4 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
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.
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).
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
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
27 * Copyright (c) 2007, 2010, Oracle and/or its affiliates. All rights reserved.
28 * Use is subject to license terms.
30 * Copyright (c) 2012, 2013, Intel Corporation.
33 * This file is part of Lustre, http://www.lustre.org/
34 * Lustre is a trademark of Sun Microsystems, Inc.
36 * lustre/fld/fld_cache.c
38 * FLD (Fids Location Database)
40 * Author: Pravin Shelar <pravin.shelar@sun.com>
41 * Author: Yury Umanets <umka@clusterfs.com>
44 #define DEBUG_SUBSYSTEM S_FLD
47 # include <libcfs/libcfs.h>
48 # include <linux/module.h>
49 # include <linux/math64.h>
50 #else /* __KERNEL__ */
51 # include <liblustre.h>
52 # include <libcfs/list.h>
55 #include <obd_support.h>
56 #include <lustre_fld.h>
57 #include "fld_internal.h"
62 struct fld_cache *fld_cache_init(const char *name,
63 int cache_size, int cache_threshold)
65 struct fld_cache *cache;
68 LASSERT(name != NULL);
69 LASSERT(cache_threshold < cache_size);
73 RETURN(ERR_PTR(-ENOMEM));
75 INIT_LIST_HEAD(&cache->fci_entries_head);
76 INIT_LIST_HEAD(&cache->fci_lru);
78 cache->fci_cache_count = 0;
79 rwlock_init(&cache->fci_lock);
81 strlcpy(cache->fci_name, name,
82 sizeof(cache->fci_name));
84 cache->fci_cache_size = cache_size;
85 cache->fci_threshold = cache_threshold;
87 /* Init fld cache info. */
88 memset(&cache->fci_stat, 0, sizeof(cache->fci_stat));
90 CDEBUG(D_INFO, "%s: FLD cache - Size: %d, Threshold: %d\n",
91 cache->fci_name, cache_size, cache_threshold);
99 void fld_cache_fini(struct fld_cache *cache)
104 LASSERT(cache != NULL);
105 fld_cache_flush(cache);
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);
114 CDEBUG(D_INFO, "FLD cache statistics (%s):\n", cache->fci_name);
115 CDEBUG(D_INFO, " Total reqs: "LPU64"\n", cache->fci_stat.fst_count);
116 CDEBUG(D_INFO, " Cache reqs: "LPU64"\n", cache->fci_stat.fst_cache);
117 CDEBUG(D_INFO, " Cache hits: "LPU64"%%\n", pct);
125 * delete given node from list.
127 void fld_cache_entry_delete(struct fld_cache *cache,
128 struct fld_cache_entry *node)
130 list_del(&node->fce_list);
131 list_del(&node->fce_lru);
132 cache->fci_cache_count--;
137 * fix list by checking new entry with NEXT entry in order.
139 static void fld_fix_new_list(struct fld_cache *cache)
141 struct fld_cache_entry *f_curr;
142 struct fld_cache_entry *f_next;
143 struct lu_seq_range *c_range;
144 struct lu_seq_range *n_range;
145 struct list_head *head = &cache->fci_entries_head;
150 list_for_each_entry_safe(f_curr, f_next, head, fce_list) {
151 c_range = &f_curr->fce_range;
152 n_range = &f_next->fce_range;
154 LASSERT(range_is_sane(c_range));
155 if (&f_next->fce_list == head)
158 if (c_range->lsr_flags != n_range->lsr_flags)
161 LASSERTF(c_range->lsr_start <= n_range->lsr_start,
162 "cur lsr_start "DRANGE" next lsr_start "DRANGE"\n",
163 PRANGE(c_range), PRANGE(n_range));
165 /* check merge possibility with next range */
166 if (c_range->lsr_end == n_range->lsr_start) {
167 if (c_range->lsr_index != n_range->lsr_index)
169 n_range->lsr_start = c_range->lsr_start;
170 fld_cache_entry_delete(cache, f_curr);
174 /* check if current range overlaps with next range. */
175 if (n_range->lsr_start < c_range->lsr_end) {
176 if (c_range->lsr_index == n_range->lsr_index) {
177 n_range->lsr_start = c_range->lsr_start;
178 n_range->lsr_end = max(c_range->lsr_end,
180 fld_cache_entry_delete(cache, f_curr);
182 if (n_range->lsr_end <= c_range->lsr_end) {
184 fld_cache_entry_delete(cache, f_curr);
186 n_range->lsr_start = c_range->lsr_end;
189 /* we could have overlap over next
190 * range too. better restart. */
194 /* kill duplicates */
195 if (c_range->lsr_start == n_range->lsr_start &&
196 c_range->lsr_end == n_range->lsr_end)
197 fld_cache_entry_delete(cache, f_curr);
204 * add node to fld cache
206 static inline void fld_cache_entry_add(struct fld_cache *cache,
207 struct fld_cache_entry *f_new,
208 struct list_head *pos)
210 list_add(&f_new->fce_list, pos);
211 list_add(&f_new->fce_lru, &cache->fci_lru);
213 cache->fci_cache_count++;
214 fld_fix_new_list(cache);
218 * Check if cache needs to be shrunk. If so - do it.
219 * Remove one entry in list and so on until cache is shrunk enough.
221 static int fld_cache_shrink(struct fld_cache *cache)
223 struct fld_cache_entry *flde;
224 struct list_head *curr;
228 LASSERT(cache != NULL);
230 if (cache->fci_cache_count < cache->fci_cache_size)
233 curr = cache->fci_lru.prev;
235 while (cache->fci_cache_count + cache->fci_threshold >
236 cache->fci_cache_size && curr != &cache->fci_lru) {
238 flde = list_entry(curr, struct fld_cache_entry, fce_lru);
240 fld_cache_entry_delete(cache, flde);
244 CDEBUG(D_INFO, "%s: FLD cache - Shrunk by "
245 "%d entries\n", cache->fci_name, num);
251 * kill all fld cache entries.
253 void fld_cache_flush(struct fld_cache *cache)
257 write_lock(&cache->fci_lock);
258 cache->fci_cache_size = 0;
259 fld_cache_shrink(cache);
260 write_unlock(&cache->fci_lock);
266 * punch hole in existing range. divide this range and add new
270 static void fld_cache_punch_hole(struct fld_cache *cache,
271 struct fld_cache_entry *f_curr,
272 struct fld_cache_entry *f_new)
274 const struct lu_seq_range *range = &f_new->fce_range;
275 const seqno_t new_start = range->lsr_start;
276 const seqno_t new_end = range->lsr_end;
277 struct fld_cache_entry *fldt;
280 OBD_ALLOC_GFP(fldt, sizeof *fldt, GFP_ATOMIC);
284 /* overlap is not allowed, so dont mess up list. */
287 /* break f_curr RANGE into three RANGES:
288 * f_curr, f_new , fldt
294 fldt->fce_range.lsr_start = new_end;
295 fldt->fce_range.lsr_end = f_curr->fce_range.lsr_end;
296 fldt->fce_range.lsr_index = f_curr->fce_range.lsr_index;
299 f_curr->fce_range.lsr_end = new_start;
301 /* add these two entries to list */
302 fld_cache_entry_add(cache, f_new, &f_curr->fce_list);
303 fld_cache_entry_add(cache, fldt, &f_new->fce_list);
305 /* no need to fixup */
310 * handle range overlap in fld cache.
312 static void fld_cache_overlap_handle(struct fld_cache *cache,
313 struct fld_cache_entry *f_curr,
314 struct fld_cache_entry *f_new)
316 const struct lu_seq_range *range = &f_new->fce_range;
317 const seqno_t new_start = range->lsr_start;
318 const seqno_t new_end = range->lsr_end;
319 const mdsno_t mdt = range->lsr_index;
321 /* this is overlap case, these case are checking overlapping with
322 * prev range only. fixup will handle overlaping with next range. */
324 if (f_curr->fce_range.lsr_index == mdt) {
325 f_curr->fce_range.lsr_start = min(f_curr->fce_range.lsr_start,
328 f_curr->fce_range.lsr_end = max(f_curr->fce_range.lsr_end,
332 fld_fix_new_list(cache);
334 } else if (new_start <= f_curr->fce_range.lsr_start &&
335 f_curr->fce_range.lsr_end <= new_end) {
336 /* case 1: new range completely overshadowed existing range.
337 * e.g. whole range migrated. update fld cache entry */
339 f_curr->fce_range = *range;
341 fld_fix_new_list(cache);
343 } else if (f_curr->fce_range.lsr_start < new_start &&
344 new_end < f_curr->fce_range.lsr_end) {
345 /* case 2: new range fit within existing range. */
347 fld_cache_punch_hole(cache, f_curr, f_new);
349 } else if (new_end <= f_curr->fce_range.lsr_end) {
351 * [new_start [c_start new_end) c_end)
354 LASSERT(new_start <= f_curr->fce_range.lsr_start);
356 f_curr->fce_range.lsr_start = new_end;
357 fld_cache_entry_add(cache, f_new, f_curr->fce_list.prev);
359 } else if (f_curr->fce_range.lsr_start <= new_start) {
361 * [c_start [new_start c_end) new_end)
364 LASSERT(f_curr->fce_range.lsr_end <= new_end);
366 f_curr->fce_range.lsr_end = new_start;
367 fld_cache_entry_add(cache, f_new, &f_curr->fce_list);
369 CERROR("NEW range ="DRANGE" curr = "DRANGE"\n",
370 PRANGE(range),PRANGE(&f_curr->fce_range));
373 struct fld_cache_entry
374 *fld_cache_entry_create(const struct lu_seq_range *range)
376 struct fld_cache_entry *f_new;
378 LASSERT(range_is_sane(range));
380 OBD_ALLOC_PTR(f_new);
382 RETURN(ERR_PTR(-ENOMEM));
384 f_new->fce_range = *range;
389 * Insert FLD entry in FLD cache.
391 * This function handles all cases of merging and breaking up of
394 int fld_cache_insert_nolock(struct fld_cache *cache,
395 struct fld_cache_entry *f_new)
397 struct fld_cache_entry *f_curr;
398 struct fld_cache_entry *n;
399 struct list_head *head;
400 struct list_head *prev = NULL;
401 const seqno_t new_start = f_new->fce_range.lsr_start;
402 const seqno_t new_end = f_new->fce_range.lsr_end;
403 __u32 new_flags = f_new->fce_range.lsr_flags;
407 * Duplicate entries are eliminated in insert op.
408 * So we don't need to search new entry before starting
412 if (!cache->fci_no_shrink)
413 fld_cache_shrink(cache);
415 head = &cache->fci_entries_head;
417 list_for_each_entry_safe(f_curr, n, head, fce_list) {
418 /* add list if next is end of list */
419 if (new_end < f_curr->fce_range.lsr_start ||
420 (new_end == f_curr->fce_range.lsr_start &&
421 new_flags != f_curr->fce_range.lsr_flags))
424 prev = &f_curr->fce_list;
425 /* check if this range is to left of new range. */
426 if (new_start < f_curr->fce_range.lsr_end &&
427 new_flags == f_curr->fce_range.lsr_flags) {
428 fld_cache_overlap_handle(cache, f_curr, f_new);
436 CDEBUG(D_INFO, "insert range "DRANGE"\n", PRANGE(&f_new->fce_range));
437 /* Add new entry to cache and lru list. */
438 fld_cache_entry_add(cache, f_new, prev);
443 int fld_cache_insert(struct fld_cache *cache,
444 const struct lu_seq_range *range)
446 struct fld_cache_entry *flde;
449 flde = fld_cache_entry_create(range);
451 RETURN(PTR_ERR(flde));
453 write_lock(&cache->fci_lock);
454 rc = fld_cache_insert_nolock(cache, flde);
455 write_unlock(&cache->fci_lock);
462 void fld_cache_delete_nolock(struct fld_cache *cache,
463 const struct lu_seq_range *range)
465 struct fld_cache_entry *flde;
466 struct fld_cache_entry *tmp;
467 struct list_head *head;
469 head = &cache->fci_entries_head;
470 list_for_each_entry_safe(flde, tmp, head, fce_list) {
471 /* add list if next is end of list */
472 if (range->lsr_start == flde->fce_range.lsr_start ||
473 (range->lsr_end == flde->fce_range.lsr_end &&
474 range->lsr_flags == flde->fce_range.lsr_flags)) {
475 fld_cache_entry_delete(cache, flde);
482 * Delete FLD entry in FLD cache.
485 void fld_cache_delete(struct fld_cache *cache,
486 const struct lu_seq_range *range)
488 write_lock(&cache->fci_lock);
489 fld_cache_delete_nolock(cache, range);
490 write_unlock(&cache->fci_lock);
493 struct fld_cache_entry *
494 fld_cache_entry_lookup_nolock(struct fld_cache *cache,
495 const struct lu_seq_range *range)
497 struct fld_cache_entry *flde;
498 struct fld_cache_entry *got = NULL;
499 struct list_head *head;
501 head = &cache->fci_entries_head;
502 list_for_each_entry(flde, head, fce_list) {
503 if (range->lsr_start == flde->fce_range.lsr_start ||
504 (range->lsr_end == flde->fce_range.lsr_end &&
505 range->lsr_flags == flde->fce_range.lsr_flags)) {
515 * lookup \a seq sequence for range in fld cache.
517 struct fld_cache_entry *
518 fld_cache_entry_lookup(struct fld_cache *cache,
519 const struct lu_seq_range *range)
521 struct fld_cache_entry *got = NULL;
524 read_lock(&cache->fci_lock);
525 got = fld_cache_entry_lookup_nolock(cache, range);
526 read_unlock(&cache->fci_lock);
532 * lookup \a seq sequence for range in fld cache.
534 int fld_cache_lookup(struct fld_cache *cache,
535 const seqno_t seq, struct lu_seq_range *range)
537 struct fld_cache_entry *flde;
538 struct fld_cache_entry *prev = NULL;
539 struct list_head *head;
542 read_lock(&cache->fci_lock);
543 head = &cache->fci_entries_head;
545 cache->fci_stat.fst_count++;
546 list_for_each_entry(flde, head, fce_list) {
547 if (flde->fce_range.lsr_start > seq) {
549 *range = prev->fce_range;
554 if (range_within(&flde->fce_range, seq)) {
555 *range = flde->fce_range;
557 cache->fci_stat.fst_cache++;
558 read_unlock(&cache->fci_lock);
562 read_unlock(&cache->fci_lock);