xfs_iomap.c 33.7 KB
Newer Older
D
Dave Chinner 已提交
1
// SPDX-License-Identifier: GPL-2.0
L
Linus Torvalds 已提交
2
/*
3
 * Copyright (c) 2000-2006 Silicon Graphics, Inc.
4
 * Copyright (c) 2016 Christoph Hellwig.
5
 * All Rights Reserved.
L
Linus Torvalds 已提交
6
 */
7
#include <linux/iomap.h>
L
Linus Torvalds 已提交
8 9
#include "xfs.h"
#include "xfs_fs.h"
10
#include "xfs_shared.h"
11 12 13
#include "xfs_format.h"
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
L
Linus Torvalds 已提交
14
#include "xfs_mount.h"
15
#include "xfs_defer.h"
L
Linus Torvalds 已提交
16
#include "xfs_inode.h"
17
#include "xfs_btree.h"
18
#include "xfs_bmap_btree.h"
L
Linus Torvalds 已提交
19
#include "xfs_bmap.h"
D
Dave Chinner 已提交
20
#include "xfs_bmap_util.h"
21
#include "xfs_errortag.h"
L
Linus Torvalds 已提交
22
#include "xfs_error.h"
23
#include "xfs_trans.h"
L
Linus Torvalds 已提交
24
#include "xfs_trans_space.h"
25
#include "xfs_inode_item.h"
L
Linus Torvalds 已提交
26
#include "xfs_iomap.h"
C
Christoph Hellwig 已提交
27
#include "xfs_trace.h"
28
#include "xfs_icache.h"
29
#include "xfs_quota.h"
30 31
#include "xfs_dquot_item.h"
#include "xfs_dquot.h"
32
#include "xfs_reflink.h"
L
Linus Torvalds 已提交
33 34 35 36 37


#define XFS_WRITEIO_ALIGN(mp,off)	(((off) >> mp->m_writeio_log) \
						<< mp->m_writeio_log)

38 39 40 41 42 43 44 45 46
void
xfs_bmbt_to_iomap(
	struct xfs_inode	*ip,
	struct iomap		*iomap,
	struct xfs_bmbt_irec	*imap)
{
	struct xfs_mount	*mp = ip->i_mount;

	if (imap->br_startblock == HOLESTARTBLOCK) {
47
		iomap->addr = IOMAP_NULL_ADDR;
48 49
		iomap->type = IOMAP_HOLE;
	} else if (imap->br_startblock == DELAYSTARTBLOCK) {
50
		iomap->addr = IOMAP_NULL_ADDR;
51 52
		iomap->type = IOMAP_DELALLOC;
	} else {
53
		iomap->addr = BBTOB(xfs_fsb_to_db(ip, imap->br_startblock));
54 55 56 57 58 59 60 61
		if (imap->br_state == XFS_EXT_UNWRITTEN)
			iomap->type = IOMAP_UNWRITTEN;
		else
			iomap->type = IOMAP_MAPPED;
	}
	iomap->offset = XFS_FSB_TO_B(mp, imap->br_startoff);
	iomap->length = XFS_FSB_TO_B(mp, imap->br_blockcount);
	iomap->bdev = xfs_find_bdev_for_inode(VFS_I(ip));
62
	iomap->dax_dev = xfs_find_daxdev_for_inode(VFS_I(ip));
63 64
}

65
xfs_extlen_t
66 67 68
xfs_eof_alignment(
	struct xfs_inode	*ip,
	xfs_extlen_t		extsize)
69
{
70 71
	struct xfs_mount	*mp = ip->i_mount;
	xfs_extlen_t		align = 0;
72

73 74 75 76 77 78 79 80 81 82 83 84 85 86
	if (!XFS_IS_REALTIME_INODE(ip)) {
		/*
		 * Round up the allocation request to a stripe unit
		 * (m_dalign) boundary if the file size is >= stripe unit
		 * size, and we are allocating past the allocation eof.
		 *
		 * If mounted with the "-o swalloc" option the alignment is
		 * increased from the strip unit size to the stripe width.
		 */
		if (mp->m_swidth && (mp->m_flags & XFS_MOUNT_SWALLOC))
			align = mp->m_swidth;
		else if (mp->m_dalign)
			align = mp->m_dalign;

87 88
		if (align && XFS_ISIZE(ip) < XFS_FSB_TO_B(mp, align))
			align = 0;
89
	}
90 91 92 93 94 95

	/*
	 * Always round up the allocation request to an extent boundary
	 * (when file on a real-time subvolume or has di_extsize hint).
	 */
	if (extsize) {
96 97
		if (align)
			align = roundup_64(align, extsize);
98 99 100 101
		else
			align = extsize;
	}

102 103 104 105 106 107 108 109 110 111 112
	return align;
}

STATIC int
xfs_iomap_eof_align_last_fsb(
	struct xfs_inode	*ip,
	xfs_extlen_t		extsize,
	xfs_fileoff_t		*last_fsb)
{
	xfs_extlen_t		align = xfs_eof_alignment(ip, extsize);

113 114
	if (align) {
		xfs_fileoff_t	new_last_fsb = roundup_64(*last_fsb, align);
115 116
		int		eof, error;

117
		error = xfs_bmap_eof(ip, new_last_fsb, XFS_DATA_FORK, &eof);
118 119 120 121 122 123 124 125
		if (error)
			return error;
		if (eof)
			*last_fsb = new_last_fsb;
	}
	return 0;
}

126
STATIC int
127
xfs_alert_fsblock_zero(
128 129 130
	xfs_inode_t	*ip,
	xfs_bmbt_irec_t	*imap)
{
131
	xfs_alert_tag(ip->i_mount, XFS_PTAG_FSBLOCK_ZERO,
132 133
			"Access to block zero in inode %llu "
			"start_block: %llx start_off: %llx "
134
			"blkcnt: %llx extent-state: %x",
135 136 137 138 139
		(unsigned long long)ip->i_ino,
		(unsigned long long)imap->br_startblock,
		(unsigned long long)imap->br_startoff,
		(unsigned long long)imap->br_blockcount,
		imap->br_state);
D
Dave Chinner 已提交
140
	return -EFSCORRUPTED;
141 142
}

C
Christoph Hellwig 已提交
143
int
L
Linus Torvalds 已提交
144 145
xfs_iomap_write_direct(
	xfs_inode_t	*ip,
146
	xfs_off_t	offset,
L
Linus Torvalds 已提交
147
	size_t		count,
148
	xfs_bmbt_irec_t *imap,
149
	int		nmaps)
L
Linus Torvalds 已提交
150 151 152 153
{
	xfs_mount_t	*mp = ip->i_mount;
	xfs_fileoff_t	offset_fsb;
	xfs_fileoff_t	last_fsb;
154
	xfs_filblks_t	count_fsb, resaligned;
L
Linus Torvalds 已提交
155
	xfs_fsblock_t	firstfsb;
156
	xfs_extlen_t	extsz;
157
	int		nimaps;
158
	int		quota_flag;
L
Linus Torvalds 已提交
159 160
	int		rt;
	xfs_trans_t	*tp;
161
	struct xfs_defer_ops dfops;
162 163
	uint		qblocks, resblks, resrtextents;
	int		error;
164
	int		lockmode;
165
	int		bmapi_flags = XFS_BMAPI_PREALLOC;
166
	uint		tflags = 0;
L
Linus Torvalds 已提交
167

168
	rt = XFS_IS_REALTIME_INODE(ip);
169
	extsz = xfs_get_extsz_hint(ip);
170 171 172
	lockmode = XFS_ILOCK_SHARED;	/* locked by caller */

	ASSERT(xfs_isilocked(ip, lockmode));
L
Linus Torvalds 已提交
173

174 175
	offset_fsb = XFS_B_TO_FSBT(mp, offset);
	last_fsb = XFS_B_TO_FSB(mp, ((xfs_ufsize_t)(offset + count)));
176
	if ((offset + count) > XFS_ISIZE(ip)) {
177 178 179 180 181 182 183 184 185
		/*
		 * Assert that the in-core extent list is present since this can
		 * call xfs_iread_extents() and we only have the ilock shared.
		 * This should be safe because the lock was held around a bmapi
		 * call in the caller and we only need it to access the in-core
		 * list.
		 */
		ASSERT(XFS_IFORK_PTR(ip, XFS_DATA_FORK)->if_flags &
								XFS_IFEXTENTS);
186
		error = xfs_iomap_eof_align_last_fsb(ip, extsz, &last_fsb);
187
		if (error)
188
			goto out_unlock;
L
Linus Torvalds 已提交
189
	} else {
190
		if (nmaps && (imap->br_startblock == HOLESTARTBLOCK))
D
Dave Chinner 已提交
191
			last_fsb = min(last_fsb, (xfs_fileoff_t)
192 193
					imap->br_blockcount +
					imap->br_startoff);
L
Linus Torvalds 已提交
194
	}
195 196
	count_fsb = last_fsb - offset_fsb;
	ASSERT(count_fsb > 0);
197
	resaligned = xfs_aligned_fsb_count(offset_fsb, count_fsb, extsz);
198 199 200 201

	if (unlikely(rt)) {
		resrtextents = qblocks = resaligned;
		resrtextents /= mp->m_sb.sb_rextsize;
202 203 204 205
		resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0);
		quota_flag = XFS_QMOPT_RES_RTBLKS;
	} else {
		resrtextents = 0;
206
		resblks = qblocks = XFS_DIOSTRAT_SPACE_RES(mp, resaligned);
207 208
		quota_flag = XFS_QMOPT_RES_REGBLKS;
	}
L
Linus Torvalds 已提交
209

210 211 212 213 214
	/*
	 * Drop the shared lock acquired by the caller, attach the dquot if
	 * necessary and move on to transaction setup.
	 */
	xfs_iunlock(ip, lockmode);
215
	error = xfs_qm_dqattach(ip);
216 217 218
	if (error)
		return error;

219 220 221 222 223 224
	/*
	 * For DAX, we do not allocate unwritten extents, but instead we zero
	 * the block before we commit the transaction.  Ideally we'd like to do
	 * this outside the transaction context, but if we commit and then crash
	 * we may not have zeroed the blocks and this will be exposed on
	 * recovery of the allocation. Hence we must zero before commit.
225
	 *
226 227 228
	 * Further, if we are mapping unwritten extents here, we need to zero
	 * and convert them to written so that we don't need an unwritten extent
	 * callback for DAX. This also means that we need to be able to dip into
229 230
	 * the reserve block pool for bmbt block allocation if there is no space
	 * left but we need to do unwritten extent conversion.
231 232 233
	 */
	if (IS_DAX(VFS_I(ip))) {
		bmapi_flags = XFS_BMAPI_CONVERT | XFS_BMAPI_ZERO;
234
		if (imap->br_state == XFS_EXT_UNWRITTEN) {
235
			tflags |= XFS_TRANS_RESERVE;
236 237
			resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0) << 1;
		}
238
	}
239 240 241
	error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, resblks, resrtextents,
			tflags, &tp);
	if (error)
E
Eric Sandeen 已提交
242
		return error;
243

244 245
	lockmode = XFS_ILOCK_EXCL;
	xfs_ilock(ip, lockmode);
L
Linus Torvalds 已提交
246

C
Christoph Hellwig 已提交
247
	error = xfs_trans_reserve_quota_nblks(tp, ip, qblocks, 0, quota_flag);
248
	if (error)
249
		goto out_trans_cancel;
L
Linus Torvalds 已提交
250

251
	xfs_trans_ijoin(tp, ip, 0);
L
Linus Torvalds 已提交
252 253

	/*
254 255
	 * From this point onwards we overwrite the imap pointer that the
	 * caller gave to us.
L
Linus Torvalds 已提交
256
	 */
257
	xfs_defer_init(&dfops, &firstfsb);
258
	nimaps = 1;
259
	error = xfs_bmapi_write(tp, ip, offset_fsb, count_fsb,
260
				bmapi_flags, &firstfsb, resblks, imap,
261
				&nimaps, &dfops);
262
	if (error)
263
		goto out_bmap_cancel;
L
Linus Torvalds 已提交
264 265

	/*
266
	 * Complete the transaction
L
Linus Torvalds 已提交
267
	 */
268
	error = xfs_defer_finish(&tp, &dfops);
269
	if (error)
270
		goto out_bmap_cancel;
271

272
	error = xfs_trans_commit(tp);
273
	if (error)
274
		goto out_unlock;
L
Linus Torvalds 已提交
275

276 277 278
	/*
	 * Copy any maps to caller's array and return any error.
	 */
L
Linus Torvalds 已提交
279
	if (nimaps == 0) {
D
Dave Chinner 已提交
280
		error = -ENOSPC;
281
		goto out_unlock;
282 283
	}

284
	if (!(imap->br_startblock || XFS_IS_REALTIME_INODE(ip)))
285
		error = xfs_alert_fsblock_zero(ip, imap);
L
Linus Torvalds 已提交
286

287
out_unlock:
288
	xfs_iunlock(ip, lockmode);
289
	return error;
L
Linus Torvalds 已提交
290

291
out_bmap_cancel:
292
	xfs_defer_cancel(&dfops);
293
	xfs_trans_unreserve_quota_nblks(tp, ip, (long)qblocks, 0, quota_flag);
294
out_trans_cancel:
295
	xfs_trans_cancel(tp);
296
	goto out_unlock;
L
Linus Torvalds 已提交
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
STATIC bool
xfs_quota_need_throttle(
	struct xfs_inode *ip,
	int type,
	xfs_fsblock_t alloc_blocks)
{
	struct xfs_dquot *dq = xfs_inode_dquot(ip, type);

	if (!dq || !xfs_this_quota_on(ip->i_mount, type))
		return false;

	/* no hi watermark, no throttle */
	if (!dq->q_prealloc_hi_wmark)
		return false;

	/* under the lo watermark, no throttle */
	if (dq->q_res_bcount + alloc_blocks < dq->q_prealloc_lo_wmark)
		return false;

	return true;
}

STATIC void
xfs_quota_calc_throttle(
	struct xfs_inode *ip,
	int type,
	xfs_fsblock_t *qblocks,
326 327
	int *qshift,
	int64_t	*qfreesp)
328 329 330 331 332
{
	int64_t freesp;
	int shift = 0;
	struct xfs_dquot *dq = xfs_inode_dquot(ip, type);

333 334
	/* no dq, or over hi wmark, squash the prealloc completely */
	if (!dq || dq->q_res_bcount >= dq->q_prealloc_hi_wmark) {
335
		*qblocks = 0;
336
		*qfreesp = 0;
337 338 339 340 341 342 343 344 345 346 347 348
		return;
	}

	freesp = dq->q_prealloc_hi_wmark - dq->q_res_bcount;
	if (freesp < dq->q_low_space[XFS_QLOWSP_5_PCNT]) {
		shift = 2;
		if (freesp < dq->q_low_space[XFS_QLOWSP_3_PCNT])
			shift += 2;
		if (freesp < dq->q_low_space[XFS_QLOWSP_1_PCNT])
			shift += 2;
	}

349 350 351
	if (freesp < *qfreesp)
		*qfreesp = freesp;

352 353 354 355 356 357 358
	/* only overwrite the throttle values if we are more aggressive */
	if ((freesp >> shift) < (*qblocks >> *qshift)) {
		*qblocks = freesp;
		*qshift = shift;
	}
}

359
/*
360 361 362 363
 * If we are doing a write at the end of the file and there are no allocations
 * past this one, then extend the allocation out to the file system's write
 * iosize.
 *
364
 * If we don't have a user specified preallocation size, dynamically increase
365
 * the preallocation size as the size of the file grows.  Cap the maximum size
366 367
 * at a single extent or less if the filesystem is near full. The closer the
 * filesystem is to full, the smaller the maximum prealocation.
368 369 370 371 372 373 374 375
 *
 * As an exception we don't do any preallocation at all if the file is smaller
 * than the minimum preallocation and we are using the default dynamic
 * preallocation scheme, as it is likely this is the only write to the file that
 * is going to be done.
 *
 * We clean up any extra space left over when the file is closed in
 * xfs_inactive().
376 377 378
 */
STATIC xfs_fsblock_t
xfs_iomap_prealloc_size(
379
	struct xfs_inode	*ip,
380 381
	loff_t			offset,
	loff_t			count,
382
	struct xfs_iext_cursor	*icur)
383
{
384
	struct xfs_mount	*mp = ip->i_mount;
385
	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
386
	xfs_fileoff_t		offset_fsb = XFS_B_TO_FSBT(mp, offset);
387
	struct xfs_bmbt_irec	prev;
388 389
	int			shift = 0;
	int64_t			freesp;
390 391
	xfs_fsblock_t		qblocks;
	int			qshift = 0;
392 393 394 395 396 397 398 399 400 401 402 403 404 405 406
	xfs_fsblock_t		alloc_blocks = 0;

	if (offset + count <= XFS_ISIZE(ip))
		return 0;

	if (!(mp->m_flags & XFS_MOUNT_DFLT_IOSIZE) &&
	    (XFS_ISIZE(ip) < XFS_FSB_TO_B(mp, mp->m_writeio_blocks)))
		return 0;

	/*
	 * If an explicit allocsize is set, the file is small, or we
	 * are writing behind a hole, then use the minimum prealloc:
	 */
	if ((mp->m_flags & XFS_MOUNT_DFLT_IOSIZE) ||
	    XFS_ISIZE(ip) < XFS_FSB_TO_B(mp, mp->m_dalign) ||
407
	    !xfs_iext_peek_prev_extent(ifp, icur, &prev) ||
408
	    prev.br_startoff + prev.br_blockcount < offset_fsb)
409
		return mp->m_writeio_blocks;
410

411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426
	/*
	 * Determine the initial size of the preallocation. We are beyond the
	 * current EOF here, but we need to take into account whether this is
	 * a sparse write or an extending write when determining the
	 * preallocation size.  Hence we need to look up the extent that ends
	 * at the current write offset and use the result to determine the
	 * preallocation size.
	 *
	 * If the extent is a hole, then preallocation is essentially disabled.
	 * Otherwise we take the size of the preceding data extent as the basis
	 * for the preallocation size. If the size of the extent is greater than
	 * half the maximum extent length, then use the current offset as the
	 * basis. This ensures that for large files the preallocation size
	 * always extends to MAXEXTLEN rather than falling short due to things
	 * like stripe unit/width alignment of real extents.
	 */
427 428
	if (prev.br_blockcount <= (MAXEXTLEN >> 1))
		alloc_blocks = prev.br_blockcount << 1;
429 430
	else
		alloc_blocks = XFS_B_TO_FSB(mp, offset);
431 432
	if (!alloc_blocks)
		goto check_writeio;
433
	qblocks = alloc_blocks;
434

435 436 437 438 439 440 441 442 443
	/*
	 * MAXEXTLEN is not a power of two value but we round the prealloc down
	 * to the nearest power of two value after throttling. To prevent the
	 * round down from unconditionally reducing the maximum supported prealloc
	 * size, we round up first, apply appropriate throttling, round down and
	 * cap the value to MAXEXTLEN.
	 */
	alloc_blocks = XFS_FILEOFF_MIN(roundup_pow_of_two(MAXEXTLEN),
				       alloc_blocks);
444

445
	freesp = percpu_counter_read_positive(&mp->m_fdblocks);
446 447 448 449 450 451 452 453 454 455
	if (freesp < mp->m_low_space[XFS_LOWSP_5_PCNT]) {
		shift = 2;
		if (freesp < mp->m_low_space[XFS_LOWSP_4_PCNT])
			shift++;
		if (freesp < mp->m_low_space[XFS_LOWSP_3_PCNT])
			shift++;
		if (freesp < mp->m_low_space[XFS_LOWSP_2_PCNT])
			shift++;
		if (freesp < mp->m_low_space[XFS_LOWSP_1_PCNT])
			shift++;
456
	}
457 458

	/*
459 460
	 * Check each quota to cap the prealloc size, provide a shift value to
	 * throttle with and adjust amount of available space.
461 462
	 */
	if (xfs_quota_need_throttle(ip, XFS_DQ_USER, alloc_blocks))
463 464
		xfs_quota_calc_throttle(ip, XFS_DQ_USER, &qblocks, &qshift,
					&freesp);
465
	if (xfs_quota_need_throttle(ip, XFS_DQ_GROUP, alloc_blocks))
466 467
		xfs_quota_calc_throttle(ip, XFS_DQ_GROUP, &qblocks, &qshift,
					&freesp);
468
	if (xfs_quota_need_throttle(ip, XFS_DQ_PROJ, alloc_blocks))
469 470
		xfs_quota_calc_throttle(ip, XFS_DQ_PROJ, &qblocks, &qshift,
					&freesp);
471 472 473 474 475 476 477 478

	/*
	 * The final prealloc size is set to the minimum of free space available
	 * in each of the quotas and the overall filesystem.
	 *
	 * The shift throttle value is set to the maximum value as determined by
	 * the global low free space values and per-quota low free space values.
	 */
D
Dave Chinner 已提交
479 480
	alloc_blocks = min(alloc_blocks, qblocks);
	shift = max(shift, qshift);
481

482 483
	if (shift)
		alloc_blocks >>= shift;
484 485 486 487 488 489 490 491
	/*
	 * rounddown_pow_of_two() returns an undefined result if we pass in
	 * alloc_blocks = 0.
	 */
	if (alloc_blocks)
		alloc_blocks = rounddown_pow_of_two(alloc_blocks);
	if (alloc_blocks > MAXEXTLEN)
		alloc_blocks = MAXEXTLEN;
492 493 494 495 496 497 498 499 500 501

	/*
	 * If we are still trying to allocate more space than is
	 * available, squash the prealloc hard. This can happen if we
	 * have a large file on a small filesystem and the above
	 * lowspace thresholds are smaller than MAXEXTLEN.
	 */
	while (alloc_blocks && alloc_blocks >= freesp)
		alloc_blocks >>= 4;
check_writeio:
502 503
	if (alloc_blocks < mp->m_writeio_blocks)
		alloc_blocks = mp->m_writeio_blocks;
504 505
	trace_xfs_iomap_prealloc_size(ip, alloc_blocks, shift,
				      mp->m_writeio_blocks);
506 507 508
	return alloc_blocks;
}

509 510 511 512 513 514
static int
xfs_file_iomap_begin_delay(
	struct inode		*inode,
	loff_t			offset,
	loff_t			count,
	struct iomap		*iomap)
L
Linus Torvalds 已提交
515
{
516 517 518 519 520 521
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;
	struct xfs_ifork	*ifp = XFS_IFORK_PTR(ip, XFS_DATA_FORK);
	xfs_fileoff_t		offset_fsb = XFS_B_TO_FSBT(mp, offset);
	xfs_fileoff_t		maxbytes_fsb =
		XFS_B_TO_FSB(mp, mp->m_super->s_maxbytes);
522
	xfs_fileoff_t		end_fsb;
523 524
	int			error = 0, eof = 0;
	struct xfs_bmbt_irec	got;
525
	struct xfs_iext_cursor	icur;
526
	xfs_fsblock_t		prealloc_blocks = 0;
527 528 529

	ASSERT(!XFS_IS_REALTIME_INODE(ip));
	ASSERT(!xfs_get_extsz_hint(ip));
530

531
	xfs_ilock(ip, XFS_ILOCK_EXCL);
L
Linus Torvalds 已提交
532

533 534 535
	if (unlikely(XFS_TEST_ERROR(
	    (XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_EXTENTS &&
	     XFS_IFORK_FORMAT(ip, XFS_DATA_FORK) != XFS_DINODE_FMT_BTREE),
536
	     mp, XFS_ERRTAG_BMAPIFORMAT))) {
537 538 539 540
		XFS_ERROR_REPORT(__func__, XFS_ERRLEVEL_LOW, mp);
		error = -EFSCORRUPTED;
		goto out_unlock;
	}
541

542
	XFS_STATS_INC(mp, xs_blk_mapw);
543

544 545 546 547
	if (!(ifp->if_flags & XFS_IFEXTENTS)) {
		error = xfs_iread_extents(NULL, ip, XFS_DATA_FORK);
		if (error)
			goto out_unlock;
L
Linus Torvalds 已提交
548 549
	}

550
	eof = !xfs_iext_lookup_extent(ip, ifp, offset_fsb, &icur, &got);
551
	if (!eof && got.br_startoff <= offset_fsb) {
552 553 554 555 556 557 558 559 560 561 562
		if (xfs_is_reflink_inode(ip)) {
			bool		shared;

			end_fsb = min(XFS_B_TO_FSB(mp, offset + count),
					maxbytes_fsb);
			xfs_trim_extent(&got, offset_fsb, end_fsb - offset_fsb);
			error = xfs_reflink_reserve_cow(ip, &got, &shared);
			if (error)
				goto out_unlock;
		}

563 564
		trace_xfs_iomap_found(ip, offset, count, 0, &got);
		goto done;
L
Linus Torvalds 已提交
565
	}
566

567
	error = xfs_qm_dqattach_locked(ip, false);
568 569 570
	if (error)
		goto out_unlock;

571
	/*
572 573 574 575 576 577 578
	 * We cap the maximum length we map here to MAX_WRITEBACK_PAGES pages
	 * to keep the chunks of work done where somewhat symmetric with the
	 * work writeback does. This is a completely arbitrary number pulled
	 * out of thin air as a best guess for initial testing.
	 *
	 * Note that the values needs to be less than 32-bits wide until
	 * the lower level functions are updated.
579
	 */
580
	count = min_t(loff_t, count, 1024 * PAGE_SIZE);
581
	end_fsb = min(XFS_B_TO_FSB(mp, offset + count), maxbytes_fsb);
582 583

	if (eof) {
584 585
		prealloc_blocks = xfs_iomap_prealloc_size(ip, offset, count,
				&icur);
586 587 588
		if (prealloc_blocks) {
			xfs_extlen_t	align;
			xfs_off_t	end_offset;
589
			xfs_fileoff_t	p_end_fsb;
590

591
			end_offset = XFS_WRITEIO_ALIGN(mp, offset + count - 1);
592 593
			p_end_fsb = XFS_B_TO_FSBT(mp, end_offset) +
					prealloc_blocks;
594 595 596

			align = xfs_eof_alignment(ip, 0);
			if (align)
597
				p_end_fsb = roundup_64(p_end_fsb, align);
598

599 600 601
			p_end_fsb = min(p_end_fsb, maxbytes_fsb);
			ASSERT(p_end_fsb > offset_fsb);
			prealloc_blocks = p_end_fsb - end_fsb;
602 603 604 605
		}
	}

retry:
606
	error = xfs_bmapi_reserve_delalloc(ip, XFS_DATA_FORK, offset_fsb,
607 608
			end_fsb - offset_fsb, prealloc_blocks, &got, &icur,
			eof);
609 610
	switch (error) {
	case 0:
611
		break;
D
Dave Chinner 已提交
612 613
	case -ENOSPC:
	case -EDQUOT:
614
		/* retry without any preallocation */
C
Christoph Hellwig 已提交
615
		trace_xfs_delalloc_enospc(ip, offset, count);
616 617
		if (prealloc_blocks) {
			prealloc_blocks = 0;
D
Dave Chinner 已提交
618
			goto retry;
619
		}
620 621 622
		/*FALLTHRU*/
	default:
		goto out_unlock;
L
Linus Torvalds 已提交
623 624
	}

625 626 627 628 629
	/*
	 * Flag newly allocated delalloc blocks with IOMAP_F_NEW so we punch
	 * them out if the write happens to fail.
	 */
	iomap->flags = IOMAP_F_NEW;
630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645
	trace_xfs_iomap_alloc(ip, offset, count, 0, &got);
done:
	if (isnullstartblock(got.br_startblock))
		got.br_startblock = DELAYSTARTBLOCK;

	if (!got.br_startblock) {
		error = xfs_alert_fsblock_zero(ip, &got);
		if (error)
			goto out_unlock;
	}

	xfs_bmbt_to_iomap(ip, iomap, &got);

out_unlock:
	xfs_iunlock(ip, XFS_ILOCK_EXCL);
	return error;
L
Linus Torvalds 已提交
646 647 648 649 650 651 652 653
}

/*
 * Pass in a delayed allocate extent, convert it to real extents;
 * return to the caller the extent we create which maps on top of
 * the originating callers request.
 *
 * Called without a lock on the inode.
654 655 656
 *
 * We no longer bother to look at the incoming map - all we have to
 * guarantee is that whatever we allocate fills the required range.
L
Linus Torvalds 已提交
657
 */
C
Christoph Hellwig 已提交
658
int
L
Linus Torvalds 已提交
659 660
xfs_iomap_write_allocate(
	xfs_inode_t	*ip,
661
	int		whichfork,
662
	xfs_off_t	offset,
663
	xfs_bmbt_irec_t *imap)
L
Linus Torvalds 已提交
664 665 666 667 668
{
	xfs_mount_t	*mp = ip->i_mount;
	xfs_fileoff_t	offset_fsb, last_block;
	xfs_fileoff_t	end_fsb, map_start_fsb;
	xfs_fsblock_t	first_block;
669
	struct xfs_defer_ops	dfops;
L
Linus Torvalds 已提交
670 671
	xfs_filblks_t	count_fsb;
	xfs_trans_t	*tp;
672
	int		nimaps;
L
Linus Torvalds 已提交
673
	int		error = 0;
C
Christoph Hellwig 已提交
674
	int		flags = XFS_BMAPI_DELALLOC;
L
Linus Torvalds 已提交
675 676
	int		nres;

677
	if (whichfork == XFS_COW_FORK)
678
		flags |= XFS_BMAPI_COWFORK | XFS_BMAPI_PREALLOC;
679

L
Linus Torvalds 已提交
680 681 682
	/*
	 * Make sure that the dquots are there.
	 */
683
	error = xfs_qm_dqattach(ip);
C
Christoph Hellwig 已提交
684
	if (error)
E
Eric Sandeen 已提交
685
		return error;
L
Linus Torvalds 已提交
686

687
	offset_fsb = XFS_B_TO_FSBT(mp, offset);
688 689
	count_fsb = imap->br_blockcount;
	map_start_fsb = imap->br_startoff;
L
Linus Torvalds 已提交
690

691
	XFS_STATS_ADD(mp, xs_xstrat_bytes, XFS_FSB_TO_B(mp, count_fsb));
L
Linus Torvalds 已提交
692 693 694 695 696 697 698 699 700 701 702 703 704

	while (count_fsb != 0) {
		/*
		 * Set up a transaction with which to allocate the
		 * backing store for the file.  Do allocations in a
		 * loop until we get some space in the range we are
		 * interested in.  The other space that might be allocated
		 * is in the delayed allocation extent on which we sit
		 * but before our buffer starts.
		 */
		nimaps = 0;
		while (nimaps == 0) {
			nres = XFS_EXTENTADD_SPACE_RES(mp, XFS_DATA_FORK);
705 706 707 708 709 710 711
			/*
			 * We have already reserved space for the extent and any
			 * indirect blocks when creating the delalloc extent,
			 * there is no need to reserve space in this transaction
			 * again.
			 */
			error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, 0,
712 713
					0, XFS_TRANS_RESERVE, &tp);
			if (error)
E
Eric Sandeen 已提交
714
				return error;
715

L
Linus Torvalds 已提交
716
			xfs_ilock(ip, XFS_ILOCK_EXCL);
717
			xfs_trans_ijoin(tp, ip, 0);
L
Linus Torvalds 已提交
718

719
			xfs_defer_init(&dfops, &first_block);
L
Linus Torvalds 已提交
720 721

			/*
722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750
			 * it is possible that the extents have changed since
			 * we did the read call as we dropped the ilock for a
			 * while. We have to be careful about truncates or hole
			 * punchs here - we are not allowed to allocate
			 * non-delalloc blocks here.
			 *
			 * The only protection against truncation is the pages
			 * for the range we are being asked to convert are
			 * locked and hence a truncate will block on them
			 * first.
			 *
			 * As a result, if we go beyond the range we really
			 * need and hit an delalloc extent boundary followed by
			 * a hole while we have excess blocks in the map, we
			 * will fill the hole incorrectly and overrun the
			 * transaction reservation.
			 *
			 * Using a single map prevents this as we are forced to
			 * check each map we look for overlap with the desired
			 * range and abort as soon as we find it. Also, given
			 * that we only return a single map, having one beyond
			 * what we can return is probably a bit silly.
			 *
			 * We also need to check that we don't go beyond EOF;
			 * this is a truncate optimisation as a truncate sets
			 * the new file size before block on the pages we
			 * currently have locked under writeback. Because they
			 * are about to be tossed, we don't need to write them
			 * back....
L
Linus Torvalds 已提交
751
			 */
752
			nimaps = 1;
753
			end_fsb = XFS_B_TO_FSB(mp, XFS_ISIZE(ip));
754
			error = xfs_bmap_last_offset(ip, &last_block,
755 756 757 758
							XFS_DATA_FORK);
			if (error)
				goto trans_cancel;

L
Linus Torvalds 已提交
759 760 761 762
			last_block = XFS_FILEOFF_MAX(last_block, end_fsb);
			if ((map_start_fsb + count_fsb) > last_block) {
				count_fsb = last_block - map_start_fsb;
				if (count_fsb == 0) {
D
Dave Chinner 已提交
763
					error = -EAGAIN;
L
Linus Torvalds 已提交
764 765 766 767
					goto trans_cancel;
				}
			}

768 769 770 771
			/*
			 * From this point onwards we overwrite the imap
			 * pointer that the caller gave to us.
			 */
772
			error = xfs_bmapi_write(tp, ip, map_start_fsb,
773
						count_fsb, flags, &first_block,
774
						nres, imap, &nimaps,
775
						&dfops);
L
Linus Torvalds 已提交
776 777 778
			if (error)
				goto trans_cancel;

779
			error = xfs_defer_finish(&tp, &dfops);
L
Linus Torvalds 已提交
780 781 782
			if (error)
				goto trans_cancel;

783
			error = xfs_trans_commit(tp);
L
Linus Torvalds 已提交
784 785 786 787 788 789 790 791 792 793
			if (error)
				goto error0;

			xfs_iunlock(ip, XFS_ILOCK_EXCL);
		}

		/*
		 * See if we were able to allocate an extent that
		 * covers at least part of the callers request
		 */
794
		if (!(imap->br_startblock || XFS_IS_REALTIME_INODE(ip)))
795
			return xfs_alert_fsblock_zero(ip, imap);
796

797 798 799
		if ((offset_fsb >= imap->br_startoff) &&
		    (offset_fsb < (imap->br_startoff +
				   imap->br_blockcount))) {
800
			XFS_STATS_INC(mp, xs_xstrat_quick);
801
			return 0;
L
Linus Torvalds 已提交
802 803
		}

804 805
		/*
		 * So far we have not mapped the requested part of the
L
Linus Torvalds 已提交
806 807
		 * file, just surrounding data, try again.
		 */
808 809
		count_fsb -= imap->br_blockcount;
		map_start_fsb = imap->br_startoff + imap->br_blockcount;
L
Linus Torvalds 已提交
810 811 812
	}

trans_cancel:
813
	xfs_defer_cancel(&dfops);
814
	xfs_trans_cancel(tp);
L
Linus Torvalds 已提交
815 816
error0:
	xfs_iunlock(ip, XFS_ILOCK_EXCL);
E
Eric Sandeen 已提交
817
	return error;
L
Linus Torvalds 已提交
818 819 820 821 822
}

int
xfs_iomap_write_unwritten(
	xfs_inode_t	*ip,
823
	xfs_off_t	offset,
824 825
	xfs_off_t	count,
	bool		update_isize)
L
Linus Torvalds 已提交
826 827 828 829 830
{
	xfs_mount_t	*mp = ip->i_mount;
	xfs_fileoff_t	offset_fsb;
	xfs_filblks_t	count_fsb;
	xfs_filblks_t	numblks_fsb;
831 832 833 834
	xfs_fsblock_t	firstfsb;
	int		nimaps;
	xfs_trans_t	*tp;
	xfs_bmbt_irec_t imap;
835
	struct xfs_defer_ops dfops;
836
	struct inode	*inode = VFS_I(ip);
837
	xfs_fsize_t	i_size;
838
	uint		resblks;
L
Linus Torvalds 已提交
839 840
	int		error;

C
Christoph Hellwig 已提交
841
	trace_xfs_unwritten_convert(ip, offset, count);
L
Linus Torvalds 已提交
842 843 844 845 846

	offset_fsb = XFS_B_TO_FSBT(mp, offset);
	count_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)offset + count);
	count_fsb = (xfs_filblks_t)(count_fsb - offset_fsb);

847 848 849 850 851 852 853 854 855 856
	/*
	 * Reserve enough blocks in this transaction for two complete extent
	 * btree splits.  We may be converting the middle part of an unwritten
	 * extent and in this case we will insert two new extents in the btree
	 * each of which could cause a full split.
	 *
	 * This reservation amount will be used in the first call to
	 * xfs_bmbt_split() to select an AG with enough space to satisfy the
	 * rest of the operation.
	 */
857
	resblks = XFS_DIOSTRAT_SPACE_RES(mp, 0) << 1;
L
Linus Torvalds 已提交
858

859
	do {
L
Linus Torvalds 已提交
860
		/*
861
		 * Set up a transaction to convert the range of extents
L
Linus Torvalds 已提交
862 863
		 * from unwritten to real. Do allocations in a loop until
		 * we have covered the range passed in.
864
		 *
865 866 867
		 * Note that we can't risk to recursing back into the filesystem
		 * here as we might be asked to write out the same inode that we
		 * complete here and might deadlock on the iolock.
L
Linus Torvalds 已提交
868
		 */
869 870 871
		error = xfs_trans_alloc(mp, &M_RES(mp)->tr_write, resblks, 0,
				XFS_TRANS_RESERVE | XFS_TRANS_NOFS, &tp);
		if (error)
E
Eric Sandeen 已提交
872
			return error;
L
Linus Torvalds 已提交
873 874

		xfs_ilock(ip, XFS_ILOCK_EXCL);
875
		xfs_trans_ijoin(tp, ip, 0);
L
Linus Torvalds 已提交
876 877 878 879

		/*
		 * Modify the unwritten extent state of the buffer.
		 */
880
		xfs_defer_init(&dfops, &firstfsb);
L
Linus Torvalds 已提交
881
		nimaps = 1;
882
		error = xfs_bmapi_write(tp, ip, offset_fsb, count_fsb,
883
					XFS_BMAPI_CONVERT, &firstfsb, resblks,
884
					&imap, &nimaps, &dfops);
L
Linus Torvalds 已提交
885 886 887
		if (error)
			goto error_on_bmapi_transaction;

888 889 890 891 892 893 894 895
		/*
		 * Log the updated inode size as we go.  We have to be careful
		 * to only log it up to the actual write offset if it is
		 * halfway into a block.
		 */
		i_size = XFS_FSB_TO_B(mp, offset_fsb + count_fsb);
		if (i_size > offset + count)
			i_size = offset + count;
896 897
		if (update_isize && i_size > i_size_read(inode))
			i_size_write(inode, i_size);
898 899 900 901 902 903
		i_size = xfs_new_eof(ip, i_size);
		if (i_size) {
			ip->i_d.di_size = i_size;
			xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
		}

904
		error = xfs_defer_finish(&tp, &dfops);
L
Linus Torvalds 已提交
905 906 907
		if (error)
			goto error_on_bmapi_transaction;

908
		error = xfs_trans_commit(tp);
L
Linus Torvalds 已提交
909 910
		xfs_iunlock(ip, XFS_ILOCK_EXCL);
		if (error)
E
Eric Sandeen 已提交
911
			return error;
912

913
		if (!(imap.br_startblock || XFS_IS_REALTIME_INODE(ip)))
914
			return xfs_alert_fsblock_zero(ip, &imap);
L
Linus Torvalds 已提交
915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930

		if ((numblks_fsb = imap.br_blockcount) == 0) {
			/*
			 * The numblks_fsb value should always get
			 * smaller, otherwise the loop is stuck.
			 */
			ASSERT(imap.br_blockcount);
			break;
		}
		offset_fsb += numblks_fsb;
		count_fsb -= numblks_fsb;
	} while (count_fsb > 0);

	return 0;

error_on_bmapi_transaction:
931
	xfs_defer_cancel(&dfops);
932
	xfs_trans_cancel(tp);
L
Linus Torvalds 已提交
933
	xfs_iunlock(ip, XFS_ILOCK_EXCL);
E
Eric Sandeen 已提交
934
	return error;
L
Linus Torvalds 已提交
935
}
936

937 938 939 940 941
static inline bool
imap_needs_alloc(
	struct inode		*inode,
	struct xfs_bmbt_irec	*imap,
	int			nimaps)
942 943 944
{
	return !nimaps ||
		imap->br_startblock == HOLESTARTBLOCK ||
945
		imap->br_startblock == DELAYSTARTBLOCK ||
946
		(IS_DAX(inode) && imap->br_state == XFS_EXT_UNWRITTEN);
947 948
}

949 950 951 952
static inline bool
needs_cow_for_zeroing(
	struct xfs_bmbt_irec	*imap,
	int			nimaps)
953 954 955 956 957 958
{
	return nimaps &&
		imap->br_startblock != HOLESTARTBLOCK &&
		imap->br_state != XFS_EXT_UNWRITTEN;
}

959 960 961 962 963
static int
xfs_ilock_for_iomap(
	struct xfs_inode	*ip,
	unsigned		flags,
	unsigned		*lockmode)
C
Christoph Hellwig 已提交
964
{
965 966
	unsigned		mode = XFS_ILOCK_SHARED;

C
Christoph Hellwig 已提交
967
	/*
968 969
	 * COW writes may allocate delalloc space or convert unwritten COW
	 * extents, so we need to make sure to take the lock exclusively here.
C
Christoph Hellwig 已提交
970
	 */
971 972 973 974 975 976 977 978 979
	if (xfs_is_reflink_inode(ip) && (flags & (IOMAP_WRITE | IOMAP_ZERO))) {
		/*
		 * FIXME: It could still overwrite on unshared extents and not
		 * need allocation.
		 */
		if (flags & IOMAP_NOWAIT)
			return -EAGAIN;
		mode = XFS_ILOCK_EXCL;
	}
980 981

	/*
982 983
	 * Extents not yet cached requires exclusive access, don't block.  This
	 * is an opencoded xfs_ilock_data_map_shared() call but with
984 985
	 * non-blocking behaviour.
	 */
986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000
	if (!(ip->i_df.if_flags & XFS_IFEXTENTS)) {
		if (flags & IOMAP_NOWAIT)
			return -EAGAIN;
		mode = XFS_ILOCK_EXCL;
	}

	if (flags & IOMAP_NOWAIT) {
		if (!xfs_ilock_nowait(ip, mode))
			return -EAGAIN;
	} else {
		xfs_ilock(ip, mode);
	}

	*lockmode = mode;
	return 0;
C
Christoph Hellwig 已提交
1001 1002
}

1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015
static int
xfs_file_iomap_begin(
	struct inode		*inode,
	loff_t			offset,
	loff_t			length,
	unsigned		flags,
	struct iomap		*iomap)
{
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;
	struct xfs_bmbt_irec	imap;
	xfs_fileoff_t		offset_fsb, end_fsb;
	int			nimaps = 1, error = 0;
1016
	bool			shared = false, trimmed = false;
1017
	unsigned		lockmode;
1018 1019 1020 1021

	if (XFS_FORCED_SHUTDOWN(mp))
		return -EIO;

C
Christoph Hellwig 已提交
1022 1023
	if (((flags & (IOMAP_WRITE | IOMAP_DIRECT)) == IOMAP_WRITE) &&
			!IS_DAX(inode) && !xfs_get_extsz_hint(ip)) {
1024
		/* Reserve delalloc blocks for regular writeback. */
1025
		return xfs_file_iomap_begin_delay(inode, offset, length, iomap);
1026 1027
	}

1028 1029 1030 1031 1032 1033 1034 1035 1036
	/*
	 * Lock the inode in the manner required for the specified operation and
	 * check for as many conditions that would result in blocking as
	 * possible. This removes most of the non-blocking checks from the
	 * mapping code below.
	 */
	error = xfs_ilock_for_iomap(ip, flags, &lockmode);
	if (error)
		return error;
G
Goldwyn Rodrigues 已提交
1037

1038
	ASSERT(offset <= mp->m_super->s_maxbytes);
1039
	if (offset > mp->m_super->s_maxbytes - length)
1040 1041 1042 1043 1044
		length = mp->m_super->s_maxbytes - offset;
	offset_fsb = XFS_B_TO_FSBT(mp, offset);
	end_fsb = XFS_B_TO_FSB(mp, offset + length);

	error = xfs_bmapi_read(ip, offset_fsb, end_fsb - offset_fsb, &imap,
1045
			       &nimaps, 0);
1046 1047
	if (error)
		goto out_unlock;
1048

1049
	if (flags & IOMAP_REPORT) {
1050 1051 1052
		/* Trim the mapping to the nearest shared extent boundary. */
		error = xfs_reflink_trim_around_shared(ip, &imap, &shared,
				&trimmed);
1053 1054 1055 1056
		if (error)
			goto out_unlock;
	}

1057 1058 1059 1060
	/* Non-modifying mapping requested, so we are done */
	if (!(flags & (IOMAP_WRITE | IOMAP_ZERO)))
		goto out_found;

1061 1062 1063 1064 1065 1066 1067 1068 1069 1070
	/*
	 * Break shared extents if necessary. Checks for non-blocking IO have
	 * been done up front, so we don't need to do them here.
	 */
	if (xfs_is_reflink_inode(ip)) {
		/* if zeroing doesn't need COW allocation, then we are done. */
		if ((flags & IOMAP_ZERO) &&
		    !needs_cow_for_zeroing(&imap, nimaps))
			goto out_found;

1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081
		if (flags & IOMAP_DIRECT) {
			/* may drop and re-acquire the ilock */
			error = xfs_reflink_allocate_cow(ip, &imap, &shared,
					&lockmode);
			if (error)
				goto out_unlock;
		} else {
			error = xfs_reflink_reserve_cow(ip, &imap, &shared);
			if (error)
				goto out_unlock;
		}
1082 1083 1084

		end_fsb = imap.br_startoff + imap.br_blockcount;
		length = XFS_FSB_TO_B(mp, end_fsb) - offset;
1085 1086
	}

1087 1088 1089
	/* Don't need to allocate over holes when doing zeroing operations. */
	if (flags & IOMAP_ZERO)
		goto out_found;
1090

1091 1092
	if (!imap_needs_alloc(inode, &imap, nimaps))
		goto out_found;
1093

1094 1095 1096 1097
	/* If nowait is set bail since we are going to make allocations. */
	if (flags & IOMAP_NOWAIT) {
		error = -EAGAIN;
		goto out_unlock;
1098 1099
	}

1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125
	/*
	 * We cap the maximum length we map to a sane size  to keep the chunks
	 * of work done where somewhat symmetric with the work writeback does.
	 * This is a completely arbitrary number pulled out of thin air as a
	 * best guess for initial testing.
	 *
	 * Note that the values needs to be less than 32-bits wide until the
	 * lower level functions are updated.
	 */
	length = min_t(loff_t, length, 1024 * PAGE_SIZE);

	/*
	 * xfs_iomap_write_direct() expects the shared lock. It is unlocked on
	 * return.
	 */
	if (lockmode == XFS_ILOCK_EXCL)
		xfs_ilock_demote(ip, lockmode);
	error = xfs_iomap_write_direct(ip, offset, length, &imap,
			nimaps);
	if (error)
		return error;

	iomap->flags = IOMAP_F_NEW;
	trace_xfs_iomap_alloc(ip, offset, length, 0, &imap);

out_finish:
1126 1127
	if (xfs_ipincount(ip) && (ip->i_itemp->ili_fsync_fields
				& ~XFS_ILOG_TIMESTAMP))
1128 1129
		iomap->flags |= IOMAP_F_DIRTY;

1130
	xfs_bmbt_to_iomap(ip, iomap, &imap);
1131

1132 1133
	if (shared)
		iomap->flags |= IOMAP_F_SHARED;
1134
	return 0;
1135 1136 1137 1138 1139 1140 1141

out_found:
	ASSERT(nimaps);
	xfs_iunlock(ip, lockmode);
	trace_xfs_iomap_found(ip, offset, length, 0, &imap);
	goto out_finish;

1142 1143 1144
out_unlock:
	xfs_iunlock(ip, lockmode);
	return error;
1145 1146 1147 1148 1149 1150 1151
}

static int
xfs_file_iomap_end_delalloc(
	struct xfs_inode	*ip,
	loff_t			offset,
	loff_t			length,
1152 1153
	ssize_t			written,
	struct iomap		*iomap)
1154 1155 1156 1157 1158 1159
{
	struct xfs_mount	*mp = ip->i_mount;
	xfs_fileoff_t		start_fsb;
	xfs_fileoff_t		end_fsb;
	int			error = 0;

1160 1161 1162 1163
	/*
	 * Behave as if the write failed if drop writes is enabled. Set the NEW
	 * flag to force delalloc cleanup.
	 */
1164
	if (XFS_TEST_ERROR(false, mp, XFS_ERRTAG_DROP_WRITES)) {
1165
		iomap->flags |= IOMAP_F_NEW;
1166
		written = 0;
1167
	}
1168

1169 1170 1171 1172 1173 1174 1175 1176 1177
	/*
	 * start_fsb refers to the first unused block after a short write. If
	 * nothing was written, round offset down to point at the first block in
	 * the range.
	 */
	if (unlikely(!written))
		start_fsb = XFS_B_TO_FSBT(mp, offset);
	else
		start_fsb = XFS_B_TO_FSB(mp, offset + written);
1178 1179 1180
	end_fsb = XFS_B_TO_FSB(mp, offset + length);

	/*
1181 1182
	 * Trim delalloc blocks if they were allocated by this write and we
	 * didn't manage to write the whole range.
1183 1184 1185 1186 1187
	 *
	 * We don't need to care about racing delalloc as we hold i_mutex
	 * across the reserve/allocate/unreserve calls. If there are delalloc
	 * blocks in the range, they are ours.
	 */
1188
	if ((iomap->flags & IOMAP_F_NEW) && start_fsb < end_fsb) {
1189 1190 1191
		truncate_pagecache_range(VFS_I(ip), XFS_FSB_TO_B(mp, start_fsb),
					 XFS_FSB_TO_B(mp, end_fsb) - 1);

1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217
		xfs_ilock(ip, XFS_ILOCK_EXCL);
		error = xfs_bmap_punch_delalloc_range(ip, start_fsb,
					       end_fsb - start_fsb);
		xfs_iunlock(ip, XFS_ILOCK_EXCL);

		if (error && !XFS_FORCED_SHUTDOWN(mp)) {
			xfs_alert(mp, "%s: unable to clean up ino %lld",
				__func__, ip->i_ino);
			return error;
		}
	}

	return 0;
}

static int
xfs_file_iomap_end(
	struct inode		*inode,
	loff_t			offset,
	loff_t			length,
	ssize_t			written,
	unsigned		flags,
	struct iomap		*iomap)
{
	if ((flags & IOMAP_WRITE) && iomap->type == IOMAP_DELALLOC)
		return xfs_file_iomap_end_delalloc(XFS_I(inode), offset,
1218
				length, written, iomap);
1219 1220 1221
	return 0;
}

1222
const struct iomap_ops xfs_iomap_ops = {
1223 1224 1225
	.iomap_begin		= xfs_file_iomap_begin,
	.iomap_end		= xfs_file_iomap_end,
};
1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245

static int
xfs_xattr_iomap_begin(
	struct inode		*inode,
	loff_t			offset,
	loff_t			length,
	unsigned		flags,
	struct iomap		*iomap)
{
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;
	xfs_fileoff_t		offset_fsb = XFS_B_TO_FSBT(mp, offset);
	xfs_fileoff_t		end_fsb = XFS_B_TO_FSB(mp, offset + length);
	struct xfs_bmbt_irec	imap;
	int			nimaps = 1, error = 0;
	unsigned		lockmode;

	if (XFS_FORCED_SHUTDOWN(mp))
		return -EIO;

1246
	lockmode = xfs_ilock_attr_map_shared(ip);
1247 1248

	/* if there are no attribute fork or extents, return ENOENT */
1249
	if (!XFS_IFORK_Q(ip) || !ip->i_d.di_anextents) {
1250 1251 1252 1253 1254 1255
		error = -ENOENT;
		goto out_unlock;
	}

	ASSERT(ip->i_d.di_aformat != XFS_DINODE_FMT_LOCAL);
	error = xfs_bmapi_read(ip, offset_fsb, end_fsb - offset_fsb, &imap,
1256
			       &nimaps, XFS_BMAPI_ATTRFORK);
1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267
out_unlock:
	xfs_iunlock(ip, lockmode);

	if (!error) {
		ASSERT(nimaps);
		xfs_bmbt_to_iomap(ip, iomap, &imap);
	}

	return error;
}

1268
const struct iomap_ops xfs_xattr_iomap_ops = {
1269 1270
	.iomap_begin		= xfs_xattr_iomap_begin,
};