Skip to contentSkip to conversation

perf probe: Fix --line option behavior

PATCH In review Namhyung Kim 4 participants 9 messages Apr 1, 2014 – Apr 14, 2014 perf events

lore
Share & tools
lore

Namhyung Kim

author maintainer
#0 lore raw

PATCH 1/1 perf probe: Fix --line option behavior

The commit 5a62257a3ddd1 ("perf probe: Replace line_list with intlist") replaced line_list to intlist but it has a problem that if a same line was added again, it'd return -EEXIST rather than 1.

Since line_range_walk_cb() only checks the result being negative, it resulted in failure or segfault sometimes.

Cc Masami Hiramatsu <[email protected]>
---
 tools/perf/util/probe-finder.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
tools/perf/util/probe-finder.c +4 -2
index df0238654698..3bf0c8cdccb7 100644
@@ -1441,13 +1441,15 @@ static int line_range_walk_cb(const char *fname, int lineno,
 			      void *data)
 {
 	struct line_finder *lf = data;
+	int err;
 
 	if ((strtailcmp(fname, lf->fname) != 0) ||
 	    (lf->lno_s > lineno || lf->lno_e < lineno))
 		return 0;
 
-	if (line_range_add_line(fname, lineno, lf->lr) < 0)
-		return -EINVAL;
+	err = line_range_add_line(fname, lineno, lf->lr);
+	if (err < 0 && err != -EEXIST)
+		return err;
 
 	return 0;
 }
Signature · 1 line
1.7.11.7
Acked-by on v1 1/1
#1 lore raw
Quoted 3 lines · Namhyung Kim

The commit 5a62257a3ddd1 ("perf probe: Replace line_list with intlist") replaced line_list to intlist but it has a problem that if a same line was added again, it'd return -EEXIST rather than 1.

Ah, right! that's a different behavior.

Quoted 2 lines

Since line_range_walk_cb() only checks the result being negative, it resulted in failure or segfault sometimes.

Could you give me an example input of the segfault? I'd like to trace it down.

Quoted 2 lines
Cc Masami Hiramatsu <[email protected]>

Thank you very much!

Quoted 27 lines
---
 tools/perf/util/probe-finder.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
tools/perf/util/probe-finder.c +4 -2
index df0238654698..3bf0c8cdccb7 100644
@@ -1441,13 +1441,15 @@ static int line_range_walk_cb(const char *fname, int lineno,
 			      void *data)
 {
 	struct line_finder *lf = data;
+	int err;
 
 	if ((strtailcmp(fname, lf->fname) != 0) ||
 	    (lf->lno_s > lineno || lf->lno_e < lineno))
 		return 0;
 
-	if (line_range_add_line(fname, lineno, lf->lr) < 0)
-		return -EINVAL;
+	err = line_range_add_line(fname, lineno, lf->lr);
+	if (err < 0 && err != -EEXIST)
+		return err;
 
 	return 0;
 }
Signature · 4 lines
Masami HIRAMATSU
IT Management Research Dept. Linux Technology Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: [email protected]
author maintainer
#2 lore raw

Hi Masami,

(Also adding Jiri in CC and changing email of acme)

Quoted 5 lines · Masami Hiramatsu
Quoted 3 lines · Namhyung Kim

The commit 5a62257a3ddd1 ("perf probe: Replace line_list with intlist") replaced line_list to intlist but it has a problem that if a same line was added again, it'd return -EEXIST rather than 1.

Quoted 7 lines

Ah, right! that's a different behavior.

Quoted 2 lines

Since line_range_walk_cb() only checks the result being negative, it resulted in failure or segfault sometimes.

Could you give me an example input of the segfault? I'd like to trace it down.

Just used current acme/perf/core.

$ git log -1
commit 28b5724a61cc9d84f2cbef6675e8a85ae4b1bc57
Author: Arnaldo Carvalho de Melo <[email protected]>
Date:   Thu Mar 27 17:36:14 2014 -0300

MAINTAINERS: Change e-mail to kernel.org one

Leaving ghostprotocols.net for old networking stuff.

...

$ perf --version
perf version 3.14.rc6.g28b572
$ perf probe -x ./perf -v -L map__load
Open Debuginfo file: /home/namhyung/project/linux/tools/perf/perf
fname: util/map.c, lineno:153
New line range: 153 to 2147483647
path: (null)
Segmentation fault (core dumped)

It seems like find_line_range_by_line() called for multiple times, line_range_walk_cb returns 0 at first but as line_range_add_line() returns -EEXIST, it freed lr->path and __show_line_range() ended up accessing to NULL path.

(gdb) bt
#0 get_real_path (new_path=0x839ff8 <params+9272>,

    comp_dir=0x1f994f0 "/home/namhyung/project/linux/tools/perf",
    raw_path=0x0) at util/probe-event.c:518
#1  __show_line_range (module=<optimized out>, lr=0x839fd8 <params+9240>)
    at util/probe-event.c:641
#2  show_line_range (lr=lr@entry=0x839fd8 <params+9240>,
    module=<optimized out>) at util/probe-event.c:699
#3  0x000000000044667b in __cmd_probe (argc=<optimized out>,
    argv=argv@entry=0x7fff7a770ba0, prefix=<optimized out>)
    at builtin-probe.c:465
#4  0x00000000004469f4 in cmd_probe (argc=<optimized out>,
    argv=0x7fff7a770ba0, prefix=<optimized out>) at builtin-probe.c:520
#5  0x00000000004241b5 in run_builtin (p=p@entry=0x7ebde0 <commands+384>,
    argc=argc@entry=6, argv=argv@entry=0x7fff7a770ba0) at perf.c:319
#6  0x0000000000423a29 in handle_internal_command (argv=0x7fff7a770ba0,
    argc=6) at perf.c:376

#7 run_argv (argv=0x7fff7a7709a0, argcp=0x7fff7a7709ac) at perf.c:420 #8 main (argc=6, argv=0x7fff7a770ba0) at perf.c:529

Thanks,
Namhyung

#3 lore raw
Quoted 43 lines · Namhyung Kim

Hi Masami,

(Also adding Jiri in CC and changing email of acme)

Quoted 5 lines · Masami Hiramatsu
Quoted 3 lines · Namhyung Kim

The commit 5a62257a3ddd1 ("perf probe: Replace line_list with intlist") replaced line_list to intlist but it has a problem that if a same line was added again, it'd return -EEXIST rather than 1.

Quoted 7 lines

Ah, right! that's a different behavior.

Quoted 2 lines

Since line_range_walk_cb() only checks the result being negative, it resulted in failure or segfault sometimes.

Could you give me an example input of the segfault? I'd like to trace it down.

Just used current acme/perf/core.

$ git log -1
commit 28b5724a61cc9d84f2cbef6675e8a85ae4b1bc57
Author: Arnaldo Carvalho de Melo <[email protected]>
Date:   Thu Mar 27 17:36:14 2014 -0300

MAINTAINERS: Change e-mail to kernel.org one

Leaving ghostprotocols.net for old networking stuff.

...

$ perf --version
perf version 3.14.rc6.g28b572
$ perf probe -x ./perf -v -L map__load
Open Debuginfo file: /home/namhyung/project/linux/tools/perf/perf
fname: util/map.c, lineno:153
New line range: 153 to 2147483647
path: (null)
Segmentation fault (core dumped)
Quoted 4 lines

It seems like find_line_range_by_line() called for multiple times, line_range_walk_cb returns 0 at first but as line_range_add_line() returns -EEXIST, it freed lr->path and __show_line_range() ended up accessing to NULL path.

OK, thanks for this nice report!
Hmm, line_range_inline_cb() must check the result of find_line_range_by_line() but ignored. That should be fixed too...

Thank you,

Quoted 31 lines

(gdb) bt
#0 get_real_path (new_path=0x839ff8 <params+9272>,

    comp_dir=0x1f994f0 "/home/namhyung/project/linux/tools/perf",
    raw_path=0x0) at util/probe-event.c:518
#1  __show_line_range (module=<optimized out>, lr=0x839fd8 <params+9240>)
    at util/probe-event.c:641
#2  show_line_range (lr=lr@entry=0x839fd8 <params+9240>,
    module=<optimized out>) at util/probe-event.c:699
#3  0x000000000044667b in __cmd_probe (argc=<optimized out>,
    argv=argv@entry=0x7fff7a770ba0, prefix=<optimized out>)
    at builtin-probe.c:465
#4  0x00000000004469f4 in cmd_probe (argc=<optimized out>,
    argv=0x7fff7a770ba0, prefix=<optimized out>) at builtin-probe.c:520
#5  0x00000000004241b5 in run_builtin (p=p@entry=0x7ebde0 <commands+384>,
    argc=argc@entry=6, argv=argv@entry=0x7fff7a770ba0) at perf.c:319
#6  0x0000000000423a29 in handle_internal_command (argv=0x7fff7a770ba0,
    argc=6) at perf.c:376

#7 run_argv (argv=0x7fff7a7709a0, argcp=0x7fff7a7709ac) at perf.c:420 #8 main (argc=6, argv=0x7fff7a770ba0) at perf.c:529

Thanks,
Namhyung
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

Signature · 4 lines
Masami HIRAMATSU
IT Management Research Dept. Linux Technology Center
Hitachi, Ltd., Yokohama Research Laboratory
E-mail: [email protected]
#4 lore raw

PATCH 1/1 perf/probe: Fix to handle errors in line_range searching

In reply to Namhyung Kim

As Namhyung reported(https://lkml.org/lkml/2014/4/1/89), current perf-probe -L option doesn't handle errors in line-range searching correctly. It causes a SEGV if an error occured in the line-range searching.

----

  $ perf probe -x ./perf -v -L map__load
  Open Debuginfo file: /home/namhyung/project/linux/tools/perf/perf
  fname: util/map.c, lineno:153
  New line range: 153 to 2147483647
  path: (null)
  Segmentation fault (core dumped)
  ----

This is because line_range_inline_cb() ignores errors from find_line_range_by_line() which means that lr->path is already freed on the error path in find_line_range_by_line(). As a result, get_real_path() accesses the lr->path and it causes a NULL pointer exception.

This fixes line_range_inline_cb() to handle the error correctly, and report it to the caller.

Anyway, this just fixes a possible SEGV bug, Namhyung's patch is also required.

Cc Arnaldo Carvalho de Melo <[email protected]>
Cc Peter Zijlstra <[email protected]>
Cc Ingo Molnar <[email protected]>
Cc Paul Mackerras <[email protected]>
Cc Namhyung Kim <[email protected]>
Cc Jiri Olsa <[email protected]>
---
 tools/perf/util/probe-finder.c |    7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
tools/perf/util/probe-finder.c +4 -3
index df02386..edcd2ff 100644
@@ -1473,14 +1473,15 @@ static int find_line_range_by_line(Dwarf_Die *sp_die, struct line_finder *lf)
 
 static int line_range_inline_cb(Dwarf_Die *in_die, void *data)
 {
-	find_line_range_by_line(in_die, data);
+	int ret = find_line_range_by_line(in_die, data);
 
 	/*
 	 * We have to check all instances of inlined function, because
 	 * some execution paths can be optimized out depends on the
-	 * function argument of instances
+	 * function argument of instances. However, if an error occurs,
+	 * it should be handled by the caller.
 	 */
-	return 0;
+	return ret < 0 ? ret : 0;
 }
 
 /* Search function definition from function name */
#5 lore raw

Re: [PATCH -tip ] [BUGFIX] perf/probe: Fix to handle errors in line_range searching

Quoted 20 lines · Masami Hiramatsu

As Namhyung reported(https://lkml.org/lkml/2014/4/1/89), current perf-probe -L option doesn't handle errors in line-range searching correctly. It causes a SEGV if an error occured in the line-range searching.

----

  $ perf probe -x ./perf -v -L map__load
  Open Debuginfo file: /home/namhyung/project/linux/tools/perf/perf
  fname: util/map.c, lineno:153
  New line range: 153 to 2147483647
  path: (null)
  Segmentation fault (core dumped)
  ----

This is because line_range_inline_cb() ignores errors from find_line_range_by_line() which means that lr->path is already freed on the error path in find_line_range_by_line(). As a result, get_real_path() accesses the lr->path and it causes a NULL pointer exception.

Quoted 5 lines

This fixes line_range_inline_cb() to handle the error correctly, and report it to the caller.

Anyway, this just fixes a possible SEGV bug, Namhyung's patch is also required.

will take both.. Namhyung, ack to Masami's change?

thanks,
jirka

author maintainer Acked-by
#6 lore raw

Re: [PATCH -tip ] [BUGFIX] perf/probe: Fix to handle errors in line_range searching

Quoted 21 lines · Jiri Olsa
Quoted 20 lines · Masami Hiramatsu

As Namhyung reported(https://lkml.org/lkml/2014/4/1/89), current perf-probe -L option doesn't handle errors in line-range searching correctly. It causes a SEGV if an error occured in the line-range searching.

----

  $ perf probe -x ./perf -v -L map__load
  Open Debuginfo file: /home/namhyung/project/linux/tools/perf/perf
  fname: util/map.c, lineno:153
  New line range: 153 to 2147483647
  path: (null)
  Segmentation fault (core dumped)
  ----

This is because line_range_inline_cb() ignores errors from find_line_range_by_line() which means that lr->path is already freed on the error path in find_line_range_by_line(). As a result, get_real_path() accesses the lr->path and it causes a NULL pointer exception.

Quoted 7 lines
Quoted 5 lines

This fixes line_range_inline_cb() to handle the error correctly, and report it to the caller.

Anyway, this just fixes a possible SEGV bug, Namhyung's patch is also required.

will take both.. Namhyung, ack to Masami's change?

Yes,

Thanks,
Namhyung

#7 lore raw

[tip:perf/urgent] perf probe: Fix to handle errors in line_range searching

In reply to Masami Hiramatsu
perf probe: Fix to handle errors in line_range searching
Commit-ID:  182c228ebcf1ac67a44e62236d8f7a8a9a3c5699
Gitweb:     http://git.kernel.org/tip/182c228ebcf1ac67a44e62236d8f7a8a9a3c5699
Author:     Masami Hiramatsu <[email protected]>
AuthorDate: Wed, 2 Apr 2014 14:48:31 +0900
Committer:  Jiri Olsa <[email protected]>
CommitDate: Mon, 14 Apr 2014 12:55:39 +0200

perf probe: Fix to handle errors in line_range searching

As Namhyung reported(https://lkml.org/lkml/2014/4/1/89), current perf-probe -L option doesn't handle errors in line-range searching correctly. It causes a SEGV if an error occured in the line-range searching.

----

  $ perf probe -x ./perf -v -L map__load
  Open Debuginfo file: /home/namhyung/project/linux/tools/perf/perf
  fname: util/map.c, lineno:153
  New line range: 153 to 2147483647
  path: (null)
  Segmentation fault (core dumped)
  ----

This is because line_range_inline_cb() ignores errors from find_line_range_by_line() which means that lr->path is already freed on the error path in find_line_range_by_line(). As a result, get_real_path() accesses the lr->path and it causes a NULL pointer exception.

This fixes line_range_inline_cb() to handle the error correctly, and report it to the caller.

Anyway, this just fixes a possible SEGV bug, Namhyung's patch is also required.

---
 tools/perf/util/probe-finder.c | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)
tools/perf/util/probe-finder.c +4 -3
index 3bf0c8c..fae274e 100644
@@ -1475,14 +1475,15 @@ static int find_line_range_by_line(Dwarf_Die *sp_die, struct line_finder *lf)
 
 static int line_range_inline_cb(Dwarf_Die *in_die, void *data)
 {
-	find_line_range_by_line(in_die, data);
+	int ret = find_line_range_by_line(in_die, data);
 
 	/*
 	 * We have to check all instances of inlined function, because
 	 * some execution paths can be optimized out depends on the
-	 * function argument of instances
+	 * function argument of instances. However, if an error occurs,
+	 * it should be handled by the caller.
 	 */
-	return 0;
+	return ret < 0 ? ret : 0;
 }
 
 /* Search function definition from function name */
#8 lore raw
In reply to Namhyung Kim
perf probe: Fix --line option behavior
Commit-ID:  202c7c123c96a1c193149b7fa2718d7fb143efb2
Gitweb:     http://git.kernel.org/tip/202c7c123c96a1c193149b7fa2718d7fb143efb2
Author:     Namhyung Kim <[email protected]>
AuthorDate: Tue, 1 Apr 2014 13:47:57 +0900
Committer:  Jiri Olsa <[email protected]>
CommitDate: Mon, 14 Apr 2014 12:55:39 +0200

perf probe: Fix --line option behavior

The commit 5a62257a3ddd1 ("perf probe: Replace line_list with intlist") replaced line_list to intlist but it has a problem that if a same line was added again, it'd return -EEXIST rather than 1.

Since line_range_walk_cb() only checks the result being negative, it resulted in failure or segfault sometimes.

---
 tools/perf/util/probe-finder.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)
tools/perf/util/probe-finder.c +4 -2
index df02386..3bf0c8c 100644
@@ -1441,13 +1441,15 @@ static int line_range_walk_cb(const char *fname, int lineno,
 			      void *data)
 {
 	struct line_finder *lf = data;
+	int err;
 
 	if ((strtailcmp(fname, lf->fname) != 0) ||
 	    (lf->lno_s > lineno || lf->lno_e < lineno))
 		return 0;
 
-	if (line_range_add_line(fname, lineno, lf->lr) < 0)
-		return -EINVAL;
+	err = line_range_add_line(fname, lineno, lf->lr);
+	if (err < 0 && err != -EEXIST)
+		return err;
 
 	return 0;
 }