perf tools: set no_aux_samples for the tracking event in PT

PATCH Applied [email protected] 4 participants 8 messages Jun 30, 2017 – Jul 26, 2017 perf events

lore

Share & tools

lore

[email protected]

author

2017-06-30 14:17 UTC #0 lore raw

PATCH v2 1/2 perf tools: set no_aux_samples for the tracking event in PT

From: Kan Liang <[email protected]>

The reason of introducing the tracking event (a dummy software event) is to collect side-band information. Additional sampling is wasteful. no_aux_samples should be set for tracking event.

Signed-off-by Kan Liang <[email protected]>

---

Changes since V1
- new patch to set no_aux_samples for the tracking event (jirka)

 tools/perf/arch/x86/util/intel-pt.c | 1 +
 1 file changed, 1 insertion(+)

tools/perf/arch/x86/util/intel-pt.c +1 -0

index f630de0..4fe1aed 100644
@@ -751,6 +751,7 @@ static int intel_pt_recording_options(struct auxtrace_record *itr,
 		tracking_evsel->attr.freq = 0;
 		tracking_evsel->attr.sample_period = 1;
 
+		tracking_evsel->no_aux_samples = true;
 		if (need_immediate)
 			tracking_evsel->immediate = true;

Signature · 1 line

2.9.4

[email protected]

author

2017-06-30 14:18 UTC #1 lore raw

PATCH v2 2/2 perf tools: always set no branch for dummy event in PT

From: Kan Liang <[email protected]>

An earlier kernel patch allowed enabling PT and LBR at the same time on Goldmont.

commit ccbebba4c6bf ("perf/x86/intel/pt: Bypass PT vs. LBR exclusivity if the core supports it")

However, users still cannot use Intel PT and LBRs simultaneously.

 $ sudo perf record -e cycles,intel_pt//u -b  -- sleep 1
 Error:
 PMU Hardware doesn't support sampling/overflow-interrupts.

PT implicitly adds dummy event in perf tool. dummy event is software event which doesn't support LBR.

Always setting no branch for dummy event in Intel PT.

Signed-off-by Kan Liang <[email protected]>

---

Changes since V1
- change the BRANCH_STACK sample bit directly (jirka)

 tools/perf/arch/x86/util/intel-pt.c | 2 ++
 1 file changed, 2 insertions(+)

tools/perf/arch/x86/util/intel-pt.c +2 -0

index 4fe1aed..2201f3b 100644
@@ -700,6 +700,7 @@ static int intel_pt_recording_options(struct auxtrace_record *itr,
 				perf_evsel__set_sample_bit(switch_evsel, TID);
 				perf_evsel__set_sample_bit(switch_evsel, TIME);
 				perf_evsel__set_sample_bit(switch_evsel, CPU);
+				perf_evsel__reset_sample_bit(switch_evsel, BRANCH_STACK);
 
 				opts->record_switch_events = false;
 				ptr->have_sched_switch = 3;

@@ -761,6 +762,7 @@ static int intel_pt_recording_options(struct auxtrace_record *itr,
 			/* And the CPU for switch events */
 			perf_evsel__set_sample_bit(tracking_evsel, CPU);
 		}
+		perf_evsel__reset_sample_bit(tracking_evsel, BRANCH_STACK);
 	}
 
 	/*

Signature · 1 line

2.9.4

Jiri Olsa

Acked-by on v2 2/2

2017-07-03 07:58 UTC #2 lore raw

Re: [PATCH V2 2/2] perf tools: always set no branch for dummy event in PT

Quoted 13 lines · On Fri, Jun 30, 2017 at 10:16:56AM -0400, [email protected] wrote:

From: Kan Liang <[email protected]>
An earlier kernel patch allowed enabling PT and LBR at the same time on Goldmont.
commit ccbebba4c6bf ("perf/x86/intel/pt: Bypass PT vs. LBR exclusivity if the core supports it")
However, users still cannot use Intel PT and LBRs simultaneously.
 $ sudo perf record -e cycles,intel_pt//u -b  -- sleep 1
 Error:
 PMU Hardware doesn't support sampling/overflow-interrupts.

Quoted 6 lines

PT implicitly adds dummy event in perf tool. dummy event is software event which doesn't support LBR.
Always setting no branch for dummy event in Intel PT.
Signed-off-by Kan Liang <[email protected]>

for the patchset:

Acked-by Jiri Olsa <[email protected]>

thanks,
jirka

Arnaldo Carvalho de Melo

maintainer Applied

2017-07-03 16:59 UTC #3 lore raw

Re: [PATCH V2 2/2] perf tools: always set no branch for dummy event in PT

Em Mon, Jul 03, 2017 at 09:58:32AM +0200, Jiri Olsa escreveu:

Quoted 17 lines · On Fri, Jun 30, 2017 at 10:16:56AM -0400, [email protected] wrote:

Quoted 16 lines
From: Kan Liang <[email protected]>
An earlier kernel patch allowed enabling PT and LBR at the same time on Goldmont.
commit ccbebba4c6bf ("perf/x86/intel/pt: Bypass PT vs. LBR exclusivity if the core supports it")
However, users still cannot use Intel PT and LBRs simultaneously.
 $ sudo perf record -e cycles,intel_pt//u -b  -- sleep 1
 Error:
 PMU Hardware doesn't support sampling/overflow-interrupts.
PT implicitly adds dummy event in perf tool. dummy event is software event which doesn't support LBR.

Quoted 7 lines

Quoted 3 lines
Always setting no branch for dummy event in Intel PT.
Signed-off-by Kan Liang <[email protected]>
for the patchset:
Acked-by Jiri Olsa <[email protected]>

Thanks, applied.

- Arnaldo

Liang, Kan

author

2017-07-19 14:28 UTC #4 lore raw

RE: [PATCH V2 2/2] perf tools: always set no branch for dummy event in PT

Quoted 22 lines

Em Mon, Jul 03, 2017 at 09:58:32AM +0200, Jiri Olsa escreveu:
Quoted 19 lines · On Fri, Jun 30, 2017 at 10:16:56AM -0400, [email protected] wrote:
Quoted 18 lines
From: Kan Liang <[email protected]>
An earlier kernel patch allowed enabling PT and LBR at the same time on Goldmont.
commit ccbebba4c6bf ("perf/x86/intel/pt: Bypass PT vs. LBR exclusivity if the core supports it")
However, users still cannot use Intel PT and LBRs simultaneously.
 $ sudo perf record -e cycles,intel_pt//u -b  -- sleep 1
 Error:
 PMU Hardware doesn't support sampling/overflow-interrupts.
PT implicitly adds dummy event in perf tool. dummy event is software event which doesn't support LBR.
Always setting no branch for dummy event in Intel PT.

Quoted 8 lines

Quoted 5 lines
Quoted 1 line
Signed-off-by Kan Liang <[email protected]>
for the patchset:
Acked-by Jiri Olsa <[email protected]>
Thanks, applied.

Hi Arnaldo,

The patch series looks miss the latest update.
Could you please have a look?

Thanks,
Kan

Arnaldo Carvalho de Melo

maintainer

2017-07-19 14:45 UTC #5 lore raw

Re: [PATCH V2 2/2] perf tools: always set no branch for dummy event in PT

Em Wed, Jul 19, 2017 at 02:28:26PM +0000, Liang, Kan escreveu:

Quoted 28 lines

Quoted 28 lines
Em Mon, Jul 03, 2017 at 09:58:32AM +0200, Jiri Olsa escreveu:
Quoted 24 lines · On Fri, Jun 30, 2017 at 10:16:56AM -0400, [email protected] wrote:
Quoted 13 lines
From: Kan Liang <[email protected]>
An earlier kernel patch allowed enabling PT and LBR at the same time on Goldmont.
commit ccbebba4c6bf ("perf/x86/intel/pt: Bypass PT vs. LBR exclusivity if the core supports it")
However, users still cannot use Intel PT and LBRs simultaneously.
 $ sudo perf record -e cycles,intel_pt//u -b  -- sleep 1
 Error:
 PMU Hardware doesn't support sampling/overflow-interrupts.
Quoted 6 lines
PT implicitly adds dummy event in perf tool. dummy event is software event which doesn't support LBR.
Always setting no branch for dummy event in Intel PT.
Signed-off-by Kan Liang <[email protected]>
for the patchset:
Acked-by Jiri Olsa <[email protected]>

Quoted 7 lines

Quoted 2 lines
Thanks, applied.
Hi Arnaldo,
The patch series looks miss the latest update.
Could you please have a look?

Will, check, thanks for verifying.

- Arnaldo

tip-bot for Kan Liang

merge bot

2017-07-26 17:16 UTC #6 lore raw

[tip:perf/core] perf intel-pt: Always set no branch for dummy event

In reply to [email protected]

perf intel-pt: Always set no branch for dummy event

Commit-ID:  91a8c5b840f2da31280e14b6268761cf14033756
Gitweb:     http://git.kernel.org/tip/91a8c5b840f2da31280e14b6268761cf14033756
Author:     Kan Liang <[email protected]>
AuthorDate: Fri, 30 Jun 2017 10:16:56 -0400
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Thu, 20 Jul 2017 09:55:51 -0300

perf intel-pt: Always set no branch for dummy event

An earlier kernel patch allowed enabling PT and LBR at the same time on Goldmont.

commit ccbebba4c6bf ("perf/x86/intel/pt: Bypass PT vs. LBR exclusivity if the core supports it")

However, users still cannot use Intel PT and LBRs simultaneously. $ sudo perf record -e cycles,intel_pt//u -b -- sleep 1 Error: PMU Hardware doesn't support sampling/overflow-interrupts.

PT implicitly adds dummy event in perf tool. dummy event is software event which doesn't support LBR.

Always setting no branch for dummy event in Intel PT.

Signed-off-by Kan Liang <[email protected]>

Acked-by Jiri Olsa <[email protected]>

Cc Adrian Hunter <[email protected]>

Cc Alexander Shishkin <[email protected]>

Cc Andi Kleen <[email protected]>

Cc Peter Zijlstra <[email protected]>

Link http://lkml.kernel.org/r/[email protected] open here

Signed-off-by Arnaldo Carvalho de Melo <[email protected]>

---
 tools/perf/arch/x86/util/intel-pt.c | 2 ++
 1 file changed, 2 insertions(+)

tools/perf/arch/x86/util/intel-pt.c +2 -0

index 4a461e8..db0ba8c 100644
@@ -701,6 +701,7 @@ static int intel_pt_recording_options(struct auxtrace_record *itr,
 				perf_evsel__set_sample_bit(switch_evsel, TID);
 				perf_evsel__set_sample_bit(switch_evsel, TIME);
 				perf_evsel__set_sample_bit(switch_evsel, CPU);
+				perf_evsel__reset_sample_bit(switch_evsel, BRANCH_STACK);
 
 				opts->record_switch_events = false;
 				ptr->have_sched_switch = 3;

@@ -762,6 +763,7 @@ static int intel_pt_recording_options(struct auxtrace_record *itr,
 			/* And the CPU for switch events */
 			perf_evsel__set_sample_bit(tracking_evsel, CPU);
 		}
+		perf_evsel__reset_sample_bit(tracking_evsel, BRANCH_STACK);
 	}
 
 	/*

tip-bot for Kan Liang

merge bot

2017-07-26 17:16 UTC #7 lore raw

[tip:perf/core] perf intel-pt: Set no_aux_samples for the tracking event

In reply to [email protected]

perf intel-pt: Set no_aux_samples for the tracking event

Commit-ID:  69d8bd8aa7d8906a1e922ae884d97f0bd7f1b269
Gitweb:     http://git.kernel.org/tip/69d8bd8aa7d8906a1e922ae884d97f0bd7f1b269
Author:     Kan Liang <[email protected]>
AuthorDate: Fri, 30 Jun 2017 10:16:55 -0400
Committer:  Arnaldo Carvalho de Melo <[email protected]>
CommitDate: Thu, 20 Jul 2017 09:55:50 -0300

perf intel-pt: Set no_aux_samples for the tracking event

The reason of introducing the tracking event (a dummy software event) is to collect side-band information. Additional sampling is wasteful. no_aux_samples should be set for tracking event.

Signed-off-by Kan Liang <[email protected]>

Acked-by Jiri Olsa <[email protected]>

Cc Adrian Hunter <[email protected]>

Cc Alexander Shishkin <[email protected]>

Cc Andi Kleen <[email protected]>

Cc Peter Zijlstra <[email protected]>

Link http://lkml.kernel.org/r/[email protected] open here

Signed-off-by Arnaldo Carvalho de Melo <[email protected]>

---
 tools/perf/arch/x86/util/intel-pt.c | 1 +
 1 file changed, 1 insertion(+)

tools/perf/arch/x86/util/intel-pt.c +1 -0

index 9535be5..4a461e8 100644
@@ -752,6 +752,7 @@ static int intel_pt_recording_options(struct auxtrace_record *itr,
 		tracking_evsel->attr.freq = 0;
 		tracking_evsel->attr.sample_period = 1;
 
+		tracking_evsel->no_aux_samples = true;
 		if (need_immediate)
 			tracking_evsel->immediate = true;