Re: 2.6.16.32 stuck in generic_file_aio_write()
From: Igmar Palsenberg
Date: Thu Dec 07 2006 - 04:59:30 EST
> I thought it was, but from my look through yout 8-billion-task backtrace,
> no task was stuck in D-state with the appropriate call trace.
I was afraid of that... Where is the lock on the i_mutex suppose
to be released ? I can't grasp the codepath from within an interrupt back
to the fs layer.
> So I don't know what's causing this. In the first trace you have at least
> four D-state kjournalds and a lot of processes stuck on an i_mutex. I
> guess it's consistent with an IO system which is losing completion
> interrupts. AFAICT in the second trace all you have is a lot of processes
> stuck on i_mutex for no obvious reason - I don't know why that would
> happen.
Is there any way to see if it is missing interrupts ? Enabling the
debugging in the areca driver isn't a good idea on this machine, it's a
heavely IO loaded machine, and the problem seems to take some time to occur.
I *does* happen less often with a 2.6.19 kernel however.
The task dump takes > 10 seconds, which causes the softlock detector to
trigger. Is there any objection to a patch which disables the lockup
detector during the dump ? It isn't a big issue, since al it does is dump
a stacktrace.
I've enabled most debugging now, I'll see of i can run both a disk and VM
stresstest.
I'll put a .config and a dmesg of the machine booting at
http://www.jdi-ict.nl/plain/ for those who want to look at it.
Regards,
Igmar
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/