* [PATCH 0/1] perf util: Do not reuse target->per_thread flag
@ 2018-01-12 23:12 Mathieu Poirier
2018-01-12 23:12 ` [PATCH 1/1] " Mathieu Poirier
0 siblings, 1 reply; 3+ messages in thread
From: Mathieu Poirier @ 2018-01-12 23:12 UTC (permalink / raw)
To: acme
Cc: peterz, mingo, alexander.shishkin, jolsa, namhyung, yao.jin,
linux-kernel
Hi Arnaldo,
Here is another stab at addressing the side effect introduced by commit
("73c0ca1eee3d perf thread_map: Enumerate all threads from /proc").
I have decided to add another variable to struct target, which keep
things simple and doesn't touch the core. Another solution could have
been to make target->per_thread an enumaration with values PER_THREAD
and ALL_THREAD, something that could be switched on in function
thread_map__new_str().
I opted for the first option as it makes the code easier to understand.
Let me know what you prefer or get back to me with a better solution if
you see one.
Thanks,
Mathieu
Mathieu Poirier (1):
perf util: Do not reuse target->per_thread flag
tools/perf/builtin-stat.c | 2 +-
tools/perf/util/evlist.c | 2 +-
tools/perf/util/target.h | 1 +
tools/perf/util/thread_map.c | 4 ++--
tools/perf/util/thread_map.h | 2 +-
5 files changed, 6 insertions(+), 5 deletions(-)
--
2.7.4
^ permalink raw reply [flat|nested] 3+ messages in thread
* [PATCH 1/1] perf util: Do not reuse target->per_thread flag
2018-01-12 23:12 [PATCH 0/1] perf util: Do not reuse target->per_thread flag Mathieu Poirier
@ 2018-01-12 23:12 ` Mathieu Poirier
2018-01-22 13:05 ` Jin, Yao
0 siblings, 1 reply; 3+ messages in thread
From: Mathieu Poirier @ 2018-01-12 23:12 UTC (permalink / raw)
To: acme
Cc: peterz, mingo, alexander.shishkin, jolsa, namhyung, yao.jin,
linux-kernel
Commit ("73c0ca1eee3d perf thread_map: Enumerate all threads from /proc")
is using the target->per_thread flag to specify that all threads in a
system should be taken into account. That is then used in function
thread_map__new_str() where all threads are added to evlist->threads.
Variable target->per_thread is also used by 'perf record' when handling
trace sessions using the --per-thread command line option. Since the
target->per_thread flag is set all threads will be added to
evlist->threads, which has the effect of creating a kernel event for each
thread in the system.
This patch address the issue by creating a new target->all_threads flag
that gets set from the 'stat' utility, avoiding any conflict with other
utilities using target->per_thread.
Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
---
tools/perf/builtin-stat.c | 2 +-
tools/perf/util/evlist.c | 2 +-
tools/perf/util/target.h | 1 +
tools/perf/util/thread_map.c | 4 ++--
tools/perf/util/thread_map.h | 2 +-
5 files changed, 6 insertions(+), 5 deletions(-)
diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
index 98bf9d32f222..87e156bfb45e 100644
--- a/tools/perf/builtin-stat.c
+++ b/tools/perf/builtin-stat.c
@@ -2831,7 +2831,7 @@ int cmd_stat(int argc, const char **argv)
target__validate(&target);
if ((stat_config.aggr_mode == AGGR_THREAD) && (target.system_wide))
- target.per_thread = true;
+ target.per_thread = target.all_threads = true;
if (perf_evlist__create_maps(evsel_list, &target) < 0) {
if (target__has_task(&target)) {
diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
index f0a5e09c4071..c239eb895612 100644
--- a/tools/perf/util/evlist.c
+++ b/tools/perf/util/evlist.c
@@ -1106,7 +1106,7 @@ int perf_evlist__create_maps(struct perf_evlist *evlist, struct target *target)
struct thread_map *threads;
threads = thread_map__new_str(target->pid, target->tid, target->uid,
- target->per_thread);
+ target->all_threads);
if (!threads)
return -1;
diff --git a/tools/perf/util/target.h b/tools/perf/util/target.h
index 6ef01a83b24e..0da8ea2b6801 100644
--- a/tools/perf/util/target.h
+++ b/tools/perf/util/target.h
@@ -15,6 +15,7 @@ struct target {
bool uses_mmap;
bool default_per_cpu;
bool per_thread;
+ bool all_threads;
};
enum target_errno {
diff --git a/tools/perf/util/thread_map.c b/tools/perf/util/thread_map.c
index 3e1038f6491c..729dad8f412d 100644
--- a/tools/perf/util/thread_map.c
+++ b/tools/perf/util/thread_map.c
@@ -323,7 +323,7 @@ struct thread_map *thread_map__new_by_tid_str(const char *tid_str)
}
struct thread_map *thread_map__new_str(const char *pid, const char *tid,
- uid_t uid, bool per_thread)
+ uid_t uid, bool all_threads)
{
if (pid)
return thread_map__new_by_pid_str(pid);
@@ -331,7 +331,7 @@ struct thread_map *thread_map__new_str(const char *pid, const char *tid,
if (!tid && uid != UINT_MAX)
return thread_map__new_by_uid(uid);
- if (per_thread)
+ if (all_threads)
return thread_map__new_all_cpus();
return thread_map__new_by_tid_str(tid);
diff --git a/tools/perf/util/thread_map.h b/tools/perf/util/thread_map.h
index 0a806b99e73c..5ec91cfd1869 100644
--- a/tools/perf/util/thread_map.h
+++ b/tools/perf/util/thread_map.h
@@ -31,7 +31,7 @@ struct thread_map *thread_map__get(struct thread_map *map);
void thread_map__put(struct thread_map *map);
struct thread_map *thread_map__new_str(const char *pid,
- const char *tid, uid_t uid, bool per_thread);
+ const char *tid, uid_t uid, bool all_threads);
struct thread_map *thread_map__new_by_tid_str(const char *tid_str);
--
2.7.4
^ permalink raw reply related [flat|nested] 3+ messages in thread
* Re: [PATCH 1/1] perf util: Do not reuse target->per_thread flag
2018-01-12 23:12 ` [PATCH 1/1] " Mathieu Poirier
@ 2018-01-22 13:05 ` Jin, Yao
0 siblings, 0 replies; 3+ messages in thread
From: Jin, Yao @ 2018-01-22 13:05 UTC (permalink / raw)
To: Mathieu Poirier, acme
Cc: peterz, mingo, alexander.shishkin, jolsa, namhyung, linux-kernel
Hi Mathieu,
Sorry, I just see this patch today.
I have tested this patch and it works.
Another idea from me is it doesn't need to add a new target->all_threads
flag.
We just use target->per-thread && target->system_wide as a condition to
check for all per-thread case.
I just think for your perf record case, the target->system_wide will not
be set. Instead, if target->per-thread and target->system_wide are both
set, that means we needs to trace on all threads, right?
Thanks
Jin Yao
On 1/13/2018 7:12 AM, Mathieu Poirier wrote:
> Commit ("73c0ca1eee3d perf thread_map: Enumerate all threads from /proc")
> is using the target->per_thread flag to specify that all threads in a
> system should be taken into account. That is then used in function
> thread_map__new_str() where all threads are added to evlist->threads.
>
> Variable target->per_thread is also used by 'perf record' when handling
> trace sessions using the --per-thread command line option. Since the
> target->per_thread flag is set all threads will be added to
> evlist->threads, which has the effect of creating a kernel event for each
> thread in the system.
>
> This patch address the issue by creating a new target->all_threads flag
> that gets set from the 'stat' utility, avoiding any conflict with other
> utilities using target->per_thread.
>
> Signed-off-by: Mathieu Poirier <mathieu.poirier@linaro.org>
> ---
> tools/perf/builtin-stat.c | 2 +-
> tools/perf/util/evlist.c | 2 +-
> tools/perf/util/target.h | 1 +
> tools/perf/util/thread_map.c | 4 ++--
> tools/perf/util/thread_map.h | 2 +-
> 5 files changed, 6 insertions(+), 5 deletions(-)
>
> diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c
> index 98bf9d32f222..87e156bfb45e 100644
> --- a/tools/perf/builtin-stat.c
> +++ b/tools/perf/builtin-stat.c
> @@ -2831,7 +2831,7 @@ int cmd_stat(int argc, const char **argv)
> target__validate(&target);
>
> if ((stat_config.aggr_mode == AGGR_THREAD) && (target.system_wide))
> - target.per_thread = true;
> + target.per_thread = target.all_threads = true;
>
> if (perf_evlist__create_maps(evsel_list, &target) < 0) {
> if (target__has_task(&target)) {
> diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c
> index f0a5e09c4071..c239eb895612 100644
> --- a/tools/perf/util/evlist.c
> +++ b/tools/perf/util/evlist.c
> @@ -1106,7 +1106,7 @@ int perf_evlist__create_maps(struct perf_evlist *evlist, struct target *target)
> struct thread_map *threads;
>
> threads = thread_map__new_str(target->pid, target->tid, target->uid,
> - target->per_thread);
> + target->all_threads);
>
> if (!threads)
> return -1;
> diff --git a/tools/perf/util/target.h b/tools/perf/util/target.h
> index 6ef01a83b24e..0da8ea2b6801 100644
> --- a/tools/perf/util/target.h
> +++ b/tools/perf/util/target.h
> @@ -15,6 +15,7 @@ struct target {
> bool uses_mmap;
> bool default_per_cpu;
> bool per_thread;
> + bool all_threads;
> };
>
> enum target_errno {
> diff --git a/tools/perf/util/thread_map.c b/tools/perf/util/thread_map.c
> index 3e1038f6491c..729dad8f412d 100644
> --- a/tools/perf/util/thread_map.c
> +++ b/tools/perf/util/thread_map.c
> @@ -323,7 +323,7 @@ struct thread_map *thread_map__new_by_tid_str(const char *tid_str)
> }
>
> struct thread_map *thread_map__new_str(const char *pid, const char *tid,
> - uid_t uid, bool per_thread)
> + uid_t uid, bool all_threads)
> {
> if (pid)
> return thread_map__new_by_pid_str(pid);
> @@ -331,7 +331,7 @@ struct thread_map *thread_map__new_str(const char *pid, const char *tid,
> if (!tid && uid != UINT_MAX)
> return thread_map__new_by_uid(uid);
>
> - if (per_thread)
> + if (all_threads)
> return thread_map__new_all_cpus();
>
> return thread_map__new_by_tid_str(tid);
> diff --git a/tools/perf/util/thread_map.h b/tools/perf/util/thread_map.h
> index 0a806b99e73c..5ec91cfd1869 100644
> --- a/tools/perf/util/thread_map.h
> +++ b/tools/perf/util/thread_map.h
> @@ -31,7 +31,7 @@ struct thread_map *thread_map__get(struct thread_map *map);
> void thread_map__put(struct thread_map *map);
>
> struct thread_map *thread_map__new_str(const char *pid,
> - const char *tid, uid_t uid, bool per_thread);
> + const char *tid, uid_t uid, bool all_threads);
>
> struct thread_map *thread_map__new_by_tid_str(const char *tid_str);
>
>
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2018-01-22 13:05 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-01-12 23:12 [PATCH 0/1] perf util: Do not reuse target->per_thread flag Mathieu Poirier
2018-01-12 23:12 ` [PATCH 1/1] " Mathieu Poirier
2018-01-22 13:05 ` Jin, Yao
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.