| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | /*
 | 
					
						
							|  |  |  |  * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> | 
					
						
							|  |  |  |  * | 
					
						
							|  |  |  |  * Parts came from builtin-{top,stat,record}.c, see those files for further | 
					
						
							|  |  |  |  * copyright notes. | 
					
						
							|  |  |  |  * | 
					
						
							|  |  |  |  * Released under the GPL v2. (and only v2, not any later version) | 
					
						
							|  |  |  |  */ | 
					
						
							| 
									
										
										
										
											2011-11-05 08:41:51 -02:00
										 |  |  | #include "util.h"
 | 
					
						
							| 
									
										
										
										
											2013-02-20 16:32:30 +01:00
										 |  |  | #include <lk/debugfs.h>
 | 
					
						
							| 
									
										
										
										
											2011-01-11 22:30:02 -02:00
										 |  |  | #include <poll.h>
 | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | #include "cpumap.h"
 | 
					
						
							|  |  |  | #include "thread_map.h"
 | 
					
						
							| 
									
										
										
										
											2012-04-26 14:15:22 +09:00
										 |  |  | #include "target.h"
 | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | #include "evlist.h"
 | 
					
						
							|  |  |  | #include "evsel.h"
 | 
					
						
							| 
									
										
										
										
											2011-11-09 08:47:15 -02:00
										 |  |  | #include <unistd.h>
 | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-11-04 09:10:59 -02:00
										 |  |  | #include "parse-events.h"
 | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | #include <sys/mman.h>
 | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-12 22:39:13 -02:00
										 |  |  | #include <linux/bitops.h>
 | 
					
						
							|  |  |  | #include <linux/hash.h>
 | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y))
 | 
					
						
							| 
									
										
										
										
											2011-03-10 11:15:54 -03:00
										 |  |  | #define SID(e, x, y) xyarray__entry(e->sample_id, x, y)
 | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  | void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus, | 
					
						
							|  |  |  | 		       struct thread_map *threads) | 
					
						
							| 
									
										
										
										
											2011-01-18 21:41:45 -02:00
										 |  |  | { | 
					
						
							|  |  |  | 	int i; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	for (i = 0; i < PERF_EVLIST__HLIST_SIZE; ++i) | 
					
						
							|  |  |  | 		INIT_HLIST_HEAD(&evlist->heads[i]); | 
					
						
							|  |  |  | 	INIT_LIST_HEAD(&evlist->entries); | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  | 	perf_evlist__set_maps(evlist, cpus, threads); | 
					
						
							| 
									
										
										
										
											2011-11-09 08:47:15 -02:00
										 |  |  | 	evlist->workload.pid = -1; | 
					
						
							| 
									
										
										
										
											2011-01-18 21:41:45 -02:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:12 +09:00
										 |  |  | struct perf_evlist *perf_evlist__new(void) | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | { | 
					
						
							|  |  |  | 	struct perf_evlist *evlist = zalloc(sizeof(*evlist)); | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-18 21:41:45 -02:00
										 |  |  | 	if (evlist != NULL) | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:12 +09:00
										 |  |  | 		perf_evlist__init(evlist, NULL, NULL); | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	return evlist; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-12-10 16:41:31 -03:00
										 |  |  | void perf_evlist__config(struct perf_evlist *evlist, | 
					
						
							|  |  |  | 			struct perf_record_opts *opts) | 
					
						
							| 
									
										
										
										
											2011-11-08 14:41:57 -02:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2012-11-12 18:34:00 +01:00
										 |  |  | 	struct perf_evsel *evsel; | 
					
						
							| 
									
										
										
										
											2012-12-10 16:41:31 -03:00
										 |  |  | 	/*
 | 
					
						
							|  |  |  | 	 * Set the evsel leader links before we configure attributes, | 
					
						
							|  |  |  | 	 * since some might depend on this info. | 
					
						
							|  |  |  | 	 */ | 
					
						
							|  |  |  | 	if (opts->group) | 
					
						
							|  |  |  | 		perf_evlist__set_leader(evlist); | 
					
						
							| 
									
										
										
										
											2011-11-08 14:41:57 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	if (evlist->cpus->map[0] < 0) | 
					
						
							|  |  |  | 		opts->no_inherit = true; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry(evsel, &evlist->entries, node) { | 
					
						
							| 
									
										
										
										
											2012-11-12 18:34:00 +01:00
										 |  |  | 		perf_evsel__config(evsel, opts); | 
					
						
							| 
									
										
										
										
											2011-11-08 14:41:57 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 		if (evlist->nr_entries > 1) | 
					
						
							| 
									
										
										
										
											2012-12-10 15:29:12 -03:00
										 |  |  | 			perf_evsel__set_sample_id(evsel); | 
					
						
							| 
									
										
										
										
											2011-11-08 14:41:57 -02:00
										 |  |  | 	} | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | static void perf_evlist__purge(struct perf_evlist *evlist) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *pos, *n; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry_safe(pos, n, &evlist->entries, node) { | 
					
						
							|  |  |  | 		list_del_init(&pos->node); | 
					
						
							|  |  |  | 		perf_evsel__delete(pos); | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	evlist->nr_entries = 0; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-18 21:41:45 -02:00
										 |  |  | void perf_evlist__exit(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2011-01-12 22:39:13 -02:00
										 |  |  | 	free(evlist->mmap); | 
					
						
							| 
									
										
										
										
											2011-01-11 22:30:02 -02:00
										 |  |  | 	free(evlist->pollfd); | 
					
						
							| 
									
										
										
										
											2011-01-18 21:41:45 -02:00
										 |  |  | 	evlist->mmap = NULL; | 
					
						
							|  |  |  | 	evlist->pollfd = NULL; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | void perf_evlist__delete(struct perf_evlist *evlist) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	perf_evlist__purge(evlist); | 
					
						
							|  |  |  | 	perf_evlist__exit(evlist); | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | 	free(evlist); | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	list_add_tail(&entry->node, &evlist->entries); | 
					
						
							|  |  |  | 	++evlist->nr_entries; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-01-27 15:34:20 +01:00
										 |  |  | void perf_evlist__splice_list_tail(struct perf_evlist *evlist, | 
					
						
							|  |  |  | 				   struct list_head *list, | 
					
						
							|  |  |  | 				   int nr_entries) | 
					
						
							| 
									
										
										
										
											2011-11-04 09:10:59 -02:00
										 |  |  | { | 
					
						
							|  |  |  | 	list_splice_tail(list, &evlist->entries); | 
					
						
							|  |  |  | 	evlist->nr_entries += nr_entries; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-08-14 16:35:48 -03:00
										 |  |  | void __perf_evlist__set_leader(struct list_head *list) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *evsel, *leader; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	leader = list_entry(list->next, struct perf_evsel, node); | 
					
						
							| 
									
										
										
										
											2013-01-22 18:09:29 +09:00
										 |  |  | 	evsel = list_entry(list->prev, struct perf_evsel, node); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	leader->nr_members = evsel->idx - leader->idx + 1; | 
					
						
							| 
									
										
										
										
											2012-08-14 16:35:48 -03:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry(evsel, list, node) { | 
					
						
							| 
									
										
										
										
											2013-01-31 13:54:37 +01:00
										 |  |  | 		evsel->leader = leader; | 
					
						
							| 
									
										
										
										
											2012-08-14 16:35:48 -03:00
										 |  |  | 	} | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | void perf_evlist__set_leader(struct perf_evlist *evlist) | 
					
						
							| 
									
										
											  
											
												perf tools: Enable grouping logic for parsed events
This patch adds a functionality that allows to create event groups
based on the way they are specified on the command line. Adding
functionality to the '{}' group syntax introduced in earlier patch.
The current '--group/-g' option behaviour remains intact. If you
specify it for record/stat/top command, all the specified events
become members of a single group with the first event as a group
leader.
With the new '{}' group syntax you can create group like:
  # perf record -e '{cycles,faults}' ls
resulting in single event group containing 'cycles' and 'faults'
events, with cycles event as group leader.
All groups are created with regards to threads and cpus. Thus
recording an event group within a 2 threads on server with
4 CPUs will create 8 separate groups.
Examples (first event in brackets is group leader):
  # 1 group (cpu-clock,task-clock)
  perf record --group -e cpu-clock,task-clock ls
  perf record -e '{cpu-clock,task-clock}' ls
  # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults)
  perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls
  # 1 group (cpu-clock,task-clock,minor-faults,major-faults)
  perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls
  perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls
  # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults)
  perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \
   -e instructions ls
  # 1 group
  # (cpu-clock,task-clock,minor-faults,major-faults,instructions)
  perf record --group -e cpu-clock,task-clock \
   -e minor-faults,major-faults -e instructions ls perf record -e
'{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls
It's possible to use standard event modifier for a group, which spans
over all events in the group and updates each event modifier settings,
for example:
  # perf record -r '{faults:k,cache-references}:p'
resulting in ':kp' modifier being used for 'faults' and ':p' modifier
being used for 'cache-references' event.
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ulrich Drepper <drepper@gmail.com>
Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2012-08-08 12:22:36 +02:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2013-01-22 18:09:29 +09:00
										 |  |  | 	if (evlist->nr_entries) { | 
					
						
							|  |  |  | 		evlist->nr_groups = evlist->nr_entries > 1 ? 1 : 0; | 
					
						
							| 
									
										
										
										
											2012-08-14 16:35:48 -03:00
										 |  |  | 		__perf_evlist__set_leader(&evlist->entries); | 
					
						
							| 
									
										
										
										
											2013-01-22 18:09:29 +09:00
										 |  |  | 	} | 
					
						
							| 
									
										
											  
											
												perf tools: Enable grouping logic for parsed events
This patch adds a functionality that allows to create event groups
based on the way they are specified on the command line. Adding
functionality to the '{}' group syntax introduced in earlier patch.
The current '--group/-g' option behaviour remains intact. If you
specify it for record/stat/top command, all the specified events
become members of a single group with the first event as a group
leader.
With the new '{}' group syntax you can create group like:
  # perf record -e '{cycles,faults}' ls
resulting in single event group containing 'cycles' and 'faults'
events, with cycles event as group leader.
All groups are created with regards to threads and cpus. Thus
recording an event group within a 2 threads on server with
4 CPUs will create 8 separate groups.
Examples (first event in brackets is group leader):
  # 1 group (cpu-clock,task-clock)
  perf record --group -e cpu-clock,task-clock ls
  perf record -e '{cpu-clock,task-clock}' ls
  # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults)
  perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls
  # 1 group (cpu-clock,task-clock,minor-faults,major-faults)
  perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls
  perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls
  # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults)
  perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \
   -e instructions ls
  # 1 group
  # (cpu-clock,task-clock,minor-faults,major-faults,instructions)
  perf record --group -e cpu-clock,task-clock \
   -e minor-faults,major-faults -e instructions ls perf record -e
'{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls
It's possible to use standard event modifier for a group, which spans
over all events in the group and updates each event modifier settings,
for example:
  # perf record -r '{faults:k,cache-references}:p'
resulting in ':kp' modifier being used for 'faults' and ':p' modifier
being used for 'cache-references' event.
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ulrich Drepper <drepper@gmail.com>
Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2012-08-08 12:22:36 +02:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | int perf_evlist__add_default(struct perf_evlist *evlist) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	struct perf_event_attr attr = { | 
					
						
							|  |  |  | 		.type = PERF_TYPE_HARDWARE, | 
					
						
							|  |  |  | 		.config = PERF_COUNT_HW_CPU_CYCLES, | 
					
						
							|  |  |  | 	}; | 
					
						
							| 
									
										
										
										
											2012-01-04 17:54:20 +01:00
										 |  |  | 	struct perf_evsel *evsel; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	event_attr_init(&attr); | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-01-04 17:54:20 +01:00
										 |  |  | 	evsel = perf_evsel__new(&attr, 0); | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | 	if (evsel == NULL) | 
					
						
							| 
									
										
										
										
											2011-06-07 18:19:36 +02:00
										 |  |  | 		goto error; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	/* use strdup() because free(evsel) assumes name is allocated */ | 
					
						
							|  |  |  | 	evsel->name = strdup("cycles"); | 
					
						
							|  |  |  | 	if (!evsel->name) | 
					
						
							|  |  |  | 		goto error_free; | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	perf_evlist__add(evlist, evsel); | 
					
						
							|  |  |  | 	return 0; | 
					
						
							| 
									
										
										
										
											2011-06-07 18:19:36 +02:00
										 |  |  | error_free: | 
					
						
							|  |  |  | 	perf_evsel__delete(evsel); | 
					
						
							|  |  |  | error: | 
					
						
							|  |  |  | 	return -ENOMEM; | 
					
						
							| 
									
										
										
										
											2011-01-11 20:56:53 -02:00
										 |  |  | } | 
					
						
							| 
									
										
										
										
											2011-01-11 22:30:02 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-10-03 11:50:55 -03:00
										 |  |  | static int perf_evlist__add_attrs(struct perf_evlist *evlist, | 
					
						
							|  |  |  | 				  struct perf_event_attr *attrs, size_t nr_attrs) | 
					
						
							| 
									
										
										
										
											2011-11-04 09:10:59 -02:00
										 |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *evsel, *n; | 
					
						
							|  |  |  | 	LIST_HEAD(head); | 
					
						
							|  |  |  | 	size_t i; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	for (i = 0; i < nr_attrs; i++) { | 
					
						
							|  |  |  | 		evsel = perf_evsel__new(attrs + i, evlist->nr_entries + i); | 
					
						
							|  |  |  | 		if (evsel == NULL) | 
					
						
							|  |  |  | 			goto out_delete_partial_list; | 
					
						
							|  |  |  | 		list_add_tail(&evsel->node, &head); | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	perf_evlist__splice_list_tail(evlist, &head, nr_attrs); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	return 0; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | out_delete_partial_list: | 
					
						
							|  |  |  | 	list_for_each_entry_safe(evsel, n, &head, node) | 
					
						
							|  |  |  | 		perf_evsel__delete(evsel); | 
					
						
							|  |  |  | 	return -1; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
											  
											
												perf stat: Initialize default events wrt exclude_{guest,host}
When no event is specified the tools use perf_evlist__add_default(), that will
call event_attr_init to initialize the KVM exclusion bits.
When the change was made to the tools so that by default guest samples would be
excluded, the changes were made just to the parsing routines and to
perf_evlist__add_default(), not to perf_evlist__add_attrs, that is used so far
just by perf stat to add multiple events, according to the level of detail
specified.
Recently the tools were changed to reconstruct the event name from all the
details in perf_event_attr, not just from .type and .config, but taking into
account all the feature bits (.exclude_{guest,host,user,kernel,etc},
.precise_ip, etc).
That is when we noticed that the default for perf stat wasn't the one for the
rest of the tools, i.e. the .exclude_guest bit wasn't being set.
I.e. the default, that doesn't call event_attr_init was showing the :HG
modifier:
  $ perf stat usleep 1
   Performance counter stats for 'usleep 1':
            0.942119 task-clock                #    0.454 CPUs utilized
                   1 context-switches          #    0.001 M/sec
                   0 CPU-migrations            #    0.000 K/sec
                 126 page-faults               #    0.134 M/sec
             693,193 cycles:HG                 #    0.736 GHz                     [40.11%]
             407,461 stalled-cycles-frontend:HG #   58.78% frontend cycles idle    [72.29%]
             365,403 stalled-cycles-backend:HG #   52.71% backend  cycles idle
             465,982 instructions:HG           #    0.67  insns per cycle
                                               #    0.87  stalled cycles per insn
              89,760 branches:HG               #   95.275 M/sec
               6,178 branch-misses:HG          #    6.88% of all branches
         0.002077228 seconds time elapsed
While if one explicitely specifies the same events, which will make the parsing code
to be called and thus event_attr_init is called:
  $ perf stat -e task-clock,context-switches,migrations,page-faults,cycles,stalled-cycles-frontend,stalled-cycles-backend,instructions,branches,branch-misses usleep 1
   Performance counter stats for 'usleep 1':
            1.040349 task-clock                #    0.500 CPUs utilized
                   2 context-switches          #    0.002 M/sec
                   0 CPU-migrations            #    0.000 K/sec
                 127 page-faults               #    0.122 M/sec
             587,966 cycles                    #    0.565 GHz                     [13.18%]
             459,167 stalled-cycles-frontend   #   78.09% frontend cycles idle
             390,249 stalled-cycles-backend    #   66.37% backend  cycles idle
             504,006 instructions              #    0.86  insns per cycle
                                               #    0.91  stalled cycles per insn
              96,455 branches                  #   92.714 M/sec
               6,522 branch-misses             #    6.76% of all branches         [96.12%]
         0.002078681 seconds time elapsed
Fix it by introducing a perf_evlist__add_default_attrs method that will call
evlist_attr_init in all the perf_event_attr entries before adding the events.
Reported-by: Ingo Molnar <mingo@kernel.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Namhyung Kim <namhyung@gmail.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-4eysr236r0pgiyum9epwxw7s@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2012-05-30 13:53:54 -03:00
										 |  |  | int __perf_evlist__add_default_attrs(struct perf_evlist *evlist, | 
					
						
							|  |  |  | 				     struct perf_event_attr *attrs, size_t nr_attrs) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	size_t i; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	for (i = 0; i < nr_attrs; i++) | 
					
						
							|  |  |  | 		event_attr_init(attrs + i); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	return perf_evlist__add_attrs(evlist, attrs, nr_attrs); | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-06-27 13:08:42 -03:00
										 |  |  | struct perf_evsel * | 
					
						
							|  |  |  | perf_evlist__find_tracepoint_by_id(struct perf_evlist *evlist, int id) | 
					
						
							| 
									
										
											  
											
												perf tools: Save some loops using perf_evlist__id2evsel
Since we already ask for PERF_SAMPLE_ID and use it to quickly find the
associated evsel, add handler func + data to struct perf_evsel to avoid
using chains of if(strcmp(event_name)) and also to avoid all the linear
list searches via trace_event_find.
To demonstrate the technique convert 'perf sched' to it:
 # perf sched record sleep 5m
And then:
 Performance counter stats for '/tmp/oldperf sched lat':
        646.929438 task-clock                #    0.999 CPUs utilized
                 9 context-switches          #    0.000 M/sec
                 0 CPU-migrations            #    0.000 M/sec
            20,901 page-faults               #    0.032 M/sec
     1,290,144,450 cycles                    #    1.994 GHz
   <not supported> stalled-cycles-frontend
   <not supported> stalled-cycles-backend
     1,606,158,439 instructions              #    1.24  insns per cycle
       339,088,395 branches                  #  524.151 M/sec
         4,550,735 branch-misses             #    1.34% of all branches
       0.647524759 seconds time elapsed
Versus:
 Performance counter stats for 'perf sched lat':
        473.564691 task-clock                #    0.999 CPUs utilized
                 9 context-switches          #    0.000 M/sec
                 0 CPU-migrations            #    0.000 M/sec
            20,903 page-faults               #    0.044 M/sec
       944,367,984 cycles                    #    1.994 GHz
   <not supported> stalled-cycles-frontend
   <not supported> stalled-cycles-backend
     1,442,385,571 instructions              #    1.53  insns per cycle
       308,383,106 branches                  #  651.195 M/sec
         4,481,784 branch-misses             #    1.45% of all branches
       0.474215751 seconds time elapsed
[root@emilia ~]#
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-1kbzpl74lwi6lavpqke2u2p3@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-11-28 17:57:40 -02:00
										 |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *evsel; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry(evsel, &evlist->entries, node) { | 
					
						
							|  |  |  | 		if (evsel->attr.type   == PERF_TYPE_TRACEPOINT && | 
					
						
							|  |  |  | 		    (int)evsel->attr.config == id) | 
					
						
							|  |  |  | 			return evsel; | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	return NULL; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-10-03 11:40:22 -03:00
										 |  |  | int perf_evlist__add_newtp(struct perf_evlist *evlist, | 
					
						
							|  |  |  | 			   const char *sys, const char *name, void *handler) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *evsel; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	evsel = perf_evsel__newtp(sys, name, evlist->nr_entries); | 
					
						
							|  |  |  | 	if (evsel == NULL) | 
					
						
							|  |  |  | 		return -1; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	evsel->handler.func = handler; | 
					
						
							|  |  |  | 	perf_evlist__add(evlist, evsel); | 
					
						
							|  |  |  | 	return 0; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-07-25 11:06:19 -03:00
										 |  |  | void perf_evlist__disable(struct perf_evlist *evlist) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	int cpu, thread; | 
					
						
							|  |  |  | 	struct perf_evsel *pos; | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	int nr_cpus = cpu_map__nr(evlist->cpus); | 
					
						
							|  |  |  | 	int nr_threads = thread_map__nr(evlist->threads); | 
					
						
							| 
									
										
										
										
											2011-07-25 11:06:19 -03:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	for (cpu = 0; cpu < nr_cpus; cpu++) { | 
					
						
							| 
									
										
										
										
											2011-07-25 11:06:19 -03:00
										 |  |  | 		list_for_each_entry(pos, &evlist->entries, node) { | 
					
						
							| 
									
										
										
										
											2012-11-29 15:38:30 +09:00
										 |  |  | 			if (!perf_evsel__is_group_leader(pos)) | 
					
						
							| 
									
										
										
										
											2012-11-12 18:34:03 +01:00
										 |  |  | 				continue; | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 			for (thread = 0; thread < nr_threads; thread++) | 
					
						
							| 
									
										
										
										
											2012-05-31 14:51:46 +09:00
										 |  |  | 				ioctl(FD(pos, cpu, thread), | 
					
						
							|  |  |  | 				      PERF_EVENT_IOC_DISABLE, 0); | 
					
						
							| 
									
										
										
										
											2011-07-25 11:06:19 -03:00
										 |  |  | 		} | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-08-25 10:17:55 -06:00
										 |  |  | void perf_evlist__enable(struct perf_evlist *evlist) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	int cpu, thread; | 
					
						
							|  |  |  | 	struct perf_evsel *pos; | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	int nr_cpus = cpu_map__nr(evlist->cpus); | 
					
						
							|  |  |  | 	int nr_threads = thread_map__nr(evlist->threads); | 
					
						
							| 
									
										
										
										
											2011-08-25 10:17:55 -06:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	for (cpu = 0; cpu < nr_cpus; cpu++) { | 
					
						
							| 
									
										
										
										
											2011-08-25 10:17:55 -06:00
										 |  |  | 		list_for_each_entry(pos, &evlist->entries, node) { | 
					
						
							| 
									
										
										
										
											2012-11-29 15:38:30 +09:00
										 |  |  | 			if (!perf_evsel__is_group_leader(pos)) | 
					
						
							| 
									
										
										
										
											2012-11-12 18:34:03 +01:00
										 |  |  | 				continue; | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 			for (thread = 0; thread < nr_threads; thread++) | 
					
						
							| 
									
										
										
										
											2012-05-31 14:51:46 +09:00
										 |  |  | 				ioctl(FD(pos, cpu, thread), | 
					
						
							|  |  |  | 				      PERF_EVENT_IOC_ENABLE, 0); | 
					
						
							| 
									
										
										
										
											2011-08-25 10:17:55 -06:00
										 |  |  | 		} | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-11-29 08:05:52 -02:00
										 |  |  | static int perf_evlist__alloc_pollfd(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-01-11 22:30:02 -02:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	int nr_cpus = cpu_map__nr(evlist->cpus); | 
					
						
							|  |  |  | 	int nr_threads = thread_map__nr(evlist->threads); | 
					
						
							|  |  |  | 	int nfds = nr_cpus * nr_threads * evlist->nr_entries; | 
					
						
							| 
									
										
										
										
											2011-01-11 22:30:02 -02:00
										 |  |  | 	evlist->pollfd = malloc(sizeof(struct pollfd) * nfds); | 
					
						
							|  |  |  | 	return evlist->pollfd != NULL ? 0 : -ENOMEM; | 
					
						
							|  |  |  | } | 
					
						
							| 
									
										
										
										
											2011-01-12 17:03:24 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	fcntl(fd, F_SETFL, O_NONBLOCK); | 
					
						
							|  |  |  | 	evlist->pollfd[evlist->nr_fds].fd = fd; | 
					
						
							|  |  |  | 	evlist->pollfd[evlist->nr_fds].events = POLLIN; | 
					
						
							|  |  |  | 	evlist->nr_fds++; | 
					
						
							|  |  |  | } | 
					
						
							| 
									
										
										
										
											2011-01-12 22:39:13 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-03-10 11:15:54 -03:00
										 |  |  | static void perf_evlist__id_hash(struct perf_evlist *evlist, | 
					
						
							|  |  |  | 				 struct perf_evsel *evsel, | 
					
						
							|  |  |  | 				 int cpu, int thread, u64 id) | 
					
						
							| 
									
										
										
										
											2011-03-04 22:29:39 -03:00
										 |  |  | { | 
					
						
							|  |  |  | 	int hash; | 
					
						
							|  |  |  | 	struct perf_sample_id *sid = SID(evsel, cpu, thread); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	sid->id = id; | 
					
						
							|  |  |  | 	sid->evsel = evsel; | 
					
						
							|  |  |  | 	hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS); | 
					
						
							|  |  |  | 	hlist_add_head(&sid->node, &evlist->heads[hash]); | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-03-10 11:15:54 -03:00
										 |  |  | void perf_evlist__id_add(struct perf_evlist *evlist, struct perf_evsel *evsel, | 
					
						
							|  |  |  | 			 int cpu, int thread, u64 id) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	perf_evlist__id_hash(evlist, evsel, cpu, thread, id); | 
					
						
							|  |  |  | 	evsel->id[evsel->ids++] = id; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | static int perf_evlist__id_add_fd(struct perf_evlist *evlist, | 
					
						
							|  |  |  | 				  struct perf_evsel *evsel, | 
					
						
							|  |  |  | 				  int cpu, int thread, int fd) | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | { | 
					
						
							|  |  |  | 	u64 read_data[4] = { 0, }; | 
					
						
							| 
									
										
										
										
											2011-03-04 22:29:39 -03:00
										 |  |  | 	int id_idx = 1; /* The first entry is the counter value */ | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	if (!(evsel->attr.read_format & PERF_FORMAT_ID) || | 
					
						
							|  |  |  | 	    read(fd, &read_data, sizeof(read_data)) == -1) | 
					
						
							|  |  |  | 		return -1; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) | 
					
						
							|  |  |  | 		++id_idx; | 
					
						
							|  |  |  | 	if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) | 
					
						
							|  |  |  | 		++id_idx; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-03-10 11:15:54 -03:00
										 |  |  | 	perf_evlist__id_add(evlist, evsel, cpu, thread, read_data[id_idx]); | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 	return 0; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-12 22:39:13 -02:00
										 |  |  | struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	struct hlist_head *head; | 
					
						
							|  |  |  | 	struct perf_sample_id *sid; | 
					
						
							|  |  |  | 	int hash; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (evlist->nr_entries == 1) | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | 		return perf_evlist__first(evlist); | 
					
						
							| 
									
										
										
										
											2011-01-12 22:39:13 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	hash = hash_64(id, PERF_EVLIST__HLIST_BITS); | 
					
						
							|  |  |  | 	head = &evlist->heads[hash]; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
											  
											
												hlist: drop the node parameter from iterators
I'm not sure why, but the hlist for each entry iterators were conceived
        list_for_each_entry(pos, head, member)
The hlist ones were greedy and wanted an extra parameter:
        hlist_for_each_entry(tpos, pos, head, member)
Why did they need an extra pos parameter? I'm not quite sure. Not only
they don't really need it, it also prevents the iterator from looking
exactly like the list iterator, which is unfortunate.
Besides the semantic patch, there was some manual work required:
 - Fix up the actual hlist iterators in linux/list.h
 - Fix up the declaration of other iterators based on the hlist ones.
 - A very small amount of places were using the 'node' parameter, this
 was modified to use 'obj->member' instead.
 - Coccinelle didn't handle the hlist_for_each_entry_safe iterator
 properly, so those had to be fixed up manually.
The semantic patch which is mostly the work of Peter Senna Tschudin is here:
@@
iterator name hlist_for_each_entry, hlist_for_each_entry_continue, hlist_for_each_entry_from, hlist_for_each_entry_rcu, hlist_for_each_entry_rcu_bh, hlist_for_each_entry_continue_rcu_bh, for_each_busy_worker, ax25_uid_for_each, ax25_for_each, inet_bind_bucket_for_each, sctp_for_each_hentry, sk_for_each, sk_for_each_rcu, sk_for_each_from, sk_for_each_safe, sk_for_each_bound, hlist_for_each_entry_safe, hlist_for_each_entry_continue_rcu, nr_neigh_for_each, nr_neigh_for_each_safe, nr_node_for_each, nr_node_for_each_safe, for_each_gfn_indirect_valid_sp, for_each_gfn_sp, for_each_host;
type T;
expression a,c,d,e;
identifier b;
statement S;
@@
-T b;
    <+... when != b
(
hlist_for_each_entry(a,
- b,
c, d) S
|
hlist_for_each_entry_continue(a,
- b,
c) S
|
hlist_for_each_entry_from(a,
- b,
c) S
|
hlist_for_each_entry_rcu(a,
- b,
c, d) S
|
hlist_for_each_entry_rcu_bh(a,
- b,
c, d) S
|
hlist_for_each_entry_continue_rcu_bh(a,
- b,
c) S
|
for_each_busy_worker(a, c,
- b,
d) S
|
ax25_uid_for_each(a,
- b,
c) S
|
ax25_for_each(a,
- b,
c) S
|
inet_bind_bucket_for_each(a,
- b,
c) S
|
sctp_for_each_hentry(a,
- b,
c) S
|
sk_for_each(a,
- b,
c) S
|
sk_for_each_rcu(a,
- b,
c) S
|
sk_for_each_from
-(a, b)
+(a)
S
+ sk_for_each_from(a) S
|
sk_for_each_safe(a,
- b,
c, d) S
|
sk_for_each_bound(a,
- b,
c) S
|
hlist_for_each_entry_safe(a,
- b,
c, d, e) S
|
hlist_for_each_entry_continue_rcu(a,
- b,
c) S
|
nr_neigh_for_each(a,
- b,
c) S
|
nr_neigh_for_each_safe(a,
- b,
c, d) S
|
nr_node_for_each(a,
- b,
c) S
|
nr_node_for_each_safe(a,
- b,
c, d) S
|
- for_each_gfn_sp(a, c, d, b) S
+ for_each_gfn_sp(a, c, d) S
|
- for_each_gfn_indirect_valid_sp(a, c, d, b) S
+ for_each_gfn_indirect_valid_sp(a, c, d) S
|
for_each_host(a,
- b,
c) S
|
for_each_host_safe(a,
- b,
c, d) S
|
for_each_mesh_entry(a,
- b,
c, d) S
)
    ...+>
[akpm@linux-foundation.org: drop bogus change from net/ipv4/raw.c]
[akpm@linux-foundation.org: drop bogus hunk from net/ipv6/raw.c]
[akpm@linux-foundation.org: checkpatch fixes]
[akpm@linux-foundation.org: fix warnings]
[akpm@linux-foudnation.org: redo intrusive kvm changes]
Tested-by: Peter Senna Tschudin <peter.senna@gmail.com>
Acked-by: Paul E. McKenney <paulmck@linux.vnet.ibm.com>
Signed-off-by: Sasha Levin <sasha.levin@oracle.com>
Cc: Wu Fengguang <fengguang.wu@intel.com>
Cc: Marcelo Tosatti <mtosatti@redhat.com>
Cc: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
											
										 
											2013-02-27 17:06:00 -08:00
										 |  |  | 	hlist_for_each_entry(sid, head, node) | 
					
						
							| 
									
										
										
										
											2011-01-12 22:39:13 -02:00
										 |  |  | 		if (sid->id == id) | 
					
						
							|  |  |  | 			return sid->evsel; | 
					
						
							| 
									
										
										
										
											2012-02-20 10:47:26 +09:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	if (!perf_evlist__sample_id_all(evlist)) | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | 		return perf_evlist__first(evlist); | 
					
						
							| 
									
										
										
										
											2012-02-20 10:47:26 +09:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-12 22:39:13 -02:00
										 |  |  | 	return NULL; | 
					
						
							|  |  |  | } | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | union perf_event *perf_evlist__mmap_read(struct perf_evlist *evlist, int idx) | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | { | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 	struct perf_mmap *md = &evlist->mmap[idx]; | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | 	unsigned int head = perf_mmap__read_head(md); | 
					
						
							|  |  |  | 	unsigned int old = md->prev; | 
					
						
							|  |  |  | 	unsigned char *data = md->base + page_size; | 
					
						
							| 
									
										
										
										
											2011-01-29 14:01:45 -02:00
										 |  |  | 	union perf_event *event = NULL; | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-29 09:08:13 -02:00
										 |  |  | 	if (evlist->overwrite) { | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | 		/*
 | 
					
						
							| 
									
										
										
										
											2011-01-29 09:08:13 -02:00
										 |  |  | 		 * If we're further behind than half the buffer, there's a chance | 
					
						
							|  |  |  | 		 * the writer will bite our tail and mess up the samples under us. | 
					
						
							|  |  |  | 		 * | 
					
						
							|  |  |  | 		 * If we somehow ended up ahead of the head, we got messed up. | 
					
						
							|  |  |  | 		 * | 
					
						
							|  |  |  | 		 * In either case, truncate and restart at head. | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | 		 */ | 
					
						
							| 
									
										
										
										
											2011-01-29 09:08:13 -02:00
										 |  |  | 		int diff = head - old; | 
					
						
							|  |  |  | 		if (diff > md->mask / 2 || diff < 0) { | 
					
						
							|  |  |  | 			fprintf(stderr, "WARNING: failed to keep up with mmap data.\n"); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 			/*
 | 
					
						
							|  |  |  | 			 * head points to a known good entry, start there. | 
					
						
							|  |  |  | 			 */ | 
					
						
							|  |  |  | 			old = head; | 
					
						
							|  |  |  | 		} | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (old != head) { | 
					
						
							|  |  |  | 		size_t size; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-29 14:01:45 -02:00
										 |  |  | 		event = (union perf_event *)&data[old & md->mask]; | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | 		size = event->header.size; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 		/*
 | 
					
						
							|  |  |  | 		 * Event straddles the mmap boundary -- header should always | 
					
						
							|  |  |  | 		 * be inside due to u64 alignment of output. | 
					
						
							|  |  |  | 		 */ | 
					
						
							|  |  |  | 		if ((old & md->mask) + size != ((old + size) & md->mask)) { | 
					
						
							|  |  |  | 			unsigned int offset = old; | 
					
						
							|  |  |  | 			unsigned int len = min(sizeof(*event), size), cpy; | 
					
						
							| 
									
										
										
										
											2013-02-05 14:12:42 -07:00
										 |  |  | 			void *dst = &md->event_copy; | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 			do { | 
					
						
							|  |  |  | 				cpy = min(md->mask + 1 - (offset & md->mask), len); | 
					
						
							|  |  |  | 				memcpy(dst, &data[offset & md->mask], cpy); | 
					
						
							|  |  |  | 				offset += cpy; | 
					
						
							|  |  |  | 				dst += cpy; | 
					
						
							|  |  |  | 				len -= cpy; | 
					
						
							|  |  |  | 			} while (len); | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-02-05 14:12:42 -07:00
										 |  |  | 			event = &md->event_copy; | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | 		} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 		old += size; | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	md->prev = old; | 
					
						
							| 
									
										
										
										
											2011-01-29 09:08:13 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	if (!evlist->overwrite) | 
					
						
							|  |  |  | 		perf_mmap__write_tail(md, old); | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-15 10:40:59 -02:00
										 |  |  | 	return event; | 
					
						
							|  |  |  | } | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  | void perf_evlist__munmap(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | { | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 	int i; | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 	for (i = 0; i < evlist->nr_mmaps; i++) { | 
					
						
							|  |  |  | 		if (evlist->mmap[i].base != NULL) { | 
					
						
							|  |  |  | 			munmap(evlist->mmap[i].base, evlist->mmap_len); | 
					
						
							|  |  |  | 			evlist->mmap[i].base = NULL; | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 		} | 
					
						
							|  |  |  | 	} | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	free(evlist->mmap); | 
					
						
							|  |  |  | 	evlist->mmap = NULL; | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-11-29 08:05:52 -02:00
										 |  |  | static int perf_evlist__alloc_mmap(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2012-09-26 12:41:14 -03:00
										 |  |  | 	evlist->nr_mmaps = cpu_map__nr(evlist->cpus); | 
					
						
							|  |  |  | 	if (cpu_map__all(evlist->cpus)) | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 		evlist->nr_mmaps = thread_map__nr(evlist->threads); | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 	evlist->mmap = zalloc(evlist->nr_mmaps * sizeof(struct perf_mmap)); | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 	return evlist->mmap != NULL ? 0 : -ENOMEM; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-06-02 10:39:43 -03:00
										 |  |  | static int __perf_evlist__mmap(struct perf_evlist *evlist, | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 			       int idx, int prot, int mask, int fd) | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | { | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 	evlist->mmap[idx].prev = 0; | 
					
						
							|  |  |  | 	evlist->mmap[idx].mask = mask; | 
					
						
							|  |  |  | 	evlist->mmap[idx].base = mmap(NULL, evlist->mmap_len, prot, | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 				      MAP_SHARED, fd, 0); | 
					
						
							| 
									
										
										
										
											2011-12-19 08:39:30 -05:00
										 |  |  | 	if (evlist->mmap[idx].base == MAP_FAILED) { | 
					
						
							|  |  |  | 		evlist->mmap[idx].base = NULL; | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 		return -1; | 
					
						
							| 
									
										
										
										
											2011-12-19 08:39:30 -05:00
										 |  |  | 	} | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	perf_evlist__add_pollfd(evlist, fd); | 
					
						
							|  |  |  | 	return 0; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | static int perf_evlist__mmap_per_cpu(struct perf_evlist *evlist, int prot, int mask) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *evsel; | 
					
						
							|  |  |  | 	int cpu, thread; | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	int nr_cpus = cpu_map__nr(evlist->cpus); | 
					
						
							|  |  |  | 	int nr_threads = thread_map__nr(evlist->threads); | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	for (cpu = 0; cpu < nr_cpus; cpu++) { | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 		int output = -1; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 		for (thread = 0; thread < nr_threads; thread++) { | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 			list_for_each_entry(evsel, &evlist->entries, node) { | 
					
						
							|  |  |  | 				int fd = FD(evsel, cpu, thread); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 				if (output == -1) { | 
					
						
							|  |  |  | 					output = fd; | 
					
						
							| 
									
										
										
										
											2011-06-02 10:39:43 -03:00
										 |  |  | 					if (__perf_evlist__mmap(evlist, cpu, | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 								prot, mask, output) < 0) | 
					
						
							|  |  |  | 						goto out_unmap; | 
					
						
							|  |  |  | 				} else { | 
					
						
							|  |  |  | 					if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, output) != 0) | 
					
						
							|  |  |  | 						goto out_unmap; | 
					
						
							|  |  |  | 				} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 				if ((evsel->attr.read_format & PERF_FORMAT_ID) && | 
					
						
							|  |  |  | 				    perf_evlist__id_add_fd(evlist, evsel, cpu, thread, fd) < 0) | 
					
						
							|  |  |  | 					goto out_unmap; | 
					
						
							|  |  |  | 			} | 
					
						
							|  |  |  | 		} | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	return 0; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | out_unmap: | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	for (cpu = 0; cpu < nr_cpus; cpu++) { | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 		if (evlist->mmap[cpu].base != NULL) { | 
					
						
							|  |  |  | 			munmap(evlist->mmap[cpu].base, evlist->mmap_len); | 
					
						
							|  |  |  | 			evlist->mmap[cpu].base = NULL; | 
					
						
							|  |  |  | 		} | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 	return -1; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | static int perf_evlist__mmap_per_thread(struct perf_evlist *evlist, int prot, int mask) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *evsel; | 
					
						
							|  |  |  | 	int thread; | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	int nr_threads = thread_map__nr(evlist->threads); | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	for (thread = 0; thread < nr_threads; thread++) { | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 		int output = -1; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 		list_for_each_entry(evsel, &evlist->entries, node) { | 
					
						
							|  |  |  | 			int fd = FD(evsel, 0, thread); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 			if (output == -1) { | 
					
						
							|  |  |  | 				output = fd; | 
					
						
							| 
									
										
										
										
											2011-06-02 10:39:43 -03:00
										 |  |  | 				if (__perf_evlist__mmap(evlist, thread, | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 							prot, mask, output) < 0) | 
					
						
							|  |  |  | 					goto out_unmap; | 
					
						
							|  |  |  | 			} else { | 
					
						
							|  |  |  | 				if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, output) != 0) | 
					
						
							|  |  |  | 					goto out_unmap; | 
					
						
							|  |  |  | 			} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 			if ((evsel->attr.read_format & PERF_FORMAT_ID) && | 
					
						
							|  |  |  | 			    perf_evlist__id_add_fd(evlist, evsel, 0, thread, fd) < 0) | 
					
						
							|  |  |  | 				goto out_unmap; | 
					
						
							|  |  |  | 		} | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	return 0; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | out_unmap: | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 	for (thread = 0; thread < nr_threads; thread++) { | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 		if (evlist->mmap[thread].base != NULL) { | 
					
						
							|  |  |  | 			munmap(evlist->mmap[thread].base, evlist->mmap_len); | 
					
						
							|  |  |  | 			evlist->mmap[thread].base = NULL; | 
					
						
							|  |  |  | 		} | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 	return -1; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | /** perf_evlist__mmap - Create per cpu maps to receive events
 | 
					
						
							|  |  |  |  * | 
					
						
							|  |  |  |  * @evlist - list of events | 
					
						
							|  |  |  |  * @pages - map length in pages | 
					
						
							|  |  |  |  * @overwrite - overwrite older events? | 
					
						
							|  |  |  |  * | 
					
						
							|  |  |  |  * If overwrite is false the user needs to signal event consuption using: | 
					
						
							|  |  |  |  * | 
					
						
							|  |  |  |  *	struct perf_mmap *m = &evlist->mmap[cpu]; | 
					
						
							|  |  |  |  *	unsigned int head = perf_mmap__read_head(m); | 
					
						
							|  |  |  |  * | 
					
						
							|  |  |  |  *	perf_mmap__write_tail(m, head) | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  |  * | 
					
						
							|  |  |  |  * Using perf_evlist__read_on_cpu does this automatically. | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  |  */ | 
					
						
							| 
									
										
										
										
											2011-11-09 09:10:47 -02:00
										 |  |  | int perf_evlist__mmap(struct perf_evlist *evlist, unsigned int pages, | 
					
						
							|  |  |  | 		      bool overwrite) | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | { | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 	struct perf_evsel *evsel; | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  | 	const struct cpu_map *cpus = evlist->cpus; | 
					
						
							|  |  |  | 	const struct thread_map *threads = evlist->threads; | 
					
						
							| 
									
										
										
										
											2011-11-09 09:10:47 -02:00
										 |  |  | 	int prot = PROT_READ | (overwrite ? 0 : PROT_WRITE), mask; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |         /* 512 kiB: default amount of unprivileged mlocked memory */ | 
					
						
							|  |  |  |         if (pages == UINT_MAX) | 
					
						
							|  |  |  |                 pages = (512 * 1024) / page_size; | 
					
						
							| 
									
										
										
										
											2011-12-19 08:39:32 -05:00
										 |  |  | 	else if (!is_power_of_2(pages)) | 
					
						
							|  |  |  | 		return -EINVAL; | 
					
						
							| 
									
										
										
										
											2011-11-09 09:10:47 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	mask = pages * page_size - 1; | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  | 	if (evlist->mmap == NULL && perf_evlist__alloc_mmap(evlist) < 0) | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 		return -ENOMEM; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  | 	if (evlist->pollfd == NULL && perf_evlist__alloc_pollfd(evlist) < 0) | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 		return -ENOMEM; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	evlist->overwrite = overwrite; | 
					
						
							|  |  |  | 	evlist->mmap_len = (pages + 1) * page_size; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry(evsel, &evlist->entries, node) { | 
					
						
							|  |  |  | 		if ((evsel->attr.read_format & PERF_FORMAT_ID) && | 
					
						
							| 
									
										
										
										
											2011-03-10 11:15:54 -03:00
										 |  |  | 		    evsel->sample_id == NULL && | 
					
						
							| 
									
										
										
										
											2012-09-26 12:41:14 -03:00
										 |  |  | 		    perf_evsel__alloc_id(evsel, cpu_map__nr(cpus), threads->nr) < 0) | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 			return -ENOMEM; | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-09-26 12:41:14 -03:00
										 |  |  | 	if (cpu_map__all(cpus)) | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 		return perf_evlist__mmap_per_thread(evlist, prot, mask); | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
											  
											
												perf evlist: Fix per thread mmap setup
The PERF_EVENT_IOC_SET_OUTPUT ioctl was returning -EINVAL when using
--pid when monitoring multithreaded apps, as we can only share a ring
buffer for events on the same thread if not doing per cpu.
Fix it by using per thread ring buffers.
Tested with:
[root@felicio ~]# tuna -t 26131 -CP | nl
  1                      thread       ctxt_switches
  2    pid SCHED_ rtpri affinity voluntary nonvoluntary             cmd
  3 26131   OTHER     0      0,1  10814276      2397830 chromium-browse
  4  642    OTHER     0      0,1     14688            0 chromium-browse
  5  26148  OTHER     0      0,1    713602       115479 chromium-browse
  6  26149  OTHER     0      0,1    801958         2262 chromium-browse
  7  26150  OTHER     0      0,1   1271128          248 chromium-browse
  8  26151  OTHER     0      0,1         3            0 chromium-browse
  9  27049  OTHER     0      0,1     36796            9 chromium-browse
 10  618    OTHER     0      0,1     14711            0 chromium-browse
 11  661    OTHER     0      0,1     14593            0 chromium-browse
 12  29048  OTHER     0      0,1     28125            0 chromium-browse
 13  26143  OTHER     0      0,1   2202789          781 chromium-browse
[root@felicio ~]#
So 11 threads under pid 26131, then:
[root@felicio ~]# perf record -F 50000 --pid 26131
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
  1 7fa4a2538000-7fa4a25b9000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  2 7fa4a25b9000-7fa4a263a000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  3 7fa4a263a000-7fa4a26bb000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  4 7fa4a26bb000-7fa4a273c000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  5 7fa4a273c000-7fa4a27bd000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  6 7fa4a27bd000-7fa4a283e000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  7 7fa4a283e000-7fa4a28bf000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  8 7fa4a28bf000-7fa4a2940000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
  9 7fa4a2940000-7fa4a29c1000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 10 7fa4a29c1000-7fa4a2a42000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
 11 7fa4a2a42000-7fa4a2ac3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
11 mmaps, one per thread since we didn't specify any CPU list, so we need one
mmap per thread and:
[root@felicio ~]# perf record -F 50000 --pid 26131
^M
^C[ perf record: Woken up 79 times to write data ]
[ perf record: Captured and wrote 20.614 MB perf.data (~900639 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	 371310 26131
     2	  96516 26148
     3	  95694 26149
     4	  95203 26150
     5	   7291 26143
     6	     87 27049
     7	     76 661
     8	     60 29048
     9	     47 618
    10	     43 642
[root@felicio ~]#
Ok, one of the threads, 26151 was quiescent, so no samples there, but all the
others are there.
Then, if I specify one CPU:
[root@felicio ~]# perf record -F 50000 --pid 26131 --cpu 1
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.680 MB perf.data (~29730 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   8444 26131
     2	   2584 26149
     3	   2518 26148
     4	   2324 26150
     5	    123 26143
     6	      9 661
     7	      9 29048
[root@felicio ~]#
This machine has two cores, so fewer threads appeared on the radar, and:
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
 1 7f484b922000-7f484b9a3000 rwxs 00000000 00:09 4064 anon_inode:[perf_event]
[root@felicio ~]#
Just one mmap, as now we can use just one per-cpu buffer instead of the
per-thread needed in the previous case.
For global profiling:
[root@felicio ~]# perf record -F 50000 -a
^C[ perf record: Woken up 26 times to write data ]
[ perf record: Captured and wrote 7.128 MB perf.data (~311412 samples) ]
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7fb49b435000-7fb49b4b6000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
     2	7fb49b4b6000-7fb49b537000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
It uses per-cpu buffers.
For just one thread:
[root@felicio ~]# perf record -F 50000 --tid 26148
^C[ perf record: Woken up 2 times to write data ]
[ perf record: Captured and wrote 0.330 MB perf.data (~14426 samples) ]
[root@felicio ~]# perf report -D | grep PERF_RECORD_SAMPLE | cut -d/ -f2 | cut -d: -f1 | sort -n | uniq -c | sort -nr | nl
     1	   9969 26148
[root@felicio ~]#
[root@felicio ~]# grep perf_event /proc/`pidof perf`/maps | nl
     1	7f286a51b000-7f286a59c000 rwxs 00000000 00:09 4064                       anon_inode:[perf_event]
[root@felicio ~]#
Tested-by: David Ahern <dsahern@gmail.com>
Tested-by: Lin Ming <ming.m.lin@intel.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
Link: http://lkml.kernel.org/r/20110426204401.GB1746@ghostprotocols.net
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2011-05-15 09:39:00 -03:00
										 |  |  | 	return perf_evlist__mmap_per_cpu(evlist, prot, mask); | 
					
						
							| 
									
										
										
										
											2011-01-30 10:46:46 -02:00
										 |  |  | } | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-04-26 14:15:19 +09:00
										 |  |  | int perf_evlist__create_maps(struct perf_evlist *evlist, | 
					
						
							|  |  |  | 			     struct perf_target *target) | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2012-04-26 14:15:19 +09:00
										 |  |  | 	evlist->threads = thread_map__new_str(target->pid, target->tid, | 
					
						
							|  |  |  | 					      target->uid); | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	if (evlist->threads == NULL) | 
					
						
							|  |  |  | 		return -1; | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-05-16 18:45:48 +09:00
										 |  |  | 	if (perf_target__has_task(target)) | 
					
						
							| 
									
										
										
										
											2012-05-07 14:09:03 +09:00
										 |  |  | 		evlist->cpus = cpu_map__dummy_new(); | 
					
						
							| 
									
										
										
										
											2012-05-16 18:45:49 +09:00
										 |  |  | 	else if (!perf_target__has_cpu(target) && !target->uses_mmap) | 
					
						
							|  |  |  | 		evlist->cpus = cpu_map__dummy_new(); | 
					
						
							| 
									
										
										
										
											2012-05-16 18:45:48 +09:00
										 |  |  | 	else | 
					
						
							|  |  |  | 		evlist->cpus = cpu_map__new(target->cpu_list); | 
					
						
							| 
									
										
										
										
											2011-01-30 11:59:43 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	if (evlist->cpus == NULL) | 
					
						
							|  |  |  | 		goto out_delete_threads; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	return 0; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | out_delete_threads: | 
					
						
							|  |  |  | 	thread_map__delete(evlist->threads); | 
					
						
							|  |  |  | 	return -1; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | void perf_evlist__delete_maps(struct perf_evlist *evlist) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	cpu_map__delete(evlist->cpus); | 
					
						
							|  |  |  | 	thread_map__delete(evlist->threads); | 
					
						
							|  |  |  | 	evlist->cpus	= NULL; | 
					
						
							|  |  |  | 	evlist->threads = NULL; | 
					
						
							|  |  |  | } | 
					
						
							| 
									
										
										
										
											2011-02-26 04:51:54 +01:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-09-26 14:43:13 -03:00
										 |  |  | int perf_evlist__apply_filters(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-02-26 04:51:54 +01:00
										 |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *evsel; | 
					
						
							| 
									
										
										
										
											2012-09-26 15:07:39 -03:00
										 |  |  | 	int err = 0; | 
					
						
							|  |  |  | 	const int ncpus = cpu_map__nr(evlist->cpus), | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 		  nthreads = thread_map__nr(evlist->threads); | 
					
						
							| 
									
										
										
										
											2011-02-26 04:51:54 +01:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry(evsel, &evlist->entries, node) { | 
					
						
							| 
									
										
										
										
											2012-09-26 15:07:39 -03:00
										 |  |  | 		if (evsel->filter == NULL) | 
					
						
							| 
									
										
										
										
											2011-02-26 04:51:54 +01:00
										 |  |  | 			continue; | 
					
						
							| 
									
										
										
										
											2012-09-26 15:07:39 -03:00
										 |  |  | 
 | 
					
						
							|  |  |  | 		err = perf_evsel__set_filter(evsel, ncpus, nthreads, evsel->filter); | 
					
						
							|  |  |  | 		if (err) | 
					
						
							|  |  |  | 			break; | 
					
						
							| 
									
										
										
										
											2011-02-26 04:51:54 +01:00
										 |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-09-26 15:07:39 -03:00
										 |  |  | 	return err; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | int perf_evlist__set_filter(struct perf_evlist *evlist, const char *filter) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *evsel; | 
					
						
							|  |  |  | 	int err = 0; | 
					
						
							|  |  |  | 	const int ncpus = cpu_map__nr(evlist->cpus), | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:14 +09:00
										 |  |  | 		  nthreads = thread_map__nr(evlist->threads); | 
					
						
							| 
									
										
										
										
											2012-09-26 15:07:39 -03:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry(evsel, &evlist->entries, node) { | 
					
						
							|  |  |  | 		err = perf_evsel__set_filter(evsel, ncpus, nthreads, filter); | 
					
						
							|  |  |  | 		if (err) | 
					
						
							|  |  |  | 			break; | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	return err; | 
					
						
							| 
									
										
										
										
											2011-02-26 04:51:54 +01:00
										 |  |  | } | 
					
						
							| 
									
										
										
										
											2011-05-21 17:49:00 +02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | bool perf_evlist__valid_sample_type(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-05-21 17:49:00 +02:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | 	struct perf_evsel *first = perf_evlist__first(evlist), *pos = first; | 
					
						
							| 
									
										
										
										
											2011-06-02 11:04:54 -03:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry_continue(pos, &evlist->entries, node) { | 
					
						
							|  |  |  | 		if (first->attr.sample_type != pos->attr.sample_type) | 
					
						
							|  |  |  | 			return false; | 
					
						
							| 
									
										
										
										
											2011-05-21 17:49:00 +02:00
										 |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-06-02 11:04:54 -03:00
										 |  |  | 	return true; | 
					
						
							| 
									
										
										
										
											2011-05-21 17:49:00 +02:00
										 |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | u64 perf_evlist__sample_type(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-06-02 11:04:54 -03:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | 	struct perf_evsel *first = perf_evlist__first(evlist); | 
					
						
							| 
									
										
										
										
											2011-06-02 11:04:54 -03:00
										 |  |  | 	return first->attr.sample_type; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | u16 perf_evlist__id_hdr_size(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-11-11 22:28:50 -02:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | 	struct perf_evsel *first = perf_evlist__first(evlist); | 
					
						
							| 
									
										
										
										
											2011-11-11 22:28:50 -02:00
										 |  |  | 	struct perf_sample *data; | 
					
						
							|  |  |  | 	u64 sample_type; | 
					
						
							|  |  |  | 	u16 size = 0; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (!first->attr.sample_id_all) | 
					
						
							|  |  |  | 		goto out; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	sample_type = first->attr.sample_type; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (sample_type & PERF_SAMPLE_TID) | 
					
						
							|  |  |  | 		size += sizeof(data->tid) * 2; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  |        if (sample_type & PERF_SAMPLE_TIME) | 
					
						
							|  |  |  | 		size += sizeof(data->time); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (sample_type & PERF_SAMPLE_ID) | 
					
						
							|  |  |  | 		size += sizeof(data->id); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (sample_type & PERF_SAMPLE_STREAM_ID) | 
					
						
							|  |  |  | 		size += sizeof(data->stream_id); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (sample_type & PERF_SAMPLE_CPU) | 
					
						
							|  |  |  | 		size += sizeof(data->cpu) * 2; | 
					
						
							|  |  |  | out: | 
					
						
							|  |  |  | 	return size; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | bool perf_evlist__valid_sample_id_all(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-05-21 17:49:00 +02:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | 	struct perf_evsel *first = perf_evlist__first(evlist), *pos = first; | 
					
						
							| 
									
										
										
										
											2011-06-02 11:04:54 -03:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry_continue(pos, &evlist->entries, node) { | 
					
						
							|  |  |  | 		if (first->attr.sample_id_all != pos->attr.sample_id_all) | 
					
						
							|  |  |  | 			return false; | 
					
						
							| 
									
										
										
										
											2011-05-21 17:49:00 +02:00
										 |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2011-06-02 11:04:54 -03:00
										 |  |  | 	return true; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | bool perf_evlist__sample_id_all(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-06-02 11:04:54 -03:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | 	struct perf_evsel *first = perf_evlist__first(evlist); | 
					
						
							| 
									
										
										
										
											2011-06-02 11:04:54 -03:00
										 |  |  | 	return first->attr.sample_id_all; | 
					
						
							| 
									
										
										
										
											2011-05-21 17:49:00 +02:00
										 |  |  | } | 
					
						
							| 
									
										
										
										
											2011-10-05 19:11:32 -03:00
										 |  |  | 
 | 
					
						
							|  |  |  | void perf_evlist__set_selected(struct perf_evlist *evlist, | 
					
						
							|  |  |  | 			       struct perf_evsel *evsel) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	evlist->selected = evsel; | 
					
						
							|  |  |  | } | 
					
						
							| 
									
										
										
										
											2011-10-25 10:42:19 -02:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-03-15 14:48:48 +09:00
										 |  |  | void perf_evlist__close(struct perf_evlist *evlist) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *evsel; | 
					
						
							|  |  |  | 	int ncpus = cpu_map__nr(evlist->cpus); | 
					
						
							|  |  |  | 	int nthreads = thread_map__nr(evlist->threads); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry_reverse(evsel, &evlist->entries, node) | 
					
						
							|  |  |  | 		perf_evsel__close(evsel, ncpus, nthreads); | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
											  
											
												perf tools: Enable grouping logic for parsed events
This patch adds a functionality that allows to create event groups
based on the way they are specified on the command line. Adding
functionality to the '{}' group syntax introduced in earlier patch.
The current '--group/-g' option behaviour remains intact. If you
specify it for record/stat/top command, all the specified events
become members of a single group with the first event as a group
leader.
With the new '{}' group syntax you can create group like:
  # perf record -e '{cycles,faults}' ls
resulting in single event group containing 'cycles' and 'faults'
events, with cycles event as group leader.
All groups are created with regards to threads and cpus. Thus
recording an event group within a 2 threads on server with
4 CPUs will create 8 separate groups.
Examples (first event in brackets is group leader):
  # 1 group (cpu-clock,task-clock)
  perf record --group -e cpu-clock,task-clock ls
  perf record -e '{cpu-clock,task-clock}' ls
  # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults)
  perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls
  # 1 group (cpu-clock,task-clock,minor-faults,major-faults)
  perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls
  perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls
  # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults)
  perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \
   -e instructions ls
  # 1 group
  # (cpu-clock,task-clock,minor-faults,major-faults,instructions)
  perf record --group -e cpu-clock,task-clock \
   -e minor-faults,major-faults -e instructions ls perf record -e
'{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls
It's possible to use standard event modifier for a group, which spans
over all events in the group and updates each event modifier settings,
for example:
  # perf record -r '{faults:k,cache-references}:p'
resulting in ':kp' modifier being used for 'faults' and ':p' modifier
being used for 'cache-references' event.
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ulrich Drepper <drepper@gmail.com>
Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2012-08-08 12:22:36 +02:00
										 |  |  | int perf_evlist__open(struct perf_evlist *evlist) | 
					
						
							| 
									
										
										
										
											2011-10-25 10:42:19 -02:00
										 |  |  | { | 
					
						
							| 
									
										
											  
											
												perf tools: Enable grouping logic for parsed events
This patch adds a functionality that allows to create event groups
based on the way they are specified on the command line. Adding
functionality to the '{}' group syntax introduced in earlier patch.
The current '--group/-g' option behaviour remains intact. If you
specify it for record/stat/top command, all the specified events
become members of a single group with the first event as a group
leader.
With the new '{}' group syntax you can create group like:
  # perf record -e '{cycles,faults}' ls
resulting in single event group containing 'cycles' and 'faults'
events, with cycles event as group leader.
All groups are created with regards to threads and cpus. Thus
recording an event group within a 2 threads on server with
4 CPUs will create 8 separate groups.
Examples (first event in brackets is group leader):
  # 1 group (cpu-clock,task-clock)
  perf record --group -e cpu-clock,task-clock ls
  perf record -e '{cpu-clock,task-clock}' ls
  # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults)
  perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls
  # 1 group (cpu-clock,task-clock,minor-faults,major-faults)
  perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls
  perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls
  # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults)
  perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \
   -e instructions ls
  # 1 group
  # (cpu-clock,task-clock,minor-faults,major-faults,instructions)
  perf record --group -e cpu-clock,task-clock \
   -e minor-faults,major-faults -e instructions ls perf record -e
'{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls
It's possible to use standard event modifier for a group, which spans
over all events in the group and updates each event modifier settings,
for example:
  # perf record -r '{faults:k,cache-references}:p'
resulting in ':kp' modifier being used for 'faults' and ':p' modifier
being used for 'cache-references' event.
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ulrich Drepper <drepper@gmail.com>
Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2012-08-08 12:22:36 +02:00
										 |  |  | 	struct perf_evsel *evsel; | 
					
						
							| 
									
										
										
										
											2013-03-15 14:48:48 +09:00
										 |  |  | 	int err; | 
					
						
							| 
									
										
										
										
											2011-10-25 10:42:19 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry(evsel, &evlist->entries, node) { | 
					
						
							| 
									
										
											  
											
												perf tools: Enable grouping logic for parsed events
This patch adds a functionality that allows to create event groups
based on the way they are specified on the command line. Adding
functionality to the '{}' group syntax introduced in earlier patch.
The current '--group/-g' option behaviour remains intact. If you
specify it for record/stat/top command, all the specified events
become members of a single group with the first event as a group
leader.
With the new '{}' group syntax you can create group like:
  # perf record -e '{cycles,faults}' ls
resulting in single event group containing 'cycles' and 'faults'
events, with cycles event as group leader.
All groups are created with regards to threads and cpus. Thus
recording an event group within a 2 threads on server with
4 CPUs will create 8 separate groups.
Examples (first event in brackets is group leader):
  # 1 group (cpu-clock,task-clock)
  perf record --group -e cpu-clock,task-clock ls
  perf record -e '{cpu-clock,task-clock}' ls
  # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults)
  perf record -e '{cpu-clock,task-clock},{minor-faults,major-faults}' ls
  # 1 group (cpu-clock,task-clock,minor-faults,major-faults)
  perf record --group -e cpu-clock,task-clock -e minor-faults,major-faults ls
  perf record -e '{cpu-clock,task-clock,minor-faults,major-faults}' ls
  # 2 groups (cpu-clock,task-clock) (minor-faults,major-faults)
  perf record -e '{cpu-clock,task-clock} -e '{minor-faults,major-faults}' \
   -e instructions ls
  # 1 group
  # (cpu-clock,task-clock,minor-faults,major-faults,instructions)
  perf record --group -e cpu-clock,task-clock \
   -e minor-faults,major-faults -e instructions ls perf record -e
'{cpu-clock,task-clock,minor-faults,major-faults,instructions}' ls
It's possible to use standard event modifier for a group, which spans
over all events in the group and updates each event modifier settings,
for example:
  # perf record -r '{faults:k,cache-references}:p'
resulting in ':kp' modifier being used for 'faults' and ':p' modifier
being used for 'cache-references' event.
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Signed-off-by: Jiri Olsa <jolsa@redhat.com>
Acked-by: Peter Zijlstra <peterz@infradead.org>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: Arnaldo Carvalho de Melo <acme@ghostprotocols.net>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ulrich Drepper <drepper@gmail.com>
Link: http://lkml.kernel.org/n/tip-ho42u0wcr8mn1otkalqi13qp@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
											
										 
											2012-08-08 12:22:36 +02:00
										 |  |  | 		err = perf_evsel__open(evsel, evlist->cpus, evlist->threads); | 
					
						
							| 
									
										
										
										
											2011-10-25 10:42:19 -02:00
										 |  |  | 		if (err < 0) | 
					
						
							|  |  |  | 			goto out_err; | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	return 0; | 
					
						
							|  |  |  | out_err: | 
					
						
							| 
									
										
										
										
											2013-03-15 14:48:48 +09:00
										 |  |  | 	perf_evlist__close(evlist); | 
					
						
							| 
									
										
										
										
											2012-02-23 12:13:36 +09:00
										 |  |  | 	errno = -err; | 
					
						
							| 
									
										
										
										
											2011-10-25 10:42:19 -02:00
										 |  |  | 	return err; | 
					
						
							|  |  |  | } | 
					
						
							| 
									
										
										
										
											2011-11-09 08:47:15 -02:00
										 |  |  | 
 | 
					
						
							|  |  |  | int perf_evlist__prepare_workload(struct perf_evlist *evlist, | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:15 +09:00
										 |  |  | 				  struct perf_target *target, | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:17 +09:00
										 |  |  | 				  const char *argv[], bool pipe_output, | 
					
						
							|  |  |  | 				  bool want_signal) | 
					
						
							| 
									
										
										
										
											2011-11-09 08:47:15 -02:00
										 |  |  | { | 
					
						
							|  |  |  | 	int child_ready_pipe[2], go_pipe[2]; | 
					
						
							|  |  |  | 	char bf; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (pipe(child_ready_pipe) < 0) { | 
					
						
							|  |  |  | 		perror("failed to create 'ready' pipe"); | 
					
						
							|  |  |  | 		return -1; | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (pipe(go_pipe) < 0) { | 
					
						
							|  |  |  | 		perror("failed to create 'go' pipe"); | 
					
						
							|  |  |  | 		goto out_close_ready_pipe; | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	evlist->workload.pid = fork(); | 
					
						
							|  |  |  | 	if (evlist->workload.pid < 0) { | 
					
						
							|  |  |  | 		perror("failed to fork"); | 
					
						
							|  |  |  | 		goto out_close_pipes; | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	if (!evlist->workload.pid) { | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:16 +09:00
										 |  |  | 		if (pipe_output) | 
					
						
							| 
									
										
										
										
											2011-11-09 08:47:15 -02:00
										 |  |  | 			dup2(2, 1); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 		close(child_ready_pipe[0]); | 
					
						
							|  |  |  | 		close(go_pipe[1]); | 
					
						
							|  |  |  | 		fcntl(go_pipe[0], F_SETFD, FD_CLOEXEC); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 		/*
 | 
					
						
							|  |  |  | 		 * Do a dummy execvp to get the PLT entry resolved, | 
					
						
							|  |  |  | 		 * so we avoid the resolver overhead on the real | 
					
						
							|  |  |  | 		 * execvp call. | 
					
						
							|  |  |  | 		 */ | 
					
						
							|  |  |  | 		execvp("", (char **)argv); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 		/*
 | 
					
						
							|  |  |  | 		 * Tell the parent we're ready to go | 
					
						
							|  |  |  | 		 */ | 
					
						
							|  |  |  | 		close(child_ready_pipe[1]); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 		/*
 | 
					
						
							|  |  |  | 		 * Wait until the parent tells us to go. | 
					
						
							|  |  |  | 		 */ | 
					
						
							|  |  |  | 		if (read(go_pipe[0], &bf, 1) == -1) | 
					
						
							|  |  |  | 			perror("unable to read pipe"); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 		execvp(argv[0], (char **)argv); | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 		perror(argv[0]); | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:17 +09:00
										 |  |  | 		if (want_signal) | 
					
						
							|  |  |  | 			kill(getppid(), SIGUSR1); | 
					
						
							| 
									
										
										
										
											2011-11-09 08:47:15 -02:00
										 |  |  | 		exit(-1); | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							| 
									
										
										
										
											2013-03-11 16:43:15 +09:00
										 |  |  | 	if (perf_target__none(target)) | 
					
						
							| 
									
										
										
										
											2011-11-09 08:47:15 -02:00
										 |  |  | 		evlist->threads->map[0] = evlist->workload.pid; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	close(child_ready_pipe[1]); | 
					
						
							|  |  |  | 	close(go_pipe[0]); | 
					
						
							|  |  |  | 	/*
 | 
					
						
							|  |  |  | 	 * wait for child to settle | 
					
						
							|  |  |  | 	 */ | 
					
						
							|  |  |  | 	if (read(child_ready_pipe[0], &bf, 1) == -1) { | 
					
						
							|  |  |  | 		perror("unable to read pipe"); | 
					
						
							|  |  |  | 		goto out_close_pipes; | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	evlist->workload.cork_fd = go_pipe[1]; | 
					
						
							|  |  |  | 	close(child_ready_pipe[0]); | 
					
						
							|  |  |  | 	return 0; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | out_close_pipes: | 
					
						
							|  |  |  | 	close(go_pipe[0]); | 
					
						
							|  |  |  | 	close(go_pipe[1]); | 
					
						
							|  |  |  | out_close_ready_pipe: | 
					
						
							|  |  |  | 	close(child_ready_pipe[0]); | 
					
						
							|  |  |  | 	close(child_ready_pipe[1]); | 
					
						
							|  |  |  | 	return -1; | 
					
						
							|  |  |  | } | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | int perf_evlist__start_workload(struct perf_evlist *evlist) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	if (evlist->workload.cork_fd > 0) { | 
					
						
							|  |  |  | 		/*
 | 
					
						
							|  |  |  | 		 * Remove the cork, let it rip! | 
					
						
							|  |  |  | 		 */ | 
					
						
							|  |  |  | 		return close(evlist->workload.cork_fd); | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	return 0; | 
					
						
							|  |  |  | } | 
					
						
							| 
									
										
										
										
											2012-08-02 11:42:57 -03:00
										 |  |  | 
 | 
					
						
							| 
									
										
										
										
											2012-08-02 12:23:46 -03:00
										 |  |  | int perf_evlist__parse_sample(struct perf_evlist *evlist, union perf_event *event, | 
					
						
							| 
									
										
										
										
											2012-09-26 12:48:18 -03:00
										 |  |  | 			      struct perf_sample *sample) | 
					
						
							| 
									
										
										
										
											2012-08-02 11:42:57 -03:00
										 |  |  | { | 
					
						
							| 
									
										
										
										
											2012-08-14 16:42:15 -03:00
										 |  |  | 	struct perf_evsel *evsel = perf_evlist__first(evlist); | 
					
						
							| 
									
										
										
										
											2012-09-26 12:48:18 -03:00
										 |  |  | 	return perf_evsel__parse_sample(evsel, event, sample); | 
					
						
							| 
									
										
										
										
											2012-08-02 11:42:57 -03:00
										 |  |  | } | 
					
						
							| 
									
										
										
										
											2012-09-06 14:54:11 -03:00
										 |  |  | 
 | 
					
						
							|  |  |  | size_t perf_evlist__fprintf(struct perf_evlist *evlist, FILE *fp) | 
					
						
							|  |  |  | { | 
					
						
							|  |  |  | 	struct perf_evsel *evsel; | 
					
						
							|  |  |  | 	size_t printed = 0; | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	list_for_each_entry(evsel, &evlist->entries, node) { | 
					
						
							|  |  |  | 		printed += fprintf(fp, "%s%s", evsel->idx ? ", " : "", | 
					
						
							|  |  |  | 				   perf_evsel__name(evsel)); | 
					
						
							|  |  |  | 	} | 
					
						
							|  |  |  | 
 | 
					
						
							|  |  |  | 	return printed + fprintf(fp, "\n");; | 
					
						
							|  |  |  | } |