Re: [PATCH] BLOCK: use disk_stat_add instead of __disk_stat_add in __end_that_request_first

[Jens Axboe <axboe@suse.de>]

On Sun, Jan 22 2006, Tejun Heo wrote:
> Unlike end_that_request_last, caller is supposed to own the request
> when it calls __end_that_request_first and thus allowed to call from
> any context without any locking.  When SCSI EH completes requests, it
> calls __end_that_request_first from kernel thread context without
> holding any lock and none of preemption/bh/irq disabled.  This results
> in the following warning.
> 
> BUG: using smp_processor_id() in preemptible [00000001] code: scsi_eh_6/927
> caller is __end_that_request_first+0xbf/0x500
>  [<c01046e3>] show_trace+0x13/0x20
>  [<c010470e>] dump_stack+0x1e/0x20
>  [<c02231c8>] debug_smp_processor_id+0xa8/0xb0
>  [<c021167f>] __end_that_request_first+0xbf/0x500
>  [<c0211ad1>] end_that_request_chunk+0x11/0x20
>  [<c03360b2>] scsi_end_request+0x32/0x110
> -- snip --
> 
> This patch makes __end_that_request_first() use undashed
> disk_stat_add() which doesn't assume preemption is disabled.

Patch is good, however I already added this identical patch to the block
git repo last week :-)

-- 
Jens Axboe

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/