[Bug 752651] decodebin: segfault on setting to NULL

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=752651

--- Comment #40 from Vincent Penquerc'h <vincent.penquerch at collabora.co.uk> ---
Created attachment 313938
  --> https://bugzilla.gnome.org/attachment.cgi?id=313938&action=edit
WIP trying to get rid of races

This is what I have at the moment. I don't get any issues with mutexes, as the
lifetime of the chain is now OK. I keep a dead boolean in the chain to avoid
calling analyze_new_pad while it's being destroyed. This is still racy though,
as the chain lock would need keeping over analyze_pad, and it can't.

I don't really get the comment about the ref on the signal connect. I don't
think that'd help anyway, as the signal might have been just triggered at the
time the chain starts getting destroyed (ie, the chain starts getting destroyed
just after pad_added_cb released the chain mutex lock, but before
analyze_new_pad is called.

I'm still seeing the "joining from same thread" errors though. Looks like
oggdemux still has a ref of 1, while its bin is now destroyed.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.