Public bug reported:

systemd can block forever in sd_event_run() after a ratelimit expiry
callback, even though an event source is already pending. PID 1 can hit
this through its mount event source and hang in the initrd.

Upstream fix:
https://github.com/systemd/systemd/commit/1a60b10de9e4a3fe59a2efa3c6a58fd72c305807
https://github.com/systemd/systemd/issues/44107

Reproduced with the upstream public-API reproducer:

- Ubuntu 24.04: systemd 255.4-1ubuntu8.18
- Ubuntu 26.04: systemd 259.9-0ubuntu0.1

$ apt-get install gcc pkg-config libsystemd-dev
$ gcc -O2 -o repro repro.c $(pkg-config --cflags --libs libsystemd)
$ ./repro
HANG: sd_event_run() blocks although a defer source is pending

A fixed build prints "OK: defer source dispatched" and exits 0.

Ubuntu 22.04 uses systemd 249 and is not affected. The v255-stable branch
does not contain the fix, so 24.04 needs a backport of 1a60b10de9e4. The
fix is already in upstream v259.10, so 26.04 can take that stable update.

Workaround for PID 1: SYSTEMD_DEFAULT_MOUNT_RATE_LIMIT_BURST=0.

** Affects: systemd (Ubuntu)
     Importance: Undecided
         Status: New

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2174036

Title:
  sd-event blocks forever after ratelimit expiry

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/systemd/+bug/2174036/+subscriptions


-- 
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs

Reply via email to