Kernel 3.0.30 hang in xfs_submit_ioend_bio

[Dave Chinner]

On Mon, Apr 30, 2012 at 10:08:20AM +0200, Stefan Priebe wrote:
> Hi,
> 
> while running several systems with 3.0.30 i'm seeing this call trace
> pretty often. Is this problem known?
> 
> [925680.203973] Call Trace:
> [925680.204170]  [<c0610ce0>] schedule+0x30/0x50
> [925680.204221]  [<c0610d71>] io_schedule+0x71/0xb0
> [925680.204273]  [<c0362420>] get_request_wait+0xa0/0x150
> [925680.204376]  [<c03625c6>] __make_request+0xf6/0x280
> [925680.204427]  [<c036143d>] generic_make_request+0x3fd/0x610
> [925680.204741]  [<c03616ba>] submit_bio+0x6a/0x100
> [925680.204894]  [<c03366a7>] xfs_submit_ioend_bio+0x47/0x70
> [925680.204946]  [<c03367a4>] xfs_submit_ioend+0xd4/0xe0
> [925680.204998]  [<c0337ba3>] xfs_vm_writepage+0x253/0x580
> [925680.205109]  [<c01ae8ca>] shrink_page_list+0x55a/0x750
> [925680.205162]  [<c01aed60>] shrink_inactive_list+0x1a0/0x350
> [925680.205214]  [<c01af321>] shrink_zone+0x411/0x570
> [925680.205265]  [<c01af9fe>] kswapd+0x57e/0x9c0

kswapd is blocking submitting IO - the request queue is full. I'd
say you've got memory pressure, lots of dirty pages and an IO
subsystem that can't keep up....

Cheers,

Dave.
-- 
Dave Chinner
david at fromorbit.com