diff options
Diffstat (limited to 'tools')
66 files changed, 5074 insertions, 2671 deletions
diff --git a/tools/perf/Documentation/perf-probe.txt b/tools/perf/Documentation/perf-probe.txt index 86b797a35aa6..81c3220e04f3 100644 --- a/tools/perf/Documentation/perf-probe.txt +++ b/tools/perf/Documentation/perf-probe.txt @@ -73,6 +73,17 @@ OPTIONS (Only for --vars) Show external defined variables in addition to local variables. +-F:: +--funcs:: + Show available functions in given module or kernel. + +--filter=FILTER:: + (Only for --vars and --funcs) Set filter. FILTER is a combination of glob + pattern, see FILTER PATTERN for detail. + Default FILTER is "!__k???tab_* & !__crc_*" for --vars, and "!_*" + for --funcs. + If several filters are specified, only the last filter is used. + -f:: --force:: Forcibly add events with existing name. @@ -135,6 +146,14 @@ e.g. This provides some sort of flexibility and robustness to probe point definitions against minor code changes. For example, actual 10th line of schedule() can be moved easily by modifying schedule(), but the same line matching 'rq=cpu_rq*' may still exist in the function.) +FILTER PATTERN +-------------- + The filter pattern is a glob matching pattern(s) to filter variables. + In addition, you can use "!" for specifying filter-out rule. You also can give several rules combined with "&" or "|", and fold those rules as one rule by using "(" ")". + +e.g. + With --filter "foo* | bar*", perf probe -V shows variables which start with "foo" or "bar". + With --filter "!foo* & *bar", perf probe -V shows variables which don't start with "foo" and end with "bar", like "fizzbar". But "foobar" is filtered out. EXAMPLES -------- diff --git a/tools/perf/Makefile b/tools/perf/Makefile index 7141c42e1469..94f73abdc56a 100644 --- a/tools/perf/Makefile +++ b/tools/perf/Makefile @@ -315,6 +315,7 @@ COMPAT_CFLAGS = COMPAT_OBJS = LIB_H = LIB_OBJS = +PYRF_OBJS = SCRIPT_PERL = SCRIPT_SH = TEST_PROGRAMS = @@ -324,6 +325,9 @@ SCRIPT_SH += perf-archive.sh grep-libs = $(filter -l%,$(1)) strip-libs = $(filter-out -l%,$(1)) +$(OUTPUT)python/perf.so: $(PYRF_OBJS) + $(QUIET_GEN)python util/setup.py --quiet build_ext --build-lib='$(OUTPUT)python' \ + --build-temp='$(OUTPUT)python/temp' # # No Perl scripts right now: # @@ -344,6 +348,8 @@ PROGRAMS += $(EXTRA_PROGRAMS) # PROGRAMS += $(OUTPUT)perf +LANG_BINDINGS = + # List built-in command $C whose implementation cmd_$C() is not in # builtin-$C.o but is linked in as part of some other command. # @@ -395,6 +401,7 @@ LIB_H += util/include/dwarf-regs.h LIB_H += util/include/asm/dwarf2.h LIB_H += util/include/asm/cpufeature.h LIB_H += perf.h +LIB_H += util/annotate.h LIB_H += util/cache.h LIB_H += util/callchain.h LIB_H += util/build-id.h @@ -402,6 +409,7 @@ LIB_H += util/debug.h LIB_H += util/debugfs.h LIB_H += util/event.h LIB_H += util/evsel.h +LIB_H += util/evlist.h LIB_H += util/exec_cmd.h LIB_H += util/types.h LIB_H += util/levenshtein.h @@ -416,6 +424,7 @@ LIB_H += util/help.h LIB_H += util/session.h LIB_H += util/strbuf.h LIB_H += util/strlist.h +LIB_H += util/strfilter.h LIB_H += util/svghelper.h LIB_H += util/run-command.h LIB_H += util/sigchain.h @@ -425,21 +434,25 @@ LIB_H += util/values.h LIB_H += util/sort.h LIB_H += util/hist.h LIB_H += util/thread.h +LIB_H += util/thread_map.h LIB_H += util/trace-event.h LIB_H += util/probe-finder.h LIB_H += util/probe-event.h LIB_H += util/pstack.h LIB_H += util/cpumap.h +LIB_H += util/top.h LIB_H += $(ARCH_INCLUDE) LIB_OBJS += $(OUTPUT)util/abspath.o LIB_OBJS += $(OUTPUT)util/alias.o +LIB_OBJS += $(OUTPUT)util/annotate.o LIB_OBJS += $(OUTPUT)util/build-id.o LIB_OBJS += $(OUTPUT)util/config.o LIB_OBJS += $(OUTPUT)util/ctype.o LIB_OBJS += $(OUTPUT)util/debugfs.o LIB_OBJS += $(OUTPUT)util/environment.o LIB_OBJS += $(OUTPUT)util/event.o +LIB_OBJS += $(OUTPUT)util/evlist.o LIB_OBJS += $(OUTPUT)util/evsel.o LIB_OBJS += $(OUTPUT)util/exec_cmd.o LIB_OBJS += $(OUTPUT)util/help.o @@ -455,6 +468,8 @@ LIB_OBJS += $(OUTPUT)util/quote.o LIB_OBJS += $(OUTPUT)util/strbuf.o LIB_OBJS += $(OUTPUT)util/string.o LIB_OBJS += $(OUTPUT)util/strlist.o +LIB_OBJS += $(OUTPUT)util/strfilter.o +LIB_OBJS += $(OUTPUT)util/top.o LIB_OBJS += $(OUTPUT)util/usage.o LIB_OBJS += $(OUTPUT)util/wrapper.o LIB_OBJS += $(OUTPUT)util/sigchain.o @@ -469,6 +484,7 @@ LIB_OBJS += $(OUTPUT)util/map.o LIB_OBJS += $(OUTPUT)util/pstack.o LIB_OBJS += $(OUTPUT)util/session.o LIB_OBJS += $(OUTPUT)util/thread.o +LIB_OBJS += $(OUTPUT)util/thread_map.o LIB_OBJS += $(OUTPUT)util/trace-event-parse.o LIB_OBJS += $(OUTPUT)util/trace-event-read.o LIB_OBJS += $(OUTPUT)util/trace-event-info.o @@ -514,6 +530,20 @@ BUILTIN_OBJS += $(OUTPUT)builtin-inject.o PERFLIBS = $(LIB_FILE) +# Files needed for the python binding, perf.so +# pyrf is just an internal name needed for all those wrappers. +# This has to be in sync with what is in the 'sources' variable in +# tools/perf/util/setup.py + +PYRF_OBJS += $(OUTPUT)util/cpumap.o +PYRF_OBJS += $(OUTPUT)util/ctype.o +PYRF_OBJS += $(OUTPUT)util/evlist.o +PYRF_OBJS += $(OUTPUT)util/evsel.o +PYRF_OBJS += $(OUTPUT)util/python.o +PYRF_OBJS += $(OUTPUT)util/thread_map.o +PYRF_OBJS += $(OUTPUT)util/util.o +PYRF_OBJS += $(OUTPUT)util/xyarray.o + # # Platform specific tweaks # @@ -595,6 +625,7 @@ else LIB_OBJS += $(OUTPUT)util/ui/browsers/annotate.o LIB_OBJS += $(OUTPUT)util/ui/browsers/hists.o LIB_OBJS += $(OUTPUT)util/ui/browsers/map.o + LIB_OBJS += $(OUTPUT)util/ui/browsers/top.o LIB_OBJS += $(OUTPUT)util/ui/helpline.o LIB_OBJS += $(OUTPUT)util/ui/progress.o LIB_OBJS += $(OUTPUT)util/ui/util.o @@ -635,12 +666,14 @@ else PYTHON_EMBED_CCOPTS = `python-config --cflags 2>/dev/null` FLAGS_PYTHON_EMBED=$(PYTHON_EMBED_CCOPTS) $(PYTHON_EMBED_LDOPTS) ifneq ($(call try-cc,$(SOURCE_PYTHON_EMBED),$(FLAGS_PYTHON_EMBED)),y) + msg := $(warning No Python.h found, install python-dev[el] to have python support in 'perf script' and to build the python bindings) BASIC_CFLAGS += -DNO_LIBPYTHON else ALL_LDFLAGS += $(PYTHON_EMBED_LDFLAGS) EXTLIBS += $(PYTHON_EMBED_LIBADD) LIB_OBJS += $(OUTPUT)util/scripting-engines/trace-event-python.o LIB_OBJS += $(OUTPUT)scripts/python/Perf-Trace-Util/Context.o + LANG_BINDINGS += $(OUTPUT)python/perf.so endif endif @@ -927,7 +960,7 @@ export TAR INSTALL DESTDIR SHELL_PATH SHELL = $(SHELL_PATH) -all:: shell_compatibility_test $(ALL_PROGRAMS) $(BUILT_INS) $(OTHER_PROGRAMS) $(OUTPUT)PERF-BUILD-OPTIONS +all:: shell_compatibility_test $(ALL_PROGRAMS) $(LANG_BINDINGS) $(BUILT_INS) $(OTHER_PROGRAMS) $(OUTPUT)PERF-BUILD-OPTIONS ifneq (,$X) $(foreach p,$(patsubst %$X,%,$(filter %$X,$(ALL_PROGRAMS) $(BUILT_INS) perf$X)), test '$p' -ef '$p$X' || $(RM) '$p';) endif @@ -1024,6 +1057,9 @@ $(OUTPUT)util/ui/browser.o: util/ui/browser.c $(OUTPUT)PERF-CFLAGS $(OUTPUT)util/ui/browsers/annotate.o: util/ui/browsers/annotate.c $(OUTPUT)PERF-CFLAGS $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< +$(OUTPUT)util/ui/browsers/top.o: util/ui/browsers/top.c $(OUTPUT)PERF-CFLAGS + $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< + $(OUTPUT)util/ui/browsers/hists.o: util/ui/browsers/hists.c $(OUTPUT)PERF-CFLAGS $(QUIET_CC)$(CC) -o $@ -c $(ALL_CFLAGS) -DENABLE_SLFUTURE_CONST $< @@ -1257,17 +1293,17 @@ distclean: clean # $(RM) configure clean: - $(RM) *.o */*.o */*/*.o */*/*/*.o $(LIB_FILE) + $(RM) $(OUTPUT){*.o,*/*.o,*/*/*.o,*/*/*/*.o,$(LIB_FILE),perf-archive} $(RM) $(ALL_PROGRAMS) $(BUILT_INS) perf$X $(RM) $(TEST_PROGRAMS) $(RM) *.spec *.pyc *.pyo */*.pyc */*.pyo $(OUTPUT)common-cmds.h TAGS tags cscope* - $(RM) -r autom4te.cache - $(RM) config.log config.mak.autogen config.mak.append config.status config.cache $(RM) -r $(PERF_TARNAME) .doc-tmp-dir $(RM) $(PERF_TARNAME).tar.gz perf-core_$(PERF_VERSION)-*.tar.gz $(RM) $(htmldocs).tar.gz $(manpages).tar.gz $(MAKE) -C Documentation/ clean $(RM) $(OUTPUT)PERF-VERSION-FILE $(OUTPUT)PERF-CFLAGS $(OUTPUT)PERF-BUILD-OPTIONS + @python util/setup.py clean --build-lib='$(OUTPUT)python' \ + --build-temp='$(OUTPUT)python/temp' .PHONY: all install clean strip .PHONY: shell_compatibility_test please_set_SHELL_PATH_to_a_more_modern_shell diff --git a/tools/perf/bench/sched-pipe.c b/tools/perf/bench/sched-pipe.c index d9ab3ce446ac..0c7454f8b8a9 100644 --- a/tools/perf/bench/sched-pipe.c +++ b/tools/perf/bench/sched-pipe.c @@ -55,7 +55,7 @@ int bench_sched_pipe(int argc, const char **argv, * discarding returned value of read(), write() * causes error in building environment for perf */ - int ret, wait_stat; + int __used ret, wait_stat; pid_t pid, retpid; argc = parse_options(argc, argv, options, diff --git a/tools/perf/builtin-annotate.c b/tools/perf/builtin-annotate.c index 8879463807e4..427182953fd7 100644 --- a/tools/perf/builtin-annotate.c +++ b/tools/perf/builtin-annotate.c @@ -9,6 +9,7 @@ #include "util/util.h" +#include "util/util.h" #include "util/color.h" #include <linux/list.h> #include "util/cache.h" @@ -18,6 +19,7 @@ #include "perf.h" #include "util/debug.h" +#include "util/annotate.h" #include "util/event.h" #include "util/parse-options.h" #include "util/parse-events.h" @@ -55,15 +57,29 @@ static int hists__add_entry(struct hists *self, struct addr_location *al) if (he == NULL) return -ENOMEM; - return hist_entry__inc_addr_samples(he, al->addr); + if (he->ms.sym != NULL) { + /* + * All aggregated on the first sym_hist. + */ + struct annotation *notes = symbol__annotation(he->ms.sym); + if (notes->src == NULL && + symbol__alloc_hist(he->ms.sym, 1) < 0) + return -ENOMEM; + + return hist_entry__inc_addr_samples(he, 0, al->addr); + } + + return 0; } -static int process_sample_event(event_t *event, struct sample_data *sample, +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct addr_location al; - if (event__preprocess_sample(event, session, &al, sample, NULL) < 0) { + if (perf_event__preprocess_sample(event, session, &al, sample, + symbol__annotate_init) < 0) { pr_warning("problem processing %d event, skipping it.\n", event->header.type); return -1; @@ -78,245 +94,10 @@ static int process_sample_event(event_t *event, struct sample_data *sample, return 0; } -static int objdump_line__print(struct objdump_line *self, - struct list_head *head, - struct hist_entry *he, u64 len) -{ - struct symbol *sym = he->ms.sym; - static const char *prev_line; - static const char *prev_color; - - if (self->offset != -1) { - const char *path = NULL; - unsigned int hits = 0; - double percent = 0.0; - const char *color; - struct sym_priv *priv = symbol__priv(sym); - struct sym_ext *sym_ext = priv->ext; - struct sym_hist *h = priv->hist; - s64 offset = self->offset; - struct objdump_line *next = objdump__get_next_ip_line(head, self); - - while (offset < (s64)len && - (next == NULL || offset < next->offset)) { - if (sym_ext) { - if (path == NULL) - path = sym_ext[offset].path; - percent += sym_ext[offset].percent; - } else - hits += h->ip[offset]; - - ++offset; - } - - if (sym_ext == NULL && h->sum) - percent = 100.0 * hits / h->sum; - - color = get_percent_color(percent); - - /* - * Also color the filename and line if needed, with - * the same color than the percentage. Don't print it - * twice for close colored ip with the same filename:line - */ - if (path) { - if (!prev_line || strcmp(prev_line, path) - || color != prev_color) { - color_fprintf(stdout, color, " %s", path); - prev_line = path; - prev_color = color; - } - } - - color_fprintf(stdout, color, " %7.2f", percent); - printf(" : "); - color_fprintf(stdout, PERF_COLOR_BLUE, "%s\n", self->line); - } else { - if (!*self->line) - printf(" :\n"); - else - printf(" : %s\n", self->line); - } - - return 0; -} - -static struct rb_root root_sym_ext; - -static void insert_source_line(struct sym_ext *sym_ext) -{ - struct sym_ext *iter; - struct rb_node **p = &root_sym_ext.rb_node; - struct rb_node *parent = NULL; - - while (*p != NULL) { - parent = *p; - iter = rb_entry(parent, struct sym_ext, node); - - if (sym_ext->percent > iter->percent) - p = &(*p)->rb_left; - else - p = &(*p)->rb_right; - } - - rb_link_node(&sym_ext->node, parent, p); - rb_insert_color(&sym_ext->node, &root_sym_ext); -} - -static void free_source_line(struct hist_entry *he, int len) -{ - struct sym_priv *priv = symbol__priv(he->ms.sym); - struct sym_ext *sym_ext = priv->ext; - int i; - - if (!sym_ext) - return; - - for (i = 0; i < len; i++) - free(sym_ext[i].path); - free(sym_ext); - - priv->ext = NULL; - root_sym_ext = RB_ROOT; -} - -/* Get the filename:line for the colored entries */ -static void -get_source_line(struct hist_entry *he, int len, const char *filename) -{ - struct symbol *sym = he->ms.sym; - u64 start; - int i; - char cmd[PATH_MAX * 2]; - struct sym_ext *sym_ext; - struct sym_priv *priv = symbol__priv(sym); - struct sym_hist *h = priv->hist; - - if (!h->sum) - return; - - sym_ext = priv->ext = calloc(len, sizeof(struct sym_ext)); - if (!priv->ext) - return; - - start = he->ms.map->unmap_ip(he->ms.map, sym->start); - - for (i = 0; i < len; i++) { - char *path = NULL; - size_t line_len; - u64 offset; - FILE *fp; - - sym_ext[i].percent = 100.0 * h->ip[i] / h->sum; - if (sym_ext[i].percent <= 0.5) - continue; - - offset = start + i; - sprintf(cmd, "addr2line -e %s %016" PRIx64, filename, offset); - fp = popen(cmd, "r"); - if (!fp) - continue; - - if (getline(&path, &line_len, fp) < 0 || !line_len) - goto next; - - sym_ext[i].path = malloc(sizeof(char) * line_len + 1); - if (!sym_ext[i].path) - goto next; - - strcpy(sym_ext[i].path, path); - insert_source_line(&sym_ext[i]); - - next: - pclose(fp); - } -} - -static void print_summary(const char *filename) +static int hist_entry__tty_annotate(struct hist_entry *he, int evidx) { - struct sym_ext *sym_ext; - struct rb_node *node; - - printf("\nSorted summary for file %s\n", filename); - printf("----------------------------------------------\n\n"); - - if (RB_EMPTY_ROOT(&root_sym_ext)) { - printf(" Nothing higher than %1.1f%%\n", MIN_GREEN); - return; - } - - node = rb_first(&root_sym_ext); - while (node) { - double percent; - const char *color; - char *path; - - sym_ext = rb_entry(node, struct sym_ext, node); - percent = sym_ext->percent; - color = get_percent_color(percent); - path = sym_ext->path; - - color_fprintf(stdout, color, " %7.2f %s", percent, path); - node = rb_next(node); - } -} - -static void hist_entry__print_hits(struct hist_entry *self) -{ - struct symbol *sym = self->ms.sym; - struct sym_priv *priv = symbol__priv(sym); - struct sym_hist *h = priv->hist; - u64 len = sym->end - sym->start, offset; - - for (offset = 0; offset < len; ++offset) - if (h->ip[offset] != 0) - printf("%*" PRIx64 ": %" PRIu64 "\n", BITS_PER_LONG / 2, - sym->start + offset, h->ip[offset]); - printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->sum", h->sum); -} - -static int hist_entry__tty_annotate(struct hist_entry *he) -{ - struct map *map = he->ms.map; - struct dso *dso = map->dso; - struct symbol *sym = he->ms.sym; - const char *filename = dso->long_name, *d_filename; - u64 len; - LIST_HEAD(head); - struct objdump_line *pos, *n; - - if (hist_entry__annotate(he, &head, 0) < 0) - return -1; - - if (full_paths) - d_filename = filename; - else - d_filename = basename(filename); - - len = sym->end - sym->start; - - if (print_line) { - get_source_line(he, len, filename); - print_summary(filename); - } - - printf("\n\n------------------------------------------------\n"); - printf(" Percent | Source code & Disassembly of %s\n", d_filename); - printf("------------------------------------------------\n"); - - if (verbose) - hist_entry__print_hits(he); - - list_for_each_entry_safe(pos, n, &head, node) { - objdump_line__print(pos, &head, he, len); - list_del(&pos->node); - objdump_line__free(pos); - } - - if (print_line) - free_source_line(he, len); - - return 0; + return symbol__tty_annotate(he->ms.sym, he->ms.map, evidx, + print_line, full_paths, 0, 0); } static void hists__find_annotations(struct hists *self) @@ -326,13 +107,13 @@ static void hists__find_annotations(struct hists *self) while (nd) { struct hist_entry *he = rb_entry(nd, struct hist_entry, rb_node); - struct sym_priv *priv; + struct annotation *notes; if (he->ms.sym == NULL || he->ms.map->dso->annotate_warned) goto find_next; - priv = symbol__priv(he->ms.sym); - if (priv->hist == NULL) { + notes = symbol__annotation(he->ms.sym); + if (notes->src == NULL) { find_next: if (key == KEY_LEFT) nd = rb_prev(nd); @@ -342,7 +123,8 @@ find_next: } if (use_browser > 0) { - key = hist_entry__tui_annotate(he); + /* For now all is aggregated on the first */ + key = hist_entry__tui_annotate(he, 0); switch (key) { case KEY_RIGHT: next = rb_next(nd); @@ -357,24 +139,25 @@ find_next: if (next != NULL) nd = next; } else { - hist_entry__tty_annotate(he); + /* For now all is aggregated on the first */ + hist_entry__tty_annotate(he, 0); nd = rb_next(nd); /* * Since we have a hist_entry per IP for the same - * symbol, free he->ms.sym->hist to signal we already + * symbol, free he->ms.sym->src to signal we already * processed this symbol. */ - free(priv->hist); - priv->hist = NULL; + free(notes->src); + notes->src = NULL; } } } static struct perf_event_ops event_ops = { .sample = process_sample_event, - .mmap = event__process_mmap, - .comm = event__process_comm, - .fork = event__process_task, + .mmap = perf_event__process_mmap, + .comm = perf_event__process_comm, + .fork = perf_event__process_task, .ordered_samples = true, .ordering_requires_timestamps = true, }; @@ -451,9 +234,9 @@ int cmd_annotate(int argc, const char **argv, const char *prefix __used) else if (use_tui) use_browser = 1; - setup_browser(); + setup_browser(true); - symbol_conf.priv_size = sizeof(struct sym_priv); + symbol_conf.priv_size = sizeof(struct annotation); symbol_conf.try_vmlinux_path = true; if (symbol__init() < 0) diff --git a/tools/perf/builtin-diff.c b/tools/perf/builtin-diff.c index 3153e492dbcc..6b7d91160ecb 100644 --- a/tools/perf/builtin-diff.c +++ b/tools/perf/builtin-diff.c @@ -30,13 +30,13 @@ static int hists__add_entry(struct hists *self, return -ENOMEM; } -static int diff__process_sample_event(event_t *event, - struct sample_data *sample, +static int diff__process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct addr_location al; - if (event__preprocess_sample(event, session, &al, sample, NULL) < 0) { + if (perf_event__preprocess_sample(event, session, &al, sample, NULL) < 0) { pr_warning("problem processing %d event, skipping it.\n", event->header.type); return -1; @@ -56,11 +56,11 @@ static int diff__process_sample_event(event_t *event, static struct perf_event_ops event_ops = { .sample = diff__process_sample_event, - .mmap = event__process_mmap, - .comm = event__process_comm, - .exit = event__process_task, - .fork = event__process_task, - .lost = event__process_lost, + .mmap = perf_event__process_mmap, + .comm = perf_event__process_comm, + .exit = perf_event__process_task, + .fork = perf_event__process_task, + .lost = perf_event__process_lost, .ordered_samples = true, .ordering_requires_timestamps = true, }; diff --git a/tools/perf/builtin-inject.c b/tools/perf/builtin-inject.c index 0c78ffa7bf67..e29f04ed3396 100644 --- a/tools/perf/builtin-inject.c +++ b/tools/perf/builtin-inject.c @@ -16,8 +16,8 @@ static char const *input_name = "-"; static bool inject_build_ids; -static int event__repipe_synth(event_t *event, - struct perf_session *session __used) +static int perf_event__repipe_synth(union perf_event *event, + struct perf_session *session __used) { uint32_t size; void *buf = event; @@ -36,41 +36,44 @@ static int event__repipe_synth(event_t *event, return 0; } -static int event__repipe(event_t *event, struct sample_data *sample __used, - struct perf_session *session) +static int perf_event__repipe(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { - return event__repipe_synth(event, session); + return perf_event__repipe_synth(event, session); } -static int event__repipe_mmap(event_t *self, struct sample_data *sample, - struct perf_session *session) +static int perf_event__repipe_mmap(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) { int err; - err = event__process_mmap(self, sample, session); - event__repipe(self, sample, session); + err = perf_event__process_mmap(event, sample, session); + perf_event__repipe(event, sample, session); return err; } -static int event__repipe_task(event_t *self, struct sample_data *sample, - struct perf_session *session) +static int perf_event__repipe_task(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) { int err; - err = event__process_task(self, sample, session); - event__repipe(self, sample, session); + err = perf_event__process_task(event, sample, session); + perf_event__repipe(event, sample, session); return err; } -static int event__repipe_tracing_data(event_t *self, - struct perf_session *session) +static int perf_event__repipe_tracing_data(union perf_event *event, + struct perf_session *session) { int err; - event__repipe_synth(self, session); - err = event__process_tracing_data(self, session); + perf_event__repipe_synth(event, session); + err = perf_event__process_tracing_data(event, session); return err; } @@ -109,8 +112,8 @@ static int dso__inject_build_id(struct dso *self, struct perf_session *session) if (self->kernel) misc = PERF_RECORD_MISC_KERNEL; - err = event__synthesize_build_id(self, misc, event__repipe, - machine, session); + err = perf_event__synthesize_build_id(self, misc, perf_event__repipe, + machine, session); if (err) { pr_err("Can't synthesize build_id event for %s\n", self->long_name); return -1; @@ -119,8 +122,9 @@ static int dso__inject_build_id(struct dso *self, struct perf_session *session) return 0; } -static int event__inject_buildid(event_t *event, struct sample_data *sample, - struct perf_session *session) +static int perf_event__inject_buildid(union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) { struct addr_location al; struct thread *thread; @@ -155,24 +159,24 @@ static int event__inject_buildid(event_t *event, struct sample_data *sample, } repipe: - event__repipe(event, sample, session); + perf_event__repipe(event, sample, session); return 0; } struct perf_event_ops inject_ops = { - .sample = event__repipe, - .mmap = event__repipe, - .comm = event__repipe, - .fork = event__repipe, - .exit = event__repipe, - .lost = event__repipe, - .read = event__repipe, - .throttle = event__repipe, - .unthrottle = event__repipe, - .attr = event__repipe_synth, - .event_type = event__repipe_synth, - .tracing_data = event__repipe_synth, - .build_id = event__repipe_synth, + .sample = perf_event__repipe, + .mmap = perf_event__repipe, + .comm = perf_event__repipe, + .fork = perf_event__repipe, + .exit = perf_event__repipe, + .lost = perf_event__repipe, + .read = perf_event__repipe, + .throttle = perf_event__repipe, + .unthrottle = perf_event__repipe, + .attr = perf_event__repipe_synth, + .event_type = perf_event__repipe_synth, + .tracing_data = perf_event__repipe_synth, + .build_id = perf_event__repipe_synth, }; extern volatile int session_done; @@ -190,10 +194,10 @@ static int __cmd_inject(void) signal(SIGINT, sig_handler); if (inject_build_ids) { - inject_ops.sample = event__inject_buildid; - inject_ops.mmap = event__repipe_mmap; - inject_ops.fork = event__repipe_task; - inject_ops.tracing_data = event__repipe_tracing_data; + inject_ops.sample = perf_event__inject_buildid; + inject_ops.mmap = perf_event__repipe_mmap; + inject_ops.fork = perf_event__repipe_task; + inject_ops.tracing_data = perf_event__repipe_tracing_data; } session = perf_session__new(input_name, O_RDONLY, false, true, &inject_ops); diff --git a/tools/perf/builtin-kmem.c b/tools/perf/builtin-kmem.c index d97256d65980..7f618f4e7b79 100644 --- a/tools/perf/builtin-kmem.c +++ b/tools/perf/builtin-kmem.c @@ -275,9 +275,8 @@ static void process_free_event(void *data, s_alloc->alloc_cpu = -1; } -static void -process_raw_event(event_t *raw_event __used, void *data, - int cpu, u64 timestamp, struct thread *thread) +static void process_raw_event(union perf_event *raw_event __used, void *data, + int cpu, u64 timestamp, struct thread *thread) { struct event *event; int type; @@ -304,7 +303,8 @@ process_raw_event(event_t *raw_event __used, void *data, } } -static int process_sample_event(event_t *event, struct sample_data *sample, +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct thread *thread = perf_session__findnew(session, event->ip.pid); @@ -325,7 +325,7 @@ static int process_sample_event(event_t *event, struct sample_data *sample, static struct perf_event_ops event_ops = { .sample = process_sample_event, - .comm = event__process_comm, + .comm = perf_event__process_comm, .ordered_samples = true, }; diff --git a/tools/perf/builtin-lock.c b/tools/perf/builtin-lock.c index 2b36defc5d73..e00d93847c44 100644 --- a/tools/perf/builtin-lock.c +++ b/tools/perf/builtin-lock.c @@ -834,14 +834,14 @@ static void dump_info(void) die("Unknown type of information\n"); } -static int process_sample_event(event_t *self, struct sample_data *sample, +static int process_sample_event(union perf_event *event, struct perf_sample *sample, struct perf_session *s) { struct thread *thread = perf_session__findnew(s, sample->tid); if (thread == NULL) { pr_debug("problem processing %d event, skipping it.\n", - self->header.type); + event->header.type); return -1; } @@ -852,7 +852,7 @@ static int process_sample_event(event_t *self, struct sample_data *sample, static struct perf_event_ops eops = { .sample = process_sample_event, - .comm = event__process_comm, + .comm = perf_event__process_comm, .ordered_samples = true, }; diff --git a/tools/perf/builtin-probe.c b/tools/perf/builtin-probe.c index add163c9f0e7..2c0e64d0b4aa 100644 --- a/tools/perf/builtin-probe.c +++ b/tools/perf/builtin-probe.c @@ -36,6 +36,7 @@ #include "builtin.h" #include "util/util.h" #include "util/strlist.h" +#include "util/strfilter.h" #include "util/symbol.h" #include "util/debug.h" #include "util/debugfs.h" @@ -43,6 +44,8 @@ #include "util/probe-finder.h" #include "util/probe-event.h" +#define DEFAULT_VAR_FILTER "!__k???tab_* & !__crc_*" +#define DEFAULT_FUNC_FILTER "!_*" #define MAX_PATH_LEN 256 /* Session management structure */ @@ -52,6 +55,7 @@ static struct { bool show_lines; bool show_vars; bool show_ext_vars; + bool show_funcs; bool mod_events; int nevents; struct perf_probe_event events[MAX_PROBES]; @@ -59,6 +63,7 @@ static struct { struct line_range line_range; const char *target_module; int max_probe_points; + struct strfilter *filter; } params; /* Parse an event definition. Note that any error must die. */ @@ -157,6 +162,27 @@ static int opt_show_vars(const struct option *opt __used, } #endif +static int opt_set_filter(const struct option *opt __used, + const char *str, int unset __used) +{ + const char *err; + + if (str) { + pr_debug2("Set filter: %s\n", str); + if (params.filter) + strfilter__delete(params.filter); + params.filter = strfilter__new(str, &err); + if (!params.filter) { + pr_err("Filter parse error at %td.\n", err - str + 1); + pr_err("Source: \"%s\"\n", str); + pr_err(" %*c\n", (int)(err - str + 1), '^'); + return -EINVAL; + } + } + + return 0; +} + static const char * const probe_usage[] = { "perf probe [<options>] 'PROBEDEF' ['PROBEDEF' ...]", "perf probe [<options>] --add 'PROBEDEF' [--add 'PROBEDEF' ...]", @@ -221,6 +247,13 @@ static const struct option options[] = { OPT__DRY_RUN(&probe_event_dry_run), OPT_INTEGER('\0', "max-probes", ¶ms.max_probe_points, "Set how many probe points can be found for a probe."), + OPT_BOOLEAN('F', "funcs", ¶ms.show_funcs, + "Show potential probe-able functions."), + OPT_CALLBACK('\0', "filter", NULL, + "[!]FILTER", "Set a filter (with --vars/funcs only)\n" + "\t\t\t(default: \"" DEFAULT_VAR_FILTER "\" for --vars,\n" + "\t\t\t \"" DEFAULT_FUNC_FILTER "\" for --funcs)", + opt_set_filter), OPT_END() }; @@ -246,7 +279,7 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) params.max_probe_points = MAX_PROBES; if ((!params.nevents && !params.dellist && !params.list_events && - !params.show_lines)) + !params.show_lines && !params.show_funcs)) usage_with_options(probe_usage, options); /* @@ -267,12 +300,41 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) pr_err(" Error: Don't use --list with --vars.\n"); usage_with_options(probe_usage, options); } + if (params.show_funcs) { + pr_err(" Error: Don't use --list with --funcs.\n"); + usage_with_options(probe_usage, options); + } ret = show_perf_probe_events(); if (ret < 0) pr_err(" Error: Failed to show event list. (%d)\n", ret); return ret; } + if (params.show_funcs) { + if (params.nevents != 0 || params.dellist) { + pr_err(" Error: Don't use --funcs with" + " --add/--del.\n"); + usage_with_options(probe_usage, options); + } + if (params.show_lines) { + pr_err(" Error: Don't use --funcs with --line.\n"); + usage_with_options(probe_usage, options); + } + if (params.show_vars) { + pr_err(" Error: Don't use --funcs with --vars.\n"); + usage_with_options(probe_usage, options); + } + if (!params.filter) + params.filter = strfilter__new(DEFAULT_FUNC_FILTER, + NULL); + ret = show_available_funcs(params.target_module, + params.filter); + strfilter__delete(params.filter); + if (ret < 0) + pr_err(" Error: Failed to show functions." + " (%d)\n", ret); + return ret; + } #ifdef DWARF_SUPPORT if (params.show_lines) { @@ -297,10 +359,16 @@ int cmd_probe(int argc, const char **argv, const char *prefix __used) " --add/--del.\n"); usage_with_options(probe_usage, options); } + if (!params.filter) + params.filter = strfilter__new(DEFAULT_VAR_FILTER, + NULL); + ret = show_available_vars(params.events, params.nevents, params.max_probe_points, params.target_module, + params.filter, params.show_ext_vars); + strfilter__delete(params.filter); if (ret < 0) pr_err(" Error: Failed to show vars. (%d)\n", ret); return ret; diff --git a/tools/perf/builtin-record.c b/tools/perf/builtin-record.c index 60cac6f92e8b..12e0e41696d9 100644 --- a/tools/perf/builtin-record.c +++ b/tools/perf/builtin-record.c @@ -18,17 +18,20 @@ #include "util/header.h" #include "util/event.h" +#include "util/evlist.h" #include "util/evsel.h" #include "util/debug.h" #include "util/session.h" #include "util/symbol.h" #include "util/cpumap.h" +#include "util/thread_map.h" #include <unistd.h> #include <sched.h> #include <sys/mman.h> #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) +#define SID(e, x, y) xyarray__entry(e->id, x, y) enum write_mode_t { WRITE_FORCE, @@ -39,14 +42,13 @@ static u64 user_interval = ULLONG_MAX; static u64 default_interval = 0; static u64 sample_type; -static struct cpu_map *cpus; static unsigned int page_size; static unsigned int mmap_pages = 128; static unsigned int user_freq = UINT_MAX; static int freq = 1000; static int output; static int pipe_output = 0; -static const char *output_name = "perf.data"; +static const char *output_name = NULL; static int group = 0; static int realtime_prio = 0; static bool nodelay = false; @@ -55,7 +57,6 @@ static bool sample_id_all_avail = true; static bool system_wide = false; static pid_t target_pid = -1; static pid_t target_tid = -1; -static struct thread_map *threads; static pid_t child_pid = -1; static bool no_inherit = false; static enum write_mode_t write_mode = WRITE_FORCE; @@ -66,51 +67,17 @@ static bool sample_address = false; static bool sample_time = false; static bool no_buildid = false; static bool no_buildid_cache = false; +static struct perf_evlist *evsel_list; static long samples = 0; static u64 bytes_written = 0; -static struct pollfd *event_array; - -static int nr_poll = 0; -static int nr_cpu = 0; - static int file_new = 1; static off_t post_processing_offset; static struct perf_session *session; static const char *cpu_list; -struct mmap_data { - void *base; - unsigned int mask; - unsigned int prev; -}; - -static struct mmap_data mmap_array[MAX_NR_CPUS]; - -static unsigned long mmap_read_head(struct mmap_data *md) -{ - struct perf_event_mmap_page *pc = md->base; - long head; - - head = pc->data_head; - rmb(); - - return head; -} - -static void mmap_write_tail(struct mmap_data *md, unsigned long tail) -{ - struct perf_event_mmap_page *pc = md->base; - - /* - * ensure all reads are done before we write the tail out. - */ - /* mb(); */ - pc->data_tail = tail; -} - static void advance_output(size_t size) { bytes_written += size; @@ -131,42 +98,26 @@ static void write_output(void *buf, size_t size) } } -static int process_synthesized_event(event_t *event, - struct sample_data *sample __used, +static int process_synthesized_event(union perf_event *event, + struct perf_sample *sample __used, struct perf_session *self __used) { write_output(event, event->header.size); return 0; } -static void mmap_read(struct mmap_data *md) +static void mmap_read(struct perf_mmap *md) { - unsigned int head = mmap_read_head(md); + unsigned int head = perf_mmap__read_head(md); unsigned int old = md->prev; unsigned char *data = md->base + page_size; unsigned long size; void *buf; - int diff; - /* - * If we're further behind than half the buffer, there's a chance - * the writer will bite our tail and mess up the samples under us. - * - * If we somehow ended up ahead of the head, we got messed up. - * - * In either case, truncate and restart at head. - */ - diff = head - old; - if (diff < 0) { - fprintf(stderr, "WARNING: failed to keep up with mmap data\n"); - /* - * head points to a known good entry, start there. - */ - old = head; - } + if (old == head) + return; - if (old != head) - samples++; + samples++; size = head - old; @@ -185,7 +136,7 @@ static void mmap_read(struct mmap_data *md) write_output(buf, size); md->prev = old; - mmap_write_tail(md, old); + perf_mmap__write_tail(md, old); } static volatile int done = 0; @@ -209,8 +160,6 @@ static void sig_atexit(void) kill(getpid(), signr); } -static int group_fd; - static struct perf_header_attr *get_header_attr(struct perf_event_attr *a, int nr) { struct perf_header_attr *h_attr; @@ -234,28 +183,47 @@ static void create_counter(struct perf_evsel *evsel, int cpu) char *filter = evsel->filter; struct perf_event_attr *attr = &evsel->attr; struct perf_header_attr *h_attr; - int track = !evsel->idx; /* only the first counter needs these */ + struct perf_sample_id *sid; int thread_index; int ret; - struct { - u64 count; - u64 time_enabled; - u64 time_running; - u64 id; - } read_data; - /* - * Check if parse_single_tracepoint_event has already asked for - * PERF_SAMPLE_TIME. - * - * XXX this is kludgy but short term fix for problems introduced by - * eac23d1c that broke 'perf script' by having different sample_types - * when using multiple tracepoint events when we use a perf binary - * that tries to use sample_id_all on an older kernel. - * - * We need to move counter creation to perf_session, support - * different sample_types, etc. - */ - bool time_needed = attr->sample_type & PERF_SAMPLE_TIME; + + for (thread_index = 0; thread_index < evsel_list->threads->nr; thread_index++) { + h_attr = get_header_attr(attr, evsel->idx); + if (h_attr == NULL) + die("nomem\n"); + + if (!file_new) { + if (memcmp(&h_attr->attr, attr, sizeof(*attr))) { + fprintf(stderr, "incompatible append\n"); + exit(-1); + } + } + + sid = SID(evsel, cpu, thread_index); + if (perf_header_attr__add_id(h_attr, sid->id) < 0) { + pr_warning("Not enough memory to add id\n"); + exit(-1); + } + + if (filter != NULL) { + ret = ioctl(FD(evsel, cpu, thread_index), + PERF_EVENT_IOC_SET_FILTER, filter); + if (ret) { + error("failed to set filter with %d (%s)\n", errno, + strerror(errno)); + exit(-1); + } + } + } + + if (!sample_type) + sample_type = attr->sample_type; +} + +static void config_attr(struct perf_evsel *evsel, struct perf_evlist *evlist) +{ + struct perf_event_attr *attr = &evsel->attr; + int track = !evsel->idx; /* only the first counter needs these */ attr->read_format = PERF_FORMAT_TOTAL_TIME_ENABLED | PERF_FORMAT_TOTAL_TIME_RUNNING | @@ -263,7 +231,7 @@ static void create_counter(struct perf_evsel *evsel, int cpu) attr->sample_type |= PERF_SAMPLE_IP | PERF_SAMPLE_TID; - if (nr_counters > 1) + if (evlist->nr_entries > 1) attr->sample_type |= PERF_SAMPLE_ID; /* @@ -315,19 +283,40 @@ static void create_counter(struct perf_evsel *evsel, int cpu) attr->mmap = track; attr->comm = track; - attr->inherit = !no_inherit; + if (target_pid == -1 && target_tid == -1 && !system_wide) { attr->disabled = 1; attr->enable_on_exec = 1; } -retry_sample_id: - attr->sample_id_all = sample_id_all_avail ? 1 : 0; +} - for (thread_index = 0; thread_index < threads->nr; thread_index++) { -try_again: - FD(evsel, nr_cpu, thread_index) = sys_perf_event_open(attr, threads->map[thread_index], cpu, group_fd, 0); +static void open_counters(struct perf_evlist *evlist) +{ + struct perf_evsel *pos; + int cpu; + + list_for_each_entry(pos, &evlist->entries, node) { + struct perf_event_attr *attr = &pos->attr; + /* + * Check if parse_single_tracepoint_event has already asked for + * PERF_SAMPLE_TIME. + * + * XXX this is kludgy but short term fix for problems introduced by + * eac23d1c that broke 'perf script' by having different sample_types + * when using multiple tracepoint events when we use a perf binary + * that tries to use sample_id_all on an older kernel. + * + * We need to move counter creation to perf_session, support + * different sample_types, etc. + */ + bool time_needed = attr->sample_type & PERF_SAMPLE_TIME; - if (FD(evsel, nr_cpu, thread_index) < 0) { + config_attr(pos, evlist); +retry_sample_id: + attr->sample_id_all = sample_id_all_avail ? 1 : 0; +try_again: + if (perf_evsel__open(pos, evlist->cpus, evlist->threads, group, + !no_inherit) < 0) { int err = errno; if (err == EPERM || err == EACCES) @@ -364,7 +353,7 @@ try_again: } printf("\n"); error("sys_perf_event_open() syscall returned with %d (%s). /bin/dmesg may provide additional information.\n", - FD(evsel, nr_cpu, thread_index), strerror(err)); + err, strerror(err)); #if defined(__i386__) || defined(__x86_64__) if (attr->type == PERF_TYPE_HARDWARE && err == EOPNOTSUPP) @@ -375,90 +364,16 @@ try_again: #endif die("No CONFIG_PERF_EVENTS=y kernel support configured?\n"); - exit(-1); - } - - h_attr = get_header_attr(attr, evsel->idx); - if (h_attr == NULL) - die("nomem\n"); - - if (!file_new) { - if (memcmp(&h_attr->attr, attr, sizeof(*attr))) { - fprintf(stderr, "incompatible append\n"); - exit(-1); - } - } - - if (read(FD(evsel, nr_cpu, thread_index), &read_data, sizeof(read_data)) == -1) { - perror("Unable to read perf file descriptor"); - exit(-1); - } - - if (perf_header_attr__add_id(h_attr, read_data.id) < 0) { - pr_warning("Not enough memory to add id\n"); - exit(-1); - } - - assert(FD(evsel, nr_cpu, thread_index) >= 0); - fcntl(FD(evsel, nr_cpu, thread_index), F_SETFL, O_NONBLOCK); - - /* - * First counter acts as the group leader: - */ - if (group && group_fd == -1) - group_fd = FD(evsel, nr_cpu, thread_index); - - if (evsel->idx || thread_index) { - struct perf_evsel *first; - first = list_entry(evsel_list.next, struct perf_evsel, node); - ret = ioctl(FD(evsel, nr_cpu, thread_index), - PERF_EVENT_IOC_SET_OUTPUT, - FD(first, nr_cpu, 0)); - if (ret) { - error("failed to set output: %d (%s)\n", errno, - strerror(errno)); - exit(-1); - } - } else { - mmap_array[nr_cpu].prev = 0; - mmap_array[nr_cpu].mask = mmap_pages*page_size - 1; - mmap_array[nr_cpu].base = mmap(NULL, (mmap_pages+1)*page_size, - PROT_READ | PROT_WRITE, MAP_SHARED, FD(evsel, nr_cpu, thread_index), 0); - if (mmap_array[nr_cpu].base == MAP_FAILED) { - error("failed to mmap with %d (%s)\n", errno, strerror(errno)); - exit(-1); - } - - event_array[nr_poll].fd = FD(evsel, nr_cpu, thread_index); - event_array[nr_poll].events = POLLIN; - nr_poll++; - } - - if (filter != NULL) { - ret = ioctl(FD(evsel, nr_cpu, thread_index), - PERF_EVENT_IOC_SET_FILTER, filter); - if (ret) { - error("failed to set filter with %d (%s)\n", errno, - strerror(errno)); - exit(-1); - } } } - if (!sample_type) - sample_type = attr->sample_type; -} - -static void open_counters(int cpu) -{ - struct perf_evsel *pos; + if (perf_evlist__mmap(evlist, mmap_pages, false) < 0) + die("failed to mmap with %d (%s)\n", errno, strerror(errno)); - group_fd = -1; - - list_for_each_entry(pos, &evsel_list, node) - create_counter(pos, cpu); - - nr_cpu++; + for (cpu = 0; cpu < evsel_list->cpus->nr; ++cpu) { + list_for_each_entry(pos, &evlist->entries, node) + create_counter(pos, cpu); + } } static int process_buildids(void) @@ -481,14 +396,14 @@ static void atexit_header(void) if (!no_buildid) process_buildids(); - perf_header__write(&session->header, output, true); + perf_header__write(&session->header, evsel_list, output, true); perf_session__delete(session); - perf_evsel_list__delete(); + perf_evlist__delete(evsel_list); symbol__exit(); } } -static void event__synthesize_guest_os(struct machine *machine, void *data) +static void perf_event__synthesize_guest_os(struct machine *machine, void *data) { int err; struct perf_session *psession = data; @@ -504,8 +419,8 @@ static void event__synthesize_guest_os(struct machine *machine, void *data) *method is used to avoid symbol missing when the first addr is *in module instead of in guest kernel. */ - err = event__synthesize_modules(process_synthesized_event, - psession, machine); + err = perf_event__synthesize_modules(process_synthesized_event, + psession, machine); if (err < 0) pr_err("Couldn't record guest kernel [%d]'s reference" " relocation symbol.\n", machine->pid); @@ -514,11 +429,12 @@ static void event__synthesize_guest_os(struct machine *machine, void *data) * We use _stext for guest kernel because guest kernel's /proc/kallsyms * have no _text sometimes. */ - err = event__synthesize_kernel_mmap(process_synthesized_event, - psession, machine, "_text"); + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + psession, machine, "_text"); if (err < 0) - err = event__synthesize_kernel_mmap(process_synthesized_event, - psession, machine, "_stext"); + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + psession, machine, + "_stext"); if (err < 0) pr_err("Couldn't record guest kernel [%d]'s reference" " relocation symbol.\n", machine->pid); @@ -533,9 +449,9 @@ static void mmap_read_all(void) { int i; - for (i = 0; i < nr_cpu; i++) { - if (mmap_array[i].base) - mmap_read(&mmap_array[i]); + for (i = 0; i < evsel_list->cpus->nr; i++) { + if (evsel_list->mmap[i].base) + mmap_read(&evsel_list->mmap[i]); } if (perf_header__has_feat(&session->header, HEADER_TRACE_INFO)) @@ -566,18 +482,26 @@ static int __cmd_record(int argc, const char **argv) exit(-1); } - if (!strcmp(output_name, "-")) - pipe_output = 1; - else if (!stat(output_name, &st) && st.st_size) { - if (write_mode == WRITE_FORCE) { - char oldname[PATH_MAX]; - snprintf(oldname, sizeof(oldname), "%s.old", - output_name); - unlink(oldname); - rename(output_name, oldname); + if (!output_name) { + if (!fstat(STDOUT_FILENO, &st) && S_ISFIFO(st.st_mode)) + pipe_output = 1; + else + output_name = "perf.data"; + } + if (output_name) { + if (!strcmp(output_name, "-")) + pipe_output = 1; + else if (!stat(output_name, &st) && st.st_size) { + if (write_mode == WRITE_FORCE) { + char oldname[PATH_MAX]; + snprintf(oldname, sizeof(oldname), "%s.old", + output_name); + unlink(oldname); + rename(output_name, oldname); + } + } else if (write_mode == WRITE_APPEND) { + write_mode = WRITE_FORCE; } - } else if (write_mode == WRITE_APPEND) { - write_mode = WRITE_FORCE; } flags = O_CREAT|O_RDWR; @@ -611,7 +535,7 @@ static int __cmd_record(int argc, const char **argv) goto out_delete_session; } - if (have_tracepoints(&evsel_list)) + if (have_tracepoints(&evsel_list->entries)) perf_header__set_feat(&session->header, HEADER_TRACE_INFO); /* @@ -659,7 +583,7 @@ static int __cmd_record(int argc, const char **argv) } if (!system_wide && target_tid == -1 && target_pid == -1) - threads->map[0] = child_pid; + evsel_list->threads->map[0] = child_pid; close(child_ready_pipe[1]); close(go_pipe[0]); @@ -673,12 +597,7 @@ static int __cmd_record(int argc, const char **argv) close(child_ready_pipe[0]); } - if (!system_wide && no_inherit && !cpu_list) { - open_counters(-1); - } else { - for (i = 0; i < cpus->nr; i++) - open_counters(cpus->map[i]); - } + open_counters(evsel_list); perf_session__set_sample_type(session, sample_type); @@ -687,7 +606,8 @@ static int __cmd_record(int argc, const char **argv) if (err < 0) return err; } else if (file_new) { - err = perf_header__write(&session->header, output, false); + err = perf_header__write(&session->header, evsel_list, + output, false); if (err < 0) return err; } @@ -697,22 +617,22 @@ static int __cmd_record(int argc, const char **argv) perf_session__set_sample_id_all(session, sample_id_all_avail); if (pipe_output) { - err = event__synthesize_attrs(&session->header, - process_synthesized_event, - session); + err = perf_event__synthesize_attrs(&session->header, + process_synthesized_event, + session); if (err < 0) { pr_err("Couldn't synthesize attrs.\n"); return err; } - err = event__synthesize_event_types(process_synthesized_event, - session); + err = perf_event__synthesize_event_types(process_synthesized_event, + session); if (err < 0) { pr_err("Couldn't synthesize event_types.\n"); return err; } - if (have_tracepoints(&evsel_list)) { + if (have_tracepoints(&evsel_list->entries)) { /* * FIXME err <= 0 here actually means that * there were no tracepoints so its not really @@ -721,9 +641,9 @@ static int __cmd_record(int argc, const char **argv) * return this more properly and also * propagate errors that now are calling die() */ - err = event__synthesize_tracing_data(output, &evsel_list, - process_synthesized_event, - session); + err = perf_event__synthesize_tracing_data(output, evsel_list, + process_synthesized_event, + session); if (err <= 0) { pr_err("Couldn't record tracing data.\n"); return err; @@ -738,31 +658,34 @@ static int __cmd_record(int argc, const char **argv) return -1; } - err = event__synthesize_kernel_mmap(process_synthesized_event, - session, machine, "_text"); + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + session, machine, "_text"); if (err < 0) - err = event__synthesize_kernel_mmap(process_synthesized_event, - session, machine, "_stext"); + err = perf_event__synthesize_kernel_mmap(process_synthesized_event, + session, machine, "_stext"); if (err < 0) pr_err("Couldn't record kernel reference relocation symbol\n" "Symbol resolution may be skewed if relocation was used (e.g. kexec).\n" "Check /proc/kallsyms permission or run as root.\n"); - err = event__synthesize_modules(process_synthesized_event, - session, machine); + err = perf_event__synthesize_modules(process_synthesized_event, + session, machine); if (err < 0) pr_err("Couldn't record kernel module information.\n" "Symbol resolution may be skewed if relocation was used (e.g. kexec).\n" "Check /proc/modules permission or run as root.\n"); if (perf_guest) - perf_session__process_machines(session, event__synthesize_guest_os); + perf_session__process_machines(session, + perf_event__synthesize_guest_os); if (!system_wide) - event__synthesize_thread_map(threads, process_synthesized_event, - session); + perf_event__synthesize_thread_map(evsel_list->threads, + process_synthesized_event, + session); else - event__synthesize_threads(process_synthesized_event, session); + perf_event__synthesize_threads(process_synthesized_event, + session); if (realtime_prio) { struct sched_param param; @@ -789,17 +712,17 @@ static int __cmd_record(int argc, const char **argv) if (hits == samples) { if (done) break; - err = poll(event_array, nr_poll, -1); + err = poll(evsel_list->pollfd, evsel_list->nr_fds, -1); waking++; } if (done) { - for (i = 0; i < nr_cpu; i++) { + for (i = 0; i < evsel_list->cpus->nr; i++) { struct perf_evsel *pos; - list_for_each_entry(pos, &evsel_list, node) { + list_for_each_entry(pos, &evsel_list->entries, node) { for (thread = 0; - thread < threads->nr; + thread < evsel_list->threads->nr; thread++) ioctl(FD(pos, i, thread), PERF_EVENT_IOC_DISABLE); @@ -838,10 +761,10 @@ static const char * const record_usage[] = { static bool force, append_file; const struct option record_options[] = { - OPT_CALLBACK('e', "event", NULL, "event", + OPT_CALLBACK('e', "event", &evsel_list, "event", "event selector. use 'perf list' to list available events", parse_events), - OPT_CALLBACK(0, "filter", NULL, "filter", + OPT_CALLBACK(0, "filter", &evsel_list, "filter", "event filter", parse_filter), OPT_INTEGER('p', "pid", &target_pid, "record events on existing process id"), @@ -892,6 +815,10 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) int err = -ENOMEM; struct perf_evsel *pos; + evsel_list = perf_evlist__new(NULL, NULL); + if (evsel_list == NULL) + return -ENOMEM; + argc = parse_options(argc, argv, record_options, record_usage, PARSE_OPT_STOP_AT_NON_OPTION); if (!argc && target_pid == -1 && target_tid == -1 && @@ -913,7 +840,8 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) if (no_buildid_cache || no_buildid) disable_buildid_cache(); - if (list_empty(&evsel_list) && perf_evsel_list__create_default() < 0) { + if (evsel_list->nr_entries == 0 && + perf_evlist__add_default(evsel_list) < 0) { pr_err("Not enough memory for event selector list\n"); goto out_symbol_exit; } @@ -921,27 +849,19 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) if (target_pid != -1) target_tid = target_pid; - threads = thread_map__new(target_pid, target_tid); - if (threads == NULL) { - pr_err("Problems finding threads of monitor\n"); + if (perf_evlist__create_maps(evsel_list, target_pid, + target_tid, cpu_list) < 0) usage_with_options(record_usage, record_options); - } - - cpus = cpu_map__new(cpu_list); - if (cpus == NULL) { - perror("failed to parse CPUs map"); - return -1; - } - list_for_each_entry(pos, &evsel_list, node) { - if (perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0) + list_for_each_entry(pos, &evsel_list->entries, node) { + if (perf_evsel__alloc_fd(pos, evsel_list->cpus->nr, + evsel_list->threads->nr) < 0) goto out_free_fd; if (perf_header__push_event(pos->attr.config, event_name(pos))) goto out_free_fd; } - event_array = malloc((sizeof(struct pollfd) * MAX_NR_CPUS * - MAX_COUNTERS * threads->nr)); - if (!event_array) + + if (perf_evlist__alloc_pollfd(evsel_list) < 0) goto out_free_fd; if (user_interval != ULLONG_MAX) @@ -959,16 +879,12 @@ int cmd_record(int argc, const char **argv, const char *prefix __used) } else { fprintf(stderr, "frequency and count are zero, aborting\n"); err = -EINVAL; - goto out_free_event_array; + goto out_free_fd; } err = __cmd_record(argc, argv); - -out_free_event_array: - free(event_array); out_free_fd: - thread_map__delete(threads); - threads = NULL; + perf_evlist__delete_maps(evsel_list); out_symbol_exit: symbol__exit(); return err; diff --git a/tools/perf/builtin-report.c b/tools/perf/builtin-report.c index c27e31f289e6..f403aced4cba 100644 --- a/tools/perf/builtin-report.c +++ b/tools/perf/builtin-report.c @@ -9,6 +9,7 @@ #include "util/util.h" +#include "util/annotate.h" #include "util/color.h" #include <linux/list.h> #include "util/cache.h" @@ -77,86 +78,96 @@ static struct hists *perf_session__hists_findnew(struct perf_session *self, return new; } -static int perf_session__add_hist_entry(struct perf_session *self, +static int perf_session__add_hist_entry(struct perf_session *session, struct addr_location *al, - struct sample_data *data) + struct perf_sample *sample) { - struct map_symbol *syms = NULL; struct symbol *parent = NULL; - int err = -ENOMEM; + int err = 0; struct hist_entry *he; struct hists *hists; struct perf_event_attr *attr; - if ((sort__has_parent || symbol_conf.use_callchain) && data->callchain) { - syms = perf_session__resolve_callchain(self, al->thread, - data->callchain, &parent); - if (syms == NULL) - return -ENOMEM; + if ((sort__has_parent || symbol_conf.use_callchain) && sample->callchain) { + err = perf_session__resolve_callchain(session, al->thread, + sample->callchain, &parent); + if (err) + return err; } - attr = perf_header__find_attr(data->id, &self->header); + attr = perf_header__find_attr(sample->id, &session->header); if (attr) - hists = perf_session__hists_findnew(self, data->id, attr->type, attr->config); + hists = perf_session__hists_findnew(session, sample->id, attr->type, attr->config); else - hists = perf_session__hists_findnew(self, data->id, 0, 0); + hists = perf_session__hists_findnew(session, sample->id, 0, 0); if (hists == NULL) - goto out_free_syms; - he = __hists__add_entry(hists, al, parent, data->period); + return -ENOMEM; + + he = __hists__add_entry(hists, al, parent, sample->period); if (he == NULL) - goto out_free_syms; - err = 0; + return -ENOMEM; + if (symbol_conf.use_callchain) { - err = callchain_append(he->callchain, data->callchain, syms, - data->period); + err = callchain_append(he->callchain, &session->callchain_cursor, + sample->period); if (err) - goto out_free_syms; + return err; } /* * Only in the newt browser we are doing integrated annotation, * so we don't allocated the extra space needed because the stdio * code will not use it. */ - if (use_browser > 0) - err = hist_entry__inc_addr_samples(he, al->addr); -out_free_syms: - free(syms); + if (al->sym != NULL && use_browser > 0) { + /* + * All aggregated on the first sym_hist. + */ + struct annotation *notes = symbol__annotation(he->ms.sym); + if (notes->src == NULL && + symbol__alloc_hist(he->ms.sym, 1) < 0) + err = -ENOMEM; + else + err = hist_entry__inc_addr_samples(he, 0, al->addr); + } + return err; } static int add_event_total(struct perf_session *session, - struct sample_data *data, + struct perf_sample *sample, struct perf_event_attr *attr) { struct hists *hists; if (attr) - hists = perf_session__hists_findnew(session, data->id, + hists = perf_session__hists_findnew(session, sample->id, attr->type, attr->config); else - hists = perf_session__hists_findnew(session, data->id, 0, 0); + hists = perf_session__hists_findnew(session, sample->id, 0, 0); if (!hists) return -ENOMEM; - hists->stats.total_period += data->period; + hists->stats.total_period += sample->period; /* * FIXME: add_event_total should be moved from here to * perf_session__process_event so that the proper hist is passed to * the event_op methods. */ hists__inc_nr_events(hists, PERF_RECORD_SAMPLE); - session->hists.stats.total_period += data->period; + session->hists.stats.total_period += sample->period; return 0; } -static int process_sample_event(event_t *event, struct sample_data *sample, +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct addr_location al; struct perf_event_attr *attr; - if (event__preprocess_sample(event, session, &al, sample, NULL) < 0) { + if (perf_event__preprocess_sample(event, session, &al, sample, + symbol__annotate_init) < 0) { fprintf(stderr, "problem processing %d event, skipping it.\n", event->header.type); return -1; @@ -180,7 +191,8 @@ static int process_sample_event(event_t *event, struct sample_data *sample, return 0; } -static int process_read_event(event_t *event, struct sample_data *sample __used, +static int process_read_event(union perf_event *event, + struct perf_sample *sample __used, struct perf_session *session __used) { struct perf_event_attr *attr; @@ -222,7 +234,7 @@ static int perf_session__setup_sample_type(struct perf_session *self) } else if (!dont_use_callchains && callchain_param.mode != CHAIN_NONE && !symbol_conf.use_callchain) { symbol_conf.use_callchain = true; - if (register_callchain_param(&callchain_param) < 0) { + if (callchain_register_param(&callchain_param) < 0) { fprintf(stderr, "Can't register callchain" " params\n"); return -EINVAL; @@ -233,17 +245,17 @@ static int perf_session__setup_sample_type(struct perf_session *self) } static struct perf_event_ops event_ops = { - .sample = process_sample_event, - .mmap = event__process_mmap, - .comm = event__process_comm, - .exit = event__process_task, - .fork = event__process_task, - .lost = event__process_lost, - .read = process_read_event, - .attr = event__process_attr, - .event_type = event__process_event_type, - .tracing_data = event__process_tracing_data, - .build_id = event__process_build_id, + .sample = process_sample_event, + .mmap = perf_event__process_mmap, + .comm = perf_event__process_comm, + .exit = perf_event__process_task, + .fork = perf_event__process_task, + .lost = perf_event__process_lost, + .read = process_read_event, + .attr = perf_event__process_attr, + .event_type = perf_event__process_event_type, + .tracing_data = perf_event__process_tracing_data, + .build_id = perf_event__process_build_id, .ordered_samples = true, .ordering_requires_timestamps = true, }; @@ -347,7 +359,7 @@ static int __cmd_report(void) } if (use_browser > 0) - hists__tui_browse_tree(&session->hists_tree, help); + hists__tui_browse_tree(&session->hists_tree, help, 0); else hists__tty_browse_tree(&session->hists_tree, help); @@ -424,7 +436,7 @@ parse_callchain_opt(const struct option *opt __used, const char *arg, if (tok2) callchain_param.print_limit = strtod(tok2, &endptr); setup: - if (register_callchain_param(&callchain_param) < 0) { + if (callchain_register_param(&callchain_param) < 0) { fprintf(stderr, "Can't register callchain params\n"); return -1; } @@ -498,7 +510,7 @@ int cmd_report(int argc, const char **argv, const char *prefix __used) use_browser = 1; if (strcmp(input_name, "-") != 0) - setup_browser(); + setup_browser(true); else use_browser = 0; /* @@ -507,7 +519,7 @@ int cmd_report(int argc, const char **argv, const char *prefix __used) * implementation. */ if (use_browser > 0) { - symbol_conf.priv_size = sizeof(struct sym_priv); + symbol_conf.priv_size = sizeof(struct annotation); /* * For searching by name on the "Browse map details". * providing it only in verbose mode not to bloat too diff --git a/tools/perf/builtin-sched.c b/tools/perf/builtin-sched.c index 29acb894e035..a32f411faeac 100644 --- a/tools/perf/builtin-sched.c +++ b/tools/perf/builtin-sched.c @@ -369,11 +369,6 @@ static void process_sched_event(struct task_desc *this_task __used, struct sched_atom *atom) { int ret = 0; - u64 now; - long long delta; - - now = get_nsecs(); - delta = start_time + atom->timestamp - now; switch (atom->type) { case SCHED_EVENT_RUN: @@ -562,7 +557,7 @@ static void wait_for_tasks(void) static void run_one_test(void) { - u64 T0, T1, delta, avg_delta, fluct, std_dev; + u64 T0, T1, delta, avg_delta, fluct; T0 = get_nsecs(); wait_for_tasks(); @@ -578,7 +573,6 @@ static void run_one_test(void) else fluct = delta - avg_delta; sum_fluct += fluct; - std_dev = sum_fluct / nr_runs / sqrt(nr_runs); if (!run_avg) run_avg = delta; run_avg = (run_avg*9 + delta)/10; @@ -799,7 +793,7 @@ replay_switch_event(struct trace_switch_event *switch_event, u64 timestamp, struct thread *thread __used) { - struct task_desc *prev, *next; + struct task_desc *prev, __used *next; u64 timestamp0; s64 delta; @@ -1404,7 +1398,7 @@ map_switch_event(struct trace_switch_event *switch_event, u64 timestamp, struct thread *thread __used) { - struct thread *sched_out, *sched_in; + struct thread *sched_out __used, *sched_in; int new_shortname; u64 timestamp0; s64 delta; @@ -1580,9 +1574,9 @@ process_sched_migrate_task_event(void *data, struct perf_session *session, event, cpu, timestamp, thread); } -static void -process_raw_event(event_t *raw_event __used, struct perf_session *session, - void *data, int cpu, u64 timestamp, struct thread *thread) +static void process_raw_event(union perf_event *raw_event __used, + struct perf_session *session, void *data, int cpu, + u64 timestamp, struct thread *thread) { struct event *event; int type; @@ -1607,7 +1601,8 @@ process_raw_event(event_t *raw_event __used, struct perf_session *session, process_sched_migrate_task_event(data, session, event, cpu, timestamp, thread); } -static int process_sample_event(event_t *event, struct sample_data *sample, +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct thread *thread; @@ -1635,9 +1630,9 @@ static int process_sample_event(event_t *event, struct sample_data *sample, static struct perf_event_ops event_ops = { .sample = process_sample_event, - .comm = event__process_comm, - .lost = event__process_lost, - .fork = event__process_task, + .comm = perf_event__process_comm, + .lost = perf_event__process_lost, + .fork = perf_event__process_task, .ordered_samples = true, }; diff --git a/tools/perf/builtin-script.c b/tools/perf/builtin-script.c index b766c2a9ac97..5f40df635dcb 100644 --- a/tools/perf/builtin-script.c +++ b/tools/perf/builtin-script.c @@ -63,7 +63,8 @@ static int cleanup_scripting(void) static char const *input_name = "perf.data"; -static int process_sample_event(event_t *event, struct sample_data *sample, +static int process_sample_event(union perf_event *event, + struct perf_sample *sample, struct perf_session *session) { struct thread *thread = perf_session__findnew(session, event->ip.pid); @@ -100,14 +101,14 @@ static int process_sample_event(event_t *event, struct sample_data *sample, } static struct perf_event_ops event_ops = { - .sample = process_sample_event, - .comm = event__process_comm, - .attr = event__process_attr, - .event_type = event__process_event_type, - .tracing_data = event__process_tracing_data, - .build_id = event__process_build_id, - .ordering_requires_timestamps = true, + .sample = process_sample_event, + .comm = perf_event__process_comm, + .attr = perf_event__process_attr, + .event_type = perf_event__process_event_type, + .tracing_data = perf_event__process_tracing_data, + .build_id = perf_event__process_build_id, .ordered_samples = true, + .ordering_requires_timestamps = true, }; extern volatile int session_done; diff --git a/tools/perf/builtin-stat.c b/tools/perf/builtin-stat.c index a482a191a0ca..806a9998fcd5 100644 --- a/tools/perf/builtin-stat.c +++ b/tools/perf/builtin-stat.c @@ -43,11 +43,13 @@ #include "util/parse-options.h" #include "util/parse-events.h" #include "util/event.h" +#include "util/evlist.h" #include "util/evsel.h" #include "util/debug.h" #include "util/header.h" #include "util/cpumap.h" #include "util/thread.h" +#include "util/thread_map.h" #include <sys/prctl.h> #include <math.h> @@ -71,8 +73,9 @@ static struct perf_event_attr default_attrs[] = { }; +struct perf_evlist *evsel_list; + static bool system_wide = false; -static struct cpu_map *cpus; static int run_idx = 0; static int run_count = 1; @@ -81,7 +84,6 @@ static bool scale = true; static bool no_aggr = false; static pid_t target_pid = -1; static pid_t target_tid = -1; -static struct thread_map *threads; static pid_t child_pid = -1; static bool null_run = false; static bool big_num = true; @@ -166,7 +168,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel) PERF_FORMAT_TOTAL_TIME_RUNNING; if (system_wide) - return perf_evsel__open_per_cpu(evsel, cpus); + return perf_evsel__open_per_cpu(evsel, evsel_list->cpus, false, false); attr->inherit = !no_inherit; if (target_pid == -1 && target_tid == -1) { @@ -174,7 +176,7 @@ static int create_perf_stat_counter(struct perf_evsel *evsel) attr->enable_on_exec = 1; } - return perf_evsel__open_per_thread(evsel, threads); + return perf_evsel__open_per_thread(evsel, evsel_list->threads, false, false); } /* @@ -199,7 +201,8 @@ static int read_counter_aggr(struct perf_evsel *counter) u64 *count = counter->counts->aggr.values; int i; - if (__perf_evsel__read(counter, cpus->nr, threads->nr, scale) < 0) + if (__perf_evsel__read(counter, evsel_list->cpus->nr, + evsel_list->threads->nr, scale) < 0) return -1; for (i = 0; i < 3; i++) @@ -232,7 +235,7 @@ static int read_counter(struct perf_evsel *counter) u64 *count; int cpu; - for (cpu = 0; cpu < cpus->nr; cpu++) { + for (cpu = 0; cpu < evsel_list->cpus->nr; cpu++) { if (__perf_evsel__read_on_cpu(counter, cpu, 0, scale) < 0) return -1; @@ -297,7 +300,7 @@ static int run_perf_stat(int argc __used, const char **argv) } if (target_tid == -1 && target_pid == -1 && !system_wide) - threads->map[0] = child_pid; + evsel_list->threads->map[0] = child_pid; /* * Wait for the child to be ready to exec. @@ -309,7 +312,7 @@ static int run_perf_stat(int argc __used, const char **argv) close(child_ready_pipe[0]); } - list_for_each_entry(counter, &evsel_list, node) { + list_for_each_entry(counter, &evsel_list->entries, node) { if (create_perf_stat_counter(counter) < 0) { if (errno == -EPERM || errno == -EACCES) { error("You may not have permission to collect %sstats.\n" @@ -347,14 +350,15 @@ static int run_perf_stat(int argc __used, const char **argv) update_stats(&walltime_nsecs_stats, t1 - t0); if (no_aggr) { - list_for_each_entry(counter, &evsel_list, node) { + list_for_each_entry(counter, &evsel_list->entries, node) { read_counter(counter); - perf_evsel__close_fd(counter, cpus->nr, 1); + perf_evsel__close_fd(counter, evsel_list->cpus->nr, 1); } } else { - list_for_each_entry(counter, &evsel_list, node) { + list_for_each_entry(counter, &evsel_list->entries, node) { read_counter_aggr(counter); - perf_evsel__close_fd(counter, cpus->nr, threads->nr); + perf_evsel__close_fd(counter, evsel_list->cpus->nr, + evsel_list->threads->nr); } } @@ -382,7 +386,7 @@ static void nsec_printout(int cpu, struct perf_evsel *evsel, double avg) if (no_aggr) sprintf(cpustr, "CPU%*d%s", csv_output ? 0 : -4, - cpus->map[cpu], csv_sep); + evsel_list->cpus->map[cpu], csv_sep); fprintf(stderr, fmt, cpustr, msecs, csv_sep, event_name(evsel)); @@ -410,7 +414,7 @@ static void abs_printout(int cpu, struct perf_evsel *evsel, double avg) if (no_aggr) sprintf(cpustr, "CPU%*d%s", csv_output ? 0 : -4, - cpus->map[cpu], csv_sep); + evsel_list->cpus->map[cpu], csv_sep); else cpu = 0; @@ -496,14 +500,14 @@ static void print_counter(struct perf_evsel *counter) u64 ena, run, val; int cpu; - for (cpu = 0; cpu < cpus->nr; cpu++) { + for (cpu = 0; cpu < evsel_list->cpus->nr; cpu++) { val = counter->counts->cpu[cpu].val; ena = counter->counts->cpu[cpu].ena; run = counter->counts->cpu[cpu].run; if (run == 0 || ena == 0) { fprintf(stderr, "CPU%*d%s%*s%s%-24s", csv_output ? 0 : -4, - cpus->map[cpu], csv_sep, + evsel_list->cpus->map[cpu], csv_sep, csv_output ? 0 : 18, "<not counted>", csv_sep, event_name(counter)); @@ -555,10 +559,10 @@ static void print_stat(int argc, const char **argv) } if (no_aggr) { - list_for_each_entry(counter, &evsel_list, node) + list_for_each_entry(counter, &evsel_list->entries, node) print_counter(counter); } else { - list_for_each_entry(counter, &evsel_list, node) + list_for_each_entry(counter, &evsel_list->entries, node) print_counter_aggr(counter); } @@ -610,7 +614,7 @@ static int stat__set_big_num(const struct option *opt __used, } static const struct option options[] = { - OPT_CALLBACK('e', "event", NULL, "event", + OPT_CALLBACK('e', "event", &evsel_list, "event", "event selector. use 'perf list' to list available events", parse_events), OPT_BOOLEAN('i', "no-inherit", &no_inherit, @@ -648,6 +652,10 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) setlocale(LC_ALL, ""); + evsel_list = perf_evlist__new(NULL, NULL); + if (evsel_list == NULL) + return -ENOMEM; + argc = parse_options(argc, argv, options, stat_usage, PARSE_OPT_STOP_AT_NON_OPTION); @@ -679,44 +687,41 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) usage_with_options(stat_usage, options); /* Set attrs and nr_counters if no event is selected and !null_run */ - if (!null_run && !nr_counters) { + if (!null_run && !evsel_list->nr_entries) { size_t c; - nr_counters = ARRAY_SIZE(default_attrs); - for (c = 0; c < ARRAY_SIZE(default_attrs); ++c) { - pos = perf_evsel__new(&default_attrs[c], - nr_counters); + pos = perf_evsel__new(&default_attrs[c], c); if (pos == NULL) goto out; - list_add(&pos->node, &evsel_list); + perf_evlist__add(evsel_list, pos); } } if (target_pid != -1) target_tid = target_pid; - threads = thread_map__new(target_pid, target_tid); - if (threads == NULL) { + evsel_list->threads = thread_map__new(target_pid, target_tid); + if (evsel_list->threads == NULL) { pr_err("Problems finding threads of monitor\n"); usage_with_options(stat_usage, options); } if (system_wide) - cpus = cpu_map__new(cpu_list); + evsel_list->cpus = cpu_map__new(cpu_list); else - cpus = cpu_map__dummy_new(); + evsel_list->cpus = cpu_map__dummy_new(); - if (cpus == NULL) { + if (evsel_list->cpus == NULL) { perror("failed to parse CPUs map"); usage_with_options(stat_usage, options); return -1; } - list_for_each_entry(pos, &evsel_list, node) { + list_for_each_entry(pos, &evsel_list->entries, node) { if (perf_evsel__alloc_stat_priv(pos) < 0 || - perf_evsel__alloc_counts(pos, cpus->nr) < 0 || - perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0) + perf_evsel__alloc_counts(pos, evsel_list->cpus->nr) < 0 || + perf_evsel__alloc_fd(pos, evsel_list->cpus->nr, evsel_list->threads->nr) < 0) goto out_free_fd; } @@ -741,11 +746,10 @@ int cmd_stat(int argc, const char **argv, const char *prefix __used) if (status != -1) print_stat(argc, argv); out_free_fd: - list_for_each_entry(pos, &evsel_list, node) + list_for_each_entry(pos, &evsel_list->entries, node) perf_evsel__free_stat_priv(pos); - perf_evsel_list__delete(); + perf_evlist__delete_maps(evsel_list); out: - thread_map__delete(threads); - threads = NULL; + perf_evlist__delete(evsel_list); return status; } diff --git a/tools/perf/builtin-test.c b/tools/perf/builtin-test.c index 5dcdba653d70..1b2106c58f66 100644 --- a/tools/perf/builtin-test.c +++ b/tools/perf/builtin-test.c @@ -7,10 +7,11 @@ #include "util/cache.h" #include "util/debug.h" +#include "util/evlist.h" #include "util/parse-options.h" -#include "util/session.h" +#include "util/parse-events.h" #include "util/symbol.h" -#include "util/thread.h" +#include "util/thread_map.h" static long page_size; @@ -238,14 +239,14 @@ out: #include "util/evsel.h" #include <sys/types.h> -static int trace_event__id(const char *event_name) +static int trace_event__id(const char *evname) { char *filename; int err = -1, fd; if (asprintf(&filename, "/sys/kernel/debug/tracing/events/syscalls/%s/id", - event_name) < 0) + evname) < 0) return -1; fd = open(filename, O_RDONLY); @@ -289,7 +290,7 @@ static int test__open_syscall_event(void) goto out_thread_map_delete; } - if (perf_evsel__open_per_thread(evsel, threads) < 0) { + if (perf_evsel__open_per_thread(evsel, threads, false, false) < 0) { pr_debug("failed to open counter: %s, " "tweak /proc/sys/kernel/perf_event_paranoid?\n", strerror(errno)); @@ -347,9 +348,9 @@ static int test__open_syscall_event_on_all_cpus(void) } cpus = cpu_map__new(NULL); - if (threads == NULL) { - pr_debug("thread_map__new\n"); - return -1; + if (cpus == NULL) { + pr_debug("cpu_map__new\n"); + goto out_thread_map_delete; } @@ -364,7 +365,7 @@ static int test__open_syscall_event_on_all_cpus(void) goto out_thread_map_delete; } - if (perf_evsel__open(evsel, cpus, threads) < 0) { + if (perf_evsel__open(evsel, cpus, threads, false, false) < 0) { pr_debug("failed to open counter: %s, " "tweak /proc/sys/kernel/perf_event_paranoid?\n", strerror(errno)); @@ -408,6 +409,8 @@ static int test__open_syscall_event_on_all_cpus(void) goto out_close_fd; } + err = 0; + for (cpu = 0; cpu < cpus->nr; ++cpu) { unsigned int expected; @@ -416,18 +419,18 @@ static int test__open_syscall_event_on_all_cpus(void) if (perf_evsel__read_on_cpu(evsel, cpu, 0) < 0) { pr_debug("perf_evsel__open_read_on_cpu\n"); - goto out_close_fd; + err = -1; + break; } expected = nr_open_calls + cpu; if (evsel->counts->cpu[cpu].val != expected) { pr_debug("perf_evsel__read_on_cpu: expected to intercept %d calls on cpu %d, got %" PRIu64 "\n", expected, cpus->map[cpu], evsel->counts->cpu[cpu].val); - goto out_close_fd; + err = -1; } } - err = 0; out_close_fd: perf_evsel__close_fd(evsel, 1, threads->nr); out_evsel_delete: @@ -437,6 +440,159 @@ out_thread_map_delete: return err; } +/* + * This test will generate random numbers of calls to some getpid syscalls, + * then establish an mmap for a group of events that are created to monitor + * the syscalls. + * + * It will receive the events, using mmap, use its PERF_SAMPLE_ID generated + * sample.id field to map back to its respective perf_evsel instance. + * + * Then it checks if the number of syscalls reported as perf events by + * the kernel corresponds to the number of syscalls made. + */ +static int test__basic_mmap(void) +{ + int err = -1; + union perf_event *event; + struct thread_map *threads; + struct cpu_map *cpus; + struct perf_evlist *evlist; + struct perf_event_attr attr = { + .type = PERF_TYPE_TRACEPOINT, + .read_format = PERF_FORMAT_ID, + .sample_type = PERF_SAMPLE_ID, + .watermark = 0, + }; + cpu_set_t cpu_set; + const char *syscall_names[] = { "getsid", "getppid", "getpgrp", + "getpgid", }; + pid_t (*syscalls[])(void) = { (void *)getsid, getppid, getpgrp, + (void*)getpgid }; +#define nsyscalls ARRAY_SIZE(syscall_names) + int ids[nsyscalls]; + unsigned int nr_events[nsyscalls], + expected_nr_events[nsyscalls], i, j; + struct perf_evsel *evsels[nsyscalls], *evsel; + + for (i = 0; i < nsyscalls; ++i) { + char name[64]; + + snprintf(name, sizeof(name), "sys_enter_%s", syscall_names[i]); + ids[i] = trace_event__id(name); + if (ids[i] < 0) { + pr_debug("Is debugfs mounted on /sys/kernel/debug?\n"); + return -1; + } + nr_events[i] = 0; + expected_nr_events[i] = random() % 257; + } + + threads = thread_map__new(-1, getpid()); + if (threads == NULL) { + pr_debug("thread_map__new\n"); + return -1; + } + + cpus = cpu_map__new(NULL); + if (cpus == NULL) { + pr_debug("cpu_map__new\n"); + goto out_free_threads; + } + + CPU_ZERO(&cpu_set); + CPU_SET(cpus->map[0], &cpu_set); + sched_setaffinity(0, sizeof(cpu_set), &cpu_set); + if (sched_setaffinity(0, sizeof(cpu_set), &cpu_set) < 0) { + pr_debug("sched_setaffinity() failed on CPU %d: %s ", + cpus->map[0], strerror(errno)); + goto out_free_cpus; + } + + evlist = perf_evlist__new(cpus, threads); + if (evlist == NULL) { + pr_debug("perf_evlist__new\n"); + goto out_free_cpus; + } + + /* anonymous union fields, can't be initialized above */ + attr.wakeup_events = 1; + attr.sample_period = 1; + + for (i = 0; i < nsyscalls; ++i) { + attr.config = ids[i]; + evsels[i] = perf_evsel__new(&attr, i); + if (evsels[i] == NULL) { + pr_debug("perf_evsel__new\n"); + goto out_free_evlist; + } + + perf_evlist__add(evlist, evsels[i]); + + if (perf_evsel__open(evsels[i], cpus, threads, false, false) < 0) { + pr_debug("failed to open counter: %s, " + "tweak /proc/sys/kernel/perf_event_paranoid?\n", + strerror(errno)); + goto out_close_fd; + } + } + + if (perf_evlist__mmap(evlist, 128, true) < 0) { + pr_debug("failed to mmap events: %d (%s)\n", errno, + strerror(errno)); + goto out_close_fd; + } + + for (i = 0; i < nsyscalls; ++i) + for (j = 0; j < expected_nr_events[i]; ++j) { + int foo = syscalls[i](); + ++foo; + } + + while ((event = perf_evlist__read_on_cpu(evlist, 0)) != NULL) { + struct perf_sample sample; + + if (event->header.type != PERF_RECORD_SAMPLE) { + pr_debug("unexpected %s event\n", + perf_event__name(event->header.type)); + goto out_munmap; + } + + perf_event__parse_sample(event, attr.sample_type, false, &sample); + evsel = perf_evlist__id2evsel(evlist, sample.id); + if (evsel == NULL) { + pr_debug("event with id %" PRIu64 + " doesn't map to an evsel\n", sample.id); + goto out_munmap; + } + nr_events[evsel->idx]++; + } + + list_for_each_entry(evsel, &evlist->entries, node) { + if (nr_events[evsel->idx] != expected_nr_events[evsel->idx]) { + pr_debug("expected %d %s events, got %d\n", + expected_nr_events[evsel->idx], + event_name(evsel), nr_events[evsel->idx]); + goto out_munmap; + } + } + + err = 0; +out_munmap: + perf_evlist__munmap(evlist); +out_close_fd: + for (i = 0; i < nsyscalls; ++i) + perf_evsel__close_fd(evsels[i], 1, threads->nr); +out_free_evlist: + perf_evlist__delete(evlist); +out_free_cpus: + cpu_map__delete(cpus); +out_free_threads: + thread_map__delete(threads); + return err; +#undef nsyscalls +} + static struct test { const char *desc; int (*func)(void); @@ -454,6 +610,10 @@ static struct test { .func = test__open_syscall_event_on_all_cpus, }, { + .desc = "read samples using the mmap interface", + .func = test__basic_mmap, + }, + { .func = NULL, }, }; diff --git a/tools/perf/builtin-timechart.c b/tools/perf/builtin-timechart.c index 746cf03cb05d..0801275c500e 100644 --- a/tools/perf/builtin-timechart.c +++ b/tools/perf/builtin-timechart.c @@ -276,21 +276,24 @@ static int cpus_cstate_state[MAX_CPUS]; static u64 cpus_pstate_start_times[MAX_CPUS]; static u64 cpus_pstate_state[MAX_CPUS]; -static int process_comm_event(event_t *event, struct sample_data *sample __used, +static int process_comm_event(union perf_event *event, + struct perf_sample *sample __used, struct perf_session *session __used) { pid_set_comm(event->comm.tid, event->comm.comm); return 0; } -static int process_fork_event(event_t *event, struct sample_data *sample __used, +static int process_fork_event(union perf_event *event, + struct perf_sample *sample __used, struct perf_session *session __used) { pid_fork(event->fork.pid, event->fork.ppid, event->fork.time); return 0; } -static int process_exit_event(event_t *event, struct sample_data *sample __used, +static int process_exit_event(union perf_event *event, + struct perf_sample *sample __used, struct perf_session *session __used) { pid_exit(event->fork.pid, event->fork.time); @@ -486,8 +489,8 @@ static void sched_switch(int cpu, u64 timestamp, struct trace_entry *te) } -static int process_sample_event(event_t *event __used, - struct sample_data *sample, +static int process_sample_event(union perf_event *event __used, + struct perf_sample *sample, struct perf_session *session) { struct trace_entry *te; diff --git a/tools/perf/builtin-top.c b/tools/perf/builtin-top.c index 5a29d9cd9486..c9fd66d4a082 100644 --- a/tools/perf/builtin-top.c +++ b/tools/perf/builtin-top.c @@ -20,11 +20,16 @@ #include "perf.h" +#include "util/annotate.h" +#include "util/cache.h" #include "util/color.h" +#include "util/evlist.h" #include "util/evsel.h" #include "util/session.h" #include "util/symbol.h" #include "util/thread.h" +#include "util/thread_map.h" +#include "util/top.h" #include "util/util.h" #include <linux/rbtree.h> #include "util/parse-options.h" @@ -45,7 +50,6 @@ #include <errno.h> #include <time.h> #include <sched.h> -#include <pthread.h> #include <sys/syscall.h> #include <sys/ioctl.h> @@ -60,85 +64,42 @@ #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) +static struct perf_top top = { + .count_filter = 5, + .delay_secs = 2, + .display_weighted = -1, + .target_pid = -1, + .target_tid = -1, + .active_symbols = LIST_HEAD_INIT(top.active_symbols), + .active_symbols_lock = PTHREAD_MUTEX_INITIALIZER, + .freq = 1000, /* 1 KHz */ +}; + static bool system_wide = false; -static int default_interval = 0; +static bool use_tui, use_stdio; -static int count_filter = 5; -static int print_entries; +static int default_interval = 0; -static int target_pid = -1; -static int target_tid = -1; -static struct thread_map *threads; static bool inherit = false; -static struct cpu_map *cpus; static int realtime_prio = 0; static bool group = false; static unsigned int page_size; -static unsigned int mmap_pages = 16; -static int freq = 1000; /* 1 KHz */ +static unsigned int mmap_pages = 128; -static int delay_secs = 2; -static bool zero = false; static bool dump_symtab = false; -static bool hide_kernel_symbols = false; -static bool hide_user_symbols = false; static struct winsize winsize; -/* - * Source - */ - -struct source_line { - u64 eip; - unsigned long count[MAX_COUNTERS]; - char *line; - struct source_line *next; -}; - static const char *sym_filter = NULL; struct sym_entry *sym_filter_entry = NULL; struct sym_entry *sym_filter_entry_sched = NULL; static int sym_pcnt_filter = 5; -static int sym_counter = 0; -static struct perf_evsel *sym_evsel = NULL; -static int display_weighted = -1; -static const char *cpu_list; - -/* - * Symbols - */ - -struct sym_entry_source { - struct source_line *source; - struct source_line *lines; - struct source_line **lines_tail; - pthread_mutex_t lock; -}; - -struct sym_entry { - struct rb_node rb_node; - struct list_head node; - unsigned long snap_count; - double weight; - int skip; - u16 name_len; - u8 origin; - struct map *map; - struct sym_entry_source *src; - unsigned long count[0]; -}; /* * Source functions */ -static inline struct symbol *sym_entry__symbol(struct sym_entry *self) -{ - return ((void *)self) + symbol_conf.priv_size; -} - void get_term_dimensions(struct winsize *ws) { char *s = getenv("LINES"); @@ -163,10 +124,10 @@ void get_term_dimensions(struct winsize *ws) static void update_print_entries(struct winsize *ws) { - print_entries = ws->ws_row; + top.print_entries = ws->ws_row; - if (print_entries > 9) - print_entries -= 9; + if (top.print_entries > 9) + top.print_entries -= 9; } static void sig_winch_handler(int sig __used) @@ -178,12 +139,9 @@ static void sig_winch_handler(int sig __used) static int parse_source(struct sym_entry *syme) { struct symbol *sym; - struct sym_entry_source *source; + struct annotation *notes; struct map *map; - FILE *file; - char command[PATH_MAX*2]; - const char *path; - u64 len; + int err = -1; if (!syme) return -1; @@ -194,411 +152,136 @@ static int parse_source(struct sym_entry *syme) /* * We can't annotate with just /proc/kallsyms */ - if (map->dso->origin == DSO__ORIG_KERNEL) + if (map->dso->origin == DSO__ORIG_KERNEL) { + pr_err("Can't annotate %s: No vmlinux file was found in the " + "path\n", sym->name); + sleep(1); return -1; - - if (syme->src == NULL) { - syme->src = zalloc(sizeof(*source)); - if (syme->src == NULL) - return -1; - pthread_mutex_init(&syme->src->lock, NULL); } - source = syme->src; - - if (source->lines) { - pthread_mutex_lock(&source->lock); + notes = symbol__annotation(sym); + if (notes->src != NULL) { + pthread_mutex_lock(¬es->lock); goto out_assign; } - path = map->dso->long_name; - - len = sym->end - sym->start; - - sprintf(command, - "objdump --start-address=%#0*" PRIx64 " --stop-address=%#0*" PRIx64 " -dS %s", - BITS_PER_LONG / 4, map__rip_2objdump(map, sym->start), - BITS_PER_LONG / 4, map__rip_2objdump(map, sym->end), path); - - file = popen(command, "r"); - if (!file) - return -1; - pthread_mutex_lock(&source->lock); - source->lines_tail = &source->lines; - while (!feof(file)) { - struct source_line *src; - size_t dummy = 0; - char *c, *sep; + pthread_mutex_lock(¬es->lock); - src = malloc(sizeof(struct source_line)); - assert(src != NULL); - memset(src, 0, sizeof(struct source_line)); - - if (getline(&src->line, &dummy, file) < 0) - break; - if (!src->line) - break; - - c = strchr(src->line, '\n'); - if (c) - *c = 0; - - src->next = NULL; - *source->lines_tail = src; - source->lines_tail = &src->next; - - src->eip = strtoull(src->line, &sep, 16); - if (*sep == ':') - src->eip = map__objdump_2ip(map, src->eip); - else /* this line has no ip info (e.g. source line) */ - src->eip = 0; + if (symbol__alloc_hist(sym, top.evlist->nr_entries) < 0) { + pr_err("Not enough memory for annotating '%s' symbol!\n", + sym->name); + sleep(1); + goto out_unlock; } - pclose(file); + + err = symbol__annotate(sym, syme->map, 0); + if (err == 0) { out_assign: sym_filter_entry = syme; - pthread_mutex_unlock(&source->lock); - return 0; + } +out_unlock: + pthread_mutex_unlock(¬es->lock); + return err; } static void __zero_source_counters(struct sym_entry *syme) { - int i; - struct source_line *line; - - line = syme->src->lines; - while (line) { - for (i = 0; i < nr_counters; i++) - line->count[i] = 0; - line = line->next; - } + struct symbol *sym = sym_entry__symbol(syme); + symbol__annotate_zero_histograms(sym); } static void record_precise_ip(struct sym_entry *syme, int counter, u64 ip) { - struct source_line *line; + struct annotation *notes; + struct symbol *sym; if (syme != sym_filter_entry) return; - if (pthread_mutex_trylock(&syme->src->lock)) - return; - - if (syme->src == NULL || syme->src->source == NULL) - goto out_unlock; - - for (line = syme->src->lines; line; line = line->next) { - /* skip lines without IP info */ - if (line->eip == 0) - continue; - if (line->eip == ip) { - line->count[counter]++; - break; - } - if (line->eip > ip) - break; - } -out_unlock: - pthread_mutex_unlock(&syme->src->lock); -} - -#define PATTERN_LEN (BITS_PER_LONG / 4 + 2) - -static void lookup_sym_source(struct sym_entry *syme) -{ - struct symbol *symbol = sym_entry__symbol(syme); - struct source_line *line; - char pattern[PATTERN_LEN + 1]; - - sprintf(pattern, "%0*" PRIx64 " <", BITS_PER_LONG / 4, - map__rip_2objdump(syme->map, symbol->start)); - - pthread_mutex_lock(&syme->src->lock); - for (line = syme->src->lines; line; line = line->next) { - if (memcmp(line->line, pattern, PATTERN_LEN) == 0) { - syme->src->source = line; - break; - } - } - pthread_mutex_unlock(&syme->src->lock); -} + sym = sym_entry__symbol(syme); + notes = symbol__annotation(sym); -static void show_lines(struct source_line *queue, int count, int total) -{ - int i; - struct source_line *line; + if (pthread_mutex_trylock(¬es->lock)) + return; - line = queue; - for (i = 0; i < count; i++) { - float pcnt = 100.0*(float)line->count[sym_counter]/(float)total; + ip = syme->map->map_ip(syme->map, ip); + symbol__inc_addr_samples(sym, syme->map, counter, ip); - printf("%8li %4.1f%%\t%s\n", line->count[sym_counter], pcnt, line->line); - line = line->next; - } + pthread_mutex_unlock(¬es->lock); } -#define TRACE_COUNT 3 - static void show_details(struct sym_entry *syme) { + struct annotation *notes; struct symbol *symbol; - struct source_line *line; - struct source_line *line_queue = NULL; - int displayed = 0; - int line_queue_count = 0, total = 0, more = 0; + int more; if (!syme) return; - if (!syme->src->source) - lookup_sym_source(syme); - - if (!syme->src->source) - return; - symbol = sym_entry__symbol(syme); - printf("Showing %s for %s\n", event_name(sym_evsel), symbol->name); - printf(" Events Pcnt (>=%d%%)\n", sym_pcnt_filter); + notes = symbol__annotation(symbol); - pthread_mutex_lock(&syme->src->lock); - line = syme->src->source; - while (line) { - total += line->count[sym_counter]; - line = line->next; - } - - line = syme->src->source; - while (line) { - float pcnt = 0.0; - - if (!line_queue_count) - line_queue = line; - line_queue_count++; - - if (line->count[sym_counter]) - pcnt = 100.0 * line->count[sym_counter] / (float)total; - if (pcnt >= (float)sym_pcnt_filter) { - if (displayed <= print_entries) - show_lines(line_queue, line_queue_count, total); - else more++; - displayed += line_queue_count; - line_queue_count = 0; - line_queue = NULL; - } else if (line_queue_count > TRACE_COUNT) { - line_queue = line_queue->next; - line_queue_count--; - } + pthread_mutex_lock(¬es->lock); - line->count[sym_counter] = zero ? 0 : line->count[sym_counter] * 7 / 8; - line = line->next; - } - pthread_mutex_unlock(&syme->src->lock); - if (more) - printf("%d lines not displayed, maybe increase display entries [e]\n", more); -} - -/* - * Symbols will be added here in event__process_sample and will get out - * after decayed. - */ -static LIST_HEAD(active_symbols); -static pthread_mutex_t active_symbols_lock = PTHREAD_MUTEX_INITIALIZER; - -/* - * Ordering weight: count-1 * count-2 * ... / count-n - */ -static double sym_weight(const struct sym_entry *sym) -{ - double weight = sym->snap_count; - int counter; - - if (!display_weighted) - return weight; - - for (counter = 1; counter < nr_counters-1; counter++) - weight *= sym->count[counter]; + if (notes->src == NULL) + goto out_unlock; - weight /= (sym->count[counter] + 1); + printf("Showing %s for %s\n", event_name(top.sym_evsel), symbol->name); + printf(" Events Pcnt (>=%d%%)\n", sym_pcnt_filter); - return weight; + more = symbol__annotate_printf(symbol, syme->map, top.sym_evsel->idx, + 0, sym_pcnt_filter, top.print_entries, 4); + if (top.zero) + symbol__annotate_zero_histogram(symbol, top.sym_evsel->idx); + else + symbol__annotate_decay_histogram(symbol, top.sym_evsel->idx); + if (more != 0) + printf("%d lines not displayed, maybe increase display entries [e]\n", more); +out_unlock: + pthread_mutex_unlock(¬es->lock); } -static long samples; -static long kernel_samples, us_samples; -static long exact_samples; -static long guest_us_samples, guest_kernel_samples; static const char CONSOLE_CLEAR[] = "[H[2J"; static void __list_insert_active_sym(struct sym_entry *syme) { - list_add(&syme->node, &active_symbols); -} - -static void list_remove_active_sym(struct sym_entry *syme) -{ - pthread_mutex_lock(&active_symbols_lock); - list_del_init(&syme->node); - pthread_mutex_unlock(&active_symbols_lock); -} - -static void rb_insert_active_sym(struct rb_root *tree, struct sym_entry *se) -{ - struct rb_node **p = &tree->rb_node; - struct rb_node *parent = NULL; - struct sym_entry *iter; - - while (*p != NULL) { - parent = *p; - iter = rb_entry(parent, struct sym_entry, rb_node); - - if (se->weight > iter->weight) - p = &(*p)->rb_left; - else - p = &(*p)->rb_right; - } - - rb_link_node(&se->rb_node, parent, p); - rb_insert_color(&se->rb_node, tree); + list_add(&syme->node, &top.active_symbols); } -static void print_sym_table(void) +static void print_sym_table(struct perf_session *session) { - int printed = 0, j; - struct perf_evsel *counter; - int snap = !display_weighted ? sym_counter : 0; - float samples_per_sec = samples/delay_secs; - float ksamples_per_sec = kernel_samples/delay_secs; - float us_samples_per_sec = (us_samples)/delay_secs; - float guest_kernel_samples_per_sec = (guest_kernel_samples)/delay_secs; - float guest_us_samples_per_sec = (guest_us_samples)/delay_secs; - float esamples_percent = (100.0*exact_samples)/samples; - float sum_ksamples = 0.0; - struct sym_entry *syme, *n; - struct rb_root tmp = RB_ROOT; + char bf[160]; + int printed = 0; struct rb_node *nd; - int sym_width = 0, dso_width = 0, dso_short_width = 0; + struct sym_entry *syme; + struct rb_root tmp = RB_ROOT; const int win_width = winsize.ws_col - 1; - - samples = us_samples = kernel_samples = exact_samples = 0; - guest_kernel_samples = guest_us_samples = 0; - - /* Sort the active symbols */ - pthread_mutex_lock(&active_symbols_lock); - syme = list_entry(active_symbols.next, struct sym_entry, node); - pthread_mutex_unlock(&active_symbols_lock); - - list_for_each_entry_safe_from(syme, n, &active_symbols, node) { - syme->snap_count = syme->count[snap]; - if (syme->snap_count != 0) { - - if ((hide_user_symbols && - syme->origin == PERF_RECORD_MISC_USER) || - (hide_kernel_symbols && - syme->origin == PERF_RECORD_MISC_KERNEL)) { - list_remove_active_sym(syme); - continue; - } - syme->weight = sym_weight(syme); - rb_insert_active_sym(&tmp, syme); - sum_ksamples += syme->snap_count; - - for (j = 0; j < nr_counters; j++) - syme->count[j] = zero ? 0 : syme->count[j] * 7 / 8; - } else - list_remove_active_sym(syme); - } + int sym_width, dso_width, dso_short_width; + float sum_ksamples = perf_top__decay_samples(&top, &tmp); puts(CONSOLE_CLEAR); - printf("%-*.*s\n", win_width, win_width, graph_dotted_line); - if (!perf_guest) { - printf(" PerfTop:%8.0f irqs/sec kernel:%4.1f%%" - " exact: %4.1f%% [", - samples_per_sec, - 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) / - samples_per_sec)), - esamples_percent); - } else { - printf(" PerfTop:%8.0f irqs/sec kernel:%4.1f%% us:%4.1f%%" - " guest kernel:%4.1f%% guest us:%4.1f%%" - " exact: %4.1f%% [", - samples_per_sec, - 100.0 - (100.0 * ((samples_per_sec-ksamples_per_sec) / - samples_per_sec)), - 100.0 - (100.0 * ((samples_per_sec-us_samples_per_sec) / - samples_per_sec)), - 100.0 - (100.0 * ((samples_per_sec - - guest_kernel_samples_per_sec) / - samples_per_sec)), - 100.0 - (100.0 * ((samples_per_sec - - guest_us_samples_per_sec) / - samples_per_sec)), - esamples_percent); - } - - if (nr_counters == 1 || !display_weighted) { - struct perf_evsel *first; - first = list_entry(evsel_list.next, struct perf_evsel, node); - printf("%" PRIu64, (uint64_t)first->attr.sample_period); - if (freq) - printf("Hz "); - else - printf(" "); - } + perf_top__header_snprintf(&top, bf, sizeof(bf)); + printf("%s\n", bf); - if (!display_weighted) - printf("%s", event_name(sym_evsel)); - else list_for_each_entry(counter, &evsel_list, node) { - if (counter->idx) - printf("/"); + perf_top__reset_sample_counters(&top); - printf("%s", event_name(counter)); - } - - printf( "], "); - - if (target_pid != -1) - printf(" (target_pid: %d", target_pid); - else if (target_tid != -1) - printf(" (target_tid: %d", target_tid); - else - printf(" (all"); + printf("%-*.*s\n", win_width, win_width, graph_dotted_line); - if (cpu_list) - printf(", CPU%s: %s)\n", cpus->nr > 1 ? "s" : "", cpu_list); - else { - if (target_tid != -1) - printf(")\n"); - else - printf(", %d CPU%s)\n", cpus->nr, cpus->nr > 1 ? "s" : ""); + if (session->hists.stats.total_lost != 0) { + color_fprintf(stdout, PERF_COLOR_RED, "WARNING:"); + printf(" LOST %" PRIu64 " events, Check IO/CPU overload\n", + session->hists.stats.total_lost); } - printf("%-*.*s\n", win_width, win_width, graph_dotted_line); - if (sym_filter_entry) { show_details(sym_filter_entry); return; } - /* - * Find the longest symbol name that will be displayed - */ - for (nd = rb_first(&tmp); nd; nd = rb_next(nd)) { - syme = rb_entry(nd, struct sym_entry, rb_node); - if (++printed > print_entries || - (int)syme->snap_count < count_filter) - continue; - - if (syme->map->dso->long_name_len > dso_width) - dso_width = syme->map->dso->long_name_len; - - if (syme->map->dso->short_name_len > dso_short_width) - dso_short_width = syme->map->dso->short_name_len; - - if (syme->name_len > sym_width) - sym_width = syme->name_len; - } - - printed = 0; + perf_top__find_widths(&top, &tmp, &dso_width, &dso_short_width, + &sym_width); if (sym_width + dso_width > winsize.ws_col - 29) { dso_width = dso_short_width; @@ -606,7 +289,7 @@ static void print_sym_table(void) sym_width = winsize.ws_col - dso_width - 29; } putchar('\n'); - if (nr_counters == 1) + if (top.evlist->nr_entries == 1) printf(" samples pcnt"); else printf(" weight samples pcnt"); @@ -615,7 +298,7 @@ static void print_sym_table(void) printf(" RIP "); printf(" %-*.*s DSO\n", sym_width, sym_width, "function"); printf(" %s _______ _____", - nr_counters == 1 ? " " : "______"); + top.evlist->nr_entries == 1 ? " " : "______"); if (verbose) printf(" ________________"); printf(" %-*.*s", sym_width, sym_width, graph_line); @@ -628,13 +311,14 @@ static void print_sym_table(void) syme = rb_entry(nd, struct sym_entry, rb_node); sym = sym_entry__symbol(syme); - if (++printed > print_entries || (int)syme->snap_count < count_filter) + if (++printed > top.print_entries || + (int)syme->snap_count < top.count_filter) continue; pcnt = 100.0 - (100.0 * ((sum_ksamples - syme->snap_count) / sum_ksamples)); - if (nr_counters == 1 || !display_weighted) + if (top.evlist->nr_entries == 1 || !top.display_weighted) printf("%20.2f ", syme->weight); else printf("%9.1f %10ld ", syme->weight, syme->snap_count); @@ -693,10 +377,8 @@ static void prompt_symbol(struct sym_entry **target, const char *msg) /* zero counters of active symbol */ if (syme) { - pthread_mutex_lock(&syme->src->lock); __zero_source_counters(syme); *target = NULL; - pthread_mutex_unlock(&syme->src->lock); } fprintf(stdout, "\n%s: ", msg); @@ -707,11 +389,11 @@ static void prompt_symbol(struct sym_entry **target, const char *msg) if (p) *p = 0; - pthread_mutex_lock(&active_symbols_lock); - syme = list_entry(active_symbols.next, struct sym_entry, node); - pthread_mutex_unlock(&active_symbols_lock); + pthread_mutex_lock(&top.active_symbols_lock); + syme = list_entry(top.active_symbols.next, struct sym_entry, node); + pthread_mutex_unlock(&top.active_symbols_lock); - list_for_each_entry_safe_from(syme, n, &active_symbols, node) { + list_for_each_entry_safe_from(syme, n, &top.active_symbols, node) { struct symbol *sym = sym_entry__symbol(syme); if (!strcmp(buf, sym->name)) { @@ -741,28 +423,28 @@ static void print_mapped_keys(void) } fprintf(stdout, "\nMapped keys:\n"); - fprintf(stdout, "\t[d] display refresh delay. \t(%d)\n", delay_secs); - fprintf(stdout, "\t[e] display entries (lines). \t(%d)\n", print_entries); + fprintf(stdout, "\t[d] display refresh delay. \t(%d)\n", top.delay_secs); + fprintf(stdout, "\t[e] display entries (lines). \t(%d)\n", top.print_entries); - if (nr_counters > 1) - fprintf(stdout, "\t[E] active event counter. \t(%s)\n", event_name(sym_evsel)); + if (top.evlist->nr_entries > 1) + fprintf(stdout, "\t[E] active event counter. \t(%s)\n", event_name(top.sym_evsel)); - fprintf(stdout, "\t[f] profile display filter (count). \t(%d)\n", count_filter); + fprintf(stdout, "\t[f] profile display filter (count). \t(%d)\n", top.count_filter); fprintf(stdout, "\t[F] annotate display filter (percent). \t(%d%%)\n", sym_pcnt_filter); fprintf(stdout, "\t[s] annotate symbol. \t(%s)\n", name?: "NULL"); fprintf(stdout, "\t[S] stop annotation.\n"); - if (nr_counters > 1) - fprintf(stdout, "\t[w] toggle display weighted/count[E]r. \t(%d)\n", display_weighted ? 1 : 0); + if (top.evlist->nr_entries > 1) + fprintf(stdout, "\t[w] toggle display weighted/count[E]r. \t(%d)\n", top.display_weighted ? 1 : 0); fprintf(stdout, "\t[K] hide kernel_symbols symbols. \t(%s)\n", - hide_kernel_symbols ? "yes" : "no"); + top.hide_kernel_symbols ? "yes" : "no"); fprintf(stdout, "\t[U] hide user symbols. \t(%s)\n", - hide_user_symbols ? "yes" : "no"); - fprintf(stdout, "\t[z] toggle sample zeroing. \t(%d)\n", zero ? 1 : 0); + top.hide_user_symbols ? "yes" : "no"); + fprintf(stdout, "\t[z] toggle sample zeroing. \t(%d)\n", top.zero ? 1 : 0); fprintf(stdout, "\t[qQ] quit.\n"); } @@ -783,7 +465,7 @@ static int key_mapped(int c) return 1; case 'E': case 'w': - return nr_counters > 1 ? 1 : 0; + return top.evlist->nr_entries > 1 ? 1 : 0; default: break; } @@ -818,47 +500,47 @@ static void handle_keypress(struct perf_session *session, int c) switch (c) { case 'd': - prompt_integer(&delay_secs, "Enter display delay"); - if (delay_secs < 1) - delay_secs = 1; + prompt_integer(&top.delay_secs, "Enter display delay"); + if (top.delay_secs < 1) + top.delay_secs = 1; break; case 'e': - prompt_integer(&print_entries, "Enter display entries (lines)"); - if (print_entries == 0) { + prompt_integer(&top.print_entries, "Enter display entries (lines)"); + if (top.print_entries == 0) { sig_winch_handler(SIGWINCH); signal(SIGWINCH, sig_winch_handler); } else signal(SIGWINCH, SIG_DFL); break; case 'E': - if (nr_counters > 1) { + if (top.evlist->nr_entries > 1) { fprintf(stderr, "\nAvailable events:"); - list_for_each_entry(sym_evsel, &evsel_list, node) - fprintf(stderr, "\n\t%d %s", sym_evsel->idx, event_name(sym_evsel)); + list_for_each_entry(top.sym_evsel, &top.evlist->entries, node) + fprintf(stderr, "\n\t%d %s", top.sym_evsel->idx, event_name(top.sym_evsel)); - prompt_integer(&sym_counter, "Enter details event counter"); + prompt_integer(&top.sym_counter, "Enter details event counter"); - if (sym_counter >= nr_counters) { - sym_evsel = list_entry(evsel_list.next, struct perf_evsel, node); - sym_counter = 0; - fprintf(stderr, "Sorry, no such event, using %s.\n", event_name(sym_evsel)); + if (top.sym_counter >= top.evlist->nr_entries) { + top.sym_evsel = list_entry(top.evlist->entries.next, struct perf_evsel, node); + top.sym_counter = 0; + fprintf(stderr, "Sorry, no such event, using %s.\n", event_name(top.sym_evsel)); sleep(1); break; } - list_for_each_entry(sym_evsel, &evsel_list, node) - if (sym_evsel->idx == sym_counter) + list_for_each_entry(top.sym_evsel, &top.evlist->entries, node) + if (top.sym_evsel->idx == top.sym_counter) break; - } else sym_counter = 0; + } else top.sym_counter = 0; break; case 'f': - prompt_integer(&count_filter, "Enter display event count filter"); + prompt_integer(&top.count_filter, "Enter display event count filter"); break; case 'F': prompt_percent(&sym_pcnt_filter, "Enter details display event filter (percent)"); break; case 'K': - hide_kernel_symbols = !hide_kernel_symbols; + top.hide_kernel_symbols = !top.hide_kernel_symbols; break; case 'q': case 'Q': @@ -875,26 +557,32 @@ static void handle_keypress(struct perf_session *session, int c) else { struct sym_entry *syme = sym_filter_entry; - pthread_mutex_lock(&syme->src->lock); sym_filter_entry = NULL; __zero_source_counters(syme); - pthread_mutex_unlock(&syme->src->lock); } break; case 'U': - hide_user_symbols = !hide_user_symbols; + top.hide_user_symbols = !top.hide_user_symbols; break; case 'w': - display_weighted = ~display_weighted; + top.display_weighted = ~top.display_weighted; break; case 'z': - zero = !zero; + top.zero = !top.zero; break; default: break; } } +static void *display_thread_tui(void *arg __used) +{ + perf_top__tui_browser(&top); + exit_browser(0); + exit(0); + return NULL; +} + static void *display_thread(void *arg __used) { struct pollfd stdin_poll = { .fd = 0, .events = POLLIN }; @@ -909,13 +597,13 @@ static void *display_thread(void *arg __used) tc.c_cc[VTIME] = 0; repeat: - delay_msecs = delay_secs * 1000; + delay_msecs = top.delay_secs * 1000; tcsetattr(0, TCSANOW, &tc); /* trash return*/ getc(stdin); do { - print_sym_table(); + print_sym_table(session); } while (!poll(&stdin_poll, 1, delay_msecs) == 1); c = getc(stdin); @@ -930,6 +618,7 @@ repeat: /* Tag samples to be skipped. */ static const char *skip_symbols[] = { "default_idle", + "native_safe_halt", "cpu_idle", "enter_idle", "exit_idle", @@ -965,7 +654,7 @@ static int symbol_filter(struct map *map, struct symbol *sym) syme = symbol__priv(sym); syme->map = map; - syme->src = NULL; + symbol__annotate_init(map, sym); if (!sym_filter_entry && sym_filter && !strcmp(name, sym_filter)) { /* schedule initial sym_filter_entry setup */ @@ -980,44 +669,40 @@ static int symbol_filter(struct map *map, struct symbol *sym) } } - if (!syme->skip) - syme->name_len = strlen(sym->name); - return 0; } -static void event__process_sample(const event_t *self, - struct sample_data *sample, - struct perf_session *session, - struct perf_evsel *evsel) +static void perf_event__process_sample(const union perf_event *event, + struct perf_sample *sample, + struct perf_session *session) { - u64 ip = self->ip.ip; + u64 ip = event->ip.ip; struct sym_entry *syme; struct addr_location al; struct machine *machine; - u8 origin = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + u8 origin = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; - ++samples; + ++top.samples; switch (origin) { case PERF_RECORD_MISC_USER: - ++us_samples; - if (hide_user_symbols) + ++top.us_samples; + if (top.hide_user_symbols) return; machine = perf_session__find_host_machine(session); break; case PERF_RECORD_MISC_KERNEL: - ++kernel_samples; - if (hide_kernel_symbols) + ++top.kernel_samples; + if (top.hide_kernel_symbols) return; machine = perf_session__find_host_machine(session); break; case PERF_RECORD_MISC_GUEST_KERNEL: - ++guest_kernel_samples; - machine = perf_session__find_machine(session, self->ip.pid); + ++top.guest_kernel_samples; + machine = perf_session__find_machine(session, event->ip.pid); break; case PERF_RECORD_MISC_GUEST_USER: - ++guest_us_samples; + ++top.guest_us_samples; /* * TODO: we don't process guest user from host side * except simple counting. @@ -1029,15 +714,15 @@ static void event__process_sample(const event_t *self, if (!machine && perf_guest) { pr_err("Can't find guest [%d]'s kernel information\n", - self->ip.pid); + event->ip.pid); return; } - if (self->header.misc & PERF_RECORD_MISC_EXACT_IP) - exact_samples++; + if (event->header.misc & PERF_RECORD_MISC_EXACT_IP) + top.exact_samples++; - if (event__preprocess_sample(self, session, &al, sample, - symbol_filter) < 0 || + if (perf_event__preprocess_sample(event, session, &al, sample, + symbol_filter) < 0 || al.filtered) return; @@ -1082,166 +767,67 @@ static void event__process_sample(const event_t *self, syme = symbol__priv(al.sym); if (!syme->skip) { - syme->count[evsel->idx]++; + struct perf_evsel *evsel; + syme->origin = origin; + evsel = perf_evlist__id2evsel(top.evlist, sample->id); + assert(evsel != NULL); + syme->count[evsel->idx]++; record_precise_ip(syme, evsel->idx, ip); - pthread_mutex_lock(&active_symbols_lock); + pthread_mutex_lock(&top.active_symbols_lock); if (list_empty(&syme->node) || !syme->node.next) __list_insert_active_sym(syme); - pthread_mutex_unlock(&active_symbols_lock); + pthread_mutex_unlock(&top.active_symbols_lock); } } -struct mmap_data { - void *base; - int mask; - unsigned int prev; -}; - -static int perf_evsel__alloc_mmap_per_thread(struct perf_evsel *evsel, - int ncpus, int nthreads) +static void perf_session__mmap_read_cpu(struct perf_session *self, int cpu) { - evsel->priv = xyarray__new(ncpus, nthreads, sizeof(struct mmap_data)); - return evsel->priv != NULL ? 0 : -ENOMEM; -} + struct perf_sample sample; + union perf_event *event; -static void perf_evsel__free_mmap(struct perf_evsel *evsel) -{ - xyarray__delete(evsel->priv); - evsel->priv = NULL; -} - -static unsigned int mmap_read_head(struct mmap_data *md) -{ - struct perf_event_mmap_page *pc = md->base; - int head; - - head = pc->data_head; - rmb(); - - return head; -} + while ((event = perf_evlist__read_on_cpu(top.evlist, cpu)) != NULL) { + perf_session__parse_sample(self, event, &sample); -static void perf_session__mmap_read_counter(struct perf_session *self, - struct perf_evsel *evsel, - int cpu, int thread_idx) -{ - struct xyarray *mmap_array = evsel->priv; - struct mmap_data *md = xyarray__entry(mmap_array, cpu, thread_idx); - unsigned int head = mmap_read_head(md); - unsigned int old = md->prev; - unsigned char *data = md->base + page_size; - struct sample_data sample; - int diff; - - /* - * If we're further behind than half the buffer, there's a chance - * the writer will bite our tail and mess up the samples under us. - * - * If we somehow ended up ahead of the head, we got messed up. - * - * In either case, truncate and restart at head. - */ - diff = head - old; - if (diff > md->mask / 2 || diff < 0) { - fprintf(stderr, "WARNING: failed to keep up with mmap data.\n"); - - /* - * head points to a known good entry, start there. - */ - old = head; - } - - for (; old != head;) { - event_t *event = (event_t *)&data[old & md->mask]; - - event_t event_copy; - - size_t size = event->header.size; - - /* - * Event straddles the mmap boundary -- header should always - * be inside due to u64 alignment of output. - */ - if ((old & md->mask) + size != ((old + size) & md->mask)) { - unsigned int offset = old; - unsigned int len = min(sizeof(*event), size), cpy; - void *dst = &event_copy; - - do { - cpy = min(md->mask + 1 - (offset & md->mask), len); - memcpy(dst, &data[offset & md->mask], cpy); - offset += cpy; - dst += cpy; - len -= cpy; - } while (len); - - event = &event_copy; - } - - event__parse_sample(event, self, &sample); if (event->header.type == PERF_RECORD_SAMPLE) - event__process_sample(event, &sample, self, evsel); + perf_event__process_sample(event, &sample, self); else - event__process(event, &sample, self); - old += size; + perf_event__process(event, &sample, self); } - - md->prev = old; } -static struct pollfd *event_array; - static void perf_session__mmap_read(struct perf_session *self) { - struct perf_evsel *counter; - int i, thread_index; - - for (i = 0; i < cpus->nr; i++) { - list_for_each_entry(counter, &evsel_list, node) { - for (thread_index = 0; - thread_index < threads->nr; - thread_index++) { - perf_session__mmap_read_counter(self, - counter, i, thread_index); - } - } - } -} + int i; -int nr_poll; -int group_fd; + for (i = 0; i < top.evlist->cpus->nr; i++) + perf_session__mmap_read_cpu(self, i); +} -static void start_counter(int i, struct perf_evsel *evsel) +static void start_counters(struct perf_evlist *evlist) { - struct xyarray *mmap_array = evsel->priv; - struct mmap_data *mm; - struct perf_event_attr *attr; - int cpu = -1; - int thread_index; - - if (target_tid == -1) - cpu = cpus->map[i]; + struct perf_evsel *counter; - attr = &evsel->attr; + list_for_each_entry(counter, &evlist->entries, node) { + struct perf_event_attr *attr = &counter->attr; - attr->sample_type = PERF_SAMPLE_IP | PERF_SAMPLE_TID; + attr->sample_type = PERF_SAMPLE_IP | PERF_SAMPLE_TID; - if (freq) { - attr->sample_type |= PERF_SAMPLE_PERIOD; - attr->freq = 1; - attr->sample_freq = freq; - } + if (top.freq) { + attr->sample_type |= PERF_SAMPLE_PERIOD; + attr->freq = 1; + attr->sample_freq = top.freq; + } - attr->inherit = (cpu < 0) && inherit; - attr->mmap = 1; + if (evlist->nr_entries > 1) { + attr->sample_type |= PERF_SAMPLE_ID; + attr->read_format |= PERF_FORMAT_ID; + } - for (thread_index = 0; thread_index < threads->nr; thread_index++) { + attr->mmap = 1; try_again: - FD(evsel, i, thread_index) = sys_perf_event_open(attr, - threads->map[thread_index], cpu, group_fd, 0); - - if (FD(evsel, i, thread_index) < 0) { + if (perf_evsel__open(counter, top.evlist->cpus, + top.evlist->threads, group, inherit) < 0) { int err = errno; if (err == EPERM || err == EACCES) @@ -1253,8 +839,8 @@ try_again: * based cpu-clock-tick sw counter, which * is always available even if no PMU support: */ - if (attr->type == PERF_TYPE_HARDWARE - && attr->config == PERF_COUNT_HW_CPU_CYCLES) { + if (attr->type == PERF_TYPE_HARDWARE && + attr->config == PERF_COUNT_HW_CPU_CYCLES) { if (verbose) warning(" ... trying to fall back to cpu-clock-ticks\n"); @@ -1264,39 +850,23 @@ try_again: goto try_again; } printf("\n"); - error("sys_perf_event_open() syscall returned with %d (%s). /bin/dmesg may provide additional information.\n", - FD(evsel, i, thread_index), strerror(err)); + error("sys_perf_event_open() syscall returned with %d " + "(%s). /bin/dmesg may provide additional information.\n", + err, strerror(err)); die("No CONFIG_PERF_EVENTS=y kernel support configured?\n"); exit(-1); } - assert(FD(evsel, i, thread_index) >= 0); - fcntl(FD(evsel, i, thread_index), F_SETFL, O_NONBLOCK); - - /* - * First counter acts as the group leader: - */ - if (group && group_fd == -1) - group_fd = FD(evsel, i, thread_index); - - event_array[nr_poll].fd = FD(evsel, i, thread_index); - event_array[nr_poll].events = POLLIN; - nr_poll++; - - mm = xyarray__entry(mmap_array, i, thread_index); - mm->prev = 0; - mm->mask = mmap_pages*page_size - 1; - mm->base = mmap(NULL, (mmap_pages+1)*page_size, - PROT_READ, MAP_SHARED, FD(evsel, i, thread_index), 0); - if (mm->base == MAP_FAILED) - die("failed to mmap with %d (%s)\n", errno, strerror(errno)); } + + if (perf_evlist__mmap(evlist, mmap_pages, false) < 0) + die("failed to mmap with %d (%s)\n", errno, strerror(errno)); } static int __cmd_top(void) { pthread_t thread; - struct perf_evsel *counter; - int i, ret; + struct perf_evsel *first; + int ret __used; /* * FIXME: perf_session__new should allow passing a O_MMAP, so that all this * mmap reading, etc is encapsulated in it. Use O_WRONLY for now. @@ -1305,23 +875,23 @@ static int __cmd_top(void) if (session == NULL) return -ENOMEM; - if (target_tid != -1) - event__synthesize_thread_map(threads, event__process, session); + if (top.target_tid != -1) + perf_event__synthesize_thread_map(top.evlist->threads, + perf_event__process, session); else - event__synthesize_threads(event__process, session); + perf_event__synthesize_threads(perf_event__process, session); - for (i = 0; i < cpus->nr; i++) { - group_fd = -1; - list_for_each_entry(counter, &evsel_list, node) - start_counter(i, counter); - } + start_counters(top.evlist); + first = list_entry(top.evlist->entries.next, struct perf_evsel, node); + perf_session__set_sample_type(session, first->attr.sample_type); /* Wait for a minimal set of events before starting the snapshot */ - poll(&event_array[0], nr_poll, 100); + poll(top.evlist->pollfd, top.evlist->nr_fds, 100); perf_session__mmap_read(session); - if (pthread_create(&thread, NULL, display_thread, session)) { + if (pthread_create(&thread, NULL, (use_browser > 0 ? display_thread_tui : + display_thread), session)) { printf("Could not create display thread.\n"); exit(-1); } @@ -1337,12 +907,12 @@ static int __cmd_top(void) } while (1) { - int hits = samples; + u64 hits = top.samples; perf_session__mmap_read(session); - if (hits == samples) - ret = poll(event_array, nr_poll, 100); + if (hits == top.samples) + ret = poll(top.evlist->pollfd, top.evlist->nr_fds, 100); } return 0; @@ -1354,31 +924,31 @@ static const char * const top_usage[] = { }; static const struct option options[] = { - OPT_CALLBACK('e', "event", NULL, "event", + OPT_CALLBACK('e', "event", &top.evlist, "event", "event selector. use 'perf list' to list available events", parse_events), OPT_INTEGER('c', "count", &default_interval, "event period to sample"), - OPT_INTEGER('p', "pid", &target_pid, + OPT_INTEGER('p', "pid", &top.target_pid, "profile events on existing process id"), - OPT_INTEGER('t', "tid", &target_tid, + OPT_INTEGER('t', "tid", &top.target_tid, "profile events on existing thread id"), OPT_BOOLEAN('a', "all-cpus", &system_wide, "system-wide collection from all CPUs"), - OPT_STRING('C', "cpu", &cpu_list, "cpu", + OPT_STRING('C', "cpu", &top.cpu_list, "cpu", "list of cpus to monitor"), OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name, "file", "vmlinux pathname"), - OPT_BOOLEAN('K', "hide_kernel_symbols", &hide_kernel_symbols, + OPT_BOOLEAN('K', "hide_kernel_symbols", &top.hide_kernel_symbols, "hide kernel symbols"), OPT_UINTEGER('m', "mmap-pages", &mmap_pages, "number of mmap data pages"), OPT_INTEGER('r', "realtime", &realtime_prio, "collect data with this RT SCHED_FIFO priority"), - OPT_INTEGER('d', "delay", &delay_secs, + OPT_INTEGER('d', "delay", &top.delay_secs, "number of seconds to delay between refreshes"), OPT_BOOLEAN('D', "dump-symtab", &dump_symtab, "dump the symbol table used for profiling"), - OPT_INTEGER('f', "count-filter", &count_filter, + OPT_INTEGER('f', "count-filter", &top.count_filter, "only display functions with more events than this"), OPT_BOOLEAN('g', "group", &group, "put the counters into a counter group"), @@ -1386,14 +956,16 @@ static const struct option options[] = { "child tasks inherit counters"), OPT_STRING('s', "sym-annotate", &sym_filter, "symbol name", "symbol to annotate"), - OPT_BOOLEAN('z', "zero", &zero, + OPT_BOOLEAN('z', "zero", &top.zero, "zero history across updates"), - OPT_INTEGER('F', "freq", &freq, + OPT_INTEGER('F', "freq", &top.freq, "profile at this frequency"), - OPT_INTEGER('E', "entries", &print_entries, + OPT_INTEGER('E', "entries", &top.print_entries, "display this many functions"), - OPT_BOOLEAN('U', "hide_user_symbols", &hide_user_symbols, + OPT_BOOLEAN('U', "hide_user_symbols", &top.hide_user_symbols, "hide user symbols"), + OPT_BOOLEAN(0, "tui", &use_tui, "Use the TUI interface"), + OPT_BOOLEAN(0, "stdio", &use_stdio, "Use the stdio interface"), OPT_INCR('v', "verbose", &verbose, "be more verbose (show counter open errors, etc)"), OPT_END() @@ -1404,64 +976,68 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) struct perf_evsel *pos; int status = -ENOMEM; + top.evlist = perf_evlist__new(NULL, NULL); + if (top.evlist == NULL) + return -ENOMEM; + page_size = sysconf(_SC_PAGE_SIZE); argc = parse_options(argc, argv, options, top_usage, 0); if (argc) usage_with_options(top_usage, options); - if (target_pid != -1) - target_tid = target_pid; + /* + * XXX For now start disabled, only using TUI if explicitely asked for. + * Change that when handle_keys equivalent gets written, live annotation + * done, etc. + */ + use_browser = 0; - threads = thread_map__new(target_pid, target_tid); - if (threads == NULL) { - pr_err("Problems finding threads of monitor\n"); - usage_with_options(top_usage, options); - } + if (use_stdio) + use_browser = 0; + else if (use_tui) + use_browser = 1; - event_array = malloc((sizeof(struct pollfd) * - MAX_NR_CPUS * MAX_COUNTERS * threads->nr)); - if (!event_array) - return -ENOMEM; + setup_browser(false); /* CPU and PID are mutually exclusive */ - if (target_tid > 0 && cpu_list) { + if (top.target_tid > 0 && top.cpu_list) { printf("WARNING: PID switch overriding CPU\n"); sleep(1); - cpu_list = NULL; + top.cpu_list = NULL; } - if (!nr_counters && perf_evsel_list__create_default() < 0) { + if (top.target_pid != -1) + top.target_tid = top.target_pid; + + if (perf_evlist__create_maps(top.evlist, top.target_pid, + top.target_tid, top.cpu_list) < 0) + usage_with_options(top_usage, options); + + if (!top.evlist->nr_entries && + perf_evlist__add_default(top.evlist) < 0) { pr_err("Not enough memory for event selector list\n"); return -ENOMEM; } - if (delay_secs < 1) - delay_secs = 1; + if (top.delay_secs < 1) + top.delay_secs = 1; /* * User specified count overrides default frequency. */ if (default_interval) - freq = 0; - else if (freq) { - default_interval = freq; + top.freq = 0; + else if (top.freq) { + default_interval = top.freq; } else { fprintf(stderr, "frequency and count are zero, aborting\n"); exit(EXIT_FAILURE); } - if (target_tid != -1) - cpus = cpu_map__dummy_new(); - else - cpus = cpu_map__new(cpu_list); - - if (cpus == NULL) - usage_with_options(top_usage, options); - - list_for_each_entry(pos, &evsel_list, node) { - if (perf_evsel__alloc_mmap_per_thread(pos, cpus->nr, threads->nr) < 0 || - perf_evsel__alloc_fd(pos, cpus->nr, threads->nr) < 0) + list_for_each_entry(pos, &top.evlist->entries, node) { + if (perf_evsel__alloc_fd(pos, top.evlist->cpus->nr, + top.evlist->threads->nr) < 0) goto out_free_fd; /* * Fill in the ones not specifically initialized via -c: @@ -1472,26 +1048,28 @@ int cmd_top(int argc, const char **argv, const char *prefix __used) pos->attr.sample_period = default_interval; } - sym_evsel = list_entry(evsel_list.next, struct perf_evsel, node); + if (perf_evlist__alloc_pollfd(top.evlist) < 0 || + perf_evlist__alloc_mmap(top.evlist) < 0) + goto out_free_fd; + + top.sym_evsel = list_entry(top.evlist->entries.next, struct perf_evsel, node); - symbol_conf.priv_size = (sizeof(struct sym_entry) + - (nr_counters + 1) * sizeof(unsigned long)); + symbol_conf.priv_size = (sizeof(struct sym_entry) + sizeof(struct annotation) + + (top.evlist->nr_entries + 1) * sizeof(unsigned long)); symbol_conf.try_vmlinux_path = (symbol_conf.vmlinux_name == NULL); if (symbol__init() < 0) return -1; get_term_dimensions(&winsize); - if (print_entries == 0) { + if (top.print_entries == 0) { update_print_entries(&winsize); signal(SIGWINCH, sig_winch_handler); } status = __cmd_top(); out_free_fd: - list_for_each_entry(pos, &evsel_list, node) - perf_evsel__free_mmap(pos); - perf_evsel_list__delete(); + perf_evlist__delete(top.evlist); return status; } diff --git a/tools/perf/perf.h b/tools/perf/perf.h index 95aaf565c704..a5fc660c1f12 100644 --- a/tools/perf/perf.h +++ b/tools/perf/perf.h @@ -94,6 +94,32 @@ void get_term_dimensions(struct winsize *ws); #include "util/types.h" #include <stdbool.h> +struct perf_mmap { + void *base; + int mask; + unsigned int prev; +}; + +static inline unsigned int perf_mmap__read_head(struct perf_mmap *mm) +{ + struct perf_event_mmap_page *pc = mm->base; + int head = pc->data_head; + rmb(); + return head; +} + +static inline void perf_mmap__write_tail(struct perf_mmap *md, + unsigned long tail) +{ + struct perf_event_mmap_page *pc = md->base; + + /* + * ensure all reads are done before we write the tail out. + */ + /* mb(); */ + pc->data_tail = tail; +} + /* * prctl(PR_TASK_PERF_EVENTS_DISABLE) will (cheaply) disable all * counters in the current task. diff --git a/tools/perf/python/twatch.py b/tools/perf/python/twatch.py new file mode 100755 index 000000000000..df638c438a9f --- /dev/null +++ b/tools/perf/python/twatch.py @@ -0,0 +1,41 @@ +#! /usr/bin/python +# -*- python -*- +# -*- coding: utf-8 -*- +# twatch - Experimental use of the perf python interface +# Copyright (C) 2011 Arnaldo Carvalho de Melo <acme@redhat.com> +# +# This application is free software; you can redistribute it and/or +# modify it under the terms of the GNU General Public License +# as published by the Free Software Foundation; version 2. +# +# This application is distributed in the hope that it will be useful, +# but WITHOUT ANY WARRANTY; without even the implied warranty of +# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU +# General Public License for more details. + +import perf + +def main(): + cpus = perf.cpu_map() + threads = perf.thread_map() + evsel = perf.evsel(task = 1, comm = 1, mmap = 0, + wakeup_events = 1, sample_period = 1, + sample_id_all = 1, + sample_type = perf.SAMPLE_PERIOD | perf.SAMPLE_TID | perf.SAMPLE_CPU | perf.SAMPLE_TID) + evsel.open(cpus = cpus, threads = threads); + evlist = perf.evlist(cpus, threads) + evlist.add(evsel) + evlist.mmap() + while True: + evlist.poll(timeout = -1) + for cpu in cpus: + event = evlist.read_on_cpu(cpu) + if not event: + continue + print "cpu: %2d, pid: %4d, tid: %4d" % (event.sample_cpu, + event.sample_pid, + event.sample_tid), + print event + +if __name__ == '__main__': + main() diff --git a/tools/perf/util/annotate.c b/tools/perf/util/annotate.c new file mode 100644 index 000000000000..02976b895f27 --- /dev/null +++ b/tools/perf/util/annotate.c @@ -0,0 +1,593 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Parts came from builtin-annotate.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ + +#include "util.h" +#include "build-id.h" +#include "color.h" +#include "cache.h" +#include "symbol.h" +#include "debug.h" +#include "annotate.h" +#include <pthread.h> + +int symbol__annotate_init(struct map *map __used, struct symbol *sym) +{ + struct annotation *notes = symbol__annotation(sym); + pthread_mutex_init(¬es->lock, NULL); + return 0; +} + +int symbol__alloc_hist(struct symbol *sym, int nevents) +{ + 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); + if (notes->src == NULL) + return -1; + notes->src->sizeof_sym_hist = sizeof_sym_hist; + notes->src->nr_histograms = nevents; + INIT_LIST_HEAD(¬es->src->source); + return 0; +} + +void symbol__annotate_zero_histograms(struct symbol *sym) +{ + struct annotation *notes = symbol__annotation(sym); + + pthread_mutex_lock(¬es->lock); + if (notes->src != NULL) + memset(notes->src->histograms, 0, + notes->src->nr_histograms * notes->src->sizeof_sym_hist); + pthread_mutex_unlock(¬es->lock); +} + +int symbol__inc_addr_samples(struct symbol *sym, struct map *map, + int evidx, u64 addr) +{ + unsigned offset; + struct annotation *notes; + struct sym_hist *h; + + notes = symbol__annotation(sym); + if (notes->src == NULL) + return -ENOMEM; + + pr_debug3("%s: addr=%#" PRIx64 "\n", __func__, map->unmap_ip(map, addr)); + + if (addr >= sym->end) + return 0; + + offset = addr - sym->start; + h = annotation__histogram(notes, evidx); + h->sum++; + h->addr[offset]++; + + pr_debug3("%#" PRIx64 " %s: period++ [addr: %#" PRIx64 ", %#" PRIx64 + ", evidx=%d] => %" PRIu64 "\n", sym->start, sym->name, + addr, addr - sym->start, evidx, h->addr[offset]); + return 0; +} + +static struct objdump_line *objdump_line__new(s64 offset, char *line, size_t privsize) +{ + struct objdump_line *self = malloc(sizeof(*self) + privsize); + + if (self != NULL) { + self->offset = offset; + self->line = line; + } + + return self; +} + +void objdump_line__free(struct objdump_line *self) +{ + free(self->line); + free(self); +} + +static void objdump__add_line(struct list_head *head, struct objdump_line *line) +{ + list_add_tail(&line->node, head); +} + +struct objdump_line *objdump__get_next_ip_line(struct list_head *head, + struct objdump_line *pos) +{ + list_for_each_entry_continue(pos, head, node) + if (pos->offset >= 0) + return pos; + + return NULL; +} + +static int objdump_line__print(struct objdump_line *oline, struct symbol *sym, + int evidx, u64 len, int min_pcnt, + int printed, int max_lines, + struct objdump_line *queue) +{ + static const char *prev_line; + static const char *prev_color; + + if (oline->offset != -1) { + const char *path = NULL; + unsigned int hits = 0; + double percent = 0.0; + const char *color; + struct annotation *notes = symbol__annotation(sym); + struct source_line *src_line = notes->src->lines; + struct sym_hist *h = annotation__histogram(notes, evidx); + s64 offset = oline->offset; + struct objdump_line *next; + + next = objdump__get_next_ip_line(¬es->src->source, oline); + + while (offset < (s64)len && + (next == NULL || offset < next->offset)) { + if (src_line) { + if (path == NULL) + path = src_line[offset].path; + percent += src_line[offset].percent; + } else + hits += h->addr[offset]; + + ++offset; + } + + if (src_line == NULL && h->sum) + percent = 100.0 * hits / h->sum; + + if (percent < min_pcnt) + return -1; + + if (max_lines && printed >= max_lines) + return 1; + + if (queue != NULL) { + list_for_each_entry_from(queue, ¬es->src->source, node) { + if (queue == oline) + break; + objdump_line__print(queue, sym, evidx, len, + 0, 0, 1, NULL); + } + } + + color = get_percent_color(percent); + + /* + * Also color the filename and line if needed, with + * the same color than the percentage. Don't print it + * twice for close colored addr with the same filename:line + */ + if (path) { + if (!prev_line || strcmp(prev_line, path) + || color != prev_color) { + color_fprintf(stdout, color, " %s", path); + prev_line = path; + prev_color = color; + } + } + + color_fprintf(stdout, color, " %7.2f", percent); + printf(" : "); + color_fprintf(stdout, PERF_COLOR_BLUE, "%s\n", oline->line); + } else if (max_lines && printed >= max_lines) + return 1; + else { + if (queue) + return -1; + + if (!*oline->line) + printf(" :\n"); + else + printf(" : %s\n", oline->line); + } + + return 0; +} + +static int symbol__parse_objdump_line(struct symbol *sym, struct map *map, + FILE *file, size_t privsize) +{ + struct annotation *notes = symbol__annotation(sym); + struct objdump_line *objdump_line; + char *line = NULL, *tmp, *tmp2, *c; + size_t line_len; + s64 line_ip, offset = -1; + + if (getline(&line, &line_len, file) < 0) + return -1; + + if (!line) + return -1; + + while (line_len != 0 && isspace(line[line_len - 1])) + line[--line_len] = '\0'; + + c = strchr(line, '\n'); + if (c) + *c = 0; + + line_ip = -1; + + /* + * Strip leading spaces: + */ + tmp = line; + while (*tmp) { + if (*tmp != ' ') + break; + tmp++; + } + + if (*tmp) { + /* + * Parse hexa addresses followed by ':' + */ + line_ip = strtoull(tmp, &tmp2, 16); + if (*tmp2 != ':' || tmp == tmp2 || tmp2[1] == '\0') + line_ip = -1; + } + + if (line_ip != -1) { + u64 start = map__rip_2objdump(map, sym->start), + end = map__rip_2objdump(map, sym->end); + + offset = line_ip - start; + if (offset < 0 || (u64)line_ip > end) + offset = -1; + } + + objdump_line = objdump_line__new(offset, line, privsize); + if (objdump_line == NULL) { + free(line); + return -1; + } + objdump__add_line(¬es->src->source, objdump_line); + + return 0; +} + +int symbol__annotate(struct symbol *sym, struct map *map, size_t privsize) +{ + struct dso *dso = map->dso; + char *filename = dso__build_id_filename(dso, NULL, 0); + bool free_filename = true; + char command[PATH_MAX * 2]; + FILE *file; + int err = 0; + char symfs_filename[PATH_MAX]; + + if (filename) { + snprintf(symfs_filename, sizeof(symfs_filename), "%s%s", + symbol_conf.symfs, filename); + } + + if (filename == NULL) { + if (dso->has_build_id) { + pr_err("Can't annotate %s: not enough memory\n", + sym->name); + return -ENOMEM; + } + goto fallback; + } else if (readlink(symfs_filename, command, sizeof(command)) < 0 || + strstr(command, "[kernel.kallsyms]") || + access(symfs_filename, R_OK)) { + free(filename); +fallback: + /* + * If we don't have build-ids or the build-id file isn't in the + * cache, or is just a kallsyms file, well, lets hope that this + * DSO is the same as when 'perf record' ran. + */ + filename = dso->long_name; + snprintf(symfs_filename, sizeof(symfs_filename), "%s%s", + symbol_conf.symfs, filename); + free_filename = false; + } + + if (dso->origin == DSO__ORIG_KERNEL) { + if (dso->annotate_warned) + goto out_free_filename; + err = -ENOENT; + dso->annotate_warned = 1; + pr_err("Can't annotate %s: No vmlinux file was found in the " + "path\n", sym->name); + goto out_free_filename; + } + + pr_debug("%s: filename=%s, sym=%s, start=%#" PRIx64 ", end=%#" PRIx64 "\n", __func__, + filename, sym->name, map->unmap_ip(map, sym->start), + map->unmap_ip(map, sym->end)); + + pr_debug("annotating [%p] %30s : [%p] %30s\n", + dso, dso->long_name, sym, sym->name); + + snprintf(command, sizeof(command), + "objdump --start-address=0x%016" PRIx64 + " --stop-address=0x%016" PRIx64 " -dS -C %s|grep -v %s|expand", + map__rip_2objdump(map, sym->start), + map__rip_2objdump(map, sym->end), + symfs_filename, filename); + + pr_debug("Executing: %s\n", command); + + file = popen(command, "r"); + if (!file) + goto out_free_filename; + + while (!feof(file)) + if (symbol__parse_objdump_line(sym, map, file, privsize) < 0) + break; + + pclose(file); +out_free_filename: + if (free_filename) + free(filename); + return err; +} + +static void insert_source_line(struct rb_root *root, struct source_line *src_line) +{ + struct source_line *iter; + struct rb_node **p = &root->rb_node; + struct rb_node *parent = NULL; + + while (*p != NULL) { + parent = *p; + iter = rb_entry(parent, struct source_line, node); + + if (src_line->percent > iter->percent) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&src_line->node, parent, p); + rb_insert_color(&src_line->node, root); +} + +static void symbol__free_source_line(struct symbol *sym, int len) +{ + struct annotation *notes = symbol__annotation(sym); + struct source_line *src_line = notes->src->lines; + int i; + + for (i = 0; i < len; i++) + free(src_line[i].path); + + free(src_line); + notes->src->lines = NULL; +} + +/* Get the filename:line for the colored entries */ +static int symbol__get_source_line(struct symbol *sym, struct map *map, + int evidx, struct rb_root *root, int len, + const char *filename) +{ + u64 start; + int i; + char cmd[PATH_MAX * 2]; + struct source_line *src_line; + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = annotation__histogram(notes, evidx); + + if (!h->sum) + return 0; + + src_line = notes->src->lines = calloc(len, sizeof(struct source_line)); + if (!notes->src->lines) + return -1; + + start = map->unmap_ip(map, sym->start); + + for (i = 0; i < len; i++) { + char *path = NULL; + size_t line_len; + u64 offset; + FILE *fp; + + src_line[i].percent = 100.0 * h->addr[i] / h->sum; + if (src_line[i].percent <= 0.5) + continue; + + offset = start + i; + sprintf(cmd, "addr2line -e %s %016" PRIx64, filename, offset); + fp = popen(cmd, "r"); + if (!fp) + continue; + + if (getline(&path, &line_len, fp) < 0 || !line_len) + goto next; + + src_line[i].path = malloc(sizeof(char) * line_len + 1); + if (!src_line[i].path) + goto next; + + strcpy(src_line[i].path, path); + insert_source_line(root, &src_line[i]); + + next: + pclose(fp); + } + + return 0; +} + +static void print_summary(struct rb_root *root, const char *filename) +{ + struct source_line *src_line; + struct rb_node *node; + + printf("\nSorted summary for file %s\n", filename); + printf("----------------------------------------------\n\n"); + + if (RB_EMPTY_ROOT(root)) { + printf(" Nothing higher than %1.1f%%\n", MIN_GREEN); + return; + } + + node = rb_first(root); + while (node) { + double percent; + const char *color; + char *path; + + src_line = rb_entry(node, struct source_line, node); + percent = src_line->percent; + color = get_percent_color(percent); + path = src_line->path; + + color_fprintf(stdout, color, " %7.2f %s", percent, path); + node = rb_next(node); + } +} + +static void symbol__annotate_hits(struct symbol *sym, int evidx) +{ + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = annotation__histogram(notes, evidx); + u64 len = sym->end - sym->start, offset; + + for (offset = 0; offset < len; ++offset) + if (h->addr[offset] != 0) + printf("%*" PRIx64 ": %" PRIu64 "\n", BITS_PER_LONG / 2, + sym->start + offset, h->addr[offset]); + printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->sum", h->sum); +} + +int symbol__annotate_printf(struct symbol *sym, struct map *map, int evidx, + bool full_paths, int min_pcnt, int max_lines, + int context) +{ + struct dso *dso = map->dso; + const char *filename = dso->long_name, *d_filename; + struct annotation *notes = symbol__annotation(sym); + struct objdump_line *pos, *queue = NULL; + int printed = 2, queue_len = 0; + int more = 0; + u64 len; + + if (full_paths) + d_filename = filename; + else + d_filename = basename(filename); + + len = sym->end - sym->start; + + printf(" Percent | Source code & Disassembly of %s\n", d_filename); + printf("------------------------------------------------\n"); + + if (verbose) + symbol__annotate_hits(sym, evidx); + + list_for_each_entry(pos, ¬es->src->source, node) { + if (context && queue == NULL) { + queue = pos; + queue_len = 0; + } + + switch (objdump_line__print(pos, sym, evidx, len, min_pcnt, + printed, max_lines, queue)) { + case 0: + ++printed; + if (context) { + printed += queue_len; + queue = NULL; + queue_len = 0; + } + break; + case 1: + /* filtered by max_lines */ + ++more; + break; + case -1: + default: + /* + * Filtered by min_pcnt or non IP lines when + * context != 0 + */ + if (!context) + break; + if (queue_len == context) + queue = list_entry(queue->node.next, typeof(*queue), node); + else + ++queue_len; + break; + } + } + + return more; +} + +void symbol__annotate_zero_histogram(struct symbol *sym, int evidx) +{ + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = annotation__histogram(notes, evidx); + + memset(h, 0, notes->src->sizeof_sym_hist); +} + +void symbol__annotate_decay_histogram(struct symbol *sym, int evidx) +{ + struct annotation *notes = symbol__annotation(sym); + struct sym_hist *h = annotation__histogram(notes, evidx); + struct objdump_line *pos; + + h->sum = 0; + + list_for_each_entry(pos, ¬es->src->source, node) { + if (pos->offset != -1) { + h->addr[pos->offset] = h->addr[pos->offset] * 7 / 8; + h->sum += h->addr[pos->offset]; + } + } +} + +void objdump_line_list__purge(struct list_head *head) +{ + struct objdump_line *pos, *n; + + list_for_each_entry_safe(pos, n, head, node) { + list_del(&pos->node); + objdump_line__free(pos); + } +} + +int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, + bool print_lines, bool full_paths, int min_pcnt, + int max_lines) +{ + struct dso *dso = map->dso; + const char *filename = dso->long_name; + struct rb_root source_line = RB_ROOT; + u64 len; + + if (symbol__annotate(sym, map, 0) < 0) + return -1; + + len = sym->end - sym->start; + + if (print_lines) { + symbol__get_source_line(sym, map, evidx, &source_line, + len, filename); + print_summary(&source_line, filename); + } + + symbol__annotate_printf(sym, map, evidx, full_paths, + min_pcnt, max_lines, 0); + if (print_lines) + symbol__free_source_line(sym, len); + + objdump_line_list__purge(&symbol__annotation(sym)->src->source); + + return 0; +} diff --git a/tools/perf/util/annotate.h b/tools/perf/util/annotate.h new file mode 100644 index 000000000000..e848803fcd48 --- /dev/null +++ b/tools/perf/util/annotate.h @@ -0,0 +1,101 @@ +#ifndef __PERF_ANNOTATE_H +#define __PERF_ANNOTATE_H + +#include <stdbool.h> +#include "types.h" +#include "symbol.h" +#include <linux/list.h> +#include <linux/rbtree.h> + +struct objdump_line { + struct list_head node; + s64 offset; + char *line; +}; + +void objdump_line__free(struct objdump_line *self); +struct objdump_line *objdump__get_next_ip_line(struct list_head *head, + struct objdump_line *pos); + +struct sym_hist { + u64 sum; + u64 addr[0]; +}; + +struct source_line { + struct rb_node node; + double percent; + char *path; +}; + +/** struct annotated_source - symbols with hits have this attached as in sannotation + * + * @histogram: Array of addr hit histograms per event being monitored + * @lines: If 'print_lines' is specified, per source code line percentages + * @source: source parsed from objdump -dS + * + * lines is allocated, percentages calculated and all sorted by percentage + * when the annotation is about to be presented, so the percentages are for + * one of the entries in the histogram array, i.e. for the event/counter being + * presented. It is deallocated right after symbol__{tui,tty,etc}_annotate + * returns. + */ +struct annotated_source { + struct list_head source; + struct source_line *lines; + int nr_histograms; + int sizeof_sym_hist; + struct sym_hist histograms[0]; +}; + +struct annotation { + pthread_mutex_t lock; + struct annotated_source *src; +}; + +struct sannotation { + struct annotation annotation; + struct symbol symbol; +}; + +static inline struct sym_hist *annotation__histogram(struct annotation *notes, int idx) +{ + return (((void *)¬es->src->histograms) + + (notes->src->sizeof_sym_hist * idx)); +} + +static inline struct annotation *symbol__annotation(struct symbol *sym) +{ + struct sannotation *a = container_of(sym, struct sannotation, symbol); + return &a->annotation; +} + +int symbol__inc_addr_samples(struct symbol *sym, struct map *map, + int evidx, u64 addr); +int symbol__alloc_hist(struct symbol *sym, int nevents); +void symbol__annotate_zero_histograms(struct symbol *sym); + +int symbol__annotate(struct symbol *sym, struct map *map, size_t privsize); +int symbol__annotate_init(struct map *map __used, struct symbol *sym); +int symbol__annotate_printf(struct symbol *sym, struct map *map, int evidx, + bool full_paths, int min_pcnt, int max_lines, + int context); +void symbol__annotate_zero_histogram(struct symbol *sym, int evidx); +void symbol__annotate_decay_histogram(struct symbol *sym, int evidx); +void objdump_line_list__purge(struct list_head *head); + +int symbol__tty_annotate(struct symbol *sym, struct map *map, int evidx, + bool print_lines, bool full_paths, int min_pcnt, + int max_lines); + +#ifdef NO_NEWT_SUPPORT +static inline int symbol__tui_annotate(struct symbol *sym __used, + struct map *map __used, int evidx __used) +{ + return 0; +} +#else +int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx); +#endif + +#endif /* __PERF_ANNOTATE_H */ diff --git a/tools/perf/util/build-id.c b/tools/perf/util/build-id.c index deffb8c96071..31f934af9861 100644 --- a/tools/perf/util/build-id.c +++ b/tools/perf/util/build-id.c @@ -14,8 +14,8 @@ #include <linux/kernel.h> #include "debug.h" -static int build_id__mark_dso_hit(event_t *event, - struct sample_data *sample __used, +static int build_id__mark_dso_hit(union perf_event *event, + struct perf_sample *sample __used, struct perf_session *session) { struct addr_location al; @@ -37,13 +37,14 @@ static int build_id__mark_dso_hit(event_t *event, return 0; } -static int event__exit_del_thread(event_t *self, struct sample_data *sample __used, - struct perf_session *session) +static int perf_event__exit_del_thread(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { - struct thread *thread = perf_session__findnew(session, self->fork.tid); + struct thread *thread = perf_session__findnew(session, event->fork.tid); - dump_printf("(%d:%d):(%d:%d)\n", self->fork.pid, self->fork.tid, - self->fork.ppid, self->fork.ptid); + 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->threads); @@ -56,9 +57,9 @@ static int event__exit_del_thread(event_t *self, struct sample_data *sample __us struct perf_event_ops build_id__mark_dso_hit_ops = { .sample = build_id__mark_dso_hit, - .mmap = event__process_mmap, - .fork = event__process_task, - .exit = event__exit_del_thread, + .mmap = perf_event__process_mmap, + .fork = perf_event__process_task, + .exit = perf_event__exit_del_thread, }; char *dso__build_id_filename(struct dso *self, char *bf, size_t size) diff --git a/tools/perf/util/cache.h b/tools/perf/util/cache.h index a7729797fd96..fc5e5a09d5b9 100644 --- a/tools/perf/util/cache.h +++ b/tools/perf/util/cache.h @@ -34,13 +34,14 @@ extern int pager_use_color; extern int use_browser; #ifdef NO_NEWT_SUPPORT -static inline void setup_browser(void) +static inline void setup_browser(bool fallback_to_pager) { - setup_pager(); + if (fallback_to_pager) + setup_pager(); } static inline void exit_browser(bool wait_for_ok __used) {} #else -void setup_browser(void); +void setup_browser(bool fallback_to_pager); void exit_browser(bool wait_for_ok); #endif diff --git a/tools/perf/util/callchain.c b/tools/perf/util/callchain.c index e12d539417b2..9f7106a8d9a4 100644 --- a/tools/perf/util/callchain.c +++ b/tools/perf/util/callchain.c @@ -1,5 +1,5 @@ /* - * Copyright (C) 2009-2010, Frederic Weisbecker <fweisbec@gmail.com> + * Copyright (C) 2009-2011, Frederic Weisbecker <fweisbec@gmail.com> * * Handle the callchains from the stream in an ad-hoc radix tree and then * sort them in an rbtree. @@ -18,7 +18,8 @@ #include "util.h" #include "callchain.h" -bool ip_callchain__valid(struct ip_callchain *chain, const event_t *event) +bool ip_callchain__valid(struct ip_callchain *chain, + const union perf_event *event) { unsigned int chain_size = event->header.size; chain_size -= (unsigned long)&event->ip.__more_data - (unsigned long)event; @@ -26,10 +27,10 @@ bool ip_callchain__valid(struct ip_callchain *chain, const event_t *event) } #define chain_for_each_child(child, parent) \ - list_for_each_entry(child, &parent->children, brothers) + list_for_each_entry(child, &parent->children, siblings) #define chain_for_each_child_safe(child, next, parent) \ - list_for_each_entry_safe(child, next, &parent->children, brothers) + list_for_each_entry_safe(child, next, &parent->children, siblings) static void rb_insert_callchain(struct rb_root *root, struct callchain_node *chain, @@ -38,14 +39,14 @@ rb_insert_callchain(struct rb_root *root, struct callchain_node *chain, struct rb_node **p = &root->rb_node; struct rb_node *parent = NULL; struct callchain_node *rnode; - u64 chain_cumul = cumul_hits(chain); + u64 chain_cumul = callchain_cumul_hits(chain); while (*p) { u64 rnode_cumul; parent = *p; rnode = rb_entry(parent, struct callchain_node, rb_node); - rnode_cumul = cumul_hits(rnode); + rnode_cumul = callchain_cumul_hits(rnode); switch (mode) { case CHAIN_FLAT: @@ -104,7 +105,7 @@ static void __sort_chain_graph_abs(struct callchain_node *node, chain_for_each_child(child, node) { __sort_chain_graph_abs(child, min_hit); - if (cumul_hits(child) >= min_hit) + if (callchain_cumul_hits(child) >= min_hit) rb_insert_callchain(&node->rb_root, child, CHAIN_GRAPH_ABS); } @@ -129,7 +130,7 @@ static void __sort_chain_graph_rel(struct callchain_node *node, chain_for_each_child(child, node) { __sort_chain_graph_rel(child, min_percent); - if (cumul_hits(child) >= min_hit) + if (callchain_cumul_hits(child) >= min_hit) rb_insert_callchain(&node->rb_root, child, CHAIN_GRAPH_REL); } @@ -143,7 +144,7 @@ sort_chain_graph_rel(struct rb_root *rb_root, struct callchain_root *chain_root, rb_root->rb_node = chain_root->node.rb_root.rb_node; } -int register_callchain_param(struct callchain_param *param) +int callchain_register_param(struct callchain_param *param) { switch (param->mode) { case CHAIN_GRAPH_ABS: @@ -189,32 +190,27 @@ create_child(struct callchain_node *parent, bool inherit_children) chain_for_each_child(next, new) next->parent = new; } - list_add_tail(&new->brothers, &parent->children); + list_add_tail(&new->siblings, &parent->children); return new; } -struct resolved_ip { - u64 ip; - struct map_symbol ms; -}; - -struct resolved_chain { - u64 nr; - struct resolved_ip ips[0]; -}; - - /* * Fill the node with callchain values */ static void -fill_node(struct callchain_node *node, struct resolved_chain *chain, int start) +fill_node(struct callchain_node *node, struct callchain_cursor *cursor) { - unsigned int i; + struct callchain_cursor_node *cursor_node; + + node->val_nr = cursor->nr - cursor->pos; + if (!node->val_nr) + pr_warning("Warning: empty node in callchain tree\n"); - for (i = start; i < chain->nr; i++) { + cursor_node = callchain_cursor_current(cursor); + + while (cursor_node) { struct callchain_list *call; call = zalloc(sizeof(*call)); @@ -222,23 +218,25 @@ fill_node(struct callchain_node *node, struct resolved_chain *chain, int start) perror("not enough memory for the code path tree"); return; } - call->ip = chain->ips[i].ip; - call->ms = chain->ips[i].ms; + call->ip = cursor_node->ip; + call->ms.sym = cursor_node->sym; + call->ms.map = cursor_node->map; list_add_tail(&call->list, &node->val); + + callchain_cursor_advance(cursor); + cursor_node = callchain_cursor_current(cursor); } - node->val_nr = chain->nr - start; - if (!node->val_nr) - pr_warning("Warning: empty node in callchain tree\n"); } static void -add_child(struct callchain_node *parent, struct resolved_chain *chain, - int start, u64 period) +add_child(struct callchain_node *parent, + struct callchain_cursor *cursor, + u64 period) { struct callchain_node *new; new = create_child(parent, false); - fill_node(new, chain, start); + fill_node(new, cursor); new->children_hit = 0; new->hit = period; @@ -250,9 +248,10 @@ add_child(struct callchain_node *parent, struct resolved_chain *chain, * Then create another child to host the given callchain of new branch */ static void -split_add_child(struct callchain_node *parent, struct resolved_chain *chain, - struct callchain_list *to_split, int idx_parents, int idx_local, - u64 period) +split_add_child(struct callchain_node *parent, + struct callchain_cursor *cursor, + struct callchain_list *to_split, + u64 idx_parents, u64 idx_local, u64 period) { struct callchain_node *new; struct list_head *old_tail; @@ -272,14 +271,14 @@ split_add_child(struct callchain_node *parent, struct resolved_chain *chain, /* split the hits */ new->hit = parent->hit; new->children_hit = parent->children_hit; - parent->children_hit = cumul_hits(new); + parent->children_hit = callchain_cumul_hits(new); new->val_nr = parent->val_nr - idx_local; parent->val_nr = idx_local; /* create a new child for the new branch if any */ - if (idx_total < chain->nr) { + if (idx_total < cursor->nr) { parent->hit = 0; - add_child(parent, chain, idx_total, period); + add_child(parent, cursor, period); parent->children_hit += period; } else { parent->hit = period; @@ -287,36 +286,41 @@ split_add_child(struct callchain_node *parent, struct resolved_chain *chain, } static int -append_chain(struct callchain_node *root, struct resolved_chain *chain, - unsigned int start, u64 period); +append_chain(struct callchain_node *root, + struct callchain_cursor *cursor, + u64 period); static void -append_chain_children(struct callchain_node *root, struct resolved_chain *chain, - unsigned int start, u64 period) +append_chain_children(struct callchain_node *root, + struct callchain_cursor *cursor, + u64 period) { struct callchain_node *rnode; /* lookup in childrens */ chain_for_each_child(rnode, root) { - unsigned int ret = append_chain(rnode, chain, start, period); + unsigned int ret = append_chain(rnode, cursor, period); if (!ret) goto inc_children_hit; } /* nothing in children, add to the current node */ - add_child(root, chain, start, period); + add_child(root, cursor, period); inc_children_hit: root->children_hit += period; } static int -append_chain(struct callchain_node *root, struct resolved_chain *chain, - unsigned int start, u64 period) +append_chain(struct callchain_node *root, + struct callchain_cursor *cursor, + u64 period) { + struct callchain_cursor_node *curr_snap = cursor->curr; struct callchain_list *cnode; - unsigned int i = start; + u64 start = cursor->pos; bool found = false; + u64 matches; /* * Lookup in the current node @@ -324,141 +328,134 @@ append_chain(struct callchain_node *root, struct resolved_chain *chain, * anywhere inside a function. */ list_for_each_entry(cnode, &root->val, list) { + struct callchain_cursor_node *node; struct symbol *sym; - if (i == chain->nr) + node = callchain_cursor_current(cursor); + if (!node) break; - sym = chain->ips[i].ms.sym; + sym = node->sym; if (cnode->ms.sym && sym) { if (cnode->ms.sym->start != sym->start) break; - } else if (cnode->ip != chain->ips[i].ip) + } else if (cnode->ip != node->ip) break; if (!found) found = true; - i++; + + callchain_cursor_advance(cursor); } /* matches not, relay on the parent */ - if (!found) + if (!found) { + cursor->curr = curr_snap; + cursor->pos = start; return -1; + } + + matches = cursor->pos - start; /* we match only a part of the node. Split it and add the new chain */ - if (i - start < root->val_nr) { - split_add_child(root, chain, cnode, start, i - start, period); + if (matches < root->val_nr) { + split_add_child(root, cursor, cnode, start, matches, period); return 0; } /* we match 100% of the path, increment the hit */ - if (i - start == root->val_nr && i == chain->nr) { + if (matches == root->val_nr && cursor->pos == cursor->nr) { root->hit += period; return 0; } /* We match the node and still have a part remaining */ - append_chain_children(root, chain, i, period); + append_chain_children(root, cursor, period); return 0; } -static void filter_context(struct ip_callchain *old, struct resolved_chain *new, - struct map_symbol *syms) -{ - int i, j = 0; - - for (i = 0; i < (int)old->nr; i++) { - if (old->ips[i] >= PERF_CONTEXT_MAX) - continue; - - new->ips[j].ip = old->ips[i]; - new->ips[j].ms = syms[i]; - j++; - } - - new->nr = j; -} - - -int callchain_append(struct callchain_root *root, struct ip_callchain *chain, - struct map_symbol *syms, u64 period) +int callchain_append(struct callchain_root *root, + struct callchain_cursor *cursor, + u64 period) { - struct resolved_chain *filtered; - - if (!chain->nr) + if (!cursor->nr) return 0; - filtered = zalloc(sizeof(*filtered) + - chain->nr * sizeof(struct resolved_ip)); - if (!filtered) - return -ENOMEM; - - filter_context(chain, filtered, syms); - - if (!filtered->nr) - goto end; + callchain_cursor_commit(cursor); - append_chain_children(&root->node, filtered, 0, period); + append_chain_children(&root->node, cursor, period); - if (filtered->nr > root->max_depth) - root->max_depth = filtered->nr; -end: - free(filtered); + if (cursor->nr > root->max_depth) + root->max_depth = cursor->nr; return 0; } static int -merge_chain_branch(struct callchain_node *dst, struct callchain_node *src, - struct resolved_chain *chain) +merge_chain_branch(struct callchain_cursor *cursor, + struct callchain_node *dst, struct callchain_node *src) { + struct callchain_cursor_node **old_last = cursor->last; struct callchain_node *child, *next_child; struct callchain_list *list, *next_list; - int old_pos = chain->nr; + int old_pos = cursor->nr; int err = 0; list_for_each_entry_safe(list, next_list, &src->val, list) { - chain->ips[chain->nr].ip = list->ip; - chain->ips[chain->nr].ms = list->ms; - chain->nr++; + callchain_cursor_append(cursor, list->ip, + list->ms.map, list->ms.sym); list_del(&list->list); free(list); } - if (src->hit) - append_chain_children(dst, chain, 0, src->hit); + if (src->hit) { + callchain_cursor_commit(cursor); + append_chain_children(dst, cursor, src->hit); + } chain_for_each_child_safe(child, next_child, src) { - err = merge_chain_branch(dst, child, chain); + err = merge_chain_branch(cursor, dst, child); if (err) break; - list_del(&child->brothers); + list_del(&child->siblings); free(child); } - chain->nr = old_pos; + cursor->nr = old_pos; + cursor->last = old_last; return err; } -int callchain_merge(struct callchain_root *dst, struct callchain_root *src) +int callchain_merge(struct callchain_cursor *cursor, + struct callchain_root *dst, struct callchain_root *src) +{ + return merge_chain_branch(cursor, &dst->node, &src->node); +} + +int callchain_cursor_append(struct callchain_cursor *cursor, + u64 ip, struct map *map, struct symbol *sym) { - struct resolved_chain *chain; - int err; + struct callchain_cursor_node *node = *cursor->last; - chain = malloc(sizeof(*chain) + - src->max_depth * sizeof(struct resolved_ip)); - if (!chain) - return -ENOMEM; + if (!node) { + node = calloc(sizeof(*node), 1); + if (!node) + return -ENOMEM; - chain->nr = 0; + *cursor->last = node; + } - err = merge_chain_branch(&dst->node, &src->node, chain); + node->ip = ip; + node->map = map; + node->sym = sym; - free(chain); + cursor->nr++; - return err; + cursor->last = &node->next; + + return 0; } diff --git a/tools/perf/util/callchain.h b/tools/perf/util/callchain.h index c15fb8c24ad2..1a79df9f739f 100644 --- a/tools/perf/util/callchain.h +++ b/tools/perf/util/callchain.h @@ -16,7 +16,7 @@ enum chain_mode { struct callchain_node { struct callchain_node *parent; - struct list_head brothers; + struct list_head siblings; struct list_head children; struct list_head val; struct rb_node rb_node; /* to sort nodes in an rbtree */ @@ -49,9 +49,30 @@ struct callchain_list { struct list_head list; }; +/* + * A callchain cursor is a single linked list that + * let one feed a callchain progressively. + * It keeps persitent allocated entries to minimize + * allocations. + */ +struct callchain_cursor_node { + u64 ip; + struct map *map; + struct symbol *sym; + struct callchain_cursor_node *next; +}; + +struct callchain_cursor { + u64 nr; + struct callchain_cursor_node *first; + struct callchain_cursor_node **last; + u64 pos; + struct callchain_cursor_node *curr; +}; + static inline void callchain_init(struct callchain_root *root) { - INIT_LIST_HEAD(&root->node.brothers); + INIT_LIST_HEAD(&root->node.siblings); INIT_LIST_HEAD(&root->node.children); INIT_LIST_HEAD(&root->node.val); @@ -61,15 +82,54 @@ static inline void callchain_init(struct callchain_root *root) root->max_depth = 0; } -static inline u64 cumul_hits(struct callchain_node *node) +static inline u64 callchain_cumul_hits(struct callchain_node *node) { return node->hit + node->children_hit; } -int register_callchain_param(struct callchain_param *param); -int callchain_append(struct callchain_root *root, struct ip_callchain *chain, - struct map_symbol *syms, u64 period); -int callchain_merge(struct callchain_root *dst, struct callchain_root *src); +int callchain_register_param(struct callchain_param *param); +int callchain_append(struct callchain_root *root, + struct callchain_cursor *cursor, + u64 period); + +int callchain_merge(struct callchain_cursor *cursor, + struct callchain_root *dst, struct callchain_root *src); + +bool ip_callchain__valid(struct ip_callchain *chain, + const union perf_event *event); +/* + * Initialize a cursor before adding entries inside, but keep + * the previously allocated entries as a cache. + */ +static inline void callchain_cursor_reset(struct callchain_cursor *cursor) +{ + cursor->nr = 0; + cursor->last = &cursor->first; +} + +int callchain_cursor_append(struct callchain_cursor *cursor, u64 ip, + struct map *map, struct symbol *sym); -bool ip_callchain__valid(struct ip_callchain *chain, const event_t *event); +/* Close a cursor writing session. Initialize for the reader */ +static inline void callchain_cursor_commit(struct callchain_cursor *cursor) +{ + cursor->curr = cursor->first; + cursor->pos = 0; +} + +/* Cursor reading iteration helpers */ +static inline struct callchain_cursor_node * +callchain_cursor_current(struct callchain_cursor *cursor) +{ + if (cursor->pos == cursor->nr) + return NULL; + + return cursor->curr; +} + +static inline void callchain_cursor_advance(struct callchain_cursor *cursor) +{ + cursor->curr = cursor->curr->next; + cursor->pos++; +} #endif /* __PERF_CALLCHAIN_H */ diff --git a/tools/perf/util/cpumap.c b/tools/perf/util/cpumap.c index 3ccaa1043383..6893eec693ab 100644 --- a/tools/perf/util/cpumap.c +++ b/tools/perf/util/cpumap.c @@ -177,3 +177,8 @@ struct cpu_map *cpu_map__dummy_new(void) return cpus; } + +void cpu_map__delete(struct cpu_map *map) +{ + free(map); +} diff --git a/tools/perf/util/cpumap.h b/tools/perf/util/cpumap.h index f7a4f42f6307..072c0a374794 100644 --- a/tools/perf/util/cpumap.h +++ b/tools/perf/util/cpumap.h @@ -8,6 +8,6 @@ struct cpu_map { struct cpu_map *cpu_map__new(const char *cpu_list); struct cpu_map *cpu_map__dummy_new(void); -void *cpu_map__delete(struct cpu_map *map); +void cpu_map__delete(struct cpu_map *map); #endif /* __PERF_CPUMAP_H */ diff --git a/tools/perf/util/debug.c b/tools/perf/util/debug.c index 01bbe8ecec3f..d4536a9e0d8c 100644 --- a/tools/perf/util/debug.c +++ b/tools/perf/util/debug.c @@ -57,7 +57,7 @@ void ui__warning(const char *format, ...) } #endif -void trace_event(event_t *event) +void trace_event(union perf_event *event) { unsigned char *raw_event = (void *)event; const char *color = PERF_COLOR_BLUE; diff --git a/tools/perf/util/debug.h b/tools/perf/util/debug.h index ca35fd66b5df..93516cf4682c 100644 --- a/tools/perf/util/debug.h +++ b/tools/perf/util/debug.h @@ -9,7 +9,7 @@ extern int verbose; extern bool quiet, dump_trace; int dump_printf(const char *fmt, ...) __attribute__((format(printf, 1, 2))); -void trace_event(event_t *event); +void trace_event(union perf_event *event); struct ui_progress; diff --git a/tools/perf/util/event.c b/tools/perf/util/event.c index 50d0a931497a..fbf5754c8866 100644 --- a/tools/perf/util/event.c +++ b/tools/perf/util/event.c @@ -6,8 +6,9 @@ #include "string.h" #include "strlist.h" #include "thread.h" +#include "thread_map.h" -static const char *event__name[] = { +static const char *perf_event__names[] = { [0] = "TOTAL", [PERF_RECORD_MMAP] = "MMAP", [PERF_RECORD_LOST] = "LOST", @@ -25,16 +26,16 @@ static const char *event__name[] = { [PERF_RECORD_FINISHED_ROUND] = "FINISHED_ROUND", }; -const char *event__get_event_name(unsigned int id) +const char *perf_event__name(unsigned int id) { - if (id >= ARRAY_SIZE(event__name)) + if (id >= ARRAY_SIZE(perf_event__names)) return "INVALID"; - if (!event__name[id]) + if (!perf_event__names[id]) return "UNKNOWN"; - return event__name[id]; + return perf_event__names[id]; } -static struct sample_data synth_sample = { +static struct perf_sample synth_sample = { .pid = -1, .tid = -1, .time = -1, @@ -43,9 +44,9 @@ static struct sample_data synth_sample = { .period = 1, }; -static pid_t event__synthesize_comm(event_t *event, pid_t pid, int full, - event__handler_t process, - struct perf_session *session) +static pid_t perf_event__synthesize_comm(union perf_event *event, pid_t pid, + int full, perf_event__handler_t process, + struct perf_session *session) { char filename[PATH_MAX]; char bf[BUFSIZ]; @@ -126,9 +127,10 @@ out: return tgid; } -static int event__synthesize_mmap_events(event_t *event, pid_t pid, pid_t tgid, - event__handler_t process, - struct perf_session *session) +static int perf_event__synthesize_mmap_events(union perf_event *event, + pid_t pid, pid_t tgid, + perf_event__handler_t process, + struct perf_session *session) { char filename[PATH_MAX]; FILE *fp; @@ -199,14 +201,14 @@ static int event__synthesize_mmap_events(event_t *event, pid_t pid, pid_t tgid, return 0; } -int event__synthesize_modules(event__handler_t process, - struct perf_session *session, - struct machine *machine) +int perf_event__synthesize_modules(perf_event__handler_t process, + struct perf_session *session, + struct machine *machine) { struct rb_node *nd; struct map_groups *kmaps = &machine->kmaps; - event_t *event = zalloc(sizeof(event->mmap) + session->id_hdr_size); - + union perf_event *event = zalloc((sizeof(event->mmap) + + session->id_hdr_size)); if (event == NULL) { pr_debug("Not enough memory synthesizing mmap event " "for kernel modules\n"); @@ -251,23 +253,24 @@ int event__synthesize_modules(event__handler_t process, return 0; } -static int __event__synthesize_thread(event_t *comm_event, event_t *mmap_event, - pid_t pid, 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_session *session) { - pid_t tgid = event__synthesize_comm(comm_event, pid, 1, process, + pid_t tgid = perf_event__synthesize_comm(comm_event, pid, 1, process, session); if (tgid == -1) return -1; - return event__synthesize_mmap_events(mmap_event, pid, tgid, + return perf_event__synthesize_mmap_events(mmap_event, pid, tgid, process, session); } -int event__synthesize_thread_map(struct thread_map *threads, - event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_thread_map(struct thread_map *threads, + perf_event__handler_t process, + struct perf_session *session) { - event_t *comm_event, *mmap_event; + union perf_event *comm_event, *mmap_event; int err = -1, thread; comm_event = malloc(sizeof(comm_event->comm) + session->id_hdr_size); @@ -294,12 +297,12 @@ out: return err; } -int event__synthesize_threads(event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_threads(perf_event__handler_t process, + struct perf_session *session) { DIR *proc; struct dirent dirent, *next; - event_t *comm_event, *mmap_event; + union perf_event *comm_event, *mmap_event; int err = -1; comm_event = malloc(sizeof(comm_event->comm) + session->id_hdr_size); @@ -357,10 +360,10 @@ static int find_symbol_cb(void *arg, const char *name, char type, return 1; } -int event__synthesize_kernel_mmap(event__handler_t process, - struct perf_session *session, - struct machine *machine, - const char *symbol_name) +int perf_event__synthesize_kernel_mmap(perf_event__handler_t process, + struct perf_session *session, + struct machine *machine, + const char *symbol_name) { size_t size; const char *filename, *mmap_name; @@ -374,8 +377,8 @@ int event__synthesize_kernel_mmap(event__handler_t process, * kernels. */ struct process_symbol_args args = { .name = symbol_name, }; - event_t *event = zalloc(sizeof(event->mmap) + session->id_hdr_size); - + union perf_event *event = zalloc((sizeof(event->mmap) + + session->id_hdr_size)); if (event == NULL) { pr_debug("Not enough memory synthesizing mmap event " "for kernel modules\n"); @@ -448,14 +451,15 @@ static int thread__set_comm_adjust(struct thread *self, const char *comm, return 0; } -int event__process_comm(event_t *self, struct sample_data *sample __used, - struct perf_session *session) +int perf_event__process_comm(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { - struct thread *thread = perf_session__findnew(session, self->comm.tid); + struct thread *thread = perf_session__findnew(session, event->comm.tid); - dump_printf(": %s:%d\n", self->comm.comm, self->comm.tid); + dump_printf(": %s:%d\n", event->comm.comm, event->comm.tid); - if (thread == NULL || thread__set_comm_adjust(thread, self->comm.comm, + if (thread == NULL || thread__set_comm_adjust(thread, event->comm.comm, &session->hists)) { dump_printf("problem processing PERF_RECORD_COMM, skipping event.\n"); return -1; @@ -464,19 +468,21 @@ int event__process_comm(event_t *self, struct sample_data *sample __used, return 0; } -int event__process_lost(event_t *self, struct sample_data *sample __used, - struct perf_session *session) +int perf_event__process_lost(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { dump_printf(": id:%" PRIu64 ": lost:%" PRIu64 "\n", - self->lost.id, self->lost.lost); - session->hists.stats.total_lost += self->lost.lost; + event->lost.id, event->lost.lost); + session->hists.stats.total_lost += event->lost.lost; return 0; } -static void event_set_kernel_mmap_len(struct map **maps, event_t *self) +static void perf_event__set_kernel_mmap_len(union perf_event *event, + struct map **maps) { - maps[MAP__FUNCTION]->start = self->mmap.start; - maps[MAP__FUNCTION]->end = self->mmap.start + self->mmap.len; + maps[MAP__FUNCTION]->start = event->mmap.start; + maps[MAP__FUNCTION]->end = event->mmap.start + event->mmap.len; /* * Be a bit paranoid here, some perf.data file came with * a zero sized synthesized MMAP event for the kernel. @@ -485,8 +491,8 @@ static void event_set_kernel_mmap_len(struct map **maps, event_t *self) maps[MAP__FUNCTION]->end = ~0ULL; } -static int event__process_kernel_mmap(event_t *self, - struct perf_session *session) +static int perf_event__process_kernel_mmap(union perf_event *event, + struct perf_session *session) { struct map *map; char kmmap_prefix[PATH_MAX]; @@ -494,9 +500,9 @@ static int event__process_kernel_mmap(event_t *self, enum dso_kernel_type kernel_type; bool is_kernel_mmap; - machine = perf_session__findnew_machine(session, self->mmap.pid); + machine = perf_session__findnew_machine(session, event->mmap.pid); if (!machine) { - pr_err("Can't find id %d's machine\n", self->mmap.pid); + pr_err("Can't find id %d's machine\n", event->mmap.pid); goto out_problem; } @@ -506,17 +512,17 @@ static int event__process_kernel_mmap(event_t *self, else kernel_type = DSO_TYPE_GUEST_KERNEL; - is_kernel_mmap = memcmp(self->mmap.filename, + is_kernel_mmap = memcmp(event->mmap.filename, kmmap_prefix, strlen(kmmap_prefix)) == 0; - if (self->mmap.filename[0] == '/' || - (!is_kernel_mmap && self->mmap.filename[0] == '[')) { + if (event->mmap.filename[0] == '/' || + (!is_kernel_mmap && event->mmap.filename[0] == '[')) { char short_module_name[1024]; char *name, *dot; - if (self->mmap.filename[0] == '/') { - name = strrchr(self->mmap.filename, '/'); + if (event->mmap.filename[0] == '/') { + name = strrchr(event->mmap.filename, '/'); if (name == NULL) goto out_problem; @@ -528,10 +534,10 @@ static int event__process_kernel_mmap(event_t *self, "[%.*s]", (int)(dot - name), name); strxfrchar(short_module_name, '-', '_'); } else - strcpy(short_module_name, self->mmap.filename); + strcpy(short_module_name, event->mmap.filename); - map = machine__new_module(machine, self->mmap.start, - self->mmap.filename); + map = machine__new_module(machine, event->mmap.start, + event->mmap.filename); if (map == NULL) goto out_problem; @@ -541,9 +547,9 @@ static int event__process_kernel_mmap(event_t *self, map->dso->short_name = name; map->dso->sname_alloc = 1; - map->end = map->start + self->mmap.len; + map->end = map->start + event->mmap.len; } else if (is_kernel_mmap) { - const char *symbol_name = (self->mmap.filename + + const char *symbol_name = (event->mmap.filename + strlen(kmmap_prefix)); /* * Should be there already, from the build-id table in @@ -558,10 +564,10 @@ static int event__process_kernel_mmap(event_t *self, if (__machine__create_kernel_maps(machine, kernel) < 0) goto out_problem; - event_set_kernel_mmap_len(machine->vmlinux_maps, self); + perf_event__set_kernel_mmap_len(event, machine->vmlinux_maps); perf_session__set_kallsyms_ref_reloc_sym(machine->vmlinux_maps, symbol_name, - self->mmap.pgoff); + event->mmap.pgoff); if (machine__is_default_guest(machine)) { /* * preload dso of guest kernel and modules @@ -575,22 +581,23 @@ out_problem: return -1; } -int event__process_mmap(event_t *self, struct sample_data *sample __used, - struct perf_session *session) +int perf_event__process_mmap(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { struct machine *machine; struct thread *thread; struct map *map; - u8 cpumode = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; int ret = 0; dump_printf(" %d/%d: [%#" PRIx64 "(%#" PRIx64 ") @ %#" PRIx64 "]: %s\n", - self->mmap.pid, self->mmap.tid, self->mmap.start, - self->mmap.len, self->mmap.pgoff, self->mmap.filename); + event->mmap.pid, event->mmap.tid, event->mmap.start, + event->mmap.len, event->mmap.pgoff, event->mmap.filename); if (cpumode == PERF_RECORD_MISC_GUEST_KERNEL || cpumode == PERF_RECORD_MISC_KERNEL) { - ret = event__process_kernel_mmap(self, session); + ret = perf_event__process_kernel_mmap(event, session); if (ret < 0) goto out_problem; return 0; @@ -599,12 +606,12 @@ int event__process_mmap(event_t *self, struct sample_data *sample __used, machine = perf_session__find_host_machine(session); if (machine == NULL) goto out_problem; - thread = perf_session__findnew(session, self->mmap.pid); + thread = perf_session__findnew(session, event->mmap.pid); if (thread == NULL) goto out_problem; - map = map__new(&machine->user_dsos, self->mmap.start, - self->mmap.len, self->mmap.pgoff, - self->mmap.pid, self->mmap.filename, + map = map__new(&machine->user_dsos, event->mmap.start, + event->mmap.len, event->mmap.pgoff, + event->mmap.pid, event->mmap.filename, MAP__FUNCTION); if (map == NULL) goto out_problem; @@ -617,16 +624,17 @@ out_problem: return 0; } -int event__process_task(event_t *self, struct sample_data *sample __used, - struct perf_session *session) +int perf_event__process_task(union perf_event *event, + struct perf_sample *sample __used, + struct perf_session *session) { - struct thread *thread = perf_session__findnew(session, self->fork.tid); - struct thread *parent = perf_session__findnew(session, self->fork.ptid); + struct thread *thread = perf_session__findnew(session, event->fork.tid); + struct thread *parent = perf_session__findnew(session, event->fork.ptid); - dump_printf("(%d:%d):(%d:%d)\n", self->fork.pid, self->fork.tid, - self->fork.ppid, self->fork.ptid); + dump_printf("(%d:%d):(%d:%d)\n", event->fork.pid, event->fork.tid, + event->fork.ppid, event->fork.ptid); - if (self->header.type == PERF_RECORD_EXIT) { + if (event->header.type == PERF_RECORD_EXIT) { perf_session__remove_thread(session, thread); return 0; } @@ -640,20 +648,22 @@ int event__process_task(event_t *self, struct sample_data *sample __used, return 0; } -int event__process(event_t *event, struct sample_data *sample, - struct perf_session *session) +int perf_event__process(union perf_event *event, struct perf_sample *sample, + struct perf_session *session) { switch (event->header.type) { case PERF_RECORD_COMM: - event__process_comm(event, sample, session); + perf_event__process_comm(event, sample, session); break; case PERF_RECORD_MMAP: - event__process_mmap(event, sample, session); + perf_event__process_mmap(event, sample, session); break; case PERF_RECORD_FORK: case PERF_RECORD_EXIT: - event__process_task(event, sample, session); + perf_event__process_task(event, sample, session); break; + case PERF_RECORD_LOST: + perf_event__process_lost(event, sample, session); default: break; } @@ -762,12 +772,14 @@ static void dso__calc_col_width(struct dso *self, struct hists *hists) self->slen_calculated = 1; } -int event__preprocess_sample(const event_t *self, struct perf_session *session, - struct addr_location *al, struct sample_data *data, - symbol_filter_t filter) +int perf_event__preprocess_sample(const union perf_event *event, + struct perf_session *session, + struct addr_location *al, + struct perf_sample *sample, + symbol_filter_t filter) { - u8 cpumode = self->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; - struct thread *thread = perf_session__findnew(session, self->ip.pid); + u8 cpumode = event->header.misc & PERF_RECORD_MISC_CPUMODE_MASK; + struct thread *thread = perf_session__findnew(session, event->ip.pid); if (thread == NULL) return -1; @@ -789,12 +801,12 @@ int event__preprocess_sample(const event_t *self, struct perf_session *session, machine__create_kernel_maps(&session->host_machine); thread__find_addr_map(thread, session, cpumode, MAP__FUNCTION, - self->ip.pid, self->ip.ip, al); + event->ip.pid, event->ip.ip, al); dump_printf(" ...... dso: %s\n", al->map ? al->map->dso->long_name : al->level == 'H' ? "[hypervisor]" : "<not found>"); al->sym = NULL; - al->cpu = data->cpu; + al->cpu = sample->cpu; if (al->map) { if (symbol_conf.dso_list && @@ -834,128 +846,3 @@ out_filtered: al->filtered = true; return 0; } - -static int event__parse_id_sample(const event_t *event, - struct perf_session *session, - struct sample_data *sample) -{ - const u64 *array; - u64 type; - - sample->cpu = sample->pid = sample->tid = -1; - sample->stream_id = sample->id = sample->time = -1ULL; - - if (!session->sample_id_all) - return 0; - - array = event->sample.array; - array += ((event->header.size - - sizeof(event->header)) / sizeof(u64)) - 1; - type = session->sample_type; - - if (type & PERF_SAMPLE_CPU) { - u32 *p = (u32 *)array; - sample->cpu = *p; - array--; - } - - if (type & PERF_SAMPLE_STREAM_ID) { - sample->stream_id = *array; - array--; - } - - if (type & PERF_SAMPLE_ID) { - sample->id = *array; - array--; - } - - if (type & PERF_SAMPLE_TIME) { - sample->time = *array; - array--; - } - - if (type & PERF_SAMPLE_TID) { - u32 *p = (u32 *)array; - sample->pid = p[0]; - sample->tid = p[1]; - } - - return 0; -} - -int event__parse_sample(const event_t *event, struct perf_session *session, - struct sample_data *data) -{ - const u64 *array; - u64 type; - - if (event->header.type != PERF_RECORD_SAMPLE) - return event__parse_id_sample(event, session, data); - - array = event->sample.array; - type = session->sample_type; - - if (type & PERF_SAMPLE_IP) { - data->ip = event->ip.ip; - array++; - } - - if (type & PERF_SAMPLE_TID) { - u32 *p = (u32 *)array; - data->pid = p[0]; - data->tid = p[1]; - array++; - } - - if (type & PERF_SAMPLE_TIME) { - data->time = *array; - array++; - } - - if (type & PERF_SAMPLE_ADDR) { - data->addr = *array; - array++; - } - - data->id = -1ULL; - if (type & PERF_SAMPLE_ID) { - data->id = *array; - array++; - } - - if (type & PERF_SAMPLE_STREAM_ID) { - data->stream_id = *array; - array++; - } - - if (type & PERF_SAMPLE_CPU) { - u32 *p = (u32 *)array; - data->cpu = *p; - array++; - } else - data->cpu = -1; - - if (type & PERF_SAMPLE_PERIOD) { - data->period = *array; - array++; - } - - if (type & PERF_SAMPLE_READ) { - pr_debug("PERF_SAMPLE_READ is unsuported for now\n"); - return -1; - } - - if (type & PERF_SAMPLE_CALLCHAIN) { - data->callchain = (struct ip_callchain *)array; - array += 1 + data->callchain->nr; - } - - if (type & PERF_SAMPLE_RAW) { - u32 *p = (u32 *)array; - data->raw_size = *p; - p++; - data->raw_data = p; - } - - return 0; -} diff --git a/tools/perf/util/event.h b/tools/perf/util/event.h index cc7b52f9b492..9c35170fb379 100644 --- a/tools/perf/util/event.h +++ b/tools/perf/util/event.h @@ -61,7 +61,7 @@ struct sample_event { u64 array[]; }; -struct sample_data { +struct perf_sample { u64 ip; u32 pid, tid; u64 time; @@ -117,7 +117,7 @@ struct tracing_data_event { u32 size; }; -typedef union event_union { +union perf_event { struct perf_event_header header; struct ip_event ip; struct mmap_event mmap; @@ -130,50 +130,54 @@ typedef union event_union { struct event_type_event event_type; struct tracing_data_event tracing_data; struct build_id_event build_id; -} event_t; +}; -void event__print_totals(void); +void perf_event__print_totals(void); struct perf_session; struct thread_map; -typedef int (*event__handler_synth_t)(event_t *event, +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, + struct perf_sample *sample, struct perf_session *session); -typedef int (*event__handler_t)(event_t *event, struct sample_data *sample, - struct perf_session *session); - -int event__synthesize_thread_map(struct thread_map *threads, - event__handler_t process, - struct perf_session *session); -int event__synthesize_threads(event__handler_t process, - struct perf_session *session); -int event__synthesize_kernel_mmap(event__handler_t process, - struct perf_session *session, - struct machine *machine, - const char *symbol_name); - -int event__synthesize_modules(event__handler_t process, - struct perf_session *session, - struct machine *machine); - -int event__process_comm(event_t *self, struct sample_data *sample, - struct perf_session *session); -int event__process_lost(event_t *self, struct sample_data *sample, - struct perf_session *session); -int event__process_mmap(event_t *self, struct sample_data *sample, - struct perf_session *session); -int event__process_task(event_t *self, struct sample_data *sample, + +int perf_event__synthesize_thread_map(struct thread_map *threads, + perf_event__handler_t process, + struct perf_session *session); +int perf_event__synthesize_threads(perf_event__handler_t process, + struct perf_session *session); +int perf_event__synthesize_kernel_mmap(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, + struct perf_session *session, + struct machine *machine); + +int perf_event__process_comm(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, + struct perf_session *session); +int perf_event__process_mmap(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, + struct perf_session *session); +int perf_event__process(union perf_event *event, struct perf_sample *sample, struct perf_session *session); -int event__process(event_t *event, struct sample_data *sample, - struct perf_session *session); struct addr_location; -int event__preprocess_sample(const event_t *self, struct perf_session *session, - struct addr_location *al, struct sample_data *data, - symbol_filter_t filter); -int event__parse_sample(const event_t *event, struct perf_session *session, - struct sample_data *sample); +int perf_event__preprocess_sample(const union perf_event *self, + struct perf_session *session, + struct addr_location *al, + struct perf_sample *sample, + symbol_filter_t filter); + +const char *perf_event__name(unsigned int id); -const char *event__get_event_name(unsigned int id); +int perf_event__parse_sample(const union perf_event *event, u64 type, + bool sample_id_all, struct perf_sample *sample); #endif /* __PERF_RECORD_H */ diff --git a/tools/perf/util/evlist.c b/tools/perf/util/evlist.c new file mode 100644 index 000000000000..95b21fece2ce --- /dev/null +++ b/tools/perf/util/evlist.c @@ -0,0 +1,350 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Parts came from builtin-{top,stat,record}.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ +#include <poll.h> +#include "cpumap.h" +#include "thread_map.h" +#include "evlist.h" +#include "evsel.h" +#include "util.h" + +#include <sys/mman.h> + +#include <linux/bitops.h> +#include <linux/hash.h> + +#define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) +#define SID(e, x, y) xyarray__entry(e->id, x, y) + +void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus, + struct thread_map *threads) +{ + int i; + + for (i = 0; i < PERF_EVLIST__HLIST_SIZE; ++i) + INIT_HLIST_HEAD(&evlist->heads[i]); + INIT_LIST_HEAD(&evlist->entries); + perf_evlist__set_maps(evlist, cpus, threads); +} + +struct perf_evlist *perf_evlist__new(struct cpu_map *cpus, + struct thread_map *threads) +{ + struct perf_evlist *evlist = zalloc(sizeof(*evlist)); + + if (evlist != NULL) + perf_evlist__init(evlist, cpus, threads); + + return evlist; +} + +static void perf_evlist__purge(struct perf_evlist *evlist) +{ + struct perf_evsel *pos, *n; + + list_for_each_entry_safe(pos, n, &evlist->entries, node) { + list_del_init(&pos->node); + perf_evsel__delete(pos); + } + + evlist->nr_entries = 0; +} + +void perf_evlist__exit(struct perf_evlist *evlist) +{ + free(evlist->mmap); + free(evlist->pollfd); + evlist->mmap = NULL; + evlist->pollfd = NULL; +} + +void perf_evlist__delete(struct perf_evlist *evlist) +{ + perf_evlist__purge(evlist); + perf_evlist__exit(evlist); + free(evlist); +} + +void perf_evlist__add(struct perf_evlist *evlist, struct perf_evsel *entry) +{ + list_add_tail(&entry->node, &evlist->entries); + ++evlist->nr_entries; +} + +int perf_evlist__add_default(struct perf_evlist *evlist) +{ + struct perf_event_attr attr = { + .type = PERF_TYPE_HARDWARE, + .config = PERF_COUNT_HW_CPU_CYCLES, + }; + struct perf_evsel *evsel = perf_evsel__new(&attr, 0); + + if (evsel == NULL) + return -ENOMEM; + + perf_evlist__add(evlist, evsel); + return 0; +} + +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); + return evlist->pollfd != NULL ? 0 : -ENOMEM; +} + +void perf_evlist__add_pollfd(struct perf_evlist *evlist, int fd) +{ + fcntl(fd, F_SETFL, O_NONBLOCK); + evlist->pollfd[evlist->nr_fds].fd = fd; + evlist->pollfd[evlist->nr_fds].events = POLLIN; + evlist->nr_fds++; +} + +static int perf_evlist__id_hash(struct perf_evlist *evlist, struct perf_evsel *evsel, + int cpu, int thread, int fd) +{ + struct perf_sample_id *sid; + u64 read_data[4] = { 0, }; + int hash, id_idx = 1; /* The first entry is the counter value */ + + if (!(evsel->attr.read_format & PERF_FORMAT_ID) || + read(fd, &read_data, sizeof(read_data)) == -1) + return -1; + + if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_ENABLED) + ++id_idx; + if (evsel->attr.read_format & PERF_FORMAT_TOTAL_TIME_RUNNING) + ++id_idx; + + sid = SID(evsel, cpu, thread); + sid->id = read_data[id_idx]; + sid->evsel = evsel; + hash = hash_64(sid->id, PERF_EVLIST__HLIST_BITS); + hlist_add_head(&sid->node, &evlist->heads[hash]); + return 0; +} + +struct perf_evsel *perf_evlist__id2evsel(struct perf_evlist *evlist, u64 id) +{ + struct hlist_head *head; + struct hlist_node *pos; + struct perf_sample_id *sid; + int hash; + + if (evlist->nr_entries == 1) + return list_entry(evlist->entries.next, struct perf_evsel, node); + + hash = hash_64(id, PERF_EVLIST__HLIST_BITS); + head = &evlist->heads[hash]; + + hlist_for_each_entry(sid, pos, head, node) + if (sid->id == id) + return sid->evsel; + return NULL; +} + +union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *evlist, int cpu) +{ + /* XXX Move this to perf.c, making it generally available */ + unsigned int page_size = sysconf(_SC_PAGE_SIZE); + struct perf_mmap *md = &evlist->mmap[cpu]; + unsigned int head = perf_mmap__read_head(md); + unsigned int old = md->prev; + unsigned char *data = md->base + page_size; + union perf_event *event = NULL; + + if (evlist->overwrite) { + /* + * If we're further behind than half the buffer, there's a chance + * the writer will bite our tail and mess up the samples under us. + * + * If we somehow ended up ahead of the head, we got messed up. + * + * In either case, truncate and restart at head. + */ + int diff = head - old; + if (diff > md->mask / 2 || diff < 0) { + fprintf(stderr, "WARNING: failed to keep up with mmap data.\n"); + + /* + * head points to a known good entry, start there. + */ + old = head; + } + } + + if (old != head) { + size_t size; + + event = (union perf_event *)&data[old & md->mask]; + size = event->header.size; + + /* + * Event straddles the mmap boundary -- header should always + * be inside due to u64 alignment of output. + */ + if ((old & md->mask) + size != ((old + size) & md->mask)) { + unsigned int offset = old; + unsigned int len = min(sizeof(*event), size), cpy; + void *dst = &evlist->event_copy; + + do { + cpy = min(md->mask + 1 - (offset & md->mask), len); + memcpy(dst, &data[offset & md->mask], cpy); + offset += cpy; + dst += cpy; + len -= cpy; + } while (len); + + event = &evlist->event_copy; + } + + old += size; + } + + md->prev = old; + + if (!evlist->overwrite) + perf_mmap__write_tail(md, old); + + return event; +} + +void perf_evlist__munmap(struct perf_evlist *evlist) +{ + int cpu; + + for (cpu = 0; cpu < evlist->cpus->nr; cpu++) { + if (evlist->mmap[cpu].base != NULL) { + munmap(evlist->mmap[cpu].base, evlist->mmap_len); + evlist->mmap[cpu].base = NULL; + } + } +} + +int perf_evlist__alloc_mmap(struct perf_evlist *evlist) +{ + evlist->mmap = zalloc(evlist->cpus->nr * sizeof(struct perf_mmap)); + return evlist->mmap != NULL ? 0 : -ENOMEM; +} + +static int __perf_evlist__mmap(struct perf_evlist *evlist, int cpu, int prot, + int mask, int fd) +{ + evlist->mmap[cpu].prev = 0; + evlist->mmap[cpu].mask = mask; + evlist->mmap[cpu].base = mmap(NULL, evlist->mmap_len, prot, + MAP_SHARED, fd, 0); + if (evlist->mmap[cpu].base == MAP_FAILED) + return -1; + + perf_evlist__add_pollfd(evlist, fd); + return 0; +} + +/** perf_evlist__mmap - Create per cpu maps to receive events + * + * @evlist - list of events + * @pages - map length in pages + * @overwrite - overwrite older events? + * + * If overwrite is false the user needs to signal event consuption using: + * + * struct perf_mmap *m = &evlist->mmap[cpu]; + * unsigned int head = perf_mmap__read_head(m); + * + * perf_mmap__write_tail(m, head) + * + * Using perf_evlist__read_on_cpu does this automatically. + */ +int perf_evlist__mmap(struct perf_evlist *evlist, int pages, bool overwrite) +{ + unsigned int page_size = sysconf(_SC_PAGE_SIZE); + int mask = pages * page_size - 1, cpu; + struct perf_evsel *first_evsel, *evsel; + const struct cpu_map *cpus = evlist->cpus; + const struct thread_map *threads = evlist->threads; + int thread, prot = PROT_READ | (overwrite ? 0 : PROT_WRITE); + + if (evlist->mmap == NULL && perf_evlist__alloc_mmap(evlist) < 0) + return -ENOMEM; + + if (evlist->pollfd == NULL && perf_evlist__alloc_pollfd(evlist) < 0) + return -ENOMEM; + + evlist->overwrite = overwrite; + evlist->mmap_len = (pages + 1) * page_size; + first_evsel = list_entry(evlist->entries.next, struct perf_evsel, node); + + list_for_each_entry(evsel, &evlist->entries, node) { + if ((evsel->attr.read_format & PERF_FORMAT_ID) && + evsel->id == NULL && + perf_evsel__alloc_id(evsel, cpus->nr, threads->nr) < 0) + return -ENOMEM; + + for (cpu = 0; cpu < cpus->nr; cpu++) { + for (thread = 0; thread < threads->nr; thread++) { + int fd = FD(evsel, cpu, thread); + + if (evsel->idx || thread) { + if (ioctl(fd, PERF_EVENT_IOC_SET_OUTPUT, + FD(first_evsel, cpu, 0)) != 0) + goto out_unmap; + } else if (__perf_evlist__mmap(evlist, cpu, prot, mask, fd) < 0) + goto out_unmap; + + if ((evsel->attr.read_format & PERF_FORMAT_ID) && + perf_evlist__id_hash(evlist, evsel, cpu, thread, fd) < 0) + goto out_unmap; + } + } + } + + return 0; + +out_unmap: + for (cpu = 0; cpu < cpus->nr; cpu++) { + if (evlist->mmap[cpu].base != NULL) { + munmap(evlist->mmap[cpu].base, evlist->mmap_len); + evlist->mmap[cpu].base = NULL; + } + } + return -1; +} + +int perf_evlist__create_maps(struct perf_evlist *evlist, pid_t target_pid, + pid_t target_tid, const char *cpu_list) +{ + evlist->threads = thread_map__new(target_pid, target_tid); + + if (evlist->threads == NULL) + return -1; + + if (target_tid != -1) + evlist->cpus = cpu_map__dummy_new(); + else + evlist->cpus = cpu_map__new(cpu_list); + + if (evlist->cpus == NULL) + goto out_delete_threads; + + return 0; + +out_delete_threads: + thread_map__delete(evlist->threads); + return -1; +} + +void perf_evlist__delete_maps(struct perf_evlist *evlist) +{ + cpu_map__delete(evlist->cpus); + thread_map__delete(evlist->threads); + evlist->cpus = NULL; + evlist->threads = NULL; +} diff --git a/tools/perf/util/evlist.h b/tools/perf/util/evlist.h new file mode 100644 index 000000000000..c9884056097c --- /dev/null +++ b/tools/perf/util/evlist.h @@ -0,0 +1,64 @@ +#ifndef __PERF_EVLIST_H +#define __PERF_EVLIST_H 1 + +#include <linux/list.h> +#include "../perf.h" +#include "event.h" + +struct pollfd; +struct thread_map; +struct cpu_map; + +#define PERF_EVLIST__HLIST_BITS 8 +#define PERF_EVLIST__HLIST_SIZE (1 << PERF_EVLIST__HLIST_BITS) + +struct perf_evlist { + struct list_head entries; + struct hlist_head heads[PERF_EVLIST__HLIST_SIZE]; + int nr_entries; + int nr_fds; + int mmap_len; + bool overwrite; + union perf_event event_copy; + struct perf_mmap *mmap; + struct pollfd *pollfd; + struct thread_map *threads; + struct cpu_map *cpus; +}; + +struct perf_evsel; + +struct perf_evlist *perf_evlist__new(struct cpu_map *cpus, + struct thread_map *threads); +void perf_evlist__init(struct perf_evlist *evlist, struct cpu_map *cpus, + struct thread_map *threads); +void perf_evlist__exit(struct perf_evlist *evlist); +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__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); + +union perf_event *perf_evlist__read_on_cpu(struct perf_evlist *self, int cpu); + +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); + +static inline void perf_evlist__set_maps(struct perf_evlist *evlist, + struct cpu_map *cpus, + struct thread_map *threads) +{ + evlist->cpus = cpus; + evlist->threads = threads; +} + +int perf_evlist__create_maps(struct perf_evlist *evlist, pid_t target_pid, + pid_t target_tid, const char *cpu_list); +void perf_evlist__delete_maps(struct perf_evlist *evlist); + +#endif /* __PERF_EVLIST_H */ diff --git a/tools/perf/util/evsel.c b/tools/perf/util/evsel.c index d8575d31ee6c..211063eed474 100644 --- a/tools/perf/util/evsel.c +++ b/tools/perf/util/evsel.c @@ -1,20 +1,34 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Parts came from builtin-{top,stat,record}.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ + #include "evsel.h" -#include "../perf.h" +#include "evlist.h" #include "util.h" #include "cpumap.h" -#include "thread.h" +#include "thread_map.h" #define FD(e, x, y) (*(int *)xyarray__entry(e->fd, x, y)) +void perf_evsel__init(struct perf_evsel *evsel, + struct perf_event_attr *attr, int idx) +{ + evsel->idx = idx; + evsel->attr = *attr; + INIT_LIST_HEAD(&evsel->node); +} + struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx) { struct perf_evsel *evsel = zalloc(sizeof(*evsel)); - if (evsel != NULL) { - evsel->idx = idx; - evsel->attr = *attr; - INIT_LIST_HEAD(&evsel->node); - } + if (evsel != NULL) + perf_evsel__init(evsel, attr, idx); return evsel; } @@ -25,6 +39,12 @@ int perf_evsel__alloc_fd(struct perf_evsel *evsel, int ncpus, int nthreads) return evsel->fd != NULL ? 0 : -ENOMEM; } +int perf_evsel__alloc_id(struct perf_evsel *evsel, int ncpus, int nthreads) +{ + evsel->id = xyarray__new(ncpus, nthreads, sizeof(struct perf_sample_id)); + return evsel->id != NULL ? 0 : -ENOMEM; +} + int perf_evsel__alloc_counts(struct perf_evsel *evsel, int ncpus) { evsel->counts = zalloc((sizeof(*evsel->counts) + @@ -38,6 +58,12 @@ void perf_evsel__free_fd(struct perf_evsel *evsel) evsel->fd = NULL; } +void perf_evsel__free_id(struct perf_evsel *evsel) +{ + xyarray__delete(evsel->id); + evsel->id = NULL; +} + void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads) { int cpu, thread; @@ -49,10 +75,16 @@ void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads) } } -void perf_evsel__delete(struct perf_evsel *evsel) +void perf_evsel__exit(struct perf_evsel *evsel) { assert(list_empty(&evsel->node)); xyarray__delete(evsel->fd); + xyarray__delete(evsel->id); +} + +void perf_evsel__delete(struct perf_evsel *evsel) +{ + perf_evsel__exit(evsel); free(evsel); } @@ -128,7 +160,7 @@ int __perf_evsel__read(struct perf_evsel *evsel, } static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, - struct thread_map *threads) + struct thread_map *threads, bool group, bool inherit) { int cpu, thread; @@ -137,12 +169,20 @@ static int __perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, return -1; for (cpu = 0; cpu < cpus->nr; cpu++) { + int group_fd = -1; + + evsel->attr.inherit = (cpus->map[cpu] < 0) && inherit; + for (thread = 0; thread < threads->nr; thread++) { FD(evsel, cpu, thread) = sys_perf_event_open(&evsel->attr, threads->map[thread], - cpus->map[cpu], -1, 0); + cpus->map[cpu], + group_fd, 0); if (FD(evsel, cpu, thread) < 0) goto out_close; + + if (group && group_fd == -1) + group_fd = FD(evsel, cpu, thread); } } @@ -175,10 +215,9 @@ static struct { .threads = { -1, }, }; -int perf_evsel__open(struct perf_evsel *evsel, - struct cpu_map *cpus, struct thread_map *threads) +int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, + struct thread_map *threads, bool group, bool inherit) { - if (cpus == NULL) { /* Work around old compiler warnings about strict aliasing */ cpus = &empty_cpu_map.map; @@ -187,15 +226,135 @@ int perf_evsel__open(struct perf_evsel *evsel, if (threads == NULL) threads = &empty_thread_map.map; - return __perf_evsel__open(evsel, cpus, threads); + return __perf_evsel__open(evsel, cpus, threads, group, inherit); +} + +int perf_evsel__open_per_cpu(struct perf_evsel *evsel, + struct cpu_map *cpus, bool group, bool inherit) +{ + return __perf_evsel__open(evsel, cpus, &empty_thread_map.map, group, inherit); } -int perf_evsel__open_per_cpu(struct perf_evsel *evsel, struct cpu_map *cpus) +int perf_evsel__open_per_thread(struct perf_evsel *evsel, + struct thread_map *threads, bool group, bool inherit) { - return __perf_evsel__open(evsel, cpus, &empty_thread_map.map); + return __perf_evsel__open(evsel, &empty_cpu_map.map, threads, group, inherit); } -int perf_evsel__open_per_thread(struct perf_evsel *evsel, struct thread_map *threads) +static int perf_event__parse_id_sample(const union perf_event *event, u64 type, + struct perf_sample *sample) { - return __perf_evsel__open(evsel, &empty_cpu_map.map, threads); + const u64 *array = event->sample.array; + + array += ((event->header.size - + sizeof(event->header)) / sizeof(u64)) - 1; + + if (type & PERF_SAMPLE_CPU) { + u32 *p = (u32 *)array; + sample->cpu = *p; + array--; + } + + if (type & PERF_SAMPLE_STREAM_ID) { + sample->stream_id = *array; + array--; + } + + if (type & PERF_SAMPLE_ID) { + sample->id = *array; + array--; + } + + if (type & PERF_SAMPLE_TIME) { + sample->time = *array; + array--; + } + + if (type & PERF_SAMPLE_TID) { + u32 *p = (u32 *)array; + sample->pid = p[0]; + sample->tid = p[1]; + } + + return 0; +} + +int perf_event__parse_sample(const union perf_event *event, u64 type, + bool sample_id_all, struct perf_sample *data) +{ + const u64 *array; + + data->cpu = data->pid = data->tid = -1; + data->stream_id = data->id = data->time = -1ULL; + + if (event->header.type != PERF_RECORD_SAMPLE) { + if (!sample_id_all) + return 0; + return perf_event__parse_id_sample(event, type, data); + } + + array = event->sample.array; + + if (type & PERF_SAMPLE_IP) { + data->ip = event->ip.ip; + array++; + } + + if (type & PERF_SAMPLE_TID) { + u32 *p = (u32 *)array; + data->pid = p[0]; + data->tid = p[1]; + array++; + } + + if (type & PERF_SAMPLE_TIME) { + data->time = *array; + array++; + } + + if (type & PERF_SAMPLE_ADDR) { + data->addr = *array; + array++; + } + + data->id = -1ULL; + if (type & PERF_SAMPLE_ID) { + data->id = *array; + array++; + } + + if (type & PERF_SAMPLE_STREAM_ID) { + data->stream_id = *array; + array++; + } + + if (type & PERF_SAMPLE_CPU) { + u32 *p = (u32 *)array; + data->cpu = *p; + array++; + } + + if (type & PERF_SAMPLE_PERIOD) { + data->period = *array; + array++; + } + + if (type & PERF_SAMPLE_READ) { + fprintf(stderr, "PERF_SAMPLE_READ is unsuported for now\n"); + return -1; + } + + if (type & PERF_SAMPLE_CALLCHAIN) { + data->callchain = (struct ip_callchain *)array; + array += 1 + data->callchain->nr; + } + + if (type & PERF_SAMPLE_RAW) { + u32 *p = (u32 *)array; + data->raw_size = *p; + p++; + data->raw_data = p; + } + + return 0; } diff --git a/tools/perf/util/evsel.h b/tools/perf/util/evsel.h index b2d755fe88a5..eecdc3aabc14 100644 --- a/tools/perf/util/evsel.h +++ b/tools/perf/util/evsel.h @@ -24,11 +24,24 @@ struct perf_counts { struct perf_counts_values cpu[]; }; +struct perf_evsel; + +/* + * Per fd, to map back from PERF_SAMPLE_ID to evsel, only used when there are + * more than one entry in the evlist. + */ +struct perf_sample_id { + struct hlist_node node; + u64 id; + struct perf_evsel *evsel; +}; + struct perf_evsel { struct list_head node; struct perf_event_attr attr; char *filter; struct xyarray *fd; + struct xyarray *id; struct perf_counts *counts; int idx; void *priv; @@ -36,19 +49,27 @@ struct perf_evsel { struct cpu_map; struct thread_map; +struct perf_evlist; struct perf_evsel *perf_evsel__new(struct perf_event_attr *attr, int idx); +void perf_evsel__init(struct perf_evsel *evsel, + struct perf_event_attr *attr, int idx); +void perf_evsel__exit(struct perf_evsel *evsel); void perf_evsel__delete(struct perf_evsel *evsel); 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); void perf_evsel__free_fd(struct perf_evsel *evsel); +void perf_evsel__free_id(struct perf_evsel *evsel); void perf_evsel__close_fd(struct perf_evsel *evsel, int ncpus, int nthreads); -int perf_evsel__open_per_cpu(struct perf_evsel *evsel, struct cpu_map *cpus); -int perf_evsel__open_per_thread(struct perf_evsel *evsel, struct thread_map *threads); -int perf_evsel__open(struct perf_evsel *evsel, - struct cpu_map *cpus, struct thread_map *threads); +int perf_evsel__open_per_cpu(struct perf_evsel *evsel, + struct cpu_map *cpus, bool group, bool inherit); +int perf_evsel__open_per_thread(struct perf_evsel *evsel, + struct thread_map *threads, bool group, bool inherit); +int perf_evsel__open(struct perf_evsel *evsel, struct cpu_map *cpus, + struct thread_map *threads, bool group, bool inherit); #define perf_evsel__match(evsel, t, c) \ (evsel->attr.type == PERF_TYPE_##t && \ diff --git a/tools/perf/util/header.c b/tools/perf/util/header.c index f6a929e74981..72c124dc5781 100644 --- a/tools/perf/util/header.c +++ b/tools/perf/util/header.c @@ -8,6 +8,7 @@ #include <linux/list.h> #include <linux/kernel.h> +#include "evlist.h" #include "util.h" #include "header.h" #include "../perf.h" @@ -428,7 +429,8 @@ static bool perf_session__read_build_ids(struct perf_session *self, bool with_hi return ret; } -static int perf_header__adds_write(struct perf_header *self, int fd) +static int perf_header__adds_write(struct perf_header *self, + struct perf_evlist *evlist, int fd) { int nr_sections; struct perf_session *session; @@ -463,7 +465,7 @@ static int perf_header__adds_write(struct perf_header *self, int fd) /* Write trace info */ trace_sec->offset = lseek(fd, 0, SEEK_CUR); - read_tracing_data(fd, &evsel_list); + read_tracing_data(fd, &evlist->entries); trace_sec->size = lseek(fd, 0, SEEK_CUR) - trace_sec->offset; } @@ -513,7 +515,8 @@ int perf_header__write_pipe(int fd) return 0; } -int perf_header__write(struct perf_header *self, int fd, bool at_exit) +int perf_header__write(struct perf_header *self, struct perf_evlist *evlist, + int fd, bool at_exit) { struct perf_file_header f_header; struct perf_file_attr f_attr; @@ -566,7 +569,7 @@ int perf_header__write(struct perf_header *self, int fd, bool at_exit) self->data_offset = lseek(fd, 0, SEEK_CUR); if (at_exit) { - err = perf_header__adds_write(self, fd); + err = perf_header__adds_write(self, evlist, fd); if (err < 0) return err; } @@ -997,11 +1000,11 @@ perf_header__find_attr(u64 id, struct perf_header *header) return NULL; } -int event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, - event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, + perf_event__handler_t process, + struct perf_session *session) { - event_t *ev; + union perf_event *ev; size_t size; int err; @@ -1028,8 +1031,9 @@ int event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, return err; } -int event__synthesize_attrs(struct perf_header *self, event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_attrs(struct perf_header *self, + perf_event__handler_t process, + struct perf_session *session) { struct perf_header_attr *attr; int i, err = 0; @@ -1037,8 +1041,8 @@ int event__synthesize_attrs(struct perf_header *self, event__handler_t process, for (i = 0; i < self->attrs; i++) { attr = self->attr[i]; - err = event__synthesize_attr(&attr->attr, attr->ids, attr->id, - process, session); + err = perf_event__synthesize_attr(&attr->attr, attr->ids, + attr->id, process, session); if (err) { pr_debug("failed to create perf header attribute\n"); return err; @@ -1048,21 +1052,22 @@ int event__synthesize_attrs(struct perf_header *self, event__handler_t process, return err; } -int event__process_attr(event_t *self, struct perf_session *session) +int perf_event__process_attr(union perf_event *event, + struct perf_session *session) { struct perf_header_attr *attr; unsigned int i, ids, n_ids; - attr = perf_header_attr__new(&self->attr.attr); + attr = perf_header_attr__new(&event->attr.attr); if (attr == NULL) return -ENOMEM; - ids = self->header.size; - ids -= (void *)&self->attr.id - (void *)self; + ids = event->header.size; + ids -= (void *)&event->attr.id - (void *)event; n_ids = ids / sizeof(u64); for (i = 0; i < n_ids; i++) { - if (perf_header_attr__add_id(attr, self->attr.id[i]) < 0) { + if (perf_header_attr__add_id(attr, event->attr.id[i]) < 0) { perf_header_attr__delete(attr); return -ENOMEM; } @@ -1078,11 +1083,11 @@ int event__process_attr(event_t *self, struct perf_session *session) return 0; } -int event__synthesize_event_type(u64 event_id, char *name, - event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_event_type(u64 event_id, char *name, + perf_event__handler_t process, + struct perf_session *session) { - event_t ev; + union perf_event ev; size_t size = 0; int err = 0; @@ -1103,8 +1108,8 @@ int event__synthesize_event_type(u64 event_id, char *name, return err; } -int event__synthesize_event_types(event__handler_t process, - struct perf_session *session) +int perf_event__synthesize_event_types(perf_event__handler_t process, + struct perf_session *session) { struct perf_trace_event_type *type; int i, err = 0; @@ -1112,8 +1117,9 @@ int event__synthesize_event_types(event__handler_t process, for (i = 0; i < event_count; i++) { type = &events[i]; - err = event__synthesize_event_type(type->event_id, type->name, - process, session); + err = perf_event__synthesize_event_type(type->event_id, + type->name, process, + session); if (err) { pr_debug("failed to create perf header event type\n"); return err; @@ -1123,28 +1129,28 @@ int event__synthesize_event_types(event__handler_t process, return err; } -int event__process_event_type(event_t *self, - struct perf_session *session __unused) +int perf_event__process_event_type(union perf_event *event, + struct perf_session *session __unused) { - if (perf_header__push_event(self->event_type.event_type.event_id, - self->event_type.event_type.name) < 0) + if (perf_header__push_event(event->event_type.event_type.event_id, + event->event_type.event_type.name) < 0) return -ENOMEM; return 0; } -int event__synthesize_tracing_data(int fd, struct list_head *pattrs, - event__handler_t process, +int perf_event__synthesize_tracing_data(int fd, struct perf_evlist *evlist, + perf_event__handler_t process, struct perf_session *session __unused) { - event_t ev; + union perf_event ev; ssize_t size = 0, aligned_size = 0, padding; - int err = 0; + int err __used = 0; memset(&ev, 0, sizeof(ev)); ev.tracing_data.header.type = PERF_RECORD_HEADER_TRACING_DATA; - size = read_tracing_data_size(fd, pattrs); + size = read_tracing_data_size(fd, &evlist->entries); if (size <= 0) return size; aligned_size = ALIGN(size, sizeof(u64)); @@ -1154,16 +1160,16 @@ int event__synthesize_tracing_data(int fd, struct list_head *pattrs, process(&ev, NULL, session); - err = read_tracing_data(fd, pattrs); + err = read_tracing_data(fd, &evlist->entries); write_padded(fd, NULL, 0, padding); return aligned_size; } -int event__process_tracing_data(event_t *self, - struct perf_session *session) +int perf_event__process_tracing_data(union perf_event *event, + struct perf_session *session) { - ssize_t size_read, padding, size = self->tracing_data.size; + ssize_t size_read, padding, size = event->tracing_data.size; off_t offset = lseek(session->fd, 0, SEEK_CUR); char buf[BUFSIZ]; @@ -1189,12 +1195,12 @@ int event__process_tracing_data(event_t *self, return size_read + padding; } -int event__synthesize_build_id(struct dso *pos, u16 misc, - event__handler_t process, - struct machine *machine, - struct perf_session *session) +int perf_event__synthesize_build_id(struct dso *pos, u16 misc, + perf_event__handler_t process, + struct machine *machine, + struct perf_session *session) { - event_t ev; + union perf_event ev; size_t len; int err = 0; @@ -1217,11 +1223,11 @@ int event__synthesize_build_id(struct dso *pos, u16 misc, return err; } -int event__process_build_id(event_t *self, - struct perf_session *session) +int perf_event__process_build_id(union perf_event *event, + struct perf_session *session) { - __event_process_build_id(&self->build_id, - self->build_id.filename, + __event_process_build_id(&event->build_id, + event->build_id.filename, session); return 0; } diff --git a/tools/perf/util/header.h b/tools/perf/util/header.h index 33f16be7b72f..f042cebcec1e 100644 --- a/tools/perf/util/header.h +++ b/tools/perf/util/header.h @@ -65,8 +65,11 @@ struct perf_header { int perf_header__init(struct perf_header *self); void perf_header__exit(struct perf_header *self); +struct perf_evlist; + int perf_header__read(struct perf_session *session, int fd); -int perf_header__write(struct perf_header *self, int fd, bool at_exit); +int perf_header__write(struct perf_header *self, struct perf_evlist *evlist, + int fd, bool at_exit); int perf_header__write_pipe(int fd); int perf_header__add_attr(struct perf_header *self, @@ -97,32 +100,32 @@ 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 event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, - event__handler_t process, - struct perf_session *session); -int event__synthesize_attrs(struct perf_header *self, - event__handler_t process, - struct perf_session *session); -int event__process_attr(event_t *self, struct perf_session *session); - -int event__synthesize_event_type(u64 event_id, char *name, - event__handler_t process, +int perf_event__synthesize_attr(struct perf_event_attr *attr, u16 ids, u64 *id, + perf_event__handler_t process, + struct perf_session *session); +int perf_event__synthesize_attrs(struct perf_header *self, + perf_event__handler_t process, struct perf_session *session); -int event__synthesize_event_types(event__handler_t process, - struct perf_session *session); -int event__process_event_type(event_t *self, - struct perf_session *session); - -int event__synthesize_tracing_data(int fd, struct list_head *pattrs, - event__handler_t process, +int perf_event__process_attr(union perf_event *event, struct perf_session *session); + +int perf_event__synthesize_event_type(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, + struct perf_session *session); +int perf_event__process_event_type(union perf_event *event, struct perf_session *session); -int event__process_tracing_data(event_t *self, - struct perf_session *session); - -int event__synthesize_build_id(struct dso *pos, u16 misc, - event__handler_t process, - struct machine *machine, - struct perf_session *session); -int event__process_build_id(event_t *self, struct perf_session *session); +int perf_event__synthesize_tracing_data(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, + perf_event__handler_t process, + struct machine *machine, + struct perf_session *session); +int perf_event__process_build_id(union perf_event *event, + struct perf_session *session); #endif /* __PERF_HEADER_H */ diff --git a/tools/perf/util/hist.c b/tools/perf/util/hist.c index 32f4f1f2f6e4..3f437236f193 100644 --- a/tools/perf/util/hist.c +++ b/tools/perf/util/hist.c @@ -1,3 +1,4 @@ +#include "annotate.h" #include "util.h" #include "build-id.h" #include "hist.h" @@ -211,7 +212,9 @@ void hist_entry__free(struct hist_entry *he) * collapse the histogram */ -static bool collapse__insert_entry(struct rb_root *root, struct hist_entry *he) +static bool hists__collapse_insert_entry(struct hists *self, + struct rb_root *root, + struct hist_entry *he) { struct rb_node **p = &root->rb_node; struct rb_node *parent = NULL; @@ -226,8 +229,11 @@ static bool collapse__insert_entry(struct rb_root *root, struct hist_entry *he) if (!cmp) { iter->period += he->period; - if (symbol_conf.use_callchain) - callchain_merge(iter->callchain, he->callchain); + if (symbol_conf.use_callchain) { + callchain_cursor_reset(&self->callchain_cursor); + callchain_merge(&self->callchain_cursor, iter->callchain, + he->callchain); + } hist_entry__free(he); return false; } @@ -262,7 +268,7 @@ void hists__collapse_resort(struct hists *self) next = rb_next(&n->rb_node); rb_erase(&n->rb_node, &self->entries); - if (collapse__insert_entry(&tmp, n)) + if (hists__collapse_insert_entry(self, &tmp, n)) hists__inc_nr_entries(self, n); } @@ -425,7 +431,7 @@ static size_t __callchain__fprintf_graph(FILE *fp, struct callchain_node *self, u64 cumul; child = rb_entry(node, struct callchain_node, rb_node); - cumul = cumul_hits(child); + cumul = callchain_cumul_hits(child); remaining -= cumul; /* @@ -944,225 +950,14 @@ void hists__filter_by_thread(struct hists *self, const struct thread *thread) } } -static int symbol__alloc_hist(struct symbol *self) -{ - struct sym_priv *priv = symbol__priv(self); - const int size = (sizeof(*priv->hist) + - (self->end - self->start) * sizeof(u64)); - - priv->hist = zalloc(size); - return priv->hist == NULL ? -1 : 0; -} - -int hist_entry__inc_addr_samples(struct hist_entry *self, u64 ip) -{ - unsigned int sym_size, offset; - struct symbol *sym = self->ms.sym; - struct sym_priv *priv; - struct sym_hist *h; - - if (!sym || !self->ms.map) - return 0; - - priv = symbol__priv(sym); - if (priv->hist == NULL && symbol__alloc_hist(sym) < 0) - return -ENOMEM; - - sym_size = sym->end - sym->start; - offset = ip - sym->start; - - pr_debug3("%s: ip=%#" PRIx64 "\n", __func__, self->ms.map->unmap_ip(self->ms.map, ip)); - - if (offset >= sym_size) - return 0; - - h = priv->hist; - h->sum++; - h->ip[offset]++; - - pr_debug3("%#" PRIx64 " %s: period++ [ip: %#" PRIx64 ", %#" PRIx64 - "] => %" PRIu64 "\n", self->ms.sym->start, self->ms.sym->name, - ip, ip - self->ms.sym->start, h->ip[offset]); - return 0; -} - -static struct objdump_line *objdump_line__new(s64 offset, char *line, size_t privsize) -{ - struct objdump_line *self = malloc(sizeof(*self) + privsize); - - if (self != NULL) { - self->offset = offset; - self->line = line; - } - - return self; -} - -void objdump_line__free(struct objdump_line *self) -{ - free(self->line); - free(self); -} - -static void objdump__add_line(struct list_head *head, struct objdump_line *line) -{ - list_add_tail(&line->node, head); -} - -struct objdump_line *objdump__get_next_ip_line(struct list_head *head, - struct objdump_line *pos) -{ - list_for_each_entry_continue(pos, head, node) - if (pos->offset >= 0) - return pos; - - return NULL; -} - -static int hist_entry__parse_objdump_line(struct hist_entry *self, FILE *file, - struct list_head *head, size_t privsize) +int hist_entry__inc_addr_samples(struct hist_entry *he, int evidx, u64 ip) { - struct symbol *sym = self->ms.sym; - struct objdump_line *objdump_line; - char *line = NULL, *tmp, *tmp2, *c; - size_t line_len; - s64 line_ip, offset = -1; - - if (getline(&line, &line_len, file) < 0) - return -1; - - if (!line) - return -1; - - while (line_len != 0 && isspace(line[line_len - 1])) - line[--line_len] = '\0'; - - c = strchr(line, '\n'); - if (c) - *c = 0; - - line_ip = -1; - - /* - * Strip leading spaces: - */ - tmp = line; - while (*tmp) { - if (*tmp != ' ') - break; - tmp++; - } - - if (*tmp) { - /* - * Parse hexa addresses followed by ':' - */ - line_ip = strtoull(tmp, &tmp2, 16); - if (*tmp2 != ':' || tmp == tmp2 || tmp2[1] == '\0') - line_ip = -1; - } - - if (line_ip != -1) { - u64 start = map__rip_2objdump(self->ms.map, sym->start), - end = map__rip_2objdump(self->ms.map, sym->end); - - offset = line_ip - start; - if (offset < 0 || (u64)line_ip > end) - offset = -1; - } - - objdump_line = objdump_line__new(offset, line, privsize); - if (objdump_line == NULL) { - free(line); - return -1; - } - objdump__add_line(head, objdump_line); - - return 0; + return symbol__inc_addr_samples(he->ms.sym, he->ms.map, evidx, ip); } -int hist_entry__annotate(struct hist_entry *self, struct list_head *head, - size_t privsize) +int hist_entry__annotate(struct hist_entry *he, size_t privsize) { - struct symbol *sym = self->ms.sym; - struct map *map = self->ms.map; - struct dso *dso = map->dso; - char *filename = dso__build_id_filename(dso, NULL, 0); - bool free_filename = true; - char command[PATH_MAX * 2]; - FILE *file; - int err = 0; - u64 len; - char symfs_filename[PATH_MAX]; - - if (filename) { - snprintf(symfs_filename, sizeof(symfs_filename), "%s%s", - symbol_conf.symfs, filename); - } - - if (filename == NULL) { - if (dso->has_build_id) { - pr_err("Can't annotate %s: not enough memory\n", - sym->name); - return -ENOMEM; - } - goto fallback; - } else if (readlink(symfs_filename, command, sizeof(command)) < 0 || - strstr(command, "[kernel.kallsyms]") || - access(symfs_filename, R_OK)) { - free(filename); -fallback: - /* - * If we don't have build-ids or the build-id file isn't in the - * cache, or is just a kallsyms file, well, lets hope that this - * DSO is the same as when 'perf record' ran. - */ - filename = dso->long_name; - snprintf(symfs_filename, sizeof(symfs_filename), "%s%s", - symbol_conf.symfs, filename); - free_filename = false; - } - - if (dso->origin == DSO__ORIG_KERNEL) { - if (dso->annotate_warned) - goto out_free_filename; - err = -ENOENT; - dso->annotate_warned = 1; - pr_err("Can't annotate %s: No vmlinux file was found in the " - "path\n", sym->name); - goto out_free_filename; - } - - pr_debug("%s: filename=%s, sym=%s, start=%#" PRIx64 ", end=%#" PRIx64 "\n", __func__, - filename, sym->name, map->unmap_ip(map, sym->start), - map->unmap_ip(map, sym->end)); - - len = sym->end - sym->start; - - pr_debug("annotating [%p] %30s : [%p] %30s\n", - dso, dso->long_name, sym, sym->name); - - snprintf(command, sizeof(command), - "objdump --start-address=0x%016" PRIx64 " --stop-address=0x%016" PRIx64 " -dS -C %s|grep -v %s|expand", - map__rip_2objdump(map, sym->start), - map__rip_2objdump(map, sym->end), - symfs_filename, filename); - - pr_debug("Executing: %s\n", command); - - file = popen(command, "r"); - if (!file) - goto out_free_filename; - - while (!feof(file)) - if (hist_entry__parse_objdump_line(self, file, head, privsize) < 0) - break; - - pclose(file); -out_free_filename: - if (free_filename) - free(filename); - return err; + return symbol__annotate(he->ms.sym, he->ms.map, privsize); } void hists__inc_nr_events(struct hists *self, u32 type) @@ -1177,7 +972,7 @@ size_t hists__fprintf_nr_events(struct hists *self, FILE *fp) size_t ret = 0; for (i = 0; i < PERF_RECORD_HEADER_MAX; ++i) { - const char *name = event__get_event_name(i); + const char *name = perf_event__name(i); if (!strcmp(name, "UNKNOWN")) continue; diff --git a/tools/perf/util/hist.h b/tools/perf/util/hist.h index ee789856a8c9..37c79089de09 100644 --- a/tools/perf/util/hist.h +++ b/tools/perf/util/hist.h @@ -9,33 +9,6 @@ extern struct callchain_param callchain_param; struct hist_entry; struct addr_location; struct symbol; -struct rb_root; - -struct objdump_line { - struct list_head node; - s64 offset; - char *line; -}; - -void objdump_line__free(struct objdump_line *self); -struct objdump_line *objdump__get_next_ip_line(struct list_head *head, - struct objdump_line *pos); - -struct sym_hist { - u64 sum; - u64 ip[0]; -}; - -struct sym_ext { - struct rb_node node; - double percent; - char *path; -}; - -struct sym_priv { - struct sym_hist *hist; - struct sym_ext *ext; -}; /* * The kernel collects the number of events it couldn't send in a stretch and @@ -77,6 +50,8 @@ struct hists { u64 event_stream; u32 type; u16 col_len[HISTC_NR_COLS]; + /* Best would be to reuse the session callchain cursor */ + struct callchain_cursor callchain_cursor; }; struct hist_entry *__hists__add_entry(struct hists *self, @@ -102,9 +77,8 @@ size_t hists__fprintf_nr_events(struct hists *self, FILE *fp); size_t hists__fprintf(struct hists *self, struct hists *pair, bool show_displacement, FILE *fp); -int hist_entry__inc_addr_samples(struct hist_entry *self, u64 ip); -int hist_entry__annotate(struct hist_entry *self, struct list_head *head, - size_t privsize); +int hist_entry__inc_addr_samples(struct hist_entry *self, int evidx, u64 addr); +int hist_entry__annotate(struct hist_entry *self, size_t privsize); void hists__filter_by_dso(struct hists *self, const struct dso *dso); void hists__filter_by_thread(struct hists *self, const struct thread *thread); @@ -116,18 +90,20 @@ bool hists__new_col_len(struct hists *self, enum hist_column col, u16 len); #ifdef NO_NEWT_SUPPORT static inline int hists__browse(struct hists *self __used, const char *helpline __used, - const char *ev_name __used) + const char *ev_name __used, int evidx __used) { return 0; } static inline int hists__tui_browse_tree(struct rb_root *self __used, - const char *help __used) + const char *help __used, + int evidx __used) { return 0; } -static inline int hist_entry__tui_annotate(struct hist_entry *self __used) +static inline int hist_entry__tui_annotate(struct hist_entry *self __used, + int evidx __used) { return 0; } @@ -136,13 +112,13 @@ static inline int hist_entry__tui_annotate(struct hist_entry *self __used) #else #include <newt.h> int hists__browse(struct hists *self, const char *helpline, - const char *ev_name); -int hist_entry__tui_annotate(struct hist_entry *self); + const char *ev_name, int evidx); +int hist_entry__tui_annotate(struct hist_entry *self, int evidx); #define KEY_LEFT NEWT_KEY_LEFT #define KEY_RIGHT NEWT_KEY_RIGHT -int hists__tui_browse_tree(struct rb_root *self, const char *help); +int hists__tui_browse_tree(struct rb_root *self, const char *help, int evidx); #endif unsigned int hists__sort_list_width(struct hists *self); diff --git a/tools/perf/util/include/linux/list.h b/tools/perf/util/include/linux/list.h index f5ca26e53fbb..356c7e467b83 100644 --- a/tools/perf/util/include/linux/list.h +++ b/tools/perf/util/include/linux/list.h @@ -1,3 +1,4 @@ +#include <linux/kernel.h> #include "../../../../include/linux/list.h" #ifndef PERF_LIST_H diff --git a/tools/perf/util/parse-events.c b/tools/perf/util/parse-events.c index 135f69baf966..cf082daa43e3 100644 --- a/tools/perf/util/parse-events.c +++ b/tools/perf/util/parse-events.c @@ -1,6 +1,7 @@ #include "../../../include/linux/hw_breakpoint.h" #include "util.h" #include "../perf.h" +#include "evlist.h" #include "evsel.h" #include "parse-options.h" #include "parse-events.h" @@ -11,10 +12,6 @@ #include "header.h" #include "debugfs.h" -int nr_counters; - -LIST_HEAD(evsel_list); - struct event_symbol { u8 type; u64 config; @@ -449,8 +446,8 @@ parse_single_tracepoint_event(char *sys_name, /* sys + ':' + event + ':' + flags*/ #define MAX_EVOPT_LEN (MAX_EVENT_LENGTH * 2 + 2 + 128) static enum event_result -parse_multiple_tracepoint_event(char *sys_name, const char *evt_exp, - char *flags) +parse_multiple_tracepoint_event(const struct option *opt, char *sys_name, + const char *evt_exp, char *flags) { char evt_path[MAXPATHLEN]; struct dirent *evt_ent; @@ -483,15 +480,16 @@ parse_multiple_tracepoint_event(char *sys_name, const char *evt_exp, if (len < 0) return EVT_FAILED; - if (parse_events(NULL, event_opt, 0)) + if (parse_events(opt, event_opt, 0)) return EVT_FAILED; } return EVT_HANDLED_ALL; } -static enum event_result parse_tracepoint_event(const char **strp, - struct perf_event_attr *attr) +static enum event_result +parse_tracepoint_event(const struct option *opt, const char **strp, + struct perf_event_attr *attr) { const char *evt_name; char *flags = NULL, *comma_loc; @@ -530,7 +528,7 @@ static enum event_result parse_tracepoint_event(const char **strp, return EVT_FAILED; if (strpbrk(evt_name, "*?")) { *strp += strlen(sys_name) + evt_length + 1; /* 1 == the ':' */ - return parse_multiple_tracepoint_event(sys_name, evt_name, + return parse_multiple_tracepoint_event(opt, sys_name, evt_name, flags); } else { return parse_single_tracepoint_event(sys_name, evt_name, @@ -740,11 +738,12 @@ parse_event_modifier(const char **strp, struct perf_event_attr *attr) * Symbolic names are (almost) exactly matched. */ static enum event_result -parse_event_symbols(const char **str, struct perf_event_attr *attr) +parse_event_symbols(const struct option *opt, const char **str, + struct perf_event_attr *attr) { enum event_result ret; - ret = parse_tracepoint_event(str, attr); + ret = parse_tracepoint_event(opt, str, attr); if (ret != EVT_FAILED) goto modifier; @@ -778,14 +777,15 @@ modifier: return ret; } -int parse_events(const struct option *opt __used, const char *str, int unset __used) +int parse_events(const struct option *opt, const char *str, int unset __used) { + struct perf_evlist *evlist = *(struct perf_evlist **)opt->value; struct perf_event_attr attr; enum event_result ret; for (;;) { memset(&attr, 0, sizeof(attr)); - ret = parse_event_symbols(&str, &attr); + ret = parse_event_symbols(opt, &str, &attr); if (ret == EVT_FAILED) return -1; @@ -794,12 +794,10 @@ int parse_events(const struct option *opt __used, const char *str, int unset __u if (ret != EVT_HANDLED_ALL) { struct perf_evsel *evsel; - evsel = perf_evsel__new(&attr, - nr_counters); + evsel = perf_evsel__new(&attr, evlist->nr_entries); if (evsel == NULL) return -1; - list_add_tail(&evsel->node, &evsel_list); - ++nr_counters; + perf_evlist__add(evlist, evsel); } if (*str == 0) @@ -813,13 +811,14 @@ int parse_events(const struct option *opt __used, const char *str, int unset __u return 0; } -int parse_filter(const struct option *opt __used, const char *str, +int parse_filter(const struct option *opt, const char *str, int unset __used) { + struct perf_evlist *evlist = *(struct perf_evlist **)opt->value; struct perf_evsel *last = NULL; - if (!list_empty(&evsel_list)) - last = list_entry(evsel_list.prev, struct perf_evsel, node); + if (evlist->nr_entries > 0) + last = list_entry(evlist->entries.prev, struct perf_evsel, node); if (last == NULL || last->attr.type != PERF_TYPE_TRACEPOINT) { fprintf(stderr, @@ -981,33 +980,3 @@ void print_events(void) exit(129); } - -int perf_evsel_list__create_default(void) -{ - struct perf_evsel *evsel; - struct perf_event_attr attr; - - memset(&attr, 0, sizeof(attr)); - attr.type = PERF_TYPE_HARDWARE; - attr.config = PERF_COUNT_HW_CPU_CYCLES; - - evsel = perf_evsel__new(&attr, 0); - - if (evsel == NULL) - return -ENOMEM; - - list_add(&evsel->node, &evsel_list); - ++nr_counters; - return 0; -} - -void perf_evsel_list__delete(void) -{ - struct perf_evsel *pos, *n; - - list_for_each_entry_safe(pos, n, &evsel_list, node) { - list_del_init(&pos->node); - perf_evsel__delete(pos); - } - nr_counters = 0; -} diff --git a/tools/perf/util/parse-events.h b/tools/perf/util/parse-events.h index 458e3ecf17af..cf7e94abb676 100644 --- a/tools/perf/util/parse-events.h +++ b/tools/perf/util/parse-events.h @@ -9,11 +9,6 @@ struct list_head; struct perf_evsel; -extern struct list_head evsel_list; - -int perf_evsel_list__create_default(void); -void perf_evsel_list__delete(void); - struct option; struct tracepoint_path { @@ -25,8 +20,6 @@ struct tracepoint_path { extern struct tracepoint_path *tracepoint_id_to_path(u64 config); extern bool have_tracepoints(struct list_head *evlist); -extern int nr_counters; - const char *event_name(struct perf_evsel *event); extern const char *__event_name(int type, u64 config); diff --git a/tools/perf/util/probe-event.c b/tools/perf/util/probe-event.c index 6e29d9c9dccc..9d237e3cff5d 100644 --- a/tools/perf/util/probe-event.c +++ b/tools/perf/util/probe-event.c @@ -31,6 +31,7 @@ #include <string.h> #include <stdarg.h> #include <limits.h> +#include <elf.h> #undef _GNU_SOURCE #include "util.h" @@ -111,7 +112,25 @@ static struct symbol *__find_kernel_function_by_name(const char *name, NULL); } -const char *kernel_get_module_path(const char *module) +static struct map *kernel_get_module_map(const char *module) +{ + struct rb_node *nd; + struct map_groups *grp = &machine.kmaps; + + if (!module) + module = "kernel"; + + for (nd = rb_first(&grp->maps[MAP__FUNCTION]); nd; nd = rb_next(nd)) { + struct map *pos = rb_entry(nd, struct map, rb_node); + if (strncmp(pos->dso->short_name + 1, module, + pos->dso->short_name_len - 2) == 0) { + return pos; + } + } + return NULL; +} + +static struct dso *kernel_get_module_dso(const char *module) { struct dso *dso; struct map *map; @@ -141,7 +160,13 @@ const char *kernel_get_module_path(const char *module) } } found: - return dso->long_name; + return dso; +} + +const char *kernel_get_module_path(const char *module) +{ + struct dso *dso = kernel_get_module_dso(module); + return (dso) ? dso->long_name : NULL; } #ifdef DWARF_SUPPORT @@ -426,12 +451,14 @@ end: } static int show_available_vars_at(int fd, struct perf_probe_event *pev, - int max_vls, bool externs) + int max_vls, struct strfilter *_filter, + bool externs) { char *buf; - int ret, i; + int ret, i, nvars; struct str_node *node; struct variable_list *vls = NULL, *vl; + const char *var; buf = synthesize_perf_probe_point(&pev->point); if (!buf) @@ -439,36 +466,45 @@ static int show_available_vars_at(int fd, struct perf_probe_event *pev, pr_debug("Searching variables at %s\n", buf); ret = find_available_vars_at(fd, pev, &vls, max_vls, externs); - if (ret > 0) { - /* Some variables were found */ - fprintf(stdout, "Available variables at %s\n", buf); - for (i = 0; i < ret; i++) { - vl = &vls[i]; - /* - * A probe point might be converted to - * several trace points. - */ - fprintf(stdout, "\t@<%s+%lu>\n", vl->point.symbol, - vl->point.offset); - free(vl->point.symbol); - if (vl->vars) { - strlist__for_each(node, vl->vars) + if (ret <= 0) { + pr_err("Failed to find variables at %s (%d)\n", buf, ret); + goto end; + } + /* Some variables are found */ + fprintf(stdout, "Available variables at %s\n", buf); + for (i = 0; i < ret; i++) { + vl = &vls[i]; + /* + * A probe point might be converted to + * several trace points. + */ + fprintf(stdout, "\t@<%s+%lu>\n", vl->point.symbol, + vl->point.offset); + free(vl->point.symbol); + nvars = 0; + if (vl->vars) { + strlist__for_each(node, vl->vars) { + var = strchr(node->s, '\t') + 1; + if (strfilter__compare(_filter, var)) { fprintf(stdout, "\t\t%s\n", node->s); - strlist__delete(vl->vars); - } else - fprintf(stdout, "(No variables)\n"); + nvars++; + } + } + strlist__delete(vl->vars); } - free(vls); - } else - pr_err("Failed to find variables at %s (%d)\n", buf, ret); - + if (nvars == 0) + fprintf(stdout, "\t\t(No matched variables)\n"); + } + free(vls); +end: free(buf); return ret; } /* Show available variables on given probe point */ int show_available_vars(struct perf_probe_event *pevs, int npevs, - int max_vls, const char *module, bool externs) + int max_vls, const char *module, + struct strfilter *_filter, bool externs) { int i, fd, ret = 0; @@ -485,7 +521,8 @@ int show_available_vars(struct perf_probe_event *pevs, int npevs, setup_pager(); for (i = 0; i < npevs && ret >= 0; i++) - ret = show_available_vars_at(fd, &pevs[i], max_vls, externs); + ret = show_available_vars_at(fd, &pevs[i], max_vls, _filter, + externs); close(fd); return ret; @@ -531,7 +568,9 @@ int show_line_range(struct line_range *lr __unused, const char *module __unused) int show_available_vars(struct perf_probe_event *pevs __unused, int npevs __unused, int max_vls __unused, - const char *module __unused, bool externs __unused) + const char *module __unused, + struct strfilter *filter __unused, + bool externs __unused) { pr_warning("Debuginfo-analysis is not supported.\n"); return -ENOSYS; @@ -1912,4 +1951,46 @@ int del_perf_probe_events(struct strlist *dellist) return ret; } +/* TODO: don't use a global variable for filter ... */ +static struct strfilter *available_func_filter; +/* + * If a symbol corresponds to a function with global binding and + * matches filter return 0. For all others return 1. + */ +static int filter_available_functions(struct map *map __unused, + struct symbol *sym) +{ + if (sym->binding == STB_GLOBAL && + strfilter__compare(available_func_filter, sym->name)) + return 0; + return 1; +} + +int show_available_funcs(const char *module, struct strfilter *_filter) +{ + struct map *map; + int ret; + + setup_pager(); + + ret = init_vmlinux(); + if (ret < 0) + return ret; + + map = kernel_get_module_map(module); + if (!map) { + pr_err("Failed to find %s map.\n", (module) ? : "kernel"); + return -EINVAL; + } + available_func_filter = _filter; + if (map__load(map, filter_available_functions)) { + pr_err("Failed to load map.\n"); + return -EINVAL; + } + if (!dso__sorted_by_name(map->dso, map->type)) + dso__sort_by_name(map->dso, map->type); + + dso__fprintf_symbols_by_name(map->dso, map->type, stdout); + return 0; +} diff --git a/tools/perf/util/probe-event.h b/tools/perf/util/probe-event.h index 5accbedfea37..3434fc9d79d5 100644 --- a/tools/perf/util/probe-event.h +++ b/tools/perf/util/probe-event.h @@ -3,6 +3,7 @@ #include <stdbool.h> #include "strlist.h" +#include "strfilter.h" extern bool probe_event_dry_run; @@ -126,7 +127,8 @@ extern int show_perf_probe_events(void); extern int show_line_range(struct line_range *lr, const char *module); extern int show_available_vars(struct perf_probe_event *pevs, int npevs, int max_probe_points, const char *module, - bool externs); + struct strfilter *filter, bool externs); +extern int show_available_funcs(const char *module, struct strfilter *filter); /* Maximum index number of event-name postfix */ diff --git a/tools/perf/util/probe-finder.c b/tools/perf/util/probe-finder.c index ab83b6ac5d65..fe461f6559f1 100644 --- a/tools/perf/util/probe-finder.c +++ b/tools/perf/util/probe-finder.c @@ -33,6 +33,7 @@ #include <ctype.h> #include <dwarf-regs.h> +#include <linux/bitops.h> #include "event.h" #include "debug.h" #include "util.h" @@ -280,6 +281,19 @@ static bool die_compare_name(Dwarf_Die *dw_die, const char *tname) return name ? (strcmp(tname, name) == 0) : false; } +/* Get callsite line number of inline-function instance */ +static int die_get_call_lineno(Dwarf_Die *in_die) +{ + Dwarf_Attribute attr; + Dwarf_Word ret; + + if (!dwarf_attr(in_die, DW_AT_call_line, &attr)) + return -ENOENT; + + dwarf_formudata(&attr, &ret); + return (int)ret; +} + /* Get type die */ static Dwarf_Die *die_get_type(Dwarf_Die *vr_die, Dwarf_Die *die_mem) { @@ -320,13 +334,23 @@ static Dwarf_Die *die_get_real_type(Dwarf_Die *vr_die, Dwarf_Die *die_mem) return vr_die; } -static bool die_is_signed_type(Dwarf_Die *tp_die) +static int die_get_attr_udata(Dwarf_Die *tp_die, unsigned int attr_name, + Dwarf_Word *result) { Dwarf_Attribute attr; + + if (dwarf_attr(tp_die, attr_name, &attr) == NULL || + dwarf_formudata(&attr, result) != 0) + return -ENOENT; + + return 0; +} + +static bool die_is_signed_type(Dwarf_Die *tp_die) +{ Dwarf_Word ret; - if (dwarf_attr(tp_die, DW_AT_encoding, &attr) == NULL || - dwarf_formudata(&attr, &ret) != 0) + if (die_get_attr_udata(tp_die, DW_AT_encoding, &ret)) return false; return (ret == DW_ATE_signed_char || ret == DW_ATE_signed || @@ -335,11 +359,29 @@ static bool die_is_signed_type(Dwarf_Die *tp_die) static int die_get_byte_size(Dwarf_Die *tp_die) { - Dwarf_Attribute attr; Dwarf_Word ret; - if (dwarf_attr(tp_die, DW_AT_byte_size, &attr) == NULL || - dwarf_formudata(&attr, &ret) != 0) + if (die_get_attr_udata(tp_die, DW_AT_byte_size, &ret)) + return 0; + + return (int)ret; +} + +static int die_get_bit_size(Dwarf_Die *tp_die) +{ + Dwarf_Word ret; + + if (die_get_attr_udata(tp_die, DW_AT_bit_size, &ret)) + return 0; + + return (int)ret; +} + +static int die_get_bit_offset(Dwarf_Die *tp_die) +{ + Dwarf_Word ret; + + if (die_get_attr_udata(tp_die, DW_AT_bit_offset, &ret)) return 0; return (int)ret; @@ -458,6 +500,151 @@ static Dwarf_Die *die_find_inlinefunc(Dwarf_Die *sp_die, Dwarf_Addr addr, return die_find_child(sp_die, __die_find_inline_cb, &addr, die_mem); } +/* Walker on lines (Note: line number will not be sorted) */ +typedef int (* line_walk_handler_t) (const char *fname, int lineno, + Dwarf_Addr addr, void *data); + +struct __line_walk_param { + const char *fname; + line_walk_handler_t handler; + void *data; + int retval; +}; + +static int __die_walk_funclines_cb(Dwarf_Die *in_die, void *data) +{ + struct __line_walk_param *lw = data; + Dwarf_Addr addr; + int lineno; + + if (dwarf_tag(in_die) == DW_TAG_inlined_subroutine) { + lineno = die_get_call_lineno(in_die); + if (lineno > 0 && dwarf_entrypc(in_die, &addr) == 0) { + lw->retval = lw->handler(lw->fname, lineno, addr, + lw->data); + if (lw->retval != 0) + return DIE_FIND_CB_FOUND; + } + } + return DIE_FIND_CB_SIBLING; +} + +/* Walk on lines of blocks included in given DIE */ +static int __die_walk_funclines(Dwarf_Die *sp_die, + line_walk_handler_t handler, void *data) +{ + struct __line_walk_param lw = { + .handler = handler, + .data = data, + .retval = 0, + }; + Dwarf_Die die_mem; + Dwarf_Addr addr; + int lineno; + + /* Handle function declaration line */ + lw.fname = dwarf_decl_file(sp_die); + if (lw.fname && dwarf_decl_line(sp_die, &lineno) == 0 && + dwarf_entrypc(sp_die, &addr) == 0) { + lw.retval = handler(lw.fname, lineno, addr, data); + if (lw.retval != 0) + goto done; + } + die_find_child(sp_die, __die_walk_funclines_cb, &lw, &die_mem); +done: + return lw.retval; +} + +static int __die_walk_culines_cb(Dwarf_Die *sp_die, void *data) +{ + struct __line_walk_param *lw = data; + + lw->retval = __die_walk_funclines(sp_die, lw->handler, lw->data); + if (lw->retval != 0) + return DWARF_CB_ABORT; + + return DWARF_CB_OK; +} + +/* + * Walk on lines inside given PDIE. If the PDIE is subprogram, walk only on + * the lines inside the subprogram, otherwise PDIE must be a CU DIE. + */ +static int die_walk_lines(Dwarf_Die *pdie, line_walk_handler_t handler, + void *data) +{ + Dwarf_Lines *lines; + Dwarf_Line *line; + Dwarf_Addr addr; + const char *fname; + int lineno, ret = 0; + Dwarf_Die die_mem, *cu_die; + size_t nlines, i; + + /* Get the CU die */ + if (dwarf_tag(pdie) == DW_TAG_subprogram) + cu_die = dwarf_diecu(pdie, &die_mem, NULL, NULL); + else + cu_die = pdie; + if (!cu_die) { + pr_debug2("Failed to get CU from subprogram\n"); + return -EINVAL; + } + + /* Get lines list in the CU */ + if (dwarf_getsrclines(cu_die, &lines, &nlines) != 0) { + pr_debug2("Failed to get source lines on this CU.\n"); + return -ENOENT; + } + pr_debug2("Get %zd lines from this CU\n", nlines); + + /* Walk on the lines on lines list */ + for (i = 0; i < nlines; i++) { + line = dwarf_onesrcline(lines, i); + if (line == NULL || + dwarf_lineno(line, &lineno) != 0 || + dwarf_lineaddr(line, &addr) != 0) { + pr_debug2("Failed to get line info. " + "Possible error in debuginfo.\n"); + continue; + } + /* Filter lines based on address */ + if (pdie != cu_die) + /* + * Address filtering + * The line is included in given function, and + * no inline block includes it. + */ + if (!dwarf_haspc(pdie, addr) || + die_find_inlinefunc(pdie, addr, &die_mem)) + continue; + /* Get source line */ + fname = dwarf_linesrc(line, NULL, NULL); + + ret = handler(fname, lineno, addr, data); + if (ret != 0) + return ret; + } + + /* + * Dwarf lines doesn't include function declarations and inlined + * subroutines. We have to check functions list or given function. + */ + if (pdie != cu_die) + ret = __die_walk_funclines(pdie, handler, data); + else { + struct __line_walk_param param = { + .handler = handler, + .data = data, + .retval = 0, + }; + dwarf_getfuncs(cu_die, __die_walk_culines_cb, ¶m, 0); + ret = param.retval; + } + + return ret; +} + struct __find_variable_param { const char *name; Dwarf_Addr addr; @@ -669,6 +856,8 @@ static_var: return 0; } +#define BYTES_TO_BITS(nb) ((nb) * BITS_PER_LONG / sizeof(long)) + static int convert_variable_type(Dwarf_Die *vr_die, struct probe_trace_arg *tvar, const char *cast) @@ -685,6 +874,14 @@ static int convert_variable_type(Dwarf_Die *vr_die, return (tvar->type == NULL) ? -ENOMEM : 0; } + if (die_get_bit_size(vr_die) != 0) { + /* This is a bitfield */ + ret = snprintf(buf, 16, "b%d@%d/%zd", die_get_bit_size(vr_die), + die_get_bit_offset(vr_die), + BYTES_TO_BITS(die_get_byte_size(vr_die))); + goto formatted; + } + if (die_get_real_type(vr_die, &type) == NULL) { pr_warning("Failed to get a type information of %s.\n", dwarf_diename(vr_die)); @@ -729,29 +926,31 @@ static int convert_variable_type(Dwarf_Die *vr_die, return (tvar->type == NULL) ? -ENOMEM : 0; } - ret = die_get_byte_size(&type) * 8; - if (ret) { - /* Check the bitwidth */ - if (ret > MAX_BASIC_TYPE_BITS) { - pr_info("%s exceeds max-bitwidth." - " Cut down to %d bits.\n", - dwarf_diename(&type), MAX_BASIC_TYPE_BITS); - ret = MAX_BASIC_TYPE_BITS; - } + ret = BYTES_TO_BITS(die_get_byte_size(&type)); + if (!ret) + /* No size ... try to use default type */ + return 0; - ret = snprintf(buf, 16, "%c%d", - die_is_signed_type(&type) ? 's' : 'u', ret); - if (ret < 0 || ret >= 16) { - if (ret >= 16) - ret = -E2BIG; - pr_warning("Failed to convert variable type: %s\n", - strerror(-ret)); - return ret; - } - tvar->type = strdup(buf); - if (tvar->type == NULL) - return -ENOMEM; + /* Check the bitwidth */ + if (ret > MAX_BASIC_TYPE_BITS) { + pr_info("%s exceeds max-bitwidth. Cut down to %d bits.\n", + dwarf_diename(&type), MAX_BASIC_TYPE_BITS); + ret = MAX_BASIC_TYPE_BITS; + } + ret = snprintf(buf, 16, "%c%d", + die_is_signed_type(&type) ? 's' : 'u', ret); + +formatted: + if (ret < 0 || ret >= 16) { + if (ret >= 16) + ret = -E2BIG; + pr_warning("Failed to convert variable type: %s\n", + strerror(-ret)); + return ret; } + tvar->type = strdup(buf); + if (tvar->type == NULL) + return -ENOMEM; return 0; } @@ -1050,157 +1249,102 @@ static int call_probe_finder(Dwarf_Die *sp_die, struct probe_finder *pf) return ret; } -/* Find probe point from its line number */ -static int find_probe_point_by_line(struct probe_finder *pf) +static int probe_point_line_walker(const char *fname, int lineno, + Dwarf_Addr addr, void *data) { - Dwarf_Lines *lines; - Dwarf_Line *line; - size_t nlines, i; - Dwarf_Addr addr; - int lineno; - int ret = 0; - - if (dwarf_getsrclines(&pf->cu_die, &lines, &nlines) != 0) { - pr_warning("No source lines found.\n"); - return -ENOENT; - } + struct probe_finder *pf = data; + int ret; - for (i = 0; i < nlines && ret == 0; i++) { - line = dwarf_onesrcline(lines, i); - if (dwarf_lineno(line, &lineno) != 0 || - lineno != pf->lno) - continue; + if (lineno != pf->lno || strtailcmp(fname, pf->fname) != 0) + return 0; - /* TODO: Get fileno from line, but how? */ - if (strtailcmp(dwarf_linesrc(line, NULL, NULL), pf->fname) != 0) - continue; + pf->addr = addr; + ret = call_probe_finder(NULL, pf); - if (dwarf_lineaddr(line, &addr) != 0) { - pr_warning("Failed to get the address of the line.\n"); - return -ENOENT; - } - pr_debug("Probe line found: line[%d]:%d addr:0x%jx\n", - (int)i, lineno, (uintmax_t)addr); - pf->addr = addr; + /* Continue if no error, because the line will be in inline function */ + return ret < 0 ?: 0; +} - ret = call_probe_finder(NULL, pf); - /* Continuing, because target line might be inlined. */ - } - return ret; +/* Find probe point from its line number */ +static int find_probe_point_by_line(struct probe_finder *pf) +{ + return die_walk_lines(&pf->cu_die, probe_point_line_walker, pf); } /* Find lines which match lazy pattern */ static int find_lazy_match_lines(struct list_head *head, const char *fname, const char *pat) { - char *fbuf, *p1, *p2; - int fd, line, nlines = -1; - struct stat st; - - fd = open(fname, O_RDONLY); - if (fd < 0) { - pr_warning("Failed to open %s: %s\n", fname, strerror(-fd)); + FILE *fp; + char *line = NULL; + size_t line_len; + ssize_t len; + int count = 0, linenum = 1; + + fp = fopen(fname, "r"); + if (!fp) { + pr_warning("Failed to open %s: %s\n", fname, strerror(errno)); return -errno; } - if (fstat(fd, &st) < 0) { - pr_warning("Failed to get the size of %s: %s\n", - fname, strerror(errno)); - nlines = -errno; - goto out_close; - } - - nlines = -ENOMEM; - fbuf = malloc(st.st_size + 2); - if (fbuf == NULL) - goto out_close; - if (read(fd, fbuf, st.st_size) < 0) { - pr_warning("Failed to read %s: %s\n", fname, strerror(errno)); - nlines = -errno; - goto out_free_fbuf; - } - fbuf[st.st_size] = '\n'; /* Dummy line */ - fbuf[st.st_size + 1] = '\0'; - p1 = fbuf; - line = 1; - nlines = 0; - while ((p2 = strchr(p1, '\n')) != NULL) { - *p2 = '\0'; - if (strlazymatch(p1, pat)) { - line_list__add_line(head, line); - nlines++; + while ((len = getline(&line, &line_len, fp)) > 0) { + + if (line[len - 1] == '\n') + line[len - 1] = '\0'; + + if (strlazymatch(line, pat)) { + line_list__add_line(head, linenum); + count++; } - line++; - p1 = p2 + 1; + linenum++; } -out_free_fbuf: - free(fbuf); -out_close: - close(fd); - return nlines; + + if (ferror(fp)) + count = -errno; + free(line); + fclose(fp); + + if (count == 0) + pr_debug("No matched lines found in %s.\n", fname); + return count; +} + +static int probe_point_lazy_walker(const char *fname, int lineno, + Dwarf_Addr addr, void *data) +{ + struct probe_finder *pf = data; + int ret; + + if (!line_list__has_line(&pf->lcache, lineno) || + strtailcmp(fname, pf->fname) != 0) + return 0; + + pr_debug("Probe line found: line:%d addr:0x%llx\n", + lineno, (unsigned long long)addr); + pf->addr = addr; + ret = call_probe_finder(NULL, pf); + + /* + * Continue if no error, because the lazy pattern will match + * to other lines + */ + return ret < 0 ?: 0; } /* Find probe points from lazy pattern */ static int find_probe_point_lazy(Dwarf_Die *sp_die, struct probe_finder *pf) { - Dwarf_Lines *lines; - Dwarf_Line *line; - size_t nlines, i; - Dwarf_Addr addr; - Dwarf_Die die_mem; - int lineno; int ret = 0; if (list_empty(&pf->lcache)) { /* Matching lazy line pattern */ ret = find_lazy_match_lines(&pf->lcache, pf->fname, pf->pev->point.lazy_line); - if (ret == 0) { - pr_debug("No matched lines found in %s.\n", pf->fname); - return 0; - } else if (ret < 0) + if (ret <= 0) return ret; } - if (dwarf_getsrclines(&pf->cu_die, &lines, &nlines) != 0) { - pr_warning("No source lines found.\n"); - return -ENOENT; - } - - for (i = 0; i < nlines && ret >= 0; i++) { - line = dwarf_onesrcline(lines, i); - - if (dwarf_lineno(line, &lineno) != 0 || - !line_list__has_line(&pf->lcache, lineno)) - continue; - - /* TODO: Get fileno from line, but how? */ - if (strtailcmp(dwarf_linesrc(line, NULL, NULL), pf->fname) != 0) - continue; - - if (dwarf_lineaddr(line, &addr) != 0) { - pr_debug("Failed to get the address of line %d.\n", - lineno); - continue; - } - if (sp_die) { - /* Address filtering 1: does sp_die include addr? */ - if (!dwarf_haspc(sp_die, addr)) - continue; - /* Address filtering 2: No child include addr? */ - if (die_find_inlinefunc(sp_die, addr, &die_mem)) - continue; - } - - pr_debug("Probe line found: line[%d]:%d addr:0x%llx\n", - (int)i, lineno, (unsigned long long)addr); - pf->addr = addr; - - ret = call_probe_finder(sp_die, pf); - /* Continuing, because target line might be inlined. */ - } - /* TODO: deallocate lines, but how? */ - return ret; + return die_walk_lines(sp_die, probe_point_lazy_walker, pf); } /* Callback parameter with return value */ @@ -1644,91 +1788,28 @@ static int line_range_add_line(const char *src, unsigned int lineno, return line_list__add_line(&lr->line_list, lineno); } -/* Search function declaration lines */ -static int line_range_funcdecl_cb(Dwarf_Die *sp_die, void *data) +static int line_range_walk_cb(const char *fname, int lineno, + Dwarf_Addr addr __used, + void *data) { - struct dwarf_callback_param *param = data; - struct line_finder *lf = param->data; - const char *src; - int lineno; + struct line_finder *lf = data; - src = dwarf_decl_file(sp_die); - if (src && strtailcmp(src, lf->fname) != 0) - return DWARF_CB_OK; - - if (dwarf_decl_line(sp_die, &lineno) != 0 || + if ((strtailcmp(fname, lf->fname) != 0) || (lf->lno_s > lineno || lf->lno_e < lineno)) - return DWARF_CB_OK; + return 0; - param->retval = line_range_add_line(src, lineno, lf->lr); - if (param->retval < 0) - return DWARF_CB_ABORT; - return DWARF_CB_OK; -} + if (line_range_add_line(fname, lineno, lf->lr) < 0) + return -EINVAL; -static int find_line_range_func_decl_lines(struct line_finder *lf) -{ - struct dwarf_callback_param param = {.data = (void *)lf, .retval = 0}; - dwarf_getfuncs(&lf->cu_die, line_range_funcdecl_cb, ¶m, 0); - return param.retval; + return 0; } /* Find line range from its line number */ static int find_line_range_by_line(Dwarf_Die *sp_die, struct line_finder *lf) { - Dwarf_Lines *lines; - Dwarf_Line *line; - size_t nlines, i; - Dwarf_Addr addr; - int lineno, ret = 0; - const char *src; - Dwarf_Die die_mem; - - line_list__init(&lf->lr->line_list); - if (dwarf_getsrclines(&lf->cu_die, &lines, &nlines) != 0) { - pr_warning("No source lines found.\n"); - return -ENOENT; - } - - /* Search probable lines on lines list */ - for (i = 0; i < nlines; i++) { - line = dwarf_onesrcline(lines, i); - if (dwarf_lineno(line, &lineno) != 0 || - (lf->lno_s > lineno || lf->lno_e < lineno)) - continue; - - if (sp_die) { - /* Address filtering 1: does sp_die include addr? */ - if (dwarf_lineaddr(line, &addr) != 0 || - !dwarf_haspc(sp_die, addr)) - continue; - - /* Address filtering 2: No child include addr? */ - if (die_find_inlinefunc(sp_die, addr, &die_mem)) - continue; - } - - /* TODO: Get fileno from line, but how? */ - src = dwarf_linesrc(line, NULL, NULL); - if (strtailcmp(src, lf->fname) != 0) - continue; - - ret = line_range_add_line(src, lineno, lf->lr); - if (ret < 0) - return ret; - } + int ret; - /* - * Dwarf lines doesn't include function declarations. We have to - * check functions list or given function. - */ - if (sp_die) { - src = dwarf_decl_file(sp_die); - if (src && dwarf_decl_line(sp_die, &lineno) == 0 && - (lf->lno_s <= lineno && lf->lno_e >= lineno)) - ret = line_range_add_line(src, lineno, lf->lr); - } else - ret = find_line_range_func_decl_lines(lf); + ret = die_walk_lines(sp_die ?: &lf->cu_die, line_range_walk_cb, lf); /* Update status */ if (ret >= 0) @@ -1758,9 +1839,6 @@ static int line_range_search_cb(Dwarf_Die *sp_die, void *data) struct line_finder *lf = param->data; struct line_range *lr = lf->lr; - pr_debug("find (%llx) %s\n", - (unsigned long long)dwarf_dieoffset(sp_die), - dwarf_diename(sp_die)); if (dwarf_tag(sp_die) == DW_TAG_subprogram && die_compare_name(sp_die, lr->function)) { lf->fname = dwarf_decl_file(sp_die); diff --git a/tools/perf/util/python.c b/tools/perf/util/python.c new file mode 100644 index 000000000000..5317ef229041 --- /dev/null +++ b/tools/perf/util/python.c @@ -0,0 +1,891 @@ +#include <Python.h> +#include <structmember.h> +#include <inttypes.h> +#include <poll.h> +#include "evlist.h" +#include "evsel.h" +#include "event.h" +#include "cpumap.h" +#include "thread_map.h" + +struct throttle_event { + struct perf_event_header header; + u64 time; + u64 id; + u64 stream_id; +}; + +PyMODINIT_FUNC initperf(void); + +#define member_def(type, member, ptype, help) \ + { #member, ptype, \ + offsetof(struct pyrf_event, event) + offsetof(struct type, member), \ + 0, help } + +#define sample_member_def(name, member, ptype, help) \ + { #name, ptype, \ + offsetof(struct pyrf_event, sample) + offsetof(struct perf_sample, member), \ + 0, help } + +struct pyrf_event { + PyObject_HEAD + struct perf_sample sample; + union perf_event event; +}; + +#define sample_members \ + sample_member_def(sample_ip, ip, T_ULONGLONG, "event type"), \ + sample_member_def(sample_pid, pid, T_INT, "event pid"), \ + sample_member_def(sample_tid, tid, T_INT, "event tid"), \ + sample_member_def(sample_time, time, T_ULONGLONG, "event timestamp"), \ + sample_member_def(sample_addr, addr, T_ULONGLONG, "event addr"), \ + sample_member_def(sample_id, id, T_ULONGLONG, "event id"), \ + sample_member_def(sample_stream_id, stream_id, T_ULONGLONG, "event stream id"), \ + sample_member_def(sample_period, period, T_ULONGLONG, "event period"), \ + sample_member_def(sample_cpu, cpu, T_UINT, "event cpu"), + +static char pyrf_mmap_event__doc[] = PyDoc_STR("perf mmap event object."); + +static PyMemberDef pyrf_mmap_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(mmap_event, pid, T_UINT, "event pid"), + member_def(mmap_event, tid, T_UINT, "event tid"), + member_def(mmap_event, start, T_ULONGLONG, "start of the map"), + member_def(mmap_event, len, T_ULONGLONG, "map length"), + member_def(mmap_event, pgoff, T_ULONGLONG, "page offset"), + member_def(mmap_event, filename, T_STRING_INPLACE, "backing store"), + { .name = NULL, }, +}; + +static PyObject *pyrf_mmap_event__repr(struct pyrf_event *pevent) +{ + PyObject *ret; + char *s; + + if (asprintf(&s, "{ type: mmap, pid: %u, tid: %u, start: %#" PRIx64 ", " + "length: %#" PRIx64 ", offset: %#" PRIx64 ", " + "filename: %s }", + pevent->event.mmap.pid, pevent->event.mmap.tid, + pevent->event.mmap.start, pevent->event.mmap.len, + pevent->event.mmap.pgoff, pevent->event.mmap.filename) < 0) { + ret = PyErr_NoMemory(); + } else { + ret = PyString_FromString(s); + free(s); + } + return ret; +} + +static PyTypeObject pyrf_mmap_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.mmap_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_mmap_event__doc, + .tp_members = pyrf_mmap_event__members, + .tp_repr = (reprfunc)pyrf_mmap_event__repr, +}; + +static char pyrf_task_event__doc[] = PyDoc_STR("perf task (fork/exit) event object."); + +static PyMemberDef pyrf_task_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(fork_event, pid, T_UINT, "event pid"), + member_def(fork_event, ppid, T_UINT, "event ppid"), + member_def(fork_event, tid, T_UINT, "event tid"), + member_def(fork_event, ptid, T_UINT, "event ptid"), + member_def(fork_event, time, T_ULONGLONG, "timestamp"), + { .name = NULL, }, +}; + +static PyObject *pyrf_task_event__repr(struct pyrf_event *pevent) +{ + return PyString_FromFormat("{ type: %s, pid: %u, ppid: %u, tid: %u, " + "ptid: %u, time: %" PRIu64 "}", + pevent->event.header.type == PERF_RECORD_FORK ? "fork" : "exit", + pevent->event.fork.pid, + pevent->event.fork.ppid, + pevent->event.fork.tid, + pevent->event.fork.ptid, + pevent->event.fork.time); +} + +static PyTypeObject pyrf_task_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.task_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_task_event__doc, + .tp_members = pyrf_task_event__members, + .tp_repr = (reprfunc)pyrf_task_event__repr, +}; + +static char pyrf_comm_event__doc[] = PyDoc_STR("perf comm event object."); + +static PyMemberDef pyrf_comm_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(comm_event, pid, T_UINT, "event pid"), + member_def(comm_event, tid, T_UINT, "event tid"), + member_def(comm_event, comm, T_STRING_INPLACE, "process name"), + { .name = NULL, }, +}; + +static PyObject *pyrf_comm_event__repr(struct pyrf_event *pevent) +{ + return PyString_FromFormat("{ type: comm, pid: %u, tid: %u, comm: %s }", + pevent->event.comm.pid, + pevent->event.comm.tid, + pevent->event.comm.comm); +} + +static PyTypeObject pyrf_comm_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.comm_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_comm_event__doc, + .tp_members = pyrf_comm_event__members, + .tp_repr = (reprfunc)pyrf_comm_event__repr, +}; + +static char pyrf_throttle_event__doc[] = PyDoc_STR("perf throttle event object."); + +static PyMemberDef pyrf_throttle_event__members[] = { + sample_members + member_def(perf_event_header, type, T_UINT, "event type"), + member_def(throttle_event, time, T_ULONGLONG, "timestamp"), + member_def(throttle_event, id, T_ULONGLONG, "event id"), + member_def(throttle_event, stream_id, T_ULONGLONG, "event stream id"), + { .name = NULL, }, +}; + +static PyObject *pyrf_throttle_event__repr(struct pyrf_event *pevent) +{ + struct throttle_event *te = (struct throttle_event *)(&pevent->event.header + 1); + + return PyString_FromFormat("{ type: %sthrottle, time: %" PRIu64 ", id: %" PRIu64 + ", stream_id: %" PRIu64 " }", + pevent->event.header.type == PERF_RECORD_THROTTLE ? "" : "un", + te->time, te->id, te->stream_id); +} + +static PyTypeObject pyrf_throttle_event__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.throttle_event", + .tp_basicsize = sizeof(struct pyrf_event), + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_throttle_event__doc, + .tp_members = pyrf_throttle_event__members, + .tp_repr = (reprfunc)pyrf_throttle_event__repr, +}; + +static int pyrf_event__setup_types(void) +{ + int err; + pyrf_mmap_event__type.tp_new = + pyrf_task_event__type.tp_new = + pyrf_comm_event__type.tp_new = + pyrf_throttle_event__type.tp_new = PyType_GenericNew; + err = PyType_Ready(&pyrf_mmap_event__type); + if (err < 0) + goto out; + err = PyType_Ready(&pyrf_task_event__type); + if (err < 0) + goto out; + err = PyType_Ready(&pyrf_comm_event__type); + if (err < 0) + goto out; + err = PyType_Ready(&pyrf_throttle_event__type); + if (err < 0) + goto out; +out: + return err; +} + +static PyTypeObject *pyrf_event__type[] = { + [PERF_RECORD_MMAP] = &pyrf_mmap_event__type, + [PERF_RECORD_LOST] = &pyrf_mmap_event__type, + [PERF_RECORD_COMM] = &pyrf_comm_event__type, + [PERF_RECORD_EXIT] = &pyrf_task_event__type, + [PERF_RECORD_THROTTLE] = &pyrf_throttle_event__type, + [PERF_RECORD_UNTHROTTLE] = &pyrf_throttle_event__type, + [PERF_RECORD_FORK] = &pyrf_task_event__type, + [PERF_RECORD_READ] = &pyrf_mmap_event__type, + [PERF_RECORD_SAMPLE] = &pyrf_mmap_event__type, +}; + +static PyObject *pyrf_event__new(union perf_event *event) +{ + struct pyrf_event *pevent; + PyTypeObject *ptype; + + if (event->header.type < PERF_RECORD_MMAP || + event->header.type > PERF_RECORD_SAMPLE) + return NULL; + + ptype = pyrf_event__type[event->header.type]; + pevent = PyObject_New(struct pyrf_event, ptype); + if (pevent != NULL) + memcpy(&pevent->event, event, event->header.size); + return (PyObject *)pevent; +} + +struct pyrf_cpu_map { + PyObject_HEAD + + struct cpu_map *cpus; +}; + +static int pyrf_cpu_map__init(struct pyrf_cpu_map *pcpus, + PyObject *args, PyObject *kwargs) +{ + static char *kwlist[] = { "cpustr", NULL, NULL, }; + char *cpustr = NULL; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|s", + kwlist, &cpustr)) + return -1; + + pcpus->cpus = cpu_map__new(cpustr); + if (pcpus->cpus == NULL) + return -1; + return 0; +} + +static void pyrf_cpu_map__delete(struct pyrf_cpu_map *pcpus) +{ + cpu_map__delete(pcpus->cpus); + pcpus->ob_type->tp_free((PyObject*)pcpus); +} + +static Py_ssize_t pyrf_cpu_map__length(PyObject *obj) +{ + struct pyrf_cpu_map *pcpus = (void *)obj; + + return pcpus->cpus->nr; +} + +static PyObject *pyrf_cpu_map__item(PyObject *obj, Py_ssize_t i) +{ + struct pyrf_cpu_map *pcpus = (void *)obj; + + if (i >= pcpus->cpus->nr) + return NULL; + + return Py_BuildValue("i", pcpus->cpus->map[i]); +} + +static PySequenceMethods pyrf_cpu_map__sequence_methods = { + .sq_length = pyrf_cpu_map__length, + .sq_item = pyrf_cpu_map__item, +}; + +static char pyrf_cpu_map__doc[] = PyDoc_STR("cpu map object."); + +static PyTypeObject pyrf_cpu_map__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.cpu_map", + .tp_basicsize = sizeof(struct pyrf_cpu_map), + .tp_dealloc = (destructor)pyrf_cpu_map__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_cpu_map__doc, + .tp_as_sequence = &pyrf_cpu_map__sequence_methods, + .tp_init = (initproc)pyrf_cpu_map__init, +}; + +static int pyrf_cpu_map__setup_types(void) +{ + pyrf_cpu_map__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_cpu_map__type); +} + +struct pyrf_thread_map { + PyObject_HEAD + + struct thread_map *threads; +}; + +static int pyrf_thread_map__init(struct pyrf_thread_map *pthreads, + PyObject *args, PyObject *kwargs) +{ + static char *kwlist[] = { "pid", "tid", NULL, NULL, }; + int pid = -1, tid = -1; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ii", + kwlist, &pid, &tid)) + return -1; + + pthreads->threads = thread_map__new(pid, tid); + if (pthreads->threads == NULL) + return -1; + return 0; +} + +static void pyrf_thread_map__delete(struct pyrf_thread_map *pthreads) +{ + thread_map__delete(pthreads->threads); + pthreads->ob_type->tp_free((PyObject*)pthreads); +} + +static Py_ssize_t pyrf_thread_map__length(PyObject *obj) +{ + struct pyrf_thread_map *pthreads = (void *)obj; + + return pthreads->threads->nr; +} + +static PyObject *pyrf_thread_map__item(PyObject *obj, Py_ssize_t i) +{ + struct pyrf_thread_map *pthreads = (void *)obj; + + if (i >= pthreads->threads->nr) + return NULL; + + return Py_BuildValue("i", pthreads->threads->map[i]); +} + +static PySequenceMethods pyrf_thread_map__sequence_methods = { + .sq_length = pyrf_thread_map__length, + .sq_item = pyrf_thread_map__item, +}; + +static char pyrf_thread_map__doc[] = PyDoc_STR("thread map object."); + +static PyTypeObject pyrf_thread_map__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.thread_map", + .tp_basicsize = sizeof(struct pyrf_thread_map), + .tp_dealloc = (destructor)pyrf_thread_map__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_thread_map__doc, + .tp_as_sequence = &pyrf_thread_map__sequence_methods, + .tp_init = (initproc)pyrf_thread_map__init, +}; + +static int pyrf_thread_map__setup_types(void) +{ + pyrf_thread_map__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_thread_map__type); +} + +struct pyrf_evsel { + PyObject_HEAD + + struct perf_evsel evsel; +}; + +static int pyrf_evsel__init(struct pyrf_evsel *pevsel, + PyObject *args, PyObject *kwargs) +{ + struct perf_event_attr attr = { + .type = PERF_TYPE_HARDWARE, + .config = PERF_COUNT_HW_CPU_CYCLES, + .sample_type = PERF_SAMPLE_PERIOD | PERF_SAMPLE_TID, + }; + static char *kwlist[] = { + "type", + "config", + "sample_freq", + "sample_period", + "sample_type", + "read_format", + "disabled", + "inherit", + "pinned", + "exclusive", + "exclude_user", + "exclude_kernel", + "exclude_hv", + "exclude_idle", + "mmap", + "comm", + "freq", + "inherit_stat", + "enable_on_exec", + "task", + "watermark", + "precise_ip", + "mmap_data", + "sample_id_all", + "wakeup_events", + "bp_type", + "bp_addr", + "bp_len", NULL, NULL, }; + u64 sample_period = 0; + u32 disabled = 0, + inherit = 0, + pinned = 0, + exclusive = 0, + exclude_user = 0, + exclude_kernel = 0, + exclude_hv = 0, + exclude_idle = 0, + mmap = 0, + comm = 0, + freq = 1, + inherit_stat = 0, + enable_on_exec = 0, + task = 0, + watermark = 0, + precise_ip = 0, + mmap_data = 0, + sample_id_all = 1; + int idx = 0; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, + "|iKiKKiiiiiiiiiiiiiiiiiiiiiKK", kwlist, + &attr.type, &attr.config, &attr.sample_freq, + &sample_period, &attr.sample_type, + &attr.read_format, &disabled, &inherit, + &pinned, &exclusive, &exclude_user, + &exclude_kernel, &exclude_hv, &exclude_idle, + &mmap, &comm, &freq, &inherit_stat, + &enable_on_exec, &task, &watermark, + &precise_ip, &mmap_data, &sample_id_all, + &attr.wakeup_events, &attr.bp_type, + &attr.bp_addr, &attr.bp_len, &idx)) + return -1; + + /* union... */ + if (sample_period != 0) { + if (attr.sample_freq != 0) + return -1; /* FIXME: throw right exception */ + attr.sample_period = sample_period; + } + + /* Bitfields */ + attr.disabled = disabled; + attr.inherit = inherit; + attr.pinned = pinned; + attr.exclusive = exclusive; + attr.exclude_user = exclude_user; + attr.exclude_kernel = exclude_kernel; + attr.exclude_hv = exclude_hv; + attr.exclude_idle = exclude_idle; + attr.mmap = mmap; + attr.comm = comm; + attr.freq = freq; + attr.inherit_stat = inherit_stat; + attr.enable_on_exec = enable_on_exec; + attr.task = task; + attr.watermark = watermark; + attr.precise_ip = precise_ip; + attr.mmap_data = mmap_data; + attr.sample_id_all = sample_id_all; + + perf_evsel__init(&pevsel->evsel, &attr, idx); + return 0; +} + +static void pyrf_evsel__delete(struct pyrf_evsel *pevsel) +{ + perf_evsel__exit(&pevsel->evsel); + pevsel->ob_type->tp_free((PyObject*)pevsel); +} + +static PyObject *pyrf_evsel__open(struct pyrf_evsel *pevsel, + PyObject *args, PyObject *kwargs) +{ + struct perf_evsel *evsel = &pevsel->evsel; + struct cpu_map *cpus = NULL; + struct thread_map *threads = NULL; + PyObject *pcpus = NULL, *pthreads = NULL; + int group = 0, overwrite = 0; + static char *kwlist[] = {"cpus", "threads", "group", "overwrite", NULL, NULL}; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|OOii", kwlist, + &pcpus, &pthreads, &group, &overwrite)) + return NULL; + + if (pthreads != NULL) + threads = ((struct pyrf_thread_map *)pthreads)->threads; + + if (pcpus != NULL) + cpus = ((struct pyrf_cpu_map *)pcpus)->cpus; + + if (perf_evsel__open(evsel, cpus, threads, group, overwrite) < 0) { + PyErr_SetFromErrno(PyExc_OSError); + return NULL; + } + + Py_INCREF(Py_None); + return Py_None; +} + +static PyMethodDef pyrf_evsel__methods[] = { + { + .ml_name = "open", + .ml_meth = (PyCFunction)pyrf_evsel__open, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("open the event selector file descriptor table.") + }, + { .ml_name = NULL, } +}; + +static char pyrf_evsel__doc[] = PyDoc_STR("perf event selector list object."); + +static PyTypeObject pyrf_evsel__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.evsel", + .tp_basicsize = sizeof(struct pyrf_evsel), + .tp_dealloc = (destructor)pyrf_evsel__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_doc = pyrf_evsel__doc, + .tp_methods = pyrf_evsel__methods, + .tp_init = (initproc)pyrf_evsel__init, +}; + +static int pyrf_evsel__setup_types(void) +{ + pyrf_evsel__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_evsel__type); +} + +struct pyrf_evlist { + PyObject_HEAD + + struct perf_evlist evlist; +}; + +static int pyrf_evlist__init(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs __used) +{ + PyObject *pcpus = NULL, *pthreads = NULL; + struct cpu_map *cpus; + struct thread_map *threads; + + if (!PyArg_ParseTuple(args, "OO", &pcpus, &pthreads)) + return -1; + + threads = ((struct pyrf_thread_map *)pthreads)->threads; + cpus = ((struct pyrf_cpu_map *)pcpus)->cpus; + perf_evlist__init(&pevlist->evlist, cpus, threads); + return 0; +} + +static void pyrf_evlist__delete(struct pyrf_evlist *pevlist) +{ + perf_evlist__exit(&pevlist->evlist); + pevlist->ob_type->tp_free((PyObject*)pevlist); +} + +static PyObject *pyrf_evlist__mmap(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + static char *kwlist[] = {"pages", "overwrite", + NULL, NULL}; + int pages = 128, overwrite = false; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|ii", kwlist, + &pages, &overwrite)) + return NULL; + + if (perf_evlist__mmap(evlist, pages, overwrite) < 0) { + PyErr_SetFromErrno(PyExc_OSError); + return NULL; + } + + Py_INCREF(Py_None); + return Py_None; +} + +static PyObject *pyrf_evlist__poll(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + static char *kwlist[] = {"timeout", NULL, NULL}; + int timeout = -1, n; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "|i", kwlist, &timeout)) + return NULL; + + n = poll(evlist->pollfd, evlist->nr_fds, timeout); + if (n < 0) { + PyErr_SetFromErrno(PyExc_OSError); + return NULL; + } + + return Py_BuildValue("i", n); +} + +static PyObject *pyrf_evlist__get_pollfd(struct pyrf_evlist *pevlist, + PyObject *args __used, PyObject *kwargs __used) +{ + struct perf_evlist *evlist = &pevlist->evlist; + PyObject *list = PyList_New(0); + int i; + + for (i = 0; i < evlist->nr_fds; ++i) { + PyObject *file; + FILE *fp = fdopen(evlist->pollfd[i].fd, "r"); + + if (fp == NULL) + goto free_list; + + file = PyFile_FromFile(fp, "perf", "r", NULL); + if (file == NULL) + goto free_list; + + if (PyList_Append(list, file) != 0) { + Py_DECREF(file); + goto free_list; + } + + Py_DECREF(file); + } + + return list; +free_list: + return PyErr_NoMemory(); +} + + +static PyObject *pyrf_evlist__add(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs __used) +{ + struct perf_evlist *evlist = &pevlist->evlist; + PyObject *pevsel; + struct perf_evsel *evsel; + + if (!PyArg_ParseTuple(args, "O", &pevsel)) + return NULL; + + Py_INCREF(pevsel); + evsel = &((struct pyrf_evsel *)pevsel)->evsel; + evsel->idx = evlist->nr_entries; + perf_evlist__add(evlist, evsel); + + return Py_BuildValue("i", evlist->nr_entries); +} + +static PyObject *pyrf_evlist__read_on_cpu(struct pyrf_evlist *pevlist, + PyObject *args, PyObject *kwargs) +{ + struct perf_evlist *evlist = &pevlist->evlist; + union perf_event *event; + int sample_id_all = 1, cpu; + static char *kwlist[] = {"sample_id_all", NULL, NULL}; + + if (!PyArg_ParseTupleAndKeywords(args, kwargs, "i|i", kwlist, + &cpu, &sample_id_all)) + return NULL; + + event = perf_evlist__read_on_cpu(evlist, cpu); + if (event != NULL) { + struct perf_evsel *first; + PyObject *pyevent = pyrf_event__new(event); + struct pyrf_event *pevent = (struct pyrf_event *)pyevent; + + if (pyevent == NULL) + return PyErr_NoMemory(); + + first = list_entry(evlist->entries.next, struct perf_evsel, node); + perf_event__parse_sample(event, first->attr.sample_type, sample_id_all, + &pevent->sample); + return pyevent; + } + + Py_INCREF(Py_None); + return Py_None; +} + +static PyMethodDef pyrf_evlist__methods[] = { + { + .ml_name = "mmap", + .ml_meth = (PyCFunction)pyrf_evlist__mmap, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("mmap the file descriptor table.") + }, + { + .ml_name = "poll", + .ml_meth = (PyCFunction)pyrf_evlist__poll, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("poll the file descriptor table.") + }, + { + .ml_name = "get_pollfd", + .ml_meth = (PyCFunction)pyrf_evlist__get_pollfd, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("get the poll file descriptor table.") + }, + { + .ml_name = "add", + .ml_meth = (PyCFunction)pyrf_evlist__add, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("adds an event selector to the list.") + }, + { + .ml_name = "read_on_cpu", + .ml_meth = (PyCFunction)pyrf_evlist__read_on_cpu, + .ml_flags = METH_VARARGS | METH_KEYWORDS, + .ml_doc = PyDoc_STR("reads an event.") + }, + { .ml_name = NULL, } +}; + +static Py_ssize_t pyrf_evlist__length(PyObject *obj) +{ + struct pyrf_evlist *pevlist = (void *)obj; + + return pevlist->evlist.nr_entries; +} + +static PyObject *pyrf_evlist__item(PyObject *obj, Py_ssize_t i) +{ + struct pyrf_evlist *pevlist = (void *)obj; + struct perf_evsel *pos; + + if (i >= pevlist->evlist.nr_entries) + return NULL; + + list_for_each_entry(pos, &pevlist->evlist.entries, node) + if (i-- == 0) + break; + + return Py_BuildValue("O", container_of(pos, struct pyrf_evsel, evsel)); +} + +static PySequenceMethods pyrf_evlist__sequence_methods = { + .sq_length = pyrf_evlist__length, + .sq_item = pyrf_evlist__item, +}; + +static char pyrf_evlist__doc[] = PyDoc_STR("perf event selector list object."); + +static PyTypeObject pyrf_evlist__type = { + PyVarObject_HEAD_INIT(NULL, 0) + .tp_name = "perf.evlist", + .tp_basicsize = sizeof(struct pyrf_evlist), + .tp_dealloc = (destructor)pyrf_evlist__delete, + .tp_flags = Py_TPFLAGS_DEFAULT|Py_TPFLAGS_BASETYPE, + .tp_as_sequence = &pyrf_evlist__sequence_methods, + .tp_doc = pyrf_evlist__doc, + .tp_methods = pyrf_evlist__methods, + .tp_init = (initproc)pyrf_evlist__init, +}; + +static int pyrf_evlist__setup_types(void) +{ + pyrf_evlist__type.tp_new = PyType_GenericNew; + return PyType_Ready(&pyrf_evlist__type); +} + +static struct { + const char *name; + int value; +} perf__constants[] = { + { "TYPE_HARDWARE", PERF_TYPE_HARDWARE }, + { "TYPE_SOFTWARE", PERF_TYPE_SOFTWARE }, + { "TYPE_TRACEPOINT", PERF_TYPE_TRACEPOINT }, + { "TYPE_HW_CACHE", PERF_TYPE_HW_CACHE }, + { "TYPE_RAW", PERF_TYPE_RAW }, + { "TYPE_BREAKPOINT", PERF_TYPE_BREAKPOINT }, + + { "COUNT_HW_CPU_CYCLES", PERF_COUNT_HW_CPU_CYCLES }, + { "COUNT_HW_INSTRUCTIONS", PERF_COUNT_HW_INSTRUCTIONS }, + { "COUNT_HW_CACHE_REFERENCES", PERF_COUNT_HW_CACHE_REFERENCES }, + { "COUNT_HW_CACHE_MISSES", PERF_COUNT_HW_CACHE_MISSES }, + { "COUNT_HW_BRANCH_INSTRUCTIONS", PERF_COUNT_HW_BRANCH_INSTRUCTIONS }, + { "COUNT_HW_BRANCH_MISSES", PERF_COUNT_HW_BRANCH_MISSES }, + { "COUNT_HW_BUS_CYCLES", PERF_COUNT_HW_BUS_CYCLES }, + { "COUNT_HW_CACHE_L1D", PERF_COUNT_HW_CACHE_L1D }, + { "COUNT_HW_CACHE_L1I", PERF_COUNT_HW_CACHE_L1I }, + { "COUNT_HW_CACHE_LL", PERF_COUNT_HW_CACHE_LL }, + { "COUNT_HW_CACHE_DTLB", PERF_COUNT_HW_CACHE_DTLB }, + { "COUNT_HW_CACHE_ITLB", PERF_COUNT_HW_CACHE_ITLB }, + { "COUNT_HW_CACHE_BPU", PERF_COUNT_HW_CACHE_BPU }, + { "COUNT_HW_CACHE_OP_READ", PERF_COUNT_HW_CACHE_OP_READ }, + { "COUNT_HW_CACHE_OP_WRITE", PERF_COUNT_HW_CACHE_OP_WRITE }, + { "COUNT_HW_CACHE_OP_PREFETCH", PERF_COUNT_HW_CACHE_OP_PREFETCH }, + { "COUNT_HW_CACHE_RESULT_ACCESS", PERF_COUNT_HW_CACHE_RESULT_ACCESS }, + { "COUNT_HW_CACHE_RESULT_MISS", PERF_COUNT_HW_CACHE_RESULT_MISS }, + + { "COUNT_SW_CPU_CLOCK", PERF_COUNT_SW_CPU_CLOCK }, + { "COUNT_SW_TASK_CLOCK", PERF_COUNT_SW_TASK_CLOCK }, + { "COUNT_SW_PAGE_FAULTS", PERF_COUNT_SW_PAGE_FAULTS }, + { "COUNT_SW_CONTEXT_SWITCHES", PERF_COUNT_SW_CONTEXT_SWITCHES }, + { "COUNT_SW_CPU_MIGRATIONS", PERF_COUNT_SW_CPU_MIGRATIONS }, + { "COUNT_SW_PAGE_FAULTS_MIN", PERF_COUNT_SW_PAGE_FAULTS_MIN }, + { "COUNT_SW_PAGE_FAULTS_MAJ", PERF_COUNT_SW_PAGE_FAULTS_MAJ }, + { "COUNT_SW_ALIGNMENT_FAULTS", PERF_COUNT_SW_ALIGNMENT_FAULTS }, + { "COUNT_SW_EMULATION_FAULTS", PERF_COUNT_SW_EMULATION_FAULTS }, + + { "SAMPLE_IP", PERF_SAMPLE_IP }, + { "SAMPLE_TID", PERF_SAMPLE_TID }, + { "SAMPLE_TIME", PERF_SAMPLE_TIME }, + { "SAMPLE_ADDR", PERF_SAMPLE_ADDR }, + { "SAMPLE_READ", PERF_SAMPLE_READ }, + { "SAMPLE_CALLCHAIN", PERF_SAMPLE_CALLCHAIN }, + { "SAMPLE_ID", PERF_SAMPLE_ID }, + { "SAMPLE_CPU", PERF_SAMPLE_CPU }, + { "SAMPLE_PERIOD", PERF_SAMPLE_PERIOD }, + { "SAMPLE_STREAM_ID", PERF_SAMPLE_STREAM_ID }, + { "SAMPLE_RAW", PERF_SAMPLE_RAW }, + + { "FORMAT_TOTAL_TIME_ENABLED", PERF_FORMAT_TOTAL_TIME_ENABLED }, + { "FORMAT_TOTAL_TIME_RUNNING", PERF_FORMAT_TOTAL_TIME_RUNNING }, + { "FORMAT_ID", PERF_FORMAT_ID }, + { "FORMAT_GROUP", PERF_FORMAT_GROUP }, + + { "RECORD_MMAP", PERF_RECORD_MMAP }, + { "RECORD_LOST", PERF_RECORD_LOST }, + { "RECORD_COMM", PERF_RECORD_COMM }, + { "RECORD_EXIT", PERF_RECORD_EXIT }, + { "RECORD_THROTTLE", PERF_RECORD_THROTTLE }, + { "RECORD_UNTHROTTLE", PERF_RECORD_UNTHROTTLE }, + { "RECORD_FORK", PERF_RECORD_FORK }, + { "RECORD_READ", PERF_RECORD_READ }, + { "RECORD_SAMPLE", PERF_RECORD_SAMPLE }, + { .name = NULL, }, +}; + +static PyMethodDef perf__methods[] = { + { .ml_name = NULL, } +}; + +PyMODINIT_FUNC initperf(void) +{ + PyObject *obj; + int i; + PyObject *dict, *module = Py_InitModule("perf", perf__methods); + + if (module == NULL || + pyrf_event__setup_types() < 0 || + pyrf_evlist__setup_types() < 0 || + pyrf_evsel__setup_types() < 0 || + pyrf_thread_map__setup_types() < 0 || + pyrf_cpu_map__setup_types() < 0) + return; + + Py_INCREF(&pyrf_evlist__type); + PyModule_AddObject(module, "evlist", (PyObject*)&pyrf_evlist__type); + + Py_INCREF(&pyrf_evsel__type); + PyModule_AddObject(module, "evsel", (PyObject*)&pyrf_evsel__type); + + Py_INCREF(&pyrf_thread_map__type); + PyModule_AddObject(module, "thread_map", (PyObject*)&pyrf_thread_map__type); + + Py_INCREF(&pyrf_cpu_map__type); + PyModule_AddObject(module, "cpu_map", (PyObject*)&pyrf_cpu_map__type); + + dict = PyModule_GetDict(module); + if (dict == NULL) + goto error; + + for (i = 0; perf__constants[i].name != NULL; i++) { + obj = PyInt_FromLong(perf__constants[i].value); + if (obj == NULL) + goto error; + PyDict_SetItemString(dict, perf__constants[i].name, obj); + Py_DECREF(obj); + } + +error: + if (PyErr_Occurred()) + PyErr_SetString(PyExc_ImportError, "perf: Init failed!"); +} diff --git a/tools/perf/util/scripting-engines/trace-event-python.c b/tools/perf/util/scripting-engines/trace-event-python.c index c6d99334bdfa..2040b8538527 100644 --- a/tools/perf/util/scripting-engines/trace-event-python.c +++ b/tools/perf/util/scripting-engines/trace-event-python.c @@ -248,8 +248,7 @@ static void python_process_event(int cpu, void *data, context = PyCObject_FromVoidPtr(scripting_context, NULL); PyTuple_SetItem(t, n++, PyString_FromString(handler_name)); - PyTuple_SetItem(t, n++, - PyCObject_FromVoidPtr(scripting_context, NULL)); + PyTuple_SetItem(t, n++, context); if (handler) { PyTuple_SetItem(t, n++, PyInt_FromLong(cpu)); diff --git a/tools/perf/util/session.c b/tools/perf/util/session.c index 105f00bfd555..a3a871f7bda3 100644 --- a/tools/perf/util/session.c +++ b/tools/perf/util/session.c @@ -67,7 +67,7 @@ out_close: static void perf_session__id_header_size(struct perf_session *session) { - struct sample_data *data; + struct perf_sample *data; u64 sample_type = session->sample_type; u16 size = 0; @@ -165,7 +165,7 @@ struct perf_session *perf_session__new(const char *filename, int mode, } else if (mode == O_WRONLY) { /* * In O_RDONLY mode this will be performed when reading the - * kernel MMAP event, in event__process_mmap(). + * kernel MMAP event, in perf_event__process_mmap(). */ if (perf_session__create_kernel_maps(self) < 0) goto out_delete; @@ -242,17 +242,16 @@ static bool symbol__match_parent_regex(struct symbol *sym) return 0; } -struct map_symbol *perf_session__resolve_callchain(struct perf_session *self, - struct thread *thread, - struct ip_callchain *chain, - struct symbol **parent) +int perf_session__resolve_callchain(struct perf_session *self, + struct thread *thread, + struct ip_callchain *chain, + struct symbol **parent) { u8 cpumode = PERF_RECORD_MISC_USER; unsigned int i; - struct map_symbol *syms = calloc(chain->nr, sizeof(*syms)); + int err; - if (!syms) - return NULL; + callchain_cursor_reset(&self->callchain_cursor); for (i = 0; i < chain->nr; i++) { u64 ip = chain->ips[i]; @@ -281,30 +280,33 @@ struct map_symbol *perf_session__resolve_callchain(struct perf_session *self, *parent = al.sym; if (!symbol_conf.use_callchain) break; - syms[i].map = al.map; - syms[i].sym = al.sym; } + + err = callchain_cursor_append(&self->callchain_cursor, + ip, al.map, al.sym); + if (err) + return err; } - return syms; + return 0; } -static int process_event_synth_stub(event_t *event __used, +static int process_event_synth_stub(union perf_event *event __used, struct perf_session *session __used) { dump_printf(": unhandled!\n"); return 0; } -static int process_event_stub(event_t *event __used, - struct sample_data *sample __used, +static int process_event_stub(union perf_event *event __used, + struct perf_sample *sample __used, struct perf_session *session __used) { dump_printf(": unhandled!\n"); return 0; } -static int process_finished_round_stub(event_t *event __used, +static int process_finished_round_stub(union perf_event *event __used, struct perf_session *session __used, struct perf_event_ops *ops __used) { @@ -312,7 +314,7 @@ static int process_finished_round_stub(event_t *event __used, return 0; } -static int process_finished_round(event_t *event, +static int process_finished_round(union perf_event *event, struct perf_session *session, struct perf_event_ops *ops); @@ -329,7 +331,7 @@ static void perf_event_ops__fill_defaults(struct perf_event_ops *handler) if (handler->exit == NULL) handler->exit = process_event_stub; if (handler->lost == NULL) - handler->lost = event__process_lost; + handler->lost = perf_event__process_lost; if (handler->read == NULL) handler->read = process_event_stub; if (handler->throttle == NULL) @@ -363,98 +365,98 @@ void mem_bswap_64(void *src, int byte_size) } } -static void event__all64_swap(event_t *self) +static void perf_event__all64_swap(union perf_event *event) { - struct perf_event_header *hdr = &self->header; - mem_bswap_64(hdr + 1, self->header.size - sizeof(*hdr)); + struct perf_event_header *hdr = &event->header; + mem_bswap_64(hdr + 1, event->header.size - sizeof(*hdr)); } -static void event__comm_swap(event_t *self) +static void perf_event__comm_swap(union perf_event *event) { - self->comm.pid = bswap_32(self->comm.pid); - self->comm.tid = bswap_32(self->comm.tid); + event->comm.pid = bswap_32(event->comm.pid); + event->comm.tid = bswap_32(event->comm.tid); } -static void event__mmap_swap(event_t *self) +static void perf_event__mmap_swap(union perf_event *event) { - self->mmap.pid = bswap_32(self->mmap.pid); - self->mmap.tid = bswap_32(self->mmap.tid); - self->mmap.start = bswap_64(self->mmap.start); - self->mmap.len = bswap_64(self->mmap.len); - self->mmap.pgoff = bswap_64(self->mmap.pgoff); + event->mmap.pid = bswap_32(event->mmap.pid); + event->mmap.tid = bswap_32(event->mmap.tid); + event->mmap.start = bswap_64(event->mmap.start); + event->mmap.len = bswap_64(event->mmap.len); + event->mmap.pgoff = bswap_64(event->mmap.pgoff); } -static void event__task_swap(event_t *self) +static void perf_event__task_swap(union perf_event *event) { - self->fork.pid = bswap_32(self->fork.pid); - self->fork.tid = bswap_32(self->fork.tid); - self->fork.ppid = bswap_32(self->fork.ppid); - self->fork.ptid = bswap_32(self->fork.ptid); - self->fork.time = bswap_64(self->fork.time); + event->fork.pid = bswap_32(event->fork.pid); + event->fork.tid = bswap_32(event->fork.tid); + event->fork.ppid = bswap_32(event->fork.ppid); + event->fork.ptid = bswap_32(event->fork.ptid); + event->fork.time = bswap_64(event->fork.time); } -static void event__read_swap(event_t *self) +static void perf_event__read_swap(union perf_event *event) { - self->read.pid = bswap_32(self->read.pid); - self->read.tid = bswap_32(self->read.tid); - self->read.value = bswap_64(self->read.value); - self->read.time_enabled = bswap_64(self->read.time_enabled); - self->read.time_running = bswap_64(self->read.time_running); - self->read.id = bswap_64(self->read.id); + event->read.pid = bswap_32(event->read.pid); + event->read.tid = bswap_32(event->read.tid); + event->read.value = bswap_64(event->read.value); + event->read.time_enabled = bswap_64(event->read.time_enabled); + event->read.time_running = bswap_64(event->read.time_running); + event->read.id = bswap_64(event->read.id); } -static void event__attr_swap(event_t *self) +static void perf_event__attr_swap(union perf_event *event) { size_t size; - self->attr.attr.type = bswap_32(self->attr.attr.type); - self->attr.attr.size = bswap_32(self->attr.attr.size); - self->attr.attr.config = bswap_64(self->attr.attr.config); - self->attr.attr.sample_period = bswap_64(self->attr.attr.sample_period); - self->attr.attr.sample_type = bswap_64(self->attr.attr.sample_type); - self->attr.attr.read_format = bswap_64(self->attr.attr.read_format); - self->attr.attr.wakeup_events = bswap_32(self->attr.attr.wakeup_events); - self->attr.attr.bp_type = bswap_32(self->attr.attr.bp_type); - self->attr.attr.bp_addr = bswap_64(self->attr.attr.bp_addr); - self->attr.attr.bp_len = bswap_64(self->attr.attr.bp_len); - - size = self->header.size; - size -= (void *)&self->attr.id - (void *)self; - mem_bswap_64(self->attr.id, size); + event->attr.attr.type = bswap_32(event->attr.attr.type); + event->attr.attr.size = bswap_32(event->attr.attr.size); + event->attr.attr.config = bswap_64(event->attr.attr.config); + event->attr.attr.sample_period = bswap_64(event->attr.attr.sample_period); + event->attr.attr.sample_type = bswap_64(event->attr.attr.sample_type); + event->attr.attr.read_format = bswap_64(event->attr.attr.read_format); + event->attr.attr.wakeup_events = bswap_32(event->attr.attr.wakeup_events); + event->attr.attr.bp_type = bswap_32(event->attr.attr.bp_type); + event->attr.attr.bp_addr = bswap_64(event->attr.attr.bp_addr); + event->attr.attr.bp_len = bswap_64(event->attr.attr.bp_len); + + size = event->header.size; + size -= (void *)&event->attr.id - (void *)event; + mem_bswap_64(event->attr.id, size); } -static void event__event_type_swap(event_t *self) +static void perf_event__event_type_swap(union perf_event *event) { - self->event_type.event_type.event_id = - bswap_64(self->event_type.event_type.event_id); + event->event_type.event_type.event_id = + bswap_64(event->event_type.event_type.event_id); } -static void event__tracing_data_swap(event_t *self) +static void perf_event__tracing_data_swap(union perf_event *event) { - self->tracing_data.size = bswap_32(self->tracing_data.size); + event->tracing_data.size = bswap_32(event->tracing_data.size); } -typedef void (*event__swap_op)(event_t *self); - -static event__swap_op event__swap_ops[] = { - [PERF_RECORD_MMAP] = event__mmap_swap, - [PERF_RECORD_COMM] = event__comm_swap, - [PERF_RECORD_FORK] = event__task_swap, - [PERF_RECORD_EXIT] = event__task_swap, - [PERF_RECORD_LOST] = event__all64_swap, - [PERF_RECORD_READ] = event__read_swap, - [PERF_RECORD_SAMPLE] = event__all64_swap, - [PERF_RECORD_HEADER_ATTR] = event__attr_swap, - [PERF_RECORD_HEADER_EVENT_TYPE] = event__event_type_swap, - [PERF_RECORD_HEADER_TRACING_DATA] = event__tracing_data_swap, - [PERF_RECORD_HEADER_BUILD_ID] = NULL, - [PERF_RECORD_HEADER_MAX] = NULL, +typedef void (*perf_event__swap_op)(union perf_event *event); + +static perf_event__swap_op perf_event__swap_ops[] = { + [PERF_RECORD_MMAP] = perf_event__mmap_swap, + [PERF_RECORD_COMM] = perf_event__comm_swap, + [PERF_RECORD_FORK] = perf_event__task_swap, + [PERF_RECORD_EXIT] = perf_event__task_swap, + [PERF_RECORD_LOST] = perf_event__all64_swap, + [PERF_RECORD_READ] = perf_event__read_swap, + [PERF_RECORD_SAMPLE] = perf_event__all64_swap, + [PERF_RECORD_HEADER_ATTR] = perf_event__attr_swap, + [PERF_RECORD_HEADER_EVENT_TYPE] = perf_event__event_type_swap, + [PERF_RECORD_HEADER_TRACING_DATA] = perf_event__tracing_data_swap, + [PERF_RECORD_HEADER_BUILD_ID] = NULL, + [PERF_RECORD_HEADER_MAX] = NULL, }; struct sample_queue { u64 timestamp; u64 file_offset; - event_t *event; + union perf_event *event; struct list_head list; }; @@ -472,8 +474,8 @@ static void perf_session_free_sample_buffers(struct perf_session *session) } static int perf_session_deliver_event(struct perf_session *session, - event_t *event, - struct sample_data *sample, + union perf_event *event, + struct perf_sample *sample, struct perf_event_ops *ops, u64 file_offset); @@ -483,7 +485,7 @@ static void flush_sample_queue(struct perf_session *s, struct ordered_samples *os = &s->ordered_samples; struct list_head *head = &os->samples; struct sample_queue *tmp, *iter; - struct sample_data sample; + struct perf_sample sample; u64 limit = os->next_flush; u64 last_ts = os->last_sample ? os->last_sample->timestamp : 0ULL; @@ -494,7 +496,7 @@ static void flush_sample_queue(struct perf_session *s, if (iter->timestamp > limit) break; - event__parse_sample(iter->event, s, &sample); + perf_session__parse_sample(s, iter->event, &sample); perf_session_deliver_event(s, iter->event, &sample, ops, iter->file_offset); @@ -550,7 +552,7 @@ static void flush_sample_queue(struct perf_session *s, * Flush every events below timestamp 7 * etc... */ -static int process_finished_round(event_t *event __used, +static int process_finished_round(union perf_event *event __used, struct perf_session *session, struct perf_event_ops *ops) { @@ -607,12 +609,12 @@ static void __queue_event(struct sample_queue *new, struct perf_session *s) #define MAX_SAMPLE_BUFFER (64 * 1024 / sizeof(struct sample_queue)) -static int perf_session_queue_event(struct perf_session *s, event_t *event, - struct sample_data *data, u64 file_offset) +static int perf_session_queue_event(struct perf_session *s, union perf_event *event, + struct perf_sample *sample, u64 file_offset) { struct ordered_samples *os = &s->ordered_samples; struct list_head *sc = &os->sample_cache; - u64 timestamp = data->time; + u64 timestamp = sample->time; struct sample_queue *new; if (!timestamp || timestamp == ~0ULL) @@ -648,7 +650,7 @@ static int perf_session_queue_event(struct perf_session *s, event_t *event, return 0; } -static void callchain__printf(struct sample_data *sample) +static void callchain__printf(struct perf_sample *sample) { unsigned int i; @@ -660,8 +662,8 @@ static void callchain__printf(struct sample_data *sample) } static void perf_session__print_tstamp(struct perf_session *session, - event_t *event, - struct sample_data *sample) + union perf_event *event, + struct perf_sample *sample) { if (event->header.type != PERF_RECORD_SAMPLE && !session->sample_id_all) { @@ -676,8 +678,8 @@ static void perf_session__print_tstamp(struct perf_session *session, printf("%" PRIu64 " ", sample->time); } -static void dump_event(struct perf_session *session, event_t *event, - u64 file_offset, struct sample_data *sample) +static void dump_event(struct perf_session *session, union perf_event *event, + u64 file_offset, struct perf_sample *sample) { if (!dump_trace) return; @@ -691,11 +693,11 @@ static void dump_event(struct perf_session *session, event_t *event, perf_session__print_tstamp(session, event, sample); printf("%#" PRIx64 " [%#x]: PERF_RECORD_%s", file_offset, - event->header.size, event__get_event_name(event->header.type)); + event->header.size, perf_event__name(event->header.type)); } -static void dump_sample(struct perf_session *session, event_t *event, - struct sample_data *sample) +static void dump_sample(struct perf_session *session, union perf_event *event, + struct perf_sample *sample) { if (!dump_trace) return; @@ -709,8 +711,8 @@ static void dump_sample(struct perf_session *session, event_t *event, } static int perf_session_deliver_event(struct perf_session *session, - event_t *event, - struct sample_data *sample, + union perf_event *event, + struct perf_sample *sample, struct perf_event_ops *ops, u64 file_offset) { @@ -743,7 +745,7 @@ static int perf_session_deliver_event(struct perf_session *session, } static int perf_session__preprocess_sample(struct perf_session *session, - event_t *event, struct sample_data *sample) + union perf_event *event, struct perf_sample *sample) { if (event->header.type != PERF_RECORD_SAMPLE || !(session->sample_type & PERF_SAMPLE_CALLCHAIN)) @@ -758,7 +760,7 @@ static int perf_session__preprocess_sample(struct perf_session *session, return 0; } -static int perf_session__process_user_event(struct perf_session *session, event_t *event, +static int perf_session__process_user_event(struct perf_session *session, union perf_event *event, struct perf_event_ops *ops, u64 file_offset) { dump_event(session, event, file_offset, NULL); @@ -783,15 +785,16 @@ static int perf_session__process_user_event(struct perf_session *session, event_ } static int perf_session__process_event(struct perf_session *session, - event_t *event, + union perf_event *event, struct perf_event_ops *ops, u64 file_offset) { - struct sample_data sample; + struct perf_sample sample; int ret; - if (session->header.needs_swap && event__swap_ops[event->header.type]) - event__swap_ops[event->header.type](event); + if (session->header.needs_swap && + perf_event__swap_ops[event->header.type]) + perf_event__swap_ops[event->header.type](event); if (event->header.type >= PERF_RECORD_HEADER_MAX) return -EINVAL; @@ -804,7 +807,7 @@ static int perf_session__process_event(struct perf_session *session, /* * For all kernel events we get the sample data */ - event__parse_sample(event, session, &sample); + perf_session__parse_sample(session, event, &sample); /* Preprocess sample records - precheck callchains */ if (perf_session__preprocess_sample(session, event, &sample)) @@ -843,7 +846,7 @@ 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) { - if (ops->lost == event__process_lost && + if (ops->lost == perf_event__process_lost && session->hists.stats.total_lost != 0) { ui__warning("Processed %" PRIu64 " events and LOST %" PRIu64 "!\n\nCheck IO/CPU overload!\n\n", @@ -875,7 +878,7 @@ volatile int session_done; static int __perf_session__process_pipe_events(struct perf_session *self, struct perf_event_ops *ops) { - event_t event; + union perf_event event; uint32_t size; int skip = 0; u64 head; @@ -956,7 +959,7 @@ int __perf_session__process_events(struct perf_session *session, struct ui_progress *progress; size_t page_size, mmap_size; char *buf, *mmaps[8]; - event_t *event; + union perf_event *event; uint32_t size; perf_event_ops__fill_defaults(ops); @@ -1001,7 +1004,7 @@ remap: file_pos = file_offset + head; more: - event = (event_t *)(buf + head); + event = (union perf_event *)(buf + head); if (session->header.needs_swap) perf_event_header__bswap(&event->header); diff --git a/tools/perf/util/session.h b/tools/perf/util/session.h index decd83f274fd..977b3a1b14aa 100644 --- a/tools/perf/util/session.h +++ b/tools/perf/util/session.h @@ -51,15 +51,17 @@ struct perf_session { int cwdlen; char *cwd; struct ordered_samples ordered_samples; - char filename[0]; + struct callchain_cursor callchain_cursor; + char filename[0]; }; struct perf_event_ops; -typedef int (*event_op)(event_t *self, struct sample_data *sample, +typedef int (*event_op)(union perf_event *self, struct perf_sample *sample, struct perf_session *session); -typedef int (*event_synth_op)(event_t *self, struct perf_session *session); -typedef int (*event_op2)(event_t *self, struct perf_session *session, +typedef int (*event_synth_op)(union perf_event *self, + struct perf_session *session); +typedef int (*event_op2)(union perf_event *self, struct perf_session *session, struct perf_event_ops *ops); struct perf_event_ops { @@ -94,10 +96,10 @@ int __perf_session__process_events(struct perf_session *self, int perf_session__process_events(struct perf_session *self, struct perf_event_ops *event_ops); -struct map_symbol *perf_session__resolve_callchain(struct perf_session *self, - struct thread *thread, - struct ip_callchain *chain, - struct symbol **parent); +int perf_session__resolve_callchain(struct perf_session *self, + struct thread *thread, + struct ip_callchain *chain, + struct symbol **parent); bool perf_session__has_traces(struct perf_session *self, const char *msg); @@ -154,4 +156,13 @@ size_t perf_session__fprintf_nr_events(struct perf_session *self, FILE *fp) { return hists__fprintf_nr_events(&self->hists, fp); } + +static inline int perf_session__parse_sample(struct perf_session *session, + const union perf_event *event, + struct perf_sample *sample) +{ + return perf_event__parse_sample(event, session->sample_type, + session->sample_id_all, sample); +} + #endif /* __PERF_SESSION_H */ diff --git a/tools/perf/util/setup.py b/tools/perf/util/setup.py new file mode 100644 index 000000000000..1947b0430c94 --- /dev/null +++ b/tools/perf/util/setup.py @@ -0,0 +1,19 @@ +#!/usr/bin/python2 + +from distutils.core import setup, Extension + +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'], + include_dirs = ['util/include'], + extra_compile_args = ['-fno-strict-aliasing', '-Wno-write-strings']) + +setup(name='perf', + version='0.1', + description='Interface with the Linux profiling infrastructure', + author='Arnaldo Carvalho de Melo', + author_email='acme@redhat.com', + license='GPLv2', + url='http://perf.wiki.kernel.org', + ext_modules=[perf]) diff --git a/tools/perf/util/strfilter.c b/tools/perf/util/strfilter.c new file mode 100644 index 000000000000..4064b7d708b8 --- /dev/null +++ b/tools/perf/util/strfilter.c @@ -0,0 +1,200 @@ +#include <ctype.h> +#include "util.h" +#include "string.h" +#include "strfilter.h" + +/* Operators */ +static const char *OP_and = "&"; /* Logical AND */ +static const char *OP_or = "|"; /* Logical OR */ +static const char *OP_not = "!"; /* Logical NOT */ + +#define is_operator(c) ((c) == '|' || (c) == '&' || (c) == '!') +#define is_separator(c) (is_operator(c) || (c) == '(' || (c) == ')') + +static void strfilter_node__delete(struct strfilter_node *self) +{ + if (self) { + if (self->p && !is_operator(*self->p)) + free((char *)self->p); + strfilter_node__delete(self->l); + strfilter_node__delete(self->r); + free(self); + } +} + +void strfilter__delete(struct strfilter *self) +{ + if (self) { + strfilter_node__delete(self->root); + free(self); + } +} + +static const char *get_token(const char *s, const char **e) +{ + const char *p; + + while (isspace(*s)) /* Skip spaces */ + s++; + + if (*s == '\0') { + p = s; + goto end; + } + + p = s + 1; + if (!is_separator(*s)) { + /* End search */ +retry: + while (*p && !is_separator(*p) && !isspace(*p)) + p++; + /* Escape and special case: '!' is also used in glob pattern */ + if (*(p - 1) == '\\' || (*p == '!' && *(p - 1) == '[')) { + p++; + goto retry; + } + } +end: + *e = p; + return s; +} + +static struct strfilter_node *strfilter_node__alloc(const char *op, + struct strfilter_node *l, + struct strfilter_node *r) +{ + struct strfilter_node *ret = zalloc(sizeof(struct strfilter_node)); + + if (ret) { + ret->p = op; + ret->l = l; + ret->r = r; + } + + return ret; +} + +static struct strfilter_node *strfilter_node__new(const char *s, + const char **ep) +{ + struct strfilter_node root, *cur, *last_op; + const char *e; + + if (!s) + return NULL; + + memset(&root, 0, sizeof(root)); + last_op = cur = &root; + + s = get_token(s, &e); + while (*s != '\0' && *s != ')') { + switch (*s) { + case '&': /* Exchg last OP->r with AND */ + if (!cur->r || !last_op->r) + goto error; + cur = strfilter_node__alloc(OP_and, last_op->r, NULL); + if (!cur) + goto nomem; + last_op->r = cur; + last_op = cur; + break; + case '|': /* Exchg the root with OR */ + if (!cur->r || !root.r) + goto error; + cur = strfilter_node__alloc(OP_or, root.r, NULL); + if (!cur) + goto nomem; + root.r = cur; + last_op = cur; + break; + case '!': /* Add NOT as a leaf node */ + if (cur->r) + goto error; + cur->r = strfilter_node__alloc(OP_not, NULL, NULL); + if (!cur->r) + goto nomem; + cur = cur->r; + break; + case '(': /* Recursively parses inside the parenthesis */ + if (cur->r) + goto error; + cur->r = strfilter_node__new(s + 1, &s); + if (!s) + goto nomem; + if (!cur->r || *s != ')') + goto error; + e = s + 1; + break; + default: + if (cur->r) + goto error; + cur->r = strfilter_node__alloc(NULL, NULL, NULL); + if (!cur->r) + goto nomem; + cur->r->p = strndup(s, e - s); + if (!cur->r->p) + goto nomem; + } + s = get_token(e, &e); + } + if (!cur->r) + goto error; + *ep = s; + return root.r; +nomem: + s = NULL; +error: + *ep = s; + strfilter_node__delete(root.r); + return NULL; +} + +/* + * Parse filter rule and return new strfilter. + * Return NULL if fail, and *ep == NULL if memory allocation failed. + */ +struct strfilter *strfilter__new(const char *rules, const char **err) +{ + struct strfilter *ret = zalloc(sizeof(struct strfilter)); + const char *ep = NULL; + + if (ret) + ret->root = strfilter_node__new(rules, &ep); + + if (!ret || !ret->root || *ep != '\0') { + if (err) + *err = ep; + strfilter__delete(ret); + ret = NULL; + } + + return ret; +} + +static bool strfilter_node__compare(struct strfilter_node *self, + const char *str) +{ + if (!self || !self->p) + return false; + + switch (*self->p) { + case '|': /* OR */ + return strfilter_node__compare(self->l, str) || + strfilter_node__compare(self->r, str); + case '&': /* AND */ + return strfilter_node__compare(self->l, str) && + strfilter_node__compare(self->r, str); + case '!': /* NOT */ + return !strfilter_node__compare(self->r, str); + default: + return strglobmatch(str, self->p); + } +} + +/* Return true if STR matches the filter rules */ +bool strfilter__compare(struct strfilter *self, const char *str) +{ + if (!self) + return false; + return strfilter_node__compare(self->root, str); +} diff --git a/tools/perf/util/strfilter.h b/tools/perf/util/strfilter.h new file mode 100644 index 000000000000..00f58a7506de --- /dev/null +++ b/tools/perf/util/strfilter.h @@ -0,0 +1,48 @@ +#ifndef __PERF_STRFILTER_H +#define __PERF_STRFILTER_H +/* General purpose glob matching filter */ + +#include <linux/list.h> +#include <stdbool.h> + +/* A node of string filter */ +struct strfilter_node { + struct strfilter_node *l; /* Tree left branche (for &,|) */ + struct strfilter_node *r; /* Tree right branche (for !,&,|) */ + const char *p; /* Operator or rule */ +}; + +/* String filter */ +struct strfilter { + struct strfilter_node *root; +}; + +/** + * strfilter__new - Create a new string filter + * @rules: Filter rule, which is a combination of glob expressions. + * @err: Pointer which points an error detected on @rules + * + * Parse @rules and return new strfilter. Return NULL if an error detected. + * In that case, *@err will indicate where it is detected, and *@err is NULL + * if a memory allocation is failed. + */ +struct strfilter *strfilter__new(const char *rules, const char **err); + +/** + * strfilter__compare - compare given string and a string filter + * @self: String filter + * @str: target string + * + * Compare @str and @self. Return true if the str match the rule + */ +bool strfilter__compare(struct strfilter *self, const char *str); + +/** + * strfilter__delete - delete a string filter + * @self: String filter to delete + * + * Delete @self. + */ +void strfilter__delete(struct strfilter *self); + +#endif diff --git a/tools/perf/util/symbol.c b/tools/perf/util/symbol.c index 7821d0e6866f..3e193f8e3061 100644 --- a/tools/perf/util/symbol.c +++ b/tools/perf/util/symbol.c @@ -1525,8 +1525,8 @@ int dso__load(struct dso *self, struct map *map, symbol_filter_t filter) symbol_conf.symfs, self->long_name); break; case DSO__ORIG_GUEST_KMODULE: - if (map->groups && map->groups->machine) - root_dir = map->groups->machine->root_dir; + if (map->groups && machine) + root_dir = machine->root_dir; else root_dir = ""; snprintf(name, size, "%s%s%s", symbol_conf.symfs, diff --git a/tools/perf/util/thread.c b/tools/perf/util/thread.c index 00f4eade2e3e..d5d3b22250f3 100644 --- a/tools/perf/util/thread.c +++ b/tools/perf/util/thread.c @@ -7,61 +7,6 @@ #include "util.h" #include "debug.h" -/* Skip "." and ".." directories */ -static int filter(const struct dirent *dir) -{ - if (dir->d_name[0] == '.') - return 0; - else - return 1; -} - -struct thread_map *thread_map__new_by_pid(pid_t pid) -{ - struct thread_map *threads; - char name[256]; - int items; - struct dirent **namelist = NULL; - int i; - - sprintf(name, "/proc/%d/task", pid); - items = scandir(name, &namelist, filter, NULL); - if (items <= 0) - return NULL; - - threads = malloc(sizeof(*threads) + sizeof(pid_t) * items); - if (threads != NULL) { - for (i = 0; i < items; i++) - threads->map[i] = atoi(namelist[i]->d_name); - threads->nr = items; - } - - for (i=0; i<items; i++) - free(namelist[i]); - free(namelist); - - return threads; -} - -struct thread_map *thread_map__new_by_tid(pid_t tid) -{ - struct thread_map *threads = malloc(sizeof(*threads) + sizeof(pid_t)); - - if (threads != NULL) { - threads->map[0] = tid; - threads->nr = 1; - } - - return threads; -} - -struct thread_map *thread_map__new(pid_t pid, pid_t tid) -{ - if (pid != -1) - return thread_map__new_by_pid(pid); - return thread_map__new_by_tid(tid); -} - static struct thread *thread__new(pid_t pid) { struct thread *self = zalloc(sizeof(*self)); diff --git a/tools/perf/util/thread.h b/tools/perf/util/thread.h index d7574101054a..e5f2401c1b5e 100644 --- a/tools/perf/util/thread.h +++ b/tools/perf/util/thread.h @@ -18,24 +18,10 @@ struct thread { int comm_len; }; -struct thread_map { - int nr; - int map[]; -}; - struct perf_session; void thread__delete(struct thread *self); -struct thread_map *thread_map__new_by_pid(pid_t pid); -struct thread_map *thread_map__new_by_tid(pid_t tid); -struct thread_map *thread_map__new(pid_t pid, pid_t tid); - -static inline void thread_map__delete(struct thread_map *threads) -{ - free(threads); -} - 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); diff --git a/tools/perf/util/thread_map.c b/tools/perf/util/thread_map.c new file mode 100644 index 000000000000..a5df131b77c3 --- /dev/null +++ b/tools/perf/util/thread_map.c @@ -0,0 +1,64 @@ +#include <dirent.h> +#include <stdlib.h> +#include <stdio.h> +#include "thread_map.h" + +/* Skip "." and ".." directories */ +static int filter(const struct dirent *dir) +{ + if (dir->d_name[0] == '.') + return 0; + else + return 1; +} + +struct thread_map *thread_map__new_by_pid(pid_t pid) +{ + struct thread_map *threads; + char name[256]; + int items; + struct dirent **namelist = NULL; + int i; + + sprintf(name, "/proc/%d/task", pid); + items = scandir(name, &namelist, filter, NULL); + if (items <= 0) + return NULL; + + threads = malloc(sizeof(*threads) + sizeof(pid_t) * items); + if (threads != NULL) { + for (i = 0; i < items; i++) + threads->map[i] = atoi(namelist[i]->d_name); + threads->nr = items; + } + + for (i=0; i<items; i++) + free(namelist[i]); + free(namelist); + + return threads; +} + +struct thread_map *thread_map__new_by_tid(pid_t tid) +{ + struct thread_map *threads = malloc(sizeof(*threads) + sizeof(pid_t)); + + if (threads != NULL) { + threads->map[0] = tid; + threads->nr = 1; + } + + return threads; +} + +struct thread_map *thread_map__new(pid_t pid, pid_t tid) +{ + if (pid != -1) + return thread_map__new_by_pid(pid); + return thread_map__new_by_tid(tid); +} + +void thread_map__delete(struct thread_map *threads) +{ + free(threads); +} diff --git a/tools/perf/util/thread_map.h b/tools/perf/util/thread_map.h new file mode 100644 index 000000000000..3cb907311409 --- /dev/null +++ b/tools/perf/util/thread_map.h @@ -0,0 +1,15 @@ +#ifndef __PERF_THREAD_MAP_H +#define __PERF_THREAD_MAP_H + +#include <sys/types.h> + +struct thread_map { + int nr; + int map[]; +}; + +struct thread_map *thread_map__new_by_pid(pid_t pid); +struct thread_map *thread_map__new_by_tid(pid_t tid); +struct thread_map *thread_map__new(pid_t pid, pid_t tid); +void thread_map__delete(struct thread_map *threads); +#endif /* __PERF_THREAD_MAP_H */ diff --git a/tools/perf/util/top.c b/tools/perf/util/top.c new file mode 100644 index 000000000000..70a9c13f4ad5 --- /dev/null +++ b/tools/perf/util/top.c @@ -0,0 +1,218 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Refactored from builtin-top.c, see that files for further copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ + +#include "cpumap.h" +#include "event.h" +#include "evlist.h" +#include "evsel.h" +#include "parse-events.h" +#include "symbol.h" +#include "top.h" +#include <inttypes.h> + +/* + * Ordering weight: count-1 * count-2 * ... / count-n + */ +static double sym_weight(const struct sym_entry *sym, struct perf_top *top) +{ + double weight = sym->snap_count; + int counter; + + if (!top->display_weighted) + return weight; + + for (counter = 1; counter < top->evlist->nr_entries - 1; counter++) + weight *= sym->count[counter]; + + weight /= (sym->count[counter] + 1); + + return weight; +} + +static void perf_top__remove_active_sym(struct perf_top *top, struct sym_entry *syme) +{ + pthread_mutex_lock(&top->active_symbols_lock); + list_del_init(&syme->node); + pthread_mutex_unlock(&top->active_symbols_lock); +} + +static void rb_insert_active_sym(struct rb_root *tree, struct sym_entry *se) +{ + struct rb_node **p = &tree->rb_node; + struct rb_node *parent = NULL; + struct sym_entry *iter; + + while (*p != NULL) { + parent = *p; + iter = rb_entry(parent, struct sym_entry, rb_node); + + if (se->weight > iter->weight) + p = &(*p)->rb_left; + else + p = &(*p)->rb_right; + } + + rb_link_node(&se->rb_node, parent, p); + rb_insert_color(&se->rb_node, tree); +} + +size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size) +{ + struct perf_evsel *counter; + float samples_per_sec = top->samples / top->delay_secs; + float ksamples_per_sec = top->kernel_samples / top->delay_secs; + float esamples_percent = (100.0 * top->exact_samples) / top->samples; + size_t ret = 0; + + if (!perf_guest) { + ret = snprintf(bf, size, + " PerfTop:%8.0f irqs/sec kernel:%4.1f%%" + " exact: %4.1f%% [", samples_per_sec, + 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) / + samples_per_sec)), + esamples_percent); + } else { + float us_samples_per_sec = top->us_samples / top->delay_secs; + float guest_kernel_samples_per_sec = top->guest_kernel_samples / top->delay_secs; + float guest_us_samples_per_sec = top->guest_us_samples / top->delay_secs; + + ret = snprintf(bf, size, + " PerfTop:%8.0f irqs/sec kernel:%4.1f%% us:%4.1f%%" + " guest kernel:%4.1f%% guest us:%4.1f%%" + " exact: %4.1f%% [", samples_per_sec, + 100.0 - (100.0 * ((samples_per_sec - ksamples_per_sec) / + samples_per_sec)), + 100.0 - (100.0 * ((samples_per_sec - us_samples_per_sec) / + samples_per_sec)), + 100.0 - (100.0 * ((samples_per_sec - + guest_kernel_samples_per_sec) / + samples_per_sec)), + 100.0 - (100.0 * ((samples_per_sec - + guest_us_samples_per_sec) / + samples_per_sec)), + esamples_percent); + } + + if (top->evlist->nr_entries == 1 || !top->display_weighted) { + struct perf_evsel *first; + first = list_entry(top->evlist->entries.next, struct perf_evsel, node); + ret += snprintf(bf + ret, size - ret, "%" PRIu64 "%s ", + (uint64_t)first->attr.sample_period, + top->freq ? "Hz" : ""); + } + + if (!top->display_weighted) { + ret += snprintf(bf + ret, size - ret, "%s", + event_name(top->sym_evsel)); + } else list_for_each_entry(counter, &top->evlist->entries, node) { + ret += snprintf(bf + ret, size - ret, "%s%s", + counter->idx ? "/" : "", event_name(counter)); + } + + ret += snprintf(bf + ret, size - ret, "], "); + + if (top->target_pid != -1) + ret += snprintf(bf + ret, size - ret, " (target_pid: %d", + top->target_pid); + else if (top->target_tid != -1) + ret += snprintf(bf + ret, size - ret, " (target_tid: %d", + top->target_tid); + else + ret += snprintf(bf + ret, size - ret, " (all"); + + if (top->cpu_list) + ret += snprintf(bf + ret, size - ret, ", CPU%s: %s)", + top->evlist->cpus->nr > 1 ? "s" : "", top->cpu_list); + else { + if (top->target_tid != -1) + ret += snprintf(bf + ret, size - ret, ")"); + else + ret += snprintf(bf + ret, size - ret, ", %d CPU%s)", + top->evlist->cpus->nr, + top->evlist->cpus->nr > 1 ? "s" : ""); + } + + return ret; +} + +void perf_top__reset_sample_counters(struct perf_top *top) +{ + top->samples = top->us_samples = top->kernel_samples = + top->exact_samples = top->guest_kernel_samples = + top->guest_us_samples = 0; +} + +float perf_top__decay_samples(struct perf_top *top, struct rb_root *root) +{ + struct sym_entry *syme, *n; + float sum_ksamples = 0.0; + int snap = !top->display_weighted ? top->sym_counter : 0, j; + + /* Sort the active symbols */ + pthread_mutex_lock(&top->active_symbols_lock); + syme = list_entry(top->active_symbols.next, struct sym_entry, node); + pthread_mutex_unlock(&top->active_symbols_lock); + + top->rb_entries = 0; + list_for_each_entry_safe_from(syme, n, &top->active_symbols, node) { + syme->snap_count = syme->count[snap]; + if (syme->snap_count != 0) { + + if ((top->hide_user_symbols && + syme->origin == PERF_RECORD_MISC_USER) || + (top->hide_kernel_symbols && + syme->origin == PERF_RECORD_MISC_KERNEL)) { + perf_top__remove_active_sym(top, syme); + continue; + } + syme->weight = sym_weight(syme, top); + + if ((int)syme->snap_count >= top->count_filter) { + rb_insert_active_sym(root, syme); + ++top->rb_entries; + } + sum_ksamples += syme->snap_count; + + for (j = 0; j < top->evlist->nr_entries; j++) + syme->count[j] = top->zero ? 0 : syme->count[j] * 7 / 8; + } else + perf_top__remove_active_sym(top, syme); + } + + return sum_ksamples; +} + +/* + * Find the longest symbol name that will be displayed + */ +void perf_top__find_widths(struct perf_top *top, struct rb_root *root, + int *dso_width, int *dso_short_width, int *sym_width) +{ + struct rb_node *nd; + int printed = 0; + + *sym_width = *dso_width = *dso_short_width = 0; + + for (nd = rb_first(root); nd; nd = rb_next(nd)) { + struct sym_entry *syme = rb_entry(nd, struct sym_entry, rb_node); + struct symbol *sym = sym_entry__symbol(syme); + + if (++printed > top->print_entries || + (int)syme->snap_count < top->count_filter) + continue; + + if (syme->map->dso->long_name_len > *dso_width) + *dso_width = syme->map->dso->long_name_len; + + if (syme->map->dso->short_name_len > *dso_short_width) + *dso_short_width = syme->map->dso->short_name_len; + + if (sym->namelen > *sym_width) + *sym_width = sym->namelen; + } +} diff --git a/tools/perf/util/top.h b/tools/perf/util/top.h new file mode 100644 index 000000000000..4f769f47e19a --- /dev/null +++ b/tools/perf/util/top.h @@ -0,0 +1,64 @@ +#ifndef __PERF_TOP_H +#define __PERF_TOP_H 1 + +#include "types.h" +#include "../perf.h" +#include <stddef.h> +#include <pthread.h> +#include <linux/list.h> +#include <linux/rbtree.h> + +struct perf_evlist; +struct perf_evsel; + +struct sym_entry { + struct rb_node rb_node; + struct list_head node; + unsigned long snap_count; + double weight; + int skip; + u8 origin; + struct map *map; + unsigned long count[0]; +}; + +static inline struct symbol *sym_entry__symbol(struct sym_entry *self) +{ + return ((void *)self) + symbol_conf.priv_size; +} + +struct perf_top { + struct perf_evlist *evlist; + /* + * Symbols will be added here in perf_event__process_sample and will + * get out after decayed. + */ + struct list_head active_symbols; + pthread_mutex_t active_symbols_lock; + u64 samples; + u64 kernel_samples, us_samples; + u64 exact_samples; + u64 guest_us_samples, guest_kernel_samples; + int print_entries, count_filter, delay_secs; + int display_weighted, freq, rb_entries, sym_counter; + pid_t target_pid, target_tid; + bool hide_kernel_symbols, hide_user_symbols, zero; + const char *cpu_list; + struct perf_evsel *sym_evsel; +}; + +size_t perf_top__header_snprintf(struct perf_top *top, char *bf, size_t size); +void perf_top__reset_sample_counters(struct perf_top *top); +float perf_top__decay_samples(struct perf_top *top, struct rb_root *root); +void perf_top__find_widths(struct perf_top *top, struct rb_root *root, + int *dso_width, int *dso_short_width, int *sym_width); + +#ifdef NO_NEWT_SUPPORT +static inline int perf_top__tui_browser(struct perf_top *top __used) +{ + return 0; +} +#else +int perf_top__tui_browser(struct perf_top *top); +#endif +#endif /* __PERF_TOP_H */ diff --git a/tools/perf/util/trace-event-parse.c b/tools/perf/util/trace-event-parse.c index 73a02223c629..d8e622dd738a 100644 --- a/tools/perf/util/trace-event-parse.c +++ b/tools/perf/util/trace-event-parse.c @@ -153,7 +153,7 @@ void parse_proc_kallsyms(char *file, unsigned int size __unused) char *next = NULL; char *addr_str; char ch; - int ret; + int ret __used; int i; line = strtok_r(file, "\n", &next); diff --git a/tools/perf/util/ui/browsers/annotate.c b/tools/perf/util/ui/browsers/annotate.c index 82b78f99251b..1aa39658539c 100644 --- a/tools/perf/util/ui/browsers/annotate.c +++ b/tools/perf/util/ui/browsers/annotate.c @@ -1,9 +1,11 @@ #include "../browser.h" #include "../helpline.h" #include "../libslang.h" +#include "../../annotate.h" #include "../../hist.h" #include "../../sort.h" #include "../../symbol.h" +#include "../../annotate.h" static void ui__error_window(const char *fmt, ...) { @@ -58,32 +60,34 @@ static void annotate_browser__write(struct ui_browser *self, void *entry, int ro } static double objdump_line__calc_percent(struct objdump_line *self, - struct list_head *head, - struct symbol *sym) + struct symbol *sym, int evidx) { double percent = 0.0; if (self->offset != -1) { int len = sym->end - sym->start; unsigned int hits = 0; - struct sym_priv *priv = symbol__priv(sym); - struct sym_ext *sym_ext = priv->ext; - struct sym_hist *h = priv->hist; + struct annotation *notes = symbol__annotation(sym); + struct source_line *src_line = notes->src->lines; + struct sym_hist *h = annotation__histogram(notes, evidx); s64 offset = self->offset; - struct objdump_line *next = objdump__get_next_ip_line(head, self); - + struct objdump_line *next; + next = objdump__get_next_ip_line(¬es->src->source, self); while (offset < (s64)len && (next == NULL || offset < next->offset)) { - if (sym_ext) { - percent += sym_ext[offset].percent; + if (src_line) { + percent += src_line[offset].percent; } else - hits += h->ip[offset]; + hits += h->addr[offset]; ++offset; } - - if (sym_ext == NULL && h->sum) + /* + * If the percentage wasn't already calculated in + * symbol__get_source_line, do it now: + */ + if (src_line == NULL && h->sum) percent = 100.0 * hits / h->sum; } @@ -136,10 +140,10 @@ static void annotate_browser__set_top(struct annotate_browser *self, static int annotate_browser__run(struct annotate_browser *self) { struct rb_node *nd; - struct hist_entry *he = self->b.priv; + struct symbol *sym = self->b.priv; int key; - if (ui_browser__show(&self->b, he->ms.sym->name, + if (ui_browser__show(&self->b, sym->name, "<-, -> or ESC: exit, TAB/shift+TAB: cycle thru samples") < 0) return -1; /* @@ -179,42 +183,47 @@ out: return key; } -int hist_entry__tui_annotate(struct hist_entry *self) +int hist_entry__tui_annotate(struct hist_entry *he, int evidx) +{ + return symbol__tui_annotate(he->ms.sym, he->ms.map, evidx); +} + +int symbol__tui_annotate(struct symbol *sym, struct map *map, int evidx) { struct objdump_line *pos, *n; struct objdump_line_rb_node *rbpos; - LIST_HEAD(head); + struct annotation *notes = symbol__annotation(sym); struct annotate_browser browser = { .b = { - .entries = &head, + .entries = ¬es->src->source, .refresh = ui_browser__list_head_refresh, .seek = ui_browser__list_head_seek, .write = annotate_browser__write, - .priv = self, + .priv = sym, }, }; int ret; - if (self->ms.sym == NULL) + if (sym == NULL) return -1; - if (self->ms.map->dso->annotate_warned) + if (map->dso->annotate_warned) return -1; - if (hist_entry__annotate(self, &head, sizeof(*rbpos)) < 0) { + if (symbol__annotate(sym, map, sizeof(*rbpos)) < 0) { ui__error_window(ui_helpline__last_msg); return -1; } ui_helpline__push("Press <- or ESC to exit"); - list_for_each_entry(pos, &head, node) { + list_for_each_entry(pos, ¬es->src->source, node) { size_t line_len = strlen(pos->line); if (browser.b.width < line_len) browser.b.width = line_len; rbpos = objdump_line__rb(pos); rbpos->idx = browser.b.nr_entries++; - rbpos->percent = objdump_line__calc_percent(pos, &head, self->ms.sym); + rbpos->percent = objdump_line__calc_percent(pos, sym, evidx); if (rbpos->percent < 0.01) continue; objdump__insert_line(&browser.entries, rbpos); @@ -229,7 +238,7 @@ int hist_entry__tui_annotate(struct hist_entry *self) browser.b.width += 18; /* Percentage */ ret = annotate_browser__run(&browser); - list_for_each_entry_safe(pos, n, &head, node) { + 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 60c463c16028..294b49538522 100644 --- a/tools/perf/util/ui/browsers/hists.c +++ b/tools/perf/util/ui/browsers/hists.c @@ -377,7 +377,7 @@ static int hist_browser__show_callchain_node_rb_tree(struct hist_browser *self, while (node) { struct callchain_node *child = rb_entry(node, struct callchain_node, rb_node); struct rb_node *next = rb_next(node); - u64 cumul = cumul_hits(child); + u64 cumul = callchain_cumul_hits(child); struct callchain_list *chain; char folded_sign = ' '; int first = true; @@ -797,7 +797,8 @@ static int hists__browser_title(struct hists *self, char *bf, size_t size, return printed; } -int hists__browse(struct hists *self, const char *helpline, const char *ev_name) +int hists__browse(struct hists *self, const char *helpline, + const char *ev_name, int evidx) { struct hist_browser *browser = hist_browser__new(self); struct pstack *fstack; @@ -935,7 +936,7 @@ do_annotate: if (he == NULL) continue; - hist_entry__tui_annotate(he); + hist_entry__tui_annotate(he, evidx); } else if (choice == browse_map) map__browse(browser->selection->map); else if (choice == zoom_dso) { @@ -984,7 +985,7 @@ out: return key; } -int hists__tui_browse_tree(struct rb_root *self, const char *help) +int hists__tui_browse_tree(struct rb_root *self, const char *help, int evidx) { struct rb_node *first = rb_first(self), *nd = first, *next; int key = 0; @@ -993,7 +994,7 @@ int hists__tui_browse_tree(struct rb_root *self, const char *help) struct hists *hists = rb_entry(nd, struct hists, rb_node); const char *ev_name = __event_name(hists->type, hists->config); - key = hists__browse(hists, help, ev_name); + key = hists__browse(hists, help, ev_name, evidx); switch (key) { case NEWT_KEY_TAB: next = rb_next(nd); diff --git a/tools/perf/util/ui/browsers/map.c b/tools/perf/util/ui/browsers/map.c index e5158369106e..8462bffe20bc 100644 --- a/tools/perf/util/ui/browsers/map.c +++ b/tools/perf/util/ui/browsers/map.c @@ -41,7 +41,7 @@ static int ui_entry__read(const char *title, char *bf, size_t size, int width) out_free_form: newtPopWindow(); newtFormDestroy(form); - return 0; + return err; } struct map_browser { diff --git a/tools/perf/util/ui/browsers/top.c b/tools/perf/util/ui/browsers/top.c new file mode 100644 index 000000000000..ca6062483a8f --- /dev/null +++ b/tools/perf/util/ui/browsers/top.c @@ -0,0 +1,136 @@ +/* + * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com> + * + * Parts came from builtin-{top,stat,record}.c, see those files for further + * copyright notes. + * + * Released under the GPL v2. (and only v2, not any later version) + */ +#include "../browser.h" +#include "../helpline.h" +#include "../libslang.h" +#include "../../evlist.h" +#include "../../hist.h" +#include "../../sort.h" +#include "../../symbol.h" +#include "../../top.h" + +struct perf_top_browser { + struct ui_browser b; + struct rb_root root; + float sum_ksamples; + int dso_width; + int dso_short_width; + int sym_width; +}; + +static void perf_top_browser__write(struct ui_browser *browser, void *entry, int row) +{ + struct perf_top_browser *top_browser = container_of(browser, struct perf_top_browser, b); + struct sym_entry *syme = rb_entry(entry, struct sym_entry, rb_node); + bool current_entry = ui_browser__is_current_entry(browser, row); + struct symbol *symbol = sym_entry__symbol(syme); + struct perf_top *top = browser->priv; + int width = browser->width; + double pcnt; + + pcnt = 100.0 - (100.0 * ((top_browser->sum_ksamples - syme->snap_count) / + top_browser->sum_ksamples)); + ui_browser__set_percent_color(browser, pcnt, current_entry); + + if (top->evlist->nr_entries == 1 || !top->display_weighted) { + slsmg_printf("%20.2f ", syme->weight); + width -= 24; + } else { + slsmg_printf("%9.1f %10ld ", syme->weight, syme->snap_count); + width -= 23; + } + + slsmg_printf("%4.1f%%", pcnt); + width -= 7; + + if (verbose) { + slsmg_printf(" %016" PRIx64, symbol->start); + width -= 17; + } + + slsmg_printf(" %-*.*s ", top_browser->sym_width, top_browser->sym_width, + symbol->name); + width -= top_browser->sym_width; + slsmg_write_nstring(width >= syme->map->dso->long_name_len ? + syme->map->dso->long_name : + syme->map->dso->short_name, width); +} + +static void perf_top_browser__update_rb_tree(struct perf_top_browser *browser) +{ + struct perf_top *top = browser->b.priv; + + browser->root = RB_ROOT; + browser->b.top = NULL; + browser->sum_ksamples = perf_top__decay_samples(top, &browser->root); + perf_top__find_widths(top, &browser->root, &browser->dso_width, + &browser->dso_short_width, + &browser->sym_width); + if (browser->sym_width + browser->dso_width > browser->b.width - 29) { + browser->dso_width = browser->dso_short_width; + if (browser->sym_width + browser->dso_width > browser->b.width - 29) + browser->sym_width = browser->b.width - browser->dso_width - 29; + } + browser->b.nr_entries = top->rb_entries; +} + +static int perf_top_browser__run(struct perf_top_browser *browser) +{ + int key; + char title[160]; + struct perf_top *top = browser->b.priv; + int delay_msecs = top->delay_secs * 1000; + + perf_top_browser__update_rb_tree(browser); + perf_top__header_snprintf(top, title, sizeof(title)); + perf_top__reset_sample_counters(top); + + if (ui_browser__show(&browser->b, title, "ESC: exit") < 0) + return -1; + + newtFormSetTimer(browser->b.form, delay_msecs); + + while (1) { + key = ui_browser__run(&browser->b); + + switch (key) { + case -1: + /* FIXME we need to check if it was es.reason == NEWT_EXIT_TIMER */ + perf_top_browser__update_rb_tree(browser); + perf_top__header_snprintf(top, title, sizeof(title)); + perf_top__reset_sample_counters(top); + ui_browser__set_color(&browser->b, NEWT_COLORSET_ROOT); + SLsmg_gotorc(0, 0); + slsmg_write_nstring(title, browser->b.width); + break; + case NEWT_KEY_TAB: + default: + goto out; + } + } +out: + ui_browser__hide(&browser->b); + return key; +} + +int perf_top__tui_browser(struct perf_top *top) +{ + struct perf_top_browser browser = { + .b = { + .entries = &browser.root, + .refresh = ui_browser__rb_tree_refresh, + .seek = ui_browser__rb_tree_seek, + .write = perf_top_browser__write, + .priv = top, + }, + }; + + ui_helpline__push("Press <- or ESC to exit"); + return perf_top_browser__run(&browser); +} diff --git a/tools/perf/util/ui/libslang.h b/tools/perf/util/ui/libslang.h index 5623da8e8080..2b63e1c9b181 100644 --- a/tools/perf/util/ui/libslang.h +++ b/tools/perf/util/ui/libslang.h @@ -13,11 +13,11 @@ #if SLANG_VERSION < 20104 #define slsmg_printf(msg, args...) \ - SLsmg_printf((char *)msg, ##args) + SLsmg_printf((char *)(msg), ##args) #define slsmg_write_nstring(msg, len) \ - SLsmg_write_nstring((char *)msg, len) + SLsmg_write_nstring((char *)(msg), len) #define sltt_set_color(obj, name, fg, bg) \ - SLtt_set_color(obj,(char *)name, (char *)fg, (char *)bg) + SLtt_set_color(obj,(char *)(name), (char *)(fg), (char *)(bg)) #else #define slsmg_printf SLsmg_printf #define slsmg_write_nstring SLsmg_write_nstring diff --git a/tools/perf/util/ui/setup.c b/tools/perf/util/ui/setup.c index 662085032eb7..fbf1a145492f 100644 --- a/tools/perf/util/ui/setup.c +++ b/tools/perf/util/ui/setup.c @@ -14,11 +14,12 @@ static void newt_suspend(void *d __used) newtResume(); } -void setup_browser(void) +void setup_browser(bool fallback_to_pager) { if (!isatty(1) || !use_browser || dump_trace) { use_browser = 0; - setup_pager(); + if (fallback_to_pager) + setup_pager(); return; } |