Just wanted to report. Stuck jiffies. No buttons work.
Music still playing. Runnning on AC power. b11


0: 0 empeg_state
1: 0 empeg_usbirq
2: 0 empeg_cs4231
3: 513918 empeg_rdsirq
4: 270 empeg_input
6: 3225 ide0
7: 550 SMC9194
17: 191 serial
20: 16552 audio-empeg
26: 1236 timer
FIQ: cs4231dma

0: 0 empeg_state
1: 0 empeg_usbirq
2: 0 empeg_cs4231
3: 527058 empeg_rdsirq
4: 270 empeg_input
6: 3225 ide0
7: 572 SMC9194
17: 191 serial
20: 16975 audio-empeg
26: 1236 timer
FIQ: cs4231dma


cpu 62 0 515 659
disk 1214 0 0 0
disk_rio 1210 0 0 0
disk_wio 4 0 0 0
disk_rblk 2420 0 0 0
disk_wblk 8 0 0 0
page 17294 327
swap 0 0
intr 587790 0 0 0 564105 270 0 3225 587 0 0 0 0 0 0 0 8 0 191 0 0 18168 0 0 0 0 0 1236 0 0 0 0 0
ctxt 54617
btime 1015076734
processes 69


cpu 62 0 515 659
disk 1214 0 0 0
disk_rio 1210 0 0 0
disk_wio 4 0 0 0
disk_rblk 2420 0 0 0
disk_wblk 8 0 0 0
page 17294 327
swap 0 0
intr 600707 0 0 0 576598 270 0 3225 609 0 0 0 0 0 0 0 8 0 191 0 0 18570 0 0 0 0 0 1236 0 0 0 0 0
ctxt 55851
btime 1015076734
processes 71


Marius (Escort Cab + Mark II)

Same here, happens every time I try to upload new music with emplode.

Display reports:

ERROR: Stuck jiffies.

Only way to fix, is power cycling the unit.

MK2a 30GB

-F

My error was while playing music, and it fixed itself after a little while (no more message on screen)

Marius (Escort Cab + Mark II)

Perfect. This is exactly the information needed.

The two successive samples you posted from /proc/interrupts show lots of system activity, but with the system timer stuck. Normally, the timer interrupt should increment 100 times per second, but your posting shows it sitting still with a count of 1236.. exactly as predicted, and now confirmed.

I'll release v223 (shortly) with the fix included (and the message removed, since they cannot happen with the fix in place). Please install, and report if the unit ever locks up again.

Cheers

Actually, I left the message in place, but it should NEVER be seen again.

You da man Mark!

Please retry ASAP with Hijack v223, and post whether it works or not for your uploads.

Thanks

I have got one of the newer versions (225?), and I haven't noticed any lockups. I left it playing overnight, and nothing
on the serial port or on the display... I guess that is good.

Marius (Escort Cab + Mark II)

Good. And v228 fixed the http lockups some folks ran into.

I'm still running 224...I was going to do the new HiJack today, but decided to do a small sync first... looking at my display, during a USB sync, I have the "Stuck Jiffies ####" with a counter going up and up.... Is this normal during a sync?

(kicking self for not doing a kernal upgrade first...)

Note... at the end of this sync... it failed to restart the player... could this be the "wibbly wobbly world?"

The "failed to restart player" is a different bug, I think.

I'm not sure how you managed to get "stuck jiffies" with v224. Are you sure about the version? Perhaps this is why so many people have trouble with USB..

???

Hmm.. it MAY have been 225, but I know it wasn't anything past or including 227. (don't know how to check and I just up'd to 230 anyway).

I remember the "failed to restart player" but was described by empeg as being becuase the player "went off into its own wibbly wobbly world" during the sync.. or something like that. Jiffies sound pretty wibbly to me.

Mmm.. I was thinking rather maybe v223 instead of v224.

See if you can do it again with any version >= v225.

Actually, the latest versions no longer even have the "stuck jiffies" message, so v225 would be best for this. There's no link for it, but it's still there on the server.

Thanks

I've downloaded it and am about to install it (225). I am planning on a lot of USB sync's today, so that will be a good test.

Update.. had another sync failure. This time with HiJack 225. No front panel display and no output to serial ...

Good. Could have been anything.

Keep us posted!

I remember the "failed to restart player" but was described by empeg as being becuase the player "went off into its own wibbly wobbly world" during the sync.. or something like that. Jiffies sound pretty wibbly to me.

The failed-to-restart-player bug (at least in the case we managed to reproduce) was nothing to do with jiffies. The jiffies thing kills serial input, whereas the player could be brought sharply back from the wibbly-wobbly world thing by sending it a ^C over serial. It's more like there was a subtle (or perhaps unsubtle) pthread_cancel problem; a workaround was applied to the code post-beta11 and will be in the next beta.

Peter

The really wicked thing about the jiffies failure, is that if it were to happen while some critical piece of software was in a loop that was dependent on the jiffie count, it would be stuck there for 19 minutes. Were this to happen in the right place, it could lock up the entire machine rather well..

But that's probably not what's going on here.

-ml

Yes, my player has locked up on boot with the jiffies problem before.

I don't know if this is related, but it might be, and I'd hate to withhold evidence.

The other day, I was transferring a ridiculously large file (a little over 220MB) from the empeg using the FTP daemon. It transferred it in a reasonable amount of time with no apparent pauses, but then, when I went back to play some more at the serial console, where the player had already been stopped (via `q') and I had been playing at the bash prompt for some time before the FTP started, there was no response to serial input. I let it sit there without doing anything more to it (as I was dealing with the file I had just downloaded and didn't feel like debugging the new problem yet), and it eventually came back. I didn't time it, but it seemed to be about 20 minutes.

Again, I don't see how this could be related, but the symptoms seem awfully similar. There was nothing going on on the empeg besides the bash prompt idling and the FTP daemon pushing all that data and whatever other processes exist all of the time.

(Edit: I should point out that I'm using an ancient version of HiJack -- v200.)

>I should point out that I'm using an ancient version of HiJack v200

Whew! Had me pulling my hair out for a few minutes there.

Yes, that would be the jiffies bug in v200.
Not in the latest versions.

The jiffies bug worked like this:

The clock interrupt handler gets invoked, signalling another "jiffie" (1/100sec) has passed. The code then begins to reset the "match register" for the next 1/100sec jiffie interval, but gets interrupted midway through by some other hardware interrupt.

If that second interrupt takes more than about .9/100sec to complete, then the original timer interrupt routine will mess up the reset when it eventually gets to run again.

The result is that the "match register" is already behind the actual counter, and will not match until the 32-bit counter wraps around (at 3.6MHz), which takes about 00:19:53 in elapsed time.

This probably won't happen so long as the second interrupt is handled very quickly. Things like ethernet and USB activity take longer to handle than other interrupts, and are more likely to trigger the bug.

And, yes, just having Hijack installed will increase the likelihood as well, since Hijack does most of it's processing during the button and display interrupt handling, lengthening those routines.

Finding the bug was hard, the fix was simple: mask ALL interrupts in the critical section of code which resets the match register in the timer interrupt handler.

Cheers.

Sorry. Didn't mean to frighten you. I just posted it because no one I saw had mentioned FTP and nothing else was going on on the machine at all. I should bother to upgrade (lazy).

Again, sorry.

I've just discovered that my mk2a will almost always experience a "stuck jiffies" moment shortly after I've run rsync to pull down a bunch of files onto it. It always finishes the job and is sat at a shell prompt before freezing though.

I'll be applying the patch, and hammering on it a bit to see if it's truly gone away. Unfortunately, I'm off to Italy for a week, so I'll not get much chance.

Hmmm.. I wouldn't normally call going to Italy "unfortunate" - at least not while living in Canada (or the UK for that matter )

Since having installed the latest versions I have been without a single jiffies-related hang of the interface. The last time I saw the problem was while running 223 - happened as soon as the player application finished loading (that was a first).

Everything seems cool right now except for the infrequent toggle of the dimmer setting for no apparent reason (not turn signals).

Bruno

Well, yeah, I meant unfortunate that I won't be able to test it, not unfortunate that I'm going to Italy .