CVE-2024-56552 (GCVE-0-2024-56552)
Vulnerability from cvelistv5
Published
2024-12-27 14:22
Modified
2026-08-05 11:45
Severity ?
VLAI Severity ?
EPSS score ?
Summary
In the Linux kernel, the following vulnerability has been resolved:
drm/xe/guc_submit: fix race around suspend_pending
Currently in some testcases we can trigger:
xe 0000:03:00.0: [drm] Assertion `exec_queue_destroyed(q)` failed!
....
WARNING: CPU: 18 PID: 2640 at drivers/gpu/drm/xe/xe_guc_submit.c:1826 xe_guc_sched_done_handler+0xa54/0xef0 [xe]
xe 0000:03:00.0: [drm] *ERROR* GT1: DEREGISTER_DONE: Unexpected engine state 0x00a1, guc_id=57
Looking at a snippet of corresponding ftrace for this GuC id we can see:
162.673311: xe_sched_msg_add: dev=0000:03:00.0, gt=1 guc_id=57, opcode=3
162.673317: xe_sched_msg_recv: dev=0000:03:00.0, gt=1 guc_id=57, opcode=3
162.673319: xe_exec_queue_scheduling_disable: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0x29, flags=0x0
162.674089: xe_exec_queue_kill: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0x29, flags=0x0
162.674108: xe_exec_queue_close: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0xa9, flags=0x0
162.674488: xe_exec_queue_scheduling_done: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0xa9, flags=0x0
162.678452: xe_exec_queue_deregister: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0xa1, flags=0x0
It looks like we try to suspend the queue (opcode=3), setting
suspend_pending and triggering a disable_scheduling. The user then
closes the queue. However the close will also forcefully signal the
suspend fence after killing the queue, later when the G2H response for
disable_scheduling comes back we have now cleared suspend_pending when
signalling the suspend fence, so the disable_scheduling now incorrectly
tries to also deregister the queue. This leads to warnings since the queue
has yet to even be marked for destruction. We also seem to trigger
errors later with trying to double unregister the same queue.
To fix this tweak the ordering when handling the response to ensure we
don't race with a disable_scheduling that didn't actually intend to
perform an unregister. The destruction path should now also correctly
wait for any pending_disable before marking as destroyed.
(cherry picked from commit f161809b362f027b6d72bd998e47f8f0bad60a2e)
References
Impacted products
{
"containers": {
"cna": {
"affected": [
{
"defaultStatus": "unaffected",
"product": "Linux",
"programFiles": [
"drivers/gpu/drm/xe/xe_guc_submit.c"
],
"repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
"vendor": "Linux",
"versions": [
{
"lessThan": "5ddcb50b700221fa7d7be2adcb3d7d7afe8633dd",
"status": "affected",
"version": "dd08ebf6c3525a7ea2186e636df064ea47281987",
"versionType": "git"
},
{
"lessThan": "87651f31ae4e6e6e7e6c7270b9b469405e747407",
"status": "affected",
"version": "dd08ebf6c3525a7ea2186e636df064ea47281987",
"versionType": "git"
}
]
},
{
"defaultStatus": "affected",
"product": "Linux",
"programFiles": [
"drivers/gpu/drm/xe/xe_guc_submit.c"
],
"repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git",
"vendor": "Linux",
"versions": [
{
"status": "affected",
"version": "6.8"
},
{
"lessThan": "6.8",
"status": "unaffected",
"version": "0",
"versionType": "semver"
},
{
"lessThanOrEqual": "6.12.*",
"status": "unaffected",
"version": "6.12.4",
"versionType": "semver"
},
{
"lessThanOrEqual": "*",
"status": "unaffected",
"version": "6.13",
"versionType": "original_commit_for_fix"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
"versionEndExcluding": "6.12.4",
"versionStartIncluding": "6.8",
"vulnerable": true
},
{
"criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*",
"versionEndExcluding": "6.13",
"versionStartIncluding": "6.8",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"descriptions": [
{
"lang": "en",
"value": "In the Linux kernel, the following vulnerability has been resolved:\n\ndrm/xe/guc_submit: fix race around suspend_pending\n\nCurrently in some testcases we can trigger:\n\nxe 0000:03:00.0: [drm] Assertion `exec_queue_destroyed(q)` failed!\n....\nWARNING: CPU: 18 PID: 2640 at drivers/gpu/drm/xe/xe_guc_submit.c:1826 xe_guc_sched_done_handler+0xa54/0xef0 [xe]\nxe 0000:03:00.0: [drm] *ERROR* GT1: DEREGISTER_DONE: Unexpected engine state 0x00a1, guc_id=57\n\nLooking at a snippet of corresponding ftrace for this GuC id we can see:\n\n162.673311: xe_sched_msg_add: dev=0000:03:00.0, gt=1 guc_id=57, opcode=3\n162.673317: xe_sched_msg_recv: dev=0000:03:00.0, gt=1 guc_id=57, opcode=3\n162.673319: xe_exec_queue_scheduling_disable: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0x29, flags=0x0\n162.674089: xe_exec_queue_kill: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0x29, flags=0x0\n162.674108: xe_exec_queue_close: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0xa9, flags=0x0\n162.674488: xe_exec_queue_scheduling_done: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0xa9, flags=0x0\n162.678452: xe_exec_queue_deregister: dev=0000:03:00.0, 1:0x2, gt=1, width=1, guc_id=57, guc_state=0xa1, flags=0x0\n\nIt looks like we try to suspend the queue (opcode=3), setting\nsuspend_pending and triggering a disable_scheduling. The user then\ncloses the queue. However the close will also forcefully signal the\nsuspend fence after killing the queue, later when the G2H response for\ndisable_scheduling comes back we have now cleared suspend_pending when\nsignalling the suspend fence, so the disable_scheduling now incorrectly\ntries to also deregister the queue. This leads to warnings since the queue\nhas yet to even be marked for destruction. We also seem to trigger\nerrors later with trying to double unregister the same queue.\n\nTo fix this tweak the ordering when handling the response to ensure we\ndon\u0027t race with a disable_scheduling that didn\u0027t actually intend to\nperform an unregister. The destruction path should now also correctly\nwait for any pending_disable before marking as destroyed.\n\n(cherry picked from commit f161809b362f027b6d72bd998e47f8f0bad60a2e)"
}
],
"metrics": [
{
"cvssV3_1": {
"baseScore": 7.8,
"baseSeverity": "HIGH",
"vectorString": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
"version": "3.1"
},
"scenarios": [
{
"lang": "en",
"value": "AV:L - The race is driven entirely through Intel Xe DRM ioctls (`XE_VM_BIND`/`XE_EXEC` triggering preempt-fence suspend, and `XE_EXEC_QUEUE_DESTROY`) on the local `/dev/dri/renderD*` node. No network or adjacent-network path reaches `xe_guc_submit.c`.\nAC:L - The attacker owns both sides of the race \u2014 it issues the operation that sets `suspend_pending` and fires `disable_scheduling()`, then closes the queue from another thread inside the GuC round-trip window, which the reported ftrace shows is ~1.2 ms wide. It reproduces in ordinary IGT test runs and can be looped across many exec queues concurrently until it lands.\nPR:L - Every ioctl on the path (`XE_VM_CREATE`, `XE_EXEC_QUEUE_CREATE`, `XE_VM_BIND`, `XE_EXEC`, `XE_EXEC_QUEUE_DESTROY`) is flagged `DRM_RENDER_ALLOW`, so any unprivileged process holding the render node \u2014 every GPU-using app on a desktop, Android, or ChromeOS system, including sandboxed browser GPU processes and containers with GPU passthrough \u2014 can reach it. No capability or DRM_MASTER is required.\nUI:N - A single unprivileged process performs the suspend-inducing operation and the queue destroy from its own threads. No victim action or cooperating user is involved.\nS:U - The corrupted state, freed objects, and resulting GT reset are all within the kernel\u0027s own security authority on the host. No VM, IOMMU, or hypervisor boundary is crossed.\nC:H - The out-of-protocol deregister lets `handle_deregister_done()` run against a queue whose lifetime the driver still owns, producing double `kfree(ge)`/`xe_exec_queue_fini(q)` and \u2014 once `release_guc_id()` recycles the id \u2014 a stale `DEREGISTER_DONE` landing on a different live queue. That use-after-free over attacker-shaped heap objects yields a read primitive into freed kernel memory.\nI:H - The same double-deregister drives `__guc_exec_queue_fini()` twice, re-running `INIT_WORK` on an already-queued work item and double-freeing `xe_guc_exec_queue`/`xe_exec_queue`, giving a write primitive into reclaimed slab memory that is exploitable for control-flow hijack via heap spraying.\nA:H - Confirmed WARN/assert splat plus permanent driver-versus-GuC state desync: the rejected `DEREGISTER_DONE` (`-EPROTO`) leaks the exec queue, its guc_id and its G2H credit forever, the follow-up `DISABLE` to an already-deregistered context times out into `xe_gt_reset_async()` killing all GPU contexts on the tile, and `show_run_ticks()`\u0027s untimed `wait_var_event()` on `pending_removal` hangs fdinfo readers in unkillable uninterruptible sleep."
}
]
}
],
"providerMetadata": {
"dateUpdated": "2026-08-05T11:45:00.626Z",
"orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"shortName": "Linux"
},
"references": [
{
"url": "https://git.kernel.org/stable/c/5ddcb50b700221fa7d7be2adcb3d7d7afe8633dd"
},
{
"url": "https://git.kernel.org/stable/c/87651f31ae4e6e6e7e6c7270b9b469405e747407"
}
],
"title": "drm/xe/guc_submit: fix race around suspend_pending",
"x_generator": {
"engine": "bippy-1.2.0"
}
}
},
"cveMetadata": {
"assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67",
"assignerShortName": "Linux",
"cveId": "CVE-2024-56552",
"datePublished": "2024-12-27T14:22:54.140Z",
"dateReserved": "2024-12-27T14:03:05.990Z",
"dateUpdated": "2026-08-05T11:45:00.626Z",
"state": "PUBLISHED"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2"
}
Loading…
Loading…
Sightings
| Author | Source | Type | Date |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or seen somewhere by the user.
- Confirmed: The vulnerability is confirmed from an analyst perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: This vulnerability was exploited and seen by the user reporting the sighting.
- Patched: This vulnerability was successfully patched by the user reporting the sighting.
- Not exploited: This vulnerability was not exploited or seen by the user reporting the sighting.
- Not confirmed: The user expresses doubt about the veracity of the vulnerability.
- Not patched: This vulnerability was not successfully patched by the user reporting the sighting.
Loading…
Loading…