Re: [PATCH] perf: implement recording/reporting per-cpu samples

[Arun Sharma <aruns@google.com>]

On Tue, May 04, 2010 at 11:16:38AM +0200, Peter Zijlstra wrote:
> > In a shared multi-core environment, users want to analyze why their
> > program was slow. In particular, if the code ran slower only on
> > certain CPUs due to interference from other programs or kernel
> > threads, they want to know that.
> 
> But for that you use perf record -a, right? So you record all cpus
> allways -- otherwise there is no telling what was happening to make it
> go slow.

The updated patch records the CPU only in the system_wide mode.

> 
> > But that's just our use case. The patch is mostly about --sort cpu
> > option. If you want to drop the part that enables PERF_SAMPLE_CPU by
> > default, that's fine by me.
> 
> Right, it would be very nice if we can avoid growing the default sample
> size. Also, your changelog needs work, please explain the full usecase
> that goes with this feature.
> 
> Explain the thing you're wanting to measure, explain the implementation
> and maybe give a short example.

Updated changelog as well.

 -Arun

>From 7ae0f1cd7d6ab0d74ab3f8a8a31b11d1cd416f36 Mon Sep 17 00:00:00 2001
From: Arun Sharma <aruns@google.com>
Date: Mon, 3 May 2010 11:45:41 -0700
Subject: [PATCH] Implement --sort cpu

In a shared multi-core environment, users want to analyze why their
program was slow. In particular, if the code ran slower only on
certain CPUs due to interference from other programs or kernel
threads, the user should be able to notice that.

Sample usage:

perf record -f -a -- sleep 3
perf report --sort cpu,comm

Workload:

program is running on 16 CPUs
Experiencing interference from an antagonist only on 4 CPUs.

  Samples: 106218177676 cycles

  Overhead  CPU          Command
  ........  ...  ...............

     6.25%  2            program
     6.24%  6            program
     6.24%  11           program
     6.24%  5            program
     6.24%  9            program
     6.24%  10           program
     6.23%  15           program
     6.23%  7            program
     6.23%  3            program
     6.23%  14           program
     6.22%  1            program
     6.20%  13           program
     3.17%  12           program
     3.15%  8            program
     3.14%  0            program
     3.13%  4            program
     3.11%  4         antagonist
     3.11%  0         antagonist
     3.10%  8         antagonist
     3.07%  12        antagonist

Signed-off-by: Arun Sharma <aruns@google.com>
---
 tools/perf/builtin-record.c |    3 +++
 tools/perf/builtin-report.c |    7 +++++++
 tools/perf/util/hist.c      |    1 +
 tools/perf/util/sort.c      |   27 +++++++++++++++++++++++++++
 tools/perf/util/sort.h      |    8 ++++++--
 tools/perf/util/symbol.h    |    1 +
 6 files changed, 45 insertions(+), 2 deletions(-)

diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 27f992a..66867b8 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -290,6 +290,9 @@ static void create_counter(int counter, int cpu)
 	if (call_graph)
 		attr->sample_type	|= PERF_SAMPLE_CALLCHAIN;
 
+	if (system_wide)
+		attr->sample_type	|= PERF_SAMPLE_CPU;
+
 	if (raw_samples) {
 		attr->sample_type	|= PERF_SAMPLE_TIME;
 		attr->sample_type	|= PERF_SAMPLE_RAW;
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 816edae..6014d70 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -190,6 +190,13 @@ static int process_sample_event(event_t *event, struct perf_session *session)
 		return -1;
 	}
 
+	if (session->sample_type & PERF_SAMPLE_CPU) {
+		dump_printf("... cpu: %d\n", data.cpu);
+		al.cpu = data.cpu;
+	} else {
+		al.cpu = -1;
+	}
+
 	if (al.filtered || (hide_unresolved && al.sym == NULL))
 		return 0;
 
diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
index ad6b22d..4640015 100644
--- a/tools/perf/util/hist.c
+++ b/tools/perf/util/hist.c
@@ -52,6 +52,7 @@ struct hist_entry *__perf_session__add_hist_entry(struct rb_root *hists,
 		},
 		.ip	= al->addr,
 		.level	= al->level,
+		.cpu    = al->cpu,
 		.count	= count,
 		.parent = sym_parent,
 	};
diff --git a/tools/perf/util/sort.c b/tools/perf/util/sort.c
index da30b30..ee70bb8 100644
--- a/tools/perf/util/sort.c
+++ b/tools/perf/util/sort.c
@@ -13,6 +13,7 @@ enum sort_type	sort__first_dimension;
 unsigned int dsos__col_width;
 unsigned int comms__col_width;
 unsigned int threads__col_width;
+unsigned int cpu__col_width;
 static unsigned int parent_symbol__col_width;
 char * field_sep;
 
@@ -28,6 +29,8 @@ static int hist_entry__sym_snprintf(struct hist_entry *self, char *bf,
 				    size_t size, unsigned int width);
 static int hist_entry__parent_snprintf(struct hist_entry *self, char *bf,
 				       size_t size, unsigned int width);
+static int hist_entry__cpu_snprintf(struct hist_entry *self, char *bf,
+				    size_t size, unsigned int width);
 
 struct sort_entry sort_thread = {
 	.se_header	= "Command:  Pid",
@@ -64,6 +67,13 @@ struct sort_entry sort_parent = {
 	.se_width	= &parent_symbol__col_width,
 };
 
+struct sort_entry sort_cpu = {
+	.se_header      = "CPU",
+	.se_cmp	        = sort__cpu_cmp,
+	.se_snprintf    = hist_entry__cpu_snprintf,
+	.se_width	= &cpu__col_width,
+};
+
 struct sort_dimension {
 	const char		*name;
 	struct sort_entry	*entry;
@@ -76,6 +86,7 @@ static struct sort_dimension sort_dimensions[] = {
 	{ .name = "dso",	.entry = &sort_dso,	},
 	{ .name = "symbol",	.entry = &sort_sym,	},
 	{ .name = "parent",	.entry = &sort_parent,	},
+	{ .name = "cpu",	.entry = &sort_cpu,	},
 };
 
 int64_t cmp_null(void *l, void *r)
@@ -242,6 +253,20 @@ static int hist_entry__parent_snprintf(struct hist_entry *self, char *bf,
 			      self->parent ? self->parent->name : "[other]");
 }
 
+/* --sort cpu */
+
+int64_t
+sort__cpu_cmp(struct hist_entry *left, struct hist_entry *right)
+{
+	return right->cpu - left->cpu;
+}
+
+static int hist_entry__cpu_snprintf(struct hist_entry *self, char *bf,
+				       size_t size, unsigned int width)
+{
+	return repsep_snprintf(bf, size, "%-*d", width, self->cpu);
+}
+
 int sort_dimension__add(const char *tok)
 {
 	unsigned int i;
@@ -281,6 +306,8 @@ int sort_dimension__add(const char *tok)
 				sort__first_dimension = SORT_SYM;
 			else if (!strcmp(sd->name, "parent"))
 				sort__first_dimension = SORT_PARENT;
+			else if (!strcmp(sd->name, "cpu"))
+				sort__first_dimension = SORT_CPU;
 		}
 
 		list_add_tail(&sd->entry->list, &hist_entry__sort_list);
diff --git a/tools/perf/util/sort.h b/tools/perf/util/sort.h
index b7c54ee..82c5596 100644
--- a/tools/perf/util/sort.h
+++ b/tools/perf/util/sort.h
@@ -60,6 +60,7 @@ struct hist_entry {
 	char			level;
 	u8			filtered;
 	struct symbol		*parent;
+	s32                     cpu;
 	union {
 		unsigned long	  position;
 		struct hist_entry *pair;
@@ -73,7 +74,8 @@ enum sort_type {
 	SORT_COMM,
 	SORT_DSO,
 	SORT_SYM,
-	SORT_PARENT
+	SORT_PARENT,
+	SORT_CPU
 };
 
 /*
@@ -102,6 +104,8 @@ extern size_t sort__thread_print(FILE *, struct hist_entry *, unsigned int);
 extern size_t sort__comm_print(FILE *, struct hist_entry *, unsigned int);
 extern size_t sort__dso_print(FILE *, struct hist_entry *, unsigned int);
 extern size_t sort__sym_print(FILE *, struct hist_entry *, unsigned int __used);
+extern size_t sort__parent_print(FILE *, struct hist_entry *, unsigned int);
+extern size_t sort__cpu_print(FILE *, struct hist_entry *, unsigned int);
 extern int64_t cmp_null(void *, void *);
 extern int64_t sort__thread_cmp(struct hist_entry *, struct hist_entry *);
 extern int64_t sort__comm_cmp(struct hist_entry *, struct hist_entry *);
@@ -109,7 +113,7 @@ extern int64_t sort__comm_collapse(struct hist_entry *, struct hist_entry *);
 extern int64_t sort__dso_cmp(struct hist_entry *, struct hist_entry *);
 extern int64_t sort__sym_cmp(struct hist_entry *, struct hist_entry *);
 extern int64_t sort__parent_cmp(struct hist_entry *, struct hist_entry *);
-extern size_t sort__parent_print(FILE *, struct hist_entry *, unsigned int);
+extern int64_t sort__cpu_cmp(struct hist_entry *, struct hist_entry *);
 extern int sort_dimension__add(const char *);
 void sort_entry__setup_elide(struct sort_entry *self, struct strlist *list,
 			     const char *list_name, FILE *fp);
diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h
index 478f5ab..4b3e09d 100644
--- a/tools/perf/util/symbol.h
+++ b/tools/perf/util/symbol.h
@@ -112,6 +112,7 @@ struct addr_location {
 	char	      level;
 	bool	      filtered;
 	unsigned int  cpumode;
+	s32           cpu;
 };
 
 enum dso_kernel_type {
-- 
1.7.0.1