Intel-XE Archive on lore.kernel.org
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "Raag Jadav" <raag.jadav@intel.com>, intel-xe@lists.freedesktop.org
Subject: Re: [PATCH v1] drm/xe/drm_ras: Protect against unplugged or wedged device
Date: Mon, 16 Nov 2026 09:07:25 +0000	[thread overview]
Message-ID: <sashiko-outbox-216583@kernel.org> (raw)
In-Reply-To: <20261116085343.1972825-1-raag.jadav@intel.com>

Thank you for your contribution! Sashiko AI review found no new issues
in this patch, but noticed 1 related pre-existing issue it does not
introduce:
- [Critical] drm/ras: use-after-free in drm_ras_nl_list_nodes_dumpit()
  [1]
--

commit f10baa8a8d01dd4d47e42403a4abd3c074961edc
Author: Raag Jadav <raag.jadav@intel.com>

drm/xe/drm_ras: Protect against unplugged or wedged device

This commit adds drm_dev_enter() and xe_wedged() checks around DRM RAS
counter and threshold callbacks and event reporting to prevent accesses on
unplugged or wedged devices.

> diff --git a/drivers/gpu/drm/xe/xe_drm_ras.c b/drivers/gpu/drm/xe/xe_drm_ras.c
> index 7554803aae11a..c0c1acb84e890 100644
> --- a/drivers/gpu/drm/xe/xe_drm_ras.c
> +++ b/drivers/gpu/drm/xe/xe_drm_ras.c
> @@ -5,6 +5,7 @@
>  
>  #include <linux/bitmap.h>
>  
> +#include <drm/drm_drv.h>
>  #include <drm/drm_managed.h>
>  #include <drm/drm_print.h>
>  #include <drm/drm_ras.h>
> @@ -12,10 +13,33 @@
>  #include "xe_device_types.h"
>  #include "xe_drm_ras.h"
>  #include "xe_ras.h"
> +#include "xe_wedge.h"
>  
>  static const char * const error_components[] = DRM_XE_RAS_ERROR_COMPONENT_NAMES;
>  static const char * const error_severity[] = DRM_XE_RAS_ERROR_SEVERITY_NAMES;
>  
> +static int drm_ras_counter_get(struct xe_device *xe, struct xe_drm_ras_counter *info,
> +			       u32 error_id, int *idx)
> +{
> +	if (xe_wedged(xe))
> +		return -ECANCELED;
> +
> +	if (!info || !info[error_id].name)
> +		return -ENOENT;
> +
> +	/* Must be last */
> +	if (!drm_dev_enter(&xe->drm, idx))
> +		return -ENODEV;

[Severity: Critical]
This problem wasn't introduced by this patch, but can
drm_ras_nl_list_nodes_dumpit() race with device teardown and trigger a
use-after-free or NULL pointer dereference on node and node->device_name
[1]?

CPU0                                    CPU1
cleanup_node()                          drm_ras_nl_list_nodes_dumpit()
                                          xa_for_each_start(&drm_ras_xa, ...)
  drm_ras_node_unregister(node)
    xa_erase(&drm_ras_xa, node->id)
  cleanup_node_param(node)
    kfree(node->device_name)
    node->device_name = NULL
  /* node freed via drmm */
                                          nla_put_u32(..., node->id)
                                          nla_put_string(...,
                                            node->device_name)  <-- UAF / NULL

drm_ras_nl_family enables parallel_ops, and xa_for_each_start() only holds
rcu_read_lock() inside xa_find() and xa_find_after(), returning node to the
loop body without any lock, RCU protection, or reference held.  Meanwhile,
cleanup_node() in xe_drm_ras.c unregisters the node with xa_erase(), frees
and clears node->device_name, and lets drmm free the drm_ras_node itself, so
a concurrent netlink dump can dereference NULL in strlen() or read freed
memory.

[1] https://sashiko.dev/#/bug/linux-f54f18de-e95f-4d99-b86c-f90c2c59b668

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/20261116085343.1972825-1-raag.jadav@intel.com?part=1

  reply	other threads:[~2026-11-16  9:07 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-16  8:53 [PATCH v1] drm/xe/drm_ras: Protect against unplugged or wedged device Raag Jadav
2026-11-16  9:07 ` sashiko-bot [this message]
2026-11-16  9:25 ` ✓ CI.KUnit: success for " Patchwork
2026-11-16  9:57 ` ✓ Xe.CI.BAT: " Patchwork
2026-11-16 13:55 ` ✓ Xe.CI.FULL: " 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=sashiko-outbox-216583@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=intel-xe@lists.freedesktop.org \
    --cc=raag.jadav@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