Re: 2.6.31rc5 RAID10 lockdep report - sysfs_nofity_dirent locking issue

[Dan Williams <dan.j.williams@intel.com>]

On Wed, Aug 5, 2009 at 10:43 PM, Neil Brown <neilb@suse.de> wrote:
> On Tuesday August 4, gregkh@suse.de wrote:
>> >
>> > Greg: does that look right?  If so I'll add a changelog entry and
>> > submit it properly (after at least a compile test...)
>>
>> Yes, it looks correct, if it passes your tests :)
>
> Thanks.
> I've tested it and documented it now and so am happy to submit it.
> Thanks,
> NeilBrown
>
>
> From: NeilBrown <neilb@suse.de>
> Date: Thu, 6 Aug 2009 15:42:08 +1000
> Subject: [PATCH] Allow sysfs_notify_dirent to be called from interrupt context.
>
> sysfs_notify_dirent is a simple atomic operation that can be used to
> alert user-space that new data can be read from a sysfs attribute.
>
> Unfortunately is cannot currently be called from non-process context
> because of it's use of spin_lock which is sometimes taken with
> interrupt enabled.
>
> So change all lockers of sysfs_open_dirent_lock to disable interrupts,
> thus making sysfs_notify_dirent safe to be called from non-process
> context (as drivers/md does in md_safemode_timeout).
>
> sysfs_get_open_dirent is (documented as being) only called from
> process context, so it uses spin_lock_irq.  Other places
> use spin_lock_irqsave.
>
> The usage for sysfs_notify_dirent in md_safemode_timeout was
> introduced in 2.6.28, so this patch is suitable for that and more
> recent kernels.
>
> Cc: stable@kernel.org
> Signed-off-by: NeilBrown <neilb@suse.de>
> ---

Greg?

Looks like this never made it upstream, and now Hans is hitting this
in his tests.

Thanks,
Dan