[Lustre-discuss] BUG: using smp_processor_id() in preemptible [00000000] code: modprobe/6024

[Andreas Dilger]

On 2010-03-22, at 05:06, Gregory Matthews wrote:
> after running racer over the weekend on some test clients I see a  
> lot of these, is this lustre or cat? There are lustre functions  
> listed in the stack trace.

It definitely wouldn't have anything to do with cat itself.   
Unfortunately, I can't see anywhere in that call stack where we are  
scheduling while atomic.  It appears (from what I can make of the  
stack trace) we are in osc_queue_async_io() and the only place we grab  
a spinlock is in a very isolated piece of code.

> BUG: scheduling while atomic: cat/4439/0x00000002
>  Call Trace:
>   [<ffffffff80422c6a>] schedule+0xf7/0x7bd
>   [<ffffffff80424c84>] _spin_unlock+0x10/0x2b
>   [<ffffffffa062e08b>] osc_queue_async_io+0x63b/0x10c0 [osc]
>   [<ffffffffa06add05>] lov_queue_async_io+0x165/0x4b0 [lov]
>   [<ffffffffa073d2e3>] ll_readpage+0xd63/0x1f60 [lustre]
>   [<ffffffff8026b087>] generic_file_aio_read+0x3c9/0x551
>   [<ffffffffa070d1fa>] ll_file_aio_read+0xf1a/0x2350 [lustre]
>   [<ffffffffa0712e79>] ll_file_read+0xb9/0xd0 [lustre]
>   [<ffffffff80294c33>] vfs_read+0xaa/0x133
>   [<ffffffff80294f18>] sys_read+0x45/0x6e


Cheers, Andreas
--
Andreas Dilger
Sr. Staff Engineer, Lustre Group
Sun Microsystems of Canada, Inc.