Always get EOS on pad probe

[longkas]

Hello,
I created a pipeline withe a videomixer, and two source as the mixer input,
I want to dynamically change one of the input when its stream gets to end,
see pipeline below or the image attached. So I add an event probe on
videoconvert "conv2" when I add that source, I will get EOS as expected the
first time , but after that every time the EOS event will push immediately
to the probe, please help me figure it, thanks


videomixer name=mix ! autovideosink \
   videotestsrc ! videoconvert ! mix. \
   videotestsrc name=src2 num-buffers=200 ! videoconvert name=conv2 ! mix. \

<http://gstreamer-devel.966125.n4.nabble.com/file/t378770/ggs122.jpg>

1) remove src2 and conv2 from pipeline
2) new src2 (with num-buffers set) and conv2, link src2-conv2-mixer
3) add probe
pad = gst_element_get_static_pad(conv2, "sink");
gst_pad_add_probe(pad, GST_PAD_PROBE_TYPE_EVENT_DOWNSTREAM,
padProbeCallback, data, NULL);

4) padProbeCallback(GstPad *pad, GstPadProbeInfo *info, gpointer user_data)
{
GstEvent *event = gst_pad_probe_info_get_event(info);
if (GST_EVENT_TYPE(event) == GST_EVENT_EOS) {
       #goto 1)
}
}



--
Sent from: http://gstreamer-devel.966125.n4.nabble.com/
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Nicolas Dufresne-5]

Le vendredi 22 mars 2019 à 07:24 -0500, longkas a écrit :
> Hello,
> I created a pipeline withe a videomixer, and two source as the mixer input,
> I want to dynamically change one of the input when its stream gets to end,
> see pipeline below or the image attached. So I add an event probe on
> videoconvert "conv2" when I add that source, I will get EOS as expected the
> first time , but after that every time the EOS event will push immediately
> to the probe, please help me figure it, thanks

EOS is a sticky event, you need to send flush-start/flush-stop sequence
on all the produced for this sub-graph to clear the EOS.

>
>
> videomixer name=mix ! autovideosink \
>    videotestsrc ! videoconvert ! mix. \
>    videotestsrc name=src2 num-buffers=200 ! videoconvert name=conv2 ! mix. \
>
> <http://gstreamer-devel.966125.n4.nabble.com/file/t378770/ggs122.jpg>
>
> 1) remove src2 and conv2 from pipeline
> 2) new src2 (with num-buffers set) and conv2, link src2-conv2-mixer
> 3) add probe
> pad = gst_element_get_static_pad(conv2, "sink");
> gst_pad_add_probe(pad, GST_PAD_PROBE_TYPE_EVENT_DOWNSTREAM,
> padProbeCallback, data, NULL);
>
> 4) padProbeCallback(GstPad *pad, GstPadProbeInfo *info, gpointer user_data)
> {
> GstEvent *event = gst_pad_probe_info_get_event(info);
> if (GST_EVENT_TYPE(event) == GST_EVENT_EOS) {
>        #goto 1)
> }
> }
>
>
>
> --
> Sent from: http://gstreamer-devel.966125.n4.nabble.com/
> _______________________________________________
> gstreamer-devel mailing list
> [hidden email]
> https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

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

[longkas]

After experiment I guess the problem is when a new sub graph is added to the
pipeline, the pipeline tries to seek data using the clock time, I tried to
gst_pad_set_offset(mixer, offsettime) but it seems not work, could you give
me suggestions



--
Sent from: http://gstreamer-devel.966125.n4.nabble.com/
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[longkas]

I find a similar question here:
http://gstreamer-devel.966125.n4.nabble.com/gst-1-2-2-remove-a-request-pad-from-videomixer-cause-SEGFAULT-td4664827.html
,  only set offset on the pad is not enough, the QOS request sent to the
pipeline will be a mess and causing problems. I have dropped the QOS request
for now, but the right approach is to modify the QOS response instead maybe?



--
Sent from: http://gstreamer-devel.966125.n4.nabble.com/
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

[Nicolas Dufresne-5]

Le sam. 23 mars 2019 07 h 11, longkas <[hidden email]> a écrit :

I find a similar question here:
http://gstreamer-devel.966125.n4.nabble.com/gst-1-2-2-remove-a-request-pad-from-videomixer-cause-SEGFAULT-td4664827.html
,  only set offset on the pad is not enough, the QOS request sent to the
pipeline will be a mess and causing problems. I have dropped the QOS request
for now, but the right approach is to modify the QOS response instead maybe?

That would match what we do in streamsynchronizer element. We found though the it can be racy in some circumstances, just be aware. Usually if you combine with a blocking probe it's ok.

--
Sent from: http://gstreamer-devel.966125.n4.nabble.com/
_______________________________________________
gstreamer-devel mailing list
[hidden email]
https://lists.freedesktop.org/mailman/listinfo/gstreamer-devel

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