file.c 19.5 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 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69
#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;

	inode_lock_shared(inode);
	/*
	 * 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)
{
	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);
}

70 71
/*
 * Called when an inode is released. Note that this is different
72
 * from ext4_file_open: open gets called at every open, but release
73 74
 * gets called only when /all/ the files are closed.
 */
75
static int ext4_release_file(struct inode *inode, struct file *filp)
76
{
77
	if (ext4_test_inode_state(inode, EXT4_STATE_DA_ALLOC_CLOSE)) {
78
		ext4_alloc_da_blocks(inode);
79
		ext4_clear_inode_state(inode, EXT4_STATE_DA_ALLOC_CLOSE);
80
	}
81 82
	/* if we are the last writer on the inode, drop the block reservation */
	if ((filp->f_mode & FMODE_WRITE) &&
83 84
			(atomic_read(&inode->i_writecount) == 1) &&
		        !EXT4_I(inode)->i_reserved_data_blocks)
85
	{
86
		down_write(&EXT4_I(inode)->i_data_sem);
87
		ext4_discard_preallocations(inode);
88
		up_write(&EXT4_I(inode)->i_data_sem);
89 90
	}
	if (is_dx(inode) && filp->private_data)
91
		ext4_htree_free_dir_info(filp->private_data);
92 93 94 95

	return 0;
}

96
static void ext4_unwritten_wait(struct inode *inode)
97 98 99
{
	wait_queue_head_t *wq = ext4_ioend_wq(inode);

100
	wait_event(*wq, (atomic_read(&EXT4_I(inode)->i_unwritten) == 0));
101 102 103 104 105 106 107 108 109 110 111 112
}

/*
 * 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
113
ext4_unaligned_aio(struct inode *inode, struct iov_iter *from, loff_t pos)
114 115 116 117
{
	struct super_block *sb = inode->i_sb;
	int blockmask = sb->s_blocksize - 1;

118
	if (pos >= i_size_read(inode))
119 120
		return 0;

A
Al Viro 已提交
121
	if ((pos | iov_iter_alignment(from)) & blockmask)
122 123 124 125 126
		return 1;

	return 0;
}

127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171
/* 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
172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206
#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;
	bool overwrite = false;

	inode_lock(inode);
	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;

	if (ext4_overwrite_io(inode, iocb->ki_pos, iov_iter_count(from))) {
		overwrite = true;
		downgrade_write(&inode->i_rwsem);
	}
	ret = dax_iomap_rw(iocb, from, &ext4_iomap_ops);
out:
	if (!overwrite)
		inode_unlock(inode);
	else
		inode_unlock_shared(inode);
	if (ret > 0)
		ret = generic_write_sync(iocb, ret);
	return ret;
}
#endif

207
static ssize_t
A
Al Viro 已提交
208
ext4_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
209
{
210
	struct inode *inode = file_inode(iocb->ki_filp);
211
	int o_direct = iocb->ki_flags & IOCB_DIRECT;
212
	int unaligned_aio = 0;
213
	int overwrite = 0;
214
	ssize_t ret;
215

J
Jan Kara 已提交
216 217 218 219 220
#ifdef CONFIG_FS_DAX
	if (IS_DAX(inode))
		return ext4_dax_write_iter(iocb, from);
#endif

221
	inode_lock(inode);
222
	ret = ext4_write_checks(iocb, from);
223 224 225
	if (ret <= 0)
		goto out;

226
	/*
227 228 229
	 * Unaligned direct AIO must be serialized among each other as zeroing
	 * of partial blocks of two competing unaligned AIOs can result in data
	 * corruption.
230
	 */
231
	if (o_direct && ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) &&
232
	    !is_sync_kiocb(iocb) &&
233 234
	    ext4_unaligned_aio(inode, from, iocb->ki_pos)) {
		unaligned_aio = 1;
235 236 237
		ext4_unwritten_wait(inode);
	}

238
	iocb->private = &overwrite;
239 240 241 242
	/* Check whether we do a DIO overwrite or not */
	if (o_direct && ext4_should_dioread_nolock(inode) && !unaligned_aio &&
	    ext4_overwrite_io(inode, iocb->ki_pos, iov_iter_count(from)))
		overwrite = 1;
243

A
Al Viro 已提交
244
	ret = __generic_file_write_iter(iocb, from);
A
Al Viro 已提交
245
	inode_unlock(inode);
246

247 248
	if (ret > 0)
		ret = generic_write_sync(iocb, ret);
249

250 251 252
	return ret;

out:
A
Al Viro 已提交
253
	inode_unlock(inode);
254
	return ret;
255 256
}

R
Ross Zwisler 已提交
257 258 259
#ifdef CONFIG_FS_DAX
static int ext4_dax_fault(struct vm_area_struct *vma, struct vm_fault *vmf)
{
260 261
	int result;
	handle_t *handle = NULL;
262 263
	struct inode *inode = file_inode(vma->vm_file);
	struct super_block *sb = inode->i_sb;
264 265 266 267 268
	bool write = vmf->flags & FAULT_FLAG_WRITE;

	if (write) {
		sb_start_pagefault(sb);
		file_update_time(vma->vm_file);
269
		down_read(&EXT4_I(inode)->i_mmap_sem);
270 271
		handle = ext4_journal_start_sb(sb, EXT4_HT_WRITE_PAGE,
						EXT4_DATA_TRANS_BLOCKS(sb));
272 273
	} else
		down_read(&EXT4_I(inode)->i_mmap_sem);
274 275 276 277

	if (IS_ERR(handle))
		result = VM_FAULT_SIGBUS;
	else
278
		result = dax_iomap_fault(vma, vmf, &ext4_iomap_ops);
279 280 281 282

	if (write) {
		if (!IS_ERR(handle))
			ext4_journal_stop(handle);
283
		up_read(&EXT4_I(inode)->i_mmap_sem);
284
		sb_end_pagefault(sb);
285 286
	} else
		up_read(&EXT4_I(inode)->i_mmap_sem);
287 288

	return result;
R
Ross Zwisler 已提交
289 290
}

M
Matthew Wilcox 已提交
291 292 293
static int ext4_dax_pmd_fault(struct vm_area_struct *vma, unsigned long addr,
						pmd_t *pmd, unsigned int flags)
{
294 295 296 297 298 299 300 301 302
	int result;
	handle_t *handle = NULL;
	struct inode *inode = file_inode(vma->vm_file);
	struct super_block *sb = inode->i_sb;
	bool write = flags & FAULT_FLAG_WRITE;

	if (write) {
		sb_start_pagefault(sb);
		file_update_time(vma->vm_file);
303
		down_read(&EXT4_I(inode)->i_mmap_sem);
304 305 306
		handle = ext4_journal_start_sb(sb, EXT4_HT_WRITE_PAGE,
				ext4_chunk_trans_blocks(inode,
							PMD_SIZE / PAGE_SIZE));
307 308
	} else
		down_read(&EXT4_I(inode)->i_mmap_sem);
309 310 311

	if (IS_ERR(handle))
		result = VM_FAULT_SIGBUS;
312 313 314 315
	else {
		result = dax_iomap_pmd_fault(vma, addr, pmd, flags,
					     &ext4_iomap_ops);
	}
316 317 318 319

	if (write) {
		if (!IS_ERR(handle))
			ext4_journal_stop(handle);
320
		up_read(&EXT4_I(inode)->i_mmap_sem);
321
		sb_end_pagefault(sb);
322 323
	} else
		up_read(&EXT4_I(inode)->i_mmap_sem);
324 325

	return result;
M
Matthew Wilcox 已提交
326 327
}

328
/*
329
 * Handle write fault for VM_MIXEDMAP mappings. Similarly to ext4_dax_fault()
330 331 332 333 334 335 336 337 338 339 340 341 342
 * 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.
 */
static int ext4_dax_pfn_mkwrite(struct vm_area_struct *vma,
				struct vm_fault *vmf)
{
	struct inode *inode = file_inode(vma->vm_file);
	struct super_block *sb = inode->i_sb;
	loff_t size;
343
	int ret;
344 345 346 347 348 349 350

	sb_start_pagefault(sb);
	file_update_time(vma->vm_file);
	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;
351 352
	else
		ret = dax_pfn_mkwrite(vma, vmf);
353 354 355 356
	up_read(&EXT4_I(inode)->i_mmap_sem);
	sb_end_pagefault(sb);

	return ret;
R
Ross Zwisler 已提交
357 358 359 360
}

static const struct vm_operations_struct ext4_dax_vm_ops = {
	.fault		= ext4_dax_fault,
M
Matthew Wilcox 已提交
361
	.pmd_fault	= ext4_dax_pmd_fault,
362
	.page_mkwrite	= ext4_dax_fault,
363
	.pfn_mkwrite	= ext4_dax_pfn_mkwrite,
R
Ross Zwisler 已提交
364 365 366 367 368
};
#else
#define ext4_dax_vm_ops	ext4_file_vm_ops
#endif

369
static const struct vm_operations_struct ext4_file_vm_ops = {
370
	.fault		= ext4_filemap_fault,
371
	.map_pages	= filemap_map_pages,
372 373 374 375 376
	.page_mkwrite   = ext4_page_mkwrite,
};

static int ext4_file_mmap(struct file *file, struct vm_area_struct *vma)
{
377 378 379
	struct inode *inode = file->f_mapping->host;

	if (ext4_encrypted_inode(inode)) {
380
		int err = fscrypt_get_encryption_info(inode);
381 382
		if (err)
			return 0;
383
		if (!fscrypt_has_encryption_key(inode))
384
			return -ENOKEY;
385
	}
386
	file_accessed(file);
R
Ross Zwisler 已提交
387 388
	if (IS_DAX(file_inode(file))) {
		vma->vm_ops = &ext4_dax_vm_ops;
M
Matthew Wilcox 已提交
389
		vma->vm_flags |= VM_MIXEDMAP | VM_HUGEPAGE;
R
Ross Zwisler 已提交
390 391 392
	} else {
		vma->vm_ops = &ext4_file_vm_ops;
	}
393 394 395
	return 0;
}

396 397 398 399 400
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;
401
	struct dentry *dir;
402 403
	struct path path;
	char buf[64], *cp;
404
	int ret;
405 406 407 408 409 410 411 412 413 414 415

	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));
416 417
		path.mnt = mnt;
		path.dentry = mnt->mnt_root;
418 419
		cp = d_path(&path, buf, sizeof(buf));
		if (!IS_ERR(cp)) {
420 421 422
			handle_t *handle;
			int err;

423
			handle = ext4_journal_start_sb(sb, EXT4_HT_MISC, 1);
424 425
			if (IS_ERR(handle))
				return PTR_ERR(handle);
426
			BUFFER_TRACE(sbi->s_sbh, "get_write_access");
427 428 429 430 431
			err = ext4_journal_get_write_access(handle, sbi->s_sbh);
			if (err) {
				ext4_journal_stop(handle);
				return err;
			}
432 433
			strlcpy(sbi->s_es->s_last_mounted, cp,
				sizeof(sbi->s_es->s_last_mounted));
434 435
			ext4_handle_dirty_super(handle, sb);
			ext4_journal_stop(handle);
436 437
		}
	}
438
	if (ext4_encrypted_inode(inode)) {
439
		ret = fscrypt_get_encryption_info(inode);
440 441
		if (ret)
			return -EACCES;
442
		if (!fscrypt_has_encryption_key(inode))
443 444
			return -ENOKEY;
	}
445

M
Miklos Szeredi 已提交
446
	dir = dget_parent(file_dentry(filp));
447
	if (ext4_encrypted_inode(d_inode(dir)) &&
448
			!fscrypt_has_permitted_context(d_inode(dir), inode)) {
449
		ext4_warning(inode->i_sb,
450
			     "Inconsistent encryption contexts: %lu/%lu",
451
			     (unsigned long) d_inode(dir)->i_ino,
452
			     (unsigned long) inode->i_ino);
453
		dput(dir);
454 455
		return -EPERM;
	}
456
	dput(dir);
457 458 459 460
	/*
	 * Set up the jbd2_inode if we are opening the inode for
	 * writing and the journal is present
	 */
461
	if (filp->f_mode & FMODE_WRITE) {
462
		ret = ext4_inode_attach_jinode(inode);
463 464
		if (ret < 0)
			return ret;
465
	}
466
	return dquot_file_open(inode, filp);
467 468
}

469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484
/*
 * 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.
 */
485 486
static int ext4_find_unwritten_pgoff(struct inode *inode,
				     int whence,
487
				     ext4_lblk_t end_blk,
488
				     loff_t *offset)
489 490
{
	struct pagevec pvec;
491
	unsigned int blkbits;
492 493
	pgoff_t index;
	pgoff_t end;
494
	loff_t endoff;
495 496 497 498
	loff_t startoff;
	loff_t lastoff;
	int found = 0;

499
	blkbits = inode->i_sb->s_blocksize_bits;
500 501
	startoff = *offset;
	lastoff = startoff;
502
	endoff = (loff_t)end_blk << blkbits;
503

504 505
	index = startoff >> PAGE_SHIFT;
	end = endoff >> PAGE_SHIFT;
506 507 508 509 510 511 512 513 514 515

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

		num = min_t(pgoff_t, end - index, PAGEVEC_SIZE);
		nr_pages = pagevec_lookup(&pvec, inode->i_mapping, index,
					  (pgoff_t)num);
		if (nr_pages == 0) {
516
			if (whence == SEEK_DATA)
517 518
				break;

519
			BUG_ON(whence != SEEK_HOLE);
520 521 522 523 524 525 526 527 528 529 530 531 532 533 534
			/*
			 * If this is the first time to go into the loop and
			 * offset is not beyond the end offset, it will be a
			 * hole at this offset
			 */
			if (lastoff == startoff || lastoff < endoff)
				found = 1;
			break;
		}

		/*
		 * If this is the first time to go into the loop and
		 * offset is smaller than the first page offset, it will be a
		 * hole at this offset.
		 */
535
		if (lastoff == startoff && whence == SEEK_HOLE &&
536 537 538 539 540 541 542 543 544 545 546 547 548
		    lastoff < page_offset(pvec.pages[0])) {
			found = 1;
			break;
		}

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

			/*
			 * If the current offset is not beyond the end of given
			 * range, it will be a hole.
			 */
549
			if (lastoff < endoff && whence == SEEK_HOLE &&
550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573
			    page->index > end) {
				found = 1;
				*offset = lastoff;
				goto out;
			}

			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 {
					if (buffer_uptodate(bh) ||
					    buffer_unwritten(bh)) {
574
						if (whence == SEEK_DATA)
575 576
							found = 1;
					} else {
577
						if (whence == SEEK_HOLE)
578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598
							found = 1;
					}
					if (found) {
						*offset = max_t(loff_t,
							startoff, lastoff);
						unlock_page(page);
						goto out;
					}
					lastoff += bh->b_size;
					bh = bh->b_this_page;
				} while (bh != head);
			}

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

		/*
		 * The no. of pages is less than our desired, that would be a
		 * hole in there.
		 */
599
		if (nr_pages < num && whence == SEEK_HOLE) {
600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619
			found = 1;
			*offset = lastoff;
			break;
		}

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

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;
620 621 622 623
	struct extent_status es;
	ext4_lblk_t start, last, end;
	loff_t dataoff, isize;
	int blkbits;
624
	int ret;
625

A
Al Viro 已提交
626
	inode_lock(inode);
627 628 629

	isize = i_size_read(inode);
	if (offset >= isize) {
A
Al Viro 已提交
630
		inode_unlock(inode);
631 632
		return -ENXIO;
	}
633 634 635 636 637 638 639 640

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

	do {
641 642 643 644 645 646 647
		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;
648
		}
649

650 651 652 653
		last = es.es_lblk;
		if (last != start)
			dataoff = (loff_t)last << blkbits;
		if (!ext4_es_is_unwritten(&es))
654 655
			break;

656 657 658 659 660
		/*
		 * 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.
		 */
661 662 663 664
		if (ext4_find_unwritten_pgoff(inode, SEEK_DATA,
					      es.es_lblk + es.es_len, &dataoff))
			break;
		last += es.es_len;
665
		dataoff = (loff_t)last << blkbits;
666
		cond_resched();
667
	} while (last <= end);
668

A
Al Viro 已提交
669
	inode_unlock(inode);
670

671 672 673 674
	if (dataoff > isize)
		return -ENXIO;

	return vfs_setpos(file, dataoff, maxsize);
675 676 677
}

/*
678
 * ext4_seek_hole() retrieves the offset for SEEK_HOLE.
679 680 681 682
 */
static loff_t ext4_seek_hole(struct file *file, loff_t offset, loff_t maxsize)
{
	struct inode *inode = file->f_mapping->host;
683 684 685 686
	struct extent_status es;
	ext4_lblk_t start, last, end;
	loff_t holeoff, isize;
	int blkbits;
687
	int ret;
688

A
Al Viro 已提交
689
	inode_lock(inode);
690 691 692

	isize = i_size_read(inode);
	if (offset >= isize) {
A
Al Viro 已提交
693
		inode_unlock(inode);
694 695 696
		return -ENXIO;
	}

697 698 699 700 701
	blkbits = inode->i_sb->s_blocksize_bits;
	start = offset >> blkbits;
	last = start;
	end = isize >> blkbits;
	holeoff = offset;
702

703
	do {
704 705 706 707
		ret = ext4_get_next_extent(inode, last, end - last + 1, &es);
		if (ret < 0) {
			inode_unlock(inode);
			return ret;
708
		}
709 710 711 712 713
		/* Found a hole? */
		if (ret == 0 || es.es_lblk > last) {
			if (last != start)
				holeoff = (loff_t)last << blkbits;
			break;
714 715 716 717 718 719
		}
		/*
		 * 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.
		 */
720 721 722 723
		if (ext4_es_is_unwritten(&es) &&
		    ext4_find_unwritten_pgoff(inode, SEEK_HOLE,
					      last + es.es_len, &holeoff))
			break;
724

725 726 727
		last += es.es_len;
		holeoff = (loff_t)last << blkbits;
		cond_resched();
728 729
	} while (last <= end);

A
Al Viro 已提交
730
	inode_unlock(inode);
731

732 733 734 735
	if (holeoff > isize)
		holeoff = isize;

	return vfs_setpos(file, holeoff, maxsize);
736 737
}

738
/*
739 740 741
 * ext4_llseek() handles both block-mapped and extent-mapped maxbytes values
 * by calling generic_file_llseek_size() with the appropriate maxbytes
 * value for each.
742
 */
743
loff_t ext4_llseek(struct file *file, loff_t offset, int whence)
744 745 746 747 748 749 750 751 752
{
	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;

753
	switch (whence) {
754 755 756
	case SEEK_SET:
	case SEEK_CUR:
	case SEEK_END:
757
		return generic_file_llseek_size(file, offset, whence,
758 759 760 761 762 763 764 765
						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;
766 767
}

768
const struct file_operations ext4_file_operations = {
769
	.llseek		= ext4_llseek,
770
	.read_iter	= ext4_file_read_iter,
A
Al Viro 已提交
771
	.write_iter	= ext4_file_write_iter,
A
Andi Kleen 已提交
772
	.unlocked_ioctl = ext4_ioctl,
773
#ifdef CONFIG_COMPAT
774
	.compat_ioctl	= ext4_compat_ioctl,
775
#endif
776
	.mmap		= ext4_file_mmap,
777
	.open		= ext4_file_open,
778 779
	.release	= ext4_release_file,
	.fsync		= ext4_sync_file,
780
	.get_unmapped_area = thp_get_unmapped_area,
781
	.splice_read	= generic_file_splice_read,
A
Al Viro 已提交
782
	.splice_write	= iter_file_splice_write,
783
	.fallocate	= ext4_fallocate,
784 785
};

786
const struct inode_operations ext4_file_inode_operations = {
787
	.setattr	= ext4_setattr,
788
	.getattr	= ext4_getattr,
789
	.listxattr	= ext4_listxattr,
790
	.get_acl	= ext4_get_acl,
791
	.set_acl	= ext4_set_acl,
792
	.fiemap		= ext4_fiemap,
793 794
};