Re: 2.6.33-rc3 -- INFO: possible recursive locking -- (s_active){++++.+}, at: [<c10d2941>] sysfs_hash_and_remove+0x3d/0x4f

[ebiederm@xmission.com (Eric W. Biederman)]

Dave Chinner <david@fromorbit.com> writes:

> On Mon, Jan 11, 2010 at 04:32:31PM -0800, Eric W. Biederman wrote:
>> Tejun Heo <tj@kernel.org> writes:
>> 
>> > On 01/11/2010 11:26 AM, Eric W. Biederman wrote:
>> >> From: Eric W. Biederman <ebiederm@xmission.com>
>> >> Date: Sun, 10 Jan 2010 18:13:35 -0800
>> >> Subject: [PATCH] sysfs: Add support for lockdep subclasses to s_active
>> >> 
>> >> We have apparently valid cases where the code for a sysfs attribute
>> >> removes other sysfs attributes.  Without support for subclasses
>> >> lockdep flags a possible recursive lock problem as it figures
>> >> the first sysfs attribute could be attempting to remove itself.
>> >> 
>> >> By adding support for sysfs subclasses we can teach lockdep to
>> >> distinguish between different types of sysfs attributes and not
>> >> get confused.
>> >> 
>> >> Signed-off-by: Eric W. Biederman <ebiederm@xmission.com>
>> >
>> > Acked-by: Tejun Heo <tj@kernel.org>
>> 
>> Now if I can just get a Tested-by  this patch will be all set ;)
>
> Hi Eric,
>
> Is this the same locking problem that this patch fixes?
>
> http://lkml.org/lkml/2010/1/11/26

It certainly looks similar.   The sysfs file that is being written
is different so I can't tell if that is a false positive because
sysfs is used for everything, or if it is a real issue.

Eric