Skip to content
This repository was archived by the owner on Jun 1, 2022. It is now read-only.

PMP Crashing when seeking #201

Closed
KrunchMuffin opened this issue Jan 10, 2016 · 15 comments
Closed

PMP Crashing when seeking #201

KrunchMuffin opened this issue Jan 10, 2016 · 15 comments

Comments

@KrunchMuffin
Copy link

v1.0.3
Raspberry Pi2

When I play new show and want to skip the intros, I will skip fwd a few times by 30s. sometimes the UI just locks up and I have to restart RPi2. I have attached the log file when it happened and before the restart.

plexmediaplayer.log.txt

@ghost
Copy link

ghost commented Jan 11, 2016

How often can you reproduce it?

@KrunchMuffin
Copy link
Author

Happens quite frequently, but it's random.

@ghost
Copy link

ghost commented Jan 13, 2016

I couldn't reproduce it yet. Will try some more.

@KrunchMuffin
Copy link
Author

Is it possibly memory leak issue? I never turn it off unless forced to and
seems to not happen for a bit after the reboot.
On Jan 13, 2016 2:57 PM, "V. Lang" [email protected] wrote:

I couldn't reproduce it yet. Will try some more.


Reply to this email directly or view it on GitHub
#201 (comment)
.

@ghost
Copy link

ghost commented Jan 13, 2016

Interesting angle. I'll try to let it run for a while.

@KrunchMuffin
Copy link
Author

happened again
plexmediaplayer.log.txt

@ghost
Copy link

ghost commented Jan 15, 2016

I've let it running and playing all day. I occasionally seeked back. Except for some mysterious sporadic failures where it showed a "playback did not start in time" error, all seemed well.

Any other circumstances I should be aware of?

(I've been testing with a 1.0.4 development version - I hope it's not just a bug that was fixed since 1.0.3, but I'm not aware of any and it's unlikely.)

@KrunchMuffin
Copy link
Author

I don't think so, other than I usually start clicking as soon as the show starts but after the info bar at the bottom disappears, so not sure if it would be a buffering thing. That's about all I can think of. I will say it was worse in 1.0.2 and got a little better in 1.0.3.

@OttoKerner
Copy link

I happened about some weirdness with seeking myself.
But it was always a movie with a relatively high bandwidth, so the server showed heavy disk activity when I clicked on 'jump forward' a few times in quick succession.
Apparently, sometimes the server can't deliver the data for the last jump fast enough so a 'time out' is probably triggered.

Part of this problem is the lack of a means to jump further than 30s, so it is almost 'natural' for a user to quickly repeat the 'jump forward' action.

@KrunchMuffin
Copy link
Author

for sure. I am binge watching shameless and I know I have to click 5x to get to the new show. 2:30 mark. Even worse with Homeland, their credits alone are 2:30.

@KrunchMuffin
Copy link
Author

Just happened again. Been watching for about 3hrs. I hit fwd about 5x, than back 3x.

I see this, not sure what it means tho.

2016-01-16 02:25:00 [ ERROR ] PlayerComponent.cpp @ 414 - ffmpeg/video: h264_mmal: Did not get output frame from MMAL.
2016-01-16 02:25:00 [ WARN ] PlayerComponent.cpp @ 412 - vd: Error while decoding frame!

I found this
https://ffmpeg.org/pipermail/ffmpeg-cvslog/2015-August/092698.html

and this
raspberrypi/firmware#496

plexmediaplayer.log.txt

@ghost
Copy link

ghost commented Jan 16, 2016

It means just that what it says: the hardware decoder did not return a frame in time. It waits with a timeout of 100ms. The reasoning is that if fed enough input data, the GPU will usually be much faster than 100ms to return a decoded frame. This could lead to race conditions, but I feel like this is the only way to handle this.

The fundamental problem is that we do not know whether the GPU is busy decoding a frame, or whether the frame was discarded due to a decoding error. Using a timeout is still better than freezing forever. In any case, it should not crash hard, even if playback fails in some way.

It's also possible that this problem is triggered by many seeks in succession (which would reset + resume the decoder, exposing potential problems). But so far I haven't been able to reproduce it even when heavily seeking around. It might be timing related, e.g. related to network speed etc. Hard to tell.

This reminds me: we actually have a crash reporting system. Can you look at the pmphelper.log and tell us which crash dumps it uploaded, if any at all?

@KrunchMuffin
Copy link
Author

Here is that log. not sure if it means one was sent or not.
pmphelper.log.txt

@KrunchMuffin
Copy link
Author

So I have 1.0.4 now and guess what happened the first time I played and tried to seek? Although the UI was still responsive and the little waiting indicator appeared, but I had to press stop and restart the show, then it happened again. frustrating.

@LongChair
Copy link
Contributor

Supposedly fixed by now.

Sign up for free to subscribe to this conversation on GitHub. Already have an account? Sign in.
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants