Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Matthew Brost" <matthew.brost@intel.com>
Cc: intel-xe@lists.freedesktop.org
Subject: Re: [PATCH v5 09/25] drm/xe: Add struct xe_pt_job_ops
Date: Fri, 04 Sep 2026 02:48:32 +0000	[thread overview]
Message-ID: <20260904024832.DB44B1F00A3F@smtp.kernel.org> (raw)
In-Reply-To: <20260904022207.3490018-10-matthew.brost@intel.com>

Thank you for your contribution! Sashiko AI review found 1 potential issue(s) to consider:

Pre-existing issues:
- [Critical] Ignored return value of xe_pt_walk_shared() in xe_pt_stage_unbind() can cause partial unbinds leading to device memory Use-After-Free.
--

commit becaaccdc6df0bb083fdb6279921f5ee714da885
Author: Matthew Brost <matthew.brost@intel.com>

drm/xe: Add struct xe_pt_job_ops

Add struct xe_pt_job_ops, a dynamically refcounted object that contains
the information required to issue a CPU bind via a job after the initial
bind IOCTL returns.

Link: https://patch.msgid.link/20260228013501.106680-10-matthew.brost@intel.com

> diff --git a/drivers/gpu/drm/xe/xe_pt.c b/drivers/gpu/drm/xe/xe_pt.c
> index e560f167fdf1b..30127ebf1b605 100644
> --- a/drivers/gpu/drm/xe/xe_pt.c
> +++ b/drivers/gpu/drm/xe/xe_pt.c

[ ... ]

> @@ -2081,13 +2094,13 @@ xe_pt_commit_prepare_unbind(struct xe_vma *vma,

[Severity: Critical]
This is a pre-existing issue, but is it safe to ignore the return value of
xe_pt_walk_shared() in xe_pt_stage_unbind() just above this diff?

If memory allocation fails during VMA unbind, the page table walk is
prematurely aborted. The function casts the error to (void) and returns the
count of successfully processed entries:

drivers/gpu/drm/xe/xe_pt.c:xe_pt_stage_unbind() {
    ...
	(void)xe_pt_walk_shared(&pt->base, pt->level, start, end,
				&xe_walk.base);
    ...
}

If the caller proceeds to unbind only that subset of PTEs, does this allow
the physical memory backing the VMA to be freed and reallocated to another
process while the GPU retains a valid PTE mapping to it?

If an unprivileged user triggers this by allocating and unmapping memory
while intentionally applying memory pressure to force an allocation failure,
can this result in a use-after-free and memory corruption?

>  static struct xe_vm_pgtable_update_op *
>  to_pt_op(struct xe_vm_pgtable_update_ops *pt_update_ops, u32 op_idx)
>  {
> -	return &pt_update_ops->ops[op_idx];
> +	return &pt_update_ops->pt_job_ops->ops[op_idx];
>  }

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20260904022207.3490018-1-matthew.brost@intel.com?part=9

  reply	other threads:[~2026-09-04  2:48 UTC|newest]

Thread overview: 41+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-09-04  2:21 [PATCH v5 00/25] CPU binds and ULLS on migration queue Matthew Brost
2026-09-04  2:21 ` [PATCH v5 01/25] drm/xe: Drop struct xe_migrate_pt_update argument from populate/clear vfuns Matthew Brost
2026-09-04  2:21 ` [PATCH v5 02/25] drm/xe: Add xe_migrate_update_pgtables_cpu_execute helper Matthew Brost
2026-09-04  2:37   ` sashiko-bot
2026-09-04  2:21 ` [PATCH v5 03/25] drm/xe: Decouple exec queue idle check from LRC Matthew Brost
2026-09-04  2:21 ` [PATCH v5 04/25] drm/xe: Add job count to GuC exec queue snapshot Matthew Brost
2026-09-04  2:21 ` [PATCH v5 05/25] drm/xe: Update xe_bo_put_deferred arguments to include writeback flag Matthew Brost
2026-09-04  2:44   ` sashiko-bot
2026-09-04  2:21 ` [PATCH v5 06/25] drm/xe: Add XE_BO_FLAG_PUT_VM_ASYNC Matthew Brost
2026-09-04  2:21 ` [PATCH v5 07/25] drm/xe: Update scheduler job layer to support PT jobs Matthew Brost
2026-09-04  2:21 ` [PATCH v5 08/25] drm/xe: Add helpers to access PT ops Matthew Brost
2026-09-04  2:21 ` [PATCH v5 09/25] drm/xe: Add struct xe_pt_job_ops Matthew Brost
2026-09-04  2:48   ` sashiko-bot [this message]
2026-09-04  2:21 ` [PATCH v5 10/25] drm/xe: Update GuC submission backend to run PT jobs Matthew Brost
2026-09-04  2:46   ` sashiko-bot
2026-09-04  2:21 ` [PATCH v5 11/25] drm/xe: Store level in struct xe_vm_pgtable_update Matthew Brost
2026-09-04  2:21 ` [PATCH v5 12/25] drm/xe: Don't use migrate exec queue for page fault binds Matthew Brost
2026-09-04  2:21 ` [PATCH v5 13/25] drm/xe: Enable CPU binds for jobs Matthew Brost
2026-09-04  2:51   ` sashiko-bot
2026-09-04  2:21 ` [PATCH v5 14/25] drm/xe: Remove unused arguments from xe_migrate_pt_update_ops Matthew Brost
2026-09-04  2:21 ` [PATCH v5 15/25] drm/xe: Make bind queues operate cross-tile Matthew Brost
2026-09-04  2:21 ` [PATCH v5 16/25] drm/xe: Add CPU bind layer Matthew Brost
2026-09-04  2:21 ` [PATCH v5 17/25] drm/xe: Add device flag to enable PT mirroring across tiles Matthew Brost
2026-09-04  2:43   ` sashiko-bot
2026-09-04  4:10     ` Matthew Brost
2026-09-04  2:22 ` [PATCH v5 18/25] drm/xe: Add xe_hw_engine_write_ring_tail Matthew Brost
2026-09-04  2:22 ` [PATCH v5 19/25] drm/xe: Add ULLS support to LRC Matthew Brost
2026-09-04  2:22 ` [PATCH v5 20/25] drm/xe: Add ULLS migration job support to migration layer Matthew Brost
2026-09-04  2:44   ` sashiko-bot
2026-09-04  4:11     ` Matthew Brost
2026-09-04  2:22 ` [PATCH v5 21/25] drm/xe: Add ULLS migration job support to ring ops Matthew Brost
2026-09-04  2:22 ` [PATCH v5 22/25] drm/xe: Add ULLS migration job support to GuC submission Matthew Brost
2026-09-04  2:49   ` sashiko-bot
2026-09-04  4:17     ` Matthew Brost
2026-09-04  2:22 ` [PATCH v5 23/25] drm/xe: Enter ULLS for migration jobs upon page fault or SVM prefetch Matthew Brost
2026-09-04  2:22 ` [PATCH v5 24/25] drm/xe: Add modparam to enable / disable ULLS on migrate queue Matthew Brost
2026-09-04  2:22 ` [PATCH v5 25/25] drm/xe: Document ULLS for migration jobs Matthew Brost
2026-09-04  2:29 ` ✗ CI.checkpatch: warning for CPU binds and ULLS on migration queue (rev7) Patchwork
2026-09-04  2:31 ` ✓ CI.KUnit: success " Patchwork
2026-09-04  3:08 ` ✓ Xe.CI.BAT: " Patchwork
2026-09-04 15:21 ` ✗ Xe.CI.FULL: failure " Patchwork

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=20260904024832.DB44B1F00A3F@smtp.kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=matthew.brost@intel.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