{"api_version":"1","generated_at":"2026-11-22T20:40:00+00:00","cve":"CVE-2026-99798","urls":{"html":"https://cve.report/CVE-2026-99798","api":"https://cve.report/api/cve/CVE-2026-99798.json","docs":"https://cve.report/api","cve_org":"https://www.cve.org/CVERecord?id=CVE-2026-99798","nvd":"https://nvd.nist.gov/vuln/detail/CVE-2026-99798"},"summary":{"title":"io_uring: defer foreign io-wq enqueues to task_work","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nio_uring: defer foreign io-wq enqueues to task_work\n\nAn IOPOLL ring without DEFER_TASKRUN can be reaped by a task other than\nthe one that submitted the I/O. If a polled request is marked\nREQ_F_REISSUE, io_free_batch_list() calls io_queue_iowq() with the\noriginal tctx. Ordinary task exit can tear down tctx->io_wq after\ndraining only the tracked requests. The live request pins the task, but\nnot its io-wq, so the foreign task can race io_wq_put_and_exit() and\nenqueue through a freed or NULL queue.\n\nForward foreign userspace enqueues to the owning task before accessing\ntctx->io_wq. Use a dedicated task_work callback so reissues still go\nstraight to io-wq, and fail them if task_work is running in cancellation\nmode. Preserve the existing PF_KTHREAD cancellation path. This also\ncovers the other callers of io_queue_iowq() without changing the\nworker-owned enqueue path.","state":"PUBLISHED","assigner":"Linux","published_at":"2026-11-20 08:18:37","updated_at":"2026-11-22 11:17:10"},"problem_types":[],"metrics":[{"version":"3.1","source":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","type":"Secondary","score":"7.8","severity":"HIGH","vector":"CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","data":{"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","attackVector":"LOCAL","attackComplexity":"LOW","privilegesRequired":"LOW","userInteraction":"NONE","scope":"UNCHANGED","confidentialityImpact":"HIGH","integrityImpact":"HIGH","availabilityImpact":"HIGH"}},{"version":"3.1","source":"CNA","type":"DECLARED","score":"7.8","severity":"HIGH","vector":"CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","data":{"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"}}],"references":[{"url":"https://git.kernel.org/stable/c/0ac8f069a8bd7a507df59c505182065ae720e946","name":"https://git.kernel.org/stable/c/0ac8f069a8bd7a507df59c505182065ae720e946","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://git.kernel.org/stable/c/5f7e7afd5418c63473f6ac3a05219e9e6e8a59d9","name":"https://git.kernel.org/stable/c/5f7e7afd5418c63473f6ac3a05219e9e6e8a59d9","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://git.kernel.org/stable/c/3de16279b43dc285ed2222aba06cebdc82fdb28b","name":"https://git.kernel.org/stable/c/3de16279b43dc285ed2222aba06cebdc82fdb28b","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://www.cve.org/CVERecord?id=CVE-2026-99798","name":"CVE Program record","refsource":"CVE.ORG","tags":["canonical"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-99798","name":"NVD vulnerability detail","refsource":"NVD","tags":["canonical","analysis"]}],"affected":[{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected d803d123948feffbd992213e144df224097f82b0 0ac8f069a8bd7a507df59c505182065ae720e946 git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected d803d123948feffbd992213e144df224097f82b0 5f7e7afd5418c63473f6ac3a05219e9e6e8a59d9 git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected d803d123948feffbd992213e144df224097f82b0 3de16279b43dc285ed2222aba06cebdc82fdb28b git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected 6.14","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 6.14 semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 6.18.61 6.18.* semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 7.3.2 7.3.* semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 7.4-rc1 * original_commit_for_fix","platforms":[]}],"timeline":[],"solutions":[],"workarounds":[],"exploits":[],"credits":[],"nvd_cpes":[],"vendor_comments":[],"enrichments":{"kev":null,"epss":{"cve_year":"2026","cve_id":"99798","cve":"CVE-2026-99798","epss":"0.003030000","percentile":"0.234640000","score_date":"2026-11-21","updated_at":"2026-11-22 00:14:17"},"legacy_qids":[]},"source_records":{"cve_program":{"containers":{"cna":{"affected":[{"defaultStatus":"unaffected","product":"Linux","programFiles":["io_uring/io_uring.c"],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","vendor":"Linux","versions":[{"lessThan":"0ac8f069a8bd7a507df59c505182065ae720e946","status":"affected","version":"d803d123948feffbd992213e144df224097f82b0","versionType":"git"},{"lessThan":"5f7e7afd5418c63473f6ac3a05219e9e6e8a59d9","status":"affected","version":"d803d123948feffbd992213e144df224097f82b0","versionType":"git"},{"lessThan":"3de16279b43dc285ed2222aba06cebdc82fdb28b","status":"affected","version":"d803d123948feffbd992213e144df224097f82b0","versionType":"git"}]},{"defaultStatus":"affected","product":"Linux","programFiles":["io_uring/io_uring.c"],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","vendor":"Linux","versions":[{"status":"affected","version":"6.14"},{"lessThan":"6.14","status":"unaffected","version":"0","versionType":"semver"},{"lessThanOrEqual":"6.18.*","status":"unaffected","version":"6.18.61","versionType":"semver"},{"lessThanOrEqual":"7.3.*","status":"unaffected","version":"7.3.2","versionType":"semver"},{"lessThanOrEqual":"*","status":"unaffected","version":"7.4-rc1","versionType":"original_commit_for_fix"}]}],"cpeApplicability":[{"nodes":[{"cpeMatch":[{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"6.18.61","versionStartIncluding":"6.14","vulnerable":true},{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"7.3.2","versionStartIncluding":"6.14","vulnerable":true},{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"7.4-rc1","versionStartIncluding":"6.14","vulnerable":true}],"negate":false,"operator":"OR"}]}],"descriptions":[{"lang":"en","value":"In the Linux kernel, the following vulnerability has been resolved:\n\nio_uring: defer foreign io-wq enqueues to task_work\n\nAn IOPOLL ring without DEFER_TASKRUN can be reaped by a task other than\nthe one that submitted the I/O. If a polled request is marked\nREQ_F_REISSUE, io_free_batch_list() calls io_queue_iowq() with the\noriginal tctx. Ordinary task exit can tear down tctx->io_wq after\ndraining only the tracked requests. The live request pins the task, but\nnot its io-wq, so the foreign task can race io_wq_put_and_exit() and\nenqueue through a freed or NULL queue.\n\nForward foreign userspace enqueues to the owning task before accessing\ntctx->io_wq. Use a dedicated task_work callback so reissues still go\nstraight to io-wq, and fail them if task_work is running in cancellation\nmode. Preserve the existing PF_KTHREAD cancellation path. This also\ncovers the other callers of io_queue_iowq() without changing the\nworker-owned enqueue path."}],"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 - A local caller of io_uring_enter() reaps an IORING_SETUP_IOPOLL ring whose setup flags leave IORING_SETUP_DEFER_TASKRUN clear, which io_allowed_run_tw() permits for a task other than the submitter. The reap runs io_iopoll_check(), io_do_iopoll(), __io_submit_flush_completions(), and io_free_batch_list(), and that last function calls io_queue_iowq() when REQ_F_REISSUE is set.\nAC:L - The attacker owns both tasks and can fill the poll-queue tags. A REQ_NOWAIT bio then fails in blk_mq_get_new_requests(), bio_wouldblock_error() completes it with BLK_STS_AGAIN, and io_complete_rw_iopoll() sets REQ_F_REISSUE. do_exit() calls io_uring_files_cancel(); io_uring_cancel_generic() returns at inflight_tracked zero, so io_uring_clean_tctx() frees io_wq while the other task runs io_free_batch_list().\nPR:L - io_uring_setup() and io_uring_enter() are open to an unprivileged uid: io_uring_allowed() returns success when sysctl kernel.io_uring_disabled is 0, the compiled default. The second reaper is another thread of that same process and needs no extra capability to call io_uring_enter() on the shared ring fd.\nUI:N - The attacker opens an O_DIRECT file they can already access, creates the IOPOLL ring, submits the read or write, and reaps it from a second thread of the same process while the submitter calls exit. Every step is the attacker's own syscall.\nS:U - The object io_wq_enqueue() uses after io_wq_destroy() is the submitter's struct io_wq in the same kernel that hosts the attacker, so the resulting privilege change stays inside that kernel.\nC:H - io_uring_clean_tctx() kfree()s tctx->io_wq in io_wq_destroy() while the live request only pins the task. io_wq_enqueue() reads wq->state and the acct from the freed slot, and io_wq_insert_work() writes kernel list pointers into it, so a reclaim of struct io_wq is a kernel memory disclosure primitive.\nI:H - On that same freed struct io_wq, io_wq_enqueue() raw_spin_lock()s acct->lock and io_wq_insert_work() writes acct->work_list, and a clear IO_WQ_BIT_EXIT leads to io_wq_create_worker(). Those are heap writes and a worker-creation primitive from attacker-reclaimed memory.\nA:H - A load of tctx->io_wq after the NULL store makes io_wq_enqueue() dereference wq->state and oops. A load in the window after kfree() and before that store locks and links the freed struct io_wq. Both outcomes crash the kernel."}]}],"providerMetadata":{"dateUpdated":"2026-11-22T11:10:11.755Z","orgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","shortName":"Linux"},"references":[{"url":"https://git.kernel.org/stable/c/0ac8f069a8bd7a507df59c505182065ae720e946"},{"url":"https://git.kernel.org/stable/c/5f7e7afd5418c63473f6ac3a05219e9e6e8a59d9"},{"url":"https://git.kernel.org/stable/c/3de16279b43dc285ed2222aba06cebdc82fdb28b"}],"title":"io_uring: defer foreign io-wq enqueues to task_work","x_generator":{"engine":"bippy-1.2.0"}}},"cveMetadata":{"assignerOrgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","assignerShortName":"Linux","cveId":"CVE-2026-99798","datePublished":"2026-11-20T07:02:21.634Z","dateReserved":"2026-09-25T10:25:14.435Z","dateUpdated":"2026-11-22T11:10:11.755Z","state":"PUBLISHED"},"dataType":"CVE_RECORD","dataVersion":"5.2"},"nvd":{"publishedDate":"2026-11-20 08:18:37","lastModifiedDate":"2026-11-22 11:17:10","problem_types":[],"metrics":{"cvssMetricV31":[{"source":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","type":"Secondary","cvssData":{"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","attackVector":"LOCAL","attackComplexity":"LOW","privilegesRequired":"LOW","userInteraction":"NONE","scope":"UNCHANGED","confidentialityImpact":"HIGH","integrityImpact":"HIGH","availabilityImpact":"HIGH"},"exploitabilityScore":1.8,"impactScore":5.9}]},"configurations":[]},"legacy_mitre":{"record":{"CveYear":"2026","CveId":"99798","Ordinal":"1","Title":"io_uring: defer foreign io-wq enqueues to task_work","CVE":"CVE-2026-99798","Year":"2026"},"notes":[{"CveYear":"2026","CveId":"99798","Ordinal":"1","NoteData":"In the Linux kernel, the following vulnerability has been resolved:\n\nio_uring: defer foreign io-wq enqueues to task_work\n\nAn IOPOLL ring without DEFER_TASKRUN can be reaped by a task other than\nthe one that submitted the I/O. If a polled request is marked\nREQ_F_REISSUE, io_free_batch_list() calls io_queue_iowq() with the\noriginal tctx. Ordinary task exit can tear down tctx->io_wq after\ndraining only the tracked requests. The live request pins the task, but\nnot its io-wq, so the foreign task can race io_wq_put_and_exit() and\nenqueue through a freed or NULL queue.\n\nForward foreign userspace enqueues to the owning task before accessing\ntctx->io_wq. Use a dedicated task_work callback so reissues still go\nstraight to io-wq, and fail them if task_work is running in cancellation\nmode. Preserve the existing PF_KTHREAD cancellation path. This also\ncovers the other callers of io_queue_iowq() without changing the\nworker-owned enqueue path.","Type":"Description","Title":"io_uring: defer foreign io-wq enqueues to task_work"}]}}}