1 /* -*- mode: c; c-basic-offset: 8; indent-tabs-mode: nil; -*-
2 * vim:expandtab:shiftwidth=8:tabstop=8:
6 * DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License version 2 only,
10 * as published by the Free Software Foundation.
12 * This program is distributed in the hope that it will be useful, but
13 * WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * General Public License version 2 for more details (a copy is included
16 * in the LICENSE file that accompanied this code).
18 * You should have received a copy of the GNU General Public License
19 * version 2 along with this program; If not, see
20 * http://www.sun.com/software/products/lustre/docs/GPLv2.pdf
22 * Please contact Sun Microsystems, Inc., 4150 Network Circle, Santa Clara,
23 * CA 95054 USA or visit www.sun.com if you need additional information or
29 * Copyright 2008 Sun Microsystems, Inc. All rights reserved
30 * Use is subject to license terms.
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>
45 # define EXPORT_SYMTAB
47 #define DEBUG_SUBSYSTEM S_FLD
50 # include <libcfs/libcfs.h>
51 # include <linux/module.h>
52 # include <linux/jbd.h>
53 # include <asm/div64.h>
54 #else /* __KERNEL__ */
55 # include <liblustre.h>
56 # include <libcfs/list.h>
60 #include <obd_class.h>
61 #include <lustre_ver.h>
62 #include <obd_support.h>
63 #include <lprocfs_status.h>
65 #include <dt_object.h>
66 #include <md_object.h>
67 #include <lustre_req_layout.h>
68 #include <lustre_fld.h>
69 #include "fld_internal.h"
74 struct fld_cache *fld_cache_init(const char *name,
75 int cache_size, int cache_threshold)
77 struct fld_cache *cache;
80 LASSERT(name != NULL);
81 LASSERT(cache_threshold < cache_size);
85 RETURN(ERR_PTR(-ENOMEM));
87 CFS_INIT_LIST_HEAD(&cache->fci_entries_head);
88 CFS_INIT_LIST_HEAD(&cache->fci_lru);
90 cache->fci_cache_count = 0;
91 spin_lock_init(&cache->fci_lock);
93 strncpy(cache->fci_name, name,
94 sizeof(cache->fci_name));
96 cache->fci_cache_size = cache_size;
97 cache->fci_threshold = cache_threshold;
99 /* Init fld cache info. */
100 memset(&cache->fci_stat, 0, sizeof(cache->fci_stat));
102 CDEBUG(D_INFO, "%s: FLD cache - Size: %d, Threshold: %d\n",
103 cache->fci_name, cache_size, cache_threshold);
111 void fld_cache_fini(struct fld_cache *cache)
116 LASSERT(cache != NULL);
117 fld_cache_flush(cache);
119 if (cache->fci_stat.fst_count > 0) {
120 pct = cache->fci_stat.fst_cache * 100;
121 do_div(pct, cache->fci_stat.fst_count);
126 CDEBUG(D_INFO, "FLD cache statistics (%s):\n", cache->fci_name);
127 CDEBUG(D_INFO, " Total reqs: "LPU64"\n", cache->fci_stat.fst_count);
128 CDEBUG(D_INFO, " Cache reqs: "LPU64"\n", cache->fci_stat.fst_cache);
129 CDEBUG(D_INFO, " Cache hits: "LPU64"%%\n", pct);
136 static inline void fld_cache_entry_delete(struct fld_cache *cache,
137 struct fld_cache_entry *node);
140 * fix list by checking new entry with NEXT entry in order.
142 static void fld_fix_new_list(struct fld_cache *cache)
144 struct fld_cache_entry *f_curr;
145 struct fld_cache_entry *f_next;
146 struct lu_seq_range *c_range;
147 struct lu_seq_range *n_range;
148 struct list_head *head = &cache->fci_entries_head;
153 list_for_each_entry_safe(f_curr, f_next, head, fce_list) {
154 c_range = &f_curr->fce_range;
155 n_range = &f_next->fce_range;
157 LASSERT(range_is_sane(c_range));
158 if (&f_next->fce_list == head)
161 LASSERT(c_range->lsr_start <= n_range->lsr_start);
163 /* check merge possibility with next range */
164 if (c_range->lsr_end == n_range->lsr_start) {
165 if (c_range->lsr_mdt != n_range->lsr_mdt)
167 n_range->lsr_start = c_range->lsr_start;
168 fld_cache_entry_delete(cache, f_curr);
172 /* check if current range overlaps with next range. */
173 if (n_range->lsr_start < c_range->lsr_end) {
175 if (c_range->lsr_mdt == n_range->lsr_mdt) {
176 n_range->lsr_start = c_range->lsr_start;
177 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 * delete given node from list.
220 static inline void fld_cache_entry_delete(struct fld_cache *cache,
221 struct fld_cache_entry *node)
223 list_del(&node->fce_list);
224 list_del(&node->fce_lru);
225 cache->fci_cache_count--;
230 * Check if cache needs to be shrunk. If so - do it.
231 * Remove one entry in list and so on until cache is shrunk enough.
233 static int fld_cache_shrink(struct fld_cache *cache)
235 struct fld_cache_entry *flde;
236 struct list_head *curr;
240 LASSERT(cache != NULL);
242 if (cache->fci_cache_count < cache->fci_cache_size)
245 curr = cache->fci_lru.prev;
247 while (cache->fci_cache_count + cache->fci_threshold >
248 cache->fci_cache_size && curr != &cache->fci_lru) {
250 flde = list_entry(curr, struct fld_cache_entry, fce_lru);
252 fld_cache_entry_delete(cache, flde);
256 CDEBUG(D_INFO, "%s: FLD cache - Shrunk by "
257 "%d entries\n", cache->fci_name, num);
263 * kill all fld cache entries.
265 void fld_cache_flush(struct fld_cache *cache)
269 spin_lock(&cache->fci_lock);
270 cache->fci_cache_size = 0;
271 fld_cache_shrink(cache);
272 spin_unlock(&cache->fci_lock);
278 * punch hole in existing range. divide this range and add new
282 void fld_cache_punch_hole(struct fld_cache *cache,
283 struct fld_cache_entry *f_curr,
284 struct fld_cache_entry *f_new)
286 const struct lu_seq_range *range = &f_new->fce_range;
287 const seqno_t new_start = range->lsr_start;
288 const seqno_t new_end = range->lsr_end;
289 struct fld_cache_entry *fldt;
292 OBD_ALLOC_GFP(fldt, sizeof *fldt, CFS_ALLOC_ATOMIC);
296 /* overlap is not allowed, so dont mess up list. */
299 /* break f_curr RANGE into three RANGES:
300 * f_curr, f_new , fldt
306 fldt->fce_range.lsr_start = new_end;
307 fldt->fce_range.lsr_end = f_curr->fce_range.lsr_end;
308 fldt->fce_range.lsr_mdt = f_curr->fce_range.lsr_mdt;
311 f_curr->fce_range.lsr_end = new_start;
313 /* add these two entries to list */
314 fld_cache_entry_add(cache, f_new, &f_curr->fce_list);
315 fld_cache_entry_add(cache, fldt, &f_new->fce_list);
317 /* no need to fixup */
322 * handle range overlap in fld cache.
324 void fld_cache_overlap_handle(struct fld_cache *cache,
325 struct fld_cache_entry *f_curr,
326 struct fld_cache_entry *f_new)
328 const struct lu_seq_range *range = &f_new->fce_range;
329 const seqno_t new_start = range->lsr_start;
330 const seqno_t new_end = range->lsr_end;
331 const mdsno_t mdt = range->lsr_mdt;
333 /* this is overlap case, these case are checking overlapping with
334 * prev range only. fixup will handle overlaping with next range. */
336 if (f_curr->fce_range.lsr_mdt == mdt) {
337 f_curr->fce_range.lsr_start = min(f_curr->fce_range.lsr_start,
340 f_curr->fce_range.lsr_end = max(f_curr->fce_range.lsr_end,
344 fld_fix_new_list(cache);
346 } else if (new_start <= f_curr->fce_range.lsr_start &&
347 f_curr->fce_range.lsr_end <= new_end) {
348 /* case 1: new range completely overshadowed existing range.
349 * e.g. whole range migrated. update fld cache entry */
351 f_curr->fce_range = *range;
353 fld_fix_new_list(cache);
355 } else if (f_curr->fce_range.lsr_start < new_start &&
356 new_end < f_curr->fce_range.lsr_end) {
357 /* case 2: new range fit within existing range. */
359 fld_cache_punch_hole(cache, f_curr, f_new);
361 } else if (new_end <= f_curr->fce_range.lsr_end) {
363 * [new_start [c_start new_end) c_end)
366 LASSERT(new_start <= f_curr->fce_range.lsr_start);
368 f_curr->fce_range.lsr_start = new_end;
369 fld_cache_entry_add(cache, f_new, f_curr->fce_list.prev);
371 } else if (f_curr->fce_range.lsr_start <= new_start) {
373 * [c_start [new_start c_end) new_end)
376 LASSERT(f_curr->fce_range.lsr_end <= new_end);
378 f_curr->fce_range.lsr_end = new_start;
379 fld_cache_entry_add(cache, f_new, &f_curr->fce_list);
381 CERROR("NEW range ="DRANGE" curr = "DRANGE"\n",
382 PRANGE(range),PRANGE(&f_curr->fce_range));
386 * Insert FLD entry in FLD cache.
388 * This function handles all cases of merging and breaking up of
391 void fld_cache_insert(struct fld_cache *cache,
392 const struct lu_seq_range *range)
394 struct fld_cache_entry *f_new;
395 struct fld_cache_entry *f_curr;
396 struct fld_cache_entry *n;
397 struct list_head *head;
398 struct list_head *prev = NULL;
399 const seqno_t new_start = range->lsr_start;
400 const seqno_t new_end = range->lsr_end;
403 LASSERT(range_is_sane(range));
405 /* Allocate new entry. */
406 OBD_ALLOC_PTR(f_new);
412 f_new->fce_range = *range;
415 * Duplicate entries are eliminated in inset op.
416 * So we don't need to search new entry before starting insertion loop.
419 spin_lock(&cache->fci_lock);
420 fld_cache_shrink(cache);
422 head = &cache->fci_entries_head;
424 list_for_each_entry_safe(f_curr, n, head, fce_list) {
425 /* add list if next is end of list */
426 if (new_end < f_curr->fce_range.lsr_start)
429 prev = &f_curr->fce_list;
430 /* check if this range is to left of new range. */
431 if (new_start < f_curr->fce_range.lsr_end) {
432 fld_cache_overlap_handle(cache, f_curr, f_new);
440 /* Add new entry to cache and lru list. */
441 fld_cache_entry_add(cache, f_new, prev);
443 spin_unlock(&cache->fci_lock);
448 * lookup \a seq sequence for range in fld cache.
450 int fld_cache_lookup(struct fld_cache *cache,
451 const seqno_t seq, struct lu_seq_range *range)
453 struct fld_cache_entry *flde;
454 struct list_head *head;
458 spin_lock(&cache->fci_lock);
459 head = &cache->fci_entries_head;
461 cache->fci_stat.fst_count++;
462 list_for_each_entry(flde, head, fce_list) {
463 if (flde->fce_range.lsr_start > seq)
466 if (range_within(&flde->fce_range, seq)) {
467 *range = flde->fce_range;
469 /* update position of this entry in lru list. */
470 list_move(&flde->fce_lru, &cache->fci_lru);
471 cache->fci_stat.fst_cache++;
472 spin_unlock(&cache->fci_lock);
476 spin_unlock(&cache->fci_lock);