1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
|
/* -*- mode: c; c-basic-offset: 8; -*-
* vim: noexpandtab sw=8 ts=8 sts=0:
*
* uptodate.c
*
* Tracking the up-to-date-ness of a local buffer_head with respect to
* the cluster.
*
* Copyright (C) 2002, 2004, 2005 Oracle. All rights reserved.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public
* License as published by the Free Software Foundation; either
* version 2 of the License, or (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
* General Public License for more details.
*
* You should have received a copy of the GNU General Public
* License along with this program; if not, write to the
* Free Software Foundation, Inc., 59 Temple Place - Suite 330,
* Boston, MA 021110-1307, USA.
*
* Standard buffer head caching flags (uptodate, etc) are insufficient
* in a clustered environment - a buffer may be marked up to date on
* our local node but could have been modified by another cluster
* member. As a result an additional (and performant) caching scheme
* is required. A further requirement is that we consume as little
* memory as possible - we never pin buffer_head structures in order
* to cache them.
*
* We track the existence of up to date buffers on the inodes which
* are associated with them. Because we don't want to pin
* buffer_heads, this is only a (strong) hint and several other checks
* are made in the I/O path to ensure that we don't use a stale or
* invalid buffer without going to disk:
* - buffer_jbd is used liberally - if a bh is in the journal on
* this node then it *must* be up to date.
* - the standard buffer_uptodate() macro is used to detect buffers
* which may be invalid (even if we have an up to date tracking
* item for them)
*
* For a full understanding of how this code works together, one
* should read the callers in dlmglue.c, the I/O functions in
* buffer_head_io.c and ocfs2_journal_access in journal.c
*/
#include <linux/fs.h>
#include <linux/types.h>
#include <linux/slab.h>
#include <linux/highmem.h>
#include <linux/buffer_head.h>
#include <linux/rbtree.h>
#ifndef CONFIG_OCFS2_COMPAT_JBD
# include <linux/jbd2.h>
#else
# include <linux/jbd.h>
#endif
#define MLOG_MASK_PREFIX ML_UPTODATE
#include <cluster/masklog.h>
#include "ocfs2.h"
#include "inode.h"
#include "uptodate.h"
struct ocfs2_meta_cache_item {
struct rb_node c_node;
sector_t c_block;
};
static struct kmem_cache *ocfs2_uptodate_cachep = NULL;
void ocfs2_metadata_cache_init(struct inode *inode)
{
struct ocfs2_inode_info *oi = OCFS2_I(inode);
struct ocfs2_caching_info *ci = &oi->ip_metadata_cache;
oi->ip_flags |= OCFS2_INODE_CACHE_INLINE;
ci->ci_num_cached = 0;
}
/* No lock taken here as 'root' is not expected to be visible to other
* processes. */
static unsigned int ocfs2_purge_copied_metadata_tree(struct rb_root *root)
{
unsigned int purged = 0;
struct rb_node *node;
struct ocfs2_meta_cache_item *item;
while ((node = rb_last(root)) != NULL) {
item = rb_entry(node, struct ocfs2_meta_cache_item, c_node);
mlog(0, "Purge item %llu\n",
(unsigned long long) item->c_block);
rb_erase(&item->c_node, root);
kmem_cache_free(ocfs2_uptodate_cachep, item);
purged++;
}
return purged;
}
/* Called from locking and called from ocfs2_clear_inode. Dump the
* cache for a given inode.
*
* This function is a few more lines longer than necessary due to some
* accounting done here, but I think it's worth tracking down those
* bugs sooner -- Mark */
void ocfs2_metadata_cache_purge(struct inode *inode)
{
struct ocfs2_inode_info *oi = OCFS2_I(inode);
unsigned int tree, to_purge, purged;
struct ocfs2_caching_info *ci = &oi->ip_metadata_cache;
struct rb_root root = RB_ROOT;
spin_lock(&oi->ip_lock);
tree = !(oi->ip_flags & OCFS2_INODE_CACHE_INLINE);
to_purge = ci->ci_num_cached;
mlog(0, "Purge %u %s items from Inode %llu\n", to_purge,
tree ? "array" : "tree", (unsigned long long)oi->ip_blkno);
/* If we're a tree, save off the root so that we can safely
* initialize the cache. We do the work to free tree members
* without the spinlock. */
if (tree)
root = ci->ci_cache.ci_tree;
ocfs2_metadata_cache_init(inode);
spin_unlock(&oi->ip_lock);
purged = ocfs2_purge_copied_metadata_tree(&root);
/* If possible, track the number wiped so that we can more
* easily detect counting errors. Unfortunately, this is only
* meaningful for trees. */
if (tree && purged != to_purge)
mlog(ML_ERROR, "Inode %llu, count = %u, purged = %u\n",
(unsigned long long)oi->ip_blkno, to_purge, purged);
}
/* Returns the index in the cache array, -1 if not found.
* Requires ip_lock. */
static int ocfs2_search_cache_array(struct ocfs2_caching_info *ci,
sector_t item)
{
int i;
for (i = 0; i < ci->ci_num_cached; i++) {
if (item == ci->ci_cache.ci_array[i])
return i;
}
return -1;
}
/* Returns the cache item if found, otherwise NULL.
* Requires ip_lock. */
static struct ocfs2_meta_cache_item *
ocfs2_search_cache_tree(struct ocfs2_caching_info *ci,
sector_t block)
{
struct rb_node * n = ci->ci_cache.ci_tree.rb_node;
struct ocfs2_meta_cache_item *item = NULL;
while (n) {
item = rb_entry(n, struct ocfs2_meta_cache_item, c_node);
if (block < item->c_block)
n = n->rb_left;
else if (block > item->c_block)
n = n->rb_right;
else
return item;
}
return NULL;
}
static int ocfs2_buffer_cached(struct ocfs2_inode_info *oi,
struct buffer_head *bh)
{
int index = -1;
struct ocfs2_meta_cache_item *item = NULL;
spin_lock(&oi->ip_lock);
mlog(0, "Inode %llu, query block %llu (inline = %u)\n",
(unsigned long long)oi->ip_blkno,
(unsigned long long) bh->b_blocknr,
!!(oi->ip_flags & OCFS2_INODE_CACHE_INLINE));
if (oi->ip_flags & OCFS2_INODE_CACHE_INLINE)
index = ocfs2_search_cache_array(&oi->ip_metadata_cache,
bh->b_blocknr);
else
item = ocfs2_search_cache_tree(&oi->ip_metadata_cache,
bh->b_blocknr);
spin_unlock(&oi->ip_lock);
mlog(0, "index = %d, item = %p\n", index, item);
return (index != -1) || (item != NULL);
}
/* Warning: even if it returns true, this does *not* guarantee that
* the block is stored in our inode metadata cache.
*
* This can be called under lock_buffer()
*/
int ocfs2_buffer_uptodate(struct inode *inode,
struct buffer_head *bh)
{
/* Doesn't matter if the bh is in our cache or not -- if it's
* not marked uptodate then we know it can't have correct
* data. */
if (!buffer_uptodate(bh))
return 0;
/* OCFS2 does not allow multiple nodes to be changing the same
* block at the same time. */
if (buffer_jbd(bh))
return 1;
/* Ok, locally the buffer is marked as up to date, now search
* our cache to see if we can trust that. */
return ocfs2_buffer_cached(OCFS2_I(inode), bh);
}
/*
* Determine whether a buffer is currently out on a read-ahead request.
* ip_io_sem should be held to serialize submitters with the logic here.
*/
int ocfs2_buffer_read_ahead(struct inode *inode,
struct buffer_head *bh)
{
return buffer_locked(bh) && ocfs2_buffer_cached(OCFS2_I(inode), bh);
}
/* Requires ip_lock */
static void ocfs2_append_cache_array(struct ocfs2_caching_info *ci,
sector_t block)
{
BUG_ON(ci->ci_num_cached >= OCFS2_INODE_MAX_CACHE_ARRAY);
mlog(0, "block %llu takes position %u\n", (unsigned long long) block,
ci->ci_num_cached);
ci->ci_cache.ci_array[ci->ci_num_cached] = block;
ci->ci_num_cached++;
}
/* By now the caller should have checked that the item does *not*
* exist in the tree.
* Requires ip_lock. */
static void __ocfs2_insert_cache_tree(struct ocfs2_caching_info *ci,
struct ocfs2_meta_cache_item *new)
{
sector_t block = new->c_block;
struct rb_node *parent = NULL;
struct rb_node **p = &ci->ci_cache.ci_tree.rb_node;
struct ocfs2_meta_cache_item *tmp;
mlog(
|