file.c 71.9 KB
Newer Older
J
Jaegeuk Kim 已提交
1
/*
J
Jaegeuk Kim 已提交
2 3 4 5 6 7 8 9 10 11 12 13 14 15
 * fs/f2fs/file.c
 *
 * Copyright (c) 2012 Samsung Electronics Co., Ltd.
 *             http://www.samsung.com/
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */
#include <linux/fs.h>
#include <linux/f2fs_fs.h>
#include <linux/stat.h>
#include <linux/buffer_head.h>
#include <linux/writeback.h>
16
#include <linux/blkdev.h>
J
Jaegeuk Kim 已提交
17 18
#include <linux/falloc.h>
#include <linux/types.h>
19
#include <linux/compat.h>
J
Jaegeuk Kim 已提交
20 21
#include <linux/uaccess.h>
#include <linux/mount.h>
22
#include <linux/pagevec.h>
23
#include <linux/uio.h>
24
#include <linux/uuid.h>
25
#include <linux/file.h>
J
Jaegeuk Kim 已提交
26 27 28 29 30 31

#include "f2fs.h"
#include "node.h"
#include "segment.h"
#include "xattr.h"
#include "acl.h"
32
#include "gc.h"
J
Jaegeuk Kim 已提交
33
#include "trace.h"
34
#include <trace/events/f2fs.h>
J
Jaegeuk Kim 已提交
35

36
static vm_fault_t f2fs_filemap_fault(struct vm_fault *vmf)
37 38
{
	struct inode *inode = file_inode(vmf->vma->vm_file);
39
	vm_fault_t ret;
40 41

	down_read(&F2FS_I(inode)->i_mmap_sem);
42
	ret = filemap_fault(vmf);
43 44
	up_read(&F2FS_I(inode)->i_mmap_sem);

45
	return ret;
46 47
}

48
static vm_fault_t f2fs_vm_page_mkwrite(struct vm_fault *vmf)
J
Jaegeuk Kim 已提交
49 50
{
	struct page *page = vmf->page;
51
	struct inode *inode = file_inode(vmf->vma->vm_file);
52
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
J
Jaegeuk Kim 已提交
53
	struct dnode_of_data dn;
54
	int err;
J
Jaegeuk Kim 已提交
55

56 57 58 59 60
	if (unlikely(f2fs_cp_error(sbi))) {
		err = -EIO;
		goto err;
	}

J
Jaegeuk Kim 已提交
61
	sb_start_pagefault(inode->i_sb);
62 63

	f2fs_bug_on(sbi, f2fs_has_inline_data(inode));
64

J
Jaegeuk Kim 已提交
65
	/* block allocation */
66
	f2fs_lock_op(sbi);
67
	set_new_dnode(&dn, inode, NULL, NULL, 0);
68
	err = f2fs_reserve_block(&dn, page->index);
69 70
	if (err) {
		f2fs_unlock_op(sbi);
71
		goto out;
72 73 74
	}
	f2fs_put_dnode(&dn);
	f2fs_unlock_op(sbi);
J
Jaegeuk Kim 已提交
75

J
Jaegeuk Kim 已提交
76
	f2fs_balance_fs(sbi, dn.node_changed);
77

78
	file_update_time(vmf->vma->vm_file);
79
	down_read(&F2FS_I(inode)->i_mmap_sem);
J
Jaegeuk Kim 已提交
80
	lock_page(page);
81
	if (unlikely(page->mapping != inode->i_mapping ||
82
			page_offset(page) > i_size_read(inode) ||
83
			!PageUptodate(page))) {
J
Jaegeuk Kim 已提交
84 85
		unlock_page(page);
		err = -EFAULT;
86
		goto out_sem;
J
Jaegeuk Kim 已提交
87 88 89 90 91 92
	}

	/*
	 * check to see if the page is mapped already (no holes)
	 */
	if (PageMappedToDisk(page))
93
		goto mapped;
J
Jaegeuk Kim 已提交
94 95

	/* page is wholly or partially inside EOF */
96
	if (((loff_t)(page->index + 1) << PAGE_SHIFT) >
C
Chao Yu 已提交
97
						i_size_read(inode)) {
98
		loff_t offset;
99

100 101
		offset = i_size_read(inode) & ~PAGE_MASK;
		zero_user_segment(page, offset, PAGE_SIZE);
J
Jaegeuk Kim 已提交
102 103
	}
	set_page_dirty(page);
104 105
	if (!PageUptodate(page))
		SetPageUptodate(page);
J
Jaegeuk Kim 已提交
106

C
Chao Yu 已提交
107 108
	f2fs_update_iostat(sbi, APP_MAPPED_IO, F2FS_BLKSIZE);

109
	trace_f2fs_vm_page_mkwrite(page, DATA);
110 111
mapped:
	/* fill the page */
112
	f2fs_wait_on_page_writeback(page, DATA, false);
113

114 115
	/* wait for GCed page writeback via META_MAPPING */
	if (f2fs_post_read_required(inode))
116
		f2fs_wait_on_block_writeback(sbi, dn.data_blkaddr);
117

118 119
out_sem:
	up_read(&F2FS_I(inode)->i_mmap_sem);
J
Jaegeuk Kim 已提交
120 121
out:
	sb_end_pagefault(inode->i_sb);
122
	f2fs_update_time(sbi, REQ_TIME);
123
err:
J
Jaegeuk Kim 已提交
124 125 126 127
	return block_page_mkwrite_return(err);
}

static const struct vm_operations_struct f2fs_file_vm_ops = {
128
	.fault		= f2fs_filemap_fault,
129
	.map_pages	= filemap_map_pages,
130
	.page_mkwrite	= f2fs_vm_page_mkwrite,
J
Jaegeuk Kim 已提交
131 132
};

133 134 135 136 137 138 139 140 141 142
static int get_parent_ino(struct inode *inode, nid_t *pino)
{
	struct dentry *dentry;

	inode = igrab(inode);
	dentry = d_find_any_alias(inode);
	iput(inode);
	if (!dentry)
		return 0;

143 144
	*pino = parent_ino(dentry);
	dput(dentry);
145 146 147
	return 1;
}

C
Chao Yu 已提交
148
static inline enum cp_reason_type need_do_checkpoint(struct inode *inode)
149
{
150
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
C
Chao Yu 已提交
151
	enum cp_reason_type cp_reason = CP_NO_NEEDED;
152

C
Chao Yu 已提交
153 154 155 156
	if (!S_ISREG(inode->i_mode))
		cp_reason = CP_NON_REGULAR;
	else if (inode->i_nlink != 1)
		cp_reason = CP_HARDLINK;
157
	else if (is_sbi_flag_set(sbi, SBI_NEED_CP))
C
Chao Yu 已提交
158
		cp_reason = CP_SB_NEED_CP;
159
	else if (file_wrong_pino(inode))
C
Chao Yu 已提交
160
		cp_reason = CP_WRONG_PINO;
C
Chao Yu 已提交
161
	else if (!f2fs_space_for_roll_forward(sbi))
C
Chao Yu 已提交
162
		cp_reason = CP_NO_SPC_ROLL;
C
Chao Yu 已提交
163
	else if (!f2fs_is_checkpointed_node(sbi, F2FS_I(inode)->i_pino))
C
Chao Yu 已提交
164
		cp_reason = CP_NODE_NEED_CP;
165
	else if (test_opt(sbi, FASTBOOT))
C
Chao Yu 已提交
166
		cp_reason = CP_FASTBOOT_MODE;
167
	else if (F2FS_OPTION(sbi).active_logs == 2)
C
Chao Yu 已提交
168
		cp_reason = CP_SPEC_LOG_NUM;
169
	else if (F2FS_OPTION(sbi).fsync_mode == FSYNC_MODE_STRICT &&
C
Chao Yu 已提交
170 171 172
		f2fs_need_dentry_mark(sbi, inode->i_ino) &&
		f2fs_exist_written_data(sbi, F2FS_I(inode)->i_pino,
							TRANS_DIR_INO))
173
		cp_reason = CP_RECOVER_DIR;
174

C
Chao Yu 已提交
175
	return cp_reason;
176 177
}

178 179 180 181 182
static bool need_inode_page_update(struct f2fs_sb_info *sbi, nid_t ino)
{
	struct page *i = find_get_page(NODE_MAPPING(sbi), ino);
	bool ret = false;
	/* But we need to avoid that there are some inode updates */
C
Chao Yu 已提交
183
	if ((i && PageDirty(i)) || f2fs_need_inode_block_update(sbi, ino))
184 185 186 187 188
		ret = true;
	f2fs_put_page(i, 0);
	return ret;
}

189 190 191 192 193 194 195 196
static void try_to_fix_pino(struct inode *inode)
{
	struct f2fs_inode_info *fi = F2FS_I(inode);
	nid_t pino;

	down_write(&fi->i_sem);
	if (file_wrong_pino(inode) && inode->i_nlink == 1 &&
			get_parent_ino(inode, &pino)) {
197
		f2fs_i_pino_write(inode, pino);
198 199
		file_got_pino(inode);
	}
200
	up_write(&fi->i_sem);
201 202
}

203 204
static int f2fs_do_sync_file(struct file *file, loff_t start, loff_t end,
						int datasync, bool atomic)
J
Jaegeuk Kim 已提交
205 206
{
	struct inode *inode = file->f_mapping->host;
207
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
J
Jaegeuk Kim 已提交
208
	nid_t ino = inode->i_ino;
J
Jaegeuk Kim 已提交
209
	int ret = 0;
C
Chao Yu 已提交
210
	enum cp_reason_type cp_reason = 0;
J
Jaegeuk Kim 已提交
211
	struct writeback_control wbc = {
212
		.sync_mode = WB_SYNC_ALL,
J
Jaegeuk Kim 已提交
213 214 215
		.nr_to_write = LONG_MAX,
		.for_reclaim = 0,
	};
216
	unsigned int seq_id = 0;
J
Jaegeuk Kim 已提交
217

218
	if (unlikely(f2fs_readonly(inode->i_sb)))
219 220
		return 0;

221
	trace_f2fs_sync_file_enter(inode);
222 223

	/* if fdatasync is triggered, let's do in-place-update */
J
Jaegeuk Kim 已提交
224
	if (datasync || get_dirty_pages(inode) <= SM_I(sbi)->min_fsync_blocks)
225
		set_inode_flag(inode, FI_NEED_IPU);
226
	ret = file_write_and_wait_range(file, start, end);
227
	clear_inode_flag(inode, FI_NEED_IPU);
228

229
	if (ret) {
C
Chao Yu 已提交
230
		trace_f2fs_sync_file_exit(inode, cp_reason, datasync, ret);
J
Jaegeuk Kim 已提交
231
		return ret;
232
	}
J
Jaegeuk Kim 已提交
233

234
	/* if the inode is dirty, let's recover all the time */
C
Chao Yu 已提交
235
	if (!f2fs_skip_inode_update(inode, datasync)) {
236
		f2fs_write_inode(inode, NULL);
237 238 239
		goto go_write;
	}

240 241 242
	/*
	 * if there is no written data, don't waste time to write recovery info.
	 */
243
	if (!is_inode_flag_set(inode, FI_APPEND_WRITE) &&
C
Chao Yu 已提交
244
			!f2fs_exist_written_data(sbi, ino, APPEND_INO)) {
245

246 247
		/* it may call write_inode just prior to fsync */
		if (need_inode_page_update(sbi, ino))
248 249
			goto go_write;

250
		if (is_inode_flag_set(inode, FI_UPDATE_WRITE) ||
C
Chao Yu 已提交
251
				f2fs_exist_written_data(sbi, ino, UPDATE_INO))
252 253 254
			goto flush_out;
		goto out;
	}
255
go_write:
256 257 258 259
	/*
	 * Both of fdatasync() and fsync() are able to be recovered from
	 * sudden-power-off.
	 */
260
	down_read(&F2FS_I(inode)->i_sem);
C
Chao Yu 已提交
261
	cp_reason = need_do_checkpoint(inode);
262
	up_read(&F2FS_I(inode)->i_sem);
263

C
Chao Yu 已提交
264
	if (cp_reason) {
J
Jaegeuk Kim 已提交
265 266
		/* all the dirty node pages should be flushed for POR */
		ret = f2fs_sync_fs(inode->i_sb, 1);
267

268 269 270 271 272
		/*
		 * We've secured consistency through sync_fs. Following pino
		 * will be used only for fsynced inodes after checkpoint.
		 */
		try_to_fix_pino(inode);
273 274
		clear_inode_flag(inode, FI_APPEND_WRITE);
		clear_inode_flag(inode, FI_UPDATE_WRITE);
275 276
		goto out;
	}
277
sync_nodes:
278
	atomic_inc(&sbi->wb_sync_req[NODE]);
279
	ret = f2fs_fsync_node_pages(sbi, inode, &wbc, atomic, &seq_id);
280
	atomic_dec(&sbi->wb_sync_req[NODE]);
281 282
	if (ret)
		goto out;
283

284
	/* if cp_error was enabled, we should avoid infinite loop */
C
Chao Yu 已提交
285 286
	if (unlikely(f2fs_cp_error(sbi))) {
		ret = -EIO;
287
		goto out;
C
Chao Yu 已提交
288
	}
289

C
Chao Yu 已提交
290
	if (f2fs_need_inode_block_update(sbi, ino)) {
291
		f2fs_mark_inode_dirty_sync(inode, true);
292 293
		f2fs_write_inode(inode, NULL);
		goto sync_nodes;
J
Jaegeuk Kim 已提交
294
	}
295

296 297 298 299 300 301 302 303 304
	/*
	 * If it's atomic_write, it's just fine to keep write ordering. So
	 * here we don't need to wait for node write completion, since we use
	 * node chain which serializes node blocks. If one of node writes are
	 * reordered, we can see simply broken chain, resulting in stopping
	 * roll-forward recovery. It means we'll recover all or none node blocks
	 * given fsync mark.
	 */
	if (!atomic) {
305
		ret = f2fs_wait_on_node_pages_writeback(sbi, seq_id);
306 307 308
		if (ret)
			goto out;
	}
309 310

	/* once recovery info is written, don't need to tack this */
C
Chao Yu 已提交
311
	f2fs_remove_ino_entry(sbi, ino, APPEND_INO);
312
	clear_inode_flag(inode, FI_APPEND_WRITE);
313
flush_out:
314
	if (!atomic && F2FS_OPTION(sbi).fsync_mode != FSYNC_MODE_NOBARRIER)
C
Chao Yu 已提交
315
		ret = f2fs_issue_flush(sbi, inode->i_ino);
316
	if (!ret) {
C
Chao Yu 已提交
317
		f2fs_remove_ino_entry(sbi, ino, UPDATE_INO);
318
		clear_inode_flag(inode, FI_UPDATE_WRITE);
C
Chao Yu 已提交
319
		f2fs_remove_ino_entry(sbi, ino, FLUSH_INO);
320
	}
321
	f2fs_update_time(sbi, REQ_TIME);
J
Jaegeuk Kim 已提交
322
out:
C
Chao Yu 已提交
323
	trace_f2fs_sync_file_exit(inode, cp_reason, datasync, ret);
324
	f2fs_trace_ios(NULL, 1);
J
Jaegeuk Kim 已提交
325 326 327
	return ret;
}

328 329
int f2fs_sync_file(struct file *file, loff_t start, loff_t end, int datasync)
{
330 331
	if (unlikely(f2fs_cp_error(F2FS_I_SB(file_inode(file)))))
		return -EIO;
332 333 334
	return f2fs_do_sync_file(file, start, end, datasync, false);
}

335 336 337
static pgoff_t __get_first_dirty_index(struct address_space *mapping,
						pgoff_t pgofs, int whence)
{
338
	struct page *page;
339 340 341 342 343 344
	int nr_pages;

	if (whence != SEEK_DATA)
		return 0;

	/* find first dirty page index */
345 346 347 348 349 350
	nr_pages = find_get_pages_tag(mapping, &pgofs, PAGECACHE_TAG_DIRTY,
				      1, &page);
	if (!nr_pages)
		return ULONG_MAX;
	pgofs = page->index;
	put_page(page);
351 352 353
	return pgofs;
}

354 355
static bool __found_offset(struct f2fs_sb_info *sbi, block_t blkaddr,
				pgoff_t dirty, pgoff_t pgofs, int whence)
356 357 358 359
{
	switch (whence) {
	case SEEK_DATA:
		if ((blkaddr == NEW_ADDR && dirty == pgofs) ||
360
			is_valid_data_blkaddr(sbi, blkaddr))
361 362 363 364 365 366 367 368 369 370
			return true;
		break;
	case SEEK_HOLE:
		if (blkaddr == NULL_ADDR)
			return true;
		break;
	}
	return false;
}

371 372 373 374 375
static loff_t f2fs_seek_block(struct file *file, loff_t offset, int whence)
{
	struct inode *inode = file->f_mapping->host;
	loff_t maxbytes = inode->i_sb->s_maxbytes;
	struct dnode_of_data dn;
376 377 378
	pgoff_t pgofs, end_offset, dirty;
	loff_t data_ofs = offset;
	loff_t isize;
379 380
	int err = 0;

A
Al Viro 已提交
381
	inode_lock(inode);
382 383 384 385 386 387

	isize = i_size_read(inode);
	if (offset >= isize)
		goto fail;

	/* handle inline data case */
C
Chao Yu 已提交
388
	if (f2fs_has_inline_data(inode) || f2fs_has_inline_dentry(inode)) {
389 390 391 392 393
		if (whence == SEEK_HOLE)
			data_ofs = isize;
		goto found;
	}

394
	pgofs = (pgoff_t)(offset >> PAGE_SHIFT);
395

396 397
	dirty = __get_first_dirty_index(inode->i_mapping, pgofs, whence);

398
	for (; data_ofs < isize; data_ofs = (loff_t)pgofs << PAGE_SHIFT) {
399
		set_new_dnode(&dn, inode, NULL, NULL, 0);
C
Chao Yu 已提交
400
		err = f2fs_get_dnode_of_data(&dn, pgofs, LOOKUP_NODE);
401 402 403
		if (err && err != -ENOENT) {
			goto fail;
		} else if (err == -ENOENT) {
A
arter97 已提交
404
			/* direct node does not exists */
405
			if (whence == SEEK_DATA) {
C
Chao Yu 已提交
406
				pgofs = f2fs_get_next_page_offset(&dn, pgofs);
407 408 409 410 411 412
				continue;
			} else {
				goto found;
			}
		}

413
		end_offset = ADDRS_PER_PAGE(dn.node_page, inode);
414 415 416 417

		/* find data/hole in dnode block */
		for (; dn.ofs_in_node < end_offset;
				dn.ofs_in_node++, pgofs++,
418
				data_ofs = (loff_t)pgofs << PAGE_SHIFT) {
419
			block_t blkaddr;
420

421 422
			blkaddr = datablock_addr(dn.inode,
					dn.node_page, dn.ofs_in_node);
423

424 425 426 427 428 429 430
			if (__is_valid_data_blkaddr(blkaddr) &&
				!f2fs_is_valid_blkaddr(F2FS_I_SB(inode),
						blkaddr, DATA_GENERIC)) {
				f2fs_put_dnode(&dn);
				goto fail;
			}

431 432
			if (__found_offset(F2FS_I_SB(inode), blkaddr, dirty,
							pgofs, whence)) {
433 434 435 436 437 438 439 440 441 442
				f2fs_put_dnode(&dn);
				goto found;
			}
		}
		f2fs_put_dnode(&dn);
	}

	if (whence == SEEK_DATA)
		goto fail;
found:
443 444
	if (whence == SEEK_HOLE && data_ofs > isize)
		data_ofs = isize;
A
Al Viro 已提交
445
	inode_unlock(inode);
446 447
	return vfs_setpos(file, data_ofs, maxbytes);
fail:
A
Al Viro 已提交
448
	inode_unlock(inode);
449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464
	return -ENXIO;
}

static loff_t f2fs_llseek(struct file *file, loff_t offset, int whence)
{
	struct inode *inode = file->f_mapping->host;
	loff_t maxbytes = inode->i_sb->s_maxbytes;

	switch (whence) {
	case SEEK_SET:
	case SEEK_CUR:
	case SEEK_END:
		return generic_file_llseek_size(file, offset, whence,
						maxbytes, i_size_read(inode));
	case SEEK_DATA:
	case SEEK_HOLE:
465 466
		if (offset < 0)
			return -ENXIO;
467 468 469 470 471 472
		return f2fs_seek_block(file, offset, whence);
	}

	return -EINVAL;
}

J
Jaegeuk Kim 已提交
473 474
static int f2fs_file_mmap(struct file *file, struct vm_area_struct *vma)
{
475
	struct inode *inode = file_inode(file);
476
	int err;
477

478 479 480
	if (unlikely(f2fs_cp_error(F2FS_I_SB(inode))))
		return -EIO;

481
	/* we don't need to use inline_data strictly */
482 483 484
	err = f2fs_convert_inline_inode(inode);
	if (err)
		return err;
485

J
Jaegeuk Kim 已提交
486 487 488 489 490
	file_accessed(file);
	vma->vm_ops = &f2fs_file_vm_ops;
	return 0;
}

491 492
static int f2fs_file_open(struct inode *inode, struct file *filp)
{
493
	int err = fscrypt_file_open(inode, filp);
494

495 496
	if (err)
		return err;
H
Hyunchul Lee 已提交
497 498 499

	filp->f_mode |= FMODE_NOWAIT;

C
Chao Yu 已提交
500
	return dquot_file_open(inode, filp);
501 502
}

C
Chao Yu 已提交
503
void f2fs_truncate_data_blocks_range(struct dnode_of_data *dn, int count)
J
Jaegeuk Kim 已提交
504
{
505
	struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
J
Jaegeuk Kim 已提交
506
	struct f2fs_node *raw_node;
C
Chao Yu 已提交
507
	int nr_free = 0, ofs = dn->ofs_in_node, len = count;
J
Jaegeuk Kim 已提交
508
	__le32 *addr;
509 510 511 512
	int base = 0;

	if (IS_INODE(dn->node_page) && f2fs_has_extra_attr(dn->inode))
		base = get_extra_isize(dn->inode);
J
Jaegeuk Kim 已提交
513

514
	raw_node = F2FS_NODE(dn->node_page);
515
	addr = blkaddr_in_node(raw_node) + base + ofs;
J
Jaegeuk Kim 已提交
516

C
Chris Fries 已提交
517
	for (; count > 0; count--, addr++, dn->ofs_in_node++) {
J
Jaegeuk Kim 已提交
518
		block_t blkaddr = le32_to_cpu(*addr);
519

J
Jaegeuk Kim 已提交
520 521 522
		if (blkaddr == NULL_ADDR)
			continue;

J
Jaegeuk Kim 已提交
523
		dn->data_blkaddr = NULL_ADDR;
C
Chao Yu 已提交
524
		f2fs_set_data_blkaddr(dn);
525 526 527 528 529

		if (__is_valid_data_blkaddr(blkaddr) &&
			!f2fs_is_valid_blkaddr(sbi, blkaddr, DATA_GENERIC))
			continue;

C
Chao Yu 已提交
530
		f2fs_invalidate_blocks(sbi, blkaddr);
531
		if (dn->ofs_in_node == 0 && IS_INODE(dn->node_page))
532
			clear_inode_flag(dn->inode, FI_FIRST_BLOCK_WRITTEN);
J
Jaegeuk Kim 已提交
533 534
		nr_free++;
	}
C
Chao Yu 已提交
535

J
Jaegeuk Kim 已提交
536
	if (nr_free) {
C
Chao Yu 已提交
537 538 539 540 541
		pgoff_t fofs;
		/*
		 * once we invalidate valid blkaddr in range [ofs, ofs + count],
		 * we will invalidate all blkaddr in the whole range.
		 */
C
Chao Yu 已提交
542
		fofs = f2fs_start_bidx_of_node(ofs_of_node(dn->node_page),
543
							dn->inode) + ofs;
C
Chao Yu 已提交
544
		f2fs_update_extent_cache_range(dn, fofs, 0, len);
545
		dec_valid_block_count(sbi, dn->inode, nr_free);
J
Jaegeuk Kim 已提交
546 547
	}
	dn->ofs_in_node = ofs;
548

549
	f2fs_update_time(sbi, REQ_TIME);
550 551
	trace_f2fs_truncate_data_blocks_range(dn->inode, dn->nid,
					 dn->ofs_in_node, nr_free);
J
Jaegeuk Kim 已提交
552 553
}

C
Chao Yu 已提交
554
void f2fs_truncate_data_blocks(struct dnode_of_data *dn)
J
Jaegeuk Kim 已提交
555
{
C
Chao Yu 已提交
556
	f2fs_truncate_data_blocks_range(dn, ADDRS_PER_BLOCK);
J
Jaegeuk Kim 已提交
557 558
}

559
static int truncate_partial_data_page(struct inode *inode, u64 from,
560
								bool cache_only)
J
Jaegeuk Kim 已提交
561
{
562
	loff_t offset = from & (PAGE_SIZE - 1);
563
	pgoff_t index = from >> PAGE_SHIFT;
564
	struct address_space *mapping = inode->i_mapping;
J
Jaegeuk Kim 已提交
565 566
	struct page *page;

567
	if (!offset && !cache_only)
568
		return 0;
J
Jaegeuk Kim 已提交
569

570
	if (cache_only) {
571
		page = find_lock_page(mapping, index);
572 573 574
		if (page && PageUptodate(page))
			goto truncate_out;
		f2fs_put_page(page, 1);
575
		return 0;
576
	}
J
Jaegeuk Kim 已提交
577

C
Chao Yu 已提交
578
	page = f2fs_get_lock_data_page(inode, index, true);
579
	if (IS_ERR(page))
580
		return PTR_ERR(page) == -ENOENT ? 0 : PTR_ERR(page);
581
truncate_out:
582
	f2fs_wait_on_page_writeback(page, DATA, true);
583
	zero_user(page, offset, PAGE_SIZE - offset);
584 585 586 587

	/* An encrypted inode should have a key and truncate the last page. */
	f2fs_bug_on(F2FS_I_SB(inode), cache_only && f2fs_encrypted_inode(inode));
	if (!cache_only)
588
		set_page_dirty(page);
J
Jaegeuk Kim 已提交
589
	f2fs_put_page(page, 1);
590
	return 0;
J
Jaegeuk Kim 已提交
591 592
}

C
Chao Yu 已提交
593
int f2fs_truncate_blocks(struct inode *inode, u64 from, bool lock)
J
Jaegeuk Kim 已提交
594
{
595
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
J
Jaegeuk Kim 已提交
596 597
	struct dnode_of_data dn;
	pgoff_t free_from;
H
Huajun Li 已提交
598
	int count = 0, err = 0;
599
	struct page *ipage;
600
	bool truncate_page = false;
J
Jaegeuk Kim 已提交
601

602 603
	trace_f2fs_truncate_blocks_enter(inode, from);

C
Chao Yu 已提交
604
	free_from = (pgoff_t)F2FS_BLK_ALIGN(from);
J
Jaegeuk Kim 已提交
605

606 607 608
	if (free_from >= sbi->max_file_blocks)
		goto free_partial;

609 610
	if (lock)
		f2fs_lock_op(sbi);
H
Huajun Li 已提交
611

C
Chao Yu 已提交
612
	ipage = f2fs_get_node_page(sbi, inode->i_ino);
613 614 615 616 617 618
	if (IS_ERR(ipage)) {
		err = PTR_ERR(ipage);
		goto out;
	}

	if (f2fs_has_inline_data(inode)) {
C
Chao Yu 已提交
619
		f2fs_truncate_inline_inode(inode, ipage, from);
620
		f2fs_put_page(ipage, 1);
621
		truncate_page = true;
622 623 624 625
		goto out;
	}

	set_new_dnode(&dn, inode, ipage, NULL, 0);
C
Chao Yu 已提交
626
	err = f2fs_get_dnode_of_data(&dn, free_from, LOOKUP_NODE_RA);
J
Jaegeuk Kim 已提交
627 628 629
	if (err) {
		if (err == -ENOENT)
			goto free_next;
630
		goto out;
631 632
	}

633
	count = ADDRS_PER_PAGE(dn.node_page, inode);
J
Jaegeuk Kim 已提交
634 635

	count -= dn.ofs_in_node;
636
	f2fs_bug_on(sbi, count < 0);
637

J
Jaegeuk Kim 已提交
638
	if (dn.ofs_in_node || IS_INODE(dn.node_page)) {
C
Chao Yu 已提交
639
		f2fs_truncate_data_blocks_range(&dn, count);
J
Jaegeuk Kim 已提交
640 641 642 643 644
		free_from += count;
	}

	f2fs_put_dnode(&dn);
free_next:
C
Chao Yu 已提交
645
	err = f2fs_truncate_inode_blocks(inode, free_from);
646 647 648
out:
	if (lock)
		f2fs_unlock_op(sbi);
649
free_partial:
650 651
	/* lastly zero out the first data page */
	if (!err)
652
		err = truncate_partial_data_page(inode, from, truncate_page);
J
Jaegeuk Kim 已提交
653

654
	trace_f2fs_truncate_blocks_exit(inode, err);
J
Jaegeuk Kim 已提交
655 656 657
	return err;
}

658
int f2fs_truncate(struct inode *inode)
J
Jaegeuk Kim 已提交
659
{
660 661
	int err;

662 663 664
	if (unlikely(f2fs_cp_error(F2FS_I_SB(inode))))
		return -EIO;

J
Jaegeuk Kim 已提交
665 666
	if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode) ||
				S_ISLNK(inode->i_mode)))
667
		return 0;
J
Jaegeuk Kim 已提交
668

669 670
	trace_f2fs_truncate(inode);

671 672 673 674 675 676
#ifdef CONFIG_F2FS_FAULT_INJECTION
	if (time_to_inject(F2FS_I_SB(inode), FAULT_TRUNCATE)) {
		f2fs_show_injection_info(FAULT_TRUNCATE);
		return -EIO;
	}
#endif
677
	/* we should check inline_data size */
678
	if (!f2fs_may_inline_data(inode)) {
679 680 681
		err = f2fs_convert_inline_inode(inode);
		if (err)
			return err;
682 683
	}

C
Chao Yu 已提交
684
	err = f2fs_truncate_blocks(inode, i_size_read(inode), true);
685 686 687
	if (err)
		return err;

688
	inode->i_mtime = inode->i_ctime = current_time(inode);
689
	f2fs_mark_inode_dirty_sync(inode, false);
690
	return 0;
J
Jaegeuk Kim 已提交
691 692
}

693
int f2fs_getattr(const struct path *path, struct kstat *stat,
C
Chao Yu 已提交
694
		 u32 request_mask, unsigned int query_flags)
J
Jaegeuk Kim 已提交
695
{
696
	struct inode *inode = d_inode(path->dentry);
C
Chao Yu 已提交
697
	struct f2fs_inode_info *fi = F2FS_I(inode);
C
Chao Yu 已提交
698
	struct f2fs_inode *ri;
C
Chao Yu 已提交
699 700
	unsigned int flags;

C
Chao Yu 已提交
701 702 703 704 705 706 707 708
	if (f2fs_has_extra_attr(inode) &&
			f2fs_sb_has_inode_crtime(inode->i_sb) &&
			F2FS_FITS_IN_INODE(ri, fi->i_extra_isize, i_crtime)) {
		stat->result_mask |= STATX_BTIME;
		stat->btime.tv_sec = fi->i_crtime.tv_sec;
		stat->btime.tv_nsec = fi->i_crtime.tv_nsec;
	}

709
	flags = fi->i_flags & F2FS_FL_USER_VISIBLE;
710
	if (flags & F2FS_APPEND_FL)
C
Chao Yu 已提交
711
		stat->attributes |= STATX_ATTR_APPEND;
712
	if (flags & F2FS_COMPR_FL)
C
Chao Yu 已提交
713 714 715
		stat->attributes |= STATX_ATTR_COMPRESSED;
	if (f2fs_encrypted_inode(inode))
		stat->attributes |= STATX_ATTR_ENCRYPTED;
716
	if (flags & F2FS_IMMUTABLE_FL)
C
Chao Yu 已提交
717
		stat->attributes |= STATX_ATTR_IMMUTABLE;
718
	if (flags & F2FS_NODUMP_FL)
C
Chao Yu 已提交
719 720 721 722 723 724 725 726
		stat->attributes |= STATX_ATTR_NODUMP;

	stat->attributes_mask |= (STATX_ATTR_APPEND |
				  STATX_ATTR_COMPRESSED |
				  STATX_ATTR_ENCRYPTED |
				  STATX_ATTR_IMMUTABLE |
				  STATX_ATTR_NODUMP);

J
Jaegeuk Kim 已提交
727
	generic_fillattr(inode, stat);
728 729 730 731 732 733

	/* we need to show initial sectors used for inline_data/dentries */
	if ((S_ISREG(inode->i_mode) && f2fs_has_inline_data(inode)) ||
					f2fs_has_inline_dentry(inode))
		stat->blocks += (stat->size + 511) >> 9;

J
Jaegeuk Kim 已提交
734 735 736 737 738 739 740 741 742 743 744 745 746
	return 0;
}

#ifdef CONFIG_F2FS_FS_POSIX_ACL
static void __setattr_copy(struct inode *inode, const struct iattr *attr)
{
	unsigned int ia_valid = attr->ia_valid;

	if (ia_valid & ATTR_UID)
		inode->i_uid = attr->ia_uid;
	if (ia_valid & ATTR_GID)
		inode->i_gid = attr->ia_gid;
	if (ia_valid & ATTR_ATIME)
747 748
		inode->i_atime = timespec64_trunc(attr->ia_atime,
						  inode->i_sb->s_time_gran);
J
Jaegeuk Kim 已提交
749
	if (ia_valid & ATTR_MTIME)
750 751
		inode->i_mtime = timespec64_trunc(attr->ia_mtime,
						  inode->i_sb->s_time_gran);
J
Jaegeuk Kim 已提交
752
	if (ia_valid & ATTR_CTIME)
753 754
		inode->i_ctime = timespec64_trunc(attr->ia_ctime,
						  inode->i_sb->s_time_gran);
J
Jaegeuk Kim 已提交
755 756 757 758 759
	if (ia_valid & ATTR_MODE) {
		umode_t mode = attr->ia_mode;

		if (!in_group_p(inode->i_gid) && !capable(CAP_FSETID))
			mode &= ~S_ISGID;
760
		set_acl_inode(inode, mode);
J
Jaegeuk Kim 已提交
761 762 763 764 765 766 767 768
	}
}
#else
#define __setattr_copy setattr_copy
#endif

int f2fs_setattr(struct dentry *dentry, struct iattr *attr)
{
769
	struct inode *inode = d_inode(dentry);
J
Jaegeuk Kim 已提交
770
	int err;
771
	bool size_changed = false;
J
Jaegeuk Kim 已提交
772

773 774 775
	if (unlikely(f2fs_cp_error(F2FS_I_SB(inode))))
		return -EIO;

776
	err = setattr_prepare(dentry, attr);
J
Jaegeuk Kim 已提交
777 778 779
	if (err)
		return err;

780 781 782 783
	err = fscrypt_prepare_setattr(dentry, attr);
	if (err)
		return err;

C
Chao Yu 已提交
784 785 786 787 788 789 790 791 792 793 794 795 796 797
	if (is_quota_modification(inode, attr)) {
		err = dquot_initialize(inode);
		if (err)
			return err;
	}
	if ((attr->ia_valid & ATTR_UID &&
		!uid_eq(attr->ia_uid, inode->i_uid)) ||
		(attr->ia_valid & ATTR_GID &&
		!gid_eq(attr->ia_gid, inode->i_gid))) {
		err = dquot_transfer(inode, attr);
		if (err)
			return err;
	}

798
	if (attr->ia_valid & ATTR_SIZE) {
799 800 801 802 803 804 805 806
		bool to_smaller = (attr->ia_size <= i_size_read(inode));

		down_write(&F2FS_I(inode)->i_mmap_sem);
		down_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);

		truncate_setsize(inode, attr->ia_size);

		if (to_smaller)
807
			err = f2fs_truncate(inode);
808 809 810 811 812 813 814 815 816
		/*
		 * do not trim all blocks after i_size if target size is
		 * larger than i_size.
		 */
		up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
		up_write(&F2FS_I(inode)->i_mmap_sem);

		if (err)
			return err;
817

818
		if (!to_smaller) {
819
			/* should convert inline inode here */
820
			if (!f2fs_may_inline_data(inode)) {
821 822 823 824
				err = f2fs_convert_inline_inode(inode);
				if (err)
					return err;
			}
825
			inode->i_mtime = inode->i_ctime = current_time(inode);
826
		}
827

828 829 830 831
		down_write(&F2FS_I(inode)->i_sem);
		F2FS_I(inode)->last_disk_size = i_size_read(inode);
		up_write(&F2FS_I(inode)->i_sem);

832
		size_changed = true;
J
Jaegeuk Kim 已提交
833 834 835 836 837
	}

	__setattr_copy(inode, attr);

	if (attr->ia_valid & ATTR_MODE) {
C
Chao Yu 已提交
838
		err = posix_acl_chmod(inode, f2fs_get_inode_mode(inode));
839 840 841
		if (err || is_inode_flag_set(inode, FI_ACL_MODE)) {
			inode->i_mode = F2FS_I(inode)->i_acl_mode;
			clear_inode_flag(inode, FI_ACL_MODE);
J
Jaegeuk Kim 已提交
842 843 844
		}
	}

845 846
	/* file size may changed here */
	f2fs_mark_inode_dirty_sync(inode, size_changed);
847 848 849 850

	/* inode change will produce dirty node pages flushed by checkpoint */
	f2fs_balance_fs(F2FS_I_SB(inode), true);

J
Jaegeuk Kim 已提交
851 852 853 854 855 856 857
	return err;
}

const struct inode_operations f2fs_file_inode_operations = {
	.getattr	= f2fs_getattr,
	.setattr	= f2fs_setattr,
	.get_acl	= f2fs_get_acl,
858
	.set_acl	= f2fs_set_acl,
J
Jaegeuk Kim 已提交
859 860 861
#ifdef CONFIG_F2FS_FS_XATTR
	.listxattr	= f2fs_listxattr,
#endif
J
Jaegeuk Kim 已提交
862
	.fiemap		= f2fs_fiemap,
J
Jaegeuk Kim 已提交
863 864
};

C
Chao Yu 已提交
865
static int fill_zero(struct inode *inode, pgoff_t index,
J
Jaegeuk Kim 已提交
866 867
					loff_t start, loff_t len)
{
868
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
J
Jaegeuk Kim 已提交
869 870 871
	struct page *page;

	if (!len)
C
Chao Yu 已提交
872
		return 0;
J
Jaegeuk Kim 已提交
873

J
Jaegeuk Kim 已提交
874
	f2fs_balance_fs(sbi, true);
875

876
	f2fs_lock_op(sbi);
C
Chao Yu 已提交
877
	page = f2fs_get_new_data_page(inode, NULL, index, false);
878
	f2fs_unlock_op(sbi);
J
Jaegeuk Kim 已提交
879

C
Chao Yu 已提交
880 881 882
	if (IS_ERR(page))
		return PTR_ERR(page);

883
	f2fs_wait_on_page_writeback(page, DATA, true);
C
Chao Yu 已提交
884 885 886 887
	zero_user(page, start, len);
	set_page_dirty(page);
	f2fs_put_page(page, 1);
	return 0;
J
Jaegeuk Kim 已提交
888 889
}

C
Chao Yu 已提交
890
int f2fs_truncate_hole(struct inode *inode, pgoff_t pg_start, pgoff_t pg_end)
J
Jaegeuk Kim 已提交
891 892 893
{
	int err;

894
	while (pg_start < pg_end) {
J
Jaegeuk Kim 已提交
895
		struct dnode_of_data dn;
896
		pgoff_t end_offset, count;
897

J
Jaegeuk Kim 已提交
898
		set_new_dnode(&dn, inode, NULL, NULL, 0);
C
Chao Yu 已提交
899
		err = f2fs_get_dnode_of_data(&dn, pg_start, LOOKUP_NODE);
J
Jaegeuk Kim 已提交
900
		if (err) {
901
			if (err == -ENOENT) {
C
Chao Yu 已提交
902 903
				pg_start = f2fs_get_next_page_offset(&dn,
								pg_start);
J
Jaegeuk Kim 已提交
904
				continue;
905
			}
J
Jaegeuk Kim 已提交
906 907 908
			return err;
		}

909
		end_offset = ADDRS_PER_PAGE(dn.node_page, inode);
910 911 912 913
		count = min(end_offset - dn.ofs_in_node, pg_end - pg_start);

		f2fs_bug_on(F2FS_I_SB(inode), count == 0 || count > end_offset);

C
Chao Yu 已提交
914
		f2fs_truncate_data_blocks_range(&dn, count);
J
Jaegeuk Kim 已提交
915
		f2fs_put_dnode(&dn);
916 917

		pg_start += count;
J
Jaegeuk Kim 已提交
918 919 920 921
	}
	return 0;
}

C
Chao Yu 已提交
922
static int punch_hole(struct inode *inode, loff_t offset, loff_t len)
J
Jaegeuk Kim 已提交
923 924 925
{
	pgoff_t pg_start, pg_end;
	loff_t off_start, off_end;
926
	int ret;
J
Jaegeuk Kim 已提交
927

928 929 930
	ret = f2fs_convert_inline_inode(inode);
	if (ret)
		return ret;
H
Huajun Li 已提交
931

932 933
	pg_start = ((unsigned long long) offset) >> PAGE_SHIFT;
	pg_end = ((unsigned long long) offset + len) >> PAGE_SHIFT;
J
Jaegeuk Kim 已提交
934

935 936
	off_start = offset & (PAGE_SIZE - 1);
	off_end = (offset + len) & (PAGE_SIZE - 1);
J
Jaegeuk Kim 已提交
937 938

	if (pg_start == pg_end) {
C
Chao Yu 已提交
939
		ret = fill_zero(inode, pg_start, off_start,
J
Jaegeuk Kim 已提交
940
						off_end - off_start);
C
Chao Yu 已提交
941 942
		if (ret)
			return ret;
J
Jaegeuk Kim 已提交
943
	} else {
C
Chao Yu 已提交
944 945
		if (off_start) {
			ret = fill_zero(inode, pg_start++, off_start,
946
						PAGE_SIZE - off_start);
C
Chao Yu 已提交
947 948 949 950 951 952 953 954
			if (ret)
				return ret;
		}
		if (off_end) {
			ret = fill_zero(inode, pg_end, 0, off_end);
			if (ret)
				return ret;
		}
J
Jaegeuk Kim 已提交
955 956 957 958

		if (pg_start < pg_end) {
			struct address_space *mapping = inode->i_mapping;
			loff_t blk_start, blk_end;
959
			struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
960

J
Jaegeuk Kim 已提交
961
			f2fs_balance_fs(sbi, true);
J
Jaegeuk Kim 已提交
962

963 964
			blk_start = (loff_t)pg_start << PAGE_SHIFT;
			blk_end = (loff_t)pg_end << PAGE_SHIFT;
965

966
			down_write(&F2FS_I(inode)->i_mmap_sem);
967 968
			down_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);

J
Jaegeuk Kim 已提交
969 970
			truncate_inode_pages_range(mapping, blk_start,
					blk_end - 1);
971

972
			f2fs_lock_op(sbi);
C
Chao Yu 已提交
973
			ret = f2fs_truncate_hole(inode, pg_start, pg_end);
974
			f2fs_unlock_op(sbi);
975 976

			up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
977
			up_write(&F2FS_I(inode)->i_mmap_sem);
J
Jaegeuk Kim 已提交
978 979 980 981 982 983
		}
	}

	return ret;
}

984 985
static int __read_out_blkaddrs(struct inode *inode, block_t *blkaddr,
				int *do_replace, pgoff_t off, pgoff_t len)
C
Chao Yu 已提交
986 987 988
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct dnode_of_data dn;
989
	int ret, done, i;
990

991
next_dnode:
992
	set_new_dnode(&dn, inode, NULL, NULL, 0);
C
Chao Yu 已提交
993
	ret = f2fs_get_dnode_of_data(&dn, off, LOOKUP_NODE_RA);
994 995 996
	if (ret && ret != -ENOENT) {
		return ret;
	} else if (ret == -ENOENT) {
997 998 999 1000 1001 1002 1003 1004 1005 1006 1007
		if (dn.max_level == 0)
			return -ENOENT;
		done = min((pgoff_t)ADDRS_PER_BLOCK - dn.ofs_in_node, len);
		blkaddr += done;
		do_replace += done;
		goto next;
	}

	done = min((pgoff_t)ADDRS_PER_PAGE(dn.node_page, inode) -
							dn.ofs_in_node, len);
	for (i = 0; i < done; i++, blkaddr++, do_replace++, dn.ofs_in_node++) {
1008 1009
		*blkaddr = datablock_addr(dn.inode,
					dn.node_page, dn.ofs_in_node);
C
Chao Yu 已提交
1010
		if (!f2fs_is_checkpointed_data(sbi, *blkaddr)) {
1011 1012 1013 1014 1015 1016

			if (test_opt(sbi, LFS)) {
				f2fs_put_dnode(&dn);
				return -ENOTSUPP;
			}

1017
			/* do not invalidate this block address */
1018
			f2fs_update_data_blkaddr(&dn, NULL_ADDR);
1019
			*do_replace = 1;
C
Chao Yu 已提交
1020
		}
1021
	}
1022 1023 1024 1025 1026 1027 1028 1029
	f2fs_put_dnode(&dn);
next:
	len -= done;
	off += done;
	if (len)
		goto next_dnode;
	return 0;
}
C
Chao Yu 已提交
1030

1031 1032 1033 1034 1035 1036
static int __roll_back_blkaddrs(struct inode *inode, block_t *blkaddr,
				int *do_replace, pgoff_t off, int len)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct dnode_of_data dn;
	int ret, i;
C
Chao Yu 已提交
1037

1038 1039 1040
	for (i = 0; i < len; i++, do_replace++, blkaddr++) {
		if (*do_replace == 0)
			continue;
C
Chao Yu 已提交
1041

1042
		set_new_dnode(&dn, inode, NULL, NULL, 0);
C
Chao Yu 已提交
1043
		ret = f2fs_get_dnode_of_data(&dn, off + i, LOOKUP_NODE_RA);
1044 1045
		if (ret) {
			dec_valid_block_count(sbi, inode, 1);
C
Chao Yu 已提交
1046
			f2fs_invalidate_blocks(sbi, *blkaddr);
1047 1048
		} else {
			f2fs_update_data_blkaddr(&dn, *blkaddr);
1049
		}
1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061
		f2fs_put_dnode(&dn);
	}
	return 0;
}

static int __clone_blkaddrs(struct inode *src_inode, struct inode *dst_inode,
			block_t *blkaddr, int *do_replace,
			pgoff_t src, pgoff_t dst, pgoff_t len, bool full)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(src_inode);
	pgoff_t i = 0;
	int ret;
1062

1063 1064 1065 1066
	while (i < len) {
		if (blkaddr[i] == NULL_ADDR && !full) {
			i++;
			continue;
1067
		}
C
Chao Yu 已提交
1068

1069 1070 1071 1072 1073
		if (do_replace[i] || blkaddr[i] == NULL_ADDR) {
			struct dnode_of_data dn;
			struct node_info ni;
			size_t new_size;
			pgoff_t ilen;
C
Chao Yu 已提交
1074

1075
			set_new_dnode(&dn, dst_inode, NULL, NULL, 0);
C
Chao Yu 已提交
1076
			ret = f2fs_get_dnode_of_data(&dn, dst + i, ALLOC_NODE);
1077 1078
			if (ret)
				return ret;
C
Chao Yu 已提交
1079

1080 1081 1082 1083 1084 1085
			ret = f2fs_get_node_info(sbi, dn.nid, &ni);
			if (ret) {
				f2fs_put_dnode(&dn);
				return ret;
			}

1086 1087 1088 1089
			ilen = min((pgoff_t)
				ADDRS_PER_PAGE(dn.node_page, dst_inode) -
						dn.ofs_in_node, len - i);
			do {
1090 1091
				dn.data_blkaddr = datablock_addr(dn.inode,
						dn.node_page, dn.ofs_in_node);
C
Chao Yu 已提交
1092
				f2fs_truncate_data_blocks_range(&dn, 1);
1093 1094 1095

				if (do_replace[i]) {
					f2fs_i_blocks_write(src_inode,
C
Chao Yu 已提交
1096
							1, false, false);
1097
					f2fs_i_blocks_write(dst_inode,
C
Chao Yu 已提交
1098
							1, true, false);
1099 1100 1101 1102 1103 1104 1105 1106 1107 1108
					f2fs_replace_block(sbi, &dn, dn.data_blkaddr,
					blkaddr[i], ni.version, true, false);

					do_replace[i] = 0;
				}
				dn.ofs_in_node++;
				i++;
				new_size = (dst + i) << PAGE_SHIFT;
				if (dst_inode->i_size < new_size)
					f2fs_i_size_write(dst_inode, new_size);
1109
			} while (--ilen && (do_replace[i] || blkaddr[i] == NULL_ADDR));
1110

1111 1112 1113 1114
			f2fs_put_dnode(&dn);
		} else {
			struct page *psrc, *pdst;

C
Chao Yu 已提交
1115 1116
			psrc = f2fs_get_lock_data_page(src_inode,
							src + i, true);
1117 1118
			if (IS_ERR(psrc))
				return PTR_ERR(psrc);
C
Chao Yu 已提交
1119
			pdst = f2fs_get_new_data_page(dst_inode, NULL, dst + i,
1120 1121 1122 1123 1124 1125 1126 1127
								true);
			if (IS_ERR(pdst)) {
				f2fs_put_page(psrc, 1);
				return PTR_ERR(pdst);
			}
			f2fs_copy_page(psrc, pdst);
			set_page_dirty(pdst);
			f2fs_put_page(pdst, 1);
1128
			f2fs_put_page(psrc, 1);
C
Chao Yu 已提交
1129

C
Chao Yu 已提交
1130 1131
			ret = f2fs_truncate_hole(src_inode,
						src + i, src + i + 1);
1132 1133 1134 1135
			if (ret)
				return ret;
			i++;
		}
1136 1137
	}
	return 0;
1138
}
C
Chao Yu 已提交
1139

1140 1141
static int __exchange_data_block(struct inode *src_inode,
			struct inode *dst_inode, pgoff_t src, pgoff_t dst,
1142
			pgoff_t len, bool full)
1143 1144 1145
{
	block_t *src_blkaddr;
	int *do_replace;
1146
	pgoff_t olen;
1147 1148
	int ret;

1149 1150
	while (len) {
		olen = min((pgoff_t)4 * ADDRS_PER_BLOCK, len);
1151

C
Chao Yu 已提交
1152
		src_blkaddr = f2fs_kvzalloc(F2FS_I_SB(src_inode),
1153 1154
					array_size(olen, sizeof(block_t)),
					GFP_KERNEL);
1155 1156
		if (!src_blkaddr)
			return -ENOMEM;
1157

C
Chao Yu 已提交
1158
		do_replace = f2fs_kvzalloc(F2FS_I_SB(src_inode),
1159 1160
					array_size(olen, sizeof(int)),
					GFP_KERNEL);
1161 1162 1163 1164
		if (!do_replace) {
			kvfree(src_blkaddr);
			return -ENOMEM;
		}
1165

1166 1167 1168 1169
		ret = __read_out_blkaddrs(src_inode, src_blkaddr,
					do_replace, src, olen);
		if (ret)
			goto roll_back;
1170

1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182
		ret = __clone_blkaddrs(src_inode, dst_inode, src_blkaddr,
					do_replace, src, dst, olen, full);
		if (ret)
			goto roll_back;

		src += olen;
		dst += olen;
		len -= olen;

		kvfree(src_blkaddr);
		kvfree(do_replace);
	}
1183 1184 1185
	return 0;

roll_back:
1186
	__roll_back_blkaddrs(src_inode, src_blkaddr, do_replace, src, olen);
1187 1188
	kvfree(src_blkaddr);
	kvfree(do_replace);
1189 1190
	return ret;
}
C
Chao Yu 已提交
1191

1192 1193 1194 1195
static int f2fs_do_collapse(struct inode *inode, pgoff_t start, pgoff_t end)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	pgoff_t nrpages = (i_size_read(inode) + PAGE_SIZE - 1) / PAGE_SIZE;
1196
	int ret;
1197

1198 1199
	f2fs_balance_fs(sbi, true);
	f2fs_lock_op(sbi);
1200 1201 1202

	f2fs_drop_extent_tree(inode);

1203 1204
	ret = __exchange_data_block(inode, inode, end, start, nrpages - end, true);
	f2fs_unlock_op(sbi);
C
Chao Yu 已提交
1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220
	return ret;
}

static int f2fs_collapse_range(struct inode *inode, loff_t offset, loff_t len)
{
	pgoff_t pg_start, pg_end;
	loff_t new_size;
	int ret;

	if (offset + len >= i_size_read(inode))
		return -EINVAL;

	/* collapse range should be aligned to block size of f2fs. */
	if (offset & (F2FS_BLKSIZE - 1) || len & (F2FS_BLKSIZE - 1))
		return -EINVAL;

1221 1222 1223
	ret = f2fs_convert_inline_inode(inode);
	if (ret)
		return ret;
1224

1225 1226
	pg_start = offset >> PAGE_SHIFT;
	pg_end = (offset + len) >> PAGE_SHIFT;
C
Chao Yu 已提交
1227

1228
	/* avoid gc operation during block exchange */
C
Chao Yu 已提交
1229
	down_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
1230

1231
	down_write(&F2FS_I(inode)->i_mmap_sem);
C
Chao Yu 已提交
1232 1233 1234
	/* write out all dirty pages from offset */
	ret = filemap_write_and_wait_range(inode->i_mapping, offset, LLONG_MAX);
	if (ret)
1235
		goto out_unlock;
1236

C
Chao Yu 已提交
1237 1238 1239 1240
	truncate_pagecache(inode, offset);

	ret = f2fs_do_collapse(inode, pg_start, pg_end);
	if (ret)
1241
		goto out_unlock;
C
Chao Yu 已提交
1242

1243 1244 1245 1246
	/* write out all moved pages, if possible */
	filemap_write_and_wait_range(inode->i_mapping, offset, LLONG_MAX);
	truncate_pagecache(inode, offset);

C
Chao Yu 已提交
1247
	new_size = i_size_read(inode) - len;
1248
	truncate_pagecache(inode, new_size);
C
Chao Yu 已提交
1249

C
Chao Yu 已提交
1250
	ret = f2fs_truncate_blocks(inode, new_size, true);
C
Chao Yu 已提交
1251
	if (!ret)
1252
		f2fs_i_size_write(inode, new_size);
1253
out_unlock:
1254
	up_write(&F2FS_I(inode)->i_mmap_sem);
C
Chao Yu 已提交
1255
	up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
C
Chao Yu 已提交
1256 1257 1258
	return ret;
}

1259 1260 1261 1262 1263 1264 1265 1266 1267 1268
static int f2fs_do_zero_range(struct dnode_of_data *dn, pgoff_t start,
								pgoff_t end)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(dn->inode);
	pgoff_t index = start;
	unsigned int ofs_in_node = dn->ofs_in_node;
	blkcnt_t count = 0;
	int ret;

	for (; index < end; index++, dn->ofs_in_node++) {
1269 1270
		if (datablock_addr(dn->inode, dn->node_page,
					dn->ofs_in_node) == NULL_ADDR)
1271 1272 1273 1274
			count++;
	}

	dn->ofs_in_node = ofs_in_node;
C
Chao Yu 已提交
1275
	ret = f2fs_reserve_new_blocks(dn, count);
1276 1277 1278 1279 1280
	if (ret)
		return ret;

	dn->ofs_in_node = ofs_in_node;
	for (index = start; index < end; index++, dn->ofs_in_node++) {
1281 1282
		dn->data_blkaddr = datablock_addr(dn->inode,
					dn->node_page, dn->ofs_in_node);
1283
		/*
C
Chao Yu 已提交
1284
		 * f2fs_reserve_new_blocks will not guarantee entire block
1285 1286 1287 1288 1289 1290 1291
		 * allocation.
		 */
		if (dn->data_blkaddr == NULL_ADDR) {
			ret = -ENOSPC;
			break;
		}
		if (dn->data_blkaddr != NEW_ADDR) {
C
Chao Yu 已提交
1292
			f2fs_invalidate_blocks(sbi, dn->data_blkaddr);
1293
			dn->data_blkaddr = NEW_ADDR;
C
Chao Yu 已提交
1294
			f2fs_set_data_blkaddr(dn);
1295 1296 1297 1298 1299 1300 1301 1302
		}
	}

	f2fs_update_extent_cache_range(dn, start, 0, index - start);

	return ret;
}

C
Chao Yu 已提交
1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316
static int f2fs_zero_range(struct inode *inode, loff_t offset, loff_t len,
								int mode)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct address_space *mapping = inode->i_mapping;
	pgoff_t index, pg_start, pg_end;
	loff_t new_size = i_size_read(inode);
	loff_t off_start, off_end;
	int ret = 0;

	ret = inode_newsize_ok(inode, (len + offset));
	if (ret)
		return ret;

1317 1318 1319
	ret = f2fs_convert_inline_inode(inode);
	if (ret)
		return ret;
C
Chao Yu 已提交
1320

1321
	down_write(&F2FS_I(inode)->i_mmap_sem);
C
Chao Yu 已提交
1322 1323
	ret = filemap_write_and_wait_range(mapping, offset, offset + len - 1);
	if (ret)
1324
		goto out_sem;
C
Chao Yu 已提交
1325

1326 1327
	pg_start = ((unsigned long long) offset) >> PAGE_SHIFT;
	pg_end = ((unsigned long long) offset + len) >> PAGE_SHIFT;
C
Chao Yu 已提交
1328

1329 1330
	off_start = offset & (PAGE_SIZE - 1);
	off_end = (offset + len) & (PAGE_SIZE - 1);
C
Chao Yu 已提交
1331 1332

	if (pg_start == pg_end) {
C
Chao Yu 已提交
1333 1334 1335
		ret = fill_zero(inode, pg_start, off_start,
						off_end - off_start);
		if (ret)
1336
			goto out_sem;
C
Chao Yu 已提交
1337

C
Chao Yu 已提交
1338 1339 1340
		new_size = max_t(loff_t, new_size, offset + len);
	} else {
		if (off_start) {
C
Chao Yu 已提交
1341
			ret = fill_zero(inode, pg_start++, off_start,
1342
						PAGE_SIZE - off_start);
C
Chao Yu 已提交
1343
			if (ret)
1344
				goto out_sem;
C
Chao Yu 已提交
1345

C
Chao Yu 已提交
1346
			new_size = max_t(loff_t, new_size,
1347
					(loff_t)pg_start << PAGE_SHIFT);
C
Chao Yu 已提交
1348 1349
		}

1350
		for (index = pg_start; index < pg_end;) {
C
Chao Yu 已提交
1351
			struct dnode_of_data dn;
1352 1353
			unsigned int end_offset;
			pgoff_t end;
C
Chao Yu 已提交
1354

1355 1356 1357 1358 1359 1360
			down_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);

			truncate_pagecache_range(inode,
				(loff_t)index << PAGE_SHIFT,
				((loff_t)pg_end << PAGE_SHIFT) - 1);

C
Chao Yu 已提交
1361 1362
			f2fs_lock_op(sbi);

1363
			set_new_dnode(&dn, inode, NULL, NULL, 0);
C
Chao Yu 已提交
1364
			ret = f2fs_get_dnode_of_data(&dn, index, ALLOC_NODE);
C
Chao Yu 已提交
1365 1366
			if (ret) {
				f2fs_unlock_op(sbi);
1367
				up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
C
Chao Yu 已提交
1368 1369 1370
				goto out;
			}

1371 1372 1373 1374
			end_offset = ADDRS_PER_PAGE(dn.node_page, inode);
			end = min(pg_end, end_offset - dn.ofs_in_node + index);

			ret = f2fs_do_zero_range(&dn, index, end);
C
Chao Yu 已提交
1375
			f2fs_put_dnode(&dn);
1376

C
Chao Yu 已提交
1377
			f2fs_unlock_op(sbi);
1378
			up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
1379 1380 1381

			f2fs_balance_fs(sbi, dn.node_changed);

1382 1383
			if (ret)
				goto out;
C
Chao Yu 已提交
1384

1385
			index = end;
C
Chao Yu 已提交
1386
			new_size = max_t(loff_t, new_size,
1387
					(loff_t)index << PAGE_SHIFT);
C
Chao Yu 已提交
1388 1389 1390
		}

		if (off_end) {
C
Chao Yu 已提交
1391 1392 1393 1394
			ret = fill_zero(inode, pg_end, 0, off_end);
			if (ret)
				goto out;

C
Chao Yu 已提交
1395 1396 1397 1398 1399
			new_size = max_t(loff_t, new_size, offset + len);
		}
	}

out:
1400 1401 1402 1403 1404 1405
	if (new_size > i_size_read(inode)) {
		if (mode & FALLOC_FL_KEEP_SIZE)
			file_set_keep_isize(inode);
		else
			f2fs_i_size_write(inode, new_size);
	}
1406 1407
out_sem:
	up_write(&F2FS_I(inode)->i_mmap_sem);
C
Chao Yu 已提交
1408 1409 1410 1411

	return ret;
}

C
Chao Yu 已提交
1412 1413 1414
static int f2fs_insert_range(struct inode *inode, loff_t offset, loff_t len)
{
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1415
	pgoff_t nr, pg_start, pg_end, delta, idx;
C
Chao Yu 已提交
1416
	loff_t new_size;
1417
	int ret = 0;
C
Chao Yu 已提交
1418 1419

	new_size = i_size_read(inode) + len;
1420 1421 1422
	ret = inode_newsize_ok(inode, new_size);
	if (ret)
		return ret;
C
Chao Yu 已提交
1423 1424 1425 1426 1427 1428 1429 1430

	if (offset >= i_size_read(inode))
		return -EINVAL;

	/* insert range should be aligned to block size of f2fs. */
	if (offset & (F2FS_BLKSIZE - 1) || len & (F2FS_BLKSIZE - 1))
		return -EINVAL;

1431 1432 1433
	ret = f2fs_convert_inline_inode(inode);
	if (ret)
		return ret;
1434

J
Jaegeuk Kim 已提交
1435
	f2fs_balance_fs(sbi, true);
1436

1437
	/* avoid gc operation during block exchange */
C
Chao Yu 已提交
1438
	down_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
1439

1440
	down_write(&F2FS_I(inode)->i_mmap_sem);
C
Chao Yu 已提交
1441
	ret = f2fs_truncate_blocks(inode, i_size_read(inode), true);
C
Chao Yu 已提交
1442
	if (ret)
1443
		goto out;
C
Chao Yu 已提交
1444 1445 1446 1447

	/* write out all dirty pages from offset */
	ret = filemap_write_and_wait_range(inode->i_mapping, offset, LLONG_MAX);
	if (ret)
1448
		goto out;
C
Chao Yu 已提交
1449 1450 1451

	truncate_pagecache(inode, offset);

1452 1453
	pg_start = offset >> PAGE_SHIFT;
	pg_end = (offset + len) >> PAGE_SHIFT;
C
Chao Yu 已提交
1454
	delta = pg_end - pg_start;
1455 1456 1457 1458 1459 1460 1461
	idx = (i_size_read(inode) + PAGE_SIZE - 1) / PAGE_SIZE;

	while (!ret && idx > pg_start) {
		nr = idx - pg_start;
		if (nr > delta)
			nr = delta;
		idx -= nr;
C
Chao Yu 已提交
1462 1463

		f2fs_lock_op(sbi);
1464 1465
		f2fs_drop_extent_tree(inode);

1466 1467
		ret = __exchange_data_block(inode, inode, idx,
					idx + delta, nr, false);
C
Chao Yu 已提交
1468 1469 1470
		f2fs_unlock_op(sbi);
	}

1471 1472 1473 1474 1475
	/* write out all moved pages, if possible */
	filemap_write_and_wait_range(inode->i_mapping, offset, LLONG_MAX);
	truncate_pagecache(inode, offset);

	if (!ret)
1476
		f2fs_i_size_write(inode, new_size);
1477 1478
out:
	up_write(&F2FS_I(inode)->i_mmap_sem);
C
Chao Yu 已提交
1479
	up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
C
Chao Yu 已提交
1480 1481 1482
	return ret;
}

J
Jaegeuk Kim 已提交
1483 1484 1485
static int expand_inode_data(struct inode *inode, loff_t offset,
					loff_t len, int mode)
{
1486
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
1487
	struct f2fs_map_blocks map = { .m_next_pgofs = NULL,
1488
			.m_next_extent = NULL, .m_seg_type = NO_CHECK_TYPE };
1489
	pgoff_t pg_end;
J
Jaegeuk Kim 已提交
1490
	loff_t new_size = i_size_read(inode);
1491
	loff_t off_end;
1492
	int err;
J
Jaegeuk Kim 已提交
1493

1494 1495 1496
	err = inode_newsize_ok(inode, (len + offset));
	if (err)
		return err;
J
Jaegeuk Kim 已提交
1497

1498 1499 1500
	err = f2fs_convert_inline_inode(inode);
	if (err)
		return err;
1501

J
Jaegeuk Kim 已提交
1502
	f2fs_balance_fs(sbi, true);
1503

1504
	pg_end = ((unsigned long long)offset + len) >> PAGE_SHIFT;
1505
	off_end = (offset + len) & (PAGE_SIZE - 1);
J
Jaegeuk Kim 已提交
1506

1507 1508 1509 1510
	map.m_lblk = ((unsigned long long)offset) >> PAGE_SHIFT;
	map.m_len = pg_end - map.m_lblk;
	if (off_end)
		map.m_len++;
1511

1512 1513
	err = f2fs_map_blocks(inode, &map, 1, F2FS_GET_BLOCK_PRE_AIO);
	if (err) {
1514
		pgoff_t last_off;
J
Jaegeuk Kim 已提交
1515

1516
		if (!map.m_len)
1517
			return err;
1518

1519 1520 1521
		last_off = map.m_lblk + map.m_len - 1;

		/* update new size to the failed position */
1522
		new_size = (last_off == pg_end) ? offset + len :
1523 1524 1525
					(loff_t)(last_off + 1) << PAGE_SHIFT;
	} else {
		new_size = ((loff_t)pg_end << PAGE_SHIFT) + off_end;
J
Jaegeuk Kim 已提交
1526 1527
	}

1528 1529 1530 1531 1532 1533
	if (new_size > i_size_read(inode)) {
		if (mode & FALLOC_FL_KEEP_SIZE)
			file_set_keep_isize(inode);
		else
			f2fs_i_size_write(inode, new_size);
	}
J
Jaegeuk Kim 已提交
1534

1535
	return err;
J
Jaegeuk Kim 已提交
1536 1537 1538 1539 1540
}

static long f2fs_fallocate(struct file *file, int mode,
				loff_t offset, loff_t len)
{
A
Al Viro 已提交
1541
	struct inode *inode = file_inode(file);
1542
	long ret = 0;
J
Jaegeuk Kim 已提交
1543

1544 1545 1546
	if (unlikely(f2fs_cp_error(F2FS_I_SB(inode))))
		return -EIO;

1547 1548 1549 1550
	/* f2fs only support ->fallocate for regular file */
	if (!S_ISREG(inode->i_mode))
		return -EINVAL;

C
Chao Yu 已提交
1551 1552
	if (f2fs_encrypted_inode(inode) &&
		(mode & (FALLOC_FL_COLLAPSE_RANGE | FALLOC_FL_INSERT_RANGE)))
1553 1554
		return -EOPNOTSUPP;

C
Chao Yu 已提交
1555
	if (mode & ~(FALLOC_FL_KEEP_SIZE | FALLOC_FL_PUNCH_HOLE |
C
Chao Yu 已提交
1556 1557
			FALLOC_FL_COLLAPSE_RANGE | FALLOC_FL_ZERO_RANGE |
			FALLOC_FL_INSERT_RANGE))
J
Jaegeuk Kim 已提交
1558 1559
		return -EOPNOTSUPP;

A
Al Viro 已提交
1560
	inode_lock(inode);
1561

1562 1563 1564 1565
	if (mode & FALLOC_FL_PUNCH_HOLE) {
		if (offset >= inode->i_size)
			goto out;

C
Chao Yu 已提交
1566
		ret = punch_hole(inode, offset, len);
C
Chao Yu 已提交
1567 1568
	} else if (mode & FALLOC_FL_COLLAPSE_RANGE) {
		ret = f2fs_collapse_range(inode, offset, len);
C
Chao Yu 已提交
1569 1570
	} else if (mode & FALLOC_FL_ZERO_RANGE) {
		ret = f2fs_zero_range(inode, offset, len, mode);
C
Chao Yu 已提交
1571 1572
	} else if (mode & FALLOC_FL_INSERT_RANGE) {
		ret = f2fs_insert_range(inode, offset, len);
C
Chao Yu 已提交
1573
	} else {
J
Jaegeuk Kim 已提交
1574
		ret = expand_inode_data(inode, offset, len, mode);
C
Chao Yu 已提交
1575
	}
J
Jaegeuk Kim 已提交
1576

1577
	if (!ret) {
1578
		inode->i_mtime = inode->i_ctime = current_time(inode);
1579
		f2fs_mark_inode_dirty_sync(inode, false);
1580
		f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
1581
	}
1582

1583
out:
A
Al Viro 已提交
1584
	inode_unlock(inode);
1585

1586
	trace_f2fs_fallocate(inode, mode, offset, len, ret);
J
Jaegeuk Kim 已提交
1587 1588 1589
	return ret;
}

1590 1591
static int f2fs_release_file(struct inode *inode, struct file *filp)
{
1592 1593 1594 1595 1596 1597 1598 1599
	/*
	 * f2fs_relase_file is called at every close calls. So we should
	 * not drop any inmemory pages by close called by other process.
	 */
	if (!(filp->f_mode & FMODE_WRITE) ||
			atomic_read(&inode->i_writecount) != 1)
		return 0;

1600 1601
	/* some remained atomic pages should discarded */
	if (f2fs_is_atomic_file(inode))
C
Chao Yu 已提交
1602
		f2fs_drop_inmem_pages(inode);
1603
	if (f2fs_is_volatile_file(inode)) {
1604
		set_inode_flag(inode, FI_DROP_CACHE);
1605
		filemap_fdatawrite(inode->i_mapping);
1606
		clear_inode_flag(inode, FI_DROP_CACHE);
1607 1608
		clear_inode_flag(inode, FI_VOLATILE_FILE);
		stat_dec_volatile_write(inode);
1609 1610 1611 1612
	}
	return 0;
}

1613
static int f2fs_file_flush(struct file *file, fl_owner_t id)
J
Jaegeuk Kim 已提交
1614
{
1615 1616 1617 1618 1619 1620 1621 1622 1623 1624
	struct inode *inode = file_inode(file);

	/*
	 * If the process doing a transaction is crashed, we should do
	 * roll-back. Otherwise, other reader/write can see corrupted database
	 * until all the writers close its file. Since this should be done
	 * before dropping file lock, it needs to do in ->flush.
	 */
	if (f2fs_is_atomic_file(inode) &&
			F2FS_I(inode)->inmem_task == current)
C
Chao Yu 已提交
1625
		f2fs_drop_inmem_pages(inode);
1626
	return 0;
J
Jaegeuk Kim 已提交
1627 1628
}

1629
static int f2fs_ioc_getflags(struct file *filp, unsigned long arg)
J
Jaegeuk Kim 已提交
1630
{
A
Al Viro 已提交
1631
	struct inode *inode = file_inode(filp);
J
Jaegeuk Kim 已提交
1632
	struct f2fs_inode_info *fi = F2FS_I(inode);
1633 1634
	unsigned int flags = fi->i_flags;

1635
	if (f2fs_encrypted_inode(inode))
1636 1637 1638 1639 1640 1641
		flags |= F2FS_ENCRYPT_FL;
	if (f2fs_has_inline_data(inode) || f2fs_has_inline_dentry(inode))
		flags |= F2FS_INLINE_DATA_FL;

	flags &= F2FS_FL_USER_VISIBLE;

1642 1643
	return put_user(flags, (int __user *)arg);
}
J
Jaegeuk Kim 已提交
1644

1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657
static int __f2fs_ioc_setflags(struct inode *inode, unsigned int flags)
{
	struct f2fs_inode_info *fi = F2FS_I(inode);
	unsigned int oldflags;

	/* Is it quota file? Do not allow user to mess with it */
	if (IS_NOQUOTA(inode))
		return -EPERM;

	flags = f2fs_mask_flags(inode->i_mode, flags);

	oldflags = fi->i_flags;

1658
	if ((flags ^ oldflags) & (F2FS_APPEND_FL | F2FS_IMMUTABLE_FL))
1659 1660 1661
		if (!capable(CAP_LINUX_IMMUTABLE))
			return -EPERM;

1662 1663
	flags = flags & F2FS_FL_USER_MODIFIABLE;
	flags |= oldflags & ~F2FS_FL_USER_MODIFIABLE;
1664 1665
	fi->i_flags = flags;

1666
	if (fi->i_flags & F2FS_PROJINHERIT_FL)
1667 1668 1669 1670 1671 1672 1673 1674 1675 1676
		set_inode_flag(inode, FI_PROJ_INHERIT);
	else
		clear_inode_flag(inode, FI_PROJ_INHERIT);

	inode->i_ctime = current_time(inode);
	f2fs_set_inode_flags(inode);
	f2fs_mark_inode_dirty_sync(inode, false);
	return 0;
}

1677 1678 1679
static int f2fs_ioc_setflags(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
1680
	unsigned int flags;
1681
	int ret;
J
Jaegeuk Kim 已提交
1682

1683 1684 1685 1686 1687 1688
	if (!inode_owner_or_capable(inode))
		return -EACCES;

	if (get_user(flags, (int __user *)arg))
		return -EFAULT;

1689 1690 1691
	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;
J
Jaegeuk Kim 已提交
1692

A
Al Viro 已提交
1693
	inode_lock(inode);
J
Jaegeuk Kim 已提交
1694

1695
	ret = __f2fs_ioc_setflags(inode, flags);
J
Jaegeuk Kim 已提交
1696

1697
	inode_unlock(inode);
1698 1699 1700
	mnt_drop_write_file(filp);
	return ret;
}
1701

C
Chao Yu 已提交
1702 1703 1704 1705 1706 1707 1708
static int f2fs_ioc_getversion(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);

	return put_user(inode->i_generation, (int __user *)arg);
}

J
Jaegeuk Kim 已提交
1709 1710 1711
static int f2fs_ioc_start_atomic_write(struct file *filp)
{
	struct inode *inode = file_inode(filp);
1712
	int ret;
J
Jaegeuk Kim 已提交
1713 1714 1715 1716

	if (!inode_owner_or_capable(inode))
		return -EACCES;

1717 1718 1719
	if (!S_ISREG(inode->i_mode))
		return -EINVAL;

1720 1721 1722 1723
	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

1724 1725
	inode_lock(inode);

C
Chao Yu 已提交
1726
	down_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
1727

1728 1729 1730
	if (f2fs_is_atomic_file(inode)) {
		if (is_inode_flag_set(inode, FI_ATOMIC_REVOKE_REQUEST))
			ret = -EINVAL;
1731
		goto out;
1732
	}
J
Jaegeuk Kim 已提交
1733

1734 1735
	ret = f2fs_convert_inline_inode(inode);
	if (ret)
1736
		goto out;
J
Jaegeuk Kim 已提交
1737

1738
	if (!get_dirty_pages(inode))
1739
		goto skip_flush;
1740 1741

	f2fs_msg(F2FS_I_SB(inode)->sb, KERN_WARNING,
1742
		"Unexpected flush for atomic writes: ino=%lu, npages=%u",
1743 1744
					inode->i_ino, get_dirty_pages(inode));
	ret = filemap_write_and_wait_range(inode->i_mapping, 0, LLONG_MAX);
1745
	if (ret)
1746
		goto out;
1747 1748
skip_flush:
	set_inode_flag(inode, FI_ATOMIC_FILE);
1749
	clear_inode_flag(inode, FI_ATOMIC_REVOKE_REQUEST);
1750
	f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
1751

1752
	F2FS_I(inode)->inmem_task = current;
1753 1754
	stat_inc_atomic_write(inode);
	stat_update_max_atomic_write(inode);
1755
out:
C
Chao Yu 已提交
1756
	up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
1757
	inode_unlock(inode);
1758
	mnt_drop_write_file(filp);
1759
	return ret;
J
Jaegeuk Kim 已提交
1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773
}

static int f2fs_ioc_commit_atomic_write(struct file *filp)
{
	struct inode *inode = file_inode(filp);
	int ret;

	if (!inode_owner_or_capable(inode))
		return -EACCES;

	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

1774 1775
	inode_lock(inode);

C
Chao Yu 已提交
1776
	down_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
1777

1778 1779
	if (f2fs_is_volatile_file(inode)) {
		ret = -EINVAL;
1780
		goto err_out;
1781
	}
1782

1783
	if (f2fs_is_atomic_file(inode)) {
C
Chao Yu 已提交
1784
		ret = f2fs_commit_inmem_pages(inode);
C
Chao Yu 已提交
1785
		if (ret)
1786
			goto err_out;
C
Chao Yu 已提交
1787

1788
		ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 0, true);
C
Chao Yu 已提交
1789 1790
		if (!ret) {
			clear_inode_flag(inode, FI_ATOMIC_FILE);
1791
			F2FS_I(inode)->i_gc_failures[GC_FAILURE_ATOMIC] = 0;
C
Chao Yu 已提交
1792
			stat_dec_atomic_write(inode);
1793
		}
1794
	} else {
1795
		ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 1, false);
1796
	}
1797
err_out:
1798 1799 1800 1801
	if (is_inode_flag_set(inode, FI_ATOMIC_REVOKE_REQUEST)) {
		clear_inode_flag(inode, FI_ATOMIC_REVOKE_REQUEST);
		ret = -EINVAL;
	}
C
Chao Yu 已提交
1802
	up_write(&F2FS_I(inode)->i_gc_rwsem[WRITE]);
1803
	inode_unlock(inode);
J
Jaegeuk Kim 已提交
1804 1805 1806 1807
	mnt_drop_write_file(filp);
	return ret;
}

1808 1809 1810
static int f2fs_ioc_start_volatile_write(struct file *filp)
{
	struct inode *inode = file_inode(filp);
1811
	int ret;
1812 1813 1814 1815

	if (!inode_owner_or_capable(inode))
		return -EACCES;

1816 1817 1818
	if (!S_ISREG(inode->i_mode))
		return -EINVAL;

1819 1820 1821 1822
	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

1823 1824
	inode_lock(inode);

1825
	if (f2fs_is_volatile_file(inode))
1826
		goto out;
1827

1828 1829
	ret = f2fs_convert_inline_inode(inode);
	if (ret)
1830
		goto out;
1831

1832 1833 1834
	stat_inc_volatile_write(inode);
	stat_update_max_volatile_write(inode);

1835
	set_inode_flag(inode, FI_VOLATILE_FILE);
1836
	f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
1837
out:
1838
	inode_unlock(inode);
1839 1840
	mnt_drop_write_file(filp);
	return ret;
1841 1842
}

1843 1844 1845
static int f2fs_ioc_release_volatile_write(struct file *filp)
{
	struct inode *inode = file_inode(filp);
1846
	int ret;
1847 1848 1849 1850

	if (!inode_owner_or_capable(inode))
		return -EACCES;

1851 1852 1853 1854
	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

1855 1856
	inode_lock(inode);

1857
	if (!f2fs_is_volatile_file(inode))
1858
		goto out;
1859

1860 1861 1862 1863
	if (!f2fs_is_first_block_written(inode)) {
		ret = truncate_partial_data_page(inode, 0, true);
		goto out;
	}
1864

1865 1866
	ret = punch_hole(inode, 0, F2FS_BLKSIZE);
out:
1867
	inode_unlock(inode);
1868 1869
	mnt_drop_write_file(filp);
	return ret;
1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883
}

static int f2fs_ioc_abort_volatile_write(struct file *filp)
{
	struct inode *inode = file_inode(filp);
	int ret;

	if (!inode_owner_or_capable(inode))
		return -EACCES;

	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

1884 1885
	inode_lock(inode);

1886
	if (f2fs_is_atomic_file(inode))
C
Chao Yu 已提交
1887
		f2fs_drop_inmem_pages(inode);
1888
	if (f2fs_is_volatile_file(inode)) {
1889
		clear_inode_flag(inode, FI_VOLATILE_FILE);
1890
		stat_dec_volatile_write(inode);
1891
		ret = f2fs_do_sync_file(filp, 0, LLONG_MAX, 0, true);
1892
	}
1893

1894 1895
	clear_inode_flag(inode, FI_ATOMIC_REVOKE_REQUEST);

1896 1897
	inode_unlock(inode);

1898
	mnt_drop_write_file(filp);
1899
	f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
1900 1901 1902
	return ret;
}

J
Jaegeuk Kim 已提交
1903 1904 1905 1906 1907 1908
static int f2fs_ioc_shutdown(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct super_block *sb = sbi->sb;
	__u32 in;
1909
	int ret = 0;
J
Jaegeuk Kim 已提交
1910 1911 1912 1913 1914 1915 1916

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	if (get_user(in, (__u32 __user *)arg))
		return -EFAULT;

1917 1918 1919 1920 1921
	if (in != F2FS_GOING_DOWN_FULLSYNC) {
		ret = mnt_want_write_file(filp);
		if (ret)
			return ret;
	}
1922

J
Jaegeuk Kim 已提交
1923 1924 1925
	switch (in) {
	case F2FS_GOING_DOWN_FULLSYNC:
		sb = freeze_bdev(sb->s_bdev);
1926 1927 1928 1929 1930
		if (IS_ERR(sb)) {
			ret = PTR_ERR(sb);
			goto out;
		}
		if (sb) {
1931
			f2fs_stop_checkpoint(sbi, false);
1932
			set_sbi_flag(sbi, SBI_IS_SHUTDOWN);
J
Jaegeuk Kim 已提交
1933 1934 1935 1936 1937
			thaw_bdev(sb->s_bdev, sb);
		}
		break;
	case F2FS_GOING_DOWN_METASYNC:
		/* do checkpoint only */
1938 1939 1940
		ret = f2fs_sync_fs(sb, 1);
		if (ret)
			goto out;
1941
		f2fs_stop_checkpoint(sbi, false);
1942
		set_sbi_flag(sbi, SBI_IS_SHUTDOWN);
J
Jaegeuk Kim 已提交
1943 1944
		break;
	case F2FS_GOING_DOWN_NOSYNC:
1945
		f2fs_stop_checkpoint(sbi, false);
1946
		set_sbi_flag(sbi, SBI_IS_SHUTDOWN);
J
Jaegeuk Kim 已提交
1947
		break;
1948
	case F2FS_GOING_DOWN_METAFLUSH:
C
Chao Yu 已提交
1949
		f2fs_sync_meta_pages(sbi, META, LONG_MAX, FS_META_IO);
1950
		f2fs_stop_checkpoint(sbi, false);
1951
		set_sbi_flag(sbi, SBI_IS_SHUTDOWN);
1952
		break;
J
Jaegeuk Kim 已提交
1953
	default:
1954 1955
		ret = -EINVAL;
		goto out;
J
Jaegeuk Kim 已提交
1956
	}
1957

C
Chao Yu 已提交
1958 1959
	f2fs_stop_gc_thread(sbi);
	f2fs_stop_discard_thread(sbi);
1960

C
Chao Yu 已提交
1961
	f2fs_drop_discard_cmd(sbi);
1962 1963
	clear_opt(sbi, DISCARD);

1964
	f2fs_update_time(sbi, REQ_TIME);
1965
out:
1966 1967
	if (in != F2FS_GOING_DOWN_FULLSYNC)
		mnt_drop_write_file(filp);
1968
	return ret;
J
Jaegeuk Kim 已提交
1969 1970
}

1971 1972 1973 1974 1975 1976 1977
static int f2fs_ioc_fitrim(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	struct super_block *sb = inode->i_sb;
	struct request_queue *q = bdev_get_queue(sb->s_bdev);
	struct fstrim_range range;
	int ret;
1978

1979 1980
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;
1981

1982 1983
	if (!blk_queue_discard(q))
		return -EOPNOTSUPP;
1984

1985 1986 1987
	if (copy_from_user(&range, (struct fstrim_range __user *)arg,
				sizeof(range)))
		return -EFAULT;
1988

1989 1990 1991 1992
	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

1993 1994 1995
	range.minlen = max((unsigned int)range.minlen,
				q->limits.discard_granularity);
	ret = f2fs_trim_fs(F2FS_SB(sb), &range);
1996
	mnt_drop_write_file(filp);
1997 1998
	if (ret < 0)
		return ret;
1999

2000 2001 2002
	if (copy_to_user((struct fstrim_range __user *)arg, &range,
				sizeof(range)))
		return -EFAULT;
2003
	f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
2004 2005 2006
	return 0;
}

2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
static bool uuid_is_nonzero(__u8 u[16])
{
	int i;

	for (i = 0; i < 16; i++)
		if (u[i])
			return true;
	return false;
}

static int f2fs_ioc_set_encryption_policy(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);

2021
	if (!f2fs_sb_has_encrypt(inode->i_sb))
2022 2023
		return -EOPNOTSUPP;

2024
	f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
2025

2026
	return fscrypt_ioctl_set_policy(filp, (const void __user *)arg);
2027 2028 2029 2030
}

static int f2fs_ioc_get_encryption_policy(struct file *filp, unsigned long arg)
{
2031
	if (!f2fs_sb_has_encrypt(file_inode(filp)->i_sb))
2032
		return -EOPNOTSUPP;
2033
	return fscrypt_ioctl_get_policy(filp, (void __user *)arg);
2034 2035 2036 2037 2038 2039 2040 2041
}

static int f2fs_ioc_get_encryption_pwsalt(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	int err;

2042
	if (!f2fs_sb_has_encrypt(inode->i_sb))
2043 2044 2045 2046 2047 2048
		return -EOPNOTSUPP;

	err = mnt_want_write_file(filp);
	if (err)
		return err;

2049
	down_write(&sbi->sb_lock);
2050 2051 2052 2053

	if (uuid_is_nonzero(sbi->raw_super->encrypt_pw_salt))
		goto got_it;

2054 2055 2056
	/* update superblock with uuid */
	generate_random_uuid(sbi->raw_super->encrypt_pw_salt);

C
Chao Yu 已提交
2057
	err = f2fs_commit_super(sbi, false);
2058 2059 2060
	if (err) {
		/* undo new data */
		memset(sbi->raw_super->encrypt_pw_salt, 0, 16);
2061
		goto out_err;
2062 2063 2064 2065
	}
got_it:
	if (copy_to_user((__u8 __user *)arg, sbi->raw_super->encrypt_pw_salt,
									16))
2066 2067
		err = -EFAULT;
out_err:
2068
	up_write(&sbi->sb_lock);
2069 2070
	mnt_drop_write_file(filp);
	return err;
2071 2072
}

2073 2074 2075 2076
static int f2fs_ioc_gc(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
C
Chao Yu 已提交
2077
	__u32 sync;
2078
	int ret;
2079 2080 2081 2082

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

C
Chao Yu 已提交
2083
	if (get_user(sync, (__u32 __user *)arg))
2084 2085
		return -EFAULT;

C
Chao Yu 已提交
2086 2087
	if (f2fs_readonly(sbi->sb))
		return -EROFS;
2088

2089 2090 2091 2092
	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

C
Chao Yu 已提交
2093
	if (!sync) {
2094 2095 2096 2097
		if (!mutex_trylock(&sbi->gc_mutex)) {
			ret = -EBUSY;
			goto out;
		}
C
Chao Yu 已提交
2098 2099
	} else {
		mutex_lock(&sbi->gc_mutex);
2100 2101
	}

2102
	ret = f2fs_gc(sbi, sync, true, NULL_SEGNO);
2103 2104 2105
out:
	mnt_drop_write_file(filp);
	return ret;
2106 2107
}

2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125
static int f2fs_ioc_gc_range(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct f2fs_gc_range range;
	u64 end;
	int ret;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	if (copy_from_user(&range, (struct f2fs_gc_range __user *)arg,
							sizeof(range)))
		return -EFAULT;

	if (f2fs_readonly(sbi->sb))
		return -EROFS;

2126 2127 2128 2129 2130
	end = range.start + range.len;
	if (range.start < MAIN_BLKADDR(sbi) || end >= MAX_BLKADDR(sbi)) {
		return -EINVAL;
	}

2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153
	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

do_more:
	if (!range.sync) {
		if (!mutex_trylock(&sbi->gc_mutex)) {
			ret = -EBUSY;
			goto out;
		}
	} else {
		mutex_lock(&sbi->gc_mutex);
	}

	ret = f2fs_gc(sbi, range.sync, true, GET_SEGNO(sbi, range.start));
	range.start += sbi->blocks_per_seg;
	if (range.start <= end)
		goto do_more;
out:
	mnt_drop_write_file(filp);
	return ret;
}

C
Chao Yu 已提交
2154
static int f2fs_ioc_write_checkpoint(struct file *filp, unsigned long arg)
2155 2156 2157
{
	struct inode *inode = file_inode(filp);
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
2158
	int ret;
2159 2160 2161 2162 2163 2164 2165

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	if (f2fs_readonly(sbi->sb))
		return -EROFS;

2166 2167 2168 2169 2170 2171 2172 2173
	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

	ret = f2fs_sync_fs(sbi->sb, 1);

	mnt_drop_write_file(filp);
	return ret;
2174 2175
}

C
Chao Yu 已提交
2176 2177 2178 2179 2180
static int f2fs_defragment_range(struct f2fs_sb_info *sbi,
					struct file *filp,
					struct f2fs_defragment *range)
{
	struct inode *inode = file_inode(filp);
2181 2182
	struct f2fs_map_blocks map = { .m_next_extent = NULL,
					.m_seg_type = NO_CHECK_TYPE };
2183
	struct extent_info ei = {0, 0, 0};
2184
	pgoff_t pg_start, pg_end, next_pgofs;
2185
	unsigned int blk_per_seg = sbi->blocks_per_seg;
C
Chao Yu 已提交
2186 2187 2188 2189 2190 2191
	unsigned int total = 0, sec_num;
	block_t blk_end = 0;
	bool fragmented = false;
	int err;

	/* if in-place-update policy is enabled, don't waste time here */
C
Chao Yu 已提交
2192
	if (f2fs_should_update_inplace(inode, NULL))
C
Chao Yu 已提交
2193 2194
		return -EINVAL;

2195 2196
	pg_start = range->start >> PAGE_SHIFT;
	pg_end = (range->start + range->len) >> PAGE_SHIFT;
C
Chao Yu 已提交
2197

J
Jaegeuk Kim 已提交
2198
	f2fs_balance_fs(sbi, true);
C
Chao Yu 已提交
2199

A
Al Viro 已提交
2200
	inode_lock(inode);
C
Chao Yu 已提交
2201 2202 2203

	/* writeback all dirty pages in the range */
	err = filemap_write_and_wait_range(inode->i_mapping, range->start,
2204
						range->start + range->len - 1);
C
Chao Yu 已提交
2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217
	if (err)
		goto out;

	/*
	 * lookup mapping info in extent cache, skip defragmenting if physical
	 * block addresses are continuous.
	 */
	if (f2fs_lookup_extent_cache(inode, pg_start, &ei)) {
		if (ei.fofs + ei.len >= pg_end)
			goto out;
	}

	map.m_lblk = pg_start;
2218
	map.m_next_pgofs = &next_pgofs;
C
Chao Yu 已提交
2219 2220 2221 2222 2223 2224 2225

	/*
	 * lookup mapping info in dnode page cache, skip defragmenting if all
	 * physical block addresses are continuous even if there are hole(s)
	 * in logical blocks.
	 */
	while (map.m_lblk < pg_end) {
F
Fan Li 已提交
2226
		map.m_len = pg_end - map.m_lblk;
2227
		err = f2fs_map_blocks(inode, &map, 0, F2FS_GET_BLOCK_DEFAULT);
C
Chao Yu 已提交
2228 2229 2230 2231
		if (err)
			goto out;

		if (!(map.m_flags & F2FS_MAP_FLAGS)) {
2232
			map.m_lblk = next_pgofs;
C
Chao Yu 已提交
2233 2234 2235
			continue;
		}

2236
		if (blk_end && blk_end != map.m_pblk)
C
Chao Yu 已提交
2237
			fragmented = true;
2238 2239 2240 2241

		/* record total count of block that we're going to move */
		total += map.m_len;

C
Chao Yu 已提交
2242 2243 2244 2245 2246 2247 2248 2249
		blk_end = map.m_pblk + map.m_len;

		map.m_lblk += map.m_len;
	}

	if (!fragmented)
		goto out;

2250
	sec_num = (total + BLKS_PER_SEC(sbi) - 1) / BLKS_PER_SEC(sbi);
C
Chao Yu 已提交
2251 2252 2253 2254 2255 2256

	/*
	 * make sure there are enough free section for LFS allocation, this can
	 * avoid defragment running in SSR mode when free section are allocated
	 * intensively
	 */
2257
	if (has_not_enough_free_secs(sbi, 0, sec_num)) {
C
Chao Yu 已提交
2258 2259 2260 2261
		err = -EAGAIN;
		goto out;
	}

2262 2263 2264 2265
	map.m_lblk = pg_start;
	map.m_len = pg_end - pg_start;
	total = 0;

C
Chao Yu 已提交
2266 2267 2268 2269 2270
	while (map.m_lblk < pg_end) {
		pgoff_t idx;
		int cnt = 0;

do_map:
F
Fan Li 已提交
2271
		map.m_len = pg_end - map.m_lblk;
2272
		err = f2fs_map_blocks(inode, &map, 0, F2FS_GET_BLOCK_DEFAULT);
C
Chao Yu 已提交
2273 2274 2275 2276
		if (err)
			goto clear_out;

		if (!(map.m_flags & F2FS_MAP_FLAGS)) {
2277
			map.m_lblk = next_pgofs;
C
Chao Yu 已提交
2278 2279 2280
			continue;
		}

2281
		set_inode_flag(inode, FI_DO_DEFRAG);
C
Chao Yu 已提交
2282 2283 2284 2285 2286

		idx = map.m_lblk;
		while (idx < map.m_lblk + map.m_len && cnt < blk_per_seg) {
			struct page *page;

C
Chao Yu 已提交
2287
			page = f2fs_get_lock_data_page(inode, idx, true);
C
Chao Yu 已提交
2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305
			if (IS_ERR(page)) {
				err = PTR_ERR(page);
				goto clear_out;
			}

			set_page_dirty(page);
			f2fs_put_page(page, 1);

			idx++;
			cnt++;
			total++;
		}

		map.m_lblk = idx;

		if (idx < pg_end && cnt < blk_per_seg)
			goto do_map;

2306
		clear_inode_flag(inode, FI_DO_DEFRAG);
C
Chao Yu 已提交
2307 2308 2309 2310 2311 2312

		err = filemap_fdatawrite(inode->i_mapping);
		if (err)
			goto out;
	}
clear_out:
2313
	clear_inode_flag(inode, FI_DO_DEFRAG);
C
Chao Yu 已提交
2314
out:
A
Al Viro 已提交
2315
	inode_unlock(inode);
C
Chao Yu 已提交
2316
	if (!err)
2317
		range->len = (u64)total << PAGE_SHIFT;
C
Chao Yu 已提交
2318 2319 2320 2321 2322 2323 2324 2325 2326 2327 2328 2329 2330
	return err;
}

static int f2fs_ioc_defragment(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct f2fs_defragment range;
	int err;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2331
	if (!S_ISREG(inode->i_mode) || f2fs_is_atomic_file(inode))
C
Chao Yu 已提交
2332 2333
		return -EINVAL;

2334 2335
	if (f2fs_readonly(sbi->sb))
		return -EROFS;
C
Chao Yu 已提交
2336 2337

	if (copy_from_user(&range, (struct f2fs_defragment __user *)arg,
2338 2339
							sizeof(range)))
		return -EFAULT;
C
Chao Yu 已提交
2340 2341

	/* verify alignment of offset & size */
2342 2343
	if (range.start & (F2FS_BLKSIZE - 1) || range.len & (F2FS_BLKSIZE - 1))
		return -EINVAL;
C
Chao Yu 已提交
2344

S
Sheng Yong 已提交
2345
	if (unlikely((range.start + range.len) >> PAGE_SHIFT >
2346 2347 2348 2349 2350 2351
					sbi->max_file_blocks))
		return -EINVAL;

	err = mnt_want_write_file(filp);
	if (err)
		return err;
S
Sheng Yong 已提交
2352

C
Chao Yu 已提交
2353
	err = f2fs_defragment_range(sbi, filp, &range);
2354 2355
	mnt_drop_write_file(filp);

2356
	f2fs_update_time(sbi, REQ_TIME);
C
Chao Yu 已提交
2357
	if (err < 0)
2358
		return err;
C
Chao Yu 已提交
2359 2360 2361

	if (copy_to_user((struct f2fs_defragment __user *)arg, &range,
							sizeof(range)))
2362 2363 2364
		return -EFAULT;

	return 0;
C
Chao Yu 已提交
2365 2366
}

2367 2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381 2382 2383
static int f2fs_move_file_range(struct file *file_in, loff_t pos_in,
			struct file *file_out, loff_t pos_out, size_t len)
{
	struct inode *src = file_inode(file_in);
	struct inode *dst = file_inode(file_out);
	struct f2fs_sb_info *sbi = F2FS_I_SB(src);
	size_t olen = len, dst_max_i_size = 0;
	size_t dst_osize;
	int ret;

	if (file_in->f_path.mnt != file_out->f_path.mnt ||
				src->i_sb != dst->i_sb)
		return -EXDEV;

	if (unlikely(f2fs_readonly(src->i_sb)))
		return -EROFS;

2384 2385
	if (!S_ISREG(src->i_mode) || !S_ISREG(dst->i_mode))
		return -EINVAL;
2386 2387 2388 2389

	if (f2fs_encrypted_inode(src) || f2fs_encrypted_inode(dst))
		return -EOPNOTSUPP;

2390 2391 2392 2393 2394 2395 2396
	if (src == dst) {
		if (pos_in == pos_out)
			return 0;
		if (pos_out > pos_in && pos_out < pos_in + len)
			return -EINVAL;
	}

2397
	inode_lock(src);
C
Chao Yu 已提交
2398
	down_write(&F2FS_I(src)->i_gc_rwsem[WRITE]);
2399
	if (src != dst) {
2400 2401 2402
		ret = -EBUSY;
		if (!inode_trylock(dst))
			goto out;
C
Chao Yu 已提交
2403
		if (!down_write_trylock(&F2FS_I(dst)->i_gc_rwsem[WRITE])) {
2404
			inode_unlock(dst);
2405 2406 2407
			goto out;
		}
	}
2408 2409 2410 2411 2412 2413 2414 2415 2416 2417 2418 2419 2420 2421 2422 2423 2424 2425 2426 2427 2428 2429 2430 2431 2432 2433 2434 2435 2436 2437 2438 2439 2440 2441 2442 2443 2444 2445 2446 2447 2448 2449 2450 2451

	ret = -EINVAL;
	if (pos_in + len > src->i_size || pos_in + len < pos_in)
		goto out_unlock;
	if (len == 0)
		olen = len = src->i_size - pos_in;
	if (pos_in + len == src->i_size)
		len = ALIGN(src->i_size, F2FS_BLKSIZE) - pos_in;
	if (len == 0) {
		ret = 0;
		goto out_unlock;
	}

	dst_osize = dst->i_size;
	if (pos_out + olen > dst->i_size)
		dst_max_i_size = pos_out + olen;

	/* verify the end result is block aligned */
	if (!IS_ALIGNED(pos_in, F2FS_BLKSIZE) ||
			!IS_ALIGNED(pos_in + len, F2FS_BLKSIZE) ||
			!IS_ALIGNED(pos_out, F2FS_BLKSIZE))
		goto out_unlock;

	ret = f2fs_convert_inline_inode(src);
	if (ret)
		goto out_unlock;

	ret = f2fs_convert_inline_inode(dst);
	if (ret)
		goto out_unlock;

	/* write out all dirty pages from offset */
	ret = filemap_write_and_wait_range(src->i_mapping,
					pos_in, pos_in + len);
	if (ret)
		goto out_unlock;

	ret = filemap_write_and_wait_range(dst->i_mapping,
					pos_out, pos_out + len);
	if (ret)
		goto out_unlock;

	f2fs_balance_fs(sbi, true);
	f2fs_lock_op(sbi);
2452 2453 2454
	ret = __exchange_data_block(src, dst, pos_in >> F2FS_BLKSIZE_BITS,
				pos_out >> F2FS_BLKSIZE_BITS,
				len >> F2FS_BLKSIZE_BITS, false);
2455 2456 2457 2458 2459 2460 2461 2462 2463

	if (!ret) {
		if (dst_max_i_size)
			f2fs_i_size_write(dst, dst_max_i_size);
		else if (dst_osize != dst->i_size)
			f2fs_i_size_write(dst, dst_osize);
	}
	f2fs_unlock_op(sbi);
out_unlock:
2464
	if (src != dst) {
C
Chao Yu 已提交
2465
		up_write(&F2FS_I(dst)->i_gc_rwsem[WRITE]);
2466
		inode_unlock(dst);
2467
	}
2468
out:
C
Chao Yu 已提交
2469
	up_write(&F2FS_I(src)->i_gc_rwsem[WRITE]);
2470 2471 2472 2473 2474 2475 2476 2477 2478 2479 2480 2481 2482 2483 2484 2485 2486 2487 2488 2489 2490 2491 2492 2493 2494 2495 2496 2497 2498 2499 2500 2501 2502 2503 2504
	inode_unlock(src);
	return ret;
}

static int f2fs_ioc_move_range(struct file *filp, unsigned long arg)
{
	struct f2fs_move_range range;
	struct fd dst;
	int err;

	if (!(filp->f_mode & FMODE_READ) ||
			!(filp->f_mode & FMODE_WRITE))
		return -EBADF;

	if (copy_from_user(&range, (struct f2fs_move_range __user *)arg,
							sizeof(range)))
		return -EFAULT;

	dst = fdget(range.dst_fd);
	if (!dst.file)
		return -EBADF;

	if (!(dst.file->f_mode & FMODE_WRITE)) {
		err = -EBADF;
		goto err_out;
	}

	err = mnt_want_write_file(filp);
	if (err)
		goto err_out;

	err = f2fs_move_file_range(filp, range.pos_in, dst.file,
					range.pos_out, range.len);

	mnt_drop_write_file(filp);
2505 2506
	if (err)
		goto err_out;
2507 2508 2509 2510 2511 2512 2513 2514 2515

	if (copy_to_user((struct f2fs_move_range __user *)arg,
						&range, sizeof(range)))
		err = -EFAULT;
err_out:
	fdput(dst);
	return err;
}

2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577
static int f2fs_ioc_flush_device(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct sit_info *sm = SIT_I(sbi);
	unsigned int start_segno = 0, end_segno = 0;
	unsigned int dev_start_segno = 0, dev_end_segno = 0;
	struct f2fs_flush_device range;
	int ret;

	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	if (f2fs_readonly(sbi->sb))
		return -EROFS;

	if (copy_from_user(&range, (struct f2fs_flush_device __user *)arg,
							sizeof(range)))
		return -EFAULT;

	if (sbi->s_ndevs <= 1 || sbi->s_ndevs - 1 <= range.dev_num ||
			sbi->segs_per_sec != 1) {
		f2fs_msg(sbi->sb, KERN_WARNING,
			"Can't flush %u in %d for segs_per_sec %u != 1\n",
				range.dev_num, sbi->s_ndevs,
				sbi->segs_per_sec);
		return -EINVAL;
	}

	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

	if (range.dev_num != 0)
		dev_start_segno = GET_SEGNO(sbi, FDEV(range.dev_num).start_blk);
	dev_end_segno = GET_SEGNO(sbi, FDEV(range.dev_num).end_blk);

	start_segno = sm->last_victim[FLUSH_DEVICE];
	if (start_segno < dev_start_segno || start_segno >= dev_end_segno)
		start_segno = dev_start_segno;
	end_segno = min(start_segno + range.segments, dev_end_segno);

	while (start_segno < end_segno) {
		if (!mutex_trylock(&sbi->gc_mutex)) {
			ret = -EBUSY;
			goto out;
		}
		sm->last_victim[GC_CB] = end_segno + 1;
		sm->last_victim[GC_GREEDY] = end_segno + 1;
		sm->last_victim[ALLOC_NEXT] = end_segno + 1;
		ret = f2fs_gc(sbi, true, true, start_segno);
		if (ret == -EAGAIN)
			ret = 0;
		else if (ret < 0)
			break;
		start_segno++;
	}
out:
	mnt_drop_write_file(filp);
	return ret;
}

2578 2579 2580 2581 2582 2583 2584 2585 2586 2587
static int f2fs_ioc_get_features(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	u32 sb_feature = le32_to_cpu(F2FS_I_SB(inode)->raw_super->feature);

	/* Must validate to set it with SQLite behavior in Android. */
	sb_feature |= F2FS_FEATURE_ATOMIC_WRITE;

	return put_user(sb_feature, (u32 __user *)arg);
}
2588

2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626
#ifdef CONFIG_QUOTA
static int f2fs_ioc_setproject(struct file *filp, __u32 projid)
{
	struct inode *inode = file_inode(filp);
	struct f2fs_inode_info *fi = F2FS_I(inode);
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);
	struct super_block *sb = sbi->sb;
	struct dquot *transfer_to[MAXQUOTAS] = {};
	struct page *ipage;
	kprojid_t kprojid;
	int err;

	if (!f2fs_sb_has_project_quota(sb)) {
		if (projid != F2FS_DEF_PROJID)
			return -EOPNOTSUPP;
		else
			return 0;
	}

	if (!f2fs_has_extra_attr(inode))
		return -EOPNOTSUPP;

	kprojid = make_kprojid(&init_user_ns, (projid_t)projid);

	if (projid_eq(kprojid, F2FS_I(inode)->i_projid))
		return 0;

	err = mnt_want_write_file(filp);
	if (err)
		return err;

	err = -EPERM;
	inode_lock(inode);

	/* Is it quota file? Do not allow user to mess with it */
	if (IS_NOQUOTA(inode))
		goto out_unlock;

C
Chao Yu 已提交
2627
	ipage = f2fs_get_node_page(sbi, inode->i_ino);
2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640
	if (IS_ERR(ipage)) {
		err = PTR_ERR(ipage);
		goto out_unlock;
	}

	if (!F2FS_FITS_IN_INODE(F2FS_INODE(ipage), fi->i_extra_isize,
								i_projid)) {
		err = -EOVERFLOW;
		f2fs_put_page(ipage, 1);
		goto out_unlock;
	}
	f2fs_put_page(ipage, 1);

2641 2642 2643
	err = dquot_initialize(inode);
	if (err)
		goto out_unlock;
2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675

	transfer_to[PRJQUOTA] = dqget(sb, make_kqid_projid(kprojid));
	if (!IS_ERR(transfer_to[PRJQUOTA])) {
		err = __dquot_transfer(inode, transfer_to);
		dqput(transfer_to[PRJQUOTA]);
		if (err)
			goto out_dirty;
	}

	F2FS_I(inode)->i_projid = kprojid;
	inode->i_ctime = current_time(inode);
out_dirty:
	f2fs_mark_inode_dirty_sync(inode, true);
out_unlock:
	inode_unlock(inode);
	mnt_drop_write_file(filp);
	return err;
}
#else
static int f2fs_ioc_setproject(struct file *filp, __u32 projid)
{
	if (projid != F2FS_DEF_PROJID)
		return -EOPNOTSUPP;
	return 0;
}
#endif

/* Transfer internal flags to xflags */
static inline __u32 f2fs_iflags_to_xflags(unsigned long iflags)
{
	__u32 xflags = 0;

2676
	if (iflags & F2FS_SYNC_FL)
2677
		xflags |= FS_XFLAG_SYNC;
2678
	if (iflags & F2FS_IMMUTABLE_FL)
2679
		xflags |= FS_XFLAG_IMMUTABLE;
2680
	if (iflags & F2FS_APPEND_FL)
2681
		xflags |= FS_XFLAG_APPEND;
2682
	if (iflags & F2FS_NODUMP_FL)
2683
		xflags |= FS_XFLAG_NODUMP;
2684
	if (iflags & F2FS_NOATIME_FL)
2685
		xflags |= FS_XFLAG_NOATIME;
2686
	if (iflags & F2FS_PROJINHERIT_FL)
2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700
		xflags |= FS_XFLAG_PROJINHERIT;
	return xflags;
}

#define F2FS_SUPPORTED_FS_XFLAGS (FS_XFLAG_SYNC | FS_XFLAG_IMMUTABLE | \
				  FS_XFLAG_APPEND | FS_XFLAG_NODUMP | \
				  FS_XFLAG_NOATIME | FS_XFLAG_PROJINHERIT)

/* Transfer xflags flags to internal */
static inline unsigned long f2fs_xflags_to_iflags(__u32 xflags)
{
	unsigned long iflags = 0;

	if (xflags & FS_XFLAG_SYNC)
2701
		iflags |= F2FS_SYNC_FL;
2702
	if (xflags & FS_XFLAG_IMMUTABLE)
2703
		iflags |= F2FS_IMMUTABLE_FL;
2704
	if (xflags & FS_XFLAG_APPEND)
2705
		iflags |= F2FS_APPEND_FL;
2706
	if (xflags & FS_XFLAG_NODUMP)
2707
		iflags |= F2FS_NODUMP_FL;
2708
	if (xflags & FS_XFLAG_NOATIME)
2709
		iflags |= F2FS_NOATIME_FL;
2710
	if (xflags & FS_XFLAG_PROJINHERIT)
2711
		iflags |= F2FS_PROJINHERIT_FL;
2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723

	return iflags;
}

static int f2fs_ioc_fsgetxattr(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	struct f2fs_inode_info *fi = F2FS_I(inode);
	struct fsxattr fa;

	memset(&fa, 0, sizeof(struct fsxattr));
	fa.fsx_xflags = f2fs_iflags_to_xflags(fi->i_flags &
2724
				F2FS_FL_USER_VISIBLE);
2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775

	if (f2fs_sb_has_project_quota(inode->i_sb))
		fa.fsx_projid = (__u32)from_kprojid(&init_user_ns,
							fi->i_projid);

	if (copy_to_user((struct fsxattr __user *)arg, &fa, sizeof(fa)))
		return -EFAULT;
	return 0;
}

static int f2fs_ioc_fssetxattr(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	struct f2fs_inode_info *fi = F2FS_I(inode);
	struct fsxattr fa;
	unsigned int flags;
	int err;

	if (copy_from_user(&fa, (struct fsxattr __user *)arg, sizeof(fa)))
		return -EFAULT;

	/* Make sure caller has proper permission */
	if (!inode_owner_or_capable(inode))
		return -EACCES;

	if (fa.fsx_xflags & ~F2FS_SUPPORTED_FS_XFLAGS)
		return -EOPNOTSUPP;

	flags = f2fs_xflags_to_iflags(fa.fsx_xflags);
	if (f2fs_mask_flags(inode->i_mode, flags) != flags)
		return -EOPNOTSUPP;

	err = mnt_want_write_file(filp);
	if (err)
		return err;

	inode_lock(inode);
	flags = (fi->i_flags & ~F2FS_FL_XFLAG_VISIBLE) |
				(flags & F2FS_FL_XFLAG_VISIBLE);
	err = __f2fs_ioc_setflags(inode, flags);
	inode_unlock(inode);
	mnt_drop_write_file(filp);
	if (err)
		return err;

	err = f2fs_ioc_setproject(filp, fa.fsx_projid);
	if (err)
		return err;

	return 0;
}
2776

2777 2778 2779 2780 2781 2782 2783
int f2fs_pin_file_control(struct inode *inode, bool inc)
{
	struct f2fs_inode_info *fi = F2FS_I(inode);
	struct f2fs_sb_info *sbi = F2FS_I_SB(inode);

	/* Use i_gc_failures for normal file as a risk signal. */
	if (inc)
2784 2785
		f2fs_i_gc_failures_write(inode,
				fi->i_gc_failures[GC_FAILURE_PIN] + 1);
2786

2787
	if (fi->i_gc_failures[GC_FAILURE_PIN] > sbi->gc_pin_file_threshold) {
2788 2789
		f2fs_msg(sbi->sb, KERN_WARNING,
			"%s: Enable GC = ino %lx after %x GC trials\n",
2790 2791
			__func__, inode->i_ino,
			fi->i_gc_failures[GC_FAILURE_PIN]);
2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821
		clear_inode_flag(inode, FI_PIN_FILE);
		return -EAGAIN;
	}
	return 0;
}

static int f2fs_ioc_set_pin_file(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	__u32 pin;
	int ret = 0;

	if (!inode_owner_or_capable(inode))
		return -EACCES;

	if (get_user(pin, (__u32 __user *)arg))
		return -EFAULT;

	if (!S_ISREG(inode->i_mode))
		return -EINVAL;

	if (f2fs_readonly(F2FS_I_SB(inode)->sb))
		return -EROFS;

	ret = mnt_want_write_file(filp);
	if (ret)
		return ret;

	inode_lock(inode);

C
Chao Yu 已提交
2822
	if (f2fs_should_update_outplace(inode, NULL)) {
C
Chao Yu 已提交
2823 2824 2825 2826
		ret = -EINVAL;
		goto out;
	}

2827 2828
	if (!pin) {
		clear_inode_flag(inode, FI_PIN_FILE);
2829
		f2fs_i_gc_failures_write(inode, 0);
2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841
		goto done;
	}

	if (f2fs_pin_file_control(inode, false)) {
		ret = -EAGAIN;
		goto out;
	}
	ret = f2fs_convert_inline_inode(inode);
	if (ret)
		goto out;

	set_inode_flag(inode, FI_PIN_FILE);
2842
	ret = F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN];
2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856
done:
	f2fs_update_time(F2FS_I_SB(inode), REQ_TIME);
out:
	inode_unlock(inode);
	mnt_drop_write_file(filp);
	return ret;
}

static int f2fs_ioc_get_pin_file(struct file *filp, unsigned long arg)
{
	struct inode *inode = file_inode(filp);
	__u32 pin = 0;

	if (is_inode_flag_set(inode, FI_PIN_FILE))
2857
		pin = F2FS_I(inode)->i_gc_failures[GC_FAILURE_PIN];
2858 2859 2860
	return put_user(pin, (u32 __user *)arg);
}

2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880
int f2fs_precache_extents(struct inode *inode)
{
	struct f2fs_inode_info *fi = F2FS_I(inode);
	struct f2fs_map_blocks map;
	pgoff_t m_next_extent;
	loff_t end;
	int err;

	if (is_inode_flag_set(inode, FI_NO_EXTENT))
		return -EOPNOTSUPP;

	map.m_lblk = 0;
	map.m_next_pgofs = NULL;
	map.m_next_extent = &m_next_extent;
	map.m_seg_type = NO_CHECK_TYPE;
	end = F2FS_I_SB(inode)->max_file_blocks;

	while (map.m_lblk < end) {
		map.m_len = end - map.m_lblk;

C
Chao Yu 已提交
2881
		down_write(&fi->i_gc_rwsem[WRITE]);
2882
		err = f2fs_map_blocks(inode, &map, 0, F2FS_GET_BLOCK_PRECACHE);
C
Chao Yu 已提交
2883
		up_write(&fi->i_gc_rwsem[WRITE]);
2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897
		if (err)
			return err;

		map.m_lblk = m_next_extent;
	}

	return err;
}

static int f2fs_ioc_precache_extents(struct file *filp, unsigned long arg)
{
	return f2fs_precache_extents(file_inode(filp));
}

2898 2899
long f2fs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
{
2900 2901 2902
	if (unlikely(f2fs_cp_error(F2FS_I_SB(file_inode(filp)))))
		return -EIO;

2903 2904 2905 2906 2907
	switch (cmd) {
	case F2FS_IOC_GETFLAGS:
		return f2fs_ioc_getflags(filp, arg);
	case F2FS_IOC_SETFLAGS:
		return f2fs_ioc_setflags(filp, arg);
C
Chao Yu 已提交
2908 2909
	case F2FS_IOC_GETVERSION:
		return f2fs_ioc_getversion(filp, arg);
J
Jaegeuk Kim 已提交
2910 2911 2912 2913
	case F2FS_IOC_START_ATOMIC_WRITE:
		return f2fs_ioc_start_atomic_write(filp);
	case F2FS_IOC_COMMIT_ATOMIC_WRITE:
		return f2fs_ioc_commit_atomic_write(filp);
2914 2915
	case F2FS_IOC_START_VOLATILE_WRITE:
		return f2fs_ioc_start_volatile_write(filp);
2916 2917 2918 2919
	case F2FS_IOC_RELEASE_VOLATILE_WRITE:
		return f2fs_ioc_release_volatile_write(filp);
	case F2FS_IOC_ABORT_VOLATILE_WRITE:
		return f2fs_ioc_abort_volatile_write(filp);
J
Jaegeuk Kim 已提交
2920 2921
	case F2FS_IOC_SHUTDOWN:
		return f2fs_ioc_shutdown(filp, arg);
2922 2923
	case FITRIM:
		return f2fs_ioc_fitrim(filp, arg);
2924 2925 2926 2927 2928 2929
	case F2FS_IOC_SET_ENCRYPTION_POLICY:
		return f2fs_ioc_set_encryption_policy(filp, arg);
	case F2FS_IOC_GET_ENCRYPTION_POLICY:
		return f2fs_ioc_get_encryption_policy(filp, arg);
	case F2FS_IOC_GET_ENCRYPTION_PWSALT:
		return f2fs_ioc_get_encryption_pwsalt(filp, arg);
2930 2931
	case F2FS_IOC_GARBAGE_COLLECT:
		return f2fs_ioc_gc(filp, arg);
2932 2933
	case F2FS_IOC_GARBAGE_COLLECT_RANGE:
		return f2fs_ioc_gc_range(filp, arg);
2934
	case F2FS_IOC_WRITE_CHECKPOINT:
C
Chao Yu 已提交
2935
		return f2fs_ioc_write_checkpoint(filp, arg);
C
Chao Yu 已提交
2936 2937
	case F2FS_IOC_DEFRAGMENT:
		return f2fs_ioc_defragment(filp, arg);
2938 2939
	case F2FS_IOC_MOVE_RANGE:
		return f2fs_ioc_move_range(filp, arg);
2940 2941
	case F2FS_IOC_FLUSH_DEVICE:
		return f2fs_ioc_flush_device(filp, arg);
2942 2943
	case F2FS_IOC_GET_FEATURES:
		return f2fs_ioc_get_features(filp, arg);
2944 2945 2946 2947
	case F2FS_IOC_FSGETXATTR:
		return f2fs_ioc_fsgetxattr(filp, arg);
	case F2FS_IOC_FSSETXATTR:
		return f2fs_ioc_fssetxattr(filp, arg);
2948 2949 2950 2951
	case F2FS_IOC_GET_PIN_FILE:
		return f2fs_ioc_get_pin_file(filp, arg);
	case F2FS_IOC_SET_PIN_FILE:
		return f2fs_ioc_set_pin_file(filp, arg);
2952 2953
	case F2FS_IOC_PRECACHE_EXTENTS:
		return f2fs_ioc_precache_extents(filp, arg);
J
Jaegeuk Kim 已提交
2954 2955 2956 2957 2958
	default:
		return -ENOTTY;
	}
}

2959 2960
static ssize_t f2fs_file_write_iter(struct kiocb *iocb, struct iov_iter *from)
{
2961 2962 2963
	struct file *file = iocb->ki_filp;
	struct inode *inode = file_inode(file);
	ssize_t ret;
2964

2965 2966 2967
	if (unlikely(f2fs_cp_error(F2FS_I_SB(inode))))
		return -EIO;

H
Hyunchul Lee 已提交
2968 2969 2970 2971 2972 2973 2974 2975 2976
	if ((iocb->ki_flags & IOCB_NOWAIT) && !(iocb->ki_flags & IOCB_DIRECT))
		return -EINVAL;

	if (!inode_trylock(inode)) {
		if (iocb->ki_flags & IOCB_NOWAIT)
			return -EAGAIN;
		inode_lock(inode);
	}

2977 2978
	ret = generic_write_checks(iocb, from);
	if (ret > 0) {
2979 2980
		bool preallocated = false;
		size_t target_size = 0;
2981
		int err;
2982

2983 2984
		if (iov_iter_fault_in_readable(from, iov_iter_count(from)))
			set_inode_flag(inode, FI_NO_PREALLOC);
2985

H
Hyunchul Lee 已提交
2986 2987 2988 2989 2990 2991
		if ((iocb->ki_flags & IOCB_NOWAIT) &&
			(iocb->ki_flags & IOCB_DIRECT)) {
				if (!f2fs_overwrite_io(inode, iocb->ki_pos,
						iov_iter_count(from)) ||
					f2fs_has_inline_data(inode) ||
					f2fs_force_buffered_io(inode, WRITE)) {
2992 2993
						clear_inode_flag(inode,
								FI_NO_PREALLOC);
H
Hyunchul Lee 已提交
2994 2995 2996 2997 2998
						inode_unlock(inode);
						return -EAGAIN;
				}

		} else {
2999 3000 3001
			preallocated = true;
			target_size = iocb->ki_pos + iov_iter_count(from);

H
Hyunchul Lee 已提交
3002 3003 3004 3005 3006 3007
			err = f2fs_preallocate_blocks(iocb, from);
			if (err) {
				clear_inode_flag(inode, FI_NO_PREALLOC);
				inode_unlock(inode);
				return err;
			}
3008
		}
3009
		ret = __generic_file_write_iter(iocb, from);
3010
		clear_inode_flag(inode, FI_NO_PREALLOC);
C
Chao Yu 已提交
3011

3012 3013 3014 3015
		/* if we couldn't write data, we should deallocate blocks. */
		if (preallocated && i_size_read(inode) < target_size)
			f2fs_truncate(inode);

C
Chao Yu 已提交
3016 3017
		if (ret > 0)
			f2fs_update_iostat(F2FS_I_SB(inode), APP_WRITE_IO, ret);
3018 3019 3020
	}
	inode_unlock(inode);

3021 3022
	if (ret > 0)
		ret = generic_write_sync(iocb, ret);
3023
	return ret;
3024 3025
}

3026 3027 3028 3029 3030 3031 3032 3033 3034 3035
#ifdef CONFIG_COMPAT
long f2fs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
	switch (cmd) {
	case F2FS_IOC32_GETFLAGS:
		cmd = F2FS_IOC_GETFLAGS;
		break;
	case F2FS_IOC32_SETFLAGS:
		cmd = F2FS_IOC_SETFLAGS;
		break;
3036 3037 3038 3039 3040 3041 3042 3043 3044 3045 3046 3047 3048
	case F2FS_IOC32_GETVERSION:
		cmd = F2FS_IOC_GETVERSION;
		break;
	case F2FS_IOC_START_ATOMIC_WRITE:
	case F2FS_IOC_COMMIT_ATOMIC_WRITE:
	case F2FS_IOC_START_VOLATILE_WRITE:
	case F2FS_IOC_RELEASE_VOLATILE_WRITE:
	case F2FS_IOC_ABORT_VOLATILE_WRITE:
	case F2FS_IOC_SHUTDOWN:
	case F2FS_IOC_SET_ENCRYPTION_POLICY:
	case F2FS_IOC_GET_ENCRYPTION_PWSALT:
	case F2FS_IOC_GET_ENCRYPTION_POLICY:
	case F2FS_IOC_GARBAGE_COLLECT:
3049
	case F2FS_IOC_GARBAGE_COLLECT_RANGE:
3050 3051
	case F2FS_IOC_WRITE_CHECKPOINT:
	case F2FS_IOC_DEFRAGMENT:
3052
	case F2FS_IOC_MOVE_RANGE:
3053
	case F2FS_IOC_FLUSH_DEVICE:
3054
	case F2FS_IOC_GET_FEATURES:
3055 3056
	case F2FS_IOC_FSGETXATTR:
	case F2FS_IOC_FSSETXATTR:
3057 3058
	case F2FS_IOC_GET_PIN_FILE:
	case F2FS_IOC_SET_PIN_FILE:
3059
	case F2FS_IOC_PRECACHE_EXTENTS:
3060
		break;
3061 3062 3063 3064 3065 3066 3067
	default:
		return -ENOIOCTLCMD;
	}
	return f2fs_ioctl(file, cmd, (unsigned long) compat_ptr(arg));
}
#endif

J
Jaegeuk Kim 已提交
3068
const struct file_operations f2fs_file_operations = {
3069
	.llseek		= f2fs_llseek,
3070
	.read_iter	= generic_file_read_iter,
3071 3072
	.write_iter	= f2fs_file_write_iter,
	.open		= f2fs_file_open,
3073
	.release	= f2fs_release_file,
J
Jaegeuk Kim 已提交
3074
	.mmap		= f2fs_file_mmap,
3075
	.flush		= f2fs_file_flush,
J
Jaegeuk Kim 已提交
3076 3077 3078
	.fsync		= f2fs_sync_file,
	.fallocate	= f2fs_fallocate,
	.unlocked_ioctl	= f2fs_ioctl,
3079 3080 3081
#ifdef CONFIG_COMPAT
	.compat_ioctl	= f2fs_compat_ioctl,
#endif
J
Jaegeuk Kim 已提交
3082
	.splice_read	= generic_file_splice_read,
A
Al Viro 已提交
3083
	.splice_write	= iter_file_splice_write,
J
Jaegeuk Kim 已提交
3084
};