Skip to contentSkip to conversation

KVM: VMX: Handle bad values on proxied writes to LBR MSRs

PATCH Applied Sean Christopherson 2 participants 3 messages May 27 – May 29

Subsystems (3)
lore
Share & tools
lore

Sean Christopherson

author maintainer
#0 lore raw

PATCH v2 1/1 KVM: VMX: Handle bad values on proxied writes to LBR MSRs

From: Xuanqing Shi <[email protected]>

Use the "safe" WRMSR API when writing LBRs on behalf of the guest (or host userspace), and propagate any errors back to the instigator, as the value being written is untrusted. E.g. if the guest (or host userspace) attempts to set reserved bits in LBR_SELECT, then KVM needs to return an error, and not WARN on the bad value.

Continue using the "unsafe" version of RDMSR, as it should be impossible to reach the helper with a completely bogus MSR, i.e. WARNing on RDMSR failure is very desirable, e.g. to make KVM bugs more visible.

unchecked MSR access error: WRMSR to 0x1c8 (tried to write 0x0000000000004000)

  Call Trace:
   intel_pmu_set_msr+0x4e0/0x7f0 [kvm_intel]
   kvm_pmu_set_msr+0x17e/0x1c0 [kvm]
   kvm_set_msr_common+0xc76/0x1440 [kvm]
   vmx_set_msr+0x5e6/0x1570 [kvm_intel]
   kvm_emulate_wrmsr+0x54/0x1d0 [kvm]
   vmx_handle_exit+0x7fc/0x970 [kvm_intel]
Fixes 1b5ac3226a1a ("KVM: vmx/pmu: Pass-through LBR msrs when the guest LBR event is ACTIVE")
[sean: rework changelog, only modify WRMSR path, tag for stable@]

---

v2:
- Rework changelog to better capture the scenario.
- Keep using the "unsafe" version of RDMSR.

v1: https://lore.kernel.org/all/[email protected]

 arch/x86/kvm/vmx/pmu_intel.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
arch/x86/kvm/vmx/pmu_intel.c +4 -2
index 27eb76e6b6a0..22138c487216 100644
@@ -308,13 +308,15 @@ static bool intel_pmu_handle_lbr_msrs_access(struct kvm_vcpu *vcpu,
 	 */
 	local_irq_disable();
 	if (lbr_desc->event->state == PERF_EVENT_STATE_ACTIVE) {
+		int err = 0;
+
 		if (read)
 			rdmsrq(index, msr_info->data);
 		else
-			wrmsrq(index, msr_info->data);
+			err = wrmsrq_safe(index, msr_info->data);
 		__set_bit(INTEL_PMC_IDX_FIXED_VLBR, vcpu_to_pmu(vcpu)->pmc_in_use);
 		local_irq_enable();
-		return true;
+		return !err;
 	}
 	clear_bit(INTEL_PMC_IDX_FIXED_VLBR, vcpu_to_pmu(vcpu)->pmc_in_use);
 	local_irq_enable();

base-commit: 9f2a49c511cb05b85745e1578e4fd425bff87f58

Signature · 1 line
2.54.0.823.g6e5bcc1fc9-goog
#1 lore raw

Thanks for picking this up and reworking the changelog!

原始邮件
发件人:Sean Christopherson <[email protected]>
发件时间:2026年5月27日 10:26
收件人:Sean Christopherson <[email protected]>, Paolo Bonzini <[email protected]>
抄送:kvm <[email protected]>, linux-kernel <[email protected]>, Xuanqing Shi <[email protected]> 主题:[PATCH v2] KVM: VMX: Handle bad values on proxied writes to LBR MSRs

From: Xuanqing Shi <[email protected]>

Use the "safe" WRMSR API when writing LBRs on behalf of the guest (or host userspace), and propagate any errors back to the instigator, as the value being written is untrusted.  E.g. if the guest (or host userspace) attempts to set reserved bits in LBR_SELECT, then KVM needs to return an error, and not WARN on the bad value.

Continue using the "unsafe" version of RDMSR, as it should be impossible to reach the helper with a completely bogus MSR, i.e. WARNing on RDMSR failure is very desirable, e.g. to make KVM bugs more visible.

  unchecked MSR access error: WRMSR to 0x1c8 (tried to write 0x0000000000004000) Call Trace:

   intel_pmu_set_msr+0x4e0/0x7f0 [kvm_intel]
   kvm_pmu_set_msr+0x17e/0x1c0 [kvm]
   kvm_set_msr_common+0xc76/0x1440 [kvm]
   vmx_set_msr+0x5e6/0x1570 [kvm_intel]
   kvm_emulate_wrmsr+0x54/0x1d0 [kvm]
   vmx_handle_exit+0x7fc/0x970 [kvm_intel]
Fixes 1b5ac3226a1a ("KVM: vmx/pmu: Pass-through LBR msrs when the guest LBR event is ACTIVE")
[sean: rework changelog, only modify WRMSR path, tag for stable@]

---

v2:
 - Rework changelog to better capture the scenario. - Keep using the "unsafe" version of RDMSR.

v1: https://lore.kernel.org/all/[email protected]

 arch/x86/kvm/vmx/pmu_intel.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kvm/vmx/pmu_intel.c b/arch/x86/kvm/vmx/pmu_intel.c index 27eb76e6b6a0..22138c487216 100644
--- a/arch/x86/kvm/vmx/pmu_intel.c
+++ b/arch/x86/kvm/vmx/pmu_intel.c
@@ -308,13 +308,15 @@ static bool intel_pmu_handle_lbr_msrs_access(struct kvm_vcpu *vcpu, */
  local_irq_disable();
  if (lbr_desc->event->state == PERF_EVENT_STATE_ACTIVE) {
+ int err = 0;
+
  if (read)
  rdmsrq(index, msr_info->data);
  else
- wrmsrq(index, msr_info->data);
+ err = wrmsrq_safe(index, msr_info->data); __set_bit(INTEL_PMC_IDX_FIXED_VLBR, vcpu_to_pmu(vcpu)->pmc_in_use); local_irq_enable();
- return true;
+ return !err;
  }
  clear_bit(INTEL_PMC_IDX_FIXED_VLBR, vcpu_to_pmu(vcpu)->pmc_in_use); local_irq_enable();

base-commit: 9f2a49c511cb05b85745e1578e4fd425bff87f58 --
2.54.0.823.g6e5bcc1fc9-goog

author maintainer Applied
#2 lore raw
In reply to Sean Christopherson
Quoted 11 lines · Sean Christopherson

Use the "safe" WRMSR API when writing LBRs on behalf of the guest (or host userspace), and propagate any errors back to the instigator, as the value being written is untrusted. E.g. if the guest (or host userspace) attempts to set reserved bits in LBR_SELECT, then KVM needs to return an error, and not WARN on the bad value.

Continue using the "unsafe" version of RDMSR, as it should be impossible to reach the helper with a completely bogus MSR, i.e. WARNing on RDMSR failure is very desirable, e.g. to make KVM bugs more visible.

[...]

Applied to kvm-x86 vmx, thanks!

[1/1] KVM: VMX: Handle bad values on proxied writes to LBR MSRs

      https://github.com/kvm-x86/linux/commit/ca674df13b19

--
https://github.com/kvm-x86/linux/tree/next