All vulnerabilities

Servers and applications

CVE-2026-99797

io_uring/cancel: drop sqd->lock while SQPOLL waits

Severity
Unrated
Published
20 Nov 2026
Last changed
20 Nov 2026
Added to UpHeron
20 Nov 2026
Fixed in
6.12.118, 6.18.64, 7.3.5
Sources
CVE Program

Description

In the Linux kernel, the following vulnerability has been resolved:

An exiting SQPOLL thread calls io_uring_cancel_generic() with sqd->lock held, and keeps it across cond_resched() and the sleep until nothing is in flight anymore. A zero-copy send notification can't be canceled and stays pending as long as the peer doesn't read, so the lock is held for as long as a user cares to. io_ring_exit_work() needs it to park the thread, and so do IORING_REGISTER_IOWQ_AFF and IOWQ_MAX_WORKERS. With 64 such rings sharing the thread, every iou_exit slot sleeps on the lock and no ring gets freed for anyone.

Drop the lock around cond_resched() and the sleep, like the main loop of the thread does. Each walk of ctx_list is still under one hold, and nothing is kept across the unlock. The thread drops its io-wq, frees its tctx and clears sqd->thread under one hold after the loop, so whoever holds the lock and finds sqd->thread set can still use them.

From the CVE Program. Shortened; the full text is there.

Affected software

SoftwareAffected versions
Linux kernelServers and applications · cna:linux|linux
  • From 6.0 before 6.12.118
  • From 6.13 before 6.18.64
  • From 6.19 before 7.3.5

About Linux kernel

Unix-like operating system kernel, basis for all Linux operating systems

From Wikidata, CC0 1.0.
GitHub stars
255,726GitHub, 15 Nov 2026
Forks
67,394GitHub, 15 Nov 2026
Developer
Linus TorvaldsWikidata
License
GNU General Public License, version 2.0Wikidata
Written in
CWikidata
First released
25 Aug 1991Wikidata
First published on GitHub
4 Sept 2011GitHub

References

  • CVE.org https://www.cve.org/CVERecord?id=CVE-2026-99797