{ "dataType": "CVE_RECORD", "dataVersion": "5.2", "cveMetadata": { "cveId": "CVE-2025-68793", "assignerOrgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67", "state": "PUBLISHED", "assignerShortName": "Linux", "dateReserved": "2025-12-24T10:30:51.037Z", "datePublished": "2026-01-13T15:29:04.877Z", "dateUpdated": "2026-08-05T12:11:29.613Z" }, "containers": { "cna": { "providerMetadata": { "orgId": "416baaa9-dc9f-4396-8d5f-8c081fb06d67", "shortName": "Linux", "dateUpdated": "2026-08-05T12:11:29.613Z" }, "descriptions": [ { "lang": "en", "value": "In the Linux kernel, the following vulnerability has been resolved:\n\ndrm/amdgpu: fix a job->pasid access race in gpu recovery\n\nAvoid a possible UAF in GPU recovery due to a race between\nthe sched timeout callback and the tdr work queue.\n\nThe gpu recovery function calls drm_sched_stop() and\nlater drm_sched_start(). drm_sched_start() restarts\nthe tdr queue which will eventually free the job. If\nthe tdr queue frees the job before time out callback\ncompletes, the job will be freed and we'll get a UAF\nwhen accessing the pasid. Cache it early to avoid the\nUAF.\n\nExample KASAN trace:\n[ 493.058141] BUG: KASAN: slab-use-after-free in amdgpu_device_gpu_recover+0x968/0x990 [amdgpu]\n[ 493.067530] Read of size 4 at addr ffff88b0ce3f794c by task kworker/u128:1/323\n[ 493.074892]\n[ 493.076485] CPU: 9 UID: 0 PID: 323 Comm: kworker/u128:1 Tainted: G E 6.16.0-1289896.2.zuul.bf4f11df81c1410bbe901c4373305a31 #1 PREEMPT(voluntary)\n[ 493.076493] Tainted: [E]=UNSIGNED_MODULE\n[ 493.076495] Hardware name: TYAN B8021G88V2HR-2T/S8021GM2NR-2T, BIOS V1.03.B10 04/01/2019\n[ 493.076500] Workqueue: amdgpu-reset-dev drm_sched_job_timedout [gpu_sched]\n[ 493.076512] Call Trace:\n[ 493.076515] \n[ 493.076518] dump_stack_lvl+0x64/0x80\n[ 493.076529] print_report+0xce/0x630\n[ 493.076536] ? _raw_spin_lock_irqsave+0x86/0xd0\n[ 493.076541] ? __pfx__raw_spin_lock_irqsave+0x10/0x10\n[ 493.076545] ? amdgpu_device_gpu_recover+0x968/0x990 [amdgpu]\n[ 493.077253] kasan_report+0xb8/0xf0\n[ 493.077258] ? amdgpu_device_gpu_recover+0x968/0x990 [amdgpu]\n[ 493.077965] amdgpu_device_gpu_recover+0x968/0x990 [amdgpu]\n[ 493.078672] ? __pfx_amdgpu_device_gpu_recover+0x10/0x10 [amdgpu]\n[ 493.079378] ? amdgpu_coredump+0x1fd/0x4c0 [amdgpu]\n[ 493.080111] amdgpu_job_timedout+0x642/0x1400 [amdgpu]\n[ 493.080903] ? pick_task_fair+0x24e/0x330\n[ 493.080910] ? __pfx_amdgpu_job_timedout+0x10/0x10 [amdgpu]\n[ 493.081702] ? _raw_spin_lock+0x75/0xc0\n[ 493.081708] ? __pfx__raw_spin_lock+0x10/0x10\n[ 493.081712] drm_sched_job_timedout+0x1b0/0x4b0 [gpu_sched]\n[ 493.081721] ? __pfx__raw_spin_lock_irq+0x10/0x10\n[ 493.081725] process_one_work+0x679/0xff0\n[ 493.081732] worker_thread+0x6ce/0xfd0\n[ 493.081736] ? __pfx_worker_thread+0x10/0x10\n[ 493.081739] kthread+0x376/0x730\n[ 493.081744] ? __pfx_kthread+0x10/0x10\n[ 493.081748] ? __pfx__raw_spin_lock_irq+0x10/0x10\n[ 493.081751] ? __pfx_kthread+0x10/0x10\n[ 493.081755] ret_from_fork+0x247/0x330\n[ 493.081761] ? __pfx_kthread+0x10/0x10\n[ 493.081764] ret_from_fork_asm+0x1a/0x30\n[ 493.081771] \n\n(cherry picked from commit 20880a3fd5dd7bca1a079534cf6596bda92e107d)" } ], "metrics": [ { "cvssV3_1": { "version": "3.1", "vectorString": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H", "baseScore": 7.8, "baseSeverity": "HIGH" }, "scenarios": [ { "lang": "en", "value": "AV:L - The vulnerability is reached through local AMDGPU command submission via a DRM render or primary device node; no network or physical protocol reaches this path.\nAC:L - The attacker controls job submission and can deliberately hang the GPU while maintaining concurrent submissions or heap spraying; the race is repeatable and requires no victim-controlled condition.\nPR:L - DRM_IOCTL_AMDGPU_CS is marked DRM_RENDER_ALLOW, and normal contexts require no capability; only ordinary unprivileged access to the AMDGPU render node is needed.\nUI:N - After the attacker submits the hanging job, the scheduler timeout, recovery, object freeing, and stale access occur automatically without another user's participation.\nS:U - The vulnerable DRM client and the affected host kernel remain under the same security authority; this is standard local kernel compromise rather than a VM or IOMMU boundary escape.\nC:H - This is a kernel slab use-after-free involving an attacker-created, variable-sized amdgpu_job allocation that can be reclaimed through heap spraying, making high-impact kernel memory disclosure defensible.\nI:H - Reclaiming the freed job with attacker-influenced allocations can support UAF exploitation and potential kernel memory manipulation or control-flow hijacking, resulting in complete integrity loss.\nA:H - The UAF was observed by KASAN in GPU recovery and can cause an oops or panic; repeated malicious hangs and recovery failures can also deny the GPU to all system users." } ] } ], "affected": [ { "product": "Linux", "vendor": "Linux", "defaultStatus": "unaffected", "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git", "programFiles": [ "drivers/gpu/drm/amd/amdgpu/amdgpu_device.c" ], "versions": [ { "version": "a72002cb181f350734108228b24c5d10d358f95a", "lessThan": "dac58c012c47cadf337a35eb05d44498c43e5cd0", "status": "affected", "versionType": "git" }, { "version": "a72002cb181f350734108228b24c5d10d358f95a", "lessThan": "77f73253015cbc7893fca1821ac3eae9eb4bc943", "status": "affected", "versionType": "git" } ] }, { "product": "Linux", "vendor": "Linux", "defaultStatus": "affected", "repo": "https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git", "programFiles": [ "drivers/gpu/drm/amd/amdgpu/amdgpu_device.c" ], "versions": [ { "version": "6.17", "status": "affected" }, { "version": "0", "lessThan": "6.17", "status": "unaffected", "versionType": "semver" }, { "version": "6.18.3", "lessThanOrEqual": "6.18.*", "status": "unaffected", "versionType": "semver" }, { "version": "6.19", "lessThanOrEqual": "*", "status": "unaffected", "versionType": "original_commit_for_fix" } ] } ], "cpeApplicability": [ { "nodes": [ { "operator": "OR", "negate": false, "cpeMatch": [ { "vulnerable": true, "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", "versionStartIncluding": "6.17", "versionEndExcluding": "6.18.3" }, { "vulnerable": true, "criteria": "cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*", "versionStartIncluding": "6.17", "versionEndExcluding": "6.19" } ] } ] } ], "references": [ { "url": "https://git.kernel.org/stable/c/dac58c012c47cadf337a35eb05d44498c43e5cd0" }, { "url": "https://git.kernel.org/stable/c/77f73253015cbc7893fca1821ac3eae9eb4bc943" } ], "title": "drm/amdgpu: fix a job->pasid access race in gpu recovery", "x_generator": { "engine": "bippy-1.2.0" } } } }