* [PATCH] perf, tools, script: Skip locking overhead in perf script
@ 2017-03-24 0:28 Andi Kleen
2017-04-04 16:17 ` Arnaldo Carvalho de Melo
0 siblings, 1 reply; 3+ messages in thread
From: Andi Kleen @ 2017-03-24 0:28 UTC (permalink / raw)
To: acme; +Cc: jolsa, linux-kernel, Andi Kleen
From: Andi Kleen <ak@linux.intel.com>
I was tired of seeing ~10% locking overhead in perf script while
doing larger PT decodes. perf script doesn't need any locking
because it is single threaded. Nearly all of it comes from
looking at the maps cache. I added a global variable to indicate
perf is running single threaded, and make perf script set that.
I only did it for locks that actually show up in profiles,
not for ones that do not. This is currently only the mapping lock.
Potentially other tools (probably all except for top/report)
could use that.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
---
tools/perf/builtin-stat.c | 2 ++
tools/perf/perf.h | 3 +++
tools/perf/util/map.c | 7 +++++--
tools/perf/util/util.c | 7 +++++++
4 files changed, 17 insertions(+), 2 deletions(-)
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index f53f449d864d..b404644f559a 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -2393,6 +2393,8 @@ int cmd_stat(int argc, const char **argv, const char *prefix __maybe_unused)
setlocale(LC_ALL, "");
+ perf_set_singlethreaded();
+
evsel_list = perf_evlist__new();
if (evsel_list == NULL)
return -ENOMEM;
diff --git a/tools/perf/perf.h b/tools/perf/perf.h
index 806c216a1078..d0fbbd168987 100644
--- a/tools/perf/perf.h
+++ b/tools/perf/perf.h
@@ -77,4 +77,7 @@ struct record_opts {
struct option;
extern const char * const *record_usage;
extern struct option *record_options;
+extern int perf_st;
+extern void perf_set_singlethreaded(void);
+
#endif
diff --git a/tools/perf/util/map.c b/tools/perf/util/map.c
index 1d9ebcf9e38e..cd8146323d8a 100644
--- a/tools/perf/util/map.c
+++ b/tools/perf/util/map.c
@@ -1,3 +1,4 @@
+#include "perf.h"
#include "symbol.h"
#include <errno.h>
#include <inttypes.h>
@@ -825,7 +826,8 @@ struct map *maps__find(struct maps *maps, u64 ip)
struct rb_node **p, *parent = NULL;
struct map *m;
- pthread_rwlock_rdlock(&maps->lock);
+ if (!perf_st)
+ pthread_rwlock_rdlock(&maps->lock);
p = &maps->entries.rb_node;
while (*p != NULL) {
@@ -841,7 +843,8 @@ struct map *maps__find(struct maps *maps, u64 ip)
m = NULL;
out:
- pthread_rwlock_unlock(&maps->lock);
+ if (!perf_st)
+ pthread_rwlock_unlock(&maps->lock);
return m;
}
diff --git a/tools/perf/util/util.c b/tools/perf/util/util.c
index d8b45cea54d0..6a3a8854da63 100644
--- a/tools/perf/util/util.c
+++ b/tools/perf/util/util.c
@@ -802,3 +802,10 @@ int unit_number__scnprintf(char *buf, size_t size, u64 n)
return scnprintf(buf, size, "%" PRIu64 "%c", n, unit[i]);
}
+
+int perf_st;
+
+void perf_set_singlethreaded(void)
+{
+ perf_st = 1;
+}
--
2.9.3
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH] perf, tools, script: Skip locking overhead in perf script
2017-03-24 0:28 [PATCH] perf, tools, script: Skip locking overhead in perf script Andi Kleen
@ 2017-04-04 16:17 ` Arnaldo Carvalho de Melo
2017-09-22 16:46 ` [tip:perf/core] perf tools: Provide mutex wrappers for pthreads rwlocks tip-bot for Arnaldo Carvalho de Melo
0 siblings, 1 reply; 3+ messages in thread
From: Arnaldo Carvalho de Melo @ 2017-04-04 16:17 UTC (permalink / raw)
To: Andi Kleen
Cc: Jiri Olsa, linux-kernel, Andi Kleen, Namhyung Kim, David Ahern,
Ingo Molnar, Wang Nan
Em Thu, Mar 23, 2017 at 05:28:07PM -0700, Andi Kleen escreveu:
> From: Andi Kleen <ak@linux.intel.com>
>
> I was tired of seeing ~10% locking overhead in perf script while
> doing larger PT decodes. perf script doesn't need any locking
> diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> index f53f449d864d..b404644f559a 100644
> --- a/tools/perf/builtin-stat.c
> +++ b/tools/perf/builtin-stat.c
Huh, should this be builtin-script? Anyway, below would be a more
complete patch, but I couldn't find improvement at least with a 50M
perf.data file with intel_pt data... will try some more later.
- Arnaldo
commit 42ee7ffd3f88047839837d9b35eef23740a86429
Author: Arnaldo Carvalho de Melo <acme@redhat.com>
Date: Tue Apr 4 13:15:04 2017 -0300
WIP
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 46acc8ece41f..e6262d36f0de 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -2493,6 +2493,8 @@ int cmd_script(int argc, const char **argv)
NULL
};
+ perf_set_singlethreaded();
+
setup_scripting();
argc = parse_options_subcommand(argc, argv, options, script_subcommands, script_usage,
diff --git a/tools/perf/util/Build b/tools/perf/util/Build
index 5c0ea11a8f0a..9319291414a5 100644
--- a/tools/perf/util/Build
+++ b/tools/perf/util/Build
@@ -75,6 +75,7 @@ libperf-y += data.o
libperf-y += tsc.o
libperf-y += cloexec.o
libperf-y += call-path.o
+libperf-y += rwsem.o
libperf-y += thread-stack.o
libperf-$(CONFIG_AUXTRACE) += auxtrace.o
libperf-$(CONFIG_AUXTRACE) += intel-pt-decoder/
diff --git a/tools/perf/util/map.c b/tools/perf/util/map.c
index c1870ac365a3..08dd01fd2b44 100644
--- a/tools/perf/util/map.c
+++ b/tools/perf/util/map.c
@@ -476,7 +476,7 @@ u64 map__objdump_2mem(struct map *map, u64 ip)
static void maps__init(struct maps *maps)
{
maps->entries = RB_ROOT;
- pthread_rwlock_init(&maps->lock, NULL);
+ init_rwsem(&maps->lock);
}
void map_groups__init(struct map_groups *mg, struct machine *machine)
@@ -505,9 +505,9 @@ static void __maps__purge(struct maps *maps)
static void maps__exit(struct maps *maps)
{
- pthread_rwlock_wrlock(&maps->lock);
+ down_write(&maps->lock);
__maps__purge(maps);
- pthread_rwlock_unlock(&maps->lock);
+ up_write(&maps->lock);
}
void map_groups__exit(struct map_groups *mg)
@@ -574,7 +574,7 @@ struct symbol *maps__find_symbol_by_name(struct maps *maps, const char *name,
struct symbol *sym;
struct rb_node *nd;
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
for (nd = rb_first(&maps->entries); nd; nd = rb_next(nd)) {
struct map *pos = rb_entry(nd, struct map, rb_node);
@@ -590,7 +590,7 @@ struct symbol *maps__find_symbol_by_name(struct maps *maps, const char *name,
sym = NULL;
out:
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
return sym;
}
@@ -626,7 +626,7 @@ static size_t maps__fprintf(struct maps *maps, FILE *fp)
size_t printed = 0;
struct rb_node *nd;
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
for (nd = rb_first(&maps->entries); nd; nd = rb_next(nd)) {
struct map *pos = rb_entry(nd, struct map, rb_node);
@@ -638,7 +638,7 @@ static size_t maps__fprintf(struct maps *maps, FILE *fp)
}
}
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
return printed;
}
@@ -670,7 +670,7 @@ static int maps__fixup_overlappings(struct maps *maps, struct map *map, FILE *fp
struct rb_node *next;
int err = 0;
- pthread_rwlock_wrlock(&maps->lock);
+ down_write(&maps->lock);
root = &maps->entries;
next = rb_first(root);
@@ -738,7 +738,7 @@ static int maps__fixup_overlappings(struct maps *maps, struct map *map, FILE *fp
err = 0;
out:
- pthread_rwlock_unlock(&maps->lock);
+ up_write(&maps->lock);
return err;
}
@@ -759,7 +759,7 @@ int map_groups__clone(struct thread *thread,
struct map *map;
struct maps *maps = &parent->maps[type];
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
for (map = maps__first(maps); map; map = map__next(map)) {
struct map *new = map__clone(map);
@@ -776,7 +776,7 @@ int map_groups__clone(struct thread *thread,
err = 0;
out_unlock:
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
return err;
}
@@ -803,9 +803,9 @@ static void __maps__insert(struct maps *maps, struct map *map)
void maps__insert(struct maps *maps, struct map *map)
{
- pthread_rwlock_wrlock(&maps->lock);
+ down_write(&maps->lock);
__maps__insert(maps, map);
- pthread_rwlock_unlock(&maps->lock);
+ up_write(&maps->lock);
}
static void __maps__remove(struct maps *maps, struct map *map)
@@ -816,9 +816,9 @@ static void __maps__remove(struct maps *maps, struct map *map)
void maps__remove(struct maps *maps, struct map *map)
{
- pthread_rwlock_wrlock(&maps->lock);
+ down_write(&maps->lock);
__maps__remove(maps, map);
- pthread_rwlock_unlock(&maps->lock);
+ up_write(&maps->lock);
}
struct map *maps__find(struct maps *maps, u64 ip)
@@ -826,7 +826,7 @@ struct map *maps__find(struct maps *maps, u64 ip)
struct rb_node **p, *parent = NULL;
struct map *m;
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
p = &maps->entries.rb_node;
while (*p != NULL) {
@@ -842,7 +842,7 @@ struct map *maps__find(struct maps *maps, u64 ip)
m = NULL;
out:
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
return m;
}
diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h
index c8a5a644c0a9..fd76534a7807 100644
--- a/tools/perf/util/map.h
+++ b/tools/perf/util/map.h
@@ -9,6 +9,7 @@
#include <stdio.h>
#include <stdbool.h>
#include <linux/types.h>
+#include "rwsem.h"
enum map_type {
MAP__FUNCTION = 0,
@@ -61,7 +62,7 @@ struct kmap {
struct maps {
struct rb_root entries;
- pthread_rwlock_t lock;
+ struct rw_semaphore lock;
};
struct map_groups {
diff --git a/tools/perf/util/rwsem.c b/tools/perf/util/rwsem.c
new file mode 100644
index 000000000000..02b0f18b96eb
--- /dev/null
+++ b/tools/perf/util/rwsem.c
@@ -0,0 +1,27 @@
+#include "util.h"
+#include "rwsem.h"
+
+int init_rwsem(struct rw_semaphore *sem)
+{
+ return perf_singlethreaded ? 0 : pthread_rwlock_init(&sem->lock, NULL);
+}
+
+int down_read(struct rw_semaphore *sem)
+{
+ return perf_singlethreaded ? 0 : pthread_rwlock_rdlock(&sem->lock);
+}
+
+int up_read(struct rw_semaphore *sem)
+{
+ return perf_singlethreaded ? 0 : pthread_rwlock_unlock(&sem->lock);
+}
+
+int down_write(struct rw_semaphore *sem)
+{
+ return perf_singlethreaded ? 0 : pthread_rwlock_wrlock(&sem->lock);
+}
+
+int up_write(struct rw_semaphore *sem)
+{
+ return perf_singlethreaded ? 0 : pthread_rwlock_unlock(&sem->lock);
+}
diff --git a/tools/perf/util/rwsem.h b/tools/perf/util/rwsem.h
new file mode 100644
index 000000000000..c06afd42dc11
--- /dev/null
+++ b/tools/perf/util/rwsem.h
@@ -0,0 +1,18 @@
+#ifndef _PERF_RWSEM_H
+#define _PERF_RWSEM_H
+
+#include <pthread.h>
+
+struct rw_semaphore {
+ pthread_rwlock_t lock;
+};
+
+int init_rwsem(struct rw_semaphore *sem);
+
+int down_read(struct rw_semaphore *sem);
+int up_read(struct rw_semaphore *sem);
+
+int down_write(struct rw_semaphore *sem);
+int up_write(struct rw_semaphore *sem);
+
+#endif /* _PERF_RWSEM_H */
diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
index 9b4d8ba22fed..d3caa5e5f845 100644
--- a/tools/perf/util/symbol.c
+++ b/tools/perf/util/symbol.c
@@ -210,7 +210,7 @@ void __map_groups__fixup_end(struct map_groups *mg, enum map_type type)
struct maps *maps = &mg->maps[type];
struct map *next, *curr;
- pthread_rwlock_wrlock(&maps->lock);
+ down_write(&maps->lock);
curr = maps__first(maps);
if (curr == NULL)
@@ -228,7 +228,7 @@ void __map_groups__fixup_end(struct map_groups *mg, enum map_type type)
curr->end = ~0ULL;
out_unlock:
- pthread_rwlock_unlock(&maps->lock);
+ up_write(&maps->lock);
}
struct symbol *symbol__new(u64 start, u64 len, u8 binding, const char *name)
@@ -1561,7 +1561,7 @@ struct map *map_groups__find_by_name(struct map_groups *mg,
struct maps *maps = &mg->maps[type];
struct map *map;
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
for (map = maps__first(maps); map; map = map__next(map)) {
if (map->dso && strcmp(map->dso->short_name, name) == 0)
@@ -1571,7 +1571,7 @@ struct map *map_groups__find_by_name(struct map_groups *mg,
map = NULL;
out_unlock:
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
return map;
}
diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c
index dcdb87a5d0a1..4da6e776a189 100644
--- a/tools/perf/util/thread.c
+++ b/tools/perf/util/thread.c
@@ -259,7 +259,7 @@ static int __thread__prepare_access(struct thread *thread)
struct maps *maps = &thread->mg->maps[i];
struct map *map;
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
for (map = maps__first(maps); map; map = map__next(map)) {
err = unwind__prepare_access(thread, map, &initialized);
@@ -267,7 +267,7 @@ static int __thread__prepare_access(struct thread *thread)
break;
}
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
}
return err;
diff --git a/tools/perf/util/util.c b/tools/perf/util/util.c
index d8b45cea54d0..a016880a55e6 100644
--- a/tools/perf/util/util.c
+++ b/tools/perf/util/util.c
@@ -20,6 +20,13 @@
#include "callchain.h"
#include "strlist.h"
+bool perf_singlethreaded;
+
+void perf_set_singlethreaded(void)
+{
+ perf_singlethreaded = true;
+}
+
#define CALLCHAIN_PARAM_DEFAULT \
.mode = CHAIN_GRAPH_ABS, \
.min_percent = 0.5, \
diff --git a/tools/perf/util/util.h b/tools/perf/util/util.h
index 7cf5752b38fd..e59aa319a2e1 100644
--- a/tools/perf/util/util.h
+++ b/tools/perf/util/util.h
@@ -380,4 +380,7 @@ struct inline_node {
struct inline_node *dso__parse_addr_inlines(struct dso *dso, u64 addr);
void inline_node__delete(struct inline_node *node);
+extern bool perf_singlethreaded;
+void perf_set_singlethreaded(void);
+
#endif /* GIT_COMPAT_UTIL_H */
^ permalink raw reply related [flat|nested] 3+ messages in thread
* [tip:perf/core] perf tools: Provide mutex wrappers for pthreads rwlocks
2017-04-04 16:17 ` Arnaldo Carvalho de Melo
@ 2017-09-22 16:46 ` tip-bot for Arnaldo Carvalho de Melo
0 siblings, 0 replies; 3+ messages in thread
From: tip-bot for Arnaldo Carvalho de Melo @ 2017-09-22 16:46 UTC (permalink / raw)
To: linux-tip-commits
Cc: jolsa, kan.liang, peterz, acme, mingo, dsahern, namhyung,
linux-kernel, adrian.hunter, tglx, ak, wangnan0, hpa
Commit-ID: 0a7c74eae307894c6c95316c382f118aef8481e8
Gitweb: http://git.kernel.org/tip/0a7c74eae307894c6c95316c382f118aef8481e8
Author: Arnaldo Carvalho de Melo <acme@redhat.com>
AuthorDate: Tue, 4 Apr 2017 13:15:04 -0300
Committer: Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Thu, 21 Sep 2017 13:28:06 -0300
perf tools: Provide mutex wrappers for pthreads rwlocks
Andi reported a performance drop in single threaded perf tools such as
'perf script' due to the growing number of locks being put in place to
allow for multithreaded tools, so wrap the POSIX threads rwlock routines
with the names used for such kinds of locks in the Linux kernel and then
allow for tools to ask for those locks to be used or not.
I.e. a tool may have a multithreaded phase and then switch to single
threaded, like the upcoming patches for the synthesizing of
PERF_RECORD_{FORK,MMAP,etc} for pre-existing processes to then switch to
single threaded mode in 'perf top'.
The init routines will not be conditional, this way starting as single
threaded to then move to multi threaded mode should be possible.
Reported-by: Andi Kleen <ak@linux.intel.com>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kan Liang <kan.liang@intel.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/20170404161739.GH12903@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
tools/perf/builtin-kvm.c | 1 -
tools/perf/builtin-script.c | 2 ++
tools/perf/util/Build | 1 +
tools/perf/util/dso.c | 12 +++++------
tools/perf/util/dso.h | 4 ++--
tools/perf/util/machine.c | 41 +++++++++++++++++++-------------------
tools/perf/util/machine.h | 3 ++-
tools/perf/util/map.c | 34 +++++++++++++++----------------
tools/perf/util/map.h | 3 ++-
tools/perf/util/rwsem.c | 32 +++++++++++++++++++++++++++++
tools/perf/util/rwsem.h | 19 ++++++++++++++++++
tools/perf/util/symbol.c | 8 ++++----
tools/perf/util/thread.c | 4 ++--
tools/perf/util/trace-event-info.c | 1 -
tools/perf/util/trace-event-read.c | 1 -
tools/perf/util/util.c | 13 ++++++++++++
tools/perf/util/util.h | 5 +++++
tools/perf/util/vdso.c | 4 ++--
18 files changed, 130 insertions(+), 58 deletions(-)
diff --git a/tools/perf/builtin-kvm.c b/tools/perf/builtin-kvm.c
index f309c37..c747a1a 100644
--- a/tools/perf/builtin-kvm.c
+++ b/tools/perf/builtin-kvm.c
@@ -34,7 +34,6 @@
#include <termios.h>
#include <semaphore.h>
#include <signal.h>
-#include <pthread.h>
#include <math.h>
static const char *get_filename_for_perf_kvm(void)
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 725dbd3..9092de0 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -2829,6 +2829,8 @@ int cmd_script(int argc, const char **argv)
NULL
};
+ perf_set_singlethreaded();
+
setup_scripting();
argc = parse_options_subcommand(argc, argv, options, script_subcommands, script_usage,
diff --git a/tools/perf/util/Build b/tools/perf/util/Build
index 71ab846..369c316 100644
--- a/tools/perf/util/Build
+++ b/tools/perf/util/Build
@@ -79,6 +79,7 @@ libperf-y += data.o
libperf-y += tsc.o
libperf-y += cloexec.o
libperf-y += call-path.o
+libperf-y += rwsem.o
libperf-y += thread-stack.o
libperf-$(CONFIG_AUXTRACE) += auxtrace.o
libperf-$(CONFIG_AUXTRACE) += intel-pt-decoder/
diff --git a/tools/perf/util/dso.c b/tools/perf/util/dso.c
index ffd7231..339e529 100644
--- a/tools/perf/util/dso.c
+++ b/tools/perf/util/dso.c
@@ -1366,9 +1366,9 @@ void __dsos__add(struct dsos *dsos, struct dso *dso)
void dsos__add(struct dsos *dsos, struct dso *dso)
{
- pthread_rwlock_wrlock(&dsos->lock);
+ down_write(&dsos->lock);
__dsos__add(dsos, dso);
- pthread_rwlock_unlock(&dsos->lock);
+ up_write(&dsos->lock);
}
struct dso *__dsos__find(struct dsos *dsos, const char *name, bool cmp_short)
@@ -1387,9 +1387,9 @@ struct dso *__dsos__find(struct dsos *dsos, const char *name, bool cmp_short)
struct dso *dsos__find(struct dsos *dsos, const char *name, bool cmp_short)
{
struct dso *dso;
- pthread_rwlock_rdlock(&dsos->lock);
+ down_read(&dsos->lock);
dso = __dsos__find(dsos, name, cmp_short);
- pthread_rwlock_unlock(&dsos->lock);
+ up_read(&dsos->lock);
return dso;
}
@@ -1416,9 +1416,9 @@ struct dso *__dsos__findnew(struct dsos *dsos, const char *name)
struct dso *dsos__findnew(struct dsos *dsos, const char *name)
{
struct dso *dso;
- pthread_rwlock_wrlock(&dsos->lock);
+ down_write(&dsos->lock);
dso = dso__get(__dsos__findnew(dsos, name));
- pthread_rwlock_unlock(&dsos->lock);
+ up_write(&dsos->lock);
return dso;
}
diff --git a/tools/perf/util/dso.h b/tools/perf/util/dso.h
index f886141..a2bbb21 100644
--- a/tools/perf/util/dso.h
+++ b/tools/perf/util/dso.h
@@ -6,7 +6,7 @@
#include <linux/rbtree.h>
#include <sys/types.h>
#include <stdbool.h>
-#include <pthread.h>
+#include "rwsem.h"
#include <linux/types.h>
#include <linux/bitops.h>
#include "map.h"
@@ -129,7 +129,7 @@ struct dso_cache {
struct dsos {
struct list_head head;
struct rb_root root; /* rbtree root sorted by long name */
- pthread_rwlock_t lock;
+ struct rw_semaphore lock;
};
struct auxtrace_cache;
diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c
index ddeea05..585b4a3 100644
--- a/tools/perf/util/machine.c
+++ b/tools/perf/util/machine.c
@@ -30,7 +30,7 @@ static void dsos__init(struct dsos *dsos)
{
INIT_LIST_HEAD(&dsos->head);
dsos->root = RB_ROOT;
- pthread_rwlock_init(&dsos->lock, NULL);
+ init_rwsem(&dsos->lock);
}
static void machine__threads_init(struct machine *machine)
@@ -40,7 +40,7 @@ static void machine__threads_init(struct machine *machine)
for (i = 0; i < THREADS__TABLE_SIZE; i++) {
struct threads *threads = &machine->threads[i];
threads->entries = RB_ROOT;
- pthread_rwlock_init(&threads->lock, NULL);
+ init_rwsem(&threads->lock);
threads->nr = 0;
INIT_LIST_HEAD(&threads->dead);
threads->last_match = NULL;
@@ -130,7 +130,7 @@ static void dsos__purge(struct dsos *dsos)
{
struct dso *pos, *n;
- pthread_rwlock_wrlock(&dsos->lock);
+ down_write(&dsos->lock);
list_for_each_entry_safe(pos, n, &dsos->head, node) {
RB_CLEAR_NODE(&pos->rb_node);
@@ -139,13 +139,13 @@ static void dsos__purge(struct dsos *dsos)
dso__put(pos);
}
- pthread_rwlock_unlock(&dsos->lock);
+ up_write(&dsos->lock);
}
static void dsos__exit(struct dsos *dsos)
{
dsos__purge(dsos);
- pthread_rwlock_destroy(&dsos->lock);
+ exit_rwsem(&dsos->lock);
}
void machine__delete_threads(struct machine *machine)
@@ -155,7 +155,7 @@ void machine__delete_threads(struct machine *machine)
for (i = 0; i < THREADS__TABLE_SIZE; i++) {
struct threads *threads = &machine->threads[i];
- pthread_rwlock_wrlock(&threads->lock);
+ down_write(&threads->lock);
nd = rb_first(&threads->entries);
while (nd) {
struct thread *t = rb_entry(nd, struct thread, rb_node);
@@ -163,7 +163,7 @@ void machine__delete_threads(struct machine *machine)
nd = rb_next(nd);
__machine__remove_thread(machine, t, false);
}
- pthread_rwlock_unlock(&threads->lock);
+ up_write(&threads->lock);
}
}
@@ -180,7 +180,7 @@ void machine__exit(struct machine *machine)
for (i = 0; i < THREADS__TABLE_SIZE; i++) {
struct threads *threads = &machine->threads[i];
- pthread_rwlock_destroy(&threads->lock);
+ exit_rwsem(&threads->lock);
}
}
@@ -482,9 +482,9 @@ struct thread *machine__findnew_thread(struct machine *machine, pid_t pid,
struct threads *threads = machine__threads(machine, tid);
struct thread *th;
- pthread_rwlock_wrlock(&threads->lock);
+ down_write(&threads->lock);
th = __machine__findnew_thread(machine, pid, tid);
- pthread_rwlock_unlock(&threads->lock);
+ up_write(&threads->lock);
return th;
}
@@ -494,9 +494,9 @@ struct thread *machine__find_thread(struct machine *machine, pid_t pid,
struct threads *threads = machine__threads(machine, tid);
struct thread *th;
- pthread_rwlock_rdlock(&threads->lock);
+ down_read(&threads->lock);
th = ____machine__findnew_thread(machine, threads, pid, tid, false);
- pthread_rwlock_unlock(&threads->lock);
+ up_read(&threads->lock);
return th;
}
@@ -588,7 +588,7 @@ static struct dso *machine__findnew_module_dso(struct machine *machine,
{
struct dso *dso;
- pthread_rwlock_wrlock(&machine->dsos.lock);
+ down_write(&machine->dsos.lock);
dso = __dsos__find(&machine->dsos, m->name, true);
if (!dso) {
@@ -602,7 +602,7 @@ static struct dso *machine__findnew_module_dso(struct machine *machine,
dso__get(dso);
out_unlock:
- pthread_rwlock_unlock(&machine->dsos.lock);
+ up_write(&machine->dsos.lock);
return dso;
}
@@ -749,7 +749,8 @@ size_t machine__fprintf(struct machine *machine, FILE *fp)
for (i = 0; i < THREADS__TABLE_SIZE; i++) {
struct threads *threads = &machine->threads[i];
- pthread_rwlock_rdlock(&threads->lock);
+
+ down_read(&threads->lock);
ret = fprintf(fp, "Threads: %u\n", threads->nr);
@@ -759,7 +760,7 @@ size_t machine__fprintf(struct machine *machine, FILE *fp)
ret += thread__fprintf(pos, fp);
}
- pthread_rwlock_unlock(&threads->lock);
+ up_read(&threads->lock);
}
return ret;
}
@@ -1319,7 +1320,7 @@ static int machine__process_kernel_mmap_event(struct machine *machine,
struct dso *kernel = NULL;
struct dso *dso;
- pthread_rwlock_rdlock(&machine->dsos.lock);
+ down_read(&machine->dsos.lock);
list_for_each_entry(dso, &machine->dsos.head, node) {
@@ -1349,7 +1350,7 @@ static int machine__process_kernel_mmap_event(struct machine *machine,
break;
}
- pthread_rwlock_unlock(&machine->dsos.lock);
+ up_read(&machine->dsos.lock);
if (kernel == NULL)
kernel = machine__findnew_dso(machine, kmmap_prefix);
@@ -1513,7 +1514,7 @@ static void __machine__remove_thread(struct machine *machine, struct thread *th,
BUG_ON(refcount_read(&th->refcnt) == 0);
if (lock)
- pthread_rwlock_wrlock(&threads->lock);
+ down_write(&threads->lock);
rb_erase_init(&th->rb_node, &threads->entries);
RB_CLEAR_NODE(&th->rb_node);
--threads->nr;
@@ -1524,7 +1525,7 @@ static void __machine__remove_thread(struct machine *machine, struct thread *th,
*/
list_add_tail(&th->node, &threads->dead);
if (lock)
- pthread_rwlock_unlock(&threads->lock);
+ up_write(&threads->lock);
thread__put(th);
}
diff --git a/tools/perf/util/machine.h b/tools/perf/util/machine.h
index fe2f058..b1cd516 100644
--- a/tools/perf/util/machine.h
+++ b/tools/perf/util/machine.h
@@ -6,6 +6,7 @@
#include "map.h"
#include "dso.h"
#include "event.h"
+#include "rwsem.h"
struct addr_location;
struct branch_stack;
@@ -28,7 +29,7 @@ struct vdso_info;
struct threads {
struct rb_root entries;
- pthread_rwlock_t lock;
+ struct rw_semaphore lock;
unsigned int nr;
struct list_head dead;
struct thread *last_match;
diff --git a/tools/perf/util/map.c b/tools/perf/util/map.c
index bdaa0a4..5792d7a 100644
--- a/tools/perf/util/map.c
+++ b/tools/perf/util/map.c
@@ -488,7 +488,7 @@ u64 map__objdump_2mem(struct map *map, u64 ip)
static void maps__init(struct maps *maps)
{
maps->entries = RB_ROOT;
- pthread_rwlock_init(&maps->lock, NULL);
+ init_rwsem(&maps->lock);
}
void map_groups__init(struct map_groups *mg, struct machine *machine)
@@ -517,9 +517,9 @@ static void __maps__purge(struct maps *maps)
static void maps__exit(struct maps *maps)
{
- pthread_rwlock_wrlock(&maps->lock);
+ down_write(&maps->lock);
__maps__purge(maps);
- pthread_rwlock_unlock(&maps->lock);
+ up_write(&maps->lock);
}
void map_groups__exit(struct map_groups *mg)
@@ -586,7 +586,7 @@ struct symbol *maps__find_symbol_by_name(struct maps *maps, const char *name,
struct symbol *sym;
struct rb_node *nd;
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
for (nd = rb_first(&maps->entries); nd; nd = rb_next(nd)) {
struct map *pos = rb_entry(nd, struct map, rb_node);
@@ -602,7 +602,7 @@ struct symbol *maps__find_symbol_by_name(struct maps *maps, const char *name,
sym = NULL;
out:
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
return sym;
}
@@ -638,7 +638,7 @@ static size_t maps__fprintf(struct maps *maps, FILE *fp)
size_t printed = 0;
struct rb_node *nd;
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
for (nd = rb_first(&maps->entries); nd; nd = rb_next(nd)) {
struct map *pos = rb_entry(nd, struct map, rb_node);
@@ -650,7 +650,7 @@ static size_t maps__fprintf(struct maps *maps, FILE *fp)
}
}
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
return printed;
}
@@ -682,7 +682,7 @@ static int maps__fixup_overlappings(struct maps *maps, struct map *map, FILE *fp
struct rb_node *next;
int err = 0;
- pthread_rwlock_wrlock(&maps->lock);
+ down_write(&maps->lock);
root = &maps->entries;
next = rb_first(root);
@@ -750,7 +750,7 @@ put_map:
err = 0;
out:
- pthread_rwlock_unlock(&maps->lock);
+ up_write(&maps->lock);
return err;
}
@@ -771,7 +771,7 @@ int map_groups__clone(struct thread *thread,
struct map *map;
struct maps *maps = &parent->maps[type];
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
for (map = maps__first(maps); map; map = map__next(map)) {
struct map *new = map__clone(map);
@@ -788,7 +788,7 @@ int map_groups__clone(struct thread *thread,
err = 0;
out_unlock:
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
return err;
}
@@ -815,9 +815,9 @@ static void __maps__insert(struct maps *maps, struct map *map)
void maps__insert(struct maps *maps, struct map *map)
{
- pthread_rwlock_wrlock(&maps->lock);
+ down_write(&maps->lock);
__maps__insert(maps, map);
- pthread_rwlock_unlock(&maps->lock);
+ up_write(&maps->lock);
}
static void __maps__remove(struct maps *maps, struct map *map)
@@ -828,9 +828,9 @@ static void __maps__remove(struct maps *maps, struct map *map)
void maps__remove(struct maps *maps, struct map *map)
{
- pthread_rwlock_wrlock(&maps->lock);
+ down_write(&maps->lock);
__maps__remove(maps, map);
- pthread_rwlock_unlock(&maps->lock);
+ up_write(&maps->lock);
}
struct map *maps__find(struct maps *maps, u64 ip)
@@ -838,7 +838,7 @@ struct map *maps__find(struct maps *maps, u64 ip)
struct rb_node **p, *parent = NULL;
struct map *m;
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
p = &maps->entries.rb_node;
while (*p != NULL) {
@@ -854,7 +854,7 @@ struct map *maps__find(struct maps *maps, u64 ip)
m = NULL;
out:
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
return m;
}
diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h
index 73aacf7..d5d7442 100644
--- a/tools/perf/util/map.h
+++ b/tools/perf/util/map.h
@@ -9,6 +9,7 @@
#include <stdio.h>
#include <stdbool.h>
#include <linux/types.h>
+#include "rwsem.h"
enum map_type {
MAP__FUNCTION = 0,
@@ -61,7 +62,7 @@ struct kmap {
struct maps {
struct rb_root entries;
- pthread_rwlock_t lock;
+ struct rw_semaphore lock;
};
struct map_groups {
diff --git a/tools/perf/util/rwsem.c b/tools/perf/util/rwsem.c
new file mode 100644
index 0000000..5e52e7b
--- /dev/null
+++ b/tools/perf/util/rwsem.c
@@ -0,0 +1,32 @@
+#include "util.h"
+#include "rwsem.h"
+
+int init_rwsem(struct rw_semaphore *sem)
+{
+ return pthread_rwlock_init(&sem->lock, NULL);
+}
+
+int exit_rwsem(struct rw_semaphore *sem)
+{
+ return pthread_rwlock_destroy(&sem->lock);
+}
+
+int down_read(struct rw_semaphore *sem)
+{
+ return perf_singlethreaded ? 0 : pthread_rwlock_rdlock(&sem->lock);
+}
+
+int up_read(struct rw_semaphore *sem)
+{
+ return perf_singlethreaded ? 0 : pthread_rwlock_unlock(&sem->lock);
+}
+
+int down_write(struct rw_semaphore *sem)
+{
+ return perf_singlethreaded ? 0 : pthread_rwlock_wrlock(&sem->lock);
+}
+
+int up_write(struct rw_semaphore *sem)
+{
+ return perf_singlethreaded ? 0 : pthread_rwlock_unlock(&sem->lock);
+}
diff --git a/tools/perf/util/rwsem.h b/tools/perf/util/rwsem.h
new file mode 100644
index 0000000..94565ad
--- /dev/null
+++ b/tools/perf/util/rwsem.h
@@ -0,0 +1,19 @@
+#ifndef _PERF_RWSEM_H
+#define _PERF_RWSEM_H
+
+#include <pthread.h>
+
+struct rw_semaphore {
+ pthread_rwlock_t lock;
+};
+
+int init_rwsem(struct rw_semaphore *sem);
+int exit_rwsem(struct rw_semaphore *sem);
+
+int down_read(struct rw_semaphore *sem);
+int up_read(struct rw_semaphore *sem);
+
+int down_write(struct rw_semaphore *sem);
+int up_write(struct rw_semaphore *sem);
+
+#endif /* _PERF_RWSEM_H */
diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
index 5909ee4..066e38a 100644
--- a/tools/perf/util/symbol.c
+++ b/tools/perf/util/symbol.c
@@ -226,7 +226,7 @@ void __map_groups__fixup_end(struct map_groups *mg, enum map_type type)
struct maps *maps = &mg->maps[type];
struct map *next, *curr;
- pthread_rwlock_wrlock(&maps->lock);
+ down_write(&maps->lock);
curr = maps__first(maps);
if (curr == NULL)
@@ -246,7 +246,7 @@ void __map_groups__fixup_end(struct map_groups *mg, enum map_type type)
curr->end = ~0ULL;
out_unlock:
- pthread_rwlock_unlock(&maps->lock);
+ up_write(&maps->lock);
}
struct symbol *symbol__new(u64 start, u64 len, u8 binding, const char *name)
@@ -1671,7 +1671,7 @@ struct map *map_groups__find_by_name(struct map_groups *mg,
struct maps *maps = &mg->maps[type];
struct map *map;
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
for (map = maps__first(maps); map; map = map__next(map)) {
if (map->dso && strcmp(map->dso->short_name, name) == 0)
@@ -1681,7 +1681,7 @@ struct map *map_groups__find_by_name(struct map_groups *mg,
map = NULL;
out_unlock:
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
return map;
}
diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c
index aee9a42..c09bdb5 100644
--- a/tools/perf/util/thread.c
+++ b/tools/perf/util/thread.c
@@ -264,7 +264,7 @@ static int __thread__prepare_access(struct thread *thread)
struct maps *maps = &thread->mg->maps[i];
struct map *map;
- pthread_rwlock_rdlock(&maps->lock);
+ down_read(&maps->lock);
for (map = maps__first(maps); map; map = map__next(map)) {
err = unwind__prepare_access(thread, map, &initialized);
@@ -272,7 +272,7 @@ static int __thread__prepare_access(struct thread *thread)
break;
}
- pthread_rwlock_unlock(&maps->lock);
+ up_read(&maps->lock);
}
return err;
diff --git a/tools/perf/util/trace-event-info.c b/tools/perf/util/trace-event-info.c
index e7d60d0..d7f2113 100644
--- a/tools/perf/util/trace-event-info.c
+++ b/tools/perf/util/trace-event-info.c
@@ -28,7 +28,6 @@
#include <sys/types.h>
#include <sys/stat.h>
#include <sys/wait.h>
-#include <pthread.h>
#include <fcntl.h>
#include <unistd.h>
#include <errno.h>
diff --git a/tools/perf/util/trace-event-read.c b/tools/perf/util/trace-event-read.c
index 8a9a677..40b4259 100644
--- a/tools/perf/util/trace-event-read.c
+++ b/tools/perf/util/trace-event-read.c
@@ -27,7 +27,6 @@
#include <sys/stat.h>
#include <sys/wait.h>
#include <sys/mman.h>
-#include <pthread.h>
#include <fcntl.h>
#include <unistd.h>
#include <errno.h>
diff --git a/tools/perf/util/util.c b/tools/perf/util/util.c
index 3c9c397..97e0c8e 100644
--- a/tools/perf/util/util.c
+++ b/tools/perf/util/util.c
@@ -23,6 +23,19 @@
/*
* XXX We need to find a better place for these things...
*/
+
+bool perf_singlethreaded = true;
+
+void perf_set_singlethreaded(void)
+{
+ perf_singlethreaded = true;
+}
+
+void perf_set_multithreaded(void)
+{
+ perf_singlethreaded = false;
+}
+
unsigned int page_size;
int cacheline_size;
diff --git a/tools/perf/util/util.h b/tools/perf/util/util.h
index 03946d5..6c7e6cc 100644
--- a/tools/perf/util/util.h
+++ b/tools/perf/util/util.h
@@ -62,4 +62,9 @@ int sched_getcpu(void);
int setns(int fd, int nstype);
#endif
+extern bool perf_singlethreaded;
+
+void perf_set_singlethreaded(void);
+void perf_set_multithreaded(void);
+
#endif /* GIT_COMPAT_UTIL_H */
diff --git a/tools/perf/util/vdso.c b/tools/perf/util/vdso.c
index d3c39ee..f5f843d 100644
--- a/tools/perf/util/vdso.c
+++ b/tools/perf/util/vdso.c
@@ -319,7 +319,7 @@ struct dso *machine__findnew_vdso(struct machine *machine,
struct vdso_info *vdso_info;
struct dso *dso = NULL;
- pthread_rwlock_wrlock(&machine->dsos.lock);
+ down_write(&machine->dsos.lock);
if (!machine->vdso_info)
machine->vdso_info = vdso_info__new();
@@ -347,7 +347,7 @@ struct dso *machine__findnew_vdso(struct machine *machine,
out_unlock:
dso__get(dso);
- pthread_rwlock_unlock(&machine->dsos.lock);
+ up_write(&machine->dsos.lock);
return dso;
}
^ permalink raw reply related [flat|nested] 3+ messages in thread
end of thread, other threads:[~2017-09-22 16:50 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-03-24 0:28 [PATCH] perf, tools, script: Skip locking overhead in perf script Andi Kleen
2017-04-04 16:17 ` Arnaldo Carvalho de Melo
2017-09-22 16:46 ` [tip:perf/core] perf tools: Provide mutex wrappers for pthreads rwlocks tip-bot for Arnaldo Carvalho de Melo
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).