From 49908a1b25d448d68fd26faca260e1850201575f Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 4 Nov 2011 16:32:25 -0400 Subject: perf: Fix parsing of __print_flags() in TP_printk() A update is made to the sched:sched_switch event that adds some logic to the first parameter of the __print_flags() that shows the state of tasks. This change cause perf to fail parsing the flags. A simple fix is needed to have the parser be able to process ops within the argument. Cc: stable@vger.kernel.org Reported-by: Andrew Vagin Signed-off-by: Steven Rostedt --- tools/perf/util/trace-event-parse.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/trace-event-parse.c b/tools/perf/util/trace-event-parse.c index 0a7ed5b5e281..6c164dc9ee95 100644 --- a/tools/perf/util/trace-event-parse.c +++ b/tools/perf/util/trace-event-parse.c @@ -1537,6 +1537,8 @@ process_flags(struct event *event, struct print_arg *arg, char **tok) field = malloc_or_die(sizeof(*field)); type = process_arg(event, field, &token); + while (type == EVENT_OP) + type = process_op(event, field, &token); if (test_type_token(type, token, EVENT_DELIM, ",")) goto out_free; -- cgit v1.2.2 From 0e2a5f10fb550835e199a3b56a80ed88232188e9 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 4 Nov 2011 08:16:58 -0200 Subject: perf python: Fix undefined symbol problem Recently we made perf_evsel__init call hists__init, which broke the perf python binding: [root@emilia linux]# ./tools/perf/python/twatch.py Traceback (most recent call last): File "./tools/perf/python/twatch.py", line 16, in import perf ImportError: /home/acme/git/build/perf/python/perf.so: undefined symbol: hists__init Fix it by moving the hists__init function to its only caller, evsel.c. This way we avoid dragging in other parts of tools/perf/util/ to the perf python binding. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-5nffmdt5mu6ozxgj54oi4qon@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evsel.c | 10 ++++++++++ tools/perf/util/hist.c | 10 ---------- tools/perf/util/hist.h | 2 -- 3 files changed, 10 insertions(+), 12 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index e42626422587..d7915d4e77cb 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -34,6 +34,16 @@ int __perf_evsel__sample_size(u64 sample_type) return size; } +static void hists__init(struct hists *hists) +{ + memset(hists, 0, sizeof(*hists)); + hists->entries_in_array[0] = hists->entries_in_array[1] = RB_ROOT; + hists->entries_in = &hists->entries_in_array[0]; + hists->entries_collapsed = RB_ROOT; + hists->entries = RB_ROOT; + pthread_mutex_init(&hists->lock, NULL); +} + void perf_evsel__init(struct perf_evsel *evsel, struct perf_event_attr *attr, int idx) { diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index a36a3fa81ffb..abef2703cd24 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -1211,13 +1211,3 @@ size_t hists__fprintf_nr_events(struct hists *hists, FILE *fp) return ret; } - -void hists__init(struct hists *hists) -{ - memset(hists, 0, sizeof(*hists)); - hists->entries_in_array[0] = hists->entries_in_array[1] = RB_ROOT; - hists->entries_in = &hists->entries_in_array[0]; - hists->entries_collapsed = RB_ROOT; - hists->entries = RB_ROOT; - pthread_mutex_init(&hists->lock, NULL); -} diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index c86c1d27bd1e..89289c8e935e 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -63,8 +63,6 @@ struct hists { struct callchain_cursor callchain_cursor; }; -void hists__init(struct hists *hists); - struct hist_entry *__hists__add_entry(struct hists *self, struct addr_location *al, struct symbol *parent, u64 period); -- cgit v1.2.2 From 47fbe53bef3b219a365ebf3eca949d6cd4c5291c Mon Sep 17 00:00:00 2001 From: David Ahern Date: Sun, 13 Nov 2011 10:45:27 -0700 Subject: perf session: Fix crash with invalid CPU list commit 5d67be9 added the option to specify a range of CPUs of interest, but does not catch an invalid CPU list: $ perf script -c foo Segmentation fault (core dumped) Cc: Anton Blanchard Link: http://lkml.kernel.org/r/1321206327-5881-1-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/session.c | 4 ++++ 1 file changed, 4 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 85c1e6b76f0a..0f4555ce9063 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -1333,6 +1333,10 @@ int perf_session__cpu_bitmap(struct perf_session *session, } map = cpu_map__new(cpu_list); + if (map == NULL) { + pr_err("Invalid cpu_list\n"); + return -1; + } for (i = 0; i < map->nr; i++) { int cpu = map->map[i]; -- cgit v1.2.2 From c168fbfb93a1c4044287858c6784f0bd1f6cfe33 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 16 Nov 2011 12:55:59 -0200 Subject: perf tools: Eliminate duplicate code and use PATH_MAX consistently No need for multiple definitions for STR() and die(), also use SuSv2's PATH_MAX instead of adding MAX_PATH. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-qpujjkw7u0bf0tr4wt55cr9y@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/cgroup.c | 15 +++++++-------- tools/perf/util/debugfs.c | 12 ++++++------ tools/perf/util/debugfs.h | 29 +++++++++-------------------- tools/perf/util/probe-finder.h | 1 - tools/perf/util/trace-event-info.c | 28 ++-------------------------- 5 files changed, 24 insertions(+), 61 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/cgroup.c b/tools/perf/util/cgroup.c index 96bee5c46008..dbe2f16b1a1a 100644 --- a/tools/perf/util/cgroup.c +++ b/tools/perf/util/cgroup.c @@ -3,7 +3,6 @@ #include "parse-options.h" #include "evsel.h" #include "cgroup.h" -#include "debugfs.h" /* MAX_PATH, STR() */ #include "evlist.h" int nr_cgroups; @@ -12,7 +11,7 @@ static int cgroupfs_find_mountpoint(char *buf, size_t maxlen) { FILE *fp; - char mountpoint[MAX_PATH+1], tokens[MAX_PATH+1], type[MAX_PATH+1]; + char mountpoint[PATH_MAX + 1], tokens[PATH_MAX + 1], type[PATH_MAX + 1]; char *token, *saved_ptr = NULL; int found = 0; @@ -25,8 +24,8 @@ cgroupfs_find_mountpoint(char *buf, size_t maxlen) * and inspect every cgroupfs mount point to find one that has * perf_event subsystem */ - while (fscanf(fp, "%*s %"STR(MAX_PATH)"s %"STR(MAX_PATH)"s %" - STR(MAX_PATH)"s %*d %*d\n", + while (fscanf(fp, "%*s %"STR(PATH_MAX)"s %"STR(PATH_MAX)"s %" + STR(PATH_MAX)"s %*d %*d\n", mountpoint, type, tokens) == 3) { if (!strcmp(type, "cgroup")) { @@ -57,15 +56,15 @@ cgroupfs_find_mountpoint(char *buf, size_t maxlen) static int open_cgroup(char *name) { - char path[MAX_PATH+1]; - char mnt[MAX_PATH+1]; + char path[PATH_MAX + 1]; + char mnt[PATH_MAX + 1]; int fd; - if (cgroupfs_find_mountpoint(mnt, MAX_PATH+1)) + if (cgroupfs_find_mountpoint(mnt, PATH_MAX + 1)) return -1; - snprintf(path, MAX_PATH, "%s/%s", mnt, name); + snprintf(path, PATH_MAX, "%s/%s", mnt, name); fd = open(path, O_RDONLY); if (fd == -1) diff --git a/tools/perf/util/debugfs.c b/tools/perf/util/debugfs.c index a88fefc0cc0a..680be3460e86 100644 --- a/tools/perf/util/debugfs.c +++ b/tools/perf/util/debugfs.c @@ -2,8 +2,10 @@ #include "debugfs.h" #include "cache.h" +#include + static int debugfs_premounted; -static char debugfs_mountpoint[MAX_PATH+1]; +static char debugfs_mountpoint[PATH_MAX + 1]; static const char *debugfs_known_mountpoints[] = { "/sys/kernel/debug/", @@ -64,9 +66,7 @@ const char *debugfs_find_mountpoint(void) if (fp == NULL) die("Can't open /proc/mounts for read"); - while (fscanf(fp, "%*s %" - STR(MAX_PATH) - "s %99s %*s %*d %*d\n", + while (fscanf(fp, "%*s %" STR(PATH_MAX) "s %99s %*s %*d %*d\n", debugfs_mountpoint, type) == 2) { if (strcmp(type, "debugfs") == 0) break; @@ -158,7 +158,7 @@ int debugfs_umount(void) int debugfs_write(const char *entry, const char *value) { - char path[MAX_PATH+1]; + char path[PATH_MAX + 1]; int ret, count; int fd; @@ -203,7 +203,7 @@ int debugfs_write(const char *entry, const char *value) */ int debugfs_read(const char *entry, char *buffer, size_t size) { - char path[MAX_PATH+1]; + char path[PATH_MAX + 1]; int ret; int fd; diff --git a/tools/perf/util/debugfs.h b/tools/perf/util/debugfs.h index 83a02879745f..8cd3fa0af880 100644 --- a/tools/perf/util/debugfs.h +++ b/tools/perf/util/debugfs.h @@ -1,25 +1,14 @@ #ifndef __DEBUGFS_H__ #define __DEBUGFS_H__ -#include - -#ifndef MAX_PATH -# define MAX_PATH 256 -#endif - -#ifndef STR -# define _STR(x) #x -# define STR(x) _STR(x) -#endif - -extern const char *debugfs_find_mountpoint(void); -extern int debugfs_valid_mountpoint(const char *debugfs); -extern int debugfs_valid_entry(const char *path); -extern char *debugfs_mount(const char *mountpoint); -extern int debugfs_umount(void); -extern int debugfs_write(const char *entry, const char *value); -extern int debugfs_read(const char *entry, char *buffer, size_t size); -extern void debugfs_force_cleanup(void); -extern int debugfs_make_path(const char *element, char *buffer, int size); +const char *debugfs_find_mountpoint(void); +int debugfs_valid_mountpoint(const char *debugfs); +int debugfs_valid_entry(const char *path); +char *debugfs_mount(const char *mountpoint); +int debugfs_umount(void); +int debugfs_write(const char *entry, const char *value); +int debugfs_read(const char *entry, char *buffer, size_t size); +void debugfs_force_cleanup(void); +int debugfs_make_path(const char *element, char *buffer, int size); #endif /* __DEBUGFS_H__ */ diff --git a/tools/perf/util/probe-finder.h b/tools/perf/util/probe-finder.h index 1132c8f0ce89..17e94d0c36f9 100644 --- a/tools/perf/util/probe-finder.h +++ b/tools/perf/util/probe-finder.h @@ -5,7 +5,6 @@ #include "util.h" #include "probe-event.h" -#define MAX_PATH_LEN 256 #define MAX_PROBE_BUFFER 1024 #define MAX_PROBES 128 diff --git a/tools/perf/util/trace-event-info.c b/tools/perf/util/trace-event-info.c index d2655f08bcc0..ac6830d8292b 100644 --- a/tools/perf/util/trace-event-info.c +++ b/tools/perf/util/trace-event-info.c @@ -18,7 +18,8 @@ * * ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ */ -#define _GNU_SOURCE +#include +#include "util.h" #include #include #include @@ -31,7 +32,6 @@ #include #include #include -#include #include #include #include @@ -44,10 +44,6 @@ #define VERSION "0.5" -#define _STR(x) #x -#define STR(x) _STR(x) -#define MAX_PATH 256 - #define TRACE_CTRL "tracing_on" #define TRACE "trace" #define AVAILABLE "available_tracers" @@ -73,26 +69,6 @@ struct events { }; - -static void die(const char *fmt, ...) -{ - va_list ap; - int ret = errno; - - if (errno) - perror("perf"); - else - ret = -1; - - va_start(ap, fmt); - fprintf(stderr, " "); - vfprintf(stderr, fmt, ap); - va_end(ap); - - fprintf(stderr, "\n"); - exit(ret); -} - void *malloc_or_die(unsigned int size) { void *data; -- cgit v1.2.2 From ebf294bf4f147aff29df5a16bfb0f8ebca15feaa Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 16 Nov 2011 14:03:07 -0200 Subject: perf tools: Simplify debugfs mountpoint handling code We don't need to have two PATH_MAX char sized arrays holding it, just one in util/debugfs.c will do. Also rename debugfs_path to tracing_events_path, as it is not the path to debugfs, that is debugfs_mountpoint. Both are now accessible. This will allow accessing this code in the perf python binding without having to drag in perf.c and util/parse-events.c. The defaults for these variables are the canonical "/sys/kernel/debug" and "/sys/kernel/debug/tracing/events/", removing the need for simple tools to call debugfs_mount(NULL). Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-ug9jvtjrsqbluuhqqxpvg30f@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/debugfs.c | 25 ++++++++++++++++++++----- tools/perf/util/debugfs.h | 4 ++++ tools/perf/util/parse-events.c | 28 +++++++++++++--------------- tools/perf/util/parse-events.h | 1 - 4 files changed, 37 insertions(+), 21 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/debugfs.c b/tools/perf/util/debugfs.c index 680be3460e86..ffc35e748e89 100644 --- a/tools/perf/util/debugfs.c +++ b/tools/perf/util/debugfs.c @@ -2,10 +2,12 @@ #include "debugfs.h" #include "cache.h" +#include #include static int debugfs_premounted; -static char debugfs_mountpoint[PATH_MAX + 1]; +char debugfs_mountpoint[PATH_MAX + 1] = "/sys/kernel/debug"; +char tracing_events_path[PATH_MAX + 1] = "/sys/kernel/debug/tracing/events"; static const char *debugfs_known_mountpoints[] = { "/sys/kernel/debug/", @@ -64,7 +66,7 @@ const char *debugfs_find_mountpoint(void) /* give up and parse /proc/mounts */ fp = fopen("/proc/mounts", "r"); if (fp == NULL) - die("Can't open /proc/mounts for read"); + return NULL; while (fscanf(fp, "%*s %" STR(PATH_MAX) "s %99s %*s %*d %*d\n", debugfs_mountpoint, type) == 2) { @@ -106,6 +108,12 @@ int debugfs_valid_entry(const char *path) return 0; } +static void debugfs_set_tracing_events_path(const char *mountpoint) +{ + snprintf(tracing_events_path, sizeof(tracing_events_path), "%s/%s", + mountpoint, "tracing/events"); +} + /* mount the debugfs somewhere if it's not mounted */ char *debugfs_mount(const char *mountpoint) @@ -113,7 +121,7 @@ char *debugfs_mount(const char *mountpoint) /* see if it's already mounted */ if (debugfs_find_mountpoint()) { debugfs_premounted = 1; - return debugfs_mountpoint; + goto out; } /* if not mounted and no argument */ @@ -129,12 +137,19 @@ char *debugfs_mount(const char *mountpoint) return NULL; /* save the mountpoint */ - strncpy(debugfs_mountpoint, mountpoint, sizeof(debugfs_mountpoint)); debugfs_found = 1; - + strncpy(debugfs_mountpoint, mountpoint, sizeof(debugfs_mountpoint)); +out: + debugfs_set_tracing_events_path(debugfs_mountpoint); return debugfs_mountpoint; } +void debugfs_set_path(const char *mountpoint) +{ + snprintf(debugfs_mountpoint, sizeof(debugfs_mountpoint), "%s", mountpoint); + debugfs_set_tracing_events_path(mountpoint); +} + /* umount the debugfs */ int debugfs_umount(void) diff --git a/tools/perf/util/debugfs.h b/tools/perf/util/debugfs.h index 8cd3fa0af880..4a878f735eb0 100644 --- a/tools/perf/util/debugfs.h +++ b/tools/perf/util/debugfs.h @@ -6,9 +6,13 @@ int debugfs_valid_mountpoint(const char *debugfs); int debugfs_valid_entry(const char *path); char *debugfs_mount(const char *mountpoint); int debugfs_umount(void); +void debugfs_set_path(const char *mountpoint); int debugfs_write(const char *entry, const char *value); int debugfs_read(const char *entry, char *buffer, size_t size); void debugfs_force_cleanup(void); int debugfs_make_path(const char *element, char *buffer, int size); +extern char debugfs_mountpoint[]; +extern char tracing_events_path[]; + #endif /* __DEBUGFS_H__ */ diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 928918b796b2..586ab3fe60f8 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -25,8 +25,6 @@ enum event_result { EVT_HANDLED_ALL }; -char debugfs_path[MAXPATHLEN]; - #define CHW(x) .type = PERF_TYPE_HARDWARE, .config = PERF_COUNT_HW_##x #define CSW(x) .type = PERF_TYPE_SOFTWARE, .config = PERF_COUNT_SW_##x @@ -140,7 +138,7 @@ static int tp_event_has_id(struct dirent *sys_dir, struct dirent *evt_dir) char evt_path[MAXPATHLEN]; int fd; - snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id", debugfs_path, + snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id", tracing_events_path, sys_dir->d_name, evt_dir->d_name); fd = open(evt_path, O_RDONLY); if (fd < 0) @@ -171,16 +169,16 @@ struct tracepoint_path *tracepoint_id_to_path(u64 config) char evt_path[MAXPATHLEN]; char dir_path[MAXPATHLEN]; - if (debugfs_valid_mountpoint(debugfs_path)) + if (debugfs_valid_mountpoint(tracing_events_path)) return NULL; - sys_dir = opendir(debugfs_path); + sys_dir = opendir(tracing_events_path); if (!sys_dir) return NULL; for_each_subsystem(sys_dir, sys_dirent, sys_next) { - snprintf(dir_path, MAXPATHLEN, "%s/%s", debugfs_path, + snprintf(dir_path, MAXPATHLEN, "%s/%s", tracing_events_path, sys_dirent.d_name); evt_dir = opendir(dir_path); if (!evt_dir) @@ -447,7 +445,7 @@ parse_single_tracepoint_event(char *sys_name, u64 id; int fd; - snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id", debugfs_path, + snprintf(evt_path, MAXPATHLEN, "%s/%s/%s/id", tracing_events_path, sys_name, evt_name); fd = open(evt_path, O_RDONLY); @@ -485,7 +483,7 @@ parse_multiple_tracepoint_event(struct perf_evlist *evlist, char *sys_name, struct dirent *evt_ent; DIR *evt_dir; - snprintf(evt_path, MAXPATHLEN, "%s/%s", debugfs_path, sys_name); + snprintf(evt_path, MAXPATHLEN, "%s/%s", tracing_events_path, sys_name); evt_dir = opendir(evt_path); if (!evt_dir) { @@ -528,7 +526,7 @@ parse_tracepoint_event(struct perf_evlist *evlist, const char **strp, char sys_name[MAX_EVENT_LENGTH]; unsigned int sys_length, evt_length; - if (debugfs_valid_mountpoint(debugfs_path)) + if (debugfs_valid_mountpoint(tracing_events_path)) return 0; evt_name = strchr(*strp, ':'); @@ -920,10 +918,10 @@ void print_tracepoint_events(const char *subsys_glob, const char *event_glob) char evt_path[MAXPATHLEN]; char dir_path[MAXPATHLEN]; - if (debugfs_valid_mountpoint(debugfs_path)) + if (debugfs_valid_mountpoint(tracing_events_path)) return; - sys_dir = opendir(debugfs_path); + sys_dir = opendir(tracing_events_path); if (!sys_dir) return; @@ -932,7 +930,7 @@ void print_tracepoint_events(const char *subsys_glob, const char *event_glob) !strglobmatch(sys_dirent.d_name, subsys_glob)) continue; - snprintf(dir_path, MAXPATHLEN, "%s/%s", debugfs_path, + snprintf(dir_path, MAXPATHLEN, "%s/%s", tracing_events_path, sys_dirent.d_name); evt_dir = opendir(dir_path); if (!evt_dir) @@ -964,16 +962,16 @@ int is_valid_tracepoint(const char *event_string) char evt_path[MAXPATHLEN]; char dir_path[MAXPATHLEN]; - if (debugfs_valid_mountpoint(debugfs_path)) + if (debugfs_valid_mountpoint(tracing_events_path)) return 0; - sys_dir = opendir(debugfs_path); + sys_dir = opendir(tracing_events_path); if (!sys_dir) return 0; for_each_subsystem(sys_dir, sys_dirent, sys_next) { - snprintf(dir_path, MAXPATHLEN, "%s/%s", debugfs_path, + snprintf(dir_path, MAXPATHLEN, "%s/%s", tracing_events_path, sys_dirent.d_name); evt_dir = opendir(dir_path); if (!evt_dir) diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h index 2f8e375e038d..7e0cbe75d5f1 100644 --- a/tools/perf/util/parse-events.h +++ b/tools/perf/util/parse-events.h @@ -39,7 +39,6 @@ void print_tracepoint_events(const char *subsys_glob, const char *event_glob); int print_hwcache_events(const char *event_glob); extern int is_valid_tracepoint(const char *event_string); -extern char debugfs_path[]; extern int valid_debugfs_mount(const char *debugfs); #endif /* __PERF_PARSE_EVENTS_H */ -- cgit v1.2.2 From 50d08e47bc04eb05502f5c86b70bbd19ef1c2778 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 4 Nov 2011 09:10:59 -0200 Subject: perf evlist: Introduce perf_evlist__add_attrs Replacing the open coded equivalents in 'perf stat'. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-1btwadnf2tds2g07hsccsdse@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 34 ++++++++++++++++++++++++++++++++++ tools/perf/util/evlist.h | 7 +++++++ 2 files changed, 41 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index fbb4b4ab9cc6..58aa1e0092bd 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -13,6 +13,8 @@ #include "evsel.h" #include "util.h" +#include "parse-events.h" + #include #include @@ -76,6 +78,14 @@ void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry) ++evlist->nr_entries; } +static void perf_evlist__splice_list_tail(struct perf_evlist *evlist, + struct list_head *list, + int nr_entries) +{ + list_splice_tail(list, &evlist->entries); + evlist->nr_entries += nr_entries; +} + int perf_evlist__add_default(struct perf_evlist *evlist) { struct perf_event_attr attr = { @@ -100,6 +110,30 @@ error: return -ENOMEM; } +int perf_evlist__add_attrs(struct perf_evlist *evlist, + struct perf_event_attr *attrs, size_t nr_attrs) +{ + struct perf_evsel *evsel, *n; + LIST_HEAD(head); + size_t i; + + for (i = 0; i < nr_attrs; i++) { + evsel = perf_evsel__new(attrs + i, evlist->nr_entries + i); + if (evsel == NULL) + goto out_delete_partial_list; + list_add_tail(&evsel->node, &head); + } + + perf_evlist__splice_list_tail(evlist, &head, nr_attrs); + + return 0; + +out_delete_partial_list: + list_for_each_entry_safe(evsel, n, &head, node) + perf_evsel__delete(evsel); + return -1; +} + void perf_evlist__disable(struct perf_evlist *evlist) { int cpu, thread; diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 1779ffef7828..57d91ff2c56a 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -2,8 +2,10 @@ #define __PERF_EVLIST_H 1 #include +#include #include "../perf.h" #include "event.h" +#include "util.h" struct pollfd; struct thread_map; @@ -39,6 +41,11 @@ void perf_evlist__delete(struct perf_evlist *evlist); void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry); int perf_evlist__add_default(struct perf_evlist *evlist); +int perf_evlist__add_attrs(struct perf_evlist *evlist, + struct perf_event_attr *attrs, size_t nr_attrs); + +#define perf_evlist__add_attrs_array(evlist, array) \ + perf_evlist__add_attrs(evlist, array, ARRAY_SIZE(array)) void perf_evlist__id_add(struct perf_evlist *evlist, struct perf_evsel *evsel, int cpu, int thread, u64 id); -- cgit v1.2.2 From a8c9ae18d810e1ae12b6ec960907e9af63171d3a Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Sat, 5 Nov 2011 08:41:51 -0200 Subject: perf evlist: Introduce add_tracepoints method Convenient way of asking for tracepoint events to be added to an existing evlist. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-0ylj4wrg54791u0baqb9swbb@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 57 +++++++++++++++++++++++++++++++++++++++++++++++- tools/perf/util/evlist.h | 5 +++++ tools/perf/util/setup.py | 3 ++- 3 files changed, 63 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 58aa1e0092bd..3bc5a287a9f9 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -6,12 +6,13 @@ * * Released under the GPL v2. (and only v2, not any later version) */ +#include "util.h" +#include "debugfs.h" #include #include "cpumap.h" #include "thread_map.h" #include "evlist.h" #include "evsel.h" -#include "util.h" #include "parse-events.h" @@ -134,6 +135,60 @@ out_delete_partial_list: return -1; } +static int trace_event__id(const char *evname) +{ + char *filename, *colon; + int err = -1, fd; + + if (asprintf(&filename, "%s/%s/id", tracing_events_path, evname) < 0) + return -1; + + colon = strrchr(filename, ':'); + if (colon != NULL) + *colon = '/'; + + fd = open(filename, O_RDONLY); + if (fd >= 0) { + char id[16]; + if (read(fd, id, sizeof(id)) > 0) + err = atoi(id); + close(fd); + } + + free(filename); + return err; +} + +int perf_evlist__add_tracepoints(struct perf_evlist *evlist, + const char *tracepoints[], + size_t nr_tracepoints) +{ + int err; + size_t i; + struct perf_event_attr *attrs = zalloc(nr_tracepoints * sizeof(*attrs)); + + if (attrs == NULL) + return -1; + + for (i = 0; i < nr_tracepoints; i++) { + err = trace_event__id(tracepoints[i]); + + if (err < 0) + goto out_free_attrs; + + attrs[i].type = PERF_TYPE_TRACEPOINT; + attrs[i].config = err; + attrs[i].sample_type = (PERF_SAMPLE_RAW | PERF_SAMPLE_TIME | + PERF_SAMPLE_CPU); + attrs[i].sample_period = 1; + } + + err = perf_evlist__add_attrs(evlist, attrs, nr_tracepoints); +out_free_attrs: + free(attrs); + return err; +} + void perf_evlist__disable(struct perf_evlist *evlist) { int cpu, thread; diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 57d91ff2c56a..ec71c82935bd 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -43,10 +43,15 @@ void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry); int perf_evlist__add_default(struct perf_evlist *evlist); int perf_evlist__add_attrs(struct perf_evlist *evlist, struct perf_event_attr *attrs, size_t nr_attrs); +int perf_evlist__add_tracepoints(struct perf_evlist *evlist, + const char *tracepoints[], size_t nr_tracepoints); #define perf_evlist__add_attrs_array(evlist, array) \ perf_evlist__add_attrs(evlist, array, ARRAY_SIZE(array)) +#define perf_evlist__add_tracepoints_array(evlist, array) \ + perf_evlist__add_tracepoints(evlist, array, ARRAY_SIZE(array)) + void perf_evlist__id_add(struct perf_evlist *evlist, struct perf_evsel *evsel, int cpu, int thread, u64 id); diff --git a/tools/perf/util/setup.py b/tools/perf/util/setup.py index 95d370074928..36d4c5619575 100644 --- a/tools/perf/util/setup.py +++ b/tools/perf/util/setup.py @@ -27,7 +27,8 @@ build_tmp = getenv('PYTHON_EXTBUILD_TMP') perf = Extension('perf', sources = ['util/python.c', 'util/ctype.c', 'util/evlist.c', 'util/evsel.c', 'util/cpumap.c', 'util/thread_map.c', - 'util/util.c', 'util/xyarray.c', 'util/cgroup.c'], + 'util/util.c', 'util/xyarray.c', 'util/cgroup.c', + 'util/debugfs.c'], include_dirs = ['util/include'], extra_compile_args = cflags, ) -- cgit v1.2.2 From 0f82ebc452f921590e216b28eee0b41f5e434a48 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 8 Nov 2011 14:41:57 -0200 Subject: perf evsel: Introduce config attr method Out of the code in 'perf record', so that we can share option parsing, etc. Eventually will be used by 'perf top', but first 'trace' will use it. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-hzjqsgnte1esk90ytq0ap98v@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 16 +++++++++++ tools/perf/util/evlist.h | 4 +++ tools/perf/util/evsel.c | 70 ++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/evsel.h | 4 +++ 4 files changed, 94 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 3bc5a287a9f9..b774341e797f 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -46,6 +46,22 @@ struct perf_evlist *perf_evlist__new(struct cpu_map *cpus, return evlist; } +void perf_evlist__config_attrs(struct perf_evlist *evlist, + struct perf_record_opts *opts) +{ + struct perf_evsel *evsel; + + if (evlist->cpus->map[0] < 0) + opts->no_inherit = true; + + list_for_each_entry(evsel, &evlist->entries, node) { + perf_evsel__config(evsel, opts); + + if (evlist->nr_entries > 1) + evsel->attr.sample_type |= PERF_SAMPLE_ID; + } +} + static void perf_evlist__purge(struct perf_evlist *evlist) { struct perf_evsel *pos, *n; diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index ec71c82935bd..231c06f8286b 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -10,6 +10,7 @@ struct pollfd; struct thread_map; struct cpu_map; +struct perf_record_opts; #define PERF_EVLIST__HLIST_BITS 8 #define PERF_EVLIST__HLIST_SIZE (1 << PERF_EVLIST__HLIST_BITS) @@ -64,6 +65,9 @@ union perf_event *perf_evlist__mmap_read(struct perf_evlist *self, int idx); int perf_evlist__open(struct perf_evlist *evlist, bool group); +void perf_evlist__config_attrs(struct perf_evlist *evlist, + struct perf_record_opts *opts); + int perf_evlist__alloc_mmap(struct perf_evlist *evlist); int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite); void perf_evlist__munmap(struct perf_evlist *evlist); diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index e42626422587..b38eaa34b28e 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -53,6 +53,76 @@ struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx) return evsel; } +void perf_evsel__config(struct perf_evsel *evsel, struct perf_record_opts *opts) +{ + struct perf_event_attr *attr = &evsel->attr; + int track = !evsel->idx; /* only the first counter needs these */ + + attr->sample_id_all = opts->sample_id_all_avail ? 1 : 0; + attr->inherit = !opts->no_inherit; + attr->read_format = PERF_FORMAT_TOTAL_TIME_ENABLED | + PERF_FORMAT_TOTAL_TIME_RUNNING | + PERF_FORMAT_ID; + + attr->sample_type |= PERF_SAMPLE_IP | PERF_SAMPLE_TID; + + /* + * We default some events to a 1 default interval. But keep + * it a weak assumption overridable by the user. + */ + if (!attr->sample_period || (opts->user_freq != UINT_MAX && + opts->user_interval != ULLONG_MAX)) { + if (opts->freq) { + attr->sample_type |= PERF_SAMPLE_PERIOD; + attr->freq = 1; + attr->sample_freq = opts->freq; + } else { + attr->sample_period = opts->default_interval; + } + } + + if (opts->no_samples) + attr->sample_freq = 0; + + if (opts->inherit_stat) + attr->inherit_stat = 1; + + if (opts->sample_address) { + attr->sample_type |= PERF_SAMPLE_ADDR; + attr->mmap_data = track; + } + + if (opts->call_graph) + attr->sample_type |= PERF_SAMPLE_CALLCHAIN; + + if (opts->system_wide) + attr->sample_type |= PERF_SAMPLE_CPU; + + if (opts->sample_id_all_avail && + (opts->sample_time || opts->system_wide || + !opts->no_inherit || opts->cpu_list)) + attr->sample_type |= PERF_SAMPLE_TIME; + + if (opts->raw_samples) { + attr->sample_type |= PERF_SAMPLE_TIME; + attr->sample_type |= PERF_SAMPLE_RAW; + attr->sample_type |= PERF_SAMPLE_CPU; + } + + if (opts->no_delay) { + attr->watermark = 0; + attr->wakeup_events = 1; + } + + attr->mmap = track; + attr->comm = track; + + if (opts->target_pid == -1 && opts->target_tid == -1 && !opts->system_wide) { + attr->disabled = 1; + attr->enable_on_exec = 1; + } +} + int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads) { int cpu, thread; diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index b1d15e6f7ae3..6421c07f5015 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -67,6 +67,7 @@ struct perf_evsel { struct cpu_map; struct thread_map; struct perf_evlist; +struct perf_record_opts; struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx); void perf_evsel__init(struct perf_evsel *evsel, @@ -74,6 +75,9 @@ void perf_evsel__init(struct perf_evsel *evsel, void perf_evsel__exit(struct perf_evsel *evsel); void perf_evsel__delete(struct perf_evsel *evsel); +void perf_evsel__config(struct perf_evsel *evsel, + struct perf_record_opts *opts); + int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads); int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads); int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus); -- cgit v1.2.2 From 35b9d88ecd8c5fb720ba0dd325262f356d0b03e7 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 9 Nov 2011 08:47:15 -0200 Subject: perf evlist: Introduce {prepare,start}_workload refactored from 'perf record' So that we can easily start a workload in other tools. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-zdsksd4aphu0nltg2lpwsw3x@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 96 ++++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/evlist.h | 10 +++++ 2 files changed, 106 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index b774341e797f..a472247af191 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -13,6 +13,7 @@ #include "thread_map.h" #include "evlist.h" #include "evsel.h" +#include #include "parse-events.h" @@ -33,6 +34,7 @@ void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus, INIT_HLIST_HEAD(&evlist->heads[i]); INIT_LIST_HEAD(&evlist->entries); perf_evlist__set_maps(evlist, cpus, threads); + evlist->workload.pid = -1; } struct perf_evlist *perf_evlist__new(struct cpu_map *cpus, @@ -674,3 +676,97 @@ out_err: return err; } + +int perf_evlist__prepare_workload(struct perf_evlist *evlist, + struct perf_record_opts *opts, + const char *argv[]) +{ + int child_ready_pipe[2], go_pipe[2]; + char bf; + + if (pipe(child_ready_pipe) < 0) { + perror("failed to create 'ready' pipe"); + return -1; + } + + if (pipe(go_pipe) < 0) { + perror("failed to create 'go' pipe"); + goto out_close_ready_pipe; + } + + evlist->workload.pid = fork(); + if (evlist->workload.pid < 0) { + perror("failed to fork"); + goto out_close_pipes; + } + + if (!evlist->workload.pid) { + if (opts->pipe_output) + dup2(2, 1); + + close(child_ready_pipe[0]); + close(go_pipe[1]); + fcntl(go_pipe[0], F_SETFD, FD_CLOEXEC); + + /* + * Do a dummy execvp to get the PLT entry resolved, + * so we avoid the resolver overhead on the real + * execvp call. + */ + execvp("", (char **)argv); + + /* + * Tell the parent we're ready to go + */ + close(child_ready_pipe[1]); + + /* + * Wait until the parent tells us to go. + */ + if (read(go_pipe[0], &bf, 1) == -1) + perror("unable to read pipe"); + + execvp(argv[0], (char **)argv); + + perror(argv[0]); + kill(getppid(), SIGUSR1); + exit(-1); + } + + if (!opts->system_wide && opts->target_tid == -1 && opts->target_pid == -1) + evlist->threads->map[0] = evlist->workload.pid; + + close(child_ready_pipe[1]); + close(go_pipe[0]); + /* + * wait for child to settle + */ + if (read(child_ready_pipe[0], &bf, 1) == -1) { + perror("unable to read pipe"); + goto out_close_pipes; + } + + evlist->workload.cork_fd = go_pipe[1]; + close(child_ready_pipe[0]); + return 0; + +out_close_pipes: + close(go_pipe[0]); + close(go_pipe[1]); +out_close_ready_pipe: + close(child_ready_pipe[0]); + close(child_ready_pipe[1]); + return -1; +} + +int perf_evlist__start_workload(struct perf_evlist *evlist) +{ + if (evlist->workload.cork_fd > 0) { + /* + * Remove the cork, let it rip! + */ + return close(evlist->workload.cork_fd); + } + + return 0; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 231c06f8286b..07d56b3e6d61 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -6,6 +6,7 @@ #include "../perf.h" #include "event.h" #include "util.h" +#include struct pollfd; struct thread_map; @@ -22,6 +23,10 @@ struct perf_evlist { int nr_fds; int nr_mmaps; int mmap_len; + struct { + int cork_fd; + pid_t pid; + } workload; bool overwrite; union perf_event event_copy; struct perf_mmap *mmap; @@ -68,6 +73,11 @@ int perf_evlist__open(struct perf_evlist *evlist, bool group); void perf_evlist__config_attrs(struct perf_evlist *evlist, struct perf_record_opts *opts); +int perf_evlist__prepare_workload(struct perf_evlist *evlist, + struct perf_record_opts *opts, + const char *argv[]); +int perf_evlist__start_workload(struct perf_evlist *evlist); + int perf_evlist__alloc_mmap(struct perf_evlist *evlist); int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite); void perf_evlist__munmap(struct perf_evlist *evlist); -- cgit v1.2.2 From 50a682ce875f91e04c81ae5680535aba5a770bad Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 9 Nov 2011 09:10:47 -0200 Subject: perf evlist: Handle default value for 'pages' on mmap method Every tool that calls this and allows the user to override the value needs this logic. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-lwscxpg57xfzahz5dmdfp9uz@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 12 +++++++++--- tools/perf/util/evlist.h | 3 ++- 2 files changed, 11 insertions(+), 4 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index a472247af191..81237dcde1fd 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -507,14 +507,20 @@ out_unmap: * * Using perf_evlist__read_on_cpu does this automatically. */ -int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite) +int perf_evlist__mmap(struct perf_evlist *evlist, unsigned int pages, + bool overwrite) { unsigned int page_size = sysconf(_SC_PAGE_SIZE); - int mask = pages * page_size - 1; struct perf_evsel *evsel; const struct cpu_map *cpus = evlist->cpus; const struct thread_map *threads = evlist->threads; - int prot = PROT_READ | (overwrite ? 0 : PROT_WRITE); + int prot = PROT_READ | (overwrite ? 0 : PROT_WRITE), mask; + + /* 512 kiB: default amount of unprivileged mlocked memory */ + if (pages == UINT_MAX) + pages = (512 * 1024) / page_size; + + mask = pages * page_size - 1; if (evlist->mmap == NULL && perf_evlist__alloc_mmap(evlist) < 0) return -ENOMEM; diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 07d56b3e6d61..17e9c80243ce 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -79,7 +79,8 @@ int perf_evlist__prepare_workload(struct perf_evlist *evlist, int perf_evlist__start_workload(struct perf_evlist *evlist); int perf_evlist__alloc_mmap(struct perf_evlist *evlist); -int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite); +int perf_evlist__mmap(struct perf_evlist *evlist, unsigned int pages, + bool overwrite); void perf_evlist__munmap(struct perf_evlist *evlist); void perf_evlist__disable(struct perf_evlist *evlist); -- cgit v1.2.2 From b424eba27160dd19577896d4520b8eebabed919f Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Wed, 9 Nov 2011 13:24:25 -0200 Subject: perf session: Move threads to struct machine The 'machine' abstraction was introduced with 'perf kvm' where we could have samples for the host and multiple guests, but at the time we ended up keeping the list of all machines threads all in session->host_machine. Move the threads rb_tree to struct machine to separate the namespaces. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-mdg7sm6j3va09vtgj49gbsrp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/build-id.c | 4 ++-- tools/perf/util/map.c | 4 ++++ tools/perf/util/map.h | 9 +++++++++ tools/perf/util/session.c | 47 +++++++++++++++++++++++++++++++++++++++------- tools/perf/util/session.h | 3 --- tools/perf/util/thread.c | 6 +++--- 6 files changed, 58 insertions(+), 15 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c index a91cd99f26ea..f2fe6ec08945 100644 --- a/tools/perf/util/build-id.c +++ b/tools/perf/util/build-id.c @@ -48,8 +48,8 @@ static int perf_event__exit_del_thread(union perf_event *event, event->fork.ppid, event->fork.ptid); if (thread) { - rb_erase(&thread->rb_node, &session->threads); - session->last_match = NULL; + rb_erase(&thread->rb_node, &session->host_machine.threads); + session->host_machine.last_match = NULL; thread__delete(thread); } diff --git a/tools/perf/util/map.c b/tools/perf/util/map.c index 78284b13e808..316aa0ab7122 100644 --- a/tools/perf/util/map.c +++ b/tools/perf/util/map.c @@ -562,6 +562,10 @@ int machine__init(struct machine *self, const char *root_dir, pid_t pid) INIT_LIST_HEAD(&self->user_dsos); INIT_LIST_HEAD(&self->kernel_dsos); + self->threads = RB_ROOT; + INIT_LIST_HEAD(&self->dead_threads); + self->last_match = NULL; + self->kmaps.machine = self; self->pid = pid; self->root_dir = strdup(root_dir); diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h index 890d85545d0f..bde6835ee257 100644 --- a/tools/perf/util/map.h +++ b/tools/perf/util/map.h @@ -62,6 +62,9 @@ struct machine { struct rb_node rb_node; pid_t pid; char *root_dir; + struct rb_root threads; + struct list_head dead_threads; + struct thread *last_match; struct list_head user_dsos; struct list_head kernel_dsos; struct map_groups kmaps; @@ -190,6 +193,12 @@ struct symbol *map_groups__find_symbol_by_name(struct map_groups *mg, struct map **mapp, symbol_filter_t filter); + +struct thread *machine__findnew_thread(struct machine *machine, pid_t pid); +void machine__remove_thread(struct machine *machine, struct thread *th); + +size_t machine__fprintf(struct machine *machine, FILE *fp); + static inline struct symbol *machine__find_kernel_symbol(struct machine *self, enum map_type type, u64 addr, diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 85c1e6b76f0a..a76666f17767 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -139,9 +139,6 @@ struct perf_session *perf_session__new(const char *filename, int mode, goto out; memcpy(self->filename, filename, len); - self->threads = RB_ROOT; - INIT_LIST_HEAD(&self->dead_threads); - self->last_match = NULL; /* * On 64bit we can mmap the data file in one go. No need for tiny mmap * slices. On 32bit we use 32MB. @@ -184,17 +181,22 @@ out_delete: return NULL; } -static void perf_session__delete_dead_threads(struct perf_session *self) +static void machine__delete_dead_threads(struct machine *machine) { struct thread *n, *t; - list_for_each_entry_safe(t, n, &self->dead_threads, node) { + list_for_each_entry_safe(t, n, &machine->dead_threads, node) { list_del(&t->node); thread__delete(t); } } -static void perf_session__delete_threads(struct perf_session *self) +static void perf_session__delete_dead_threads(struct perf_session *session) +{ + machine__delete_dead_threads(&session->host_machine); +} + +static void machine__delete_threads(struct machine *self) { struct rb_node *nd = rb_first(&self->threads); @@ -207,6 +209,11 @@ static void perf_session__delete_threads(struct perf_session *self) } } +static void perf_session__delete_threads(struct perf_session *session) +{ + machine__delete_threads(&session->host_machine); +} + void perf_session__delete(struct perf_session *self) { perf_session__destroy_kernel_maps(self); @@ -217,7 +224,7 @@ void perf_session__delete(struct perf_session *self) free(self); } -void perf_session__remove_thread(struct perf_session *self, struct thread *th) +void machine__remove_thread(struct machine *self, struct thread *th) { self->last_match = NULL; rb_erase(&th->rb_node, &self->threads); @@ -884,6 +891,11 @@ void perf_event_header__bswap(struct perf_event_header *self) self->size = bswap_16(self->size); } +struct thread *perf_session__findnew(struct perf_session *session, pid_t pid) +{ + return machine__findnew_thread(&session->host_machine, pid); +} + static struct thread *perf_session__register_idle_thread(struct perf_session *self) { struct thread *thread = perf_session__findnew(self, 0); @@ -1224,6 +1236,27 @@ size_t perf_session__fprintf_nr_events(struct perf_session *session, FILE *fp) return ret; } +size_t perf_session__fprintf(struct perf_session *session, FILE *fp) +{ + /* + * FIXME: Here we have to actually print all the machines in this + * session, not just the host... + */ + return machine__fprintf(&session->host_machine, fp); +} + +void perf_session__remove_thread(struct perf_session *session, + struct thread *th) +{ + /* + * FIXME: This one makes no sense, we need to remove the thread from + * the machine it belongs to, perf_session can have many machines, so + * doing it always on ->host_machine is wrong. Fix when auditing all + * the 'perf kvm' code. + */ + machine__remove_thread(&session->host_machine, th); +} + struct perf_evsel *perf_session__find_first_evtype(struct perf_session *session, unsigned int type) { diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index 6e393c98eb34..76d462d3bef7 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -30,9 +30,6 @@ struct perf_session { struct perf_header header; unsigned long size; unsigned long mmap_window; - struct rb_root threads; - struct list_head dead_threads; - struct thread *last_match; struct machine host_machine; struct rb_root machines; struct perf_evlist *evlist; diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c index d5d3b22250f3..fb4b7ea6752f 100644 --- a/tools/perf/util/thread.c +++ b/tools/perf/util/thread.c @@ -61,7 +61,7 @@ static size_t thread__fprintf(struct thread *self, FILE *fp) map_groups__fprintf(&self->mg, verbose, fp); } -struct thread *perf_session__findnew(struct perf_session *self, pid_t pid) +struct thread *machine__findnew_thread(struct machine *self, pid_t pid) { struct rb_node **p = &self->threads.rb_node; struct rb_node *parent = NULL; @@ -125,12 +125,12 @@ int thread__fork(struct thread *self, struct thread *parent) return 0; } -size_t perf_session__fprintf(struct perf_session *self, FILE *fp) +size_t machine__fprintf(struct machine *machine, FILE *fp) { size_t ret = 0; struct rb_node *nd; - for (nd = rb_first(&self->threads); nd; nd = rb_next(nd)) { + for (nd = rb_first(&machine->threads); nd; nd = rb_next(nd)) { struct thread *pos = rb_entry(nd, struct thread, rb_node); ret += thread__fprintf(pos, fp); -- cgit v1.2.2 From 18b552350515188a732db6ccdb81e9cefb8b58c9 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 11 Nov 2011 22:08:07 -0200 Subject: perf ui progress: Fix divide by zero Happens in a perf.data file where one of the events had no samples. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-j7st3oyiotvfxqde2nc41kxb@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/ui/progress.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/ui/progress.c b/tools/perf/util/ui/progress.c index 295e366b6311..13aa64e50e11 100644 --- a/tools/perf/util/ui/progress.c +++ b/tools/perf/util/ui/progress.c @@ -14,6 +14,9 @@ void ui_progress__update(u64 curr, u64 total, const char *title) if (use_browser <= 0) return; + if (total == 0) + return; + ui__refresh_dimensions(true); pthread_mutex_lock(&ui__lock); y = SLtt_Screen_Rows / 2 - 2; -- cgit v1.2.2 From d04b35f8085f0d4c5c874515b8f65e7664357148 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 11 Nov 2011 22:17:32 -0200 Subject: perf symbols: Add nr_events to symbol_conf Since symbol__alloc_hists need it, to avoid passing it around in many functions have it in the symbol_conf struct. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-cwv8ysvpywzjq4v3xtbd4zwv@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/annotate.c | 6 +++--- tools/perf/util/annotate.h | 5 ++--- tools/perf/util/header.c | 2 ++ tools/perf/util/hist.h | 3 +-- tools/perf/util/symbol.h | 1 + tools/perf/util/ui/browsers/annotate.c | 16 +++++++--------- tools/perf/util/ui/browsers/hists.c | 2 +- 7 files changed, 17 insertions(+), 18 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 119e996035c8..376e643f7066 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -25,17 +25,17 @@ int symbol__annotate_init(struct map *map __used, struct symbol *sym) return 0; } -int symbol__alloc_hist(struct symbol *sym, int nevents) +int symbol__alloc_hist(struct symbol *sym) { struct annotation *notes = symbol__annotation(sym); size_t sizeof_sym_hist = (sizeof(struct sym_hist) + (sym->end - sym->start) * sizeof(u64)); - notes->src = zalloc(sizeof(*notes->src) + nevents * sizeof_sym_hist); + notes->src = zalloc(sizeof(*notes->src) + symbol_conf.nr_events * sizeof_sym_hist); if (notes->src == NULL) return -1; notes->src->sizeof_sym_hist = sizeof_sym_hist; - notes->src->nr_histograms = nevents; + notes->src->nr_histograms = symbol_conf.nr_events; INIT_LIST_HEAD(¬es->src->source); return 0; } diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h index d9072523d342..efa5dc82bfae 100644 --- a/tools/perf/util/annotate.h +++ b/tools/perf/util/annotate.h @@ -72,7 +72,7 @@ static inline struct annotation *symbol__annotation(struct symbol *sym) int symbol__inc_addr_samples(struct symbol *sym, struct map *map, int evidx, u64 addr); -int symbol__alloc_hist(struct symbol *sym, int nevents); +int symbol__alloc_hist(struct symbol *sym); void symbol__annotate_zero_histograms(struct symbol *sym); int symbol__annotate(struct symbol *sym, struct map *map, size_t privsize); @@ -99,8 +99,7 @@ static inline int symbol__tui_annotate(struct symbol *sym __used, } #else int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx, - int nr_events, void(*timer)(void *arg), void *arg, - int delay_secs); + void(*timer)(void *arg), void *arg, int delay_secs); #endif extern const char *disassembler_style; diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index bcd05d05b4f0..41424a16be8e 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -2041,6 +2041,8 @@ int perf_session__read_header(struct perf_session *session, int fd) lseek(fd, tmp, SEEK_SET); } + symbol_conf.nr_events = nr_attrs; + if (f_header.event_types.size) { lseek(fd, f_header.event_types.offset, SEEK_SET); events = malloc(f_header.event_types.size); diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index c86c1d27bd1e..6676d558b2a7 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -119,7 +119,6 @@ int perf_evlist__tui_browse_hists(struct perf_evlist *evlist __used, static inline int hist_entry__tui_annotate(struct hist_entry *self __used, int evidx __used, - int nr_events __used, void(*timer)(void *arg) __used, void *arg __used, int delay_secs __used) @@ -130,7 +129,7 @@ static inline int hist_entry__tui_annotate(struct hist_entry *self __used, #define K_RIGHT -2 #else #include "ui/keysyms.h" -int hist_entry__tui_annotate(struct hist_entry *he, int evidx, int nr_events, +int hist_entry__tui_annotate(struct hist_entry *he, int evidx, void(*timer)(void *arg), void *arg, int delay_secs); int perf_evlist__tui_browse_hists(struct perf_evlist *evlist, const char *help, diff --git a/tools/perf/util/symbol.h b/tools/perf/util/symbol.h index 29f8d742e92f..123c2e14353e 100644 --- a/tools/perf/util/symbol.h +++ b/tools/perf/util/symbol.h @@ -68,6 +68,7 @@ struct strlist; struct symbol_conf { unsigned short priv_size; + unsigned short nr_events; bool try_vmlinux_path, use_modules, sort_by_name, diff --git a/tools/perf/util/ui/browsers/annotate.c b/tools/perf/util/ui/browsers/annotate.c index 0575905d1205..295a9c93f945 100644 --- a/tools/perf/util/ui/browsers/annotate.c +++ b/tools/perf/util/ui/browsers/annotate.c @@ -224,7 +224,7 @@ static bool annotate_browser__toggle_source(struct annotate_browser *browser) } static int annotate_browser__run(struct annotate_browser *self, int evidx, - int nr_events, void(*timer)(void *arg), + void(*timer)(void *arg), void *arg, int delay_secs) { struct rb_node *nd = NULL; @@ -328,8 +328,7 @@ static int annotate_browser__run(struct annotate_browser *self, int evidx, notes = symbol__annotation(target); pthread_mutex_lock(¬es->lock); - if (notes->src == NULL && - symbol__alloc_hist(target, nr_events) < 0) { + if (notes->src == NULL && symbol__alloc_hist(target) < 0) { pthread_mutex_unlock(¬es->lock); ui__warning("Not enough memory for annotating '%s' symbol!\n", target->name); @@ -337,7 +336,7 @@ static int annotate_browser__run(struct annotate_browser *self, int evidx, } pthread_mutex_unlock(¬es->lock); - symbol__tui_annotate(target, ms->map, evidx, nr_events, + symbol__tui_annotate(target, ms->map, evidx, timer, arg, delay_secs); } continue; @@ -358,15 +357,15 @@ out: return key; } -int hist_entry__tui_annotate(struct hist_entry *he, int evidx, int nr_events, +int hist_entry__tui_annotate(struct hist_entry *he, int evidx, void(*timer)(void *arg), void *arg, int delay_secs) { - return symbol__tui_annotate(he->ms.sym, he->ms.map, evidx, nr_events, + return symbol__tui_annotate(he->ms.sym, he->ms.map, evidx, timer, arg, delay_secs); } int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx, - int nr_events, void(*timer)(void *arg), void *arg, + void(*timer)(void *arg), void *arg, int delay_secs) { struct objdump_line *pos, *n; @@ -419,8 +418,7 @@ int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx, browser.b.nr_entries = browser.nr_entries; browser.b.entries = ¬es->src->source, browser.b.width += 18; /* Percentage */ - ret = annotate_browser__run(&browser, evidx, nr_events, - timer, arg, delay_secs); + ret = annotate_browser__run(&browser, evidx, timer, arg, delay_secs); list_for_each_entry_safe(pos, n, ¬es->src->source, node) { list_del(&pos->node); objdump_line__free(pos); diff --git a/tools/perf/util/ui/browsers/hists.c b/tools/perf/util/ui/browsers/hists.c index d0c94b459685..1212a386a033 100644 --- a/tools/perf/util/ui/browsers/hists.c +++ b/tools/perf/util/ui/browsers/hists.c @@ -1020,7 +1020,7 @@ do_annotate: * Don't let this be freed, say, by hists__decay_entry. */ he->used = true; - err = hist_entry__tui_annotate(he, evsel->idx, nr_events, + err = hist_entry__tui_annotate(he, evsel->idx, timer, arg, delay_secs); he->used = false; ui_browser__update_nr_entries(&browser->b, browser->hists->nr_entries); -- cgit v1.2.2 From 81e36bffad95e015af9741b5b1ee16afe08aab05 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 11 Nov 2011 22:28:50 -0200 Subject: perf evlist: Introduce id_hdr_size method out of perf_session We will need this when not using perf_session in cases like 'perf top' and strace where no perf.data file is created nor consumed. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-za923wjc41q5xot5vrhuhj3j@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 32 ++++++++++++++++++++++++++++++++ tools/perf/util/evlist.h | 1 + tools/perf/util/session.c | 29 +---------------------------- 3 files changed, 34 insertions(+), 28 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 81237dcde1fd..d44e3df13a8f 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -625,6 +625,38 @@ u64 perf_evlist__sample_type(const struct perf_evlist *evlist) return first->attr.sample_type; } +u16 perf_evlist__id_hdr_size(const struct perf_evlist *evlist) +{ + struct perf_evsel *first; + struct perf_sample *data; + u64 sample_type; + u16 size = 0; + + first = list_entry(evlist->entries.next, struct perf_evsel, node); + + if (!first->attr.sample_id_all) + goto out; + + sample_type = first->attr.sample_type; + + if (sample_type & PERF_SAMPLE_TID) + size += sizeof(data->tid) * 2; + + if (sample_type & PERF_SAMPLE_TIME) + size += sizeof(data->time); + + if (sample_type & PERF_SAMPLE_ID) + size += sizeof(data->id); + + if (sample_type & PERF_SAMPLE_STREAM_ID) + size += sizeof(data->stream_id); + + if (sample_type & PERF_SAMPLE_CPU) + size += sizeof(data->cpu) * 2; +out: + return size; +} + bool perf_evlist__valid_sample_id_all(const struct perf_evlist *evlist) { struct perf_evsel *pos, *first; diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 17e9c80243ce..2202e7b04103 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -104,6 +104,7 @@ int perf_evlist__set_filters(struct perf_evlist *evlist); u64 perf_evlist__sample_type(const struct perf_evlist *evlist); bool perf_evlist__sample_id_all(const const struct perf_evlist *evlist); +u16 perf_evlist__id_hdr_size(const struct perf_evlist *evlist); bool perf_evlist__valid_sample_type(const struct perf_evlist *evlist); bool perf_evlist__valid_sample_id_all(const struct perf_evlist *evlist); diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index a76666f17767..675e080f66b6 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -78,39 +78,12 @@ out_close: return -1; } -static void perf_session__id_header_size(struct perf_session *session) -{ - struct perf_sample *data; - u64 sample_type = session->sample_type; - u16 size = 0; - - if (!session->sample_id_all) - goto out; - - if (sample_type & PERF_SAMPLE_TID) - size += sizeof(data->tid) * 2; - - if (sample_type & PERF_SAMPLE_TIME) - size += sizeof(data->time); - - if (sample_type & PERF_SAMPLE_ID) - size += sizeof(data->id); - - if (sample_type & PERF_SAMPLE_STREAM_ID) - size += sizeof(data->stream_id); - - if (sample_type & PERF_SAMPLE_CPU) - size += sizeof(data->cpu) * 2; -out: - session->id_hdr_size = size; -} - void perf_session__update_sample_type(struct perf_session *self) { self->sample_type = perf_evlist__sample_type(self->evlist); self->sample_size = __perf_evsel__sample_size(self->sample_type); self->sample_id_all = perf_evlist__sample_id_all(self->evlist); - perf_session__id_header_size(self); + self->id_hdr_size = perf_evlist__id_hdr_size(self->evlist); } int perf_session__create_kernel_maps(struct perf_session *self) -- cgit v1.2.2 From 10d0f086df77f3ff259b46cb501362dbaf2c7989 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 11 Nov 2011 22:45:41 -0200 Subject: perf event: perf_event_ops->attr() manipulates only an evlist Removing another case where a perf_session is required when processing events. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-ug1wtjbnva4bxwknflkkrlrh@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/header.c | 19 ++++++++----------- tools/perf/util/header.h | 2 +- tools/perf/util/session.c | 16 ++++++++++++++-- tools/perf/util/session.h | 6 ++++-- 4 files changed, 27 insertions(+), 16 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 41424a16be8e..1fa97dd21200 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -2120,23 +2120,23 @@ int perf_session__synthesize_attrs(struct perf_session *session, } int perf_event__process_attr(union perf_event *event, - struct perf_session *session) + struct perf_evlist **pevlist) { unsigned int i, ids, n_ids; struct perf_evsel *evsel; + struct perf_evlist *evlist = *pevlist; - if (session->evlist == NULL) { - session->evlist = perf_evlist__new(NULL, NULL); - if (session->evlist == NULL) + if (evlist == NULL) { + *pevlist = evlist = perf_evlist__new(NULL, NULL); + if (evlist == NULL) return -ENOMEM; } - evsel = perf_evsel__new(&event->attr.attr, - session->evlist->nr_entries); + evsel = perf_evsel__new(&event->attr.attr, evlist->nr_entries); if (evsel == NULL) return -ENOMEM; - perf_evlist__add(session->evlist, evsel); + perf_evlist__add(evlist, evsel); ids = event->header.size; ids -= (void *)&event->attr.id - (void *)event; @@ -2150,12 +2150,9 @@ int perf_event__process_attr(union perf_event *event, return -ENOMEM; for (i = 0; i < n_ids; i++) { - perf_evlist__id_add(session->evlist, evsel, 0, i, - event->attr.id[i]); + perf_evlist__id_add(evlist, evsel, 0, i, event->attr.id[i]); } - perf_session__update_sample_type(session); - return 0; } diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index 3d5a742f4a2a..0a88982bc392 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -101,7 +101,7 @@ int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, struct perf_session *session); int perf_session__synthesize_attrs(struct perf_session *session, perf_event__handler_t process); -int perf_event__process_attr(union perf_event *event, struct perf_session *session); +int perf_event__process_attr(union perf_event *event, struct perf_evlist **pevlist); int perf_event__synthesize_event_type(u64 event_id, char *name, perf_event__handler_t process, diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 675e080f66b6..6e7d5f54b37d 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -277,6 +277,13 @@ static int process_event_synth_stub(union perf_event *event __used, return 0; } +static int process_event_synth_attr_stub(union perf_event *event __used, + struct perf_evlist **pevlist __used) +{ + dump_printf(": unhandled!\n"); + return 0; +} + static int process_event_sample_stub(union perf_event *event __used, struct perf_sample *sample __used, struct perf_evsel *evsel __used, @@ -327,7 +334,7 @@ static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) if (handler->unthrottle == NULL) handler->unthrottle = process_event_stub; if (handler->attr == NULL) - handler->attr = process_event_synth_stub; + handler->attr = process_event_synth_attr_stub; if (handler->event_type == NULL) handler->event_type = process_event_synth_stub; if (handler->tracing_data == NULL) @@ -794,12 +801,17 @@ static int perf_session__preprocess_sample(struct perf_session *session, static int perf_session__process_user_event(struct perf_session *session, union perf_event *event, struct perf_event_ops *ops, u64 file_offset) { + int err; + dump_event(session, event, file_offset, NULL); /* These events are processed right away */ switch (event->header.type) { case PERF_RECORD_HEADER_ATTR: - return ops->attr(event, session); + err = ops->attr(event, &session->evlist); + if (err == 0) + perf_session__update_sample_type(session); + return err; case PERF_RECORD_HEADER_EVENT_TYPE: return ops->event_type(event, session); case PERF_RECORD_HEADER_TRACING_DATA: diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index 76d462d3bef7..13bd5e0a0691 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -63,6 +63,8 @@ typedef int (*event_op)(union perf_event *self, struct perf_sample *sample, struct perf_session *session); typedef int (*event_synth_op)(union perf_event *self, struct perf_session *session); +typedef int (*event_attr_op)(union perf_event *event, + struct perf_evlist **pevlist); typedef int (*event_op2)(union perf_event *self, struct perf_session *session, struct perf_event_ops *ops); @@ -76,8 +78,8 @@ struct perf_event_ops { read, throttle, unthrottle; - event_synth_op attr, - event_type, + event_attr_op attr; + event_synth_op event_type, tracing_data, build_id; event_op2 finished_round; -- cgit v1.2.2 From 246d4ce8107ea16521384c8b2a8fcff354ef2b7c Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 11 Nov 2011 23:10:26 -0200 Subject: perf session: Remove superfluous callchain_cursor member Since we have it in evsel->hists.callchain_cursor, remove it from perf_session. One more step in disentangling several places from requiring a perf_session pointer. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-rxr5dj3di7ckyfmnz0naku1z@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/session.c | 12 ++++++------ tools/perf/util/session.h | 5 ++--- 2 files changed, 8 insertions(+), 9 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 6e7d5f54b37d..734358b51ed1 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -216,7 +216,7 @@ static bool symbol__match_parent_regex(struct symbol *sym) return 0; } -int perf_session__resolve_callchain(struct perf_session *self, +int perf_session__resolve_callchain(struct perf_session *self, struct perf_evsel *evsel, struct thread *thread, struct ip_callchain *chain, struct symbol **parent) @@ -225,7 +225,7 @@ int perf_session__resolve_callchain(struct perf_session *self, unsigned int i; int err; - callchain_cursor_reset(&self->callchain_cursor); + callchain_cursor_reset(&evsel->hists.callchain_cursor); for (i = 0; i < chain->nr; i++) { u64 ip; @@ -261,7 +261,7 @@ int perf_session__resolve_callchain(struct perf_session *self, break; } - err = callchain_cursor_append(&self->callchain_cursor, + err = callchain_cursor_append(&evsel->hists.callchain_cursor, ip, al.map, al.sym); if (err) return err; @@ -1254,14 +1254,14 @@ struct perf_evsel *perf_session__find_first_evtype(struct perf_session *session, return NULL; } -void perf_session__print_ip(union perf_event *event, +void perf_session__print_ip(union perf_event *event, struct perf_evsel *evsel, struct perf_sample *sample, struct perf_session *session, int print_sym, int print_dso) { struct addr_location al; const char *symname, *dsoname; - struct callchain_cursor *cursor = &session->callchain_cursor; + struct callchain_cursor *cursor = &evsel->hists.callchain_cursor; struct callchain_cursor_node *node; if (perf_event__preprocess_sample(event, session, &al, sample, @@ -1273,7 +1273,7 @@ void perf_session__print_ip(union perf_event *event, if (symbol_conf.use_callchain && sample->callchain) { - if (perf_session__resolve_callchain(session, al.thread, + if (perf_session__resolve_callchain(session, evsel, al.thread, sample->callchain, NULL) != 0) { if (verbose) error("Failed to resolve callchain. Skipping\n"); diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index 13bd5e0a0691..d2f430367713 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -50,7 +50,6 @@ struct perf_session { int cwdlen; char *cwd; struct ordered_samples ordered_samples; - struct callchain_cursor callchain_cursor; char filename[0]; }; @@ -100,7 +99,7 @@ int __perf_session__process_events(struct perf_session *self, int perf_session__process_events(struct perf_session *self, struct perf_event_ops *event_ops); -int perf_session__resolve_callchain(struct perf_session *self, +int perf_session__resolve_callchain(struct perf_session *self, struct perf_evsel *evsel, struct thread *thread, struct ip_callchain *chain, struct symbol **parent); @@ -169,7 +168,7 @@ static inline int perf_session__parse_sample(struct perf_session *session, struct perf_evsel *perf_session__find_first_evtype(struct perf_session *session, unsigned int type); -void perf_session__print_ip(union perf_event *event, +void perf_session__print_ip(union perf_event *event, struct perf_evsel *evsel, struct perf_sample *sample, struct perf_session *session, int print_sym, int print_dso); -- cgit v1.2.2 From d20deb64e0490ee9442b5181bc08a62d2cadcb90 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 25 Nov 2011 08:19:45 -0200 Subject: perf tools: Pass tool context in the the perf_event_ops functions So that we don't need to have that many globals. Next steps will remove the 'session' pointer, that in most cases is not needed. Then we can rename perf_event_ops to 'perf_tool' that better describes this class hierarchy. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-wp4djox7x6w1i2bab1pt4xxp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/build-id.c | 7 +++-- tools/perf/util/callchain.h | 3 +++ tools/perf/util/event.c | 66 ++++++++++++++++++++++++++------------------- tools/perf/util/event.h | 38 +++++++++++++++++--------- tools/perf/util/header.c | 36 +++++++++++++++---------- tools/perf/util/header.h | 27 ++++++++++++------- tools/perf/util/session.c | 60 ++++++++++++++++++++++++----------------- tools/perf/util/session.h | 23 +++++++++------- tools/perf/util/top.h | 3 ++- 9 files changed, 163 insertions(+), 100 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c index f2fe6ec08945..0e4de1865013 100644 --- a/tools/perf/util/build-id.c +++ b/tools/perf/util/build-id.c @@ -13,8 +13,10 @@ #include "symbol.h" #include #include "debug.h" +#include "session.h" -static int build_id__mark_dso_hit(union perf_event *event, +static int build_id__mark_dso_hit(struct perf_event_ops *ops __used, + union perf_event *event, struct perf_sample *sample __used, struct perf_evsel *evsel __used, struct perf_session *session) @@ -38,7 +40,8 @@ static int build_id__mark_dso_hit(union perf_event *event, return 0; } -static int perf_event__exit_del_thread(union perf_event *event, +static int perf_event__exit_del_thread(struct perf_event_ops *ops __used, + union perf_event *event, struct perf_sample *sample __used, struct perf_session *session) { diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index 9b4ff16cac96..7f9c0f1ae3a9 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -101,6 +101,9 @@ int callchain_append(struct callchain_root *root, int callchain_merge(struct callchain_cursor *cursor, struct callchain_root *dst, struct callchain_root *src); +struct ip_callchain; +union perf_event; + bool ip_callchain__valid(struct ip_callchain *chain, const union perf_event *event); /* diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 437f8ca679a0..4800f38c7277 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -44,7 +44,8 @@ static struct perf_sample synth_sample = { .period = 1, }; -static pid_t perf_event__synthesize_comm(union perf_event *event, pid_t pid, +static pid_t perf_event__synthesize_comm(struct perf_event_ops *ops, + union perf_event *event, pid_t pid, int full, perf_event__handler_t process, struct perf_session *session) { @@ -99,7 +100,7 @@ out_race: if (!full) { event->comm.tid = pid; - process(event, &synth_sample, session); + process(ops, event, &synth_sample, session); goto out; } @@ -117,7 +118,7 @@ out_race: event->comm.tid = pid; - process(event, &synth_sample, session); + process(ops, event, &synth_sample, session); } closedir(tasks); @@ -127,7 +128,8 @@ out: return tgid; } -static int perf_event__synthesize_mmap_events(union perf_event *event, +static int perf_event__synthesize_mmap_events(struct perf_event_ops *ops, + union perf_event *event, pid_t pid, pid_t tgid, perf_event__handler_t process, struct perf_session *session) @@ -198,7 +200,7 @@ static int perf_event__synthesize_mmap_events(union perf_event *event, event->mmap.pid = tgid; event->mmap.tid = pid; - process(event, &synth_sample, session); + process(ops, event, &synth_sample, session); } } @@ -206,7 +208,8 @@ static int perf_event__synthesize_mmap_events(union perf_event *event, return 0; } -int perf_event__synthesize_modules(perf_event__handler_t process, +int perf_event__synthesize_modules(struct perf_event_ops *ops, + perf_event__handler_t process, struct perf_session *session, struct machine *machine) { @@ -251,7 +254,7 @@ int perf_event__synthesize_modules(perf_event__handler_t process, memcpy(event->mmap.filename, pos->dso->long_name, pos->dso->long_name_len + 1); - process(event, &synth_sample, session); + process(ops, event, &synth_sample, session); } free(event); @@ -261,17 +264,19 @@ int perf_event__synthesize_modules(perf_event__handler_t process, static int __event__synthesize_thread(union perf_event *comm_event, union perf_event *mmap_event, pid_t pid, perf_event__handler_t process, + struct perf_event_ops *ops, struct perf_session *session) { - pid_t tgid = perf_event__synthesize_comm(comm_event, pid, 1, process, + pid_t tgid = perf_event__synthesize_comm(ops, comm_event, pid, 1, process, session); if (tgid == -1) return -1; - return perf_event__synthesize_mmap_events(mmap_event, pid, tgid, + return perf_event__synthesize_mmap_events(ops, mmap_event, pid, tgid, process, session); } -int perf_event__synthesize_thread_map(struct thread_map *threads, +int perf_event__synthesize_thread_map(struct perf_event_ops *ops, + struct thread_map *threads, perf_event__handler_t process, struct perf_session *session) { @@ -290,7 +295,7 @@ int perf_event__synthesize_thread_map(struct thread_map *threads, for (thread = 0; thread < threads->nr; ++thread) { if (__event__synthesize_thread(comm_event, mmap_event, threads->map[thread], - process, session)) { + process, ops, session)) { err = -1; break; } @@ -302,7 +307,8 @@ out: return err; } -int perf_event__synthesize_threads(perf_event__handler_t process, +int perf_event__synthesize_threads(struct perf_event_ops *ops, + perf_event__handler_t process, struct perf_session *session) { DIR *proc; @@ -330,7 +336,7 @@ int perf_event__synthesize_threads(perf_event__handler_t process, continue; __event__synthesize_thread(comm_event, mmap_event, pid, - process, session); + process, ops, session); } closedir(proc); @@ -365,7 +371,8 @@ static int find_symbol_cb(void *arg, const char *name, char type, return 1; } -int perf_event__synthesize_kernel_mmap(perf_event__handler_t process, +int perf_event__synthesize_kernel_mmap(struct perf_event_ops *ops, + perf_event__handler_t process, struct perf_session *session, struct machine *machine, const char *symbol_name) @@ -423,13 +430,14 @@ int perf_event__synthesize_kernel_mmap(perf_event__handler_t process, event->mmap.len = map->end - event->mmap.start; event->mmap.pid = machine->pid; - err = process(event, &synth_sample, session); + err = process(ops, event, &synth_sample, session); free(event); return err; } -int perf_event__process_comm(union perf_event *event, +int perf_event__process_comm(struct perf_event_ops *ops __used, + union perf_event *event, struct perf_sample *sample __used, struct perf_session *session) { @@ -445,7 +453,8 @@ int perf_event__process_comm(union perf_event *event, return 0; } -int perf_event__process_lost(union perf_event *event, +int perf_event__process_lost(struct perf_event_ops *ops __used, + union perf_event *event, struct perf_sample *sample __used, struct perf_session *session) { @@ -468,7 +477,8 @@ static void perf_event__set_kernel_mmap_len(union perf_event *event, maps[MAP__FUNCTION]->end = ~0ULL; } -static int perf_event__process_kernel_mmap(union perf_event *event, +static int perf_event__process_kernel_mmap(struct perf_event_ops *ops __used, + union perf_event *event, struct perf_session *session) { struct map *map; @@ -567,7 +577,8 @@ out_problem: return -1; } -int perf_event__process_mmap(union perf_event *event, +int perf_event__process_mmap(struct perf_event_ops *ops, + union perf_event *event, struct perf_sample *sample __used, struct perf_session *session) { @@ -583,7 +594,7 @@ int perf_event__process_mmap(union perf_event *event, if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL || cpumode == PERF_RECORD_MISC_KERNEL) { - ret = perf_event__process_kernel_mmap(event, session); + ret = perf_event__process_kernel_mmap(ops, event, session); if (ret < 0) goto out_problem; return 0; @@ -610,7 +621,8 @@ out_problem: return 0; } -int perf_event__process_task(union perf_event *event, +int perf_event__process_task(struct perf_event_ops *ops __used, + union perf_event *event, struct perf_sample *sample __used, struct perf_session *session) { @@ -634,22 +646,22 @@ int perf_event__process_task(union perf_event *event, return 0; } -int perf_event__process(union perf_event *event, struct perf_sample *sample, - struct perf_session *session) +int perf_event__process(struct perf_event_ops *ops, union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { switch (event->header.type) { case PERF_RECORD_COMM: - perf_event__process_comm(event, sample, session); + perf_event__process_comm(ops, event, sample, session); break; case PERF_RECORD_MMAP: - perf_event__process_mmap(event, sample, session); + perf_event__process_mmap(ops, event, sample, session); break; case PERF_RECORD_FORK: case PERF_RECORD_EXIT: - perf_event__process_task(event, sample, session); + perf_event__process_task(ops, event, sample, session); break; case PERF_RECORD_LOST: - perf_event__process_lost(event, sample, session); + perf_event__process_lost(ops, event, sample, session); default: break; } diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index 357a85b85248..669409d35710 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -141,38 +141,52 @@ union perf_event { void perf_event__print_totals(void); +struct perf_event_ops; struct perf_session; struct thread_map; -typedef int (*perf_event__handler_synth_t)(union perf_event *event, - struct perf_session *session); -typedef int (*perf_event__handler_t)(union perf_event *event, +typedef int (*perf_event__handler_t)(struct perf_event_ops *ops, + union perf_event *event, struct perf_sample *sample, struct perf_session *session); -int perf_event__synthesize_thread_map(struct thread_map *threads, +int perf_event__synthesize_thread_map(struct perf_event_ops *ops, + struct thread_map *threads, perf_event__handler_t process, struct perf_session *session); -int perf_event__synthesize_threads(perf_event__handler_t process, +int perf_event__synthesize_threads(struct perf_event_ops *ops, + perf_event__handler_t process, struct perf_session *session); -int perf_event__synthesize_kernel_mmap(perf_event__handler_t process, +int perf_event__synthesize_kernel_mmap(struct perf_event_ops *ops, + perf_event__handler_t process, struct perf_session *session, struct machine *machine, const char *symbol_name); -int perf_event__synthesize_modules(perf_event__handler_t process, +int perf_event__synthesize_modules(struct perf_event_ops *ops, + perf_event__handler_t process, struct perf_session *session, struct machine *machine); -int perf_event__process_comm(union perf_event *event, struct perf_sample *sample, +int perf_event__process_comm(struct perf_event_ops *ops, + union perf_event *event, + struct perf_sample *sample, struct perf_session *session); -int perf_event__process_lost(union perf_event *event, struct perf_sample *sample, +int perf_event__process_lost(struct perf_event_ops *ops, + union perf_event *event, + struct perf_sample *sample, struct perf_session *session); -int perf_event__process_mmap(union perf_event *event, struct perf_sample *sample, +int perf_event__process_mmap(struct perf_event_ops *ops, + union perf_event *event, + struct perf_sample *sample, struct perf_session *session); -int perf_event__process_task(union perf_event *event, struct perf_sample *sample, +int perf_event__process_task(struct perf_event_ops *ops, + union perf_event *event, + struct perf_sample *sample, struct perf_session *session); -int perf_event__process(union perf_event *event, struct perf_sample *sample, +int perf_event__process(struct perf_event_ops *ops, + union perf_event *event, + struct perf_sample *sample, struct perf_session *session); struct addr_location; diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 1fa97dd21200..ab3a2b0e8f06 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -2070,7 +2070,8 @@ out_delete_evlist: return -ENOMEM; } -int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, +int perf_event__synthesize_attr(struct perf_event_ops *ops, + struct perf_event_attr *attr, u16 ids, u64 *id, perf_event__handler_t process, struct perf_session *session) { @@ -2094,21 +2095,22 @@ int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, ev->attr.header.type = PERF_RECORD_HEADER_ATTR; ev->attr.header.size = size; - err = process(ev, NULL, session); + err = process(ops, ev, NULL, session); free(ev); return err; } -int perf_session__synthesize_attrs(struct perf_session *session, +int perf_event__synthesize_attrs(struct perf_event_ops *ops, + struct perf_session *session, perf_event__handler_t process) { struct perf_evsel *attr; int err = 0; list_for_each_entry(attr, &session->evlist->entries, node) { - err = perf_event__synthesize_attr(&attr->attr, attr->ids, + err = perf_event__synthesize_attr(ops, &attr->attr, attr->ids, attr->id, process, session); if (err) { pr_debug("failed to create perf header attribute\n"); @@ -2156,7 +2158,8 @@ int perf_event__process_attr(union perf_event *event, return 0; } -int perf_event__synthesize_event_type(u64 event_id, char *name, +int perf_event__synthesize_event_type(struct perf_event_ops *ops, + u64 event_id, char *name, perf_event__handler_t process, struct perf_session *session) { @@ -2176,12 +2179,13 @@ int perf_event__synthesize_event_type(u64 event_id, char *name, ev.event_type.header.size = sizeof(ev.event_type) - (sizeof(ev.event_type.event_type.name) - size); - err = process(&ev, NULL, session); + err = process(ops, &ev, NULL, session); return err; } -int perf_event__synthesize_event_types(perf_event__handler_t process, +int perf_event__synthesize_event_types(struct perf_event_ops *ops, + perf_event__handler_t process, struct perf_session *session) { struct perf_trace_event_type *type; @@ -2190,7 +2194,7 @@ int perf_event__synthesize_event_types(perf_event__handler_t process, for (i = 0; i < event_count; i++) { type = &events[i]; - err = perf_event__synthesize_event_type(type->event_id, + err = perf_event__synthesize_event_type(ops, type->event_id, type->name, process, session); if (err) { @@ -2202,7 +2206,8 @@ int perf_event__synthesize_event_types(perf_event__handler_t process, return err; } -int perf_event__process_event_type(union perf_event *event, +int perf_event__process_event_type(struct perf_event_ops *ops __unused, + union perf_event *event, struct perf_session *session __unused) { if (perf_header__push_event(event->event_type.event_type.event_id, @@ -2212,7 +2217,8 @@ int perf_event__process_event_type(union perf_event *event, return 0; } -int perf_event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, +int perf_event__synthesize_tracing_data(struct perf_event_ops *ops, int fd, + struct perf_evlist *evlist, perf_event__handler_t process, struct perf_session *session __unused) { @@ -2245,7 +2251,7 @@ int perf_event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, ev.tracing_data.header.size = sizeof(ev.tracing_data); ev.tracing_data.size = aligned_size; - process(&ev, NULL, session); + process(ops, &ev, NULL, session); /* * The put function will copy all the tracing data @@ -2287,7 +2293,8 @@ int perf_event__process_tracing_data(union perf_event *event, return size_read + padding; } -int perf_event__synthesize_build_id(struct dso *pos, u16 misc, +int perf_event__synthesize_build_id(struct perf_event_ops *ops, + struct dso *pos, u16 misc, perf_event__handler_t process, struct machine *machine, struct perf_session *session) @@ -2310,12 +2317,13 @@ int perf_event__synthesize_build_id(struct dso *pos, u16 misc, ev.build_id.header.size = sizeof(ev.build_id) + len; memcpy(&ev.build_id.filename, pos->long_name, pos->long_name_len); - err = process(&ev, NULL, session); + err = process(ops, &ev, NULL, session); return err; } -int perf_event__process_build_id(union perf_event *event, +int perf_event__process_build_id(struct perf_event_ops *ops __used, + union perf_event *event, struct perf_session *session) { __event_process_build_id(&event->build_id, diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index 0a88982bc392..54dae5f09556 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -68,6 +68,7 @@ struct perf_header { }; struct perf_evlist; +struct perf_session; int perf_session__read_header(struct perf_session *session, int fd); int perf_session__write_header(struct perf_session *session, @@ -96,32 +97,40 @@ int build_id_cache__add_s(const char *sbuild_id, const char *debugdir, const char *name, bool is_kallsyms); int build_id_cache__remove_s(const char *sbuild_id, const char *debugdir); -int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, +int perf_event__synthesize_attr(struct perf_event_ops *ops, + struct perf_event_attr *attr, u16 ids, u64 *id, perf_event__handler_t process, struct perf_session *session); -int perf_session__synthesize_attrs(struct perf_session *session, - perf_event__handler_t process); +int perf_event__synthesize_attrs(struct perf_event_ops *ops, + struct perf_session *session, + perf_event__handler_t process); int perf_event__process_attr(union perf_event *event, struct perf_evlist **pevlist); -int perf_event__synthesize_event_type(u64 event_id, char *name, +int perf_event__synthesize_event_type(struct perf_event_ops *ops, + u64 event_id, char *name, perf_event__handler_t process, struct perf_session *session); -int perf_event__synthesize_event_types(perf_event__handler_t process, +int perf_event__synthesize_event_types(struct perf_event_ops *ops, + perf_event__handler_t process, struct perf_session *session); -int perf_event__process_event_type(union perf_event *event, +int perf_event__process_event_type(struct perf_event_ops *ops, + union perf_event *event, struct perf_session *session); -int perf_event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, +int perf_event__synthesize_tracing_data(struct perf_event_ops *ops, + int fd, struct perf_evlist *evlist, perf_event__handler_t process, struct perf_session *session); int perf_event__process_tracing_data(union perf_event *event, struct perf_session *session); -int perf_event__synthesize_build_id(struct dso *pos, u16 misc, +int perf_event__synthesize_build_id(struct perf_event_ops *ops, + struct dso *pos, u16 misc, perf_event__handler_t process, struct machine *machine, struct perf_session *session); -int perf_event__process_build_id(union perf_event *event, +int perf_event__process_build_id(struct perf_event_ops *ops, + union perf_event *event, struct perf_session *session); /* diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 734358b51ed1..a36023a66779 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -270,13 +270,21 @@ int perf_session__resolve_callchain(struct perf_session *self, struct perf_evsel return 0; } -static int process_event_synth_stub(union perf_event *event __used, +static int process_event_synth_stub(struct perf_event_ops *ops __used, + union perf_event *event __used, struct perf_session *session __used) { dump_printf(": unhandled!\n"); return 0; } +static int process_event_synth_tracing_data_stub(union perf_event *event __used, + struct perf_session *session __used) +{ + dump_printf(": unhandled!\n"); + return 0; +} + static int process_event_synth_attr_stub(union perf_event *event __used, struct perf_evlist **pevlist __used) { @@ -284,7 +292,8 @@ static int process_event_synth_attr_stub(union perf_event *event __used, return 0; } -static int process_event_sample_stub(union perf_event *event __used, +static int process_event_sample_stub(struct perf_event_ops *ops __used, + union perf_event *event __used, struct perf_sample *sample __used, struct perf_evsel *evsel __used, struct perf_session *session __used) @@ -293,7 +302,8 @@ static int process_event_sample_stub(union perf_event *event __used, return 0; } -static int process_event_stub(union perf_event *event __used, +static int process_event_stub(struct perf_event_ops *ops __used, + union perf_event *event __used, struct perf_sample *sample __used, struct perf_session *session __used) { @@ -301,17 +311,17 @@ static int process_event_stub(union perf_event *event __used, return 0; } -static int process_finished_round_stub(union perf_event *event __used, - struct perf_session *session __used, - struct perf_event_ops *ops __used) +static int process_finished_round_stub(struct perf_event_ops *ops __used, + union perf_event *event __used, + struct perf_session *session __used) { dump_printf(": unhandled!\n"); return 0; } -static int process_finished_round(union perf_event *event, - struct perf_session *session, - struct perf_event_ops *ops); +static int process_finished_round(struct perf_event_ops *ops, + union perf_event *event, + struct perf_session *session); static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) { @@ -338,7 +348,7 @@ static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) if (handler->event_type == NULL) handler->event_type = process_event_synth_stub; if (handler->tracing_data == NULL) - handler->tracing_data = process_event_synth_stub; + handler->tracing_data = process_event_synth_tracing_data_stub; if (handler->build_id == NULL) handler->build_id = process_event_synth_stub; if (handler->finished_round == NULL) { @@ -565,9 +575,9 @@ static void flush_sample_queue(struct perf_session *s, * Flush every events below timestamp 7 * etc... */ -static int process_finished_round(union perf_event *event __used, - struct perf_session *session, - struct perf_event_ops *ops) +static int process_finished_round(struct perf_event_ops *ops, + union perf_event *event __used, + struct perf_session *session) { flush_sample_queue(session, ops); session->ordered_samples.next_flush = session->ordered_samples.max_timestamp; @@ -759,23 +769,23 @@ static int perf_session_deliver_event(struct perf_session *session, ++session->hists.stats.nr_unknown_id; return -1; } - return ops->sample(event, sample, evsel, session); + return ops->sample(ops, event, sample, evsel, session); case PERF_RECORD_MMAP: - return ops->mmap(event, sample, session); + return ops->mmap(ops, event, sample, session); case PERF_RECORD_COMM: - return ops->comm(event, sample, session); + return ops->comm(ops, event, sample, session); case PERF_RECORD_FORK: - return ops->fork(event, sample, session); + return ops->fork(ops, event, sample, session); case PERF_RECORD_EXIT: - return ops->exit(event, sample, session); + return ops->exit(ops, event, sample, session); case PERF_RECORD_LOST: - return ops->lost(event, sample, session); + return ops->lost(ops, event, sample, session); case PERF_RECORD_READ: - return ops->read(event, sample, session); + return ops->read(ops, event, sample, session); case PERF_RECORD_THROTTLE: - return ops->throttle(event, sample, session); + return ops->throttle(ops, event, sample, session); case PERF_RECORD_UNTHROTTLE: - return ops->unthrottle(event, sample, session); + return ops->unthrottle(ops, event, sample, session); default: ++session->hists.stats.nr_unknown_events; return -1; @@ -813,15 +823,15 @@ static int perf_session__process_user_event(struct perf_session *session, union perf_session__update_sample_type(session); return err; case PERF_RECORD_HEADER_EVENT_TYPE: - return ops->event_type(event, session); + return ops->event_type(ops, event, session); case PERF_RECORD_HEADER_TRACING_DATA: /* setup for reading amidst mmap */ lseek(session->fd, file_offset, SEEK_SET); return ops->tracing_data(event, session); case PERF_RECORD_HEADER_BUILD_ID: - return ops->build_id(event, session); + return ops->build_id(ops, event, session); case PERF_RECORD_FINISHED_ROUND: - return ops->finished_round(event, session, ops); + return ops->finished_round(ops, event, session); default: return -EINVAL; } diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index d2f430367713..6de3d1368900 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -56,16 +56,18 @@ struct perf_session { struct perf_evsel; struct perf_event_ops; -typedef int (*event_sample)(union perf_event *event, struct perf_sample *sample, +typedef int (*event_sample)(struct perf_event_ops *ops, + union perf_event *event, struct perf_sample *sample, struct perf_evsel *evsel, struct perf_session *session); -typedef int (*event_op)(union perf_event *self, struct perf_sample *sample, +typedef int (*event_op)(struct perf_event_ops *ops, union perf_event *event, + struct perf_sample *sample, struct perf_session *session); typedef int (*event_synth_op)(union perf_event *self, struct perf_session *session); typedef int (*event_attr_op)(union perf_event *event, struct perf_evlist **pevlist); -typedef int (*event_op2)(union perf_event *self, struct perf_session *session, - struct perf_event_ops *ops); +typedef int (*event_op2)(struct perf_event_ops *ops, union perf_event *event, + struct perf_session *session); struct perf_event_ops { event_sample sample; @@ -78,10 +80,10 @@ struct perf_event_ops { throttle, unthrottle; event_attr_op attr; - event_synth_op event_type, - tracing_data, - build_id; - event_op2 finished_round; + event_synth_op tracing_data; + event_op2 event_type, + build_id, + finished_round; bool ordered_samples; bool ordering_requires_timestamps; }; @@ -142,10 +144,11 @@ struct machine *perf_session__findnew_machine(struct perf_session *self, pid_t p static inline void perf_session__process_machines(struct perf_session *self, + struct perf_event_ops *ops, machine__process_t process) { - process(&self->host_machine, self); - return machines__process(&self->machines, process, self); + process(&self->host_machine, ops); + return machines__process(&self->machines, process, ops); } size_t perf_session__fprintf_dsos(struct perf_session *self, FILE *fp); diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h index 399650967958..44eda6fc6b33 100644 --- a/tools/perf/util/top.h +++ b/tools/perf/util/top.h @@ -2,14 +2,15 @@ #define __PERF_TOP_H 1 #include "types.h" +#include "session.h" #include "../perf.h" #include struct perf_evlist; struct perf_evsel; -struct perf_session; struct perf_top { + struct perf_event_ops ops; struct perf_evlist *evlist; /* * Symbols will be added here in perf_event__process_sample and will -- cgit v1.2.2 From 743eb868657bdb1b26c7b24077ca21c67c82c777 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 28 Nov 2011 07:56:39 -0200 Subject: perf tools: Resolve machine earlier and pass it to perf_event_ops Reducing the exposure of perf_session further, so that we can use the classes in cases where no perf.data file is created. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-stua66dcscsezzrcdugvbmvd@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/build-id.c | 16 +-- tools/perf/util/event.c | 151 +++++++++------------ tools/perf/util/event.h | 21 ++- tools/perf/util/header.c | 28 ++-- tools/perf/util/header.h | 16 +-- tools/perf/util/map.h | 10 ++ .../perf/util/scripting-engines/trace-event-perl.c | 4 +- .../util/scripting-engines/trace-event-python.c | 4 +- tools/perf/util/session.c | 91 +++++++------ tools/perf/util/session.h | 30 ++-- tools/perf/util/thread.h | 14 +- tools/perf/util/trace-event-scripting.c | 2 +- tools/perf/util/trace-event.h | 8 +- 13 files changed, 195 insertions(+), 200 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c index 0e4de1865013..2f84c4802aca 100644 --- a/tools/perf/util/build-id.c +++ b/tools/perf/util/build-id.c @@ -19,11 +19,11 @@ static int build_id__mark_dso_hit(struct perf_event_ops *ops __used, union perf_event *event, struct perf_sample *sample __used, struct perf_evsel *evsel __used, - struct perf_session *session) + struct machine *machine) { struct addr_location al; u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; - struct thread *thread = perf_session__findnew(session, event->ip.pid); + struct thread *thread = machine__findnew_thread(machine, event->ip.pid); if (thread == NULL) { pr_err("problem processing %d event, skipping it.\n", @@ -31,8 +31,8 @@ static int build_id__mark_dso_hit(struct perf_event_ops *ops __used, return -1; } - thread__find_addr_map(thread, session, cpumode, MAP__FUNCTION, - event->ip.pid, event->ip.ip, &al); + thread__find_addr_map(thread, machine, cpumode, MAP__FUNCTION, + event->ip.ip, &al); if (al.map != NULL) al.map->dso->hit = 1; @@ -43,16 +43,16 @@ static int build_id__mark_dso_hit(struct perf_event_ops *ops __used, static int perf_event__exit_del_thread(struct perf_event_ops *ops __used, union perf_event *event, struct perf_sample *sample __used, - struct perf_session *session) + struct machine *machine) { - struct thread *thread = perf_session__findnew(session, event->fork.tid); + struct thread *thread = machine__findnew_thread(machine, event->fork.tid); dump_printf("(%d:%d):(%d:%d)\n", event->fork.pid, event->fork.tid, event->fork.ppid, event->fork.ptid); if (thread) { - rb_erase(&thread->rb_node, &session->host_machine.threads); - session->host_machine.last_match = NULL; + rb_erase(&thread->rb_node, &machine->threads); + machine->last_match = NULL; thread__delete(thread); } diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 4800f38c7277..0cdc811c48e2 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -1,7 +1,6 @@ #include #include "event.h" #include "debug.h" -#include "session.h" #include "sort.h" #include "string.h" #include "strlist.h" @@ -47,7 +46,7 @@ static struct perf_sample synth_sample = { static pid_t perf_event__synthesize_comm(struct perf_event_ops *ops, union perf_event *event, pid_t pid, int full, perf_event__handler_t process, - struct perf_session *session) + struct machine *machine) { char filename[PATH_MAX]; char bf[BUFSIZ]; @@ -93,14 +92,14 @@ out_race: event->comm.header.type = PERF_RECORD_COMM; size = ALIGN(size, sizeof(u64)); - memset(event->comm.comm + size, 0, session->id_hdr_size); + memset(event->comm.comm + size, 0, machine->id_hdr_size); event->comm.header.size = (sizeof(event->comm) - (sizeof(event->comm.comm) - size) + - session->id_hdr_size); + machine->id_hdr_size); if (!full) { event->comm.tid = pid; - process(ops, event, &synth_sample, session); + process(ops, event, &synth_sample, machine); goto out; } @@ -118,7 +117,7 @@ out_race: event->comm.tid = pid; - process(ops, event, &synth_sample, session); + process(ops, event, &synth_sample, machine); } closedir(tasks); @@ -132,7 +131,7 @@ static int perf_event__synthesize_mmap_events(struct perf_event_ops *ops, union perf_event *event, pid_t pid, pid_t tgid, perf_event__handler_t process, - struct perf_session *session) + struct machine *machine) { char filename[PATH_MAX]; FILE *fp; @@ -195,12 +194,12 @@ static int perf_event__synthesize_mmap_events(struct perf_event_ops *ops, event->mmap.len -= event->mmap.start; event->mmap.header.size = (sizeof(event->mmap) - (sizeof(event->mmap.filename) - size)); - memset(event->mmap.filename + size, 0, session->id_hdr_size); - event->mmap.header.size += session->id_hdr_size; + memset(event->mmap.filename + size, 0, machine->id_hdr_size); + event->mmap.header.size += machine->id_hdr_size; event->mmap.pid = tgid; event->mmap.tid = pid; - process(ops, event, &synth_sample, session); + process(ops, event, &synth_sample, machine); } } @@ -210,13 +209,12 @@ static int perf_event__synthesize_mmap_events(struct perf_event_ops *ops, int perf_event__synthesize_modules(struct perf_event_ops *ops, perf_event__handler_t process, - struct perf_session *session, struct machine *machine) { struct rb_node *nd; struct map_groups *kmaps = &machine->kmaps; union perf_event *event = zalloc((sizeof(event->mmap) + - session->id_hdr_size)); + machine->id_hdr_size)); if (event == NULL) { pr_debug("Not enough memory synthesizing mmap event " "for kernel modules\n"); @@ -246,15 +244,15 @@ int perf_event__synthesize_modules(struct perf_event_ops *ops, event->mmap.header.type = PERF_RECORD_MMAP; event->mmap.header.size = (sizeof(event->mmap) - (sizeof(event->mmap.filename) - size)); - memset(event->mmap.filename + size, 0, session->id_hdr_size); - event->mmap.header.size += session->id_hdr_size; + memset(event->mmap.filename + size, 0, machine->id_hdr_size); + event->mmap.header.size += machine->id_hdr_size; event->mmap.start = pos->start; event->mmap.len = pos->end - pos->start; event->mmap.pid = machine->pid; memcpy(event->mmap.filename, pos->dso->long_name, pos->dso->long_name_len + 1); - process(ops, event, &synth_sample, session); + process(ops, event, &synth_sample, machine); } free(event); @@ -265,29 +263,29 @@ static int __event__synthesize_thread(union perf_event *comm_event, union perf_event *mmap_event, pid_t pid, perf_event__handler_t process, struct perf_event_ops *ops, - struct perf_session *session) + struct machine *machine) { - pid_t tgid = perf_event__synthesize_comm(ops, comm_event, pid, 1, process, - session); + pid_t tgid = perf_event__synthesize_comm(ops, comm_event, pid, 1, + process, machine); if (tgid == -1) return -1; return perf_event__synthesize_mmap_events(ops, mmap_event, pid, tgid, - process, session); + process, machine); } int perf_event__synthesize_thread_map(struct perf_event_ops *ops, struct thread_map *threads, perf_event__handler_t process, - struct perf_session *session) + struct machine *machine) { union perf_event *comm_event, *mmap_event; int err = -1, thread; - comm_event = malloc(sizeof(comm_event->comm) + session->id_hdr_size); + comm_event = malloc(sizeof(comm_event->comm) + machine->id_hdr_size); if (comm_event == NULL) goto out; - mmap_event = malloc(sizeof(mmap_event->mmap) + session->id_hdr_size); + mmap_event = malloc(sizeof(mmap_event->mmap) + machine->id_hdr_size); if (mmap_event == NULL) goto out_free_comm; @@ -295,7 +293,7 @@ int perf_event__synthesize_thread_map(struct perf_event_ops *ops, for (thread = 0; thread < threads->nr; ++thread) { if (__event__synthesize_thread(comm_event, mmap_event, threads->map[thread], - process, ops, session)) { + process, ops, machine)) { err = -1; break; } @@ -309,18 +307,18 @@ out: int perf_event__synthesize_threads(struct perf_event_ops *ops, perf_event__handler_t process, - struct perf_session *session) + struct machine *machine) { DIR *proc; struct dirent dirent, *next; union perf_event *comm_event, *mmap_event; int err = -1; - comm_event = malloc(sizeof(comm_event->comm) + session->id_hdr_size); + comm_event = malloc(sizeof(comm_event->comm) + machine->id_hdr_size); if (comm_event == NULL) goto out; - mmap_event = malloc(sizeof(mmap_event->mmap) + session->id_hdr_size); + mmap_event = malloc(sizeof(mmap_event->mmap) + machine->id_hdr_size); if (mmap_event == NULL) goto out_free_comm; @@ -336,7 +334,7 @@ int perf_event__synthesize_threads(struct perf_event_ops *ops, continue; __event__synthesize_thread(comm_event, mmap_event, pid, - process, ops, session); + process, ops, machine); } closedir(proc); @@ -373,7 +371,6 @@ static int find_symbol_cb(void *arg, const char *name, char type, int perf_event__synthesize_kernel_mmap(struct perf_event_ops *ops, perf_event__handler_t process, - struct perf_session *session, struct machine *machine, const char *symbol_name) { @@ -390,7 +387,7 @@ int perf_event__synthesize_kernel_mmap(struct perf_event_ops *ops, */ struct process_symbol_args args = { .name = symbol_name, }; union perf_event *event = zalloc((sizeof(event->mmap) + - session->id_hdr_size)); + machine->id_hdr_size)); if (event == NULL) { pr_debug("Not enough memory synthesizing mmap event " "for kernel modules\n"); @@ -424,13 +421,13 @@ int perf_event__synthesize_kernel_mmap(struct perf_event_ops *ops, size = ALIGN(size, sizeof(u64)); event->mmap.header.type = PERF_RECORD_MMAP; event->mmap.header.size = (sizeof(event->mmap) - - (sizeof(event->mmap.filename) - size) + session->id_hdr_size); + (sizeof(event->mmap.filename) - size) + machine->id_hdr_size); event->mmap.pgoff = args.start; event->mmap.start = map->start; event->mmap.len = map->end - event->mmap.start; event->mmap.pid = machine->pid; - err = process(ops, event, &synth_sample, session); + err = process(ops, event, &synth_sample, machine); free(event); return err; @@ -439,9 +436,9 @@ int perf_event__synthesize_kernel_mmap(struct perf_event_ops *ops, int perf_event__process_comm(struct perf_event_ops *ops __used, union perf_event *event, struct perf_sample *sample __used, - struct perf_session *session) + struct machine *machine) { - struct thread *thread = perf_session__findnew(session, event->comm.tid); + struct thread *thread = machine__findnew_thread(machine, event->comm.tid); dump_printf(": %s:%d\n", event->comm.comm, event->comm.tid); @@ -456,11 +453,10 @@ int perf_event__process_comm(struct perf_event_ops *ops __used, int perf_event__process_lost(struct perf_event_ops *ops __used, union perf_event *event, struct perf_sample *sample __used, - struct perf_session *session) + struct machine *machine __used) { dump_printf(": id:%" PRIu64 ": lost:%" PRIu64 "\n", event->lost.id, event->lost.lost); - session->hists.stats.total_lost += event->lost.lost; return 0; } @@ -479,20 +475,13 @@ static void perf_event__set_kernel_mmap_len(union perf_event *event, static int perf_event__process_kernel_mmap(struct perf_event_ops *ops __used, union perf_event *event, - struct perf_session *session) + struct machine *machine) { struct map *map; char kmmap_prefix[PATH_MAX]; - struct machine *machine; enum dso_kernel_type kernel_type; bool is_kernel_mmap; - machine = perf_session__findnew_machine(session, event->mmap.pid); - if (!machine) { - pr_err("Can't find id %d's machine\n", event->mmap.pid); - goto out_problem; - } - machine__mmap_name(machine, kmmap_prefix, sizeof(kmmap_prefix)); if (machine__is_host(machine)) kernel_type = DSO_TYPE_KERNEL; @@ -559,9 +548,9 @@ static int perf_event__process_kernel_mmap(struct perf_event_ops *ops __used, * time /proc/sys/kernel/kptr_restrict was non zero. */ if (event->mmap.pgoff != 0) { - perf_session__set_kallsyms_ref_reloc_sym(machine->vmlinux_maps, - symbol_name, - event->mmap.pgoff); + maps__set_kallsyms_ref_reloc_sym(machine->vmlinux_maps, + symbol_name, + event->mmap.pgoff); } if (machine__is_default_guest(machine)) { @@ -580,9 +569,8 @@ out_problem: int perf_event__process_mmap(struct perf_event_ops *ops, union perf_event *event, struct perf_sample *sample __used, - struct perf_session *session) + struct machine *machine) { - struct machine *machine; struct thread *thread; struct map *map; u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; @@ -594,16 +582,13 @@ int perf_event__process_mmap(struct perf_event_ops *ops, if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL || cpumode == PERF_RECORD_MISC_KERNEL) { - ret = perf_event__process_kernel_mmap(ops, event, session); + ret = perf_event__process_kernel_mmap(ops, event, machine); if (ret < 0) goto out_problem; return 0; } - machine = perf_session__find_host_machine(session); - if (machine == NULL) - goto out_problem; - thread = perf_session__findnew(session, event->mmap.pid); + thread = machine__findnew_thread(machine, event->mmap.pid); if (thread == NULL) goto out_problem; map = map__new(&machine->user_dsos, event->mmap.start, @@ -624,16 +609,16 @@ out_problem: int perf_event__process_task(struct perf_event_ops *ops __used, union perf_event *event, struct perf_sample *sample __used, - struct perf_session *session) + struct machine *machine) { - struct thread *thread = perf_session__findnew(session, event->fork.tid); - struct thread *parent = perf_session__findnew(session, event->fork.ptid); + struct thread *thread = machine__findnew_thread(machine, event->fork.tid); + struct thread *parent = machine__findnew_thread(machine, event->fork.ptid); dump_printf("(%d:%d):(%d:%d)\n", event->fork.pid, event->fork.tid, event->fork.ppid, event->fork.ptid); if (event->header.type == PERF_RECORD_EXIT) { - perf_session__remove_thread(session, thread); + machine__remove_thread(machine, thread); return 0; } @@ -647,21 +632,21 @@ int perf_event__process_task(struct perf_event_ops *ops __used, } int perf_event__process(struct perf_event_ops *ops, union perf_event *event, - struct perf_sample *sample, struct perf_session *session) + struct perf_sample *sample, struct machine *machine) { switch (event->header.type) { case PERF_RECORD_COMM: - perf_event__process_comm(ops, event, sample, session); + perf_event__process_comm(ops, event, sample, machine); break; case PERF_RECORD_MMAP: - perf_event__process_mmap(ops, event, sample, session); + perf_event__process_mmap(ops, event, sample, machine); break; case PERF_RECORD_FORK: case PERF_RECORD_EXIT: - perf_event__process_task(ops, event, sample, session); + perf_event__process_task(ops, event, sample, machine); break; case PERF_RECORD_LOST: - perf_event__process_lost(ops, event, sample, session); + perf_event__process_lost(ops, event, sample, machine); default: break; } @@ -670,36 +655,29 @@ int perf_event__process(struct perf_event_ops *ops, union perf_event *event, } void thread__find_addr_map(struct thread *self, - struct perf_session *session, u8 cpumode, - enum map_type type, pid_t pid, u64 addr, + struct machine *machine, u8 cpumode, + enum map_type type, u64 addr, struct addr_location *al) { struct map_groups *mg = &self->mg; - struct machine *machine = NULL; al->thread = self; al->addr = addr; al->cpumode = cpumode; al->filtered = false; + if (machine == NULL) { + al->map = NULL; + return; + } + if (cpumode == PERF_RECORD_MISC_KERNEL && perf_host) { al->level = 'k'; - machine = perf_session__find_host_machine(session); - if (machine == NULL) { - al->map = NULL; - return; - } mg = &machine->kmaps; } else if (cpumode == PERF_RECORD_MISC_USER && perf_host) { al->level = '.'; - machine = perf_session__find_host_machine(session); } else if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL && perf_guest) { al->level = 'g'; - machine = perf_session__find_machine(session, pid); - if (machine == NULL) { - al->map = NULL; - return; - } mg = &machine->kmaps; } else { /* @@ -745,13 +723,12 @@ try_again: al->addr = al->map->map_ip(al->map, al->addr); } -void thread__find_addr_location(struct thread *self, - struct perf_session *session, u8 cpumode, - enum map_type type, pid_t pid, u64 addr, +void thread__find_addr_location(struct thread *thread, struct machine *machine, + u8 cpumode, enum map_type type, u64 addr, struct addr_location *al, symbol_filter_t filter) { - thread__find_addr_map(self, session, cpumode, type, pid, addr, al); + thread__find_addr_map(thread, machine, cpumode, type, addr, al); if (al->map != NULL) al->sym = map__find_symbol(al->map, al->addr, filter); else @@ -759,13 +736,13 @@ void thread__find_addr_location(struct thread *self, } int perf_event__preprocess_sample(const union perf_event *event, - struct perf_session *session, + struct machine *machine, struct addr_location *al, struct perf_sample *sample, symbol_filter_t filter) { u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; - struct thread *thread = perf_session__findnew(session, event->ip.pid); + struct thread *thread = machine__findnew_thread(machine, event->ip.pid); if (thread == NULL) return -1; @@ -776,18 +753,18 @@ int perf_event__preprocess_sample(const union perf_event *event, dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid); /* - * Have we already created the kernel maps for the host machine? + * Have we already created the kernel maps for this machine? * * This should have happened earlier, when we processed the kernel MMAP * events, but for older perf.data files there was no such thing, so do * it now. */ if (cpumode == PERF_RECORD_MISC_KERNEL && - session->host_machine.vmlinux_maps[MAP__FUNCTION] == NULL) - machine__create_kernel_maps(&session->host_machine); + machine->vmlinux_maps[MAP__FUNCTION] == NULL) + machine__create_kernel_maps(machine); - thread__find_addr_map(thread, session, cpumode, MAP__FUNCTION, - event->ip.pid, event->ip.ip, al); + thread__find_addr_map(thread, machine, cpumode, MAP__FUNCTION, + event->ip.ip, al); dump_printf(" ...... dso: %s\n", al->map ? al->map->dso->long_name : al->level == 'H' ? "[hypervisor]" : ""); diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index 669409d35710..1564877e8703 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -142,56 +142,53 @@ union perf_event { void perf_event__print_totals(void); struct perf_event_ops; -struct perf_session; struct thread_map; typedef int (*perf_event__handler_t)(struct perf_event_ops *ops, union perf_event *event, struct perf_sample *sample, - struct perf_session *session); + struct machine *machine); int perf_event__synthesize_thread_map(struct perf_event_ops *ops, struct thread_map *threads, perf_event__handler_t process, - struct perf_session *session); + struct machine *machine); int perf_event__synthesize_threads(struct perf_event_ops *ops, perf_event__handler_t process, - struct perf_session *session); + struct machine *machine); int perf_event__synthesize_kernel_mmap(struct perf_event_ops *ops, perf_event__handler_t process, - struct perf_session *session, struct machine *machine, const char *symbol_name); int perf_event__synthesize_modules(struct perf_event_ops *ops, perf_event__handler_t process, - struct perf_session *session, struct machine *machine); int perf_event__process_comm(struct perf_event_ops *ops, union perf_event *event, struct perf_sample *sample, - struct perf_session *session); + struct machine *machine); int perf_event__process_lost(struct perf_event_ops *ops, union perf_event *event, struct perf_sample *sample, - struct perf_session *session); + struct machine *machine); int perf_event__process_mmap(struct perf_event_ops *ops, union perf_event *event, struct perf_sample *sample, - struct perf_session *session); + struct machine *machine); int perf_event__process_task(struct perf_event_ops *ops, union perf_event *event, struct perf_sample *sample, - struct perf_session *session); + struct machine *machine); int perf_event__process(struct perf_event_ops *ops, union perf_event *event, struct perf_sample *sample, - struct perf_session *session); + struct machine *machine); struct addr_location; int perf_event__preprocess_sample(const union perf_event *self, - struct perf_session *session, + struct machine *machine, struct addr_location *al, struct perf_sample *sample, symbol_filter_t filter); diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index ab3a2b0e8f06..db280d6ca898 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -2072,8 +2072,7 @@ out_delete_evlist: int perf_event__synthesize_attr(struct perf_event_ops *ops, struct perf_event_attr *attr, u16 ids, u64 *id, - perf_event__handler_t process, - struct perf_session *session) + perf_event__handler_t process) { union perf_event *ev; size_t size; @@ -2095,7 +2094,7 @@ int perf_event__synthesize_attr(struct perf_event_ops *ops, ev->attr.header.type = PERF_RECORD_HEADER_ATTR; ev->attr.header.size = size; - err = process(ops, ev, NULL, session); + err = process(ops, ev, NULL, NULL); free(ev); @@ -2111,7 +2110,7 @@ int perf_event__synthesize_attrs(struct perf_event_ops *ops, list_for_each_entry(attr, &session->evlist->entries, node) { err = perf_event__synthesize_attr(ops, &attr->attr, attr->ids, - attr->id, process, session); + attr->id, process); if (err) { pr_debug("failed to create perf header attribute\n"); return err; @@ -2161,7 +2160,7 @@ int perf_event__process_attr(union perf_event *event, int perf_event__synthesize_event_type(struct perf_event_ops *ops, u64 event_id, char *name, perf_event__handler_t process, - struct perf_session *session) + struct machine *machine) { union perf_event ev; size_t size = 0; @@ -2179,14 +2178,14 @@ int perf_event__synthesize_event_type(struct perf_event_ops *ops, ev.event_type.header.size = sizeof(ev.event_type) - (sizeof(ev.event_type.event_type.name) - size); - err = process(ops, &ev, NULL, session); + err = process(ops, &ev, NULL, machine); return err; } int perf_event__synthesize_event_types(struct perf_event_ops *ops, perf_event__handler_t process, - struct perf_session *session) + struct machine *machine) { struct perf_trace_event_type *type; int i, err = 0; @@ -2196,7 +2195,7 @@ int perf_event__synthesize_event_types(struct perf_event_ops *ops, err = perf_event__synthesize_event_type(ops, type->event_id, type->name, process, - session); + machine); if (err) { pr_debug("failed to create perf header event type\n"); return err; @@ -2207,8 +2206,7 @@ int perf_event__synthesize_event_types(struct perf_event_ops *ops, } int perf_event__process_event_type(struct perf_event_ops *ops __unused, - union perf_event *event, - struct perf_session *session __unused) + union perf_event *event) { if (perf_header__push_event(event->event_type.event_type.event_id, event->event_type.event_type.name) < 0) @@ -2219,8 +2217,7 @@ int perf_event__process_event_type(struct perf_event_ops *ops __unused, int perf_event__synthesize_tracing_data(struct perf_event_ops *ops, int fd, struct perf_evlist *evlist, - perf_event__handler_t process, - struct perf_session *session __unused) + perf_event__handler_t process) { union perf_event ev; struct tracing_data *tdata; @@ -2251,7 +2248,7 @@ int perf_event__synthesize_tracing_data(struct perf_event_ops *ops, int fd, ev.tracing_data.header.size = sizeof(ev.tracing_data); ev.tracing_data.size = aligned_size; - process(ops, &ev, NULL, session); + process(ops, &ev, NULL, NULL); /* * The put function will copy all the tracing data @@ -2296,8 +2293,7 @@ int perf_event__process_tracing_data(union perf_event *event, int perf_event__synthesize_build_id(struct perf_event_ops *ops, struct dso *pos, u16 misc, perf_event__handler_t process, - struct machine *machine, - struct perf_session *session) + struct machine *machine) { union perf_event ev; size_t len; @@ -2317,7 +2313,7 @@ int perf_event__synthesize_build_id(struct perf_event_ops *ops, ev.build_id.header.size = sizeof(ev.build_id) + len; memcpy(&ev.build_id.filename, pos->long_name, pos->long_name_len); - err = process(ops, &ev, NULL, session); + err = process(ops, &ev, NULL, machine); return err; } diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index 54dae5f09556..a604962fc431 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -99,8 +99,7 @@ int build_id_cache__remove_s(const char *sbuild_id, const char *debugdir); int perf_event__synthesize_attr(struct perf_event_ops *ops, struct perf_event_attr *attr, u16 ids, u64 *id, - perf_event__handler_t process, - struct perf_session *session); + perf_event__handler_t process); int perf_event__synthesize_attrs(struct perf_event_ops *ops, struct perf_session *session, perf_event__handler_t process); @@ -109,26 +108,23 @@ int perf_event__process_attr(union perf_event *event, struct perf_evlist **pevli int perf_event__synthesize_event_type(struct perf_event_ops *ops, u64 event_id, char *name, perf_event__handler_t process, - struct perf_session *session); + struct machine *machine); int perf_event__synthesize_event_types(struct perf_event_ops *ops, perf_event__handler_t process, - struct perf_session *session); + struct machine *machine); int perf_event__process_event_type(struct perf_event_ops *ops, - union perf_event *event, - struct perf_session *session); + union perf_event *event); int perf_event__synthesize_tracing_data(struct perf_event_ops *ops, int fd, struct perf_evlist *evlist, - perf_event__handler_t process, - struct perf_session *session); + perf_event__handler_t process); int perf_event__process_tracing_data(union perf_event *event, struct perf_session *session); int perf_event__synthesize_build_id(struct perf_event_ops *ops, struct dso *pos, u16 misc, perf_event__handler_t process, - struct machine *machine, - struct perf_session *session); + struct machine *machine); int perf_event__process_build_id(struct perf_event_ops *ops, union perf_event *event, struct perf_session *session); diff --git a/tools/perf/util/map.h b/tools/perf/util/map.h index bde6835ee257..2b8017f8a930 100644 --- a/tools/perf/util/map.h +++ b/tools/perf/util/map.h @@ -18,9 +18,11 @@ enum map_type { extern const char *map_type__name[MAP__NR_TYPES]; struct dso; +struct ip_callchain; struct ref_reloc_sym; struct map_groups; struct machine; +struct perf_evsel; struct map { union { @@ -61,6 +63,7 @@ struct map_groups { struct machine { struct rb_node rb_node; pid_t pid; + u16 id_hdr_size; char *root_dir; struct rb_root threads; struct list_head dead_threads; @@ -151,6 +154,13 @@ int machine__init(struct machine *self, const char *root_dir, pid_t pid); void machine__exit(struct machine *self); void machine__delete(struct machine *self); +int machine__resolve_callchain(struct machine *machine, + struct perf_evsel *evsel, struct thread *thread, + struct ip_callchain *chain, + struct symbol **parent); +int maps__set_kallsyms_ref_reloc_sym(struct map **maps, const char *symbol_name, + u64 addr); + /* * Default guest kernel is defined by parameter --guestkallsyms * and --guestmodules diff --git a/tools/perf/util/scripting-engines/trace-event-perl.c b/tools/perf/util/scripting-engines/trace-event-perl.c index 74350ffb57fe..a82ce4303ff5 100644 --- a/tools/perf/util/scripting-engines/trace-event-perl.c +++ b/tools/perf/util/scripting-engines/trace-event-perl.c @@ -27,6 +27,8 @@ #include "../../perf.h" #include "../util.h" +#include "../thread.h" +#include "../event.h" #include "../trace-event.h" #include @@ -248,7 +250,7 @@ static inline struct event *find_cache_event(int type) static void perl_process_event(union perf_event *pevent __unused, struct perf_sample *sample, struct perf_evsel *evsel, - struct perf_session *session __unused, + struct machine *machine __unused, struct thread *thread) { struct format_field *field; diff --git a/tools/perf/util/scripting-engines/trace-event-python.c b/tools/perf/util/scripting-engines/trace-event-python.c index 6ccf70e8d8f2..0b2a48783172 100644 --- a/tools/perf/util/scripting-engines/trace-event-python.c +++ b/tools/perf/util/scripting-engines/trace-event-python.c @@ -29,6 +29,8 @@ #include "../../perf.h" #include "../util.h" +#include "../event.h" +#include "../thread.h" #include "../trace-event.h" PyMODINIT_FUNC initperf_trace_context(void); @@ -207,7 +209,7 @@ static inline struct event *find_cache_event(int type) static void python_process_event(union perf_event *pevent __unused, struct perf_sample *sample, struct perf_evsel *evsel __unused, - struct perf_session *session __unused, + struct machine *machine __unused, struct thread *thread) { PyObject *handler, *retval, *context, *t, *obj, *dict = NULL; diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index a36023a66779..be33606386bf 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -84,6 +84,7 @@ void perf_session__update_sample_type(struct perf_session *self) self->sample_size = __perf_evsel__sample_size(self->sample_type); self->sample_id_all = perf_evlist__sample_id_all(self->evlist); self->id_hdr_size = perf_evlist__id_hdr_size(self->evlist); + self->host_machine.id_hdr_size = self->id_hdr_size; } int perf_session__create_kernel_maps(struct perf_session *self) @@ -216,10 +217,10 @@ static bool symbol__match_parent_regex(struct symbol *sym) return 0; } -int perf_session__resolve_callchain(struct perf_session *self, struct perf_evsel *evsel, - struct thread *thread, - struct ip_callchain *chain, - struct symbol **parent) +int machine__resolve_callchain(struct machine *self, struct perf_evsel *evsel, + struct thread *thread, + struct ip_callchain *chain, + struct symbol **parent) { u8 cpumode = PERF_RECORD_MISC_USER; unsigned int i; @@ -252,7 +253,7 @@ int perf_session__resolve_callchain(struct perf_session *self, struct perf_evsel al.filtered = false; thread__find_addr_location(thread, self, cpumode, - MAP__FUNCTION, thread->pid, ip, &al, NULL); + MAP__FUNCTION, ip, &al, NULL); if (al.sym != NULL) { if (sort__has_parent && !*parent && symbol__match_parent_regex(al.sym)) @@ -270,14 +271,6 @@ int perf_session__resolve_callchain(struct perf_session *self, struct perf_evsel return 0; } -static int process_event_synth_stub(struct perf_event_ops *ops __used, - union perf_event *event __used, - struct perf_session *session __used) -{ - dump_printf(": unhandled!\n"); - return 0; -} - static int process_event_synth_tracing_data_stub(union perf_event *event __used, struct perf_session *session __used) { @@ -296,7 +289,7 @@ static int process_event_sample_stub(struct perf_event_ops *ops __used, union perf_event *event __used, struct perf_sample *sample __used, struct perf_evsel *evsel __used, - struct perf_session *session __used) + struct machine *machine __used) { dump_printf(": unhandled!\n"); return 0; @@ -305,7 +298,7 @@ static int process_event_sample_stub(struct perf_event_ops *ops __used, static int process_event_stub(struct perf_event_ops *ops __used, union perf_event *event __used, struct perf_sample *sample __used, - struct perf_session *session __used) + struct machine *machine __used) { dump_printf(": unhandled!\n"); return 0; @@ -313,7 +306,14 @@ static int process_event_stub(struct perf_event_ops *ops __used, static int process_finished_round_stub(struct perf_event_ops *ops __used, union perf_event *event __used, - struct perf_session *session __used) + struct perf_session *perf_session __used) +{ + dump_printf(": unhandled!\n"); + return 0; +} + +static int process_event_type_stub(struct perf_event_ops *ops __used, + union perf_event *event __used) { dump_printf(": unhandled!\n"); return 0; @@ -338,7 +338,7 @@ static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) if (handler->lost == NULL) handler->lost = perf_event__process_lost; if (handler->read == NULL) - handler->read = process_event_stub; + handler->read = process_event_sample_stub; if (handler->throttle == NULL) handler->throttle = process_event_stub; if (handler->unthrottle == NULL) @@ -346,11 +346,11 @@ static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) if (handler->attr == NULL) handler->attr = process_event_synth_attr_stub; if (handler->event_type == NULL) - handler->event_type = process_event_synth_stub; + handler->event_type = process_event_type_stub; if (handler->tracing_data == NULL) handler->tracing_data = process_event_synth_tracing_data_stub; if (handler->build_id == NULL) - handler->build_id = process_event_synth_stub; + handler->build_id = process_finished_round_stub; if (handler->finished_round == NULL) { if (handler->ordered_samples) handler->finished_round = process_finished_round; @@ -734,6 +734,18 @@ static void dump_sample(struct perf_session *session, union perf_event *event, callchain__printf(sample); } +static struct machine * + perf_session__find_machine_for_cpumode(struct perf_session *session, + union perf_event *event) +{ + const u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + + if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL && perf_guest) + return perf_session__find_machine(session, event->ip.pid); + + return perf_session__find_host_machine(session); +} + static int perf_session_deliver_event(struct perf_session *session, union perf_event *event, struct perf_sample *sample, @@ -741,6 +753,7 @@ static int perf_session_deliver_event(struct perf_session *session, u64 file_offset) { struct perf_evsel *evsel; + struct machine *machine; dump_event(session, event, file_offset, sample); @@ -762,6 +775,8 @@ static int perf_session_deliver_event(struct perf_session *session, hists__inc_nr_events(&evsel->hists, event->header.type); } + machine = perf_session__find_machine_for_cpumode(session, event); + switch (event->header.type) { case PERF_RECORD_SAMPLE: dump_sample(session, event, sample); @@ -769,23 +784,25 @@ static int perf_session_deliver_event(struct perf_session *session, ++session->hists.stats.nr_unknown_id; return -1; } - return ops->sample(ops, event, sample, evsel, session); + return ops->sample(ops, event, sample, evsel, machine); case PERF_RECORD_MMAP: - return ops->mmap(ops, event, sample, session); + return ops->mmap(ops, event, sample, machine); case PERF_RECORD_COMM: - return ops->comm(ops, event, sample, session); + return ops->comm(ops, event, sample, machine); case PERF_RECORD_FORK: - return ops->fork(ops, event, sample, session); + return ops->fork(ops, event, sample, machine); case PERF_RECORD_EXIT: - return ops->exit(ops, event, sample, session); + return ops->exit(ops, event, sample, machine); case PERF_RECORD_LOST: - return ops->lost(ops, event, sample, session); + if (ops->lost == perf_event__process_lost) + session->hists.stats.total_lost += event->lost.lost; + return ops->lost(ops, event, sample, machine); case PERF_RECORD_READ: - return ops->read(ops, event, sample, session); + return ops->read(ops, event, sample, evsel, machine); case PERF_RECORD_THROTTLE: - return ops->throttle(ops, event, sample, session); + return ops->throttle(ops, event, sample, machine); case PERF_RECORD_UNTHROTTLE: - return ops->unthrottle(ops, event, sample, session); + return ops->unthrottle(ops, event, sample, machine); default: ++session->hists.stats.nr_unknown_events; return -1; @@ -823,7 +840,7 @@ static int perf_session__process_user_event(struct perf_session *session, union perf_session__update_sample_type(session); return err; case PERF_RECORD_HEADER_EVENT_TYPE: - return ops->event_type(ops, event, session); + return ops->event_type(ops, event); case PERF_RECORD_HEADER_TRACING_DATA: /* setup for reading amidst mmap */ lseek(session->fd, file_offset, SEEK_SET); @@ -1170,9 +1187,8 @@ bool perf_session__has_traces(struct perf_session *self, const char *msg) return true; } -int perf_session__set_kallsyms_ref_reloc_sym(struct map **maps, - const char *symbol_name, - u64 addr) +int maps__set_kallsyms_ref_reloc_sym(struct map **maps, + const char *symbol_name, u64 addr) { char *bracket; enum map_type i; @@ -1264,17 +1280,16 @@ struct perf_evsel *perf_session__find_first_evtype(struct perf_session *session, return NULL; } -void perf_session__print_ip(union perf_event *event, struct perf_evsel *evsel, - struct perf_sample *sample, - struct perf_session *session, - int print_sym, int print_dso) +void perf_event__print_ip(union perf_event *event, struct perf_sample *sample, + struct machine *machine, struct perf_evsel *evsel, + int print_sym, int print_dso) { struct addr_location al; const char *symname, *dsoname; struct callchain_cursor *cursor = &evsel->hists.callchain_cursor; struct callchain_cursor_node *node; - if (perf_event__preprocess_sample(event, session, &al, sample, + if (perf_event__preprocess_sample(event, machine, &al, sample, NULL) < 0) { error("problem processing %d event, skipping it.\n", event->header.type); @@ -1283,7 +1298,7 @@ void perf_session__print_ip(union perf_event *event, struct perf_evsel *evsel, if (symbol_conf.use_callchain && sample->callchain) { - if (perf_session__resolve_callchain(session, evsel, al.thread, + if (machine__resolve_callchain(machine, evsel, al.thread, sample->callchain, NULL) != 0) { if (verbose) error("Failed to resolve callchain. Skipping\n"); diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index 6de3d1368900..1c5823c7d6dc 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -58,32 +58,34 @@ struct perf_event_ops; typedef int (*event_sample)(struct perf_event_ops *ops, union perf_event *event, struct perf_sample *sample, - struct perf_evsel *evsel, struct perf_session *session); + struct perf_evsel *evsel, struct machine *machine); typedef int (*event_op)(struct perf_event_ops *ops, union perf_event *event, struct perf_sample *sample, - struct perf_session *session); + struct machine *machine); typedef int (*event_synth_op)(union perf_event *self, struct perf_session *session); typedef int (*event_attr_op)(union perf_event *event, struct perf_evlist **pevlist); +typedef int (*event_simple_op)(struct perf_event_ops *ops, + union perf_event *event); typedef int (*event_op2)(struct perf_event_ops *ops, union perf_event *event, struct perf_session *session); struct perf_event_ops { - event_sample sample; + event_sample sample, + read; event_op mmap, comm, fork, exit, lost, - read, throttle, unthrottle; event_attr_op attr; event_synth_op tracing_data; - event_op2 event_type, - build_id, - finished_round; + event_simple_op event_type; + event_op2 finished_round, + build_id; bool ordered_samples; bool ordering_requires_timestamps; }; @@ -108,10 +110,6 @@ int perf_session__resolve_callchain(struct perf_session *self, struct perf_evsel bool perf_session__has_traces(struct perf_session *self, const char *msg); -int perf_session__set_kallsyms_ref_reloc_sym(struct map **maps, - const char *symbol_name, - u64 addr); - void mem_bswap_64(void *src, int byte_size); void perf_event__attr_swap(struct perf_event_attr *attr); @@ -151,6 +149,9 @@ void perf_session__process_machines(struct perf_session *self, return machines__process(&self->machines, process, ops); } +struct thread *perf_session__findnew(struct perf_session *self, pid_t pid); +size_t perf_session__fprintf(struct perf_session *self, FILE *fp); + size_t perf_session__fprintf_dsos(struct perf_session *self, FILE *fp); size_t perf_session__fprintf_dsos_buildid(struct perf_session *self, @@ -171,10 +172,9 @@ static inline int perf_session__parse_sample(struct perf_session *session, struct perf_evsel *perf_session__find_first_evtype(struct perf_session *session, unsigned int type); -void perf_session__print_ip(union perf_event *event, struct perf_evsel *evsel, - struct perf_sample *sample, - struct perf_session *session, - int print_sym, int print_dso); +void perf_event__print_ip(union perf_event *event, struct perf_sample *sample, + struct machine *machine, struct perf_evsel *evsel, + int print_sym, int print_dso); int perf_session__cpu_bitmap(struct perf_session *session, const char *cpu_list, unsigned long *cpu_bitmap); diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h index e5f2401c1b5e..70c2c13ff679 100644 --- a/tools/perf/util/thread.h +++ b/tools/perf/util/thread.h @@ -18,16 +18,14 @@ struct thread { int comm_len; }; -struct perf_session; +struct machine; void thread__delete(struct thread *self); int thread__set_comm(struct thread *self, const char *comm); int thread__comm_len(struct thread *self); -struct thread *perf_session__findnew(struct perf_session *self, pid_t pid); void thread__insert_map(struct thread *self, struct map *map); int thread__fork(struct thread *self, struct thread *parent); -size_t perf_session__fprintf(struct perf_session *self, FILE *fp); static inline struct map *thread__find_map(struct thread *self, enum map_type type, u64 addr) @@ -35,14 +33,12 @@ static inline struct map *thread__find_map(struct thread *self, return self ? map_groups__find(&self->mg, type, addr) : NULL; } -void thread__find_addr_map(struct thread *self, - struct perf_session *session, u8 cpumode, - enum map_type type, pid_t pid, u64 addr, +void thread__find_addr_map(struct thread *thread, struct machine *machine, + u8 cpumode, enum map_type type, u64 addr, struct addr_location *al); -void thread__find_addr_location(struct thread *self, - struct perf_session *session, u8 cpumode, - enum map_type type, pid_t pid, u64 addr, +void thread__find_addr_location(struct thread *thread, struct machine *machine, + u8 cpumode, enum map_type type, u64 addr, struct addr_location *al, symbol_filter_t filter); #endif /* __PERF_THREAD_H */ diff --git a/tools/perf/util/trace-event-scripting.c b/tools/perf/util/trace-event-scripting.c index c9dcbec7d800..a3fdf55f317b 100644 --- a/tools/perf/util/trace-event-scripting.c +++ b/tools/perf/util/trace-event-scripting.c @@ -39,7 +39,7 @@ static int stop_script_unsupported(void) static void process_event_unsupported(union perf_event *event __unused, struct perf_sample *sample __unused, struct perf_evsel *evsel __unused, - struct perf_session *session __unused, + struct machine *machine __unused, struct thread *thread __unused) { } diff --git a/tools/perf/util/trace-event.h b/tools/perf/util/trace-event.h index a84100817649..58ae14c5baac 100644 --- a/tools/perf/util/trace-event.h +++ b/tools/perf/util/trace-event.h @@ -3,7 +3,11 @@ #include #include "parse-events.h" -#include "session.h" + +struct machine; +struct perf_sample; +union perf_event; +struct thread; #define __unused __attribute__((unused)) @@ -292,7 +296,7 @@ struct scripting_ops { void (*process_event) (union perf_event *event, struct perf_sample *sample, struct perf_evsel *evsel, - struct perf_session *session, + struct machine *machine, struct thread *thread); int (*generate_script) (const char *outfile); }; -- cgit v1.2.2 From 45694aa7702bc44d538a3bcb51bb2bb96cf190c0 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 28 Nov 2011 08:30:20 -0200 Subject: perf tools: Rename perf_event_ops to perf_tool To better reflect that it became the base class for all tools, that must be in each tool struct and where common stuff will be put. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-qgpc4msetqlwr8y2k7537cxe@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/build-id.c | 7 +- tools/perf/util/build-id.h | 2 +- tools/perf/util/event.c | 54 +++++++-------- tools/perf/util/event.h | 22 +++--- tools/perf/util/header.c | 28 ++++---- tools/perf/util/header.h | 16 ++--- tools/perf/util/session.c | 163 +++++++++++++++++++++++---------------------- tools/perf/util/session.h | 49 ++------------ tools/perf/util/tool.h | 45 +++++++++++++ tools/perf/util/top.h | 7 +- 10 files changed, 203 insertions(+), 190 deletions(-) create mode 100644 tools/perf/util/tool.h (limited to 'tools/perf/util') diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c index 2f84c4802aca..dff9c7a725f4 100644 --- a/tools/perf/util/build-id.c +++ b/tools/perf/util/build-id.c @@ -14,8 +14,9 @@ #include #include "debug.h" #include "session.h" +#include "tool.h" -static int build_id__mark_dso_hit(struct perf_event_ops *ops __used, +static int build_id__mark_dso_hit(struct perf_tool *tool __used, union perf_event *event, struct perf_sample *sample __used, struct perf_evsel *evsel __used, @@ -40,7 +41,7 @@ static int build_id__mark_dso_hit(struct perf_event_ops *ops __used, return 0; } -static int perf_event__exit_del_thread(struct perf_event_ops *ops __used, +static int perf_event__exit_del_thread(struct perf_tool *tool __used, union perf_event *event, struct perf_sample *sample __used, struct machine *machine) @@ -59,7 +60,7 @@ static int perf_event__exit_del_thread(struct perf_event_ops *ops __used, return 0; } -struct perf_event_ops build_id__mark_dso_hit_ops = { +struct perf_tool build_id__mark_dso_hit_ops = { .sample = build_id__mark_dso_hit, .mmap = perf_event__process_mmap, .fork = perf_event__process_task, diff --git a/tools/perf/util/build-id.h b/tools/perf/util/build-id.h index 5dafb00eaa06..a993ba87d996 100644 --- a/tools/perf/util/build-id.h +++ b/tools/perf/util/build-id.h @@ -3,7 +3,7 @@ #include "session.h" -extern struct perf_event_ops build_id__mark_dso_hit_ops; +extern struct perf_tool build_id__mark_dso_hit_ops; char *dso__build_id_filename(struct dso *self, char *bf, size_t size); diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 0cdc811c48e2..0ebbe7641335 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -43,7 +43,7 @@ static struct perf_sample synth_sample = { .period = 1, }; -static pid_t perf_event__synthesize_comm(struct perf_event_ops *ops, +static pid_t perf_event__synthesize_comm(struct perf_tool *tool, union perf_event *event, pid_t pid, int full, perf_event__handler_t process, struct machine *machine) @@ -99,7 +99,7 @@ out_race: if (!full) { event->comm.tid = pid; - process(ops, event, &synth_sample, machine); + process(tool, event, &synth_sample, machine); goto out; } @@ -117,7 +117,7 @@ out_race: event->comm.tid = pid; - process(ops, event, &synth_sample, machine); + process(tool, event, &synth_sample, machine); } closedir(tasks); @@ -127,7 +127,7 @@ out: return tgid; } -static int perf_event__synthesize_mmap_events(struct perf_event_ops *ops, +static int perf_event__synthesize_mmap_events(struct perf_tool *tool, union perf_event *event, pid_t pid, pid_t tgid, perf_event__handler_t process, @@ -199,7 +199,7 @@ static int perf_event__synthesize_mmap_events(struct perf_event_ops *ops, event->mmap.pid = tgid; event->mmap.tid = pid; - process(ops, event, &synth_sample, machine); + process(tool, event, &synth_sample, machine); } } @@ -207,7 +207,7 @@ static int perf_event__synthesize_mmap_events(struct perf_event_ops *ops, return 0; } -int perf_event__synthesize_modules(struct perf_event_ops *ops, +int perf_event__synthesize_modules(struct perf_tool *tool, perf_event__handler_t process, struct machine *machine) { @@ -252,7 +252,7 @@ int perf_event__synthesize_modules(struct perf_event_ops *ops, memcpy(event->mmap.filename, pos->dso->long_name, pos->dso->long_name_len + 1); - process(ops, event, &synth_sample, machine); + process(tool, event, &synth_sample, machine); } free(event); @@ -262,18 +262,18 @@ int perf_event__synthesize_modules(struct perf_event_ops *ops, static int __event__synthesize_thread(union perf_event *comm_event, union perf_event *mmap_event, pid_t pid, perf_event__handler_t process, - struct perf_event_ops *ops, + struct perf_tool *tool, struct machine *machine) { - pid_t tgid = perf_event__synthesize_comm(ops, comm_event, pid, 1, + pid_t tgid = perf_event__synthesize_comm(tool, comm_event, pid, 1, process, machine); if (tgid == -1) return -1; - return perf_event__synthesize_mmap_events(ops, mmap_event, pid, tgid, + return perf_event__synthesize_mmap_events(tool, mmap_event, pid, tgid, process, machine); } -int perf_event__synthesize_thread_map(struct perf_event_ops *ops, +int perf_event__synthesize_thread_map(struct perf_tool *tool, struct thread_map *threads, perf_event__handler_t process, struct machine *machine) @@ -293,7 +293,7 @@ int perf_event__synthesize_thread_map(struct perf_event_ops *ops, for (thread = 0; thread < threads->nr; ++thread) { if (__event__synthesize_thread(comm_event, mmap_event, threads->map[thread], - process, ops, machine)) { + process, tool, machine)) { err = -1; break; } @@ -305,7 +305,7 @@ out: return err; } -int perf_event__synthesize_threads(struct perf_event_ops *ops, +int perf_event__synthesize_threads(struct perf_tool *tool, perf_event__handler_t process, struct machine *machine) { @@ -334,7 +334,7 @@ int perf_event__synthesize_threads(struct perf_event_ops *ops, continue; __event__synthesize_thread(comm_event, mmap_event, pid, - process, ops, machine); + process, tool, machine); } closedir(proc); @@ -369,7 +369,7 @@ static int find_symbol_cb(void *arg, const char *name, char type, return 1; } -int perf_event__synthesize_kernel_mmap(struct perf_event_ops *ops, +int perf_event__synthesize_kernel_mmap(struct perf_tool *tool, perf_event__handler_t process, struct machine *machine, const char *symbol_name) @@ -427,13 +427,13 @@ int perf_event__synthesize_kernel_mmap(struct perf_event_ops *ops, event->mmap.len = map->end - event->mmap.start; event->mmap.pid = machine->pid; - err = process(ops, event, &synth_sample, machine); + err = process(tool, event, &synth_sample, machine); free(event); return err; } -int perf_event__process_comm(struct perf_event_ops *ops __used, +int perf_event__process_comm(struct perf_tool *tool __used, union perf_event *event, struct perf_sample *sample __used, struct machine *machine) @@ -450,7 +450,7 @@ int perf_event__process_comm(struct perf_event_ops *ops __used, return 0; } -int perf_event__process_lost(struct perf_event_ops *ops __used, +int perf_event__process_lost(struct perf_tool *tool __used, union perf_event *event, struct perf_sample *sample __used, struct machine *machine __used) @@ -473,7 +473,7 @@ static void perf_event__set_kernel_mmap_len(union perf_event *event, maps[MAP__FUNCTION]->end = ~0ULL; } -static int perf_event__process_kernel_mmap(struct perf_event_ops *ops __used, +static int perf_event__process_kernel_mmap(struct perf_tool *tool __used, union perf_event *event, struct machine *machine) { @@ -566,7 +566,7 @@ out_problem: return -1; } -int perf_event__process_mmap(struct perf_event_ops *ops, +int perf_event__process_mmap(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample __used, struct machine *machine) @@ -582,7 +582,7 @@ int perf_event__process_mmap(struct perf_event_ops *ops, if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL || cpumode == PERF_RECORD_MISC_KERNEL) { - ret = perf_event__process_kernel_mmap(ops, event, machine); + ret = perf_event__process_kernel_mmap(tool, event, machine); if (ret < 0) goto out_problem; return 0; @@ -606,7 +606,7 @@ out_problem: return 0; } -int perf_event__process_task(struct perf_event_ops *ops __used, +int perf_event__process_task(struct perf_tool *tool __used, union perf_event *event, struct perf_sample *sample __used, struct machine *machine) @@ -631,22 +631,22 @@ int perf_event__process_task(struct perf_event_ops *ops __used, return 0; } -int perf_event__process(struct perf_event_ops *ops, union perf_event *event, +int perf_event__process(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample, struct machine *machine) { switch (event->header.type) { case PERF_RECORD_COMM: - perf_event__process_comm(ops, event, sample, machine); + perf_event__process_comm(tool, event, sample, machine); break; case PERF_RECORD_MMAP: - perf_event__process_mmap(ops, event, sample, machine); + perf_event__process_mmap(tool, event, sample, machine); break; case PERF_RECORD_FORK: case PERF_RECORD_EXIT: - perf_event__process_task(ops, event, sample, machine); + perf_event__process_task(tool, event, sample, machine); break; case PERF_RECORD_LOST: - perf_event__process_lost(ops, event, sample, machine); + perf_event__process_lost(tool, event, sample, machine); default: break; } diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index 1564877e8703..d8499e7cf641 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -141,47 +141,47 @@ union perf_event { void perf_event__print_totals(void); -struct perf_event_ops; +struct perf_tool; struct thread_map; -typedef int (*perf_event__handler_t)(struct perf_event_ops *ops, +typedef int (*perf_event__handler_t)(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample, struct machine *machine); -int perf_event__synthesize_thread_map(struct perf_event_ops *ops, +int perf_event__synthesize_thread_map(struct perf_tool *tool, struct thread_map *threads, perf_event__handler_t process, struct machine *machine); -int perf_event__synthesize_threads(struct perf_event_ops *ops, +int perf_event__synthesize_threads(struct perf_tool *tool, perf_event__handler_t process, struct machine *machine); -int perf_event__synthesize_kernel_mmap(struct perf_event_ops *ops, +int perf_event__synthesize_kernel_mmap(struct perf_tool *tool, perf_event__handler_t process, struct machine *machine, const char *symbol_name); -int perf_event__synthesize_modules(struct perf_event_ops *ops, +int perf_event__synthesize_modules(struct perf_tool *tool, perf_event__handler_t process, struct machine *machine); -int perf_event__process_comm(struct perf_event_ops *ops, +int perf_event__process_comm(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample, struct machine *machine); -int perf_event__process_lost(struct perf_event_ops *ops, +int perf_event__process_lost(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample, struct machine *machine); -int perf_event__process_mmap(struct perf_event_ops *ops, +int perf_event__process_mmap(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample, struct machine *machine); -int perf_event__process_task(struct perf_event_ops *ops, +int perf_event__process_task(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample, struct machine *machine); -int perf_event__process(struct perf_event_ops *ops, +int perf_event__process(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample, struct machine *machine); diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index db280d6ca898..9272f3a20cac 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -2070,7 +2070,7 @@ out_delete_evlist: return -ENOMEM; } -int perf_event__synthesize_attr(struct perf_event_ops *ops, +int perf_event__synthesize_attr(struct perf_tool *tool, struct perf_event_attr *attr, u16 ids, u64 *id, perf_event__handler_t process) { @@ -2094,14 +2094,14 @@ int perf_event__synthesize_attr(struct perf_event_ops *ops, ev->attr.header.type = PERF_RECORD_HEADER_ATTR; ev->attr.header.size = size; - err = process(ops, ev, NULL, NULL); + err = process(tool, ev, NULL, NULL); free(ev); return err; } -int perf_event__synthesize_attrs(struct perf_event_ops *ops, +int perf_event__synthesize_attrs(struct perf_tool *tool, struct perf_session *session, perf_event__handler_t process) { @@ -2109,7 +2109,7 @@ int perf_event__synthesize_attrs(struct perf_event_ops *ops, int err = 0; list_for_each_entry(attr, &session->evlist->entries, node) { - err = perf_event__synthesize_attr(ops, &attr->attr, attr->ids, + err = perf_event__synthesize_attr(tool, &attr->attr, attr->ids, attr->id, process); if (err) { pr_debug("failed to create perf header attribute\n"); @@ -2157,7 +2157,7 @@ int perf_event__process_attr(union perf_event *event, return 0; } -int perf_event__synthesize_event_type(struct perf_event_ops *ops, +int perf_event__synthesize_event_type(struct perf_tool *tool, u64 event_id, char *name, perf_event__handler_t process, struct machine *machine) @@ -2178,12 +2178,12 @@ int perf_event__synthesize_event_type(struct perf_event_ops *ops, ev.event_type.header.size = sizeof(ev.event_type) - (sizeof(ev.event_type.event_type.name) - size); - err = process(ops, &ev, NULL, machine); + err = process(tool, &ev, NULL, machine); return err; } -int perf_event__synthesize_event_types(struct perf_event_ops *ops, +int perf_event__synthesize_event_types(struct perf_tool *tool, perf_event__handler_t process, struct machine *machine) { @@ -2193,7 +2193,7 @@ int perf_event__synthesize_event_types(struct perf_event_ops *ops, for (i = 0; i < event_count; i++) { type = &events[i]; - err = perf_event__synthesize_event_type(ops, type->event_id, + err = perf_event__synthesize_event_type(tool, type->event_id, type->name, process, machine); if (err) { @@ -2205,7 +2205,7 @@ int perf_event__synthesize_event_types(struct perf_event_ops *ops, return err; } -int perf_event__process_event_type(struct perf_event_ops *ops __unused, +int perf_event__process_event_type(struct perf_tool *tool __unused, union perf_event *event) { if (perf_header__push_event(event->event_type.event_type.event_id, @@ -2215,7 +2215,7 @@ int perf_event__process_event_type(struct perf_event_ops *ops __unused, return 0; } -int perf_event__synthesize_tracing_data(struct perf_event_ops *ops, int fd, +int perf_event__synthesize_tracing_data(struct perf_tool *tool, int fd, struct perf_evlist *evlist, perf_event__handler_t process) { @@ -2248,7 +2248,7 @@ int perf_event__synthesize_tracing_data(struct perf_event_ops *ops, int fd, ev.tracing_data.header.size = sizeof(ev.tracing_data); ev.tracing_data.size = aligned_size; - process(ops, &ev, NULL, NULL); + process(tool, &ev, NULL, NULL); /* * The put function will copy all the tracing data @@ -2290,7 +2290,7 @@ int perf_event__process_tracing_data(union perf_event *event, return size_read + padding; } -int perf_event__synthesize_build_id(struct perf_event_ops *ops, +int perf_event__synthesize_build_id(struct perf_tool *tool, struct dso *pos, u16 misc, perf_event__handler_t process, struct machine *machine) @@ -2313,12 +2313,12 @@ int perf_event__synthesize_build_id(struct perf_event_ops *ops, ev.build_id.header.size = sizeof(ev.build_id) + len; memcpy(&ev.build_id.filename, pos->long_name, pos->long_name_len); - err = process(ops, &ev, NULL, machine); + err = process(tool, &ev, NULL, machine); return err; } -int perf_event__process_build_id(struct perf_event_ops *ops __used, +int perf_event__process_build_id(struct perf_tool *tool __used, union perf_event *event, struct perf_session *session) { diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index a604962fc431..09365b32098e 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -97,35 +97,35 @@ int build_id_cache__add_s(const char *sbuild_id, const char *debugdir, const char *name, bool is_kallsyms); int build_id_cache__remove_s(const char *sbuild_id, const char *debugdir); -int perf_event__synthesize_attr(struct perf_event_ops *ops, +int perf_event__synthesize_attr(struct perf_tool *tool, struct perf_event_attr *attr, u16 ids, u64 *id, perf_event__handler_t process); -int perf_event__synthesize_attrs(struct perf_event_ops *ops, +int perf_event__synthesize_attrs(struct perf_tool *tool, struct perf_session *session, perf_event__handler_t process); int perf_event__process_attr(union perf_event *event, struct perf_evlist **pevlist); -int perf_event__synthesize_event_type(struct perf_event_ops *ops, +int perf_event__synthesize_event_type(struct perf_tool *tool, u64 event_id, char *name, perf_event__handler_t process, struct machine *machine); -int perf_event__synthesize_event_types(struct perf_event_ops *ops, +int perf_event__synthesize_event_types(struct perf_tool *tool, perf_event__handler_t process, struct machine *machine); -int perf_event__process_event_type(struct perf_event_ops *ops, +int perf_event__process_event_type(struct perf_tool *tool, union perf_event *event); -int perf_event__synthesize_tracing_data(struct perf_event_ops *ops, +int perf_event__synthesize_tracing_data(struct perf_tool *tool, int fd, struct perf_evlist *evlist, perf_event__handler_t process); int perf_event__process_tracing_data(union perf_event *event, struct perf_session *session); -int perf_event__synthesize_build_id(struct perf_event_ops *ops, +int perf_event__synthesize_build_id(struct perf_tool *tool, struct dso *pos, u16 misc, perf_event__handler_t process, struct machine *machine); -int perf_event__process_build_id(struct perf_event_ops *ops, +int perf_event__process_build_id(struct perf_tool *tool, union perf_event *event, struct perf_session *session); diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index be33606386bf..7d159088c4ac 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -10,6 +10,7 @@ #include "evlist.h" #include "evsel.h" #include "session.h" +#include "tool.h" #include "sort.h" #include "util.h" #include "cpumap.h" @@ -104,7 +105,7 @@ static void perf_session__destroy_kernel_maps(struct perf_session *self) struct perf_session *perf_session__new(const char *filename, int mode, bool force, bool repipe, - struct perf_event_ops *ops) + struct perf_tool *tool) { size_t len = filename ? strlen(filename) + 1 : 0; struct perf_session *self = zalloc(sizeof(*self) + len); @@ -142,10 +143,10 @@ struct perf_session *perf_session__new(const char *filename, int mode, goto out_delete; } - if (ops && ops->ordering_requires_timestamps && - ops->ordered_samples && !self->sample_id_all) { + if (tool && tool->ordering_requires_timestamps && + tool->ordered_samples && !self->sample_id_all) { dump_printf("WARNING: No sample_id_all support, falling back to unordered processing\n"); - ops->ordered_samples = false; + tool->ordered_samples = false; } out: @@ -285,7 +286,7 @@ static int process_event_synth_attr_stub(union perf_event *event __used, return 0; } -static int process_event_sample_stub(struct perf_event_ops *ops __used, +static int process_event_sample_stub(struct perf_tool *tool __used, union perf_event *event __used, struct perf_sample *sample __used, struct perf_evsel *evsel __used, @@ -295,7 +296,7 @@ static int process_event_sample_stub(struct perf_event_ops *ops __used, return 0; } -static int process_event_stub(struct perf_event_ops *ops __used, +static int process_event_stub(struct perf_tool *tool __used, union perf_event *event __used, struct perf_sample *sample __used, struct machine *machine __used) @@ -304,7 +305,7 @@ static int process_event_stub(struct perf_event_ops *ops __used, return 0; } -static int process_finished_round_stub(struct perf_event_ops *ops __used, +static int process_finished_round_stub(struct perf_tool *tool __used, union perf_event *event __used, struct perf_session *perf_session __used) { @@ -312,50 +313,50 @@ static int process_finished_round_stub(struct perf_event_ops *ops __used, return 0; } -static int process_event_type_stub(struct perf_event_ops *ops __used, +static int process_event_type_stub(struct perf_tool *tool __used, union perf_event *event __used) { dump_printf(": unhandled!\n"); return 0; } -static int process_finished_round(struct perf_event_ops *ops, +static int process_finished_round(struct perf_tool *tool, union perf_event *event, struct perf_session *session); -static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) +static void perf_tool__fill_defaults(struct perf_tool *tool) { - if (handler->sample == NULL) - handler->sample = process_event_sample_stub; - if (handler->mmap == NULL) - handler->mmap = process_event_stub; - if (handler->comm == NULL) - handler->comm = process_event_stub; - if (handler->fork == NULL) - handler->fork = process_event_stub; - if (handler->exit == NULL) - handler->exit = process_event_stub; - if (handler->lost == NULL) - handler->lost = perf_event__process_lost; - if (handler->read == NULL) - handler->read = process_event_sample_stub; - if (handler->throttle == NULL) - handler->throttle = process_event_stub; - if (handler->unthrottle == NULL) - handler->unthrottle = process_event_stub; - if (handler->attr == NULL) - handler->attr = process_event_synth_attr_stub; - if (handler->event_type == NULL) - handler->event_type = process_event_type_stub; - if (handler->tracing_data == NULL) - handler->tracing_data = process_event_synth_tracing_data_stub; - if (handler->build_id == NULL) - handler->build_id = process_finished_round_stub; - if (handler->finished_round == NULL) { - if (handler->ordered_samples) - handler->finished_round = process_finished_round; + if (tool->sample == NULL) + tool->sample = process_event_sample_stub; + if (tool->mmap == NULL) + tool->mmap = process_event_stub; + if (tool->comm == NULL) + tool->comm = process_event_stub; + if (tool->fork == NULL) + tool->fork = process_event_stub; + if (tool->exit == NULL) + tool->exit = process_event_stub; + if (tool->lost == NULL) + tool->lost = perf_event__process_lost; + if (tool->read == NULL) + tool->read = process_event_sample_stub; + if (tool->throttle == NULL) + tool->throttle = process_event_stub; + if (tool->unthrottle == NULL) + tool->unthrottle = process_event_stub; + if (tool->attr == NULL) + tool->attr = process_event_synth_attr_stub; + if (tool->event_type == NULL) + tool->event_type = process_event_type_stub; + if (tool->tracing_data == NULL) + tool->tracing_data = process_event_synth_tracing_data_stub; + if (tool->build_id == NULL) + tool->build_id = process_finished_round_stub; + if (tool->finished_round == NULL) { + if (tool->ordered_samples) + tool->finished_round = process_finished_round; else - handler->finished_round = process_finished_round_stub; + tool->finished_round = process_finished_round_stub; } } @@ -487,11 +488,11 @@ static void perf_session_free_sample_buffers(struct perf_session *session) static int perf_session_deliver_event(struct perf_session *session, union perf_event *event, struct perf_sample *sample, - struct perf_event_ops *ops, + struct perf_tool *tool, u64 file_offset); static void flush_sample_queue(struct perf_session *s, - struct perf_event_ops *ops) + struct perf_tool *tool) { struct ordered_samples *os = &s->ordered_samples; struct list_head *head = &os->samples; @@ -502,7 +503,7 @@ static void flush_sample_queue(struct perf_session *s, unsigned idx = 0, progress_next = os->nr_samples / 16; int ret; - if (!ops->ordered_samples || !limit) + if (!tool->ordered_samples || !limit) return; list_for_each_entry_safe(iter, tmp, head, list) { @@ -513,7 +514,7 @@ static void flush_sample_queue(struct perf_session *s, if (ret) pr_err("Can't parse sample, err = %d\n", ret); else - perf_session_deliver_event(s, iter->event, &sample, ops, + perf_session_deliver_event(s, iter->event, &sample, tool, iter->file_offset); os->last_flush = iter->timestamp; @@ -575,11 +576,11 @@ static void flush_sample_queue(struct perf_session *s, * Flush every events below timestamp 7 * etc... */ -static int process_finished_round(struct perf_event_ops *ops, +static int process_finished_round(struct perf_tool *tool, union perf_event *event __used, struct perf_session *session) { - flush_sample_queue(session, ops); + flush_sample_queue(session, tool); session->ordered_samples.next_flush = session->ordered_samples.max_timestamp; return 0; @@ -749,7 +750,7 @@ static struct machine * static int perf_session_deliver_event(struct perf_session *session, union perf_event *event, struct perf_sample *sample, - struct perf_event_ops *ops, + struct perf_tool *tool, u64 file_offset) { struct perf_evsel *evsel; @@ -784,25 +785,25 @@ static int perf_session_deliver_event(struct perf_session *session, ++session->hists.stats.nr_unknown_id; return -1; } - return ops->sample(ops, event, sample, evsel, machine); + return tool->sample(tool, event, sample, evsel, machine); case PERF_RECORD_MMAP: - return ops->mmap(ops, event, sample, machine); + return tool->mmap(tool, event, sample, machine); case PERF_RECORD_COMM: - return ops->comm(ops, event, sample, machine); + return tool->comm(tool, event, sample, machine); case PERF_RECORD_FORK: - return ops->fork(ops, event, sample, machine); + return tool->fork(tool, event, sample, machine); case PERF_RECORD_EXIT: - return ops->exit(ops, event, sample, machine); + return tool->exit(tool, event, sample, machine); case PERF_RECORD_LOST: - if (ops->lost == perf_event__process_lost) + if (tool->lost == perf_event__process_lost) session->hists.stats.total_lost += event->lost.lost; - return ops->lost(ops, event, sample, machine); + return tool->lost(tool, event, sample, machine); case PERF_RECORD_READ: - return ops->read(ops, event, sample, evsel, machine); + return tool->read(tool, event, sample, evsel, machine); case PERF_RECORD_THROTTLE: - return ops->throttle(ops, event, sample, machine); + return tool->throttle(tool, event, sample, machine); case PERF_RECORD_UNTHROTTLE: - return ops->unthrottle(ops, event, sample, machine); + return tool->unthrottle(tool, event, sample, machine); default: ++session->hists.stats.nr_unknown_events; return -1; @@ -826,7 +827,7 @@ static int perf_session__preprocess_sample(struct perf_session *session, } static int perf_session__process_user_event(struct perf_session *session, union perf_event *event, - struct perf_event_ops *ops, u64 file_offset) + struct perf_tool *tool, u64 file_offset) { int err; @@ -835,20 +836,20 @@ static int perf_session__process_user_event(struct perf_session *session, union /* These events are processed right away */ switch (event->header.type) { case PERF_RECORD_HEADER_ATTR: - err = ops->attr(event, &session->evlist); + err = tool->attr(event, &session->evlist); if (err == 0) perf_session__update_sample_type(session); return err; case PERF_RECORD_HEADER_EVENT_TYPE: - return ops->event_type(ops, event); + return tool->event_type(tool, event); case PERF_RECORD_HEADER_TRACING_DATA: /* setup for reading amidst mmap */ lseek(session->fd, file_offset, SEEK_SET); - return ops->tracing_data(event, session); + return tool->tracing_data(event, session); case PERF_RECORD_HEADER_BUILD_ID: - return ops->build_id(ops, event, session); + return tool->build_id(tool, event, session); case PERF_RECORD_FINISHED_ROUND: - return ops->finished_round(ops, event, session); + return tool->finished_round(tool, event, session); default: return -EINVAL; } @@ -856,7 +857,7 @@ static int perf_session__process_user_event(struct perf_session *session, union static int perf_session__process_event(struct perf_session *session, union perf_event *event, - struct perf_event_ops *ops, + struct perf_tool *tool, u64 file_offset) { struct perf_sample sample; @@ -872,7 +873,7 @@ static int perf_session__process_event(struct perf_session *session, hists__inc_nr_events(&session->hists, event->header.type); if (event->header.type >= PERF_RECORD_USER_TYPE_START) - return perf_session__process_user_event(session, event, ops, file_offset); + return perf_session__process_user_event(session, event, tool, file_offset); /* * For all kernel events we get the sample data @@ -885,14 +886,14 @@ static int perf_session__process_event(struct perf_session *session, if (perf_session__preprocess_sample(session, event, &sample)) return 0; - if (ops->ordered_samples) { + if (tool->ordered_samples) { ret = perf_session_queue_event(session, event, &sample, file_offset); if (ret != -ETIME) return ret; } - return perf_session_deliver_event(session, event, &sample, ops, + return perf_session_deliver_event(session, event, &sample, tool, file_offset); } @@ -921,9 +922,9 @@ static struct thread *perf_session__register_idle_thread(struct perf_session *se } static void perf_session__warn_about_errors(const struct perf_session *session, - const struct perf_event_ops *ops) + const struct perf_tool *tool) { - if (ops->lost == perf_event__process_lost && + if (tool->lost == perf_event__process_lost && session->hists.stats.nr_events[PERF_RECORD_LOST] != 0) { ui__warning("Processed %d events and lost %d chunks!\n\n" "Check IO/CPU overload!\n\n", @@ -958,7 +959,7 @@ static void perf_session__warn_about_errors(const struct perf_session *session, volatile int session_done; static int __perf_session__process_pipe_events(struct perf_session *self, - struct perf_event_ops *ops) + struct perf_tool *tool) { union perf_event event; uint32_t size; @@ -967,7 +968,7 @@ static int __perf_session__process_pipe_events(struct perf_session *self, int err; void *p; - perf_event_ops__fill_defaults(ops); + perf_tool__fill_defaults(tool); head = 0; more: @@ -1004,7 +1005,7 @@ more: } if (size == 0 || - (skip = perf_session__process_event(self, &event, ops, head)) < 0) { + (skip = perf_session__process_event(self, &event, tool, head)) < 0) { dump_printf("%#" PRIx64 " [%#x]: skipping unknown header type: %d\n", head, event.header.size, event.header.type); /* @@ -1027,7 +1028,7 @@ more: done: err = 0; out_err: - perf_session__warn_about_errors(self, ops); + perf_session__warn_about_errors(self, tool); perf_session_free_sample_buffers(self); return err; } @@ -1058,7 +1059,7 @@ fetch_mmaped_event(struct perf_session *session, int __perf_session__process_events(struct perf_session *session, u64 data_offset, u64 data_size, - u64 file_size, struct perf_event_ops *ops) + u64 file_size, struct perf_tool *tool) { u64 head, page_offset, file_offset, file_pos, progress_next; int err, mmap_prot, mmap_flags, map_idx = 0; @@ -1067,7 +1068,7 @@ int __perf_session__process_events(struct perf_session *session, union perf_event *event; uint32_t size; - perf_event_ops__fill_defaults(ops); + perf_tool__fill_defaults(tool); page_size = sysconf(_SC_PAGESIZE); @@ -1122,7 +1123,7 @@ more: size = event->header.size; if (size == 0 || - perf_session__process_event(session, event, ops, file_pos) < 0) { + perf_session__process_event(session, event, tool, file_pos) < 0) { dump_printf("%#" PRIx64 " [%#x]: skipping unknown header type: %d\n", file_offset + head, event->header.size, event->header.type); @@ -1151,15 +1152,15 @@ more: err = 0; /* do the final flush for ordered samples */ session->ordered_samples.next_flush = ULLONG_MAX; - flush_sample_queue(session, ops); + flush_sample_queue(session, tool); out_err: - perf_session__warn_about_errors(session, ops); + perf_session__warn_about_errors(session, tool); perf_session_free_sample_buffers(session); return err; } int perf_session__process_events(struct perf_session *self, - struct perf_event_ops *ops) + struct perf_tool *tool) { int err; @@ -1170,9 +1171,9 @@ int perf_session__process_events(struct perf_session *self, err = __perf_session__process_events(self, self->header.data_offset, self->header.data_size, - self->size, ops); + self->size, tool); else - err = __perf_session__process_pipe_events(self, ops); + err = __perf_session__process_pipe_events(self, tool); return err; } diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index 1c5823c7d6dc..30e9c6b6fc3c 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -53,55 +53,20 @@ struct perf_session { char filename[0]; }; -struct perf_evsel; -struct perf_event_ops; - -typedef int (*event_sample)(struct perf_event_ops *ops, - union perf_event *event, struct perf_sample *sample, - struct perf_evsel *evsel, struct machine *machine); -typedef int (*event_op)(struct perf_event_ops *ops, union perf_event *event, - struct perf_sample *sample, - struct machine *machine); -typedef int (*event_synth_op)(union perf_event *self, - struct perf_session *session); -typedef int (*event_attr_op)(union perf_event *event, - struct perf_evlist **pevlist); -typedef int (*event_simple_op)(struct perf_event_ops *ops, - union perf_event *event); -typedef int (*event_op2)(struct perf_event_ops *ops, union perf_event *event, - struct perf_session *session); - -struct perf_event_ops { - event_sample sample, - read; - event_op mmap, - comm, - fork, - exit, - lost, - throttle, - unthrottle; - event_attr_op attr; - event_synth_op tracing_data; - event_simple_op event_type; - event_op2 finished_round, - build_id; - bool ordered_samples; - bool ordering_requires_timestamps; -}; +struct perf_tool; struct perf_session *perf_session__new(const char *filename, int mode, bool force, bool repipe, - struct perf_event_ops *ops); + struct perf_tool *tool); void perf_session__delete(struct perf_session *self); void perf_event_header__bswap(struct perf_event_header *self); int __perf_session__process_events(struct perf_session *self, u64 data_offset, u64 data_size, u64 size, - struct perf_event_ops *ops); + struct perf_tool *tool); int perf_session__process_events(struct perf_session *self, - struct perf_event_ops *event_ops); + struct perf_tool *tool); int perf_session__resolve_callchain(struct perf_session *self, struct perf_evsel *evsel, struct thread *thread, @@ -142,11 +107,11 @@ struct machine *perf_session__findnew_machine(struct perf_session *self, pid_t p static inline void perf_session__process_machines(struct perf_session *self, - struct perf_event_ops *ops, + struct perf_tool *tool, machine__process_t process) { - process(&self->host_machine, ops); - return machines__process(&self->machines, process, ops); + process(&self->host_machine, tool); + return machines__process(&self->machines, process, tool); } struct thread *perf_session__findnew(struct perf_session *self, pid_t pid); diff --git a/tools/perf/util/tool.h b/tools/perf/util/tool.h new file mode 100644 index 000000000000..89ff1b551a74 --- /dev/null +++ b/tools/perf/util/tool.h @@ -0,0 +1,45 @@ +#ifndef __PERF_TOOL_H +#define __PERF_TOOL_H + +struct perf_session; +struct perf_evsel; +struct perf_tool; +struct machine; + +typedef int (*event_sample)(struct perf_tool *tool, union perf_event *event, + struct perf_sample *sample, + struct perf_evsel *evsel, struct machine *machine); + +typedef int (*event_op)(struct perf_tool *tool, union perf_event *event, + struct perf_sample *sample, struct machine *machine); + +typedef int (*event_attr_op)(union perf_event *event, + struct perf_evlist **pevlist); +typedef int (*event_simple_op)(struct perf_tool *tool, union perf_event *event); + +typedef int (*event_synth_op)(union perf_event *event, + struct perf_session *session); + +typedef int (*event_op2)(struct perf_tool *tool, union perf_event *event, + struct perf_session *session); + +struct perf_tool { + event_sample sample, + read; + event_op mmap, + comm, + fork, + exit, + lost, + throttle, + unthrottle; + event_attr_op attr; + event_synth_op tracing_data; + event_simple_op event_type; + event_op2 finished_round, + build_id; + bool ordered_samples; + bool ordering_requires_timestamps; +}; + +#endif /* __PERF_TOOL_H */ diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h index 44eda6fc6b33..40430ec5c267 100644 --- a/tools/perf/util/top.h +++ b/tools/perf/util/top.h @@ -1,16 +1,17 @@ #ifndef __PERF_TOP_H #define __PERF_TOP_H 1 +#include "tool.h" #include "types.h" -#include "session.h" -#include "../perf.h" #include +#include struct perf_evlist; struct perf_evsel; +struct perf_session; struct perf_top { - struct perf_event_ops ops; + struct perf_tool tool; struct perf_evlist *evlist; /* * Symbols will be added here in perf_event__process_sample and will -- cgit v1.2.2 From 1758af10cf13d156014035b29ff50eab3773d849 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 28 Nov 2011 09:37:05 -0200 Subject: perf top: Stop using globals for tool state Use its 'perf_tool' base class instead. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-i33q40wwvk2zna8fd36ex6sm@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/top.h | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h index 40430ec5c267..a248f3c2c60d 100644 --- a/tools/perf/util/top.h +++ b/tools/perf/util/top.h @@ -25,10 +25,26 @@ struct perf_top { int freq; pid_t target_pid, target_tid; bool hide_kernel_symbols, hide_user_symbols, zero; + bool system_wide; + bool use_tui, use_stdio; + bool sort_has_symbols; + bool dont_use_callchains; + bool kptr_restrict_warned; + bool vmlinux_warned; + bool inherit; + bool group; + bool sample_id_all_avail; + bool dump_symtab; const char *cpu_list; struct hist_entry *sym_filter_entry; struct perf_evsel *sym_evsel; struct perf_session *session; + struct winsize winsize; + unsigned int mmap_pages; + int default_interval; + int realtime_prio; + int sym_pcnt_filter; + const char *sym_filter; }; size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size); -- cgit v1.2.2 From ee29be625bd7b115d45eba4b0526ff3e24bf3ca0 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Mon, 28 Nov 2011 17:57:40 -0200 Subject: perf tools: Save some loops using perf_evlist__id2evsel Since we already ask for PERF_SAMPLE_ID and use it to quickly find the associated evsel, add handler func + data to struct perf_evsel to avoid using chains of if(strcmp(event_name)) and also to avoid all the linear list searches via trace_event_find. To demonstrate the technique convert 'perf sched' to it: # perf sched record sleep 5m And then: Performance counter stats for '/tmp/oldperf sched lat': 646.929438 task-clock # 0.999 CPUs utilized 9 context-switches # 0.000 M/sec 0 CPU-migrations # 0.000 M/sec 20,901 page-faults # 0.032 M/sec 1,290,144,450 cycles # 1.994 GHz stalled-cycles-frontend stalled-cycles-backend 1,606,158,439 instructions # 1.24 insns per cycle 339,088,395 branches # 524.151 M/sec 4,550,735 branch-misses # 1.34% of all branches 0.647524759 seconds time elapsed Versus: Performance counter stats for 'perf sched lat': 473.564691 task-clock # 0.999 CPUs utilized 9 context-switches # 0.000 M/sec 0 CPU-migrations # 0.000 M/sec 20,903 page-faults # 0.044 M/sec 944,367,984 cycles # 1.994 GHz stalled-cycles-frontend stalled-cycles-backend 1,442,385,571 instructions # 1.53 insns per cycle 308,383,106 branches # 651.195 M/sec 4,481,784 branch-misses # 1.45% of all branches 0.474215751 seconds time elapsed [root@emilia ~]# Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-1kbzpl74lwi6lavpqke2u2p3@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 42 ++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/evlist.h | 11 +++++++++++ tools/perf/util/evsel.h | 4 ++++ tools/perf/util/tool.h | 5 +++++ 4 files changed, 62 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index d44e3df13a8f..b36f26fe767a 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -207,6 +207,48 @@ out_free_attrs: return err; } +static struct perf_evsel * + perf_evlist__find_tracepoint_by_id(struct perf_evlist *evlist, int id) +{ + struct perf_evsel *evsel; + + list_for_each_entry(evsel, &evlist->entries, node) { + if (evsel->attr.type == PERF_TYPE_TRACEPOINT && + (int)evsel->attr.config == id) + return evsel; + } + + return NULL; +} + +int perf_evlist__set_tracepoints_handlers(struct perf_evlist *evlist, + const struct perf_evsel_str_handler *assocs, + size_t nr_assocs) +{ + struct perf_evsel *evsel; + int err; + size_t i; + + for (i = 0; i < nr_assocs; i++) { + err = trace_event__id(assocs[i].name); + if (err < 0) + goto out; + + evsel = perf_evlist__find_tracepoint_by_id(evlist, err); + if (evsel == NULL) + continue; + + err = -EEXIST; + if (evsel->handler.func != NULL) + goto out; + evsel->handler.func = assocs[i].handler; + } + + err = 0; +out: + return err; +} + void perf_evlist__disable(struct perf_evlist *evlist) { int cpu, thread; diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index 2202e7b04103..f94ed7e0d987 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -36,6 +36,11 @@ struct perf_evlist { struct perf_evsel *selected; }; +struct perf_evsel_str_handler { + const char *name; + void *handler; +}; + struct perf_evsel; struct perf_evlist *perf_evlist__new(struct cpu_map *cpus, @@ -51,6 +56,9 @@ int perf_evlist__add_attrs(struct perf_evlist *evlist, struct perf_event_attr *attrs, size_t nr_attrs); int perf_evlist__add_tracepoints(struct perf_evlist *evlist, const char *tracepoints[], size_t nr_tracepoints); +int perf_evlist__set_tracepoints_handlers(struct perf_evlist *evlist, + const struct perf_evsel_str_handler *assocs, + size_t nr_assocs); #define perf_evlist__add_attrs_array(evlist, array) \ perf_evlist__add_attrs(evlist, array, ARRAY_SIZE(array)) @@ -58,6 +66,9 @@ int perf_evlist__add_tracepoints(struct perf_evlist *evlist, #define perf_evlist__add_tracepoints_array(evlist, array) \ perf_evlist__add_tracepoints(evlist, array, ARRAY_SIZE(array)) +#define perf_evlist__set_tracepoints_handlers_array(evlist, array) \ + perf_evlist__set_tracepoints_handlers(evlist, array, ARRAY_SIZE(array)) + void perf_evlist__id_add(struct perf_evlist *evlist, struct perf_evsel *evsel, int cpu, int thread, u64 id); diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index 6421c07f5015..326b8e4d5035 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -61,6 +61,10 @@ struct perf_evsel { off_t id_offset; }; struct cgroup_sel *cgrp; + struct { + void *func; + void *data; + } handler; bool supported; }; diff --git a/tools/perf/util/tool.h b/tools/perf/util/tool.h index 89ff1b551a74..b0e1aadba8d5 100644 --- a/tools/perf/util/tool.h +++ b/tools/perf/util/tool.h @@ -1,8 +1,13 @@ #ifndef __PERF_TOOL_H #define __PERF_TOOL_H +#include + struct perf_session; +union perf_event; +struct perf_evlist; struct perf_evsel; +struct perf_sample; struct perf_tool; struct machine; -- cgit v1.2.2 From 806fb63007447622dd61d9767b4403919737e120 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Tue, 29 Nov 2011 08:05:52 -0200 Subject: perf evlist: Always do automatic allocation of pollfd and mmap structures At first tools were required to do that, but while writing the python bindings to simplify the API I made them auto-allocate when needed. This just makes record, stat and top use that auto allocation, simplifying them a bit. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-iokhcvkzzijr3keioubx8hlq@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 4 ++-- tools/perf/util/evlist.h | 2 -- 2 files changed, 2 insertions(+), 4 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index b36f26fe767a..8b19e7a1e881 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -275,7 +275,7 @@ void perf_evlist__enable(struct perf_evlist *evlist) } } -int perf_evlist__alloc_pollfd(struct perf_evlist *evlist) +static int perf_evlist__alloc_pollfd(struct perf_evlist *evlist) { int nfds = evlist->cpus->nr * evlist->threads->nr * evlist->nr_entries; evlist->pollfd = malloc(sizeof(struct pollfd) * nfds); @@ -431,7 +431,7 @@ void perf_evlist__munmap(struct perf_evlist *evlist) evlist->mmap = NULL; } -int perf_evlist__alloc_mmap(struct perf_evlist *evlist) +static int perf_evlist__alloc_mmap(struct perf_evlist *evlist) { evlist->nr_mmaps = evlist->cpus->nr; if (evlist->cpus->map[0] == -1) diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h index f94ed7e0d987..8922aeed0467 100644 --- a/tools/perf/util/evlist.h +++ b/tools/perf/util/evlist.h @@ -72,7 +72,6 @@ int perf_evlist__set_tracepoints_handlers(struct perf_evlist *evlist, void perf_evlist__id_add(struct perf_evlist *evlist, struct perf_evsel *evsel, int cpu, int thread, u64 id); -int perf_evlist__alloc_pollfd(struct perf_evlist *evlist); void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd); struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id); @@ -89,7 +88,6 @@ int perf_evlist__prepare_workload(struct perf_evlist *evlist, const char *argv[]); int perf_evlist__start_workload(struct perf_evlist *evlist); -int perf_evlist__alloc_mmap(struct perf_evlist *evlist); int perf_evlist__mmap(struct perf_evlist *evlist, unsigned int pages, bool overwrite); void perf_evlist__munmap(struct perf_evlist *evlist); -- cgit v1.2.2 From 482ad89745f7121020f6cee38aa4e894a4e7d642 Mon Sep 17 00:00:00 2001 From: Arnaldo Carvalho de Melo Date: Fri, 2 Dec 2011 11:06:37 -0200 Subject: perf event: Introduce perf_event__fprintf So that tools like 'perf test' can print the events when in verbose mode, for instance. Cc: David Ahern Cc: Frederic Weisbecker Cc: Mike Galbraith Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/n/tip-xnovdqfi25nc48gy6604k7yp@git.kernel.org Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/event.c | 54 +++++++++++++++++++++++++++++++++++++++++++------ tools/perf/util/event.h | 6 ++++++ 2 files changed, 54 insertions(+), 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 0ebbe7641335..97c479bcb0dc 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -433,6 +433,11 @@ int perf_event__synthesize_kernel_mmap(struct perf_tool *tool, return err; } +size_t perf_event__fprintf_comm(union perf_event *event, FILE *fp) +{ + return fprintf(fp, ": %s:%d\n", event->comm.comm, event->comm.tid); +} + int perf_event__process_comm(struct perf_tool *tool __used, union perf_event *event, struct perf_sample *sample __used, @@ -440,7 +445,8 @@ int perf_event__process_comm(struct perf_tool *tool __used, { struct thread *thread = machine__findnew_thread(machine, event->comm.tid); - dump_printf(": %s:%d\n", event->comm.comm, event->comm.tid); + if (dump_trace) + perf_event__fprintf_comm(event, stdout); if (thread == NULL || thread__set_comm(thread, event->comm.comm)) { dump_printf("problem processing PERF_RECORD_COMM, skipping event.\n"); @@ -566,6 +572,13 @@ out_problem: return -1; } +size_t perf_event__fprintf_mmap(union perf_event *event, FILE *fp) +{ + return fprintf(fp, " %d/%d: [%#" PRIx64 "(%#" PRIx64 ") @ %#" PRIx64 "]: %s\n", + event->mmap.pid, event->mmap.tid, event->mmap.start, + event->mmap.len, event->mmap.pgoff, event->mmap.filename); +} + int perf_event__process_mmap(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample __used, @@ -576,9 +589,8 @@ int perf_event__process_mmap(struct perf_tool *tool, u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; int ret = 0; - dump_printf(" %d/%d: [%#" PRIx64 "(%#" PRIx64 ") @ %#" PRIx64 "]: %s\n", - event->mmap.pid, event->mmap.tid, event->mmap.start, - event->mmap.len, event->mmap.pgoff, event->mmap.filename); + if (dump_trace) + perf_event__fprintf_mmap(event, stdout); if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL || cpumode == PERF_RECORD_MISC_KERNEL) { @@ -606,6 +618,13 @@ out_problem: return 0; } +size_t perf_event__fprintf_task(union perf_event *event, FILE *fp) +{ + return fprintf(fp, "(%d:%d):(%d:%d)\n", + event->fork.pid, event->fork.tid, + event->fork.ppid, event->fork.ptid); +} + int perf_event__process_task(struct perf_tool *tool __used, union perf_event *event, struct perf_sample *sample __used, @@ -614,8 +633,8 @@ int perf_event__process_task(struct perf_tool *tool __used, struct thread *thread = machine__findnew_thread(machine, event->fork.tid); struct thread *parent = machine__findnew_thread(machine, event->fork.ptid); - dump_printf("(%d:%d):(%d:%d)\n", event->fork.pid, event->fork.tid, - event->fork.ppid, event->fork.ptid); + if (dump_trace) + perf_event__fprintf_task(event, stdout); if (event->header.type == PERF_RECORD_EXIT) { machine__remove_thread(machine, thread); @@ -631,6 +650,29 @@ int perf_event__process_task(struct perf_tool *tool __used, return 0; } +size_t perf_event__fprintf(union perf_event *event, FILE *fp) +{ + size_t ret = fprintf(fp, "PERF_RECORD_%s", + perf_event__name(event->header.type)); + + switch (event->header.type) { + case PERF_RECORD_COMM: + ret += perf_event__fprintf_comm(event, fp); + break; + case PERF_RECORD_FORK: + case PERF_RECORD_EXIT: + ret += perf_event__fprintf_task(event, fp); + break; + case PERF_RECORD_MMAP: + ret += perf_event__fprintf_mmap(event, fp); + break; + default: + ret += fprintf(fp, "\n"); + } + + return ret; +} + int perf_event__process(struct perf_tool *tool, union perf_event *event, struct perf_sample *sample, struct machine *machine) { diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index d8499e7cf641..0d80201ce844 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -2,6 +2,7 @@ #define __PERF_RECORD_H #include +#include #include "../perf.h" #include "map.h" @@ -199,4 +200,9 @@ int perf_event__parse_sample(const union perf_event *event, u64 type, int sample_size, bool sample_id_all, struct perf_sample *sample, bool swapped); +size_t perf_event__fprintf_comm(union perf_event *event, FILE *fp); +size_t perf_event__fprintf_mmap(union perf_event *event, FILE *fp); +size_t perf_event__fprintf_task(union perf_event *event, FILE *fp); +size_t perf_event__fprintf(union perf_event *event, FILE *fp); + #endif /* __PERF_RECORD_H */ -- cgit v1.2.2 From d06c27b22aa66e48e32f03f9387328a9af9b0625 Mon Sep 17 00:00:00 2001 From: Steven Rostedt Date: Fri, 4 Nov 2011 16:32:25 -0400 Subject: perf: Fix parsing of __print_flags() in TP_printk() A update is made to the sched:sched_switch event that adds some logic to the first parameter of the __print_flags() that shows the state of tasks. This change cause perf to fail parsing the flags. A simple fix is needed to have the parser be able to process ops within the argument. Cc: stable@vger.kernel.org Reported-by: Andrew Vagin Signed-off-by: Steven Rostedt --- tools/perf/util/trace-event-parse.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/trace-event-parse.c b/tools/perf/util/trace-event-parse.c index 0a7ed5b5e281..6c164dc9ee95 100644 --- a/tools/perf/util/trace-event-parse.c +++ b/tools/perf/util/trace-event-parse.c @@ -1537,6 +1537,8 @@ process_flags(struct event *event, struct print_arg *arg, char **tok) field = malloc_or_die(sizeof(*field)); type = process_arg(event, field, &token); + while (type == EVENT_OP) + type = process_op(event, field, &token); if (test_type_token(type, token, EVENT_DELIM, ",")) goto out_free; -- cgit v1.2.2 From 6340cfed48c478cf67451ddcd2daa3438e642632 Mon Sep 17 00:00:00 2001 From: Andrew Vagin Date: Mon, 28 Nov 2011 12:03:29 +0300 Subject: perf header: Use event_name() to get an event name perf_evsel.name may be not initialized Cc: David Ahern Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: Arun Sharma Cc: devel@openvz.org Link: http://lkml.kernel.org/r/1322471015-107825-2-git-send-email-avagin@openvz.org Signed-off-by: Andrew Vagin Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/header.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index bcd05d05b4f0..33c17a2b2a81 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -388,7 +388,7 @@ static int write_event_desc(int fd, struct perf_header *h __used, /* * write event string as passed on cmdline */ - ret = do_write_string(fd, attr->name); + ret = do_write_string(fd, event_name(attr)); if (ret < 0) return ret; /* -- cgit v1.2.2 From 74eec26facadbe6dbc0621bc862892c915c4534f Mon Sep 17 00:00:00 2001 From: Andrew Vagin Date: Mon, 28 Nov 2011 12:03:31 +0300 Subject: perf tools: Add ability to synthesize event according to a sample It's the counterpart of perf_session__parse_sample. v2: fixed mistakes found by David Ahern. v3: s/data/sample/ s/perf_event__change_sample/perf_event__synthesize_sample Reviewed-by: David Ahern Cc: Arun Sharma Cc: David Ahern Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: devel@openvz.org Link: http://lkml.kernel.org/r/1323266161-394927-3-git-send-email-avagin@openvz.org Signed-off-by: Andrew Vagin Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/event.h | 3 ++ tools/perf/util/evsel.c | 79 +++++++++++++++++++++++++++++++++++++++++++++++ tools/perf/util/session.h | 8 +++++ 3 files changed, 90 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index 0d80201ce844..cbdeaad9c5e5 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -199,6 +199,9 @@ const char *perf_event__name(unsigned int id); int perf_event__parse_sample(const union perf_event *event, u64 type, int sample_size, bool sample_id_all, struct perf_sample *sample, bool swapped); +int perf_event__synthesize_sample(union perf_event *event, u64 type, + const struct perf_sample *sample, + bool swapped); size_t perf_event__fprintf_comm(union perf_event *event, FILE *fp); size_t perf_event__fprintf_mmap(union perf_event *event, FILE *fp); diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index ee68d6944e61..4a8c8b02e9cc 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -574,3 +574,82 @@ int perf_event__parse_sample(const union perf_event *event, u64 type, return 0; } + +int perf_event__synthesize_sample(union perf_event *event, u64 type, + const struct perf_sample *sample, + bool swapped) +{ + u64 *array; + + /* + * used for cross-endian analysis. See git commit 65014ab3 + * for why this goofiness is needed. + */ + union { + u64 val64; + u32 val32[2]; + } u; + + array = event->sample.array; + + if (type & PERF_SAMPLE_IP) { + event->ip.ip = sample->ip; + array++; + } + + if (type & PERF_SAMPLE_TID) { + u.val32[0] = sample->pid; + u.val32[1] = sample->tid; + if (swapped) { + /* + * Inverse of what is done in perf_event__parse_sample + */ + u.val32[0] = bswap_32(u.val32[0]); + u.val32[1] = bswap_32(u.val32[1]); + u.val64 = bswap_64(u.val64); + } + + *array = u.val64; + array++; + } + + if (type & PERF_SAMPLE_TIME) { + *array = sample->time; + array++; + } + + if (type & PERF_SAMPLE_ADDR) { + *array = sample->addr; + array++; + } + + if (type & PERF_SAMPLE_ID) { + *array = sample->id; + array++; + } + + if (type & PERF_SAMPLE_STREAM_ID) { + *array = sample->stream_id; + array++; + } + + if (type & PERF_SAMPLE_CPU) { + u.val32[0] = sample->cpu; + if (swapped) { + /* + * Inverse of what is done in perf_event__parse_sample + */ + u.val32[0] = bswap_32(u.val32[0]); + u.val64 = bswap_64(u.val64); + } + *array = u.val64; + array++; + } + + if (type & PERF_SAMPLE_PERIOD) { + *array = sample->period; + array++; + } + + return 0; +} diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index 30e9c6b6fc3c..fb696124ad61 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -134,6 +134,14 @@ static inline int perf_session__parse_sample(struct perf_session *session, session->header.needs_swap); } +static inline int perf_session__synthesize_sample(struct perf_session *session, + union perf_event *event, + const struct perf_sample *sample) +{ + return perf_event__synthesize_sample(event, session->sample_type, + sample, session->header.needs_swap); +} + struct perf_evsel *perf_session__find_first_evtype(struct perf_session *session, unsigned int type); -- cgit v1.2.2 From 3e76ac78b08479e84a3eca3fb1b3066fb8230461 Mon Sep 17 00:00:00 2001 From: Andrew Vagin Date: Tue, 20 Dec 2011 17:32:45 +0300 Subject: perf record: Add ability to record event period The problem is that when SAMPLE_PERIOD is not set, the kernel generates a number of samples in proportion to an event's period. Number of these samples may be too big and the kernel throttles all samples above a defined limit. E.g.: I want to trace when a process sleeps. I created a process which sleeps for 1ms and for 4ms. perf got 100 events in both cases. swapper 0 [000] 1141.371830: sched_stat_sleep: comm=foo pid=1801 delay=1386750 [ns] swapper 0 [000] 1141.369444: sched_stat_sleep: comm=foo pid=1801 delay=4499585 [ns] In the first case a kernel want to send 4499585 events and in the second case it wants to send 1386750 events. perf-reports shows that process sleeps in both places equal time. Instead of this we can get only one sample with an attribute period. As result we have less data transferring between kernel and user-space and we avoid throttling of samples. The patch "events: Don't divide events if it has field period" added a kernel part of this functionality. Acked-by: Arun Sharma Cc: Arun Sharma Cc: David Ahern Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Cc: devel@openvz.org Link: http://lkml.kernel.org/r/1324391565-1369947-1-git-send-email-avagin@openvz.org Signed-off-by: Andrew Vagin Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evsel.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index 4a8c8b02e9cc..60ad0286759e 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -108,6 +108,9 @@ void perf_evsel__config(struct perf_evsel *evsel, struct perf_record_opts *opts) if (opts->system_wide) attr->sample_type |= PERF_SAMPLE_CPU; + if (opts->period) + attr->sample_type |= PERF_SAMPLE_PERIOD; + if (opts->sample_id_all_avail && (opts->sample_time || opts->system_wide || !opts->no_inherit || opts->cpu_list)) -- cgit v1.2.2 From f3bda2c9a689b38c059f7cb2d761ff58a2996370 Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Thu, 15 Dec 2011 17:32:39 +0100 Subject: perf evsel: Fix uninitialized memory access to struct perf_sample Memory in struct perf_sample is not fully initialized during parsing. Depending on sampling data some parts may left unchanged. Zero out struct perf_sample first to avoid access to uninitialized memory. Cc: Ingo Molnar Link: http://lkml.kernel.org/r/1323966762-8574-2-git-send-email-robert.richter@amd.com Signed-off-by: Robert Richter Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evsel.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index 60ad0286759e..667f3b78bb2c 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -460,7 +460,7 @@ int perf_event__parse_sample(const union perf_event *event, u64 type, u32 val32[2]; } u; - + memset(data, 0, sizeof(*data)); data->cpu = data->pid = data->tid = -1; data->stream_id = data->id = data->time = -1ULL; -- cgit v1.2.2 From 301b195db179241da8be25f345f3c4e64960f1d5 Mon Sep 17 00:00:00 2001 From: Nelson Elhage Date: Mon, 19 Dec 2011 08:39:30 -0500 Subject: perf evlist: Fix errno value reporting on failed mmap On failure, perf_evlist__mmap_per_{cpu,thread} will try to munmap() every map that doesn't have a NULL base. This will fail with EINVAL if one of them has base == MAP_FAILED, clobbering errno, so that perf_evlist__map will return EINVAL on any failure regardless of the root cause. Fix this by resetting failed maps to a NULL base. Acked-by: Namhyung Kim Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1324301972-22740-2-git-send-email-nelhage@nelhage.com Signed-off-by: Nelson Elhage Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 4 +++- 1 file changed, 3 insertions(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 8b19e7a1e881..963d63dde457 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -447,8 +447,10 @@ static int __perf_evlist__mmap(struct perf_evlist *evlist, evlist->mmap[idx].mask = mask; evlist->mmap[idx].base = mmap(NULL, evlist->mmap_len, prot, MAP_SHARED, fd, 0); - if (evlist->mmap[idx].base == MAP_FAILED) + if (evlist->mmap[idx].base == MAP_FAILED) { + evlist->mmap[idx].base = NULL; return -1; + } perf_evlist__add_pollfd(evlist, fd); return 0; -- cgit v1.2.2 From 2b600f9578852d12af59420011e3dadfaa58b043 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 13 Dec 2011 00:16:51 +0900 Subject: perf symbols: Get rid of duplicated snprintf() The 'path' variable is set on a upper line, don't need to do it again. Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1323703017-6060-3-git-send-email-namhyung@gmail.com Signed-off-by: Namhyung Kim Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/symbol.c | 7 +------ 1 file changed, 1 insertion(+), 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 632b50c7bc26..e54b13d4c357 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -1757,7 +1757,7 @@ static int map_groups__set_modules_path_dir(struct map_groups *mg, struct stat st; /*sshfs might return bad dent->d_type, so we have to stat*/ - sprintf(path, "%s/%s", dir_name, dent->d_name); + snprintf(path, sizeof(path), "%s/%s", dir_name, dent->d_name); if (stat(path, &st)) continue; @@ -1766,8 +1766,6 @@ static int map_groups__set_modules_path_dir(struct map_groups *mg, !strcmp(dent->d_name, "..")) continue; - snprintf(path, sizeof(path), "%s/%s", - dir_name, dent->d_name); ret = map_groups__set_modules_path_dir(mg, path); if (ret < 0) goto out; @@ -1788,9 +1786,6 @@ static int map_groups__set_modules_path_dir(struct map_groups *mg, if (map == NULL) continue; - snprintf(path, sizeof(path), "%s/%s", - dir_name, dent->d_name); - long_name = strdup(path); if (long_name == NULL) { ret = -1; -- cgit v1.2.2 From d74c896b7e3250a07f7d0315eecdd2ae1a7bc3c3 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 13 Dec 2011 00:16:52 +0900 Subject: perf symbols: Fix error path on symbol__init() The order of freeing comm_list and dso_list should be reversed. Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1323703017-6060-4-git-send-email-namhyung@gmail.com Signed-off-by: Namhyung Kim Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/symbol.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index e54b13d4c357..215d50f2042e 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -2604,10 +2604,10 @@ int symbol__init(void) symbol_conf.initialized = true; return 0; -out_free_dso_list: - strlist__delete(symbol_conf.dso_list); out_free_comm_list: strlist__delete(symbol_conf.comm_list); +out_free_dso_list: + strlist__delete(symbol_conf.dso_list); return -1; } -- cgit v1.2.2 From 0161d82e9b740caa90f508138d1ae1b9d981b6d3 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 13 Dec 2011 00:16:53 +0900 Subject: perf tools: Fix a memory leak on perf_read_values_destroy After freeing each elements of the @values->value, we should free itself too. Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1323703017-6060-5-git-send-email-namhyung@gmail.com Signed-off-by: Namhyung Kim Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/values.c | 1 + 1 file changed, 1 insertion(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/values.c b/tools/perf/util/values.c index bdd33470b235..697c8b4e59cc 100644 --- a/tools/perf/util/values.c +++ b/tools/perf/util/values.c @@ -32,6 +32,7 @@ void perf_read_values_destroy(struct perf_read_values *values) for (i = 0; i < values->threads; i++) free(values->value[i]); + free(values->value); free(values->pid); free(values->tid); free(values->counterrawid); -- cgit v1.2.2 From 5f9273d64a5ccbd3c2b4446cc8b71123ed5d6366 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 13 Dec 2011 22:52:03 +0900 Subject: perf tools: Remove stale git headlines from top comment These files are part of PERF not GIT although they're come from there :) Cc: Ingo Molnar Cc: Johannes Schindelin Cc: Linus Torvalds Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1323784323-2150-1-git-send-email-namhyung@gmail.com Signed-off-by: Namhyung Kim Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/config.c | 5 ++++- tools/perf/util/usage.c | 5 ++++- 2 files changed, 8 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/config.c b/tools/perf/util/config.c index 80d9598db31a..0deac6a14b65 100644 --- a/tools/perf/util/config.c +++ b/tools/perf/util/config.c @@ -1,5 +1,8 @@ /* - * GIT - The information manager from hell + * config.c + * + * Helper functions for parsing config items. + * Originally copied from GIT source. * * Copyright (C) Linus Torvalds, 2005 * Copyright (C) Johannes Schindelin, 2005 diff --git a/tools/perf/util/usage.c b/tools/perf/util/usage.c index e16bf9a707e8..d76d1c0ff98f 100644 --- a/tools/perf/util/usage.c +++ b/tools/perf/util/usage.c @@ -1,5 +1,8 @@ /* - * GIT - The information manager from hell + * usage.c + * + * Various reporting routines. + * Originally copied from GIT source. * * Copyright (C) Linus Torvalds, 2005 */ -- cgit v1.2.2 From cb8f4e9aa37c469ddd80dda51469f327606c0118 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Tue, 13 Dec 2011 00:16:55 +0900 Subject: perf events: Tidy up perf_event__preprocess_sample Use local variable 'dso' to reduce typing a bit and rearrange the if condition. Also NULL check of al->map in the condition is not necessary. Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1323703017-6060-7-git-send-email-namhyung@gmail.com Signed-off-by: Namhyung Kim Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/event.c | 13 +++++++------ 1 file changed, 7 insertions(+), 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 97c479bcb0dc..b7c7f39a8f6d 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -814,13 +814,14 @@ int perf_event__preprocess_sample(const union perf_event *event, al->cpu = sample->cpu; if (al->map) { + struct dso *dso = al->map->dso; + if (symbol_conf.dso_list && - (!al->map || !al->map->dso || - !(strlist__has_entry(symbol_conf.dso_list, - al->map->dso->short_name) || - (al->map->dso->short_name != al->map->dso->long_name && - strlist__has_entry(symbol_conf.dso_list, - al->map->dso->long_name))))) + (!dso || !(strlist__has_entry(symbol_conf.dso_list, + dso->short_name) || + (dso->short_name != dso->long_name && + strlist__has_entry(symbol_conf.dso_list, + dso->long_name))))) goto out_filtered; al->sym = map__find_symbol(al->map, al->addr, filter); -- cgit v1.2.2 From f1ac18af219835fd5b8e19c14d2dd75c55f78737 Mon Sep 17 00:00:00 2001 From: Stephane Eranian Date: Sun, 11 Dec 2011 00:28:54 +0100 Subject: perf: Add support for PERF_HW_COUNT_REF_CPU_CYCLES Add new generic hw event: ref-cycles, which maps to PERF_HW_COUNT_REF_CPUCYCLES: $ perf stat -e ref-cycles ls Signed-off-by: Stephane Eranian Signed-off-by: Peter Zijlstra Link: http://lkml.kernel.org/r/1323559734-3488-5-git-send-email-eranian@google.com Signed-off-by: Ingo Molnar --- tools/perf/util/parse-events.c | 2 ++ 1 file changed, 2 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 586ab3fe60f8..531c283fc0c5 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -38,6 +38,7 @@ static struct event_symbol event_symbols[] = { { CHW(BRANCH_INSTRUCTIONS), "branch-instructions", "branches" }, { CHW(BRANCH_MISSES), "branch-misses", "" }, { CHW(BUS_CYCLES), "bus-cycles", "" }, + { CHW(REF_CPU_CYCLES), "ref-cycles", "" }, { CSW(CPU_CLOCK), "cpu-clock", "" }, { CSW(TASK_CLOCK), "task-clock", "" }, @@ -68,6 +69,7 @@ static const char *hw_event_names[PERF_COUNT_HW_MAX] = { "bus-cycles", "stalled-cycles-frontend", "stalled-cycles-backend", + "ref-cycles", }; static const char *sw_event_names[PERF_COUNT_SW_MAX] = { -- cgit v1.2.2 From defd8d38773cf9e01c69a903d04d5895b78ee74f Mon Sep 17 00:00:00 2001 From: David Ahern Date: Thu, 22 Dec 2011 11:30:01 -0700 Subject: perf tools: Fix comm for processes with named threads perf does not properly handle monitoring of processes with named threads. For example: $ ps -C myapp -L PID LWP TTY TIME CMD 25118 25118 ? 00:00:00 myapp 25118 25119 ? 00:00:00 myapp:worker perf record -e cs -c 1 -fo /tmp/perf.data -p 25118 -- sleep 10 perf report --stdio -i /tmp/perf.data 100.00% myapp:worker [kernel.kallsyms] [k] perf_event_task_sched_out The process name is set to the name of the last thread it finds for the process. The Problem: perf-top and perf-record both create a thread_map of threads to be monitored. That map is used in perf_event__synthesize_thread_map which loops over the entries in thread_map and calls __event__synthesize_thread to generate COMM and MMAP events. __event__synthesize_thread calls perf_event__synthesize_comm which opens /proc/pid/status, reads the name of the task and its thread group id. That's all fine. The problem is that it then reads /proc/pid/task and generates COMM events for each task it finds - but using the name found in /proc/pid/status where pid is the thread of interest. The end result (looping over thread_map + synthesizing comm events for each thread each time) means the name of the last thread processed sets the name for all threads in the process - which is not good for multithreaded processes with named threads. The Fix: perf_event__synthesize_comm has an input argument (full) that decides whether to process task entries for each pid it is passed. It currently never set to 0 (perf_event__synthesize_comm has a single caller and it always passes the value 1). Let's fix that. Add the full input argument to __event__synthesize_thread which passes it to perf_event__synthesize_comm. For thread/process monitoring set full to 0 which means COMM and MMAP events are only generated for the pid passed to it. For system wide monitoring set full to 1 so that COMM events are generated for all threads in a process. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1324578603-12762-2-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/event.c | 37 ++++++++++++++++++++++++++++++++----- 1 file changed, 32 insertions(+), 5 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index b7c7f39a8f6d..a5787260181a 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -261,11 +261,12 @@ int perf_event__synthesize_modules(struct perf_tool *tool, static int __event__synthesize_thread(union perf_event *comm_event, union perf_event *mmap_event, - pid_t pid, perf_event__handler_t process, + pid_t pid, int full, + perf_event__handler_t process, struct perf_tool *tool, struct machine *machine) { - pid_t tgid = perf_event__synthesize_comm(tool, comm_event, pid, 1, + pid_t tgid = perf_event__synthesize_comm(tool, comm_event, pid, full, process, machine); if (tgid == -1) return -1; @@ -279,7 +280,7 @@ int perf_event__synthesize_thread_map(struct perf_tool *tool, struct machine *machine) { union perf_event *comm_event, *mmap_event; - int err = -1, thread; + int err = -1, thread, j; comm_event = malloc(sizeof(comm_event->comm) + machine->id_hdr_size); if (comm_event == NULL) @@ -292,11 +293,37 @@ int perf_event__synthesize_thread_map(struct perf_tool *tool, err = 0; for (thread = 0; thread < threads->nr; ++thread) { if (__event__synthesize_thread(comm_event, mmap_event, - threads->map[thread], + threads->map[thread], 0, process, tool, machine)) { err = -1; break; } + + /* + * comm.pid is set to thread group id by + * perf_event__synthesize_comm + */ + if ((int) comm_event->comm.pid != threads->map[thread]) { + bool need_leader = true; + + /* is thread group leader in thread_map? */ + for (j = 0; j < threads->nr; ++j) { + if ((int) comm_event->comm.pid == threads->map[j]) { + need_leader = false; + break; + } + } + + /* if not, generate events for it */ + if (need_leader && + __event__synthesize_thread(comm_event, + mmap_event, + comm_event->comm.pid, 0, + process, tool, machine)) { + err = -1; + break; + } + } } free(mmap_event); out_free_comm: @@ -333,7 +360,7 @@ int perf_event__synthesize_threads(struct perf_tool *tool, if (*end) /* only interested in proper numerical dirents */ continue; - __event__synthesize_thread(comm_event, mmap_event, pid, + __event__synthesize_thread(comm_event, mmap_event, pid, 1, process, tool, machine); } -- cgit v1.2.2 From f5faf726184a6a5ca1735f610cb97e509fce33e2 Mon Sep 17 00:00:00 2001 From: David Ahern Date: Thu, 22 Dec 2011 11:30:02 -0700 Subject: perf tools: Look up thread names for system wide profiling This handles multithreaded processes with named threads when doing system wide profiling: the comm for each thread is looked up allowing them to be different from the thread group leader. v2: - fixed sizeof arg to perf_event__get_comm_tgid Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1324578603-12762-3-git-send-email-dsahern@gmail.com Signed-off-by: David Ahern Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/event.c | 75 ++++++++++++++++++++++++++++++++++--------------- 1 file changed, 53 insertions(+), 22 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index a5787260181a..73ddaf06b8e7 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -43,37 +43,27 @@ static struct perf_sample synth_sample = { .period = 1, }; -static pid_t perf_event__synthesize_comm(struct perf_tool *tool, - union perf_event *event, pid_t pid, - int full, perf_event__handler_t process, - struct machine *machine) +static pid_t perf_event__get_comm_tgid(pid_t pid, char *comm, size_t len) { char filename[PATH_MAX]; char bf[BUFSIZ]; FILE *fp; size_t size = 0; - DIR *tasks; - struct dirent dirent, *next; - pid_t tgid = 0; + pid_t tgid = -1; snprintf(filename, sizeof(filename), "/proc/%d/status", pid); fp = fopen(filename, "r"); if (fp == NULL) { -out_race: - /* - * We raced with a task exiting - just return: - */ pr_debug("couldn't open %s\n", filename); return 0; } - memset(&event->comm, 0, sizeof(event->comm)); - - while (!event->comm.comm[0] || !event->comm.pid) { + while (!comm[0] || (tgid < 0)) { if (fgets(bf, sizeof(bf), fp) == NULL) { - pr_warning("couldn't get COMM and pgid, malformed %s\n", filename); - goto out; + pr_warning("couldn't get COMM and pgid, malformed %s\n", + filename); + break; } if (memcmp(bf, "Name:", 5) == 0) { @@ -81,16 +71,46 @@ out_race: while (*name && isspace(*name)) ++name; size = strlen(name) - 1; - memcpy(event->comm.comm, name, size++); + if (size >= len) + size = len - 1; + memcpy(comm, name, size); + } else if (memcmp(bf, "Tgid:", 5) == 0) { char *tgids = bf + 5; while (*tgids && isspace(*tgids)) ++tgids; - tgid = event->comm.pid = atoi(tgids); + tgid = atoi(tgids); } } + fclose(fp); + + return tgid; +} + +static pid_t perf_event__synthesize_comm(struct perf_tool *tool, + union perf_event *event, pid_t pid, + int full, + perf_event__handler_t process, + struct machine *machine) +{ + char filename[PATH_MAX]; + size_t size; + DIR *tasks; + struct dirent dirent, *next; + pid_t tgid; + + memset(&event->comm, 0, sizeof(event->comm)); + + tgid = perf_event__get_comm_tgid(pid, event->comm.comm, + sizeof(event->comm.comm)); + if (tgid < 0) + goto out; + + event->comm.pid = tgid; event->comm.header.type = PERF_RECORD_COMM; + + size = strlen(event->comm.comm) + 1; size = ALIGN(size, sizeof(u64)); memset(event->comm.comm + size, 0, machine->id_hdr_size); event->comm.header.size = (sizeof(event->comm) - @@ -106,8 +126,10 @@ out_race: snprintf(filename, sizeof(filename), "/proc/%d/task", pid); tasks = opendir(filename); - if (tasks == NULL) - goto out_race; + if (tasks == NULL) { + pr_debug("couldn't open %s\n", filename); + return 0; + } while (!readdir_r(tasks, &dirent, &next) && next) { char *end; @@ -115,6 +137,17 @@ out_race: if (*end) continue; + /* already have tgid; jut want to update the comm */ + (void) perf_event__get_comm_tgid(pid, event->comm.comm, + sizeof(event->comm.comm)); + + size = strlen(event->comm.comm) + 1; + size = ALIGN(size, sizeof(u64)); + memset(event->comm.comm + size, 0, machine->id_hdr_size); + event->comm.header.size = (sizeof(event->comm) - + (sizeof(event->comm.comm) - size) + + machine->id_hdr_size); + event->comm.tid = pid; process(tool, event, &synth_sample, machine); @@ -122,8 +155,6 @@ out_race: closedir(tasks); out: - fclose(fp); - return tgid; } -- cgit v1.2.2 From f41612f43be9575e1160460b08c3a760e6e27e1b Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Fri, 23 Dec 2011 14:08:04 +0100 Subject: perf tools: Fix truncated annotation MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit I get such truncated annotation results in 'perf top': : Disassembly of section .text: ▒ : ▒ : ffffffff810966a8 : ▒ 4.94 : ffffffff810966a8: movslq %edi,%rdi ▒ 3.70 : ffffffff810966ab: mov $0x13700,%rax ▒ 0.00 : ffffffff810966b2: add -0x7e32cb00(,%rdi,8),%rax ▒ 8.64 : ffffffff810966ba: mov 0x7e0(%rax),%eax ▒ 82.72 : ffffffff810966c0: cltq ▒ Note the missing 'retq' which is there in the original function: ffffffff810966a8 : ffffffff810966a8: 48 63 ff movslq %edi,%rdi ffffffff810966ab: 48 c7 c0 00 37 01 00 mov $0x13700,%rax ffffffff810966b2: 48 03 04 fd 00 35 cd add -0x7e32cb00(,%rdi,8),%rax ffffffff810966b9: 81 ffffffff810966ba: 8b 80 e0 07 00 00 mov 0x7e0(%rax),%eax ffffffff810966c0: 48 98 cltq ffffffff810966c2: c3 retq ffffffff810966c3 : I'm using a fairly recent binutils: GNU objdump version 2.21.51.0.6-2.fc16 20110118 AFAICS the bug is simply that sym->end points to the last byte of the symbol in question - while objdump's --stop-address expects the last byte plus 1 to disassemble the full range. Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/20111223130804.GA24305@elte.hu Signed-off-by: Ingo Molnar Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/annotate.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c index 376e643f7066..011ed2676604 100644 --- a/tools/perf/util/annotate.c +++ b/tools/perf/util/annotate.c @@ -334,7 +334,7 @@ fallback: disassembler_style ? "-M " : "", disassembler_style ? disassembler_style : "", map__rip_2objdump(map, sym->start), - map__rip_2objdump(map, sym->end), + map__rip_2objdump(map, sym->end+1), symbol_conf.annotate_asm_raw ? "" : "--no-show-raw", symbol_conf.annotate_src ? "-S" : "", symfs_filename, filename); -- cgit v1.2.2 From 41d0d933494ce10eb77758a1168b08e317c42e8e Mon Sep 17 00:00:00 2001 From: Nelson Elhage Date: Mon, 19 Dec 2011 08:39:32 -0500 Subject: perf: builtin-record: Document and check that mmap_pages must be a power of two. Now that we automatically point users at it, let's provide them some guidance so that they hopefully don't just get mysterious EINVAL's from the kernel. Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1324301972-22740-4-git-send-email-nelhage@nelhage.com Signed-off-by: Nelson Elhage [ committer note: Made it work after 50a682c ] Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/evlist.c | 2 ++ tools/perf/util/util.h | 11 +++++++++++ 2 files changed, 13 insertions(+) (limited to 'tools/perf/util') diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c index 963d63dde457..fa1837088ca8 100644 --- a/tools/perf/util/evlist.c +++ b/tools/perf/util/evlist.c @@ -563,6 +563,8 @@ int perf_evlist__mmap(struct perf_evlist *evlist, unsigned int pages, /* 512 kiB: default amount of unprivileged mlocked memory */ if (pages == UINT_MAX) pages = (512 * 1024) / page_size; + else if (!is_power_of_2(pages)) + return -EINVAL; mask = pages * page_size - 1; diff --git a/tools/perf/util/util.h b/tools/perf/util/util.h index 0128906bac88..37be34dff798 100644 --- a/tools/perf/util/util.h +++ b/tools/perf/util/util.h @@ -245,4 +245,15 @@ int readn(int fd, void *buf, size_t size); #define _STR(x) #x #define STR(x) _STR(x) +/* + * Determine whether some value is a power of two, where zero is + * *not* considered a power of two. + */ + +static inline __attribute__((const)) +bool is_power_of_2(unsigned long n) +{ + return (n != 0 && ((n & (n - 1)) == 0)); +} + #endif -- cgit v1.2.2 From 8cdfa78a885d94a79205d183a611ebc4876d6f33 Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Wed, 7 Dec 2011 10:02:56 +0100 Subject: perf tools: Improve macros for struct feature_ops Reducing duplication and line size by extending function names for print and write from a single name. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1323248577-11268-7-git-send-email-robert.richter@amd.com Signed-off-by: Robert Richter Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/header.c | 40 ++++++++++++++++++++++------------------ 1 file changed, 22 insertions(+), 18 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 5b01449152ef..4c48be80dcdf 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -1065,26 +1065,30 @@ struct feature_ops { bool full_only; }; -#define FEAT_OPA(n, w, p) \ - [n] = { .name = #n, .write = w, .print = p } -#define FEAT_OPF(n, w, p) \ - [n] = { .name = #n, .write = w, .print = p, .full_only = true } +#define FEAT_OPA(n, func) \ + [n] = { .name = #n, .write = write_##func, .print = print_##func } +#define FEAT_OPF(n, func) \ + [n] = { .name = #n, .write = write_##func, .print = print_##func, .full_only = true } + +/* feature_ops not implemented: */ +#define print_trace_info NULL +#define print_build_id NULL static const struct feature_ops feat_ops[HEADER_LAST_FEATURE] = { - FEAT_OPA(HEADER_TRACE_INFO, write_trace_info, NULL), - FEAT_OPA(HEADER_BUILD_ID, write_build_id, NULL), - FEAT_OPA(HEADER_HOSTNAME, write_hostname, print_hostname), - FEAT_OPA(HEADER_OSRELEASE, write_osrelease, print_osrelease), - FEAT_OPA(HEADER_VERSION, write_version, print_version), - FEAT_OPA(HEADER_ARCH, write_arch, print_arch), - FEAT_OPA(HEADER_NRCPUS, write_nrcpus, print_nrcpus), - FEAT_OPA(HEADER_CPUDESC, write_cpudesc, print_cpudesc), - FEAT_OPA(HEADER_CPUID, write_cpuid, print_cpuid), - FEAT_OPA(HEADER_TOTAL_MEM, write_total_mem, print_total_mem), - FEAT_OPA(HEADER_EVENT_DESC, write_event_desc, print_event_desc), - FEAT_OPA(HEADER_CMDLINE, write_cmdline, print_cmdline), - FEAT_OPF(HEADER_CPU_TOPOLOGY, write_cpu_topology, print_cpu_topology), - FEAT_OPF(HEADER_NUMA_TOPOLOGY, write_numa_topology, print_numa_topology), + FEAT_OPA(HEADER_TRACE_INFO, trace_info), + FEAT_OPA(HEADER_BUILD_ID, build_id), + FEAT_OPA(HEADER_HOSTNAME, hostname), + FEAT_OPA(HEADER_OSRELEASE, osrelease), + FEAT_OPA(HEADER_VERSION, version), + FEAT_OPA(HEADER_ARCH, arch), + FEAT_OPA(HEADER_NRCPUS, nrcpus), + FEAT_OPA(HEADER_CPUDESC, cpudesc), + FEAT_OPA(HEADER_CPUID, cpuid), + FEAT_OPA(HEADER_TOTAL_MEM, total_mem), + FEAT_OPA(HEADER_EVENT_DESC, event_desc), + FEAT_OPA(HEADER_CMDLINE, cmdline), + FEAT_OPF(HEADER_CPU_TOPOLOGY, cpu_topology), + FEAT_OPF(HEADER_NUMA_TOPOLOGY, numa_topology), }; struct header_print_data { -- cgit v1.2.2 From f7a8a1336416883dc0ccd96c17c604e34de61c25 Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Wed, 7 Dec 2011 10:02:51 +0100 Subject: perf tools: Continue processing header on unknown features A feature may be unknown if perf.data is created and parsed on different perf tool versions. This should not stop the header to be processed, instead continue processing it. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1323248577-11268-2-git-send-email-robert.richter@amd.com Signed-off-by: Robert Richter Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/header.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 4c48be80dcdf..428a4a2ce350 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -1109,7 +1109,7 @@ static int perf_file_section__fprintf_info(struct perf_file_section *section, } if (feat < HEADER_TRACE_INFO || feat >= HEADER_LAST_FEATURE) { pr_warning("unknown feature %d\n", feat); - return -1; + return 0; } if (!feat_ops[feat].print) return 0; -- cgit v1.2.2 From 002c4fd92d772becf8745b9cbcebe5c95fe6dad0 Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Wed, 7 Dec 2011 10:02:52 +0100 Subject: perf tools: Fix out-of-bound access to struct perf_session If filename is NULL there is an out-of-bound access to struct perf_session if it would be used with perf_session__open(). Shouldn't actually happen in current implementation as filename is always !NULL. Fixing this by always null-terminating filename. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1323248577-11268-3-git-send-email-robert.richter@amd.com Signed-off-by: Robert Richter Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/session.c | 2 +- tools/perf/util/session.h | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index d9318d8a9ba1..ea17dfb85baa 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -107,7 +107,7 @@ struct perf_session *perf_session__new(const char *filename, int mode, bool force, bool repipe, struct perf_tool *tool) { - size_t len = filename ? strlen(filename) + 1 : 0; + size_t len = filename ? strlen(filename) : 0; struct perf_session *self = zalloc(sizeof(*self) + len); if (self == NULL) diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index fb696124ad61..37bc38381fb6 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -50,7 +50,7 @@ struct perf_session { int cwdlen; char *cwd; struct ordered_samples ordered_samples; - char filename[0]; + char filename[1]; }; struct perf_tool; -- cgit v1.2.2 From 1b5495043d5bc058def21f9b66fd8feaa794eb44 Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Wed, 7 Dec 2011 10:02:53 +0100 Subject: perf tools: Moving code in some files Needed for later changes. No modified functionality. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1323248577-11268-4-git-send-email-robert.richter@amd.com Signed-off-by: Robert Richter Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/header.c | 495 +++++++++++++++++++++++------------------------ 1 file changed, 246 insertions(+), 249 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 428a4a2ce350..609d79b5fb5e 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -28,9 +28,6 @@ static struct perf_trace_event_type *events; static u32 header_argc; static const char **header_argv; -static int dsos__write_buildid_table(struct perf_header *header, int fd); -static int perf_session__cache_build_ids(struct perf_session *session); - int perf_header__push_event(u64 id, const char *name) { if (strlen(name) > MAX_EVENT_NAME) @@ -187,6 +184,252 @@ perf_header__set_cmdline(int argc, const char **argv) return 0; } +#define dsos__for_each_with_build_id(pos, head) \ + list_for_each_entry(pos, head, node) \ + if (!pos->has_build_id) \ + continue; \ + else + +static int __dsos__write_buildid_table(struct list_head *head, pid_t pid, + u16 misc, int fd) +{ + struct dso *pos; + + dsos__for_each_with_build_id(pos, head) { + int err; + struct build_id_event b; + size_t len; + + if (!pos->hit) + continue; + len = pos->long_name_len + 1; + len = ALIGN(len, NAME_ALIGN); + memset(&b, 0, sizeof(b)); + memcpy(&b.build_id, pos->build_id, sizeof(pos->build_id)); + b.pid = pid; + b.header.misc = misc; + b.header.size = sizeof(b) + len; + err = do_write(fd, &b, sizeof(b)); + if (err < 0) + return err; + err = write_padded(fd, pos->long_name, + pos->long_name_len + 1, len); + if (err < 0) + return err; + } + + return 0; +} + +static int machine__write_buildid_table(struct machine *machine, int fd) +{ + int err; + u16 kmisc = PERF_RECORD_MISC_KERNEL, + umisc = PERF_RECORD_MISC_USER; + + if (!machine__is_host(machine)) { + kmisc = PERF_RECORD_MISC_GUEST_KERNEL; + umisc = PERF_RECORD_MISC_GUEST_USER; + } + + err = __dsos__write_buildid_table(&machine->kernel_dsos, machine->pid, + kmisc, fd); + if (err == 0) + err = __dsos__write_buildid_table(&machine->user_dsos, + machine->pid, umisc, fd); + return err; +} + +static int dsos__write_buildid_table(struct perf_header *header, int fd) +{ + struct perf_session *session = container_of(header, + struct perf_session, header); + struct rb_node *nd; + int err = machine__write_buildid_table(&session->host_machine, fd); + + if (err) + return err; + + for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { + struct machine *pos = rb_entry(nd, struct machine, rb_node); + err = machine__write_buildid_table(pos, fd); + if (err) + break; + } + return err; +} + +int build_id_cache__add_s(const char *sbuild_id, const char *debugdir, + const char *name, bool is_kallsyms) +{ + const size_t size = PATH_MAX; + char *realname, *filename = zalloc(size), + *linkname = zalloc(size), *targetname; + int len, err = -1; + + if (is_kallsyms) { + if (symbol_conf.kptr_restrict) { + pr_debug("Not caching a kptr_restrict'ed /proc/kallsyms\n"); + return 0; + } + realname = (char *)name; + } else + realname = realpath(name, NULL); + + if (realname == NULL || filename == NULL || linkname == NULL) + goto out_free; + + len = snprintf(filename, size, "%s%s%s", + debugdir, is_kallsyms ? "/" : "", realname); + if (mkdir_p(filename, 0755)) + goto out_free; + + snprintf(filename + len, sizeof(filename) - len, "/%s", sbuild_id); + + if (access(filename, F_OK)) { + if (is_kallsyms) { + if (copyfile("/proc/kallsyms", filename)) + goto out_free; + } else if (link(realname, filename) && copyfile(name, filename)) + goto out_free; + } + + len = snprintf(linkname, size, "%s/.build-id/%.2s", + debugdir, sbuild_id); + + if (access(linkname, X_OK) && mkdir_p(linkname, 0755)) + goto out_free; + + snprintf(linkname + len, size - len, "/%s", sbuild_id + 2); + targetname = filename + strlen(debugdir) - 5; + memcpy(targetname, "../..", 5); + + if (symlink(targetname, linkname) == 0) + err = 0; +out_free: + if (!is_kallsyms) + free(realname); + free(filename); + free(linkname); + return err; +} + +static int build_id_cache__add_b(const u8 *build_id, size_t build_id_size, + const char *name, const char *debugdir, + bool is_kallsyms) +{ + char sbuild_id[BUILD_ID_SIZE * 2 + 1]; + + build_id__sprintf(build_id, build_id_size, sbuild_id); + + return build_id_cache__add_s(sbuild_id, debugdir, name, is_kallsyms); +} + +int build_id_cache__remove_s(const char *sbuild_id, const char *debugdir) +{ + const size_t size = PATH_MAX; + char *filename = zalloc(size), + *linkname = zalloc(size); + int err = -1; + + if (filename == NULL || linkname == NULL) + goto out_free; + + snprintf(linkname, size, "%s/.build-id/%.2s/%s", + debugdir, sbuild_id, sbuild_id + 2); + + if (access(linkname, F_OK)) + goto out_free; + + if (readlink(linkname, filename, size - 1) < 0) + goto out_free; + + if (unlink(linkname)) + goto out_free; + + /* + * Since the link is relative, we must make it absolute: + */ + snprintf(linkname, size, "%s/.build-id/%.2s/%s", + debugdir, sbuild_id, filename); + + if (unlink(linkname)) + goto out_free; + + err = 0; +out_free: + free(filename); + free(linkname); + return err; +} + +static int dso__cache_build_id(struct dso *dso, const char *debugdir) +{ + bool is_kallsyms = dso->kernel && dso->long_name[0] != '/'; + + return build_id_cache__add_b(dso->build_id, sizeof(dso->build_id), + dso->long_name, debugdir, is_kallsyms); +} + +static int __dsos__cache_build_ids(struct list_head *head, const char *debugdir) +{ + struct dso *pos; + int err = 0; + + dsos__for_each_with_build_id(pos, head) + if (dso__cache_build_id(pos, debugdir)) + err = -1; + + return err; +} + +static int machine__cache_build_ids(struct machine *machine, const char *debugdir) +{ + int ret = __dsos__cache_build_ids(&machine->kernel_dsos, debugdir); + ret |= __dsos__cache_build_ids(&machine->user_dsos, debugdir); + return ret; +} + +static int perf_session__cache_build_ids(struct perf_session *session) +{ + struct rb_node *nd; + int ret; + char debugdir[PATH_MAX]; + + snprintf(debugdir, sizeof(debugdir), "%s", buildid_dir); + + if (mkdir(debugdir, 0755) != 0 && errno != EEXIST) + return -1; + + ret = machine__cache_build_ids(&session->host_machine, debugdir); + + for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { + struct machine *pos = rb_entry(nd, struct machine, rb_node); + ret |= machine__cache_build_ids(pos, debugdir); + } + return ret ? -1 : 0; +} + +static bool machine__read_build_ids(struct machine *machine, bool with_hits) +{ + bool ret = __dsos__read_build_ids(&machine->kernel_dsos, with_hits); + ret |= __dsos__read_build_ids(&machine->user_dsos, with_hits); + return ret; +} + +static bool perf_session__read_build_ids(struct perf_session *session, bool with_hits) +{ + struct rb_node *nd; + bool ret = machine__read_build_ids(&session->host_machine, with_hits); + + for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { + struct machine *pos = rb_entry(nd, struct machine, rb_node); + ret |= machine__read_build_ids(pos, with_hits); + } + + return ret; +} + static int write_trace_info(int fd, struct perf_header *h __used, struct perf_evlist *evlist) { @@ -1136,252 +1379,6 @@ int perf_header__fprintf_info(struct perf_session *session, FILE *fp, bool full) return 0; } -#define dsos__for_each_with_build_id(pos, head) \ - list_for_each_entry(pos, head, node) \ - if (!pos->has_build_id) \ - continue; \ - else - -static int __dsos__write_buildid_table(struct list_head *head, pid_t pid, - u16 misc, int fd) -{ - struct dso *pos; - - dsos__for_each_with_build_id(pos, head) { - int err; - struct build_id_event b; - size_t len; - - if (!pos->hit) - continue; - len = pos->long_name_len + 1; - len = ALIGN(len, NAME_ALIGN); - memset(&b, 0, sizeof(b)); - memcpy(&b.build_id, pos->build_id, sizeof(pos->build_id)); - b.pid = pid; - b.header.misc = misc; - b.header.size = sizeof(b) + len; - err = do_write(fd, &b, sizeof(b)); - if (err < 0) - return err; - err = write_padded(fd, pos->long_name, - pos->long_name_len + 1, len); - if (err < 0) - return err; - } - - return 0; -} - -static int machine__write_buildid_table(struct machine *machine, int fd) -{ - int err; - u16 kmisc = PERF_RECORD_MISC_KERNEL, - umisc = PERF_RECORD_MISC_USER; - - if (!machine__is_host(machine)) { - kmisc = PERF_RECORD_MISC_GUEST_KERNEL; - umisc = PERF_RECORD_MISC_GUEST_USER; - } - - err = __dsos__write_buildid_table(&machine->kernel_dsos, machine->pid, - kmisc, fd); - if (err == 0) - err = __dsos__write_buildid_table(&machine->user_dsos, - machine->pid, umisc, fd); - return err; -} - -static int dsos__write_buildid_table(struct perf_header *header, int fd) -{ - struct perf_session *session = container_of(header, - struct perf_session, header); - struct rb_node *nd; - int err = machine__write_buildid_table(&session->host_machine, fd); - - if (err) - return err; - - for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { - struct machine *pos = rb_entry(nd, struct machine, rb_node); - err = machine__write_buildid_table(pos, fd); - if (err) - break; - } - return err; -} - -int build_id_cache__add_s(const char *sbuild_id, const char *debugdir, - const char *name, bool is_kallsyms) -{ - const size_t size = PATH_MAX; - char *realname, *filename = zalloc(size), - *linkname = zalloc(size), *targetname; - int len, err = -1; - - if (is_kallsyms) { - if (symbol_conf.kptr_restrict) { - pr_debug("Not caching a kptr_restrict'ed /proc/kallsyms\n"); - return 0; - } - realname = (char *)name; - } else - realname = realpath(name, NULL); - - if (realname == NULL || filename == NULL || linkname == NULL) - goto out_free; - - len = snprintf(filename, size, "%s%s%s", - debugdir, is_kallsyms ? "/" : "", realname); - if (mkdir_p(filename, 0755)) - goto out_free; - - snprintf(filename + len, sizeof(filename) - len, "/%s", sbuild_id); - - if (access(filename, F_OK)) { - if (is_kallsyms) { - if (copyfile("/proc/kallsyms", filename)) - goto out_free; - } else if (link(realname, filename) && copyfile(name, filename)) - goto out_free; - } - - len = snprintf(linkname, size, "%s/.build-id/%.2s", - debugdir, sbuild_id); - - if (access(linkname, X_OK) && mkdir_p(linkname, 0755)) - goto out_free; - - snprintf(linkname + len, size - len, "/%s", sbuild_id + 2); - targetname = filename + strlen(debugdir) - 5; - memcpy(targetname, "../..", 5); - - if (symlink(targetname, linkname) == 0) - err = 0; -out_free: - if (!is_kallsyms) - free(realname); - free(filename); - free(linkname); - return err; -} - -static int build_id_cache__add_b(const u8 *build_id, size_t build_id_size, - const char *name, const char *debugdir, - bool is_kallsyms) -{ - char sbuild_id[BUILD_ID_SIZE * 2 + 1]; - - build_id__sprintf(build_id, build_id_size, sbuild_id); - - return build_id_cache__add_s(sbuild_id, debugdir, name, is_kallsyms); -} - -int build_id_cache__remove_s(const char *sbuild_id, const char *debugdir) -{ - const size_t size = PATH_MAX; - char *filename = zalloc(size), - *linkname = zalloc(size); - int err = -1; - - if (filename == NULL || linkname == NULL) - goto out_free; - - snprintf(linkname, size, "%s/.build-id/%.2s/%s", - debugdir, sbuild_id, sbuild_id + 2); - - if (access(linkname, F_OK)) - goto out_free; - - if (readlink(linkname, filename, size - 1) < 0) - goto out_free; - - if (unlink(linkname)) - goto out_free; - - /* - * Since the link is relative, we must make it absolute: - */ - snprintf(linkname, size, "%s/.build-id/%.2s/%s", - debugdir, sbuild_id, filename); - - if (unlink(linkname)) - goto out_free; - - err = 0; -out_free: - free(filename); - free(linkname); - return err; -} - -static int dso__cache_build_id(struct dso *dso, const char *debugdir) -{ - bool is_kallsyms = dso->kernel && dso->long_name[0] != '/'; - - return build_id_cache__add_b(dso->build_id, sizeof(dso->build_id), - dso->long_name, debugdir, is_kallsyms); -} - -static int __dsos__cache_build_ids(struct list_head *head, const char *debugdir) -{ - struct dso *pos; - int err = 0; - - dsos__for_each_with_build_id(pos, head) - if (dso__cache_build_id(pos, debugdir)) - err = -1; - - return err; -} - -static int machine__cache_build_ids(struct machine *machine, const char *debugdir) -{ - int ret = __dsos__cache_build_ids(&machine->kernel_dsos, debugdir); - ret |= __dsos__cache_build_ids(&machine->user_dsos, debugdir); - return ret; -} - -static int perf_session__cache_build_ids(struct perf_session *session) -{ - struct rb_node *nd; - int ret; - char debugdir[PATH_MAX]; - - snprintf(debugdir, sizeof(debugdir), "%s", buildid_dir); - - if (mkdir(debugdir, 0755) != 0 && errno != EEXIST) - return -1; - - ret = machine__cache_build_ids(&session->host_machine, debugdir); - - for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { - struct machine *pos = rb_entry(nd, struct machine, rb_node); - ret |= machine__cache_build_ids(pos, debugdir); - } - return ret ? -1 : 0; -} - -static bool machine__read_build_ids(struct machine *machine, bool with_hits) -{ - bool ret = __dsos__read_build_ids(&machine->kernel_dsos, with_hits); - ret |= __dsos__read_build_ids(&machine->user_dsos, with_hits); - return ret; -} - -static bool perf_session__read_build_ids(struct perf_session *session, bool with_hits) -{ - struct rb_node *nd; - bool ret = machine__read_build_ids(&session->host_machine, with_hits); - - for (nd = rb_first(&session->machines); nd; nd = rb_next(nd)) { - struct machine *pos = rb_entry(nd, struct machine, rb_node); - ret |= machine__read_build_ids(pos, with_hits); - } - - return ret; -} - static int do_write_feat(int fd, struct perf_header *h, int type, struct perf_file_section **p, struct perf_evlist *evlist) -- cgit v1.2.2 From efad14150a0b4429f37da7245001a8096ef7ee38 Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Wed, 7 Dec 2011 10:02:54 +0100 Subject: perf report: Accept fifos as input file The default input file for perf report is not handled the same way as perf record does it for its output file. This leads to unexpected behavior of perf report, etc. E.g.: # perf record -a -e cpu-cycles sleep 2 | perf report | cat failed to open perf.data: No such file or directory (try 'perf record' first) While perf record writes to a fifo, perf report expects perf.data to be read. This patch changes this to accept fifos as input file. Applies to the following commands: perf annotate perf buildid-list perf evlist perf kmem perf lock perf report perf sched perf script perf timechart Also fixes char const* -> const char* type declaration for filename strings. v2: * Prevent potential null pointer access to input_name in builtin-report.c. Needed due to removal of patch "perf report: Setup browser if stdout is a pipe" Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1323248577-11268-5-git-send-email-robert.richter@amd.com Signed-off-by: Robert Richter Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/session.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index ea17dfb85baa..cc5e6be46d86 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -107,8 +107,19 @@ struct perf_session *perf_session__new(const char *filename, int mode, bool force, bool repipe, struct perf_tool *tool) { - size_t len = filename ? strlen(filename) : 0; - struct perf_session *self = zalloc(sizeof(*self) + len); + struct perf_session *self; + struct stat st; + size_t len; + + if (!filename || !strlen(filename)) { + if (!fstat(STDIN_FILENO, &st) && S_ISFIFO(st.st_mode)) + filename = "-"; + else + filename = "perf.data"; + } + + len = strlen(filename); + self = zalloc(sizeof(*self) + len); if (self == NULL) goto out; -- cgit v1.2.2 From e20960c0271f91aead94746872fd976326a703b3 Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Wed, 7 Dec 2011 10:02:55 +0100 Subject: perf tools: Unify handling of features when writing feature section The features HEADER_TRACE_INFO and HEADER_BUILD_ID are handled different when writing the feature section. All other features are simply disabled on failure and writing the section goes on without returning an error. There is no reason for these special cases. This patch unifies handling of the features. This should be ok since all features can be parsed independently. Offset and size of a feature's block is stored in struct perf_file_ section right after the data block of perf.data (see perf_session__ write_header()). Thus, if a feature does not exist then other features can be processed anyway. Also moving special code for HEADER_BUILD_ID out to write_build_id(). v2: * perf record throws an error now if buildids may not be generated, which can be disabled with the --no-buildid option. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1323248577-11268-6-git-send-email-robert.richter@amd.com Signed-off-by: Robert Richter Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/header.c | 14 +++++--------- 1 file changed, 5 insertions(+), 9 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 609d79b5fb5e..71326836921b 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -445,6 +445,9 @@ static int write_build_id(int fd, struct perf_header *h, session = container_of(h, struct perf_session, header); + if (!perf_session__read_build_ids(session, true)) + return -1; + err = dsos__write_buildid_table(h, fd); if (err < 0) { pr_debug("failed to write buildid table\n"); @@ -1417,10 +1420,6 @@ static int perf_header__adds_write(struct perf_header *header, session = container_of(header, struct perf_session, header); - if (perf_header__has_feat(header, HEADER_BUILD_ID && - !perf_session__read_build_ids(session, true))) - perf_header__clear_feat(header, HEADER_BUILD_ID); - nr_sections = bitmap_weight(header->adds_features, HEADER_FEAT_BITS); if (!nr_sections) return 0; @@ -1436,13 +1435,11 @@ static int perf_header__adds_write(struct perf_header *header, err = do_write_feat(fd, header, HEADER_TRACE_INFO, &p, evlist); if (err) - goto out_free; + perf_header__clear_feat(header, HEADER_TRACE_INFO); err = do_write_feat(fd, header, HEADER_BUILD_ID, &p, evlist); - if (err) { + if (err) perf_header__clear_feat(header, HEADER_BUILD_ID); - goto out_free; - } err = do_write_feat(fd, header, HEADER_HOSTNAME, &p, evlist); if (err) @@ -1500,7 +1497,6 @@ static int perf_header__adds_write(struct perf_header *header, err = do_write(fd, feat_sec, sec_size); if (err < 0) pr_debug("failed to write feature section\n"); -out_free: free(feat_sec); return err; } -- cgit v1.2.2 From b1e5a9bee3c342dd3281aef76d1be1044dd8addf Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Wed, 7 Dec 2011 10:02:57 +0100 Subject: perf tools: Use for_each_set_bit() to iterate over feature flags This patch introduces the for_each_set_bit() macro and modifies feature implementation to use it. Cc: Frederic Weisbecker Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1323248577-11268-8-git-send-email-robert.richter@amd.com Signed-off-by: Robert Richter Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/header.c | 118 ++++++++------------------------- tools/perf/util/header.h | 6 +- tools/perf/util/include/linux/bitops.h | 118 +++++++++++++++++++++++++++++++++ 3 files changed, 149 insertions(+), 93 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index 71326836921b..e509a9dea00b 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -8,6 +8,7 @@ #include #include #include +#include #include #include "evlist.h" @@ -1353,7 +1354,7 @@ static int perf_file_section__fprintf_info(struct perf_file_section *section, "%d, continuing...\n", section->offset, feat); return 0; } - if (feat < HEADER_TRACE_INFO || feat >= HEADER_LAST_FEATURE) { + if (feat >= HEADER_LAST_FEATURE) { pr_warning("unknown feature %d\n", feat); return 0; } @@ -1390,6 +1391,8 @@ static int do_write_feat(int fd, struct perf_header *h, int type, int ret = 0; if (perf_header__has_feat(h, type)) { + if (!feat_ops[type].write) + return -1; (*p)->offset = lseek(fd, 0, SEEK_CUR); @@ -1416,6 +1419,7 @@ static int perf_header__adds_write(struct perf_header *header, struct perf_file_section *feat_sec, *p; int sec_size; u64 sec_start; + int feat; int err; session = container_of(header, struct perf_session, header); @@ -1433,61 +1437,10 @@ static int perf_header__adds_write(struct perf_header *header, sec_start = header->data_offset + header->data_size; lseek(fd, sec_start + sec_size, SEEK_SET); - err = do_write_feat(fd, header, HEADER_TRACE_INFO, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_TRACE_INFO); - - err = do_write_feat(fd, header, HEADER_BUILD_ID, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_BUILD_ID); - - err = do_write_feat(fd, header, HEADER_HOSTNAME, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_HOSTNAME); - - err = do_write_feat(fd, header, HEADER_OSRELEASE, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_OSRELEASE); - - err = do_write_feat(fd, header, HEADER_VERSION, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_VERSION); - - err = do_write_feat(fd, header, HEADER_ARCH, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_ARCH); - - err = do_write_feat(fd, header, HEADER_NRCPUS, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_NRCPUS); - - err = do_write_feat(fd, header, HEADER_CPUDESC, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_CPUDESC); - - err = do_write_feat(fd, header, HEADER_CPUID, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_CPUID); - - err = do_write_feat(fd, header, HEADER_TOTAL_MEM, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_TOTAL_MEM); - - err = do_write_feat(fd, header, HEADER_CMDLINE, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_CMDLINE); - - err = do_write_feat(fd, header, HEADER_EVENT_DESC, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_EVENT_DESC); - - err = do_write_feat(fd, header, HEADER_CPU_TOPOLOGY, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_CPU_TOPOLOGY); - - err = do_write_feat(fd, header, HEADER_NUMA_TOPOLOGY, &p, evlist); - if (err) - perf_header__clear_feat(header, HEADER_NUMA_TOPOLOGY); + for_each_set_bit(feat, header->adds_features, HEADER_FEAT_BITS) { + if (do_write_feat(fd, header, feat, &p, evlist)) + perf_header__clear_feat(header, feat); + } lseek(fd, sec_start, SEEK_SET); /* @@ -1634,20 +1587,20 @@ static int perf_header__getbuffer64(struct perf_header *header, int perf_header__process_sections(struct perf_header *header, int fd, void *data, int (*process)(struct perf_file_section *section, - struct perf_header *ph, - int feat, int fd, void *data)) + struct perf_header *ph, + int feat, int fd, void *data)) { - struct perf_file_section *feat_sec; + struct perf_file_section *feat_sec, *sec; int nr_sections; int sec_size; - int idx = 0; - int err = -1, feat = 1; + int feat; + int err; nr_sections = bitmap_weight(header->adds_features, HEADER_FEAT_BITS); if (!nr_sections) return 0; - feat_sec = calloc(sizeof(*feat_sec), nr_sections); + feat_sec = sec = calloc(sizeof(*feat_sec), nr_sections); if (!feat_sec) return -1; @@ -1655,20 +1608,16 @@ int perf_header__process_sections(struct perf_header *header, int fd, lseek(fd, header->data_offset + header->data_size, SEEK_SET); - if (perf_header__getbuffer64(header, fd, feat_sec, sec_size)) + err = perf_header__getbuffer64(header, fd, feat_sec, sec_size); + if (err < 0) goto out_free; - err = 0; - while (idx < nr_sections && feat < HEADER_LAST_FEATURE) { - if (perf_header__has_feat(header, feat)) { - struct perf_file_section *sec = &feat_sec[idx++]; - - err = process(sec, header, feat, fd, data); - if (err < 0) - break; - } - ++feat; + for_each_set_bit(feat, header->adds_features, HEADER_LAST_FEATURE) { + err = process(sec++, header, feat, fd, data); + if (err < 0) + goto out_free; } + err = 0; out_free: free(feat_sec); return err; @@ -1903,32 +1852,21 @@ static int perf_file_section__process(struct perf_file_section *section, return 0; } + if (feat >= HEADER_LAST_FEATURE) { + pr_debug("unknown feature %d, continuing...\n", feat); + return 0; + } + switch (feat) { case HEADER_TRACE_INFO: trace_report(fd, false); break; - case HEADER_BUILD_ID: if (perf_header__read_build_ids(ph, fd, section->offset, section->size)) pr_debug("Failed to read buildids, continuing...\n"); break; - - case HEADER_HOSTNAME: - case HEADER_OSRELEASE: - case HEADER_VERSION: - case HEADER_ARCH: - case HEADER_NRCPUS: - case HEADER_CPUDESC: - case HEADER_CPUID: - case HEADER_TOTAL_MEM: - case HEADER_CMDLINE: - case HEADER_EVENT_DESC: - case HEADER_CPU_TOPOLOGY: - case HEADER_NUMA_TOPOLOGY: - break; - default: - pr_debug("unknown feature %d, continuing...\n", feat); + break; } return 0; diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index 09365b32098e..ac4ec956024e 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -10,7 +10,8 @@ #include enum { - HEADER_TRACE_INFO = 1, + HEADER_RESERVED = 0, /* always cleared */ + HEADER_TRACE_INFO = 1, HEADER_BUILD_ID, HEADER_HOSTNAME, @@ -27,10 +28,9 @@ enum { HEADER_NUMA_TOPOLOGY, HEADER_LAST_FEATURE, + HEADER_FEAT_BITS = 256, }; -#define HEADER_FEAT_BITS 256 - struct perf_file_section { u64 offset; u64 size; diff --git a/tools/perf/util/include/linux/bitops.h b/tools/perf/util/include/linux/bitops.h index 305c8484f200..62cdee78db7b 100644 --- a/tools/perf/util/include/linux/bitops.h +++ b/tools/perf/util/include/linux/bitops.h @@ -9,6 +9,17 @@ #define BITS_PER_BYTE 8 #define BITS_TO_LONGS(nr) DIV_ROUND_UP(nr, BITS_PER_BYTE * sizeof(long)) +#define for_each_set_bit(bit, addr, size) \ + for ((bit) = find_first_bit((addr), (size)); \ + (bit) < (size); \ + (bit) = find_next_bit((addr), (size), (bit) + 1)) + +/* same as for_each_set_bit() but use bit as value to start with */ +#define for_each_set_bit_cont(bit, addr, size) \ + for ((bit) = find_next_bit((addr), (size), (bit)); \ + (bit) < (size); \ + (bit) = find_next_bit((addr), (size), (bit) + 1)) + static inline void set_bit(int nr, unsigned long *addr) { addr[nr / BITS_PER_LONG] |= 1UL << (nr % BITS_PER_LONG); @@ -30,4 +41,111 @@ static inline unsigned long hweight_long(unsigned long w) return sizeof(w) == 4 ? hweight32(w) : hweight64(w); } +#define BITOP_WORD(nr) ((nr) / BITS_PER_LONG) + +/** + * __ffs - find first bit in word. + * @word: The word to search + * + * Undefined if no bit exists, so code should check against 0 first. + */ +static __always_inline unsigned long __ffs(unsigned long word) +{ + int num = 0; + +#if BITS_PER_LONG == 64 + if ((word & 0xffffffff) == 0) { + num += 32; + word >>= 32; + } +#endif + if ((word & 0xffff) == 0) { + num += 16; + word >>= 16; + } + if ((word & 0xff) == 0) { + num += 8; + word >>= 8; + } + if ((word & 0xf) == 0) { + num += 4; + word >>= 4; + } + if ((word & 0x3) == 0) { + num += 2; + word >>= 2; + } + if ((word & 0x1) == 0) + num += 1; + return num; +} + +/* + * Find the first set bit in a memory region. + */ +static inline unsigned long +find_first_bit(const unsigned long *addr, unsigned long size) +{ + const unsigned long *p = addr; + unsigned long result = 0; + unsigned long tmp; + + while (size & ~(BITS_PER_LONG-1)) { + if ((tmp = *(p++))) + goto found; + result += BITS_PER_LONG; + size -= BITS_PER_LONG; + } + if (!size) + return result; + + tmp = (*p) & (~0UL >> (BITS_PER_LONG - size)); + if (tmp == 0UL) /* Are any bits set? */ + return result + size; /* Nope. */ +found: + return result + __ffs(tmp); +} + +/* + * Find the next set bit in a memory region. + */ +static inline unsigned long +find_next_bit(const unsigned long *addr, unsigned long size, unsigned long offset) +{ + const unsigned long *p = addr + BITOP_WORD(offset); + unsigned long result = offset & ~(BITS_PER_LONG-1); + unsigned long tmp; + + if (offset >= size) + return size; + size -= result; + offset %= BITS_PER_LONG; + if (offset) { + tmp = *(p++); + tmp &= (~0UL << offset); + if (size < BITS_PER_LONG) + goto found_first; + if (tmp) + goto found_middle; + size -= BITS_PER_LONG; + result += BITS_PER_LONG; + } + while (size & ~(BITS_PER_LONG-1)) { + if ((tmp = *(p++))) + goto found_middle; + result += BITS_PER_LONG; + size -= BITS_PER_LONG; + } + if (!size) + return result; + tmp = *p; + +found_first: + tmp &= (~0UL >> (BITS_PER_LONG - size)); + if (tmp == 0UL) /* Are any bits set? */ + return result + size; /* Nope. */ +found_middle: + return result + __ffs(tmp); +} + #endif -- cgit v1.2.2 From 37a058ea006de0cc24553637afa788594a975176 Mon Sep 17 00:00:00 2001 From: Robert Richter Date: Thu, 15 Dec 2011 18:23:43 +0100 Subject: perf script: Add generic perl handler to process events The current perf scripting facility only supports tracepoints. This patch implements a generic perl handler to support other events than tracepoints too. This patch introduces a function process_event() that is called by perf for each sample. The function is called with byte streams as arguments containing information about the event, its attributes, the sample and raw data. Perl's unpack() function can easily be used for byte decoding. The following is the default implementation for process_event() that can also be generated with perf script: # Packed byte string args of process_event(): # # $event: union perf_event util/event.h # $attr: struct perf_event_attr linux/perf_event.h # $sample: struct perf_sample util/event.h # $raw_data: perf_sample->raw_data util/event.h sub process_event { my ($event, $attr, $sample, $raw_data) = @_; my @event = unpack("LSS", $event); my @attr = unpack("LLQQQQQLLQQ", $attr); my @sample = unpack("QLLQQQQQLL", $sample); my @raw_data = unpack("C*", $raw_data); use Data::Dumper; print Dumper \@event, \@attr, \@sample, \@raw_data; } Cc: Ingo Molnar Cc: Peter Zijlstra Cc: Stephane Eranian Link: http://lkml.kernel.org/r/1323969824-9711-4-git-send-email-robert.richter@amd.com Signed-off-by: Robert Richter Signed-off-by: Arnaldo Carvalho de Melo --- .../perf/util/scripting-engines/trace-event-perl.c | 73 ++++++++++++++++++++-- 1 file changed, 67 insertions(+), 6 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/scripting-engines/trace-event-perl.c b/tools/perf/util/scripting-engines/trace-event-perl.c index a82ce4303ff5..e30749e38a9b 100644 --- a/tools/perf/util/scripting-engines/trace-event-perl.c +++ b/tools/perf/util/scripting-engines/trace-event-perl.c @@ -30,6 +30,7 @@ #include "../thread.h" #include "../event.h" #include "../trace-event.h" +#include "../evsel.h" #include #include @@ -247,11 +248,11 @@ static inline struct event *find_cache_event(int type) return event; } -static void perl_process_event(union perf_event *pevent __unused, - struct perf_sample *sample, - struct perf_evsel *evsel, - struct machine *machine __unused, - struct thread *thread) +static void perl_process_tracepoint(union perf_event *pevent __unused, + struct perf_sample *sample, + struct perf_evsel *evsel, + struct machine *machine __unused, + struct thread *thread) { struct format_field *field; static char handler[256]; @@ -267,6 +268,9 @@ static void perl_process_event(union perf_event *pevent __unused, dSP; + if (evsel->attr.type != PERF_TYPE_TRACEPOINT) + return; + type = trace_parse_common_type(data); event = find_cache_event(type); @@ -334,6 +338,42 @@ static void perl_process_event(union perf_event *pevent __unused, LEAVE; } +static void perl_process_event_generic(union perf_event *pevent __unused, + struct perf_sample *sample, + struct perf_evsel *evsel __unused, + struct machine *machine __unused, + struct thread *thread __unused) +{ + dSP; + + if (!get_cv("process_event", 0)) + return; + + ENTER; + SAVETMPS; + PUSHMARK(SP); + XPUSHs(sv_2mortal(newSVpvn((const char *)pevent, pevent->header.size))); + XPUSHs(sv_2mortal(newSVpvn((const char *)&evsel->attr, sizeof(evsel->attr)))); + XPUSHs(sv_2mortal(newSVpvn((const char *)sample, sizeof(*sample)))); + XPUSHs(sv_2mortal(newSVpvn((const char *)sample->raw_data, sample->raw_size))); + PUTBACK; + call_pv("process_event", G_SCALAR); + SPAGAIN; + PUTBACK; + FREETMPS; + LEAVE; +} + +static void perl_process_event(union perf_event *pevent, + struct perf_sample *sample, + struct perf_evsel *evsel, + struct machine *machine, + struct thread *thread) +{ + perl_process_tracepoint(pevent, sample, evsel, machine, thread); + perl_process_event_generic(pevent, sample, evsel, machine, thread); +} + static void run_start_sub(void) { dSP; /* access to Perl stack */ @@ -555,7 +595,28 @@ static int perl_generate_script(const char *outfile) fprintf(ofp, "sub print_header\n{\n" "\tmy ($event_name, $cpu, $secs, $nsecs, $pid, $comm) = @_;\n\n" "\tprintf(\"%%-20s %%5u %%05u.%%09u %%8u %%-20s \",\n\t " - "$event_name, $cpu, $secs, $nsecs, $pid, $comm);\n}"); + "$event_name, $cpu, $secs, $nsecs, $pid, $comm);\n}\n"); + + fprintf(ofp, + "\n# Packed byte string args of process_event():\n" + "#\n" + "# $event:\tunion perf_event\tutil/event.h\n" + "# $attr:\tstruct perf_event_attr\tlinux/perf_event.h\n" + "# $sample:\tstruct perf_sample\tutil/event.h\n" + "# $raw_data:\tperf_sample->raw_data\tutil/event.h\n" + "\n" + "sub process_event\n" + "{\n" + "\tmy ($event, $attr, $sample, $raw_data) = @_;\n" + "\n" + "\tmy @event\t= unpack(\"LSS\", $event);\n" + "\tmy @attr\t= unpack(\"LLQQQQQLLQQ\", $attr);\n" + "\tmy @sample\t= unpack(\"QLLQQQQQLL\", $sample);\n" + "\tmy @raw_data\t= unpack(\"C*\", $raw_data);\n" + "\n" + "\tuse Data::Dumper;\n" + "\tprint Dumper \\@event, \\@attr, \\@sample, \\@raw_data;\n" + "}\n"); fclose(ofp); -- cgit v1.2.2 From f2328062726d36e562f1458d6346b77aa048acad Mon Sep 17 00:00:00 2001 From: Ingo Molnar Date: Thu, 29 Dec 2011 21:26:17 +0100 Subject: perf tools: Fix feature-bits rework fallout, remove unused variable Cc: Arnaldo Carvalho de Melo Cc: Frederic Weisbecker Cc: Peter Zijlstra Cc: Stephane Eranian Cc: Robert Richter Link: http://lkml.kernel.org/n/tip-lfckuwbl8m1ykb7t9ydsxe4r@git.kernel.org Signed-off-by: Ingo Molnar --- tools/perf/util/header.c | 3 --- 1 file changed, 3 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index e509a9dea00b..3e7e0b09c12c 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -1415,15 +1415,12 @@ static int perf_header__adds_write(struct perf_header *header, struct perf_evlist *evlist, int fd) { int nr_sections; - struct perf_session *session; struct perf_file_section *feat_sec, *p; int sec_size; u64 sec_start; int feat; int err; - session = container_of(header, struct perf_session, header); - nr_sections = bitmap_weight(header->adds_features, HEADER_FEAT_BITS); if (!nr_sections) return 0; -- cgit v1.2.2 From 29c9862f1b818bf4caa4c48a30dbe5f25c84ee08 Mon Sep 17 00:00:00 2001 From: Namhyung Kim Date: Wed, 28 Dec 2011 00:35:48 +0900 Subject: perf session: Remove impossible condition check The 'size' cannot be 0 because it was set to 8 on the above line in case it was 0 and never changed. Cc: Ingo Molnar Cc: Paul Mackerras Cc: Peter Zijlstra Link: http://lkml.kernel.org/r/1325000151-4463-1-git-send-email-namhyung@gmail.com Signed-off-by: Namhyung Kim Signed-off-by: Arnaldo Carvalho de Melo --- tools/perf/util/session.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'tools/perf/util') diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index cc5e6be46d86..b5ca2558c7bb 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -1015,8 +1015,7 @@ more: } } - if (size == 0 || - (skip = perf_session__process_event(self, &event, tool, head)) < 0) { + if ((skip = perf_session__process_event(self, &event, tool, head)) < 0) { dump_printf("%#" PRIx64 " [%#x]: skipping unknown header type: %d\n", head, event.header.size, event.header.type); /* -- cgit v1.2.2