Messages in this thread | | | Date | Mon, 20 Aug 2012 17:58:55 +0200 | From | Oleg Nesterov <> | Subject | Re: lockdep trace from posix timers |
| |
On 08/20, Peter Zijlstra wrote: > > On Mon, 2012-08-20 at 17:43 +0200, Oleg Nesterov wrote: > > BTW, do you think it is really important to try to avoid ->pi_lock? > > It is for me, I'm doing task_work_add() from under rq->lock.. :/ I could > fudge around and delay, but not having to would be nice.
Aha, got it.
At first glance, bit_spin_lock() can work. I'll try to make the patches tomorrow.
Oleg.
| |