perf event, x86: Use better stalled cycles metric

Use the UOPS_EXECUTED.*,c=1,i=1 event on Intel CPUs - it is a rather
good indicator of CPU execution stalls, more sensitive and more inclusive
than the 0xa2 resource stalls event (which does not count nearly as many
stall types).

Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Link: http://lkml.kernel.org/n/tip-7y40wib8n1eqio7hjpn2dsrm@git.kernel.org
Signed-off-by: Ingo Molnar <mingo@elte.hu>

diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 067a48b..1ea9422 100644 (file)
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -1413,8 +1413,8 @@ static __init int intel_pmu_init(void)
                x86_pmu.enable_all = intel_pmu_nhm_enable_all;
                x86_pmu.extra_regs = intel_nehalem_extra_regs;
 
-               /* Install the stalled-cycles event: 0xff: All reasons, 0xa2: Resource stalls */
-               intel_perfmon_event_map[PERF_COUNT_HW_STALLED_CYCLES] = 0xffa2;
+               /* Install the stalled-cycles event: UOPS_EXECUTED.CORE_ACTIVE_CYCLES,c=1,i=1 */
+               intel_perfmon_event_map[PERF_COUNT_HW_STALLED_CYCLES] = 0x1803fb1;
 
                if (ebx & 0x40) {
                        /*