xfs_file.c 35.9 KB
Newer Older
L
Linus Torvalds 已提交
1
/*
2 3
 * Copyright (c) 2000-2005 Silicon Graphics, Inc.
 * All Rights Reserved.
L
Linus Torvalds 已提交
4
 *
5 6
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as
L
Linus Torvalds 已提交
7 8
 * published by the Free Software Foundation.
 *
9 10 11 12
 * This program is distributed in the hope that it would be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
L
Linus Torvalds 已提交
13
 *
14 15 16
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write the Free Software Foundation,
 * Inc.,  51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
L
Linus Torvalds 已提交
17 18
 */
#include "xfs.h"
19
#include "xfs_fs.h"
20
#include "xfs_shared.h"
21
#include "xfs_format.h"
22 23
#include "xfs_log_format.h"
#include "xfs_trans_resv.h"
L
Linus Torvalds 已提交
24
#include "xfs_sb.h"
25
#include "xfs_ag.h"
L
Linus Torvalds 已提交
26
#include "xfs_mount.h"
27 28
#include "xfs_da_format.h"
#include "xfs_da_btree.h"
L
Linus Torvalds 已提交
29
#include "xfs_inode.h"
30
#include "xfs_trans.h"
31
#include "xfs_inode_item.h"
32
#include "xfs_bmap.h"
D
Dave Chinner 已提交
33
#include "xfs_bmap_util.h"
L
Linus Torvalds 已提交
34
#include "xfs_error.h"
35
#include "xfs_dir2.h"
D
Dave Chinner 已提交
36
#include "xfs_dir2_priv.h"
37
#include "xfs_ioctl.h"
38
#include "xfs_trace.h"
39
#include "xfs_log.h"
40
#include "xfs_dinode.h"
L
Linus Torvalds 已提交
41

42
#include <linux/aio.h>
L
Linus Torvalds 已提交
43
#include <linux/dcache.h>
44
#include <linux/falloc.h>
45
#include <linux/pagevec.h>
L
Linus Torvalds 已提交
46

47
static const struct vm_operations_struct xfs_file_vm_ops;
L
Linus Torvalds 已提交
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
/*
 * Locking primitives for read and write IO paths to ensure we consistently use
 * and order the inode->i_mutex, ip->i_lock and ip->i_iolock.
 */
static inline void
xfs_rw_ilock(
	struct xfs_inode	*ip,
	int			type)
{
	if (type & XFS_IOLOCK_EXCL)
		mutex_lock(&VFS_I(ip)->i_mutex);
	xfs_ilock(ip, type);
}

static inline void
xfs_rw_iunlock(
	struct xfs_inode	*ip,
	int			type)
{
	xfs_iunlock(ip, type);
	if (type & XFS_IOLOCK_EXCL)
		mutex_unlock(&VFS_I(ip)->i_mutex);
}

static inline void
xfs_rw_ilock_demote(
	struct xfs_inode	*ip,
	int			type)
{
	xfs_ilock_demote(ip, type);
	if (type & XFS_IOLOCK_EXCL)
		mutex_unlock(&VFS_I(ip)->i_mutex);
}

83 84 85 86 87 88 89 90 91 92
/*
 *	xfs_iozero
 *
 *	xfs_iozero clears the specified range of buffer supplied,
 *	and marks all the affected blocks as valid and modified.  If
 *	an affected block is not allocated, it will be allocated.  If
 *	an affected block is not completely overwritten, and is not
 *	valid before the operation, it will be read from disk before
 *	being partially zeroed.
 */
93
int
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
xfs_iozero(
	struct xfs_inode	*ip,	/* inode			*/
	loff_t			pos,	/* offset in file		*/
	size_t			count)	/* size of data to zero		*/
{
	struct page		*page;
	struct address_space	*mapping;
	int			status;

	mapping = VFS_I(ip)->i_mapping;
	do {
		unsigned offset, bytes;
		void *fsdata;

		offset = (pos & (PAGE_CACHE_SIZE -1)); /* Within page */
		bytes = PAGE_CACHE_SIZE - offset;
		if (bytes > count)
			bytes = count;

		status = pagecache_write_begin(NULL, mapping, pos, bytes,
					AOP_FLAG_UNINTERRUPTIBLE,
					&page, &fsdata);
		if (status)
			break;

		zero_user(page, offset, bytes);

		status = pagecache_write_end(NULL, mapping, pos, bytes, bytes,
					page, fsdata);
		WARN_ON(status <= 0); /* can't return less than zero! */
		pos += bytes;
		count -= bytes;
		status = 0;
	} while (count);

	return (-status);
}

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
/*
 * Fsync operations on directories are much simpler than on regular files,
 * as there is no file data to flush, and thus also no need for explicit
 * cache flush operations, and there are no non-transaction metadata updates
 * on directories either.
 */
STATIC int
xfs_dir_fsync(
	struct file		*file,
	loff_t			start,
	loff_t			end,
	int			datasync)
{
	struct xfs_inode	*ip = XFS_I(file->f_mapping->host);
	struct xfs_mount	*mp = ip->i_mount;
	xfs_lsn_t		lsn = 0;

	trace_xfs_dir_fsync(ip);

	xfs_ilock(ip, XFS_ILOCK_SHARED);
	if (xfs_ipincount(ip))
		lsn = ip->i_itemp->ili_last_lsn;
	xfs_iunlock(ip, XFS_ILOCK_SHARED);

	if (!lsn)
		return 0;
158
	return -_xfs_log_force_lsn(mp, lsn, XFS_LOG_SYNC, NULL);
159 160
}

161 162 163
STATIC int
xfs_file_fsync(
	struct file		*file,
164 165
	loff_t			start,
	loff_t			end,
166 167
	int			datasync)
{
168 169
	struct inode		*inode = file->f_mapping->host;
	struct xfs_inode	*ip = XFS_I(inode);
170
	struct xfs_mount	*mp = ip->i_mount;
171 172
	int			error = 0;
	int			log_flushed = 0;
173
	xfs_lsn_t		lsn = 0;
174

C
Christoph Hellwig 已提交
175
	trace_xfs_file_fsync(ip);
176

177 178 179 180
	error = filemap_write_and_wait_range(inode->i_mapping, start, end);
	if (error)
		return error;

181
	if (XFS_FORCED_SHUTDOWN(mp))
182 183 184 185
		return -XFS_ERROR(EIO);

	xfs_iflags_clear(ip, XFS_ITRUNCATED);

186 187 188 189 190 191 192 193 194 195 196 197 198 199
	if (mp->m_flags & XFS_MOUNT_BARRIER) {
		/*
		 * If we have an RT and/or log subvolume we need to make sure
		 * to flush the write cache the device used for file data
		 * first.  This is to ensure newly written file data make
		 * it to disk before logging the new inode size in case of
		 * an extending write.
		 */
		if (XFS_IS_REALTIME_INODE(ip))
			xfs_blkdev_issue_flush(mp->m_rtdev_targp);
		else if (mp->m_logdev_targp != mp->m_ddev_targp)
			xfs_blkdev_issue_flush(mp->m_ddev_targp);
	}

200
	/*
C
Christoph Hellwig 已提交
201 202
	 * All metadata updates are logged, which means that we just have
	 * to flush the log up to the latest LSN that touched the inode.
203 204
	 */
	xfs_ilock(ip, XFS_ILOCK_SHARED);
205 206 207 208 209
	if (xfs_ipincount(ip)) {
		if (!datasync ||
		    (ip->i_itemp->ili_fields & ~XFS_ILOG_TIMESTAMP))
			lsn = ip->i_itemp->ili_last_lsn;
	}
C
Christoph Hellwig 已提交
210
	xfs_iunlock(ip, XFS_ILOCK_SHARED);
211

C
Christoph Hellwig 已提交
212
	if (lsn)
213 214
		error = _xfs_log_force_lsn(mp, lsn, XFS_LOG_SYNC, &log_flushed);

215 216 217 218 219 220 221 222 223 224 225 226
	/*
	 * If we only have a single device, and the log force about was
	 * a no-op we might have to flush the data device cache here.
	 * This can only happen for fdatasync/O_DSYNC if we were overwriting
	 * an already allocated file and thus do not have any metadata to
	 * commit.
	 */
	if ((mp->m_flags & XFS_MOUNT_BARRIER) &&
	    mp->m_logdev_targp == mp->m_ddev_targp &&
	    !XFS_IS_REALTIME_INODE(ip) &&
	    !log_flushed)
		xfs_blkdev_issue_flush(mp->m_ddev_targp);
227 228 229 230

	return -error;
}

231
STATIC ssize_t
A
Al Viro 已提交
232
xfs_file_read_iter(
233
	struct kiocb		*iocb,
A
Al Viro 已提交
234
	struct iov_iter		*to)
235 236 237
{
	struct file		*file = iocb->ki_filp;
	struct inode		*inode = file->f_mapping->host;
238 239
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;
A
Al Viro 已提交
240
	size_t			size = iov_iter_count(to);
241
	ssize_t			ret = 0;
242
	int			ioflags = 0;
243
	xfs_fsize_t		n;
A
Al Viro 已提交
244
	loff_t			pos = iocb->ki_pos;
245 246 247

	XFS_STATS_INC(xs_read_calls);

248 249 250 251 252
	if (unlikely(file->f_flags & O_DIRECT))
		ioflags |= IO_ISDIRECT;
	if (file->f_mode & FMODE_NOCMTIME)
		ioflags |= IO_INVIS;

253 254 255 256
	if (unlikely(ioflags & IO_ISDIRECT)) {
		xfs_buftarg_t	*target =
			XFS_IS_REALTIME_INODE(ip) ?
				mp->m_rtdev_targp : mp->m_ddev_targp;
257 258
		/* DIO must be aligned to device logical sector size */
		if ((pos | size) & target->bt_logical_sectormask) {
D
Dave Chinner 已提交
259
			if (pos == i_size_read(inode))
260
				return 0;
261 262 263 264
			return -XFS_ERROR(EINVAL);
		}
	}

D
Dave Chinner 已提交
265
	n = mp->m_super->s_maxbytes - pos;
266
	if (n <= 0 || size == 0)
267 268 269 270 271 272 273 274
		return 0;

	if (n < size)
		size = n;

	if (XFS_FORCED_SHUTDOWN(mp))
		return -EIO;

275 276 277 278 279 280 281 282 283 284 285 286 287
	/*
	 * Locking is a bit tricky here. If we take an exclusive lock
	 * for direct IO, we effectively serialise all new concurrent
	 * read IO to this file and block it behind IO that is currently in
	 * progress because IO in progress holds the IO lock shared. We only
	 * need to hold the lock exclusive to blow away the page cache, so
	 * only take lock exclusively if the page cache needs invalidation.
	 * This allows the normal direct IO case of no page cache pages to
	 * proceeed concurrently without serialisation.
	 */
	xfs_rw_ilock(ip, XFS_IOLOCK_SHARED);
	if ((ioflags & IO_ISDIRECT) && inode->i_mapping->nrpages) {
		xfs_rw_iunlock(ip, XFS_IOLOCK_SHARED);
288 289
		xfs_rw_ilock(ip, XFS_IOLOCK_EXCL);

290
		if (inode->i_mapping->nrpages) {
291
			ret = filemap_write_and_wait_range(
D
Dave Chinner 已提交
292 293
							VFS_I(ip)->i_mapping,
							pos, -1);
294 295 296 297
			if (ret) {
				xfs_rw_iunlock(ip, XFS_IOLOCK_EXCL);
				return ret;
			}
D
Dave Chinner 已提交
298
			truncate_pagecache_range(VFS_I(ip), pos, -1);
299
		}
300
		xfs_rw_ilock_demote(ip, XFS_IOLOCK_EXCL);
301
	}
302

D
Dave Chinner 已提交
303
	trace_xfs_file_read(ip, size, pos, ioflags);
304

A
Al Viro 已提交
305
	ret = generic_file_read_iter(iocb, to);
306 307 308
	if (ret > 0)
		XFS_STATS_ADD(xs_read_bytes, ret);

309
	xfs_rw_iunlock(ip, XFS_IOLOCK_SHARED);
310 311 312
	return ret;
}

313 314
STATIC ssize_t
xfs_file_splice_read(
315 316 317 318
	struct file		*infilp,
	loff_t			*ppos,
	struct pipe_inode_info	*pipe,
	size_t			count,
319
	unsigned int		flags)
320
{
321 322
	struct xfs_inode	*ip = XFS_I(infilp->f_mapping->host);
	int			ioflags = 0;
323 324 325
	ssize_t			ret;

	XFS_STATS_INC(xs_read_calls);
326 327 328 329

	if (infilp->f_mode & FMODE_NOCMTIME)
		ioflags |= IO_INVIS;

330 331 332
	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
		return -EIO;

333
	xfs_rw_ilock(ip, XFS_IOLOCK_SHARED);
334 335 336 337 338 339 340

	trace_xfs_file_splice_read(ip, count, *ppos, ioflags);

	ret = generic_file_splice_read(infilp, ppos, pipe, count, flags);
	if (ret > 0)
		XFS_STATS_ADD(xs_read_bytes, ret);

341
	xfs_rw_iunlock(ip, XFS_IOLOCK_SHARED);
342 343 344 345
	return ret;
}

/*
346 347 348 349
 * This routine is called to handle zeroing any space in the last block of the
 * file that is beyond the EOF.  We do this since the size is being increased
 * without writing anything to that block and we don't want to read the
 * garbage on the disk.
350 351 352
 */
STATIC int				/* error (positive) */
xfs_zero_last_block(
353 354 355
	struct xfs_inode	*ip,
	xfs_fsize_t		offset,
	xfs_fsize_t		isize)
356
{
357 358 359 360 361 362 363
	struct xfs_mount	*mp = ip->i_mount;
	xfs_fileoff_t		last_fsb = XFS_B_TO_FSBT(mp, isize);
	int			zero_offset = XFS_B_FSB_OFFSET(mp, isize);
	int			zero_len;
	int			nimaps = 1;
	int			error = 0;
	struct xfs_bmbt_irec	imap;
364

365
	xfs_ilock(ip, XFS_ILOCK_EXCL);
D
Dave Chinner 已提交
366
	error = xfs_bmapi_read(ip, last_fsb, 1, &imap, &nimaps, 0);
367
	xfs_iunlock(ip, XFS_ILOCK_EXCL);
D
Dave Chinner 已提交
368
	if (error)
369
		return error;
370

371
	ASSERT(nimaps > 0);
372

373 374 375 376
	/*
	 * If the block underlying isize is just a hole, then there
	 * is nothing to zero.
	 */
377
	if (imap.br_startblock == HOLESTARTBLOCK)
378 379 380 381 382
		return 0;

	zero_len = mp->m_sb.sb_blocksize - zero_offset;
	if (isize + zero_len > offset)
		zero_len = offset - isize;
383
	return xfs_iozero(ip, isize, zero_len);
384 385 386
}

/*
387 388 389 390 391 392 393 394 395
 * Zero any on disk space between the current EOF and the new, larger EOF.
 *
 * This handles the normal case of zeroing the remainder of the last block in
 * the file and the unusual case of zeroing blocks out beyond the size of the
 * file.  This second case only happens with fixed size extents and when the
 * system crashes before the inode size was updated but after blocks were
 * allocated.
 *
 * Expects the iolock to be held exclusive, and will take the ilock internally.
396 397 398
 */
int					/* error (positive) */
xfs_zero_eof(
399 400 401
	struct xfs_inode	*ip,
	xfs_off_t		offset,		/* starting I/O offset */
	xfs_fsize_t		isize)		/* current inode size */
402
{
403 404 405 406 407 408 409 410 411 412 413 414
	struct xfs_mount	*mp = ip->i_mount;
	xfs_fileoff_t		start_zero_fsb;
	xfs_fileoff_t		end_zero_fsb;
	xfs_fileoff_t		zero_count_fsb;
	xfs_fileoff_t		last_fsb;
	xfs_fileoff_t		zero_off;
	xfs_fsize_t		zero_len;
	int			nimaps;
	int			error = 0;
	struct xfs_bmbt_irec	imap;

	ASSERT(xfs_isilocked(ip, XFS_IOLOCK_EXCL));
415 416 417 418
	ASSERT(offset > isize);

	/*
	 * First handle zeroing the block on which isize resides.
419
	 *
420 421
	 * We only zero a part of that block so it is handled specially.
	 */
422 423 424 425
	if (XFS_B_FSB_OFFSET(mp, isize) != 0) {
		error = xfs_zero_last_block(ip, offset, isize);
		if (error)
			return error;
426 427 428
	}

	/*
429 430 431 432 433 434 435
	 * Calculate the range between the new size and the old where blocks
	 * needing to be zeroed may exist.
	 *
	 * To get the block where the last byte in the file currently resides,
	 * we need to subtract one from the size and truncate back to a block
	 * boundary.  We subtract 1 in case the size is exactly on a block
	 * boundary.
436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452
	 */
	last_fsb = isize ? XFS_B_TO_FSBT(mp, isize - 1) : (xfs_fileoff_t)-1;
	start_zero_fsb = XFS_B_TO_FSB(mp, (xfs_ufsize_t)isize);
	end_zero_fsb = XFS_B_TO_FSBT(mp, offset - 1);
	ASSERT((xfs_sfiloff_t)last_fsb < (xfs_sfiloff_t)start_zero_fsb);
	if (last_fsb == end_zero_fsb) {
		/*
		 * The size was only incremented on its last block.
		 * We took care of that above, so just return.
		 */
		return 0;
	}

	ASSERT(start_zero_fsb <= end_zero_fsb);
	while (start_zero_fsb <= end_zero_fsb) {
		nimaps = 1;
		zero_count_fsb = end_zero_fsb - start_zero_fsb + 1;
453 454

		xfs_ilock(ip, XFS_ILOCK_EXCL);
D
Dave Chinner 已提交
455 456
		error = xfs_bmapi_read(ip, start_zero_fsb, zero_count_fsb,
					  &imap, &nimaps, 0);
457 458
		xfs_iunlock(ip, XFS_ILOCK_EXCL);
		if (error)
459
			return error;
460

461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479
		ASSERT(nimaps > 0);

		if (imap.br_state == XFS_EXT_UNWRITTEN ||
		    imap.br_startblock == HOLESTARTBLOCK) {
			start_zero_fsb = imap.br_startoff + imap.br_blockcount;
			ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
			continue;
		}

		/*
		 * There are blocks we need to zero.
		 */
		zero_off = XFS_FSB_TO_B(mp, start_zero_fsb);
		zero_len = XFS_FSB_TO_B(mp, imap.br_blockcount);

		if ((zero_off + zero_len) > offset)
			zero_len = offset - zero_off;

		error = xfs_iozero(ip, zero_off, zero_len);
480 481
		if (error)
			return error;
482 483 484 485 486 487 488 489

		start_zero_fsb = imap.br_startoff + imap.br_blockcount;
		ASSERT(start_zero_fsb <= (end_zero_fsb + 1));
	}

	return 0;
}

490 491 492
/*
 * Common pre-write limit and setup checks.
 *
493 494 495
 * Called with the iolocked held either shared and exclusive according to
 * @iolock, and returns with it held.  Might upgrade the iolock to exclusive
 * if called for a direct write beyond i_size.
496 497 498 499 500 501 502 503 504 505 506 507
 */
STATIC ssize_t
xfs_file_aio_write_checks(
	struct file		*file,
	loff_t			*pos,
	size_t			*count,
	int			*iolock)
{
	struct inode		*inode = file->f_mapping->host;
	struct xfs_inode	*ip = XFS_I(inode);
	int			error = 0;

508
restart:
509
	error = generic_write_checks(file, pos, count, S_ISBLK(inode->i_mode));
510
	if (error)
511 512 513 514 515
		return error;

	/*
	 * If the offset is beyond the size of the file, we need to zero any
	 * blocks that fall between the existing EOF and the start of this
516
	 * write.  If zeroing is needed and we are currently holding the
517 518
	 * iolock shared, we need to update it to exclusive which implies
	 * having to redo all checks before.
519
	 */
520
	if (*pos > i_size_read(inode)) {
521
		if (*iolock == XFS_IOLOCK_SHARED) {
522
			xfs_rw_iunlock(ip, *iolock);
523
			*iolock = XFS_IOLOCK_EXCL;
524
			xfs_rw_ilock(ip, *iolock);
525 526
			goto restart;
		}
527
		error = -xfs_zero_eof(ip, *pos, i_size_read(inode));
528 529
		if (error)
			return error;
530
	}
531

C
Christoph Hellwig 已提交
532 533 534 535 536 537
	/*
	 * Updating the timestamps will grab the ilock again from
	 * xfs_fs_dirty_inode, so we have to call it after dropping the
	 * lock above.  Eventually we should look into a way to avoid
	 * the pointless lock roundtrip.
	 */
538 539 540 541 542
	if (likely(!(file->f_mode & FMODE_NOCMTIME))) {
		error = file_update_time(file);
		if (error)
			return error;
	}
C
Christoph Hellwig 已提交
543

544 545 546 547 548 549 550 551
	/*
	 * If we're writing the file then make sure to clear the setuid and
	 * setgid bits if the process is not being run by root.  This keeps
	 * people from modifying setuid and setgid binaries.
	 */
	return file_remove_suid(file);
}

552 553 554 555
/*
 * xfs_file_dio_aio_write - handle direct IO writes
 *
 * Lock the inode appropriately to prepare for and issue a direct IO write.
556
 * By separating it from the buffered write path we remove all the tricky to
557 558
 * follow locking changes and looping.
 *
559 560 561 562 563 564 565 566 567 568 569 570 571
 * If there are cached pages or we're extending the file, we need IOLOCK_EXCL
 * until we're sure the bytes at the new EOF have been zeroed and/or the cached
 * pages are flushed out.
 *
 * In most cases the direct IO writes will be done holding IOLOCK_SHARED
 * allowing them to be done in parallel with reads and other direct IO writes.
 * However, if the IO is not aligned to filesystem blocks, the direct IO layer
 * needs to do sub-block zeroing and that requires serialisation against other
 * direct IOs to the same block. In this case we need to serialise the
 * submission of the unaligned IOs so that we don't get racing block zeroing in
 * the dio layer.  To avoid the problem with aio, we also need to wait for
 * outstanding IOs to complete so that unwritten extent conversion is completed
 * before we try to map the overlapping block. This is currently implemented by
C
Christoph Hellwig 已提交
572
 * hitting it with a big hammer (i.e. inode_dio_wait()).
573
 *
574 575 576 577 578 579
 * Returns with locks held indicated by @iolock and errors indicated by
 * negative return values.
 */
STATIC ssize_t
xfs_file_dio_aio_write(
	struct kiocb		*iocb,
580
	struct iov_iter		*from)
581 582 583 584 585 586 587
{
	struct file		*file = iocb->ki_filp;
	struct address_space	*mapping = file->f_mapping;
	struct inode		*inode = mapping->host;
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;
	ssize_t			ret = 0;
588
	int			unaligned_io = 0;
589
	int			iolock;
590 591
	size_t			count = iov_iter_count(from);
	loff_t			pos = iocb->ki_pos;
592 593 594
	struct xfs_buftarg	*target = XFS_IS_REALTIME_INODE(ip) ?
					mp->m_rtdev_targp : mp->m_ddev_targp;

595 596
	/* DIO must be aligned to device logical sector size */
	if ((pos | count) & target->bt_logical_sectormask)
597 598
		return -XFS_ERROR(EINVAL);

599
	/* "unaligned" here means not aligned to a filesystem block */
600 601 602
	if ((pos & mp->m_blockmask) || ((pos + count) & mp->m_blockmask))
		unaligned_io = 1;

603 604 605 606 607 608 609 610
	/*
	 * We don't need to take an exclusive lock unless there page cache needs
	 * to be invalidated or unaligned IO is being executed. We don't need to
	 * consider the EOF extension case here because
	 * xfs_file_aio_write_checks() will relock the inode as necessary for
	 * EOF zeroing cases and fill out the new inode size as appropriate.
	 */
	if (unaligned_io || mapping->nrpages)
611
		iolock = XFS_IOLOCK_EXCL;
612
	else
613 614
		iolock = XFS_IOLOCK_SHARED;
	xfs_rw_ilock(ip, iolock);
615 616 617 618 619 620

	/*
	 * Recheck if there are cached pages that need invalidate after we got
	 * the iolock to protect against other threads adding new pages while
	 * we were waiting for the iolock.
	 */
621 622 623 624
	if (mapping->nrpages && iolock == XFS_IOLOCK_SHARED) {
		xfs_rw_iunlock(ip, iolock);
		iolock = XFS_IOLOCK_EXCL;
		xfs_rw_ilock(ip, iolock);
625
	}
626

627
	ret = xfs_file_aio_write_checks(file, &pos, &count, &iolock);
628
	if (ret)
629
		goto out;
630
	iov_iter_truncate(from, count);
631 632

	if (mapping->nrpages) {
633
		ret = filemap_write_and_wait_range(VFS_I(ip)->i_mapping,
D
Dave Chinner 已提交
634
						    pos, -1);
635
		if (ret)
636
			goto out;
D
Dave Chinner 已提交
637
		truncate_pagecache_range(VFS_I(ip), pos, -1);
638 639
	}

640 641 642 643 644
	/*
	 * If we are doing unaligned IO, wait for all other IO to drain,
	 * otherwise demote the lock if we had to flush cached pages
	 */
	if (unaligned_io)
C
Christoph Hellwig 已提交
645
		inode_dio_wait(inode);
646
	else if (iolock == XFS_IOLOCK_EXCL) {
647
		xfs_rw_ilock_demote(ip, XFS_IOLOCK_EXCL);
648
		iolock = XFS_IOLOCK_SHARED;
649 650 651
	}

	trace_xfs_file_direct_write(ip, count, iocb->ki_pos, 0);
652
	ret = generic_file_direct_write(iocb, from, pos);
653

654 655 656
out:
	xfs_rw_iunlock(ip, iolock);

657 658 659 660 661
	/* No fallback to buffered IO on errors for XFS. */
	ASSERT(ret < 0 || ret == count);
	return ret;
}

662
STATIC ssize_t
663
xfs_file_buffered_aio_write(
664
	struct kiocb		*iocb,
665
	struct iov_iter		*from)
666 667 668 669
{
	struct file		*file = iocb->ki_filp;
	struct address_space	*mapping = file->f_mapping;
	struct inode		*inode = mapping->host;
670
	struct xfs_inode	*ip = XFS_I(inode);
671 672
	ssize_t			ret;
	int			enospc = 0;
673
	int			iolock = XFS_IOLOCK_EXCL;
674 675
	loff_t			pos = iocb->ki_pos;
	size_t			count = iov_iter_count(from);
676

677
	xfs_rw_ilock(ip, iolock);
678

679
	ret = xfs_file_aio_write_checks(file, &pos, &count, &iolock);
680
	if (ret)
681
		goto out;
682

683
	iov_iter_truncate(from, count);
684 685 686 687
	/* We can write back this queue in page reclaim */
	current->backing_dev_info = mapping->backing_dev_info;

write_retry:
688
	trace_xfs_file_buffered_write(ip, count, iocb->ki_pos, 0);
689
	ret = generic_perform_write(file, from, pos);
690 691
	if (likely(ret >= 0))
		iocb->ki_pos = pos + ret;
692
	/*
D
Dave Chinner 已提交
693 694 695
	 * If we just got an ENOSPC, try to write back all dirty inodes to
	 * convert delalloc space to free up some of the excess reserved
	 * metadata space.
696 697 698
	 */
	if (ret == -ENOSPC && !enospc) {
		enospc = 1;
D
Dave Chinner 已提交
699 700
		xfs_flush_inodes(ip->i_mount);
		goto write_retry;
701
	}
702

703
	current->backing_dev_info = NULL;
704 705
out:
	xfs_rw_iunlock(ip, iolock);
706 707 708 709
	return ret;
}

STATIC ssize_t
A
Al Viro 已提交
710
xfs_file_write_iter(
711
	struct kiocb		*iocb,
A
Al Viro 已提交
712
	struct iov_iter		*from)
713 714 715 716 717 718
{
	struct file		*file = iocb->ki_filp;
	struct address_space	*mapping = file->f_mapping;
	struct inode		*inode = mapping->host;
	struct xfs_inode	*ip = XFS_I(inode);
	ssize_t			ret;
A
Al Viro 已提交
719
	size_t			ocount = iov_iter_count(from);
720 721 722 723 724 725

	XFS_STATS_INC(xs_write_calls);

	if (ocount == 0)
		return 0;

A
Al Viro 已提交
726 727
	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
		return -EIO;
728 729

	if (unlikely(file->f_flags & O_DIRECT))
A
Al Viro 已提交
730
		ret = xfs_file_dio_aio_write(iocb, from);
731
	else
A
Al Viro 已提交
732
		ret = xfs_file_buffered_aio_write(iocb, from);
733

734 735
	if (ret > 0) {
		ssize_t err;
736

737
		XFS_STATS_ADD(xs_write_bytes, ret);
738

739
		/* Handle various SYNC-type writes */
740
		err = generic_write_sync(file, iocb->ki_pos - ret, ret);
741 742
		if (err < 0)
			ret = err;
743
	}
744
	return ret;
745 746
}

747 748
STATIC long
xfs_file_fallocate(
749 750 751 752
	struct file		*file,
	int			mode,
	loff_t			offset,
	loff_t			len)
753
{
754 755 756 757 758
	struct inode		*inode = file_inode(file);
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_trans	*tp;
	long			error;
	loff_t			new_size = 0;
759

760 761
	if (!S_ISREG(inode->i_mode))
		return -EINVAL;
762
	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE |
763
		     FALLOC_FL_COLLAPSE_RANGE | FALLOC_FL_ZERO_RANGE))
764 765 766
		return -EOPNOTSUPP;

	xfs_ilock(ip, XFS_IOLOCK_EXCL);
767 768 769 770
	if (mode & FALLOC_FL_PUNCH_HOLE) {
		error = xfs_free_file_space(ip, offset, len);
		if (error)
			goto out_unlock;
771 772 773 774
	} else if (mode & FALLOC_FL_COLLAPSE_RANGE) {
		unsigned blksize_mask = (1 << inode->i_blkbits) - 1;

		if (offset & blksize_mask || len & blksize_mask) {
775
			error = EINVAL;
776 777 778
			goto out_unlock;
		}

779 780 781 782 783
		/*
		 * There is no need to overlap collapse range with EOF,
		 * in which case it is effectively a truncate operation
		 */
		if (offset + len >= i_size_read(inode)) {
784
			error = EINVAL;
785 786 787
			goto out_unlock;
		}

788 789 790 791 792
		new_size = i_size_read(inode) - len;

		error = xfs_collapse_file_space(ip, offset, len);
		if (error)
			goto out_unlock;
793 794 795 796 797 798 799 800
	} else {
		if (!(mode & FALLOC_FL_KEEP_SIZE) &&
		    offset + len > i_size_read(inode)) {
			new_size = offset + len;
			error = -inode_newsize_ok(inode, new_size);
			if (error)
				goto out_unlock;
		}
801

802 803 804 805 806
		if (mode & FALLOC_FL_ZERO_RANGE)
			error = xfs_zero_file_space(ip, offset, len);
		else
			error = xfs_alloc_file_space(ip, offset, len,
						     XFS_BMAPI_PREALLOC);
807 808 809 810
		if (error)
			goto out_unlock;
	}

811 812 813 814 815 816 817 818 819 820 821 822
	tp = xfs_trans_alloc(ip->i_mount, XFS_TRANS_WRITEID);
	error = xfs_trans_reserve(tp, &M_RES(ip->i_mount)->tr_writeid, 0, 0);
	if (error) {
		xfs_trans_cancel(tp, 0);
		goto out_unlock;
	}

	xfs_ilock(ip, XFS_ILOCK_EXCL);
	xfs_trans_ijoin(tp, ip, XFS_ILOCK_EXCL);
	ip->i_d.di_mode &= ~S_ISUID;
	if (ip->i_d.di_mode & S_IXGRP)
		ip->i_d.di_mode &= ~S_ISGID;
823

824
	if (!(mode & (FALLOC_FL_PUNCH_HOLE | FALLOC_FL_COLLAPSE_RANGE)))
825 826 827 828 829 830 831 832
		ip->i_d.di_flags |= XFS_DIFLAG_PREALLOC;

	xfs_trans_ichgtime(tp, ip, XFS_ICHGTIME_MOD | XFS_ICHGTIME_CHG);
	xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);

	if (file->f_flags & O_DSYNC)
		xfs_trans_set_sync(tp);
	error = xfs_trans_commit(tp, 0);
833 834 835 836 837 838 839 840 841
	if (error)
		goto out_unlock;

	/* Change file size if needed */
	if (new_size) {
		struct iattr iattr;

		iattr.ia_valid = ATTR_SIZE;
		iattr.ia_size = new_size;
842
		error = xfs_setattr_size(ip, &iattr);
843 844 845 846
	}

out_unlock:
	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
847
	return -error;
848 849 850
}


L
Linus Torvalds 已提交
851
STATIC int
852
xfs_file_open(
L
Linus Torvalds 已提交
853
	struct inode	*inode,
854
	struct file	*file)
L
Linus Torvalds 已提交
855
{
856
	if (!(file->f_flags & O_LARGEFILE) && i_size_read(inode) > MAX_NON_LFS)
L
Linus Torvalds 已提交
857
		return -EFBIG;
858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879
	if (XFS_FORCED_SHUTDOWN(XFS_M(inode->i_sb)))
		return -EIO;
	return 0;
}

STATIC int
xfs_dir_open(
	struct inode	*inode,
	struct file	*file)
{
	struct xfs_inode *ip = XFS_I(inode);
	int		mode;
	int		error;

	error = xfs_file_open(inode, file);
	if (error)
		return error;

	/*
	 * If there are any blocks, read-ahead block 0 as we're almost
	 * certain to have the next operation be a read there.
	 */
880
	mode = xfs_ilock_data_map_shared(ip);
881
	if (ip->i_d.di_nextents > 0)
882
		xfs_dir3_data_readahead(ip, 0, -1);
883 884
	xfs_iunlock(ip, mode);
	return 0;
L
Linus Torvalds 已提交
885 886 887
}

STATIC int
888
xfs_file_release(
L
Linus Torvalds 已提交
889 890 891
	struct inode	*inode,
	struct file	*filp)
{
892
	return -xfs_release(XFS_I(inode));
L
Linus Torvalds 已提交
893 894 895
}

STATIC int
896
xfs_file_readdir(
A
Al Viro 已提交
897 898
	struct file	*file,
	struct dir_context *ctx)
L
Linus Torvalds 已提交
899
{
A
Al Viro 已提交
900
	struct inode	*inode = file_inode(file);
901
	xfs_inode_t	*ip = XFS_I(inode);
C
Christoph Hellwig 已提交
902 903 904 905 906 907 908 909 910 911 912 913 914
	int		error;
	size_t		bufsize;

	/*
	 * The Linux API doesn't pass down the total size of the buffer
	 * we read into down to the filesystem.  With the filldir concept
	 * it's not needed for correct information, but the XFS dir2 leaf
	 * code wants an estimate of the buffer size to calculate it's
	 * readahead window and size the buffers used for mapping to
	 * physical blocks.
	 *
	 * Try to give it an estimate that's good enough, maybe at some
	 * point we can change the ->readdir prototype to include the
E
Eric Sandeen 已提交
915
	 * buffer size.  For now we use the current glibc buffer size.
C
Christoph Hellwig 已提交
916
	 */
E
Eric Sandeen 已提交
917
	bufsize = (size_t)min_t(loff_t, 32768, ip->i_d.di_size);
C
Christoph Hellwig 已提交
918

A
Al Viro 已提交
919
	error = xfs_readdir(ip, ctx, bufsize);
C
Christoph Hellwig 已提交
920 921 922
	if (error)
		return -error;
	return 0;
L
Linus Torvalds 已提交
923 924 925
}

STATIC int
926
xfs_file_mmap(
L
Linus Torvalds 已提交
927 928 929
	struct file	*filp,
	struct vm_area_struct *vma)
{
930
	vma->vm_ops = &xfs_file_vm_ops;
931

932
	file_accessed(filp);
L
Linus Torvalds 已提交
933 934 935
	return 0;
}

936 937 938 939 940 941 942 943 944
/*
 * mmap()d file has taken write protection fault and is being made
 * writable. We can set the page state up correctly for a writable
 * page, which means we can do correct delalloc accounting (ENOSPC
 * checking!) and unwritten extent mapping.
 */
STATIC int
xfs_vm_page_mkwrite(
	struct vm_area_struct	*vma,
945
	struct vm_fault		*vmf)
946
{
947
	return block_page_mkwrite(vma, vmf, xfs_get_blocks);
948 949
}

950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 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 1049 1050 1051 1052 1053 1054 1055 1056 1057 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 1096 1097 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 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167
/*
 * This type is designed to indicate the type of offset we would like
 * to search from page cache for either xfs_seek_data() or xfs_seek_hole().
 */
enum {
	HOLE_OFF = 0,
	DATA_OFF,
};

/*
 * Lookup the desired type of offset from the given page.
 *
 * On success, return true and the offset argument will point to the
 * start of the region that was found.  Otherwise this function will
 * return false and keep the offset argument unchanged.
 */
STATIC bool
xfs_lookup_buffer_offset(
	struct page		*page,
	loff_t			*offset,
	unsigned int		type)
{
	loff_t			lastoff = page_offset(page);
	bool			found = false;
	struct buffer_head	*bh, *head;

	bh = head = page_buffers(page);
	do {
		/*
		 * Unwritten extents that have data in the page
		 * cache covering them can be identified by the
		 * BH_Unwritten state flag.  Pages with multiple
		 * buffers might have a mix of holes, data and
		 * unwritten extents - any buffer with valid
		 * data in it should have BH_Uptodate flag set
		 * on it.
		 */
		if (buffer_unwritten(bh) ||
		    buffer_uptodate(bh)) {
			if (type == DATA_OFF)
				found = true;
		} else {
			if (type == HOLE_OFF)
				found = true;
		}

		if (found) {
			*offset = lastoff;
			break;
		}
		lastoff += bh->b_size;
	} while ((bh = bh->b_this_page) != head);

	return found;
}

/*
 * This routine is called to find out and return a data or hole offset
 * from the page cache for unwritten extents according to the desired
 * type for xfs_seek_data() or xfs_seek_hole().
 *
 * The argument offset is used to tell where we start to search from the
 * page cache.  Map is used to figure out the end points of the range to
 * lookup pages.
 *
 * Return true if the desired type of offset was found, and the argument
 * offset is filled with that address.  Otherwise, return false and keep
 * offset unchanged.
 */
STATIC bool
xfs_find_get_desired_pgoff(
	struct inode		*inode,
	struct xfs_bmbt_irec	*map,
	unsigned int		type,
	loff_t			*offset)
{
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;
	struct pagevec		pvec;
	pgoff_t			index;
	pgoff_t			end;
	loff_t			endoff;
	loff_t			startoff = *offset;
	loff_t			lastoff = startoff;
	bool			found = false;

	pagevec_init(&pvec, 0);

	index = startoff >> PAGE_CACHE_SHIFT;
	endoff = XFS_FSB_TO_B(mp, map->br_startoff + map->br_blockcount);
	end = endoff >> PAGE_CACHE_SHIFT;
	do {
		int		want;
		unsigned	nr_pages;
		unsigned int	i;

		want = min_t(pgoff_t, end - index, PAGEVEC_SIZE);
		nr_pages = pagevec_lookup(&pvec, inode->i_mapping, index,
					  want);
		/*
		 * No page mapped into given range.  If we are searching holes
		 * and if this is the first time we got into the loop, it means
		 * that the given offset is landed in a hole, return it.
		 *
		 * If we have already stepped through some block buffers to find
		 * holes but they all contains data.  In this case, the last
		 * offset is already updated and pointed to the end of the last
		 * mapped page, if it does not reach the endpoint to search,
		 * that means there should be a hole between them.
		 */
		if (nr_pages == 0) {
			/* Data search found nothing */
			if (type == DATA_OFF)
				break;

			ASSERT(type == HOLE_OFF);
			if (lastoff == startoff || lastoff < endoff) {
				found = true;
				*offset = lastoff;
			}
			break;
		}

		/*
		 * At lease we found one page.  If this is the first time we
		 * step into the loop, and if the first page index offset is
		 * greater than the given search offset, a hole was found.
		 */
		if (type == HOLE_OFF && lastoff == startoff &&
		    lastoff < page_offset(pvec.pages[0])) {
			found = true;
			break;
		}

		for (i = 0; i < nr_pages; i++) {
			struct page	*page = pvec.pages[i];
			loff_t		b_offset;

			/*
			 * At this point, the page may be truncated or
			 * invalidated (changing page->mapping to NULL),
			 * or even swizzled back from swapper_space to tmpfs
			 * file mapping. However, page->index will not change
			 * because we have a reference on the page.
			 *
			 * Searching done if the page index is out of range.
			 * If the current offset is not reaches the end of
			 * the specified search range, there should be a hole
			 * between them.
			 */
			if (page->index > end) {
				if (type == HOLE_OFF && lastoff < endoff) {
					*offset = lastoff;
					found = true;
				}
				goto out;
			}

			lock_page(page);
			/*
			 * Page truncated or invalidated(page->mapping == NULL).
			 * We can freely skip it and proceed to check the next
			 * page.
			 */
			if (unlikely(page->mapping != inode->i_mapping)) {
				unlock_page(page);
				continue;
			}

			if (!page_has_buffers(page)) {
				unlock_page(page);
				continue;
			}

			found = xfs_lookup_buffer_offset(page, &b_offset, type);
			if (found) {
				/*
				 * The found offset may be less than the start
				 * point to search if this is the first time to
				 * come here.
				 */
				*offset = max_t(loff_t, startoff, b_offset);
				unlock_page(page);
				goto out;
			}

			/*
			 * We either searching data but nothing was found, or
			 * searching hole but found a data buffer.  In either
			 * case, probably the next page contains the desired
			 * things, update the last offset to it so.
			 */
			lastoff = page_offset(page) + PAGE_SIZE;
			unlock_page(page);
		}

		/*
		 * The number of returned pages less than our desired, search
		 * done.  In this case, nothing was found for searching data,
		 * but we found a hole behind the last offset.
		 */
		if (nr_pages < want) {
			if (type == HOLE_OFF) {
				*offset = lastoff;
				found = true;
			}
			break;
		}

		index = pvec.pages[i - 1]->index + 1;
		pagevec_release(&pvec);
	} while (index <= end);

out:
	pagevec_release(&pvec);
	return found;
}

1168 1169 1170
STATIC loff_t
xfs_seek_data(
	struct file		*file,
1171
	loff_t			start)
1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182
{
	struct inode		*inode = file->f_mapping->host;
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;
	loff_t			uninitialized_var(offset);
	xfs_fsize_t		isize;
	xfs_fileoff_t		fsbno;
	xfs_filblks_t		end;
	uint			lock;
	int			error;

1183
	lock = xfs_ilock_data_map_shared(ip);
1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194

	isize = i_size_read(inode);
	if (start >= isize) {
		error = ENXIO;
		goto out_unlock;
	}

	/*
	 * Try to read extents from the first block indicated
	 * by fsbno to the end block of the file.
	 */
1195
	fsbno = XFS_B_TO_FSBT(mp, start);
1196
	end = XFS_B_TO_FSB(mp, isize);
1197 1198 1199 1200
	for (;;) {
		struct xfs_bmbt_irec	map[2];
		int			nmap = 2;
		unsigned int		i;
1201

1202 1203 1204 1205
		error = xfs_bmapi_read(ip, fsbno, end - fsbno, map, &nmap,
				       XFS_BMAPI_ENTIRE);
		if (error)
			goto out_unlock;
1206

1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238
		/* No extents at given offset, must be beyond EOF */
		if (nmap == 0) {
			error = ENXIO;
			goto out_unlock;
		}

		for (i = 0; i < nmap; i++) {
			offset = max_t(loff_t, start,
				       XFS_FSB_TO_B(mp, map[i].br_startoff));

			/* Landed in a data extent */
			if (map[i].br_startblock == DELAYSTARTBLOCK ||
			    (map[i].br_state == XFS_EXT_NORM &&
			     !isnullstartblock(map[i].br_startblock)))
				goto out;

			/*
			 * Landed in an unwritten extent, try to search data
			 * from page cache.
			 */
			if (map[i].br_state == XFS_EXT_UNWRITTEN) {
				if (xfs_find_get_desired_pgoff(inode, &map[i],
							DATA_OFF, &offset))
					goto out;
			}
		}

		/*
		 * map[0] is hole or its an unwritten extent but
		 * without data in page cache.  Probably means that
		 * we are reading after EOF if nothing in map[1].
		 */
1239 1240 1241 1242 1243
		if (nmap == 1) {
			error = ENXIO;
			goto out_unlock;
		}

1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255
		ASSERT(i > 1);

		/*
		 * Nothing was found, proceed to the next round of search
		 * if reading offset not beyond or hit EOF.
		 */
		fsbno = map[i - 1].br_startoff + map[i - 1].br_blockcount;
		start = XFS_FSB_TO_B(mp, fsbno);
		if (start >= isize) {
			error = ENXIO;
			goto out_unlock;
		}
1256 1257
	}

1258
out:
J
Jie Liu 已提交
1259
	offset = vfs_setpos(file, offset, inode->i_sb->s_maxbytes);
1260 1261

out_unlock:
1262
	xfs_iunlock(ip, lock);
1263 1264 1265 1266 1267 1268 1269 1270 1271

	if (error)
		return -error;
	return offset;
}

STATIC loff_t
xfs_seek_hole(
	struct file		*file,
1272
	loff_t			start)
1273 1274 1275 1276 1277 1278 1279
{
	struct inode		*inode = file->f_mapping->host;
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;
	loff_t			uninitialized_var(offset);
	xfs_fsize_t		isize;
	xfs_fileoff_t		fsbno;
1280
	xfs_filblks_t		end;
1281 1282 1283 1284 1285 1286
	uint			lock;
	int			error;

	if (XFS_FORCED_SHUTDOWN(mp))
		return -XFS_ERROR(EIO);

1287
	lock = xfs_ilock_data_map_shared(ip);
1288 1289 1290 1291 1292 1293 1294 1295

	isize = i_size_read(inode);
	if (start >= isize) {
		error = ENXIO;
		goto out_unlock;
	}

	fsbno = XFS_B_TO_FSBT(mp, start);
1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331
	end = XFS_B_TO_FSB(mp, isize);

	for (;;) {
		struct xfs_bmbt_irec	map[2];
		int			nmap = 2;
		unsigned int		i;

		error = xfs_bmapi_read(ip, fsbno, end - fsbno, map, &nmap,
				       XFS_BMAPI_ENTIRE);
		if (error)
			goto out_unlock;

		/* No extents at given offset, must be beyond EOF */
		if (nmap == 0) {
			error = ENXIO;
			goto out_unlock;
		}

		for (i = 0; i < nmap; i++) {
			offset = max_t(loff_t, start,
				       XFS_FSB_TO_B(mp, map[i].br_startoff));

			/* Landed in a hole */
			if (map[i].br_startblock == HOLESTARTBLOCK)
				goto out;

			/*
			 * Landed in an unwritten extent, try to search hole
			 * from page cache.
			 */
			if (map[i].br_state == XFS_EXT_UNWRITTEN) {
				if (xfs_find_get_desired_pgoff(inode, &map[i],
							HOLE_OFF, &offset))
					goto out;
			}
		}
1332 1333

		/*
1334 1335 1336 1337 1338
		 * map[0] contains data or its unwritten but contains
		 * data in page cache, probably means that we are
		 * reading after EOF.  We should fix offset to point
		 * to the end of the file(i.e., there is an implicit
		 * hole at the end of any file).
1339
		 */
1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356
		if (nmap == 1) {
			offset = isize;
			break;
		}

		ASSERT(i > 1);

		/*
		 * Both mappings contains data, proceed to the next round of
		 * search if the current reading offset not beyond or hit EOF.
		 */
		fsbno = map[i - 1].br_startoff + map[i - 1].br_blockcount;
		start = XFS_FSB_TO_B(mp, fsbno);
		if (start >= isize) {
			offset = isize;
			break;
		}
1357 1358
	}

1359 1360 1361 1362 1363 1364 1365 1366
out:
	/*
	 * At this point, we must have found a hole.  However, the returned
	 * offset may be bigger than the file size as it may be aligned to
	 * page boundary for unwritten extents, we need to deal with this
	 * situation in particular.
	 */
	offset = min_t(loff_t, offset, isize);
J
Jie Liu 已提交
1367
	offset = vfs_setpos(file, offset, inode->i_sb->s_maxbytes);
1368 1369

out_unlock:
1370
	xfs_iunlock(ip, lock);
1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388

	if (error)
		return -error;
	return offset;
}

STATIC loff_t
xfs_file_llseek(
	struct file	*file,
	loff_t		offset,
	int		origin)
{
	switch (origin) {
	case SEEK_END:
	case SEEK_CUR:
	case SEEK_SET:
		return generic_file_llseek(file, offset, origin);
	case SEEK_DATA:
1389
		return xfs_seek_data(file, offset);
1390
	case SEEK_HOLE:
1391
		return xfs_seek_hole(file, offset);
1392 1393 1394 1395 1396
	default:
		return -EINVAL;
	}
}

1397
const struct file_operations xfs_file_operations = {
1398
	.llseek		= xfs_file_llseek,
A
Al Viro 已提交
1399
	.read		= new_sync_read,
A
Al Viro 已提交
1400
	.write		= new_sync_write,
A
Al Viro 已提交
1401
	.read_iter	= xfs_file_read_iter,
A
Al Viro 已提交
1402
	.write_iter	= xfs_file_write_iter,
1403
	.splice_read	= xfs_file_splice_read,
A
Al Viro 已提交
1404
	.splice_write	= iter_file_splice_write,
1405
	.unlocked_ioctl	= xfs_file_ioctl,
L
Linus Torvalds 已提交
1406
#ifdef CONFIG_COMPAT
1407
	.compat_ioctl	= xfs_file_compat_ioctl,
L
Linus Torvalds 已提交
1408
#endif
1409 1410 1411 1412
	.mmap		= xfs_file_mmap,
	.open		= xfs_file_open,
	.release	= xfs_file_release,
	.fsync		= xfs_file_fsync,
1413
	.fallocate	= xfs_file_fallocate,
L
Linus Torvalds 已提交
1414 1415
};

1416
const struct file_operations xfs_dir_file_operations = {
1417
	.open		= xfs_dir_open,
L
Linus Torvalds 已提交
1418
	.read		= generic_read_dir,
A
Al Viro 已提交
1419
	.iterate	= xfs_file_readdir,
1420
	.llseek		= generic_file_llseek,
1421
	.unlocked_ioctl	= xfs_file_ioctl,
1422
#ifdef CONFIG_COMPAT
1423
	.compat_ioctl	= xfs_file_compat_ioctl,
1424
#endif
1425
	.fsync		= xfs_dir_fsync,
L
Linus Torvalds 已提交
1426 1427
};

1428
static const struct vm_operations_struct xfs_file_vm_ops = {
1429
	.fault		= filemap_fault,
1430
	.map_pages	= filemap_map_pages,
1431
	.page_mkwrite	= xfs_vm_page_mkwrite,
1432
	.remap_pages	= generic_file_remap_pages,
1433
};