xref: /linux/tools/mm/page_owner_sort.c (revision 4b132aacb0768ac1e652cf517097ea6f237214b9) 
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 <sys/types.h>
17 #include <sys/stat.h>
18 #include <fcntl.h>
19 #include <unistd.h>
20 #include <string.h>
21 #include <regex.h>
22 #include <errno.h>
23 #include <linux/types.h>
24 #include <getopt.h>
25 
26 #define bool int
27 #define true 1
28 #define false 0
29 #define TASK_COMM_LEN 16
30 
31 struct block_list {
32 	char *txt;
33 	char *comm; // task command name
34 	char *stacktrace;
35 	__u64 ts_nsec;
36 	int len;
37 	int num;
38 	int page_num;
39 	pid_t pid;
40 	pid_t tgid;
41 	int allocator;
42 };
43 enum FILTER_BIT {
44 	FILTER_PID = 1<<1,
45 	FILTER_TGID = 1<<2,
46 	FILTER_COMM = 1<<3
47 };
48 enum CULL_BIT {
49 	CULL_PID = 1<<1,
50 	CULL_TGID = 1<<2,
51 	CULL_COMM = 1<<3,
52 	CULL_STACKTRACE = 1<<4,
53 	CULL_ALLOCATOR = 1<<5
54 };
55 enum ALLOCATOR_BIT {
56 	ALLOCATOR_CMA = 1<<1,
57 	ALLOCATOR_SLAB = 1<<2,
58 	ALLOCATOR_VMALLOC = 1<<3,
59 	ALLOCATOR_OTHERS = 1<<4
60 };
61 enum ARG_TYPE {
62 	ARG_TXT, ARG_COMM, ARG_STACKTRACE, ARG_ALLOC_TS, ARG_CULL_TIME,
63 	ARG_PAGE_NUM, ARG_PID, ARG_TGID, ARG_UNKNOWN, ARG_ALLOCATOR
64 };
65 enum SORT_ORDER {
66 	SORT_ASC = 1,
67 	SORT_DESC = -1,
68 };
69 enum COMP_FLAG {
70 	COMP_NO_FLAG = 0,
71 	COMP_ALLOC = 1<<0,
72 	COMP_PAGE_NUM = 1<<1,
73 	COMP_PID = 1<<2,
74 	COMP_STACK = 1<<3,
75 	COMP_NUM = 1<<4,
76 	COMP_TGID = 1<<5,
77 	COMP_COMM = 1<<6
78 };
79 struct filter_condition {
80 	pid_t *pids;
81 	pid_t *tgids;
82 	char **comms;
83 	int pids_size;
84 	int tgids_size;
85 	int comms_size;
86 };
87 struct sort_condition {
88 	int (**cmps)(const void *, const void *);
89 	int *signs;
90 	int size;
91 };
92 static struct filter_condition fc;
93 static struct sort_condition sc;
94 static regex_t order_pattern;
95 static regex_t pid_pattern;
96 static regex_t tgid_pattern;
97 static regex_t comm_pattern;
98 static regex_t ts_nsec_pattern;
99 static struct block_list *list;
100 static int list_size;
101 static int max_size;
102 static int cull;
103 static int filter;
104 static bool debug_on;
105 
106 static void set_single_cmp(int (*cmp)(const void *, const void *), int sign);
107 
108 int read_block(char *buf, char *ext_buf, int buf_size, FILE *fin)
109 {
110 	char *curr = buf, *const buf_end = buf + buf_size;
111 
112 	while (buf_end - curr > 1 && fgets(curr, buf_end - curr, fin)) {
113 		if (*curr == '\n') { /* empty line */
114 			return curr - buf;
115 		}
116 		if (!strncmp(curr, "PFN", 3)) {
117 			strcpy(ext_buf, curr);
118 			continue;
119 		}
120 		curr += strlen(curr);
121 	}
122 
123 	return -1; /* EOF or no space left in buf. */
124 }
125 
126 static int compare_txt(const void *p1, const void *p2)
127 {
128 	const struct block_list *l1 = p1, *l2 = p2;
129 
130 	return strcmp(l1->txt, l2->txt);
131 }
132 
133 static int compare_stacktrace(const void *p1, const void *p2)
134 {
135 	const struct block_list *l1 = p1, *l2 = p2;
136 
137 	return strcmp(l1->stacktrace, l2->stacktrace);
138 }
139 
140 static int compare_num(const void *p1, const void *p2)
141 {
142 	const struct block_list *l1 = p1, *l2 = p2;
143 
144 	return l1->num - l2->num;
145 }
146 
147 static int compare_page_num(const void *p1, const void *p2)
148 {
149 	const struct block_list *l1 = p1, *l2 = p2;
150 
151 	return l1->page_num - l2->page_num;
152 }
153 
154 static int compare_pid(const void *p1, const void *p2)
155 {
156 	const struct block_list *l1 = p1, *l2 = p2;
157 
158 	return l1->pid - l2->pid;
159 }
160 
161 static int compare_tgid(const void *p1, const void *p2)
162 {
163 	const struct block_list *l1 = p1, *l2 = p2;
164 
165 	return l1->tgid - l2->tgid;
166 }
167 
168 static int compare_allocator(const void *p1, const void *p2)
169 {
170 	const struct block_list *l1 = p1, *l2 = p2;
171 
172 	return l1->allocator - l2->allocator;
173 }
174 
175 static int compare_comm(const void *p1, const void *p2)
176 {
177 	const struct block_list *l1 = p1, *l2 = p2;
178 
179 	return strcmp(l1->comm, l2->comm);
180 }
181 
182 static int compare_ts(const void *p1, const void *p2)
183 {
184 	const struct block_list *l1 = p1, *l2 = p2;
185 
186 	return l1->ts_nsec < l2->ts_nsec ? -1 : 1;
187 }
188 
189 static int compare_cull_condition(const void *p1, const void *p2)
190 {
191 	if (cull == 0)
192 		return compare_txt(p1, p2);
193 	if ((cull & CULL_STACKTRACE) && compare_stacktrace(p1, p2))
194 		return compare_stacktrace(p1, p2);
195 	if ((cull & CULL_PID) && compare_pid(p1, p2))
196 		return compare_pid(p1, p2);
197 	if ((cull & CULL_TGID) && compare_tgid(p1, p2))
198 		return compare_tgid(p1, p2);
199 	if ((cull & CULL_COMM) && compare_comm(p1, p2))
200 		return compare_comm(p1, p2);
201 	if ((cull & CULL_ALLOCATOR) && compare_allocator(p1, p2))
202 		return compare_allocator(p1, p2);
203 	return 0;
204 }
205 
206 static int compare_sort_condition(const void *p1, const void *p2)
207 {
208 	int cmp = 0;
209 
210 	for (int i = 0; i < sc.size; ++i)
211 		if (cmp == 0)
212 			cmp = sc.signs[i] * sc.cmps[i](p1, p2);
213 	return cmp;
214 }
215 
216 static int remove_pattern(regex_t *pattern, char *buf, int len)
217 {
218 	regmatch_t pmatch[2];
219 	int err;
220 
221 	err = regexec(pattern, buf, 2, pmatch, REG_NOTBOL);
222 	if (err != 0 || pmatch[1].rm_so == -1)
223 		return len;
224 
225 	memcpy(buf + pmatch[1].rm_so,
226 		buf + pmatch[1].rm_eo, len - pmatch[1].rm_eo);
227 
228 	return len - (pmatch[1].rm_eo - pmatch[1].rm_so);
229 }
230 
231 static int search_pattern(regex_t *pattern, char *pattern_str, char *buf)
232 {
233 	int err, val_len;
234 	regmatch_t pmatch[2];
235 
236 	err = regexec(pattern, buf, 2, pmatch, REG_NOTBOL);
237 	if (err != 0 || pmatch[1].rm_so == -1) {
238 		if (debug_on)
239 			fprintf(stderr, "no matching pattern in %s\n", buf);
240 		return -1;
241 	}
242 	val_len = pmatch[1].rm_eo - pmatch[1].rm_so;
243 
244 	memcpy(pattern_str, buf + pmatch[1].rm_so, val_len);
245 
246 	return 0;
247 }
248 
249 static bool check_regcomp(regex_t *pattern, const char *regex)
250 {
251 	int err;
252 
253 	err = regcomp(pattern, regex, REG_EXTENDED | REG_NEWLINE);
254 	if (err != 0 || pattern->re_nsub != 1) {
255 		fprintf(stderr, "Invalid pattern %s code %d\n", regex, err);
256 		return false;
257 	}
258 	return true;
259 }
260 
261 static char **explode(char sep, const char *str, int *size)
262 {
263 	int count = 0, len = strlen(str);
264 	int lastindex = -1, j = 0;
265 
266 	for (int i = 0; i < len; i++)
267 		if (str[i] == sep)
268 			count++;
269 	char **ret = calloc(++count, sizeof(char *));
270 
271 	for (int i = 0; i < len; i++) {
272 		if (str[i] == sep) {
273 			ret[j] = calloc(i - lastindex, sizeof(char));
274 			memcpy(ret[j++], str + lastindex + 1, i - lastindex - 1);
275 			lastindex = i;
276 		}
277 	}
278 	if (lastindex <= len - 1) {
279 		ret[j] = calloc(len - lastindex, sizeof(char));
280 		memcpy(ret[j++], str + lastindex + 1, strlen(str) - 1 - lastindex);
281 	}
282 	*size = j;
283 	return ret;
284 }
285 
286 static void free_explode(char **arr, int size)
287 {
288 	for (int i = 0; i < size; i++)
289 		free(arr[i]);
290 	free(arr);
291 }
292 
293 # define FIELD_BUFF 25
294 
295 static int get_page_num(char *buf)
296 {
297 	int order_val;
298 	char order_str[FIELD_BUFF] = {0};
299 	char *endptr;
300 
301 	search_pattern(&order_pattern, order_str, buf);
302 	errno = 0;
303 	order_val = strtol(order_str, &endptr, 10);
304 	if (order_val > 64 || errno != 0 || endptr == order_str || *endptr != '\0') {
305 		if (debug_on)
306 			fprintf(stderr, "wrong order in follow buf:\n%s\n", buf);
307 		return 0;
308 	}
309 
310 	return 1 << order_val;
311 }
312 
313 static pid_t get_pid(char *buf)
314 {
315 	pid_t pid;
316 	char pid_str[FIELD_BUFF] = {0};
317 	char *endptr;
318 
319 	search_pattern(&pid_pattern, pid_str, buf);
320 	errno = 0;
321 	pid = strtol(pid_str, &endptr, 10);
322 	if (errno != 0 || endptr == pid_str || *endptr != '\0') {
323 		if (debug_on)
324 			fprintf(stderr, "wrong/invalid pid in follow buf:\n%s\n", buf);
325 		return -1;
326 	}
327 
328 	return pid;
329 
330 }
331 
332 static pid_t get_tgid(char *buf)
333 {
334 	pid_t tgid;
335 	char tgid_str[FIELD_BUFF] = {0};
336 	char *endptr;
337 
338 	search_pattern(&tgid_pattern, tgid_str, buf);
339 	errno = 0;
340 	tgid = strtol(tgid_str, &endptr, 10);
341 	if (errno != 0 || endptr == tgid_str || *endptr != '\0') {
342 		if (debug_on)
343 			fprintf(stderr, "wrong/invalid tgid in follow buf:\n%s\n", buf);
344 		return -1;
345 	}
346 
347 	return tgid;
348 
349 }
350 
351 static __u64 get_ts_nsec(char *buf)
352 {
353 	__u64 ts_nsec;
354 	char ts_nsec_str[FIELD_BUFF] = {0};
355 	char *endptr;
356 
357 	search_pattern(&ts_nsec_pattern, ts_nsec_str, buf);
358 	errno = 0;
359 	ts_nsec = strtoull(ts_nsec_str, &endptr, 10);
360 	if (errno != 0 || endptr == ts_nsec_str || *endptr != '\0') {
361 		if (debug_on)
362 			fprintf(stderr, "wrong ts_nsec in follow buf:\n%s\n", buf);
363 		return -1;
364 	}
365 
366 	return ts_nsec;
367 }
368 
369 static char *get_comm(char *buf)
370 {
371 	char *comm_str = malloc(TASK_COMM_LEN);
372 
373 	memset(comm_str, 0, TASK_COMM_LEN);
374 
375 	search_pattern(&comm_pattern, comm_str, buf);
376 	errno = 0;
377 	if (errno != 0) {
378 		if (debug_on)
379 			fprintf(stderr, "wrong comm in follow buf:\n%s\n", buf);
380 		return NULL;
381 	}
382 
383 	return comm_str;
384 }
385 
386 static int get_arg_type(const char *arg)
387 {
388 	if (!strcmp(arg, "pid") || !strcmp(arg, "p"))
389 		return ARG_PID;
390 	else if (!strcmp(arg, "tgid") || !strcmp(arg, "tg"))
391 		return ARG_TGID;
392 	else if (!strcmp(arg, "name") || !strcmp(arg, "n"))
393 		return  ARG_COMM;
394 	else if (!strcmp(arg, "stacktrace") || !strcmp(arg, "st"))
395 		return ARG_STACKTRACE;
396 	else if (!strcmp(arg, "txt") || !strcmp(arg, "T"))
397 		return ARG_TXT;
398 	else if (!strcmp(arg, "alloc_ts") || !strcmp(arg, "at"))
399 		return ARG_ALLOC_TS;
400 	else if (!strcmp(arg, "allocator") || !strcmp(arg, "ator"))
401 		return ARG_ALLOCATOR;
402 	else {
403 		return ARG_UNKNOWN;
404 	}
405 }
406 
407 static int get_allocator(const char *buf, const char *migrate_info)
408 {
409 	char *tmp, *first_line, *second_line;
410 	int allocator = 0;
411 
412 	if (strstr(migrate_info, "CMA"))
413 		allocator |= ALLOCATOR_CMA;
414 	if (strstr(migrate_info, "slab"))
415 		allocator |= ALLOCATOR_SLAB;
416 	tmp = strstr(buf, "__vmalloc_node_range");
417 	if (tmp) {
418 		second_line = tmp;
419 		while (*tmp != '\n')
420 			tmp--;
421 		tmp--;
422 		while (*tmp != '\n')
423 			tmp--;
424 		first_line = ++tmp;
425 		tmp = strstr(tmp, "alloc_pages");
426 		if (tmp && first_line <= tmp && tmp < second_line)
427 			allocator |= ALLOCATOR_VMALLOC;
428 	}
429 	if (allocator == 0)
430 		allocator = ALLOCATOR_OTHERS;
431 	return allocator;
432 }
433 
434 static bool match_num_list(int num, int *list, int list_size)
435 {
436 	for (int i = 0; i < list_size; ++i)
437 		if (list[i] == num)
438 			return true;
439 	return false;
440 }
441 
442 static bool match_str_list(const char *str, char **list, int list_size)
443 {
444 	for (int i = 0; i < list_size; ++i)
445 		if (!strcmp(list[i], str))
446 			return true;
447 	return false;
448 }
449 
450 static bool is_need(char *buf)
451 {
452 	if ((filter & FILTER_PID) && !match_num_list(get_pid(buf), fc.pids, fc.pids_size))
453 		return false;
454 	if ((filter & FILTER_TGID) &&
455 		!match_num_list(get_tgid(buf), fc.tgids, fc.tgids_size))
456 		return false;
457 
458 	char *comm = get_comm(buf);
459 
460 	if ((filter & FILTER_COMM) &&
461 	!match_str_list(comm, fc.comms, fc.comms_size)) {
462 		free(comm);
463 		return false;
464 	}
465 	free(comm);
466 	return true;
467 }
468 
469 static bool add_list(char *buf, int len, char *ext_buf)
470 {
471 	if (list_size == max_size) {
472 		fprintf(stderr, "max_size too small??\n");
473 		return false;
474 	}
475 	if (!is_need(buf))
476 		return true;
477 	list[list_size].pid = get_pid(buf);
478 	list[list_size].tgid = get_tgid(buf);
479 	list[list_size].comm = get_comm(buf);
480 	list[list_size].txt = malloc(len+1);
481 	if (!list[list_size].txt) {
482 		fprintf(stderr, "Out of memory\n");
483 		return false;
484 	}
485 	memcpy(list[list_size].txt, buf, len);
486 	if (sc.cmps[0] != compare_ts) {
487 		len = remove_pattern(&ts_nsec_pattern, list[list_size].txt, len);
488 	}
489 	list[list_size].txt[len] = 0;
490 	list[list_size].len = len;
491 	list[list_size].num = 1;
492 	list[list_size].page_num = get_page_num(buf);
493 
494 	list[list_size].stacktrace = strchr(list[list_size].txt, '\n') ?: "";
495 	if (*list[list_size].stacktrace == '\n')
496 		list[list_size].stacktrace++;
497 	list[list_size].ts_nsec = get_ts_nsec(buf);
498 	list[list_size].allocator = get_allocator(buf, ext_buf);
499 	list_size++;
500 	if (list_size % 1000 == 0) {
501 		printf("loaded %d\r", list_size);
502 		fflush(stdout);
503 	}
504 	return true;
505 }
506 
507 static bool parse_cull_args(const char *arg_str)
508 {
509 	int size = 0;
510 	char **args = explode(',', arg_str, &size);
511 
512 	for (int i = 0; i < size; ++i) {
513 		int arg_type = get_arg_type(args[i]);
514 
515 		if (arg_type == ARG_PID)
516 			cull |= CULL_PID;
517 		else if (arg_type == ARG_TGID)
518 			cull |= CULL_TGID;
519 		else if (arg_type == ARG_COMM)
520 			cull |= CULL_COMM;
521 		else if (arg_type == ARG_STACKTRACE)
522 			cull |= CULL_STACKTRACE;
523 		else if (arg_type == ARG_ALLOCATOR)
524 			cull |= CULL_ALLOCATOR;
525 		else {
526 			free_explode(args, size);
527 			return false;
528 		}
529 	}
530 	free_explode(args, size);
531 	if (sc.size == 0)
532 		set_single_cmp(compare_num, SORT_DESC);
533 	return true;
534 }
535 
536 static void set_single_cmp(int (*cmp)(const void *, const void *), int sign)
537 {
538 	if (sc.signs == NULL || sc.size < 1)
539 		sc.signs = calloc(1, sizeof(int));
540 	sc.signs[0] = sign;
541 	if (sc.cmps == NULL || sc.size < 1)
542 		sc.cmps = calloc(1, sizeof(int *));
543 	sc.cmps[0] = cmp;
544 	sc.size = 1;
545 }
546 
547 static bool parse_sort_args(const char *arg_str)
548 {
549 	int size = 0;
550 
551 	if (sc.size != 0) { /* reset sort_condition */
552 		free(sc.signs);
553 		free(sc.cmps);
554 		size = 0;
555 	}
556 
557 	char **args = explode(',', arg_str, &size);
558 
559 	sc.signs = calloc(size, sizeof(int));
560 	sc.cmps = calloc(size, sizeof(int *));
561 	for (int i = 0; i < size; ++i) {
562 		int offset = 0;
563 
564 		sc.signs[i] = SORT_ASC;
565 		if (args[i][0] == '-' || args[i][0] == '+') {
566 			if (args[i][0] == '-')
567 				sc.signs[i] = SORT_DESC;
568 			offset = 1;
569 		}
570 
571 		int arg_type = get_arg_type(args[i]+offset);
572 
573 		if (arg_type == ARG_PID)
574 			sc.cmps[i] = compare_pid;
575 		else if (arg_type == ARG_TGID)
576 			sc.cmps[i] = compare_tgid;
577 		else if (arg_type == ARG_COMM)
578 			sc.cmps[i] = compare_comm;
579 		else if (arg_type == ARG_STACKTRACE)
580 			sc.cmps[i] = compare_stacktrace;
581 		else if (arg_type == ARG_ALLOC_TS)
582 			sc.cmps[i] = compare_ts;
583 		else if (arg_type == ARG_TXT)
584 			sc.cmps[i] = compare_txt;
585 		else if (arg_type == ARG_ALLOCATOR)
586 			sc.cmps[i] = compare_allocator;
587 		else {
588 			free_explode(args, size);
589 			sc.size = 0;
590 			return false;
591 		}
592 	}
593 	sc.size = size;
594 	free_explode(args, size);
595 	return true;
596 }
597 
598 static int *parse_nums_list(char *arg_str, int *list_size)
599 {
600 	int size = 0;
601 	char **args = explode(',', arg_str, &size);
602 	int *list = calloc(size, sizeof(int));
603 
604 	errno = 0;
605 	for (int i = 0; i < size; ++i) {
606 		char *endptr = NULL;
607 
608 		list[i] = strtol(args[i], &endptr, 10);
609 		if (errno != 0 || endptr == args[i] || *endptr != '\0') {
610 			free(list);
611 			return NULL;
612 		}
613 	}
614 	*list_size = size;
615 	free_explode(args, size);
616 	return list;
617 }
618 
619 static void print_allocator(FILE *out, int allocator)
620 {
621 	fprintf(out, "allocated by ");
622 	if (allocator & ALLOCATOR_CMA)
623 		fprintf(out, "CMA ");
624 	if (allocator & ALLOCATOR_SLAB)
625 		fprintf(out, "SLAB ");
626 	if (allocator & ALLOCATOR_VMALLOC)
627 		fprintf(out, "VMALLOC ");
628 	if (allocator & ALLOCATOR_OTHERS)
629 		fprintf(out, "OTHERS ");
630 }
631 
632 #define BUF_SIZE	(128 * 1024)
633 
634 static void usage(void)
635 {
636 	printf("Usage: ./page_owner_sort [OPTIONS] <input> <output>\n"
637 		"-a\t\t\tSort by memory allocation time.\n"
638 		"-m\t\t\tSort by total memory.\n"
639 		"-n\t\t\tSort by task command name.\n"
640 		"-p\t\t\tSort by pid.\n"
641 		"-P\t\t\tSort by tgid.\n"
642 		"-s\t\t\tSort by the stacktrace.\n"
643 		"-t\t\t\tSort by number of times record is seen (default).\n\n"
644 		"--pid <pidlist>\t\tSelect by pid. This selects the information"
645 		" of\n\t\t\tblocks whose process ID numbers appear in <pidlist>.\n"
646 		"--tgid <tgidlist>\tSelect by tgid. This selects the information"
647 		" of\n\t\t\tblocks whose Thread Group ID numbers appear in "
648 		"<tgidlist>.\n"
649 		"--name <cmdlist>\tSelect by command name. This selects the"
650 		" information\n\t\t\tof blocks whose command name appears in"
651 		" <cmdlist>.\n"
652 		"--cull <rules>\t\tCull by user-defined rules. <rules> is a "
653 		"single\n\t\t\targument in the form of a comma-separated list "
654 		"with some\n\t\t\tcommon fields predefined (pid, tgid, comm, "
655 		"stacktrace, allocator)\n"
656 		"--sort <order>\t\tSpecify sort order as: [+|-]key[,[+|-]key[,...]]\n"
657 	);
658 }
659 
660 int main(int argc, char **argv)
661 {
662 	FILE *fin, *fout;
663 	char *buf, *ext_buf;
664 	int i, count, compare_flag;
665 	struct stat st;
666 	int opt;
667 	struct option longopts[] = {
668 		{ "pid", required_argument, NULL, 1 },
669 		{ "tgid", required_argument, NULL, 2 },
670 		{ "name", required_argument, NULL, 3 },
671 		{ "cull",  required_argument, NULL, 4 },
672 		{ "sort",  required_argument, NULL, 5 },
673 		{ 0, 0, 0, 0},
674 	};
675 
676 	compare_flag = COMP_NO_FLAG;
677 
678 	while ((opt = getopt_long(argc, argv, "admnpstP", 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 1:
705 			filter = filter | FILTER_PID;
706 			fc.pids = parse_nums_list(optarg, &fc.pids_size);
707 			if (fc.pids == NULL) {
708 				fprintf(stderr, "wrong/invalid pid in from the command line:%s\n",
709 						optarg);
710 				exit(1);
711 			}
712 			break;
713 		case 2:
714 			filter = filter | FILTER_TGID;
715 			fc.tgids = parse_nums_list(optarg, &fc.tgids_size);
716 			if (fc.tgids == NULL) {
717 				fprintf(stderr, "wrong/invalid tgid in from the command line:%s\n",
718 						optarg);
719 				exit(1);
720 			}
721 			break;
722 		case 3:
723 			filter = filter | FILTER_COMM;
724 			fc.comms = explode(',', optarg, &fc.comms_size);
725 			break;
726 		case 4:
727 			if (!parse_cull_args(optarg)) {
728 				fprintf(stderr, "wrong argument after --cull option:%s\n",
729 						optarg);
730 				exit(1);
731 			}
732 			break;
733 		case 5:
734 			if (!parse_sort_args(optarg)) {
735 				fprintf(stderr, "wrong argument after --sort option:%s\n",
736 						optarg);
737 				exit(1);
738 			}
739 			break;
740 		default:
741 			usage();
742 			exit(1);
743 		}
744 
745 	if (optind >= (argc - 1)) {
746 		usage();
747 		exit(1);
748 	}
749 
750 	/* Only one compare option is allowed, yet we also want handle the
751 	 * default case were no option is provided, but we still want to
752 	 * match the behavior of the -t option (compare by number of times
753 	 * a record is seen
754 	 */
755 	switch (compare_flag) {
756 	case COMP_ALLOC:
757 		set_single_cmp(compare_ts, SORT_ASC);
758 		break;
759 	case COMP_PAGE_NUM:
760 		set_single_cmp(compare_page_num, SORT_DESC);
761 		break;
762 	case COMP_PID:
763 		set_single_cmp(compare_pid, SORT_ASC);
764 		break;
765 	case COMP_STACK:
766 		set_single_cmp(compare_stacktrace, SORT_ASC);
767 		break;
768 	case COMP_NO_FLAG:
769 	case COMP_NUM:
770 		set_single_cmp(compare_num, SORT_DESC);
771 		break;
772 	case COMP_TGID:
773 		set_single_cmp(compare_tgid, SORT_ASC);
774 		break;
775 	case COMP_COMM:
776 		set_single_cmp(compare_comm, SORT_ASC);
777 		break;
778 	default:
779 		usage();
780 		exit(1);
781 	}
782 
783 	fin = fopen(argv[optind], "r");
784 	fout = fopen(argv[optind + 1], "w");
785 	if (!fin || !fout) {
786 		usage();
787 		perror("open: ");
788 		exit(1);
789 	}
790 
791 	if (!check_regcomp(&order_pattern, "order\\s*([0-9]*),"))
792 		goto out_order;
793 	if (!check_regcomp(&pid_pattern, "pid\\s*([0-9]*),"))
794 		goto out_pid;
795 	if (!check_regcomp(&tgid_pattern, "tgid\\s*([0-9]*) "))
796 		goto out_tgid;
797 	if (!check_regcomp(&comm_pattern, "tgid\\s*[0-9]*\\s*\\((.*)\\),\\s*ts"))
798 		goto out_comm;
799 	if (!check_regcomp(&ts_nsec_pattern, "ts\\s*([0-9]*)\\s*ns"))
800 		goto out_ts;
801 
802 	fstat(fileno(fin), &st);
803 	max_size = st.st_size / 100; /* hack ... */
804 
805 	list = malloc(max_size * sizeof(*list));
806 	buf = malloc(BUF_SIZE);
807 	ext_buf = malloc(BUF_SIZE);
808 	if (!list || !buf || !ext_buf) {
809 		fprintf(stderr, "Out of memory\n");
810 		goto out_free;
811 	}
812 
813 	for ( ; ; ) {
814 		int buf_len = read_block(buf, ext_buf, BUF_SIZE, fin);
815 
816 		if (buf_len < 0)
817 			break;
818 		if (!add_list(buf, buf_len, ext_buf))
819 			goto out_free;
820 	}
821 
822 	printf("loaded %d\n", list_size);
823 
824 	printf("sorting ....\n");
825 
826 	qsort(list, list_size, sizeof(list[0]), compare_cull_condition);
827 
828 	printf("culling\n");
829 
830 	for (i = count = 0; i < list_size; i++) {
831 		if (count == 0 ||
832 		    compare_cull_condition((void *)(&list[count-1]), (void *)(&list[i])) != 0) {
833 			list[count++] = list[i];
834 		} else {
835 			list[count-1].num += list[i].num;
836 			list[count-1].page_num += list[i].page_num;
837 		}
838 	}
839 
840 	qsort(list, count, sizeof(list[0]), compare_sort_condition);
841 
842 	for (i = 0; i < count; i++) {
843 		if (cull == 0) {
844 			fprintf(fout, "%d times, %d pages, ", list[i].num, list[i].page_num);
845 			print_allocator(fout, list[i].allocator);
846 			fprintf(fout, ":\n%s\n", list[i].txt);
847 		}
848 		else {
849 			fprintf(fout, "%d times, %d pages",
850 					list[i].num, list[i].page_num);
851 			if (cull & CULL_PID || filter & FILTER_PID)
852 				fprintf(fout, ", PID %d", list[i].pid);
853 			if (cull & CULL_TGID || filter & FILTER_TGID)
854 				fprintf(fout, ", TGID %d", list[i].tgid);
855 			if (cull & CULL_COMM || filter & FILTER_COMM)
856 				fprintf(fout, ", task_comm_name: %s", list[i].comm);
857 			if (cull & CULL_ALLOCATOR) {
858 				fprintf(fout, ", ");
859 				print_allocator(fout, list[i].allocator);
860 			}
861 			if (cull & CULL_STACKTRACE)
862 				fprintf(fout, ":\n%s", list[i].stacktrace);
863 			fprintf(fout, "\n");
864 		}
865 	}
866 
867 out_free:
868 	if (ext_buf)
869 		free(ext_buf);
870 	if (buf)
871 		free(buf);
872 	if (list)
873 		free(list);
874 out_ts:
875 	regfree(&ts_nsec_pattern);
876 out_comm:
877 	regfree(&comm_pattern);
878 out_tgid:
879 	regfree(&tgid_pattern);
880 out_pid:
881 	regfree(&pid_pattern);
882 out_order:
883 	regfree(&order_pattern);
884 
885 	return 0;
886 }
887