summaryrefslogtreecommitdiffstats
path: root/fs/jfs/jfs_extent.c
blob: f65bd6b35412b936319a80bca37e4fd0a5aaf95c (plain)
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
590
591
592
593
594
595
596
597
598
599
600
601
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
// SPDX-License-Identifier: GPL-2.0-or-later
/*
 *   Copyright (C) International Business Machines Corp., 2000-2004
 */

#include <linux/fs.h>
#include <linux/quotaops.h>
#include "jfs_incore.h"
#include "jfs_inode.h"
#include "jfs_superblock.h"
#include "jfs_dmap.h"
#include "jfs_extent.h"
#include "jfs_debug.h"

/*
 * forward references
 */
static int extBalloc(struct inode *, s64, s64 *, s64 *);
#ifdef _NOTYET
static int extBrealloc(struct inode *, s64, s64, s64 *, s64 *);
#endif
static s64 extRoundDown(s64 nb);

#define DPD(a)		(printk("(a): %d\n",(a)))
#define DPC(a)		(printk("(a): %c\n",(a)))
#define DPL1(a)					\
{						\
	if ((a) >> 32)				\
		printk("(a): %x%08x  ",(a));	\
	else					\
		printk("(a): %x  ",(a) << 32);	\
}
#define DPL(a)					\
{						\
	if ((a) >> 32)				\
		printk("(a): %x%08x\n",(a));	\
	else					\
		printk("(a): %x\n",(a) << 32);	\
}

#define DPD1(a)		(printk("(a): %d  ",(a)))
#define DPX(a)		(printk("(a): %08x\n",(a)))
#define DPX1(a)		(printk("(a): %08x  ",(a)))
#define DPS(a)		(printk("%s\n",(a)))
#define DPE(a)		(printk("\nENTERING: %s\n",(a)))
#define DPE1(a)		(printk("\nENTERING: %s",(a)))
#define DPS1(a)		(printk("  %s  ",(a)))


/*
 * NAME:	extAlloc()
 *
 * FUNCTION:	allocate an extent for a specified page range within a
 *		file.
 *
 * PARAMETERS:
 *	ip	- the inode of the file.
 *	xlen	- requested extent length.
 *	pno	- the starting page number with the file.
 *	xp	- pointer to an xad.  on entry, xad describes an
 *		  extent that is used as an allocation hint if the
 *		  xaddr of the xad is non-zero.  on successful exit,
 *		  the xad describes the newly allocated extent.
 *	abnr	- bool indicating whether the newly allocated extent
 *		  should be marked as allocated but not recorded.
 *
 * RETURN VALUES:
 *	0	- success
 *	-EIO	- i/o error.
 *	-ENOSPC	- insufficient disk resources.
 */
int
extAlloc(struct inode *ip, s64 xlen, s64 pno, xad_t * xp, bool abnr)
{
	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
	s64 nxlen, nxaddr, xoff, hint, xaddr = 0;
	int rc;
	int xflag;

	/* This blocks if we are low on resources */
	txBeginAnon(ip->i_sb);

	/* Avoid race with jfs_commit_inode() */
	mutex_lock(&JFS_IP(ip)->commit_mutex);

	/* validate extent length */
	if (xlen > MAXXLEN)
		xlen = MAXXLEN;

	/* get the page's starting extent offset */
	xoff = pno << sbi->l2nbperpage;

	/* check if an allocation hint was provided */
	if ((hint = addressXAD(xp))) {
		/* get the size of the extent described by the hint */
		nxlen = lengthXAD(xp);

		/* check if the hint is for the portion of the file
		 * immediately previous to the current allocation
		 * request and if hint extent has the same abnr
		 * value as the current request.  if so, we can
		 * extend the hint extent to include the current
		 * extent if we can allocate the blocks immediately
		 * following the hint extent.
		 */
		if (offsetXAD(xp) + nxlen == xoff &&
		    abnr == ((xp->flag & XAD_NOTRECORDED) ? true : false))
			xaddr = hint + nxlen;

		/* adjust the hint to the last block of the extent */
		hint += (nxlen - 1);
	}

	/* allocate the disk blocks for the extent.  initially, extBalloc()
	 * will try to allocate disk blocks for the requested size (xlen).
	 * if this fails (xlen contiguous free blocks not available), it'll
	 * try to allocate a smaller number of blocks (producing a smaller
	 * extent), with this smaller number of blocks consisting of the
	 * requested number of blocks rounded down to the next smaller
	 * power of 2 number (i.e. 16 -> 8).  it'll continue to round down
	 * and retry the allocation until the number of blocks to allocate
	 * is smaller than the number of blocks per page.
	 */
	nxlen = xlen;
	if ((rc = extBalloc(ip, hint ? hint : INOHINT(ip), &nxlen, &nxaddr))) {
		mutex_unlock(&JFS_IP(ip)->commit_mutex);
		return (rc);
	}

	/* Allocate blocks to quota. */
	rc = dquot_alloc_block(ip, nxlen);
	if (rc) {
		dbFree(ip, nxaddr, (s64) nxlen);
		mutex_unlock(&JFS_IP(ip)->commit_mutex);
		return rc;
	}

	/* determine the value of the extent flag */
	xflag = abnr ? XAD_NOTRECORDED : 0;

	/* if we can extend the hint extent to cover the current request,
	 * extend it.  otherwise, insert a new extent to
	 * cover the current request.
	 */
	if (xaddr && xaddr == nxaddr)
		rc = xtExtend(0, ip, xoff, (int) nxlen, 0);
	else
		rc = xtInsert(0, ip, xflag, xoff, (int) nxlen, &nxaddr, 0);

	/* if the extend or insert failed,
	 * free the newly allocated blocks and return the error.
	 */
	if (rc) {
		dbFree(ip, nxaddr, nxlen);
		dquot_free_block(ip, nxlen);
		mutex_unlock(&JFS_IP(ip)->commit_mutex);
		return (rc);
	}

	/* set the results of the extent allocation */
	XADaddress(xp, nxaddr);
	XADlength(xp, nxlen);
	XADoffset(xp, xoff);
	xp->flag = xflag;

	mark_inode_dirty(ip);

	mutex_unlock(&JFS_IP(ip)->commit_mutex);
	/*
	 * COMMIT_SyncList flags an anonymous tlock on page that is on
	 * sync list.
	 * We need to commit the inode to get the page written disk.
	 */
	if (test_and_clear_cflag(COMMIT_Synclist,ip))
		jfs_commit_inode(ip, 0);

	return (0);
}


#ifdef _NOTYET
/*
 * NAME:	extRealloc()
 *
 * FUNCTION:	extend the allocation of a file extent containing a
 *		partial back last page.
 *
 * PARAMETERS:
 *	ip	- the inode of the file.
 *	cp	- cbuf for the partial backed last page.
 *	xlen	- request size of the resulting extent.
 *	xp	- pointer to an xad. on successful exit, the xad
 *		  describes the newly allocated extent.
 *	abnr	- bool indicating whether the newly allocated extent
 *		  should be marked as allocated but not recorded.
 *
 * RETURN VALUES:
 *	0	- success
 *	-EIO	- i/o error.
 *	-ENOSPC	- insufficient disk resources.
 */
int extRealloc(struct inode *ip, s64 nxlen, xad_t * xp, bool abnr)
{
	struct super_block *sb = ip->i_sb;
	s64 xaddr, xlen, nxaddr, delta, xoff;
	s64 ntail, nextend, ninsert;
	int rc, nbperpage = JFS_SBI(sb)->nbperpage;
	int xflag;

	/* This blocks if we are low on resources */
	txBeginAnon(ip->i_sb);

	mutex_lock(&JFS_IP(ip)->commit_mutex);
	/* validate extent length */
	if (nxlen > MAXXLEN)
		nxlen = MAXXLEN;

	/* get the extend (partial) page's disk block address and
	 * number of blocks.
	 */
	xaddr = addressXAD(xp);
	xlen = lengthXAD(xp);
	xoff = offsetXAD(xp);

	/* if the extend page is abnr and if the request is for
	 * the extent to be allocated and recorded,
	 * make the page allocated and recorded.
	 */
	if ((xp->flag & XAD_NOTRECORDED) && !abnr) {
		xp->flag = 0;
		if ((rc = xtUpdate(0, ip, xp)))
			goto exit;
	}

	/* try to allocated the request number of blocks for the
	 * extent.  dbRealloc() first tries to satisfy the request
	 * by extending the allocation in place. otherwise, it will
	 * try to allocate a new set of blocks large enough for the
	 * request.  in satisfying a request, dbReAlloc() may allocate
	 * less than what was request but will always allocate enough
	 * space as to satisfy the extend page.
	 */
	if ((rc = extBrealloc(ip, xaddr, xlen, &nxlen, &nxaddr)))
		goto exit;

	/* Allocat blocks to quota. */
	rc = dquot_alloc_block(ip, nxlen);
	if (rc) {
		dbFree(ip, nxaddr, (s64) nxlen);
		mutex_unlock(&JFS_IP(ip)->commit_mutex);
		return rc;
	}

	delta = nxlen - xlen;

	/* check if the extend page is not abnr but the request is abnr
	 * and the allocated disk space is for more than one page.  if this
	 * is the case, there is a miss match of abnr between the extend page
	 * and the one or more pages following the extend page.  as a result,
	 * two extents will have to be manipulated. the first will be that
	 * of the extent of the extend page and will be manipulated thru
	 * an xtExtend() or an xtTailgate(), depending upon whether the
	 * disk allocation occurred as an inplace extension.  the second
	 * extent will be manipulated (created) through an xtInsert() and
	 * will be for the pages following the extend page.
	 */
	if (abnr && (!(xp->flag & XAD_NOTRECORDED)) && (nxlen > nbperpage)) {
		ntail = nbperpage;
		nextend = ntail - xlen;
		ninsert = nxlen - nbperpage;

		xflag = XAD_NOTRECORDED;
	} else {
		ntail = nxlen;
		nextend = delta;
		ninsert = 0;

		xflag = xp->flag;
	}

	/* if we were able to extend the disk allocation in place,
	 * extend the extent.  otherwise, move the extent to a
	 * new disk location.
	 */
	if (xaddr == nxaddr) {
		/* extend the extent */
		if ((rc = xtExtend(0, ip, xoff + xlen, (int) nextend, 0))) {
			dbFree(ip, xaddr + xlen, delta);
			dquot_free_block(ip, nxlen);
			goto exit;
		}
	} else {
		/*
		 * move the extent to a new location:
		 *
		 * xtTailgate() accounts for relocated tail extent;
		 */
		if ((rc = xtTailgate(0, ip, xoff, (int) ntail, nxaddr, 0))) {
			dbFree(ip, nxaddr, nxlen);
			dquot_free_block(ip, nxlen);
			goto exit;
		}
	}


	/* check if we need to also insert a new extent */
	if (ninsert) {
		/* perform the insert.  if it fails, free the blocks
		 * to be inserted and make it appear that we only did
		 * the xtExtend() or xtTailgate() above.
		 */
		xaddr = nxaddr + ntail;
		if (xtInsert (0, ip, xflag, xoff + ntail, (int) ninsert,
			      &xaddr, 0)) {
			dbFree(ip, xaddr, (s64) ninsert);
			delta = nextend;
			nxlen = ntail;
			xflag = 0;
		}
	}

	/* set the return results */
	XADaddress(xp, nxaddr);
	XADlength(xp, nxlen);
	XADoffset(xp, xoff);
	xp->flag = xflag;

	mark_inode_dirty(ip);
exit:
	mutex_unlock(&JFS_IP(ip)->commit_mutex);
	return (rc);
}
#endif			/* _NOTYET */


/*
 * NAME:	extHint()
 *
 * FUNCTION:	produce an extent allocation hint for a file offset.
 *
 * PARAMETERS:
 *	ip	- the inode of the file.
 *	offset  - file offset for which the hint is needed.
 *	xp	- pointer to the xad that is to be filled in with
 *		  the hint.
 *
 * RETURN VALUES:
 *	0	- success
 *	-EIO	- i/o error.
 */
int extHint(struct inode *ip, s64 offset, xad_t * xp)
{
	struct super_block *sb = ip->i_sb;
	int nbperpage = JFS_SBI(sb)->nbperpage;
	s64 prev;
	int rc = 0;
	s64 xaddr;
	int xlen;
	int xflag;

	/* init the hint as "no hint provided" */
	XADaddress(xp, 0);

	/* determine the starting extent offset of the page previous
	 * to the page containing the offset.
	 */
	prev = ((offset & ~POFFSET) >> JFS_SBI(sb)->l2bsize) - nbperpage;

	/* if the offset is in the first page of the file, no hint provided.
	 */
	if (prev < 0)
		goto out;

	rc = xtLookup(ip, prev, nbperpage, &xflag, &xaddr, &xlen, 0);

	if ((rc == 0) && xlen) {
		if (xlen != nbperpage) {
			jfs_error(ip->i_sb, "corrupt xtree\n");
			rc = -EIO;
		}
		XADaddress(xp, xaddr);
		XADlength(xp, xlen);
		XADoffset(xp, prev);
		/*
		 * only preserve the abnr flag within the xad flags
		 * of the returned hint.
		 */
		xp->flag  = xflag & XAD_NOTRECORDED;
	} else
		rc = 0;

out:
	return (rc);
}


/*
 * NAME:	extRecord()
 *
 * FUNCTION:	change a page with a file from not recorded to recorded.
 *
 * PARAMETERS:
 *	ip	- inode of the file.
 *	cp	- cbuf of the file page.
 *
 * RETURN VALUES:
 *	0	- success
 *	-EIO	- i/o error.
 *	-ENOSPC	- insufficient disk resources.
 */
int extRecord(struct inode *ip, xad_t * xp)
{
	int rc;

	txBeginAnon(ip->i_sb);

	mutex_lock(&JFS_IP(ip)->commit_mutex);

	/* update the extent */
	rc = xtUpdate(0, ip, xp);

	mutex_unlock(&JFS_IP(ip)->commit_mutex);
	return rc;
}


#ifdef _NOTYET
/*
 * NAME:	extFill()
 *
 * FUNCTION:	allocate disk space for a file page that represents
 *		a file hole.
 *
 * PARAMETERS:
 *	ip	- the inode of the file.
 *	cp	- cbuf of the file page represent the hole.
 *
 * RETURN VALUES:
 *	0	- success
 *	-EIO	- i/o error.
 *	-ENOSPC	- insufficient disk resources.
 */
int extFill(struct inode *ip, xad_t * xp)
{
	int rc, nbperpage = JFS_SBI(ip->i_sb)->nbperpage;
	s64 blkno = offsetXAD(xp) >> ip->i_blkbits;

//	assert(ISSPARSE(ip));

	/* initialize the extent allocation hint */
	XADaddress(xp, 0);

	/* allocate an extent to fill the hole */
	if ((rc = extAlloc(ip, nbperpage, blkno, xp, false)))
		return (rc);

	assert(lengthPXD(xp) == nbperpage);

	return (0);
}
#endif			/* _NOTYET */


/*
 * NAME:	extBalloc()
 *
 * FUNCTION:	allocate disk blocks to form an extent.
 *
 *		initially, we will try to allocate disk blocks for the
 *		requested size (nblocks).  if this fails (nblocks
 *		contiguous free blocks not available), we'll try to allocate
 *		a smaller number of blocks (producing a smaller extent), with
 *		this smaller number of blocks consisting of the requested
 *		number of blocks rounded down to the next smaller power of 2
 *		number (i.e. 16 -> 8).  we'll continue to round down and
 *		retry the allocation until the number of blocks to allocate
 *		is smaller than the number of blocks per page.
 *
 * PARAMETERS:
 *	ip	 - the inode of the file.
 *	hint	 - disk block number to be used as an allocation hint.
 *	*nblocks - pointer to an s64 value.  on entry, this value specifies
 *		   the desired number of block to be allocated. on successful
 *		   exit, this value is set to the number of blocks actually
 *		   allocated.
 *	blkno	 - pointer to a block address that is filled in on successful
 *		   return with the starting block number of the newly
 *		   allocated block range.
 *
 * RETURN VALUES:
 *	0	- success
 *	-EIO	- i/o error.
 *	-ENOSPC	- insufficient disk resources.
 */
static int
extBalloc(struct inode *ip, s64 hint, s64 * nblocks, s64 * blkno)
{
	struct jfs_inode_info *ji = JFS_IP(ip);
	struct jfs_sb_info *sbi = JFS_SBI(ip->i_sb);
	s64 nb, nblks, daddr, max;
	int rc, nbperpage = sbi->nbperpage;
	struct bmap *bmp = sbi->bmap;
	int ag;

	/* get the number of blocks to initially attempt to allocate.
	 * we'll first try the number of blocks requested unless this
	 * number is greater than the maximum number of contiguous free
	 * blocks in the map. in that case, we'll start off with the
	 * maximum free.
	 */
	max = (s64) 1 << bmp->db_maxfreebud;
	if (*nblocks >= max && *nblocks > nbperpage)
		nb = nblks = (max > nbperpage) ? max : nbperpage;
	else
		nb = nblks = *nblocks;

	/* try to allocate blocks */
	while ((rc = dbAlloc(ip, hint, nb, &daddr)) != 0) {
		/* if something other than an out of space error,
		 * stop and return this error.
		 */
		if (rc != -ENOSPC)
			return (rc);

		/* decrease the allocation request size */
		nb = min(nblks, extRoundDown(nb));

		/* give up if we cannot cover a page */
		if (nb < nbperpage)
			return (rc);
	}

	*nblocks = nb;
	*blkno = daddr;

	if (S_ISREG(ip->i_mode) && (ji->fileset == FILESYSTEM_I)) {
		ag = BLKTOAG(daddr, sbi);
		spin_lock_irq(&ji->ag_lock);
		if (ji->active_ag == -1) {
			atomic_inc(&bmp->db_active[ag]);
			ji->active_ag = ag;
		} else if (ji->active_ag != ag) {
			atomic_dec(&bmp->db_active[ji->active_ag]);
			atomic_inc(&bmp->db_active[ag]);
			ji->active_ag = ag;
		}
		spin_unlock_irq(&ji->ag_lock);
	}

	return (0);
}


#ifdef _NOTYET
/*
 * NAME:	extBrealloc()
 *
 * FUNCTION:	attempt to extend an extent's allocation.
 *
 *		Initially, we will try to extend the extent's allocation
 *		in place.  If this fails, we'll try to move the extent
 *		to a new set of blocks.  If moving the extent, we initially
 *		will try to allocate disk blocks for the requested size
 *		(newnblks).  if this fails (new contiguous free blocks not
 *		available), we'll try to allocate a smaller number of
 *		blocks (producing a smaller extent), with this smaller
 *		number of blocks consisting of the requested number of
 *		blocks rounded down to the next smaller power of 2
 *		number (i.e. 16 -> 8).  We'll continue to round down and
 *		retry the allocation until the number of blocks to allocate
 *		is smaller than the number of blocks per page.
 *
 * PARAMETERS:
 *	ip	 - the inode of the file.
 *	blkno	 - starting block number of the extents current allocation.
 *	nblks	 - number of blocks within the extents current allocation.
 *	newnblks - pointer to a s64 value.  on entry, this value is the
 *		   the new desired extent size (number of blocks).  on
 *		   successful exit, this value is set to the extent's actual
 *		   new size (new number of blocks).
 *	newblkno - the starting block number of the extents new allocation.
 *
 * RETURN VALUES:
 *	0	- success
 *	-EIO	- i/o error.
 *	-ENOSPC	- insufficient disk resources.
 */
static int
extBrealloc(struct inode *ip,
	    s64 blkno, s64 nblks, s64 * newnblks, s64 * newblkno)
{
	int rc;

	/* try to extend in place */
	if ((rc = dbExtend(ip, blkno, nblks, *newnblks - nblks)) == 0) {
		*newblkno = blkno;
		return (0);
	} else {
		if (rc != -ENOSPC)
			return (rc);
	}

	/* in place extension not possible.
	 * try to move the extent to a new set of blocks.
	 */
	return (extBalloc(ip, blkno, newnblks, newblkno));
}
#endif			/* _NOTYET */


/*
 * NAME:	extRoundDown()
 *
 * FUNCTION:	round down a specified number of blocks to the next
 *		smallest power of 2 number.
 *
 * PARAMETERS:
 *	nb	- the inode of the file.
 *
 * RETURN VALUES:
 *	next smallest power of 2 number.
 */
static s64 extRoundDown(s64 nb)
{
	int i;
	u64 m, k;

	for (i = 0, m = (u64) 1 << 63; i < 64; i++, m >>= 1) {
		if (m & nb)
			break;
	}

	i = 63 - i;
	k = (u64) 1 << i;
	k = ((k - 1) & nb) ? k : k >> 1;

	return (k);
}
OpenPOWER on IntegriCloud