[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20250628045017.1361563-22-irogers@google.com>
Date: Fri, 27 Jun 2025 21:50:15 -0700
From: Ian Rogers <irogers@...gle.com>
To: Peter Zijlstra <peterz@...radead.org>, Ingo Molnar <mingo@...hat.com>,
Arnaldo Carvalho de Melo <acme@...nel.org>, Namhyung Kim <namhyung@...nel.org>,
Mark Rutland <mark.rutland@....com>,
Alexander Shishkin <alexander.shishkin@...ux.intel.com>, Jiri Olsa <jolsa@...nel.org>,
Ian Rogers <irogers@...gle.com>, Adrian Hunter <adrian.hunter@...el.com>,
Kan Liang <kan.liang@...ux.intel.com>, John Garry <john.g.garry@...cle.com>,
Will Deacon <will@...nel.org>, James Clark <james.clark@...aro.org>,
Mike Leach <mike.leach@...aro.org>, Leo Yan <leo.yan@...ux.dev>,
"Masami Hiramatsu (Google)" <mhiramat@...nel.org>, Ravi Bangoria <ravi.bangoria@....com>,
Charlie Jenkins <charlie@...osinc.com>, Colin Ian King <colin.i.king@...il.com>,
Andi Kleen <ak@...ux.intel.com>, Dmitry Vyukov <dvyukov@...gle.com>,
Graham Woodward <graham.woodward@....com>, Ilkka Koskinen <ilkka@...amperecomputing.com>,
Zhongqiu Han <quic_zhonhan@...cinc.com>, Yicong Yang <yangyicong@...ilicon.com>,
Athira Rajeev <atrajeev@...ux.ibm.com>, Kajol Jain <kjain@...ux.ibm.com>,
Li Huafei <lihuafei1@...wei.com>, "Steinar H. Gunderson" <sesse@...gle.com>,
Stephen Brennan <stephen.s.brennan@...cle.com>, Chun-Tse Shao <ctshao@...gle.com>,
Yujie Liu <yujie.liu@...el.com>, "Dr. David Alan Gilbert" <linux@...blig.org>, Levi Yun <yeoreum.yun@....com>,
Howard Chu <howardchu95@...il.com>, Weilin Wang <weilin.wang@...el.com>,
Thomas Falcon <thomas.falcon@...el.com>, Matt Fleming <matt@...dmodwrite.com>,
Veronika Molnarova <vmolnaro@...hat.com>,
"Krzysztof Łopatowski" <krzysztof.m.lopatowski@...il.com>, Zixian Cai <fzczx123@...il.com>,
Steve Clevenger <scclevenger@...amperecomputing.com>, Ben Gainey <ben.gainey@....com>,
Chaitanya S Prakash <chaitanyas.prakash@....com>, Martin Liska <martin.liska@....com>,
"Martin Liška" <m.liska@...link.cz>, Song Liu <song@...nel.org>, linux-kernel@...r.kernel.org,
linux-perf-users@...r.kernel.org, linux-arm-kernel@...ts.infradead.org
Subject: [PATCH v5 21/23] perf sample: Remove arch notion of sample parsing
By definition arch sample parsing and synthesis will inhibit certain
kinds of cross-platform record then analysis (report, script,
etc.). Remove arch_perf_parse_sample_weight and
arch_perf_synthesize_sample_weight replacing with a common
implementation. Combine perf_sample p_stage_cyc and retire_lat as
weight3 to capture the differing uses regardless of compiled for
architecture.
Signed-off-by: Ian Rogers <irogers@...gle.com>
---
tools/perf/arch/powerpc/util/event.c | 26 ---------------------
tools/perf/arch/x86/tests/sample-parsing.c | 4 ++--
tools/perf/arch/x86/util/event.c | 27 ----------------------
tools/perf/builtin-script.c | 2 +-
tools/perf/util/dlfilter.c | 2 +-
tools/perf/util/event.h | 2 --
tools/perf/util/evsel.c | 17 ++++++++++----
tools/perf/util/hist.c | 4 ++--
tools/perf/util/hist.h | 3 ++-
tools/perf/util/intel-tpebs.c | 4 ++--
tools/perf/util/sample.h | 6 ++---
tools/perf/util/session.c | 2 +-
tools/perf/util/sort.c | 7 +++---
tools/perf/util/synthetic-events.c | 10 ++++++--
14 files changed, 36 insertions(+), 80 deletions(-)
diff --git a/tools/perf/arch/powerpc/util/event.c b/tools/perf/arch/powerpc/util/event.c
index 77d8cc2b5691..024ac8b54c33 100644
--- a/tools/perf/arch/powerpc/util/event.c
+++ b/tools/perf/arch/powerpc/util/event.c
@@ -11,32 +11,6 @@
#include "../../../util/debug.h"
#include "../../../util/sample.h"
-void arch_perf_parse_sample_weight(struct perf_sample *data,
- const __u64 *array, u64 type)
-{
- union perf_sample_weight weight;
-
- weight.full = *array;
- if (type & PERF_SAMPLE_WEIGHT)
- data->weight = weight.full;
- else {
- data->weight = weight.var1_dw;
- data->ins_lat = weight.var2_w;
- data->p_stage_cyc = weight.var3_w;
- }
-}
-
-void arch_perf_synthesize_sample_weight(const struct perf_sample *data,
- __u64 *array, u64 type)
-{
- *array = data->weight;
-
- if (type & PERF_SAMPLE_WEIGHT_STRUCT) {
- *array &= 0xffffffff;
- *array |= ((u64)data->ins_lat << 32);
- }
-}
-
const char *arch_perf_header_entry(const char *se_header)
{
if (!strcmp(se_header, "Local INSTR Latency"))
diff --git a/tools/perf/arch/x86/tests/sample-parsing.c b/tools/perf/arch/x86/tests/sample-parsing.c
index a061e8619267..22feec23e53d 100644
--- a/tools/perf/arch/x86/tests/sample-parsing.c
+++ b/tools/perf/arch/x86/tests/sample-parsing.c
@@ -29,7 +29,7 @@ static bool samples_same(const struct perf_sample *s1,
{
if (type & PERF_SAMPLE_WEIGHT_STRUCT) {
COMP(ins_lat);
- COMP(retire_lat);
+ COMP(weight3);
}
return true;
@@ -50,7 +50,7 @@ static int do_test(u64 sample_type)
struct perf_sample sample = {
.weight = 101,
.ins_lat = 102,
- .retire_lat = 103,
+ .weight3 = 103,
};
struct perf_sample sample_out;
size_t i, sz, bufsz;
diff --git a/tools/perf/arch/x86/util/event.c b/tools/perf/arch/x86/util/event.c
index a0400707180c..576c1c36046c 100644
--- a/tools/perf/arch/x86/util/event.c
+++ b/tools/perf/arch/x86/util/event.c
@@ -92,33 +92,6 @@ int perf_event__synthesize_extra_kmaps(const struct perf_tool *tool,
#endif
-void arch_perf_parse_sample_weight(struct perf_sample *data,
- const __u64 *array, u64 type)
-{
- union perf_sample_weight weight;
-
- weight.full = *array;
- if (type & PERF_SAMPLE_WEIGHT)
- data->weight = weight.full;
- else {
- data->weight = weight.var1_dw;
- data->ins_lat = weight.var2_w;
- data->retire_lat = weight.var3_w;
- }
-}
-
-void arch_perf_synthesize_sample_weight(const struct perf_sample *data,
- __u64 *array, u64 type)
-{
- *array = data->weight;
-
- if (type & PERF_SAMPLE_WEIGHT_STRUCT) {
- *array &= 0xffffffff;
- *array |= ((u64)data->ins_lat << 32);
- *array |= ((u64)data->retire_lat << 48);
- }
-}
-
const char *arch_perf_header_entry(const char *se_header)
{
if (!strcmp(se_header, "Local Pipeline Stage Cycle"))
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 2c25eda4be26..f7bdf9a00d00 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -2253,7 +2253,7 @@ static void process_event(struct perf_script *script,
fprintf(fp, "%16" PRIu16, sample->ins_lat);
if (PRINT_FIELD(RETIRE_LAT))
- fprintf(fp, "%16" PRIu16, sample->retire_lat);
+ fprintf(fp, "%16" PRIu16, sample->weight3);
if (PRINT_FIELD(CGROUP)) {
const char *cgrp_name;
diff --git a/tools/perf/util/dlfilter.c b/tools/perf/util/dlfilter.c
index ddacef881af2..c0afcbd954f8 100644
--- a/tools/perf/util/dlfilter.c
+++ b/tools/perf/util/dlfilter.c
@@ -513,6 +513,7 @@ int dlfilter__do_filter_event(struct dlfilter *d,
d->d_addr_al = &d_addr_al;
d_sample.size = sizeof(d_sample);
+ d_sample.p_stage_cyc = sample->weight3;
d_ip_al.size = 0; /* To indicate d_ip_al is not initialized */
d_addr_al.size = 0; /* To indicate d_addr_al is not initialized */
@@ -526,7 +527,6 @@ int dlfilter__do_filter_event(struct dlfilter *d,
ASSIGN(period);
ASSIGN(weight);
ASSIGN(ins_lat);
- ASSIGN(p_stage_cyc);
ASSIGN(transaction);
ASSIGN(insn_cnt);
ASSIGN(cyc_cnt);
diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h
index 67ad4a2014bc..b13385a6068b 100644
--- a/tools/perf/util/event.h
+++ b/tools/perf/util/event.h
@@ -391,8 +391,6 @@ extern unsigned int proc_map_timeout;
#define PAGE_SIZE_NAME_LEN 32
char *get_page_size_name(u64 size, char *str);
-void arch_perf_parse_sample_weight(struct perf_sample *data, const __u64 *array, u64 type);
-void arch_perf_synthesize_sample_weight(const struct perf_sample *data, __u64 *array, u64 type);
const char *arch_perf_header_entry(const char *se_header);
int arch_support_sort_key(const char *sort_key);
diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index 988fe5c4137a..9f1d7ec0386f 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -2847,11 +2847,18 @@ perf_event__check_size(union perf_event *event, unsigned int sample_size)
return 0;
}
-void __weak arch_perf_parse_sample_weight(struct perf_sample *data,
- const __u64 *array,
- u64 type __maybe_unused)
+static void perf_parse_sample_weight(struct perf_sample *data, const __u64 *array, u64 type)
{
- data->weight = *array;
+ union perf_sample_weight weight;
+
+ weight.full = *array;
+ if (type & PERF_SAMPLE_WEIGHT_STRUCT) {
+ data->weight = weight.var1_dw;
+ data->ins_lat = weight.var2_w;
+ data->weight3 = weight.var3_w;
+ } else {
+ data->weight = weight.full;
+ }
}
u64 evsel__bitfield_swap_branch_flags(u64 value)
@@ -3237,7 +3244,7 @@ int evsel__parse_sample(struct evsel *evsel, union perf_event *event,
if (type & PERF_SAMPLE_WEIGHT_TYPE) {
OVERFLOW_CHECK_u64(array);
- arch_perf_parse_sample_weight(data, array, type);
+ perf_parse_sample_weight(data, array, type);
array++;
}
diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
index afc6855327ab..64ff427040c3 100644
--- a/tools/perf/util/hist.c
+++ b/tools/perf/util/hist.c
@@ -829,7 +829,7 @@ __hists__add_entry(struct hists *hists,
.period = sample->period,
.weight1 = sample->weight,
.weight2 = sample->ins_lat,
- .weight3 = sample->p_stage_cyc,
+ .weight3 = sample->weight3,
.latency = al->latency,
},
.parent = sym_parent,
@@ -846,7 +846,7 @@ __hists__add_entry(struct hists *hists,
.time = hist_time(sample->time),
.weight = sample->weight,
.ins_lat = sample->ins_lat,
- .p_stage_cyc = sample->p_stage_cyc,
+ .weight3 = sample->weight3,
.simd_flags = sample->simd_flags,
}, *he = hists__findnew_entry(hists, &entry, al, sample_self);
diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h
index c64254088fc7..70438d03ca9c 100644
--- a/tools/perf/util/hist.h
+++ b/tools/perf/util/hist.h
@@ -255,7 +255,8 @@ struct hist_entry {
u64 code_page_size;
u64 weight;
u64 ins_lat;
- u64 p_stage_cyc;
+ /** @weight3: On x86 holds retire_lat, on powerpc holds p_stage_cyc. */
+ u64 weight3;
s32 socket;
s32 cpu;
int parallelism;
diff --git a/tools/perf/util/intel-tpebs.c b/tools/perf/util/intel-tpebs.c
index 3b92ebf5c112..8c9aee157ec4 100644
--- a/tools/perf/util/intel-tpebs.c
+++ b/tools/perf/util/intel-tpebs.c
@@ -210,8 +210,8 @@ static int process_sample_event(const struct perf_tool *tool __maybe_unused,
* latency value will be used. Save the number of samples and the sum of
* retire latency value for each event.
*/
- t->last = sample->retire_lat;
- update_stats(&t->stats, sample->retire_lat);
+ t->last = sample->weight3;
+ update_stats(&t->stats, sample->weight3);
mutex_unlock(tpebs_mtx_get());
return 0;
}
diff --git a/tools/perf/util/sample.h b/tools/perf/util/sample.h
index 0e96240052e9..fae834144ef4 100644
--- a/tools/perf/util/sample.h
+++ b/tools/perf/util/sample.h
@@ -104,10 +104,8 @@ struct perf_sample {
u8 cpumode;
u16 misc;
u16 ins_lat;
- union {
- u16 p_stage_cyc;
- u16 retire_lat;
- };
+ /** @weight3: On x86 holds retire_lat, on powerpc holds p_stage_cyc. */
+ u16 weight3;
bool no_hw_idx; /* No hw_idx collected in branch_stack */
char insn[MAX_INSN];
void *raw_data;
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index df03743ac2bf..daf72362d26a 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -1100,7 +1100,7 @@ static void dump_sample(struct evsel *evsel, union perf_event *event,
printf("... weight: %" PRIu64 "", sample->weight);
if (sample_type & PERF_SAMPLE_WEIGHT_STRUCT) {
printf(",0x%"PRIx16"", sample->ins_lat);
- printf(",0x%"PRIx16"", sample->p_stage_cyc);
+ printf(",0x%"PRIx16"", sample->weight3);
}
printf("\n");
}
diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c
index 7969d64a47bf..0ba2ce1b1c07 100644
--- a/tools/perf/util/sort.c
+++ b/tools/perf/util/sort.c
@@ -1884,21 +1884,20 @@ struct sort_entry sort_global_ins_lat = {
static int64_t
sort__p_stage_cyc_cmp(struct hist_entry *left, struct hist_entry *right)
{
- return left->p_stage_cyc - right->p_stage_cyc;
+ return left->weight3 - right->weight3;
}
static int hist_entry__global_p_stage_cyc_snprintf(struct hist_entry *he, char *bf,
size_t size, unsigned int width)
{
- return repsep_snprintf(bf, size, "%-*u", width,
- he->p_stage_cyc * he->stat.nr_events);
+ return repsep_snprintf(bf, size, "%-*u", width, he->weight3 * he->stat.nr_events);
}
static int hist_entry__p_stage_cyc_snprintf(struct hist_entry *he, char *bf,
size_t size, unsigned int width)
{
- return repsep_snprintf(bf, size, "%-*u", width, he->p_stage_cyc);
+ return repsep_snprintf(bf, size, "%-*u", width, he->weight3);
}
struct sort_entry sort_local_p_stage_cyc = {
diff --git a/tools/perf/util/synthetic-events.c b/tools/perf/util/synthetic-events.c
index 638d7dd7fa4b..6afdcfd14224 100644
--- a/tools/perf/util/synthetic-events.c
+++ b/tools/perf/util/synthetic-events.c
@@ -1573,10 +1573,16 @@ size_t perf_event__sample_event_size(const struct perf_sample *sample, u64 type,
return result;
}
-void __weak arch_perf_synthesize_sample_weight(const struct perf_sample *data,
+static void perf_synthesize_sample_weight(const struct perf_sample *data,
__u64 *array, u64 type __maybe_unused)
{
*array = data->weight;
+
+ if (type & PERF_SAMPLE_WEIGHT_STRUCT) {
+ *array &= 0xffffffff;
+ *array |= ((u64)data->ins_lat << 32);
+ *array |= ((u64)data->weight3 << 48);
+ }
}
static __u64 *copy_read_group_values(__u64 *array, __u64 read_format,
@@ -1736,7 +1742,7 @@ int perf_event__synthesize_sample(union perf_event *event, u64 type, u64 read_fo
}
if (type & PERF_SAMPLE_WEIGHT_TYPE) {
- arch_perf_synthesize_sample_weight(sample, array, type);
+ perf_synthesize_sample_weight(sample, array, type);
array++;
}
--
2.50.0.727.gbf7dc18ff4-goog
Powered by blists - more mailing lists