Re: [PATCH] perf report: Add option to collapse undesired parts ofcall graph

From: Greg Price
Date: Sat Jun 22 2013 - 23:22:57 EST


For example, in an application with an expensive function
implemented with deeply nested recursive calls, the default
call-graph presentation is dominated by the different callchains
within that function. By treating the function as a black box,
we can collect the callchains leading into the function and
compactly identify what to blame for expensive calls.

For example, in this report the callers of garbage_collect() are
scattered across the tree:
$ perf report -d ruby 2>- | grep -m10 ^[^#]*[a-z]
22.03% ruby [.] gc_mark
--- gc_mark
|--59.40%-- mark_keyvalue
| st_foreach
| gc_mark_children
| |--99.75%-- rb_gc_mark
| | rb_vm_mark
| | gc_mark_children
| | gc_marks
| | |--99.00%-- garbage_collect

If we make garbage_collect() a black box, its callers are coalesced:
$ perf report --blackbox garbage_collect -d ruby 2>- | grep -m10 ^[^#]*[a-z]
72.92% ruby [.] garbage_collect
--- garbage_collect
vm_xmalloc
|--47.08%-- ruby_xmalloc
| st_insert2
| rb_hash_aset
| |--98.45%-- features_index_add
| | rb_provide_feature
| | rb_require_safe
| | vm_call_method

Cc: Arnaldo Carvalho de Melo <acme@xxxxxxxxxxxxxxxxxx>
Cc: Peter Zijlstra <a.p.zijlstra@xxxxxxxxx>
Cc: Paul Mackerras <paulus@xxxxxxxxx>
Cc: Ingo Molnar <mingo@xxxxxxxxxx>
Cc: Jiri Olsa <jolsa@xxxxxxxxxx>
Cc: David Ahern <dsahern@xxxxxxxxx>
Signed-off-by: Greg Price <price@xxxxxxx>
---

On Fri, Jan 11, 2013 at 02:27:36AM -0300, Arnaldo Carvalho de Melo wrote:
> Looks like an interesting feature, will review this soon,

Rebased on top of v3.10-rc7, please take a look at your convenience.


tools/perf/builtin-report.c | 19 ++++++++++++++++---
tools/perf/builtin-top.c | 3 +--
tools/perf/util/machine.c | 26 +++++++++++++++++---------
tools/perf/util/machine.h | 9 ++++++++-
tools/perf/util/session.c | 3 +--
5 files changed, 43 insertions(+), 17 deletions(-)

diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c
index bd0ca81..bf0d860 100644
--- a/tools/perf/builtin-report.c
+++ b/tools/perf/builtin-report.c
@@ -83,7 +83,7 @@ static int perf_report__add_mem_hist_entry(struct perf_tool *tool,
if ((sort__has_parent || symbol_conf.use_callchain) &&
sample->callchain) {
err = machine__resolve_callchain(machine, evsel, al->thread,
- sample, &parent);
+ sample, &parent, al);
if (err)
return err;
}
@@ -174,7 +174,7 @@ static int perf_report__add_branch_hist_entry(struct perf_tool *tool,
if ((sort__has_parent || symbol_conf.use_callchain)
&& sample->callchain) {
err = machine__resolve_callchain(machine, evsel, al->thread,
- sample, &parent);
+ sample, &parent, al);
if (err)
return err;
}
@@ -245,7 +245,7 @@ static int perf_evsel__add_hist_entry(struct perf_evsel *evsel,

if ((sort__has_parent || symbol_conf.use_callchain) && sample->callchain) {
err = machine__resolve_callchain(machine, evsel, al->thread,
- sample, &parent);
+ sample, &parent, al);
if (err)
return err;
}
@@ -764,6 +764,8 @@ int cmd_report(int argc, const char **argv, const char *prefix __maybe_unused)
"Default: fractal,0.5,callee", &parse_callchain_opt, callchain_default_opt),
OPT_BOOLEAN('G', "inverted", &report.inverted_callchain,
"alias for inverted call graph"),
+ OPT_STRING(0, "blackbox", &blackbox_pattern, "regex",
+ "functions to treat as black boxes in call graphs, collapsing callees"),
OPT_STRING('d', "dsos", &symbol_conf.dso_list_str, "dso[,dso...]",
"only consider symbols in these dsos"),
OPT_STRING('c', "comms", &symbol_conf.comm_list_str, "comm[,comm...]",
@@ -918,6 +920,17 @@ repeat:
} else
symbol_conf.exclude_other = false;

+ if (blackbox_pattern) {
+ int err = regcomp(&blackbox_regex, blackbox_pattern, REG_EXTENDED);
+ if (err) {
+ char buf[BUFSIZ];
+ regerror(err, &blackbox_regex, buf, sizeof(buf));
+ pr_err("Invalid blackbox regex: %s\n%s", blackbox_pattern, buf);
+ goto error;
+ }
+ have_blackbox = 1;
+ }
+
if (argc) {
/*
* Special case: if there's an argument left then assume that
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 67bdb9f..abec83d 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -775,8 +775,7 @@ static void perf_event__process_sample(struct perf_tool *tool,
sample->callchain) {
err = machine__resolve_callchain(machine, evsel,
al.thread, sample,
- &parent);
-
+ &parent, NULL);
if (err)
return;
}
diff --git a/tools/perf/util/machine.c b/tools/perf/util/machine.c
index b2ecad6..a14489c 100644
--- a/tools/perf/util/machine.c
+++ b/tools/perf/util/machine.c
@@ -11,6 +11,10 @@
#include <stdbool.h>
#include "unwind.h"

+regex_t blackbox_regex;
+const char *blackbox_pattern;
+int have_blackbox = 0;
+
int machine__init(struct machine *machine, const char *root_dir, pid_t pid)
{
map_groups__init(&machine->kmaps);
@@ -1058,11 +1062,10 @@ int machine__process_event(struct machine *machine, union perf_event *event)
return ret;
}

-static bool symbol__match_parent_regex(struct symbol *sym)
+static bool symbol__match_regex(struct symbol *sym, regex_t *regex)
{
- if (sym->name && !regexec(&parent_regex, sym->name, 0, NULL, 0))
+ if (sym->name && !regexec(regex, sym->name, 0, NULL, 0))
return 1;
-
return 0;
}

@@ -1159,8 +1162,8 @@ struct branch_info *machine__resolve_bstack(struct machine *machine,
static int machine__resolve_callchain_sample(struct machine *machine,
struct thread *thread,
struct ip_callchain *chain,
- struct symbol **parent)
-
+ struct symbol **parent,
+ struct addr_location *root_al)
{
u8 cpumode = PERF_RECORD_MISC_USER;
unsigned int i;
@@ -1211,8 +1214,13 @@ static int machine__resolve_callchain_sample(struct machine *machine,
MAP__FUNCTION, ip, &al, NULL);
if (al.sym != NULL) {
if (sort__has_parent && !*parent &&
- symbol__match_parent_regex(al.sym))
+ symbol__match_regex(al.sym, &parent_regex))
*parent = al.sym;
+ else if (have_blackbox && root_al &&
+ symbol__match_regex(al.sym, &blackbox_regex)) {
+ *root_al = al;
+ callchain_cursor_reset(&callchain_cursor);
+ }
if (!symbol_conf.use_callchain)
break;
}
@@ -1237,15 +1245,15 @@ int machine__resolve_callchain(struct machine *machine,
struct perf_evsel *evsel,
struct thread *thread,
struct perf_sample *sample,
- struct symbol **parent)
-
+ struct symbol **parent,
+ struct addr_location *root_al)
{
int ret;

callchain_cursor_reset(&callchain_cursor);

ret = machine__resolve_callchain_sample(machine, thread,
- sample->callchain, parent);
+ sample->callchain, parent, root_al);
if (ret)
return ret;

diff --git a/tools/perf/util/machine.h b/tools/perf/util/machine.h
index 7794068..6f0310a 100644
--- a/tools/perf/util/machine.h
+++ b/tools/perf/util/machine.h
@@ -3,8 +3,14 @@

#include <sys/types.h>
#include <linux/rbtree.h>
+#include <regex.h>
#include "map.h"

+extern regex_t blackbox_regex;
+extern const char *blackbox_pattern;
+extern int have_blackbox;
+
+struct addr_location;
struct branch_stack;
struct perf_evsel;
struct perf_sample;
@@ -83,7 +89,8 @@ int machine__resolve_callchain(struct machine *machine,
struct perf_evsel *evsel,
struct thread *thread,
struct perf_sample *sample,
- struct symbol **parent);
+ struct symbol **parent,
+ struct addr_location *root_al);

/*
* Default guest kernel is defined by parameter --guestkallsyms
diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c
index cf1fe01..7024950 100644
--- a/tools/perf/util/session.c
+++ b/tools/perf/util/session.c
@@ -1397,9 +1397,8 @@ void perf_evsel__print_ip(struct perf_evsel *evsel, union perf_event *event,

if (symbol_conf.use_callchain && sample->callchain) {

-
if (machine__resolve_callchain(machine, evsel, al.thread,
- sample, NULL) != 0) {
+ sample, NULL, NULL) != 0) {
if (verbose)
error("Failed to resolve callchain. Skipping\n");
return;
--
1.8.2
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/