mirrors/linux-stable
1
0
Fork 0
mirror of https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git synced 2024-10-02 07:04:24 +00:00
Code Issues Releases Wiki Activity

KVM: SVM: Don't apply SEV+SMAP workaround on code fetch or PT access

Browse source 
Resume the guest instead of synthesizing a triple fault shutdown if the
instruction bytes buffer is empty due to the #NPF being on the code fetch
itself or on a page table access.  The SMAP errata applies if and only if
the code fetch was successful and ucode's subsequent data read from the
code page encountered a SMAP violation.  In practice, the guest is likely
hosed either way, but crashing the guest on a code fetch to emulated MMIO
is technically wrong according to the behavior described in the APM.

Signed-off-by: Sean Christopherson <seanjc@google.com>
Reviewed-by: Liam Merwick <liam.merwick@oracle.com>
Message-Id: <20220120010719.711476-9-seanjc@google.com>
Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>
This commit is contained in:
Sean Christopherson 2022-01-20 01:07:18 +00:00 committed by Paolo Bonzini
parent 04c40f344d
commit 3280cc22ae
1 changed files with 34 additions and 9 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

43
arch/x86/kvm/svm/svm.c
View file

@ -4263,6 +4263,7 @@ static bool svm_can_emulate_instruction(struct kvm_vcpu *vcpu, int emul_type,
{
bool smep, smap, is_user;
unsigned long cr4;
u64 error_code;
/* Emulation is always possible when KVM has access to all guest state. */
if (!sev_guest(vcpu->kvm))
@ -4328,22 +4329,31 @@ static bool svm_can_emulate_instruction(struct kvm_vcpu *vcpu, int emul_type,
* loap uop with CPL=0 privileges. If the load hits a SMAP #PF, ucode
* gives up and does not fill the instruction bytes buffer.
*
* Detection:
* KVM reaches this point if the VM is an SEV guest, the CPU supports
* DecodeAssist, a #NPF was raised, KVM's page fault handler triggered
* emulation (e.g. for MMIO), and the CPU returned 0 in GuestIntrBytes
* field of the VMCB.
* As above, KVM reaches this point iff the VM is an SEV guest, the CPU
* supports DecodeAssist, a #NPF was raised, KVM's page fault handler
* triggered emulation (e.g. for MMIO), and the CPU returned 0 in the
* GuestIntrBytes field of the VMCB.
*
* This does _not_ mean that the erratum has been encountered, as the
* DecodeAssist will also fail if the load for CS:RIP hits a legitimate
* #PF, e.g. if the guest attempt to execute from emulated MMIO and
* encountered a reserved/not-present #PF.
*
* To reduce the likelihood of false positives, take action if and only
* if CR4.SMAP=1 (obviously required to hit the erratum) and CR4.SMEP=0
* or CPL=3. If SMEP=1 and CPL!=3, the erratum cannot have been hit as
* the guest would have encountered a SMEP violation #PF, not a #NPF.
* To hit the erratum, the following conditions must be true:
* 1. CR4.SMAP=1 (obviously).
* 2. CR4.SMEP=0 || CPL=3. If SMEP=1 and CPL<3, the erratum cannot
* have been hit as the guest would have encountered a SMEP
* violation #PF, not a #NPF.
* 3. The #NPF is not due to a code fetch, in which case failure to
* retrieve the instruction bytes is legitimate (see abvoe).
*
* In addition, don't apply the erratum workaround if the #NPF occurred
* while translating guest page tables (see below).
*/
error_code = to_svm(vcpu)->vmcb->control.exit_info_1;
if (error_code & (PFERR_GUEST_PAGE_MASK | PFERR_FETCH_MASK))
goto resume_guest;
cr4 = kvm_read_cr4(vcpu);
smep = cr4 & X86_CR4_SMEP;
smap = cr4 & X86_CR4_SMAP;
@ -4353,6 +4363,21 @@ static bool svm_can_emulate_instruction(struct kvm_vcpu *vcpu, int emul_type,
kvm_make_request(KVM_REQ_TRIPLE_FAULT, vcpu);
}
resume_guest:
/*
* If the erratum was not hit, simply resume the guest and let it fault
* again. While awful, e.g. the vCPU may get stuck in an infinite loop
* if the fault is at CPL=0, it's the lesser of all evils. Exiting to
* userspace will kill the guest, and letting the emulator read garbage
* will yield random behavior and potentially corrupt the guest.
*
* Simply resuming the guest is technically not a violation of the SEV
* architecture. AMD's APM states that all code fetches and page table
* accesses for SEV guest are encrypted, regardless of the C-Bit. The
* APM also states that encrypted accesses to MMIO are "ignored", but
* doesn't explicitly define "ignored", i.e. doing nothing and letting
* the guest spin is technically "ignoring" the access.
*/
return false;
}