|
pte_list_remove(rmap_head, sptep);
|
2018-12-06 13:21:08 +00:00
|
|
|
|
|
|
|
if (kvm_available_flush_tlb_with_range())
|
|
|
|
kvm_flush_remote_tlbs_with_address(kvm, sp->gfn,
|
|
|
|
KVM_PAGES_PER_HPAGE(sp->role.level));
|
|
|
|
else
|
|
|
|
need_tlb_flush = 1;
|
|
|
|
|
2015-05-13 06:42:20 +00:00
|
|
|
goto restart;
|
|
|
|
}
|
2015-04-03 07:40:25 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return need_tlb_flush;
|
|
|
|
}
|
|
|
|
|
|
|
|
void kvm_mmu_zap_collapsible_sptes(struct kvm *kvm,
|
2015-05-18 11:20:23 +00:00
|
|
|
const struct kvm_memory_slot *memslot)
|
2015-04-03 07:40:25 +00:00
|
|
|
{
|
2015-05-18 11:20:23 +00:00
|
|
|
/* FIXME: const-ify all uses of struct kvm_memory_slot. */
|
2015-04-03 07:40:25 +00:00
|
|
|
spin_lock(&kvm->mmu_lock);
|
2015-05-18 11:20:23 +00:00
|
|
|
slot_handle_leaf(kvm, (struct kvm_memory_slot *)memslot,
|
|
|
|
kvm_mmu_zap_collapsible_spte, true);
|
2015-04-03 07:40:25 +00:00
|
|
|
spin_unlock(&kvm->mmu_lock);
|
|
|
|
}
|
|
|
|
|
2015-01-28 02:54:24 +00:00
|
|
|
void kvm_mmu_slot_leaf_clear_dirty(struct kvm *kvm,
|
|
|
|
struct kvm_memory_slot *memslot)
|
|
|
|
{
|
2015-05-13 06:42:24 +00:00
|
|
|
bool flush;
|
2015-01-28 02:54:24 +00:00
|
|
|
|
|
|
|
spin_lock(&kvm->mmu_lock);
|
2015-05-13 06:42:24 +00:00
|
|
|
flush = slot_handle_leaf(kvm, memslot, __rmap_clear_dirty, false);
|
2015-01-28 02:54:24 +00:00
|
|
|
spin_unlock(&kvm->mmu_lock);
|
|
|
|
|
|
|
|
lockdep_assert_held(&kvm->slots_lock);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* It's also safe to flush TLBs out of mmu lock here as currently this
|
|
|
|
* function is only used for dirty logging, in which case flushing TLB
|
|
|
|
* out of mmu lock also guarantees no dirty pages will be lost in
|
|
|
|
* dirty_bitmap.
|
|
|
|
*/
|
|
|
|
if (flush)
|
2018-12-06 13:21:09 +00:00
|
|
|
kvm_flush_remote_tlbs_with_address(kvm, memslot->base_gfn,
|
|
|
|
memslot->npages);
|
2015-01-28 02:54:24 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_slot_leaf_clear_dirty);
|
|
|
|
|
|
|
|
void kvm_mmu_slot_largepage_remove_write_access(struct kvm *kvm,
|
|
|
|
struct kvm_memory_slot *memslot)
|
|
|
|
{
|
2015-05-13 06:42:24 +00:00
|
|
|
bool flush;
|
2015-01-28 02:54:24 +00:00
|
|
|
|
|
|
|
spin_lock(&kvm->mmu_lock);
|
2015-05-13 06:42:24 +00:00
|
|
|
flush = slot_handle_large_level(kvm, memslot, slot_rmap_write_protect,
|
|
|
|
false);
|
2015-01-28 02:54:24 +00:00
|
|
|
spin_unlock(&kvm->mmu_lock);
|
|
|
|
|
|
|
|
/* see kvm_mmu_slot_remove_write_access */
|
|
|
|
lockdep_assert_held(&kvm->slots_lock);
|
|
|
|
|
|
|
|
if (flush)
|
2018-12-06 13:21:09 +00:00
|
|
|
kvm_flush_remote_tlbs_with_address(kvm, memslot->base_gfn,
|
|
|
|
memslot->npages);
|
2015-01-28 02:54:24 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_slot_largepage_remove_write_access);
|
|
|
|
|
|
|
|
void kvm_mmu_slot_set_dirty(struct kvm *kvm,
|
|
|
|
struct kvm_memory_slot *memslot)
|
|
|
|
{
|
2015-05-13 06:42:24 +00:00
|
|
|
bool flush;
|
2015-01-28 02:54:24 +00:00
|
|
|
|
|
|
|
spin_lock(&kvm->mmu_lock);
|
2015-05-13 06:42:24 +00:00
|
|
|
flush = slot_handle_all_level(kvm, memslot, __rmap_set_dirty, false);
|
2015-01-28 02:54:24 +00:00
|
|
|
spin_unlock(&kvm->mmu_lock);
|
|
|
|
|
|
|
|
lockdep_assert_held(&kvm->slots_lock);
|
|
|
|
|
|
|
|
/* see kvm_mmu_slot_leaf_clear_dirty */
|
|
|
|
if (flush)
|
2018-12-06 13:21:09 +00:00
|
|
|
kvm_flush_remote_tlbs_with_address(kvm, memslot->base_gfn,
|
|
|
|
memslot->npages);
|
2015-01-28 02:54:24 +00:00
|
|
|
}
|
|
|
|
EXPORT_SYMBOL_GPL(kvm_mmu_slot_set_dirty);
|
|
|
|
|
2013-05-31 00:36:27 +00:00
|
|
|
#define BATCH_ZAP_PAGES 10
|
2013-05-31 00:36:22 +00:00
|
|
|
static void kvm_zap_obsolete_pages(struct kvm *kvm)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp, *node;
|
2013-05-31 00:36:27 +00:00
|
|
|
int batch = 0;
|
2013-05-31 00:36:22 +00:00
|
|
|
|
|
|
|
restart:
|
|
|
|
list_for_each_entry_safe_reverse(sp, node,
|
|
|
|
&kvm->arch.active_mmu_pages, link) {
|
2013-05-31 00:36:27 +00:00
|
|
|
int ret;
|
|
|
|
|
2013-05-31 00:36:22 +00:00
|
|
|
/*
|
|
|
|
* No obsolete page exists before new created page since
|
|
|
|
* active_mmu_pages is the FIFO list.
|
|
|
|
*/
|
|
|
|
if (!is_obsolete_sp(kvm, sp))
|
|
|
|
break;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Since we are reversely walking the list and the invalid
|
|
|
|
* list will be moved to the head, skip the invalid page
|
|
|
|
* can help us to avoid the infinity list walking.
|
|
|
|
*/
|
|
|
|
if (sp->role.invalid)
|
|
|
|
continue;
|
|
|
|
|
2013-05-31 00:36:28 +00:00
|
|
|
/*
|
|
|
|
* Need not flush tlb since we only zap the sp with invalid
|
|
|
|
* generation number.
|
|
|
|
*/
|
2013-05-31 00:36:27 +00:00
|
|
|
if (batch >= BATCH_ZAP_PAGES &&
|
2013-05-31 00:36:28 +00:00
|
|
|
cond_resched_lock(&kvm->mmu_lock)) {
|
2013-05-31 00:36:27 +00:00
|
|
|
batch = 0;
|
2013-05-31 00:36:22 +00:00
|
|
|
goto restart;
|
|
|
|
}
|
|
|
|
|
2013-05-31 00:36:29 +00:00
|
|
|
ret = kvm_mmu_prepare_zap_page(kvm, sp,
|
|
|
|
&kvm->arch.zapped_obsolete_pages);
|
2013-05-31 00:36:27 +00:00
|
|
|
batch += ret;
|
|
|
|
|
|
|
|
if (ret)
|
2013-05-31 00:36:22 +00:00
|
|
|
goto restart;
|
|
|
|
}
|
|
|
|
|
2013-05-31 00:36:28 +00:00
|
|
|
/*
|
|
|
|
* Should flush tlb before free page tables since lockless-walking
|
|
|
|
* may use the pages.
|
|
|
|
*/
|
2013-05-31 00:36:29 +00:00
|
|
|
kvm_mmu_commit_zap_page(kvm, &kvm->arch.zapped_obsolete_pages);
|
2013-05-31 00:36:22 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Fast invalidate all shadow pages and use lock-break technique
|
|
|
|
* to zap obsolete pages.
|
|
|
|
*
|
|
|
|
* It's required when memslot is being deleted or VM is being
|
|
|
|
* destroyed, in these cases, we should ensure that KVM MMU does
|
|
|
|
* not use any resource of the being-deleted slot or all slots
|
|
|
|
* after calling the function.
|
|
|
|
*/
|
|
|
|
void kvm_mmu_invalidate_zap_all_pages(struct kvm *kvm)
|
|
|
|
{
|
|
|
|
spin_lock(&kvm->mmu_lock);
|
2013-05-31 00:36:25 +00:00
|
|
|
trace_kvm_mmu_invalidate_zap_all_pages(kvm);
|
2013-05-31 00:36:22 +00:00
|
|
|
kvm->arch.mmu_valid_gen++;
|
|
|
|
|
2013-05-31 00:36:28 +00:00
|
|
|
/*
|
|
|
|
* Notify all vcpus to reload its shadow page table
|
|
|
|
* and flush TLB. Then all vcpus will switch to new
|
|
|
|
* shadow page table with the new mmu_valid_gen.
|
|
|
|
*
|
|
|
|
* Note: we should do this under the protection of
|
|
|
|
* mmu-lock, otherwise, vcpu would purge shadow page
|
|
|
|
* but miss tlb flush.
|
|
|
|
*/
|
|
|
|
kvm_reload_remote_mmus(kvm);
|
|
|
|
|
2013-05-31 00:36:22 +00:00
|
|
|
kvm_zap_obsolete_pages(kvm);
|
|
|
|
spin_unlock(&kvm->mmu_lock);
|
|
|
|
}
|
|
|
|
|
2013-05-31 00:36:29 +00:00
|
|
|
static bool kvm_has_zapped_obsolete_pages(struct kvm *kvm)
|
|
|
|
{
|
|
|
|
return unlikely(!list_empty_careful(&kvm->arch.zapped_obsolete_pages));
|
|
|
|
}
|
|
|
|
|
2019-02-05 21:01:23 +00:00
|
|
|
static void kvm_mmu_zap_mmio_sptes(struct kvm *kvm)
|
|
|
|
{
|
|
|
|
struct kvm_mmu_page *sp, *node;
|
|
|
|
LIST_HEAD(invalid_list);
|
|
|
|
|
|
|
|
spin_lock(&kvm->mmu_lock);
|
|
|
|
restart:
|
|
|
|
list_for_each_entry_safe(sp, node, &kvm->arch.active_mmu_pages, link) {
|
|
|
|
if (!sp->mmio_cached)
|
|
|
|
continue;
|
2019-02-05 21:01:24 +00:00
|
|
|
if (kvm_mmu_prepare_zap_page(kvm, sp, &invalid_list) ||
|
|
|
|
cond_resched_lock(&kvm->mmu_lock))
|
2019-02-05 21:01:23 +00:00
|
|
|
goto restart;
|
|
|
|
}
|
|
|
|
|
|
|
|
kvm_mmu_commit_zap_page(kvm, &invalid_list);
|
|
|
|
spin_unlock(&kvm->mmu_lock);
|
|
|
|
}
|
|
|
|
|
2019-02-05 20:54:17 +00:00
|
|
|
void kvm_mmu_invalidate_mmio_sptes(struct kvm *kvm, u64 gen)
|
2013-06-07 08:51:26 +00:00
|
|
|
{
|
2019-02-05 21:01:18 +00:00
|
|
|
WARN_ON(gen & KVM_MEMSLOT_GEN_UPDATE_IN_PROGRESS);
|
2019-02-05 21:01:12 +00:00
|
|
|
|
2019-02-05 21:01:18 +00:00
|
|
|
gen &= MMIO_SPTE_GEN_MASK;
|
2019-02-05 21:01:12 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Generation numbers are incremented in multiples of the number of
|
|
|
|
* address spaces in order to provide unique generations across all
|
|
|
|
* address spaces. Strip what is effectively the address space
|
|
|
|
* modifier prior to checking for a wrap of the MMIO generation so
|
|
|
|
* that a wrap in any address space is detected.
|
|
|
|
*/
|
|
|
|
gen &= ~((u64)KVM_ADDRESS_SPACE_NUM - 1);
|
|
|
|
|
2013-06-07 08:51:26 +00:00
|
|
|
/*
|
2019-02-05 21:01:12 +00:00
|
|
|
* The very rare case: if the MMIO generation number has wrapped,
|
2013-06-07 08:51:26 +00:00
|
|
|
* zap all shadow pages.
|
|
|
|
*/
|
2019-02-05 21:01:12 +00:00
|
|
|
if (unlikely(gen == 0)) {
|
2016-11-15 06:36:18 +00:00
|
|
|
kvm_debug_ratelimited("kvm: zapping shadow pages for mmio generation wraparound\n");
|
2019-02-05 21:01:23 +00:00
|
|
|
kvm_mmu_zap_mmio_sptes(kvm);
|
2013-06-20 16:34:31 +00:00
|
|
|
}
|
2013-06-07 08:51:26 +00:00
|
|
|
}
|
|
|
|
|
2013-08-28 00:18:14 +00:00
|
|
|
static unsigned long
|
|
|
|
mmu_shrink_scan(struct shrinker *shrink, struct shrink_control *sc)
|
2008-03-30 12:17:21 +00:00
|
|
|
{
|
|
|
|
struct kvm *kvm;
|
2011-05-25 00:12:27 +00:00
|
|
|
int nr_to_scan = sc->nr_to_scan;
|
2013-08-28 00:18:14 +00:00
|
|
|
unsigned long freed = 0;
|
2008-03-30 12:17:21 +00:00
|
|
|
|
2013-09-25 11:53:07 +00:00
|
|
|
spin_lock(&kvm_lock);
|
2008-03-30 12:17:21 +00:00
|
|
|
|
|
|
|
list_for_each_entry(kvm, &vm_list, vm_list) {
|
2011-12-02 17:35:24 +00:00
|
|
|
int idx;
|
2010-06-04 13:55:29 +00:00
|
|
|
LIST_HEAD(invalid_list);
|
2008-03-30 12:17:21 +00:00
|
|
|
|
2012-08-20 09:35:39 +00:00
|
|
|
/*
|
|
|
|
* Never scan more than sc->nr_to_scan VM instances.
|
|
|
|
* Will not hit this condition practically since we do not try
|
|
|
|
* to shrink more than one VM and it is very unlikely to see
|
|
|
|
* !n_used_mmu_pages so many times.
|
|
|
|
*/
|
|
|
|
if (!nr_to_scan--)
|
|
|
|
break;
|
2012-06-04 11:53:23 +00:00
|
|
|
/*
|
|
|
|
* n_used_mmu_pages is accessed without holding kvm->mmu_lock
|
|
|
|
* here. We may skip a VM instance errorneosly, but we do not
|
|
|
|
* want to shrink a VM that only started to populate its MMU
|
|
|
|
* anyway.
|
|
|
|
*/
|
2013-05-31 00:36:29 +00:00
|
|
|
if (!kvm->arch.n_used_mmu_pages &&
|
|
|
|
!kvm_has_zapped_obsolete_pages(kvm))
|
2012-06-04 11:53:23 +00:00
|
|
|
continue;
|
|
|
|
|
2009-12-23 16:35:25 +00:00
|
|
|
idx = srcu_read_lock(&kvm->srcu);
|
2008-03-30 12:17:21 +00:00
|
|
|
spin_lock(&kvm->mmu_lock);
|
|
|
|
|
2013-05-31 00:36:29 +00:00
|
|
|
if (kvm_has_zapped_obsolete_pages(kvm)) {
|
|
|
|
kvm_mmu_commit_zap_page(kvm,
|
|
|
|
&kvm->arch.zapped_obsolete_pages);
|
|
|
|
goto unlock;
|
|
|
|
}
|
|
|
|
|
2013-08-28 00:18:14 +00:00
|
|
|
if (prepare_zap_oldest_mmu_page(kvm, &invalid_list))
|
|
|
|
freed++;
|
2010-06-04 13:55:29 +00:00
|
|
|
kvm_mmu_commit_zap_page(kvm, &invalid_list);
|
2012-06-04 11:53:23 +00:00
|
|
|
|
2013-05-31 00:36:29 +00:00
|
|
|
unlock:
|
2008-03-30 12:17:21 +00:00
|
|
|
spin_unlock(&kvm->mmu_lock);
|
2009-12-23 16:35:25 +00:00
|
|
|
srcu_read_unlock(&kvm->srcu, idx);
|
2012-06-04 11:53:23 +00:00
|
|
|
|
2013-08-28 00:18:14 +00:00
|
|
|
/*
|
|
|
|
* unfair on small ones
|
|
|
|
* per-vm shrinkers cry out
|
|
|
|
* sadness comes quickly
|
|
|
|
*/
|
2012-06-04 11:53:23 +00:00
|
|
|
list_move_tail(&kvm->vm_list, &vm_list);
|
|
|
|
break;
|
2008-03-30 12:17:21 +00:00
|
|
|
}
|
|
|
|
|
2013-09-25 11:53:07 +00:00
|
|
|
spin_unlock(&kvm_lock);
|
2013-08-28 00:18:14 +00:00
|
|
|
return freed;
|
|
|
|
}
|
|
|
|
|
|
|
|
static unsigned long
|
|
|
|
mmu_shrink_count(struct shrinker *shrink, struct shrink_control *sc)
|
|
|
|
{
|
KVM: create aggregate kvm_total_used_mmu_pages value
Of slab shrinkers, the VM code says:
* Note that 'shrink' will be passed nr_to_scan == 0 when the VM is
* querying the cache size, so a fastpath for that case is appropriate.
and it *means* it. Look at how it calls the shrinkers:
nr_before = (*shrinker->shrink)(0, gfp_mask);
shrink_ret = (*shrinker->shrink)(this_scan, gfp_mask);
So, if you do anything stupid in your shrinker, the VM will doubly
punish you.
The mmu_shrink() function takes the global kvm_lock, then acquires
every VM's kvm->mmu_lock in sequence. If we have 100 VMs, then
we're going to take 101 locks. We do it twice, so each call takes
202 locks. If we're under memory pressure, we can have each cpu
trying to do this. It can get really hairy, and we've seen lock
spinning in mmu_shrink() be the dominant entry in profiles.
This is guaranteed to optimize at least half of those lock
aquisitions away. It removes the need to take any of the locks
when simply trying to count objects.
A 'percpu_counter' can be a large object, but we only have one
of these for the entire system. There are not any better
alternatives at the moment, especially ones that handle CPU
hotplug.
Signed-off-by: Dave Hansen <dave@linux.vnet.ibm.com>
Signed-off-by: Tim Pepper <lnxninja@linux.vnet.ibm.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
2010-08-20 01:11:37 +00:00
|
|
|
return percpu_counter_read_positive(&kvm_total_used_mmu_pages);
|
2008-03-30 12:17:21 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static struct shrinker mmu_shrinker = {
|
2013-08-28 00:18:14 +00:00
|
|
|
.count_objects = mmu_shrink_count,
|
|
|
|
.scan_objects = mmu_shrink_scan,
|
2008-03-30 12:17:21 +00:00
|
|
|
.seeks = DEFAULT_SEEKS * 10,
|
|
|
|
};
|
|
|
|
|
2008-05-22 08:37:48 +00:00
|
|
|
static void mmu_destroy_caches(void)
|
2007-04-15 13:31:09 +00:00
|
|
|
{
|
2017-10-08 03:15:23 +00:00
|
|
|
kmem_cache_destroy(pte_list_desc_cache);
|
|
|
|
kmem_cache_destroy(mmu_page_header_cache);
|
2007-04-15 13:31:09 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
int kvm_mmu_module_init(void)
|
|
|
|
{
|
2018-01-10 16:26:59 +00:00
|
|
|
int ret = -ENOMEM;
|
|
|
|
|
2018-10-08 19:28:10 +00:00
|
|
|
/*
|
|
|
|
* MMU roles use union aliasing which is, generally speaking, an
|
|
|
|
* undefined behavior. However, we supposedly know how compilers behave
|
|
|
|
* and the current status quo is unlikely to change. Guardians below are
|
|
|
|
* supposed to let us know if the assumption becomes false.
|
|
|
|
*/
|
|
|
|
BUILD_BUG_ON(sizeof(union kvm_mmu_page_role) != sizeof(u32));
|
|
|
|
BUILD_BUG_ON(sizeof(union kvm_mmu_extended_role) != sizeof(u32));
|
|
|
|
BUILD_BUG_ON(sizeof(union kvm_mmu_role) != sizeof(u64));
|
|
|
|
|
2018-08-14 17:15:34 +00:00
|
|
|
kvm_mmu_reset_all_pte_masks();
|
2016-12-07 00:46:16 +00:00
|
|
|
|
2011-05-15 15:26:20 +00:00
|
|
|
pte_list_desc_cache = kmem_cache_create("pte_list_desc",
|
|
|
|
sizeof(struct pte_list_desc),
|
2017-10-06 01:07:24 +00:00
|
|
|
0, SLAB_ACCOUNT, NULL);
|
2011-05-15 15:26:20 +00:00
|
|
|
if (!pte_list_desc_cache)
|
2018-01-10 16:26:59 +00:00
|
|
|
goto out;
|
2007-04-15 13:31:09 +00:00
|
|
|
|
2007-05-30 09:34:53 +00:00
|
|
|
mmu_page_header_cache = kmem_cache_create("kvm_mmu_page_header",
|
|
|
|
sizeof(struct kvm_mmu_page),
|
2017-10-06 01:07:24 +00:00
|
|
|
0, SLAB_ACCOUNT, NULL);
|
2007-05-30 09:34:53 +00:00
|
|
|
if (!mmu_page_header_cache)
|
2018-01-10 16:26:59 +00:00
|
|
|
goto out;
|
2007-05-30 09:34:53 +00:00
|
|
|
|
2014-09-08 00:51:29 +00:00
|
|
|
if (percpu_counter_init(&kvm_total_used_mmu_pages, 0, GFP_KERNEL))
|
2018-01-10 16:26:59 +00:00
|
|
|
goto out;
|
2010-08-23 08:13:15 +00:00
|
|
|
|
2018-01-10 16:26:59 +00:00
|
|
|
ret = register_shrinker(&mmu_shrinker);
|
|
|
|
if (ret)
|
|
|
|
goto out;
|
2008-03-30 12:17:21 +00:00
|
|
|
|
2007-04-15 13:31:09 +00:00
|
|
|
return 0;
|
|
|
|
|
2018-01-10 16:26:59 +00:00
|
|
|
out:
|
2008-03-30 12:17:21 +00:00
|
|
|
mmu_destroy_caches();
|
2018-01-10 16:26:59 +00:00
|
|
|
return ret;
|
2007-04-15 13:31:09 +00:00
|
|
|
}
|
|
|
|
|
2007-11-20 05:11:38 +00:00
|
|
|
/*
|
2018-10-04 15:45:00 +00:00
|
|
|
* Calculate mmu pages needed for kvm.
|
2007-11-20 05:11:38 +00:00
|
|
|
*/
|
|
|
|
unsigned int kvm_mmu_calculate_mmu_pages(struct kvm *kvm)
|
|
|
|
{
|
|
|
|
unsigned int nr_mmu_pages;
|
|
|
|
unsigned int nr_pages = 0;
|
2009-12-23 16:35:21 +00:00
|
|
|
struct kvm_memslots *slots;
|
2011-11-24 09:39:18 +00:00
|
|
|
struct kvm_memory_slot *memslot;
|
2015-05-18 11:33:16 +00:00
|
|
|
int i;
|
2007-11-20 05:11:38 +00:00
|
|
|
|
2015-05-18 11:33:16 +00:00
|
|
|
for (i = 0; i < KVM_ADDRESS_SPACE_NUM; i++) {
|
|
|
|
slots = __kvm_memslots(kvm, i);
|
2010-04-19 09:41:23 +00:00
|
|
|
|
2015-05-18 11:33:16 +00:00
|
|
|
kvm_for_each_memslot(memslot, slots)
|
|
|
|
nr_pages += memslot->npages;
|
|
|
|
}
|
2007-11-20 05:11:38 +00:00
|
|
|
|
|
|
|
nr_mmu_pages = nr_pages * KVM_PERMILLE_MMU_PAGES / 1000;
|
|
|
|
nr_mmu_pages = max(nr_mmu_pages,
|
2015-05-18 11:33:16 +00:00
|
|
|
(unsigned int) KVM_MIN_ALLOC_MMU_PAGES);
|
2007-11-20 05:11:38 +00:00
|
|
|
|
|
|
|
return nr_mmu_pages;
|
|
|
|
}
|
|
|
|
|
2010-09-27 10:07:07 +00:00
|
|
|
void kvm_mmu_destroy(struct kvm_vcpu *vcpu)
|
|
|
|
{
|
2013-10-02 14:56:12 +00:00
|
|
|
kvm_mmu_unload(vcpu);
|
2010-09-27 10:07:07 +00:00
|
|
|
free_mmu_pages(vcpu);
|
|
|
|
mmu_free_memory_caches(vcpu);
|
2010-12-23 08:08:35 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
void kvm_mmu_module_exit(void)
|
|
|
|
{
|
|
|
|
mmu_destroy_caches();
|
|
|
|
percpu_counter_destroy(&kvm_total_used_mmu_pages);
|
|
|
|
unregister_shrinker(&mmu_shrinker);
|
2010-09-27 10:07:07 +00:00
|
|
|
mmu_audit_disable();
|
|
|
|
}
|