Loading...
1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com>
4 *
5 * Parts came from builtin-annotate.c, see those files for further
6 * copyright notes.
7 */
8
9#include <errno.h>
10#include <inttypes.h>
11#include <libgen.h>
12#include <stdlib.h>
13#include "util.h" // hex_width()
14#include "ui/ui.h"
15#include "sort.h"
16#include "build-id.h"
17#include "color.h"
18#include "config.h"
19#include "dso.h"
20#include "env.h"
21#include "map.h"
22#include "maps.h"
23#include "symbol.h"
24#include "srcline.h"
25#include "units.h"
26#include "debug.h"
27#include "annotate.h"
28#include "annotate-data.h"
29#include "evsel.h"
30#include "evlist.h"
31#include "bpf-event.h"
32#include "bpf-utils.h"
33#include "block-range.h"
34#include "string2.h"
35#include "dwarf-regs.h"
36#include "util/event.h"
37#include "util/sharded_mutex.h"
38#include "arch/common.h"
39#include "namespaces.h"
40#include <regex.h>
41#include <linux/bitops.h>
42#include <linux/kernel.h>
43#include <linux/string.h>
44#include <linux/zalloc.h>
45#include <subcmd/parse-options.h>
46#include <subcmd/run-command.h>
47
48/* FIXME: For the HE_COLORSET */
49#include "ui/browser.h"
50
51/*
52 * FIXME: Using the same values as slang.h,
53 * but that header may not be available everywhere
54 */
55#define LARROW_CHAR ((unsigned char)',')
56#define RARROW_CHAR ((unsigned char)'+')
57#define DARROW_CHAR ((unsigned char)'.')
58#define UARROW_CHAR ((unsigned char)'-')
59
60#include <linux/ctype.h>
61
62/* global annotation options */
63struct annotation_options annotate_opts;
64
65static regex_t file_lineno;
66
67static struct ins_ops *ins__find(struct arch *arch, const char *name);
68static void ins__sort(struct arch *arch);
69static int disasm_line__parse(char *line, const char **namep, char **rawp);
70static int call__scnprintf(struct ins *ins, char *bf, size_t size,
71 struct ins_operands *ops, int max_ins_name);
72static int jump__scnprintf(struct ins *ins, char *bf, size_t size,
73 struct ins_operands *ops, int max_ins_name);
74
75struct arch {
76 const char *name;
77 struct ins *instructions;
78 size_t nr_instructions;
79 size_t nr_instructions_allocated;
80 struct ins_ops *(*associate_instruction_ops)(struct arch *arch, const char *name);
81 bool sorted_instructions;
82 bool initialized;
83 const char *insn_suffix;
84 void *priv;
85 unsigned int model;
86 unsigned int family;
87 int (*init)(struct arch *arch, char *cpuid);
88 bool (*ins_is_fused)(struct arch *arch, const char *ins1,
89 const char *ins2);
90 struct {
91 char comment_char;
92 char skip_functions_char;
93 char register_char;
94 char memory_ref_char;
95 } objdump;
96};
97
98static struct ins_ops call_ops;
99static struct ins_ops dec_ops;
100static struct ins_ops jump_ops;
101static struct ins_ops mov_ops;
102static struct ins_ops nop_ops;
103static struct ins_ops lock_ops;
104static struct ins_ops ret_ops;
105
106/* Data type collection debug statistics */
107struct annotated_data_stat ann_data_stat;
108LIST_HEAD(ann_insn_stat);
109
110static int arch__grow_instructions(struct arch *arch)
111{
112 struct ins *new_instructions;
113 size_t new_nr_allocated;
114
115 if (arch->nr_instructions_allocated == 0 && arch->instructions)
116 goto grow_from_non_allocated_table;
117
118 new_nr_allocated = arch->nr_instructions_allocated + 128;
119 new_instructions = realloc(arch->instructions, new_nr_allocated * sizeof(struct ins));
120 if (new_instructions == NULL)
121 return -1;
122
123out_update_instructions:
124 arch->instructions = new_instructions;
125 arch->nr_instructions_allocated = new_nr_allocated;
126 return 0;
127
128grow_from_non_allocated_table:
129 new_nr_allocated = arch->nr_instructions + 128;
130 new_instructions = calloc(new_nr_allocated, sizeof(struct ins));
131 if (new_instructions == NULL)
132 return -1;
133
134 memcpy(new_instructions, arch->instructions, arch->nr_instructions);
135 goto out_update_instructions;
136}
137
138static int arch__associate_ins_ops(struct arch* arch, const char *name, struct ins_ops *ops)
139{
140 struct ins *ins;
141
142 if (arch->nr_instructions == arch->nr_instructions_allocated &&
143 arch__grow_instructions(arch))
144 return -1;
145
146 ins = &arch->instructions[arch->nr_instructions];
147 ins->name = strdup(name);
148 if (!ins->name)
149 return -1;
150
151 ins->ops = ops;
152 arch->nr_instructions++;
153
154 ins__sort(arch);
155 return 0;
156}
157
158#include "arch/arc/annotate/instructions.c"
159#include "arch/arm/annotate/instructions.c"
160#include "arch/arm64/annotate/instructions.c"
161#include "arch/csky/annotate/instructions.c"
162#include "arch/loongarch/annotate/instructions.c"
163#include "arch/mips/annotate/instructions.c"
164#include "arch/x86/annotate/instructions.c"
165#include "arch/powerpc/annotate/instructions.c"
166#include "arch/riscv64/annotate/instructions.c"
167#include "arch/s390/annotate/instructions.c"
168#include "arch/sparc/annotate/instructions.c"
169
170static struct arch architectures[] = {
171 {
172 .name = "arc",
173 .init = arc__annotate_init,
174 },
175 {
176 .name = "arm",
177 .init = arm__annotate_init,
178 },
179 {
180 .name = "arm64",
181 .init = arm64__annotate_init,
182 },
183 {
184 .name = "csky",
185 .init = csky__annotate_init,
186 },
187 {
188 .name = "mips",
189 .init = mips__annotate_init,
190 .objdump = {
191 .comment_char = '#',
192 },
193 },
194 {
195 .name = "x86",
196 .init = x86__annotate_init,
197 .instructions = x86__instructions,
198 .nr_instructions = ARRAY_SIZE(x86__instructions),
199 .insn_suffix = "bwlq",
200 .objdump = {
201 .comment_char = '#',
202 .register_char = '%',
203 .memory_ref_char = '(',
204 },
205 },
206 {
207 .name = "powerpc",
208 .init = powerpc__annotate_init,
209 },
210 {
211 .name = "riscv64",
212 .init = riscv64__annotate_init,
213 },
214 {
215 .name = "s390",
216 .init = s390__annotate_init,
217 .objdump = {
218 .comment_char = '#',
219 },
220 },
221 {
222 .name = "sparc",
223 .init = sparc__annotate_init,
224 .objdump = {
225 .comment_char = '#',
226 },
227 },
228 {
229 .name = "loongarch",
230 .init = loongarch__annotate_init,
231 .objdump = {
232 .comment_char = '#',
233 },
234 },
235};
236
237static void ins__delete(struct ins_operands *ops)
238{
239 if (ops == NULL)
240 return;
241 zfree(&ops->source.raw);
242 zfree(&ops->source.name);
243 zfree(&ops->target.raw);
244 zfree(&ops->target.name);
245}
246
247static int ins__raw_scnprintf(struct ins *ins, char *bf, size_t size,
248 struct ins_operands *ops, int max_ins_name)
249{
250 return scnprintf(bf, size, "%-*s %s", max_ins_name, ins->name, ops->raw);
251}
252
253int ins__scnprintf(struct ins *ins, char *bf, size_t size,
254 struct ins_operands *ops, int max_ins_name)
255{
256 if (ins->ops->scnprintf)
257 return ins->ops->scnprintf(ins, bf, size, ops, max_ins_name);
258
259 return ins__raw_scnprintf(ins, bf, size, ops, max_ins_name);
260}
261
262bool ins__is_fused(struct arch *arch, const char *ins1, const char *ins2)
263{
264 if (!arch || !arch->ins_is_fused)
265 return false;
266
267 return arch->ins_is_fused(arch, ins1, ins2);
268}
269
270static int call__parse(struct arch *arch, struct ins_operands *ops, struct map_symbol *ms)
271{
272 char *endptr, *tok, *name;
273 struct map *map = ms->map;
274 struct addr_map_symbol target = {
275 .ms = { .map = map, },
276 };
277
278 ops->target.addr = strtoull(ops->raw, &endptr, 16);
279
280 name = strchr(endptr, '<');
281 if (name == NULL)
282 goto indirect_call;
283
284 name++;
285
286 if (arch->objdump.skip_functions_char &&
287 strchr(name, arch->objdump.skip_functions_char))
288 return -1;
289
290 tok = strchr(name, '>');
291 if (tok == NULL)
292 return -1;
293
294 *tok = '\0';
295 ops->target.name = strdup(name);
296 *tok = '>';
297
298 if (ops->target.name == NULL)
299 return -1;
300find_target:
301 target.addr = map__objdump_2mem(map, ops->target.addr);
302
303 if (maps__find_ams(ms->maps, &target) == 0 &&
304 map__rip_2objdump(target.ms.map, map__map_ip(target.ms.map, target.addr)) == ops->target.addr)
305 ops->target.sym = target.ms.sym;
306
307 return 0;
308
309indirect_call:
310 tok = strchr(endptr, '*');
311 if (tok != NULL) {
312 endptr++;
313
314 /* Indirect call can use a non-rip register and offset: callq *0x8(%rbx).
315 * Do not parse such instruction. */
316 if (strstr(endptr, "(%r") == NULL)
317 ops->target.addr = strtoull(endptr, NULL, 16);
318 }
319 goto find_target;
320}
321
322static int call__scnprintf(struct ins *ins, char *bf, size_t size,
323 struct ins_operands *ops, int max_ins_name)
324{
325 if (ops->target.sym)
326 return scnprintf(bf, size, "%-*s %s", max_ins_name, ins->name, ops->target.sym->name);
327
328 if (ops->target.addr == 0)
329 return ins__raw_scnprintf(ins, bf, size, ops, max_ins_name);
330
331 if (ops->target.name)
332 return scnprintf(bf, size, "%-*s %s", max_ins_name, ins->name, ops->target.name);
333
334 return scnprintf(bf, size, "%-*s *%" PRIx64, max_ins_name, ins->name, ops->target.addr);
335}
336
337static struct ins_ops call_ops = {
338 .parse = call__parse,
339 .scnprintf = call__scnprintf,
340};
341
342bool ins__is_call(const struct ins *ins)
343{
344 return ins->ops == &call_ops || ins->ops == &s390_call_ops || ins->ops == &loongarch_call_ops;
345}
346
347/*
348 * Prevents from matching commas in the comment section, e.g.:
349 * ffff200008446e70: b.cs ffff2000084470f4 <generic_exec_single+0x314> // b.hs, b.nlast
350 *
351 * and skip comma as part of function arguments, e.g.:
352 * 1d8b4ac <linemap_lookup(line_maps const*, unsigned int)+0xcc>
353 */
354static inline const char *validate_comma(const char *c, struct ins_operands *ops)
355{
356 if (ops->jump.raw_comment && c > ops->jump.raw_comment)
357 return NULL;
358
359 if (ops->jump.raw_func_start && c > ops->jump.raw_func_start)
360 return NULL;
361
362 return c;
363}
364
365static int jump__parse(struct arch *arch, struct ins_operands *ops, struct map_symbol *ms)
366{
367 struct map *map = ms->map;
368 struct symbol *sym = ms->sym;
369 struct addr_map_symbol target = {
370 .ms = { .map = map, },
371 };
372 const char *c = strchr(ops->raw, ',');
373 u64 start, end;
374
375 ops->jump.raw_comment = strchr(ops->raw, arch->objdump.comment_char);
376 ops->jump.raw_func_start = strchr(ops->raw, '<');
377
378 c = validate_comma(c, ops);
379
380 /*
381 * Examples of lines to parse for the _cpp_lex_token@@Base
382 * function:
383 *
384 * 1159e6c: jne 115aa32 <_cpp_lex_token@@Base+0xf92>
385 * 1159e8b: jne c469be <cpp_named_operator2name@@Base+0xa72>
386 *
387 * The first is a jump to an offset inside the same function,
388 * the second is to another function, i.e. that 0xa72 is an
389 * offset in the cpp_named_operator2name@@base function.
390 */
391 /*
392 * skip over possible up to 2 operands to get to address, e.g.:
393 * tbnz w0, #26, ffff0000083cd190 <security_file_permission+0xd0>
394 */
395 if (c++ != NULL) {
396 ops->target.addr = strtoull(c, NULL, 16);
397 if (!ops->target.addr) {
398 c = strchr(c, ',');
399 c = validate_comma(c, ops);
400 if (c++ != NULL)
401 ops->target.addr = strtoull(c, NULL, 16);
402 }
403 } else {
404 ops->target.addr = strtoull(ops->raw, NULL, 16);
405 }
406
407 target.addr = map__objdump_2mem(map, ops->target.addr);
408 start = map__unmap_ip(map, sym->start);
409 end = map__unmap_ip(map, sym->end);
410
411 ops->target.outside = target.addr < start || target.addr > end;
412
413 /*
414 * FIXME: things like this in _cpp_lex_token (gcc's cc1 program):
415
416 cpp_named_operator2name@@Base+0xa72
417
418 * Point to a place that is after the cpp_named_operator2name
419 * boundaries, i.e. in the ELF symbol table for cc1
420 * cpp_named_operator2name is marked as being 32-bytes long, but it in
421 * fact is much larger than that, so we seem to need a symbols__find()
422 * routine that looks for >= current->start and < next_symbol->start,
423 * possibly just for C++ objects?
424 *
425 * For now lets just make some progress by marking jumps to outside the
426 * current function as call like.
427 *
428 * Actual navigation will come next, with further understanding of how
429 * the symbol searching and disassembly should be done.
430 */
431 if (maps__find_ams(ms->maps, &target) == 0 &&
432 map__rip_2objdump(target.ms.map, map__map_ip(target.ms.map, target.addr)) == ops->target.addr)
433 ops->target.sym = target.ms.sym;
434
435 if (!ops->target.outside) {
436 ops->target.offset = target.addr - start;
437 ops->target.offset_avail = true;
438 } else {
439 ops->target.offset_avail = false;
440 }
441
442 return 0;
443}
444
445static int jump__scnprintf(struct ins *ins, char *bf, size_t size,
446 struct ins_operands *ops, int max_ins_name)
447{
448 const char *c;
449
450 if (!ops->target.addr || ops->target.offset < 0)
451 return ins__raw_scnprintf(ins, bf, size, ops, max_ins_name);
452
453 if (ops->target.outside && ops->target.sym != NULL)
454 return scnprintf(bf, size, "%-*s %s", max_ins_name, ins->name, ops->target.sym->name);
455
456 c = strchr(ops->raw, ',');
457 c = validate_comma(c, ops);
458
459 if (c != NULL) {
460 const char *c2 = strchr(c + 1, ',');
461
462 c2 = validate_comma(c2, ops);
463 /* check for 3-op insn */
464 if (c2 != NULL)
465 c = c2;
466 c++;
467
468 /* mirror arch objdump's space-after-comma style */
469 if (*c == ' ')
470 c++;
471 }
472
473 return scnprintf(bf, size, "%-*s %.*s%" PRIx64, max_ins_name,
474 ins->name, c ? c - ops->raw : 0, ops->raw,
475 ops->target.offset);
476}
477
478static void jump__delete(struct ins_operands *ops __maybe_unused)
479{
480 /*
481 * The ops->jump.raw_comment and ops->jump.raw_func_start belong to the
482 * raw string, don't free them.
483 */
484}
485
486static struct ins_ops jump_ops = {
487 .free = jump__delete,
488 .parse = jump__parse,
489 .scnprintf = jump__scnprintf,
490};
491
492bool ins__is_jump(const struct ins *ins)
493{
494 return ins->ops == &jump_ops || ins->ops == &loongarch_jump_ops;
495}
496
497static int comment__symbol(char *raw, char *comment, u64 *addrp, char **namep)
498{
499 char *endptr, *name, *t;
500
501 if (strstr(raw, "(%rip)") == NULL)
502 return 0;
503
504 *addrp = strtoull(comment, &endptr, 16);
505 if (endptr == comment)
506 return 0;
507 name = strchr(endptr, '<');
508 if (name == NULL)
509 return -1;
510
511 name++;
512
513 t = strchr(name, '>');
514 if (t == NULL)
515 return 0;
516
517 *t = '\0';
518 *namep = strdup(name);
519 *t = '>';
520
521 return 0;
522}
523
524static int lock__parse(struct arch *arch, struct ins_operands *ops, struct map_symbol *ms)
525{
526 ops->locked.ops = zalloc(sizeof(*ops->locked.ops));
527 if (ops->locked.ops == NULL)
528 return 0;
529
530 if (disasm_line__parse(ops->raw, &ops->locked.ins.name, &ops->locked.ops->raw) < 0)
531 goto out_free_ops;
532
533 ops->locked.ins.ops = ins__find(arch, ops->locked.ins.name);
534
535 if (ops->locked.ins.ops == NULL)
536 goto out_free_ops;
537
538 if (ops->locked.ins.ops->parse &&
539 ops->locked.ins.ops->parse(arch, ops->locked.ops, ms) < 0)
540 goto out_free_ops;
541
542 return 0;
543
544out_free_ops:
545 zfree(&ops->locked.ops);
546 return 0;
547}
548
549static int lock__scnprintf(struct ins *ins, char *bf, size_t size,
550 struct ins_operands *ops, int max_ins_name)
551{
552 int printed;
553
554 if (ops->locked.ins.ops == NULL)
555 return ins__raw_scnprintf(ins, bf, size, ops, max_ins_name);
556
557 printed = scnprintf(bf, size, "%-*s ", max_ins_name, ins->name);
558 return printed + ins__scnprintf(&ops->locked.ins, bf + printed,
559 size - printed, ops->locked.ops, max_ins_name);
560}
561
562static void lock__delete(struct ins_operands *ops)
563{
564 struct ins *ins = &ops->locked.ins;
565
566 if (ins->ops && ins->ops->free)
567 ins->ops->free(ops->locked.ops);
568 else
569 ins__delete(ops->locked.ops);
570
571 zfree(&ops->locked.ops);
572 zfree(&ops->target.raw);
573 zfree(&ops->target.name);
574}
575
576static struct ins_ops lock_ops = {
577 .free = lock__delete,
578 .parse = lock__parse,
579 .scnprintf = lock__scnprintf,
580};
581
582/*
583 * Check if the operand has more than one registers like x86 SIB addressing:
584 * 0x1234(%rax, %rbx, 8)
585 *
586 * But it doesn't care segment selectors like %gs:0x5678(%rcx), so just check
587 * the input string after 'memory_ref_char' if exists.
588 */
589static bool check_multi_regs(struct arch *arch, const char *op)
590{
591 int count = 0;
592
593 if (arch->objdump.register_char == 0)
594 return false;
595
596 if (arch->objdump.memory_ref_char) {
597 op = strchr(op, arch->objdump.memory_ref_char);
598 if (op == NULL)
599 return false;
600 }
601
602 while ((op = strchr(op, arch->objdump.register_char)) != NULL) {
603 count++;
604 op++;
605 }
606
607 return count > 1;
608}
609
610static int mov__parse(struct arch *arch, struct ins_operands *ops, struct map_symbol *ms __maybe_unused)
611{
612 char *s = strchr(ops->raw, ','), *target, *comment, prev;
613
614 if (s == NULL)
615 return -1;
616
617 *s = '\0';
618
619 /*
620 * x86 SIB addressing has something like 0x8(%rax, %rcx, 1)
621 * then it needs to have the closing parenthesis.
622 */
623 if (strchr(ops->raw, '(')) {
624 *s = ',';
625 s = strchr(ops->raw, ')');
626 if (s == NULL || s[1] != ',')
627 return -1;
628 *++s = '\0';
629 }
630
631 ops->source.raw = strdup(ops->raw);
632 *s = ',';
633
634 if (ops->source.raw == NULL)
635 return -1;
636
637 ops->source.multi_regs = check_multi_regs(arch, ops->source.raw);
638
639 target = skip_spaces(++s);
640 comment = strchr(s, arch->objdump.comment_char);
641
642 if (comment != NULL)
643 s = comment - 1;
644 else
645 s = strchr(s, '\0') - 1;
646
647 while (s > target && isspace(s[0]))
648 --s;
649 s++;
650 prev = *s;
651 *s = '\0';
652
653 ops->target.raw = strdup(target);
654 *s = prev;
655
656 if (ops->target.raw == NULL)
657 goto out_free_source;
658
659 ops->target.multi_regs = check_multi_regs(arch, ops->target.raw);
660
661 if (comment == NULL)
662 return 0;
663
664 comment = skip_spaces(comment);
665 comment__symbol(ops->source.raw, comment + 1, &ops->source.addr, &ops->source.name);
666 comment__symbol(ops->target.raw, comment + 1, &ops->target.addr, &ops->target.name);
667
668 return 0;
669
670out_free_source:
671 zfree(&ops->source.raw);
672 return -1;
673}
674
675static int mov__scnprintf(struct ins *ins, char *bf, size_t size,
676 struct ins_operands *ops, int max_ins_name)
677{
678 return scnprintf(bf, size, "%-*s %s,%s", max_ins_name, ins->name,
679 ops->source.name ?: ops->source.raw,
680 ops->target.name ?: ops->target.raw);
681}
682
683static struct ins_ops mov_ops = {
684 .parse = mov__parse,
685 .scnprintf = mov__scnprintf,
686};
687
688static int dec__parse(struct arch *arch __maybe_unused, struct ins_operands *ops, struct map_symbol *ms __maybe_unused)
689{
690 char *target, *comment, *s, prev;
691
692 target = s = ops->raw;
693
694 while (s[0] != '\0' && !isspace(s[0]))
695 ++s;
696 prev = *s;
697 *s = '\0';
698
699 ops->target.raw = strdup(target);
700 *s = prev;
701
702 if (ops->target.raw == NULL)
703 return -1;
704
705 comment = strchr(s, arch->objdump.comment_char);
706 if (comment == NULL)
707 return 0;
708
709 comment = skip_spaces(comment);
710 comment__symbol(ops->target.raw, comment + 1, &ops->target.addr, &ops->target.name);
711
712 return 0;
713}
714
715static int dec__scnprintf(struct ins *ins, char *bf, size_t size,
716 struct ins_operands *ops, int max_ins_name)
717{
718 return scnprintf(bf, size, "%-*s %s", max_ins_name, ins->name,
719 ops->target.name ?: ops->target.raw);
720}
721
722static struct ins_ops dec_ops = {
723 .parse = dec__parse,
724 .scnprintf = dec__scnprintf,
725};
726
727static int nop__scnprintf(struct ins *ins __maybe_unused, char *bf, size_t size,
728 struct ins_operands *ops __maybe_unused, int max_ins_name)
729{
730 return scnprintf(bf, size, "%-*s", max_ins_name, "nop");
731}
732
733static struct ins_ops nop_ops = {
734 .scnprintf = nop__scnprintf,
735};
736
737static struct ins_ops ret_ops = {
738 .scnprintf = ins__raw_scnprintf,
739};
740
741bool ins__is_ret(const struct ins *ins)
742{
743 return ins->ops == &ret_ops;
744}
745
746bool ins__is_lock(const struct ins *ins)
747{
748 return ins->ops == &lock_ops;
749}
750
751static int ins__key_cmp(const void *name, const void *insp)
752{
753 const struct ins *ins = insp;
754
755 return strcmp(name, ins->name);
756}
757
758static int ins__cmp(const void *a, const void *b)
759{
760 const struct ins *ia = a;
761 const struct ins *ib = b;
762
763 return strcmp(ia->name, ib->name);
764}
765
766static void ins__sort(struct arch *arch)
767{
768 const int nmemb = arch->nr_instructions;
769
770 qsort(arch->instructions, nmemb, sizeof(struct ins), ins__cmp);
771}
772
773static struct ins_ops *__ins__find(struct arch *arch, const char *name)
774{
775 struct ins *ins;
776 const int nmemb = arch->nr_instructions;
777
778 if (!arch->sorted_instructions) {
779 ins__sort(arch);
780 arch->sorted_instructions = true;
781 }
782
783 ins = bsearch(name, arch->instructions, nmemb, sizeof(struct ins), ins__key_cmp);
784 if (ins)
785 return ins->ops;
786
787 if (arch->insn_suffix) {
788 char tmp[32];
789 char suffix;
790 size_t len = strlen(name);
791
792 if (len == 0 || len >= sizeof(tmp))
793 return NULL;
794
795 suffix = name[len - 1];
796 if (strchr(arch->insn_suffix, suffix) == NULL)
797 return NULL;
798
799 strcpy(tmp, name);
800 tmp[len - 1] = '\0'; /* remove the suffix and check again */
801
802 ins = bsearch(tmp, arch->instructions, nmemb, sizeof(struct ins), ins__key_cmp);
803 }
804 return ins ? ins->ops : NULL;
805}
806
807static struct ins_ops *ins__find(struct arch *arch, const char *name)
808{
809 struct ins_ops *ops = __ins__find(arch, name);
810
811 if (!ops && arch->associate_instruction_ops)
812 ops = arch->associate_instruction_ops(arch, name);
813
814 return ops;
815}
816
817static int arch__key_cmp(const void *name, const void *archp)
818{
819 const struct arch *arch = archp;
820
821 return strcmp(name, arch->name);
822}
823
824static int arch__cmp(const void *a, const void *b)
825{
826 const struct arch *aa = a;
827 const struct arch *ab = b;
828
829 return strcmp(aa->name, ab->name);
830}
831
832static void arch__sort(void)
833{
834 const int nmemb = ARRAY_SIZE(architectures);
835
836 qsort(architectures, nmemb, sizeof(struct arch), arch__cmp);
837}
838
839static struct arch *arch__find(const char *name)
840{
841 const int nmemb = ARRAY_SIZE(architectures);
842 static bool sorted;
843
844 if (!sorted) {
845 arch__sort();
846 sorted = true;
847 }
848
849 return bsearch(name, architectures, nmemb, sizeof(struct arch), arch__key_cmp);
850}
851
852bool arch__is(struct arch *arch, const char *name)
853{
854 return !strcmp(arch->name, name);
855}
856
857static struct annotated_source *annotated_source__new(void)
858{
859 struct annotated_source *src = zalloc(sizeof(*src));
860
861 if (src != NULL)
862 INIT_LIST_HEAD(&src->source);
863
864 return src;
865}
866
867static __maybe_unused void annotated_source__delete(struct annotated_source *src)
868{
869 if (src == NULL)
870 return;
871 zfree(&src->histograms);
872 free(src);
873}
874
875static int annotated_source__alloc_histograms(struct annotated_source *src,
876 size_t size, int nr_hists)
877{
878 size_t sizeof_sym_hist;
879
880 /*
881 * Add buffer of one element for zero length symbol.
882 * When sample is taken from first instruction of
883 * zero length symbol, perf still resolves it and
884 * shows symbol name in perf report and allows to
885 * annotate it.
886 */
887 if (size == 0)
888 size = 1;
889
890 /* Check for overflow when calculating sizeof_sym_hist */
891 if (size > (SIZE_MAX - sizeof(struct sym_hist)) / sizeof(struct sym_hist_entry))
892 return -1;
893
894 sizeof_sym_hist = (sizeof(struct sym_hist) + size * sizeof(struct sym_hist_entry));
895
896 /* Check for overflow in zalloc argument */
897 if (sizeof_sym_hist > SIZE_MAX / nr_hists)
898 return -1;
899
900 src->sizeof_sym_hist = sizeof_sym_hist;
901 src->nr_histograms = nr_hists;
902 src->histograms = calloc(nr_hists, sizeof_sym_hist) ;
903 return src->histograms ? 0 : -1;
904}
905
906void symbol__annotate_zero_histograms(struct symbol *sym)
907{
908 struct annotation *notes = symbol__annotation(sym);
909
910 annotation__lock(notes);
911 if (notes->src != NULL) {
912 memset(notes->src->histograms, 0,
913 notes->src->nr_histograms * notes->src->sizeof_sym_hist);
914 }
915 if (notes->branch && notes->branch->cycles_hist) {
916 memset(notes->branch->cycles_hist, 0,
917 symbol__size(sym) * sizeof(struct cyc_hist));
918 }
919 annotation__unlock(notes);
920}
921
922static int __symbol__account_cycles(struct cyc_hist *ch,
923 u64 start,
924 unsigned offset, unsigned cycles,
925 unsigned have_start)
926{
927 /*
928 * For now we can only account one basic block per
929 * final jump. But multiple could be overlapping.
930 * Always account the longest one. So when
931 * a shorter one has been already seen throw it away.
932 *
933 * We separately always account the full cycles.
934 */
935 ch[offset].num_aggr++;
936 ch[offset].cycles_aggr += cycles;
937
938 if (cycles > ch[offset].cycles_max)
939 ch[offset].cycles_max = cycles;
940
941 if (ch[offset].cycles_min) {
942 if (cycles && cycles < ch[offset].cycles_min)
943 ch[offset].cycles_min = cycles;
944 } else
945 ch[offset].cycles_min = cycles;
946
947 if (!have_start && ch[offset].have_start)
948 return 0;
949 if (ch[offset].num) {
950 if (have_start && (!ch[offset].have_start ||
951 ch[offset].start > start)) {
952 ch[offset].have_start = 0;
953 ch[offset].cycles = 0;
954 ch[offset].num = 0;
955 if (ch[offset].reset < 0xffff)
956 ch[offset].reset++;
957 } else if (have_start &&
958 ch[offset].start < start)
959 return 0;
960 }
961
962 if (ch[offset].num < NUM_SPARKS)
963 ch[offset].cycles_spark[ch[offset].num] = cycles;
964
965 ch[offset].have_start = have_start;
966 ch[offset].start = start;
967 ch[offset].cycles += cycles;
968 ch[offset].num++;
969 return 0;
970}
971
972static int __symbol__inc_addr_samples(struct map_symbol *ms,
973 struct annotated_source *src, int evidx, u64 addr,
974 struct perf_sample *sample)
975{
976 struct symbol *sym = ms->sym;
977 unsigned offset;
978 struct sym_hist *h;
979
980 pr_debug3("%s: addr=%#" PRIx64 "\n", __func__, map__unmap_ip(ms->map, addr));
981
982 if ((addr < sym->start || addr >= sym->end) &&
983 (addr != sym->end || sym->start != sym->end)) {
984 pr_debug("%s(%d): ERANGE! sym->name=%s, start=%#" PRIx64 ", addr=%#" PRIx64 ", end=%#" PRIx64 "\n",
985 __func__, __LINE__, sym->name, sym->start, addr, sym->end);
986 return -ERANGE;
987 }
988
989 offset = addr - sym->start;
990 h = annotated_source__histogram(src, evidx);
991 if (h == NULL) {
992 pr_debug("%s(%d): ENOMEM! sym->name=%s, start=%#" PRIx64 ", addr=%#" PRIx64 ", end=%#" PRIx64 ", func: %d\n",
993 __func__, __LINE__, sym->name, sym->start, addr, sym->end, sym->type == STT_FUNC);
994 return -ENOMEM;
995 }
996 h->nr_samples++;
997 h->addr[offset].nr_samples++;
998 h->period += sample->period;
999 h->addr[offset].period += sample->period;
1000
1001 pr_debug3("%#" PRIx64 " %s: period++ [addr: %#" PRIx64 ", %#" PRIx64
1002 ", evidx=%d] => nr_samples: %" PRIu64 ", period: %" PRIu64 "\n",
1003 sym->start, sym->name, addr, addr - sym->start, evidx,
1004 h->addr[offset].nr_samples, h->addr[offset].period);
1005 return 0;
1006}
1007
1008struct annotated_branch *annotation__get_branch(struct annotation *notes)
1009{
1010 if (notes == NULL)
1011 return NULL;
1012
1013 if (notes->branch == NULL)
1014 notes->branch = zalloc(sizeof(*notes->branch));
1015
1016 return notes->branch;
1017}
1018
1019static struct cyc_hist *symbol__cycles_hist(struct symbol *sym)
1020{
1021 struct annotation *notes = symbol__annotation(sym);
1022 struct annotated_branch *branch;
1023
1024 branch = annotation__get_branch(notes);
1025 if (branch == NULL)
1026 return NULL;
1027
1028 if (branch->cycles_hist == NULL) {
1029 const size_t size = symbol__size(sym);
1030
1031 branch->cycles_hist = calloc(size, sizeof(struct cyc_hist));
1032 }
1033
1034 return branch->cycles_hist;
1035}
1036
1037struct annotated_source *symbol__hists(struct symbol *sym, int nr_hists)
1038{
1039 struct annotation *notes = symbol__annotation(sym);
1040
1041 if (notes->src == NULL) {
1042 notes->src = annotated_source__new();
1043 if (notes->src == NULL)
1044 return NULL;
1045 goto alloc_histograms;
1046 }
1047
1048 if (notes->src->histograms == NULL) {
1049alloc_histograms:
1050 annotated_source__alloc_histograms(notes->src, symbol__size(sym),
1051 nr_hists);
1052 }
1053
1054 return notes->src;
1055}
1056
1057static int symbol__inc_addr_samples(struct map_symbol *ms,
1058 struct evsel *evsel, u64 addr,
1059 struct perf_sample *sample)
1060{
1061 struct symbol *sym = ms->sym;
1062 struct annotated_source *src;
1063
1064 if (sym == NULL)
1065 return 0;
1066 src = symbol__hists(sym, evsel->evlist->core.nr_entries);
1067 return src ? __symbol__inc_addr_samples(ms, src, evsel->core.idx, addr, sample) : 0;
1068}
1069
1070static int symbol__account_cycles(u64 addr, u64 start,
1071 struct symbol *sym, unsigned cycles)
1072{
1073 struct cyc_hist *cycles_hist;
1074 unsigned offset;
1075
1076 if (sym == NULL)
1077 return 0;
1078 cycles_hist = symbol__cycles_hist(sym);
1079 if (cycles_hist == NULL)
1080 return -ENOMEM;
1081 if (addr < sym->start || addr >= sym->end)
1082 return -ERANGE;
1083
1084 if (start) {
1085 if (start < sym->start || start >= sym->end)
1086 return -ERANGE;
1087 if (start >= addr)
1088 start = 0;
1089 }
1090 offset = addr - sym->start;
1091 return __symbol__account_cycles(cycles_hist,
1092 start ? start - sym->start : 0,
1093 offset, cycles,
1094 !!start);
1095}
1096
1097int addr_map_symbol__account_cycles(struct addr_map_symbol *ams,
1098 struct addr_map_symbol *start,
1099 unsigned cycles)
1100{
1101 u64 saddr = 0;
1102 int err;
1103
1104 if (!cycles)
1105 return 0;
1106
1107 /*
1108 * Only set start when IPC can be computed. We can only
1109 * compute it when the basic block is completely in a single
1110 * function.
1111 * Special case the case when the jump is elsewhere, but
1112 * it starts on the function start.
1113 */
1114 if (start &&
1115 (start->ms.sym == ams->ms.sym ||
1116 (ams->ms.sym &&
1117 start->addr == ams->ms.sym->start + map__start(ams->ms.map))))
1118 saddr = start->al_addr;
1119 if (saddr == 0)
1120 pr_debug2("BB with bad start: addr %"PRIx64" start %"PRIx64" sym %"PRIx64" saddr %"PRIx64"\n",
1121 ams->addr,
1122 start ? start->addr : 0,
1123 ams->ms.sym ? ams->ms.sym->start + map__start(ams->ms.map) : 0,
1124 saddr);
1125 err = symbol__account_cycles(ams->al_addr, saddr, ams->ms.sym, cycles);
1126 if (err)
1127 pr_debug2("account_cycles failed %d\n", err);
1128 return err;
1129}
1130
1131static unsigned annotation__count_insn(struct annotation *notes, u64 start, u64 end)
1132{
1133 unsigned n_insn = 0;
1134 u64 offset;
1135
1136 for (offset = start; offset <= end; offset++) {
1137 if (notes->src->offsets[offset])
1138 n_insn++;
1139 }
1140 return n_insn;
1141}
1142
1143static void annotated_branch__delete(struct annotated_branch *branch)
1144{
1145 if (branch) {
1146 zfree(&branch->cycles_hist);
1147 free(branch);
1148 }
1149}
1150
1151static void annotation__count_and_fill(struct annotation *notes, u64 start, u64 end, struct cyc_hist *ch)
1152{
1153 unsigned n_insn;
1154 unsigned int cover_insn = 0;
1155 u64 offset;
1156
1157 n_insn = annotation__count_insn(notes, start, end);
1158 if (n_insn && ch->num && ch->cycles) {
1159 struct annotated_branch *branch;
1160 float ipc = n_insn / ((double)ch->cycles / (double)ch->num);
1161
1162 /* Hide data when there are too many overlaps. */
1163 if (ch->reset >= 0x7fff)
1164 return;
1165
1166 for (offset = start; offset <= end; offset++) {
1167 struct annotation_line *al = notes->src->offsets[offset];
1168
1169 if (al && al->cycles && al->cycles->ipc == 0.0) {
1170 al->cycles->ipc = ipc;
1171 cover_insn++;
1172 }
1173 }
1174
1175 branch = annotation__get_branch(notes);
1176 if (cover_insn && branch) {
1177 branch->hit_cycles += ch->cycles;
1178 branch->hit_insn += n_insn * ch->num;
1179 branch->cover_insn += cover_insn;
1180 }
1181 }
1182}
1183
1184static int annotation__compute_ipc(struct annotation *notes, size_t size)
1185{
1186 int err = 0;
1187 s64 offset;
1188
1189 if (!notes->branch || !notes->branch->cycles_hist)
1190 return 0;
1191
1192 notes->branch->total_insn = annotation__count_insn(notes, 0, size - 1);
1193 notes->branch->hit_cycles = 0;
1194 notes->branch->hit_insn = 0;
1195 notes->branch->cover_insn = 0;
1196
1197 annotation__lock(notes);
1198 for (offset = size - 1; offset >= 0; --offset) {
1199 struct cyc_hist *ch;
1200
1201 ch = ¬es->branch->cycles_hist[offset];
1202 if (ch && ch->cycles) {
1203 struct annotation_line *al;
1204
1205 al = notes->src->offsets[offset];
1206 if (al && al->cycles == NULL) {
1207 al->cycles = zalloc(sizeof(*al->cycles));
1208 if (al->cycles == NULL) {
1209 err = ENOMEM;
1210 break;
1211 }
1212 }
1213 if (ch->have_start)
1214 annotation__count_and_fill(notes, ch->start, offset, ch);
1215 if (al && ch->num_aggr) {
1216 al->cycles->avg = ch->cycles_aggr / ch->num_aggr;
1217 al->cycles->max = ch->cycles_max;
1218 al->cycles->min = ch->cycles_min;
1219 }
1220 }
1221 }
1222
1223 if (err) {
1224 while (++offset < (s64)size) {
1225 struct cyc_hist *ch = ¬es->branch->cycles_hist[offset];
1226
1227 if (ch && ch->cycles) {
1228 struct annotation_line *al = notes->src->offsets[offset];
1229 if (al)
1230 zfree(&al->cycles);
1231 }
1232 }
1233 }
1234
1235 annotation__unlock(notes);
1236 return 0;
1237}
1238
1239int addr_map_symbol__inc_samples(struct addr_map_symbol *ams, struct perf_sample *sample,
1240 struct evsel *evsel)
1241{
1242 return symbol__inc_addr_samples(&ams->ms, evsel, ams->al_addr, sample);
1243}
1244
1245int hist_entry__inc_addr_samples(struct hist_entry *he, struct perf_sample *sample,
1246 struct evsel *evsel, u64 ip)
1247{
1248 return symbol__inc_addr_samples(&he->ms, evsel, ip, sample);
1249}
1250
1251static void disasm_line__init_ins(struct disasm_line *dl, struct arch *arch, struct map_symbol *ms)
1252{
1253 dl->ins.ops = ins__find(arch, dl->ins.name);
1254
1255 if (!dl->ins.ops)
1256 return;
1257
1258 if (dl->ins.ops->parse && dl->ins.ops->parse(arch, &dl->ops, ms) < 0)
1259 dl->ins.ops = NULL;
1260}
1261
1262static int disasm_line__parse(char *line, const char **namep, char **rawp)
1263{
1264 char tmp, *name = skip_spaces(line);
1265
1266 if (name[0] == '\0')
1267 return -1;
1268
1269 *rawp = name + 1;
1270
1271 while ((*rawp)[0] != '\0' && !isspace((*rawp)[0]))
1272 ++*rawp;
1273
1274 tmp = (*rawp)[0];
1275 (*rawp)[0] = '\0';
1276 *namep = strdup(name);
1277
1278 if (*namep == NULL)
1279 goto out;
1280
1281 (*rawp)[0] = tmp;
1282 *rawp = strim(*rawp);
1283
1284 return 0;
1285
1286out:
1287 return -1;
1288}
1289
1290struct annotate_args {
1291 struct arch *arch;
1292 struct map_symbol ms;
1293 struct evsel *evsel;
1294 struct annotation_options *options;
1295 s64 offset;
1296 char *line;
1297 int line_nr;
1298 char *fileloc;
1299};
1300
1301static void annotation_line__init(struct annotation_line *al,
1302 struct annotate_args *args,
1303 int nr)
1304{
1305 al->offset = args->offset;
1306 al->line = strdup(args->line);
1307 al->line_nr = args->line_nr;
1308 al->fileloc = args->fileloc;
1309 al->data_nr = nr;
1310}
1311
1312static void annotation_line__exit(struct annotation_line *al)
1313{
1314 zfree_srcline(&al->path);
1315 zfree(&al->line);
1316 zfree(&al->cycles);
1317}
1318
1319static size_t disasm_line_size(int nr)
1320{
1321 struct annotation_line *al;
1322
1323 return (sizeof(struct disasm_line) + (sizeof(al->data[0]) * nr));
1324}
1325
1326/*
1327 * Allocating the disasm annotation line data with
1328 * following structure:
1329 *
1330 * -------------------------------------------
1331 * struct disasm_line | struct annotation_line
1332 * -------------------------------------------
1333 *
1334 * We have 'struct annotation_line' member as last member
1335 * of 'struct disasm_line' to have an easy access.
1336 */
1337static struct disasm_line *disasm_line__new(struct annotate_args *args)
1338{
1339 struct disasm_line *dl = NULL;
1340 int nr = 1;
1341
1342 if (evsel__is_group_event(args->evsel))
1343 nr = args->evsel->core.nr_members;
1344
1345 dl = zalloc(disasm_line_size(nr));
1346 if (!dl)
1347 return NULL;
1348
1349 annotation_line__init(&dl->al, args, nr);
1350 if (dl->al.line == NULL)
1351 goto out_delete;
1352
1353 if (args->offset != -1) {
1354 if (disasm_line__parse(dl->al.line, &dl->ins.name, &dl->ops.raw) < 0)
1355 goto out_free_line;
1356
1357 disasm_line__init_ins(dl, args->arch, &args->ms);
1358 }
1359
1360 return dl;
1361
1362out_free_line:
1363 zfree(&dl->al.line);
1364out_delete:
1365 free(dl);
1366 return NULL;
1367}
1368
1369void disasm_line__free(struct disasm_line *dl)
1370{
1371 if (dl->ins.ops && dl->ins.ops->free)
1372 dl->ins.ops->free(&dl->ops);
1373 else
1374 ins__delete(&dl->ops);
1375 zfree(&dl->ins.name);
1376 annotation_line__exit(&dl->al);
1377 free(dl);
1378}
1379
1380int disasm_line__scnprintf(struct disasm_line *dl, char *bf, size_t size, bool raw, int max_ins_name)
1381{
1382 if (raw || !dl->ins.ops)
1383 return scnprintf(bf, size, "%-*s %s", max_ins_name, dl->ins.name, dl->ops.raw);
1384
1385 return ins__scnprintf(&dl->ins, bf, size, &dl->ops, max_ins_name);
1386}
1387
1388void annotation__exit(struct annotation *notes)
1389{
1390 annotated_source__delete(notes->src);
1391 annotated_branch__delete(notes->branch);
1392}
1393
1394static struct sharded_mutex *sharded_mutex;
1395
1396static void annotation__init_sharded_mutex(void)
1397{
1398 /* As many mutexes as there are CPUs. */
1399 sharded_mutex = sharded_mutex__new(cpu__max_present_cpu().cpu);
1400}
1401
1402static size_t annotation__hash(const struct annotation *notes)
1403{
1404 return (size_t)notes;
1405}
1406
1407static struct mutex *annotation__get_mutex(const struct annotation *notes)
1408{
1409 static pthread_once_t once = PTHREAD_ONCE_INIT;
1410
1411 pthread_once(&once, annotation__init_sharded_mutex);
1412 if (!sharded_mutex)
1413 return NULL;
1414
1415 return sharded_mutex__get_mutex(sharded_mutex, annotation__hash(notes));
1416}
1417
1418void annotation__lock(struct annotation *notes)
1419 NO_THREAD_SAFETY_ANALYSIS
1420{
1421 struct mutex *mutex = annotation__get_mutex(notes);
1422
1423 if (mutex)
1424 mutex_lock(mutex);
1425}
1426
1427void annotation__unlock(struct annotation *notes)
1428 NO_THREAD_SAFETY_ANALYSIS
1429{
1430 struct mutex *mutex = annotation__get_mutex(notes);
1431
1432 if (mutex)
1433 mutex_unlock(mutex);
1434}
1435
1436bool annotation__trylock(struct annotation *notes)
1437{
1438 struct mutex *mutex = annotation__get_mutex(notes);
1439
1440 if (!mutex)
1441 return false;
1442
1443 return mutex_trylock(mutex);
1444}
1445
1446
1447static void annotation_line__add(struct annotation_line *al, struct list_head *head)
1448{
1449 list_add_tail(&al->node, head);
1450}
1451
1452struct annotation_line *
1453annotation_line__next(struct annotation_line *pos, struct list_head *head)
1454{
1455 list_for_each_entry_continue(pos, head, node)
1456 if (pos->offset >= 0)
1457 return pos;
1458
1459 return NULL;
1460}
1461
1462static const char *annotate__address_color(struct block_range *br)
1463{
1464 double cov = block_range__coverage(br);
1465
1466 if (cov >= 0) {
1467 /* mark red for >75% coverage */
1468 if (cov > 0.75)
1469 return PERF_COLOR_RED;
1470
1471 /* mark dull for <1% coverage */
1472 if (cov < 0.01)
1473 return PERF_COLOR_NORMAL;
1474 }
1475
1476 return PERF_COLOR_MAGENTA;
1477}
1478
1479static const char *annotate__asm_color(struct block_range *br)
1480{
1481 double cov = block_range__coverage(br);
1482
1483 if (cov >= 0) {
1484 /* mark dull for <1% coverage */
1485 if (cov < 0.01)
1486 return PERF_COLOR_NORMAL;
1487 }
1488
1489 return PERF_COLOR_BLUE;
1490}
1491
1492static void annotate__branch_printf(struct block_range *br, u64 addr)
1493{
1494 bool emit_comment = true;
1495
1496 if (!br)
1497 return;
1498
1499#if 1
1500 if (br->is_target && br->start == addr) {
1501 struct block_range *branch = br;
1502 double p;
1503
1504 /*
1505 * Find matching branch to our target.
1506 */
1507 while (!branch->is_branch)
1508 branch = block_range__next(branch);
1509
1510 p = 100 *(double)br->entry / branch->coverage;
1511
1512 if (p > 0.1) {
1513 if (emit_comment) {
1514 emit_comment = false;
1515 printf("\t#");
1516 }
1517
1518 /*
1519 * The percentage of coverage joined at this target in relation
1520 * to the next branch.
1521 */
1522 printf(" +%.2f%%", p);
1523 }
1524 }
1525#endif
1526 if (br->is_branch && br->end == addr) {
1527 double p = 100*(double)br->taken / br->coverage;
1528
1529 if (p > 0.1) {
1530 if (emit_comment) {
1531 emit_comment = false;
1532 printf("\t#");
1533 }
1534
1535 /*
1536 * The percentage of coverage leaving at this branch, and
1537 * its prediction ratio.
1538 */
1539 printf(" -%.2f%% (p:%.2f%%)", p, 100*(double)br->pred / br->taken);
1540 }
1541 }
1542}
1543
1544static int disasm_line__print(struct disasm_line *dl, u64 start, int addr_fmt_width)
1545{
1546 s64 offset = dl->al.offset;
1547 const u64 addr = start + offset;
1548 struct block_range *br;
1549
1550 br = block_range__find(addr);
1551 color_fprintf(stdout, annotate__address_color(br), " %*" PRIx64 ":", addr_fmt_width, addr);
1552 color_fprintf(stdout, annotate__asm_color(br), "%s", dl->al.line);
1553 annotate__branch_printf(br, addr);
1554 return 0;
1555}
1556
1557static int
1558annotation_line__print(struct annotation_line *al, struct symbol *sym, u64 start,
1559 struct evsel *evsel, u64 len, int min_pcnt, int printed,
1560 int max_lines, struct annotation_line *queue, int addr_fmt_width,
1561 int percent_type)
1562{
1563 struct disasm_line *dl = container_of(al, struct disasm_line, al);
1564 static const char *prev_line;
1565
1566 if (al->offset != -1) {
1567 double max_percent = 0.0;
1568 int i, nr_percent = 1;
1569 const char *color;
1570 struct annotation *notes = symbol__annotation(sym);
1571
1572 for (i = 0; i < al->data_nr; i++) {
1573 double percent;
1574
1575 percent = annotation_data__percent(&al->data[i],
1576 percent_type);
1577
1578 if (percent > max_percent)
1579 max_percent = percent;
1580 }
1581
1582 if (al->data_nr > nr_percent)
1583 nr_percent = al->data_nr;
1584
1585 if (max_percent < min_pcnt)
1586 return -1;
1587
1588 if (max_lines && printed >= max_lines)
1589 return 1;
1590
1591 if (queue != NULL) {
1592 list_for_each_entry_from(queue, ¬es->src->source, node) {
1593 if (queue == al)
1594 break;
1595 annotation_line__print(queue, sym, start, evsel, len,
1596 0, 0, 1, NULL, addr_fmt_width,
1597 percent_type);
1598 }
1599 }
1600
1601 color = get_percent_color(max_percent);
1602
1603 for (i = 0; i < nr_percent; i++) {
1604 struct annotation_data *data = &al->data[i];
1605 double percent;
1606
1607 percent = annotation_data__percent(data, percent_type);
1608 color = get_percent_color(percent);
1609
1610 if (symbol_conf.show_total_period)
1611 color_fprintf(stdout, color, " %11" PRIu64,
1612 data->he.period);
1613 else if (symbol_conf.show_nr_samples)
1614 color_fprintf(stdout, color, " %7" PRIu64,
1615 data->he.nr_samples);
1616 else
1617 color_fprintf(stdout, color, " %7.2f", percent);
1618 }
1619
1620 printf(" : ");
1621
1622 disasm_line__print(dl, start, addr_fmt_width);
1623
1624 /*
1625 * Also color the filename and line if needed, with
1626 * the same color than the percentage. Don't print it
1627 * twice for close colored addr with the same filename:line
1628 */
1629 if (al->path) {
1630 if (!prev_line || strcmp(prev_line, al->path)) {
1631 color_fprintf(stdout, color, " // %s", al->path);
1632 prev_line = al->path;
1633 }
1634 }
1635
1636 printf("\n");
1637 } else if (max_lines && printed >= max_lines)
1638 return 1;
1639 else {
1640 int width = symbol_conf.show_total_period ? 12 : 8;
1641
1642 if (queue)
1643 return -1;
1644
1645 if (evsel__is_group_event(evsel))
1646 width *= evsel->core.nr_members;
1647
1648 if (!*al->line)
1649 printf(" %*s:\n", width, " ");
1650 else
1651 printf(" %*s: %-*d %s\n", width, " ", addr_fmt_width, al->line_nr, al->line);
1652 }
1653
1654 return 0;
1655}
1656
1657/*
1658 * symbol__parse_objdump_line() parses objdump output (with -d --no-show-raw)
1659 * which looks like following
1660 *
1661 * 0000000000415500 <_init>:
1662 * 415500: sub $0x8,%rsp
1663 * 415504: mov 0x2f5ad5(%rip),%rax # 70afe0 <_DYNAMIC+0x2f8>
1664 * 41550b: test %rax,%rax
1665 * 41550e: je 415515 <_init+0x15>
1666 * 415510: callq 416e70 <__gmon_start__@plt>
1667 * 415515: add $0x8,%rsp
1668 * 415519: retq
1669 *
1670 * it will be parsed and saved into struct disasm_line as
1671 * <offset> <name> <ops.raw>
1672 *
1673 * The offset will be a relative offset from the start of the symbol and -1
1674 * means that it's not a disassembly line so should be treated differently.
1675 * The ops.raw part will be parsed further according to type of the instruction.
1676 */
1677static int symbol__parse_objdump_line(struct symbol *sym,
1678 struct annotate_args *args,
1679 char *parsed_line, int *line_nr, char **fileloc)
1680{
1681 struct map *map = args->ms.map;
1682 struct annotation *notes = symbol__annotation(sym);
1683 struct disasm_line *dl;
1684 char *tmp;
1685 s64 line_ip, offset = -1;
1686 regmatch_t match[2];
1687
1688 /* /filename:linenr ? Save line number and ignore. */
1689 if (regexec(&file_lineno, parsed_line, 2, match, 0) == 0) {
1690 *line_nr = atoi(parsed_line + match[1].rm_so);
1691 free(*fileloc);
1692 *fileloc = strdup(parsed_line);
1693 return 0;
1694 }
1695
1696 /* Process hex address followed by ':'. */
1697 line_ip = strtoull(parsed_line, &tmp, 16);
1698 if (parsed_line != tmp && tmp[0] == ':' && tmp[1] != '\0') {
1699 u64 start = map__rip_2objdump(map, sym->start),
1700 end = map__rip_2objdump(map, sym->end);
1701
1702 offset = line_ip - start;
1703 if ((u64)line_ip < start || (u64)line_ip >= end)
1704 offset = -1;
1705 else
1706 parsed_line = tmp + 1;
1707 }
1708
1709 args->offset = offset;
1710 args->line = parsed_line;
1711 args->line_nr = *line_nr;
1712 args->fileloc = *fileloc;
1713 args->ms.sym = sym;
1714
1715 dl = disasm_line__new(args);
1716 (*line_nr)++;
1717
1718 if (dl == NULL)
1719 return -1;
1720
1721 if (!disasm_line__has_local_offset(dl)) {
1722 dl->ops.target.offset = dl->ops.target.addr -
1723 map__rip_2objdump(map, sym->start);
1724 dl->ops.target.offset_avail = true;
1725 }
1726
1727 /* kcore has no symbols, so add the call target symbol */
1728 if (dl->ins.ops && ins__is_call(&dl->ins) && !dl->ops.target.sym) {
1729 struct addr_map_symbol target = {
1730 .addr = dl->ops.target.addr,
1731 .ms = { .map = map, },
1732 };
1733
1734 if (!maps__find_ams(args->ms.maps, &target) &&
1735 target.ms.sym->start == target.al_addr)
1736 dl->ops.target.sym = target.ms.sym;
1737 }
1738
1739 annotation_line__add(&dl->al, ¬es->src->source);
1740 return 0;
1741}
1742
1743static __attribute__((constructor)) void symbol__init_regexpr(void)
1744{
1745 regcomp(&file_lineno, "^/[^:]+:([0-9]+)", REG_EXTENDED);
1746}
1747
1748static void delete_last_nop(struct symbol *sym)
1749{
1750 struct annotation *notes = symbol__annotation(sym);
1751 struct list_head *list = ¬es->src->source;
1752 struct disasm_line *dl;
1753
1754 while (!list_empty(list)) {
1755 dl = list_entry(list->prev, struct disasm_line, al.node);
1756
1757 if (dl->ins.ops) {
1758 if (dl->ins.ops != &nop_ops)
1759 return;
1760 } else {
1761 if (!strstr(dl->al.line, " nop ") &&
1762 !strstr(dl->al.line, " nopl ") &&
1763 !strstr(dl->al.line, " nopw "))
1764 return;
1765 }
1766
1767 list_del_init(&dl->al.node);
1768 disasm_line__free(dl);
1769 }
1770}
1771
1772int symbol__strerror_disassemble(struct map_symbol *ms, int errnum, char *buf, size_t buflen)
1773{
1774 struct dso *dso = map__dso(ms->map);
1775
1776 BUG_ON(buflen == 0);
1777
1778 if (errnum >= 0) {
1779 str_error_r(errnum, buf, buflen);
1780 return 0;
1781 }
1782
1783 switch (errnum) {
1784 case SYMBOL_ANNOTATE_ERRNO__NO_VMLINUX: {
1785 char bf[SBUILD_ID_SIZE + 15] = " with build id ";
1786 char *build_id_msg = NULL;
1787
1788 if (dso->has_build_id) {
1789 build_id__sprintf(&dso->bid, bf + 15);
1790 build_id_msg = bf;
1791 }
1792 scnprintf(buf, buflen,
1793 "No vmlinux file%s\nwas found in the path.\n\n"
1794 "Note that annotation using /proc/kcore requires CAP_SYS_RAWIO capability.\n\n"
1795 "Please use:\n\n"
1796 " perf buildid-cache -vu vmlinux\n\n"
1797 "or:\n\n"
1798 " --vmlinux vmlinux\n", build_id_msg ?: "");
1799 }
1800 break;
1801 case SYMBOL_ANNOTATE_ERRNO__NO_LIBOPCODES_FOR_BPF:
1802 scnprintf(buf, buflen, "Please link with binutils's libopcode to enable BPF annotation");
1803 break;
1804 case SYMBOL_ANNOTATE_ERRNO__ARCH_INIT_REGEXP:
1805 scnprintf(buf, buflen, "Problems with arch specific instruction name regular expressions.");
1806 break;
1807 case SYMBOL_ANNOTATE_ERRNO__ARCH_INIT_CPUID_PARSING:
1808 scnprintf(buf, buflen, "Problems while parsing the CPUID in the arch specific initialization.");
1809 break;
1810 case SYMBOL_ANNOTATE_ERRNO__BPF_INVALID_FILE:
1811 scnprintf(buf, buflen, "Invalid BPF file: %s.", dso->long_name);
1812 break;
1813 case SYMBOL_ANNOTATE_ERRNO__BPF_MISSING_BTF:
1814 scnprintf(buf, buflen, "The %s BPF file has no BTF section, compile with -g or use pahole -J.",
1815 dso->long_name);
1816 break;
1817 default:
1818 scnprintf(buf, buflen, "Internal error: Invalid %d error code\n", errnum);
1819 break;
1820 }
1821
1822 return 0;
1823}
1824
1825static int dso__disassemble_filename(struct dso *dso, char *filename, size_t filename_size)
1826{
1827 char linkname[PATH_MAX];
1828 char *build_id_filename;
1829 char *build_id_path = NULL;
1830 char *pos;
1831 int len;
1832
1833 if (dso->symtab_type == DSO_BINARY_TYPE__KALLSYMS &&
1834 !dso__is_kcore(dso))
1835 return SYMBOL_ANNOTATE_ERRNO__NO_VMLINUX;
1836
1837 build_id_filename = dso__build_id_filename(dso, NULL, 0, false);
1838 if (build_id_filename) {
1839 __symbol__join_symfs(filename, filename_size, build_id_filename);
1840 free(build_id_filename);
1841 } else {
1842 if (dso->has_build_id)
1843 return ENOMEM;
1844 goto fallback;
1845 }
1846
1847 build_id_path = strdup(filename);
1848 if (!build_id_path)
1849 return ENOMEM;
1850
1851 /*
1852 * old style build-id cache has name of XX/XXXXXXX.. while
1853 * new style has XX/XXXXXXX../{elf,kallsyms,vdso}.
1854 * extract the build-id part of dirname in the new style only.
1855 */
1856 pos = strrchr(build_id_path, '/');
1857 if (pos && strlen(pos) < SBUILD_ID_SIZE - 2)
1858 dirname(build_id_path);
1859
1860 if (dso__is_kcore(dso))
1861 goto fallback;
1862
1863 len = readlink(build_id_path, linkname, sizeof(linkname) - 1);
1864 if (len < 0)
1865 goto fallback;
1866
1867 linkname[len] = '\0';
1868 if (strstr(linkname, DSO__NAME_KALLSYMS) ||
1869 access(filename, R_OK)) {
1870fallback:
1871 /*
1872 * If we don't have build-ids or the build-id file isn't in the
1873 * cache, or is just a kallsyms file, well, lets hope that this
1874 * DSO is the same as when 'perf record' ran.
1875 */
1876 if (dso->kernel && dso->long_name[0] == '/')
1877 snprintf(filename, filename_size, "%s", dso->long_name);
1878 else
1879 __symbol__join_symfs(filename, filename_size, dso->long_name);
1880
1881 mutex_lock(&dso->lock);
1882 if (access(filename, R_OK) && errno == ENOENT && dso->nsinfo) {
1883 char *new_name = dso__filename_with_chroot(dso, filename);
1884 if (new_name) {
1885 strlcpy(filename, new_name, filename_size);
1886 free(new_name);
1887 }
1888 }
1889 mutex_unlock(&dso->lock);
1890 }
1891
1892 free(build_id_path);
1893 return 0;
1894}
1895
1896#if defined(HAVE_LIBBFD_SUPPORT) && defined(HAVE_LIBBPF_SUPPORT)
1897#define PACKAGE "perf"
1898#include <bfd.h>
1899#include <dis-asm.h>
1900#include <bpf/bpf.h>
1901#include <bpf/btf.h>
1902#include <bpf/libbpf.h>
1903#include <linux/btf.h>
1904#include <tools/dis-asm-compat.h>
1905
1906static int symbol__disassemble_bpf(struct symbol *sym,
1907 struct annotate_args *args)
1908{
1909 struct annotation *notes = symbol__annotation(sym);
1910 struct bpf_prog_linfo *prog_linfo = NULL;
1911 struct bpf_prog_info_node *info_node;
1912 int len = sym->end - sym->start;
1913 disassembler_ftype disassemble;
1914 struct map *map = args->ms.map;
1915 struct perf_bpil *info_linear;
1916 struct disassemble_info info;
1917 struct dso *dso = map__dso(map);
1918 int pc = 0, count, sub_id;
1919 struct btf *btf = NULL;
1920 char tpath[PATH_MAX];
1921 size_t buf_size;
1922 int nr_skip = 0;
1923 char *buf;
1924 bfd *bfdf;
1925 int ret;
1926 FILE *s;
1927
1928 if (dso->binary_type != DSO_BINARY_TYPE__BPF_PROG_INFO)
1929 return SYMBOL_ANNOTATE_ERRNO__BPF_INVALID_FILE;
1930
1931 pr_debug("%s: handling sym %s addr %" PRIx64 " len %" PRIx64 "\n", __func__,
1932 sym->name, sym->start, sym->end - sym->start);
1933
1934 memset(tpath, 0, sizeof(tpath));
1935 perf_exe(tpath, sizeof(tpath));
1936
1937 bfdf = bfd_openr(tpath, NULL);
1938 if (bfdf == NULL)
1939 abort();
1940
1941 if (!bfd_check_format(bfdf, bfd_object))
1942 abort();
1943
1944 s = open_memstream(&buf, &buf_size);
1945 if (!s) {
1946 ret = errno;
1947 goto out;
1948 }
1949 init_disassemble_info_compat(&info, s,
1950 (fprintf_ftype) fprintf,
1951 fprintf_styled);
1952 info.arch = bfd_get_arch(bfdf);
1953 info.mach = bfd_get_mach(bfdf);
1954
1955 info_node = perf_env__find_bpf_prog_info(dso->bpf_prog.env,
1956 dso->bpf_prog.id);
1957 if (!info_node) {
1958 ret = SYMBOL_ANNOTATE_ERRNO__BPF_MISSING_BTF;
1959 goto out;
1960 }
1961 info_linear = info_node->info_linear;
1962 sub_id = dso->bpf_prog.sub_id;
1963
1964 info.buffer = (void *)(uintptr_t)(info_linear->info.jited_prog_insns);
1965 info.buffer_length = info_linear->info.jited_prog_len;
1966
1967 if (info_linear->info.nr_line_info)
1968 prog_linfo = bpf_prog_linfo__new(&info_linear->info);
1969
1970 if (info_linear->info.btf_id) {
1971 struct btf_node *node;
1972
1973 node = perf_env__find_btf(dso->bpf_prog.env,
1974 info_linear->info.btf_id);
1975 if (node)
1976 btf = btf__new((__u8 *)(node->data),
1977 node->data_size);
1978 }
1979
1980 disassemble_init_for_target(&info);
1981
1982#ifdef DISASM_FOUR_ARGS_SIGNATURE
1983 disassemble = disassembler(info.arch,
1984 bfd_big_endian(bfdf),
1985 info.mach,
1986 bfdf);
1987#else
1988 disassemble = disassembler(bfdf);
1989#endif
1990 if (disassemble == NULL)
1991 abort();
1992
1993 fflush(s);
1994 do {
1995 const struct bpf_line_info *linfo = NULL;
1996 struct disasm_line *dl;
1997 size_t prev_buf_size;
1998 const char *srcline;
1999 u64 addr;
2000
2001 addr = pc + ((u64 *)(uintptr_t)(info_linear->info.jited_ksyms))[sub_id];
2002 count = disassemble(pc, &info);
2003
2004 if (prog_linfo)
2005 linfo = bpf_prog_linfo__lfind_addr_func(prog_linfo,
2006 addr, sub_id,
2007 nr_skip);
2008
2009 if (linfo && btf) {
2010 srcline = btf__name_by_offset(btf, linfo->line_off);
2011 nr_skip++;
2012 } else
2013 srcline = NULL;
2014
2015 fprintf(s, "\n");
2016 prev_buf_size = buf_size;
2017 fflush(s);
2018
2019 if (!annotate_opts.hide_src_code && srcline) {
2020 args->offset = -1;
2021 args->line = strdup(srcline);
2022 args->line_nr = 0;
2023 args->fileloc = NULL;
2024 args->ms.sym = sym;
2025 dl = disasm_line__new(args);
2026 if (dl) {
2027 annotation_line__add(&dl->al,
2028 ¬es->src->source);
2029 }
2030 }
2031
2032 args->offset = pc;
2033 args->line = buf + prev_buf_size;
2034 args->line_nr = 0;
2035 args->fileloc = NULL;
2036 args->ms.sym = sym;
2037 dl = disasm_line__new(args);
2038 if (dl)
2039 annotation_line__add(&dl->al, ¬es->src->source);
2040
2041 pc += count;
2042 } while (count > 0 && pc < len);
2043
2044 ret = 0;
2045out:
2046 free(prog_linfo);
2047 btf__free(btf);
2048 fclose(s);
2049 bfd_close(bfdf);
2050 return ret;
2051}
2052#else // defined(HAVE_LIBBFD_SUPPORT) && defined(HAVE_LIBBPF_SUPPORT)
2053static int symbol__disassemble_bpf(struct symbol *sym __maybe_unused,
2054 struct annotate_args *args __maybe_unused)
2055{
2056 return SYMBOL_ANNOTATE_ERRNO__NO_LIBOPCODES_FOR_BPF;
2057}
2058#endif // defined(HAVE_LIBBFD_SUPPORT) && defined(HAVE_LIBBPF_SUPPORT)
2059
2060static int
2061symbol__disassemble_bpf_image(struct symbol *sym,
2062 struct annotate_args *args)
2063{
2064 struct annotation *notes = symbol__annotation(sym);
2065 struct disasm_line *dl;
2066
2067 args->offset = -1;
2068 args->line = strdup("to be implemented");
2069 args->line_nr = 0;
2070 args->fileloc = NULL;
2071 dl = disasm_line__new(args);
2072 if (dl)
2073 annotation_line__add(&dl->al, ¬es->src->source);
2074
2075 zfree(&args->line);
2076 return 0;
2077}
2078
2079/*
2080 * Possibly create a new version of line with tabs expanded. Returns the
2081 * existing or new line, storage is updated if a new line is allocated. If
2082 * allocation fails then NULL is returned.
2083 */
2084static char *expand_tabs(char *line, char **storage, size_t *storage_len)
2085{
2086 size_t i, src, dst, len, new_storage_len, num_tabs;
2087 char *new_line;
2088 size_t line_len = strlen(line);
2089
2090 for (num_tabs = 0, i = 0; i < line_len; i++)
2091 if (line[i] == '\t')
2092 num_tabs++;
2093
2094 if (num_tabs == 0)
2095 return line;
2096
2097 /*
2098 * Space for the line and '\0', less the leading and trailing
2099 * spaces. Each tab may introduce 7 additional spaces.
2100 */
2101 new_storage_len = line_len + 1 + (num_tabs * 7);
2102
2103 new_line = malloc(new_storage_len);
2104 if (new_line == NULL) {
2105 pr_err("Failure allocating memory for tab expansion\n");
2106 return NULL;
2107 }
2108
2109 /*
2110 * Copy regions starting at src and expand tabs. If there are two
2111 * adjacent tabs then 'src == i', the memcpy is of size 0 and the spaces
2112 * are inserted.
2113 */
2114 for (i = 0, src = 0, dst = 0; i < line_len && num_tabs; i++) {
2115 if (line[i] == '\t') {
2116 len = i - src;
2117 memcpy(&new_line[dst], &line[src], len);
2118 dst += len;
2119 new_line[dst++] = ' ';
2120 while (dst % 8 != 0)
2121 new_line[dst++] = ' ';
2122 src = i + 1;
2123 num_tabs--;
2124 }
2125 }
2126
2127 /* Expand the last region. */
2128 len = line_len - src;
2129 memcpy(&new_line[dst], &line[src], len);
2130 dst += len;
2131 new_line[dst] = '\0';
2132
2133 free(*storage);
2134 *storage = new_line;
2135 *storage_len = new_storage_len;
2136 return new_line;
2137
2138}
2139
2140static int symbol__disassemble(struct symbol *sym, struct annotate_args *args)
2141{
2142 struct annotation_options *opts = &annotate_opts;
2143 struct map *map = args->ms.map;
2144 struct dso *dso = map__dso(map);
2145 char *command;
2146 FILE *file;
2147 char symfs_filename[PATH_MAX];
2148 struct kcore_extract kce;
2149 bool delete_extract = false;
2150 bool decomp = false;
2151 int lineno = 0;
2152 char *fileloc = NULL;
2153 int nline;
2154 char *line;
2155 size_t line_len;
2156 const char *objdump_argv[] = {
2157 "/bin/sh",
2158 "-c",
2159 NULL, /* Will be the objdump command to run. */
2160 "--",
2161 NULL, /* Will be the symfs path. */
2162 NULL,
2163 };
2164 struct child_process objdump_process;
2165 int err = dso__disassemble_filename(dso, symfs_filename, sizeof(symfs_filename));
2166
2167 if (err)
2168 return err;
2169
2170 pr_debug("%s: filename=%s, sym=%s, start=%#" PRIx64 ", end=%#" PRIx64 "\n", __func__,
2171 symfs_filename, sym->name, map__unmap_ip(map, sym->start),
2172 map__unmap_ip(map, sym->end));
2173
2174 pr_debug("annotating [%p] %30s : [%p] %30s\n",
2175 dso, dso->long_name, sym, sym->name);
2176
2177 if (dso->binary_type == DSO_BINARY_TYPE__BPF_PROG_INFO) {
2178 return symbol__disassemble_bpf(sym, args);
2179 } else if (dso->binary_type == DSO_BINARY_TYPE__BPF_IMAGE) {
2180 return symbol__disassemble_bpf_image(sym, args);
2181 } else if (dso__is_kcore(dso)) {
2182 kce.kcore_filename = symfs_filename;
2183 kce.addr = map__rip_2objdump(map, sym->start);
2184 kce.offs = sym->start;
2185 kce.len = sym->end - sym->start;
2186 if (!kcore_extract__create(&kce)) {
2187 delete_extract = true;
2188 strlcpy(symfs_filename, kce.extract_filename,
2189 sizeof(symfs_filename));
2190 }
2191 } else if (dso__needs_decompress(dso)) {
2192 char tmp[KMOD_DECOMP_LEN];
2193
2194 if (dso__decompress_kmodule_path(dso, symfs_filename,
2195 tmp, sizeof(tmp)) < 0)
2196 return -1;
2197
2198 decomp = true;
2199 strcpy(symfs_filename, tmp);
2200 }
2201
2202 err = asprintf(&command,
2203 "%s %s%s --start-address=0x%016" PRIx64
2204 " --stop-address=0x%016" PRIx64
2205 " %s -d %s %s %s %c%s%c %s%s -C \"$1\"",
2206 opts->objdump_path ?: "objdump",
2207 opts->disassembler_style ? "-M " : "",
2208 opts->disassembler_style ?: "",
2209 map__rip_2objdump(map, sym->start),
2210 map__rip_2objdump(map, sym->end),
2211 opts->show_linenr ? "-l" : "",
2212 opts->show_asm_raw ? "" : "--no-show-raw-insn",
2213 opts->annotate_src ? "-S" : "",
2214 opts->prefix ? "--prefix " : "",
2215 opts->prefix ? '"' : ' ',
2216 opts->prefix ?: "",
2217 opts->prefix ? '"' : ' ',
2218 opts->prefix_strip ? "--prefix-strip=" : "",
2219 opts->prefix_strip ?: "");
2220
2221 if (err < 0) {
2222 pr_err("Failure allocating memory for the command to run\n");
2223 goto out_remove_tmp;
2224 }
2225
2226 pr_debug("Executing: %s\n", command);
2227
2228 objdump_argv[2] = command;
2229 objdump_argv[4] = symfs_filename;
2230
2231 /* Create a pipe to read from for stdout */
2232 memset(&objdump_process, 0, sizeof(objdump_process));
2233 objdump_process.argv = objdump_argv;
2234 objdump_process.out = -1;
2235 objdump_process.err = -1;
2236 objdump_process.no_stderr = 1;
2237 if (start_command(&objdump_process)) {
2238 pr_err("Failure starting to run %s\n", command);
2239 err = -1;
2240 goto out_free_command;
2241 }
2242
2243 file = fdopen(objdump_process.out, "r");
2244 if (!file) {
2245 pr_err("Failure creating FILE stream for %s\n", command);
2246 /*
2247 * If we were using debug info should retry with
2248 * original binary.
2249 */
2250 err = -1;
2251 goto out_close_stdout;
2252 }
2253
2254 /* Storage for getline. */
2255 line = NULL;
2256 line_len = 0;
2257
2258 nline = 0;
2259 while (!feof(file)) {
2260 const char *match;
2261 char *expanded_line;
2262
2263 if (getline(&line, &line_len, file) < 0 || !line)
2264 break;
2265
2266 /* Skip lines containing "filename:" */
2267 match = strstr(line, symfs_filename);
2268 if (match && match[strlen(symfs_filename)] == ':')
2269 continue;
2270
2271 expanded_line = strim(line);
2272 expanded_line = expand_tabs(expanded_line, &line, &line_len);
2273 if (!expanded_line)
2274 break;
2275
2276 /*
2277 * The source code line number (lineno) needs to be kept in
2278 * across calls to symbol__parse_objdump_line(), so that it
2279 * can associate it with the instructions till the next one.
2280 * See disasm_line__new() and struct disasm_line::line_nr.
2281 */
2282 if (symbol__parse_objdump_line(sym, args, expanded_line,
2283 &lineno, &fileloc) < 0)
2284 break;
2285 nline++;
2286 }
2287 free(line);
2288 free(fileloc);
2289
2290 err = finish_command(&objdump_process);
2291 if (err)
2292 pr_err("Error running %s\n", command);
2293
2294 if (nline == 0) {
2295 err = -1;
2296 pr_err("No output from %s\n", command);
2297 }
2298
2299 /*
2300 * kallsyms does not have symbol sizes so there may a nop at the end.
2301 * Remove it.
2302 */
2303 if (dso__is_kcore(dso))
2304 delete_last_nop(sym);
2305
2306 fclose(file);
2307
2308out_close_stdout:
2309 close(objdump_process.out);
2310
2311out_free_command:
2312 free(command);
2313
2314out_remove_tmp:
2315 if (decomp)
2316 unlink(symfs_filename);
2317
2318 if (delete_extract)
2319 kcore_extract__delete(&kce);
2320
2321 return err;
2322}
2323
2324static void calc_percent(struct sym_hist *sym_hist,
2325 struct hists *hists,
2326 struct annotation_data *data,
2327 s64 offset, s64 end)
2328{
2329 unsigned int hits = 0;
2330 u64 period = 0;
2331
2332 while (offset < end) {
2333 hits += sym_hist->addr[offset].nr_samples;
2334 period += sym_hist->addr[offset].period;
2335 ++offset;
2336 }
2337
2338 if (sym_hist->nr_samples) {
2339 data->he.period = period;
2340 data->he.nr_samples = hits;
2341 data->percent[PERCENT_HITS_LOCAL] = 100.0 * hits / sym_hist->nr_samples;
2342 }
2343
2344 if (hists->stats.nr_non_filtered_samples)
2345 data->percent[PERCENT_HITS_GLOBAL] = 100.0 * hits / hists->stats.nr_non_filtered_samples;
2346
2347 if (sym_hist->period)
2348 data->percent[PERCENT_PERIOD_LOCAL] = 100.0 * period / sym_hist->period;
2349
2350 if (hists->stats.total_period)
2351 data->percent[PERCENT_PERIOD_GLOBAL] = 100.0 * period / hists->stats.total_period;
2352}
2353
2354static void annotation__calc_percent(struct annotation *notes,
2355 struct evsel *leader, s64 len)
2356{
2357 struct annotation_line *al, *next;
2358 struct evsel *evsel;
2359
2360 list_for_each_entry(al, ¬es->src->source, node) {
2361 s64 end;
2362 int i = 0;
2363
2364 if (al->offset == -1)
2365 continue;
2366
2367 next = annotation_line__next(al, ¬es->src->source);
2368 end = next ? next->offset : len;
2369
2370 for_each_group_evsel(evsel, leader) {
2371 struct hists *hists = evsel__hists(evsel);
2372 struct annotation_data *data;
2373 struct sym_hist *sym_hist;
2374
2375 BUG_ON(i >= al->data_nr);
2376
2377 sym_hist = annotation__histogram(notes, evsel->core.idx);
2378 data = &al->data[i++];
2379
2380 calc_percent(sym_hist, hists, data, al->offset, end);
2381 }
2382 }
2383}
2384
2385void symbol__calc_percent(struct symbol *sym, struct evsel *evsel)
2386{
2387 struct annotation *notes = symbol__annotation(sym);
2388
2389 annotation__calc_percent(notes, evsel, symbol__size(sym));
2390}
2391
2392static int evsel__get_arch(struct evsel *evsel, struct arch **parch)
2393{
2394 struct perf_env *env = evsel__env(evsel);
2395 const char *arch_name = perf_env__arch(env);
2396 struct arch *arch;
2397 int err;
2398
2399 if (!arch_name)
2400 return errno;
2401
2402 *parch = arch = arch__find(arch_name);
2403 if (arch == NULL) {
2404 pr_err("%s: unsupported arch %s\n", __func__, arch_name);
2405 return ENOTSUP;
2406 }
2407
2408 if (arch->init) {
2409 err = arch->init(arch, env ? env->cpuid : NULL);
2410 if (err) {
2411 pr_err("%s: failed to initialize %s arch priv area\n",
2412 __func__, arch->name);
2413 return err;
2414 }
2415 }
2416 return 0;
2417}
2418
2419int symbol__annotate(struct map_symbol *ms, struct evsel *evsel,
2420 struct arch **parch)
2421{
2422 struct symbol *sym = ms->sym;
2423 struct annotation *notes = symbol__annotation(sym);
2424 struct annotate_args args = {
2425 .evsel = evsel,
2426 .options = &annotate_opts,
2427 };
2428 struct arch *arch = NULL;
2429 int err;
2430
2431 err = evsel__get_arch(evsel, &arch);
2432 if (err < 0)
2433 return err;
2434
2435 if (parch)
2436 *parch = arch;
2437
2438 args.arch = arch;
2439 args.ms = *ms;
2440 if (annotate_opts.full_addr)
2441 notes->start = map__objdump_2mem(ms->map, ms->sym->start);
2442 else
2443 notes->start = map__rip_2objdump(ms->map, ms->sym->start);
2444
2445 return symbol__disassemble(sym, &args);
2446}
2447
2448static void insert_source_line(struct rb_root *root, struct annotation_line *al)
2449{
2450 struct annotation_line *iter;
2451 struct rb_node **p = &root->rb_node;
2452 struct rb_node *parent = NULL;
2453 unsigned int percent_type = annotate_opts.percent_type;
2454 int i, ret;
2455
2456 while (*p != NULL) {
2457 parent = *p;
2458 iter = rb_entry(parent, struct annotation_line, rb_node);
2459
2460 ret = strcmp(iter->path, al->path);
2461 if (ret == 0) {
2462 for (i = 0; i < al->data_nr; i++) {
2463 iter->data[i].percent_sum += annotation_data__percent(&al->data[i],
2464 percent_type);
2465 }
2466 return;
2467 }
2468
2469 if (ret < 0)
2470 p = &(*p)->rb_left;
2471 else
2472 p = &(*p)->rb_right;
2473 }
2474
2475 for (i = 0; i < al->data_nr; i++) {
2476 al->data[i].percent_sum = annotation_data__percent(&al->data[i],
2477 percent_type);
2478 }
2479
2480 rb_link_node(&al->rb_node, parent, p);
2481 rb_insert_color(&al->rb_node, root);
2482}
2483
2484static int cmp_source_line(struct annotation_line *a, struct annotation_line *b)
2485{
2486 int i;
2487
2488 for (i = 0; i < a->data_nr; i++) {
2489 if (a->data[i].percent_sum == b->data[i].percent_sum)
2490 continue;
2491 return a->data[i].percent_sum > b->data[i].percent_sum;
2492 }
2493
2494 return 0;
2495}
2496
2497static void __resort_source_line(struct rb_root *root, struct annotation_line *al)
2498{
2499 struct annotation_line *iter;
2500 struct rb_node **p = &root->rb_node;
2501 struct rb_node *parent = NULL;
2502
2503 while (*p != NULL) {
2504 parent = *p;
2505 iter = rb_entry(parent, struct annotation_line, rb_node);
2506
2507 if (cmp_source_line(al, iter))
2508 p = &(*p)->rb_left;
2509 else
2510 p = &(*p)->rb_right;
2511 }
2512
2513 rb_link_node(&al->rb_node, parent, p);
2514 rb_insert_color(&al->rb_node, root);
2515}
2516
2517static void resort_source_line(struct rb_root *dest_root, struct rb_root *src_root)
2518{
2519 struct annotation_line *al;
2520 struct rb_node *node;
2521
2522 node = rb_first(src_root);
2523 while (node) {
2524 struct rb_node *next;
2525
2526 al = rb_entry(node, struct annotation_line, rb_node);
2527 next = rb_next(node);
2528 rb_erase(node, src_root);
2529
2530 __resort_source_line(dest_root, al);
2531 node = next;
2532 }
2533}
2534
2535static void print_summary(struct rb_root *root, const char *filename)
2536{
2537 struct annotation_line *al;
2538 struct rb_node *node;
2539
2540 printf("\nSorted summary for file %s\n", filename);
2541 printf("----------------------------------------------\n\n");
2542
2543 if (RB_EMPTY_ROOT(root)) {
2544 printf(" Nothing higher than %1.1f%%\n", MIN_GREEN);
2545 return;
2546 }
2547
2548 node = rb_first(root);
2549 while (node) {
2550 double percent, percent_max = 0.0;
2551 const char *color;
2552 char *path;
2553 int i;
2554
2555 al = rb_entry(node, struct annotation_line, rb_node);
2556 for (i = 0; i < al->data_nr; i++) {
2557 percent = al->data[i].percent_sum;
2558 color = get_percent_color(percent);
2559 color_fprintf(stdout, color, " %7.2f", percent);
2560
2561 if (percent > percent_max)
2562 percent_max = percent;
2563 }
2564
2565 path = al->path;
2566 color = get_percent_color(percent_max);
2567 color_fprintf(stdout, color, " %s\n", path);
2568
2569 node = rb_next(node);
2570 }
2571}
2572
2573static void symbol__annotate_hits(struct symbol *sym, struct evsel *evsel)
2574{
2575 struct annotation *notes = symbol__annotation(sym);
2576 struct sym_hist *h = annotation__histogram(notes, evsel->core.idx);
2577 u64 len = symbol__size(sym), offset;
2578
2579 for (offset = 0; offset < len; ++offset)
2580 if (h->addr[offset].nr_samples != 0)
2581 printf("%*" PRIx64 ": %" PRIu64 "\n", BITS_PER_LONG / 2,
2582 sym->start + offset, h->addr[offset].nr_samples);
2583 printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->nr_samples", h->nr_samples);
2584}
2585
2586static int annotated_source__addr_fmt_width(struct list_head *lines, u64 start)
2587{
2588 char bf[32];
2589 struct annotation_line *line;
2590
2591 list_for_each_entry_reverse(line, lines, node) {
2592 if (line->offset != -1)
2593 return scnprintf(bf, sizeof(bf), "%" PRIx64, start + line->offset);
2594 }
2595
2596 return 0;
2597}
2598
2599int symbol__annotate_printf(struct map_symbol *ms, struct evsel *evsel)
2600{
2601 struct map *map = ms->map;
2602 struct symbol *sym = ms->sym;
2603 struct dso *dso = map__dso(map);
2604 char *filename;
2605 const char *d_filename;
2606 const char *evsel_name = evsel__name(evsel);
2607 struct annotation *notes = symbol__annotation(sym);
2608 struct sym_hist *h = annotation__histogram(notes, evsel->core.idx);
2609 struct annotation_line *pos, *queue = NULL;
2610 struct annotation_options *opts = &annotate_opts;
2611 u64 start = map__rip_2objdump(map, sym->start);
2612 int printed = 2, queue_len = 0, addr_fmt_width;
2613 int more = 0;
2614 bool context = opts->context;
2615 u64 len;
2616 int width = symbol_conf.show_total_period ? 12 : 8;
2617 int graph_dotted_len;
2618 char buf[512];
2619
2620 filename = strdup(dso->long_name);
2621 if (!filename)
2622 return -ENOMEM;
2623
2624 if (opts->full_path)
2625 d_filename = filename;
2626 else
2627 d_filename = basename(filename);
2628
2629 len = symbol__size(sym);
2630
2631 if (evsel__is_group_event(evsel)) {
2632 width *= evsel->core.nr_members;
2633 evsel__group_desc(evsel, buf, sizeof(buf));
2634 evsel_name = buf;
2635 }
2636
2637 graph_dotted_len = printf(" %-*.*s| Source code & Disassembly of %s for %s (%" PRIu64 " samples, "
2638 "percent: %s)\n",
2639 width, width, symbol_conf.show_total_period ? "Period" :
2640 symbol_conf.show_nr_samples ? "Samples" : "Percent",
2641 d_filename, evsel_name, h->nr_samples,
2642 percent_type_str(opts->percent_type));
2643
2644 printf("%-*.*s----\n",
2645 graph_dotted_len, graph_dotted_len, graph_dotted_line);
2646
2647 if (verbose > 0)
2648 symbol__annotate_hits(sym, evsel);
2649
2650 addr_fmt_width = annotated_source__addr_fmt_width(¬es->src->source, start);
2651
2652 list_for_each_entry(pos, ¬es->src->source, node) {
2653 int err;
2654
2655 if (context && queue == NULL) {
2656 queue = pos;
2657 queue_len = 0;
2658 }
2659
2660 err = annotation_line__print(pos, sym, start, evsel, len,
2661 opts->min_pcnt, printed, opts->max_lines,
2662 queue, addr_fmt_width, opts->percent_type);
2663
2664 switch (err) {
2665 case 0:
2666 ++printed;
2667 if (context) {
2668 printed += queue_len;
2669 queue = NULL;
2670 queue_len = 0;
2671 }
2672 break;
2673 case 1:
2674 /* filtered by max_lines */
2675 ++more;
2676 break;
2677 case -1:
2678 default:
2679 /*
2680 * Filtered by min_pcnt or non IP lines when
2681 * context != 0
2682 */
2683 if (!context)
2684 break;
2685 if (queue_len == context)
2686 queue = list_entry(queue->node.next, typeof(*queue), node);
2687 else
2688 ++queue_len;
2689 break;
2690 }
2691 }
2692
2693 free(filename);
2694
2695 return more;
2696}
2697
2698static void FILE__set_percent_color(void *fp __maybe_unused,
2699 double percent __maybe_unused,
2700 bool current __maybe_unused)
2701{
2702}
2703
2704static int FILE__set_jumps_percent_color(void *fp __maybe_unused,
2705 int nr __maybe_unused, bool current __maybe_unused)
2706{
2707 return 0;
2708}
2709
2710static int FILE__set_color(void *fp __maybe_unused, int color __maybe_unused)
2711{
2712 return 0;
2713}
2714
2715static void FILE__printf(void *fp, const char *fmt, ...)
2716{
2717 va_list args;
2718
2719 va_start(args, fmt);
2720 vfprintf(fp, fmt, args);
2721 va_end(args);
2722}
2723
2724static void FILE__write_graph(void *fp, int graph)
2725{
2726 const char *s;
2727 switch (graph) {
2728
2729 case DARROW_CHAR: s = "↓"; break;
2730 case UARROW_CHAR: s = "↑"; break;
2731 case LARROW_CHAR: s = "←"; break;
2732 case RARROW_CHAR: s = "→"; break;
2733 default: s = "?"; break;
2734 }
2735
2736 fputs(s, fp);
2737}
2738
2739static int symbol__annotate_fprintf2(struct symbol *sym, FILE *fp)
2740{
2741 struct annotation *notes = symbol__annotation(sym);
2742 struct annotation_write_ops wops = {
2743 .first_line = true,
2744 .obj = fp,
2745 .set_color = FILE__set_color,
2746 .set_percent_color = FILE__set_percent_color,
2747 .set_jumps_percent_color = FILE__set_jumps_percent_color,
2748 .printf = FILE__printf,
2749 .write_graph = FILE__write_graph,
2750 };
2751 struct annotation_line *al;
2752
2753 list_for_each_entry(al, ¬es->src->source, node) {
2754 if (annotation_line__filter(al))
2755 continue;
2756 annotation_line__write(al, notes, &wops);
2757 fputc('\n', fp);
2758 wops.first_line = false;
2759 }
2760
2761 return 0;
2762}
2763
2764int map_symbol__annotation_dump(struct map_symbol *ms, struct evsel *evsel)
2765{
2766 const char *ev_name = evsel__name(evsel);
2767 char buf[1024];
2768 char *filename;
2769 int err = -1;
2770 FILE *fp;
2771
2772 if (asprintf(&filename, "%s.annotation", ms->sym->name) < 0)
2773 return -1;
2774
2775 fp = fopen(filename, "w");
2776 if (fp == NULL)
2777 goto out_free_filename;
2778
2779 if (evsel__is_group_event(evsel)) {
2780 evsel__group_desc(evsel, buf, sizeof(buf));
2781 ev_name = buf;
2782 }
2783
2784 fprintf(fp, "%s() %s\nEvent: %s\n\n",
2785 ms->sym->name, map__dso(ms->map)->long_name, ev_name);
2786 symbol__annotate_fprintf2(ms->sym, fp);
2787
2788 fclose(fp);
2789 err = 0;
2790out_free_filename:
2791 free(filename);
2792 return err;
2793}
2794
2795void symbol__annotate_zero_histogram(struct symbol *sym, int evidx)
2796{
2797 struct annotation *notes = symbol__annotation(sym);
2798 struct sym_hist *h = annotation__histogram(notes, evidx);
2799
2800 memset(h, 0, notes->src->sizeof_sym_hist);
2801}
2802
2803void symbol__annotate_decay_histogram(struct symbol *sym, int evidx)
2804{
2805 struct annotation *notes = symbol__annotation(sym);
2806 struct sym_hist *h = annotation__histogram(notes, evidx);
2807 int len = symbol__size(sym), offset;
2808
2809 h->nr_samples = 0;
2810 for (offset = 0; offset < len; ++offset) {
2811 h->addr[offset].nr_samples = h->addr[offset].nr_samples * 7 / 8;
2812 h->nr_samples += h->addr[offset].nr_samples;
2813 }
2814}
2815
2816void annotated_source__purge(struct annotated_source *as)
2817{
2818 struct annotation_line *al, *n;
2819
2820 list_for_each_entry_safe(al, n, &as->source, node) {
2821 list_del_init(&al->node);
2822 disasm_line__free(disasm_line(al));
2823 }
2824}
2825
2826static size_t disasm_line__fprintf(struct disasm_line *dl, FILE *fp)
2827{
2828 size_t printed;
2829
2830 if (dl->al.offset == -1)
2831 return fprintf(fp, "%s\n", dl->al.line);
2832
2833 printed = fprintf(fp, "%#" PRIx64 " %s", dl->al.offset, dl->ins.name);
2834
2835 if (dl->ops.raw[0] != '\0') {
2836 printed += fprintf(fp, "%.*s %s\n", 6 - (int)printed, " ",
2837 dl->ops.raw);
2838 }
2839
2840 return printed + fprintf(fp, "\n");
2841}
2842
2843size_t disasm__fprintf(struct list_head *head, FILE *fp)
2844{
2845 struct disasm_line *pos;
2846 size_t printed = 0;
2847
2848 list_for_each_entry(pos, head, al.node)
2849 printed += disasm_line__fprintf(pos, fp);
2850
2851 return printed;
2852}
2853
2854bool disasm_line__is_valid_local_jump(struct disasm_line *dl, struct symbol *sym)
2855{
2856 if (!dl || !dl->ins.ops || !ins__is_jump(&dl->ins) ||
2857 !disasm_line__has_local_offset(dl) || dl->ops.target.offset < 0 ||
2858 dl->ops.target.offset >= (s64)symbol__size(sym))
2859 return false;
2860
2861 return true;
2862}
2863
2864void annotation__mark_jump_targets(struct annotation *notes, struct symbol *sym)
2865{
2866 u64 offset, size = symbol__size(sym);
2867
2868 /* PLT symbols contain external offsets */
2869 if (strstr(sym->name, "@plt"))
2870 return;
2871
2872 for (offset = 0; offset < size; ++offset) {
2873 struct annotation_line *al = notes->src->offsets[offset];
2874 struct disasm_line *dl;
2875
2876 dl = disasm_line(al);
2877
2878 if (!disasm_line__is_valid_local_jump(dl, sym))
2879 continue;
2880
2881 al = notes->src->offsets[dl->ops.target.offset];
2882
2883 /*
2884 * FIXME: Oops, no jump target? Buggy disassembler? Or do we
2885 * have to adjust to the previous offset?
2886 */
2887 if (al == NULL)
2888 continue;
2889
2890 if (++al->jump_sources > notes->max_jump_sources)
2891 notes->max_jump_sources = al->jump_sources;
2892 }
2893}
2894
2895void annotation__set_offsets(struct annotation *notes, s64 size)
2896{
2897 struct annotation_line *al;
2898 struct annotated_source *src = notes->src;
2899
2900 src->max_line_len = 0;
2901 src->nr_entries = 0;
2902 src->nr_asm_entries = 0;
2903
2904 list_for_each_entry(al, &src->source, node) {
2905 size_t line_len = strlen(al->line);
2906
2907 if (src->max_line_len < line_len)
2908 src->max_line_len = line_len;
2909 al->idx = src->nr_entries++;
2910 if (al->offset != -1) {
2911 al->idx_asm = src->nr_asm_entries++;
2912 /*
2913 * FIXME: short term bandaid to cope with assembly
2914 * routines that comes with labels in the same column
2915 * as the address in objdump, sigh.
2916 *
2917 * E.g. copy_user_generic_unrolled
2918 */
2919 if (al->offset < size)
2920 notes->src->offsets[al->offset] = al;
2921 } else
2922 al->idx_asm = -1;
2923 }
2924}
2925
2926static inline int width_jumps(int n)
2927{
2928 if (n >= 100)
2929 return 5;
2930 if (n / 10)
2931 return 2;
2932 return 1;
2933}
2934
2935static int annotation__max_ins_name(struct annotation *notes)
2936{
2937 int max_name = 0, len;
2938 struct annotation_line *al;
2939
2940 list_for_each_entry(al, ¬es->src->source, node) {
2941 if (al->offset == -1)
2942 continue;
2943
2944 len = strlen(disasm_line(al)->ins.name);
2945 if (max_name < len)
2946 max_name = len;
2947 }
2948
2949 return max_name;
2950}
2951
2952void annotation__init_column_widths(struct annotation *notes, struct symbol *sym)
2953{
2954 notes->widths.addr = notes->widths.target =
2955 notes->widths.min_addr = hex_width(symbol__size(sym));
2956 notes->widths.max_addr = hex_width(sym->end);
2957 notes->widths.jumps = width_jumps(notes->max_jump_sources);
2958 notes->widths.max_ins_name = annotation__max_ins_name(notes);
2959}
2960
2961void annotation__update_column_widths(struct annotation *notes)
2962{
2963 if (annotate_opts.use_offset)
2964 notes->widths.target = notes->widths.min_addr;
2965 else if (annotate_opts.full_addr)
2966 notes->widths.target = BITS_PER_LONG / 4;
2967 else
2968 notes->widths.target = notes->widths.max_addr;
2969
2970 notes->widths.addr = notes->widths.target;
2971
2972 if (annotate_opts.show_nr_jumps)
2973 notes->widths.addr += notes->widths.jumps + 1;
2974}
2975
2976void annotation__toggle_full_addr(struct annotation *notes, struct map_symbol *ms)
2977{
2978 annotate_opts.full_addr = !annotate_opts.full_addr;
2979
2980 if (annotate_opts.full_addr)
2981 notes->start = map__objdump_2mem(ms->map, ms->sym->start);
2982 else
2983 notes->start = map__rip_2objdump(ms->map, ms->sym->start);
2984
2985 annotation__update_column_widths(notes);
2986}
2987
2988static void annotation__calc_lines(struct annotation *notes, struct map *map,
2989 struct rb_root *root)
2990{
2991 struct annotation_line *al;
2992 struct rb_root tmp_root = RB_ROOT;
2993
2994 list_for_each_entry(al, ¬es->src->source, node) {
2995 double percent_max = 0.0;
2996 int i;
2997
2998 for (i = 0; i < al->data_nr; i++) {
2999 double percent;
3000
3001 percent = annotation_data__percent(&al->data[i],
3002 annotate_opts.percent_type);
3003
3004 if (percent > percent_max)
3005 percent_max = percent;
3006 }
3007
3008 if (percent_max <= 0.5)
3009 continue;
3010
3011 al->path = get_srcline(map__dso(map), notes->start + al->offset, NULL,
3012 false, true, notes->start + al->offset);
3013 insert_source_line(&tmp_root, al);
3014 }
3015
3016 resort_source_line(root, &tmp_root);
3017}
3018
3019static void symbol__calc_lines(struct map_symbol *ms, struct rb_root *root)
3020{
3021 struct annotation *notes = symbol__annotation(ms->sym);
3022
3023 annotation__calc_lines(notes, ms->map, root);
3024}
3025
3026int symbol__tty_annotate2(struct map_symbol *ms, struct evsel *evsel)
3027{
3028 struct dso *dso = map__dso(ms->map);
3029 struct symbol *sym = ms->sym;
3030 struct rb_root source_line = RB_ROOT;
3031 struct hists *hists = evsel__hists(evsel);
3032 char buf[1024];
3033 int err;
3034
3035 err = symbol__annotate2(ms, evsel, NULL);
3036 if (err) {
3037 char msg[BUFSIZ];
3038
3039 dso->annotate_warned = true;
3040 symbol__strerror_disassemble(ms, err, msg, sizeof(msg));
3041 ui__error("Couldn't annotate %s:\n%s", sym->name, msg);
3042 return -1;
3043 }
3044
3045 if (annotate_opts.print_lines) {
3046 srcline_full_filename = annotate_opts.full_path;
3047 symbol__calc_lines(ms, &source_line);
3048 print_summary(&source_line, dso->long_name);
3049 }
3050
3051 hists__scnprintf_title(hists, buf, sizeof(buf));
3052 fprintf(stdout, "%s, [percent: %s]\n%s() %s\n",
3053 buf, percent_type_str(annotate_opts.percent_type), sym->name,
3054 dso->long_name);
3055 symbol__annotate_fprintf2(sym, stdout);
3056
3057 annotated_source__purge(symbol__annotation(sym)->src);
3058
3059 return 0;
3060}
3061
3062int symbol__tty_annotate(struct map_symbol *ms, struct evsel *evsel)
3063{
3064 struct dso *dso = map__dso(ms->map);
3065 struct symbol *sym = ms->sym;
3066 struct rb_root source_line = RB_ROOT;
3067 int err;
3068
3069 err = symbol__annotate(ms, evsel, NULL);
3070 if (err) {
3071 char msg[BUFSIZ];
3072
3073 dso->annotate_warned = true;
3074 symbol__strerror_disassemble(ms, err, msg, sizeof(msg));
3075 ui__error("Couldn't annotate %s:\n%s", sym->name, msg);
3076 return -1;
3077 }
3078
3079 symbol__calc_percent(sym, evsel);
3080
3081 if (annotate_opts.print_lines) {
3082 srcline_full_filename = annotate_opts.full_path;
3083 symbol__calc_lines(ms, &source_line);
3084 print_summary(&source_line, dso->long_name);
3085 }
3086
3087 symbol__annotate_printf(ms, evsel);
3088
3089 annotated_source__purge(symbol__annotation(sym)->src);
3090
3091 return 0;
3092}
3093
3094bool ui__has_annotation(void)
3095{
3096 return use_browser == 1 && perf_hpp_list.sym;
3097}
3098
3099
3100static double annotation_line__max_percent(struct annotation_line *al,
3101 struct annotation *notes,
3102 unsigned int percent_type)
3103{
3104 double percent_max = 0.0;
3105 int i;
3106
3107 for (i = 0; i < notes->nr_events; i++) {
3108 double percent;
3109
3110 percent = annotation_data__percent(&al->data[i],
3111 percent_type);
3112
3113 if (percent > percent_max)
3114 percent_max = percent;
3115 }
3116
3117 return percent_max;
3118}
3119
3120static void disasm_line__write(struct disasm_line *dl, struct annotation *notes,
3121 void *obj, char *bf, size_t size,
3122 void (*obj__printf)(void *obj, const char *fmt, ...),
3123 void (*obj__write_graph)(void *obj, int graph))
3124{
3125 if (dl->ins.ops && dl->ins.ops->scnprintf) {
3126 if (ins__is_jump(&dl->ins)) {
3127 bool fwd;
3128
3129 if (dl->ops.target.outside)
3130 goto call_like;
3131 fwd = dl->ops.target.offset > dl->al.offset;
3132 obj__write_graph(obj, fwd ? DARROW_CHAR : UARROW_CHAR);
3133 obj__printf(obj, " ");
3134 } else if (ins__is_call(&dl->ins)) {
3135call_like:
3136 obj__write_graph(obj, RARROW_CHAR);
3137 obj__printf(obj, " ");
3138 } else if (ins__is_ret(&dl->ins)) {
3139 obj__write_graph(obj, LARROW_CHAR);
3140 obj__printf(obj, " ");
3141 } else {
3142 obj__printf(obj, " ");
3143 }
3144 } else {
3145 obj__printf(obj, " ");
3146 }
3147
3148 disasm_line__scnprintf(dl, bf, size, !annotate_opts.use_offset, notes->widths.max_ins_name);
3149}
3150
3151static void ipc_coverage_string(char *bf, int size, struct annotation *notes)
3152{
3153 double ipc = 0.0, coverage = 0.0;
3154 struct annotated_branch *branch = annotation__get_branch(notes);
3155
3156 if (branch && branch->hit_cycles)
3157 ipc = branch->hit_insn / ((double)branch->hit_cycles);
3158
3159 if (branch && branch->total_insn) {
3160 coverage = branch->cover_insn * 100.0 /
3161 ((double)branch->total_insn);
3162 }
3163
3164 scnprintf(bf, size, "(Average IPC: %.2f, IPC Coverage: %.1f%%)",
3165 ipc, coverage);
3166}
3167
3168static void __annotation_line__write(struct annotation_line *al, struct annotation *notes,
3169 bool first_line, bool current_entry, bool change_color, int width,
3170 void *obj, unsigned int percent_type,
3171 int (*obj__set_color)(void *obj, int color),
3172 void (*obj__set_percent_color)(void *obj, double percent, bool current),
3173 int (*obj__set_jumps_percent_color)(void *obj, int nr, bool current),
3174 void (*obj__printf)(void *obj, const char *fmt, ...),
3175 void (*obj__write_graph)(void *obj, int graph))
3176
3177{
3178 double percent_max = annotation_line__max_percent(al, notes, percent_type);
3179 int pcnt_width = annotation__pcnt_width(notes),
3180 cycles_width = annotation__cycles_width(notes);
3181 bool show_title = false;
3182 char bf[256];
3183 int printed;
3184
3185 if (first_line && (al->offset == -1 || percent_max == 0.0)) {
3186 if (notes->branch && al->cycles) {
3187 if (al->cycles->ipc == 0.0 && al->cycles->avg == 0)
3188 show_title = true;
3189 } else
3190 show_title = true;
3191 }
3192
3193 if (al->offset != -1 && percent_max != 0.0) {
3194 int i;
3195
3196 for (i = 0; i < notes->nr_events; i++) {
3197 double percent;
3198
3199 percent = annotation_data__percent(&al->data[i], percent_type);
3200
3201 obj__set_percent_color(obj, percent, current_entry);
3202 if (symbol_conf.show_total_period) {
3203 obj__printf(obj, "%11" PRIu64 " ", al->data[i].he.period);
3204 } else if (symbol_conf.show_nr_samples) {
3205 obj__printf(obj, "%6" PRIu64 " ",
3206 al->data[i].he.nr_samples);
3207 } else {
3208 obj__printf(obj, "%6.2f ", percent);
3209 }
3210 }
3211 } else {
3212 obj__set_percent_color(obj, 0, current_entry);
3213
3214 if (!show_title)
3215 obj__printf(obj, "%-*s", pcnt_width, " ");
3216 else {
3217 obj__printf(obj, "%-*s", pcnt_width,
3218 symbol_conf.show_total_period ? "Period" :
3219 symbol_conf.show_nr_samples ? "Samples" : "Percent");
3220 }
3221 }
3222
3223 if (notes->branch) {
3224 if (al->cycles && al->cycles->ipc)
3225 obj__printf(obj, "%*.2f ", ANNOTATION__IPC_WIDTH - 1, al->cycles->ipc);
3226 else if (!show_title)
3227 obj__printf(obj, "%*s", ANNOTATION__IPC_WIDTH, " ");
3228 else
3229 obj__printf(obj, "%*s ", ANNOTATION__IPC_WIDTH - 1, "IPC");
3230
3231 if (!annotate_opts.show_minmax_cycle) {
3232 if (al->cycles && al->cycles->avg)
3233 obj__printf(obj, "%*" PRIu64 " ",
3234 ANNOTATION__CYCLES_WIDTH - 1, al->cycles->avg);
3235 else if (!show_title)
3236 obj__printf(obj, "%*s",
3237 ANNOTATION__CYCLES_WIDTH, " ");
3238 else
3239 obj__printf(obj, "%*s ",
3240 ANNOTATION__CYCLES_WIDTH - 1,
3241 "Cycle");
3242 } else {
3243 if (al->cycles) {
3244 char str[32];
3245
3246 scnprintf(str, sizeof(str),
3247 "%" PRIu64 "(%" PRIu64 "/%" PRIu64 ")",
3248 al->cycles->avg, al->cycles->min,
3249 al->cycles->max);
3250
3251 obj__printf(obj, "%*s ",
3252 ANNOTATION__MINMAX_CYCLES_WIDTH - 1,
3253 str);
3254 } else if (!show_title)
3255 obj__printf(obj, "%*s",
3256 ANNOTATION__MINMAX_CYCLES_WIDTH,
3257 " ");
3258 else
3259 obj__printf(obj, "%*s ",
3260 ANNOTATION__MINMAX_CYCLES_WIDTH - 1,
3261 "Cycle(min/max)");
3262 }
3263
3264 if (show_title && !*al->line) {
3265 ipc_coverage_string(bf, sizeof(bf), notes);
3266 obj__printf(obj, "%*s", ANNOTATION__AVG_IPC_WIDTH, bf);
3267 }
3268 }
3269
3270 obj__printf(obj, " ");
3271
3272 if (!*al->line)
3273 obj__printf(obj, "%-*s", width - pcnt_width - cycles_width, " ");
3274 else if (al->offset == -1) {
3275 if (al->line_nr && annotate_opts.show_linenr)
3276 printed = scnprintf(bf, sizeof(bf), "%-*d ", notes->widths.addr + 1, al->line_nr);
3277 else
3278 printed = scnprintf(bf, sizeof(bf), "%-*s ", notes->widths.addr, " ");
3279 obj__printf(obj, bf);
3280 obj__printf(obj, "%-*s", width - printed - pcnt_width - cycles_width + 1, al->line);
3281 } else {
3282 u64 addr = al->offset;
3283 int color = -1;
3284
3285 if (!annotate_opts.use_offset)
3286 addr += notes->start;
3287
3288 if (!annotate_opts.use_offset) {
3289 printed = scnprintf(bf, sizeof(bf), "%" PRIx64 ": ", addr);
3290 } else {
3291 if (al->jump_sources &&
3292 annotate_opts.offset_level >= ANNOTATION__OFFSET_JUMP_TARGETS) {
3293 if (annotate_opts.show_nr_jumps) {
3294 int prev;
3295 printed = scnprintf(bf, sizeof(bf), "%*d ",
3296 notes->widths.jumps,
3297 al->jump_sources);
3298 prev = obj__set_jumps_percent_color(obj, al->jump_sources,
3299 current_entry);
3300 obj__printf(obj, bf);
3301 obj__set_color(obj, prev);
3302 }
3303print_addr:
3304 printed = scnprintf(bf, sizeof(bf), "%*" PRIx64 ": ",
3305 notes->widths.target, addr);
3306 } else if (ins__is_call(&disasm_line(al)->ins) &&
3307 annotate_opts.offset_level >= ANNOTATION__OFFSET_CALL) {
3308 goto print_addr;
3309 } else if (annotate_opts.offset_level == ANNOTATION__MAX_OFFSET_LEVEL) {
3310 goto print_addr;
3311 } else {
3312 printed = scnprintf(bf, sizeof(bf), "%-*s ",
3313 notes->widths.addr, " ");
3314 }
3315 }
3316
3317 if (change_color)
3318 color = obj__set_color(obj, HE_COLORSET_ADDR);
3319 obj__printf(obj, bf);
3320 if (change_color)
3321 obj__set_color(obj, color);
3322
3323 disasm_line__write(disasm_line(al), notes, obj, bf, sizeof(bf), obj__printf, obj__write_graph);
3324
3325 obj__printf(obj, "%-*s", width - pcnt_width - cycles_width - 3 - printed, bf);
3326 }
3327
3328}
3329
3330void annotation_line__write(struct annotation_line *al, struct annotation *notes,
3331 struct annotation_write_ops *wops)
3332{
3333 __annotation_line__write(al, notes, wops->first_line, wops->current_entry,
3334 wops->change_color, wops->width, wops->obj,
3335 annotate_opts.percent_type,
3336 wops->set_color, wops->set_percent_color,
3337 wops->set_jumps_percent_color, wops->printf,
3338 wops->write_graph);
3339}
3340
3341int symbol__annotate2(struct map_symbol *ms, struct evsel *evsel,
3342 struct arch **parch)
3343{
3344 struct symbol *sym = ms->sym;
3345 struct annotation *notes = symbol__annotation(sym);
3346 size_t size = symbol__size(sym);
3347 int nr_pcnt = 1, err;
3348
3349 notes->src->offsets = zalloc(size * sizeof(struct annotation_line *));
3350 if (notes->src->offsets == NULL)
3351 return ENOMEM;
3352
3353 if (evsel__is_group_event(evsel))
3354 nr_pcnt = evsel->core.nr_members;
3355
3356 err = symbol__annotate(ms, evsel, parch);
3357 if (err)
3358 goto out_free_offsets;
3359
3360 symbol__calc_percent(sym, evsel);
3361
3362 annotation__set_offsets(notes, size);
3363 annotation__mark_jump_targets(notes, sym);
3364
3365 err = annotation__compute_ipc(notes, size);
3366 if (err)
3367 goto out_free_offsets;
3368
3369 annotation__init_column_widths(notes, sym);
3370 notes->nr_events = nr_pcnt;
3371
3372 annotation__update_column_widths(notes);
3373 sym->annotate2 = 1;
3374
3375 return 0;
3376
3377out_free_offsets:
3378 zfree(¬es->src->offsets);
3379 return err;
3380}
3381
3382static int annotation__config(const char *var, const char *value, void *data)
3383{
3384 struct annotation_options *opt = data;
3385
3386 if (!strstarts(var, "annotate."))
3387 return 0;
3388
3389 if (!strcmp(var, "annotate.offset_level")) {
3390 perf_config_u8(&opt->offset_level, "offset_level", value);
3391
3392 if (opt->offset_level > ANNOTATION__MAX_OFFSET_LEVEL)
3393 opt->offset_level = ANNOTATION__MAX_OFFSET_LEVEL;
3394 else if (opt->offset_level < ANNOTATION__MIN_OFFSET_LEVEL)
3395 opt->offset_level = ANNOTATION__MIN_OFFSET_LEVEL;
3396 } else if (!strcmp(var, "annotate.hide_src_code")) {
3397 opt->hide_src_code = perf_config_bool("hide_src_code", value);
3398 } else if (!strcmp(var, "annotate.jump_arrows")) {
3399 opt->jump_arrows = perf_config_bool("jump_arrows", value);
3400 } else if (!strcmp(var, "annotate.show_linenr")) {
3401 opt->show_linenr = perf_config_bool("show_linenr", value);
3402 } else if (!strcmp(var, "annotate.show_nr_jumps")) {
3403 opt->show_nr_jumps = perf_config_bool("show_nr_jumps", value);
3404 } else if (!strcmp(var, "annotate.show_nr_samples")) {
3405 symbol_conf.show_nr_samples = perf_config_bool("show_nr_samples",
3406 value);
3407 } else if (!strcmp(var, "annotate.show_total_period")) {
3408 symbol_conf.show_total_period = perf_config_bool("show_total_period",
3409 value);
3410 } else if (!strcmp(var, "annotate.use_offset")) {
3411 opt->use_offset = perf_config_bool("use_offset", value);
3412 } else if (!strcmp(var, "annotate.disassembler_style")) {
3413 opt->disassembler_style = strdup(value);
3414 if (!opt->disassembler_style) {
3415 pr_err("Not enough memory for annotate.disassembler_style\n");
3416 return -1;
3417 }
3418 } else if (!strcmp(var, "annotate.objdump")) {
3419 opt->objdump_path = strdup(value);
3420 if (!opt->objdump_path) {
3421 pr_err("Not enough memory for annotate.objdump\n");
3422 return -1;
3423 }
3424 } else if (!strcmp(var, "annotate.addr2line")) {
3425 symbol_conf.addr2line_path = strdup(value);
3426 if (!symbol_conf.addr2line_path) {
3427 pr_err("Not enough memory for annotate.addr2line\n");
3428 return -1;
3429 }
3430 } else if (!strcmp(var, "annotate.demangle")) {
3431 symbol_conf.demangle = perf_config_bool("demangle", value);
3432 } else if (!strcmp(var, "annotate.demangle_kernel")) {
3433 symbol_conf.demangle_kernel = perf_config_bool("demangle_kernel", value);
3434 } else {
3435 pr_debug("%s variable unknown, ignoring...", var);
3436 }
3437
3438 return 0;
3439}
3440
3441void annotation_options__init(void)
3442{
3443 struct annotation_options *opt = &annotate_opts;
3444
3445 memset(opt, 0, sizeof(*opt));
3446
3447 /* Default values. */
3448 opt->use_offset = true;
3449 opt->jump_arrows = true;
3450 opt->annotate_src = true;
3451 opt->offset_level = ANNOTATION__OFFSET_JUMP_TARGETS;
3452 opt->percent_type = PERCENT_PERIOD_LOCAL;
3453}
3454
3455void annotation_options__exit(void)
3456{
3457 zfree(&annotate_opts.disassembler_style);
3458 zfree(&annotate_opts.objdump_path);
3459}
3460
3461void annotation_config__init(void)
3462{
3463 perf_config(annotation__config, &annotate_opts);
3464}
3465
3466static unsigned int parse_percent_type(char *str1, char *str2)
3467{
3468 unsigned int type = (unsigned int) -1;
3469
3470 if (!strcmp("period", str1)) {
3471 if (!strcmp("local", str2))
3472 type = PERCENT_PERIOD_LOCAL;
3473 else if (!strcmp("global", str2))
3474 type = PERCENT_PERIOD_GLOBAL;
3475 }
3476
3477 if (!strcmp("hits", str1)) {
3478 if (!strcmp("local", str2))
3479 type = PERCENT_HITS_LOCAL;
3480 else if (!strcmp("global", str2))
3481 type = PERCENT_HITS_GLOBAL;
3482 }
3483
3484 return type;
3485}
3486
3487int annotate_parse_percent_type(const struct option *opt __maybe_unused, const char *_str,
3488 int unset __maybe_unused)
3489{
3490 unsigned int type;
3491 char *str1, *str2;
3492 int err = -1;
3493
3494 str1 = strdup(_str);
3495 if (!str1)
3496 return -ENOMEM;
3497
3498 str2 = strchr(str1, '-');
3499 if (!str2)
3500 goto out;
3501
3502 *str2++ = 0;
3503
3504 type = parse_percent_type(str1, str2);
3505 if (type == (unsigned int) -1)
3506 type = parse_percent_type(str2, str1);
3507 if (type != (unsigned int) -1) {
3508 annotate_opts.percent_type = type;
3509 err = 0;
3510 }
3511
3512out:
3513 free(str1);
3514 return err;
3515}
3516
3517int annotate_check_args(void)
3518{
3519 struct annotation_options *args = &annotate_opts;
3520
3521 if (args->prefix_strip && !args->prefix) {
3522 pr_err("--prefix-strip requires --prefix\n");
3523 return -1;
3524 }
3525 return 0;
3526}
3527
3528/*
3529 * Get register number and access offset from the given instruction.
3530 * It assumes AT&T x86 asm format like OFFSET(REG). Maybe it needs
3531 * to revisit the format when it handles different architecture.
3532 * Fills @reg and @offset when return 0.
3533 */
3534static int extract_reg_offset(struct arch *arch, const char *str,
3535 struct annotated_op_loc *op_loc)
3536{
3537 char *p;
3538 char *regname;
3539
3540 if (arch->objdump.register_char == 0)
3541 return -1;
3542
3543 /*
3544 * It should start from offset, but it's possible to skip 0
3545 * in the asm. So 0(%rax) should be same as (%rax).
3546 *
3547 * However, it also start with a segment select register like
3548 * %gs:0x18(%rbx). In that case it should skip the part.
3549 */
3550 if (*str == arch->objdump.register_char) {
3551 while (*str && !isdigit(*str) &&
3552 *str != arch->objdump.memory_ref_char)
3553 str++;
3554 }
3555
3556 op_loc->offset = strtol(str, &p, 0);
3557
3558 p = strchr(p, arch->objdump.register_char);
3559 if (p == NULL)
3560 return -1;
3561
3562 regname = strdup(p);
3563 if (regname == NULL)
3564 return -1;
3565
3566 op_loc->reg = get_dwarf_regnum(regname, 0);
3567 free(regname);
3568 return 0;
3569}
3570
3571/**
3572 * annotate_get_insn_location - Get location of instruction
3573 * @arch: the architecture info
3574 * @dl: the target instruction
3575 * @loc: a buffer to save the data
3576 *
3577 * Get detailed location info (register and offset) in the instruction.
3578 * It needs both source and target operand and whether it accesses a
3579 * memory location. The offset field is meaningful only when the
3580 * corresponding mem flag is set.
3581 *
3582 * Some examples on x86:
3583 *
3584 * mov (%rax), %rcx # src_reg = rax, src_mem = 1, src_offset = 0
3585 * # dst_reg = rcx, dst_mem = 0
3586 *
3587 * mov 0x18, %r8 # src_reg = -1, dst_reg = r8
3588 */
3589int annotate_get_insn_location(struct arch *arch, struct disasm_line *dl,
3590 struct annotated_insn_loc *loc)
3591{
3592 struct ins_operands *ops;
3593 struct annotated_op_loc *op_loc;
3594 int i;
3595
3596 if (!strcmp(dl->ins.name, "lock"))
3597 ops = dl->ops.locked.ops;
3598 else
3599 ops = &dl->ops;
3600
3601 if (ops == NULL)
3602 return -1;
3603
3604 memset(loc, 0, sizeof(*loc));
3605
3606 for_each_insn_op_loc(loc, i, op_loc) {
3607 const char *insn_str = ops->source.raw;
3608
3609 if (i == INSN_OP_TARGET)
3610 insn_str = ops->target.raw;
3611
3612 /* Invalidate the register by default */
3613 op_loc->reg = -1;
3614
3615 if (insn_str == NULL)
3616 continue;
3617
3618 if (strchr(insn_str, arch->objdump.memory_ref_char)) {
3619 op_loc->mem_ref = true;
3620 extract_reg_offset(arch, insn_str, op_loc);
3621 } else {
3622 char *s = strdup(insn_str);
3623
3624 if (s) {
3625 op_loc->reg = get_dwarf_regnum(s, 0);
3626 free(s);
3627 }
3628 }
3629 }
3630
3631 return 0;
3632}
3633
3634static void symbol__ensure_annotate(struct map_symbol *ms, struct evsel *evsel)
3635{
3636 struct disasm_line *dl, *tmp_dl;
3637 struct annotation *notes;
3638
3639 notes = symbol__annotation(ms->sym);
3640 if (!list_empty(¬es->src->source))
3641 return;
3642
3643 if (symbol__annotate(ms, evsel, NULL) < 0)
3644 return;
3645
3646 /* remove non-insn disasm lines for simplicity */
3647 list_for_each_entry_safe(dl, tmp_dl, ¬es->src->source, al.node) {
3648 if (dl->al.offset == -1) {
3649 list_del(&dl->al.node);
3650 free(dl);
3651 }
3652 }
3653}
3654
3655static struct disasm_line *find_disasm_line(struct symbol *sym, u64 ip)
3656{
3657 struct disasm_line *dl;
3658 struct annotation *notes;
3659
3660 notes = symbol__annotation(sym);
3661
3662 list_for_each_entry(dl, ¬es->src->source, al.node) {
3663 if (sym->start + dl->al.offset == ip)
3664 return dl;
3665 }
3666 return NULL;
3667}
3668
3669static struct annotated_item_stat *annotate_data_stat(struct list_head *head,
3670 const char *name)
3671{
3672 struct annotated_item_stat *istat;
3673
3674 list_for_each_entry(istat, head, list) {
3675 if (!strcmp(istat->name, name))
3676 return istat;
3677 }
3678
3679 istat = zalloc(sizeof(*istat));
3680 if (istat == NULL)
3681 return NULL;
3682
3683 istat->name = strdup(name);
3684 if (istat->name == NULL) {
3685 free(istat);
3686 return NULL;
3687 }
3688
3689 list_add_tail(&istat->list, head);
3690 return istat;
3691}
3692
3693/**
3694 * hist_entry__get_data_type - find data type for given hist entry
3695 * @he: hist entry
3696 *
3697 * This function first annotates the instruction at @he->ip and extracts
3698 * register and offset info from it. Then it searches the DWARF debug
3699 * info to get a variable and type information using the address, register,
3700 * and offset.
3701 */
3702struct annotated_data_type *hist_entry__get_data_type(struct hist_entry *he)
3703{
3704 struct map_symbol *ms = &he->ms;
3705 struct evsel *evsel = hists_to_evsel(he->hists);
3706 struct arch *arch;
3707 struct disasm_line *dl;
3708 struct annotated_insn_loc loc;
3709 struct annotated_op_loc *op_loc;
3710 struct annotated_data_type *mem_type;
3711 struct annotated_item_stat *istat;
3712 u64 ip = he->ip;
3713 int i;
3714
3715 ann_data_stat.total++;
3716
3717 if (ms->map == NULL || ms->sym == NULL) {
3718 ann_data_stat.no_sym++;
3719 return NULL;
3720 }
3721
3722 if (!symbol_conf.init_annotation) {
3723 ann_data_stat.no_sym++;
3724 return NULL;
3725 }
3726
3727 if (evsel__get_arch(evsel, &arch) < 0) {
3728 ann_data_stat.no_insn++;
3729 return NULL;
3730 }
3731
3732 /* Make sure it runs objdump to get disasm of the function */
3733 symbol__ensure_annotate(ms, evsel);
3734
3735 /*
3736 * Get a disasm to extract the location from the insn.
3737 * This is too slow...
3738 */
3739 dl = find_disasm_line(ms->sym, ip);
3740 if (dl == NULL) {
3741 ann_data_stat.no_insn++;
3742 return NULL;
3743 }
3744
3745 istat = annotate_data_stat(&ann_insn_stat, dl->ins.name);
3746 if (istat == NULL) {
3747 ann_data_stat.no_insn++;
3748 return NULL;
3749 }
3750
3751 if (annotate_get_insn_location(arch, dl, &loc) < 0) {
3752 ann_data_stat.no_insn_ops++;
3753 istat->bad++;
3754 return NULL;
3755 }
3756
3757 for_each_insn_op_loc(&loc, i, op_loc) {
3758 if (!op_loc->mem_ref)
3759 continue;
3760
3761 mem_type = find_data_type(ms, ip, op_loc->reg, op_loc->offset);
3762 if (mem_type)
3763 istat->good++;
3764 else
3765 istat->bad++;
3766
3767 if (symbol_conf.annotate_data_sample) {
3768 annotated_data_type__update_samples(mem_type, evsel,
3769 op_loc->offset,
3770 he->stat.nr_events,
3771 he->stat.period);
3772 }
3773 he->mem_type_off = op_loc->offset;
3774 return mem_type;
3775 }
3776
3777 ann_data_stat.no_mem_ops++;
3778 istat->bad++;
3779 return NULL;
3780}
1// SPDX-License-Identifier: GPL-2.0-only
2/*
3 * Copyright (C) 2011, Red Hat Inc, Arnaldo Carvalho de Melo <acme@redhat.com>
4 *
5 * Parts came from builtin-annotate.c, see those files for further
6 * copyright notes.
7 */
8
9#include <errno.h>
10#include <inttypes.h>
11#include <libgen.h>
12#include <stdlib.h>
13#include "util.h" // hex_width()
14#include "ui/ui.h"
15#include "sort.h"
16#include "build-id.h"
17#include "color.h"
18#include "config.h"
19#include "dso.h"
20#include "env.h"
21#include "map.h"
22#include "maps.h"
23#include "symbol.h"
24#include "srcline.h"
25#include "units.h"
26#include "debug.h"
27#include "annotate.h"
28#include "annotate-data.h"
29#include "evsel.h"
30#include "evlist.h"
31#include "bpf-event.h"
32#include "bpf-utils.h"
33#include "block-range.h"
34#include "string2.h"
35#include "dwarf-regs.h"
36#include "util/event.h"
37#include "util/sharded_mutex.h"
38#include "arch/common.h"
39#include "namespaces.h"
40#include "thread.h"
41#include "hashmap.h"
42#include <regex.h>
43#include <linux/bitops.h>
44#include <linux/kernel.h>
45#include <linux/string.h>
46#include <linux/zalloc.h>
47#include <subcmd/parse-options.h>
48#include <subcmd/run-command.h>
49
50/* FIXME: For the HE_COLORSET */
51#include "ui/browser.h"
52
53/*
54 * FIXME: Using the same values as slang.h,
55 * but that header may not be available everywhere
56 */
57#define LARROW_CHAR ((unsigned char)',')
58#define RARROW_CHAR ((unsigned char)'+')
59#define DARROW_CHAR ((unsigned char)'.')
60#define UARROW_CHAR ((unsigned char)'-')
61
62#include <linux/ctype.h>
63
64/* global annotation options */
65struct annotation_options annotate_opts;
66
67static regex_t file_lineno;
68
69static struct ins_ops *ins__find(struct arch *arch, const char *name);
70static void ins__sort(struct arch *arch);
71static int disasm_line__parse(char *line, const char **namep, char **rawp);
72static int call__scnprintf(struct ins *ins, char *bf, size_t size,
73 struct ins_operands *ops, int max_ins_name);
74static int jump__scnprintf(struct ins *ins, char *bf, size_t size,
75 struct ins_operands *ops, int max_ins_name);
76
77struct arch {
78 const char *name;
79 struct ins *instructions;
80 size_t nr_instructions;
81 size_t nr_instructions_allocated;
82 struct ins_ops *(*associate_instruction_ops)(struct arch *arch, const char *name);
83 bool sorted_instructions;
84 bool initialized;
85 const char *insn_suffix;
86 void *priv;
87 unsigned int model;
88 unsigned int family;
89 int (*init)(struct arch *arch, char *cpuid);
90 bool (*ins_is_fused)(struct arch *arch, const char *ins1,
91 const char *ins2);
92 struct {
93 char comment_char;
94 char skip_functions_char;
95 char register_char;
96 char memory_ref_char;
97 } objdump;
98};
99
100static struct ins_ops call_ops;
101static struct ins_ops dec_ops;
102static struct ins_ops jump_ops;
103static struct ins_ops mov_ops;
104static struct ins_ops nop_ops;
105static struct ins_ops lock_ops;
106static struct ins_ops ret_ops;
107
108/* Data type collection debug statistics */
109struct annotated_data_stat ann_data_stat;
110LIST_HEAD(ann_insn_stat);
111
112/* Pseudo data types */
113struct annotated_data_type stackop_type = {
114 .self = {
115 .type_name = (char *)"(stack operation)",
116 .children = LIST_HEAD_INIT(stackop_type.self.children),
117 },
118};
119
120static int arch__grow_instructions(struct arch *arch)
121{
122 struct ins *new_instructions;
123 size_t new_nr_allocated;
124
125 if (arch->nr_instructions_allocated == 0 && arch->instructions)
126 goto grow_from_non_allocated_table;
127
128 new_nr_allocated = arch->nr_instructions_allocated + 128;
129 new_instructions = realloc(arch->instructions, new_nr_allocated * sizeof(struct ins));
130 if (new_instructions == NULL)
131 return -1;
132
133out_update_instructions:
134 arch->instructions = new_instructions;
135 arch->nr_instructions_allocated = new_nr_allocated;
136 return 0;
137
138grow_from_non_allocated_table:
139 new_nr_allocated = arch->nr_instructions + 128;
140 new_instructions = calloc(new_nr_allocated, sizeof(struct ins));
141 if (new_instructions == NULL)
142 return -1;
143
144 memcpy(new_instructions, arch->instructions, arch->nr_instructions);
145 goto out_update_instructions;
146}
147
148static int arch__associate_ins_ops(struct arch* arch, const char *name, struct ins_ops *ops)
149{
150 struct ins *ins;
151
152 if (arch->nr_instructions == arch->nr_instructions_allocated &&
153 arch__grow_instructions(arch))
154 return -1;
155
156 ins = &arch->instructions[arch->nr_instructions];
157 ins->name = strdup(name);
158 if (!ins->name)
159 return -1;
160
161 ins->ops = ops;
162 arch->nr_instructions++;
163
164 ins__sort(arch);
165 return 0;
166}
167
168#include "arch/arc/annotate/instructions.c"
169#include "arch/arm/annotate/instructions.c"
170#include "arch/arm64/annotate/instructions.c"
171#include "arch/csky/annotate/instructions.c"
172#include "arch/loongarch/annotate/instructions.c"
173#include "arch/mips/annotate/instructions.c"
174#include "arch/x86/annotate/instructions.c"
175#include "arch/powerpc/annotate/instructions.c"
176#include "arch/riscv64/annotate/instructions.c"
177#include "arch/s390/annotate/instructions.c"
178#include "arch/sparc/annotate/instructions.c"
179
180static struct arch architectures[] = {
181 {
182 .name = "arc",
183 .init = arc__annotate_init,
184 },
185 {
186 .name = "arm",
187 .init = arm__annotate_init,
188 },
189 {
190 .name = "arm64",
191 .init = arm64__annotate_init,
192 },
193 {
194 .name = "csky",
195 .init = csky__annotate_init,
196 },
197 {
198 .name = "mips",
199 .init = mips__annotate_init,
200 .objdump = {
201 .comment_char = '#',
202 },
203 },
204 {
205 .name = "x86",
206 .init = x86__annotate_init,
207 .instructions = x86__instructions,
208 .nr_instructions = ARRAY_SIZE(x86__instructions),
209 .insn_suffix = "bwlq",
210 .objdump = {
211 .comment_char = '#',
212 .register_char = '%',
213 .memory_ref_char = '(',
214 },
215 },
216 {
217 .name = "powerpc",
218 .init = powerpc__annotate_init,
219 },
220 {
221 .name = "riscv64",
222 .init = riscv64__annotate_init,
223 },
224 {
225 .name = "s390",
226 .init = s390__annotate_init,
227 .objdump = {
228 .comment_char = '#',
229 },
230 },
231 {
232 .name = "sparc",
233 .init = sparc__annotate_init,
234 .objdump = {
235 .comment_char = '#',
236 },
237 },
238 {
239 .name = "loongarch",
240 .init = loongarch__annotate_init,
241 .objdump = {
242 .comment_char = '#',
243 },
244 },
245};
246
247static void ins__delete(struct ins_operands *ops)
248{
249 if (ops == NULL)
250 return;
251 zfree(&ops->source.raw);
252 zfree(&ops->source.name);
253 zfree(&ops->target.raw);
254 zfree(&ops->target.name);
255}
256
257static int ins__raw_scnprintf(struct ins *ins, char *bf, size_t size,
258 struct ins_operands *ops, int max_ins_name)
259{
260 return scnprintf(bf, size, "%-*s %s", max_ins_name, ins->name, ops->raw);
261}
262
263int ins__scnprintf(struct ins *ins, char *bf, size_t size,
264 struct ins_operands *ops, int max_ins_name)
265{
266 if (ins->ops->scnprintf)
267 return ins->ops->scnprintf(ins, bf, size, ops, max_ins_name);
268
269 return ins__raw_scnprintf(ins, bf, size, ops, max_ins_name);
270}
271
272bool ins__is_fused(struct arch *arch, const char *ins1, const char *ins2)
273{
274 if (!arch || !arch->ins_is_fused)
275 return false;
276
277 return arch->ins_is_fused(arch, ins1, ins2);
278}
279
280static int call__parse(struct arch *arch, struct ins_operands *ops, struct map_symbol *ms)
281{
282 char *endptr, *tok, *name;
283 struct map *map = ms->map;
284 struct addr_map_symbol target = {
285 .ms = { .map = map, },
286 };
287
288 ops->target.addr = strtoull(ops->raw, &endptr, 16);
289
290 name = strchr(endptr, '<');
291 if (name == NULL)
292 goto indirect_call;
293
294 name++;
295
296 if (arch->objdump.skip_functions_char &&
297 strchr(name, arch->objdump.skip_functions_char))
298 return -1;
299
300 tok = strchr(name, '>');
301 if (tok == NULL)
302 return -1;
303
304 *tok = '\0';
305 ops->target.name = strdup(name);
306 *tok = '>';
307
308 if (ops->target.name == NULL)
309 return -1;
310find_target:
311 target.addr = map__objdump_2mem(map, ops->target.addr);
312
313 if (maps__find_ams(ms->maps, &target) == 0 &&
314 map__rip_2objdump(target.ms.map, map__map_ip(target.ms.map, target.addr)) == ops->target.addr)
315 ops->target.sym = target.ms.sym;
316
317 return 0;
318
319indirect_call:
320 tok = strchr(endptr, '*');
321 if (tok != NULL) {
322 endptr++;
323
324 /* Indirect call can use a non-rip register and offset: callq *0x8(%rbx).
325 * Do not parse such instruction. */
326 if (strstr(endptr, "(%r") == NULL)
327 ops->target.addr = strtoull(endptr, NULL, 16);
328 }
329 goto find_target;
330}
331
332static int call__scnprintf(struct ins *ins, char *bf, size_t size,
333 struct ins_operands *ops, int max_ins_name)
334{
335 if (ops->target.sym)
336 return scnprintf(bf, size, "%-*s %s", max_ins_name, ins->name, ops->target.sym->name);
337
338 if (ops->target.addr == 0)
339 return ins__raw_scnprintf(ins, bf, size, ops, max_ins_name);
340
341 if (ops->target.name)
342 return scnprintf(bf, size, "%-*s %s", max_ins_name, ins->name, ops->target.name);
343
344 return scnprintf(bf, size, "%-*s *%" PRIx64, max_ins_name, ins->name, ops->target.addr);
345}
346
347static struct ins_ops call_ops = {
348 .parse = call__parse,
349 .scnprintf = call__scnprintf,
350};
351
352bool ins__is_call(const struct ins *ins)
353{
354 return ins->ops == &call_ops || ins->ops == &s390_call_ops || ins->ops == &loongarch_call_ops;
355}
356
357/*
358 * Prevents from matching commas in the comment section, e.g.:
359 * ffff200008446e70: b.cs ffff2000084470f4 <generic_exec_single+0x314> // b.hs, b.nlast
360 *
361 * and skip comma as part of function arguments, e.g.:
362 * 1d8b4ac <linemap_lookup(line_maps const*, unsigned int)+0xcc>
363 */
364static inline const char *validate_comma(const char *c, struct ins_operands *ops)
365{
366 if (ops->jump.raw_comment && c > ops->jump.raw_comment)
367 return NULL;
368
369 if (ops->jump.raw_func_start && c > ops->jump.raw_func_start)
370 return NULL;
371
372 return c;
373}
374
375static int jump__parse(struct arch *arch, struct ins_operands *ops, struct map_symbol *ms)
376{
377 struct map *map = ms->map;
378 struct symbol *sym = ms->sym;
379 struct addr_map_symbol target = {
380 .ms = { .map = map, },
381 };
382 const char *c = strchr(ops->raw, ',');
383 u64 start, end;
384
385 ops->jump.raw_comment = strchr(ops->raw, arch->objdump.comment_char);
386 ops->jump.raw_func_start = strchr(ops->raw, '<');
387
388 c = validate_comma(c, ops);
389
390 /*
391 * Examples of lines to parse for the _cpp_lex_token@@Base
392 * function:
393 *
394 * 1159e6c: jne 115aa32 <_cpp_lex_token@@Base+0xf92>
395 * 1159e8b: jne c469be <cpp_named_operator2name@@Base+0xa72>
396 *
397 * The first is a jump to an offset inside the same function,
398 * the second is to another function, i.e. that 0xa72 is an
399 * offset in the cpp_named_operator2name@@base function.
400 */
401 /*
402 * skip over possible up to 2 operands to get to address, e.g.:
403 * tbnz w0, #26, ffff0000083cd190 <security_file_permission+0xd0>
404 */
405 if (c++ != NULL) {
406 ops->target.addr = strtoull(c, NULL, 16);
407 if (!ops->target.addr) {
408 c = strchr(c, ',');
409 c = validate_comma(c, ops);
410 if (c++ != NULL)
411 ops->target.addr = strtoull(c, NULL, 16);
412 }
413 } else {
414 ops->target.addr = strtoull(ops->raw, NULL, 16);
415 }
416
417 target.addr = map__objdump_2mem(map, ops->target.addr);
418 start = map__unmap_ip(map, sym->start);
419 end = map__unmap_ip(map, sym->end);
420
421 ops->target.outside = target.addr < start || target.addr > end;
422
423 /*
424 * FIXME: things like this in _cpp_lex_token (gcc's cc1 program):
425
426 cpp_named_operator2name@@Base+0xa72
427
428 * Point to a place that is after the cpp_named_operator2name
429 * boundaries, i.e. in the ELF symbol table for cc1
430 * cpp_named_operator2name is marked as being 32-bytes long, but it in
431 * fact is much larger than that, so we seem to need a symbols__find()
432 * routine that looks for >= current->start and < next_symbol->start,
433 * possibly just for C++ objects?
434 *
435 * For now lets just make some progress by marking jumps to outside the
436 * current function as call like.
437 *
438 * Actual navigation will come next, with further understanding of how
439 * the symbol searching and disassembly should be done.
440 */
441 if (maps__find_ams(ms->maps, &target) == 0 &&
442 map__rip_2objdump(target.ms.map, map__map_ip(target.ms.map, target.addr)) == ops->target.addr)
443 ops->target.sym = target.ms.sym;
444
445 if (!ops->target.outside) {
446 ops->target.offset = target.addr - start;
447 ops->target.offset_avail = true;
448 } else {
449 ops->target.offset_avail = false;
450 }
451
452 return 0;
453}
454
455static int jump__scnprintf(struct ins *ins, char *bf, size_t size,
456 struct ins_operands *ops, int max_ins_name)
457{
458 const char *c;
459
460 if (!ops->target.addr || ops->target.offset < 0)
461 return ins__raw_scnprintf(ins, bf, size, ops, max_ins_name);
462
463 if (ops->target.outside && ops->target.sym != NULL)
464 return scnprintf(bf, size, "%-*s %s", max_ins_name, ins->name, ops->target.sym->name);
465
466 c = strchr(ops->raw, ',');
467 c = validate_comma(c, ops);
468
469 if (c != NULL) {
470 const char *c2 = strchr(c + 1, ',');
471
472 c2 = validate_comma(c2, ops);
473 /* check for 3-op insn */
474 if (c2 != NULL)
475 c = c2;
476 c++;
477
478 /* mirror arch objdump's space-after-comma style */
479 if (*c == ' ')
480 c++;
481 }
482
483 return scnprintf(bf, size, "%-*s %.*s%" PRIx64, max_ins_name,
484 ins->name, c ? c - ops->raw : 0, ops->raw,
485 ops->target.offset);
486}
487
488static void jump__delete(struct ins_operands *ops __maybe_unused)
489{
490 /*
491 * The ops->jump.raw_comment and ops->jump.raw_func_start belong to the
492 * raw string, don't free them.
493 */
494}
495
496static struct ins_ops jump_ops = {
497 .free = jump__delete,
498 .parse = jump__parse,
499 .scnprintf = jump__scnprintf,
500};
501
502bool ins__is_jump(const struct ins *ins)
503{
504 return ins->ops == &jump_ops || ins->ops == &loongarch_jump_ops;
505}
506
507static int comment__symbol(char *raw, char *comment, u64 *addrp, char **namep)
508{
509 char *endptr, *name, *t;
510
511 if (strstr(raw, "(%rip)") == NULL)
512 return 0;
513
514 *addrp = strtoull(comment, &endptr, 16);
515 if (endptr == comment)
516 return 0;
517 name = strchr(endptr, '<');
518 if (name == NULL)
519 return -1;
520
521 name++;
522
523 t = strchr(name, '>');
524 if (t == NULL)
525 return 0;
526
527 *t = '\0';
528 *namep = strdup(name);
529 *t = '>';
530
531 return 0;
532}
533
534static int lock__parse(struct arch *arch, struct ins_operands *ops, struct map_symbol *ms)
535{
536 ops->locked.ops = zalloc(sizeof(*ops->locked.ops));
537 if (ops->locked.ops == NULL)
538 return 0;
539
540 if (disasm_line__parse(ops->raw, &ops->locked.ins.name, &ops->locked.ops->raw) < 0)
541 goto out_free_ops;
542
543 ops->locked.ins.ops = ins__find(arch, ops->locked.ins.name);
544
545 if (ops->locked.ins.ops == NULL)
546 goto out_free_ops;
547
548 if (ops->locked.ins.ops->parse &&
549 ops->locked.ins.ops->parse(arch, ops->locked.ops, ms) < 0)
550 goto out_free_ops;
551
552 return 0;
553
554out_free_ops:
555 zfree(&ops->locked.ops);
556 return 0;
557}
558
559static int lock__scnprintf(struct ins *ins, char *bf, size_t size,
560 struct ins_operands *ops, int max_ins_name)
561{
562 int printed;
563
564 if (ops->locked.ins.ops == NULL)
565 return ins__raw_scnprintf(ins, bf, size, ops, max_ins_name);
566
567 printed = scnprintf(bf, size, "%-*s ", max_ins_name, ins->name);
568 return printed + ins__scnprintf(&ops->locked.ins, bf + printed,
569 size - printed, ops->locked.ops, max_ins_name);
570}
571
572static void lock__delete(struct ins_operands *ops)
573{
574 struct ins *ins = &ops->locked.ins;
575
576 if (ins->ops && ins->ops->free)
577 ins->ops->free(ops->locked.ops);
578 else
579 ins__delete(ops->locked.ops);
580
581 zfree(&ops->locked.ops);
582 zfree(&ops->target.raw);
583 zfree(&ops->target.name);
584}
585
586static struct ins_ops lock_ops = {
587 .free = lock__delete,
588 .parse = lock__parse,
589 .scnprintf = lock__scnprintf,
590};
591
592/*
593 * Check if the operand has more than one registers like x86 SIB addressing:
594 * 0x1234(%rax, %rbx, 8)
595 *
596 * But it doesn't care segment selectors like %gs:0x5678(%rcx), so just check
597 * the input string after 'memory_ref_char' if exists.
598 */
599static bool check_multi_regs(struct arch *arch, const char *op)
600{
601 int count = 0;
602
603 if (arch->objdump.register_char == 0)
604 return false;
605
606 if (arch->objdump.memory_ref_char) {
607 op = strchr(op, arch->objdump.memory_ref_char);
608 if (op == NULL)
609 return false;
610 }
611
612 while ((op = strchr(op, arch->objdump.register_char)) != NULL) {
613 count++;
614 op++;
615 }
616
617 return count > 1;
618}
619
620static int mov__parse(struct arch *arch, struct ins_operands *ops, struct map_symbol *ms __maybe_unused)
621{
622 char *s = strchr(ops->raw, ','), *target, *comment, prev;
623
624 if (s == NULL)
625 return -1;
626
627 *s = '\0';
628
629 /*
630 * x86 SIB addressing has something like 0x8(%rax, %rcx, 1)
631 * then it needs to have the closing parenthesis.
632 */
633 if (strchr(ops->raw, '(')) {
634 *s = ',';
635 s = strchr(ops->raw, ')');
636 if (s == NULL || s[1] != ',')
637 return -1;
638 *++s = '\0';
639 }
640
641 ops->source.raw = strdup(ops->raw);
642 *s = ',';
643
644 if (ops->source.raw == NULL)
645 return -1;
646
647 ops->source.multi_regs = check_multi_regs(arch, ops->source.raw);
648
649 target = skip_spaces(++s);
650 comment = strchr(s, arch->objdump.comment_char);
651
652 if (comment != NULL)
653 s = comment - 1;
654 else
655 s = strchr(s, '\0') - 1;
656
657 while (s > target && isspace(s[0]))
658 --s;
659 s++;
660 prev = *s;
661 *s = '\0';
662
663 ops->target.raw = strdup(target);
664 *s = prev;
665
666 if (ops->target.raw == NULL)
667 goto out_free_source;
668
669 ops->target.multi_regs = check_multi_regs(arch, ops->target.raw);
670
671 if (comment == NULL)
672 return 0;
673
674 comment = skip_spaces(comment);
675 comment__symbol(ops->source.raw, comment + 1, &ops->source.addr, &ops->source.name);
676 comment__symbol(ops->target.raw, comment + 1, &ops->target.addr, &ops->target.name);
677
678 return 0;
679
680out_free_source:
681 zfree(&ops->source.raw);
682 return -1;
683}
684
685static int mov__scnprintf(struct ins *ins, char *bf, size_t size,
686 struct ins_operands *ops, int max_ins_name)
687{
688 return scnprintf(bf, size, "%-*s %s,%s", max_ins_name, ins->name,
689 ops->source.name ?: ops->source.raw,
690 ops->target.name ?: ops->target.raw);
691}
692
693static struct ins_ops mov_ops = {
694 .parse = mov__parse,
695 .scnprintf = mov__scnprintf,
696};
697
698static int dec__parse(struct arch *arch __maybe_unused, struct ins_operands *ops, struct map_symbol *ms __maybe_unused)
699{
700 char *target, *comment, *s, prev;
701
702 target = s = ops->raw;
703
704 while (s[0] != '\0' && !isspace(s[0]))
705 ++s;
706 prev = *s;
707 *s = '\0';
708
709 ops->target.raw = strdup(target);
710 *s = prev;
711
712 if (ops->target.raw == NULL)
713 return -1;
714
715 comment = strchr(s, arch->objdump.comment_char);
716 if (comment == NULL)
717 return 0;
718
719 comment = skip_spaces(comment);
720 comment__symbol(ops->target.raw, comment + 1, &ops->target.addr, &ops->target.name);
721
722 return 0;
723}
724
725static int dec__scnprintf(struct ins *ins, char *bf, size_t size,
726 struct ins_operands *ops, int max_ins_name)
727{
728 return scnprintf(bf, size, "%-*s %s", max_ins_name, ins->name,
729 ops->target.name ?: ops->target.raw);
730}
731
732static struct ins_ops dec_ops = {
733 .parse = dec__parse,
734 .scnprintf = dec__scnprintf,
735};
736
737static int nop__scnprintf(struct ins *ins __maybe_unused, char *bf, size_t size,
738 struct ins_operands *ops __maybe_unused, int max_ins_name)
739{
740 return scnprintf(bf, size, "%-*s", max_ins_name, "nop");
741}
742
743static struct ins_ops nop_ops = {
744 .scnprintf = nop__scnprintf,
745};
746
747static struct ins_ops ret_ops = {
748 .scnprintf = ins__raw_scnprintf,
749};
750
751bool ins__is_ret(const struct ins *ins)
752{
753 return ins->ops == &ret_ops;
754}
755
756bool ins__is_lock(const struct ins *ins)
757{
758 return ins->ops == &lock_ops;
759}
760
761static int ins__key_cmp(const void *name, const void *insp)
762{
763 const struct ins *ins = insp;
764
765 return strcmp(name, ins->name);
766}
767
768static int ins__cmp(const void *a, const void *b)
769{
770 const struct ins *ia = a;
771 const struct ins *ib = b;
772
773 return strcmp(ia->name, ib->name);
774}
775
776static void ins__sort(struct arch *arch)
777{
778 const int nmemb = arch->nr_instructions;
779
780 qsort(arch->instructions, nmemb, sizeof(struct ins), ins__cmp);
781}
782
783static struct ins_ops *__ins__find(struct arch *arch, const char *name)
784{
785 struct ins *ins;
786 const int nmemb = arch->nr_instructions;
787
788 if (!arch->sorted_instructions) {
789 ins__sort(arch);
790 arch->sorted_instructions = true;
791 }
792
793 ins = bsearch(name, arch->instructions, nmemb, sizeof(struct ins), ins__key_cmp);
794 if (ins)
795 return ins->ops;
796
797 if (arch->insn_suffix) {
798 char tmp[32];
799 char suffix;
800 size_t len = strlen(name);
801
802 if (len == 0 || len >= sizeof(tmp))
803 return NULL;
804
805 suffix = name[len - 1];
806 if (strchr(arch->insn_suffix, suffix) == NULL)
807 return NULL;
808
809 strcpy(tmp, name);
810 tmp[len - 1] = '\0'; /* remove the suffix and check again */
811
812 ins = bsearch(tmp, arch->instructions, nmemb, sizeof(struct ins), ins__key_cmp);
813 }
814 return ins ? ins->ops : NULL;
815}
816
817static struct ins_ops *ins__find(struct arch *arch, const char *name)
818{
819 struct ins_ops *ops = __ins__find(arch, name);
820
821 if (!ops && arch->associate_instruction_ops)
822 ops = arch->associate_instruction_ops(arch, name);
823
824 return ops;
825}
826
827static int arch__key_cmp(const void *name, const void *archp)
828{
829 const struct arch *arch = archp;
830
831 return strcmp(name, arch->name);
832}
833
834static int arch__cmp(const void *a, const void *b)
835{
836 const struct arch *aa = a;
837 const struct arch *ab = b;
838
839 return strcmp(aa->name, ab->name);
840}
841
842static void arch__sort(void)
843{
844 const int nmemb = ARRAY_SIZE(architectures);
845
846 qsort(architectures, nmemb, sizeof(struct arch), arch__cmp);
847}
848
849static struct arch *arch__find(const char *name)
850{
851 const int nmemb = ARRAY_SIZE(architectures);
852 static bool sorted;
853
854 if (!sorted) {
855 arch__sort();
856 sorted = true;
857 }
858
859 return bsearch(name, architectures, nmemb, sizeof(struct arch), arch__key_cmp);
860}
861
862bool arch__is(struct arch *arch, const char *name)
863{
864 return !strcmp(arch->name, name);
865}
866
867/* symbol histogram: key = offset << 16 | evsel->core.idx */
868static size_t sym_hist_hash(long key, void *ctx __maybe_unused)
869{
870 return (key >> 16) + (key & 0xffff);
871}
872
873static bool sym_hist_equal(long key1, long key2, void *ctx __maybe_unused)
874{
875 return key1 == key2;
876}
877
878static struct annotated_source *annotated_source__new(void)
879{
880 struct annotated_source *src = zalloc(sizeof(*src));
881
882 if (src != NULL)
883 INIT_LIST_HEAD(&src->source);
884
885 return src;
886}
887
888static __maybe_unused void annotated_source__delete(struct annotated_source *src)
889{
890 struct hashmap_entry *cur;
891 size_t bkt;
892
893 if (src == NULL)
894 return;
895
896 if (src->samples) {
897 hashmap__for_each_entry(src->samples, cur, bkt)
898 zfree(&cur->pvalue);
899 hashmap__free(src->samples);
900 }
901 zfree(&src->histograms);
902 free(src);
903}
904
905static int annotated_source__alloc_histograms(struct annotated_source *src,
906 int nr_hists)
907{
908 src->nr_histograms = nr_hists;
909 src->histograms = calloc(nr_hists, sizeof(*src->histograms));
910
911 if (src->histograms == NULL)
912 return -1;
913
914 src->samples = hashmap__new(sym_hist_hash, sym_hist_equal, NULL);
915 if (src->samples == NULL)
916 zfree(&src->histograms);
917
918 return src->histograms ? 0 : -1;
919}
920
921void symbol__annotate_zero_histograms(struct symbol *sym)
922{
923 struct annotation *notes = symbol__annotation(sym);
924
925 annotation__lock(notes);
926 if (notes->src != NULL) {
927 memset(notes->src->histograms, 0,
928 notes->src->nr_histograms * sizeof(*notes->src->histograms));
929 hashmap__clear(notes->src->samples);
930 }
931 if (notes->branch && notes->branch->cycles_hist) {
932 memset(notes->branch->cycles_hist, 0,
933 symbol__size(sym) * sizeof(struct cyc_hist));
934 }
935 annotation__unlock(notes);
936}
937
938static int __symbol__account_cycles(struct cyc_hist *ch,
939 u64 start,
940 unsigned offset, unsigned cycles,
941 unsigned have_start)
942{
943 /*
944 * For now we can only account one basic block per
945 * final jump. But multiple could be overlapping.
946 * Always account the longest one. So when
947 * a shorter one has been already seen throw it away.
948 *
949 * We separately always account the full cycles.
950 */
951 ch[offset].num_aggr++;
952 ch[offset].cycles_aggr += cycles;
953
954 if (cycles > ch[offset].cycles_max)
955 ch[offset].cycles_max = cycles;
956
957 if (ch[offset].cycles_min) {
958 if (cycles && cycles < ch[offset].cycles_min)
959 ch[offset].cycles_min = cycles;
960 } else
961 ch[offset].cycles_min = cycles;
962
963 if (!have_start && ch[offset].have_start)
964 return 0;
965 if (ch[offset].num) {
966 if (have_start && (!ch[offset].have_start ||
967 ch[offset].start > start)) {
968 ch[offset].have_start = 0;
969 ch[offset].cycles = 0;
970 ch[offset].num = 0;
971 if (ch[offset].reset < 0xffff)
972 ch[offset].reset++;
973 } else if (have_start &&
974 ch[offset].start < start)
975 return 0;
976 }
977
978 if (ch[offset].num < NUM_SPARKS)
979 ch[offset].cycles_spark[ch[offset].num] = cycles;
980
981 ch[offset].have_start = have_start;
982 ch[offset].start = start;
983 ch[offset].cycles += cycles;
984 ch[offset].num++;
985 return 0;
986}
987
988static int __symbol__inc_addr_samples(struct map_symbol *ms,
989 struct annotated_source *src, int evidx, u64 addr,
990 struct perf_sample *sample)
991{
992 struct symbol *sym = ms->sym;
993 long hash_key;
994 u64 offset;
995 struct sym_hist *h;
996 struct sym_hist_entry *entry;
997
998 pr_debug3("%s: addr=%#" PRIx64 "\n", __func__, map__unmap_ip(ms->map, addr));
999
1000 if ((addr < sym->start || addr >= sym->end) &&
1001 (addr != sym->end || sym->start != sym->end)) {
1002 pr_debug("%s(%d): ERANGE! sym->name=%s, start=%#" PRIx64 ", addr=%#" PRIx64 ", end=%#" PRIx64 "\n",
1003 __func__, __LINE__, sym->name, sym->start, addr, sym->end);
1004 return -ERANGE;
1005 }
1006
1007 offset = addr - sym->start;
1008 h = annotated_source__histogram(src, evidx);
1009 if (h == NULL) {
1010 pr_debug("%s(%d): ENOMEM! sym->name=%s, start=%#" PRIx64 ", addr=%#" PRIx64 ", end=%#" PRIx64 ", func: %d\n",
1011 __func__, __LINE__, sym->name, sym->start, addr, sym->end, sym->type == STT_FUNC);
1012 return -ENOMEM;
1013 }
1014
1015 hash_key = offset << 16 | evidx;
1016 if (!hashmap__find(src->samples, hash_key, &entry)) {
1017 entry = zalloc(sizeof(*entry));
1018 if (entry == NULL)
1019 return -ENOMEM;
1020
1021 if (hashmap__add(src->samples, hash_key, entry) < 0)
1022 return -ENOMEM;
1023 }
1024
1025 h->nr_samples++;
1026 h->period += sample->period;
1027 entry->nr_samples++;
1028 entry->period += sample->period;
1029
1030 pr_debug3("%#" PRIx64 " %s: period++ [addr: %#" PRIx64 ", %#" PRIx64
1031 ", evidx=%d] => nr_samples: %" PRIu64 ", period: %" PRIu64 "\n",
1032 sym->start, sym->name, addr, addr - sym->start, evidx,
1033 entry->nr_samples, entry->period);
1034 return 0;
1035}
1036
1037struct annotated_branch *annotation__get_branch(struct annotation *notes)
1038{
1039 if (notes == NULL)
1040 return NULL;
1041
1042 if (notes->branch == NULL)
1043 notes->branch = zalloc(sizeof(*notes->branch));
1044
1045 return notes->branch;
1046}
1047
1048static struct cyc_hist *symbol__cycles_hist(struct symbol *sym)
1049{
1050 struct annotation *notes = symbol__annotation(sym);
1051 struct annotated_branch *branch;
1052
1053 branch = annotation__get_branch(notes);
1054 if (branch == NULL)
1055 return NULL;
1056
1057 if (branch->cycles_hist == NULL) {
1058 const size_t size = symbol__size(sym);
1059
1060 branch->cycles_hist = calloc(size, sizeof(struct cyc_hist));
1061 }
1062
1063 return branch->cycles_hist;
1064}
1065
1066struct annotated_source *symbol__hists(struct symbol *sym, int nr_hists)
1067{
1068 struct annotation *notes = symbol__annotation(sym);
1069
1070 if (notes->src == NULL) {
1071 notes->src = annotated_source__new();
1072 if (notes->src == NULL)
1073 return NULL;
1074 goto alloc_histograms;
1075 }
1076
1077 if (notes->src->histograms == NULL) {
1078alloc_histograms:
1079 annotated_source__alloc_histograms(notes->src, nr_hists);
1080 }
1081
1082 return notes->src;
1083}
1084
1085static int symbol__inc_addr_samples(struct map_symbol *ms,
1086 struct evsel *evsel, u64 addr,
1087 struct perf_sample *sample)
1088{
1089 struct symbol *sym = ms->sym;
1090 struct annotated_source *src;
1091
1092 if (sym == NULL)
1093 return 0;
1094 src = symbol__hists(sym, evsel->evlist->core.nr_entries);
1095 return src ? __symbol__inc_addr_samples(ms, src, evsel->core.idx, addr, sample) : 0;
1096}
1097
1098static int symbol__account_cycles(u64 addr, u64 start,
1099 struct symbol *sym, unsigned cycles)
1100{
1101 struct cyc_hist *cycles_hist;
1102 unsigned offset;
1103
1104 if (sym == NULL)
1105 return 0;
1106 cycles_hist = symbol__cycles_hist(sym);
1107 if (cycles_hist == NULL)
1108 return -ENOMEM;
1109 if (addr < sym->start || addr >= sym->end)
1110 return -ERANGE;
1111
1112 if (start) {
1113 if (start < sym->start || start >= sym->end)
1114 return -ERANGE;
1115 if (start >= addr)
1116 start = 0;
1117 }
1118 offset = addr - sym->start;
1119 return __symbol__account_cycles(cycles_hist,
1120 start ? start - sym->start : 0,
1121 offset, cycles,
1122 !!start);
1123}
1124
1125int addr_map_symbol__account_cycles(struct addr_map_symbol *ams,
1126 struct addr_map_symbol *start,
1127 unsigned cycles)
1128{
1129 u64 saddr = 0;
1130 int err;
1131
1132 if (!cycles)
1133 return 0;
1134
1135 /*
1136 * Only set start when IPC can be computed. We can only
1137 * compute it when the basic block is completely in a single
1138 * function.
1139 * Special case the case when the jump is elsewhere, but
1140 * it starts on the function start.
1141 */
1142 if (start &&
1143 (start->ms.sym == ams->ms.sym ||
1144 (ams->ms.sym &&
1145 start->addr == ams->ms.sym->start + map__start(ams->ms.map))))
1146 saddr = start->al_addr;
1147 if (saddr == 0)
1148 pr_debug2("BB with bad start: addr %"PRIx64" start %"PRIx64" sym %"PRIx64" saddr %"PRIx64"\n",
1149 ams->addr,
1150 start ? start->addr : 0,
1151 ams->ms.sym ? ams->ms.sym->start + map__start(ams->ms.map) : 0,
1152 saddr);
1153 err = symbol__account_cycles(ams->al_addr, saddr, ams->ms.sym, cycles);
1154 if (err)
1155 pr_debug2("account_cycles failed %d\n", err);
1156 return err;
1157}
1158
1159static unsigned annotation__count_insn(struct annotation *notes, u64 start, u64 end)
1160{
1161 unsigned n_insn = 0;
1162 u64 offset;
1163
1164 for (offset = start; offset <= end; offset++) {
1165 if (notes->src->offsets[offset])
1166 n_insn++;
1167 }
1168 return n_insn;
1169}
1170
1171static void annotated_branch__delete(struct annotated_branch *branch)
1172{
1173 if (branch) {
1174 zfree(&branch->cycles_hist);
1175 free(branch);
1176 }
1177}
1178
1179static void annotation__count_and_fill(struct annotation *notes, u64 start, u64 end, struct cyc_hist *ch)
1180{
1181 unsigned n_insn;
1182 unsigned int cover_insn = 0;
1183 u64 offset;
1184
1185 n_insn = annotation__count_insn(notes, start, end);
1186 if (n_insn && ch->num && ch->cycles) {
1187 struct annotated_branch *branch;
1188 float ipc = n_insn / ((double)ch->cycles / (double)ch->num);
1189
1190 /* Hide data when there are too many overlaps. */
1191 if (ch->reset >= 0x7fff)
1192 return;
1193
1194 for (offset = start; offset <= end; offset++) {
1195 struct annotation_line *al = notes->src->offsets[offset];
1196
1197 if (al && al->cycles && al->cycles->ipc == 0.0) {
1198 al->cycles->ipc = ipc;
1199 cover_insn++;
1200 }
1201 }
1202
1203 branch = annotation__get_branch(notes);
1204 if (cover_insn && branch) {
1205 branch->hit_cycles += ch->cycles;
1206 branch->hit_insn += n_insn * ch->num;
1207 branch->cover_insn += cover_insn;
1208 }
1209 }
1210}
1211
1212static int annotation__compute_ipc(struct annotation *notes, size_t size)
1213{
1214 int err = 0;
1215 s64 offset;
1216
1217 if (!notes->branch || !notes->branch->cycles_hist)
1218 return 0;
1219
1220 notes->branch->total_insn = annotation__count_insn(notes, 0, size - 1);
1221 notes->branch->hit_cycles = 0;
1222 notes->branch->hit_insn = 0;
1223 notes->branch->cover_insn = 0;
1224
1225 annotation__lock(notes);
1226 for (offset = size - 1; offset >= 0; --offset) {
1227 struct cyc_hist *ch;
1228
1229 ch = ¬es->branch->cycles_hist[offset];
1230 if (ch && ch->cycles) {
1231 struct annotation_line *al;
1232
1233 al = notes->src->offsets[offset];
1234 if (al && al->cycles == NULL) {
1235 al->cycles = zalloc(sizeof(*al->cycles));
1236 if (al->cycles == NULL) {
1237 err = ENOMEM;
1238 break;
1239 }
1240 }
1241 if (ch->have_start)
1242 annotation__count_and_fill(notes, ch->start, offset, ch);
1243 if (al && ch->num_aggr) {
1244 al->cycles->avg = ch->cycles_aggr / ch->num_aggr;
1245 al->cycles->max = ch->cycles_max;
1246 al->cycles->min = ch->cycles_min;
1247 }
1248 }
1249 }
1250
1251 if (err) {
1252 while (++offset < (s64)size) {
1253 struct cyc_hist *ch = ¬es->branch->cycles_hist[offset];
1254
1255 if (ch && ch->cycles) {
1256 struct annotation_line *al = notes->src->offsets[offset];
1257 if (al)
1258 zfree(&al->cycles);
1259 }
1260 }
1261 }
1262
1263 annotation__unlock(notes);
1264 return 0;
1265}
1266
1267int addr_map_symbol__inc_samples(struct addr_map_symbol *ams, struct perf_sample *sample,
1268 struct evsel *evsel)
1269{
1270 return symbol__inc_addr_samples(&ams->ms, evsel, ams->al_addr, sample);
1271}
1272
1273int hist_entry__inc_addr_samples(struct hist_entry *he, struct perf_sample *sample,
1274 struct evsel *evsel, u64 ip)
1275{
1276 return symbol__inc_addr_samples(&he->ms, evsel, ip, sample);
1277}
1278
1279static void disasm_line__init_ins(struct disasm_line *dl, struct arch *arch, struct map_symbol *ms)
1280{
1281 dl->ins.ops = ins__find(arch, dl->ins.name);
1282
1283 if (!dl->ins.ops)
1284 return;
1285
1286 if (dl->ins.ops->parse && dl->ins.ops->parse(arch, &dl->ops, ms) < 0)
1287 dl->ins.ops = NULL;
1288}
1289
1290static int disasm_line__parse(char *line, const char **namep, char **rawp)
1291{
1292 char tmp, *name = skip_spaces(line);
1293
1294 if (name[0] == '\0')
1295 return -1;
1296
1297 *rawp = name + 1;
1298
1299 while ((*rawp)[0] != '\0' && !isspace((*rawp)[0]))
1300 ++*rawp;
1301
1302 tmp = (*rawp)[0];
1303 (*rawp)[0] = '\0';
1304 *namep = strdup(name);
1305
1306 if (*namep == NULL)
1307 goto out;
1308
1309 (*rawp)[0] = tmp;
1310 *rawp = strim(*rawp);
1311
1312 return 0;
1313
1314out:
1315 return -1;
1316}
1317
1318struct annotate_args {
1319 struct arch *arch;
1320 struct map_symbol ms;
1321 struct evsel *evsel;
1322 struct annotation_options *options;
1323 s64 offset;
1324 char *line;
1325 int line_nr;
1326 char *fileloc;
1327};
1328
1329static void annotation_line__init(struct annotation_line *al,
1330 struct annotate_args *args,
1331 int nr)
1332{
1333 al->offset = args->offset;
1334 al->line = strdup(args->line);
1335 al->line_nr = args->line_nr;
1336 al->fileloc = args->fileloc;
1337 al->data_nr = nr;
1338}
1339
1340static void annotation_line__exit(struct annotation_line *al)
1341{
1342 zfree_srcline(&al->path);
1343 zfree(&al->line);
1344 zfree(&al->cycles);
1345}
1346
1347static size_t disasm_line_size(int nr)
1348{
1349 struct annotation_line *al;
1350
1351 return (sizeof(struct disasm_line) + (sizeof(al->data[0]) * nr));
1352}
1353
1354/*
1355 * Allocating the disasm annotation line data with
1356 * following structure:
1357 *
1358 * -------------------------------------------
1359 * struct disasm_line | struct annotation_line
1360 * -------------------------------------------
1361 *
1362 * We have 'struct annotation_line' member as last member
1363 * of 'struct disasm_line' to have an easy access.
1364 */
1365static struct disasm_line *disasm_line__new(struct annotate_args *args)
1366{
1367 struct disasm_line *dl = NULL;
1368 int nr = 1;
1369
1370 if (evsel__is_group_event(args->evsel))
1371 nr = args->evsel->core.nr_members;
1372
1373 dl = zalloc(disasm_line_size(nr));
1374 if (!dl)
1375 return NULL;
1376
1377 annotation_line__init(&dl->al, args, nr);
1378 if (dl->al.line == NULL)
1379 goto out_delete;
1380
1381 if (args->offset != -1) {
1382 if (disasm_line__parse(dl->al.line, &dl->ins.name, &dl->ops.raw) < 0)
1383 goto out_free_line;
1384
1385 disasm_line__init_ins(dl, args->arch, &args->ms);
1386 }
1387
1388 return dl;
1389
1390out_free_line:
1391 zfree(&dl->al.line);
1392out_delete:
1393 free(dl);
1394 return NULL;
1395}
1396
1397void disasm_line__free(struct disasm_line *dl)
1398{
1399 if (dl->ins.ops && dl->ins.ops->free)
1400 dl->ins.ops->free(&dl->ops);
1401 else
1402 ins__delete(&dl->ops);
1403 zfree(&dl->ins.name);
1404 annotation_line__exit(&dl->al);
1405 free(dl);
1406}
1407
1408int disasm_line__scnprintf(struct disasm_line *dl, char *bf, size_t size, bool raw, int max_ins_name)
1409{
1410 if (raw || !dl->ins.ops)
1411 return scnprintf(bf, size, "%-*s %s", max_ins_name, dl->ins.name, dl->ops.raw);
1412
1413 return ins__scnprintf(&dl->ins, bf, size, &dl->ops, max_ins_name);
1414}
1415
1416void annotation__exit(struct annotation *notes)
1417{
1418 annotated_source__delete(notes->src);
1419 annotated_branch__delete(notes->branch);
1420}
1421
1422static struct sharded_mutex *sharded_mutex;
1423
1424static void annotation__init_sharded_mutex(void)
1425{
1426 /* As many mutexes as there are CPUs. */
1427 sharded_mutex = sharded_mutex__new(cpu__max_present_cpu().cpu);
1428}
1429
1430static size_t annotation__hash(const struct annotation *notes)
1431{
1432 return (size_t)notes;
1433}
1434
1435static struct mutex *annotation__get_mutex(const struct annotation *notes)
1436{
1437 static pthread_once_t once = PTHREAD_ONCE_INIT;
1438
1439 pthread_once(&once, annotation__init_sharded_mutex);
1440 if (!sharded_mutex)
1441 return NULL;
1442
1443 return sharded_mutex__get_mutex(sharded_mutex, annotation__hash(notes));
1444}
1445
1446void annotation__lock(struct annotation *notes)
1447 NO_THREAD_SAFETY_ANALYSIS
1448{
1449 struct mutex *mutex = annotation__get_mutex(notes);
1450
1451 if (mutex)
1452 mutex_lock(mutex);
1453}
1454
1455void annotation__unlock(struct annotation *notes)
1456 NO_THREAD_SAFETY_ANALYSIS
1457{
1458 struct mutex *mutex = annotation__get_mutex(notes);
1459
1460 if (mutex)
1461 mutex_unlock(mutex);
1462}
1463
1464bool annotation__trylock(struct annotation *notes)
1465{
1466 struct mutex *mutex = annotation__get_mutex(notes);
1467
1468 if (!mutex)
1469 return false;
1470
1471 return mutex_trylock(mutex);
1472}
1473
1474
1475static void annotation_line__add(struct annotation_line *al, struct list_head *head)
1476{
1477 list_add_tail(&al->node, head);
1478}
1479
1480struct annotation_line *
1481annotation_line__next(struct annotation_line *pos, struct list_head *head)
1482{
1483 list_for_each_entry_continue(pos, head, node)
1484 if (pos->offset >= 0)
1485 return pos;
1486
1487 return NULL;
1488}
1489
1490static const char *annotate__address_color(struct block_range *br)
1491{
1492 double cov = block_range__coverage(br);
1493
1494 if (cov >= 0) {
1495 /* mark red for >75% coverage */
1496 if (cov > 0.75)
1497 return PERF_COLOR_RED;
1498
1499 /* mark dull for <1% coverage */
1500 if (cov < 0.01)
1501 return PERF_COLOR_NORMAL;
1502 }
1503
1504 return PERF_COLOR_MAGENTA;
1505}
1506
1507static const char *annotate__asm_color(struct block_range *br)
1508{
1509 double cov = block_range__coverage(br);
1510
1511 if (cov >= 0) {
1512 /* mark dull for <1% coverage */
1513 if (cov < 0.01)
1514 return PERF_COLOR_NORMAL;
1515 }
1516
1517 return PERF_COLOR_BLUE;
1518}
1519
1520static void annotate__branch_printf(struct block_range *br, u64 addr)
1521{
1522 bool emit_comment = true;
1523
1524 if (!br)
1525 return;
1526
1527#if 1
1528 if (br->is_target && br->start == addr) {
1529 struct block_range *branch = br;
1530 double p;
1531
1532 /*
1533 * Find matching branch to our target.
1534 */
1535 while (!branch->is_branch)
1536 branch = block_range__next(branch);
1537
1538 p = 100 *(double)br->entry / branch->coverage;
1539
1540 if (p > 0.1) {
1541 if (emit_comment) {
1542 emit_comment = false;
1543 printf("\t#");
1544 }
1545
1546 /*
1547 * The percentage of coverage joined at this target in relation
1548 * to the next branch.
1549 */
1550 printf(" +%.2f%%", p);
1551 }
1552 }
1553#endif
1554 if (br->is_branch && br->end == addr) {
1555 double p = 100*(double)br->taken / br->coverage;
1556
1557 if (p > 0.1) {
1558 if (emit_comment) {
1559 emit_comment = false;
1560 printf("\t#");
1561 }
1562
1563 /*
1564 * The percentage of coverage leaving at this branch, and
1565 * its prediction ratio.
1566 */
1567 printf(" -%.2f%% (p:%.2f%%)", p, 100*(double)br->pred / br->taken);
1568 }
1569 }
1570}
1571
1572static int disasm_line__print(struct disasm_line *dl, u64 start, int addr_fmt_width)
1573{
1574 s64 offset = dl->al.offset;
1575 const u64 addr = start + offset;
1576 struct block_range *br;
1577
1578 br = block_range__find(addr);
1579 color_fprintf(stdout, annotate__address_color(br), " %*" PRIx64 ":", addr_fmt_width, addr);
1580 color_fprintf(stdout, annotate__asm_color(br), "%s", dl->al.line);
1581 annotate__branch_printf(br, addr);
1582 return 0;
1583}
1584
1585static int
1586annotation_line__print(struct annotation_line *al, struct symbol *sym, u64 start,
1587 struct evsel *evsel, u64 len, int min_pcnt, int printed,
1588 int max_lines, struct annotation_line *queue, int addr_fmt_width,
1589 int percent_type)
1590{
1591 struct disasm_line *dl = container_of(al, struct disasm_line, al);
1592 static const char *prev_line;
1593
1594 if (al->offset != -1) {
1595 double max_percent = 0.0;
1596 int i, nr_percent = 1;
1597 const char *color;
1598 struct annotation *notes = symbol__annotation(sym);
1599
1600 for (i = 0; i < al->data_nr; i++) {
1601 double percent;
1602
1603 percent = annotation_data__percent(&al->data[i],
1604 percent_type);
1605
1606 if (percent > max_percent)
1607 max_percent = percent;
1608 }
1609
1610 if (al->data_nr > nr_percent)
1611 nr_percent = al->data_nr;
1612
1613 if (max_percent < min_pcnt)
1614 return -1;
1615
1616 if (max_lines && printed >= max_lines)
1617 return 1;
1618
1619 if (queue != NULL) {
1620 list_for_each_entry_from(queue, ¬es->src->source, node) {
1621 if (queue == al)
1622 break;
1623 annotation_line__print(queue, sym, start, evsel, len,
1624 0, 0, 1, NULL, addr_fmt_width,
1625 percent_type);
1626 }
1627 }
1628
1629 color = get_percent_color(max_percent);
1630
1631 for (i = 0; i < nr_percent; i++) {
1632 struct annotation_data *data = &al->data[i];
1633 double percent;
1634
1635 percent = annotation_data__percent(data, percent_type);
1636 color = get_percent_color(percent);
1637
1638 if (symbol_conf.show_total_period)
1639 color_fprintf(stdout, color, " %11" PRIu64,
1640 data->he.period);
1641 else if (symbol_conf.show_nr_samples)
1642 color_fprintf(stdout, color, " %7" PRIu64,
1643 data->he.nr_samples);
1644 else
1645 color_fprintf(stdout, color, " %7.2f", percent);
1646 }
1647
1648 printf(" : ");
1649
1650 disasm_line__print(dl, start, addr_fmt_width);
1651
1652 /*
1653 * Also color the filename and line if needed, with
1654 * the same color than the percentage. Don't print it
1655 * twice for close colored addr with the same filename:line
1656 */
1657 if (al->path) {
1658 if (!prev_line || strcmp(prev_line, al->path)) {
1659 color_fprintf(stdout, color, " // %s", al->path);
1660 prev_line = al->path;
1661 }
1662 }
1663
1664 printf("\n");
1665 } else if (max_lines && printed >= max_lines)
1666 return 1;
1667 else {
1668 int width = symbol_conf.show_total_period ? 12 : 8;
1669
1670 if (queue)
1671 return -1;
1672
1673 if (evsel__is_group_event(evsel))
1674 width *= evsel->core.nr_members;
1675
1676 if (!*al->line)
1677 printf(" %*s:\n", width, " ");
1678 else
1679 printf(" %*s: %-*d %s\n", width, " ", addr_fmt_width, al->line_nr, al->line);
1680 }
1681
1682 return 0;
1683}
1684
1685/*
1686 * symbol__parse_objdump_line() parses objdump output (with -d --no-show-raw)
1687 * which looks like following
1688 *
1689 * 0000000000415500 <_init>:
1690 * 415500: sub $0x8,%rsp
1691 * 415504: mov 0x2f5ad5(%rip),%rax # 70afe0 <_DYNAMIC+0x2f8>
1692 * 41550b: test %rax,%rax
1693 * 41550e: je 415515 <_init+0x15>
1694 * 415510: callq 416e70 <__gmon_start__@plt>
1695 * 415515: add $0x8,%rsp
1696 * 415519: retq
1697 *
1698 * it will be parsed and saved into struct disasm_line as
1699 * <offset> <name> <ops.raw>
1700 *
1701 * The offset will be a relative offset from the start of the symbol and -1
1702 * means that it's not a disassembly line so should be treated differently.
1703 * The ops.raw part will be parsed further according to type of the instruction.
1704 */
1705static int symbol__parse_objdump_line(struct symbol *sym,
1706 struct annotate_args *args,
1707 char *parsed_line, int *line_nr, char **fileloc)
1708{
1709 struct map *map = args->ms.map;
1710 struct annotation *notes = symbol__annotation(sym);
1711 struct disasm_line *dl;
1712 char *tmp;
1713 s64 line_ip, offset = -1;
1714 regmatch_t match[2];
1715
1716 /* /filename:linenr ? Save line number and ignore. */
1717 if (regexec(&file_lineno, parsed_line, 2, match, 0) == 0) {
1718 *line_nr = atoi(parsed_line + match[1].rm_so);
1719 free(*fileloc);
1720 *fileloc = strdup(parsed_line);
1721 return 0;
1722 }
1723
1724 /* Process hex address followed by ':'. */
1725 line_ip = strtoull(parsed_line, &tmp, 16);
1726 if (parsed_line != tmp && tmp[0] == ':' && tmp[1] != '\0') {
1727 u64 start = map__rip_2objdump(map, sym->start),
1728 end = map__rip_2objdump(map, sym->end);
1729
1730 offset = line_ip - start;
1731 if ((u64)line_ip < start || (u64)line_ip >= end)
1732 offset = -1;
1733 else
1734 parsed_line = tmp + 1;
1735 }
1736
1737 args->offset = offset;
1738 args->line = parsed_line;
1739 args->line_nr = *line_nr;
1740 args->fileloc = *fileloc;
1741 args->ms.sym = sym;
1742
1743 dl = disasm_line__new(args);
1744 (*line_nr)++;
1745
1746 if (dl == NULL)
1747 return -1;
1748
1749 if (!disasm_line__has_local_offset(dl)) {
1750 dl->ops.target.offset = dl->ops.target.addr -
1751 map__rip_2objdump(map, sym->start);
1752 dl->ops.target.offset_avail = true;
1753 }
1754
1755 /* kcore has no symbols, so add the call target symbol */
1756 if (dl->ins.ops && ins__is_call(&dl->ins) && !dl->ops.target.sym) {
1757 struct addr_map_symbol target = {
1758 .addr = dl->ops.target.addr,
1759 .ms = { .map = map, },
1760 };
1761
1762 if (!maps__find_ams(args->ms.maps, &target) &&
1763 target.ms.sym->start == target.al_addr)
1764 dl->ops.target.sym = target.ms.sym;
1765 }
1766
1767 annotation_line__add(&dl->al, ¬es->src->source);
1768 return 0;
1769}
1770
1771static __attribute__((constructor)) void symbol__init_regexpr(void)
1772{
1773 regcomp(&file_lineno, "^/[^:]+:([0-9]+)", REG_EXTENDED);
1774}
1775
1776static void delete_last_nop(struct symbol *sym)
1777{
1778 struct annotation *notes = symbol__annotation(sym);
1779 struct list_head *list = ¬es->src->source;
1780 struct disasm_line *dl;
1781
1782 while (!list_empty(list)) {
1783 dl = list_entry(list->prev, struct disasm_line, al.node);
1784
1785 if (dl->ins.ops) {
1786 if (dl->ins.ops != &nop_ops)
1787 return;
1788 } else {
1789 if (!strstr(dl->al.line, " nop ") &&
1790 !strstr(dl->al.line, " nopl ") &&
1791 !strstr(dl->al.line, " nopw "))
1792 return;
1793 }
1794
1795 list_del_init(&dl->al.node);
1796 disasm_line__free(dl);
1797 }
1798}
1799
1800int symbol__strerror_disassemble(struct map_symbol *ms, int errnum, char *buf, size_t buflen)
1801{
1802 struct dso *dso = map__dso(ms->map);
1803
1804 BUG_ON(buflen == 0);
1805
1806 if (errnum >= 0) {
1807 str_error_r(errnum, buf, buflen);
1808 return 0;
1809 }
1810
1811 switch (errnum) {
1812 case SYMBOL_ANNOTATE_ERRNO__NO_VMLINUX: {
1813 char bf[SBUILD_ID_SIZE + 15] = " with build id ";
1814 char *build_id_msg = NULL;
1815
1816 if (dso->has_build_id) {
1817 build_id__sprintf(&dso->bid, bf + 15);
1818 build_id_msg = bf;
1819 }
1820 scnprintf(buf, buflen,
1821 "No vmlinux file%s\nwas found in the path.\n\n"
1822 "Note that annotation using /proc/kcore requires CAP_SYS_RAWIO capability.\n\n"
1823 "Please use:\n\n"
1824 " perf buildid-cache -vu vmlinux\n\n"
1825 "or:\n\n"
1826 " --vmlinux vmlinux\n", build_id_msg ?: "");
1827 }
1828 break;
1829 case SYMBOL_ANNOTATE_ERRNO__NO_LIBOPCODES_FOR_BPF:
1830 scnprintf(buf, buflen, "Please link with binutils's libopcode to enable BPF annotation");
1831 break;
1832 case SYMBOL_ANNOTATE_ERRNO__ARCH_INIT_REGEXP:
1833 scnprintf(buf, buflen, "Problems with arch specific instruction name regular expressions.");
1834 break;
1835 case SYMBOL_ANNOTATE_ERRNO__ARCH_INIT_CPUID_PARSING:
1836 scnprintf(buf, buflen, "Problems while parsing the CPUID in the arch specific initialization.");
1837 break;
1838 case SYMBOL_ANNOTATE_ERRNO__BPF_INVALID_FILE:
1839 scnprintf(buf, buflen, "Invalid BPF file: %s.", dso->long_name);
1840 break;
1841 case SYMBOL_ANNOTATE_ERRNO__BPF_MISSING_BTF:
1842 scnprintf(buf, buflen, "The %s BPF file has no BTF section, compile with -g or use pahole -J.",
1843 dso->long_name);
1844 break;
1845 default:
1846 scnprintf(buf, buflen, "Internal error: Invalid %d error code\n", errnum);
1847 break;
1848 }
1849
1850 return 0;
1851}
1852
1853static int dso__disassemble_filename(struct dso *dso, char *filename, size_t filename_size)
1854{
1855 char linkname[PATH_MAX];
1856 char *build_id_filename;
1857 char *build_id_path = NULL;
1858 char *pos;
1859 int len;
1860
1861 if (dso->symtab_type == DSO_BINARY_TYPE__KALLSYMS &&
1862 !dso__is_kcore(dso))
1863 return SYMBOL_ANNOTATE_ERRNO__NO_VMLINUX;
1864
1865 build_id_filename = dso__build_id_filename(dso, NULL, 0, false);
1866 if (build_id_filename) {
1867 __symbol__join_symfs(filename, filename_size, build_id_filename);
1868 free(build_id_filename);
1869 } else {
1870 if (dso->has_build_id)
1871 return ENOMEM;
1872 goto fallback;
1873 }
1874
1875 build_id_path = strdup(filename);
1876 if (!build_id_path)
1877 return ENOMEM;
1878
1879 /*
1880 * old style build-id cache has name of XX/XXXXXXX.. while
1881 * new style has XX/XXXXXXX../{elf,kallsyms,vdso}.
1882 * extract the build-id part of dirname in the new style only.
1883 */
1884 pos = strrchr(build_id_path, '/');
1885 if (pos && strlen(pos) < SBUILD_ID_SIZE - 2)
1886 dirname(build_id_path);
1887
1888 if (dso__is_kcore(dso))
1889 goto fallback;
1890
1891 len = readlink(build_id_path, linkname, sizeof(linkname) - 1);
1892 if (len < 0)
1893 goto fallback;
1894
1895 linkname[len] = '\0';
1896 if (strstr(linkname, DSO__NAME_KALLSYMS) ||
1897 access(filename, R_OK)) {
1898fallback:
1899 /*
1900 * If we don't have build-ids or the build-id file isn't in the
1901 * cache, or is just a kallsyms file, well, lets hope that this
1902 * DSO is the same as when 'perf record' ran.
1903 */
1904 if (dso->kernel && dso->long_name[0] == '/')
1905 snprintf(filename, filename_size, "%s", dso->long_name);
1906 else
1907 __symbol__join_symfs(filename, filename_size, dso->long_name);
1908
1909 mutex_lock(&dso->lock);
1910 if (access(filename, R_OK) && errno == ENOENT && dso->nsinfo) {
1911 char *new_name = dso__filename_with_chroot(dso, filename);
1912 if (new_name) {
1913 strlcpy(filename, new_name, filename_size);
1914 free(new_name);
1915 }
1916 }
1917 mutex_unlock(&dso->lock);
1918 }
1919
1920 free(build_id_path);
1921 return 0;
1922}
1923
1924#if defined(HAVE_LIBBFD_SUPPORT) && defined(HAVE_LIBBPF_SUPPORT)
1925#define PACKAGE "perf"
1926#include <bfd.h>
1927#include <dis-asm.h>
1928#include <bpf/bpf.h>
1929#include <bpf/btf.h>
1930#include <bpf/libbpf.h>
1931#include <linux/btf.h>
1932#include <tools/dis-asm-compat.h>
1933
1934static int symbol__disassemble_bpf(struct symbol *sym,
1935 struct annotate_args *args)
1936{
1937 struct annotation *notes = symbol__annotation(sym);
1938 struct bpf_prog_linfo *prog_linfo = NULL;
1939 struct bpf_prog_info_node *info_node;
1940 int len = sym->end - sym->start;
1941 disassembler_ftype disassemble;
1942 struct map *map = args->ms.map;
1943 struct perf_bpil *info_linear;
1944 struct disassemble_info info;
1945 struct dso *dso = map__dso(map);
1946 int pc = 0, count, sub_id;
1947 struct btf *btf = NULL;
1948 char tpath[PATH_MAX];
1949 size_t buf_size;
1950 int nr_skip = 0;
1951 char *buf;
1952 bfd *bfdf;
1953 int ret;
1954 FILE *s;
1955
1956 if (dso->binary_type != DSO_BINARY_TYPE__BPF_PROG_INFO)
1957 return SYMBOL_ANNOTATE_ERRNO__BPF_INVALID_FILE;
1958
1959 pr_debug("%s: handling sym %s addr %" PRIx64 " len %" PRIx64 "\n", __func__,
1960 sym->name, sym->start, sym->end - sym->start);
1961
1962 memset(tpath, 0, sizeof(tpath));
1963 perf_exe(tpath, sizeof(tpath));
1964
1965 bfdf = bfd_openr(tpath, NULL);
1966 if (bfdf == NULL)
1967 abort();
1968
1969 if (!bfd_check_format(bfdf, bfd_object))
1970 abort();
1971
1972 s = open_memstream(&buf, &buf_size);
1973 if (!s) {
1974 ret = errno;
1975 goto out;
1976 }
1977 init_disassemble_info_compat(&info, s,
1978 (fprintf_ftype) fprintf,
1979 fprintf_styled);
1980 info.arch = bfd_get_arch(bfdf);
1981 info.mach = bfd_get_mach(bfdf);
1982
1983 info_node = perf_env__find_bpf_prog_info(dso->bpf_prog.env,
1984 dso->bpf_prog.id);
1985 if (!info_node) {
1986 ret = SYMBOL_ANNOTATE_ERRNO__BPF_MISSING_BTF;
1987 goto out;
1988 }
1989 info_linear = info_node->info_linear;
1990 sub_id = dso->bpf_prog.sub_id;
1991
1992 info.buffer = (void *)(uintptr_t)(info_linear->info.jited_prog_insns);
1993 info.buffer_length = info_linear->info.jited_prog_len;
1994
1995 if (info_linear->info.nr_line_info)
1996 prog_linfo = bpf_prog_linfo__new(&info_linear->info);
1997
1998 if (info_linear->info.btf_id) {
1999 struct btf_node *node;
2000
2001 node = perf_env__find_btf(dso->bpf_prog.env,
2002 info_linear->info.btf_id);
2003 if (node)
2004 btf = btf__new((__u8 *)(node->data),
2005 node->data_size);
2006 }
2007
2008 disassemble_init_for_target(&info);
2009
2010#ifdef DISASM_FOUR_ARGS_SIGNATURE
2011 disassemble = disassembler(info.arch,
2012 bfd_big_endian(bfdf),
2013 info.mach,
2014 bfdf);
2015#else
2016 disassemble = disassembler(bfdf);
2017#endif
2018 if (disassemble == NULL)
2019 abort();
2020
2021 fflush(s);
2022 do {
2023 const struct bpf_line_info *linfo = NULL;
2024 struct disasm_line *dl;
2025 size_t prev_buf_size;
2026 const char *srcline;
2027 u64 addr;
2028
2029 addr = pc + ((u64 *)(uintptr_t)(info_linear->info.jited_ksyms))[sub_id];
2030 count = disassemble(pc, &info);
2031
2032 if (prog_linfo)
2033 linfo = bpf_prog_linfo__lfind_addr_func(prog_linfo,
2034 addr, sub_id,
2035 nr_skip);
2036
2037 if (linfo && btf) {
2038 srcline = btf__name_by_offset(btf, linfo->line_off);
2039 nr_skip++;
2040 } else
2041 srcline = NULL;
2042
2043 fprintf(s, "\n");
2044 prev_buf_size = buf_size;
2045 fflush(s);
2046
2047 if (!annotate_opts.hide_src_code && srcline) {
2048 args->offset = -1;
2049 args->line = strdup(srcline);
2050 args->line_nr = 0;
2051 args->fileloc = NULL;
2052 args->ms.sym = sym;
2053 dl = disasm_line__new(args);
2054 if (dl) {
2055 annotation_line__add(&dl->al,
2056 ¬es->src->source);
2057 }
2058 }
2059
2060 args->offset = pc;
2061 args->line = buf + prev_buf_size;
2062 args->line_nr = 0;
2063 args->fileloc = NULL;
2064 args->ms.sym = sym;
2065 dl = disasm_line__new(args);
2066 if (dl)
2067 annotation_line__add(&dl->al, ¬es->src->source);
2068
2069 pc += count;
2070 } while (count > 0 && pc < len);
2071
2072 ret = 0;
2073out:
2074 free(prog_linfo);
2075 btf__free(btf);
2076 fclose(s);
2077 bfd_close(bfdf);
2078 return ret;
2079}
2080#else // defined(HAVE_LIBBFD_SUPPORT) && defined(HAVE_LIBBPF_SUPPORT)
2081static int symbol__disassemble_bpf(struct symbol *sym __maybe_unused,
2082 struct annotate_args *args __maybe_unused)
2083{
2084 return SYMBOL_ANNOTATE_ERRNO__NO_LIBOPCODES_FOR_BPF;
2085}
2086#endif // defined(HAVE_LIBBFD_SUPPORT) && defined(HAVE_LIBBPF_SUPPORT)
2087
2088static int
2089symbol__disassemble_bpf_image(struct symbol *sym,
2090 struct annotate_args *args)
2091{
2092 struct annotation *notes = symbol__annotation(sym);
2093 struct disasm_line *dl;
2094
2095 args->offset = -1;
2096 args->line = strdup("to be implemented");
2097 args->line_nr = 0;
2098 args->fileloc = NULL;
2099 dl = disasm_line__new(args);
2100 if (dl)
2101 annotation_line__add(&dl->al, ¬es->src->source);
2102
2103 zfree(&args->line);
2104 return 0;
2105}
2106
2107/*
2108 * Possibly create a new version of line with tabs expanded. Returns the
2109 * existing or new line, storage is updated if a new line is allocated. If
2110 * allocation fails then NULL is returned.
2111 */
2112static char *expand_tabs(char *line, char **storage, size_t *storage_len)
2113{
2114 size_t i, src, dst, len, new_storage_len, num_tabs;
2115 char *new_line;
2116 size_t line_len = strlen(line);
2117
2118 for (num_tabs = 0, i = 0; i < line_len; i++)
2119 if (line[i] == '\t')
2120 num_tabs++;
2121
2122 if (num_tabs == 0)
2123 return line;
2124
2125 /*
2126 * Space for the line and '\0', less the leading and trailing
2127 * spaces. Each tab may introduce 7 additional spaces.
2128 */
2129 new_storage_len = line_len + 1 + (num_tabs * 7);
2130
2131 new_line = malloc(new_storage_len);
2132 if (new_line == NULL) {
2133 pr_err("Failure allocating memory for tab expansion\n");
2134 return NULL;
2135 }
2136
2137 /*
2138 * Copy regions starting at src and expand tabs. If there are two
2139 * adjacent tabs then 'src == i', the memcpy is of size 0 and the spaces
2140 * are inserted.
2141 */
2142 for (i = 0, src = 0, dst = 0; i < line_len && num_tabs; i++) {
2143 if (line[i] == '\t') {
2144 len = i - src;
2145 memcpy(&new_line[dst], &line[src], len);
2146 dst += len;
2147 new_line[dst++] = ' ';
2148 while (dst % 8 != 0)
2149 new_line[dst++] = ' ';
2150 src = i + 1;
2151 num_tabs--;
2152 }
2153 }
2154
2155 /* Expand the last region. */
2156 len = line_len - src;
2157 memcpy(&new_line[dst], &line[src], len);
2158 dst += len;
2159 new_line[dst] = '\0';
2160
2161 free(*storage);
2162 *storage = new_line;
2163 *storage_len = new_storage_len;
2164 return new_line;
2165
2166}
2167
2168static int symbol__disassemble(struct symbol *sym, struct annotate_args *args)
2169{
2170 struct annotation_options *opts = &annotate_opts;
2171 struct map *map = args->ms.map;
2172 struct dso *dso = map__dso(map);
2173 char *command;
2174 FILE *file;
2175 char symfs_filename[PATH_MAX];
2176 struct kcore_extract kce;
2177 bool delete_extract = false;
2178 bool decomp = false;
2179 int lineno = 0;
2180 char *fileloc = NULL;
2181 int nline;
2182 char *line;
2183 size_t line_len;
2184 const char *objdump_argv[] = {
2185 "/bin/sh",
2186 "-c",
2187 NULL, /* Will be the objdump command to run. */
2188 "--",
2189 NULL, /* Will be the symfs path. */
2190 NULL,
2191 };
2192 struct child_process objdump_process;
2193 int err = dso__disassemble_filename(dso, symfs_filename, sizeof(symfs_filename));
2194
2195 if (err)
2196 return err;
2197
2198 pr_debug("%s: filename=%s, sym=%s, start=%#" PRIx64 ", end=%#" PRIx64 "\n", __func__,
2199 symfs_filename, sym->name, map__unmap_ip(map, sym->start),
2200 map__unmap_ip(map, sym->end));
2201
2202 pr_debug("annotating [%p] %30s : [%p] %30s\n",
2203 dso, dso->long_name, sym, sym->name);
2204
2205 if (dso->binary_type == DSO_BINARY_TYPE__BPF_PROG_INFO) {
2206 return symbol__disassemble_bpf(sym, args);
2207 } else if (dso->binary_type == DSO_BINARY_TYPE__BPF_IMAGE) {
2208 return symbol__disassemble_bpf_image(sym, args);
2209 } else if (dso__is_kcore(dso)) {
2210 kce.kcore_filename = symfs_filename;
2211 kce.addr = map__rip_2objdump(map, sym->start);
2212 kce.offs = sym->start;
2213 kce.len = sym->end - sym->start;
2214 if (!kcore_extract__create(&kce)) {
2215 delete_extract = true;
2216 strlcpy(symfs_filename, kce.extract_filename,
2217 sizeof(symfs_filename));
2218 }
2219 } else if (dso__needs_decompress(dso)) {
2220 char tmp[KMOD_DECOMP_LEN];
2221
2222 if (dso__decompress_kmodule_path(dso, symfs_filename,
2223 tmp, sizeof(tmp)) < 0)
2224 return -1;
2225
2226 decomp = true;
2227 strcpy(symfs_filename, tmp);
2228 }
2229
2230 err = asprintf(&command,
2231 "%s %s%s --start-address=0x%016" PRIx64
2232 " --stop-address=0x%016" PRIx64
2233 " %s -d %s %s %s %c%s%c %s%s -C \"$1\"",
2234 opts->objdump_path ?: "objdump",
2235 opts->disassembler_style ? "-M " : "",
2236 opts->disassembler_style ?: "",
2237 map__rip_2objdump(map, sym->start),
2238 map__rip_2objdump(map, sym->end),
2239 opts->show_linenr ? "-l" : "",
2240 opts->show_asm_raw ? "" : "--no-show-raw-insn",
2241 opts->annotate_src ? "-S" : "",
2242 opts->prefix ? "--prefix " : "",
2243 opts->prefix ? '"' : ' ',
2244 opts->prefix ?: "",
2245 opts->prefix ? '"' : ' ',
2246 opts->prefix_strip ? "--prefix-strip=" : "",
2247 opts->prefix_strip ?: "");
2248
2249 if (err < 0) {
2250 pr_err("Failure allocating memory for the command to run\n");
2251 goto out_remove_tmp;
2252 }
2253
2254 pr_debug("Executing: %s\n", command);
2255
2256 objdump_argv[2] = command;
2257 objdump_argv[4] = symfs_filename;
2258
2259 /* Create a pipe to read from for stdout */
2260 memset(&objdump_process, 0, sizeof(objdump_process));
2261 objdump_process.argv = objdump_argv;
2262 objdump_process.out = -1;
2263 objdump_process.err = -1;
2264 objdump_process.no_stderr = 1;
2265 if (start_command(&objdump_process)) {
2266 pr_err("Failure starting to run %s\n", command);
2267 err = -1;
2268 goto out_free_command;
2269 }
2270
2271 file = fdopen(objdump_process.out, "r");
2272 if (!file) {
2273 pr_err("Failure creating FILE stream for %s\n", command);
2274 /*
2275 * If we were using debug info should retry with
2276 * original binary.
2277 */
2278 err = -1;
2279 goto out_close_stdout;
2280 }
2281
2282 /* Storage for getline. */
2283 line = NULL;
2284 line_len = 0;
2285
2286 nline = 0;
2287 while (!feof(file)) {
2288 const char *match;
2289 char *expanded_line;
2290
2291 if (getline(&line, &line_len, file) < 0 || !line)
2292 break;
2293
2294 /* Skip lines containing "filename:" */
2295 match = strstr(line, symfs_filename);
2296 if (match && match[strlen(symfs_filename)] == ':')
2297 continue;
2298
2299 expanded_line = strim(line);
2300 expanded_line = expand_tabs(expanded_line, &line, &line_len);
2301 if (!expanded_line)
2302 break;
2303
2304 /*
2305 * The source code line number (lineno) needs to be kept in
2306 * across calls to symbol__parse_objdump_line(), so that it
2307 * can associate it with the instructions till the next one.
2308 * See disasm_line__new() and struct disasm_line::line_nr.
2309 */
2310 if (symbol__parse_objdump_line(sym, args, expanded_line,
2311 &lineno, &fileloc) < 0)
2312 break;
2313 nline++;
2314 }
2315 free(line);
2316 free(fileloc);
2317
2318 err = finish_command(&objdump_process);
2319 if (err)
2320 pr_err("Error running %s\n", command);
2321
2322 if (nline == 0) {
2323 err = -1;
2324 pr_err("No output from %s\n", command);
2325 }
2326
2327 /*
2328 * kallsyms does not have symbol sizes so there may a nop at the end.
2329 * Remove it.
2330 */
2331 if (dso__is_kcore(dso))
2332 delete_last_nop(sym);
2333
2334 fclose(file);
2335
2336out_close_stdout:
2337 close(objdump_process.out);
2338
2339out_free_command:
2340 free(command);
2341
2342out_remove_tmp:
2343 if (decomp)
2344 unlink(symfs_filename);
2345
2346 if (delete_extract)
2347 kcore_extract__delete(&kce);
2348
2349 return err;
2350}
2351
2352static void calc_percent(struct annotation *notes,
2353 struct evsel *evsel,
2354 struct annotation_data *data,
2355 s64 offset, s64 end)
2356{
2357 struct hists *hists = evsel__hists(evsel);
2358 int evidx = evsel->core.idx;
2359 struct sym_hist *sym_hist = annotation__histogram(notes, evidx);
2360 unsigned int hits = 0;
2361 u64 period = 0;
2362
2363 while (offset < end) {
2364 struct sym_hist_entry *entry;
2365
2366 entry = annotated_source__hist_entry(notes->src, evidx, offset);
2367 if (entry) {
2368 hits += entry->nr_samples;
2369 period += entry->period;
2370 }
2371 ++offset;
2372 }
2373
2374 if (sym_hist->nr_samples) {
2375 data->he.period = period;
2376 data->he.nr_samples = hits;
2377 data->percent[PERCENT_HITS_LOCAL] = 100.0 * hits / sym_hist->nr_samples;
2378 }
2379
2380 if (hists->stats.nr_non_filtered_samples)
2381 data->percent[PERCENT_HITS_GLOBAL] = 100.0 * hits / hists->stats.nr_non_filtered_samples;
2382
2383 if (sym_hist->period)
2384 data->percent[PERCENT_PERIOD_LOCAL] = 100.0 * period / sym_hist->period;
2385
2386 if (hists->stats.total_period)
2387 data->percent[PERCENT_PERIOD_GLOBAL] = 100.0 * period / hists->stats.total_period;
2388}
2389
2390static void annotation__calc_percent(struct annotation *notes,
2391 struct evsel *leader, s64 len)
2392{
2393 struct annotation_line *al, *next;
2394 struct evsel *evsel;
2395
2396 list_for_each_entry(al, ¬es->src->source, node) {
2397 s64 end;
2398 int i = 0;
2399
2400 if (al->offset == -1)
2401 continue;
2402
2403 next = annotation_line__next(al, ¬es->src->source);
2404 end = next ? next->offset : len;
2405
2406 for_each_group_evsel(evsel, leader) {
2407 struct annotation_data *data;
2408
2409 BUG_ON(i >= al->data_nr);
2410
2411 data = &al->data[i++];
2412
2413 calc_percent(notes, evsel, data, al->offset, end);
2414 }
2415 }
2416}
2417
2418void symbol__calc_percent(struct symbol *sym, struct evsel *evsel)
2419{
2420 struct annotation *notes = symbol__annotation(sym);
2421
2422 annotation__calc_percent(notes, evsel, symbol__size(sym));
2423}
2424
2425static int evsel__get_arch(struct evsel *evsel, struct arch **parch)
2426{
2427 struct perf_env *env = evsel__env(evsel);
2428 const char *arch_name = perf_env__arch(env);
2429 struct arch *arch;
2430 int err;
2431
2432 if (!arch_name)
2433 return errno;
2434
2435 *parch = arch = arch__find(arch_name);
2436 if (arch == NULL) {
2437 pr_err("%s: unsupported arch %s\n", __func__, arch_name);
2438 return ENOTSUP;
2439 }
2440
2441 if (arch->init) {
2442 err = arch->init(arch, env ? env->cpuid : NULL);
2443 if (err) {
2444 pr_err("%s: failed to initialize %s arch priv area\n",
2445 __func__, arch->name);
2446 return err;
2447 }
2448 }
2449 return 0;
2450}
2451
2452int symbol__annotate(struct map_symbol *ms, struct evsel *evsel,
2453 struct arch **parch)
2454{
2455 struct symbol *sym = ms->sym;
2456 struct annotation *notes = symbol__annotation(sym);
2457 struct annotate_args args = {
2458 .evsel = evsel,
2459 .options = &annotate_opts,
2460 };
2461 struct arch *arch = NULL;
2462 int err;
2463
2464 err = evsel__get_arch(evsel, &arch);
2465 if (err < 0)
2466 return err;
2467
2468 if (parch)
2469 *parch = arch;
2470
2471 if (!list_empty(¬es->src->source))
2472 return 0;
2473
2474 args.arch = arch;
2475 args.ms = *ms;
2476 if (annotate_opts.full_addr)
2477 notes->start = map__objdump_2mem(ms->map, ms->sym->start);
2478 else
2479 notes->start = map__rip_2objdump(ms->map, ms->sym->start);
2480
2481 return symbol__disassemble(sym, &args);
2482}
2483
2484static void insert_source_line(struct rb_root *root, struct annotation_line *al)
2485{
2486 struct annotation_line *iter;
2487 struct rb_node **p = &root->rb_node;
2488 struct rb_node *parent = NULL;
2489 unsigned int percent_type = annotate_opts.percent_type;
2490 int i, ret;
2491
2492 while (*p != NULL) {
2493 parent = *p;
2494 iter = rb_entry(parent, struct annotation_line, rb_node);
2495
2496 ret = strcmp(iter->path, al->path);
2497 if (ret == 0) {
2498 for (i = 0; i < al->data_nr; i++) {
2499 iter->data[i].percent_sum += annotation_data__percent(&al->data[i],
2500 percent_type);
2501 }
2502 return;
2503 }
2504
2505 if (ret < 0)
2506 p = &(*p)->rb_left;
2507 else
2508 p = &(*p)->rb_right;
2509 }
2510
2511 for (i = 0; i < al->data_nr; i++) {
2512 al->data[i].percent_sum = annotation_data__percent(&al->data[i],
2513 percent_type);
2514 }
2515
2516 rb_link_node(&al->rb_node, parent, p);
2517 rb_insert_color(&al->rb_node, root);
2518}
2519
2520static int cmp_source_line(struct annotation_line *a, struct annotation_line *b)
2521{
2522 int i;
2523
2524 for (i = 0; i < a->data_nr; i++) {
2525 if (a->data[i].percent_sum == b->data[i].percent_sum)
2526 continue;
2527 return a->data[i].percent_sum > b->data[i].percent_sum;
2528 }
2529
2530 return 0;
2531}
2532
2533static void __resort_source_line(struct rb_root *root, struct annotation_line *al)
2534{
2535 struct annotation_line *iter;
2536 struct rb_node **p = &root->rb_node;
2537 struct rb_node *parent = NULL;
2538
2539 while (*p != NULL) {
2540 parent = *p;
2541 iter = rb_entry(parent, struct annotation_line, rb_node);
2542
2543 if (cmp_source_line(al, iter))
2544 p = &(*p)->rb_left;
2545 else
2546 p = &(*p)->rb_right;
2547 }
2548
2549 rb_link_node(&al->rb_node, parent, p);
2550 rb_insert_color(&al->rb_node, root);
2551}
2552
2553static void resort_source_line(struct rb_root *dest_root, struct rb_root *src_root)
2554{
2555 struct annotation_line *al;
2556 struct rb_node *node;
2557
2558 node = rb_first(src_root);
2559 while (node) {
2560 struct rb_node *next;
2561
2562 al = rb_entry(node, struct annotation_line, rb_node);
2563 next = rb_next(node);
2564 rb_erase(node, src_root);
2565
2566 __resort_source_line(dest_root, al);
2567 node = next;
2568 }
2569}
2570
2571static void print_summary(struct rb_root *root, const char *filename)
2572{
2573 struct annotation_line *al;
2574 struct rb_node *node;
2575
2576 printf("\nSorted summary for file %s\n", filename);
2577 printf("----------------------------------------------\n\n");
2578
2579 if (RB_EMPTY_ROOT(root)) {
2580 printf(" Nothing higher than %1.1f%%\n", MIN_GREEN);
2581 return;
2582 }
2583
2584 node = rb_first(root);
2585 while (node) {
2586 double percent, percent_max = 0.0;
2587 const char *color;
2588 char *path;
2589 int i;
2590
2591 al = rb_entry(node, struct annotation_line, rb_node);
2592 for (i = 0; i < al->data_nr; i++) {
2593 percent = al->data[i].percent_sum;
2594 color = get_percent_color(percent);
2595 color_fprintf(stdout, color, " %7.2f", percent);
2596
2597 if (percent > percent_max)
2598 percent_max = percent;
2599 }
2600
2601 path = al->path;
2602 color = get_percent_color(percent_max);
2603 color_fprintf(stdout, color, " %s\n", path);
2604
2605 node = rb_next(node);
2606 }
2607}
2608
2609static void symbol__annotate_hits(struct symbol *sym, struct evsel *evsel)
2610{
2611 int evidx = evsel->core.idx;
2612 struct annotation *notes = symbol__annotation(sym);
2613 struct sym_hist *h = annotation__histogram(notes, evidx);
2614 u64 len = symbol__size(sym), offset;
2615
2616 for (offset = 0; offset < len; ++offset) {
2617 struct sym_hist_entry *entry;
2618
2619 entry = annotated_source__hist_entry(notes->src, evidx, offset);
2620 if (entry && entry->nr_samples != 0)
2621 printf("%*" PRIx64 ": %" PRIu64 "\n", BITS_PER_LONG / 2,
2622 sym->start + offset, entry->nr_samples);
2623 }
2624 printf("%*s: %" PRIu64 "\n", BITS_PER_LONG / 2, "h->nr_samples", h->nr_samples);
2625}
2626
2627static int annotated_source__addr_fmt_width(struct list_head *lines, u64 start)
2628{
2629 char bf[32];
2630 struct annotation_line *line;
2631
2632 list_for_each_entry_reverse(line, lines, node) {
2633 if (line->offset != -1)
2634 return scnprintf(bf, sizeof(bf), "%" PRIx64, start + line->offset);
2635 }
2636
2637 return 0;
2638}
2639
2640int symbol__annotate_printf(struct map_symbol *ms, struct evsel *evsel)
2641{
2642 struct map *map = ms->map;
2643 struct symbol *sym = ms->sym;
2644 struct dso *dso = map__dso(map);
2645 char *filename;
2646 const char *d_filename;
2647 const char *evsel_name = evsel__name(evsel);
2648 struct annotation *notes = symbol__annotation(sym);
2649 struct sym_hist *h = annotation__histogram(notes, evsel->core.idx);
2650 struct annotation_line *pos, *queue = NULL;
2651 struct annotation_options *opts = &annotate_opts;
2652 u64 start = map__rip_2objdump(map, sym->start);
2653 int printed = 2, queue_len = 0, addr_fmt_width;
2654 int more = 0;
2655 bool context = opts->context;
2656 u64 len;
2657 int width = symbol_conf.show_total_period ? 12 : 8;
2658 int graph_dotted_len;
2659 char buf[512];
2660
2661 filename = strdup(dso->long_name);
2662 if (!filename)
2663 return -ENOMEM;
2664
2665 if (opts->full_path)
2666 d_filename = filename;
2667 else
2668 d_filename = basename(filename);
2669
2670 len = symbol__size(sym);
2671
2672 if (evsel__is_group_event(evsel)) {
2673 width *= evsel->core.nr_members;
2674 evsel__group_desc(evsel, buf, sizeof(buf));
2675 evsel_name = buf;
2676 }
2677
2678 graph_dotted_len = printf(" %-*.*s| Source code & Disassembly of %s for %s (%" PRIu64 " samples, "
2679 "percent: %s)\n",
2680 width, width, symbol_conf.show_total_period ? "Period" :
2681 symbol_conf.show_nr_samples ? "Samples" : "Percent",
2682 d_filename, evsel_name, h->nr_samples,
2683 percent_type_str(opts->percent_type));
2684
2685 printf("%-*.*s----\n",
2686 graph_dotted_len, graph_dotted_len, graph_dotted_line);
2687
2688 if (verbose > 0)
2689 symbol__annotate_hits(sym, evsel);
2690
2691 addr_fmt_width = annotated_source__addr_fmt_width(¬es->src->source, start);
2692
2693 list_for_each_entry(pos, ¬es->src->source, node) {
2694 int err;
2695
2696 if (context && queue == NULL) {
2697 queue = pos;
2698 queue_len = 0;
2699 }
2700
2701 err = annotation_line__print(pos, sym, start, evsel, len,
2702 opts->min_pcnt, printed, opts->max_lines,
2703 queue, addr_fmt_width, opts->percent_type);
2704
2705 switch (err) {
2706 case 0:
2707 ++printed;
2708 if (context) {
2709 printed += queue_len;
2710 queue = NULL;
2711 queue_len = 0;
2712 }
2713 break;
2714 case 1:
2715 /* filtered by max_lines */
2716 ++more;
2717 break;
2718 case -1:
2719 default:
2720 /*
2721 * Filtered by min_pcnt or non IP lines when
2722 * context != 0
2723 */
2724 if (!context)
2725 break;
2726 if (queue_len == context)
2727 queue = list_entry(queue->node.next, typeof(*queue), node);
2728 else
2729 ++queue_len;
2730 break;
2731 }
2732 }
2733
2734 free(filename);
2735
2736 return more;
2737}
2738
2739static void FILE__set_percent_color(void *fp __maybe_unused,
2740 double percent __maybe_unused,
2741 bool current __maybe_unused)
2742{
2743}
2744
2745static int FILE__set_jumps_percent_color(void *fp __maybe_unused,
2746 int nr __maybe_unused, bool current __maybe_unused)
2747{
2748 return 0;
2749}
2750
2751static int FILE__set_color(void *fp __maybe_unused, int color __maybe_unused)
2752{
2753 return 0;
2754}
2755
2756static void FILE__printf(void *fp, const char *fmt, ...)
2757{
2758 va_list args;
2759
2760 va_start(args, fmt);
2761 vfprintf(fp, fmt, args);
2762 va_end(args);
2763}
2764
2765static void FILE__write_graph(void *fp, int graph)
2766{
2767 const char *s;
2768 switch (graph) {
2769
2770 case DARROW_CHAR: s = "↓"; break;
2771 case UARROW_CHAR: s = "↑"; break;
2772 case LARROW_CHAR: s = "←"; break;
2773 case RARROW_CHAR: s = "→"; break;
2774 default: s = "?"; break;
2775 }
2776
2777 fputs(s, fp);
2778}
2779
2780static int symbol__annotate_fprintf2(struct symbol *sym, FILE *fp)
2781{
2782 struct annotation *notes = symbol__annotation(sym);
2783 struct annotation_write_ops wops = {
2784 .first_line = true,
2785 .obj = fp,
2786 .set_color = FILE__set_color,
2787 .set_percent_color = FILE__set_percent_color,
2788 .set_jumps_percent_color = FILE__set_jumps_percent_color,
2789 .printf = FILE__printf,
2790 .write_graph = FILE__write_graph,
2791 };
2792 struct annotation_line *al;
2793
2794 list_for_each_entry(al, ¬es->src->source, node) {
2795 if (annotation_line__filter(al))
2796 continue;
2797 annotation_line__write(al, notes, &wops);
2798 fputc('\n', fp);
2799 wops.first_line = false;
2800 }
2801
2802 return 0;
2803}
2804
2805int map_symbol__annotation_dump(struct map_symbol *ms, struct evsel *evsel)
2806{
2807 const char *ev_name = evsel__name(evsel);
2808 char buf[1024];
2809 char *filename;
2810 int err = -1;
2811 FILE *fp;
2812
2813 if (asprintf(&filename, "%s.annotation", ms->sym->name) < 0)
2814 return -1;
2815
2816 fp = fopen(filename, "w");
2817 if (fp == NULL)
2818 goto out_free_filename;
2819
2820 if (evsel__is_group_event(evsel)) {
2821 evsel__group_desc(evsel, buf, sizeof(buf));
2822 ev_name = buf;
2823 }
2824
2825 fprintf(fp, "%s() %s\nEvent: %s\n\n",
2826 ms->sym->name, map__dso(ms->map)->long_name, ev_name);
2827 symbol__annotate_fprintf2(ms->sym, fp);
2828
2829 fclose(fp);
2830 err = 0;
2831out_free_filename:
2832 free(filename);
2833 return err;
2834}
2835
2836void symbol__annotate_zero_histogram(struct symbol *sym, int evidx)
2837{
2838 struct annotation *notes = symbol__annotation(sym);
2839 struct sym_hist *h = annotation__histogram(notes, evidx);
2840
2841 memset(h, 0, sizeof(*notes->src->histograms) * notes->src->nr_histograms);
2842}
2843
2844void symbol__annotate_decay_histogram(struct symbol *sym, int evidx)
2845{
2846 struct annotation *notes = symbol__annotation(sym);
2847 struct sym_hist *h = annotation__histogram(notes, evidx);
2848 int len = symbol__size(sym), offset;
2849
2850 h->nr_samples = 0;
2851 for (offset = 0; offset < len; ++offset) {
2852 struct sym_hist_entry *entry;
2853
2854 entry = annotated_source__hist_entry(notes->src, evidx, offset);
2855 if (entry == NULL)
2856 continue;
2857
2858 entry->nr_samples = entry->nr_samples * 7 / 8;
2859 h->nr_samples += entry->nr_samples;
2860 }
2861}
2862
2863void annotated_source__purge(struct annotated_source *as)
2864{
2865 struct annotation_line *al, *n;
2866
2867 list_for_each_entry_safe(al, n, &as->source, node) {
2868 list_del_init(&al->node);
2869 disasm_line__free(disasm_line(al));
2870 }
2871}
2872
2873static size_t disasm_line__fprintf(struct disasm_line *dl, FILE *fp)
2874{
2875 size_t printed;
2876
2877 if (dl->al.offset == -1)
2878 return fprintf(fp, "%s\n", dl->al.line);
2879
2880 printed = fprintf(fp, "%#" PRIx64 " %s", dl->al.offset, dl->ins.name);
2881
2882 if (dl->ops.raw[0] != '\0') {
2883 printed += fprintf(fp, "%.*s %s\n", 6 - (int)printed, " ",
2884 dl->ops.raw);
2885 }
2886
2887 return printed + fprintf(fp, "\n");
2888}
2889
2890size_t disasm__fprintf(struct list_head *head, FILE *fp)
2891{
2892 struct disasm_line *pos;
2893 size_t printed = 0;
2894
2895 list_for_each_entry(pos, head, al.node)
2896 printed += disasm_line__fprintf(pos, fp);
2897
2898 return printed;
2899}
2900
2901bool disasm_line__is_valid_local_jump(struct disasm_line *dl, struct symbol *sym)
2902{
2903 if (!dl || !dl->ins.ops || !ins__is_jump(&dl->ins) ||
2904 !disasm_line__has_local_offset(dl) || dl->ops.target.offset < 0 ||
2905 dl->ops.target.offset >= (s64)symbol__size(sym))
2906 return false;
2907
2908 return true;
2909}
2910
2911void annotation__mark_jump_targets(struct annotation *notes, struct symbol *sym)
2912{
2913 u64 offset, size = symbol__size(sym);
2914
2915 /* PLT symbols contain external offsets */
2916 if (strstr(sym->name, "@plt"))
2917 return;
2918
2919 for (offset = 0; offset < size; ++offset) {
2920 struct annotation_line *al = notes->src->offsets[offset];
2921 struct disasm_line *dl;
2922
2923 dl = disasm_line(al);
2924
2925 if (!disasm_line__is_valid_local_jump(dl, sym))
2926 continue;
2927
2928 al = notes->src->offsets[dl->ops.target.offset];
2929
2930 /*
2931 * FIXME: Oops, no jump target? Buggy disassembler? Or do we
2932 * have to adjust to the previous offset?
2933 */
2934 if (al == NULL)
2935 continue;
2936
2937 if (++al->jump_sources > notes->max_jump_sources)
2938 notes->max_jump_sources = al->jump_sources;
2939 }
2940}
2941
2942void annotation__set_offsets(struct annotation *notes, s64 size)
2943{
2944 struct annotation_line *al;
2945 struct annotated_source *src = notes->src;
2946
2947 src->max_line_len = 0;
2948 src->nr_entries = 0;
2949 src->nr_asm_entries = 0;
2950
2951 list_for_each_entry(al, &src->source, node) {
2952 size_t line_len = strlen(al->line);
2953
2954 if (src->max_line_len < line_len)
2955 src->max_line_len = line_len;
2956 al->idx = src->nr_entries++;
2957 if (al->offset != -1) {
2958 al->idx_asm = src->nr_asm_entries++;
2959 /*
2960 * FIXME: short term bandaid to cope with assembly
2961 * routines that comes with labels in the same column
2962 * as the address in objdump, sigh.
2963 *
2964 * E.g. copy_user_generic_unrolled
2965 */
2966 if (al->offset < size)
2967 notes->src->offsets[al->offset] = al;
2968 } else
2969 al->idx_asm = -1;
2970 }
2971}
2972
2973static inline int width_jumps(int n)
2974{
2975 if (n >= 100)
2976 return 5;
2977 if (n / 10)
2978 return 2;
2979 return 1;
2980}
2981
2982static int annotation__max_ins_name(struct annotation *notes)
2983{
2984 int max_name = 0, len;
2985 struct annotation_line *al;
2986
2987 list_for_each_entry(al, ¬es->src->source, node) {
2988 if (al->offset == -1)
2989 continue;
2990
2991 len = strlen(disasm_line(al)->ins.name);
2992 if (max_name < len)
2993 max_name = len;
2994 }
2995
2996 return max_name;
2997}
2998
2999void annotation__init_column_widths(struct annotation *notes, struct symbol *sym)
3000{
3001 notes->widths.addr = notes->widths.target =
3002 notes->widths.min_addr = hex_width(symbol__size(sym));
3003 notes->widths.max_addr = hex_width(sym->end);
3004 notes->widths.jumps = width_jumps(notes->max_jump_sources);
3005 notes->widths.max_ins_name = annotation__max_ins_name(notes);
3006}
3007
3008void annotation__update_column_widths(struct annotation *notes)
3009{
3010 if (annotate_opts.use_offset)
3011 notes->widths.target = notes->widths.min_addr;
3012 else if (annotate_opts.full_addr)
3013 notes->widths.target = BITS_PER_LONG / 4;
3014 else
3015 notes->widths.target = notes->widths.max_addr;
3016
3017 notes->widths.addr = notes->widths.target;
3018
3019 if (annotate_opts.show_nr_jumps)
3020 notes->widths.addr += notes->widths.jumps + 1;
3021}
3022
3023void annotation__toggle_full_addr(struct annotation *notes, struct map_symbol *ms)
3024{
3025 annotate_opts.full_addr = !annotate_opts.full_addr;
3026
3027 if (annotate_opts.full_addr)
3028 notes->start = map__objdump_2mem(ms->map, ms->sym->start);
3029 else
3030 notes->start = map__rip_2objdump(ms->map, ms->sym->start);
3031
3032 annotation__update_column_widths(notes);
3033}
3034
3035static void annotation__calc_lines(struct annotation *notes, struct map_symbol *ms,
3036 struct rb_root *root)
3037{
3038 struct annotation_line *al;
3039 struct rb_root tmp_root = RB_ROOT;
3040
3041 list_for_each_entry(al, ¬es->src->source, node) {
3042 double percent_max = 0.0;
3043 u64 addr;
3044 int i;
3045
3046 for (i = 0; i < al->data_nr; i++) {
3047 double percent;
3048
3049 percent = annotation_data__percent(&al->data[i],
3050 annotate_opts.percent_type);
3051
3052 if (percent > percent_max)
3053 percent_max = percent;
3054 }
3055
3056 if (percent_max <= 0.5)
3057 continue;
3058
3059 addr = map__rip_2objdump(ms->map, ms->sym->start);
3060 al->path = get_srcline(map__dso(ms->map), addr + al->offset, NULL,
3061 false, true, ms->sym->start + al->offset);
3062 insert_source_line(&tmp_root, al);
3063 }
3064
3065 resort_source_line(root, &tmp_root);
3066}
3067
3068static void symbol__calc_lines(struct map_symbol *ms, struct rb_root *root)
3069{
3070 struct annotation *notes = symbol__annotation(ms->sym);
3071
3072 annotation__calc_lines(notes, ms, root);
3073}
3074
3075int symbol__tty_annotate2(struct map_symbol *ms, struct evsel *evsel)
3076{
3077 struct dso *dso = map__dso(ms->map);
3078 struct symbol *sym = ms->sym;
3079 struct rb_root source_line = RB_ROOT;
3080 struct hists *hists = evsel__hists(evsel);
3081 char buf[1024];
3082 int err;
3083
3084 err = symbol__annotate2(ms, evsel, NULL);
3085 if (err) {
3086 char msg[BUFSIZ];
3087
3088 dso->annotate_warned = true;
3089 symbol__strerror_disassemble(ms, err, msg, sizeof(msg));
3090 ui__error("Couldn't annotate %s:\n%s", sym->name, msg);
3091 return -1;
3092 }
3093
3094 if (annotate_opts.print_lines) {
3095 srcline_full_filename = annotate_opts.full_path;
3096 symbol__calc_lines(ms, &source_line);
3097 print_summary(&source_line, dso->long_name);
3098 }
3099
3100 hists__scnprintf_title(hists, buf, sizeof(buf));
3101 fprintf(stdout, "%s, [percent: %s]\n%s() %s\n",
3102 buf, percent_type_str(annotate_opts.percent_type), sym->name,
3103 dso->long_name);
3104 symbol__annotate_fprintf2(sym, stdout);
3105
3106 annotated_source__purge(symbol__annotation(sym)->src);
3107
3108 return 0;
3109}
3110
3111int symbol__tty_annotate(struct map_symbol *ms, struct evsel *evsel)
3112{
3113 struct dso *dso = map__dso(ms->map);
3114 struct symbol *sym = ms->sym;
3115 struct rb_root source_line = RB_ROOT;
3116 int err;
3117
3118 err = symbol__annotate(ms, evsel, NULL);
3119 if (err) {
3120 char msg[BUFSIZ];
3121
3122 dso->annotate_warned = true;
3123 symbol__strerror_disassemble(ms, err, msg, sizeof(msg));
3124 ui__error("Couldn't annotate %s:\n%s", sym->name, msg);
3125 return -1;
3126 }
3127
3128 symbol__calc_percent(sym, evsel);
3129
3130 if (annotate_opts.print_lines) {
3131 srcline_full_filename = annotate_opts.full_path;
3132 symbol__calc_lines(ms, &source_line);
3133 print_summary(&source_line, dso->long_name);
3134 }
3135
3136 symbol__annotate_printf(ms, evsel);
3137
3138 annotated_source__purge(symbol__annotation(sym)->src);
3139
3140 return 0;
3141}
3142
3143bool ui__has_annotation(void)
3144{
3145 return use_browser == 1 && perf_hpp_list.sym;
3146}
3147
3148
3149static double annotation_line__max_percent(struct annotation_line *al,
3150 struct annotation *notes,
3151 unsigned int percent_type)
3152{
3153 double percent_max = 0.0;
3154 int i;
3155
3156 for (i = 0; i < notes->nr_events; i++) {
3157 double percent;
3158
3159 percent = annotation_data__percent(&al->data[i],
3160 percent_type);
3161
3162 if (percent > percent_max)
3163 percent_max = percent;
3164 }
3165
3166 return percent_max;
3167}
3168
3169static void disasm_line__write(struct disasm_line *dl, struct annotation *notes,
3170 void *obj, char *bf, size_t size,
3171 void (*obj__printf)(void *obj, const char *fmt, ...),
3172 void (*obj__write_graph)(void *obj, int graph))
3173{
3174 if (dl->ins.ops && dl->ins.ops->scnprintf) {
3175 if (ins__is_jump(&dl->ins)) {
3176 bool fwd;
3177
3178 if (dl->ops.target.outside)
3179 goto call_like;
3180 fwd = dl->ops.target.offset > dl->al.offset;
3181 obj__write_graph(obj, fwd ? DARROW_CHAR : UARROW_CHAR);
3182 obj__printf(obj, " ");
3183 } else if (ins__is_call(&dl->ins)) {
3184call_like:
3185 obj__write_graph(obj, RARROW_CHAR);
3186 obj__printf(obj, " ");
3187 } else if (ins__is_ret(&dl->ins)) {
3188 obj__write_graph(obj, LARROW_CHAR);
3189 obj__printf(obj, " ");
3190 } else {
3191 obj__printf(obj, " ");
3192 }
3193 } else {
3194 obj__printf(obj, " ");
3195 }
3196
3197 disasm_line__scnprintf(dl, bf, size, !annotate_opts.use_offset, notes->widths.max_ins_name);
3198}
3199
3200static void ipc_coverage_string(char *bf, int size, struct annotation *notes)
3201{
3202 double ipc = 0.0, coverage = 0.0;
3203 struct annotated_branch *branch = annotation__get_branch(notes);
3204
3205 if (branch && branch->hit_cycles)
3206 ipc = branch->hit_insn / ((double)branch->hit_cycles);
3207
3208 if (branch && branch->total_insn) {
3209 coverage = branch->cover_insn * 100.0 /
3210 ((double)branch->total_insn);
3211 }
3212
3213 scnprintf(bf, size, "(Average IPC: %.2f, IPC Coverage: %.1f%%)",
3214 ipc, coverage);
3215}
3216
3217static void __annotation_line__write(struct annotation_line *al, struct annotation *notes,
3218 bool first_line, bool current_entry, bool change_color, int width,
3219 void *obj, unsigned int percent_type,
3220 int (*obj__set_color)(void *obj, int color),
3221 void (*obj__set_percent_color)(void *obj, double percent, bool current),
3222 int (*obj__set_jumps_percent_color)(void *obj, int nr, bool current),
3223 void (*obj__printf)(void *obj, const char *fmt, ...),
3224 void (*obj__write_graph)(void *obj, int graph))
3225
3226{
3227 double percent_max = annotation_line__max_percent(al, notes, percent_type);
3228 int pcnt_width = annotation__pcnt_width(notes),
3229 cycles_width = annotation__cycles_width(notes);
3230 bool show_title = false;
3231 char bf[256];
3232 int printed;
3233
3234 if (first_line && (al->offset == -1 || percent_max == 0.0)) {
3235 if (notes->branch && al->cycles) {
3236 if (al->cycles->ipc == 0.0 && al->cycles->avg == 0)
3237 show_title = true;
3238 } else
3239 show_title = true;
3240 }
3241
3242 if (al->offset != -1 && percent_max != 0.0) {
3243 int i;
3244
3245 for (i = 0; i < notes->nr_events; i++) {
3246 double percent;
3247
3248 percent = annotation_data__percent(&al->data[i], percent_type);
3249
3250 obj__set_percent_color(obj, percent, current_entry);
3251 if (symbol_conf.show_total_period) {
3252 obj__printf(obj, "%11" PRIu64 " ", al->data[i].he.period);
3253 } else if (symbol_conf.show_nr_samples) {
3254 obj__printf(obj, "%6" PRIu64 " ",
3255 al->data[i].he.nr_samples);
3256 } else {
3257 obj__printf(obj, "%6.2f ", percent);
3258 }
3259 }
3260 } else {
3261 obj__set_percent_color(obj, 0, current_entry);
3262
3263 if (!show_title)
3264 obj__printf(obj, "%-*s", pcnt_width, " ");
3265 else {
3266 obj__printf(obj, "%-*s", pcnt_width,
3267 symbol_conf.show_total_period ? "Period" :
3268 symbol_conf.show_nr_samples ? "Samples" : "Percent");
3269 }
3270 }
3271
3272 if (notes->branch) {
3273 if (al->cycles && al->cycles->ipc)
3274 obj__printf(obj, "%*.2f ", ANNOTATION__IPC_WIDTH - 1, al->cycles->ipc);
3275 else if (!show_title)
3276 obj__printf(obj, "%*s", ANNOTATION__IPC_WIDTH, " ");
3277 else
3278 obj__printf(obj, "%*s ", ANNOTATION__IPC_WIDTH - 1, "IPC");
3279
3280 if (!annotate_opts.show_minmax_cycle) {
3281 if (al->cycles && al->cycles->avg)
3282 obj__printf(obj, "%*" PRIu64 " ",
3283 ANNOTATION__CYCLES_WIDTH - 1, al->cycles->avg);
3284 else if (!show_title)
3285 obj__printf(obj, "%*s",
3286 ANNOTATION__CYCLES_WIDTH, " ");
3287 else
3288 obj__printf(obj, "%*s ",
3289 ANNOTATION__CYCLES_WIDTH - 1,
3290 "Cycle");
3291 } else {
3292 if (al->cycles) {
3293 char str[32];
3294
3295 scnprintf(str, sizeof(str),
3296 "%" PRIu64 "(%" PRIu64 "/%" PRIu64 ")",
3297 al->cycles->avg, al->cycles->min,
3298 al->cycles->max);
3299
3300 obj__printf(obj, "%*s ",
3301 ANNOTATION__MINMAX_CYCLES_WIDTH - 1,
3302 str);
3303 } else if (!show_title)
3304 obj__printf(obj, "%*s",
3305 ANNOTATION__MINMAX_CYCLES_WIDTH,
3306 " ");
3307 else
3308 obj__printf(obj, "%*s ",
3309 ANNOTATION__MINMAX_CYCLES_WIDTH - 1,
3310 "Cycle(min/max)");
3311 }
3312
3313 if (show_title && !*al->line) {
3314 ipc_coverage_string(bf, sizeof(bf), notes);
3315 obj__printf(obj, "%*s", ANNOTATION__AVG_IPC_WIDTH, bf);
3316 }
3317 }
3318
3319 obj__printf(obj, " ");
3320
3321 if (!*al->line)
3322 obj__printf(obj, "%-*s", width - pcnt_width - cycles_width, " ");
3323 else if (al->offset == -1) {
3324 if (al->line_nr && annotate_opts.show_linenr)
3325 printed = scnprintf(bf, sizeof(bf), "%-*d ", notes->widths.addr + 1, al->line_nr);
3326 else
3327 printed = scnprintf(bf, sizeof(bf), "%-*s ", notes->widths.addr, " ");
3328 obj__printf(obj, bf);
3329 obj__printf(obj, "%-*s", width - printed - pcnt_width - cycles_width + 1, al->line);
3330 } else {
3331 u64 addr = al->offset;
3332 int color = -1;
3333
3334 if (!annotate_opts.use_offset)
3335 addr += notes->start;
3336
3337 if (!annotate_opts.use_offset) {
3338 printed = scnprintf(bf, sizeof(bf), "%" PRIx64 ": ", addr);
3339 } else {
3340 if (al->jump_sources &&
3341 annotate_opts.offset_level >= ANNOTATION__OFFSET_JUMP_TARGETS) {
3342 if (annotate_opts.show_nr_jumps) {
3343 int prev;
3344 printed = scnprintf(bf, sizeof(bf), "%*d ",
3345 notes->widths.jumps,
3346 al->jump_sources);
3347 prev = obj__set_jumps_percent_color(obj, al->jump_sources,
3348 current_entry);
3349 obj__printf(obj, bf);
3350 obj__set_color(obj, prev);
3351 }
3352print_addr:
3353 printed = scnprintf(bf, sizeof(bf), "%*" PRIx64 ": ",
3354 notes->widths.target, addr);
3355 } else if (ins__is_call(&disasm_line(al)->ins) &&
3356 annotate_opts.offset_level >= ANNOTATION__OFFSET_CALL) {
3357 goto print_addr;
3358 } else if (annotate_opts.offset_level == ANNOTATION__MAX_OFFSET_LEVEL) {
3359 goto print_addr;
3360 } else {
3361 printed = scnprintf(bf, sizeof(bf), "%-*s ",
3362 notes->widths.addr, " ");
3363 }
3364 }
3365
3366 if (change_color)
3367 color = obj__set_color(obj, HE_COLORSET_ADDR);
3368 obj__printf(obj, bf);
3369 if (change_color)
3370 obj__set_color(obj, color);
3371
3372 disasm_line__write(disasm_line(al), notes, obj, bf, sizeof(bf), obj__printf, obj__write_graph);
3373
3374 obj__printf(obj, "%-*s", width - pcnt_width - cycles_width - 3 - printed, bf);
3375 }
3376
3377}
3378
3379void annotation_line__write(struct annotation_line *al, struct annotation *notes,
3380 struct annotation_write_ops *wops)
3381{
3382 __annotation_line__write(al, notes, wops->first_line, wops->current_entry,
3383 wops->change_color, wops->width, wops->obj,
3384 annotate_opts.percent_type,
3385 wops->set_color, wops->set_percent_color,
3386 wops->set_jumps_percent_color, wops->printf,
3387 wops->write_graph);
3388}
3389
3390int symbol__annotate2(struct map_symbol *ms, struct evsel *evsel,
3391 struct arch **parch)
3392{
3393 struct symbol *sym = ms->sym;
3394 struct annotation *notes = symbol__annotation(sym);
3395 size_t size = symbol__size(sym);
3396 int nr_pcnt = 1, err;
3397
3398 notes->src->offsets = zalloc(size * sizeof(struct annotation_line *));
3399 if (notes->src->offsets == NULL)
3400 return ENOMEM;
3401
3402 if (evsel__is_group_event(evsel))
3403 nr_pcnt = evsel->core.nr_members;
3404
3405 err = symbol__annotate(ms, evsel, parch);
3406 if (err)
3407 goto out_free_offsets;
3408
3409 symbol__calc_percent(sym, evsel);
3410
3411 annotation__set_offsets(notes, size);
3412 annotation__mark_jump_targets(notes, sym);
3413
3414 err = annotation__compute_ipc(notes, size);
3415 if (err)
3416 goto out_free_offsets;
3417
3418 annotation__init_column_widths(notes, sym);
3419 notes->nr_events = nr_pcnt;
3420
3421 annotation__update_column_widths(notes);
3422 sym->annotate2 = 1;
3423
3424 return 0;
3425
3426out_free_offsets:
3427 zfree(¬es->src->offsets);
3428 return err;
3429}
3430
3431static int annotation__config(const char *var, const char *value, void *data)
3432{
3433 struct annotation_options *opt = data;
3434
3435 if (!strstarts(var, "annotate."))
3436 return 0;
3437
3438 if (!strcmp(var, "annotate.offset_level")) {
3439 perf_config_u8(&opt->offset_level, "offset_level", value);
3440
3441 if (opt->offset_level > ANNOTATION__MAX_OFFSET_LEVEL)
3442 opt->offset_level = ANNOTATION__MAX_OFFSET_LEVEL;
3443 else if (opt->offset_level < ANNOTATION__MIN_OFFSET_LEVEL)
3444 opt->offset_level = ANNOTATION__MIN_OFFSET_LEVEL;
3445 } else if (!strcmp(var, "annotate.hide_src_code")) {
3446 opt->hide_src_code = perf_config_bool("hide_src_code", value);
3447 } else if (!strcmp(var, "annotate.jump_arrows")) {
3448 opt->jump_arrows = perf_config_bool("jump_arrows", value);
3449 } else if (!strcmp(var, "annotate.show_linenr")) {
3450 opt->show_linenr = perf_config_bool("show_linenr", value);
3451 } else if (!strcmp(var, "annotate.show_nr_jumps")) {
3452 opt->show_nr_jumps = perf_config_bool("show_nr_jumps", value);
3453 } else if (!strcmp(var, "annotate.show_nr_samples")) {
3454 symbol_conf.show_nr_samples = perf_config_bool("show_nr_samples",
3455 value);
3456 } else if (!strcmp(var, "annotate.show_total_period")) {
3457 symbol_conf.show_total_period = perf_config_bool("show_total_period",
3458 value);
3459 } else if (!strcmp(var, "annotate.use_offset")) {
3460 opt->use_offset = perf_config_bool("use_offset", value);
3461 } else if (!strcmp(var, "annotate.disassembler_style")) {
3462 opt->disassembler_style = strdup(value);
3463 if (!opt->disassembler_style) {
3464 pr_err("Not enough memory for annotate.disassembler_style\n");
3465 return -1;
3466 }
3467 } else if (!strcmp(var, "annotate.objdump")) {
3468 opt->objdump_path = strdup(value);
3469 if (!opt->objdump_path) {
3470 pr_err("Not enough memory for annotate.objdump\n");
3471 return -1;
3472 }
3473 } else if (!strcmp(var, "annotate.addr2line")) {
3474 symbol_conf.addr2line_path = strdup(value);
3475 if (!symbol_conf.addr2line_path) {
3476 pr_err("Not enough memory for annotate.addr2line\n");
3477 return -1;
3478 }
3479 } else if (!strcmp(var, "annotate.demangle")) {
3480 symbol_conf.demangle = perf_config_bool("demangle", value);
3481 } else if (!strcmp(var, "annotate.demangle_kernel")) {
3482 symbol_conf.demangle_kernel = perf_config_bool("demangle_kernel", value);
3483 } else {
3484 pr_debug("%s variable unknown, ignoring...", var);
3485 }
3486
3487 return 0;
3488}
3489
3490void annotation_options__init(void)
3491{
3492 struct annotation_options *opt = &annotate_opts;
3493
3494 memset(opt, 0, sizeof(*opt));
3495
3496 /* Default values. */
3497 opt->use_offset = true;
3498 opt->jump_arrows = true;
3499 opt->annotate_src = true;
3500 opt->offset_level = ANNOTATION__OFFSET_JUMP_TARGETS;
3501 opt->percent_type = PERCENT_PERIOD_LOCAL;
3502}
3503
3504void annotation_options__exit(void)
3505{
3506 zfree(&annotate_opts.disassembler_style);
3507 zfree(&annotate_opts.objdump_path);
3508}
3509
3510void annotation_config__init(void)
3511{
3512 perf_config(annotation__config, &annotate_opts);
3513}
3514
3515static unsigned int parse_percent_type(char *str1, char *str2)
3516{
3517 unsigned int type = (unsigned int) -1;
3518
3519 if (!strcmp("period", str1)) {
3520 if (!strcmp("local", str2))
3521 type = PERCENT_PERIOD_LOCAL;
3522 else if (!strcmp("global", str2))
3523 type = PERCENT_PERIOD_GLOBAL;
3524 }
3525
3526 if (!strcmp("hits", str1)) {
3527 if (!strcmp("local", str2))
3528 type = PERCENT_HITS_LOCAL;
3529 else if (!strcmp("global", str2))
3530 type = PERCENT_HITS_GLOBAL;
3531 }
3532
3533 return type;
3534}
3535
3536int annotate_parse_percent_type(const struct option *opt __maybe_unused, const char *_str,
3537 int unset __maybe_unused)
3538{
3539 unsigned int type;
3540 char *str1, *str2;
3541 int err = -1;
3542
3543 str1 = strdup(_str);
3544 if (!str1)
3545 return -ENOMEM;
3546
3547 str2 = strchr(str1, '-');
3548 if (!str2)
3549 goto out;
3550
3551 *str2++ = 0;
3552
3553 type = parse_percent_type(str1, str2);
3554 if (type == (unsigned int) -1)
3555 type = parse_percent_type(str2, str1);
3556 if (type != (unsigned int) -1) {
3557 annotate_opts.percent_type = type;
3558 err = 0;
3559 }
3560
3561out:
3562 free(str1);
3563 return err;
3564}
3565
3566int annotate_check_args(void)
3567{
3568 struct annotation_options *args = &annotate_opts;
3569
3570 if (args->prefix_strip && !args->prefix) {
3571 pr_err("--prefix-strip requires --prefix\n");
3572 return -1;
3573 }
3574 return 0;
3575}
3576
3577/*
3578 * Get register number and access offset from the given instruction.
3579 * It assumes AT&T x86 asm format like OFFSET(REG). Maybe it needs
3580 * to revisit the format when it handles different architecture.
3581 * Fills @reg and @offset when return 0.
3582 */
3583static int extract_reg_offset(struct arch *arch, const char *str,
3584 struct annotated_op_loc *op_loc)
3585{
3586 char *p;
3587 char *regname;
3588
3589 if (arch->objdump.register_char == 0)
3590 return -1;
3591
3592 /*
3593 * It should start from offset, but it's possible to skip 0
3594 * in the asm. So 0(%rax) should be same as (%rax).
3595 *
3596 * However, it also start with a segment select register like
3597 * %gs:0x18(%rbx). In that case it should skip the part.
3598 */
3599 if (*str == arch->objdump.register_char) {
3600 while (*str && !isdigit(*str) &&
3601 *str != arch->objdump.memory_ref_char)
3602 str++;
3603 }
3604
3605 op_loc->offset = strtol(str, &p, 0);
3606
3607 p = strchr(p, arch->objdump.register_char);
3608 if (p == NULL)
3609 return -1;
3610
3611 regname = strdup(p);
3612 if (regname == NULL)
3613 return -1;
3614
3615 op_loc->reg1 = get_dwarf_regnum(regname, 0);
3616 free(regname);
3617
3618 /* Get the second register */
3619 if (op_loc->multi_regs) {
3620 p = strchr(p + 1, arch->objdump.register_char);
3621 if (p == NULL)
3622 return -1;
3623
3624 regname = strdup(p);
3625 if (regname == NULL)
3626 return -1;
3627
3628 op_loc->reg2 = get_dwarf_regnum(regname, 0);
3629 free(regname);
3630 }
3631 return 0;
3632}
3633
3634/**
3635 * annotate_get_insn_location - Get location of instruction
3636 * @arch: the architecture info
3637 * @dl: the target instruction
3638 * @loc: a buffer to save the data
3639 *
3640 * Get detailed location info (register and offset) in the instruction.
3641 * It needs both source and target operand and whether it accesses a
3642 * memory location. The offset field is meaningful only when the
3643 * corresponding mem flag is set. The reg2 field is meaningful only
3644 * when multi_regs flag is set.
3645 *
3646 * Some examples on x86:
3647 *
3648 * mov (%rax), %rcx # src_reg1 = rax, src_mem = 1, src_offset = 0
3649 * # dst_reg1 = rcx, dst_mem = 0
3650 *
3651 * mov 0x18, %r8 # src_reg1 = -1, src_mem = 0
3652 * # dst_reg1 = r8, dst_mem = 0
3653 *
3654 * mov %rsi, 8(%rbx,%rcx,4) # src_reg1 = rsi, src_mem = 0, dst_multi_regs = 0
3655 * # dst_reg1 = rbx, dst_reg2 = rcx, dst_mem = 1
3656 * # dst_multi_regs = 1, dst_offset = 8
3657 */
3658int annotate_get_insn_location(struct arch *arch, struct disasm_line *dl,
3659 struct annotated_insn_loc *loc)
3660{
3661 struct ins_operands *ops;
3662 struct annotated_op_loc *op_loc;
3663 int i;
3664
3665 if (!strcmp(dl->ins.name, "lock"))
3666 ops = dl->ops.locked.ops;
3667 else
3668 ops = &dl->ops;
3669
3670 if (ops == NULL)
3671 return -1;
3672
3673 memset(loc, 0, sizeof(*loc));
3674
3675 for_each_insn_op_loc(loc, i, op_loc) {
3676 const char *insn_str = ops->source.raw;
3677 bool multi_regs = ops->source.multi_regs;
3678
3679 if (i == INSN_OP_TARGET) {
3680 insn_str = ops->target.raw;
3681 multi_regs = ops->target.multi_regs;
3682 }
3683
3684 /* Invalidate the register by default */
3685 op_loc->reg1 = -1;
3686 op_loc->reg2 = -1;
3687
3688 if (insn_str == NULL)
3689 continue;
3690
3691 if (strchr(insn_str, arch->objdump.memory_ref_char)) {
3692 op_loc->mem_ref = true;
3693 op_loc->multi_regs = multi_regs;
3694 extract_reg_offset(arch, insn_str, op_loc);
3695 } else {
3696 char *s = strdup(insn_str);
3697
3698 if (s) {
3699 op_loc->reg1 = get_dwarf_regnum(s, 0);
3700 free(s);
3701 }
3702 }
3703 }
3704
3705 return 0;
3706}
3707
3708static void symbol__ensure_annotate(struct map_symbol *ms, struct evsel *evsel)
3709{
3710 struct disasm_line *dl, *tmp_dl;
3711 struct annotation *notes;
3712
3713 notes = symbol__annotation(ms->sym);
3714 if (!list_empty(¬es->src->source))
3715 return;
3716
3717 if (symbol__annotate(ms, evsel, NULL) < 0)
3718 return;
3719
3720 /* remove non-insn disasm lines for simplicity */
3721 list_for_each_entry_safe(dl, tmp_dl, ¬es->src->source, al.node) {
3722 if (dl->al.offset == -1) {
3723 list_del(&dl->al.node);
3724 free(dl);
3725 }
3726 }
3727}
3728
3729static struct disasm_line *find_disasm_line(struct symbol *sym, u64 ip)
3730{
3731 struct disasm_line *dl;
3732 struct annotation *notes;
3733
3734 notes = symbol__annotation(sym);
3735
3736 list_for_each_entry(dl, ¬es->src->source, al.node) {
3737 if (sym->start + dl->al.offset == ip) {
3738 /*
3739 * llvm-objdump places "lock" in a separate line and
3740 * in that case, we want to get the next line.
3741 */
3742 if (!strcmp(dl->ins.name, "lock") && *dl->ops.raw == '\0') {
3743 ip++;
3744 continue;
3745 }
3746 return dl;
3747 }
3748 }
3749 return NULL;
3750}
3751
3752static struct annotated_item_stat *annotate_data_stat(struct list_head *head,
3753 const char *name)
3754{
3755 struct annotated_item_stat *istat;
3756
3757 list_for_each_entry(istat, head, list) {
3758 if (!strcmp(istat->name, name))
3759 return istat;
3760 }
3761
3762 istat = zalloc(sizeof(*istat));
3763 if (istat == NULL)
3764 return NULL;
3765
3766 istat->name = strdup(name);
3767 if (istat->name == NULL) {
3768 free(istat);
3769 return NULL;
3770 }
3771
3772 list_add_tail(&istat->list, head);
3773 return istat;
3774}
3775
3776static bool is_stack_operation(struct arch *arch, struct disasm_line *dl)
3777{
3778 if (arch__is(arch, "x86")) {
3779 if (!strncmp(dl->ins.name, "push", 4) ||
3780 !strncmp(dl->ins.name, "pop", 3) ||
3781 !strncmp(dl->ins.name, "ret", 3))
3782 return true;
3783 }
3784
3785 return false;
3786}
3787
3788u64 annotate_calc_pcrel(struct map_symbol *ms, u64 ip, int offset,
3789 struct disasm_line *dl)
3790{
3791 struct annotation *notes;
3792 struct disasm_line *next;
3793 u64 addr;
3794
3795 notes = symbol__annotation(ms->sym);
3796 /*
3797 * PC-relative addressing starts from the next instruction address
3798 * But the IP is for the current instruction. Since disasm_line
3799 * doesn't have the instruction size, calculate it using the next
3800 * disasm_line. If it's the last one, we can use symbol's end
3801 * address directly.
3802 */
3803 if (&dl->al.node == notes->src->source.prev)
3804 addr = ms->sym->end + offset;
3805 else {
3806 next = list_next_entry(dl, al.node);
3807 addr = ip + (next->al.offset - dl->al.offset) + offset;
3808 }
3809 return map__rip_2objdump(ms->map, addr);
3810}
3811
3812/**
3813 * hist_entry__get_data_type - find data type for given hist entry
3814 * @he: hist entry
3815 *
3816 * This function first annotates the instruction at @he->ip and extracts
3817 * register and offset info from it. Then it searches the DWARF debug
3818 * info to get a variable and type information using the address, register,
3819 * and offset.
3820 */
3821struct annotated_data_type *hist_entry__get_data_type(struct hist_entry *he)
3822{
3823 struct map_symbol *ms = &he->ms;
3824 struct evsel *evsel = hists_to_evsel(he->hists);
3825 struct arch *arch;
3826 struct disasm_line *dl;
3827 struct annotated_insn_loc loc;
3828 struct annotated_op_loc *op_loc;
3829 struct annotated_data_type *mem_type;
3830 struct annotated_item_stat *istat;
3831 u64 ip = he->ip, addr = 0;
3832 const char *var_name = NULL;
3833 int var_offset;
3834 int i;
3835
3836 ann_data_stat.total++;
3837
3838 if (ms->map == NULL || ms->sym == NULL) {
3839 ann_data_stat.no_sym++;
3840 return NULL;
3841 }
3842
3843 if (!symbol_conf.init_annotation) {
3844 ann_data_stat.no_sym++;
3845 return NULL;
3846 }
3847
3848 if (evsel__get_arch(evsel, &arch) < 0) {
3849 ann_data_stat.no_insn++;
3850 return NULL;
3851 }
3852
3853 /* Make sure it runs objdump to get disasm of the function */
3854 symbol__ensure_annotate(ms, evsel);
3855
3856 /*
3857 * Get a disasm to extract the location from the insn.
3858 * This is too slow...
3859 */
3860 dl = find_disasm_line(ms->sym, ip);
3861 if (dl == NULL) {
3862 ann_data_stat.no_insn++;
3863 return NULL;
3864 }
3865
3866retry:
3867 istat = annotate_data_stat(&ann_insn_stat, dl->ins.name);
3868 if (istat == NULL) {
3869 ann_data_stat.no_insn++;
3870 return NULL;
3871 }
3872
3873 if (annotate_get_insn_location(arch, dl, &loc) < 0) {
3874 ann_data_stat.no_insn_ops++;
3875 istat->bad++;
3876 return NULL;
3877 }
3878
3879 if (is_stack_operation(arch, dl)) {
3880 istat->good++;
3881 he->mem_type_off = 0;
3882 return &stackop_type;
3883 }
3884
3885 for_each_insn_op_loc(&loc, i, op_loc) {
3886 if (!op_loc->mem_ref)
3887 continue;
3888
3889 /* Recalculate IP because of LOCK prefix or insn fusion */
3890 ip = ms->sym->start + dl->al.offset;
3891
3892 var_offset = op_loc->offset;
3893
3894 /* PC-relative addressing */
3895 if (op_loc->reg1 == DWARF_REG_PC) {
3896 struct addr_location al;
3897 struct symbol *var;
3898 u64 map_addr;
3899
3900 addr = annotate_calc_pcrel(ms, ip, op_loc->offset, dl);
3901 /* Kernel symbols might be relocated */
3902 map_addr = addr + map__reloc(ms->map);
3903
3904 addr_location__init(&al);
3905 var = thread__find_symbol_fb(he->thread, he->cpumode,
3906 map_addr, &al);
3907 if (var) {
3908 var_name = var->name;
3909 /* Calculate type offset from the start of variable */
3910 var_offset = map_addr - map__unmap_ip(al.map, var->start);
3911 }
3912 addr_location__exit(&al);
3913 }
3914
3915 mem_type = find_data_type(ms, ip, op_loc, addr, var_name);
3916 if (mem_type)
3917 istat->good++;
3918 else
3919 istat->bad++;
3920
3921 if (mem_type && var_name)
3922 op_loc->offset = var_offset;
3923
3924 if (symbol_conf.annotate_data_sample) {
3925 annotated_data_type__update_samples(mem_type, evsel,
3926 op_loc->offset,
3927 he->stat.nr_events,
3928 he->stat.period);
3929 }
3930 he->mem_type_off = op_loc->offset;
3931 return mem_type;
3932 }
3933
3934 /*
3935 * Some instructions can be fused and the actual memory access came
3936 * from the previous instruction.
3937 */
3938 if (dl->al.offset > 0) {
3939 struct disasm_line *prev_dl;
3940
3941 prev_dl = list_prev_entry(dl, al.node);
3942 if (ins__is_fused(arch, prev_dl->ins.name, dl->ins.name)) {
3943 dl = prev_dl;
3944 goto retry;
3945 }
3946 }
3947
3948 ann_data_stat.no_mem_ops++;
3949 istat->bad++;
3950 return NULL;
3951}