perf_counter patches
Get series with b4
b4 am [email protected]Versions newest first
- v1 Latest6 patches 18 repliesIn review
Patches in v1 No applies recorded yet
| # | Subject and files | Size | Replies | Applied |
|---|---|---|---|---|
| 1/6 | perf stat: define MATCH_EVENT for easy attrs checkingtools/ | +10 | 1 reply | |
| 2/6 | perf stat: treat same behaviour for all CYCLES and CLOCKStools/ | +34 | 3 replies | |
| 3/6 | perf_counter: Add Generalized Hardware vectored co-processor support for AMDarch/ include/ kernel/ +1 more | +87 | 20 replies | |
| 4/6 | perf_counter: Add Generalized Hardware interrupt support for AMDarch/ include/ kernel/ +1 more | +78 | 9 replies | |
| 5/6 | perf_counter: Add hardware vector events for nehalemarch/ | +12 | 0 replies | |
| 6/6 | perf_counter: Add hardware interrupt events for nehalem, core2 and atomarch/ | +20 | 0 replies |
Reviews trailers per patch
No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.
Cover letter [GIT-PULL -tip][PATCH 0/6] perf_counter patches
Ingo,
Please pull perf_counter patches :
The following changes since commit 092304de242705abf24edcb0fc7beed4c4276865:
Ingo Molnar (1):
Merge branch 'perfcounters/urgent'
are available in the git repository at:
git://git.kernel.org/pub/scm/linux/kernel/git/jaswinder/linux-2.6-tip.git master
Jaswinder Singh Rajput (6):
perf stat: define MATCH_EVENT for easy attrs checking
perf stat: treat same behaviour for all CYCLES and CLOCKS
perf_counter: Add Generalized Hardware vectored co-processor support for AMD
perf_counter: Add Generalized Hardware interrupt support for AMD
perf_counter: Add hardware vector events for nehalem
perf_counter: Add hardware interrupt events for nehalem, core2 and atom
arch/x86/kernel/cpu/perf_counter.c | 95 ++++++++++++++++++++++++++++++++++++ include/linux/perf_counter.h | 27 ++++++++++ kernel/perf_counter.c | 2 + tools/perf/builtin-stat.c | 60 ++++++++++++++--------- tools/perf/util/parse-events.c | 73 +++++++++++++++++++++++++++ 5 files changed, 233 insertions(+), 24 deletions(-)
Complete diff:
arch/x86/kernel/cpu/perf_counter.c +95 -0
@@ -372,6 +372,42 @@ static const u64 atom_hw_cache_event_ids }, }; +/* +* Generalized hw vectored co-processor event table +*/ + +static u64 __read_mostly hw_vector_event_ids[PERF_COUNT_HW_VECTOR_MAX ]; + +static const u64 nehalem_hw_vector_event_ids[] = +{ + [PERF_COUNT_HW_VECTOR_ADD ] =0x01B1 ,/* UOPS_EXECUTED.PORT0 */ + [PERF_COUNT_HW_VECTOR_MULTIPLY ] =0x0214 ,/* ARITH.MUL */ + [PERF_COUNT_HW_VECTOR_DIVIDE ] =0x0114 ,/* ARITH.CYCLES_DIV_BUSY */ + [PERF_COUNT_HW_VECTOR_IDLE_CYCLES ] =0x0 , + [PERF_COUNT_HW_VECTOR_STALL_CYCLES ] =0x60A2 ,/* RESOURCE_STALLS.FPCW|MXCSR*/ + [PERF_COUNT_HW_VECTOR_OPS ] =0x0710 ,/* FP_COMP_OPS_EXE.X87|MMX|SSE_FP*/ +}; + +/* +* Generalized hw interrupt event table +*/ + +static u64 __read_mostly hw_interrupt_event_ids[PERF_COUNT_HW_INTERRUPT_MAX ]; + +static const u64 nehalem_hw_interrupt_event_ids[] = +{ + [PERF_COUNT_HW_INTERRUPT ] =0x011D ,/* HW_INT.RCV */ + [PERF_COUNT_HW_INTERRUPT_MASK ] =0x021D ,/* HW_INT.CYCLES_MASKED */ + [PERF_COUNT_HW_INTERRUPT_PENDING_MASK ]=0x041D ,/* HW_INT.CYCLES_PENDING_AND_MASKED*/ +}; + +static const u64 core2_atom_hw_interrupt_event_ids[] = +{ + [PERF_COUNT_HW_INTERRUPT ] =0x00C8 ,/* HW_INT_RCV */ + [PERF_COUNT_HW_INTERRUPT_MASK ] =0x01C6 ,/* CYCLES_INT_MASKED.CYCLES_INT_MASKED*/ + [PERF_COUNT_HW_INTERRUPT_PENDING_MASK ]=0x02C6 ,/* CYCLES_INT_MASKED.CYCLES_INT_PENDING_AND_MASKED*/ +}; +static u64 intel_pmu_raw_event (u64 event) {#define CORE_EVNTSEL_EVENT_MASK 0x000000FFULL
@@ -481,6 +517,25 @@ static const u64 amd_hw_cache_event_ids }, }; +static const u64 amd_hw_vector_event_ids[] = +{ + [PERF_COUNT_HW_VECTOR_ADD ] =0x0100 ,/* Dispatched FPU Add */ + [PERF_COUNT_HW_VECTOR_MULTIPLY ] =0x0200 ,/* Dispatched FPU Multiply */ + [PERF_COUNT_HW_VECTOR_DIVIDE ] =0x0400 ,/* Dispatched FPU Store */ + [PERF_COUNT_HW_VECTOR_IDLE_CYCLES ] =0x0001 ,/* FPU Empty cycles */ + [PERF_COUNT_HW_VECTOR_STALL_CYCLES ] =0x00D7 ,/* Dispatch stall for FPU */ + [PERF_COUNT_HW_VECTOR_OPS ] =0x0FCB ,/* Retired x87|(MMX & 3Dnow) +|SSE & SSE2) Instructions */ +}; + + +static const u64 amd_hw_interrupt_event_ids[] = +{ + [PERF_COUNT_HW_INTERRUPT ] =0x00CF ,/* Interrupts Taken */ + [PERF_COUNT_HW_INTERRUPT_MASK ] =0x00CD ,/* Interrupts-Masked Cycles*/ + [PERF_COUNT_HW_INTERRUPT_PENDING_MASK ]=0x00CE ,/* Int Mask+Pending Cycles */ +}; +/* * AMD Performance Monitor K7 and later. */
@@ -659,6 +714,28 @@ set_ext_hw_attr(struct hw_perf_counter *hwc, struct perf_counter_attr *attr)return 0 ; } +static inline int +set_hw_vector_attr (struct hw_perf_counter *hwc,struct perf_counter_attr *attr) +{ +if (attr->config >=PERF_COUNT_HW_VECTOR_MAX ) +return -EINVAL ; + + hwc->config |= hw_vector_event_ids[attr->config]; + +return 0 ; +} + +static inline int +set_hw_interrupt_attr (struct hw_perf_counter *hwc,struct perf_counter_attr *attr) +{ +if (attr->config >=PERF_COUNT_HW_INTERRUPT_MAX ) +return -EINVAL ; + + hwc->config |= hw_interrupt_event_ids[attr->config]; + +return 0 ; +} +/* * Setup the hardware configuration for a given attr_type */
@@ -716,6 +793,12 @@ static int __hw_perf_counter_init(struct perf_counter *counter)if (attr->type ==PERF_TYPE_HW_CACHE )return set_ext_hw_attr (hwc, attr); +if (attr->type ==PERF_TYPE_HW_VECTOR ) +return set_hw_vector_attr (hwc, attr); + +if (attr->type ==PERF_TYPE_HW_INTERRUPT ) +return set_hw_interrupt_attr (hwc, attr); +if (attr->config >= x86_pmu.max_events)return -EINVAL ;/*
@@ -1437,6 +1520,8 @@ static int intel_pmu_init(void)case 29 :/* six-core 45 nm xeon "Dunnington" */ memcpy (hw_cache_event_ids, core2_hw_cache_event_ids,sizeof (hw_cache_event_ids)); +memcpy (hw_interrupt_event_ids, core2_atom_hw_interrupt_event_ids, +sizeof (hw_interrupt_event_ids));pr_cont ("Core2 events, " );break ;
@@ -1444,12 +1529,18 @@ static int intel_pmu_init(void)case 26 :memcpy (hw_cache_event_ids, nehalem_hw_cache_event_ids,sizeof (hw_cache_event_ids)); +memcpy (hw_vector_event_ids, nehalem_hw_vector_event_ids, +sizeof (hw_vector_event_ids)); +memcpy (hw_interrupt_event_ids, nehalem_hw_interrupt_event_ids, +sizeof (hw_interrupt_event_ids));pr_cont ("Nehalem/Corei7 events, " );break ;case 28 :memcpy (hw_cache_event_ids, atom_hw_cache_event_ids,sizeof (hw_cache_event_ids)); +memcpy (hw_interrupt_event_ids, core2_atom_hw_interrupt_event_ids, +sizeof (hw_interrupt_event_ids));pr_cont ("Atom events, " );break ;
@@ -1468,6 +1559,10 @@ static int amd_pmu_init(void)/* Events are common for all AMDs */ memcpy (hw_cache_event_ids, amd_hw_cache_event_ids,sizeof (hw_cache_event_ids)); +memcpy (hw_vector_event_ids, amd_hw_vector_event_ids, +sizeof (hw_vector_event_ids)); +memcpy (hw_interrupt_event_ids, amd_hw_interrupt_event_ids, +sizeof (hw_interrupt_event_ids));return 0 ; }
include/linux/perf_counter.h +27 -0
@@ -31,6 +31,8 @@ enum perf_type_id {PERF_TYPE_TRACEPOINT =2 ,PERF_TYPE_HW_CACHE =3 ,PERF_TYPE_RAW =4 , +PERF_TYPE_HW_VECTOR =5 , +PERF_TYPE_HW_INTERRUPT =6 , PERF_TYPE_MAX, /* non-ABI */ };
@@ -89,6 +91,31 @@ enum perf_hw_cache_op_result_id { };/* +* Generalized hardware vectored co-processor counters: +*/ +enum perf_hw_vector_id { +PERF_COUNT_HW_VECTOR_ADD =0 , +PERF_COUNT_HW_VECTOR_MULTIPLY =1 , +PERF_COUNT_HW_VECTOR_DIVIDE =2 , +PERF_COUNT_HW_VECTOR_IDLE_CYCLES =3 , +PERF_COUNT_HW_VECTOR_STALL_CYCLES =4 , +PERF_COUNT_HW_VECTOR_OPS =5 , + +PERF_COUNT_HW_VECTOR_MAX ,/* non-ABI */ +}; + +/* +* Generalized hardware inturrupt counters: +*/ +enum perf_hw_interrupt_id { +PERF_COUNT_HW_INTERRUPT =0 , +PERF_COUNT_HW_INTERRUPT_MASK =1 , +PERF_COUNT_HW_INTERRUPT_PENDING_MASK =2 , + +PERF_COUNT_HW_INTERRUPT_MAX ,/* non-ABI */ +}; + +/* * Special "software" counters provided by the kernel, even if the hardware * does not support performance counters. These counters measure various * physical and sw events of the kernel (and allow the profiling of them as
kernel/perf_counter.c +2 -0
@@ -3838,6 +3838,8 @@ perf_counter_alloc(struct perf_counter_attr *attr,case PERF_TYPE_RAW :case PERF_TYPE_HARDWARE :case PERF_TYPE_HW_CACHE : +case PERF_TYPE_HW_VECTOR : +case PERF_TYPE_HW_INTERRUPT : pmu =hw_perf_counter_init (counter);break ;
tools/perf/builtin-stat.c +36 -24
97 lines not on this page · raw message
tools/perf/util/parse-events.c +73 -0
97 lines not on this page · raw message