From 342955593af08f185279e074b3d77719d2f23b82 Mon Sep 17 00:00:00 2001
From: Shawn Bohrer <shawn.bohrer@gmail.com>
Date: Tue, 30 Nov 2010 19:57:11 -0600
Subject: perf diff: Fix displacement and modules options short flag

The --displacement and --modules options to perf diff both use -m as a
short flag.  Change --displacement to use -M since other perf commands
use -m, --modules.

Cc: Ingo Molnar <mingo@elte.hu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
LKML-Reference: <1291168642-11402-4-git-send-email-shawn.bohrer@gmail.com>
Signed-off-by: Shawn Bohrer <shawn.bohrer@gmail.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/builtin-diff.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

(limited to 'tools/perf/builtin-diff.c')

diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index fca1d440291..221b823bc26 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/builtin-diff.c
@@ -173,7 +173,7 @@ static const char * const diff_usage[] = {
 static const struct option options[] = {
 	OPT_INCR('v', "verbose", &verbose,
 		    "be more verbose (show symbol address, etc)"),
-	OPT_BOOLEAN('m', "displacement", &show_displacement,
+	OPT_BOOLEAN('M', "displacement", &show_displacement,
 		    "Show position displacement relative to baseline"),
 	OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
 		    "dump raw trace in ASCII"),
-- 
cgit v1.2.3-70-g09d2


From 640c03ce837fe8d4b56342aba376ea0da3960459 Mon Sep 17 00:00:00 2001
From: Arnaldo Carvalho de Melo <acme@redhat.com>
Date: Thu, 2 Dec 2010 14:10:21 -0200
Subject: perf session: Parse sample earlier
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

At perf_session__process_event, so that we reduce the number of lines in eache
tool sample processing routine that now receives a sample_data pointer already
parsed.

This will also be useful in the next patch, where we'll allow sample the
identity fields in MMAP, FORK, EXIT, etc, when it will be possible to see (cpu,
timestamp) just after before every event.

Also validate callchains in perf_session__process_event, i.e. as early as
possible, and keep a counter of the number of events discarded due to invalid
callchains, warning the user about it if it happens.

There is an assumption that was kept that all events have the same sample_type,
that will be dealt with in the future, when this preexisting limitation will be
removed.

Tested-by: Thomas Gleixner <tglx@linutronix.de>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Acked-by: Ian Munsie <imunsie@au1.ibm.com>
Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Ian Munsie <imunsie@au1.ibm.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Stephane Eranian <eranian@google.com>
LKML-Reference: <1291318772-30880-4-git-send-email-acme@infradead.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/builtin-annotate.c  |   6 +--
 tools/perf/builtin-diff.c      |  11 +++--
 tools/perf/builtin-inject.c    |  39 ++++++++++------
 tools/perf/builtin-kmem.c      |  21 ++-------
 tools/perf/builtin-lock.c      |  12 ++---
 tools/perf/builtin-record.c    |   7 +++
 tools/perf/builtin-report.c    |  15 +++---
 tools/perf/builtin-sched.c     |  21 +++------
 tools/perf/builtin-script.c    |  34 +++++---------
 tools/perf/builtin-timechart.c |  40 ++++++++--------
 tools/perf/builtin-top.c       |   4 +-
 tools/perf/util/build-id.c     |   7 ++-
 tools/perf/util/event.c        |  66 +++++++++++---------------
 tools/perf/util/event.h        |  20 +++++---
 tools/perf/util/header.c       |  11 ++---
 tools/perf/util/hist.h         |   2 +
 tools/perf/util/session.c      | 103 +++++++++++++++++++++++++++++++----------
 tools/perf/util/session.h      |   9 ++--
 18 files changed, 236 insertions(+), 192 deletions(-)

(limited to 'tools/perf/builtin-diff.c')

diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c
index 6d5604d8df9..569a2761b90 100644
--- a/tools/perf/builtin-annotate.c
+++ b/tools/perf/builtin-annotate.c
@@ -58,12 +58,12 @@ static int hists__add_entry(struct hists *self, struct addr_location *al)
 	return hist_entry__inc_addr_samples(he, al->addr);
 }
 
-static int process_sample_event(event_t *event, struct perf_session *session)
+static int process_sample_event(event_t *event, struct sample_data *sample,
+				struct perf_session *session)
 {
 	struct addr_location al;
-	struct sample_data data;
 
-	if (event__preprocess_sample(event, session, &al, &data, NULL) < 0) {
+	if (event__preprocess_sample(event, session, &al, sample, NULL) < 0) {
 		pr_warning("problem processing %d event, skipping it.\n",
 			   event->header.type);
 		return -1;
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index 221b823bc26..5e1a043aae0 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/builtin-diff.c
@@ -30,12 +30,13 @@ static int hists__add_entry(struct hists *self,
 	return -ENOMEM;
 }
 
-static int diff__process_sample_event(event_t *event, struct perf_session *session)
+static int diff__process_sample_event(event_t *event,
+				      struct sample_data *sample,
+				      struct perf_session *session)
 {
 	struct addr_location al;
-	struct sample_data data = { .period = 1, };
 
-	if (event__preprocess_sample(event, session, &al, &data, NULL) < 0) {
+	if (event__preprocess_sample(event, session, &al, sample, NULL) < 0) {
 		pr_warning("problem processing %d event, skipping it.\n",
 			   event->header.type);
 		return -1;
@@ -44,12 +45,12 @@ static int diff__process_sample_event(event_t *event, struct perf_session *sessi
 	if (al.filtered || al.sym == NULL)
 		return 0;
 
-	if (hists__add_entry(&session->hists, &al, data.period)) {
+	if (hists__add_entry(&session->hists, &al, sample->period)) {
 		pr_warning("problem incrementing symbol period, skipping event\n");
 		return -1;
 	}
 
-	session->hists.stats.total_period += data.period;
+	session->hists.stats.total_period += sample->period;
 	return 0;
 }
 
diff --git a/tools/perf/builtin-inject.c b/tools/perf/builtin-inject.c
index 8e3e47b064c..4b66b857941 100644
--- a/tools/perf/builtin-inject.c
+++ b/tools/perf/builtin-inject.c
@@ -16,8 +16,8 @@
 static char		const *input_name = "-";
 static bool		inject_build_ids;
 
-static int event__repipe(event_t *event __used,
-			 struct perf_session *session __used)
+static int event__repipe_synth(event_t *event,
+			       struct perf_session *session __used)
 {
 	uint32_t size;
 	void *buf = event;
@@ -36,22 +36,30 @@ static int event__repipe(event_t *event __used,
 	return 0;
 }
 
-static int event__repipe_mmap(event_t *self, struct perf_session *session)
+static int event__repipe(event_t *event, struct sample_data *sample __used,
+			 struct perf_session *session)
+{
+	return event__repipe_synth(event, session);
+}
+
+static int event__repipe_mmap(event_t *self, struct sample_data *sample,
+			      struct perf_session *session)
 {
 	int err;
 
-	err = event__process_mmap(self, session);
-	event__repipe(self, session);
+	err = event__process_mmap(self, sample, session);
+	event__repipe(self, sample, session);
 
 	return err;
 }
 
-static int event__repipe_task(event_t *self, struct perf_session *session)
+static int event__repipe_task(event_t *self, struct sample_data *sample,
+			      struct perf_session *session)
 {
 	int err;
 
-	err = event__process_task(self, session);
-	event__repipe(self, session);
+	err = event__process_task(self, sample, session);
+	event__repipe(self, sample, session);
 
 	return err;
 }
@@ -61,7 +69,7 @@ static int event__repipe_tracing_data(event_t *self,
 {
 	int err;
 
-	event__repipe(self, session);
+	event__repipe_synth(self, session);
 	err = event__process_tracing_data(self, session);
 
 	return err;
@@ -111,7 +119,8 @@ static int dso__inject_build_id(struct dso *self, struct perf_session *session)
 	return 0;
 }
 
-static int event__inject_buildid(event_t *event, struct perf_session *session)
+static int event__inject_buildid(event_t *event, struct sample_data *sample,
+				 struct perf_session *session)
 {
 	struct addr_location al;
 	struct thread *thread;
@@ -146,7 +155,7 @@ static int event__inject_buildid(event_t *event, struct perf_session *session)
 	}
 
 repipe:
-	event__repipe(event, session);
+	event__repipe(event, sample, session);
 	return 0;
 }
 
@@ -160,10 +169,10 @@ struct perf_event_ops inject_ops = {
 	.read		= event__repipe,
 	.throttle	= event__repipe,
 	.unthrottle	= event__repipe,
-	.attr		= event__repipe,
-	.event_type 	= event__repipe,
-	.tracing_data 	= event__repipe,
-	.build_id 	= event__repipe,
+	.attr		= event__repipe_synth,
+	.event_type 	= event__repipe_synth,
+	.tracing_data 	= event__repipe_synth,
+	.build_id 	= event__repipe_synth,
 };
 
 extern volatile int session_done;
diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c
index 31f60a2535e..d0a652e9c70 100644
--- a/tools/perf/builtin-kmem.c
+++ b/tools/perf/builtin-kmem.c
@@ -304,22 +304,11 @@ process_raw_event(event_t *raw_event __used, void *data,
 	}
 }
 
-static int process_sample_event(event_t *event, struct perf_session *session)
+static int process_sample_event(event_t *event, struct sample_data *sample,
+				struct perf_session *session)
 {
-	struct sample_data data;
-	struct thread *thread;
+	struct thread *thread = perf_session__findnew(session, event->ip.pid);
 
-	memset(&data, 0, sizeof(data));
-	data.time = -1;
-	data.cpu = -1;
-	data.period = 1;
-
-	event__parse_sample(event, session->sample_type, &data);
-
-	dump_printf("(IP, %d): %d/%d: %#Lx period: %Ld\n", event->header.misc,
-		    data.pid, data.tid, data.ip, data.period);
-
-	thread = perf_session__findnew(session, event->ip.pid);
 	if (thread == NULL) {
 		pr_debug("problem processing %d event, skipping it.\n",
 			 event->header.type);
@@ -328,8 +317,8 @@ static int process_sample_event(event_t *event, struct perf_session *session)
 
 	dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid);
 
-	process_raw_event(event, data.raw_data, data.cpu,
-			  data.time, thread);
+	process_raw_event(event, sample->raw_data, sample->cpu,
+			  sample->time, thread);
 
 	return 0;
 }
diff --git a/tools/perf/builtin-lock.c b/tools/perf/builtin-lock.c
index 8452a2ae219..92d3da52a5c 100644
--- a/tools/perf/builtin-lock.c
+++ b/tools/perf/builtin-lock.c
@@ -834,22 +834,18 @@ static void dump_info(void)
 		die("Unknown type of information\n");
 }
 
-static int process_sample_event(event_t *self, struct perf_session *s)
+static int process_sample_event(event_t *self, struct sample_data *sample,
+				struct perf_session *s)
 {
-	struct sample_data data;
-	struct thread *thread;
+	struct thread *thread = perf_session__findnew(s, sample->tid);
 
-	bzero(&data, sizeof(data));
-	event__parse_sample(self, s->sample_type, &data);
-
-	thread = perf_session__findnew(s, data.tid);
 	if (thread == NULL) {
 		pr_debug("problem processing %d event, skipping it.\n",
 			self->header.type);
 		return -1;
 	}
 
-	process_raw_event(data.raw_data, data.cpu, data.time, thread);
+	process_raw_event(sample->raw_data, sample->cpu, sample->time, thread);
 
 	return 0;
 }
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 024e1441d76..b34de9291c2 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -36,6 +36,7 @@ static int			*fd[MAX_NR_CPUS][MAX_COUNTERS];
 
 static u64			user_interval			= ULLONG_MAX;
 static u64			default_interval		=      0;
+static u64			sample_type;
 
 static int			nr_cpus				=      0;
 static unsigned int		page_size;
@@ -129,6 +130,7 @@ static void write_output(void *buf, size_t size)
 }
 
 static int process_synthesized_event(event_t *event,
+				     struct sample_data *sample __used,
 				     struct perf_session *self __used)
 {
 	write_output(event, event->header.size);
@@ -287,6 +289,9 @@ static void create_counter(int counter, int cpu)
 		attr->sample_type	|= PERF_SAMPLE_CPU;
 	}
 
+	if (!sample_type)
+		sample_type = attr->sample_type;
+
 	attr->mmap		= track;
 	attr->comm		= track;
 	attr->inherit		= !no_inherit;
@@ -642,6 +647,8 @@ static int __cmd_record(int argc, const char **argv)
 			open_counters(cpumap[i]);
 	}
 
+	perf_session__set_sample_type(session, sample_type);
+
 	if (pipe_output) {
 		err = perf_header__write_pipe(output);
 		if (err < 0)
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 5de405d4523..904519fba43 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -150,13 +150,13 @@ static int add_event_total(struct perf_session *session,
 	return 0;
 }
 
-static int process_sample_event(event_t *event, struct perf_session *session)
+static int process_sample_event(event_t *event, struct sample_data *sample,
+				struct perf_session *session)
 {
-	struct sample_data data = { .period = 1, };
 	struct addr_location al;
 	struct perf_event_attr *attr;
 
-	if (event__preprocess_sample(event, session, &al, &data, NULL) < 0) {
+	if (event__preprocess_sample(event, session, &al, sample, NULL) < 0) {
 		fprintf(stderr, "problem processing %d event, skipping it.\n",
 			event->header.type);
 		return -1;
@@ -165,14 +165,14 @@ static int process_sample_event(event_t *event, struct perf_session *session)
 	if (al.filtered || (hide_unresolved && al.sym == NULL))
 		return 0;
 
-	if (perf_session__add_hist_entry(session, &al, &data)) {
+	if (perf_session__add_hist_entry(session, &al, sample)) {
 		pr_debug("problem incrementing symbol period, skipping event\n");
 		return -1;
 	}
 
-	attr = perf_header__find_attr(data.id, &session->header);
+	attr = perf_header__find_attr(sample->id, &session->header);
 
-	if (add_event_total(session, &data, attr)) {
+	if (add_event_total(session, sample, attr)) {
 		pr_debug("problem adding event period\n");
 		return -1;
 	}
@@ -180,7 +180,8 @@ static int process_sample_event(event_t *event, struct perf_session *session)
 	return 0;
 }
 
-static int process_read_event(event_t *event, struct perf_session *session __used)
+static int process_read_event(event_t *event, struct sample_data *sample __used,
+			      struct perf_session *session __used)
 {
 	struct perf_event_attr *attr;
 
diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
index 26523c93979..73d1e30b0fe 100644
--- a/tools/perf/builtin-sched.c
+++ b/tools/perf/builtin-sched.c
@@ -1606,25 +1606,15 @@ process_raw_event(event_t *raw_event __used, struct perf_session *session,
 		process_sched_migrate_task_event(data, session, event, cpu, timestamp, thread);
 }
 
-static int process_sample_event(event_t *event, struct perf_session *session)
+static int process_sample_event(event_t *event, struct sample_data *sample,
+				struct perf_session *session)
 {
-	struct sample_data data;
 	struct thread *thread;
 
 	if (!(session->sample_type & PERF_SAMPLE_RAW))
 		return 0;
 
-	memset(&data, 0, sizeof(data));
-	data.time = -1;
-	data.cpu = -1;
-	data.period = -1;
-
-	event__parse_sample(event, session->sample_type, &data);
-
-	dump_printf("(IP, %d): %d/%d: %#Lx period: %Ld\n", event->header.misc,
-		    data.pid, data.tid, data.ip, data.period);
-
-	thread = perf_session__findnew(session, data.pid);
+	thread = perf_session__findnew(session, sample->pid);
 	if (thread == NULL) {
 		pr_debug("problem processing %d event, skipping it.\n",
 			 event->header.type);
@@ -1633,10 +1623,11 @@ static int process_sample_event(event_t *event, struct perf_session *session)
 
 	dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid);
 
-	if (profile_cpu != -1 && profile_cpu != (int)data.cpu)
+	if (profile_cpu != -1 && profile_cpu != (int)sample->cpu)
 		return 0;
 
-	process_raw_event(event, session, data.raw_data, data.cpu, data.time, thread);
+	process_raw_event(event, session, sample->raw_data, sample->cpu,
+			  sample->time, thread);
 
 	return 0;
 }
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 4539551ab40..683a30572cc 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -63,22 +63,11 @@ static int cleanup_scripting(void)
 
 static char const		*input_name = "perf.data";
 
-static int process_sample_event(event_t *event, struct perf_session *session)
+static int process_sample_event(event_t *event, struct sample_data *sample,
+				struct perf_session *session)
 {
-	struct sample_data data;
-	struct thread *thread;
+	struct thread *thread = perf_session__findnew(session, event->ip.pid);
 
-	memset(&data, 0, sizeof(data));
-	data.time = -1;
-	data.cpu = -1;
-	data.period = 1;
-
-	event__parse_sample(event, session->sample_type, &data);
-
-	dump_printf("(IP, %d): %d/%d: %#Lx period: %Ld\n", event->header.misc,
-		    data.pid, data.tid, data.ip, data.period);
-
-	thread = perf_session__findnew(session, event->ip.pid);
 	if (thread == NULL) {
 		pr_debug("problem processing %d event, skipping it.\n",
 			 event->header.type);
@@ -87,13 +76,13 @@ static int process_sample_event(event_t *event, struct perf_session *session)
 
 	if (session->sample_type & PERF_SAMPLE_RAW) {
 		if (debug_mode) {
-			if (data.time < last_timestamp) {
+			if (sample->time < last_timestamp) {
 				pr_err("Samples misordered, previous: %llu "
 					"this: %llu\n", last_timestamp,
-					data.time);
+					sample->time);
 				nr_unordered++;
 			}
-			last_timestamp = data.time;
+			last_timestamp = sample->time;
 			return 0;
 		}
 		/*
@@ -101,18 +90,19 @@ static int process_sample_event(event_t *event, struct perf_session *session)
 		 * field, although it should be the same than this perf
 		 * event pid
 		 */
-		scripting_ops->process_event(data.cpu, data.raw_data,
-					     data.raw_size,
-					     data.time, thread->comm);
+		scripting_ops->process_event(sample->cpu, sample->raw_data,
+					     sample->raw_size,
+					     sample->time, thread->comm);
 	}
 
-	session->hists.stats.total_period += data.period;
+	session->hists.stats.total_period += sample->period;
 	return 0;
 }
 
 static u64 nr_lost;
 
-static int process_lost_event(event_t *event, struct perf_session *session __used)
+static int process_lost_event(event_t *event, struct sample_data *sample __used,
+			      struct perf_session *session __used)
 {
 	nr_lost += event->lost.lost;
 
diff --git a/tools/perf/builtin-timechart.c b/tools/perf/builtin-timechart.c
index 9bcc38f0b70..1f158dc0334 100644
--- a/tools/perf/builtin-timechart.c
+++ b/tools/perf/builtin-timechart.c
@@ -272,19 +272,22 @@ static int cpus_cstate_state[MAX_CPUS];
 static u64 cpus_pstate_start_times[MAX_CPUS];
 static u64 cpus_pstate_state[MAX_CPUS];
 
-static int process_comm_event(event_t *event, struct perf_session *session __used)
+static int process_comm_event(event_t *event, struct sample_data *sample __used,
+			      struct perf_session *session __used)
 {
 	pid_set_comm(event->comm.tid, event->comm.comm);
 	return 0;
 }
 
-static int process_fork_event(event_t *event, struct perf_session *session __used)
+static int process_fork_event(event_t *event, struct sample_data *sample __used,
+			      struct perf_session *session __used)
 {
 	pid_fork(event->fork.pid, event->fork.ppid, event->fork.time);
 	return 0;
 }
 
-static int process_exit_event(event_t *event, struct perf_session *session __used)
+static int process_exit_event(event_t *event, struct sample_data *sample __used,
+			      struct perf_session *session __used)
 {
 	pid_exit(event->fork.pid, event->fork.time);
 	return 0;
@@ -470,24 +473,21 @@ static void sched_switch(int cpu, u64 timestamp, struct trace_entry *te)
 }
 
 
-static int process_sample_event(event_t *event, struct perf_session *session)
+static int process_sample_event(event_t *event __used,
+				struct sample_data *sample,
+				struct perf_session *session)
 {
-	struct sample_data data;
 	struct trace_entry *te;
 
-	memset(&data, 0, sizeof(data));
-
-	event__parse_sample(event, session->sample_type, &data);
-
 	if (session->sample_type & PERF_SAMPLE_TIME) {
-		if (!first_time || first_time > data.time)
-			first_time = data.time;
-		if (last_time < data.time)
-			last_time = data.time;
+		if (!first_time || first_time > sample->time)
+			first_time = sample->time;
+		if (last_time < sample->time)
+			last_time = sample->time;
 	}
 
-	te = (void *)data.raw_data;
-	if (session->sample_type & PERF_SAMPLE_RAW && data.raw_size > 0) {
+	te = (void *)sample->raw_data;
+	if (session->sample_type & PERF_SAMPLE_RAW && sample->raw_size > 0) {
 		char *event_str;
 		struct power_entry *pe;
 
@@ -499,19 +499,19 @@ static int process_sample_event(event_t *event, struct perf_session *session)
 			return 0;
 
 		if (strcmp(event_str, "power:power_start") == 0)
-			c_state_start(pe->cpu_id, data.time, pe->value);
+			c_state_start(pe->cpu_id, sample->time, pe->value);
 
 		if (strcmp(event_str, "power:power_end") == 0)
-			c_state_end(pe->cpu_id, data.time);
+			c_state_end(pe->cpu_id, sample->time);
 
 		if (strcmp(event_str, "power:power_frequency") == 0)
-			p_state_change(pe->cpu_id, data.time, pe->value);
+			p_state_change(pe->cpu_id, sample->time, pe->value);
 
 		if (strcmp(event_str, "sched:sched_wakeup") == 0)
-			sched_wakeup(data.cpu, data.time, data.pid, te);
+			sched_wakeup(sample->cpu, sample->time, sample->pid, te);
 
 		if (strcmp(event_str, "sched:sched_switch") == 0)
-			sched_switch(data.cpu, data.time, te);
+			sched_switch(sample->cpu, sample->time, te);
 	}
 	return 0;
 }
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 3d2b47d5121..5aa29e1e855 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -1025,6 +1025,8 @@ static void event__process_sample(const event_t *self,
 	if (self->header.misc & PERF_RECORD_MISC_EXACT_IP)
 		exact_samples++;
 
+	event__parse_sample(self, session->sample_type, &data);
+
 	if (event__preprocess_sample(self, session, &al, &data,
 				     symbol_filter) < 0 ||
 	    al.filtered)
@@ -1155,7 +1157,7 @@ static void perf_session__mmap_read_counter(struct perf_session *self,
 		if (event->header.type == PERF_RECORD_SAMPLE)
 			event__process_sample(event, self, md->counter);
 		else
-			event__process(event, self);
+			event__process(event, NULL, self);
 		old += size;
 	}
 
diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c
index e437edb7241..deffb8c9607 100644
--- a/tools/perf/util/build-id.c
+++ b/tools/perf/util/build-id.c
@@ -14,7 +14,9 @@
 #include <linux/kernel.h>
 #include "debug.h"
 
-static int build_id__mark_dso_hit(event_t *event, struct perf_session *session)
+static int build_id__mark_dso_hit(event_t *event,
+				  struct sample_data *sample __used,
+				  struct perf_session *session)
 {
 	struct addr_location al;
 	u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
@@ -35,7 +37,8 @@ static int build_id__mark_dso_hit(event_t *event, struct perf_session *session)
 	return 0;
 }
 
-static int event__exit_del_thread(event_t *self, struct perf_session *session)
+static int event__exit_del_thread(event_t *self, struct sample_data *sample __used,
+				  struct perf_session *session)
 {
 	struct thread *thread = perf_session__findnew(session, self->fork.tid);
 
diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c
index 7260db75b93..34510f44197 100644
--- a/tools/perf/util/event.c
+++ b/tools/perf/util/event.c
@@ -24,6 +24,15 @@ const char *event__name[] = {
 	[PERF_RECORD_HEADER_BUILD_ID]	 = "BUILD_ID",
 };
 
+static struct sample_data synth_sample = {
+	.pid	   = -1,
+	.tid	   = -1,
+	.time	   = -1,
+	.stream_id = -1,
+	.cpu	   = -1,
+	.period	   = 1,
+};
+
 static pid_t event__synthesize_comm(pid_t pid, int full,
 				    event__handler_t process,
 				    struct perf_session *session)
@@ -75,7 +84,7 @@ out_race:
 	if (!full) {
 		ev.comm.tid = pid;
 
-		process(&ev, session);
+		process(&ev, &synth_sample, session);
 		goto out_fclose;
 	}
 
@@ -93,7 +102,7 @@ out_race:
 
 		ev.comm.tid = pid;
 
-		process(&ev, session);
+		process(&ev, &synth_sample, session);
 	}
 	closedir(tasks);
 
@@ -173,7 +182,7 @@ static int event__synthesize_mmap_events(pid_t pid, pid_t tgid,
 			ev.mmap.pid = tgid;
 			ev.mmap.tid = pid;
 
-			process(&ev, session);
+			process(&ev, &synth_sample, session);
 		}
 	}
 
@@ -219,7 +228,7 @@ int event__synthesize_modules(event__handler_t process,
 
 		memcpy(ev.mmap.filename, pos->dso->long_name,
 		       pos->dso->long_name_len + 1);
-		process(&ev, session);
+		process(&ev, &synth_sample, session);
 	}
 
 	return 0;
@@ -331,7 +340,7 @@ int event__synthesize_kernel_mmap(event__handler_t process,
 	ev.mmap.len   = map->end - ev.mmap.start;
 	ev.mmap.pid   = machine->pid;
 
-	return process(&ev, session);
+	return process(&ev, &synth_sample, session);
 }
 
 static void thread__comm_adjust(struct thread *self, struct hists *hists)
@@ -361,7 +370,8 @@ static int thread__set_comm_adjust(struct thread *self, const char *comm,
 	return 0;
 }
 
-int event__process_comm(event_t *self, struct perf_session *session)
+int event__process_comm(event_t *self, struct sample_data *sample __used,
+			struct perf_session *session)
 {
 	struct thread *thread = perf_session__findnew(session, self->comm.tid);
 
@@ -376,7 +386,8 @@ int event__process_comm(event_t *self, struct perf_session *session)
 	return 0;
 }
 
-int event__process_lost(event_t *self, struct perf_session *session)
+int event__process_lost(event_t *self, struct sample_data *sample __used,
+			struct perf_session *session)
 {
 	dump_printf(": id:%Ld: lost:%Ld\n", self->lost.id, self->lost.lost);
 	session->hists.stats.total_lost += self->lost.lost;
@@ -485,7 +496,8 @@ out_problem:
 	return -1;
 }
 
-int event__process_mmap(event_t *self, struct perf_session *session)
+int event__process_mmap(event_t *self, struct sample_data *sample __used,
+			struct perf_session *session)
 {
 	struct machine *machine;
 	struct thread *thread;
@@ -526,7 +538,8 @@ out_problem:
 	return 0;
 }
 
-int event__process_task(event_t *self, struct perf_session *session)
+int event__process_task(event_t *self, struct sample_data *sample __used,
+			struct perf_session *session)
 {
 	struct thread *thread = perf_session__findnew(session, self->fork.tid);
 	struct thread *parent = perf_session__findnew(session, self->fork.ptid);
@@ -548,18 +561,19 @@ int event__process_task(event_t *self, struct perf_session *session)
 	return 0;
 }
 
-int event__process(event_t *event, struct perf_session *session)
+int event__process(event_t *event, struct sample_data *sample,
+		   struct perf_session *session)
 {
 	switch (event->header.type) {
 	case PERF_RECORD_COMM:
-		event__process_comm(event, session);
+		event__process_comm(event, sample, session);
 		break;
 	case PERF_RECORD_MMAP:
-		event__process_mmap(event, session);
+		event__process_mmap(event, sample, session);
 		break;
 	case PERF_RECORD_FORK:
 	case PERF_RECORD_EXIT:
-		event__process_task(event, session);
+		event__process_task(event, sample, session);
 		break;
 	default:
 		break;
@@ -674,32 +688,8 @@ int event__preprocess_sample(const event_t *self, struct perf_session *session,
 			     symbol_filter_t filter)
 {
 	u8 cpumode = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK;
-	struct thread *thread;
-
-	event__parse_sample(self, session->sample_type, data);
+	struct thread *thread = perf_session__findnew(session, self->ip.pid);
 
-	dump_printf("(IP, %d): %d/%d: %#Lx period: %Ld cpu:%d\n",
-		    self->header.misc, data->pid, data->tid, data->ip,
-		    data->period, data->cpu);
-
-	if (session->sample_type & PERF_SAMPLE_CALLCHAIN) {
-		unsigned int i;
-
-		dump_printf("... chain: nr:%Lu\n", data->callchain->nr);
-
-		if (!ip_callchain__valid(data->callchain, self)) {
-			pr_debug("call-chain problem with event, "
-				 "skipping it.\n");
-			goto out_filtered;
-		}
-
-		if (dump_trace) {
-			for (i = 0; i < data->callchain->nr; i++)
-				dump_printf("..... %2d: %016Lx\n",
-					    i, data->callchain->ips[i]);
-		}
-	}
-	thread = perf_session__findnew(session, self->ip.pid);
 	if (thread == NULL)
 		return -1;
 
diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h
index 8e790dae702..08c400b83d5 100644
--- a/tools/perf/util/event.h
+++ b/tools/perf/util/event.h
@@ -135,7 +135,10 @@ void event__print_totals(void);
 
 struct perf_session;
 
-typedef int (*event__handler_t)(event_t *event, struct perf_session *session);
+typedef int (*event__handler_synth_t)(event_t *event, 
+				      struct perf_session *session);
+typedef int (*event__handler_t)(event_t *event, struct sample_data *sample,
+				struct perf_session *session);
 
 int event__synthesize_thread(pid_t pid, event__handler_t process,
 			     struct perf_session *session);
@@ -150,11 +153,16 @@ int event__synthesize_modules(event__handler_t process,
 			      struct perf_session *session,
 			      struct machine *machine);
 
-int event__process_comm(event_t *self, struct perf_session *session);
-int event__process_lost(event_t *self, struct perf_session *session);
-int event__process_mmap(event_t *self, struct perf_session *session);
-int event__process_task(event_t *self, struct perf_session *session);
-int event__process(event_t *event, struct perf_session *session);
+int event__process_comm(event_t *self, struct sample_data *sample,
+			struct perf_session *session);
+int event__process_lost(event_t *self, struct sample_data *sample,
+			struct perf_session *session);
+int event__process_mmap(event_t *self, struct sample_data *sample,
+			struct perf_session *session);
+int event__process_task(event_t *self, struct sample_data *sample,
+			struct perf_session *session);
+int event__process(event_t *event, struct sample_data *sample,
+		   struct perf_session *session);
 
 struct addr_location;
 int event__preprocess_sample(const event_t *self, struct perf_session *session,
diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c
index f65d7dc127b..fe652f3b0aa 100644
--- a/tools/perf/util/header.c
+++ b/tools/perf/util/header.c
@@ -993,15 +993,14 @@ int event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id,
 	ev->attr.header.type = PERF_RECORD_HEADER_ATTR;
 	ev->attr.header.size = size;
 
-	err = process(ev, session);
+	err = process(ev, NULL, session);
 
 	free(ev);
 
 	return err;
 }
 
-int event__synthesize_attrs(struct perf_header *self,
-			    event__handler_t process,
+int event__synthesize_attrs(struct perf_header *self, event__handler_t process,
 			    struct perf_session *session)
 {
 	struct perf_header_attr	*attr;
@@ -1071,7 +1070,7 @@ int event__synthesize_event_type(u64 event_id, char *name,
 	ev.event_type.header.size = sizeof(ev.event_type) -
 		(sizeof(ev.event_type.event_type.name) - size);
 
-	err = process(&ev, session);
+	err = process(&ev, NULL, session);
 
 	return err;
 }
@@ -1126,7 +1125,7 @@ int event__synthesize_tracing_data(int fd, struct perf_event_attr *pattrs,
 	ev.tracing_data.header.size = sizeof(ev.tracing_data);
 	ev.tracing_data.size = aligned_size;
 
-	process(&ev, session);
+	process(&ev, NULL, session);
 
 	err = read_tracing_data(fd, pattrs, nb_events);
 	write_padded(fd, NULL, 0, padding);
@@ -1186,7 +1185,7 @@ int event__synthesize_build_id(struct dso *pos, u16 misc,
 	ev.build_id.header.size = sizeof(ev.build_id) + len;
 	memcpy(&ev.build_id.filename, pos->long_name, pos->long_name_len);
 
-	err = process(&ev, session);
+	err = process(&ev, NULL, session);
 
 	return err;
 }
diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h
index 587d375d343..ee789856a8c 100644
--- a/tools/perf/util/hist.h
+++ b/tools/perf/util/hist.h
@@ -52,8 +52,10 @@ struct sym_priv {
 struct events_stats {
 	u64 total_period;
 	u64 total_lost;
+	u64 total_invalid_chains;
 	u32 nr_events[PERF_RECORD_HEADER_MAX];
 	u32 nr_unknown_events;
+	u32 nr_invalid_chains;
 };
 
 enum hist_column {
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index 52672dad1fe..08ec018966a 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -70,6 +70,11 @@ void perf_session__update_sample_type(struct perf_session *self)
 	self->sample_type = perf_header__sample_type(&self->header);
 }
 
+void perf_session__set_sample_type(struct perf_session *session, u64 type)
+{
+	session->sample_type = type;
+}
+
 int perf_session__create_kernel_maps(struct perf_session *self)
 {
 	int ret = machine__create_kernel_maps(&self->host_machine);
@@ -240,7 +245,15 @@ struct map_symbol *perf_session__resolve_callchain(struct perf_session *self,
 	return syms;
 }
 
+static int process_event_synth_stub(event_t *event __used,
+				    struct perf_session *session __used)
+{
+	dump_printf(": unhandled!\n");
+	return 0;
+}
+
 static int process_event_stub(event_t *event __used,
+			      struct sample_data *sample __used,
 			      struct perf_session *session __used)
 {
 	dump_printf(": unhandled!\n");
@@ -280,13 +293,13 @@ static void perf_event_ops__fill_defaults(struct perf_event_ops *handler)
 	if (handler->unthrottle == NULL)
 		handler->unthrottle = process_event_stub;
 	if (handler->attr == NULL)
-		handler->attr = process_event_stub;
+		handler->attr = process_event_synth_stub;
 	if (handler->event_type == NULL)
-		handler->event_type = process_event_stub;
+		handler->event_type = process_event_synth_stub;
 	if (handler->tracing_data == NULL)
-		handler->tracing_data = process_event_stub;
+		handler->tracing_data = process_event_synth_stub;
 	if (handler->build_id == NULL)
-		handler->build_id = process_event_stub;
+		handler->build_id = process_event_synth_stub;
 	if (handler->finished_round == NULL) {
 		if (handler->ordered_samples)
 			handler->finished_round = process_finished_round;
@@ -419,6 +432,7 @@ static void flush_sample_queue(struct perf_session *s,
 	struct ordered_samples *os = &s->ordered_samples;
 	struct list_head *head = &os->samples;
 	struct sample_queue *tmp, *iter;
+	struct sample_data sample;
 	u64 limit = os->next_flush;
 	u64 last_ts = os->last_sample ? os->last_sample->timestamp : 0ULL;
 
@@ -429,7 +443,8 @@ static void flush_sample_queue(struct perf_session *s,
 		if (iter->timestamp > limit)
 			break;
 
-		ops->sample(iter->event, s);
+		event__parse_sample(iter->event, s->sample_type, &sample);
+		ops->sample(iter->event, &sample, s);
 
 		os->last_flush = iter->timestamp;
 		list_del(&iter->list);
@@ -578,20 +593,29 @@ static int queue_sample_event(event_t *event, struct sample_data *data,
 	return 0;
 }
 
-static int perf_session__process_sample(event_t *event, struct perf_session *s,
+static int perf_session__process_sample(event_t *event,
+					struct sample_data *sample,
+					struct perf_session *s,
 					struct perf_event_ops *ops)
 {
-	struct sample_data data;
-
 	if (!ops->ordered_samples)
-		return ops->sample(event, s);
+		return ops->sample(event, sample, s);
+
+	queue_sample_event(event, sample, s);
+	return 0;
+}
 
-	bzero(&data, sizeof(struct sample_data));
-	event__parse_sample(event, s->sample_type, &data);
+static void callchain__dump(struct sample_data *sample)
+{
+	unsigned int i;
 
-	queue_sample_event(event, &data, s);
+	if (!dump_trace)
+		return;
 
-	return 0;
+	printf("... chain: nr:%Lu\n", sample->callchain->nr);
+
+	for (i = 0; i < sample->callchain->nr; i++)
+		printf("..... %2d: %016Lx\n", i, sample->callchain->ips[i]);
 }
 
 static int perf_session__process_event(struct perf_session *self,
@@ -599,8 +623,16 @@ static int perf_session__process_event(struct perf_session *self,
 				       struct perf_event_ops *ops,
 				       u64 file_offset)
 {
+	struct sample_data sample;
+
 	trace_event(event);
 
+	if (self->header.needs_swap && event__swap_ops[event->header.type])
+		event__swap_ops[event->header.type](event);
+
+	if (event->header.type == PERF_RECORD_SAMPLE)
+		event__parse_sample(event, self->sample_type, &sample);
+
 	if (event->header.type < PERF_RECORD_HEADER_MAX) {
 		dump_printf("%#Lx [%#x]: PERF_RECORD_%s",
 			    file_offset, event->header.size,
@@ -608,28 +640,41 @@ static int perf_session__process_event(struct perf_session *self,
 		hists__inc_nr_events(&self->hists, event->header.type);
 	}
 
-	if (self->header.needs_swap && event__swap_ops[event->header.type])
-		event__swap_ops[event->header.type](event);
-
 	switch (event->header.type) {
 	case PERF_RECORD_SAMPLE:
-		return perf_session__process_sample(event, self, ops);
+		dump_printf("(IP, %d): %d/%d: %#Lx period: %Ld\n", event->header.misc,
+			    sample.pid, sample.tid, sample.ip, sample.period);
+
+		if (self->sample_type & PERF_SAMPLE_CALLCHAIN) {
+			if (!ip_callchain__valid(sample.callchain, event)) {
+				pr_debug("call-chain problem with event, "
+					 "skipping it.\n");
+				++self->hists.stats.nr_invalid_chains;
+				self->hists.stats.total_invalid_chains += sample.period;
+				return 0;
+			}
+
+			callchain__dump(&sample);
+		}
+
+		return perf_session__process_sample(event, &sample, self, ops);
+
 	case PERF_RECORD_MMAP:
-		return ops->mmap(event, self);
+		return ops->mmap(event, &sample, self);
 	case PERF_RECORD_COMM:
-		return ops->comm(event, self);
+		return ops->comm(event, &sample, self);
 	case PERF_RECORD_FORK:
-		return ops->fork(event, self);
+		return ops->fork(event, &sample, self);
 	case PERF_RECORD_EXIT:
-		return ops->exit(event, self);
+		return ops->exit(event, &sample, self);
 	case PERF_RECORD_LOST:
-		return ops->lost(event, self);
+		return ops->lost(event, &sample, self);
 	case PERF_RECORD_READ:
-		return ops->read(event, self);
+		return ops->read(event, &sample, self);
 	case PERF_RECORD_THROTTLE:
-		return ops->throttle(event, self);
+		return ops->throttle(event, &sample, self);
 	case PERF_RECORD_UNTHROTTLE:
-		return ops->unthrottle(event, self);
+		return ops->unthrottle(event, &sample, self);
 	case PERF_RECORD_HEADER_ATTR:
 		return ops->attr(event, self);
 	case PERF_RECORD_HEADER_EVENT_TYPE:
@@ -894,6 +939,14 @@ out_err:
 			    session->hists.stats.nr_unknown_events);
 	}
 
+ 	if (session->hists.stats.nr_invalid_chains != 0) {
+ 		ui__warning("Found invalid callchains!\n\n"
+ 			    "%u out of %u events were discarded for this reason.\n\n"
+ 			    "Consider reporting to linux-kernel@vger.kernel.org.\n\n",
+ 			    session->hists.stats.nr_invalid_chains,
+ 			    session->hists.stats.nr_events[PERF_RECORD_SAMPLE]);
+ 	}
+
 	perf_session_free_sample_buffers(session);
 	return err;
 }
diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h
index 5bf6efa3788..4578f86a620 100644
--- a/tools/perf/util/session.h
+++ b/tools/perf/util/session.h
@@ -54,7 +54,9 @@ struct perf_session {
 
 struct perf_event_ops;
 
-typedef int (*event_op)(event_t *self, struct perf_session *session);
+typedef int (*event_op)(event_t *self, struct sample_data *sample,
+			struct perf_session *session);
+typedef int (*event_synth_op)(event_t *self, struct perf_session *session);
 typedef int (*event_op2)(event_t *self, struct perf_session *session,
 			 struct perf_event_ops *ops);
 
@@ -67,8 +69,8 @@ struct perf_event_ops {
 			lost,
 			read,
 			throttle,
-			unthrottle,
-			attr,
+			unthrottle;
+	event_synth_op	attr,
 			event_type,
 			tracing_data,
 			build_id;
@@ -104,6 +106,7 @@ int perf_session__create_kernel_maps(struct perf_session *self);
 
 int do_read(int fd, void *buf, size_t size);
 void perf_session__update_sample_type(struct perf_session *self);
+void perf_session__set_sample_type(struct perf_session *session, u64 type);
 void perf_session__remove_thread(struct perf_session *self, struct thread *th);
 
 static inline
-- 
cgit v1.2.3-70-g09d2


From 21ef97f05a7da5bc23b26cb34d6746f83ca9bf20 Mon Sep 17 00:00:00 2001
From: Ian Munsie <imunsie@au1.ibm.com>
Date: Fri, 10 Dec 2010 14:09:16 +1100
Subject: perf session: Fallback to unordered processing if no sample_id_all

If we are running the new perf on an old kernel without support for
sample_id_all, we should fall back to the old unordered processing of
events. If we didn't than we would *always* process events without
timestamps out of order, whether or not we hit a reordering race. In
other words, instead of there being a chance of not attributing samples
correctly, we would guarantee that samples would not be attributed.

While processing all events without timestamps before events with
timestamps may seem like an intuitive solution, it falls down as
PERF_RECORD_EXIT events would also be processed before any samples.
Even with a workaround for that case, samples before/after an exec would
not be attributed correctly.

This patch allows commands to indicate whether they need to fall back to
unordered processing, so that commands that do not care about timestamps
on every event will not be affected. If we do fallback, this will print
out a warning if report -D was invoked.

This patch adds the test in perf_session__new so that we only need to
test once per session. Commands that do not use an event_ops (such as
record and top) can simply pass NULL in it's place.

Acked-by: Thomas Gleixner <tglx@linutronix.de>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
LKML-Reference: <1291951882-sup-6069@au1.ibm.com>
Signed-off-by: Ian Munsie <imunsie@au1.ibm.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/builtin-annotate.c     |  2 +-
 tools/perf/builtin-buildid-list.c |  3 ++-
 tools/perf/builtin-diff.c         |  4 ++--
 tools/perf/builtin-inject.c       |  2 +-
 tools/perf/builtin-kmem.c         |  3 ++-
 tools/perf/builtin-lock.c         |  2 +-
 tools/perf/builtin-record.c       |  2 +-
 tools/perf/builtin-report.c       |  2 +-
 tools/perf/builtin-sched.c        |  3 ++-
 tools/perf/builtin-script.c       |  2 +-
 tools/perf/builtin-timechart.c    |  3 ++-
 tools/perf/builtin-top.c          |  2 +-
 tools/perf/util/session.c         | 11 ++++++++++-
 tools/perf/util/session.h         |  5 ++++-
 14 files changed, 31 insertions(+), 15 deletions(-)

(limited to 'tools/perf/builtin-diff.c')

diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c
index 569a2761b90..48dbab4b482 100644
--- a/tools/perf/builtin-annotate.c
+++ b/tools/perf/builtin-annotate.c
@@ -382,7 +382,7 @@ static int __cmd_annotate(void)
 	int ret;
 	struct perf_session *session;
 
-	session = perf_session__new(input_name, O_RDONLY, force, false);
+	session = perf_session__new(input_name, O_RDONLY, force, false, &event_ops);
 	if (session == NULL)
 		return -ENOMEM;
 
diff --git a/tools/perf/builtin-buildid-list.c b/tools/perf/builtin-buildid-list.c
index 44a47e13bd6..3b06f9ca263 100644
--- a/tools/perf/builtin-buildid-list.c
+++ b/tools/perf/builtin-buildid-list.c
@@ -39,7 +39,8 @@ static int __cmd_buildid_list(void)
 	int err = -1;
 	struct perf_session *session;
 
-	session = perf_session__new(input_name, O_RDONLY, force, false);
+	session = perf_session__new(input_name, O_RDONLY, force, false,
+				    &build_id__mark_dso_hit_ops);
 	if (session == NULL)
 		return -1;
 
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index 5e1a043aae0..af84e1c0519 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/builtin-diff.c
@@ -142,8 +142,8 @@ static int __cmd_diff(void)
 	int ret, i;
 	struct perf_session *session[2];
 
-	session[0] = perf_session__new(input_old, O_RDONLY, force, false);
-	session[1] = perf_session__new(input_new, O_RDONLY, force, false);
+	session[0] = perf_session__new(input_old, O_RDONLY, force, false, &event_ops);
+	session[1] = perf_session__new(input_new, O_RDONLY, force, false, &event_ops);
 	if (session[0] == NULL || session[1] == NULL)
 		return -ENOMEM;
 
diff --git a/tools/perf/builtin-inject.c b/tools/perf/builtin-inject.c
index 4b66b857941..0c78ffa7bf6 100644
--- a/tools/perf/builtin-inject.c
+++ b/tools/perf/builtin-inject.c
@@ -196,7 +196,7 @@ static int __cmd_inject(void)
 		inject_ops.tracing_data	= event__repipe_tracing_data;
 	}
 
-	session = perf_session__new(input_name, O_RDONLY, false, true);
+	session = perf_session__new(input_name, O_RDONLY, false, true, &inject_ops);
 	if (session == NULL)
 		return -ENOMEM;
 
diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c
index c9620ff6496..def7ddc2fd4 100644
--- a/tools/perf/builtin-kmem.c
+++ b/tools/perf/builtin-kmem.c
@@ -481,7 +481,8 @@ static void sort_result(void)
 static int __cmd_kmem(void)
 {
 	int err = -EINVAL;
-	struct perf_session *session = perf_session__new(input_name, O_RDONLY, 0, false);
+	struct perf_session *session = perf_session__new(input_name, O_RDONLY,
+							 0, false, &event_ops);
 	if (session == NULL)
 		return -ENOMEM;
 
diff --git a/tools/perf/builtin-lock.c b/tools/perf/builtin-lock.c
index b41b4492b1c..b9c6e543297 100644
--- a/tools/perf/builtin-lock.c
+++ b/tools/perf/builtin-lock.c
@@ -858,7 +858,7 @@ static struct perf_event_ops eops = {
 
 static int read_events(void)
 {
-	session = perf_session__new(input_name, O_RDONLY, 0, false);
+	session = perf_session__new(input_name, O_RDONLY, 0, false, &eops);
 	if (!session)
 		die("Initializing perf session failed\n");
 
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index e9be6ae87a2..efd1b3c3d4a 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -572,7 +572,7 @@ static int __cmd_record(int argc, const char **argv)
 	}
 
 	session = perf_session__new(output_name, O_WRONLY,
-				    write_mode == WRITE_FORCE, false);
+				    write_mode == WRITE_FORCE, false, NULL);
 	if (session == NULL) {
 		pr_err("Not enough memory for reading perf file header\n");
 		return -1;
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index b6a2a899aa8..fd4c4500cd1 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -308,7 +308,7 @@ static int __cmd_report(void)
 
 	signal(SIGINT, sig_handler);
 
-	session = perf_session__new(input_name, O_RDONLY, force, false);
+	session = perf_session__new(input_name, O_RDONLY, force, false, &event_ops);
 	if (session == NULL)
 		return -ENOMEM;
 
diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c
index c7753940aea..7a4ebeb8b01 100644
--- a/tools/perf/builtin-sched.c
+++ b/tools/perf/builtin-sched.c
@@ -1643,7 +1643,8 @@ static struct perf_event_ops event_ops = {
 static int read_events(void)
 {
 	int err = -EINVAL;
-	struct perf_session *session = perf_session__new(input_name, O_RDONLY, 0, false);
+	struct perf_session *session = perf_session__new(input_name, O_RDONLY,
+							 0, false, &event_ops);
 	if (session == NULL)
 		return -ENOMEM;
 
diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c
index 54f1ea808db..6ef65c04ab9 100644
--- a/tools/perf/builtin-script.c
+++ b/tools/perf/builtin-script.c
@@ -779,7 +779,7 @@ int cmd_script(int argc, const char **argv, const char *prefix __used)
 	if (!script_name)
 		setup_pager();
 
-	session = perf_session__new(input_name, O_RDONLY, 0, false);
+	session = perf_session__new(input_name, O_RDONLY, 0, false, &event_ops);
 	if (session == NULL)
 		return -ENOMEM;
 
diff --git a/tools/perf/builtin-timechart.c b/tools/perf/builtin-timechart.c
index d2fc46103f8..459b5e3db26 100644
--- a/tools/perf/builtin-timechart.c
+++ b/tools/perf/builtin-timechart.c
@@ -937,7 +937,8 @@ static struct perf_event_ops event_ops = {
 
 static int __cmd_timechart(void)
 {
-	struct perf_session *session = perf_session__new(input_name, O_RDONLY, 0, false);
+	struct perf_session *session = perf_session__new(input_name, O_RDONLY,
+							 0, false, &event_ops);
 	int ret = -EINVAL;
 
 	if (session == NULL)
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 0515ce9d3d3..ae15f046c40 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -1272,7 +1272,7 @@ static int __cmd_top(void)
 	 * FIXME: perf_session__new should allow passing a O_MMAP, so that all this
 	 * mmap reading, etc is encapsulated in it. Use O_WRONLY for now.
 	 */
-	struct perf_session *session = perf_session__new(NULL, O_WRONLY, false, false);
+	struct perf_session *session = perf_session__new(NULL, O_WRONLY, false, false, NULL);
 	if (session == NULL)
 		return -ENOMEM;
 
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index b59abf5aba3..0f7e544544f 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -125,7 +125,9 @@ static void perf_session__destroy_kernel_maps(struct perf_session *self)
 	machines__destroy_guest_kernel_maps(&self->machines);
 }
 
-struct perf_session *perf_session__new(const char *filename, int mode, bool force, bool repipe)
+struct perf_session *perf_session__new(const char *filename, int mode,
+				       bool force, bool repipe,
+				       struct perf_event_ops *ops)
 {
 	size_t len = filename ? strlen(filename) + 1 : 0;
 	struct perf_session *self = zalloc(sizeof(*self) + len);
@@ -170,6 +172,13 @@ struct perf_session *perf_session__new(const char *filename, int mode, bool forc
 	}
 
 	perf_session__update_sample_type(self);
+
+	if (ops && ops->ordering_requires_timestamps &&
+	    ops->ordered_samples && !self->sample_id_all) {
+		dump_printf("WARNING: No sample_id_all support, falling back to unordered processing\n");
+		ops->ordered_samples = false;
+	}
+
 out:
 	return self;
 out_free:
diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h
index ac36f99f14a..ffe4b98db8f 100644
--- a/tools/perf/util/session.h
+++ b/tools/perf/util/session.h
@@ -78,9 +78,12 @@ struct perf_event_ops {
 			build_id;
 	event_op2	finished_round;
 	bool		ordered_samples;
+	bool		ordering_requires_timestamps;
 };
 
-struct perf_session *perf_session__new(const char *filename, int mode, bool force, bool repipe);
+struct perf_session *perf_session__new(const char *filename, int mode,
+				       bool force, bool repipe,
+				       struct perf_event_ops *ops);
 void perf_session__delete(struct perf_session *self);
 
 void perf_event_header__bswap(struct perf_event_header *self);
-- 
cgit v1.2.3-70-g09d2


From eac23d1c384b55e4bbb89ea9e5a6bb77fb4d1140 Mon Sep 17 00:00:00 2001
From: Ian Munsie <imunsie@au1.ibm.com>
Date: Thu, 9 Dec 2010 16:33:53 +1100
Subject: perf record,report,annotate,diff: Process events in order

This patch changes perf report to ask for the ID info on all events be
default if recording from multiple CPUs.

Perf report, annotate and diff will now process the events in order if
the kernel is able to provide timestamps on all events. This ensures
that events such as COMM and MMAP which are necessary to correctly
interpret samples are processed prior to those samples so that they are
attributed correctly.

Before:
 # perf record ./cachetest
 # perf report

 # Events: 6K cycles
 #
 # Overhead  Command      Shared Object                           Symbol
 # ........  .......  .................  ...............................
 #
     74.11%    :3259  [unknown]          [k] 0x4a6c
      1.50%  cachetest  ld-2.11.2.so       [.] 0x1777c
      1.46%    :3259  [kernel.kallsyms]  [k] .perf_event_mmap_ctx
      1.25%    :3259  [kernel.kallsyms]  [k] restore
      0.74%    :3259  [kernel.kallsyms]  [k] ._raw_spin_lock
      0.71%    :3259  [kernel.kallsyms]  [k] .filemap_fault
      0.66%    :3259  [kernel.kallsyms]  [k] .memset
      0.54%  cachetest  [kernel.kallsyms]  [k] .sha_transform
      0.54%    :3259  [kernel.kallsyms]  [k] .copy_4K_page
      0.54%    :3259  [kernel.kallsyms]  [k] .find_get_page
      0.52%    :3259  [kernel.kallsyms]  [k] .trace_hardirqs_off
      0.50%    :3259  [kernel.kallsyms]  [k] .__do_fault
<SNIP>

After:
 # perf report

 # Events: 6K cycles
 #
 # Overhead  Command      Shared Object                           Symbol
 # ........  .......  .................  ...............................
 #
     44.28%  cachetest  cachetest          [.] sumArrayNaive
     22.53%  cachetest  cachetest          [.] sumArrayOptimal
      6.59%  cachetest  ld-2.11.2.so       [.] 0x1777c
      2.13%  cachetest  [unknown]          [k] 0x340
      1.46%  cachetest  [kernel.kallsyms]  [k] .perf_event_mmap_ctx
      1.25%  cachetest  [kernel.kallsyms]  [k] restore
      0.74%  cachetest  [kernel.kallsyms]  [k] ._raw_spin_lock
      0.71%  cachetest  [kernel.kallsyms]  [k] .filemap_fault
      0.66%  cachetest  [kernel.kallsyms]  [k] .memset
      0.54%  cachetest  [kernel.kallsyms]  [k] .copy_4K_page
      0.54%  cachetest  [kernel.kallsyms]  [k] .find_get_page
      0.54%  cachetest  [kernel.kallsyms]  [k] .sha_transform
      0.52%  cachetest  [kernel.kallsyms]  [k] .trace_hardirqs_off
      0.50%  cachetest  [kernel.kallsyms]  [k] .__do_fault
<SNIP>

Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
LKML-Reference: <1291872833-839-1-git-send-email-imunsie@au1.ibm.com>
Signed-off-by: Ian Munsie <imunsie@au1.ibm.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/builtin-annotate.c | 2 ++
 tools/perf/builtin-diff.c     | 2 ++
 tools/perf/builtin-record.c   | 5 ++++-
 tools/perf/builtin-report.c   | 2 ++
 4 files changed, 10 insertions(+), 1 deletion(-)

(limited to 'tools/perf/builtin-diff.c')

diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c
index 48dbab4b482..c056cdc0691 100644
--- a/tools/perf/builtin-annotate.c
+++ b/tools/perf/builtin-annotate.c
@@ -375,6 +375,8 @@ static struct perf_event_ops event_ops = {
 	.mmap	= event__process_mmap,
 	.comm	= event__process_comm,
 	.fork	= event__process_task,
+	.ordered_samples = true,
+	.ordering_requires_timestamps = true,
 };
 
 static int __cmd_annotate(void)
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index af84e1c0519..97846dcafc6 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/builtin-diff.c
@@ -61,6 +61,8 @@ static struct perf_event_ops event_ops = {
 	.exit	= event__process_task,
 	.fork	= event__process_task,
 	.lost	= event__process_lost,
+	.ordered_samples = true,
+	.ordering_requires_timestamps = true,
 };
 
 static void perf_session__insert_hist_entry_by_name(struct rb_root *root,
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index efd1b3c3d4a..5149e3deb7b 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -285,7 +285,7 @@ static void create_counter(int counter, int cpu)
 	if (system_wide)
 		attr->sample_type	|= PERF_SAMPLE_CPU;
 
-	if (sample_time)
+	if (sample_time || system_wide || !no_inherit || cpu_list)
 		attr->sample_type	|= PERF_SAMPLE_TIME;
 
 	if (raw_samples) {
@@ -327,6 +327,9 @@ try_again:
 				 * Old kernel, no attr->sample_id_type_all field
 				 */
 				sample_id_all_avail = false;
+				if (!sample_time && !raw_samples)
+					attr->sample_type &= ~PERF_SAMPLE_TIME;
+
 				goto retry_sample_id;
 			}
 
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index fd4c4500cd1..4af7ce6e155 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -244,6 +244,8 @@ static struct perf_event_ops event_ops = {
 	.event_type = event__process_event_type,
 	.tracing_data = event__process_tracing_data,
 	.build_id = event__process_build_id,
+	.ordered_samples = true,
+	.ordering_requires_timestamps = true,
 };
 
 extern volatile int session_done;
-- 
cgit v1.2.3-70-g09d2


From ec5761eab318e50e69fcf8e63e9edaef5949c067 Mon Sep 17 00:00:00 2001
From: David Ahern <daahern@cisco.com>
Date: Thu, 9 Dec 2010 13:27:07 -0700
Subject: perf symbols: Add symfs option for off-box analysis using specified
 tree

The symfs argument allows analysis of perf.data file using a locally accessible
filesystem tree with debug symbols - e.g., tree created during image builds,
sshfs mount, loop mounted KVM disk images, USB keys, initrds, etc. Anything
with an OS tree can be analyzed from anywhere without the need to populate a
local data store with build-ids.

Commiter notes:

o Fixed up symfs="/" variants handling.

o prefixed DSO__ORIG_GUEST_KMODULE case with symfs too, avoiding use of files
  outside the symfs directory.

LKML-Reference: <1291926427-28846-1-git-send-email-daahern@cisco.com>
Signed-off-by: David Ahern <daahern@cisco.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
---
 tools/perf/Documentation/perf-diff.txt      |  2 +
 tools/perf/Documentation/perf-report.txt    |  3 ++
 tools/perf/Documentation/perf-timechart.txt |  2 +
 tools/perf/builtin-diff.c                   |  2 +
 tools/perf/builtin-report.c                 |  2 +
 tools/perf/builtin-timechart.c              |  2 +
 tools/perf/util/hist.c                      | 14 ++++--
 tools/perf/util/symbol.c                    | 72 ++++++++++++++++++++++-------
 tools/perf/util/symbol.h                    |  1 +
 9 files changed, 81 insertions(+), 19 deletions(-)

(limited to 'tools/perf/builtin-diff.c')

diff --git a/tools/perf/Documentation/perf-diff.txt b/tools/perf/Documentation/perf-diff.txt
index 6a9ec2b3531..74d7481ed7a 100644
--- a/tools/perf/Documentation/perf-diff.txt
+++ b/tools/perf/Documentation/perf-diff.txt
@@ -66,6 +66,8 @@ OPTIONS
 --force::
        Don't complain, do it.
 
+--symfs=<directory>::
+        Look for files with symbols relative to this directory.
 
 SEE ALSO
 --------
diff --git a/tools/perf/Documentation/perf-report.txt b/tools/perf/Documentation/perf-report.txt
index fefea77ec6e..8ba03d6e539 100644
--- a/tools/perf/Documentation/perf-report.txt
+++ b/tools/perf/Documentation/perf-report.txt
@@ -116,6 +116,9 @@ OPTIONS
 --force::
         Don't complain, do it.
 
+--symfs=<directory>::
+        Look for files with symbols relative to this directory.
+
 SEE ALSO
 --------
 linkperf:perf-stat[1]
diff --git a/tools/perf/Documentation/perf-timechart.txt b/tools/perf/Documentation/perf-timechart.txt
index 4b1788355ec..d7b79e2ba2a 100644
--- a/tools/perf/Documentation/perf-timechart.txt
+++ b/tools/perf/Documentation/perf-timechart.txt
@@ -38,6 +38,8 @@ OPTIONS
 --process::
         Select the processes to display, by name or PID
 
+--symfs=<directory>::
+        Look for files with symbols relative to this directory.
 
 SEE ALSO
 --------
diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c
index 97846dcafc6..3153e492dbc 100644
--- a/tools/perf/builtin-diff.c
+++ b/tools/perf/builtin-diff.c
@@ -194,6 +194,8 @@ static const struct option options[] = {
 	OPT_STRING('t', "field-separator", &symbol_conf.field_sep, "separator",
 		   "separator for columns, no spaces will be added between "
 		   "columns '.' is reserved."),
+	OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory",
+		    "Look for files with symbols relative to this directory"),
 	OPT_END()
 };
 
diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index 4af7ce6e155..75183a4518e 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -483,6 +483,8 @@ static const struct option options[] = {
 		   "columns '.' is reserved."),
 	OPT_BOOLEAN('U', "hide-unresolved", &hide_unresolved,
 		    "Only display entries resolved to a symbol"),
+	OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory",
+		    "Look for files with symbols relative to this directory"),
 	OPT_END()
 };
 
diff --git a/tools/perf/builtin-timechart.c b/tools/perf/builtin-timechart.c
index 459b5e3db26..d75084bccdb 100644
--- a/tools/perf/builtin-timechart.c
+++ b/tools/perf/builtin-timechart.c
@@ -1022,6 +1022,8 @@ static const struct option options[] = {
 	OPT_CALLBACK('p', "process", NULL, "process",
 		      "process selector. Pass a pid or process name.",
 		       parse_process),
+	OPT_STRING(0, "symfs", &symbol_conf.symfs, "directory",
+		    "Look for files with symbols relative to this directory"),
 	OPT_END()
 };
 
diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c
index a3b84160c42..d5036700a43 100644
--- a/tools/perf/util/hist.c
+++ b/tools/perf/util/hist.c
@@ -1092,6 +1092,12 @@ int hist_entry__annotate(struct hist_entry *self, struct list_head *head,
 	FILE *file;
 	int err = 0;
 	u64 len;
+	char symfs_filename[PATH_MAX];
+
+	if (filename) {
+		snprintf(symfs_filename, sizeof(symfs_filename), "%s%s",
+			 symbol_conf.symfs, filename);
+	}
 
 	if (filename == NULL) {
 		if (dso->has_build_id) {
@@ -1100,9 +1106,9 @@ int hist_entry__annotate(struct hist_entry *self, struct list_head *head,
 			return -ENOMEM;
 		}
 		goto fallback;
-	} else if (readlink(filename, command, sizeof(command)) < 0 ||
+	} else if (readlink(symfs_filename, command, sizeof(command)) < 0 ||
 		   strstr(command, "[kernel.kallsyms]") ||
-		   access(filename, R_OK)) {
+		   access(symfs_filename, R_OK)) {
 		free(filename);
 fallback:
 		/*
@@ -1111,6 +1117,8 @@ fallback:
 		 * DSO is the same as when 'perf record' ran.
 		 */
 		filename = dso->long_name;
+		snprintf(symfs_filename, sizeof(symfs_filename), "%s%s",
+			 symbol_conf.symfs, filename);
 		free_filename = false;
 	}
 
@@ -1137,7 +1145,7 @@ fallback:
 		 "objdump --start-address=0x%016Lx --stop-address=0x%016Lx -dS -C %s|grep -v %s|expand",
 		 map__rip_2objdump(map, sym->start),
 		 map__rip_2objdump(map, sym->end),
-		 filename, filename);
+		 symfs_filename, filename);
 
 	pr_debug("Executing: %s\n", command);
 
diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
index ceefa6568de..561db6361f5 100644
--- a/tools/perf/util/symbol.c
+++ b/tools/perf/util/symbol.c
@@ -41,6 +41,7 @@ struct symbol_conf symbol_conf = {
 	.exclude_other	  = true,
 	.use_modules	  = true,
 	.try_vmlinux_path = true,
+	.symfs            = "",
 };
 
 int dso__name_len(const struct dso *self)
@@ -839,8 +840,11 @@ static int dso__synthesize_plt_symbols(struct  dso *self, struct map *map,
 	char sympltname[1024];
 	Elf *elf;
 	int nr = 0, symidx, fd, err = 0;
+	char name[PATH_MAX];
 
-	fd = open(self->long_name, O_RDONLY);
+	snprintf(name, sizeof(name), "%s%s",
+		 symbol_conf.symfs, self->long_name);
+	fd = open(name, O_RDONLY);
 	if (fd < 0)
 		goto out;
 
@@ -1452,16 +1456,19 @@ int dso__load(struct dso *self, struct map *map, symbol_filter_t filter)
 	     self->origin++) {
 		switch (self->origin) {
 		case DSO__ORIG_BUILD_ID_CACHE:
-			if (dso__build_id_filename(self, name, size) == NULL)
+			/* skip the locally configured cache if a symfs is given */
+			if (symbol_conf.symfs[0] ||
+			    (dso__build_id_filename(self, name, size) == NULL)) {
 				continue;
+			}
 			break;
 		case DSO__ORIG_FEDORA:
-			snprintf(name, size, "/usr/lib/debug%s.debug",
-				 self->long_name);
+			snprintf(name, size, "%s/usr/lib/debug%s.debug",
+				 symbol_conf.symfs, self->long_name);
 			break;
 		case DSO__ORIG_UBUNTU:
-			snprintf(name, size, "/usr/lib/debug%s",
-				 self->long_name);
+			snprintf(name, size, "%s/usr/lib/debug%s",
+				 symbol_conf.symfs, self->long_name);
 			break;
 		case DSO__ORIG_BUILDID: {
 			char build_id_hex[BUILD_ID_SIZE * 2 + 1];
@@ -1473,19 +1480,26 @@ int dso__load(struct dso *self, struct map *map, symbol_filter_t filter)
 					  sizeof(self->build_id),
 					  build_id_hex);
 			snprintf(name, size,
-				 "/usr/lib/debug/.build-id/%.2s/%s.debug",
-				 build_id_hex, build_id_hex + 2);
+				 "%s/usr/lib/debug/.build-id/%.2s/%s.debug",
+				 symbol_conf.symfs, build_id_hex, build_id_hex + 2);
 			}
 			break;
 		case DSO__ORIG_DSO:
-			snprintf(name, size, "%s", self->long_name);
+			snprintf(name, size, "%s%s",
+			     symbol_conf.symfs, self->long_name);
 			break;
 		case DSO__ORIG_GUEST_KMODULE:
 			if (map->groups && map->groups->machine)
 				root_dir = map->groups->machine->root_dir;
 			else
 				root_dir = "";
-			snprintf(name, size, "%s%s", root_dir, self->long_name);
+			snprintf(name, size, "%s%s%s", symbol_conf.symfs,
+				 root_dir, self->long_name);
+			break;
+
+		case DSO__ORIG_KMODULE:
+			snprintf(name, size, "%s%s", symbol_conf.symfs,
+				 self->long_name);
 			break;
 
 		default:
@@ -1784,17 +1798,20 @@ static int dso__load_vmlinux(struct dso *self, struct map *map,
 			     const char *vmlinux, symbol_filter_t filter)
 {
 	int err = -1, fd;
+	char symfs_vmlinux[PATH_MAX];
 
-	fd = open(vmlinux, O_RDONLY);
+	snprintf(symfs_vmlinux, sizeof(symfs_vmlinux), "%s/%s",
+		 symbol_conf.symfs, vmlinux);
+	fd = open(symfs_vmlinux, O_RDONLY);
 	if (fd < 0)
 		return -1;
 
 	dso__set_loaded(self, map->type);
-	err = dso__load_sym(self, map, vmlinux, fd, filter, 0, 0);
+	err = dso__load_sym(self, map, symfs_vmlinux, fd, filter, 0, 0);
 	close(fd);
 
 	if (err > 0)
-		pr_debug("Using %s for symbols\n", vmlinux);
+		pr_debug("Using %s for symbols\n", symfs_vmlinux);
 
 	return err;
 }
@@ -1872,6 +1889,10 @@ static int dso__load_kernel_sym(struct dso *self, struct map *map,
 			goto out_fixup;
 	}
 
+	/* do not try local files if a symfs was given */
+	if (symbol_conf.symfs[0] != 0)
+		return -1;
+
 	/*
 	 * Say the kernel DSO was created when processing the build-id header table,
 	 * we have a build-id, so check if it is the same as the running kernel,
@@ -2262,9 +2283,6 @@ static int vmlinux_path__init(void)
 	struct utsname uts;
 	char bf[PATH_MAX];
 
-	if (uname(&uts) < 0)
-		return -1;
-
 	vmlinux_path = malloc(sizeof(char *) * 5);
 	if (vmlinux_path == NULL)
 		return -1;
@@ -2277,6 +2295,14 @@ static int vmlinux_path__init(void)
 	if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
 		goto out_fail;
 	++vmlinux_path__nr_entries;
+
+	/* only try running kernel version if no symfs was given */
+	if (symbol_conf.symfs[0] != 0)
+		return 0;
+
+	if (uname(&uts) < 0)
+		return -1;
+
 	snprintf(bf, sizeof(bf), "/boot/vmlinux-%s", uts.release);
 	vmlinux_path[vmlinux_path__nr_entries] = strdup(bf);
 	if (vmlinux_path[vmlinux_path__nr_entries] == NULL)
@@ -2336,6 +2362,8 @@ static int setup_list(struct strlist **list, const char *list_str,
 
 int symbol__init(void)
 {
+	const char *symfs;
+
 	if (symbol_conf.initialized)
 		return 0;
 
@@ -2364,6 +2392,18 @@ int symbol__init(void)
 		       symbol_conf.sym_list_str, "symbol") < 0)
 		goto out_free_comm_list;
 
+	/*
+	 * A path to symbols of "/" is identical to ""
+	 * reset here for simplicity.
+	 */
+	symfs = realpath(symbol_conf.symfs, NULL);
+	if (symfs == NULL)
+		symfs = symbol_conf.symfs;
+	if (strcmp(symfs, "/") == 0)
+		symbol_conf.symfs = "";
+	if (symfs != symbol_conf.symfs)
+		free((void *)symfs);
+
 	symbol_conf.initialized = true;
 	return 0;
 
diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h
index 12defbe18c1..bcd2f986927 100644
--- a/tools/perf/util/symbol.h
+++ b/tools/perf/util/symbol.h
@@ -86,6 +86,7 @@ struct symbol_conf {
        struct strlist	*dso_list,
 			*comm_list,
 			*sym_list;
+	const char	*symfs;
 };
 
 extern struct symbol_conf symbol_conf;
-- 
cgit v1.2.3-70-g09d2