License cleanup: add SPDX GPL-2.0 license identifier to files with no license
Many source files in the tree are missing licensing information, which
makes it harder for compliance tools to determine the correct license.
By default all files without license information are under the default
license of the kernel, which is GPL version 2.
Update the files which contain no license information with the 'GPL-2.0'
SPDX license identifier. The SPDX identifier is a legally binding
shorthand, which can be used instead of the full boiler plate text.
This patch is based on work done by Thomas Gleixner and Kate Stewart and
Philippe Ombredanne.
How this work was done:
Patches were generated and checked against linux-4.14-rc6 for a subset of
the use cases:
- file had no licensing information it it.
- file was a */uapi/* one with no licensing information in it,
- file was a */uapi/* one with existing licensing information,
Further patches will be generated in subsequent months to fix up cases
where non-standard license headers were used, and references to license
had to be inferred by heuristics based on keywords.
The analysis to determine which SPDX License Identifier to be applied to
a file was done in a spreadsheet of side by side results from of the
output of two independent scanners (ScanCode & Windriver) producing SPDX
tag:value files created by Philippe Ombredanne. Philippe prepared the
base worksheet, and did an initial spot review of a few 1000 files.
The 4.13 kernel was the starting point of the analysis with 60,537 files
assessed. Kate Stewart did a file by file comparison of the scanner
results in the spreadsheet to determine which SPDX license identifier(s)
to be applied to the file. She confirmed any determination that was not
immediately clear with lawyers working with the Linux Foundation.
Criteria used to select files for SPDX license identifier tagging was:
- Files considered eligible had to be source code files.
- Make and config files were included as candidates if they contained >5
lines of source
- File already had some variant of a license header in it (even if <5
lines).
All documentation files were explicitly excluded.
The following heuristics were used to determine which SPDX license
identifiers to apply.
- when both scanners couldn't find any license traces, file was
considered to have no license information in it, and the top level
COPYING file license applied.
For non */uapi/* files that summary was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 11139
and resulted in the first patch in this series.
If that file was a */uapi/* path one, it was "GPL-2.0 WITH
Linux-syscall-note" otherwise it was "GPL-2.0". Results of that was:
SPDX license identifier # files
---------------------------------------------------|-------
GPL-2.0 WITH Linux-syscall-note 930
and resulted in the second patch in this series.
- if a file had some form of licensing information in it, and was one
of the */uapi/* ones, it was denoted with the Linux-syscall-note if
any GPL family license was found in the file or had no licensing in
it (per prior point). Results summary:
SPDX license identifier # files
---------------------------------------------------|------
GPL-2.0 WITH Linux-syscall-note 270
GPL-2.0+ WITH Linux-syscall-note 169
((GPL-2.0 WITH Linux-syscall-note) OR BSD-2-Clause) 21
((GPL-2.0 WITH Linux-syscall-note) OR BSD-3-Clause) 17
LGPL-2.1+ WITH Linux-syscall-note 15
GPL-1.0+ WITH Linux-syscall-note 14
((GPL-2.0+ WITH Linux-syscall-note) OR BSD-3-Clause) 5
LGPL-2.0+ WITH Linux-syscall-note 4
LGPL-2.1 WITH Linux-syscall-note 3
((GPL-2.0 WITH Linux-syscall-note) OR MIT) 3
((GPL-2.0 WITH Linux-syscall-note) AND MIT) 1
and that resulted in the third patch in this series.
- when the two scanners agreed on the detected license(s), that became
the concluded license(s).
- when there was disagreement between the two scanners (one detected a
license but the other didn't, or they both detected different
licenses) a manual inspection of the file occurred.
- In most cases a manual inspection of the information in the file
resulted in a clear resolution of the license that should apply (and
which scanner probably needed to revisit its heuristics).
- When it was not immediately clear, the license identifier was
confirmed with lawyers working with the Linux Foundation.
- If there was any question as to the appropriate license identifier,
the file was flagged for further research and to be revisited later
in time.
In total, over 70 hours of logged manual review was done on the
spreadsheet to determine the SPDX license identifiers to apply to the
source files by Kate, Philippe, Thomas and, in some cases, confirmation
by lawyers working with the Linux Foundation.
Kate also obtained a third independent scan of the 4.13 code base from
FOSSology, and compared selected files where the other two scanners
disagreed against that SPDX file, to see if there was new insights. The
Windriver scanner is based on an older version of FOSSology in part, so
they are related.
Thomas did random spot checks in about 500 files from the spreadsheets
for the uapi headers and agreed with SPDX license identifier in the
files he inspected. For the non-uapi files Thomas did random spot checks
in about 15000 files.
In initial set of patches against 4.14-rc6, 3 files were found to have
copy/paste license identifier errors, and have been fixed to reflect the
correct identifier.
Additionally Philippe spent 10 hours this week doing a detailed manual
inspection and review of the 12,461 patched files from the initial patch
version early this week with:
- a full scancode scan run, collecting the matched texts, detected
license ids and scores
- reviewing anything where there was a license detected (about 500+
files) to ensure that the applied SPDX license was correct
- reviewing anything where there was no detection but the patch license
was not GPL-2.0 WITH Linux-syscall-note to ensure that the applied
SPDX license was correct
This produced a worksheet with 20 files needing minor correction. This
worksheet was then exported into 3 different .csv files for the
different types of files to be modified.
These .csv files were then reviewed by Greg. Thomas wrote a script to
parse the csv files and add the proper SPDX tag to the file, in the
format that the file expected. This script was further refined by Greg
based on the output to detect more types of files automatically and to
distinguish between header and source .c files (which need different
comment types.) Finally Greg ran the script using the .csv files to
generate the patches.
Reviewed-by: Kate Stewart <kstewart@linuxfoundation.org>
Reviewed-by: Philippe Ombredanne <pombredanne@nexb.com>
Reviewed-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2017-11-01 14:07:57 +00:00
|
|
|
// SPDX-License-Identifier: GPL-2.0
|
2017-04-18 13:46:11 +00:00
|
|
|
#include <errno.h>
|
2017-04-17 18:23:08 +00:00
|
|
|
#include <inttypes.h>
|
2017-04-18 15:33:30 +00:00
|
|
|
#include <regex.h>
|
2014-06-01 13:38:29 +00:00
|
|
|
#include <sys/mman.h>
|
2009-09-24 16:02:49 +00:00
|
|
|
#include "sort.h"
|
2010-07-20 17:42:52 +00:00
|
|
|
#include "hist.h"
|
2013-09-13 07:28:57 +00:00
|
|
|
#include "comm.h"
|
2013-04-03 12:26:19 +00:00
|
|
|
#include "symbol.h"
|
2017-04-20 00:34:35 +00:00
|
|
|
#include "thread.h"
|
2014-03-03 02:46:55 +00:00
|
|
|
#include "evsel.h"
|
2015-12-22 17:07:01 +00:00
|
|
|
#include "evlist.h"
|
2017-04-18 13:57:25 +00:00
|
|
|
#include "strlist.h"
|
2015-12-22 17:07:01 +00:00
|
|
|
#include <traceevent/event-parse.h>
|
2016-02-24 08:46:46 +00:00
|
|
|
#include "mem-events.h"
|
2017-04-17 14:39:06 +00:00
|
|
|
#include <linux/kernel.h>
|
2009-09-24 16:02:49 +00:00
|
|
|
|
|
|
|
regex_t parent_regex;
|
2010-05-17 19:22:41 +00:00
|
|
|
const char default_parent_pattern[] = "^sys_|^do_page_fault";
|
|
|
|
const char *parent_pattern = default_parent_pattern;
|
2016-08-12 23:41:01 +00:00
|
|
|
const char *default_sort_order = "comm,dso,symbol";
|
2015-07-18 15:24:53 +00:00
|
|
|
const char default_branch_sort_order[] = "comm,dso_from,symbol_from,symbol_to,cycles";
|
2014-03-18 02:31:39 +00:00
|
|
|
const char default_mem_sort_order[] = "local_weight,mem,sym,dso,symbol_daddr,dso_daddr,snoop,tlb,locked";
|
|
|
|
const char default_top_sort_order[] = "dso,symbol";
|
|
|
|
const char default_diff_sort_order[] = "dso,symbol";
|
2015-12-22 17:07:10 +00:00
|
|
|
const char default_tracepoint_sort_order[] = "trace";
|
2014-03-18 02:31:39 +00:00
|
|
|
const char *sort_order;
|
2014-03-04 01:46:34 +00:00
|
|
|
const char *field_order;
|
2012-12-07 05:48:05 +00:00
|
|
|
regex_t ignore_callees_regex;
|
|
|
|
int have_ignore_callees = 0;
|
2013-04-01 11:35:20 +00:00
|
|
|
enum sort_mode sort__mode = SORT_MODE__NORMAL;
|
perf tools: Bind callchains to the first sort dimension column
Currently, the callchains are displayed using a constant left
margin. So depending on the current sort dimension
configuration, callchains may appear to be well attached to the
first sort dimension column field which is mostly the case,
except when the first dimension of sorting is done by comm,
because these are right aligned.
This patch binds the callchain to the first letter in the first
column, whatever type of column it is (dso, comm, symbol).
Before:
0.80% perf [k] __lock_acquire
__lock_acquire
lock_acquire
|
|--58.33%-- _spin_lock
| |
| |--28.57%-- inotify_should_send_event
| | fsnotify
| | __fsnotify_parent
After:
0.80% perf [k] __lock_acquire
__lock_acquire
lock_acquire
|
|--58.33%-- _spin_lock
| |
| |--28.57%-- inotify_should_send_event
| | fsnotify
| | __fsnotify_parent
Also, for clarity, we don't put anymore the callchain as is but:
- If we have a top level ancestor in the callchain, start it
with a first ascii hook.
Before:
0.80% perf [kernel] [k] __lock_acquire
__lock_acquire
lock_acquire
|
|--58.33%-- _spin_lock
| |
| |--28.57%-- inotify_should_send_event
| | fsnotify
[..] [..]
After:
0.80% perf [kernel] [k] __lock_acquire
|
--- __lock_acquire
lock_acquire
|
|--58.33%-- _spin_lock
| |
| |--28.57%-- inotify_should_send_event
| | fsnotify
[..] [..]
- Otherwise, if we have several top level ancestors, then
display these like we did before:
1.69% Xorg
|
|--21.21%-- vread_hpet
| 0x7fffd85b46fc
| 0x7fffd85b494d
| 0x7f4fafb4e54d
|
|--15.15%-- exaOffscreenAlloc
|
|--9.09%-- I830WaitLpRing
Signed-off-by: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Anton Blanchard <anton@samba.org>
LKML-Reference: <1256246604-17156-2-git-send-email-fweisbec@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-10-22 21:23:23 +00:00
|
|
|
|
2016-02-12 14:27:51 +00:00
|
|
|
/*
|
|
|
|
* Replaces all occurrences of a char used with the:
|
|
|
|
*
|
|
|
|
* -t, --field-separator
|
|
|
|
*
|
|
|
|
* option, that uses a special separator character and don't pad with spaces,
|
|
|
|
* replacing all occurances of this separator in symbol names (and other
|
|
|
|
* output) with a '.' character, that thus it's the only non valid separator.
|
|
|
|
*/
|
2010-03-31 14:33:40 +00:00
|
|
|
static int repsep_snprintf(char *bf, size_t size, const char *fmt, ...)
|
2009-09-24 16:02:49 +00:00
|
|
|
{
|
|
|
|
int n;
|
|
|
|
va_list ap;
|
|
|
|
|
|
|
|
va_start(ap, fmt);
|
2010-03-31 14:33:40 +00:00
|
|
|
n = vsnprintf(bf, size, fmt, ap);
|
2012-09-06 15:46:56 +00:00
|
|
|
if (symbol_conf.field_sep && n > 0) {
|
2010-03-31 14:33:40 +00:00
|
|
|
char *sep = bf;
|
|
|
|
|
|
|
|
while (1) {
|
2012-09-06 15:46:56 +00:00
|
|
|
sep = strchr(sep, *symbol_conf.field_sep);
|
2010-03-31 14:33:40 +00:00
|
|
|
if (sep == NULL)
|
|
|
|
break;
|
|
|
|
*sep = '.';
|
2009-09-24 16:02:49 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
va_end(ap);
|
perf tools: Incorrect use of snprintf results in SEGV
I have a workload where perf top scribbles over the stack and we SEGV.
What makes it interesting is that an snprintf is causing this.
The workload is a c++ gem that has method names over 3000 characters
long, but snprintf is designed to avoid overrunning buffers. So what
went wrong?
The problem is we assume snprintf returns the number of characters
written:
ret += repsep_snprintf(bf + ret, size - ret, "[%c] ", self->level);
...
ret += repsep_snprintf(bf + ret, size - ret, "%s", self->ms.sym->name);
Unfortunately this is not how snprintf works. snprintf returns the
number of characters that would have been written if there was enough
space. In the above case, if the first snprintf returns a value larger
than size, we pass a negative size into the second snprintf and happily
scribble over the stack. If you have 3000 character c++ methods thats a
lot of stack to trample.
This patch fixes repsep_snprintf by clamping the value at size - 1 which
is the maximum snprintf can write before adding the NULL terminator.
I get the sinking feeling that there are a lot of other uses of snprintf
that have this same bug, we should audit them all.
Cc: David Ahern <dsahern@gmail.com>
Cc: Eric B Munson <emunson@mgebm.net>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Yanmin Zhang <yanmin_zhang@linux.intel.com>
Cc: stable@kernel.org
Link: http://lkml.kernel.org/r/20120307114249.44275ca3@kryten
Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2012-03-07 00:42:49 +00:00
|
|
|
|
|
|
|
if (n >= (int)size)
|
|
|
|
return size - 1;
|
2009-09-24 16:02:49 +00:00
|
|
|
return n;
|
|
|
|
}
|
|
|
|
|
2013-09-11 12:46:56 +00:00
|
|
|
static int64_t cmp_null(const void *l, const void *r)
|
2011-06-29 01:14:52 +00:00
|
|
|
{
|
|
|
|
if (!l && !r)
|
|
|
|
return 0;
|
|
|
|
else if (!l)
|
|
|
|
return -1;
|
|
|
|
else
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* --sort pid */
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__thread_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2013-07-04 13:20:31 +00:00
|
|
|
return right->thread->tid - left->thread->tid;
|
2011-06-29 01:14:52 +00:00
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__thread_snprintf(struct hist_entry *he, char *bf,
|
2010-03-31 14:33:40 +00:00
|
|
|
size_t size, unsigned int width)
|
2009-09-24 16:02:49 +00:00
|
|
|
{
|
2013-09-11 12:46:56 +00:00
|
|
|
const char *comm = thread__comm_str(he->thread);
|
2014-07-31 05:47:38 +00:00
|
|
|
|
2016-06-20 21:58:19 +00:00
|
|
|
width = max(7U, width) - 8;
|
|
|
|
return repsep_snprintf(bf, size, "%7d:%-*.*s", he->thread->tid,
|
2014-07-31 05:47:38 +00:00
|
|
|
width, width, comm ?: "");
|
2009-09-24 16:02:49 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 15:13:37 +00:00
|
|
|
static int hist_entry__thread_filter(struct hist_entry *he, int type, const void *arg)
|
|
|
|
{
|
|
|
|
const struct thread *th = arg;
|
|
|
|
|
|
|
|
if (type != HIST_FILTER__THREAD)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
return th && he->thread != th;
|
|
|
|
}
|
|
|
|
|
2011-06-29 01:14:52 +00:00
|
|
|
struct sort_entry sort_thread = {
|
2016-06-20 21:58:19 +00:00
|
|
|
.se_header = " Pid:Command",
|
2011-06-29 01:14:52 +00:00
|
|
|
.se_cmp = sort__thread_cmp,
|
|
|
|
.se_snprintf = hist_entry__thread_snprintf,
|
2016-02-24 15:13:37 +00:00
|
|
|
.se_filter = hist_entry__thread_filter,
|
2011-06-29 01:14:52 +00:00
|
|
|
.se_width_idx = HISTC_THREAD,
|
|
|
|
};
|
|
|
|
|
|
|
|
/* --sort comm */
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__comm_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2013-09-11 15:18:09 +00:00
|
|
|
/* Compare the addr that should be unique among comm */
|
2015-05-15 15:54:28 +00:00
|
|
|
return strcmp(comm__str(right->comm), comm__str(left->comm));
|
2011-06-29 01:14:52 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__comm_collapse(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2013-09-13 07:28:57 +00:00
|
|
|
/* Compare the addr that should be unique among comm */
|
2015-05-15 15:54:28 +00:00
|
|
|
return strcmp(comm__str(right->comm), comm__str(left->comm));
|
2011-06-29 01:14:52 +00:00
|
|
|
}
|
|
|
|
|
2014-03-04 02:01:41 +00:00
|
|
|
static int64_t
|
|
|
|
sort__comm_sort(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
return strcmp(comm__str(right->comm), comm__str(left->comm));
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__comm_snprintf(struct hist_entry *he, char *bf,
|
2010-03-31 14:33:40 +00:00
|
|
|
size_t size, unsigned int width)
|
2009-09-24 16:02:49 +00:00
|
|
|
{
|
2014-07-31 05:47:38 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width, comm__str(he->comm));
|
2009-09-24 16:02:49 +00:00
|
|
|
}
|
|
|
|
|
2012-12-27 09:11:38 +00:00
|
|
|
struct sort_entry sort_comm = {
|
|
|
|
.se_header = "Command",
|
|
|
|
.se_cmp = sort__comm_cmp,
|
|
|
|
.se_collapse = sort__comm_collapse,
|
2014-03-04 02:01:41 +00:00
|
|
|
.se_sort = sort__comm_sort,
|
2012-12-27 09:11:38 +00:00
|
|
|
.se_snprintf = hist_entry__comm_snprintf,
|
2016-02-24 15:13:37 +00:00
|
|
|
.se_filter = hist_entry__thread_filter,
|
2012-12-27 09:11:38 +00:00
|
|
|
.se_width_idx = HISTC_COMM,
|
|
|
|
};
|
|
|
|
|
|
|
|
/* --sort dso */
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
static int64_t _sort__dso_cmp(struct map *map_l, struct map *map_r)
|
|
|
|
{
|
|
|
|
struct dso *dso_l = map_l ? map_l->dso : NULL;
|
|
|
|
struct dso *dso_r = map_r ? map_r->dso : NULL;
|
|
|
|
const char *dso_name_l, *dso_name_r;
|
|
|
|
|
|
|
|
if (!dso_l || !dso_r)
|
2014-03-04 02:01:41 +00:00
|
|
|
return cmp_null(dso_r, dso_l);
|
2012-02-09 22:21:01 +00:00
|
|
|
|
2017-02-17 08:17:38 +00:00
|
|
|
if (verbose > 0) {
|
2012-02-09 22:21:01 +00:00
|
|
|
dso_name_l = dso_l->long_name;
|
|
|
|
dso_name_r = dso_r->long_name;
|
|
|
|
} else {
|
|
|
|
dso_name_l = dso_l->short_name;
|
|
|
|
dso_name_r = dso_r->short_name;
|
|
|
|
}
|
|
|
|
|
|
|
|
return strcmp(dso_name_l, dso_name_r);
|
|
|
|
}
|
|
|
|
|
2011-06-29 01:14:52 +00:00
|
|
|
static int64_t
|
2009-09-24 16:02:49 +00:00
|
|
|
sort__dso_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2014-03-04 02:01:41 +00:00
|
|
|
return _sort__dso_cmp(right->ms.map, left->ms.map);
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
2009-09-24 16:02:49 +00:00
|
|
|
|
2012-12-27 09:11:38 +00:00
|
|
|
static int _hist_entry__dso_snprintf(struct map *map, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
if (map && map->dso) {
|
2017-02-17 08:17:38 +00:00
|
|
|
const char *dso_name = verbose > 0 ? map->dso->long_name :
|
|
|
|
map->dso->short_name;
|
2014-07-31 05:47:38 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width, dso_name);
|
2012-12-27 09:11:38 +00:00
|
|
|
}
|
|
|
|
|
2014-07-31 05:47:38 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width, "[unknown]");
|
2012-12-27 09:11:38 +00:00
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__dso_snprintf(struct hist_entry *he, char *bf,
|
2012-12-27 09:11:38 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2013-10-22 22:01:31 +00:00
|
|
|
return _hist_entry__dso_snprintf(he->ms.map, bf, size, width);
|
2012-12-27 09:11:38 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 15:13:37 +00:00
|
|
|
static int hist_entry__dso_filter(struct hist_entry *he, int type, const void *arg)
|
|
|
|
{
|
|
|
|
const struct dso *dso = arg;
|
|
|
|
|
|
|
|
if (type != HIST_FILTER__DSO)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
return dso && (!he->ms.map || he->ms.map->dso != dso);
|
|
|
|
}
|
|
|
|
|
2012-12-27 09:11:38 +00:00
|
|
|
struct sort_entry sort_dso = {
|
|
|
|
.se_header = "Shared Object",
|
|
|
|
.se_cmp = sort__dso_cmp,
|
|
|
|
.se_snprintf = hist_entry__dso_snprintf,
|
2016-02-24 15:13:37 +00:00
|
|
|
.se_filter = hist_entry__dso_filter,
|
2012-12-27 09:11:38 +00:00
|
|
|
.se_width_idx = HISTC_DSO,
|
|
|
|
};
|
|
|
|
|
|
|
|
/* --sort symbol */
|
2009-09-24 16:02:49 +00:00
|
|
|
|
2013-12-18 05:21:09 +00:00
|
|
|
static int64_t _sort__addr_cmp(u64 left_ip, u64 right_ip)
|
|
|
|
{
|
|
|
|
return (int64_t)(right_ip - left_ip);
|
|
|
|
}
|
|
|
|
|
2013-02-06 05:57:15 +00:00
|
|
|
static int64_t _sort__sym_cmp(struct symbol *sym_l, struct symbol *sym_r)
|
2012-02-09 22:21:01 +00:00
|
|
|
{
|
|
|
|
if (!sym_l || !sym_r)
|
|
|
|
return cmp_null(sym_l, sym_r);
|
|
|
|
|
|
|
|
if (sym_l == sym_r)
|
|
|
|
return 0;
|
|
|
|
|
perf report: Compare symbol name for inlined frames when sorting
Similar to the callstack frame matching, we also have to compare the
symbol name when sorting hist entries. The reason is twofold: On one
hand, multiple inlined functions will use the same symbol start/end
values of the parent, non-inlined symbol.
As such, all of these symbols often end up missing from top-level
report, as they get merged with the non-inlined frame. On the other
hand, multiple different functions may end up inlining the same
function, and we need to aggregate these values properly.
Before:
~~~~~
perf report --stdio --inline -g none
# Children Self Command Shared Object Symbol
# ........ ........ ............ ............. ...................................
#
100.00% 39.69% cpp-inlining cpp-inlining [.] main
100.00% 0.00% cpp-inlining cpp-inlining [.] _start
100.00% 0.00% cpp-inlining libc-2.25.so [.] __libc_start_main
97.03% 0.00% cpp-inlining cpp-inlining [.] std::norm<double> (inlined)
59.53% 4.26% cpp-inlining libm-2.25.so [.] hypot
55.21% 55.08% cpp-inlining libm-2.25.so [.] __hypot_finite
0.52% 0.52% cpp-inlining libm-2.25.so [.] cabs
~~~~~
After:
~~~~~
perf report --stdio --inline -g none
# Children Self Command Shared Object Symbol
# ........ ........ ............ ............. ...................................................................................................................................
#
100.00% 39.69% cpp-inlining cpp-inlining [.] main
100.00% 0.00% cpp-inlining cpp-inlining [.] _start
100.00% 0.00% cpp-inlining libc-2.25.so [.] __libc_start_main
62.57% 0.00% cpp-inlining cpp-inlining [.] std::_Norm_helper<true>::_S_do_it<double> (inlined)
62.57% 0.00% cpp-inlining cpp-inlining [.] std::__complex_abs (inlined)
62.57% 0.00% cpp-inlining cpp-inlining [.] std::abs<double> (inlined)
62.57% 0.00% cpp-inlining cpp-inlining [.] std::norm<double> (inlined)
59.53% 4.26% cpp-inlining libm-2.25.so [.] hypot
55.21% 55.08% cpp-inlining libm-2.25.so [.] __hypot_finite
34.46% 0.00% cpp-inlining cpp-inlining [.] std::uniform_real_distribution<double>::operator()<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> > (inlined)
32.39% 0.00% cpp-inlining cpp-inlining [.] std::__detail::_Adaptor<std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>, double>::operator() (inlined)
32.39% 0.00% cpp-inlining cpp-inlining [.] std::generate_canonical<double, 53ul, std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul> > (inlined)
12.29% 0.00% cpp-inlining cpp-inlining [.] std::__detail::_Mod<unsigned long, 2147483647ul, 16807ul, 0ul, true, true>::__calc (inlined)
12.29% 0.00% cpp-inlining cpp-inlining [.] std::__detail::__mod<unsigned long, 2147483647ul, 16807ul, 0ul> (inlined)
12.29% 0.00% cpp-inlining cpp-inlining [.] std::linear_congruential_engine<unsigned long, 16807ul, 0ul, 2147483647ul>::operator() (inlined)
0.52% 0.52% cpp-inlining libm-2.25.so [.] cabs
~~~~~
Signed-off-by: Milian Wolff <milian.wolff@kdab.com>
Reviewed-by: Jiri Olsa <jolsa@redhat.com>
Reviewed-by: Namhyung Kim <namhyung@kernel.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Yao Jin <yao.jin@linux.intel.com>
Link: http://lkml.kernel.org/r/20171009203310.17362-11-milian.wolff@kdab.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-10-09 20:33:04 +00:00
|
|
|
if (sym_l->inlined || sym_r->inlined)
|
|
|
|
return strcmp(sym_l->name, sym_r->name);
|
|
|
|
|
2015-06-17 23:41:10 +00:00
|
|
|
if (sym_l->start != sym_r->start)
|
|
|
|
return (int64_t)(sym_r->start - sym_l->start);
|
2012-02-09 22:21:01 +00:00
|
|
|
|
2015-06-17 23:41:10 +00:00
|
|
|
return (int64_t)(sym_r->end - sym_l->end);
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
|
|
|
|
2012-12-27 09:11:38 +00:00
|
|
|
static int64_t
|
|
|
|
sort__sym_cmp(struct hist_entry *left, struct hist_entry *right)
|
2012-02-09 22:21:01 +00:00
|
|
|
{
|
2013-10-15 02:01:56 +00:00
|
|
|
int64_t ret;
|
|
|
|
|
2012-12-27 09:11:38 +00:00
|
|
|
if (!left->ms.sym && !right->ms.sym)
|
2013-12-18 05:21:09 +00:00
|
|
|
return _sort__addr_cmp(left->ip, right->ip);
|
2009-09-24 16:02:49 +00:00
|
|
|
|
2013-10-15 02:01:56 +00:00
|
|
|
/*
|
|
|
|
* comparing symbol address alone is not enough since it's a
|
|
|
|
* relative address within a dso.
|
|
|
|
*/
|
2016-05-03 11:54:45 +00:00
|
|
|
if (!hists__has(left->hists, dso) || hists__has(right->hists, dso)) {
|
2013-12-18 05:21:10 +00:00
|
|
|
ret = sort__dso_cmp(left, right);
|
|
|
|
if (ret != 0)
|
|
|
|
return ret;
|
|
|
|
}
|
2013-10-15 02:01:56 +00:00
|
|
|
|
2013-02-06 05:57:15 +00:00
|
|
|
return _sort__sym_cmp(left->ms.sym, right->ms.sym);
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
|
|
|
|
2014-03-04 02:01:41 +00:00
|
|
|
static int64_t
|
|
|
|
sort__sym_sort(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
if (!left->ms.sym || !right->ms.sym)
|
|
|
|
return cmp_null(left->ms.sym, right->ms.sym);
|
|
|
|
|
|
|
|
return strcmp(right->ms.sym->name, left->ms.sym->name);
|
|
|
|
}
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
static int _hist_entry__sym_snprintf(struct map *map, struct symbol *sym,
|
|
|
|
u64 ip, char level, char *bf, size_t size,
|
2012-12-27 09:11:39 +00:00
|
|
|
unsigned int width)
|
2012-02-09 22:21:01 +00:00
|
|
|
{
|
|
|
|
size_t ret = 0;
|
|
|
|
|
2017-02-17 08:17:38 +00:00
|
|
|
if (verbose > 0) {
|
2012-02-09 22:21:01 +00:00
|
|
|
char o = map ? dso__symtab_origin(map->dso) : '!';
|
|
|
|
ret += repsep_snprintf(bf, size, "%-#*llx %c ",
|
2013-04-01 11:35:19 +00:00
|
|
|
BITS_PER_LONG / 4 + 2, ip, o);
|
2009-10-02 06:29:58 +00:00
|
|
|
}
|
2009-09-24 16:02:49 +00:00
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
ret += repsep_snprintf(bf + ret, size - ret, "[%c] ", level);
|
2013-01-24 15:10:35 +00:00
|
|
|
if (sym && map) {
|
|
|
|
if (map->type == MAP__VARIABLE) {
|
|
|
|
ret += repsep_snprintf(bf + ret, size - ret, "%s", sym->name);
|
|
|
|
ret += repsep_snprintf(bf + ret, size - ret, "+0x%llx",
|
2013-01-24 15:10:42 +00:00
|
|
|
ip - map->unmap_ip(map, sym->start));
|
2013-01-24 15:10:35 +00:00
|
|
|
} else {
|
perf hists: Do column alignment on the format iterator
We were doing column alignment in the format function for each cell,
returning a string padded with spaces so that when the next column is
printed the cursor is at its column alignment.
This ends up needlessly printing trailing spaces, do it at the format
iterator, that is where we know if it is needed, i.e. if there is more
columns to be printed.
This eliminates the need for triming lines when doing a dump using 'P'
in the TUI browser and also produces far saner results with things like
piping 'perf report' to 'less'.
Right now only the formatters for sym->name and the 'locked' column
(perf mem report), that are the ones that end up at the end of lines
in the default 'perf report', 'perf top' and 'perf mem report' tools,
the others will be done in a subsequent patch.
In the end the 'width' parameter for the formatters now mean, in
'printf' terms, the 'precision', where before it was the field 'width'.
Reported-by: Dave Jones <davej@codemonkey.org.uk>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lkml.kernel.org/n/tip-s7iwl2gj23w92l6tibnrcqzr@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2016-02-11 20:14:13 +00:00
|
|
|
ret += repsep_snprintf(bf + ret, size - ret, "%.*s",
|
2013-01-24 15:10:35 +00:00
|
|
|
width - ret,
|
|
|
|
sym->name);
|
2017-10-09 20:33:01 +00:00
|
|
|
if (sym->inlined)
|
|
|
|
ret += repsep_snprintf(bf + ret, size - ret,
|
|
|
|
" (inlined)");
|
2013-01-24 15:10:35 +00:00
|
|
|
}
|
|
|
|
} else {
|
2012-02-09 22:21:01 +00:00
|
|
|
size_t len = BITS_PER_LONG / 4;
|
|
|
|
ret += repsep_snprintf(bf + ret, size - ret, "%-#.*llx",
|
|
|
|
len, ip);
|
|
|
|
}
|
|
|
|
|
perf hists: Do column alignment on the format iterator
We were doing column alignment in the format function for each cell,
returning a string padded with spaces so that when the next column is
printed the cursor is at its column alignment.
This ends up needlessly printing trailing spaces, do it at the format
iterator, that is where we know if it is needed, i.e. if there is more
columns to be printed.
This eliminates the need for triming lines when doing a dump using 'P'
in the TUI browser and also produces far saner results with things like
piping 'perf report' to 'less'.
Right now only the formatters for sym->name and the 'locked' column
(perf mem report), that are the ones that end up at the end of lines
in the default 'perf report', 'perf top' and 'perf mem report' tools,
the others will be done in a subsequent patch.
In the end the 'width' parameter for the formatters now mean, in
'printf' terms, the 'precision', where before it was the field 'width'.
Reported-by: Dave Jones <davej@codemonkey.org.uk>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lkml.kernel.org/n/tip-s7iwl2gj23w92l6tibnrcqzr@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2016-02-11 20:14:13 +00:00
|
|
|
return ret;
|
2009-09-24 16:02:49 +00:00
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__sym_snprintf(struct hist_entry *he, char *bf,
|
2012-12-27 09:11:39 +00:00
|
|
|
size_t size, unsigned int width)
|
2012-02-09 22:21:01 +00:00
|
|
|
{
|
2013-10-22 22:01:31 +00:00
|
|
|
return _hist_entry__sym_snprintf(he->ms.map, he->ms.sym, he->ip,
|
|
|
|
he->level, bf, size, width);
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
2009-09-24 16:02:49 +00:00
|
|
|
|
2016-02-24 15:13:37 +00:00
|
|
|
static int hist_entry__sym_filter(struct hist_entry *he, int type, const void *arg)
|
|
|
|
{
|
|
|
|
const char *sym = arg;
|
|
|
|
|
|
|
|
if (type != HIST_FILTER__SYMBOL)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
return sym && (!he->ms.sym || !strstr(he->ms.sym->name, sym));
|
|
|
|
}
|
|
|
|
|
2011-06-29 01:14:52 +00:00
|
|
|
struct sort_entry sort_sym = {
|
|
|
|
.se_header = "Symbol",
|
|
|
|
.se_cmp = sort__sym_cmp,
|
2014-03-04 02:01:41 +00:00
|
|
|
.se_sort = sort__sym_sort,
|
2011-06-29 01:14:52 +00:00
|
|
|
.se_snprintf = hist_entry__sym_snprintf,
|
2016-02-24 15:13:37 +00:00
|
|
|
.se_filter = hist_entry__sym_filter,
|
2011-06-29 01:14:52 +00:00
|
|
|
.se_width_idx = HISTC_SYMBOL,
|
|
|
|
};
|
2009-09-24 16:02:49 +00:00
|
|
|
|
2012-05-30 13:33:24 +00:00
|
|
|
/* --sort srcline */
|
|
|
|
|
2016-07-10 13:47:40 +00:00
|
|
|
char *hist_entry__get_srcline(struct hist_entry *he)
|
2016-02-22 00:31:51 +00:00
|
|
|
{
|
|
|
|
struct map *map = he->ms.map;
|
|
|
|
|
|
|
|
if (!map)
|
|
|
|
return SRCLINE_UNKNOWN;
|
|
|
|
|
|
|
|
return get_srcline(map->dso, map__rip_2objdump(map, he->ip),
|
2017-12-29 16:26:52 +00:00
|
|
|
he->ms.sym, true, true, he->ip);
|
2016-02-22 00:31:51 +00:00
|
|
|
}
|
|
|
|
|
2012-05-30 13:33:24 +00:00
|
|
|
static int64_t
|
|
|
|
sort__srcline_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2016-02-22 00:31:51 +00:00
|
|
|
if (!left->srcline)
|
|
|
|
left->srcline = hist_entry__get_srcline(left);
|
|
|
|
if (!right->srcline)
|
|
|
|
right->srcline = hist_entry__get_srcline(right);
|
|
|
|
|
2014-03-04 02:01:41 +00:00
|
|
|
return strcmp(right->srcline, left->srcline);
|
2012-05-30 13:33:24 +00:00
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__srcline_snprintf(struct hist_entry *he, char *bf,
|
2014-07-31 05:47:38 +00:00
|
|
|
size_t size, unsigned int width)
|
2012-05-30 13:33:24 +00:00
|
|
|
{
|
2016-02-22 00:31:51 +00:00
|
|
|
if (!he->srcline)
|
|
|
|
he->srcline = hist_entry__get_srcline(he);
|
|
|
|
|
2016-02-22 00:32:33 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-.*s", width, he->srcline);
|
2012-05-30 13:33:24 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_srcline = {
|
|
|
|
.se_header = "Source:Line",
|
|
|
|
.se_cmp = sort__srcline_cmp,
|
|
|
|
.se_snprintf = hist_entry__srcline_snprintf,
|
|
|
|
.se_width_idx = HISTC_SRCLINE,
|
|
|
|
};
|
|
|
|
|
2016-05-20 20:15:08 +00:00
|
|
|
/* --sort srcline_from */
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__srcline_from_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
if (!left->branch_info->srcline_from) {
|
|
|
|
struct map *map = left->branch_info->from.map;
|
|
|
|
if (!map)
|
|
|
|
left->branch_info->srcline_from = SRCLINE_UNKNOWN;
|
|
|
|
else
|
|
|
|
left->branch_info->srcline_from = get_srcline(map->dso,
|
|
|
|
map__rip_2objdump(map,
|
|
|
|
left->branch_info->from.al_addr),
|
2017-03-18 21:49:28 +00:00
|
|
|
left->branch_info->from.sym,
|
2017-12-29 16:26:52 +00:00
|
|
|
true, true,
|
|
|
|
left->branch_info->from.al_addr);
|
2016-05-20 20:15:08 +00:00
|
|
|
}
|
|
|
|
if (!right->branch_info->srcline_from) {
|
|
|
|
struct map *map = right->branch_info->from.map;
|
|
|
|
if (!map)
|
|
|
|
right->branch_info->srcline_from = SRCLINE_UNKNOWN;
|
|
|
|
else
|
|
|
|
right->branch_info->srcline_from = get_srcline(map->dso,
|
|
|
|
map__rip_2objdump(map,
|
|
|
|
right->branch_info->from.al_addr),
|
2017-03-18 21:49:28 +00:00
|
|
|
right->branch_info->from.sym,
|
2017-12-29 16:26:52 +00:00
|
|
|
true, true,
|
|
|
|
right->branch_info->from.al_addr);
|
2016-05-20 20:15:08 +00:00
|
|
|
}
|
|
|
|
return strcmp(right->branch_info->srcline_from, left->branch_info->srcline_from);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__srcline_from_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width, he->branch_info->srcline_from);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_srcline_from = {
|
|
|
|
.se_header = "From Source:Line",
|
|
|
|
.se_cmp = sort__srcline_from_cmp,
|
|
|
|
.se_snprintf = hist_entry__srcline_from_snprintf,
|
|
|
|
.se_width_idx = HISTC_SRCLINE_FROM,
|
|
|
|
};
|
|
|
|
|
|
|
|
/* --sort srcline_to */
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__srcline_to_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
if (!left->branch_info->srcline_to) {
|
|
|
|
struct map *map = left->branch_info->to.map;
|
|
|
|
if (!map)
|
|
|
|
left->branch_info->srcline_to = SRCLINE_UNKNOWN;
|
|
|
|
else
|
|
|
|
left->branch_info->srcline_to = get_srcline(map->dso,
|
|
|
|
map__rip_2objdump(map,
|
|
|
|
left->branch_info->to.al_addr),
|
2017-03-18 21:49:28 +00:00
|
|
|
left->branch_info->from.sym,
|
2017-12-29 16:26:52 +00:00
|
|
|
true, true,
|
|
|
|
left->branch_info->to.al_addr);
|
2016-05-20 20:15:08 +00:00
|
|
|
}
|
|
|
|
if (!right->branch_info->srcline_to) {
|
|
|
|
struct map *map = right->branch_info->to.map;
|
|
|
|
if (!map)
|
|
|
|
right->branch_info->srcline_to = SRCLINE_UNKNOWN;
|
|
|
|
else
|
|
|
|
right->branch_info->srcline_to = get_srcline(map->dso,
|
|
|
|
map__rip_2objdump(map,
|
|
|
|
right->branch_info->to.al_addr),
|
2017-03-18 21:49:28 +00:00
|
|
|
right->branch_info->to.sym,
|
2017-12-29 16:26:52 +00:00
|
|
|
true, true,
|
|
|
|
right->branch_info->to.al_addr);
|
2016-05-20 20:15:08 +00:00
|
|
|
}
|
|
|
|
return strcmp(right->branch_info->srcline_to, left->branch_info->srcline_to);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__srcline_to_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width, he->branch_info->srcline_to);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_srcline_to = {
|
|
|
|
.se_header = "To Source:Line",
|
|
|
|
.se_cmp = sort__srcline_to_cmp,
|
|
|
|
.se_snprintf = hist_entry__srcline_to_snprintf,
|
|
|
|
.se_width_idx = HISTC_SRCLINE_TO,
|
|
|
|
};
|
|
|
|
|
2015-08-07 22:54:24 +00:00
|
|
|
/* --sort srcfile */
|
|
|
|
|
|
|
|
static char no_srcfile[1];
|
|
|
|
|
2016-02-22 00:31:51 +00:00
|
|
|
static char *hist_entry__get_srcfile(struct hist_entry *e)
|
2015-08-07 22:54:24 +00:00
|
|
|
{
|
|
|
|
char *sf, *p;
|
|
|
|
struct map *map = e->ms.map;
|
|
|
|
|
2016-02-22 00:31:51 +00:00
|
|
|
if (!map)
|
|
|
|
return no_srcfile;
|
|
|
|
|
2015-09-01 18:47:19 +00:00
|
|
|
sf = __get_srcline(map->dso, map__rip_2objdump(map, e->ip),
|
2017-12-29 16:26:52 +00:00
|
|
|
e->ms.sym, false, true, true, e->ip);
|
2015-08-11 13:36:55 +00:00
|
|
|
if (!strcmp(sf, SRCLINE_UNKNOWN))
|
|
|
|
return no_srcfile;
|
2015-08-07 22:54:24 +00:00
|
|
|
p = strchr(sf, ':');
|
|
|
|
if (p && *sf) {
|
|
|
|
*p = 0;
|
|
|
|
return sf;
|
|
|
|
}
|
|
|
|
free(sf);
|
|
|
|
return no_srcfile;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__srcfile_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2016-02-22 00:31:51 +00:00
|
|
|
if (!left->srcfile)
|
|
|
|
left->srcfile = hist_entry__get_srcfile(left);
|
|
|
|
if (!right->srcfile)
|
|
|
|
right->srcfile = hist_entry__get_srcfile(right);
|
|
|
|
|
2015-08-07 22:54:24 +00:00
|
|
|
return strcmp(right->srcfile, left->srcfile);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__srcfile_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2016-02-22 00:31:51 +00:00
|
|
|
if (!he->srcfile)
|
|
|
|
he->srcfile = hist_entry__get_srcfile(he);
|
|
|
|
|
2016-02-22 00:32:33 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-.*s", width, he->srcfile);
|
2015-08-07 22:54:24 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_srcfile = {
|
|
|
|
.se_header = "Source File",
|
|
|
|
.se_cmp = sort__srcfile_cmp,
|
|
|
|
.se_snprintf = hist_entry__srcfile_snprintf,
|
|
|
|
.se_width_idx = HISTC_SRCFILE,
|
|
|
|
};
|
|
|
|
|
2009-09-24 16:02:49 +00:00
|
|
|
/* --sort parent */
|
|
|
|
|
2011-06-29 01:14:52 +00:00
|
|
|
static int64_t
|
2009-09-24 16:02:49 +00:00
|
|
|
sort__parent_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
struct symbol *sym_l = left->parent;
|
|
|
|
struct symbol *sym_r = right->parent;
|
|
|
|
|
|
|
|
if (!sym_l || !sym_r)
|
|
|
|
return cmp_null(sym_l, sym_r);
|
|
|
|
|
2014-03-04 02:01:41 +00:00
|
|
|
return strcmp(sym_r->name, sym_l->name);
|
2009-09-24 16:02:49 +00:00
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__parent_snprintf(struct hist_entry *he, char *bf,
|
2010-03-31 14:33:40 +00:00
|
|
|
size_t size, unsigned int width)
|
2009-09-24 16:02:49 +00:00
|
|
|
{
|
2014-07-31 05:47:38 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width,
|
2013-10-22 22:01:31 +00:00
|
|
|
he->parent ? he->parent->name : "[other]");
|
2009-09-24 16:02:49 +00:00
|
|
|
}
|
|
|
|
|
2011-06-29 01:14:52 +00:00
|
|
|
struct sort_entry sort_parent = {
|
|
|
|
.se_header = "Parent symbol",
|
|
|
|
.se_cmp = sort__parent_cmp,
|
|
|
|
.se_snprintf = hist_entry__parent_snprintf,
|
|
|
|
.se_width_idx = HISTC_PARENT,
|
|
|
|
};
|
|
|
|
|
2010-06-04 14:27:10 +00:00
|
|
|
/* --sort cpu */
|
|
|
|
|
2011-06-29 01:14:52 +00:00
|
|
|
static int64_t
|
2010-06-04 14:27:10 +00:00
|
|
|
sort__cpu_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
return right->cpu - left->cpu;
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__cpu_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
2010-06-04 14:27:10 +00:00
|
|
|
{
|
2014-07-31 05:47:38 +00:00
|
|
|
return repsep_snprintf(bf, size, "%*.*d", width, width, he->cpu);
|
2010-06-04 14:27:10 +00:00
|
|
|
}
|
|
|
|
|
2011-06-29 01:14:52 +00:00
|
|
|
struct sort_entry sort_cpu = {
|
|
|
|
.se_header = "CPU",
|
|
|
|
.se_cmp = sort__cpu_cmp,
|
|
|
|
.se_snprintf = hist_entry__cpu_snprintf,
|
|
|
|
.se_width_idx = HISTC_CPU,
|
|
|
|
};
|
|
|
|
|
perf tools: Add 'cgroup_id' sort order keyword
This patch introduces a cgroup identifier entry field in perf report to
identify or distinguish data of different cgroups. It uses the device
number and inode number of cgroup namespace, included in perf data with
the new PERF_RECORD_NAMESPACES event, as cgroup identifier.
With the assumption that each container is created with it's own cgroup
namespace, this allows assessment/analysis of multiple containers at
once.
A simple test for this would be to clone a few processes passing
SIGCHILD & CLONE_NEWCROUP flags to each of them, execute shell and run
different workloads on each of those contexts, while running perf
record command with --namespaces option.
Shown below is the output of perf report, sorted with cgroup identifier,
on perf.data generated with the above test scenario, clearly indicating
one context's considerable use of kernel memory in comparison with
others:
$ perf report -s cgroup_id,sample --stdio
#
# Total Lost Samples: 0
#
# Samples: 5K of event 'kmem:kmalloc'
# Event count (approx.): 5965
#
# Overhead cgroup id (dev/inode) Samples
# ........ ..................... ............
#
81.27% 3/0xeffffffb 4848
16.24% 3/0xf00000d0 969
1.16% 3/0xf00000ce 69
0.82% 3/0xf00000cf 49
0.50% 0/0x0 30
While this is a start, there is further scope of improving this. For
example, instead of cgroup namespace's device and inode numbers, dev
and inode numbers of some or all namespaces may be used to distinguish
which processes are running in a given container context.
Also, scripts to map device and inode info to containers sounds
plausible for better tracing of containers.
Signed-off-by: Hari Bathini <hbathini@linux.vnet.ibm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Starovoitov <ast@fb.com>
Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>
Cc: Aravinda Prasad <aravinda@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sargun Dhillon <sargun@sargun.me>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/148891933338.25309.756882900782042645.stgit@hbathini.in.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-03-07 20:42:13 +00:00
|
|
|
/* --sort cgroup_id */
|
|
|
|
|
|
|
|
static int64_t _sort__cgroup_dev_cmp(u64 left_dev, u64 right_dev)
|
|
|
|
{
|
|
|
|
return (int64_t)(right_dev - left_dev);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t _sort__cgroup_inode_cmp(u64 left_ino, u64 right_ino)
|
|
|
|
{
|
|
|
|
return (int64_t)(right_ino - left_ino);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__cgroup_id_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
int64_t ret;
|
|
|
|
|
|
|
|
ret = _sort__cgroup_dev_cmp(right->cgroup_id.dev, left->cgroup_id.dev);
|
|
|
|
if (ret != 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
return _sort__cgroup_inode_cmp(right->cgroup_id.ino,
|
|
|
|
left->cgroup_id.ino);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__cgroup_id_snprintf(struct hist_entry *he,
|
|
|
|
char *bf, size_t size,
|
|
|
|
unsigned int width __maybe_unused)
|
|
|
|
{
|
|
|
|
return repsep_snprintf(bf, size, "%lu/0x%lx", he->cgroup_id.dev,
|
|
|
|
he->cgroup_id.ino);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_cgroup_id = {
|
|
|
|
.se_header = "cgroup id (dev/inode)",
|
|
|
|
.se_cmp = sort__cgroup_id_cmp,
|
|
|
|
.se_snprintf = hist_entry__cgroup_id_snprintf,
|
|
|
|
.se_width_idx = HISTC_CGROUP_ID,
|
|
|
|
};
|
|
|
|
|
2015-09-04 14:45:43 +00:00
|
|
|
/* --sort socket */
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__socket_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
return right->socket - left->socket;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__socket_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
return repsep_snprintf(bf, size, "%*.*d", width, width-3, he->socket);
|
|
|
|
}
|
|
|
|
|
2016-02-24 15:13:37 +00:00
|
|
|
static int hist_entry__socket_filter(struct hist_entry *he, int type, const void *arg)
|
|
|
|
{
|
|
|
|
int sk = *(const int *)arg;
|
|
|
|
|
|
|
|
if (type != HIST_FILTER__SOCKET)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
return sk >= 0 && he->socket != sk;
|
|
|
|
}
|
|
|
|
|
2015-09-04 14:45:43 +00:00
|
|
|
struct sort_entry sort_socket = {
|
|
|
|
.se_header = "Socket",
|
|
|
|
.se_cmp = sort__socket_cmp,
|
|
|
|
.se_snprintf = hist_entry__socket_snprintf,
|
2016-02-24 15:13:37 +00:00
|
|
|
.se_filter = hist_entry__socket_filter,
|
2015-09-04 14:45:43 +00:00
|
|
|
.se_width_idx = HISTC_SOCKET,
|
|
|
|
};
|
|
|
|
|
2015-12-22 17:07:04 +00:00
|
|
|
/* --sort trace */
|
|
|
|
|
|
|
|
static char *get_trace_output(struct hist_entry *he)
|
|
|
|
{
|
|
|
|
struct trace_seq seq;
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
struct pevent_record rec = {
|
|
|
|
.data = he->raw_data,
|
|
|
|
.size = he->raw_size,
|
|
|
|
};
|
|
|
|
|
|
|
|
evsel = hists_to_evsel(he->hists);
|
|
|
|
|
|
|
|
trace_seq_init(&seq);
|
2015-12-22 17:07:05 +00:00
|
|
|
if (symbol_conf.raw_trace) {
|
|
|
|
pevent_print_fields(&seq, he->raw_data, he->raw_size,
|
|
|
|
evsel->tp_format);
|
|
|
|
} else {
|
|
|
|
pevent_event_info(&seq, evsel->tp_format, &rec);
|
|
|
|
}
|
2016-08-05 15:37:21 +00:00
|
|
|
/*
|
|
|
|
* Trim the buffer, it starts at 4KB and we're not going to
|
|
|
|
* add anything more to this buffer.
|
|
|
|
*/
|
|
|
|
return realloc(seq.buffer, seq.len + 1);
|
2015-12-22 17:07:04 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__trace_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
|
|
|
|
evsel = hists_to_evsel(left->hists);
|
|
|
|
if (evsel->attr.type != PERF_TYPE_TRACEPOINT)
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
if (left->trace_output == NULL)
|
|
|
|
left->trace_output = get_trace_output(left);
|
|
|
|
if (right->trace_output == NULL)
|
|
|
|
right->trace_output = get_trace_output(right);
|
|
|
|
|
|
|
|
return strcmp(right->trace_output, left->trace_output);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__trace_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
|
|
|
|
evsel = hists_to_evsel(he->hists);
|
|
|
|
if (evsel->attr.type != PERF_TYPE_TRACEPOINT)
|
2016-02-22 00:32:33 +00:00
|
|
|
return scnprintf(bf, size, "%-.*s", width, "N/A");
|
2015-12-22 17:07:04 +00:00
|
|
|
|
|
|
|
if (he->trace_output == NULL)
|
|
|
|
he->trace_output = get_trace_output(he);
|
2016-02-22 00:32:33 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-.*s", width, he->trace_output);
|
2015-12-22 17:07:04 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_trace = {
|
|
|
|
.se_header = "Trace output",
|
|
|
|
.se_cmp = sort__trace_cmp,
|
|
|
|
.se_snprintf = hist_entry__trace_snprintf,
|
|
|
|
.se_width_idx = HISTC_TRACE,
|
|
|
|
};
|
|
|
|
|
2012-12-27 09:11:38 +00:00
|
|
|
/* sort keys for branch stacks */
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
static int64_t
|
|
|
|
sort__dso_from_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2014-10-16 14:07:07 +00:00
|
|
|
if (!left->branch_info || !right->branch_info)
|
|
|
|
return cmp_null(left->branch_info, right->branch_info);
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
return _sort__dso_cmp(left->branch_info->from.map,
|
|
|
|
right->branch_info->from.map);
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__dso_from_snprintf(struct hist_entry *he, char *bf,
|
2012-02-09 22:21:01 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2014-10-16 14:07:07 +00:00
|
|
|
if (he->branch_info)
|
|
|
|
return _hist_entry__dso_snprintf(he->branch_info->from.map,
|
|
|
|
bf, size, width);
|
|
|
|
else
|
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width, "N/A");
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 15:13:37 +00:00
|
|
|
static int hist_entry__dso_from_filter(struct hist_entry *he, int type,
|
|
|
|
const void *arg)
|
|
|
|
{
|
|
|
|
const struct dso *dso = arg;
|
|
|
|
|
|
|
|
if (type != HIST_FILTER__DSO)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
return dso && (!he->branch_info || !he->branch_info->from.map ||
|
|
|
|
he->branch_info->from.map->dso != dso);
|
|
|
|
}
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
static int64_t
|
|
|
|
sort__dso_to_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2014-10-16 14:07:06 +00:00
|
|
|
if (!left->branch_info || !right->branch_info)
|
|
|
|
return cmp_null(left->branch_info, right->branch_info);
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
return _sort__dso_cmp(left->branch_info->to.map,
|
|
|
|
right->branch_info->to.map);
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__dso_to_snprintf(struct hist_entry *he, char *bf,
|
2012-02-09 22:21:01 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2014-10-16 14:07:06 +00:00
|
|
|
if (he->branch_info)
|
|
|
|
return _hist_entry__dso_snprintf(he->branch_info->to.map,
|
|
|
|
bf, size, width);
|
|
|
|
else
|
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width, "N/A");
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 15:13:37 +00:00
|
|
|
static int hist_entry__dso_to_filter(struct hist_entry *he, int type,
|
|
|
|
const void *arg)
|
|
|
|
{
|
|
|
|
const struct dso *dso = arg;
|
|
|
|
|
|
|
|
if (type != HIST_FILTER__DSO)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
return dso && (!he->branch_info || !he->branch_info->to.map ||
|
|
|
|
he->branch_info->to.map->dso != dso);
|
|
|
|
}
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
static int64_t
|
|
|
|
sort__sym_from_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
struct addr_map_symbol *from_l = &left->branch_info->from;
|
|
|
|
struct addr_map_symbol *from_r = &right->branch_info->from;
|
|
|
|
|
2014-10-16 14:07:05 +00:00
|
|
|
if (!left->branch_info || !right->branch_info)
|
|
|
|
return cmp_null(left->branch_info, right->branch_info);
|
|
|
|
|
|
|
|
from_l = &left->branch_info->from;
|
|
|
|
from_r = &right->branch_info->from;
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
if (!from_l->sym && !from_r->sym)
|
2013-12-18 05:21:09 +00:00
|
|
|
return _sort__addr_cmp(from_l->addr, from_r->addr);
|
2012-02-09 22:21:01 +00:00
|
|
|
|
2013-02-06 05:57:15 +00:00
|
|
|
return _sort__sym_cmp(from_l->sym, from_r->sym);
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__sym_to_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2014-10-16 14:07:04 +00:00
|
|
|
struct addr_map_symbol *to_l, *to_r;
|
|
|
|
|
|
|
|
if (!left->branch_info || !right->branch_info)
|
|
|
|
return cmp_null(left->branch_info, right->branch_info);
|
|
|
|
|
|
|
|
to_l = &left->branch_info->to;
|
|
|
|
to_r = &right->branch_info->to;
|
2012-02-09 22:21:01 +00:00
|
|
|
|
|
|
|
if (!to_l->sym && !to_r->sym)
|
2013-12-18 05:21:09 +00:00
|
|
|
return _sort__addr_cmp(to_l->addr, to_r->addr);
|
2012-02-09 22:21:01 +00:00
|
|
|
|
2013-02-06 05:57:15 +00:00
|
|
|
return _sort__sym_cmp(to_l->sym, to_r->sym);
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__sym_from_snprintf(struct hist_entry *he, char *bf,
|
2012-12-27 09:11:39 +00:00
|
|
|
size_t size, unsigned int width)
|
2012-02-09 22:21:01 +00:00
|
|
|
{
|
2014-10-16 14:07:05 +00:00
|
|
|
if (he->branch_info) {
|
|
|
|
struct addr_map_symbol *from = &he->branch_info->from;
|
2012-02-09 22:21:01 +00:00
|
|
|
|
2014-10-16 14:07:05 +00:00
|
|
|
return _hist_entry__sym_snprintf(from->map, from->sym, from->addr,
|
|
|
|
he->level, bf, size, width);
|
|
|
|
}
|
|
|
|
|
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width, "N/A");
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__sym_to_snprintf(struct hist_entry *he, char *bf,
|
2012-12-27 09:11:39 +00:00
|
|
|
size_t size, unsigned int width)
|
2012-02-09 22:21:01 +00:00
|
|
|
{
|
2014-10-16 14:07:04 +00:00
|
|
|
if (he->branch_info) {
|
|
|
|
struct addr_map_symbol *to = &he->branch_info->to;
|
|
|
|
|
|
|
|
return _hist_entry__sym_snprintf(to->map, to->sym, to->addr,
|
|
|
|
he->level, bf, size, width);
|
|
|
|
}
|
2012-02-09 22:21:01 +00:00
|
|
|
|
2014-10-16 14:07:04 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width, "N/A");
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
|
|
|
|
2016-02-24 15:13:37 +00:00
|
|
|
static int hist_entry__sym_from_filter(struct hist_entry *he, int type,
|
|
|
|
const void *arg)
|
|
|
|
{
|
|
|
|
const char *sym = arg;
|
|
|
|
|
|
|
|
if (type != HIST_FILTER__SYMBOL)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
return sym && !(he->branch_info && he->branch_info->from.sym &&
|
|
|
|
strstr(he->branch_info->from.sym->name, sym));
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__sym_to_filter(struct hist_entry *he, int type,
|
|
|
|
const void *arg)
|
|
|
|
{
|
|
|
|
const char *sym = arg;
|
|
|
|
|
|
|
|
if (type != HIST_FILTER__SYMBOL)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
return sym && !(he->branch_info && he->branch_info->to.sym &&
|
|
|
|
strstr(he->branch_info->to.sym->name, sym));
|
|
|
|
}
|
|
|
|
|
2012-12-27 09:11:38 +00:00
|
|
|
struct sort_entry sort_dso_from = {
|
|
|
|
.se_header = "Source Shared Object",
|
|
|
|
.se_cmp = sort__dso_from_cmp,
|
|
|
|
.se_snprintf = hist_entry__dso_from_snprintf,
|
2016-02-24 15:13:37 +00:00
|
|
|
.se_filter = hist_entry__dso_from_filter,
|
2012-12-27 09:11:38 +00:00
|
|
|
.se_width_idx = HISTC_DSO_FROM,
|
|
|
|
};
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
struct sort_entry sort_dso_to = {
|
|
|
|
.se_header = "Target Shared Object",
|
|
|
|
.se_cmp = sort__dso_to_cmp,
|
|
|
|
.se_snprintf = hist_entry__dso_to_snprintf,
|
2016-02-24 15:13:37 +00:00
|
|
|
.se_filter = hist_entry__dso_to_filter,
|
2012-02-09 22:21:01 +00:00
|
|
|
.se_width_idx = HISTC_DSO_TO,
|
|
|
|
};
|
|
|
|
|
|
|
|
struct sort_entry sort_sym_from = {
|
|
|
|
.se_header = "Source Symbol",
|
|
|
|
.se_cmp = sort__sym_from_cmp,
|
|
|
|
.se_snprintf = hist_entry__sym_from_snprintf,
|
2016-02-24 15:13:37 +00:00
|
|
|
.se_filter = hist_entry__sym_from_filter,
|
2012-02-09 22:21:01 +00:00
|
|
|
.se_width_idx = HISTC_SYMBOL_FROM,
|
|
|
|
};
|
|
|
|
|
|
|
|
struct sort_entry sort_sym_to = {
|
|
|
|
.se_header = "Target Symbol",
|
|
|
|
.se_cmp = sort__sym_to_cmp,
|
|
|
|
.se_snprintf = hist_entry__sym_to_snprintf,
|
2016-02-24 15:13:37 +00:00
|
|
|
.se_filter = hist_entry__sym_to_filter,
|
2012-02-09 22:21:01 +00:00
|
|
|
.se_width_idx = HISTC_SYMBOL_TO,
|
|
|
|
};
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__mispredict_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2014-10-16 14:07:03 +00:00
|
|
|
unsigned char mp, p;
|
2012-02-09 22:21:01 +00:00
|
|
|
|
2014-10-16 14:07:03 +00:00
|
|
|
if (!left->branch_info || !right->branch_info)
|
|
|
|
return cmp_null(left->branch_info, right->branch_info);
|
|
|
|
|
|
|
|
mp = left->branch_info->flags.mispred != right->branch_info->flags.mispred;
|
|
|
|
p = left->branch_info->flags.predicted != right->branch_info->flags.predicted;
|
2012-02-09 22:21:01 +00:00
|
|
|
return mp || p;
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__mispredict_snprintf(struct hist_entry *he, char *bf,
|
2012-02-09 22:21:01 +00:00
|
|
|
size_t size, unsigned int width){
|
|
|
|
static const char *out = "N/A";
|
|
|
|
|
2014-10-16 14:07:03 +00:00
|
|
|
if (he->branch_info) {
|
|
|
|
if (he->branch_info->flags.predicted)
|
|
|
|
out = "N";
|
|
|
|
else if (he->branch_info->flags.mispred)
|
|
|
|
out = "Y";
|
|
|
|
}
|
2012-02-09 22:21:01 +00:00
|
|
|
|
2014-07-31 05:47:38 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*.*s", width, width, out);
|
2012-02-09 22:21:01 +00:00
|
|
|
}
|
|
|
|
|
2015-07-18 15:24:46 +00:00
|
|
|
static int64_t
|
|
|
|
sort__cycles_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2017-03-13 08:31:48 +00:00
|
|
|
if (!left->branch_info || !right->branch_info)
|
|
|
|
return cmp_null(left->branch_info, right->branch_info);
|
|
|
|
|
2015-07-18 15:24:46 +00:00
|
|
|
return left->branch_info->flags.cycles -
|
|
|
|
right->branch_info->flags.cycles;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__cycles_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2017-03-13 08:31:48 +00:00
|
|
|
if (!he->branch_info)
|
|
|
|
return scnprintf(bf, size, "%-.*s", width, "N/A");
|
2015-07-18 15:24:46 +00:00
|
|
|
if (he->branch_info->flags.cycles == 0)
|
|
|
|
return repsep_snprintf(bf, size, "%-*s", width, "-");
|
|
|
|
return repsep_snprintf(bf, size, "%-*hd", width,
|
|
|
|
he->branch_info->flags.cycles);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_cycles = {
|
|
|
|
.se_header = "Basic Block Cycles",
|
|
|
|
.se_cmp = sort__cycles_cmp,
|
|
|
|
.se_snprintf = hist_entry__cycles_snprintf,
|
|
|
|
.se_width_idx = HISTC_CYCLES,
|
|
|
|
};
|
|
|
|
|
2013-01-24 15:10:35 +00:00
|
|
|
/* --sort daddr_sym */
|
2016-09-22 15:36:34 +00:00
|
|
|
int64_t
|
2013-01-24 15:10:35 +00:00
|
|
|
sort__daddr_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
uint64_t l = 0, r = 0;
|
|
|
|
|
|
|
|
if (left->mem_info)
|
|
|
|
l = left->mem_info->daddr.addr;
|
|
|
|
if (right->mem_info)
|
|
|
|
r = right->mem_info->daddr.addr;
|
|
|
|
|
|
|
|
return (int64_t)(r - l);
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__daddr_snprintf(struct hist_entry *he, char *bf,
|
2013-01-24 15:10:35 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
uint64_t addr = 0;
|
|
|
|
struct map *map = NULL;
|
|
|
|
struct symbol *sym = NULL;
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
if (he->mem_info) {
|
|
|
|
addr = he->mem_info->daddr.addr;
|
|
|
|
map = he->mem_info->daddr.map;
|
|
|
|
sym = he->mem_info->daddr.sym;
|
2013-01-24 15:10:35 +00:00
|
|
|
}
|
2013-10-22 22:01:31 +00:00
|
|
|
return _hist_entry__sym_snprintf(map, sym, addr, he->level, bf, size,
|
2013-01-24 15:10:35 +00:00
|
|
|
width);
|
|
|
|
}
|
|
|
|
|
2016-09-22 15:36:34 +00:00
|
|
|
int64_t
|
2015-10-05 18:06:07 +00:00
|
|
|
sort__iaddr_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
uint64_t l = 0, r = 0;
|
|
|
|
|
|
|
|
if (left->mem_info)
|
|
|
|
l = left->mem_info->iaddr.addr;
|
|
|
|
if (right->mem_info)
|
|
|
|
r = right->mem_info->iaddr.addr;
|
|
|
|
|
|
|
|
return (int64_t)(r - l);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__iaddr_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
uint64_t addr = 0;
|
|
|
|
struct map *map = NULL;
|
|
|
|
struct symbol *sym = NULL;
|
|
|
|
|
|
|
|
if (he->mem_info) {
|
|
|
|
addr = he->mem_info->iaddr.addr;
|
|
|
|
map = he->mem_info->iaddr.map;
|
|
|
|
sym = he->mem_info->iaddr.sym;
|
|
|
|
}
|
|
|
|
return _hist_entry__sym_snprintf(map, sym, addr, he->level, bf, size,
|
|
|
|
width);
|
|
|
|
}
|
|
|
|
|
2013-01-24 15:10:35 +00:00
|
|
|
static int64_t
|
|
|
|
sort__dso_daddr_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
struct map *map_l = NULL;
|
|
|
|
struct map *map_r = NULL;
|
|
|
|
|
|
|
|
if (left->mem_info)
|
|
|
|
map_l = left->mem_info->daddr.map;
|
|
|
|
if (right->mem_info)
|
|
|
|
map_r = right->mem_info->daddr.map;
|
|
|
|
|
|
|
|
return _sort__dso_cmp(map_l, map_r);
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__dso_daddr_snprintf(struct hist_entry *he, char *bf,
|
2013-01-24 15:10:35 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
struct map *map = NULL;
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
if (he->mem_info)
|
|
|
|
map = he->mem_info->daddr.map;
|
2013-01-24 15:10:35 +00:00
|
|
|
|
|
|
|
return _hist_entry__dso_snprintf(map, bf, size, width);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__locked_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
union perf_mem_data_src data_src_l;
|
|
|
|
union perf_mem_data_src data_src_r;
|
|
|
|
|
|
|
|
if (left->mem_info)
|
|
|
|
data_src_l = left->mem_info->data_src;
|
|
|
|
else
|
|
|
|
data_src_l.mem_lock = PERF_MEM_LOCK_NA;
|
|
|
|
|
|
|
|
if (right->mem_info)
|
|
|
|
data_src_r = right->mem_info->data_src;
|
|
|
|
else
|
|
|
|
data_src_r.mem_lock = PERF_MEM_LOCK_NA;
|
|
|
|
|
|
|
|
return (int64_t)(data_src_r.mem_lock - data_src_l.mem_lock);
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__locked_snprintf(struct hist_entry *he, char *bf,
|
2013-01-24 15:10:35 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2016-02-24 08:46:49 +00:00
|
|
|
char out[10];
|
2013-01-24 15:10:35 +00:00
|
|
|
|
2016-02-24 08:46:49 +00:00
|
|
|
perf_mem__lck_scnprintf(out, sizeof(out), he->mem_info);
|
perf hists: Do column alignment on the format iterator
We were doing column alignment in the format function for each cell,
returning a string padded with spaces so that when the next column is
printed the cursor is at its column alignment.
This ends up needlessly printing trailing spaces, do it at the format
iterator, that is where we know if it is needed, i.e. if there is more
columns to be printed.
This eliminates the need for triming lines when doing a dump using 'P'
in the TUI browser and also produces far saner results with things like
piping 'perf report' to 'less'.
Right now only the formatters for sym->name and the 'locked' column
(perf mem report), that are the ones that end up at the end of lines
in the default 'perf report', 'perf top' and 'perf mem report' tools,
the others will be done in a subsequent patch.
In the end the 'width' parameter for the formatters now mean, in
'printf' terms, the 'precision', where before it was the field 'width'.
Reported-by: Dave Jones <davej@codemonkey.org.uk>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Link: http://lkml.kernel.org/n/tip-s7iwl2gj23w92l6tibnrcqzr@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2016-02-11 20:14:13 +00:00
|
|
|
return repsep_snprintf(bf, size, "%.*s", width, out);
|
2013-01-24 15:10:35 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__tlb_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
union perf_mem_data_src data_src_l;
|
|
|
|
union perf_mem_data_src data_src_r;
|
|
|
|
|
|
|
|
if (left->mem_info)
|
|
|
|
data_src_l = left->mem_info->data_src;
|
|
|
|
else
|
|
|
|
data_src_l.mem_dtlb = PERF_MEM_TLB_NA;
|
|
|
|
|
|
|
|
if (right->mem_info)
|
|
|
|
data_src_r = right->mem_info->data_src;
|
|
|
|
else
|
|
|
|
data_src_r.mem_dtlb = PERF_MEM_TLB_NA;
|
|
|
|
|
|
|
|
return (int64_t)(data_src_r.mem_dtlb - data_src_l.mem_dtlb);
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__tlb_snprintf(struct hist_entry *he, char *bf,
|
2013-01-24 15:10:35 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
char out[64];
|
|
|
|
|
2016-02-24 08:46:46 +00:00
|
|
|
perf_mem__tlb_scnprintf(out, sizeof(out), he->mem_info);
|
2013-01-24 15:10:35 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*s", width, out);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__lvl_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
union perf_mem_data_src data_src_l;
|
|
|
|
union perf_mem_data_src data_src_r;
|
|
|
|
|
|
|
|
if (left->mem_info)
|
|
|
|
data_src_l = left->mem_info->data_src;
|
|
|
|
else
|
|
|
|
data_src_l.mem_lvl = PERF_MEM_LVL_NA;
|
|
|
|
|
|
|
|
if (right->mem_info)
|
|
|
|
data_src_r = right->mem_info->data_src;
|
|
|
|
else
|
|
|
|
data_src_r.mem_lvl = PERF_MEM_LVL_NA;
|
|
|
|
|
|
|
|
return (int64_t)(data_src_r.mem_lvl - data_src_l.mem_lvl);
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__lvl_snprintf(struct hist_entry *he, char *bf,
|
2013-01-24 15:10:35 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
char out[64];
|
|
|
|
|
2016-02-24 08:46:47 +00:00
|
|
|
perf_mem__lvl_scnprintf(out, sizeof(out), he->mem_info);
|
2013-01-24 15:10:35 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*s", width, out);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__snoop_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
union perf_mem_data_src data_src_l;
|
|
|
|
union perf_mem_data_src data_src_r;
|
|
|
|
|
|
|
|
if (left->mem_info)
|
|
|
|
data_src_l = left->mem_info->data_src;
|
|
|
|
else
|
|
|
|
data_src_l.mem_snoop = PERF_MEM_SNOOP_NA;
|
|
|
|
|
|
|
|
if (right->mem_info)
|
|
|
|
data_src_r = right->mem_info->data_src;
|
|
|
|
else
|
|
|
|
data_src_r.mem_snoop = PERF_MEM_SNOOP_NA;
|
|
|
|
|
|
|
|
return (int64_t)(data_src_r.mem_snoop - data_src_l.mem_snoop);
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__snoop_snprintf(struct hist_entry *he, char *bf,
|
2013-01-24 15:10:35 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
char out[64];
|
|
|
|
|
2016-02-24 08:46:48 +00:00
|
|
|
perf_mem__snp_scnprintf(out, sizeof(out), he->mem_info);
|
2013-01-24 15:10:35 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*s", width, out);
|
|
|
|
}
|
|
|
|
|
2016-09-22 15:36:34 +00:00
|
|
|
int64_t
|
2014-06-01 13:38:29 +00:00
|
|
|
sort__dcacheline_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
u64 l, r;
|
|
|
|
struct map *l_map, *r_map;
|
|
|
|
|
|
|
|
if (!left->mem_info) return -1;
|
|
|
|
if (!right->mem_info) return 1;
|
|
|
|
|
|
|
|
/* group event types together */
|
|
|
|
if (left->cpumode > right->cpumode) return -1;
|
|
|
|
if (left->cpumode < right->cpumode) return 1;
|
|
|
|
|
|
|
|
l_map = left->mem_info->daddr.map;
|
|
|
|
r_map = right->mem_info->daddr.map;
|
|
|
|
|
|
|
|
/* if both are NULL, jump to sort on al_addr instead */
|
|
|
|
if (!l_map && !r_map)
|
|
|
|
goto addr;
|
|
|
|
|
|
|
|
if (!l_map) return -1;
|
|
|
|
if (!r_map) return 1;
|
|
|
|
|
|
|
|
if (l_map->maj > r_map->maj) return -1;
|
|
|
|
if (l_map->maj < r_map->maj) return 1;
|
|
|
|
|
|
|
|
if (l_map->min > r_map->min) return -1;
|
|
|
|
if (l_map->min < r_map->min) return 1;
|
|
|
|
|
|
|
|
if (l_map->ino > r_map->ino) return -1;
|
|
|
|
if (l_map->ino < r_map->ino) return 1;
|
|
|
|
|
|
|
|
if (l_map->ino_generation > r_map->ino_generation) return -1;
|
|
|
|
if (l_map->ino_generation < r_map->ino_generation) return 1;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Addresses with no major/minor numbers are assumed to be
|
|
|
|
* anonymous in userspace. Sort those on pid then address.
|
|
|
|
*
|
|
|
|
* The kernel and non-zero major/minor mapped areas are
|
|
|
|
* assumed to be unity mapped. Sort those on address.
|
|
|
|
*/
|
|
|
|
|
|
|
|
if ((left->cpumode != PERF_RECORD_MISC_KERNEL) &&
|
|
|
|
(!(l_map->flags & MAP_SHARED)) &&
|
|
|
|
!l_map->maj && !l_map->min && !l_map->ino &&
|
|
|
|
!l_map->ino_generation) {
|
|
|
|
/* userspace anonymous */
|
|
|
|
|
|
|
|
if (left->thread->pid_ > right->thread->pid_) return -1;
|
|
|
|
if (left->thread->pid_ < right->thread->pid_) return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
addr:
|
|
|
|
/* al_addr does all the right addr - start + offset calculations */
|
|
|
|
l = cl_address(left->mem_info->daddr.al_addr);
|
|
|
|
r = cl_address(right->mem_info->daddr.al_addr);
|
|
|
|
|
|
|
|
if (l > r) return -1;
|
|
|
|
if (l < r) return 1;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__dcacheline_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
|
|
|
|
uint64_t addr = 0;
|
|
|
|
struct map *map = NULL;
|
|
|
|
struct symbol *sym = NULL;
|
|
|
|
char level = he->level;
|
|
|
|
|
|
|
|
if (he->mem_info) {
|
|
|
|
addr = cl_address(he->mem_info->daddr.al_addr);
|
|
|
|
map = he->mem_info->daddr.map;
|
|
|
|
sym = he->mem_info->daddr.sym;
|
|
|
|
|
|
|
|
/* print [s] for shared data mmaps */
|
|
|
|
if ((he->cpumode != PERF_RECORD_MISC_KERNEL) &&
|
|
|
|
map && (map->type == MAP__VARIABLE) &&
|
|
|
|
(map->flags & MAP_SHARED) &&
|
|
|
|
(map->maj || map->min || map->ino ||
|
|
|
|
map->ino_generation))
|
|
|
|
level = 's';
|
|
|
|
else if (!map)
|
|
|
|
level = 'X';
|
|
|
|
}
|
|
|
|
return _hist_entry__sym_snprintf(map, sym, addr, level, bf, size,
|
|
|
|
width);
|
|
|
|
}
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
struct sort_entry sort_mispredict = {
|
|
|
|
.se_header = "Branch Mispredicted",
|
|
|
|
.se_cmp = sort__mispredict_cmp,
|
|
|
|
.se_snprintf = hist_entry__mispredict_snprintf,
|
|
|
|
.se_width_idx = HISTC_MISPREDICT,
|
|
|
|
};
|
|
|
|
|
2013-01-24 15:10:29 +00:00
|
|
|
static u64 he_weight(struct hist_entry *he)
|
|
|
|
{
|
|
|
|
return he->stat.nr_events ? he->stat.weight / he->stat.nr_events : 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__local_weight_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
return he_weight(left) - he_weight(right);
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__local_weight_snprintf(struct hist_entry *he, char *bf,
|
2013-01-24 15:10:29 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2013-10-22 22:01:31 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*llu", width, he_weight(he));
|
2013-01-24 15:10:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_local_weight = {
|
|
|
|
.se_header = "Local Weight",
|
|
|
|
.se_cmp = sort__local_weight_cmp,
|
|
|
|
.se_snprintf = hist_entry__local_weight_snprintf,
|
|
|
|
.se_width_idx = HISTC_LOCAL_WEIGHT,
|
|
|
|
};
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__global_weight_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
return left->stat.weight - right->stat.weight;
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__global_weight_snprintf(struct hist_entry *he, char *bf,
|
2013-01-24 15:10:29 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2013-10-22 22:01:31 +00:00
|
|
|
return repsep_snprintf(bf, size, "%-*llu", width, he->stat.weight);
|
2013-01-24 15:10:29 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_global_weight = {
|
|
|
|
.se_header = "Weight",
|
|
|
|
.se_cmp = sort__global_weight_cmp,
|
|
|
|
.se_snprintf = hist_entry__global_weight_snprintf,
|
|
|
|
.se_width_idx = HISTC_GLOBAL_WEIGHT,
|
|
|
|
};
|
|
|
|
|
2013-01-24 15:10:35 +00:00
|
|
|
struct sort_entry sort_mem_daddr_sym = {
|
|
|
|
.se_header = "Data Symbol",
|
|
|
|
.se_cmp = sort__daddr_cmp,
|
|
|
|
.se_snprintf = hist_entry__daddr_snprintf,
|
|
|
|
.se_width_idx = HISTC_MEM_DADDR_SYMBOL,
|
|
|
|
};
|
|
|
|
|
2015-10-05 18:06:07 +00:00
|
|
|
struct sort_entry sort_mem_iaddr_sym = {
|
|
|
|
.se_header = "Code Symbol",
|
|
|
|
.se_cmp = sort__iaddr_cmp,
|
|
|
|
.se_snprintf = hist_entry__iaddr_snprintf,
|
|
|
|
.se_width_idx = HISTC_MEM_IADDR_SYMBOL,
|
|
|
|
};
|
|
|
|
|
2013-01-24 15:10:35 +00:00
|
|
|
struct sort_entry sort_mem_daddr_dso = {
|
|
|
|
.se_header = "Data Object",
|
|
|
|
.se_cmp = sort__dso_daddr_cmp,
|
|
|
|
.se_snprintf = hist_entry__dso_daddr_snprintf,
|
2016-06-14 18:19:12 +00:00
|
|
|
.se_width_idx = HISTC_MEM_DADDR_DSO,
|
2013-01-24 15:10:35 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
struct sort_entry sort_mem_locked = {
|
|
|
|
.se_header = "Locked",
|
|
|
|
.se_cmp = sort__locked_cmp,
|
|
|
|
.se_snprintf = hist_entry__locked_snprintf,
|
|
|
|
.se_width_idx = HISTC_MEM_LOCKED,
|
|
|
|
};
|
|
|
|
|
|
|
|
struct sort_entry sort_mem_tlb = {
|
|
|
|
.se_header = "TLB access",
|
|
|
|
.se_cmp = sort__tlb_cmp,
|
|
|
|
.se_snprintf = hist_entry__tlb_snprintf,
|
|
|
|
.se_width_idx = HISTC_MEM_TLB,
|
|
|
|
};
|
|
|
|
|
|
|
|
struct sort_entry sort_mem_lvl = {
|
|
|
|
.se_header = "Memory access",
|
|
|
|
.se_cmp = sort__lvl_cmp,
|
|
|
|
.se_snprintf = hist_entry__lvl_snprintf,
|
|
|
|
.se_width_idx = HISTC_MEM_LVL,
|
|
|
|
};
|
|
|
|
|
|
|
|
struct sort_entry sort_mem_snoop = {
|
|
|
|
.se_header = "Snoop",
|
|
|
|
.se_cmp = sort__snoop_cmp,
|
|
|
|
.se_snprintf = hist_entry__snoop_snprintf,
|
|
|
|
.se_width_idx = HISTC_MEM_SNOOP,
|
|
|
|
};
|
|
|
|
|
2014-06-01 13:38:29 +00:00
|
|
|
struct sort_entry sort_mem_dcacheline = {
|
|
|
|
.se_header = "Data Cacheline",
|
|
|
|
.se_cmp = sort__dcacheline_cmp,
|
|
|
|
.se_snprintf = hist_entry__dcacheline_snprintf,
|
|
|
|
.se_width_idx = HISTC_MEM_DCACHELINE,
|
|
|
|
};
|
|
|
|
|
2017-08-29 17:11:09 +00:00
|
|
|
static int64_t
|
|
|
|
sort__phys_daddr_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
uint64_t l = 0, r = 0;
|
|
|
|
|
|
|
|
if (left->mem_info)
|
|
|
|
l = left->mem_info->daddr.phys_addr;
|
|
|
|
if (right->mem_info)
|
|
|
|
r = right->mem_info->daddr.phys_addr;
|
|
|
|
|
|
|
|
return (int64_t)(r - l);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__phys_daddr_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
uint64_t addr = 0;
|
|
|
|
size_t ret = 0;
|
|
|
|
size_t len = BITS_PER_LONG / 4;
|
|
|
|
|
|
|
|
addr = he->mem_info->daddr.phys_addr;
|
|
|
|
|
|
|
|
ret += repsep_snprintf(bf + ret, size - ret, "[%c] ", he->level);
|
|
|
|
|
|
|
|
ret += repsep_snprintf(bf + ret, size - ret, "%-#.*llx", len, addr);
|
|
|
|
|
|
|
|
ret += repsep_snprintf(bf + ret, size - ret, "%-*s", width - ret, "");
|
|
|
|
|
|
|
|
if (ret > width)
|
|
|
|
bf[width] = '\0';
|
|
|
|
|
|
|
|
return width;
|
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_mem_phys_daddr = {
|
|
|
|
.se_header = "Data Physical Address",
|
|
|
|
.se_cmp = sort__phys_daddr_cmp,
|
|
|
|
.se_snprintf = hist_entry__phys_daddr_snprintf,
|
|
|
|
.se_width_idx = HISTC_MEM_PHYS_DADDR,
|
|
|
|
};
|
|
|
|
|
2013-09-20 14:40:41 +00:00
|
|
|
static int64_t
|
|
|
|
sort__abort_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2014-10-16 14:07:01 +00:00
|
|
|
if (!left->branch_info || !right->branch_info)
|
|
|
|
return cmp_null(left->branch_info, right->branch_info);
|
|
|
|
|
2013-09-20 14:40:41 +00:00
|
|
|
return left->branch_info->flags.abort !=
|
|
|
|
right->branch_info->flags.abort;
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__abort_snprintf(struct hist_entry *he, char *bf,
|
2013-09-20 14:40:41 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2014-10-16 14:07:01 +00:00
|
|
|
static const char *out = "N/A";
|
|
|
|
|
|
|
|
if (he->branch_info) {
|
|
|
|
if (he->branch_info->flags.abort)
|
|
|
|
out = "A";
|
|
|
|
else
|
|
|
|
out = ".";
|
|
|
|
}
|
2013-09-20 14:40:41 +00:00
|
|
|
|
|
|
|
return repsep_snprintf(bf, size, "%-*s", width, out);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_abort = {
|
|
|
|
.se_header = "Transaction abort",
|
|
|
|
.se_cmp = sort__abort_cmp,
|
|
|
|
.se_snprintf = hist_entry__abort_snprintf,
|
|
|
|
.se_width_idx = HISTC_ABORT,
|
|
|
|
};
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__in_tx_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
2014-10-16 14:07:02 +00:00
|
|
|
if (!left->branch_info || !right->branch_info)
|
|
|
|
return cmp_null(left->branch_info, right->branch_info);
|
|
|
|
|
2013-09-20 14:40:41 +00:00
|
|
|
return left->branch_info->flags.in_tx !=
|
|
|
|
right->branch_info->flags.in_tx;
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__in_tx_snprintf(struct hist_entry *he, char *bf,
|
2013-09-20 14:40:41 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2014-10-16 14:07:02 +00:00
|
|
|
static const char *out = "N/A";
|
2013-09-20 14:40:41 +00:00
|
|
|
|
2014-10-16 14:07:02 +00:00
|
|
|
if (he->branch_info) {
|
|
|
|
if (he->branch_info->flags.in_tx)
|
|
|
|
out = "T";
|
|
|
|
else
|
|
|
|
out = ".";
|
|
|
|
}
|
2013-09-20 14:40:41 +00:00
|
|
|
|
|
|
|
return repsep_snprintf(bf, size, "%-*s", width, out);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_in_tx = {
|
|
|
|
.se_header = "Branch in transaction",
|
|
|
|
.se_cmp = sort__in_tx_cmp,
|
|
|
|
.se_snprintf = hist_entry__in_tx_snprintf,
|
|
|
|
.se_width_idx = HISTC_IN_TX,
|
|
|
|
};
|
|
|
|
|
2013-09-20 14:40:43 +00:00
|
|
|
static int64_t
|
|
|
|
sort__transaction_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
return left->transaction - right->transaction;
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline char *add_str(char *p, const char *str)
|
|
|
|
{
|
|
|
|
strcpy(p, str);
|
|
|
|
return p + strlen(str);
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct txbit {
|
|
|
|
unsigned flag;
|
|
|
|
const char *name;
|
|
|
|
int skip_for_len;
|
|
|
|
} txbits[] = {
|
|
|
|
{ PERF_TXN_ELISION, "EL ", 0 },
|
|
|
|
{ PERF_TXN_TRANSACTION, "TX ", 1 },
|
|
|
|
{ PERF_TXN_SYNC, "SYNC ", 1 },
|
|
|
|
{ PERF_TXN_ASYNC, "ASYNC ", 0 },
|
|
|
|
{ PERF_TXN_RETRY, "RETRY ", 0 },
|
|
|
|
{ PERF_TXN_CONFLICT, "CON ", 0 },
|
|
|
|
{ PERF_TXN_CAPACITY_WRITE, "CAP-WRITE ", 1 },
|
|
|
|
{ PERF_TXN_CAPACITY_READ, "CAP-READ ", 0 },
|
|
|
|
{ 0, NULL, 0 }
|
|
|
|
};
|
|
|
|
|
|
|
|
int hist_entry__transaction_len(void)
|
|
|
|
{
|
|
|
|
int i;
|
|
|
|
int len = 0;
|
|
|
|
|
|
|
|
for (i = 0; txbits[i].name; i++) {
|
|
|
|
if (!txbits[i].skip_for_len)
|
|
|
|
len += strlen(txbits[i].name);
|
|
|
|
}
|
|
|
|
len += 4; /* :XX<space> */
|
|
|
|
return len;
|
|
|
|
}
|
|
|
|
|
2013-10-22 22:01:31 +00:00
|
|
|
static int hist_entry__transaction_snprintf(struct hist_entry *he, char *bf,
|
2013-09-20 14:40:43 +00:00
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
2013-10-22 22:01:31 +00:00
|
|
|
u64 t = he->transaction;
|
2013-09-20 14:40:43 +00:00
|
|
|
char buf[128];
|
|
|
|
char *p = buf;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
buf[0] = 0;
|
|
|
|
for (i = 0; txbits[i].name; i++)
|
|
|
|
if (txbits[i].flag & t)
|
|
|
|
p = add_str(p, txbits[i].name);
|
|
|
|
if (t && !(t & (PERF_TXN_SYNC|PERF_TXN_ASYNC)))
|
|
|
|
p = add_str(p, "NEITHER ");
|
|
|
|
if (t & PERF_TXN_ABORT_MASK) {
|
|
|
|
sprintf(p, ":%" PRIx64,
|
|
|
|
(t & PERF_TXN_ABORT_MASK) >>
|
|
|
|
PERF_TXN_ABORT_SHIFT);
|
|
|
|
p += strlen(p);
|
|
|
|
}
|
|
|
|
|
|
|
|
return repsep_snprintf(bf, size, "%-*s", width, buf);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_transaction = {
|
|
|
|
.se_header = "Transaction ",
|
|
|
|
.se_cmp = sort__transaction_cmp,
|
|
|
|
.se_snprintf = hist_entry__transaction_snprintf,
|
|
|
|
.se_width_idx = HISTC_TRANSACTION,
|
|
|
|
};
|
|
|
|
|
2017-02-24 13:32:56 +00:00
|
|
|
/* --sort symbol_size */
|
|
|
|
|
|
|
|
static int64_t _sort__sym_size_cmp(struct symbol *sym_l, struct symbol *sym_r)
|
|
|
|
{
|
|
|
|
int64_t size_l = sym_l != NULL ? symbol__size(sym_l) : 0;
|
|
|
|
int64_t size_r = sym_r != NULL ? symbol__size(sym_r) : 0;
|
|
|
|
|
|
|
|
return size_l < size_r ? -1 :
|
|
|
|
size_l == size_r ? 0 : 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t
|
|
|
|
sort__sym_size_cmp(struct hist_entry *left, struct hist_entry *right)
|
|
|
|
{
|
|
|
|
return _sort__sym_size_cmp(right->ms.sym, left->ms.sym);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int _hist_entry__sym_size_snprintf(struct symbol *sym, char *bf,
|
|
|
|
size_t bf_size, unsigned int width)
|
|
|
|
{
|
|
|
|
if (sym)
|
|
|
|
return repsep_snprintf(bf, bf_size, "%*d", width, symbol__size(sym));
|
|
|
|
|
|
|
|
return repsep_snprintf(bf, bf_size, "%*s", width, "unknown");
|
|
|
|
}
|
|
|
|
|
|
|
|
static int hist_entry__sym_size_snprintf(struct hist_entry *he, char *bf,
|
|
|
|
size_t size, unsigned int width)
|
|
|
|
{
|
|
|
|
return _hist_entry__sym_size_snprintf(he->ms.sym, bf, size, width);
|
|
|
|
}
|
|
|
|
|
|
|
|
struct sort_entry sort_sym_size = {
|
|
|
|
.se_header = "Symbol size",
|
|
|
|
.se_cmp = sort__sym_size_cmp,
|
|
|
|
.se_snprintf = hist_entry__sym_size_snprintf,
|
|
|
|
.se_width_idx = HISTC_SYM_SIZE,
|
|
|
|
};
|
|
|
|
|
|
|
|
|
2011-06-29 01:14:52 +00:00
|
|
|
struct sort_dimension {
|
|
|
|
const char *name;
|
|
|
|
struct sort_entry *entry;
|
|
|
|
int taken;
|
|
|
|
};
|
|
|
|
|
2012-02-09 22:21:01 +00:00
|
|
|
#define DIM(d, n, func) [d] = { .name = n, .entry = &(func) }
|
|
|
|
|
2012-12-27 09:11:46 +00:00
|
|
|
static struct sort_dimension common_sort_dimensions[] = {
|
2012-02-09 22:21:01 +00:00
|
|
|
DIM(SORT_PID, "pid", sort_thread),
|
|
|
|
DIM(SORT_COMM, "comm", sort_comm),
|
|
|
|
DIM(SORT_DSO, "dso", sort_dso),
|
|
|
|
DIM(SORT_SYM, "symbol", sort_sym),
|
|
|
|
DIM(SORT_PARENT, "parent", sort_parent),
|
|
|
|
DIM(SORT_CPU, "cpu", sort_cpu),
|
2015-09-04 14:45:43 +00:00
|
|
|
DIM(SORT_SOCKET, "socket", sort_socket),
|
2012-05-30 13:33:24 +00:00
|
|
|
DIM(SORT_SRCLINE, "srcline", sort_srcline),
|
2015-08-07 22:54:24 +00:00
|
|
|
DIM(SORT_SRCFILE, "srcfile", sort_srcfile),
|
2013-07-18 22:58:53 +00:00
|
|
|
DIM(SORT_LOCAL_WEIGHT, "local_weight", sort_local_weight),
|
|
|
|
DIM(SORT_GLOBAL_WEIGHT, "weight", sort_global_weight),
|
2013-09-20 14:40:43 +00:00
|
|
|
DIM(SORT_TRANSACTION, "transaction", sort_transaction),
|
2015-12-22 17:07:04 +00:00
|
|
|
DIM(SORT_TRACE, "trace", sort_trace),
|
2017-02-24 13:32:56 +00:00
|
|
|
DIM(SORT_SYM_SIZE, "symbol_size", sort_sym_size),
|
perf tools: Add 'cgroup_id' sort order keyword
This patch introduces a cgroup identifier entry field in perf report to
identify or distinguish data of different cgroups. It uses the device
number and inode number of cgroup namespace, included in perf data with
the new PERF_RECORD_NAMESPACES event, as cgroup identifier.
With the assumption that each container is created with it's own cgroup
namespace, this allows assessment/analysis of multiple containers at
once.
A simple test for this would be to clone a few processes passing
SIGCHILD & CLONE_NEWCROUP flags to each of them, execute shell and run
different workloads on each of those contexts, while running perf
record command with --namespaces option.
Shown below is the output of perf report, sorted with cgroup identifier,
on perf.data generated with the above test scenario, clearly indicating
one context's considerable use of kernel memory in comparison with
others:
$ perf report -s cgroup_id,sample --stdio
#
# Total Lost Samples: 0
#
# Samples: 5K of event 'kmem:kmalloc'
# Event count (approx.): 5965
#
# Overhead cgroup id (dev/inode) Samples
# ........ ..................... ............
#
81.27% 3/0xeffffffb 4848
16.24% 3/0xf00000d0 969
1.16% 3/0xf00000ce 69
0.82% 3/0xf00000cf 49
0.50% 0/0x0 30
While this is a start, there is further scope of improving this. For
example, instead of cgroup namespace's device and inode numbers, dev
and inode numbers of some or all namespaces may be used to distinguish
which processes are running in a given container context.
Also, scripts to map device and inode info to containers sounds
plausible for better tracing of containers.
Signed-off-by: Hari Bathini <hbathini@linux.vnet.ibm.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexander Shishkin <alexander.shishkin@linux.intel.com>
Cc: Alexei Starovoitov <ast@fb.com>
Cc: Ananth N Mavinakayanahalli <ananth@linux.vnet.ibm.com>
Cc: Aravinda Prasad <aravinda@linux.vnet.ibm.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: Eric Biederman <ebiederm@xmission.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Sargun Dhillon <sargun@sargun.me>
Cc: Steven Rostedt <rostedt@goodmis.org>
Link: http://lkml.kernel.org/r/148891933338.25309.756882900782042645.stgit@hbathini.in.ibm.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2017-03-07 20:42:13 +00:00
|
|
|
DIM(SORT_CGROUP_ID, "cgroup_id", sort_cgroup_id),
|
2011-06-29 01:14:52 +00:00
|
|
|
};
|
|
|
|
|
2012-12-27 09:11:46 +00:00
|
|
|
#undef DIM
|
|
|
|
|
|
|
|
#define DIM(d, n, func) [d - __SORT_BRANCH_STACK] = { .name = n, .entry = &(func) }
|
|
|
|
|
|
|
|
static struct sort_dimension bstack_sort_dimensions[] = {
|
|
|
|
DIM(SORT_DSO_FROM, "dso_from", sort_dso_from),
|
|
|
|
DIM(SORT_DSO_TO, "dso_to", sort_dso_to),
|
|
|
|
DIM(SORT_SYM_FROM, "symbol_from", sort_sym_from),
|
|
|
|
DIM(SORT_SYM_TO, "symbol_to", sort_sym_to),
|
|
|
|
DIM(SORT_MISPREDICT, "mispredict", sort_mispredict),
|
2013-09-20 14:40:41 +00:00
|
|
|
DIM(SORT_IN_TX, "in_tx", sort_in_tx),
|
|
|
|
DIM(SORT_ABORT, "abort", sort_abort),
|
2015-07-18 15:24:46 +00:00
|
|
|
DIM(SORT_CYCLES, "cycles", sort_cycles),
|
2016-05-20 20:15:08 +00:00
|
|
|
DIM(SORT_SRCLINE_FROM, "srcline_from", sort_srcline_from),
|
|
|
|
DIM(SORT_SRCLINE_TO, "srcline_to", sort_srcline_to),
|
2012-12-27 09:11:46 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
#undef DIM
|
|
|
|
|
2013-04-03 12:26:11 +00:00
|
|
|
#define DIM(d, n, func) [d - __SORT_MEMORY_MODE] = { .name = n, .entry = &(func) }
|
|
|
|
|
|
|
|
static struct sort_dimension memory_sort_dimensions[] = {
|
|
|
|
DIM(SORT_MEM_DADDR_SYMBOL, "symbol_daddr", sort_mem_daddr_sym),
|
2015-10-05 18:06:07 +00:00
|
|
|
DIM(SORT_MEM_IADDR_SYMBOL, "symbol_iaddr", sort_mem_iaddr_sym),
|
2013-04-03 12:26:11 +00:00
|
|
|
DIM(SORT_MEM_DADDR_DSO, "dso_daddr", sort_mem_daddr_dso),
|
|
|
|
DIM(SORT_MEM_LOCKED, "locked", sort_mem_locked),
|
|
|
|
DIM(SORT_MEM_TLB, "tlb", sort_mem_tlb),
|
|
|
|
DIM(SORT_MEM_LVL, "mem", sort_mem_lvl),
|
|
|
|
DIM(SORT_MEM_SNOOP, "snoop", sort_mem_snoop),
|
2014-06-01 13:38:29 +00:00
|
|
|
DIM(SORT_MEM_DCACHELINE, "dcacheline", sort_mem_dcacheline),
|
2017-08-29 17:11:09 +00:00
|
|
|
DIM(SORT_MEM_PHYS_DADDR, "phys_daddr", sort_mem_phys_daddr),
|
2013-04-03 12:26:11 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
#undef DIM
|
|
|
|
|
2014-03-04 00:06:42 +00:00
|
|
|
struct hpp_dimension {
|
|
|
|
const char *name;
|
|
|
|
struct perf_hpp_fmt *fmt;
|
|
|
|
int taken;
|
|
|
|
};
|
|
|
|
|
|
|
|
#define DIM(d, n) { .name = n, .fmt = &perf_hpp__format[d], }
|
|
|
|
|
|
|
|
static struct hpp_dimension hpp_sort_dimensions[] = {
|
|
|
|
DIM(PERF_HPP__OVERHEAD, "overhead"),
|
|
|
|
DIM(PERF_HPP__OVERHEAD_SYS, "overhead_sys"),
|
|
|
|
DIM(PERF_HPP__OVERHEAD_US, "overhead_us"),
|
|
|
|
DIM(PERF_HPP__OVERHEAD_GUEST_SYS, "overhead_guest_sys"),
|
|
|
|
DIM(PERF_HPP__OVERHEAD_GUEST_US, "overhead_guest_us"),
|
2013-10-30 07:06:59 +00:00
|
|
|
DIM(PERF_HPP__OVERHEAD_ACC, "overhead_children"),
|
2014-03-04 00:06:42 +00:00
|
|
|
DIM(PERF_HPP__SAMPLES, "sample"),
|
|
|
|
DIM(PERF_HPP__PERIOD, "period"),
|
|
|
|
};
|
|
|
|
|
|
|
|
#undef DIM
|
|
|
|
|
2014-03-03 02:46:55 +00:00
|
|
|
struct hpp_sort_entry {
|
|
|
|
struct perf_hpp_fmt hpp;
|
|
|
|
struct sort_entry *se;
|
|
|
|
};
|
|
|
|
|
2014-07-31 05:47:37 +00:00
|
|
|
void perf_hpp__reset_sort_width(struct perf_hpp_fmt *fmt, struct hists *hists)
|
2014-03-20 02:18:54 +00:00
|
|
|
{
|
|
|
|
struct hpp_sort_entry *hse;
|
|
|
|
|
|
|
|
if (!perf_hpp__is_sort_entry(fmt))
|
|
|
|
return;
|
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
2014-07-31 05:47:40 +00:00
|
|
|
hists__new_col_len(hists, hse->se->se_width_idx, strlen(fmt->name));
|
2014-03-20 02:18:54 +00:00
|
|
|
}
|
|
|
|
|
2014-03-03 02:46:55 +00:00
|
|
|
static int __sort__hpp_header(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
|
2016-08-07 15:28:30 +00:00
|
|
|
struct hists *hists, int line __maybe_unused,
|
|
|
|
int *span __maybe_unused)
|
2014-03-03 02:46:55 +00:00
|
|
|
{
|
|
|
|
struct hpp_sort_entry *hse;
|
2014-07-31 05:47:38 +00:00
|
|
|
size_t len = fmt->user_len;
|
2014-03-03 02:46:55 +00:00
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
|
|
|
|
2014-07-31 05:47:38 +00:00
|
|
|
if (!len)
|
2016-06-14 18:19:19 +00:00
|
|
|
len = hists__col_len(hists, hse->se->se_width_idx);
|
2014-07-31 05:47:38 +00:00
|
|
|
|
2014-07-31 05:47:40 +00:00
|
|
|
return scnprintf(hpp->buf, hpp->size, "%-*.*s", len, len, fmt->name);
|
2014-03-03 02:46:55 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int __sort__hpp_width(struct perf_hpp_fmt *fmt,
|
|
|
|
struct perf_hpp *hpp __maybe_unused,
|
2016-06-14 18:19:20 +00:00
|
|
|
struct hists *hists)
|
2014-03-03 02:46:55 +00:00
|
|
|
{
|
|
|
|
struct hpp_sort_entry *hse;
|
2014-07-31 05:47:38 +00:00
|
|
|
size_t len = fmt->user_len;
|
2014-03-03 02:46:55 +00:00
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
|
|
|
|
2014-07-31 05:47:38 +00:00
|
|
|
if (!len)
|
2016-06-14 18:19:20 +00:00
|
|
|
len = hists__col_len(hists, hse->se->se_width_idx);
|
2014-07-31 05:47:38 +00:00
|
|
|
|
|
|
|
return len;
|
2014-03-03 02:46:55 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static int __sort__hpp_entry(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
|
|
|
|
struct hist_entry *he)
|
|
|
|
{
|
|
|
|
struct hpp_sort_entry *hse;
|
2014-07-31 05:47:38 +00:00
|
|
|
size_t len = fmt->user_len;
|
2014-03-03 02:46:55 +00:00
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
2014-07-31 05:47:38 +00:00
|
|
|
|
|
|
|
if (!len)
|
|
|
|
len = hists__col_len(he->hists, hse->se->se_width_idx);
|
2014-03-03 02:46:55 +00:00
|
|
|
|
|
|
|
return hse->se->se_snprintf(he, hpp->buf, hpp->size, len);
|
|
|
|
}
|
|
|
|
|
2015-01-08 00:45:46 +00:00
|
|
|
static int64_t __sort__hpp_cmp(struct perf_hpp_fmt *fmt,
|
|
|
|
struct hist_entry *a, struct hist_entry *b)
|
|
|
|
{
|
|
|
|
struct hpp_sort_entry *hse;
|
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
|
|
|
return hse->se->se_cmp(a, b);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t __sort__hpp_collapse(struct perf_hpp_fmt *fmt,
|
|
|
|
struct hist_entry *a, struct hist_entry *b)
|
|
|
|
{
|
|
|
|
struct hpp_sort_entry *hse;
|
|
|
|
int64_t (*collapse_fn)(struct hist_entry *, struct hist_entry *);
|
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
|
|
|
collapse_fn = hse->se->se_collapse ?: hse->se->se_cmp;
|
|
|
|
return collapse_fn(a, b);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t __sort__hpp_sort(struct perf_hpp_fmt *fmt,
|
|
|
|
struct hist_entry *a, struct hist_entry *b)
|
|
|
|
{
|
|
|
|
struct hpp_sort_entry *hse;
|
|
|
|
int64_t (*sort_fn)(struct hist_entry *, struct hist_entry *);
|
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
|
|
|
sort_fn = hse->se->se_sort ?: hse->se->se_cmp;
|
|
|
|
return sort_fn(a, b);
|
|
|
|
}
|
|
|
|
|
2016-01-18 09:24:03 +00:00
|
|
|
bool perf_hpp__is_sort_entry(struct perf_hpp_fmt *format)
|
|
|
|
{
|
|
|
|
return format->header == __sort__hpp_header;
|
|
|
|
}
|
|
|
|
|
2016-03-09 13:46:57 +00:00
|
|
|
#define MK_SORT_ENTRY_CHK(key) \
|
|
|
|
bool perf_hpp__is_ ## key ## _entry(struct perf_hpp_fmt *fmt) \
|
|
|
|
{ \
|
|
|
|
struct hpp_sort_entry *hse; \
|
|
|
|
\
|
|
|
|
if (!perf_hpp__is_sort_entry(fmt)) \
|
|
|
|
return false; \
|
|
|
|
\
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp); \
|
|
|
|
return hse->se == &sort_ ## key ; \
|
|
|
|
}
|
|
|
|
|
|
|
|
MK_SORT_ENTRY_CHK(trace)
|
|
|
|
MK_SORT_ENTRY_CHK(srcline)
|
|
|
|
MK_SORT_ENTRY_CHK(srcfile)
|
|
|
|
MK_SORT_ENTRY_CHK(thread)
|
|
|
|
MK_SORT_ENTRY_CHK(comm)
|
|
|
|
MK_SORT_ENTRY_CHK(dso)
|
|
|
|
MK_SORT_ENTRY_CHK(sym)
|
2016-02-24 15:13:33 +00:00
|
|
|
|
|
|
|
|
2016-01-18 09:24:03 +00:00
|
|
|
static bool __sort__hpp_equal(struct perf_hpp_fmt *a, struct perf_hpp_fmt *b)
|
|
|
|
{
|
|
|
|
struct hpp_sort_entry *hse_a;
|
|
|
|
struct hpp_sort_entry *hse_b;
|
|
|
|
|
|
|
|
if (!perf_hpp__is_sort_entry(a) || !perf_hpp__is_sort_entry(b))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
hse_a = container_of(a, struct hpp_sort_entry, hpp);
|
|
|
|
hse_b = container_of(b, struct hpp_sort_entry, hpp);
|
|
|
|
|
|
|
|
return hse_a->se == hse_b->se;
|
|
|
|
}
|
|
|
|
|
2016-01-18 09:24:09 +00:00
|
|
|
static void hse_free(struct perf_hpp_fmt *fmt)
|
|
|
|
{
|
|
|
|
struct hpp_sort_entry *hse;
|
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
|
|
|
free(hse);
|
|
|
|
}
|
|
|
|
|
2014-03-04 01:46:34 +00:00
|
|
|
static struct hpp_sort_entry *
|
2016-03-07 19:44:43 +00:00
|
|
|
__sort_dimension__alloc_hpp(struct sort_dimension *sd, int level)
|
2014-03-03 02:46:55 +00:00
|
|
|
{
|
|
|
|
struct hpp_sort_entry *hse;
|
|
|
|
|
|
|
|
hse = malloc(sizeof(*hse));
|
|
|
|
if (hse == NULL) {
|
|
|
|
pr_err("Memory allocation failed\n");
|
2014-03-04 01:46:34 +00:00
|
|
|
return NULL;
|
2014-03-03 02:46:55 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
hse->se = sd->entry;
|
2014-07-31 05:47:40 +00:00
|
|
|
hse->hpp.name = sd->entry->se_header;
|
2014-03-03 02:46:55 +00:00
|
|
|
hse->hpp.header = __sort__hpp_header;
|
|
|
|
hse->hpp.width = __sort__hpp_width;
|
|
|
|
hse->hpp.entry = __sort__hpp_entry;
|
|
|
|
hse->hpp.color = NULL;
|
|
|
|
|
2015-01-08 00:45:46 +00:00
|
|
|
hse->hpp.cmp = __sort__hpp_cmp;
|
|
|
|
hse->hpp.collapse = __sort__hpp_collapse;
|
|
|
|
hse->hpp.sort = __sort__hpp_sort;
|
2016-01-18 09:24:03 +00:00
|
|
|
hse->hpp.equal = __sort__hpp_equal;
|
2016-01-18 09:24:09 +00:00
|
|
|
hse->hpp.free = hse_free;
|
2014-03-03 02:46:55 +00:00
|
|
|
|
|
|
|
INIT_LIST_HEAD(&hse->hpp.list);
|
|
|
|
INIT_LIST_HEAD(&hse->hpp.sort_list);
|
perf tools: Move elide bool into perf_hpp_fmt struct
After output/sort fields refactoring, it's expensive
to check the elide bool in its current location inside
the 'struct sort_entry'.
The perf_hpp__should_skip function gets highly noticable in
workloads with high number of output/sort fields, like for:
$ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio
Performance report:
9.70% perf [.] perf_hpp__should_skip
Moving the elide bool into the 'struct perf_hpp_fmt', which
makes the perf_hpp__should_skip just single struct read.
Got speedup of around 22% for my test perf.data workload.
The change should not harm any other workload types.
Performance counter stats for (10 runs):
before:
358,319,732,626 cycles ( +- 0.55% )
467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% )
150.943975206 seconds time elapsed ( +- 0.62% )
now:
278,785,972,990 cycles ( +- 0.12% )
370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% )
116.416670507 seconds time elapsed ( +- 0.31% )
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20140601142622.GA9131@krava.brq.redhat.com
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
2014-05-23 15:15:47 +00:00
|
|
|
hse->hpp.elide = false;
|
2014-07-31 05:47:37 +00:00
|
|
|
hse->hpp.len = 0;
|
2014-07-31 05:47:38 +00:00
|
|
|
hse->hpp.user_len = 0;
|
2016-03-07 19:44:43 +00:00
|
|
|
hse->hpp.level = level;
|
2014-03-03 02:46:55 +00:00
|
|
|
|
2014-03-04 01:46:34 +00:00
|
|
|
return hse;
|
|
|
|
}
|
|
|
|
|
2016-01-18 09:24:09 +00:00
|
|
|
static void hpp_free(struct perf_hpp_fmt *fmt)
|
|
|
|
{
|
|
|
|
free(fmt);
|
|
|
|
}
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
static struct perf_hpp_fmt *__hpp_dimension__alloc_hpp(struct hpp_dimension *hd,
|
|
|
|
int level)
|
2016-01-18 09:24:07 +00:00
|
|
|
{
|
|
|
|
struct perf_hpp_fmt *fmt;
|
|
|
|
|
|
|
|
fmt = memdup(hd->fmt, sizeof(*fmt));
|
|
|
|
if (fmt) {
|
|
|
|
INIT_LIST_HEAD(&fmt->list);
|
|
|
|
INIT_LIST_HEAD(&fmt->sort_list);
|
2016-01-18 09:24:09 +00:00
|
|
|
fmt->free = hpp_free;
|
2016-03-07 19:44:43 +00:00
|
|
|
fmt->level = level;
|
2016-01-18 09:24:07 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return fmt;
|
|
|
|
}
|
|
|
|
|
2016-02-24 15:13:37 +00:00
|
|
|
int hist_entry__filter(struct hist_entry *he, int type, const void *arg)
|
|
|
|
{
|
|
|
|
struct perf_hpp_fmt *fmt;
|
|
|
|
struct hpp_sort_entry *hse;
|
2016-03-09 13:46:56 +00:00
|
|
|
int ret = -1;
|
|
|
|
int r;
|
|
|
|
|
|
|
|
perf_hpp_list__for_each_format(he->hpp_list, fmt) {
|
|
|
|
if (!perf_hpp__is_sort_entry(fmt))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
|
|
|
if (hse->se->se_filter == NULL)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* hist entry is filtered if any of sort key in the hpp list
|
|
|
|
* is applied. But it should skip non-matched filter types.
|
|
|
|
*/
|
|
|
|
r = hse->se->se_filter(he, type, arg);
|
|
|
|
if (r >= 0) {
|
|
|
|
if (ret < 0)
|
|
|
|
ret = 0;
|
|
|
|
ret |= r;
|
|
|
|
}
|
|
|
|
}
|
2016-02-24 15:13:37 +00:00
|
|
|
|
2016-03-09 13:46:56 +00:00
|
|
|
return ret;
|
2016-02-24 15:13:37 +00:00
|
|
|
}
|
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
static int __sort_dimension__add_hpp_sort(struct sort_dimension *sd,
|
|
|
|
struct perf_hpp_list *list,
|
|
|
|
int level)
|
2014-03-04 01:46:34 +00:00
|
|
|
{
|
2016-03-07 19:44:43 +00:00
|
|
|
struct hpp_sort_entry *hse = __sort_dimension__alloc_hpp(sd, level);
|
2014-03-04 01:46:34 +00:00
|
|
|
|
|
|
|
if (hse == NULL)
|
|
|
|
return -1;
|
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
perf_hpp_list__register_sort_field(list, &hse->hpp);
|
2014-03-03 02:46:55 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
static int __sort_dimension__add_hpp_output(struct sort_dimension *sd,
|
|
|
|
struct perf_hpp_list *list)
|
2014-03-04 01:46:34 +00:00
|
|
|
{
|
2016-03-07 19:44:43 +00:00
|
|
|
struct hpp_sort_entry *hse = __sort_dimension__alloc_hpp(sd, 0);
|
2014-03-04 01:46:34 +00:00
|
|
|
|
|
|
|
if (hse == NULL)
|
|
|
|
return -1;
|
|
|
|
|
2016-01-18 09:24:16 +00:00
|
|
|
perf_hpp_list__column_register(list, &hse->hpp);
|
2014-03-04 01:46:34 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
struct hpp_dynamic_entry {
|
|
|
|
struct perf_hpp_fmt hpp;
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
struct format_field *field;
|
|
|
|
unsigned dynamic_len;
|
2015-12-22 17:07:05 +00:00
|
|
|
bool raw_trace;
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
};
|
|
|
|
|
|
|
|
static int hde_width(struct hpp_dynamic_entry *hde)
|
|
|
|
{
|
|
|
|
if (!hde->hpp.len) {
|
|
|
|
int len = hde->dynamic_len;
|
|
|
|
int namelen = strlen(hde->field->name);
|
|
|
|
int fieldlen = hde->field->size;
|
|
|
|
|
|
|
|
if (namelen > len)
|
|
|
|
len = namelen;
|
|
|
|
|
|
|
|
if (!(hde->field->flags & FIELD_IS_STRING)) {
|
|
|
|
/* length for print hex numbers */
|
|
|
|
fieldlen = hde->field->size * 2 + 2;
|
|
|
|
}
|
|
|
|
if (fieldlen > len)
|
|
|
|
len = fieldlen;
|
|
|
|
|
|
|
|
hde->hpp.len = len;
|
|
|
|
}
|
|
|
|
return hde->hpp.len;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:03 +00:00
|
|
|
static void update_dynamic_len(struct hpp_dynamic_entry *hde,
|
|
|
|
struct hist_entry *he)
|
|
|
|
{
|
|
|
|
char *str, *pos;
|
|
|
|
struct format_field *field = hde->field;
|
|
|
|
size_t namelen;
|
|
|
|
bool last = false;
|
|
|
|
|
2015-12-22 17:07:05 +00:00
|
|
|
if (hde->raw_trace)
|
|
|
|
return;
|
|
|
|
|
2015-12-22 17:07:03 +00:00
|
|
|
/* parse pretty print result and update max length */
|
|
|
|
if (!he->trace_output)
|
|
|
|
he->trace_output = get_trace_output(he);
|
|
|
|
|
|
|
|
namelen = strlen(field->name);
|
|
|
|
str = he->trace_output;
|
|
|
|
|
|
|
|
while (str) {
|
|
|
|
pos = strchr(str, ' ');
|
|
|
|
if (pos == NULL) {
|
|
|
|
last = true;
|
|
|
|
pos = str + strlen(str);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!strncmp(str, field->name, namelen)) {
|
|
|
|
size_t len;
|
|
|
|
|
|
|
|
str += namelen + 1;
|
|
|
|
len = pos - str;
|
|
|
|
|
|
|
|
if (len > hde->dynamic_len)
|
|
|
|
hde->dynamic_len = len;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (last)
|
|
|
|
str = NULL;
|
|
|
|
else
|
|
|
|
str = pos + 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
static int __sort__hde_header(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
|
2016-08-07 15:28:27 +00:00
|
|
|
struct hists *hists __maybe_unused,
|
2016-08-07 15:28:30 +00:00
|
|
|
int line __maybe_unused,
|
|
|
|
int *span __maybe_unused)
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
{
|
|
|
|
struct hpp_dynamic_entry *hde;
|
|
|
|
size_t len = fmt->user_len;
|
|
|
|
|
|
|
|
hde = container_of(fmt, struct hpp_dynamic_entry, hpp);
|
|
|
|
|
|
|
|
if (!len)
|
|
|
|
len = hde_width(hde);
|
|
|
|
|
|
|
|
return scnprintf(hpp->buf, hpp->size, "%*.*s", len, len, hde->field->name);
|
|
|
|
}
|
|
|
|
|
|
|
|
static int __sort__hde_width(struct perf_hpp_fmt *fmt,
|
|
|
|
struct perf_hpp *hpp __maybe_unused,
|
2016-06-14 18:19:20 +00:00
|
|
|
struct hists *hists __maybe_unused)
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
{
|
|
|
|
struct hpp_dynamic_entry *hde;
|
|
|
|
size_t len = fmt->user_len;
|
|
|
|
|
|
|
|
hde = container_of(fmt, struct hpp_dynamic_entry, hpp);
|
|
|
|
|
|
|
|
if (!len)
|
|
|
|
len = hde_width(hde);
|
|
|
|
|
|
|
|
return len;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:08 +00:00
|
|
|
bool perf_hpp__defined_dynamic_entry(struct perf_hpp_fmt *fmt, struct hists *hists)
|
|
|
|
{
|
|
|
|
struct hpp_dynamic_entry *hde;
|
|
|
|
|
|
|
|
hde = container_of(fmt, struct hpp_dynamic_entry, hpp);
|
|
|
|
|
|
|
|
return hists_to_evsel(hists) == hde->evsel;
|
|
|
|
}
|
|
|
|
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
static int __sort__hde_entry(struct perf_hpp_fmt *fmt, struct perf_hpp *hpp,
|
|
|
|
struct hist_entry *he)
|
|
|
|
{
|
|
|
|
struct hpp_dynamic_entry *hde;
|
|
|
|
size_t len = fmt->user_len;
|
2015-12-22 17:07:03 +00:00
|
|
|
char *str, *pos;
|
|
|
|
struct format_field *field;
|
|
|
|
size_t namelen;
|
|
|
|
bool last = false;
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
int ret;
|
|
|
|
|
|
|
|
hde = container_of(fmt, struct hpp_dynamic_entry, hpp);
|
|
|
|
|
|
|
|
if (!len)
|
|
|
|
len = hde_width(hde);
|
|
|
|
|
2015-12-22 17:07:05 +00:00
|
|
|
if (hde->raw_trace)
|
|
|
|
goto raw_field;
|
2015-12-22 17:07:03 +00:00
|
|
|
|
2016-02-26 18:52:46 +00:00
|
|
|
if (!he->trace_output)
|
|
|
|
he->trace_output = get_trace_output(he);
|
|
|
|
|
2015-12-22 17:07:05 +00:00
|
|
|
field = hde->field;
|
2015-12-22 17:07:03 +00:00
|
|
|
namelen = strlen(field->name);
|
|
|
|
str = he->trace_output;
|
|
|
|
|
|
|
|
while (str) {
|
|
|
|
pos = strchr(str, ' ');
|
|
|
|
if (pos == NULL) {
|
|
|
|
last = true;
|
|
|
|
pos = str + strlen(str);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (!strncmp(str, field->name, namelen)) {
|
|
|
|
str += namelen + 1;
|
|
|
|
str = strndup(str, pos - str);
|
|
|
|
|
|
|
|
if (str == NULL)
|
|
|
|
return scnprintf(hpp->buf, hpp->size,
|
|
|
|
"%*.*s", len, len, "ERROR");
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (last)
|
|
|
|
str = NULL;
|
|
|
|
else
|
|
|
|
str = pos + 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (str == NULL) {
|
|
|
|
struct trace_seq seq;
|
2015-12-22 17:07:05 +00:00
|
|
|
raw_field:
|
2015-12-22 17:07:03 +00:00
|
|
|
trace_seq_init(&seq);
|
|
|
|
pevent_print_field(&seq, he->raw_data, hde->field);
|
|
|
|
str = seq.buffer;
|
|
|
|
}
|
|
|
|
|
|
|
|
ret = scnprintf(hpp->buf, hpp->size, "%*.*s", len, len, str);
|
|
|
|
free(str);
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int64_t __sort__hde_cmp(struct perf_hpp_fmt *fmt,
|
|
|
|
struct hist_entry *a, struct hist_entry *b)
|
|
|
|
{
|
|
|
|
struct hpp_dynamic_entry *hde;
|
|
|
|
struct format_field *field;
|
|
|
|
unsigned offset, size;
|
|
|
|
|
|
|
|
hde = container_of(fmt, struct hpp_dynamic_entry, hpp);
|
|
|
|
|
2016-02-26 18:52:47 +00:00
|
|
|
if (b == NULL) {
|
|
|
|
update_dynamic_len(hde, a);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
field = hde->field;
|
|
|
|
if (field->flags & FIELD_IS_DYNAMIC) {
|
|
|
|
unsigned long long dyn;
|
|
|
|
|
|
|
|
pevent_read_number_field(field, a->raw_data, &dyn);
|
|
|
|
offset = dyn & 0xffff;
|
|
|
|
size = (dyn >> 16) & 0xffff;
|
|
|
|
|
|
|
|
/* record max width for output */
|
|
|
|
if (size > hde->dynamic_len)
|
|
|
|
hde->dynamic_len = size;
|
|
|
|
} else {
|
|
|
|
offset = field->offset;
|
|
|
|
size = field->size;
|
|
|
|
}
|
|
|
|
|
|
|
|
return memcmp(a->raw_data + offset, b->raw_data + offset, size);
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:08 +00:00
|
|
|
bool perf_hpp__is_dynamic_entry(struct perf_hpp_fmt *fmt)
|
|
|
|
{
|
|
|
|
return fmt->cmp == __sort__hde_cmp;
|
|
|
|
}
|
|
|
|
|
2016-02-21 14:22:35 +00:00
|
|
|
static bool __sort__hde_equal(struct perf_hpp_fmt *a, struct perf_hpp_fmt *b)
|
|
|
|
{
|
|
|
|
struct hpp_dynamic_entry *hde_a;
|
|
|
|
struct hpp_dynamic_entry *hde_b;
|
|
|
|
|
|
|
|
if (!perf_hpp__is_dynamic_entry(a) || !perf_hpp__is_dynamic_entry(b))
|
|
|
|
return false;
|
|
|
|
|
|
|
|
hde_a = container_of(a, struct hpp_dynamic_entry, hpp);
|
|
|
|
hde_b = container_of(b, struct hpp_dynamic_entry, hpp);
|
|
|
|
|
|
|
|
return hde_a->field == hde_b->field;
|
|
|
|
}
|
|
|
|
|
2016-01-18 09:24:09 +00:00
|
|
|
static void hde_free(struct perf_hpp_fmt *fmt)
|
|
|
|
{
|
|
|
|
struct hpp_dynamic_entry *hde;
|
|
|
|
|
|
|
|
hde = container_of(fmt, struct hpp_dynamic_entry, hpp);
|
|
|
|
free(hde);
|
|
|
|
}
|
|
|
|
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
static struct hpp_dynamic_entry *
|
2016-03-07 19:44:43 +00:00
|
|
|
__alloc_dynamic_entry(struct perf_evsel *evsel, struct format_field *field,
|
|
|
|
int level)
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
{
|
|
|
|
struct hpp_dynamic_entry *hde;
|
|
|
|
|
|
|
|
hde = malloc(sizeof(*hde));
|
|
|
|
if (hde == NULL) {
|
|
|
|
pr_debug("Memory allocation failed\n");
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
hde->evsel = evsel;
|
|
|
|
hde->field = field;
|
|
|
|
hde->dynamic_len = 0;
|
|
|
|
|
|
|
|
hde->hpp.name = field->name;
|
|
|
|
hde->hpp.header = __sort__hde_header;
|
|
|
|
hde->hpp.width = __sort__hde_width;
|
|
|
|
hde->hpp.entry = __sort__hde_entry;
|
|
|
|
hde->hpp.color = NULL;
|
|
|
|
|
|
|
|
hde->hpp.cmp = __sort__hde_cmp;
|
|
|
|
hde->hpp.collapse = __sort__hde_cmp;
|
|
|
|
hde->hpp.sort = __sort__hde_cmp;
|
2016-02-21 14:22:35 +00:00
|
|
|
hde->hpp.equal = __sort__hde_equal;
|
2016-01-18 09:24:09 +00:00
|
|
|
hde->hpp.free = hde_free;
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
|
|
|
|
INIT_LIST_HEAD(&hde->hpp.list);
|
|
|
|
INIT_LIST_HEAD(&hde->hpp.sort_list);
|
|
|
|
hde->hpp.elide = false;
|
|
|
|
hde->hpp.len = 0;
|
|
|
|
hde->hpp.user_len = 0;
|
2016-03-07 19:44:43 +00:00
|
|
|
hde->hpp.level = level;
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
|
|
|
|
return hde;
|
|
|
|
}
|
|
|
|
|
2016-03-07 19:44:45 +00:00
|
|
|
struct perf_hpp_fmt *perf_hpp_fmt__dup(struct perf_hpp_fmt *fmt)
|
|
|
|
{
|
|
|
|
struct perf_hpp_fmt *new_fmt = NULL;
|
|
|
|
|
|
|
|
if (perf_hpp__is_sort_entry(fmt)) {
|
|
|
|
struct hpp_sort_entry *hse, *new_hse;
|
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
|
|
|
new_hse = memdup(hse, sizeof(*hse));
|
|
|
|
if (new_hse)
|
|
|
|
new_fmt = &new_hse->hpp;
|
|
|
|
} else if (perf_hpp__is_dynamic_entry(fmt)) {
|
|
|
|
struct hpp_dynamic_entry *hde, *new_hde;
|
|
|
|
|
|
|
|
hde = container_of(fmt, struct hpp_dynamic_entry, hpp);
|
|
|
|
new_hde = memdup(hde, sizeof(*hde));
|
|
|
|
if (new_hde)
|
|
|
|
new_fmt = &new_hde->hpp;
|
|
|
|
} else {
|
|
|
|
new_fmt = memdup(fmt, sizeof(*fmt));
|
|
|
|
}
|
|
|
|
|
|
|
|
INIT_LIST_HEAD(&new_fmt->list);
|
|
|
|
INIT_LIST_HEAD(&new_fmt->sort_list);
|
|
|
|
|
|
|
|
return new_fmt;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:06 +00:00
|
|
|
static int parse_field_name(char *str, char **event, char **field, char **opt)
|
|
|
|
{
|
|
|
|
char *event_name, *field_name, *opt_name;
|
|
|
|
|
|
|
|
event_name = str;
|
|
|
|
field_name = strchr(str, '.');
|
|
|
|
|
|
|
|
if (field_name) {
|
|
|
|
*field_name++ = '\0';
|
|
|
|
} else {
|
|
|
|
event_name = NULL;
|
|
|
|
field_name = str;
|
|
|
|
}
|
|
|
|
|
|
|
|
opt_name = strchr(field_name, '/');
|
|
|
|
if (opt_name)
|
|
|
|
*opt_name++ = '\0';
|
|
|
|
|
|
|
|
*event = event_name;
|
|
|
|
*field = field_name;
|
|
|
|
*opt = opt_name;
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* find match evsel using a given event name. The event name can be:
|
2016-01-05 10:58:35 +00:00
|
|
|
* 1. '%' + event index (e.g. '%1' for first event)
|
|
|
|
* 2. full event name (e.g. sched:sched_switch)
|
|
|
|
* 3. partial event name (should not contain ':')
|
2015-12-22 17:07:06 +00:00
|
|
|
*/
|
|
|
|
static struct perf_evsel *find_evsel(struct perf_evlist *evlist, char *event_name)
|
|
|
|
{
|
|
|
|
struct perf_evsel *evsel = NULL;
|
|
|
|
struct perf_evsel *pos;
|
|
|
|
bool full_name;
|
|
|
|
|
|
|
|
/* case 1 */
|
|
|
|
if (event_name[0] == '%') {
|
|
|
|
int nr = strtol(event_name+1, NULL, 0);
|
|
|
|
|
|
|
|
if (nr > evlist->nr_entries)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
evsel = perf_evlist__first(evlist);
|
|
|
|
while (--nr > 0)
|
|
|
|
evsel = perf_evsel__next(evsel);
|
|
|
|
|
|
|
|
return evsel;
|
|
|
|
}
|
|
|
|
|
|
|
|
full_name = !!strchr(event_name, ':');
|
2016-06-23 14:26:15 +00:00
|
|
|
evlist__for_each_entry(evlist, pos) {
|
2016-01-05 10:58:35 +00:00
|
|
|
/* case 2 */
|
2015-12-22 17:07:06 +00:00
|
|
|
if (full_name && !strcmp(pos->name, event_name))
|
|
|
|
return pos;
|
2016-01-05 10:58:35 +00:00
|
|
|
/* case 3 */
|
2015-12-22 17:07:06 +00:00
|
|
|
if (!full_name && strstr(pos->name, event_name)) {
|
|
|
|
if (evsel) {
|
|
|
|
pr_debug("'%s' event is ambiguous: it can be %s or %s\n",
|
|
|
|
event_name, evsel->name, pos->name);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
evsel = pos;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return evsel;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:07 +00:00
|
|
|
static int __dynamic_dimension__add(struct perf_evsel *evsel,
|
|
|
|
struct format_field *field,
|
2016-03-07 19:44:43 +00:00
|
|
|
bool raw_trace, int level)
|
2015-12-22 17:07:07 +00:00
|
|
|
{
|
|
|
|
struct hpp_dynamic_entry *hde;
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
hde = __alloc_dynamic_entry(evsel, field, level);
|
2015-12-22 17:07:07 +00:00
|
|
|
if (hde == NULL)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
|
|
|
hde->raw_trace = raw_trace;
|
|
|
|
|
|
|
|
perf_hpp__register_sort_field(&hde->hpp);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
static int add_evsel_fields(struct perf_evsel *evsel, bool raw_trace, int level)
|
2015-12-22 17:07:09 +00:00
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
struct format_field *field;
|
|
|
|
|
|
|
|
field = evsel->tp_format->format.fields;
|
|
|
|
while (field) {
|
2016-03-07 19:44:43 +00:00
|
|
|
ret = __dynamic_dimension__add(evsel, field, raw_trace, level);
|
2015-12-22 17:07:09 +00:00
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
field = field->next;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
static int add_all_dynamic_fields(struct perf_evlist *evlist, bool raw_trace,
|
|
|
|
int level)
|
2015-12-22 17:07:09 +00:00
|
|
|
{
|
|
|
|
int ret;
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
|
2016-06-23 14:26:15 +00:00
|
|
|
evlist__for_each_entry(evlist, evsel) {
|
2015-12-22 17:07:09 +00:00
|
|
|
if (evsel->attr.type != PERF_TYPE_TRACEPOINT)
|
|
|
|
continue;
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
ret = add_evsel_fields(evsel, raw_trace, level);
|
2015-12-22 17:07:09 +00:00
|
|
|
if (ret < 0)
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-01-05 10:58:35 +00:00
|
|
|
static int add_all_matching_fields(struct perf_evlist *evlist,
|
2016-03-07 19:44:43 +00:00
|
|
|
char *field_name, bool raw_trace, int level)
|
2016-01-05 10:58:35 +00:00
|
|
|
{
|
|
|
|
int ret = -ESRCH;
|
|
|
|
struct perf_evsel *evsel;
|
|
|
|
struct format_field *field;
|
|
|
|
|
2016-06-23 14:26:15 +00:00
|
|
|
evlist__for_each_entry(evlist, evsel) {
|
2016-01-05 10:58:35 +00:00
|
|
|
if (evsel->attr.type != PERF_TYPE_TRACEPOINT)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
field = pevent_find_any_field(evsel->tp_format, field_name);
|
|
|
|
if (field == NULL)
|
|
|
|
continue;
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
ret = __dynamic_dimension__add(evsel, field, raw_trace, level);
|
2016-01-05 10:58:35 +00:00
|
|
|
if (ret < 0)
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
static int add_dynamic_entry(struct perf_evlist *evlist, const char *tok,
|
|
|
|
int level)
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
{
|
2015-12-22 17:07:06 +00:00
|
|
|
char *str, *event_name, *field_name, *opt_name;
|
|
|
|
struct perf_evsel *evsel;
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
struct format_field *field;
|
2015-12-22 17:07:05 +00:00
|
|
|
bool raw_trace = symbol_conf.raw_trace;
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
if (evlist == NULL)
|
|
|
|
return -ENOENT;
|
|
|
|
|
|
|
|
str = strdup(tok);
|
|
|
|
if (str == NULL)
|
|
|
|
return -ENOMEM;
|
|
|
|
|
2015-12-22 17:07:06 +00:00
|
|
|
if (parse_field_name(str, &event_name, &field_name, &opt_name) < 0) {
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
ret = -EINVAL;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:06 +00:00
|
|
|
if (opt_name) {
|
|
|
|
if (strcmp(opt_name, "raw")) {
|
|
|
|
pr_debug("unsupported field option %s\n", opt_name);
|
2015-12-22 17:07:05 +00:00
|
|
|
ret = -EINVAL;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
raw_trace = true;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:09 +00:00
|
|
|
if (!strcmp(field_name, "trace_fields")) {
|
2016-03-07 19:44:43 +00:00
|
|
|
ret = add_all_dynamic_fields(evlist, raw_trace, level);
|
2015-12-22 17:07:09 +00:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2016-01-05 10:58:35 +00:00
|
|
|
if (event_name == NULL) {
|
2016-03-07 19:44:43 +00:00
|
|
|
ret = add_all_matching_fields(evlist, field_name, raw_trace, level);
|
2016-01-05 10:58:35 +00:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:06 +00:00
|
|
|
evsel = find_evsel(evlist, event_name);
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
if (evsel == NULL) {
|
|
|
|
pr_debug("Cannot find event: %s\n", event_name);
|
|
|
|
ret = -ENOENT;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (evsel->attr.type != PERF_TYPE_TRACEPOINT) {
|
|
|
|
pr_debug("%s is not a tracepoint event\n", event_name);
|
|
|
|
ret = -EINVAL;
|
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:07 +00:00
|
|
|
if (!strcmp(field_name, "*")) {
|
2016-03-07 19:44:43 +00:00
|
|
|
ret = add_evsel_fields(evsel, raw_trace, level);
|
2015-12-22 17:07:07 +00:00
|
|
|
} else {
|
|
|
|
field = pevent_find_any_field(evsel->tp_format, field_name);
|
|
|
|
if (field == NULL) {
|
|
|
|
pr_debug("Cannot find event field for %s.%s\n",
|
|
|
|
event_name, field_name);
|
|
|
|
return -ENOENT;
|
|
|
|
}
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
ret = __dynamic_dimension__add(evsel, field, raw_trace, level);
|
2015-12-22 17:07:07 +00:00
|
|
|
}
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
|
|
|
|
out:
|
|
|
|
free(str);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
static int __sort_dimension__add(struct sort_dimension *sd,
|
|
|
|
struct perf_hpp_list *list,
|
|
|
|
int level)
|
2013-04-03 12:26:10 +00:00
|
|
|
{
|
|
|
|
if (sd->taken)
|
2014-03-03 02:46:55 +00:00
|
|
|
return 0;
|
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
if (__sort_dimension__add_hpp_sort(sd, list, level) < 0)
|
2014-03-03 02:46:55 +00:00
|
|
|
return -1;
|
2013-04-03 12:26:10 +00:00
|
|
|
|
|
|
|
if (sd->entry->se_collapse)
|
2016-05-03 11:54:42 +00:00
|
|
|
list->need_collapse = 1;
|
2013-04-03 12:26:10 +00:00
|
|
|
|
|
|
|
sd->taken = 1;
|
2014-03-03 02:46:55 +00:00
|
|
|
|
|
|
|
return 0;
|
2013-04-03 12:26:10 +00:00
|
|
|
}
|
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
static int __hpp_dimension__add(struct hpp_dimension *hd,
|
|
|
|
struct perf_hpp_list *list,
|
|
|
|
int level)
|
2014-03-04 00:06:42 +00:00
|
|
|
{
|
2016-01-18 09:24:07 +00:00
|
|
|
struct perf_hpp_fmt *fmt;
|
2014-03-04 00:06:42 +00:00
|
|
|
|
2016-01-18 09:24:07 +00:00
|
|
|
if (hd->taken)
|
|
|
|
return 0;
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
fmt = __hpp_dimension__alloc_hpp(hd, level);
|
2016-01-18 09:24:07 +00:00
|
|
|
if (!fmt)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
hd->taken = 1;
|
2016-03-09 10:04:17 +00:00
|
|
|
perf_hpp_list__register_sort_field(list, fmt);
|
2014-03-04 00:06:42 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-01-18 09:24:16 +00:00
|
|
|
static int __sort_dimension__add_output(struct perf_hpp_list *list,
|
|
|
|
struct sort_dimension *sd)
|
2014-03-04 01:46:34 +00:00
|
|
|
{
|
|
|
|
if (sd->taken)
|
|
|
|
return 0;
|
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
if (__sort_dimension__add_hpp_output(sd, list) < 0)
|
2014-03-04 01:46:34 +00:00
|
|
|
return -1;
|
|
|
|
|
|
|
|
sd->taken = 1;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-01-18 09:24:16 +00:00
|
|
|
static int __hpp_dimension__add_output(struct perf_hpp_list *list,
|
|
|
|
struct hpp_dimension *hd)
|
2014-03-04 01:46:34 +00:00
|
|
|
{
|
2016-01-18 09:24:07 +00:00
|
|
|
struct perf_hpp_fmt *fmt;
|
2014-03-04 01:46:34 +00:00
|
|
|
|
2016-01-18 09:24:07 +00:00
|
|
|
if (hd->taken)
|
|
|
|
return 0;
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
fmt = __hpp_dimension__alloc_hpp(hd, 0);
|
2016-01-18 09:24:07 +00:00
|
|
|
if (!fmt)
|
|
|
|
return -1;
|
|
|
|
|
|
|
|
hd->taken = 1;
|
2016-01-18 09:24:16 +00:00
|
|
|
perf_hpp_list__column_register(list, fmt);
|
2014-03-04 01:46:34 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2015-10-06 12:25:11 +00:00
|
|
|
int hpp_dimension__add_output(unsigned col)
|
|
|
|
{
|
|
|
|
BUG_ON(col >= PERF_HPP__MAX_INDEX);
|
2016-01-18 09:24:16 +00:00
|
|
|
return __hpp_dimension__add_output(&perf_hpp_list, &hpp_sort_dimensions[col]);
|
2015-10-06 12:25:11 +00:00
|
|
|
}
|
|
|
|
|
2016-09-22 15:36:33 +00:00
|
|
|
int sort_dimension__add(struct perf_hpp_list *list, const char *tok,
|
|
|
|
struct perf_evlist *evlist,
|
|
|
|
int level)
|
2009-09-24 16:02:49 +00:00
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
|
2012-12-27 09:11:46 +00:00
|
|
|
for (i = 0; i < ARRAY_SIZE(common_sort_dimensions); i++) {
|
|
|
|
struct sort_dimension *sd = &common_sort_dimensions[i];
|
2009-09-24 16:02:49 +00:00
|
|
|
|
|
|
|
if (strncasecmp(tok, sd->name, strlen(tok)))
|
|
|
|
continue;
|
2012-12-27 09:11:46 +00:00
|
|
|
|
2009-09-24 16:02:49 +00:00
|
|
|
if (sd->entry == &sort_parent) {
|
|
|
|
int ret = regcomp(&parent_regex, parent_pattern, REG_EXTENDED);
|
|
|
|
if (ret) {
|
|
|
|
char err[BUFSIZ];
|
|
|
|
|
|
|
|
regerror(ret, &parent_regex, err, sizeof(err));
|
2010-04-02 15:30:57 +00:00
|
|
|
pr_err("Invalid regex: %s\n%s", parent_pattern, err);
|
|
|
|
return -EINVAL;
|
2009-09-24 16:02:49 +00:00
|
|
|
}
|
2016-05-03 11:54:43 +00:00
|
|
|
list->parent = 1;
|
2013-04-05 01:26:36 +00:00
|
|
|
} else if (sd->entry == &sort_sym) {
|
2016-05-03 11:54:44 +00:00
|
|
|
list->sym = 1;
|
perf diff: Support for different binaries
Currently, the perf diff only works with same binaries. That's because
it compares the symbol start address. It doesn't work if the perf.data
comes from different binaries. This patch matches the symbol names.
Actually, perf diff once intended to compare the symbol names. The
commit as below can look for a pair by name.
604c5c92972d (perf diff: Change the default sort order to "dso,symbol")
However, at that time, perf diff used a global list of dsos. That means
the binaries which has same name can only be loaded once. That's a
problem for comparing different binaries.
For example, we have an old binary and an updated binary. They very
likely have same name and most of the functions, so only dsos from old
binary will be loaded. When processing the data from updated binary,
perf still use the symbol information from old binary. That's wrong.
Then the commit as below used IP to replace symbol name.
9c443dfdd31e ("perf diff: Fix support for all --sort combinations")
>From that time, perf diff starts to compare the symbol address.
The global dsos is discarded from a patch in 2010.
a1645ce12adb ("perf: 'perf kvm' tool for monitoring guest performance
from host")
However, at that time, perf diff already compared by address. So perf
diff cannot work for different binaries as well.
This patch actually rolls back the perf diff to original design. The
document is also changed, so everybody knows the original design is to
compare the symbol names.
Here are some examples:
The only difference between example_v1.c and example_v2.c is the
location of f2 and f3. There is no change in behavior, but the previous
perf diff display the wrong differential profile.
example_v1.c
noinline void f3(void)
{
volatile int i;
for (i = 0; i < 10000;) {
if(i%2)
i++;
else
i++;
}
}
noinline void f2(void)
{
volatile int a = 100, b, c;
for (b = 0; b < 10000; b++)
c = a * b;
}
noinline void f1(void)
{
f2();
f3();
}
int main()
{
int i;
for (i = 0; i < 100000; i++)
f1();
}
example_v2.c
noinline void f2(void)
{
volatile int a = 100, b, c;
for (b = 0; b < 10000; b++)
c = a * b;
}
noinline void f3(void)
{
volatile int i;
for (i = 0; i < 10000;) {
if(i%2)
i++;
else
i++;
}
}
noinline void f1(void)
{
f2();
f3();
}
int main()
{
int i;
for (i = 0; i < 100000; i++)
f1();
}
[lk@localhost perf_diff]$ gcc example_v1.c -o example
[lk@localhost perf_diff]$ perf record -o example_v1.data ./example
[ perf record: Woken up 4 times to write data ]
[ perf record: Captured and wrote 0.813 MB example_v1.data (~35522 samples) ]
[lk@localhost perf_diff]$ gcc example_v2.c -o example
[lk@localhost perf_diff]$ perf record -o example_v2.data ./example
[ perf record: Woken up 4 times to write data ]
[ perf record: Captured and wrote 0.824 MB example_v2.data (~36015 samples) ]
Old perf diff result:
[lk@localhost perf_diff]$ perf diff example_v1.data example_v2.data
Event 'cycles'
Baseline Delta Shared Object Symbol
........ ....... ................ ...............................
[kernel.vmlinux] [k] __perf_event_task_sched_out
0.00% [kernel.vmlinux] [k] apic_timer_interrupt
[kernel.vmlinux] [k] idle_cpu
[kernel.vmlinux] [k] intel_pstate_timer_func
[kernel.vmlinux] [k] native_read_msr_safe
0.00% [kernel.vmlinux] [k] native_read_tsc
0.00% [kernel.vmlinux] [k] native_write_msr_safe
[kernel.vmlinux] [k] ntp_tick_length
0.00% [kernel.vmlinux] [k] rb_erase
0.00% [kernel.vmlinux] [k] tick_sched_timer
0.00% [kernel.vmlinux] [k] unmap_single_vma
0.00% [kernel.vmlinux] [k] update_wall_time
0.00% example [.] f1
46.24% example [.] f2
53.71% -7.55% example [.] f3
+53.81% example [.] f3
0.02% example [.] main
New perf diff result:
[lk@localhost perf_diff]$ perf diff example_v1.data example_v2.data
[kernel.vmlinux] [k] __perf_event_task_sched_out
0.00% [kernel.vmlinux] [k] apic_timer_interrupt
[kernel.vmlinux] [k] idle_cpu
[kernel.vmlinux] [k] intel_pstate_timer_func
[kernel.vmlinux] [k] native_read_msr_safe
0.00% [kernel.vmlinux] [k] native_read_tsc
0.00% [kernel.vmlinux] [k] native_write_msr_safe
[kernel.vmlinux] [k] ntp_tick_length
0.00% [kernel.vmlinux] [k] rb_erase
0.00% [kernel.vmlinux] [k] tick_sched_timer
0.00% [kernel.vmlinux] [k] unmap_single_vma
0.00% [kernel.vmlinux] [k] update_wall_time
0.00% example [.] f1
46.24% -0.08% example [.] f2
53.71% +0.11% example [.] f3
0.02% example [.] main
Signed-off-by: Kan Liang <kan.liang@intel.com>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Andi Kleen <ak@linux.intel.com>
Link: http://lkml.kernel.org/r/1423460384-11645-1-git-send-email-kan.liang@intel.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-02-09 05:39:44 +00:00
|
|
|
/*
|
|
|
|
* perf diff displays the performance difference amongst
|
|
|
|
* two or more perf.data files. Those files could come
|
|
|
|
* from different binaries. So we should not compare
|
|
|
|
* their ips, but the name of symbol.
|
|
|
|
*/
|
|
|
|
if (sort__mode == SORT_MODE__DIFF)
|
|
|
|
sd->entry->se_collapse = sort__sym_sort;
|
|
|
|
|
2013-12-18 05:21:10 +00:00
|
|
|
} else if (sd->entry == &sort_dso) {
|
2016-05-03 11:54:45 +00:00
|
|
|
list->dso = 1;
|
2015-09-04 14:45:43 +00:00
|
|
|
} else if (sd->entry == &sort_socket) {
|
2016-05-03 11:54:46 +00:00
|
|
|
list->socket = 1;
|
2016-01-21 22:13:24 +00:00
|
|
|
} else if (sd->entry == &sort_thread) {
|
2016-05-03 11:54:47 +00:00
|
|
|
list->thread = 1;
|
2016-03-09 14:20:51 +00:00
|
|
|
} else if (sd->entry == &sort_comm) {
|
2016-05-03 11:54:48 +00:00
|
|
|
list->comm = 1;
|
2009-09-24 16:02:49 +00:00
|
|
|
}
|
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
return __sort_dimension__add(sd, list, level);
|
2009-09-24 16:02:49 +00:00
|
|
|
}
|
2012-12-27 09:11:46 +00:00
|
|
|
|
2014-03-04 00:06:42 +00:00
|
|
|
for (i = 0; i < ARRAY_SIZE(hpp_sort_dimensions); i++) {
|
|
|
|
struct hpp_dimension *hd = &hpp_sort_dimensions[i];
|
|
|
|
|
|
|
|
if (strncasecmp(tok, hd->name, strlen(tok)))
|
|
|
|
continue;
|
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
return __hpp_dimension__add(hd, list, level);
|
2014-03-04 00:06:42 +00:00
|
|
|
}
|
|
|
|
|
2012-12-27 09:11:46 +00:00
|
|
|
for (i = 0; i < ARRAY_SIZE(bstack_sort_dimensions); i++) {
|
|
|
|
struct sort_dimension *sd = &bstack_sort_dimensions[i];
|
|
|
|
|
|
|
|
if (strncasecmp(tok, sd->name, strlen(tok)))
|
|
|
|
continue;
|
|
|
|
|
2013-04-01 11:35:20 +00:00
|
|
|
if (sort__mode != SORT_MODE__BRANCH)
|
2012-12-27 09:11:46 +00:00
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
if (sd->entry == &sort_sym_from || sd->entry == &sort_sym_to)
|
2016-05-03 11:54:44 +00:00
|
|
|
list->sym = 1;
|
2012-12-27 09:11:46 +00:00
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
__sort_dimension__add(sd, list, level);
|
2012-12-27 09:11:46 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2013-04-03 12:26:11 +00:00
|
|
|
for (i = 0; i < ARRAY_SIZE(memory_sort_dimensions); i++) {
|
|
|
|
struct sort_dimension *sd = &memory_sort_dimensions[i];
|
|
|
|
|
|
|
|
if (strncasecmp(tok, sd->name, strlen(tok)))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
if (sort__mode != SORT_MODE__MEMORY)
|
|
|
|
return -EINVAL;
|
|
|
|
|
2016-07-15 13:08:43 +00:00
|
|
|
if (sd->entry == &sort_mem_dcacheline && cacheline_size == 0)
|
|
|
|
return -EINVAL;
|
|
|
|
|
2013-04-03 12:26:11 +00:00
|
|
|
if (sd->entry == &sort_mem_daddr_sym)
|
2016-05-03 11:54:44 +00:00
|
|
|
list->sym = 1;
|
2013-04-03 12:26:11 +00:00
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
__sort_dimension__add(sd, list, level);
|
2013-04-03 12:26:11 +00:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-03-07 19:44:43 +00:00
|
|
|
if (!add_dynamic_entry(evlist, tok, level))
|
perf tools: Add dynamic sort key for tracepoint events
The existing sort keys are less useful for tracepoint events in that
they are always sampled at the same place, the function where the
tracepoint is located.
For example, a 'perf report' on sched:sched_switch event looks like the
following:
# Overhead Command Shared Object Symbol
# ........ ............... ................ ..............
#
47.22% swapper [kernel.vmlinux] [k] __schedule
21.67% transmission-gt [kernel.vmlinux] [k] __schedule
8.23% netctl-auto [kernel.vmlinux] [k] __schedule
5.53% kworker/0:1H [kernel.vmlinux] [k] __schedule
1.98% Xephyr [kernel.vmlinux] [k] __schedule
1.33% irq/33-iwlwifi [kernel.vmlinux] [k] __schedule
1.17% wpa_cli [kernel.vmlinux] [k] __schedule
1.13% rcu_preempt [kernel.vmlinux] [k] __schedule
0.85% ksoftirqd/0 [kernel.vmlinux] [k] __schedule
0.77% Timer [kernel.vmlinux] [k] __schedule
In fact, tracepoints have meaningful information in their fields but
there's no way to use in 'perf report' currently. The dynamic sort keys
are introduced in this patc to overcome this limitation.
The sched:sched_switch events have following fields:
# sudo cat /sys/kernel/debug/tracing/events/sched/sched_switch/format
name: sched_switch
ID: 268
format:
field:unsigned short common_type; offset:0; size:2; signed:0;
field:unsigned char common_flags; offset:2; size:1; signed:0;
field:unsigned char common_preempt_count; offset:3; size:1; signed:0;
field:int common_pid; offset:4; size:4; signed:1;
field:char prev_comm[16]; offset:8; size:16; signed:1;
field:pid_t prev_pid; offset:24; size:4; signed:1;
field:int prev_prio; offset:28; size:4; signed:1;
field:long prev_state; offset:32; size:8; signed:1;
field:char next_comm[16]; offset:40; size:16; signed:1;
field:pid_t next_pid; offset:56; size:4; signed:1;
field:int next_prio; offset:60; size:4; signed:1;
print fmt: "prev_comm=%s prev_pid=%d prev_prio=%d prev_state=%s%s ==>
next_comm=%s next_pid=%d next_prio=%d",
REC->prev_comm, REC->prev_pid, REC->prev_prio,
REC->prev_state & (2048-1) ? __print_flags(REC->prev_state & (2048-1),
"|", { 1, "S"} , { 2, "D" }, { 4, "T" }, { 8, "t" }, { 16, "Z" }, { 32, "X" },
{ 64, "x" }, { 128, "K"}, { 256, "W" }, { 512, "P" }, { 1024, "N" }) : "R",
REC->prev_state & 2048 ? "+" : "", REC->next_comm, REC->next_pid, REC->next_prio
With dynamic sort keys, you can use <event.field> as a sort key. Those
dynamic keys are checked and created on demand. For instance, below is
to sort by next_pid field output on the same data file:
$ perf report -s comm,sched:sched_switch.next_pid --stdio
...
# Overhead Command next_pid
# ........ ............... ..........
#
21.23% transmission-gt 0
20.86% swapper 17773
6.62% netctl-auto 0
5.25% swapper 109
5.21% kworker/0:1H 0
1.98% Xephyr 0
1.98% swapper 6524
1.98% swapper 27478
1.37% swapper 27476
1.17% swapper 233
Multiple dynamic sort keys are also supported:
$ perf report -s comm,sched:sched_switch.next_pid,sched:sched_switch.next_comm --stdio
...
# Overhead Command next_pid next_comm
# ........ ............... .......... ................
#
20.86% swapper 17773 transmission-gt
9.64% transmission-gt 0 swapper/0
9.16% transmission-gt 0 swapper/2
5.25% swapper 109 kworker/0:1H
5.21% kworker/0:1H 0 swapper/0
2.14% netctl-auto 0 swapper/2
1.98% netctl-auto 0 swapper/0
1.98% swapper 6524 Xephyr
1.98% swapper 27478 netctl-auto
1.78% transmission-gt 0 swapper/3
1.53% Xephyr 0 swapper/0
1.29% netctl-auto 0 swapper/1
1.29% swapper 27476 netctl-auto
1.21% netctl-auto 0 swapper/3
1.17% swapper 233 irq/33-iwlwifi
Note that pid 0 exists for each cpu so have comm of 'swapper/N'.
Signed-off-by: Namhyung Kim <namhyung@kernel.org>
Acked-by: Jiri Olsa <jolsa@kernel.org>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Andi Kleen <andi@firstfloor.org>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Steven Rostedt <rostedt@goodmis.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1450804030-29193-6-git-send-email-namhyung@kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-12-22 17:07:02 +00:00
|
|
|
return 0;
|
|
|
|
|
2009-09-24 16:02:49 +00:00
|
|
|
return -ESRCH;
|
|
|
|
}
|
2009-12-14 22:09:29 +00:00
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
static int setup_sort_list(struct perf_hpp_list *list, char *str,
|
|
|
|
struct perf_evlist *evlist)
|
2016-01-18 09:24:10 +00:00
|
|
|
{
|
|
|
|
char *tmp, *tok;
|
|
|
|
int ret = 0;
|
2016-03-07 19:44:43 +00:00
|
|
|
int level = 0;
|
2016-03-07 19:44:47 +00:00
|
|
|
int next_level = 1;
|
|
|
|
bool in_group = false;
|
|
|
|
|
|
|
|
do {
|
|
|
|
tok = str;
|
|
|
|
tmp = strpbrk(str, "{}, ");
|
|
|
|
if (tmp) {
|
|
|
|
if (in_group)
|
|
|
|
next_level = level;
|
|
|
|
else
|
|
|
|
next_level = level + 1;
|
|
|
|
|
|
|
|
if (*tmp == '{')
|
|
|
|
in_group = true;
|
|
|
|
else if (*tmp == '}')
|
|
|
|
in_group = false;
|
|
|
|
|
|
|
|
*tmp = '\0';
|
|
|
|
str = tmp + 1;
|
|
|
|
}
|
2016-01-18 09:24:10 +00:00
|
|
|
|
2016-03-07 19:44:47 +00:00
|
|
|
if (*tok) {
|
2016-03-09 10:04:17 +00:00
|
|
|
ret = sort_dimension__add(list, tok, evlist, level);
|
2016-03-07 19:44:47 +00:00
|
|
|
if (ret == -EINVAL) {
|
2016-07-15 13:08:43 +00:00
|
|
|
if (!cacheline_size && !strncasecmp(tok, "dcacheline", strlen(tok)))
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("The \"dcacheline\" --sort key needs to know the cacheline size and it couldn't be determined on this system");
|
2016-07-15 13:08:43 +00:00
|
|
|
else
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("Invalid --sort key: `%s'", tok);
|
2016-03-07 19:44:47 +00:00
|
|
|
break;
|
|
|
|
} else if (ret == -ESRCH) {
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("Unknown --sort key: `%s'", tok);
|
2016-03-07 19:44:47 +00:00
|
|
|
break;
|
|
|
|
}
|
2016-01-18 09:24:10 +00:00
|
|
|
}
|
2016-03-07 19:44:47 +00:00
|
|
|
|
|
|
|
level = next_level;
|
|
|
|
} while (tmp);
|
2016-01-18 09:24:10 +00:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:10 +00:00
|
|
|
static const char *get_default_sort_order(struct perf_evlist *evlist)
|
2014-03-18 02:31:39 +00:00
|
|
|
{
|
|
|
|
const char *default_sort_orders[] = {
|
|
|
|
default_sort_order,
|
|
|
|
default_branch_sort_order,
|
|
|
|
default_mem_sort_order,
|
|
|
|
default_top_sort_order,
|
|
|
|
default_diff_sort_order,
|
2015-12-22 17:07:10 +00:00
|
|
|
default_tracepoint_sort_order,
|
2014-03-18 02:31:39 +00:00
|
|
|
};
|
2015-12-22 17:07:10 +00:00
|
|
|
bool use_trace = true;
|
|
|
|
struct perf_evsel *evsel;
|
2014-03-18 02:31:39 +00:00
|
|
|
|
|
|
|
BUG_ON(sort__mode >= ARRAY_SIZE(default_sort_orders));
|
|
|
|
|
2017-07-21 05:11:57 +00:00
|
|
|
if (evlist == NULL || perf_evlist__empty(evlist))
|
2015-12-22 17:07:10 +00:00
|
|
|
goto out_no_evlist;
|
|
|
|
|
2016-06-23 14:26:15 +00:00
|
|
|
evlist__for_each_entry(evlist, evsel) {
|
2015-12-22 17:07:10 +00:00
|
|
|
if (evsel->attr.type != PERF_TYPE_TRACEPOINT) {
|
|
|
|
use_trace = false;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
if (use_trace) {
|
|
|
|
sort__mode = SORT_MODE__TRACEPOINT;
|
|
|
|
if (symbol_conf.raw_trace)
|
|
|
|
return "trace_fields";
|
|
|
|
}
|
|
|
|
out_no_evlist:
|
2014-03-18 02:31:39 +00:00
|
|
|
return default_sort_orders[sort__mode];
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:10 +00:00
|
|
|
static int setup_sort_order(struct perf_evlist *evlist)
|
2014-08-23 12:59:48 +00:00
|
|
|
{
|
|
|
|
char *new_sort_order;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Append '+'-prefixed sort order to the default sort
|
|
|
|
* order string.
|
|
|
|
*/
|
|
|
|
if (!sort_order || is_strict_order(sort_order))
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
if (sort_order[1] == '\0') {
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("Invalid --sort key: `+'");
|
2014-08-23 12:59:48 +00:00
|
|
|
return -EINVAL;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* We allocate new sort_order string, but we never free it,
|
|
|
|
* because it's checked over the rest of the code.
|
|
|
|
*/
|
|
|
|
if (asprintf(&new_sort_order, "%s,%s",
|
2015-12-22 17:07:10 +00:00
|
|
|
get_default_sort_order(evlist), sort_order + 1) < 0) {
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("Not enough memory to set up --sort");
|
2014-08-23 12:59:48 +00:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
|
|
|
sort_order = new_sort_order;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2016-01-07 09:14:08 +00:00
|
|
|
/*
|
|
|
|
* Adds 'pre,' prefix into 'str' is 'pre' is
|
|
|
|
* not already part of 'str'.
|
|
|
|
*/
|
|
|
|
static char *prefix_if_not_in(const char *pre, char *str)
|
|
|
|
{
|
|
|
|
char *n;
|
|
|
|
|
|
|
|
if (!str || strstr(str, pre))
|
|
|
|
return str;
|
|
|
|
|
|
|
|
if (asprintf(&n, "%s,%s", pre, str) < 0)
|
|
|
|
return NULL;
|
|
|
|
|
|
|
|
free(str);
|
|
|
|
return n;
|
|
|
|
}
|
|
|
|
|
|
|
|
static char *setup_overhead(char *keys)
|
|
|
|
{
|
2016-05-10 14:26:24 +00:00
|
|
|
if (sort__mode == SORT_MODE__DIFF)
|
|
|
|
return keys;
|
|
|
|
|
2016-01-07 09:14:08 +00:00
|
|
|
keys = prefix_if_not_in("overhead", keys);
|
|
|
|
|
|
|
|
if (symbol_conf.cumulate_callchain)
|
|
|
|
keys = prefix_if_not_in("overhead_children", keys);
|
|
|
|
|
|
|
|
return keys;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:01 +00:00
|
|
|
static int __setup_sorting(struct perf_evlist *evlist)
|
2009-12-14 22:09:29 +00:00
|
|
|
{
|
2016-01-18 09:24:10 +00:00
|
|
|
char *str;
|
2014-08-23 12:59:48 +00:00
|
|
|
const char *sort_keys;
|
2013-02-06 05:57:16 +00:00
|
|
|
int ret = 0;
|
2009-12-14 22:09:29 +00:00
|
|
|
|
2015-12-22 17:07:10 +00:00
|
|
|
ret = setup_sort_order(evlist);
|
2014-08-23 12:59:48 +00:00
|
|
|
if (ret)
|
|
|
|
return ret;
|
|
|
|
|
|
|
|
sort_keys = sort_order;
|
2014-03-04 01:46:34 +00:00
|
|
|
if (sort_keys == NULL) {
|
2014-08-22 13:58:38 +00:00
|
|
|
if (is_strict_order(field_order)) {
|
2014-03-04 01:46:34 +00:00
|
|
|
/*
|
|
|
|
* If user specified field order but no sort order,
|
|
|
|
* we'll honor it and not add default sort orders.
|
|
|
|
*/
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2015-12-22 17:07:10 +00:00
|
|
|
sort_keys = get_default_sort_order(evlist);
|
2014-03-04 01:46:34 +00:00
|
|
|
}
|
2014-03-18 02:31:39 +00:00
|
|
|
|
|
|
|
str = strdup(sort_keys);
|
2013-02-06 05:57:17 +00:00
|
|
|
if (str == NULL) {
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("Not enough memory to setup sort keys");
|
2013-02-06 05:57:17 +00:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
2016-01-07 09:14:08 +00:00
|
|
|
/*
|
|
|
|
* Prepend overhead fields for backward compatibility.
|
|
|
|
*/
|
|
|
|
if (!is_strict_order(field_order)) {
|
|
|
|
str = setup_overhead(str);
|
|
|
|
if (str == NULL) {
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("Not enough memory to setup overhead keys");
|
2016-01-07 09:14:08 +00:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2016-03-09 10:04:17 +00:00
|
|
|
ret = setup_sort_list(&perf_hpp_list, str, evlist);
|
2009-12-14 22:09:29 +00:00
|
|
|
|
|
|
|
free(str);
|
2013-02-06 05:57:16 +00:00
|
|
|
return ret;
|
2009-12-14 22:09:29 +00:00
|
|
|
}
|
perf diff: Use perf_session__fprintf_hists just like 'perf record'
That means that almost everything you can do with 'perf report'
can be done with 'perf diff', for instance:
$ perf record -f find / > /dev/null
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.062 MB perf.data (~2699
samples) ] $ perf record -f find / > /dev/null
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.062 MB perf.data (~2687
samples) ] perf diff | head -8
9.02% +1.00% find libc-2.10.1.so [.] _IO_vfprintf_internal
2.91% -1.00% find [kernel] [k] __kmalloc
2.85% -1.00% find [kernel] [k] ext4_htree_store_dirent
1.99% -1.00% find [kernel] [k] _atomic_dec_and_lock
2.44% find [kernel] [k] half_md4_transform
$
So if you want to zoom into libc:
$ perf diff --dsos libc-2.10.1.so | head -8
37.34% find [.] _IO_vfprintf_internal
10.34% find [.] __GI_memmove
8.25% +2.00% find [.] _int_malloc
5.07% -1.00% find [.] __GI_mempcpy
7.62% +2.00% find [.] _int_free
$
And if there were multiple commands using libc, it is also
possible to aggregate them all by using --sort symbol:
$ perf diff --dsos libc-2.10.1.so --sort symbol | head -8
37.34% [.] _IO_vfprintf_internal
10.34% [.] __GI_memmove
8.25% +2.00% [.] _int_malloc
5.07% -1.00% [.] __GI_mempcpy
7.62% +2.00% [.] _int_free
$
The displacement column now is off by default, to use it:
perf diff -m --dsos libc-2.10.1.so --sort symbol | head -8
37.34% [.] _IO_vfprintf_internal
10.34% [.] __GI_memmove
8.25% +2.00% [.] _int_malloc
5.07% -1.00% +2 [.] __GI_mempcpy
7.62% +2.00% -1 [.] _int_free
$
Using -t/--field-separator can be used for scripting:
$ perf diff -t, -m --dsos libc-2.10.1.so --sort symbol | head -8
37.34, , ,[.] _IO_vfprintf_internal
10.34, , ,[.] __GI_memmove
8.25,+2.00%, ,[.] _int_malloc
5.07,-1.00%, +2,[.] __GI_mempcpy
7.62,+2.00%, -1,[.] _int_free
6.99,+1.00%, -1,[.] _IO_new_file_xsputn
1.89,-2.00%, +4,[.] __readdir64
$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1260978567-550-1-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 15:49:27 +00:00
|
|
|
|
perf tools: Move elide bool into perf_hpp_fmt struct
After output/sort fields refactoring, it's expensive
to check the elide bool in its current location inside
the 'struct sort_entry'.
The perf_hpp__should_skip function gets highly noticable in
workloads with high number of output/sort fields, like for:
$ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio
Performance report:
9.70% perf [.] perf_hpp__should_skip
Moving the elide bool into the 'struct perf_hpp_fmt', which
makes the perf_hpp__should_skip just single struct read.
Got speedup of around 22% for my test perf.data workload.
The change should not harm any other workload types.
Performance counter stats for (10 runs):
before:
358,319,732,626 cycles ( +- 0.55% )
467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% )
150.943975206 seconds time elapsed ( +- 0.62% )
now:
278,785,972,990 cycles ( +- 0.12% )
370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% )
116.416670507 seconds time elapsed ( +- 0.31% )
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20140601142622.GA9131@krava.brq.redhat.com
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
2014-05-23 15:15:47 +00:00
|
|
|
void perf_hpp__set_elide(int idx, bool elide)
|
2014-03-18 04:00:59 +00:00
|
|
|
{
|
perf tools: Move elide bool into perf_hpp_fmt struct
After output/sort fields refactoring, it's expensive
to check the elide bool in its current location inside
the 'struct sort_entry'.
The perf_hpp__should_skip function gets highly noticable in
workloads with high number of output/sort fields, like for:
$ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio
Performance report:
9.70% perf [.] perf_hpp__should_skip
Moving the elide bool into the 'struct perf_hpp_fmt', which
makes the perf_hpp__should_skip just single struct read.
Got speedup of around 22% for my test perf.data workload.
The change should not harm any other workload types.
Performance counter stats for (10 runs):
before:
358,319,732,626 cycles ( +- 0.55% )
467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% )
150.943975206 seconds time elapsed ( +- 0.62% )
now:
278,785,972,990 cycles ( +- 0.12% )
370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% )
116.416670507 seconds time elapsed ( +- 0.31% )
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20140601142622.GA9131@krava.brq.redhat.com
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
2014-05-23 15:15:47 +00:00
|
|
|
struct perf_hpp_fmt *fmt;
|
|
|
|
struct hpp_sort_entry *hse;
|
2014-03-18 04:00:59 +00:00
|
|
|
|
2016-01-18 09:24:17 +00:00
|
|
|
perf_hpp_list__for_each_format(&perf_hpp_list, fmt) {
|
perf tools: Move elide bool into perf_hpp_fmt struct
After output/sort fields refactoring, it's expensive
to check the elide bool in its current location inside
the 'struct sort_entry'.
The perf_hpp__should_skip function gets highly noticable in
workloads with high number of output/sort fields, like for:
$ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio
Performance report:
9.70% perf [.] perf_hpp__should_skip
Moving the elide bool into the 'struct perf_hpp_fmt', which
makes the perf_hpp__should_skip just single struct read.
Got speedup of around 22% for my test perf.data workload.
The change should not harm any other workload types.
Performance counter stats for (10 runs):
before:
358,319,732,626 cycles ( +- 0.55% )
467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% )
150.943975206 seconds time elapsed ( +- 0.62% )
now:
278,785,972,990 cycles ( +- 0.12% )
370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% )
116.416670507 seconds time elapsed ( +- 0.31% )
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20140601142622.GA9131@krava.brq.redhat.com
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
2014-05-23 15:15:47 +00:00
|
|
|
if (!perf_hpp__is_sort_entry(fmt))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
|
|
|
if (hse->se->se_width_idx == idx) {
|
|
|
|
fmt->elide = elide;
|
|
|
|
break;
|
|
|
|
}
|
2014-03-18 04:00:59 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
perf tools: Move elide bool into perf_hpp_fmt struct
After output/sort fields refactoring, it's expensive
to check the elide bool in its current location inside
the 'struct sort_entry'.
The perf_hpp__should_skip function gets highly noticable in
workloads with high number of output/sort fields, like for:
$ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio
Performance report:
9.70% perf [.] perf_hpp__should_skip
Moving the elide bool into the 'struct perf_hpp_fmt', which
makes the perf_hpp__should_skip just single struct read.
Got speedup of around 22% for my test perf.data workload.
The change should not harm any other workload types.
Performance counter stats for (10 runs):
before:
358,319,732,626 cycles ( +- 0.55% )
467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% )
150.943975206 seconds time elapsed ( +- 0.62% )
now:
278,785,972,990 cycles ( +- 0.12% )
370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% )
116.416670507 seconds time elapsed ( +- 0.31% )
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20140601142622.GA9131@krava.brq.redhat.com
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
2014-05-23 15:15:47 +00:00
|
|
|
static bool __get_elide(struct strlist *list, const char *list_name, FILE *fp)
|
perf diff: Use perf_session__fprintf_hists just like 'perf record'
That means that almost everything you can do with 'perf report'
can be done with 'perf diff', for instance:
$ perf record -f find / > /dev/null
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.062 MB perf.data (~2699
samples) ] $ perf record -f find / > /dev/null
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.062 MB perf.data (~2687
samples) ] perf diff | head -8
9.02% +1.00% find libc-2.10.1.so [.] _IO_vfprintf_internal
2.91% -1.00% find [kernel] [k] __kmalloc
2.85% -1.00% find [kernel] [k] ext4_htree_store_dirent
1.99% -1.00% find [kernel] [k] _atomic_dec_and_lock
2.44% find [kernel] [k] half_md4_transform
$
So if you want to zoom into libc:
$ perf diff --dsos libc-2.10.1.so | head -8
37.34% find [.] _IO_vfprintf_internal
10.34% find [.] __GI_memmove
8.25% +2.00% find [.] _int_malloc
5.07% -1.00% find [.] __GI_mempcpy
7.62% +2.00% find [.] _int_free
$
And if there were multiple commands using libc, it is also
possible to aggregate them all by using --sort symbol:
$ perf diff --dsos libc-2.10.1.so --sort symbol | head -8
37.34% [.] _IO_vfprintf_internal
10.34% [.] __GI_memmove
8.25% +2.00% [.] _int_malloc
5.07% -1.00% [.] __GI_mempcpy
7.62% +2.00% [.] _int_free
$
The displacement column now is off by default, to use it:
perf diff -m --dsos libc-2.10.1.so --sort symbol | head -8
37.34% [.] _IO_vfprintf_internal
10.34% [.] __GI_memmove
8.25% +2.00% [.] _int_malloc
5.07% -1.00% +2 [.] __GI_mempcpy
7.62% +2.00% -1 [.] _int_free
$
Using -t/--field-separator can be used for scripting:
$ perf diff -t, -m --dsos libc-2.10.1.so --sort symbol | head -8
37.34, , ,[.] _IO_vfprintf_internal
10.34, , ,[.] __GI_memmove
8.25,+2.00%, ,[.] _int_malloc
5.07,-1.00%, +2,[.] __GI_mempcpy
7.62,+2.00%, -1,[.] _int_free
6.99,+1.00%, -1,[.] _IO_new_file_xsputn
1.89,-2.00%, +4,[.] __readdir64
$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1260978567-550-1-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 15:49:27 +00:00
|
|
|
{
|
|
|
|
if (list && strlist__nr_entries(list) == 1) {
|
|
|
|
if (fp != NULL)
|
|
|
|
fprintf(fp, "# %s: %s\n", list_name,
|
|
|
|
strlist__entry(list, 0)->s);
|
perf tools: Move elide bool into perf_hpp_fmt struct
After output/sort fields refactoring, it's expensive
to check the elide bool in its current location inside
the 'struct sort_entry'.
The perf_hpp__should_skip function gets highly noticable in
workloads with high number of output/sort fields, like for:
$ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio
Performance report:
9.70% perf [.] perf_hpp__should_skip
Moving the elide bool into the 'struct perf_hpp_fmt', which
makes the perf_hpp__should_skip just single struct read.
Got speedup of around 22% for my test perf.data workload.
The change should not harm any other workload types.
Performance counter stats for (10 runs):
before:
358,319,732,626 cycles ( +- 0.55% )
467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% )
150.943975206 seconds time elapsed ( +- 0.62% )
now:
278,785,972,990 cycles ( +- 0.12% )
370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% )
116.416670507 seconds time elapsed ( +- 0.31% )
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20140601142622.GA9131@krava.brq.redhat.com
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
2014-05-23 15:15:47 +00:00
|
|
|
return true;
|
|
|
|
}
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool get_elide(int idx, FILE *output)
|
|
|
|
{
|
|
|
|
switch (idx) {
|
|
|
|
case HISTC_SYMBOL:
|
|
|
|
return __get_elide(symbol_conf.sym_list, "symbol", output);
|
|
|
|
case HISTC_DSO:
|
|
|
|
return __get_elide(symbol_conf.dso_list, "dso", output);
|
|
|
|
case HISTC_COMM:
|
|
|
|
return __get_elide(symbol_conf.comm_list, "comm", output);
|
|
|
|
default:
|
|
|
|
break;
|
perf diff: Use perf_session__fprintf_hists just like 'perf record'
That means that almost everything you can do with 'perf report'
can be done with 'perf diff', for instance:
$ perf record -f find / > /dev/null
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.062 MB perf.data (~2699
samples) ] $ perf record -f find / > /dev/null
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.062 MB perf.data (~2687
samples) ] perf diff | head -8
9.02% +1.00% find libc-2.10.1.so [.] _IO_vfprintf_internal
2.91% -1.00% find [kernel] [k] __kmalloc
2.85% -1.00% find [kernel] [k] ext4_htree_store_dirent
1.99% -1.00% find [kernel] [k] _atomic_dec_and_lock
2.44% find [kernel] [k] half_md4_transform
$
So if you want to zoom into libc:
$ perf diff --dsos libc-2.10.1.so | head -8
37.34% find [.] _IO_vfprintf_internal
10.34% find [.] __GI_memmove
8.25% +2.00% find [.] _int_malloc
5.07% -1.00% find [.] __GI_mempcpy
7.62% +2.00% find [.] _int_free
$
And if there were multiple commands using libc, it is also
possible to aggregate them all by using --sort symbol:
$ perf diff --dsos libc-2.10.1.so --sort symbol | head -8
37.34% [.] _IO_vfprintf_internal
10.34% [.] __GI_memmove
8.25% +2.00% [.] _int_malloc
5.07% -1.00% [.] __GI_mempcpy
7.62% +2.00% [.] _int_free
$
The displacement column now is off by default, to use it:
perf diff -m --dsos libc-2.10.1.so --sort symbol | head -8
37.34% [.] _IO_vfprintf_internal
10.34% [.] __GI_memmove
8.25% +2.00% [.] _int_malloc
5.07% -1.00% +2 [.] __GI_mempcpy
7.62% +2.00% -1 [.] _int_free
$
Using -t/--field-separator can be used for scripting:
$ perf diff -t, -m --dsos libc-2.10.1.so --sort symbol | head -8
37.34, , ,[.] _IO_vfprintf_internal
10.34, , ,[.] __GI_memmove
8.25,+2.00%, ,[.] _int_malloc
5.07,-1.00%, +2,[.] __GI_mempcpy
7.62,+2.00%, -1,[.] _int_free
6.99,+1.00%, -1,[.] _IO_new_file_xsputn
1.89,-2.00%, +4,[.] __readdir64
$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1260978567-550-1-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 15:49:27 +00:00
|
|
|
}
|
perf tools: Move elide bool into perf_hpp_fmt struct
After output/sort fields refactoring, it's expensive
to check the elide bool in its current location inside
the 'struct sort_entry'.
The perf_hpp__should_skip function gets highly noticable in
workloads with high number of output/sort fields, like for:
$ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio
Performance report:
9.70% perf [.] perf_hpp__should_skip
Moving the elide bool into the 'struct perf_hpp_fmt', which
makes the perf_hpp__should_skip just single struct read.
Got speedup of around 22% for my test perf.data workload.
The change should not harm any other workload types.
Performance counter stats for (10 runs):
before:
358,319,732,626 cycles ( +- 0.55% )
467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% )
150.943975206 seconds time elapsed ( +- 0.62% )
now:
278,785,972,990 cycles ( +- 0.12% )
370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% )
116.416670507 seconds time elapsed ( +- 0.31% )
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20140601142622.GA9131@krava.brq.redhat.com
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
2014-05-23 15:15:47 +00:00
|
|
|
|
|
|
|
if (sort__mode != SORT_MODE__BRANCH)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
switch (idx) {
|
|
|
|
case HISTC_SYMBOL_FROM:
|
|
|
|
return __get_elide(symbol_conf.sym_from_list, "sym_from", output);
|
|
|
|
case HISTC_SYMBOL_TO:
|
|
|
|
return __get_elide(symbol_conf.sym_to_list, "sym_to", output);
|
|
|
|
case HISTC_DSO_FROM:
|
|
|
|
return __get_elide(symbol_conf.dso_from_list, "dso_from", output);
|
|
|
|
case HISTC_DSO_TO:
|
|
|
|
return __get_elide(symbol_conf.dso_to_list, "dso_to", output);
|
|
|
|
default:
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
|
|
|
return false;
|
perf diff: Use perf_session__fprintf_hists just like 'perf record'
That means that almost everything you can do with 'perf report'
can be done with 'perf diff', for instance:
$ perf record -f find / > /dev/null
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.062 MB perf.data (~2699
samples) ] $ perf record -f find / > /dev/null
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.062 MB perf.data (~2687
samples) ] perf diff | head -8
9.02% +1.00% find libc-2.10.1.so [.] _IO_vfprintf_internal
2.91% -1.00% find [kernel] [k] __kmalloc
2.85% -1.00% find [kernel] [k] ext4_htree_store_dirent
1.99% -1.00% find [kernel] [k] _atomic_dec_and_lock
2.44% find [kernel] [k] half_md4_transform
$
So if you want to zoom into libc:
$ perf diff --dsos libc-2.10.1.so | head -8
37.34% find [.] _IO_vfprintf_internal
10.34% find [.] __GI_memmove
8.25% +2.00% find [.] _int_malloc
5.07% -1.00% find [.] __GI_mempcpy
7.62% +2.00% find [.] _int_free
$
And if there were multiple commands using libc, it is also
possible to aggregate them all by using --sort symbol:
$ perf diff --dsos libc-2.10.1.so --sort symbol | head -8
37.34% [.] _IO_vfprintf_internal
10.34% [.] __GI_memmove
8.25% +2.00% [.] _int_malloc
5.07% -1.00% [.] __GI_mempcpy
7.62% +2.00% [.] _int_free
$
The displacement column now is off by default, to use it:
perf diff -m --dsos libc-2.10.1.so --sort symbol | head -8
37.34% [.] _IO_vfprintf_internal
10.34% [.] __GI_memmove
8.25% +2.00% [.] _int_malloc
5.07% -1.00% +2 [.] __GI_mempcpy
7.62% +2.00% -1 [.] _int_free
$
Using -t/--field-separator can be used for scripting:
$ perf diff -t, -m --dsos libc-2.10.1.so --sort symbol | head -8
37.34, , ,[.] _IO_vfprintf_internal
10.34, , ,[.] __GI_memmove
8.25,+2.00%, ,[.] _int_malloc
5.07,-1.00%, +2,[.] __GI_mempcpy
7.62,+2.00%, -1,[.] _int_free
6.99,+1.00%, -1,[.] _IO_new_file_xsputn
1.89,-2.00%, +4,[.] __readdir64
$
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Paul Mackerras <paulus@samba.org>
LKML-Reference: <1260978567-550-1-git-send-email-acme@infradead.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
2009-12-16 15:49:27 +00:00
|
|
|
}
|
2013-04-03 12:26:19 +00:00
|
|
|
|
|
|
|
void sort__setup_elide(FILE *output)
|
|
|
|
{
|
2014-05-19 05:19:30 +00:00
|
|
|
struct perf_hpp_fmt *fmt;
|
|
|
|
struct hpp_sort_entry *hse;
|
2013-11-08 08:53:42 +00:00
|
|
|
|
2016-01-18 09:24:17 +00:00
|
|
|
perf_hpp_list__for_each_format(&perf_hpp_list, fmt) {
|
perf tools: Move elide bool into perf_hpp_fmt struct
After output/sort fields refactoring, it's expensive
to check the elide bool in its current location inside
the 'struct sort_entry'.
The perf_hpp__should_skip function gets highly noticable in
workloads with high number of output/sort fields, like for:
$ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio
Performance report:
9.70% perf [.] perf_hpp__should_skip
Moving the elide bool into the 'struct perf_hpp_fmt', which
makes the perf_hpp__should_skip just single struct read.
Got speedup of around 22% for my test perf.data workload.
The change should not harm any other workload types.
Performance counter stats for (10 runs):
before:
358,319,732,626 cycles ( +- 0.55% )
467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% )
150.943975206 seconds time elapsed ( +- 0.62% )
now:
278,785,972,990 cycles ( +- 0.12% )
370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% )
116.416670507 seconds time elapsed ( +- 0.31% )
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20140601142622.GA9131@krava.brq.redhat.com
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
2014-05-23 15:15:47 +00:00
|
|
|
if (!perf_hpp__is_sort_entry(fmt))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
hse = container_of(fmt, struct hpp_sort_entry, hpp);
|
|
|
|
fmt->elide = get_elide(hse->se->se_width_idx, output);
|
2013-04-03 12:26:19 +00:00
|
|
|
}
|
|
|
|
|
2013-11-08 08:53:42 +00:00
|
|
|
/*
|
|
|
|
* It makes no sense to elide all of sort entries.
|
|
|
|
* Just revert them to show up again.
|
|
|
|
*/
|
2016-01-18 09:24:17 +00:00
|
|
|
perf_hpp_list__for_each_format(&perf_hpp_list, fmt) {
|
2014-05-19 05:19:30 +00:00
|
|
|
if (!perf_hpp__is_sort_entry(fmt))
|
|
|
|
continue;
|
|
|
|
|
perf tools: Move elide bool into perf_hpp_fmt struct
After output/sort fields refactoring, it's expensive
to check the elide bool in its current location inside
the 'struct sort_entry'.
The perf_hpp__should_skip function gets highly noticable in
workloads with high number of output/sort fields, like for:
$ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio
Performance report:
9.70% perf [.] perf_hpp__should_skip
Moving the elide bool into the 'struct perf_hpp_fmt', which
makes the perf_hpp__should_skip just single struct read.
Got speedup of around 22% for my test perf.data workload.
The change should not harm any other workload types.
Performance counter stats for (10 runs):
before:
358,319,732,626 cycles ( +- 0.55% )
467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% )
150.943975206 seconds time elapsed ( +- 0.62% )
now:
278,785,972,990 cycles ( +- 0.12% )
370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% )
116.416670507 seconds time elapsed ( +- 0.31% )
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20140601142622.GA9131@krava.brq.redhat.com
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
2014-05-23 15:15:47 +00:00
|
|
|
if (!fmt->elide)
|
2013-11-08 08:53:42 +00:00
|
|
|
return;
|
|
|
|
}
|
|
|
|
|
2016-01-18 09:24:17 +00:00
|
|
|
perf_hpp_list__for_each_format(&perf_hpp_list, fmt) {
|
2014-05-19 05:19:30 +00:00
|
|
|
if (!perf_hpp__is_sort_entry(fmt))
|
|
|
|
continue;
|
|
|
|
|
perf tools: Move elide bool into perf_hpp_fmt struct
After output/sort fields refactoring, it's expensive
to check the elide bool in its current location inside
the 'struct sort_entry'.
The perf_hpp__should_skip function gets highly noticable in
workloads with high number of output/sort fields, like for:
$ perf report -i perf-test.data -F overhead,sample,period,comm,pid,dso,symbol,cpu --stdio
Performance report:
9.70% perf [.] perf_hpp__should_skip
Moving the elide bool into the 'struct perf_hpp_fmt', which
makes the perf_hpp__should_skip just single struct read.
Got speedup of around 22% for my test perf.data workload.
The change should not harm any other workload types.
Performance counter stats for (10 runs):
before:
358,319,732,626 cycles ( +- 0.55% )
467,129,581,515 instructions # 1.30 insns per cycle ( +- 0.00% )
150.943975206 seconds time elapsed ( +- 0.62% )
now:
278,785,972,990 cycles ( +- 0.12% )
370,146,797,640 instructions # 1.33 insns per cycle ( +- 0.00% )
116.416670507 seconds time elapsed ( +- 0.31% )
Acked-by: Namhyung Kim <namhyung@kernel.org>
Cc: Arnaldo Carvalho de Melo <acme@kernel.org>
Cc: Corey Ashford <cjashfor@linux.vnet.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/r/20140601142622.GA9131@krava.brq.redhat.com
Signed-off-by: Jiri Olsa <jolsa@kernel.org>
2014-05-23 15:15:47 +00:00
|
|
|
fmt->elide = false;
|
2014-05-19 05:19:30 +00:00
|
|
|
}
|
2013-04-03 12:26:19 +00:00
|
|
|
}
|
2014-03-04 01:46:34 +00:00
|
|
|
|
2016-09-22 15:36:33 +00:00
|
|
|
int output_field_add(struct perf_hpp_list *list, char *tok)
|
2014-03-04 01:46:34 +00:00
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(common_sort_dimensions); i++) {
|
|
|
|
struct sort_dimension *sd = &common_sort_dimensions[i];
|
|
|
|
|
|
|
|
if (strncasecmp(tok, sd->name, strlen(tok)))
|
|
|
|
continue;
|
|
|
|
|
2016-01-18 09:24:16 +00:00
|
|
|
return __sort_dimension__add_output(list, sd);
|
2014-03-04 01:46:34 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(hpp_sort_dimensions); i++) {
|
|
|
|
struct hpp_dimension *hd = &hpp_sort_dimensions[i];
|
|
|
|
|
|
|
|
if (strncasecmp(tok, hd->name, strlen(tok)))
|
|
|
|
continue;
|
|
|
|
|
2016-01-18 09:24:16 +00:00
|
|
|
return __hpp_dimension__add_output(list, hd);
|
2014-03-04 01:46:34 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(bstack_sort_dimensions); i++) {
|
|
|
|
struct sort_dimension *sd = &bstack_sort_dimensions[i];
|
|
|
|
|
|
|
|
if (strncasecmp(tok, sd->name, strlen(tok)))
|
|
|
|
continue;
|
|
|
|
|
2016-01-18 09:24:16 +00:00
|
|
|
return __sort_dimension__add_output(list, sd);
|
2014-03-04 01:46:34 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(memory_sort_dimensions); i++) {
|
|
|
|
struct sort_dimension *sd = &memory_sort_dimensions[i];
|
|
|
|
|
|
|
|
if (strncasecmp(tok, sd->name, strlen(tok)))
|
|
|
|
continue;
|
|
|
|
|
2016-01-18 09:24:16 +00:00
|
|
|
return __sort_dimension__add_output(list, sd);
|
2014-03-04 01:46:34 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
return -ESRCH;
|
|
|
|
}
|
|
|
|
|
2016-01-18 09:24:16 +00:00
|
|
|
static int setup_output_list(struct perf_hpp_list *list, char *str)
|
2016-01-18 09:24:11 +00:00
|
|
|
{
|
|
|
|
char *tmp, *tok;
|
|
|
|
int ret = 0;
|
|
|
|
|
|
|
|
for (tok = strtok_r(str, ", ", &tmp);
|
|
|
|
tok; tok = strtok_r(NULL, ", ", &tmp)) {
|
2016-01-18 09:24:16 +00:00
|
|
|
ret = output_field_add(list, tok);
|
2016-01-18 09:24:11 +00:00
|
|
|
if (ret == -EINVAL) {
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("Invalid --fields key: `%s'", tok);
|
2016-01-18 09:24:11 +00:00
|
|
|
break;
|
|
|
|
} else if (ret == -ESRCH) {
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("Unknown --fields key: `%s'", tok);
|
2016-01-18 09:24:11 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2016-09-22 15:36:32 +00:00
|
|
|
void reset_dimensions(void)
|
2014-03-04 01:46:34 +00:00
|
|
|
{
|
|
|
|
unsigned int i;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(common_sort_dimensions); i++)
|
|
|
|
common_sort_dimensions[i].taken = 0;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(hpp_sort_dimensions); i++)
|
|
|
|
hpp_sort_dimensions[i].taken = 0;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(bstack_sort_dimensions); i++)
|
|
|
|
bstack_sort_dimensions[i].taken = 0;
|
|
|
|
|
|
|
|
for (i = 0; i < ARRAY_SIZE(memory_sort_dimensions); i++)
|
|
|
|
memory_sort_dimensions[i].taken = 0;
|
|
|
|
}
|
|
|
|
|
2014-08-22 13:58:38 +00:00
|
|
|
bool is_strict_order(const char *order)
|
|
|
|
{
|
|
|
|
return order && (*order != '+');
|
|
|
|
}
|
|
|
|
|
2014-03-04 01:46:34 +00:00
|
|
|
static int __setup_output_field(void)
|
|
|
|
{
|
2016-01-18 09:24:11 +00:00
|
|
|
char *str, *strp;
|
2014-08-22 13:58:38 +00:00
|
|
|
int ret = -EINVAL;
|
2014-03-04 01:46:34 +00:00
|
|
|
|
|
|
|
if (field_order == NULL)
|
|
|
|
return 0;
|
|
|
|
|
2014-08-22 13:58:38 +00:00
|
|
|
strp = str = strdup(field_order);
|
2014-03-04 01:46:34 +00:00
|
|
|
if (str == NULL) {
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("Not enough memory to setup output fields");
|
2014-03-04 01:46:34 +00:00
|
|
|
return -ENOMEM;
|
|
|
|
}
|
|
|
|
|
2014-08-22 13:58:38 +00:00
|
|
|
if (!is_strict_order(field_order))
|
|
|
|
strp++;
|
|
|
|
|
|
|
|
if (!strlen(strp)) {
|
2017-06-27 14:22:31 +00:00
|
|
|
pr_err("Invalid --fields key: `+'");
|
2014-08-22 13:58:38 +00:00
|
|
|
goto out;
|
|
|
|
}
|
|
|
|
|
2016-01-18 09:24:16 +00:00
|
|
|
ret = setup_output_list(&perf_hpp_list, strp);
|
2014-03-04 01:46:34 +00:00
|
|
|
|
2014-08-22 13:58:38 +00:00
|
|
|
out:
|
2014-03-04 01:46:34 +00:00
|
|
|
free(str);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2016-03-02 12:58:00 +00:00
|
|
|
int setup_sorting(struct perf_evlist *evlist)
|
|
|
|
{
|
|
|
|
int err;
|
|
|
|
|
|
|
|
err = __setup_sorting(evlist);
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
|
|
|
|
if (parent_pattern != default_parent_pattern) {
|
2016-03-09 10:04:17 +00:00
|
|
|
err = sort_dimension__add(&perf_hpp_list, "parent", evlist, -1);
|
2016-03-02 12:58:00 +00:00
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
}
|
|
|
|
|
2014-03-04 01:46:34 +00:00
|
|
|
reset_dimensions();
|
|
|
|
|
|
|
|
/*
|
|
|
|
* perf diff doesn't use default hpp output fields.
|
|
|
|
*/
|
|
|
|
if (sort__mode != SORT_MODE__DIFF)
|
|
|
|
perf_hpp__init();
|
|
|
|
|
|
|
|
err = __setup_output_field();
|
|
|
|
if (err < 0)
|
|
|
|
return err;
|
|
|
|
|
|
|
|
/* copy sort keys to output fields */
|
2016-01-18 09:24:21 +00:00
|
|
|
perf_hpp__setup_output_field(&perf_hpp_list);
|
2014-03-04 01:46:34 +00:00
|
|
|
/* and then copy output fields to sort keys */
|
2016-01-18 09:24:21 +00:00
|
|
|
perf_hpp__append_sort_keys(&perf_hpp_list);
|
2014-03-04 01:46:34 +00:00
|
|
|
|
2016-03-07 19:44:45 +00:00
|
|
|
/* setup hists-specific output fields */
|
|
|
|
if (perf_hpp__setup_hists_formats(&perf_hpp_list, evlist) < 0)
|
|
|
|
return -1;
|
|
|
|
|
2014-03-04 01:46:34 +00:00
|
|
|
return 0;
|
|
|
|
}
|
2014-05-07 09:42:24 +00:00
|
|
|
|
|
|
|
void reset_output_field(void)
|
|
|
|
{
|
2016-05-03 11:54:42 +00:00
|
|
|
perf_hpp_list.need_collapse = 0;
|
2016-05-03 11:54:43 +00:00
|
|
|
perf_hpp_list.parent = 0;
|
2016-05-03 11:54:44 +00:00
|
|
|
perf_hpp_list.sym = 0;
|
2016-05-03 11:54:45 +00:00
|
|
|
perf_hpp_list.dso = 0;
|
2014-05-07 09:42:24 +00:00
|
|
|
|
2014-05-23 01:59:01 +00:00
|
|
|
field_order = NULL;
|
|
|
|
sort_order = NULL;
|
|
|
|
|
2014-05-07 09:42:24 +00:00
|
|
|
reset_dimensions();
|
2016-01-18 09:24:21 +00:00
|
|
|
perf_hpp__reset_output_field(&perf_hpp_list);
|
2014-05-07 09:42:24 +00:00
|
|
|
}
|