[GIT PULL 0/5] perf/core improvements and fixes

[GIT PULL 0/5] perf/core improvements and fixes

[Arnaldo Carvalho de Melo]

Hi Ingo,

	Please consider pulling,

- Arnaldo

The following changes since commit 712df65ccb63da08a484bf57c40b250dfd4103a7:

  perf/x86/intel/uncore: Fix multi-segment problem of perf_event_intel_uncore (2015-10-06 17:31:51 +0200)

are available in the git repository at:

  git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux.git tags/perf-core-for-mingo

for you to fetch changes up to ae938802443732e77d01f8d5b52b900b9327ff30:

  perf python: Support the PERF_RECORD_SWITCH event (2015-10-07 19:41:50 -0300)

----------------------------------------------------------------
perf/core improvements and fixes:

User visible:

- Adding a field via 'perf report -F' that already is enabled makes
  the tool get stuck in a loop, fix it (Jiri Olsa)

Infrastructure:

- Support PERF_RECORD_SWITCH in the python binding (Arnaldo Carvalho de Melo)

- Fix handling read() result using a signed variable, found with Coccinelle
  (Andrzej Hajda)

Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>

----------------------------------------------------------------
Andrzej Hajda (1):
      perf tools: Fix handling read result using a signed variable

Arnaldo Carvalho de Melo (1):
      perf python: Support the PERF_RECORD_SWITCH event

Jiri Olsa (3):
      perf tools: Get rid of superfluos call to reset_dimensions
      perf tools: Introduce hpp_dimension__add_output function
      perf tools: Use hpp_dimension__add_output to register hpp columns

 tools/perf/python/twatch.py | 23 +++++++++++++++---
 tools/perf/ui/hist.c        | 16 ++++++------
 tools/perf/util/event.c     |  3 ++-
 tools/perf/util/python.c    | 59 ++++++++++++++++++++++++++++++++++++++++++---
 tools/perf/util/sort.c      |  8 ++++--
 tools/perf/util/sort.h      |  2 ++
 6 files changed, 93 insertions(+), 18 deletions(-)

[PATCH 1/5] perf tools: Get rid of superfluos call to reset_dimensions

[Arnaldo Carvalho de Melo]

From: Jiri Olsa <jolsa@kernel.org>

There's no need to call reset_dimensions within __setup_output_field
function. It's already called in its caller setup_sorting right before
perf_hpp__init, which will be changed in following patch to respect
taken dimension.

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1444134312-29136-2-git-send-email-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/sort.c | 2 --
 1 file changed, 2 deletions(-)

diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c
index ee94b728fca4..8521e3c89b8d 100644
--- a/tools/perf/util/sort.c
+++ b/tools/perf/util/sort.c
@@ -1917,8 +1917,6 @@ static int __setup_output_field(void)
 	if (field_order == NULL)
 		return 0;
 
-	reset_dimensions();
-
 	strp = str = strdup(field_order);
 	if (str == NULL) {
 		error("Not enough memory to setup output fields");
-- 
2.1.0

[PATCH 2/5] perf tools: Introduce hpp_dimension__add_output function

[Arnaldo Carvalho de Melo]

From: Jiri Olsa <jolsa@kernel.org>

This function will allow to register output column from ui code and
respect taken sort/output dimensions.

Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1444134312-29136-3-git-send-email-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/sort.c | 6 ++++++
 tools/perf/util/sort.h | 2 ++
 2 files changed, 8 insertions(+)

diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c
index 8521e3c89b8d..2d8ccd4d9e1b 100644
--- a/tools/perf/util/sort.c
+++ b/tools/perf/util/sort.c
@@ -1577,6 +1577,12 @@ static int __hpp_dimension__add_output(struct hpp_dimension *hd)
 	return 0;
 }
 
+int hpp_dimension__add_output(unsigned col)
+{
+	BUG_ON(col >= PERF_HPP__MAX_INDEX);
+	return __hpp_dimension__add_output(&hpp_sort_dimensions[col]);
+}
+
 int sort_dimension__add(const char *tok)
 {
 	unsigned int i;
diff --git a/tools/perf/util/sort.h b/tools/perf/util/sort.h
index 33b3d30e18d3..31228851e397 100644
--- a/tools/perf/util/sort.h
+++ b/tools/perf/util/sort.h
@@ -234,4 +234,6 @@ void perf_hpp__set_elide(int idx, bool elide);
 int report_parse_ignore_callees_opt(const struct option *opt, const char *arg, int unset);
 
 bool is_strict_order(const char *order);
+
+int hpp_dimension__add_output(unsigned col);
 #endif	/* __PERF_SORT_H */
-- 
2.1.0

[PATCH 3/5] perf tools: Use hpp_dimension__add_output to register hpp columns

[Arnaldo Carvalho de Melo]

From: Jiri Olsa <jolsa@kernel.org>

The perf_hpp__init currently does not respect sorting dimensions and the
setup_sorting function could endup queueing same format twice. That
screwed up the perf_hpp__list and got stuck in loop within
perf_hpp__setup_output_field function.

  $ perf report -F +overhead

  0x00000000004c1355 in perf_hpp__is_sort_entry (format=format@entry=0x880440 <perf_hpp.format>) at util/sort.c:1506
  1506    {

     #0  0x00000000004c1355 in perf_hpp__is_sort_entry (format=format@entry=0x880440 <perf_hpp.format>) at util/sort.c:1506
     #1  0x00000000004c139d in perf_hpp__same_sort_entry (a=a@entry=0x880440 <perf_hpp.format>, b=b@entry=0x2bb2fe0) at util/sort.c:1380
     #2  0x00000000004f8d3c in perf_hpp__setup_output_field () at ui/hist.c:554
     #3  0x00000000004c1d1e in setup_sorting () at util/sort.c:1984
     #4  0x000000000042efbf in cmd_report (argc=0, argv=0x7ffea5a0e790, prefix=<optimized out>) at builtin-report.c:874
     #5  0x0000000000476f13 in run_builtin (p=p@entry=0x875628 <commands+168>, argc=argc@entry=3, argv=argv@entry=0x7ffea5a0e790) at perf.c:385
     #6  0x000000000047710b in handle_internal_command (argc=3, argv=0x7ffea5a0e790) at perf.c:445
     #7  0x0000000000477176 in run_argv (argcp=argcp@entry=0x7ffea5a0e5fc, argv=argv@entry=0x7ffea5a0e5f0) at perf.c:489
     #8  0x00000000004773e7 in main (argc=3, argv=0x7ffea5a0e790) at perf.c:606

Using hpp_dimension__add_output function to register the output column.
It will also mark the dimension as taken and omit above stuck.

Reported-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Link: http://lkml.kernel.org/r/1444134312-29136-4-git-send-email-jolsa@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/ui/hist.c | 16 ++++++++--------
 1 file changed, 8 insertions(+), 8 deletions(-)

diff --git a/tools/perf/ui/hist.c b/tools/perf/ui/hist.c
index 25d608394d74..5029ba2b55af 100644
--- a/tools/perf/ui/hist.c
+++ b/tools/perf/ui/hist.c
@@ -463,27 +463,27 @@ void perf_hpp__init(void)
 		return;
 
 	if (symbol_conf.cumulate_callchain) {
-		perf_hpp__column_enable(PERF_HPP__OVERHEAD_ACC);
+		hpp_dimension__add_output(PERF_HPP__OVERHEAD_ACC);
 		perf_hpp__format[PERF_HPP__OVERHEAD].name = "Self";
 	}
 
-	perf_hpp__column_enable(PERF_HPP__OVERHEAD);
+	hpp_dimension__add_output(PERF_HPP__OVERHEAD);
 
 	if (symbol_conf.show_cpu_utilization) {
-		perf_hpp__column_enable(PERF_HPP__OVERHEAD_SYS);
-		perf_hpp__column_enable(PERF_HPP__OVERHEAD_US);
+		hpp_dimension__add_output(PERF_HPP__OVERHEAD_SYS);
+		hpp_dimension__add_output(PERF_HPP__OVERHEAD_US);
 
 		if (perf_guest) {
-			perf_hpp__column_enable(PERF_HPP__OVERHEAD_GUEST_SYS);
-			perf_hpp__column_enable(PERF_HPP__OVERHEAD_GUEST_US);
+			hpp_dimension__add_output(PERF_HPP__OVERHEAD_GUEST_SYS);
+			hpp_dimension__add_output(PERF_HPP__OVERHEAD_GUEST_US);
 		}
 	}
 
 	if (symbol_conf.show_nr_samples)
-		perf_hpp__column_enable(PERF_HPP__SAMPLES);
+		hpp_dimension__add_output(PERF_HPP__SAMPLES);
 
 	if (symbol_conf.show_total_period)
-		perf_hpp__column_enable(PERF_HPP__PERIOD);
+		hpp_dimension__add_output(PERF_HPP__PERIOD);
 
 	/* prepend overhead field for backward compatiblity.  */
 	list = &perf_hpp__format[PERF_HPP__OVERHEAD].sort_list;
-- 
2.1.0

[PATCH 4/5] perf tools: Fix handling read result using a signed variable

[Arnaldo Carvalho de Melo]

From: Andrzej Hajda <a.hajda@samsung.com>

The function can return negative value, assigning it to unsigned
variable can cause memory corruption.

The problem has been detected using proposed semantic patch
scripts/coccinelle/tests/unsigned_lesser_than_zero.cocci [1].

[1]: http://permalink.gmane.org/gmane.linux.kernel/2038576

Signed-off-by: Andrzej Hajda <a.hajda@samsung.com>
Cc: Bartlomiej Zolnierkiewicz <b.zolnierkie@samsung.com>
Cc: Marek Szyprowski <m.szyprowski@samsung.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: kernel-janitors@vger.kernel.org
Link: http://lkml.kernel.org/r/1444122017-16856-1-git-send-email-a.hajda@samsung.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/util/event.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c
index cb98b5af9e17..8b10621b415c 100644
--- a/tools/perf/util/event.c
+++ b/tools/perf/util/event.c
@@ -67,7 +67,8 @@ static int perf_event__get_comm_ids(pid_t pid, char *comm, size_t len,
 	char filename[PATH_MAX];
 	char bf[4096];
 	int fd;
-	size_t size = 0, n;
+	size_t size = 0;
+	ssize_t n;
 	char *nl, *name, *tgids, *ppids;
 
 	*tgid = -1;
-- 
2.1.0

[PATCH 5/5] perf python: Support the PERF_RECORD_SWITCH

[Arnaldo Carvalho de Melo]

From: Arnaldo Carvalho de Melo <acme@redhat.com>

To test it check tools/perf/python/twatch.py, after following the
instructions there to enable context_switch, output looks like:

  [root@zoo linux]# tools/perf/python/twatch.py
  cpu: 1, pid: 31463, tid: 31463 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31463, switch_out: 0 }
  cpu: 2, pid: 31463, tid: 31496 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31496, switch_out: 0 }
  cpu: 2, pid: 31463, tid: 31496 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31496, switch_out: 1 }
  cpu: 3, pid: 31463, tid: 31527 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31527, switch_out: 0 }
  cpu: 1, pid: 31463, tid: 31463 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31463, switch_out: 1 }
  cpu: 3, pid: 31463, tid: 31527 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31527, switch_out: 1 }
  cpu: 1, pid: 31463, tid: 31463 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31463, switch_out: 0 }
  ^CTraceback (most recent call last):
    File "tools/perf/python/twatch.py", line 67, in <module>
      main(context_switch = 1, thread = 31463)
    File "tools/perf/python/twatch.py", line 40, in main
      evlist.poll(timeout = -1)
  KeyboardInterrupt
  [root@zoo linux]#

Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: Borislav Petkov <bp@suse.de>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Guy Streeter <streeter@redhat.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/n/tip-1ukistmpamc5z717k80ctcp2@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/python/twatch.py | 23 +++++++++++++++---
 tools/perf/util/python.c    | 59 ++++++++++++++++++++++++++++++++++++++++++---
 2 files changed, 75 insertions(+), 7 deletions(-)

diff --git a/tools/perf/python/twatch.py b/tools/perf/python/twatch.py
index b9d508336ae6..c235c22b107a 100755
--- a/tools/perf/python/twatch.py
+++ b/tools/perf/python/twatch.py
@@ -15,14 +15,14 @@
 
 import perf
 
-def main():
+def main(context_switch = 0, thread = -1):
 	cpus = perf.cpu_map()
-	threads = perf.thread_map()
+	threads = perf.thread_map(thread)
 	evsel = perf.evsel(type	  = perf.TYPE_SOFTWARE,
 			   config = perf.COUNT_SW_DUMMY,
 			   task = 1, comm = 1, mmap = 0, freq = 0,
 			   wakeup_events = 1, watermark = 1,
-			   sample_id_all = 1,
+			   sample_id_all = 1, context_switch = context_switch,
 			   sample_type = perf.SAMPLE_PERIOD | perf.SAMPLE_TID | perf.SAMPLE_CPU)
 
 	"""What we want are just the PERF_RECORD_ lifetime events for threads,
@@ -48,4 +48,21 @@ def main():
 			print event
 
 if __name__ == '__main__':
+    """
+	To test the PERF_RECORD_SWITCH record, pick a pid and replace
+	in the following line.
+
+	Example output:
+
+cpu: 3, pid: 31463, tid: 31593 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31593, switch_out: 1 }
+cpu: 1, pid: 31463, tid: 31489 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31489, switch_out: 1 }
+cpu: 2, pid: 31463, tid: 31496 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31496, switch_out: 1 }
+cpu: 3, pid: 31463, tid: 31491 { type: context_switch, next_prev_pid: 31463, next_prev_tid: 31491, switch_out: 0 }
+
+	It is possible as well to use event.misc & perf.PERF_RECORD_MISC_SWITCH_OUT
+	to figure out if this is a context switch in or out of the monitored threads.
+
+	If bored, please add command line option parsing support for these options :-)
+    """
+    # main(context_switch = 1, thread = 31463)
     main()
diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c
index 6324fe6b161e..98f127abfa42 100644
--- a/tools/perf/util/python.c
+++ b/tools/perf/util/python.c
@@ -67,6 +67,7 @@ static char pyrf_mmap_event__doc[] = PyDoc_STR("perf mmap event object.");
 static PyMemberDef pyrf_mmap_event__members[] = {
 	sample_members
 	member_def(perf_event_header, type, T_UINT, "event type"),
+	member_def(perf_event_header, misc, T_UINT, "event misc"),
 	member_def(mmap_event, pid, T_UINT, "event pid"),
 	member_def(mmap_event, tid, T_UINT, "event tid"),
 	member_def(mmap_event, start, T_ULONGLONG, "start of the map"),
@@ -297,6 +298,43 @@ static PyTypeObject pyrf_sample_event__type = {
 	.tp_repr	= (reprfunc)pyrf_sample_event__repr,
 };
 
+static char pyrf_context_switch_event__doc[] = PyDoc_STR("perf context_switch event object.");
+
+static PyMemberDef pyrf_context_switch_event__members[] = {
+	sample_members
+	member_def(perf_event_header, type, T_UINT, "event type"),
+	member_def(context_switch_event, next_prev_pid, T_UINT, "next/prev pid"),
+	member_def(context_switch_event, next_prev_tid, T_UINT, "next/prev tid"),
+	{ .name = NULL, },
+};
+
+static PyObject *pyrf_context_switch_event__repr(struct pyrf_event *pevent)
+{
+	PyObject *ret;
+	char *s;
+
+	if (asprintf(&s, "{ type: context_switch, next_prev_pid: %u, next_prev_tid: %u, switch_out: %u }",
+		     pevent->event.context_switch.next_prev_pid,
+		     pevent->event.context_switch.next_prev_tid,
+		     !!(pevent->event.header.misc & PERF_RECORD_MISC_SWITCH_OUT)) < 0) {
+		ret = PyErr_NoMemory();
+	} else {
+		ret = PyString_FromString(s);
+		free(s);
+	}
+	return ret;
+}
+
+static PyTypeObject pyrf_context_switch_event__type = {
+	PyVarObject_HEAD_INIT(NULL, 0)
+	.tp_name	= "perf.context_switch_event",
+	.tp_basicsize	= sizeof(struct pyrf_event),
+	.tp_flags	= Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE,
+	.tp_doc		= pyrf_context_switch_event__doc,
+	.tp_members	= pyrf_context_switch_event__members,
+	.tp_repr	= (reprfunc)pyrf_context_switch_event__repr,
+};
+
 static int pyrf_event__setup_types(void)
 {
 	int err;
@@ -306,6 +344,7 @@ static int pyrf_event__setup_types(void)
 	pyrf_lost_event__type.tp_new =
 	pyrf_read_event__type.tp_new =
 	pyrf_sample_event__type.tp_new =
+	pyrf_context_switch_event__type.tp_new =
 	pyrf_throttle_event__type.tp_new = PyType_GenericNew;
 	err = PyType_Ready(&pyrf_mmap_event__type);
 	if (err < 0)
@@ -328,6 +367,9 @@ static int pyrf_event__setup_types(void)
 	err = PyType_Ready(&pyrf_sample_event__type);
 	if (err < 0)
 		goto out;
+	err = PyType_Ready(&pyrf_context_switch_event__type);
+	if (err < 0)
+		goto out;
 out:
 	return err;
 }
@@ -342,6 +384,8 @@ static PyTypeObject *pyrf_event__type[] = {
 	[PERF_RECORD_FORK]	 = &pyrf_task_event__type,
 	[PERF_RECORD_READ]	 = &pyrf_read_event__type,
 	[PERF_RECORD_SAMPLE]	 = &pyrf_sample_event__type,
+	[PERF_RECORD_SWITCH]	 = &pyrf_context_switch_event__type,
+	[PERF_RECORD_SWITCH_CPU_WIDE]  = &pyrf_context_switch_event__type,
 };
 
 static PyObject *pyrf_event__new(union perf_event *event)
@@ -349,8 +393,10 @@ static PyObject *pyrf_event__new(union perf_event *event)
 	struct pyrf_event *pevent;
 	PyTypeObject *ptype;
 
-	if (event->header.type < PERF_RECORD_MMAP ||
-	    event->header.type > PERF_RECORD_SAMPLE)
+	if ((event->header.type < PERF_RECORD_MMAP ||
+	     event->header.type > PERF_RECORD_SAMPLE) &&
+	    !(event->header.type == PERF_RECORD_SWITCH ||
+	      event->header.type == PERF_RECORD_SWITCH_CPU_WIDE))
 		return NULL;
 
 	ptype = pyrf_event__type[event->header.type];
@@ -528,6 +574,7 @@ static int pyrf_evsel__init(struct pyrf_evsel *pevsel,
 		"exclude_hv",
 		"exclude_idle",
 		"mmap",
+		"context_switch",
 		"comm",
 		"freq",
 		"inherit_stat",
@@ -553,6 +600,7 @@ static int pyrf_evsel__init(struct pyrf_evsel *pevsel,
 	    exclude_hv = 0,
 	    exclude_idle = 0,
 	    mmap = 0,
+	    context_switch = 0,
 	    comm = 0,
 	    freq = 1,
 	    inherit_stat = 0,
@@ -565,13 +613,13 @@ static int pyrf_evsel__init(struct pyrf_evsel *pevsel,
 	int idx = 0;
 
 	if (!PyArg_ParseTupleAndKeywords(args, kwargs,
-					 "|iKiKKiiiiiiiiiiiiiiiiiiiiiKK", kwlist,
+					 "|iKiKKiiiiiiiiiiiiiiiiiiiiiiKK", kwlist,
 					 &attr.type, &attr.config, &attr.sample_freq,
 					 &sample_period, &attr.sample_type,
 					 &attr.read_format, &disabled, &inherit,
 					 &pinned, &exclusive, &exclude_user,
 					 &exclude_kernel, &exclude_hv, &exclude_idle,
-					 &mmap, &comm, &freq, &inherit_stat,
+					 &mmap, &context_switch, &comm, &freq, &inherit_stat,
 					 &enable_on_exec, &task, &watermark,
 					 &precise_ip, &mmap_data, &sample_id_all,
 					 &attr.wakeup_events, &attr.bp_type,
@@ -595,6 +643,7 @@ static int pyrf_evsel__init(struct pyrf_evsel *pevsel,
 	attr.exclude_hv	    = exclude_hv;
 	attr.exclude_idle   = exclude_idle;
 	attr.mmap	    = mmap;
+	attr.context_switch = context_switch;
 	attr.comm	    = comm;
 	attr.freq	    = freq;
 	attr.inherit_stat   = inherit_stat;
@@ -1019,6 +1068,8 @@ static struct {
 	PERF_CONST(RECORD_LOST_SAMPLES),
 	PERF_CONST(RECORD_SWITCH),
 	PERF_CONST(RECORD_SWITCH_CPU_WIDE),
+
+	PERF_CONST(RECORD_MISC_SWITCH_OUT),
 	{ .name = NULL, },
 };
 
-- 
2.1.0