xref: /linux/tools/mm/page_owner_sort.c (revision c31f4aa8fed048fa70e742c4bb49bb48dc489ab3) 
1 // SPDX-License-Identifier: GPL-2.0
2 /*
3  * User-space helper to sort the output of /sys/kernel/debug/page_owner
4  *
5  * Example use:
6  * cat /sys/kernel/debug/page_owner > page_owner_full.txt
7  * ./page_owner_sort page_owner_full.txt sorted_page_owner.txt
8  * Or sort by total memory:
9  * ./page_owner_sort -m page_owner_full.txt sorted_page_owner.txt
10  *
11  * See Documentation/mm/page_owner.rst
12 */
13 
14 #include <stdio.h>
15 #include <stdlib.h>
16 #include <stdbool.h>
17 #include <sys/types.h>
18 #include <sys/stat.h>
19 #include <fcntl.h>
20 #include <unistd.h>
21 #include <string.h>
22 #include <regex.h>
23 #include <errno.h>
24 #include <linux/types.h>
25 #include <getopt.h>
26 
27 #define TASK_COMM_LEN 16
28 
29 struct block_list {
30 	char *txt;
31 	char *comm; // task command name
32 	char *stacktrace;
33 	__u64 ts_nsec;
34 	int len;
35 	int num;
36 	int page_num;
37 	pid_t pid;
38 	pid_t tgid;
39 	int allocator;
40 };
41 enum FILTER_BIT {
42 	FILTER_PID = 1<<1,
43 	FILTER_TGID = 1<<2,
44 	FILTER_COMM = 1<<3
45 };
46 enum CULL_BIT {
47 	CULL_PID = 1<<1,
48 	CULL_TGID = 1<<2,
49 	CULL_COMM = 1<<3,
50 	CULL_STACKTRACE = 1<<4,
51 	CULL_ALLOCATOR = 1<<5
52 };
53 enum ALLOCATOR_BIT {
54 	ALLOCATOR_CMA = 1<<1,
55 	ALLOCATOR_SLAB = 1<<2,
56 	ALLOCATOR_VMALLOC = 1<<3,
57 	ALLOCATOR_OTHERS = 1<<4
58 };
59 enum ARG_TYPE {
60 	ARG_TXT, ARG_COMM, ARG_STACKTRACE, ARG_ALLOC_TS, ARG_CULL_TIME,
61 	ARG_PAGE_NUM, ARG_PID, ARG_TGID, ARG_UNKNOWN, ARG_ALLOCATOR
62 };
63 enum SORT_ORDER {
64 	SORT_ASC = 1,
65 	SORT_DESC = -1,
66 };
67 enum COMP_FLAG {
68 	COMP_NO_FLAG = 0,
69 	COMP_ALLOC = 1<<0,
70 	COMP_PAGE_NUM = 1<<1,
71 	COMP_PID = 1<<2,
72 	COMP_STACK = 1<<3,
73 	COMP_NUM = 1<<4,
74 	COMP_TGID = 1<<5,
75 	COMP_COMM = 1<<6
76 };
77 struct filter_condition {
78 	pid_t *pids;
79 	pid_t *tgids;
80 	char **comms;
81 	int pids_size;
82 	int tgids_size;
83 	int comms_size;
84 };
85 struct sort_condition {
86 	int (**cmps)(const void *, const void *);
87 	int *signs;
88 	int size;
89 };
90 static struct filter_condition fc;
91 static struct sort_condition sc;
92 static regex_t order_pattern;
93 static regex_t pid_pattern;
94 static regex_t tgid_pattern;
95 static regex_t comm_pattern;
96 static regex_t ts_nsec_pattern;
97 static struct block_list *list;
98 static int list_size;
99 static int max_size;
100 static int cull;
101 static int filter;
102 static bool debug_on;
103 
104 static void set_single_cmp(int (*cmp)(const void *, const void *), int sign);
105 
106 int read_block(char *buf, char *ext_buf, int buf_size, FILE *fin)
107 {
108 	char *curr = buf, *const buf_end = buf + buf_size;
109 
110 	while (buf_end - curr > 1 && fgets(curr, buf_end - curr, fin)) {
111 		if (*curr == '\n') { /* empty line */
112 			return curr - buf;
113 		}
114 		if (!strncmp(curr, "PFN", 3)) {
115 			strcpy(ext_buf, curr);
116 			continue;
117 		}
118 		curr += strlen(curr);
119 	}
120 
121 	return -1; /* EOF or no space left in buf. */
122 }
123 
124 static int compare_txt(const void *p1, const void *p2)
125 {
126 	const struct block_list *l1 = p1, *l2 = p2;
127 
128 	return strcmp(l1->txt, l2->txt);
129 }
130 
131 static int compare_stacktrace(const void *p1, const void *p2)
132 {
133 	const struct block_list *l1 = p1, *l2 = p2;
134 
135 	return strcmp(l1->stacktrace, l2->stacktrace);
136 }
137 
138 static int compare_num(const void *p1, const void *p2)
139 {
140 	const struct block_list *l1 = p1, *l2 = p2;
141 
142 	return l1->num - l2->num;
143 }
144 
145 static int compare_page_num(const void *p1, const void *p2)
146 {
147 	const struct block_list *l1 = p1, *l2 = p2;
148 
149 	return l1->page_num - l2->page_num;
150 }
151 
152 static int compare_pid(const void *p1, const void *p2)
153 {
154 	const struct block_list *l1 = p1, *l2 = p2;
155 
156 	return l1->pid - l2->pid;
157 }
158 
159 static int compare_tgid(const void *p1, const void *p2)
160 {
161 	const struct block_list *l1 = p1, *l2 = p2;
162 
163 	return l1->tgid - l2->tgid;
164 }
165 
166 static int compare_allocator(const void *p1, const void *p2)
167 {
168 	const struct block_list *l1 = p1, *l2 = p2;
169 
170 	return l1->allocator - l2->allocator;
171 }
172 
173 static int compare_comm(const void *p1, const void *p2)
174 {
175 	const struct block_list *l1 = p1, *l2 = p2;
176 
177 	return strcmp(l1->comm, l2->comm);
178 }
179 
180 static int compare_ts(const void *p1, const void *p2)
181 {
182 	const struct block_list *l1 = p1, *l2 = p2;
183 
184 	return l1->ts_nsec < l2->ts_nsec ? -1 : 1;
185 }
186 
187 static int compare_cull_condition(const void *p1, const void *p2)
188 {
189 	if (cull == 0)
190 		return compare_txt(p1, p2);
191 	if ((cull & CULL_STACKTRACE) && compare_stacktrace(p1, p2))
192 		return compare_stacktrace(p1, p2);
193 	if ((cull & CULL_PID) && compare_pid(p1, p2))
194 		return compare_pid(p1, p2);
195 	if ((cull & CULL_TGID) && compare_tgid(p1, p2))
196 		return compare_tgid(p1, p2);
197 	if ((cull & CULL_COMM) && compare_comm(p1, p2))
198 		return compare_comm(p1, p2);
199 	if ((cull & CULL_ALLOCATOR) && compare_allocator(p1, p2))
200 		return compare_allocator(p1, p2);
201 	return 0;
202 }
203 
204 static int compare_sort_condition(const void *p1, const void *p2)
205 {
206 	int cmp = 0;
207 
208 	for (int i = 0; i < sc.size; ++i)
209 		if (cmp == 0)
210 			cmp = sc.signs[i] * sc.cmps[i](p1, p2);
211 	return cmp;
212 }
213 
214 static int remove_pattern(regex_t *pattern, char *buf, int len)
215 {
216 	regmatch_t pmatch[2];
217 	int err;
218 
219 	err = regexec(pattern, buf, 2, pmatch, REG_NOTBOL);
220 	if (err != 0 || pmatch[1].rm_so == -1)
221 		return len;
222 
223 	memcpy(buf + pmatch[1].rm_so,
224 		buf + pmatch[1].rm_eo, len - pmatch[1].rm_eo);
225 
226 	return len - (pmatch[1].rm_eo - pmatch[1].rm_so);
227 }
228 
229 static int search_pattern(regex_t *pattern, char *pattern_str, char *buf)
230 {
231 	int err, val_len;
232 	regmatch_t pmatch[2];
233 
234 	err = regexec(pattern, buf, 2, pmatch, REG_NOTBOL);
235 	if (err != 0 || pmatch[1].rm_so == -1) {
236 		if (debug_on)
237 			fprintf(stderr, "no matching pattern in %s\n", buf);
238 		return -1;
239 	}
240 	val_len = pmatch[1].rm_eo - pmatch[1].rm_so;
241 
242 	memcpy(pattern_str, buf + pmatch[1].rm_so, val_len);
243 
244 	return 0;
245 }
246 
247 static bool check_regcomp(regex_t *pattern, const char *regex)
248 {
249 	int err;
250 
251 	err = regcomp(pattern, regex, REG_EXTENDED | REG_NEWLINE);
252 	if (err != 0 || pattern->re_nsub != 1) {
253 		fprintf(stderr, "Invalid pattern %s code %d\n", regex, err);
254 		return false;
255 	}
256 	return true;
257 }
258 
259 static char **explode(char sep, const char *str, int *size)
260 {
261 	int count = 0, len = strlen(str);
262 	int lastindex = -1, j = 0;
263 
264 	for (int i = 0; i < len; i++)
265 		if (str[i] == sep)
266 			count++;
267 	char **ret = calloc(++count, sizeof(char *));
268 
269 	for (int i = 0; i < len; i++) {
270 		if (str[i] == sep) {
271 			ret[j] = calloc(i - lastindex, sizeof(char));
272 			memcpy(ret[j++], str + lastindex + 1, i - lastindex - 1);
273 			lastindex = i;
274 		}
275 	}
276 	if (lastindex <= len - 1) {
277 		ret[j] = calloc(len - lastindex, sizeof(char));
278 		memcpy(ret[j++], str + lastindex + 1, strlen(str) - 1 - lastindex);
279 	}
280 	*size = j;
281 	return ret;
282 }
283 
284 static void free_explode(char **arr, int size)
285 {
286 	for (int i = 0; i < size; i++)
287 		free(arr[i]);
288 	free(arr);
289 }
290 
291 # define FIELD_BUFF 25
292 
293 static int get_page_num(char *buf)
294 {
295 	int order_val;
296 	char order_str[FIELD_BUFF] = {0};
297 	char *endptr;
298 
299 	search_pattern(&order_pattern, order_str, buf);
300 	errno = 0;
301 	order_val = strtol(order_str, &endptr, 10);
302 	if (order_val > 64 || errno != 0 || endptr == order_str || *endptr != '\0') {
303 		if (debug_on)
304 			fprintf(stderr, "wrong order in follow buf:\n%s\n", buf);
305 		return 0;
306 	}
307 
308 	return 1 << order_val;
309 }
310 
311 static pid_t get_pid(char *buf)
312 {
313 	pid_t pid;
314 	char pid_str[FIELD_BUFF] = {0};
315 	char *endptr;
316 
317 	search_pattern(&pid_pattern, pid_str, buf);
318 	errno = 0;
319 	pid = strtol(pid_str, &endptr, 10);
320 	if (errno != 0 || endptr == pid_str || *endptr != '\0') {
321 		if (debug_on)
322 			fprintf(stderr, "wrong/invalid pid in follow buf:\n%s\n", buf);
323 		return -1;
324 	}
325 
326 	return pid;
327 
328 }
329 
330 static pid_t get_tgid(char *buf)
331 {
332 	pid_t tgid;
333 	char tgid_str[FIELD_BUFF] = {0};
334 	char *endptr;
335 
336 	search_pattern(&tgid_pattern, tgid_str, buf);
337 	errno = 0;
338 	tgid = strtol(tgid_str, &endptr, 10);
339 	if (errno != 0 || endptr == tgid_str || *endptr != '\0') {
340 		if (debug_on)
341 			fprintf(stderr, "wrong/invalid tgid in follow buf:\n%s\n", buf);
342 		return -1;
343 	}
344 
345 	return tgid;
346 
347 }
348 
349 static __u64 get_ts_nsec(char *buf)
350 {
351 	__u64 ts_nsec;
352 	char ts_nsec_str[FIELD_BUFF] = {0};
353 	char *endptr;
354 
355 	search_pattern(&ts_nsec_pattern, ts_nsec_str, buf);
356 	errno = 0;
357 	ts_nsec = strtoull(ts_nsec_str, &endptr, 10);
358 	if (errno != 0 || endptr == ts_nsec_str || *endptr != '\0') {
359 		if (debug_on)
360 			fprintf(stderr, "wrong ts_nsec in follow buf:\n%s\n", buf);
361 		return -1;
362 	}
363 
364 	return ts_nsec;
365 }
366 
367 static char *get_comm(char *buf)
368 {
369 	char *comm_str = malloc(TASK_COMM_LEN);
370 
371 	memset(comm_str, 0, TASK_COMM_LEN);
372 
373 	search_pattern(&comm_pattern, comm_str, buf);
374 	errno = 0;
375 	if (errno != 0) {
376 		if (debug_on)
377 			fprintf(stderr, "wrong comm in follow buf:\n%s\n", buf);
378 		free(comm_str);
379 		return NULL;
380 	}
381 
382 	return comm_str;
383 }
384 
385 static int get_arg_type(const char *arg)
386 {
387 	if (!strcmp(arg, "pid") || !strcmp(arg, "p"))
388 		return ARG_PID;
389 	else if (!strcmp(arg, "tgid") || !strcmp(arg, "tg"))
390 		return ARG_TGID;
391 	else if (!strcmp(arg, "name") || !strcmp(arg, "n"))
392 		return  ARG_COMM;
393 	else if (!strcmp(arg, "stacktrace") || !strcmp(arg, "st"))
394 		return ARG_STACKTRACE;
395 	else if (!strcmp(arg, "txt") || !strcmp(arg, "T"))
396 		return ARG_TXT;
397 	else if (!strcmp(arg, "alloc_ts") || !strcmp(arg, "at"))
398 		return ARG_ALLOC_TS;
399 	else if (!strcmp(arg, "allocator") || !strcmp(arg, "ator"))
400 		return ARG_ALLOCATOR;
401 	else {
402 		return ARG_UNKNOWN;
403 	}
404 }
405 
406 static int get_allocator(const char *buf, const char *migrate_info)
407 {
408 	char *tmp, *first_line, *second_line;
409 	int allocator = 0;
410 
411 	if (strstr(migrate_info, "CMA"))
412 		allocator |= ALLOCATOR_CMA;
413 	if (strstr(migrate_info, "slab"))
414 		allocator |= ALLOCATOR_SLAB;
415 	tmp = strstr(buf, "__vmalloc_node_range");
416 	if (tmp) {
417 		second_line = tmp;
418 		while (*tmp != '\n')
419 			tmp--;
420 		tmp--;
421 		while (*tmp != '\n')
422 			tmp--;
423 		first_line = ++tmp;
424 		tmp = strstr(tmp, "alloc_pages");
425 		if (tmp && first_line <= tmp && tmp < second_line)
426 			allocator |= ALLOCATOR_VMALLOC;
427 	}
428 	if (allocator == 0)
429 		allocator = ALLOCATOR_OTHERS;
430 	return allocator;
431 }
432 
433 static bool match_num_list(int num, int *list, int list_size)
434 {
435 	for (int i = 0; i < list_size; ++i)
436 		if (list[i] == num)
437 			return true;
438 	return false;
439 }
440 
441 static bool match_str_list(const char *str, char **list, int list_size)
442 {
443 	for (int i = 0; i < list_size; ++i)
444 		if (!strcmp(list[i], str))
445 			return true;
446 	return false;
447 }
448 
449 static bool is_need(char *buf)
450 {
451 	if ((filter & FILTER_PID) && !match_num_list(get_pid(buf), fc.pids, fc.pids_size))
452 		return false;
453 	if ((filter & FILTER_TGID) &&
454 		!match_num_list(get_tgid(buf), fc.tgids, fc.tgids_size))
455 		return false;
456 
457 	char *comm = get_comm(buf);
458 
459 	if ((filter & FILTER_COMM) &&
460 	!match_str_list(comm, fc.comms, fc.comms_size)) {
461 		free(comm);
462 		return false;
463 	}
464 	free(comm);
465 	return true;
466 }
467 
468 static bool add_list(char *buf, int len, char *ext_buf)
469 {
470 	if (list_size == max_size) {
471 		fprintf(stderr, "max_size too small??\n");
472 		return false;
473 	}
474 	if (!is_need(buf))
475 		return true;
476 	list[list_size].pid = get_pid(buf);
477 	list[list_size].tgid = get_tgid(buf);
478 	list[list_size].comm = get_comm(buf);
479 	list[list_size].txt = malloc(len+1);
480 	if (!list[list_size].txt) {
481 		fprintf(stderr, "Out of memory\n");
482 		return false;
483 	}
484 	memcpy(list[list_size].txt, buf, len);
485 	if (sc.cmps[0] != compare_ts) {
486 		len = remove_pattern(&ts_nsec_pattern, list[list_size].txt, len);
487 	}
488 	list[list_size].txt[len] = 0;
489 	list[list_size].len = len;
490 	list[list_size].num = 1;
491 	list[list_size].page_num = get_page_num(buf);
492 
493 	list[list_size].stacktrace = strchr(list[list_size].txt, '\n') ?: "";
494 	if (*list[list_size].stacktrace == '\n')
495 		list[list_size].stacktrace++;
496 	list[list_size].ts_nsec = get_ts_nsec(buf);
497 	list[list_size].allocator = get_allocator(buf, ext_buf);
498 	list_size++;
499 	if (list_size % 1000 == 0) {
500 		printf("loaded %d\r", list_size);
501 		fflush(stdout);
502 	}
503 	return true;
504 }
505 
506 static bool parse_cull_args(const char *arg_str)
507 {
508 	int size = 0;
509 	char **args = explode(',', arg_str, &size);
510 
511 	for (int i = 0; i < size; ++i) {
512 		int arg_type = get_arg_type(args[i]);
513 
514 		if (arg_type == ARG_PID)
515 			cull |= CULL_PID;
516 		else if (arg_type == ARG_TGID)
517 			cull |= CULL_TGID;
518 		else if (arg_type == ARG_COMM)
519 			cull |= CULL_COMM;
520 		else if (arg_type == ARG_STACKTRACE)
521 			cull |= CULL_STACKTRACE;
522 		else if (arg_type == ARG_ALLOCATOR)
523 			cull |= CULL_ALLOCATOR;
524 		else {
525 			free_explode(args, size);
526 			return false;
527 		}
528 	}
529 	free_explode(args, size);
530 	if (sc.size == 0)
531 		set_single_cmp(compare_num, SORT_DESC);
532 	return true;
533 }
534 
535 static void set_single_cmp(int (*cmp)(const void *, const void *), int sign)
536 {
537 	if (sc.signs == NULL || sc.size < 1)
538 		sc.signs = calloc(1, sizeof(int));
539 	sc.signs[0] = sign;
540 	if (sc.cmps == NULL || sc.size < 1)
541 		sc.cmps = calloc(1, sizeof(int *));
542 	sc.cmps[0] = cmp;
543 	sc.size = 1;
544 }
545 
546 static bool parse_sort_args(const char *arg_str)
547 {
548 	int size = 0;
549 
550 	if (sc.size != 0) { /* reset sort_condition */
551 		free(sc.signs);
552 		free(sc.cmps);
553 		size = 0;
554 	}
555 
556 	char **args = explode(',', arg_str, &size);
557 
558 	sc.signs = calloc(size, sizeof(int));
559 	sc.cmps = calloc(size, sizeof(int *));
560 	for (int i = 0; i < size; ++i) {
561 		int offset = 0;
562 
563 		sc.signs[i] = SORT_ASC;
564 		if (args[i][0] == '-' || args[i][0] == '+') {
565 			if (args[i][0] == '-')
566 				sc.signs[i] = SORT_DESC;
567 			offset = 1;
568 		}
569 
570 		int arg_type = get_arg_type(args[i]+offset);
571 
572 		if (arg_type == ARG_PID)
573 			sc.cmps[i] = compare_pid;
574 		else if (arg_type == ARG_TGID)
575 			sc.cmps[i] = compare_tgid;
576 		else if (arg_type == ARG_COMM)
577 			sc.cmps[i] = compare_comm;
578 		else if (arg_type == ARG_STACKTRACE)
579 			sc.cmps[i] = compare_stacktrace;
580 		else if (arg_type == ARG_ALLOC_TS)
581 			sc.cmps[i] = compare_ts;
582 		else if (arg_type == ARG_TXT)
583 			sc.cmps[i] = compare_txt;
584 		else if (arg_type == ARG_ALLOCATOR)
585 			sc.cmps[i] = compare_allocator;
586 		else {
587 			free_explode(args, size);
588 			sc.size = 0;
589 			return false;
590 		}
591 	}
592 	sc.size = size;
593 	free_explode(args, size);
594 	return true;
595 }
596 
597 static int *parse_nums_list(char *arg_str, int *list_size)
598 {
599 	int size = 0;
600 	char **args = explode(',', arg_str, &size);
601 	int *list = calloc(size, sizeof(int));
602 
603 	errno = 0;
604 	for (int i = 0; i < size; ++i) {
605 		char *endptr = NULL;
606 
607 		list[i] = strtol(args[i], &endptr, 10);
608 		if (errno != 0 || endptr == args[i] || *endptr != '\0') {
609 			free(list);
610 			return NULL;
611 		}
612 	}
613 	*list_size = size;
614 	free_explode(args, size);
615 	return list;
616 }
617 
618 static void print_allocator(FILE *out, int allocator)
619 {
620 	fprintf(out, "allocated by ");
621 	if (allocator & ALLOCATOR_CMA)
622 		fprintf(out, "CMA ");
623 	if (allocator & ALLOCATOR_SLAB)
624 		fprintf(out, "SLAB ");
625 	if (allocator & ALLOCATOR_VMALLOC)
626 		fprintf(out, "VMALLOC ");
627 	if (allocator & ALLOCATOR_OTHERS)
628 		fprintf(out, "OTHERS ");
629 }
630 
631 #define BUF_SIZE	(128 * 1024)
632 
633 static void usage(void)
634 {
635 	printf("Usage: ./page_owner_sort [OPTIONS] <input> <output>\n"
636 		"-a\t\t\tSort by memory allocation time.\n"
637 		"-m\t\t\tSort by total memory.\n"
638 		"-n\t\t\tSort by task command name.\n"
639 		"-p\t\t\tSort by pid.\n"
640 		"-P\t\t\tSort by tgid.\n"
641 		"-s\t\t\tSort by the stacktrace.\n"
642 		"-t\t\t\tSort by number of times record is seen (default).\n\n"
643 		"--pid <pidlist>\t\tSelect by pid. This selects the information"
644 		" of\n\t\t\tblocks whose process ID numbers appear in <pidlist>.\n"
645 		"--tgid <tgidlist>\tSelect by tgid. This selects the information"
646 		" of\n\t\t\tblocks whose Thread Group ID numbers appear in "
647 		"<tgidlist>.\n"
648 		"--name <cmdlist>\tSelect by command name. This selects the"
649 		" information\n\t\t\tof blocks whose command name appears in"
650 		" <cmdlist>.\n"
651 		"--cull <rules>\t\tCull by user-defined rules. <rules> is a "
652 		"single\n\t\t\targument in the form of a comma-separated list "
653 		"with some\n\t\t\tcommon fields predefined (pid, tgid, comm, "
654 		"stacktrace, allocator)\n"
655 		"--sort <order>\t\tSpecify sort order as: [+|-]key[,[+|-]key[,...]]\n"
656 	);
657 }
658 
659 int main(int argc, char **argv)
660 {
661 	FILE *fin, *fout;
662 	char *buf, *ext_buf;
663 	int i, count, compare_flag;
664 	struct stat st;
665 	int opt;
666 	struct option longopts[] = {
667 		{ "pid", required_argument, NULL, 1 },
668 		{ "tgid", required_argument, NULL, 2 },
669 		{ "name", required_argument, NULL, 3 },
670 		{ "cull", required_argument, NULL, 4 },
671 		{ "sort", required_argument, NULL, 5 },
672 		{ "help", no_argument, NULL, 'h' },
673 		{ 0, 0, 0, 0},
674 	};
675 
676 	compare_flag = COMP_NO_FLAG;
677 
678 	while ((opt = getopt_long(argc, argv, "admnpstPh", longopts, NULL)) != -1)
679 		switch (opt) {
680 		case 'a':
681 			compare_flag |= COMP_ALLOC;
682 			break;
683 		case 'd':
684 			debug_on = true;
685 			break;
686 		case 'm':
687 			compare_flag |= COMP_PAGE_NUM;
688 			break;
689 		case 'p':
690 			compare_flag |= COMP_PID;
691 			break;
692 		case 's':
693 			compare_flag |= COMP_STACK;
694 			break;
695 		case 't':
696 			compare_flag |= COMP_NUM;
697 			break;
698 		case 'P':
699 			compare_flag |= COMP_TGID;
700 			break;
701 		case 'n':
702 			compare_flag |= COMP_COMM;
703 			break;
704 		case 'h':
705 			usage();
706 			exit(0);
707 		case 1:
708 			filter = filter | FILTER_PID;
709 			fc.pids = parse_nums_list(optarg, &fc.pids_size);
710 			if (fc.pids == NULL) {
711 				fprintf(stderr, "wrong/invalid pid in from the command line:%s\n",
712 						optarg);
713 				exit(1);
714 			}
715 			break;
716 		case 2:
717 			filter = filter | FILTER_TGID;
718 			fc.tgids = parse_nums_list(optarg, &fc.tgids_size);
719 			if (fc.tgids == NULL) {
720 				fprintf(stderr, "wrong/invalid tgid in from the command line:%s\n",
721 						optarg);
722 				exit(1);
723 			}
724 			break;
725 		case 3:
726 			filter = filter | FILTER_COMM;
727 			fc.comms = explode(',', optarg, &fc.comms_size);
728 			break;
729 		case 4:
730 			if (!parse_cull_args(optarg)) {
731 				fprintf(stderr, "wrong argument after --cull option:%s\n",
732 						optarg);
733 				exit(1);
734 			}
735 			break;
736 		case 5:
737 			if (!parse_sort_args(optarg)) {
738 				fprintf(stderr, "wrong argument after --sort option:%s\n",
739 						optarg);
740 				exit(1);
741 			}
742 			break;
743 		default:
744 			usage();
745 			exit(1);
746 		}
747 
748 	if (optind >= (argc - 1)) {
749 		usage();
750 		exit(1);
751 	}
752 
753 	/* Only one compare option is allowed, yet we also want handle the
754 	 * default case were no option is provided, but we still want to
755 	 * match the behavior of the -t option (compare by number of times
756 	 * a record is seen
757 	 */
758 	switch (compare_flag) {
759 	case COMP_ALLOC:
760 		set_single_cmp(compare_ts, SORT_ASC);
761 		break;
762 	case COMP_PAGE_NUM:
763 		set_single_cmp(compare_page_num, SORT_DESC);
764 		break;
765 	case COMP_PID:
766 		set_single_cmp(compare_pid, SORT_ASC);
767 		break;
768 	case COMP_STACK:
769 		set_single_cmp(compare_stacktrace, SORT_ASC);
770 		break;
771 	case COMP_NO_FLAG:
772 	case COMP_NUM:
773 		set_single_cmp(compare_num, SORT_DESC);
774 		break;
775 	case COMP_TGID:
776 		set_single_cmp(compare_tgid, SORT_ASC);
777 		break;
778 	case COMP_COMM:
779 		set_single_cmp(compare_comm, SORT_ASC);
780 		break;
781 	default:
782 		usage();
783 		exit(1);
784 	}
785 
786 	fin = fopen(argv[optind], "r");
787 	fout = fopen(argv[optind + 1], "w");
788 	if (!fin || !fout) {
789 		usage();
790 		perror("open: ");
791 		exit(1);
792 	}
793 
794 	if (!check_regcomp(&order_pattern, "order\\s*([0-9]*),"))
795 		goto out_order;
796 	if (!check_regcomp(&pid_pattern, "pid\\s*([0-9]*),"))
797 		goto out_pid;
798 	if (!check_regcomp(&tgid_pattern, "tgid\\s*([0-9]*) "))
799 		goto out_tgid;
800 	if (!check_regcomp(&comm_pattern, "tgid\\s*[0-9]*\\s*\\((.*)\\),\\s*ts"))
801 		goto out_comm;
802 	if (!check_regcomp(&ts_nsec_pattern, "ts\\s*([0-9]*)\\s*ns"))
803 		goto out_ts;
804 
805 	fstat(fileno(fin), &st);
806 	max_size = st.st_size / 100; /* hack ... */
807 
808 	list = malloc(max_size * sizeof(*list));
809 	buf = malloc(BUF_SIZE);
810 	ext_buf = malloc(BUF_SIZE);
811 	if (!list || !buf || !ext_buf) {
812 		fprintf(stderr, "Out of memory\n");
813 		goto out_free;
814 	}
815 
816 	for ( ; ; ) {
817 		int buf_len = read_block(buf, ext_buf, BUF_SIZE, fin);
818 
819 		if (buf_len < 0)
820 			break;
821 		if (!add_list(buf, buf_len, ext_buf))
822 			goto out_free;
823 	}
824 
825 	printf("loaded %d\n", list_size);
826 
827 	printf("sorting ....\n");
828 
829 	qsort(list, list_size, sizeof(list[0]), compare_cull_condition);
830 
831 	printf("culling\n");
832 
833 	for (i = count = 0; i < list_size; i++) {
834 		if (count == 0 ||
835 		    compare_cull_condition((void *)(&list[count-1]), (void *)(&list[i])) != 0) {
836 			list[count++] = list[i];
837 		} else {
838 			list[count-1].num += list[i].num;
839 			list[count-1].page_num += list[i].page_num;
840 		}
841 	}
842 
843 	qsort(list, count, sizeof(list[0]), compare_sort_condition);
844 
845 	for (i = 0; i < count; i++) {
846 		if (cull == 0) {
847 			fprintf(fout, "%d times, %d pages, ", list[i].num, list[i].page_num);
848 			print_allocator(fout, list[i].allocator);
849 			fprintf(fout, ":\n%s\n", list[i].txt);
850 		}
851 		else {
852 			fprintf(fout, "%d times, %d pages",
853 					list[i].num, list[i].page_num);
854 			if (cull & CULL_PID || filter & FILTER_PID)
855 				fprintf(fout, ", PID %d", list[i].pid);
856 			if (cull & CULL_TGID || filter & FILTER_TGID)
857 				fprintf(fout, ", TGID %d", list[i].tgid);
858 			if (cull & CULL_COMM || filter & FILTER_COMM)
859 				fprintf(fout, ", task_comm_name: %s", list[i].comm);
860 			if (cull & CULL_ALLOCATOR) {
861 				fprintf(fout, ", ");
862 				print_allocator(fout, list[i].allocator);
863 			}
864 			if (cull & CULL_STACKTRACE)
865 				fprintf(fout, ":\n%s", list[i].stacktrace);
866 			fprintf(fout, "\n");
867 		}
868 	}
869 
870 out_free:
871 	if (ext_buf)
872 		free(ext_buf);
873 	if (buf)
874 		free(buf);
875 	if (list)
876 		free(list);
877 out_ts:
878 	regfree(&ts_nsec_pattern);
879 out_comm:
880 	regfree(&comm_pattern);
881 out_tgid:
882 	regfree(&tgid_pattern);
883 out_pid:
884 	regfree(&pid_pattern);
885 out_order:
886 	regfree(&order_pattern);
887 
888 	return 0;
889 }
890