Messages in this thread | | | Subject | Re: [PATCH] sched: avoid unnecessary overflow in sched_clock | From | Peter Zijlstra <> | Date | Wed, 16 Nov 2011 09:56:22 +0100 |
| |
On Tue, 2011-11-15 at 23:08 -0800, Paul Turner wrote: > Probably a good idea -- This especially sucks rocks in the sched_clock_stable==1 > case; resulting in it coming straight back out of sched_clock_cpu() and trashing > rq->clock.
There's another problem there in that TSC will wrap too soon and things will blow up in your face that way.
| |