[PATCH v1 0/4] perf hwmon related improvements

[PATCH v1 0/4] perf hwmon related improvements

[Ian Rogers]

Try to do less scanning of files/directories with or without hwmon
PMUs. Don't merge all events with the same name, only merge those on
the same PMU ignoring suffix. Refactor uniquification so the evsels
with the same name as other evsels in the evlist are uniquified.

Ian Rogers (4):
  perf evsel: Reduce scanning core PMUs in is_hybrid
  perf pmus: Restructure pmu_read_sysfs to scan fewer PMUs
  perf stat: Don't merge counters purely on name
  perf stat: Changes to event name uniquification

 tools/perf/util/evsel.c        |   4 +-
 tools/perf/util/evsel.h        |   1 +
 tools/perf/util/pmu.h          |   2 +
 tools/perf/util/pmus.c         | 144 +++++++++++++++++++++------------
 tools/perf/util/stat-display.c | 111 +++++++++++++++++--------
 tools/perf/util/stat.c         |  13 ++-
 6 files changed, 187 insertions(+), 88 deletions(-)

-- 
2.48.0.rc2.279.g1de40edade-goog

[PATCH v1 1/4] perf evsel: Reduce scanning core PMUs in is_hybrid

[Ian Rogers]

evsel__is_hybrid returns true if there are multiple core PMUs and the
evsel is for a core PMU. Determining the number of core PMUs can
require loading/scanning PMUs. There's no point doing the scanning if
evsel for the is_hybrid test isn't core so reorder the tests to reduce
PMU scanning.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/evsel.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c
index bc144388f892..ae20691bbfae 100644
--- a/tools/perf/util/evsel.c
+++ b/tools/perf/util/evsel.c
@@ -3856,10 +3856,10 @@ void evsel__zero_per_pkg(struct evsel *evsel)
  */
 bool evsel__is_hybrid(const struct evsel *evsel)
 {
-	if (perf_pmus__num_core_pmus() == 1)
+	if (!evsel->core.is_pmu_core)
 		return false;
 
-	return evsel->core.is_pmu_core;
+	return perf_pmus__num_core_pmus() > 1;
 }
 
 struct evsel *evsel__leader(const struct evsel *evsel)
-- 
2.48.0.rc2.279.g1de40edade-goog

[PATCH v1 2/4] perf pmus: Restructure pmu_read_sysfs to scan fewer PMUs

[Ian Rogers]

Rather than scanning core or all PMUs, allow pmu_read_sysfs to read
some combination of core, other, hwmon and tool PMUs. The PMUs that
should be read and are already read are held as bitmaps. It is known
that a "hwmon_" prefix is necessary for a hwmon PMU's name, similarly
with "tool", so only scan those PMUs in situations the PMU name or the
PMU's type number make sense to.

The number of openat system calls reduces from 276 to 98 for a hwmon
event. The number of openats for regular perf events isn't changed.

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/pmu.h  |   2 +
 tools/perf/util/pmus.c | 144 ++++++++++++++++++++++++++---------------
 2 files changed, 95 insertions(+), 51 deletions(-)

diff --git a/tools/perf/util/pmu.h b/tools/perf/util/pmu.h
index dbed6c243a5e..edd36c20aedc 100644
--- a/tools/perf/util/pmu.h
+++ b/tools/perf/util/pmu.h
@@ -37,6 +37,8 @@ struct perf_pmu_caps {
 };
 
 enum {
+	PERF_PMU_TYPE_PE_START    = 0,
+	PERF_PMU_TYPE_PE_END      = 0xFFFEFFFF,
 	PERF_PMU_TYPE_HWMON_START = 0xFFFF0000,
 	PERF_PMU_TYPE_HWMON_END   = 0xFFFFFFFD,
 	PERF_PMU_TYPE_TOOL = 0xFFFFFFFE,
diff --git a/tools/perf/util/pmus.c b/tools/perf/util/pmus.c
index b493da0d22ef..3e3ffafcad71 100644
--- a/tools/perf/util/pmus.c
+++ b/tools/perf/util/pmus.c
@@ -37,10 +37,23 @@
  */
 static LIST_HEAD(core_pmus);
 static LIST_HEAD(other_pmus);
-static bool read_sysfs_core_pmus;
-static bool read_sysfs_all_pmus;
+enum perf_tool_pmu_type {
+	PERF_TOOL_PMU_TYPE_PE_CORE,
+#define PERF_TOOL_PMU_TYPE_PE_CORE_MASK (1 << PERF_TOOL_PMU_TYPE_PE_CORE)
+	PERF_TOOL_PMU_TYPE_PE_OTHER,
+#define PERF_TOOL_PMU_TYPE_PE_OTHER_MASK (1 << PERF_TOOL_PMU_TYPE_PE_OTHER)
+	PERF_TOOL_PMU_TYPE_TOOL,
+#define PERF_TOOL_PMU_TYPE_TOOL_MASK (1 << PERF_TOOL_PMU_TYPE_TOOL)
+	PERF_TOOL_PMU_TYPE_HWMON,
+#define PERF_TOOL_PMU_TYPE_HWMON_MASK (1 << PERF_TOOL_PMU_TYPE_HWMON)
+#define PERF_TOOL_PMU_TYPE_ALL_MASK (PERF_TOOL_PMU_TYPE_PE_CORE_MASK |	\
+					PERF_TOOL_PMU_TYPE_PE_OTHER_MASK | \
+					PERF_TOOL_PMU_TYPE_TOOL_MASK |	\
+					PERF_TOOL_PMU_TYPE_HWMON_MASK)
+};
+static unsigned int read_pmu_types;
 
-static void pmu_read_sysfs(bool core_only);
+static void pmu_read_sysfs(unsigned int to_read_pmus);
 
 size_t pmu_name_len_no_suffix(const char *str)
 {
@@ -102,8 +115,7 @@ void perf_pmus__destroy(void)
 
 		perf_pmu__delete(pmu);
 	}
-	read_sysfs_core_pmus = false;
-	read_sysfs_all_pmus = false;
+	read_pmu_types = 0;
 }
 
 static struct perf_pmu *pmu_find(const char *name)
@@ -129,6 +141,7 @@ struct perf_pmu *perf_pmus__find(const char *name)
 	struct perf_pmu *pmu;
 	int dirfd;
 	bool core_pmu;
+	unsigned int to_read_pmus = 0;
 
 	/*
 	 * Once PMU is loaded it stays in the list,
@@ -139,11 +152,11 @@ struct perf_pmu *perf_pmus__find(const char *name)
 	if (pmu)
 		return pmu;
 
-	if (read_sysfs_all_pmus)
+	if (read_pmu_types == PERF_TOOL_PMU_TYPE_ALL_MASK)
 		return NULL;
 
 	core_pmu = is_pmu_core(name);
-	if (core_pmu && read_sysfs_core_pmus)
+	if (core_pmu && (read_pmu_types & PERF_TOOL_PMU_TYPE_PE_CORE_MASK))
 		return NULL;
 
 	dirfd = perf_pmu__event_source_devices_fd();
@@ -151,15 +164,27 @@ struct perf_pmu *perf_pmus__find(const char *name)
 			       /*eager_load=*/false);
 	close(dirfd);
 
-	if (!pmu) {
-		/*
-		 * Looking up an inidividual PMU failed. This may mean name is
-		 * an alias, so read the PMUs from sysfs and try to find again.
-		 */
-		pmu_read_sysfs(core_pmu);
+	if (pmu)
+		return pmu;
+
+	/* Looking up an individual perf event PMU failed, check if a tool PMU should be read. */
+	if (!strncmp(name, "hwmon_", 6))
+		to_read_pmus |= PERF_TOOL_PMU_TYPE_HWMON_MASK;
+	else if (!strcmp(name, "tool"))
+		to_read_pmus |= PERF_TOOL_PMU_TYPE_TOOL_MASK;
+
+	if (to_read_pmus) {
+		pmu_read_sysfs(to_read_pmus);
 		pmu = pmu_find(name);
+		if (pmu)
+			return pmu;
 	}
-	return pmu;
+	/* Read all necessary PMUs from sysfs and see if the PMU is found. */
+	to_read_pmus = PERF_TOOL_PMU_TYPE_PE_CORE_MASK;
+	if (!core_pmu)
+		to_read_pmus |= PERF_TOOL_PMU_TYPE_PE_OTHER_MASK;
+	pmu_read_sysfs(to_read_pmus);
+	return pmu_find(name);
 }
 
 static struct perf_pmu *perf_pmu__find2(int dirfd, const char *name)
@@ -176,11 +201,11 @@ static struct perf_pmu *perf_pmu__find2(int dirfd, const char *name)
 	if (pmu)
 		return pmu;
 
-	if (read_sysfs_all_pmus)
+	if (read_pmu_types == PERF_TOOL_PMU_TYPE_ALL_MASK)
 		return NULL;
 
 	core_pmu = is_pmu_core(name);
-	if (core_pmu && read_sysfs_core_pmus)
+	if (core_pmu && (read_pmu_types & PERF_TOOL_PMU_TYPE_PE_CORE_MASK))
 		return NULL;
 
 	return perf_pmu__lookup(core_pmu ? &core_pmus : &other_pmus, dirfd, name,
@@ -197,52 +222,60 @@ static int pmus_cmp(void *priv __maybe_unused,
 }
 
 /* Add all pmus in sysfs to pmu list: */
-static void pmu_read_sysfs(bool core_only)
+static void pmu_read_sysfs(unsigned int to_read_types)
 {
-	int fd;
-	DIR *dir;
-	struct dirent *dent;
 	struct perf_pmu *tool_pmu;
 
-	if (read_sysfs_all_pmus || (core_only && read_sysfs_core_pmus))
+	if ((read_pmu_types & to_read_types) == to_read_types) {
+		/* All requested PMU types have been read. */
 		return;
+	}
 
-	fd = perf_pmu__event_source_devices_fd();
-	if (fd < 0)
-		return;
+	if (to_read_types & (PERF_TOOL_PMU_TYPE_PE_CORE_MASK | PERF_TOOL_PMU_TYPE_PE_OTHER_MASK)) {
+		int fd = perf_pmu__event_source_devices_fd();
+		DIR *dir;
+		struct dirent *dent;
+		bool core_only = (to_read_types & PERF_TOOL_PMU_TYPE_PE_OTHER_MASK) == 0;
 
-	dir = fdopendir(fd);
-	if (!dir) {
-		close(fd);
-		return;
-	}
+		if (fd < 0)
+			goto skip_pe_pmus;
 
-	while ((dent = readdir(dir))) {
-		if (!strcmp(dent->d_name, ".") || !strcmp(dent->d_name, ".."))
-			continue;
-		if (core_only && !is_pmu_core(dent->d_name))
-			continue;
-		/* add to static LIST_HEAD(core_pmus) or LIST_HEAD(other_pmus): */
-		perf_pmu__find2(fd, dent->d_name);
-	}
+		dir = fdopendir(fd);
+		if (!dir) {
+			close(fd);
+			goto skip_pe_pmus;
+		}
 
-	closedir(dir);
-	if (list_empty(&core_pmus)) {
+		while ((dent = readdir(dir))) {
+			if (!strcmp(dent->d_name, ".") || !strcmp(dent->d_name, ".."))
+				continue;
+			if (core_only && !is_pmu_core(dent->d_name))
+				continue;
+			/* add to static LIST_HEAD(core_pmus) or LIST_HEAD(other_pmus): */
+			perf_pmu__find2(fd, dent->d_name);
+		}
+
+		closedir(dir);
+	}
+skip_pe_pmus:
+	if ((to_read_types & PERF_TOOL_PMU_TYPE_PE_CORE_MASK) && list_empty(&core_pmus)) {
 		if (!perf_pmu__create_placeholder_core_pmu(&core_pmus))
 			pr_err("Failure to set up any core PMUs\n");
 	}
 	list_sort(NULL, &core_pmus, pmus_cmp);
-	if (!core_only) {
+
+	if ((to_read_types & PERF_TOOL_PMU_TYPE_TOOL_MASK) != 0 &&
+	    (read_pmu_types & PERF_TOOL_PMU_TYPE_TOOL_MASK) == 0) {
 		tool_pmu = perf_pmus__tool_pmu();
 		list_add_tail(&tool_pmu->list, &other_pmus);
-		perf_pmus__read_hwmon_pmus(&other_pmus);
 	}
+	if ((to_read_types & PERF_TOOL_PMU_TYPE_HWMON_MASK) != 0 &&
+	    (read_pmu_types & PERF_TOOL_PMU_TYPE_HWMON_MASK) == 0)
+		perf_pmus__read_hwmon_pmus(&other_pmus);
+
 	list_sort(NULL, &other_pmus, pmus_cmp);
-	if (!list_empty(&core_pmus)) {
-		read_sysfs_core_pmus = true;
-		if (!core_only)
-			read_sysfs_all_pmus = true;
-	}
+
+	read_pmu_types |= to_read_types;
 }
 
 static struct perf_pmu *__perf_pmus__find_by_type(unsigned int type)
@@ -263,12 +296,21 @@ static struct perf_pmu *__perf_pmus__find_by_type(unsigned int type)
 
 struct perf_pmu *perf_pmus__find_by_type(unsigned int type)
 {
+	unsigned int to_read_pmus;
 	struct perf_pmu *pmu = __perf_pmus__find_by_type(type);
 
-	if (pmu || read_sysfs_all_pmus)
+	if (pmu || (read_pmu_types == PERF_TOOL_PMU_TYPE_ALL_MASK))
 		return pmu;
 
-	pmu_read_sysfs(/*core_only=*/false);
+	if (type >= PERF_PMU_TYPE_PE_START && type <= PERF_PMU_TYPE_PE_END) {
+		to_read_pmus = PERF_TOOL_PMU_TYPE_PE_CORE_MASK |
+			PERF_TOOL_PMU_TYPE_PE_OTHER_MASK;
+	} else if (type >= PERF_PMU_TYPE_HWMON_START && type <= PERF_PMU_TYPE_HWMON_END) {
+		to_read_pmus = PERF_TOOL_PMU_TYPE_HWMON_MASK;
+	} else {
+		to_read_pmus = PERF_TOOL_PMU_TYPE_TOOL_MASK;
+	}
+	pmu_read_sysfs(to_read_pmus);
 	pmu = __perf_pmus__find_by_type(type);
 	return pmu;
 }
@@ -282,7 +324,7 @@ struct perf_pmu *perf_pmus__scan(struct perf_pmu *pmu)
 	bool use_core_pmus = !pmu || pmu->is_core;
 
 	if (!pmu) {
-		pmu_read_sysfs(/*core_only=*/false);
+		pmu_read_sysfs(PERF_TOOL_PMU_TYPE_ALL_MASK);
 		pmu = list_prepare_entry(pmu, &core_pmus, list);
 	}
 	if (use_core_pmus) {
@@ -300,7 +342,7 @@ struct perf_pmu *perf_pmus__scan(struct perf_pmu *pmu)
 struct perf_pmu *perf_pmus__scan_core(struct perf_pmu *pmu)
 {
 	if (!pmu) {
-		pmu_read_sysfs(/*core_only=*/true);
+		pmu_read_sysfs(PERF_TOOL_PMU_TYPE_PE_CORE_MASK);
 		return list_first_entry_or_null(&core_pmus, typeof(*pmu), list);
 	}
 	list_for_each_entry_continue(pmu, &core_pmus, list)
@@ -316,7 +358,7 @@ static struct perf_pmu *perf_pmus__scan_skip_duplicates(struct perf_pmu *pmu)
 	const char *last_pmu_name = (pmu && pmu->name) ? pmu->name : "";
 
 	if (!pmu) {
-		pmu_read_sysfs(/*core_only=*/false);
+		pmu_read_sysfs(PERF_TOOL_PMU_TYPE_ALL_MASK);
 		pmu = list_prepare_entry(pmu, &core_pmus, list);
 	} else
 		last_pmu_name_len = pmu_name_len_no_suffix(pmu->name ?: "");
-- 
2.48.0.rc2.279.g1de40edade-goog

[PATCH v1 3/4] perf stat: Don't merge counters purely on name

[Ian Rogers]

Counter merging was added in commit 942c5593393d ("perf stat: Add
perf_stat_merge_counters()"), however, it merges events with the same
name on different PMUs regardless of whether the different PMUs are
actually of the same type (ie they differ only in the suffix on the
PMU). For hwmon events there may be a temp1 event on every PMU, but
the PMU names are all unique and don't differ just by a suffix. The
merging is over eager and will merge all the hwmon counters together
meaning an aggregated and very large temp1 value is shown. The same
would be true for say cache events and memory controller events where
the PMUs differ but the event names are the same.

Fix the problem by correctly saying two PMUs alias when they differ
only by suffix.

Note, there is an overlap with evsel's merged_stat with aggregation
and the evsel's metric_leader where aggregation happens for metrics.

Fixes: 942c5593393d ("perf stat: Add perf_stat_merge_counters()")
Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/stat.c | 13 +++++++++++--
 1 file changed, 11 insertions(+), 2 deletions(-)

diff --git a/tools/perf/util/stat.c b/tools/perf/util/stat.c
index 7c2ccdcc3fdb..cfdb6d55df9b 100644
--- a/tools/perf/util/stat.c
+++ b/tools/perf/util/stat.c
@@ -535,7 +535,10 @@ static int evsel__merge_aggr_counters(struct evsel *evsel, struct evsel *alias)
 
 	return 0;
 }
-/* events should have the same name, scale, unit, cgroup but on different PMUs */
+/*
+ * Events should have the same name, scale, unit, cgroup but on different core
+ * PMUs or on different but matching uncore PMUs.
+ */
 static bool evsel__is_alias(struct evsel *evsel_a, struct evsel *evsel_b)
 {
 	if (strcmp(evsel__name(evsel_a), evsel__name(evsel_b)))
@@ -553,7 +556,13 @@ static bool evsel__is_alias(struct evsel *evsel_a, struct evsel *evsel_b)
 	if (evsel__is_clock(evsel_a) != evsel__is_clock(evsel_b))
 		return false;
 
-	return evsel_a->pmu != evsel_b->pmu;
+	if (evsel_a->pmu == evsel_b->pmu || evsel_a->pmu == NULL || evsel_b->pmu == NULL)
+		return false;
+
+	if (evsel_a->pmu->is_core)
+		return evsel_b->pmu->is_core;
+
+	return pmu__name_match(evsel_a->pmu, evsel_b->pmu->name);
 }
 
 static void evsel__merge_aliases(struct evsel *evsel)
-- 
2.48.0.rc2.279.g1de40edade-goog

[PATCH v1 4/4] perf stat: Changes to event name uniquification

[Ian Rogers]

The existing logic would disable uniquification on an evlist or enable
it per evsel, this is unfortunate as uniquification is most needed
when events have the same name and so the whole evlist must be
considered.  Change the initial disable uniquify on an evlist
processing to also set a needs_uniquify flag, for cases like the
matching event names. This must be done as an initial pass as
uniquification of an event name will change the behavior of the
check. Keep the per counter uniquification but now only uniquify event
names when the needs_uniquify flag is set.

Before this change a hwmon like temp1 wouldn't be uniquified and
afterwards it will (ie the PMU is added to the temp1 event's name).

Signed-off-by: Ian Rogers <irogers@google.com>
---
 tools/perf/util/evsel.h        |   1 +
 tools/perf/util/stat-display.c | 111 +++++++++++++++++++++++----------
 2 files changed, 79 insertions(+), 33 deletions(-)

diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h
index 5e789fa80590..d91621b6c8c2 100644
--- a/tools/perf/util/evsel.h
+++ b/tools/perf/util/evsel.h
@@ -119,6 +119,7 @@ struct evsel {
 	bool			errored;
 	bool			needs_auxtrace_mmap;
 	bool			default_metricgroup; /* A member of the Default metricgroup */
+	bool			needs_uniquify;
 	struct hashmap		*per_pkg_mask;
 	int			err;
 	int			script_output_type;
diff --git a/tools/perf/util/stat-display.c b/tools/perf/util/stat-display.c
index ba79f73e1cf5..e65c7e9f15d1 100644
--- a/tools/perf/util/stat-display.c
+++ b/tools/perf/util/stat-display.c
@@ -929,12 +929,16 @@ static void printout(struct perf_stat_config *config, struct outstate *os,
 	}
 }
 
-static void uniquify_event_name(struct evsel *counter)
+static void evsel__uniquify_counter(struct evsel *counter)
 {
 	const char *name, *pmu_name;
 	char *new_name, *config;
 	int ret;
 
+	/* No uniquification necessary. */
+	if (!counter->needs_uniquify)
+		return;
+
 	/* The evsel was already uniquified. */
 	if (counter->uniquified_name)
 		return;
@@ -942,19 +946,6 @@ static void uniquify_event_name(struct evsel *counter)
 	/* Avoid checking to uniquify twice. */
 	counter->uniquified_name = true;
 
-	/* The evsel has a "name=" config term or is from libpfm. */
-	if (counter->use_config_name || counter->is_libpfm_event)
-		return;
-
-	/* Legacy no PMU event, don't uniquify. */
-	if  (!counter->pmu ||
-	     (counter->pmu->type < PERF_TYPE_MAX && counter->pmu->type != PERF_TYPE_RAW))
-		return;
-
-	/* A sysfs or json event replacing a legacy event, don't uniquify. */
-	if (counter->pmu->is_core && counter->alternate_hw_config != PERF_COUNT_HW_MAX)
-		return;
-
 	name = evsel__name(counter);
 	pmu_name = counter->pmu->name;
 	/* Already prefixed by the PMU name. */
@@ -993,17 +984,6 @@ static void uniquify_event_name(struct evsel *counter)
 	}
 }
 
-static bool hybrid_uniquify(struct evsel *evsel, struct perf_stat_config *config)
-{
-	return evsel__is_hybrid(evsel) && !config->hybrid_merge;
-}
-
-static void uniquify_counter(struct perf_stat_config *config, struct evsel *counter)
-{
-	if (config->aggr_mode == AGGR_NONE || hybrid_uniquify(counter, config))
-		uniquify_event_name(counter);
-}
-
 /**
  * should_skip_zero_count() - Check if the event should print 0 values.
  * @config: The perf stat configuration (including aggregation mode).
@@ -1089,7 +1069,7 @@ static void print_counter_aggrdata(struct perf_stat_config *config,
 	if (counter->merged_stat)
 		return;
 
-	uniquify_counter(config, counter);
+	evsel__uniquify_counter(counter);
 
 	val = aggr->counts.val;
 	ena = aggr->counts.ena;
@@ -1670,7 +1650,8 @@ static void print_cgroup_counter(struct perf_stat_config *config, struct evlist
 		print_metric_end(config, os);
 }
 
-static void disable_uniquify(struct evlist *evlist)
+/* Should uniquify be disabled for the evlist? */
+static bool evlist__disable_uniquify(const struct evlist *evlist)
 {
 	struct evsel *counter;
 	struct perf_pmu *last_pmu = NULL;
@@ -1679,20 +1660,84 @@ static void disable_uniquify(struct evlist *evlist)
 	evlist__for_each_entry(evlist, counter) {
 		/* If PMUs vary then uniquify can be useful. */
 		if (!first && counter->pmu != last_pmu)
-			return;
+			return false;
 		first = false;
 		if (counter->pmu) {
 			/* Allow uniquify for uncore PMUs. */
 			if (!counter->pmu->is_core)
-				return;
+				return false;
 			/* Keep hybrid event names uniquified for clarity. */
 			if (perf_pmus__num_core_pmus() > 1)
-				return;
+				return false;
+		}
+	}
+	return true;
+}
+
+static void evsel__set_needs_uniquify(struct evsel *counter, const struct perf_stat_config *config)
+{
+	struct evsel *evsel;
+
+	if (counter->merged_stat) {
+		/* Counter won't be shown. */
+		return;
+	}
+
+	if (counter->use_config_name || counter->is_libpfm_event) {
+		/* Original name will be used. */
+		return;
+	}
+
+	if  (counter->core.attr.type < PERF_TYPE_MAX && counter->core.attr.type != PERF_TYPE_RAW) {
+		/* Legacy event, don't uniquify. */
+		return;
+	}
+
+	if (counter->pmu && counter->pmu->is_core &&
+	    counter->alternate_hw_config != PERF_COUNT_HW_MAX) {
+		/* A sysfs or json event replacing a legacy event, don't uniquify. */
+		return;
+	}
+
+	if (config->aggr_mode == AGGR_NONE) {
+		/* Always unique with no aggregation. */
+		counter->needs_uniquify = true;
+		return;
+	}
+
+	if (!config->hybrid_merge && evsel__is_hybrid(counter)) {
+		/* Unique hybrid counters necessary. */
+		counter->needs_uniquify = true;
+		return;
+	}
+
+	/*
+	 * Do other non-merged events in the evlist have the same name? If so
+	 * uniquify is necessary.
+	 */
+	evlist__for_each_entry(counter->evlist, evsel) {
+		if (evsel == counter || evsel->merged_stat)
+			continue;
+
+		if (evsel__name_is(counter, evsel__name(evsel))) {
+			counter->needs_uniquify = true;
+			return;
 		}
 	}
-	evlist__for_each_entry_continue(evlist, counter) {
-		counter->uniquified_name = true;
+}
+
+static void evlist__set_needs_uniquify(struct evlist *evlist, const struct perf_stat_config *config)
+{
+	struct evsel *counter;
+
+	if (evlist__disable_uniquify(evlist)) {
+		evlist__for_each_entry(evlist, counter)
+			counter->uniquified_name = true;
+		return;
 	}
+
+	evlist__for_each_entry(evlist, counter)
+		evsel__set_needs_uniquify(counter, config);
 }
 
 void evlist__print_counters(struct evlist *evlist, struct perf_stat_config *config,
@@ -1706,7 +1751,7 @@ void evlist__print_counters(struct evlist *evlist, struct perf_stat_config *conf
 		.first = true,
 	};
 
-	disable_uniquify(evlist);
+	evlist__set_needs_uniquify(evlist, config);
 
 	if (config->iostat_run)
 		evlist->selected = evlist__first(evlist);
-- 
2.48.0.rc2.279.g1de40edade-goog