Messages in this thread | | | Date | Thu, 1 Aug 2013 15:35:49 +0200 | From | Peter Zijlstra <> | Subject | Re: [PATCH 6/8] perf: Account freq events per cpu |
| |
On Thu, Aug 01, 2013 at 03:31:55PM +0200, Peter Zijlstra wrote: > the branch_stack thing tests ATTACH_TASK, which should also be good, but > leaves me wonder wth they do for those that are attached to tasks.
Ah found it, in intel_pmu_lbr_enable() we test cpuc->lbr_context against event->ctx and wipe the LBR if they differ, that should indeed avoid the leak but isn't sufficient to guarantee sane data as the context swap for inherited counters can foil this.
| |