2011-03-28 18:08:29

by Arnaldo Carvalho de Melo

[permalink] [raw]
Subject: [GIT PULL 0/1] perf/urgent fix

Hi Ingo,

Please consider pulling from:

git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6 perf/urgent

I need to go over symbol.c to clean up the way the long name
is set/freed, but Andrew's solves a real problem and thus I think it
should go in now.

Regards,

- Arnaldo

Andrew Lutomirski (1):
perf symbols: Fix vsyscall symbol lookup

tools/perf/util/event.c | 2 +-
tools/perf/util/symbol.c | 3 +++
2 files changed, 4 insertions(+), 1 deletions(-)


2011-03-28 18:08:10

by Arnaldo Carvalho de Melo

[permalink] [raw]
Subject: [PATCH 1/1] perf symbols: Fix vsyscall symbol lookup

From: Andrew Lutomirski <[email protected]>

Perf can't currently trace into the vsyscall page. It looks like it was
meant to work.

Tested on 2.6.38 and today's -git.

The bug is easy to reproduce. Compile this:

int main()
{
int i;
struct timespec t;
for(i = 0; i < 10000000; i++)
clock_gettime(CLOCK_MONOTONIC, &t);
return 0;
}

and run it through perf record; perf report. The top entry shows
"[unknown]" and you can't zoom in.

It looks like there are two issues. The first is a that a test for user
mode executing in kernel space is backwards. (That's the first hunk
below). The second (I think) is that something's wrong with the code
that generates lots of little struct dso objects for different sections
-- when it runs on vmlinux it results in bogus long_name values which
cause objdump to fail.

Cc: Ingo Molnar <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Peter Zijlstra <[email protected]>
LPU-Reference: <[email protected]>
Signed-off-by: Andy Lutomirski <[email protected]>
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
---
tools/perf/util/event.c | 2 +-
tools/perf/util/symbol.c | 3 +++
2 files changed, 4 insertions(+), 1 deletions(-)

diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c
index 2b15c36..1023f67 100644
--- a/tools/perf/util/event.c
+++ b/tools/perf/util/event.c
@@ -710,7 +710,7 @@ try_again:
* in the whole kernel symbol list.
*/
if ((long long)al->addr < 0 &&
- cpumode == PERF_RECORD_MISC_KERNEL &&
+ cpumode == PERF_RECORD_MISC_USER &&
machine && mg != &machine->kmaps) {
mg = &machine->kmaps;
goto try_again;
diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c
index 17df793..8f73907 100644
--- a/tools/perf/util/symbol.c
+++ b/tools/perf/util/symbol.c
@@ -1196,6 +1196,8 @@ static int dso__load_sym(struct dso *self, struct map *map, const char *name,
if (curr_dso == NULL)
goto out_elf_end;
curr_dso->kernel = self->kernel;
+ curr_dso->long_name = self->long_name;
+ curr_dso->long_name_len = self->long_name_len;
curr_map = map__new2(start, curr_dso,
map->type);
if (curr_map == NULL) {
@@ -1842,6 +1844,7 @@ int dso__load_vmlinux(struct dso *self, struct map *map,
if (fd < 0)
return -1;

+ dso__set_long_name(self, (char *)vmlinux);
dso__set_loaded(self, map->type);
err = dso__load_sym(self, map, symfs_vmlinux, fd, filter, 0, 0);
close(fd);
--
1.6.2.5

2011-03-29 07:12:22

by Ingo Molnar

[permalink] [raw]
Subject: Re: [GIT PULL 0/1] perf/urgent fix


* Arnaldo Carvalho de Melo <[email protected]> wrote:

> Hi Ingo,
>
> Please consider pulling from:
>
> git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6 perf/urgent
>
> I need to go over symbol.c to clean up the way the long name
> is set/freed, but Andrew's solves a real problem and thus I think it
> should go in now.
>
> Regards,
>
> - Arnaldo
>
> Andrew Lutomirski (1):
> perf symbols: Fix vsyscall symbol lookup
>
> tools/perf/util/event.c | 2 +-
> tools/perf/util/symbol.c | 3 +++
> 2 files changed, 4 insertions(+), 1 deletions(-)

Pulled, thanks Arnaldo!

Ingo