Thanks for reporting this. Interesting to know it might be related to
hibernation. Did you do something non-standard to allow hibernation on
your machine? I understand it isn't officially supported anymore.
I can't reproduce this crash myself. Can you describe what you were
doing when you noticed the crash. Playing audio over Bluetooth before
suspending? Anything like that you can think of?
It would be useful to see your full journal the next time you get a
crash.
In your crash reports I see Pipewire is passing garbage to libdbus where
the crash eventually happens,
#8 0x00007cf04ccb35d7 in bluez_register_endpoint_legacy
(adapter=adapter@entry=0x5f992e1a2ee0,
direction=direction@entry=SPA_BT_MEDIA_SOURCE, uuid=<optimized out>,
uuid@entry=0x7cf04ccfa880 "0000110a-0000-1000-8000-00805f9b34fb",
codec=codec@entry=0x7cf04cbdfc20 <a2dp_codec_sbc>) at
../spa/plugins/bluez5/bluez5-dbus.c:4600
monitor = 0x5f992e18e408
path = 0x5f992e1a2fa0 "\362U\263\327\234_"
object_path = 0x5f992e1a5460 "/MediaEndpoint/A2DPSource/sbc"
Each of your reports points at the same garbage, "\362U\263\327\234_",
which is interesting.
The source data of that "path" variable come from BlueZ's
org.freedesktop.DBus.ObjectManager::GetManagedObjects method and/or the
org.freedesktop.DBus.ObjectManager::InterfacesAdded signal.
Exactly what the sequence of signals / method calls is unclear from the
static traces. A journal log, ideally with Pipewire debugging enabled,
would potentially help narrow it down.
Once interfaces have been advertised by BlueZ, the object paths sent are
copied into Pipewire's SPA module for BlueZ. There they are managed as
spa_bt_adapter objects in an adapter_list.
So the corruption would appear to be coming from one of,
1. BlueZ messages over DBus
2. DBus demarshalling code in Pipewire
3. BlueZ adapter management code in Pipewire
1. seems unlikely, journal logs will confirm.
Reviewing the code in bluez5-dbus.c I would not be surprised at all if
there's a double free or race condition somewhere in there, making case
3 the most likely to me.
I would be great if you can report this upstream, you're in the best
position to answer any questions. You will likely need to follow
https://gitlab.freedesktop.org/pipewire/pipewire/-/wikis/Troubleshooting
to provide all the information they may ask for.
** Changed in: wireplumber (Ubuntu)
Status: New => Incomplete
** Also affects: pipewire (Ubuntu)
Importance: Undecided
Status: New
** Changed in: wireplumber (Ubuntu)
Status: Incomplete => Invalid
** Changed in: pipewire (Ubuntu)
Status: New => Incomplete
** Also affects: bluez (Ubuntu)
Importance: Undecided
Status: New
** Changed in: bluez (Ubuntu)
Status: New => Incomplete
--
You received this bug notification because you are a member of Ubuntu
Bugs, which is subscribed to Ubuntu.
https://bugs.launchpad.net/bugs/2139019
Title:
Wireplumber 0.5.10-3ubuntu1 DBus crashes on hibernation resume
To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/bluez/+bug/2139019/+subscriptions
--
ubuntu-bugs mailing list
[email protected]
https://lists.ubuntu.com/mailman/listinfo/ubuntu-bugs
