From: sashiko-bot@kernel.org
To: gregkh@linuxfoundation.org
Cc: intel-xe@lists.freedesktop.org
Subject: Re: Patch "drm/xe: Fix shrinker accounting double-subtraction on nested external pins" has been added to the 6.18-stable tree
Date: Mon, 12 Oct 2026 15:53:25 +0000 [thread overview]
Message-ID: <sashiko-outbox-168365@kernel.org> (raw)
In-Reply-To: <2026101225-abacus-errand-7132@gregkh>
Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:
- [High] drm/xe: list corruption and UAF on fb unpin of externally pinned BOs
--
commit 877c19cc3340fbb074cda1d52d8508b429d1bf00
Author: Thomas Hellström <thomas.hellstrom@linux.intel.com>
drm/xe: Fix shrinker accounting double-subtraction on nested external pins
This patch centralizes shrinker accounting into xe_bo_account_pin() and
xe_bo_account_unpin(). It ensures accounting is only updated on the outermost
pin and the final unpin, addressing a double-subtraction issue on nested
external pins.
Link: https://patch.msgid.link/20261002100830.12297-2-thomas.hellstrom@linux.intel.com
> diff --git a/drivers/gpu/drm/xe/display/xe_fb_pin.c b/drivers/gpu/drm/xe/display/xe_fb_pin.c
> index 1fd4a815e784b..7e2b4dbef4d76 100644
> --- a/drivers/gpu/drm/xe/display/xe_fb_pin.c
> +++ b/drivers/gpu/drm/xe/display/xe_fb_pin.c
[ ... ]
> @@ -362,7 +362,7 @@ static void __xe_unpin_fb_vma(struct i915_vma *vma)
> xe_ggtt_node_remove(vma->node, false);
>
> ttm_bo_reserve(&vma->bo->ttm, false, false, NULL);
> - ttm_bo_unpin(&vma->bo->ttm);
> + xe_bo_unpin_account(vma->bo);
> ttm_bo_unreserve(&vma->bo->ttm);
> kfree(vma);
> }
[Severity: High]
Does this code leave a stale pinned_link entry if a BO is pinned both externally
and as a framebuffer?
If a BO has a pin_count of 2 from being pinned both externally and as a
framebuffer, xe_bo_unpin_external() in drivers/gpu/drm/xe/xe_bo.c will reduce
the count to 1 and skip list removal because of its bo->ttm.pin_count == 1
check.
When __xe_unpin_fb_vma() in drivers/gpu/drm/xe/display/xe_fb_pin.c is later
called for the final unpin (dropping pin_count to 0), it only calls
xe_bo_unpin_account(). It lacks the logic to acquire xe->pinned.lock and call
list_del_init(&bo->pinned_link).
Could this result in the BO being freed while its pinned_link remains in
xe->pinned.late.external, leading to list corruption and a Use-After-Free
if the list is iterated later? I note the commit message mentions this was
addressed in a separate follow-up patch, but I am including it here as it
was flagged during review.
--
Sashiko AI review · https://sashiko.dev/#/patchset/2026101225-abacus-errand-7132@gregkh?part=1
prev parent reply other threads:[~2026-10-12 15:53 UTC|newest]
Thread overview: 2+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-10-12 15:38 Patch "drm/xe: Fix shrinker accounting double-subtraction on nested external pins" has been added to the 6.18-stable tree gregkh
2026-10-12 15:53 ` sashiko-bot [this message]
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-168365@kernel.org \
--to=sashiko-bot@kernel.org \
--cc=gregkh@linuxfoundation.org \
--cc=intel-xe@lists.freedesktop.org \
--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