SRU justification for input-remapper 2.2.0-1ubuntu0.26.04.1 (resolute).
Test build available in ppa:chris-bainbridge/sru; requesting review by
universe-sponsors.

[Impact]

* input-remapper-control, run by the installed udev rule for every input
  device add event, crashes with an unhandled gi.repository.GLib.GError
  when the system D-Bus is unavailable. The crash was reported on 26.04
  (LP #2174519) and matches upstream issue
  https://github.com/sezanzeb/input-remapper/issues/1148 : the udev rule
  runs input-remapper-control during coldplug, before
  /run/dbus/system_bus_socket exists (and whenever the bus is stopped or
  restarting). In 2.2.0, Daemon.connect() calls SystemBus() outside the
  try block that handles GLib.GError, so the exception escapes and
  Apport records a crash report; the autoload attempt also fails.

* The fix moves the SystemBus() call inside the existing GLib.GError
  handlers (initial connect and fallback retry loop). When the bus is
  unavailable the helper now logs an error and exits 5 instead of
  raising an unhandled exception. Behaviour when the bus is reachable is
  unchanged.

* The equivalent change for the dasbus-based code was submitted upstream
  (https://github.com/sezanzeb/input-remapper/pull/1396, refs #1148) and
  is already uploaded to the tenacious development series as
  2.2.1-1ubuntu1 (devel-first requirement).

[Test Case]

* Reproduce with the unpatched package (2.2.0-1):

    sudo unshare -m sh -c 'mount -t tmpfs tmpfs /run/dbus && input-
remapper-control --command autoload --device /dev/input/event0'; echo
exit=$?

  Result: unhandled traceback ending in
  "gi.repository.GLib.GError: g-io-error-quark: Could not connect: No
  such file or directory (1)", exit code 1, Apport crash report.

* With the patched package the same command logs

    ERROR: Service not running? g-io-error-quark: Could not connect: No such 
file or directory (1)
    ERROR: Daemon missing

  and exits 5, with no traceback and no Apport report. Verified on
  26.04.1 with the 2.2.0-1ubuntu0.26.04.1 PPA build.

* Regression check: input-remapper-daemon starts normally, remapping
  autoloads at login (session autostart) and on device hotplug, and
  input-remapper-gtk starts normally. The full upstream unit test suite
  passes on the patched source (538 tests, run in an Ubuntu 26.04 VM as
  a non-root user), including two new regression tests covering the
  no-system-bus path.

[Regression Potential]

* Low. The patch only moves an existing call inside an existing
  exception handler; no other code path changes. If the bus is
  unavailable, the behaviour changes from "unhandled exception + Apport
  report" to "logged error + exit 5" (the fallback path still retries
  and exits 8 if the bus remains unavailable, as before). Remapping
  itself is unaffected: the session autostart performs the real
  autoload after login, and hotplug autoload works once the bus is up.

** Bug watch added: github.com/sezanzeb/input-remapper/issues #1148
   https://github.com/sezanzeb/input-remapper/issues/1148

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2174519

Title:
  input-remapper-control crashed with gi.repository.GLib.GError in
  bus_get() during udev autoload

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/input-remapper/+bug/2174519/+subscriptions


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

Reply via email to