Re: [PATCH] x86, perf, p4: Counter corruption when using lots of perf groups

2014-02-03 Thread Don Zickus
On Mon, Feb 03, 2014 at 10:19:29AM +0400, Cyrill Gorcunov wrote: > On Wed, Jan 29, 2014 at 03:17:17PM -0500, Don Zickus wrote: > > > > I am not entirely sure on the corruption path, but what happens is: > > > > > > > > o perf schedules a group with p4_pmu_schedule_events() > > > > o inside p4_pmu_

Re: [PATCH] x86, perf, p4: Counter corruption when using lots of perf groups

2014-02-02 Thread Cyrill Gorcunov
On Wed, Jan 29, 2014 at 03:17:17PM -0500, Don Zickus wrote: > > > I am not entirely sure on the corruption path, but what happens is: > > > > > > o perf schedules a group with p4_pmu_schedule_events() > > > o inside p4_pmu_schedule_events(), it notices an hwc pointer is being > > > reused > > >

Re: [PATCH] x86, perf, p4: Counter corruption when using lots of perf groups

2014-01-29 Thread Don Zickus
On Thu, Jan 30, 2014 at 12:06:57AM +0400, Cyrill Gorcunov wrote: > On Wed, Jan 29, 2014 at 02:37:50PM -0500, Don Zickus wrote: > > On a P4 box stressing perf with > > > > ./perf record -o perf.data ./perf stat -v ./perf bench all > > > > it was noticed that a slew of unknown NMIs would pop out ra

Re: [PATCH] x86, perf, p4: Counter corruption when using lots of perf groups

2014-01-29 Thread Cyrill Gorcunov
On Wed, Jan 29, 2014 at 02:37:50PM -0500, Don Zickus wrote: > On a P4 box stressing perf with > > ./perf record -o perf.data ./perf stat -v ./perf bench all > > it was noticed that a slew of unknown NMIs would pop out rather quickly. > > Painfully debugging this ancient platform, led me to notic

[PATCH] x86, perf, p4: Counter corruption when using lots of perf groups

2014-01-29 Thread Don Zickus
On a P4 box stressing perf with ./perf record -o perf.data ./perf stat -v ./perf bench all it was noticed that a slew of unknown NMIs would pop out rather quickly. Painfully debugging this ancient platform, led me to notice cross cpu counter corruption. The P4 machine is special in that it has