perf intel-pt: Add PEBS-via-PT side-band
Subsystems (2)
Adrian Hunter
reviewerv3 cover perf intel-pt: Add PEBS-via-PT side-band
Hi
PEBS output to Intel Processor Trace was introduced with Atom processors based on Tremont. Currently there is software support only for a single PEBS-via-PT event.
Here is support for multiple PEBS-via-PT events.
The first patch is the kernel change which adds a new event, namely PERF_RECORD_AUX_OUTPUT_HW_ID which contains the counter index. That is output when the PEBS-via-PT event is enabled in hardware. There is an optimization, to report the index only when it changes for the event. That will work only so long as all PEBS-via-PT events are scheduled together, which they are for a recording session because they are in a single group.
Also no attribute bit is used to select the new event, so a new kernel is not compatible with older perf tools. The assumption being that PEBS-via-PT is sufficiently esoteric that users will not be troubled by this.
The second patch adds the usual boiler plate to perf tools for a new event.
The third patch adds support for processing the new event by perf tool's Intel PT decoder.
Changes in V3:
perf/x86: Add new event for AUX output counter index
Do not set assign callback unless x86_pmu.intel_cap.pebs_output_pt_available
Changes in V2:
perf/x86: Add new event for AUX output counter index
Use callback from x86_assign_hw_event
Adrian Hunter (3):
perf/x86: Add new event for AUX output counter index perf tools: Add support for PERF_RECORD_AUX_OUTPUT_HW_ID perf intel-pt: Add support for PERF_RECORD_AUX_OUTPUT_HW_ID
arch/x86/events/core.c | 6 +++ arch/x86/events/intel/core.c | 16 ++++++ arch/x86/events/perf_event.h | 1 + include/linux/perf_event.h | 1 + include/uapi/linux/perf_event.h | 15 ++++++ kernel/events/core.c | 30 +++++++++++ tools/include/uapi/linux/perf_event.h | 15 ++++++ tools/lib/perf/include/perf/event.h | 6 +++ tools/perf/Documentation/perf-intel-pt.txt | 7 ++- tools/perf/builtin-inject.c | 4 +- tools/perf/builtin-record.c | 2 +- tools/perf/util/event.c | 18 +++++++ tools/perf/util/event.h | 5 ++ tools/perf/util/intel-pt.c | 85 ++++++++++++++++++++++++++++-- tools/perf/util/machine.c | 10 ++++ tools/perf/util/machine.h | 2 + tools/perf/util/session.c | 5 ++ tools/perf/util/tool.h | 1 + 18 files changed, 222 insertions(+), 7 deletions(-)
Regards
Adrian
Adrian Hunter
reviewerPATCH v3 1/3 perf/x86: Add new event for AUX output counter index
PEBS-via-PT records contain a mask of applicable counters. To identify which event belongs to which counter, a side-band event is needed. Until now, there has been no side-band event, and consequently users were limited to using a single event.
Add such a side-band event. Note the event is optimised to output only when the counter index changes for an event. That works only so long as all PEBS-via-PT events are scheduled together, which they are for a recording session because they are in a single group.
Also no attribute bit is used to select the new event, so a new kernel is not compatible with older perf tools. The assumption being that PEBS-via-PT is sufficiently esoteric that users will not be troubled by this.
---
Changes in V3:
Do not set assign callback unless x86_pmu.intel_cap.pebs_output_pt_available
Changes in V2:
Use callback from x86_assign_hw_event
arch/x86/events/core.c | 6 ++++++ arch/x86/events/intel/core.c | 16 ++++++++++++++++ arch/x86/events/perf_event.h | 1 + include/linux/perf_event.h | 1 + include/uapi/linux/perf_event.h | 15 +++++++++++++++ kernel/events/core.c | 30 ++++++++++++++++++++++++++++++ 6 files changed, 69 insertions(+)
arch/x86/events/core.c +6 -0
@@ -66,6 +66,8 @@ DEFINE_STATIC_CALL_NULL(x86_pmu_enable_all, *x86_pmu.enable_all);DEFINE_STATIC_CALL_NULL (x86_pmu_enable, *x86_pmu.enable);DEFINE_STATIC_CALL_NULL (x86_pmu_disable, *x86_pmu.disable); +DEFINE_STATIC_CALL_NULL (x86_pmu_assign, *x86_pmu.assign); +DEFINE_STATIC_CALL_NULL (x86_pmu_add, *x86_pmu.add);DEFINE_STATIC_CALL_NULL (x86_pmu_del, *x86_pmu.del);DEFINE_STATIC_CALL_NULL (x86_pmu_read, *x86_pmu.read);
@@ -1215,6 +1217,8 @@ static inline void x86_assign_hw_event(struct perf_event *event, hwc->last_cpu =smp_processor_id (); hwc->last_tag = ++cpuc->tags[i]; +static_call_cond (x86_pmu_assign)(event, idx); +switch (hwc->idx) {case INTEL_PMC_IDX_FIXED_BTS :case INTEL_PMC_IDX_FIXED_VLBR :
@@ -2005,6 +2009,8 @@ static void x86_pmu_static_call_update(void)static_call_update (x86_pmu_enable, x86_pmu.enable);static_call_update (x86_pmu_disable, x86_pmu.disable); +static_call_update (x86_pmu_assign, x86_pmu.assign); +static_call_update (x86_pmu_add, x86_pmu.add);static_call_update (x86_pmu_del, x86_pmu.del);static_call_update (x86_pmu_read, x86_pmu.read);
arch/x86/events/intel/core.c +16 -0
@@ -2402,6 +2402,12 @@ static void intel_pmu_disable_event(struct perf_event *event)intel_pmu_pebs_disable (event); } +static void intel_pmu_assign_event (struct perf_event *event,int idx) +{ +if (is_pebs_pt (event)) +perf_report_aux_output_id (event, idx); +} +static void intel_pmu_del_event (struct perf_event *event) {if (needs_branch_stack (event))
@@ -4494,8 +4500,16 @@ static int intel_pmu_check_period(struct perf_event *event, u64 value)return intel_pmu_has_bts_period (event, value) ? -EINVAL :0 ; } +static void intel_aux_output_init (void ) +{ +/* Refer also intel_pmu_aux_output_match() */ +if (x86_pmu.intel_cap.pebs_output_pt_available) + x86_pmu.assign = intel_pmu_assign_event; +} +static int intel_pmu_aux_output_match (struct perf_event *event) { +/* intel_pmu_assign_event() is needed, refer intel_aux_output_init() */ if (!x86_pmu.intel_cap.pebs_output_pt_available)return 0 ;
@@ -6301,6 +6315,8 @@ __init int intel_pmu_init(void)if (is_hybrid ())intel_pmu_check_hybrid_pmus ((u64 )fixed_mask); +intel_aux_output_init (); +return 0 ; }
arch/x86/events/perf_event.h +1 -0
@@ -726,6 +726,7 @@ struct x86_pmu {void (*enable_all)(int added);void (*enable)(struct perf_event *);void (*disable)(struct perf_event *); +void (*assign)(struct perf_event *event,int idx);void (*add)(struct perf_event *);void (*del)(struct perf_event *);void (*read)(struct perf_event *event);
include/linux/perf_event.h +1 -0
@@ -1398,6 +1398,7 @@ perf_event_addr_filters(struct perf_event *event) }extern void perf_event_addr_filters_sync (struct perf_event *event); +extern void perf_report_aux_output_id (struct perf_event *event,u64 hw_id); extern int perf_output_begin(struct perf_output_handle *handle, struct perf_sample_data *data,
include/uapi/linux/perf_event.h +15 -0
@@ -1141,6 +1141,21 @@ enum perf_event_type {*/ PERF_RECORD_TEXT_POKE =20 , +/* +* Data written to the AUX area by hardware due to aux_output, may need +* to be matched to the event by an architecture-specific hardware ID. +* This records the hardware ID, but requires sample_id to provide the +* event ID. e.g. Intel PT uses this record to disambiguate PEBS-via-PT +* records from multiple events. +* +* struct { +* struct perf_event_header header; +* u64 hw_id; +* struct sample_id sample_id; +* }; +*/ +PERF_RECORD_AUX_OUTPUT_HW_ID =21 , + PERF_RECORD_MAX, /* non-ABI */ };
kernel/events/core.c +30 -0
@@ -9073,6 +9073,36 @@ static void perf_log_itrace_start(struct perf_event *event)perf_output_end (&handle); } +void perf_report_aux_output_id (struct perf_event *event,u64 hw_id) +{ +struct perf_output_handle handle; +struct perf_sample_data sample; +struct perf_aux_event { +struct perf_event_header header; +u64 hw_id; + } rec; +int ret; + +if (event->parent) + event = event->parent; + + rec.header.type =PERF_RECORD_AUX_OUTPUT_HW_ID ; + rec.header.misc =0 ; + rec.header.size =sizeof (rec); + rec.hw_id = hw_id; + +perf_event_header__init_id (&rec.header, &sample, event); + ret =perf_output_begin (&handle, &sample, event, rec.header.size); + +if (ret) +return ; + +perf_output_put (&handle, rec); +perf_event__output_id_sample (event, &handle, &sample); + +perf_output_end (&handle); +} +static int __perf_event_account_interrupt (struct perf_event *event,int throttle) {
Signature · 1 line
2.17.1
Re: [PATCH V3 1/3] perf/x86: Add new event for AUX output counter index
Quoted 82 lines · Adrian Hunter
PEBS-via-PT records contain a mask of applicable counters. To identify which event belongs to which counter, a side-band event is needed. Until now, there has been no side-band event, and consequently users were limited to using a single event.
Add such a side-band event. Note the event is optimised to output only when the counter index changes for an event. That works only so long as all PEBS-via-PT events are scheduled together, which they are for a recording session because they are in a single group.
Also no attribute bit is used to select the new event, so a new kernel is not compatible with older perf tools. The assumption being that PEBS-via-PT is sufficiently esoteric that users will not be troubled by this.
Signed-off-by Adrian Hunter <[email protected]>---
Changes in V3:
Do not set assign callback unless x86_pmu.intel_cap.pebs_output_pt_availableChanges in V2:
Use callback fromx86_assign_hw_eventarch/x86/events/core.c | 6 ++++++ arch/x86/events/intel/core.c | 16 ++++++++++++++++ arch/x86/events/perf_event.h | 1 + include/linux/perf_event.h | 1 + include/uapi/linux/perf_event.h | 15 +++++++++++++++ kernel/events/core.c | 30 ++++++++++++++++++++++++++++++ 6 files changed, 69 insertions(+)arch/x86/events/core.c +6
-0@@ -66,6 +66,8 @@ DEFINE_STATIC_CALL_NULL(x86_pmu_enable_all, *x86_pmu.enable_all);DEFINE_STATIC_CALL_NULL (x86_pmu_enable, *x86_pmu.enable);DEFINE_STATIC_CALL_NULL (x86_pmu_disable, *x86_pmu.disable); +DEFINE_STATIC_CALL_NULL (x86_pmu_assign, *x86_pmu.assign); +DEFINE_STATIC_CALL_NULL (x86_pmu_add, *x86_pmu.add);DEFINE_STATIC_CALL_NULL (x86_pmu_del, *x86_pmu.del);DEFINE_STATIC_CALL_NULL (x86_pmu_read, *x86_pmu.read);@@ -1215,6 +1217,8 @@ static inline void x86_assign_hw_event(struct perf_event *event, hwc->last_cpu =smp_processor_id (); hwc->last_tag = ++cpuc->tags[i]; +static_call_cond (x86_pmu_assign)(event, idx); +switch (hwc->idx) {case INTEL_PMC_IDX_FIXED_BTS :case INTEL_PMC_IDX_FIXED_VLBR :@@ -2005,6 +2009,8 @@ static void x86_pmu_static_call_update(void)static_call_update (x86_pmu_enable, x86_pmu.enable);static_call_update (x86_pmu_disable, x86_pmu.disable); +static_call_update (x86_pmu_assign, x86_pmu.assign); +static_call_update (x86_pmu_add, x86_pmu.add);static_call_update (x86_pmu_del, x86_pmu.del);static_call_update (x86_pmu_read, x86_pmu.read);arch/x86/events/intel/core.c +6
-0@@ -2402,6 +2402,12 @@ static void intel_pmu_disable_event(struct perf_event *event)intel_pmu_pebs_disable (event); } +static void intel_pmu_assign_event (struct perf_event *event,int idx) +{ +if (is_pebs_pt (event)) +perf_report_aux_output_id (event, idx); +} +static void intel_pmu_del_event (struct perf_event *event) {if (needs_branch_stack (event))
arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4508
Quoted 9 lines
arch/x86/events/intel/core.c +6
-0@@ -4494,8 +4500,16 @@ static int intel_pmu_check_period(struct perf_event *event, u64 value)return intel_pmu_has_bts_period (event, value) ? -EINVAL :0 ; } +static void intel_aux_output_init (void ) +{ +/* Refer also intel_pmu_aux_output_match() */ +if (x86_pmu.intel_cap.pebs_output_pt_available) + x86_pmu.assign = intel_pmu_assign_event; +}
For a hybrid machine, x86_pmu.intel_cap.pebs_output_pt_available is always cleared. We probably need the PMU specific pmu->intel_cap.pebs_output_pt_available here.
arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4515
Quoted 7 lines
arch/x86/events/intel/core.c +2
-0+static int intel_pmu_aux_output_match (struct perf_event *event) { +/* intel_pmu_assign_event() is needed, refer intel_aux_output_init() */ if (!x86_pmu.intel_cap.pebs_output_pt_available)return 0 ;
For a hybrid machine, this always return 0. I think we need to fix it first?
Thanks,
Kan
Quoted 101 lines
arch/x86/events/intel/core.c +2
-0@@ -6301,6 +6315,8 @@ __init int intel_pmu_init(void)if (is_hybrid ())intel_pmu_check_hybrid_pmus ((u64 )fixed_mask); +intel_aux_output_init (); +return 0 ; }arch/x86/events/perf_event.h +1
-0@@ -726,6 +726,7 @@ struct x86_pmu {void (*enable_all)(int added);void (*enable)(struct perf_event *);void (*disable)(struct perf_event *); +void (*assign)(struct perf_event *event,int idx);void (*add)(struct perf_event *);void (*del)(struct perf_event *);void (*read)(struct perf_event *event);include/linux/perf_event.h +1
-0@@ -1398,6 +1398,7 @@ perf_event_addr_filters(struct perf_event *event) }extern void perf_event_addr_filters_sync (struct perf_event *event); +extern void perf_report_aux_output_id (struct perf_event *event,u64 hw_id); extern int perf_output_begin(struct perf_output_handle *handle, struct perf_sample_data *data,include/uapi/linux/perf_event.h +15
-0@@ -1141,6 +1141,21 @@ enum perf_event_type {*/ PERF_RECORD_TEXT_POKE =20 , +/* +* Data written to the AUX area by hardware due to aux_output, may need +* to be matched to the event by an architecture-specific hardware ID. +* This records the hardware ID, but requires sample_id to provide the +* event ID. e.g. Intel PT uses this record to disambiguate PEBS-via-PT +* records from multiple events. +* +* struct { +* struct perf_event_header header; +* u64 hw_id; +* struct sample_id sample_id; +* }; +*/ +PERF_RECORD_AUX_OUTPUT_HW_ID =21 , + PERF_RECORD_MAX, /* non-ABI */ };kernel/events/core.c +30
-0@@ -9073,6 +9073,36 @@ static void perf_log_itrace_start(struct perf_event *event)perf_output_end (&handle); } +void perf_report_aux_output_id (struct perf_event *event,u64 hw_id) +{ +struct perf_output_handle handle; +struct perf_sample_data sample; +struct perf_aux_event { +struct perf_event_header header; +u64 hw_id; + } rec; +int ret; + +if (event->parent) + event = event->parent; + + rec.header.type =PERF_RECORD_AUX_OUTPUT_HW_ID ; + rec.header.misc =0 ; + rec.header.size =sizeof (rec); + rec.hw_id = hw_id; + +perf_event_header__init_id (&rec.header, &sample, event); + ret =perf_output_begin (&handle, &sample, event, rec.header.size); + +if (ret) +return ; + +perf_output_put (&handle, rec); +perf_event__output_id_sample (event, &handle, &sample); + +perf_output_end (&handle); +} +static int __perf_event_account_interrupt (struct perf_event *event,int throttle) {
Peter Zijlstra
maintainerRe: [PATCH V3 1/3] perf/x86: Add new event for AUX output counter index
Quoted 16 lines · Kan
arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4507
Quoted 9 lines · Adrian Hunter
arch/x86/events/intel/core.c +6
-0@@ -4494,8 +4500,16 @@ static int intel_pmu_check_period(struct perf_event *event, u64 value)return intel_pmu_has_bts_period (event, value) ? -EINVAL :0 ; } +static void intel_aux_output_init (void ) +{ +/* Refer also intel_pmu_aux_output_match() */ +if (x86_pmu.intel_cap.pebs_output_pt_available) + x86_pmu.assign = intel_pmu_assign_event; +}For a hybrid machine, x86_pmu.intel_cap.pebs_output_pt_available is always cleared. We probably need the PMU specific
pmu->intel_cap.pebs_output_pt_availablehere.
Quoted 9 lines
arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4515
Quoted 7 lines
arch/x86/events/intel/core.c +2
-0+static int intel_pmu_aux_output_match (struct perf_event *event) { +/* intel_pmu_assign_event() is needed, refer intel_aux_output_init() */ if (!x86_pmu.intel_cap.pebs_output_pt_available)return 0 ;For a hybrid machine, this always return 0. I think we need to fix it first?
AFAICT the patch is correct for !hybrid, and the hybrid PT muck can then also fix this up, right?
Re: [PATCH V3 1/3] perf/x86: Add new event for AUX output counter index
Quoted 27 lines · Peter Zijlstra
Quoted 16 lines · Kan
arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4507
Quoted 9 lines · Adrian Hunter
arch/x86/events/intel/core.c +6
-0@@ -4494,8 +4500,16 @@ static int intel_pmu_check_period(struct perf_event *event, u64 value)return intel_pmu_has_bts_period (event, value) ? -EINVAL :0 ; } +static void intel_aux_output_init (void ) +{ +/* Refer also intel_pmu_aux_output_match() */ +if (x86_pmu.intel_cap.pebs_output_pt_available) + x86_pmu.assign = intel_pmu_assign_event; +}For a hybrid machine, x86_pmu.intel_cap.pebs_output_pt_available is always cleared. We probably need the PMU specific
pmu->intel_cap.pebs_output_pt_availablehere.Quoted 9 lines
arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4515
Quoted 7 lines
arch/x86/events/intel/core.c +2
-0+static int intel_pmu_aux_output_match (struct perf_event *event) { +/* intel_pmu_assign_event() is needed, refer intel_aux_output_init() */ if (!x86_pmu.intel_cap.pebs_output_pt_available)return 0 ;For a hybrid machine, this always return 0. I think we need to fix it first?
Quoted 3 lines
AFAICT the patch is correct for !hybrid, and the hybrid PT muck can then also fix this up, right?
Yes, for !hybrid, the patch is good.
Since PEBS via PT is temporarily disabled for hybrid for now, the patch set should not bring any issues with hybrid either. The hybrid PT can be fixed separately.
Thanks,
Kan
Adrian Hunter
reviewerRe: [PATCH V3 1/3] perf/x86: Add new event for AUX output counter index
Quoted 30 lines · Kan
Quoted 27 lines · Peter Zijlstra
Quoted 25 lines · Kan
arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4507
Quoted 9 lines · Adrian Hunter
arch/x86/events/intel/core.c +6
-0@@ -4494,8 +4500,16 @@ static int intel_pmu_check_period(struct perf_event *event, u64 value) return intel_pmu_has_bts_period(event, value) ? -EINVAL : 0; } +static void intel_aux_output_init (void ) +{ + /* Refer also intel_pmu_aux_output_match() */ + if (x86_pmu.intel_cap.pebs_output_pt_available) + x86_pmu.assign = intel_pmu_assign_event; +}For a hybrid machine, x86_pmu.intel_cap.pebs_output_pt_available is always cleared. We probably need the PMU specific
pmu->intel_cap.pebs_output_pt_availablehere.arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4515
Quoted 7 lines
arch/x86/events/intel/core.c +2
-0+ static int intel_pmu_aux_output_match(struct perf_event *event) { + /* intel_pmu_assign_event() is needed, refer intel_aux_output_init() */ if (!x86_pmu.intel_cap.pebs_output_pt_available) return 0;For a hybrid machine, this always return 0. I think we need to fix it first?
Quoted 8 lines
Quoted 3 lines
AFAICT the patch is correct for !hybrid, and the hybrid PT muck can then also fix this up, right?
Yes, for !hybrid, the patch is good.
Since PEBS via PT is temporarily disabled for hybrid for now, the patch set should not bring any issues with hybrid either. The hybrid PT can be fixed separately.
I don't have much time to look at the hybrid case right now.
Would it be OK to go ahead with these patches?
Alexander Shishkin
reviewerRe: [PATCH V3 1/3] perf/x86: Add new event for AUX output counter index
Quoted 35 lines · Adrian Hunter
Quoted 34 lines · Kan
Quoted 30 lines · Peter Zijlstra
Quoted 24 lines · Kan
arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4507
Quoted 9 lines · Adrian Hunter
arch/x86/events/intel/core.c +6
-0@@ -4494,8 +4500,16 @@ static int intel_pmu_check_period(struct perf_event *event, u64 value) return intel_pmu_has_bts_period(event, value) ? -EINVAL : 0; } +static void intel_aux_output_init (void ) +{ + /* Refer also intel_pmu_aux_output_match() */ + if (x86_pmu.intel_cap.pebs_output_pt_available) + x86_pmu.assign = intel_pmu_assign_event; +}For a hybrid machine, x86_pmu.intel_cap.pebs_output_pt_available is always cleared. We probably need the PMU specific
pmu->intel_cap.pebs_output_pt_availablehere.arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4515
Quoted 7 lines
arch/x86/events/intel/core.c +2
-0+ static int intel_pmu_aux_output_match(struct perf_event *event) { + /* intel_pmu_assign_event() is needed, refer intel_aux_output_init() */ if (!x86_pmu.intel_cap.pebs_output_pt_available) return 0;Quoted 1 line
For a hybrid machine, this always return 0. I think we need to fix it first?
AFAICT the patch is correct for !hybrid, and the hybrid PT muck can then also fix this up, right?
Quoted 8 lines
Quoted 4 lines
Yes, for !hybrid, the patch is good.
Since PEBS via PT is temporarily disabled for hybrid for now, the patch set should not bring any issues with hybrid either. The hybrid PT can be fixed separately.
I don't have much time to look at the hybrid case right now.
Would it be OK to go ahead with these patches?
I'll deal with the PEBS-via-PT on hybrid. As it stands right now, this patchset is good.
Regards,
--
Alex
Adrian Hunter
reviewerRe: [PATCH V3 1/3] perf/x86: Add new event for AUX output counter index
Quoted 42 lines · Alexander Shishkin
Quoted 40 lines · Adrian Hunter
Quoted 38 lines · Kan
Quoted 25 lines · Peter Zijlstra
Quoted 24 lines · Kan
arch/x86/events/intel/core.c static int intel_pmu_check_period(struct perf_event *event, u64 value) · line 4507
Quoted 9 lines · Adrian Hunter
arch/x86/events/intel/core.c +6
-0@@ -4494,8 +4500,16 @@ static int intel_pmu_check_period(struct perf_event *event, u64 value) return intel_pmu_has_bts_period(event, value) ? -EINVAL : 0; } +static void intel_aux_output_init (void ) +{ + /* Refer also intel_pmu_aux_output_match() */ + if (x86_pmu.intel_cap.pebs_output_pt_available) + x86_pmu.assign = intel_pmu_assign_event; +}For a hybrid machine, x86_pmu.intel_cap.pebs_output_pt_available is always cleared. We probably need the PMU specific
pmu->intel_cap.pebs_output_pt_availablehere.Quoted 7 lines
arch/x86/events/intel/core.c +2
-0+ static int intel_pmu_aux_output_match(struct perf_event *event) { + /* intel_pmu_assign_event() is needed, refer intel_aux_output_init() */ if (!x86_pmu.intel_cap.pebs_output_pt_available) return 0;Quoted 5 lines
Quoted 1 line
For a hybrid machine, this always return 0. I think we need to fix it first?
AFAICT the patch is correct for !hybrid, and the hybrid PT muck can then also fix this up, right?
Yes, for !hybrid, the patch is good.
Since PEBS via PT is temporarily disabled for hybrid for now, the patch set should not bring any issues with hybrid either. The hybrid PT can be fixed separately.
Quoted 6 lines
Quoted 3 lines
I don't have much time to look at the hybrid case right now.
Would it be OK to go ahead with these patches?
I'll deal with the PEBS-via-PT on hybrid. As it stands right now, this patchset is good.
Will anyone takes these patches? Perhaps Arnaldo if no one objects? The patches still seem to apply cleanly.
Re: [PATCH V3 1/3] perf/x86: Add new event for AUX output counter index
Quoted 2 lines · Adrian Hunter
Will anyone takes these patches? Perhaps Arnaldo if no one objects? The patches still seem to apply cleanly.
I've picked up the first. But if acme wants to route the whole lot:
Arnaldo Carvalho de Melo
maintainerRe: [PATCH V3 1/3] perf/x86: Add new event for AUX output counter index
Em Wed, Oct 13, 2021 at 04:17:49PM +0200, Peter Zijlstra escreveu:
Quoted 8 lines · Adrian Hunter
Quoted 3 lines
Will anyone takes these patches? Perhaps Arnaldo if no one objects? The patches still seem to apply cleanly.
I've picked up the first. But if acme wants to route the whole lot:
I'll pick the rest, thanks.
- Arnaldo
[tip: perf/core] perf/x86: Add new event for AUX output counter index
In reply to Adrian HunterThe following commit has been merged into the perf/core branch of tip:
The following commit has been merged into the perf/core branch of tip:
Commit-ID: 8b8ff8cc3b8155c18162e8b1f70e1230db176862 Gitweb: https://git.kernel.org/tip/8b8ff8cc3b8155c18162e8b1f70e1230db176862 Author: Adrian Hunter <[email protected]> AuthorDate: Tue, 07 Sep 2021 19:39:01 +03:00 Committer: Peter Zijlstra <[email protected]> CommitterDate: Fri, 15 Oct 2021 11:25:31 +02:00
perf/x86: Add new event for AUX output counter index
PEBS-via-PT records contain a mask of applicable counters. To identify which event belongs to which counter, a side-band event is needed. Until now, there has been no side-band event, and consequently users were limited to using a single event.
Add such a side-band event. Note the event is optimised to output only when the counter index changes for an event. That works only so long as all PEBS-via-PT events are scheduled together, which they are for a recording session because they are in a single group.
Also no attribute bit is used to select the new event, so a new kernel is not compatible with older perf tools. The assumption being that PEBS-via-PT is sufficiently esoteric that users will not be troubled by this.
--- arch/x86/events/core.c | 6 ++++++ arch/x86/events/intel/core.c | 16 ++++++++++++++++ arch/x86/events/perf_event.h | 1 + include/linux/perf_event.h | 1 + include/uapi/linux/perf_event.h | 15 +++++++++++++++ kernel/events/core.c | 30 ++++++++++++++++++++++++++++++ 6 files changed, 69 insertions(+)
arch/x86/events/core.c +6 -0
@@ -66,6 +66,8 @@ DEFINE_STATIC_CALL_NULL(x86_pmu_enable_all, *x86_pmu.enable_all);DEFINE_STATIC_CALL_NULL (x86_pmu_enable, *x86_pmu.enable);DEFINE_STATIC_CALL_NULL (x86_pmu_disable, *x86_pmu.disable); +DEFINE_STATIC_CALL_NULL (x86_pmu_assign, *x86_pmu.assign); +DEFINE_STATIC_CALL_NULL (x86_pmu_add, *x86_pmu.add);DEFINE_STATIC_CALL_NULL (x86_pmu_del, *x86_pmu.del);DEFINE_STATIC_CALL_NULL (x86_pmu_read, *x86_pmu.read);
@@ -1215,6 +1217,8 @@ static inline void x86_assign_hw_event(struct perf_event *event, hwc->last_cpu =smp_processor_id (); hwc->last_tag = ++cpuc->tags[i]; +static_call_cond (x86_pmu_assign)(event, idx); +switch (hwc->idx) {case INTEL_PMC_IDX_FIXED_BTS :case INTEL_PMC_IDX_FIXED_VLBR :
@@ -2005,6 +2009,8 @@ static void x86_pmu_static_call_update(void)static_call_update (x86_pmu_enable, x86_pmu.enable);static_call_update (x86_pmu_disable, x86_pmu.disable); +static_call_update (x86_pmu_assign, x86_pmu.assign); +static_call_update (x86_pmu_add, x86_pmu.add);static_call_update (x86_pmu_del, x86_pmu.del);static_call_update (x86_pmu_read, x86_pmu.read);
arch/x86/events/intel/core.c +16 -0
@@ -2402,6 +2402,12 @@ static void intel_pmu_disable_event(struct perf_event *event)intel_pmu_pebs_disable (event); } +static void intel_pmu_assign_event (struct perf_event *event,int idx) +{ +if (is_pebs_pt (event)) +perf_report_aux_output_id (event, idx); +} +static void intel_pmu_del_event (struct perf_event *event) {if (needs_branch_stack (event))
@@ -4494,8 +4500,16 @@ static int intel_pmu_check_period(struct perf_event *event, u64 value)return intel_pmu_has_bts_period (event, value) ? -EINVAL :0 ; } +static void intel_aux_output_init (void ) +{ +/* Refer also intel_pmu_aux_output_match() */ +if (x86_pmu.intel_cap.pebs_output_pt_available) + x86_pmu.assign = intel_pmu_assign_event; +} +static int intel_pmu_aux_output_match (struct perf_event *event) { +/* intel_pmu_assign_event() is needed, refer intel_aux_output_init() */ if (!x86_pmu.intel_cap.pebs_output_pt_available)return 0 ;
@@ -6301,6 +6315,8 @@ __init int intel_pmu_init(void)if (is_hybrid ())intel_pmu_check_hybrid_pmus ((u64 )fixed_mask); +intel_aux_output_init (); +return 0 ; }
arch/x86/events/perf_event.h +1 -0
@@ -726,6 +726,7 @@ struct x86_pmu {void (*enable_all)(int added);void (*enable)(struct perf_event *);void (*disable)(struct perf_event *); +void (*assign)(struct perf_event *event,int idx);void (*add)(struct perf_event *);void (*del)(struct perf_event *);void (*read)(struct perf_event *event);
include/linux/perf_event.h +1 -0
@@ -1397,6 +1397,7 @@ perf_event_addr_filters(struct perf_event *event) }extern void perf_event_addr_filters_sync (struct perf_event *event); +extern void perf_report_aux_output_id (struct perf_event *event,u64 hw_id); extern int perf_output_begin(struct perf_output_handle *handle, struct perf_sample_data *data,
include/uapi/linux/perf_event.h +15 -0
@@ -1141,6 +1141,21 @@ enum perf_event_type {*/ PERF_RECORD_TEXT_POKE =20 , +/* +* Data written to the AUX area by hardware due to aux_output, may need +* to be matched to the event by an architecture-specific hardware ID. +* This records the hardware ID, but requires sample_id to provide the +* event ID. e.g. Intel PT uses this record to disambiguate PEBS-via-PT +* records from multiple events. +* +* struct { +* struct perf_event_header header; +* u64 hw_id; +* struct sample_id sample_id; +* }; +*/ +PERF_RECORD_AUX_OUTPUT_HW_ID =21 , + PERF_RECORD_MAX, /* non-ABI */ };
kernel/events/core.c +30 -0
@@ -9062,6 +9062,36 @@ static void perf_log_itrace_start(struct perf_event *event)perf_output_end (&handle); } +void perf_report_aux_output_id (struct perf_event *event,u64 hw_id) +{ +struct perf_output_handle handle; +struct perf_sample_data sample; +struct perf_aux_event { +struct perf_event_header header; +u64 hw_id; + } rec; +int ret; + +if (event->parent) + event = event->parent; + + rec.header.type =PERF_RECORD_AUX_OUTPUT_HW_ID ; + rec.header.misc =0 ; + rec.header.size =sizeof (rec); + rec.hw_id = hw_id; + +perf_event_header__init_id (&rec.header, &sample, event); + ret =perf_output_begin (&handle, &sample, event, rec.header.size); + +if (ret) +return ; + +perf_output_put (&handle, rec); +perf_event__output_id_sample (event, &handle, &sample); + +perf_output_end (&handle); +} +static int __perf_event_account_interrupt (struct perf_event *event,int throttle) {
Adrian Hunter
reviewerPATCH v3 2/3 perf tools: Add support for PERF_RECORD_AUX_OUTPUT_HW_ID
In reply to Adrian HunterThe PERF_RECORD_AUX_OUTPUT_HW_ID event provides a way to match AUX output data like Intel PT PEBS-via-PT back to the event that it came from, by providing a hardware ID that is present in the AUX output.
--- tools/include/uapi/linux/perf_event.h | 15 +++++++++++++++ tools/lib/perf/include/perf/event.h | 6 ++++++ tools/perf/builtin-inject.c | 4 +++- tools/perf/builtin-record.c | 2 +- tools/perf/util/event.c | 18 ++++++++++++++++++ tools/perf/util/event.h | 5 +++++ tools/perf/util/machine.c | 10 ++++++++++ tools/perf/util/machine.h | 2 ++ tools/perf/util/session.c | 5 +++++ tools/perf/util/tool.h | 1 + 10 files changed, 66 insertions(+), 2 deletions(-)
tools/include/uapi/linux/perf_event.h +15 -0
@@ -1141,6 +1141,21 @@ enum perf_event_type {*/ PERF_RECORD_TEXT_POKE =20 , +/* +* Data written to the AUX area by hardware due to aux_output, may need +* to be matched to the event by an architecture-specific hardware ID. +* This records the hardware ID, but requires sample_id to provide the +* event ID. e.g. Intel PT uses this record to disambiguate PEBS-via-PT +* records from multiple events. +* +* struct { +* struct perf_event_header header; +* u64 hw_id; +* struct sample_id sample_id; +* }; +*/ +PERF_RECORD_AUX_OUTPUT_HW_ID =21 , + PERF_RECORD_MAX, /* non-ABI */ };
tools/lib/perf/include/perf/event.h +6 -0
@@ -289,6 +289,11 @@ struct perf_record_itrace_start {__u32 tid; }; +struct perf_record_aux_output_hw_id { +struct perf_event_header header; +__u64 hw_id; +}; +struct perf_record_thread_map_entry {__u64 pid;char comm[16 ];
@@ -414,6 +419,7 @@ union perf_event {struct perf_record_auxtrace_error auxtrace_error;struct perf_record_aux aux;struct perf_record_itrace_start itrace_start; +struct perf_record_aux_output_hw_id aux_output_hw_id;struct perf_record_switch context_switch;struct perf_record_thread_map thread_map;struct perf_record_cpu_map cpu_map;
tools/perf/builtin-inject.c +3 -1
@@ -815,7 +815,8 @@ static int __cmd_inject(struct perf_inject *inject) inject->tool.auxtrace_info = perf_event__process_auxtrace_info; inject->tool.auxtrace = perf_event__process_auxtrace; inject->tool.aux = perf_event__drop_aux; - inject->tool.itrace_start = perf_event__drop_aux, + inject->tool.itrace_start = perf_event__drop_aux; + inject->tool.aux_output_hw_id = perf_event__drop_aux; inject->tool.ordered_events = true; inject->tool.ordering_requires_timestamps = true;/* Allow space in the header for new attributes */
@@ -882,6 +883,7 @@ int cmd_inject(int argc, const char **argv) .lost_samples = perf_event__repipe, .aux = perf_event__repipe, .itrace_start = perf_event__repipe, + .aux_output_hw_id = perf_event__repipe, .context_switch = perf_event__repipe, .throttle = perf_event__repipe, .unthrottle = perf_event__repipe,
tools/perf/builtin-record.c +1 -1
@@ -1409,7 +1409,7 @@ static int record__synthesize(struct record *rec, bool tail) goto out;/* Synthesize id_index before auxtrace_info */ -if (rec->opts.auxtrace_sample_mode) { +if (rec->opts.auxtrace_sample_mode || rec->opts.full_auxtrace) { err =perf_event__synthesize_id_index (tool, process_synthesized_event, session->evlist, machine);
tools/perf/util/event.c +18 -0
@@ -57,6 +57,7 @@ static const char *perf_event__names[] = { [PERF_RECORD_BPF_EVENT] = "BPF_EVENT", [PERF_RECORD_CGROUP] = "CGROUP", [PERF_RECORD_TEXT_POKE] = "TEXT_POKE", + [PERF_RECORD_AUX_OUTPUT_HW_ID] = "AUX_OUTPUT_HW_ID", [PERF_RECORD_HEADER_ATTR] = "ATTR", [PERF_RECORD_HEADER_EVENT_TYPE] = "EVENT_TYPE", [PERF_RECORD_HEADER_TRACING_DATA] = "TRACING_DATA",
@@ -237,6 +238,14 @@ int perf_event__process_itrace_start(struct perf_tool *tool __maybe_unused,return machine__process_itrace_start_event (machine, event); } +int perf_event__process_aux_output_hw_id (struct perf_tool *tool__maybe_unused , +union perf_event *event, +struct perf_sample *sample__maybe_unused , +struct machine *machine) +{ +return machine__process_aux_output_hw_id_event (machine, event); +} + int perf_event__process_lost_samples(struct perf_tool *tool __maybe_unused, union perf_event *event, struct perf_sample *sample,
@@ -407,6 +416,12 @@ size_t perf_event__fprintf_itrace_start(union perf_event *event, FILE *fp) event->itrace_start.pid, event->itrace_start.tid); } +size_t perf_event__fprintf_aux_output_hw_id (union perf_event *event,FILE *fp) +{ +return fprintf (fp," hw_id: %#" PRI_lx64"\n" , + event->aux_output_hw_id.hw_id); +} +size_t perf_event__fprintf_switch (union perf_event *event,FILE *fp) {bool out = event->header.misc &PERF_RECORD_MISC_SWITCH_OUT ;
@@ -534,6 +549,9 @@ size_t perf_event__fprintf(union perf_event *event, struct machine *machine, FILcase PERF_RECORD_TEXT_POKE : ret +=perf_event__fprintf_text_poke (event, machine, fp);break ; +case PERF_RECORD_AUX_OUTPUT_HW_ID : + ret +=perf_event__fprintf_aux_output_hw_id (event, fp); +break ;default : ret +=fprintf (fp,"\n" ); }
tools/perf/util/event.h +5 -0
@@ -330,6 +330,10 @@ int perf_event__process_itrace_start(struct perf_tool *tool,union perf_event *event, struct perf_sample *sample, struct machine *machine); +int perf_event__process_aux_output_hw_id (struct perf_tool *tool, +union perf_event *event, +struct perf_sample *sample, +struct machine *machine); int perf_event__process_switch(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample,
@@ -397,6 +401,7 @@ size_t perf_event__fprintf_mmap2(union perf_event *event, FILE *fp);size_t perf_event__fprintf_task (union perf_event *event,FILE *fp);size_t perf_event__fprintf_aux (union perf_event *event,FILE *fp);size_t perf_event__fprintf_itrace_start (union perf_event *event,FILE *fp); +size_t perf_event__fprintf_aux_output_hw_id (union perf_event *event,FILE *fp);size_t perf_event__fprintf_switch (union perf_event *event,FILE *fp);size_t perf_event__fprintf_thread_map (union perf_event *event,FILE *fp);size_t perf_event__fprintf_cpu_map (union perf_event *event,FILE *fp);
tools/perf/util/machine.c +10 -0
@@ -755,6 +755,14 @@ int machine__process_itrace_start_event(struct machine *machine __maybe_unused,return 0 ; } +int machine__process_aux_output_hw_id_event (struct machine *machine__maybe_unused , +union perf_event *event) +{ +if (dump_trace) +perf_event__fprintf_aux_output_hw_id (event, stdout); +return 0 ; +} +int machine__process_switch_event (struct machine *machine__maybe_unused ,union perf_event *event) {
@@ -2028,6 +2036,8 @@ int machine__process_event(struct machine *machine, union perf_event *event, ret =machine__process_bpf (machine, event, sample);break ;case PERF_RECORD_TEXT_POKE : ret =machine__process_text_poke (machine, event, sample);break ; +case PERF_RECORD_AUX_OUTPUT_HW_ID : + ret =machine__process_aux_output_hw_id_event (machine, event);break ;default : ret =-1 ;break ;
tools/perf/util/machine.h +2 -0
@@ -124,6 +124,8 @@ int machine__process_aux_event(struct machine *machine, union perf_event *event);int machine__process_itrace_start_event (struct machine *machine,union perf_event *event); +int machine__process_aux_output_hw_id_event (struct machine *machine, +union perf_event *event);int machine__process_switch_event (struct machine *machine,union perf_event *event); int machine__process_namespaces_event(struct machine *machine,
tools/perf/util/session.c +5 -0
@@ -509,6 +509,8 @@ void perf_tool__fill_defaults(struct perf_tool *tool) tool->bpf = perf_event__process_bpf;if (tool->text_poke ==NULL ) tool->text_poke = perf_event__process_text_poke; +if (tool->aux_output_hw_id ==NULL ) + tool->aux_output_hw_id = perf_event__process_aux_output_hw_id;if (tool->read ==NULL ) tool->read = process_event_sample_stub;if (tool->throttle ==NULL )
@@ -1000,6 +1002,7 @@ static perf_event__swap_op perf_event__swap_ops[] = { [PERF_RECORD_NAMESPACES] = perf_event__namespaces_swap, [PERF_RECORD_CGROUP] = perf_event__cgroup_swap, [PERF_RECORD_TEXT_POKE] = perf_event__text_poke_swap, + [PERF_RECORD_AUX_OUTPUT_HW_ID] = perf_event__all64_swap, [PERF_RECORD_HEADER_ATTR] = perf_event__hdr_attr_swap, [PERF_RECORD_HEADER_EVENT_TYPE] = perf_event__event_type_swap, [PERF_RECORD_HEADER_TRACING_DATA] = perf_event__tracing_data_swap,
@@ -1556,6 +1559,8 @@ static int machines__deliver_event(struct machines *machines,return tool->bpf (tool, event, sample, machine);case PERF_RECORD_TEXT_POKE :return tool->text_poke (tool, event, sample, machine); +case PERF_RECORD_AUX_OUTPUT_HW_ID : +return tool->aux_output_hw_id (tool, event, sample, machine);default : ++evlist->stats.nr_unknown_events;return -1 ;
tools/perf/util/tool.h +1 -0
@@ -53,6 +53,7 @@ struct perf_tool { lost_samples, aux, itrace_start, + aux_output_hw_id, context_switch, throttle, unthrottle,
Signature · 1 line
2.17.1
Adrian Hunter
reviewerPATCH v3 3/3 perf intel-pt: Add support for PERF_RECORD_AUX_OUTPUT_HW_ID
In reply to Adrian HunterOriginally, software only supported redirecting at most one PEBS event to Intel PT (PEBS-via-PT) because it was not able to differentiate one event from another. To overcome that, add support for the PERF_RECORD_AUX_OUTPUT_HW_ID side-band event.
--- tools/perf/Documentation/perf-intel-pt.txt | 7 +- tools/perf/util/intel-pt.c | 85 +++++++++++++++++++++- 2 files changed, 87 insertions(+), 5 deletions(-)
tools/perf/Documentation/perf-intel-pt.txt +6 -1
@@ -1144,7 +1144,12 @@ Recording is selected by using the aux-output config term e.g. perf record -c 10000 -e '{intel_pt/branch=0/,cycles/aux-output/ppp}' uname -Note that currently, software only supports redirecting at most one PEBS event. +Originally, software only supported redirecting at most one PEBS event because it +was not able to differentiate one event from another. To overcome that, more recent +kernels and perf tools add support for the PERF_RECORD_AUX_OUTPUT_HW_ID side-band event. +To check for the presence of that event in a PEBS-via-PT trace: + + perf script -D --no-itrace | grep PERF_RECORD_AUX_OUTPUT_HW_ID To display PEBS events from the Intel PT trace, use the itrace 'o' option e.g.
tools/perf/util/intel-pt.c +81 -4
141 lines not on this page · raw message
Signature · 1 line
2.17.1