2009-06-03 01:37:05 +04:00
/*
* perf . c
*
* Performance analysis utility .
*
* This is the main hub from which the sub - commands ( perf stat ,
* perf top , perf record , perf report , etc . ) are started .
*/
2009-04-20 17:00:56 +04:00
# include "builtin.h"
2009-06-03 01:37:05 +04:00
2015-09-09 16:37:01 +03:00
# include "util/env.h"
2015-12-15 18:39:39 +03:00
# include <subcmd/exec-cmd.h>
2016-06-23 11:55:17 +03:00
# include "util/config.h"
2009-04-27 10:02:14 +04:00
# include "util/quote.h"
2015-12-15 18:39:39 +03:00
# include <subcmd/run-command.h>
2009-07-21 22:16:29 +04:00
# include "util/parse-events.h"
2015-12-15 18:39:39 +03:00
# include <subcmd/parse-options.h>
perf tools: Enable passing bpf object file to --event
By introducing new rules in tools/perf/util/parse-events.[ly], this
patch enables 'perf record --event bpf_file.o' to select events by an
eBPF object file. It calls parse_events_load_bpf() to load that file,
which uses bpf__prepare_load() and finally calls bpf_object__open() for
the object files.
After applying this patch, commands like:
# perf record --event foo.o sleep
become possible.
However, at this point it is unable to link any useful things onto the
evsel list because the creating of probe points and BPF program
attaching have not been implemented. Before real events are possible to
be extracted, to avoid perf report error because of empty evsel list,
this patch link a dummy evsel. The dummy event related code will be
removed when probing and extracting code is ready.
Commiter notes:
Using it:
$ ls -la foo.o
ls: cannot access foo.o: No such file or directory
$ perf record --event foo.o sleep
libbpf: failed to open foo.o: No such file or directory
event syntax error: 'foo.o'
\___ BPF object file 'foo.o' is invalid
(add -v to see detail)
Run 'perf list' for a list of valid events
Usage: perf record [<options>] [<command>]
or: perf record [<options>] -- <command> [<options>]
-e, --event <event> event selector. use 'perf list' to list available events
$
$ file /tmp/build/perf/perf.o
/tmp/build/perf/perf.o: ELF 64-bit LSB relocatable, x86-64, version 1 (SYSV), not stripped
$ perf record --event /tmp/build/perf/perf.o sleep
libbpf: /tmp/build/perf/perf.o is not an eBPF object file
event syntax error: '/tmp/build/perf/perf.o'
\___ BPF object file '/tmp/build/perf/perf.o' is invalid
(add -v to see detail)
Run 'perf list' for a list of valid events
Usage: perf record [<options>] [<command>]
or: perf record [<options>] -- <command> [<options>]
-e, --event <event> event selector. use 'perf list' to list available events
$
$ file /tmp/foo.o
/tmp/foo.o: ELF 64-bit LSB relocatable, no machine, version 1 (SYSV), not stripped
$ perf record --event /tmp/foo.o sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.013 MB perf.data ]
$ perf evlist
/tmp/foo.o
$ perf evlist -v
/tmp/foo.o: type: 1, size: 112, config: 0x9, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|PERIOD, disabled: 1, inherit: 1, mmap: 1, comm: 1, freq: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1
$
So, type 1 is PERF_TYPE_SOFTWARE, config 0x9 is PERF_COUNT_SW_DUMMY, ok.
$ perf report --stdio
Error:
The perf.data file has no samples!
# To display the perf.data header info, please use --header/--header-only options.
#
$
Signed-off-by: Wang Nan <wangnan0@huawei.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexei Starovoitov <ast@plumgrid.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: David Ahern <dsahern@gmail.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kaixu Xia <xiakaixu@huawei.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Zefan Li <lizefan@huawei.com>
Cc: pi3orama@163.com
Link: http://lkml.kernel.org/r/1444826502-49291-4-git-send-email-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-10-14 15:41:14 +03:00
# include "util/bpf-loader.h"
2014-07-17 14:55:00 +04:00
# include "util/debug.h"
2017-04-25 21:30:47 +03:00
# include "util/event.h"
2016-04-27 16:16:24 +03:00
# include <api/fs/fs.h>
2015-09-02 10:56:34 +03:00
# include <api/fs/tracing_path.h>
2017-04-18 16:46:11 +03:00
# include <errno.h>
2012-09-08 04:43:19 +04:00
# include <pthread.h>
2017-04-19 21:49:18 +03:00
# include <signal.h>
2016-01-07 14:41:53 +03:00
# include <stdlib.h>
# include <time.h>
2017-04-20 02:57:47 +03:00
# include <sys/types.h>
# include <sys/stat.h>
# include <unistd.h>
2017-04-17 17:39:06 +03:00
# include <linux/kernel.h>
2009-04-20 17:00:56 +04:00
const char perf_usage_string [ ] =
2014-07-18 11:11:30 +04:00
" perf [--version] [--help] [OPTIONS] COMMAND [ARGS] " ;
2009-04-20 17:00:56 +04:00
const char perf_more_info_string [ ] =
" See 'perf help COMMAND' for more information on a specific command. " ;
static int use_pager = - 1 ;
2012-10-30 07:56:02 +04:00
const char * input_name ;
perf tui: Allow disabling the TUI on a per command basis in ~/.perfconfig
Using the same scheme as for git's/perf's pager setup, i.e. if one
doesn't want to, on a newt enabled perf binary, to disable the TUI for
'perf report', its just a matter of doing:
[root@doppio linux-2.6-tip]# printf "[tui]\n\nreport = off\n" >
/root/.perfconfig
[root@doppio linux-2.6-tip]# cat /root/.perfconfig
[tui]
report = off
[root@doppio linux-2.6-tip]#
System wide settings are also possible, by editing /etc/perfconfig, etc,
i.e. the git machinery for config files applies to perf as well, so when
in doubt where to put your settings, consult the git documentation, if
it fails, please let us know.
Suggested-by: Ingo Molnar <mingo@elte.hu>
Discussed-with: Stephane Eranian <eranian@google.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-21 05:01:10 +04:00
2012-08-09 18:31:51 +04:00
struct cmd_struct {
const char * cmd ;
2017-03-27 17:47:20 +03:00
int ( * fn ) ( int , const char * * ) ;
2012-08-09 18:31:51 +04:00
int option ;
} ;
static struct cmd_struct commands [ ] = {
{ " buildid-cache " , cmd_buildid_cache , 0 } ,
{ " buildid-list " , cmd_buildid_list , 0 } ,
2015-11-17 16:53:21 +03:00
{ " config " , cmd_config , 0 } ,
2016-09-22 18:36:38 +03:00
{ " c2c " , cmd_c2c , 0 } ,
2012-08-09 18:31:51 +04:00
{ " diff " , cmd_diff , 0 } ,
{ " evlist " , cmd_evlist , 0 } ,
{ " help " , cmd_help , 0 } ,
2017-01-05 21:33:32 +03:00
{ " kallsyms " , cmd_kallsyms , 0 } ,
2012-08-09 18:31:51 +04:00
{ " list " , cmd_list , 0 } ,
{ " record " , cmd_record , 0 } ,
{ " report " , cmd_report , 0 } ,
{ " bench " , cmd_bench , 0 } ,
{ " stat " , cmd_stat , 0 } ,
{ " timechart " , cmd_timechart , 0 } ,
{ " top " , cmd_top , 0 } ,
{ " annotate " , cmd_annotate , 0 } ,
{ " version " , cmd_version , 0 } ,
{ " script " , cmd_script , 0 } ,
{ " sched " , cmd_sched , 0 } ,
2013-09-30 14:07:11 +04:00
# ifdef HAVE_LIBELF_SUPPORT
2012-08-09 18:31:51 +04:00
{ " probe " , cmd_probe , 0 } ,
2012-08-06 08:41:21 +04:00
# endif
2012-08-09 18:31:51 +04:00
{ " kmem " , cmd_kmem , 0 } ,
{ " lock " , cmd_lock , 0 } ,
{ " kvm " , cmd_kvm , 0 } ,
{ " test " , cmd_test , 0 } ,
2013-09-30 14:07:11 +04:00
# ifdef HAVE_LIBAUDIT_SUPPORT
2012-09-27 03:05:56 +04:00
{ " trace " , cmd_trace , 0 } ,
2012-09-27 15:23:38 +04:00
# endif
2012-08-09 18:31:51 +04:00
{ " inject " , cmd_inject , 0 } ,
2013-01-24 19:10:38 +04:00
{ " mem " , cmd_mem , 0 } ,
2015-02-21 01:16:59 +03:00
{ " data " , cmd_data , 0 } ,
2013-03-07 16:45:20 +04:00
{ " ftrace " , cmd_ftrace , 0 } ,
2012-08-09 18:31:51 +04:00
} ;
2009-04-20 17:00:56 +04:00
struct pager_config {
const char * cmd ;
int val ;
} ;
static int pager_command_config ( const char * var , const char * value , void * data )
{
struct pager_config * c = data ;
2017-07-20 21:27:39 +03:00
if ( strstarts ( var , " pager. " ) & & ! strcmp ( var + 6 , c - > cmd ) )
2009-04-20 17:00:56 +04:00
c - > val = perf_config_bool ( var , value ) ;
return 0 ;
}
/* returns 0 for "no pager", 1 for "use pager", and -1 for "not specified" */
2017-04-20 02:46:41 +03:00
static int check_pager_config ( const char * cmd )
2009-04-20 17:00:56 +04:00
{
2017-01-24 19:44:10 +03:00
int err ;
2009-04-20 17:00:56 +04:00
struct pager_config c ;
c . cmd = cmd ;
c . val = - 1 ;
2017-01-24 19:44:10 +03:00
err = perf_config ( pager_command_config , & c ) ;
return err ? : c . val ;
2009-04-20 17:00:56 +04:00
}
2012-11-12 06:50:17 +04:00
static int browser_command_config ( const char * var , const char * value , void * data )
perf tui: Allow disabling the TUI on a per command basis in ~/.perfconfig
Using the same scheme as for git's/perf's pager setup, i.e. if one
doesn't want to, on a newt enabled perf binary, to disable the TUI for
'perf report', its just a matter of doing:
[root@doppio linux-2.6-tip]# printf "[tui]\n\nreport = off\n" >
/root/.perfconfig
[root@doppio linux-2.6-tip]# cat /root/.perfconfig
[tui]
report = off
[root@doppio linux-2.6-tip]#
System wide settings are also possible, by editing /etc/perfconfig, etc,
i.e. the git machinery for config files applies to perf as well, so when
in doubt where to put your settings, consult the git documentation, if
it fails, please let us know.
Suggested-by: Ingo Molnar <mingo@elte.hu>
Discussed-with: Stephane Eranian <eranian@google.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-21 05:01:10 +04:00
{
struct pager_config * c = data ;
2017-07-20 21:27:39 +03:00
if ( strstarts ( var , " tui. " ) & & ! strcmp ( var + 4 , c - > cmd ) )
perf tui: Allow disabling the TUI on a per command basis in ~/.perfconfig
Using the same scheme as for git's/perf's pager setup, i.e. if one
doesn't want to, on a newt enabled perf binary, to disable the TUI for
'perf report', its just a matter of doing:
[root@doppio linux-2.6-tip]# printf "[tui]\n\nreport = off\n" >
/root/.perfconfig
[root@doppio linux-2.6-tip]# cat /root/.perfconfig
[tui]
report = off
[root@doppio linux-2.6-tip]#
System wide settings are also possible, by editing /etc/perfconfig, etc,
i.e. the git machinery for config files applies to perf as well, so when
in doubt where to put your settings, consult the git documentation, if
it fails, please let us know.
Suggested-by: Ingo Molnar <mingo@elte.hu>
Discussed-with: Stephane Eranian <eranian@google.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-21 05:01:10 +04:00
c - > val = perf_config_bool ( var , value ) ;
2017-07-20 21:27:39 +03:00
if ( strstarts ( var , " gtk. " ) & & ! strcmp ( var + 4 , c - > cmd ) )
2012-11-12 06:50:17 +04:00
c - > val = perf_config_bool ( var , value ) ? 2 : 0 ;
perf tui: Allow disabling the TUI on a per command basis in ~/.perfconfig
Using the same scheme as for git's/perf's pager setup, i.e. if one
doesn't want to, on a newt enabled perf binary, to disable the TUI for
'perf report', its just a matter of doing:
[root@doppio linux-2.6-tip]# printf "[tui]\n\nreport = off\n" >
/root/.perfconfig
[root@doppio linux-2.6-tip]# cat /root/.perfconfig
[tui]
report = off
[root@doppio linux-2.6-tip]#
System wide settings are also possible, by editing /etc/perfconfig, etc,
i.e. the git machinery for config files applies to perf as well, so when
in doubt where to put your settings, consult the git documentation, if
it fails, please let us know.
Suggested-by: Ingo Molnar <mingo@elte.hu>
Discussed-with: Stephane Eranian <eranian@google.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-21 05:01:10 +04:00
return 0 ;
}
2012-11-12 06:50:17 +04:00
/*
* returns 0 for " no tui " , 1 for " use tui " , 2 for " use gtk " ,
* and - 1 for " not specified "
*/
static int check_browser_config ( const char * cmd )
perf tui: Allow disabling the TUI on a per command basis in ~/.perfconfig
Using the same scheme as for git's/perf's pager setup, i.e. if one
doesn't want to, on a newt enabled perf binary, to disable the TUI for
'perf report', its just a matter of doing:
[root@doppio linux-2.6-tip]# printf "[tui]\n\nreport = off\n" >
/root/.perfconfig
[root@doppio linux-2.6-tip]# cat /root/.perfconfig
[tui]
report = off
[root@doppio linux-2.6-tip]#
System wide settings are also possible, by editing /etc/perfconfig, etc,
i.e. the git machinery for config files applies to perf as well, so when
in doubt where to put your settings, consult the git documentation, if
it fails, please let us know.
Suggested-by: Ingo Molnar <mingo@elte.hu>
Discussed-with: Stephane Eranian <eranian@google.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-21 05:01:10 +04:00
{
2017-01-24 19:44:10 +03:00
int err ;
perf tui: Allow disabling the TUI on a per command basis in ~/.perfconfig
Using the same scheme as for git's/perf's pager setup, i.e. if one
doesn't want to, on a newt enabled perf binary, to disable the TUI for
'perf report', its just a matter of doing:
[root@doppio linux-2.6-tip]# printf "[tui]\n\nreport = off\n" >
/root/.perfconfig
[root@doppio linux-2.6-tip]# cat /root/.perfconfig
[tui]
report = off
[root@doppio linux-2.6-tip]#
System wide settings are also possible, by editing /etc/perfconfig, etc,
i.e. the git machinery for config files applies to perf as well, so when
in doubt where to put your settings, consult the git documentation, if
it fails, please let us know.
Suggested-by: Ingo Molnar <mingo@elte.hu>
Discussed-with: Stephane Eranian <eranian@google.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-21 05:01:10 +04:00
struct pager_config c ;
c . cmd = cmd ;
c . val = - 1 ;
2017-01-24 19:44:10 +03:00
err = perf_config ( browser_command_config , & c ) ;
return err ? : c . val ;
perf tui: Allow disabling the TUI on a per command basis in ~/.perfconfig
Using the same scheme as for git's/perf's pager setup, i.e. if one
doesn't want to, on a newt enabled perf binary, to disable the TUI for
'perf report', its just a matter of doing:
[root@doppio linux-2.6-tip]# printf "[tui]\n\nreport = off\n" >
/root/.perfconfig
[root@doppio linux-2.6-tip]# cat /root/.perfconfig
[tui]
report = off
[root@doppio linux-2.6-tip]#
System wide settings are also possible, by editing /etc/perfconfig, etc,
i.e. the git machinery for config files applies to perf as well, so when
in doubt where to put your settings, consult the git documentation, if
it fails, please let us know.
Suggested-by: Ingo Molnar <mingo@elte.hu>
Discussed-with: Stephane Eranian <eranian@google.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-21 05:01:10 +04:00
}
2010-01-28 02:05:55 +03:00
static void commit_pager_choice ( void )
{
2009-04-20 17:00:56 +04:00
switch ( use_pager ) {
case 0 :
2015-12-15 18:39:35 +03:00
setenv ( PERF_PAGER_ENVIRONMENT , " cat " , 1 ) ;
2009-04-20 17:00:56 +04:00
break ;
case 1 :
/* setup_pager(); */
break ;
default :
break ;
}
}
perf tools: Fix the bash completion problem of 'perf --*'
The perf-completion.sh uses a predefined string '--help --version
--exec-path --html-path --paginate --no-pager --perf-dir --work-tree
--debugfs-dir' for the bash completion of 'perf --*', which has two
problems:
Problem 1: If the options of perf are changed (see handle_options() in
perf.c), the perf-completion.sh has to be changed at the same time. If
not, the bash completion of 'perf --*' and the options which perf
really supports will be inconsistent.
Problem 2: When typing another single character after 'perf --', e.g.
'h', and hit TAB key to get the bash completion of 'perf --h', the
character 'h' disappears at once. This is not what we want, we wish the
bash completion can return '--help --html-path' and then we can
continue to choose one.
To solve this problem, we add '--list-opts' to perf, which now supports
'perf --list-opts' directly, and its result can be used in bash
completion now.
Example:
Before this patch:
$ perf --h <-- hit TAB key after character 'h'
$ perf -- <-- 'h' disappears and no required result
After this patch:
$ perf --h <-- hit TAB key after character 'h'
--help --html-path <-- the required result
Signed-off-by: Yunlong Song <yunlong.song@huawei.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1425032491-20224-8-git-send-email-yunlong.song@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-02-27 13:21:31 +03:00
struct option options [ ] = {
OPT_ARGUMENT ( " help " , " help " ) ,
OPT_ARGUMENT ( " version " , " version " ) ,
OPT_ARGUMENT ( " exec-path " , " exec-path " ) ,
OPT_ARGUMENT ( " html-path " , " html-path " ) ,
OPT_ARGUMENT ( " paginate " , " paginate " ) ,
OPT_ARGUMENT ( " no-pager " , " no-pager " ) ,
OPT_ARGUMENT ( " debugfs-dir " , " debugfs-dir " ) ,
OPT_ARGUMENT ( " buildid-dir " , " buildid-dir " ) ,
OPT_ARGUMENT ( " list-cmds " , " list-cmds " ) ,
OPT_ARGUMENT ( " list-opts " , " list-opts " ) ,
OPT_ARGUMENT ( " debug " , " debug " ) ,
OPT_END ( )
} ;
2010-01-28 02:05:55 +03:00
static int handle_options ( const char * * * argv , int * argc , int * envchanged )
2009-04-20 17:00:56 +04:00
{
int handled = 0 ;
while ( * argc > 0 ) {
const char * cmd = ( * argv ) [ 0 ] ;
if ( cmd [ 0 ] ! = ' - ' )
break ;
/*
* For legacy reasons , the " version " and " help "
* commands can be written with " -- " prepended
* to make them look like flags .
*/
if ( ! strcmp ( cmd , " --help " ) | | ! strcmp ( cmd , " --version " ) )
break ;
2015-10-05 21:06:09 +03:00
/*
* Shortcut for ' - h ' and ' - v ' options to invoke help
* and version command .
*/
if ( ! strcmp ( cmd , " -h " ) ) {
( * argv ) [ 0 ] = " --help " ;
break ;
}
if ( ! strcmp ( cmd , " -v " ) ) {
( * argv ) [ 0 ] = " --version " ;
break ;
}
2009-04-20 17:00:56 +04:00
/*
* Check remaining flags .
*/
2017-07-20 21:27:39 +03:00
if ( strstarts ( cmd , CMD_EXEC_PATH ) ) {
2009-10-13 12:18:16 +04:00
cmd + = strlen ( CMD_EXEC_PATH ) ;
2009-04-20 17:00:56 +04:00
if ( * cmd = = ' = ' )
2015-12-15 18:39:37 +03:00
set_argv_exec_path ( cmd + 1 ) ;
2009-04-20 17:00:56 +04:00
else {
2015-12-15 18:39:37 +03:00
puts ( get_argv_exec_path ( ) ) ;
2009-04-20 17:00:56 +04:00
exit ( 0 ) ;
}
} else if ( ! strcmp ( cmd , " --html-path " ) ) {
puts ( system_path ( PERF_HTML_PATH ) ) ;
exit ( 0 ) ;
} else if ( ! strcmp ( cmd , " -p " ) | | ! strcmp ( cmd , " --paginate " ) ) {
use_pager = 1 ;
} else if ( ! strcmp ( cmd , " --no-pager " ) ) {
use_pager = 0 ;
if ( envchanged )
* envchanged = 1 ;
2009-07-21 22:16:29 +04:00
} else if ( ! strcmp ( cmd , " --debugfs-dir " ) ) {
if ( * argc < 2 ) {
fprintf ( stderr , " No directory given for --debugfs-dir. \n " ) ;
usage ( perf_usage_string ) ;
}
2015-09-02 10:56:33 +03:00
tracing_path_set ( ( * argv ) [ 1 ] ) ;
2009-07-21 22:16:29 +04:00
if ( envchanged )
* envchanged = 1 ;
( * argv ) + + ;
( * argc ) - - ;
2014-12-01 22:06:24 +03:00
} else if ( ! strcmp ( cmd , " --buildid-dir " ) ) {
if ( * argc < 2 ) {
fprintf ( stderr , " No directory given for --buildid-dir. \n " ) ;
usage ( perf_usage_string ) ;
}
set_buildid_dir ( ( * argv ) [ 1 ] ) ;
if ( envchanged )
* envchanged = 1 ;
( * argv ) + + ;
( * argc ) - - ;
2017-07-20 21:27:39 +03:00
} else if ( strstarts ( cmd , CMD_DEBUGFS_DIR ) ) {
2015-09-02 10:56:33 +03:00
tracing_path_set ( cmd + strlen ( CMD_DEBUGFS_DIR ) ) ;
2015-08-26 16:46:45 +03:00
fprintf ( stderr , " dir: %s \n " , tracing_path ) ;
2009-07-21 22:16:29 +04:00
if ( envchanged )
* envchanged = 1 ;
2012-08-09 18:31:51 +04:00
} else if ( ! strcmp ( cmd , " --list-cmds " ) ) {
unsigned int i ;
for ( i = 0 ; i < ARRAY_SIZE ( commands ) ; i + + ) {
struct cmd_struct * p = commands + i ;
printf ( " %s " , p - > cmd ) ;
}
2015-02-27 13:21:29 +03:00
putchar ( ' \n ' ) ;
2012-08-09 18:31:51 +04:00
exit ( 0 ) ;
perf tools: Fix the bash completion problem of 'perf --*'
The perf-completion.sh uses a predefined string '--help --version
--exec-path --html-path --paginate --no-pager --perf-dir --work-tree
--debugfs-dir' for the bash completion of 'perf --*', which has two
problems:
Problem 1: If the options of perf are changed (see handle_options() in
perf.c), the perf-completion.sh has to be changed at the same time. If
not, the bash completion of 'perf --*' and the options which perf
really supports will be inconsistent.
Problem 2: When typing another single character after 'perf --', e.g.
'h', and hit TAB key to get the bash completion of 'perf --h', the
character 'h' disappears at once. This is not what we want, we wish the
bash completion can return '--help --html-path' and then we can
continue to choose one.
To solve this problem, we add '--list-opts' to perf, which now supports
'perf --list-opts' directly, and its result can be used in bash
completion now.
Example:
Before this patch:
$ perf --h <-- hit TAB key after character 'h'
$ perf -- <-- 'h' disappears and no required result
After this patch:
$ perf --h <-- hit TAB key after character 'h'
--help --html-path <-- the required result
Signed-off-by: Yunlong Song <yunlong.song@huawei.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/r/1425032491-20224-8-git-send-email-yunlong.song@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-02-27 13:21:31 +03:00
} else if ( ! strcmp ( cmd , " --list-opts " ) ) {
unsigned int i ;
for ( i = 0 ; i < ARRAY_SIZE ( options ) - 1 ; i + + ) {
struct option * p = options + i ;
printf ( " --%s " , p - > long_name ) ;
}
putchar ( ' \n ' ) ;
exit ( 0 ) ;
2014-07-17 14:55:00 +04:00
} else if ( ! strcmp ( cmd , " --debug " ) ) {
if ( * argc < 2 ) {
fprintf ( stderr , " No variable specified for --debug. \n " ) ;
usage ( perf_usage_string ) ;
}
if ( perf_debug_option ( ( * argv ) [ 1 ] ) )
usage ( perf_usage_string ) ;
( * argv ) + + ;
( * argc ) - - ;
2009-04-20 17:00:56 +04:00
} else {
fprintf ( stderr , " Unknown option: %s \n " , cmd ) ;
usage ( perf_usage_string ) ;
}
( * argv ) + + ;
( * argc ) - - ;
handled + + ;
}
return handled ;
}
# define RUN_SETUP (1<<0)
# define USE_PAGER (1<<1)
static int run_builtin ( struct cmd_struct * p , int argc , const char * * argv )
{
int status ;
struct stat st ;
2014-08-14 06:22:32 +04:00
char sbuf [ STRERR_BUFSIZE ] ;
2009-04-20 17:00:56 +04:00
perf tui: Allow disabling the TUI on a per command basis in ~/.perfconfig
Using the same scheme as for git's/perf's pager setup, i.e. if one
doesn't want to, on a newt enabled perf binary, to disable the TUI for
'perf report', its just a matter of doing:
[root@doppio linux-2.6-tip]# printf "[tui]\n\nreport = off\n" >
/root/.perfconfig
[root@doppio linux-2.6-tip]# cat /root/.perfconfig
[tui]
report = off
[root@doppio linux-2.6-tip]#
System wide settings are also possible, by editing /etc/perfconfig, etc,
i.e. the git machinery for config files applies to perf as well, so when
in doubt where to put your settings, consult the git documentation, if
it fails, please let us know.
Suggested-by: Ingo Molnar <mingo@elte.hu>
Discussed-with: Stephane Eranian <eranian@google.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-21 05:01:10 +04:00
if ( use_browser = = - 1 )
2012-11-12 06:50:17 +04:00
use_browser = check_browser_config ( p - > cmd ) ;
perf tui: Allow disabling the TUI on a per command basis in ~/.perfconfig
Using the same scheme as for git's/perf's pager setup, i.e. if one
doesn't want to, on a newt enabled perf binary, to disable the TUI for
'perf report', its just a matter of doing:
[root@doppio linux-2.6-tip]# printf "[tui]\n\nreport = off\n" >
/root/.perfconfig
[root@doppio linux-2.6-tip]# cat /root/.perfconfig
[tui]
report = off
[root@doppio linux-2.6-tip]#
System wide settings are also possible, by editing /etc/perfconfig, etc,
i.e. the git machinery for config files applies to perf as well, so when
in doubt where to put your settings, consult the git documentation, if
it fails, please let us know.
Suggested-by: Ingo Molnar <mingo@elte.hu>
Discussed-with: Stephane Eranian <eranian@google.com>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-05-21 05:01:10 +04:00
2009-04-20 17:00:56 +04:00
if ( use_pager = = - 1 & & p - > option & RUN_SETUP )
use_pager = check_pager_config ( p - > cmd ) ;
if ( use_pager = = - 1 & & p - > option & USE_PAGER )
use_pager = 1 ;
commit_pager_choice ( ) ;
2015-12-08 07:21:46 +03:00
perf_env__set_cmdline ( & perf_env , argc , argv ) ;
2017-03-27 17:47:20 +03:00
status = p - > fn ( argc , argv ) ;
2016-06-23 17:14:31 +03:00
perf_config__exit ( ) ;
2010-03-22 19:10:25 +03:00
exit_browser ( status ) ;
2015-09-09 16:37:01 +03:00
perf_env__exit ( & perf_env ) ;
perf tools: Enable passing bpf object file to --event
By introducing new rules in tools/perf/util/parse-events.[ly], this
patch enables 'perf record --event bpf_file.o' to select events by an
eBPF object file. It calls parse_events_load_bpf() to load that file,
which uses bpf__prepare_load() and finally calls bpf_object__open() for
the object files.
After applying this patch, commands like:
# perf record --event foo.o sleep
become possible.
However, at this point it is unable to link any useful things onto the
evsel list because the creating of probe points and BPF program
attaching have not been implemented. Before real events are possible to
be extracted, to avoid perf report error because of empty evsel list,
this patch link a dummy evsel. The dummy event related code will be
removed when probing and extracting code is ready.
Commiter notes:
Using it:
$ ls -la foo.o
ls: cannot access foo.o: No such file or directory
$ perf record --event foo.o sleep
libbpf: failed to open foo.o: No such file or directory
event syntax error: 'foo.o'
\___ BPF object file 'foo.o' is invalid
(add -v to see detail)
Run 'perf list' for a list of valid events
Usage: perf record [<options>] [<command>]
or: perf record [<options>] -- <command> [<options>]
-e, --event <event> event selector. use 'perf list' to list available events
$
$ file /tmp/build/perf/perf.o
/tmp/build/perf/perf.o: ELF 64-bit LSB relocatable, x86-64, version 1 (SYSV), not stripped
$ perf record --event /tmp/build/perf/perf.o sleep
libbpf: /tmp/build/perf/perf.o is not an eBPF object file
event syntax error: '/tmp/build/perf/perf.o'
\___ BPF object file '/tmp/build/perf/perf.o' is invalid
(add -v to see detail)
Run 'perf list' for a list of valid events
Usage: perf record [<options>] [<command>]
or: perf record [<options>] -- <command> [<options>]
-e, --event <event> event selector. use 'perf list' to list available events
$
$ file /tmp/foo.o
/tmp/foo.o: ELF 64-bit LSB relocatable, no machine, version 1 (SYSV), not stripped
$ perf record --event /tmp/foo.o sleep 1
[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.013 MB perf.data ]
$ perf evlist
/tmp/foo.o
$ perf evlist -v
/tmp/foo.o: type: 1, size: 112, config: 0x9, { sample_period, sample_freq }: 4000, sample_type: IP|TID|TIME|PERIOD, disabled: 1, inherit: 1, mmap: 1, comm: 1, freq: 1, enable_on_exec: 1, task: 1, sample_id_all: 1, exclude_guest: 1, mmap2: 1, comm_exec: 1
$
So, type 1 is PERF_TYPE_SOFTWARE, config 0x9 is PERF_COUNT_SW_DUMMY, ok.
$ perf report --stdio
Error:
The perf.data file has no samples!
# To display the perf.data header info, please use --header/--header-only options.
#
$
Signed-off-by: Wang Nan <wangnan0@huawei.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Alexei Starovoitov <ast@plumgrid.com>
Cc: Brendan Gregg <brendan.d.gregg@gmail.com>
Cc: Daniel Borkmann <daniel@iogearbox.net>
Cc: David Ahern <dsahern@gmail.com>
Cc: He Kuang <hekuang@huawei.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Kaixu Xia <xiakaixu@huawei.com>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: Zefan Li <lizefan@huawei.com>
Cc: pi3orama@163.com
Link: http://lkml.kernel.org/r/1444826502-49291-4-git-send-email-wangnan0@huawei.com
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-10-14 15:41:14 +03:00
bpf__clear ( ) ;
2010-03-22 19:10:25 +03:00
2009-04-20 17:00:56 +04:00
if ( status )
return status & 0xff ;
/* Somebody closed stdout? */
if ( fstat ( fileno ( stdout ) , & st ) )
return 0 ;
/* Ignore write errors for pipes and sockets.. */
if ( S_ISFIFO ( st . st_mode ) | | S_ISSOCK ( st . st_mode ) )
return 0 ;
2013-01-24 22:18:54 +04:00
status = 1 ;
2009-04-20 17:00:56 +04:00
/* Check for ENOSPC and EIO errors.. */
2013-01-24 22:18:54 +04:00
if ( fflush ( stdout ) ) {
2014-08-14 06:22:32 +04:00
fprintf ( stderr , " write failure on standard output: %s " ,
tools: Introduce str_error_r()
The tools so far have been using the strerror_r() GNU variant, that
returns a string, be it the buffer passed or something else.
But that, besides being tricky in cases where we expect that the
function using strerror_r() returns the error formatted in a provided
buffer (we have to check if it returned something else and copy that
instead), breaks the build on systems not using glibc, like Alpine
Linux, where musl libc is used.
So, introduce yet another wrapper, str_error_r(), that has the GNU
interface, but uses the portable XSI variant of strerror_r(), so that
users rest asured that the provided buffer is used and it is what is
returned.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/n/tip-d4t42fnf48ytlk8rjxs822tf@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2016-07-06 17:56:20 +03:00
str_error_r ( errno , sbuf , sizeof ( sbuf ) ) ) ;
2013-01-24 22:18:54 +04:00
goto out ;
}
if ( ferror ( stdout ) ) {
fprintf ( stderr , " unknown write failure on standard output " ) ;
goto out ;
}
if ( fclose ( stdout ) ) {
2014-08-14 06:22:32 +04:00
fprintf ( stderr , " close failed on standard output: %s " ,
tools: Introduce str_error_r()
The tools so far have been using the strerror_r() GNU variant, that
returns a string, be it the buffer passed or something else.
But that, besides being tricky in cases where we expect that the
function using strerror_r() returns the error formatted in a provided
buffer (we have to check if it returned something else and copy that
instead), breaks the build on systems not using glibc, like Alpine
Linux, where musl libc is used.
So, introduce yet another wrapper, str_error_r(), that has the GNU
interface, but uses the portable XSI variant of strerror_r(), so that
users rest asured that the provided buffer is used and it is what is
returned.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/n/tip-d4t42fnf48ytlk8rjxs822tf@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2016-07-06 17:56:20 +03:00
str_error_r ( errno , sbuf , sizeof ( sbuf ) ) ) ;
2013-01-24 22:18:54 +04:00
goto out ;
}
status = 0 ;
out :
return status ;
2009-04-20 17:00:56 +04:00
}
static void handle_internal_command ( int argc , const char * * argv )
{
const char * cmd = argv [ 0 ] ;
2009-07-01 14:37:06 +04:00
unsigned int i ;
2009-04-20 17:00:56 +04:00
/* Turn "perf cmd --help" into "perf help cmd" */
if ( argc > 1 & & ! strcmp ( argv [ 1 ] , " --help " ) ) {
argv [ 1 ] = argv [ 0 ] ;
argv [ 0 ] = cmd = " help " ;
}
for ( i = 0 ; i < ARRAY_SIZE ( commands ) ; i + + ) {
struct cmd_struct * p = commands + i ;
if ( strcmp ( p - > cmd , cmd ) )
continue ;
exit ( run_builtin ( p , argc , argv ) ) ;
}
}
static void execv_dashed_external ( const char * * argv )
{
2016-02-25 16:14:50 +03:00
char * cmd ;
2009-04-20 17:00:56 +04:00
const char * tmp ;
int status ;
2016-02-25 16:14:50 +03:00
if ( asprintf ( & cmd , " perf-%s " , argv [ 0 ] ) < 0 )
goto do_die ;
2009-04-20 17:00:56 +04:00
/*
* argv [ 0 ] must be the perf command , but the argv array
* belongs to the caller , and may be reused in
* subsequent loop iterations . Save argv [ 0 ] and
* restore it on error .
*/
tmp = argv [ 0 ] ;
2016-02-25 16:14:50 +03:00
argv [ 0 ] = cmd ;
2009-04-20 17:00:56 +04:00
/*
* if we fail because the command is not found , it is
* OK to return . Otherwise , we just pass along the status code .
*/
status = run_command_v_opt ( argv , 0 ) ;
if ( status ! = - ERR_RUN_COMMAND_EXEC ) {
2016-02-25 16:14:50 +03:00
if ( IS_RUN_COMMAND_ERR ( status ) ) {
do_die :
2017-04-04 17:36:22 +03:00
pr_err ( " FATAL: unable to run '%s' " , argv [ 0 ] ) ;
status = - 128 ;
2016-02-25 16:14:50 +03:00
}
2009-04-20 17:00:56 +04:00
exit ( - status ) ;
}
errno = ENOENT ; /* as if we called execvp */
argv [ 0 ] = tmp ;
2016-02-25 16:14:50 +03:00
zfree ( & cmd ) ;
2009-04-20 17:00:56 +04:00
}
static int run_argv ( int * argcp , const char * * * argv )
{
2017-03-28 17:19:59 +03:00
/* See if it's an internal command */
handle_internal_command ( * argcp , * argv ) ;
2009-04-20 17:00:56 +04:00
2017-03-28 17:19:59 +03:00
/* .. then try the external ones */
execv_dashed_external ( * argv ) ;
return 0 ;
2009-04-20 17:00:56 +04:00
}
2011-10-13 15:52:46 +04:00
static void pthread__block_sigwinch ( void )
{
sigset_t set ;
sigemptyset ( & set ) ;
sigaddset ( & set , SIGWINCH ) ;
pthread_sigmask ( SIG_BLOCK , & set , NULL ) ;
}
void pthread__unblock_sigwinch ( void )
{
sigset_t set ;
sigemptyset ( & set ) ;
sigaddset ( & set , SIGWINCH ) ;
pthread_sigmask ( SIG_UNBLOCK , & set , NULL ) ;
}
2016-07-12 17:01:17 +03:00
# ifdef _SC_LEVEL1_DCACHE_LINESIZE
# define cache_line_size(cacheline_sizep) *cacheline_sizep = sysconf(_SC_LEVEL1_DCACHE_LINESIZE)
# else
static void cache_line_size ( int * cacheline_sizep )
{
if ( sysfs__read_int ( " devices/system/cpu/cpu0/cache/index0/coherency_line_size " , cacheline_sizep ) )
2016-07-15 16:05:19 +03:00
pr_debug ( " cannot determine cache line size " ) ;
2016-07-12 17:01:17 +03:00
}
# endif
2009-04-20 17:00:56 +04:00
int main ( int argc , const char * * argv )
{
2017-01-24 19:44:10 +03:00
int err ;
2009-04-20 17:00:56 +04:00
const char * cmd ;
2014-08-14 06:22:32 +04:00
char sbuf [ STRERR_BUFSIZE ] ;
2016-04-27 16:16:24 +03:00
int value ;
2009-04-20 17:00:56 +04:00
2015-12-15 18:39:35 +03:00
/* libsubcmd init */
exec_cmd_init ( " perf " , PREFIX , PERF_EXEC_PATH , EXEC_PATH_ENVIRONMENT ) ;
pager_init ( PERF_PAGER_ENVIRONMENT ) ;
2013-09-12 20:39:35 +04:00
/* The page_size is placed in util object. */
2012-10-06 21:57:10 +04:00
page_size = sysconf ( _SC_PAGE_SIZE ) ;
2016-07-12 17:01:17 +03:00
cache_line_size ( & cacheline_size ) ;
2012-10-06 21:57:10 +04:00
2016-04-27 16:16:24 +03:00
if ( sysctl__read_int ( " kernel/perf_event_max_stack " , & value ) = = 0 )
sysctl_perf_event_max_stack = value ;
2016-05-17 03:16:54 +03:00
if ( sysctl__read_int ( " kernel/perf_event_max_contexts_per_stack " , & value ) = = 0 )
sysctl_perf_event_max_contexts_per_stack = value ;
2015-12-15 18:39:37 +03:00
cmd = extract_argv0_path ( argv [ 0 ] ) ;
2009-04-20 17:00:56 +04:00
if ( ! cmd )
cmd = " perf-help " ;
2015-09-02 10:56:33 +03:00
2016-01-07 14:41:53 +03:00
srandom ( time ( NULL ) ) ;
2016-06-23 17:14:31 +03:00
perf_config__init ( ) ;
2017-01-24 19:44:10 +03:00
err = perf_config ( perf_default_config , NULL ) ;
if ( err )
return err ;
2016-03-27 20:22:18 +03:00
set_buildid_dir ( NULL ) ;
2016-02-26 12:31:51 +03:00
2015-09-02 10:56:33 +03:00
/* get debugfs/tracefs mount point from /proc/mounts */
tracing_path_mount ( ) ;
2009-04-20 17:00:56 +04:00
/*
* " perf-xxxx " is the same as " perf xxxx " , but we obviously :
*
* - cannot take flags in between the " perf " and the " xxxx " .
* - cannot execute it externally ( since it would just do
* the same thing over again )
*
2017-09-11 14:14:22 +03:00
* So we just directly call the internal command handler . If that one
* fails to handle this , then maybe we just run a renamed perf binary
* that contains a dash in its name . To handle this scenario , we just
* fall through and ignore the " xxxx " part of the command string .
2009-04-20 17:00:56 +04:00
*/
2017-07-20 21:27:39 +03:00
if ( strstarts ( cmd , " perf- " ) ) {
2009-05-25 16:45:24 +04:00
cmd + = 5 ;
2009-04-20 17:00:56 +04:00
argv [ 0 ] = cmd ;
handle_internal_command ( argc , argv ) ;
2017-09-11 14:14:22 +03:00
/*
* If the command is handled , the above function does not
* return undo changes and fall through in such a case .
*/
cmd - = 5 ;
argv [ 0 ] = cmd ;
2009-04-20 17:00:56 +04:00
}
2017-07-20 21:27:39 +03:00
if ( strstarts ( cmd , " trace " ) ) {
2014-05-26 23:02:29 +04:00
# ifdef HAVE_LIBAUDIT_SUPPORT
perf trace: Add 'trace' alias to 'perf trace'
Make 'perf trace' more accessible by aliasing it to just 'trace':
[root@zoo linux]# trace --duration 15 -a -e futex sleep 1
110.092 (16.188 ms): libvirtd/1166 futex(uaddr: 0x185b344, op: WAIT|PRIV, val: 174293 ) = 0
110.101 (15.903 ms): libvirtd/1171 futex(uaddr: 0x185b3dc, op: WAIT|PRIV, val: 139265 ) = 0
111.594 (15.776 ms): libvirtd/1165 futex(uaddr: 0x185b344, op: WAIT|PRIV, val: 174295 ) = 0
111.610 (15.969 ms): libvirtd/1169 futex(uaddr: 0x185b3dc, op: WAIT|PRIV, val: 139267 ) = 0
113.556 (16.216 ms): libvirtd/1168 futex(uaddr: 0x185b3dc, op: WAIT|PRIV, val: 139269 ) = 0
291.265 (199.508 ms): chromium-brows/15830 futex(uaddr: 0x7fff2986bcb4, op: WAIT_BITSET|PRIV|CLKRT, val: 1, utime: 0x7fff2986bab0, val3: 4294967295) = -1 ETIMEDOUT Connection timed out
360.354 (69.053 ms): chromium-brows/15830 futex(uaddr: 0x7fff2986bcb4, op: WAIT_BITSET|PRIV|CLKRT, val: 1, utime: 0x7fff2986bab0, val3: 4294967295) = -1 ETIMEDOUT Connection timed out
[root@zoo linux]#
I.e. looking for futex calls that take at least 15ms, system wide, during a one
second window. Now to get callchains into 'trace' to figure out what are those
locks :-)
Requested-by: Ingo Molnar <mingo@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-ch4smqz8b5fmgrte7c5e4fuw@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2013-09-24 18:56:36 +04:00
setup_path ( ) ;
argv [ 0 ] = " trace " ;
2017-03-27 17:47:20 +03:00
return cmd_trace ( argc , argv ) ;
2014-05-26 23:02:29 +04:00
# else
fprintf ( stderr ,
" trace command not available: missing audit-libs devel package at build time. \n " ) ;
goto out ;
perf trace: Add 'trace' alias to 'perf trace'
Make 'perf trace' more accessible by aliasing it to just 'trace':
[root@zoo linux]# trace --duration 15 -a -e futex sleep 1
110.092 (16.188 ms): libvirtd/1166 futex(uaddr: 0x185b344, op: WAIT|PRIV, val: 174293 ) = 0
110.101 (15.903 ms): libvirtd/1171 futex(uaddr: 0x185b3dc, op: WAIT|PRIV, val: 139265 ) = 0
111.594 (15.776 ms): libvirtd/1165 futex(uaddr: 0x185b344, op: WAIT|PRIV, val: 174295 ) = 0
111.610 (15.969 ms): libvirtd/1169 futex(uaddr: 0x185b3dc, op: WAIT|PRIV, val: 139267 ) = 0
113.556 (16.216 ms): libvirtd/1168 futex(uaddr: 0x185b3dc, op: WAIT|PRIV, val: 139269 ) = 0
291.265 (199.508 ms): chromium-brows/15830 futex(uaddr: 0x7fff2986bcb4, op: WAIT_BITSET|PRIV|CLKRT, val: 1, utime: 0x7fff2986bab0, val3: 4294967295) = -1 ETIMEDOUT Connection timed out
360.354 (69.053 ms): chromium-brows/15830 futex(uaddr: 0x7fff2986bcb4, op: WAIT_BITSET|PRIV|CLKRT, val: 1, utime: 0x7fff2986bab0, val3: 4294967295) = -1 ETIMEDOUT Connection timed out
[root@zoo linux]#
I.e. looking for futex calls that take at least 15ms, system wide, during a one
second window. Now to get callchains into 'trace' to figure out what are those
locks :-)
Requested-by: Ingo Molnar <mingo@kernel.org>
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Frederic Weisbecker <fweisbec@gmail.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Mike Galbraith <efault@gmx.de>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Link: http://lkml.kernel.org/n/tip-ch4smqz8b5fmgrte7c5e4fuw@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2013-09-24 18:56:36 +04:00
# endif
2014-05-26 23:02:29 +04:00
}
2009-04-20 17:00:56 +04:00
/* Look for flags.. */
argv + + ;
argc - - ;
handle_options ( & argv , & argc , NULL ) ;
commit_pager_choice ( ) ;
perf buildid: add perfconfig option to specify buildid cache dir
This patch adds the ability to specify an alternate directory to store the
buildid cache (buildids, copy of binaries). By default, it is hardcoded to
$HOME/.debug. This directory contains immutable data. The layout of the
directory is such that no conflicts in filenames are possible. A modification
in a file, yields a different buildid and thus a different location in the
subdir hierarchy.
You may want to put the buildid cache elsewhere because of disk space
limitation or simply to share the cache between users. It is also useful for
remote collect vs. local analysis of profiles.
This patch adds a new config option to the perfconfig file. Under the tag
'buildid', there is a dir option. For instance, if you have:
$ cat /etc/perfconfig
[buildid]
dir = /var/cache/perf-buildid
All buildids and binaries are be saved in the directory specified. The perf
record, buildid-list, buildid-cache, report, annotate, and archive commands
will it to pull information out.
The option can be set in the system-wide perfconfig file or in the
$HOME/.perfconfig file.
Cc: David S. Miller <davem@davemloft.net>
Cc: Frédéric Weisbecker <fweisbec@gmail.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Paul Mackerras <paulus@samba.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Stephane Eranian <eranian@google.com>
Cc: Tom Zanussi <tzanussi@gmail.com>
LKML-Reference: <4c055fb7.df0ce30a.5f0d.ffffae52@mx.google.com>
Signed-off-by: Stephane Eranian <eranian@google.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-06-01 23:25:01 +04:00
2009-04-20 17:00:56 +04:00
if ( argc > 0 ) {
2017-07-20 21:27:39 +03:00
if ( strstarts ( argv [ 0 ] , " -- " ) )
2009-04-20 17:00:56 +04:00
argv [ 0 ] + = 2 ;
} else {
/* The user didn't specify a command; give them help */
2009-03-13 05:20:49 +03:00
printf ( " \n usage: %s \n \n " , perf_usage_string ) ;
2009-04-20 17:00:56 +04:00
list_common_cmds_help ( ) ;
2009-03-13 05:20:49 +03:00
printf ( " \n %s \n \n " , perf_more_info_string ) ;
2013-01-24 22:18:54 +04:00
goto out ;
2009-04-20 17:00:56 +04:00
}
cmd = argv [ 0 ] ;
2012-10-31 18:52:47 +04:00
test_attr__init ( ) ;
2009-04-20 17:00:56 +04:00
/*
* We use PATH to find perf commands , but we prepend some higher
2010-01-18 18:02:48 +03:00
* precedence paths : the " --exec-path " option , the PERF_EXEC_PATH
2009-04-20 17:00:56 +04:00
* environment , and the $ ( perfexecdir ) from the Makefile at build
* time .
*/
setup_path ( ) ;
2011-10-13 15:52:46 +04:00
/*
* Block SIGWINCH notifications so that the thread that wants it can
* unblock and get syscalls like select interrupted instead of waiting
* forever while the signal goes to some other non interested thread .
*/
pthread__block_sigwinch ( ) ;
2009-04-20 17:00:56 +04:00
2016-02-14 19:03:45 +03:00
perf_debug_setup ( ) ;
2009-04-20 17:00:56 +04:00
while ( 1 ) {
2010-01-28 02:05:55 +03:00
static int done_help ;
2017-03-28 17:19:59 +03:00
run_argv ( & argc , & argv ) ;
2009-06-06 17:19:13 +04:00
2009-04-20 17:00:56 +04:00
if ( errno ! = ENOENT )
break ;
2009-06-06 17:19:13 +04:00
2009-04-20 17:00:56 +04:00
if ( ! done_help ) {
cmd = argv [ 0 ] = help_unknown_cmd ( cmd ) ;
done_help = 1 ;
} else
break ;
}
fprintf ( stderr , " Failed to run command '%s': %s \n " ,
tools: Introduce str_error_r()
The tools so far have been using the strerror_r() GNU variant, that
returns a string, be it the buffer passed or something else.
But that, besides being tricky in cases where we expect that the
function using strerror_r() returns the error formatted in a provided
buffer (we have to check if it returned something else and copy that
instead), breaks the build on systems not using glibc, like Alpine
Linux, where musl libc is used.
So, introduce yet another wrapper, str_error_r(), that has the GNU
interface, but uses the portable XSI variant of strerror_r(), so that
users rest asured that the provided buffer is used and it is what is
returned.
Cc: Adrian Hunter <adrian.hunter@intel.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Jiri Olsa <jolsa@kernel.org>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Wang Nan <wangnan0@huawei.com>
Link: http://lkml.kernel.org/n/tip-d4t42fnf48ytlk8rjxs822tf@git.kernel.org
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2016-07-06 17:56:20 +03:00
cmd , str_error_r ( errno , sbuf , sizeof ( sbuf ) ) ) ;
2013-01-24 22:18:54 +04:00
out :
2009-04-20 17:00:56 +04:00
return 1 ;
}