Commit graph

97 commits

Author SHA1 Message Date
Andrei Sarakeev
558f9a2a6f videomixer: Fix synchronization if dynamically changing the FPS
https://bugzilla.gnome.org/show_bug.cgi?id=735859
2014-09-04 11:34:26 +03:00
Sebastian Dröge
ce1d4d9f21 videomixer: Use the best width/height/etc if downstream can handle that
Before it was always using whatever downstream preferred, while
the code and documentation claimed something different.

https://bugzilla.gnome.org/show_bug.cgi?id=727180
2014-08-14 16:36:44 +03:00
Ravi Kiran K N
61fe02a018 videomixer: Avoid double free of VideoConvert
https://bugzilla.gnome.org/show_bug.cgi?id=734764
2014-08-14 15:31:48 +03:00
Sebastian Rasmussen
1fa61632fe videomixer: Unref allowed caps after usage
Fixes https://bugzilla.gnome.org/show_bug.cgi?id=734474
2014-08-08 15:59:36 -03:00
Tim-Philipp Müller
deeef84d2c videomixer: fix double unlock in segment seek segment code path
We only want to unlock if we push an event downstream and
jump to done_unlock label afterwards. We would also unlock
in case of a segment seek and then unlock again later, and
nothing good can come of that.

(This code looks a bit dodgy anyway though, shouldn't it
also bail out with FLOW_EOS here in case of a segment seek
scenario, just without the event?)
2014-07-04 20:26:46 +01:00
Vincent Penquerc'h
bbb1a8de1f videomixer: reset QoS on segment event
https://bugzilla.gnome.org/show_bug.cgi?id=732540
2014-07-01 16:35:05 +01:00
Thibault Saunier
45b9ef1825 videomixer: Declare as Compositor in 'klass' 2014-06-26 17:49:23 +02:00
David Fernandez
4ed74d3ab0 videomixer2: Solve segmentation fault when src caps are configured
Change function pointers to NULL while holding the lock to avoid
race conditions

https://bugzilla.gnome.org/show_bug.cgi?id=701110
2014-06-25 16:44:38 +02:00
Tim-Philipp Müller
77badda6b9 videomixer: name collectpads object based on videomixer name
Makes it easier to track things in debug logs when there
are multiple mixers and muxers.
2014-04-16 21:40:45 +01:00
Tim-Philipp Müller
f8d15b1e56 videomixer: better logging of incoming events
The pad and parent names are already logged as part of logging
the object. Instead log the full event details.
2014-04-16 21:38:35 +01:00
Sebastian Dröge
74c23f0f4f videomixer: Create hashtable only when we actually use it
In error cases we previously returned without freeing it.
2014-04-16 17:33:46 +02:00
Sebastian Dröge
d3a2b3c73a videomixer: Chain up to the parent class' dispose function 2014-04-16 17:30:59 +02:00
Mathieu Duponchelle
6cf0f19c14 videomixer: Port to new collectpads API
See: https://bugzilla.gnome.org/show_bug.cgi?id=724705
2014-03-16 17:44:40 +01:00
Thiago Santos
373eceef7c videomixer2: store video info with buffers to keep it in sync
Instead the queued buffer might have an old caps while the pad
is already storing the information for a new caps. Mixing those
while handling buffers will often lead to issues

https://bugzilla.gnome.org/show_bug.cgi?id=725948
2014-03-11 00:49:19 -03:00
Mathieu Duponchelle
532598e360 videomixer: explicitly fail when alpha information would have been lost. 2013-11-27 16:35:46 +01:00
MathieuDuponchelle
83f8ee1d41 videomixer2: Merge tag events to send them in collected.
Otherwise there were race conditions where we would send tags
on a flushing srcpad.

We have a test for that in GES, but this should be tested
systematically with harness in the future as I believe it
is useful for exactly that kind of cases.

https://bugzilla.gnome.org/show_bug.cgi?id=708165
2013-11-22 18:54:35 -03:00
Mathieu Duponchelle
3d780c5c6d videomixer: Check if the pad needs reconfiguration in collected
https://bugzilla.gnome.org/show_bug.cgi?id=709384
2013-10-04 10:53:26 +02:00
Thibault Saunier
9f4a8ccdf4 videomixer: Do not check if caps are empty when they are NULL
In the case the caps are actually NULL, we should just concider it the
same way as empty caps in that case.
2013-09-11 14:33:31 -03:00
Mathieu Duponchelle
8db40a8c7f videomixer: Add colorspace conversion
https://bugzilla.gnome.org/show_bug.cgi?id=704950
2013-09-10 10:37:23 +02:00
Mathieu Duponchelle
707e39fe7a videomixer: Don't send reconfigure event when formats or PAR are different
It is racy with multiple pads.

https://bugzilla.gnome.org/show_bug.cgi?id=704950
2013-09-10 10:36:48 +02:00
Mathieu Duponchelle
8db3648544 videomixer: Bundle private copies of videoconvert code
Ideally, this would be part of libgstvideo.
Prefixes videoconvert symbols with videomixer_.

https://bugzilla.gnome.org/show_bug.cgi?id=704950
2013-09-10 10:36:30 +02:00
Mathieu Duponchelle
b68f419b6f videomixer: Don't set EOS to FALSE when the collectpad *is* EOS
https://bugzilla.gnome.org/show_bug.cgi?id=707238
2013-09-04 11:09:04 +02:00
Mathieu Duponchelle
5d21f8f2e3 videomixer: don't send flush_stop twice.
If we get flush start and a seek we need to only send flush_stop once.

More info at #706441
2013-08-23 20:17:11 -04:00
Thibault Saunier
e47ffb203b videomixer: Do not send flush_stop ourself after a flush_start
When we receive a flush_start, we should wait for the next flush_stop
and foward it, not create a flush_stop ourself.
2013-08-17 11:40:27 +02:00
Thibault Saunier
6c349d6ec3 videomixer: Send EOS if buf_end >= segment.stop
That means the whole segment is already played, and we are sure we
are EOS at that point.

Also handle segment seeks, and do not send EOS in that case.
2013-08-11 19:05:18 +02:00
Thibault Saunier
c5fa4666b7 videomixer: Make sure to send EOS if the buffer end time equals the segment end time
Otherwize EOS never gets sent in that particular case.
2013-08-06 12:21:33 +02:00
Mathieu Duponchelle
d67a671bfb videomixer: use gst_util_uint64_scale*_round.
There could be a case where:
      1) you do a new set_caps after buffers have been processed.
      2) ts_offset gets set to a different value, eg 0.033333333
      3) your pads get EOS, but the check dor that doesn't work
         because you use ts_offset + a truncated value < segment.stop
      4) so in the next collected, you end up comparing for example:
      0.9999999999 > 1., which is false and means you don't send EOS.

Also adds scale_round in two other places where it potentially could
have caused problems.
2013-07-21 19:21:57 -04:00
Sebastian Dröge
3d0988f46f videomixer: Fix handling of buffers without a duration
We'll have to pop buffer from collectpads and store it
internally only to get the timestamp of the next buffer.
If we continue to keep it in collectpads, no new buffer
to calculate the end time will ever arrive.

https://bugzilla.gnome.org/show_bug.cgi?id=703743
2013-07-09 12:42:17 +02:00
Sebastian Dröge
9e9d2ce098 videomixer: Fix negotiation with 0/1 framerates
https://bugzilla.gnome.org/show_bug.cgi?id=703743
2013-07-09 11:53:28 +02:00
Sebastian Dröge
01cc493944 Revert "videomixer: When all sinkpads are eos, update output segment stop and forward it"
This reverts commit 2d3910fc79.

It's not solving any problem and instead causes code to fall apart.

https://bugzilla.gnome.org/show_bug.cgi?id=701519
2013-06-12 18:25:59 +02:00
Mathieu Duponchelle
6e23f1fec4 videomixer: check last end_time after conversion to running segment
The last end_time was saved after conversion, so the comparison
had to be made after conversion for it to make sense.

https://bugzilla.gnome.org/show_bug.cgi?id=701385
2013-06-11 21:03:35 +02:00
Mathieu Duponchelle
4243714301 videomixer: add mix->segment.start to output_end_time
When the segment start is not 0, this created a situation where
the output_end_time is inferior to output_start_time, and the duration
of the next buffer ended up underflowing.

https://bugzilla.gnome.org/show_bug.cgi?id=701385
2013-06-11 21:03:03 +02:00
Sebastian Dröge
e2e1d1a158 videomixer: Add FIXME comment about the DURATION query from adder
Currently the code just takes with maximum upstream duration, which
is wrong. It should be the maximum upstream duration in running time.
2013-05-30 23:56:38 +02:00
Mathieu Duponchelle
5223868caa videomixer: Set a reference to mix->current_caps as the QUERY_CAPS result. 2013-05-30 15:36:48 -04:00
Nicolas Dufresne
cd30a81ee3 videomixer: Maintain z-order when new pad are added
https://bugzilla.gnome.org/show_bug.cgi?id=701109
2013-05-27 22:43:25 -04:00
Thibault Saunier
7a3df1ab31 videomixer: Always handle flush_stop_pending atomically
It is not protected with the COLLECT_PADS_STREAM_LOCK anymore
2013-05-25 12:20:08 -04:00
Thibault Saunier
608bd3e2db videomixer: Do not take COLLECT_PADS_STREAM_LOCK when unnecessary
Collectpad takes the lock itself when receiving serialized events
and we should not take it for not serialized ones
2013-05-25 11:03:31 -04:00
Nicolas Dufresne
d8c5e31657 videomixer: Don't hold stream-lock while pushing non-serialized events
https://bugzilla.gnome.org/show_bug.cgi?id=700868
2013-05-23 09:20:04 -04:00
Nicolas Dufresne
a7e0f251ca videomixer: Don't hold object lock while sending events
https://bugzilla.gnome.org/show_bug.cgi?id=700868
2013-05-23 09:20:04 -04:00
Thibault Saunier
18ef4f18d0 videomixer: Send a reconfigure event upstream if sinkpad caps are not usable
https://bugzilla.gnome.org/show_bug.cgi?id=684237
2013-05-21 12:15:36 -04:00
Mathieu Duponchelle
2d3910fc79 videomixer: When all sinkpads are eos, update output segment stop and forward it
https://bugzilla.gnome.org/show_bug.cgi?id=699793
2013-05-20 21:06:56 +02:00
Mathieu Duponchelle
521c9a7b5d videomixer: Don't reset the output segment on flush stop
Only init it when getting from READY to PAUSED, and change it on seek events.

https://bugzilla.gnome.org/show_bug.cgi?id=699793
2013-05-20 21:03:03 +02:00
Thibault Saunier
86b106091c videomixer: Send caps event from the streaming thread
This way we avoid races in caps negotiation and we make sure
that the caps are sent after stream-start.

https://bugzilla.gnome.org/show_bug.cgi?id=684237
2013-05-19 09:28:04 -04:00
Thibault Saunier
718f9004d0 videomixer: Do not send flush_stop when receiving a seek
There is no reason to send a flush-stop when receiving a seek event.
In the case of a flushing seek, we could eventually want to, but in
the code path were we check if the seek is "flushing", we have the
following comment that makes sense:

"we can't send FLUSH_STOP here since upstream could start pushing data
after we unlock mix->collect.
We set flush_stop_pending to TRUE instead and send FLUSH_STOP after
forwarding the seek upstream or from gst_videomixer_collected,
whichever happens first."

https://bugzilla.gnome.org/show_bug.cgi?id=684237
2013-05-19 09:28:04 -04:00
Thibault Saunier
85b6852deb videomixer2: Protect flush_stop_pending with the collectpad stream lock
And make sure to expect a flush-stop after a flush-start

https://bugzilla.gnome.org/show_bug.cgi?id=684237
2013-05-19 09:28:04 -04:00
Sebastian Dröge
3110b7cc31 Revert "videomixer2: Take into account new segments"
This reverts commit 84ae670ab4.

Actually this is not how it is supposed to work. videomixer
creates a [0,-1] segment and then puts frames of the different
streams there based on their running times in their own segments.
2013-05-09 16:26:19 +02:00
Mathieu Duponchelle
84ae670ab4 videomixer2: Take into account new segments
Also forward the event downstream on the next opportunity.

https://bugzilla.gnome.org/show_bug.cgi?id=699793
2013-05-09 16:18:54 +02:00
Sebastian Dröge
1588cda9a1 videomixer2: Send stream-start before caps event
https://bugzilla.gnome.org/show_bug.cgi?id=699895
2013-05-08 16:02:46 +02:00
Mathieu Duponchelle
6b153ce385 videomixer: send stream-start event. 2013-04-25 16:09:34 -03:00
Olivier Crête
24bb263d54 videomixer: Don't unref query, we don't own it
Fixes double-unref bug. Bug found by Youness Alaoui
2013-04-16 19:29:48 -04:00