{"api_version":"1","generated_at":"2026-10-01T10:27:32+00:00","cve":"CVE-2026-89875","urls":{"html":"https://cve.report/CVE-2026-89875","api":"https://cve.report/api/cve/CVE-2026-89875.json","docs":"https://cve.report/api","cve_org":"https://www.cve.org/CVERecord?id=CVE-2026-89875","nvd":"https://nvd.nist.gov/vuln/detail/CVE-2026-89875"},"summary":{"title":"media: ti: vpe: quiesce overflow recovery before freeing streams","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nmedia: ti: vpe: quiesce overflow recovery before freeing streams\n\nThe VIP overflow recovery worker is armed from the hardirq handler when a\nFIFO overflow is detected, and the list-complete path looks the stream up\nthrough the VPDMA list private pointer. Both keep touching stream, port\nand device state; the recovery worker also resets the parser and VPDMA,\nrepopulates the descriptor list, and re-enables the per-list IRQs.\n\nvip_stop_streaming() masks and clears the per-list IRQs, but it neither\nsynchronizes the hardirq handler nor disables recovery_work. An overflow\nIRQ that has already queued recovery_work, or a list-complete IRQ in\nflight when the stream is torn down, can therefore still dereference the\nstream after its resources are released: the descriptor list is freed by\nvip_release_stream() on file release, and the stream itself by\nfree_stream() on unbind/remove.\n\nDrain the recovery worker and the IRQ handler at both teardown points\nthrough a shared vip_quiesce_stream() helper, before any stream-owned\nresource is released. disable_work_sync() cancels pending recovery_work,\ndrains a running instance, and raises its disable depth, so a subsequent\nschedule_work() issued by a racing IRQ handler is rejected at the\nworkqueue scheduler: recovery_work cannot be requeued after\ndisable_work_sync() takes effect. The worker may still re-enable the\nper-list IRQs before disable_work_sync() returns; disable_irqs() then\nmasks those sources and synchronize_irq() waits for any in-flight handler\nthat still dereferences stream state. In vip_stop_streaming() the helper\nruns before the parser is stopped, since a worker drained by\ndisable_work_sync() may re-enable the parser before exiting and would\notherwise undo the stop. recovery_work is created disabled and enabled in\nvip_start_streaming() before IRQs, pairing the enable with the teardown\ndisable across the streaming lifecycle.\n\nThis issue was found by an in-house static analysis tool and confirmed\nby manual code review.","state":"PUBLISHED","assigner":"Linux","published_at":"2026-09-16 11:16:55","updated_at":"2026-09-16 15:18:14"},"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/fd5b4a14c87b21e1c9eaee603b01bb7fa7db45c8","name":"https://git.kernel.org/stable/c/fd5b4a14c87b21e1c9eaee603b01bb7fa7db45c8","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://git.kernel.org/stable/c/aeaacc3001449d44b4ab7da56331121d1f3b137b","name":"https://git.kernel.org/stable/c/aeaacc3001449d44b4ab7da56331121d1f3b137b","refsource":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","tags":[],"title":"","mime":"","httpstatus":"","archivestatus":"0"},{"url":"https://www.cve.org/CVERecord?id=CVE-2026-89875","name":"CVE Program record","refsource":"CVE.ORG","tags":["canonical"]},{"url":"https://nvd.nist.gov/vuln/detail/CVE-2026-89875","name":"NVD vulnerability detail","refsource":"NVD","tags":["canonical","analysis"]}],"affected":[{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected fc2873aa4a21355593b55eb49ff534d31f81c584 fd5b4a14c87b21e1c9eaee603b01bb7fa7db45c8 git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected fc2873aa4a21355593b55eb49ff534d31f81c584 aeaacc3001449d44b4ab7da56331121d1f3b137b git","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"affected 7.0","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 7.0 semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 7.2.5 7.2.* semver","platforms":[]},{"source":"CNA","vendor":"Linux","product":"Linux","version":"unaffected 7.3-rc1 * original_commit_for_fix","platforms":[]}],"timeline":[],"solutions":[],"workarounds":[],"exploits":[],"credits":[],"nvd_cpes":[],"vendor_comments":[],"enrichments":{"kev":null,"epss":null,"legacy_qids":[]},"source_records":{"cve_program":{"containers":{"cna":{"affected":[{"defaultStatus":"unaffected","product":"Linux","programFiles":["drivers/media/platform/ti/vpe/vip.c"],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","vendor":"Linux","versions":[{"lessThan":"fd5b4a14c87b21e1c9eaee603b01bb7fa7db45c8","status":"affected","version":"fc2873aa4a21355593b55eb49ff534d31f81c584","versionType":"git"},{"lessThan":"aeaacc3001449d44b4ab7da56331121d1f3b137b","status":"affected","version":"fc2873aa4a21355593b55eb49ff534d31f81c584","versionType":"git"}]},{"defaultStatus":"affected","product":"Linux","programFiles":["drivers/media/platform/ti/vpe/vip.c"],"repo":"https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git","vendor":"Linux","versions":[{"status":"affected","version":"7.0"},{"lessThan":"7.0","status":"unaffected","version":"0","versionType":"semver"},{"lessThanOrEqual":"7.2.*","status":"unaffected","version":"7.2.5","versionType":"semver"},{"lessThanOrEqual":"*","status":"unaffected","version":"7.3-rc1","versionType":"original_commit_for_fix"}]}],"cpeApplicability":[{"nodes":[{"cpeMatch":[{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"7.2.5","versionStartIncluding":"7.0","vulnerable":true},{"criteria":"cpe:2.3:o:linux:linux_kernel:*:*:*:*:*:*:*:*","versionEndExcluding":"7.3-rc1","versionStartIncluding":"7.0","vulnerable":true}],"negate":false,"operator":"OR"}]}],"descriptions":[{"lang":"en","value":"In the Linux kernel, the following vulnerability has been resolved:\n\nmedia: ti: vpe: quiesce overflow recovery before freeing streams\n\nThe VIP overflow recovery worker is armed from the hardirq handler when a\nFIFO overflow is detected, and the list-complete path looks the stream up\nthrough the VPDMA list private pointer. Both keep touching stream, port\nand device state; the recovery worker also resets the parser and VPDMA,\nrepopulates the descriptor list, and re-enables the per-list IRQs.\n\nvip_stop_streaming() masks and clears the per-list IRQs, but it neither\nsynchronizes the hardirq handler nor disables recovery_work. An overflow\nIRQ that has already queued recovery_work, or a list-complete IRQ in\nflight when the stream is torn down, can therefore still dereference the\nstream after its resources are released: the descriptor list is freed by\nvip_release_stream() on file release, and the stream itself by\nfree_stream() on unbind/remove.\n\nDrain the recovery worker and the IRQ handler at both teardown points\nthrough a shared vip_quiesce_stream() helper, before any stream-owned\nresource is released. disable_work_sync() cancels pending recovery_work,\ndrains a running instance, and raises its disable depth, so a subsequent\nschedule_work() issued by a racing IRQ handler is rejected at the\nworkqueue scheduler: recovery_work cannot be requeued after\ndisable_work_sync() takes effect. The worker may still re-enable the\nper-list IRQs before disable_work_sync() returns; disable_irqs() then\nmasks those sources and synchronize_irq() waits for any in-flight handler\nthat still dereferences stream state. In vip_stop_streaming() the helper\nruns before the parser is stopped, since a worker drained by\ndisable_work_sync() may re-enable the parser before exiting and would\notherwise undo the stop. recovery_work is created disabled and enabled in\nvip_start_streaming() before IRQs, pairing the enable with the teardown\ndisable across the streaming lifecycle.\n\nThis issue was found by an in-house static analysis tool and confirmed\nby manual code review."}],"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 bug is reached only through the TI VIP V4L2 capture node: a local process open()s /dev/videoX, STREAMONs to arm parser/list-complete IRQs and overflow recovery, then STREAMOFF or close() tears the stream down. It is not reachable from network packets or a remote peer.\nAC:L - The attacker starts streaming (which generates list-complete IRQs and can queue recovery_work) and then STREAMOFF/close, racing the hardirq and an already-queued worker that the old teardown never drained. Both sides of the race are attacker-induced and retryable, so this UAF is Low complexity.\nPR:L - Exploitation needs only an unprivileged local user who can open the VIP /dev/videoX node and issue V4L2 streaming ioctls. On DRA7/AM57 embedded, industrial, and automotive systems that is typically the video group or a camera app, not root or init-namespace capabilities.\nUI:N - The attacker performs open, STREAMON, and STREAMOFF/close themselves. No separate victim action such as mounting a filesystem or plugging in a device is required.\nS:U - The UAF corrupts host-kernel stream and VPDMA descriptor state used by the same kernel that serves the ioctl. That is standard kernel memory corruption/privilege escalation, not a VM, IOMMU, or sandbox boundary crossing.\nC:H - This is a use-after-free of the VPDMA descriptor list (freed on last file release) and of the stream object (freed on unbind/remove). The IRQ and recovery worker then read those objects, which per CVSS kernel UAF guidance enables kernel memory disclosure.\nI:H - vip_overflow_recovery_work and the list-complete path write the freed descriptor buffer, rebuild DTDs, and resubmit VPDMA lists, giving a kernel write/DMA primitive suitable for control-flow hijack. UAF memory corruption is scored High integrity.\nA:H - The same UAF and unlocked post_bufs/vidq walks from hardirq vs teardown oops or panic the kernel even when not fully exploited, so availability impact is High."}]}],"providerMetadata":{"dateUpdated":"2026-09-16T14:39:29.374Z","orgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","shortName":"Linux"},"references":[{"url":"https://git.kernel.org/stable/c/fd5b4a14c87b21e1c9eaee603b01bb7fa7db45c8"},{"url":"https://git.kernel.org/stable/c/aeaacc3001449d44b4ab7da56331121d1f3b137b"}],"title":"media: ti: vpe: quiesce overflow recovery before freeing streams","x_generator":{"engine":"bippy-1.2.0"}}},"cveMetadata":{"assignerOrgId":"416baaa9-dc9f-4396-8d5f-8c081fb06d67","assignerShortName":"Linux","cveId":"CVE-2026-89875","datePublished":"2026-09-16T10:31:42.777Z","dateReserved":"2026-09-11T19:38:34.772Z","dateUpdated":"2026-09-16T14:39:29.374Z","state":"PUBLISHED"},"dataType":"CVE_RECORD","dataVersion":"5.2"},"nvd":{"publishedDate":"2026-09-16 11:16:55","lastModifiedDate":"2026-09-16 15:18:14","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":"89875","Ordinal":"1","Title":"media: ti: vpe: quiesce overflow recovery before freeing streams","CVE":"CVE-2026-89875","Year":"2026"},"notes":[{"CveYear":"2026","CveId":"89875","Ordinal":"1","NoteData":"In the Linux kernel, the following vulnerability has been resolved:\n\nmedia: ti: vpe: quiesce overflow recovery before freeing streams\n\nThe VIP overflow recovery worker is armed from the hardirq handler when a\nFIFO overflow is detected, and the list-complete path looks the stream up\nthrough the VPDMA list private pointer. Both keep touching stream, port\nand device state; the recovery worker also resets the parser and VPDMA,\nrepopulates the descriptor list, and re-enables the per-list IRQs.\n\nvip_stop_streaming() masks and clears the per-list IRQs, but it neither\nsynchronizes the hardirq handler nor disables recovery_work. An overflow\nIRQ that has already queued recovery_work, or a list-complete IRQ in\nflight when the stream is torn down, can therefore still dereference the\nstream after its resources are released: the descriptor list is freed by\nvip_release_stream() on file release, and the stream itself by\nfree_stream() on unbind/remove.\n\nDrain the recovery worker and the IRQ handler at both teardown points\nthrough a shared vip_quiesce_stream() helper, before any stream-owned\nresource is released. disable_work_sync() cancels pending recovery_work,\ndrains a running instance, and raises its disable depth, so a subsequent\nschedule_work() issued by a racing IRQ handler is rejected at the\nworkqueue scheduler: recovery_work cannot be requeued after\ndisable_work_sync() takes effect. The worker may still re-enable the\nper-list IRQs before disable_work_sync() returns; disable_irqs() then\nmasks those sources and synchronize_irq() waits for any in-flight handler\nthat still dereferences stream state. In vip_stop_streaming() the helper\nruns before the parser is stopped, since a worker drained by\ndisable_work_sync() may re-enable the parser before exiting and would\notherwise undo the stop. recovery_work is created disabled and enabled in\nvip_start_streaming() before IRQs, pairing the enable with the teardown\ndisable across the streaming lifecycle.\n\nThis issue was found by an in-house static analysis tool and confirmed\nby manual code review.","Type":"Description","Title":"media: ti: vpe: quiesce overflow recovery before freeing streams"}]}}}