perf-probe updates - data-structure support, etc.
cover perf-probe updates - data-structure support, etc.
Hi Ingo,
Here are several updates of perf-probe.
This updates includes data structure accessing support --list option enhancement and --dry-run option support.
- --dry-run allows users to run perf-probe just for checking that the command works correctly.
- --list option enhancement allows user to check where the probes are put on (with filename/line number).
- data structure accessing support allows user to trace the contents of local variables/function parameters.
Long-term TODOs (future features):
- Enhance probe-finder to decode call frame instructions.
- Support tracing static variables (non global)
- Support variable basic types from debuginfo (e.g. char, int, ...)
- Support array element (var[N])
- Support string/dynamic arrays (*var, var[N..M])
- Support the type of return value
- Support force type-casting ((type)var)
- Support sys_perf_counter_open (for non-root users)
- Support dynamic array-indexing (var[var2])
Miscs:
- Better support for probes on modules
- More debugger like enhancements(%next, --disasm, etc.)
Thank you,
---
Masami Hiramatsu (10):
perf probe: Accessing members in data structures perf probe: List probes with line number and file name perf probe: Introduce kprobe_trace_event and perf_probe_event perf probe: Add --dry-run option
perf probe: Introduce die_find_child() function perf probe: Rename some die_get_* functions
perf probe: Rename session to param
perf probe: Move add-probe routine under util/ perf probe: Use wrapper functions
perf tools: Introduce xzalloc() for detecting out of memory
tools/perf/Documentation/perf-probe.txt | 5 tools/perf/builtin-probe.c | 191 +------ tools/perf/util/probe-event.c | 830 +++++++++++++++++++++++-------- tools/perf/util/probe-event.h | 123 ++++- tools/perf/util/probe-finder.c | 459 ++++++++++++----- tools/perf/util/probe-finder.h | 60 +- tools/perf/util/util.h | 7 7 files changed, 1115 insertions(+), 560 deletions(-)
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
PATCH 1/10 perf tools: Introduce xzalloc() for detecting out of memory
Introducing xzalloc() which wrapping zalloc() for detecting out of memory.
---
tools/perf/util/util.h | 7 +++++++ 1 files changed, 7 insertions(+), 0 deletions(-)
tools/perf/util/util.h +7 -0
@@ -295,6 +295,13 @@ extern void *xmemdupz(const void *data, size_t len);extern char *xstrndup (const char *str,size_t len);extern void *xrealloc (void *ptr,size_t size) __attribute__((weak)); +static inline void *xzalloc (size_t size) +{ +void *buf; + buf =xmalloc (size); +return memset (buf,0 , size); +} +static inline void *zalloc (size_t size) {return calloc (1 , size);
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
[tip:perf/core] perf tools: Introduce xzalloc() for detecting out of memory conditions
Commit-ID: a1d37d5285bcda07f9c0b80a2634ca20ab545297 Gitweb: http://git.kernel.org/tip/a1d37d5285bcda07f9c0b80a2634ca20ab545297 Author: Masami Hiramatsu <[email protected]> AuthorDate: Tue, 16 Mar 2010 18:05:21 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 11:32:29 +0100
perf tools: Introduce xzalloc() for detecting out of memory conditions
Introducing xzalloc() which wrapping zalloc() for detecting out of memory conditions.
LKML-Reference: <[email protected]>
--- tools/perf/util/util.h | 9 +++++++++ 1 files changed, 9 insertions(+), 0 deletions(-)
tools/perf/util/util.h +9 -0
@@ -295,6 +295,13 @@ extern void *xmemdupz(const void *data, size_t len);extern char *xstrndup (const char *str,size_t len);extern void *xrealloc (void *ptr,size_t size) __attribute__((weak)); +static inline void *xzalloc (size_t size) +{ +void *buf =xmalloc (size); + +return memset (buf,0 , size); +} +static inline void *zalloc (size_t size) {return calloc (1 , size);
@@ -309,6 +316,7 @@ static inline int has_extension(const char *filename, const char *ext) {size_t len =strlen (filename);size_t extlen =strlen (ext); +return len > extlen && !memcmp (filename + len - extlen, ext, extlen); }
@@ -322,6 +330,7 @@ static inline int has_extension(const char *filename, const char *ext)#undef isalnum#undef tolower#undef toupper +extern unsigned char sane_ctype[256 ];#define GIT_SPACE 0x01#define GIT_DIGIT 0x02
PATCH 2/10 perf probe: Use wrapper functions
In reply to Masami HiramatsuUse wrapperf functions as much as possible, to check out of memory in perf probe.
---
tools/perf/builtin-probe.c | 4 +-- tools/perf/util/probe-event.c | 46 +++++++++++++++------------------------- tools/perf/util/probe-finder.c | 14 +++++------- 3 files changed, 24 insertions(+), 40 deletions(-)
tools/perf/builtin-probe.c +1 -3
@@ -87,9 +87,7 @@ static void parse_probe_event_argv(int argc, const char **argv) len =0 ;for (i =0 ; i < argc; i++) len +=strlen (argv[i]) +1 ; - buf =zalloc (len +1 ); -if (!buf) -die ("Failed to allocate memory for binding arguments." ); + buf =xzalloc (len +1 ); len =0 ;for (i =0 ; i < argc; i++) len +=sprintf (&buf[len],"%s " , argv[i]);
tools/perf/util/probe-event.c +17 -29
127 lines not on this page · raw message
tools/perf/util/probe-finder.c +6 -8
@@ -125,8 +125,7 @@ static void line_list__add_line(struct list_head *head, unsigned int line) p = head; found:pr_debug ("line list: add a line %u\n" , line); - ln =zalloc (sizeof (struct line_node )); -DIE_IF (ln ==NULL ); + ln =xzalloc (sizeof (struct line_node )); ln->line = line;INIT_LIST_HEAD (&ln->list);list_add (&ln->list, p);
@@ -416,7 +415,7 @@ static void show_probe_point(Dwarf_Die *sp_die, struct probe_finder *pf) (unsigned long)(pf->addr - eaddr));/* Copy the function name if possible */ if (!pp->function) { - pp->function =strdup (name); + pp->function =xstrdup (name); pp->offset = (size_t )(pf->addr - eaddr); } }else {
@@ -425,7 +424,7 @@ static void show_probe_point(Dwarf_Die *sp_die, struct probe_finder *pf) (uintmax_t)pf->addr);if (!pp->function) {/* TODO: Use _stext */ - pp->function =strdup ("" ); + pp->function =xstrdup ("" ); pp->offset = (size_t )pf->addr; } }
@@ -456,7 +455,7 @@ static void show_probe_point(Dwarf_Die *sp_die, struct probe_finder *pf)if (pp->found ==MAX_PROBES )die ("Too many( > %d) probe point found.\n" ,MAX_PROBES ); - pp->probes[pp->found] =strdup (tmp); + pp->probes[pp->found] =xstrdup (tmp); pp->found++; }
@@ -506,8 +505,7 @@ static int find_lazy_match_lines(struct list_head *head,if (fd <0 )die ("failed to open %s" , fname);DIE_IF (fstat (fd, &st) <0 ); - fbuf =malloc (st.st_size +2 ); -DIE_IF (fbuf ==NULL ); + fbuf =xmalloc (st.st_size +2 );DIE_IF (read (fd, fbuf, st.st_size) <0 );close (fd); fbuf[st.st_size] ='\n' ;/* Dummy line */
@@ -727,7 +725,7 @@ static void find_line_range_by_line(Dwarf_Die *sp_die, struct line_finder *lf)/* Copy real path */ if (!lf->lr->path) - lf->lr->path =strdup (src); + lf->lr->path =xstrdup (src);line_list__add_line (&lf->lr->line_list, (unsigned int )lineno); }/* Update status */
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
[tip:perf/core] perf probe: Use wrapper functions
Commit-ID: 31facc5f1ac674fbcc29f212377e589396bb934c Gitweb: http://git.kernel.org/tip/31facc5f1ac674fbcc29f212377e589396bb934c Author: Masami Hiramatsu <[email protected]> AuthorDate: Tue, 16 Mar 2010 18:05:30 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 11:32:29 +0100
perf probe: Use wrapper functions
Use wrapped functions as much as possible, to check out of memory conditions in perf probe.
LKML-Reference: <[email protected]>
--- tools/perf/builtin-probe.c | 4 +-- tools/perf/util/probe-event.c | 46 ++++++++++++++------------------------- tools/perf/util/probe-finder.c | 14 +++++------- 3 files changed, 24 insertions(+), 40 deletions(-)
tools/perf/builtin-probe.c +1 -3
@@ -87,9 +87,7 @@ static void parse_probe_event_argv(int argc, const char **argv) len =0 ;for (i =0 ; i < argc; i++) len +=strlen (argv[i]) +1 ; - buf =zalloc (len +1 ); -if (!buf) -die ("Failed to allocate memory for binding arguments." ); + buf =xzalloc (len +1 ); len =0 ;for (i =0 ; i < argc; i++) len +=sprintf (&buf[len],"%s " , argv[i]);
tools/perf/util/probe-event.c +17 -29
127 lines not on this page · raw message
tools/perf/util/probe-finder.c +6 -8
@@ -125,8 +125,7 @@ static void line_list__add_line(struct list_head *head, unsigned int line) p = head; found:pr_debug ("line list: add a line %u\n" , line); - ln =zalloc (sizeof (struct line_node )); -DIE_IF (ln ==NULL ); + ln =xzalloc (sizeof (struct line_node )); ln->line = line;INIT_LIST_HEAD (&ln->list);list_add (&ln->list, p);
@@ -416,7 +415,7 @@ static void show_probe_point(Dwarf_Die *sp_die, struct probe_finder *pf) (unsigned long)(pf->addr - eaddr));/* Copy the function name if possible */ if (!pp->function) { - pp->function =strdup (name); + pp->function =xstrdup (name); pp->offset = (size_t )(pf->addr - eaddr); } }else {
@@ -425,7 +424,7 @@ static void show_probe_point(Dwarf_Die *sp_die, struct probe_finder *pf) (uintmax_t)pf->addr);if (!pp->function) {/* TODO: Use _stext */ - pp->function =strdup ("" ); + pp->function =xstrdup ("" ); pp->offset = (size_t )pf->addr; } }
@@ -456,7 +455,7 @@ static void show_probe_point(Dwarf_Die *sp_die, struct probe_finder *pf)if (pp->found ==MAX_PROBES )die ("Too many( > %d) probe point found.\n" ,MAX_PROBES ); - pp->probes[pp->found] =strdup (tmp); + pp->probes[pp->found] =xstrdup (tmp); pp->found++; }
@@ -506,8 +505,7 @@ static int find_lazy_match_lines(struct list_head *head,if (fd <0 )die ("failed to open %s" , fname);DIE_IF (fstat (fd, &st) <0 ); - fbuf =malloc (st.st_size +2 ); -DIE_IF (fbuf ==NULL ); + fbuf =xmalloc (st.st_size +2 );DIE_IF (read (fd, fbuf, st.st_size) <0 );close (fd); fbuf[st.st_size] ='\n' ;/* Dummy line */
@@ -727,7 +725,7 @@ static void find_line_range_by_line(Dwarf_Die *sp_die, struct line_finder *lf)/* Copy real path */ if (!lf->lr->path) - lf->lr->path =strdup (src); + lf->lr->path =xstrdup (src);line_list__add_line (&lf->lr->line_list, (unsigned int )lineno); }/* Update status */
PATCH 3/10 perf probe: Move add-probe routine under util/
In reply to Masami HiramatsuMove add-probe routine into util/probe_event.c. This simplifies main routine for reducing maintenance cost.
---
tools/perf/builtin-probe.c | 126 +------------------------------------- tools/perf/util/probe-event.c | 137 ++++++++++++++++++++++++++++++++++++++++- tools/perf/util/probe-event.h | 2 - 3 files changed, 139 insertions(+), 126 deletions(-)
tools/perf/builtin-probe.c +3 -123
@@ -36,11 +36,9 @@#include "builtin.h" #include "util/util.h" #include "util/strlist.h" -#include "util/event.h" +#include "util/symbol.h" #include "util/debug.h" #include "util/debugfs.h" -#include "util/symbol.h" -#include "util/thread.h" #include "util/parse-options.h" #include "util/parse-events.h" /* For debugfs_path */ #include "util/probe-finder.h"
@@ -57,8 +55,6 @@ static struct {int nr_probe;struct probe_point probes[MAX_PROBES ];struct strlist *dellist; -struct map_groups kmap_groups; -struct map *kmaps[MAP__NR_TYPES ];struct line_range line_range; } session;
@@ -114,29 +110,7 @@ static int opt_del_probe_event(const struct option *opt __used,return 0 ; } -/* Currently just checking function name from symbol map */ -static void evaluate_probe_point (struct probe_point *pp) -{ -struct symbol *sym; - sym =map__find_symbol_by_name (session.kmaps[MAP__FUNCTION ], - pp->function,NULL ); -if (!sym) -die ("Kernel symbol \'%s\' not found - probe not added." , - pp->function); -} -#ifndef NO_DWARF_SUPPORT -static int open_vmlinux (void ) -{ -if (map__load (session.kmaps[MAP__FUNCTION ],NULL ) <0 ) { -pr_debug ("Failed to load kernel map.\n" ); -return -EINVAL ; - } -pr_debug ("Try to open %s\n" , - session.kmaps[MAP__FUNCTION ]->dso->long_name); -return open (session.kmaps[MAP__FUNCTION ]->dso->long_name,O_RDONLY ); -} -static int opt_show_lines (const struct option *opt__used ,const char *str,int unset__used ) {
@@ -204,31 +178,8 @@ static const struct option options[] = { OPT_END() }; -/* Initialize symbol maps for vmlinux */ -static void init_vmlinux (void ) -{ - symbol_conf.sort_by_name = true; -if (symbol_conf.vmlinux_name ==NULL ) - symbol_conf.try_vmlinux_path = true; -else -pr_debug ("Use vmlinux: %s\n" , symbol_conf.vmlinux_name); -if (symbol__init () <0 ) -die ("Failed to init symbol map." ); - -map_groups__init (&session.kmap_groups); -if (map_groups__create_kernel_maps (&session.kmap_groups, - session.kmaps) <0 ) -die ("Failed to create kernel maps." ); -} -int cmd_probe (int argc,const char **argv,const char *prefix__used ) { -int i, ret; -#ifndef NO_DWARF_SUPPORT -int fd; -#endif -struct probe_point *pp; - argc =parse_options (argc, argv, options, probe_usage,PARSE_OPT_STOP_AT_NON_OPTION );if (argc >0 ) {
@@ -267,14 +218,7 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) " --add/--del.\n");usage_with_options (probe_usage, options); } -init_vmlinux (); - fd =open_vmlinux (); -if (fd <0 ) -die ("Could not open debuginfo file." ); - ret =find_line_range (fd, &session.line_range); -if (ret <=0 ) -die ("Source line is not found.\n" ); -close (fd); +show_line_range (&session.line_range);return 0 ; }
@@ -287,72 +231,8 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)return 0 ; } -/* Add probes */ -init_vmlinux (); - - if (session.need_dwarf) -#ifdef NO_DWARF_SUPPORT -die ("Debuginfo-analysis is not supported" ); -#else /* !NO_DWARF_SUPPORT */ -pr_debug ("Some probes require debuginfo.\n" ); - - fd =open_vmlinux (); -if (fd <0 ) { -if (session.need_dwarf) -die ("Could not open debuginfo file." ); - -pr_debug ("Could not open vmlinux/module file." -" Try to use symbols.\n" ); - goto end_dwarf; - } - -/* Searching probe points */ -for (i =0 ; i < session.nr_probe; i++) { - pp = &session.probes[i]; -if (pp->found) -continue ; - -lseek (fd,SEEK_SET ,0 ); - ret =find_probe_point (fd, pp); -if (ret >0 ) -continue ; -if (ret ==0 ) {/* No error but failed to find probe point. */ -synthesize_perf_probe_point (pp); -die ("Probe point '%s' not found. - probe not added." , - pp->probes[0 ]); - } -/* Error path */ -if (session.need_dwarf) { -if (ret == -ENOENT ) -pr_warning ("No dwarf info found in the vmlinux - please rebuild with CONFIG_DEBUG_INFO=y.\n" ); -die ("Could not analyze debuginfo." ); - } -pr_debug ("An error occurred in debuginfo analysis." -" Try to use symbols.\n" ); -break ; - } -close (fd); - -end_dwarf: -#endif /* !NO_DWARF_SUPPORT */ - -/* Synthesize probes without dwarf */ -for (i =0 ; i < session.nr_probe; i++) { - pp = &session.probes[i]; -if (pp->found)/* This probe is already found. */ -continue ; - -evaluate_probe_point (pp); - ret =synthesize_trace_kprobe_event (pp); -if (ret == -E2BIG ) -die ("probe point definition becomes too long." ); -else if (ret <0 ) -die ("Failed to synthesize a probe point." ); - } - -/* Settng up probe points */ add_trace_kprobe_events (session.probes, session.nr_probe, - session.force_add); + session.force_add, session.need_dwarf);return 0 ; }
tools/perf/util/probe-event.c +135 -2
170 lines not on this page · raw message
tools/perf/util/probe-event.h +1 -1
@@ -13,7 +13,7 @@ extern int synthesize_perf_probe_event(struct probe_point *pp);extern void parse_trace_kprobe_event (const char *str,struct probe_point *pp);extern int synthesize_trace_kprobe_event (struct probe_point *pp);extern void add_trace_kprobe_events (struct probe_point *probes,int nr_probes, -bool force_add); +bool force_add,bool need_dwarf);extern void del_trace_kprobe_events (struct strlist *dellist);extern void show_perf_probe_events (void );extern void show_line_range (struct line_range *lr);
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
[tip:perf/core] perf probe: Move add-probe routine to util/
Commit-ID: e0faa8d35845bb1893cf9e608a5a5d92e9390bf0 Gitweb: http://git.kernel.org/tip/e0faa8d35845bb1893cf9e608a5a5d92e9390bf0 Author: Masami Hiramatsu <[email protected]> AuthorDate: Tue, 16 Mar 2010 18:05:37 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 11:32:30 +0100
perf probe: Move add-probe routine to util/
Move add-probe routine to util/probe_event.c. This simplifies main routine for reducing maintenance cost.
LKML-Reference: <[email protected]>
--- tools/perf/builtin-probe.c | 126 +------------------------------------ tools/perf/util/probe-event.c | 137 ++++++++++++++++++++++++++++++++++++++++- tools/perf/util/probe-event.h | 2 +- 3 files changed, 139 insertions(+), 126 deletions(-)
tools/perf/builtin-probe.c +3 -123
@@ -36,11 +36,9 @@#include "builtin.h" #include "util/util.h" #include "util/strlist.h" -#include "util/event.h" +#include "util/symbol.h" #include "util/debug.h" #include "util/debugfs.h" -#include "util/symbol.h" -#include "util/thread.h" #include "util/parse-options.h" #include "util/parse-events.h" /* For debugfs_path */ #include "util/probe-finder.h"
@@ -57,8 +55,6 @@ static struct {int nr_probe;struct probe_point probes[MAX_PROBES ];struct strlist *dellist; -struct map_groups kmap_groups; -struct map *kmaps[MAP__NR_TYPES ];struct line_range line_range; } session;
@@ -114,29 +110,7 @@ static int opt_del_probe_event(const struct option *opt __used,return 0 ; } -/* Currently just checking function name from symbol map */ -static void evaluate_probe_point (struct probe_point *pp) -{ -struct symbol *sym; - sym =map__find_symbol_by_name (session.kmaps[MAP__FUNCTION ], - pp->function,NULL ); -if (!sym) -die ("Kernel symbol \'%s\' not found - probe not added." , - pp->function); -} -#ifndef NO_DWARF_SUPPORT -static int open_vmlinux (void ) -{ -if (map__load (session.kmaps[MAP__FUNCTION ],NULL ) <0 ) { -pr_debug ("Failed to load kernel map.\n" ); -return -EINVAL ; - } -pr_debug ("Try to open %s\n" , - session.kmaps[MAP__FUNCTION ]->dso->long_name); -return open (session.kmaps[MAP__FUNCTION ]->dso->long_name,O_RDONLY ); -} -static int opt_show_lines (const struct option *opt__used ,const char *str,int unset__used ) {
@@ -204,31 +178,8 @@ static const struct option options[] = { OPT_END() }; -/* Initialize symbol maps for vmlinux */ -static void init_vmlinux (void ) -{ - symbol_conf.sort_by_name = true; -if (symbol_conf.vmlinux_name ==NULL ) - symbol_conf.try_vmlinux_path = true; -else -pr_debug ("Use vmlinux: %s\n" , symbol_conf.vmlinux_name); -if (symbol__init () <0 ) -die ("Failed to init symbol map." ); - -map_groups__init (&session.kmap_groups); -if (map_groups__create_kernel_maps (&session.kmap_groups, - session.kmaps) <0 ) -die ("Failed to create kernel maps." ); -} -int cmd_probe (int argc,const char **argv,const char *prefix__used ) { -int i, ret; -#ifndef NO_DWARF_SUPPORT -int fd; -#endif -struct probe_point *pp; - argc =parse_options (argc, argv, options, probe_usage,PARSE_OPT_STOP_AT_NON_OPTION );if (argc >0 ) {
@@ -267,14 +218,7 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) " --add/--del.\n");usage_with_options (probe_usage, options); } -init_vmlinux (); - fd =open_vmlinux (); -if (fd <0 ) -die ("Could not open debuginfo file." ); - ret =find_line_range (fd, &session.line_range); -if (ret <=0 ) -die ("Source line is not found.\n" ); -close (fd); +show_line_range (&session.line_range);return 0 ; }
@@ -287,72 +231,8 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)return 0 ; } -/* Add probes */ -init_vmlinux (); - - if (session.need_dwarf) -#ifdef NO_DWARF_SUPPORT -die ("Debuginfo-analysis is not supported" ); -#else /* !NO_DWARF_SUPPORT */ -pr_debug ("Some probes require debuginfo.\n" ); - - fd =open_vmlinux (); -if (fd <0 ) { -if (session.need_dwarf) -die ("Could not open debuginfo file." ); - -pr_debug ("Could not open vmlinux/module file." -" Try to use symbols.\n" ); - goto end_dwarf; - } - -/* Searching probe points */ -for (i =0 ; i < session.nr_probe; i++) { - pp = &session.probes[i]; -if (pp->found) -continue ; - -lseek (fd,SEEK_SET ,0 ); - ret =find_probe_point (fd, pp); -if (ret >0 ) -continue ; -if (ret ==0 ) {/* No error but failed to find probe point. */ -synthesize_perf_probe_point (pp); -die ("Probe point '%s' not found. - probe not added." , - pp->probes[0 ]); - } -/* Error path */ -if (session.need_dwarf) { -if (ret == -ENOENT ) -pr_warning ("No dwarf info found in the vmlinux - please rebuild with CONFIG_DEBUG_INFO=y.\n" ); -die ("Could not analyze debuginfo." ); - } -pr_debug ("An error occurred in debuginfo analysis." -" Try to use symbols.\n" ); -break ; - } -close (fd); - -end_dwarf: -#endif /* !NO_DWARF_SUPPORT */ - -/* Synthesize probes without dwarf */ -for (i =0 ; i < session.nr_probe; i++) { - pp = &session.probes[i]; -if (pp->found)/* This probe is already found. */ -continue ; - -evaluate_probe_point (pp); - ret =synthesize_trace_kprobe_event (pp); -if (ret == -E2BIG ) -die ("probe point definition becomes too long." ); -else if (ret <0 ) -die ("Failed to synthesize a probe point." ); - } - -/* Settng up probe points */ add_trace_kprobe_events (session.probes, session.nr_probe, - session.force_add); + session.force_add, session.need_dwarf);return 0 ; }
tools/perf/util/probe-event.c +135 -2
170 lines not on this page · raw message
tools/perf/util/probe-event.h +1 -1
@@ -13,7 +13,7 @@ extern int synthesize_perf_probe_event(struct probe_point *pp);extern void parse_trace_kprobe_event (const char *str,struct probe_point *pp);extern int synthesize_trace_kprobe_event (struct probe_point *pp);extern void add_trace_kprobe_events (struct probe_point *probes,int nr_probes, -bool force_add); +bool force_add,bool need_dwarf);extern void del_trace_kprobe_events (struct strlist *dellist);extern void show_perf_probe_events (void );extern void show_line_range (struct line_range *lr);
PATCH 4/10 perf probe: Rename session to param
In reply to Masami HiramatsuSince this name 'session' conflicts with 'perf_session', and this structure just holds parameters anymore.
---
tools/perf/builtin-probe.c | 54 ++++++++++++++++++++++---------------------- 1 files changed, 27 insertions(+), 27 deletions(-)
tools/perf/builtin-probe.c +27 -27
@@ -56,20 +56,20 @@ static struct {struct probe_point probes[MAX_PROBES ];struct strlist *dellist;struct line_range line_range; -} session; +} params;/* Parse an event definition. Note that any error must die. */ static void parse_probe_event (const char *str) { -struct probe_point *pp = &session.probes[session.nr_probe]; +struct probe_point *pp = ¶ms.probes[params.nr_probe]; -pr_debug ("probe-definition(%d): %s\n" , session.nr_probe, str); -if (++session.nr_probe ==MAX_PROBES ) +pr_debug ("probe-definition(%d): %s\n" , params.nr_probe, str); +if (++params.nr_probe ==MAX_PROBES )die ("Too many probes (> %d) are specified." ,MAX_PROBES );/* Parse perf-probe event into probe_point */ -parse_perf_probe_event (str, pp, &session.need_dwarf); +parse_perf_probe_event (str, pp, ¶ms.need_dwarf);pr_debug ("%d arguments\n" , pp->nr_args); }
@@ -103,9 +103,9 @@ static int opt_del_probe_event(const struct option *opt __used, const char *str, int unset __used) { if (str) { - if (!session.dellist) - session.dellist = strlist__new(true, NULL); -strlist__add (session.dellist, str); + if (!params.dellist) + params.dellist = strlist__new(true, NULL); +strlist__add (params.dellist, str); }return 0 ; }
@@ -115,9 +115,9 @@ static int opt_show_lines(const struct option *opt __used, const char *str, int unset __used) { if (str) - parse_line_range_desc(str, &session.line_range); -INIT_LIST_HEAD (&session.line_range.line_list); - session.show_lines = true; + parse_line_range_desc(str, ¶ms.line_range); +INIT_LIST_HEAD (¶ms.line_range.line_list); + params.show_lines = true;return 0 ; }#endif
@@ -140,7 +140,7 @@ static const struct option options[] = {OPT_STRING ('k' ,"vmlinux" , &symbol_conf.vmlinux_name,"file" ,"vmlinux pathname" ), #endif -OPT_BOOLEAN ('l' ,"list" , &session.list_events, +OPT_BOOLEAN ('l' ,"list" , ¶ms.list_events,"list up current probe events" ),OPT_CALLBACK ('d' ,"del" ,NULL ,"[GROUP:]EVENT" ,"delete a probe event." , opt_del_probe_event),
@@ -168,7 +168,7 @@ static const struct option options[] = {#endif "\t\t\tkprobe-tracer argument format.)\n", opt_add_probe_event), - OPT_BOOLEAN('f', "force", &session.force_add, "forcibly add events" + OPT_BOOLEAN('f', "force", ¶ms.force_add, "forcibly add events" " with existing name"),#ifndef NO_DWARF_SUPPORT OPT_CALLBACK('L', "line", NULL,
@@ -190,20 +190,20 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)parse_probe_event_argv (argc, argv); } -if ((!session.nr_probe && !session.dellist && !session.list_events && - !session.show_lines)) +if ((!params.nr_probe && !params.dellist && !params.list_events && + !params.show_lines))usage_with_options (probe_usage, options);if (debugfs_valid_mountpoint (debugfs_path) <0 )die ("Failed to find debugfs path." ); -if (session.list_events) { -if (session.nr_probe !=0 || session.dellist) { +if (params.list_events) { +if (params.nr_probe !=0 || params.dellist) {pr_warning (" Error: Don't use --list with" " --add/--del.\n" );usage_with_options (probe_usage, options); } -if (session.show_lines) { +if (params.show_lines) {pr_warning (" Error: Don't use --list with --line.\n" );usage_with_options (probe_usage, options); }
@@ -212,27 +212,27 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) }#ifndef NO_DWARF_SUPPORT -if (session.show_lines) { -if (session.nr_probe !=0 || session.dellist) { +if (params.show_lines) { +if (params.nr_probe !=0 || params.dellist) {pr_warning (" Error: Don't use --line with" " --add/--del.\n" );usage_with_options (probe_usage, options); } -show_line_range (&session.line_range); +show_line_range (¶ms.line_range);return 0 ; }#endif -if (session.dellist) { -del_trace_kprobe_events (session.dellist); -strlist__delete (session.dellist); -if (session.nr_probe ==0 ) +if (params.dellist) { +del_trace_kprobe_events (params.dellist); +strlist__delete (params.dellist); +if (params.nr_probe ==0 )return 0 ; } -add_trace_kprobe_events (session.probes, session.nr_probe, - session.force_add, session.need_dwarf); +add_trace_kprobe_events (params.probes, params.nr_probe, + params.force_add, params.need_dwarf);return 0 ; }
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
[tip:perf/core] perf probe: Rename session to param
Commit-ID: 12a1fadb41b5a6733c36b488b881fb19a28c92d3 Gitweb: http://git.kernel.org/tip/12a1fadb41b5a6733c36b488b881fb19a28c92d3 Author: Masami Hiramatsu <[email protected]> AuthorDate: Tue, 16 Mar 2010 18:05:44 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 11:32:30 +0100
perf probe: Rename session to param
Since this name 'session' conflicts with 'perf_session', and this structure just holds parameters anymore.
LKML-Reference: <[email protected]>
--- tools/perf/builtin-probe.c | 54 ++++++++++++++++++++++---------------------- 1 files changed, 27 insertions(+), 27 deletions(-)
tools/perf/builtin-probe.c +27 -27
@@ -56,20 +56,20 @@ static struct {struct probe_point probes[MAX_PROBES ];struct strlist *dellist;struct line_range line_range; -} session; +} params;/* Parse an event definition. Note that any error must die. */ static void parse_probe_event (const char *str) { -struct probe_point *pp = &session.probes[session.nr_probe]; +struct probe_point *pp = ¶ms.probes[params.nr_probe]; -pr_debug ("probe-definition(%d): %s\n" , session.nr_probe, str); -if (++session.nr_probe ==MAX_PROBES ) +pr_debug ("probe-definition(%d): %s\n" , params.nr_probe, str); +if (++params.nr_probe ==MAX_PROBES )die ("Too many probes (> %d) are specified." ,MAX_PROBES );/* Parse perf-probe event into probe_point */ -parse_perf_probe_event (str, pp, &session.need_dwarf); +parse_perf_probe_event (str, pp, ¶ms.need_dwarf);pr_debug ("%d arguments\n" , pp->nr_args); }
@@ -103,9 +103,9 @@ static int opt_del_probe_event(const struct option *opt __used, const char *str, int unset __used) { if (str) { - if (!session.dellist) - session.dellist = strlist__new(true, NULL); -strlist__add (session.dellist, str); + if (!params.dellist) + params.dellist = strlist__new(true, NULL); +strlist__add (params.dellist, str); }return 0 ; }
@@ -115,9 +115,9 @@ static int opt_show_lines(const struct option *opt __used, const char *str, int unset __used) { if (str) - parse_line_range_desc(str, &session.line_range); -INIT_LIST_HEAD (&session.line_range.line_list); - session.show_lines = true; + parse_line_range_desc(str, ¶ms.line_range); +INIT_LIST_HEAD (¶ms.line_range.line_list); + params.show_lines = true;return 0 ; }#endif
@@ -140,7 +140,7 @@ static const struct option options[] = {OPT_STRING ('k' ,"vmlinux" , &symbol_conf.vmlinux_name,"file" ,"vmlinux pathname" ), #endif -OPT_BOOLEAN ('l' ,"list" , &session.list_events, +OPT_BOOLEAN ('l' ,"list" , ¶ms.list_events,"list up current probe events" ),OPT_CALLBACK ('d' ,"del" ,NULL ,"[GROUP:]EVENT" ,"delete a probe event." , opt_del_probe_event),
@@ -168,7 +168,7 @@ static const struct option options[] = {#endif "\t\t\tkprobe-tracer argument format.)\n", opt_add_probe_event), - OPT_BOOLEAN('f', "force", &session.force_add, "forcibly add events" + OPT_BOOLEAN('f', "force", ¶ms.force_add, "forcibly add events" " with existing name"),#ifndef NO_DWARF_SUPPORT OPT_CALLBACK('L', "line", NULL,
@@ -190,20 +190,20 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)parse_probe_event_argv (argc, argv); } -if ((!session.nr_probe && !session.dellist && !session.list_events && - !session.show_lines)) +if ((!params.nr_probe && !params.dellist && !params.list_events && + !params.show_lines))usage_with_options (probe_usage, options);if (debugfs_valid_mountpoint (debugfs_path) <0 )die ("Failed to find debugfs path." ); -if (session.list_events) { -if (session.nr_probe !=0 || session.dellist) { +if (params.list_events) { +if (params.nr_probe !=0 || params.dellist) {pr_warning (" Error: Don't use --list with" " --add/--del.\n" );usage_with_options (probe_usage, options); } -if (session.show_lines) { +if (params.show_lines) {pr_warning (" Error: Don't use --list with --line.\n" );usage_with_options (probe_usage, options); }
@@ -212,27 +212,27 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) }#ifndef NO_DWARF_SUPPORT -if (session.show_lines) { -if (session.nr_probe !=0 || session.dellist) { +if (params.show_lines) { +if (params.nr_probe !=0 || params.dellist) {pr_warning (" Error: Don't use --line with" " --add/--del.\n" );usage_with_options (probe_usage, options); } -show_line_range (&session.line_range); +show_line_range (¶ms.line_range);return 0 ; }#endif -if (session.dellist) { -del_trace_kprobe_events (session.dellist); -strlist__delete (session.dellist); -if (session.nr_probe ==0 ) +if (params.dellist) { +del_trace_kprobe_events (params.dellist); +strlist__delete (params.dellist); +if (params.nr_probe ==0 )return 0 ; } -add_trace_kprobe_events (session.probes, session.nr_probe, - session.force_add, session.need_dwarf); +add_trace_kprobe_events (params.probes, params.nr_probe, + params.force_add, params.need_dwarf);return 0 ; }
PATCH 5/10 perf probe: Rename some die_get_* functions
In reply to Masami HiramatsuRename die_get_real_subprogram and die_get_inlinefunc to die_find_real_subprogram and die_find_inlinefunc respectively, because these functions search its children. After that, 'die_get_' means getting a property of that die, and 'die_find_' means searching DIE-tree to get an appropriate child die.
---
tools/perf/util/probe-finder.c | 16 ++++++++-------- 1 files changed, 8 insertions(+), 8 deletions(-)
tools/perf/util/probe-finder.c +8 -8
@@ -204,8 +204,8 @@ static int __die_search_func_cb(Dwarf_Die *fn_die, void *data) }/* Search a real subprogram including this line, */ -static Dwarf_Die *die_get_real_subprogram (Dwarf_Die *cu_die,Dwarf_Addr addr, -Dwarf_Die *die_mem) +static Dwarf_Die *die_find_real_subprogram (Dwarf_Die *cu_die,Dwarf_Addr addr, +Dwarf_Die *die_mem) {struct __addr_die_search_param ad; ad.addr = addr;
@@ -218,8 +218,8 @@ static Dwarf_Die *die_get_real_subprogram(Dwarf_Die *cu_die, Dwarf_Addr addr, }/* Similar to dwarf_getfuncs, but returns inlined_subroutine if exists. */ -static Dwarf_Die *die_get_inlinefunc (Dwarf_Die *sp_die,Dwarf_Addr addr, -Dwarf_Die *die_mem) +static Dwarf_Die *die_find_inlinefunc (Dwarf_Die *sp_die,Dwarf_Addr addr, +Dwarf_Die *die_mem) {Dwarf_Die child_die;int ret;
@@ -233,7 +233,7 @@ static Dwarf_Die *die_get_inlinefunc(Dwarf_Die *sp_die, Dwarf_Addr addr, dwarf_haspc(die_mem, addr))return die_mem; -if (die_get_inlinefunc (die_mem, addr, &child_die)) { +if (die_find_inlinefunc (die_mem, addr, &child_die)) {memcpy (die_mem, &child_die,sizeof (Dwarf_Die));return die_mem; }
@@ -401,7 +401,7 @@ static void show_probe_point(Dwarf_Die *sp_die, struct probe_finder *pf)/* If no real subprogram, find a real one */ if (!sp_die ||dwarf_tag (sp_die) != DW_TAG_subprogram) { - sp_die =die_get_real_subprogram (&pf->cu_die, + sp_die =die_find_real_subprogram (&pf->cu_die, pf->addr, &die_mem);if (!sp_die)die ("Probe point is not found in subprograms." );
@@ -564,7 +564,7 @@ static void find_probe_point_lazy(Dwarf_Die *sp_die, struct probe_finder *pf)if (!dwarf_haspc (sp_die, addr))continue ;/* Address filtering 2: No child include addr? */ -if (die_get_inlinefunc (sp_die, addr, &die_mem)) +if (die_find_inlinefunc (sp_die, addr, &die_mem))continue ; }
@@ -714,7 +714,7 @@ static void find_line_range_by_line(Dwarf_Die *sp_die, struct line_finder *lf)continue ;/* Address filtering 2: No child include addr? */ -if (die_get_inlinefunc (sp_die, addr, &die_mem)) +if (die_find_inlinefunc (sp_die, addr, &die_mem))continue ; }
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
[tip:perf/core] perf probe: Rename some die_get_* functions
Commit-ID: 95a3e4c4e21de1920a2ddb54bfc57c0af7e2561e Gitweb: http://git.kernel.org/tip/95a3e4c4e21de1920a2ddb54bfc57c0af7e2561e Author: Masami Hiramatsu <[email protected]> AuthorDate: Tue, 16 Mar 2010 18:05:51 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 11:32:30 +0100
perf probe: Rename some die_get_* functions
Rename die_get_real_subprogram and die_get_inlinefunc to die_find_real_subprogram and die_find_inlinefunc respectively, because these functions search its children. After that, 'die_get_' means getting a property of that die, and 'die_find_' means searching DIE-tree to get an appropriate child die.
LKML-Reference: <[email protected]>
--- tools/perf/util/probe-finder.c | 16 ++++++++-------- 1 files changed, 8 insertions(+), 8 deletions(-)
tools/perf/util/probe-finder.c +8 -8
@@ -204,8 +204,8 @@ static int __die_search_func_cb(Dwarf_Die *fn_die, void *data) }/* Search a real subprogram including this line, */ -static Dwarf_Die *die_get_real_subprogram (Dwarf_Die *cu_die,Dwarf_Addr addr, -Dwarf_Die *die_mem) +static Dwarf_Die *die_find_real_subprogram (Dwarf_Die *cu_die,Dwarf_Addr addr, +Dwarf_Die *die_mem) {struct __addr_die_search_param ad; ad.addr = addr;
@@ -218,8 +218,8 @@ static Dwarf_Die *die_get_real_subprogram(Dwarf_Die *cu_die, Dwarf_Addr addr, }/* Similar to dwarf_getfuncs, but returns inlined_subroutine if exists. */ -static Dwarf_Die *die_get_inlinefunc (Dwarf_Die *sp_die,Dwarf_Addr addr, -Dwarf_Die *die_mem) +static Dwarf_Die *die_find_inlinefunc (Dwarf_Die *sp_die,Dwarf_Addr addr, +Dwarf_Die *die_mem) {Dwarf_Die child_die;int ret;
@@ -233,7 +233,7 @@ static Dwarf_Die *die_get_inlinefunc(Dwarf_Die *sp_die, Dwarf_Addr addr, dwarf_haspc(die_mem, addr))return die_mem; -if (die_get_inlinefunc (die_mem, addr, &child_die)) { +if (die_find_inlinefunc (die_mem, addr, &child_die)) {memcpy (die_mem, &child_die,sizeof (Dwarf_Die));return die_mem; }
@@ -401,7 +401,7 @@ static void show_probe_point(Dwarf_Die *sp_die, struct probe_finder *pf)/* If no real subprogram, find a real one */ if (!sp_die ||dwarf_tag (sp_die) != DW_TAG_subprogram) { - sp_die =die_get_real_subprogram (&pf->cu_die, + sp_die =die_find_real_subprogram (&pf->cu_die, pf->addr, &die_mem);if (!sp_die)die ("Probe point is not found in subprograms." );
@@ -564,7 +564,7 @@ static void find_probe_point_lazy(Dwarf_Die *sp_die, struct probe_finder *pf)if (!dwarf_haspc (sp_die, addr))continue ;/* Address filtering 2: No child include addr? */ -if (die_get_inlinefunc (sp_die, addr, &die_mem)) +if (die_find_inlinefunc (sp_die, addr, &die_mem))continue ; }
@@ -714,7 +714,7 @@ static void find_line_range_by_line(Dwarf_Die *sp_die, struct line_finder *lf)continue ;/* Address filtering 2: No child include addr? */ -if (die_get_inlinefunc (sp_die, addr, &die_mem)) +if (die_find_inlinefunc (sp_die, addr, &die_mem))continue ; }
PATCH 6/10 perf probe: Introduce die_find_child() function
In reply to Masami HiramatsuIntroduce die_find_child() function to integrate DIE-tree searching functions.
---
tools/perf/util/probe-finder.c | 136 ++++++++++++++++++++++++---------------- 1 files changed, 80 insertions(+), 56 deletions(-)
tools/perf/util/probe-finder.c +80 -56
@@ -186,6 +186,62 @@ static const char *cu_find_realpath(Dwarf_Die *cu_die, const char *fname)return src; } +/* Compare diename and tname */ +static bool die_compare_name (Dwarf_Die *dw_die,const char *tname) +{ +const char *name; + name =dwarf_diename (dw_die); +DIE_IF (name ==NULL ); +return strcmp (tname, name); +} + +/* Get entry pc(or low pc, 1st entry of ranges) of the die */ +static Dwarf_Addr die_get_entrypc (Dwarf_Die *dw_die) +{ +Dwarf_Addr epc; +int ret; + + ret =dwarf_entrypc (dw_die, &epc); +DIE_IF (ret ==-1 ); +return epc; +} + +/* Return values for die_find callbacks */ +enum { +DIE_FIND_CB_FOUND =0 ,/* End of Search */ +DIE_FIND_CB_CHILD =1 ,/* Search only children */ +DIE_FIND_CB_SIBLING =2 ,/* Search only siblings */ +DIE_FIND_CB_CONTINUE =3 ,/* Search children and siblings */ +}; + +/* Search a child die */ +static Dwarf_Die *die_find_child (Dwarf_Die *rt_die, +int (*callback)(Dwarf_Die *,void *), +void *data,Dwarf_Die *die_mem) +{ +Dwarf_Die child_die; +int ret; + + ret =dwarf_child (rt_die, die_mem); +if (ret !=0 ) +return NULL ; + +do { + ret =callback (die_mem, data); +if (ret ==DIE_FIND_CB_FOUND ) +return die_mem; + +if ((ret &DIE_FIND_CB_CHILD ) && +die_find_child (die_mem, callback, data, &child_die)) { +memcpy (die_mem, &child_die,sizeof (Dwarf_Die)); +return die_mem; + } + }while ((ret &DIE_FIND_CB_SIBLING ) && +dwarf_siblingof (die_mem, die_mem) ==0 ); + +return NULL ; +} +struct __addr_die_search_param {Dwarf_Addr addr;Dwarf_Die *die_mem;
@@ -217,77 +273,45 @@ static Dwarf_Die *die_find_real_subprogram(Dwarf_Die *cu_die, Dwarf_Addr addr,return die_mem; } -/* Similar to dwarf_getfuncs, but returns inlined_subroutine if exists. */ -static Dwarf_Die *die_find_inlinefunc (Dwarf_Die *sp_die,Dwarf_Addr addr, -Dwarf_Die *die_mem) +/* die_find callback for inline function search */ +static int __die_find_inline_cb (Dwarf_Die *die_mem,void *data) { -Dwarf_Die child_die; -int ret; +Dwarf_Addr *addr = data; - ret =dwarf_child (sp_die, die_mem); -if (ret !=0 ) -return NULL ; - -do { -if (dwarf_tag (die_mem) == DW_TAG_inlined_subroutine && -dwarf_haspc (die_mem, addr)) -return die_mem; +if (dwarf_tag (die_mem) == DW_TAG_inlined_subroutine && +dwarf_haspc (die_mem, *addr)) +return DIE_FIND_CB_FOUND ; -if (die_find_inlinefunc (die_mem, addr, &child_die)) { -memcpy (die_mem, &child_die,sizeof (Dwarf_Die)); -return die_mem; - } - }while (dwarf_siblingof (die_mem, die_mem) ==0 ); - -return NULL ; +return DIE_FIND_CB_CONTINUE ; } -/* Compare diename and tname */ -static bool die_compare_name (Dwarf_Die *dw_die,const char *tname) +/* Similar to dwarf_getfuncs, but returns inlined_subroutine if exists. */ +static Dwarf_Die *die_find_inlinefunc (Dwarf_Die *sp_die,Dwarf_Addr addr, +Dwarf_Die *die_mem) { -const char *name; - name =dwarf_diename (dw_die); -DIE_IF (name ==NULL ); -return strcmp (tname, name); +return die_find_child (sp_die, __die_find_inline_cb, &addr, die_mem); } -/* Get entry pc(or low pc, 1st entry of ranges) of the die */ -static Dwarf_Addr die_get_entrypc (Dwarf_Die *dw_die) +static int __die_find_variable_cb (Dwarf_Die *die_mem,void *data) { -Dwarf_Addr epc; -int ret; +const char *name = data; +int tag; - ret =dwarf_entrypc (dw_die, &epc); -DIE_IF (ret ==-1 ); -return epc; + tag =dwarf_tag (die_mem); +if ((tag == DW_TAG_formal_parameter || + tag == DW_TAG_variable) && + (die_compare_name (die_mem, name) ==0 )) +return DIE_FIND_CB_FOUND ; + +return DIE_FIND_CB_CONTINUE ; } -/* Get a variable die */ +/* Find a variable called 'name' */ static Dwarf_Die *die_find_variable (Dwarf_Die *sp_die,const char *name,Dwarf_Die *die_mem) { -Dwarf_Die child_die; -int tag; -int ret; - - ret =dwarf_child (sp_die, die_mem); -if (ret !=0 ) -return NULL ; - -do { - tag =dwarf_tag (die_mem); -if ((tag == DW_TAG_formal_parameter || - tag == DW_TAG_variable) && - (die_compare_name (die_mem, name) ==0 )) -return die_mem; - -if (die_find_variable (die_mem, name, &child_die)) { -memcpy (die_mem, &child_die,sizeof (Dwarf_Die)); -return die_mem; - } - }while (dwarf_siblingof (die_mem, die_mem) ==0 ); - -return NULL ; +return die_find_child (sp_die, __die_find_variable_cb, (void *)name, + die_mem); }/*
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
[tip:perf/core] perf probe: Introduce die_find_child() function
Commit-ID: 016f262e4fb10c6ecff709317098912f94a21efa Gitweb: http://git.kernel.org/tip/016f262e4fb10c6ecff709317098912f94a21efa Author: Masami Hiramatsu <[email protected]> AuthorDate: Tue, 16 Mar 2010 18:05:58 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 11:32:31 +0100
perf probe: Introduce die_find_child() function
Introduce die_find_child() function to integrate DIE-tree searching functions.
LKML-Reference: <[email protected]>
--- tools/perf/util/probe-finder.c | 136 +++++++++++++++++++++++---------------- 1 files changed, 80 insertions(+), 56 deletions(-)
tools/perf/util/probe-finder.c +80 -56
@@ -186,6 +186,62 @@ static const char *cu_find_realpath(Dwarf_Die *cu_die, const char *fname)return src; } +/* Compare diename and tname */ +static bool die_compare_name (Dwarf_Die *dw_die,const char *tname) +{ +const char *name; + name =dwarf_diename (dw_die); +DIE_IF (name ==NULL ); +return strcmp (tname, name); +} + +/* Get entry pc(or low pc, 1st entry of ranges) of the die */ +static Dwarf_Addr die_get_entrypc (Dwarf_Die *dw_die) +{ +Dwarf_Addr epc; +int ret; + + ret =dwarf_entrypc (dw_die, &epc); +DIE_IF (ret ==-1 ); +return epc; +} + +/* Return values for die_find callbacks */ +enum { +DIE_FIND_CB_FOUND =0 ,/* End of Search */ +DIE_FIND_CB_CHILD =1 ,/* Search only children */ +DIE_FIND_CB_SIBLING =2 ,/* Search only siblings */ +DIE_FIND_CB_CONTINUE =3 ,/* Search children and siblings */ +}; + +/* Search a child die */ +static Dwarf_Die *die_find_child (Dwarf_Die *rt_die, +int (*callback)(Dwarf_Die *,void *), +void *data,Dwarf_Die *die_mem) +{ +Dwarf_Die child_die; +int ret; + + ret =dwarf_child (rt_die, die_mem); +if (ret !=0 ) +return NULL ; + +do { + ret =callback (die_mem, data); +if (ret ==DIE_FIND_CB_FOUND ) +return die_mem; + +if ((ret &DIE_FIND_CB_CHILD ) && +die_find_child (die_mem, callback, data, &child_die)) { +memcpy (die_mem, &child_die,sizeof (Dwarf_Die)); +return die_mem; + } + }while ((ret &DIE_FIND_CB_SIBLING ) && +dwarf_siblingof (die_mem, die_mem) ==0 ); + +return NULL ; +} +struct __addr_die_search_param {Dwarf_Addr addr;Dwarf_Die *die_mem;
@@ -217,77 +273,45 @@ static Dwarf_Die *die_find_real_subprogram(Dwarf_Die *cu_die, Dwarf_Addr addr,return die_mem; } -/* Similar to dwarf_getfuncs, but returns inlined_subroutine if exists. */ -static Dwarf_Die *die_find_inlinefunc (Dwarf_Die *sp_die,Dwarf_Addr addr, -Dwarf_Die *die_mem) +/* die_find callback for inline function search */ +static int __die_find_inline_cb (Dwarf_Die *die_mem,void *data) { -Dwarf_Die child_die; -int ret; +Dwarf_Addr *addr = data; - ret =dwarf_child (sp_die, die_mem); -if (ret !=0 ) -return NULL ; - -do { -if (dwarf_tag (die_mem) == DW_TAG_inlined_subroutine && -dwarf_haspc (die_mem, addr)) -return die_mem; +if (dwarf_tag (die_mem) == DW_TAG_inlined_subroutine && +dwarf_haspc (die_mem, *addr)) +return DIE_FIND_CB_FOUND ; -if (die_find_inlinefunc (die_mem, addr, &child_die)) { -memcpy (die_mem, &child_die,sizeof (Dwarf_Die)); -return die_mem; - } - }while (dwarf_siblingof (die_mem, die_mem) ==0 ); - -return NULL ; +return DIE_FIND_CB_CONTINUE ; } -/* Compare diename and tname */ -static bool die_compare_name (Dwarf_Die *dw_die,const char *tname) +/* Similar to dwarf_getfuncs, but returns inlined_subroutine if exists. */ +static Dwarf_Die *die_find_inlinefunc (Dwarf_Die *sp_die,Dwarf_Addr addr, +Dwarf_Die *die_mem) { -const char *name; - name =dwarf_diename (dw_die); -DIE_IF (name ==NULL ); -return strcmp (tname, name); +return die_find_child (sp_die, __die_find_inline_cb, &addr, die_mem); } -/* Get entry pc(or low pc, 1st entry of ranges) of the die */ -static Dwarf_Addr die_get_entrypc (Dwarf_Die *dw_die) +static int __die_find_variable_cb (Dwarf_Die *die_mem,void *data) { -Dwarf_Addr epc; -int ret; +const char *name = data; +int tag; - ret =dwarf_entrypc (dw_die, &epc); -DIE_IF (ret ==-1 ); -return epc; + tag =dwarf_tag (die_mem); +if ((tag == DW_TAG_formal_parameter || + tag == DW_TAG_variable) && + (die_compare_name (die_mem, name) ==0 )) +return DIE_FIND_CB_FOUND ; + +return DIE_FIND_CB_CONTINUE ; } -/* Get a variable die */ +/* Find a variable called 'name' */ static Dwarf_Die *die_find_variable (Dwarf_Die *sp_die,const char *name,Dwarf_Die *die_mem) { -Dwarf_Die child_die; -int tag; -int ret; - - ret =dwarf_child (sp_die, die_mem); -if (ret !=0 ) -return NULL ; - -do { - tag =dwarf_tag (die_mem); -if ((tag == DW_TAG_formal_parameter || - tag == DW_TAG_variable) && - (die_compare_name (die_mem, name) ==0 )) -return die_mem; - -if (die_find_variable (die_mem, name, &child_die)) { -memcpy (die_mem, &child_die,sizeof (Dwarf_Die)); -return die_mem; - } - }while (dwarf_siblingof (die_mem, die_mem) ==0 ); - -return NULL ; +return die_find_child (sp_die, __die_find_variable_cb, (void *)name, + die_mem); }/*
PATCH 7/10 perf probe: Add --dry-run option
In reply to Masami HiramatsuAdd --dry-run option for debugging and testing.
---
tools/perf/Documentation/perf-probe.txt | 5 +++++ tools/perf/builtin-probe.c | 1 + tools/perf/util/probe-event.c | 24 ++++++++++++++++-------- tools/perf/util/probe-event.h | 2 ++ 4 files changed, 24 insertions(+), 8 deletions(-)
tools/perf/Documentation/perf-probe.txt +5 -0
@@ -57,6 +57,11 @@ OPTIONS --force:: Forcibly add events with existing name. +-n:: +--dry-run:: + Dry run. With this option, --add and --del doesn't execute actual + adding and removal operations. + PROBE SYNTAX ------------ Probe points are defined by following syntax.
tools/perf/builtin-probe.c +1 -0
@@ -175,6 +175,7 @@ static const struct option options[] = { "FUNC[:RLN[+NUM|:RLN2]]|SRC:ALN[+NUM|:ALN2]", "Show source code lines.", opt_show_lines),#endif +OPT__DRY_RUN (&probe_event_dry_run), OPT_END() };
tools/perf/util/probe-event.c +16 -8
@@ -49,6 +49,8 @@#define MAX_PROBE_ARGS 128#define PERFPROBE_GROUP "probe" +bool probe_event_dry_run;/* Dry run flag */ + #define semantic_error(msg ...) die("Semantic error :" msg)/* If there is no space to write, returns -E2BIG. */
@@ -430,7 +432,7 @@ error:return ret; } -static int open_kprobe_events (int flags,int mode) +static int open_kprobe_events (bool readwrite) {char buf[PATH_MAX ];int ret;
@@ -439,7 +441,11 @@ static int open_kprobe_events(int flags, int mode)if (ret <0 )die ("Failed to make kprobe_events path." ); - ret =open (buf, flags, mode); +if (readwrite && !probe_event_dry_run) + ret =open (buf,O_RDWR ,O_APPEND ); +else + ret =open (buf,O_RDONLY ,0 ); +if (ret <0 ) {if (errno ==ENOENT ) die("kprobe_events file does not exist -"
@@ -535,7 +541,7 @@ void show_perf_probe_events(void)setup_pager ();memset (&pp,0 ,sizeof (pp)); - fd =open_kprobe_events (O_RDONLY ,0 ); + fd =open_kprobe_events (false); rawlist =get_trace_kprobe_event_rawlist (fd);close (fd);
@@ -585,9 +591,11 @@ static void write_trace_kprobe_event(int fd, const char *buf)int ret;pr_debug ("Writing event: %s\n" , buf); - ret =write (fd, buf,strlen (buf)); -if (ret <=0 ) -die ("Failed to write event: %s" ,strerror (errno)); +if (!probe_event_dry_run) { + ret =write (fd, buf,strlen (buf)); +if (ret <=0 ) +die ("Failed to write event: %s" ,strerror (errno)); + } } static void get_new_event_name(char *buf, size_t len, const char *base,
@@ -630,7 +638,7 @@ static void __add_trace_kprobe_events(struct probe_point *probes,struct strlist *namelist;bool allow_suffix; - fd =open_kprobe_events (O_RDWR ,O_APPEND ); + fd =open_kprobe_events (true);/* Get current event names */ namelist =get_perf_event_names (fd, false);
@@ -814,7 +822,7 @@ void del_trace_kprobe_events(struct strlist *dellist)struct str_node *ent;struct strlist *namelist; - fd =open_kprobe_events (O_RDWR ,O_APPEND ); + fd =open_kprobe_events (true);/* Get current event names */ namelist =get_perf_event_names (fd, true);
tools/perf/util/probe-event.h +2 -0
@@ -5,6 +5,8 @@#include "probe-finder.h" #include "strlist.h" +extern bool probe_event_dry_run; +extern void parse_line_range_desc (const char *arg,struct line_range *lr);extern void parse_perf_probe_event (const char *str,struct probe_point *pp,bool *need_dwarf);
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
[tip:perf/core] perf probe: Add --dry-run option
Commit-ID: f4d7da499e4fc1fdff8f26fdeb1a058d475a7a6c Gitweb: http://git.kernel.org/tip/f4d7da499e4fc1fdff8f26fdeb1a058d475a7a6c Author: Masami Hiramatsu <[email protected]> AuthorDate: Tue, 16 Mar 2010 18:06:05 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 11:32:31 +0100
perf probe: Add --dry-run option
Add --dry-run option for debugging and testing.
LKML-Reference: <[email protected]>
--- tools/perf/Documentation/perf-probe.txt | 5 +++++ tools/perf/builtin-probe.c | 1 + tools/perf/util/probe-event.c | 24 ++++++++++++++++-------- tools/perf/util/probe-event.h | 2 ++ 4 files changed, 24 insertions(+), 8 deletions(-)
tools/perf/Documentation/perf-probe.txt +5 -0
@@ -57,6 +57,11 @@ OPTIONS --force:: Forcibly add events with existing name. +-n:: +--dry-run:: + Dry run. With this option, --add and --del doesn't execute actual + adding and removal operations. + PROBE SYNTAX ------------ Probe points are defined by following syntax.
tools/perf/builtin-probe.c +1 -0
@@ -175,6 +175,7 @@ static const struct option options[] = { "FUNC[:RLN[+NUM|:RLN2]]|SRC:ALN[+NUM|:ALN2]", "Show source code lines.", opt_show_lines),#endif +OPT__DRY_RUN (&probe_event_dry_run), OPT_END() };
tools/perf/util/probe-event.c +16 -8
@@ -49,6 +49,8 @@#define MAX_PROBE_ARGS 128#define PERFPROBE_GROUP "probe" +bool probe_event_dry_run;/* Dry run flag */ + #define semantic_error(msg ...) die("Semantic error :" msg)/* If there is no space to write, returns -E2BIG. */
@@ -430,7 +432,7 @@ error:return ret; } -static int open_kprobe_events (int flags,int mode) +static int open_kprobe_events (bool readwrite) {char buf[PATH_MAX ];int ret;
@@ -439,7 +441,11 @@ static int open_kprobe_events(int flags, int mode)if (ret <0 )die ("Failed to make kprobe_events path." ); - ret =open (buf, flags, mode); +if (readwrite && !probe_event_dry_run) + ret =open (buf,O_RDWR ,O_APPEND ); +else + ret =open (buf,O_RDONLY ,0 ); +if (ret <0 ) {if (errno ==ENOENT ) die("kprobe_events file does not exist -"
@@ -535,7 +541,7 @@ void show_perf_probe_events(void)setup_pager ();memset (&pp,0 ,sizeof (pp)); - fd =open_kprobe_events (O_RDONLY ,0 ); + fd =open_kprobe_events (false); rawlist =get_trace_kprobe_event_rawlist (fd);close (fd);
@@ -585,9 +591,11 @@ static void write_trace_kprobe_event(int fd, const char *buf)int ret;pr_debug ("Writing event: %s\n" , buf); - ret =write (fd, buf,strlen (buf)); -if (ret <=0 ) -die ("Failed to write event: %s" ,strerror (errno)); +if (!probe_event_dry_run) { + ret =write (fd, buf,strlen (buf)); +if (ret <=0 ) +die ("Failed to write event: %s" ,strerror (errno)); + } } static void get_new_event_name(char *buf, size_t len, const char *base,
@@ -630,7 +638,7 @@ static void __add_trace_kprobe_events(struct probe_point *probes,struct strlist *namelist;bool allow_suffix; - fd =open_kprobe_events (O_RDWR ,O_APPEND ); + fd =open_kprobe_events (true);/* Get current event names */ namelist =get_perf_event_names (fd, false);
@@ -814,7 +822,7 @@ void del_trace_kprobe_events(struct strlist *dellist)struct str_node *ent;struct strlist *namelist; - fd =open_kprobe_events (O_RDWR ,O_APPEND ); + fd =open_kprobe_events (true);/* Get current event names */ namelist =get_perf_event_names (fd, true);
tools/perf/util/probe-event.h +2 -0
@@ -5,6 +5,8 @@#include "probe-finder.h" #include "strlist.h" +extern bool probe_event_dry_run; +extern void parse_line_range_desc (const char *arg,struct line_range *lr);extern void parse_perf_probe_event (const char *str,struct probe_point *pp,bool *need_dwarf);
PATCH 8/10 perf probe: Introduce kprobe_trace_event and perf_probe_event
In reply to Masami HiramatsuIntroduce kprobe_trace_event and perf_probe_event and replace old probe_point structure with it. probe_point structure is not enough flexible nor extensible. New data structures will help implementing further features.
---
tools/perf/builtin-probe.c | 30 +- tools/perf/util/probe-event.c | 626 ++++++++++++++++++++++++++-------------- tools/perf/util/probe-event.h | 111 ++++++- tools/perf/util/probe-finder.c | 140 ++++----- tools/perf/util/probe-finder.h | 58 +--- 5 files changed, 589 insertions(+), 376 deletions(-)
tools/perf/builtin-probe.c +14 -16
@@ -48,12 +48,11 @@/* Session management structure */ static struct { -bool need_dwarf;bool list_events;bool force_add;bool show_lines; -int nr_probe; -struct probe_point probes[MAX_PROBES ]; +int nevents; +struct perf_probe_event events[MAX_PROBES ];struct strlist *dellist;struct line_range line_range; } params;
@@ -62,16 +61,16 @@ static struct {/* Parse an event definition. Note that any error must die. */ static void parse_probe_event (const char *str) { -struct probe_point *pp = ¶ms.probes[params.nr_probe]; +struct perf_probe_event *pev = ¶ms.events[params.nevents]; -pr_debug ("probe-definition(%d): %s\n" , params.nr_probe, str); -if (++params.nr_probe ==MAX_PROBES ) +pr_debug ("probe-definition(%d): %s\n" , params.nevents, str); +if (++params.nevents ==MAX_PROBES )die ("Too many probes (> %d) are specified." ,MAX_PROBES ); -/* Parse perf-probe event into probe_point */ -parse_perf_probe_event (str, pp, ¶ms.need_dwarf); +/* Parse a perf-probe command into event */ +parse_perf_probe_command (str, pev); -pr_debug ("%d arguments\n" , pp->nr_args); +pr_debug ("%d arguments\n" , pev->nargs); }static void parse_probe_event_argv (int argc,const char **argv)
@@ -191,7 +190,7 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)parse_probe_event_argv (argc, argv); } -if ((!params.nr_probe && !params.dellist && !params.list_events && +if ((!params.nevents && !params.dellist && !params.list_events && !params.show_lines))usage_with_options (probe_usage, options);
@@ -199,7 +198,7 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)die ("Failed to find debugfs path." );if (params.list_events) { -if (params.nr_probe !=0 || params.dellist) { +if (params.nevents !=0 || params.dellist) {pr_warning (" Error: Don't use --list with" " --add/--del.\n" );usage_with_options (probe_usage, options);
@@ -214,7 +213,7 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)#ifndef NO_DWARF_SUPPORT if (params.show_lines) { -if (params.nr_probe !=0 || params.dellist) { +if (params.nevents !=0 || params.dellist) {pr_warning (" Error: Don't use --line with" " --add/--del.\n" );usage_with_options (probe_usage, options);
@@ -226,14 +225,13 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)#endif if (params.dellist) { -del_trace_kprobe_events (params.dellist); +del_perf_probe_events (params.dellist);strlist__delete (params.dellist); -if (params.nr_probe ==0 ) +if (params.nevents ==0 )return 0 ; } -add_trace_kprobe_events (params.probes, params.nr_probe, - params.force_add, params.need_dwarf); +add_perf_probe_events (params.events, params.nevents, params.force_add);return 0 ; }
tools/perf/util/probe-event.c +400 -226
873 lines not on this page · raw message
tools/perf/util/probe-event.h +100 -11
124 lines not on this page · raw message
tools/perf/util/probe-finder.c +63 -77
291 lines not on this page · raw message
tools/perf/util/probe-finder.h +12 -46
85 lines not on this page · raw message
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
[tip:perf/core] perf probe: Introduce kprobe_trace_event and perf_probe_event
Commit-ID: 4235b0454ebeefc2295ad8417e18a8761425b19e Gitweb: http://git.kernel.org/tip/4235b0454ebeefc2295ad8417e18a8761425b19e Author: Masami Hiramatsu <[email protected]> AuthorDate: Tue, 16 Mar 2010 18:06:12 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 11:32:31 +0100
perf probe: Introduce kprobe_trace_event and perf_probe_event
Introduce kprobe_trace_event and perf_probe_event and replace old probe_point structure with it. probe_point structure is not enough flexible nor extensible. New data structures will help implementing further features.
LKML-Reference: <[email protected]>
--- tools/perf/builtin-probe.c | 30 +- tools/perf/util/probe-event.c | 628 +++++++++++++++++++++++++--------------- tools/perf/util/probe-event.h | 111 +++++++- tools/perf/util/probe-finder.c | 140 ++++----- tools/perf/util/probe-finder.h | 58 +--- 5 files changed, 590 insertions(+), 377 deletions(-)
tools/perf/builtin-probe.c +14 -16
@@ -48,12 +48,11 @@/* Session management structure */ static struct { -bool need_dwarf;bool list_events;bool force_add;bool show_lines; -int nr_probe; -struct probe_point probes[MAX_PROBES ]; +int nevents; +struct perf_probe_event events[MAX_PROBES ];struct strlist *dellist;struct line_range line_range; } params;
@@ -62,16 +61,16 @@ static struct {/* Parse an event definition. Note that any error must die. */ static void parse_probe_event (const char *str) { -struct probe_point *pp = ¶ms.probes[params.nr_probe]; +struct perf_probe_event *pev = ¶ms.events[params.nevents]; -pr_debug ("probe-definition(%d): %s\n" , params.nr_probe, str); -if (++params.nr_probe ==MAX_PROBES ) +pr_debug ("probe-definition(%d): %s\n" , params.nevents, str); +if (++params.nevents ==MAX_PROBES )die ("Too many probes (> %d) are specified." ,MAX_PROBES ); -/* Parse perf-probe event into probe_point */ -parse_perf_probe_event (str, pp, ¶ms.need_dwarf); +/* Parse a perf-probe command into event */ +parse_perf_probe_command (str, pev); -pr_debug ("%d arguments\n" , pp->nr_args); +pr_debug ("%d arguments\n" , pev->nargs); }static void parse_probe_event_argv (int argc,const char **argv)
@@ -191,7 +190,7 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)parse_probe_event_argv (argc, argv); } -if ((!params.nr_probe && !params.dellist && !params.list_events && +if ((!params.nevents && !params.dellist && !params.list_events && !params.show_lines))usage_with_options (probe_usage, options);
@@ -199,7 +198,7 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)die ("Failed to find debugfs path." );if (params.list_events) { -if (params.nr_probe !=0 || params.dellist) { +if (params.nevents !=0 || params.dellist) {pr_warning (" Error: Don't use --list with" " --add/--del.\n" );usage_with_options (probe_usage, options);
@@ -214,7 +213,7 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)#ifndef NO_DWARF_SUPPORT if (params.show_lines) { -if (params.nr_probe !=0 || params.dellist) { +if (params.nevents !=0 || params.dellist) {pr_warning (" Error: Don't use --line with" " --add/--del.\n" );usage_with_options (probe_usage, options);
@@ -226,14 +225,13 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used)#endif if (params.dellist) { -del_trace_kprobe_events (params.dellist); +del_perf_probe_events (params.dellist);strlist__delete (params.dellist); -if (params.nr_probe ==0 ) +if (params.nevents ==0 )return 0 ; } -add_trace_kprobe_events (params.probes, params.nr_probe, - params.force_add, params.need_dwarf); +add_perf_probe_events (params.events, params.nevents, params.force_add);return 0 ; }
tools/perf/util/probe-event.c +400 -226
873 lines not on this page · raw message
tools/perf/util/probe-event.h +100 -11
124 lines not on this page · raw message
tools/perf/util/probe-finder.c +63 -77
291 lines not on this page · raw message
tools/perf/util/probe-finder.h +12 -46
85 lines not on this page · raw message
[tip:perf/core] perf probe: Fix !dwarf build
In reply to Masami HiramatsuCommit-ID: 3b0d516463f8deb897a55cb81e9dbbe58a2490ed Gitweb: http://git.kernel.org/tip/3b0d516463f8deb897a55cb81e9dbbe58a2490ed Author: Ingo Molnar <[email protected]> AuthorDate: Wed, 17 Mar 2010 12:13:28 +0100 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 12:13:28 +0100
perf probe: Fix !dwarf build
Fix the !drawf build.
This uses the existing NO_DWARF_SUPPORT mechanism we use for that, but it's really fragile and needs a cleanup. (in a separate patch)
1) Such uses:
#ifndef NO_DWARF_SUPPORT
are double inverted logic a'la 'not not'. Instead the flag should be called DWARF_SUPPORT.
2) Furthermore, assymetric #ifdef polluted code flow like:
if (need_dwarf)
#ifdef NO_DWARF_SUPPORT
die("Debuginfo-analysis is not supported");
#else /* !NO_DWARF_SUPPORT */
pr_debug("Some probes require debuginfo.\n");
fd = open_vmlinux();is very fragile and not acceptable. Instead of that helper functions should be created and the dwarf/no-dwarf logic should be separated more cleanly.
3) Local variable #ifdefs like this:
#ifndef NO_DWARF_SUPPORT
int fd;#endif
Are fragile as well and should be eliminated. Helper functions achieve that too.
Cc: Masami Hiramatsu <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Peter Zijlstra <[email protected]>
LKML-Reference: <[email protected]>
--- tools/perf/util/probe-event.c | 6 ++++++ 1 files changed, 6 insertions(+), 0 deletions(-)
tools/perf/util/probe-event.c +6 -0
@@ -1012,8 +1012,10 @@ end_dwarf:if (!sym)die ("Kernel symbol \'%s\' not found - probe not added." , tev->point.symbol); +#ifndef NO_DWARF_SUPPORT found:close (fd); +#endif return ntevs; }
@@ -1172,10 +1174,13 @@ void show_line_range(struct line_range *lr)unsigned int l =1 ;struct line_node *ln;FILE *fp; +#ifndef NO_DWARF_SUPPORT int fd, ret; +#endif /* Search a line range */ init_vmlinux (); +#ifndef NO_DWARF_SUPPORT fd =open_vmlinux ();if (fd <0 )die ("Could not open debuginfo file." );
@@ -1183,6 +1188,7 @@ void show_line_range(struct line_range *lr)if (ret <=0 )die ("Source line is not found.\n" );close (fd); +#endif setup_pager ();
Masami Hiramatsu
Acked-byRe: [tip:perf/core] perf probe: Fix !dwarf build
Oops, thank you,
Quoted 93 lines · tip-bot for Ingo Molnar
Commit-ID: 3b0d516463f8deb897a55cb81e9dbbe58a2490ed Gitweb: http://git.kernel.org/tip/3b0d516463f8deb897a55cb81e9dbbe58a2490ed Author: Ingo Molnar <[email protected]> AuthorDate: Wed, 17 Mar 2010 12:13:28 +0100 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 12:13:28 +0100perf probe: Fix !dwarf build
Fix the !drawf build.
This uses the existing
NO_DWARF_SUPPORTmechanism we use for that, but it's really fragile and needs a cleanup. (in a separate patch)1) Such uses:
#ifndef
NO_DWARF_SUPPORTare double inverted logic a'la 'not not'. Instead the flag should be called
DWARF_SUPPORT.2) Furthermore, assymetric #ifdef polluted code flow like:
if (
need_dwarf)#ifdef NO_DWARF_SUPPORT die("Debuginfo-analysis is not supported"); #else /* !NO_DWARF_SUPPORT */ pr_debug("Some probes require debuginfo.\n"); fd = open_vmlinux();is very fragile and not acceptable. Instead of that helper functions should be created and the dwarf/no-dwarf logic should be separated more cleanly.
3) Local variable #ifdefs like this:
#ifndef NO_DWARF_SUPPORT int fd;#endif
Are fragile as well and should be eliminated. Helper functions achieve that too.
Cc: Masami Hiramatsu <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Peter Zijlstra <[email protected]>
LKML-Reference: <[email protected]>Signed-off-by Ingo Molnar <[email protected]>--- tools/perf/util/probe-event.c | 6 ++++++ 1 files changed, 6 insertions(+), 0 deletions(-)tools/perf/util/probe-event.c +6
-0@@ -1012,8 +1012,10 @@ end_dwarf:if (!sym)die ("Kernel symbol \'%s\' not found - probe not added." , tev->point.symbol); +#ifndef NO_DWARF_SUPPORT found:close (fd); +#endif return ntevs; }@@ -1172,10 +1174,13 @@ void show_line_range(struct line_range *lr)unsigned int l =1 ;struct line_node *ln;FILE *fp; +#ifndef NO_DWARF_SUPPORT int fd, ret; +#endif /* Search a line range */ init_vmlinux (); +#ifndef NO_DWARF_SUPPORT fd =open_vmlinux ();if (fd <0 )die ("Could not open debuginfo file." );@@ -1183,6 +1188,7 @@ void show_line_range(struct line_range *lr)if (ret <=0 )die ("Source line is not found.\n" );close (fd); +#endif setup_pager ();
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
[tip:perf/core] perf events: Fix false positive build warning with older GCC's
In reply to Masami HiramatsuCommit-ID: 55632770d7298835645489828af87f854c47749c Gitweb: http://git.kernel.org/tip/55632770d7298835645489828af87f854c47749c Author: Ingo Molnar <[email protected]> AuthorDate: Thu, 18 Mar 2010 16:51:16 +0100 Committer: Ingo Molnar <[email protected]> CommitDate: Thu, 18 Mar 2010 17:03:24 +0100
perf events: Fix false positive build warning with older GCC's
gcc 4.2.1 produces:
util/probe-event.c: In function 'add_perf_probe_events':
util/probe-event.c:883: warning: 'tev' may be used uninitialized in this function
make: *** [util/probe-event.o] Error 1
Newer GCCs get this right.
To work it around, initialize the variable to NULL so that older GCCs see it as initialized too.
Cc: Masami Hiramatsu <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Peter Zijlstra <[email protected]>
LKML-Reference: <[email protected]>
--- tools/perf/util/probe-event.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)
tools/perf/util/probe-event.c +1 -1
@@ -880,7 +880,7 @@ static void __add_kprobe_trace_events(struct perf_probe_event *pev, int ntevs, bool allow_suffix) { int i, fd; -struct kprobe_trace_event *tev; +struct kprobe_trace_event *tev =NULL ;char buf[64 ];const char *event, *group;struct strlist *namelist;
Re: [tip:perf/core] perf events: Fix false positive build warning with older GCC's
Quoted 11 lines · tip-bot for Ingo Molnar
Commit-ID: 55632770d7298835645489828af87f854c47749c Gitweb: http://git.kernel.org/tip/55632770d7298835645489828af87f854c47749c Author: Ingo Molnar <[email protected]> AuthorDate: Thu, 18 Mar 2010 16:51:16 +0100 Committer: Ingo Molnar <[email protected]> CommitDate: Thu, 18 Mar 2010 17:03:24 +0100perf events: Fix false positive build warning with older GCC's
gcc 4.2.1 produces:
Quoted 8 lines
util/probe-event.c: In function 'add_perf_probe_events':
util/probe-event.c:883: warning: 'tev' may be used uninitialized in this function
make: *** [util/probe-event.o] Error 1Newer GCCs get this right.
To work it around, initialize the variable to NULL so that older GCCs see it as initialized too.
Hmm, indeed. this code implies that ntevs > 0, it seems that we'd better check it first (e.g. 'if(ntevs <= 0) return;') too.
Thank you,
Quoted 31 lines
Cc: Masami Hiramatsu <[email protected]>
Cc: Frederic Weisbecker <[email protected]>
Cc: Arnaldo Carvalho de Melo <[email protected]>
Cc: Paul Mackerras <[email protected]>
Cc: Mike Galbraith <[email protected]>
Cc: Peter Zijlstra <[email protected]>
LKML-Reference: <[email protected]>Signed-off-by Ingo Molnar <[email protected]>--- tools/perf/util/probe-event.c | 2 +- 1 files changed, 1 insertions(+), 1 deletions(-)tools/perf/util/probe-event.c +1
-2@@ -880,7 +880,7 @@ static void __add_kprobe_trace_events(struct perf_probe_event *pev, int ntevs, bool allow_suffix) { int i, fd; -struct kprobe_trace_event *tev; +struct kprobe_trace_event *tev =NULL ;char buf[64 ];const char *event, *group;struct strlist *namelist; --To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
PATCH 9/10 perf probe: List probes with line number and file name
In reply to Masami HiramatsuImprove --list to show current exist probes with line number and file name. This enables user easily to check which line is already probed.
e.g.
./perf probe --list
probe:vfs_read (on vfs_read:[email protected]/fs/read_write.c)
---
tools/perf/util/probe-event.c | 55 ++++++++++++++++++++++++------- tools/perf/util/probe-finder.c | 70 ++++++++++++++++++++++++++++++++++++++++ tools/perf/util/probe-finder.h | 4 ++ 3 files changed, 116 insertions(+), 13 deletions(-)
tools/perf/util/probe-event.c +42 -13
@@ -70,7 +70,6 @@ static int e_snprintf(char *str, size_t size, const char *format, ...)return ret; } -static struct map_groups kmap_groups;static struct map *kmaps[MAP__NR_TYPES ];
@@ -357,27 +356,39 @@ void parse_kprobe_trace_command(const char *cmd, struct kprobe_trace_event *tev)/* Compose only probe point (not argument) */ static char *synthesize_perf_probe_point (struct perf_probe_point *pp) { -char *buf; -char offs[64 ] ="" , line[64 ] ="" ; -int ret; +char *buf, *tmp; +char offs[32 ] ="" , line[32 ] ="" , file[32 ] ="" ; +int ret, len; buf =xzalloc (MAX_CMDLEN );if (pp->offset) { - ret =e_snprintf (offs,64 ,"+%lu" , pp->offset); + ret =e_snprintf (offs,32 ,"+%lu" , pp->offset);if (ret <=0 ) goto error; }if (pp->line) { - ret =e_snprintf (line,64 ,":%d" , pp->line); + ret =e_snprintf (line,32 ,":%d" , pp->line); +if (ret <=0 ) + goto error; + } +if (pp->file) { + len =strlen (pp->file) -32 ; +if (len <0 ) + len =0 ; + tmp =strchr (pp->file + len,'/' ); +if (!tmp) + tmp = pp->file + len -1 ; + ret =e_snprintf (file,32 ,"@%s" , tmp +1 );if (ret <=0 ) goto error; }if (pp->function) - ret =e_snprintf (buf,MAX_CMDLEN ,"%s%s%s%s" , pp->function, - offs, pp->retprobe ?"%return" :"" , line); + ret =e_snprintf (buf,MAX_CMDLEN ,"%s%s%s%s%s" , pp->function, + offs, pp->retprobe ?"%return" :"" , line, + file);else - ret =e_snprintf (buf,MAX_CMDLEN ,"%s%s" , pp->file, line); + ret =e_snprintf (buf,MAX_CMDLEN ,"%s%s" , file, line);if (ret <=0 ) goto error;
@@ -511,15 +522,32 @@ void convert_to_perf_probe_event(struct kprobe_trace_event *tev, {char buf[64 ];int i; +#ifndef NO_DWARF_SUPPORT +struct symbol *sym; +int fd, ret =0 ; - pev->event =xstrdup (tev->event); - pev->group =xstrdup (tev->group); - + sym =map__find_symbol_by_name (kmaps[MAP__FUNCTION ], + tev->point.symbol,NULL ); +if (sym) { + fd =open_vmlinux (); + ret =find_perf_probe_point (fd, sym->start + tev->point.offset, + &pev->point); +close (fd); + } +if (ret <=0 ) { + pev->point.function =xstrdup (tev->point.symbol); + pev->point.offset = tev->point.offset; + } +#else /* Convert trace_point to probe_point */ pev->point.function =xstrdup (tev->point.symbol); pev->point.offset = tev->point.offset; +#endif pev->point.retprobe = tev->point.retprobe; + pev->event =xstrdup (tev->event); + pev->group =xstrdup (tev->group); +/* Convert trace_arg to probe_arg */ pev->nargs = tev->nargs; pev->args =xzalloc (sizeof (struct perf_probe_arg ) * pev->nargs);
@@ -650,7 +678,7 @@ static void show_perf_probe_event(struct perf_probe_event *pev) ret =e_snprintf (buf,128 ,"%s:%s" , pev->group, pev->event);if (ret <0 )die ("Failed to copy event: %s" ,strerror (-ret)); -printf (" %-40s (on %s" , buf, place); +printf (" %-20s (on %s" , buf, place);if (pev->nargs >0 ) {printf (" with" );
@@ -671,6 +699,7 @@ void show_perf_probe_events(void)struct str_node *ent;setup_pager (); +init_vmlinux ();memset (&tev,0 ,sizeof (tev));memset (&pev,0 ,sizeof (pev));
tools/perf/util/probe-finder.c +70 -0
76 lines not on this page · raw message
tools/perf/util/probe-finder.h +4 -0
@@ -20,6 +20,10 @@ static inline int is_c_varname(const char *name)extern int find_kprobe_trace_events (int fd,struct perf_probe_event *pev,struct kprobe_trace_event **tevs); +/* Find a perf_probe_point from debuginfo */ +extern int find_perf_probe_point (int fd,unsigned long addr, +struct perf_probe_point *ppt); +extern int find_line_range (int fd,struct line_range *lr);#include <dwarf.h>
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
[tip:perf/core] perf probe: List probes with line number and file name
Commit-ID: fb1587d869a399554220e166d4b90b581a8ade01 Gitweb: http://git.kernel.org/tip/fb1587d869a399554220e166d4b90b581a8ade01 Author: Masami Hiramatsu <[email protected]> AuthorDate: Tue, 16 Mar 2010 18:06:19 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 11:32:32 +0100
perf probe: List probes with line number and file name
Improve --list to show current exist probes with line number and file name. This enables user easily to check which line is already probed.
for example:
./perf probe --list
probe:vfs_read (on vfs_read:[email protected]/fs/read_write.c)
LKML-Reference: <[email protected]>
--- tools/perf/util/probe-event.c | 57 ++++++++++++++++++++++++-------- tools/perf/util/probe-finder.c | 70 ++++++++++++++++++++++++++++++++++++++++ tools/perf/util/probe-finder.h | 4 ++ 3 files changed, 117 insertions(+), 14 deletions(-)
tools/perf/util/probe-event.c +42 -13
@@ -70,7 +70,6 @@ static int e_snprintf(char *str, size_t size, const char *format, ...)return ret; } -static struct map_groups kmap_groups;static struct map *kmaps[MAP__NR_TYPES ];
@@ -357,27 +356,39 @@ void parse_kprobe_trace_command(const char *cmd, struct kprobe_trace_event *tev)/* Compose only probe point (not argument) */ static char *synthesize_perf_probe_point (struct perf_probe_point *pp) { -char *buf; -char offs[64 ] ="" , line[64 ] ="" ; -int ret; +char *buf, *tmp; +char offs[32 ] ="" , line[32 ] ="" , file[32 ] ="" ; +int ret, len; buf =xzalloc (MAX_CMDLEN );if (pp->offset) { - ret =e_snprintf (offs,64 ,"+%lu" , pp->offset); + ret =e_snprintf (offs,32 ,"+%lu" , pp->offset);if (ret <=0 ) goto error; }if (pp->line) { - ret =e_snprintf (line,64 ,":%d" , pp->line); + ret =e_snprintf (line,32 ,":%d" , pp->line); +if (ret <=0 ) + goto error; + } +if (pp->file) { + len =strlen (pp->file) -32 ; +if (len <0 ) + len =0 ; + tmp =strchr (pp->file + len,'/' ); +if (!tmp) + tmp = pp->file + len -1 ; + ret =e_snprintf (file,32 ,"@%s" , tmp +1 );if (ret <=0 ) goto error; }if (pp->function) - ret =e_snprintf (buf,MAX_CMDLEN ,"%s%s%s%s" , pp->function, - offs, pp->retprobe ?"%return" :"" , line); + ret =e_snprintf (buf,MAX_CMDLEN ,"%s%s%s%s%s" , pp->function, + offs, pp->retprobe ?"%return" :"" , line, + file);else - ret =e_snprintf (buf,MAX_CMDLEN ,"%s%s" , pp->file, line); + ret =e_snprintf (buf,MAX_CMDLEN ,"%s%s" , file, line);if (ret <=0 ) goto error;
@@ -511,15 +522,32 @@ void convert_to_perf_probe_event(struct kprobe_trace_event *tev, {char buf[64 ];int i; +#ifndef NO_DWARF_SUPPORT +struct symbol *sym; +int fd, ret =0 ; - pev->event =xstrdup (tev->event); - pev->group =xstrdup (tev->group); - + sym =map__find_symbol_by_name (kmaps[MAP__FUNCTION ], + tev->point.symbol,NULL ); +if (sym) { + fd =open_vmlinux (); + ret =find_perf_probe_point (fd, sym->start + tev->point.offset, + &pev->point); +close (fd); + } +if (ret <=0 ) { + pev->point.function =xstrdup (tev->point.symbol); + pev->point.offset = tev->point.offset; + } +#else /* Convert trace_point to probe_point */ pev->point.function =xstrdup (tev->point.symbol); pev->point.offset = tev->point.offset; +#endif pev->point.retprobe = tev->point.retprobe; + pev->event =xstrdup (tev->event); + pev->group =xstrdup (tev->group); +/* Convert trace_arg to probe_arg */ pev->nargs = tev->nargs; pev->args =xzalloc (sizeof (struct perf_probe_arg ) * pev->nargs);
@@ -650,7 +678,7 @@ static void show_perf_probe_event(struct perf_probe_event *pev) ret =e_snprintf (buf,128 ,"%s:%s" , pev->group, pev->event);if (ret <0 )die ("Failed to copy event: %s" ,strerror (-ret)); -printf (" %-40s (on %s" , buf, place); +printf (" %-20s (on %s" , buf, place);if (pev->nargs >0 ) {printf (" with" );
@@ -671,6 +699,7 @@ void show_perf_probe_events(void)struct str_node *ent;setup_pager (); +init_vmlinux ();memset (&tev,0 ,sizeof (tev));memset (&pev,0 ,sizeof (pev));
tools/perf/util/probe-finder.c +70 -0
76 lines not on this page · raw message
tools/perf/util/probe-finder.h +4 -0
@@ -20,6 +20,10 @@ static inline int is_c_varname(const char *name)extern int find_kprobe_trace_events (int fd,struct perf_probe_event *pev,struct kprobe_trace_event **tevs); +/* Find a perf_probe_point from debuginfo */ +extern int find_perf_probe_point (int fd,unsigned long addr, +struct perf_probe_point *ppt); +extern int find_line_range (int fd,struct line_range *lr);#include <dwarf.h>
PATCH 10/10 perf probe: Accessing members in data structures
In reply to Masami HiramatsuSupport accessing members in the data structures. With this, perf-probe accepts data-structure members(IOW, it now accepts dot '.' and arrow '->' operators) as probe arguemnts.
e.g.
./perf probe --add 'schedule:44 rq->curr'
./perf probe --add 'vfs_read file->f_op->read file->f_path.dentry'
Note that '>' can be interpreted as redirection in command-line.
---
tools/perf/util/probe-event.c | 90 +++++++++++++++++++++++++++++++++- tools/perf/util/probe-event.h | 12 ++++- tools/perf/util/probe-finder.c | 105 ++++++++++++++++++++++++++++++++++++++++ 3 files changed, 201 insertions(+), 6 deletions(-)
tools/perf/util/probe-event.c +86 -4
@@ -262,6 +262,49 @@ static void parse_perf_probe_point(char *arg, struct perf_probe_event *pev) pp->lazy_line); } +/* Parse perf-probe event argument */ +static void parse_perf_probe_arg (const char *str,struct perf_probe_arg *arg) +{ +const char *tmp; +struct perf_probe_arg_field **fieldp; + +pr_debug ("parsing arg: %s into " , str); + + tmp =strpbrk (str,"-." ); +if (!is_c_varname (str) || !tmp) { +/* A variable, register, symbol or special value */ + arg->name =xstrdup (str); +pr_debug ("%s\n" , arg->name); +return ; + } + +/* Structure fields */ + arg->name =xstrndup (str, tmp - str); +pr_debug ("%s, " , arg->name); + fieldp = &arg->field; + +do { + *fieldp =xzalloc (sizeof (struct perf_probe_arg_field )); +if (*tmp =='.' ) { + str = tmp +1 ; + (*fieldp)->ref = false; + }else if (tmp[1 ] =='>' ) { + str = tmp +2 ; + (*fieldp)->ref = true; + }else +semantic_error ("Argument parse error: %s" , str); + + tmp =strpbrk (str,"-." ); +if (tmp) { + (*fieldp)->name =xstrndup (str, tmp - str); +pr_debug ("%s(%d), " , (*fieldp)->name, (*fieldp)->ref); + fieldp = &(*fieldp)->next; + } + }while (tmp); + (*fieldp)->name =xstrdup (str); +pr_debug ("%s(%d)\n" , (*fieldp)->name, (*fieldp)->ref); +} +/* Parse perf-probe event command */ void parse_perf_probe_command (const char *cmd,struct perf_probe_event *pev) {
@@ -281,7 +324,7 @@ void parse_perf_probe_command(const char *cmd, struct perf_probe_event *pev) pev->nargs = argc -1 ; pev->args =xzalloc (sizeof (struct perf_probe_arg ) * pev->nargs);for (i =0 ; i < pev->nargs; i++) { - pev->args[i].name =xstrdup (argv[i +1 ]); +parse_perf_probe_arg (argv[i +1 ], &pev->args[i]);if (is_c_varname (pev->args[i].name) && pev->point.retprobe)semantic_error ("You can't specify local variable for" " kretprobe" );
@@ -353,6 +396,33 @@ void parse_kprobe_trace_command(const char *cmd, struct kprobe_trace_event *tev)argv_free (argv); } +/* Compose only probe arg */ +int synthesize_perf_probe_arg (struct perf_probe_arg *pa,char *buf,size_t len) +{ +struct perf_probe_arg_field *field = pa->field; +int ret; +char *tmp = buf; + + ret =e_snprintf (tmp, len,"%s" , pa->name); +if (ret <=0 ) + goto error; + tmp += ret; + len -= ret; + +while (field) { + ret =e_snprintf (tmp, len,"%s%s" , field->ref ?"->" :"." , + field->name); +if (ret <=0 ) + goto error; + tmp += ret; + len -= ret; + field = field->next; + } +return tmp - buf; +error: +die ("Failed to synthesize perf probe argument: %s" ,strerror (-ret)); +} +/* Compose only probe point (not argument) */ static char *synthesize_perf_probe_point (struct perf_probe_point *pp) {
@@ -563,6 +633,7 @@ void convert_to_perf_probe_event(struct kprobe_trace_event *tev,void clear_perf_probe_event (struct perf_probe_event *pev) {struct perf_probe_point *pp = &pev->point; +struct perf_probe_arg_field *field, *next;int i;if (pev->event)
@@ -575,9 +646,18 @@ void clear_perf_probe_event(struct perf_probe_event *pev)free (pp->function);if (pp->lazy_line)free (pp->lazy_line); -for (i =0 ; i < pev->nargs; i++) +for (i =0 ; i < pev->nargs; i++) {if (pev->args[i].name)free (pev->args[i].name); + field = pev->args[i].field; +while (field) { + next = field->next; +if (field->name) +free (field->name); +free (field); + field = next; + } + }if (pev->args)free (pev->args);memset (pev,0 ,sizeof (*pev));
@@ -682,8 +762,10 @@ static void show_perf_probe_event(struct perf_probe_event *pev)if (pev->nargs >0 ) {printf (" with" ); -for (i =0 ; i < pev->nargs; i++) -printf (" %s" , pev->args[i].name); +for (i =0 ; i < pev->nargs; i++) { +synthesize_perf_probe_arg (&pev->args[i], buf,128 ); +printf (" %s" , buf); + } }printf (")\n" );free (place);
tools/perf/util/probe-event.h +11 -1
@@ -45,9 +45,17 @@ struct perf_probe_point {bool retprobe;/* Return probe flag */ }; +/* Perf probe probing argument field chain */ +struct perf_probe_arg_field { +struct perf_probe_arg_field *next;/* Next field */ +char *name;/* Name of the field */ +bool ref;/* Referencing flag */ +}; +/* Perf probe probing argument */ struct perf_probe_arg { -char *name;/* Argument name */ +char *name;/* Argument name */ +struct perf_probe_arg_field *field;/* Structure fields */ };/* Perf probe probing event (point + arg) */
@@ -86,6 +94,8 @@ extern void parse_kprobe_trace_command(const char *cmd,/* Events to command string */ extern char *synthesize_perf_probe_command (struct perf_probe_event *pev);extern char *synthesize_kprobe_trace_command (struct kprobe_trace_event *tev); +extern int synthesize_perf_probe_arg (struct perf_probe_arg *pa,char *buf, +size_t len);/* Check the perf_probe_event needs debuginfo */ extern bool perf_probe_event_need_dwarf (struct perf_probe_event *pev);
tools/perf/util/probe-finder.c +104 -1
141 lines not on this page · raw message
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
Re: [PATCH -tip 10/10] perf probe: Accessing members in data structures
Quoted 11 lines · Masami Hiramatsu
Support accessing members in the data structures. With this, perf-probe accepts data-structure
members(IOW, it now accepts dot '.' and arrow '->' operators)as probe arguemnts.e.g.
./perf probe --add 'schedule:44
rq->curr'./perf probe --add '
vfs_readfile->f_op->readfile->f_path.dentry'Note that '>' can be interpreted as redirection in command-line.
If you find that a problem then you can do like SystemTap does and allow '.' in place of '->'. In the code you already use the perf_probe_arg_field ref flag only to check that the DIE gives you the same information. So you could just drop that and use any separator. Then you decide based on whether you see a DW_TAG_pointer_type. This gives the user some extra flexibility by letting them not having to care about specifying extra type information already available elsewhere.
Cheers,
Mark
Re: [PATCH -tip 10/10] perf probe: Accessing members in data structures
Quoted 13 lines · Mark Wielaard
Quoted 11 lines · Masami Hiramatsu
Support accessing members in the data structures. With this, perf-probe accepts data-structure
members(IOW, it now accepts dot '.' and arrow '->' operators)as probe arguemnts.e.g.
./perf probe --add 'schedule:44
rq->curr'./perf probe --add '
vfs_readfile->f_op->readfile->f_path.dentry'Note that '>' can be interpreted as redirection in command-line.
Quoted 7 lines
If you find that a problem then you can do like SystemTap does and allow '.' in place of '->'. In the code you already use the
perf_probe_arg_fieldref flag only to check that the DIE gives you the same information. So you could just drop that and use any separator. Then you decide based on whether you see aDW_TAG_pointer_type. This gives the user some extra flexibility by letting them not having to care about specifying extra type information already available elsewhere.
Thanks, when designing this feature, I considered it too.
Since perf probe already support displaying source code by --line option, users will read the probed code itself and try to probe it. In that case, I think they naturally use '.' and '->' as they read (they might try to copy & paste it).
So, I think that it would be good to support both of '.' and '->' as they are used in the code, because it will not confuse users.
Thank you,
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
Re: [PATCH -tip 10/10] perf probe: Accessing members in data structures
Quoted 29 lines · Masami Hiramatsu
Quoted 13 lines · Mark Wielaard
Quoted 11 lines · Masami Hiramatsu
Support accessing members in the data structures. With this, perf-probe accepts data-structure
members(IOW, it now accepts dot '.' and arrow '->' operators)as probe arguemnts.e.g.
./perf probe --add 'schedule:44
rq->curr'./perf probe --add '
vfs_readfile->f_op->readfile->f_path.dentry'Note that '>' can be interpreted as redirection in command-line.
Quoted 7 lines
If you find that a problem then you can do like SystemTap does and allow '.' in place of '->'. In the code you already use the
perf_probe_arg_fieldref flag only to check that the DIE gives you the same information. So you could just drop that and use any separator. Then you decide based on whether you see aDW_TAG_pointer_type. This gives the user some extra flexibility by letting them not having to care about specifying extra type information already available elsewhere.Thanks, when designing this feature, I considered it too.
Since perf probe already support displaying source code by --line option, users will read the probed code itself and try to probe it. In that case, I think they naturally use '.' and '->' as they read (they might try to copy & paste it).
Quoted 4 lines
So, I think that it would be good to support both of '.' and '->' as they are used in the code, because it will not confuse users.
Thank you,
Agreed.
And lets people use what is common for them: expressions that follow C rules in the context.
And those who will be more familiar with perf probe will know they can use the simplified "." based scheme.
Re: [PATCH -tip 10/10] perf probe: Accessing members in data structures
Quoted 36 lines · Frederic Weisbecker
Quoted 29 lines · Masami Hiramatsu
Quoted 13 lines · Mark Wielaard
Quoted 11 lines · Masami Hiramatsu
Support accessing members in the data structures. With this, perf-probe accepts data-structure
members(IOW, it now accepts dot '.' and arrow '->' operators)as probe arguemnts.e.g.
./perf probe --add 'schedule:44
rq->curr'./perf probe --add '
vfs_readfile->f_op->readfile->f_path.dentry'Note that '>' can be interpreted as redirection in command-line.
Quoted 7 lines
If you find that a problem then you can do like SystemTap does and allow '.' in place of '->'. In the code you already use the
perf_probe_arg_fieldref flag only to check that the DIE gives you the same information. So you could just drop that and use any separator. Then you decide based on whether you see aDW_TAG_pointer_type. This gives the user some extra flexibility by letting them not having to care about specifying extra type information already available elsewhere.Thanks, when designing this feature, I considered it too.
Since perf probe already support displaying source code by --line option, users will read the probed code itself and try to probe it. In that case, I think they naturally use '.' and '->' as they read (they might try to copy & paste it).
Quoted 4 lines
So, I think that it would be good to support both of '.' and '->' as they are used in the code, because it will not confuse users.
Thank you,
Quoted 7 lines
Agreed.
And lets people use what is common for them: expressions that follow C rules in the context.
And those who will be more familiar with perf probe will know they can use the simplified "." based scheme.
Hi Frederic, Steven,
BTW, currently perf trace can't parse these 'data structure' type field yet.
e.g.
execute following commands;
./perf probe vfs_read 'file->f_mode'
./perf record -e probe:vfs_read -afR ls -l ./perf trace
Warning: Error: expected ';' but read '->'
Warning: failed to read event format for vfs_read
Because perf trace expects that the field name should be a C-variable name. Moreover, if perf probe supports array element, it becomes more problematic, e.g. 'field:u8 array[8]'! oops! what does the field contain? an array with 8 elements? or 8th element of 'array' array? :(
Should I suppose that trace event format only accept C-style variable name for field?
Thank you,
Signature · 2 lines
Masami Hiramatsu e-mail: [email protected]
Peter Zijlstra
maintainerRe: [PATCH -tip 10/10] perf probe: Accessing members in data structures
In reply to Frederic WeisbeckerQuoted 36 lines · Frederic Weisbecker
Quoted 29 lines · Masami Hiramatsu
Quoted 13 lines · Mark Wielaard
Quoted 11 lines · Masami Hiramatsu
Support accessing members in the data structures. With this, perf-probe accepts data-structure
members(IOW, it now accepts dot '.' and arrow '->' operators)as probe arguemnts.e.g.
./perf probe --add 'schedule:44
rq->curr'./perf probe --add '
vfs_readfile->f_op->readfile->f_path.dentry'Note that '>' can be interpreted as redirection in command-line.
Quoted 7 lines
If you find that a problem then you can do like SystemTap does and allow '.' in place of '->'. In the code you already use the
perf_probe_arg_fieldref flag only to check that the DIE gives you the same information. So you could just drop that and use any separator. Then you decide based on whether you see aDW_TAG_pointer_type. This gives the user some extra flexibility by letting them not having to care about specifying extra type information already available elsewhere.Thanks, when designing this feature, I considered it too.
Since perf probe already support displaying source code by --line option, users will read the probed code itself and try to probe it. In that case, I think they naturally use '.' and '->' as they read (they might try to copy & paste it).
Quoted 4 lines
So, I think that it would be good to support both of '.' and '->' as they are used in the code, because it will not confuse users.
Thank you,
Quoted 7 lines
Agreed.
And lets people use what is common for them: expressions that follow C rules in the context.
And those who will be more familiar with perf probe will know they can use the simplified "." based scheme.
I'd expect a syntax error when I mix up '.' and '->'.
Re: [PATCH -tip 10/10] perf probe: Accessing members in data structures
Em Tue, Mar 23, 2010 at 04:55:13PM +0100, Peter Zijlstra escreveu:
Quoted 21 lines · Frederic Weisbecker
Quoted 19 lines · Masami Hiramatsu
Quoted 18 lines
Quoted 7 lines · Mark Wielaard
If you find that a problem then you can do like SystemTap does and allow '.' in place of '->'. In the code you already use the
perf_probe_arg_fieldref flag only to check that the DIE gives you the same information. So you could just drop that and use any separator. Then you decide based on whether you see aDW_TAG_pointer_type. This gives the user some extra flexibility by letting them not having to care about specifying extra type information already available elsewhere.Thanks, when designing this feature, I considered it too.
Since perf probe already support displaying source code by --line option, users will read the probed code itself and try to probe it. In that case, I think they naturally use '.' and '->' as they read (they might try to copy & paste it).
So, I think that it would be good to support both of '.' and '->' as they are used in the code, because it will not confuse users.
Quoted 7 lines
Quoted 5 lines
And lets people use what is common for them: expressions that follow C rules in the context.
And those who will be more familiar with perf probe will know they can use the simplified "." based scheme.
I'd expect a syntax error when I mix up '.' and '->'.
Yup, I'd stick with following the C language rules.
- Arnaldo
[tip:perf/core] perf probe: Add data structure member access support
In reply to Masami HiramatsuCommit-ID: 7df2f32956cf0f1a45df38cd0e0fe0c3467580e8 Gitweb: http://git.kernel.org/tip/7df2f32956cf0f1a45df38cd0e0fe0c3467580e8 Author: Masami Hiramatsu <[email protected]> AuthorDate: Tue, 16 Mar 2010 18:06:26 -0400 Committer: Ingo Molnar <[email protected]> CommitDate: Wed, 17 Mar 2010 12:11:15 +0100
perf probe: Add data structure member access support
Support accessing members in the data structures. With this, perf-probe accepts data-structure members(IOW, it now accepts dot '.' and arrow '->' operators) as probe arguemnts.
e.g.
./perf probe --add 'schedule:44 rq->curr'
./perf probe --add 'vfs_read file->f_op->read file->f_path.dentry'
Note that '>' can be interpreted as redirection in command-line.
LKML-Reference: <[email protected]>
--- tools/perf/util/probe-event.c | 90 ++++++++++++++++++++++++++++++++-- tools/perf/util/probe-event.h | 12 ++++- tools/perf/util/probe-finder.c | 105 +++++++++++++++++++++++++++++++++++++++- 3 files changed, 201 insertions(+), 6 deletions(-)
tools/perf/util/probe-event.c +86 -4
@@ -262,6 +262,49 @@ static void parse_perf_probe_point(char *arg, struct perf_probe_event *pev) pp->lazy_line); } +/* Parse perf-probe event argument */ +static void parse_perf_probe_arg (const char *str,struct perf_probe_arg *arg) +{ +const char *tmp; +struct perf_probe_arg_field **fieldp; + +pr_debug ("parsing arg: %s into " , str); + + tmp =strpbrk (str,"-." ); +if (!is_c_varname (str) || !tmp) { +/* A variable, register, symbol or special value */ + arg->name =xstrdup (str); +pr_debug ("%s\n" , arg->name); +return ; + } + +/* Structure fields */ + arg->name =xstrndup (str, tmp - str); +pr_debug ("%s, " , arg->name); + fieldp = &arg->field; + +do { + *fieldp =xzalloc (sizeof (struct perf_probe_arg_field )); +if (*tmp =='.' ) { + str = tmp +1 ; + (*fieldp)->ref = false; + }else if (tmp[1 ] =='>' ) { + str = tmp +2 ; + (*fieldp)->ref = true; + }else +semantic_error ("Argument parse error: %s" , str); + + tmp =strpbrk (str,"-." ); +if (tmp) { + (*fieldp)->name =xstrndup (str, tmp - str); +pr_debug ("%s(%d), " , (*fieldp)->name, (*fieldp)->ref); + fieldp = &(*fieldp)->next; + } + }while (tmp); + (*fieldp)->name =xstrdup (str); +pr_debug ("%s(%d)\n" , (*fieldp)->name, (*fieldp)->ref); +} +/* Parse perf-probe event command */ void parse_perf_probe_command (const char *cmd,struct perf_probe_event *pev) {
@@ -281,7 +324,7 @@ void parse_perf_probe_command(const char *cmd, struct perf_probe_event *pev) pev->nargs = argc -1 ; pev->args =xzalloc (sizeof (struct perf_probe_arg ) * pev->nargs);for (i =0 ; i < pev->nargs; i++) { - pev->args[i].name =xstrdup (argv[i +1 ]); +parse_perf_probe_arg (argv[i +1 ], &pev->args[i]);if (is_c_varname (pev->args[i].name) && pev->point.retprobe)semantic_error ("You can't specify local variable for" " kretprobe" );
@@ -353,6 +396,33 @@ void parse_kprobe_trace_command(const char *cmd, struct kprobe_trace_event *tev)argv_free (argv); } +/* Compose only probe arg */ +int synthesize_perf_probe_arg (struct perf_probe_arg *pa,char *buf,size_t len) +{ +struct perf_probe_arg_field *field = pa->field; +int ret; +char *tmp = buf; + + ret =e_snprintf (tmp, len,"%s" , pa->name); +if (ret <=0 ) + goto error; + tmp += ret; + len -= ret; + +while (field) { + ret =e_snprintf (tmp, len,"%s%s" , field->ref ?"->" :"." , + field->name); +if (ret <=0 ) + goto error; + tmp += ret; + len -= ret; + field = field->next; + } +return tmp - buf; +error: +die ("Failed to synthesize perf probe argument: %s" ,strerror (-ret)); +} +/* Compose only probe point (not argument) */ static char *synthesize_perf_probe_point (struct perf_probe_point *pp) {
@@ -563,6 +633,7 @@ void convert_to_perf_probe_event(struct kprobe_trace_event *tev,void clear_perf_probe_event (struct perf_probe_event *pev) {struct perf_probe_point *pp = &pev->point; +struct perf_probe_arg_field *field, *next;int i;if (pev->event)
@@ -575,9 +646,18 @@ void clear_perf_probe_event(struct perf_probe_event *pev)free (pp->function);if (pp->lazy_line)free (pp->lazy_line); -for (i =0 ; i < pev->nargs; i++) +for (i =0 ; i < pev->nargs; i++) {if (pev->args[i].name)free (pev->args[i].name); + field = pev->args[i].field; +while (field) { + next = field->next; +if (field->name) +free (field->name); +free (field); + field = next; + } + }if (pev->args)free (pev->args);memset (pev,0 ,sizeof (*pev));
@@ -682,8 +762,10 @@ static void show_perf_probe_event(struct perf_probe_event *pev)if (pev->nargs >0 ) {printf (" with" ); -for (i =0 ; i < pev->nargs; i++) -printf (" %s" , pev->args[i].name); +for (i =0 ; i < pev->nargs; i++) { +synthesize_perf_probe_arg (&pev->args[i], buf,128 ); +printf (" %s" , buf); + } }printf (")\n" );free (place);
tools/perf/util/probe-event.h +11 -1
@@ -45,9 +45,17 @@ struct perf_probe_point {bool retprobe;/* Return probe flag */ }; +/* Perf probe probing argument field chain */ +struct perf_probe_arg_field { +struct perf_probe_arg_field *next;/* Next field */ +char *name;/* Name of the field */ +bool ref;/* Referencing flag */ +}; +/* Perf probe probing argument */ struct perf_probe_arg { -char *name;/* Argument name */ +char *name;/* Argument name */ +struct perf_probe_arg_field *field;/* Structure fields */ };/* Perf probe probing event (point + arg) */
@@ -86,6 +94,8 @@ extern void parse_kprobe_trace_command(const char *cmd,/* Events to command string */ extern char *synthesize_perf_probe_command (struct perf_probe_event *pev);extern char *synthesize_kprobe_trace_command (struct kprobe_trace_event *tev); +extern int synthesize_perf_probe_arg (struct perf_probe_arg *pa,char *buf, +size_t len);/* Check the perf_probe_event needs debuginfo */ extern bool perf_probe_event_need_dwarf (struct perf_probe_event *pev);
tools/perf/util/probe-finder.c +104 -1
141 lines not on this page · raw message
Ingo Molnar
AppliedQuoted 6 lines · Masami Hiramatsu
Hi Ingo,
Here are several updates of perf-probe.
This updates includes data structure accessing support --list option enhancement and --dry-run option support.
Quoted 8 lines
- --dry-run allows users to run perf-probe just for checking that the command works correctly.
- --list option enhancement allows user to check where the probes are put on (with filename/line number).
- data structure accessing support allows user to trace the contents of local variables/function parameters.
Very nice! I think tracing a data structure value at a given point in the source will be a rather popular feature.
Quoted 11 lines
Long-term TODOs (future features):
- Enhance probe-finder to decode call frame instructions.
- Support tracing static variables (non global)
- Support variable basic types from debuginfo (e.g. char, int, ...)
- Support array element (var[N])
- Support string/dynamic arrays (*var, var[N..M])
- Support the type of return value
- Support force type-casting ((type)var)
- Supportsys_perf_counter_open(for non-root users)
- Support dynamic array-indexing (var[var2])
Quoted 3 lines
Miscs:
- Better support for probes on modules
- More debugger likeenhancements(%next, --disasm, etc.)
Sounds good :-)
Quoted 13 lines
Masami Hiramatsu (10):
perf probe: Accessing members in data structures perf probe: List probes with line number and file name perf probe: Introducekprobe_trace_eventandperf_probe_eventperf probe: Add --dry-run option
perf probe: Introducedie_find_child()function perf probe: Rename some die_get_* functions
perf probe: Rename session to param
perf probe: Move add-probe routine under util/ perf probe: Use wrapper functions
perf tools: Introducexzalloc()for detecting out of memory
Quoted 8 lines
tools/perf/Documentation/perf-probe.txt | 5 tools/perf/builtin-probe.c | 191 +------ tools/perf/util/probe-event.c | 830 +++++++++++++++++++++++-------- tools/perf/util/probe-event.h | 123 ++++- tools/perf/util/probe-finder.c | 459 ++++++++++++----- tools/perf/util/probe-finder.h | 60 +- tools/perf/util/util.h | 7 7 files changed, 1115 insertions(+), 560 deletions(-)
Applied, thanks!
Ingo