After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 725047 - tsdemux offsetting timestamp
tsdemux offsetting timestamp
Status: RESOLVED OBSOLETE
Product: GStreamer
Classification: Platform
Component: gst-plugins-bad
1.2.2
Other Linux
: Normal normal
: 1.5.1
Assigned To: GStreamer Maintainers
GStreamer Maintainers
Depends on:
Blocks:
 
 
Reported: 2014-02-24 09:40 UTC by RajuB
Modified: 2014-12-01 14:25 UTC
See Also:
GNOME target: ---
GNOME version: ---


Attachments
PCR/PTS/DTS on pid 0x02DB (61.64 KB, image/png)
2014-02-27 13:36 UTC, Edward Hervey
Details

Description RajuB 2014-02-24 09:40:05 UTC
I am using gstreamer-1.2.2 version. I have a MPTS file, if I run the below given pipeline timestamp is getting offseted to 3 hours. Does any path in tsdemux is broken in 1.2.2 version??

gst-launch filesrc location=timestamp_issue.ts ! tsdemux program-number=1310 ! video/mpeg ! checksumsink

you reproduce the issue using this file: http://goo.gl/DKYnS7

Thanks in advance.
Comment 1 Edward Hervey 2014-02-25 07:12:03 UTC
Also happens in master
Comment 2 Edward Hervey 2014-02-25 07:25:03 UTC
FWIW, there *is* a 3 hour discontinuity in that file.

pcr 0x02db 1151437431105 (11:50:45.830781666) offset:12130324
pcr 0x02db 1444639404849 (14:51:45.163142555) offset:12442968

We do detect the PCR GAP correctly ... but that PCR arrives after the already-shifted PTS.

Not sure how to handle that efficiently. Detect if PTS is "in line" with the previous ones ? If not return GST_CLOCK_TIME_NONE ? Don't push it and wait for next PCR to recalculate proper ts ?
Comment 3 Edward Hervey 2014-02-27 13:15:44 UTC
How was this file created ? Where does it come from ?


It's full of a *lot* of weirdness.
Comment 4 RajuB 2014-02-27 13:26:24 UTC
Hello Edward, thank you very much for your help. This file was captured from a TV channel. The file I have uploaded is a very small piece of bigger TS (65GB) file. When I analyse this stream, I found that timestamp is abruptly jumping to huge values. I can upload 200 Mb File snippet which could be helpful for your analysis.

Thanks for your time. Good day.
Comment 5 Edward Hervey 2014-02-27 13:33:12 UTC
Can you upload a bigger snippet somewhere ? It might help understand why i'm seeing such erratic PTS/PCR values.
Comment 6 Edward Hervey 2014-02-27 13:36:24 UTC
Created attachment 270471 [details]
PCR/PTS/DTS on pid 0x02DB

This is a zoom on PID 0x02DB just after the PCR jump.

the PCR jump in itself is not that much of a problem. What's a lot more concerning is the weird PTS behaviour (the staircase effect).

The PTS/DTS should be at a (more or less) constant offset from the PCR. In this case ... it's not.
Comment 7 RajuB 2014-03-01 09:15:12 UTC
Edward, I lost the bigger file chunk I had. Now I am capturing the live content again. Once it is done, I will upload the file and let you know the link for downloading. Thank you so much for your time. Good day.
Comment 8 RajuB 2014-03-07 06:37:11 UTC
Hello Edward, My apologies for delay in replying. Here I am attaching 200 MB file chunk using which you can reproduce this issue.

File link: http://goo.gl/DRFMNw

I hope it will help you to analyse tsdemux behavior. Many thanks for looking into this issue. Good day.
Comment 9 Tim-Philipp Müller 2014-11-24 17:32:34 UTC
Is this still a problem with recent versions (1.4.x or git master)?
Comment 10 Edward Hervey 2014-12-01 14:25:13 UTC
Both those samples work fine in git master and 1.4, in pull and push mode.

Closing bug