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@google.com>
Link: https://lore.kernel.org/r/20250724163302.596743-21-irogers@google.com
Signed-off-by: Namhyung Kim <namhyung@kernel.org>

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"))

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;

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"))

tools/perf/builtin-script.c

@@ -2252,7 +2252,7 @@ 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;

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);

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);
 

tools/perf/util/evsel.c

@@ -2880,11 +2880,18 @@ 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)
@@ -3270,7 +3277,7 @@ 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++;
 	}
 

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);
 

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;

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;
 }

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;

tools/perf/util/session.c

@@ -1099,7 +1099,7 @@ 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");
 	}

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 = {

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++;
 	}