[SOLVED] Random Player Unresponsiveness

[Kernle 32DLL]

Hi ho,

Since a few days I experienced some situations, where rockbox enters a strange state of delicate unresponsiveness. This happens while a music track is running. The unresponsiveness has so far appeared when I was browsing files, looking at the "playing now" screen, and even when the player was locked and I did no input at all - so I can assume it's an overall problem.

The unresponsiveness shows itself in such, that the current playback stops (without any symbol indicating so). I can navigate through all menus with no probs, even select other tracks, but they won’t start (or at least "hang" at 00:00). As such, I have to restart my player. If I'm unlucky, that problem may appear again just some seconds after I restarted the player and chosen a new track to listen to. The lower bar, which normally shows which track is played next, shows a constant "loading..." by the way.

There were also numerous times, that under the same circumstances I got a slightly different outcome. Instead of the track just "pausing", I got a terrible loud electrical noise. Fun fact: I could start and stop that noise as if I was listening to music, but as the playback bar won’t move, I don't guess the actual track is getting played anymore.

I personally think, those are the same errors, only that in case one I don't hear the noise. Without much knowledge of rockbox, i would say this error is related to the function that "caches" which track to play next (aka "that lower bar"). I use the Ultimate skin by the way.

I hope you can help me track the error, since it is REALLY nasty if this happens, and you are standing in the subway, people all around you, and you just can't reach down to your mp3 player to either restart it, or just to turn off that terrible noise. Not nice indeed.

So long,
Kernle

PS: I use the Vibe 500 port

[soap]

I'm not 100% sure I understand your question fully.

This appears to have started a few days ago, I get that.  Did it start after installing a current build, or had you not upgraded recently prior to the onset of symptoms?

[Kernle 32DLL]

Well, the problem started a few days after I installed rockbox (current build). But as this "problem" occurs very random, I can assume that the problem was there from the beginning.

I managed to get some more information about the problems. Both seem buffer related, as I got some way strange PCM values in the buffering thread screen. The "silence" problem, where the music just stops, gives me a on the left side of the "/" a value that is far more greater than the value on the right side. Also, this value stays till I restart the player. The "noise" problem, essentially the same as the silence problem but with terrible noise instead of silence, gives me a very great, negative value on the left side.

Both problems have in common that if I jump to a different track after the problem started, rockbox seems to fail to get the IDv? data, as I only get a bunch of "?" instead of title information, bitrate, etc.

Sadly i don't have the means to create a screenshot or anything else more useful (unless rockbox has a build-in way to do this).

Hope you can help me

So long,
Kernle

[Chronon]

Rockbox method is:
System > Debug (Keep out!) > Screendump

Then insert the USB cable to grab whatever is on the screen.

[Kernle 32DLL]

Gotcha, I'll provide you with some pictures then as soon as I can reproduce the error somehow when I'm at home.

Edit:
Happend during today's train ride. Luckily I had my Laptop with me, and as such was able to take some dumps. I dumped everything I fought useful. You can grab the archive here. Be aware that this archive is on my private webspace, and as such will only be uploaded for several weeks. So, if this problem points out to be a long-term problem, please make a copy to your own PC or webspace.

The second picture surely will be the most useful I think, as it clearly shows the incorrect PCM values. A note about the first picture: By "lower bar" I mean that bar that reads "Afternoons and Coffespoons"in the picture, just that you get what i meant earlier.

[Kernle 32DLL]

I think my mp3-player is evil. While the odds for the "silence" problem increased, the "noise" problem nearly vanished (didn't had it for days). Strange :/

[Kernle 32DLL]

I guess triple posts are not very nice, but the same problem is still consiting. Only thing I managed to find out so far, was the fact that setting the "Anti-Skip Buffer" to maximum (15 minutes I think) causes the silence problem to occur at nearly every start. I have to restart the player 3-5 times, till I am finally able to listen to my music. Of course I tried to set the Anti-Skip Buffer to it's lessest value, too (5 secs i guess), but it didn't change a lot.

I persist with my idea of a PCM lockup. Is there any way to provoke, or at least trigger a complete PCM buffer reset in rockbox? That would at least help me having to restart my player that often (or at the very least prove my idea of the pcm lockup wrong).

Any ideas?

So long,
Kernle

PS: The download link for the dump still works.

[Chronon]

The lack of corroboration from other users suggests one of a couple of possibilities:
1) The problems you are seeing are unique to your setup and have to do with some peculiarity of your player (file system damage, hardware problems, etc.).
2) The Rockbox build for Vibe 500 is exhibiting problems that are not apparent on other players (a bug in a driver, perhaps).

Are these problems persisting across multiple builds?  Have you checked your file system for damage?  (A damaged file system can manifest in many strange ways.) 

It would be helpful if other users of this player can corroborate this behavior.

[b0hoon]

It's strange. I've never experienced this problem. I'll try to walk with my vibe everywhere and reproduce it, but now it works without any problems. I have some version after change to eabi.

What version of Rockbox do you have?

What format of music do you use. It may be specific problem with the codec. Are sure that your files (or file system) are ok?

[Kernle 32DLL]

Yep, my files are 100% okay. I use standard MP3 files, with varying bitrates from 128 to 320. This problem occurs with static and variable bitrates, alike.

I use the 3.6 build (r27293-100705)

I find it really disturbing which values the pcm buffer takes. I would like to stick to that, so what could cause the pcm to say.. crash? And what would it take to recover it? One intersting thing is, as I mentioned, that i really have to restart the player to continue listening to my music. Selecting another track does not accomplish anything (intersting thing to know is, when i change to a different track, that all the ID Tag Info is shown as 's, as if the information couldn't be loaded).

Several times, but not always, I get a "playlist control file invalid" message after restarting the player. This may or may not have something to do with it, just mentioning it. Another thing I noticed is if the player has crashed as I mentioned it, it takes unusually long to shut down the player (around 2-3 times longer).

So long,
Kernle

PS: b0hoon, may sending my settings file help you?
PPS: I think my file system is okay, but just to be sure, I gonna reformat my player this weekend.

[Chronon]

You don't need to format.  Running a program like chkdsk or fsck.vfat (Linux) will detect problems and in most cases can correct problems it finds.

[b0hoon]

Tested and molested through the last four days and i can't reproduce this  (r28078-100914). The Ultima theme is a big mess, but it's probably not updated to the new wps grammar.

(a bug in a driver, perhaps).

Yes. It's possible, but when i was porting Rockbox to this player i only added define to an existing driver for audio - that's all. So it should affect other players (Mrobe 100, Irivers H10, Gogears HDD) the same way.

Kernle, i know that you modified your player with a new battery and had to dismantle it. Are you sure this adhesive blocks which are used to ground some parts of the casing, etc. are in the right place and don't touch some parts of the mainboard where they shouldn't?

Several times, but not always, I get a "playlist control file invalid" message after restarting the player. This may or may not have something to do with it, just mentioning it.

I think it's not related. I get it on my vibe and my ipod video sometimes too.

PS: b0hoon, may sending my settings file help you?

Yes, i can do some tests with your settings, sure. It will be more accurate.
Maybe you should try the current build, but please do backup of your existing version just in case (zip).

[Chronon]

(a bug in a driver, perhaps).

Yes. It's possible, but when i was porting Rockbox to this player i only added define to an existing driver for audio - that's all. So it should affect other players (Mrobe 100, Irivers H10, Gogears HDD) the same way.

Okay.  I didn't assign a very high probability to this possibility in my own assessments.  I was just trying to present various possibilities that could account for the observed behaviors.  Given this information as well as your inability to reproduce this behavior it appears more likely that it depends on some specific property of Kernle's setup (such as the possibility of a grounding problem that you raised).

[Kernle 32DLL]

Well, I think you may now officially slay me for not having tested that out earlier, but I updated my vibe just before last weekend (r28092-100916 now), and the problem has not occoured once since then. Sorry for making such a big thing out of it, I really should have tested that in the beginning.

So long,
Kernle

[b0hoon]

Glad to hear that!

I'm still a bit curious what was wrong with r27293 and when it returned to normal.

PS: Kernle: and i'm still waiting for your report about the new battery in the other thread.