Messages in this thread | | | Subject | Re: [BUG] perf_events: ctx_flexible_sched_in() | From | Peter Zijlstra <> | Date | Mon, 01 Feb 2010 13:27:03 +0100 |
| |
On Mon, 2010-02-01 at 13:20 +0100, Stephane Eranian wrote: > The function does not allow maximizing PMU usage because of > the way can_add_hw is managed. Basically, as soon as a group > fail to be scheduled in, then no other group can. I believe this > is not optimum. You need to skip the group that fails and keep > scanning the list. There may be other groups which can be > scheduled.
Yeah, I saw that too, we need a new hw_ callback for that, or more structured error values out of ->enable that distinguish between this event won't fit and pmu full.
| |