[PATCH] possible scheduler deadlock in 2.6.16

[Nick Piggin]

Anton Blanchard wrote:

> One way to solve this is to always take runqueues in cpu id order. To do
> this we add a cpu variable to the runqueue and check it in the
> double runqueue locking functions.
> 
> Thoughts?
> 

You're right. I can't think of a better fix, although we've been trying
to avoid adding cpu to the runqueue structure.

I was going to suggest moving more work into wake_sleeping_dependent
instead, but cores with 4 and more threads now make that less desirable
I suppose.

-- 
SUSE Labs, Novell Inc.
Send instant messages to your online friends http://au.messenger.yahoo.com