xfs_file.c 30.5 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_bit.h"
L
Linus Torvalds 已提交
21
#include "xfs_log.h"
22
#include "xfs_inum.h"
L
Linus Torvalds 已提交
23
#include "xfs_sb.h"
24
#include "xfs_ag.h"
L
Linus Torvalds 已提交
25 26 27 28 29 30
#include "xfs_trans.h"
#include "xfs_mount.h"
#include "xfs_bmap_btree.h"
#include "xfs_alloc.h"
#include "xfs_dinode.h"
#include "xfs_inode.h"
31
#include "xfs_inode_item.h"
32
#include "xfs_bmap.h"
L
Linus Torvalds 已提交
33
#include "xfs_error.h"
34
#include "xfs_vnodeops.h"
35
#include "xfs_da_btree.h"
36
#include "xfs_ioctl.h"
37
#include "xfs_trace.h"
L
Linus Torvalds 已提交
38 39

#include <linux/dcache.h>
40
#include <linux/falloc.h>
L
Linus Torvalds 已提交
41

42
static const struct vm_operations_struct xfs_file_vm_ops;
L
Linus Torvalds 已提交
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
/*
 * 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);
}

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
/*
 *	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.
 */
STATIC int
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);
}

127 128 129
STATIC int
xfs_file_fsync(
	struct file		*file,
130 131
	loff_t			start,
	loff_t			end,
132 133
	int			datasync)
{
134 135
	struct inode		*inode = file->f_mapping->host;
	struct xfs_inode	*ip = XFS_I(inode);
136
	struct xfs_mount	*mp = ip->i_mount;
137 138 139 140
	struct xfs_trans	*tp;
	int			error = 0;
	int			log_flushed = 0;

C
Christoph Hellwig 已提交
141
	trace_xfs_file_fsync(ip);
142

143 144 145 146
	error = filemap_write_and_wait_range(inode->i_mapping, start, end);
	if (error)
		return error;

147
	if (XFS_FORCED_SHUTDOWN(mp))
148 149 150 151
		return -XFS_ERROR(EIO);

	xfs_iflags_clear(ip, XFS_ITRUNCATED);

152 153 154 155 156 157 158 159 160 161 162 163 164 165
	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);
	}

166 167 168 169 170 171 172 173 174 175 176 177 178 179
	/*
	 * We always need to make sure that the required inode state is safe on
	 * disk.  The inode might be clean but we still might need to force the
	 * log because of committed transactions that haven't hit the disk yet.
	 * Likewise, there could be unflushed non-transactional changes to the
	 * inode core that have to go to disk and this requires us to issue
	 * a synchronous transaction to capture these changes correctly.
	 *
	 * This code relies on the assumption that if the i_update_core field
	 * of the inode is clear and the inode is unpinned then it is clean
	 * and no action is required.
	 */
	xfs_ilock(ip, XFS_ILOCK_SHARED);

180 181 182 183 184 185 186 187 188
	/*
	 * First check if the VFS inode is marked dirty.  All the dirtying
	 * of non-transactional updates no goes through mark_inode_dirty*,
	 * which allows us to distinguish beteeen pure timestamp updates
	 * and i_size updates which need to be caught for fdatasync.
	 * After that also theck for the dirty state in the XFS inode, which
	 * might gets cleared when the inode gets written out via the AIL
	 * or xfs_iflush_cluster.
	 */
189 190
	if (((inode->i_state & I_DIRTY_DATASYNC) ||
	    ((inode->i_state & I_DIRTY_SYNC) && !datasync)) &&
191
	    ip->i_update_core) {
192 193 194 195 196
		/*
		 * Kick off a transaction to log the inode core to get the
		 * updates.  The sync transaction will also force the log.
		 */
		xfs_iunlock(ip, XFS_ILOCK_SHARED);
197
		tp = xfs_trans_alloc(mp, XFS_TRANS_FSYNC_TS);
198
		error = xfs_trans_reserve(tp, 0,
199
				XFS_FSYNC_TS_LOG_RES(mp), 0, 0, 0);
200 201 202 203 204 205 206 207 208 209 210 211 212 213 214
		if (error) {
			xfs_trans_cancel(tp, 0);
			return -error;
		}
		xfs_ilock(ip, XFS_ILOCK_EXCL);

		/*
		 * Note - it's possible that we might have pushed ourselves out
		 * of the way during trans_reserve which would flush the inode.
		 * But there's no guarantee that the inode buffer has actually
		 * gone out yet (it's delwri).	Plus the buffer could be pinned
		 * anyway if it's part of an inode in another recent
		 * transaction.	 So we play it safe and fire off the
		 * transaction anyway.
		 */
215
		xfs_trans_ijoin(tp, ip);
216 217 218 219 220 221 222 223 224 225 226 227 228 229 230
		xfs_trans_log_inode(tp, ip, XFS_ILOG_CORE);
		xfs_trans_set_sync(tp);
		error = _xfs_trans_commit(tp, 0, &log_flushed);

		xfs_iunlock(ip, XFS_ILOCK_EXCL);
	} else {
		/*
		 * Timestamps/size haven't changed since last inode flush or
		 * inode transaction commit.  That means either nothing got
		 * written or a transaction committed which caught the updates.
		 * If the latter happened and the transaction hasn't hit the
		 * disk yet, the inode will be still be pinned.  If it is,
		 * force the log.
		 */
		if (xfs_ipincount(ip)) {
231
			error = _xfs_log_force_lsn(mp,
232 233
					ip->i_itemp->ili_last_lsn,
					XFS_LOG_SYNC, &log_flushed);
234
		}
235
		xfs_iunlock(ip, XFS_ILOCK_SHARED);
236 237
	}

238 239 240 241 242 243 244 245 246 247 248 249
	/*
	 * 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);
250 251 252 253

	return -error;
}

254 255
STATIC ssize_t
xfs_file_aio_read(
256 257
	struct kiocb		*iocb,
	const struct iovec	*iovp,
258 259
	unsigned long		nr_segs,
	loff_t			pos)
260 261 262
{
	struct file		*file = iocb->ki_filp;
	struct inode		*inode = file->f_mapping->host;
263 264
	struct xfs_inode	*ip = XFS_I(inode);
	struct xfs_mount	*mp = ip->i_mount;
265 266
	size_t			size = 0;
	ssize_t			ret = 0;
267
	int			ioflags = 0;
268 269 270 271 272
	xfs_fsize_t		n;
	unsigned long		seg;

	XFS_STATS_INC(xs_read_calls);

273 274 275 276 277 278 279
	BUG_ON(iocb->ki_pos != pos);

	if (unlikely(file->f_flags & O_DIRECT))
		ioflags |= IO_ISDIRECT;
	if (file->f_mode & FMODE_NOCMTIME)
		ioflags |= IO_INVIS;

280
	/* START copy & waste from filemap.c */
281
	for (seg = 0; seg < nr_segs; seg++) {
282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297
		const struct iovec *iv = &iovp[seg];

		/*
		 * If any segment has a negative length, or the cumulative
		 * length ever wraps negative then return -EINVAL.
		 */
		size += iv->iov_len;
		if (unlikely((ssize_t)(size|iv->iov_len) < 0))
			return XFS_ERROR(-EINVAL);
	}
	/* END copy & waste from filemap.c */

	if (unlikely(ioflags & IO_ISDIRECT)) {
		xfs_buftarg_t	*target =
			XFS_IS_REALTIME_INODE(ip) ?
				mp->m_rtdev_targp : mp->m_ddev_targp;
298
		if ((iocb->ki_pos & target->bt_smask) ||
299
		    (size & target->bt_smask)) {
300 301
			if (iocb->ki_pos == ip->i_size)
				return 0;
302 303 304 305
			return -XFS_ERROR(EINVAL);
		}
	}

306 307
	n = XFS_MAXIOFFSET(mp) - iocb->ki_pos;
	if (n <= 0 || size == 0)
308 309 310 311 312 313 314 315
		return 0;

	if (n < size)
		size = n;

	if (XFS_FORCED_SHUTDOWN(mp))
		return -EIO;

316 317 318 319 320 321 322 323 324 325 326 327 328
	/*
	 * 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);
329 330
		xfs_rw_ilock(ip, XFS_IOLOCK_EXCL);

331 332 333 334
		if (inode->i_mapping->nrpages) {
			ret = -xfs_flushinval_pages(ip,
					(iocb->ki_pos & PAGE_CACHE_MASK),
					-1, FI_REMAPF_LOCKED);
335 336 337 338
			if (ret) {
				xfs_rw_iunlock(ip, XFS_IOLOCK_EXCL);
				return ret;
			}
339
		}
340
		xfs_rw_ilock_demote(ip, XFS_IOLOCK_EXCL);
341
	}
342

343
	trace_xfs_file_read(ip, size, iocb->ki_pos, ioflags);
344

345
	ret = generic_file_aio_read(iocb, iovp, nr_segs, iocb->ki_pos);
346 347 348
	if (ret > 0)
		XFS_STATS_ADD(xs_read_bytes, ret);

349
	xfs_rw_iunlock(ip, XFS_IOLOCK_SHARED);
350 351 352
	return ret;
}

353 354
STATIC ssize_t
xfs_file_splice_read(
355 356 357 358
	struct file		*infilp,
	loff_t			*ppos,
	struct pipe_inode_info	*pipe,
	size_t			count,
359
	unsigned int		flags)
360
{
361 362
	struct xfs_inode	*ip = XFS_I(infilp->f_mapping->host);
	int			ioflags = 0;
363 364 365
	ssize_t			ret;

	XFS_STATS_INC(xs_read_calls);
366 367 368 369

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

370 371 372
	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
		return -EIO;

373
	xfs_rw_ilock(ip, XFS_IOLOCK_SHARED);
374 375 376 377 378 379 380

	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);

381
	xfs_rw_iunlock(ip, XFS_IOLOCK_SHARED);
382 383 384
	return ret;
}

385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401
STATIC void
xfs_aio_write_isize_update(
	struct inode	*inode,
	loff_t		*ppos,
	ssize_t		bytes_written)
{
	struct xfs_inode	*ip = XFS_I(inode);
	xfs_fsize_t		isize = i_size_read(inode);

	if (bytes_written > 0)
		XFS_STATS_ADD(xs_write_bytes, bytes_written);

	if (unlikely(bytes_written < 0 && bytes_written != -EFAULT &&
					*ppos > isize))
		*ppos = isize;

	if (*ppos > ip->i_size) {
402
		xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
403 404
		if (*ppos > ip->i_size)
			ip->i_size = *ppos;
405
		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
406 407 408
	}
}

409 410
/*
 * If this was a direct or synchronous I/O that failed (such as ENOSPC) then
L
Lucas De Marchi 已提交
411
 * part of the I/O may have been written to disk before the error occurred.  In
412 413 414 415 416
 * this case the on-disk file size may have been adjusted beyond the in-memory
 * file size and now needs to be truncated back.
 */
STATIC void
xfs_aio_write_newsize_update(
417 418
	struct xfs_inode	*ip,
	xfs_fsize_t		new_size)
419
{
420
	if (new_size == ip->i_new_size) {
421
		xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
422 423
		if (new_size == ip->i_new_size)
			ip->i_new_size = 0;
424 425
		if (ip->i_d.di_size > ip->i_size)
			ip->i_d.di_size = ip->i_size;
426
		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
427 428 429
	}
}

430 431 432 433 434 435 436 437
/*
 * xfs_file_splice_write() does not use xfs_rw_ilock() because
 * generic_file_splice_write() takes the i_mutex itself. This, in theory,
 * couuld cause lock inversions between the aio_write path and the splice path
 * if someone is doing concurrent splice(2) based writes and write(2) based
 * writes to the same inode. The only real way to fix this is to re-implement
 * the generic code here with correct locking orders.
 */
438 439
STATIC ssize_t
xfs_file_splice_write(
440 441 442 443
	struct pipe_inode_info	*pipe,
	struct file		*outfilp,
	loff_t			*ppos,
	size_t			count,
444
	unsigned int		flags)
445 446
{
	struct inode		*inode = outfilp->f_mapping->host;
447
	struct xfs_inode	*ip = XFS_I(inode);
448
	xfs_fsize_t		new_size;
449 450
	int			ioflags = 0;
	ssize_t			ret;
451 452

	XFS_STATS_INC(xs_write_calls);
453 454 455 456

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

457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472
	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
		return -EIO;

	xfs_ilock(ip, XFS_IOLOCK_EXCL);

	new_size = *ppos + count;

	xfs_ilock(ip, XFS_ILOCK_EXCL);
	if (new_size > ip->i_size)
		ip->i_new_size = new_size;
	xfs_iunlock(ip, XFS_ILOCK_EXCL);

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

	ret = generic_file_splice_write(pipe, outfilp, ppos, count, flags);

473
	xfs_aio_write_isize_update(inode, ppos, ret);
474
	xfs_aio_write_newsize_update(ip, new_size);
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
	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
	return ret;
}

/*
 * 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 anyone to read the garbage on the disk.
 */
STATIC int				/* error (positive) */
xfs_zero_last_block(
	xfs_inode_t	*ip,
	xfs_fsize_t	offset,
	xfs_fsize_t	isize)
{
	xfs_fileoff_t	last_fsb;
	xfs_mount_t	*mp = ip->i_mount;
	int		nimaps;
	int		zero_offset;
	int		zero_len;
	int		error = 0;
	xfs_bmbt_irec_t	imap;

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL));

	zero_offset = XFS_B_FSB_OFFSET(mp, isize);
	if (zero_offset == 0) {
		/*
		 * There are no extra bytes in the last block on disk to
		 * zero, so return.
		 */
		return 0;
	}

	last_fsb = XFS_B_TO_FSBT(mp, isize);
	nimaps = 1;
	error = xfs_bmapi(NULL, ip, last_fsb, 1, 0, NULL, 0, &imap,
513
			  &nimaps, NULL);
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
	if (error) {
		return error;
	}
	ASSERT(nimaps > 0);
	/*
	 * If the block underlying isize is just a hole, then there
	 * is nothing to zero.
	 */
	if (imap.br_startblock == HOLESTARTBLOCK) {
		return 0;
	}
	/*
	 * Zero the part of the last block beyond the EOF, and write it
	 * out sync.  We need to drop the ilock while we do this so we
	 * don't deadlock when the buffer cache calls back to us.
	 */
	xfs_iunlock(ip, XFS_ILOCK_EXCL);

	zero_len = mp->m_sb.sb_blocksize - zero_offset;
	if (isize + zero_len > offset)
		zero_len = offset - isize;
	error = xfs_iozero(ip, isize, zero_len);

	xfs_ilock(ip, XFS_ILOCK_EXCL);
	ASSERT(error >= 0);
	return error;
}

/*
 * 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.  If fill is set,
 * then any holes in the range are filled and zeroed.  If not, the holes
 * are left alone as holes.
 */

int					/* error (positive) */
xfs_zero_eof(
	xfs_inode_t	*ip,
	xfs_off_t	offset,		/* starting I/O offset */
	xfs_fsize_t	isize)		/* current inode size */
{
	xfs_mount_t	*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;
	xfs_bmbt_irec_t	imap;

	ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
	ASSERT(offset > isize);

	/*
	 * First handle zeroing the block on which isize resides.
	 * We only zero a part of that block so it is handled specially.
	 */
	error = xfs_zero_last_block(ip, offset, isize);
	if (error) {
		ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
		return error;
	}

	/*
	 * 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.
	 */
	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;
		error = xfs_bmapi(NULL, ip, start_zero_fsb, zero_count_fsb,
608
				  0, NULL, 0, &imap, &nimaps, NULL);
609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660
		if (error) {
			ASSERT(xfs_isilocked(ip, XFS_ILOCK_EXCL|XFS_IOLOCK_EXCL));
			return error;
		}
		ASSERT(nimaps > 0);

		if (imap.br_state == XFS_EXT_UNWRITTEN ||
		    imap.br_startblock == HOLESTARTBLOCK) {
			/*
			 * This loop handles initializing pages that were
			 * partially initialized by the code below this
			 * loop. It basically zeroes the part of the page
			 * that sits on a hole and sets the page as P_HOLE
			 * and calls remapf if it is a mapped file.
			 */
			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.
		 * Drop the inode lock while we're doing the I/O.
		 * We'll still have the iolock to protect us.
		 */
		xfs_iunlock(ip, XFS_ILOCK_EXCL);

		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);
		if (error) {
			goto out_lock;
		}

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

		xfs_ilock(ip, XFS_ILOCK_EXCL);
	}

	return 0;

out_lock:
	xfs_ilock(ip, XFS_ILOCK_EXCL);
	ASSERT(error >= 0);
	return error;
}

661 662 663 664 665 666 667 668 669 670
/*
 * Common pre-write limit and setup checks.
 *
 * Returns with iolock held according to @iolock.
 */
STATIC ssize_t
xfs_file_aio_write_checks(
	struct file		*file,
	loff_t			*pos,
	size_t			*count,
671
	xfs_fsize_t		*new_sizep,
672 673 674 675 676 677 678
	int			*iolock)
{
	struct inode		*inode = file->f_mapping->host;
	struct xfs_inode	*ip = XFS_I(inode);
	xfs_fsize_t		new_size;
	int			error = 0;

679
	xfs_rw_ilock(ip, XFS_ILOCK_EXCL);
680 681
	*new_sizep = 0;
restart:
682 683 684 685 686 687 688 689 690 691 692 693 694
	error = generic_write_checks(file, pos, count, S_ISBLK(inode->i_mode));
	if (error) {
		xfs_rw_iunlock(ip, XFS_ILOCK_EXCL | *iolock);
		*iolock = 0;
		return error;
	}

	if (likely(!(file->f_mode & FMODE_NOCMTIME)))
		file_update_time(file);

	/*
	 * 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
695 696 697 698 699 700
	 * write. There is no need to issue zeroing if another in-flght IO ends
	 * at or before this one If zeronig is needed and we are currently
	 * holding the iolock shared, we need to update it to exclusive which
	 * involves dropping all locks and relocking to maintain correct locking
	 * order. If we do this, restart the function to ensure all checks and
	 * values are still valid.
701
	 */
702 703 704 705 706 707 708 709
	if ((ip->i_new_size && *pos > ip->i_new_size) ||
	    (!ip->i_new_size && *pos > ip->i_size)) {
		if (*iolock == XFS_IOLOCK_SHARED) {
			xfs_rw_iunlock(ip, XFS_ILOCK_EXCL | *iolock);
			*iolock = XFS_IOLOCK_EXCL;
			xfs_rw_ilock(ip, XFS_ILOCK_EXCL | *iolock);
			goto restart;
		}
710
		error = -xfs_zero_eof(ip, *pos, ip->i_size);
711 712 713 714 715 716 717 718 719 720 721 722 723
	}

	/*
	 * If this IO extends beyond EOF, we may need to update ip->i_new_size.
	 * We have already zeroed space beyond EOF (if necessary).  Only update
	 * ip->i_new_size if this IO ends beyond any other in-flight writes.
	 */
	new_size = *pos + *count;
	if (new_size > ip->i_size) {
		if (new_size > ip->i_new_size)
			ip->i_new_size = new_size;
		*new_sizep = new_size;
	}
724 725 726 727 728 729 730 731 732 733 734 735 736 737

	xfs_rw_iunlock(ip, XFS_ILOCK_EXCL);
	if (error)
		return error;

	/*
	 * 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);

}

738 739 740 741
/*
 * xfs_file_dio_aio_write - handle direct IO writes
 *
 * Lock the inode appropriately to prepare for and issue a direct IO write.
742
 * By separating it from the buffered write path we remove all the tricky to
743 744
 * follow locking changes and looping.
 *
745 746 747 748 749 750 751 752 753 754 755 756 757
 * 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
758
 * hitting it with a big hammer (i.e. inode_dio_wait()).
759
 *
760 761 762 763 764 765 766 767 768 769
 * 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,
	const struct iovec	*iovp,
	unsigned long		nr_segs,
	loff_t			pos,
	size_t			ocount,
770
	xfs_fsize_t		*new_size,
771 772 773 774 775 776 777 778 779
	int			*iolock)
{
	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;
	size_t			count = ocount;
780
	int			unaligned_io = 0;
781 782 783 784 785 786 787
	struct xfs_buftarg	*target = XFS_IS_REALTIME_INODE(ip) ?
					mp->m_rtdev_targp : mp->m_ddev_targp;

	*iolock = 0;
	if ((pos & target->bt_smask) || (count & target->bt_smask))
		return -XFS_ERROR(EINVAL);

788 789 790
	if ((pos & mp->m_blockmask) || ((pos + count) & mp->m_blockmask))
		unaligned_io = 1;

791 792 793 794 795 796 797 798
	/*
	 * 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)
799 800 801
		*iolock = XFS_IOLOCK_EXCL;
	else
		*iolock = XFS_IOLOCK_SHARED;
802 803 804 805 806 807 808 809 810 811 812 813
	xfs_rw_ilock(ip, *iolock);

	/*
	 * 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.
	 */
	if (mapping->nrpages && *iolock == XFS_IOLOCK_SHARED) {
		xfs_rw_iunlock(ip, *iolock);
		*iolock = XFS_IOLOCK_EXCL;
		xfs_rw_ilock(ip, *iolock);
	}
814

815
	ret = xfs_file_aio_write_checks(file, &pos, &count, new_size, iolock);
816
	if (ret)
817 818 819 820 821 822 823 824 825
		return ret;

	if (mapping->nrpages) {
		ret = -xfs_flushinval_pages(ip, (pos & PAGE_CACHE_MASK), -1,
							FI_REMAPF_LOCKED);
		if (ret)
			return ret;
	}

826 827 828 829 830
	/*
	 * 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
831
		inode_dio_wait(inode);
832
	else if (*iolock == XFS_IOLOCK_EXCL) {
833 834 835 836 837 838 839 840 841 842 843 844 845
		xfs_rw_ilock_demote(ip, XFS_IOLOCK_EXCL);
		*iolock = XFS_IOLOCK_SHARED;
	}

	trace_xfs_file_direct_write(ip, count, iocb->ki_pos, 0);
	ret = generic_file_direct_write(iocb, iovp,
			&nr_segs, pos, &iocb->ki_pos, count, ocount);

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

846
STATIC ssize_t
847
xfs_file_buffered_aio_write(
848 849
	struct kiocb		*iocb,
	const struct iovec	*iovp,
850
	unsigned long		nr_segs,
851 852
	loff_t			pos,
	size_t			ocount,
853
	xfs_fsize_t		*new_size,
854
	int			*iolock)
855 856 857 858
{
	struct file		*file = iocb->ki_filp;
	struct address_space	*mapping = file->f_mapping;
	struct inode		*inode = mapping->host;
859
	struct xfs_inode	*ip = XFS_I(inode);
860 861 862
	ssize_t			ret;
	int			enospc = 0;
	size_t			count = ocount;
863

864
	*iolock = XFS_IOLOCK_EXCL;
865
	xfs_rw_ilock(ip, *iolock);
866

867
	ret = xfs_file_aio_write_checks(file, &pos, &count, new_size, iolock);
868
	if (ret)
869
		return ret;
870 871 872 873 874

	/* We can write back this queue in page reclaim */
	current->backing_dev_info = mapping->backing_dev_info;

write_retry:
875 876 877 878 879 880 881 882 883 884 885 886 887
	trace_xfs_file_buffered_write(ip, count, iocb->ki_pos, 0);
	ret = generic_file_buffered_write(iocb, iovp, nr_segs,
			pos, &iocb->ki_pos, count, ret);
	/*
	 * if we just got an ENOSPC, flush the inode now we aren't holding any
	 * page locks and retry *once*
	 */
	if (ret == -ENOSPC && !enospc) {
		ret = -xfs_flush_pages(ip, 0, -1, 0, FI_NONE);
		if (ret)
			return ret;
		enospc = 1;
		goto write_retry;
888 889
	}
	current->backing_dev_info = NULL;
890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906
	return ret;
}

STATIC ssize_t
xfs_file_aio_write(
	struct kiocb		*iocb,
	const struct iovec	*iovp,
	unsigned long		nr_segs,
	loff_t			pos)
{
	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;
	int			iolock;
	size_t			ocount = 0;
907
	xfs_fsize_t		new_size = 0;
908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926

	XFS_STATS_INC(xs_write_calls);

	BUG_ON(iocb->ki_pos != pos);

	ret = generic_segment_checks(iovp, &nr_segs, &ocount, VERIFY_READ);
	if (ret)
		return ret;

	if (ocount == 0)
		return 0;

	xfs_wait_for_freeze(ip->i_mount, SB_FREEZE_WRITE);

	if (XFS_FORCED_SHUTDOWN(ip->i_mount))
		return -EIO;

	if (unlikely(file->f_flags & O_DIRECT))
		ret = xfs_file_dio_aio_write(iocb, iovp, nr_segs, pos,
927
						ocount, &new_size, &iolock);
928 929
	else
		ret = xfs_file_buffered_aio_write(iocb, iovp, nr_segs, pos,
930
						ocount, &new_size, &iolock);
931

932
	xfs_aio_write_isize_update(inode, &iocb->ki_pos, ret);
933 934

	if (ret <= 0)
935
		goto out_unlock;
936 937 938 939

	/* Handle various SYNC-type writes */
	if ((file->f_flags & O_DSYNC) || IS_SYNC(inode)) {
		loff_t end = pos + ret - 1;
940
		int error;
941

942
		xfs_rw_iunlock(ip, iolock);
943
		error = xfs_file_fsync(file, pos, end,
944
				      (file->f_flags & __O_SYNC) ? 0 : 1);
945
		xfs_rw_ilock(ip, iolock);
946 947
		if (error)
			ret = error;
948 949
	}

950
out_unlock:
951
	xfs_aio_write_newsize_update(ip, new_size);
952
	xfs_rw_iunlock(ip, iolock);
953
	return ret;
954 955
}

956 957 958 959 960 961 962 963 964 965 966 967 968
STATIC long
xfs_file_fallocate(
	struct file	*file,
	int		mode,
	loff_t		offset,
	loff_t		len)
{
	struct inode	*inode = file->f_path.dentry->d_inode;
	long		error;
	loff_t		new_size = 0;
	xfs_flock64_t	bf;
	xfs_inode_t	*ip = XFS_I(inode);
	int		cmd = XFS_IOC_RESVSP;
969
	int		attr_flags = XFS_ATTR_NOLOCK;
970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991

	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE))
		return -EOPNOTSUPP;

	bf.l_whence = 0;
	bf.l_start = offset;
	bf.l_len = len;

	xfs_ilock(ip, XFS_IOLOCK_EXCL);

	if (mode & FALLOC_FL_PUNCH_HOLE)
		cmd = XFS_IOC_UNRESVSP;

	/* check the new inode size is valid before allocating */
	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;
	}

992 993 994 995
	if (file->f_flags & O_DSYNC)
		attr_flags |= XFS_ATTR_SYNC;

	error = -xfs_change_file_space(ip, cmd, &bf, 0, attr_flags);
996 997 998 999 1000 1001 1002 1003 1004
	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;
C
Christoph Hellwig 已提交
1005
		error = -xfs_setattr_size(ip, &iattr, XFS_ATTR_NOLOCK);
1006 1007 1008 1009 1010 1011 1012 1013
	}

out_unlock:
	xfs_iunlock(ip, XFS_IOLOCK_EXCL);
	return error;
}


L
Linus Torvalds 已提交
1014
STATIC int
1015
xfs_file_open(
L
Linus Torvalds 已提交
1016
	struct inode	*inode,
1017
	struct file	*file)
L
Linus Torvalds 已提交
1018
{
1019
	if (!(file->f_flags & O_LARGEFILE) && i_size_read(inode) > MAX_NON_LFS)
L
Linus Torvalds 已提交
1020
		return -EFBIG;
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
	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.
	 */
	mode = xfs_ilock_map_shared(ip);
	if (ip->i_d.di_nextents > 0)
		xfs_da_reada_buf(NULL, ip, 0, XFS_DATA_FORK);
	xfs_iunlock(ip, mode);
	return 0;
L
Linus Torvalds 已提交
1048 1049 1050
}

STATIC int
1051
xfs_file_release(
L
Linus Torvalds 已提交
1052 1053 1054
	struct inode	*inode,
	struct file	*filp)
{
1055
	return -xfs_release(XFS_I(inode));
L
Linus Torvalds 已提交
1056 1057 1058
}

STATIC int
1059
xfs_file_readdir(
L
Linus Torvalds 已提交
1060 1061 1062 1063
	struct file	*filp,
	void		*dirent,
	filldir_t	filldir)
{
C
Christoph Hellwig 已提交
1064
	struct inode	*inode = filp->f_path.dentry->d_inode;
1065
	xfs_inode_t	*ip = XFS_I(inode);
C
Christoph Hellwig 已提交
1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078
	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
1079
	 * buffer size.  For now we use the current glibc buffer size.
C
Christoph Hellwig 已提交
1080
	 */
E
Eric Sandeen 已提交
1081
	bufsize = (size_t)min_t(loff_t, 32768, ip->i_d.di_size);
C
Christoph Hellwig 已提交
1082

1083
	error = xfs_readdir(ip, dirent, bufsize,
C
Christoph Hellwig 已提交
1084 1085 1086 1087
				(xfs_off_t *)&filp->f_pos, filldir);
	if (error)
		return -error;
	return 0;
L
Linus Torvalds 已提交
1088 1089 1090
}

STATIC int
1091
xfs_file_mmap(
L
Linus Torvalds 已提交
1092 1093 1094
	struct file	*filp,
	struct vm_area_struct *vma)
{
1095
	vma->vm_ops = &xfs_file_vm_ops;
N
Nick Piggin 已提交
1096
	vma->vm_flags |= VM_CAN_NONLINEAR;
1097

1098
	file_accessed(filp);
L
Linus Torvalds 已提交
1099 1100 1101
	return 0;
}

1102 1103 1104 1105 1106 1107 1108 1109 1110
/*
 * 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,
1111
	struct vm_fault		*vmf)
1112
{
1113
	return block_page_mkwrite(vma, vmf, xfs_get_blocks);
1114 1115
}

1116
const struct file_operations xfs_file_operations = {
L
Linus Torvalds 已提交
1117 1118
	.llseek		= generic_file_llseek,
	.read		= do_sync_read,
1119
	.write		= do_sync_write,
1120 1121
	.aio_read	= xfs_file_aio_read,
	.aio_write	= xfs_file_aio_write,
1122 1123
	.splice_read	= xfs_file_splice_read,
	.splice_write	= xfs_file_splice_write,
1124
	.unlocked_ioctl	= xfs_file_ioctl,
L
Linus Torvalds 已提交
1125
#ifdef CONFIG_COMPAT
1126
	.compat_ioctl	= xfs_file_compat_ioctl,
L
Linus Torvalds 已提交
1127
#endif
1128 1129 1130 1131
	.mmap		= xfs_file_mmap,
	.open		= xfs_file_open,
	.release	= xfs_file_release,
	.fsync		= xfs_file_fsync,
1132
	.fallocate	= xfs_file_fallocate,
L
Linus Torvalds 已提交
1133 1134
};

1135
const struct file_operations xfs_dir_file_operations = {
1136
	.open		= xfs_dir_open,
L
Linus Torvalds 已提交
1137
	.read		= generic_read_dir,
1138
	.readdir	= xfs_file_readdir,
1139
	.llseek		= generic_file_llseek,
1140
	.unlocked_ioctl	= xfs_file_ioctl,
1141
#ifdef CONFIG_COMPAT
1142
	.compat_ioctl	= xfs_file_compat_ioctl,
1143
#endif
1144
	.fsync		= xfs_file_fsync,
L
Linus Torvalds 已提交
1145 1146
};

1147
static const struct vm_operations_struct xfs_file_vm_ops = {
1148
	.fault		= filemap_fault,
1149
	.page_mkwrite	= xfs_vm_page_mkwrite,
1150
};