Skip to contentSkip to conversation

perf: honoring the architectural performance monitoring version

PATCH In review Imre Palik 2 participants 3 messages Jun 15, 2015 – Jun 18, 2015

Subsystems (2)
lore
Share & tools
lore

Imre Palik

author
#0 lore raw

PATCH v3 1/1 perf: honoring the architectural performance monitoring version

From: "Palik, Imre" <[email protected]>

Architectural performance monitoring version 1 doesn't support fixed counters. Currently, even if a hypervisor advertises support for architectural performance monitoring version 1, perf may still tries to use the fixed counters, as the constraints are set up based on the CPU model.

This patch ensures that perf honors the architectural performance monitoring version returned by CPUID, and it only uses the fixed counters for version two and above.

Some of the ideas in this patch are coming from Peter Zijlstra.

Cc Anthony Liguori <[email protected]>
---
 arch/x86/kernel/cpu/perf_event_intel.c |   13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)
arch/x86/kernel/cpu/perf_event_intel.c +7 -6
index a1e35c9..c54bb97 100644
@@ -3322,13 +3322,14 @@ __init int intel_pmu_init(void)
 		 * counter, so do not extend mask to generic counters
 		 */
 		for_each_event_constraint(c, x86_pmu.event_constraints) {
-			if (c->cmask != FIXED_EVENT_FLAGS
-			    || c->idxmsk64 == INTEL_PMC_MSK_FIXED_REF_CYCLES) {
-				continue;
+			if (c->cmask == FIXED_EVENT_FLAGS
+			    && c->idxmsk64 != INTEL_PMC_MSK_FIXED_REF_CYCLES) {
+				c->idxmsk64 |=
+					(1ULL << x86_pmu.num_counters) - 1;
 			}
-
-			c->idxmsk64 |= (1ULL << x86_pmu.num_counters) - 1;
-			c->weight += x86_pmu.num_counters;
+			c->idxmsk64 &=
+				~(~0UL << (INTEL_PMC_IDX_FIXED + x86_pmu.num_counters_fixed));
+			c->weight = hweight64(c->idxmsk64);
 		}
 	}
 
Signature · 1 line
1.7.9.5
maintainer
#1 lore raw
Quoted 7 lines · Imre Palik

From: "Palik, Imre" <[email protected]>

Architectural performance monitoring version 1 doesn't support fixed counters. Currently, even if a hypervisor advertises support for architectural performance monitoring version 1, perf may still tries to use the fixed counters, as the constraints are set up based on the CPU model.

Quoted 5 lines

This patch ensures that perf honors the architectural performance monitoring version returned by CPUID, and it only uses the fixed counters for version two and above.

Some of the ideas in this patch are coming from Peter Zijlstra.

What's the difference from -v2? I'm not really seeing it.

author
#2 lore raw
Quoted 8 lines · Peter Zijlstra
Quoted 7 lines · Imre Palik

From: "Palik, Imre" <[email protected]>

Architectural performance monitoring version 1 doesn't support fixed counters. Currently, even if a hypervisor advertises support for architectural performance monitoring version 1, perf may still tries to use the fixed counters, as the constraints are set up based on the CPU model.

Quoted 8 lines
Quoted 5 lines

This patch ensures that perf honors the architectural performance monitoring version returned by CPUID, and it only uses the fixed counters for version two and above.

Some of the ideas in this patch are coming from Peter Zijlstra.

What's the difference from -v2? I'm not really seeing it.

I made the mistake of sending two different versions as v2. v3 is the same as the second v2.

I sent this to clear up possible confusion. Sorry if it caused more of it.