1 /*
2 * GPL HEADER START
3 *
4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
5 *
6 * This program is free software; you can redistribute it and/or modify
7 * it under the terms of the GNU General Public License version 2 only,
8 * as published by the Free Software Foundation.
9 *
10 * This program is distributed in the hope that it will be useful, but
11 * WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
13 * General Public License version 2 for more details (a copy is included
14 * in the LICENSE file that accompanied this code).
15 *
16 * You should have received a copy of the GNU General Public License
17 * version 2 along with this program; If not, see
18 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
19 *
20 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
21 * CA 95054 USA or visit www.sun.com if you need additional information or
22 * have any questions.
23 *
24 * GPL HEADER END
25 */
26 /*
27 * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
28 * Use is subject to license terms.
29 *
30 * Copyright (c) 2012, 2013, Intel Corporation.
31 */
32 /*
33 * This file is part of Lustre, http://www.lustre.org/
34 * Lustre is a trademark of Sun Microsystems, Inc.
35 *
36 * lustre/fld/fld_cache.c
37 *
38 * FLD (Fids Location Database)
39 *
40 * Author: Pravin Shelar <pravin.shelar@sun.com>
41 * Author: Yury Umanets <umka@clusterfs.com>
42 */
43
44 #define DEBUG_SUBSYSTEM S_FLD
45
46 #include "../../include/linux/libcfs/libcfs.h"
47 #include <linux/module.h>
48 #include <asm/div64.h>
49
50 #include "../include/obd.h"
51 #include "../include/obd_class.h"
52 #include "../include/lustre_ver.h"
53 #include "../include/obd_support.h"
54 #include "../include/lprocfs_status.h"
55
56 #include "../include/dt_object.h"
57 #include "../include/lustre_req_layout.h"
58 #include "../include/lustre_fld.h"
59 #include "fld_internal.h"
60
61 /**
62 * create fld cache.
63 */
fld_cache_init(const char * name,int cache_size,int cache_threshold)64 struct fld_cache *fld_cache_init(const char *name,
65 int cache_size, int cache_threshold)
66 {
67 struct fld_cache *cache;
68
69 LASSERT(name != NULL);
70 LASSERT(cache_threshold < cache_size);
71
72 OBD_ALLOC_PTR(cache);
73 if (cache == NULL)
74 return ERR_PTR(-ENOMEM);
75
76 INIT_LIST_HEAD(&cache->fci_entries_head);
77 INIT_LIST_HEAD(&cache->fci_lru);
78
79 cache->fci_cache_count = 0;
80 rwlock_init(&cache->fci_lock);
81
82 strlcpy(cache->fci_name, name,
83 sizeof(cache->fci_name));
84
85 cache->fci_cache_size = cache_size;
86 cache->fci_threshold = cache_threshold;
87
88 /* Init fld cache info. */
89 memset(&cache->fci_stat, 0, sizeof(cache->fci_stat));
90
91 CDEBUG(D_INFO, "%s: FLD cache - Size: %d, Threshold: %d\n",
92 cache->fci_name, cache_size, cache_threshold);
93
94 return cache;
95 }
96
97 /**
98 * destroy fld cache.
99 */
fld_cache_fini(struct fld_cache * cache)100 void fld_cache_fini(struct fld_cache *cache)
101 {
102 __u64 pct;
103
104 LASSERT(cache != NULL);
105 fld_cache_flush(cache);
106
107 if (cache->fci_stat.fst_count > 0) {
108 pct = cache->fci_stat.fst_cache * 100;
109 do_div(pct, cache->fci_stat.fst_count);
110 } else {
111 pct = 0;
112 }
113
114 CDEBUG(D_INFO, "FLD cache statistics (%s):\n", cache->fci_name);
115 CDEBUG(D_INFO, " Total reqs: %llu\n", cache->fci_stat.fst_count);
116 CDEBUG(D_INFO, " Cache reqs: %llu\n", cache->fci_stat.fst_cache);
117 CDEBUG(D_INFO, " Cache hits: %llu%%\n", pct);
118
119 OBD_FREE_PTR(cache);
120 }
121
122 /**
123 * delete given node from list.
124 */
fld_cache_entry_delete(struct fld_cache * cache,struct fld_cache_entry * node)125 void fld_cache_entry_delete(struct fld_cache *cache,
126 struct fld_cache_entry *node)
127 {
128 list_del(&node->fce_list);
129 list_del(&node->fce_lru);
130 cache->fci_cache_count--;
131 OBD_FREE_PTR(node);
132 }
133
134 /**
135 * fix list by checking new entry with NEXT entry in order.
136 */
fld_fix_new_list(struct fld_cache * cache)137 static void fld_fix_new_list(struct fld_cache *cache)
138 {
139 struct fld_cache_entry *f_curr;
140 struct fld_cache_entry *f_next;
141 struct lu_seq_range *c_range;
142 struct lu_seq_range *n_range;
143 struct list_head *head = &cache->fci_entries_head;
144
145 restart_fixup:
146
147 list_for_each_entry_safe(f_curr, f_next, head, fce_list) {
148 c_range = &f_curr->fce_range;
149 n_range = &f_next->fce_range;
150
151 LASSERT(range_is_sane(c_range));
152 if (&f_next->fce_list == head)
153 break;
154
155 if (c_range->lsr_flags != n_range->lsr_flags)
156 continue;
157
158 LASSERTF(c_range->lsr_start <= n_range->lsr_start,
159 "cur lsr_start "DRANGE" next lsr_start "DRANGE"\n",
160 PRANGE(c_range), PRANGE(n_range));
161
162 /* check merge possibility with next range */
163 if (c_range->lsr_end == n_range->lsr_start) {
164 if (c_range->lsr_index != n_range->lsr_index)
165 continue;
166 n_range->lsr_start = c_range->lsr_start;
167 fld_cache_entry_delete(cache, f_curr);
168 continue;
169 }
170
171 /* check if current range overlaps with next range. */
172 if (n_range->lsr_start < c_range->lsr_end) {
173 if (c_range->lsr_index == n_range->lsr_index) {
174 n_range->lsr_start = c_range->lsr_start;
175 n_range->lsr_end = max(c_range->lsr_end,
176 n_range->lsr_end);
177 fld_cache_entry_delete(cache, f_curr);
178 } else {
179 if (n_range->lsr_end <= c_range->lsr_end) {
180 *n_range = *c_range;
181 fld_cache_entry_delete(cache, f_curr);
182 } else
183 n_range->lsr_start = c_range->lsr_end;
184 }
185
186 /* we could have overlap over next
187 * range too. better restart. */
188 goto restart_fixup;
189 }
190
191 /* kill duplicates */
192 if (c_range->lsr_start == n_range->lsr_start &&
193 c_range->lsr_end == n_range->lsr_end)
194 fld_cache_entry_delete(cache, f_curr);
195 }
196 }
197
198 /**
199 * add node to fld cache
200 */
fld_cache_entry_add(struct fld_cache * cache,struct fld_cache_entry * f_new,struct list_head * pos)201 static inline void fld_cache_entry_add(struct fld_cache *cache,
202 struct fld_cache_entry *f_new,
203 struct list_head *pos)
204 {
205 list_add(&f_new->fce_list, pos);
206 list_add(&f_new->fce_lru, &cache->fci_lru);
207
208 cache->fci_cache_count++;
209 fld_fix_new_list(cache);
210 }
211
212 /**
213 * Check if cache needs to be shrunk. If so - do it.
214 * Remove one entry in list and so on until cache is shrunk enough.
215 */
fld_cache_shrink(struct fld_cache * cache)216 static int fld_cache_shrink(struct fld_cache *cache)
217 {
218 struct fld_cache_entry *flde;
219 struct list_head *curr;
220 int num = 0;
221
222 LASSERT(cache != NULL);
223
224 if (cache->fci_cache_count < cache->fci_cache_size)
225 return 0;
226
227 curr = cache->fci_lru.prev;
228
229 while (cache->fci_cache_count + cache->fci_threshold >
230 cache->fci_cache_size && curr != &cache->fci_lru) {
231
232 flde = list_entry(curr, struct fld_cache_entry, fce_lru);
233 curr = curr->prev;
234 fld_cache_entry_delete(cache, flde);
235 num++;
236 }
237
238 CDEBUG(D_INFO, "%s: FLD cache - Shrunk by %d entries\n",
239 cache->fci_name, num);
240
241 return 0;
242 }
243
244 /**
245 * kill all fld cache entries.
246 */
fld_cache_flush(struct fld_cache * cache)247 void fld_cache_flush(struct fld_cache *cache)
248 {
249 write_lock(&cache->fci_lock);
250 cache->fci_cache_size = 0;
251 fld_cache_shrink(cache);
252 write_unlock(&cache->fci_lock);
253 }
254
255 /**
256 * punch hole in existing range. divide this range and add new
257 * entry accordingly.
258 */
259
fld_cache_punch_hole(struct fld_cache * cache,struct fld_cache_entry * f_curr,struct fld_cache_entry * f_new)260 static void fld_cache_punch_hole(struct fld_cache *cache,
261 struct fld_cache_entry *f_curr,
262 struct fld_cache_entry *f_new)
263 {
264 const struct lu_seq_range *range = &f_new->fce_range;
265 const u64 new_start = range->lsr_start;
266 const u64 new_end = range->lsr_end;
267 struct fld_cache_entry *fldt;
268
269 OBD_ALLOC_GFP(fldt, sizeof(*fldt), GFP_ATOMIC);
270 if (!fldt) {
271 OBD_FREE_PTR(f_new);
272 /* overlap is not allowed, so dont mess up list. */
273 return;
274 }
275 /* break f_curr RANGE into three RANGES:
276 * f_curr, f_new , fldt
277 */
278
279 /* f_new = *range */
280
281 /* fldt */
282 fldt->fce_range.lsr_start = new_end;
283 fldt->fce_range.lsr_end = f_curr->fce_range.lsr_end;
284 fldt->fce_range.lsr_index = f_curr->fce_range.lsr_index;
285
286 /* f_curr */
287 f_curr->fce_range.lsr_end = new_start;
288
289 /* add these two entries to list */
290 fld_cache_entry_add(cache, f_new, &f_curr->fce_list);
291 fld_cache_entry_add(cache, fldt, &f_new->fce_list);
292
293 /* no need to fixup */
294 }
295
296 /**
297 * handle range overlap in fld cache.
298 */
fld_cache_overlap_handle(struct fld_cache * cache,struct fld_cache_entry * f_curr,struct fld_cache_entry * f_new)299 static void fld_cache_overlap_handle(struct fld_cache *cache,
300 struct fld_cache_entry *f_curr,
301 struct fld_cache_entry *f_new)
302 {
303 const struct lu_seq_range *range = &f_new->fce_range;
304 const u64 new_start = range->lsr_start;
305 const u64 new_end = range->lsr_end;
306 const u32 mdt = range->lsr_index;
307
308 /* this is overlap case, these case are checking overlapping with
309 * prev range only. fixup will handle overlapping with next range. */
310
311 if (f_curr->fce_range.lsr_index == mdt) {
312 f_curr->fce_range.lsr_start = min(f_curr->fce_range.lsr_start,
313 new_start);
314
315 f_curr->fce_range.lsr_end = max(f_curr->fce_range.lsr_end,
316 new_end);
317
318 OBD_FREE_PTR(f_new);
319 fld_fix_new_list(cache);
320
321 } else if (new_start <= f_curr->fce_range.lsr_start &&
322 f_curr->fce_range.lsr_end <= new_end) {
323 /* case 1: new range completely overshadowed existing range.
324 * e.g. whole range migrated. update fld cache entry */
325
326 f_curr->fce_range = *range;
327 OBD_FREE_PTR(f_new);
328 fld_fix_new_list(cache);
329
330 } else if (f_curr->fce_range.lsr_start < new_start &&
331 new_end < f_curr->fce_range.lsr_end) {
332 /* case 2: new range fit within existing range. */
333
334 fld_cache_punch_hole(cache, f_curr, f_new);
335
336 } else if (new_end <= f_curr->fce_range.lsr_end) {
337 /* case 3: overlap:
338 * [new_start [c_start new_end) c_end)
339 */
340
341 LASSERT(new_start <= f_curr->fce_range.lsr_start);
342
343 f_curr->fce_range.lsr_start = new_end;
344 fld_cache_entry_add(cache, f_new, f_curr->fce_list.prev);
345
346 } else if (f_curr->fce_range.lsr_start <= new_start) {
347 /* case 4: overlap:
348 * [c_start [new_start c_end) new_end)
349 */
350
351 LASSERT(f_curr->fce_range.lsr_end <= new_end);
352
353 f_curr->fce_range.lsr_end = new_start;
354 fld_cache_entry_add(cache, f_new, &f_curr->fce_list);
355 } else
356 CERROR("NEW range ="DRANGE" curr = "DRANGE"\n",
357 PRANGE(range), PRANGE(&f_curr->fce_range));
358 }
359
360 struct fld_cache_entry
fld_cache_entry_create(const struct lu_seq_range * range)361 *fld_cache_entry_create(const struct lu_seq_range *range)
362 {
363 struct fld_cache_entry *f_new;
364
365 LASSERT(range_is_sane(range));
366
367 OBD_ALLOC_PTR(f_new);
368 if (!f_new)
369 return ERR_PTR(-ENOMEM);
370
371 f_new->fce_range = *range;
372 return f_new;
373 }
374
375 /**
376 * Insert FLD entry in FLD cache.
377 *
378 * This function handles all cases of merging and breaking up of
379 * ranges.
380 */
fld_cache_insert_nolock(struct fld_cache * cache,struct fld_cache_entry * f_new)381 int fld_cache_insert_nolock(struct fld_cache *cache,
382 struct fld_cache_entry *f_new)
383 {
384 struct fld_cache_entry *f_curr;
385 struct fld_cache_entry *n;
386 struct list_head *head;
387 struct list_head *prev = NULL;
388 const u64 new_start = f_new->fce_range.lsr_start;
389 const u64 new_end = f_new->fce_range.lsr_end;
390 __u32 new_flags = f_new->fce_range.lsr_flags;
391
392 /*
393 * Duplicate entries are eliminated in insert op.
394 * So we don't need to search new entry before starting
395 * insertion loop.
396 */
397
398 if (!cache->fci_no_shrink)
399 fld_cache_shrink(cache);
400
401 head = &cache->fci_entries_head;
402
403 list_for_each_entry_safe(f_curr, n, head, fce_list) {
404 /* add list if next is end of list */
405 if (new_end < f_curr->fce_range.lsr_start ||
406 (new_end == f_curr->fce_range.lsr_start &&
407 new_flags != f_curr->fce_range.lsr_flags))
408 break;
409
410 prev = &f_curr->fce_list;
411 /* check if this range is to left of new range. */
412 if (new_start < f_curr->fce_range.lsr_end &&
413 new_flags == f_curr->fce_range.lsr_flags) {
414 fld_cache_overlap_handle(cache, f_curr, f_new);
415 goto out;
416 }
417 }
418
419 if (prev == NULL)
420 prev = head;
421
422 CDEBUG(D_INFO, "insert range "DRANGE"\n", PRANGE(&f_new->fce_range));
423 /* Add new entry to cache and lru list. */
424 fld_cache_entry_add(cache, f_new, prev);
425 out:
426 return 0;
427 }
428
fld_cache_insert(struct fld_cache * cache,const struct lu_seq_range * range)429 int fld_cache_insert(struct fld_cache *cache,
430 const struct lu_seq_range *range)
431 {
432 struct fld_cache_entry *flde;
433 int rc;
434
435 flde = fld_cache_entry_create(range);
436 if (IS_ERR(flde))
437 return PTR_ERR(flde);
438
439 write_lock(&cache->fci_lock);
440 rc = fld_cache_insert_nolock(cache, flde);
441 write_unlock(&cache->fci_lock);
442 if (rc)
443 OBD_FREE_PTR(flde);
444
445 return rc;
446 }
447
fld_cache_delete_nolock(struct fld_cache * cache,const struct lu_seq_range * range)448 void fld_cache_delete_nolock(struct fld_cache *cache,
449 const struct lu_seq_range *range)
450 {
451 struct fld_cache_entry *flde;
452 struct fld_cache_entry *tmp;
453 struct list_head *head;
454
455 head = &cache->fci_entries_head;
456 list_for_each_entry_safe(flde, tmp, head, fce_list) {
457 /* add list if next is end of list */
458 if (range->lsr_start == flde->fce_range.lsr_start ||
459 (range->lsr_end == flde->fce_range.lsr_end &&
460 range->lsr_flags == flde->fce_range.lsr_flags)) {
461 fld_cache_entry_delete(cache, flde);
462 break;
463 }
464 }
465 }
466
467 /**
468 * Delete FLD entry in FLD cache.
469 *
470 */
fld_cache_delete(struct fld_cache * cache,const struct lu_seq_range * range)471 void fld_cache_delete(struct fld_cache *cache,
472 const struct lu_seq_range *range)
473 {
474 write_lock(&cache->fci_lock);
475 fld_cache_delete_nolock(cache, range);
476 write_unlock(&cache->fci_lock);
477 }
478
479 struct fld_cache_entry
fld_cache_entry_lookup_nolock(struct fld_cache * cache,struct lu_seq_range * range)480 *fld_cache_entry_lookup_nolock(struct fld_cache *cache,
481 struct lu_seq_range *range)
482 {
483 struct fld_cache_entry *flde;
484 struct fld_cache_entry *got = NULL;
485 struct list_head *head;
486
487 head = &cache->fci_entries_head;
488 list_for_each_entry(flde, head, fce_list) {
489 if (range->lsr_start == flde->fce_range.lsr_start ||
490 (range->lsr_end == flde->fce_range.lsr_end &&
491 range->lsr_flags == flde->fce_range.lsr_flags)) {
492 got = flde;
493 break;
494 }
495 }
496
497 return got;
498 }
499
500 /**
501 * lookup \a seq sequence for range in fld cache.
502 */
503 struct fld_cache_entry
fld_cache_entry_lookup(struct fld_cache * cache,struct lu_seq_range * range)504 *fld_cache_entry_lookup(struct fld_cache *cache, struct lu_seq_range *range)
505 {
506 struct fld_cache_entry *got = NULL;
507
508 read_lock(&cache->fci_lock);
509 got = fld_cache_entry_lookup_nolock(cache, range);
510 read_unlock(&cache->fci_lock);
511 return got;
512 }
513
514 /**
515 * lookup \a seq sequence for range in fld cache.
516 */
fld_cache_lookup(struct fld_cache * cache,const u64 seq,struct lu_seq_range * range)517 int fld_cache_lookup(struct fld_cache *cache,
518 const u64 seq, struct lu_seq_range *range)
519 {
520 struct fld_cache_entry *flde;
521 struct fld_cache_entry *prev = NULL;
522 struct list_head *head;
523
524 read_lock(&cache->fci_lock);
525 head = &cache->fci_entries_head;
526
527 cache->fci_stat.fst_count++;
528 list_for_each_entry(flde, head, fce_list) {
529 if (flde->fce_range.lsr_start > seq) {
530 if (prev != NULL)
531 *range = prev->fce_range;
532 break;
533 }
534
535 prev = flde;
536 if (range_within(&flde->fce_range, seq)) {
537 *range = flde->fce_range;
538
539 cache->fci_stat.fst_cache++;
540 read_unlock(&cache->fci_lock);
541 return 0;
542 }
543 }
544 read_unlock(&cache->fci_lock);
545 return -ENOENT;
546 }
547