On Tue, 27 May 2008 10:10:39 -0300
Mauro Carvalho Chehab <mchehab@infradead.org> wrote:
So we're replacing the big kernel lock with the big v4l2 lock. That
might help the situation, but you'd need to be sure to serialize
against other calls (open(), for example) which are also currently done
under the BKL.
I don't think that stress tools are the way to eliminate the BKL.
You'll never find all the problems that way. There's really no way to
avoid the task of actually *looking* at each driver and ensuring that
it has its act together with regard to locking.
jon
--