Messages in this thread | | | Date | Wed, 29 Oct 2008 14:54:46 -0700 | From | Arjan van de Ven <> | Subject | Re: 2.6.28-rc2-mm1: possible circular locking |
| |
On Wed, 29 Oct 2008 13:58:40 -0700 Andrew Morton <akpm@linux-foundation.org> wrote:
> We've been calling schedule_on_each_cpu() from within > lru_add_drain_all() for ages. What changed to cause all this > to start happening?
what started to get these out of the weed is that copy_*_user() is now annotated to (potentially) take the mmap sem (which it does if there's a fault)... previously you had to actually fault to get the lock dependency noticed.
-- Arjan van de Ven Intel Open Source Technology Centre For development, discussion and tips for power savings, visit atty://www.lesswatts.org
| |