Skip to contentSkip to conversation

perf-top: Wait for a minimal set of events before reading first snapshot

PATCH In review Frederic Weisbecker 1 participant 2 messages Jun 6, 2009 – Jun 7, 2009 perf events

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 perf-top: Wait for a minimal set of events before reading first snapshot

The first snapshot reading often occur before any events have been read in the mapped perfcounter files.

Just wait until we have at least one event before starting the snapshot, or the delay before the first set of entries to be displayed may be long in case of low refresh rate.

Note: we could also use a semaphore to wait before "print_entries" number of eveents is reached, but again this value is tunable and we can't ensure we will even reach it. Also we could base on a default mimimum set of entries for the first refresh, say 15, but again, the minimal sample is tunable, and we could end up displaying nothing until we have a minimal default set of events, which can take some time in case of high samples filters.

Hence this simple solution which partially covers the default case.

[ Impact: display first events once they are available on perf top ]

Cc Peter Zijlstra <[email protected]>
Cc Mike Galbraith <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>
---
 tools/perf/builtin-top.c |   22 +++++++++++++++++-----
 1 files changed, 17 insertions(+), 5 deletions(-)
tools/perf/builtin-top.c +17 -5
index f2e7312..fdc1d58 100644
@@ -420,7 +420,7 @@ static unsigned int mmap_read_head(struct mmap_data *md)
 
 struct timeval last_read, this_read;
 
-static void mmap_read(struct mmap_data *md)
+static void mmap_read_counter(struct mmap_data *md)
 {
 	unsigned int head = mmap_read_head(md);
 	unsigned int old = md->prev;
@@ -517,6 +517,16 @@ static void mmap_read(struct mmap_data *md)
 static struct pollfd event_array[MAX_NR_CPUS * MAX_COUNTERS];
 static struct mmap_data mmap_array[MAX_NR_CPUS][MAX_COUNTERS];
 
+static void mmap_read(void)
+{
+	int i, counter;
+
+	for (i = 0; i < nr_cpus; i++) {
+		for (counter = 0; counter < nr_counters; counter++)
+			mmap_read_counter(&mmap_array[i][counter]);
+	}
+}
+
 static int __cmd_top(void)
 {
 	struct perf_counter_attr *attr;
@@ -571,6 +581,11 @@ static int __cmd_top(void)
 		}
 	}
 
+	/* Wait for a minimal set of events before starting the snapshot */
+	poll(event_array, nr_poll, 100);
+
+	mmap_read();
+
 	if (pthread_create(&thread, NULL, display_thread, NULL)) {
 		printf("Could not create display thread.\n");
 		exit(-1);
@@ -589,10 +604,7 @@ static int __cmd_top(void)
 	while (1) {
 		int hits = samples;
 
-		for (i = 0; i < nr_cpus; i++) {
-			for (counter = 0; counter < nr_counters; counter++)
-				mmap_read(&mmap_array[i][counter]);
-		}
+		mmap_read();
 
 		if (hits == samples)
 			ret = poll(event_array, nr_poll, 100);
Signature · 1 line
1.6.2.3
#1 lore raw

[tip:perfcounters/core] perf top: Wait for a minimal set of events before reading first snapshot

Commit-ID:  2f01190aa62fe9dd0a98205927b9f09fd191c017
Gitweb:     http://git.kernel.org/tip/2f01190aa62fe9dd0a98205927b9f09fd191c017
Author:     Frederic Weisbecker <[email protected]>
AuthorDate: Sat, 6 Jun 2009 23:10:43 +0200
Committer:  Ingo Molnar <[email protected]>
CommitDate: Sun, 7 Jun 2009 09:32:44 +0200

perf top: Wait for a minimal set of events before reading first snapshot

The first snapshot reading often occur before any events have been read in the mapped perfcounter files.

Just wait until we have at least one event before starting the snapshot, or the delay before the first set of entries to be displayed may be long in case of low refresh rate.

Note: we could also use a semaphore to wait before "print_entries" number of eveents is reached, but again this value is tunable and we can't ensure we will even reach it. Also we could base on a default mimimum set of entries for the first refresh, say 15, but again, the minimal sample is tunable, and we could end up displaying nothing until we have a minimal default set of events, which can take some time in case of high samples filters.

Hence this simple solution which partially covers the default case.

[ Impact: fix display artifacts in perf top ]

Cc Peter Zijlstra <[email protected]>
Cc Mike Galbraith <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Corey Ashford <[email protected]>
Cc Marcelo Tosatti <[email protected]>
Cc Arnaldo Carvalho de Melo <[email protected]>

LKML-Reference: <[email protected]>

---
 tools/perf/builtin-top.c |   22 +++++++++++++++++-----
 1 files changed, 17 insertions(+), 5 deletions(-)
tools/perf/builtin-top.c +17 -5
index f2e7312..fdc1d58 100644
@@ -420,7 +420,7 @@ static unsigned int mmap_read_head(struct mmap_data *md)
 
 struct timeval last_read, this_read;
 
-static void mmap_read(struct mmap_data *md)
+static void mmap_read_counter(struct mmap_data *md)
 {
 	unsigned int head = mmap_read_head(md);
 	unsigned int old = md->prev;
@@ -517,6 +517,16 @@ static void mmap_read(struct mmap_data *md)
 static struct pollfd event_array[MAX_NR_CPUS * MAX_COUNTERS];
 static struct mmap_data mmap_array[MAX_NR_CPUS][MAX_COUNTERS];
 
+static void mmap_read(void)
+{
+	int i, counter;
+
+	for (i = 0; i < nr_cpus; i++) {
+		for (counter = 0; counter < nr_counters; counter++)
+			mmap_read_counter(&mmap_array[i][counter]);
+	}
+}
+
 static int __cmd_top(void)
 {
 	struct perf_counter_attr *attr;
@@ -571,6 +581,11 @@ static int __cmd_top(void)
 		}
 	}
 
+	/* Wait for a minimal set of events before starting the snapshot */
+	poll(event_array, nr_poll, 100);
+
+	mmap_read();
+
 	if (pthread_create(&thread, NULL, display_thread, NULL)) {
 		printf("Could not create display thread.\n");
 		exit(-1);
@@ -589,10 +604,7 @@ static int __cmd_top(void)
 	while (1) {
 		int hits = samples;
 
-		for (i = 0; i < nr_cpus; i++) {
-			for (counter = 0; counter < nr_counters; counter++)
-				mmap_read(&mmap_array[i][counter]);
-		}
+		mmap_read();
 
 		if (hits == samples)
 			ret = poll(event_array, nr_poll, 100);