Opened 12 years ago
Closed 12 years ago
Last modified 12 years ago
#10962 closed Bug Report - General (Fixed)
Elapsed time set to zero when pausing HD recordings; set to the correct value when play resumes
Reported by: | Owned by: | Jim Stichnoth | |
---|---|---|---|
Priority: | minor | Milestone: | 0.25.4 |
Component: | MythTV - Video Playback | Version: | 0.25-fixes |
Severity: | low | Keywords: | |
Cc: | Ticket locked: | no |
Description
MythTV box receiving UK Freesat with S1/S2 tuners.
While watching a recording from a HD channel the OSD shows the correct elapsed time and the correct total time, but when playback is paused the elapsed time is set to zero. When playback is resumed the elapsed time is reset to the correct value and everything is ok again.
Recordings of SD channels behave as expected: the correct elapsed time is displayed while paused.
Attachments (5)
Change History (25)
comment:1 Changed 12 years ago by
Milestone: | → unknown |
---|
comment:2 Changed 12 years ago by
Owner: | set to stuartm |
---|---|
Status: | new → accepted |
comment:3 Changed 12 years ago by
Milestone: | unknown → 0.26 |
---|---|
Resolution: | → Works for me |
Status: | accepted → closed |
I can't reproduce this with 0.26 for BBC 1 HD, BBC HD, ITV1 HD, C4 HD nor NHK HD on Freesat. I'm going to say this is fixed but feel free to re-open if you are able to reproduce with 0.26 on release.
If you do re-open can you please include a screenshot so we know that we're all talking about the same thing? It could also be a theme specific issue so try another theme first.
comment:4 Changed 12 years ago by
OK - I'll wait for the 0.26 release. FWIW I did confirm that the issue was there with several different themes.
Changed 12 years ago by
Attachment: | paused.png added |
---|
Screenshot while paused (elapsed time is zero :-( )
comment:5 Changed 12 years ago by
I have now updated to 0.26 and I still see this problem. I have attached an updated mythbackend --version
output, and screenshots of the issue.
play.png shows that during playback the display is as expected, paused.png shows how the displayed elapsed time is set to zero when playback is paused.
comment:6 Changed 12 years ago by
Resolution: | Works for me |
---|---|
Status: | closed → new |
I can confirm that O. Lupton's screenshots reflect exactly the issue I see on 0.26.
comment:7 Changed 12 years ago by
Owner: | changed from stuartm to Jim Stichnoth |
---|---|
Status: | new → accepted |
It works for me, but the screenshots are very convincing. I've spent some time pondering this, but I still can't come up with a plausible explanation.
Could you provide a mythfrontend log that demonstrates the problem? Please run with "-v all,norefcount --loglevel debug", which may produce a very large log file but hopefully all available clues will be intact.
Also, if you could provide a link to a ~50MB sample, that would be very helpful.
comment:8 Changed 12 years ago by
I will do that as soon as I get a chance, hopefully tonight. For what it's worth, I *believe* I started seeing this when I split my installation into a fully separate frontend & backend, though that may also be confounded by a few other things:
- I believe the backend server is running Ubuntu 12.04 LTS and the frontend is running 12.10.
- I started seeing it when I built a new frontend with new hardware and a fresh install.
- I am using the mythbuntu repos on both frontend & backend, using packages that track the 0.26-fixes branch.
Changed 12 years ago by
Attachment: | mythfrontend.2013-03-21-11-21-47.log.bz2 added |
---|
sample logfile showing pause behaviour
comment:9 Changed 12 years ago by
I have been seeing the same behaviour on my system and just attached a log file with the requested options. Search down to "2013-03-21 11:22:17.791617" to find a "UpdateOSDSeekMessage(Paused, 0)" call that looks like the culprit. The sequence was start frontend, navigate to recordings screen, play the second show in the list (Conan), pause after a second or two, unpause, exit after a few seconds, exit all the way out of the application. It segfaults on the way out but I think that's due to memory corruption issues with the intel graphics drivers.
I'm running master from March 12 (v0.27-pre2-766-g20c0935) with a few non-playback patches. Mostly to fix problems with bonjour or thread safety, you can get the list by searching for open tickets with my name attached.
Like Geoff I started seeing the issue when I built my new machine which is a ivy bridge i7 on debian testing. And, as mentioned, I'm using the on-chip graphics if that makes any difference.
It's easy for me to update to current master and test patches should you want more logging etc.
Oh, the recording is a ATSC mpeg2 but the problem also happens with pvr-250/350 recordings also so if you have any of those around they'll trigger it. Basically pausing any mpeg2 seems to show the problem. I also tried pausing a dvd but the correct time stayed on the osd. There doesn't seem to be any problem with h.264 files, I tried a bunch of those too.
comment:10 Changed 12 years ago by
Oh, interesting! I'm also using VA-API + on-board graphics on an Ivy Bridge i5.
comment:11 Changed 12 years ago by
Might it be that the VA-API decoder is involved in reporting elapsed time, and the original reporter's playback profile was using VA-API for HD recordings but not SD?
(I'm also using ATSC mpeg2 recordings.)
comment:12 Changed 12 years ago by
Thanks for the diagnosis. I can now reproduce the problem when I use the VAAPI playback profile.
comment:14 Changed 12 years ago by
George, please test the latest commit to Master. If that works and doesn't cause any regressions, I'll backport it to fixes.
comment:15 Changed 12 years ago by
Jim, I wish there was a way for me to thank you and write out that I appreciate the fix without Trac flagging my post as blogspam.
comment:16 Changed 12 years ago by
I just tried a fresh master build against all the files that gave me problems this morning and they're all pausing fine now + no regressions on files that paused OK before. Thanks for the quick fix!
comment:19 Changed 12 years ago by
Milestone: | 0.26 → 0.25.4 |
---|---|
Resolution: | → Fixed |
Status: | accepted → closed |
comment:20 Changed 12 years ago by
Confirmed that this fix is present and working downstream in Mythbuntu 0.26-fixes. Thanks again!
mythbackend --version