From: Ravi Bangoria <ravi.bangoria@linux.ibm.com> To: linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org Cc: eranian@google.com, peterz@infradead.org, mpe@ellerman.id.au, paulus@samba.org, mingo@redhat.com, acme@kernel.org, mark.rutland@arm.com, alexander.shishkin@linux.intel.com, jolsa@redhat.com, namhyung@kernel.org, adrian.hunter@intel.com, ak@linux.intel.com, kan.liang@linux.intel.com, alexey.budankov@linux.intel.com, yao.jin@linux.intel.com, robert.richter@amd.com, kim.phillips@amd.com, maddy@linux.ibm.com, ravi.bangoria@linux.ibm.com, Madhavan Srinivasan <maddy@linux.vnet.ibm.com> Subject: [RFC 05/11] perf tools: Enable record and script to record and show hazard data Date: Mon, 2 Mar 2020 10:53:49 +0530 [thread overview] Message-ID: <20200302052355.36365-6-ravi.bangoria@linux.ibm.com> (raw) In-Reply-To: <20200302052355.36365-1-ravi.bangoria@linux.ibm.com> From: Madhavan Srinivasan <maddy@linux.vnet.ibm.com> Introduce new perf record option "--hazard" to capture cpu pipeline hazard data. Also enable perf script -D to dump raw values of it. Sample o/p: $ ./perf record -e r4010e --hazard -- ls $ ./perf script -D ... PERF_RECORD_SAMPLE(IP, 0x2): ... hazard information: Inst Type 0x1 Inst Cache 0x1 Hazard Stage 0x4 Hazard Reason 0x3 Stall Stage 0x4 Stall Reason 0x2 Signed-off-by: Madhavan Srinivasan <maddy@linux.vnet.ibm.com> Signed-off-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com> --- tools/perf/Documentation/perf-record.txt | 3 +++ tools/perf/builtin-record.c | 1 + tools/perf/util/event.h | 1 + tools/perf/util/evsel.c | 10 ++++++++++ tools/perf/util/perf_event_attr_fprintf.c | 1 + tools/perf/util/record.h | 1 + tools/perf/util/session.c | 16 ++++++++++++++++ 7 files changed, 33 insertions(+) diff --git a/tools/perf/Documentation/perf-record.txt b/tools/perf/Documentation/perf-record.txt index b23a4012a606..e7bd1b6938ce 100644 --- a/tools/perf/Documentation/perf-record.txt +++ b/tools/perf/Documentation/perf-record.txt @@ -283,6 +283,9 @@ OPTIONS --phys-data:: Record the sample physical addresses. +--hazard:: + Record processor pipeline hazard and stall information. + -T:: --timestamp:: Record the sample timestamps. Use it with 'perf report -D' to see the diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 4c301466101b..6bd32d7bc4e9 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -2301,6 +2301,7 @@ static struct option __record_options[] = { OPT_BOOLEAN('s', "stat", &record.opts.inherit_stat, "per thread counts"), OPT_BOOLEAN('d', "data", &record.opts.sample_address, "Record the sample addresses"), + OPT_BOOLEAN(0, "hazard", &record.opts.hazard, "Record processor pipeline hazard and stall information"), OPT_BOOLEAN(0, "phys-data", &record.opts.sample_phys_addr, "Record the sample physical addresses"), OPT_BOOLEAN(0, "sample-cpu", &record.opts.sample_cpu, "Record the sample cpu"), diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index 85223159737c..ff0f03253a95 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -148,6 +148,7 @@ struct perf_sample { struct stack_dump user_stack; struct sample_read read; struct aux_sample aux_sample; + struct perf_pipeline_haz_data *pipeline_haz; }; #define PERF_MEM_DATA_SRC_NONE \ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index c8dc4450884c..e37ed7929c2c 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -1080,6 +1080,9 @@ void perf_evsel__config(struct evsel *evsel, struct record_opts *opts, if (opts->sample_phys_addr) perf_evsel__set_sample_bit(evsel, PHYS_ADDR); + if (opts->hazard) + perf_evsel__set_sample_bit(evsel, PIPELINE_HAZ); + if (opts->no_buffering) { attr->watermark = 0; attr->wakeup_events = 1; @@ -2265,6 +2268,13 @@ int perf_evsel__parse_sample(struct evsel *evsel, union perf_event *event, array = (void *)array + sz; } + if (type & PERF_SAMPLE_PIPELINE_HAZ) { + sz = sizeof(struct perf_pipeline_haz_data); + OVERFLOW_CHECK(array, sz, max_size); + data->pipeline_haz = (struct perf_pipeline_haz_data *)array; + array = (void *)array + sz; + } + return 0; } diff --git a/tools/perf/util/perf_event_attr_fprintf.c b/tools/perf/util/perf_event_attr_fprintf.c index 651203126c71..d97e755c886b 100644 --- a/tools/perf/util/perf_event_attr_fprintf.c +++ b/tools/perf/util/perf_event_attr_fprintf.c @@ -35,6 +35,7 @@ static void __p_sample_type(char *buf, size_t size, u64 value) bit_name(BRANCH_STACK), bit_name(REGS_USER), bit_name(STACK_USER), bit_name(IDENTIFIER), bit_name(REGS_INTR), bit_name(DATA_SRC), bit_name(WEIGHT), bit_name(PHYS_ADDR), bit_name(AUX), + bit_name(PIPELINE_HAZ), { .name = NULL, } }; #undef bit_name diff --git a/tools/perf/util/record.h b/tools/perf/util/record.h index 5421fd2ad383..f1678a0bc8ce 100644 --- a/tools/perf/util/record.h +++ b/tools/perf/util/record.h @@ -67,6 +67,7 @@ struct record_opts { int affinity; int mmap_flush; unsigned int comp_level; + bool hazard; }; extern const char * const *record_usage; diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index d0d7d25b23e3..834ca7df2349 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -1153,6 +1153,19 @@ static void stack_user__printf(struct stack_dump *dump) dump->size, dump->offset); } +static void pipeline_hazard__printf(struct perf_sample *sample) +{ + struct perf_pipeline_haz_data *haz = sample->pipeline_haz; + + printf("... hazard information:\n"); + printf(".... Inst Type 0x%" PRIx32 "\n", haz->itype); + printf(".... Inst Cache 0x%" PRIx32 "\n", haz->icache); + printf(".... Hazard Stage 0x%" PRIx32 "\n", haz->hazard_stage); + printf(".... Hazard Reason 0x%" PRIx32 "\n", haz->hazard_reason); + printf(".... Stall Stage 0x%" PRIx32 "\n", haz->stall_stage); + printf(".... Stall Reason 0x%" PRIx32 "\n", haz->stall_reason); +} + static void perf_evlist__print_tstamp(struct evlist *evlist, union perf_event *event, struct perf_sample *sample) @@ -1251,6 +1264,9 @@ static void dump_sample(struct evsel *evsel, union perf_event *event, if (sample_type & PERF_SAMPLE_STACK_USER) stack_user__printf(&sample->user_stack); + if (sample_type & PERF_SAMPLE_PIPELINE_HAZ) + pipeline_hazard__printf(sample); + if (sample_type & PERF_SAMPLE_WEIGHT) printf("... weight: %" PRIu64 "\n", sample->weight); -- 2.21.1
WARNING: multiple messages have this Message-ID (diff)
From: Ravi Bangoria <ravi.bangoria@linux.ibm.com> To: linuxppc-dev@lists.ozlabs.org, linux-kernel@vger.kernel.org Cc: mark.rutland@arm.com, ravi.bangoria@linux.ibm.com, ak@linux.intel.com, maddy@linux.ibm.com, peterz@infradead.org, alexey.budankov@linux.intel.com, Madhavan Srinivasan <maddy@linux.vnet.ibm.com>, adrian.hunter@intel.com, acme@kernel.org, alexander.shishkin@linux.intel.com, yao.jin@linux.intel.com, mingo@redhat.com, paulus@samba.org, eranian@google.com, robert.richter@amd.com, namhyung@kernel.org, kim.phillips@amd.com, jolsa@redhat.com, kan.liang@linux.intel.com Subject: [RFC 05/11] perf tools: Enable record and script to record and show hazard data Date: Mon, 2 Mar 2020 10:53:49 +0530 [thread overview] Message-ID: <20200302052355.36365-6-ravi.bangoria@linux.ibm.com> (raw) In-Reply-To: <20200302052355.36365-1-ravi.bangoria@linux.ibm.com> From: Madhavan Srinivasan <maddy@linux.vnet.ibm.com> Introduce new perf record option "--hazard" to capture cpu pipeline hazard data. Also enable perf script -D to dump raw values of it. Sample o/p: $ ./perf record -e r4010e --hazard -- ls $ ./perf script -D ... PERF_RECORD_SAMPLE(IP, 0x2): ... hazard information: Inst Type 0x1 Inst Cache 0x1 Hazard Stage 0x4 Hazard Reason 0x3 Stall Stage 0x4 Stall Reason 0x2 Signed-off-by: Madhavan Srinivasan <maddy@linux.vnet.ibm.com> Signed-off-by: Ravi Bangoria <ravi.bangoria@linux.ibm.com> --- tools/perf/Documentation/perf-record.txt | 3 +++ tools/perf/builtin-record.c | 1 + tools/perf/util/event.h | 1 + tools/perf/util/evsel.c | 10 ++++++++++ tools/perf/util/perf_event_attr_fprintf.c | 1 + tools/perf/util/record.h | 1 + tools/perf/util/session.c | 16 ++++++++++++++++ 7 files changed, 33 insertions(+) diff --git a/tools/perf/Documentation/perf-record.txt b/tools/perf/Documentation/perf-record.txt index b23a4012a606..e7bd1b6938ce 100644 --- a/tools/perf/Documentation/perf-record.txt +++ b/tools/perf/Documentation/perf-record.txt @@ -283,6 +283,9 @@ OPTIONS --phys-data:: Record the sample physical addresses. +--hazard:: + Record processor pipeline hazard and stall information. + -T:: --timestamp:: Record the sample timestamps. Use it with 'perf report -D' to see the diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 4c301466101b..6bd32d7bc4e9 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -2301,6 +2301,7 @@ static struct option __record_options[] = { OPT_BOOLEAN('s', "stat", &record.opts.inherit_stat, "per thread counts"), OPT_BOOLEAN('d', "data", &record.opts.sample_address, "Record the sample addresses"), + OPT_BOOLEAN(0, "hazard", &record.opts.hazard, "Record processor pipeline hazard and stall information"), OPT_BOOLEAN(0, "phys-data", &record.opts.sample_phys_addr, "Record the sample physical addresses"), OPT_BOOLEAN(0, "sample-cpu", &record.opts.sample_cpu, "Record the sample cpu"), diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index 85223159737c..ff0f03253a95 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -148,6 +148,7 @@ struct perf_sample { struct stack_dump user_stack; struct sample_read read; struct aux_sample aux_sample; + struct perf_pipeline_haz_data *pipeline_haz; }; #define PERF_MEM_DATA_SRC_NONE \ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index c8dc4450884c..e37ed7929c2c 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -1080,6 +1080,9 @@ void perf_evsel__config(struct evsel *evsel, struct record_opts *opts, if (opts->sample_phys_addr) perf_evsel__set_sample_bit(evsel, PHYS_ADDR); + if (opts->hazard) + perf_evsel__set_sample_bit(evsel, PIPELINE_HAZ); + if (opts->no_buffering) { attr->watermark = 0; attr->wakeup_events = 1; @@ -2265,6 +2268,13 @@ int perf_evsel__parse_sample(struct evsel *evsel, union perf_event *event, array = (void *)array + sz; } + if (type & PERF_SAMPLE_PIPELINE_HAZ) { + sz = sizeof(struct perf_pipeline_haz_data); + OVERFLOW_CHECK(array, sz, max_size); + data->pipeline_haz = (struct perf_pipeline_haz_data *)array; + array = (void *)array + sz; + } + return 0; } diff --git a/tools/perf/util/perf_event_attr_fprintf.c b/tools/perf/util/perf_event_attr_fprintf.c index 651203126c71..d97e755c886b 100644 --- a/tools/perf/util/perf_event_attr_fprintf.c +++ b/tools/perf/util/perf_event_attr_fprintf.c @@ -35,6 +35,7 @@ static void __p_sample_type(char *buf, size_t size, u64 value) bit_name(BRANCH_STACK), bit_name(REGS_USER), bit_name(STACK_USER), bit_name(IDENTIFIER), bit_name(REGS_INTR), bit_name(DATA_SRC), bit_name(WEIGHT), bit_name(PHYS_ADDR), bit_name(AUX), + bit_name(PIPELINE_HAZ), { .name = NULL, } }; #undef bit_name diff --git a/tools/perf/util/record.h b/tools/perf/util/record.h index 5421fd2ad383..f1678a0bc8ce 100644 --- a/tools/perf/util/record.h +++ b/tools/perf/util/record.h @@ -67,6 +67,7 @@ struct record_opts { int affinity; int mmap_flush; unsigned int comp_level; + bool hazard; }; extern const char * const *record_usage; diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index d0d7d25b23e3..834ca7df2349 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -1153,6 +1153,19 @@ static void stack_user__printf(struct stack_dump *dump) dump->size, dump->offset); } +static void pipeline_hazard__printf(struct perf_sample *sample) +{ + struct perf_pipeline_haz_data *haz = sample->pipeline_haz; + + printf("... hazard information:\n"); + printf(".... Inst Type 0x%" PRIx32 "\n", haz->itype); + printf(".... Inst Cache 0x%" PRIx32 "\n", haz->icache); + printf(".... Hazard Stage 0x%" PRIx32 "\n", haz->hazard_stage); + printf(".... Hazard Reason 0x%" PRIx32 "\n", haz->hazard_reason); + printf(".... Stall Stage 0x%" PRIx32 "\n", haz->stall_stage); + printf(".... Stall Reason 0x%" PRIx32 "\n", haz->stall_reason); +} + static void perf_evlist__print_tstamp(struct evlist *evlist, union perf_event *event, struct perf_sample *sample) @@ -1251,6 +1264,9 @@ static void dump_sample(struct evsel *evsel, union perf_event *event, if (sample_type & PERF_SAMPLE_STACK_USER) stack_user__printf(&sample->user_stack); + if (sample_type & PERF_SAMPLE_PIPELINE_HAZ) + pipeline_hazard__printf(sample); + if (sample_type & PERF_SAMPLE_WEIGHT) printf("... weight: %" PRIu64 "\n", sample->weight); -- 2.21.1
next prev parent reply other threads:[~2020-03-02 5:24 UTC|newest] Thread overview: 73+ messages / expand[flat|nested] mbox.gz Atom feed top 2020-03-02 5:23 [RFC 00/11] perf: Enhancing perf to export processor hazard information Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 5:23 ` [RFC 01/11] powerpc/perf: Simplify ISA207_SIER macros Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 5:23 ` [RFC 02/11] perf/core: Data structure to present hazard data Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 9:55 ` Peter Zijlstra 2020-03-02 9:55 ` Peter Zijlstra 2020-03-02 14:23 ` maddy 2020-03-02 14:23 ` maddy 2020-03-02 14:48 ` Mark Rutland 2020-03-02 14:48 ` Mark Rutland 2020-03-03 14:32 ` Ravi Bangoria 2020-03-03 14:32 ` Ravi Bangoria 2020-03-02 14:54 ` Mark Rutland 2020-03-02 14:54 ` Mark Rutland 2020-03-03 14:31 ` Ravi Bangoria 2020-03-03 14:31 ` Ravi Bangoria 2020-03-02 5:23 ` [RFC 03/11] powerpc/perf: Arch specific definitions for pipeline Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 5:23 ` [RFC 04/11] powerpc/perf: Arch support to expose Hazard data Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria [this message] 2020-03-02 5:23 ` [RFC 05/11] perf tools: Enable record and script to record and show hazard data Ravi Bangoria 2020-03-02 5:23 ` [RFC 06/11] perf hists: Make a room for hazard info in struct hist_entry Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 5:23 ` [RFC 07/11] perf hazard: Functions to convert generic hazard data to arch specific string Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 5:23 ` [RFC 08/11] perf report: Enable hazard mode Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 5:23 ` [RFC 09/11] perf annotate: Introduce type for annotation_line Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 5:23 ` [RFC 10/11] perf annotate: Preparation for hazard Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 5:23 ` [RFC 11/11] perf annotate: Show hazard data in tui mode Ravi Bangoria 2020-03-02 5:23 ` Ravi Bangoria 2020-03-02 10:13 ` [RFC 00/11] perf: Enhancing perf to export processor hazard information Peter Zijlstra 2020-03-02 10:13 ` Peter Zijlstra 2020-03-02 20:21 ` Stephane Eranian 2020-03-02 20:21 ` Stephane Eranian 2020-03-02 22:25 ` Kim Phillips 2020-03-02 22:25 ` Kim Phillips 2020-03-05 4:46 ` Ravi Bangoria 2020-03-05 4:46 ` Ravi Bangoria 2020-03-05 22:06 ` Kim Phillips 2020-03-05 22:06 ` Kim Phillips 2020-03-11 16:00 ` Ravi Bangoria 2020-03-12 22:38 ` Kim Phillips 2020-03-12 22:38 ` Kim Phillips 2020-03-17 6:50 ` maddy 2020-03-17 6:50 ` maddy 2020-03-18 17:35 ` Kim Phillips 2020-03-18 17:35 ` Kim Phillips 2020-03-19 11:22 ` Michael Ellerman 2020-03-19 11:22 ` Michael Ellerman 2020-03-26 10:19 ` maddy 2020-03-26 10:19 ` maddy 2020-03-26 19:48 ` Kim Phillips 2020-03-26 19:48 ` Kim Phillips 2020-04-20 7:09 ` Madhavan Srinivasan 2020-04-20 7:09 ` Madhavan Srinivasan 2020-04-27 7:18 ` Madhavan Srinivasan 2020-04-27 7:18 ` Madhavan Srinivasan 2020-03-05 4:28 ` maddy 2020-03-05 4:28 ` maddy 2020-03-03 1:33 ` Andi Kleen 2020-03-03 1:33 ` Andi Kleen 2020-03-05 5:06 ` Ravi Bangoria 2020-03-05 5:06 ` Ravi Bangoria 2020-03-02 21:08 ` Paul Clarke 2020-03-02 21:08 ` Paul Clarke 2020-03-05 5:06 ` Ravi Bangoria 2020-03-05 5:06 ` Ravi Bangoria
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=20200302052355.36365-6-ravi.bangoria@linux.ibm.com \ --to=ravi.bangoria@linux.ibm.com \ --cc=acme@kernel.org \ --cc=adrian.hunter@intel.com \ --cc=ak@linux.intel.com \ --cc=alexander.shishkin@linux.intel.com \ --cc=alexey.budankov@linux.intel.com \ --cc=eranian@google.com \ --cc=jolsa@redhat.com \ --cc=kan.liang@linux.intel.com \ --cc=kim.phillips@amd.com \ --cc=linux-kernel@vger.kernel.org \ --cc=linuxppc-dev@lists.ozlabs.org \ --cc=maddy@linux.ibm.com \ --cc=maddy@linux.vnet.ibm.com \ --cc=mark.rutland@arm.com \ --cc=mingo@redhat.com \ --cc=mpe@ellerman.id.au \ --cc=namhyung@kernel.org \ --cc=paulus@samba.org \ --cc=peterz@infradead.org \ --cc=robert.richter@amd.com \ --cc=yao.jin@linux.intel.com \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: linkBe sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes, see mirroring instructions on how to clone and mirror all data and code used by this external index.