[ Upstream commit beb8d93b3e423043e079ef3dda19dad7b28467a8 ]
A previous fix to prevent KVM from consuming stale VMCS state after a failed VM-Entry inadvertantly blocked KVM's handling of machine checks that occur during VM-Entry.
Per Intel's SDM, a #MC during VM-Entry is handled in one of three ways, depending on when the #MC is recognoized. As it pertains to this bug fix, the third case explicitly states EXIT_REASON_MCE_DURING_VMENTRY is handled like any other VM-Exit during VM-Entry, i.e. sets bit 31 to indicate the VM-Entry failed.
If a machine-check event occurs during a VM entry, one of the following occurs: - The machine-check event is handled as if it occurred before the VM entry: ... - The machine-check event is handled after VM entry completes: ... - A VM-entry failure occurs as described in Section 26.7. The basic exit reason is 41, for "VM-entry failure due to machine-check event".
Explicitly handle EXIT_REASON_MCE_DURING_VMENTRY as a one-off case in vmx_vcpu_run() instead of binning it into vmx_complete_atomic_exit(). Doing so allows vmx_vcpu_run() to handle VMX_EXIT_REASONS_FAILED_VMENTRY in a sane fashion and also simplifies vmx_complete_atomic_exit() since VMCS.VM_EXIT_INTR_INFO is guaranteed to be fresh.
Fixes: b060ca3b2e9e7 ("kvm: vmx: Handle VMLAUNCH/VMRESUME failure properly") Cc: stable@vger.kernel.org Signed-off-by: Sean Christopherson sean.j.christopherson@intel.com Reviewed-by: Jim Mattson jmattson@google.com Signed-off-by: Paolo Bonzini pbonzini@redhat.com Signed-off-by: Sasha Levin sashal@kernel.org --- arch/x86/kvm/vmx.c | 20 ++++++++------------ 1 file changed, 8 insertions(+), 12 deletions(-)
diff --git a/arch/x86/kvm/vmx.c b/arch/x86/kvm/vmx.c index e4bba840a0708..82253d31842a2 100644 --- a/arch/x86/kvm/vmx.c +++ b/arch/x86/kvm/vmx.c @@ -10438,28 +10438,21 @@ static void vmx_apicv_post_state_restore(struct kvm_vcpu *vcpu)
static void vmx_complete_atomic_exit(struct vcpu_vmx *vmx) { - u32 exit_intr_info = 0; - u16 basic_exit_reason = (u16)vmx->exit_reason; - - if (!(basic_exit_reason == EXIT_REASON_MCE_DURING_VMENTRY - || basic_exit_reason == EXIT_REASON_EXCEPTION_NMI)) + if (vmx->exit_reason != EXIT_REASON_EXCEPTION_NMI) return;
- if (!(vmx->exit_reason & VMX_EXIT_REASONS_FAILED_VMENTRY)) - exit_intr_info = vmcs_read32(VM_EXIT_INTR_INFO); - vmx->exit_intr_info = exit_intr_info; + vmx->exit_intr_info = vmcs_read32(VM_EXIT_INTR_INFO);
/* if exit due to PF check for async PF */ - if (is_page_fault(exit_intr_info)) + if (is_page_fault(vmx->exit_intr_info)) vmx->vcpu.arch.apf.host_apf_reason = kvm_read_and_reset_pf_reason();
/* Handle machine checks before interrupts are enabled */ - if (basic_exit_reason == EXIT_REASON_MCE_DURING_VMENTRY || - is_machine_check(exit_intr_info)) + if (is_machine_check(vmx->exit_intr_info)) kvm_machine_check();
/* We need to handle NMIs before interrupts are enabled */ - if (is_nmi(exit_intr_info)) { + if (is_nmi(vmx->exit_intr_info)) { kvm_before_interrupt(&vmx->vcpu); asm("int $2"); kvm_after_interrupt(&vmx->vcpu); @@ -10980,6 +10973,9 @@ static void __noclone vmx_vcpu_run(struct kvm_vcpu *vcpu) vmx->idt_vectoring_info = 0;
vmx->exit_reason = vmx->fail ? 0xdead : vmcs_read32(VM_EXIT_REASON); + if ((u16)vmx->exit_reason == EXIT_REASON_MCE_DURING_VMENTRY) + kvm_machine_check(); + if (vmx->fail || (vmx->exit_reason & VMX_EXIT_REASONS_FAILED_VMENTRY)) return;