[PATCH] perf, x86: Handle extended offcore mask on Haswell

[PATCH] perf, x86: Handle extended offcore mask on Haswell

[Andi Kleen]

From: Andi Kleen <ak@linux.intel.com>

HSW-EP has a larger offcore mask than the client Haswell CPUs.
It is the same mask as on Sandy/IvyBridge-EP.

On the client parts some bits were also missing compared
to Sandy/IvyBridge, in particular the bits to match on a L4
cache hit.

Specifying extra bits is ignored by the CPU.

So use the snbep extended mask, which is a superset of the
client and the server, for all of Haswell.

This allows specifying a number of extra offcore events, like
for example for HSW-EP.

% perf stat -e cpu/event=0xb7,umask=0x1,offcore_rsp=0x3fffc00100,name=offcore_response_pf_l3_rfo_l3_miss_any_response/ true

which were <not counted> before.

Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
 arch/x86/kernel/cpu/perf_event_intel.c | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c
index 2502d0d..42cd9bc 100644
--- a/arch/x86/kernel/cpu/perf_event_intel.c
+++ b/arch/x86/kernel/cpu/perf_event_intel.c
@@ -2552,7 +2552,10 @@ __init int intel_pmu_init(void)
 
 		x86_pmu.event_constraints = intel_hsw_event_constraints;
 		x86_pmu.pebs_constraints = intel_hsw_pebs_event_constraints;
-		x86_pmu.extra_regs = intel_snb_extra_regs;
+		if (boot_cpu_data.x86_model == 63)
+			x86_pmu.extra_regs = intel_snbep_extra_regs;
+		else
+			x86_pmu.extra_regs = intel_snb_extra_regs;
 		x86_pmu.pebs_aliases = intel_pebs_aliases_snb;
 		/* all extra regs are per-cpu when HT is on */
 		x86_pmu.er_flags |= ERF_HAS_RSP_1;
-- 
1.9.3

Re: [PATCH] perf, x86: Handle extended offcore mask on Haswell

[Andi Kleen]

Andi Kleen <andi@firstfloor.org> writes:

> From: Andi Kleen <ak@linux.intel.com>
>
> HSW-EP has a larger offcore mask than the client Haswell CPUs.
> It is the same mask as on Sandy/IvyBridge-EP.

Sorry I posted the wrong patch, not matching the description.
Use the followon patch instead.

-Andi

-- 
ak@linux.intel.com -- Speaking for myself only