inode.c 176.4 KB
Newer Older
1
// SPDX-License-Identifier: GPL-2.0
2
/*
3
 *  linux/fs/ext4/inode.c
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
 *
 * 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/inode.c
 *
 *  Copyright (C) 1991, 1992  Linus Torvalds
 *
 *  64-bit file support on 64-bit platforms by Jakub Jelinek
 *	(jj@sunsite.ms.mff.cuni.cz)
 *
19
 *  Assorted race fixes, rewrite of ext4_get_block() by Al Viro, 2000
20 21 22 23 24 25
 */

#include <linux/fs.h>
#include <linux/time.h>
#include <linux/highuid.h>
#include <linux/pagemap.h>
26
#include <linux/dax.h>
27 28 29 30
#include <linux/quotaops.h>
#include <linux/string.h>
#include <linux/buffer_head.h>
#include <linux/writeback.h>
31
#include <linux/pagevec.h>
32
#include <linux/mpage.h>
33
#include <linux/namei.h>
34 35
#include <linux/uio.h>
#include <linux/bio.h>
36
#include <linux/workqueue.h>
37
#include <linux/kernel.h>
38
#include <linux/printk.h>
39
#include <linux/slab.h>
40
#include <linux/bitops.h>
41
#include <linux/iomap.h>
42

43
#include "ext4_jbd2.h"
44 45
#include "xattr.h"
#include "acl.h"
46
#include "truncate.h"
47

48 49
#include <trace/events/ext4.h>

50 51
#define MPAGE_DA_EXTENT_TAIL 0x01

52 53 54 55 56
static __u32 ext4_inode_csum(struct inode *inode, struct ext4_inode *raw,
			      struct ext4_inode_info *ei)
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
	__u32 csum;
57 58 59
	__u16 dummy_csum = 0;
	int offset = offsetof(struct ext4_inode, i_checksum_lo);
	unsigned int csum_size = sizeof(dummy_csum);
60

61 62 63 64 65
	csum = ext4_chksum(sbi, ei->i_csum_seed, (__u8 *)raw, offset);
	csum = ext4_chksum(sbi, csum, (__u8 *)&dummy_csum, csum_size);
	offset += csum_size;
	csum = ext4_chksum(sbi, csum, (__u8 *)raw + offset,
			   EXT4_GOOD_OLD_INODE_SIZE - offset);
66

67 68 69 70 71 72 73 74 75 76
	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE) {
		offset = offsetof(struct ext4_inode, i_checksum_hi);
		csum = ext4_chksum(sbi, csum, (__u8 *)raw +
				   EXT4_GOOD_OLD_INODE_SIZE,
				   offset - EXT4_GOOD_OLD_INODE_SIZE);
		if (EXT4_FITS_IN_INODE(raw, ei, i_checksum_hi)) {
			csum = ext4_chksum(sbi, csum, (__u8 *)&dummy_csum,
					   csum_size);
			offset += csum_size;
		}
77 78
		csum = ext4_chksum(sbi, csum, (__u8 *)raw + offset,
				   EXT4_INODE_SIZE(inode->i_sb) - offset);
79 80 81 82 83 84 85 86 87 88 89 90
	}

	return csum;
}

static int ext4_inode_csum_verify(struct inode *inode, struct ext4_inode *raw,
				  struct ext4_inode_info *ei)
{
	__u32 provided, calculated;

	if (EXT4_SB(inode->i_sb)->s_es->s_creator_os !=
	    cpu_to_le32(EXT4_OS_LINUX) ||
91
	    !ext4_has_metadata_csum(inode->i_sb))
92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111
		return 1;

	provided = le16_to_cpu(raw->i_checksum_lo);
	calculated = ext4_inode_csum(inode, raw, ei);
	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE &&
	    EXT4_FITS_IN_INODE(raw, ei, i_checksum_hi))
		provided |= ((__u32)le16_to_cpu(raw->i_checksum_hi)) << 16;
	else
		calculated &= 0xFFFF;

	return provided == calculated;
}

static void ext4_inode_csum_set(struct inode *inode, struct ext4_inode *raw,
				struct ext4_inode_info *ei)
{
	__u32 csum;

	if (EXT4_SB(inode->i_sb)->s_es->s_creator_os !=
	    cpu_to_le32(EXT4_OS_LINUX) ||
112
	    !ext4_has_metadata_csum(inode->i_sb))
113 114 115 116 117 118 119 120 121
		return;

	csum = ext4_inode_csum(inode, raw, ei);
	raw->i_checksum_lo = cpu_to_le16(csum & 0xFFFF);
	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE &&
	    EXT4_FITS_IN_INODE(raw, ei, i_checksum_hi))
		raw->i_checksum_hi = cpu_to_le16(csum >> 16);
}

122 123 124
static inline int ext4_begin_ordered_truncate(struct inode *inode,
					      loff_t new_size)
{
125
	trace_ext4_begin_ordered_truncate(inode, new_size);
126 127 128 129 130 131 132 133 134 135 136
	/*
	 * If jinode is zero, then we never opened the file for
	 * writing, so there's no need to call
	 * jbd2_journal_begin_ordered_truncate() since there's no
	 * outstanding writes we need to flush.
	 */
	if (!EXT4_I(inode)->jinode)
		return 0;
	return jbd2_journal_begin_ordered_truncate(EXT4_JOURNAL(inode),
						   EXT4_I(inode)->jinode,
						   new_size);
137 138
}

139 140
static void ext4_invalidatepage(struct page *page, unsigned int offset,
				unsigned int length);
141 142
static int __ext4_journalled_writepage(struct page *page, unsigned int len);
static int ext4_bh_delay_or_unwritten(handle_t *handle, struct buffer_head *bh);
T
Tahsin Erdogan 已提交
143 144
static int ext4_meta_trans_blocks(struct inode *inode, int lblocks,
				  int pextents);
145

146 147
/*
 * Test whether an inode is a fast symlink.
148
 * A fast symlink has its symlink data stored in ext4_inode_info->i_data.
149
 */
150
int ext4_inode_is_fast_symlink(struct inode *inode)
151
{
152 153
	return S_ISLNK(inode->i_mode) && inode->i_size &&
	       (inode->i_size < EXT4_N_BLOCKS * 4);
154 155 156 157 158 159 160
}

/*
 * Restart the transaction associated with *handle.  This does a commit,
 * so before we call here everything must be consistently dirtied against
 * this transaction.
 */
161
int ext4_truncate_restart_trans(handle_t *handle, struct inode *inode,
162
				 int nblocks)
163
{
164 165 166
	int ret;

	/*
167
	 * Drop i_data_sem to avoid deadlock with ext4_map_blocks.  At this
168 169 170 171
	 * moment, get_block can be called only for blocks inside i_size since
	 * page cache has been already dropped and writes are blocked by
	 * i_mutex. So we can safely drop the i_data_sem here.
	 */
172
	BUG_ON(EXT4_JOURNAL(inode) == NULL);
173
	jbd_debug(2, "restarting handle %p\n", handle);
174
	up_write(&EXT4_I(inode)->i_data_sem);
175
	ret = ext4_journal_restart(handle, nblocks);
176
	down_write(&EXT4_I(inode)->i_data_sem);
177
	ext4_discard_preallocations(inode);
178 179

	return ret;
180 181 182 183 184
}

/*
 * Called at the last iput() if i_nlink is zero.
 */
A
Al Viro 已提交
185
void ext4_evict_inode(struct inode *inode)
186 187
{
	handle_t *handle;
188
	int err;
A
Andreas Dilger 已提交
189
	int extra_credits = 3;
190
	struct ext4_xattr_inode_array *ea_inode_array = NULL;
191

192
	trace_ext4_evict_inode(inode);
193

A
Al Viro 已提交
194
	if (inode->i_nlink) {
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212
		/*
		 * When journalling data dirty buffers are tracked only in the
		 * journal. So although mm thinks everything is clean and
		 * ready for reaping the inode might still have some pages to
		 * write in the running transaction or waiting to be
		 * checkpointed. Thus calling jbd2_journal_invalidatepage()
		 * (via truncate_inode_pages()) to discard these buffers can
		 * cause data loss. Also even if we did not discard these
		 * buffers, we would have no way to find them after the inode
		 * is reaped and thus user could see stale data if he tries to
		 * read them before the transaction is checkpointed. So be
		 * careful and force everything to disk here... We use
		 * ei->i_datasync_tid to store the newest transaction
		 * containing inode's data.
		 *
		 * Note that directories do not have this problem because they
		 * don't use page cache.
		 */
213 214
		if (inode->i_ino != EXT4_JOURNAL_INO &&
		    ext4_should_journal_data(inode) &&
215 216
		    (S_ISLNK(inode->i_mode) || S_ISREG(inode->i_mode)) &&
		    inode->i_data.nrpages) {
217 218 219
			journal_t *journal = EXT4_SB(inode->i_sb)->s_journal;
			tid_t commit_tid = EXT4_I(inode)->i_datasync_tid;

220
			jbd2_complete_transaction(journal, commit_tid);
221 222
			filemap_write_and_wait(&inode->i_data);
		}
223
		truncate_inode_pages_final(&inode->i_data);
J
Jan Kara 已提交
224

A
Al Viro 已提交
225 226 227
		goto no_delete;
	}

228 229 230
	if (is_bad_inode(inode))
		goto no_delete;
	dquot_initialize(inode);
231

232 233
	if (ext4_should_order_data(inode))
		ext4_begin_ordered_truncate(inode, 0);
234
	truncate_inode_pages_final(&inode->i_data);
235

236 237 238 239 240
	/*
	 * Protect us against freezing - iput() caller didn't have to have any
	 * protection against it
	 */
	sb_start_intwrite(inode->i_sb);
A
Andreas Dilger 已提交
241

242 243 244 245 246
	if (!IS_NOQUOTA(inode))
		extra_credits += EXT4_MAXQUOTAS_DEL_BLOCKS(inode->i_sb);

	handle = ext4_journal_start(inode, EXT4_HT_TRUNCATE,
				 ext4_blocks_for_truncate(inode)+extra_credits);
247
	if (IS_ERR(handle)) {
248
		ext4_std_error(inode->i_sb, PTR_ERR(handle));
249 250 251 252 253
		/*
		 * If we're going to skip the normal cleanup, we still need to
		 * make sure that the in-core orphan linked list is properly
		 * cleaned up.
		 */
254
		ext4_orphan_del(NULL, inode);
255
		sb_end_intwrite(inode->i_sb);
256 257
		goto no_delete;
	}
258

259
	if (IS_SYNC(inode))
260
		ext4_handle_sync(handle);
261 262 263 264 265 266 267 268 269 270

	/*
	 * Set inode->i_size to 0 before calling ext4_truncate(). We need
	 * special handling of symlinks here because i_size is used to
	 * determine whether ext4_inode_info->i_data contains symlink data or
	 * block mappings. Setting i_size to 0 will remove its fast symlink
	 * status. Erase i_data so that it becomes a valid empty block map.
	 */
	if (ext4_inode_is_fast_symlink(inode))
		memset(EXT4_I(inode)->i_data, 0, sizeof(EXT4_I(inode)->i_data));
271
	inode->i_size = 0;
272 273
	err = ext4_mark_inode_dirty(handle, inode);
	if (err) {
274
		ext4_warning(inode->i_sb,
275 276 277
			     "couldn't mark inode dirty (err %d)", err);
		goto stop_handle;
	}
278 279 280 281 282 283 284 285 286
	if (inode->i_blocks) {
		err = ext4_truncate(inode);
		if (err) {
			ext4_error(inode->i_sb,
				   "couldn't truncate inode %lu (err %d)",
				   inode->i_ino, err);
			goto stop_handle;
		}
	}
287

288 289 290 291 292 293 294 295 296 297 298
	/* Remove xattr references. */
	err = ext4_xattr_delete_inode(handle, inode, &ea_inode_array,
				      extra_credits);
	if (err) {
		ext4_warning(inode->i_sb, "xattr delete (err %d)", err);
stop_handle:
		ext4_journal_stop(handle);
		ext4_orphan_del(NULL, inode);
		sb_end_intwrite(inode->i_sb);
		ext4_xattr_inode_array_free(ea_inode_array);
		goto no_delete;
299 300
	}

301
	/*
302
	 * Kill off the orphan record which ext4_truncate created.
303
	 * AKPM: I think this can be inside the above `if'.
304
	 * Note that ext4_orphan_del() has to be able to cope with the
305
	 * deletion of a non-existent orphan - this is because we don't
306
	 * know if ext4_truncate() actually created an orphan record.
307 308
	 * (Well, we could do this if we need to, but heck - it works)
	 */
309 310
	ext4_orphan_del(handle, inode);
	EXT4_I(inode)->i_dtime	= get_seconds();
311 312 313 314 315 316 317 318

	/*
	 * One subtle ordering requirement: if anything has gone wrong
	 * (transaction abort, IO errors, whatever), then we can still
	 * do these next steps (the fs will already have been marked as
	 * having errors), but we can't free the inode if the mark_dirty
	 * fails.
	 */
319
	if (ext4_mark_inode_dirty(handle, inode))
320
		/* If that failed, just do the required in-core inode clear. */
A
Al Viro 已提交
321
		ext4_clear_inode(inode);
322
	else
323 324
		ext4_free_inode(handle, inode);
	ext4_journal_stop(handle);
325
	sb_end_intwrite(inode->i_sb);
326
	ext4_xattr_inode_array_free(ea_inode_array);
327 328
	return;
no_delete:
A
Al Viro 已提交
329
	ext4_clear_inode(inode);	/* We must guarantee clearing of inode... */
330 331
}

332 333
#ifdef CONFIG_QUOTA
qsize_t *ext4_get_reserved_space(struct inode *inode)
334
{
335
	return &EXT4_I(inode)->i_reserved_quota;
336
}
337
#endif
338

339 340 341 342
/*
 * Called with i_data_sem down, which is important since we can call
 * ext4_discard_preallocations() from here.
 */
343 344
void ext4_da_update_reserve_space(struct inode *inode,
					int used, int quota_claim)
345 346
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
347 348 349
	struct ext4_inode_info *ei = EXT4_I(inode);

	spin_lock(&ei->i_block_reservation_lock);
350
	trace_ext4_da_update_reserve_space(inode, used, quota_claim);
351
	if (unlikely(used > ei->i_reserved_data_blocks)) {
352
		ext4_warning(inode->i_sb, "%s: ino %lu, used %d "
353
			 "with only %d reserved data blocks",
354 355 356 357 358
			 __func__, inode->i_ino, used,
			 ei->i_reserved_data_blocks);
		WARN_ON(1);
		used = ei->i_reserved_data_blocks;
	}
359

360 361
	/* Update per-inode reservations */
	ei->i_reserved_data_blocks -= used;
362
	percpu_counter_sub(&sbi->s_dirtyclusters_counter, used);
363

364
	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
365

366 367
	/* Update quota subsystem for data blocks */
	if (quota_claim)
368
		dquot_claim_block(inode, EXT4_C2B(sbi, used));
369
	else {
370 371 372
		/*
		 * We did fallocate with an offset that is already delayed
		 * allocated. So on delayed allocated writeback we should
373
		 * not re-claim the quota for fallocated blocks.
374
		 */
375
		dquot_release_reservation_block(inode, EXT4_C2B(sbi, used));
376
	}
377 378 379 380 381 382

	/*
	 * If we have done all the pending block allocations and if
	 * there aren't any writers on the inode, we can discard the
	 * inode's preallocations.
	 */
383 384
	if ((ei->i_reserved_data_blocks == 0) &&
	    (atomic_read(&inode->i_writecount) == 0))
385
		ext4_discard_preallocations(inode);
386 387
}

388
static int __check_block_validity(struct inode *inode, const char *func,
389 390
				unsigned int line,
				struct ext4_map_blocks *map)
391
{
392 393
	if (!ext4_data_block_valid(EXT4_SB(inode->i_sb), map->m_pblk,
				   map->m_len)) {
394 395 396 397
		ext4_error_inode(inode, func, line, map->m_pblk,
				 "lblock %lu mapped to illegal pblock "
				 "(length %d)", (unsigned long) map->m_lblk,
				 map->m_len);
398
		return -EFSCORRUPTED;
399 400 401 402
	}
	return 0;
}

J
Jan Kara 已提交
403 404 405 406 407 408
int ext4_issue_zeroout(struct inode *inode, ext4_lblk_t lblk, ext4_fsblk_t pblk,
		       ext4_lblk_t len)
{
	int ret;

	if (ext4_encrypted_inode(inode))
409
		return fscrypt_zeroout_range(inode, lblk, pblk, len);
J
Jan Kara 已提交
410 411 412 413 414 415 416 417

	ret = sb_issue_zeroout(inode->i_sb, pblk, len, GFP_NOFS);
	if (ret > 0)
		ret = 0;

	return ret;
}

418
#define check_block_validity(inode, map)	\
419
	__check_block_validity((inode), __func__, __LINE__, (map))
420

421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437
#ifdef ES_AGGRESSIVE_TEST
static void ext4_map_blocks_es_recheck(handle_t *handle,
				       struct inode *inode,
				       struct ext4_map_blocks *es_map,
				       struct ext4_map_blocks *map,
				       int flags)
{
	int retval;

	map->m_flags = 0;
	/*
	 * There is a race window that the result is not the same.
	 * e.g. xfstests #223 when dioread_nolock enables.  The reason
	 * is that we lookup a block mapping in extent status tree with
	 * out taking i_data_sem.  So at the time the unwritten extent
	 * could be converted.
	 */
438
	down_read(&EXT4_I(inode)->i_data_sem);
439 440 441 442 443 444 445
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
		retval = ext4_ext_map_blocks(handle, inode, map, flags &
					     EXT4_GET_BLOCKS_KEEP_SIZE);
	} else {
		retval = ext4_ind_map_blocks(handle, inode, map, flags &
					     EXT4_GET_BLOCKS_KEEP_SIZE);
	}
446
	up_read((&EXT4_I(inode)->i_data_sem));
447 448 449 450 451 452 453 454

	/*
	 * We don't check m_len because extent will be collpased in status
	 * tree.  So the m_len might not equal.
	 */
	if (es_map->m_lblk != map->m_lblk ||
	    es_map->m_flags != map->m_flags ||
	    es_map->m_pblk != map->m_pblk) {
455
		printk("ES cache assertion failed for inode: %lu "
456 457 458 459 460 461 462 463 464 465
		       "es_cached ex [%d/%d/%llu/%x] != "
		       "found ex [%d/%d/%llu/%x] retval %d flags %x\n",
		       inode->i_ino, es_map->m_lblk, es_map->m_len,
		       es_map->m_pblk, es_map->m_flags, map->m_lblk,
		       map->m_len, map->m_pblk, map->m_flags,
		       retval, flags);
	}
}
#endif /* ES_AGGRESSIVE_TEST */

466
/*
467
 * The ext4_map_blocks() function tries to look up the requested blocks,
468
 * and returns if the blocks are already mapped.
469 470 471 472 473
 *
 * Otherwise it takes the write lock of the i_data_sem and allocate blocks
 * and store the allocated blocks in the result buffer head and mark it
 * mapped.
 *
474 475
 * If file type is extents based, it will call ext4_ext_map_blocks(),
 * Otherwise, call with ext4_ind_map_blocks() to handle indirect mapping
476 477
 * based files
 *
478 479 480
 * On success, it returns the number of blocks being mapped or allocated.  if
 * create==0 and the blocks are pre-allocated and unwritten, the resulting @map
 * is marked as unwritten. If the create == 1, it will mark @map as mapped.
481 482
 *
 * It returns 0 if plain look up failed (blocks have not been allocated), in
483 484
 * that case, @map is returned as unmapped but we still do fill map->m_len to
 * indicate the length of a hole starting at map->m_lblk.
485 486 487
 *
 * It returns the error in case of allocation failure.
 */
488 489
int ext4_map_blocks(handle_t *handle, struct inode *inode,
		    struct ext4_map_blocks *map, int flags)
490
{
491
	struct extent_status es;
492
	int retval;
493
	int ret = 0;
494 495 496 497 498
#ifdef ES_AGGRESSIVE_TEST
	struct ext4_map_blocks orig_map;

	memcpy(&orig_map, map, sizeof(*map));
#endif
499

500 501 502 503
	map->m_flags = 0;
	ext_debug("ext4_map_blocks(): inode %lu, flag %d, max_blocks %u,"
		  "logical block %lu\n", inode->i_ino, flags, map->m_len,
		  (unsigned long) map->m_lblk);
504

505 506 507 508 509 510
	/*
	 * ext4_map_blocks returns an int, and m_len is an unsigned int
	 */
	if (unlikely(map->m_len > INT_MAX))
		map->m_len = INT_MAX;

511 512
	/* We can handle the block number less than EXT_MAX_BLOCKS */
	if (unlikely(map->m_lblk >= EXT_MAX_BLOCKS))
513
		return -EFSCORRUPTED;
514

515 516 517 518 519 520 521 522 523 524 525 526
	/* Lookup extent status tree firstly */
	if (ext4_es_lookup_extent(inode, map->m_lblk, &es)) {
		if (ext4_es_is_written(&es) || ext4_es_is_unwritten(&es)) {
			map->m_pblk = ext4_es_pblock(&es) +
					map->m_lblk - es.es_lblk;
			map->m_flags |= ext4_es_is_written(&es) ?
					EXT4_MAP_MAPPED : EXT4_MAP_UNWRITTEN;
			retval = es.es_len - (map->m_lblk - es.es_lblk);
			if (retval > map->m_len)
				retval = map->m_len;
			map->m_len = retval;
		} else if (ext4_es_is_delayed(&es) || ext4_es_is_hole(&es)) {
527 528 529 530 531
			map->m_pblk = 0;
			retval = es.es_len - (map->m_lblk - es.es_lblk);
			if (retval > map->m_len)
				retval = map->m_len;
			map->m_len = retval;
532 533 534 535
			retval = 0;
		} else {
			BUG_ON(1);
		}
536 537 538 539
#ifdef ES_AGGRESSIVE_TEST
		ext4_map_blocks_es_recheck(handle, inode, map,
					   &orig_map, flags);
#endif
540 541 542
		goto found;
	}

543
	/*
544 545
	 * Try to see if we can get the block without requesting a new
	 * file system block.
546
	 */
547
	down_read(&EXT4_I(inode)->i_data_sem);
548
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
549 550
		retval = ext4_ext_map_blocks(handle, inode, map, flags &
					     EXT4_GET_BLOCKS_KEEP_SIZE);
551
	} else {
552 553
		retval = ext4_ind_map_blocks(handle, inode, map, flags &
					     EXT4_GET_BLOCKS_KEEP_SIZE);
554
	}
555
	if (retval > 0) {
556
		unsigned int status;
557

558 559 560 561 562 563
		if (unlikely(retval != map->m_len)) {
			ext4_warning(inode->i_sb,
				     "ES len assertion failed for inode "
				     "%lu: retval %d != map->m_len %d",
				     inode->i_ino, retval, map->m_len);
			WARN_ON(1);
564 565
		}

566 567 568
		status = map->m_flags & EXT4_MAP_UNWRITTEN ?
				EXTENT_STATUS_UNWRITTEN : EXTENT_STATUS_WRITTEN;
		if (!(flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) &&
569
		    !(status & EXTENT_STATUS_WRITTEN) &&
570 571 572 573 574 575 576 577
		    ext4_find_delalloc_range(inode, map->m_lblk,
					     map->m_lblk + map->m_len - 1))
			status |= EXTENT_STATUS_DELAYED;
		ret = ext4_es_insert_extent(inode, map->m_lblk,
					    map->m_len, map->m_pblk, status);
		if (ret < 0)
			retval = ret;
	}
578
	up_read((&EXT4_I(inode)->i_data_sem));
579

580
found:
581
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
582
		ret = check_block_validity(inode, map);
583 584 585 586
		if (ret != 0)
			return ret;
	}

587
	/* If it is only a block(s) look up */
588
	if ((flags & EXT4_GET_BLOCKS_CREATE) == 0)
589 590 591 592 593 594
		return retval;

	/*
	 * Returns if the blocks have already allocated
	 *
	 * Note that if blocks have been preallocated
595
	 * ext4_ext_get_block() returns the create = 0
596 597
	 * with buffer head unmapped.
	 */
598
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED)
599 600 601 602 603 604 605
		/*
		 * If we need to convert extent to unwritten
		 * we continue and do the actual work in
		 * ext4_ext_map_blocks()
		 */
		if (!(flags & EXT4_GET_BLOCKS_CONVERT_UNWRITTEN))
			return retval;
606

607
	/*
608 609
	 * Here we clear m_flags because after allocating an new extent,
	 * it will be set again.
610
	 */
611
	map->m_flags &= ~EXT4_MAP_FLAGS;
612

613
	/*
614
	 * New blocks allocate and/or writing to unwritten extent
615
	 * will possibly result in updating i_data, so we take
616
	 * the write lock of i_data_sem, and call get_block()
617
	 * with create == 1 flag.
618
	 */
619
	down_write(&EXT4_I(inode)->i_data_sem);
620

621 622 623 624
	/*
	 * We need to check for EXT4 here because migrate
	 * could have changed the inode type in between
	 */
625
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
626
		retval = ext4_ext_map_blocks(handle, inode, map, flags);
627
	} else {
628
		retval = ext4_ind_map_blocks(handle, inode, map, flags);
629

630
		if (retval > 0 && map->m_flags & EXT4_MAP_NEW) {
631 632 633 634 635
			/*
			 * We allocated new blocks which will result in
			 * i_data's format changing.  Force the migrate
			 * to fail by clearing migrate flags
			 */
636
			ext4_clear_inode_state(inode, EXT4_STATE_EXT_MIGRATE);
637
		}
638

639 640 641 642 643 644 645
		/*
		 * Update reserved blocks/metadata blocks after successful
		 * block allocation which had been deferred till now. We don't
		 * support fallocate for non extent files. So we can update
		 * reserve space here.
		 */
		if ((retval > 0) &&
646
			(flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE))
647 648
			ext4_da_update_reserve_space(inode, retval, 1);
	}
649

650
	if (retval > 0) {
651
		unsigned int status;
652

653 654 655 656 657 658
		if (unlikely(retval != map->m_len)) {
			ext4_warning(inode->i_sb,
				     "ES len assertion failed for inode "
				     "%lu: retval %d != map->m_len %d",
				     inode->i_ino, retval, map->m_len);
			WARN_ON(1);
659 660
		}

661 662 663
		/*
		 * We have to zeroout blocks before inserting them into extent
		 * status tree. Otherwise someone could look them up there and
664 665 666
		 * use them before they are really zeroed. We also have to
		 * unmap metadata before zeroing as otherwise writeback can
		 * overwrite zeros with stale data from block device.
667 668 669 670
		 */
		if (flags & EXT4_GET_BLOCKS_ZERO &&
		    map->m_flags & EXT4_MAP_MAPPED &&
		    map->m_flags & EXT4_MAP_NEW) {
671 672
			clean_bdev_aliases(inode->i_sb->s_bdev, map->m_pblk,
					   map->m_len);
673 674 675 676 677 678 679 680
			ret = ext4_issue_zeroout(inode, map->m_lblk,
						 map->m_pblk, map->m_len);
			if (ret) {
				retval = ret;
				goto out_sem;
			}
		}

681 682 683 684 685 686 687
		/*
		 * If the extent has been zeroed out, we don't need to update
		 * extent status tree.
		 */
		if ((flags & EXT4_GET_BLOCKS_PRE_IO) &&
		    ext4_es_lookup_extent(inode, map->m_lblk, &es)) {
			if (ext4_es_is_written(&es))
688
				goto out_sem;
689
		}
690 691 692
		status = map->m_flags & EXT4_MAP_UNWRITTEN ?
				EXTENT_STATUS_UNWRITTEN : EXTENT_STATUS_WRITTEN;
		if (!(flags & EXT4_GET_BLOCKS_DELALLOC_RESERVE) &&
693
		    !(status & EXTENT_STATUS_WRITTEN) &&
694 695 696 697 698
		    ext4_find_delalloc_range(inode, map->m_lblk,
					     map->m_lblk + map->m_len - 1))
			status |= EXTENT_STATUS_DELAYED;
		ret = ext4_es_insert_extent(inode, map->m_lblk, map->m_len,
					    map->m_pblk, status);
699
		if (ret < 0) {
700
			retval = ret;
701 702
			goto out_sem;
		}
703 704
	}

705
out_sem:
706
	up_write((&EXT4_I(inode)->i_data_sem));
707
	if (retval > 0 && map->m_flags & EXT4_MAP_MAPPED) {
708
		ret = check_block_validity(inode, map);
709 710
		if (ret != 0)
			return ret;
J
Jan Kara 已提交
711 712 713 714 715 716 717 718 719

		/*
		 * Inodes with freshly allocated blocks where contents will be
		 * visible after transaction commit must be on transaction's
		 * ordered data list.
		 */
		if (map->m_flags & EXT4_MAP_NEW &&
		    !(map->m_flags & EXT4_MAP_UNWRITTEN) &&
		    !(flags & EXT4_GET_BLOCKS_ZERO) &&
T
Tahsin Erdogan 已提交
720
		    !ext4_is_quota_file(inode) &&
J
Jan Kara 已提交
721
		    ext4_should_order_data(inode)) {
722 723 724 725
			if (flags & EXT4_GET_BLOCKS_IO_SUBMIT)
				ret = ext4_jbd2_inode_add_wait(handle, inode);
			else
				ret = ext4_jbd2_inode_add_write(handle, inode);
J
Jan Kara 已提交
726 727 728
			if (ret)
				return ret;
		}
729
	}
730 731 732
	return retval;
}

J
Jan Kara 已提交
733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760
/*
 * Update EXT4_MAP_FLAGS in bh->b_state. For buffer heads attached to pages
 * we have to be careful as someone else may be manipulating b_state as well.
 */
static void ext4_update_bh_state(struct buffer_head *bh, unsigned long flags)
{
	unsigned long old_state;
	unsigned long new_state;

	flags &= EXT4_MAP_FLAGS;

	/* Dummy buffer_head? Set non-atomically. */
	if (!bh->b_page) {
		bh->b_state = (bh->b_state & ~EXT4_MAP_FLAGS) | flags;
		return;
	}
	/*
	 * Someone else may be modifying b_state. Be careful! This is ugly but
	 * once we get rid of using bh as a container for mapping information
	 * to pass to / from get_block functions, this can go away.
	 */
	do {
		old_state = READ_ONCE(bh->b_state);
		new_state = (old_state & ~EXT4_MAP_FLAGS) | flags;
	} while (unlikely(
		 cmpxchg(&bh->b_state, old_state, new_state) != old_state));
}

761 762
static int _ext4_get_block(struct inode *inode, sector_t iblock,
			   struct buffer_head *bh, int flags)
763
{
764
	struct ext4_map_blocks map;
765
	int ret = 0;
766

T
Tao Ma 已提交
767 768 769
	if (ext4_has_inline_data(inode))
		return -ERANGE;

770 771 772
	map.m_lblk = iblock;
	map.m_len = bh->b_size >> inode->i_blkbits;

773 774
	ret = ext4_map_blocks(ext4_journal_current_handle(), inode, &map,
			      flags);
J
Jan Kara 已提交
775
	if (ret > 0) {
776
		map_bh(bh, inode->i_sb, map.m_pblk);
J
Jan Kara 已提交
777
		ext4_update_bh_state(bh, map.m_flags);
778
		bh->b_size = inode->i_sb->s_blocksize * map.m_len;
J
Jan Kara 已提交
779
		ret = 0;
780 781 782
	} else if (ret == 0) {
		/* hole case, need to fill in bh->b_size */
		bh->b_size = inode->i_sb->s_blocksize * map.m_len;
783 784 785 786
	}
	return ret;
}

787 788 789 790 791 792 793
int ext4_get_block(struct inode *inode, sector_t iblock,
		   struct buffer_head *bh, int create)
{
	return _ext4_get_block(inode, iblock, bh,
			       create ? EXT4_GET_BLOCKS_CREATE : 0);
}

794 795 796 797 798 799 800 801 802 803 804 805 806 807
/*
 * Get block function used when preparing for buffered write if we require
 * creating an unwritten extent if blocks haven't been allocated.  The extent
 * will be converted to written after the IO is complete.
 */
int ext4_get_block_unwritten(struct inode *inode, sector_t iblock,
			     struct buffer_head *bh_result, int create)
{
	ext4_debug("ext4_get_block_unwritten: inode %lu, create flag %d\n",
		   inode->i_ino, create);
	return _ext4_get_block(inode, iblock, bh_result,
			       EXT4_GET_BLOCKS_IO_CREATE_EXT);
}

808 809 810
/* Maximum number of blocks we map for direct IO at once. */
#define DIO_MAX_BLOCKS 4096

811 812 813 814 815 816 817
/*
 * Get blocks function for the cases that need to start a transaction -
 * generally difference cases of direct IO and DAX IO. It also handles retries
 * in case of ENOSPC.
 */
static int ext4_get_block_trans(struct inode *inode, sector_t iblock,
				struct buffer_head *bh_result, int flags)
818 819
{
	int dio_credits;
820 821 822
	handle_t *handle;
	int retries = 0;
	int ret;
823 824 825 826 827 828

	/* Trim mapping request to maximum we can map at once for DIO */
	if (bh_result->b_size >> inode->i_blkbits > DIO_MAX_BLOCKS)
		bh_result->b_size = DIO_MAX_BLOCKS << inode->i_blkbits;
	dio_credits = ext4_chunk_trans_blocks(inode,
				      bh_result->b_size >> inode->i_blkbits);
829 830 831 832 833 834 835 836 837 838 839
retry:
	handle = ext4_journal_start(inode, EXT4_HT_MAP_BLOCKS, dio_credits);
	if (IS_ERR(handle))
		return PTR_ERR(handle);

	ret = _ext4_get_block(inode, iblock, bh_result, flags);
	ext4_journal_stop(handle);

	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
		goto retry;
	return ret;
840 841
}

842 843 844 845
/* Get block function for DIO reads and writes to inodes without extents */
int ext4_dio_get_block(struct inode *inode, sector_t iblock,
		       struct buffer_head *bh, int create)
{
846 847 848
	/* We don't expect handle for direct IO */
	WARN_ON_ONCE(ext4_journal_current_handle());

849 850 851
	if (!create)
		return _ext4_get_block(inode, iblock, bh, 0);
	return ext4_get_block_trans(inode, iblock, bh, EXT4_GET_BLOCKS_CREATE);
852 853 854
}

/*
855
 * Get block function for AIO DIO writes when we create unwritten extent if
856 857 858
 * blocks are not allocated yet. The extent will be converted to written
 * after IO is complete.
 */
859 860
static int ext4_dio_get_block_unwritten_async(struct inode *inode,
		sector_t iblock, struct buffer_head *bh_result,	int create)
861
{
862 863 864 865 866
	int ret;

	/* We don't expect handle for direct IO */
	WARN_ON_ONCE(ext4_journal_current_handle());

867 868
	ret = ext4_get_block_trans(inode, iblock, bh_result,
				   EXT4_GET_BLOCKS_IO_CREATE_EXT);
869

870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886
	/*
	 * When doing DIO using unwritten extents, we need io_end to convert
	 * unwritten extents to written on IO completion. We allocate io_end
	 * once we spot unwritten extent and store it in b_private. Generic
	 * DIO code keeps b_private set and furthermore passes the value to
	 * our completion callback in 'private' argument.
	 */
	if (!ret && buffer_unwritten(bh_result)) {
		if (!bh_result->b_private) {
			ext4_io_end_t *io_end;

			io_end = ext4_init_io_end(inode, GFP_KERNEL);
			if (!io_end)
				return -ENOMEM;
			bh_result->b_private = io_end;
			ext4_set_io_unwritten_flag(inode, io_end);
		}
887 888 889 890
		set_buffer_defer_completion(bh_result);
	}

	return ret;
891 892
}

893 894 895
/*
 * Get block function for non-AIO DIO writes when we create unwritten extent if
 * blocks are not allocated yet. The extent will be converted to written
896
 * after IO is complete by ext4_direct_IO_write().
897 898 899 900 901 902 903 904 905
 */
static int ext4_dio_get_block_unwritten_sync(struct inode *inode,
		sector_t iblock, struct buffer_head *bh_result,	int create)
{
	int ret;

	/* We don't expect handle for direct IO */
	WARN_ON_ONCE(ext4_journal_current_handle());

906 907
	ret = ext4_get_block_trans(inode, iblock, bh_result,
				   EXT4_GET_BLOCKS_IO_CREATE_EXT);
908 909 910

	/*
	 * Mark inode as having pending DIO writes to unwritten extents.
911
	 * ext4_direct_IO_write() checks this flag and converts extents to
912 913 914 915 916 917 918 919
	 * written.
	 */
	if (!ret && buffer_unwritten(bh_result))
		ext4_set_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN);

	return ret;
}

920 921 922 923 924 925 926
static int ext4_dio_get_block_overwrite(struct inode *inode, sector_t iblock,
		   struct buffer_head *bh_result, int create)
{
	int ret;

	ext4_debug("ext4_dio_get_block_overwrite: inode %lu, create flag %d\n",
		   inode->i_ino, create);
927 928 929
	/* We don't expect handle for direct IO */
	WARN_ON_ONCE(ext4_journal_current_handle());

930 931 932 933 934
	ret = _ext4_get_block(inode, iblock, bh_result, 0);
	/*
	 * Blocks should have been preallocated! ext4_file_write_iter() checks
	 * that.
	 */
935
	WARN_ON_ONCE(!buffer_mapped(bh_result) || buffer_unwritten(bh_result));
936 937 938 939 940

	return ret;
}


941 942 943
/*
 * `handle' can be NULL if create is zero
 */
944
struct buffer_head *ext4_getblk(handle_t *handle, struct inode *inode,
945
				ext4_lblk_t block, int map_flags)
946
{
947 948
	struct ext4_map_blocks map;
	struct buffer_head *bh;
949
	int create = map_flags & EXT4_GET_BLOCKS_CREATE;
950
	int err;
951 952 953

	J_ASSERT(handle != NULL || create == 0);

954 955
	map.m_lblk = block;
	map.m_len = 1;
956
	err = ext4_map_blocks(handle, inode, &map, map_flags);
957

958 959
	if (err == 0)
		return create ? ERR_PTR(-ENOSPC) : NULL;
960
	if (err < 0)
961
		return ERR_PTR(err);
962 963

	bh = sb_getblk(inode->i_sb, map.m_pblk);
964 965
	if (unlikely(!bh))
		return ERR_PTR(-ENOMEM);
966 967 968
	if (map.m_flags & EXT4_MAP_NEW) {
		J_ASSERT(create != 0);
		J_ASSERT(handle != NULL);
969

970 971 972 973 974 975 976 977 978
		/*
		 * Now that we do not always journal data, we should
		 * keep in mind whether this should always journal the
		 * new buffer as metadata.  For now, regular file
		 * writes use ext4_get_block instead, so it's not a
		 * problem.
		 */
		lock_buffer(bh);
		BUFFER_TRACE(bh, "call get_create_access");
979 980 981 982 983 984
		err = ext4_journal_get_create_access(handle, bh);
		if (unlikely(err)) {
			unlock_buffer(bh);
			goto errout;
		}
		if (!buffer_uptodate(bh)) {
985 986
			memset(bh->b_data, 0, inode->i_sb->s_blocksize);
			set_buffer_uptodate(bh);
987
		}
988 989 990
		unlock_buffer(bh);
		BUFFER_TRACE(bh, "call ext4_handle_dirty_metadata");
		err = ext4_handle_dirty_metadata(handle, inode, bh);
991 992 993
		if (unlikely(err))
			goto errout;
	} else
994 995
		BUFFER_TRACE(bh, "not a new buffer");
	return bh;
996 997 998
errout:
	brelse(bh);
	return ERR_PTR(err);
999 1000
}

1001
struct buffer_head *ext4_bread(handle_t *handle, struct inode *inode,
1002
			       ext4_lblk_t block, int map_flags)
1003
{
1004
	struct buffer_head *bh;
1005

1006
	bh = ext4_getblk(handle, inode, block, map_flags);
1007
	if (IS_ERR(bh))
1008
		return bh;
1009
	if (!bh || buffer_uptodate(bh))
1010
		return bh;
1011
	ll_rw_block(REQ_OP_READ, REQ_META | REQ_PRIO, 1, &bh);
1012 1013 1014 1015
	wait_on_buffer(bh);
	if (buffer_uptodate(bh))
		return bh;
	put_bh(bh);
1016
	return ERR_PTR(-EIO);
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
/* Read a contiguous batch of blocks. */
int ext4_bread_batch(struct inode *inode, ext4_lblk_t block, int bh_count,
		     bool wait, struct buffer_head **bhs)
{
	int i, err;

	for (i = 0; i < bh_count; i++) {
		bhs[i] = ext4_getblk(NULL, inode, block + i, 0 /* map_flags */);
		if (IS_ERR(bhs[i])) {
			err = PTR_ERR(bhs[i]);
			bh_count = i;
			goto out_brelse;
		}
	}

	for (i = 0; i < bh_count; i++)
		/* Note that NULL bhs[i] is valid because of holes. */
		if (bhs[i] && !buffer_uptodate(bhs[i]))
			ll_rw_block(REQ_OP_READ, REQ_META | REQ_PRIO, 1,
				    &bhs[i]);

	if (!wait)
		return 0;

	for (i = 0; i < bh_count; i++)
		if (bhs[i])
			wait_on_buffer(bhs[i]);

	for (i = 0; i < bh_count; i++) {
		if (bhs[i] && !buffer_uptodate(bhs[i])) {
			err = -EIO;
			goto out_brelse;
		}
	}
	return 0;

out_brelse:
	for (i = 0; i < bh_count; i++) {
		brelse(bhs[i]);
		bhs[i] = NULL;
	}
	return err;
}

1063 1064 1065 1066 1067 1068 1069
int ext4_walk_page_buffers(handle_t *handle,
			   struct buffer_head *head,
			   unsigned from,
			   unsigned to,
			   int *partial,
			   int (*fn)(handle_t *handle,
				     struct buffer_head *bh))
1070 1071 1072 1073 1074 1075 1076
{
	struct buffer_head *bh;
	unsigned block_start, block_end;
	unsigned blocksize = head->b_size;
	int err, ret = 0;
	struct buffer_head *next;

1077 1078
	for (bh = head, block_start = 0;
	     ret == 0 && (bh != head || !block_start);
1079
	     block_start = block_end, bh = next) {
1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096
		next = bh->b_this_page;
		block_end = block_start + blocksize;
		if (block_end <= from || block_start >= to) {
			if (partial && !buffer_uptodate(bh))
				*partial = 1;
			continue;
		}
		err = (*fn)(handle, bh);
		if (!ret)
			ret = err;
	}
	return ret;
}

/*
 * To preserve ordering, it is essential that the hole instantiation and
 * the data write be encapsulated in a single transaction.  We cannot
1097
 * close off a transaction and start a new one between the ext4_get_block()
1098
 * and the commit_write().  So doing the jbd2_journal_start at the start of
1099 1100
 * prepare_write() is the right place.
 *
1101 1102 1103 1104
 * Also, this function can nest inside ext4_writepage().  In that case, we
 * *know* that ext4_writepage() has generated enough buffer credits to do the
 * whole page.  So we won't block on the journal in that case, which is good,
 * because the caller may be PF_MEMALLOC.
1105
 *
1106
 * By accident, ext4 can be reentered when a transaction is open via
1107 1108 1109 1110 1111 1112
 * quota file writes.  If we were to commit the transaction while thus
 * reentered, there can be a deadlock - we would be holding a quota
 * lock, and the commit would never complete if another thread had a
 * transaction open and was blocking on the quota lock - a ranking
 * violation.
 *
1113
 * So what we do is to rely on the fact that jbd2_journal_stop/journal_start
1114 1115 1116 1117
 * will _not_ run commit under these circumstances because handle->h_ref
 * is elevated.  We'll still have enough credits for the tiny quotafile
 * write.
 */
1118 1119
int do_journal_get_write_access(handle_t *handle,
				struct buffer_head *bh)
1120
{
1121 1122 1123
	int dirty = buffer_dirty(bh);
	int ret;

1124 1125
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
1126
	/*
C
Christoph Hellwig 已提交
1127
	 * __block_write_begin() could have dirtied some buffers. Clean
1128 1129
	 * the dirty bit as jbd2_journal_get_write_access() could complain
	 * otherwise about fs integrity issues. Setting of the dirty bit
C
Christoph Hellwig 已提交
1130
	 * by __block_write_begin() isn't a real problem here as we clear
1131 1132 1133 1134 1135
	 * the bit before releasing a page lock and thus writeback cannot
	 * ever write the buffer.
	 */
	if (dirty)
		clear_buffer_dirty(bh);
1136
	BUFFER_TRACE(bh, "get write access");
1137 1138 1139 1140
	ret = ext4_journal_get_write_access(handle, bh);
	if (!ret && dirty)
		ret = ext4_handle_dirty_metadata(handle, NULL, bh);
	return ret;
1141 1142
}

1143 1144 1145 1146
#ifdef CONFIG_EXT4_FS_ENCRYPTION
static int ext4_block_write_begin(struct page *page, loff_t pos, unsigned len,
				  get_block_t *get_block)
{
1147
	unsigned from = pos & (PAGE_SIZE - 1);
1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158
	unsigned to = from + len;
	struct inode *inode = page->mapping->host;
	unsigned block_start, block_end;
	sector_t block;
	int err = 0;
	unsigned blocksize = inode->i_sb->s_blocksize;
	unsigned bbits;
	struct buffer_head *bh, *head, *wait[2], **wait_bh = wait;
	bool decrypt = false;

	BUG_ON(!PageLocked(page));
1159 1160
	BUG_ON(from > PAGE_SIZE);
	BUG_ON(to > PAGE_SIZE);
1161 1162 1163 1164 1165 1166
	BUG_ON(from > to);

	if (!page_has_buffers(page))
		create_empty_buffers(page, blocksize, 0);
	head = page_buffers(page);
	bbits = ilog2(blocksize);
1167
	block = (sector_t)page->index << (PAGE_SHIFT - bbits);
1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186

	for (bh = head, block_start = 0; bh != head || !block_start;
	    block++, block_start = block_end, bh = bh->b_this_page) {
		block_end = block_start + blocksize;
		if (block_end <= from || block_start >= to) {
			if (PageUptodate(page)) {
				if (!buffer_uptodate(bh))
					set_buffer_uptodate(bh);
			}
			continue;
		}
		if (buffer_new(bh))
			clear_buffer_new(bh);
		if (!buffer_mapped(bh)) {
			WARN_ON(bh->b_size != blocksize);
			err = get_block(inode, block, bh, 1);
			if (err)
				break;
			if (buffer_new(bh)) {
1187
				clean_bdev_bh_alias(bh);
1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207
				if (PageUptodate(page)) {
					clear_buffer_new(bh);
					set_buffer_uptodate(bh);
					mark_buffer_dirty(bh);
					continue;
				}
				if (block_end > to || block_start < from)
					zero_user_segments(page, to, block_end,
							   block_start, from);
				continue;
			}
		}
		if (PageUptodate(page)) {
			if (!buffer_uptodate(bh))
				set_buffer_uptodate(bh);
			continue;
		}
		if (!buffer_uptodate(bh) && !buffer_delay(bh) &&
		    !buffer_unwritten(bh) &&
		    (block_start < from || block_end > to)) {
1208
			ll_rw_block(REQ_OP_READ, 0, 1, &bh);
1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224
			*wait_bh++ = bh;
			decrypt = ext4_encrypted_inode(inode) &&
				S_ISREG(inode->i_mode);
		}
	}
	/*
	 * If we issued read requests, let them complete.
	 */
	while (wait_bh > wait) {
		wait_on_buffer(*--wait_bh);
		if (!buffer_uptodate(*wait_bh))
			err = -EIO;
	}
	if (unlikely(err))
		page_zero_new_buffers(page, from, to);
	else if (decrypt)
1225
		err = fscrypt_decrypt_page(page->mapping->host, page,
1226
				PAGE_SIZE, 0, page->index);
1227 1228 1229 1230
	return err;
}
#endif

N
Nick Piggin 已提交
1231
static int ext4_write_begin(struct file *file, struct address_space *mapping,
1232 1233
			    loff_t pos, unsigned len, unsigned flags,
			    struct page **pagep, void **fsdata)
1234
{
1235
	struct inode *inode = mapping->host;
1236
	int ret, needed_blocks;
1237 1238
	handle_t *handle;
	int retries = 0;
1239
	struct page *page;
1240
	pgoff_t index;
1241
	unsigned from, to;
N
Nick Piggin 已提交
1242

1243 1244 1245
	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
		return -EIO;

1246
	trace_ext4_write_begin(inode, pos, len, flags);
1247 1248 1249 1250 1251
	/*
	 * Reserve one block more for addition to orphan list in case
	 * we allocate blocks but write fails for some reason
	 */
	needed_blocks = ext4_writepage_trans_blocks(inode) + 1;
1252 1253
	index = pos >> PAGE_SHIFT;
	from = pos & (PAGE_SIZE - 1);
1254
	to = from + len;
1255

1256 1257 1258 1259
	if (ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA)) {
		ret = ext4_try_to_write_inline_data(mapping, inode, pos, len,
						    flags, pagep);
		if (ret < 0)
1260 1261 1262
			return ret;
		if (ret == 1)
			return 0;
1263 1264
	}

1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278
	/*
	 * grab_cache_page_write_begin() can take a long time if the
	 * system is thrashing due to memory pressure, or if the page
	 * is being written back.  So grab it first before we start
	 * the transaction handle.  This also allows us to allocate
	 * the page (if needed) without using GFP_NOFS.
	 */
retry_grab:
	page = grab_cache_page_write_begin(mapping, index, flags);
	if (!page)
		return -ENOMEM;
	unlock_page(page);

retry_journal:
1279
	handle = ext4_journal_start(inode, EXT4_HT_WRITE_PAGE, needed_blocks);
1280
	if (IS_ERR(handle)) {
1281
		put_page(page);
1282
		return PTR_ERR(handle);
1283
	}
1284

1285 1286 1287 1288
	lock_page(page);
	if (page->mapping != mapping) {
		/* The page got truncated from under us */
		unlock_page(page);
1289
		put_page(page);
1290
		ext4_journal_stop(handle);
1291
		goto retry_grab;
1292
	}
1293 1294
	/* In case writeback began while the page was unlocked */
	wait_for_stable_page(page);
1295

1296 1297 1298
#ifdef CONFIG_EXT4_FS_ENCRYPTION
	if (ext4_should_dioread_nolock(inode))
		ret = ext4_block_write_begin(page, pos, len,
1299
					     ext4_get_block_unwritten);
1300 1301 1302 1303
	else
		ret = ext4_block_write_begin(page, pos, len,
					     ext4_get_block);
#else
1304
	if (ext4_should_dioread_nolock(inode))
1305 1306
		ret = __block_write_begin(page, pos, len,
					  ext4_get_block_unwritten);
1307
	else
1308
		ret = __block_write_begin(page, pos, len, ext4_get_block);
1309
#endif
N
Nick Piggin 已提交
1310
	if (!ret && ext4_should_journal_data(inode)) {
1311 1312 1313
		ret = ext4_walk_page_buffers(handle, page_buffers(page),
					     from, to, NULL,
					     do_journal_get_write_access);
1314
	}
N
Nick Piggin 已提交
1315 1316

	if (ret) {
1317
		unlock_page(page);
1318
		/*
1319
		 * __block_write_begin may have instantiated a few blocks
1320 1321
		 * outside i_size.  Trim these off again. Don't need
		 * i_size_read because we hold i_mutex.
1322 1323 1324
		 *
		 * Add inode to orphan list in case we crash before
		 * truncate finishes
1325
		 */
1326
		if (pos + len > inode->i_size && ext4_can_truncate(inode))
1327 1328 1329 1330
			ext4_orphan_add(handle, inode);

		ext4_journal_stop(handle);
		if (pos + len > inode->i_size) {
1331
			ext4_truncate_failed_write(inode);
1332
			/*
1333
			 * If truncate failed early the inode might
1334 1335 1336 1337 1338 1339 1340
			 * still be on the orphan list; we need to
			 * make sure the inode is removed from the
			 * orphan list in that case.
			 */
			if (inode->i_nlink)
				ext4_orphan_del(NULL, inode);
		}
N
Nick Piggin 已提交
1341

1342 1343 1344
		if (ret == -ENOSPC &&
		    ext4_should_retry_alloc(inode->i_sb, &retries))
			goto retry_journal;
1345
		put_page(page);
1346 1347 1348
		return ret;
	}
	*pagep = page;
1349 1350 1351
	return ret;
}

N
Nick Piggin 已提交
1352 1353
/* For write_end() in data=journal mode */
static int write_end_fn(handle_t *handle, struct buffer_head *bh)
1354
{
1355
	int ret;
1356 1357 1358
	if (!buffer_mapped(bh) || buffer_freed(bh))
		return 0;
	set_buffer_uptodate(bh);
1359 1360 1361 1362
	ret = ext4_handle_dirty_metadata(handle, NULL, bh);
	clear_buffer_meta(bh);
	clear_buffer_prio(bh);
	return ret;
1363 1364
}

1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375
/*
 * We need to pick up the new inode size which generic_commit_write gave us
 * `file' can be NULL - eg, when called from page_symlink().
 *
 * ext4 never places buffers on inode->i_mapping->private_list.  metadata
 * buffers are managed internally.
 */
static int ext4_write_end(struct file *file,
			  struct address_space *mapping,
			  loff_t pos, unsigned len, unsigned copied,
			  struct page *page, void *fsdata)
1376 1377
{
	handle_t *handle = ext4_journal_current_handle();
1378
	struct inode *inode = mapping->host;
1379
	loff_t old_size = inode->i_size;
1380 1381 1382 1383
	int ret = 0, ret2;
	int i_size_changed = 0;

	trace_ext4_write_end(inode, pos, len, copied);
1384 1385 1386
	if (ext4_has_inline_data(inode)) {
		ret = ext4_write_inline_data_end(inode, pos, len,
						 copied, page);
1387 1388 1389
		if (ret < 0) {
			unlock_page(page);
			put_page(page);
1390
			goto errout;
1391
		}
1392 1393
		copied = ret;
	} else
1394 1395
		copied = block_write_end(file, mapping, pos,
					 len, copied, page, fsdata);
1396
	/*
1397
	 * it's important to update i_size while still holding page lock:
1398 1399
	 * page writeout could otherwise come in and zero beyond i_size.
	 */
1400
	i_size_changed = ext4_update_inode_size(inode, pos + copied);
1401
	unlock_page(page);
1402
	put_page(page);
1403

1404 1405
	if (old_size < pos)
		pagecache_isize_extended(inode, old_size, pos);
1406 1407 1408 1409 1410 1411 1412 1413 1414
	/*
	 * Don't mark the inode dirty under page lock. First, it unnecessarily
	 * makes the holding time of page lock longer. Second, it forces lock
	 * ordering of page lock and transaction start for journaling
	 * filesystems.
	 */
	if (i_size_changed)
		ext4_mark_inode_dirty(handle, inode);

1415
	if (pos + len > inode->i_size && ext4_can_truncate(inode))
1416 1417 1418 1419 1420
		/* if we have allocated more blocks and copied
		 * less. We will have blocks allocated outside
		 * inode->i_size. So truncate them
		 */
		ext4_orphan_add(handle, inode);
1421
errout:
1422
	ret2 = ext4_journal_stop(handle);
1423 1424
	if (!ret)
		ret = ret2;
N
Nick Piggin 已提交
1425

1426
	if (pos + len > inode->i_size) {
1427
		ext4_truncate_failed_write(inode);
1428
		/*
1429
		 * If truncate failed early the inode might still be
1430 1431 1432 1433 1434 1435 1436
		 * on the orphan list; we need to make sure the inode
		 * is removed from the orphan list in that case.
		 */
		if (inode->i_nlink)
			ext4_orphan_del(NULL, inode);
	}

N
Nick Piggin 已提交
1437
	return ret ? ret : copied;
1438 1439
}

1440 1441 1442 1443 1444
/*
 * This is a private version of page_zero_new_buffers() which doesn't
 * set the buffer to be dirty, since in data=journalled mode we need
 * to call ext4_handle_dirty_metadata() instead.
 */
1445 1446 1447
static void ext4_journalled_zero_new_buffers(handle_t *handle,
					    struct page *page,
					    unsigned from, unsigned to)
1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463
{
	unsigned int block_start = 0, block_end;
	struct buffer_head *head, *bh;

	bh = head = page_buffers(page);
	do {
		block_end = block_start + bh->b_size;
		if (buffer_new(bh)) {
			if (block_end > from && block_start < to) {
				if (!PageUptodate(page)) {
					unsigned start, size;

					start = max(from, block_start);
					size = min(to, block_end) - start;

					zero_user(page, start, size);
1464
					write_end_fn(handle, bh);
1465 1466 1467 1468 1469 1470 1471 1472 1473
				}
				clear_buffer_new(bh);
			}
		}
		block_start = block_end;
		bh = bh->b_this_page;
	} while (bh != head);
}

N
Nick Piggin 已提交
1474
static int ext4_journalled_write_end(struct file *file,
1475 1476 1477
				     struct address_space *mapping,
				     loff_t pos, unsigned len, unsigned copied,
				     struct page *page, void *fsdata)
1478
{
1479
	handle_t *handle = ext4_journal_current_handle();
N
Nick Piggin 已提交
1480
	struct inode *inode = mapping->host;
1481
	loff_t old_size = inode->i_size;
1482 1483
	int ret = 0, ret2;
	int partial = 0;
N
Nick Piggin 已提交
1484
	unsigned from, to;
1485
	int size_changed = 0;
1486

1487
	trace_ext4_journalled_write_end(inode, pos, len, copied);
1488
	from = pos & (PAGE_SIZE - 1);
N
Nick Piggin 已提交
1489 1490
	to = from + len;

1491 1492
	BUG_ON(!ext4_handle_valid(handle));

1493 1494 1495 1496 1497 1498 1499 1500 1501 1502
	if (ext4_has_inline_data(inode)) {
		ret = ext4_write_inline_data_end(inode, pos, len,
						 copied, page);
		if (ret < 0) {
			unlock_page(page);
			put_page(page);
			goto errout;
		}
		copied = ret;
	} else if (unlikely(copied < len) && !PageUptodate(page)) {
1503 1504 1505 1506 1507 1508
		copied = 0;
		ext4_journalled_zero_new_buffers(handle, page, from, to);
	} else {
		if (unlikely(copied < len))
			ext4_journalled_zero_new_buffers(handle, page,
							 from + copied, to);
1509
		ret = ext4_walk_page_buffers(handle, page_buffers(page), from,
1510 1511
					     from + copied, &partial,
					     write_end_fn);
1512 1513 1514
		if (!partial)
			SetPageUptodate(page);
	}
1515
	size_changed = ext4_update_inode_size(inode, pos + copied);
1516
	ext4_set_inode_state(inode, EXT4_STATE_JDATA);
1517
	EXT4_I(inode)->i_datasync_tid = handle->h_transaction->t_tid;
1518
	unlock_page(page);
1519
	put_page(page);
1520

1521 1522 1523
	if (old_size < pos)
		pagecache_isize_extended(inode, old_size, pos);

1524
	if (size_changed) {
1525
		ret2 = ext4_mark_inode_dirty(handle, inode);
1526 1527 1528
		if (!ret)
			ret = ret2;
	}
N
Nick Piggin 已提交
1529

1530
	if (pos + len > inode->i_size && ext4_can_truncate(inode))
1531 1532 1533 1534 1535 1536
		/* if we have allocated more blocks and copied
		 * less. We will have blocks allocated outside
		 * inode->i_size. So truncate them
		 */
		ext4_orphan_add(handle, inode);

1537
errout:
1538
	ret2 = ext4_journal_stop(handle);
1539 1540
	if (!ret)
		ret = ret2;
1541
	if (pos + len > inode->i_size) {
1542
		ext4_truncate_failed_write(inode);
1543
		/*
1544
		 * If truncate failed early the inode might still be
1545 1546 1547 1548 1549 1550
		 * on the orphan list; we need to make sure the inode
		 * is removed from the orphan list in that case.
		 */
		if (inode->i_nlink)
			ext4_orphan_del(NULL, inode);
	}
N
Nick Piggin 已提交
1551 1552

	return ret ? ret : copied;
1553
}
1554

1555
/*
1556
 * Reserve space for a single cluster
1557
 */
1558
static int ext4_da_reserve_space(struct inode *inode)
1559
{
1560
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
1561
	struct ext4_inode_info *ei = EXT4_I(inode);
1562
	int ret;
1563 1564 1565 1566 1567 1568 1569 1570 1571

	/*
	 * We will charge metadata quota at writeout time; this saves
	 * us from metadata over-estimation, though we may go over by
	 * a small amount in the end.  Here we just reserve for data.
	 */
	ret = dquot_reserve_block(inode, EXT4_C2B(sbi, 1));
	if (ret)
		return ret;
1572

1573
	spin_lock(&ei->i_block_reservation_lock);
1574
	if (ext4_claim_free_clusters(sbi, 1, 0)) {
1575 1576
		spin_unlock(&ei->i_block_reservation_lock);
		dquot_release_reservation_block(inode, EXT4_C2B(sbi, 1));
1577 1578
		return -ENOSPC;
	}
1579
	ei->i_reserved_data_blocks++;
1580
	trace_ext4_da_reserve_space(inode);
1581
	spin_unlock(&ei->i_block_reservation_lock);
1582

1583 1584 1585
	return 0;       /* success */
}

1586
static void ext4_da_release_space(struct inode *inode, int to_free)
1587 1588
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
1589
	struct ext4_inode_info *ei = EXT4_I(inode);
1590

1591 1592 1593
	if (!to_free)
		return;		/* Nothing to release, exit */

1594
	spin_lock(&EXT4_I(inode)->i_block_reservation_lock);
1595

L
Li Zefan 已提交
1596
	trace_ext4_da_release_space(inode, to_free);
1597
	if (unlikely(to_free > ei->i_reserved_data_blocks)) {
1598
		/*
1599 1600 1601 1602
		 * if there aren't enough reserved blocks, then the
		 * counter is messed up somewhere.  Since this
		 * function is called from invalidate page, it's
		 * harmless to return without any action.
1603
		 */
1604
		ext4_warning(inode->i_sb, "ext4_da_release_space: "
1605
			 "ino %lu, to_free %d with only %d reserved "
1606
			 "data blocks", inode->i_ino, to_free,
1607 1608 1609
			 ei->i_reserved_data_blocks);
		WARN_ON(1);
		to_free = ei->i_reserved_data_blocks;
1610
	}
1611
	ei->i_reserved_data_blocks -= to_free;
1612

1613
	/* update fs dirty data blocks counter */
1614
	percpu_counter_sub(&sbi->s_dirtyclusters_counter, to_free);
1615 1616

	spin_unlock(&EXT4_I(inode)->i_block_reservation_lock);
1617

1618
	dquot_release_reservation_block(inode, EXT4_C2B(sbi, to_free));
1619 1620 1621
}

static void ext4_da_page_release_reservation(struct page *page,
1622 1623
					     unsigned int offset,
					     unsigned int length)
1624
{
1625
	int to_release = 0, contiguous_blks = 0;
1626 1627
	struct buffer_head *head, *bh;
	unsigned int curr_off = 0;
1628 1629
	struct inode *inode = page->mapping->host;
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
1630
	unsigned int stop = offset + length;
1631
	int num_clusters;
1632
	ext4_fsblk_t lblk;
1633

1634
	BUG_ON(stop > PAGE_SIZE || stop < length);
1635

1636 1637 1638 1639 1640
	head = page_buffers(page);
	bh = head;
	do {
		unsigned int next_off = curr_off + bh->b_size;

1641 1642 1643
		if (next_off > stop)
			break;

1644 1645
		if ((offset <= curr_off) && (buffer_delay(bh))) {
			to_release++;
1646
			contiguous_blks++;
1647
			clear_buffer_delay(bh);
1648 1649
		} else if (contiguous_blks) {
			lblk = page->index <<
1650
			       (PAGE_SHIFT - inode->i_blkbits);
1651 1652 1653 1654
			lblk += (curr_off >> inode->i_blkbits) -
				contiguous_blks;
			ext4_es_remove_extent(inode, lblk, contiguous_blks);
			contiguous_blks = 0;
1655 1656 1657
		}
		curr_off = next_off;
	} while ((bh = bh->b_this_page) != head);
1658

1659
	if (contiguous_blks) {
1660
		lblk = page->index << (PAGE_SHIFT - inode->i_blkbits);
1661 1662
		lblk += (curr_off >> inode->i_blkbits) - contiguous_blks;
		ext4_es_remove_extent(inode, lblk, contiguous_blks);
1663 1664
	}

1665 1666 1667 1668
	/* If we have released all the blocks belonging to a cluster, then we
	 * need to release the reserved space for that cluster. */
	num_clusters = EXT4_NUM_B2C(sbi, to_release);
	while (num_clusters > 0) {
1669
		lblk = (page->index << (PAGE_SHIFT - inode->i_blkbits)) +
1670 1671
			((num_clusters - 1) << sbi->s_cluster_bits);
		if (sbi->s_cluster_ratio == 1 ||
1672
		    !ext4_find_delalloc_cluster(inode, lblk))
1673 1674 1675 1676
			ext4_da_release_space(inode, 1);

		num_clusters--;
	}
1677
}
1678

1679 1680 1681 1682
/*
 * Delayed allocation stuff
 */

J
Jan Kara 已提交
1683 1684 1685
struct mpage_da_data {
	struct inode *inode;
	struct writeback_control *wbc;
1686

J
Jan Kara 已提交
1687 1688 1689
	pgoff_t first_page;	/* The first page to write */
	pgoff_t next_page;	/* Current page to examine */
	pgoff_t last_page;	/* Last page to examine */
1690
	/*
J
Jan Kara 已提交
1691 1692 1693
	 * Extent to map - this can be after first_page because that can be
	 * fully mapped. We somewhat abuse m_flags to store whether the extent
	 * is delalloc or unwritten.
1694
	 */
J
Jan Kara 已提交
1695 1696
	struct ext4_map_blocks map;
	struct ext4_io_submit io_submit;	/* IO submission data */
1697
	unsigned int do_map:1;
J
Jan Kara 已提交
1698
};
1699

J
Jan Kara 已提交
1700 1701
static void mpage_release_unused_pages(struct mpage_da_data *mpd,
				       bool invalidate)
1702 1703 1704 1705 1706 1707
{
	int nr_pages, i;
	pgoff_t index, end;
	struct pagevec pvec;
	struct inode *inode = mpd->inode;
	struct address_space *mapping = inode->i_mapping;
J
Jan Kara 已提交
1708 1709 1710 1711

	/* This is necessary when next_page == 0. */
	if (mpd->first_page >= mpd->next_page)
		return;
1712

1713 1714
	index = mpd->first_page;
	end   = mpd->next_page - 1;
J
Jan Kara 已提交
1715 1716
	if (invalidate) {
		ext4_lblk_t start, last;
1717 1718
		start = index << (PAGE_SHIFT - inode->i_blkbits);
		last = end << (PAGE_SHIFT - inode->i_blkbits);
J
Jan Kara 已提交
1719 1720
		ext4_es_remove_extent(inode, start, last - start + 1);
	}
1721

1722
	pagevec_init(&pvec, 0);
1723
	while (index <= end) {
1724
		nr_pages = pagevec_lookup_range(&pvec, mapping, &index, end);
1725 1726 1727 1728
		if (nr_pages == 0)
			break;
		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];
1729

1730 1731
			BUG_ON(!PageLocked(page));
			BUG_ON(PageWriteback(page));
J
Jan Kara 已提交
1732
			if (invalidate) {
1733 1734
				if (page_mapped(page))
					clear_page_dirty_for_io(page);
1735
				block_invalidatepage(page, 0, PAGE_SIZE);
J
Jan Kara 已提交
1736 1737
				ClearPageUptodate(page);
			}
1738 1739
			unlock_page(page);
		}
1740
		pagevec_release(&pvec);
1741 1742 1743
	}
}

1744 1745 1746
static void ext4_print_free_blocks(struct inode *inode)
{
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
1747
	struct super_block *sb = inode->i_sb;
1748
	struct ext4_inode_info *ei = EXT4_I(inode);
1749 1750

	ext4_msg(sb, KERN_CRIT, "Total free blocks count %lld",
1751
	       EXT4_C2B(EXT4_SB(inode->i_sb),
1752
			ext4_count_free_clusters(sb)));
1753 1754
	ext4_msg(sb, KERN_CRIT, "Free/Dirty block details");
	ext4_msg(sb, KERN_CRIT, "free_blocks=%lld",
1755
	       (long long) EXT4_C2B(EXT4_SB(sb),
1756
		percpu_counter_sum(&sbi->s_freeclusters_counter)));
1757
	ext4_msg(sb, KERN_CRIT, "dirty_blocks=%lld",
1758
	       (long long) EXT4_C2B(EXT4_SB(sb),
1759
		percpu_counter_sum(&sbi->s_dirtyclusters_counter)));
1760 1761
	ext4_msg(sb, KERN_CRIT, "Block reservation details");
	ext4_msg(sb, KERN_CRIT, "i_reserved_data_blocks=%u",
1762
		 ei->i_reserved_data_blocks);
1763 1764 1765
	return;
}

1766
static int ext4_bh_delay_or_unwritten(handle_t *handle, struct buffer_head *bh)
1767
{
1768
	return (buffer_delay(bh) || buffer_unwritten(bh)) && buffer_dirty(bh);
1769 1770
}

1771 1772 1773 1774 1775 1776 1777 1778 1779 1780
/*
 * This function is grabs code from the very beginning of
 * ext4_map_blocks, but assumes that the caller is from delayed write
 * time. This function looks up the requested blocks and sets the
 * buffer delay bit under the protection of i_data_sem.
 */
static int ext4_da_map_blocks(struct inode *inode, sector_t iblock,
			      struct ext4_map_blocks *map,
			      struct buffer_head *bh)
{
1781
	struct extent_status es;
1782 1783
	int retval;
	sector_t invalid_block = ~((sector_t) 0xffff);
1784 1785 1786 1787 1788
#ifdef ES_AGGRESSIVE_TEST
	struct ext4_map_blocks orig_map;

	memcpy(&orig_map, map, sizeof(*map));
#endif
1789 1790 1791 1792 1793 1794 1795 1796

	if (invalid_block < ext4_blocks_count(EXT4_SB(inode->i_sb)->s_es))
		invalid_block = ~0;

	map->m_flags = 0;
	ext_debug("ext4_da_map_blocks(): inode %lu, max_blocks %u,"
		  "logical block %lu\n", inode->i_ino, map->m_len,
		  (unsigned long) map->m_lblk);
1797 1798 1799 1800 1801

	/* Lookup extent status tree firstly */
	if (ext4_es_lookup_extent(inode, iblock, &es)) {
		if (ext4_es_is_hole(&es)) {
			retval = 0;
1802
			down_read(&EXT4_I(inode)->i_data_sem);
1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828
			goto add_delayed;
		}

		/*
		 * Delayed extent could be allocated by fallocate.
		 * So we need to check it.
		 */
		if (ext4_es_is_delayed(&es) && !ext4_es_is_unwritten(&es)) {
			map_bh(bh, inode->i_sb, invalid_block);
			set_buffer_new(bh);
			set_buffer_delay(bh);
			return 0;
		}

		map->m_pblk = ext4_es_pblock(&es) + iblock - es.es_lblk;
		retval = es.es_len - (iblock - es.es_lblk);
		if (retval > map->m_len)
			retval = map->m_len;
		map->m_len = retval;
		if (ext4_es_is_written(&es))
			map->m_flags |= EXT4_MAP_MAPPED;
		else if (ext4_es_is_unwritten(&es))
			map->m_flags |= EXT4_MAP_UNWRITTEN;
		else
			BUG_ON(1);

1829 1830 1831
#ifdef ES_AGGRESSIVE_TEST
		ext4_map_blocks_es_recheck(NULL, inode, map, &orig_map, 0);
#endif
1832 1833 1834
		return retval;
	}

1835 1836 1837 1838
	/*
	 * Try to see if we can get the block without requesting a new
	 * file system block.
	 */
1839
	down_read(&EXT4_I(inode)->i_data_sem);
1840
	if (ext4_has_inline_data(inode))
1841
		retval = 0;
1842
	else if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
1843
		retval = ext4_ext_map_blocks(NULL, inode, map, 0);
1844
	else
1845
		retval = ext4_ind_map_blocks(NULL, inode, map, 0);
1846

1847
add_delayed:
1848
	if (retval == 0) {
1849
		int ret;
1850 1851 1852 1853
		/*
		 * XXX: __block_prepare_write() unmaps passed block,
		 * is it OK?
		 */
1854 1855 1856 1857 1858
		/*
		 * If the block was allocated from previously allocated cluster,
		 * then we don't need to reserve it again. However we still need
		 * to reserve metadata for every block we're going to write.
		 */
1859
		if (EXT4_SB(inode->i_sb)->s_cluster_ratio == 1 ||
1860
		    !ext4_find_delalloc_cluster(inode, map->m_lblk)) {
1861
			ret = ext4_da_reserve_space(inode);
1862
			if (ret) {
1863
				/* not enough space to reserve */
1864
				retval = ret;
1865
				goto out_unlock;
1866
			}
1867 1868
		}

1869 1870 1871 1872
		ret = ext4_es_insert_extent(inode, map->m_lblk, map->m_len,
					    ~0, EXTENT_STATUS_DELAYED);
		if (ret) {
			retval = ret;
1873
			goto out_unlock;
1874
		}
1875

1876 1877 1878
		map_bh(bh, inode->i_sb, invalid_block);
		set_buffer_new(bh);
		set_buffer_delay(bh);
1879 1880
	} else if (retval > 0) {
		int ret;
1881
		unsigned int status;
1882

1883 1884 1885 1886 1887 1888
		if (unlikely(retval != map->m_len)) {
			ext4_warning(inode->i_sb,
				     "ES len assertion failed for inode "
				     "%lu: retval %d != map->m_len %d",
				     inode->i_ino, retval, map->m_len);
			WARN_ON(1);
1889 1890
		}

1891 1892 1893 1894 1895 1896
		status = map->m_flags & EXT4_MAP_UNWRITTEN ?
				EXTENT_STATUS_UNWRITTEN : EXTENT_STATUS_WRITTEN;
		ret = ext4_es_insert_extent(inode, map->m_lblk, map->m_len,
					    map->m_pblk, status);
		if (ret != 0)
			retval = ret;
1897 1898 1899 1900 1901 1902 1903 1904
	}

out_unlock:
	up_read((&EXT4_I(inode)->i_data_sem));

	return retval;
}

1905
/*
1906
 * This is a special get_block_t callback which is used by
1907 1908
 * ext4_da_write_begin().  It will either return mapped block or
 * reserve space for a single block.
1909 1910 1911 1912 1913 1914 1915
 *
 * For delayed buffer_head we have BH_Mapped, BH_New, BH_Delay set.
 * We also have b_blocknr = -1 and b_bdev initialized properly
 *
 * For unwritten buffer_head we have BH_Mapped, BH_New, BH_Unwritten set.
 * We also have b_blocknr = physicalblock mapping unwritten extent and b_bdev
 * initialized properly.
1916
 */
1917 1918
int ext4_da_get_block_prep(struct inode *inode, sector_t iblock,
			   struct buffer_head *bh, int create)
1919
{
1920
	struct ext4_map_blocks map;
1921 1922 1923
	int ret = 0;

	BUG_ON(create == 0);
1924 1925 1926 1927
	BUG_ON(bh->b_size != inode->i_sb->s_blocksize);

	map.m_lblk = iblock;
	map.m_len = 1;
1928 1929 1930 1931 1932 1933

	/*
	 * first, we need to know whether the block is allocated already
	 * preallocated blocks are unmapped but should treated
	 * the same as allocated blocks.
	 */
1934 1935
	ret = ext4_da_map_blocks(inode, iblock, &map, bh);
	if (ret <= 0)
1936
		return ret;
1937

1938
	map_bh(bh, inode->i_sb, map.m_pblk);
J
Jan Kara 已提交
1939
	ext4_update_bh_state(bh, map.m_flags);
1940 1941 1942 1943 1944 1945 1946 1947 1948

	if (buffer_unwritten(bh)) {
		/* A delayed write to unwritten bh should be marked
		 * new and mapped.  Mapped ensures that we don't do
		 * get_block multiple times when we write to the same
		 * offset and new ensures that we do proper zero out
		 * for partial write.
		 */
		set_buffer_new(bh);
1949
		set_buffer_mapped(bh);
1950 1951
	}
	return 0;
1952
}
1953

1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970
static int bget_one(handle_t *handle, struct buffer_head *bh)
{
	get_bh(bh);
	return 0;
}

static int bput_one(handle_t *handle, struct buffer_head *bh)
{
	put_bh(bh);
	return 0;
}

static int __ext4_journalled_writepage(struct page *page,
				       unsigned int len)
{
	struct address_space *mapping = page->mapping;
	struct inode *inode = mapping->host;
1971
	struct buffer_head *page_bufs = NULL;
1972
	handle_t *handle = NULL;
1973 1974 1975
	int ret = 0, err = 0;
	int inline_data = ext4_has_inline_data(inode);
	struct buffer_head *inode_bh = NULL;
1976

1977
	ClearPageChecked(page);
1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993

	if (inline_data) {
		BUG_ON(page->index != 0);
		BUG_ON(len > ext4_get_max_inline_size(inode));
		inode_bh = ext4_journalled_write_inline_data(inode, len, page);
		if (inode_bh == NULL)
			goto out;
	} else {
		page_bufs = page_buffers(page);
		if (!page_bufs) {
			BUG();
			goto out;
		}
		ext4_walk_page_buffers(handle, page_bufs, 0, len,
				       NULL, bget_one);
	}
1994 1995 1996 1997 1998 1999
	/*
	 * We need to release the page lock before we start the
	 * journal, so grab a reference so the page won't disappear
	 * out from under us.
	 */
	get_page(page);
2000 2001
	unlock_page(page);

2002 2003
	handle = ext4_journal_start(inode, EXT4_HT_WRITE_PAGE,
				    ext4_writepage_trans_blocks(inode));
2004 2005
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
2006 2007
		put_page(page);
		goto out_no_pagelock;
2008
	}
2009 2010
	BUG_ON(!ext4_handle_valid(handle));

2011 2012 2013 2014 2015 2016 2017 2018 2019
	lock_page(page);
	put_page(page);
	if (page->mapping != mapping) {
		/* The page got truncated from under us */
		ext4_journal_stop(handle);
		ret = 0;
		goto out;
	}

2020
	if (inline_data) {
2021
		BUFFER_TRACE(inode_bh, "get write access");
2022
		ret = ext4_journal_get_write_access(handle, inode_bh);
2023

2024 2025 2026 2027 2028 2029 2030 2031 2032
		err = ext4_handle_dirty_metadata(handle, inode, inode_bh);

	} else {
		ret = ext4_walk_page_buffers(handle, page_bufs, 0, len, NULL,
					     do_journal_get_write_access);

		err = ext4_walk_page_buffers(handle, page_bufs, 0, len, NULL,
					     write_end_fn);
	}
2033 2034
	if (ret == 0)
		ret = err;
2035
	EXT4_I(inode)->i_datasync_tid = handle->h_transaction->t_tid;
2036 2037 2038 2039
	err = ext4_journal_stop(handle);
	if (!ret)
		ret = err;

2040
	if (!ext4_has_inline_data(inode))
2041
		ext4_walk_page_buffers(NULL, page_bufs, 0, len,
2042
				       NULL, bput_one);
2043
	ext4_set_inode_state(inode, EXT4_STATE_JDATA);
2044
out:
2045 2046
	unlock_page(page);
out_no_pagelock:
2047
	brelse(inode_bh);
2048 2049 2050
	return ret;
}

2051
/*
2052 2053 2054 2055
 * Note that we don't need to start a transaction unless we're journaling data
 * because we should have holes filled from ext4_page_mkwrite(). We even don't
 * need to file the inode to the transaction's list in ordered mode because if
 * we are writing back data added by write(), the inode is already there and if
L
Lucas De Marchi 已提交
2056
 * we are writing back data modified via mmap(), no one guarantees in which
2057 2058 2059 2060
 * transaction the data will hit the disk. In case we are journaling data, we
 * cannot start transaction directly because transaction start ranks above page
 * lock so we have to do some magic.
 *
2061
 * This function can get called via...
2062
 *   - ext4_writepages after taking page lock (have journal handle)
2063
 *   - journal_submit_inode_data_buffers (no journal handle)
2064
 *   - shrink_page_list via the kswapd/direct reclaim (no journal handle)
2065
 *   - grab_page_cache when doing write_begin (have journal handle)
2066 2067 2068 2069 2070 2071 2072 2073 2074
 *
 * We don't do any block allocation in this function. If we have page with
 * multiple blocks we need to write those buffer_heads that are mapped. This
 * is important for mmaped based write. So if we do with blocksize 1K
 * truncate(f, 1024);
 * a = mmap(f, 0, 4096);
 * a[0] = 'a';
 * truncate(f, 4096);
 * we have in the page first buffer_head mapped via page_mkwrite call back
2075
 * but other buffer_heads would be unmapped but dirty (dirty done via the
2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090
 * do_wp_page). So writepage should write the first block. If we modify
 * the mmap area beyond 1024 we will again get a page_fault and the
 * page_mkwrite callback will do the block allocation and mark the
 * buffer_heads mapped.
 *
 * We redirty the page if we have any buffer_heads that is either delay or
 * unwritten in the page.
 *
 * We can get recursively called as show below.
 *
 *	ext4_writepage() -> kmalloc() -> __alloc_pages() -> page_launder() ->
 *		ext4_writepage()
 *
 * But since we don't do any block allocation we should not deadlock.
 * Page also have the dirty flag cleared so we don't get recurive page_lock.
2091
 */
2092
static int ext4_writepage(struct page *page,
2093
			  struct writeback_control *wbc)
2094
{
2095
	int ret = 0;
2096
	loff_t size;
2097
	unsigned int len;
2098
	struct buffer_head *page_bufs = NULL;
2099
	struct inode *inode = page->mapping->host;
2100
	struct ext4_io_submit io_submit;
2101
	bool keep_towrite = false;
2102

2103 2104 2105 2106 2107 2108
	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb)))) {
		ext4_invalidatepage(page, 0, PAGE_SIZE);
		unlock_page(page);
		return -EIO;
	}

L
Lukas Czerner 已提交
2109
	trace_ext4_writepage(page);
2110
	size = i_size_read(inode);
2111 2112
	if (page->index == size >> PAGE_SHIFT)
		len = size & ~PAGE_MASK;
2113
	else
2114
		len = PAGE_SIZE;
2115

T
Theodore Ts'o 已提交
2116 2117
	page_bufs = page_buffers(page);
	/*
2118 2119 2120 2121 2122
	 * We cannot do block allocation or other extent handling in this
	 * function. If there are buffers needing that, we have to redirty
	 * the page. But we may reach here when we do a journal commit via
	 * journal_submit_inode_data_buffers() and in that case we must write
	 * allocated buffers to achieve data=ordered mode guarantees.
2123 2124 2125 2126 2127 2128 2129 2130 2131 2132
	 *
	 * Also, if there is only one buffer per page (the fs block
	 * size == the page size), if one buffer needs block
	 * allocation or needs to modify the extent tree to clear the
	 * unwritten flag, we know that the page can't be written at
	 * all, so we might as well refuse the write immediately.
	 * Unfortunately if the block size != page size, we can't as
	 * easily detect this case using ext4_walk_page_buffers(), but
	 * for the extremely common case, this is an optimization that
	 * skips a useless round trip through ext4_bio_write_page().
T
Theodore Ts'o 已提交
2133
	 */
2134 2135
	if (ext4_walk_page_buffers(NULL, page_bufs, 0, len, NULL,
				   ext4_bh_delay_or_unwritten)) {
2136
		redirty_page_for_writepage(wbc, page);
2137
		if ((current->flags & PF_MEMALLOC) ||
2138
		    (inode->i_sb->s_blocksize == PAGE_SIZE)) {
2139 2140 2141 2142 2143 2144 2145
			/*
			 * For memory cleaning there's no point in writing only
			 * some buffers. So just bail out. Warn if we came here
			 * from direct reclaim.
			 */
			WARN_ON_ONCE((current->flags & (PF_MEMALLOC|PF_KSWAPD))
							== PF_MEMALLOC);
2146 2147 2148
			unlock_page(page);
			return 0;
		}
2149
		keep_towrite = true;
T
Theodore Ts'o 已提交
2150
	}
2151

2152
	if (PageChecked(page) && ext4_should_journal_data(inode))
2153 2154 2155 2156
		/*
		 * It's mmapped pagecache.  Add buffers and journal it.  There
		 * doesn't seem much point in redirtying the page here.
		 */
2157
		return __ext4_journalled_writepage(page, len);
2158

J
Jan Kara 已提交
2159 2160 2161 2162 2163 2164 2165
	ext4_io_submit_init(&io_submit, wbc);
	io_submit.io_end = ext4_init_io_end(inode, GFP_NOFS);
	if (!io_submit.io_end) {
		redirty_page_for_writepage(wbc, page);
		unlock_page(page);
		return -ENOMEM;
	}
2166
	ret = ext4_bio_write_page(&io_submit, page, len, wbc, keep_towrite);
2167
	ext4_io_submit(&io_submit);
J
Jan Kara 已提交
2168 2169
	/* Drop io_end reference we got from init */
	ext4_put_io_end_defer(io_submit.io_end);
2170 2171 2172
	return ret;
}

2173 2174 2175
static int mpage_submit_page(struct mpage_da_data *mpd, struct page *page)
{
	int len;
2176
	loff_t size;
2177 2178 2179
	int err;

	BUG_ON(page->index != mpd->first_page);
2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194
	clear_page_dirty_for_io(page);
	/*
	 * We have to be very careful here!  Nothing protects writeback path
	 * against i_size changes and the page can be writeably mapped into
	 * page tables. So an application can be growing i_size and writing
	 * data through mmap while writeback runs. clear_page_dirty_for_io()
	 * write-protects our page in page tables and the page cannot get
	 * written to again until we release page lock. So only after
	 * clear_page_dirty_for_io() we are safe to sample i_size for
	 * ext4_bio_write_page() to zero-out tail of the written page. We rely
	 * on the barrier provided by TestClearPageDirty in
	 * clear_page_dirty_for_io() to make sure i_size is really sampled only
	 * after page tables are updated.
	 */
	size = i_size_read(mpd->inode);
2195 2196
	if (page->index == size >> PAGE_SHIFT)
		len = size & ~PAGE_MASK;
2197
	else
2198
		len = PAGE_SIZE;
2199
	err = ext4_bio_write_page(&mpd->io_submit, page, len, mpd->wbc, false);
2200 2201 2202 2203 2204 2205 2206
	if (!err)
		mpd->wbc->nr_to_write--;
	mpd->first_page++;

	return err;
}

J
Jan Kara 已提交
2207 2208
#define BH_FLAGS ((1 << BH_Unwritten) | (1 << BH_Delay))

2209
/*
2210 2211
 * mballoc gives us at most this number of blocks...
 * XXX: That seems to be only a limitation of ext4_mb_normalize_request().
2212
 * The rest of mballoc seems to handle chunks up to full group size.
2213
 */
2214
#define MAX_WRITEPAGES_EXTENT_LEN 2048
2215

J
Jan Kara 已提交
2216 2217 2218 2219 2220
/*
 * mpage_add_bh_to_extent - try to add bh to extent of blocks to map
 *
 * @mpd - extent of blocks
 * @lblk - logical number of the block in the file
2221
 * @bh - buffer head we want to add to the extent
J
Jan Kara 已提交
2222
 *
2223 2224 2225 2226 2227 2228
 * The function is used to collect contig. blocks in the same state. If the
 * buffer doesn't require mapping for writeback and we haven't started the
 * extent of buffers to map yet, the function returns 'true' immediately - the
 * caller can write the buffer right away. Otherwise the function returns true
 * if the block has been added to the extent, false if the block couldn't be
 * added.
J
Jan Kara 已提交
2229
 */
2230 2231
static bool mpage_add_bh_to_extent(struct mpage_da_data *mpd, ext4_lblk_t lblk,
				   struct buffer_head *bh)
J
Jan Kara 已提交
2232 2233 2234
{
	struct ext4_map_blocks *map = &mpd->map;

2235 2236 2237 2238 2239 2240 2241 2242
	/* Buffer that doesn't need mapping for writeback? */
	if (!buffer_dirty(bh) || !buffer_mapped(bh) ||
	    (!buffer_delay(bh) && !buffer_unwritten(bh))) {
		/* So far no extent to map => we write the buffer right away */
		if (map->m_len == 0)
			return true;
		return false;
	}
J
Jan Kara 已提交
2243 2244 2245

	/* First block in the extent? */
	if (map->m_len == 0) {
2246 2247 2248
		/* We cannot map unless handle is started... */
		if (!mpd->do_map)
			return false;
J
Jan Kara 已提交
2249 2250
		map->m_lblk = lblk;
		map->m_len = 1;
2251 2252
		map->m_flags = bh->b_state & BH_FLAGS;
		return true;
J
Jan Kara 已提交
2253 2254
	}

2255 2256 2257 2258
	/* Don't go larger than mballoc is willing to allocate */
	if (map->m_len >= MAX_WRITEPAGES_EXTENT_LEN)
		return false;

J
Jan Kara 已提交
2259 2260
	/* Can we merge the block to our big extent? */
	if (lblk == map->m_lblk + map->m_len &&
2261
	    (bh->b_state & BH_FLAGS) == map->m_flags) {
J
Jan Kara 已提交
2262
		map->m_len++;
2263
		return true;
J
Jan Kara 已提交
2264
	}
2265
	return false;
J
Jan Kara 已提交
2266 2267
}

2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287
/*
 * mpage_process_page_bufs - submit page buffers for IO or add them to extent
 *
 * @mpd - extent of blocks for mapping
 * @head - the first buffer in the page
 * @bh - buffer we should start processing from
 * @lblk - logical number of the block in the file corresponding to @bh
 *
 * Walk through page buffers from @bh upto @head (exclusive) and either submit
 * the page for IO if all buffers in this page were mapped and there's no
 * accumulated extent of buffers to map or add buffers in the page to the
 * extent of buffers to map. The function returns 1 if the caller can continue
 * by processing the next page, 0 if it should stop adding buffers to the
 * extent to map because we cannot extend it anymore. It can also return value
 * < 0 in case of error during IO submission.
 */
static int mpage_process_page_bufs(struct mpage_da_data *mpd,
				   struct buffer_head *head,
				   struct buffer_head *bh,
				   ext4_lblk_t lblk)
J
Jan Kara 已提交
2288 2289
{
	struct inode *inode = mpd->inode;
2290
	int err;
F
Fabian Frederick 已提交
2291
	ext4_lblk_t blocks = (i_size_read(inode) + i_blocksize(inode) - 1)
J
Jan Kara 已提交
2292 2293 2294 2295 2296
							>> inode->i_blkbits;

	do {
		BUG_ON(buffer_locked(bh));

2297
		if (lblk >= blocks || !mpage_add_bh_to_extent(mpd, lblk, bh)) {
J
Jan Kara 已提交
2298 2299
			/* Found extent to map? */
			if (mpd->map.m_len)
2300
				return 0;
2301 2302 2303
			/* Buffer needs mapping and handle is not started? */
			if (!mpd->do_map)
				return 0;
2304
			/* Everything mapped so far and we hit EOF */
2305
			break;
J
Jan Kara 已提交
2306 2307
		}
	} while (lblk++, (bh = bh->b_this_page) != head);
2308 2309 2310 2311 2312 2313 2314
	/* So far everything mapped? Submit the page for IO. */
	if (mpd->map.m_len == 0) {
		err = mpage_submit_page(mpd, head->b_page);
		if (err < 0)
			return err;
	}
	return lblk < blocks;
J
Jan Kara 已提交
2315 2316 2317 2318 2319 2320 2321 2322 2323 2324 2325
}

/*
 * mpage_map_buffers - update buffers corresponding to changed extent and
 *		       submit fully mapped pages for IO
 *
 * @mpd - description of extent to map, on return next extent to map
 *
 * Scan buffers corresponding to changed extent (we expect corresponding pages
 * to be already locked) and update buffer state according to new extent state.
 * We map delalloc buffers to their physical location, clear unwritten bits,
2326
 * and mark buffers as uninit when we perform writes to unwritten extents
J
Jan Kara 已提交
2327 2328 2329 2330 2331 2332 2333 2334 2335 2336
 * and do extent conversion after IO is finished. If the last page is not fully
 * mapped, we update @map to the next extent in the last page that needs
 * mapping. Otherwise we submit the page for IO.
 */
static int mpage_map_and_submit_buffers(struct mpage_da_data *mpd)
{
	struct pagevec pvec;
	int nr_pages, i;
	struct inode *inode = mpd->inode;
	struct buffer_head *head, *bh;
2337
	int bpp_bits = PAGE_SHIFT - inode->i_blkbits;
J
Jan Kara 已提交
2338 2339 2340 2341 2342 2343 2344 2345 2346 2347 2348 2349
	pgoff_t start, end;
	ext4_lblk_t lblk;
	sector_t pblock;
	int err;

	start = mpd->map.m_lblk >> bpp_bits;
	end = (mpd->map.m_lblk + mpd->map.m_len - 1) >> bpp_bits;
	lblk = start << bpp_bits;
	pblock = mpd->map.m_pblk;

	pagevec_init(&pvec, 0);
	while (start <= end) {
2350
		nr_pages = pagevec_lookup_range(&pvec, inode->i_mapping,
2351
						&start, end);
J
Jan Kara 已提交
2352 2353 2354 2355 2356 2357 2358 2359 2360 2361 2362 2363 2364 2365 2366 2367
		if (nr_pages == 0)
			break;
		for (i = 0; i < nr_pages; i++) {
			struct page *page = pvec.pages[i];

			bh = head = page_buffers(page);
			do {
				if (lblk < mpd->map.m_lblk)
					continue;
				if (lblk >= mpd->map.m_lblk + mpd->map.m_len) {
					/*
					 * Buffer after end of mapped extent.
					 * Find next buffer in the page to map.
					 */
					mpd->map.m_len = 0;
					mpd->map.m_flags = 0;
2368 2369 2370 2371 2372 2373 2374 2375 2376
					/*
					 * FIXME: If dioread_nolock supports
					 * blocksize < pagesize, we need to make
					 * sure we add size mapped so far to
					 * io_end->size as the following call
					 * can submit the page for IO.
					 */
					err = mpage_process_page_bufs(mpd, head,
								      bh, lblk);
J
Jan Kara 已提交
2377
					pagevec_release(&pvec);
2378 2379 2380
					if (err > 0)
						err = 0;
					return err;
J
Jan Kara 已提交
2381 2382 2383 2384 2385 2386
				}
				if (buffer_delay(bh)) {
					clear_buffer_delay(bh);
					bh->b_blocknr = pblock++;
				}
				clear_buffer_unwritten(bh);
2387
			} while (lblk++, (bh = bh->b_this_page) != head);
J
Jan Kara 已提交
2388 2389 2390 2391 2392 2393

			/*
			 * FIXME: This is going to break if dioread_nolock
			 * supports blocksize < pagesize as we will try to
			 * convert potentially unmapped parts of inode.
			 */
2394
			mpd->io_submit.io_end->size += PAGE_SIZE;
J
Jan Kara 已提交
2395 2396 2397 2398 2399 2400 2401 2402 2403 2404 2405 2406 2407 2408 2409 2410 2411 2412 2413 2414
			/* Page fully mapped - let IO run! */
			err = mpage_submit_page(mpd, page);
			if (err < 0) {
				pagevec_release(&pvec);
				return err;
			}
		}
		pagevec_release(&pvec);
	}
	/* Extent fully mapped and matches with page boundary. We are done. */
	mpd->map.m_len = 0;
	mpd->map.m_flags = 0;
	return 0;
}

static int mpage_map_one_extent(handle_t *handle, struct mpage_da_data *mpd)
{
	struct inode *inode = mpd->inode;
	struct ext4_map_blocks *map = &mpd->map;
	int get_blocks_flags;
2415
	int err, dioread_nolock;
J
Jan Kara 已提交
2416 2417 2418 2419

	trace_ext4_da_write_pages_extent(inode, map);
	/*
	 * Call ext4_map_blocks() to allocate any delayed allocation blocks, or
2420
	 * to convert an unwritten extent to be initialized (in the case
J
Jan Kara 已提交
2421 2422 2423 2424 2425 2426 2427
	 * where we have written into one or more preallocated blocks).  It is
	 * possible that we're going to need more metadata blocks than
	 * previously reserved. However we must not fail because we're in
	 * writeback and there is nothing we can do about it so it might result
	 * in data loss.  So use reserved blocks to allocate metadata if
	 * possible.
	 *
2428 2429 2430 2431
	 * We pass in the magic EXT4_GET_BLOCKS_DELALLOC_RESERVE if
	 * the blocks in question are delalloc blocks.  This indicates
	 * that the blocks and quotas has already been checked when
	 * the data was copied into the page cache.
J
Jan Kara 已提交
2432 2433
	 */
	get_blocks_flags = EXT4_GET_BLOCKS_CREATE |
2434 2435
			   EXT4_GET_BLOCKS_METADATA_NOFAIL |
			   EXT4_GET_BLOCKS_IO_SUBMIT;
2436 2437
	dioread_nolock = ext4_should_dioread_nolock(inode);
	if (dioread_nolock)
J
Jan Kara 已提交
2438 2439 2440 2441 2442 2443 2444
		get_blocks_flags |= EXT4_GET_BLOCKS_IO_CREATE_EXT;
	if (map->m_flags & (1 << BH_Delay))
		get_blocks_flags |= EXT4_GET_BLOCKS_DELALLOC_RESERVE;

	err = ext4_map_blocks(handle, inode, map, get_blocks_flags);
	if (err < 0)
		return err;
2445
	if (dioread_nolock && (map->m_flags & EXT4_MAP_UNWRITTEN)) {
2446 2447 2448 2449 2450
		if (!mpd->io_submit.io_end->handle &&
		    ext4_handle_valid(handle)) {
			mpd->io_submit.io_end->handle = handle->h_rsv_handle;
			handle->h_rsv_handle = NULL;
		}
J
Jan Kara 已提交
2451
		ext4_set_io_unwritten_flag(inode, mpd->io_submit.io_end);
2452
	}
J
Jan Kara 已提交
2453 2454 2455

	BUG_ON(map->m_len == 0);
	if (map->m_flags & EXT4_MAP_NEW) {
2456 2457
		clean_bdev_aliases(inode->i_sb->s_bdev, map->m_pblk,
				   map->m_len);
J
Jan Kara 已提交
2458 2459 2460 2461 2462 2463 2464 2465 2466 2467
	}
	return 0;
}

/*
 * mpage_map_and_submit_extent - map extent starting at mpd->lblk of length
 *				 mpd->len and submit pages underlying it for IO
 *
 * @handle - handle for journal operations
 * @mpd - extent to map
2468 2469 2470
 * @give_up_on_write - we set this to true iff there is a fatal error and there
 *                     is no hope of writing the data. The caller should discard
 *                     dirty pages to avoid infinite loops.
J
Jan Kara 已提交
2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482
 *
 * The function maps extent starting at mpd->lblk of length mpd->len. If it is
 * delayed, blocks are allocated, if it is unwritten, we may need to convert
 * them to initialized or split the described range from larger unwritten
 * extent. Note that we need not map all the described range since allocation
 * can return less blocks or the range is covered by more unwritten extents. We
 * cannot map more because we are limited by reserved transaction credits. On
 * the other hand we always make sure that the last touched page is fully
 * mapped so that it can be written out (and thus forward progress is
 * guaranteed). After mapping we submit all mapped pages for IO.
 */
static int mpage_map_and_submit_extent(handle_t *handle,
2483 2484
				       struct mpage_da_data *mpd,
				       bool *give_up_on_write)
J
Jan Kara 已提交
2485 2486 2487 2488 2489
{
	struct inode *inode = mpd->inode;
	struct ext4_map_blocks *map = &mpd->map;
	int err;
	loff_t disksize;
2490
	int progress = 0;
J
Jan Kara 已提交
2491 2492 2493

	mpd->io_submit.io_end->offset =
				((loff_t)map->m_lblk) << inode->i_blkbits;
2494
	do {
J
Jan Kara 已提交
2495 2496 2497 2498
		err = mpage_map_one_extent(handle, mpd);
		if (err < 0) {
			struct super_block *sb = inode->i_sb;

2499 2500
			if (ext4_forced_shutdown(EXT4_SB(sb)) ||
			    EXT4_SB(sb)->s_mount_flags & EXT4_MF_FS_ABORTED)
2501
				goto invalidate_dirty_pages;
J
Jan Kara 已提交
2502
			/*
2503 2504 2505
			 * Let the uper layers retry transient errors.
			 * In the case of ENOSPC, if ext4_count_free_blocks()
			 * is non-zero, a commit should free up blocks.
J
Jan Kara 已提交
2506
			 */
2507
			if ((err == -ENOMEM) ||
2508 2509 2510
			    (err == -ENOSPC && ext4_count_free_clusters(sb))) {
				if (progress)
					goto update_disksize;
2511
				return err;
2512
			}
2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526
			ext4_msg(sb, KERN_CRIT,
				 "Delayed block allocation failed for "
				 "inode %lu at logical offset %llu with"
				 " max blocks %u with error %d",
				 inode->i_ino,
				 (unsigned long long)map->m_lblk,
				 (unsigned)map->m_len, -err);
			ext4_msg(sb, KERN_CRIT,
				 "This should not happen!! Data will "
				 "be lost\n");
			if (err == -ENOSPC)
				ext4_print_free_blocks(inode);
		invalidate_dirty_pages:
			*give_up_on_write = true;
J
Jan Kara 已提交
2527 2528
			return err;
		}
2529
		progress = 1;
J
Jan Kara 已提交
2530 2531 2532 2533 2534 2535
		/*
		 * Update buffer state, submit mapped pages, and get us new
		 * extent to map
		 */
		err = mpage_map_and_submit_buffers(mpd);
		if (err < 0)
2536
			goto update_disksize;
2537
	} while (map->m_len);
J
Jan Kara 已提交
2538

2539
update_disksize:
2540 2541 2542 2543
	/*
	 * Update on-disk size after IO is submitted.  Races with
	 * truncate are avoided by checking i_size under i_data_sem.
	 */
2544
	disksize = ((loff_t)mpd->first_page) << PAGE_SHIFT;
J
Jan Kara 已提交
2545 2546
	if (disksize > EXT4_I(inode)->i_disksize) {
		int err2;
2547 2548 2549 2550 2551 2552 2553 2554 2555
		loff_t i_size;

		down_write(&EXT4_I(inode)->i_data_sem);
		i_size = i_size_read(inode);
		if (disksize > i_size)
			disksize = i_size;
		if (disksize > EXT4_I(inode)->i_disksize)
			EXT4_I(inode)->i_disksize = disksize;
		up_write(&EXT4_I(inode)->i_data_sem);
2556
		err2 = ext4_mark_inode_dirty(handle, inode);
J
Jan Kara 已提交
2557 2558 2559 2560 2561 2562 2563 2564 2565 2566
		if (err2)
			ext4_error(inode->i_sb,
				   "Failed to mark inode %lu dirty",
				   inode->i_ino);
		if (!err)
			err = err2;
	}
	return err;
}

2567 2568
/*
 * Calculate the total number of credits to reserve for one writepages
2569
 * iteration. This is called from ext4_writepages(). We map an extent of
2570
 * up to MAX_WRITEPAGES_EXTENT_LEN blocks and then we go on and finish mapping
2571 2572 2573
 * the last partial page. So in total we can map MAX_WRITEPAGES_EXTENT_LEN +
 * bpp - 1 blocks in bpp different extents.
 */
2574 2575
static int ext4_da_writepages_trans_blocks(struct inode *inode)
{
2576
	int bpp = ext4_journal_blocks_per_page(inode);
2577

2578 2579
	return ext4_meta_trans_blocks(inode,
				MAX_WRITEPAGES_EXTENT_LEN + bpp - 1, bpp);
2580
}
2581

2582
/*
J
Jan Kara 已提交
2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598
 * mpage_prepare_extent_to_map - find & lock contiguous range of dirty pages
 * 				 and underlying extent to map
 *
 * @mpd - where to look for pages
 *
 * Walk dirty pages in the mapping. If they are fully mapped, submit them for
 * IO immediately. When we find a page which isn't mapped we start accumulating
 * extent of buffers underlying these pages that needs mapping (formed by
 * either delayed or unwritten buffers). We also lock the pages containing
 * these buffers. The extent found is returned in @mpd structure (starting at
 * mpd->lblk with length mpd->len blocks).
 *
 * Note that this function can attach bios to one io_end structure which are
 * neither logically nor physically contiguous. Although it may seem as an
 * unnecessary complication, it is actually inevitable in blocksize < pagesize
 * case as we need to track IO to all buffers underlying a page in one io_end.
2599
 */
J
Jan Kara 已提交
2600
static int mpage_prepare_extent_to_map(struct mpage_da_data *mpd)
2601
{
J
Jan Kara 已提交
2602 2603 2604
	struct address_space *mapping = mpd->inode->i_mapping;
	struct pagevec pvec;
	unsigned int nr_pages;
2605
	long left = mpd->wbc->nr_to_write;
J
Jan Kara 已提交
2606 2607 2608 2609 2610 2611 2612
	pgoff_t index = mpd->first_page;
	pgoff_t end = mpd->last_page;
	int tag;
	int i, err = 0;
	int blkbits = mpd->inode->i_blkbits;
	ext4_lblk_t lblk;
	struct buffer_head *head;
2613

J
Jan Kara 已提交
2614
	if (mpd->wbc->sync_mode == WB_SYNC_ALL || mpd->wbc->tagged_writepages)
2615 2616 2617 2618
		tag = PAGECACHE_TAG_TOWRITE;
	else
		tag = PAGECACHE_TAG_DIRTY;

J
Jan Kara 已提交
2619 2620 2621
	pagevec_init(&pvec, 0);
	mpd->map.m_len = 0;
	mpd->next_page = index;
2622
	while (index <= end) {
2623
		nr_pages = pagevec_lookup_tag(&pvec, mapping, &index, tag,
2624 2625
			      min(end - index, (pgoff_t)PAGEVEC_SIZE-1) + 1);
		if (nr_pages == 0)
J
Jan Kara 已提交
2626
			goto out;
2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637

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

			/*
			 * 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.
			 */
2638 2639
			if (page->index > end)
				goto out;
2640

2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651
			/*
			 * Accumulated enough dirty pages? This doesn't apply
			 * to WB_SYNC_ALL mode. For integrity sync we have to
			 * keep going because someone may be concurrently
			 * dirtying pages, and we might have synced a lot of
			 * newly appeared dirty pages, but have not synced all
			 * of the old dirty pages.
			 */
			if (mpd->wbc->sync_mode == WB_SYNC_NONE && left <= 0)
				goto out;

J
Jan Kara 已提交
2652 2653 2654
			/* If we can't merge this page, we are done. */
			if (mpd->map.m_len > 0 && mpd->next_page != page->index)
				goto out;
2655

2656 2657
			lock_page(page);
			/*
J
Jan Kara 已提交
2658 2659 2660 2661 2662
			 * If the page is no longer dirty, or its mapping no
			 * longer corresponds to inode we are writing (which
			 * means it has been truncated or invalidated), or the
			 * page is already under writeback and we are not doing
			 * a data integrity writeback, skip the page
2663
			 */
2664 2665
			if (!PageDirty(page) ||
			    (PageWriteback(page) &&
J
Jan Kara 已提交
2666
			     (mpd->wbc->sync_mode == WB_SYNC_NONE)) ||
2667
			    unlikely(page->mapping != mapping)) {
2668 2669 2670 2671
				unlock_page(page);
				continue;
			}

2672
			wait_on_page_writeback(page);
2673 2674
			BUG_ON(PageWriteback(page));

J
Jan Kara 已提交
2675
			if (mpd->map.m_len == 0)
2676 2677
				mpd->first_page = page->index;
			mpd->next_page = page->index + 1;
2678
			/* Add all dirty buffers to mpd */
J
Jan Kara 已提交
2679
			lblk = ((ext4_lblk_t)page->index) <<
2680
				(PAGE_SHIFT - blkbits);
2681
			head = page_buffers(page);
2682 2683
			err = mpage_process_page_bufs(mpd, head, head, lblk);
			if (err <= 0)
J
Jan Kara 已提交
2684
				goto out;
2685
			err = 0;
2686
			left--;
2687 2688 2689 2690
		}
		pagevec_release(&pvec);
		cond_resched();
	}
2691
	return 0;
2692 2693
out:
	pagevec_release(&pvec);
J
Jan Kara 已提交
2694
	return err;
2695 2696
}

2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707
static int __writepage(struct page *page, struct writeback_control *wbc,
		       void *data)
{
	struct address_space *mapping = data;
	int ret = ext4_writepage(page, wbc);
	mapping_set_error(mapping, ret);
	return ret;
}

static int ext4_writepages(struct address_space *mapping,
			   struct writeback_control *wbc)
2708
{
J
Jan Kara 已提交
2709 2710
	pgoff_t	writeback_index = 0;
	long nr_to_write = wbc->nr_to_write;
2711
	int range_whole = 0;
J
Jan Kara 已提交
2712
	int cycled = 1;
2713
	handle_t *handle = NULL;
2714
	struct mpage_da_data mpd;
2715
	struct inode *inode = mapping->host;
2716
	int needed_blocks, rsv_blocks = 0, ret = 0;
2717
	struct ext4_sb_info *sbi = EXT4_SB(mapping->host->i_sb);
J
Jan Kara 已提交
2718
	bool done;
S
Shaohua Li 已提交
2719
	struct blk_plug plug;
2720
	bool give_up_on_write = false;
2721

2722 2723 2724
	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
		return -EIO;

2725
	percpu_down_read(&sbi->s_journal_flag_rwsem);
2726
	trace_ext4_writepages(inode, wbc);
2727

2728 2729 2730 2731 2732
	if (dax_mapping(mapping)) {
		ret = dax_writeback_mapping_range(mapping, inode->i_sb->s_bdev,
						  wbc);
		goto out_writepages;
	}
2733

2734 2735 2736 2737 2738
	/*
	 * No pages to write? This is mainly a kludge to avoid starting
	 * a transaction for special inodes like journal inode on last iput()
	 * because that could violate lock ordering on umount
	 */
2739
	if (!mapping->nrpages || !mapping_tagged(mapping, PAGECACHE_TAG_DIRTY))
2740
		goto out_writepages;
2741

2742 2743 2744 2745 2746 2747
	if (ext4_should_journal_data(inode)) {
		struct blk_plug plug;

		blk_start_plug(&plug);
		ret = write_cache_pages(mapping, wbc, __writepage, mapping);
		blk_finish_plug(&plug);
2748
		goto out_writepages;
2749 2750
	}

2751 2752 2753 2754
	/*
	 * If the filesystem has aborted, it is read-only, so return
	 * right away instead of dumping stack traces later on that
	 * will obscure the real source of the problem.  We test
2755
	 * EXT4_MF_FS_ABORTED instead of sb->s_flag's MS_RDONLY because
2756
	 * the latter could be true if the filesystem is mounted
2757
	 * read-only, and in that case, ext4_writepages should
2758 2759 2760
	 * *never* be called, so if that ever happens, we would want
	 * the stack trace.
	 */
2761 2762
	if (unlikely(ext4_forced_shutdown(EXT4_SB(mapping->host->i_sb)) ||
		     sbi->s_mount_flags & EXT4_MF_FS_ABORTED)) {
2763 2764 2765
		ret = -EROFS;
		goto out_writepages;
	}
2766

2767 2768
	if (ext4_should_dioread_nolock(inode)) {
		/*
2769
		 * We may need to convert up to one extent per block in
2770 2771
		 * the page and we may dirty the inode.
		 */
2772
		rsv_blocks = 1 + (PAGE_SIZE >> inode->i_blkbits);
2773 2774
	}

J
Jan Kara 已提交
2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792
	/*
	 * If we have inline data and arrive here, it means that
	 * we will soon create the block for the 1st page, so
	 * we'd better clear the inline data here.
	 */
	if (ext4_has_inline_data(inode)) {
		/* Just inode will be modified... */
		handle = ext4_journal_start(inode, EXT4_HT_INODE, 1);
		if (IS_ERR(handle)) {
			ret = PTR_ERR(handle);
			goto out_writepages;
		}
		BUG_ON(ext4_test_inode_state(inode,
				EXT4_STATE_MAY_INLINE_DATA));
		ext4_destroy_inline_data(handle, inode);
		ext4_journal_stop(handle);
	}

2793 2794
	if (wbc->range_start == 0 && wbc->range_end == LLONG_MAX)
		range_whole = 1;
2795

2796
	if (wbc->range_cyclic) {
J
Jan Kara 已提交
2797 2798
		writeback_index = mapping->writeback_index;
		if (writeback_index)
2799
			cycled = 0;
J
Jan Kara 已提交
2800 2801
		mpd.first_page = writeback_index;
		mpd.last_page = -1;
2802
	} else {
2803 2804
		mpd.first_page = wbc->range_start >> PAGE_SHIFT;
		mpd.last_page = wbc->range_end >> PAGE_SHIFT;
2805
	}
2806

J
Jan Kara 已提交
2807 2808 2809
	mpd.inode = inode;
	mpd.wbc = wbc;
	ext4_io_submit_init(&mpd.io_submit, wbc);
2810
retry:
2811
	if (wbc->sync_mode == WB_SYNC_ALL || wbc->tagged_writepages)
J
Jan Kara 已提交
2812 2813
		tag_pages_for_writeback(mapping, mpd.first_page, mpd.last_page);
	done = false;
S
Shaohua Li 已提交
2814
	blk_start_plug(&plug);
2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837

	/*
	 * First writeback pages that don't need mapping - we can avoid
	 * starting a transaction unnecessarily and also avoid being blocked
	 * in the block layer on device congestion while having transaction
	 * started.
	 */
	mpd.do_map = 0;
	mpd.io_submit.io_end = ext4_init_io_end(inode, GFP_KERNEL);
	if (!mpd.io_submit.io_end) {
		ret = -ENOMEM;
		goto unplug;
	}
	ret = mpage_prepare_extent_to_map(&mpd);
	/* Submit prepared bio */
	ext4_io_submit(&mpd.io_submit);
	ext4_put_io_end_defer(mpd.io_submit.io_end);
	mpd.io_submit.io_end = NULL;
	/* Unlock pages we didn't use */
	mpage_release_unused_pages(&mpd, false);
	if (ret < 0)
		goto unplug;

J
Jan Kara 已提交
2838 2839 2840 2841 2842 2843 2844
	while (!done && mpd.first_page <= mpd.last_page) {
		/* For each extent of pages we use new io_end */
		mpd.io_submit.io_end = ext4_init_io_end(inode, GFP_KERNEL);
		if (!mpd.io_submit.io_end) {
			ret = -ENOMEM;
			break;
		}
2845 2846

		/*
J
Jan Kara 已提交
2847 2848 2849 2850 2851
		 * We have two constraints: We find one extent to map and we
		 * must always write out whole page (makes a difference when
		 * blocksize < pagesize) so that we don't block on IO when we
		 * try to write out the rest of the page. Journalled mode is
		 * not supported by delalloc.
2852 2853
		 */
		BUG_ON(ext4_should_journal_data(inode));
2854
		needed_blocks = ext4_da_writepages_trans_blocks(inode);
2855

J
Jan Kara 已提交
2856
		/* start a new transaction */
2857 2858
		handle = ext4_journal_start_with_reserve(inode,
				EXT4_HT_WRITE_PAGE, needed_blocks, rsv_blocks);
2859 2860
		if (IS_ERR(handle)) {
			ret = PTR_ERR(handle);
2861
			ext4_msg(inode->i_sb, KERN_CRIT, "%s: jbd2_start: "
2862
			       "%ld pages, ino %lu; err %d", __func__,
2863
				wbc->nr_to_write, inode->i_ino, ret);
J
Jan Kara 已提交
2864 2865
			/* Release allocated io_end */
			ext4_put_io_end(mpd.io_submit.io_end);
2866
			mpd.io_submit.io_end = NULL;
J
Jan Kara 已提交
2867
			break;
2868
		}
2869
		mpd.do_map = 1;
2870

J
Jan Kara 已提交
2871 2872 2873 2874
		trace_ext4_da_write_pages(inode, mpd.first_page, mpd.wbc);
		ret = mpage_prepare_extent_to_map(&mpd);
		if (!ret) {
			if (mpd.map.m_len)
2875 2876
				ret = mpage_map_and_submit_extent(handle, &mpd,
					&give_up_on_write);
J
Jan Kara 已提交
2877 2878 2879 2880 2881 2882 2883 2884 2885
			else {
				/*
				 * We scanned the whole range (or exhausted
				 * nr_to_write), submitted what was mapped and
				 * didn't find anything needing mapping. We are
				 * done.
				 */
				done = true;
			}
2886
		}
2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899
		/*
		 * Caution: If the handle is synchronous,
		 * ext4_journal_stop() can wait for transaction commit
		 * to finish which may depend on writeback of pages to
		 * complete or on page lock to be released.  In that
		 * case, we have to wait until after after we have
		 * submitted all the IO, released page locks we hold,
		 * and dropped io_end reference (for extent conversion
		 * to be able to complete) before stopping the handle.
		 */
		if (!ext4_handle_valid(handle) || handle->h_sync == 0) {
			ext4_journal_stop(handle);
			handle = NULL;
2900
			mpd.do_map = 0;
2901
		}
J
Jan Kara 已提交
2902 2903 2904
		/* Submit prepared bio */
		ext4_io_submit(&mpd.io_submit);
		/* Unlock pages we didn't use */
2905
		mpage_release_unused_pages(&mpd, give_up_on_write);
2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917
		/*
		 * Drop our io_end reference we got from init. We have
		 * to be careful and use deferred io_end finishing if
		 * we are still holding the transaction as we can
		 * release the last reference to io_end which may end
		 * up doing unwritten extent conversion.
		 */
		if (handle) {
			ext4_put_io_end_defer(mpd.io_submit.io_end);
			ext4_journal_stop(handle);
		} else
			ext4_put_io_end(mpd.io_submit.io_end);
2918
		mpd.io_submit.io_end = NULL;
J
Jan Kara 已提交
2919 2920 2921 2922

		if (ret == -ENOSPC && sbi->s_journal) {
			/*
			 * Commit the transaction which would
2923 2924 2925
			 * free blocks released in the transaction
			 * and try again
			 */
2926
			jbd2_journal_force_commit_nested(sbi->s_journal);
2927
			ret = 0;
J
Jan Kara 已提交
2928 2929 2930 2931
			continue;
		}
		/* Fatal error - ENOMEM, EIO... */
		if (ret)
2932
			break;
2933
	}
2934
unplug:
S
Shaohua Li 已提交
2935
	blk_finish_plug(&plug);
2936
	if (!ret && !cycled && wbc->nr_to_write > 0) {
2937
		cycled = 1;
J
Jan Kara 已提交
2938 2939
		mpd.last_page = writeback_index - 1;
		mpd.first_page = 0;
2940 2941
		goto retry;
	}
2942 2943 2944 2945

	/* Update index */
	if (wbc->range_cyclic || (range_whole && wbc->nr_to_write > 0))
		/*
J
Jan Kara 已提交
2946
		 * Set the writeback_index so that range_cyclic
2947 2948
		 * mode will write it back later
		 */
J
Jan Kara 已提交
2949
		mapping->writeback_index = mpd.first_page;
2950

2951
out_writepages:
2952 2953
	trace_ext4_writepages_result(inode, wbc, ret,
				     nr_to_write - wbc->nr_to_write);
2954
	percpu_up_read(&sbi->s_journal_flag_rwsem);
2955
	return ret;
2956 2957
}

2958 2959
static int ext4_nonda_switch(struct super_block *sb)
{
2960
	s64 free_clusters, dirty_clusters;
2961 2962 2963 2964 2965
	struct ext4_sb_info *sbi = EXT4_SB(sb);

	/*
	 * switch to non delalloc mode if we are running low
	 * on free block. The free block accounting via percpu
2966
	 * counters can get slightly wrong with percpu_counter_batch getting
2967 2968 2969 2970
	 * accumulated on each CPU without updating global counters
	 * Delalloc need an accurate free block accounting. So switch
	 * to non delalloc when we are near to error range.
	 */
2971 2972 2973 2974
	free_clusters =
		percpu_counter_read_positive(&sbi->s_freeclusters_counter);
	dirty_clusters =
		percpu_counter_read_positive(&sbi->s_dirtyclusters_counter);
2975 2976 2977
	/*
	 * Start pushing delalloc when 1/2 of free blocks are dirty.
	 */
2978
	if (dirty_clusters && (free_clusters < 2 * dirty_clusters))
2979
		try_to_writeback_inodes_sb(sb, WB_REASON_FS_FREE_SPACE);
2980

2981 2982
	if (2 * free_clusters < 3 * dirty_clusters ||
	    free_clusters < (dirty_clusters + EXT4_FREECLUSTERS_WATERMARK)) {
2983
		/*
2984 2985
		 * free block count is less than 150% of dirty blocks
		 * or free blocks is less than watermark
2986 2987 2988 2989 2990 2991
		 */
		return 1;
	}
	return 0;
}

2992 2993 2994
/* We always reserve for an inode update; the superblock could be there too */
static int ext4_da_write_credits(struct inode *inode, loff_t pos, unsigned len)
{
2995
	if (likely(ext4_has_feature_large_file(inode->i_sb)))
2996 2997 2998 2999 3000 3001 3002 3003 3004
		return 1;

	if (pos + len <= 0x7fffffffULL)
		return 1;

	/* We might need to update the superblock to set LARGE_FILE */
	return 2;
}

3005
static int ext4_da_write_begin(struct file *file, struct address_space *mapping,
3006 3007
			       loff_t pos, unsigned len, unsigned flags,
			       struct page **pagep, void **fsdata)
3008
{
3009
	int ret, retries = 0;
3010 3011 3012 3013 3014
	struct page *page;
	pgoff_t index;
	struct inode *inode = mapping->host;
	handle_t *handle;

3015 3016 3017
	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
		return -EIO;

3018
	index = pos >> PAGE_SHIFT;
3019

3020 3021
	if (ext4_nonda_switch(inode->i_sb) ||
	    S_ISLNK(inode->i_mode)) {
3022 3023 3024 3025 3026
		*fsdata = (void *)FALL_BACK_TO_NONDELALLOC;
		return ext4_write_begin(file, mapping, pos,
					len, flags, pagep, fsdata);
	}
	*fsdata = (void *)0;
3027
	trace_ext4_da_write_begin(inode, pos, len, flags);
3028 3029 3030 3031 3032 3033

	if (ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA)) {
		ret = ext4_da_write_inline_data_begin(mapping, inode,
						      pos, len, flags,
						      pagep, fsdata);
		if (ret < 0)
3034 3035 3036
			return ret;
		if (ret == 1)
			return 0;
3037 3038
	}

3039 3040 3041 3042 3043 3044 3045 3046 3047 3048 3049 3050 3051
	/*
	 * grab_cache_page_write_begin() can take a long time if the
	 * system is thrashing due to memory pressure, or if the page
	 * is being written back.  So grab it first before we start
	 * the transaction handle.  This also allows us to allocate
	 * the page (if needed) without using GFP_NOFS.
	 */
retry_grab:
	page = grab_cache_page_write_begin(mapping, index, flags);
	if (!page)
		return -ENOMEM;
	unlock_page(page);

3052 3053 3054 3055 3056 3057
	/*
	 * With delayed allocation, we don't log the i_disksize update
	 * if there is delayed block allocation. But we still need
	 * to journalling the i_disksize update if writes to the end
	 * of file which has an already mapped buffer.
	 */
3058
retry_journal:
3059 3060
	handle = ext4_journal_start(inode, EXT4_HT_WRITE_PAGE,
				ext4_da_write_credits(inode, pos, len));
3061
	if (IS_ERR(handle)) {
3062
		put_page(page);
3063
		return PTR_ERR(handle);
3064 3065
	}

3066 3067 3068 3069
	lock_page(page);
	if (page->mapping != mapping) {
		/* The page got truncated from under us */
		unlock_page(page);
3070
		put_page(page);
3071
		ext4_journal_stop(handle);
3072
		goto retry_grab;
3073
	}
3074
	/* In case writeback began while the page was unlocked */
3075
	wait_for_stable_page(page);
3076

3077 3078 3079 3080
#ifdef CONFIG_EXT4_FS_ENCRYPTION
	ret = ext4_block_write_begin(page, pos, len,
				     ext4_da_get_block_prep);
#else
3081
	ret = __block_write_begin(page, pos, len, ext4_da_get_block_prep);
3082
#endif
3083 3084 3085
	if (ret < 0) {
		unlock_page(page);
		ext4_journal_stop(handle);
3086 3087 3088 3089 3090 3091
		/*
		 * block_write_begin may have instantiated a few blocks
		 * outside i_size.  Trim these off again. Don't need
		 * i_size_read because we hold i_mutex.
		 */
		if (pos + len > inode->i_size)
3092
			ext4_truncate_failed_write(inode);
3093 3094 3095 3096 3097

		if (ret == -ENOSPC &&
		    ext4_should_retry_alloc(inode->i_sb, &retries))
			goto retry_journal;

3098
		put_page(page);
3099
		return ret;
3100 3101
	}

3102
	*pagep = page;
3103 3104 3105
	return ret;
}

3106 3107 3108 3109 3110
/*
 * Check if we should update i_disksize
 * when write to the end of file but not require block allocation
 */
static int ext4_da_should_update_i_disksize(struct page *page,
3111
					    unsigned long offset)
3112 3113 3114 3115 3116 3117 3118 3119 3120
{
	struct buffer_head *bh;
	struct inode *inode = page->mapping->host;
	unsigned int idx;
	int i;

	bh = page_buffers(page);
	idx = offset >> inode->i_blkbits;

3121
	for (i = 0; i < idx; i++)
3122 3123
		bh = bh->b_this_page;

3124
	if (!buffer_mapped(bh) || (buffer_delay(bh)) || buffer_unwritten(bh))
3125 3126 3127 3128
		return 0;
	return 1;
}

3129
static int ext4_da_write_end(struct file *file,
3130 3131 3132
			     struct address_space *mapping,
			     loff_t pos, unsigned len, unsigned copied,
			     struct page *page, void *fsdata)
3133 3134 3135 3136 3137
{
	struct inode *inode = mapping->host;
	int ret = 0, ret2;
	handle_t *handle = ext4_journal_current_handle();
	loff_t new_i_size;
3138
	unsigned long start, end;
3139 3140
	int write_mode = (int)(unsigned long)fsdata;

3141 3142 3143
	if (write_mode == FALL_BACK_TO_NONDELALLOC)
		return ext4_write_end(file, mapping, pos,
				      len, copied, page, fsdata);
3144

3145
	trace_ext4_da_write_end(inode, pos, len, copied);
3146
	start = pos & (PAGE_SIZE - 1);
3147
	end = start + copied - 1;
3148 3149 3150 3151 3152 3153 3154

	/*
	 * generic_write_end() will run mark_inode_dirty() if i_size
	 * changes.  So let's piggyback the i_disksize mark_inode_dirty
	 * into that.
	 */
	new_i_size = pos + copied;
3155
	if (copied && new_i_size > EXT4_I(inode)->i_disksize) {
3156 3157
		if (ext4_has_inline_data(inode) ||
		    ext4_da_should_update_i_disksize(page, end)) {
3158
			ext4_update_i_disksize(inode, new_i_size);
3159 3160 3161 3162 3163
			/* We need to mark inode dirty even if
			 * new_i_size is less that inode->i_size
			 * bu greater than i_disksize.(hint delalloc)
			 */
			ext4_mark_inode_dirty(handle, inode);
3164
		}
3165
	}
3166 3167 3168 3169 3170 3171 3172 3173

	if (write_mode != CONVERT_INLINE_DATA &&
	    ext4_test_inode_state(inode, EXT4_STATE_MAY_INLINE_DATA) &&
	    ext4_has_inline_data(inode))
		ret2 = ext4_da_write_inline_data_end(inode, pos, len, copied,
						     page);
	else
		ret2 = generic_write_end(file, mapping, pos, len, copied,
3174
							page, fsdata);
3175

3176 3177 3178 3179 3180 3181 3182 3183 3184 3185
	copied = ret2;
	if (ret2 < 0)
		ret = ret2;
	ret2 = ext4_journal_stop(handle);
	if (!ret)
		ret = ret2;

	return ret ? ret : copied;
}

3186 3187
static void ext4_da_invalidatepage(struct page *page, unsigned int offset,
				   unsigned int length)
3188 3189 3190 3191 3192 3193 3194 3195
{
	/*
	 * Drop reserved blocks
	 */
	BUG_ON(!PageLocked(page));
	if (!page_has_buffers(page))
		goto out;

3196
	ext4_da_page_release_reservation(page, offset, length);
3197 3198

out:
3199
	ext4_invalidatepage(page, offset, length);
3200 3201 3202 3203

	return;
}

3204 3205 3206 3207 3208
/*
 * Force all delayed allocation blocks to be allocated for a given inode.
 */
int ext4_alloc_da_blocks(struct inode *inode)
{
3209 3210
	trace_ext4_alloc_da_blocks(inode);

3211
	if (!EXT4_I(inode)->i_reserved_data_blocks)
3212 3213 3214 3215 3216 3217 3218 3219
		return 0;

	/*
	 * We do something simple for now.  The filemap_flush() will
	 * also start triggering a write of the data blocks, which is
	 * not strictly speaking necessary (and for users of
	 * laptop_mode, not even desirable).  However, to do otherwise
	 * would require replicating code paths in:
3220
	 *
3221
	 * ext4_writepages() ->
3222 3223 3224 3225 3226 3227 3228 3229 3230 3231 3232
	 *    write_cache_pages() ---> (via passed in callback function)
	 *        __mpage_da_writepage() -->
	 *           mpage_add_bh_to_extent()
	 *           mpage_da_map_blocks()
	 *
	 * The problem is that write_cache_pages(), located in
	 * mm/page-writeback.c, marks pages clean in preparation for
	 * doing I/O, which is not desirable if we're not planning on
	 * doing I/O at all.
	 *
	 * We could call write_cache_pages(), and then redirty all of
3233
	 * the pages by calling redirty_page_for_writepage() but that
3234 3235
	 * would be ugly in the extreme.  So instead we would need to
	 * replicate parts of the code in the above functions,
L
Lucas De Marchi 已提交
3236
	 * simplifying them because we wouldn't actually intend to
3237 3238 3239
	 * write out the pages, but rather only collect contiguous
	 * logical block extents, call the multi-block allocator, and
	 * then update the buffer heads with the block allocations.
3240
	 *
3241 3242 3243 3244 3245 3246
	 * For now, though, we'll cheat by calling filemap_flush(),
	 * which will map the blocks, and start the I/O, but not
	 * actually wait for the I/O to complete.
	 */
	return filemap_flush(inode->i_mapping);
}
3247

3248 3249 3250 3251 3252
/*
 * bmap() is special.  It gets used by applications such as lilo and by
 * the swapper to find the on-disk block of a specific piece of data.
 *
 * Naturally, this is dangerous if the block concerned is still in the
3253
 * journal.  If somebody makes a swapfile on an ext4 data-journaling
3254 3255 3256 3257 3258 3259 3260 3261
 * filesystem and enables swap, then they may get a nasty shock when the
 * data getting swapped to that swapfile suddenly gets overwritten by
 * the original zero's written out previously to the journal and
 * awaiting writeback in the kernel's buffer cache.
 *
 * So, if we see any bmap calls here on a modified, data-journaled file,
 * take extra steps to flush any blocks which might be in the cache.
 */
3262
static sector_t ext4_bmap(struct address_space *mapping, sector_t block)
3263 3264 3265 3266 3267
{
	struct inode *inode = mapping->host;
	journal_t *journal;
	int err;

T
Tao Ma 已提交
3268 3269 3270 3271 3272 3273
	/*
	 * We can get here for an inline file via the FIBMAP ioctl
	 */
	if (ext4_has_inline_data(inode))
		return 0;

3274 3275 3276 3277 3278 3279 3280 3281 3282 3283
	if (mapping_tagged(mapping, PAGECACHE_TAG_DIRTY) &&
			test_opt(inode->i_sb, DELALLOC)) {
		/*
		 * With delalloc we want to sync the file
		 * so that we can make sure we allocate
		 * blocks for file
		 */
		filemap_write_and_wait(mapping);
	}

3284 3285
	if (EXT4_JOURNAL(inode) &&
	    ext4_test_inode_state(inode, EXT4_STATE_JDATA)) {
3286 3287 3288 3289 3290 3291 3292 3293 3294 3295 3296
		/*
		 * This is a REALLY heavyweight approach, but the use of
		 * bmap on dirty files is expected to be extremely rare:
		 * only if we run lilo or swapon on a freshly made file
		 * do we expect this to happen.
		 *
		 * (bmap requires CAP_SYS_RAWIO so this does not
		 * represent an unprivileged user DOS attack --- we'd be
		 * in trouble if mortal users could trigger this path at
		 * will.)
		 *
3297
		 * NB. EXT4_STATE_JDATA is not set on files other than
3298 3299 3300 3301 3302 3303
		 * regular files.  If somebody wants to bmap a directory
		 * or symlink and gets confused because the buffer
		 * hasn't yet been flushed to disk, they deserve
		 * everything they get.
		 */

3304
		ext4_clear_inode_state(inode, EXT4_STATE_JDATA);
3305
		journal = EXT4_JOURNAL(inode);
3306 3307 3308
		jbd2_journal_lock_updates(journal);
		err = jbd2_journal_flush(journal);
		jbd2_journal_unlock_updates(journal);
3309 3310 3311 3312 3313

		if (err)
			return 0;
	}

3314
	return generic_block_bmap(mapping, block, ext4_get_block);
3315 3316
}

3317
static int ext4_readpage(struct file *file, struct page *page)
3318
{
T
Tao Ma 已提交
3319 3320 3321
	int ret = -EAGAIN;
	struct inode *inode = page->mapping->host;

3322
	trace_ext4_readpage(page);
T
Tao Ma 已提交
3323 3324 3325 3326 3327

	if (ext4_has_inline_data(inode))
		ret = ext4_readpage_inline(inode, page);

	if (ret == -EAGAIN)
3328
		return ext4_mpage_readpages(page->mapping, NULL, page, 1);
T
Tao Ma 已提交
3329 3330

	return ret;
3331 3332 3333
}

static int
3334
ext4_readpages(struct file *file, struct address_space *mapping,
3335 3336
		struct list_head *pages, unsigned nr_pages)
{
T
Tao Ma 已提交
3337 3338 3339 3340 3341 3342
	struct inode *inode = mapping->host;

	/* If the file has inline data, no need to do readpages. */
	if (ext4_has_inline_data(inode))
		return 0;

3343
	return ext4_mpage_readpages(mapping, pages, NULL, nr_pages);
3344 3345
}

3346 3347
static void ext4_invalidatepage(struct page *page, unsigned int offset,
				unsigned int length)
3348
{
3349
	trace_ext4_invalidatepage(page, offset, length);
3350

3351 3352 3353
	/* No journalling happens on data buffers when this function is used */
	WARN_ON(page_has_buffers(page) && buffer_jbd(page_buffers(page)));

3354
	block_invalidatepage(page, offset, length);
3355 3356
}

3357
static int __ext4_journalled_invalidatepage(struct page *page,
3358 3359
					    unsigned int offset,
					    unsigned int length)
3360 3361 3362
{
	journal_t *journal = EXT4_JOURNAL(page->mapping->host);

3363
	trace_ext4_journalled_invalidatepage(page, offset, length);
3364

3365 3366 3367
	/*
	 * If it's a full truncate we just forget about the pending dirtying
	 */
3368
	if (offset == 0 && length == PAGE_SIZE)
3369 3370
		ClearPageChecked(page);

3371
	return jbd2_journal_invalidatepage(journal, page, offset, length);
3372 3373 3374 3375
}

/* Wrapper for aops... */
static void ext4_journalled_invalidatepage(struct page *page,
3376 3377
					   unsigned int offset,
					   unsigned int length)
3378
{
3379
	WARN_ON(__ext4_journalled_invalidatepage(page, offset, length) < 0);
3380 3381
}

3382
static int ext4_releasepage(struct page *page, gfp_t wait)
3383
{
3384
	journal_t *journal = EXT4_JOURNAL(page->mapping->host);
3385

3386 3387
	trace_ext4_releasepage(page);

3388 3389
	/* Page has dirty journalled data -> cannot release */
	if (PageChecked(page))
3390
		return 0;
3391 3392 3393 3394
	if (journal)
		return jbd2_journal_try_to_free_buffers(journal, page, wait);
	else
		return try_to_free_buffers(page);
3395 3396
}

3397 3398 3399
static int ext4_iomap_begin(struct inode *inode, loff_t offset, loff_t length,
			    unsigned flags, struct iomap *iomap)
{
3400
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
3401 3402 3403 3404
	unsigned int blkbits = inode->i_blkbits;
	unsigned long first_block = offset >> blkbits;
	unsigned long last_block = (offset + length - 1) >> blkbits;
	struct ext4_map_blocks map;
3405
	bool delalloc = false;
3406 3407
	int ret;

3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421

	if (flags & IOMAP_REPORT) {
		if (ext4_has_inline_data(inode)) {
			ret = ext4_inline_data_iomap(inode, iomap);
			if (ret != -EAGAIN) {
				if (ret == 0 && offset >= iomap->length)
					ret = -ENOENT;
				return ret;
			}
		}
	} else {
		if (WARN_ON_ONCE(ext4_has_inline_data(inode)))
			return -ERANGE;
	}
3422 3423 3424 3425

	map.m_lblk = first_block;
	map.m_len = last_block - first_block + 1;

3426
	if (flags & IOMAP_REPORT) {
J
Jan Kara 已提交
3427
		ret = ext4_map_blocks(NULL, inode, &map, 0);
3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452
		if (ret < 0)
			return ret;

		if (ret == 0) {
			ext4_lblk_t end = map.m_lblk + map.m_len - 1;
			struct extent_status es;

			ext4_es_find_delayed_extent_range(inode, map.m_lblk, end, &es);

			if (!es.es_len || es.es_lblk > end) {
				/* entire range is a hole */
			} else if (es.es_lblk > map.m_lblk) {
				/* range starts with a hole */
				map.m_len = es.es_lblk - map.m_lblk;
			} else {
				ext4_lblk_t offs = 0;

				if (es.es_lblk < map.m_lblk)
					offs = map.m_lblk - es.es_lblk;
				map.m_lblk = es.es_lblk + offs;
				map.m_len = es.es_len - offs;
				delalloc = true;
			}
		}
	} else if (flags & IOMAP_WRITE) {
J
Jan Kara 已提交
3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472 3473 3474 3475 3476 3477 3478 3479 3480 3481 3482 3483
		int dio_credits;
		handle_t *handle;
		int retries = 0;

		/* Trim mapping request to maximum we can map at once for DIO */
		if (map.m_len > DIO_MAX_BLOCKS)
			map.m_len = DIO_MAX_BLOCKS;
		dio_credits = ext4_chunk_trans_blocks(inode, map.m_len);
retry:
		/*
		 * Either we allocate blocks and then we don't get unwritten
		 * extent so we have reserved enough credits, or the blocks
		 * are already allocated and unwritten and in that case
		 * extent conversion fits in the credits as well.
		 */
		handle = ext4_journal_start(inode, EXT4_HT_MAP_BLOCKS,
					    dio_credits);
		if (IS_ERR(handle))
			return PTR_ERR(handle);

		ret = ext4_map_blocks(handle, inode, &map,
				      EXT4_GET_BLOCKS_CREATE_ZERO);
		if (ret < 0) {
			ext4_journal_stop(handle);
			if (ret == -ENOSPC &&
			    ext4_should_retry_alloc(inode->i_sb, &retries))
				goto retry;
			return ret;
		}

		/*
3484
		 * If we added blocks beyond i_size, we need to make sure they
J
Jan Kara 已提交
3485
		 * will get truncated if we crash before updating i_size in
3486 3487 3488 3489 3490
		 * ext4_iomap_end(). For faults we don't need to do that (and
		 * even cannot because for orphan list operations inode_lock is
		 * required) - if we happen to instantiate block beyond i_size,
		 * it is because we race with truncate which has already added
		 * the inode to the orphan list.
J
Jan Kara 已提交
3491
		 */
3492 3493
		if (!(flags & IOMAP_FAULT) && first_block + map.m_len >
		    (i_size_read(inode) + (1 << blkbits) - 1) >> blkbits) {
J
Jan Kara 已提交
3494 3495 3496 3497 3498 3499 3500 3501 3502
			int err;

			err = ext4_orphan_add(handle, inode);
			if (err < 0) {
				ext4_journal_stop(handle);
				return err;
			}
		}
		ext4_journal_stop(handle);
3503 3504 3505 3506
	} else {
		ret = ext4_map_blocks(NULL, inode, &map, 0);
		if (ret < 0)
			return ret;
J
Jan Kara 已提交
3507
	}
3508 3509

	iomap->flags = 0;
3510 3511
	iomap->bdev = inode->i_sb->s_bdev;
	iomap->dax_dev = sbi->s_daxdev;
3512
	iomap->offset = first_block << blkbits;
3513
	iomap->length = (u64)map.m_len << blkbits;
3514 3515

	if (ret == 0) {
3516
		iomap->type = delalloc ? IOMAP_DELALLOC : IOMAP_HOLE;
3517
		iomap->addr = IOMAP_NULL_ADDR;
3518 3519 3520 3521 3522 3523 3524 3525 3526
	} else {
		if (map.m_flags & EXT4_MAP_MAPPED) {
			iomap->type = IOMAP_MAPPED;
		} else if (map.m_flags & EXT4_MAP_UNWRITTEN) {
			iomap->type = IOMAP_UNWRITTEN;
		} else {
			WARN_ON_ONCE(1);
			return -EIO;
		}
3527
		iomap->addr = (u64)map.m_pblk << blkbits;
3528 3529 3530 3531
	}

	if (map.m_flags & EXT4_MAP_NEW)
		iomap->flags |= IOMAP_F_NEW;
3532

3533 3534 3535
	return 0;
}

J
Jan Kara 已提交
3536 3537 3538 3539 3540 3541 3542 3543
static int ext4_iomap_end(struct inode *inode, loff_t offset, loff_t length,
			  ssize_t written, unsigned flags, struct iomap *iomap)
{
	int ret = 0;
	handle_t *handle;
	int blkbits = inode->i_blkbits;
	bool truncate = false;

3544
	if (!(flags & IOMAP_WRITE) || (flags & IOMAP_FAULT))
J
Jan Kara 已提交
3545 3546 3547 3548 3549 3550 3551 3552 3553 3554 3555 3556 3557 3558 3559 3560 3561 3562 3563 3564 3565 3566 3567 3568 3569 3570 3571 3572 3573 3574 3575 3576 3577 3578 3579 3580 3581 3582 3583 3584 3585 3586 3587
		return 0;

	handle = ext4_journal_start(inode, EXT4_HT_INODE, 2);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		goto orphan_del;
	}
	if (ext4_update_inode_size(inode, offset + written))
		ext4_mark_inode_dirty(handle, inode);
	/*
	 * We may need to truncate allocated but not written blocks beyond EOF.
	 */
	if (iomap->offset + iomap->length > 
	    ALIGN(inode->i_size, 1 << blkbits)) {
		ext4_lblk_t written_blk, end_blk;

		written_blk = (offset + written) >> blkbits;
		end_blk = (offset + length) >> blkbits;
		if (written_blk < end_blk && ext4_can_truncate(inode))
			truncate = true;
	}
	/*
	 * Remove inode from orphan list if we were extending a inode and
	 * everything went fine.
	 */
	if (!truncate && inode->i_nlink &&
	    !list_empty(&EXT4_I(inode)->i_orphan))
		ext4_orphan_del(handle, inode);
	ext4_journal_stop(handle);
	if (truncate) {
		ext4_truncate_failed_write(inode);
orphan_del:
		/*
		 * If truncate failed early the inode might still be on the
		 * orphan list; we need to make sure the inode is removed from
		 * the orphan list in that case.
		 */
		if (inode->i_nlink)
			ext4_orphan_del(NULL, inode);
	}
	return ret;
}

3588
const struct iomap_ops ext4_iomap_ops = {
3589
	.iomap_begin		= ext4_iomap_begin,
J
Jan Kara 已提交
3590
	.iomap_end		= ext4_iomap_end,
3591 3592
};

3593
static int ext4_end_io_dio(struct kiocb *iocb, loff_t offset,
3594
			    ssize_t size, void *private)
3595
{
3596
        ext4_io_end_t *io_end = private;
3597

J
Jan Kara 已提交
3598
	/* if not async direct IO just return */
3599
	if (!io_end)
3600
		return 0;
3601

3602
	ext_debug("ext4_end_io_dio(): io_end 0x%p "
3603
		  "for inode %lu, iocb 0x%p, offset %llu, size %zd\n",
3604
		  io_end, io_end->inode->i_ino, iocb, offset, size);
3605

3606 3607 3608 3609 3610 3611 3612 3613
	/*
	 * Error during AIO DIO. We cannot convert unwritten extents as the
	 * data was not written. Just clear the unwritten flag and drop io_end.
	 */
	if (size <= 0) {
		ext4_clear_io_unwritten_flag(io_end);
		size = 0;
	}
3614 3615
	io_end->offset = offset;
	io_end->size = size;
3616
	ext4_put_io_end(io_end);
3617 3618

	return 0;
3619
}
3620

3621
/*
J
Jan Kara 已提交
3622 3623 3624
 * Handling of direct IO writes.
 *
 * For ext4 extent files, ext4 will do direct-io write even to holes,
3625 3626 3627
 * preallocated extents, and those write extend the file, no need to
 * fall back to buffered IO.
 *
3628
 * For holes, we fallocate those blocks, mark them as unwritten
3629
 * If those blocks were preallocated, we mark sure they are split, but
3630
 * still keep the range to write as unwritten.
3631
 *
3632
 * The unwritten extents will be converted to written when DIO is completed.
3633
 * For async direct IO, since the IO may still pending when return, we
L
Lucas De Marchi 已提交
3634
 * set up an end_io call back function, which will do the conversion
3635
 * when async direct IO completed.
3636 3637 3638 3639 3640 3641
 *
 * If the O_DIRECT write will extend the file then add this inode to the
 * orphan list.  So recovery will truncate it back to the original size
 * if the machine crashes during the write.
 *
 */
3642
static ssize_t ext4_direct_IO_write(struct kiocb *iocb, struct iov_iter *iter)
3643 3644 3645
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;
J
Jan Kara 已提交
3646
	struct ext4_inode_info *ei = EXT4_I(inode);
3647
	ssize_t ret;
3648
	loff_t offset = iocb->ki_pos;
3649
	size_t count = iov_iter_count(iter);
3650 3651 3652
	int overwrite = 0;
	get_block_t *get_block_func = NULL;
	int dio_flags = 0;
3653
	loff_t final_size = offset + count;
J
Jan Kara 已提交
3654 3655
	int orphan = 0;
	handle_t *handle;
3656

J
Jan Kara 已提交
3657 3658 3659 3660 3661 3662 3663 3664 3665 3666 3667 3668 3669 3670 3671 3672
	if (final_size > inode->i_size) {
		/* Credits for sb + inode write */
		handle = ext4_journal_start(inode, EXT4_HT_INODE, 2);
		if (IS_ERR(handle)) {
			ret = PTR_ERR(handle);
			goto out;
		}
		ret = ext4_orphan_add(handle, inode);
		if (ret) {
			ext4_journal_stop(handle);
			goto out;
		}
		orphan = 1;
		ei->i_disksize = inode->i_size;
		ext4_journal_stop(handle);
	}
3673

3674
	BUG_ON(iocb->private == NULL);
3675

3676 3677 3678 3679 3680
	/*
	 * Make all waiters for direct IO properly wait also for extent
	 * conversion. This also disallows race between truncate() and
	 * overwrite DIO as i_dio_count needs to be incremented under i_mutex.
	 */
J
Jan Kara 已提交
3681
	inode_dio_begin(inode);
3682

3683 3684
	/* If we do a overwrite dio, i_mutex locking can be released */
	overwrite = *((int *)iocb->private);
3685

3686
	if (overwrite)
A
Al Viro 已提交
3687
		inode_unlock(inode);
3688

3689
	/*
J
Jan Kara 已提交
3690
	 * For extent mapped files we could direct write to holes and fallocate.
3691
	 *
3692 3693 3694
	 * Allocated blocks to fill the hole are marked as unwritten to prevent
	 * parallel buffered read to expose the stale data before DIO complete
	 * the data IO.
3695
	 *
3696 3697
	 * As to previously fallocated extents, ext4 get_block will just simply
	 * mark the buffer mapped but still keep the extents unwritten.
3698
	 *
3699 3700 3701 3702
	 * For non AIO case, we will convert those unwritten extents to written
	 * after return back from blockdev_direct_IO. That way we save us from
	 * allocating io_end structure and also the overhead of offloading
	 * the extent convertion to a workqueue.
3703 3704 3705 3706 3707 3708 3709
	 *
	 * For async DIO, the conversion needs to be deferred when the
	 * IO is completed. The ext4 end_io callback function will be
	 * called to take care of the conversion work.  Here for async
	 * case, we allocate an io_end structure to hook to the iocb.
	 */
	iocb->private = NULL;
3710
	if (overwrite)
3711
		get_block_func = ext4_dio_get_block_overwrite;
3712
	else if (!ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS) ||
F
Fabian Frederick 已提交
3713
		   round_down(offset, i_blocksize(inode)) >= inode->i_size) {
J
Jan Kara 已提交
3714 3715 3716
		get_block_func = ext4_dio_get_block;
		dio_flags = DIO_LOCKING | DIO_SKIP_HOLES;
	} else if (is_sync_kiocb(iocb)) {
3717 3718
		get_block_func = ext4_dio_get_block_unwritten_sync;
		dio_flags = DIO_LOCKING;
3719
	} else {
3720
		get_block_func = ext4_dio_get_block_unwritten_async;
3721 3722
		dio_flags = DIO_LOCKING;
	}
3723 3724 3725
	ret = __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev, iter,
				   get_block_func, ext4_end_io_dio, NULL,
				   dio_flags);
3726

J
Jan Kara 已提交
3727
	if (ret > 0 && !overwrite && ext4_test_inode_state(inode,
3728 3729 3730 3731 3732 3733
						EXT4_STATE_DIO_UNWRITTEN)) {
		int err;
		/*
		 * for non AIO case, since the IO is already
		 * completed, we could do the conversion right here
		 */
3734
		err = ext4_convert_unwritten_extents(NULL, inode,
3735 3736 3737 3738 3739
						     offset, ret);
		if (err < 0)
			ret = err;
		ext4_clear_inode_state(inode, EXT4_STATE_DIO_UNWRITTEN);
	}
3740

J
Jan Kara 已提交
3741
	inode_dio_end(inode);
3742
	/* take i_mutex locking again if we do a ovewrite dio */
3743
	if (overwrite)
A
Al Viro 已提交
3744
		inode_lock(inode);
3745

J
Jan Kara 已提交
3746 3747 3748 3749 3750 3751 3752 3753 3754 3755 3756 3757 3758 3759 3760 3761 3762 3763 3764 3765 3766 3767 3768 3769 3770 3771 3772 3773 3774 3775 3776 3777 3778 3779 3780 3781 3782 3783 3784 3785 3786 3787 3788 3789
	if (ret < 0 && final_size > inode->i_size)
		ext4_truncate_failed_write(inode);

	/* Handle extending of i_size after direct IO write */
	if (orphan) {
		int err;

		/* Credits for sb + inode write */
		handle = ext4_journal_start(inode, EXT4_HT_INODE, 2);
		if (IS_ERR(handle)) {
			/* This is really bad luck. We've written the data
			 * but cannot extend i_size. Bail out and pretend
			 * the write failed... */
			ret = PTR_ERR(handle);
			if (inode->i_nlink)
				ext4_orphan_del(NULL, inode);

			goto out;
		}
		if (inode->i_nlink)
			ext4_orphan_del(handle, inode);
		if (ret > 0) {
			loff_t end = offset + ret;
			if (end > inode->i_size) {
				ei->i_disksize = end;
				i_size_write(inode, end);
				/*
				 * We're going to return a positive `ret'
				 * here due to non-zero-length I/O, so there's
				 * no way of reporting error returns from
				 * ext4_mark_inode_dirty() to userspace.  So
				 * ignore it.
				 */
				ext4_mark_inode_dirty(handle, inode);
			}
		}
		err = ext4_journal_stop(handle);
		if (ret == 0)
			ret = err;
	}
out:
	return ret;
}

3790
static ssize_t ext4_direct_IO_read(struct kiocb *iocb, struct iov_iter *iter)
J
Jan Kara 已提交
3791
{
J
Jan Kara 已提交
3792 3793
	struct address_space *mapping = iocb->ki_filp->f_mapping;
	struct inode *inode = mapping->host;
3794
	size_t count = iov_iter_count(iter);
J
Jan Kara 已提交
3795 3796
	ssize_t ret;

J
Jan Kara 已提交
3797 3798 3799 3800 3801 3802
	/*
	 * Shared inode_lock is enough for us - it protects against concurrent
	 * writes & truncates and since we take care of writing back page cache,
	 * we are protected against page writeback as well.
	 */
	inode_lock_shared(inode);
3803
	ret = filemap_write_and_wait_range(mapping, iocb->ki_pos,
3804
					   iocb->ki_pos + count - 1);
3805 3806 3807 3808
	if (ret)
		goto out_unlock;
	ret = __blockdev_direct_IO(iocb, inode, inode->i_sb->s_bdev,
				   iter, ext4_dio_get_block, NULL, NULL, 0);
J
Jan Kara 已提交
3809 3810
out_unlock:
	inode_unlock_shared(inode);
3811
	return ret;
3812 3813
}

3814
static ssize_t ext4_direct_IO(struct kiocb *iocb, struct iov_iter *iter)
3815 3816 3817
{
	struct file *file = iocb->ki_filp;
	struct inode *inode = file->f_mapping->host;
3818
	size_t count = iov_iter_count(iter);
3819
	loff_t offset = iocb->ki_pos;
3820
	ssize_t ret;
3821

3822 3823 3824 3825 3826
#ifdef CONFIG_EXT4_FS_ENCRYPTION
	if (ext4_encrypted_inode(inode) && S_ISREG(inode->i_mode))
		return 0;
#endif

3827 3828 3829 3830 3831 3832
	/*
	 * If we are doing data journalling we don't support O_DIRECT
	 */
	if (ext4_should_journal_data(inode))
		return 0;

T
Tao Ma 已提交
3833 3834 3835 3836
	/* Let buffer I/O handle the inline data case. */
	if (ext4_has_inline_data(inode))
		return 0;

3837 3838 3839 3840
	/* DAX uses iomap path now */
	if (WARN_ON_ONCE(IS_DAX(inode)))
		return 0;

3841
	trace_ext4_direct_IO_enter(inode, offset, count, iov_iter_rw(iter));
J
Jan Kara 已提交
3842
	if (iov_iter_rw(iter) == READ)
3843
		ret = ext4_direct_IO_read(iocb, iter);
3844
	else
3845
		ret = ext4_direct_IO_write(iocb, iter);
3846
	trace_ext4_direct_IO_exit(inode, offset, count, iov_iter_rw(iter), ret);
3847
	return ret;
3848 3849
}

3850
/*
3851
 * Pages can be marked dirty completely asynchronously from ext4's journalling
3852 3853 3854 3855 3856 3857 3858 3859 3860 3861 3862
 * activity.  By filemap_sync_pte(), try_to_unmap_one(), etc.  We cannot do
 * much here because ->set_page_dirty is called under VFS locks.  The page is
 * not necessarily locked.
 *
 * We cannot just dirty the page and leave attached buffers clean, because the
 * buffers' dirty state is "definitive".  We cannot just set the buffers dirty
 * or jbddirty because all the journalling code will explode.
 *
 * So what we do is to mark the page "pending dirty" and next time writepage
 * is called, propagate that into the buffers appropriately.
 */
3863
static int ext4_journalled_set_page_dirty(struct page *page)
3864 3865 3866 3867 3868
{
	SetPageChecked(page);
	return __set_page_dirty_nobuffers(page);
}

3869 3870 3871 3872 3873 3874 3875
static int ext4_set_page_dirty(struct page *page)
{
	WARN_ON_ONCE(!PageLocked(page) && !PageDirty(page));
	WARN_ON_ONCE(!page_has_buffers(page));
	return __set_page_dirty_buffers(page);
}

3876
static const struct address_space_operations ext4_aops = {
3877 3878
	.readpage		= ext4_readpage,
	.readpages		= ext4_readpages,
3879
	.writepage		= ext4_writepage,
3880
	.writepages		= ext4_writepages,
3881
	.write_begin		= ext4_write_begin,
3882
	.write_end		= ext4_write_end,
3883
	.set_page_dirty		= ext4_set_page_dirty,
3884 3885 3886 3887 3888 3889
	.bmap			= ext4_bmap,
	.invalidatepage		= ext4_invalidatepage,
	.releasepage		= ext4_releasepage,
	.direct_IO		= ext4_direct_IO,
	.migratepage		= buffer_migrate_page,
	.is_partially_uptodate  = block_is_partially_uptodate,
3890
	.error_remove_page	= generic_error_remove_page,
3891 3892
};

3893
static const struct address_space_operations ext4_journalled_aops = {
3894 3895
	.readpage		= ext4_readpage,
	.readpages		= ext4_readpages,
3896
	.writepage		= ext4_writepage,
3897
	.writepages		= ext4_writepages,
3898 3899 3900 3901
	.write_begin		= ext4_write_begin,
	.write_end		= ext4_journalled_write_end,
	.set_page_dirty		= ext4_journalled_set_page_dirty,
	.bmap			= ext4_bmap,
3902
	.invalidatepage		= ext4_journalled_invalidatepage,
3903
	.releasepage		= ext4_releasepage,
3904
	.direct_IO		= ext4_direct_IO,
3905
	.is_partially_uptodate  = block_is_partially_uptodate,
3906
	.error_remove_page	= generic_error_remove_page,
3907 3908
};

3909
static const struct address_space_operations ext4_da_aops = {
3910 3911
	.readpage		= ext4_readpage,
	.readpages		= ext4_readpages,
3912
	.writepage		= ext4_writepage,
3913
	.writepages		= ext4_writepages,
3914 3915
	.write_begin		= ext4_da_write_begin,
	.write_end		= ext4_da_write_end,
3916
	.set_page_dirty		= ext4_set_page_dirty,
3917 3918 3919 3920 3921 3922
	.bmap			= ext4_bmap,
	.invalidatepage		= ext4_da_invalidatepage,
	.releasepage		= ext4_releasepage,
	.direct_IO		= ext4_direct_IO,
	.migratepage		= buffer_migrate_page,
	.is_partially_uptodate  = block_is_partially_uptodate,
3923
	.error_remove_page	= generic_error_remove_page,
3924 3925
};

3926
void ext4_set_aops(struct inode *inode)
3927
{
3928 3929 3930 3931 3932
	switch (ext4_inode_journal_mode(inode)) {
	case EXT4_INODE_ORDERED_DATA_MODE:
	case EXT4_INODE_WRITEBACK_DATA_MODE:
		break;
	case EXT4_INODE_JOURNAL_DATA_MODE:
3933
		inode->i_mapping->a_ops = &ext4_journalled_aops;
3934
		return;
3935 3936 3937
	default:
		BUG();
	}
3938 3939 3940 3941
	if (test_opt(inode->i_sb, DELALLOC))
		inode->i_mapping->a_ops = &ext4_da_aops;
	else
		inode->i_mapping->a_ops = &ext4_aops;
3942 3943
}

R
Ross Zwisler 已提交
3944
static int __ext4_block_zero_page_range(handle_t *handle,
3945 3946
		struct address_space *mapping, loff_t from, loff_t length)
{
3947 3948
	ext4_fsblk_t index = from >> PAGE_SHIFT;
	unsigned offset = from & (PAGE_SIZE-1);
R
Ross Zwisler 已提交
3949
	unsigned blocksize, pos;
3950 3951 3952 3953 3954 3955
	ext4_lblk_t iblock;
	struct inode *inode = mapping->host;
	struct buffer_head *bh;
	struct page *page;
	int err = 0;

3956
	page = find_or_create_page(mapping, from >> PAGE_SHIFT,
3957
				   mapping_gfp_constraint(mapping, ~__GFP_FS));
3958 3959 3960 3961 3962
	if (!page)
		return -ENOMEM;

	blocksize = inode->i_sb->s_blocksize;

3963
	iblock = index << (PAGE_SHIFT - inode->i_sb->s_blocksize_bits);
3964 3965 3966 3967 3968 3969 3970 3971 3972 3973 3974 3975 3976 3977 3978 3979 3980 3981 3982 3983 3984 3985 3986 3987 3988 3989 3990 3991 3992 3993 3994 3995

	if (!page_has_buffers(page))
		create_empty_buffers(page, blocksize, 0);

	/* Find the buffer that contains "offset" */
	bh = page_buffers(page);
	pos = blocksize;
	while (offset >= pos) {
		bh = bh->b_this_page;
		iblock++;
		pos += blocksize;
	}
	if (buffer_freed(bh)) {
		BUFFER_TRACE(bh, "freed: skip");
		goto unlock;
	}
	if (!buffer_mapped(bh)) {
		BUFFER_TRACE(bh, "unmapped");
		ext4_get_block(inode, iblock, bh, 0);
		/* unmapped? It's a hole - nothing to do */
		if (!buffer_mapped(bh)) {
			BUFFER_TRACE(bh, "still unmapped");
			goto unlock;
		}
	}

	/* Ok, it's mapped. Make sure it's up-to-date */
	if (PageUptodate(page))
		set_buffer_uptodate(bh);

	if (!buffer_uptodate(bh)) {
		err = -EIO;
3996
		ll_rw_block(REQ_OP_READ, 0, 1, &bh);
3997 3998 3999 4000
		wait_on_buffer(bh);
		/* Uhhuh. Read error. Complain and punt. */
		if (!buffer_uptodate(bh))
			goto unlock;
4001 4002 4003
		if (S_ISREG(inode->i_mode) &&
		    ext4_encrypted_inode(inode)) {
			/* We expect the key to be set. */
4004
			BUG_ON(!fscrypt_has_encryption_key(inode));
4005
			BUG_ON(blocksize != PAGE_SIZE);
4006
			WARN_ON_ONCE(fscrypt_decrypt_page(page->mapping->host,
4007
						page, PAGE_SIZE, 0, page->index));
4008
		}
4009 4010 4011 4012 4013 4014 4015 4016 4017 4018 4019 4020
	}
	if (ext4_should_journal_data(inode)) {
		BUFFER_TRACE(bh, "get write access");
		err = ext4_journal_get_write_access(handle, bh);
		if (err)
			goto unlock;
	}
	zero_user(page, offset, length);
	BUFFER_TRACE(bh, "zeroed end of block");

	if (ext4_should_journal_data(inode)) {
		err = ext4_handle_dirty_metadata(handle, inode, bh);
4021
	} else {
4022
		err = 0;
4023
		mark_buffer_dirty(bh);
J
Jan Kara 已提交
4024
		if (ext4_should_order_data(inode))
4025
			err = ext4_jbd2_inode_add_write(handle, inode);
4026
	}
4027 4028 4029

unlock:
	unlock_page(page);
4030
	put_page(page);
4031 4032 4033
	return err;
}

R
Ross Zwisler 已提交
4034 4035 4036 4037 4038 4039 4040 4041 4042 4043 4044
/*
 * ext4_block_zero_page_range() zeros out a mapping of length 'length'
 * starting from file offset 'from'.  The range to be zero'd must
 * be contained with in one block.  If the specified range exceeds
 * the end of the block it will be shortened to end of the block
 * that cooresponds to 'from'
 */
static int ext4_block_zero_page_range(handle_t *handle,
		struct address_space *mapping, loff_t from, loff_t length)
{
	struct inode *inode = mapping->host;
4045
	unsigned offset = from & (PAGE_SIZE-1);
R
Ross Zwisler 已提交
4046 4047 4048 4049 4050 4051 4052 4053 4054 4055
	unsigned blocksize = inode->i_sb->s_blocksize;
	unsigned max = blocksize - (offset & (blocksize - 1));

	/*
	 * correct length if it does not fall between
	 * 'from' and the end of the block
	 */
	if (length > max || length < 0)
		length = max;

4056 4057 4058 4059
	if (IS_DAX(inode)) {
		return iomap_zero_range(inode, from, length, NULL,
					&ext4_iomap_ops);
	}
R
Ross Zwisler 已提交
4060 4061 4062
	return __ext4_block_zero_page_range(handle, mapping, from, length);
}

4063 4064 4065 4066 4067 4068
/*
 * ext4_block_truncate_page() zeroes out a mapping from file offset `from'
 * up to the end of the block which corresponds to `from'.
 * This required during truncate. We need to physically zero the tail end
 * of that block so it doesn't yield old data if the file is later grown.
 */
4069
static int ext4_block_truncate_page(handle_t *handle,
4070 4071
		struct address_space *mapping, loff_t from)
{
4072
	unsigned offset = from & (PAGE_SIZE-1);
4073 4074 4075 4076
	unsigned length;
	unsigned blocksize;
	struct inode *inode = mapping->host;

4077 4078 4079 4080
	/* If we are processing an encrypted inode during orphan list handling */
	if (ext4_encrypted_inode(inode) && !fscrypt_has_encryption_key(inode))
		return 0;

4081 4082 4083 4084 4085 4086
	blocksize = inode->i_sb->s_blocksize;
	length = blocksize - (offset & (blocksize - 1));

	return ext4_block_zero_page_range(handle, mapping, from, length);
}

4087 4088 4089 4090 4091
int ext4_zero_partial_blocks(handle_t *handle, struct inode *inode,
			     loff_t lstart, loff_t length)
{
	struct super_block *sb = inode->i_sb;
	struct address_space *mapping = inode->i_mapping;
4092
	unsigned partial_start, partial_end;
4093 4094 4095 4096
	ext4_fsblk_t start, end;
	loff_t byte_end = (lstart + length - 1);
	int err = 0;

4097 4098 4099
	partial_start = lstart & (sb->s_blocksize - 1);
	partial_end = byte_end & (sb->s_blocksize - 1);

4100 4101 4102 4103
	start = lstart >> sb->s_blocksize_bits;
	end = byte_end >> sb->s_blocksize_bits;

	/* Handle partial zero within the single block */
4104 4105
	if (start == end &&
	    (partial_start || (partial_end != sb->s_blocksize - 1))) {
4106 4107 4108 4109 4110
		err = ext4_block_zero_page_range(handle, mapping,
						 lstart, length);
		return err;
	}
	/* Handle partial zero out on the start of the range */
4111
	if (partial_start) {
4112 4113 4114 4115 4116 4117
		err = ext4_block_zero_page_range(handle, mapping,
						 lstart, sb->s_blocksize);
		if (err)
			return err;
	}
	/* Handle partial zero out on the end of the range */
4118
	if (partial_end != sb->s_blocksize - 1)
4119
		err = ext4_block_zero_page_range(handle, mapping,
4120 4121
						 byte_end - partial_end,
						 partial_end + 1);
4122 4123 4124
	return err;
}

4125 4126 4127 4128 4129 4130 4131 4132 4133 4134 4135
int ext4_can_truncate(struct inode *inode)
{
	if (S_ISREG(inode->i_mode))
		return 1;
	if (S_ISDIR(inode->i_mode))
		return 1;
	if (S_ISLNK(inode->i_mode))
		return !ext4_inode_is_fast_symlink(inode);
	return 0;
}

4136 4137 4138 4139 4140 4141 4142 4143 4144 4145 4146 4147
/*
 * We have to make sure i_disksize gets properly updated before we truncate
 * page cache due to hole punching or zero range. Otherwise i_disksize update
 * can get lost as it may have been postponed to submission of writeback but
 * that will never happen after we truncate page cache.
 */
int ext4_update_disksize_before_punch(struct inode *inode, loff_t offset,
				      loff_t len)
{
	handle_t *handle;
	loff_t size = i_size_read(inode);

A
Al Viro 已提交
4148
	WARN_ON(!inode_is_locked(inode));
4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160 4161 4162 4163 4164
	if (offset > size || offset + len < size)
		return 0;

	if (EXT4_I(inode)->i_disksize >= size)
		return 0;

	handle = ext4_journal_start(inode, EXT4_HT_MISC, 1);
	if (IS_ERR(handle))
		return PTR_ERR(handle);
	ext4_update_i_disksize(inode, size);
	ext4_mark_inode_dirty(handle, inode);
	ext4_journal_stop(handle);

	return 0;
}

4165
/*
4166
 * ext4_punch_hole: punches a hole in a file by releasing the blocks
4167 4168 4169 4170 4171 4172
 * associated with the given offset and length
 *
 * @inode:  File inode
 * @offset: The offset where the hole will begin
 * @len:    The length of the hole
 *
4173
 * Returns: 0 on success or negative on failure
4174 4175
 */

4176
int ext4_punch_hole(struct inode *inode, loff_t offset, loff_t length)
4177
{
T
Theodore Ts'o 已提交
4178 4179 4180
	struct super_block *sb = inode->i_sb;
	ext4_lblk_t first_block, stop_block;
	struct address_space *mapping = inode->i_mapping;
4181
	loff_t first_block_offset, last_block_offset;
T
Theodore Ts'o 已提交
4182 4183 4184 4185
	handle_t *handle;
	unsigned int credits;
	int ret = 0;

4186
	if (!S_ISREG(inode->i_mode))
4187
		return -EOPNOTSUPP;
4188

4189
	trace_ext4_punch_hole(inode, offset, length, 0);
4190

T
Theodore Ts'o 已提交
4191 4192 4193 4194
	/*
	 * Write out all dirty pages to avoid race conditions
	 * Then release them.
	 */
4195
	if (mapping_tagged(mapping, PAGECACHE_TAG_DIRTY)) {
T
Theodore Ts'o 已提交
4196 4197 4198 4199 4200 4201
		ret = filemap_write_and_wait_range(mapping, offset,
						   offset + length - 1);
		if (ret)
			return ret;
	}

A
Al Viro 已提交
4202
	inode_lock(inode);
4203

T
Theodore Ts'o 已提交
4204 4205 4206 4207 4208 4209 4210 4211 4212 4213
	/* No need to punch hole beyond i_size */
	if (offset >= inode->i_size)
		goto out_mutex;

	/*
	 * If the hole extends beyond i_size, set the hole
	 * to end after the page that contains i_size
	 */
	if (offset + length > inode->i_size) {
		length = inode->i_size +
4214
		   PAGE_SIZE - (inode->i_size & (PAGE_SIZE - 1)) -
T
Theodore Ts'o 已提交
4215 4216 4217
		   offset;
	}

4218 4219 4220 4221 4222 4223 4224 4225 4226 4227 4228 4229
	if (offset & (sb->s_blocksize - 1) ||
	    (offset + length) & (sb->s_blocksize - 1)) {
		/*
		 * Attach jinode to inode for jbd2 if we do any zeroing of
		 * partial block
		 */
		ret = ext4_inode_attach_jinode(inode);
		if (ret < 0)
			goto out_mutex;

	}

4230 4231 4232 4233 4234 4235 4236 4237 4238
	/* Wait all existing dio workers, newcomers will block on i_mutex */
	ext4_inode_block_unlocked_dio(inode);
	inode_dio_wait(inode);

	/*
	 * Prevent page faults from reinstantiating pages we have released from
	 * page cache.
	 */
	down_write(&EXT4_I(inode)->i_mmap_sem);
4239 4240
	first_block_offset = round_up(offset, sb->s_blocksize);
	last_block_offset = round_down((offset + length), sb->s_blocksize) - 1;
T
Theodore Ts'o 已提交
4241

4242
	/* Now release the pages and zero block aligned part of pages*/
4243 4244 4245 4246
	if (last_block_offset > first_block_offset) {
		ret = ext4_update_disksize_before_punch(inode, offset, length);
		if (ret)
			goto out_dio;
4247 4248
		truncate_pagecache_range(inode, first_block_offset,
					 last_block_offset);
4249
	}
T
Theodore Ts'o 已提交
4250 4251 4252 4253 4254 4255 4256 4257 4258 4259 4260 4261

	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
		credits = ext4_writepage_trans_blocks(inode);
	else
		credits = ext4_blocks_for_truncate(inode);
	handle = ext4_journal_start(inode, EXT4_HT_TRUNCATE, credits);
	if (IS_ERR(handle)) {
		ret = PTR_ERR(handle);
		ext4_std_error(sb, ret);
		goto out_dio;
	}

4262 4263 4264 4265
	ret = ext4_zero_partial_blocks(handle, inode, offset,
				       length);
	if (ret)
		goto out_stop;
T
Theodore Ts'o 已提交
4266 4267 4268 4269 4270 4271 4272 4273 4274 4275 4276 4277 4278 4279 4280 4281 4282 4283 4284 4285 4286 4287 4288

	first_block = (offset + sb->s_blocksize - 1) >>
		EXT4_BLOCK_SIZE_BITS(sb);
	stop_block = (offset + length) >> EXT4_BLOCK_SIZE_BITS(sb);

	/* If there are no blocks to remove, return now */
	if (first_block >= stop_block)
		goto out_stop;

	down_write(&EXT4_I(inode)->i_data_sem);
	ext4_discard_preallocations(inode);

	ret = ext4_es_remove_extent(inode, first_block,
				    stop_block - first_block);
	if (ret) {
		up_write(&EXT4_I(inode)->i_data_sem);
		goto out_stop;
	}

	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
		ret = ext4_ext_remove_space(inode, first_block,
					    stop_block - 1);
	else
4289
		ret = ext4_ind_remove_space(handle, inode, first_block,
T
Theodore Ts'o 已提交
4290 4291
					    stop_block);

T
Theodore Ts'o 已提交
4292
	up_write(&EXT4_I(inode)->i_data_sem);
T
Theodore Ts'o 已提交
4293 4294
	if (IS_SYNC(inode))
		ext4_handle_sync(handle);
4295

4296
	inode->i_mtime = inode->i_ctime = current_time(inode);
T
Theodore Ts'o 已提交
4297
	ext4_mark_inode_dirty(handle, inode);
4298 4299
	if (ret >= 0)
		ext4_update_inode_fsync_trans(handle, inode, 1);
T
Theodore Ts'o 已提交
4300 4301 4302
out_stop:
	ext4_journal_stop(handle);
out_dio:
4303
	up_write(&EXT4_I(inode)->i_mmap_sem);
T
Theodore Ts'o 已提交
4304 4305
	ext4_inode_resume_unlocked_dio(inode);
out_mutex:
A
Al Viro 已提交
4306
	inode_unlock(inode);
T
Theodore Ts'o 已提交
4307
	return ret;
4308 4309
}

4310 4311 4312 4313 4314 4315 4316 4317 4318 4319 4320 4321 4322 4323 4324 4325 4326 4327 4328 4329 4330 4331 4332 4333 4334
int ext4_inode_attach_jinode(struct inode *inode)
{
	struct ext4_inode_info *ei = EXT4_I(inode);
	struct jbd2_inode *jinode;

	if (ei->jinode || !EXT4_SB(inode->i_sb)->s_journal)
		return 0;

	jinode = jbd2_alloc_inode(GFP_KERNEL);
	spin_lock(&inode->i_lock);
	if (!ei->jinode) {
		if (!jinode) {
			spin_unlock(&inode->i_lock);
			return -ENOMEM;
		}
		ei->jinode = jinode;
		jbd2_journal_init_jbd_inode(ei->jinode, inode);
		jinode = NULL;
	}
	spin_unlock(&inode->i_lock);
	if (unlikely(jinode != NULL))
		jbd2_free_inode(jinode);
	return 0;
}

4335
/*
4336
 * ext4_truncate()
4337
 *
4338 4339
 * We block out ext4_get_block() block instantiations across the entire
 * transaction, and VFS/VM ensures that ext4_truncate() cannot run
4340 4341
 * simultaneously on behalf of the same inode.
 *
4342
 * As we work through the truncate and commit bits of it to the journal there
4343 4344 4345 4346 4347 4348 4349 4350 4351 4352 4353 4354 4355
 * is one core, guiding principle: the file's tree must always be consistent on
 * disk.  We must be able to restart the truncate after a crash.
 *
 * The file's tree may be transiently inconsistent in memory (although it
 * probably isn't), but whenever we close off and commit a journal transaction,
 * the contents of (the filesystem + the journal) must be consistent and
 * restartable.  It's pretty simple, really: bottom up, right to left (although
 * left-to-right works OK too).
 *
 * Note that at recovery time, journal replay occurs *before* the restart of
 * truncate against the orphan inode list.
 *
 * The committed inode has the new, desired i_size (which is the same as
4356
 * i_disksize in this case).  After a crash, ext4_orphan_cleanup() will see
4357
 * that this inode's truncate did not complete and it will again call
4358 4359
 * ext4_truncate() to have another go.  So there will be instantiated blocks
 * to the right of the truncation point in a crashed ext4 filesystem.  But
4360
 * that's fine - as long as they are linked from the inode, the post-crash
4361
 * ext4_truncate() run will find them and release them.
4362
 */
4363
int ext4_truncate(struct inode *inode)
4364
{
T
Theodore Ts'o 已提交
4365 4366
	struct ext4_inode_info *ei = EXT4_I(inode);
	unsigned int credits;
4367
	int err = 0;
T
Theodore Ts'o 已提交
4368 4369 4370
	handle_t *handle;
	struct address_space *mapping = inode->i_mapping;

4371 4372
	/*
	 * There is a possibility that we're either freeing the inode
M
Matthew Wilcox 已提交
4373
	 * or it's a completely new inode. In those cases we might not
4374 4375 4376
	 * have i_mutex locked because it's not necessary.
	 */
	if (!(inode->i_state & (I_NEW|I_FREEING)))
A
Al Viro 已提交
4377
		WARN_ON(!inode_is_locked(inode));
4378 4379
	trace_ext4_truncate_enter(inode);

4380
	if (!ext4_can_truncate(inode))
4381
		return 0;
4382

4383
	ext4_clear_inode_flag(inode, EXT4_INODE_EOFBLOCKS);
4384

4385
	if (inode->i_size == 0 && !test_opt(inode->i_sb, NO_AUTO_DA_ALLOC))
4386
		ext4_set_inode_state(inode, EXT4_STATE_DA_ALLOC_CLOSE);
4387

4388 4389 4390
	if (ext4_has_inline_data(inode)) {
		int has_inline = 1;

4391 4392 4393
		err = ext4_inline_data_truncate(inode, &has_inline);
		if (err)
			return err;
4394
		if (has_inline)
4395
			return 0;
4396 4397
	}

4398 4399 4400
	/* If we zero-out tail of the page, we have to create jinode for jbd2 */
	if (inode->i_size & (inode->i_sb->s_blocksize - 1)) {
		if (ext4_inode_attach_jinode(inode) < 0)
4401
			return 0;
4402 4403
	}

T
Theodore Ts'o 已提交
4404 4405 4406 4407 4408 4409
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
		credits = ext4_writepage_trans_blocks(inode);
	else
		credits = ext4_blocks_for_truncate(inode);

	handle = ext4_journal_start(inode, EXT4_HT_TRUNCATE, credits);
4410 4411
	if (IS_ERR(handle))
		return PTR_ERR(handle);
T
Theodore Ts'o 已提交
4412

4413 4414
	if (inode->i_size & (inode->i_sb->s_blocksize - 1))
		ext4_block_truncate_page(handle, mapping, inode->i_size);
T
Theodore Ts'o 已提交
4415 4416 4417 4418 4419 4420 4421 4422 4423 4424

	/*
	 * We add the inode to the orphan list, so that if this
	 * truncate spans multiple transactions, and we crash, we will
	 * resume the truncate when the filesystem recovers.  It also
	 * marks the inode dirty, to catch the new size.
	 *
	 * Implication: the file must always be in a sane, consistent
	 * truncatable state while each transaction commits.
	 */
4425 4426
	err = ext4_orphan_add(handle, inode);
	if (err)
T
Theodore Ts'o 已提交
4427 4428 4429 4430 4431 4432
		goto out_stop;

	down_write(&EXT4_I(inode)->i_data_sem);

	ext4_discard_preallocations(inode);

4433
	if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))
4434
		err = ext4_ext_truncate(handle, inode);
4435
	else
T
Theodore Ts'o 已提交
4436 4437 4438
		ext4_ind_truncate(handle, inode);

	up_write(&ei->i_data_sem);
4439 4440
	if (err)
		goto out_stop;
T
Theodore Ts'o 已提交
4441 4442 4443 4444 4445 4446 4447 4448 4449

	if (IS_SYNC(inode))
		ext4_handle_sync(handle);

out_stop:
	/*
	 * If this was a simple ftruncate() and the file will remain alive,
	 * then we need to clear up the orphan record which we created above.
	 * However, if this was a real unlink then we were called by
4450
	 * ext4_evict_inode(), and we allow that function to clean up the
T
Theodore Ts'o 已提交
4451 4452 4453 4454 4455
	 * orphan info for us.
	 */
	if (inode->i_nlink)
		ext4_orphan_del(handle, inode);

4456
	inode->i_mtime = inode->i_ctime = current_time(inode);
T
Theodore Ts'o 已提交
4457 4458
	ext4_mark_inode_dirty(handle, inode);
	ext4_journal_stop(handle);
4459

4460
	trace_ext4_truncate_exit(inode);
4461
	return err;
4462 4463 4464
}

/*
4465
 * ext4_get_inode_loc returns with an extra refcount against the inode's
4466 4467 4468 4469
 * underlying buffer_head on success. If 'in_mem' is true, we have all
 * data in memory that is needed to recreate the on-disk version of this
 * inode.
 */
4470 4471
static int __ext4_get_inode_loc(struct inode *inode,
				struct ext4_iloc *iloc, int in_mem)
4472
{
4473 4474 4475 4476 4477 4478
	struct ext4_group_desc	*gdp;
	struct buffer_head	*bh;
	struct super_block	*sb = inode->i_sb;
	ext4_fsblk_t		block;
	int			inodes_per_block, inode_offset;

A
Aneesh Kumar K.V 已提交
4479
	iloc->bh = NULL;
4480
	if (!ext4_valid_inum(sb, inode->i_ino))
4481
		return -EFSCORRUPTED;
4482

4483 4484 4485
	iloc->block_group = (inode->i_ino - 1) / EXT4_INODES_PER_GROUP(sb);
	gdp = ext4_get_group_desc(sb, iloc->block_group, NULL);
	if (!gdp)
4486 4487
		return -EIO;

4488 4489 4490
	/*
	 * Figure out the offset within the block group inode table
	 */
4491
	inodes_per_block = EXT4_SB(sb)->s_inodes_per_block;
4492 4493 4494 4495 4496 4497
	inode_offset = ((inode->i_ino - 1) %
			EXT4_INODES_PER_GROUP(sb));
	block = ext4_inode_table(sb, gdp) + (inode_offset / inodes_per_block);
	iloc->offset = (inode_offset % inodes_per_block) * EXT4_INODE_SIZE(sb);

	bh = sb_getblk(sb, block);
4498
	if (unlikely(!bh))
4499
		return -ENOMEM;
4500 4501
	if (!buffer_uptodate(bh)) {
		lock_buffer(bh);
4502 4503 4504 4505 4506 4507 4508 4509 4510 4511

		/*
		 * If the buffer has the write error flag, we have failed
		 * to write out another inode in the same block.  In this
		 * case, we don't have to read the block because we may
		 * read the old inode data successfully.
		 */
		if (buffer_write_io_error(bh) && !buffer_uptodate(bh))
			set_buffer_uptodate(bh);

4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524
		if (buffer_uptodate(bh)) {
			/* someone brought it uptodate while we waited */
			unlock_buffer(bh);
			goto has_buffer;
		}

		/*
		 * If we have all information of the inode in memory and this
		 * is the only valid inode in the block, we need not read the
		 * block.
		 */
		if (in_mem) {
			struct buffer_head *bitmap_bh;
4525
			int i, start;
4526

4527
			start = inode_offset & ~(inodes_per_block - 1);
4528

4529 4530
			/* Is the inode bitmap in cache? */
			bitmap_bh = sb_getblk(sb, ext4_inode_bitmap(sb, gdp));
4531
			if (unlikely(!bitmap_bh))
4532 4533 4534 4535 4536 4537 4538 4539 4540 4541 4542
				goto make_io;

			/*
			 * If the inode bitmap isn't in cache then the
			 * optimisation may end up performing two reads instead
			 * of one, so skip it.
			 */
			if (!buffer_uptodate(bitmap_bh)) {
				brelse(bitmap_bh);
				goto make_io;
			}
4543
			for (i = start; i < start + inodes_per_block; i++) {
4544 4545
				if (i == inode_offset)
					continue;
4546
				if (ext4_test_bit(i, bitmap_bh->b_data))
4547 4548 4549
					break;
			}
			brelse(bitmap_bh);
4550
			if (i == start + inodes_per_block) {
4551 4552 4553 4554 4555 4556 4557 4558 4559
				/* all other inodes are free, so skip I/O */
				memset(bh->b_data, 0, bh->b_size);
				set_buffer_uptodate(bh);
				unlock_buffer(bh);
				goto has_buffer;
			}
		}

make_io:
4560 4561 4562 4563 4564 4565 4566
		/*
		 * If we need to do any I/O, try to pre-readahead extra
		 * blocks from the inode table.
		 */
		if (EXT4_SB(sb)->s_inode_readahead_blks) {
			ext4_fsblk_t b, end, table;
			unsigned num;
4567
			__u32 ra_blks = EXT4_SB(sb)->s_inode_readahead_blks;
4568 4569

			table = ext4_inode_table(sb, gdp);
T
Theodore Ts'o 已提交
4570
			/* s_inode_readahead_blks is always a power of 2 */
4571
			b = block & ~((ext4_fsblk_t) ra_blks - 1);
4572 4573
			if (table > b)
				b = table;
4574
			end = b + ra_blks;
4575
			num = EXT4_INODES_PER_GROUP(sb);
4576
			if (ext4_has_group_desc_csum(sb))
4577
				num -= ext4_itable_unused_count(sb, gdp);
4578 4579 4580 4581 4582 4583 4584
			table += num / inodes_per_block;
			if (end > table)
				end = table;
			while (b <= end)
				sb_breadahead(sb, b++);
		}

4585 4586 4587 4588 4589
		/*
		 * There are other valid inodes in the buffer, this inode
		 * has in-inode xattrs, or we don't have this inode in memory.
		 * Read the block from disk.
		 */
4590
		trace_ext4_load_inode(inode);
4591 4592
		get_bh(bh);
		bh->b_end_io = end_buffer_read_sync;
4593
		submit_bh(REQ_OP_READ, REQ_META | REQ_PRIO, bh);
4594 4595
		wait_on_buffer(bh);
		if (!buffer_uptodate(bh)) {
4596 4597
			EXT4_ERROR_INODE_BLOCK(inode, block,
					       "unable to read itable block");
4598 4599 4600 4601 4602 4603 4604 4605 4606
			brelse(bh);
			return -EIO;
		}
	}
has_buffer:
	iloc->bh = bh;
	return 0;
}

4607
int ext4_get_inode_loc(struct inode *inode, struct ext4_iloc *iloc)
4608 4609
{
	/* We have all inode data except xattrs in memory here. */
4610
	return __ext4_get_inode_loc(inode, iloc,
4611
		!ext4_test_inode_state(inode, EXT4_STATE_XATTR));
4612 4613
}

R
Ross Zwisler 已提交
4614 4615 4616 4617 4618 4619 4620 4621 4622 4623 4624 4625 4626 4627 4628
static bool ext4_should_use_dax(struct inode *inode)
{
	if (!test_opt(inode->i_sb, DAX))
		return false;
	if (!S_ISREG(inode->i_mode))
		return false;
	if (ext4_should_journal_data(inode))
		return false;
	if (ext4_has_inline_data(inode))
		return false;
	if (ext4_encrypted_inode(inode))
		return false;
	return true;
}

4629
void ext4_set_inode_flags(struct inode *inode)
4630
{
4631
	unsigned int flags = EXT4_I(inode)->i_flags;
4632
	unsigned int new_fl = 0;
4633

4634
	if (flags & EXT4_SYNC_FL)
4635
		new_fl |= S_SYNC;
4636
	if (flags & EXT4_APPEND_FL)
4637
		new_fl |= S_APPEND;
4638
	if (flags & EXT4_IMMUTABLE_FL)
4639
		new_fl |= S_IMMUTABLE;
4640
	if (flags & EXT4_NOATIME_FL)
4641
		new_fl |= S_NOATIME;
4642
	if (flags & EXT4_DIRSYNC_FL)
4643
		new_fl |= S_DIRSYNC;
R
Ross Zwisler 已提交
4644
	if (ext4_should_use_dax(inode))
R
Ross Zwisler 已提交
4645
		new_fl |= S_DAX;
4646 4647
	if (flags & EXT4_ENCRYPT_FL)
		new_fl |= S_ENCRYPTED;
4648
	inode_set_flags(inode, new_fl,
4649 4650
			S_SYNC|S_APPEND|S_IMMUTABLE|S_NOATIME|S_DIRSYNC|S_DAX|
			S_ENCRYPTED);
4651 4652
}

4653
static blkcnt_t ext4_inode_blocks(struct ext4_inode *raw_inode,
4654
				  struct ext4_inode_info *ei)
4655 4656
{
	blkcnt_t i_blocks ;
A
Aneesh Kumar K.V 已提交
4657 4658
	struct inode *inode = &(ei->vfs_inode);
	struct super_block *sb = inode->i_sb;
4659

4660
	if (ext4_has_feature_huge_file(sb)) {
4661 4662 4663
		/* we are using combined 48 bit field */
		i_blocks = ((u64)le16_to_cpu(raw_inode->i_blocks_high)) << 32 |
					le32_to_cpu(raw_inode->i_blocks_lo);
4664
		if (ext4_test_inode_flag(inode, EXT4_INODE_HUGE_FILE)) {
A
Aneesh Kumar K.V 已提交
4665 4666 4667 4668 4669
			/* i_blocks represent file system block size */
			return i_blocks  << (inode->i_blkbits - 9);
		} else {
			return i_blocks;
		}
4670 4671 4672 4673
	} else {
		return le32_to_cpu(raw_inode->i_blocks_lo);
	}
}
4674

4675 4676 4677 4678 4679 4680
static inline void ext4_iget_extra_inode(struct inode *inode,
					 struct ext4_inode *raw_inode,
					 struct ext4_inode_info *ei)
{
	__le32 *magic = (void *)raw_inode +
			EXT4_GOOD_OLD_INODE_SIZE + ei->i_extra_isize;
4681 4682 4683
	if (EXT4_GOOD_OLD_INODE_SIZE + ei->i_extra_isize + sizeof(__le32) <=
	    EXT4_INODE_SIZE(inode->i_sb) &&
	    *magic == cpu_to_le32(EXT4_XATTR_MAGIC)) {
4684
		ext4_set_inode_state(inode, EXT4_STATE_XATTR);
4685
		ext4_find_inline_data_nolock(inode);
4686 4687
	} else
		EXT4_I(inode)->i_inline_off = 0;
4688 4689
}

L
Li Xi 已提交
4690 4691
int ext4_get_projid(struct inode *inode, kprojid_t *projid)
{
K
Kaho Ng 已提交
4692
	if (!ext4_has_feature_project(inode->i_sb))
L
Li Xi 已提交
4693 4694 4695 4696 4697
		return -EOPNOTSUPP;
	*projid = EXT4_I(inode)->i_projid;
	return 0;
}

4698
struct inode *ext4_iget(struct super_block *sb, unsigned long ino)
4699
{
4700 4701
	struct ext4_iloc iloc;
	struct ext4_inode *raw_inode;
4702 4703
	struct ext4_inode_info *ei;
	struct inode *inode;
4704
	journal_t *journal = EXT4_SB(sb)->s_journal;
4705
	long ret;
4706
	loff_t size;
4707
	int block;
4708 4709
	uid_t i_uid;
	gid_t i_gid;
L
Li Xi 已提交
4710
	projid_t i_projid;
4711

4712 4713 4714 4715 4716 4717 4718
	inode = iget_locked(sb, ino);
	if (!inode)
		return ERR_PTR(-ENOMEM);
	if (!(inode->i_state & I_NEW))
		return inode;

	ei = EXT4_I(inode);
4719
	iloc.bh = NULL;
4720

4721 4722
	ret = __ext4_get_inode_loc(inode, &iloc, 0);
	if (ret < 0)
4723
		goto bad_inode;
4724
	raw_inode = ext4_raw_inode(&iloc);
4725 4726 4727 4728

	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE) {
		ei->i_extra_isize = le16_to_cpu(raw_inode->i_extra_isize);
		if (EXT4_GOOD_OLD_INODE_SIZE + ei->i_extra_isize >
4729 4730 4731 4732 4733 4734
			EXT4_INODE_SIZE(inode->i_sb) ||
		    (ei->i_extra_isize & 3)) {
			EXT4_ERROR_INODE(inode,
					 "bad extra_isize %u (inode size %u)",
					 ei->i_extra_isize,
					 EXT4_INODE_SIZE(inode->i_sb));
4735
			ret = -EFSCORRUPTED;
4736 4737 4738 4739 4740 4741
			goto bad_inode;
		}
	} else
		ei->i_extra_isize = 0;

	/* Precompute checksum seed for inode metadata */
4742
	if (ext4_has_metadata_csum(sb)) {
4743 4744 4745 4746 4747 4748 4749 4750 4751 4752 4753 4754
		struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
		__u32 csum;
		__le32 inum = cpu_to_le32(inode->i_ino);
		__le32 gen = raw_inode->i_generation;
		csum = ext4_chksum(sbi, sbi->s_csum_seed, (__u8 *)&inum,
				   sizeof(inum));
		ei->i_csum_seed = ext4_chksum(sbi, csum, (__u8 *)&gen,
					      sizeof(gen));
	}

	if (!ext4_inode_csum_verify(inode, raw_inode, ei)) {
		EXT4_ERROR_INODE(inode, "checksum invalid");
4755
		ret = -EFSBADCRC;
4756 4757 4758
		goto bad_inode;
	}

4759
	inode->i_mode = le16_to_cpu(raw_inode->i_mode);
4760 4761
	i_uid = (uid_t)le16_to_cpu(raw_inode->i_uid_low);
	i_gid = (gid_t)le16_to_cpu(raw_inode->i_gid_low);
K
Kaho Ng 已提交
4762
	if (ext4_has_feature_project(sb) &&
L
Li Xi 已提交
4763 4764 4765 4766 4767 4768
	    EXT4_INODE_SIZE(sb) > EXT4_GOOD_OLD_INODE_SIZE &&
	    EXT4_FITS_IN_INODE(raw_inode, ei, i_projid))
		i_projid = (projid_t)le32_to_cpu(raw_inode->i_projid);
	else
		i_projid = EXT4_DEF_PROJID;

4769
	if (!(test_opt(inode->i_sb, NO_UID32))) {
4770 4771
		i_uid |= le16_to_cpu(raw_inode->i_uid_high) << 16;
		i_gid |= le16_to_cpu(raw_inode->i_gid_high) << 16;
4772
	}
4773 4774
	i_uid_write(inode, i_uid);
	i_gid_write(inode, i_gid);
L
Li Xi 已提交
4775
	ei->i_projid = make_kprojid(&init_user_ns, i_projid);
M
Miklos Szeredi 已提交
4776
	set_nlink(inode, le16_to_cpu(raw_inode->i_links_count));
4777

4778
	ext4_clear_state_flags(ei);	/* Only relevant on 32-bit archs */
4779
	ei->i_inline_off = 0;
4780 4781 4782 4783 4784 4785 4786 4787
	ei->i_dir_start_lookup = 0;
	ei->i_dtime = le32_to_cpu(raw_inode->i_dtime);
	/* We now have enough fields to check if the inode was active or not.
	 * This is needed because nfsd might try to access dead inodes
	 * the test is that same one that e2fsck uses
	 * NeilBrown 1999oct15
	 */
	if (inode->i_nlink == 0) {
4788 4789 4790
		if ((inode->i_mode == 0 ||
		     !(EXT4_SB(inode->i_sb)->s_mount_state & EXT4_ORPHAN_FS)) &&
		    ino != EXT4_BOOT_LOADER_INO) {
4791
			/* this inode is deleted */
4792
			ret = -ESTALE;
4793 4794 4795 4796 4797
			goto bad_inode;
		}
		/* The only unlinked inodes we let through here have
		 * valid i_mode and are being read by the orphan
		 * recovery code: that's fine, we're about to complete
4798 4799 4800
		 * the process of deleting those.
		 * OR it is the EXT4_BOOT_LOADER_INO which is
		 * not initialized on a new filesystem. */
4801 4802
	}
	ei->i_flags = le32_to_cpu(raw_inode->i_flags);
4803
	inode->i_blocks = ext4_inode_blocks(raw_inode, ei);
4804
	ei->i_file_acl = le32_to_cpu(raw_inode->i_file_acl_lo);
4805
	if (ext4_has_feature_64bit(sb))
B
Badari Pulavarty 已提交
4806 4807
		ei->i_file_acl |=
			((__u64)le16_to_cpu(raw_inode->i_file_acl_high)) << 32;
A
Artem Blagodarenko 已提交
4808
	inode->i_size = ext4_isize(sb, raw_inode);
4809 4810 4811 4812 4813
	if ((size = i_size_read(inode)) < 0) {
		EXT4_ERROR_INODE(inode, "bad i_size value: %lld", size);
		ret = -EFSCORRUPTED;
		goto bad_inode;
	}
4814
	ei->i_disksize = inode->i_size;
4815 4816 4817
#ifdef CONFIG_QUOTA
	ei->i_reserved_quota = 0;
#endif
4818 4819
	inode->i_generation = le32_to_cpu(raw_inode->i_generation);
	ei->i_block_group = iloc.block_group;
4820
	ei->i_last_alloc_group = ~0;
4821 4822 4823 4824
	/*
	 * NOTE! The in-memory inode i_data array is in little-endian order
	 * even on big-endian machines: we do NOT byteswap the block numbers!
	 */
4825
	for (block = 0; block < EXT4_N_BLOCKS; block++)
4826 4827 4828
		ei->i_data[block] = raw_inode->i_block[block];
	INIT_LIST_HEAD(&ei->i_orphan);

4829 4830 4831 4832 4833 4834 4835 4836 4837 4838 4839
	/*
	 * Set transaction id's of transactions that have to be committed
	 * to finish f[data]sync. We set them to currently running transaction
	 * as we cannot be sure that the inode or some of its metadata isn't
	 * part of the transaction - the inode could have been reclaimed and
	 * now it is reread from disk.
	 */
	if (journal) {
		transaction_t *transaction;
		tid_t tid;

4840
		read_lock(&journal->j_state_lock);
4841 4842 4843 4844 4845 4846 4847 4848
		if (journal->j_running_transaction)
			transaction = journal->j_running_transaction;
		else
			transaction = journal->j_committing_transaction;
		if (transaction)
			tid = transaction->t_tid;
		else
			tid = journal->j_commit_sequence;
4849
		read_unlock(&journal->j_state_lock);
4850 4851 4852 4853
		ei->i_sync_tid = tid;
		ei->i_datasync_tid = tid;
	}

4854
	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE) {
4855 4856
		if (ei->i_extra_isize == 0) {
			/* The extra space is currently unused. Use it. */
4857
			BUILD_BUG_ON(sizeof(struct ext4_inode) & 3);
4858 4859
			ei->i_extra_isize = sizeof(struct ext4_inode) -
					    EXT4_GOOD_OLD_INODE_SIZE;
4860
		} else {
4861
			ext4_iget_extra_inode(inode, raw_inode, ei);
4862
		}
4863
	}
4864

K
Kalpak Shah 已提交
4865 4866 4867 4868 4869
	EXT4_INODE_GET_XTIME(i_ctime, inode, raw_inode);
	EXT4_INODE_GET_XTIME(i_mtime, inode, raw_inode);
	EXT4_INODE_GET_XTIME(i_atime, inode, raw_inode);
	EXT4_EINODE_GET_XTIME(i_crtime, ei, raw_inode);

4870
	if (likely(!test_opt2(inode->i_sb, HURD_COMPAT))) {
4871 4872 4873 4874 4875 4876
		inode->i_version = le32_to_cpu(raw_inode->i_disk_version);
		if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE) {
			if (EXT4_FITS_IN_INODE(raw_inode, ei, i_version_hi))
				inode->i_version |=
		    (__u64)(le32_to_cpu(raw_inode->i_version_hi)) << 32;
		}
4877 4878
	}

4879
	ret = 0;
4880
	if (ei->i_file_acl &&
4881
	    !ext4_data_block_valid(EXT4_SB(sb), ei->i_file_acl, 1)) {
4882 4883
		EXT4_ERROR_INODE(inode, "bad extended attribute block %llu",
				 ei->i_file_acl);
4884
		ret = -EFSCORRUPTED;
4885
		goto bad_inode;
4886 4887 4888 4889 4890 4891 4892 4893 4894 4895 4896 4897 4898
	} else if (!ext4_has_inline_data(inode)) {
		if (ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)) {
			if ((S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
			    (S_ISLNK(inode->i_mode) &&
			     !ext4_inode_is_fast_symlink(inode))))
				/* Validate extent which is part of inode */
				ret = ext4_ext_check_inode(inode);
		} else if (S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
			   (S_ISLNK(inode->i_mode) &&
			    !ext4_inode_is_fast_symlink(inode))) {
			/* Validate block references which are part of inode */
			ret = ext4_ind_check_inode(inode);
		}
4899
	}
4900
	if (ret)
4901
		goto bad_inode;
4902

4903
	if (S_ISREG(inode->i_mode)) {
4904
		inode->i_op = &ext4_file_inode_operations;
4905
		inode->i_fop = &ext4_file_operations;
4906
		ext4_set_aops(inode);
4907
	} else if (S_ISDIR(inode->i_mode)) {
4908 4909
		inode->i_op = &ext4_dir_inode_operations;
		inode->i_fop = &ext4_dir_operations;
4910
	} else if (S_ISLNK(inode->i_mode)) {
4911 4912 4913 4914
		if (ext4_encrypted_inode(inode)) {
			inode->i_op = &ext4_encrypted_symlink_inode_operations;
			ext4_set_aops(inode);
		} else if (ext4_inode_is_fast_symlink(inode)) {
A
Al Viro 已提交
4915
			inode->i_link = (char *)ei->i_data;
4916
			inode->i_op = &ext4_fast_symlink_inode_operations;
4917 4918 4919
			nd_terminate_link(ei->i_data, inode->i_size,
				sizeof(ei->i_data) - 1);
		} else {
4920 4921
			inode->i_op = &ext4_symlink_inode_operations;
			ext4_set_aops(inode);
4922
		}
4923
		inode_nohighmem(inode);
4924 4925
	} else if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode) ||
	      S_ISFIFO(inode->i_mode) || S_ISSOCK(inode->i_mode)) {
4926
		inode->i_op = &ext4_special_inode_operations;
4927 4928 4929 4930 4931 4932
		if (raw_inode->i_block[0])
			init_special_inode(inode, inode->i_mode,
			   old_decode_dev(le32_to_cpu(raw_inode->i_block[0])));
		else
			init_special_inode(inode, inode->i_mode,
			   new_decode_dev(le32_to_cpu(raw_inode->i_block[1])));
4933 4934
	} else if (ino == EXT4_BOOT_LOADER_INO) {
		make_bad_inode(inode);
4935
	} else {
4936
		ret = -EFSCORRUPTED;
4937
		EXT4_ERROR_INODE(inode, "bogus i_mode (%o)", inode->i_mode);
4938
		goto bad_inode;
4939
	}
4940
	brelse(iloc.bh);
4941
	ext4_set_inode_flags(inode);
T
Tahsin Erdogan 已提交
4942

4943 4944
	unlock_new_inode(inode);
	return inode;
4945 4946

bad_inode:
4947
	brelse(iloc.bh);
4948 4949
	iget_failed(inode);
	return ERR_PTR(ret);
4950 4951
}

4952 4953 4954
struct inode *ext4_iget_normal(struct super_block *sb, unsigned long ino)
{
	if (ino < EXT4_FIRST_INO(sb) && ino != EXT4_ROOT_INO)
4955
		return ERR_PTR(-EFSCORRUPTED);
4956 4957 4958
	return ext4_iget(sb, ino);
}

4959 4960 4961 4962 4963 4964 4965 4966 4967 4968
static int ext4_inode_blocks_set(handle_t *handle,
				struct ext4_inode *raw_inode,
				struct ext4_inode_info *ei)
{
	struct inode *inode = &(ei->vfs_inode);
	u64 i_blocks = inode->i_blocks;
	struct super_block *sb = inode->i_sb;

	if (i_blocks <= ~0U) {
		/*
4969
		 * i_blocks can be represented in a 32 bit variable
4970 4971
		 * as multiple of 512 bytes
		 */
A
Aneesh Kumar K.V 已提交
4972
		raw_inode->i_blocks_lo   = cpu_to_le32(i_blocks);
4973
		raw_inode->i_blocks_high = 0;
4974
		ext4_clear_inode_flag(inode, EXT4_INODE_HUGE_FILE);
4975 4976
		return 0;
	}
4977
	if (!ext4_has_feature_huge_file(sb))
4978 4979 4980
		return -EFBIG;

	if (i_blocks <= 0xffffffffffffULL) {
4981 4982 4983 4984
		/*
		 * i_blocks can be represented in a 48 bit variable
		 * as multiple of 512 bytes
		 */
A
Aneesh Kumar K.V 已提交
4985
		raw_inode->i_blocks_lo   = cpu_to_le32(i_blocks);
4986
		raw_inode->i_blocks_high = cpu_to_le16(i_blocks >> 32);
4987
		ext4_clear_inode_flag(inode, EXT4_INODE_HUGE_FILE);
4988
	} else {
4989
		ext4_set_inode_flag(inode, EXT4_INODE_HUGE_FILE);
A
Aneesh Kumar K.V 已提交
4990 4991 4992 4993
		/* i_block is stored in file system block size */
		i_blocks = i_blocks >> (inode->i_blkbits - 9);
		raw_inode->i_blocks_lo   = cpu_to_le32(i_blocks);
		raw_inode->i_blocks_high = cpu_to_le16(i_blocks >> 32);
4994
	}
4995
	return 0;
4996 4997
}

4998 4999 5000 5001 5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014 5015 5016 5017 5018 5019 5020 5021 5022 5023 5024 5025 5026 5027 5028 5029 5030 5031 5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047
struct other_inode {
	unsigned long		orig_ino;
	struct ext4_inode	*raw_inode;
};

static int other_inode_match(struct inode * inode, unsigned long ino,
			     void *data)
{
	struct other_inode *oi = (struct other_inode *) data;

	if ((inode->i_ino != ino) ||
	    (inode->i_state & (I_FREEING | I_WILL_FREE | I_NEW |
			       I_DIRTY_SYNC | I_DIRTY_DATASYNC)) ||
	    ((inode->i_state & I_DIRTY_TIME) == 0))
		return 0;
	spin_lock(&inode->i_lock);
	if (((inode->i_state & (I_FREEING | I_WILL_FREE | I_NEW |
				I_DIRTY_SYNC | I_DIRTY_DATASYNC)) == 0) &&
	    (inode->i_state & I_DIRTY_TIME)) {
		struct ext4_inode_info	*ei = EXT4_I(inode);

		inode->i_state &= ~(I_DIRTY_TIME | I_DIRTY_TIME_EXPIRED);
		spin_unlock(&inode->i_lock);

		spin_lock(&ei->i_raw_lock);
		EXT4_INODE_SET_XTIME(i_ctime, inode, oi->raw_inode);
		EXT4_INODE_SET_XTIME(i_mtime, inode, oi->raw_inode);
		EXT4_INODE_SET_XTIME(i_atime, inode, oi->raw_inode);
		ext4_inode_csum_set(inode, oi->raw_inode, ei);
		spin_unlock(&ei->i_raw_lock);
		trace_ext4_other_inode_update_time(inode, oi->orig_ino);
		return -1;
	}
	spin_unlock(&inode->i_lock);
	return -1;
}

/*
 * Opportunistically update the other time fields for other inodes in
 * the same inode table block.
 */
static void ext4_update_other_inodes_time(struct super_block *sb,
					  unsigned long orig_ino, char *buf)
{
	struct other_inode oi;
	unsigned long ino;
	int i, inodes_per_block = EXT4_SB(sb)->s_inodes_per_block;
	int inode_size = EXT4_INODE_SIZE(sb);

	oi.orig_ino = orig_ino;
5048 5049 5050 5051 5052 5053
	/*
	 * Calculate the first inode in the inode table block.  Inode
	 * numbers are one-based.  That is, the first inode in a block
	 * (assuming 4k blocks and 256 byte inodes) is (n*16 + 1).
	 */
	ino = ((orig_ino - 1) & ~(inodes_per_block - 1)) + 1;
5054 5055 5056 5057 5058 5059 5060 5061
	for (i = 0; i < inodes_per_block; i++, ino++, buf += inode_size) {
		if (ino == orig_ino)
			continue;
		oi.raw_inode = (struct ext4_inode *) buf;
		(void) find_inode_nowait(sb, ino, other_inode_match, &oi);
	}
}

5062 5063 5064 5065 5066 5067 5068
/*
 * Post the struct inode info into an on-disk inode location in the
 * buffer-cache.  This gobbles the caller's reference to the
 * buffer_head in the inode location struct.
 *
 * The caller must have write access to iloc->bh.
 */
5069
static int ext4_do_update_inode(handle_t *handle,
5070
				struct inode *inode,
5071
				struct ext4_iloc *iloc)
5072
{
5073 5074
	struct ext4_inode *raw_inode = ext4_raw_inode(iloc);
	struct ext4_inode_info *ei = EXT4_I(inode);
5075
	struct buffer_head *bh = iloc->bh;
5076
	struct super_block *sb = inode->i_sb;
5077
	int err = 0, rc, block;
5078
	int need_datasync = 0, set_large_file = 0;
5079 5080
	uid_t i_uid;
	gid_t i_gid;
L
Li Xi 已提交
5081
	projid_t i_projid;
5082

5083 5084 5085
	spin_lock(&ei->i_raw_lock);

	/* For fields not tracked in the in-memory inode,
5086
	 * initialise them to zero for new inodes. */
5087
	if (ext4_test_inode_state(inode, EXT4_STATE_NEW))
5088
		memset(raw_inode, 0, EXT4_SB(inode->i_sb)->s_inode_size);
5089 5090

	raw_inode->i_mode = cpu_to_le16(inode->i_mode);
5091 5092
	i_uid = i_uid_read(inode);
	i_gid = i_gid_read(inode);
L
Li Xi 已提交
5093
	i_projid = from_kprojid(&init_user_ns, ei->i_projid);
5094
	if (!(test_opt(inode->i_sb, NO_UID32))) {
5095 5096
		raw_inode->i_uid_low = cpu_to_le16(low_16_bits(i_uid));
		raw_inode->i_gid_low = cpu_to_le16(low_16_bits(i_gid));
5097 5098 5099 5100
/*
 * Fix up interoperability with old kernels. Otherwise, old inodes get
 * re-used with the upper 16 bits of the uid/gid intact
 */
5101 5102 5103 5104
		if (ei->i_dtime && list_empty(&ei->i_orphan)) {
			raw_inode->i_uid_high = 0;
			raw_inode->i_gid_high = 0;
		} else {
5105
			raw_inode->i_uid_high =
5106
				cpu_to_le16(high_16_bits(i_uid));
5107
			raw_inode->i_gid_high =
5108
				cpu_to_le16(high_16_bits(i_gid));
5109 5110
		}
	} else {
5111 5112
		raw_inode->i_uid_low = cpu_to_le16(fs_high2lowuid(i_uid));
		raw_inode->i_gid_low = cpu_to_le16(fs_high2lowgid(i_gid));
5113 5114 5115 5116
		raw_inode->i_uid_high = 0;
		raw_inode->i_gid_high = 0;
	}
	raw_inode->i_links_count = cpu_to_le16(inode->i_nlink);
K
Kalpak Shah 已提交
5117 5118 5119 5120 5121 5122

	EXT4_INODE_SET_XTIME(i_ctime, inode, raw_inode);
	EXT4_INODE_SET_XTIME(i_mtime, inode, raw_inode);
	EXT4_INODE_SET_XTIME(i_atime, inode, raw_inode);
	EXT4_EINODE_SET_XTIME(i_crtime, ei, raw_inode);

5123 5124
	err = ext4_inode_blocks_set(handle, raw_inode, ei);
	if (err) {
5125
		spin_unlock(&ei->i_raw_lock);
5126
		goto out_brelse;
5127
	}
5128
	raw_inode->i_dtime = cpu_to_le32(ei->i_dtime);
5129
	raw_inode->i_flags = cpu_to_le32(ei->i_flags & 0xFFFFFFFF);
5130
	if (likely(!test_opt2(inode->i_sb, HURD_COMPAT)))
B
Badari Pulavarty 已提交
5131 5132
		raw_inode->i_file_acl_high =
			cpu_to_le16(ei->i_file_acl >> 32);
5133
	raw_inode->i_file_acl_lo = cpu_to_le32(ei->i_file_acl);
A
Artem Blagodarenko 已提交
5134
	if (ei->i_disksize != ext4_isize(inode->i_sb, raw_inode)) {
5135 5136 5137
		ext4_isize_set(raw_inode, ei->i_disksize);
		need_datasync = 1;
	}
5138
	if (ei->i_disksize > 0x7fffffffULL) {
5139
		if (!ext4_has_feature_large_file(sb) ||
5140
				EXT4_SB(sb)->s_es->s_rev_level ==
5141 5142
		    cpu_to_le32(EXT4_GOOD_OLD_REV))
			set_large_file = 1;
5143 5144 5145 5146 5147 5148 5149 5150 5151 5152 5153 5154 5155
	}
	raw_inode->i_generation = cpu_to_le32(inode->i_generation);
	if (S_ISCHR(inode->i_mode) || S_ISBLK(inode->i_mode)) {
		if (old_valid_dev(inode->i_rdev)) {
			raw_inode->i_block[0] =
				cpu_to_le32(old_encode_dev(inode->i_rdev));
			raw_inode->i_block[1] = 0;
		} else {
			raw_inode->i_block[0] = 0;
			raw_inode->i_block[1] =
				cpu_to_le32(new_encode_dev(inode->i_rdev));
			raw_inode->i_block[2] = 0;
		}
5156
	} else if (!ext4_has_inline_data(inode)) {
5157 5158
		for (block = 0; block < EXT4_N_BLOCKS; block++)
			raw_inode->i_block[block] = ei->i_data[block];
5159
	}
5160

5161
	if (likely(!test_opt2(inode->i_sb, HURD_COMPAT))) {
5162 5163 5164 5165 5166 5167 5168 5169
		raw_inode->i_disk_version = cpu_to_le32(inode->i_version);
		if (ei->i_extra_isize) {
			if (EXT4_FITS_IN_INODE(raw_inode, ei, i_version_hi))
				raw_inode->i_version_hi =
					cpu_to_le32(inode->i_version >> 32);
			raw_inode->i_extra_isize =
				cpu_to_le16(ei->i_extra_isize);
		}
5170
	}
L
Li Xi 已提交
5171

K
Kaho Ng 已提交
5172
	BUG_ON(!ext4_has_feature_project(inode->i_sb) &&
L
Li Xi 已提交
5173 5174 5175 5176 5177 5178
	       i_projid != EXT4_DEF_PROJID);

	if (EXT4_INODE_SIZE(inode->i_sb) > EXT4_GOOD_OLD_INODE_SIZE &&
	    EXT4_FITS_IN_INODE(raw_inode, ei, i_projid))
		raw_inode->i_projid = cpu_to_le32(i_projid);

5179
	ext4_inode_csum_set(inode, raw_inode, ei);
5180
	spin_unlock(&ei->i_raw_lock);
5181 5182 5183
	if (inode->i_sb->s_flags & MS_LAZYTIME)
		ext4_update_other_inodes_time(inode->i_sb, inode->i_ino,
					      bh->b_data);
5184

5185
	BUFFER_TRACE(bh, "call ext4_handle_dirty_metadata");
5186
	rc = ext4_handle_dirty_metadata(handle, NULL, bh);
5187 5188
	if (!err)
		err = rc;
5189
	ext4_clear_inode_state(inode, EXT4_STATE_NEW);
5190
	if (set_large_file) {
5191
		BUFFER_TRACE(EXT4_SB(sb)->s_sbh, "get write access");
5192 5193 5194 5195
		err = ext4_journal_get_write_access(handle, EXT4_SB(sb)->s_sbh);
		if (err)
			goto out_brelse;
		ext4_update_dynamic_rev(sb);
5196
		ext4_set_feature_large_file(sb);
5197 5198 5199
		ext4_handle_sync(handle);
		err = ext4_handle_dirty_super(handle, sb);
	}
5200
	ext4_update_inode_fsync_trans(handle, inode, need_datasync);
5201
out_brelse:
5202
	brelse(bh);
5203
	ext4_std_error(inode->i_sb, err);
5204 5205 5206 5207
	return err;
}

/*
5208
 * ext4_write_inode()
5209 5210 5211
 *
 * We are called from a few places:
 *
5212
 * - Within generic_file_aio_write() -> generic_write_sync() for O_SYNC files.
5213
 *   Here, there will be no transaction running. We wait for any running
5214
 *   transaction to commit.
5215
 *
5216 5217
 * - Within flush work (sys_sync(), kupdate and such).
 *   We wait on commit, if told to.
5218
 *
5219 5220
 * - Within iput_final() -> write_inode_now()
 *   We wait on commit, if told to.
5221 5222 5223
 *
 * In all cases it is actually safe for us to return without doing anything,
 * because the inode has been copied into a raw inode buffer in
5224 5225
 * ext4_mark_inode_dirty().  This is a correctness thing for WB_SYNC_ALL
 * writeback.
5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236
 *
 * Note that we are absolutely dependent upon all inode dirtiers doing the
 * right thing: they *must* call mark_inode_dirty() after dirtying info in
 * which we are interested.
 *
 * It would be a bug for them to not do this.  The code:
 *
 *	mark_inode_dirty(inode)
 *	stuff();
 *	inode->i_size = expr;
 *
5237 5238 5239
 * is in error because write_inode() could occur while `stuff()' is running,
 * and the new i_size will be lost.  Plus the inode will no longer be on the
 * superblock's dirty inode list.
5240
 */
5241
int ext4_write_inode(struct inode *inode, struct writeback_control *wbc)
5242
{
5243 5244
	int err;

5245
	if (WARN_ON_ONCE(current->flags & PF_MEMALLOC))
5246 5247
		return 0;

5248 5249 5250 5251 5252 5253
	if (EXT4_SB(inode->i_sb)->s_journal) {
		if (ext4_journal_current_handle()) {
			jbd_debug(1, "called recursively, non-PF_MEMALLOC!\n");
			dump_stack();
			return -EIO;
		}
5254

5255 5256 5257 5258 5259 5260
		/*
		 * No need to force transaction in WB_SYNC_NONE mode. Also
		 * ext4_sync_fs() will force the commit after everything is
		 * written.
		 */
		if (wbc->sync_mode != WB_SYNC_ALL || wbc->for_sync)
5261 5262 5263 5264 5265
			return 0;

		err = ext4_force_commit(inode->i_sb);
	} else {
		struct ext4_iloc iloc;
5266

5267
		err = __ext4_get_inode_loc(inode, &iloc, 0);
5268 5269
		if (err)
			return err;
5270 5271 5272 5273 5274
		/*
		 * sync(2) will flush the whole buffer cache. No need to do
		 * it here separately for each inode.
		 */
		if (wbc->sync_mode == WB_SYNC_ALL && !wbc->for_sync)
5275 5276
			sync_dirty_buffer(iloc.bh);
		if (buffer_req(iloc.bh) && !buffer_uptodate(iloc.bh)) {
5277 5278
			EXT4_ERROR_INODE_BLOCK(inode, iloc.bh->b_blocknr,
					 "IO error syncing inode");
5279 5280
			err = -EIO;
		}
5281
		brelse(iloc.bh);
5282 5283
	}
	return err;
5284 5285
}

5286 5287 5288 5289 5290 5291 5292 5293 5294 5295 5296 5297 5298
/*
 * In data=journal mode ext4_journalled_invalidatepage() may fail to invalidate
 * buffers that are attached to a page stradding i_size and are undergoing
 * commit. In that case we have to wait for commit to finish and try again.
 */
static void ext4_wait_for_tail_page_commit(struct inode *inode)
{
	struct page *page;
	unsigned offset;
	journal_t *journal = EXT4_SB(inode->i_sb)->s_journal;
	tid_t commit_tid = 0;
	int ret;

5299
	offset = inode->i_size & (PAGE_SIZE - 1);
5300 5301
	/*
	 * All buffers in the last page remain valid? Then there's nothing to
5302
	 * do. We do the check mainly to optimize the common PAGE_SIZE ==
5303 5304
	 * blocksize case
	 */
F
Fabian Frederick 已提交
5305
	if (offset > PAGE_SIZE - i_blocksize(inode))
5306 5307 5308
		return;
	while (1) {
		page = find_lock_page(inode->i_mapping,
5309
				      inode->i_size >> PAGE_SHIFT);
5310 5311
		if (!page)
			return;
5312
		ret = __ext4_journalled_invalidatepage(page, offset,
5313
						PAGE_SIZE - offset);
5314
		unlock_page(page);
5315
		put_page(page);
5316 5317 5318 5319 5320 5321 5322 5323 5324 5325 5326 5327
		if (ret != -EBUSY)
			return;
		commit_tid = 0;
		read_lock(&journal->j_state_lock);
		if (journal->j_committing_transaction)
			commit_tid = journal->j_committing_transaction->t_tid;
		read_unlock(&journal->j_state_lock);
		if (commit_tid)
			jbd2_log_wait_commit(journal, commit_tid);
	}
}

5328
/*
5329
 * ext4_setattr()
5330 5331 5332 5333 5334 5335 5336 5337 5338 5339 5340 5341 5342
 *
 * Called from notify_change.
 *
 * We want to trap VFS attempts to truncate the file as soon as
 * possible.  In particular, we want to make sure that when the VFS
 * shrinks i_size, we put the inode on the orphan list and modify
 * i_disksize immediately, so that during the subsequent flushing of
 * dirty pages and freeing of disk blocks, we can guarantee that any
 * commit will leave the blocks being flushed in an unused state on
 * disk.  (On recovery, the inode will get truncated and the blocks will
 * be freed, so we have a strong guarantee that no future commit will
 * leave these blocks visible to the user.)
 *
5343 5344 5345 5346 5347 5348 5349 5350
 * Another thing we have to assure is that if we are in ordered mode
 * and inode is still attached to the committing transaction, we must
 * we start writeout of all the dirty pages which are being truncated.
 * This way we are sure that all the data written in the previous
 * transaction are already on disk (truncate waits for pages under
 * writeback).
 *
 * Called with inode->i_mutex down.
5351
 */
5352
int ext4_setattr(struct dentry *dentry, struct iattr *attr)
5353
{
5354
	struct inode *inode = d_inode(dentry);
5355
	int error, rc = 0;
5356
	int orphan = 0;
5357 5358
	const unsigned int ia_valid = attr->ia_valid;

5359 5360 5361
	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
		return -EIO;

5362
	error = setattr_prepare(dentry, attr);
5363 5364 5365
	if (error)
		return error;

5366 5367 5368 5369
	error = fscrypt_prepare_setattr(dentry, attr);
	if (error)
		return error;

5370 5371 5372 5373 5374
	if (is_quota_modification(inode, attr)) {
		error = dquot_initialize(inode);
		if (error)
			return error;
	}
5375 5376
	if ((ia_valid & ATTR_UID && !uid_eq(attr->ia_uid, inode->i_uid)) ||
	    (ia_valid & ATTR_GID && !gid_eq(attr->ia_gid, inode->i_gid))) {
5377 5378 5379 5380
		handle_t *handle;

		/* (user+group)*(old+new) structure, inode write (sb,
		 * inode block, ? - but truncate inode update has it) */
5381 5382 5383
		handle = ext4_journal_start(inode, EXT4_HT_QUOTA,
			(EXT4_MAXQUOTAS_INIT_BLOCKS(inode->i_sb) +
			 EXT4_MAXQUOTAS_DEL_BLOCKS(inode->i_sb)) + 3);
5384 5385 5386 5387
		if (IS_ERR(handle)) {
			error = PTR_ERR(handle);
			goto err_out;
		}
5388 5389 5390 5391 5392

		/* dquot_transfer() calls back ext4_get_inode_usage() which
		 * counts xattr inode references.
		 */
		down_read(&EXT4_I(inode)->xattr_sem);
5393
		error = dquot_transfer(inode, attr);
5394 5395
		up_read(&EXT4_I(inode)->xattr_sem);

5396
		if (error) {
5397
			ext4_journal_stop(handle);
5398 5399 5400 5401 5402 5403 5404 5405
			return error;
		}
		/* Update corresponding info in inode so that everything is in
		 * one transaction */
		if (attr->ia_valid & ATTR_UID)
			inode->i_uid = attr->ia_uid;
		if (attr->ia_valid & ATTR_GID)
			inode->i_gid = attr->ia_gid;
5406 5407
		error = ext4_mark_inode_dirty(handle, inode);
		ext4_journal_stop(handle);
5408 5409
	}

5410
	if (attr->ia_valid & ATTR_SIZE) {
5411
		handle_t *handle;
5412 5413
		loff_t oldsize = inode->i_size;
		int shrink = (attr->ia_size <= inode->i_size);
5414

5415
		if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS))) {
5416 5417
			struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);

5418 5419
			if (attr->ia_size > sbi->s_bitmap_maxbytes)
				return -EFBIG;
5420
		}
5421 5422
		if (!S_ISREG(inode->i_mode))
			return -EINVAL;
C
Christoph Hellwig 已提交
5423 5424 5425 5426

		if (IS_I_VERSION(inode) && attr->ia_size != inode->i_size)
			inode_inc_iversion(inode);

5427
		if (ext4_should_order_data(inode) &&
5428
		    (attr->ia_size < inode->i_size)) {
5429
			error = ext4_begin_ordered_truncate(inode,
5430
							    attr->ia_size);
5431 5432 5433 5434
			if (error)
				goto err_out;
		}
		if (attr->ia_size != inode->i_size) {
5435 5436 5437 5438 5439
			handle = ext4_journal_start(inode, EXT4_HT_INODE, 3);
			if (IS_ERR(handle)) {
				error = PTR_ERR(handle);
				goto err_out;
			}
5440
			if (ext4_handle_valid(handle) && shrink) {
5441 5442 5443
				error = ext4_orphan_add(handle, inode);
				orphan = 1;
			}
E
Eryu Guan 已提交
5444 5445 5446 5447 5448
			/*
			 * Update c/mtime on truncate up, ext4_truncate() will
			 * update c/mtime in shrink case below
			 */
			if (!shrink) {
5449
				inode->i_mtime = current_time(inode);
E
Eryu Guan 已提交
5450 5451
				inode->i_ctime = inode->i_mtime;
			}
5452
			down_write(&EXT4_I(inode)->i_data_sem);
5453 5454 5455 5456
			EXT4_I(inode)->i_disksize = attr->ia_size;
			rc = ext4_mark_inode_dirty(handle, inode);
			if (!error)
				error = rc;
5457 5458 5459 5460 5461 5462 5463 5464
			/*
			 * We have to update i_size under i_data_sem together
			 * with i_disksize to avoid races with writeback code
			 * running ext4_wb_update_i_disksize().
			 */
			if (!error)
				i_size_write(inode, attr->ia_size);
			up_write(&EXT4_I(inode)->i_data_sem);
5465 5466
			ext4_journal_stop(handle);
			if (error) {
5467 5468
				if (orphan)
					ext4_orphan_del(NULL, inode);
5469 5470
				goto err_out;
			}
5471
		}
5472 5473
		if (!shrink)
			pagecache_isize_extended(inode, oldsize, inode->i_size);
5474

5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486
		/*
		 * Blocks are going to be removed from the inode. Wait
		 * for dio in flight.  Temporarily disable
		 * dioread_nolock to prevent livelock.
		 */
		if (orphan) {
			if (!ext4_should_journal_data(inode)) {
				ext4_inode_block_unlocked_dio(inode);
				inode_dio_wait(inode);
				ext4_inode_resume_unlocked_dio(inode);
			} else
				ext4_wait_for_tail_page_commit(inode);
5487
		}
5488
		down_write(&EXT4_I(inode)->i_mmap_sem);
5489 5490 5491 5492
		/*
		 * Truncate pagecache after we've waited for commit
		 * in data=journal mode to make pages freeable.
		 */
R
Ross Zwisler 已提交
5493
		truncate_pagecache(inode, inode->i_size);
5494 5495 5496 5497 5498
		if (shrink) {
			rc = ext4_truncate(inode);
			if (rc)
				error = rc;
		}
5499
		up_write(&EXT4_I(inode)->i_mmap_sem);
5500
	}
5501

5502
	if (!error) {
C
Christoph Hellwig 已提交
5503 5504 5505 5506 5507 5508 5509 5510
		setattr_copy(inode, attr);
		mark_inode_dirty(inode);
	}

	/*
	 * If the call to ext4_truncate failed to get a transaction handle at
	 * all, we need to clean up the in-core orphan list manually.
	 */
5511
	if (orphan && inode->i_nlink)
5512
		ext4_orphan_del(NULL, inode);
5513

5514
	if (!error && (ia_valid & ATTR_MODE))
5515
		rc = posix_acl_chmod(inode, inode->i_mode);
5516 5517

err_out:
5518
	ext4_std_error(inode->i_sb, error);
5519 5520 5521 5522 5523
	if (!error)
		error = rc;
	return error;
}

5524 5525
int ext4_getattr(const struct path *path, struct kstat *stat,
		 u32 request_mask, unsigned int query_flags)
5526
{
D
David Howells 已提交
5527 5528 5529 5530 5531 5532 5533 5534 5535 5536 5537 5538 5539 5540 5541 5542 5543 5544 5545 5546 5547 5548
	struct inode *inode = d_inode(path->dentry);
	struct ext4_inode *raw_inode;
	struct ext4_inode_info *ei = EXT4_I(inode);
	unsigned int flags;

	if (EXT4_FITS_IN_INODE(raw_inode, ei, i_crtime)) {
		stat->result_mask |= STATX_BTIME;
		stat->btime.tv_sec = ei->i_crtime.tv_sec;
		stat->btime.tv_nsec = ei->i_crtime.tv_nsec;
	}

	flags = ei->i_flags & EXT4_FL_USER_VISIBLE;
	if (flags & EXT4_APPEND_FL)
		stat->attributes |= STATX_ATTR_APPEND;
	if (flags & EXT4_COMPR_FL)
		stat->attributes |= STATX_ATTR_COMPRESSED;
	if (flags & EXT4_ENCRYPT_FL)
		stat->attributes |= STATX_ATTR_ENCRYPTED;
	if (flags & EXT4_IMMUTABLE_FL)
		stat->attributes |= STATX_ATTR_IMMUTABLE;
	if (flags & EXT4_NODUMP_FL)
		stat->attributes |= STATX_ATTR_NODUMP;
5549

5550 5551 5552 5553 5554 5555
	stat->attributes_mask |= (STATX_ATTR_APPEND |
				  STATX_ATTR_COMPRESSED |
				  STATX_ATTR_ENCRYPTED |
				  STATX_ATTR_IMMUTABLE |
				  STATX_ATTR_NODUMP);

5556
	generic_fillattr(inode, stat);
D
David Howells 已提交
5557 5558 5559 5560 5561 5562 5563 5564 5565 5566
	return 0;
}

int ext4_file_getattr(const struct path *path, struct kstat *stat,
		      u32 request_mask, unsigned int query_flags)
{
	struct inode *inode = d_inode(path->dentry);
	u64 delalloc_blocks;

	ext4_getattr(path, stat, request_mask, query_flags);
5567

5568 5569 5570 5571
	/*
	 * If there is inline data in the inode, the inode will normally not
	 * have data blocks allocated (it may have an external xattr block).
	 * Report at least one sector for such files, so tools like tar, rsync,
T
Theodore Ts'o 已提交
5572
	 * others don't incorrectly think the file is completely sparse.
5573 5574 5575 5576
	 */
	if (unlikely(ext4_has_inline_data(inode)))
		stat->blocks += (stat->size + 511) >> 9;

5577 5578 5579 5580 5581 5582 5583 5584 5585 5586
	/*
	 * We can't update i_blocks if the block allocation is delayed
	 * otherwise in the case of system crash before the real block
	 * allocation is done, we will have i_blocks inconsistent with
	 * on-disk file blocks.
	 * We always keep i_blocks updated together with real
	 * allocation. But to not confuse with user, stat
	 * will return the blocks that include the delayed allocation
	 * blocks for this file.
	 */
5587
	delalloc_blocks = EXT4_C2B(EXT4_SB(inode->i_sb),
5588 5589
				   EXT4_I(inode)->i_reserved_data_blocks);
	stat->blocks += delalloc_blocks << (inode->i_sb->s_blocksize_bits - 9);
5590 5591
	return 0;
}
5592

5593 5594
static int ext4_index_trans_blocks(struct inode *inode, int lblocks,
				   int pextents)
5595
{
5596
	if (!(ext4_test_inode_flag(inode, EXT4_INODE_EXTENTS)))
5597 5598
		return ext4_ind_trans_blocks(inode, lblocks);
	return ext4_ext_index_trans_blocks(inode, pextents);
5599
}
5600

5601
/*
5602 5603 5604
 * Account for index blocks, block groups bitmaps and block group
 * descriptor blocks if modify datablocks and index blocks
 * worse case, the indexs blocks spread over different block groups
5605
 *
5606
 * If datablocks are discontiguous, they are possible to spread over
5607
 * different block groups too. If they are contiguous, with flexbg,
5608
 * they could still across block group boundary.
5609
 *
5610 5611
 * Also account for superblock, inode, quota and xattr blocks
 */
T
Tahsin Erdogan 已提交
5612
static int ext4_meta_trans_blocks(struct inode *inode, int lblocks,
5613
				  int pextents)
5614
{
5615 5616
	ext4_group_t groups, ngroups = ext4_get_groups_count(inode->i_sb);
	int gdpblocks;
5617 5618 5619 5620
	int idxblocks;
	int ret = 0;

	/*
5621 5622
	 * How many index blocks need to touch to map @lblocks logical blocks
	 * to @pextents physical extents?
5623
	 */
5624
	idxblocks = ext4_index_trans_blocks(inode, lblocks, pextents);
5625 5626 5627 5628 5629 5630 5631

	ret = idxblocks;

	/*
	 * Now let's see how many group bitmaps and group descriptors need
	 * to account
	 */
5632
	groups = idxblocks + pextents;
5633
	gdpblocks = groups;
5634 5635
	if (groups > ngroups)
		groups = ngroups;
5636 5637 5638 5639 5640 5641 5642 5643 5644 5645 5646 5647 5648
	if (groups > EXT4_SB(inode->i_sb)->s_gdb_count)
		gdpblocks = EXT4_SB(inode->i_sb)->s_gdb_count;

	/* bitmaps and block group descriptor blocks */
	ret += groups + gdpblocks;

	/* Blocks for super block, inode, quota and xattr blocks */
	ret += EXT4_META_TRANS_BLOCKS(inode->i_sb);

	return ret;
}

/*
L
Lucas De Marchi 已提交
5649
 * Calculate the total number of credits to reserve to fit
5650 5651
 * the modification of a single pages into a single transaction,
 * which may include multiple chunks of block allocations.
5652
 *
5653
 * This could be called via ext4_write_begin()
5654
 *
5655
 * We need to consider the worse case, when
5656
 * one new block per extent.
5657
 */
A
Alex Tomas 已提交
5658
int ext4_writepage_trans_blocks(struct inode *inode)
5659
{
5660
	int bpp = ext4_journal_blocks_per_page(inode);
5661 5662
	int ret;

5663
	ret = ext4_meta_trans_blocks(inode, bpp, bpp);
A
Alex Tomas 已提交
5664

5665
	/* Account for data blocks for journalled mode */
5666
	if (ext4_should_journal_data(inode))
5667
		ret += bpp;
5668 5669
	return ret;
}
5670 5671 5672 5673 5674

/*
 * Calculate the journal credits for a chunk of data modification.
 *
 * This is called from DIO, fallocate or whoever calling
5675
 * ext4_map_blocks() to map/allocate a chunk of contiguous disk blocks.
5676 5677 5678 5679 5680 5681 5682 5683 5684
 *
 * journal buffers for data blocks are not included here, as DIO
 * and fallocate do no need to journal data buffers.
 */
int ext4_chunk_trans_blocks(struct inode *inode, int nrblocks)
{
	return ext4_meta_trans_blocks(inode, nrblocks, 1);
}

5685
/*
5686
 * The caller must have previously called ext4_reserve_inode_write().
5687 5688
 * Give this, we know that the caller already has write access to iloc->bh.
 */
5689
int ext4_mark_iloc_dirty(handle_t *handle,
5690
			 struct inode *inode, struct ext4_iloc *iloc)
5691 5692 5693
{
	int err = 0;

5694 5695 5696
	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
		return -EIO;

5697
	if (IS_I_VERSION(inode))
5698 5699
		inode_inc_iversion(inode);

5700 5701 5702
	/* the do_update_inode consumes one bh->b_count */
	get_bh(iloc->bh);

5703
	/* ext4_do_update_inode() does jbd2_journal_dirty_metadata */
5704
	err = ext4_do_update_inode(handle, inode, iloc);
5705 5706 5707 5708 5709 5710 5711 5712 5713 5714
	put_bh(iloc->bh);
	return err;
}

/*
 * On success, We end up with an outstanding reference count against
 * iloc->bh.  This _must_ be cleaned up later.
 */

int
5715 5716
ext4_reserve_inode_write(handle_t *handle, struct inode *inode,
			 struct ext4_iloc *iloc)
5717
{
5718 5719
	int err;

5720 5721 5722
	if (unlikely(ext4_forced_shutdown(EXT4_SB(inode->i_sb))))
		return -EIO;

5723 5724 5725 5726 5727 5728 5729
	err = ext4_get_inode_loc(inode, iloc);
	if (!err) {
		BUFFER_TRACE(iloc->bh, "get_write_access");
		err = ext4_journal_get_write_access(handle, iloc->bh);
		if (err) {
			brelse(iloc->bh);
			iloc->bh = NULL;
5730 5731
		}
	}
5732
	ext4_std_error(inode->i_sb, err);
5733 5734 5735
	return err;
}

5736 5737 5738 5739 5740 5741 5742 5743 5744 5745 5746 5747 5748 5749 5750 5751 5752 5753 5754 5755 5756 5757 5758 5759 5760 5761 5762 5763 5764 5765 5766 5767 5768 5769 5770 5771
static int __ext4_expand_extra_isize(struct inode *inode,
				     unsigned int new_extra_isize,
				     struct ext4_iloc *iloc,
				     handle_t *handle, int *no_expand)
{
	struct ext4_inode *raw_inode;
	struct ext4_xattr_ibody_header *header;
	int error;

	raw_inode = ext4_raw_inode(iloc);

	header = IHDR(inode, raw_inode);

	/* No extended attributes present */
	if (!ext4_test_inode_state(inode, EXT4_STATE_XATTR) ||
	    header->h_magic != cpu_to_le32(EXT4_XATTR_MAGIC)) {
		memset((void *)raw_inode + EXT4_GOOD_OLD_INODE_SIZE +
		       EXT4_I(inode)->i_extra_isize, 0,
		       new_extra_isize - EXT4_I(inode)->i_extra_isize);
		EXT4_I(inode)->i_extra_isize = new_extra_isize;
		return 0;
	}

	/* try to expand with EAs present */
	error = ext4_expand_extra_isize_ea(inode, new_extra_isize,
					   raw_inode, handle);
	if (error) {
		/*
		 * Inode size expansion failed; don't try again
		 */
		*no_expand = 1;
	}

	return error;
}

5772 5773 5774 5775
/*
 * Expand an inode by new_extra_isize bytes.
 * Returns 0 on success or negative error number on failure.
 */
5776 5777 5778 5779
static int ext4_try_to_expand_extra_isize(struct inode *inode,
					  unsigned int new_extra_isize,
					  struct ext4_iloc iloc,
					  handle_t *handle)
5780
{
5781 5782
	int no_expand;
	int error;
5783

5784 5785 5786 5787 5788 5789 5790 5791 5792 5793 5794 5795 5796 5797 5798 5799
	if (ext4_test_inode_state(inode, EXT4_STATE_NO_EXPAND))
		return -EOVERFLOW;

	/*
	 * In nojournal mode, we can immediately attempt to expand
	 * the inode.  When journaled, we first need to obtain extra
	 * buffer credits since we may write into the EA block
	 * with this same handle. If journal_extend fails, then it will
	 * only result in a minor loss of functionality for that inode.
	 * If this is felt to be critical, then e2fsck should be run to
	 * force a large enough s_min_extra_isize.
	 */
	if (ext4_handle_valid(handle) &&
	    jbd2_journal_extend(handle,
				EXT4_DATA_TRANS_BLOCKS(inode->i_sb)) != 0)
		return -ENOSPC;
5800

5801
	if (ext4_write_trylock_xattr(inode, &no_expand) == 0)
5802
		return -EBUSY;
5803

5804 5805 5806
	error = __ext4_expand_extra_isize(inode, new_extra_isize, &iloc,
					  handle, &no_expand);
	ext4_write_unlock_xattr(inode, &no_expand);
5807

5808 5809
	return error;
}
5810

5811 5812 5813 5814 5815 5816 5817 5818 5819 5820 5821
int ext4_expand_extra_isize(struct inode *inode,
			    unsigned int new_extra_isize,
			    struct ext4_iloc *iloc)
{
	handle_t *handle;
	int no_expand;
	int error, rc;

	if (ext4_test_inode_state(inode, EXT4_STATE_NO_EXPAND)) {
		brelse(iloc->bh);
		return -EOVERFLOW;
5822 5823
	}

5824 5825 5826 5827 5828 5829 5830 5831 5832 5833 5834 5835
	handle = ext4_journal_start(inode, EXT4_HT_INODE,
				    EXT4_DATA_TRANS_BLOCKS(inode->i_sb));
	if (IS_ERR(handle)) {
		error = PTR_ERR(handle);
		brelse(iloc->bh);
		return error;
	}

	ext4_write_lock_xattr(inode, &no_expand);

	BUFFER_TRACE(iloc.bh, "get_write_access");
	error = ext4_journal_get_write_access(handle, iloc->bh);
5836
	if (error) {
5837 5838
		brelse(iloc->bh);
		goto out_stop;
5839
	}
5840

5841 5842 5843 5844 5845 5846 5847 5848 5849 5850
	error = __ext4_expand_extra_isize(inode, new_extra_isize, iloc,
					  handle, &no_expand);

	rc = ext4_mark_iloc_dirty(handle, inode, iloc);
	if (!error)
		error = rc;

	ext4_write_unlock_xattr(inode, &no_expand);
out_stop:
	ext4_journal_stop(handle);
5851
	return error;
5852 5853
}

5854 5855 5856 5857 5858 5859 5860 5861 5862 5863 5864 5865 5866
/*
 * What we do here is to mark the in-core inode as clean with respect to inode
 * dirtiness (it may still be data-dirty).
 * This means that the in-core inode may be reaped by prune_icache
 * without having to perform any I/O.  This is a very good thing,
 * because *any* task may call prune_icache - even ones which
 * have a transaction open against a different journal.
 *
 * Is this cheating?  Not really.  Sure, we haven't written the
 * inode out, but prune_icache isn't a user-visible syncing function.
 * Whenever the user wants stuff synced (sys_sync, sys_msync, sys_fsync)
 * we start and wait on commits.
 */
5867
int ext4_mark_inode_dirty(handle_t *handle, struct inode *inode)
5868
{
5869
	struct ext4_iloc iloc;
5870
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
5871
	int err;
5872 5873

	might_sleep();
5874
	trace_ext4_mark_inode_dirty(inode, _RET_IP_);
5875
	err = ext4_reserve_inode_write(handle, inode, &iloc);
5876 5877
	if (err)
		return err;
5878 5879 5880 5881 5882

	if (EXT4_I(inode)->i_extra_isize < sbi->s_want_extra_isize)
		ext4_try_to_expand_extra_isize(inode, sbi->s_want_extra_isize,
					       iloc, handle);

5883
	return ext4_mark_iloc_dirty(handle, inode, &iloc);
5884 5885 5886
}

/*
5887
 * ext4_dirty_inode() is called from __mark_inode_dirty()
5888 5889 5890 5891 5892
 *
 * We're really interested in the case where a file is being extended.
 * i_size has been changed by generic_commit_write() and we thus need
 * to include the updated inode in the current transaction.
 *
5893
 * Also, dquot_alloc_block() will always dirty the inode when blocks
5894 5895 5896 5897 5898
 * are allocated to the file.
 *
 * If the inode is marked synchronous, we don't honour that here - doing
 * so would cause a commit on atime updates, which we don't bother doing.
 * We handle synchronous inodes at the highest possible level.
5899 5900 5901 5902
 *
 * If only the I_DIRTY_TIME flag is set, we can skip everything.  If
 * I_DIRTY_TIME and I_DIRTY_SYNC is set, the only inode fields we need
 * to copy into the on-disk inode structure are the timestamp files.
5903
 */
5904
void ext4_dirty_inode(struct inode *inode, int flags)
5905 5906 5907
{
	handle_t *handle;

5908 5909
	if (flags == I_DIRTY_TIME)
		return;
5910
	handle = ext4_journal_start(inode, EXT4_HT_INODE, 2);
5911 5912
	if (IS_ERR(handle))
		goto out;
5913 5914 5915

	ext4_mark_inode_dirty(handle, inode);

5916
	ext4_journal_stop(handle);
5917 5918 5919 5920 5921 5922 5923 5924
out:
	return;
}

#if 0
/*
 * Bind an inode's backing buffer_head into this transaction, to prevent
 * it from being flushed to disk early.  Unlike
5925
 * ext4_reserve_inode_write, this leaves behind no bh reference and
5926 5927 5928
 * returns no iloc structure, so the caller needs to repeat the iloc
 * lookup to mark the inode dirty later.
 */
5929
static int ext4_pin_inode(handle_t *handle, struct inode *inode)
5930
{
5931
	struct ext4_iloc iloc;
5932 5933 5934

	int err = 0;
	if (handle) {
5935
		err = ext4_get_inode_loc(inode, &iloc);
5936 5937
		if (!err) {
			BUFFER_TRACE(iloc.bh, "get_write_access");
5938
			err = jbd2_journal_get_write_access(handle, iloc.bh);
5939
			if (!err)
5940
				err = ext4_handle_dirty_metadata(handle,
5941
								 NULL,
5942
								 iloc.bh);
5943 5944 5945
			brelse(iloc.bh);
		}
	}
5946
	ext4_std_error(inode->i_sb, err);
5947 5948 5949 5950
	return err;
}
#endif

5951
int ext4_change_inode_journal_flag(struct inode *inode, int val)
5952 5953 5954 5955
{
	journal_t *journal;
	handle_t *handle;
	int err;
5956
	struct ext4_sb_info *sbi = EXT4_SB(inode->i_sb);
5957 5958 5959 5960 5961 5962 5963 5964 5965 5966 5967

	/*
	 * We have to be very careful here: changing a data block's
	 * journaling status dynamically is dangerous.  If we write a
	 * data block to the journal, change the status and then delete
	 * that block, we risk forgetting to revoke the old log record
	 * from the journal and so a subsequent replay can corrupt data.
	 * So, first we make sure that the journal is empty and that
	 * nobody is changing anything.
	 */

5968
	journal = EXT4_JOURNAL(inode);
5969 5970
	if (!journal)
		return 0;
5971
	if (is_journal_aborted(journal))
5972 5973
		return -EROFS;

5974 5975 5976 5977
	/* Wait for all existing dio workers */
	ext4_inode_block_unlocked_dio(inode);
	inode_dio_wait(inode);

5978 5979 5980 5981 5982 5983 5984 5985 5986 5987 5988 5989 5990 5991 5992 5993 5994 5995
	/*
	 * Before flushing the journal and switching inode's aops, we have
	 * to flush all dirty data the inode has. There can be outstanding
	 * delayed allocations, there can be unwritten extents created by
	 * fallocate or buffered writes in dioread_nolock mode covered by
	 * dirty data which can be converted only after flushing the dirty
	 * data (and journalled aops don't know how to handle these cases).
	 */
	if (val) {
		down_write(&EXT4_I(inode)->i_mmap_sem);
		err = filemap_write_and_wait(inode->i_mapping);
		if (err < 0) {
			up_write(&EXT4_I(inode)->i_mmap_sem);
			ext4_inode_resume_unlocked_dio(inode);
			return err;
		}
	}

5996
	percpu_down_write(&sbi->s_journal_flag_rwsem);
5997
	jbd2_journal_lock_updates(journal);
5998 5999 6000 6001 6002 6003 6004 6005 6006 6007

	/*
	 * OK, there are no updates running now, and all cached data is
	 * synced to disk.  We are now in a completely consistent state
	 * which doesn't have anything in the journal, and we know that
	 * no filesystem updates are running, so it is safe to modify
	 * the inode's in-core data-journaling state flag now.
	 */

	if (val)
6008
		ext4_set_inode_flag(inode, EXT4_INODE_JOURNAL_DATA);
6009
	else {
6010 6011 6012
		err = jbd2_journal_flush(journal);
		if (err < 0) {
			jbd2_journal_unlock_updates(journal);
6013
			percpu_up_write(&sbi->s_journal_flag_rwsem);
6014 6015 6016
			ext4_inode_resume_unlocked_dio(inode);
			return err;
		}
6017
		ext4_clear_inode_flag(inode, EXT4_INODE_JOURNAL_DATA);
6018
	}
6019
	ext4_set_aops(inode);
6020

6021
	jbd2_journal_unlock_updates(journal);
6022 6023
	percpu_up_write(&sbi->s_journal_flag_rwsem);

6024 6025
	if (val)
		up_write(&EXT4_I(inode)->i_mmap_sem);
6026
	ext4_inode_resume_unlocked_dio(inode);
6027 6028 6029

	/* Finally we can mark the inode as dirty. */

6030
	handle = ext4_journal_start(inode, EXT4_HT_INODE, 1);
6031 6032 6033
	if (IS_ERR(handle))
		return PTR_ERR(handle);

6034
	err = ext4_mark_inode_dirty(handle, inode);
6035
	ext4_handle_sync(handle);
6036 6037
	ext4_journal_stop(handle);
	ext4_std_error(inode->i_sb, err);
6038 6039 6040

	return err;
}
6041 6042 6043 6044 6045 6046

static int ext4_bh_unmapped(handle_t *handle, struct buffer_head *bh)
{
	return !buffer_mapped(bh);
}

6047
int ext4_page_mkwrite(struct vm_fault *vmf)
6048
{
6049
	struct vm_area_struct *vma = vmf->vma;
6050
	struct page *page = vmf->page;
6051 6052
	loff_t size;
	unsigned long len;
6053
	int ret;
6054
	struct file *file = vma->vm_file;
A
Al Viro 已提交
6055
	struct inode *inode = file_inode(file);
6056
	struct address_space *mapping = inode->i_mapping;
6057 6058 6059
	handle_t *handle;
	get_block_t *get_block;
	int retries = 0;
6060

6061
	sb_start_pagefault(inode->i_sb);
6062
	file_update_time(vma->vm_file);
6063 6064

	down_read(&EXT4_I(inode)->i_mmap_sem);
6065 6066 6067 6068 6069

	ret = ext4_convert_inline_data(inode);
	if (ret)
		goto out_ret;

6070 6071 6072 6073 6074
	/* Delalloc case is easy... */
	if (test_opt(inode->i_sb, DELALLOC) &&
	    !ext4_should_journal_data(inode) &&
	    !ext4_nonda_switch(inode->i_sb)) {
		do {
6075
			ret = block_page_mkwrite(vma, vmf,
6076 6077 6078 6079
						   ext4_da_get_block_prep);
		} while (ret == -ENOSPC &&
		       ext4_should_retry_alloc(inode->i_sb, &retries));
		goto out_ret;
6080
	}
6081 6082

	lock_page(page);
6083 6084 6085 6086 6087 6088
	size = i_size_read(inode);
	/* Page got truncated from under us? */
	if (page->mapping != mapping || page_offset(page) > size) {
		unlock_page(page);
		ret = VM_FAULT_NOPAGE;
		goto out;
6089
	}
6090

6091 6092
	if (page->index == size >> PAGE_SHIFT)
		len = size & ~PAGE_MASK;
6093
	else
6094
		len = PAGE_SIZE;
6095
	/*
6096 6097
	 * Return if we have all the buffers mapped. This avoids the need to do
	 * journal_start/journal_stop which can block and take a long time
6098
	 */
6099
	if (page_has_buffers(page)) {
6100 6101 6102
		if (!ext4_walk_page_buffers(NULL, page_buffers(page),
					    0, len, NULL,
					    ext4_bh_unmapped)) {
6103
			/* Wait so that we don't change page under IO */
6104
			wait_for_stable_page(page);
6105 6106
			ret = VM_FAULT_LOCKED;
			goto out;
6107
		}
6108
	}
6109
	unlock_page(page);
6110 6111
	/* OK, we need to fill the hole... */
	if (ext4_should_dioread_nolock(inode))
6112
		get_block = ext4_get_block_unwritten;
6113 6114 6115
	else
		get_block = ext4_get_block;
retry_alloc:
6116 6117
	handle = ext4_journal_start(inode, EXT4_HT_WRITE_PAGE,
				    ext4_writepage_trans_blocks(inode));
6118
	if (IS_ERR(handle)) {
6119
		ret = VM_FAULT_SIGBUS;
6120 6121
		goto out;
	}
6122
	ret = block_page_mkwrite(vma, vmf, get_block);
6123
	if (!ret && ext4_should_journal_data(inode)) {
6124
		if (ext4_walk_page_buffers(handle, page_buffers(page), 0,
6125
			  PAGE_SIZE, NULL, do_journal_get_write_access)) {
6126 6127
			unlock_page(page);
			ret = VM_FAULT_SIGBUS;
6128
			ext4_journal_stop(handle);
6129 6130 6131 6132 6133 6134 6135 6136 6137 6138
			goto out;
		}
		ext4_set_inode_state(inode, EXT4_STATE_JDATA);
	}
	ext4_journal_stop(handle);
	if (ret == -ENOSPC && ext4_should_retry_alloc(inode->i_sb, &retries))
		goto retry_alloc;
out_ret:
	ret = block_page_mkwrite_return(ret);
out:
6139
	up_read(&EXT4_I(inode)->i_mmap_sem);
6140
	sb_end_pagefault(inode->i_sb);
6141 6142
	return ret;
}
6143

6144
int ext4_filemap_fault(struct vm_fault *vmf)
6145
{
6146
	struct inode *inode = file_inode(vmf->vma->vm_file);
6147 6148 6149
	int err;

	down_read(&EXT4_I(inode)->i_mmap_sem);
6150
	err = filemap_fault(vmf);
6151 6152 6153 6154
	up_read(&EXT4_I(inode)->i_mmap_sem);

	return err;
}