[Bug 689191] d3dvideosink improvements

GStreamer (bugzilla.gnome.org) bugzilla at gnome.org
Thu Jan 10 02:03:43 PST 2013


https://bugzilla.gnome.org/show_bug.cgi?id=689191
  GStreamer | gst-plugins-bad | 0.10.x

Sebastian Dröge <slomo> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever Confirmed|0                           |1

--- Comment #14 from Sebastian Dröge <slomo at circular-chaos.org> 2013-01-10 10:03:39 UTC ---
(In reply to comment #12)
> Sebastian: While I'm at it, and I apologize for hijacking this thread, I saw
> another commit 'remove scary "while (object.refcount > 0) release (object);"'
> -- I agree that's scary but the reason for that being in there is due to the
> fact that video card manufacturers actually implement part of the DirectX API
> and some of their implementations are just plain buggy and they forget to
> unreference causing a leak. It's not for covering up a logic error.
> 
> In an ideal world, I agree it's a bad idea. Perhaps it would be better to fail
> fast instead of ignoring the problem and allowing the memory leak.
>
> When the sink was first written I had to use it on systems that ran 24/7 for
> years and so slow memory leaks over time become huge problems.
> 
> The original code was tried on several video cards over a few different
> vendors. ATI cards were typically the worst offenders IIRC. And that hack
> eliminated the leaks. Just my $0.02...

The problem with this hack is that it's completely not threadsafe and fragile
:) If it was used to work around bugs in drivers, first of all this should be
reported to the vendor providing this driver. The leak won't be only happening
in GStreamer but also in other applications. I would be fine with
re-introducing it, but only actually use that hack if an environment variable
is set.

-- 
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the QA contact for the bug.
You are the assignee for the bug.


More information about the gstreamer-bugs mailing list