file.c 18.7 KB
Newer Older
1
/*
2
 *  linux/fs/ext4/file.c
3 4 5 6 7 8 9 10 11 12 13 14
 *
 * Copyright (C) 1992, 1993, 1994, 1995
 * Remy Card (card@masi.ibp.fr)
 * Laboratoire MASI - Institut Blaise Pascal
 * Universite Pierre et Marie Curie (Paris VI)
 *
 *  from
 *
 *  linux/fs/minix/file.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *
15
 *  ext4 fs regular file handling primitives
16 17 18 19 20 21 22
 *
 *  64-bit file support on 64-bit platforms by Jakub Jelinek
 *	(jj@sunsite.ms.mff.cuni.cz)
 */

#include <linux/time.h>
#include <linux/fs.h>
23 24
#include <linux/mount.h>
#include <linux/path.h>
25
#include <linux/dax.h>
26
#include <linux/quotaops.h>
27
#include <linux/pagevec.h>
28
#include <linux/uio.h>
29 30
#include "ext4.h"
#include "ext4_jbd2.h"
31 32 33
#include "xattr.h"
#include "acl.h"

34 35 36 37 38 39
#ifdef CONFIG_FS_DAX
static ssize_t ext4_dax_read_iter(struct kiocb *iocb, struct iov_iter *to)
{
	struct inode *inode = file_inode(iocb->ki_filp);
	ssize_t ret;

G
Goldwyn Rodrigues 已提交
40 41 42 43 44
	if (!inode_trylock_shared(inode)) {
		if (iocb->ki_flags & IOCB_NOWAIT)
			return -EAGAIN;
		inode_lock_shared(inode);
	}
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63
	/*
	 * Recheck under inode lock - at this point we are sure it cannot
	 * change anymore
	 */
	if (!IS_DAX(inode)) {
		inode_unlock_shared(inode);
		/* Fallback to buffered IO in case we cannot support DAX */
		return generic_file_read_iter(iocb, to);
	}
	ret = dax_iomap_rw(iocb, to, &ext4_iomap_ops);
	inode_unlock_shared(inode);

	file_accessed(iocb->ki_filp);
	return ret;
}
#endif

static ssize_t ext4_file_read_iter(struct kiocb *iocb, struct iov_iter *to)
{
64 65 66
	if (unlikely(ext4_forced_shutdown(EXT4_SB(file_inode(iocb->ki_filp)->i_sb))))
		return -EIO;

67 68 69 70 71 72 73 74 75 76
	if (!iov_iter_count(to))
		return 0; /* skip atime */

#ifdef CONFIG_FS_DAX
	if (IS_DAX(file_inode(iocb->ki_filp)))
		return ext4_dax_read_iter(iocb, to);
#endif
	return generic_file_read_iter(iocb, to);
}

77 78
/*
 * Called when an inode is released. Note that this is different
79
 * from ext4_file_open: open gets called at every open, but release
80 81
 * gets called only when /all/ the files are closed.
 */
82
static int ext4_release_file(struct inode *inode, struct file *filp)
83
{
84
	if (ext4_test_inode_state(inode, EXT4_STATE_DA_ALLOC_CLOSE)) {
85
		ext4_alloc_da_blocks(inode);
86
		ext4_clear_inode_state(inode, EXT4_STATE_DA_ALLOC_CLOSE);
87
	}
88 89
	/* if we are the last writer on the inode, drop the block reservation */
	if ((filp->f_mode & FMODE_WRITE) &&
90 91
			(atomic_read(&inode->i_writecount) == 1) &&
		        !EXT4_I(inode)->i_reserved_data_blocks)
92
	{
93
		down_write(&EXT4_I(inode)->i_data_sem);
94
		ext4_discard_preallocations(inode);
95
		up_write(&EXT4_I(inode)->i_data_sem);
96 97
	}
	if (is_dx(inode) && filp->private_data)
98
		ext4_htree_free_dir_info(filp->private_data);
99 100 101 102

	return 0;
}

103
static void ext4_unwritten_wait(struct inode *inode)
104 105 106
{
	wait_queue_head_t *wq = ext4_ioend_wq(inode);

107
	wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_unwritten) == 0));
108 109 110 111 112 113 114 115 116 117 118 119
}

/*
 * This tests whether the IO in question is block-aligned or not.
 * Ext4 utilizes unwritten extents when hole-filling during direct IO, and they
 * are converted to written only after the IO is complete.  Until they are
 * mapped, these blocks appear as holes, so dio_zero_block() will assume that
 * it needs to zero out portions of the start and/or end block.  If 2 AIO
 * threads are at work on the same unwritten block, they must be synchronized
 * or one thread will zero the other's data, causing corruption.
 */
static int
A
Al Viro 已提交
120
ext4_unaligned_aio(struct inode *inode, struct iov_iter *from, loff_t pos)
121 122 123 124
{
	struct super_block *sb = inode->i_sb;
	int blockmask = sb->s_blocksize - 1;

125
	if (pos >= i_size_read(inode))
126 127
		return 0;

A
Al Viro 已提交
128
	if ((pos | iov_iter_alignment(from)) & blockmask)
129 130 131 132 133
		return 1;

	return 0;
}

134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178
/* Is IO overwriting allocated and initialized blocks? */
static bool ext4_overwrite_io(struct inode *inode, loff_t pos, loff_t len)
{
	struct ext4_map_blocks map;
	unsigned int blkbits = inode->i_blkbits;
	int err, blklen;

	if (pos + len > i_size_read(inode))
		return false;

	map.m_lblk = pos >> blkbits;
	map.m_len = EXT4_MAX_BLOCKS(len, pos, blkbits);
	blklen = map.m_len;

	err = ext4_map_blocks(NULL, inode, &map, 0);
	/*
	 * 'err==len' means that all of the blocks have been preallocated,
	 * regardless of whether they have been initialized or not. To exclude
	 * unwritten extents, we need to check m_flags.
	 */
	return err == blklen && (map.m_flags & EXT4_MAP_MAPPED);
}

static ssize_t ext4_write_checks(struct kiocb *iocb, struct iov_iter *from)
{
	struct inode *inode = file_inode(iocb->ki_filp);
	ssize_t ret;

	ret = generic_write_checks(iocb, from);
	if (ret <= 0)
		return ret;
	/*
	 * If we have encountered a bitmap-format file, the size limit
	 * is smaller than s_maxbytes, which is for extent-mapped files.
	 */
	if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))) {
		struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);

		if (iocb->ki_pos >= sbi->s_bitmap_maxbytes)
			return -EFBIG;
		iov_iter_truncate(from, sbi->s_bitmap_maxbytes - iocb->ki_pos);
	}
	return iov_iter_count(from);
}

J
Jan Kara 已提交
179 180 181 182 183 184 185
#ifdef CONFIG_FS_DAX
static ssize_t
ext4_dax_write_iter(struct kiocb *iocb, struct iov_iter *from)
{
	struct inode *inode = file_inode(iocb->ki_filp);
	ssize_t ret;

G
Goldwyn Rodrigues 已提交
186 187 188 189 190
	if (!inode_trylock(inode)) {
		if (iocb->ki_flags & IOCB_NOWAIT)
			return -EAGAIN;
		inode_lock(inode);
	}
J
Jan Kara 已提交
191 192 193 194 195 196 197 198 199 200 201 202
	ret = ext4_write_checks(iocb, from);
	if (ret <= 0)
		goto out;
	ret = file_remove_privs(iocb->ki_filp);
	if (ret)
		goto out;
	ret = file_update_time(iocb->ki_filp);
	if (ret)
		goto out;

	ret = dax_iomap_rw(iocb, from, &ext4_iomap_ops);
out:
C
Christoph Hellwig 已提交
203
	inode_unlock(inode);
J
Jan Kara 已提交
204 205 206 207 208 209
	if (ret > 0)
		ret = generic_write_sync(iocb, ret);
	return ret;
}
#endif

210
static ssize_t
A
Al Viro 已提交
211
ext4_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
212
{
213
	struct inode *inode = file_inode(iocb->ki_filp);
214
	int o_direct = iocb->ki_flags & IOCB_DIRECT;
215
	int unaligned_aio = 0;
216
	int overwrite = 0;
217
	ssize_t ret;
218

219 220 221
	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
		return -EIO;

J
Jan Kara 已提交
222 223 224 225 226
#ifdef CONFIG_FS_DAX
	if (IS_DAX(inode))
		return ext4_dax_write_iter(iocb, from);
#endif

G
Goldwyn Rodrigues 已提交
227 228 229 230 231 232
	if (!inode_trylock(inode)) {
		if (iocb->ki_flags & IOCB_NOWAIT)
			return -EAGAIN;
		inode_lock(inode);
	}

233
	ret = ext4_write_checks(iocb, from);
234 235 236
	if (ret <= 0)
		goto out;

237
	/*
238 239 240
	 * Unaligned direct AIO must be serialized among each other as zeroing
	 * of partial blocks of two competing unaligned AIOs can result in data
	 * corruption.
241
	 */
242
	if (o_direct && ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) &&
243
	    !is_sync_kiocb(iocb) &&
244 245
	    ext4_unaligned_aio(inode, from, iocb->ki_pos)) {
		unaligned_aio = 1;
246 247 248
		ext4_unwritten_wait(inode);
	}

249
	iocb->private = &overwrite;
250
	/* Check whether we do a DIO overwrite or not */
G
Goldwyn Rodrigues 已提交
251 252 253 254 255 256 257 258 259
	if (o_direct && !unaligned_aio) {
		if (ext4_overwrite_io(inode, iocb->ki_pos, iov_iter_count(from))) {
			if (ext4_should_dioread_nolock(inode))
				overwrite = 1;
		} else if (iocb->ki_flags & IOCB_NOWAIT) {
			ret = -EAGAIN;
			goto out;
		}
	}
260

A
Al Viro 已提交
261
	ret = __generic_file_write_iter(iocb, from);
A
Al Viro 已提交
262
	inode_unlock(inode);
263

264 265
	if (ret > 0)
		ret = generic_write_sync(iocb, ret);
266

267 268 269
	return ret;

out:
A
Al Viro 已提交
270
	inode_unlock(inode);
271
	return ret;
272 273
}

R
Ross Zwisler 已提交
274
#ifdef CONFIG_FS_DAX
275 276
static int ext4_dax_huge_fault(struct vm_fault *vmf,
		enum page_entry_size pe_size)
R
Ross Zwisler 已提交
277
{
278
	int result;
279
	handle_t *handle = NULL;
280
	struct inode *inode = file_inode(vmf->vma->vm_file);
281
	struct super_block *sb = inode->i_sb;
282 283 284 285
	bool write = vmf->flags & FAULT_FLAG_WRITE;

	if (write) {
		sb_start_pagefault(sb);
286
		file_update_time(vmf->vma->vm_file);
287 288 289 290 291
		down_read(&EXT4_I(inode)->i_mmap_sem);
		handle = ext4_journal_start_sb(sb, EXT4_HT_WRITE_PAGE,
					       EXT4_DATA_TRANS_BLOCKS(sb));
	} else {
		down_read(&EXT4_I(inode)->i_mmap_sem);
J
Jan Kara 已提交
292
	}
293 294 295 296 297 298 299 300
	if (!IS_ERR(handle))
		result = dax_iomap_fault(vmf, pe_size, &ext4_iomap_ops);
	else
		result = VM_FAULT_SIGBUS;
	if (write) {
		if (!IS_ERR(handle))
			ext4_journal_stop(handle);
		up_read(&EXT4_I(inode)->i_mmap_sem);
301
		sb_end_pagefault(sb);
302 303 304
	} else {
		up_read(&EXT4_I(inode)->i_mmap_sem);
	}
305 306

	return result;
R
Ross Zwisler 已提交
307 308
}

309 310 311 312 313
static int ext4_dax_fault(struct vm_fault *vmf)
{
	return ext4_dax_huge_fault(vmf, PE_SIZE_PTE);
}

314
/*
315
 * Handle write fault for VM_MIXEDMAP mappings. Similarly to ext4_dax_fault()
316 317 318 319 320 321 322
 * handler we check for races agaist truncate. Note that since we cycle through
 * i_mmap_sem, we are sure that also any hole punching that began before we
 * were called is finished by now and so if it included part of the file we
 * are working on, our pte will get unmapped and the check for pte_same() in
 * wp_pfn_shared() fails. Thus fault gets retried and things work out as
 * desired.
 */
323
static int ext4_dax_pfn_mkwrite(struct vm_fault *vmf)
324
{
325
	struct inode *inode = file_inode(vmf->vma->vm_file);
326 327
	struct super_block *sb = inode->i_sb;
	loff_t size;
328
	int ret;
329 330

	sb_start_pagefault(sb);
331
	file_update_time(vmf->vma->vm_file);
332 333 334 335
	down_read(&EXT4_I(inode)->i_mmap_sem);
	size = (i_size_read(inode) + PAGE_SIZE - 1) >> PAGE_SHIFT;
	if (vmf->pgoff >= size)
		ret = VM_FAULT_SIGBUS;
336
	else
337
		ret = dax_pfn_mkwrite(vmf);
338 339 340 341
	up_read(&EXT4_I(inode)->i_mmap_sem);
	sb_end_pagefault(sb);

	return ret;
R
Ross Zwisler 已提交
342 343 344 345
}

static const struct vm_operations_struct ext4_dax_vm_ops = {
	.fault		= ext4_dax_fault,
346
	.huge_fault	= ext4_dax_huge_fault,
347
	.page_mkwrite	= ext4_dax_fault,
348
	.pfn_mkwrite	= ext4_dax_pfn_mkwrite,
R
Ross Zwisler 已提交
349 350 351 352 353
};
#else
#define ext4_dax_vm_ops	ext4_file_vm_ops
#endif

354
static const struct vm_operations_struct ext4_file_vm_ops = {
355
	.fault		= ext4_filemap_fault,
356
	.map_pages	= filemap_map_pages,
357 358 359 360 361
	.page_mkwrite   = ext4_page_mkwrite,
};

static int ext4_file_mmap(struct file *file, struct vm_area_struct *vma)
{
362 363
	struct inode *inode = file->f_mapping->host;

364 365 366
	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
		return -EIO;

367
	file_accessed(file);
R
Ross Zwisler 已提交
368 369
	if (IS_DAX(file_inode(file))) {
		vma->vm_ops = &ext4_dax_vm_ops;
M
Matthew Wilcox 已提交
370
		vma->vm_flags |= VM_MIXEDMAP | VM_HUGEPAGE;
R
Ross Zwisler 已提交
371 372 373
	} else {
		vma->vm_ops = &ext4_file_vm_ops;
	}
374 375 376
	return 0;
}

377 378 379 380 381
static int ext4_file_open(struct inode * inode, struct file * filp)
{
	struct super_block *sb = inode->i_sb;
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	struct vfsmount *mnt = filp->f_path.mnt;
382
	struct dentry *dir;
383 384
	struct path path;
	char buf[64], *cp;
385
	int ret;
386

387 388 389
	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
		return -EIO;

390 391 392 393 394 395 396 397 398 399
	if (unlikely(!(sbi->s_mount_flags & EXT4_MF_MNTDIR_SAMPLED) &&
		     !(sb->s_flags & MS_RDONLY))) {
		sbi->s_mount_flags |= EXT4_MF_MNTDIR_SAMPLED;
		/*
		 * Sample where the filesystem has been mounted and
		 * store it in the superblock for sysadmin convenience
		 * when trying to sort through large numbers of block
		 * devices or filesystem images.
		 */
		memset(buf, 0, sizeof(buf));
400 401
		path.mnt = mnt;
		path.dentry = mnt->mnt_root;
402 403
		cp = d_path(&path, buf, sizeof(buf));
		if (!IS_ERR(cp)) {
404 405 406
			handle_t *handle;
			int err;

407
			handle = ext4_journal_start_sb(sb, EXT4_HT_MISC, 1);
408 409
			if (IS_ERR(handle))
				return PTR_ERR(handle);
410
			BUFFER_TRACE(sbi->s_sbh, "get_write_access");
411 412 413 414 415
			err = ext4_journal_get_write_access(handle, sbi->s_sbh);
			if (err) {
				ext4_journal_stop(handle);
				return err;
			}
416 417
			strlcpy(sbi->s_es->s_last_mounted, cp,
				sizeof(sbi->s_es->s_last_mounted));
418 419
			ext4_handle_dirty_super(handle, sb);
			ext4_journal_stop(handle);
420 421
		}
	}
422
	if (ext4_encrypted_inode(inode)) {
423
		ret = fscrypt_get_encryption_info(inode);
424 425
		if (ret)
			return -EACCES;
426
		if (!fscrypt_has_encryption_key(inode))
427 428
			return -ENOKEY;
	}
429

M
Miklos Szeredi 已提交
430
	dir = dget_parent(file_dentry(filp));
431
	if (ext4_encrypted_inode(d_inode(dir)) &&
432
			!fscrypt_has_permitted_context(d_inode(dir), inode)) {
433
		ext4_warning(inode->i_sb,
434
			     "Inconsistent encryption contexts: %lu/%lu",
435
			     (unsigned long) d_inode(dir)->i_ino,
436
			     (unsigned long) inode->i_ino);
437
		dput(dir);
438 439
		return -EPERM;
	}
440
	dput(dir);
441 442 443 444
	/*
	 * Set up the jbd2_inode if we are opening the inode for
	 * writing and the journal is present
	 */
445
	if (filp->f_mode & FMODE_WRITE) {
446
		ret = ext4_inode_attach_jinode(inode);
447 448
		if (ret < 0)
			return ret;
449
	}
G
Goldwyn Rodrigues 已提交
450 451 452 453

	/* Set the flags to support nowait AIO */
	filp->f_mode |= FMODE_AIO_NOWAIT;

454
	return dquot_file_open(inode, filp);
455 456
}

457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472
/*
 * Here we use ext4_map_blocks() to get a block mapping for a extent-based
 * file rather than ext4_ext_walk_space() because we can introduce
 * SEEK_DATA/SEEK_HOLE for block-mapped and extent-mapped file at the same
 * function.  When extent status tree has been fully implemented, it will
 * track all extent status for a file and we can directly use it to
 * retrieve the offset for SEEK_DATA/SEEK_HOLE.
 */

/*
 * When we retrieve the offset for SEEK_DATA/SEEK_HOLE, we would need to
 * lookup page cache to check whether or not there has some data between
 * [startoff, endoff] because, if this range contains an unwritten extent,
 * we determine this extent as a data or a hole according to whether the
 * page cache has data or not.
 */
473 474
static int ext4_find_unwritten_pgoff(struct inode *inode,
				     int whence,
475
				     ext4_lblk_t end_blk,
476
				     loff_t *offset)
477 478
{
	struct pagevec pvec;
479
	unsigned int blkbits;
480 481
	pgoff_t index;
	pgoff_t end;
482
	loff_t endoff;
483 484 485 486
	loff_t startoff;
	loff_t lastoff;
	int found = 0;

487
	blkbits = inode->i_sb->s_blocksize_bits;
488 489
	startoff = *offset;
	lastoff = startoff;
490
	endoff = (loff_t)end_blk << blkbits;
491

492
	index = startoff >> PAGE_SHIFT;
493
	end = (endoff - 1) >> PAGE_SHIFT;
494 495 496 497 498 499

	pagevec_init(&pvec, 0);
	do {
		int i, num;
		unsigned long nr_pages;

500
		num = min_t(pgoff_t, end - index, PAGEVEC_SIZE - 1) + 1;
501 502
		nr_pages = pagevec_lookup(&pvec, inode->i_mapping, index,
					  (pgoff_t)num);
J
Jan Kara 已提交
503
		if (nr_pages == 0)
504 505 506 507 508 509 510
			break;

		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];
			struct buffer_head *bh, *head;

			/*
J
Jan Kara 已提交
511 512
			 * If current offset is smaller than the page offset,
			 * there is a hole at this offset.
513
			 */
J
Jan Kara 已提交
514 515
			if (whence == SEEK_HOLE && lastoff < endoff &&
			    lastoff < page_offset(pvec.pages[i])) {
516 517 518 519 520
				found = 1;
				*offset = lastoff;
				goto out;
			}

J
Jan Kara 已提交
521 522 523
			if (page->index > end)
				goto out;

524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539
			lock_page(page);

			if (unlikely(page->mapping != inode->i_mapping)) {
				unlock_page(page);
				continue;
			}

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

			if (page_has_buffers(page)) {
				lastoff = page_offset(page);
				bh = head = page_buffers(page);
				do {
540 541
					if (lastoff + bh->b_size <= startoff)
						goto next;
542 543
					if (buffer_uptodate(bh) ||
					    buffer_unwritten(bh)) {
544
						if (whence == SEEK_DATA)
545 546
							found = 1;
					} else {
547
						if (whence == SEEK_HOLE)
548 549 550 551 552 553 554 555
							found = 1;
					}
					if (found) {
						*offset = max_t(loff_t,
							startoff, lastoff);
						unlock_page(page);
						goto out;
					}
556
next:
557 558 559 560 561 562 563 564 565
					lastoff += bh->b_size;
					bh = bh->b_this_page;
				} while (bh != head);
			}

			lastoff = page_offset(page) + PAGE_SIZE;
			unlock_page(page);
		}

J
Jan Kara 已提交
566 567
		/* The no. of pages is less than our desired, we are done. */
		if (nr_pages < num)
568 569 570 571 572 573
			break;

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

J
Jan Kara 已提交
574 575 576 577
	if (whence == SEEK_HOLE && lastoff < endoff) {
		found = 1;
		*offset = lastoff;
	}
578 579 580 581 582 583 584 585 586 587 588
out:
	pagevec_release(&pvec);
	return found;
}

/*
 * ext4_seek_data() retrieves the offset for SEEK_DATA.
 */
static loff_t ext4_seek_data(struct file *file, loff_t offset, loff_t maxsize)
{
	struct inode *inode = file->f_mapping->host;
589 590 591 592
	struct extent_status es;
	ext4_lblk_t start, last, end;
	loff_t dataoff, isize;
	int blkbits;
593
	int ret;
594

A
Al Viro 已提交
595
	inode_lock(inode);
596 597 598

	isize = i_size_read(inode);
	if (offset >= isize) {
A
Al Viro 已提交
599
		inode_unlock(inode);
600 601
		return -ENXIO;
	}
602 603 604 605 606 607 608 609

	blkbits = inode->i_sb->s_blocksize_bits;
	start = offset >> blkbits;
	last = start;
	end = isize >> blkbits;
	dataoff = offset;

	do {
610 611 612 613 614 615 616
		ret = ext4_get_next_extent(inode, last, end - last + 1, &es);
		if (ret <= 0) {
			/* No extent found -> no data */
			if (ret == 0)
				ret = -ENXIO;
			inode_unlock(inode);
			return ret;
617
		}
618

619 620 621 622
		last = es.es_lblk;
		if (last != start)
			dataoff = (loff_t)last << blkbits;
		if (!ext4_es_is_unwritten(&es))
623 624
			break;

625 626 627 628 629
		/*
		 * If there is a unwritten extent at this offset,
		 * it will be as a data or a hole according to page
		 * cache that has data or not.
		 */
630 631 632 633
		if (ext4_find_unwritten_pgoff(inode, SEEK_DATA,
					      es.es_lblk + es.es_len, &dataoff))
			break;
		last += es.es_len;
634
		dataoff = (loff_t)last << blkbits;
635
		cond_resched();
636
	} while (last <= end);
637

A
Al Viro 已提交
638
	inode_unlock(inode);
639

640 641 642 643
	if (dataoff > isize)
		return -ENXIO;

	return vfs_setpos(file, dataoff, maxsize);
644 645 646
}

/*
647
 * ext4_seek_hole() retrieves the offset for SEEK_HOLE.
648 649 650 651
 */
static loff_t ext4_seek_hole(struct file *file, loff_t offset, loff_t maxsize)
{
	struct inode *inode = file->f_mapping->host;
652 653 654 655
	struct extent_status es;
	ext4_lblk_t start, last, end;
	loff_t holeoff, isize;
	int blkbits;
656
	int ret;
657

A
Al Viro 已提交
658
	inode_lock(inode);
659 660 661

	isize = i_size_read(inode);
	if (offset >= isize) {
A
Al Viro 已提交
662
		inode_unlock(inode);
663 664 665
		return -ENXIO;
	}

666 667 668 669 670
	blkbits = inode->i_sb->s_blocksize_bits;
	start = offset >> blkbits;
	last = start;
	end = isize >> blkbits;
	holeoff = offset;
671

672
	do {
673 674 675 676
		ret = ext4_get_next_extent(inode, last, end - last + 1, &es);
		if (ret < 0) {
			inode_unlock(inode);
			return ret;
677
		}
678 679 680 681 682
		/* Found a hole? */
		if (ret == 0 || es.es_lblk > last) {
			if (last != start)
				holeoff = (loff_t)last << blkbits;
			break;
683 684 685 686 687 688
		}
		/*
		 * If there is a unwritten extent at this offset,
		 * it will be as a data or a hole according to page
		 * cache that has data or not.
		 */
689 690 691 692
		if (ext4_es_is_unwritten(&es) &&
		    ext4_find_unwritten_pgoff(inode, SEEK_HOLE,
					      last + es.es_len, &holeoff))
			break;
693

694 695 696
		last += es.es_len;
		holeoff = (loff_t)last << blkbits;
		cond_resched();
697 698
	} while (last <= end);

A
Al Viro 已提交
699
	inode_unlock(inode);
700

701 702 703 704
	if (holeoff > isize)
		holeoff = isize;

	return vfs_setpos(file, holeoff, maxsize);
705 706
}

707
/*
708 709 710
 * ext4_llseek() handles both block-mapped and extent-mapped maxbytes values
 * by calling generic_file_llseek_size() with the appropriate maxbytes
 * value for each.
711
 */
712
loff_t ext4_llseek(struct file *file, loff_t offset, int whence)
713 714 715 716 717 718 719 720 721
{
	struct inode *inode = file->f_mapping->host;
	loff_t maxbytes;

	if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
		maxbytes = EXT4_SB(inode->i_sb)->s_bitmap_maxbytes;
	else
		maxbytes = inode->i_sb->s_maxbytes;

722
	switch (whence) {
723 724 725
	case SEEK_SET:
	case SEEK_CUR:
	case SEEK_END:
726
		return generic_file_llseek_size(file, offset, whence,
727 728 729 730 731 732 733 734
						maxbytes, i_size_read(inode));
	case SEEK_DATA:
		return ext4_seek_data(file, offset, maxbytes);
	case SEEK_HOLE:
		return ext4_seek_hole(file, offset, maxbytes);
	}

	return -EINVAL;
735 736
}

737
const struct file_operations ext4_file_operations = {
738
	.llseek		= ext4_llseek,
739
	.read_iter	= ext4_file_read_iter,
A
Al Viro 已提交
740
	.write_iter	= ext4_file_write_iter,
A
Andi Kleen 已提交
741
	.unlocked_ioctl = ext4_ioctl,
742
#ifdef CONFIG_COMPAT
743
	.compat_ioctl	= ext4_compat_ioctl,
744
#endif
745
	.mmap		= ext4_file_mmap,
746
	.open		= ext4_file_open,
747 748
	.release	= ext4_release_file,
	.fsync		= ext4_sync_file,
749
	.get_unmapped_area = thp_get_unmapped_area,
750
	.splice_read	= generic_file_splice_read,
A
Al Viro 已提交
751
	.splice_write	= iter_file_splice_write,
752
	.fallocate	= ext4_fallocate,
753 754
};

755
const struct inode_operations ext4_file_inode_operations = {
756
	.setattr	= ext4_setattr,
D
David Howells 已提交
757
	.getattr	= ext4_file_getattr,
758
	.listxattr	= ext4_listxattr,
759
	.get_acl	= ext4_get_acl,
760
	.set_acl	= ext4_set_acl,
761
	.fiemap		= ext4_fiemap,
762 763
};