Re: [PATCH 25/31] KVM: x86/mmu: Take memory protection attributes into account during faults

[sashiko-bot@kernel.org]

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [Medium] kvm: trace buffer spam on AMD with GMET enabled
--

commit e48ad04487bab345ee62e42a2457d15e6edd7180
Author: Nicolas Saenz Julienne <nsaenz@amazon.com>
Subject: KVM: x86/mmu: Take memory protection attributes into account during faults

This commit adds support for memory protection attributes during guest page
faults, injecting a user-space -EFAULT exit for prohibited accesses. It also
adds a trace event that is intended to trigger only when memory protection
attributes restrict access.

> diff --git a/arch/x86/kvm/mmu/mmu.c b/arch/x86/kvm/mmu/mmu.c
> index 3c926f95f87be..6bbecbae2f8c4 100644
> --- a/arch/x86/kvm/mmu/mmu.c
> +++ b/arch/x86/kvm/mmu/mmu.c
[ ... ]
> +static inline unsigned kvm_get_gfn_protections(struct kvm_vcpu *vcpu, gfn_t gfn)
> +{
> +	struct kvm *kvm = vcpu->kvm;
> +	unsigned int access = vcpu->arch.mmu->root_role.access;
> +	unsigned long attrs = kvm_get_memory_attributes(kvm, gfn);
> +	if (!attrs)
> +		return access;
> +
[ ... ]
> +static int kvm_faultin_memory_protections(struct kvm_vcpu *vcpu,
> +					  struct kvm_page_fault *fault)
> +{
> +	unsigned access;
> +
> +	/* Memory attributes don't apply to MMIO regions */
> +	if (unlikely(!fault->slot))
> +		return RET_PF_CONTINUE;
> +
> +	access = kvm_get_gfn_protections(vcpu, fault->gfn);
> +	if (access == ACC_ALL)
> +		return RET_PF_CONTINUE;
> +
> +	trace_kvm_faultin_memory_protections(vcpu, fault, access);

[Severity: Medium]
Will this check effectively prevent trace buffer spam on AMD processors
with Guest Mode Execute Trap (GMET) enabled?

When GMET is enabled, KVM sets the root_role.access for the TDP MMU to
ACC_ALL & ~ACC_USER_MASK. Since kvm_get_gfn_protections() initializes the
access mask using vcpu->arch.mmu->root_role.access, it will return the
restricted mask even if no memory attributes are present.

If access is restricted by GMET, the condition checking if access == ACC_ALL
evaluates to false, which causes the trace event to trigger unconditionally
on every page fault. Could this lead to noticeable performance regressions
during tracing on these systems?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260918135030.171564-2-pbonzini@redhat.com?part=25