ioctl.c 137.9 KB
Newer Older
C
Christoph Hellwig 已提交
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
/*
 * Copyright (C) 2007 Oracle.  All rights reserved.
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public
 * License v2 as published by the Free Software Foundation.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * General Public License for more details.
 *
 * You should have received a copy of the GNU General Public
 * License along with this program; if not, write to the
 * Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 * Boston, MA 021110-1307, USA.
 */

#include <linux/kernel.h>
#include <linux/bio.h>
#include <linux/buffer_head.h>
#include <linux/file.h>
#include <linux/fs.h>
24
#include <linux/fsnotify.h>
C
Christoph Hellwig 已提交
25 26 27 28 29 30
#include <linux/pagemap.h>
#include <linux/highmem.h>
#include <linux/time.h>
#include <linux/init.h>
#include <linux/string.h>
#include <linux/backing-dev.h>
31
#include <linux/mount.h>
C
Christoph Hellwig 已提交
32
#include <linux/mpage.h>
33
#include <linux/namei.h>
C
Christoph Hellwig 已提交
34 35 36 37
#include <linux/swap.h>
#include <linux/writeback.h>
#include <linux/compat.h>
#include <linux/bit_spinlock.h>
38
#include <linux/security.h>
C
Christoph Hellwig 已提交
39
#include <linux/xattr.h>
40
#include <linux/mm.h>
41
#include <linux/slab.h>
42
#include <linux/blkdev.h>
43
#include <linux/uuid.h>
44
#include <linux/btrfs.h>
M
Mark Fasheh 已提交
45
#include <linux/uaccess.h>
C
Christoph Hellwig 已提交
46 47 48 49 50 51
#include "ctree.h"
#include "disk-io.h"
#include "transaction.h"
#include "btrfs_inode.h"
#include "print-tree.h"
#include "volumes.h"
52
#include "locking.h"
53
#include "inode-map.h"
54
#include "backref.h"
55
#include "rcu-string.h"
56
#include "send.h"
57
#include "dev-replace.h"
58
#include "props.h"
59
#include "sysfs.h"
J
Josef Bacik 已提交
60
#include "qgroup.h"
61
#include "tree-log.h"
62
#include "compression.h"
C
Christoph Hellwig 已提交
63

64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88
#ifdef CONFIG_64BIT
/* If we have a 32-bit userspace and 64-bit kernel, then the UAPI
 * structures are incorrect, as the timespec structure from userspace
 * is 4 bytes too small. We define these alternatives here to teach
 * the kernel about the 32-bit struct packing.
 */
struct btrfs_ioctl_timespec_32 {
	__u64 sec;
	__u32 nsec;
} __attribute__ ((__packed__));

struct btrfs_ioctl_received_subvol_args_32 {
	char	uuid[BTRFS_UUID_SIZE];	/* in */
	__u64	stransid;		/* in */
	__u64	rtransid;		/* out */
	struct btrfs_ioctl_timespec_32 stime; /* in */
	struct btrfs_ioctl_timespec_32 rtime; /* out */
	__u64	flags;			/* in */
	__u64	reserved[16];		/* in */
} __attribute__ ((__packed__));

#define BTRFS_IOC_SET_RECEIVED_SUBVOL_32 _IOWR(BTRFS_IOCTL_MAGIC, 37, \
				struct btrfs_ioctl_received_subvol_args_32)
#endif

89 90 91 92 93 94 95 96 97 98 99 100 101
#if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
struct btrfs_ioctl_send_args_32 {
	__s64 send_fd;			/* in */
	__u64 clone_sources_count;	/* in */
	compat_uptr_t clone_sources;	/* in */
	__u64 parent_root;		/* in */
	__u64 flags;			/* in */
	__u64 reserved[4];		/* in */
} __attribute__ ((__packed__));

#define BTRFS_IOC_SEND_32 _IOW(BTRFS_IOCTL_MAGIC, 38, \
			       struct btrfs_ioctl_send_args_32)
#endif
102

M
Mark Fasheh 已提交
103
static int btrfs_clone(struct inode *src, struct inode *inode,
104 105
		       u64 off, u64 olen, u64 olen_aligned, u64 destoff,
		       int no_time_update);
M
Mark Fasheh 已提交
106

107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136
/* Mask out flags that are inappropriate for the given type of inode. */
static inline __u32 btrfs_mask_flags(umode_t mode, __u32 flags)
{
	if (S_ISDIR(mode))
		return flags;
	else if (S_ISREG(mode))
		return flags & ~FS_DIRSYNC_FL;
	else
		return flags & (FS_NODUMP_FL | FS_NOATIME_FL);
}

/*
 * Export inode flags to the format expected by the FS_IOC_GETFLAGS ioctl.
 */
static unsigned int btrfs_flags_to_ioctl(unsigned int flags)
{
	unsigned int iflags = 0;

	if (flags & BTRFS_INODE_SYNC)
		iflags |= FS_SYNC_FL;
	if (flags & BTRFS_INODE_IMMUTABLE)
		iflags |= FS_IMMUTABLE_FL;
	if (flags & BTRFS_INODE_APPEND)
		iflags |= FS_APPEND_FL;
	if (flags & BTRFS_INODE_NODUMP)
		iflags |= FS_NODUMP_FL;
	if (flags & BTRFS_INODE_NOATIME)
		iflags |= FS_NOATIME_FL;
	if (flags & BTRFS_INODE_DIRSYNC)
		iflags |= FS_DIRSYNC_FL;
L
Li Zefan 已提交
137 138 139
	if (flags & BTRFS_INODE_NODATACOW)
		iflags |= FS_NOCOW_FL;

140
	if (flags & BTRFS_INODE_NOCOMPRESS)
L
Li Zefan 已提交
141
		iflags |= FS_NOCOMP_FL;
142 143
	else if (flags & BTRFS_INODE_COMPRESS)
		iflags |= FS_COMPR_FL;
144 145 146 147 148 149 150 151 152 153

	return iflags;
}

/*
 * Update inode->i_flags based on the btrfs internal flags.
 */
void btrfs_update_iflags(struct inode *inode)
{
	struct btrfs_inode *ip = BTRFS_I(inode);
154
	unsigned int new_fl = 0;
155 156

	if (ip->flags & BTRFS_INODE_SYNC)
157
		new_fl |= S_SYNC;
158
	if (ip->flags & BTRFS_INODE_IMMUTABLE)
159
		new_fl |= S_IMMUTABLE;
160
	if (ip->flags & BTRFS_INODE_APPEND)
161
		new_fl |= S_APPEND;
162
	if (ip->flags & BTRFS_INODE_NOATIME)
163
		new_fl |= S_NOATIME;
164
	if (ip->flags & BTRFS_INODE_DIRSYNC)
165 166 167 168 169
		new_fl |= S_DIRSYNC;

	set_mask_bits(&inode->i_flags,
		      S_SYNC | S_APPEND | S_IMMUTABLE | S_NOATIME | S_DIRSYNC,
		      new_fl);
170 171 172 173
}

static int btrfs_ioctl_getflags(struct file *file, void __user *arg)
{
A
Al Viro 已提交
174
	struct btrfs_inode *ip = BTRFS_I(file_inode(file));
175 176 177 178 179 180 181
	unsigned int flags = btrfs_flags_to_ioctl(ip->flags);

	if (copy_to_user(arg, &flags, sizeof(flags)))
		return -EFAULT;
	return 0;
}

182 183 184 185 186
static int check_flags(unsigned int flags)
{
	if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \
		      FS_NOATIME_FL | FS_NODUMP_FL | \
		      FS_SYNC_FL | FS_DIRSYNC_FL | \
L
Li Zefan 已提交
187 188
		      FS_NOCOMP_FL | FS_COMPR_FL |
		      FS_NOCOW_FL))
189 190 191 192 193 194 195 196
		return -EOPNOTSUPP;

	if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL))
		return -EINVAL;

	return 0;
}

197 198
static int btrfs_ioctl_setflags(struct file *file, void __user *arg)
{
A
Al Viro 已提交
199
	struct inode *inode = file_inode(file);
200
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
201 202 203 204 205
	struct btrfs_inode *ip = BTRFS_I(inode);
	struct btrfs_root *root = ip->root;
	struct btrfs_trans_handle *trans;
	unsigned int flags, oldflags;
	int ret;
206 207
	u64 ip_oldflags;
	unsigned int i_oldflags;
208
	umode_t mode;
209

210 211 212
	if (!inode_owner_or_capable(inode))
		return -EPERM;

L
Li Zefan 已提交
213 214 215
	if (btrfs_root_readonly(root))
		return -EROFS;

216 217 218
	if (copy_from_user(&flags, arg, sizeof(flags)))
		return -EFAULT;

219 220 221
	ret = check_flags(flags);
	if (ret)
		return ret;
C
Christoph Hellwig 已提交
222

223 224 225 226
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;

A
Al Viro 已提交
227
	inode_lock(inode);
228

229 230
	ip_oldflags = ip->flags;
	i_oldflags = inode->i_flags;
231
	mode = inode->i_mode;
232

233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265
	flags = btrfs_mask_flags(inode->i_mode, flags);
	oldflags = btrfs_flags_to_ioctl(ip->flags);
	if ((flags ^ oldflags) & (FS_APPEND_FL | FS_IMMUTABLE_FL)) {
		if (!capable(CAP_LINUX_IMMUTABLE)) {
			ret = -EPERM;
			goto out_unlock;
		}
	}

	if (flags & FS_SYNC_FL)
		ip->flags |= BTRFS_INODE_SYNC;
	else
		ip->flags &= ~BTRFS_INODE_SYNC;
	if (flags & FS_IMMUTABLE_FL)
		ip->flags |= BTRFS_INODE_IMMUTABLE;
	else
		ip->flags &= ~BTRFS_INODE_IMMUTABLE;
	if (flags & FS_APPEND_FL)
		ip->flags |= BTRFS_INODE_APPEND;
	else
		ip->flags &= ~BTRFS_INODE_APPEND;
	if (flags & FS_NODUMP_FL)
		ip->flags |= BTRFS_INODE_NODUMP;
	else
		ip->flags &= ~BTRFS_INODE_NODUMP;
	if (flags & FS_NOATIME_FL)
		ip->flags |= BTRFS_INODE_NOATIME;
	else
		ip->flags &= ~BTRFS_INODE_NOATIME;
	if (flags & FS_DIRSYNC_FL)
		ip->flags |= BTRFS_INODE_DIRSYNC;
	else
		ip->flags &= ~BTRFS_INODE_DIRSYNC;
266 267 268 269 270 271 272 273 274 275 276 277 278 279 280
	if (flags & FS_NOCOW_FL) {
		if (S_ISREG(mode)) {
			/*
			 * It's safe to turn csums off here, no extents exist.
			 * Otherwise we want the flag to reflect the real COW
			 * status of the file and will not set it.
			 */
			if (inode->i_size == 0)
				ip->flags |= BTRFS_INODE_NODATACOW
					   | BTRFS_INODE_NODATASUM;
		} else {
			ip->flags |= BTRFS_INODE_NODATACOW;
		}
	} else {
		/*
281
		 * Revert back under same assumptions as above
282 283 284 285 286 287 288 289 290
		 */
		if (S_ISREG(mode)) {
			if (inode->i_size == 0)
				ip->flags &= ~(BTRFS_INODE_NODATACOW
				             | BTRFS_INODE_NODATASUM);
		} else {
			ip->flags &= ~BTRFS_INODE_NODATACOW;
		}
	}
291

292 293 294 295 296 297 298 299
	/*
	 * The COMPRESS flag can only be changed by users, while the NOCOMPRESS
	 * flag may be changed automatically if compression code won't make
	 * things smaller.
	 */
	if (flags & FS_NOCOMP_FL) {
		ip->flags &= ~BTRFS_INODE_COMPRESS;
		ip->flags |= BTRFS_INODE_NOCOMPRESS;
300 301 302 303

		ret = btrfs_set_prop(inode, "btrfs.compression", NULL, 0, 0);
		if (ret && ret != -ENODATA)
			goto out_drop;
304
	} else if (flags & FS_COMPR_FL) {
305 306
		const char *comp;

307 308
		ip->flags |= BTRFS_INODE_COMPRESS;
		ip->flags &= ~BTRFS_INODE_NOCOMPRESS;
309

310
		if (fs_info->compress_type == BTRFS_COMPRESS_LZO)
311
			comp = "lzo";
N
Nick Terrell 已提交
312
		else if (fs_info->compress_type == BTRFS_COMPRESS_ZLIB)
313
			comp = "zlib";
N
Nick Terrell 已提交
314 315
		else
			comp = "zstd";
316 317 318 319 320
		ret = btrfs_set_prop(inode, "btrfs.compression",
				     comp, strlen(comp), 0);
		if (ret)
			goto out_drop;

L
Li Zefan 已提交
321
	} else {
322 323 324
		ret = btrfs_set_prop(inode, "btrfs.compression", NULL, 0, 0);
		if (ret && ret != -ENODATA)
			goto out_drop;
L
Li Zefan 已提交
325
		ip->flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS);
326
	}
327

328
	trans = btrfs_start_transaction(root, 1);
329 330 331 332
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out_drop;
	}
333

334
	btrfs_update_iflags(inode);
335
	inode_inc_iversion(inode);
336
	inode->i_ctime = current_time(inode);
337 338
	ret = btrfs_update_inode(trans, root, inode);

339
	btrfs_end_transaction(trans);
340 341 342 343 344
 out_drop:
	if (ret) {
		ip->flags = ip_oldflags;
		inode->i_flags = i_oldflags;
	}
345 346

 out_unlock:
A
Al Viro 已提交
347
	inode_unlock(inode);
348
	mnt_drop_write_file(file);
349
	return ret;
350 351 352 353
}

static int btrfs_ioctl_getversion(struct file *file, int __user *arg)
{
A
Al Viro 已提交
354
	struct inode *inode = file_inode(file);
355 356 357

	return put_user(inode->i_generation, arg);
}
C
Christoph Hellwig 已提交
358

359 360
static noinline int btrfs_ioctl_fitrim(struct file *file, void __user *arg)
{
361 362
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
363 364 365 366 367
	struct btrfs_device *device;
	struct request_queue *q;
	struct fstrim_range range;
	u64 minlen = ULLONG_MAX;
	u64 num_devices = 0;
368
	u64 total_bytes = btrfs_super_total_bytes(fs_info->super_copy);
369 370 371 372 373
	int ret;

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

374 375 376
	rcu_read_lock();
	list_for_each_entry_rcu(device, &fs_info->fs_devices->devices,
				dev_list) {
377 378 379 380 381
		if (!device->bdev)
			continue;
		q = bdev_get_queue(device->bdev);
		if (blk_queue_discard(q)) {
			num_devices++;
382
			minlen = min_t(u64, q->limits.discard_granularity,
383 384 385
				     minlen);
		}
	}
386
	rcu_read_unlock();
387

388 389 390 391
	if (!num_devices)
		return -EOPNOTSUPP;
	if (copy_from_user(&range, arg, sizeof(range)))
		return -EFAULT;
392 393
	if (range.start > total_bytes ||
	    range.len < fs_info->sb->s_blocksize)
394
		return -EINVAL;
395

396
	range.len = min(range.len, total_bytes - range.start);
397
	range.minlen = max(range.minlen, minlen);
398
	ret = btrfs_trim_fs(fs_info, &range);
399 400 401 402 403 404 405 406 407
	if (ret < 0)
		return ret;

	if (copy_to_user(arg, &range, sizeof(range)))
		return -EFAULT;

	return 0;
}

408 409
int btrfs_is_empty_uuid(u8 *uuid)
{
C
Chris Mason 已提交
410 411 412 413 414 415 416
	int i;

	for (i = 0; i < BTRFS_UUID_SIZE; i++) {
		if (uuid[i])
			return 0;
	}
	return 1;
417 418
}

419
static noinline int create_subvol(struct inode *dir,
420
				  struct dentry *dentry,
421
				  const char *name, int namelen,
A
Arne Jansen 已提交
422
				  u64 *async_transid,
423
				  struct btrfs_qgroup_inherit *inherit)
C
Christoph Hellwig 已提交
424
{
425
	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
C
Christoph Hellwig 已提交
426 427
	struct btrfs_trans_handle *trans;
	struct btrfs_key key;
428
	struct btrfs_root_item *root_item;
C
Christoph Hellwig 已提交
429 430
	struct btrfs_inode_item *inode_item;
	struct extent_buffer *leaf;
431
	struct btrfs_root *root = BTRFS_I(dir)->root;
432
	struct btrfs_root *new_root;
433
	struct btrfs_block_rsv block_rsv;
434
	struct timespec cur_time = current_time(dir);
435
	struct inode *inode;
C
Christoph Hellwig 已提交
436 437 438 439
	int ret;
	int err;
	u64 objectid;
	u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID;
440
	u64 index = 0;
441
	u64 qgroup_reserved;
442
	uuid_le new_uuid;
C
Christoph Hellwig 已提交
443

444 445 446 447
	root_item = kzalloc(sizeof(*root_item), GFP_KERNEL);
	if (!root_item)
		return -ENOMEM;

448
	ret = btrfs_find_free_objectid(fs_info->tree_root, &objectid);
449
	if (ret)
450
		goto fail_free;
451

452 453
	/*
	 * Don't create subvolume whose level is not zero. Or qgroup will be
454
	 * screwed up since it assumes subvolume qgroup's level to be 0.
455
	 */
456 457 458 459
	if (btrfs_qgroup_level(objectid)) {
		ret = -ENOSPC;
		goto fail_free;
	}
460

461
	btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP);
J
Josef Bacik 已提交
462
	/*
463 464
	 * The same as the snapshot creation, please see the comment
	 * of create_snapshot().
J
Josef Bacik 已提交
465
	 */
466
	ret = btrfs_subvolume_reserve_metadata(root, &block_rsv,
467
					       8, &qgroup_reserved, false);
468
	if (ret)
469
		goto fail_free;
470 471 472 473

	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
474
		btrfs_subvolume_release_metadata(fs_info, &block_rsv);
475
		goto fail_free;
476 477 478
	}
	trans->block_rsv = &block_rsv;
	trans->bytes_reserved = block_rsv.size;
C
Christoph Hellwig 已提交
479

480
	ret = btrfs_qgroup_inherit(trans, fs_info, 0, objectid, inherit);
A
Arne Jansen 已提交
481 482 483
	if (ret)
		goto fail;

484
	leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0);
485 486 487 488
	if (IS_ERR(leaf)) {
		ret = PTR_ERR(leaf);
		goto fail;
	}
C
Christoph Hellwig 已提交
489

490
	memzero_extent_buffer(leaf, 0, sizeof(struct btrfs_header));
C
Christoph Hellwig 已提交
491 492
	btrfs_set_header_bytenr(leaf, leaf->start);
	btrfs_set_header_generation(leaf, trans->transid);
493
	btrfs_set_header_backref_rev(leaf, BTRFS_MIXED_BACKREF_REV);
C
Christoph Hellwig 已提交
494 495
	btrfs_set_header_owner(leaf, objectid);

496 497
	write_extent_buffer_fsid(leaf, fs_info->fsid);
	write_extent_buffer_chunk_tree_uuid(leaf, fs_info->chunk_tree_uuid);
C
Christoph Hellwig 已提交
498 499
	btrfs_mark_buffer_dirty(leaf);

500
	inode_item = &root_item->inode;
501 502 503
	btrfs_set_stack_inode_generation(inode_item, 1);
	btrfs_set_stack_inode_size(inode_item, 3);
	btrfs_set_stack_inode_nlink(inode_item, 1);
504
	btrfs_set_stack_inode_nbytes(inode_item,
505
				     fs_info->nodesize);
506
	btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755);
C
Christoph Hellwig 已提交
507

508 509
	btrfs_set_root_flags(root_item, 0);
	btrfs_set_root_limit(root_item, 0);
510
	btrfs_set_stack_inode_flags(inode_item, BTRFS_INODE_ROOT_ITEM_INIT);
511

512 513 514 515 516 517
	btrfs_set_root_bytenr(root_item, leaf->start);
	btrfs_set_root_generation(root_item, trans->transid);
	btrfs_set_root_level(root_item, 0);
	btrfs_set_root_refs(root_item, 1);
	btrfs_set_root_used(root_item, leaf->len);
	btrfs_set_root_last_snapshot(root_item, 0);
C
Christoph Hellwig 已提交
518

519 520
	btrfs_set_root_generation_v2(root_item,
			btrfs_root_generation(root_item));
521
	uuid_le_gen(&new_uuid);
522 523 524 525 526 527
	memcpy(root_item->uuid, new_uuid.b, BTRFS_UUID_SIZE);
	btrfs_set_stack_timespec_sec(&root_item->otime, cur_time.tv_sec);
	btrfs_set_stack_timespec_nsec(&root_item->otime, cur_time.tv_nsec);
	root_item->ctime = root_item->otime;
	btrfs_set_root_ctransid(root_item, trans->transid);
	btrfs_set_root_otransid(root_item, trans->transid);
C
Christoph Hellwig 已提交
528

529
	btrfs_tree_unlock(leaf);
C
Christoph Hellwig 已提交
530 531 532
	free_extent_buffer(leaf);
	leaf = NULL;

533
	btrfs_set_root_dirid(root_item, new_dirid);
C
Christoph Hellwig 已提交
534 535

	key.objectid = objectid;
536
	key.offset = 0;
537
	key.type = BTRFS_ROOT_ITEM_KEY;
538
	ret = btrfs_insert_root(trans, fs_info->tree_root, &key,
539
				root_item);
C
Christoph Hellwig 已提交
540 541 542
	if (ret)
		goto fail;

543
	key.offset = (u64)-1;
544
	new_root = btrfs_read_fs_root_no_name(fs_info, &key);
545 546
	if (IS_ERR(new_root)) {
		ret = PTR_ERR(new_root);
547
		btrfs_abort_transaction(trans, ret);
548 549
		goto fail;
	}
550 551 552

	btrfs_record_root_in_trans(trans, new_root);

553
	ret = btrfs_create_subvol_root(trans, new_root, root, new_dirid);
554 555
	if (ret) {
		/* We potentially lose an unused inode item here */
556
		btrfs_abort_transaction(trans, ret);
557 558 559
		goto fail;
	}

560 561 562 563
	mutex_lock(&new_root->objectid_mutex);
	new_root->highest_objectid = new_dirid;
	mutex_unlock(&new_root->objectid_mutex);

C
Christoph Hellwig 已提交
564 565 566
	/*
	 * insert the directory item
	 */
567
	ret = btrfs_set_inode_index(BTRFS_I(dir), &index);
568
	if (ret) {
569
		btrfs_abort_transaction(trans, ret);
570 571
		goto fail;
	}
572 573

	ret = btrfs_insert_dir_item(trans, root,
574
				    name, namelen, BTRFS_I(dir), &key,
575
				    BTRFS_FT_DIR, index);
576
	if (ret) {
577
		btrfs_abort_transaction(trans, ret);
C
Christoph Hellwig 已提交
578
		goto fail;
579
	}
580

581
	btrfs_i_size_write(BTRFS_I(dir), dir->i_size + namelen * 2);
582 583 584
	ret = btrfs_update_inode(trans, root, dir);
	BUG_ON(ret);

585
	ret = btrfs_add_root_ref(trans, fs_info,
586
				 objectid, root->root_key.objectid,
587
				 btrfs_ino(BTRFS_I(dir)), index, name, namelen);
588
	BUG_ON(ret);
C
Christoph Hellwig 已提交
589

590
	ret = btrfs_uuid_tree_add(trans, fs_info, root_item->uuid,
591
				  BTRFS_UUID_KEY_SUBVOL, objectid);
592
	if (ret)
593
		btrfs_abort_transaction(trans, ret);
594

C
Christoph Hellwig 已提交
595
fail:
596
	kfree(root_item);
597 598
	trans->block_rsv = NULL;
	trans->bytes_reserved = 0;
599
	btrfs_subvolume_release_metadata(fs_info, &block_rsv);
600

S
Sage Weil 已提交
601 602
	if (async_transid) {
		*async_transid = trans->transid;
603
		err = btrfs_commit_transaction_async(trans, 1);
604
		if (err)
605
			err = btrfs_commit_transaction(trans);
S
Sage Weil 已提交
606
	} else {
607
		err = btrfs_commit_transaction(trans);
S
Sage Weil 已提交
608
	}
C
Christoph Hellwig 已提交
609 610
	if (err && !ret)
		ret = err;
611

612 613
	if (!ret) {
		inode = btrfs_lookup_dentry(dir, dentry);
614 615
		if (IS_ERR(inode))
			return PTR_ERR(inode);
616 617
		d_instantiate(dentry, inode);
	}
C
Christoph Hellwig 已提交
618
	return ret;
619 620 621 622

fail_free:
	kfree(root_item);
	return ret;
C
Christoph Hellwig 已提交
623 624
}

625
static int create_snapshot(struct btrfs_root *root, struct inode *dir,
626
			   struct dentry *dentry,
627 628
			   u64 *async_transid, bool readonly,
			   struct btrfs_qgroup_inherit *inherit)
C
Christoph Hellwig 已提交
629
{
630
	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
631
	struct inode *inode;
C
Christoph Hellwig 已提交
632 633
	struct btrfs_pending_snapshot *pending_snapshot;
	struct btrfs_trans_handle *trans;
634
	int ret;
C
Christoph Hellwig 已提交
635

636
	if (!test_bit(BTRFS_ROOT_REF_COWS, &root->state))
C
Christoph Hellwig 已提交
637 638
		return -EINVAL;

639
	pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_KERNEL);
640 641 642
	if (!pending_snapshot)
		return -ENOMEM;

643
	pending_snapshot->root_item = kzalloc(sizeof(struct btrfs_root_item),
644
			GFP_KERNEL);
645 646
	pending_snapshot->path = btrfs_alloc_path();
	if (!pending_snapshot->root_item || !pending_snapshot->path) {
647 648 649 650
		ret = -ENOMEM;
		goto free_pending;
	}

651
	atomic_inc(&root->will_be_snapshotted);
652
	smp_mb__after_atomic();
653 654 655
	/* wait for no snapshot writes */
	wait_event(root->subv_writers->wait,
		   percpu_counter_sum(&root->subv_writers->counter) == 0);
656

657 658
	ret = btrfs_start_delalloc_inodes(root, 0);
	if (ret)
659
		goto dec_and_free;
660

661
	btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1);
662

663 664
	btrfs_init_block_rsv(&pending_snapshot->block_rsv,
			     BTRFS_BLOCK_RSV_TEMP);
665 666 667 668 669 670
	/*
	 * 1 - parent dir inode
	 * 2 - dir entries
	 * 1 - root item
	 * 2 - root ref/backref
	 * 1 - root of snapshot
671
	 * 1 - UUID item
672 673
	 */
	ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root,
674
					&pending_snapshot->block_rsv, 8,
675 676
					&pending_snapshot->qgroup_reserved,
					false);
677
	if (ret)
678
		goto dec_and_free;
679

680
	pending_snapshot->dentry = dentry;
C
Christoph Hellwig 已提交
681
	pending_snapshot->root = root;
L
Li Zefan 已提交
682
	pending_snapshot->readonly = readonly;
683
	pending_snapshot->dir = dir;
684
	pending_snapshot->inherit = inherit;
685

686
	trans = btrfs_start_transaction(root, 0);
687 688 689 690 691
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto fail;
	}

692
	spin_lock(&fs_info->trans_lock);
C
Christoph Hellwig 已提交
693 694
	list_add(&pending_snapshot->list,
		 &trans->transaction->pending_snapshots);
695
	spin_unlock(&fs_info->trans_lock);
S
Sage Weil 已提交
696 697
	if (async_transid) {
		*async_transid = trans->transid;
698
		ret = btrfs_commit_transaction_async(trans, 1);
699
		if (ret)
700
			ret = btrfs_commit_transaction(trans);
S
Sage Weil 已提交
701
	} else {
702
		ret = btrfs_commit_transaction(trans);
S
Sage Weil 已提交
703
	}
704
	if (ret)
705
		goto fail;
706 707 708 709 710

	ret = pending_snapshot->error;
	if (ret)
		goto fail;

711 712 713 714
	ret = btrfs_orphan_cleanup(pending_snapshot->snap);
	if (ret)
		goto fail;

715
	inode = btrfs_lookup_dentry(d_inode(dentry->d_parent), dentry);
716 717 718 719
	if (IS_ERR(inode)) {
		ret = PTR_ERR(inode);
		goto fail;
	}
720

721 722 723
	d_instantiate(dentry, inode);
	ret = 0;
fail:
724
	btrfs_subvolume_release_metadata(fs_info, &pending_snapshot->block_rsv);
725
dec_and_free:
726 727
	if (atomic_dec_and_test(&root->will_be_snapshotted))
		wake_up_atomic_t(&root->will_be_snapshotted);
728 729
free_pending:
	kfree(pending_snapshot->root_item);
730
	btrfs_free_path(pending_snapshot->path);
731 732
	kfree(pending_snapshot);

C
Christoph Hellwig 已提交
733 734 735
	return ret;
}

736 737 738 739 740 741 742 743 744 745 746
/*  copy of may_delete in fs/namei.c()
 *	Check whether we can remove a link victim from directory dir, check
 *  whether the type of victim is right.
 *  1. We can't do it if dir is read-only (done in permission())
 *  2. We should have write and exec permissions on dir
 *  3. We can't remove anything from append-only dir
 *  4. We can't do anything with immutable dir (done in permission())
 *  5. If the sticky bit on dir is set we should either
 *	a. be owner of dir, or
 *	b. be owner of victim, or
 *	c. have CAP_FOWNER capability
747
 *  6. If the victim is append-only or immutable we can't do anything with
748 749 750 751 752 753 754 755
 *     links pointing to it.
 *  7. If we were asked to remove a directory and victim isn't one - ENOTDIR.
 *  8. If we were asked to remove a non-directory and victim isn't one - EISDIR.
 *  9. We can't remove a root or mountpoint.
 * 10. We don't allow removal of NFS sillyrenamed files; it's handled by
 *     nfs_async_unlink().
 */

756
static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir)
757 758 759
{
	int error;

760
	if (d_really_is_negative(victim))
761 762
		return -ENOENT;

763
	BUG_ON(d_inode(victim->d_parent) != dir);
764
	audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE);
765 766 767 768 769 770

	error = inode_permission(dir, MAY_WRITE | MAY_EXEC);
	if (error)
		return error;
	if (IS_APPEND(dir))
		return -EPERM;
771 772
	if (check_sticky(dir, d_inode(victim)) || IS_APPEND(d_inode(victim)) ||
	    IS_IMMUTABLE(d_inode(victim)) || IS_SWAPFILE(d_inode(victim)))
773 774
		return -EPERM;
	if (isdir) {
775
		if (!d_is_dir(victim))
776 777 778
			return -ENOTDIR;
		if (IS_ROOT(victim))
			return -EBUSY;
779
	} else if (d_is_dir(victim))
780 781 782 783 784 785 786 787
		return -EISDIR;
	if (IS_DEADDIR(dir))
		return -ENOENT;
	if (victim->d_flags & DCACHE_NFSFS_RENAMED)
		return -EBUSY;
	return 0;
}

788 789 790
/* copy of may_create in fs/namei.c() */
static inline int btrfs_may_create(struct inode *dir, struct dentry *child)
{
791
	if (d_really_is_positive(child))
792 793 794 795 796 797 798 799 800 801 802
		return -EEXIST;
	if (IS_DEADDIR(dir))
		return -ENOENT;
	return inode_permission(dir, MAY_WRITE | MAY_EXEC);
}

/*
 * Create a new subvolume below @parent.  This is largely modeled after
 * sys_mkdirat and vfs_mkdir, but we only do a single component lookup
 * inside this filesystem so it's quite a bit simpler.
 */
A
Al Viro 已提交
803
static noinline int btrfs_mksubvol(const struct path *parent,
804
				   const char *name, int namelen,
S
Sage Weil 已提交
805
				   struct btrfs_root *snap_src,
A
Arne Jansen 已提交
806
				   u64 *async_transid, bool readonly,
807
				   struct btrfs_qgroup_inherit *inherit)
808
{
809 810
	struct inode *dir = d_inode(parent->dentry);
	struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb);
811 812 813
	struct dentry *dentry;
	int error;

814 815 816
	error = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT);
	if (error == -EINTR)
		return error;
817 818 819 820 821 822

	dentry = lookup_one_len(name, parent->dentry, namelen);
	error = PTR_ERR(dentry);
	if (IS_ERR(dentry))
		goto out_unlock;

823
	error = btrfs_may_create(dir, dentry);
824
	if (error)
825
		goto out_dput;
826

C
Chris Mason 已提交
827 828 829 830 831 832 833 834 835 836
	/*
	 * even if this name doesn't exist, we may get hash collisions.
	 * check for them now when we can safely fail
	 */
	error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root,
					       dir->i_ino, name,
					       namelen);
	if (error)
		goto out_dput;

837
	down_read(&fs_info->subvol_sem);
838 839 840 841

	if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0)
		goto out_up_read;

842
	if (snap_src) {
843
		error = create_snapshot(snap_src, dir, dentry,
A
Arne Jansen 已提交
844
					async_transid, readonly, inherit);
845
	} else {
846 847
		error = create_subvol(dir, dentry, name, namelen,
				      async_transid, inherit);
848
	}
849 850 851
	if (!error)
		fsnotify_mkdir(dir, dentry);
out_up_read:
852
	up_read(&fs_info->subvol_sem);
853 854 855
out_dput:
	dput(dentry);
out_unlock:
A
Al Viro 已提交
856
	inode_unlock(dir);
857 858 859
	return error;
}

C
Chris Mason 已提交
860 861 862 863 864 865 866
/*
 * When we're defragging a range, we don't want to kick it off again
 * if it is really just waiting for delalloc to send it down.
 * If we find a nice big extent or delalloc range for the bytes in the
 * file you want to defrag, we return 0 to let you know to skip this
 * part of the file
 */
867
static int check_defrag_in_cache(struct inode *inode, u64 offset, u32 thresh)
C
Chris Mason 已提交
868 869 870 871 872 873 874
{
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct extent_map *em = NULL;
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
	u64 end;

	read_lock(&em_tree->lock);
875
	em = lookup_extent_mapping(em_tree, offset, PAGE_SIZE);
C
Chris Mason 已提交
876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901
	read_unlock(&em_tree->lock);

	if (em) {
		end = extent_map_end(em);
		free_extent_map(em);
		if (end - offset > thresh)
			return 0;
	}
	/* if we already have a nice delalloc here, just stop */
	thresh /= 2;
	end = count_range_bits(io_tree, &offset, offset + thresh,
			       thresh, EXTENT_DELALLOC, 1);
	if (end >= thresh)
		return 0;
	return 1;
}

/*
 * helper function to walk through a file and find extents
 * newer than a specific transid, and smaller than thresh.
 *
 * This is used by the defragging code to find new and small
 * extents
 */
static int find_new_extents(struct btrfs_root *root,
			    struct inode *inode, u64 newer_than,
902
			    u64 *off, u32 thresh)
C
Chris Mason 已提交
903 904 905 906 907 908 909
{
	struct btrfs_path *path;
	struct btrfs_key min_key;
	struct extent_buffer *leaf;
	struct btrfs_file_extent_item *extent;
	int type;
	int ret;
910
	u64 ino = btrfs_ino(BTRFS_I(inode));
C
Chris Mason 已提交
911 912 913 914 915

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

916
	min_key.objectid = ino;
C
Chris Mason 已提交
917 918 919
	min_key.type = BTRFS_EXTENT_DATA_KEY;
	min_key.offset = *off;

920
	while (1) {
921
		ret = btrfs_search_forward(root, &min_key, path, newer_than);
C
Chris Mason 已提交
922 923
		if (ret != 0)
			goto none;
924
process_slot:
925
		if (min_key.objectid != ino)
C
Chris Mason 已提交
926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942
			goto none;
		if (min_key.type != BTRFS_EXTENT_DATA_KEY)
			goto none;

		leaf = path->nodes[0];
		extent = btrfs_item_ptr(leaf, path->slots[0],
					struct btrfs_file_extent_item);

		type = btrfs_file_extent_type(leaf, extent);
		if (type == BTRFS_FILE_EXTENT_REG &&
		    btrfs_file_extent_num_bytes(leaf, extent) < thresh &&
		    check_defrag_in_cache(inode, min_key.offset, thresh)) {
			*off = min_key.offset;
			btrfs_free_path(path);
			return 0;
		}

943 944 945 946 947 948
		path->slots[0]++;
		if (path->slots[0] < btrfs_header_nritems(leaf)) {
			btrfs_item_key_to_cpu(leaf, &min_key, path->slots[0]);
			goto process_slot;
		}

C
Chris Mason 已提交
949 950 951 952 953 954 955 956 957 958 959
		if (min_key.offset == (u64)-1)
			goto none;

		min_key.offset++;
		btrfs_release_path(path);
	}
none:
	btrfs_free_path(path);
	return -ENOENT;
}

L
Li Zefan 已提交
960
static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start)
961 962
{
	struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree;
L
Li Zefan 已提交
963 964
	struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree;
	struct extent_map *em;
965
	u64 len = PAGE_SIZE;
966

L
Li Zefan 已提交
967 968 969 970
	/*
	 * hopefully we have this extent in the tree already, try without
	 * the full extent lock
	 */
971
	read_lock(&em_tree->lock);
L
Li Zefan 已提交
972
	em = lookup_extent_mapping(em_tree, start, len);
973 974
	read_unlock(&em_tree->lock);

L
Li Zefan 已提交
975
	if (!em) {
976 977 978
		struct extent_state *cached = NULL;
		u64 end = start + len - 1;

L
Li Zefan 已提交
979
		/* get the big lock and read metadata off disk */
980
		lock_extent_bits(io_tree, start, end, &cached);
981
		em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start, len, 0);
982
		unlock_extent_cached(io_tree, start, end, &cached, GFP_NOFS);
L
Li Zefan 已提交
983 984 985 986 987 988 989

		if (IS_ERR(em))
			return NULL;
	}

	return em;
}
990

L
Li Zefan 已提交
991 992 993 994 995 996 997 998 999 1000
static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em)
{
	struct extent_map *next;
	bool ret = true;

	/* this is the last extent */
	if (em->start + em->len >= i_size_read(inode))
		return false;

	next = defrag_lookup_extent(inode, em->start + em->len);
1001 1002 1003
	if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE)
		ret = false;
	else if ((em->block_start + em->block_len == next->block_start) &&
1004
		 (em->block_len > SZ_128K && next->block_len > SZ_128K))
L
Li Zefan 已提交
1005 1006 1007
		ret = false;

	free_extent_map(next);
1008 1009 1010
	return ret;
}

1011
static int should_defrag_range(struct inode *inode, u64 start, u32 thresh,
1012 1013
			       u64 *last_len, u64 *skip, u64 *defrag_end,
			       int compress)
1014
{
L
Li Zefan 已提交
1015
	struct extent_map *em;
1016
	int ret = 1;
L
Li Zefan 已提交
1017
	bool next_mergeable = true;
1018
	bool prev_mergeable = true;
1019 1020

	/*
1021
	 * make sure that once we start defragging an extent, we keep on
1022 1023 1024 1025 1026 1027 1028
	 * defragging it
	 */
	if (start < *defrag_end)
		return 1;

	*skip = 0;

L
Li Zefan 已提交
1029 1030 1031
	em = defrag_lookup_extent(inode, start);
	if (!em)
		return 0;
1032 1033

	/* this will cover holes, and inline extents */
1034
	if (em->block_start >= EXTENT_MAP_LAST_BYTE) {
1035
		ret = 0;
1036 1037 1038
		goto out;
	}

1039 1040 1041
	if (!*defrag_end)
		prev_mergeable = false;

L
Li Zefan 已提交
1042
	next_mergeable = defrag_check_next_extent(inode, em);
1043
	/*
L
Li Zefan 已提交
1044 1045
	 * we hit a real extent, if it is big or the next extent is not a
	 * real extent, don't bother defragging it
1046
	 */
1047
	if (!compress && (*last_len == 0 || *last_len >= thresh) &&
1048
	    (em->len >= thresh || (!next_mergeable && !prev_mergeable)))
1049
		ret = 0;
1050
out:
1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070
	/*
	 * last_len ends up being a counter of how many bytes we've defragged.
	 * every time we choose not to defrag an extent, we reset *last_len
	 * so that the next tiny extent will force a defrag.
	 *
	 * The end result of this is that tiny extents before a single big
	 * extent will force at least part of that big extent to be defragged.
	 */
	if (ret) {
		*defrag_end = extent_map_end(em);
	} else {
		*last_len = 0;
		*skip = extent_map_end(em);
		*defrag_end = 0;
	}

	free_extent_map(em);
	return ret;
}

C
Chris Mason 已提交
1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085
/*
 * it doesn't do much good to defrag one or two pages
 * at a time.  This pulls in a nice chunk of pages
 * to COW and defrag.
 *
 * It also makes sure the delalloc code has enough
 * dirty data to avoid making new small extents as part
 * of the defrag
 *
 * It's a good idea to start RA on this range
 * before calling this.
 */
static int cluster_pages_for_defrag(struct inode *inode,
				    struct page **pages,
				    unsigned long start_index,
1086
				    unsigned long num_pages)
C
Christoph Hellwig 已提交
1087
{
C
Chris Mason 已提交
1088 1089 1090 1091
	unsigned long file_end;
	u64 isize = i_size_read(inode);
	u64 page_start;
	u64 page_end;
1092
	u64 page_cnt;
C
Chris Mason 已提交
1093 1094 1095
	int ret;
	int i;
	int i_done;
1096
	struct btrfs_ordered_extent *ordered;
C
Chris Mason 已提交
1097
	struct extent_state *cached_state = NULL;
1098
	struct extent_io_tree *tree;
1099
	struct extent_changeset *data_reserved = NULL;
1100
	gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping);
C
Chris Mason 已提交
1101

1102
	file_end = (isize - 1) >> PAGE_SHIFT;
1103 1104 1105 1106
	if (!isize || start_index > file_end)
		return 0;

	page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1);
C
Chris Mason 已提交
1107

1108
	ret = btrfs_delalloc_reserve_space(inode, &data_reserved,
1109 1110
			start_index << PAGE_SHIFT,
			page_cnt << PAGE_SHIFT);
C
Chris Mason 已提交
1111 1112 1113
	if (ret)
		return ret;
	i_done = 0;
1114
	tree = &BTRFS_I(inode)->io_tree;
C
Chris Mason 已提交
1115 1116

	/* step one, lock all the pages */
1117
	for (i = 0; i < page_cnt; i++) {
C
Chris Mason 已提交
1118
		struct page *page;
1119
again:
1120
		page = find_or_create_page(inode->i_mapping,
1121
					   start_index + i, mask);
C
Chris Mason 已提交
1122 1123 1124
		if (!page)
			break;

1125
		page_start = page_offset(page);
1126
		page_end = page_start + PAGE_SIZE - 1;
1127
		while (1) {
1128
			lock_extent_bits(tree, page_start, page_end,
1129
					 &cached_state);
1130 1131
			ordered = btrfs_lookup_ordered_extent(inode,
							      page_start);
1132 1133
			unlock_extent_cached(tree, page_start, page_end,
					     &cached_state, GFP_NOFS);
1134 1135 1136 1137 1138 1139 1140
			if (!ordered)
				break;

			unlock_page(page);
			btrfs_start_ordered_extent(inode, ordered, 1);
			btrfs_put_ordered_extent(ordered);
			lock_page(page);
1141 1142 1143 1144 1145 1146
			/*
			 * we unlocked the page above, so we need check if
			 * it was released or not.
			 */
			if (page->mapping != inode->i_mapping) {
				unlock_page(page);
1147
				put_page(page);
1148 1149
				goto again;
			}
1150 1151
		}

C
Chris Mason 已提交
1152 1153 1154 1155 1156
		if (!PageUptodate(page)) {
			btrfs_readpage(NULL, page);
			lock_page(page);
			if (!PageUptodate(page)) {
				unlock_page(page);
1157
				put_page(page);
C
Chris Mason 已提交
1158 1159 1160 1161
				ret = -EIO;
				break;
			}
		}
1162 1163 1164

		if (page->mapping != inode->i_mapping) {
			unlock_page(page);
1165
			put_page(page);
1166 1167 1168
			goto again;
		}

C
Chris Mason 已提交
1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185
		pages[i] = page;
		i_done++;
	}
	if (!i_done || ret)
		goto out;

	if (!(inode->i_sb->s_flags & MS_ACTIVE))
		goto out;

	/*
	 * so now we have a nice long stream of locked
	 * and up to date pages, lets wait on them
	 */
	for (i = 0; i < i_done; i++)
		wait_on_page_writeback(pages[i]);

	page_start = page_offset(pages[0]);
1186
	page_end = page_offset(pages[i_done - 1]) + PAGE_SIZE;
C
Chris Mason 已提交
1187 1188

	lock_extent_bits(&BTRFS_I(inode)->io_tree,
1189
			 page_start, page_end - 1, &cached_state);
C
Chris Mason 已提交
1190 1191
	clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start,
			  page_end - 1, EXTENT_DIRTY | EXTENT_DELALLOC |
1192 1193
			  EXTENT_DO_ACCOUNTING | EXTENT_DEFRAG, 0, 0,
			  &cached_state, GFP_NOFS);
C
Chris Mason 已提交
1194

1195
	if (i_done != page_cnt) {
1196 1197 1198
		spin_lock(&BTRFS_I(inode)->lock);
		BTRFS_I(inode)->outstanding_extents++;
		spin_unlock(&BTRFS_I(inode)->lock);
1199
		btrfs_delalloc_release_space(inode, data_reserved,
1200 1201
				start_index << PAGE_SHIFT,
				(page_cnt - i_done) << PAGE_SHIFT);
C
Chris Mason 已提交
1202 1203 1204
	}


1205
	set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1,
1206
			  &cached_state);
C
Chris Mason 已提交
1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217

	unlock_extent_cached(&BTRFS_I(inode)->io_tree,
			     page_start, page_end - 1, &cached_state,
			     GFP_NOFS);

	for (i = 0; i < i_done; i++) {
		clear_page_dirty_for_io(pages[i]);
		ClearPageChecked(pages[i]);
		set_page_extent_mapped(pages[i]);
		set_page_dirty(pages[i]);
		unlock_page(pages[i]);
1218
		put_page(pages[i]);
C
Chris Mason 已提交
1219
	}
1220
	extent_changeset_free(data_reserved);
C
Chris Mason 已提交
1221 1222 1223 1224
	return i_done;
out:
	for (i = 0; i < i_done; i++) {
		unlock_page(pages[i]);
1225
		put_page(pages[i]);
C
Chris Mason 已提交
1226
	}
1227
	btrfs_delalloc_release_space(inode, data_reserved,
1228 1229
			start_index << PAGE_SHIFT,
			page_cnt << PAGE_SHIFT);
1230
	extent_changeset_free(data_reserved);
C
Chris Mason 已提交
1231 1232 1233 1234 1235 1236 1237 1238
	return ret;

}

int btrfs_defrag_file(struct inode *inode, struct file *file,
		      struct btrfs_ioctl_defrag_range_args *range,
		      u64 newer_than, unsigned long max_to_defrag)
{
1239
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
C
Chris Mason 已提交
1240 1241
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct file_ra_state *ra = NULL;
C
Christoph Hellwig 已提交
1242
	unsigned long last_index;
1243
	u64 isize = i_size_read(inode);
1244 1245 1246
	u64 last_len = 0;
	u64 skip = 0;
	u64 defrag_end = 0;
C
Chris Mason 已提交
1247
	u64 newer_off = range->start;
C
Christoph Hellwig 已提交
1248
	unsigned long i;
1249
	unsigned long ra_index = 0;
C
Christoph Hellwig 已提交
1250
	int ret;
C
Chris Mason 已提交
1251
	int defrag_count = 0;
1252
	int compress_type = BTRFS_COMPRESS_ZLIB;
1253
	u32 extent_thresh = range->extent_thresh;
1254
	unsigned long max_cluster = SZ_256K >> PAGE_SHIFT;
1255
	unsigned long cluster = max_cluster;
1256
	u64 new_align = ~((u64)SZ_128K - 1);
C
Chris Mason 已提交
1257
	struct page **pages = NULL;
1258
	bool do_compress = range->flags & BTRFS_DEFRAG_RANGE_COMPRESS;
C
Chris Mason 已提交
1259

1260 1261 1262 1263 1264
	if (isize == 0)
		return 0;

	if (range->start >= isize)
		return -EINVAL;
1265

1266
	if (do_compress) {
1267 1268 1269 1270 1271
		if (range->compress_type > BTRFS_COMPRESS_TYPES)
			return -EINVAL;
		if (range->compress_type)
			compress_type = range->compress_type;
	}
C
Christoph Hellwig 已提交
1272

1273
	if (extent_thresh == 0)
1274
		extent_thresh = SZ_256K;
1275

C
Chris Mason 已提交
1276
	/*
1277 1278 1279
	 * If we were not given a file, allocate a readahead context. As
	 * readahead is just an optimization, defrag will work without it so
	 * we don't error out.
C
Chris Mason 已提交
1280 1281
	 */
	if (!file) {
1282
		ra = kzalloc(sizeof(*ra), GFP_KERNEL);
1283 1284
		if (ra)
			file_ra_state_init(ra, inode->i_mapping);
C
Chris Mason 已提交
1285 1286 1287 1288
	} else {
		ra = &file->f_ra;
	}

1289
	pages = kmalloc_array(max_cluster, sizeof(struct page *), GFP_KERNEL);
C
Chris Mason 已提交
1290 1291 1292 1293 1294 1295
	if (!pages) {
		ret = -ENOMEM;
		goto out_ra;
	}

	/* find the last page to defrag */
C
Chris Mason 已提交
1296
	if (range->start + range->len > range->start) {
1297
		last_index = min_t(u64, isize - 1,
1298
			 range->start + range->len - 1) >> PAGE_SHIFT;
C
Chris Mason 已提交
1299
	} else {
1300
		last_index = (isize - 1) >> PAGE_SHIFT;
C
Chris Mason 已提交
1301 1302
	}

C
Chris Mason 已提交
1303 1304
	if (newer_than) {
		ret = find_new_extents(root, inode, newer_than,
1305
				       &newer_off, SZ_64K);
C
Chris Mason 已提交
1306 1307 1308 1309 1310 1311
		if (!ret) {
			range->start = newer_off;
			/*
			 * we always align our defrag to help keep
			 * the extents in the file evenly spaced
			 */
1312
			i = (newer_off & new_align) >> PAGE_SHIFT;
C
Chris Mason 已提交
1313 1314 1315
		} else
			goto out_ra;
	} else {
1316
		i = range->start >> PAGE_SHIFT;
C
Chris Mason 已提交
1317 1318
	}
	if (!max_to_defrag)
1319
		max_to_defrag = last_index - i + 1;
C
Chris Mason 已提交
1320

L
Li Zefan 已提交
1321 1322 1323 1324 1325 1326 1327
	/*
	 * make writeback starts from i, so the defrag range can be
	 * written sequentially.
	 */
	if (i < inode->i_mapping->writeback_index)
		inode->i_mapping->writeback_index = i;

1328
	while (i <= last_index && defrag_count < max_to_defrag &&
1329
	       (i < DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE))) {
C
Chris Mason 已提交
1330 1331 1332 1333 1334 1335 1336
		/*
		 * make sure we stop running if someone unmounts
		 * the FS
		 */
		if (!(inode->i_sb->s_flags & MS_ACTIVE))
			break;

1337 1338
		if (btrfs_defrag_cancelled(fs_info)) {
			btrfs_debug(fs_info, "defrag_file cancelled");
1339 1340 1341 1342
			ret = -EAGAIN;
			break;
		}

1343
		if (!should_defrag_range(inode, (u64)i << PAGE_SHIFT,
L
Li Zefan 已提交
1344
					 extent_thresh, &last_len, &skip,
1345
					 &defrag_end, do_compress)){
1346 1347 1348 1349 1350
			unsigned long next;
			/*
			 * the should_defrag function tells us how much to skip
			 * bump our counter by the suggested amount
			 */
1351
			next = DIV_ROUND_UP(skip, PAGE_SIZE);
1352 1353 1354
			i = max(i + 1, next);
			continue;
		}
1355 1356

		if (!newer_than) {
1357 1358
			cluster = (PAGE_ALIGN(defrag_end) >>
				   PAGE_SHIFT) - i;
1359 1360 1361 1362 1363 1364 1365
			cluster = min(cluster, max_cluster);
		} else {
			cluster = max_cluster;
		}

		if (i + cluster > ra_index) {
			ra_index = max(i, ra_index);
1366
			if (ra)
1367 1368
				page_cache_sync_readahead(inode->i_mapping, ra,
						file, ra_index, cluster);
1369
			ra_index += cluster;
1370
		}
1371

A
Al Viro 已提交
1372
		inode_lock(inode);
1373
		if (do_compress)
1374
			BTRFS_I(inode)->defrag_compress = compress_type;
1375
		ret = cluster_pages_for_defrag(inode, pages, i, cluster);
1376
		if (ret < 0) {
A
Al Viro 已提交
1377
			inode_unlock(inode);
C
Chris Mason 已提交
1378
			goto out_ra;
1379
		}
C
Chris Mason 已提交
1380 1381

		defrag_count += ret;
1382
		balance_dirty_pages_ratelimited(inode->i_mapping);
A
Al Viro 已提交
1383
		inode_unlock(inode);
C
Chris Mason 已提交
1384 1385 1386 1387 1388

		if (newer_than) {
			if (newer_off == (u64)-1)
				break;

1389 1390 1391
			if (ret > 0)
				i += ret;

C
Chris Mason 已提交
1392
			newer_off = max(newer_off + 1,
1393
					(u64)i << PAGE_SHIFT);
C
Chris Mason 已提交
1394

1395 1396
			ret = find_new_extents(root, inode, newer_than,
					       &newer_off, SZ_64K);
C
Chris Mason 已提交
1397 1398
			if (!ret) {
				range->start = newer_off;
1399
				i = (newer_off & new_align) >> PAGE_SHIFT;
C
Chris Mason 已提交
1400 1401
			} else {
				break;
C
Christoph Hellwig 已提交
1402
			}
C
Chris Mason 已提交
1403
		} else {
1404
			if (ret > 0) {
L
Li Zefan 已提交
1405
				i += ret;
1406
				last_len += ret << PAGE_SHIFT;
1407
			} else {
L
Li Zefan 已提交
1408
				i++;
1409 1410
				last_len = 0;
			}
C
Christoph Hellwig 已提交
1411 1412 1413
		}
	}

1414
	if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) {
C
Chris Mason 已提交
1415
		filemap_flush(inode->i_mapping);
1416 1417 1418 1419
		if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT,
			     &BTRFS_I(inode)->runtime_flags))
			filemap_flush(inode->i_mapping);
	}
C
Chris Mason 已提交
1420

1421
	if (range->compress_type == BTRFS_COMPRESS_LZO) {
1422
		btrfs_set_fs_incompat(fs_info, COMPRESS_LZO);
N
Nick Terrell 已提交
1423 1424
	} else if (range->compress_type == BTRFS_COMPRESS_ZSTD) {
		btrfs_set_fs_incompat(fs_info, COMPRESS_ZSTD);
1425 1426
	}

1427
	ret = defrag_count;
1428

C
Chris Mason 已提交
1429
out_ra:
1430
	if (do_compress) {
A
Al Viro 已提交
1431
		inode_lock(inode);
1432
		BTRFS_I(inode)->defrag_compress = BTRFS_COMPRESS_NONE;
A
Al Viro 已提交
1433
		inode_unlock(inode);
1434
	}
C
Chris Mason 已提交
1435 1436 1437
	if (!file)
		kfree(ra);
	kfree(pages);
1438
	return ret;
C
Christoph Hellwig 已提交
1439 1440
}

1441
static noinline int btrfs_ioctl_resize(struct file *file,
1442
					void __user *arg)
C
Christoph Hellwig 已提交
1443
{
1444 1445
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
C
Christoph Hellwig 已提交
1446 1447 1448
	u64 new_size;
	u64 old_size;
	u64 devid = 1;
1449
	struct btrfs_root *root = BTRFS_I(inode)->root;
C
Christoph Hellwig 已提交
1450 1451 1452 1453
	struct btrfs_ioctl_vol_args *vol_args;
	struct btrfs_trans_handle *trans;
	struct btrfs_device *device = NULL;
	char *sizestr;
1454
	char *retptr;
C
Christoph Hellwig 已提交
1455 1456 1457 1458
	char *devstr = NULL;
	int ret = 0;
	int mod = 0;

1459 1460 1461
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

1462 1463 1464 1465
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;

1466
	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
1467
		mnt_drop_write_file(file);
1468
		return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
1469 1470
	}

1471
	mutex_lock(&fs_info->volume_mutex);
L
Li Zefan 已提交
1472
	vol_args = memdup_user(arg, sizeof(*vol_args));
1473 1474 1475 1476
	if (IS_ERR(vol_args)) {
		ret = PTR_ERR(vol_args);
		goto out;
	}
1477 1478

	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
C
Christoph Hellwig 已提交
1479 1480 1481 1482 1483 1484 1485

	sizestr = vol_args->name;
	devstr = strchr(sizestr, ':');
	if (devstr) {
		sizestr = devstr + 1;
		*devstr = '\0';
		devstr = vol_args->name;
1486 1487 1488
		ret = kstrtoull(devstr, 10, &devid);
		if (ret)
			goto out_free;
1489 1490 1491 1492
		if (!devid) {
			ret = -EINVAL;
			goto out_free;
		}
1493
		btrfs_info(fs_info, "resizing devid %llu", devid);
C
Christoph Hellwig 已提交
1494
	}
M
Miao Xie 已提交
1495

1496
	device = btrfs_find_device(fs_info, devid, NULL, NULL);
C
Christoph Hellwig 已提交
1497
	if (!device) {
1498 1499
		btrfs_info(fs_info, "resizer unable to find device %llu",
			   devid);
1500
		ret = -ENODEV;
1501
		goto out_free;
C
Christoph Hellwig 已提交
1502
	}
M
Miao Xie 已提交
1503 1504

	if (!device->writeable) {
1505
		btrfs_info(fs_info,
1506
			   "resizer unable to apply on readonly device %llu",
1507
		       devid);
1508
		ret = -EPERM;
L
Liu Bo 已提交
1509 1510 1511
		goto out_free;
	}

C
Christoph Hellwig 已提交
1512 1513 1514 1515 1516 1517 1518 1519 1520 1521
	if (!strcmp(sizestr, "max"))
		new_size = device->bdev->bd_inode->i_size;
	else {
		if (sizestr[0] == '-') {
			mod = -1;
			sizestr++;
		} else if (sizestr[0] == '+') {
			mod = 1;
			sizestr++;
		}
1522 1523
		new_size = memparse(sizestr, &retptr);
		if (*retptr != '\0' || new_size == 0) {
C
Christoph Hellwig 已提交
1524
			ret = -EINVAL;
1525
			goto out_free;
C
Christoph Hellwig 已提交
1526 1527 1528
		}
	}

1529
	if (device->is_tgtdev_for_dev_replace) {
1530
		ret = -EPERM;
1531 1532 1533
		goto out_free;
	}

1534
	old_size = btrfs_device_get_total_bytes(device);
C
Christoph Hellwig 已提交
1535 1536 1537 1538

	if (mod < 0) {
		if (new_size > old_size) {
			ret = -EINVAL;
1539
			goto out_free;
C
Christoph Hellwig 已提交
1540 1541 1542
		}
		new_size = old_size - new_size;
	} else if (mod > 0) {
1543
		if (new_size > ULLONG_MAX - old_size) {
1544
			ret = -ERANGE;
1545 1546
			goto out_free;
		}
C
Christoph Hellwig 已提交
1547 1548 1549
		new_size = old_size + new_size;
	}

1550
	if (new_size < SZ_256M) {
C
Christoph Hellwig 已提交
1551
		ret = -EINVAL;
1552
		goto out_free;
C
Christoph Hellwig 已提交
1553 1554 1555
	}
	if (new_size > device->bdev->bd_inode->i_size) {
		ret = -EFBIG;
1556
		goto out_free;
C
Christoph Hellwig 已提交
1557 1558
	}

1559
	new_size = round_down(new_size, fs_info->sectorsize);
C
Christoph Hellwig 已提交
1560

1561 1562
	btrfs_info_in_rcu(fs_info, "new size for %s is %llu",
			  rcu_str_deref(device->name), new_size);
C
Christoph Hellwig 已提交
1563 1564

	if (new_size > old_size) {
1565
		trans = btrfs_start_transaction(root, 0);
1566 1567
		if (IS_ERR(trans)) {
			ret = PTR_ERR(trans);
1568
			goto out_free;
1569
		}
C
Christoph Hellwig 已提交
1570
		ret = btrfs_grow_device(trans, device, new_size);
1571
		btrfs_commit_transaction(trans);
1572
	} else if (new_size < old_size) {
C
Christoph Hellwig 已提交
1573
		ret = btrfs_shrink_device(device, new_size);
1574
	} /* equal, nothing need to do */
C
Christoph Hellwig 已提交
1575

1576
out_free:
C
Christoph Hellwig 已提交
1577
	kfree(vol_args);
1578
out:
1579
	mutex_unlock(&fs_info->volume_mutex);
1580
	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
1581
	mnt_drop_write_file(file);
C
Christoph Hellwig 已提交
1582 1583 1584
	return ret;
}

S
Sage Weil 已提交
1585
static noinline int btrfs_ioctl_snap_create_transid(struct file *file,
1586
				const char *name, unsigned long fd, int subvol,
A
Arne Jansen 已提交
1587
				u64 *transid, bool readonly,
1588
				struct btrfs_qgroup_inherit *inherit)
C
Christoph Hellwig 已提交
1589 1590
{
	int namelen;
1591
	int ret = 0;
C
Christoph Hellwig 已提交
1592

1593 1594 1595
	if (!S_ISDIR(file_inode(file)->i_mode))
		return -ENOTDIR;

1596 1597 1598 1599
	ret = mnt_want_write_file(file);
	if (ret)
		goto out;

S
Sage Weil 已提交
1600 1601
	namelen = strlen(name);
	if (strchr(name, '/')) {
C
Christoph Hellwig 已提交
1602
		ret = -EINVAL;
1603
		goto out_drop_write;
C
Christoph Hellwig 已提交
1604 1605
	}

1606 1607 1608
	if (name[0] == '.' &&
	   (namelen == 1 || (name[1] == '.' && namelen == 2))) {
		ret = -EEXIST;
1609
		goto out_drop_write;
1610 1611
	}

1612
	if (subvol) {
S
Sage Weil 已提交
1613
		ret = btrfs_mksubvol(&file->f_path, name, namelen,
A
Arne Jansen 已提交
1614
				     NULL, transid, readonly, inherit);
1615
	} else {
1616
		struct fd src = fdget(fd);
1617
		struct inode *src_inode;
1618
		if (!src.file) {
1619
			ret = -EINVAL;
1620
			goto out_drop_write;
1621 1622
		}

A
Al Viro 已提交
1623 1624
		src_inode = file_inode(src.file);
		if (src_inode->i_sb != file_inode(file)->i_sb) {
J
Josef Bacik 已提交
1625
			btrfs_info(BTRFS_I(file_inode(file))->root->fs_info,
1626
				   "Snapshot src from another FS");
1627
			ret = -EXDEV;
1628 1629 1630 1631 1632 1633
		} else if (!inode_owner_or_capable(src_inode)) {
			/*
			 * Subvolume creation is not restricted, but snapshots
			 * are limited to own subvolumes only
			 */
			ret = -EPERM;
1634 1635 1636 1637
		} else {
			ret = btrfs_mksubvol(&file->f_path, name, namelen,
					     BTRFS_I(src_inode)->root,
					     transid, readonly, inherit);
1638
		}
1639
		fdput(src);
1640
	}
1641 1642
out_drop_write:
	mnt_drop_write_file(file);
C
Christoph Hellwig 已提交
1643
out:
S
Sage Weil 已提交
1644 1645 1646 1647
	return ret;
}

static noinline int btrfs_ioctl_snap_create(struct file *file,
1648
					    void __user *arg, int subvol)
S
Sage Weil 已提交
1649
{
1650
	struct btrfs_ioctl_vol_args *vol_args;
S
Sage Weil 已提交
1651 1652
	int ret;

1653 1654 1655
	if (!S_ISDIR(file_inode(file)->i_mode))
		return -ENOTDIR;

1656 1657 1658 1659
	vol_args = memdup_user(arg, sizeof(*vol_args));
	if (IS_ERR(vol_args))
		return PTR_ERR(vol_args);
	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
S
Sage Weil 已提交
1660

1661
	ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
L
Li Zefan 已提交
1662
					      vol_args->fd, subvol,
A
Arne Jansen 已提交
1663
					      NULL, false, NULL);
1664

1665 1666 1667
	kfree(vol_args);
	return ret;
}
1668

1669 1670 1671 1672 1673 1674 1675
static noinline int btrfs_ioctl_snap_create_v2(struct file *file,
					       void __user *arg, int subvol)
{
	struct btrfs_ioctl_vol_args_v2 *vol_args;
	int ret;
	u64 transid = 0;
	u64 *ptr = NULL;
L
Li Zefan 已提交
1676
	bool readonly = false;
A
Arne Jansen 已提交
1677
	struct btrfs_qgroup_inherit *inherit = NULL;
1678

1679 1680 1681
	if (!S_ISDIR(file_inode(file)->i_mode))
		return -ENOTDIR;

1682 1683 1684 1685
	vol_args = memdup_user(arg, sizeof(*vol_args));
	if (IS_ERR(vol_args))
		return PTR_ERR(vol_args);
	vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
1686

L
Li Zefan 已提交
1687
	if (vol_args->flags &
A
Arne Jansen 已提交
1688 1689
	    ~(BTRFS_SUBVOL_CREATE_ASYNC | BTRFS_SUBVOL_RDONLY |
	      BTRFS_SUBVOL_QGROUP_INHERIT)) {
L
Li Zefan 已提交
1690
		ret = -EOPNOTSUPP;
D
Dan Carpenter 已提交
1691
		goto free_args;
S
Sage Weil 已提交
1692
	}
1693 1694 1695

	if (vol_args->flags & BTRFS_SUBVOL_CREATE_ASYNC)
		ptr = &transid;
L
Li Zefan 已提交
1696 1697
	if (vol_args->flags & BTRFS_SUBVOL_RDONLY)
		readonly = true;
A
Arne Jansen 已提交
1698
	if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) {
1699
		if (vol_args->size > PAGE_SIZE) {
A
Arne Jansen 已提交
1700
			ret = -EINVAL;
D
Dan Carpenter 已提交
1701
			goto free_args;
A
Arne Jansen 已提交
1702 1703 1704 1705
		}
		inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size);
		if (IS_ERR(inherit)) {
			ret = PTR_ERR(inherit);
D
Dan Carpenter 已提交
1706
			goto free_args;
A
Arne Jansen 已提交
1707 1708
		}
	}
1709 1710

	ret = btrfs_ioctl_snap_create_transid(file, vol_args->name,
A
Arne Jansen 已提交
1711
					      vol_args->fd, subvol, ptr,
1712
					      readonly, inherit);
D
Dan Carpenter 已提交
1713 1714
	if (ret)
		goto free_inherit;
1715

D
Dan Carpenter 已提交
1716 1717 1718 1719
	if (ptr && copy_to_user(arg +
				offsetof(struct btrfs_ioctl_vol_args_v2,
					transid),
				ptr, sizeof(*ptr)))
1720
		ret = -EFAULT;
D
Dan Carpenter 已提交
1721 1722

free_inherit:
A
Arne Jansen 已提交
1723
	kfree(inherit);
D
Dan Carpenter 已提交
1724 1725
free_args:
	kfree(vol_args);
C
Christoph Hellwig 已提交
1726 1727 1728
	return ret;
}

1729 1730 1731
static noinline int btrfs_ioctl_subvol_getflags(struct file *file,
						void __user *arg)
{
A
Al Viro 已提交
1732
	struct inode *inode = file_inode(file);
1733
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
1734 1735 1736 1737
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret = 0;
	u64 flags = 0;

1738
	if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID)
1739 1740
		return -EINVAL;

1741
	down_read(&fs_info->subvol_sem);
1742 1743
	if (btrfs_root_readonly(root))
		flags |= BTRFS_SUBVOL_RDONLY;
1744
	up_read(&fs_info->subvol_sem);
1745 1746 1747 1748 1749 1750 1751 1752 1753 1754

	if (copy_to_user(arg, &flags, sizeof(flags)))
		ret = -EFAULT;

	return ret;
}

static noinline int btrfs_ioctl_subvol_setflags(struct file *file,
					      void __user *arg)
{
A
Al Viro 已提交
1755
	struct inode *inode = file_inode(file);
1756
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
1757 1758 1759 1760 1761 1762
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;
	u64 root_flags;
	u64 flags;
	int ret = 0;

1763 1764 1765
	if (!inode_owner_or_capable(inode))
		return -EPERM;

1766 1767 1768
	ret = mnt_want_write_file(file);
	if (ret)
		goto out;
1769

1770
	if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) {
1771 1772 1773
		ret = -EINVAL;
		goto out_drop_write;
	}
1774

1775 1776 1777 1778
	if (copy_from_user(&flags, arg, sizeof(flags))) {
		ret = -EFAULT;
		goto out_drop_write;
	}
1779

1780 1781 1782 1783
	if (flags & BTRFS_SUBVOL_CREATE_ASYNC) {
		ret = -EINVAL;
		goto out_drop_write;
	}
1784

1785 1786 1787 1788
	if (flags & ~BTRFS_SUBVOL_RDONLY) {
		ret = -EOPNOTSUPP;
		goto out_drop_write;
	}
1789

1790
	down_write(&fs_info->subvol_sem);
1791 1792 1793

	/* nothing to do */
	if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root))
1794
		goto out_drop_sem;
1795 1796

	root_flags = btrfs_root_flags(&root->root_item);
1797
	if (flags & BTRFS_SUBVOL_RDONLY) {
1798 1799
		btrfs_set_root_flags(&root->root_item,
				     root_flags | BTRFS_ROOT_SUBVOL_RDONLY);
1800 1801 1802 1803 1804 1805 1806 1807
	} else {
		/*
		 * Block RO -> RW transition if this subvolume is involved in
		 * send
		 */
		spin_lock(&root->root_item_lock);
		if (root->send_in_progress == 0) {
			btrfs_set_root_flags(&root->root_item,
1808
				     root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY);
1809 1810 1811
			spin_unlock(&root->root_item_lock);
		} else {
			spin_unlock(&root->root_item_lock);
1812 1813 1814
			btrfs_warn(fs_info,
				   "Attempt to set subvolume %llu read-write during send",
				   root->root_key.objectid);
1815 1816 1817 1818
			ret = -EPERM;
			goto out_drop_sem;
		}
	}
1819 1820 1821 1822 1823 1824 1825

	trans = btrfs_start_transaction(root, 1);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out_reset;
	}

1826
	ret = btrfs_update_root(trans, fs_info->tree_root,
1827
				&root->root_key, &root->root_item);
1828 1829 1830 1831 1832 1833
	if (ret < 0) {
		btrfs_end_transaction(trans);
		goto out_reset;
	}

	ret = btrfs_commit_transaction(trans);
1834 1835 1836 1837

out_reset:
	if (ret)
		btrfs_set_root_flags(&root->root_item, root_flags);
1838
out_drop_sem:
1839
	up_write(&fs_info->subvol_sem);
1840 1841 1842
out_drop_write:
	mnt_drop_write_file(file);
out:
1843 1844 1845
	return ret;
}

1846 1847 1848 1849 1850
/*
 * helper to check if the subvolume references other subvolumes
 */
static noinline int may_destroy_subvol(struct btrfs_root *root)
{
1851
	struct btrfs_fs_info *fs_info = root->fs_info;
1852
	struct btrfs_path *path;
1853
	struct btrfs_dir_item *di;
1854
	struct btrfs_key key;
1855
	u64 dir_id;
1856 1857 1858 1859 1860 1861
	int ret;

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

1862
	/* Make sure this root isn't set as the default subvol */
1863 1864
	dir_id = btrfs_super_root_dir(fs_info->super_copy);
	di = btrfs_lookup_dir_item(NULL, fs_info->tree_root, path,
1865 1866 1867 1868
				   dir_id, "default", 7, 0);
	if (di && !IS_ERR(di)) {
		btrfs_dir_item_key_to_cpu(path->nodes[0], di, &key);
		if (key.objectid == root->root_key.objectid) {
1869
			ret = -EPERM;
1870
			btrfs_err(fs_info,
J
Jeff Mahoney 已提交
1871 1872
				  "deleting default subvolume %llu is not allowed",
				  key.objectid);
1873 1874 1875 1876 1877
			goto out;
		}
		btrfs_release_path(path);
	}

1878 1879 1880 1881
	key.objectid = root->root_key.objectid;
	key.type = BTRFS_ROOT_REF_KEY;
	key.offset = (u64)-1;

1882
	ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0);
1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899
	if (ret < 0)
		goto out;
	BUG_ON(ret == 0);

	ret = 0;
	if (path->slots[0] > 0) {
		path->slots[0]--;
		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
		if (key.objectid == root->root_key.objectid &&
		    key.type == BTRFS_ROOT_REF_KEY)
			ret = -ENOTEMPTY;
	}
out:
	btrfs_free_path(path);
	return ret;
}

1900 1901 1902
static noinline int key_in_sk(struct btrfs_key *key,
			      struct btrfs_ioctl_search_key *sk)
{
1903 1904 1905 1906 1907 1908 1909 1910 1911
	struct btrfs_key test;
	int ret;

	test.objectid = sk->min_objectid;
	test.type = sk->min_type;
	test.offset = sk->min_offset;

	ret = btrfs_comp_cpu_keys(key, &test);
	if (ret < 0)
1912
		return 0;
1913 1914 1915 1916 1917 1918 1919

	test.objectid = sk->max_objectid;
	test.type = sk->max_type;
	test.offset = sk->max_offset;

	ret = btrfs_comp_cpu_keys(key, &test);
	if (ret > 0)
1920 1921 1922 1923
		return 0;
	return 1;
}

1924
static noinline int copy_to_sk(struct btrfs_path *path,
1925 1926
			       struct btrfs_key *key,
			       struct btrfs_ioctl_search_key *sk,
1927
			       size_t *buf_size,
1928
			       char __user *ubuf,
1929 1930 1931 1932 1933 1934
			       unsigned long *sk_offset,
			       int *num_found)
{
	u64 found_transid;
	struct extent_buffer *leaf;
	struct btrfs_ioctl_search_header sh;
1935
	struct btrfs_key test;
1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956
	unsigned long item_off;
	unsigned long item_len;
	int nritems;
	int i;
	int slot;
	int ret = 0;

	leaf = path->nodes[0];
	slot = path->slots[0];
	nritems = btrfs_header_nritems(leaf);

	if (btrfs_header_generation(leaf) > sk->max_transid) {
		i = nritems;
		goto advance_key;
	}
	found_transid = btrfs_header_generation(leaf);

	for (i = slot; i < nritems; i++) {
		item_off = btrfs_item_ptr_offset(leaf, i);
		item_len = btrfs_item_size_nr(leaf, i);

1957 1958 1959 1960
		btrfs_item_key_to_cpu(leaf, key, i);
		if (!key_in_sk(key, sk))
			continue;

1961
		if (sizeof(sh) + item_len > *buf_size) {
1962 1963 1964 1965 1966 1967 1968 1969 1970 1971
			if (*num_found) {
				ret = 1;
				goto out;
			}

			/*
			 * return one empty item back for v1, which does not
			 * handle -EOVERFLOW
			 */

1972
			*buf_size = sizeof(sh) + item_len;
1973
			item_len = 0;
1974 1975
			ret = -EOVERFLOW;
		}
1976

1977
		if (sizeof(sh) + item_len + *sk_offset > *buf_size) {
1978
			ret = 1;
1979
			goto out;
1980 1981 1982 1983 1984 1985 1986 1987 1988
		}

		sh.objectid = key->objectid;
		sh.offset = key->offset;
		sh.type = key->type;
		sh.len = item_len;
		sh.transid = found_transid;

		/* copy search result header */
1989 1990 1991 1992 1993
		if (copy_to_user(ubuf + *sk_offset, &sh, sizeof(sh))) {
			ret = -EFAULT;
			goto out;
		}

1994 1995 1996
		*sk_offset += sizeof(sh);

		if (item_len) {
1997
			char __user *up = ubuf + *sk_offset;
1998
			/* copy the item */
1999 2000 2001 2002 2003 2004
			if (read_extent_buffer_to_user(leaf, up,
						       item_off, item_len)) {
				ret = -EFAULT;
				goto out;
			}

2005 2006
			*sk_offset += item_len;
		}
2007
		(*num_found)++;
2008

2009 2010 2011
		if (ret) /* -EOVERFLOW from above */
			goto out;

2012 2013 2014 2015
		if (*num_found >= sk->nr_items) {
			ret = 1;
			goto out;
		}
2016 2017
	}
advance_key:
2018
	ret = 0;
2019 2020 2021 2022 2023 2024
	test.objectid = sk->max_objectid;
	test.type = sk->max_type;
	test.offset = sk->max_offset;
	if (btrfs_comp_cpu_keys(key, &test) >= 0)
		ret = 1;
	else if (key->offset < (u64)-1)
2025
		key->offset++;
2026
	else if (key->type < (u8)-1) {
2027
		key->offset = 0;
2028
		key->type++;
2029
	} else if (key->objectid < (u64)-1) {
2030 2031
		key->offset = 0;
		key->type = 0;
2032
		key->objectid++;
2033 2034
	} else
		ret = 1;
2035
out:
2036 2037 2038 2039 2040 2041 2042 2043 2044
	/*
	 *  0: all items from this leaf copied, continue with next
	 *  1: * more items can be copied, but unused buffer is too small
	 *     * all items were found
	 *     Either way, it will stops the loop which iterates to the next
	 *     leaf
	 *  -EOVERFLOW: item was to large for buffer
	 *  -EFAULT: could not copy extent buffer back to userspace
	 */
2045 2046 2047 2048
	return ret;
}

static noinline int search_ioctl(struct inode *inode,
2049
				 struct btrfs_ioctl_search_key *sk,
2050
				 size_t *buf_size,
2051
				 char __user *ubuf)
2052
{
2053
	struct btrfs_fs_info *info = btrfs_sb(inode->i_sb);
2054 2055 2056 2057 2058 2059 2060
	struct btrfs_root *root;
	struct btrfs_key key;
	struct btrfs_path *path;
	int ret;
	int num_found = 0;
	unsigned long sk_offset = 0;

2061 2062
	if (*buf_size < sizeof(struct btrfs_ioctl_search_header)) {
		*buf_size = sizeof(struct btrfs_ioctl_search_header);
2063
		return -EOVERFLOW;
2064
	}
2065

2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087
	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

	if (sk->tree_id == 0) {
		/* search the root of the inode that was passed */
		root = BTRFS_I(inode)->root;
	} else {
		key.objectid = sk->tree_id;
		key.type = BTRFS_ROOT_ITEM_KEY;
		key.offset = (u64)-1;
		root = btrfs_read_fs_root_no_name(info, &key);
		if (IS_ERR(root)) {
			btrfs_free_path(path);
			return -ENOENT;
		}
	}

	key.objectid = sk->min_objectid;
	key.type = sk->min_type;
	key.offset = sk->min_offset;

2088
	while (1) {
2089
		ret = btrfs_search_forward(root, &key, path, sk->min_transid);
2090 2091 2092 2093 2094
		if (ret != 0) {
			if (ret > 0)
				ret = 0;
			goto err;
		}
2095
		ret = copy_to_sk(path, &key, sk, buf_size, ubuf,
2096
				 &sk_offset, &num_found);
2097
		btrfs_release_path(path);
2098
		if (ret)
2099 2100 2101
			break;

	}
2102 2103
	if (ret > 0)
		ret = 0;
2104 2105 2106 2107 2108 2109 2110 2111 2112
err:
	sk->nr_items = num_found;
	btrfs_free_path(path);
	return ret;
}

static noinline int btrfs_ioctl_tree_search(struct file *file,
					   void __user *argp)
{
2113 2114
	struct btrfs_ioctl_search_args __user *uargs;
	struct btrfs_ioctl_search_key sk;
2115 2116 2117
	struct inode *inode;
	int ret;
	size_t buf_size;
2118 2119 2120 2121

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

2122 2123 2124 2125
	uargs = (struct btrfs_ioctl_search_args __user *)argp;

	if (copy_from_user(&sk, &uargs->key, sizeof(sk)))
		return -EFAULT;
2126

2127
	buf_size = sizeof(uargs->buf);
2128

A
Al Viro 已提交
2129
	inode = file_inode(file);
2130
	ret = search_ioctl(inode, &sk, &buf_size, uargs->buf);
2131 2132 2133 2134 2135 2136 2137 2138

	/*
	 * In the origin implementation an overflow is handled by returning a
	 * search header with a len of zero, so reset ret.
	 */
	if (ret == -EOVERFLOW)
		ret = 0;

2139
	if (ret == 0 && copy_to_user(&uargs->key, &sk, sizeof(sk)))
2140 2141 2142 2143
		ret = -EFAULT;
	return ret;
}

G
Gerhard Heift 已提交
2144 2145 2146 2147 2148 2149 2150 2151
static noinline int btrfs_ioctl_tree_search_v2(struct file *file,
					       void __user *argp)
{
	struct btrfs_ioctl_search_args_v2 __user *uarg;
	struct btrfs_ioctl_search_args_v2 args;
	struct inode *inode;
	int ret;
	size_t buf_size;
2152
	const size_t buf_limit = SZ_16M;
G
Gerhard Heift 已提交
2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169

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

	/* copy search header and buffer size */
	uarg = (struct btrfs_ioctl_search_args_v2 __user *)argp;
	if (copy_from_user(&args, uarg, sizeof(args)))
		return -EFAULT;

	buf_size = args.buf_size;

	/* limit result size to 16MB */
	if (buf_size > buf_limit)
		buf_size = buf_limit;

	inode = file_inode(file);
	ret = search_ioctl(inode, &args.key, &buf_size,
2170
			   (char __user *)(&uarg->buf[0]));
G
Gerhard Heift 已提交
2171 2172 2173 2174 2175 2176
	if (ret == 0 && copy_to_user(&uarg->key, &args.key, sizeof(args.key)))
		ret = -EFAULT;
	else if (ret == -EOVERFLOW &&
		copy_to_user(&uarg->buf_size, &buf_size, sizeof(buf_size)))
		ret = -EFAULT;

2177 2178 2179
	return ret;
}

2180
/*
2181 2182 2183
 * Search INODE_REFs to identify path name of 'dirid' directory
 * in a 'tree_id' tree. and sets path name to 'name'.
 */
2184 2185 2186 2187 2188
static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info,
				u64 tree_id, u64 dirid, char *name)
{
	struct btrfs_root *root;
	struct btrfs_key key;
2189
	char *ptr;
2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206
	int ret = -1;
	int slot;
	int len;
	int total_len = 0;
	struct btrfs_inode_ref *iref;
	struct extent_buffer *l;
	struct btrfs_path *path;

	if (dirid == BTRFS_FIRST_FREE_OBJECTID) {
		name[0]='\0';
		return 0;
	}

	path = btrfs_alloc_path();
	if (!path)
		return -ENOMEM;

2207
	ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX];
2208 2209 2210 2211 2212 2213

	key.objectid = tree_id;
	key.type = BTRFS_ROOT_ITEM_KEY;
	key.offset = (u64)-1;
	root = btrfs_read_fs_root_no_name(info, &key);
	if (IS_ERR(root)) {
2214
		btrfs_err(info, "could not find root %llu", tree_id);
2215 2216
		ret = -ENOENT;
		goto out;
2217 2218 2219 2220
	}

	key.objectid = dirid;
	key.type = BTRFS_INODE_REF_KEY;
2221
	key.offset = (u64)-1;
2222

2223
	while (1) {
2224 2225 2226
		ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
		if (ret < 0)
			goto out;
2227 2228 2229 2230 2231 2232 2233 2234 2235 2236
		else if (ret > 0) {
			ret = btrfs_previous_item(root, path, dirid,
						  BTRFS_INODE_REF_KEY);
			if (ret < 0)
				goto out;
			else if (ret > 0) {
				ret = -ENOENT;
				goto out;
			}
		}
2237 2238 2239 2240 2241 2242 2243 2244 2245

		l = path->nodes[0];
		slot = path->slots[0];
		btrfs_item_key_to_cpu(l, &key, slot);

		iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref);
		len = btrfs_inode_ref_name_len(l, iref);
		ptr -= len + 1;
		total_len += len + 1;
2246 2247
		if (ptr < name) {
			ret = -ENAMETOOLONG;
2248
			goto out;
2249
		}
2250 2251

		*(ptr + len) = '/';
2252
		read_extent_buffer(l, ptr, (unsigned long)(iref + 1), len);
2253 2254 2255 2256

		if (key.offset == BTRFS_FIRST_FREE_OBJECTID)
			break;

2257
		btrfs_release_path(path);
2258
		key.objectid = key.offset;
2259
		key.offset = (u64)-1;
2260 2261
		dirid = key.objectid;
	}
2262
	memmove(name, ptr, total_len);
2263
	name[total_len] = '\0';
2264 2265 2266
	ret = 0;
out:
	btrfs_free_path(path);
2267 2268 2269 2270 2271 2272 2273 2274
	return ret;
}

static noinline int btrfs_ioctl_ino_lookup(struct file *file,
					   void __user *argp)
{
	 struct btrfs_ioctl_ino_lookup_args *args;
	 struct inode *inode;
2275
	int ret = 0;
2276

J
Julia Lawall 已提交
2277 2278 2279
	args = memdup_user(argp, sizeof(*args));
	if (IS_ERR(args))
		return PTR_ERR(args);
2280

A
Al Viro 已提交
2281
	inode = file_inode(file);
2282

2283 2284 2285 2286
	/*
	 * Unprivileged query to obtain the containing subvolume root id. The
	 * path is reset so it's consistent with btrfs_search_path_in_tree.
	 */
2287 2288 2289
	if (args->treeid == 0)
		args->treeid = BTRFS_I(inode)->root->root_key.objectid;

2290 2291 2292 2293 2294 2295 2296 2297 2298 2299
	if (args->objectid == BTRFS_FIRST_FREE_OBJECTID) {
		args->name[0] = 0;
		goto out;
	}

	if (!capable(CAP_SYS_ADMIN)) {
		ret = -EPERM;
		goto out;
	}

2300 2301 2302 2303
	ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info,
					args->treeid, args->objectid,
					args->name);

2304
out:
2305 2306 2307 2308
	if (ret == 0 && copy_to_user(argp, args, sizeof(*args)))
		ret = -EFAULT;

	kfree(args);
2309 2310 2311
	return ret;
}

2312 2313 2314
static noinline int btrfs_ioctl_snap_destroy(struct file *file,
					     void __user *arg)
{
A
Al Viro 已提交
2315
	struct dentry *parent = file->f_path.dentry;
2316
	struct btrfs_fs_info *fs_info = btrfs_sb(parent->d_sb);
2317
	struct dentry *dentry;
2318
	struct inode *dir = d_inode(parent);
2319 2320 2321 2322 2323
	struct inode *inode;
	struct btrfs_root *root = BTRFS_I(dir)->root;
	struct btrfs_root *dest = NULL;
	struct btrfs_ioctl_vol_args *vol_args;
	struct btrfs_trans_handle *trans;
2324
	struct btrfs_block_rsv block_rsv;
2325
	u64 root_flags;
2326
	u64 qgroup_reserved;
2327 2328 2329 2330
	int namelen;
	int ret;
	int err = 0;

2331 2332 2333
	if (!S_ISDIR(dir->i_mode))
		return -ENOTDIR;

2334 2335 2336 2337 2338 2339 2340 2341 2342 2343 2344 2345
	vol_args = memdup_user(arg, sizeof(*vol_args));
	if (IS_ERR(vol_args))
		return PTR_ERR(vol_args);

	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
	namelen = strlen(vol_args->name);
	if (strchr(vol_args->name, '/') ||
	    strncmp(vol_args->name, "..", namelen) == 0) {
		err = -EINVAL;
		goto out;
	}

2346
	err = mnt_want_write_file(file);
2347 2348 2349
	if (err)
		goto out;

2350

2351 2352 2353
	err = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT);
	if (err == -EINTR)
		goto out_drop_write;
2354 2355 2356 2357 2358 2359
	dentry = lookup_one_len(vol_args->name, parent, namelen);
	if (IS_ERR(dentry)) {
		err = PTR_ERR(dentry);
		goto out_unlock_dir;
	}

2360
	if (d_really_is_negative(dentry)) {
2361 2362 2363 2364
		err = -ENOENT;
		goto out_dput;
	}

2365
	inode = d_inode(dentry);
2366
	dest = BTRFS_I(inode)->root;
2367
	if (!capable(CAP_SYS_ADMIN)) {
2368 2369 2370 2371 2372 2373 2374 2375 2376 2377 2378 2379 2380 2381
		/*
		 * Regular user.  Only allow this with a special mount
		 * option, when the user has write+exec access to the
		 * subvol root, and when rmdir(2) would have been
		 * allowed.
		 *
		 * Note that this is _not_ check that the subvol is
		 * empty or doesn't contain data that we wouldn't
		 * otherwise be able to delete.
		 *
		 * Users who want to delete empty subvols should try
		 * rmdir(2).
		 */
		err = -EPERM;
2382
		if (!btrfs_test_opt(fs_info, USER_SUBVOL_RM_ALLOWED))
2383 2384 2385 2386 2387 2388 2389 2390 2391 2392 2393 2394 2395 2396 2397 2398 2399 2400
			goto out_dput;

		/*
		 * Do not allow deletion if the parent dir is the same
		 * as the dir to be deleted.  That means the ioctl
		 * must be called on the dentry referencing the root
		 * of the subvol, not a random directory contained
		 * within it.
		 */
		err = -EINVAL;
		if (root == dest)
			goto out_dput;

		err = inode_permission(inode, MAY_WRITE | MAY_EXEC);
		if (err)
			goto out_dput;
	}

2401 2402 2403 2404 2405
	/* check if subvolume may be deleted by a user */
	err = btrfs_may_delete(dir, dentry, 1);
	if (err)
		goto out_dput;

2406
	if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) {
2407 2408 2409 2410
		err = -EINVAL;
		goto out_dput;
	}

A
Al Viro 已提交
2411
	inode_lock(inode);
2412 2413 2414 2415 2416 2417 2418

	/*
	 * Don't allow to delete a subvolume with send in progress. This is
	 * inside the i_mutex so the error handling that has to drop the bit
	 * again is not run concurrently.
	 */
	spin_lock(&dest->root_item_lock);
2419 2420 2421
	root_flags = btrfs_root_flags(&dest->root_item);
	if (dest->send_in_progress == 0) {
		btrfs_set_root_flags(&dest->root_item,
2422 2423 2424 2425
				root_flags | BTRFS_ROOT_SUBVOL_DEAD);
		spin_unlock(&dest->root_item_lock);
	} else {
		spin_unlock(&dest->root_item_lock);
2426 2427 2428
		btrfs_warn(fs_info,
			   "Attempt to delete subvolume %llu during send",
			   dest->root_key.objectid);
2429
		err = -EPERM;
2430
		goto out_unlock_inode;
2431 2432
	}

2433
	down_write(&fs_info->subvol_sem);
2434 2435 2436 2437 2438

	err = may_destroy_subvol(dest);
	if (err)
		goto out_up_write;

2439 2440 2441 2442 2443 2444
	btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP);
	/*
	 * One for dir inode, two for dir entries, two for root
	 * ref/backref.
	 */
	err = btrfs_subvolume_reserve_metadata(root, &block_rsv,
2445
					       5, &qgroup_reserved, true);
2446 2447 2448
	if (err)
		goto out_up_write;

2449 2450 2451
	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		err = PTR_ERR(trans);
2452
		goto out_release;
2453
	}
2454 2455
	trans->block_rsv = &block_rsv;
	trans->bytes_reserved = block_rsv.size;
2456

2457
	btrfs_record_snapshot_destroy(trans, BTRFS_I(dir));
2458

2459 2460 2461 2462
	ret = btrfs_unlink_subvol(trans, root, dir,
				dest->root_key.objectid,
				dentry->d_name.name,
				dentry->d_name.len);
2463 2464
	if (ret) {
		err = ret;
2465
		btrfs_abort_transaction(trans, ret);
2466 2467
		goto out_end_trans;
	}
2468 2469 2470 2471 2472 2473 2474 2475

	btrfs_record_root_in_trans(trans, dest);

	memset(&dest->root_item.drop_progress, 0,
		sizeof(dest->root_item.drop_progress));
	dest->root_item.drop_level = 0;
	btrfs_set_root_refs(&dest->root_item, 0);

2476
	if (!test_and_set_bit(BTRFS_ROOT_ORPHAN_ITEM_INSERTED, &dest->state)) {
2477
		ret = btrfs_insert_orphan_item(trans,
2478
					fs_info->tree_root,
2479
					dest->root_key.objectid);
2480
		if (ret) {
2481
			btrfs_abort_transaction(trans, ret);
2482 2483 2484
			err = ret;
			goto out_end_trans;
		}
2485
	}
2486

2487
	ret = btrfs_uuid_tree_rem(trans, fs_info, dest->root_item.uuid,
2488
				  BTRFS_UUID_KEY_SUBVOL,
2489 2490
				  dest->root_key.objectid);
	if (ret && ret != -ENOENT) {
2491
		btrfs_abort_transaction(trans, ret);
2492 2493 2494 2495
		err = ret;
		goto out_end_trans;
	}
	if (!btrfs_is_empty_uuid(dest->root_item.received_uuid)) {
2496
		ret = btrfs_uuid_tree_rem(trans, fs_info,
2497 2498 2499 2500
					  dest->root_item.received_uuid,
					  BTRFS_UUID_KEY_RECEIVED_SUBVOL,
					  dest->root_key.objectid);
		if (ret && ret != -ENOENT) {
2501
			btrfs_abort_transaction(trans, ret);
2502 2503 2504 2505 2506
			err = ret;
			goto out_end_trans;
		}
	}

2507
out_end_trans:
2508 2509
	trans->block_rsv = NULL;
	trans->bytes_reserved = 0;
2510
	ret = btrfs_end_transaction(trans);
2511 2512
	if (ret && !err)
		err = ret;
2513
	inode->i_flags |= S_DEAD;
2514
out_release:
2515
	btrfs_subvolume_release_metadata(fs_info, &block_rsv);
2516
out_up_write:
2517
	up_write(&fs_info->subvol_sem);
2518 2519
	if (err) {
		spin_lock(&dest->root_item_lock);
2520 2521
		root_flags = btrfs_root_flags(&dest->root_item);
		btrfs_set_root_flags(&dest->root_item,
2522 2523 2524
				root_flags & ~BTRFS_ROOT_SUBVOL_DEAD);
		spin_unlock(&dest->root_item_lock);
	}
2525
out_unlock_inode:
A
Al Viro 已提交
2526
	inode_unlock(inode);
2527
	if (!err) {
2528
		d_invalidate(dentry);
2529 2530
		btrfs_invalidate_inodes(dest);
		d_delete(dentry);
2531
		ASSERT(dest->send_in_progress == 0);
2532 2533

		/* the last ref */
2534 2535 2536
		if (dest->ino_cache_inode) {
			iput(dest->ino_cache_inode);
			dest->ino_cache_inode = NULL;
2537
		}
2538 2539 2540 2541
	}
out_dput:
	dput(dentry);
out_unlock_dir:
A
Al Viro 已提交
2542
	inode_unlock(dir);
2543
out_drop_write:
A
Al Viro 已提交
2544
	mnt_drop_write_file(file);
2545 2546 2547 2548 2549
out:
	kfree(vol_args);
	return err;
}

C
Chris Mason 已提交
2550
static int btrfs_ioctl_defrag(struct file *file, void __user *argp)
C
Christoph Hellwig 已提交
2551
{
A
Al Viro 已提交
2552
	struct inode *inode = file_inode(file);
C
Christoph Hellwig 已提交
2553
	struct btrfs_root *root = BTRFS_I(inode)->root;
C
Chris Mason 已提交
2554
	struct btrfs_ioctl_defrag_range_args *range;
Y
Yan Zheng 已提交
2555 2556
	int ret;

2557 2558 2559
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
L
Li Zefan 已提交
2560

2561 2562 2563
	if (btrfs_root_readonly(root)) {
		ret = -EROFS;
		goto out;
2564
	}
C
Christoph Hellwig 已提交
2565 2566 2567

	switch (inode->i_mode & S_IFMT) {
	case S_IFDIR:
2568 2569 2570 2571
		if (!capable(CAP_SYS_ADMIN)) {
			ret = -EPERM;
			goto out;
		}
2572
		ret = btrfs_defrag_root(root);
C
Christoph Hellwig 已提交
2573 2574
		break;
	case S_IFREG:
2575 2576 2577 2578
		if (!(file->f_mode & FMODE_WRITE)) {
			ret = -EINVAL;
			goto out;
		}
C
Chris Mason 已提交
2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590

		range = kzalloc(sizeof(*range), GFP_KERNEL);
		if (!range) {
			ret = -ENOMEM;
			goto out;
		}

		if (argp) {
			if (copy_from_user(range, argp,
					   sizeof(*range))) {
				ret = -EFAULT;
				kfree(range);
2591
				goto out;
C
Chris Mason 已提交
2592 2593 2594 2595 2596 2597 2598 2599 2600 2601
			}
			/* compression requires us to start the IO */
			if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) {
				range->flags |= BTRFS_DEFRAG_RANGE_START_IO;
				range->extent_thresh = (u32)-1;
			}
		} else {
			/* the rest are all set to zero by kzalloc */
			range->len = (u64)-1;
		}
A
Al Viro 已提交
2602
		ret = btrfs_defrag_file(file_inode(file), file,
C
Chris Mason 已提交
2603 2604 2605
					range, 0, 0);
		if (ret > 0)
			ret = 0;
C
Chris Mason 已提交
2606
		kfree(range);
C
Christoph Hellwig 已提交
2607
		break;
2608 2609
	default:
		ret = -EINVAL;
C
Christoph Hellwig 已提交
2610
	}
2611
out:
2612
	mnt_drop_write_file(file);
2613
	return ret;
C
Christoph Hellwig 已提交
2614 2615
}

2616
static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg)
C
Christoph Hellwig 已提交
2617 2618 2619 2620
{
	struct btrfs_ioctl_vol_args *vol_args;
	int ret;

2621 2622 2623
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2624
	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags))
2625
		return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
2626

2627
	mutex_lock(&fs_info->volume_mutex);
L
Li Zefan 已提交
2628
	vol_args = memdup_user(arg, sizeof(*vol_args));
2629 2630 2631 2632
	if (IS_ERR(vol_args)) {
		ret = PTR_ERR(vol_args);
		goto out;
	}
C
Christoph Hellwig 已提交
2633

2634
	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2635
	ret = btrfs_init_new_device(fs_info, vol_args->name);
C
Christoph Hellwig 已提交
2636

A
Anand Jain 已提交
2637
	if (!ret)
2638
		btrfs_info(fs_info, "disk added %s", vol_args->name);
A
Anand Jain 已提交
2639

C
Christoph Hellwig 已提交
2640
	kfree(vol_args);
2641
out:
2642
	mutex_unlock(&fs_info->volume_mutex);
2643
	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
C
Christoph Hellwig 已提交
2644 2645 2646
	return ret;
}

2647
static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg)
C
Christoph Hellwig 已提交
2648
{
2649 2650
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
2651
	struct btrfs_ioctl_vol_args_v2 *vol_args;
C
Christoph Hellwig 已提交
2652 2653
	int ret;

2654 2655 2656
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2657 2658 2659
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
Y
Yan Zheng 已提交
2660

L
Li Zefan 已提交
2661
	vol_args = memdup_user(arg, sizeof(*vol_args));
2662 2663
	if (IS_ERR(vol_args)) {
		ret = PTR_ERR(vol_args);
D
Dan Carpenter 已提交
2664
		goto err_drop;
2665
	}
C
Christoph Hellwig 已提交
2666

2667
	/* Check for compatibility reject unknown flags */
2668 2669
	if (vol_args->flags & ~BTRFS_VOL_ARG_V2_FLAGS_SUPPORTED)
		return -EOPNOTSUPP;
C
Christoph Hellwig 已提交
2670

2671
	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
2672 2673 2674 2675
		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
		goto out;
	}

2676
	mutex_lock(&fs_info->volume_mutex);
2677
	if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) {
2678
		ret = btrfs_rm_device(fs_info, NULL, vol_args->devid);
2679 2680
	} else {
		vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0';
2681
		ret = btrfs_rm_device(fs_info, vol_args->name, 0);
2682
	}
2683
	mutex_unlock(&fs_info->volume_mutex);
2684
	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
2685

2686
	if (!ret) {
2687
		if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID)
2688
			btrfs_info(fs_info, "device deleted: id %llu",
2689 2690
					vol_args->devid);
		else
2691
			btrfs_info(fs_info, "device deleted: %s",
2692 2693
					vol_args->name);
	}
2694 2695
out:
	kfree(vol_args);
D
Dan Carpenter 已提交
2696
err_drop:
2697
	mnt_drop_write_file(file);
C
Christoph Hellwig 已提交
2698 2699 2700
	return ret;
}

2701
static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg)
C
Christoph Hellwig 已提交
2702
{
2703 2704
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
C
Christoph Hellwig 已提交
2705 2706 2707
	struct btrfs_ioctl_vol_args *vol_args;
	int ret;

2708 2709 2710
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

2711 2712 2713
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
Y
Yan Zheng 已提交
2714

2715
	if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
2716
		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
2717 2718 2719 2720 2721 2722
		goto out_drop_write;
	}

	vol_args = memdup_user(arg, sizeof(*vol_args));
	if (IS_ERR(vol_args)) {
		ret = PTR_ERR(vol_args);
2723 2724 2725
		goto out;
	}

2726
	vol_args->name[BTRFS_PATH_NAME_MAX] = '\0';
2727
	mutex_lock(&fs_info->volume_mutex);
2728
	ret = btrfs_rm_device(fs_info, vol_args->name, 0);
2729
	mutex_unlock(&fs_info->volume_mutex);
2730

2731
	if (!ret)
2732
		btrfs_info(fs_info, "disk deleted %s", vol_args->name);
2733
	kfree(vol_args);
2734
out:
2735
	clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
2736
out_drop_write:
2737
	mnt_drop_write_file(file);
2738

C
Christoph Hellwig 已提交
2739 2740 2741
	return ret;
}

2742 2743
static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info,
				void __user *arg)
J
Jan Schmidt 已提交
2744
{
2745
	struct btrfs_ioctl_fs_info_args *fi_args;
J
Jan Schmidt 已提交
2746
	struct btrfs_device *device;
2747
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
2748
	int ret = 0;
J
Jan Schmidt 已提交
2749

2750 2751 2752 2753
	fi_args = kzalloc(sizeof(*fi_args), GFP_KERNEL);
	if (!fi_args)
		return -ENOMEM;

2754
	mutex_lock(&fs_devices->device_list_mutex);
2755
	fi_args->num_devices = fs_devices->num_devices;
2756
	memcpy(&fi_args->fsid, fs_info->fsid, sizeof(fi_args->fsid));
J
Jan Schmidt 已提交
2757

2758
	list_for_each_entry(device, &fs_devices->devices, dev_list) {
2759 2760
		if (device->devid > fi_args->max_id)
			fi_args->max_id = device->devid;
J
Jan Schmidt 已提交
2761 2762 2763
	}
	mutex_unlock(&fs_devices->device_list_mutex);

2764 2765 2766
	fi_args->nodesize = fs_info->nodesize;
	fi_args->sectorsize = fs_info->sectorsize;
	fi_args->clone_alignment = fs_info->sectorsize;
2767

2768 2769
	if (copy_to_user(arg, fi_args, sizeof(*fi_args)))
		ret = -EFAULT;
J
Jan Schmidt 已提交
2770

2771 2772
	kfree(fi_args);
	return ret;
J
Jan Schmidt 已提交
2773 2774
}

2775 2776
static long btrfs_ioctl_dev_info(struct btrfs_fs_info *fs_info,
				 void __user *arg)
J
Jan Schmidt 已提交
2777 2778 2779
{
	struct btrfs_ioctl_dev_info_args *di_args;
	struct btrfs_device *dev;
2780
	struct btrfs_fs_devices *fs_devices = fs_info->fs_devices;
J
Jan Schmidt 已提交
2781 2782 2783 2784 2785 2786 2787
	int ret = 0;
	char *s_uuid = NULL;

	di_args = memdup_user(arg, sizeof(*di_args));
	if (IS_ERR(di_args))
		return PTR_ERR(di_args);

2788
	if (!btrfs_is_empty_uuid(di_args->uuid))
J
Jan Schmidt 已提交
2789 2790 2791
		s_uuid = di_args->uuid;

	mutex_lock(&fs_devices->device_list_mutex);
2792
	dev = btrfs_find_device(fs_info, di_args->devid, s_uuid, NULL);
J
Jan Schmidt 已提交
2793 2794 2795 2796 2797 2798 2799

	if (!dev) {
		ret = -ENODEV;
		goto out;
	}

	di_args->devid = dev->devid;
2800 2801
	di_args->bytes_used = btrfs_device_get_bytes_used(dev);
	di_args->total_bytes = btrfs_device_get_total_bytes(dev);
J
Jan Schmidt 已提交
2802
	memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid));
2803
	if (dev->name) {
2804 2805 2806 2807 2808 2809
		struct rcu_string *name;

		rcu_read_lock();
		name = rcu_dereference(dev->name);
		strncpy(di_args->path, name->str, sizeof(di_args->path));
		rcu_read_unlock();
2810 2811
		di_args->path[sizeof(di_args->path) - 1] = 0;
	} else {
2812
		di_args->path[0] = '\0';
2813
	}
J
Jan Schmidt 已提交
2814 2815

out:
2816
	mutex_unlock(&fs_devices->device_list_mutex);
J
Jan Schmidt 已提交
2817 2818 2819 2820 2821 2822 2823
	if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args)))
		ret = -EFAULT;

	kfree(di_args);
	return ret;
}

2824
static struct page *extent_same_get_page(struct inode *inode, pgoff_t index)
M
Mark Fasheh 已提交
2825 2826 2827 2828 2829
{
	struct page *page;

	page = grab_cache_page(inode->i_mapping, index);
	if (!page)
2830
		return ERR_PTR(-ENOMEM);
M
Mark Fasheh 已提交
2831 2832

	if (!PageUptodate(page)) {
2833 2834 2835 2836 2837
		int ret;

		ret = btrfs_readpage(NULL, page);
		if (ret)
			return ERR_PTR(ret);
M
Mark Fasheh 已提交
2838 2839 2840
		lock_page(page);
		if (!PageUptodate(page)) {
			unlock_page(page);
2841
			put_page(page);
2842 2843 2844 2845
			return ERR_PTR(-EIO);
		}
		if (page->mapping != inode->i_mapping) {
			unlock_page(page);
2846
			put_page(page);
2847
			return ERR_PTR(-EAGAIN);
M
Mark Fasheh 已提交
2848 2849 2850 2851 2852 2853
		}
	}

	return page;
}

2854 2855 2856 2857
static int gather_extent_pages(struct inode *inode, struct page **pages,
			       int num_pages, u64 off)
{
	int i;
2858
	pgoff_t index = off >> PAGE_SHIFT;
2859 2860

	for (i = 0; i < num_pages; i++) {
2861
again:
2862
		pages[i] = extent_same_get_page(inode, index + i);
2863 2864 2865 2866 2867 2868 2869 2870
		if (IS_ERR(pages[i])) {
			int err = PTR_ERR(pages[i]);

			if (err == -EAGAIN)
				goto again;
			pages[i] = NULL;
			return err;
		}
2871 2872 2873 2874
	}
	return 0;
}

2875 2876
static int lock_extent_range(struct inode *inode, u64 off, u64 len,
			     bool retry_range_locking)
2877
{
2878 2879 2880 2881 2882 2883 2884 2885
	/*
	 * Do any pending delalloc/csum calculations on inode, one way or
	 * another, and lock file content.
	 * The locking order is:
	 *
	 *   1) pages
	 *   2) range in the inode's io tree
	 */
2886 2887 2888 2889 2890
	while (1) {
		struct btrfs_ordered_extent *ordered;
		lock_extent(&BTRFS_I(inode)->io_tree, off, off + len - 1);
		ordered = btrfs_lookup_first_ordered_extent(inode,
							    off + len - 1);
2891 2892 2893
		if ((!ordered ||
		     ordered->file_offset + ordered->len <= off ||
		     ordered->file_offset >= off + len) &&
2894
		    !test_range_bit(&BTRFS_I(inode)->io_tree, off,
2895 2896 2897
				    off + len - 1, EXTENT_DELALLOC, 0, NULL)) {
			if (ordered)
				btrfs_put_ordered_extent(ordered);
2898
			break;
2899
		}
2900 2901 2902
		unlock_extent(&BTRFS_I(inode)->io_tree, off, off + len - 1);
		if (ordered)
			btrfs_put_ordered_extent(ordered);
2903 2904
		if (!retry_range_locking)
			return -EAGAIN;
2905 2906
		btrfs_wait_ordered_range(inode, off, len);
	}
2907
	return 0;
2908 2909
}

2910
static void btrfs_double_inode_unlock(struct inode *inode1, struct inode *inode2)
M
Mark Fasheh 已提交
2911
{
A
Al Viro 已提交
2912 2913
	inode_unlock(inode1);
	inode_unlock(inode2);
M
Mark Fasheh 已提交
2914 2915
}

2916 2917 2918 2919 2920
static void btrfs_double_inode_lock(struct inode *inode1, struct inode *inode2)
{
	if (inode1 < inode2)
		swap(inode1, inode2);

A
Al Viro 已提交
2921 2922
	inode_lock_nested(inode1, I_MUTEX_PARENT);
	inode_lock_nested(inode2, I_MUTEX_CHILD);
2923 2924 2925 2926 2927 2928 2929 2930 2931
}

static void btrfs_double_extent_unlock(struct inode *inode1, u64 loff1,
				      struct inode *inode2, u64 loff2, u64 len)
{
	unlock_extent(&BTRFS_I(inode1)->io_tree, loff1, loff1 + len - 1);
	unlock_extent(&BTRFS_I(inode2)->io_tree, loff2, loff2 + len - 1);
}

2932 2933 2934
static int btrfs_double_extent_lock(struct inode *inode1, u64 loff1,
				    struct inode *inode2, u64 loff2, u64 len,
				    bool retry_range_locking)
M
Mark Fasheh 已提交
2935
{
2936 2937
	int ret;

M
Mark Fasheh 已提交
2938 2939 2940 2941
	if (inode1 < inode2) {
		swap(inode1, inode2);
		swap(loff1, loff2);
	}
2942 2943 2944 2945 2946 2947 2948 2949
	ret = lock_extent_range(inode1, loff1, len, retry_range_locking);
	if (ret)
		return ret;
	ret = lock_extent_range(inode2, loff2, len, retry_range_locking);
	if (ret)
		unlock_extent(&BTRFS_I(inode1)->io_tree, loff1,
			      loff1 + len - 1);
	return ret;
2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964
}

struct cmp_pages {
	int		num_pages;
	struct page	**src_pages;
	struct page	**dst_pages;
};

static void btrfs_cmp_data_free(struct cmp_pages *cmp)
{
	int i;
	struct page *pg;

	for (i = 0; i < cmp->num_pages; i++) {
		pg = cmp->src_pages[i];
2965 2966
		if (pg) {
			unlock_page(pg);
2967
			put_page(pg);
2968
		}
2969
		pg = cmp->dst_pages[i];
2970 2971
		if (pg) {
			unlock_page(pg);
2972
			put_page(pg);
2973
		}
2974 2975 2976 2977 2978 2979 2980 2981 2982 2983
	}
	kfree(cmp->src_pages);
	kfree(cmp->dst_pages);
}

static int btrfs_cmp_data_prepare(struct inode *src, u64 loff,
				  struct inode *dst, u64 dst_loff,
				  u64 len, struct cmp_pages *cmp)
{
	int ret;
2984
	int num_pages = PAGE_ALIGN(len) >> PAGE_SHIFT;
2985 2986 2987 2988 2989 2990 2991 2992
	struct page **src_pgarr, **dst_pgarr;

	/*
	 * We must gather up all the pages before we initiate our
	 * extent locking. We use an array for the page pointers. Size
	 * of the array is bounded by len, which is in turn bounded by
	 * BTRFS_MAX_DEDUPE_LEN.
	 */
2993 2994
	src_pgarr = kcalloc(num_pages, sizeof(struct page *), GFP_KERNEL);
	dst_pgarr = kcalloc(num_pages, sizeof(struct page *), GFP_KERNEL);
2995 2996 2997 2998
	if (!src_pgarr || !dst_pgarr) {
		kfree(src_pgarr);
		kfree(dst_pgarr);
		return -ENOMEM;
M
Mark Fasheh 已提交
2999
	}
3000 3001 3002 3003
	cmp->num_pages = num_pages;
	cmp->src_pages = src_pgarr;
	cmp->dst_pages = dst_pgarr;

3004 3005 3006 3007 3008 3009 3010 3011 3012 3013 3014
	/*
	 * If deduping ranges in the same inode, locking rules make it mandatory
	 * to always lock pages in ascending order to avoid deadlocks with
	 * concurrent tasks (such as starting writeback/delalloc).
	 */
	if (src == dst && dst_loff < loff) {
		swap(src_pgarr, dst_pgarr);
		swap(loff, dst_loff);
	}

	ret = gather_extent_pages(src, src_pgarr, cmp->num_pages, loff);
3015 3016 3017
	if (ret)
		goto out;

3018
	ret = gather_extent_pages(dst, dst_pgarr, cmp->num_pages, dst_loff);
3019 3020 3021 3022

out:
	if (ret)
		btrfs_cmp_data_free(cmp);
3023
	return ret;
M
Mark Fasheh 已提交
3024 3025
}

3026
static int btrfs_cmp_data(u64 len, struct cmp_pages *cmp)
M
Mark Fasheh 已提交
3027 3028
{
	int ret = 0;
3029
	int i;
M
Mark Fasheh 已提交
3030
	struct page *src_page, *dst_page;
3031
	unsigned int cmp_len = PAGE_SIZE;
M
Mark Fasheh 已提交
3032 3033
	void *addr, *dst_addr;

3034
	i = 0;
M
Mark Fasheh 已提交
3035
	while (len) {
3036
		if (len < PAGE_SIZE)
M
Mark Fasheh 已提交
3037 3038
			cmp_len = len;

3039 3040 3041 3042
		BUG_ON(i >= cmp->num_pages);

		src_page = cmp->src_pages[i];
		dst_page = cmp->dst_pages[i];
3043 3044
		ASSERT(PageLocked(src_page));
		ASSERT(PageLocked(dst_page));
3045

M
Mark Fasheh 已提交
3046 3047 3048 3049 3050 3051 3052
		addr = kmap_atomic(src_page);
		dst_addr = kmap_atomic(dst_page);

		flush_dcache_page(src_page);
		flush_dcache_page(dst_page);

		if (memcmp(addr, dst_addr, cmp_len))
3053
			ret = -EBADE;
M
Mark Fasheh 已提交
3054 3055 3056 3057 3058 3059 3060 3061

		kunmap_atomic(addr);
		kunmap_atomic(dst_addr);

		if (ret)
			break;

		len -= cmp_len;
3062
		i++;
M
Mark Fasheh 已提交
3063 3064 3065 3066 3067
	}

	return ret;
}

3068 3069
static int extent_same_check_offsets(struct inode *inode, u64 off, u64 *plen,
				     u64 olen)
M
Mark Fasheh 已提交
3070
{
3071
	u64 len = *plen;
M
Mark Fasheh 已提交
3072 3073
	u64 bs = BTRFS_I(inode)->root->fs_info->sb->s_blocksize;

3074
	if (off + olen > inode->i_size || off + olen < off)
M
Mark Fasheh 已提交
3075
		return -EINVAL;
3076 3077 3078 3079 3080

	/* if we extend to eof, continue to block boundary */
	if (off + len == inode->i_size)
		*plen = len = ALIGN(inode->i_size, bs) - off;

M
Mark Fasheh 已提交
3081 3082 3083 3084 3085 3086 3087
	/* Check that we are block aligned - btrfs_clone() requires this */
	if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs))
		return -EINVAL;

	return 0;
}

3088
static int btrfs_extent_same(struct inode *src, u64 loff, u64 olen,
M
Mark Fasheh 已提交
3089 3090 3091
			     struct inode *dst, u64 dst_loff)
{
	int ret;
3092
	u64 len = olen;
3093
	struct cmp_pages cmp;
3094
	bool same_inode = (src == dst);
M
Mark Fasheh 已提交
3095 3096
	u64 same_lock_start = 0;
	u64 same_lock_len = 0;
M
Mark Fasheh 已提交
3097

3098 3099 3100
	if (len == 0)
		return 0;

3101
	if (same_inode)
A
Al Viro 已提交
3102
		inode_lock(src);
3103 3104
	else
		btrfs_double_inode_lock(src, dst);
M
Mark Fasheh 已提交
3105

3106 3107 3108
	ret = extent_same_check_offsets(src, loff, &len, olen);
	if (ret)
		goto out_unlock;
M
Mark Fasheh 已提交
3109

3110 3111 3112 3113 3114
	ret = extent_same_check_offsets(dst, dst_loff, &len, olen);
	if (ret)
		goto out_unlock;

	if (same_inode) {
M
Mark Fasheh 已提交
3115 3116 3117 3118 3119 3120 3121 3122 3123 3124 3125 3126 3127 3128 3129 3130 3131 3132 3133 3134 3135 3136 3137 3138 3139 3140 3141 3142
		/*
		 * Single inode case wants the same checks, except we
		 * don't want our length pushed out past i_size as
		 * comparing that data range makes no sense.
		 *
		 * extent_same_check_offsets() will do this for an
		 * unaligned length at i_size, so catch it here and
		 * reject the request.
		 *
		 * This effectively means we require aligned extents
		 * for the single-inode case, whereas the other cases
		 * allow an unaligned length so long as it ends at
		 * i_size.
		 */
		if (len != olen) {
			ret = -EINVAL;
			goto out_unlock;
		}

		/* Check for overlapping ranges */
		if (dst_loff + len > loff && dst_loff < loff + len) {
			ret = -EINVAL;
			goto out_unlock;
		}

		same_lock_start = min_t(u64, loff, dst_loff);
		same_lock_len = max_t(u64, loff, dst_loff) + len - same_lock_start;
	}
M
Mark Fasheh 已提交
3143 3144 3145 3146 3147 3148 3149 3150

	/* don't make the dst file partly checksummed */
	if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
	    (BTRFS_I(dst)->flags & BTRFS_INODE_NODATASUM)) {
		ret = -EINVAL;
		goto out_unlock;
	}

3151
again:
3152 3153 3154 3155
	ret = btrfs_cmp_data_prepare(src, loff, dst, dst_loff, olen, &cmp);
	if (ret)
		goto out_unlock;

M
Mark Fasheh 已提交
3156
	if (same_inode)
3157 3158
		ret = lock_extent_range(src, same_lock_start, same_lock_len,
					false);
M
Mark Fasheh 已提交
3159
	else
3160 3161 3162 3163 3164 3165 3166 3167 3168 3169 3170 3171 3172 3173 3174 3175 3176 3177 3178 3179 3180 3181 3182 3183 3184 3185 3186 3187 3188 3189 3190
		ret = btrfs_double_extent_lock(src, loff, dst, dst_loff, len,
					       false);
	/*
	 * If one of the inodes has dirty pages in the respective range or
	 * ordered extents, we need to flush dellaloc and wait for all ordered
	 * extents in the range. We must unlock the pages and the ranges in the
	 * io trees to avoid deadlocks when flushing delalloc (requires locking
	 * pages) and when waiting for ordered extents to complete (they require
	 * range locking).
	 */
	if (ret == -EAGAIN) {
		/*
		 * Ranges in the io trees already unlocked. Now unlock all
		 * pages before waiting for all IO to complete.
		 */
		btrfs_cmp_data_free(&cmp);
		if (same_inode) {
			btrfs_wait_ordered_range(src, same_lock_start,
						 same_lock_len);
		} else {
			btrfs_wait_ordered_range(src, loff, len);
			btrfs_wait_ordered_range(dst, dst_loff, len);
		}
		goto again;
	}
	ASSERT(ret == 0);
	if (WARN_ON(ret)) {
		/* ranges in the io trees already unlocked */
		btrfs_cmp_data_free(&cmp);
		return ret;
	}
3191

3192
	/* pass original length for comparison so we stay within i_size */
3193
	ret = btrfs_cmp_data(olen, &cmp);
M
Mark Fasheh 已提交
3194
	if (ret == 0)
3195
		ret = btrfs_clone(src, dst, loff, olen, len, dst_loff, 1);
M
Mark Fasheh 已提交
3196

M
Mark Fasheh 已提交
3197 3198 3199 3200 3201
	if (same_inode)
		unlock_extent(&BTRFS_I(src)->io_tree, same_lock_start,
			      same_lock_start + same_lock_len - 1);
	else
		btrfs_double_extent_unlock(src, loff, dst, dst_loff, len);
3202 3203

	btrfs_cmp_data_free(&cmp);
M
Mark Fasheh 已提交
3204
out_unlock:
M
Mark Fasheh 已提交
3205
	if (same_inode)
A
Al Viro 已提交
3206
		inode_unlock(src);
M
Mark Fasheh 已提交
3207 3208
	else
		btrfs_double_inode_unlock(src, dst);
M
Mark Fasheh 已提交
3209 3210 3211 3212

	return ret;
}

3213
#define BTRFS_MAX_DEDUPE_LEN	SZ_16M
M
Mark Fasheh 已提交
3214

3215 3216
ssize_t btrfs_dedupe_file_range(struct file *src_file, u64 loff, u64 olen,
				struct file *dst_file, u64 dst_loff)
M
Mark Fasheh 已提交
3217
{
3218 3219
	struct inode *src = file_inode(src_file);
	struct inode *dst = file_inode(dst_file);
M
Mark Fasheh 已提交
3220
	u64 bs = BTRFS_I(src)->root->fs_info->sb->s_blocksize;
3221
	ssize_t res;
M
Mark Fasheh 已提交
3222

3223 3224
	if (olen > BTRFS_MAX_DEDUPE_LEN)
		olen = BTRFS_MAX_DEDUPE_LEN;
M
Mark Fasheh 已提交
3225

3226
	if (WARN_ON_ONCE(bs < PAGE_SIZE)) {
M
Mark Fasheh 已提交
3227 3228 3229 3230 3231
		/*
		 * Btrfs does not support blocksize < page_size. As a
		 * result, btrfs_cmp_data() won't correctly handle
		 * this situation without an update.
		 */
3232
		return -EINVAL;
M
Mark Fasheh 已提交
3233 3234
	}

3235 3236 3237 3238
	res = btrfs_extent_same(src, loff, olen, dst, dst_loff);
	if (res)
		return res;
	return olen;
M
Mark Fasheh 已提交
3239 3240
}

3241 3242 3243 3244
static int clone_finish_inode_update(struct btrfs_trans_handle *trans,
				     struct inode *inode,
				     u64 endoff,
				     const u64 destoff,
3245 3246
				     const u64 olen,
				     int no_time_update)
3247 3248 3249 3250 3251
{
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret;

	inode_inc_iversion(inode);
3252
	if (!no_time_update)
3253
		inode->i_mtime = inode->i_ctime = current_time(inode);
3254 3255 3256 3257 3258 3259 3260
	/*
	 * We round up to the block size at eof when determining which
	 * extents to clone above, but shouldn't round up the file size.
	 */
	if (endoff > destoff + olen)
		endoff = destoff + olen;
	if (endoff > inode->i_size)
3261
		btrfs_i_size_write(BTRFS_I(inode), endoff);
3262 3263 3264

	ret = btrfs_update_inode(trans, root, inode);
	if (ret) {
3265
		btrfs_abort_transaction(trans, ret);
3266
		btrfs_end_transaction(trans);
3267 3268
		goto out;
	}
3269
	ret = btrfs_end_transaction(trans);
3270 3271 3272 3273
out:
	return ret;
}

3274
static void clone_update_extent_map(struct btrfs_inode *inode,
3275 3276 3277 3278 3279
				    const struct btrfs_trans_handle *trans,
				    const struct btrfs_path *path,
				    const u64 hole_offset,
				    const u64 hole_len)
{
3280
	struct extent_map_tree *em_tree = &inode->extent_tree;
3281 3282 3283 3284 3285
	struct extent_map *em;
	int ret;

	em = alloc_extent_map();
	if (!em) {
3286
		set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags);
3287 3288 3289
		return;
	}

3290 3291 3292 3293 3294
	if (path) {
		struct btrfs_file_extent_item *fi;

		fi = btrfs_item_ptr(path->nodes[0], path->slots[0],
				    struct btrfs_file_extent_item);
3295
		btrfs_extent_item_to_extent_map(inode, path, fi, false, em);
3296 3297 3298 3299
		em->generation = -1;
		if (btrfs_file_extent_type(path->nodes[0], fi) ==
		    BTRFS_FILE_EXTENT_INLINE)
			set_bit(BTRFS_INODE_NEEDS_FULL_SYNC,
3300
					&inode->runtime_flags);
3301 3302 3303 3304 3305 3306 3307 3308 3309 3310 3311 3312 3313 3314 3315 3316 3317 3318 3319 3320
	} else {
		em->start = hole_offset;
		em->len = hole_len;
		em->ram_bytes = em->len;
		em->orig_start = hole_offset;
		em->block_start = EXTENT_MAP_HOLE;
		em->block_len = 0;
		em->orig_block_len = 0;
		em->compress_type = BTRFS_COMPRESS_NONE;
		em->generation = trans->transid;
	}

	while (1) {
		write_lock(&em_tree->lock);
		ret = add_extent_mapping(em_tree, em, 1);
		write_unlock(&em_tree->lock);
		if (ret != -EEXIST) {
			free_extent_map(em);
			break;
		}
3321
		btrfs_drop_extent_cache(inode, em->start,
3322 3323 3324
					em->start + em->len - 1, 0);
	}

3325
	if (ret)
3326
		set_bit(BTRFS_INODE_NEEDS_FULL_SYNC, &inode->runtime_flags);
3327 3328
}

3329 3330 3331 3332 3333 3334 3335 3336 3337 3338 3339 3340 3341 3342 3343 3344 3345 3346 3347 3348 3349 3350 3351 3352 3353
/*
 * Make sure we do not end up inserting an inline extent into a file that has
 * already other (non-inline) extents. If a file has an inline extent it can
 * not have any other extents and the (single) inline extent must start at the
 * file offset 0. Failing to respect these rules will lead to file corruption,
 * resulting in EIO errors on read/write operations, hitting BUG_ON's in mm, etc
 *
 * We can have extents that have been already written to disk or we can have
 * dirty ranges still in delalloc, in which case the extent maps and items are
 * created only when we run delalloc, and the delalloc ranges might fall outside
 * the range we are currently locking in the inode's io tree. So we check the
 * inode's i_size because of that (i_size updates are done while holding the
 * i_mutex, which we are holding here).
 * We also check to see if the inode has a size not greater than "datal" but has
 * extents beyond it, due to an fallocate with FALLOC_FL_KEEP_SIZE (and we are
 * protected against such concurrent fallocate calls by the i_mutex).
 *
 * If the file has no extents but a size greater than datal, do not allow the
 * copy because we would need turn the inline extent into a non-inline one (even
 * with NO_HOLES enabled). If we find our destination inode only has one inline
 * extent, just overwrite it with the source inline extent if its size is less
 * than the source extent's size, or we could copy the source inline extent's
 * data into the destination inode's inline extent if the later is greater then
 * the former.
 */
3354
static int clone_copy_inline_extent(struct inode *dst,
3355 3356 3357 3358 3359 3360 3361 3362 3363
				    struct btrfs_trans_handle *trans,
				    struct btrfs_path *path,
				    struct btrfs_key *new_key,
				    const u64 drop_start,
				    const u64 datal,
				    const u64 skip,
				    const u64 size,
				    char *inline_data)
{
3364
	struct btrfs_fs_info *fs_info = btrfs_sb(dst->i_sb);
3365 3366
	struct btrfs_root *root = BTRFS_I(dst)->root;
	const u64 aligned_end = ALIGN(new_key->offset + datal,
3367
				      fs_info->sectorsize);
3368 3369 3370 3371 3372 3373
	int ret;
	struct btrfs_key key;

	if (new_key->offset > 0)
		return -EOPNOTSUPP;

3374
	key.objectid = btrfs_ino(BTRFS_I(dst));
3375 3376 3377 3378 3379 3380 3381 3382 3383 3384 3385 3386 3387 3388
	key.type = BTRFS_EXTENT_DATA_KEY;
	key.offset = 0;
	ret = btrfs_search_slot(NULL, root, &key, path, 0, 0);
	if (ret < 0) {
		return ret;
	} else if (ret > 0) {
		if (path->slots[0] >= btrfs_header_nritems(path->nodes[0])) {
			ret = btrfs_next_leaf(root, path);
			if (ret < 0)
				return ret;
			else if (ret > 0)
				goto copy_inline_extent;
		}
		btrfs_item_key_to_cpu(path->nodes[0], &key, path->slots[0]);
3389
		if (key.objectid == btrfs_ino(BTRFS_I(dst)) &&
3390 3391 3392 3393 3394 3395 3396 3397 3398 3399 3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413 3414 3415 3416 3417 3418 3419 3420 3421 3422
		    key.type == BTRFS_EXTENT_DATA_KEY) {
			ASSERT(key.offset > 0);
			return -EOPNOTSUPP;
		}
	} else if (i_size_read(dst) <= datal) {
		struct btrfs_file_extent_item *ei;
		u64 ext_len;

		/*
		 * If the file size is <= datal, make sure there are no other
		 * extents following (can happen do to an fallocate call with
		 * the flag FALLOC_FL_KEEP_SIZE).
		 */
		ei = btrfs_item_ptr(path->nodes[0], path->slots[0],
				    struct btrfs_file_extent_item);
		/*
		 * If it's an inline extent, it can not have other extents
		 * following it.
		 */
		if (btrfs_file_extent_type(path->nodes[0], ei) ==
		    BTRFS_FILE_EXTENT_INLINE)
			goto copy_inline_extent;

		ext_len = btrfs_file_extent_num_bytes(path->nodes[0], ei);
		if (ext_len > aligned_end)
			return -EOPNOTSUPP;

		ret = btrfs_next_item(root, path);
		if (ret < 0) {
			return ret;
		} else if (ret == 0) {
			btrfs_item_key_to_cpu(path->nodes[0], &key,
					      path->slots[0]);
3423
			if (key.objectid == btrfs_ino(BTRFS_I(dst)) &&
3424 3425 3426 3427 3428 3429 3430 3431 3432 3433 3434 3435 3436 3437 3438 3439 3440 3441 3442 3443 3444 3445 3446 3447 3448 3449 3450 3451 3452 3453 3454 3455 3456 3457 3458 3459 3460 3461 3462 3463 3464 3465 3466 3467 3468 3469 3470 3471 3472
			    key.type == BTRFS_EXTENT_DATA_KEY)
				return -EOPNOTSUPP;
		}
	}

copy_inline_extent:
	/*
	 * We have no extent items, or we have an extent at offset 0 which may
	 * or may not be inlined. All these cases are dealt the same way.
	 */
	if (i_size_read(dst) > datal) {
		/*
		 * If the destination inode has an inline extent...
		 * This would require copying the data from the source inline
		 * extent into the beginning of the destination's inline extent.
		 * But this is really complex, both extents can be compressed
		 * or just one of them, which would require decompressing and
		 * re-compressing data (which could increase the new compressed
		 * size, not allowing the compressed data to fit anymore in an
		 * inline extent).
		 * So just don't support this case for now (it should be rare,
		 * we are not really saving space when cloning inline extents).
		 */
		return -EOPNOTSUPP;
	}

	btrfs_release_path(path);
	ret = btrfs_drop_extents(trans, root, dst, drop_start, aligned_end, 1);
	if (ret)
		return ret;
	ret = btrfs_insert_empty_item(trans, root, path, new_key, size);
	if (ret)
		return ret;

	if (skip) {
		const u32 start = btrfs_file_extent_calc_inline_size(0);

		memmove(inline_data + start, inline_data + start + skip, datal);
	}

	write_extent_buffer(path->nodes[0], inline_data,
			    btrfs_item_ptr_offset(path->nodes[0],
						  path->slots[0]),
			    size);
	inode_add_bytes(dst, datal);

	return 0;
}

3473 3474 3475 3476 3477 3478 3479
/**
 * btrfs_clone() - clone a range from inode file to another
 *
 * @src: Inode to clone from
 * @inode: Inode to clone to
 * @off: Offset within source to start clone from
 * @olen: Original length, passed by user, of range to clone
3480
 * @olen_aligned: Block-aligned value of olen
3481
 * @destoff: Offset within @inode to start clone
3482
 * @no_time_update: Whether to update mtime/ctime on the target inode
3483 3484
 */
static int btrfs_clone(struct inode *src, struct inode *inode,
3485
		       const u64 off, const u64 olen, const u64 olen_aligned,
3486
		       const u64 destoff, int no_time_update)
C
Christoph Hellwig 已提交
3487
{
3488
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
C
Christoph Hellwig 已提交
3489
	struct btrfs_root *root = BTRFS_I(inode)->root;
3490
	struct btrfs_path *path = NULL;
C
Christoph Hellwig 已提交
3491
	struct extent_buffer *leaf;
3492 3493
	struct btrfs_trans_handle *trans;
	char *buf = NULL;
Y
Yan Zheng 已提交
3494
	struct btrfs_key key;
C
Christoph Hellwig 已提交
3495 3496
	u32 nritems;
	int slot;
Y
Yan Zheng 已提交
3497
	int ret;
3498 3499
	const u64 len = olen_aligned;
	u64 last_dest_end = destoff;
Y
Yan Zheng 已提交
3500 3501

	ret = -ENOMEM;
3502 3503 3504
	buf = kvmalloc(fs_info->nodesize, GFP_KERNEL);
	if (!buf)
		return ret;
Y
Yan Zheng 已提交
3505 3506 3507

	path = btrfs_alloc_path();
	if (!path) {
3508
		kvfree(buf);
3509
		return ret;
3510 3511
	}

3512
	path->reada = READA_FORWARD;
3513
	/* clone data */
3514
	key.objectid = btrfs_ino(BTRFS_I(src));
Y
Yan Zheng 已提交
3515
	key.type = BTRFS_EXTENT_DATA_KEY;
3516
	key.offset = off;
C
Christoph Hellwig 已提交
3517 3518

	while (1) {
3519
		u64 next_key_min_offset = key.offset + 1;
3520

C
Christoph Hellwig 已提交
3521 3522 3523 3524
		/*
		 * note the key will change type as we walk through the
		 * tree.
		 */
3525
		path->leave_spinning = 1;
3526 3527
		ret = btrfs_search_slot(NULL, BTRFS_I(src)->root, &key, path,
				0, 0);
C
Christoph Hellwig 已提交
3528 3529
		if (ret < 0)
			goto out;
3530 3531 3532 3533 3534 3535 3536 3537 3538 3539 3540
		/*
		 * First search, if no extent item that starts at offset off was
		 * found but the previous item is an extent item, it's possible
		 * it might overlap our target range, therefore process it.
		 */
		if (key.offset == off && ret > 0 && path->slots[0] > 0) {
			btrfs_item_key_to_cpu(path->nodes[0], &key,
					      path->slots[0] - 1);
			if (key.type == BTRFS_EXTENT_DATA_KEY)
				path->slots[0]--;
		}
C
Christoph Hellwig 已提交
3541

Y
Yan Zheng 已提交
3542
		nritems = btrfs_header_nritems(path->nodes[0]);
3543
process_slot:
Y
Yan Zheng 已提交
3544
		if (path->slots[0] >= nritems) {
3545
			ret = btrfs_next_leaf(BTRFS_I(src)->root, path);
C
Christoph Hellwig 已提交
3546 3547 3548 3549
			if (ret < 0)
				goto out;
			if (ret > 0)
				break;
Y
Yan Zheng 已提交
3550
			nritems = btrfs_header_nritems(path->nodes[0]);
C
Christoph Hellwig 已提交
3551 3552 3553 3554
		}
		leaf = path->nodes[0];
		slot = path->slots[0];

Y
Yan Zheng 已提交
3555
		btrfs_item_key_to_cpu(leaf, &key, slot);
3556
		if (key.type > BTRFS_EXTENT_DATA_KEY ||
3557
		    key.objectid != btrfs_ino(BTRFS_I(src)))
C
Christoph Hellwig 已提交
3558 3559
			break;

3560
		if (key.type == BTRFS_EXTENT_DATA_KEY) {
3561 3562
			struct btrfs_file_extent_item *extent;
			int type;
Z
Zheng Yan 已提交
3563 3564
			u32 size;
			struct btrfs_key new_key;
3565 3566 3567
			u64 disko = 0, diskl = 0;
			u64 datao = 0, datal = 0;
			u8 comp;
3568
			u64 drop_start;
Z
Zheng Yan 已提交
3569

3570 3571 3572 3573
			extent = btrfs_item_ptr(leaf, slot,
						struct btrfs_file_extent_item);
			comp = btrfs_file_extent_compression(leaf, extent);
			type = btrfs_file_extent_type(leaf, extent);
3574 3575
			if (type == BTRFS_FILE_EXTENT_REG ||
			    type == BTRFS_FILE_EXTENT_PREALLOC) {
C
Chris Mason 已提交
3576 3577 3578 3579
				disko = btrfs_file_extent_disk_bytenr(leaf,
								      extent);
				diskl = btrfs_file_extent_disk_num_bytes(leaf,
								 extent);
3580
				datao = btrfs_file_extent_offset(leaf, extent);
C
Chris Mason 已提交
3581 3582
				datal = btrfs_file_extent_num_bytes(leaf,
								    extent);
3583 3584 3585 3586 3587
			} else if (type == BTRFS_FILE_EXTENT_INLINE) {
				/* take upper bound, may be compressed */
				datal = btrfs_file_extent_ram_bytes(leaf,
								    extent);
			}
Z
Zheng Yan 已提交
3588

3589 3590 3591 3592 3593 3594
			/*
			 * The first search might have left us at an extent
			 * item that ends before our target range's start, can
			 * happen if we have holes and NO_HOLES feature enabled.
			 */
			if (key.offset + datal <= off) {
3595 3596
				path->slots[0]++;
				goto process_slot;
3597 3598
			} else if (key.offset >= off + len) {
				break;
3599
			}
3600
			next_key_min_offset = key.offset + datal;
3601 3602 3603 3604 3605 3606 3607
			size = btrfs_item_size_nr(leaf, slot);
			read_extent_buffer(leaf, buf,
					   btrfs_item_ptr_offset(leaf, slot),
					   size);

			btrfs_release_path(path);
			path->leave_spinning = 0;
3608

Z
Zheng Yan 已提交
3609
			memcpy(&new_key, &key, sizeof(new_key));
3610
			new_key.objectid = btrfs_ino(BTRFS_I(inode));
3611 3612 3613 3614
			if (off <= key.offset)
				new_key.offset = key.offset + destoff - off;
			else
				new_key.offset = destoff;
Z
Zheng Yan 已提交
3615

3616 3617 3618 3619 3620 3621 3622 3623 3624 3625 3626 3627
			/*
			 * Deal with a hole that doesn't have an extent item
			 * that represents it (NO_HOLES feature enabled).
			 * This hole is either in the middle of the cloning
			 * range or at the beginning (fully overlaps it or
			 * partially overlaps it).
			 */
			if (new_key.offset != last_dest_end)
				drop_start = last_dest_end;
			else
				drop_start = new_key.offset;

3628 3629 3630 3631 3632 3633
			/*
			 * 1 - adjusting old extent (we may have to split it)
			 * 1 - add new extent
			 * 1 - inode update
			 */
			trans = btrfs_start_transaction(root, 3);
3634 3635 3636 3637 3638
			if (IS_ERR(trans)) {
				ret = PTR_ERR(trans);
				goto out;
			}

3639 3640
			if (type == BTRFS_FILE_EXTENT_REG ||
			    type == BTRFS_FILE_EXTENT_PREALLOC) {
3641 3642 3643 3644 3645
				/*
				 *    a  | --- range to clone ---|  b
				 * | ------------- extent ------------- |
				 */

3646
				/* subtract range b */
3647 3648 3649
				if (key.offset + datal > off + len)
					datal = off + len - key.offset;

3650
				/* subtract range a */
3651 3652 3653 3654 3655
				if (off > key.offset) {
					datao += off - key.offset;
					datal -= off - key.offset;
				}

J
Josef Bacik 已提交
3656
				ret = btrfs_drop_extents(trans, root, inode,
3657
							 drop_start,
3658
							 new_key.offset + datal,
3659
							 1);
3660
				if (ret) {
3661
					if (ret != -EOPNOTSUPP)
3662
						btrfs_abort_transaction(trans,
3663
									ret);
3664
					btrfs_end_transaction(trans);
3665 3666
					goto out;
				}
3667

3668 3669
				ret = btrfs_insert_empty_item(trans, root, path,
							      &new_key, size);
3670
				if (ret) {
3671
					btrfs_abort_transaction(trans, ret);
3672
					btrfs_end_transaction(trans);
3673 3674
					goto out;
				}
3675 3676 3677 3678

				leaf = path->nodes[0];
				slot = path->slots[0];
				write_extent_buffer(leaf, buf,
Z
Zheng Yan 已提交
3679 3680
					    btrfs_item_ptr_offset(leaf, slot),
					    size);
Y
Yan Zheng 已提交
3681

3682
				extent = btrfs_item_ptr(leaf, slot,
C
Christoph Hellwig 已提交
3683
						struct btrfs_file_extent_item);
3684 3685 3686 3687 3688 3689 3690 3691 3692

				/* disko == 0 means it's a hole */
				if (!disko)
					datao = 0;

				btrfs_set_file_extent_offset(leaf, extent,
							     datao);
				btrfs_set_file_extent_num_bytes(leaf, extent,
								datal);
J
Josef Bacik 已提交
3693

3694 3695
				if (disko) {
					inode_add_bytes(inode, datal);
3696
					ret = btrfs_inc_extent_ref(trans,
3697
							root,
3698 3699
							disko, diskl, 0,
							root->root_key.objectid,
3700
							btrfs_ino(BTRFS_I(inode)),
3701
							new_key.offset - datao);
3702 3703 3704
					if (ret) {
						btrfs_abort_transaction(trans,
									ret);
3705
						btrfs_end_transaction(trans);
3706 3707 3708
						goto out;

					}
C
Christoph Hellwig 已提交
3709
				}
3710 3711 3712
			} else if (type == BTRFS_FILE_EXTENT_INLINE) {
				u64 skip = 0;
				u64 trim = 0;
3713

3714 3715 3716 3717
				if (off > key.offset) {
					skip = off - key.offset;
					new_key.offset += skip;
				}
C
Chris Mason 已提交
3718

L
Liu Bo 已提交
3719 3720
				if (key.offset + datal > off + len)
					trim = key.offset + datal - (off + len);
C
Chris Mason 已提交
3721

3722 3723
				if (comp && (skip || trim)) {
					ret = -EINVAL;
3724
					btrfs_end_transaction(trans);
3725 3726 3727 3728
					goto out;
				}
				size -= skip + trim;
				datal -= skip + trim;
3729

3730
				ret = clone_copy_inline_extent(inode,
3731 3732 3733 3734 3735
							       trans, path,
							       &new_key,
							       drop_start,
							       datal,
							       skip, size, buf);
3736
				if (ret) {
3737
					if (ret != -EOPNOTSUPP)
3738
						btrfs_abort_transaction(trans,
3739
									ret);
3740
					btrfs_end_transaction(trans);
3741 3742
					goto out;
				}
3743 3744
				leaf = path->nodes[0];
				slot = path->slots[0];
C
Christoph Hellwig 已提交
3745
			}
3746

3747 3748
			/* If we have an implicit hole (NO_HOLES feature). */
			if (drop_start < new_key.offset)
3749
				clone_update_extent_map(BTRFS_I(inode), trans,
3750
						NULL, drop_start,
3751 3752
						new_key.offset - drop_start);

3753 3754
			clone_update_extent_map(BTRFS_I(inode), trans,
					path, 0, 0);
3755

3756
			btrfs_mark_buffer_dirty(leaf);
3757
			btrfs_release_path(path);
3758

3759
			last_dest_end = ALIGN(new_key.offset + datal,
3760
					      fs_info->sectorsize);
3761 3762
			ret = clone_finish_inode_update(trans, inode,
							last_dest_end,
3763 3764
							destoff, olen,
							no_time_update);
3765
			if (ret)
3766
				goto out;
3767 3768
			if (new_key.offset + datal >= destoff + len)
				break;
3769
		}
3770
		btrfs_release_path(path);
3771
		key.offset = next_key_min_offset;
3772 3773 3774 3775 3776

		if (fatal_signal_pending(current)) {
			ret = -EINTR;
			goto out;
		}
C
Christoph Hellwig 已提交
3777 3778
	}
	ret = 0;
3779

3780 3781 3782 3783 3784 3785 3786 3787 3788 3789 3790 3791 3792 3793 3794 3795 3796 3797 3798 3799
	if (last_dest_end < destoff + len) {
		/*
		 * We have an implicit hole (NO_HOLES feature is enabled) that
		 * fully or partially overlaps our cloning range at its end.
		 */
		btrfs_release_path(path);

		/*
		 * 1 - remove extent(s)
		 * 1 - inode update
		 */
		trans = btrfs_start_transaction(root, 2);
		if (IS_ERR(trans)) {
			ret = PTR_ERR(trans);
			goto out;
		}
		ret = btrfs_drop_extents(trans, root, inode,
					 last_dest_end, destoff + len, 1);
		if (ret) {
			if (ret != -EOPNOTSUPP)
3800
				btrfs_abort_transaction(trans, ret);
3801
			btrfs_end_transaction(trans);
3802 3803
			goto out;
		}
3804 3805 3806
		clone_update_extent_map(BTRFS_I(inode), trans, NULL,
				last_dest_end,
				destoff + len - last_dest_end);
3807
		ret = clone_finish_inode_update(trans, inode, destoff + len,
3808
						destoff, olen, no_time_update);
3809 3810
	}

C
Christoph Hellwig 已提交
3811
out:
3812
	btrfs_free_path(path);
3813
	kvfree(buf);
3814 3815 3816
	return ret;
}

3817 3818
static noinline int btrfs_clone_files(struct file *file, struct file *file_src,
					u64 off, u64 olen, u64 destoff)
3819
{
A
Al Viro 已提交
3820
	struct inode *inode = file_inode(file);
3821
	struct inode *src = file_inode(file_src);
3822
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
3823 3824 3825
	struct btrfs_root *root = BTRFS_I(inode)->root;
	int ret;
	u64 len = olen;
3826
	u64 bs = fs_info->sb->s_blocksize;
3827
	int same_inode = src == inode;
3828 3829 3830 3831 3832 3833 3834

	/*
	 * TODO:
	 * - split compressed inline extents.  annoying: we need to
	 *   decompress into destination's address_space (the file offset
	 *   may change, so source mapping won't do), then recompress (or
	 *   otherwise reinsert) a subrange.
3835 3836 3837
	 *
	 * - split destination inode's inline extents.  The inline extents can
	 *   be either compressed or non-compressed.
3838 3839 3840 3841 3842
	 */

	if (btrfs_root_readonly(root))
		return -EROFS;

3843 3844 3845
	if (file_src->f_path.mnt != file->f_path.mnt ||
	    src->i_sb != inode->i_sb)
		return -EXDEV;
3846 3847 3848 3849

	/* don't make the dst file partly checksummed */
	if ((BTRFS_I(src)->flags & BTRFS_INODE_NODATASUM) !=
	    (BTRFS_I(inode)->flags & BTRFS_INODE_NODATASUM))
3850
		return -EINVAL;
3851 3852

	if (S_ISDIR(src->i_mode) || S_ISDIR(inode->i_mode))
3853
		return -EISDIR;
3854 3855

	if (!same_inode) {
3856
		btrfs_double_inode_lock(src, inode);
3857
	} else {
A
Al Viro 已提交
3858
		inode_lock(src);
3859 3860 3861 3862 3863 3864 3865 3866 3867 3868 3869 3870
	}

	/* determine range to clone */
	ret = -EINVAL;
	if (off + len > src->i_size || off + len < off)
		goto out_unlock;
	if (len == 0)
		olen = len = src->i_size - off;
	/* if we extend to eof, continue to block boundary */
	if (off + len == src->i_size)
		len = ALIGN(src->i_size, bs) - off;

3871 3872 3873 3874 3875
	if (len == 0) {
		ret = 0;
		goto out_unlock;
	}

3876 3877 3878 3879 3880 3881 3882 3883 3884 3885 3886 3887 3888 3889 3890 3891 3892
	/* verify the end result is block aligned */
	if (!IS_ALIGNED(off, bs) || !IS_ALIGNED(off + len, bs) ||
	    !IS_ALIGNED(destoff, bs))
		goto out_unlock;

	/* verify if ranges are overlapped within the same file */
	if (same_inode) {
		if (destoff + len > off && destoff < off + len)
			goto out_unlock;
	}

	if (destoff > inode->i_size) {
		ret = btrfs_cont_expand(inode, inode->i_size, destoff);
		if (ret)
			goto out_unlock;
	}

3893 3894 3895 3896 3897 3898 3899 3900 3901 3902
	/*
	 * Lock the target range too. Right after we replace the file extent
	 * items in the fs tree (which now point to the cloned data), we might
	 * have a worker replace them with extent items relative to a write
	 * operation that was issued before this clone operation (i.e. confront
	 * with inode.c:btrfs_finish_ordered_io).
	 */
	if (same_inode) {
		u64 lock_start = min_t(u64, off, destoff);
		u64 lock_len = max_t(u64, off, destoff) + len - lock_start;
3903

3904
		ret = lock_extent_range(src, lock_start, lock_len, true);
3905
	} else {
3906 3907 3908 3909 3910 3911 3912
		ret = btrfs_double_extent_lock(src, off, inode, destoff, len,
					       true);
	}
	ASSERT(ret == 0);
	if (WARN_ON(ret)) {
		/* ranges in the io trees already unlocked */
		goto out_unlock;
3913
	}
3914

3915
	ret = btrfs_clone(src, inode, off, olen, len, destoff, 0);
3916

3917 3918 3919 3920 3921 3922
	if (same_inode) {
		u64 lock_start = min_t(u64, off, destoff);
		u64 lock_end = max_t(u64, off, destoff) + len - 1;

		unlock_extent(&BTRFS_I(src)->io_tree, lock_start, lock_end);
	} else {
3923
		btrfs_double_extent_unlock(src, off, inode, destoff, len);
3924 3925 3926 3927 3928
	}
	/*
	 * Truncate page cache pages so that future reads will see the cloned
	 * data immediately and not the previous data.
	 */
3929
	truncate_inode_pages_range(&inode->i_data,
3930 3931
				round_down(destoff, PAGE_SIZE),
				round_up(destoff + len, PAGE_SIZE) - 1);
C
Christoph Hellwig 已提交
3932
out_unlock:
3933 3934 3935
	if (!same_inode)
		btrfs_double_inode_unlock(src, inode);
	else
A
Al Viro 已提交
3936
		inode_unlock(src);
3937 3938 3939
	return ret;
}

3940 3941
int btrfs_clone_file_range(struct file *src_file, loff_t off,
		struct file *dst_file, loff_t destoff, u64 len)
3942
{
3943
	return btrfs_clone_files(dst_file, src_file, off, len, destoff);
3944 3945
}

C
Christoph Hellwig 已提交
3946 3947 3948 3949 3950 3951
/*
 * there are many ways the trans_start and trans_end ioctls can lead
 * to deadlocks.  They should only be used by applications that
 * basically own the machine, and have a very in depth understanding
 * of all the possible deadlocks and enospc problems.
 */
3952
static long btrfs_ioctl_trans_start(struct file *file)
C
Christoph Hellwig 已提交
3953
{
A
Al Viro 已提交
3954
	struct inode *inode = file_inode(file);
3955
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
C
Christoph Hellwig 已提交
3956 3957
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_trans_handle *trans;
3958
	struct btrfs_file_private *private;
3959
	int ret;
3960
	static bool warned = false;
C
Christoph Hellwig 已提交
3961

3962
	ret = -EPERM;
3963
	if (!capable(CAP_SYS_ADMIN))
3964
		goto out;
3965

3966 3967 3968 3969 3970 3971 3972 3973 3974 3975
	if (!warned) {
		btrfs_warn(fs_info,
			"Userspace transaction mechanism is considered "
			"deprecated and slated to be removed in 4.17. "
			"If you have a valid use case please "
			"speak up on the mailing list");
		WARN_ON(1);
		warned = true;
	}

3976
	ret = -EINPROGRESS;
3977 3978
	private = file->private_data;
	if (private && private->trans)
C
Christoph Hellwig 已提交
3979
		goto out;
3980 3981 3982 3983 3984 3985 3986
	if (!private) {
		private = kzalloc(sizeof(struct btrfs_file_private),
				  GFP_KERNEL);
		if (!private)
			return -ENOMEM;
		file->private_data = private;
	}
3987

L
Li Zefan 已提交
3988 3989 3990 3991
	ret = -EROFS;
	if (btrfs_root_readonly(root))
		goto out;

3992
	ret = mnt_want_write_file(file);
Y
Yan Zheng 已提交
3993 3994 3995
	if (ret)
		goto out;

3996
	atomic_inc(&fs_info->open_ioctl_trans);
3997

3998
	ret = -ENOMEM;
3999
	trans = btrfs_start_ioctl_transaction(root);
4000
	if (IS_ERR(trans))
4001 4002
		goto out_drop;

4003
	private->trans = trans;
4004 4005 4006
	return 0;

out_drop:
4007
	atomic_dec(&fs_info->open_ioctl_trans);
A
Al Viro 已提交
4008
	mnt_drop_write_file(file);
C
Christoph Hellwig 已提交
4009 4010 4011 4012
out:
	return ret;
}

4013 4014
static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp)
{
A
Al Viro 已提交
4015
	struct inode *inode = file_inode(file);
4016
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
4017 4018 4019 4020 4021 4022 4023 4024 4025
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_root *new_root;
	struct btrfs_dir_item *di;
	struct btrfs_trans_handle *trans;
	struct btrfs_path *path;
	struct btrfs_key location;
	struct btrfs_disk_key disk_key;
	u64 objectid = 0;
	u64 dir_id;
4026
	int ret;
4027 4028 4029 4030

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

4031 4032 4033 4034 4035 4036 4037 4038
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;

	if (copy_from_user(&objectid, argp, sizeof(objectid))) {
		ret = -EFAULT;
		goto out;
	}
4039 4040

	if (!objectid)
4041
		objectid = BTRFS_FS_TREE_OBJECTID;
4042 4043 4044 4045 4046

	location.objectid = objectid;
	location.type = BTRFS_ROOT_ITEM_KEY;
	location.offset = (u64)-1;

4047
	new_root = btrfs_read_fs_root_no_name(fs_info, &location);
4048 4049 4050 4051
	if (IS_ERR(new_root)) {
		ret = PTR_ERR(new_root);
		goto out;
	}
4052 4053 4054 4055
	if (!is_fstree(new_root->objectid)) {
		ret = -ENOENT;
		goto out;
	}
4056 4057

	path = btrfs_alloc_path();
4058 4059 4060 4061
	if (!path) {
		ret = -ENOMEM;
		goto out;
	}
4062 4063 4064
	path->leave_spinning = 1;

	trans = btrfs_start_transaction(root, 1);
4065
	if (IS_ERR(trans)) {
4066
		btrfs_free_path(path);
4067 4068
		ret = PTR_ERR(trans);
		goto out;
4069 4070
	}

4071 4072
	dir_id = btrfs_super_root_dir(fs_info->super_copy);
	di = btrfs_lookup_dir_item(trans, fs_info->tree_root, path,
4073
				   dir_id, "default", 7, 1);
4074
	if (IS_ERR_OR_NULL(di)) {
4075
		btrfs_free_path(path);
4076
		btrfs_end_transaction(trans);
4077
		btrfs_err(fs_info,
J
Jeff Mahoney 已提交
4078
			  "Umm, you don't have the default diritem, this isn't going to work");
4079 4080
		ret = -ENOENT;
		goto out;
4081 4082 4083 4084 4085 4086 4087
	}

	btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key);
	btrfs_set_dir_item_key(path->nodes[0], di, &disk_key);
	btrfs_mark_buffer_dirty(path->nodes[0]);
	btrfs_free_path(path);

4088
	btrfs_set_fs_incompat(fs_info, DEFAULT_SUBVOL);
4089
	btrfs_end_transaction(trans);
4090 4091 4092
out:
	mnt_drop_write_file(file);
	return ret;
4093 4094
}

4095 4096
void btrfs_get_block_group_info(struct list_head *groups_list,
				struct btrfs_ioctl_space_info *space)
4097 4098 4099 4100 4101 4102 4103 4104 4105 4106 4107 4108 4109 4110
{
	struct btrfs_block_group_cache *block_group;

	space->total_bytes = 0;
	space->used_bytes = 0;
	space->flags = 0;
	list_for_each_entry(block_group, groups_list, list) {
		space->flags = block_group->flags;
		space->total_bytes += block_group->key.offset;
		space->used_bytes +=
			btrfs_block_group_used(&block_group->item);
	}
}

4111 4112
static long btrfs_ioctl_space_info(struct btrfs_fs_info *fs_info,
				   void __user *arg)
J
Josef Bacik 已提交
4113 4114 4115 4116
{
	struct btrfs_ioctl_space_args space_args;
	struct btrfs_ioctl_space_info space;
	struct btrfs_ioctl_space_info *dest;
4117
	struct btrfs_ioctl_space_info *dest_orig;
4118
	struct btrfs_ioctl_space_info __user *user_dest;
J
Josef Bacik 已提交
4119
	struct btrfs_space_info *info;
4120 4121 4122 4123 4124 4125
	static const u64 types[] = {
		BTRFS_BLOCK_GROUP_DATA,
		BTRFS_BLOCK_GROUP_SYSTEM,
		BTRFS_BLOCK_GROUP_METADATA,
		BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA
	};
4126
	int num_types = 4;
4127
	int alloc_size;
J
Josef Bacik 已提交
4128
	int ret = 0;
4129
	u64 slot_count = 0;
4130
	int i, c;
J
Josef Bacik 已提交
4131 4132 4133 4134 4135 4136

	if (copy_from_user(&space_args,
			   (struct btrfs_ioctl_space_args __user *)arg,
			   sizeof(space_args)))
		return -EFAULT;

4137 4138 4139 4140 4141
	for (i = 0; i < num_types; i++) {
		struct btrfs_space_info *tmp;

		info = NULL;
		rcu_read_lock();
4142
		list_for_each_entry_rcu(tmp, &fs_info->space_info,
4143 4144 4145 4146 4147 4148 4149 4150 4151 4152 4153 4154 4155 4156 4157 4158 4159 4160
					list) {
			if (tmp->flags == types[i]) {
				info = tmp;
				break;
			}
		}
		rcu_read_unlock();

		if (!info)
			continue;

		down_read(&info->groups_sem);
		for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
			if (!list_empty(&info->block_groups[c]))
				slot_count++;
		}
		up_read(&info->groups_sem);
	}
4161

4162 4163 4164 4165 4166
	/*
	 * Global block reserve, exported as a space_info
	 */
	slot_count++;

4167 4168 4169 4170 4171
	/* space_slots == 0 means they are asking for a count */
	if (space_args.space_slots == 0) {
		space_args.total_spaces = slot_count;
		goto out;
	}
4172

4173
	slot_count = min_t(u64, space_args.space_slots, slot_count);
4174

4175
	alloc_size = sizeof(*dest) * slot_count;
4176

4177 4178 4179
	/* we generally have at most 6 or so space infos, one for each raid
	 * level.  So, a whole page should be more than enough for everyone
	 */
4180
	if (alloc_size > PAGE_SIZE)
4181 4182
		return -ENOMEM;

J
Josef Bacik 已提交
4183
	space_args.total_spaces = 0;
4184
	dest = kmalloc(alloc_size, GFP_KERNEL);
4185 4186 4187
	if (!dest)
		return -ENOMEM;
	dest_orig = dest;
J
Josef Bacik 已提交
4188

4189
	/* now we have a buffer to copy into */
4190 4191 4192
	for (i = 0; i < num_types; i++) {
		struct btrfs_space_info *tmp;

4193 4194 4195
		if (!slot_count)
			break;

4196 4197
		info = NULL;
		rcu_read_lock();
4198
		list_for_each_entry_rcu(tmp, &fs_info->space_info,
4199 4200 4201 4202 4203 4204 4205
					list) {
			if (tmp->flags == types[i]) {
				info = tmp;
				break;
			}
		}
		rcu_read_unlock();
4206

4207 4208 4209 4210 4211
		if (!info)
			continue;
		down_read(&info->groups_sem);
		for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) {
			if (!list_empty(&info->block_groups[c])) {
4212 4213
				btrfs_get_block_group_info(
					&info->block_groups[c], &space);
4214 4215 4216
				memcpy(dest, &space, sizeof(space));
				dest++;
				space_args.total_spaces++;
4217
				slot_count--;
4218
			}
4219 4220
			if (!slot_count)
				break;
4221 4222
		}
		up_read(&info->groups_sem);
J
Josef Bacik 已提交
4223 4224
	}

4225 4226 4227 4228
	/*
	 * Add global block reserve
	 */
	if (slot_count) {
4229
		struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv;
4230 4231 4232 4233 4234 4235 4236 4237 4238 4239

		spin_lock(&block_rsv->lock);
		space.total_bytes = block_rsv->size;
		space.used_bytes = block_rsv->size - block_rsv->reserved;
		spin_unlock(&block_rsv->lock);
		space.flags = BTRFS_SPACE_INFO_GLOBAL_RSV;
		memcpy(dest, &space, sizeof(space));
		space_args.total_spaces++;
	}

D
Daniel J Blueman 已提交
4240
	user_dest = (struct btrfs_ioctl_space_info __user *)
4241 4242 4243 4244 4245 4246 4247 4248
		(arg + sizeof(struct btrfs_ioctl_space_args));

	if (copy_to_user(user_dest, dest_orig, alloc_size))
		ret = -EFAULT;

	kfree(dest_orig);
out:
	if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args)))
J
Josef Bacik 已提交
4249 4250 4251 4252 4253
		ret = -EFAULT;

	return ret;
}

C
Christoph Hellwig 已提交
4254 4255 4256 4257 4258 4259 4260 4261
/*
 * there are many ways the trans_start and trans_end ioctls can lead
 * to deadlocks.  They should only be used by applications that
 * basically own the machine, and have a very in depth understanding
 * of all the possible deadlocks and enospc problems.
 */
long btrfs_ioctl_trans_end(struct file *file)
{
A
Al Viro 已提交
4262
	struct inode *inode = file_inode(file);
C
Christoph Hellwig 已提交
4263
	struct btrfs_root *root = BTRFS_I(inode)->root;
4264
	struct btrfs_file_private *private = file->private_data;
C
Christoph Hellwig 已提交
4265

4266
	if (!private || !private->trans)
4267
		return -EINVAL;
4268

4269 4270
	btrfs_end_transaction(private->trans);
	private->trans = NULL;
4271

J
Josef Bacik 已提交
4272
	atomic_dec(&root->fs_info->open_ioctl_trans);
4273

A
Al Viro 已提交
4274
	mnt_drop_write_file(file);
4275
	return 0;
C
Christoph Hellwig 已提交
4276 4277
}

4278 4279
static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root,
					    void __user *argp)
4280 4281 4282
{
	struct btrfs_trans_handle *trans;
	u64 transid;
T
Tsutomu Itoh 已提交
4283
	int ret;
4284

M
Miao Xie 已提交
4285
	trans = btrfs_attach_transaction_barrier(root);
4286 4287 4288 4289 4290 4291 4292 4293
	if (IS_ERR(trans)) {
		if (PTR_ERR(trans) != -ENOENT)
			return PTR_ERR(trans);

		/* No running transaction, don't bother */
		transid = root->fs_info->last_trans_committed;
		goto out;
	}
4294
	transid = trans->transid;
4295
	ret = btrfs_commit_transaction_async(trans, 0);
4296
	if (ret) {
4297
		btrfs_end_transaction(trans);
T
Tsutomu Itoh 已提交
4298
		return ret;
4299
	}
4300
out:
4301 4302 4303 4304 4305 4306
	if (argp)
		if (copy_to_user(argp, &transid, sizeof(transid)))
			return -EFAULT;
	return 0;
}

4307
static noinline long btrfs_ioctl_wait_sync(struct btrfs_fs_info *fs_info,
4308
					   void __user *argp)
4309 4310 4311 4312 4313 4314 4315 4316 4317
{
	u64 transid;

	if (argp) {
		if (copy_from_user(&transid, argp, sizeof(transid)))
			return -EFAULT;
	} else {
		transid = 0;  /* current trans */
	}
4318
	return btrfs_wait_for_commit(fs_info, transid);
4319 4320
}

M
Miao Xie 已提交
4321
static long btrfs_ioctl_scrub(struct file *file, void __user *arg)
J
Jan Schmidt 已提交
4322
{
4323
	struct btrfs_fs_info *fs_info = btrfs_sb(file_inode(file)->i_sb);
J
Jan Schmidt 已提交
4324
	struct btrfs_ioctl_scrub_args *sa;
M
Miao Xie 已提交
4325
	int ret;
J
Jan Schmidt 已提交
4326 4327 4328 4329 4330 4331 4332 4333

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

	sa = memdup_user(arg, sizeof(*sa));
	if (IS_ERR(sa))
		return PTR_ERR(sa);

M
Miao Xie 已提交
4334 4335 4336 4337 4338 4339
	if (!(sa->flags & BTRFS_SCRUB_READONLY)) {
		ret = mnt_want_write_file(file);
		if (ret)
			goto out;
	}

4340
	ret = btrfs_scrub_dev(fs_info, sa->devid, sa->start, sa->end,
4341 4342
			      &sa->progress, sa->flags & BTRFS_SCRUB_READONLY,
			      0);
J
Jan Schmidt 已提交
4343 4344 4345 4346

	if (copy_to_user(arg, sa, sizeof(*sa)))
		ret = -EFAULT;

M
Miao Xie 已提交
4347 4348 4349
	if (!(sa->flags & BTRFS_SCRUB_READONLY))
		mnt_drop_write_file(file);
out:
J
Jan Schmidt 已提交
4350 4351 4352 4353
	kfree(sa);
	return ret;
}

4354
static long btrfs_ioctl_scrub_cancel(struct btrfs_fs_info *fs_info)
J
Jan Schmidt 已提交
4355 4356 4357 4358
{
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

4359
	return btrfs_scrub_cancel(fs_info);
J
Jan Schmidt 已提交
4360 4361
}

4362
static long btrfs_ioctl_scrub_progress(struct btrfs_fs_info *fs_info,
J
Jan Schmidt 已提交
4363 4364 4365 4366 4367 4368 4369 4370 4371 4372 4373 4374
				       void __user *arg)
{
	struct btrfs_ioctl_scrub_args *sa;
	int ret;

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

	sa = memdup_user(arg, sizeof(*sa));
	if (IS_ERR(sa))
		return PTR_ERR(sa);

4375
	ret = btrfs_scrub_progress(fs_info, sa->devid, &sa->progress);
J
Jan Schmidt 已提交
4376 4377 4378 4379 4380 4381 4382 4383

	if (copy_to_user(arg, sa, sizeof(*sa)))
		ret = -EFAULT;

	kfree(sa);
	return ret;
}

4384
static long btrfs_ioctl_get_dev_stats(struct btrfs_fs_info *fs_info,
4385
				      void __user *arg)
4386 4387 4388 4389 4390 4391 4392 4393
{
	struct btrfs_ioctl_get_dev_stats *sa;
	int ret;

	sa = memdup_user(arg, sizeof(*sa));
	if (IS_ERR(sa))
		return PTR_ERR(sa);

4394 4395 4396 4397 4398
	if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) {
		kfree(sa);
		return -EPERM;
	}

4399
	ret = btrfs_get_dev_stats(fs_info, sa);
4400 4401 4402 4403 4404 4405 4406 4407

	if (copy_to_user(arg, sa, sizeof(*sa)))
		ret = -EFAULT;

	kfree(sa);
	return ret;
}

4408 4409
static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info,
				    void __user *arg)
4410 4411 4412 4413 4414 4415 4416 4417 4418 4419 4420 4421 4422
{
	struct btrfs_ioctl_dev_replace_args *p;
	int ret;

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

	p = memdup_user(arg, sizeof(*p));
	if (IS_ERR(p))
		return PTR_ERR(p);

	switch (p->cmd) {
	case BTRFS_IOCTL_DEV_REPLACE_CMD_START:
4423
		if (sb_rdonly(fs_info->sb)) {
4424 4425 4426
			ret = -EROFS;
			goto out;
		}
4427
		if (test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
4428
			ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
4429
		} else {
4430
			ret = btrfs_dev_replace_by_ioctl(fs_info, p);
4431
			clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
4432 4433 4434
		}
		break;
	case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS:
4435
		btrfs_dev_replace_status(fs_info, p);
4436 4437 4438
		ret = 0;
		break;
	case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL:
4439
		ret = btrfs_dev_replace_cancel(fs_info, p);
4440 4441 4442 4443 4444 4445 4446 4447
		break;
	default:
		ret = -EINVAL;
		break;
	}

	if (copy_to_user(arg, p, sizeof(*p)))
		ret = -EFAULT;
4448
out:
4449 4450 4451 4452
	kfree(p);
	return ret;
}

4453 4454 4455 4456
static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg)
{
	int ret = 0;
	int i;
4457
	u64 rel_ptr;
4458
	int size;
4459
	struct btrfs_ioctl_ino_path_args *ipa = NULL;
4460 4461 4462
	struct inode_fs_paths *ipath = NULL;
	struct btrfs_path *path;

4463
	if (!capable(CAP_DAC_READ_SEARCH))
4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489 4490 4491
		return -EPERM;

	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto out;
	}

	ipa = memdup_user(arg, sizeof(*ipa));
	if (IS_ERR(ipa)) {
		ret = PTR_ERR(ipa);
		ipa = NULL;
		goto out;
	}

	size = min_t(u32, ipa->size, 4096);
	ipath = init_ipath(size, root, path);
	if (IS_ERR(ipath)) {
		ret = PTR_ERR(ipath);
		ipath = NULL;
		goto out;
	}

	ret = paths_from_inode(ipa->inum, ipath);
	if (ret < 0)
		goto out;

	for (i = 0; i < ipath->fspath->elem_cnt; ++i) {
4492 4493
		rel_ptr = ipath->fspath->val[i] -
			  (u64)(unsigned long)ipath->fspath->val;
4494
		ipath->fspath->val[i] = rel_ptr;
4495 4496
	}

4497 4498
	ret = copy_to_user((void __user *)(unsigned long)ipa->fspath,
			   ipath->fspath, size);
4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531
	if (ret) {
		ret = -EFAULT;
		goto out;
	}

out:
	btrfs_free_path(path);
	free_ipath(ipath);
	kfree(ipa);

	return ret;
}

static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx)
{
	struct btrfs_data_container *inodes = ctx;
	const size_t c = 3 * sizeof(u64);

	if (inodes->bytes_left >= c) {
		inodes->bytes_left -= c;
		inodes->val[inodes->elem_cnt] = inum;
		inodes->val[inodes->elem_cnt + 1] = offset;
		inodes->val[inodes->elem_cnt + 2] = root;
		inodes->elem_cnt += 3;
	} else {
		inodes->bytes_missing += c - inodes->bytes_left;
		inodes->bytes_left = 0;
		inodes->elem_missed += 3;
	}

	return 0;
}

4532
static long btrfs_ioctl_logical_to_ino(struct btrfs_fs_info *fs_info,
4533
					void __user *arg, int version)
4534 4535 4536 4537 4538 4539
{
	int ret = 0;
	int size;
	struct btrfs_ioctl_logical_ino_args *loi;
	struct btrfs_data_container *inodes = NULL;
	struct btrfs_path *path = NULL;
4540
	bool ignore_offset;
4541 4542 4543 4544 4545

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

	loi = memdup_user(arg, sizeof(*loi));
4546 4547
	if (IS_ERR(loi))
		return PTR_ERR(loi);
4548

4549 4550
	if (version == 1) {
		ignore_offset = false;
4551
		size = min_t(u32, loi->size, SZ_64K);
4552 4553 4554 4555 4556 4557 4558 4559 4560 4561 4562 4563
	} else {
		/* All reserved bits must be 0 for now */
		if (memchr_inv(loi->reserved, 0, sizeof(loi->reserved))) {
			ret = -EINVAL;
			goto out_loi;
		}
		/* Only accept flags we have defined so far */
		if (loi->flags & ~(BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET)) {
			ret = -EINVAL;
			goto out_loi;
		}
		ignore_offset = loi->flags & BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET;
4564
		size = min_t(u32, loi->size, SZ_16M);
4565 4566
	}

4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579
	path = btrfs_alloc_path();
	if (!path) {
		ret = -ENOMEM;
		goto out;
	}

	inodes = init_data_container(size);
	if (IS_ERR(inodes)) {
		ret = PTR_ERR(inodes);
		inodes = NULL;
		goto out;
	}

4580
	ret = iterate_inodes_from_logical(loi->logical, fs_info, path,
4581
					  build_ino_list, inodes, ignore_offset);
L
Liu Bo 已提交
4582
	if (ret == -EINVAL)
4583 4584 4585 4586
		ret = -ENOENT;
	if (ret < 0)
		goto out;

4587 4588
	ret = copy_to_user((void __user *)(unsigned long)loi->inodes, inodes,
			   size);
4589 4590 4591 4592 4593
	if (ret)
		ret = -EFAULT;

out:
	btrfs_free_path(path);
4594
	kvfree(inodes);
4595
out_loi:
4596 4597 4598 4599 4600
	kfree(loi);

	return ret;
}

4601
void update_ioctl_balance_args(struct btrfs_fs_info *fs_info, int lock,
4602 4603 4604 4605 4606 4607
			       struct btrfs_ioctl_balance_args *bargs)
{
	struct btrfs_balance_control *bctl = fs_info->balance_ctl;

	bargs->flags = bctl->flags;

4608 4609 4610 4611
	if (atomic_read(&fs_info->balance_running))
		bargs->state |= BTRFS_BALANCE_STATE_RUNNING;
	if (atomic_read(&fs_info->balance_pause_req))
		bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ;
4612 4613
	if (atomic_read(&fs_info->balance_cancel_req))
		bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ;
4614

4615 4616 4617
	memcpy(&bargs->data, &bctl->data, sizeof(bargs->data));
	memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta));
	memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys));
4618 4619 4620 4621 4622 4623 4624 4625

	if (lock) {
		spin_lock(&fs_info->balance_lock);
		memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
		spin_unlock(&fs_info->balance_lock);
	} else {
		memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat));
	}
4626 4627
}

4628
static long btrfs_ioctl_balance(struct file *file, void __user *arg)
4629
{
A
Al Viro 已提交
4630
	struct btrfs_root *root = BTRFS_I(file_inode(file))->root;
4631 4632 4633
	struct btrfs_fs_info *fs_info = root->fs_info;
	struct btrfs_ioctl_balance_args *bargs;
	struct btrfs_balance_control *bctl;
4634
	bool need_unlock; /* for mut. excl. ops lock */
4635 4636 4637 4638 4639
	int ret;

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

4640
	ret = mnt_want_write_file(file);
4641 4642 4643
	if (ret)
		return ret;

4644
again:
4645
	if (!test_and_set_bit(BTRFS_FS_EXCL_OP, &fs_info->flags)) {
4646 4647 4648 4649 4650 4651 4652
		mutex_lock(&fs_info->volume_mutex);
		mutex_lock(&fs_info->balance_mutex);
		need_unlock = true;
		goto locked;
	}

	/*
4653
	 * mut. excl. ops lock is locked.  Three possibilities:
4654 4655 4656 4657
	 *   (1) some other op is running
	 *   (2) balance is running
	 *   (3) balance is paused -- special case (think resume)
	 */
4658
	mutex_lock(&fs_info->balance_mutex);
4659 4660 4661 4662 4663 4664 4665 4666 4667 4668 4669 4670 4671 4672 4673 4674 4675 4676 4677 4678 4679 4680 4681 4682 4683 4684 4685
	if (fs_info->balance_ctl) {
		/* this is either (2) or (3) */
		if (!atomic_read(&fs_info->balance_running)) {
			mutex_unlock(&fs_info->balance_mutex);
			if (!mutex_trylock(&fs_info->volume_mutex))
				goto again;
			mutex_lock(&fs_info->balance_mutex);

			if (fs_info->balance_ctl &&
			    !atomic_read(&fs_info->balance_running)) {
				/* this is (3) */
				need_unlock = false;
				goto locked;
			}

			mutex_unlock(&fs_info->balance_mutex);
			mutex_unlock(&fs_info->volume_mutex);
			goto again;
		} else {
			/* this is (2) */
			mutex_unlock(&fs_info->balance_mutex);
			ret = -EINPROGRESS;
			goto out;
		}
	} else {
		/* this is (1) */
		mutex_unlock(&fs_info->balance_mutex);
4686
		ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS;
4687 4688 4689 4690
		goto out;
	}

locked:
4691
	BUG_ON(!test_bit(BTRFS_FS_EXCL_OP, &fs_info->flags));
4692 4693 4694 4695 4696

	if (arg) {
		bargs = memdup_user(arg, sizeof(*bargs));
		if (IS_ERR(bargs)) {
			ret = PTR_ERR(bargs);
4697
			goto out_unlock;
4698
		}
4699 4700 4701 4702 4703 4704 4705 4706 4707 4708 4709 4710 4711 4712

		if (bargs->flags & BTRFS_BALANCE_RESUME) {
			if (!fs_info->balance_ctl) {
				ret = -ENOTCONN;
				goto out_bargs;
			}

			bctl = fs_info->balance_ctl;
			spin_lock(&fs_info->balance_lock);
			bctl->flags |= BTRFS_BALANCE_RESUME;
			spin_unlock(&fs_info->balance_lock);

			goto do_balance;
		}
4713 4714 4715 4716
	} else {
		bargs = NULL;
	}

4717
	if (fs_info->balance_ctl) {
4718 4719 4720 4721
		ret = -EINPROGRESS;
		goto out_bargs;
	}

4722
	bctl = kzalloc(sizeof(*bctl), GFP_KERNEL);
4723 4724 4725 4726 4727 4728 4729 4730 4731 4732 4733 4734
	if (!bctl) {
		ret = -ENOMEM;
		goto out_bargs;
	}

	bctl->fs_info = fs_info;
	if (arg) {
		memcpy(&bctl->data, &bargs->data, sizeof(bctl->data));
		memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta));
		memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys));

		bctl->flags = bargs->flags;
4735 4736 4737
	} else {
		/* balance everything - no filters */
		bctl->flags |= BTRFS_BALANCE_TYPE_MASK;
4738 4739
	}

4740 4741
	if (bctl->flags & ~(BTRFS_BALANCE_ARGS_MASK | BTRFS_BALANCE_TYPE_MASK)) {
		ret = -EINVAL;
4742
		goto out_bctl;
4743 4744
	}

4745
do_balance:
4746
	/*
4747
	 * Ownership of bctl and filesystem flag BTRFS_FS_EXCL_OP
4748 4749
	 * goes to to btrfs_balance.  bctl is freed in __cancel_balance,
	 * or, if restriper was paused all the way until unmount, in
4750
	 * free_fs_info.  The flag is cleared in __cancel_balance.
4751
	 */
4752 4753 4754
	need_unlock = false;

	ret = btrfs_balance(bctl, bargs);
4755
	bctl = NULL;
4756

4757 4758 4759 4760 4761
	if (arg) {
		if (copy_to_user(arg, bargs, sizeof(*bargs)))
			ret = -EFAULT;
	}

4762 4763
out_bctl:
	kfree(bctl);
4764 4765
out_bargs:
	kfree(bargs);
4766
out_unlock:
4767 4768
	mutex_unlock(&fs_info->balance_mutex);
	mutex_unlock(&fs_info->volume_mutex);
4769
	if (need_unlock)
4770
		clear_bit(BTRFS_FS_EXCL_OP, &fs_info->flags);
4771
out:
4772
	mnt_drop_write_file(file);
4773 4774 4775
	return ret;
}

4776
static long btrfs_ioctl_balance_ctl(struct btrfs_fs_info *fs_info, int cmd)
4777 4778 4779 4780 4781 4782
{
	if (!capable(CAP_SYS_ADMIN))
		return -EPERM;

	switch (cmd) {
	case BTRFS_BALANCE_CTL_PAUSE:
4783
		return btrfs_pause_balance(fs_info);
4784
	case BTRFS_BALANCE_CTL_CANCEL:
4785
		return btrfs_cancel_balance(fs_info);
4786 4787 4788 4789 4790
	}

	return -EINVAL;
}

4791
static long btrfs_ioctl_balance_progress(struct btrfs_fs_info *fs_info,
4792 4793 4794 4795 4796 4797 4798 4799 4800 4801 4802 4803 4804 4805
					 void __user *arg)
{
	struct btrfs_ioctl_balance_args *bargs;
	int ret = 0;

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

	mutex_lock(&fs_info->balance_mutex);
	if (!fs_info->balance_ctl) {
		ret = -ENOTCONN;
		goto out;
	}

4806
	bargs = kzalloc(sizeof(*bargs), GFP_KERNEL);
4807 4808 4809 4810 4811 4812 4813 4814 4815 4816 4817 4818 4819 4820 4821 4822
	if (!bargs) {
		ret = -ENOMEM;
		goto out;
	}

	update_ioctl_balance_args(fs_info, 1, bargs);

	if (copy_to_user(arg, bargs, sizeof(*bargs)))
		ret = -EFAULT;

	kfree(bargs);
out:
	mutex_unlock(&fs_info->balance_mutex);
	return ret;
}

4823
static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg)
A
Arne Jansen 已提交
4824
{
4825 4826
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
A
Arne Jansen 已提交
4827 4828 4829 4830 4831 4832 4833 4834
	struct btrfs_ioctl_quota_ctl_args *sa;
	struct btrfs_trans_handle *trans = NULL;
	int ret;
	int err;

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

4835 4836 4837
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
A
Arne Jansen 已提交
4838 4839

	sa = memdup_user(arg, sizeof(*sa));
4840 4841 4842 4843
	if (IS_ERR(sa)) {
		ret = PTR_ERR(sa);
		goto drop_write;
	}
A
Arne Jansen 已提交
4844

4845 4846
	down_write(&fs_info->subvol_sem);
	trans = btrfs_start_transaction(fs_info->tree_root, 2);
J
Jan Schmidt 已提交
4847 4848 4849
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out;
A
Arne Jansen 已提交
4850 4851 4852 4853
	}

	switch (sa->cmd) {
	case BTRFS_QUOTA_CTL_ENABLE:
4854
		ret = btrfs_quota_enable(trans, fs_info);
A
Arne Jansen 已提交
4855 4856
		break;
	case BTRFS_QUOTA_CTL_DISABLE:
4857
		ret = btrfs_quota_disable(trans, fs_info);
A
Arne Jansen 已提交
4858 4859 4860 4861 4862 4863
		break;
	default:
		ret = -EINVAL;
		break;
	}

4864
	err = btrfs_commit_transaction(trans);
J
Jan Schmidt 已提交
4865 4866
	if (err && !ret)
		ret = err;
A
Arne Jansen 已提交
4867 4868
out:
	kfree(sa);
4869
	up_write(&fs_info->subvol_sem);
4870 4871
drop_write:
	mnt_drop_write_file(file);
A
Arne Jansen 已提交
4872 4873 4874
	return ret;
}

4875
static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg)
A
Arne Jansen 已提交
4876
{
4877 4878 4879
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
A
Arne Jansen 已提交
4880 4881 4882 4883 4884 4885 4886 4887
	struct btrfs_ioctl_qgroup_assign_args *sa;
	struct btrfs_trans_handle *trans;
	int ret;
	int err;

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

4888 4889 4890
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
A
Arne Jansen 已提交
4891 4892

	sa = memdup_user(arg, sizeof(*sa));
4893 4894 4895 4896
	if (IS_ERR(sa)) {
		ret = PTR_ERR(sa);
		goto drop_write;
	}
A
Arne Jansen 已提交
4897 4898 4899 4900 4901 4902 4903 4904

	trans = btrfs_join_transaction(root);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out;
	}

	if (sa->assign) {
4905
		ret = btrfs_add_qgroup_relation(trans, fs_info,
A
Arne Jansen 已提交
4906 4907
						sa->src, sa->dst);
	} else {
4908
		ret = btrfs_del_qgroup_relation(trans, fs_info,
A
Arne Jansen 已提交
4909 4910 4911
						sa->src, sa->dst);
	}

4912
	/* update qgroup status and info */
4913
	err = btrfs_run_qgroups(trans, fs_info);
4914
	if (err < 0)
4915 4916
		btrfs_handle_fs_error(fs_info, err,
				      "failed to update qgroup status and info");
4917
	err = btrfs_end_transaction(trans);
A
Arne Jansen 已提交
4918 4919 4920 4921 4922
	if (err && !ret)
		ret = err;

out:
	kfree(sa);
4923 4924
drop_write:
	mnt_drop_write_file(file);
A
Arne Jansen 已提交
4925 4926 4927
	return ret;
}

4928
static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg)
A
Arne Jansen 已提交
4929
{
4930 4931 4932
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
A
Arne Jansen 已提交
4933 4934 4935 4936 4937 4938 4939 4940
	struct btrfs_ioctl_qgroup_create_args *sa;
	struct btrfs_trans_handle *trans;
	int ret;
	int err;

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

4941 4942 4943
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
A
Arne Jansen 已提交
4944 4945

	sa = memdup_user(arg, sizeof(*sa));
4946 4947 4948 4949
	if (IS_ERR(sa)) {
		ret = PTR_ERR(sa);
		goto drop_write;
	}
A
Arne Jansen 已提交
4950

M
Miao Xie 已提交
4951 4952 4953 4954 4955
	if (!sa->qgroupid) {
		ret = -EINVAL;
		goto out;
	}

A
Arne Jansen 已提交
4956 4957 4958 4959 4960 4961 4962
	trans = btrfs_join_transaction(root);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out;
	}

	if (sa->create) {
4963
		ret = btrfs_create_qgroup(trans, fs_info, sa->qgroupid);
A
Arne Jansen 已提交
4964
	} else {
4965
		ret = btrfs_remove_qgroup(trans, fs_info, sa->qgroupid);
A
Arne Jansen 已提交
4966 4967
	}

4968
	err = btrfs_end_transaction(trans);
A
Arne Jansen 已提交
4969 4970 4971 4972 4973
	if (err && !ret)
		ret = err;

out:
	kfree(sa);
4974 4975
drop_write:
	mnt_drop_write_file(file);
A
Arne Jansen 已提交
4976 4977 4978
	return ret;
}

4979
static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg)
A
Arne Jansen 已提交
4980
{
4981 4982 4983
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
A
Arne Jansen 已提交
4984 4985 4986 4987 4988 4989 4990 4991 4992
	struct btrfs_ioctl_qgroup_limit_args *sa;
	struct btrfs_trans_handle *trans;
	int ret;
	int err;
	u64 qgroupid;

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

4993 4994 4995
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;
A
Arne Jansen 已提交
4996 4997

	sa = memdup_user(arg, sizeof(*sa));
4998 4999 5000 5001
	if (IS_ERR(sa)) {
		ret = PTR_ERR(sa);
		goto drop_write;
	}
A
Arne Jansen 已提交
5002 5003 5004 5005 5006 5007 5008 5009 5010 5011 5012 5013 5014

	trans = btrfs_join_transaction(root);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out;
	}

	qgroupid = sa->qgroupid;
	if (!qgroupid) {
		/* take the current subvol as qgroup */
		qgroupid = root->root_key.objectid;
	}

5015
	ret = btrfs_limit_qgroup(trans, fs_info, qgroupid, &sa->lim);
A
Arne Jansen 已提交
5016

5017
	err = btrfs_end_transaction(trans);
A
Arne Jansen 已提交
5018 5019 5020 5021 5022
	if (err && !ret)
		ret = err;

out:
	kfree(sa);
5023 5024
drop_write:
	mnt_drop_write_file(file);
A
Arne Jansen 已提交
5025 5026 5027
	return ret;
}

J
Jan Schmidt 已提交
5028 5029
static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg)
{
5030 5031
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
J
Jan Schmidt 已提交
5032 5033 5034 5035 5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052
	struct btrfs_ioctl_quota_rescan_args *qsa;
	int ret;

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

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

	qsa = memdup_user(arg, sizeof(*qsa));
	if (IS_ERR(qsa)) {
		ret = PTR_ERR(qsa);
		goto drop_write;
	}

	if (qsa->flags) {
		ret = -EINVAL;
		goto out;
	}

5053
	ret = btrfs_qgroup_rescan(fs_info);
J
Jan Schmidt 已提交
5054 5055 5056 5057 5058 5059 5060 5061 5062 5063

out:
	kfree(qsa);
drop_write:
	mnt_drop_write_file(file);
	return ret;
}

static long btrfs_ioctl_quota_rescan_status(struct file *file, void __user *arg)
{
5064 5065
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
J
Jan Schmidt 已提交
5066 5067 5068 5069 5070 5071
	struct btrfs_ioctl_quota_rescan_args *qsa;
	int ret = 0;

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

5072
	qsa = kzalloc(sizeof(*qsa), GFP_KERNEL);
J
Jan Schmidt 已提交
5073 5074 5075
	if (!qsa)
		return -ENOMEM;

5076
	if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) {
J
Jan Schmidt 已提交
5077
		qsa->flags = 1;
5078
		qsa->progress = fs_info->qgroup_rescan_progress.objectid;
J
Jan Schmidt 已提交
5079 5080 5081 5082 5083 5084 5085 5086 5087
	}

	if (copy_to_user(arg, qsa, sizeof(*qsa)))
		ret = -EFAULT;

	kfree(qsa);
	return ret;
}

5088 5089
static long btrfs_ioctl_quota_rescan_wait(struct file *file, void __user *arg)
{
5090 5091
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
5092 5093 5094 5095

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

5096
	return btrfs_qgroup_wait_for_completion(fs_info, true);
5097 5098
}

5099 5100
static long _btrfs_ioctl_set_received_subvol(struct file *file,
					    struct btrfs_ioctl_received_subvol_args *sa)
5101
{
A
Al Viro 已提交
5102
	struct inode *inode = file_inode(file);
5103
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
5104 5105 5106
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_root_item *root_item = &root->root_item;
	struct btrfs_trans_handle *trans;
5107
	struct timespec ct = current_time(inode);
5108
	int ret = 0;
5109
	int received_uuid_changed;
5110

5111 5112 5113
	if (!inode_owner_or_capable(inode))
		return -EPERM;

5114 5115 5116 5117
	ret = mnt_want_write_file(file);
	if (ret < 0)
		return ret;

5118
	down_write(&fs_info->subvol_sem);
5119

5120
	if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) {
5121 5122 5123 5124 5125 5126 5127 5128 5129
		ret = -EINVAL;
		goto out;
	}

	if (btrfs_root_readonly(root)) {
		ret = -EROFS;
		goto out;
	}

5130 5131 5132 5133 5134
	/*
	 * 1 - root item
	 * 2 - uuid items (received uuid + subvol uuid)
	 */
	trans = btrfs_start_transaction(root, 3);
5135 5136 5137 5138 5139 5140 5141 5142 5143 5144
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		trans = NULL;
		goto out;
	}

	sa->rtransid = trans->transid;
	sa->rtime.sec = ct.tv_sec;
	sa->rtime.nsec = ct.tv_nsec;

5145 5146 5147 5148
	received_uuid_changed = memcmp(root_item->received_uuid, sa->uuid,
				       BTRFS_UUID_SIZE);
	if (received_uuid_changed &&
	    !btrfs_is_empty_uuid(root_item->received_uuid))
5149
		btrfs_uuid_tree_rem(trans, fs_info, root_item->received_uuid,
5150 5151
				    BTRFS_UUID_KEY_RECEIVED_SUBVOL,
				    root->root_key.objectid);
5152 5153 5154
	memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE);
	btrfs_set_root_stransid(root_item, sa->stransid);
	btrfs_set_root_rtransid(root_item, sa->rtransid);
5155 5156 5157 5158
	btrfs_set_stack_timespec_sec(&root_item->stime, sa->stime.sec);
	btrfs_set_stack_timespec_nsec(&root_item->stime, sa->stime.nsec);
	btrfs_set_stack_timespec_sec(&root_item->rtime, sa->rtime.sec);
	btrfs_set_stack_timespec_nsec(&root_item->rtime, sa->rtime.nsec);
5159

5160
	ret = btrfs_update_root(trans, fs_info->tree_root,
5161 5162
				&root->root_key, &root->root_item);
	if (ret < 0) {
5163
		btrfs_end_transaction(trans);
5164
		goto out;
5165 5166
	}
	if (received_uuid_changed && !btrfs_is_empty_uuid(sa->uuid)) {
5167
		ret = btrfs_uuid_tree_add(trans, fs_info, sa->uuid,
5168 5169 5170
					  BTRFS_UUID_KEY_RECEIVED_SUBVOL,
					  root->root_key.objectid);
		if (ret < 0 && ret != -EEXIST) {
5171
			btrfs_abort_transaction(trans, ret);
5172
			btrfs_end_transaction(trans);
5173
			goto out;
5174 5175
		}
	}
5176
	ret = btrfs_commit_transaction(trans);
5177
out:
5178
	up_write(&fs_info->subvol_sem);
5179 5180 5181 5182 5183 5184 5185 5186 5187 5188 5189 5190 5191
	mnt_drop_write_file(file);
	return ret;
}

#ifdef CONFIG_64BIT
static long btrfs_ioctl_set_received_subvol_32(struct file *file,
						void __user *arg)
{
	struct btrfs_ioctl_received_subvol_args_32 *args32 = NULL;
	struct btrfs_ioctl_received_subvol_args *args64 = NULL;
	int ret = 0;

	args32 = memdup_user(arg, sizeof(*args32));
5192 5193
	if (IS_ERR(args32))
		return PTR_ERR(args32);
5194

5195
	args64 = kmalloc(sizeof(*args64), GFP_KERNEL);
5196 5197
	if (!args64) {
		ret = -ENOMEM;
5198 5199 5200 5201 5202 5203 5204 5205 5206 5207 5208 5209 5210 5211 5212 5213 5214 5215 5216 5217 5218 5219 5220 5221 5222 5223 5224 5225 5226 5227 5228 5229 5230 5231 5232 5233 5234 5235 5236 5237 5238 5239 5240
		goto out;
	}

	memcpy(args64->uuid, args32->uuid, BTRFS_UUID_SIZE);
	args64->stransid = args32->stransid;
	args64->rtransid = args32->rtransid;
	args64->stime.sec = args32->stime.sec;
	args64->stime.nsec = args32->stime.nsec;
	args64->rtime.sec = args32->rtime.sec;
	args64->rtime.nsec = args32->rtime.nsec;
	args64->flags = args32->flags;

	ret = _btrfs_ioctl_set_received_subvol(file, args64);
	if (ret)
		goto out;

	memcpy(args32->uuid, args64->uuid, BTRFS_UUID_SIZE);
	args32->stransid = args64->stransid;
	args32->rtransid = args64->rtransid;
	args32->stime.sec = args64->stime.sec;
	args32->stime.nsec = args64->stime.nsec;
	args32->rtime.sec = args64->rtime.sec;
	args32->rtime.nsec = args64->rtime.nsec;
	args32->flags = args64->flags;

	ret = copy_to_user(arg, args32, sizeof(*args32));
	if (ret)
		ret = -EFAULT;

out:
	kfree(args32);
	kfree(args64);
	return ret;
}
#endif

static long btrfs_ioctl_set_received_subvol(struct file *file,
					    void __user *arg)
{
	struct btrfs_ioctl_received_subvol_args *sa = NULL;
	int ret = 0;

	sa = memdup_user(arg, sizeof(*sa));
5241 5242
	if (IS_ERR(sa))
		return PTR_ERR(sa);
5243 5244 5245 5246 5247 5248

	ret = _btrfs_ioctl_set_received_subvol(file, sa);

	if (ret)
		goto out;

5249 5250 5251 5252 5253 5254 5255 5256 5257
	ret = copy_to_user(arg, sa, sizeof(*sa));
	if (ret)
		ret = -EFAULT;

out:
	kfree(sa);
	return ret;
}

5258 5259
static int btrfs_ioctl_get_fslabel(struct file *file, void __user *arg)
{
5260 5261
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
5262
	size_t len;
5263
	int ret;
5264 5265
	char label[BTRFS_LABEL_SIZE];

5266 5267 5268
	spin_lock(&fs_info->super_lock);
	memcpy(label, fs_info->super_copy->label, BTRFS_LABEL_SIZE);
	spin_unlock(&fs_info->super_lock);
5269 5270

	len = strnlen(label, BTRFS_LABEL_SIZE);
5271 5272

	if (len == BTRFS_LABEL_SIZE) {
5273 5274 5275
		btrfs_warn(fs_info,
			   "label is too long, return the first %zu bytes",
			   --len);
5276 5277 5278 5279 5280 5281 5282
	}

	ret = copy_to_user(arg, label, len);

	return ret ? -EFAULT : 0;
}

5283 5284
static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg)
{
5285 5286 5287 5288
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_super_block *super_block = fs_info->super_copy;
5289 5290 5291 5292 5293 5294 5295 5296 5297 5298 5299
	struct btrfs_trans_handle *trans;
	char label[BTRFS_LABEL_SIZE];
	int ret;

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

	if (copy_from_user(label, arg, sizeof(label)))
		return -EFAULT;

	if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) {
5300
		btrfs_err(fs_info,
J
Jeff Mahoney 已提交
5301 5302
			  "unable to set label with more than %d bytes",
			  BTRFS_LABEL_SIZE - 1);
5303 5304 5305 5306 5307 5308 5309 5310 5311 5312 5313 5314 5315
		return -EINVAL;
	}

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

	trans = btrfs_start_transaction(root, 0);
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out_unlock;
	}

5316
	spin_lock(&fs_info->super_lock);
5317
	strcpy(super_block->label, label);
5318
	spin_unlock(&fs_info->super_lock);
5319
	ret = btrfs_commit_transaction(trans);
5320 5321 5322 5323 5324 5325

out_unlock:
	mnt_drop_write_file(file);
	return ret;
}

5326 5327 5328 5329 5330
#define INIT_FEATURE_FLAGS(suffix) \
	{ .compat_flags = BTRFS_FEATURE_COMPAT_##suffix, \
	  .compat_ro_flags = BTRFS_FEATURE_COMPAT_RO_##suffix, \
	  .incompat_flags = BTRFS_FEATURE_INCOMPAT_##suffix }

5331
int btrfs_ioctl_get_supported_features(void __user *arg)
5332
{
D
David Sterba 已提交
5333
	static const struct btrfs_ioctl_feature_flags features[3] = {
5334 5335 5336 5337 5338 5339 5340 5341 5342 5343 5344 5345 5346
		INIT_FEATURE_FLAGS(SUPP),
		INIT_FEATURE_FLAGS(SAFE_SET),
		INIT_FEATURE_FLAGS(SAFE_CLEAR)
	};

	if (copy_to_user(arg, &features, sizeof(features)))
		return -EFAULT;

	return 0;
}

static int btrfs_ioctl_get_features(struct file *file, void __user *arg)
{
5347 5348 5349
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_super_block *super_block = fs_info->super_copy;
5350 5351 5352 5353 5354 5355 5356 5357 5358 5359 5360 5361
	struct btrfs_ioctl_feature_flags features;

	features.compat_flags = btrfs_super_compat_flags(super_block);
	features.compat_ro_flags = btrfs_super_compat_ro_flags(super_block);
	features.incompat_flags = btrfs_super_incompat_flags(super_block);

	if (copy_to_user(arg, &features, sizeof(features)))
		return -EFAULT;

	return 0;
}

5362
static int check_feature_bits(struct btrfs_fs_info *fs_info,
5363
			      enum btrfs_feature_set set,
5364 5365 5366
			      u64 change_mask, u64 flags, u64 supported_flags,
			      u64 safe_set, u64 safe_clear)
{
5367 5368
	const char *type = btrfs_feature_set_names[set];
	char *names;
5369 5370 5371 5372 5373 5374
	u64 disallowed, unsupported;
	u64 set_mask = flags & change_mask;
	u64 clear_mask = ~flags & change_mask;

	unsupported = set_mask & ~supported_flags;
	if (unsupported) {
5375 5376
		names = btrfs_printable_features(set, unsupported);
		if (names) {
5377 5378 5379
			btrfs_warn(fs_info,
				   "this kernel does not support the %s feature bit%s",
				   names, strchr(names, ',') ? "s" : "");
5380 5381
			kfree(names);
		} else
5382 5383 5384
			btrfs_warn(fs_info,
				   "this kernel does not support %s bits 0x%llx",
				   type, unsupported);
5385 5386 5387 5388 5389
		return -EOPNOTSUPP;
	}

	disallowed = set_mask & ~safe_set;
	if (disallowed) {
5390 5391
		names = btrfs_printable_features(set, disallowed);
		if (names) {
5392 5393 5394
			btrfs_warn(fs_info,
				   "can't set the %s feature bit%s while mounted",
				   names, strchr(names, ',') ? "s" : "");
5395 5396
			kfree(names);
		} else
5397 5398 5399
			btrfs_warn(fs_info,
				   "can't set %s bits 0x%llx while mounted",
				   type, disallowed);
5400 5401 5402 5403 5404
		return -EPERM;
	}

	disallowed = clear_mask & ~safe_clear;
	if (disallowed) {
5405 5406
		names = btrfs_printable_features(set, disallowed);
		if (names) {
5407 5408 5409
			btrfs_warn(fs_info,
				   "can't clear the %s feature bit%s while mounted",
				   names, strchr(names, ',') ? "s" : "");
5410 5411
			kfree(names);
		} else
5412 5413 5414
			btrfs_warn(fs_info,
				   "can't clear %s bits 0x%llx while mounted",
				   type, disallowed);
5415 5416 5417 5418 5419 5420
		return -EPERM;
	}

	return 0;
}

5421 5422
#define check_feature(fs_info, change_mask, flags, mask_base)	\
check_feature_bits(fs_info, FEAT_##mask_base, change_mask, flags,	\
5423 5424 5425 5426 5427 5428
		   BTRFS_FEATURE_ ## mask_base ## _SUPP,	\
		   BTRFS_FEATURE_ ## mask_base ## _SAFE_SET,	\
		   BTRFS_FEATURE_ ## mask_base ## _SAFE_CLEAR)

static int btrfs_ioctl_set_features(struct file *file, void __user *arg)
{
5429 5430 5431 5432
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
	struct btrfs_super_block *super_block = fs_info->super_copy;
5433 5434 5435 5436 5437 5438 5439 5440 5441 5442 5443 5444 5445 5446 5447 5448
	struct btrfs_ioctl_feature_flags flags[2];
	struct btrfs_trans_handle *trans;
	u64 newflags;
	int ret;

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

	if (copy_from_user(flags, arg, sizeof(flags)))
		return -EFAULT;

	/* Nothing to do */
	if (!flags[0].compat_flags && !flags[0].compat_ro_flags &&
	    !flags[0].incompat_flags)
		return 0;

5449
	ret = check_feature(fs_info, flags[0].compat_flags,
5450 5451 5452 5453
			    flags[1].compat_flags, COMPAT);
	if (ret)
		return ret;

5454
	ret = check_feature(fs_info, flags[0].compat_ro_flags,
5455 5456 5457 5458
			    flags[1].compat_ro_flags, COMPAT_RO);
	if (ret)
		return ret;

5459
	ret = check_feature(fs_info, flags[0].incompat_flags,
5460 5461 5462 5463
			    flags[1].incompat_flags, INCOMPAT);
	if (ret)
		return ret;

5464 5465 5466 5467
	ret = mnt_want_write_file(file);
	if (ret)
		return ret;

5468
	trans = btrfs_start_transaction(root, 0);
5469 5470 5471 5472
	if (IS_ERR(trans)) {
		ret = PTR_ERR(trans);
		goto out_drop_write;
	}
5473

5474
	spin_lock(&fs_info->super_lock);
5475 5476 5477 5478 5479 5480 5481 5482 5483 5484 5485 5486 5487 5488
	newflags = btrfs_super_compat_flags(super_block);
	newflags |= flags[0].compat_flags & flags[1].compat_flags;
	newflags &= ~(flags[0].compat_flags & ~flags[1].compat_flags);
	btrfs_set_super_compat_flags(super_block, newflags);

	newflags = btrfs_super_compat_ro_flags(super_block);
	newflags |= flags[0].compat_ro_flags & flags[1].compat_ro_flags;
	newflags &= ~(flags[0].compat_ro_flags & ~flags[1].compat_ro_flags);
	btrfs_set_super_compat_ro_flags(super_block, newflags);

	newflags = btrfs_super_incompat_flags(super_block);
	newflags |= flags[0].incompat_flags & flags[1].incompat_flags;
	newflags &= ~(flags[0].incompat_flags & ~flags[1].incompat_flags);
	btrfs_set_super_incompat_flags(super_block, newflags);
5489
	spin_unlock(&fs_info->super_lock);
5490

5491
	ret = btrfs_commit_transaction(trans);
5492 5493 5494 5495
out_drop_write:
	mnt_drop_write_file(file);

	return ret;
5496 5497
}

5498 5499 5500 5501 5502 5503 5504 5505 5506 5507 5508 5509 5510 5511 5512 5513 5514 5515 5516 5517 5518 5519 5520 5521 5522 5523 5524 5525 5526 5527 5528 5529 5530 5531 5532
static int _btrfs_ioctl_send(struct file *file, void __user *argp, bool compat)
{
	struct btrfs_ioctl_send_args *arg;
	int ret;

	if (compat) {
#if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
		struct btrfs_ioctl_send_args_32 args32;

		ret = copy_from_user(&args32, argp, sizeof(args32));
		if (ret)
			return -EFAULT;
		arg = kzalloc(sizeof(*arg), GFP_KERNEL);
		if (!arg)
			return -ENOMEM;
		arg->send_fd = args32.send_fd;
		arg->clone_sources_count = args32.clone_sources_count;
		arg->clone_sources = compat_ptr(args32.clone_sources);
		arg->parent_root = args32.parent_root;
		arg->flags = args32.flags;
		memcpy(arg->reserved, args32.reserved,
		       sizeof(args32.reserved));
#else
		return -ENOTTY;
#endif
	} else {
		arg = memdup_user(argp, sizeof(*arg));
		if (IS_ERR(arg))
			return PTR_ERR(arg);
	}
	ret = btrfs_ioctl_send(file, arg);
	kfree(arg);
	return ret;
}

C
Christoph Hellwig 已提交
5533 5534 5535
long btrfs_ioctl(struct file *file, unsigned int
		cmd, unsigned long arg)
{
5536 5537 5538
	struct inode *inode = file_inode(file);
	struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb);
	struct btrfs_root *root = BTRFS_I(inode)->root;
5539
	void __user *argp = (void __user *)arg;
C
Christoph Hellwig 已提交
5540 5541

	switch (cmd) {
5542 5543 5544 5545 5546 5547
	case FS_IOC_GETFLAGS:
		return btrfs_ioctl_getflags(file, argp);
	case FS_IOC_SETFLAGS:
		return btrfs_ioctl_setflags(file, argp);
	case FS_IOC_GETVERSION:
		return btrfs_ioctl_getversion(file, argp);
5548 5549
	case FITRIM:
		return btrfs_ioctl_fitrim(file, argp);
C
Christoph Hellwig 已提交
5550
	case BTRFS_IOC_SNAP_CREATE:
5551
		return btrfs_ioctl_snap_create(file, argp, 0);
5552
	case BTRFS_IOC_SNAP_CREATE_V2:
5553
		return btrfs_ioctl_snap_create_v2(file, argp, 0);
5554
	case BTRFS_IOC_SUBVOL_CREATE:
5555
		return btrfs_ioctl_snap_create(file, argp, 1);
A
Arne Jansen 已提交
5556 5557
	case BTRFS_IOC_SUBVOL_CREATE_V2:
		return btrfs_ioctl_snap_create_v2(file, argp, 1);
5558 5559
	case BTRFS_IOC_SNAP_DESTROY:
		return btrfs_ioctl_snap_destroy(file, argp);
5560 5561 5562 5563
	case BTRFS_IOC_SUBVOL_GETFLAGS:
		return btrfs_ioctl_subvol_getflags(file, argp);
	case BTRFS_IOC_SUBVOL_SETFLAGS:
		return btrfs_ioctl_subvol_setflags(file, argp);
5564 5565
	case BTRFS_IOC_DEFAULT_SUBVOL:
		return btrfs_ioctl_default_subvol(file, argp);
C
Christoph Hellwig 已提交
5566
	case BTRFS_IOC_DEFRAG:
C
Chris Mason 已提交
5567 5568 5569
		return btrfs_ioctl_defrag(file, NULL);
	case BTRFS_IOC_DEFRAG_RANGE:
		return btrfs_ioctl_defrag(file, argp);
C
Christoph Hellwig 已提交
5570
	case BTRFS_IOC_RESIZE:
5571
		return btrfs_ioctl_resize(file, argp);
C
Christoph Hellwig 已提交
5572
	case BTRFS_IOC_ADD_DEV:
5573
		return btrfs_ioctl_add_dev(fs_info, argp);
C
Christoph Hellwig 已提交
5574
	case BTRFS_IOC_RM_DEV:
5575
		return btrfs_ioctl_rm_dev(file, argp);
5576 5577
	case BTRFS_IOC_RM_DEV_V2:
		return btrfs_ioctl_rm_dev_v2(file, argp);
J
Jan Schmidt 已提交
5578
	case BTRFS_IOC_FS_INFO:
5579
		return btrfs_ioctl_fs_info(fs_info, argp);
J
Jan Schmidt 已提交
5580
	case BTRFS_IOC_DEV_INFO:
5581
		return btrfs_ioctl_dev_info(fs_info, argp);
C
Christoph Hellwig 已提交
5582
	case BTRFS_IOC_BALANCE:
5583
		return btrfs_ioctl_balance(file, NULL);
C
Christoph Hellwig 已提交
5584 5585 5586 5587
	case BTRFS_IOC_TRANS_START:
		return btrfs_ioctl_trans_start(file);
	case BTRFS_IOC_TRANS_END:
		return btrfs_ioctl_trans_end(file);
5588 5589
	case BTRFS_IOC_TREE_SEARCH:
		return btrfs_ioctl_tree_search(file, argp);
G
Gerhard Heift 已提交
5590 5591
	case BTRFS_IOC_TREE_SEARCH_V2:
		return btrfs_ioctl_tree_search_v2(file, argp);
5592 5593
	case BTRFS_IOC_INO_LOOKUP:
		return btrfs_ioctl_ino_lookup(file, argp);
5594 5595 5596
	case BTRFS_IOC_INO_PATHS:
		return btrfs_ioctl_ino_to_path(root, argp);
	case BTRFS_IOC_LOGICAL_INO:
5597 5598 5599
		return btrfs_ioctl_logical_to_ino(fs_info, argp, 1);
	case BTRFS_IOC_LOGICAL_INO_V2:
		return btrfs_ioctl_logical_to_ino(fs_info, argp, 2);
J
Josef Bacik 已提交
5600
	case BTRFS_IOC_SPACE_INFO:
5601
		return btrfs_ioctl_space_info(fs_info, argp);
5602 5603 5604
	case BTRFS_IOC_SYNC: {
		int ret;

5605
		ret = btrfs_start_delalloc_roots(fs_info, 0, -1);
5606 5607
		if (ret)
			return ret;
5608
		ret = btrfs_sync_fs(inode->i_sb, 1);
5609 5610
		/*
		 * The transaction thread may want to do more work,
5611
		 * namely it pokes the cleaner kthread that will start
5612 5613
		 * processing uncleaned subvols.
		 */
5614
		wake_up_process(fs_info->transaction_kthread);
5615 5616
		return ret;
	}
5617
	case BTRFS_IOC_START_SYNC:
5618
		return btrfs_ioctl_start_sync(root, argp);
5619
	case BTRFS_IOC_WAIT_SYNC:
5620
		return btrfs_ioctl_wait_sync(fs_info, argp);
J
Jan Schmidt 已提交
5621
	case BTRFS_IOC_SCRUB:
M
Miao Xie 已提交
5622
		return btrfs_ioctl_scrub(file, argp);
J
Jan Schmidt 已提交
5623
	case BTRFS_IOC_SCRUB_CANCEL:
5624
		return btrfs_ioctl_scrub_cancel(fs_info);
J
Jan Schmidt 已提交
5625
	case BTRFS_IOC_SCRUB_PROGRESS:
5626
		return btrfs_ioctl_scrub_progress(fs_info, argp);
5627
	case BTRFS_IOC_BALANCE_V2:
5628
		return btrfs_ioctl_balance(file, argp);
5629
	case BTRFS_IOC_BALANCE_CTL:
5630
		return btrfs_ioctl_balance_ctl(fs_info, arg);
5631
	case BTRFS_IOC_BALANCE_PROGRESS:
5632
		return btrfs_ioctl_balance_progress(fs_info, argp);
5633 5634
	case BTRFS_IOC_SET_RECEIVED_SUBVOL:
		return btrfs_ioctl_set_received_subvol(file, argp);
5635 5636 5637 5638
#ifdef CONFIG_64BIT
	case BTRFS_IOC_SET_RECEIVED_SUBVOL_32:
		return btrfs_ioctl_set_received_subvol_32(file, argp);
#endif
5639
	case BTRFS_IOC_SEND:
5640 5641 5642 5643 5644
		return _btrfs_ioctl_send(file, argp, false);
#if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT)
	case BTRFS_IOC_SEND_32:
		return _btrfs_ioctl_send(file, argp, true);
#endif
5645
	case BTRFS_IOC_GET_DEV_STATS:
5646
		return btrfs_ioctl_get_dev_stats(fs_info, argp);
A
Arne Jansen 已提交
5647
	case BTRFS_IOC_QUOTA_CTL:
5648
		return btrfs_ioctl_quota_ctl(file, argp);
A
Arne Jansen 已提交
5649
	case BTRFS_IOC_QGROUP_ASSIGN:
5650
		return btrfs_ioctl_qgroup_assign(file, argp);
A
Arne Jansen 已提交
5651
	case BTRFS_IOC_QGROUP_CREATE:
5652
		return btrfs_ioctl_qgroup_create(file, argp);
A
Arne Jansen 已提交
5653
	case BTRFS_IOC_QGROUP_LIMIT:
5654
		return btrfs_ioctl_qgroup_limit(file, argp);
J
Jan Schmidt 已提交
5655 5656 5657 5658
	case BTRFS_IOC_QUOTA_RESCAN:
		return btrfs_ioctl_quota_rescan(file, argp);
	case BTRFS_IOC_QUOTA_RESCAN_STATUS:
		return btrfs_ioctl_quota_rescan_status(file, argp);
5659 5660
	case BTRFS_IOC_QUOTA_RESCAN_WAIT:
		return btrfs_ioctl_quota_rescan_wait(file, argp);
5661
	case BTRFS_IOC_DEV_REPLACE:
5662
		return btrfs_ioctl_dev_replace(fs_info, argp);
5663 5664
	case BTRFS_IOC_GET_FSLABEL:
		return btrfs_ioctl_get_fslabel(file, argp);
5665 5666
	case BTRFS_IOC_SET_FSLABEL:
		return btrfs_ioctl_set_fslabel(file, argp);
5667
	case BTRFS_IOC_GET_SUPPORTED_FEATURES:
5668
		return btrfs_ioctl_get_supported_features(argp);
5669 5670 5671 5672
	case BTRFS_IOC_GET_FEATURES:
		return btrfs_ioctl_get_features(file, argp);
	case BTRFS_IOC_SET_FEATURES:
		return btrfs_ioctl_set_features(file, argp);
C
Christoph Hellwig 已提交
5673 5674 5675 5676
	}

	return -ENOTTY;
}
5677 5678 5679 5680

#ifdef CONFIG_COMPAT
long btrfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg)
{
5681 5682 5683 5684
	/*
	 * These all access 32-bit values anyway so no further
	 * handling is necessary.
	 */
5685 5686 5687 5688 5689 5690 5691 5692 5693 5694 5695 5696 5697 5698 5699
	switch (cmd) {
	case FS_IOC32_GETFLAGS:
		cmd = FS_IOC_GETFLAGS;
		break;
	case FS_IOC32_SETFLAGS:
		cmd = FS_IOC_SETFLAGS;
		break;
	case FS_IOC32_GETVERSION:
		cmd = FS_IOC_GETVERSION;
		break;
	}

	return btrfs_ioctl(file, cmd, (unsigned long) compat_ptr(arg));
}
#endif