Re: [PATCH] possible scheduler deadlock in 2.6.16

Nick Piggin wrote:

Anton Blanchard wrote:

quoted

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.

But now that it's there it will enable further optimizations in parts of
sched.c, wouldn't it?  E.g. there's a number of functions that get
passed both the run queue and the CPI id as arguments and these could be
simplified.

Peter
-- 
Peter Williams                                   pwil3058@bigpond.net.au

"Learning, n. The kind of ignorance distinguishing the studious."
  -- Ambrose Bierce