Loading...
1// SPDX-License-Identifier: GPL-2.0
2#include <errno.h>
3#include <linux/kernel.h>
4#include <linux/types.h>
5#include <inttypes.h>
6#include <stdlib.h>
7#include <unistd.h>
8#include <stdio.h>
9#include <string.h>
10#include <sys/param.h>
11#include <perf/cpumap.h>
12#include <perf/evlist.h>
13#include <perf/mmap.h>
14
15#include "debug.h"
16#include "dso.h"
17#include "env.h"
18#include "parse-events.h"
19#include "evlist.h"
20#include "evsel.h"
21#include "thread_map.h"
22#include "machine.h"
23#include "map.h"
24#include "symbol.h"
25#include "event.h"
26#include "record.h"
27#include "util/mmap.h"
28#include "util/string2.h"
29#include "util/synthetic-events.h"
30#include "util/util.h"
31#include "thread.h"
32
33#include "tests.h"
34
35#include <linux/ctype.h>
36
37#define BUFSZ 1024
38#define READLEN 128
39
40struct state {
41 u64 done[1024];
42 size_t done_cnt;
43};
44
45static size_t read_objdump_chunk(const char **line, unsigned char **buf,
46 size_t *buf_len)
47{
48 size_t bytes_read = 0;
49 unsigned char *chunk_start = *buf;
50
51 /* Read bytes */
52 while (*buf_len > 0) {
53 char c1, c2;
54
55 /* Get 2 hex digits */
56 c1 = *(*line)++;
57 if (!isxdigit(c1))
58 break;
59 c2 = *(*line)++;
60 if (!isxdigit(c2))
61 break;
62
63 /* Store byte and advance buf */
64 **buf = (hex(c1) << 4) | hex(c2);
65 (*buf)++;
66 (*buf_len)--;
67 bytes_read++;
68
69 /* End of chunk? */
70 if (isspace(**line))
71 break;
72 }
73
74 /*
75 * objdump will display raw insn as LE if code endian
76 * is LE and bytes_per_chunk > 1. In that case reverse
77 * the chunk we just read.
78 *
79 * see disassemble_bytes() at binutils/objdump.c for details
80 * how objdump chooses display endian)
81 */
82 if (bytes_read > 1 && !host_is_bigendian()) {
83 unsigned char *chunk_end = chunk_start + bytes_read - 1;
84 unsigned char tmp;
85
86 while (chunk_start < chunk_end) {
87 tmp = *chunk_start;
88 *chunk_start = *chunk_end;
89 *chunk_end = tmp;
90 chunk_start++;
91 chunk_end--;
92 }
93 }
94
95 return bytes_read;
96}
97
98static size_t read_objdump_line(const char *line, unsigned char *buf,
99 size_t buf_len)
100{
101 const char *p;
102 size_t ret, bytes_read = 0;
103
104 /* Skip to a colon */
105 p = strchr(line, ':');
106 if (!p)
107 return 0;
108 p++;
109
110 /* Skip initial spaces */
111 while (*p) {
112 if (!isspace(*p))
113 break;
114 p++;
115 }
116
117 do {
118 ret = read_objdump_chunk(&p, &buf, &buf_len);
119 bytes_read += ret;
120 p++;
121 } while (ret > 0);
122
123 /* return number of successfully read bytes */
124 return bytes_read;
125}
126
127static int read_objdump_output(FILE *f, void *buf, size_t *len, u64 start_addr)
128{
129 char *line = NULL;
130 size_t line_len, off_last = 0;
131 ssize_t ret;
132 int err = 0;
133 u64 addr, last_addr = start_addr;
134
135 while (off_last < *len) {
136 size_t off, read_bytes, written_bytes;
137 unsigned char tmp[BUFSZ];
138
139 ret = getline(&line, &line_len, f);
140 if (feof(f))
141 break;
142 if (ret < 0) {
143 pr_debug("getline failed\n");
144 err = -1;
145 break;
146 }
147
148 /* read objdump data into temporary buffer */
149 read_bytes = read_objdump_line(line, tmp, sizeof(tmp));
150 if (!read_bytes)
151 continue;
152
153 if (sscanf(line, "%"PRIx64, &addr) != 1)
154 continue;
155 if (addr < last_addr) {
156 pr_debug("addr going backwards, read beyond section?\n");
157 break;
158 }
159 last_addr = addr;
160
161 /* copy it from temporary buffer to 'buf' according
162 * to address on current objdump line */
163 off = addr - start_addr;
164 if (off >= *len)
165 break;
166 written_bytes = MIN(read_bytes, *len - off);
167 memcpy(buf + off, tmp, written_bytes);
168 off_last = off + written_bytes;
169 }
170
171 /* len returns number of bytes that could not be read */
172 *len -= off_last;
173
174 free(line);
175
176 return err;
177}
178
179static int read_via_objdump(const char *filename, u64 addr, void *buf,
180 size_t len)
181{
182 char cmd[PATH_MAX * 2];
183 const char *fmt;
184 FILE *f;
185 int ret;
186
187 fmt = "%s -z -d --start-address=0x%"PRIx64" --stop-address=0x%"PRIx64" %s";
188 ret = snprintf(cmd, sizeof(cmd), fmt, test_objdump_path, addr, addr + len,
189 filename);
190 if (ret <= 0 || (size_t)ret >= sizeof(cmd))
191 return -1;
192
193 pr_debug("Objdump command is: %s\n", cmd);
194
195 /* Ignore objdump errors */
196 strcat(cmd, " 2>/dev/null");
197
198 f = popen(cmd, "r");
199 if (!f) {
200 pr_debug("popen failed\n");
201 return -1;
202 }
203
204 ret = read_objdump_output(f, buf, &len, addr);
205 if (len) {
206 pr_debug("objdump read too few bytes: %zd\n", len);
207 if (!ret)
208 ret = len;
209 }
210
211 pclose(f);
212
213 return ret;
214}
215
216static void dump_buf(unsigned char *buf, size_t len)
217{
218 size_t i;
219
220 for (i = 0; i < len; i++) {
221 pr_debug("0x%02x ", buf[i]);
222 if (i % 16 == 15)
223 pr_debug("\n");
224 }
225 pr_debug("\n");
226}
227
228static int read_object_code(u64 addr, size_t len, u8 cpumode,
229 struct thread *thread, struct state *state)
230{
231 struct addr_location al;
232 unsigned char buf1[BUFSZ] = {0};
233 unsigned char buf2[BUFSZ] = {0};
234 size_t ret_len;
235 u64 objdump_addr;
236 const char *objdump_name;
237 char decomp_name[KMOD_DECOMP_LEN];
238 bool decomp = false;
239 int ret, err = 0;
240 struct dso *dso;
241
242 pr_debug("Reading object code for memory address: %#"PRIx64"\n", addr);
243
244 addr_location__init(&al);
245 if (!thread__find_map(thread, cpumode, addr, &al) || !map__dso(al.map)) {
246 if (cpumode == PERF_RECORD_MISC_HYPERVISOR) {
247 pr_debug("Hypervisor address can not be resolved - skipping\n");
248 goto out;
249 }
250
251 pr_debug("thread__find_map failed\n");
252 err = -1;
253 goto out;
254 }
255 dso = map__dso(al.map);
256 pr_debug("File is: %s\n", dso->long_name);
257
258 if (dso->symtab_type == DSO_BINARY_TYPE__KALLSYMS && !dso__is_kcore(dso)) {
259 pr_debug("Unexpected kernel address - skipping\n");
260 goto out;
261 }
262
263 pr_debug("On file address is: %#"PRIx64"\n", al.addr);
264
265 if (len > BUFSZ)
266 len = BUFSZ;
267
268 /* Do not go off the map */
269 if (addr + len > map__end(al.map))
270 len = map__end(al.map) - addr;
271
272 /*
273 * Some architectures (ex: powerpc) have stubs (trampolines) in kernel
274 * modules to manage long jumps. Check if the ip offset falls in stubs
275 * sections for kernel modules. And skip module address after text end
276 */
277 if (dso->is_kmod && al.addr > dso->text_end) {
278 pr_debug("skipping the module address %#"PRIx64" after text end\n", al.addr);
279 goto out;
280 }
281
282 /* Read the object code using perf */
283 ret_len = dso__data_read_offset(dso, maps__machine(thread__maps(thread)),
284 al.addr, buf1, len);
285 if (ret_len != len) {
286 pr_debug("dso__data_read_offset failed\n");
287 err = -1;
288 goto out;
289 }
290
291 /*
292 * Converting addresses for use by objdump requires more information.
293 * map__load() does that. See map__rip_2objdump() for details.
294 */
295 if (map__load(al.map)) {
296 err = -1;
297 goto out;
298 }
299
300 /* objdump struggles with kcore - try each map only once */
301 if (dso__is_kcore(dso)) {
302 size_t d;
303
304 for (d = 0; d < state->done_cnt; d++) {
305 if (state->done[d] == map__start(al.map)) {
306 pr_debug("kcore map tested already");
307 pr_debug(" - skipping\n");
308 goto out;
309 }
310 }
311 if (state->done_cnt >= ARRAY_SIZE(state->done)) {
312 pr_debug("Too many kcore maps - skipping\n");
313 goto out;
314 }
315 state->done[state->done_cnt++] = map__start(al.map);
316 }
317
318 objdump_name = dso->long_name;
319 if (dso__needs_decompress(dso)) {
320 if (dso__decompress_kmodule_path(dso, objdump_name,
321 decomp_name,
322 sizeof(decomp_name)) < 0) {
323 pr_debug("decompression failed\n");
324 err = -1;
325 goto out;
326 }
327
328 decomp = true;
329 objdump_name = decomp_name;
330 }
331
332 /* Read the object code using objdump */
333 objdump_addr = map__rip_2objdump(al.map, al.addr);
334 ret = read_via_objdump(objdump_name, objdump_addr, buf2, len);
335
336 if (decomp)
337 unlink(objdump_name);
338
339 if (ret > 0) {
340 /*
341 * The kernel maps are inaccurate - assume objdump is right in
342 * that case.
343 */
344 if (cpumode == PERF_RECORD_MISC_KERNEL ||
345 cpumode == PERF_RECORD_MISC_GUEST_KERNEL) {
346 len -= ret;
347 if (len) {
348 pr_debug("Reducing len to %zu\n", len);
349 } else if (dso__is_kcore(dso)) {
350 /*
351 * objdump cannot handle very large segments
352 * that may be found in kcore.
353 */
354 pr_debug("objdump failed for kcore");
355 pr_debug(" - skipping\n");
356 } else {
357 err = -1;
358 }
359 goto out;
360 }
361 }
362 if (ret < 0) {
363 pr_debug("read_via_objdump failed\n");
364 err = -1;
365 goto out;
366 }
367
368 /* The results should be identical */
369 if (memcmp(buf1, buf2, len)) {
370 pr_debug("Bytes read differ from those read by objdump\n");
371 pr_debug("buf1 (dso):\n");
372 dump_buf(buf1, len);
373 pr_debug("buf2 (objdump):\n");
374 dump_buf(buf2, len);
375 err = -1;
376 goto out;
377 }
378 pr_debug("Bytes read match those read by objdump\n");
379out:
380 addr_location__exit(&al);
381 return err;
382}
383
384static int process_sample_event(struct machine *machine,
385 struct evlist *evlist,
386 union perf_event *event, struct state *state)
387{
388 struct perf_sample sample;
389 struct thread *thread;
390 int ret;
391
392 if (evlist__parse_sample(evlist, event, &sample)) {
393 pr_debug("evlist__parse_sample failed\n");
394 return -1;
395 }
396
397 thread = machine__findnew_thread(machine, sample.pid, sample.tid);
398 if (!thread) {
399 pr_debug("machine__findnew_thread failed\n");
400 return -1;
401 }
402
403 ret = read_object_code(sample.ip, READLEN, sample.cpumode, thread, state);
404 thread__put(thread);
405 return ret;
406}
407
408static int process_event(struct machine *machine, struct evlist *evlist,
409 union perf_event *event, struct state *state)
410{
411 if (event->header.type == PERF_RECORD_SAMPLE)
412 return process_sample_event(machine, evlist, event, state);
413
414 if (event->header.type == PERF_RECORD_THROTTLE ||
415 event->header.type == PERF_RECORD_UNTHROTTLE)
416 return 0;
417
418 if (event->header.type < PERF_RECORD_MAX) {
419 int ret;
420
421 ret = machine__process_event(machine, event, NULL);
422 if (ret < 0)
423 pr_debug("machine__process_event failed, event type %u\n",
424 event->header.type);
425 return ret;
426 }
427
428 return 0;
429}
430
431static int process_events(struct machine *machine, struct evlist *evlist,
432 struct state *state)
433{
434 union perf_event *event;
435 struct mmap *md;
436 int i, ret;
437
438 for (i = 0; i < evlist->core.nr_mmaps; i++) {
439 md = &evlist->mmap[i];
440 if (perf_mmap__read_init(&md->core) < 0)
441 continue;
442
443 while ((event = perf_mmap__read_event(&md->core)) != NULL) {
444 ret = process_event(machine, evlist, event, state);
445 perf_mmap__consume(&md->core);
446 if (ret < 0)
447 return ret;
448 }
449 perf_mmap__read_done(&md->core);
450 }
451 return 0;
452}
453
454static int comp(const void *a, const void *b)
455{
456 return *(int *)a - *(int *)b;
457}
458
459static void do_sort_something(void)
460{
461 int buf[40960], i;
462
463 for (i = 0; i < (int)ARRAY_SIZE(buf); i++)
464 buf[i] = ARRAY_SIZE(buf) - i - 1;
465
466 qsort(buf, ARRAY_SIZE(buf), sizeof(int), comp);
467
468 for (i = 0; i < (int)ARRAY_SIZE(buf); i++) {
469 if (buf[i] != i) {
470 pr_debug("qsort failed\n");
471 break;
472 }
473 }
474}
475
476static void sort_something(void)
477{
478 int i;
479
480 for (i = 0; i < 10; i++)
481 do_sort_something();
482}
483
484static void syscall_something(void)
485{
486 int pipefd[2];
487 int i;
488
489 for (i = 0; i < 1000; i++) {
490 if (pipe(pipefd) < 0) {
491 pr_debug("pipe failed\n");
492 break;
493 }
494 close(pipefd[1]);
495 close(pipefd[0]);
496 }
497}
498
499static void fs_something(void)
500{
501 const char *test_file_name = "temp-perf-code-reading-test-file--";
502 FILE *f;
503 int i;
504
505 for (i = 0; i < 1000; i++) {
506 f = fopen(test_file_name, "w+");
507 if (f) {
508 fclose(f);
509 unlink(test_file_name);
510 }
511 }
512}
513
514static void do_something(void)
515{
516 fs_something();
517
518 sort_something();
519
520 syscall_something();
521}
522
523enum {
524 TEST_CODE_READING_OK,
525 TEST_CODE_READING_NO_VMLINUX,
526 TEST_CODE_READING_NO_KCORE,
527 TEST_CODE_READING_NO_ACCESS,
528 TEST_CODE_READING_NO_KERNEL_OBJ,
529};
530
531static int do_test_code_reading(bool try_kcore)
532{
533 struct machine *machine;
534 struct thread *thread;
535 struct record_opts opts = {
536 .mmap_pages = UINT_MAX,
537 .user_freq = UINT_MAX,
538 .user_interval = ULLONG_MAX,
539 .freq = 500,
540 .target = {
541 .uses_mmap = true,
542 },
543 };
544 struct state state = {
545 .done_cnt = 0,
546 };
547 struct perf_thread_map *threads = NULL;
548 struct perf_cpu_map *cpus = NULL;
549 struct evlist *evlist = NULL;
550 struct evsel *evsel = NULL;
551 int err = -1, ret;
552 pid_t pid;
553 struct map *map;
554 bool have_vmlinux, have_kcore;
555 struct dso *dso;
556 const char *events[] = { "cycles", "cycles:u", "cpu-clock", "cpu-clock:u", NULL };
557 int evidx = 0;
558
559 pid = getpid();
560
561 machine = machine__new_host();
562 machine->env = &perf_env;
563
564 ret = machine__create_kernel_maps(machine);
565 if (ret < 0) {
566 pr_debug("machine__create_kernel_maps failed\n");
567 goto out_err;
568 }
569
570 /* Force the use of kallsyms instead of vmlinux to try kcore */
571 if (try_kcore)
572 symbol_conf.kallsyms_name = "/proc/kallsyms";
573
574 /* Load kernel map */
575 map = machine__kernel_map(machine);
576 ret = map__load(map);
577 if (ret < 0) {
578 pr_debug("map__load failed\n");
579 goto out_err;
580 }
581 dso = map__dso(map);
582 have_vmlinux = dso__is_vmlinux(dso);
583 have_kcore = dso__is_kcore(dso);
584
585 /* 2nd time through we just try kcore */
586 if (try_kcore && !have_kcore)
587 return TEST_CODE_READING_NO_KCORE;
588
589 /* No point getting kernel events if there is no kernel object */
590 if (!have_vmlinux && !have_kcore)
591 evidx++;
592
593 threads = thread_map__new_by_tid(pid);
594 if (!threads) {
595 pr_debug("thread_map__new_by_tid failed\n");
596 goto out_err;
597 }
598
599 ret = perf_event__synthesize_thread_map(NULL, threads,
600 perf_event__process, machine,
601 true, false);
602 if (ret < 0) {
603 pr_debug("perf_event__synthesize_thread_map failed\n");
604 goto out_err;
605 }
606
607 thread = machine__findnew_thread(machine, pid, pid);
608 if (!thread) {
609 pr_debug("machine__findnew_thread failed\n");
610 goto out_put;
611 }
612
613 cpus = perf_cpu_map__new_online_cpus();
614 if (!cpus) {
615 pr_debug("perf_cpu_map__new failed\n");
616 goto out_put;
617 }
618
619 while (events[evidx]) {
620 const char *str;
621
622 evlist = evlist__new();
623 if (!evlist) {
624 pr_debug("evlist__new failed\n");
625 goto out_put;
626 }
627
628 perf_evlist__set_maps(&evlist->core, cpus, threads);
629
630 str = events[evidx];
631 pr_debug("Parsing event '%s'\n", str);
632 ret = parse_event(evlist, str);
633 if (ret < 0) {
634 pr_debug("parse_events failed\n");
635 goto out_put;
636 }
637
638 evlist__config(evlist, &opts, NULL);
639
640 evsel = evlist__first(evlist);
641
642 evsel->core.attr.comm = 1;
643 evsel->core.attr.disabled = 1;
644 evsel->core.attr.enable_on_exec = 0;
645
646 ret = evlist__open(evlist);
647 if (ret < 0) {
648 evidx++;
649
650 if (events[evidx] == NULL && verbose > 0) {
651 char errbuf[512];
652 evlist__strerror_open(evlist, errno, errbuf, sizeof(errbuf));
653 pr_debug("perf_evlist__open() failed!\n%s\n", errbuf);
654 }
655
656 /*
657 * Both cpus and threads are now owned by evlist
658 * and will be freed by following perf_evlist__set_maps
659 * call. Getting reference to keep them alive.
660 */
661 perf_cpu_map__get(cpus);
662 perf_thread_map__get(threads);
663 perf_evlist__set_maps(&evlist->core, NULL, NULL);
664 evlist__delete(evlist);
665 evlist = NULL;
666 continue;
667 }
668 break;
669 }
670
671 if (events[evidx] == NULL)
672 goto out_put;
673
674 ret = evlist__mmap(evlist, UINT_MAX);
675 if (ret < 0) {
676 pr_debug("evlist__mmap failed\n");
677 goto out_put;
678 }
679
680 evlist__enable(evlist);
681
682 do_something();
683
684 evlist__disable(evlist);
685
686 ret = process_events(machine, evlist, &state);
687 if (ret < 0)
688 goto out_put;
689
690 if (!have_vmlinux && !have_kcore && !try_kcore)
691 err = TEST_CODE_READING_NO_KERNEL_OBJ;
692 else if (!have_vmlinux && !try_kcore)
693 err = TEST_CODE_READING_NO_VMLINUX;
694 else if (strstr(events[evidx], ":u"))
695 err = TEST_CODE_READING_NO_ACCESS;
696 else
697 err = TEST_CODE_READING_OK;
698out_put:
699 thread__put(thread);
700out_err:
701 evlist__delete(evlist);
702 perf_cpu_map__put(cpus);
703 perf_thread_map__put(threads);
704 machine__delete(machine);
705
706 return err;
707}
708
709static int test__code_reading(struct test_suite *test __maybe_unused, int subtest __maybe_unused)
710{
711 int ret;
712
713 ret = do_test_code_reading(false);
714 if (!ret)
715 ret = do_test_code_reading(true);
716
717 switch (ret) {
718 case TEST_CODE_READING_OK:
719 return 0;
720 case TEST_CODE_READING_NO_VMLINUX:
721 pr_debug("no vmlinux\n");
722 return 0;
723 case TEST_CODE_READING_NO_KCORE:
724 pr_debug("no kcore\n");
725 return 0;
726 case TEST_CODE_READING_NO_ACCESS:
727 pr_debug("no access\n");
728 return 0;
729 case TEST_CODE_READING_NO_KERNEL_OBJ:
730 pr_debug("no kernel obj\n");
731 return 0;
732 default:
733 return -1;
734 };
735}
736
737DEFINE_SUITE("Object code reading", code_reading);
1// SPDX-License-Identifier: GPL-2.0
2#include <errno.h>
3#include <linux/kernel.h>
4#include <linux/types.h>
5#include <inttypes.h>
6#include <stdlib.h>
7#include <unistd.h>
8#include <stdio.h>
9#include <string.h>
10#include <sys/param.h>
11#include <perf/cpumap.h>
12#include <perf/evlist.h>
13#include <perf/mmap.h>
14
15#include "debug.h"
16#include "dso.h"
17#include "env.h"
18#include "parse-events.h"
19#include "trace-event.h"
20#include "evlist.h"
21#include "evsel.h"
22#include "thread_map.h"
23#include "machine.h"
24#include "map.h"
25#include "symbol.h"
26#include "event.h"
27#include "record.h"
28#include "util/mmap.h"
29#include "util/synthetic-events.h"
30#include "thread.h"
31
32#include "tests.h"
33
34#include <linux/ctype.h>
35
36#define BUFSZ 1024
37#define READLEN 128
38
39struct state {
40 u64 done[1024];
41 size_t done_cnt;
42};
43
44static unsigned int hex(char c)
45{
46 if (c >= '0' && c <= '9')
47 return c - '0';
48 if (c >= 'a' && c <= 'f')
49 return c - 'a' + 10;
50 return c - 'A' + 10;
51}
52
53static size_t read_objdump_chunk(const char **line, unsigned char **buf,
54 size_t *buf_len)
55{
56 size_t bytes_read = 0;
57 unsigned char *chunk_start = *buf;
58
59 /* Read bytes */
60 while (*buf_len > 0) {
61 char c1, c2;
62
63 /* Get 2 hex digits */
64 c1 = *(*line)++;
65 if (!isxdigit(c1))
66 break;
67 c2 = *(*line)++;
68 if (!isxdigit(c2))
69 break;
70
71 /* Store byte and advance buf */
72 **buf = (hex(c1) << 4) | hex(c2);
73 (*buf)++;
74 (*buf_len)--;
75 bytes_read++;
76
77 /* End of chunk? */
78 if (isspace(**line))
79 break;
80 }
81
82 /*
83 * objdump will display raw insn as LE if code endian
84 * is LE and bytes_per_chunk > 1. In that case reverse
85 * the chunk we just read.
86 *
87 * see disassemble_bytes() at binutils/objdump.c for details
88 * how objdump chooses display endian)
89 */
90 if (bytes_read > 1 && !bigendian()) {
91 unsigned char *chunk_end = chunk_start + bytes_read - 1;
92 unsigned char tmp;
93
94 while (chunk_start < chunk_end) {
95 tmp = *chunk_start;
96 *chunk_start = *chunk_end;
97 *chunk_end = tmp;
98 chunk_start++;
99 chunk_end--;
100 }
101 }
102
103 return bytes_read;
104}
105
106static size_t read_objdump_line(const char *line, unsigned char *buf,
107 size_t buf_len)
108{
109 const char *p;
110 size_t ret, bytes_read = 0;
111
112 /* Skip to a colon */
113 p = strchr(line, ':');
114 if (!p)
115 return 0;
116 p++;
117
118 /* Skip initial spaces */
119 while (*p) {
120 if (!isspace(*p))
121 break;
122 p++;
123 }
124
125 do {
126 ret = read_objdump_chunk(&p, &buf, &buf_len);
127 bytes_read += ret;
128 p++;
129 } while (ret > 0);
130
131 /* return number of successfully read bytes */
132 return bytes_read;
133}
134
135static int read_objdump_output(FILE *f, void *buf, size_t *len, u64 start_addr)
136{
137 char *line = NULL;
138 size_t line_len, off_last = 0;
139 ssize_t ret;
140 int err = 0;
141 u64 addr, last_addr = start_addr;
142
143 while (off_last < *len) {
144 size_t off, read_bytes, written_bytes;
145 unsigned char tmp[BUFSZ];
146
147 ret = getline(&line, &line_len, f);
148 if (feof(f))
149 break;
150 if (ret < 0) {
151 pr_debug("getline failed\n");
152 err = -1;
153 break;
154 }
155
156 /* read objdump data into temporary buffer */
157 read_bytes = read_objdump_line(line, tmp, sizeof(tmp));
158 if (!read_bytes)
159 continue;
160
161 if (sscanf(line, "%"PRIx64, &addr) != 1)
162 continue;
163 if (addr < last_addr) {
164 pr_debug("addr going backwards, read beyond section?\n");
165 break;
166 }
167 last_addr = addr;
168
169 /* copy it from temporary buffer to 'buf' according
170 * to address on current objdump line */
171 off = addr - start_addr;
172 if (off >= *len)
173 break;
174 written_bytes = MIN(read_bytes, *len - off);
175 memcpy(buf + off, tmp, written_bytes);
176 off_last = off + written_bytes;
177 }
178
179 /* len returns number of bytes that could not be read */
180 *len -= off_last;
181
182 free(line);
183
184 return err;
185}
186
187static int read_via_objdump(const char *filename, u64 addr, void *buf,
188 size_t len)
189{
190 char cmd[PATH_MAX * 2];
191 const char *fmt;
192 FILE *f;
193 int ret;
194
195 fmt = "%s -z -d --start-address=0x%"PRIx64" --stop-address=0x%"PRIx64" %s";
196 ret = snprintf(cmd, sizeof(cmd), fmt, "objdump", addr, addr + len,
197 filename);
198 if (ret <= 0 || (size_t)ret >= sizeof(cmd))
199 return -1;
200
201 pr_debug("Objdump command is: %s\n", cmd);
202
203 /* Ignore objdump errors */
204 strcat(cmd, " 2>/dev/null");
205
206 f = popen(cmd, "r");
207 if (!f) {
208 pr_debug("popen failed\n");
209 return -1;
210 }
211
212 ret = read_objdump_output(f, buf, &len, addr);
213 if (len) {
214 pr_debug("objdump read too few bytes: %zd\n", len);
215 if (!ret)
216 ret = len;
217 }
218
219 pclose(f);
220
221 return ret;
222}
223
224static void dump_buf(unsigned char *buf, size_t len)
225{
226 size_t i;
227
228 for (i = 0; i < len; i++) {
229 pr_debug("0x%02x ", buf[i]);
230 if (i % 16 == 15)
231 pr_debug("\n");
232 }
233 pr_debug("\n");
234}
235
236static int read_object_code(u64 addr, size_t len, u8 cpumode,
237 struct thread *thread, struct state *state)
238{
239 struct addr_location al;
240 unsigned char buf1[BUFSZ];
241 unsigned char buf2[BUFSZ];
242 size_t ret_len;
243 u64 objdump_addr;
244 const char *objdump_name;
245 char decomp_name[KMOD_DECOMP_LEN];
246 bool decomp = false;
247 int ret;
248
249 pr_debug("Reading object code for memory address: %#"PRIx64"\n", addr);
250
251 if (!thread__find_map(thread, cpumode, addr, &al) || !al.map->dso) {
252 if (cpumode == PERF_RECORD_MISC_HYPERVISOR) {
253 pr_debug("Hypervisor address can not be resolved - skipping\n");
254 return 0;
255 }
256
257 pr_debug("thread__find_map failed\n");
258 return -1;
259 }
260
261 pr_debug("File is: %s\n", al.map->dso->long_name);
262
263 if (al.map->dso->symtab_type == DSO_BINARY_TYPE__KALLSYMS &&
264 !dso__is_kcore(al.map->dso)) {
265 pr_debug("Unexpected kernel address - skipping\n");
266 return 0;
267 }
268
269 pr_debug("On file address is: %#"PRIx64"\n", al.addr);
270
271 if (len > BUFSZ)
272 len = BUFSZ;
273
274 /* Do not go off the map */
275 if (addr + len > al.map->end)
276 len = al.map->end - addr;
277
278 /* Read the object code using perf */
279 ret_len = dso__data_read_offset(al.map->dso, thread->maps->machine,
280 al.addr, buf1, len);
281 if (ret_len != len) {
282 pr_debug("dso__data_read_offset failed\n");
283 return -1;
284 }
285
286 /*
287 * Converting addresses for use by objdump requires more information.
288 * map__load() does that. See map__rip_2objdump() for details.
289 */
290 if (map__load(al.map))
291 return -1;
292
293 /* objdump struggles with kcore - try each map only once */
294 if (dso__is_kcore(al.map->dso)) {
295 size_t d;
296
297 for (d = 0; d < state->done_cnt; d++) {
298 if (state->done[d] == al.map->start) {
299 pr_debug("kcore map tested already");
300 pr_debug(" - skipping\n");
301 return 0;
302 }
303 }
304 if (state->done_cnt >= ARRAY_SIZE(state->done)) {
305 pr_debug("Too many kcore maps - skipping\n");
306 return 0;
307 }
308 state->done[state->done_cnt++] = al.map->start;
309 }
310
311 objdump_name = al.map->dso->long_name;
312 if (dso__needs_decompress(al.map->dso)) {
313 if (dso__decompress_kmodule_path(al.map->dso, objdump_name,
314 decomp_name,
315 sizeof(decomp_name)) < 0) {
316 pr_debug("decompression failed\n");
317 return -1;
318 }
319
320 decomp = true;
321 objdump_name = decomp_name;
322 }
323
324 /* Read the object code using objdump */
325 objdump_addr = map__rip_2objdump(al.map, al.addr);
326 ret = read_via_objdump(objdump_name, objdump_addr, buf2, len);
327
328 if (decomp)
329 unlink(objdump_name);
330
331 if (ret > 0) {
332 /*
333 * The kernel maps are inaccurate - assume objdump is right in
334 * that case.
335 */
336 if (cpumode == PERF_RECORD_MISC_KERNEL ||
337 cpumode == PERF_RECORD_MISC_GUEST_KERNEL) {
338 len -= ret;
339 if (len) {
340 pr_debug("Reducing len to %zu\n", len);
341 } else if (dso__is_kcore(al.map->dso)) {
342 /*
343 * objdump cannot handle very large segments
344 * that may be found in kcore.
345 */
346 pr_debug("objdump failed for kcore");
347 pr_debug(" - skipping\n");
348 return 0;
349 } else {
350 return -1;
351 }
352 }
353 }
354 if (ret < 0) {
355 pr_debug("read_via_objdump failed\n");
356 return -1;
357 }
358
359 /* The results should be identical */
360 if (memcmp(buf1, buf2, len)) {
361 pr_debug("Bytes read differ from those read by objdump\n");
362 pr_debug("buf1 (dso):\n");
363 dump_buf(buf1, len);
364 pr_debug("buf2 (objdump):\n");
365 dump_buf(buf2, len);
366 return -1;
367 }
368 pr_debug("Bytes read match those read by objdump\n");
369
370 return 0;
371}
372
373static int process_sample_event(struct machine *machine,
374 struct evlist *evlist,
375 union perf_event *event, struct state *state)
376{
377 struct perf_sample sample;
378 struct thread *thread;
379 int ret;
380
381 if (perf_evlist__parse_sample(evlist, event, &sample)) {
382 pr_debug("perf_evlist__parse_sample failed\n");
383 return -1;
384 }
385
386 thread = machine__findnew_thread(machine, sample.pid, sample.tid);
387 if (!thread) {
388 pr_debug("machine__findnew_thread failed\n");
389 return -1;
390 }
391
392 ret = read_object_code(sample.ip, READLEN, sample.cpumode, thread, state);
393 thread__put(thread);
394 return ret;
395}
396
397static int process_event(struct machine *machine, struct evlist *evlist,
398 union perf_event *event, struct state *state)
399{
400 if (event->header.type == PERF_RECORD_SAMPLE)
401 return process_sample_event(machine, evlist, event, state);
402
403 if (event->header.type == PERF_RECORD_THROTTLE ||
404 event->header.type == PERF_RECORD_UNTHROTTLE)
405 return 0;
406
407 if (event->header.type < PERF_RECORD_MAX) {
408 int ret;
409
410 ret = machine__process_event(machine, event, NULL);
411 if (ret < 0)
412 pr_debug("machine__process_event failed, event type %u\n",
413 event->header.type);
414 return ret;
415 }
416
417 return 0;
418}
419
420static int process_events(struct machine *machine, struct evlist *evlist,
421 struct state *state)
422{
423 union perf_event *event;
424 struct mmap *md;
425 int i, ret;
426
427 for (i = 0; i < evlist->core.nr_mmaps; i++) {
428 md = &evlist->mmap[i];
429 if (perf_mmap__read_init(&md->core) < 0)
430 continue;
431
432 while ((event = perf_mmap__read_event(&md->core)) != NULL) {
433 ret = process_event(machine, evlist, event, state);
434 perf_mmap__consume(&md->core);
435 if (ret < 0)
436 return ret;
437 }
438 perf_mmap__read_done(&md->core);
439 }
440 return 0;
441}
442
443static int comp(const void *a, const void *b)
444{
445 return *(int *)a - *(int *)b;
446}
447
448static void do_sort_something(void)
449{
450 int buf[40960], i;
451
452 for (i = 0; i < (int)ARRAY_SIZE(buf); i++)
453 buf[i] = ARRAY_SIZE(buf) - i - 1;
454
455 qsort(buf, ARRAY_SIZE(buf), sizeof(int), comp);
456
457 for (i = 0; i < (int)ARRAY_SIZE(buf); i++) {
458 if (buf[i] != i) {
459 pr_debug("qsort failed\n");
460 break;
461 }
462 }
463}
464
465static void sort_something(void)
466{
467 int i;
468
469 for (i = 0; i < 10; i++)
470 do_sort_something();
471}
472
473static void syscall_something(void)
474{
475 int pipefd[2];
476 int i;
477
478 for (i = 0; i < 1000; i++) {
479 if (pipe(pipefd) < 0) {
480 pr_debug("pipe failed\n");
481 break;
482 }
483 close(pipefd[1]);
484 close(pipefd[0]);
485 }
486}
487
488static void fs_something(void)
489{
490 const char *test_file_name = "temp-perf-code-reading-test-file--";
491 FILE *f;
492 int i;
493
494 for (i = 0; i < 1000; i++) {
495 f = fopen(test_file_name, "w+");
496 if (f) {
497 fclose(f);
498 unlink(test_file_name);
499 }
500 }
501}
502
503#ifdef __s390x__
504#include "header.h" // for get_cpuid()
505#endif
506
507static const char *do_determine_event(bool excl_kernel)
508{
509 const char *event = excl_kernel ? "cycles:u" : "cycles";
510
511#ifdef __s390x__
512 char cpuid[128], model[16], model_c[16], cpum_cf_v[16];
513 unsigned int family;
514 int ret, cpum_cf_a;
515
516 if (get_cpuid(cpuid, sizeof(cpuid)))
517 goto out_clocks;
518 ret = sscanf(cpuid, "%*[^,],%u,%[^,],%[^,],%[^,],%x", &family, model_c,
519 model, cpum_cf_v, &cpum_cf_a);
520 if (ret != 5) /* Not available */
521 goto out_clocks;
522 if (excl_kernel && (cpum_cf_a & 4))
523 return event;
524 if (!excl_kernel && (cpum_cf_a & 2))
525 return event;
526
527 /* Fall through: missing authorization */
528out_clocks:
529 event = excl_kernel ? "cpu-clock:u" : "cpu-clock";
530
531#endif
532 return event;
533}
534
535static void do_something(void)
536{
537 fs_something();
538
539 sort_something();
540
541 syscall_something();
542}
543
544enum {
545 TEST_CODE_READING_OK,
546 TEST_CODE_READING_NO_VMLINUX,
547 TEST_CODE_READING_NO_KCORE,
548 TEST_CODE_READING_NO_ACCESS,
549 TEST_CODE_READING_NO_KERNEL_OBJ,
550};
551
552static int do_test_code_reading(bool try_kcore)
553{
554 struct machine *machine;
555 struct thread *thread;
556 struct record_opts opts = {
557 .mmap_pages = UINT_MAX,
558 .user_freq = UINT_MAX,
559 .user_interval = ULLONG_MAX,
560 .freq = 500,
561 .target = {
562 .uses_mmap = true,
563 },
564 };
565 struct state state = {
566 .done_cnt = 0,
567 };
568 struct perf_thread_map *threads = NULL;
569 struct perf_cpu_map *cpus = NULL;
570 struct evlist *evlist = NULL;
571 struct evsel *evsel = NULL;
572 int err = -1, ret;
573 pid_t pid;
574 struct map *map;
575 bool have_vmlinux, have_kcore, excl_kernel = false;
576
577 pid = getpid();
578
579 machine = machine__new_host();
580 machine->env = &perf_env;
581
582 ret = machine__create_kernel_maps(machine);
583 if (ret < 0) {
584 pr_debug("machine__create_kernel_maps failed\n");
585 goto out_err;
586 }
587
588 /* Force the use of kallsyms instead of vmlinux to try kcore */
589 if (try_kcore)
590 symbol_conf.kallsyms_name = "/proc/kallsyms";
591
592 /* Load kernel map */
593 map = machine__kernel_map(machine);
594 ret = map__load(map);
595 if (ret < 0) {
596 pr_debug("map__load failed\n");
597 goto out_err;
598 }
599 have_vmlinux = dso__is_vmlinux(map->dso);
600 have_kcore = dso__is_kcore(map->dso);
601
602 /* 2nd time through we just try kcore */
603 if (try_kcore && !have_kcore)
604 return TEST_CODE_READING_NO_KCORE;
605
606 /* No point getting kernel events if there is no kernel object */
607 if (!have_vmlinux && !have_kcore)
608 excl_kernel = true;
609
610 threads = thread_map__new_by_tid(pid);
611 if (!threads) {
612 pr_debug("thread_map__new_by_tid failed\n");
613 goto out_err;
614 }
615
616 ret = perf_event__synthesize_thread_map(NULL, threads,
617 perf_event__process, machine, false);
618 if (ret < 0) {
619 pr_debug("perf_event__synthesize_thread_map failed\n");
620 goto out_err;
621 }
622
623 thread = machine__findnew_thread(machine, pid, pid);
624 if (!thread) {
625 pr_debug("machine__findnew_thread failed\n");
626 goto out_put;
627 }
628
629 cpus = perf_cpu_map__new(NULL);
630 if (!cpus) {
631 pr_debug("perf_cpu_map__new failed\n");
632 goto out_put;
633 }
634
635 while (1) {
636 const char *str;
637
638 evlist = evlist__new();
639 if (!evlist) {
640 pr_debug("perf_evlist__new failed\n");
641 goto out_put;
642 }
643
644 perf_evlist__set_maps(&evlist->core, cpus, threads);
645
646 str = do_determine_event(excl_kernel);
647 pr_debug("Parsing event '%s'\n", str);
648 ret = parse_events(evlist, str, NULL);
649 if (ret < 0) {
650 pr_debug("parse_events failed\n");
651 goto out_put;
652 }
653
654 perf_evlist__config(evlist, &opts, NULL);
655
656 evsel = evlist__first(evlist);
657
658 evsel->core.attr.comm = 1;
659 evsel->core.attr.disabled = 1;
660 evsel->core.attr.enable_on_exec = 0;
661
662 ret = evlist__open(evlist);
663 if (ret < 0) {
664 if (!excl_kernel) {
665 excl_kernel = true;
666 /*
667 * Both cpus and threads are now owned by evlist
668 * and will be freed by following perf_evlist__set_maps
669 * call. Getting refference to keep them alive.
670 */
671 perf_cpu_map__get(cpus);
672 perf_thread_map__get(threads);
673 perf_evlist__set_maps(&evlist->core, NULL, NULL);
674 evlist__delete(evlist);
675 evlist = NULL;
676 continue;
677 }
678
679 if (verbose > 0) {
680 char errbuf[512];
681 evlist__strerror_open(evlist, errno, errbuf, sizeof(errbuf));
682 pr_debug("perf_evlist__open() failed!\n%s\n", errbuf);
683 }
684
685 goto out_put;
686 }
687 break;
688 }
689
690 ret = evlist__mmap(evlist, UINT_MAX);
691 if (ret < 0) {
692 pr_debug("evlist__mmap failed\n");
693 goto out_put;
694 }
695
696 evlist__enable(evlist);
697
698 do_something();
699
700 evlist__disable(evlist);
701
702 ret = process_events(machine, evlist, &state);
703 if (ret < 0)
704 goto out_put;
705
706 if (!have_vmlinux && !have_kcore && !try_kcore)
707 err = TEST_CODE_READING_NO_KERNEL_OBJ;
708 else if (!have_vmlinux && !try_kcore)
709 err = TEST_CODE_READING_NO_VMLINUX;
710 else if (excl_kernel)
711 err = TEST_CODE_READING_NO_ACCESS;
712 else
713 err = TEST_CODE_READING_OK;
714out_put:
715 thread__put(thread);
716out_err:
717
718 if (evlist) {
719 evlist__delete(evlist);
720 } else {
721 perf_cpu_map__put(cpus);
722 perf_thread_map__put(threads);
723 }
724 machine__delete_threads(machine);
725 machine__delete(machine);
726
727 return err;
728}
729
730int test__code_reading(struct test *test __maybe_unused, int subtest __maybe_unused)
731{
732 int ret;
733
734 ret = do_test_code_reading(false);
735 if (!ret)
736 ret = do_test_code_reading(true);
737
738 switch (ret) {
739 case TEST_CODE_READING_OK:
740 return 0;
741 case TEST_CODE_READING_NO_VMLINUX:
742 pr_debug("no vmlinux\n");
743 return 0;
744 case TEST_CODE_READING_NO_KCORE:
745 pr_debug("no kcore\n");
746 return 0;
747 case TEST_CODE_READING_NO_ACCESS:
748 pr_debug("no access\n");
749 return 0;
750 case TEST_CODE_READING_NO_KERNEL_OBJ:
751 pr_debug("no kernel obj\n");
752 return 0;
753 default:
754 return -1;
755 };
756}