OMXPlayer randomly hangs #12
Comments
|
Log file can be obtained from http://hosting.cmalton.me.uk/chrism/omxplayer.log.gz |
|
Does sending q to stop work at this point? Really I'll need a file that exhibits the problem. |
|
Haven't tested that - will do so next time it hangs. The script I use to do the conversion to x624 is as follows: #! /bin/bash
FILEBASE=`echo $1 | sed -E -e 's/\.(.*)$//g'`
ffmpeg -i $1 -vcodec libx264 -s 1920x1080 -aspect '16:9' -vb 4M -acodec libmp3lame -ab 128k ${FILEBASE}-hd.mp4All of the files are encoded using that. Like I say, this happens on one of N files (where N is any number!), and only started happening since early April (the version of OMXplayer we used at the last event in May worked fine!). The file is seemingly random, although there are a couple that cause the issue more often than others. |
|
I've got clearance to send over one of the videos that tends to cause the problem. It's a whopping 11MB and we'd prefer not to send it completely in the clear. Could you ping me over an email address I can send the link to? |
|
I can confirm that sending a single q at this point will cause OMXplayer to stop and quit correctly, so this seems to not be a full hang per-say, more OMXplayer running off the end of the video and carrying on! |
|
Looking at it, it seems huceke#178 is possibly related. |
|
Also: raspberrypi/firmware#185 I'll try an out-of-band firmware update tomorrow. |
|
Out-of-band (i.e. non-APT) firmware installed from raspberrypi's git repo. |
|
|
|
Like I say, this isn't 100% reproducible - it occurs at some point after so many videos have run. I couldn't tell you how many times I have to run OMXPlayer (it seems to be ~45 mins to 1hr in, which is around 100 plays (based on average duration of the clips we are running) Version is as follows: I've currently got a workaround in place that has a 1 minute 40 second timeout, and if OMXPlayer hasn't quit by that point, it quits OMXPlayer and then the next video plays. Does that help matters? |
|
I've been experiencing similar problems and I can reproduce it. I have thousands of short h264 encoded (directly from iPhone with no transcoding) videos that are played in a loop one after another in xbmc. I noticed in recent Raspbmc and OpenELEC builds that the video stream would frequently hang which never used to happen in the past. There were no messages in xbmc.log that gave any indication of the fact this was happening or why--videos just stopped playing. To debug, I built a video loop using omxplayer directly and cutting xbmc out of the picture, but encountered the same problem. What I've found is that on most videos omxplayer shuts down and returns to the console once it finishes playing, but on some videos omxplayer doesn't finish and just stalls...usually toward the end of the video with about one second to go. If you press 'q' omxplayer quits gracefully. Separate to the shell script, I see the same behavior in a version of pyomxplayer I've used---pyomxplayer never receives an EOF from omxplayer on certain videos. I'm using an updated/upgraded 2013-05-25-wheezy-raspbian and the very recent 4fa7d64 build of omxplayer. Let me know if you'd like me to PM a copy of the video causing problems--a typical one is about 60 MB. |
|
top of the log file is below for command: omxplayer -o local -s - g IMG_2314.MOV |
|
@andyjagoe: Thanks for confirming this isn't just me. I've had to bodge around it in my software - which is highly unpleasant. Look forward to hearing from @popcornmix when the problem has been identified. More than happy to try testing builds now - I have about a month and a half to get this under control for the project I'm working on now. |
|
@cjsoftuk: No problem. Are your freezes usually at the end of the video as On Sun, Jul 7, 2013 at 11:19 PM, cjsoftuk notifications@github.com wrote:
|
|
@andyjagoe: Yes, all my freezes that I've noticed were at the end as well. On 08/07/13 08:45, andyjagoe wrote:
|
|
@andyjagoe |
|
@popcornmix It is shared to your gmail account. Let me know if you have any problems accessing it. |
|
@andyjagoe |
|
I've committed a fix that fixes @andyjagoe's file. |
|
Thanks @popcornmix. Hugely appreciate the quick response. I'm traveling but will be able to try it out in a day or so. |
|
Fix is working great and I'm seeing no further problems. Thanks for the quick response. I'm ok to close this issue, but I'll leave that to you or to @cjsoftuk since he hasn't commented yet and he opened it. |
|
Cheers for the quick fix @popcornmix - I'll try and test it out as soon as I can. I see that the sconde.net builds are now up to date with this fix patched in, so I'll be trying it out by leaving a Pi running tomorrow when I go out to work. |
|
OK, I've left it running, and all seems good at the moment, I'll let you know when I get back from work if it's still getting stuck. |
|
@popcornmix This doesn't seem to have fixed my issue unfortunately! I was just about to go out to work and I've noticed it's got stuck (And this time it was the first time it played that video I sent you). |
|
OK, this is a curious one. I've set off a test run calling just "omxplayer -g /content/Combat-hd.mp4" and we're at iteration 30 without any issue so far. I'll try and pin this down further. |
|
Yep - seems to be with the way I'm calling omxplayer now (i.e. omxplayer and SDL interacting badly!) - I'd like to keep this open for the moment @popcornmix until such point as I can pin down what's actually responsible (SDL or OMXplayer or python) Last night we got to 300+ iterations of the same video without faolure. |
|
Okay. Let us know when you have more info. |
|
@cjsoftuk did you get any more info on this? I just experienced this issue again with another file and (unfortunately) cannot repeat it. The hang was very curious in that it was clearly the end of the video, but omxplayer was continuing to stream stats (I used the '-s' option) and increment the M: value. @popcornmix ...any thoughts on how we might best to trap this issue? Not sure if my omxplayer.log was of any value for the last file you helped with. |
|
Also...is it possible to specify a path / filename for the '-g' option? I'm running my loop now so next time it hangs I'll have the omxplayer log file. |
|
@andyjagoe No way to change the destination directory of log file other than changing current directory. |
|
Hi @jehutting currently I am running ArchLinux on my RPi's, so I will try to do something with your debpkg, but dunno if this idea works (running debian packages on arch). |
|
@jehutting Hopefully it will be in a @skgsergio build soon for more to test. |
|
@popcornmix Your commit looks OK to me. Thanks. @aawray Arch Linux...google'd around and it should be possible to install a debian package with I found http://archlinuxarm.org/forum/viewtopic.php?f=59&t=6638&p=36868&hilit=debian#p36868 with some interesting code. So what I did
After this I was able to run |
|
I'm installing for testing now. Looping a 10 minute mp4 which has been hanging something awful this week. A stable 24 hours would be a nice relief. - except I'm not setup for cross compiling. Ehhh... |
|
@jehutting, great inspiration, thank you! I will try this out! |
|
@jehutting good news :) since 11th Feb my test rig has played back 8,474 videos none of which needed omxpayer to be killed; previous to that, there was an average of 38 kills per day out of 1,200 playbacks per day) Thanks for your work 👍 Do you have a view on whether this binary is suitable for a "stable" release? |
|
@jehutting I've had this running for 6 days and it's still running smoothly. I'm hesitantly relieved to say it's working great. Well done... |
|
@jehutting I've run all my test rigs without any failures. Great work Jozef |
|
@jehutting I also confirm that it works! Yay! |
|
@cjsoftuk |
|
I'll run some tests over the weekend and then yes, sounds probably ok. Anyone able to give me an updated deb for latest raspbian? Chris On 21 February 2014 11:31:02 GMT+00:00, popcornmix notifications@github.com wrote:
Sent from my Android phone with K-9 Mail. Please excuse my brevity. |
|
http://omxplayer.sconde.net/ has been updated with latest deb. |
|
Cheers @popcornmix - Will leave it running this weekend. If I haven't hit this issue by Monday, I'll give the all-clear. |
|
I've just set the test running on my RasPi with the production software I won't be able to check on it until Sunday, but will let you know if Regards, Chris |
|
Thanks @aawray @johnspackman @XionTech @KenT2 @andyjagoe for testing my package, and for the responses that it all works fine. I finally got the dbus changes tested; dbus is still working ;). I still have to test the keyboard changes regarding to the key configuration. |
|
Generally looking good - think I had one hang last night but I'm not sure if that was OMXPlayer or some external factor. Leaving to run while I'm in the office today and will let people know if I hit this issue again. |
|
OK, seems to be a LOT, LOT better than it used to be. The odd video does still hang though, and I'm now seeing this in dmesg..... |
|
When the hang happens, sending q to OMXPlayer isn't enough it seems. I have to SIGKILL it to get it to die. I'll leave it running another 24 hours - although I do wonder if it's the videos themselves. I'll try and get the OK to put them somewhere for people to experiment with. |
|
@cjsoftuk |
|
@popcornmix Yep - figured it may be a different issue. Overclock is off I believe, and it's on a dedicated 1.2A PSU. I just find it curious that it's a recoverable panic, rather than a fatal failure. Will close this issue and then look at this under a new issue on the relevant project. |
|
Filed an issue at raspberrypi/firmware#253 |
Got a project that I can't disclose much info about, unfortunately, however.....
We use omxplayer and Raspberry Pi as part of the project, and we play lots of videos, back-to-back (i.e. OMXPlayer quits, and our project starts a new one almost immediately).
However, from time to time (usually within an hour), OMXPlayer will hang.
We do have a custom kernel we use, which I can post the .config for later today and I plan to swap back to stock kernel if the issue recurs again today (I'm currently in the middle of doing live demos to prospective customers). Other than the custom kernel, we're up to date as far as the bootloader goes and all other packages (inc OMXPlayer - 0.3.0 installed to try and fix this issue).
Usually, the hang will occur at the end of the video. OMXPlayer doesn't quit, doesn't print any messages to the log.
We're invoking omxplayer as follows:
DISPLAY=:0 /usr/bin/omxplayer -o local --win "0 0 160 96" /content/file.mp4 </tmp/omxplayer
/tmp/omxplayer is a fifo that is created by my program.
Can anyone shed any light on the matter? Is there any more info you need?
Thanks for your help in advance.
Cheers,
Chris
The text was updated successfully, but these errors were encountered: