[PATCH 20/23] perf ftrace: Add option for function filtering

[Arnaldo Carvalho de Melo]

From: Namhyung Kim <namhyung@xxxxxxxxxx>

The -T/--trace-funcs and -N/--notrace-funcs options are to specify
functions to enable/disable tracing dynamically.

The -G/--graph-funcs and -g/--nograph-funcs options are to set filters
for function graph tracer.

For example, to trace fault handling functions only:

  $ sudo perf ftrace -T *fault hello
   0)               |  __do_page_fault() {
   0)               |    handle_mm_fault() {
   0)   2.117 us    |      __handle_mm_fault();
   0)   3.627 us    |    }
   0)   7.811 us    |  }
   0)               |  __do_page_fault() {
   0)               |    handle_mm_fault() {
   0)   2.014 us    |      __handle_mm_fault();
   0)   2.424 us    |    }
   0)   2.951 us    |  }
   ...

To trace all functions executed in __do_page_fault:

  $ sudo perf ftrace -G __do_page_fault hello
   2)               |  __do_page_fault() {
   3)   0.060 us    |    down_read_trylock();
   3)               |    find_vma() {
   3)   0.075 us    |      vmacache_find();
   3)   0.053 us    |      vmacache_update();
   3)   1.246 us    |    }
   3)               |    handle_mm_fault() {
   3)   0.063 us    |      __rcu_read_lock();
   3)   0.056 us    |      mem_cgroup_from_task();
   3)   0.057 us    |      __rcu_read_unlock();
   3)               |      __handle_mm_fault() {
   3)               |        filemap_map_pages() {
   3)   0.058 us    |          __rcu_read_lock();
   3)               |          alloc_set_pte() {
   ...

But don't want to show details in handle_mm_fault:

  $ sudo perf ftrace -G __do_page_fault -g handle_mm_fault hello
   3)               |  __do_page_fault() {
   3)   0.049 us    |    down_read_trylock();
   3)               |    find_vma() {
   3)   0.048 us    |      vmacache_find();
   3)   0.041 us    |      vmacache_update();
   3)   0.680 us    |    }
   3)   0.036 us    |    up_read();
   3)   4.547 us    |  } /* __do_page_fault */
   ...

Signed-off-by: Namhyung Kim <namhyung@xxxxxxxxxx>
Tested-by: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
Cc: Frederic Weisbecker <fweisbec@xxxxxxxxx>
Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
Cc: Masami Hiramatsu <mhiramat@xxxxxxxxxx>
Cc: Peter Zijlstra <peterz@xxxxxxxxxxxxx>
Cc: Steven Rostedt <rostedt@xxxxxxxxxxx>
Cc: kernel-team@xxxxxxx
Link: http://lkml.kernel.org/r/20170618142302.25390-3-namhyung@xxxxxxxxxx
Signed-off-by: Arnaldo Carvalho de Melo <acme@xxxxxxxxxx>
---
 tools/perf/Documentation/perf-ftrace.txt |  30 ++++++++
 tools/perf/builtin-ftrace.c              | 117 +++++++++++++++++++++++++++++--
 2 files changed, 141 insertions(+), 6 deletions(-)

diff --git a/tools/perf/Documentation/perf-ftrace.txt b/tools/perf/Documentation/perf-ftrace.txt
index 6e6a8b22c859..78d6126ca485 100644
--- a/tools/perf/Documentation/perf-ftrace.txt
+++ b/tools/perf/Documentation/perf-ftrace.txt
@@ -48,6 +48,36 @@ OPTIONS
 	Ranges of CPUs are specified with -: 0-2.
 	Default is to trace on all online CPUs.
 
+-T::
+--trace-funcs=::
+	Only trace functions given by the argument.  Multiple functions
+	can be given by using this option more than once.  The function
+	argument also can be a glob pattern.  It will be passed to
+	'set_ftrace_filter' in tracefs.
+
+-N::
+--notrace-funcs=::
+	Do not trace functions given by the argument.  Like -T option,
+	this can be used more than once to specify multiple functions
+	(or glob patterns).  It will be passed to 'set_ftrace_notrace'
+	in tracefs.
+
+-G::
+--graph-funcs=::
+	Set graph filter on the given function (or a glob pattern).
+	This is useful for the function_graph tracer only and enables
+	tracing for functions executed from the given function.
+	This can be used more than once to specify multiple functions.
+	It will be passed to 'set_graph_function' in tracefs.
+
+-g::
+--nograph-funcs=::
+	Set graph notrace filter on the given function (or a glob pattern).
+	Like -G option, this is useful for the function_graph tracer only
+	and disables tracing for function executed from the given function.
+	This can be used more than once to specify multiple functions.
+	It will be passed to 'set_graph_notrace' in tracefs.
+
 
 SEE ALSO
 --------
diff --git a/tools/perf/builtin-ftrace.c b/tools/perf/builtin-ftrace.c
index 982b98ee639e..3285375ce3c2 100644
--- a/tools/perf/builtin-ftrace.c
+++ b/tools/perf/builtin-ftrace.c
@@ -28,9 +28,18 @@
 #define DEFAULT_TRACER  "function_graph"
 
 struct perf_ftrace {
-	struct perf_evlist *evlist;
-	struct target target;
-	const char *tracer;
+	struct perf_evlist	*evlist;
+	struct target		target;
+	const char		*tracer;
+	struct list_head	filters;
+	struct list_head	notrace;
+	struct list_head	graph_funcs;
+	struct list_head	nograph_funcs;
+};
+
+struct filter_entry {
+	struct list_head	list;
+	char			name[];
 };
 
 static bool done;
@@ -104,6 +113,7 @@ static int append_tracing_file(const char *name, const char *val)
 }
 
 static int reset_tracing_cpu(void);
+static void reset_tracing_filters(void);
 
 static int reset_tracing_files(struct perf_ftrace *ftrace __maybe_unused)
 {
@@ -119,6 +129,7 @@ static int reset_tracing_files(struct perf_ftrace *ftrace __maybe_unused)
 	if (reset_tracing_cpu() < 0)
 		return -1;
 
+	reset_tracing_filters();
 	return 0;
 }
 
@@ -184,6 +195,48 @@ static int reset_tracing_cpu(void)
 	return ret;
 }
 
+static int __set_tracing_filter(const char *filter_file, struct list_head *funcs)
+{
+	struct filter_entry *pos;
+
+	list_for_each_entry(pos, funcs, list) {
+		if (append_tracing_file(filter_file, pos->name) < 0)
+			return -1;
+	}
+
+	return 0;
+}
+
+static int set_tracing_filters(struct perf_ftrace *ftrace)
+{
+	int ret;
+
+	ret = __set_tracing_filter("set_ftrace_filter", &ftrace->filters);
+	if (ret < 0)
+		return ret;
+
+	ret = __set_tracing_filter("set_ftrace_notrace", &ftrace->notrace);
+	if (ret < 0)
+		return ret;
+
+	ret = __set_tracing_filter("set_graph_function", &ftrace->graph_funcs);
+	if (ret < 0)
+		return ret;
+
+	/* old kernels do not have this filter */
+	__set_tracing_filter("set_graph_notrace", &ftrace->nograph_funcs);
+
+	return ret;
+}
+
+static void reset_tracing_filters(void)
+{
+	write_tracing_file("set_ftrace_filter", " ");
+	write_tracing_file("set_ftrace_notrace", " ");
+	write_tracing_file("set_graph_function", " ");
+	write_tracing_file("set_graph_notrace", " ");
+}
+
 static int __cmd_ftrace(struct perf_ftrace *ftrace, int argc, const char **argv)
 {
 	char *trace_file;
@@ -226,6 +279,11 @@ static int __cmd_ftrace(struct perf_ftrace *ftrace, int argc, const char **argv)
 		goto out_reset;
 	}
 
+	if (set_tracing_filters(ftrace) < 0) {
+		pr_err("failed to set tracing filters\n");
+		goto out_reset;
+	}
+
 	if (write_tracing_file("current_tracer", ftrace->tracer) < 0) {
 		pr_err("failed to set current_tracer to %s\n", ftrace->tracer);
 		goto out_reset;
@@ -310,6 +368,32 @@ static int perf_ftrace_config(const char *var, const char *value, void *cb)
 	return -1;
 }
 
+static int parse_filter_func(const struct option *opt, const char *str,
+			     int unset __maybe_unused)
+{
+	struct list_head *head = opt->value;
+	struct filter_entry *entry;
+
+	entry = malloc(sizeof(*entry) + strlen(str) + 1);
+	if (entry == NULL)
+		return -ENOMEM;
+
+	strcpy(entry->name, str);
+	list_add_tail(&entry->list, head);
+
+	return 0;
+}
+
+static void delete_filter_func(struct list_head *head)
+{
+	struct filter_entry *pos, *tmp;
+
+	list_for_each_entry_safe(pos, tmp, head, list) {
+		list_del(&pos->list);
+		free(pos);
+	}
+}
+
 int cmd_ftrace(int argc, const char **argv)
 {
 	int ret;
@@ -333,9 +417,22 @@ int cmd_ftrace(int argc, const char **argv)
 		    "system-wide collection from all CPUs"),
 	OPT_STRING('C', "cpu", &ftrace.target.cpu_list, "cpu",
 		    "list of cpus to monitor"),
+	OPT_CALLBACK('T', "trace-funcs", &ftrace.filters, "func",
+		     "trace given functions only", parse_filter_func),
+	OPT_CALLBACK('N', "notrace-funcs", &ftrace.notrace, "func",
+		     "do not trace given functions", parse_filter_func),
+	OPT_CALLBACK('G', "graph-funcs", &ftrace.graph_funcs, "func",
+		     "Set graph filter on given functions", parse_filter_func),
+	OPT_CALLBACK('g', "nograph-funcs", &ftrace.nograph_funcs, "func",
+		     "Set nograph filter on given functions", parse_filter_func),
 	OPT_END()
 	};
 
+	INIT_LIST_HEAD(&ftrace.filters);
+	INIT_LIST_HEAD(&ftrace.notrace);
+	INIT_LIST_HEAD(&ftrace.graph_funcs);
+	INIT_LIST_HEAD(&ftrace.nograph_funcs);
+
 	ret = perf_config(perf_ftrace_config, &ftrace);
 	if (ret < 0)
 		return -1;
@@ -351,12 +448,14 @@ int cmd_ftrace(int argc, const char **argv)
 
 		target__strerror(&ftrace.target, ret, errbuf, 512);
 		pr_err("%s\n", errbuf);
-		return -EINVAL;
+		goto out_delete_filters;
 	}
 
 	ftrace.evlist = perf_evlist__new();
-	if (ftrace.evlist == NULL)
-		return -ENOMEM;
+	if (ftrace.evlist == NULL) {
+		ret = -ENOMEM;
+		goto out_delete_filters;
+	}
 
 	ret = perf_evlist__create_maps(ftrace.evlist, &ftrace.target);
 	if (ret < 0)
@@ -367,5 +466,11 @@ int cmd_ftrace(int argc, const char **argv)
 out_delete_evlist:
 	perf_evlist__delete(ftrace.evlist);
 
+out_delete_filters:
+	delete_filter_func(&ftrace.filters);
+	delete_filter_func(&ftrace.notrace);
+	delete_filter_func(&ftrace.graph_funcs);
+	delete_filter_func(&ftrace.nograph_funcs);
+
 	return ret;
 }
-- 
2.9.4