After updating to 4.7.0 I get distorted video files #1820
Replies: 9 comments 11 replies
-
|
This looks like some bytes come in wrong order into extpipe. And your found error may be related Few questions:
|
Beta Was this translation helpful? Give feedback.
-
|
Good afternoon! Thanks for the answer!
|
Beta Was this translation helpful? Give feedback.
-
|
And how comfortable are you with compiling motion from sources? |
Beta Was this translation helpful? Give feedback.
-
|
See point 0 of the previous answer for the answer to this question. And this happens very quickly on fairly old hardware. After all, the program is very compact. |
Beta Was this translation helpful? Give feedback.
-
|
Here is a patch that will show how many bytes actually has been written into the extpipe: Please reproduce the issue and collect the full log again, this time it will be bigger, so maybe upload it to gist |
Beta Was this translation helpful? Give feedback.
-
|
I'll do this in the evening, when everyone leaves me alone :) |
Beta Was this translation helpful? Give feedback.
-
|
In general, I wrote a line from the documentation in movie_extpipe and not a single failure occurred on any of the cameras, everything works great! Now I have changed the format to matroska, leaving the codec and everything else, I observe... There are no errors in the logs either. I'll see how it behaves with matroska. Current line: My old line: |
Beta Was this translation helpful? Give feedback.
-
|
In general, I tested with this line: And everything works perfectly, on all cameras! I applied your patch and returned the old my line to libx265 and now I'm watching (I'm waiting for statistics when they make a few videos). |
Beta Was this translation helpful? Give feedback.
-
|
I've only managed to catch it once so far. With the patch it works much more stable, apparently due to delays in writing to the log (my guess). |
Beta Was this translation helpful? Give feedback.
-
|
Ok, can you try to revert this commit 629b3ba, build and try again? |
Beta Was this translation helpful? Give feedback.
-
|
I'm not good at big programming, but did I understand correctly that I should have added the red highlight to this function? It wasn't there, was it? In general, I added and rebuilt, now we are waiting for the result... |
Beta Was this translation helpful? Give feedback.
-
|
Ah, sorry, I should have been more specific, this command would do the thing:
|
Beta Was this translation helpful? Give feedback.
-
|
Forgive my ignorance, but I couldn't do it. Can I have detailed instructions for fools or just this file that I will replace? |
Beta Was this translation helpful? Give feedback.
-
|
To revert the commit, you have to restore deleted lines from that file. Either manually, or using git. Whatever is best for you |
Beta Was this translation helpful? Give feedback.
-
|
Thanks for the clarification! I did it the old fashioned way, made a patch and applied it. I cleared the log, now I wait and watch. I'll finish assembling the stand over the weekend and test it with one camera. One more thing, after applying the first patch that expanded the details in the log, crashes occurred only on the second camera, where there is the largest human traffic. Maybe it's a coincidence, maybe not. I'll check it at the stand later. |
Beta Was this translation helpful? Give feedback.
-
|
Hello! Looks like you managed to find and solve the problem! Thank you! For yesterday I recorded: Everything without a hitch! Today at 7 am I changed the log level from 8 to 5 (which I use daily) for the purity of the experiment, a total of 15 videos have been recorded so far and there are no failures. As for the stand with one camera, the release build (without patches) fails, with a patch for extended logging a little less frequently. There are no glitches with the latest patch! Dear tosiara, during experiments Motion sometimes failed with a segmentation fault. This happens randomly and as far as I remember version 4.6.0 behaved the same way, I just didn’t have time to report then. The problem only appears when the program is shut down and there is nothing in the logs, only in dmesg. In this regard, I would like to ask what actions should I take to create a report? |
Beta Was this translation helpful? Give feedback.
-
|
@Mr-Dave could you please take a look at this issue? Some edge case related to signals and system interrupts |
Beta Was this translation helpful? Give feedback.
-
|
Instead of reverting that commit, it may be as simple as commenting out this line. I think the zombies would only be for child processes. |
Beta Was this translation helpful? Give feedback.
-
|
Hello! I did as you said, on the very first video the frame again was disrupted and there was a 'green'. Those remained as it was. |
Beta Was this translation helpful? Give feedback.
-
Hello team!
After updating from 4.6.0 to 4.7.0, sometimes (quite often) distortions appear in the recorded files, it looks like a frame is broken. This happens randomly on any of the three cameras. In version 4.6.0 everything worked perfectly! For now I returned version 4.6.0. Photos are saved normally, regardless of the version. Please tell me what should I do? Should I create a bug report?
My system:
Calculate Linux 23
Kernel 6.6.32-calculate
v4l2 capture card based on cx25821
Driver cx25821 native from kernel.
And 3 analog b/w cameras 720x576
Normally frame:
Distorted frame:
The error I found was:
Oct 8 07:05:01 box motion[7041]: [2:ml2] [ERR] [EVT] event_extpipe_put: Error writing in pipe , state error 1: System call interruptedLog:
Beta Was this translation helpful? Give feedback.
All reactions