mirror of https://lore.kernel.org/lkml/
 help / color / mirror / Atom feed
From: Yosry Ahmed <yosry@kernel.org>
To: Sean Christopherson <seanjc@google.com>
Cc: Paolo Bonzini <pbonzini@redhat.com>,
	Jim Mattson <jmattson@google.com>,
	kvm@vger.kernel.org, linux-kernel@vger.kernel.org,
	Yosry Ahmed <yosry@kernel.org>
Subject: [PATCH 0/4] KVM: x86/mmu: Don't sync cached roots created for a different MMU context
Date: Tue, 13 Oct 2026 23:17:24 +0000	[thread overview]
Message-ID: <20261013231728.883001-1-yosry@kernel.org> (raw)

When invalidating an address, KVM walks and syncs cached roots using the
current MMU context, but cached roots may have been created for a
different context after a nested transition.  With TDP disabled, this
lets a guest trigger a BUG_ON() and crash the host (see patch 3 for
details).

Patch 1 downgrades the BUG_ON() to a WARN_ON_ONCE(), patch 2 is prep
work, patch 3 is the fix, and patch 4 drops a TODO that isn't worth
pursuing.

Sashiko flagged the issue while reviewing the INVLPGA emulation patch in
my nested SVM TLB series [1].  The host crash is easy to reproduce with an
AI-generated repro (with TDP disabled), and no longer triggers with this
series.

[1] https://lore.kernel.org/kvm/sashiko-outbox-163479@kernel.org/

Yosry Ahmed (4):
  KVM: x86/mmu: WARN, don't BUG, when walking a cached root with a PAE
    MMU
  KVM: x86/mmu: Drop for_each_shadow_entry_using_root()
  KVM: x86/mmu: Don't sync cached roots created for a different MMU
    context
  KVM: nVMX: Drop the TODO to sync only the affected SPTEs for INVVPID

 arch/x86/kvm/mmu/mmu.c    | 76 ++++++++++++++++++++++++++++-----------
 arch/x86/kvm/vmx/nested.c |  2 --
 2 files changed, 55 insertions(+), 23 deletions(-)


base-commit: 6bd2905303c58679e303941b7d8ae8c074cd95ce
-- 
2.56.0.385.gd3acb90ef8-goog


             reply	other threads:[~2026-10-13 23:17 UTC|newest]

Thread overview: 9+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-10-13 23:17 Yosry Ahmed [this message]
2026-10-13 23:17 ` [PATCH 1/4] KVM: x86/mmu: WARN, don't BUG, when walking a cached root with a PAE MMU Yosry Ahmed
2026-10-14 12:48   ` Jim Mattson
2026-10-14 14:27     ` Yosry Ahmed
2026-10-14 14:40       ` Jim Mattson
2026-10-13 23:17 ` [PATCH 2/4] KVM: x86/mmu: Drop for_each_shadow_entry_using_root() Yosry Ahmed
2026-10-14 12:54   ` Jim Mattson
2026-10-13 23:17 ` [PATCH 3/4] KVM: x86/mmu: Don't sync cached roots created for a different MMU context Yosry Ahmed
2026-10-13 23:17 ` [PATCH 4/4] KVM: nVMX: Drop the TODO to sync only the affected SPTEs for INVVPID Yosry Ahmed

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=20261013231728.883001-1-yosry@kernel.org \
    --to=yosry@kernel.org \
    --cc=jmattson@google.com \
    --cc=kvm@vger.kernel.org \
    --cc=linux-kernel@vger.kernel.org \
    --cc=pbonzini@redhat.com \
    --cc=seanjc@google.com \
    /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

all inboxes | Powered by JetHome®