Implement --switch-output-event

tools/perf/util/evlist.c +1 -7

-int perf_evlist__add_sb_event(struct evlist **evlist,
+int perf_evlist__add_sb_event(struct evlist *evlist,
 			      struct perf_event_attr *attr,
 			      perf_evsel__sb_cb_t cb,
 			      void *data)
 {
 	struct evsel *evsel;
-	bool new_evlist = (*evlist) == NULL;
-
-	if (*evlist == NULL)
-		*evlist = evlist__new();
-	if (*evlist == NULL)
-		return -1;

tools/perf/util/evlist.c +1 -6

 	evsel->side_band.cb = cb;
 	evsel->side_band.data = data;
-	evlist__add(*evlist, evsel);
+	evlist__add(evlist, evsel);
 	return 0;
-
 out_err:
-	if (new_evlist) {
-		evlist__delete(*evlist);
-		*evlist = NULL;
-	}
 	return -1;
 }

tools/perf/util/evlist.h +1 -1

index f5bd5c386df1..0f02408fff3e 100644
@@ -107,7 +107,7 @@ int __perf_evlist__add_default_attrs(struct evlist *evlist,
 
 int perf_evlist__add_dummy(struct evlist *evlist);
 
-int perf_evlist__add_sb_event(struct evlist **evlist,
+int perf_evlist__add_sb_event(struct evlist *evlist,
 			      struct perf_event_attr *attr,
 			      perf_evsel__sb_cb_t cb,
 			      void *data);

Signature · 1 line

2.21.1

SNIP

Quoted 19 lines

tools/perf/util/evlist.c +1 -7

-int perf_evlist__add_sb_event(struct evlist **evlist,
+int perf_evlist__add_sb_event(struct evlist *evlist,
 			      struct perf_event_attr *attr,
 			      perf_evsel__sb_cb_t cb,
 			      void *data)
 {
 	struct evsel *evsel;
-	bool new_evlist = (*evlist) == NULL;
-
-	if (*evlist == NULL)
-		*evlist = evlist__new();
-	if (*evlist == NULL)
-		return -1;
 
 	if (!attr->sample_id_all) {
 		pr_warning("enabling sample_id_all for all side band events\n");
 		attr->sample_id_all = 1;
 	}

tools/perf/util/evlist.c struct evsel *perf_evlist__reset_weak_group(struct evlist *evsel_list, · line 1721

Quoted 4 lines

tools/perf/util/evlist.c +1 -1

-	evsel = perf_evsel__new_idx(attr, (*evlist)->core.nr_entries);
+	evsel = perf_evsel__new_idx(attr, evlist->core.nr_entries);
 	if (!evsel)
 		goto out_err;

we can return -1 right here

jirka

Quoted 31 lines

tools/perf/util/evlist.c +1 -6

 	evsel->side_band.cb = cb;
 	evsel->side_band.data = data;
-	evlist__add(*evlist, evsel);
+	evlist__add(evlist, evsel);
 	return 0;
-
 out_err:
-	if (new_evlist) {
-		evlist__delete(*evlist);
-		*evlist = NULL;
-	}
 	return -1;
 }

tools/perf/util/evlist.h +1 -1

index f5bd5c386df1..0f02408fff3e 100644
@@ -107,7 +107,7 @@ int __perf_evlist__add_default_attrs(struct evlist *evlist,
 
 int perf_evlist__add_dummy(struct evlist *evlist);
 
-int perf_evlist__add_sb_event(struct evlist **evlist,
+int perf_evlist__add_sb_event(struct evlist *evlist,
 			      struct perf_event_attr *attr,
 			      perf_evsel__sb_cb_t cb,
 			      void *data);

Signature · 1 line

2.21.1

From: Arnaldo Carvalho de Melo <[email protected]>

I.e. so far we had just one event in that side band thread, a dummy one with attr.bpf_event set, so that 'perf record' can go ahead and ask the kernel for further information about BPF programs being loaded.

Allow for more than one event to be there, so that we can use it as well for the upcoming --switch-output-event feature.

Cc Adrian Hunter <[email protected]>

Cc Jiri Olsa <[email protected]>

Cc Namhyung Kim <[email protected]>

Cc Song Liu <[email protected]>

Link http://lore.kernel.org/lkml/[email protected] open here

Signed-off-by Arnaldo Carvalho de Melo <[email protected]>

---
 tools/perf/util/evlist.c | 22 ++++++++++++++++++++++
 tools/perf/util/evlist.h |  1 +
 2 files changed, 23 insertions(+)

tools/perf/util/evlist.c +13 -0

index 1d0d36da223b..849058766757 100644
@@ -1777,6 +1777,19 @@ static void *perf_evlist__poll_thread(void *arg)
 	return NULL;
 }
 
+void evlist__set_cb(struct evlist *evlist, perf_evsel__sb_cb_t cb, void *data)
+{
+	struct evsel *evsel;
+
+	evlist__for_each_entry(evlist, evsel) {
+		evsel->core.attr.sample_id_all	  = 1;
+		evsel->core.attr.watermark	  = 1;
+		evsel->core.attr.wakeup_watermark = 1;
+		evsel->side_band.cb   = cb;
+		evsel->side_band.data = data;
+	}
+}
+
 int perf_evlist__start_sb_thread(struct evlist *evlist,
 				 struct target *target)
 {

Quoted 3 lines · Arnaldo Carvalho de Melo

From: Arnaldo Carvalho de Melo <[email protected]>

I.e. so far we had just one event in that side band thread, a dummy

one

Quoted 1 line

with attr.bpf_event set, so that 'perf record' can go ahead and ask

the

Quoted 10 lines

kernel for further information about BPF programs being loaded.

Allow for more than one event to be there, so that we can use it as well for the upcoming --switch-output-event feature.

Cc: Adrian Hunter <[email protected]>
Cc: Jiri Olsa <[email protected]>
Cc: Namhyung Kim <[email protected]>
Cc: Song Liu <[email protected]>
Link:

http://lore.kernel.org/lkml/[email protected]

tools/perf/util/evlist.c static void *perf_evlist__poll_thread(void

Quoted 11 lines

Signed-off-by Arnaldo Carvalho de Melo <[email protected]>

---
 tools/perf/util/evlist.c | 22 ++++++++++++++++++++++
 tools/perf/util/evlist.h |  1 +
 2 files changed, 23 insertions(+)

tools/perf/util/evlist.c +0 -0

index 1d0d36da223b..849058766757 100644
@@ -1777,6 +1777,19 @@ static void *perf_evlist__poll_thread(void

*arg)

tools/perf/util/evlist.c static void *perf_evlist__poll_thread(void *arg)

Quoted 4 lines

tools/perf/util/evlist.c +1 -0

 	return NULL;
 }
 
+void evlist__set_cb(struct evlist *evlist, perf_evsel__sb_cb_t cb,

void *data)

tools/perf/util/evlist.c int perf_evlist__start_sb_thread(struct evlist

Quoted 16 lines

tools/perf/util/evlist.c +12 -0

+{
+	struct evsel *evsel;
+
+	evlist__for_each_entry(evlist, evsel) {
+		evsel->core.attr.sample_id_all	  = 1;
+		evsel->core.attr.watermark	  = 1;
+		evsel->core.attr.wakeup_watermark = 1;
+		evsel->side_band.cb   = cb;
+		evsel->side_band.data = data;
+	}
+}
+
 int perf_evlist__start_sb_thread(struct evlist *evlist,
 				 struct target *target)
 {

@@ -1788,6 +1801,15 @@ int perf_evlist__start_sb_thread(struct evlist

*evlist,

tools/perf/util/evlist.c int perf_evlist__start_sb_thread(struct evlist *evlist, · line 1805

Quoted 5 lines

tools/perf/util/evlist.c +2 -0

 	if (perf_evlist__create_maps(evlist, target))
 		goto out_delete_evlist;
 
+	if (evlist->core.nr_entries > 1) {
+		bool can_sample_identifier = perf_can_sample_identifier();

I just found this breaks python, because perf_can_sample_identifier is defined in util/record.c

	19: 'import perf' in python                               :
	--- start ---
	test child forked, pid 1808205
	Traceback (most recent call last):
	  File "<stdin>", line 1, in <module>
	ImportError: python/perf.so: undefined symbol:

perf_can_sample_identifier
test child finished with -1
---- end ----
'import perf' in python: FAILED!

jirka

From: Arnaldo Carvalho de Melo <[email protected]>

Now we can use it with --overwrite to have a flight recorder mode that gets snapshot requests from arbitrary events that are processed in the side band thread together with the PERF_RECORD_BPF_EVENT processing.

Example:

To collect scheduler events until a recvmmsg syscall happens, system wide:

[root@five a]# rm -f perf.data.2020042717*
[root@five a]# perf record --overwrite -e sched:*switch,syscalls:*recvmmsg --switch-output-event syscalls:sys_enter_recvmmsg [ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042717585458 ]
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042717590235 ]
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042717590398 ]
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Dump perf.data.2020042717590511 ]
[ perf record: Captured and wrote 7.244 MB perf.data.<timestamp> ]

So in the above case we had 3 snapshots, the fourth was forced by control+C:

[root@five a]# ls -la
total 20440
drwxr-xr-x. 2 root root 4096 Apr 27 17:59 .
dr-xr-x---. 12 root root 4096 Apr 27 17:46 ..

  -rw-------.  1 root root 3936125 Apr 27 17:58 perf.data.2020042717585458
  -rw-------.  1 root root 5074869 Apr 27 17:59 perf.data.2020042717590235
  -rw-------.  1 root root 4291037 Apr 27 17:59 perf.data.2020042717590398
  -rw-------.  1 root root 7617037 Apr 27 17:59 perf.data.2020042717590511

[root@five a]#

One can make this more precise by adding the switch output event to the main -e events list, as since this is done asynchronously, a few events after the signal event will appear in the snapshots, as can be seen with:

[root@five a]# rm -f perf.data.20200427175*
[root@five a]# perf record --overwrite -e sched:*switch,syscalls:*recvmmsg --switch-output-event syscalls:sys_enter_recvmmsg [ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042718024203 ]
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042718024301 ]
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042718024484 ]
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Dump perf.data.2020042718024562 ]
[ perf record: Captured and wrote 7.337 MB perf.data.<timestamp> ]
[root@five a]# perf script -i perf.data.2020042718024203 | tail -15

       PacerThread 148586 [005] 122.830729: sched:sched_switch: prev_comm=PacerThread prev_pid=148586...
           swapper      0 [000] 122.833588: sched:sched_switch: prev_comm=swapper/0 prev_pid=...
    NetworkManager   1251 [000] 122.833619: syscalls:sys_enter_recvmmsg: fd: 0x0000001c, mmsg: 0x7ffe83054a1...
           swapper      0 [002] 122.833624: sched:sched_switch: prev_comm=swapper/2 prev_pid=...
           swapper      0 [003] 122.833624: sched:sched_switch: prev_comm=swapper/3 prev_pid=...
    NetworkManager   1251 [000] 122.833626: syscalls:sys_exit_recvmmsg: 0x1
   kworker/3:3-eve 158946 [003] 122.833628: sched:sched_switch: prev_comm=kworker/3:3 prev_pid=15894...
           swapper      0 [004] 122.833641: sched:sched_switch: prev_comm=swapper/4 prev_pid=...
    NetworkManager   1251 [000] 122.833642: sched:sched_switch: prev_comm=NetworkManage...
              perf 228273 [002] 122.833645: sched:sched_switch: prev_comm=perf prev_pid=22827...
           swapper      0 [011] 122.833646: sched:sched_switch: prev_comm=swapper/1...
           swapper      0 [002] 122.833648: sched:sched_switch: prev_comm=swapper/...
   kworker/0:2-eve 207387 [000] 122.833648: sched:sched_switch: prev_comm=kworker/0:2 prev_pid=20738...
   kworker/2:3-eve 232038 [002] 122.833652: sched:sched_switch: prev_comm=kworker/2:3 prev_pid=23203...
              perf 235825 [003] 122.833653: sched:sched_switch: prev_comm=perf prev_pid=23582...

[root@five a]#

Cc Adrian Hunter <[email protected]>

Cc Jiri Olsa <[email protected]>

Cc Namhyung Kim <[email protected]>

Cc Song Liu <[email protected]>

Cc Wang Nan <[email protected]>

Link http://lore.kernel.org/lkml/[email protected] open here

Link http://lore.kernel.org/lkml/[email protected] open here

Signed-off-by Arnaldo Carvalho de Melo <[email protected]>

---
 tools/perf/Documentation/perf-record.txt | 13 ++++++++
 tools/perf/builtin-record.c              | 41 +++++++++++++++++++++---
 2 files changed, 50 insertions(+), 4 deletions(-)

tools/perf/Documentation/perf-record.txt +0 -0

index 6e8b4649307c..561ef55743e2 100644

tools/perf/Documentation/perf-record.txt +8 -0

+Uses the same syntax as --event, it will just not be recorded, serving only to
+switch the perf.data file as soon as the --switch-output event is processed by
+a separate sideband thread.
+
+This sideband thread is also used to other purposes, like processing the
+PERF_RECORD_BPF_EVENT records as they happen, asking the kernel for extra BPF
+information, etc.
+
 --switch-max-files=N::
 
 When rotating perf.data with --switch-output, only keep N files.

tools/perf/builtin-record.c +34 -4

index ed2244847400..7a6a89972691 100644
@@ -87,7 +87,9 @@ struct record {
        struct evlist   *evlist;
        struct perf_session     *session;
        struct evlist           *sb_evlist;
+       pthread_t               thread_id;
        int                     realtime_prio;
+       bool                    switch_output_event_set;
        bool                    no_buildid;
        bool                    no_buildid_set;
        bool                    no_buildid_cache;

@@ -1436,6 +1438,13 @@ static int record__synthesize(struct record *rec, bool tail)
        return err;
 }
 
+static int record__process_signal_event(union perf_event *event __maybe_unused, void *data)
+{
+       struct record *rec = data;
+       pthread_kill(rec->thread_id, SIGUSR2);
+       return 0;
+}
+
 static int __cmd_record(struct record *rec, int argc, const char **argv)
 {
        int err;

@@ -1580,12 +1589,24 @@ static int __cmd_record(struct record *rec, int argc, const char **argv)
                goto out_child;
        }
 
-       if (!opts->no_bpf_event) {
-               rec->sb_evlist = evlist__new();
+       if (rec->sb_evlist != NULL) {
+               /*
+                * We get here if --switch-output-event populated the
+                * sb_evlist, so associate a callback that will send a SIGUSR2
+                * to the main thread.
+                */
+               evlist__set_cb(rec->sb_evlist, record__process_signal_event, rec);
+               rec->thread_id = pthread_self();
+       }
 
+       if (!opts->no_bpf_event) {
                if (rec->sb_evlist == NULL) {
-                       pr_err("Couldn't create side band evlist.\n.");
-                       goto out_child;
+                       rec->sb_evlist = evlist__new();
+
+                       if (rec->sb_evlist == NULL) {
+                               pr_err("Couldn't create side band evlist.\n.");
+                               goto out_child;
+                       }
                }
 
                if (evlist__add_bpf_sb_event(rec->sb_evlist, &session->header.env)) {

@@ -2179,10 +2200,19 @@ static int switch_output_setup(struct record *rec)
        };
        unsigned long val;
 
+       /*
+        * If we're using --switch-output-events, then we imply its
+        * --switch-output=signal, as we'll send a SIGUSR2 from the side band
+        *  thread to its parent.
+        */
+       if (rec->switch_output_event_set)
+               goto do_signal;
+
        if (!s->set)
                return 0;
 
        if (!strcmp(s->str, "signal")) {
+do_signal:
                s->signal = true;
                pr_debug("switch-output with SIGUSR2 signal\n");
                goto enabled;

tools/perf/builtin-record.c +1 -1

+                        parse_events_option_new_evlist),
        OPT_INTEGER(0, "switch-max-files", &record.switch_output.num_files,
                   "Limit number of switch output generated files"),
        OPT_BOOLEAN(0, "dry-run", &dry_run,
--

2.21.1

Quoted 92 lines · Arnaldo Carvalho de Melo

From: Arnaldo Carvalho de Melo <[email protected]>

Now we can use it with --overwrite to have a flight recorder mode that gets snapshot requests from arbitrary events that are processed in the side band thread together with the PERF_RECORD_BPF_EVENT processing.

Example:

To collect scheduler events until a recvmmsg syscall happens, system wide:

[root@five a]# rm -f perf.data.2020042717*
[root@five a]# perf record --overwrite -e sched:*switch,syscalls:*recvmmsg --switch-output-event syscalls:sys_enter_recvmmsg [ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042717585458 ]
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042717590235 ]
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042717590398 ]
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Dump perf.data.2020042717590511 ]
[ perf record: Captured and wrote 7.244 MB perf.data.<timestamp> ]

So in the above case we had 3 snapshots, the fourth was forced by control+C:

[root@five a]# ls -la
total 20440
drwxr-xr-x. 2 root root 4096 Apr 27 17:59 .
dr-xr-x---. 12 root root 4096 Apr 27 17:46 ..

  -rw-------.  1 root root 3936125 Apr 27 17:58 perf.data.2020042717585458
  -rw-------.  1 root root 5074869 Apr 27 17:59 perf.data.2020042717590235
  -rw-------.  1 root root 4291037 Apr 27 17:59 perf.data.2020042717590398
  -rw-------.  1 root root 7617037 Apr 27 17:59 perf.data.2020042717590511

[root@five a]#

One can make this more precise by adding the switch output event to the main -e events list, as since this is done asynchronously, a few events after the signal event will appear in the snapshots, as can be seen with:

[root@five a]# rm -f perf.data.20200427175*
[root@five a]# perf record --overwrite -e sched:*switch,syscalls:*recvmmsg --switch-output-event syscalls:sys_enter_recvmmsg [ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042718024203 ]
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042718024301 ]
[ perf record: dump data: Woken up 1 times ]
[ perf record: Dump perf.data.2020042718024484 ]
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Dump perf.data.2020042718024562 ]
[ perf record: Captured and wrote 7.337 MB perf.data.<timestamp> ]
[root@five a]# perf script -i perf.data.2020042718024203 | tail -15

       PacerThread 148586 [005] 122.830729: sched:sched_switch: prev_comm=PacerThread prev_pid=148586...
           swapper      0 [000] 122.833588: sched:sched_switch: prev_comm=swapper/0 prev_pid=...
    NetworkManager   1251 [000] 122.833619: syscalls:sys_enter_recvmmsg: fd: 0x0000001c, mmsg: 0x7ffe83054a1...
           swapper      0 [002] 122.833624: sched:sched_switch: prev_comm=swapper/2 prev_pid=...
           swapper      0 [003] 122.833624: sched:sched_switch: prev_comm=swapper/3 prev_pid=...
    NetworkManager   1251 [000] 122.833626: syscalls:sys_exit_recvmmsg: 0x1
   kworker/3:3-eve 158946 [003] 122.833628: sched:sched_switch: prev_comm=kworker/3:3 prev_pid=15894...
           swapper      0 [004] 122.833641: sched:sched_switch: prev_comm=swapper/4 prev_pid=...
    NetworkManager   1251 [000] 122.833642: sched:sched_switch: prev_comm=NetworkManage...
              perf 228273 [002] 122.833645: sched:sched_switch: prev_comm=perf prev_pid=22827...
           swapper      0 [011] 122.833646: sched:sched_switch: prev_comm=swapper/1...
           swapper      0 [002] 122.833648: sched:sched_switch: prev_comm=swapper/...
   kworker/0:2-eve 207387 [000] 122.833648: sched:sched_switch: prev_comm=kworker/0:2 prev_pid=20738...
   kworker/2:3-eve 232038 [002] 122.833652: sched:sched_switch: prev_comm=kworker/2:3 prev_pid=23203...
              perf 235825 [003] 122.833653: sched:sched_switch: prev_comm=perf prev_pid=23582...

[root@five a]#

Cc Adrian Hunter <[email protected]>

Cc Jiri Olsa <[email protected]>

Cc Namhyung Kim <[email protected]>

Cc Song Liu <[email protected]>

Cc Wang Nan <[email protected]>

Link http://lore.kernel.org/lkml/[email protected] open here

Link http://lore.kernel.org/lkml/[email protected] open here

Signed-off-by Arnaldo Carvalho de Melo <[email protected]>

---
 tools/perf/Documentation/perf-record.txt | 13 ++++++++
 tools/perf/builtin-record.c              | 41 +++++++++++++++++++++---
 2 files changed, 50 insertions(+), 4 deletions(-)

tools/perf/Documentation/perf-record.txt +0 -0

index 6e8b4649307c..561ef55743e2 100644
@@ -556,6 +556,19 @@ overhead. You can still switch them on with:
 
   --switch-output --no-no-buildid  --no-no-buildid-cache
 

tools/perf/Documentation/perf-record.txt overhead. You can still switch them on with: · line 562

Quoted 4 lines

tools/perf/Documentation/perf-record.txt +4 -0

+--switch-output-event::
+Events that will cause the switch of the perf.data file, auto-selecting
+--switch-output=signal, the results are similar as internally the side band
+thread will also send a SIGUSR2 to the main one.

I found this paragraph a little hard, perhaps:
A list of events that when they occur cause the output perf.data file to be ended and a new one created. The signal event, --switch-output=signal, is auto-selected as SIGUSR2 is used internally by the thread monitoring the events.

Quoted 89 lines

tools/perf/Documentation/perf-record.txt +8 -0

+Uses the same syntax as --event, it will just not be recorded, serving only to
+switch the perf.data file as soon as the --switch-output event is processed by
+a separate sideband thread.
+
+This sideband thread is also used to other purposes, like processing the
+PERF_RECORD_BPF_EVENT records as they happen, asking the kernel for extra BPF
+information, etc.
+
 --switch-max-files=N::
 
 When rotating perf.data with --switch-output, only keep N files.

tools/perf/builtin-record.c +34 -4

index ed2244847400..7a6a89972691 100644
@@ -87,7 +87,9 @@ struct record {
        struct evlist   *evlist;
        struct perf_session     *session;
        struct evlist           *sb_evlist;
+       pthread_t               thread_id;
        int                     realtime_prio;
+       bool                    switch_output_event_set;
        bool                    no_buildid;
        bool                    no_buildid_set;
        bool                    no_buildid_cache;

@@ -1436,6 +1438,13 @@ static int record__synthesize(struct record *rec, bool tail)
        return err;
 }
 
+static int record__process_signal_event(union perf_event *event __maybe_unused, void *data)
+{
+       struct record *rec = data;
+       pthread_kill(rec->thread_id, SIGUSR2);
+       return 0;
+}
+
 static int __cmd_record(struct record *rec, int argc, const char **argv)
 {
        int err;

@@ -1580,12 +1589,24 @@ static int __cmd_record(struct record *rec, int argc, const char **argv)
                goto out_child;
        }
 
-       if (!opts->no_bpf_event) {
-               rec->sb_evlist = evlist__new();
+       if (rec->sb_evlist != NULL) {
+               /*
+                * We get here if --switch-output-event populated the
+                * sb_evlist, so associate a callback that will send a SIGUSR2
+                * to the main thread.
+                */
+               evlist__set_cb(rec->sb_evlist, record__process_signal_event, rec);
+               rec->thread_id = pthread_self();
+       }
 
+       if (!opts->no_bpf_event) {
                if (rec->sb_evlist == NULL) {
-                       pr_err("Couldn't create side band evlist.\n.");
-                       goto out_child;
+                       rec->sb_evlist = evlist__new();
+
+                       if (rec->sb_evlist == NULL) {
+                               pr_err("Couldn't create side band evlist.\n.");
+                               goto out_child;
+                       }
                }
 
                if (evlist__add_bpf_sb_event(rec->sb_evlist, &session->header.env)) {

@@ -2179,10 +2200,19 @@ static int switch_output_setup(struct record *rec)
        };
        unsigned long val;
 
+       /*
+        * If we're using --switch-output-events, then we imply its
+        * --switch-output=signal, as we'll send a SIGUSR2 from the side band
+        *  thread to its parent.
+        */
+       if (rec->switch_output_event_set)
+               goto do_signal;
+
        if (!s->set)
                return 0;
 
        if (!strcmp(s->str, "signal")) {
+do_signal:
                s->signal = true;
                pr_debug("switch-output with SIGUSR2 signal\n");
                goto enabled;

tools/perf/builtin-record.c static struct option __record_options[] = { · line 2474

Quoted 5 lines

tools/perf/builtin-record.c +2 -0

@@ -2440,6 +2470,9 @@ static struct option __record_options[] = {
                          &record.switch_output.set, "signal or size[BKMG] or time[smhd]",
                          "Switch output when receiving SIGUSR2 (signal) or cross a size or time threshold",
                          "signal"),
+       OPT_CALLBACK_SET(0, "switch-output-event", &record.sb_evlist, &record.switch_output_event_set, "switch output event",
+                        "switch output event selector. use 'perf list' to list available events",

Perhaps:
"A list of events, see 'perf list', that when they occur cause the end of one perf.data file and the creation of another"

Thanks,
Ian

Quoted 7 lines

tools/perf/builtin-record.c +1 -1

+                        parse_events_option_new_evlist),
        OPT_INTEGER(0, "switch-max-files", &record.switch_output.num_files,
                   "Limit number of switch output generated files"),
        OPT_BOOLEAN(0, "dry-run", &dry_run,
--

2.21.1

Hi guys,

Please consider reviewing, this addresses comments by Jiri in
the V1.

Again, the example provided is too simple, using 'perf probe' to
put probes in specific places in some workload to then get any other event close to the time the trigger hits comes to mind as well, using the signal was just to reuse the pre-existing logic and keep the patchkit small.

One other thing that occurred to me while testing is that this
can be combined with 'perf report/perf script' --switch-off option:

  $ perf report -h --switch-off

Usage: perf report [<options>]

--switch-off <event>
Stop considering events after the ocurrence of this event

$

To remove from consideration the events that end up being
recorded in the ring buffer after the --switch-output-event but gets in the ring buffer because we process the --switch-output-event asynchronously.

Its available at:

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git perf/switch-output-event

Best regards,

- Arnaldo

Arnaldo Carvalho de Melo (8):
perf record: Move sb_evlist to 'struct record'
perf top: Move sb_evlist to 'struct perf_top'
perf bpf: Decouple creating the evlist from adding the SB event perf parse-events: Add parse_events_option() variant that creates evlist perf evlist: Allow reusing the side band thread for more purposes
libsubcmd: Introduce OPT_CALLBACK_SET()
perf record: Introduce --switch-output-event
perf record: Move side band evlist setup to separate routine

 tools/lib/subcmd/parse-options.h         |  2 +
 tools/perf/Documentation/perf-record.txt | 13 ++++
 tools/perf/builtin-record.c              | 75 ++++++++++++++++++++----
 tools/perf/builtin-top.c                 | 20 +++++--
 tools/perf/util/bpf-event.c              |  3 +-
 tools/perf/util/bpf-event.h              |  7 +--
 tools/perf/util/evlist.c                 | 39 +++++++-----
 tools/perf/util/evlist.h                 |  3 +-
 tools/perf/util/parse-events.c           | 23 ++++++++
 tools/perf/util/parse-events.h           |  1 +
 tools/perf/util/top.h                    |  2 +-
 11 files changed, 151 insertions(+), 37 deletions(-)

Signature · 1 line

2.21.1