playback crashes on 5g (60GB) ipod

[argent]

So I just got my ipod refurbished.  Got it home, and, of course, having a practically new ipod, the first thing I needed to do was install rockbox, which I had been using before, but uninstalled.  (actually, not even the music I put on the apple OS was on it, so it might be an entirely new hard drive).  

So I downloaded ipodpatcher, ran it at the command prompt, and hit i to install the bootloader, peachy-keen.  Proceeded to download the latest build, and unzip it to the root of my ipod (using 7zip for windows).  Add some music, eject, reboot, and I'm running rockbox.  All's well so far.  So I go to start playing the music (in mp3; without changing any of the default settings), and it plays fine for about 10 seconds, then crashes.  (Data abort at some hex, prefetch abort at some hex, Unknown instruction at some hex, etc.  Various errors result, at various hexes)

Okay, maybe it's the default theme: that's had problems before with skippy playback.  So I try changing the theme (iCatcher and UniCatcher), and it still crashes, albeit about 30 seconds into the song now.  I go download the "dark" theme, and it still crashes.  So that probably didn't have anything to do with it.  

Disk mode, and connect.  Maybe it's the latest daily build's fault.  So I delete the .rockbox folder, and unzip the oldest archived build (05-30-2007) to my ipod.  Reboot, same problem.  

Well, I guess that since my brother's ipod (same model as mine) works fine with a (admittedly rather old; he refuses to update his; it's probably from sometime last year) rockbox daily, I grab his ipod, remove my .rockbox, copy his to mine, and reboot.  It works!  But I'm stubborn or stupid and don't want to be stuck with a build from last year.  So I try updating, again from the latest daily build.  And I make a mistake.  I started to copy it onto my brother's ipod first.  Crappity.  Delete his .rockbox, copy mine to his (which I just copied from his, so it shouldn't have anything wrong with it, more on that later), and then proceed to copy the update where I wanted it: my ipod.  Reboot, and it crashes, of course.  

Okay, delete my .rockbox, copy his to mine again.  Reboot.  Codec error.  Uh-oh.  
Reboot his.  Codec error.  Uh-oh.  Interesting, though, too, this practically proves there's either something wrong with my ipod, or one of my cables: something corrupted the data, otherwise he wouldn't be getting an error where he wasn't before.  
At the moment, all my music is in mp3 (most of it I deleted to make room for other stuff) so that's the only codec I can try.  I really regret not deleting the codecs here and at least trying patching them from another build.  But I didn't, I just updated his to the oldest archived daily (05-30-2007), and reboot, and now he gets my crashes.  Oops.  I try the second-to-last daily (fresh install), and it still crashes.  

Well, I reboot his to the Apple OS, and proceed to add some tunes to it with sharepod, so I don't feel as bad.  Now that I think about it, I do the same to mine.  Will the Apple OS crash?  Nope.  I haven't tried watching videos or playing games on it yet, though.  Hell, there's nothing on mine, why not download itunes, do a restore, and try a fresh install of rockbox?  So I do, and this time I change the loader to loader2.  It still crashes a few seconds into playback.  Fine, I'll do another restore, and settle for the Apple OS for the time being.  It's about 1 a.m. when I'm doing this, anyway, so I was tired as hell.  

There's a bunch of un-backed-up information on my brother's though, and, long story short, I won't be able to back it up until monday.  Once I do, I'm fairly certain an itunes restore and fresh rockbox will clean his up beautifully (and I should have backups of some fairly old builds, too, so I can try those).  And I'll have another cable, so if the cable is the problem, that should fix it.  

Things I should probably try but haven't yet:  
copy rockbox to my ipod with both cables, and verify data integrity by hashing it or something.  (anybody know a good program for doing this?)
do that, and reboot the ipod before checking for integrity.  (if the first test comes out all right)
fresh format & install on my brother's ipod.  (I'll have to wait until monday, like I said)

So at the moment, it looks like either one of my two cables is bad, or my ipod is bad.  Which would suck, because it doesn't seem to want to crash under the Apple OS, and that means I might actually have to pay to get the damn thing repaired this time, as opposed to getting it fixed under warrantee.  

[Febs]

So I downloaded ipodpatcher, ran it at the command prompt, and hit i to install the bootloader, peachy-keen.  Proceeded to download the latest build, and unzip it to the root of my ipod (using 7zip for windows).  Add some music, eject, reboot, and I'm running rockbox.  All's well so far.  So I go to start playing the music (in mp3; without changing any of the default settings), and it plays fine for about 10 seconds, then crashes.  (Data abort at some hex, prefetch abort at some hex, Unknown instruction at some hex, etc.  Various errors result, at various hexes)

Double check to make sure that you have downloaded the most up-to-date bootloader, and then download and re-install a current build.

You may want to try using the Rockbox Utility, as it will perform both of these steps for you automatically:  http://www.rockbox.org/twiki/bin/view/Main/RockboxUtility

[argent]

Cool, very handy little utility.  I used it to install the bootloader first (as I no longer have a need of Loader2, I never used ipodlinux and decided not to install it on this one).

Good, or maybe bad news.  

It works, with the 30gb daily build (actually, that's what the rockbox utility downloads, too; all 5gs are lumped together).  So I'm thinking "wtf?" about now.  The only difference I can think of between the two (besides HD capacity) would be the amount of RAM.  Any chance that's the problem?

So I'm pretty happy with it right now, and if it's going to become too big a pain, I'm happy with it where it's at.  Of course, not all is resolved completely: I still have no idea why the 60gb build wouldn't work.  And, as that goes, what could be done to change that.  

And, of course, thank you very much for the help.  

[Chronon]

The only difference between them is that the 30GB build utilizes 32MB of RAM, not the full 64MB that the 60 (and 80) GB models have.  

I would try downloading the 64MB version again now that you have the most recent bootloader running.

[argent]

I have, and it crashes, and, to be quite honest, that irks me.  Both with and without deleting the old .rockbox folder beforehand.  

Without deleting it, though, it played through one 5-minute song that I'd listened to before, then crashed 14 seconds into the second, which I had not.  Odd.  

I can live with the 30GB version, and would naturally prefer the 60/80, but hey.  At this point it looks like it's my ipod's fault, so I don't have much a choice, anyway.  

I'm still curious about something.  I did get the thing refurbished, which was described to me as (in this case) a hard-drive transplant.  Any chance I've got a 60GB in a 30GB's body?  Any records of a 30GB person who's been trying to run the 60GB build?  Because if they're similar symptoms... (I tried a forum search, but only got one unresolved topic).

[Febs]

Do you mean a 30 GB in a 60 GB's body?  It sounds like that is entirely possible.

[argent]

A 60GB hard drive, with 32MB of RAM.  If that's the case there isn't really anything I can do about it anyway.  

So the question would be, how can I check the amount of RAM in it?  I'm not entirely opposed to the option of opening it up, but I'd need to know what to look for.  Still, a software solution would obviously be best.  

If nothing comes to mind, though, it's cool.  Whether or not I actually have 32MB of RAM, it's for all practical purposes 32MB, so it's really academic at this point.  

[David Marsh]

yes i have this problem.  it think it because the cpu is now running at 78Mhz and not 75Mhz.

I was running a build from 22 Jun 2007 and installed 3 July 2007.  then it started crashing quite soon into a track.  i might be lucky and get 3 tracks before it crashed.  previously it didnt crash at all.

I have reverted back to 22 jun 2007 build and the CPU freq is back to 75Mhz.  You can check this in the debug menu and select 'CPU frequency' option.  You can also see this in 'view audio thread' during which on the 3 july build it crashed with:

Code: [Select]

Data abort 
at 00030784 

Is the cpu meant to be runing faster?  is it just reporting an incorrect value?  

[soap]

The speed change was intentional and part of changes to the PP5020 code made to correct the crashing on iPod 4th gen / Mini 1st gen targets.

[David Marsh]

Thats all very well but if it causes 5g ipods to fail then it needs to be backed out.  I am currently running 1-july-2007 build and that seems to be holding up quite well right now.  This was the build before any CPU changes were made.  I will roll forward each daily build until the symptoms reappear then we can narrow down the exact cause, and hopefully find a solution.  The builds for the 4th & 5th were causing crashing, so effectively I only have to install 2nd and 3rd july daily builds to find the one that causing the problem.

[Llorean]

Actually, if it causes the 5G iPods to fail, then there's probably something else wrong with the code. The 78mhz thing is a result of doing things more properly. As well, it causes several other iPod models to stop failing. So more needs to be investigated either way.

But I don't see that you've determined it was that change that caused it, just guessed wildly after a multiple day range of changes caused only your iPod to stop working, but not many other peoples. Considering the first post in this thread happened June 29, it clearly wasn't the same cause for them too.

[David Marsh]

Sorry I had overlooked the original date of this post.  I was trying not to post again about a problem that had potentially already been reported.  You are quite right to say that this may not be a symptom of the CPU changes but a manifestation of a problem elsewhere within the source, highlighted by these changes.  All I can really do is identify in which build the problem first occurred so that we can get to the bottom of it.  Iw as unaware that the 78MHz change had taken place.  I usually download the latest build everytime I sync my iPod with the music folders on my PC (its a shell script I wrote as I'm running linux and no iTunes), which could be every day or every few weeks hence why my earlier build was from 22 June.

The build for 1st July seems to be ok.  My iPod ran for over 3 hours without any crashes (whereas before it would crash within about 3 tracks).  I am running 2nd July build now. I will let you know how it goes.

If I find which build first causes me problems, is there a way to enable some sort of stack trace to  work out where the crash occurred, so I can feed that back to help whoever can write a fix?

[robin0800]

Discussion in irc with soap last two hours if interested.

[amiconn]

My 2 commits from July 2nd and 3rd regarding the PP5020 and PP5022 clock setup were done in order to do things more properly in general.

They stop the PP5020 models (iPod 4G, iPod color/photo, iPod Mini G1, Iriver H10) crashing on clock change. Unfortunately they also caused some of the PP5022 targets (iPod Video, iPod nano, iPod Mini G2) to become unstable. I found the cause and committed a fix today.

I also changed the max. CPU clock on PP5020 and PP5022 to 78MHz. This is unrelated to the stability issues, but desired because it provides a bit more CPU power when needed. Battery life is unaffected as long as the CPU isn't 100% boosted (which means it would not have been able to cope at 75MHz). I will probably go for 80MHz soon.

[David Marsh]

Nicely sorted there amiconn.   Thanks for the prompt fix.  I'm a happy bunny again