Same crash here on Ubuntu 26.04, with an identical stack trace and identical
library offsets. Adding measurements that may help, including a workaround that
does work and two that do not.

ENVIRONMENT
  Ubuntu 26.04 (upgraded from 25.10 on 2026-11-13)
  pidgin                           1:2.14.14-1ubuntu3
  libgstreamer1.0-0                1.28.2-1            (core)
  libgstreamer-plugins-base1.0-0   1.28.2-1ubuntu0.3
  gstreamer1.0-pipewire            1.6.2-1ubuntu1.2
  Session: Wayland (KDE Plasma), GTK2 build of Pidgin

STACK (systemd-coredump, five crashes, byte-identical offsets every time)
  #0  gst_device_get_display_name (libgstreamer-1.0.so.0 + 0x6d1b5)
  #1  n/a                         (libpurple.so.0 + 0x73519)
  #2  n/a                         (libgstreamer-1.0.so.0 + 0x60f21)
  #3-#5                           (libglib-2.0.so.0, g_main_loop_run)
  #6  gtk_main                    (libgtk-x11-2.0.so.0 + 0x155971)

  SegvReason: writing unknown VMA
  Debug output ends with:
    gst_device_get_display_name: assertion 'GST_IS_DEVICE (device)' failed

  Frame #1 is very likely purple_media_manager_unregister_gst_device(), based on
  the resolved frame in bug #2136313 - which also carries the root-cause 
analysis
  (libpurple passing an untyped object from the 'gst-device' property of a
  PurpleMediaElementInfo into GStreamer's type check). Bugs #2136313, #2150783
  and this one look like the same defect reached through three different
  triggers.

A TRIGGER OBSERVATION THAT MAY WIDEN THE PICTURE
  Here the crashes happened during ordinary chat usage - no USB device was
  unplugged by the user, and the machine was not suspended (it had been running
  continuously). Five crashes: four on 2026-11-15 within 80 minutes (21:23:51,
  21:24:52, 22:16:19, 22:44:39) and one on 2026-11-16 at 22:06:55, the last one
  after 76 seconds of normal use.
  Whether some device de-registered itself (PipeWire, or the built-in camera,
  see below) is unverified here - I can only say no user action triggered it.

WORKAROUND THAT WORKS (measured, with a control run)
  Make GStreamer plugins unreachable, so no device monitor is created:
      GST_PLUGIN_SYSTEM_PATH=/nonexistent GST_PLUGIN_PATH=/nonexistent pidgin
  Measured: 0 devices reported instead of 6.
  Control: 343 seconds of normal chat usage with no crash, versus a crash after
  76 seconds in an otherwise identical run on the same machine on the same day
  with GStreamer active.
  Cost: no notification sounds and no audio/video calls. Chat and file transfer
  are unaffected. For chat-only users this is a usable stopgap.

TWO THINGS THAT DO NOT WORK (so nobody has to retry them)
  1. GST_PLUGIN_FEATURE_RANK=<provider>:NONE has no effect - neither for a 
single
     provider nor for all eight present on this system. Measured: 6 devices
     before and 6 after. Device providers appear not to be affected by feature
     ranking.
  2. The 2026-11-16 security update of gst-plugins-base1.0 (1.28.2-1ubuntu0.2 ->
     1ubuntu0.3) does not help. The crash recurred afterwards with 
byte-identical
     offsets, because the affected library libgstreamer1.0-0 (source package
     gstreamer1.0) was not part of that update and is still at 1.28.2-1.

RULED OUT HERE (measured, not assumed)
  * Profile: running with an empty profile (pidgin -c /tmp/pidgin-probe -d)
    crashes the same way.
  * Plugins: ~/.purple/plugins is empty.
  * Package integrity: dpkg -V over pidgin, pidgin-data, libpurple0, pidgin-otr
    and libgstreamer1.0-0 returns rc=0; dpkg -C returns rc=0; ldd reports no
    missing libraries.

DEVICE ENVIRONMENT
  A standalone GStreamer device monitor in the same session enumerates 24 
devices
  and calls get_display_name() on every one WITHOUT crashing. So enumeration
  alone is not sufficient - the difference is in libpurple's callback handling.
  Possibly relevant: the built-in camera is reported TWICE, once via v4l2
  ("HP True Vision FHD Camera (V4L2)", /dev/video0, 17 caps structures) and once
  via the PipeWire/USB provider (12 caps structures). Several audio sinks appear
  four times each. ALSA emits errors during enumeration:
      alsalib error: Parameter DEV must be an integer
      alsalib error: Unknown PCM hw:0,3:{AES0 0x02 AES1 0x82 AES2 0x00 AES3 
0x02}

-- 
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2161135

Title:
  Pidgin crashes when disconnecting USB audio device

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/pidgin/+bug/2161135/+subscriptions


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

Reply via email to