file.c 52.4 KB
Newer Older
Linus Torvalds's avatar
Linus Torvalds committed
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
/*
 * Copyright 2000 by Hans Reiser, licensing governed by reiserfs/README
 */

#include <linux/time.h>
#include <linux/reiserfs_fs.h>
#include <linux/reiserfs_acl.h>
#include <linux/reiserfs_xattr.h>
#include <linux/smp_lock.h>
#include <asm/uaccess.h>
#include <linux/pagemap.h>
#include <linux/swap.h>
#include <linux/writeback.h>
#include <linux/blkdev.h>
#include <linux/buffer_head.h>
#include <linux/quotaops.h>

/*
** We pack the tails of files on file close, not at the time they are written.
** This implies an unnecessary copy of the tail and an unnecessary indirect item
** insertion/balancing, for files that are written in one write.
** It avoids unnecessary tail packings (balances) for files that are written in
** multiple writes and are small enough to have tails.
** 
** file_release is called by the VFS layer when the file is closed.  If
** this is the last open file descriptor, and the file
** small enough to have a tail, and the tail is currently in an
** unformatted node, the tail is converted back into a direct item.
** 
** We use reiserfs_truncate_file to pack the tail, since it already has
** all the conditions coded.  
*/
33
static int reiserfs_file_release(struct inode *inode, struct file *filp)
Linus Torvalds's avatar
Linus Torvalds committed
34
35
{

36
37
38
	struct reiserfs_transaction_handle th;
	int err;
	int jbegin_failure = 0;
Linus Torvalds's avatar
Linus Torvalds committed
39

40
	BUG_ON(!S_ISREG(inode->i_mode));
Linus Torvalds's avatar
Linus Torvalds committed
41

42
43
44
45
46
47
48
	/* fast out for when nothing needs to be done */
	if ((atomic_read(&inode->i_count) > 1 ||
	     !(REISERFS_I(inode)->i_flags & i_pack_on_close_mask) ||
	     !tail_has_to_be_packed(inode)) &&
	    REISERFS_I(inode)->i_prealloc_count <= 0) {
		return 0;
	}
Linus Torvalds's avatar
Linus Torvalds committed
49

50
	mutex_lock(&inode->i_mutex);
51
	reiserfs_write_lock(inode->i_sb);
52
53
54
55
56
57
	/* freeing preallocation only involves relogging blocks that
	 * are already in the current transaction.  preallocation gets
	 * freed at the end of each transaction, so it is impossible for
	 * us to log any additional blocks (including quota blocks)
	 */
	err = journal_begin(&th, inode->i_sb, 1);
Linus Torvalds's avatar
Linus Torvalds committed
58
	if (err) {
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
		/* uh oh, we can't allow the inode to go away while there
		 * is still preallocation blocks pending.  Try to join the
		 * aborted transaction
		 */
		jbegin_failure = err;
		err = journal_join_abort(&th, inode->i_sb, 1);

		if (err) {
			/* hmpf, our choices here aren't good.  We can pin the inode
			 * which will disallow unmount from every happening, we can
			 * do nothing, which will corrupt random memory on unmount,
			 * or we can forcibly remove the file from the preallocation
			 * list, which will leak blocks on disk.  Lets pin the inode
			 * and let the admin know what is going on.
			 */
			igrab(inode);
			reiserfs_warning(inode->i_sb,
					 "pinning inode %lu because the "
Alexey Dobriyan's avatar
Alexey Dobriyan committed
77
78
					 "preallocation can't be freed",
					 inode->i_ino);
79
80
			goto out;
		}
Linus Torvalds's avatar
Linus Torvalds committed
81
	}
82
	reiserfs_update_inode_transaction(inode);
Linus Torvalds's avatar
Linus Torvalds committed
83
84

#ifdef REISERFS_PREALLOCATE
85
	reiserfs_discard_prealloc(&th, inode);
Linus Torvalds's avatar
Linus Torvalds committed
86
#endif
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
	err = journal_end(&th, inode->i_sb, 1);

	/* copy back the error code from journal_begin */
	if (!err)
		err = jbegin_failure;

	if (!err && atomic_read(&inode->i_count) <= 1 &&
	    (REISERFS_I(inode)->i_flags & i_pack_on_close_mask) &&
	    tail_has_to_be_packed(inode)) {
		/* if regular file is released by last holder and it has been
		   appended (we append by unformatted node only) or its direct
		   item(s) had to be converted, then it may have to be
		   indirect2direct converted */
		err = reiserfs_truncate_file(inode, 0);
	}
      out:
103
	mutex_unlock(&inode->i_mutex);
104
105
	reiserfs_write_unlock(inode->i_sb);
	return err;
Linus Torvalds's avatar
Linus Torvalds committed
106
107
}

108
109
110
static void reiserfs_vfs_truncate_file(struct inode *inode)
{
	reiserfs_truncate_file(inode, 1);
Linus Torvalds's avatar
Linus Torvalds committed
111
112
113
114
115
116
117
118
119
}

/* Sync a reiserfs file. */

/*
 * FIXME: sync_mapping_buffers() never has anything to sync.  Can
 * be removed...
 */

120
121
122
123
124
125
126
static int reiserfs_sync_file(struct file *p_s_filp,
			      struct dentry *p_s_dentry, int datasync)
{
	struct inode *p_s_inode = p_s_dentry->d_inode;
	int n_err;
	int barrier_done;

127
	BUG_ON(!S_ISREG(p_s_inode->i_mode));
128
129
130
131
	n_err = sync_mapping_buffers(p_s_inode->i_mapping);
	reiserfs_write_lock(p_s_inode->i_sb);
	barrier_done = reiserfs_commit_for_inode(p_s_inode);
	reiserfs_write_unlock(p_s_inode->i_sb);
132
	if (barrier_done != 1 && reiserfs_barrier_flush(p_s_inode->i_sb))
133
134
135
136
		blkdev_issue_flush(p_s_inode->i_sb->s_bdev, NULL);
	if (barrier_done < 0)
		return barrier_done;
	return (n_err < 0) ? -EIO : 0;
Linus Torvalds's avatar
Linus Torvalds committed
137
138
139
140
141
142
143
144
145
146
147
}

/* I really do not want to play with memory shortage right now, so
   to simplify the code, we are not going to write more than this much pages at
   a time. This still should considerably improve performance compared to 4k
   at a time case. This is 32 pages of 4k size. */
#define REISERFS_WRITE_PAGES_AT_A_TIME (128 * 1024) / PAGE_CACHE_SIZE

/* Allocates blocks for a file to fulfil write request.
   Maps all unmapped but prepared pages from the list.
   Updates metadata with newly allocated blocknumbers as needed */
148
149
150
151
152
153
154
155
156
157
158
159
160
static int reiserfs_allocate_blocks_for_region(struct reiserfs_transaction_handle *th, struct inode *inode,	/* Inode we work with */
					       loff_t pos,	/* Writing position */
					       int num_pages,	/* number of pages write going
								   to touch */
					       int write_bytes,	/* amount of bytes to write */
					       struct page **prepared_pages,	/* array of
										   prepared pages
										 */
					       int blocks_to_allocate	/* Amount of blocks we
									   need to allocate to
									   fit the data into file
									 */
    )
Linus Torvalds's avatar
Linus Torvalds committed
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
	struct cpu_key key;	// cpu key of item that we are going to deal with
	struct item_head *ih;	// pointer to item head that we are going to deal with
	struct buffer_head *bh;	// Buffer head that contains items that we are going to deal with
	__le32 *item;		// pointer to item we are going to deal with
	INITIALIZE_PATH(path);	// path to item, that we are going to deal with.
	b_blocknr_t *allocated_blocks;	// Pointer to a place where allocated blocknumbers would be stored.
	reiserfs_blocknr_hint_t hint;	// hint structure for block allocator.
	size_t res;		// return value of various functions that we call.
	int curr_block;		// current block used to keep track of unmapped blocks.
	int i;			// loop counter
	int itempos;		// position in item
	unsigned int from = (pos & (PAGE_CACHE_SIZE - 1));	// writing position in
	// first page
	unsigned int to = ((pos + write_bytes - 1) & (PAGE_CACHE_SIZE - 1)) + 1;	/* last modified byte offset in last page */
	__u64 hole_size;	// amount of blocks for a file hole, if it needed to be created.
	int modifying_this_item = 0;	// Flag for items traversal code to keep track
	// of the fact that we already prepared
	// current block for journal
	int will_prealloc = 0;
	RFALSE(!blocks_to_allocate,
	       "green-9004: tried to allocate zero blocks?");

	/* only preallocate if this is a small write */
	if (REISERFS_I(inode)->i_prealloc_count ||
	    (!(write_bytes & (inode->i_sb->s_blocksize - 1)) &&
	     blocks_to_allocate <
	     REISERFS_SB(inode->i_sb)->s_alloc_options.preallocsize))
		will_prealloc =
		    REISERFS_SB(inode->i_sb)->s_alloc_options.preallocsize;

	allocated_blocks = kmalloc((blocks_to_allocate + will_prealloc) *
				   sizeof(b_blocknr_t), GFP_NOFS);
194
195
	if (!allocated_blocks)
		return -ENOMEM;
196
197
198
199
200
201
202
203
204
205

	/* First we compose a key to point at the writing position, we want to do
	   that outside of any locking region. */
	make_cpu_key(&key, inode, pos + 1, TYPE_ANY, 3 /*key length */ );

	/* If we came here, it means we absolutely need to open a transaction,
	   since we need to allocate some blocks */
	reiserfs_write_lock(inode->i_sb);	// Journaling stuff and we need that.
	res = journal_begin(th, inode->i_sb, JOURNAL_PER_BALANCE_CNT * 3 + 1 + 2 * REISERFS_QUOTA_TRANS_BLOCKS(inode->i_sb));	// Wish I know if this number enough
	if (res)
Linus Torvalds's avatar
Linus Torvalds committed
206
		goto error_exit;
207
	reiserfs_update_inode_transaction(inode);
Linus Torvalds's avatar
Linus Torvalds committed
208

209
210
211
212
	/* Look for the in-tree position of our write, need path for block allocator */
	res = search_for_position_by_key(inode->i_sb, &key, &path);
	if (res == IO_ERROR) {
		res = -EIO;
Linus Torvalds's avatar
Linus Torvalds committed
213
214
215
		goto error_exit;
	}

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
	/* Allocate blocks */
	/* First fill in "hint" structure for block allocator */
	hint.th = th;		// transaction handle.
	hint.path = &path;	// Path, so that block allocator can determine packing locality or whatever it needs to determine.
	hint.inode = inode;	// Inode is needed by block allocator too.
	hint.search_start = 0;	// We have no hint on where to search free blocks for block allocator.
	hint.key = key.on_disk_key;	// on disk key of file.
	hint.block = inode->i_blocks >> (inode->i_sb->s_blocksize_bits - 9);	// Number of disk blocks this file occupies already.
	hint.formatted_node = 0;	// We are allocating blocks for unformatted node.
	hint.preallocate = will_prealloc;

	/* Call block allocator to allocate blocks */
	res =
	    reiserfs_allocate_blocknrs(&hint, allocated_blocks,
				       blocks_to_allocate, blocks_to_allocate);
	if (res != CARRY_ON) {
		if (res == NO_DISK_SPACE) {
			/* We flush the transaction in case of no space. This way some
			   blocks might become free */
			SB_JOURNAL(inode->i_sb)->j_must_wait = 1;
			res = restart_transaction(th, inode, &path);
			if (res)
				goto error_exit;

			/* We might have scheduled, so search again */
			res =
			    search_for_position_by_key(inode->i_sb, &key,
						       &path);
			if (res == IO_ERROR) {
				res = -EIO;
				goto error_exit;
			}
Linus Torvalds's avatar
Linus Torvalds committed
248

249
250
251
252
253
254
			/* update changed info for hint structure. */
			res =
			    reiserfs_allocate_blocknrs(&hint, allocated_blocks,
						       blocks_to_allocate,
						       blocks_to_allocate);
			if (res != CARRY_ON) {
255
				res = res == QUOTA_EXCEEDED ? -EDQUOT : -ENOSPC;
256
257
258
259
				pathrelse(&path);
				goto error_exit;
			}
		} else {
260
			res = res == QUOTA_EXCEEDED ? -EDQUOT : -ENOSPC;
261
262
263
264
265
266
267
			pathrelse(&path);
			goto error_exit;
		}
	}
#ifdef __BIG_ENDIAN
	// Too bad, I have not found any way to convert a given region from
	// cpu format to little endian format
Linus Torvalds's avatar
Linus Torvalds committed
268
	{
269
270
271
		int i;
		for (i = 0; i < blocks_to_allocate; i++)
			allocated_blocks[i] = cpu_to_le32(allocated_blocks[i]);
Linus Torvalds's avatar
Linus Torvalds committed
272
	}
273
#endif
Linus Torvalds's avatar
Linus Torvalds committed
274

275
276
277
278
279
280
	/* Blocks allocating well might have scheduled and tree might have changed,
	   let's search the tree again */
	/* find where in the tree our write should go */
	res = search_for_position_by_key(inode->i_sb, &key, &path);
	if (res == IO_ERROR) {
		res = -EIO;
Linus Torvalds's avatar
Linus Torvalds committed
281
		goto error_exit_free_blocks;
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
	}

	bh = get_last_bh(&path);	// Get a bufferhead for last element in path.
	ih = get_ih(&path);	// Get a pointer to last item head in path.
	item = get_item(&path);	// Get a pointer to last item in path

	/* Let's see what we have found */
	if (res != POSITION_FOUND) {	/* position not found, this means that we
					   might need to append file with holes
					   first */
		// Since we are writing past the file's end, we need to find out if
		// there is a hole that needs to be inserted before our writing
		// position, and how many blocks it is going to cover (we need to
		//  populate pointers to file blocks representing the hole with zeros)

		{
			int item_offset = 1;
			/*
			 * if ih is stat data, its offset is 0 and we don't want to
			 * add 1 to pos in the hole_size calculation
			 */
			if (is_statdata_le_ih(ih))
				item_offset = 0;
			hole_size = (pos + item_offset -
				     (le_key_k_offset
				      (get_inode_item_key_version(inode),
				       &(ih->ih_key)) + op_bytes_number(ih,
									inode->
									i_sb->
									s_blocksize)))
			    >> inode->i_sb->s_blocksize_bits;
		}

		if (hole_size > 0) {
			int to_paste = min_t(__u64, hole_size, MAX_ITEM_LEN(inode->i_sb->s_blocksize) / UNFM_P_SIZE);	// How much data to insert first time.
			/* area filled with zeroes, to supply as list of zero blocknumbers
			   We allocate it outside of loop just in case loop would spin for
			   several iterations. */
320
			char *zeros = kzalloc(to_paste * UNFM_P_SIZE, GFP_ATOMIC);	// We cannot insert more than MAX_ITEM_LEN bytes anyway.
321
322
323
			if (!zeros) {
				res = -ENOMEM;
				goto error_exit_free_blocks;
Linus Torvalds's avatar
Linus Torvalds committed
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
			do {
				to_paste =
				    min_t(__u64, hole_size,
					  MAX_ITEM_LEN(inode->i_sb->
						       s_blocksize) /
					  UNFM_P_SIZE);
				if (is_indirect_le_ih(ih)) {
					/* Ok, there is existing indirect item already. Need to append it */
					/* Calculate position past inserted item */
					make_cpu_key(&key, inode,
						     le_key_k_offset
						     (get_inode_item_key_version
						      (inode),
						      &(ih->ih_key)) +
						     op_bytes_number(ih,
								     inode->
								     i_sb->
								     s_blocksize),
						     TYPE_INDIRECT, 3);
					res =
					    reiserfs_paste_into_item(th, &path,
								     &key,
								     inode,
								     (char *)
								     zeros,
								     UNFM_P_SIZE
								     *
								     to_paste);
					if (res) {
						kfree(zeros);
						goto error_exit_free_blocks;
					}
				} else if (is_statdata_le_ih(ih)) {
					/* No existing item, create it */
					/* item head for new item */
					struct item_head ins_ih;

					/* create a key for our new item */
					make_cpu_key(&key, inode, 1,
						     TYPE_INDIRECT, 3);

					/* Create new item head for our new item */
					make_le_item_head(&ins_ih, &key,
							  key.version, 1,
							  TYPE_INDIRECT,
							  to_paste *
							  UNFM_P_SIZE,
							  0 /* free space */ );

					/* Find where such item should live in the tree */
					res =
					    search_item(inode->i_sb, &key,
							&path);
					if (res != ITEM_NOT_FOUND) {
						/* item should not exist, otherwise we have error */
						if (res != -ENOSPC) {
							reiserfs_warning(inode->
									 i_sb,
									 "green-9008: search_by_key (%K) returned %d",
									 &key,
									 res);
						}
						res = -EIO;
						kfree(zeros);
						goto error_exit_free_blocks;
					}
					res =
					    reiserfs_insert_item(th, &path,
								 &key, &ins_ih,
								 inode,
								 (char *)zeros);
				} else {
					reiserfs_panic(inode->i_sb,
						       "green-9011: Unexpected key type %K\n",
						       &key);
				}
				if (res) {
					kfree(zeros);
					goto error_exit_free_blocks;
				}
				/* Now we want to check if transaction is too full, and if it is
				   we restart it. This will also free the path. */
				if (journal_transaction_should_end
				    (th, th->t_blocks_allocated)) {
409
410
					inode->i_size = cpu_key_k_offset(&key) +
						(to_paste << inode->i_blkbits);
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
					res =
					    restart_transaction(th, inode,
								&path);
					if (res) {
						pathrelse(&path);
						kfree(zeros);
						goto error_exit;
					}
				}

				/* Well, need to recalculate path and stuff */
				set_cpu_key_k_offset(&key,
						     cpu_key_k_offset(&key) +
						     (to_paste << inode->
						      i_blkbits));
				res =
				    search_for_position_by_key(inode->i_sb,
							       &key, &path);
				if (res == IO_ERROR) {
					res = -EIO;
					kfree(zeros);
					goto error_exit_free_blocks;
				}
				bh = get_last_bh(&path);
				ih = get_ih(&path);
				item = get_item(&path);
				hole_size -= to_paste;
			} while (hole_size);
			kfree(zeros);
Linus Torvalds's avatar
Linus Torvalds committed
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
	}
	// Go through existing indirect items first
	// replace all zeroes with blocknumbers from list
	// Note that if no corresponding item was found, by previous search,
	// it means there are no existing in-tree representation for file area
	// we are going to overwrite, so there is nothing to scan through for holes.
	for (curr_block = 0, itempos = path.pos_in_item;
	     curr_block < blocks_to_allocate && res == POSITION_FOUND;) {
	      retry:

		if (itempos >= ih_item_len(ih) / UNFM_P_SIZE) {
			/* We run out of data in this indirect item, let's look for another
			   one. */
			/* First if we are already modifying current item, log it */
			if (modifying_this_item) {
				journal_mark_dirty(th, inode->i_sb, bh);
				modifying_this_item = 0;
			}
			/* Then set the key to look for a new indirect item (offset of old
			   item is added to old item length */
			set_cpu_key_k_offset(&key,
					     le_key_k_offset
					     (get_inode_item_key_version(inode),
					      &(ih->ih_key)) +
					     op_bytes_number(ih,
							     inode->i_sb->
							     s_blocksize));
			/* Search ofor position of new key in the tree. */
			res =
			    search_for_position_by_key(inode->i_sb, &key,
						       &path);
			if (res == IO_ERROR) {
				res = -EIO;
				goto error_exit_free_blocks;
			}
			bh = get_last_bh(&path);
			ih = get_ih(&path);
			item = get_item(&path);
			itempos = path.pos_in_item;
			continue;	// loop to check all kinds of conditions and so on.
Linus Torvalds's avatar
Linus Torvalds committed
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
		/* Ok, we have correct position in item now, so let's see if it is
		   representing file hole (blocknumber is zero) and fill it if needed */
		if (!item[itempos]) {
			/* Ok, a hole. Now we need to check if we already prepared this
			   block to be journaled */
			while (!modifying_this_item) {	// loop until succeed
				/* Well, this item is not journaled yet, so we must prepare
				   it for journal first, before we can change it */
				struct item_head tmp_ih;	// We copy item head of found item,
				// here to detect if fs changed under
				// us while we were preparing for
				// journal.
				int fs_gen;	// We store fs generation here to find if someone
				// changes fs under our feet

				copy_item_head(&tmp_ih, ih);	// Remember itemhead
				fs_gen = get_generation(inode->i_sb);	// remember fs generation
				reiserfs_prepare_for_journal(inode->i_sb, bh, 1);	// Prepare a buffer within which indirect item is stored for changing.
				if (fs_changed(fs_gen, inode->i_sb)
				    && item_moved(&tmp_ih, &path)) {
					// Sigh, fs was changed under us, we need to look for new
					// location of item we are working with

					/* unmark prepaerd area as journaled and search for it's
					   new position */
					reiserfs_restore_prepared_buffer(inode->
									 i_sb,
									 bh);
					res =
					    search_for_position_by_key(inode->
								       i_sb,
								       &key,
								       &path);
					if (res == IO_ERROR) {
						res = -EIO;
						goto error_exit_free_blocks;
					}
					bh = get_last_bh(&path);
					ih = get_ih(&path);
					item = get_item(&path);
					itempos = path.pos_in_item;
					goto retry;
				}
				modifying_this_item = 1;
			}
			item[itempos] = allocated_blocks[curr_block];	// Assign new block
			curr_block++;
Linus Torvalds's avatar
Linus Torvalds committed
529
		}
530
		itempos++;
Linus Torvalds's avatar
Linus Torvalds committed
531
	}
532
533
534
535

	if (modifying_this_item) {	// We need to log last-accessed block, if it
		// was modified, but not logged yet.
		journal_mark_dirty(th, inode->i_sb, bh);
Linus Torvalds's avatar
Linus Torvalds committed
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
590
591
592
593
594
595
596
597
598
599
600
601

	if (curr_block < blocks_to_allocate) {
		// Oh, well need to append to indirect item, or to create indirect item
		// if there weren't any
		if (is_indirect_le_ih(ih)) {
			// Existing indirect item - append. First calculate key for append
			// position. We do not need to recalculate path as it should
			// already point to correct place.
			make_cpu_key(&key, inode,
				     le_key_k_offset(get_inode_item_key_version
						     (inode),
						     &(ih->ih_key)) +
				     op_bytes_number(ih,
						     inode->i_sb->s_blocksize),
				     TYPE_INDIRECT, 3);
			res =
			    reiserfs_paste_into_item(th, &path, &key, inode,
						     (char *)(allocated_blocks +
							      curr_block),
						     UNFM_P_SIZE *
						     (blocks_to_allocate -
						      curr_block));
			if (res) {
				goto error_exit_free_blocks;
			}
		} else if (is_statdata_le_ih(ih)) {
			// Last found item was statdata. That means we need to create indirect item.
			struct item_head ins_ih;	/* itemhead for new item */

			/* create a key for our new item */
			make_cpu_key(&key, inode, 1, TYPE_INDIRECT, 3);	// Position one,
			// because that's
			// where first
			// indirect item
			// begins
			/* Create new item head for our new item */
			make_le_item_head(&ins_ih, &key, key.version, 1,
					  TYPE_INDIRECT,
					  (blocks_to_allocate -
					   curr_block) * UNFM_P_SIZE,
					  0 /* free space */ );
			/* Find where such item should live in the tree */
			res = search_item(inode->i_sb, &key, &path);
			if (res != ITEM_NOT_FOUND) {
				/* Well, if we have found such item already, or some error
				   occured, we need to warn user and return error */
				if (res != -ENOSPC) {
					reiserfs_warning(inode->i_sb,
							 "green-9009: search_by_key (%K) "
							 "returned %d", &key,
							 res);
				}
				res = -EIO;
				goto error_exit_free_blocks;
			}
			/* Insert item into the tree with the data as its body */
			res =
			    reiserfs_insert_item(th, &path, &key, &ins_ih,
						 inode,
						 (char *)(allocated_blocks +
							  curr_block));
		} else {
			reiserfs_panic(inode->i_sb,
				       "green-9010: unexpected item type for key %K\n",
				       &key);
Linus Torvalds's avatar
Linus Torvalds committed
602
603
		}
	}
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
	// the caller is responsible for closing the transaction
	// unless we return an error, they are also responsible for logging
	// the inode.
	//
	pathrelse(&path);
	/*
	 * cleanup prellocation from previous writes
	 * if this is a partial block write
	 */
	if (write_bytes & (inode->i_sb->s_blocksize - 1))
		reiserfs_discard_prealloc(th, inode);
	reiserfs_write_unlock(inode->i_sb);

	// go through all the pages/buffers and map the buffers to newly allocated
	// blocks (so that system knows where to write these pages later).
	curr_block = 0;
	for (i = 0; i < num_pages; i++) {
		struct page *page = prepared_pages[i];	//current page
		struct buffer_head *head = page_buffers(page);	// first buffer for a page
		int block_start, block_end;	// in-page offsets for buffers.

		if (!page_buffers(page))
			reiserfs_panic(inode->i_sb,
				       "green-9005: No buffers for prepared page???");

		/* For each buffer in page */
		for (bh = head, block_start = 0; bh != head || !block_start;
		     block_start = block_end, bh = bh->b_this_page) {
			if (!bh)
				reiserfs_panic(inode->i_sb,
					       "green-9006: Allocated but absent buffer for a page?");
			block_end = block_start + inode->i_sb->s_blocksize;
			if (i == 0 && block_end <= from)
				/* if this buffer is before requested data to map, skip it */
				continue;
			if (i == num_pages - 1 && block_start >= to)
				/* If this buffer is after requested data to map, abort
				   processing of current page */
				break;

			if (!buffer_mapped(bh)) {	// Ok, unmapped buffer, need to map it
				map_bh(bh, inode->i_sb,
				       le32_to_cpu(allocated_blocks
						   [curr_block]));
				curr_block++;
				set_buffer_new(bh);
			}
Linus Torvalds's avatar
Linus Torvalds committed
651
652
653
		}
	}

654
655
	RFALSE(curr_block > blocks_to_allocate,
	       "green-9007: Used too many blocks? weird");
Linus Torvalds's avatar
Linus Torvalds committed
656

657
658
	kfree(allocated_blocks);
	return 0;
Linus Torvalds's avatar
Linus Torvalds committed
659
660

// Need to deal with transaction here.
661
662
663
664
665
666
667
668
669
670
671
      error_exit_free_blocks:
	pathrelse(&path);
	// free blocks
	for (i = 0; i < blocks_to_allocate; i++)
		reiserfs_free_block(th, inode, le32_to_cpu(allocated_blocks[i]),
				    1);

      error_exit:
	if (th->t_trans_id) {
		int err;
		// update any changes we made to blk count
672
		mark_inode_dirty(inode);
673
674
675
676
677
678
679
680
681
682
683
		err =
		    journal_end(th, inode->i_sb,
				JOURNAL_PER_BALANCE_CNT * 3 + 1 +
				2 * REISERFS_QUOTA_TRANS_BLOCKS(inode->i_sb));
		if (err)
			res = err;
	}
	reiserfs_write_unlock(inode->i_sb);
	kfree(allocated_blocks);

	return res;
Linus Torvalds's avatar
Linus Torvalds committed
684
685
686
}

/* Unlock pages prepared by reiserfs_prepare_file_region_for_write */
687
688
689
690
static void reiserfs_unprepare_pages(struct page **prepared_pages,	/* list of locked pages */
				     size_t num_pages /* amount of pages */ )
{
	int i;			// loop counter
Linus Torvalds's avatar
Linus Torvalds committed
691

692
693
	for (i = 0; i < num_pages; i++) {
		struct page *page = prepared_pages[i];
Linus Torvalds's avatar
Linus Torvalds committed
694

695
696
697
698
		try_to_free_buffers(page);
		unlock_page(page);
		page_cache_release(page);
	}
Linus Torvalds's avatar
Linus Torvalds committed
699
700
701
702
}

/* This function will copy data from userspace to specified pages within
   supplied byte range */
703
704
705
706
707
708
709
710
711
712
static int reiserfs_copy_from_user_to_file_region(loff_t pos,	/* In-file position */
						  int num_pages,	/* Number of pages affected */
						  int write_bytes,	/* Amount of bytes to write */
						  struct page **prepared_pages,	/* pointer to 
										   array to
										   prepared pages
										 */
						  const char __user * buf	/* Pointer to user-supplied
										   data */
    )
Linus Torvalds's avatar
Linus Torvalds committed
713
{
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
	long page_fault = 0;	// status of copy_from_user.
	int i;			// loop counter.
	int offset;		// offset in page

	for (i = 0, offset = (pos & (PAGE_CACHE_SIZE - 1)); i < num_pages;
	     i++, offset = 0) {
		size_t count = min_t(size_t, PAGE_CACHE_SIZE - offset, write_bytes);	// How much of bytes to write to this page
		struct page *page = prepared_pages[i];	// Current page we process.

		fault_in_pages_readable(buf, count);

		/* Copy data from userspace to the current page */
		kmap(page);
		page_fault = __copy_from_user(page_address(page) + offset, buf, count);	// Copy the data.
		/* Flush processor's dcache for this page */
		flush_dcache_page(page);
		kunmap(page);
		buf += count;
		write_bytes -= count;

		if (page_fault)
			break;	// Was there a fault? abort.
	}

	return page_fault ? -EFAULT : 0;
Linus Torvalds's avatar
Linus Torvalds committed
739
740
741
742
}

/* taken fs/buffer.c:__block_commit_write */
int reiserfs_commit_page(struct inode *inode, struct page *page,
743
			 unsigned from, unsigned to)
Linus Torvalds's avatar
Linus Torvalds committed
744
{
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
	unsigned block_start, block_end;
	int partial = 0;
	unsigned blocksize;
	struct buffer_head *bh, *head;
	unsigned long i_size_index = inode->i_size >> PAGE_CACHE_SHIFT;
	int new;
	int logit = reiserfs_file_data_log(inode);
	struct super_block *s = inode->i_sb;
	int bh_per_page = PAGE_CACHE_SIZE / s->s_blocksize;
	struct reiserfs_transaction_handle th;
	int ret = 0;

	th.t_trans_id = 0;
	blocksize = 1 << inode->i_blkbits;

	if (logit) {
		reiserfs_write_lock(s);
		ret = journal_begin(&th, s, bh_per_page + 1);
		if (ret)
			goto drop_write_lock;
		reiserfs_update_inode_transaction(inode);
	}
	for (bh = head = page_buffers(page), block_start = 0;
	     bh != head || !block_start;
	     block_start = block_end, bh = bh->b_this_page) {

		new = buffer_new(bh);
		clear_buffer_new(bh);
		block_end = block_start + blocksize;
		if (block_end <= from || block_start >= to) {
			if (!buffer_uptodate(bh))
				partial = 1;
		} else {
			set_buffer_uptodate(bh);
			if (logit) {
				reiserfs_prepare_for_journal(s, bh, 1);
				journal_mark_dirty(&th, s, bh);
			} else if (!buffer_dirty(bh)) {
				mark_buffer_dirty(bh);
				/* do data=ordered on any page past the end
				 * of file and any buffer marked BH_New.
				 */
				if (reiserfs_data_ordered(inode->i_sb) &&
				    (new || page->index >= i_size_index)) {
					reiserfs_add_ordered_list(inode, bh);
				}
			}
		}
Linus Torvalds's avatar
Linus Torvalds committed
793
	}
794
795
796
797
798
799
800
801
802
803
804
805
806
807
	if (logit) {
		ret = journal_end(&th, s, bh_per_page + 1);
	      drop_write_lock:
		reiserfs_write_unlock(s);
	}
	/*
	 * If this is a partial write which happened to make all buffers
	 * uptodate then we can optimize away a bogus readpage() for
	 * the next read(). Here we 'discover' whether the page went
	 * uptodate as a result of this (potentially partial) write.
	 */
	if (!partial)
		SetPageUptodate(page);
	return ret;
Linus Torvalds's avatar
Linus Torvalds committed
808
809
810
811
812
}

/* Submit pages for write. This was separated from actual file copying
   because we might want to allocate block numbers in-between.
   This function assumes that caller will adjust file size to correct value. */
813
814
815
816
817
static int reiserfs_submit_file_region_for_write(struct reiserfs_transaction_handle *th, struct inode *inode, loff_t pos,	/* Writing position offset */
						 size_t num_pages,	/* Number of pages to write */
						 size_t write_bytes,	/* number of bytes to write */
						 struct page **prepared_pages	/* list of pages */
    )
Linus Torvalds's avatar
Linus Torvalds committed
818
{
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
	int status;		// return status of block_commit_write.
	int retval = 0;		// Return value we are going to return.
	int i;			// loop counter
	int offset;		// Writing offset in page.
	int orig_write_bytes = write_bytes;
	int sd_update = 0;

	for (i = 0, offset = (pos & (PAGE_CACHE_SIZE - 1)); i < num_pages;
	     i++, offset = 0) {
		int count = min_t(int, PAGE_CACHE_SIZE - offset, write_bytes);	// How much of bytes to write to this page
		struct page *page = prepared_pages[i];	// Current page we process.

		status =
		    reiserfs_commit_page(inode, page, offset, offset + count);
		if (status)
			retval = status;	// To not overcomplicate matters We are going to
		// submit all the pages even if there was error.
		// we only remember error status to report it on
		// exit.
		write_bytes -= count;
	}
	/* now that we've gotten all the ordered buffers marked dirty,
	 * we can safely update i_size and close any running transaction
	 */
	if (pos + orig_write_bytes > inode->i_size) {
		inode->i_size = pos + orig_write_bytes;	// Set new size
		/* If the file have grown so much that tail packing is no
		 * longer possible, reset "need to pack" flag */
		if ((have_large_tails(inode->i_sb) &&
		     inode->i_size > i_block_size(inode) * 4) ||
		    (have_small_tails(inode->i_sb) &&
		     inode->i_size > i_block_size(inode)))
			REISERFS_I(inode)->i_flags &= ~i_pack_on_close_mask;
		else if ((have_large_tails(inode->i_sb) &&
			  inode->i_size < i_block_size(inode) * 4) ||
			 (have_small_tails(inode->i_sb) &&
			  inode->i_size < i_block_size(inode)))
			REISERFS_I(inode)->i_flags |= i_pack_on_close_mask;

		if (th->t_trans_id) {
			reiserfs_write_lock(inode->i_sb);
860
861
			// this sets the proper flags for O_SYNC to trigger a commit
			mark_inode_dirty(inode);
862
			reiserfs_write_unlock(inode->i_sb);
863
864
865
		} else {
			reiserfs_write_lock(inode->i_sb);
			reiserfs_update_inode_transaction(inode);
866
			mark_inode_dirty(inode);
867
868
			reiserfs_write_unlock(inode->i_sb);
		}
869
870
871

		sd_update = 1;
	}
Linus Torvalds's avatar
Linus Torvalds committed
872
	if (th->t_trans_id) {
873
874
		reiserfs_write_lock(inode->i_sb);
		if (!sd_update)
875
			mark_inode_dirty(inode);
876
877
878
879
880
881
		status = journal_end(th, th->t_super, th->t_blocks_allocated);
		if (status)
			retval = status;
		reiserfs_write_unlock(inode->i_sb);
	}
	th->t_trans_id = 0;
Linus Torvalds's avatar
Linus Torvalds committed
882

883
884
885
886
887
888
889
890
891
892
893
	/* 
	 * we have to unlock the pages after updating i_size, otherwise
	 * we race with writepage
	 */
	for (i = 0; i < num_pages; i++) {
		struct page *page = prepared_pages[i];
		unlock_page(page);
		mark_page_accessed(page);
		page_cache_release(page);
	}
	return retval;
Linus Torvalds's avatar
Linus Torvalds committed
894
895
896
897
}

/* Look if passed writing region is going to touch file's tail
   (if it is present). And if it is, convert the tail to unformatted node */
898
899
900
901
static int reiserfs_check_for_tail_and_convert(struct inode *inode,	/* inode to deal with */
					       loff_t pos,	/* Writing position */
					       int write_bytes	/* amount of bytes to write */
    )
Linus Torvalds's avatar
Linus Torvalds committed
902
{
903
904
905
906
907
908
909
910
	INITIALIZE_PATH(path);	// needed for search_for_position
	struct cpu_key key;	// Key that would represent last touched writing byte.
	struct item_head *ih;	// item header of found block;
	int res;		// Return value of various functions we call.
	int cont_expand_offset;	// We will put offset for generic_cont_expand here
	// This can be int just because tails are created
	// only for small files.

Linus Torvalds's avatar
Linus Torvalds committed
911
/* this embodies a dependency on a particular tail policy */
912
913
914
915
916
	if (inode->i_size >= inode->i_sb->s_blocksize * 4) {
		/* such a big files do not have tails, so we won't bother ourselves
		   to look for tails, simply return */
		return 0;
	}
Linus Torvalds's avatar
Linus Torvalds committed
917

918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
	reiserfs_write_lock(inode->i_sb);
	/* find the item containing the last byte to be written, or if
	 * writing past the end of the file then the last item of the
	 * file (and then we check its type). */
	make_cpu_key(&key, inode, pos + write_bytes + 1, TYPE_ANY,
		     3 /*key length */ );
	res = search_for_position_by_key(inode->i_sb, &key, &path);
	if (res == IO_ERROR) {
		reiserfs_write_unlock(inode->i_sb);
		return -EIO;
	}
	ih = get_ih(&path);
	res = 0;
	if (is_direct_le_ih(ih)) {
		/* Ok, closest item is file tail (tails are stored in "direct"
		 * items), so we need to unpack it. */
		/* To not overcomplicate matters, we just call generic_cont_expand
		   which will in turn call other stuff and finally will boil down to
		   reiserfs_get_block() that would do necessary conversion. */
		cont_expand_offset =
		    le_key_k_offset(get_inode_item_key_version(inode),
				    &(ih->ih_key));
		pathrelse(&path);
		res = generic_cont_expand(inode, cont_expand_offset);
	} else
		pathrelse(&path);

	reiserfs_write_unlock(inode->i_sb);
	return res;
Linus Torvalds's avatar
Linus Torvalds committed
947
948
949
950
951
952
953
954
955
956
}

/* This function locks pages starting from @pos for @inode.
   @num_pages pages are locked and stored in
   @prepared_pages array. Also buffers are allocated for these pages.
   First and last page of the region is read if it is overwritten only
   partially. If last page did not exist before write (file hole or file
   append), it is zeroed, then. 
   Returns number of unallocated blocks that should be allocated to cover
   new file data.*/
957
958
959
960
961
962
963
964
965
966
967
968
static int reiserfs_prepare_file_region_for_write(struct inode *inode
						  /* Inode of the file */ ,
						  loff_t pos,	/* position in the file */
						  size_t num_pages,	/* number of pages to
									   prepare */
						  size_t write_bytes,	/* Amount of bytes to be
									   overwritten from
									   @pos */
						  struct page **prepared_pages	/* pointer to array
										   where to store
										   prepared pages */
    )
Linus Torvalds's avatar
Linus Torvalds committed
969
{
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
	int res = 0;		// Return values of different functions we call.
	unsigned long index = pos >> PAGE_CACHE_SHIFT;	// Offset in file in pages.
	int from = (pos & (PAGE_CACHE_SIZE - 1));	// Writing offset in first page
	int to = ((pos + write_bytes - 1) & (PAGE_CACHE_SIZE - 1)) + 1;
	/* offset of last modified byte in last
	   page */
	struct address_space *mapping = inode->i_mapping;	// Pages are mapped here.
	int i;			// Simple counter
	int blocks = 0;		/* Return value (blocks that should be allocated) */
	struct buffer_head *bh, *head;	// Current bufferhead and first bufferhead
	// of a page.
	unsigned block_start, block_end;	// Starting and ending offsets of current
	// buffer in the page.
	struct buffer_head *wait[2], **wait_bh = wait;	// Buffers for page, if
	// Page appeared to be not up
	// to date. Note how we have
	// at most 2 buffers, this is
	// because we at most may
	// partially overwrite two
	// buffers for one page. One at                                                 // the beginning of write area
	// and one at the end.
	// Everything inthe middle gets                                                 // overwritten totally.

	struct cpu_key key;	// cpu key of item that we are going to deal with
	struct item_head *ih = NULL;	// pointer to item head that we are going to deal with
	struct buffer_head *itembuf = NULL;	// Buffer head that contains items that we are going to deal with
	INITIALIZE_PATH(path);	// path to item, that we are going to deal with.
	__le32 *item = NULL;	// pointer to item we are going to deal with
	int item_pos = -1;	/* Position in indirect item */

	if (num_pages < 1) {
		reiserfs_warning(inode->i_sb,
				 "green-9001: reiserfs_prepare_file_region_for_write "
				 "called with zero number of pages to process");
		return -EFAULT;
Linus Torvalds's avatar
Linus Torvalds committed
1005
1006
	}

1007
1008
1009
1010
1011
1012
1013
1014
	/* We have 2 loops for pages. In first loop we grab and lock the pages, so
	   that nobody would touch these until we release the pages. Then
	   we'd start to deal with mapping buffers to blocks. */
	for (i = 0; i < num_pages; i++) {
		prepared_pages[i] = grab_cache_page(mapping, index + i);	// locks the page
		if (!prepared_pages[i]) {
			res = -ENOMEM;
			goto failed_page_grabbing;
Linus Torvalds's avatar
Linus Torvalds committed
1015
		}
1016
1017
1018
1019
		if (!page_has_buffers(prepared_pages[i]))
			create_empty_buffers(prepared_pages[i],
					     inode->i_sb->s_blocksize, 0);
	}
Linus Torvalds's avatar
Linus Torvalds committed
1020

1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
	/* Let's count amount of blocks for a case where all the blocks
	   overwritten are new (we will substract already allocated blocks later) */
	if (num_pages > 2)
		/* These are full-overwritten pages so we count all the blocks in
		   these pages are counted as needed to be allocated */
		blocks =
		    (num_pages - 2) << (PAGE_CACHE_SHIFT - inode->i_blkbits);

	/* count blocks needed for first page (possibly partially written) */
	blocks += ((PAGE_CACHE_SIZE - from) >> inode->i_blkbits) + !!(from & (inode->i_sb->s_blocksize - 1));	/* roundup */

	/* Now we account for last page. If last page == first page (we
	   overwrite only one page), we substract all the blocks past the
	   last writing position in a page out of already calculated number
	   of blocks */
	blocks += ((num_pages > 1) << (PAGE_CACHE_SHIFT - inode->i_blkbits)) -
	    ((PAGE_CACHE_SIZE - to) >> inode->i_blkbits);
	/* Note how we do not roundup here since partial blocks still
	   should be allocated */

	/* Now if all the write area lies past the file end, no point in
	   maping blocks, since there is none, so we just zero out remaining
	   parts of first and last pages in write area (if needed) */
	if ((pos & ~((loff_t) PAGE_CACHE_SIZE - 1)) > inode->i_size) {
		if (from != 0) {	/* First page needs to be partially zeroed */
			char *kaddr = kmap_atomic(prepared_pages[0], KM_USER0);
			memset(kaddr, 0, from);
			kunmap_atomic(kaddr, KM_USER0);
1049
			flush_dcache_page(prepared_pages[0]);
1050
1051
1052
1053
1054
1055
1056
		}
		if (to != PAGE_CACHE_SIZE) {	/* Last page needs to be partially zeroed */
			char *kaddr =
			    kmap_atomic(prepared_pages[num_pages - 1],
					KM_USER0);
			memset(kaddr + to, 0, PAGE_CACHE_SIZE - to);
			kunmap_atomic(kaddr, KM_USER0);
1057
			flush_dcache_page(prepared_pages[num_pages - 1]);
Linus Torvalds's avatar
Linus Torvalds committed
1058
1059
		}

1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
		/* Since all blocks are new - use already calculated value */
		return blocks;
	}

	/* Well, since we write somewhere into the middle of a file, there is
	   possibility we are writing over some already allocated blocks, so
	   let's map these blocks and substract number of such blocks out of blocks
	   we need to allocate (calculated above) */
	/* Mask write position to start on blocksize, we do it out of the
	   loop for performance reasons */
	pos &= ~((loff_t) inode->i_sb->s_blocksize - 1);
	/* Set cpu key to the starting position in a file (on left block boundary) */
	make_cpu_key(&key, inode,
		     1 + ((pos) & ~((loff_t) inode->i_sb->s_blocksize - 1)),
		     TYPE_ANY, 3 /*key length */ );

	reiserfs_write_lock(inode->i_sb);	// We need that for at least search_by_key()
	for (i = 0; i < num_pages; i++) {

		head = page_buffers(prepared_pages[i]);
		/* For each buffer in the page */
		for (bh = head, block_start = 0; bh != head || !block_start;
		     block_start = block_end, bh = bh->b_this_page) {
			if (!bh)
				reiserfs_panic(inode->i_sb,
					       "green-9002: Allocated but absent buffer for a page?");
			/* Find where this buffer ends */
			block_end = block_start + inode->i_sb->s_blocksize;
			if (i == 0 && block_end <= from)
				/* if this buffer is before requested data to map, skip it */
				continue;

			if (i == num_pages - 1 && block_start >= to) {
				/* If this buffer is after requested data to map, abort
				   processing of current page */
				break;
Linus Torvalds's avatar
Linus Torvalds committed
1096
1097
			}

1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
			if (buffer_mapped(bh) && bh->b_blocknr != 0) {
				/* This is optimisation for a case where buffer is mapped
				   and have blocknumber assigned. In case significant amount
				   of such buffers are present, we may avoid some amount
				   of search_by_key calls.
				   Probably it would be possible to move parts of this code
				   out of BKL, but I afraid that would overcomplicate code
				   without any noticeable benefit.
				 */
				item_pos++;
				/* Update the key */
				set_cpu_key_k_offset(&key,
						     cpu_key_k_offset(&key) +
						     inode->i_sb->s_blocksize);
				blocks--;	// Decrease the amount of blocks that need to be
				// allocated
				continue;	// Go to the next buffer
			}
Linus Torvalds's avatar
Linus Torvalds committed
1116

1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
			if (!itembuf ||	/* if first iteration */
			    item_pos >= ih_item_len(ih) / UNFM_P_SIZE) {	/* or if we progressed past the
										   current unformatted_item */
				/* Try to find next item */
				res =
				    search_for_position_by_key(inode->i_sb,
							       &key, &path);
				/* Abort if no more items */
				if (res != POSITION_FOUND) {
					/* make sure later loops don't use this item */
					itembuf = NULL;
					item = NULL;
					break;
				}

				/* Update information about current indirect item */
				itembuf = get_last_bh(&path);
				ih = get_ih(&path);
				item = get_item(&path);
				item_pos = path.pos_in_item;

				RFALSE(!is_indirect_le_ih(ih),
				       "green-9003: indirect item expected");
			}
Linus Torvalds's avatar
Linus Torvalds committed
1141

1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
			/* See if there is some block associated with the file
			   at that position, map the buffer to this block */
			if (get_block_num(item, item_pos)) {
				map_bh(bh, inode->i_sb,
				       get_block_num(item, item_pos));
				blocks--;	// Decrease the amount of blocks that need to be
				// allocated
			}
			item_pos++;
			/* Update the key */
			set_cpu_key_k_offset(&key,
					     cpu_key_k_offset(&key) +
					     inode->i_sb->s_blocksize);
Linus Torvalds's avatar
Linus Torvalds committed
1155
1156
		}
	}
1157
1158
	pathrelse(&path);	// Free the path
	reiserfs_write_unlock(inode->i_sb);
Linus Torvalds's avatar
Linus Torvalds committed
1159
1160
1161
1162

	/* Now zero out unmappend buffers for the first and last pages of
	   write area or issue read requests if page is mapped. */
	/* First page, see if it is not uptodate */
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
	if (!PageUptodate(prepared_pages[0])) {
		head = page_buffers(prepared_pages[0]);

		/* For each buffer in page */
		for (bh = head, block_start = 0; bh != head || !block_start;
		     block_start = block_end, bh = bh->b_this_page) {

			if (!bh)
				reiserfs_panic(inode->i_sb,
					       "green-9002: Allocated but absent buffer for a page?");
			/* Find where this buffer ends */
			block_end = block_start + inode->i_sb->s_blocksize;
			if (block_end <= from)
				/* if this buffer is before requested data to map, skip it */
				continue;
			if (block_start < from) {	/* Aha, our partial buffer */
				if (buffer_mapped(bh)) {	/* If it is mapped, we need to
								   issue READ request for it to
								   not loose data */
					ll_rw_block(READ, 1, &bh);
					*wait_bh++ = bh;
				} else {	/* Not mapped, zero it */
					char *kaddr =
					    kmap_atomic(prepared_pages[0],
							KM_USER0);
					memset(kaddr + block_start, 0,
					       from - block_start);
					kunmap_atomic(kaddr, KM_USER0);
1191
					flush_dcache_page(prepared_pages[0]);
1192
1193
1194
					set_buffer_uptodate(bh);
				}
			}
Linus Torvalds's avatar
Linus Torvalds committed
1195
1196
1197
1198
		}
	}

	/* Last page, see if it is not uptodate, or if the last page is past the end of the file. */
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
	if (!PageUptodate(prepared_pages[num_pages - 1]) ||
	    ((pos + write_bytes) >> PAGE_CACHE_SHIFT) >
	    (inode->i_size >> PAGE_CACHE_SHIFT)) {
		head = page_buffers(prepared_pages[num_pages - 1]);

		/* for each buffer in page */
		for (bh = head, block_start = 0; bh != head || !block_start;
		     block_start = block_end, bh = bh->b_this_page) {

			if (!bh)
				reiserfs_panic(inode->i_sb,
					       "green-9002: Allocated but absent buffer for a page?");
			/* Find where this buffer ends */
			block_end = block_start + inode->i_sb->s_blocksize;
			if (block_start >= to)
				/* if this buffer is after requested data to map, skip it */
				break;
			if (block_end > to) {	/* Aha, our partial buffer */
				if (buffer_mapped(bh)) {	/* If it is mapped, we need to
								   issue READ request for it to
								   not loose data */
					ll_rw_block(READ, 1, &bh);
					*wait_bh++ = bh;
				} else {	/* Not mapped, zero it */
					char *kaddr =
					    kmap_atomic(prepared_pages
							[num_pages - 1],
							KM_USER0);
					memset(kaddr + to, 0, block_end - to);
					kunmap_atomic(kaddr, KM_USER0);
1229
					flush_dcache_page(prepared_pages[num_pages - 1]);
1230
1231
1232
					set_buffer_uptodate(bh);
				}
			}
Linus Torvalds's avatar
Linus Torvalds committed
1233
1234
1235
		}
	}

1236
1237
1238
1239
1240
1241
1242
	/* Wait for read requests we made to happen, if necessary */
	while (wait_bh > wait) {
		wait_on_buffer(*--wait_bh);
		if (!buffer_uptodate(*wait_bh)) {
			res = -EIO;
			goto failed_read;
		}
Linus Torvalds's avatar
Linus Torvalds committed
1243
	}
1244
1245
1246
1247
1248
1249
1250

	return blocks;
      failed_page_grabbing:
	num_pages = i;
      failed_read:
	reiserfs_unprepare_pages(prepared_pages, num_pages);
	return res;
Linus Torvalds's avatar
Linus Torvalds committed
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
}

/* Write @count bytes at position @ppos in a file indicated by @file
   from the buffer @buf.  

   generic_file_write() is only appropriate for filesystems that are not seeking to optimize performance and want
   something simple that works.  It is not for serious use by general purpose filesystems, excepting the one that it was
   written for (ext2/3).  This is for several reasons:

   * It has no understanding of any filesystem specific optimizations.

   * It enters the filesystem repeatedly for each page that is written.

   * It depends on reiserfs_get_block() function which if implemented by reiserfs performs costly search_by_key
   * operation for each page it is supplied with. By contrast reiserfs_file_write() feeds as much as possible at a time
   * to reiserfs which allows for fewer tree traversals.

   * Each indirect pointer insertion takes a lot of cpu, because it involves memory moves inside of blocks.

   * Asking the block allocation code for blocks one at a time is slightly less efficient.

   All of these reasons for not using only generic file write were understood back when reiserfs was first miscoded to
   use it, but we were in a hurry to make code freeze, and so it couldn't be revised then.  This new code should make
   things right finally.

   Future Features: providing search_by_key with hints.

*/
1279
1280
1281
1282
1283
1284
1285
static ssize_t reiserfs_file_write(struct file *file,	/* the file we are going to write into */
				   const char __user * buf,	/*  pointer to user supplied data
								   (in userspace) */
				   size_t count,	/* amount of bytes to write */
				   loff_t * ppos	/* pointer to position in file that we start writing at. Should be updated to
							 * new current position before returning. */
				   )
Linus Torvalds's avatar
Linus Torvalds committed
1286
{
1287
1288
1289
1290
	size_t already_written = 0;	// Number of bytes already written to the file.
	loff_t pos;		// Current position in the file.
	ssize_t res;		// return value of various functions that we call.
	int err = 0;
1291
	struct inode *inode = file->f_path.dentry->d_inode;	// Inode of the file that we are writing to.
1292
1293
1294
1295
1296
1297
	/* To simplify coding at this time, we store
	   locked pages in array for now */
	struct page *prepared_pages[REISERFS_WRITE_PAGES_AT_A_TIME];
	struct reiserfs_transaction_handle th;
	th.t_trans_id = 0;

1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
1311
1312
1313
1314
	/* If a filesystem is converted from 3.5 to 3.6, we'll have v3.5 items
	* lying around (most of the disk, in fact). Despite the filesystem
	* now being a v3.6 format, the old items still can't support large
	* file sizes. Catch this case here, as the rest of the VFS layer is
	* oblivious to the different limitations between old and new items.
	* reiserfs_setattr catches this for truncates. This chunk is lifted
	* from generic_write_checks. */
	if (get_inode_item_key_version (inode) == KEY_FORMAT_3_5 &&
	    *ppos + count > MAX_NON_LFS) {
		if (*ppos >= MAX_NON_LFS) {
			send_sig(SIGXFSZ, current, 0);
			return -EFBIG;
		}
		if (count > MAX_NON_LFS - (unsigned long)*ppos)
			count = MAX_NON_LFS - (unsigned long)*ppos;
	}

1315
1316
	if (file->f_flags & O_DIRECT)
		return do_sync_write(file, buf, count, ppos);
Linus Torvalds's avatar
Linus Torvalds committed
1317

1318
1319
1320
1321
1322
1323
	if (unlikely((ssize_t) count < 0))
		return -EINVAL;

	if (unlikely(!access_ok(VERIFY_READ, buf, count)))
		return -EFAULT;

1324
	mutex_lock(&inode->i_mutex);	// locks the entire file for just us
1325
1326
1327
1328
1329
1330
1331
1332
1333
1334
1335
1336
1337

	pos = *ppos;

	/* Check if we can write to specified region of file, file
	   is not overly big and this kind of stuff. Adjust pos and
	   count, if needed */
	res = generic_write_checks(file, &pos, &count, 0);
	if (res)
		goto out;

	if (count == 0)
		goto out;

1338
	res = remove_suid(file->f_path.dentry);
1339
1340
1341
	if (res)
		goto out;

1342
	file_update_time(file);
1343
1344
1345
1346
1347
1348
1349
1350
1351
1352
1353
1354
1355
1356
1357
1358
1359
1360
1361
1362
1363
1364
1365
1366
1367
1368
1369
1370
1371
1372
1373
1374
1375
1376
1377
1378
1379
1380
1381
1382
1383
1384
1385
1386
1387
1388
1389
1390
1391
1392
1393
1394
1395
1396
1397
1398
1399
1400
1401
1402
1403
1404
1405
1406
1407
1408
1409
1410
1411
1412
1413
1414
1415
1416
1417
1418
1419
1420
1421

	// Ok, we are done with all the checks.

	// Now we should start real work

	/* If we are going to write past the file's packed tail or if we are going
	   to overwrite part of the tail, we need that tail to be converted into
	   unformatted node */
	res = reiserfs_check_for_tail_and_convert(inode, pos, count);
	if (res)
		goto out;

	while (count > 0) {
		/* This is the main loop in which we running until some error occures
		   or until we write all of the data. */
		size_t num_pages;	/* amount of pages we are going to write this iteration */
		size_t write_bytes;	/* amount of bytes to write during this iteration */
		size_t blocks_to_allocate;	/* how much blocks we need to allocate for this iteration */

		/*  (pos & (PAGE_CACHE_SIZE-1)) is an idiom for offset into a page of pos */
		num_pages = !!((pos + count) & (PAGE_CACHE_SIZE - 1)) +	/* round up partial
									   pages */
		    ((count +
		      (pos & (PAGE_CACHE_SIZE - 1))) >> PAGE_CACHE_SHIFT);
		/* convert size to amount of
		   pages */
		reiserfs_write_lock(inode->i_sb);
		if (num_pages > REISERFS_WRITE_PAGES_AT_A_TIME
		    || num_pages > reiserfs_can_fit_pages(inode->i_sb)) {
			/* If we were asked to write more data than we want to or if there
			   is not that much space, then we shorten amount of data to write
			   for this iteration. */
			num_pages =
			    min_t(size_t, REISERFS_WRITE_PAGES_AT_A_TIME,
				  reiserfs_can_fit_pages(inode->i_sb));
			/* Also we should not forget to set size in bytes accordingly */
			write_bytes = (num_pages << PAGE_CACHE_SHIFT) -
			    (pos & (PAGE_CACHE_SIZE - 1));
			/* If position is not on the
			   start of the page, we need
			   to substract the offset
			   within page */
		} else
			write_bytes = count;

		/* reserve the blocks to be allocated later, so that later on
		   we still have the space to write the blocks to */
		reiserfs_claim_blocks_to_be_allocated(inode->i_sb,
						      num_pages <<
						      (PAGE_CACHE_SHIFT -
						       inode->i_blkbits));
		reiserfs_write_unlock(inode->i_sb);

		if (!num_pages) {	/* If we do not have enough space even for a single page... */
			if (pos >
			    inode->i_size + inode->i_sb->s_blocksize -
			    (pos & (inode->i_sb->s_blocksize - 1))) {
				res = -ENOSPC;
				break;	// In case we are writing past the end of the last file block, break.
			}
			// Otherwise we are possibly overwriting the file, so
			// let's set write size to be equal or less than blocksize.
			// This way we get it correctly for file holes.
			// But overwriting files on absolutelly full volumes would not
			// be very efficient. Well, people are not supposed to fill
			// 100% of disk space anyway.
			write_bytes =
			    min_t(size_t, count,
				  inode->i_sb->s_blocksize -
				  (pos & (inode->i_sb->s_blocksize - 1)));
			num_pages = 1;
			// No blocks were claimed before, so do it now.
			reiserfs_claim_blocks_to_be_allocated(inode->i_sb,
							      1 <<
							      (PAGE_CACHE_SHIFT
							       -
							       inode->
							       i_blkbits));
		}
Linus Torvalds's avatar
Linus Torvalds committed
1422

1423
1424
1425
1426
		/* Prepare for writing into the region, read in all the
		   partially overwritten pages, if needed. And lock the pages,
		   so that nobody else can access these until we are done.
		   We get number of actual blocks needed as a result. */
1427
1428
1429
1430
1431
		res = reiserfs_prepare_file_region_for_write(inode, pos,
							     num_pages,
							     write_bytes,
							     prepared_pages);
		if (res < 0) {
1432
1433
1434
1435
1436
1437
			reiserfs_release_claimed_blocks(inode->i_sb,
							num_pages <<
							(PAGE_CACHE_SHIFT -
							 inode->i_blkbits));
			break;
		}
Linus Torvalds's avatar
Linus Torvalds committed
1438

1439
1440
		blocks_to_allocate = res;

1441
1442
1443
1444
1445
1446
1447
1448
1449
1450
1451
1452
1453
1454
1455
1456
1457
		/* First we correct our estimate of how many blocks we need */
		reiserfs_release_claimed_blocks(inode->i_sb,
						(num_pages <<
						 (PAGE_CACHE_SHIFT -
						  inode->i_sb->
						  s_blocksize_bits)) -
						blocks_to_allocate);

		if (blocks_to_allocate > 0) {	/*We only allocate blocks if we need to */
			/* Fill in all the possible holes and append the file if needed */
			res =
			    reiserfs_allocate_blocks_for_region(&th, inode, pos,
								num_pages,
								write_bytes,
								prepared_pages,
								blocks_to_allocate);
		}
Linus Torvalds's avatar
Linus Torvalds committed
1458

1459
1460
1461
1462
1463
1464
1465
1466
		/* well, we have allocated the blocks, so it is time to free
		   the reservation we made earlier. */
		reiserfs_release_claimed_blocks(inode->i_sb,
						blocks_to_allocate);
		if (res) {
			reiserfs_unprepare_pages(prepared_pages, num_pages);
			break;
		}
Linus Torvalds's avatar
Linus Torvalds committed
1467

1468
1469
1470
/* NOTE that allocating blocks and filling blocks can be done in reverse order
   and probably we would do that just to get rid of garbage in files after a
   crash */
Linus Torvalds's avatar
Linus Torvalds committed
1471

1472
1473
1474
1475
1476
1477
1478
1479
1480
		/* Copy data from user-supplied buffer to file's pages */
		res =
		    reiserfs_copy_from_user_to_file_region(pos, num_pages,
							   write_bytes,
							   prepared_pages, buf);
		if (res) {
			reiserfs_unprepare_pages(prepared_pages, num_pages);
			break;
		}
Linus Torvalds's avatar
Linus Torvalds committed
1481

1482
1483
1484
1485
1486
1487
1488
1489
1490
1491
1492
1493
1494
		/* Send the pages to disk and unlock them. */
		res =
		    reiserfs_submit_file_region_for_write(&th, inode, pos,
							  num_pages,
							  write_bytes,
							  prepared_pages);
		if (res)
			break;

		already_written += write_bytes;
		buf += write_bytes;
		*ppos = pos += write_bytes;
		count -= write_bytes;
1495
		balance_dirty_pages_ratelimited_nr(inode->i_mapping, num_pages);
Linus Torvalds's avatar
Linus Torvalds committed
1496
1497
	}

1498
1499
1500
1501
1502
1503
1504
1505
1506
	/* this is only true on error */
	if (th.t_trans_id) {
		reiserfs_write_lock(inode->i_sb);
		err = journal_end(&th, th.t_super, th.t_blocks_allocated);
		reiserfs_write_unlock(inode->i_sb);
		if (err) {
			res = err;
			goto out;
		}
Linus Torvalds's avatar
Linus Torvalds committed
1507
1508
	}

1509
1510
1511
1512
	if (likely(res >= 0) &&
	    (unlikely((file->f_flags & O_SYNC) || IS_SYNC(inode))))
		res = generic_osync_inode(inode, file->f_mapping,
		                          OSYNC_METADATA | OSYNC_DATA);
Linus Torvalds's avatar
Linus Torvalds committed
1513

1514
	mutex_unlock(&inode->i_mutex);
1515
1516
	reiserfs_async_progress_wait(inode->i_sb);
	return (already_written != 0) ? already_written : res;
Linus Torvalds's avatar
Linus Torvalds committed
1517

1518
      out:
1519
	mutex_unlock(&inode->i_mutex);	// unlock the file on exit.
1520
	return res;
Linus Torvalds's avatar
Linus Torvalds committed
1521
1522
}

1523
const struct file_operations reiserfs_file_operations = {
1524
	.read = do_sync_read,
1525
1526
	.write = reiserfs_file_write,
	.ioctl = reiserfs_ioctl,
1527
1528
1529
#ifdef CONFIG_COMPAT
	.compat_ioctl = reiserfs_compat_ioctl,
#endif
1530
	.mmap = generic_file_mmap,
1531
	.open = generic_file_open,
1532
1533
1534
1535
	.release = reiserfs_file_release,
	.fsync = reiserfs_sync_file,
	.sendfile = generic_file_sendfile,
	.aio_read = generic_file_aio_read,
1536
	.aio_write = generic_file_aio_write,
1537
1538
	.splice_read = generic_file_splice_read,
	.splice_write = generic_file_splice_write,
Linus Torvalds's avatar
Linus Torvalds committed
1539
1540
};

1541
1542
1543
1544
1545
1546
1547
1548
struct inode_operations reiserfs_file_inode_operations = {
	.truncate = reiserfs_vfs_truncate_file,
	.setattr = reiserfs_setattr,
	.setxattr = reiserfs_setxattr,
	.getxattr = reiserfs_getxattr,
	.listxattr = reiserfs_listxattr,
	.removexattr = reiserfs_removexattr,
	.permission = reiserfs_permission,
Linus Torvalds's avatar
Linus Torvalds committed
1549
};