summaryrefslogtreecommitdiff
path: root/tools/perf/builtin-test.c
blob: 02958d6a807ffa4b461b50eb2aba23b01f827219 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
/*
 * builtin-test.c
 *
 * Builtin regression testing command: ever growing number of sanity tests
 */
#include "builtin.h"

#include "util/cache.h"
#include "util/debug.h"
#include "util/parse-options.h"
#include "util/session.h"
#include "util/symbol.h"
#include "util/thread.h"

static long page_size;

static int vmlinux_matches_kallsyms_filter(struct map *map __used, struct symbol *sym)
{
	bool *visited = symbol__priv(sym);
	*visited = true;
	return 0;
}

static int test__vmlinux_matches_kallsyms(void)
{
	int err = -1;
	struct rb_node *nd;
	struct symbol *sym;
	struct map *kallsyms_map, *vmlinux_map;
	struct machine kallsyms, vmlinux;
	enum map_type type = MAP__FUNCTION;
	struct ref_reloc_sym ref_reloc_sym = { .name = "_stext", };

	/*
	 * Step 1:
	 *
	 * Init the machines that will hold kernel, modules obtained from
	 * both vmlinux + .ko files and from /proc/kallsyms split by modules.
	 */
	machine__init(&kallsyms, "", HOST_KERNEL_ID);
	machine__init(&vmlinux, "", HOST_KERNEL_ID);

	/*
	 * Step 2:
	 *
	 * Create the kernel maps for kallsyms and the DSO where we will then
	 * load /proc/kallsyms. Also create the modules maps from /proc/modules
	 * and find the .ko files that match them in /lib/modules/`uname -r`/.
	 */
	if (machine__create_kernel_maps(&kallsyms) < 0) {
		pr_debug("machine__create_kernel_maps ");
		return -1;
	}

	/*
	 * Step 3:
	 *
	 * Load and split /proc/kallsyms into multiple maps, one per module.
	 */
	if (machine__load_kallsyms(&kallsyms, "/proc/kallsyms", type, NULL) <= 0) {
		pr_debug("dso__load_kallsyms ");
		goto out;
	}

	/*
	 * Step 4:
	 *
	 * kallsyms will be internally on demand sorted by name so that we can
	 * find the reference relocation * symbol, i.e. the symbol we will use
	 * to see if the running kernel was relocated by checking if it has the
	 * same value in the vmlinux file we load.
	 */
	kallsyms_map = machine__kernel_map(&kallsyms, type);

	sym = map__find_symbol_by_name(kallsyms_map, ref_reloc_sym.name, NULL);
	if (sym == NULL) {
		pr_debug("dso__find_symbol_by_name ");
		goto out;
	}

	ref_reloc_sym.addr = sym->start;

	/*
	 * Step 5:
	 *
	 * Now repeat step 2, this time for the vmlinux file we'll auto-locate.
	 */
	if (machine__create_kernel_maps(&vmlinux) < 0) {
		pr_debug("machine__create_kernel_maps ");
		goto out;
	}

	vmlinux_map = machine__kernel_map(&vmlinux, type);
	map__kmap(vmlinux_map)->ref_reloc_sym = &ref_reloc_sym;

	/*
	 * Step 6:
	 *
	 * Locate a vmlinux file in the vmlinux path that has a buildid that
	 * matches the one of the running kernel.
	 *
	 * While doing that look if we find the ref reloc symbol, if we find it
	 * we'll have its ref_reloc_symbol.unrelocated_addr and then
	 * maps__reloc_vmlinux will notice and set proper ->[un]map_ip routines
	 * to fixup the symbols.
	 */
	if (machine__load_vmlinux_path(&vmlinux, type,
				       vmlinux_matches_kallsyms_filter) <= 0) {
		pr_debug("machine__load_vmlinux_path ");
		goto out;
	}

	err = 0;
	/*
	 * Step 7:
	 *
	 * Now look at the symbols in the vmlinux DSO and check if we find all of them
	 * in the kallsyms dso. For the ones that are in both, check its names and
	 * end addresses too.
	 */
	for (nd = rb_first(&vmlinux_map->dso->symbols[type]); nd; nd = rb_next(nd)) {
		struct symbol *pair, *first_pair;
		bool backwards = true;

		sym  = rb_entry(nd, struct symbol, rb_node);

		if (sym->start == sym->end)
			continue;

		first_pair = machine__find_kernel_symbol(&kallsyms, type, sym->start, NULL, NULL);
		pair = first_pair;

		if (pair && pair->start == sym->start) {
next_pair:
			if (strcmp(sym->name, pair->name) == 0) {
				/*
				 * kallsyms don't have the symbol end, so we
				 * set that by using the next symbol start - 1,
				 * in some cases we get this up to a page
				 * wrong, trace_kmalloc when I was developing
				 * this code was one such example, 2106 bytes
				 * off the real size. More than that and we
				 * _really_ have a problem.
				 */
				s64 skew = sym->end - pair->end;
				if (llabs(skew) < page_size)
					continue;

				pr_debug("%#Lx: diff end addr for %s v: %#Lx k: %#Lx\n",
					 sym->start, sym->name, sym->end, pair->end);
			} else {
				struct rb_node *nnd;
detour:
				nnd = backwards ? rb_prev(&pair->rb_node) :
						  rb_next(&pair->rb_node);
				if (nnd) {
					struct symbol *next = rb_entry(nnd, struct symbol, rb_node);

					if (next->start == sym->start) {
						pair = next;
						goto next_pair;
					}
				}

				if (backwards) {
					backwards = false;
					pair = first_pair;
					goto detour;
				}

				pr_debug("%#Lx: diff name v: %s k: %s\n",
					 sym->start, sym->name, pair->name);
			}
		} else
			pr_debug("%#Lx: %s not on kallsyms\n", sym->start, sym->name);

		err = -1;
	}

	if (!verbose)
		goto out;

	pr_info("Maps only in vmlinux:\n");

	for (nd = rb_first(&vmlinux.kmaps.maps[type]); nd; nd = rb_next(nd)) {
		struct map *pos = rb_entry(nd, struct map, rb_node), *pair;
		/*
		 * If it is the kernel, kallsyms is always "[kernel.kallsyms]", while
		 * the kernel will have the path for the vmlinux file being used,
		 * so use the short name, less descriptive but the same ("[kernel]" in
		 * both cases.
		 */
		pair = map_groups__find_by_name(&kallsyms.kmaps, type,
						(pos->dso->kernel ?
							pos->dso->short_name :
							pos->dso->name));
		if (pair)
			pair->priv = 1;
		else
			map__fprintf(pos, stderr);
	}

	pr_info("Maps in vmlinux with a different name in kallsyms:\n");

	for (nd = rb_first(&vmlinux.kmaps.maps[type]); nd; nd = rb_next(nd)) {
		struct map *pos = rb_entry(nd, struct map, rb_node), *pair;

		pair = map_groups__find(&kallsyms.kmaps, type, pos->start);
		if (pair == NULL || pair->priv)
			continue;

		if (pair->start == pos->start) {
			pair->priv = 1;
			pr_info(" %Lx-%Lx %Lx %s in kallsyms as",
				pos->start, pos->end, pos->pgoff, pos->dso->name);
			if (pos->pgoff != pair->pgoff || pos->end != pair->end)
				pr_info(": \n*%Lx-%Lx %Lx",
					pair->start, pair->end, pair->pgoff);
			pr_info(" %s\n", pair->dso->name);
			pair->priv = 1;
		}
	}

	pr_info("Maps only in kallsyms:\n");

	for (nd = rb_first(&kallsyms.kmaps.maps[type]);
	     nd; nd = rb_next(nd)) {
		struct map *pos = rb_entry(nd, struct map, rb_node);

		if (!pos->priv)
			map__fprintf(pos, stderr);
	}
out:
	return err;
}

#include "util/cpumap.h"
#include "util/evsel.h"
#include <sys/types.h>

static int trace_event__id(const char *event_name)
{
	char *filename;
	int err = -1, fd;

	if (asprintf(&filename,
		     "/sys/kernel/debug/tracing/events/syscalls/%s/id",
		     event_name) < 0)
		return -1;

	fd = open(filename, O_RDONLY);
	if (fd >= 0) {
		char id[16];
		if (read(fd, id, sizeof(id)) > 0)
			err = atoi(id);
		close(fd);
	}

	free(filename);
	return err;
}

static int test__open_syscall_event(void)
{
	int err = -1, fd;
	struct thread_map *threads;
	struct perf_evsel *evsel;
	struct perf_event_attr attr;
	unsigned int nr_open_calls = 111, i;
	int id = trace_event__id("sys_enter_open");

	if (id < 0) {
		pr_debug("is debugfs mounted on /sys/kernel/debug?\n");
		return -1;
	}

	threads = thread_map__new(-1, getpid());
	if (threads == NULL) {
		pr_debug("thread_map__new\n");
		return -1;
	}

	memset(&attr, 0, sizeof(attr));
	attr.type = PERF_TYPE_TRACEPOINT;
	attr.config = id;
	evsel = perf_evsel__new(&attr, 0);
	if (evsel == NULL) {
		pr_debug("perf_evsel__new\n");
		goto out_thread_map_delete;
	}

	if (perf_evsel__open_per_thread(evsel, threads) < 0) {
		pr_debug("failed to open counter: %s, "
			 "tweak /proc/sys/kernel/perf_event_paranoid?\n",
			 strerror(errno));
		goto out_evsel_delete;
	}

	for (i = 0; i < nr_open_calls; ++i) {
		fd = open("/etc/passwd", O_RDONLY);
		close(fd);
	}

	if (perf_evsel__read_on_cpu(evsel, 0, 0) < 0) {
		pr_debug("perf_evsel__open_read_on_cpu\n");
		goto out_close_fd;
	}

	if (evsel->counts->cpu[0].val != nr_open_calls) {
		pr_debug("perf_evsel__read_on_cpu: expected to intercept %d calls, got %Ld\n",
			 nr_open_calls, evsel->counts->cpu[0].val);
		goto out_close_fd;
	}
	
	err = 0;
out_close_fd:
	perf_evsel__close_fd(evsel, 1, threads->nr);
out_evsel_delete:
	perf_evsel__delete(evsel);
out_thread_map_delete:
	thread_map__delete(threads);
	return err;
}

#include <sched.h>

static int test__open_syscall_event_on_all_cpus(void)
{
	int err = -1, fd, cpu;
	struct thread_map *threads;
	struct cpu_map *cpus;
	struct perf_evsel *evsel;
	struct perf_event_attr attr;
	unsigned int nr_open_calls = 111, i;
	cpu_set_t *cpu_set;
	size_t cpu_set_size;
	int id = trace_event__id("sys_enter_open");

	if (id < 0) {
		pr_debug("is debugfs mounted on /sys/kernel/debug?\n");
		return -1;
	}

	threads = thread_map__new(-1, getpid());
	if (threads == NULL) {
		pr_debug("thread_map__new\n");
		return -1;
	}

	cpus = cpu_map__new(NULL);
	if (threads == NULL) {
		pr_debug("thread_map__new\n");
		return -1;
	}

	cpu_set = CPU_ALLOC(cpus->nr);

	if (cpu_set == NULL)
		goto out_thread_map_delete;

	cpu_set_size = CPU_ALLOC_SIZE(cpus->nr);
	CPU_ZERO_S(cpu_set_size, cpu_set);

	memset(&attr, 0, sizeof(attr));
	attr.type = PERF_TYPE_TRACEPOINT;
	attr.config = id;
	evsel = perf_evsel__new(&attr, 0);
	if (evsel == NULL) {
		pr_debug("perf_evsel__new\n");
		goto out_cpu_free;
	}

	if (perf_evsel__open(evsel, cpus, threads) < 0) {
		pr_debug("failed to open counter: %s, "
			 "tweak /proc/sys/kernel/perf_event_paranoid?\n",
			 strerror(errno));
		goto out_evsel_delete;
	}

	for (cpu = 0; cpu < cpus->nr; ++cpu) {
		unsigned int ncalls = nr_open_calls + cpu;

		CPU_SET_S(cpus->map[cpu], cpu_set_size, cpu_set);
		if (sched_setaffinity(0, cpu_set_size, cpu_set) < 0) {
			pr_debug("sched_setaffinity() failed on CPU %d: %s ",
				 cpus->map[cpu],
				 strerror(errno));
			goto out_close_fd;
		}
		for (i = 0; i < ncalls; ++i) {
			fd = open("/etc/passwd", O_RDONLY);
			close(fd);
		}
		CPU_CLR_S(cpus->map[cpu], cpu_set_size, cpu_set);
	}

	/*
	 * Here we need to explicitely preallocate the counts, as if
	 * we use the auto allocation it will allocate just for 1 cpu,
	 * as we start by cpu 0.
	 */
	if (perf_evsel__alloc_counts(evsel, cpus->nr) < 0) {
		pr_debug("perf_evsel__alloc_counts(ncpus=%d)\n", cpus->nr);
		goto out_close_fd;
	}

	for (cpu = 0; cpu < cpus->nr; ++cpu) {
		unsigned int expected;

		if (perf_evsel__read_on_cpu(evsel, cpu, 0) < 0) {
			pr_debug("perf_evsel__open_read_on_cpu\n");
			goto out_close_fd;
		}

		expected = nr_open_calls + cpu;
		if (evsel->counts->cpu[cpu].val != expected) {
			pr_debug("perf_evsel__read_on_cpu: expected to intercept %d calls on cpu %d, got %Ld\n",
				 expected, cpus->map[cpu], evsel->counts->cpu[cpu].val);
			goto out_close_fd;
		}
	}

	err = 0;
out_close_fd:
	perf_evsel__close_fd(evsel, 1, threads->nr);
out_evsel_delete:
	perf_evsel__delete(evsel);
out_cpu_free:
	CPU_FREE(cpu_set);
out_thread_map_delete:
	thread_map__delete(threads);
	return err;
}

static struct test {
	const char *desc;
	int (*func)(void);
} tests[] = {
	{
		.desc = "vmlinux symtab matches kallsyms",
		.func = test__vmlinux_matches_kallsyms,
	},
	{
		.desc = "detect open syscall event",
		.func = test__open_syscall_event,
	},
	{
		.desc = "detect open syscall event on all cpus",
		.func = test__open_syscall_event_on_all_cpus,
	},
	{
		.func = NULL,
	},
};

static int __cmd_test(void)
{
	int i = 0;

	page_size = sysconf(_SC_PAGE_SIZE);

	while (tests[i].func) {
		int err;
		pr_info("%2d: %s:", i + 1, tests[i].desc);
		pr_debug("\n--- start ---\n");
		err = tests[i].func();
		pr_debug("---- end ----\n%s:", tests[i].desc);
		pr_info(" %s\n", err ? "FAILED!\n" : "Ok");
		++i;
	}

	return 0;
}

static const char * const test_usage[] = {
	"perf test [<options>]",
	NULL,
};

static const struct option test_options[] = {
	OPT_INTEGER('v', "verbose", &verbose,
		    "be more verbose (show symbol address, etc)"),
	OPT_END()
};

int cmd_test(int argc, const char **argv, const char *prefix __used)
{
	argc = parse_options(argc, argv, test_options, test_usage, 0);
	if (argc)
		usage_with_options(test_usage, test_options);

	symbol_conf.priv_size = sizeof(int);
	symbol_conf.sort_by_name = true;
	symbol_conf.try_vmlinux_path = true;

	if (symbol__init() < 0)
		return -1;

	setup_pager();

	return __cmd_test();
}