Linux (among the others) has checks to make sure that certain features
aren't enabled on a certain family/model/stepping if the microcode version
isn't greater than or equal to a known good version.

By exposing the real microcode version, we're preventing buggy guests that
don't check that they are running virtualized (i.e., they should trust the
hypervisor) from disabling features that are effectively not buggy.
Suggested-by: Filippo Sironi <sironi@amazon.de>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Liran Alon <liran.alon@oracle.com>
Cc: Nadav Amit <nadav.amit@gmail.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>

Introduce kvm_get_msr_feature() to handle the msrs which are supported
by different vendors and sharing the same emulation logic.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Liran Alon <liran.alon@oracle.com>
Cc: Nadav Amit <nadav.amit@gmail.com>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>

In order to determine if LFENCE is a serializing instruction on AMD
processors, MSR 0xc0011029 (MSR_F10H_DECFG) must be read and the state
of bit 1 checked.  This patch will add support to allow a guest to
properly make this determination.

Add the MSR feature callback operation to svm.c and add MSR 0xc0011029
to the list of MSR-based features.  If LFENCE is serializing, then the
feature is supported, allowing the hypervisor to set the value of the
MSR that guest will see.  Support is also added to write (hypervisor only)
and read the MSR value for the guest.  A write by the guest will result in
a #GP.  A read by the guest will return the value as set by the host.  In
this way, the support to expose the feature to the guest is controlled by
the hypervisor.
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>

Provide a new KVM capability that allows bits within MSRs to be recognized
as features.  Two new ioctls are added to the /dev/kvm ioctl routine to
retrieve the list of these MSRs and then retrieve their values. A kvm_x86_ops
callback is used to determine support for the listed MSR-based features.
Signed-off-by: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
[Tweaked documentation. - Radim]
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>

Pull KVM fixes from Paolo Bonzini:
 "s390:
   - optimization for the exitless interrupt support that was merged in 4.16-rc1
   - improve the branch prediction blocking for nested KVM
   - replace some jump tables with switch statements to improve expoline performance
   - fixes for multiple epoch facility

  ARM:
   - fix the interaction of userspace irqchip VMs with in-kernel irqchip VMs
   - make sure we can build 32-bit KVM/ARM with gcc-8.

  x86:
   - fixes for AMD SEV
   - fixes for Intel nested VMX, emulated UMIP and a dump_stack() on VM startup
   - fixes for async page fault migration
   - small optimization to PV TLB flush (new in 4.16-rc1)
   - syzkaller fixes

  Generic:
   - compiler warning fixes
   - syzkaller fixes
   - more improvements to the kvm_stat tool

  Two more small Spectre fixes are going to reach you via Ingo"

* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (40 commits)
  KVM: SVM: Fix SEV LAUNCH_SECRET command
  KVM: SVM: install RSM intercept
  KVM: SVM: no need to call access_ok() in LAUNCH_MEASURE command
  include: psp-sev: Capitalize invalid length enum
  crypto: ccp: Fix sparse, use plain integer as NULL pointer
  KVM: X86: Avoid traversing all the cpus for pv tlb flush when steal time is disabled
  x86/kvm: Make parse_no_xxx __init for kvm
  KVM: x86: fix backward migration with async_PF
  kvm: fix warning for non-x86 builds
  kvm: fix warning for CONFIG_HAVE_KVM_EVENTFD builds
  tools/kvm_stat: print 'Total' line for multiple events only
  tools/kvm_stat: group child events indented after parent
  tools/kvm_stat: separate drilldown and fields filtering
  tools/kvm_stat: eliminate extra guest/pid selection dialog
  tools/kvm_stat: mark private methods as such
  tools/kvm_stat: fix debugfs handling
  tools/kvm_stat: print error on invalid regex
  tools/kvm_stat: fix crash when filtering out all non-child trace events
  tools/kvm_stat: avoid 'is' for equality checks
  tools/kvm_stat: use a more pythonic way to iterate over dictionaries
  ...

Pull Xtensa fixes from Max Filippov:
 "Two fixes for reserved memory/DMA buffers allocation in high memory on
  xtensa architecture

   - fix memory accounting when reserved memory is in high memory region

   - fix DMA allocation from high memory"

* tag 'xtensa-20180225' of git://github.com/jcmvbkbc/linux-xtensa:
  xtensa: support DMA buffers in high memory
  xtensa: fix high memory/reserved memory collision

Pull x86 fixes from Thomas Gleixner:
 "A small set of fixes:

   - UAPI data type correction for hyperv

   - correct the cpu cores field in /proc/cpuinfo on CPU hotplug

   - return proper error code in the resctrl file system failure path to
     avoid silent subsequent failures

   - correct a subtle accounting issue in the new vector allocation code
     which went unnoticed for a while and caused suspend/resume
     failures"

* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/topology: Update the 'cpu cores' field in /proc/cpuinfo correctly across CPU hotplug operations
  x86/topology: Fix function name in documentation
  x86/intel_rdt: Fix incorrect returned value when creating rdgroup sub-directory in resctrl file system
  x86/apic/vector: Handle vector release on CPU unplug correctly
  genirq/matrix: Handle CPU offlining proper
  x86/headers/UAPI: Use __u64 instead of u64 in <uapi/asm/hyperv.h>

Pull perf fix from Thomas Gleixner:
 "A single commit which shuts up a bogus GCC-8 warning"

* 'perf-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  x86/oprofile: Fix bogus GCC-8 warning in nmi_setup()

Pull locking fixes from Thomas Gleixner:
 "Three patches to fix memory ordering issues on ALPHA and a comment to
  clarify the usage scope of a mutex internal function"

* 'locking-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  locking/xchg/alpha: Fix xchg() and cmpxchg() memory ordering bugs
  locking/xchg/alpha: Clean up barrier usage by using smp_mb() in place of __ASM__MB
  locking/xchg/alpha: Add unconditional memory barrier to cmpxchg()
  locking/mutex: Add comment to __mutex_owner() to deter usage

Pull cleanup patchlet from Thomas Gleixner:
 "A single commit removing a bunch of bogus double semicolons all over
  the tree"

* 'core-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
  treewide/trivial: Remove ';;$' typo noise

Pull NFS client bugfixes from Trond Myklebust:

 - fix a broken cast in nfs4_callback_recallany()

 - fix an Oops during NFSv4 migration events

 - make struct nlmclnt_fl_close_lock_ops static

* tag 'nfs-for-4.16-3' of git://git.linux-nfs.org/projects/trondmy/linux-nfs:
  NFS: make struct nlmclnt_fl_close_lock_ops static
  nfs: system crashes after NFS4ERR_MOVED recovery
  NFSv4: Fix broken cast in nfs4_callback_recallany()

Pull powerpc fixes from Michael Ellerman:

 - Add handling for a missing instruction in our 32-bit BPF JIT so that
   it can be used for seccomp filtering.

 - Add a missing NULL pointer check before a function call in new EEH
   code.

 - Fix an error path in the new ocxl driver to correctly return EFAULT.

 - The support for the new ibm,drc-info device tree property turns out
   to need several fixes, so for now we just stop advertising to
   firmware that we support it until the bugs can be ironed out.

 - One fix for the new drmem code which was incorrectly modifying the
   device tree in place.

 - Finally two fixes for the RFI flush support, so that firmware can
   advertise to us that it should be disabled entirely so as not to
   affect performance.

Thanks to: Bharata B Rao, Frederic Barrat, Juan J. Alvarez, Mark Lord,
Michael Bringmann.

* tag 'powerpc-4.16-4' of git://git.kernel.org/pub/scm/linux/kernel/git/powerpc/linux:
  powerpc/powernv: Support firmware disable of RFI flush
  powerpc/pseries: Support firmware disable of RFI flush
  powerpc/mm/drmem: Fix unexpected flag value in ibm,dynamic-memory-v2
  powerpc/bpf/jit: Fix 32-bit JIT for seccomp_data access
  powerpc/pseries: Revert support for ibm,drc-info devtree property
  powerpc/pseries: Fix duplicate firmware feature for DRC_INFO
  ocxl: Fix potential bad errno on irq allocation
  powerpc/eeh: Fix crashes in eeh_report_resume()

The SEV LAUNCH_SECRET command fails with error code 'invalid param'
because we missed filling the guest and header system physical address
while issuing the command.

Fixes: 9f5b5b95 (KVM: SVM: Add support for SEV LAUNCH_SECRET command)
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: linux-kernel@vger.kernel.org
Cc: Joerg Roedel <joro@8bytes.org>
Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

RSM instruction is used by the SMM handler to return from SMM mode.
Currently, rsm causes a #UD - which results in instruction fetch, decode,
and emulate. By installing the RSM intercept we can avoid the instruction
fetch since we know that #VMEXIT was due to rsm.

The patch is required for the SEV guest, because in case of SEV guest
memory is encrypted with guest-specific key and hypervisor will not
able to fetch the instruction bytes from the guest memory.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Joerg Roedel <joro@8bytes.org>
Cc: Borislav Petkov <bp@suse.de>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Using the access_ok() to validate the input before issuing the SEV
command does not buy us anything in this case. If userland is
giving us a garbage pointer then copy_to_user() will catch it when we try
to return the measurement.
Suggested-by: Al Viro <viro@ZenIV.linux.org.uk>
Fixes: 0d0736f7 (KVM: SVM: Add support for KVM_SEV_LAUNCH_MEASURE ...)
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: linux-kernel@vger.kernel.org
Cc: Joerg Roedel <joro@8bytes.org>
Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Commit 1d57b17c ("crypto: ccp: Define SEV userspace ioctl and command
id") added the invalid length enum but we missed capitalizing it.

Fixes: 1d57b17c (crypto: ccp: Define SEV userspace ioctl ...)
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
CC: Gary R Hook <gary.hook@amd.com>
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Fix sparse warning: Using plain integer as NULL pointer. Replaces
assignment of 0 to pointer with NULL assignment.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: Herbert Xu <herbert@gondor.apana.org.au>
Cc: Gary Hook <gary.hook@amd.com>
Cc: Tom Lendacky <thomas.lendacky@amd.com>
Cc: linux-crypto@vger.kernel.org
Cc: linux-kernel@vger.kernel.org
Signed-off-by: Brijesh Singh <brijesh.singh@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Avoid traversing all the cpus for pv tlb flush when steal time
is disabled since pv tlb flush depends on the field in steal time
for shared data.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim KrÄmář <rkrcmar@redhat.com>
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

The early_param() is only called during kernel initialization, So Linux
marks the functions of it with __init macro to save memory.

But it forgot to mark the parse_no_kvmapf/stealacc/kvmclock_vsyscall,
So, Make them __init as well.

Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: rkrcmar@redhat.com
Cc: kvm@vger.kernel.org
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Juergen Gross <jgross@suse.com>
Cc: x86@kernel.org
Signed-off-by: Dou Liyang <douly.fnst@cn.fujitsu.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Guests on new hypersiors might set KVM_ASYNC_PF_DELIVERY_AS_PF_VMEXIT
bit when enabling async_PF, but this bit is reserved on old hypervisors,
which results in a failure upon migration.

To avoid breaking different cases, we are checking for CPUID feature bit
before enabling the feature and nothing else.

Fixes: 52a5c155 ("KVM: async_pf: Let guest support delivery of async_pf from guest mode")
Cc: <stable@vger.kernel.org>
Reviewed-by: Wanpeng Li <wanpengli@tencent.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Fix the following sparse warning by moving the prototype
of kvm_arch_mmu_notifier_invalidate_range() to linux/kvm_host.h .

  CHECK   arch/s390/kvm/../../../virt/kvm/kvm_main.c
arch/s390/kvm/../../../virt/kvm/kvm_main.c:138:13: warning: symbol 'kvm_arch_mmu_notifier_invalidate_range' was not declared. Should it be static?
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Move the kvm_arch_irq_routing_update() prototype outside of
ifdef CONFIG_HAVE_KVM_EVENTFD guards to fix the following sparse warning:

arch/s390/kvm/../../../virt/kvm/irqchip.c:171:28: warning: symbol 'kvm_arch_irq_routing_update' was not declared. Should it be static?
Signed-off-by: Sebastian Ott <sebott@linux.vnet.ibm.com>
Acked-by: Christian Borntraeger <borntraeger@de.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

The 'Total' line looks a bit weird when we have a single event only. This
can happen e.g. due to filters. Therefore suppress when there's only a
single event in the output.
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

We keep the current logic that sorts all events (parent and child), but
re-shuffle the events afterwards, grouping the children after the
respective parent. Note that the percentage column for child events
gives the percentage of the parent's total.
Since we rework the logic anyway, we modify the total average
calculation to use the raw numbers instead of the (rounded) averages.
Note that this can result in differing numbers (between total average
and the sum of the individual averages) due to rounding errors.
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Drilldown (i.e. toggle display of child trace events) was implemented by
overriding the fields filter. This resulted in inconsistencies: E.g. when
drilldown was not active, adding a filter that also matches child trace
events would not only filter fields according to the filter, but also add
in the child trace events matching the filter. E.g. on x86, setting
'kvm_userspace_exit' as the fields filter after startup would result in
display of kvm_userspace_exit(DCR), although that wasn't previously
present - not exactly what one would expect from a filter.
This patch addresses the issue by keeping drilldown and fields filter
separate. While at it, we also fix a PEP8 issue by adding a blank line
at one place (since we're in the area...).
We implement this by adding a framework that also allows to define a
taxonomy among the debugfs events to identify child trace events. I.e.
drilldown using 'x' can now also work with debugfs. A respective parent-
child relationship is only known for S390 at the moment, but could be
added adjusting other platforms' ARCH.dbg_is_child() methods
accordingly.
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

We can do with a single dialog that takes both, pids and guest names.
Note that we keep both interactive commands, 'p' and 'g' for now, to
avoid confusion among users used to a specific key.

While at it, we improve on some minor glitches regarding curses usage,
e.g. cursor still visible when not supposed to be.
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Helps quite a bit reading the code when it's obvious when a method is
intended for internal use only.
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Te checks for debugfs assumed that debugfs is always mounted at
/sys/kernel/debug - which is likely, but not guaranteed. This is addressed
by checking /proc/mounts for the actual location.
Furthermore, when debugfs was mounted, but the kvm module not loaded, a
misleading error pointing towards debugfs not present was given.
To reproduce,
(a) run kvm_stat with debugfs mounted at a place different from
    /sys/kernel/debug
(b) run kvm_stat with debugfs mounted but kvm module not loaded
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Entering an invalid regular expression did not produce any indication of an
error so far.
To reproduce, press 'f' and enter 'foo(' (with an unescaped bracket).
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

When we apply a filter that will only leave child trace events, we
receive a ZeroDivisionError when calculating the percentages.
In that case, provide percentages based on child events only.
To reproduce, run 'kvm_stat -f .*[\(].*'.
Signed-off-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Use '==' for equality checks and 'is' when comparing identities.

An example where '==' and 'is' behave differently:
>>> a = 4242
>>> a == 4242
True
>>> a is 4242
False
Signed-off-by: Marc Hartmayer <mhartmay@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

If it's clear that the values of a dictionary will be used then use
the '.items()' method.
Signed-off-by: Marc Hartmayer <mhartmay@linux.vnet.ibm.com>
Tested-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
[Include fix for logging mode by Stefan Raspl]
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Use a namedtuple for storing the values as it allows to access the
fields of a tuple via names. This makes the overall code much easier
to read and to understand. Access by index is still possible as
before.
Signed-off-by: Marc Hartmayer <mhartmay@linux.vnet.ibm.com>
Tested-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

The 'sortkey' function references a value in its enclosing
scope (closure). This is not common practice for a sort key function
so let's replace it. Additionally, the function 'sorted' has already a
parameter for reversing the result therefore the inversion of the
values is unneeded. The check for stats[x][1] is also superfluous as
it's ensured that this value is initialized with 0.
Signed-off-by: Marc Hartmayer <mhartmay@linux.vnet.ibm.com>
Tested-by: Stefan Raspl <raspl@linux.vnet.ibm.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Reported by syzkaller:

   WARNING: CPU: 6 PID: 2434 at arch/x86/kvm/vmx.c:6660 handle_ept_misconfig+0x54/0x1e0 [kvm_intel]
   CPU: 6 PID: 2434 Comm: repro_test Not tainted 4.15.0+ #4
   RIP: 0010:handle_ept_misconfig+0x54/0x1e0 [kvm_intel]
   Call Trace:
    vmx_handle_exit+0xbd/0xe20 [kvm_intel]
    kvm_arch_vcpu_ioctl_run+0xdaf/0x1d50 [kvm]
    kvm_vcpu_ioctl+0x3e9/0x720 [kvm]
    do_vfs_ioctl+0xa4/0x6a0
    SyS_ioctl+0x79/0x90
    entry_SYSCALL_64_fastpath+0x25/0x9c

The testcase creates a first thread to issue KVM_SMI ioctl, and then creates
a second thread to mmap and operate on the same vCPU.  This triggers a race
condition when running the testcase with multiple threads. Sometimes one thread
exits with a triple fault while another thread mmaps and operates on the same
vCPU.  Because CS=0x3000/IP=0x8000 is not mapped, accessing the SMI handler
results in an EPT misconfig. This patch fixes it by returning RET_PF_EMULATE
in kvm_handle_bad_page(), which will go on to cause an emulation failure and an
exit with KVM_EXIT_INTERNAL_ERROR.

Reported-by: syzbot+c1d9517cab094dae65e446c0c5b4de6c40f4dc58@syzkaller.appspotmail.com
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Wanpeng Li <wanpengli@tencent.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Although L2 is in halt state, it will be in the active state after
VM entry if the VM entry is vectoring according to SDM 26.6.2 Activity
State. Halting the vcpu here means the event won't be injected to L2
and this decision isn't reported to L1. Thus L0 drops an event that
should be injected to L2.

Cc: Liran Alon <liran.alon@oracle.com>
Reviewed-by: Liran Alon <liran.alon@oracle.com>
Signed-off-by: Chao Gao <chao.gao@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

Reported by syzkaller:

    pte_list_remove: ffff9714eb1f8078 0->BUG
    ------------[ cut here ]------------
    kernel BUG at arch/x86/kvm/mmu.c:1157!
    invalid opcode: 0000 [#1] SMP
    RIP: 0010:pte_list_remove+0x11b/0x120 [kvm]
    Call Trace:
     drop_spte+0x83/0xb0 [kvm]
     mmu_page_zap_pte+0xcc/0xe0 [kvm]
     kvm_mmu_prepare_zap_page+0x81/0x4a0 [kvm]
     kvm_mmu_invalidate_zap_all_pages+0x159/0x220 [kvm]
     kvm_arch_flush_shadow_all+0xe/0x10 [kvm]
     kvm_mmu_notifier_release+0x6c/0xa0 [kvm]
     ? kvm_mmu_notifier_release+0x5/0xa0 [kvm]
     __mmu_notifier_release+0x79/0x110
     ? __mmu_notifier_release+0x5/0x110
     exit_mmap+0x15a/0x170
     ? do_exit+0x281/0xcb0
     mmput+0x66/0x160
     do_exit+0x2c9/0xcb0
     ? __context_tracking_exit.part.5+0x4a/0x150
     do_group_exit+0x50/0xd0
     SyS_exit_group+0x14/0x20
     do_syscall_64+0x73/0x1f0
     entry_SYSCALL64_slow_path+0x25/0x25

The reason is that when creates new memslot, there is no guarantee for new
memslot not overlap with private memslots. This can be triggered by the
following program:

   #include <fcntl.h>
   #include <pthread.h>
   #include <setjmp.h>
   #include <signal.h>
   #include <stddef.h>
   #include <stdint.h>
   #include <stdio.h>
   #include <stdlib.h>
   #include <string.h>
   #include <sys/ioctl.h>
   #include <sys/stat.h>
   #include <sys/syscall.h>
   #include <sys/types.h>
   #include <unistd.h>
   #include <linux/kvm.h>

   long r[16];

   int main()
   {
	void *p = valloc(0x4000);

	r[2] = open("/dev/kvm", 0);
	r[3] = ioctl(r[2], KVM_CREATE_VM, 0x0ul);

	uint64_t addr = 0xf000;
	ioctl(r[3], KVM_SET_IDENTITY_MAP_ADDR, &addr);
	r[6] = ioctl(r[3], KVM_CREATE_VCPU, 0x0ul);
	ioctl(r[3], KVM_SET_TSS_ADDR, 0x0ul);
	ioctl(r[6], KVM_RUN, 0);
	ioctl(r[6], KVM_RUN, 0);

	struct kvm_userspace_memory_region mr = {
		.slot = 0,
		.flags = KVM_MEM_LOG_DIRTY_PAGES,
		.guest_phys_addr = 0xf000,
		.memory_size = 0x4000,
		.userspace_addr = (uintptr_t) p
	};
	ioctl(r[3], KVM_SET_USER_MEMORY_REGION, &mr);
	return 0;
   }

This patch fixes the bug by not adding a new memslot even if it
overlaps with private memslots.
Reported-by: Dmitry Vyukov <dvyukov@google.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Dmitry Vyukov <dvyukov@google.com>
Cc: Eric Biggers <ebiggers3@gmail.com>
Cc: stable@vger.kernel.org
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
---
 virt/kvm/kvm_main.c | 3 +--
 1 file changed, 1 insertion(+), 2 deletions(-)

On x86, special KVM memslots such as the TSS region have anonymous
memory mappings created on behalf of userspace, and these mappings are
removed when the VM is destroyed.

It is however possible for removing these mappings via vm_munmap() to
fail.  This can most easily happen if the thread receives SIGKILL while
it's waiting to acquire ->mmap_sem.   This triggers the 'WARN_ON(r < 0)'
in __x86_set_memory_region().  syzkaller was able to hit this, using
'exit()' to send the SIGKILL.  Note that while the vm_munmap() failure
results in the mapping not being removed immediately, it is not leaked
forever but rather will be freed when the process exits.

It's not really possible to handle this failure properly, so almost
every other caller of vm_munmap() doesn't check the return value.  It's
a limitation of having the kernel manage these mappings rather than
userspace.

So just remove the WARN_ON() so that users can't spam the kernel log
with this warning.

Fixes: f0d648bd ("KVM: x86: map/unmap private slots in __x86_set_memory_region")
Reported-by: syzbot <syzkaller@googlegroups.com>
Signed-off-by: Eric Biggers <ebiggers@google.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>

L1 might want to use SECONDARY_EXEC_DESC, so we must not clear the VMCS
bit if UMIP is not being emulated.

We must still set the bit when emulating UMIP as the feature can be
passed to L2 where L0 will do the emulation and because L2 can change
CR4 without a VM exit, we should clear the bit if UMIP is disabled.

Fixes: 0367f205 ("KVM: vmx: add support for emulating UMIP")
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>