perf tools: Reorganize event processing routines, lotsa dups killed
[safe/jmp/linux-2.6] / tools / perf / builtin-annotate.c
1 /*
2  * builtin-annotate.c
3  *
4  * Builtin annotate command: Analyze the perf.data input file,
5  * look up and read DSOs and symbol information and display
6  * a histogram of results, along various sorting keys.
7  */
8 #include "builtin.h"
9
10 #include "util/util.h"
11
12 #include "util/color.h"
13 #include <linux/list.h>
14 #include "util/cache.h"
15 #include <linux/rbtree.h>
16 #include "util/symbol.h"
17 #include "util/string.h"
18
19 #include "perf.h"
20 #include "util/debug.h"
21
22 #include "util/event.h"
23 #include "util/parse-options.h"
24 #include "util/parse-events.h"
25 #include "util/thread.h"
26 #include "util/sort.h"
27 #include "util/hist.h"
28
29 static char             const *input_name = "perf.data";
30
31 static int              force;
32 static int              input;
33
34 static int              full_paths;
35
36 static int              print_line;
37
38 static unsigned long    page_size;
39 static unsigned long    mmap_window = 32;
40
41 struct sym_hist {
42         u64             sum;
43         u64             ip[0];
44 };
45
46 struct sym_ext {
47         struct rb_node  node;
48         double          percent;
49         char            *path;
50 };
51
52 struct sym_priv {
53         struct sym_hist *hist;
54         struct sym_ext  *ext;
55 };
56
57 static struct symbol_conf symbol_conf = {
58         .priv_size        = sizeof(struct sym_priv),
59         .try_vmlinux_path = true,
60 };
61
62 static const char *sym_hist_filter;
63
64 static int symbol_filter(struct map *map __used, struct symbol *sym)
65 {
66         if (sym_hist_filter == NULL ||
67             strcmp(sym->name, sym_hist_filter) == 0) {
68                 struct sym_priv *priv = symbol__priv(sym);
69                 const int size = (sizeof(*priv->hist) +
70                                   (sym->end - sym->start) * sizeof(u64));
71
72                 priv->hist = malloc(size);
73                 if (priv->hist)
74                         memset(priv->hist, 0, size);
75                 return 0;
76         }
77         /*
78          * FIXME: We should really filter it out, as we don't want to go thru symbols
79          * we're not interested, and if a DSO ends up with no symbols, delete it too,
80          * but right now the kernel loading routines in symbol.c bail out if no symbols
81          * are found, fix it later.
82          */
83         return 0;
84 }
85
86 /*
87  * collect histogram counts
88  */
89 static void hist_hit(struct hist_entry *he, u64 ip)
90 {
91         unsigned int sym_size, offset;
92         struct symbol *sym = he->sym;
93         struct sym_priv *priv;
94         struct sym_hist *h;
95
96         he->count++;
97
98         if (!sym || !he->map)
99                 return;
100
101         priv = symbol__priv(sym);
102         if (!priv->hist)
103                 return;
104
105         sym_size = sym->end - sym->start;
106         offset = ip - sym->start;
107
108         if (verbose)
109                 fprintf(stderr, "%s: ip=%Lx\n", __func__,
110                         he->map->unmap_ip(he->map, ip));
111
112         if (offset >= sym_size)
113                 return;
114
115         h = priv->hist;
116         h->sum++;
117         h->ip[offset]++;
118
119         if (verbose >= 3)
120                 printf("%p %s: count++ [ip: %p, %08Lx] => %Ld\n",
121                         (void *)(unsigned long)he->sym->start,
122                         he->sym->name,
123                         (void *)(unsigned long)ip, ip - he->sym->start,
124                         h->ip[offset]);
125 }
126
127 static int hist_entry__add(struct thread *thread, struct map *map,
128                            struct symbol *sym, u64 ip, u64 count, char level)
129 {
130         bool hit;
131         struct hist_entry *he = __hist_entry__add(thread, map, sym, NULL, ip,
132                                                   count, level, &hit);
133         if (he == NULL)
134                 return -ENOMEM;
135         hist_hit(he, ip);
136         return 0;
137 }
138
139 static int process_sample_event(event_t *event)
140 {
141         char level;
142         u64 ip = event->ip.ip;
143         struct map *map = NULL;
144         struct symbol *sym = NULL;
145         struct thread *thread = threads__findnew(event->ip.pid);
146
147         dump_printf("(IP, %d): %d: %p\n", event->header.misc,
148                     event->ip.pid, (void *)(long)ip);
149
150         if (thread == NULL) {
151                 fprintf(stderr, "problem processing %d event, skipping it.\n",
152                         event->header.type);
153                 return -1;
154         }
155
156         dump_printf(" ... thread: %s:%d\n", thread->comm, thread->pid);
157
158         if (event->header.misc & PERF_RECORD_MISC_KERNEL) {
159                 level = 'k';
160                 sym = kernel_maps__find_function(ip, &map, symbol_filter);
161                 dump_printf(" ...... dso: %s\n",
162                             map ? map->dso->long_name : "<not found>");
163         } else if (event->header.misc & PERF_RECORD_MISC_USER) {
164                 level = '.';
165                 map = thread__find_map(thread, MAP__FUNCTION, ip);
166                 if (map != NULL) {
167                         ip = map->map_ip(map, ip);
168                         sym = map__find_symbol(map, ip, symbol_filter);
169                 } else {
170                         /*
171                          * If this is outside of all known maps,
172                          * and is a negative address, try to look it
173                          * up in the kernel dso, as it might be a
174                          * vsyscall or vdso (which executes in user-mode).
175                          *
176                          * XXX This is nasty, we should have a symbol list in
177                          * the "[vdso]" dso, but for now lets use the old
178                          * trick of looking in the whole kernel symbol list.
179                          */
180                         if ((long long)ip < 0)
181                                 sym = kernel_maps__find_function(ip, &map,
182                                                                  symbol_filter);
183                 }
184                 dump_printf(" ...... dso: %s\n",
185                             map ? map->dso->long_name : "<not found>");
186         } else {
187                 level = 'H';
188                 dump_printf(" ...... dso: [hypervisor]\n");
189         }
190
191         if (hist_entry__add(thread, map, sym, ip, 1, level)) {
192                 fprintf(stderr, "problem incrementing symbol count, "
193                                 "skipping event\n");
194                 return -1;
195         }
196
197         return 0;
198 }
199
200 static int event__process(event_t *self)
201 {
202         switch (self->header.type) {
203         case PERF_RECORD_SAMPLE:
204                 return process_sample_event(self);
205
206         case PERF_RECORD_MMAP:
207                 return event__process_mmap(self);
208
209         case PERF_RECORD_COMM:
210                 return event__process_comm(self);
211
212         case PERF_RECORD_FORK:
213                 return event__process_task(self);
214         /*
215          * We dont process them right now but they are fine:
216          */
217
218         case PERF_RECORD_THROTTLE:
219         case PERF_RECORD_UNTHROTTLE:
220                 return 0;
221
222         default:
223                 return -1;
224         }
225
226         return 0;
227 }
228
229 static int parse_line(FILE *file, struct hist_entry *he, u64 len)
230 {
231         struct symbol *sym = he->sym;
232         char *line = NULL, *tmp, *tmp2;
233         static const char *prev_line;
234         static const char *prev_color;
235         unsigned int offset;
236         size_t line_len;
237         u64 start;
238         s64 line_ip;
239         int ret;
240         char *c;
241
242         if (getline(&line, &line_len, file) < 0)
243                 return -1;
244         if (!line)
245                 return -1;
246
247         c = strchr(line, '\n');
248         if (c)
249                 *c = 0;
250
251         line_ip = -1;
252         offset = 0;
253         ret = -2;
254
255         /*
256          * Strip leading spaces:
257          */
258         tmp = line;
259         while (*tmp) {
260                 if (*tmp != ' ')
261                         break;
262                 tmp++;
263         }
264
265         if (*tmp) {
266                 /*
267                  * Parse hexa addresses followed by ':'
268                  */
269                 line_ip = strtoull(tmp, &tmp2, 16);
270                 if (*tmp2 != ':')
271                         line_ip = -1;
272         }
273
274         start = he->map->unmap_ip(he->map, sym->start);
275
276         if (line_ip != -1) {
277                 const char *path = NULL;
278                 unsigned int hits = 0;
279                 double percent = 0.0;
280                 const char *color;
281                 struct sym_priv *priv = symbol__priv(sym);
282                 struct sym_ext *sym_ext = priv->ext;
283                 struct sym_hist *h = priv->hist;
284
285                 offset = line_ip - start;
286                 if (offset < len)
287                         hits = h->ip[offset];
288
289                 if (offset < len && sym_ext) {
290                         path = sym_ext[offset].path;
291                         percent = sym_ext[offset].percent;
292                 } else if (h->sum)
293                         percent = 100.0 * hits / h->sum;
294
295                 color = get_percent_color(percent);
296
297                 /*
298                  * Also color the filename and line if needed, with
299                  * the same color than the percentage. Don't print it
300                  * twice for close colored ip with the same filename:line
301                  */
302                 if (path) {
303                         if (!prev_line || strcmp(prev_line, path)
304                                        || color != prev_color) {
305                                 color_fprintf(stdout, color, " %s", path);
306                                 prev_line = path;
307                                 prev_color = color;
308                         }
309                 }
310
311                 color_fprintf(stdout, color, " %7.2f", percent);
312                 printf(" :      ");
313                 color_fprintf(stdout, PERF_COLOR_BLUE, "%s\n", line);
314         } else {
315                 if (!*line)
316                         printf("         :\n");
317                 else
318                         printf("         :      %s\n", line);
319         }
320
321         return 0;
322 }
323
324 static struct rb_root root_sym_ext;
325
326 static void insert_source_line(struct sym_ext *sym_ext)
327 {
328         struct sym_ext *iter;
329         struct rb_node **p = &root_sym_ext.rb_node;
330         struct rb_node *parent = NULL;
331
332         while (*p != NULL) {
333                 parent = *p;
334                 iter = rb_entry(parent, struct sym_ext, node);
335
336                 if (sym_ext->percent > iter->percent)
337                         p = &(*p)->rb_left;
338                 else
339                         p = &(*p)->rb_right;
340         }
341
342         rb_link_node(&sym_ext->node, parent, p);
343         rb_insert_color(&sym_ext->node, &root_sym_ext);
344 }
345
346 static void free_source_line(struct hist_entry *he, int len)
347 {
348         struct sym_priv *priv = symbol__priv(he->sym);
349         struct sym_ext *sym_ext = priv->ext;
350         int i;
351
352         if (!sym_ext)
353                 return;
354
355         for (i = 0; i < len; i++)
356                 free(sym_ext[i].path);
357         free(sym_ext);
358
359         priv->ext = NULL;
360         root_sym_ext = RB_ROOT;
361 }
362
363 /* Get the filename:line for the colored entries */
364 static void
365 get_source_line(struct hist_entry *he, int len, const char *filename)
366 {
367         struct symbol *sym = he->sym;
368         u64 start;
369         int i;
370         char cmd[PATH_MAX * 2];
371         struct sym_ext *sym_ext;
372         struct sym_priv *priv = symbol__priv(sym);
373         struct sym_hist *h = priv->hist;
374
375         if (!h->sum)
376                 return;
377
378         sym_ext = priv->ext = calloc(len, sizeof(struct sym_ext));
379         if (!priv->ext)
380                 return;
381
382         start = he->map->unmap_ip(he->map, sym->start);
383
384         for (i = 0; i < len; i++) {
385                 char *path = NULL;
386                 size_t line_len;
387                 u64 offset;
388                 FILE *fp;
389
390                 sym_ext[i].percent = 100.0 * h->ip[i] / h->sum;
391                 if (sym_ext[i].percent <= 0.5)
392                         continue;
393
394                 offset = start + i;
395                 sprintf(cmd, "addr2line -e %s %016llx", filename, offset);
396                 fp = popen(cmd, "r");
397                 if (!fp)
398                         continue;
399
400                 if (getline(&path, &line_len, fp) < 0 || !line_len)
401                         goto next;
402
403                 sym_ext[i].path = malloc(sizeof(char) * line_len + 1);
404                 if (!sym_ext[i].path)
405                         goto next;
406
407                 strcpy(sym_ext[i].path, path);
408                 insert_source_line(&sym_ext[i]);
409
410         next:
411                 pclose(fp);
412         }
413 }
414
415 static void print_summary(const char *filename)
416 {
417         struct sym_ext *sym_ext;
418         struct rb_node *node;
419
420         printf("\nSorted summary for file %s\n", filename);
421         printf("----------------------------------------------\n\n");
422
423         if (RB_EMPTY_ROOT(&root_sym_ext)) {
424                 printf(" Nothing higher than %1.1f%%\n", MIN_GREEN);
425                 return;
426         }
427
428         node = rb_first(&root_sym_ext);
429         while (node) {
430                 double percent;
431                 const char *color;
432                 char *path;
433
434                 sym_ext = rb_entry(node, struct sym_ext, node);
435                 percent = sym_ext->percent;
436                 color = get_percent_color(percent);
437                 path = sym_ext->path;
438
439                 color_fprintf(stdout, color, " %7.2f %s", percent, path);
440                 node = rb_next(node);
441         }
442 }
443
444 static void annotate_sym(struct hist_entry *he)
445 {
446         struct map *map = he->map;
447         struct dso *dso = map->dso;
448         struct symbol *sym = he->sym;
449         const char *filename = dso->long_name, *d_filename;
450         u64 len;
451         char command[PATH_MAX*2];
452         FILE *file;
453
454         if (!filename)
455                 return;
456
457         if (verbose)
458                 fprintf(stderr, "%s: filename=%s, sym=%s, start=%Lx, end=%Lx\n",
459                         __func__, filename, sym->name,
460                         map->unmap_ip(map, sym->start),
461                         map->unmap_ip(map, sym->end));
462
463         if (full_paths)
464                 d_filename = filename;
465         else
466                 d_filename = basename(filename);
467
468         len = sym->end - sym->start;
469
470         if (print_line) {
471                 get_source_line(he, len, filename);
472                 print_summary(filename);
473         }
474
475         printf("\n\n------------------------------------------------\n");
476         printf(" Percent |      Source code & Disassembly of %s\n", d_filename);
477         printf("------------------------------------------------\n");
478
479         if (verbose >= 2)
480                 printf("annotating [%p] %30s : [%p] %30s\n",
481                        dso, dso->long_name, sym, sym->name);
482
483         sprintf(command, "objdump --start-address=0x%016Lx --stop-address=0x%016Lx -dS %s|grep -v %s",
484                 map->unmap_ip(map, sym->start), map->unmap_ip(map, sym->end),
485                 filename, filename);
486
487         if (verbose >= 3)
488                 printf("doing: %s\n", command);
489
490         file = popen(command, "r");
491         if (!file)
492                 return;
493
494         while (!feof(file)) {
495                 if (parse_line(file, he, len) < 0)
496                         break;
497         }
498
499         pclose(file);
500         if (print_line)
501                 free_source_line(he, len);
502 }
503
504 static void find_annotations(void)
505 {
506         struct rb_node *nd;
507
508         for (nd = rb_first(&output_hists); nd; nd = rb_next(nd)) {
509                 struct hist_entry *he = rb_entry(nd, struct hist_entry, rb_node);
510                 struct sym_priv *priv;
511
512                 if (he->sym == NULL)
513                         continue;
514
515                 priv = symbol__priv(he->sym);
516                 if (priv->hist == NULL)
517                         continue;
518
519                 annotate_sym(he);
520                 /*
521                  * Since we have a hist_entry per IP for the same symbol, free
522                  * he->sym->hist to signal we already processed this symbol.
523                  */
524                 free(priv->hist);
525                 priv->hist = NULL;
526         }
527 }
528
529 static int __cmd_annotate(void)
530 {
531         int ret, rc = EXIT_FAILURE;
532         unsigned long offset = 0;
533         unsigned long head = 0;
534         struct stat input_stat;
535         event_t *event;
536         uint32_t size;
537         char *buf;
538
539         register_idle_thread();
540
541         input = open(input_name, O_RDONLY);
542         if (input < 0) {
543                 perror("failed to open file");
544                 exit(-1);
545         }
546
547         ret = fstat(input, &input_stat);
548         if (ret < 0) {
549                 perror("failed to stat file");
550                 exit(-1);
551         }
552
553         if (!force && input_stat.st_uid && (input_stat.st_uid != geteuid())) {
554                 fprintf(stderr, "file: %s not owned by current user or root\n", input_name);
555                 exit(-1);
556         }
557
558         if (!input_stat.st_size) {
559                 fprintf(stderr, "zero-sized file, nothing to do!\n");
560                 exit(0);
561         }
562
563 remap:
564         buf = (char *)mmap(NULL, page_size * mmap_window, PROT_READ,
565                            MAP_SHARED, input, offset);
566         if (buf == MAP_FAILED) {
567                 perror("failed to mmap file");
568                 exit(-1);
569         }
570
571 more:
572         event = (event_t *)(buf + head);
573
574         size = event->header.size;
575         if (!size)
576                 size = 8;
577
578         if (head + event->header.size >= page_size * mmap_window) {
579                 unsigned long shift = page_size * (head / page_size);
580                 int munmap_ret;
581
582                 munmap_ret = munmap(buf, page_size * mmap_window);
583                 assert(munmap_ret == 0);
584
585                 offset += shift;
586                 head -= shift;
587                 goto remap;
588         }
589
590         size = event->header.size;
591
592         dump_printf("%p [%p]: event: %d\n",
593                         (void *)(offset + head),
594                         (void *)(long)event->header.size,
595                         event->header.type);
596
597         if (!size || event__process(event) < 0) {
598
599                 dump_printf("%p [%p]: skipping unknown header type: %d\n",
600                         (void *)(offset + head),
601                         (void *)(long)(event->header.size),
602                         event->header.type);
603                 /*
604                  * assume we lost track of the stream, check alignment, and
605                  * increment a single u64 in the hope to catch on again 'soon'.
606                  */
607
608                 if (unlikely(head & 7))
609                         head &= ~7ULL;
610
611                 size = 8;
612         }
613
614         head += size;
615
616         if (offset + head < (unsigned long)input_stat.st_size)
617                 goto more;
618
619         rc = EXIT_SUCCESS;
620         close(input);
621
622
623         if (dump_trace) {
624                 event__print_totals();
625                 return 0;
626         }
627
628         if (verbose > 3)
629                 threads__fprintf(stdout);
630
631         if (verbose > 2)
632                 dsos__fprintf(stdout);
633
634         collapse__resort();
635         output__resort(event__total[0]);
636
637         find_annotations();
638
639         return rc;
640 }
641
642 static const char * const annotate_usage[] = {
643         "perf annotate [<options>] <command>",
644         NULL
645 };
646
647 static const struct option options[] = {
648         OPT_STRING('i', "input", &input_name, "file",
649                     "input file name"),
650         OPT_STRING('s', "symbol", &sym_hist_filter, "symbol",
651                     "symbol to annotate"),
652         OPT_BOOLEAN('f', "force", &force, "don't complain, do it"),
653         OPT_BOOLEAN('v', "verbose", &verbose,
654                     "be more verbose (show symbol address, etc)"),
655         OPT_BOOLEAN('D', "dump-raw-trace", &dump_trace,
656                     "dump raw trace in ASCII"),
657         OPT_STRING('k', "vmlinux", &symbol_conf.vmlinux_name,
658                    "file", "vmlinux pathname"),
659         OPT_BOOLEAN('m', "modules", &symbol_conf.use_modules,
660                     "load module symbols - WARNING: use only with -k and LIVE kernel"),
661         OPT_BOOLEAN('l', "print-line", &print_line,
662                     "print matching source lines (may be slow)"),
663         OPT_BOOLEAN('P', "full-paths", &full_paths,
664                     "Don't shorten the displayed pathnames"),
665         OPT_END()
666 };
667
668 static void setup_sorting(void)
669 {
670         char *tmp, *tok, *str = strdup(sort_order);
671
672         for (tok = strtok_r(str, ", ", &tmp);
673                         tok; tok = strtok_r(NULL, ", ", &tmp)) {
674                 if (sort_dimension__add(tok) < 0) {
675                         error("Unknown --sort key: `%s'", tok);
676                         usage_with_options(annotate_usage, options);
677                 }
678         }
679
680         free(str);
681 }
682
683 int cmd_annotate(int argc, const char **argv, const char *prefix __used)
684 {
685         if (symbol__init(&symbol_conf) < 0)
686                 return -1;
687
688         page_size = getpagesize();
689
690         argc = parse_options(argc, argv, options, annotate_usage, 0);
691
692         setup_sorting();
693
694         if (argc) {
695                 /*
696                  * Special case: if there's an argument left then assume tha
697                  * it's a symbol filter:
698                  */
699                 if (argc > 1)
700                         usage_with_options(annotate_usage, options);
701
702                 sym_hist_filter = argv[0];
703         }
704
705         setup_pager();
706
707         if (field_sep && *field_sep == '.') {
708                 fputs("'.' is the only non valid --field-separator argument\n",
709                                 stderr);
710                 exit(129);
711         }
712
713         return __cmd_annotate();
714 }