2009-10-09 01:17:38 +04:00
# ifndef _PROBE_FINDER_H
# define _PROBE_FINDER_H
2010-02-25 16:35:42 +03:00
# include <stdbool.h>
2009-12-28 02:37:00 +03:00
# include "util.h"
2014-02-06 09:32:09 +04:00
# include "intlist.h"
2010-03-17 01:06:12 +03:00
# include "probe-event.h"
2009-12-28 02:37:00 +03:00
2009-12-17 01:16:19 +03:00
# define MAX_PROBE_BUFFER 1024
# define MAX_PROBES 128
2013-10-11 11:10:23 +04:00
# define MAX_PROBE_ARGS 128
2009-10-09 01:17:38 +04:00
perf probe: Support $params special probe argument
$params is similar to $vars but matches only function parameters not
local variables.
Thus, this is useful for tracing function parameter changing or tracing
function call with parameters.
Testing it:
# perf probe tcp_sendmsg '$params'
Added new event:
probe:tcp_sendmsg (on tcp_sendmsg with $params)
You can now use it in all perf tools, such as:
perf record -e probe:tcp_sendmsg -aR sleep 1
# perf probe -l
probe:tcp_sendmsg (on tcp_sendmsg@acme/git/linux/net/ipv4/tcp.c with iocb sk msg size)
# perf record -a -e probe:*
press some random letters to generate TCP (sshd) traffic...
^C[ perf record: Woken up 1 times to write data ]
[ perf record: Captured and wrote 0.223 MB perf.data (6 samples) ]
# perf script
sshd 6385 [2] 3.907529: probe:tcp_sendmsg: iocb=0xffff8800ac4cfe70 sk=0xffff88042196c140 msg=0xffff8800ac4cfda8 size=0x24
sshd 6385 [2] 4.138973: probe:tcp_sendmsg: iocb=0xffff8800ac4cfe70 sk=0xffff88042196c140 msg=0xffff8800ac4cfda8 size=0x24
sshd 6385 [2] 4.378966: probe:tcp_sendmsg: iocb=0xffff8800ac4cfe70 sk=0xffff88042196c140 msg=0xffff8800ac4cfda8 size=0x24
sshd 6385 [2] 4.603681: probe:tcp_sendmsg: iocb=0xffff8800ac4cfe70 sk=0xffff88042196c140 msg=0xffff8800ac4cfda8 size=0x24
sshd 6385 [2] 4.818455: probe:tcp_sendmsg: iocb=0xffff8800ac4cfe70 sk=0xffff88042196c140 msg=0xffff8800ac4cfda8 size=0x24
sshd 6385 [2] 5.043603: probe:tcp_sendmsg: iocb=0xffff8800ac4cfe70 sk=0xffff88042196c140 msg=0xffff8800ac4cfda8 size=0x24
# cat /sys/kernel/debug/tracing/events/probe/tcp_sendmsg/format
name: tcp_sendmsg
ID: 1927
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:unsigned long __probe_ip; offset:8; size:8; signed:0;
field:u64 iocb; offset:16; size:8; signed:0;
field:u64 sk; offset:24; size:8; signed:0;
field:u64 msg; offset:32; size:8; signed:0;
field:u64 size; offset:40; size:8; signed:0;
print fmt: "(%lx) iocb=0x%Lx sk=0x%Lx msg=0x%Lx size=0x%Lx", REC->__probe_ip, REC->iocb, REC->sk, REC->msg, REC->size
#
Do some system wide tracing of this probe + write syscalls:
# perf trace -e write --ev probe:* --filter-pids 6385
462.612 (0.010 ms): bash/19153 write(fd: 1</dev/pts/1>, buf: 0x7f7556c78000, count: 29 ) = 29
462.701 (0.027 ms): sshd/19152 write(fd: 3<socket:[63117]>, buf: 0x7f78dd12e160, count: 68 ) ...
462.701 ( ): probe:tcp_sendmsg:(ffffffff8163db30) iocb=0xffff8803ebec7e70 sk=0xffff88042196ab80 msg=0xffff8803ebec7da8 size=0x44)
462.710 (0.035 ms): sshd/19152 ... [continued]: write()) = 68
462.787 (0.009 ms): bash/19153 write(fd: 2</dev/pts/1>, buf: 0x7f7556c77000, count: 22 ) = 22
462.865 (0.002 ms): sshd/19152 write(fd: 3<socket:[63117]>, buf: 0x7f78dd12e160, count: 68 ) ...
462.865 ( ): probe:tcp_sendmsg:(ffffffff8163db30) iocb=0xffff8803ebec7e70 sk=0xffff88042196ab80 msg=0xffff8803ebec7da8 size=0x44)
462.873 (0.010 ms): sshd/19152 ... [continued]: write()) = 68
Signed-off-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Tested-by: Arnaldo Carvalho de Melo <acme@redhat.com>
Cc: Ananth N Mavinakayanahalli <ananth@in.ibm.com>
Cc: David Ahern <dsahern@gmail.com>
Cc: Hemant Kumar <hemant@linux.vnet.ibm.com>
Cc: Jiri Olsa <jolsa@redhat.com>
Cc: Namhyung Kim <namhyung@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Link: http://lkml.kernel.org/r/20150506124653.4961.59806.stgit@localhost.localdomain
[ Add some examples to the changelog message showing how to use it ]
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2015-05-06 15:46:53 +03:00
# define PROBE_ARG_VARS "$vars"
# define PROBE_ARG_PARAMS "$params"
2009-10-09 01:17:38 +04:00
static inline int is_c_varname ( const char * name )
{
/* TODO */
return isalpha ( name [ 0 ] ) | | name [ 0 ] = = ' _ ' ;
}
2013-09-30 14:07:11 +04:00
# ifdef HAVE_DWARF_SUPPORT
2011-06-27 11:27:39 +04:00
# include "dwarf-aux.h"
/* TODO: export debuginfo data structure even if no dwarf support */
/* debug information structure */
struct debuginfo {
Dwarf * dbg ;
2013-09-25 17:16:16 +04:00
Dwfl_Module * mod ;
2011-06-27 11:27:39 +04:00
Dwfl * dwfl ;
Dwarf_Addr bias ;
} ;
2014-02-06 09:32:27 +04:00
/* This also tries to open distro debuginfo */
2011-06-27 11:27:39 +04:00
extern struct debuginfo * debuginfo__new ( const char * path ) ;
2013-11-05 22:32:36 +04:00
extern void debuginfo__delete ( struct debuginfo * dbg ) ;
2011-06-27 11:27:39 +04:00
2010-07-29 18:13:51 +04:00
/* Find probe_trace_events specified by perf_probe_event from debuginfo */
2013-11-05 22:32:36 +04:00
extern int debuginfo__find_trace_events ( struct debuginfo * dbg ,
2011-06-27 11:27:39 +04:00
struct perf_probe_event * pev ,
2015-05-08 04:03:31 +03:00
struct probe_trace_event * * tevs ) ;
2010-03-17 01:06:12 +03:00
2010-03-17 01:06:19 +03:00
/* Find a perf_probe_point from debuginfo */
2013-11-05 22:32:36 +04:00
extern int debuginfo__find_probe_point ( struct debuginfo * dbg ,
2011-06-27 11:27:39 +04:00
unsigned long addr ,
struct perf_probe_point * ppt ) ;
2010-03-17 01:06:19 +03:00
2010-10-21 14:13:23 +04:00
/* Find a line range */
2013-11-05 22:32:36 +04:00
extern int debuginfo__find_line_range ( struct debuginfo * dbg ,
2011-06-27 11:27:39 +04:00
struct line_range * lr ) ;
2009-10-09 01:17:38 +04:00
2010-10-21 14:13:23 +04:00
/* Find available variables */
2013-11-05 22:32:36 +04:00
extern int debuginfo__find_available_vars_at ( struct debuginfo * dbg ,
2011-06-27 11:27:39 +04:00
struct perf_probe_event * pev ,
2015-05-08 04:03:31 +03:00
struct variable_list * * vls ) ;
2009-10-09 01:17:38 +04:00
2015-03-13 08:18:40 +03:00
/* Find a src file from a DWARF tag path */
int get_real_path ( const char * raw_path , const char * comp_dir ,
char * * new_path ) ;
2009-10-09 01:17:38 +04:00
struct probe_finder {
2010-03-17 01:06:12 +03:00
struct perf_probe_event * pev ; /* Target probe event */
2010-10-21 14:13:23 +04:00
/* Callback when a probe point is found */
2011-08-11 15:02:59 +04:00
int ( * callback ) ( Dwarf_Die * sc_die , struct probe_finder * pf ) ;
2009-10-09 01:17:38 +04:00
/* For function searching */
perf tools: Reorganize some structs to save space
Using 'pahole --packable' I found some structs that could be reorganized
to eliminate alignment holes, in some cases getting them to be cacheline
multiples.
[acme@doppio linux-2.6-tip]$ codiff perf.old ~/bin/perf
builtin-annotate.c:
struct perf_session | -8
struct perf_header | -8
2 structs changed
builtin-diff.c:
struct sample_data | -8
1 struct changed
diff__process_sample_event | -8
1 function changed, 8 bytes removed, diff: -8
builtin-sched.c:
struct sched_atom | -8
1 struct changed
builtin-timechart.c:
struct per_pid | -8
1 struct changed
cmd_timechart | -16
1 function changed, 16 bytes removed, diff: -16
builtin-probe.c:
struct perf_probe_point | -8
struct perf_probe_event | -8
2 structs changed
opt_add_probe_event | -3
1 function changed, 3 bytes removed, diff: -3
util/probe-finder.c:
struct probe_finder | -8
1 struct changed
find_kprobe_trace_events | -16
1 function changed, 16 bytes removed, diff: -16
/home/acme/bin/perf:
4 functions changed, 43 bytes removed, diff: -43
[acme@doppio linux-2.6-tip]$
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: <new-submission>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2010-04-05 19:53:45 +04:00
int lno ; /* Line number */
2010-02-25 16:35:42 +03:00
Dwarf_Addr addr ; /* Address */
2010-03-17 01:06:12 +03:00
const char * fname ; /* Real file name */
2010-02-25 16:35:42 +03:00
Dwarf_Die cu_die ; /* Current CU */
perf probe: Add fastpath to do lookup by function name
v3 -> v2:
- Make pubname_search_cb more generic
- Add fastpath to find_probes also
v2 -> v1:
- Don't compare file names with cu_find_realpath(...), instead, compare
them with the name returned by dwarf_decl_file(sp_die)
The vmlinux file may have thousands of CUs.
We can lookup function name from .debug_pubnames section
to avoid the slow loop on CUs.
1. Improvement data for find_line_range
./perf stat -e cycles -r 10 -- ./perf probe -k /home/mlin/vmlinux \
-s /home/mlin/linux-2.6 \
--line csum_partial_copy_to_user > tmp.log
before patch applied
=====================
847,988,276 cycles
0.355075856 seconds time elapsed
after patch applied
=====================
206,102,622 cycles
0.086883555 seconds time elapsed
2. Improvement data for find_probes
./perf stat -e cycles -r 10 -- ./perf probe -k /home/mlin/vmlinux \
-s /home/mlin/linux-2.6 \
--vars csum_partial_copy_to_user > tmp.log
before patch applied
=====================
848,490,844 cycles
0.355307901 seconds time elapsed
after patch applied
=====================
205,684,469 cycles
0.086694010 seconds time elapsed
Acked-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: linux-kernel <linux-kernel@vger.kernel.org>
LKML-Reference: <1301041668.14111.52.camel@minggr.sh.intel.com>
Signed-off-by: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-03-25 11:27:48 +03:00
Dwarf_Die sp_die ;
2014-02-06 09:32:09 +04:00
struct intlist * lcache ; /* Line cache for lazy match */
2009-10-09 01:17:38 +04:00
/* For variable searching */
2010-05-10 21:12:07 +04:00
# if _ELFUTILS_PREREQ(0, 142)
2010-04-12 21:17:29 +04:00
Dwarf_CFI * cfi ; /* Call Frame Information */
2010-05-10 21:12:07 +04:00
# endif
2010-02-25 16:35:42 +03:00
Dwarf_Op * fb_ops ; /* Frame base attribute */
2010-03-17 01:06:12 +03:00
struct perf_probe_arg * pvar ; /* Current target variable */
2010-07-29 18:13:51 +04:00
struct probe_trace_arg * tvar ; /* Current result variable */
2009-10-09 01:17:38 +04:00
} ;
2010-01-06 17:45:34 +03:00
2010-10-21 14:13:23 +04:00
struct trace_event_finder {
struct probe_finder pf ;
2013-09-25 17:16:16 +04:00
Dwfl_Module * mod ; /* For solving symbols */
2010-10-21 14:13:23 +04:00
struct probe_trace_event * tevs ; /* Found trace events */
int ntevs ; /* Number of trace events */
int max_tevs ; /* Max number of trace events */
} ;
struct available_var_finder {
struct probe_finder pf ;
2013-09-25 17:16:16 +04:00
Dwfl_Module * mod ; /* For solving symbols */
2010-10-21 14:13:23 +04:00
struct variable_list * vls ; /* Found variable lists */
int nvls ; /* Number of variable lists */
int max_vls ; /* Max no. of variable lists */
2010-10-21 14:13:35 +04:00
bool child ; /* Search child scopes */
2010-10-21 14:13:23 +04:00
} ;
2010-01-06 17:45:34 +03:00
struct line_finder {
2010-02-25 16:35:42 +03:00
struct line_range * lr ; /* Target line range */
const char * fname ; /* File name */
int lno_s ; /* Start line number */
int lno_e ; /* End line number */
Dwarf_Die cu_die ; /* Current CU */
perf probe: Add fastpath to do lookup by function name
v3 -> v2:
- Make pubname_search_cb more generic
- Add fastpath to find_probes also
v2 -> v1:
- Don't compare file names with cu_find_realpath(...), instead, compare
them with the name returned by dwarf_decl_file(sp_die)
The vmlinux file may have thousands of CUs.
We can lookup function name from .debug_pubnames section
to avoid the slow loop on CUs.
1. Improvement data for find_line_range
./perf stat -e cycles -r 10 -- ./perf probe -k /home/mlin/vmlinux \
-s /home/mlin/linux-2.6 \
--line csum_partial_copy_to_user > tmp.log
before patch applied
=====================
847,988,276 cycles
0.355075856 seconds time elapsed
after patch applied
=====================
206,102,622 cycles
0.086883555 seconds time elapsed
2. Improvement data for find_probes
./perf stat -e cycles -r 10 -- ./perf probe -k /home/mlin/vmlinux \
-s /home/mlin/linux-2.6 \
--vars csum_partial_copy_to_user > tmp.log
before patch applied
=====================
848,490,844 cycles
0.355307901 seconds time elapsed
after patch applied
=====================
205,684,469 cycles
0.086694010 seconds time elapsed
Acked-by: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Masami Hiramatsu <masami.hiramatsu.pt@hitachi.com>
Cc: Peter Zijlstra <a.p.zijlstra@chello.nl>
Cc: linux-kernel <linux-kernel@vger.kernel.org>
LKML-Reference: <1301041668.14111.52.camel@minggr.sh.intel.com>
Signed-off-by: Lin Ming <ming.m.lin@intel.com>
Signed-off-by: Arnaldo Carvalho de Melo <acme@redhat.com>
2011-03-25 11:27:48 +03:00
Dwarf_Die sp_die ;
2010-01-06 17:45:34 +03:00
int found ;
} ;
2013-09-30 14:07:11 +04:00
# endif /* HAVE_DWARF_SUPPORT */
2009-10-09 01:17:38 +04:00
# endif /*_PROBE_FINDER_H */