{"id":"AZL-115723","summary":"CVE-2026-99796 affecting package kernel 6.6.161.1-1","details":"In the Linux kernel, the following vulnerability has been resolved:\n\nio_uring/poll: cancel all matching requests in one pass\n\nWith IORING_ASYNC_CANCEL_ALL, __io_async_cancel() calls io_try_cancel()\nuntil nothing is found, and io_poll_cancel() cancels one request per\ncall, starting each lookup at the beginning of the hash. A cancelled\nrequest stays hashed until its task work runs, so every lookup walks\npast the ones already cancelled. Cancelling K of N pending polls visits\nabout K*N/2 requests in one io_uring_enter(), with uring_lock held.\n\nA single CANCEL_FD|ALL on 80000 polls on one eventfd takes 17.06s,\n0.02s with this change. With voluntary preemption the soft lockup\nwatchdog fires:\n\n  watchdog: BUG: soft lockup - CPU#1 stuck for 26s! [repro:100]\n  RIP: 0010:io_cancel_req_match+0x19/0xa0\n   io_poll_cancel+0x73/0x170\n   io_try_cancel+0x7b/0xf0\n   __io_async_cancel+0x41/0x110\n\nWalk the hash once and cancel every match, as io_cancel_remove() does\nfor futex and waitid. Cancel by user_data only needs its own bucket.\n__io_async_cancel() still loops until nothing is found, so a request\nthat io-wq parks in the poll hash meanwhile is still cancelled.\n\nA cancel that matches one request now walks the whole hash twice instead\nof stopping at the match.","modified":"2026-11-21T14:16:17.969500897Z","published":"2026-11-20T08:18:37Z","upstream":["CVE-2026-99796"],"references":[{"type":"WEB","url":"https://nvd.nist.gov/vuln/detail/CVE-2026-99796"}],"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-115723.json"}}],"schema_version":"1.9.2"}