Merge commit 'stark/master'
[adu.git] / adu.c
1 #include "adu.h"
2 #include <dirent.h> /* readdir() */
3
4 #include "gcc-compat.h"
5 #include "osl.h"
6 #include "fd.h"
7 #include "hash.h"
8 #include "string.h"
9 #include "error.h"
10
11 DEFINE_ERRLIST;
12
13 /** evaluates to 1 if x < y, to -1 if x > y and to 0 if x == y */
14 #define NUM_COMPARE(x, y) ((int)((x) < (y)) - (int)((x) > (y)))
15
16
17 /**
18 * The log function.
19 *
20 * \param ll Loglevel.
21 * \param fml Usual format string.
22 *
23 * All XXX_LOG() macros use this function.
24 */
25 __printf_2_3 void __log(int ll, const char* fmt,...)
26 {
27 va_list argp;
28 FILE *outfd;
29 struct tm *tm;
30 time_t t1;
31 char str[255] = "";
32
33 if (ll < 4)
34 return;
35 outfd = stderr;
36 time(&t1);
37 tm = localtime(&t1);
38 strftime(str, sizeof(str), "%b %d %H:%M:%S", tm);
39 fprintf(outfd, "%s ", str);
40 va_start(argp, fmt);
41 vfprintf(outfd, fmt, argp);
42 va_end(argp);
43 }
44
45 /**
46 * Compare the size of two directories
47 *
48 * \param obj1 Pointer to the first object.
49 * \param obj2 Pointer to the second object.
50 *
51 * This function first compares the size values as usual integers. If they compare as
52 * equal, the address of \a obj1 and \a obj2 are compared. So this compare function
53 * returns zero if and only if \a obj1 and \a obj2 point to the same memory area.
54 */
55 static int size_compare(const struct osl_object *obj1, const struct osl_object *obj2)
56 {
57 uint64_t d1 = *(uint64_t *)obj1->data;
58 uint64_t d2 = *(uint64_t *)obj2->data;
59 int ret = NUM_COMPARE(d2, d1);
60
61 if (ret)
62 return ret;
63 //INFO_LOG("addresses: %p, %p\n", obj1->data, obj2->data);
64 return NUM_COMPARE(obj2->data, obj1->data);
65 }
66
67 /**
68 * Compare two osl objects of string type.
69 *
70 * \param obj1 Pointer to the first object.
71 * \param obj2 Pointer to the second object.
72 *
73 * In any case, only \p MIN(obj1->size, obj2->size) characters of each string
74 * are taken into account.
75 *
76 * \return It returns an integer less than, equal to, or greater than zero if
77 * \a obj1 is found, respectively, to be less than, to match, or be greater than
78 * obj2.
79 *
80 * \sa strcmp(3), strncmp(3), osl_compare_func.
81 */
82 int string_compare(const struct osl_object *obj1, const struct osl_object *obj2)
83 {
84 const char *str1 = (const char *)obj1->data;
85 const char *str2 = (const char *)obj2->data;
86 return strncmp(str1, str2, MIN(obj1->size, obj2->size));
87 }
88
89 /** The columns of the directory table. */
90 enum dir_table_columns {
91 /** The name of the directory. */
92 DT_NAME,
93 /** The dir count number. */
94 DT_NUM,
95 /** The number of bytes of all regular files. */
96 DT_BYTES,
97 /** The number of all regular files. */
98 DT_FILES,
99 /** Number of columns in this table. */
100 NUM_DT_COLUMNS
101 };
102
103 static struct osl_column_description dir_table_cols[] = {
104 [DT_NAME] = {
105 .storage_type = OSL_MAPPED_STORAGE,
106 .storage_flags = OSL_RBTREE | OSL_UNIQUE,
107 .name = "dir",
108 .compare_function = string_compare,
109 },
110 [DT_NUM] = {
111 .storage_type = OSL_MAPPED_STORAGE,
112 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
113 .name = "num",
114 .compare_function = uint32_compare,
115 .data_size = sizeof(uint32_t)
116 },
117 [DT_BYTES] = {
118 .storage_type = OSL_MAPPED_STORAGE,
119 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
120 .compare_function = size_compare,
121 .name = "num_bytes",
122 .data_size = sizeof(uint64_t)
123 },
124 [DT_FILES] = {
125 .storage_type = OSL_MAPPED_STORAGE,
126 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
127 .compare_function = size_compare,
128 .name = "num_files",
129 .data_size = sizeof(uint64_t)
130 }
131 };
132
133 static struct osl_table_description dir_table_desc = {
134 .name = "dir_table",
135 .num_columns = NUM_DT_COLUMNS,
136 .flags = 0,
137 .column_descriptions = dir_table_cols,
138 .dir = "/tmp/adu"
139 };
140
141 /** The columns of the id table. */
142 enum id_table_columns {
143 /** The user id. */
144 IDT_UID,
145 /** The number of bytes of all regular files owned by this id. */
146 IDT_BYTES,
147 /** The number of regular files owned by this id. */
148 IDT_FILES,
149 /** The user table for this uid. */
150 IDT_TABLE,
151 /** Number of columns in this table. */
152 NUM_IDT_COLUMNS
153 };
154
155 static struct osl_column_description id_table_cols[] = {
156 [IDT_UID] = {
157 .storage_type = OSL_MAPPED_STORAGE,
158 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
159 .name = "uid",
160 .compare_function = uint32_compare,
161 .data_size = sizeof(uint32_t)
162 },
163 [IDT_BYTES] = {
164 .storage_type = OSL_MAPPED_STORAGE,
165 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
166 .compare_function = size_compare,
167 .name = "num_bytes",
168 .data_size = sizeof(uint64_t)
169 },
170 [IDT_FILES] = {
171 .storage_type = OSL_MAPPED_STORAGE,
172 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
173 .compare_function = size_compare,
174 .name = "num_filess",
175 .data_size = sizeof(uint64_t)
176 },
177 [IDT_TABLE] = {
178 .storage_type = OSL_NO_STORAGE,
179 .storage_flags = OSL_FIXED_SIZE | OSL_UNIQUE,
180 .name = "user_table",
181 .data_size = sizeof(void *)
182 }
183 };
184
185 static struct osl_table_description id_table_desc = {
186 .name = "id_table",
187 .num_columns = NUM_IDT_COLUMNS,
188 .flags = 0,
189 .column_descriptions = id_table_cols,
190 .dir = "/tmp/adu"
191 };
192
193 /** The columns of the id table. */
194 enum user_table_columns {
195 /** The numer of the directory. */
196 UT_DIR_NUM,
197 /** The number of bytes of all regular files in this dir owned by this id. */
198 UT_BYTES,
199 /** The number of files in this dir owned by this id. */
200 UT_FILES,
201 /** Number of columns in this table. */
202 NUM_UT_COLUMNS
203 };
204
205 static struct osl_column_description user_table_cols[] = {
206 [UT_DIR_NUM] = {
207 .storage_type = OSL_MAPPED_STORAGE,
208 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE | OSL_UNIQUE,
209 .name = "dir_num",
210 .compare_function = uint32_compare,
211 .data_size = sizeof(uint32_t)
212 },
213 [IDT_BYTES] = {
214 .storage_type = OSL_MAPPED_STORAGE,
215 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
216 .compare_function = size_compare,
217 .name = "num_bytes",
218 .data_size = sizeof(uint64_t)
219 },
220 [IDT_FILES] = {
221 .storage_type = OSL_MAPPED_STORAGE,
222 .storage_flags = OSL_RBTREE | OSL_FIXED_SIZE,
223 .compare_function = size_compare,
224 .name = "num_files",
225 .data_size = sizeof(uint64_t)
226 },
227 };
228
229 static struct osl_table_description user_table_desc = {
230 .num_columns = NUM_UT_COLUMNS,
231 .flags = 0,
232 .column_descriptions = user_table_cols,
233 .dir = "/tmp/adu"
234 };
235 static struct osl_table *dir_table;
236 static struct osl_table *id_table;
237
238 static int create_tables(void)
239 {
240 int ret = osl_create_table(&dir_table_desc);
241 if (ret < 0)
242 return ret;
243 ret = osl_create_table(&id_table_desc);
244 if (ret < 0)
245 return ret;
246 return 1;
247 }
248
249 int add_directory(char *dirname, uint32_t dir_num, uint64_t *dir_size,
250 uint64_t *dir_files)
251 {
252 struct osl_object dir_objects[NUM_DT_COLUMNS];
253
254 INFO_LOG("adding #%u: %s\n", dir_num, dirname);
255 dir_objects[DT_NAME].data = dirname;
256 dir_objects[DT_NAME].size = strlen(dirname) + 1;
257 dir_objects[DT_NUM].data = &dir_num;
258 dir_objects[DT_NUM].size = sizeof(dir_num);
259 dir_objects[DT_BYTES].data = dir_size;
260 dir_objects[DT_BYTES].size = sizeof(*dir_size);
261 dir_objects[DT_FILES].data = dir_files;
262 dir_objects[DT_FILES].size = sizeof(*dir_files);
263
264 return osl_add_row(dir_table, dir_objects);
265 }
266
267 int create_and_open_user_table(uint32_t uid, struct osl_table **t)
268 {
269 int ret;
270 struct osl_table_description *desc = para_malloc(sizeof(*desc));
271
272 desc->num_columns = NUM_UT_COLUMNS;
273 desc->flags = 0;
274 desc->column_descriptions = user_table_cols;
275 desc->dir = para_strdup("/tmp/adu");
276 desc->name = make_message("%u", uid);
277 INFO_LOG("................................. %u\n", uid);
278 // user_table_desc.name = make_message("%u", uid);
279 ret = osl_create_table(desc);
280 if (ret < 0)
281 return ret;
282 return osl_open_table(desc, t);
283 }
284
285 static int insert_id_row(uint32_t uid, struct osl_table *t, struct osl_row **row)
286 {
287 struct osl_object objects[NUM_IDT_COLUMNS];
288 uint64_t num = 0;
289
290 struct osl_table **table_ptr = para_malloc(sizeof(*table_ptr));
291 *table_ptr = t;
292
293 INFO_LOG("§§§§§§§§§§§§§§§§§§§§§ uid: %d, t: %p\n", uid, t);
294 objects[IDT_UID].data = &uid;
295 objects[IDT_UID].size = sizeof(uid);
296 objects[IDT_BYTES].data = &num;
297 objects[IDT_BYTES].size = sizeof(num);
298 objects[IDT_FILES].data = &num;
299 objects[IDT_FILES].size = sizeof(num);
300 objects[IDT_TABLE].data = table_ptr;
301 objects[IDT_TABLE].size = sizeof(*table_ptr);
302 return osl_add_and_get_row(id_table, objects, row);
303 }
304
305 static int get_user_table(struct osl_row *row, struct osl_table **t)
306 {
307 struct osl_object obj;
308
309 int ret = osl_get_object(id_table, row, IDT_TABLE, &obj);
310 if (ret < 0)
311 return ret;
312 *t = *(struct osl_table **)obj.data;
313 INFO_LOG("^^^^^^^^^^^^^^^^^^ t: %p\n", *t);
314 return 1;
315 }
316
317 static int add_id_bytes(struct osl_row *row, uint64_t *add)
318 {
319 uint64_t num;
320 struct osl_object obj1, obj2 = {.data = &num, .size = sizeof(num)};
321
322 /* update number of bytes */
323 int ret = osl_get_object(id_table, row, IDT_BYTES, &obj1);
324 if (ret < 0)
325 return ret;
326 num = *(uint64_t *)obj1.data + *add;
327 ret = osl_update_object(id_table, row, IDT_BYTES, &obj2);
328 if (ret < 0)
329 return ret;
330 /* increment number of files */
331 ret = osl_get_object(id_table, row, IDT_FILES, &obj1);
332 if (ret < 0)
333 return ret;
334 num = *(uint64_t *)obj1.data + 1;
335 return osl_update_object(id_table, row, IDT_FILES, &obj2);
336 }
337
338 static int update_user_row(struct osl_table *t, uint32_t dir_num,
339 uint64_t *add)
340 {
341 struct osl_row *row;
342 struct osl_object obj = {.data = &dir_num, .size = sizeof(dir_num)};
343
344 int ret = osl_get_row(t, UT_DIR_NUM, &obj, &row);
345
346 if (ret < 0 && ret != -E_RB_KEY_NOT_FOUND)
347 return ret;
348 if (ret < 0) { /* this is the first file we add */
349 struct osl_object objects[NUM_UT_COLUMNS];
350 uint64_t num_files = 1;
351
352 objects[UT_DIR_NUM].data = &dir_num;
353 objects[UT_DIR_NUM].size = sizeof(dir_num);
354 objects[UT_BYTES].data = add;
355 objects[UT_BYTES].size = sizeof(*add);
356 objects[UT_FILES].data = &num_files;
357 objects[UT_FILES].size = sizeof(num_files);
358 INFO_LOG("######################### ret: %d\n", ret);
359 ret = osl_add_row(t, objects);
360 INFO_LOG("######################### ret: %d\n", ret);
361 return ret;
362 } else { /* add size and increment file count */
363 uint64_t num;
364 struct osl_object obj1, obj2 = {.data = &num, .size = sizeof(num)};
365
366 ret = osl_get_object(t, row, UT_BYTES, &obj1);
367 if (ret < 0)
368 return ret;
369 num = *(uint64_t *)obj1.data + *add;
370 ret = osl_update_object(t, row, UT_BYTES, &obj2);
371 if (ret < 0)
372 return ret;
373 ret = osl_get_object(t, row, UT_FILES, &obj1);
374 if (ret < 0)
375 return ret;
376 num = *(uint64_t *)obj1.data + 1;
377 return osl_update_object(t, row, UT_FILES, &obj2);
378 }
379 }
380
381 static uint32_t dir_num;
382
383 int scan_dir(char *dirname)
384 {
385 DIR *dir;
386 struct dirent *entry;
387 int ret, cwd_fd, ret2;
388 uint64_t dir_size = 0, dir_files = 0;
389 struct osl_object obj;
390
391 INFO_LOG("----------------- %s\n", dirname);
392 ret = para_opendir(dirname, &dir, &cwd_fd);
393 if (ret < 0) {
394 if (ret != -ERRNO_TO_ERROR(EACCES))
395 return ret;
396 WARNING_LOG("permission denied for %s\n", dirname);
397 return 1;
398 }
399 while ((entry = readdir(dir))) {
400 mode_t m;
401 char *tmp;
402 struct stat s;
403 uint32_t uid;
404 uint64_t size;
405 struct osl_row *id_row;
406 struct osl_table *user_table;
407
408 if (!strcmp(entry->d_name, "."))
409 continue;
410 if (!strcmp(entry->d_name, ".."))
411 continue;
412 if (lstat(entry->d_name, &s) == -1)
413 continue;
414 m = s.st_mode;
415 if (!S_ISREG(m) && !S_ISDIR(m))
416 continue;
417 tmp = make_message("%s/%s", dirname, entry->d_name);
418 if (S_ISDIR(m)) {
419 ret = scan_dir(tmp);
420 free(tmp);
421 if (ret < 0)
422 goto out;
423 continue;
424 }
425 /* regular file */
426 size = s.st_size;
427 dir_size += size;
428 dir_files++;
429 uid = s.st_uid;
430 INFO_LOG("++++++++++++++++++++++++++ %s, uid: %u\n", entry->d_name, uid);
431 obj.data = &uid;
432 obj.size = sizeof(uid);
433 ret = osl_get_row(id_table, IDT_UID, &obj, &id_row);
434 if (ret < 0 && ret != -E_RB_KEY_NOT_FOUND)
435 goto out;
436 if (ret < 0) {
437 ret = create_and_open_user_table(uid, &user_table);
438 if (ret < 0)
439 goto out;
440 ret = insert_id_row(uid, user_table, &id_row);
441 if (ret < 0)
442 goto out;
443 } else {
444 ret = get_user_table(id_row, &user_table);
445 if (ret < 0)
446 goto out;
447 }
448 ret = add_id_bytes(id_row, &size);
449 if (ret < 0)
450 goto out;
451 INFO_LOG("user_table: %p\n", user_table);
452 ret = update_user_row(user_table, dir_num, &size);
453 INFO_LOG("update_user ret: %d\n", ret);
454 if (ret < 0)
455 goto out;
456 }
457 ret = add_directory(dirname, dir_num++, &dir_size, &dir_files);
458 out:
459 closedir(dir);
460 ret2 = para_fchdir(cwd_fd);
461 if (ret2 < 0 && ret >= 0)
462 ret = ret2;
463 close(cwd_fd);
464 return ret;
465 }
466
467 static int get_dir_name(struct osl_row *row, char **name)
468 {
469 struct osl_object obj;
470 int ret = osl_get_object(dir_table, row, DT_NAME, &obj);
471
472 if (ret < 0)
473 return ret;
474 *name = obj.data;
475 return 1;
476 }
477
478 static int print_dirname_and_size(struct osl_row *row, void *data)
479 {
480 unsigned *count = data;
481 struct osl_object obj;
482 char *name;
483 int ret;
484
485 if ((*count)++ > 100)
486 return -E_LOOP_COMPLETE;
487 ret = get_dir_name(row, &name);
488 if (ret < 0)
489 return ret;
490 ret = osl_get_object(dir_table, row, DT_BYTES, &obj);
491 if (ret < 0)
492 return ret;
493 printf("%s\t%llu\n", name, *(long long unsigned *)obj.data);
494 return 1;
495 }
496
497 static int print_dirname_and_file_count(struct osl_row *row, void *data)
498 {
499 unsigned *count = data;
500 struct osl_object obj;
501 char *name;
502 int ret;
503
504 if ((*count)++ > 100)
505 return -E_LOOP_COMPLETE;
506 ret = get_dir_name(row, &name);
507 if (ret < 0)
508 return ret;
509 ret = osl_get_object(dir_table, row, DT_FILES, &obj);
510 if (ret < 0)
511 return ret;
512 printf("%s\t%llu\n", name, *(long long unsigned *)obj.data);
513 return 1;
514 }
515
516 static int print_id_stats(struct osl_row *row, __a_unused void *data)
517 {
518 struct osl_object obj;
519 uint32_t uid;
520 uint64_t bytes, files;
521 int ret = osl_get_object(id_table, row, IDT_UID, &obj);
522
523 if (ret < 0)
524 return ret;
525 uid = *(uint32_t *)obj.data;
526 ret = osl_get_object(id_table, row, IDT_BYTES, &obj);
527 if (ret < 0)
528 return ret;
529 bytes = *(uint64_t *)obj.data;
530 ret = osl_get_object(id_table, row, IDT_FILES, &obj);
531 if (ret < 0)
532 return ret;
533 files = *(uint64_t *)obj.data;
534
535 printf("%u\t%llu\t%llu\n", (unsigned)uid, (long long unsigned)files,
536 (long long unsigned)bytes);
537 return 1;
538 }
539
540 struct id_dir_stat_info {
541 unsigned count;
542 struct osl_table *user_table;
543 };
544
545 static int print_big_dir(struct osl_row *row, void *data)
546 {
547 struct id_dir_stat_info *info = data;
548 info->count++;
549 int ret;
550 struct osl_row *dir_row;
551 char *dirname;
552 uint64_t bytes;
553 struct osl_object obj;
554
555 if (info->count > 10)
556 return -E_LOOP_COMPLETE;
557 ret = osl_get_object(info->user_table, row, UT_BYTES, &obj);
558 if (ret < 0)
559 return ret;
560 bytes = *(uint64_t *)obj.data;
561 ret = osl_get_object(info->user_table, row, UT_DIR_NUM, &obj);
562 if (ret < 0)
563 return ret;
564 ret = osl_get_row(dir_table, DT_NUM, &obj, &dir_row);
565 if (ret < 0)
566 return ret;
567 ret = osl_get_object(dir_table, dir_row, DT_NAME, &obj);
568 if (ret < 0)
569 return ret;
570 dirname = obj.data;
571 printf("%s: %llu\n", dirname, (long long unsigned)bytes);
572 return 1;
573 }
574
575 static int print_id_dir_stats(struct osl_row *row, __a_unused void *data)
576 {
577 struct osl_object obj;
578 uint32_t uid;
579 int ret = osl_get_object(id_table, row, IDT_UID, &obj);
580 struct id_dir_stat_info info = {.count = 0};
581
582 if (ret < 0)
583 return ret;
584 uid = *(uint32_t *)obj.data;
585
586 ret = osl_get_object(id_table, row, IDT_TABLE, &obj);
587 if (ret < 0)
588 return ret;
589 info.user_table = *(struct osl_table **)obj.data;
590
591 printf("************************* Big dirs owned by uid %u\n", (unsigned) uid);
592 osl_rbtree_loop_reverse(info.user_table, IDT_BYTES, &info, print_big_dir);
593 return 1;
594 }
595
596 static int print_statistics(void)
597 {
598 unsigned count = 0;
599 int ret;
600
601 printf("************************* Biggest dirs\n");
602 ret = osl_rbtree_loop_reverse(dir_table, DT_BYTES, &count, print_dirname_and_size);
603 if (ret < 0 && ret != -E_LOOP_COMPLETE)
604 return ret;
605 count = 0;
606 printf("************************* dirs containing many files\n");
607 ret = osl_rbtree_loop_reverse(dir_table, DT_FILES, &count, print_dirname_and_file_count);
608 if (ret < 0 && ret != -E_LOOP_COMPLETE)
609 return ret;
610
611 printf("************************* dirs stats by owner\n");
612 ret = osl_rbtree_loop(id_table, IDT_BYTES, NULL, print_id_stats);
613 if (ret < 0)
614 return ret;
615
616 return osl_rbtree_loop(id_table, IDT_BYTES, NULL, print_id_dir_stats);
617 }
618
619
620 int main(int argc, char **argv)
621 {
622 int ret = create_tables();
623 if (ret < 0)
624 goto out;
625 ret = osl_open_table(&dir_table_desc, &dir_table);
626 if (ret < 0)
627 goto out;
628 ret = osl_open_table(&id_table_desc, &id_table);
629 if (ret < 0)
630 goto out;
631 ret = -E_SYNTAX;
632 if (argc != 2)
633 goto out;
634 ret = scan_dir(argv[1]);
635 if (ret < 0)
636 goto out;
637 print_statistics();
638 out:
639 if (ret < 0) {
640 ERROR_LOG("%s\n", error_txt(-ret));
641 return -EXIT_FAILURE;
642 }
643 return EXIT_SUCCESS;
644 }
645