Kernel KVM virtualization development
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Paolo Bonzini" <pbonzini@redhat.com>, kvm@vger.kernel.org
Subject: Re: [PATCH] KVM: SEV: never retry after #NPF on paging structures
Date: Tue, 17 Nov 2026 19:28:59 +0000	[thread overview]
Message-ID: <sashiko-outbox-219473@kernel.org> (raw)
In-Reply-To: <20261117185922.466255-1-pbonzini@redhat.com>

Thank you for your contribution! Sashiko AI review found 2 potential
issues to consider:
- [Medium] KVM: SVM: unchecked exit_info_1 read in
  svm_check_emulate_instruction()
- [Low] KVM: SEV: commit subject says never retry instead of never
  emulate
It also noticed 1 related pre-existing issue this patch does not
introduce:
- [Medium] KVM: SVM: wrong exit_int_info mask in
  svm_check_emulate_instruction() [1]
--

commit 79cd3998685bf7eaec1673e862ee32458c3ae3ee
Author: Paolo Bonzini <pbonzini@redhat.com>

KVM: SEV: never retry after #NPF on paging structures

This commit moves the PFERR_GUEST_PAGE_MASK check in
svm_check_emulate_instruction() ahead of the insn_len check so that KVM
resumes the SEV guest instead of attempting instruction emulation after a
#NPF on guest page tables.

> KVM: SEV: never retry after #NPF on paging structures

[Severity: Low]
Should the commit subject say "never emulate" (or "always retry") instead
of "never retry"?  Jumping to resume_guest returns X86EMUL_RETRY_INSTR to
resume the guest and retry the instruction rather than emulating it.

> diff --git a/arch/x86/kvm/svm/svm.c b/arch/x86/kvm/svm/svm.c
> index fe0cba4731b2e..65b15999cf9d8 100644
> --- a/arch/x86/kvm/svm/svm.c
> +++ b/arch/x86/kvm/svm/svm.c
> @@ -5192,19 +5192,29 @@ static int svm_check_emulate_instruction(struct kvm_vcpu *vcpu, int emul_type,
>  	}
>  
>  	/*
> -	 * Emulate for SEV guests if the insn buffer is not empty.  The buffer
> -	 * will be empty if the DecodeAssist microcode cannot fetch bytes for
> -	 * the faulting instruction because the code fetch itself faulted, e.g.
> -	 * the guest attempted to fetch from emulated MMIO or a guest page
> -	 * table used to translate CS:RIP resides in emulated MMIO.
> +	 * Under SEV, page tables are encrypted and cannot be read by the
> +	 * hypervisor, so there is no way to resolve a fault on page tables.
>  	 */
> +	error_code = svm->vmcb->control.exit_info_1;
> +	if (error_code & PFERR_GUEST_PAGE_MASK)
> +		goto resume_guest;
> +
>  	if (likely(insn_len))
>  		return X86EMUL_CONTINUE;

[Severity: Medium]
Can moving the exit_info_1 check ahead of the insn_len check without
checking emul_type & EMULTYPE_PF cause non-page-fault emulations to consume
a stale exit_info_1 value?

When an SEV guest executes RSM, rsm_interception() calls
kvm_emulate_instruction_from_buffer(vcpu, rsm_ins_bytes, 2) with
emul_type == 0 and insn_len == 2:

arch/x86/kvm/svm/svm.c:rsm_interception() {
	return kvm_emulate_instruction_from_buffer(vcpu, rsm_ins_bytes, 2);
}

Previously, that path returned X86EMUL_CONTINUE at if (likely(insn_len))
before reading exit_info_1.  Because SVM_EXIT_RSM does not define
EXITINFO1, a stale PFERR_GUEST_PAGE_MASK bit left in
vmcb->control.exit_info_1 from an earlier #NPF causes
svm_check_emulate_instruction() to return X86EMUL_RETRY_INSTR and loop on
RSM instead of emulating it or injecting #UD.

>  	/*
> -	 * Detect and workaround Errata 1096 Fam_17h_00_0Fh.
> +	 * The insn buffer was empty; the DecodeAssist microcode could not fetch
> +	 * bytes for the faulting instruction because the code fetch itself faulted,
> +	 * e.g. the guest attempted to fetch from emulated MMIO or a guest page
> +	 * table used to translate CS:RIP resides in emulated MMIO.
>  	 *
> -	 * Errata:
> +	 * A data fault instead could be Errata 1096 Fam_17h_00_0Fh, so only
> +	 * retry for a fetch fault and otherwise try to workaround it.
> +	 */
> +	if (error_code & PFERR_FETCH_MASK)
> +		goto resume_guest;
> +
> +	/*
>  	 * When CPU raises #NPF on guest data access and vCPU CR4.SMAP=1, it is
>  	 * possible that CPU microcode implementing DecodeAssist will fail to
>  	 * read guest memory at CS:RIP and vmcb.GuestIntrBytes will incorrectly
> @@ -5231,12 +5241,8 @@ static int svm_check_emulate_instruction(struct kvm_vcpu *vcpu, int emul_type,
>  	 *       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).
> +	 * while translating guest page tables (see above).
>  	 */
> -	error_code = svm->vmcb->control.exit_info_1;
> -	if (error_code & (PFERR_GUEST_PAGE_MASK | PFERR_FETCH_MASK))
> -		goto resume_guest;
> -
>  	smep = kvm_is_cr4_bit_set(vcpu, X86_CR4_SMEP);
>  	smap = kvm_is_cr4_bit_set(vcpu, X86_CR4_SMAP);
>  	is_user = svm_get_cpl(vcpu) == 3;

[Severity: Medium]
This problem wasn't introduced by this patch, but at the start of
svm_check_emulate_instruction() the event-vectoring check [1] masks
exit_int_info with SVM_EXITINTINFO_TYPE_MASK instead of
SVM_EXITINTINFO_VALID:

arch/x86/kvm/svm/svm.c:svm_check_emulate_instruction() {
	...
	/* Check that emulation is possible during event vectoring */
	if ((svm->vmcb->control.exit_int_info & SVM_EXITINTINFO_TYPE_MASK) &&
	    !kvm_can_emulate_event_vectoring(emul_type))
		return X86EMUL_UNHANDLEABLE_VECTORING;
	...
}

Because SVM_EXITINTINFO_TYPE_INTR is 0, this test misses exits that occur
while vectoring a hardware interrupt, allowing the vCPU to emulate or skip
the interrupted instruction instead of returning
X86EMUL_UNHANDLEABLE_VECTORING.

[1] https://sashiko.dev/#/bug/linux-7ce5444c-332c-4505-82c7-e1440dea4b14

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261117185922.466255-1-pbonzini@redhat.com?part=1

  reply	other threads:[~2026-11-17 19:29 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-17 18:59 [PATCH] KVM: SEV: never retry after #NPF on paging structures Paolo Bonzini
2026-11-17 19:28 ` sashiko-bot [this message]
2026-11-17 20:07 ` Sasha Levin

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=sashiko-outbox-219473@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=kvm@vger.kernel.org \
    --cc=pbonzini@redhat.com \
    --cc=sashiko-reviews@lists.linux.dev \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox