Hi Ingo,
Please consider pulling from:
git://git.kernel.org/pub/scm/linux/kernel/git/acme/linux-2.6 perf/core
This one has everything in acme/perf/urgent, i.e. starts the next
tools/ devel cycle :-)
Regards,
- Arnaldo
Arnaldo Carvalho de Melo (1):
perf tools: Fixup exit path when not able to open events
David Ahern (1):
perf tools: Emit clearer message for sys_perf_event_open ENOENT
return
Lin Ming (1):
perf probe: Add fastpath to do lookup by function name
tools/perf/builtin-record.c | 19 +++++++---
tools/perf/builtin-top.c | 44 ++++++++++++++++--------
tools/perf/util/debug.c | 10 +++++
tools/perf/util/debug.h | 1 +
tools/perf/util/probe-finder.c | 72 ++++++++++++++++++++++++++++++++++++++++
tools/perf/util/probe-finder.h | 2 +
6 files changed, 128 insertions(+), 20 deletions(-)
From: David Ahern <[email protected]>
Resend of patch sent back in January 2011 in light of recent confusion around
unsupported events for a given platform.
Improve sys_perf_event_open ENOENT return handling in top and record, just
like 5a3446b does for stat.
Retry of Arnaldo's patch using ui_warning instead of die which allows the
fallback from hardware cycles to software clock.
Cc: Frederic Weisbecker <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Thomas Gleixner <[email protected]>
Cc: [email protected]
LKML-Reference: <[email protected]>
Signed-off-by: David Ahern <[email protected]>
[ committer note: Some adjustments to make it apply to newer codebase ]
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
---
tools/perf/builtin-record.c | 10 +++++++++-
tools/perf/builtin-top.c | 6 ++++++
2 files changed, 15 insertions(+), 1 deletions(-)
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 623695e..db6adec 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -301,11 +301,19 @@ try_again:
&& attr->config == PERF_COUNT_HW_CPU_CYCLES) {
if (verbose)
- warning(" ... trying to fall back to cpu-clock-ticks\n");
+ ui__warning("The cycles event is not supported, "
+ "trying to fall back to cpu-clock-ticks\n");
attr->type = PERF_TYPE_SOFTWARE;
attr->config = PERF_COUNT_SW_CPU_CLOCK;
goto try_again;
}
+
+ if (err == ENOENT) {
+ ui__warning("The %s event is not supported.\n",
+ event_name(pos));
+ exit(EXIT_FAILURE);
+ }
+
printf("\n");
error("sys_perf_event_open() syscall returned with %d (%s). /bin/dmesg may provide additional information.\n",
err, strerror(err));
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 935fc4f..fc1273e 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -870,6 +870,12 @@ try_again:
goto try_again;
}
+ if (err == ENOENT) {
+ ui__warning("The %s event is not supported.\n",
+ event_name(counter));
+ goto out_err;
+ }
+
ui__warning("The sys_perf_event_open() syscall "
"returned with %d (%s). /bin/dmesg "
"may provide additional information.\n"
--
1.6.2.5
From: Lin Ming <[email protected]>
v3 -> v2:
- Make pubname_search_cb more generic
- Add fastpath to find_probes also
v2 -> v1:
- Don't compare file names with cu_find_realpath(...), instead, compare
them with the name returned by dwarf_decl_file(sp_die)
The vmlinux file may have thousands of CUs.
We can lookup function name from .debug_pubnames section
to avoid the slow loop on CUs.
1. Improvement data for find_line_range
./perf stat -e cycles -r 10 -- ./perf probe -k /home/mlin/vmlinux \
-s /home/mlin/linux-2.6 \
--line csum_partial_copy_to_user > tmp.log
before patch applied
=====================
847,988,276 cycles
0.355075856 seconds time elapsed
after patch applied
=====================
206,102,622 cycles
0.086883555 seconds time elapsed
2. Improvement data for find_probes
./perf stat -e cycles -r 10 -- ./perf probe -k /home/mlin/vmlinux \
-s /home/mlin/linux-2.6 \
--vars csum_partial_copy_to_user > tmp.log
before patch applied
=====================
848,490,844 cycles
0.355307901 seconds time elapsed
after patch applied
=====================
205,684,469 cycles
0.086694010 seconds time elapsed
Acked-by: Masami Hiramatsu <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Masami Hiramatsu <[email protected]>
Cc: Peter Zijlstra <[email protected]>
LKML-Reference: <[email protected]>
Signed-off-by: Lin Ming <[email protected]>
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
---
tools/perf/util/probe-finder.c | 72 ++++++++++++++++++++++++++++++++++++++++
tools/perf/util/probe-finder.h | 2 +
2 files changed, 74 insertions(+), 0 deletions(-)
diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c
index 194f9e2..ff416b8 100644
--- a/tools/perf/util/probe-finder.c
+++ b/tools/perf/util/probe-finder.c
@@ -1435,6 +1435,38 @@ static int find_probe_point_by_func(struct probe_finder *pf)
return _param.retval;
}
+struct pubname_callback_param {
+ char *function;
+ char *file;
+ Dwarf_Die *cu_die;
+ Dwarf_Die *sp_die;
+ int found;
+};
+
+static int pubname_search_cb(Dwarf *dbg, Dwarf_Global *gl, void *data)
+{
+ struct pubname_callback_param *param = data;
+
+ if (dwarf_offdie(dbg, gl->die_offset, param->sp_die)) {
+ if (dwarf_tag(param->sp_die) != DW_TAG_subprogram)
+ return DWARF_CB_OK;
+
+ if (die_compare_name(param->sp_die, param->function)) {
+ if (!dwarf_offdie(dbg, gl->cu_offset, param->cu_die))
+ return DWARF_CB_OK;
+
+ if (param->file &&
+ strtailcmp(param->file, dwarf_decl_file(param->sp_die)))
+ return DWARF_CB_OK;
+
+ param->found = 1;
+ return DWARF_CB_ABORT;
+ }
+ }
+
+ return DWARF_CB_OK;
+}
+
/* Find probe points from debuginfo */
static int find_probes(int fd, struct probe_finder *pf)
{
@@ -1461,6 +1493,27 @@ static int find_probes(int fd, struct probe_finder *pf)
off = 0;
line_list__init(&pf->lcache);
+
+ /* Fastpath: lookup by function name from .debug_pubnames section */
+ if (pp->function) {
+ struct pubname_callback_param pubname_param = {
+ .function = pp->function,
+ .file = pp->file,
+ .cu_die = &pf->cu_die,
+ .sp_die = &pf->sp_die,
+ };
+ struct dwarf_callback_param probe_param = {
+ .data = pf,
+ };
+
+ dwarf_getpubnames(dbg, pubname_search_cb, &pubname_param, 0);
+ if (pubname_param.found) {
+ ret = probe_point_search_cb(&pf->sp_die, &probe_param);
+ if (ret)
+ goto found;
+ }
+ }
+
/* Loop on CUs (Compilation Unit) */
while (!dwarf_nextcu(dbg, off, &noff, &cuhl, NULL, NULL, NULL)) {
/* Get the DIE(Debugging Information Entry) of this CU */
@@ -1488,6 +1541,8 @@ static int find_probes(int fd, struct probe_finder *pf)
}
off = noff;
}
+
+found:
line_list__free(&pf->lcache);
if (dwfl)
dwfl_end(dwfl);
@@ -1895,6 +1950,22 @@ int find_line_range(int fd, struct line_range *lr)
return -EBADF;
}
+ /* Fastpath: lookup by function name from .debug_pubnames section */
+ if (lr->function) {
+ struct pubname_callback_param pubname_param = {
+ .function = lr->function, .file = lr->file,
+ .cu_die = &lf.cu_die, .sp_die = &lf.sp_die, .found = 0};
+ struct dwarf_callback_param line_range_param = {
+ .data = (void *)&lf, .retval = 0};
+
+ dwarf_getpubnames(dbg, pubname_search_cb, &pubname_param, 0);
+ if (pubname_param.found) {
+ line_range_search_cb(&lf.sp_die, &line_range_param);
+ if (lf.found)
+ goto found;
+ }
+ }
+
/* Loop on CUs (Compilation Unit) */
while (!lf.found && ret >= 0) {
if (dwarf_nextcu(dbg, off, &noff, &cuhl, NULL, NULL, NULL) != 0)
@@ -1923,6 +1994,7 @@ int find_line_range(int fd, struct line_range *lr)
off = noff;
}
+found:
/* Store comp_dir */
if (lf.found) {
comp_dir = cu_get_comp_dir(&lf.cu_die);
diff --git a/tools/perf/util/probe-finder.h b/tools/perf/util/probe-finder.h
index beaefc3..605730a 100644
--- a/tools/perf/util/probe-finder.h
+++ b/tools/perf/util/probe-finder.h
@@ -49,6 +49,7 @@ struct probe_finder {
Dwarf_Addr addr; /* Address */
const char *fname; /* Real file name */
Dwarf_Die cu_die; /* Current CU */
+ Dwarf_Die sp_die;
struct list_head lcache; /* Line cache for lazy match */
/* For variable searching */
@@ -83,6 +84,7 @@ struct line_finder {
int lno_s; /* Start line number */
int lno_e; /* End line number */
Dwarf_Die cu_die; /* Current CU */
+ Dwarf_Die sp_die;
int found;
};
--
1.6.2.5
From: Arnaldo Carvalho de Melo <[email protected]>
We have to deal with the TUI mode in perf top, so that we don't end up
with a garbled screen when, say, a non root user on a machine with a
paranoid setting (the default) tries to use 'perf top'.
Introduce a ui__warning_paranoid() routine shared by top and record that
tells the user the valid values for /proc/sys/kernel/perf_event_paranoid.
Cc: David Ahern <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Ingo Molnar <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Peter Zijlstra <[email protected]>
Cc: Stephane Eranian <[email protected]>
Cc: Tom Zanussi <[email protected]>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <[email protected]>
---
tools/perf/builtin-record.c | 9 ++++-----
tools/perf/builtin-top.c | 38 ++++++++++++++++++++++++--------------
tools/perf/util/debug.c | 10 ++++++++++
tools/perf/util/debug.h | 1 +
4 files changed, 39 insertions(+), 19 deletions(-)
diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c
index 6febcc1..623695e 100644
--- a/tools/perf/builtin-record.c
+++ b/tools/perf/builtin-record.c
@@ -275,11 +275,10 @@ try_again:
!no_inherit) < 0) {
int err = errno;
- if (err == EPERM || err == EACCES)
- die("Permission error - are you root?\n"
- "\t Consider tweaking"
- " /proc/sys/kernel/perf_event_paranoid.\n");
- else if (err == ENODEV && cpu_list) {
+ if (err == EPERM || err == EACCES) {
+ ui__warning_paranoid();
+ exit(EXIT_FAILURE);
+ } else if (err == ENODEV && cpu_list) {
die("No such device - did you specify"
" an out-of-range profile CPU?\n");
} else if (err == EINVAL && sample_id_all_avail) {
diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c
index 676b4fb..935fc4f 100644
--- a/tools/perf/builtin-top.c
+++ b/tools/perf/builtin-top.c
@@ -850,10 +850,10 @@ try_again:
top.evlist->threads, group, inherit) < 0) {
int err = errno;
- if (err == EPERM || err == EACCES)
- die("Permission error - are you root?\n"
- "\t Consider tweaking"
- " /proc/sys/kernel/perf_event_paranoid.\n");
+ if (err == EPERM || err == EACCES) {
+ ui__warning_paranoid();
+ goto out_err;
+ }
/*
* If it's cycles then fall back to hrtimer
* based cpu-clock-tick sw counter, which
@@ -861,25 +861,35 @@ try_again:
*/
if (attr->type == PERF_TYPE_HARDWARE &&
attr->config == PERF_COUNT_HW_CPU_CYCLES) {
-
if (verbose)
- warning(" ... trying to fall back to cpu-clock-ticks\n");
+ ui__warning("Cycles event not supported,\n"
+ "trying to fall back to cpu-clock-ticks\n");
attr->type = PERF_TYPE_SOFTWARE;
attr->config = PERF_COUNT_SW_CPU_CLOCK;
goto try_again;
}
- printf("\n");
- error("sys_perf_event_open() syscall returned with %d "
- "(%s). /bin/dmesg may provide additional information.\n",
- err, strerror(err));
- die("No CONFIG_PERF_EVENTS=y kernel support configured?\n");
- exit(-1);
+
+ ui__warning("The sys_perf_event_open() syscall "
+ "returned with %d (%s). /bin/dmesg "
+ "may provide additional information.\n"
+ "No CONFIG_PERF_EVENTS=y kernel support "
+ "configured?\n", err, strerror(err));
+ goto out_err;
}
}
- if (perf_evlist__mmap(evlist, mmap_pages, false) < 0)
- die("failed to mmap with %d (%s)\n", errno, strerror(errno));
+ if (perf_evlist__mmap(evlist, mmap_pages, false) < 0) {
+ ui__warning("Failed to mmap with %d (%s)\n",
+ errno, strerror(errno));
+ goto out_err;
+ }
+
+ return;
+
+out_err:
+ exit_browser(0);
+ exit(0);
}
static int __cmd_top(void)
diff --git a/tools/perf/util/debug.c b/tools/perf/util/debug.c
index d4536a9..155749d 100644
--- a/tools/perf/util/debug.c
+++ b/tools/perf/util/debug.c
@@ -57,6 +57,16 @@ void ui__warning(const char *format, ...)
}
#endif
+void ui__warning_paranoid(void)
+{
+ ui__warning("Permission error - are you root?\n"
+ "Consider tweaking /proc/sys/kernel/perf_event_paranoid:\n"
+ " -1 - Not paranoid at all\n"
+ " 0 - Disallow raw tracepoint access for unpriv\n"
+ " 1 - Disallow cpu events for unpriv\n"
+ " 2 - Disallow kernel profiling for unpriv\n");
+}
+
void trace_event(union perf_event *event)
{
unsigned char *raw_event = (void *)event;
diff --git a/tools/perf/util/debug.h b/tools/perf/util/debug.h
index 93516cf..fd53db4 100644
--- a/tools/perf/util/debug.h
+++ b/tools/perf/util/debug.h
@@ -36,5 +36,6 @@ int ui_helpline__show_help(const char *format, va_list ap);
#endif
void ui__warning(const char *format, ...) __attribute__((format(printf, 1, 2)));
+void ui__warning_paranoid(void);
#endif /* __PERF_DEBUG_H */
--
1.6.2.5
* 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/core
>
> This one has everything in acme/perf/urgent, i.e. starts the next
> tools/ devel cycle :-)
>
> Regards,
>
> - Arnaldo
>
> Arnaldo Carvalho de Melo (1):
> perf tools: Fixup exit path when not able to open events
>
> David Ahern (1):
> perf tools: Emit clearer message for sys_perf_event_open ENOENT
> return
>
> Lin Ming (1):
> perf probe: Add fastpath to do lookup by function name
>
> tools/perf/builtin-record.c | 19 +++++++---
> tools/perf/builtin-top.c | 44 ++++++++++++++++--------
> tools/perf/util/debug.c | 10 +++++
> tools/perf/util/debug.h | 1 +
> tools/perf/util/probe-finder.c | 72 ++++++++++++++++++++++++++++++++++++++++
> tools/perf/util/probe-finder.h | 2 +
> 6 files changed, 128 insertions(+), 20 deletions(-)
Pulled, thanks Arnaldo!
Note, i merged these two also into perf/urgent:
ca6a42586fae: perf tools: Emit clearer message for sys_perf_event_open ENOENT return
c286c419c784: perf tools: Fixup exit path when not able to open events
as both are simple and fix real bugs.
Thanks,
Ingo