[pulseaudio-tickets] [Bug 98617] unloading module-remap-sink while module-combine-sink is using it causes Assertion '*_head == _item' failed at modules/module-combine-sink.c:818, function output_remove_within_thread().

[bugzilla-daemon at freedesktop.org]

https://bugs.freedesktop.org/show_bug.cgi?id=98617

--- Comment #2 from Tanu Kaskinen <tanuk at iki.fi> ---
I was able to reproduce the crash with the provided script. In case someone
else tries to reproduce this too, note that I first failed to reproduce when I
issued the commands manually, so it's best to use the script. I would guess the
manual commands failed to cause a crash, because the sinks had time to suspend
before running the unload-module command.

Here are my investigation results so far:

When module-remap-sink is unloaded, it has to remove the virtual sink input
that connects it to the master sink. When the "detach" callback of the sink
input is called, module-remap-sink propagates the detach call to all sink
inputs connected to it, so all sink inputs connected to the sink tree get
detached. Then later, when module-combine-sink removes its virtual sink input,
the detach callback gets called again. module-combine-sink assumes that the
detach callback will be called only once, and the result of an incorrect
assumption is a crash.

I think it's reasonable to assume that the detach callback won't be called for
streams that are already detached. I'll try if skipping the detach callback for
already-detached sink inputs works.

-- 
You are receiving this mail because:
You are the assignee for the bug.
You are the QA Contact for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.freedesktop.org/archives/pulseaudio-bugs/attachments/20161207/3555364e/attachment.html>