<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
Hello Stefan,<br>
<br>
many many thanks, I had a deeper look into GStreamer queue element
and putting it into our pipeline with the appropriate property
values fixed our issues.<br>
<br>
Your help was very precious to us!<br>
<br>
Best regards,<br>
<br>
Matteo<br>
<br>
On 08/02/2012 20:14, Stefan Sauer wrote:
<blockquote cite="mid:4F32C9A5.5040905@hora-obscura.de" type="cite">
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
On 02/08/2012 05:04 PM, Matteo Pampolini wrote:
<blockquote cite="mid:4F329D0C.5090204@selexelsag.com" type="cite">
<meta http-equiv="content-type" content="text/html;
charset=ISO-8859-1">
<font size="-1">Hello there,<br>
<br>
my name is Matteo and together with my group I'm developing a
quite complex video-surveillance application based on
GStreamer (Windows). For simplicity you can imagine a window
with a grid layout with many concurrent videos showing in
their respective area and fetching data through RTSP/RTP.<br>
<br>
We noticed that if we make any change that in some way slows
down or stops the visualization the memory heap starts
increasing very fast until the process inevitably crashes.<br>
<br>
Then I tried to insulate the issue by simply using gst-launch
with an RTSP/RTP source and putting a very silly<br>
<br>
while(1)<br>
Sleep(500);<br>
<br>
inside the show_frame function of videosink plugin (in our
case d3dvideosink), and the problem is easily reproduced.<br>
</font></blockquote>
Are you having unbound queues in your pipeline? The videosink will
usually discard late buffers. If it does not run, upstream
elements would be stalled. If there are queues in the upstream
graph, the blocking would only go to the queue and the elements on
the other side would fill up the queue. If the queue is limited it
would stop the upstream elements once its full.<br>
<br>
Stefan<br>
<blockquote cite="mid:4F329D0C.5090204@selexelsag.com" type="cite"><font
size="-1"> <br>
Can you please try to give me an idea of the lifetime of
GstBuffer through the pipeline? I mean, the heap issue is
surely caused by a continuous allocation of GstBuffer's by the
RTP source without a corresponding free when they are
consumed, but who is in charge of this? Why isn't there a
mechanism that checks the consumer status and eventually stops
the producer?<br>
<br>
Many thanks in advance for your kind support,<br>
<br>
Matteo <br>
</font>
<pre wrap=""><fieldset class="mimeAttachmentHeader"></fieldset>
_______________________________________________
gstreamer-devel mailing list
<a moz-do-not-send="true" class="moz-txt-link-abbreviated" href="mailto:gstreamer-devel@lists.freedesktop.org">gstreamer-devel@lists.freedesktop.org</a>
<a moz-do-not-send="true" class="moz-txt-link-freetext" href="http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel">http://lists.freedesktop.org/mailman/listinfo/gstreamer-devel</a>
</pre>
</blockquote>
<br>
</blockquote>
<br>
</body>
</html>