[PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[Wang Nan]

After converting perf.data to CTF, we lost pid-tid-comm mapping. Which
makes inconvience. For example, in perf script output we know which
process issue an event like this:

 compiz 19361 [001] 3275709.313929:  raw_syscalls:sys_exit: NR 7 = 0

After converting to CTF, we only get this:

  [3275709.313929985] (+0.110646118) raw_syscalls:sys_exit: { cpu_id = 1 }, { perf_ip = 0xFFFFFFFF8107B2E8, perf_tid = 19361, perf_pid = 19361, perf_id = 18920, perf_period = 1, common_type = 16, common_flags = 0, common_preempt_count = 1, common_pid = 19361, id = 7, ret = 0 }

Currently, if we want to find the name and parent of a process, we
have to collect 'sched:sched_switch' event.

This patch set adds a '--all' option to 'perf convert', converts comm,
fork and exit events to CTF output. CTF user now can track the mapping
by their own.

v1 -> v2: Report number of sample and non-sample events when finish.
          rename opts.non_sample to opts.all.

Wang Nan (7):
  perf ctf: Add value_set_string() helper
  perf ctf: Pass convert options through opts structure
  perf ctf: Add 'all' option
  perf ctf: Prepare collect non-sample events
  perf ctf: Generate comm event to CTF output
  perf ctf: Add '--all' option for 'perf data convert'
  perf ctf: Generate fork and exit events to CTF output

 tools/perf/Documentation/perf-data.txt |   4 +
 tools/perf/builtin-data.c              |  11 +-
 tools/perf/util/data-convert-bt.c      | 195 ++++++++++++++++++++++++++++++++-
 tools/perf/util/data-convert-bt.h      |   4 +-
 tools/perf/util/data-convert.h         |   9 ++
 5 files changed, 216 insertions(+), 7 deletions(-)
 create mode 100644 tools/perf/util/data-convert.h

-- 
1.8.3.4

[PATCH v2 1/7] perf ctf: Add value_set_string() helper

[Wang Nan]

There are many value_set_##x helper for integer, but only for integer.
This patch adds value_set_string() helper to help following commits
create string fields.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
---
 tools/perf/util/data-convert-bt.c | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index 4b59879..7908278 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -140,6 +140,36 @@ FUNC_VALUE_SET(s64)
 FUNC_VALUE_SET(u64)
 __FUNC_VALUE_SET(u64_hex, u64)
 
+static int string_set_value(struct bt_ctf_field *field, const char *string);
+static __maybe_unused int
+value_set_string(struct ctf_writer *cw, struct bt_ctf_event *event,
+		 const char *name, const char *string)
+{
+	struct bt_ctf_field_type *type = cw->data.string;
+	struct bt_ctf_field *field;
+	int ret = 0;
+
+	field = bt_ctf_field_create(type);
+	if (!field) {
+		pr_err("failed to create a field %s\n", name);
+		return -1;
+	}
+
+	ret = string_set_value(field, string);
+	if (ret) {
+		pr_err("failed to set value %s\n", name);
+		goto err_put_field;
+	}
+
+	ret = bt_ctf_event_set_payload(event, name, field);
+	if (ret)
+		pr_err("failed to set payload %s\n", name);
+
+err_put_field:
+	bt_ctf_field_put(field);
+	return ret;
+}
+
 static struct bt_ctf_field_type*
 get_tracepoint_field_type(struct ctf_writer *cw, struct format_field *field)
 {
-- 
1.8.3.4

[tip:perf/core] perf data ctf: Add value_set_string() helper

[tip-bot for Wang Nan]

Commit-ID:  069ee5c488d161f539bb897b1bc64b83f9773221
Gitweb:     http://git.kernel.org/tip/069ee5c488d161f539bb897b1bc64b83f9773221
Author:     Wang Nan <wangnan0@huawei.com>
AuthorDate: Fri, 24 Jun 2016 11:22:06 +0000
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Tue, 28 Jun 2016 10:54:55 -0300

perf data ctf: Add value_set_string() helper

There are many value_set_##x helper for integer, but only for integer.
This patch adds value_set_string() helper to help following commits
create string fields.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Zefan Li <lizefan@huawei.com>
Cc: pi3orama@163.com
Link: http://lkml.kernel.org/r/1466767332-114472-2-git-send-email-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/data-convert-bt.c | 30 ++++++++++++++++++++++++++++++
 1 file changed, 30 insertions(+)

diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index 7b1bc24..4b68e7b 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -141,6 +141,36 @@ FUNC_VALUE_SET(s64)
 FUNC_VALUE_SET(u64)
 __FUNC_VALUE_SET(u64_hex, u64)
 
+static int string_set_value(struct bt_ctf_field *field, const char *string);
+static __maybe_unused int
+value_set_string(struct ctf_writer *cw, struct bt_ctf_event *event,
+		 const char *name, const char *string)
+{
+	struct bt_ctf_field_type *type = cw->data.string;
+	struct bt_ctf_field *field;
+	int ret = 0;
+
+	field = bt_ctf_field_create(type);
+	if (!field) {
+		pr_err("failed to create a field %s\n", name);
+		return -1;
+	}
+
+	ret = string_set_value(field, string);
+	if (ret) {
+		pr_err("failed to set value %s\n", name);
+		goto err_put_field;
+	}
+
+	ret = bt_ctf_event_set_payload(event, name, field);
+	if (ret)
+		pr_err("failed to set payload %s\n", name);
+
+err_put_field:
+	bt_ctf_field_put(field);
+	return ret;
+}
+
 static struct bt_ctf_field_type*
 get_tracepoint_field_type(struct ctf_writer *cw, struct format_field *field)
 {

[PATCH v2 2/7] perf ctf: Pass convert options through opts structure

[Wang Nan]

Following commits will add new option to 'perf data convert'. All options
should be grouped into a structure and passed to low level converter
(currently there's only one converter).

Introduce data-convert.h and define 'struct perf_data_convert_opts' in
it. Pass 'force' through opts.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
---
 tools/perf/builtin-data.c         | 9 ++++++---
 tools/perf/util/data-convert-bt.c | 5 +++--
 tools/perf/util/data-convert-bt.h | 4 +++-
 tools/perf/util/data-convert.h    | 8 ++++++++
 4 files changed, 20 insertions(+), 6 deletions(-)
 create mode 100644 tools/perf/util/data-convert.h

diff --git a/tools/perf/builtin-data.c b/tools/perf/builtin-data.c
index b97bc15..38111a9 100644
--- a/tools/perf/builtin-data.c
+++ b/tools/perf/builtin-data.c
@@ -3,6 +3,7 @@
 #include "perf.h"
 #include "debug.h"
 #include <subcmd/parse-options.h>
+#include "data-convert.h"
 #include "data-convert-bt.h"
 
 typedef int (*data_cmd_fn_t)(int argc, const char **argv, const char *prefix);
@@ -53,14 +54,16 @@ static int cmd_data_convert(int argc, const char **argv,
 			    const char *prefix __maybe_unused)
 {
 	const char *to_ctf     = NULL;
-	bool force = false;
+	struct perf_data_convert_opts opts = {
+		.force = false,
+	};
 	const struct option options[] = {
 		OPT_INCR('v', "verbose", &verbose, "be more verbose"),
 		OPT_STRING('i', "input", &input_name, "file", "input file name"),
 #ifdef HAVE_LIBBABELTRACE_SUPPORT
 		OPT_STRING(0, "to-ctf", &to_ctf, NULL, "Convert to CTF format"),
 #endif
-		OPT_BOOLEAN('f', "force", &force, "don't complain, do it"),
+		OPT_BOOLEAN('f', "force", &opts.force, "don't complain, do it"),
 		OPT_END()
 	};
 
@@ -78,7 +81,7 @@ static int cmd_data_convert(int argc, const char **argv,
 
 	if (to_ctf) {
 #ifdef HAVE_LIBBABELTRACE_SUPPORT
-		return bt_convert__perf2ctf(input_name, to_ctf, force);
+		return bt_convert__perf2ctf(input_name, to_ctf, &opts);
 #else
 		pr_err("The libbabeltrace support is not compiled in.\n");
 		return -1;
diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index 7908278..a5e0f68 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -1303,13 +1303,14 @@ static int convert__config(const char *var, const char *value, void *cb)
 	return 0;
 }
 
-int bt_convert__perf2ctf(const char *input, const char *path, bool force)
+int bt_convert__perf2ctf(const char *input, const char *path,
+			 struct perf_data_convert_opts *opts)
 {
 	struct perf_session *session;
 	struct perf_data_file file = {
 		.path = input,
 		.mode = PERF_DATA_MODE_READ,
-		.force = force,
+		.force = opts->force,
 	};
 	struct convert c = {
 		.tool = {
diff --git a/tools/perf/util/data-convert-bt.h b/tools/perf/util/data-convert-bt.h
index 4c20434..9a3b587 100644
--- a/tools/perf/util/data-convert-bt.h
+++ b/tools/perf/util/data-convert-bt.h
@@ -1,8 +1,10 @@
 #ifndef __DATA_CONVERT_BT_H
 #define __DATA_CONVERT_BT_H
+#include "data-convert.h"
 #ifdef HAVE_LIBBABELTRACE_SUPPORT
 
-int bt_convert__perf2ctf(const char *input_name, const char *to_ctf, bool force);
+int bt_convert__perf2ctf(const char *input_name, const char *to_ctf,
+			 struct perf_data_convert_opts *opts);
 
 #endif /* HAVE_LIBBABELTRACE_SUPPORT */
 #endif /* __DATA_CONVERT_BT_H */
diff --git a/tools/perf/util/data-convert.h b/tools/perf/util/data-convert.h
new file mode 100644
index 0000000..97cfd36
--- /dev/null
+++ b/tools/perf/util/data-convert.h
@@ -0,0 +1,8 @@
+#ifndef __DATA_CONVERT_H
+#define __DATA_CONVERT_H
+
+struct perf_data_convert_opts {
+	bool force;
+};
+
+#endif /* __DATA_CONVERT_H */
-- 
1.8.3.4

[tip:perf/core] perf data ctf: Pass convert options through opts structure

[tip-bot for Wang Nan]

Commit-ID:  3275f68e50290acd04612c6af41173fe83fdf4b0
Gitweb:     http://git.kernel.org/tip/3275f68e50290acd04612c6af41173fe83fdf4b0
Author:     Wang Nan <wangnan0@huawei.com>
AuthorDate: Fri, 24 Jun 2016 11:22:07 +0000
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Tue, 28 Jun 2016 10:54:55 -0300

perf data ctf: Pass convert options through opts structure

Following commits will add new option to 'perf data convert'. All options
should be grouped into a structure and passed to low level converter
(currently there's only one converter).

Introduce data-convert.h and define 'struct perf_data_convert_opts' in
it. Pass 'force' through opts.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Zefan Li <lizefan@huawei.com>
Cc: pi3orama@163.com
Link: http://lkml.kernel.org/r/1466767332-114472-3-git-send-email-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/builtin-data.c         | 9 ++++++---
 tools/perf/util/data-convert-bt.c | 5 +++--
 tools/perf/util/data-convert-bt.h | 4 +++-
 tools/perf/util/data-convert.h    | 8 ++++++++
 4 files changed, 20 insertions(+), 6 deletions(-)

diff --git a/tools/perf/builtin-data.c b/tools/perf/builtin-data.c
index b97bc15..38111a9 100644
--- a/tools/perf/builtin-data.c
+++ b/tools/perf/builtin-data.c
@@ -3,6 +3,7 @@
 #include "perf.h"
 #include "debug.h"
 #include <subcmd/parse-options.h>
+#include "data-convert.h"
 #include "data-convert-bt.h"
 
 typedef int (*data_cmd_fn_t)(int argc, const char **argv, const char *prefix);
@@ -53,14 +54,16 @@ static int cmd_data_convert(int argc, const char **argv,
 			    const char *prefix __maybe_unused)
 {
 	const char *to_ctf     = NULL;
-	bool force = false;
+	struct perf_data_convert_opts opts = {
+		.force = false,
+	};
 	const struct option options[] = {
 		OPT_INCR('v', "verbose", &verbose, "be more verbose"),
 		OPT_STRING('i', "input", &input_name, "file", "input file name"),
 #ifdef HAVE_LIBBABELTRACE_SUPPORT
 		OPT_STRING(0, "to-ctf", &to_ctf, NULL, "Convert to CTF format"),
 #endif
-		OPT_BOOLEAN('f', "force", &force, "don't complain, do it"),
+		OPT_BOOLEAN('f', "force", &opts.force, "don't complain, do it"),
 		OPT_END()
 	};
 
@@ -78,7 +81,7 @@ static int cmd_data_convert(int argc, const char **argv,
 
 	if (to_ctf) {
 #ifdef HAVE_LIBBABELTRACE_SUPPORT
-		return bt_convert__perf2ctf(input_name, to_ctf, force);
+		return bt_convert__perf2ctf(input_name, to_ctf, &opts);
 #else
 		pr_err("The libbabeltrace support is not compiled in.\n");
 		return -1;
diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index 4b68e7b..09571b3 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -1304,13 +1304,14 @@ static int convert__config(const char *var, const char *value, void *cb)
 	return 0;
 }
 
-int bt_convert__perf2ctf(const char *input, const char *path, bool force)
+int bt_convert__perf2ctf(const char *input, const char *path,
+			 struct perf_data_convert_opts *opts)
 {
 	struct perf_session *session;
 	struct perf_data_file file = {
 		.path = input,
 		.mode = PERF_DATA_MODE_READ,
-		.force = force,
+		.force = opts->force,
 	};
 	struct convert c = {
 		.tool = {
diff --git a/tools/perf/util/data-convert-bt.h b/tools/perf/util/data-convert-bt.h
index 4c20434..9a3b587 100644
--- a/tools/perf/util/data-convert-bt.h
+++ b/tools/perf/util/data-convert-bt.h
@@ -1,8 +1,10 @@
 #ifndef __DATA_CONVERT_BT_H
 #define __DATA_CONVERT_BT_H
+#include "data-convert.h"
 #ifdef HAVE_LIBBABELTRACE_SUPPORT
 
-int bt_convert__perf2ctf(const char *input_name, const char *to_ctf, bool force);
+int bt_convert__perf2ctf(const char *input_name, const char *to_ctf,
+			 struct perf_data_convert_opts *opts);
 
 #endif /* HAVE_LIBBABELTRACE_SUPPORT */
 #endif /* __DATA_CONVERT_BT_H */
diff --git a/tools/perf/util/data-convert.h b/tools/perf/util/data-convert.h
new file mode 100644
index 0000000..97cfd36
--- /dev/null
+++ b/tools/perf/util/data-convert.h
@@ -0,0 +1,8 @@
+#ifndef __DATA_CONVERT_H
+#define __DATA_CONVERT_H
+
+struct perf_data_convert_opts {
+	bool force;
+};
+
+#endif /* __DATA_CONVERT_H */

[PATCH v2 3/7] perf ctf: Add 'all' option

[Wang Nan]

If 'all' option is selected, 'perf data convert' should convert
not only samples, but non-sample events such as comm and fork. Add this
option in perf_data_convert_opts. Following commits will add cmdline
option to select it.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
---
 tools/perf/builtin-data.c      | 1 +
 tools/perf/util/data-convert.h | 1 +
 2 files changed, 2 insertions(+)

diff --git a/tools/perf/builtin-data.c b/tools/perf/builtin-data.c
index 38111a9..ddfe3ac4 100644
--- a/tools/perf/builtin-data.c
+++ b/tools/perf/builtin-data.c
@@ -56,6 +56,7 @@ static int cmd_data_convert(int argc, const char **argv,
 	const char *to_ctf     = NULL;
 	struct perf_data_convert_opts opts = {
 		.force = false,
+		.all = false,
 	};
 	const struct option options[] = {
 		OPT_INCR('v', "verbose", &verbose, "be more verbose"),
diff --git a/tools/perf/util/data-convert.h b/tools/perf/util/data-convert.h
index 97cfd36..5314962 100644
--- a/tools/perf/util/data-convert.h
+++ b/tools/perf/util/data-convert.h
@@ -3,6 +3,7 @@
 
 struct perf_data_convert_opts {
 	bool force;
+	bool all;
 };
 
 #endif /* __DATA_CONVERT_H */
-- 
1.8.3.4

[tip:perf/core] perf data ctf: Add 'all' option

[tip-bot for Wang Nan]

Commit-ID:  f02a6489d1e181c6c2731e80ff37024a130c326a
Gitweb:     http://git.kernel.org/tip/f02a6489d1e181c6c2731e80ff37024a130c326a
Author:     Wang Nan <wangnan0@huawei.com>
AuthorDate: Fri, 24 Jun 2016 11:22:08 +0000
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Tue, 28 Jun 2016 10:54:56 -0300

perf data ctf: Add 'all' option

If 'all' option is selected, 'perf data convert' should convert not only
samples, but non-sample events such as comm and fork. Add this option in
perf_data_convert_opts. Following commits will add cmdline option to
select it.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Zefan Li <lizefan@huawei.com>
Cc: pi3orama@163.com
Link: http://lkml.kernel.org/r/1466767332-114472-4-git-send-email-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/builtin-data.c      | 1 +
 tools/perf/util/data-convert.h | 1 +
 2 files changed, 2 insertions(+)

diff --git a/tools/perf/builtin-data.c b/tools/perf/builtin-data.c
index 38111a9..ddfe3ac4 100644
--- a/tools/perf/builtin-data.c
+++ b/tools/perf/builtin-data.c
@@ -56,6 +56,7 @@ static int cmd_data_convert(int argc, const char **argv,
 	const char *to_ctf     = NULL;
 	struct perf_data_convert_opts opts = {
 		.force = false,
+		.all = false,
 	};
 	const struct option options[] = {
 		OPT_INCR('v', "verbose", &verbose, "be more verbose"),
diff --git a/tools/perf/util/data-convert.h b/tools/perf/util/data-convert.h
index 97cfd36..5314962 100644
--- a/tools/perf/util/data-convert.h
+++ b/tools/perf/util/data-convert.h
@@ -3,6 +3,7 @@
 
 struct perf_data_convert_opts {
 	bool force;
+	bool all;
 };
 
 #endif /* __DATA_CONVERT_H */

[PATCH v2 4/7] perf ctf: Prepare collect non-sample events

[Wang Nan]

Following commits are going to allow 'perf data convert' to collect not
only samples, but also non-sample events like comm and fork. In this
patch we count non-sample events using c.non_sample_count, and prepare
to print number of both type of events like:

 # ~/perf data convert --all --to-ctf ./out.ctf
 [ perf data convert: Converted 'perf.data' into CTF data './out.ctf' ]
 [ perf data convert: Converted and wrote 0.846 MB (6508 samples, 686 non-samples) ]

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
---
 tools/perf/util/data-convert-bt.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index a5e0f68..bf4c15f 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -76,6 +76,7 @@ struct convert {
 
 	u64			events_size;
 	u64			events_count;
+	u64			non_sample_count;
 
 	/* Ordered events configured queue size. */
 	u64			queue_size;
@@ -1368,10 +1369,15 @@ int bt_convert__perf2ctf(const char *input, const char *path,
 		file.path, path);
 
 	fprintf(stderr,
-		"[ perf data convert: Converted and wrote %.3f MB (%" PRIu64 " samples) ]\n",
+		"[ perf data convert: Converted and wrote %.3f MB (%" PRIu64 " samples",
 		(double) c.events_size / 1024.0 / 1024.0,
 		c.events_count);
 
+	if (!c.non_sample_count)
+		fprintf(stderr, ") ]\n");
+	else
+		fprintf(stderr, ", %" PRIu64 " non-samples) ]\n", c.non_sample_count);
+
 	cleanup_events(session);
 	perf_session__delete(session);
 	ctf_writer__cleanup(cw);
-- 
1.8.3.4

[tip:perf/core] perf data ctf: Prepare collect non-sample events

[tip-bot for Wang Nan]

Commit-ID:  8ee4c46c5ec2481dd18098c5604f791ff911d427
Gitweb:     http://git.kernel.org/tip/8ee4c46c5ec2481dd18098c5604f791ff911d427
Author:     Wang Nan <wangnan0@huawei.com>
AuthorDate: Fri, 24 Jun 2016 11:22:09 +0000
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Tue, 28 Jun 2016 10:54:56 -0300

perf data ctf: Prepare collect non-sample events

Following commits are going to allow 'perf data convert' to collect not
only samples, but also non-sample events like comm and fork. In this
patch we count non-sample events using c.non_sample_count, and prepare
to print number of both type of events like:

  # ~/perf data convert --all --to-ctf ./out.ctf
  [ perf data convert: Converted 'perf.data' into CTF data './out.ctf' ]
  [ perf data convert: Converted and wrote 0.846 MB (6508 samples, 686 non-samples) ]

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Zefan Li <lizefan@huawei.com>
Cc: pi3orama@163.com
Link: http://lkml.kernel.org/r/1466767332-114472-5-git-send-email-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/data-convert-bt.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)

diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index 09571b3..3b3ac7c 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -77,6 +77,7 @@ struct convert {
 
 	u64			events_size;
 	u64			events_count;
+	u64			non_sample_count;
 
 	/* Ordered events configured queue size. */
 	u64			queue_size;
@@ -1369,10 +1370,15 @@ int bt_convert__perf2ctf(const char *input, const char *path,
 		file.path, path);
 
 	fprintf(stderr,
-		"[ perf data convert: Converted and wrote %.3f MB (%" PRIu64 " samples) ]\n",
+		"[ perf data convert: Converted and wrote %.3f MB (%" PRIu64 " samples",
 		(double) c.events_size / 1024.0 / 1024.0,
 		c.events_count);
 
+	if (!c.non_sample_count)
+		fprintf(stderr, ") ]\n");
+	else
+		fprintf(stderr, ", %" PRIu64 " non-samples) ]\n", c.non_sample_count);
+
 	cleanup_events(session);
 	perf_session__delete(session);
 	ctf_writer__cleanup(cw);

[PATCH v2 5/7] perf ctf: Generate comm event to CTF output

[Wang Nan]

If 'all' is selected, convert comm event to output CTF stream.

setup_non_sample_events() is called if non_sample is selected. It creates
a comm_class for comm event.

Use macros to generate and process_comm_event and add_comm_event. These
macros can be reused for other non-sample events.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
---
 tools/perf/util/data-convert-bt.c | 110 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 110 insertions(+)

diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index bf4c15f..31bc81a 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -68,6 +68,7 @@ struct ctf_writer {
 		};
 		struct bt_ctf_field_type *array[6];
 	} data;
+	struct bt_ctf_event_class	*comm_class;
 };
 
 struct convert {
@@ -762,6 +763,57 @@ static int process_sample_event(struct perf_tool *tool,
 	return cs ? 0 : -1;
 }
 
+#define __NON_SAMPLE_SET_FIELD(_name, _type, _field) 	\
+do {							\
+	ret = value_set_##_type(cw, event, #_field, _event->_name._field);\
+	if (ret)					\
+		return -1;				\
+} while(0)
+
+#define __FUNC_PROCESS_NON_SAMPLE(_name, body) 	\
+static int process_##_name##_event(struct perf_tool *tool,	\
+				   union perf_event *_event,	\
+				   struct perf_sample *sample,	\
+				   struct machine *machine)	\
+{								\
+	struct convert *c = container_of(tool, struct convert, tool);\
+	struct ctf_writer *cw = &c->writer;			\
+	struct bt_ctf_event_class *event_class = cw->_name##_class;\
+	struct bt_ctf_event *event;				\
+	struct ctf_stream *cs;					\
+	int ret;						\
+								\
+	c->non_sample_count++;					\
+	c->events_size += _event->header.size;			\
+	event = bt_ctf_event_create(event_class);		\
+	if (!event) {						\
+		pr_err("Failed to create an CTF event\n");	\
+		return -1;					\
+	}							\
+								\
+	bt_ctf_clock_set_time(cw->clock, sample->time);		\
+	body							\
+	cs = ctf_stream(cw, 0);					\
+	if (cs) {						\
+		if (is_flush_needed(cs))			\
+			ctf_stream__flush(cs);			\
+								\
+		cs->count++;					\
+		bt_ctf_stream_append_event(cs->stream, event);	\
+	}							\
+	bt_ctf_event_put(event);				\
+								\
+	return perf_event__process_##_name(tool, _event, sample, machine);\
+}
+
+__FUNC_PROCESS_NON_SAMPLE(comm,
+	__NON_SAMPLE_SET_FIELD(comm, u32, pid);
+	__NON_SAMPLE_SET_FIELD(comm, u32, tid);
+	__NON_SAMPLE_SET_FIELD(comm, string, comm);
+)
+#undef __NON_SAMPLE_SET_FIELD
+#undef __FUNC_PROCESS_NON_SAMPLE
+
 /* If dup < 0, add a prefix. Else, add _dupl_X suffix. */
 static char *change_name(char *name, char *orig_name, int dup)
 {
@@ -1036,6 +1088,58 @@ static int setup_events(struct ctf_writer *cw, struct perf_session *session)
 	return 0;
 }
 
+#define __NON_SAMPLE_ADD_FIELD(t, n)						\
+	do {							\
+		pr2("  field '%s'\n", #n);			\
+		if (bt_ctf_event_class_add_field(event_class, cw->data.t, #n)) {\
+			pr_err("Failed to add field '%s';\n", #n);\
+			return -1;				\
+		}						\
+	} while(0)
+
+#define __FUNC_ADD_NON_SAMPLE_EVENT_CLASS(_name, body) 		\
+static int add_##_name##_event(struct ctf_writer *cw)		\
+{								\
+	struct bt_ctf_event_class *event_class;			\
+	int ret;						\
+								\
+	pr("Adding "#_name" event\n");				\
+	event_class = bt_ctf_event_class_create("perf_" #_name);\
+	if (!event_class)					\
+		return -1;					\
+	body							\
+								\
+	ret = bt_ctf_stream_class_add_event_class(cw->stream_class, event_class);\
+	if (ret) {						\
+		pr("Failed to add event class '"#_name"' into stream.\n");\
+		return ret;					\
+	}							\
+								\
+	cw->_name##_class = event_class;			\
+	bt_ctf_event_class_put(event_class);			\
+	return 0;						\
+}
+
+__FUNC_ADD_NON_SAMPLE_EVENT_CLASS(comm,
+	__NON_SAMPLE_ADD_FIELD(u32, pid);
+	__NON_SAMPLE_ADD_FIELD(u32, tid);
+	__NON_SAMPLE_ADD_FIELD(string, comm);
+)
+
+#undef __NON_SAMPLE_ADD_FIELD
+#undef __FUNC_ADD_NON_SAMPLE_EVENT_CLASS
+
+static int setup_non_sample_events(struct ctf_writer *cw,
+				   struct perf_session *session __maybe_unused)
+{
+	int ret;
+
+	ret = add_comm_event(cw);
+	if (ret)
+		return ret;
+	return 0;
+}
+
 static void cleanup_events(struct perf_session *session)
 {
 	struct perf_evlist *evlist = session->evlist;
@@ -1331,6 +1435,9 @@ int bt_convert__perf2ctf(const char *input, const char *path,
 	struct ctf_writer *cw = &c.writer;
 	int err = -1;
 
+	if (opts->all)
+		c.tool.comm = process_comm_event;
+
 	perf_config(convert__config, &c);
 
 	/* CTF writer */
@@ -1355,6 +1462,9 @@ int bt_convert__perf2ctf(const char *input, const char *path,
 	if (setup_events(cw, session))
 		goto free_session;
 
+	if (opts->all && setup_non_sample_events(cw, session))
+		goto free_session;
+
 	if (setup_streams(cw, session))
 		goto free_session;
 
-- 
1.8.3.4

[tip:perf/core] perf data ctf: Generate comm event to CTF output

[tip-bot for Wang Nan]

Commit-ID:  f5a08ceda55bee91f879d2ac19edeb4a8916d04f
Gitweb:     http://git.kernel.org/tip/f5a08ceda55bee91f879d2ac19edeb4a8916d04f
Author:     Wang Nan <wangnan0@huawei.com>
AuthorDate: Fri, 24 Jun 2016 11:22:10 +0000
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Tue, 28 Jun 2016 10:54:57 -0300

perf data ctf: Generate comm event to CTF output

If 'all' is selected, convert comm event to output CTF stream.

setup_non_sample_events() is called if non_sample is selected. It
creates a comm_class for comm event.

Use macros to generate and process_comm_event and add_comm_event. These
macros can be reused for other non-sample events.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Zefan Li <lizefan@huawei.com>
Cc: pi3orama@163.com
Link: http://lkml.kernel.org/r/1466767332-114472-6-git-send-email-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/data-convert-bt.c | 110 ++++++++++++++++++++++++++++++++++++++
 1 file changed, 110 insertions(+)

diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index 3b3ac7c..5dd62ba 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -69,6 +69,7 @@ struct ctf_writer {
 		};
 		struct bt_ctf_field_type *array[6];
 	} data;
+	struct bt_ctf_event_class	*comm_class;
 };
 
 struct convert {
@@ -763,6 +764,57 @@ static int process_sample_event(struct perf_tool *tool,
 	return cs ? 0 : -1;
 }
 
+#define __NON_SAMPLE_SET_FIELD(_name, _type, _field) 	\
+do {							\
+	ret = value_set_##_type(cw, event, #_field, _event->_name._field);\
+	if (ret)					\
+		return -1;				\
+} while(0)
+
+#define __FUNC_PROCESS_NON_SAMPLE(_name, body) 	\
+static int process_##_name##_event(struct perf_tool *tool,	\
+				   union perf_event *_event,	\
+				   struct perf_sample *sample,	\
+				   struct machine *machine)	\
+{								\
+	struct convert *c = container_of(tool, struct convert, tool);\
+	struct ctf_writer *cw = &c->writer;			\
+	struct bt_ctf_event_class *event_class = cw->_name##_class;\
+	struct bt_ctf_event *event;				\
+	struct ctf_stream *cs;					\
+	int ret;						\
+								\
+	c->non_sample_count++;					\
+	c->events_size += _event->header.size;			\
+	event = bt_ctf_event_create(event_class);		\
+	if (!event) {						\
+		pr_err("Failed to create an CTF event\n");	\
+		return -1;					\
+	}							\
+								\
+	bt_ctf_clock_set_time(cw->clock, sample->time);		\
+	body							\
+	cs = ctf_stream(cw, 0);					\
+	if (cs) {						\
+		if (is_flush_needed(cs))			\
+			ctf_stream__flush(cs);			\
+								\
+		cs->count++;					\
+		bt_ctf_stream_append_event(cs->stream, event);	\
+	}							\
+	bt_ctf_event_put(event);				\
+								\
+	return perf_event__process_##_name(tool, _event, sample, machine);\
+}
+
+__FUNC_PROCESS_NON_SAMPLE(comm,
+	__NON_SAMPLE_SET_FIELD(comm, u32, pid);
+	__NON_SAMPLE_SET_FIELD(comm, u32, tid);
+	__NON_SAMPLE_SET_FIELD(comm, string, comm);
+)
+#undef __NON_SAMPLE_SET_FIELD
+#undef __FUNC_PROCESS_NON_SAMPLE
+
 /* If dup < 0, add a prefix. Else, add _dupl_X suffix. */
 static char *change_name(char *name, char *orig_name, int dup)
 {
@@ -1037,6 +1089,58 @@ static int setup_events(struct ctf_writer *cw, struct perf_session *session)
 	return 0;
 }
 
+#define __NON_SAMPLE_ADD_FIELD(t, n)						\
+	do {							\
+		pr2("  field '%s'\n", #n);			\
+		if (bt_ctf_event_class_add_field(event_class, cw->data.t, #n)) {\
+			pr_err("Failed to add field '%s';\n", #n);\
+			return -1;				\
+		}						\
+	} while(0)
+
+#define __FUNC_ADD_NON_SAMPLE_EVENT_CLASS(_name, body) 		\
+static int add_##_name##_event(struct ctf_writer *cw)		\
+{								\
+	struct bt_ctf_event_class *event_class;			\
+	int ret;						\
+								\
+	pr("Adding "#_name" event\n");				\
+	event_class = bt_ctf_event_class_create("perf_" #_name);\
+	if (!event_class)					\
+		return -1;					\
+	body							\
+								\
+	ret = bt_ctf_stream_class_add_event_class(cw->stream_class, event_class);\
+	if (ret) {						\
+		pr("Failed to add event class '"#_name"' into stream.\n");\
+		return ret;					\
+	}							\
+								\
+	cw->_name##_class = event_class;			\
+	bt_ctf_event_class_put(event_class);			\
+	return 0;						\
+}
+
+__FUNC_ADD_NON_SAMPLE_EVENT_CLASS(comm,
+	__NON_SAMPLE_ADD_FIELD(u32, pid);
+	__NON_SAMPLE_ADD_FIELD(u32, tid);
+	__NON_SAMPLE_ADD_FIELD(string, comm);
+)
+
+#undef __NON_SAMPLE_ADD_FIELD
+#undef __FUNC_ADD_NON_SAMPLE_EVENT_CLASS
+
+static int setup_non_sample_events(struct ctf_writer *cw,
+				   struct perf_session *session __maybe_unused)
+{
+	int ret;
+
+	ret = add_comm_event(cw);
+	if (ret)
+		return ret;
+	return 0;
+}
+
 static void cleanup_events(struct perf_session *session)
 {
 	struct perf_evlist *evlist = session->evlist;
@@ -1332,6 +1436,9 @@ int bt_convert__perf2ctf(const char *input, const char *path,
 	struct ctf_writer *cw = &c.writer;
 	int err = -1;
 
+	if (opts->all)
+		c.tool.comm = process_comm_event;
+
 	perf_config(convert__config, &c);
 
 	/* CTF writer */
@@ -1356,6 +1463,9 @@ int bt_convert__perf2ctf(const char *input, const char *path,
 	if (setup_events(cw, session))
 		goto free_session;
 
+	if (opts->all && setup_non_sample_events(cw, session))
+		goto free_session;
+
 	if (setup_streams(cw, session))
 		goto free_session;
 

[PATCH v2 6/7] perf ctf: Add '--all' option for 'perf data convert'

[Wang Nan]

After this patch, 'perf data convert' convert comm events to output
CTF stream.

Result:

 # perf record -a sleep 1
 [ perf record: Woken up 1 times to write data ]
 [ perf record: Captured and wrote 0.378 MB perf.data (73 samples)  ]

 # perf data convert --to-ctf ./out.ctf
 [ perf data convert: Converted 'perf.data' into CTF data './out.ctf' ]
 [ perf data convert: Converted and wrote 0.003 MB (73 samples) ]

 # babeltrace --clock-seconds ./out.ctf/
 [10627.402515791] (+?.?????????) cycles:ppp: { cpu_id = 0 }, { perf_ip = 0xFFFFFFFF81065AF4, perf_tid = 0, perf_pid = 0, perf_period = 1 }
 [10627.402518972] (+0.000003181) cycles:ppp: { cpu_id = 0 }, { perf_ip = 0xFFFFFFFF81065AF4, perf_tid = 0, perf_pid = 0, perf_period = 1 }
 ...    // only sample event is converted

 # perf data convert --all --to-ctf ./out.ctf
 [ perf data convert: Converted 'perf.data' into CTF data './out.ctf' ]
 [ perf data convert: Converted and wrote 0.023 MB (73 samples, 384 non-samples) ]

 # babeltrace --clock-seconds ./out.ctf/
 [  0.000000000] (+?.?????????) perf_comm: { cpu_id = 0 }, { pid = 1, tid = 1, comm = "init" }
 [  0.000000000] (+0.000000000) perf_comm: { cpu_id = 0 }, { pid = 2, tid = 2, comm = "kthreadd" }
 [  0.000000000] (+0.000000000) perf_comm: { cpu_id = 0 }, { pid = 3, tid = 3, comm = "ksoftirqd/0" }
 ...    // comm events are converted
 [10627.402515791] (+10627.402515791) cycles:ppp: { cpu_id = 0 }, { perf_ip = 0xFFFFFFFF81065AF4, perf_tid = 0, perf_pid = 0, perf_period = 1 }
 [10627.402518972] (+0.000003181) cycles:ppp: { cpu_id = 0 }, { perf_ip = 0xFFFFFFFF81065AF4, perf_tid = 0, perf_pid = 0, perf_period = 1 }
 ...    // samples are also converted

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
---
 tools/perf/Documentation/perf-data.txt | 4 ++++
 tools/perf/builtin-data.c              | 1 +
 2 files changed, 5 insertions(+)

diff --git a/tools/perf/Documentation/perf-data.txt b/tools/perf/Documentation/perf-data.txt
index be8fa1a..f0796a4 100644
--- a/tools/perf/Documentation/perf-data.txt
+++ b/tools/perf/Documentation/perf-data.txt
@@ -34,6 +34,10 @@ OPTIONS for 'convert'
 --verbose::
         Be more verbose (show counter open errors, etc).
 
+--all::
+	Convert all events, including non-sample events (comm, fork, ...), to output.
+	Default is off, only convert samples.
+
 SEE ALSO
 --------
 linkperf:perf[1]
diff --git a/tools/perf/builtin-data.c b/tools/perf/builtin-data.c
index ddfe3ac4..7ad6e17 100644
--- a/tools/perf/builtin-data.c
+++ b/tools/perf/builtin-data.c
@@ -65,6 +65,7 @@ static int cmd_data_convert(int argc, const char **argv,
 		OPT_STRING(0, "to-ctf", &to_ctf, NULL, "Convert to CTF format"),
 #endif
 		OPT_BOOLEAN('f', "force", &opts.force, "don't complain, do it"),
+		OPT_BOOLEAN(0, "all", &opts.all, "Convert all events"),
 		OPT_END()
 	};
 
-- 
1.8.3.4

[tip:perf/core] perf data ctf: Add '--all' option for 'perf data convert'

[tip-bot for Wang Nan]

Commit-ID:  9e1a7ea19f9f8e3e40c5ad1a5cc3615c1746ae7b
Gitweb:     http://git.kernel.org/tip/9e1a7ea19f9f8e3e40c5ad1a5cc3615c1746ae7b
Author:     Wang Nan <wangnan0@huawei.com>
AuthorDate: Fri, 24 Jun 2016 11:22:11 +0000
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Tue, 28 Jun 2016 10:54:57 -0300

perf data ctf: Add '--all' option for 'perf data convert'

After this patch, 'perf data convert' convert comm events to output CTF
stream.

Result:

  # perf record -a sleep 1
  [ perf record: Woken up 1 times to write data ]
  [ perf record: Captured and wrote 0.378 MB perf.data (73 samples)  ]

  # perf data convert --to-ctf ./out.ctf
  [ perf data convert: Converted 'perf.data' into CTF data './out.ctf' ]
  [ perf data convert: Converted and wrote 0.003 MB (73 samples) ]

  # babeltrace --clock-seconds ./out.ctf/
  [10627.402515791] (+?.?????????) cycles:ppp: { cpu_id = 0 }, { perf_ip = 0xFFFFFFFF81065AF4, perf_tid = 0, perf_pid = 0, perf_period = 1 }
  [10627.402518972] (+0.000003181) cycles:ppp: { cpu_id = 0 }, { perf_ip = 0xFFFFFFFF81065AF4, perf_tid = 0, perf_pid = 0, perf_period = 1 }
  ...    // only sample event is converted

  # perf data convert --all --to-ctf ./out.ctf
  [ perf data convert: Converted 'perf.data' into CTF data './out.ctf' ]
  [ perf data convert: Converted and wrote 0.023 MB (73 samples, 384 non-samples) ]

  # babeltrace --clock-seconds ./out.ctf/
  [  0.000000000] (+?.?????????) perf_comm: { cpu_id = 0 }, { pid = 1, tid = 1, comm = "init" }
  [  0.000000000] (+0.000000000) perf_comm: { cpu_id = 0 }, { pid = 2, tid = 2, comm = "kthreadd" }
  [  0.000000000] (+0.000000000) perf_comm: { cpu_id = 0 }, { pid = 3, tid = 3, comm = "ksoftirqd/0" }
  ...    // comm events are converted
  [10627.402515791] (+10627.402515791) cycles:ppp: { cpu_id = 0 }, { perf_ip = 0xFFFFFFFF81065AF4, perf_tid = 0, perf_pid = 0, perf_period = 1 }
  [10627.402518972] (+0.000003181) cycles:ppp: { cpu_id = 0 }, { perf_ip = 0xFFFFFFFF81065AF4, perf_tid = 0, perf_pid = 0, perf_period = 1 }
  ...    // samples are also converted

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Zefan Li <lizefan@huawei.com>
Cc: pi3orama@163.com
Link: http://lkml.kernel.org/r/1466767332-114472-7-git-send-email-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/Documentation/perf-data.txt | 4 ++++
 tools/perf/builtin-data.c              | 1 +
 2 files changed, 5 insertions(+)

diff --git a/tools/perf/Documentation/perf-data.txt b/tools/perf/Documentation/perf-data.txt
index be8fa1a..f0796a4 100644
--- a/tools/perf/Documentation/perf-data.txt
+++ b/tools/perf/Documentation/perf-data.txt
@@ -34,6 +34,10 @@ OPTIONS for 'convert'
 --verbose::
         Be more verbose (show counter open errors, etc).
 
+--all::
+	Convert all events, including non-sample events (comm, fork, ...), to output.
+	Default is off, only convert samples.
+
 SEE ALSO
 --------
 linkperf:perf[1]
diff --git a/tools/perf/builtin-data.c b/tools/perf/builtin-data.c
index ddfe3ac4..7ad6e17 100644
--- a/tools/perf/builtin-data.c
+++ b/tools/perf/builtin-data.c
@@ -65,6 +65,7 @@ static int cmd_data_convert(int argc, const char **argv,
 		OPT_STRING(0, "to-ctf", &to_ctf, NULL, "Convert to CTF format"),
 #endif
 		OPT_BOOLEAN('f', "force", &opts.force, "don't complain, do it"),
+		OPT_BOOLEAN(0, "all", &opts.all, "Convert all events"),
 		OPT_END()
 	};
 

[PATCH v2 7/7] perf ctf: Generate fork and exit events to CTF output

[Wang Nan]

If 'all' is selected, convert fork and exit events to output CTF stream.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
---
 tools/perf/util/data-convert-bt.c | 44 ++++++++++++++++++++++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)

diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index 31bc81a..acd11d4 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -69,6 +69,8 @@ struct ctf_writer {
 		struct bt_ctf_field_type *array[6];
 	} data;
 	struct bt_ctf_event_class	*comm_class;
+	struct bt_ctf_event_class	*exit_class;
+	struct bt_ctf_event_class	*fork_class;
 };
 
 struct convert {
@@ -811,6 +813,21 @@ __FUNC_PROCESS_NON_SAMPLE(comm,
 	__NON_SAMPLE_SET_FIELD(comm, u32, tid);
 	__NON_SAMPLE_SET_FIELD(comm, string, comm);
 )
+__FUNC_PROCESS_NON_SAMPLE(fork,
+	__NON_SAMPLE_SET_FIELD(fork, u32, pid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, ppid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, tid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, ptid);
+	__NON_SAMPLE_SET_FIELD(fork, u64, time);
+)
+
+__FUNC_PROCESS_NON_SAMPLE(exit,
+	__NON_SAMPLE_SET_FIELD(fork, u32, pid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, ppid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, tid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, ptid);
+	__NON_SAMPLE_SET_FIELD(fork, u64, time);
+)
 #undef __NON_SAMPLE_SET_FIELD
 #undef __FUNC_PROCESS_NON_SAMPLE
 
@@ -1126,6 +1143,22 @@ __FUNC_ADD_NON_SAMPLE_EVENT_CLASS(comm,
 	__NON_SAMPLE_ADD_FIELD(string, comm);
 )
 
+__FUNC_ADD_NON_SAMPLE_EVENT_CLASS(fork,
+	__NON_SAMPLE_ADD_FIELD(u32, pid);
+	__NON_SAMPLE_ADD_FIELD(u32, ppid);
+	__NON_SAMPLE_ADD_FIELD(u32, tid);
+	__NON_SAMPLE_ADD_FIELD(u32, ptid);
+	__NON_SAMPLE_ADD_FIELD(u64, time);
+)
+
+__FUNC_ADD_NON_SAMPLE_EVENT_CLASS(exit,
+	__NON_SAMPLE_ADD_FIELD(u32, pid);
+	__NON_SAMPLE_ADD_FIELD(u32, ppid);
+	__NON_SAMPLE_ADD_FIELD(u32, tid);
+	__NON_SAMPLE_ADD_FIELD(u32, ptid);
+	__NON_SAMPLE_ADD_FIELD(u64, time);
+)
+
 #undef __NON_SAMPLE_ADD_FIELD
 #undef __FUNC_ADD_NON_SAMPLE_EVENT_CLASS
 
@@ -1137,6 +1170,12 @@ static int setup_non_sample_events(struct ctf_writer *cw,
 	ret = add_comm_event(cw);
 	if (ret)
 		return ret;
+	ret = add_exit_event(cw);
+	if (ret)
+		return ret;
+	ret = add_fork_event(cw);
+	if (ret)
+		return ret;
 	return 0;
 }
 
@@ -1435,8 +1474,11 @@ int bt_convert__perf2ctf(const char *input, const char *path,
 	struct ctf_writer *cw = &c.writer;
 	int err = -1;
 
-	if (opts->all)
+	if (opts->all) {
 		c.tool.comm = process_comm_event;
+		c.tool.exit = process_exit_event;
+		c.tool.fork = process_fork_event;
+	}
 
 	perf_config(convert__config, &c);
 
-- 
1.8.3.4

[tip:perf/core] perf data ctf: Generate fork and exit events to CTF output

[tip-bot for Wang Nan]

Commit-ID:  ebccba3fe0a02f622f80e6be0e8ecb1a9a3ed983
Gitweb:     http://git.kernel.org/tip/ebccba3fe0a02f622f80e6be0e8ecb1a9a3ed983
Author:     Wang Nan <wangnan0@huawei.com>
AuthorDate: Fri, 24 Jun 2016 11:22:12 +0000
Committer:  Arnaldo Carvalho de Melo <acme@redhat.com>
CommitDate: Tue, 28 Jun 2016 10:54:58 -0300

perf data ctf: Generate fork and exit events to CTF output

If 'all' is selected, convert fork and exit events to output CTF stream.

Signed-off-by: Wang Nan <wangnan0@huawei.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Cc: Zefan Li <lizefan@huawei.com>
Cc: pi3orama@163.com
Link: http://lkml.kernel.org/r/1466767332-114472-8-git-send-email-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/data-convert-bt.c | 44 ++++++++++++++++++++++++++++++++++++++-
 1 file changed, 43 insertions(+), 1 deletion(-)

diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index 5dd62ba..4f979bb 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -70,6 +70,8 @@ struct ctf_writer {
 		struct bt_ctf_field_type *array[6];
 	} data;
 	struct bt_ctf_event_class	*comm_class;
+	struct bt_ctf_event_class	*exit_class;
+	struct bt_ctf_event_class	*fork_class;
 };
 
 struct convert {
@@ -812,6 +814,21 @@ __FUNC_PROCESS_NON_SAMPLE(comm,
 	__NON_SAMPLE_SET_FIELD(comm, u32, tid);
 	__NON_SAMPLE_SET_FIELD(comm, string, comm);
 )
+__FUNC_PROCESS_NON_SAMPLE(fork,
+	__NON_SAMPLE_SET_FIELD(fork, u32, pid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, ppid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, tid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, ptid);
+	__NON_SAMPLE_SET_FIELD(fork, u64, time);
+)
+
+__FUNC_PROCESS_NON_SAMPLE(exit,
+	__NON_SAMPLE_SET_FIELD(fork, u32, pid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, ppid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, tid);
+	__NON_SAMPLE_SET_FIELD(fork, u32, ptid);
+	__NON_SAMPLE_SET_FIELD(fork, u64, time);
+)
 #undef __NON_SAMPLE_SET_FIELD
 #undef __FUNC_PROCESS_NON_SAMPLE
 
@@ -1127,6 +1144,22 @@ __FUNC_ADD_NON_SAMPLE_EVENT_CLASS(comm,
 	__NON_SAMPLE_ADD_FIELD(string, comm);
 )
 
+__FUNC_ADD_NON_SAMPLE_EVENT_CLASS(fork,
+	__NON_SAMPLE_ADD_FIELD(u32, pid);
+	__NON_SAMPLE_ADD_FIELD(u32, ppid);
+	__NON_SAMPLE_ADD_FIELD(u32, tid);
+	__NON_SAMPLE_ADD_FIELD(u32, ptid);
+	__NON_SAMPLE_ADD_FIELD(u64, time);
+)
+
+__FUNC_ADD_NON_SAMPLE_EVENT_CLASS(exit,
+	__NON_SAMPLE_ADD_FIELD(u32, pid);
+	__NON_SAMPLE_ADD_FIELD(u32, ppid);
+	__NON_SAMPLE_ADD_FIELD(u32, tid);
+	__NON_SAMPLE_ADD_FIELD(u32, ptid);
+	__NON_SAMPLE_ADD_FIELD(u64, time);
+)
+
 #undef __NON_SAMPLE_ADD_FIELD
 #undef __FUNC_ADD_NON_SAMPLE_EVENT_CLASS
 
@@ -1138,6 +1171,12 @@ static int setup_non_sample_events(struct ctf_writer *cw,
 	ret = add_comm_event(cw);
 	if (ret)
 		return ret;
+	ret = add_exit_event(cw);
+	if (ret)
+		return ret;
+	ret = add_fork_event(cw);
+	if (ret)
+		return ret;
 	return 0;
 }
 
@@ -1436,8 +1475,11 @@ int bt_convert__perf2ctf(const char *input, const char *path,
 	struct ctf_writer *cw = &c.writer;
 	int err = -1;
 
-	if (opts->all)
+	if (opts->all) {
 		c.tool.comm = process_comm_event;
+		c.tool.exit = process_exit_event;
+		c.tool.fork = process_fork_event;
+	}
 
 	perf_config(convert__config, &c);
 

Re: [PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[Jiri Olsa]

On Fri, Jun 24, 2016 at 11:22:05AM +0000, Wang Nan wrote:
> After converting perf.data to CTF, we lost pid-tid-comm mapping. Which
> makes inconvience. For example, in perf script output we know which
> process issue an event like this:
> 
>  compiz 19361 [001] 3275709.313929:  raw_syscalls:sys_exit: NR 7 = 0
> 
> After converting to CTF, we only get this:
> 
>   [3275709.313929985] (+0.110646118) raw_syscalls:sys_exit: { cpu_id = 1 }, { perf_ip = 0xFFFFFFFF8107B2E8, perf_tid = 19361, perf_pid = 19361, perf_id = 18920, perf_period = 1, common_type = 16, common_flags = 0, common_preempt_count = 1, common_pid = 19361, id = 7, ret = 0 }
> 
> Currently, if we want to find the name and parent of a process, we
> have to collect 'sched:sched_switch' event.
> 
> This patch set adds a '--all' option to 'perf convert', converts comm,
> fork and exit events to CTF output. CTF user now can track the mapping
> by their own.
> 
> v1 -> v2: Report number of sample and non-sample events when finish.
>           rename opts.non_sample to opts.all.
> 
> Wang Nan (7):
>   perf ctf: Add value_set_string() helper
>   perf ctf: Pass convert options through opts structure
>   perf ctf: Add 'all' option
>   perf ctf: Prepare collect non-sample events
>   perf ctf: Generate comm event to CTF output
>   perf ctf: Add '--all' option for 'perf data convert'
>   perf ctf: Generate fork and exit events to CTF output

Acked-by: Jiri Olsa <jolsa@kernel.org>

thanks,
jirka

Re: [PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[Arnaldo Carvalho de Melo]

Em Fri, Jun 24, 2016 at 02:28:51PM +0200, Jiri Olsa escreveu:
> On Fri, Jun 24, 2016 at 11:22:05AM +0000, Wang Nan wrote:
> > After converting perf.data to CTF, we lost pid-tid-comm mapping. Which
> > makes inconvience. For example, in perf script output we know which
> > process issue an event like this:
> > 
> >  compiz 19361 [001] 3275709.313929:  raw_syscalls:sys_exit: NR 7 = 0
> > 
> > After converting to CTF, we only get this:
> > 
> >   [3275709.313929985] (+0.110646118) raw_syscalls:sys_exit: { cpu_id = 1 }, { perf_ip = 0xFFFFFFFF8107B2E8, perf_tid = 19361, perf_pid = 19361, perf_id = 18920, perf_period = 1, common_type = 16, common_flags = 0, common_preempt_count = 1, common_pid = 19361, id = 7, ret = 0 }
> > 
> > Currently, if we want to find the name and parent of a process, we
> > have to collect 'sched:sched_switch' event.
> > 
> > This patch set adds a '--all' option to 'perf convert', converts comm,
> > fork and exit events to CTF output. CTF user now can track the mapping
> > by their own.
> > 
> > v1 -> v2: Report number of sample and non-sample events when finish.
> >           rename opts.non_sample to opts.all.
> > 
> > Wang Nan (7):
> >   perf ctf: Add value_set_string() helper
> >   perf ctf: Pass convert options through opts structure
> >   perf ctf: Add 'all' option
> >   perf ctf: Prepare collect non-sample events
> >   perf ctf: Generate comm event to CTF output
> >   perf ctf: Add '--all' option for 'perf data convert'
> >   perf ctf: Generate fork and exit events to CTF output
> 
> Acked-by: Jiri Olsa <jolsa@kernel.org>

Thanks, applied.

- Arnaldo
- Arnaldo

Re: [PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[Jiri Olsa]

On Fri, Jun 24, 2016 at 11:22:05AM +0000, Wang Nan wrote:
> After converting perf.data to CTF, we lost pid-tid-comm mapping. Which
> makes inconvience. For example, in perf script output we know which
> process issue an event like this:
> 
>  compiz 19361 [001] 3275709.313929:  raw_syscalls:sys_exit: NR 7 = 0
> 
> After converting to CTF, we only get this:
> 
>   [3275709.313929985] (+0.110646118) raw_syscalls:sys_exit: { cpu_id = 1 }, { perf_ip = 0xFFFFFFFF8107B2E8, perf_tid = 19361, perf_pid = 19361, perf_id = 18920, perf_period = 1, common_type = 16, common_flags = 0, common_preempt_count = 1, common_pid = 19361, id = 7, ret = 0 }
> 
> Currently, if we want to find the name and parent of a process, we
> have to collect 'sched:sched_switch' event.
> 
> This patch set adds a '--all' option to 'perf convert', converts comm,
> fork and exit events to CTF output. CTF user now can track the mapping
> by their own.
> 
> v1 -> v2: Report number of sample and non-sample events when finish.
>           rename opts.non_sample to opts.all.
> 
> Wang Nan (7):
>   perf ctf: Add value_set_string() helper
>   perf ctf: Pass convert options through opts structure
>   perf ctf: Add 'all' option
>   perf ctf: Prepare collect non-sample events
>   perf ctf: Generate comm event to CTF output
>   perf ctf: Add '--all' option for 'perf data convert'
>   perf ctf: Generate fork and exit events to CTF output

I can't compile unless I can include config.h

[jolsa@krava perf]$ make LIBBABELTRACE_DIR=/opt/libbabeltrace/ LIBBABELTRACE=1
  BUILD:   Doing 'make -j4' parallel build
  CC       util/data-convert-bt.o
util/data-convert-bt.c: In function ‘convert__config’:
util/data-convert-bt.c:1299:19: error: implicit declaration of function ‘perf_config_u64’ [-Werror=implicit-function-declaration]
   c->queue_size = perf_config_u64(var, value);
                   ^
util/data-convert-bt.c:1299:3: error: nested extern declaration of ‘perf_config_u64’ [-Werror=nested-externs]
   c->queue_size = perf_config_u64(var, value);
   ^
util/data-convert-bt.c: In function ‘bt_convert__perf2ctf’:
util/data-convert-bt.c:1332:2: error: implicit declaration of function ‘perf_config’ [-Werror=implicit-function-declaration]
  perf_config(convert__config, &c);
  ^
util/data-convert-bt.c:1332:2: error: nested extern declaration of ‘perf_config’ [-Werror=nested-externs]
cc1: all warnings being treated as errors


your compiler's not that strict I guess ;-)
I'll post it shortly

jirka

---
diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
index 79082782e7d2..4b68e7b9ee0c 100644
--- a/tools/perf/util/data-convert-bt.c
+++ b/tools/perf/util/data-convert-bt.c
@@ -26,6 +26,7 @@
 #include "evlist.h"
 #include "evsel.h"
 #include "machine.h"
+#include "config.h"
 
 #define pr_N(n, fmt, ...) \
 	eprintf(n, debug_data_convert, fmt, ##__VA_ARGS__)

Re: [PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[pi3orama]

发自我的 iPhone

> 在 2016年6月24日，下午8:29，Jiri Olsa <jolsa@redhat.com> 写道：
> 
>> On Fri, Jun 24, 2016 at 11:22:05AM +0000, Wang Nan wrote:
>> After converting perf.data to CTF, we lost pid-tid-comm mapping. Which
>> makes inconvience. For example, in perf script output we know which
>> process issue an event like this:
>> 
>> compiz 19361 [001] 3275709.313929:  raw_syscalls:sys_exit: NR 7 = 0
>> 
>> After converting to CTF, we only get this:
>> 
>>  [3275709.313929985] (+0.110646118) raw_syscalls:sys_exit: { cpu_id = 1 }, { perf_ip = 0xFFFFFFFF8107B2E8, perf_tid = 19361, perf_pid = 19361, perf_id = 18920, perf_period = 1, common_type = 16, common_flags = 0, common_preempt_count = 1, common_pid = 19361, id = 7, ret = 0 }
>> 
>> Currently, if we want to find the name and parent of a process, we
>> have to collect 'sched:sched_switch' event.
>> 
>> This patch set adds a '--all' option to 'perf convert', converts comm,
>> fork and exit events to CTF output. CTF user now can track the mapping
>> by their own.
>> 
>> v1 -> v2: Report number of sample and non-sample events when finish.
>>          rename opts.non_sample to opts.all.
>> 
>> Wang Nan (7):
>>  perf ctf: Add value_set_string() helper
>>  perf ctf: Pass convert options through opts structure
>>  perf ctf: Add 'all' option
>>  perf ctf: Prepare collect non-sample events
>>  perf ctf: Generate comm event to CTF output
>>  perf ctf: Add '--all' option for 'perf data convert'
>>  perf ctf: Generate fork and exit events to CTF output
> 
> I can't compile unless I can include config.h
> 
> [jolsa@krava perf]$ make LIBBABELTRACE_DIR=/opt/libbabeltrace/ LIBBABELTRACE=1
>  BUILD:   Doing 'make -j4' parallel build
>  CC       util/data-convert-bt.o
> util/data-convert-bt.c: In function ‘convert__config’:
> util/data-convert-bt.c:1299:19: error: implicit declaration of function ‘perf_config_u64’ [-Werror=implicit-function-declaration]
>   c->queue_size = perf_config_u64(var, value);
>                   ^
> util/data-convert-bt.c:1299:3: error: nested extern declaration of ‘perf_config_u64’ [-Werror=nested-externs]
>   c->queue_size = perf_config_u64(var, value);
>   ^
> util/data-convert-bt.c: In function ‘bt_convert__perf2ctf’:
> util/data-convert-bt.c:1332:2: error: implicit declaration of function ‘perf_config’ [-Werror=implicit-function-declaration]
>  perf_config(convert__config, &c);
>  ^
> util/data-convert-bt.c:1332:2: error: nested extern declaration of ‘perf_config’ [-Werror=nested-externs]
> cc1: all warnings being treated as errors
> 

Strange.

The error message seems unrelated to my
patch, right?

Thank you.

> 
> your compiler's not that strict I guess ;-)
> I'll post it shortly
> 
> jirka
> 
> ---
> diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
> index 79082782e7d2..4b68e7b9ee0c 100644
> --- a/tools/perf/util/data-convert-bt.c
> +++ b/tools/perf/util/data-convert-bt.c
> @@ -26,6 +26,7 @@
> #include "evlist.h"
> #include "evsel.h"
> #include "machine.h"
> +#include "config.h"
> 
> #define pr_N(n, fmt, ...) \
>    eprintf(n, debug_data_convert, fmt, ##__VA_ARGS__)

Re: [PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[Jiri Olsa]

On Fri, Jun 24, 2016 at 08:33:27PM +0800, pi3orama wrote:

SNIP

> > I can't compile unless I can include config.h
> > 
> > [jolsa@krava perf]$ make LIBBABELTRACE_DIR=/opt/libbabeltrace/ LIBBABELTRACE=1
> >  BUILD:   Doing 'make -j4' parallel build
> >  CC       util/data-convert-bt.o
> > util/data-convert-bt.c: In function ‘convert__config’:
> > util/data-convert-bt.c:1299:19: error: implicit declaration of function ‘perf_config_u64’ [-Werror=implicit-function-declaration]
> >   c->queue_size = perf_config_u64(var, value);
> >                   ^
> > util/data-convert-bt.c:1299:3: error: nested extern declaration of ‘perf_config_u64’ [-Werror=nested-externs]
> >   c->queue_size = perf_config_u64(var, value);
> >   ^
> > util/data-convert-bt.c: In function ‘bt_convert__perf2ctf’:
> > util/data-convert-bt.c:1332:2: error: implicit declaration of function ‘perf_config’ [-Werror=implicit-function-declaration]
> >  perf_config(convert__config, &c);
> >  ^
> > util/data-convert-bt.c:1332:2: error: nested extern declaration of ‘perf_config’ [-Werror=nested-externs]
> > cc1: all warnings being treated as errors
> > 
> 
> Strange.
> 
> The error message seems unrelated to my
> patch, right?

yep, it's unrelated and I just posted the fix

I haven't enabled/compiled this code for a while and that's
what I've got.. I assume we're using different compilers

jirka

Re: [PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[Arnaldo Carvalho de Melo]

Em Fri, Jun 24, 2016 at 02:29:07PM +0200, Jiri Olsa escreveu:
> On Fri, Jun 24, 2016 at 11:22:05AM +0000, Wang Nan wrote:
> > After converting perf.data to CTF, we lost pid-tid-comm mapping. Which
> > makes inconvience. For example, in perf script output we know which
> > process issue an event like this:
> > 
> >  compiz 19361 [001] 3275709.313929:  raw_syscalls:sys_exit: NR 7 = 0
> > 
> > After converting to CTF, we only get this:
> > 
> >   [3275709.313929985] (+0.110646118) raw_syscalls:sys_exit: { cpu_id = 1 }, { perf_ip = 0xFFFFFFFF8107B2E8, perf_tid = 19361, perf_pid = 19361, perf_id = 18920, perf_period = 1, common_type = 16, common_flags = 0, common_preempt_count = 1, common_pid = 19361, id = 7, ret = 0 }
> > 
> > Currently, if we want to find the name and parent of a process, we
> > have to collect 'sched:sched_switch' event.
> > 
> > This patch set adds a '--all' option to 'perf convert', converts comm,
> > fork and exit events to CTF output. CTF user now can track the mapping
> > by their own.
> > 
> > v1 -> v2: Report number of sample and non-sample events when finish.
> >           rename opts.non_sample to opts.all.
> > 
> > Wang Nan (7):
> >   perf ctf: Add value_set_string() helper
> >   perf ctf: Pass convert options through opts structure
> >   perf ctf: Add 'all' option
> >   perf ctf: Prepare collect non-sample events
> >   perf ctf: Generate comm event to CTF output
> >   perf ctf: Add '--all' option for 'perf data convert'
> >   perf ctf: Generate fork and exit events to CTF output
> 
> I can't compile unless I can include config.h

Waiting for this fix to proceed, but it seems we don't have this covered
in 'build-test', right? ;-)

- Arnaldo
 
> [jolsa@krava perf]$ make LIBBABELTRACE_DIR=/opt/libbabeltrace/ LIBBABELTRACE=1
>   BUILD:   Doing 'make -j4' parallel build
>   CC       util/data-convert-bt.o
> util/data-convert-bt.c: In function ‘convert__config’:
> util/data-convert-bt.c:1299:19: error: implicit declaration of function ‘perf_config_u64’ [-Werror=implicit-function-declaration]
>    c->queue_size = perf_config_u64(var, value);
>                    ^
> util/data-convert-bt.c:1299:3: error: nested extern declaration of ‘perf_config_u64’ [-Werror=nested-externs]
>    c->queue_size = perf_config_u64(var, value);
>    ^
> util/data-convert-bt.c: In function ‘bt_convert__perf2ctf’:
> util/data-convert-bt.c:1332:2: error: implicit declaration of function ‘perf_config’ [-Werror=implicit-function-declaration]
>   perf_config(convert__config, &c);
>   ^
> util/data-convert-bt.c:1332:2: error: nested extern declaration of ‘perf_config’ [-Werror=nested-externs]
> cc1: all warnings being treated as errors
> 
> 
> your compiler's not that strict I guess ;-)
> I'll post it shortly
> 
> jirka
> 
> ---
> diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
> index 79082782e7d2..4b68e7b9ee0c 100644
> --- a/tools/perf/util/data-convert-bt.c
> +++ b/tools/perf/util/data-convert-bt.c
> @@ -26,6 +26,7 @@
>  #include "evlist.h"
>  #include "evsel.h"
>  #include "machine.h"
> +#include "config.h"
>  
>  #define pr_N(n, fmt, ...) \
>  	eprintf(n, debug_data_convert, fmt, ##__VA_ARGS__)

Re: [PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[pi3orama]

发自我的 iPhone

> 在 2016年6月24日，下午8:41，Arnaldo Carvalho de Melo <acme@kernel.org> 写道：
> 
> Em Fri, Jun 24, 2016 at 02:29:07PM +0200, Jiri Olsa escreveu:
>> On Fri, Jun 24, 2016 at 11:22:05AM +0000, Wang Nan wrote:
>>> After converting perf.data to CTF, we lost pid-tid-comm mapping. Which
>>> makes inconvience. For example, in perf script output we know which
>>> process issue an event like this:
>>> 
>>> compiz 19361 [001] 3275709.313929:  raw_syscalls:sys_exit: NR 7 = 0
>>> 
>>> After converting to CTF, we only get this:
>>> 
>>>  [3275709.313929985] (+0.110646118) raw_syscalls:sys_exit: { cpu_id = 1 }, { perf_ip = 0xFFFFFFFF8107B2E8, perf_tid = 19361, perf_pid = 19361, perf_id = 18920, perf_period = 1, common_type = 16, common_flags = 0, common_preempt_count = 1, common_pid = 19361, id = 7, ret = 0 }
>>> 
>>> Currently, if we want to find the name and parent of a process, we
>>> have to collect 'sched:sched_switch' event.
>>> 
>>> This patch set adds a '--all' option to 'perf convert', converts comm,
>>> fork and exit events to CTF output. CTF user now can track the mapping
>>> by their own.
>>> 
>>> v1 -> v2: Report number of sample and non-sample events when finish.
>>>          rename opts.non_sample to opts.all.
>>> 
>>> Wang Nan (7):
>>>  perf ctf: Add value_set_string() helper
>>>  perf ctf: Pass convert options through opts structure
>>>  perf ctf: Add 'all' option
>>>  perf ctf: Prepare collect non-sample events
>>>  perf ctf: Generate comm event to CTF output
>>>  perf ctf: Add '--all' option for 'perf data convert'
>>>  perf ctf: Generate fork and exit events to CTF output
>> 
>> I can't compile unless I can include config.h
> 
> Waiting for this fix to proceed, but it seems we don't have this covered
> in 'build-test', right? ;-)
> 

CTF support is off by default. When can we
turn it on like other options?

Thank you.

> - Arnaldo
> 
>> [jolsa@krava perf]$ make LIBBABELTRACE_DIR=/opt/libbabeltrace/ LIBBABELTRACE=1
>>  BUILD:   Doing 'make -j4' parallel build
>>  CC       util/data-convert-bt.o
>> util/data-convert-bt.c: In function ‘convert__config’:
>> util/data-convert-bt.c:1299:19: error: implicit declaration of function ‘perf_config_u64’ [-Werror=implicit-function-declaration]
>>   c->queue_size = perf_config_u64(var, value);
>>                   ^
>> util/data-convert-bt.c:1299:3: error: nested extern declaration of ‘perf_config_u64’ [-Werror=nested-externs]
>>   c->queue_size = perf_config_u64(var, value);
>>   ^
>> util/data-convert-bt.c: In function ‘bt_convert__perf2ctf’:
>> util/data-convert-bt.c:1332:2: error: implicit declaration of function ‘perf_config’ [-Werror=implicit-function-declaration]
>>  perf_config(convert__config, &c);
>>  ^
>> util/data-convert-bt.c:1332:2: error: nested extern declaration of ‘perf_config’ [-Werror=nested-externs]
>> cc1: all warnings being treated as errors
>> 
>> 
>> your compiler's not that strict I guess ;-)
>> I'll post it shortly
>> 
>> jirka
>> 
>> ---
>> diff --git a/tools/perf/util/data-convert-bt.c b/tools/perf/util/data-convert-bt.c
>> index 79082782e7d2..4b68e7b9ee0c 100644
>> --- a/tools/perf/util/data-convert-bt.c
>> +++ b/tools/perf/util/data-convert-bt.c
>> @@ -26,6 +26,7 @@
>> #include "evlist.h"
>> #include "evsel.h"
>> #include "machine.h"
>> +#include "config.h"
>> 
>> #define pr_N(n, fmt, ...) \
>>    eprintf(n, debug_data_convert, fmt, ##__VA_ARGS__)

Re: [PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[Jiri Olsa]

On Fri, Jun 24, 2016 at 08:43:56PM +0800, pi3orama wrote:
> 
> 
> 发自我的 iPhone
> 
> > 在 2016年6月24日，下午8:41，Arnaldo Carvalho de Melo <acme@kernel.org> 写道：
> > 
> > Em Fri, Jun 24, 2016 at 02:29:07PM +0200, Jiri Olsa escreveu:
> >> On Fri, Jun 24, 2016 at 11:22:05AM +0000, Wang Nan wrote:
> >>> After converting perf.data to CTF, we lost pid-tid-comm mapping. Which
> >>> makes inconvience. For example, in perf script output we know which
> >>> process issue an event like this:
> >>> 
> >>> compiz 19361 [001] 3275709.313929:  raw_syscalls:sys_exit: NR 7 = 0
> >>> 
> >>> After converting to CTF, we only get this:
> >>> 
> >>>  [3275709.313929985] (+0.110646118) raw_syscalls:sys_exit: { cpu_id = 1 }, { perf_ip = 0xFFFFFFFF8107B2E8, perf_tid = 19361, perf_pid = 19361, perf_id = 18920, perf_period = 1, common_type = 16, common_flags = 0, common_preempt_count = 1, common_pid = 19361, id = 7, ret = 0 }
> >>> 
> >>> Currently, if we want to find the name and parent of a process, we
> >>> have to collect 'sched:sched_switch' event.
> >>> 
> >>> This patch set adds a '--all' option to 'perf convert', converts comm,
> >>> fork and exit events to CTF output. CTF user now can track the mapping
> >>> by their own.
> >>> 
> >>> v1 -> v2: Report number of sample and non-sample events when finish.
> >>>          rename opts.non_sample to opts.all.
> >>> 
> >>> Wang Nan (7):
> >>>  perf ctf: Add value_set_string() helper
> >>>  perf ctf: Pass convert options through opts structure
> >>>  perf ctf: Add 'all' option
> >>>  perf ctf: Prepare collect non-sample events
> >>>  perf ctf: Generate comm event to CTF output
> >>>  perf ctf: Add '--all' option for 'perf data convert'
> >>>  perf ctf: Generate fork and exit events to CTF output
> >> 
> >> I can't compile unless I can include config.h
> > 
> > Waiting for this fix to proceed, but it seems we don't have this covered
> > in 'build-test', right? ;-)
> > 
> 
> CTF support is off by default. When can we
> turn it on like other options?

we switched it off because the latest version we needed
wasn't part of main distros.. it might have changed now

jirka

Re: [PATCH v2 0/7] perf ctf: Convert comm, fork and exit events to CTF

[Jiri Olsa]

On Fri, Jun 24, 2016 at 09:41:13AM -0300, Arnaldo Carvalho de Melo wrote:
> Em Fri, Jun 24, 2016 at 02:29:07PM +0200, Jiri Olsa escreveu:
> > On Fri, Jun 24, 2016 at 11:22:05AM +0000, Wang Nan wrote:
> > > After converting perf.data to CTF, we lost pid-tid-comm mapping. Which
> > > makes inconvience. For example, in perf script output we know which
> > > process issue an event like this:
> > > 
> > >  compiz 19361 [001] 3275709.313929:  raw_syscalls:sys_exit: NR 7 = 0
> > > 
> > > After converting to CTF, we only get this:
> > > 
> > >   [3275709.313929985] (+0.110646118) raw_syscalls:sys_exit: { cpu_id = 1 }, { perf_ip = 0xFFFFFFFF8107B2E8, perf_tid = 19361, perf_pid = 19361, perf_id = 18920, perf_period = 1, common_type = 16, common_flags = 0, common_preempt_count = 1, common_pid = 19361, id = 7, ret = 0 }
> > > 
> > > Currently, if we want to find the name and parent of a process, we
> > > have to collect 'sched:sched_switch' event.
> > > 
> > > This patch set adds a '--all' option to 'perf convert', converts comm,
> > > fork and exit events to CTF output. CTF user now can track the mapping
> > > by their own.
> > > 
> > > v1 -> v2: Report number of sample and non-sample events when finish.
> > >           rename opts.non_sample to opts.all.
> > > 
> > > Wang Nan (7):
> > >   perf ctf: Add value_set_string() helper
> > >   perf ctf: Pass convert options through opts structure
> > >   perf ctf: Add 'all' option
> > >   perf ctf: Prepare collect non-sample events
> > >   perf ctf: Generate comm event to CTF output
> > >   perf ctf: Add '--all' option for 'perf data convert'
> > >   perf ctf: Generate fork and exit events to CTF output
> > 
> > I can't compile unless I can include config.h
> 
> Waiting for this fix to proceed, but it seems we don't have this covered

just posted it ;-)

> in 'build-test', right? ;-)

right, but we dont cover babletrace stuff in tests/make

jirka