Quantcast

decodebin decodegroup/decodechain management ==> was Re: observing problems with dynamic pipeline

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

decodebin decodegroup/decodechain management ==> was Re: observing problems with dynamic pipeline

Nitin Mahajan
Updating the subject after further analysis.
Refining problem statement:
decodebin2 is nased on DecodeGroup which contains DecodeChain per stream (video/audio)
as I explained earlier, my requirement is to remove one stream from a Decodegroup (remove audio DecodeChain) by pushing EOS on demux audio src pad and remove the pad itself.
Then create Decodechain with new audio stream by creating new demuxer audio src pad.
In this process, the problem by going through decodebin design seems that the new decodegroup will be a separate decodegroup which essentially means that an audio stream (audio chain) specific DecodeGroup. That would result into 2 separate DecodeGroups which means that A/V playback can not work as both streams are in separate DecodeGroup. Is it correct understanding ?
In order to solve this problem, probably it would mean to delete all pads of original DecoderGroup and recreate new DecodeGroup with all streams again Older Video Stream and new audio stream. But this would involve a video switch and with an objective to just change audio, the overall exp will be probably a black screen or video freeze when switching across decodegroups. Is it correct understanding ?

Also do we have strict sequence to follow to add new pad, send eos on old pads and removal of old pads in order that original DecodeGroup is drained successfully and switch to new DecodeGroup is successful ?

Also noted some documentation on DecodeBin3 

Looking forward to understand from the decodebin design perspective as currently struggling to even remove older DecoderGroup and create new DecodeGroup, in my example it seems that there are 2 Decodegroups with video stream in original and audio stream in latest one.

Thanks in advance for your help to chelp understand decodebin design and usage.

BR
Nitin


On Sat, Apr 29, 2017 at 2:14 PM, Nitin Mahajan <[hidden email]> wrote:
Hello folks,

I'm facing some problem while playing around with with dynamic pipelines.
Need your help / suggestions based on below scenario in case observer something wrong in approach.

I have a gstreamer demuxer plugin (not the standard plugin) with 1 video and 1 audio pad.
This plugin has 2 source pads: video pad, audio pad. (always 2 pads, doesn't have multiple audio pads per stream as per regular demuxers)

For each pad there is dedicated gst_buffer_pool allocated which contains pool of buffers to host Video and Audio ES post demuxing. In this configuration A/V playback is working OK (the pipeline is constructed with playbin)

Now on top of this I have to support switching to another audio stream present in the container (MP4) at run time when gstreamer pipeline is in playing state (with same reason as explained above that number of src pads is fixed)

demux has to switch to another audio stream as mentioned above which has different codec type. So following the dynamic pipeline principles, following steps are done in sequence to attain this objective:
While injection to demux video pad is ongoing (I guess it should be ok to let continue video inject because we are considering deletion and creation of new src pad for audio only chain) ,older audio pad is removed by following below steps: 
flush the pipeline, send eos event and remove pad followed by no more pad notification:
1. gst_pad_push_event (audio_stream->pad, gst_event_new_flush_start ());
2. gst_pad_push_event (audio_stream->pad, gst_event_new_flush_stop (TRUE));
3. gst_pad_push_event (audio_stream->pad, gst_event_new_eos ());
4. Flush the buffer pool associated associated with audio stream
5. gst_element_remove_pad (xx, audio_stream->pad);
6. gst_element_no_more_pads (xx);

And then a new audio pad is created with following operations
1. gst_pad_set_active pad
2. gst_pad_use_fixed_caps
3. gst_pad_push_event gst_event_new_stream_start
4. gst_pad_set_caps pad
5. gst_element_add_pad
6. new segment event
7. caps event
8. push buffers

gstreamer version: 1.4.5

Note that all above steps are done in sequence and in same thread context which we can call as streaming thread as per gstreamer design. Being sequential operations there is no possibility to push buffers or events on audio stream src pad (old and then new) from any other thread.
But during this period I have another thread which keeps on pushing the buffers on video src pad (which although little different from standard gstreamer streaming thread concept but since the ongoing operations are on audio stream pad so it should not create problem .. problem with buffer timestamps in case ?

The expectation was, after above steps, we can inject audio buffers on new audio pad and overall there should be just change in audio stream with no problem on video.
But the observations are not as per expectation !!!!

Video is stalled for some time when older audio pad is deleted and new audio pad is created.
Thereafter video playback is fine but there is no audio and from demux perspective audio packets are pushed without any error on new audio pad.
And the pipeline dump gives an idea that older chain is still not deleted and new chain is not built yet.

Taking dump of GST pipeline it seems that older pipeline is not deleted and new pipeline from new audio pad is partially created upto proxypad16 inside decodebin,
This pad is connected with Gstaacparse src pad but the chain meets dead end at this stage !!!!

gst_pad_push() is always successful on new audio pad. The buffers which are pushed on audio pad are allocated from buffer pool.

Any suggestions for debugging will be helpful especially with dynamic pipeline management (gstreamer 1.4.5)

Thanks
Nitin


_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
Reply | Threaded
Open this post in threaded view
|  
Report Content as Inappropriate

Re: decodebin decodegroup/decodechain management ==> was Re: observing problems with dynamic pipeline

Nitin Mahajan
Hello,

Gentle reminder. Your suggestions/opinion would certainly be of great help.

Best Regards
Nitin

On Mon, May 1, 2017 at 8:36 PM, Nitin Mahajan <[hidden email]> wrote:
Updating the subject after further analysis.
Refining problem statement:
decodebin2 is nased on DecodeGroup which contains DecodeChain per stream (video/audio)
as I explained earlier, my requirement is to remove one stream from a Decodegroup (remove audio DecodeChain) by pushing EOS on demux audio src pad and remove the pad itself.
Then create Decodechain with new audio stream by creating new demuxer audio src pad.
In this process, the problem by going through decodebin design seems that the new decodegroup will be a separate decodegroup which essentially means that an audio stream (audio chain) specific DecodeGroup. That would result into 2 separate DecodeGroups which means that A/V playback can not work as both streams are in separate DecodeGroup. Is it correct understanding ?
In order to solve this problem, probably it would mean to delete all pads of original DecoderGroup and recreate new DecodeGroup with all streams again Older Video Stream and new audio stream. But this would involve a video switch and with an objective to just change audio, the overall exp will be probably a black screen or video freeze when switching across decodegroups. Is it correct understanding ?

Also do we have strict sequence to follow to add new pad, send eos on old pads and removal of old pads in order that original DecodeGroup is drained successfully and switch to new DecodeGroup is successful ?

Also noted some documentation on DecodeBin3 

Looking forward to understand from the decodebin design perspective as currently struggling to even remove older DecoderGroup and create new DecodeGroup, in my example it seems that there are 2 Decodegroups with video stream in original and audio stream in latest one.

Thanks in advance for your help to chelp understand decodebin design and usage.

BR
Nitin


On Sat, Apr 29, 2017 at 2:14 PM, Nitin Mahajan <[hidden email]> wrote:
Hello folks,

I'm facing some problem while playing around with with dynamic pipelines.
Need your help / suggestions based on below scenario in case observer something wrong in approach.

I have a gstreamer demuxer plugin (not the standard plugin) with 1 video and 1 audio pad.
This plugin has 2 source pads: video pad, audio pad. (always 2 pads, doesn't have multiple audio pads per stream as per regular demuxers)

For each pad there is dedicated gst_buffer_pool allocated which contains pool of buffers to host Video and Audio ES post demuxing. In this configuration A/V playback is working OK (the pipeline is constructed with playbin)

Now on top of this I have to support switching to another audio stream present in the container (MP4) at run time when gstreamer pipeline is in playing state (with same reason as explained above that number of src pads is fixed)

demux has to switch to another audio stream as mentioned above which has different codec type. So following the dynamic pipeline principles, following steps are done in sequence to attain this objective:
While injection to demux video pad is ongoing (I guess it should be ok to let continue video inject because we are considering deletion and creation of new src pad for audio only chain) ,older audio pad is removed by following below steps: 
flush the pipeline, send eos event and remove pad followed by no more pad notification:
1. gst_pad_push_event (audio_stream->pad, gst_event_new_flush_start ());
2. gst_pad_push_event (audio_stream->pad, gst_event_new_flush_stop (TRUE));
3. gst_pad_push_event (audio_stream->pad, gst_event_new_eos ());
4. Flush the buffer pool associated associated with audio stream
5. gst_element_remove_pad (xx, audio_stream->pad);
6. gst_element_no_more_pads (xx);

And then a new audio pad is created with following operations
1. gst_pad_set_active pad
2. gst_pad_use_fixed_caps
3. gst_pad_push_event gst_event_new_stream_start
4. gst_pad_set_caps pad
5. gst_element_add_pad
6. new segment event
7. caps event
8. push buffers

gstreamer version: 1.4.5

Note that all above steps are done in sequence and in same thread context which we can call as streaming thread as per gstreamer design. Being sequential operations there is no possibility to push buffers or events on audio stream src pad (old and then new) from any other thread.
But during this period I have another thread which keeps on pushing the buffers on video src pad (which although little different from standard gstreamer streaming thread concept but since the ongoing operations are on audio stream pad so it should not create problem .. problem with buffer timestamps in case ?

The expectation was, after above steps, we can inject audio buffers on new audio pad and overall there should be just change in audio stream with no problem on video.
But the observations are not as per expectation !!!!

Video is stalled for some time when older audio pad is deleted and new audio pad is created.
Thereafter video playback is fine but there is no audio and from demux perspective audio packets are pushed without any error on new audio pad.
And the pipeline dump gives an idea that older chain is still not deleted and new chain is not built yet.

Taking dump of GST pipeline it seems that older pipeline is not deleted and new pipeline from new audio pad is partially created upto proxypad16 inside decodebin,
This pad is connected with Gstaacparse src pad but the chain meets dead end at this stage !!!!

gst_pad_push() is always successful on new audio pad. The buffers which are pushed on audio pad are allocated from buffer pool.

Any suggestions for debugging will be helpful especially with dynamic pipeline management (gstreamer 1.4.5)

Thanks
Nitin



_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel
Loading...