2520 replies to this topic

[betacentauri]

I have tested the old version of the gst1.0 branch (so gstreamer 1.0.2) with my patch (see attached patch for gstreamer base plugins).

 - With the patch mp3 playback works, but the progress on the timeline is too fast. Within 10 seconds playback (from start of file) timeline shows over 1 minute.

 - Playback of mkv (without dts track) has the same problem. But there's no stuttering.

 - Playback of vob with dts track didn't work.

 - Playback of mkv with dts soundtrack didn't work. Afterwards the box didn't respond and I had to switch it off.

 - Switching between AC3 and AAC tracks in a mkv shows the same problem as under 0.10. AC3 sound don't work.

 

To the merlinmusicplayer: There are more things that need to be patched to get it working. I can provide a patch for it.

 

 

 

Attached Files

•  revert-0dfdd9186e143daa568521c4e55c9923e5cbc466.patch.txt   558bytes   14 downloads

[athoik]

Hello, 

 

I believe we need change enigma2.bb to use the new SRC_URL because currently we are making tests with an old version of enigma2.

 

Spoiler

-SRC_URI = "git://openpli.git.sourceforge.net/gitroot/openpli/enigma2;protocol=git;branch=${ENIGMA2_BRANCH}"

+SRC_URI = "git://git.code.sf.net/p/openpli/enigma2;protocol=git;branch=${ENIGMA2_BRANCH}"

 

I don't know if changing SRC_URI is enough, i am doing a fresh build because bitbake enigma2 didn't work.

Edited by athoik, 5 July 2013 - 16:24.

[betacentauri]

If you change a bitbake file increase PR or PRINC values. This causes a rebuild of that package.

 

(I already use current repositories)

[athoik]

If you change a bitbake file increase PR or PRINC values. This causes a rebuild of that package.

 

(I already use current repositories)

 

Did you change enigma2-plugins too?

[betacentauri]

Yes, I think I have changed it everywhere.

[athoik]

Yes, I think I have changed it everywhere.

 

Can you create a patch to apply against gst 1.0 branch?

 

PS. gst 1.0 branch is upgraded to 1.0.7 GStreamer.

 

PPS. did you change glib to newer version also?

Edited by athoik, 5 July 2013 - 17:18.

[betacentauri]

It's not sooo easy. Give me some time...

 

I didn't change glib.

[athoik]

It's not sooo easy. Give me some time...

 

I didn't change glib.

 

Yes i know how hard is... I am also making changes locally, so its good not to make the same changes double.

[betacentauri]

Here's the patch.

Included:

- Switch to current repositories

- mp3 patch

 

Not included:

- Revert of glib to old version

- The changes for dreambox media sink

- merlinmusicplayer fixes (only the one is included which prevents compiler to abort)

 

You need to build it from scratch, because I have not increased PR.

Perhaps you wait a little bit with building. I take a look at the glib issue.

Attached Files

•  0001-gst-1.0-Change-to-current-repositories.patch.txt   13.5KB   24 downloads

[athoik]

Not included:

- Revert of glib to old version

 

We cannot revert to old glib version. GStreamer 1.0 requires minimum glib 2.32.

[betacentauri]

Yes, you're right

[athoik]

Dear all,

 

 

There was an assertion error on dvbmediasink (probably Sjaaky was trying to find also what is happening).

 

Gstreamer logs an error like 'unmap: assert failed. This is not a buffer'. In order to find out I added a lot of debug output and commented unmaps and unrefs. It is worse now, because all video's fail after a few seconds, because the system runs out of memory.

(enigma2:XXX): GStreamer-CRITICAL **: gst_buffer_unmap: assertion `GST_IS_BUFFER(buffer)' failed

(gst-launch-1.0:YYY): GStreamer-CRITICAL **: gst_buffer_unmap: assertion `GST_IS_BUFFER (buffer)' failed

This is happening because we are setting the buffer equal to tmpbuff in some places (895, 955).

Calling unref on tmpbuf after calling unmap on buffer seems to solve the assertion problem.

 0001-Fix-GST_IS_BUFFER-assertion.patch.txt   1.4KB   9 downloads

Edited by athoik, 6 July 2013 - 17:07.

[athoik]

Hello,

 

 

Do we need to change the following to enigma2 (servicemp3.cpp line 1922) also, because audio caps were simplified in GStreamer 1.0?

 

-else if ( gst_structure_has_name (structure, "audio/x-raw-int") )

+else if ( gst_structure_has_name (structure, "audio/x-raw-int") || gst_structure_has_name (structure, "audio/x-raw") )

 

Video and audio caps were simplified. audio/x-raw-int and audio/x-raw-float are now all under the audio/x-raw media type. Similarly, video/x-raw-rgb and video/x-raw-yuv are now video/x-raw.

 

http://gstreamer.fre...orting-1.0.html

[pieterg]

Calling unref on tmpbuf after calling unmap on buffer seems to solve the assertion problem.

Does that mean we can uncomment the commented unrefs / unmaps in Sjaaky's commit
http://sourceforge.n...d5086791213968/
as well?
Or do we have more buffer badness still?

[Sjaaky]

Those were all added in search of a solution for the assertion failure. If that is fixed by unref tmpbuf, the rest of the unmap, unref statements should probably be reverted to http://sourceforge.n...e6634d98fae364/

[athoik]

Does that mean we can uncomment the commented unrefs / unmaps in Sjaaky's commit
http://sourceforge.n...d5086791213968/
as well?
Or do we have more buffer badness still?

 

I think it was the only buffer badness.

 

The commented unrefs where inserted by Sjaaky on this commit http://sourceforge.n...d5086791213968/ so we need to delete debug messages and unrefs.

 

Those were all added in search of a solution for the assertion failure. If that is fixed by unref tmpbuf, the rest of the unmap, unref statements should probably be reverted to http://sourceforge.n...e6634d98fae364/

 

Going back to http://sourceforge.n...e6634d98fae364/ we need to re-apply  the event handler for GST_EVENT_CAPS right?

Edited by athoik, 7 July 2013 - 07:37.

[pieterg]

Yes, we cannot revert the whole comnit. (@Sjaaky, you might want to take a look at git-cola, very convenient tool allowing to stage and commit parts of a sourcefile)

[Sjaaky]

Yes my bad. I didn't have a lot of time and a few people asked for gst 1.0 improvements, so I though just commit it and be done with it.

[athoik]

Please use this patch. It will remove extra debug inserted by Sjaaky and hopefully will eliminate the assertion issue.

 

 0001-Fix-GST_IS_BUFFER-Assertion-And-Remove-Useless-Code.patch.txt   6.02KB   10 downloads

Edited by athoik, 7 July 2013 - 19:19.

[athoik]

Issues till now:

 

1. Using OpenPLi dvbmediasink i have a problem with audio/x-raw. Dreaboxes are using different bypass ioctl (0xf). On non Dreaboxes probably will work like a charm.

 

2. Subtitles seem to have a problem on syncronazation

 

3. Trying to play hls streams i am getting missing H.264 decoder.

 

root@dm800se:~# gst-launch-1.0 playbin uri=http://devimages.apple.com/iphone/samples/bipbop/gear1/prog_index.m3u8
Setting pipeline to PAUSED ...
Pipeline is PREROLLING ...
Pipeline is PREROLLED ...
Setting pipeline to PLAYING ...
Buffering, setting pipeline to PAUSED ...
Done buffering, setting pipeline to PLAYING ...
Missing element: H.264 decoder
WARNING: from element /GstPlayBin:playbin0/GstURIDecodeBin:uridecodebin0: No decoder available for type 'video/x-h264, stream-format=(string)byte-stream, alignment=(string)nal'.
Additional debug info:
gsturidecodebin.c(880): unknown_type_cb (): /GstPlayBin:playbin0/GstURIDecodeBin:uridecodebin0
WARNING: from element /GstPlayBin:playbin0/GstPlaySink:playsink: No volume control found
Additional debug info:
gstplaysink.c(2649): gen_audio_chain (): /GstPlayBin:playbin0/GstPlaySink:playsink:
Volume/mute is not available
New clock: GstSystemClock
^CCaught interrupt -- handling interrupt.
Interrupt: Stopping pipeline ...
Execution ended after 10017752335 ns.
Setting pipeline to PAUSED ...
Setting pipeline to READY ...
Setting pipeline to NULL ...
Freeing pipeline ...

 

Maybe stream-format=(string)byte-stream is causing a problem?

 

4. DTS downmix is not working (i tried also gst-plugins-bad-dtsdec, but it downstream to x-raw so i cannot hear something because of issue 1)