Re: [Nbd] [PATCH 2/7] block: loop: don't hold lo_ctl_mutex in lo_open

[Ming Lei <ming.lei@...1301...>]

On Wed, Apr 8, 2015 at 10:00 PM, Jarod Wilson <jarod@...696...> wrote:
> On Wed, Apr 08, 2015 at 09:40:34AM -0400, Jarod Wilson wrote:
>> On Wed, Apr 08, 2015 at 02:50:59PM +0800, Ming Lei wrote:
>> > Hi Jarod,
>> >
>> > On Wed, Apr 8, 2015 at 2:23 PM, Jarod Wilson <jarod@...696...> wrote:
>> > > From: Ming Lei <ming.lei@...1301...>
>> > >
>> > > The lo_ctl_mutex is held for running all ioctl handlers, and
>> > > in some ioctl handlers, ioctl_by_bdev(BLKRRPART) is called for
>> > > rereading partitions, which requires bd_mutex.
>> > >
>> > > So it is easy to cause failure because trylock(bd_mutex) may
>> > > fail inside blkdev_reread_part(), and follows the lock context:
>> > >
>> > > blkid or other application:
>> > >         ->open()
>> > >                 ->mutex_lock(bd_mutex)
>> > >                 ->lo_open()
>> > >                         ->mutex_lock(lo_ctl_mutex)
>> > >
>> > > losetup(set fd ioctl):
>> > >         ->mutex_lock(lo_ctl_mutex)
>> > >         ->ioctl_by_bdev(BLKRRPART)
>> > >                 ->trylock(bd_mutex)
>> > >
>> > > This patch trys to eliminate the ABBA lock dependency by removing
>> > > lo_ctl_mutext in lo_open() with the following approach:
>> > >
>> > > 1) introduce lo_open_mutex to protect lo_refcnt and avoid acquiring
>> > > lo_ctl_mutex in lo_open():
>> >
>> > It is a bit quick since I said the lo_open_mutex can be removed,
>> > and Christoph agreed that too.
>> >
>> > So looks we still need to post another version, :-)
>>
>> Ah. I missed that bit. Just trying to keep up momentum.
>
> Are you working on that atomic_t version then, or should I dive into it?

It has been ready, and I will post them out once it passes my tests.

Thanks,