From 5e176213a6b2bc5146820c79542d37290434a3c4 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Mon, 14 Sep 2015 14:47:02 -0700 Subject: perf/x86/intel: Make the CYCLE_ACTIVITY.* constraint on Broadwell more specific The counter constraint for CYCLE_ACTIVITY.* on Broadwell covered all CYCLE_ACTIVITY.* sub events, and forced them on counter 2. But actually only one sub event (umask 8) needs to be on counter 2, all others do not have any constraint. Only force that subevent. This fixes groups with multiple CYCLE_ACTIVITY.* events, for example: % perf stat -x, -e '{cpu/event=0xa3,umask=0x6,cmask=6/,\ cpu/event=0xa2,umask=0x8/,\ cpu/event=0xa3,umask=0x4,cmask=4/,cpu/event=0xb1,umask=0x1,cmask=1/}' true 122150,,cpu/event=0xa3,umask=0x6,cmask=6/,846486,100.00 16483,,cpu/event=0xa2,umask=0x8/,846486,100.00 252280,,cpu/event=0xa3,umask=0x4,cmask=4/,846486,100.00 233604,,cpu/event=0xb1,umask=0x1,cmask=1/,846486,100.00 % Without this patch the third result would be Signed-off-by: Andi Kleen Signed-off-by: Peter Zijlstra (Intel) Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/1442267222-16464-1-git-send-email-andi@firstfloor.org Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/perf_event_intel.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c index 3fefebfbdf4b..1d84b41ba932 100644 --- a/arch/x86/kernel/cpu/perf_event_intel.c +++ b/arch/x86/kernel/cpu/perf_event_intel.c @@ -250,7 +250,7 @@ struct event_constraint intel_bdw_event_constraints[] = { FIXED_EVENT_CONSTRAINT(0x003c, 1), /* CPU_CLK_UNHALTED.CORE */ FIXED_EVENT_CONSTRAINT(0x0300, 2), /* CPU_CLK_UNHALTED.REF */ INTEL_UEVENT_CONSTRAINT(0x148, 0x4), /* L1D_PEND_MISS.PENDING */ - INTEL_EVENT_CONSTRAINT(0xa3, 0x4), /* CYCLE_ACTIVITY.* */ + INTEL_UEVENT_CONSTRAINT(0x8a3, 0x4), /* CYCLE_ACTIVITY.CYCLES_L1D_MISS */ EVENT_CONSTRAINT_END }; -- cgit v1.2.3 From d0dc8494cd6904f8ad035d9ad97f313948f35d0c Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Wed, 9 Sep 2015 14:53:59 -0700 Subject: perf/x86/intel/pebs: Add PEBS frontend profiling for Skylake Skylake has a new FRONTEND_LATENCY PEBS event to accurately profile frontend problems (like ITLB or decoding issues). The new event is configured through a separate MSR, which selects a range of sub events. Define the extra MSR as a extra reg and export support for it through sysfs. To avoid duplicating the existing tables use a new function to add new entries to existing tables. Signed-off-by: Andi Kleen Signed-off-by: Peter Zijlstra (Intel) Acked-by: Thomas Gleixner Cc: Linus Torvalds Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1435707205-6676-4-git-send-email-andi@firstfloor.org Signed-off-by: Ingo Molnar --- arch/x86/include/asm/msr-index.h | 2 ++ arch/x86/kernel/cpu/perf_event.h | 1 + arch/x86/kernel/cpu/perf_event_intel.c | 11 ++++++++++- 3 files changed, 13 insertions(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h index c1c0a1c14344..54390bc140dd 100644 --- a/arch/x86/include/asm/msr-index.h +++ b/arch/x86/include/asm/msr-index.h @@ -141,6 +141,8 @@ #define DEBUGCTLMSR_BTS_OFF_USR (1UL << 10) #define DEBUGCTLMSR_FREEZE_LBRS_ON_PMI (1UL << 11) +#define MSR_PEBS_FRONTEND 0x000003f7 + #define MSR_IA32_POWER_CTL 0x000001fc #define MSR_IA32_MC0_CTL 0x00000400 diff --git a/arch/x86/kernel/cpu/perf_event.h b/arch/x86/kernel/cpu/perf_event.h index 5edf6d868fc1..165be83a7fa4 100644 --- a/arch/x86/kernel/cpu/perf_event.h +++ b/arch/x86/kernel/cpu/perf_event.h @@ -47,6 +47,7 @@ enum extra_reg_type { EXTRA_REG_RSP_1 = 1, /* offcore_response_1 */ EXTRA_REG_LBR = 2, /* lbr_select */ EXTRA_REG_LDLAT = 3, /* ld_lat_threshold */ + EXTRA_REG_FE = 4, /* fe_* */ EXTRA_REG_MAX /* number of entries needed */ }; diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c index 1d84b41ba932..ef74c9d05f82 100644 --- a/arch/x86/kernel/cpu/perf_event_intel.c +++ b/arch/x86/kernel/cpu/perf_event_intel.c @@ -205,6 +205,7 @@ static struct extra_reg intel_skl_extra_regs[] __read_mostly = { INTEL_UEVENT_EXTRA_REG(0x01b7, MSR_OFFCORE_RSP_0, 0x3fffff8fffull, RSP_0), INTEL_UEVENT_EXTRA_REG(0x01bb, MSR_OFFCORE_RSP_1, 0x3fffff8fffull, RSP_1), INTEL_UEVENT_PEBS_LDLAT_EXTRA_REG(0x01cd), + INTEL_UEVENT_EXTRA_REG(0x01c6, MSR_PEBS_FRONTEND, 0x3fff17, FE), EVENT_EXTRA_END }; @@ -2891,6 +2892,8 @@ PMU_FORMAT_ATTR(offcore_rsp, "config1:0-63"); PMU_FORMAT_ATTR(ldlat, "config1:0-15"); +PMU_FORMAT_ATTR(frontend, "config1:0-23"); + static struct attribute *intel_arch3_formats_attr[] = { &format_attr_event.attr, &format_attr_umask.attr, @@ -2907,6 +2910,11 @@ static struct attribute *intel_arch3_formats_attr[] = { NULL, }; +static struct attribute *skl_format_attr[] = { + &format_attr_frontend.attr, + NULL, +}; + static __initconst const struct x86_pmu core_pmu = { .name = "core", .handle_irq = x86_pmu_handle_irq, @@ -3516,7 +3524,8 @@ __init int intel_pmu_init(void) x86_pmu.hw_config = hsw_hw_config; x86_pmu.get_event_constraints = hsw_get_event_constraints; - x86_pmu.cpu_events = hsw_events_attrs; + x86_pmu.format_attrs = merge_attr(intel_arch3_formats_attr, + skl_format_attr); WARN_ON(!x86_pmu.format_attrs); x86_pmu.cpu_events = hsw_events_attrs; pr_cont("Skylake events, "); -- cgit v1.2.3 From dfe1f3cb312624928052413928d88b0ee3492216 Mon Sep 17 00:00:00 2001 From: Andi Kleen Date: Wed, 9 Sep 2015 14:54:00 -0700 Subject: perf/x86/intel: Fix Skylake FRONTEND MSR extrareg mask Stephane pointed out that the extrareg mask was one bit too short. The bubble width field was truncated by one bit. Fix that here. Also add some extra comments on the reserved bits inside the event select code. Reported-by: Stephane Eranian Signed-off-by: Andi Kleen Signed-off-by: Peter Zijlstra (Intel) Acked-by: Thomas Gleixner Cc: Linus Torvalds Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1441835640-21347-3-git-send-email-andi@firstfloor.org Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/perf_event_intel.c | 6 +++++- 1 file changed, 5 insertions(+), 1 deletion(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/cpu/perf_event_intel.c b/arch/x86/kernel/cpu/perf_event_intel.c index ef74c9d05f82..f63360be2238 100644 --- a/arch/x86/kernel/cpu/perf_event_intel.c +++ b/arch/x86/kernel/cpu/perf_event_intel.c @@ -205,7 +205,11 @@ static struct extra_reg intel_skl_extra_regs[] __read_mostly = { INTEL_UEVENT_EXTRA_REG(0x01b7, MSR_OFFCORE_RSP_0, 0x3fffff8fffull, RSP_0), INTEL_UEVENT_EXTRA_REG(0x01bb, MSR_OFFCORE_RSP_1, 0x3fffff8fffull, RSP_1), INTEL_UEVENT_PEBS_LDLAT_EXTRA_REG(0x01cd), - INTEL_UEVENT_EXTRA_REG(0x01c6, MSR_PEBS_FRONTEND, 0x3fff17, FE), + /* + * Note the low 8 bits eventsel code is not a continuous field, containing + * some #GPing bits. These are masked out. + */ + INTEL_UEVENT_EXTRA_REG(0x01c6, MSR_PEBS_FRONTEND, 0x7fff17, FE), EVENT_EXTRA_END }; -- cgit v1.2.3 From 7e5560a5648ab2bce7199c73b9c2a51b846f5541 Mon Sep 17 00:00:00 2001 From: Geliang Tang Date: Thu, 24 Sep 2015 04:48:53 -0700 Subject: perf/x86: Change test_aperfmperf() and test_intel() to static Fixes the following sparse warnings: arch/x86/kernel/cpu/perf_event_msr.c:13:6: warning: symbol 'test_aperfmperf' was not declared. Should it be static? arch/x86/kernel/cpu/perf_event_msr.c:18:6: warning: symbol 'test_intel' was not declared. Should it be static? Signed-off-by: Geliang Tang Cc: Arnaldo Carvalho de Melo Cc: Linus Torvalds Cc: Peter Zijlstra Cc: Peter Zijlstra Cc: Thomas Gleixner Link: http://lkml.kernel.org/r/4588e8ab09638458f2451af572827108be3b4a36.1443123796.git.geliangtang@163.com Signed-off-by: Ingo Molnar --- arch/x86/kernel/cpu/perf_event_msr.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'arch/x86') diff --git a/arch/x86/kernel/cpu/perf_event_msr.c b/arch/x86/kernel/cpu/perf_event_msr.c index 086b12eae794..f32ac13934f2 100644 --- a/arch/x86/kernel/cpu/perf_event_msr.c +++ b/arch/x86/kernel/cpu/perf_event_msr.c @@ -10,12 +10,12 @@ enum perf_msr_id { PERF_MSR_EVENT_MAX, }; -bool test_aperfmperf(int idx) +static bool test_aperfmperf(int idx) { return boot_cpu_has(X86_FEATURE_APERFMPERF); } -bool test_intel(int idx) +static bool test_intel(int idx) { if (boot_cpu_data.x86_vendor != X86_VENDOR_INTEL || boot_cpu_data.x86 != 6) -- cgit v1.2.3