Re: Deadlocks due to per-process plugging

[Mike Galbraith <efault@gmx.de>]

On Mon, 2012-07-16 at 13:24 +0200, Mike Galbraith wrote:

> Box disagrees.
> 
> Waiting for device /dev/cciss/c0d0p6 to appear:  ok
> fsck from util-linux 2.19.1
> [/sbin/fsck.ext3 (1) -- /] fsck.ext3 -a /dev/cciss/c0d0p6 
> SLES11: clean, 141001/305824 files, 1053733/1222940 blocks
> fsck succeeded. Mounting root device read-write.
> Mounting root /dev/cciss/c0d0p6
> mount -o rw,acl,user_xattr -t [   27.558934] jbd: disagrees about
> version of symbol rt_spin_trylock
> ext3 /dev/cciss/[   27.558938] jbd: Unknown symbol rt_spin_trylock (err
> -22)
> c0d0p6 /root
> [   27.558959] jbd: disagrees about version of symbol
> __rt_spin_lock_init
> [   27.558961] jbd: Unknown symbol __rt_spin_lock_init (err -22)
> [   27.558986] jbd: disagrees about version of symbol rt_spin_unlock
> [   27.558988] jbd: Unknown symbol rt_spin_unlock (err -22)
> [   27.558990] jbd: disagrees about version of symbol rt_spin_lock
> [   27.558992] jbd: Unknown symbol rt_spin_lock (err -22)
> [   27.559004] jbd: disagrees about version of symbol __rt_mutex_init
> [   27.559006] jbd: Unknown symbol __rt_mutex_init (err -22)
> modprobe: FATAL: Error inserting ext3
> (/lib/modules/3.0.35-rt56-rt/extra/ext3.ko): Invalid argument
> 
> mount: unknown filesystem type 'ext3'
> could not mount root filesystem -- exiting to /bin/sh
> $

Weird.  Make mrproper didn't help.  Whacking every trace that the damn
tree, kernel and modules ever existed did.  Box is grinding away at ext3
dbench testcase.  At 7 hours per run, it'll take a while.  This time
tomorrow box should still be happily grinding away.

-Mike