Commit 4f350c6dbc (kvm: nVMX: Handle deferred early VMLAUNCH/VMRESUME failure
properly) can result in L1(run kvm-unit-tests/run_tests.sh vmx_controls in L1)
null pointer deference and also L0 calltrace when EPT=0 on both L0 and L1.
In L1:
BUG: unable to handle kernel paging request at ffffffffc015bf8f
IP: vmx_vcpu_run+0x202/0x510 [kvm_intel]
PGD 146e13067 P4D 146e13067 PUD 146e15067 PMD 3d2686067 PTE 3d4af9161
Oops: 0003 [#1] PREEMPT SMP
CPU: 2 PID: 1798 Comm: qemu-system-x86 Not tainted 4.14.0-rc4+ #6
RIP: 0010:vmx_vcpu_run+0x202/0x510 [kvm_intel]
Call Trace:
WARNING: kernel stack frame pointer at ffffb86f4988bc18 in qemu-system-x86:1798 has bad value 0000000000000002
In L0:
-----------[ cut here ]------------
WARNING: CPU: 6 PID: 4460 at /home/kernel/linux/arch/x86/kvm//vmx.c:9845 vmx_inject_page_fault_nested+0x130/0x140 [kvm_intel]
CPU: 6 PID: 4460 Comm: qemu-system-x86 Tainted: G OE 4.14.0-rc7+ #25
RIP: 0010:vmx_inject_page_fault_nested+0x130/0x140 [kvm_intel]
Call Trace:
paging64_page_fault+0x500/0xde0 [kvm]
? paging32_gva_to_gpa_nested+0x120/0x120 [kvm]
? nonpaging_page_fault+0x3b0/0x3b0 [kvm]
? __asan_storeN+0x12/0x20
? paging64_gva_to_gpa+0xb0/0x120 [kvm]
? paging64_walk_addr_generic+0x11a0/0x11a0 [kvm]
? lock_acquire+0x2c0/0x2c0
? vmx_read_guest_seg_ar+0x97/0x100 [kvm_intel]
? vmx_get_segment+0x2a6/0x310 [kvm_intel]
? sched_clock+0x1f/0x30
? check_chain_key+0x137/0x1e0
? __lock_acquire+0x83c/0x2420
? kvm_multiple_exception+0xf2/0x220 [kvm]
? debug_check_no_locks_freed+0x240/0x240
? debug_smp_processor_id+0x17/0x20
? __lock_is_held+0x9e/0x100
kvm_mmu_page_fault+0x90/0x180 [kvm]
kvm_handle_page_fault+0x15c/0x310 [kvm]
? __lock_is_held+0x9e/0x100
handle_exception+0x3c7/0x4d0 [kvm_intel]
vmx_handle_exit+0x103/0x1010 [kvm_intel]
? kvm_arch_vcpu_ioctl_run+0x1628/0x2e20 [kvm]
The commit avoids to load host state of vmcs12 as vmcs01's guest state
since vmcs12 is not modified (except for the VM-instruction error field)
if the checking of vmcs control area fails. However, the mmu context is
switched to nested mmu in prepare_vmcs02() and it will not be reloaded
since load_vmcs12_host_state() is skipped when nested VMLAUNCH/VMRESUME
fails. This patch fixes it by reloading mmu context when nested
VMLAUNCH/VMRESUME fails.
Reviewed-by: Jim Mattson <jmattson@google.com>
Reviewed-by: Krish Sadhukhan <krish.sadhukhan@oracle.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Jim Mattson <jmattson@google.com>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
According to the SDM, if the "load IA32_BNDCFGS" VM-entry controls is 1, the
following checks are performed on the field for the IA32_BNDCFGS MSR:
- Bits reserved in the IA32_BNDCFGS MSR must be 0.
- The linear address in bits 63:12 must be canonical.
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Jim Mattson <jmattson@google.com>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
In case of instruction-decode failure or emulation failure,
x86_emulate_instruction() will call reexecute_instruction() which will
attempt to use the cr2 value passed to x86_emulate_instruction().
However, when x86_emulate_instruction() is called from
emulate_instruction(), cr2 is not passed (passed as 0) and therefore
it doesn't make sense to execute reexecute_instruction() logic at all.
Fixes: 51d8b66199 ("KVM: cleanup emulate_instruction")
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Instruction emulation after trapping a #UD exception can result in an
MMIO access, for example when emulating a MOVBE on a processor that
doesn't support the instruction. In this case, the #UD vmexit handler
must exit to user mode, but there wasn't any code to do so. Add it for
both VMX and SVM.
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Wanpeng Li <wanpeng.li@hotmail.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
When running L2, #UD should be intercepted by L1 or just forwarded
directly to L2. It should not reach L0 x86 emulator.
Therefore, set intercept for #UD only based on L1 exception-bitmap.
Also add WARN_ON_ONCE() on L0 #UD intercept handlers to make sure
it is never reached while running L2.
This improves commit ae1f576707 ("KVM: nVMX: Do not emulate #UD while
in guest mode") by removing an unnecessary exit from L2 to L0 on #UD
when L1 doesn't intercept it.
In addition, SVM L0 #UD intercept handler doesn't handle correctly the
case it is raised from L2. In this case, it should forward the #UD to
guest instead of x86 emulator. As done in VMX #UD intercept handler.
This commit fixes this issue as-well.
Signed-off-by: Liran Alon <liran.alon@oracle.com>
Reviewed-by: Nikita Leshenko <nikita.leshchenko@oracle.com>
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Reviewed-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
To simplify testing of these rarely used code paths, add a module parameter
that turns it on. One eventinj.flat test (NMI after iret) fails when
loading kvm_intel with vnmi=0.
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
This is more or less a revert of commit 2c82878b0c ("KVM: VMX: require
virtual NMI support", 2017-03-27); it turns out that Core 2 Duo machines
only had virtual NMIs in some SKUs.
The revert is not trivial because in the meanwhile there have been several
fixes to nested NMI injection. Therefore, the entire vNMI state is moved
to struct loaded_vmcs.
Another change compared to before the patch is a simplification here:
if (unlikely(!cpu_has_virtual_nmis() && vmx->soft_vnmi_blocked &&
!(is_guest_mode(vcpu) && nested_cpu_has_virtual_nmis(
get_vmcs12(vcpu))))) {
The final condition here is always true (because nested_cpu_has_virtual_nmis
is always false) and is removed.
Fixes: 2c82878b0c
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1490803
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Common:
- Python 3 support in kvm_stat
- Accounting of slabs to kmemcg
ARM:
- Optimized arch timer handling for KVM/ARM
- Improvements to the VGIC ITS code and introduction of an ITS reset
ioctl
- Unification of the 32-bit fault injection logic
- More exact external abort matching logic
PPC:
- Support for running hashed page table (HPT) MMU mode on a host that
is using the radix MMU mode; single threaded mode on POWER 9 is
added as a pre-requisite
- Resolution of merge conflicts with the last second 4.14 HPT fixes
- Fixes and cleanups
s390:
- Some initial preparation patches for exitless interrupts and crypto
- New capability for AIS migration
- Fixes
x86:
- Improved emulation of LAPIC timer mode changes, MCi_STATUS MSRs, and
after-reset state
- Refined dependencies for VMX features
- Fixes for nested SMI injection
- A lot of cleanups
-----BEGIN PGP SIGNATURE-----
iQEcBAABCAAGBQJaDayXAAoJEED/6hsPKofo/3UH/3HvlcHt+ADTkCU1/iiKAs+i
0zngIOXIxgHDnV0ww6bV+Znww0BzTYgKCAXX76z603jdpDwG/pzQQcbLDF5ZoJnD
sQtF10gZinWaRsHlfbLqjrHGL2pGDHO1UKBKLJ0bAIyORPZBxs7i+VmrY/blnr9c
0wsybJ8RbvwAxjsDL5jeX/z4NehPupmKUc4Lf0eZdSHwVOf9sjn+MP6jJ0r2JcIb
D+zddPBiLStzN97t4gZpQsrlj3LKrDS+6hY+1TjSvlh+yHKFVFh58VhLm4DuDeb5
bYOAlWJ/gAWEzfvr5Ld+Nd7SqWWn/14logPkQ4gcU4BI/neAOzk4c6hJfCHl1nk=
=593n
-----END PGP SIGNATURE-----
Merge tag 'kvm-4.15-1' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Pull KVM updates from Radim Krčmář:
"First batch of KVM changes for 4.15
Common:
- Python 3 support in kvm_stat
- Accounting of slabs to kmemcg
ARM:
- Optimized arch timer handling for KVM/ARM
- Improvements to the VGIC ITS code and introduction of an ITS reset
ioctl
- Unification of the 32-bit fault injection logic
- More exact external abort matching logic
PPC:
- Support for running hashed page table (HPT) MMU mode on a host that
is using the radix MMU mode; single threaded mode on POWER 9 is
added as a pre-requisite
- Resolution of merge conflicts with the last second 4.14 HPT fixes
- Fixes and cleanups
s390:
- Some initial preparation patches for exitless interrupts and crypto
- New capability for AIS migration
- Fixes
x86:
- Improved emulation of LAPIC timer mode changes, MCi_STATUS MSRs,
and after-reset state
- Refined dependencies for VMX features
- Fixes for nested SMI injection
- A lot of cleanups"
* tag 'kvm-4.15-1' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (89 commits)
KVM: s390: provide a capability for AIS state migration
KVM: s390: clear_io_irq() requests are not expected for adapter interrupts
KVM: s390: abstract conversion between isc and enum irq_types
KVM: s390: vsie: use common code functions for pinning
KVM: s390: SIE considerations for AP Queue virtualization
KVM: s390: document memory ordering for kvm_s390_vcpu_wakeup
KVM: PPC: Book3S HV: Cosmetic post-merge cleanups
KVM: arm/arm64: fix the incompatible matching for external abort
KVM: arm/arm64: Unify 32bit fault injection
KVM: arm/arm64: vgic-its: Implement KVM_DEV_ARM_ITS_CTRL_RESET
KVM: arm/arm64: Document KVM_DEV_ARM_ITS_CTRL_RESET
KVM: arm/arm64: vgic-its: Free caches when GITS_BASER Valid bit is cleared
KVM: arm/arm64: vgic-its: New helper functions to free the caches
KVM: arm/arm64: vgic-its: Remove kvm_its_unmap_device
arm/arm64: KVM: Load the timer state when enabling the timer
KVM: arm/arm64: Rework kvm_timer_should_fire
KVM: arm/arm64: Get rid of kvm_timer_flush_hwstate
KVM: arm/arm64: Avoid phys timer emulation in vcpu entry/exit
KVM: arm/arm64: Move phys_timer_emulate function
KVM: arm/arm64: Use kvm_arm_timer_set/get_reg for guest register traps
...
In kvm_apic_set_state() we update the hardware virtualized APIC after
the full APIC state has been overwritten. Do the same, when the full
APIC state has been reset in kvm_lapic_reset().
This updates some hardware state that was previously forgotten, as
far as I can tell. Also, this allows removing some APIC-related reset
code from vmx_vcpu_reset().
Signed-off-by: Jan H. Schönherr <jschoenh@amazon.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Parts of the posted interrupt descriptor configure host behavior,
such as the notification vector and destination. Overwriting them
with zero as done during vCPU reset breaks posted interrupts.
KVM (re-)writes these fields on certain occasions and belatedly fixes
the situation in many cases. However, if you have a guest configured
with "idle=poll", for example, the fields might stay zero forever.
Do not reset the full descriptor in vmx_vcpu_reset(). Instead,
reset only the outstanding notifications and leave everything
else untouched.
Signed-off-by: Jan H. Schönherr <jschoenh@amazon.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In my setup, EPT is not exposed to L1, the VPID capability is exposed and
can be observed by vmxcap tool in L1:
INVVPID supported yes
Individual-address INVVPID yes
Single-context INVVPID yes
All-context INVVPID yes
Single-context-retaining-globals INVVPID yes
However, the module parameter of VPID observed in L1 is always N, the
cpu_has_vmx_invvpid() check in L1 KVM fails since vmx_capability.vpid
is 0 and it is not read from MSR due to EPT is not exposed.
The VPID can be used to tag linear mappings when EPT is not enabled. However,
current logic just detects VPID capability if EPT is enabled, this patch
fixes it.
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Jim Mattson <jmattson@google.com>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
I can use vmxcap tool to observe "EPTP Switching yes" even if EPT is not
exposed to L1.
EPT switching is advertised unconditionally since it is emulated, however,
it can be treated as an extended feature for EPT and it should not be
advertised if EPT itself is not exposed. This patch fixes it.
Reviewed-by: David Hildenbrand <david@redhat.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Jim Mattson <jmattson@google.com>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Commit 05cade71cf ("KVM: nSVM: fix SMI injection in guest mode") made
KVM mask SMI if GIF=0 but it didn't do anything to unmask it when GIF is
enabled.
The issue manifests for me as a significantly longer boot time of Windows
guests when running with SMM-enabled OVMF.
This commit fixes it by intercepting STGI instead of requesting immediate
exit if the reason why SMM was masked is GIF.
Fixes: 05cade71cf ("KVM: nSVM: fix SMI injection in guest mode")
Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Entering SMM while running in guest mode wasn't working very well because several
pieces of the vcpu state were left set up for nested operation.
Some of the issues observed:
* L1 was getting unexpected VM exits (using L1 interception controls but running
in SMM execution environment)
* SMM handler couldn't write to vmx_set_cr4 because of incorrect validity checks
predicated on nested.vmxon
* MMU was confused (walk_mmu was still set to nested_mmu)
Intel SDM actually prescribes the logical processor to "leave VMX operation" upon
entering SMM in 34.14.1 Default Treatment of SMI Delivery. What we need to do is
basically get out of guest mode and set nested.vmxon to false for the duration of
SMM. All this completely transparent to L1, i.e. L1 is not given control and no
L1 observable state changes.
To avoid code duplication this commit takes advantage of the existing nested
vmexit and run functionality, perhaps at the cost of efficiency. To get out of
guest mode, nested_vmx_vmexit with exit_reason == -1 is called, a trick already
used in vmx_leave_nested. Re-entering is cleaner, using enter_vmx_non_root_mode.
This commit fixes running Windows Server 2016 with Hyper-V enabled in a VM with
OVMF firmware (OVMF_CODE-need-smm.fd).
Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Intel SDM 27.5.2 Loading Host Segment and Descriptor-Table Registers:
"The GDTR and IDTR limits are each set to FFFFH."
Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Similar to NMI, there may be ISA specific reasons why an SMI cannot be
injected into the guest. This commit adds a new smi_allowed callback to
be implemented in following commits.
Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Entering and exiting SMM may require ISA specific handling under certain
circumstances. This commit adds two new callbacks with empty implementations.
Actual functionality will be added in following commits.
* pre_enter_smm() is to be called when injecting an SMM, before any
SMM related vcpu state has been changed
* pre_leave_smm() is to be called when emulating the RSM instruction,
when the vcpu is in real mode and before any SMM related vcpu state
has been restored
Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
It has always annoyed me a bit how SVM_EXIT_NPF is handled by
pf_interception. This is also the only reason behind the
under-documented need_unprotect argument to kvm_handle_page_fault.
Let NPF go straight to kvm_mmu_page_fault, just like VMX
does in handle_ept_violation and handle_ept_misconfig.
Reviewed-by: Brijesh Singh <brijesh.singh@amd.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
SDM mentioned:
"If either the “unrestricted guest†VM-execution control or the “mode-based
execute control for EPT†VM- execution control is 1, the “enable EPTâ€
VM-execution control must also be 1."
However, we can still observe unrestricted_guest is Y after inserting the kvm-intel.ko
w/ ept=N. It depends on later starts a guest in order that the function
vmx_compute_secondary_exec_control() can be executed, then both the module parameter
and exec control fields will be amended.
This patch fixes it by amending module parameter immediately during vmcs data setup.
Reviewed-by: Jim Mattson <jmattson@google.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Jim Mattson <jmattson@google.com>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
- XCR0 is reset to 1 by RESET but not INIT
- XSS is zeroed by both RESET and INIT
- BNDCFGU, BND0-BND3, BNDCFGS, BNDSTATUS are zeroed by both RESET and INIT
This patch does this according to SDM.
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Cc: Jim Mattson <jmattson@google.com>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Let's just name these according to the SDM. This should make it clearer
that the are used to enable exiting and not the feature itself.
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
No need for another enable_ept check. kvm->arch.ept_identity_map_addr
only has to be inititalized once. Having alloc_identity_pagetable() is
overkill and dropping BUG_ONs is always nice.
Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
They are inititally 0, so no need to reset them to 0.
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
vcpu->cpu is not cleared when doing a vmx_vcpu_put/load, so this can be
dropped.
Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Without this, we won't be able to do any flushes, so let's just require
it. Should be absent in very strange configurations.
Suggested-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
ept_* function should only be called with enable_ept being set.
Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
This function is only called with enable_ept.
Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
Make it a void and drop error handling code.
Reviewed-by: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
KVM doesn't expose the PLE capability to the L1 hypervisor, however,
ple_window still shows the default value on L1 hypervisor. This patch
fixes it by clearing all the PLE related module parameter if there is
no PLE capability.
Reviewed-by: Konrad Rzeszutek Wilk <konrad.wilk@oracle.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
When KVM emulates an exit from L2 to L1, it loads L1 CR4 into the
guest CR4. Before this CR4 loading, the guest CR4 refers to L2
CR4. Because these two CR4's are in different levels of guest, we
should vmx_set_cr4() rather than kvm_set_cr4() here. The latter, which
is used to handle guest writes to its CR4, checks the guest change to
CR4 and may fail if the change is invalid.
The failure may cause trouble. Consider we start
a L1 guest with non-zero L1 PCID in use,
(i.e. L1 CR4.PCIDE == 1 && L1 CR3.PCID != 0)
and
a L2 guest with L2 PCID disabled,
(i.e. L2 CR4.PCIDE == 0)
and following events may happen:
1. If kvm_set_cr4() is used in load_vmcs12_host_state() to load L1 CR4
into guest CR4 (in VMCS01) for L2 to L1 exit, it will fail because
of PCID check. As a result, the guest CR4 recorded in L0 KVM (i.e.
vcpu->arch.cr4) is left to the value of L2 CR4.
2. Later, if L1 attempts to change its CR4, e.g., clearing VMXE bit,
kvm_set_cr4() in L0 KVM will think L1 also wants to enable PCID,
because the wrong L2 CR4 is used by L0 KVM as L1 CR4. As L1
CR3.PCID != 0, L0 KVM will inject GP to L1 guest.
Fixes: 4704d0befb ("KVM: nVMX: Exiting from L2 to L1")
Cc: qemu-stable@nongnu.org
Signed-off-by: Haozhong Zhang <haozhong.zhang@intel.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
that was finally triggered by PCID support.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2.0.22 (GNU/Linux)
iQEcBAABAgAGBQJZzmJqAAoJEL/70l94x66D99oH/R4hOMfzDxFOgW3LnaCQJvwo
n1+tH3as0dfdkpggZ+UmJuKnbVJ0625+qozenrdYkKtk1YiyIIQWG3vdsz4HBfzp
CYK2NVVymf0dg8DQaluz6iB1R28ke12PggzyFv01s1QyENBDA8J38pslZarPM2OA
tnpRKC6B59/VmRD0PWS6yRmTXY+HfzWlWg4JMraq2VdybbEXJhh8BNfjjNn30DkZ
SW8kHq60AUd5Arhb3cmiPiXZCQ7odqF2u2mEcBmnA9hAacaGEheSzKCUOaEIjmZV
5/jTyG1tZkN7CbrG81ryuoa8A6qTOSyHxo1QkzAmE/p+s2IzGfzzLqmtfIsAWkE=
=1lM1
-----END PGP SIGNATURE-----
Merge tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Pull kvm fixes from Paolo Bonzini:
"Mixed bugfixes. Perhaps the most interesting one is a latent bug that
was finally triggered by PCID support"
* tag 'for-linus' of git://git.kernel.org/pub/scm/virt/kvm/kvm:
kvm/x86: Handle async PF in RCU read-side critical sections
KVM: nVMX: Fix nested #PF intends to break L1's vmlauch/vmresume
KVM: VMX: use cmpxchg64
KVM: VMX: simplify and fix vmx_vcpu_pi_load
KVM: VMX: avoid double list add with VT-d posted interrupts
KVM: VMX: extract __pi_post_block
KVM: PPC: Book3S HV: Check for updated HDSISR on P9 HDSI exception
KVM: nVMX: fix HOST_CR3/HOST_CR4 cache
------------[ cut here ]------------
WARNING: CPU: 4 PID: 5280 at /home/kernel/linux/arch/x86/kvm//vmx.c:11394 nested_vmx_vmexit+0xc2b/0xd70 [kvm_intel]
CPU: 4 PID: 5280 Comm: qemu-system-x86 Tainted: G W OE 4.13.0+ #17
RIP: 0010:nested_vmx_vmexit+0xc2b/0xd70 [kvm_intel]
Call Trace:
? emulator_read_emulated+0x15/0x20 [kvm]
? segmented_read+0xae/0xf0 [kvm]
vmx_inject_page_fault_nested+0x60/0x70 [kvm_intel]
? vmx_inject_page_fault_nested+0x60/0x70 [kvm_intel]
x86_emulate_instruction+0x733/0x810 [kvm]
vmx_handle_exit+0x2f4/0xda0 [kvm_intel]
? kvm_arch_vcpu_ioctl_run+0xd2f/0x1c60 [kvm]
kvm_arch_vcpu_ioctl_run+0xdab/0x1c60 [kvm]
? kvm_arch_vcpu_load+0x62/0x230 [kvm]
kvm_vcpu_ioctl+0x340/0x700 [kvm]
? kvm_vcpu_ioctl+0x340/0x700 [kvm]
? __fget+0xfc/0x210
do_vfs_ioctl+0xa4/0x6a0
? __fget+0x11d/0x210
SyS_ioctl+0x79/0x90
entry_SYSCALL_64_fastpath+0x23/0xc2
A nested #PF is triggered during L0 emulating instruction for L2. However, it
doesn't consider we should not break L1's vmlauch/vmresme. This patch fixes
it by queuing the #PF exception instead ,requesting an immediate VM exit from
L2 and keeping the exception for L1 pending for a subsequent nested VM exit.
This should actually work all the time, making vmx_inject_page_fault_nested
totally unnecessary. However, that's not working yet, so this patch can work
around the issue in the meanwhile.
Cc: Paolo Bonzini <pbonzini@redhat.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Signed-off-by: Wanpeng Li <wanpeng.li@hotmail.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The simplify part: do not touch pi_desc.nv, we can set it when the
VCPU is first created. Likewise, pi_desc.sn is only handled by
vmx_vcpu_pi_load, do not touch it in __pi_post_block.
The fix part: do not check kvm_arch_has_assigned_device, instead
check the SN bit to figure out whether vmx_vcpu_pi_put ran before.
This matches what the previous patch did in pi_post_block.
Cc: Huangweidong <weidong.huang@huawei.com>
Cc: Gonglei <arei.gonglei@huawei.com>
Cc: wangxin <wangxinxin.wang@huawei.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Tested-by: Longpeng (Mike) <longpeng2@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
In some cases, for example involving hot-unplug of assigned
devices, pi_post_block can forget to remove the vCPU from the
blocked_vcpu_list. When this happens, the next call to
pi_pre_block corrupts the list.
Fix this in two ways. First, check vcpu->pre_pcpu in pi_pre_block
and WARN instead of adding the element twice in the list. Second,
always do the list removal in pi_post_block if vcpu->pre_pcpu is
set (not -1).
The new code keeps interrupts disabled for the whole duration of
pi_pre_block/pi_post_block. This is not strictly necessary, but
easier to follow. For the same reason, PI.ON is checked only
after the cmpxchg, and to handle it we just call the post-block
code. This removes duplication of the list removal code.
Cc: Huangweidong <weidong.huang@huawei.com>
Cc: Gonglei <arei.gonglei@huawei.com>
Cc: wangxin <wangxinxin.wang@huawei.com>
Cc: Radim Krčmář <rkrcmar@redhat.com>
Tested-by: Longpeng (Mike) <longpeng2@huawei.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Pull x86 fixes from Ingo Molnar:
"Another round of CR3/PCID related fixes (I think this addresses all
but one of the known problems with PCID support), an objtool fix plus
a Clang fix that (finally) solves all Clang quirks to build a bootable
x86 kernel as-is"
* 'x86-urgent-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/tip:
x86/asm: Fix inline asm call constraints for Clang
objtool: Handle another GCC stack pointer adjustment bug
x86/mm/32: Load a sane CR3 before cpu_init() on secondary CPUs
x86/mm/32: Move setup_clear_cpu_cap(X86_FEATURE_PCID) earlier
x86/mm/64: Stop using CR3.PCID == 0 in ASID-aware code
x86/mm: Factor out CR3-building code
For inline asm statements which have a CALL instruction, we list the
stack pointer as a constraint to convince GCC to ensure the frame
pointer is set up first:
static inline void foo()
{
register void *__sp asm(_ASM_SP);
asm("call bar" : "+r" (__sp))
}
Unfortunately, that pattern causes Clang to corrupt the stack pointer.
The fix is easy: convert the stack pointer register variable to a global
variable.
It should be noted that the end result is different based on the GCC
version. With GCC 6.4, this patch has exactly the same result as
before:
defconfig defconfig-nofp distro distro-nofp
before 9820389 9491555 8816046 8516940
after 9820389 9491555 8816046 8516940
With GCC 7.2, however, GCC's behavior has changed. It now changes its
behavior based on the conversion of the register variable to a global.
That somehow convinces it to *always* set up the frame pointer before
inserting *any* inline asm. (Therefore, listing the variable as an
output constraint is a no-op and is no longer necessary.) It's a bit
overkill, but the performance impact should be negligible. And in fact,
there's a nice improvement with frame pointers disabled:
defconfig defconfig-nofp distro distro-nofp
before 9796316 9468236 9076191 8790305
after 9796957 9464267 9076381 8785949
So in summary, while listing the stack pointer as an output constraint
is no longer necessary for newer versions of GCC, it's still needed for
older versions.
Suggested-by: Andrey Ryabinin <aryabinin@virtuozzo.com>
Reported-by: Matthias Kaehlcke <mka@chromium.org>
Signed-off-by: Josh Poimboeuf <jpoimboe@redhat.com>
Cc: Alexander Potapenko <glider@google.com>
Cc: Andy Lutomirski <luto@kernel.org>
Cc: Arnd Bergmann <arnd@arndb.de>
Cc: Dmitriy Vyukov <dvyukov@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Miguel Bernal Marin <miguel.bernal.marin@linux.intel.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Link: http://lkml.kernel.org/r/3db862e970c432ae823cf515c52b54fec8270e0e.1505942196.git.jpoimboe@redhat.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
For nested virt we maintain multiple VMCS that can run on a vCPU. So it is
incorrect to keep vmcs_host_cr3 and vmcs_host_cr4, whose purpose is caching
the value of the rarely changing HOST_CR3 and HOST_CR4 VMCS fields, in
vCPU-wide data structures.
Hyper-V nested on KVM runs into this consistently for me with PCID enabled.
CR3 is updated with a new value, unlikely(cr3 != vmx->host_state.vmcs_host_cr3)
fires, and the currently loaded VMCS is updated. Then we switch from L2 to
L1 and the next exit reverts CR3 to its old value.
Fixes: d6e41f1151 ("x86/mm, KVM: Teach KVM's VMX code that CR3 isn't a constant")
Signed-off-by: Ladi Prosek <lprosek@redhat.com>
Cc: stable@vger.kernel.org
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
WARN_ON_ONCE(pi_test_sn(&vmx->pi_desc)) in kvm_vcpu_trigger_posted_interrupt()
intends to detect the violation of invariant that VT-d PI notification
event is not suppressed when vcpu is in the guest mode. Because the
two checks for the target vcpu mode and the target suppress field
cannot be performed atomically, the target vcpu mode may change in
between. If that does happen, WARN_ON_ONCE() here may raise false
alarms.
As the previous patch fixed the real invariant breaker, remove this
WARN_ON_ONCE() to avoid false alarms, and document the allowed cases
instead.
Signed-off-by: Haozhong Zhang <haozhong.zhang@intel.com>
Reported-by: "Ramamurthy, Venkatesh" <venkatesh.ramamurthy@intel.com>
Reported-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Fixes: 28b835d60f ("KVM: Update Posted-Interrupts Descriptor when vCPU is preempted")
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
In kvm_vcpu_trigger_posted_interrupt() and pi_pre_block(), KVM
assumes that PI notification events should not be suppressed when the
target vCPU is not blocked.
vmx_update_pi_irte() sets the SN field before changing an interrupt
from posting to remapping, but it does not check the vCPU mode.
Therefore, the change of SN field may break above the assumption.
Besides, I don't see reasons to suppress notification events here, so
remove the changes of SN field to avoid race condition.
Signed-off-by: Haozhong Zhang <haozhong.zhang@intel.com>
Reported-by: "Ramamurthy, Venkatesh" <venkatesh.ramamurthy@intel.com>
Reported-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Paolo Bonzini <pbonzini@redhat.com>
Fixes: 28b835d60f ("KVM: Update Posted-Interrupts Descriptor when vCPU is preempted")
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
When emulating a nested VM-entry from L1 to L2, several control field
validation checks are deferred to the hardware. Should one of these
validation checks fail, vcpu_vmx_run will set the vmx->fail flag. When
this happens, the L2 guest state is not loaded (even in part), and
execution should continue in L1 with the next instruction after the
VMLAUNCH/VMRESUME.
The VMCS12 is not modified (except for the VM-instruction error
field), the VMCS12 MSR save/load lists are not processed, and the CPU
state is not loaded from the VMCS12 host area. Moreover, the vmcs02
exit reason is stale, so it should not be consulted for any reason.
Signed-off-by: Jim Mattson <jmattson@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
On an early VMLAUNCH/VMRESUME failure (i.e. one which sets the
VM-instruction error field of the current VMCS), the launch state of
the current VMCS is not set to "launched," and the VM-exit information
fields of the current VMCS (including IDT-vectoring information and
exit reason) are stale.
On a late VMLAUNCH/VMRESUME failure (i.e. one which sets the high bit
of the exit reason field), the launch state of the current VMCS is not
set to "launched," and only two of the VM-exit information fields of
the current VMCS are modified (exit reason and exit
qualification). The remaining VM-exit information fields of the
current VMCS (including IDT-vectoring information, in particular) are
stale.
Signed-off-by: Jim Mattson <jmattson@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
After a successful VM-entry, RFLAGS is cleared, with the exception of
bit 1, which is always set. This is handled by load_vmcs12_host_state.
Signed-off-by: Jim Mattson <jmattson@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
The value of the guest_irq argument to vmx_update_pi_irte() is
ultimately coming from a KVM_IRQFD API call. Do not BUG() in
vmx_update_pi_irte() if the value is out-of bounds. (Especially,
since KVM as a whole seems to hang after that.)
Instead, print a message only once if we find that we don't have a
route for a certain IRQ (which can be out-of-bounds or within the
array).
This fixes CVE-2017-1000252.
Fixes: efc644048e ("KVM: x86: Update IRTE for posted-interrupts")
Signed-off-by: Jan H. Schönherr <jschoenh@amazon.de>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
If L1 does not specify the "use TPR shadow" VM-execution control in
vmcs12, then L0 must specify the "CR8-load exiting" and "CR8-store
exiting" VM-execution controls in vmcs02. Failure to do so will give
the L2 VM unrestricted read/write access to the hardware CR8.
This fixes CVE-2017-12154.
Signed-off-by: Jim Mattson <jmattson@google.com>
Reviewed-by: David Hildenbrand <david@redhat.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
Modify struct kvm_x86_ops.arch.apicv_active() to take struct kvm_vcpu
pointer as parameter in preparation to subsequent changes.
Signed-off-by: Suravee Suthikulpanit <suravee.suthikulpanit@amd.com>
Signed-off-by: Radim Krčmář <rkrcmar@redhat.com>
