{"id":"AZL-114932","summary":"CVE-2026-99468 affecting package kernel 6.6.161.1-1","details":"In the Linux kernel, the following vulnerability has been resolved:\n\nmedia: imx-jpeg: cancel timeout worker when streaming stops\n\nEach per-fd context ctx owns a delayed_work (ctx-\u003etask_timer, callback\nmxc_jpeg_device_run_timeout) armed via schedule_delayed_work() at the end\nof mxc_jpeg_device_run() to recover a stalled encode/decode job. The only\nexisting cancellation is cancel_delayed_work() in the frame-done IRQ\nhandler, which de-queues a pending work item but does not wait for a\ncallback that has already started, and it only runs when a frame completes.\n\nWhen the fd is closed while a job is in flight (the frame-done IRQ has not\nfired yet), nothing syncs the worker before mxc_jpeg_release() frees ctx\nwith kfree() after v4l2_m2m_ctx_release(). A queued or executing\nmxc_jpeg_device_run_timeout() can then recover ctx through\ncontainer_of(&ctx-\u003etask_timer) and dereference it (ctx-\u003emxc_jpeg,\nslot_data, dev_warn) after ctx has been freed.\n\nCancel the worker from mxc_jpeg_stop_streaming(). The cancel cannot live\nin mxc_jpeg_release(): mxc_jpeg_device_run() arms the timer while holding\nonly hw_lock, not the mxc_jpeg-\u003elock mutex that release holds, so a cancel\nin release could still race a concurrent mxc_jpeg_device_run() that\nre-arms the timer afterwards. mxc_jpeg_stop_streaming() instead runs inside\nv4l2_m2m_ctx_release() -\u003e vb2_queue_release(), i.e. after\nv4l2_m2m_cancel_job() has set TRANS_ABORT and waited for any in-flight job\nto finish (so __v4l2_m2m_try_queue() will not queue and v4l2_m2m_try_run()\nwill not run any further job for this context, which prevents\nmxc_jpeg_device_run() from re-arming the timer) and before the m2m context\nis freed. cancel_delayed_work_sync() removes a pending work item and waits\nfor a running callback, so the worker can no longer race with the\nsubsequent kfree(). The cancel is placed before the buffer-release loop so\na concurrently running timeout callback cannot race with it over the same\nbuffers. If the frame-done IRQ canceled a still-pending timer, this cancel\nis a no-op; if the timeout callback has already started, it waits for the\ncallback to finish. The same mxc_jpeg_stop_streaming() call is also\nreached from VIDIOC_STREAMOFF, which drains the worker early, although\nSTREAMOFF itself does not free ctx -- the use-after-free arises only\nwhen the fd is later closed.\n\nThis bug was found by static analysis.","modified":"2026-11-20T14:16:06.748665389Z","published":"2026-11-19T15:18:01Z","upstream":["CVE-2026-99468"],"references":[{"type":"WEB","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-99468"}],"affected":[{"package":{"name":"kernel","ecosystem":"Azure Linux:3","purl":"pkg:rpm/azure-linux/kernel"},"ranges":[{"type":"ECOSYSTEM","events":[{"introduced":"0"},{"last_affected":"6.6.161.1-1"}]}],"database_specific":{"source":"https://github.com/microsoft/AzureLinuxVulnerabilityData/blob/main/osv/AZL-114932.json"}}],"schema_version":"1.9.2"}