]>
Commit | Line | Data |
---|---|---|
c1d7c514 | 1 | // SPDX-License-Identifier: GPL-2.0 |
f46b5a66 CH |
2 | /* |
3 | * Copyright (C) 2007 Oracle. All rights reserved. | |
f46b5a66 CH |
4 | */ |
5 | ||
6 | #include <linux/kernel.h> | |
7 | #include <linux/bio.h> | |
f46b5a66 CH |
8 | #include <linux/file.h> |
9 | #include <linux/fs.h> | |
cb8e7090 | 10 | #include <linux/fsnotify.h> |
f46b5a66 CH |
11 | #include <linux/pagemap.h> |
12 | #include <linux/highmem.h> | |
13 | #include <linux/time.h> | |
f46b5a66 | 14 | #include <linux/string.h> |
f46b5a66 | 15 | #include <linux/backing-dev.h> |
cb8e7090 | 16 | #include <linux/mount.h> |
cb8e7090 | 17 | #include <linux/namei.h> |
f46b5a66 | 18 | #include <linux/writeback.h> |
f46b5a66 | 19 | #include <linux/compat.h> |
cb8e7090 | 20 | #include <linux/security.h> |
f46b5a66 | 21 | #include <linux/xattr.h> |
f54de068 | 22 | #include <linux/mm.h> |
5a0e3ad6 | 23 | #include <linux/slab.h> |
f7039b1d | 24 | #include <linux/blkdev.h> |
8ea05e3a | 25 | #include <linux/uuid.h> |
55e301fd | 26 | #include <linux/btrfs.h> |
416161db | 27 | #include <linux/uaccess.h> |
ae5e165d | 28 | #include <linux/iversion.h> |
97fc2977 | 29 | #include <linux/fileattr.h> |
f46b5a66 CH |
30 | #include "ctree.h" |
31 | #include "disk-io.h" | |
949964c9 | 32 | #include "export.h" |
f46b5a66 CH |
33 | #include "transaction.h" |
34 | #include "btrfs_inode.h" | |
f46b5a66 CH |
35 | #include "print-tree.h" |
36 | #include "volumes.h" | |
925baedd | 37 | #include "locking.h" |
d7728c96 | 38 | #include "backref.h" |
606686ee | 39 | #include "rcu-string.h" |
31db9f7c | 40 | #include "send.h" |
3f6bcfbd | 41 | #include "dev-replace.h" |
63541927 | 42 | #include "props.h" |
3b02a68a | 43 | #include "sysfs.h" |
fcebe456 | 44 | #include "qgroup.h" |
1ec9a1ae | 45 | #include "tree-log.h" |
ebb8765b | 46 | #include "compression.h" |
8719aaae | 47 | #include "space-info.h" |
86736342 | 48 | #include "delalloc-space.h" |
aac0023c | 49 | #include "block-group.h" |
f46b5a66 | 50 | |
abccd00f HM |
51 | #ifdef CONFIG_64BIT |
52 | /* If we have a 32-bit userspace and 64-bit kernel, then the UAPI | |
53 | * structures are incorrect, as the timespec structure from userspace | |
54 | * is 4 bytes too small. We define these alternatives here to teach | |
55 | * the kernel about the 32-bit struct packing. | |
56 | */ | |
57 | struct btrfs_ioctl_timespec_32 { | |
58 | __u64 sec; | |
59 | __u32 nsec; | |
60 | } __attribute__ ((__packed__)); | |
61 | ||
62 | struct btrfs_ioctl_received_subvol_args_32 { | |
63 | char uuid[BTRFS_UUID_SIZE]; /* in */ | |
64 | __u64 stransid; /* in */ | |
65 | __u64 rtransid; /* out */ | |
66 | struct btrfs_ioctl_timespec_32 stime; /* in */ | |
67 | struct btrfs_ioctl_timespec_32 rtime; /* out */ | |
68 | __u64 flags; /* in */ | |
69 | __u64 reserved[16]; /* in */ | |
70 | } __attribute__ ((__packed__)); | |
71 | ||
72 | #define BTRFS_IOC_SET_RECEIVED_SUBVOL_32 _IOWR(BTRFS_IOCTL_MAGIC, 37, \ | |
73 | struct btrfs_ioctl_received_subvol_args_32) | |
74 | #endif | |
75 | ||
2351f431 JB |
76 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) |
77 | struct btrfs_ioctl_send_args_32 { | |
78 | __s64 send_fd; /* in */ | |
79 | __u64 clone_sources_count; /* in */ | |
80 | compat_uptr_t clone_sources; /* in */ | |
81 | __u64 parent_root; /* in */ | |
82 | __u64 flags; /* in */ | |
83 | __u64 reserved[4]; /* in */ | |
84 | } __attribute__ ((__packed__)); | |
85 | ||
86 | #define BTRFS_IOC_SEND_32 _IOW(BTRFS_IOCTL_MAGIC, 38, \ | |
87 | struct btrfs_ioctl_send_args_32) | |
88 | #endif | |
abccd00f | 89 | |
6cbff00f | 90 | /* Mask out flags that are inappropriate for the given type of inode. */ |
1905a0f7 DS |
91 | static unsigned int btrfs_mask_fsflags_for_type(struct inode *inode, |
92 | unsigned int flags) | |
6cbff00f | 93 | { |
1905a0f7 | 94 | if (S_ISDIR(inode->i_mode)) |
6cbff00f | 95 | return flags; |
1905a0f7 | 96 | else if (S_ISREG(inode->i_mode)) |
6cbff00f CH |
97 | return flags & ~FS_DIRSYNC_FL; |
98 | else | |
99 | return flags & (FS_NODUMP_FL | FS_NOATIME_FL); | |
100 | } | |
101 | ||
102 | /* | |
a157d4fd DS |
103 | * Export internal inode flags to the format expected by the FS_IOC_GETFLAGS |
104 | * ioctl. | |
6cbff00f | 105 | */ |
a157d4fd | 106 | static unsigned int btrfs_inode_flags_to_fsflags(unsigned int flags) |
6cbff00f CH |
107 | { |
108 | unsigned int iflags = 0; | |
109 | ||
110 | if (flags & BTRFS_INODE_SYNC) | |
111 | iflags |= FS_SYNC_FL; | |
112 | if (flags & BTRFS_INODE_IMMUTABLE) | |
113 | iflags |= FS_IMMUTABLE_FL; | |
114 | if (flags & BTRFS_INODE_APPEND) | |
115 | iflags |= FS_APPEND_FL; | |
116 | if (flags & BTRFS_INODE_NODUMP) | |
117 | iflags |= FS_NODUMP_FL; | |
118 | if (flags & BTRFS_INODE_NOATIME) | |
119 | iflags |= FS_NOATIME_FL; | |
120 | if (flags & BTRFS_INODE_DIRSYNC) | |
121 | iflags |= FS_DIRSYNC_FL; | |
d0092bdd LZ |
122 | if (flags & BTRFS_INODE_NODATACOW) |
123 | iflags |= FS_NOCOW_FL; | |
124 | ||
13f48dc9 | 125 | if (flags & BTRFS_INODE_NOCOMPRESS) |
d0092bdd | 126 | iflags |= FS_NOCOMP_FL; |
13f48dc9 ST |
127 | else if (flags & BTRFS_INODE_COMPRESS) |
128 | iflags |= FS_COMPR_FL; | |
6cbff00f CH |
129 | |
130 | return iflags; | |
131 | } | |
132 | ||
133 | /* | |
134 | * Update inode->i_flags based on the btrfs internal flags. | |
135 | */ | |
7b6a221e | 136 | void btrfs_sync_inode_flags_to_i_flags(struct inode *inode) |
6cbff00f | 137 | { |
5c57b8b6 | 138 | struct btrfs_inode *binode = BTRFS_I(inode); |
3cc79392 | 139 | unsigned int new_fl = 0; |
6cbff00f | 140 | |
5c57b8b6 | 141 | if (binode->flags & BTRFS_INODE_SYNC) |
3cc79392 | 142 | new_fl |= S_SYNC; |
5c57b8b6 | 143 | if (binode->flags & BTRFS_INODE_IMMUTABLE) |
3cc79392 | 144 | new_fl |= S_IMMUTABLE; |
5c57b8b6 | 145 | if (binode->flags & BTRFS_INODE_APPEND) |
3cc79392 | 146 | new_fl |= S_APPEND; |
5c57b8b6 | 147 | if (binode->flags & BTRFS_INODE_NOATIME) |
3cc79392 | 148 | new_fl |= S_NOATIME; |
5c57b8b6 | 149 | if (binode->flags & BTRFS_INODE_DIRSYNC) |
3cc79392 FM |
150 | new_fl |= S_DIRSYNC; |
151 | ||
152 | set_mask_bits(&inode->i_flags, | |
153 | S_SYNC | S_APPEND | S_IMMUTABLE | S_NOATIME | S_DIRSYNC, | |
154 | new_fl); | |
6cbff00f CH |
155 | } |
156 | ||
f37c563b DS |
157 | /* |
158 | * Check if @flags are a supported and valid set of FS_*_FL flags and that | |
159 | * the old and new flags are not conflicting | |
160 | */ | |
161 | static int check_fsflags(unsigned int old_flags, unsigned int flags) | |
75e7cb7f LB |
162 | { |
163 | if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \ | |
164 | FS_NOATIME_FL | FS_NODUMP_FL | \ | |
165 | FS_SYNC_FL | FS_DIRSYNC_FL | \ | |
e1e8fb6a LZ |
166 | FS_NOCOMP_FL | FS_COMPR_FL | |
167 | FS_NOCOW_FL)) | |
75e7cb7f LB |
168 | return -EOPNOTSUPP; |
169 | ||
f37c563b | 170 | /* COMPR and NOCOMP on new/old are valid */ |
75e7cb7f LB |
171 | if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL)) |
172 | return -EINVAL; | |
173 | ||
f37c563b DS |
174 | if ((flags & FS_COMPR_FL) && (flags & FS_NOCOW_FL)) |
175 | return -EINVAL; | |
176 | ||
177 | /* NOCOW and compression options are mutually exclusive */ | |
178 | if ((old_flags & FS_NOCOW_FL) && (flags & (FS_COMPR_FL | FS_NOCOMP_FL))) | |
179 | return -EINVAL; | |
180 | if ((flags & FS_NOCOW_FL) && (old_flags & (FS_COMPR_FL | FS_NOCOMP_FL))) | |
181 | return -EINVAL; | |
182 | ||
75e7cb7f LB |
183 | return 0; |
184 | } | |
185 | ||
d206e9c9 NA |
186 | static int check_fsflags_compatible(struct btrfs_fs_info *fs_info, |
187 | unsigned int flags) | |
188 | { | |
189 | if (btrfs_is_zoned(fs_info) && (flags & FS_NOCOW_FL)) | |
190 | return -EPERM; | |
191 | ||
192 | return 0; | |
193 | } | |
194 | ||
97fc2977 MS |
195 | /* |
196 | * Set flags/xflags from the internal inode flags. The remaining items of | |
197 | * fsxattr are zeroed. | |
198 | */ | |
199 | int btrfs_fileattr_get(struct dentry *dentry, struct fileattr *fa) | |
6cbff00f | 200 | { |
97fc2977 MS |
201 | struct btrfs_inode *binode = BTRFS_I(d_inode(dentry)); |
202 | ||
203 | fileattr_fill_flags(fa, btrfs_inode_flags_to_fsflags(binode->flags)); | |
204 | return 0; | |
205 | } | |
206 | ||
207 | int btrfs_fileattr_set(struct user_namespace *mnt_userns, | |
208 | struct dentry *dentry, struct fileattr *fa) | |
209 | { | |
210 | struct inode *inode = d_inode(dentry); | |
0b246afa | 211 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
5c57b8b6 DS |
212 | struct btrfs_inode *binode = BTRFS_I(inode); |
213 | struct btrfs_root *root = binode->root; | |
6cbff00f | 214 | struct btrfs_trans_handle *trans; |
5aca2842 | 215 | unsigned int fsflags, old_fsflags; |
6cbff00f | 216 | int ret; |
ff9fef55 | 217 | const char *comp = NULL; |
f37c563b | 218 | u32 binode_flags; |
6cbff00f | 219 | |
b83cc969 LZ |
220 | if (btrfs_root_readonly(root)) |
221 | return -EROFS; | |
222 | ||
97fc2977 MS |
223 | if (fileattr_has_fsx(fa)) |
224 | return -EOPNOTSUPP; | |
e7848683 | 225 | |
97fc2977 | 226 | fsflags = btrfs_mask_fsflags_for_type(inode, fa->flags); |
5aca2842 | 227 | old_fsflags = btrfs_inode_flags_to_fsflags(binode->flags); |
f37c563b DS |
228 | ret = check_fsflags(old_fsflags, fsflags); |
229 | if (ret) | |
97fc2977 | 230 | return ret; |
f37c563b | 231 | |
d206e9c9 NA |
232 | ret = check_fsflags_compatible(fs_info, fsflags); |
233 | if (ret) | |
97fc2977 | 234 | return ret; |
d206e9c9 | 235 | |
f37c563b | 236 | binode_flags = binode->flags; |
5c57b8b6 | 237 | if (fsflags & FS_SYNC_FL) |
d2b8fcfe | 238 | binode_flags |= BTRFS_INODE_SYNC; |
6cbff00f | 239 | else |
d2b8fcfe | 240 | binode_flags &= ~BTRFS_INODE_SYNC; |
5c57b8b6 | 241 | if (fsflags & FS_IMMUTABLE_FL) |
d2b8fcfe | 242 | binode_flags |= BTRFS_INODE_IMMUTABLE; |
6cbff00f | 243 | else |
d2b8fcfe | 244 | binode_flags &= ~BTRFS_INODE_IMMUTABLE; |
5c57b8b6 | 245 | if (fsflags & FS_APPEND_FL) |
d2b8fcfe | 246 | binode_flags |= BTRFS_INODE_APPEND; |
6cbff00f | 247 | else |
d2b8fcfe | 248 | binode_flags &= ~BTRFS_INODE_APPEND; |
5c57b8b6 | 249 | if (fsflags & FS_NODUMP_FL) |
d2b8fcfe | 250 | binode_flags |= BTRFS_INODE_NODUMP; |
6cbff00f | 251 | else |
d2b8fcfe | 252 | binode_flags &= ~BTRFS_INODE_NODUMP; |
5c57b8b6 | 253 | if (fsflags & FS_NOATIME_FL) |
d2b8fcfe | 254 | binode_flags |= BTRFS_INODE_NOATIME; |
6cbff00f | 255 | else |
d2b8fcfe | 256 | binode_flags &= ~BTRFS_INODE_NOATIME; |
97fc2977 MS |
257 | |
258 | /* If coming from FS_IOC_FSSETXATTR then skip unconverted flags */ | |
259 | if (!fa->flags_valid) { | |
260 | /* 1 item for the inode */ | |
261 | trans = btrfs_start_transaction(root, 1); | |
9b8a233b RH |
262 | if (IS_ERR(trans)) |
263 | return PTR_ERR(trans); | |
97fc2977 MS |
264 | goto update_flags; |
265 | } | |
266 | ||
5c57b8b6 | 267 | if (fsflags & FS_DIRSYNC_FL) |
d2b8fcfe | 268 | binode_flags |= BTRFS_INODE_DIRSYNC; |
6cbff00f | 269 | else |
d2b8fcfe | 270 | binode_flags &= ~BTRFS_INODE_DIRSYNC; |
5c57b8b6 | 271 | if (fsflags & FS_NOCOW_FL) { |
44e5194b | 272 | if (S_ISREG(inode->i_mode)) { |
7e97b8da DS |
273 | /* |
274 | * It's safe to turn csums off here, no extents exist. | |
275 | * Otherwise we want the flag to reflect the real COW | |
276 | * status of the file and will not set it. | |
277 | */ | |
278 | if (inode->i_size == 0) | |
d2b8fcfe AJ |
279 | binode_flags |= BTRFS_INODE_NODATACOW | |
280 | BTRFS_INODE_NODATASUM; | |
7e97b8da | 281 | } else { |
d2b8fcfe | 282 | binode_flags |= BTRFS_INODE_NODATACOW; |
7e97b8da DS |
283 | } |
284 | } else { | |
285 | /* | |
01327610 | 286 | * Revert back under same assumptions as above |
7e97b8da | 287 | */ |
44e5194b | 288 | if (S_ISREG(inode->i_mode)) { |
7e97b8da | 289 | if (inode->i_size == 0) |
d2b8fcfe AJ |
290 | binode_flags &= ~(BTRFS_INODE_NODATACOW | |
291 | BTRFS_INODE_NODATASUM); | |
7e97b8da | 292 | } else { |
d2b8fcfe | 293 | binode_flags &= ~BTRFS_INODE_NODATACOW; |
7e97b8da DS |
294 | } |
295 | } | |
6cbff00f | 296 | |
75e7cb7f LB |
297 | /* |
298 | * The COMPRESS flag can only be changed by users, while the NOCOMPRESS | |
299 | * flag may be changed automatically if compression code won't make | |
300 | * things smaller. | |
301 | */ | |
5c57b8b6 | 302 | if (fsflags & FS_NOCOMP_FL) { |
d2b8fcfe AJ |
303 | binode_flags &= ~BTRFS_INODE_COMPRESS; |
304 | binode_flags |= BTRFS_INODE_NOCOMPRESS; | |
5c57b8b6 | 305 | } else if (fsflags & FS_COMPR_FL) { |
63541927 | 306 | |
97fc2977 MS |
307 | if (IS_SWAPFILE(inode)) |
308 | return -ETXTBSY; | |
eede2bf3 | 309 | |
d2b8fcfe AJ |
310 | binode_flags |= BTRFS_INODE_COMPRESS; |
311 | binode_flags &= ~BTRFS_INODE_NOCOMPRESS; | |
63541927 | 312 | |
93370509 DS |
313 | comp = btrfs_compress_type2str(fs_info->compress_type); |
314 | if (!comp || comp[0] == 0) | |
315 | comp = btrfs_compress_type2str(BTRFS_COMPRESS_ZLIB); | |
ebcb904d | 316 | } else { |
d2b8fcfe | 317 | binode_flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS); |
75e7cb7f | 318 | } |
6cbff00f | 319 | |
ff9fef55 AJ |
320 | /* |
321 | * 1 for inode item | |
322 | * 2 for properties | |
323 | */ | |
324 | trans = btrfs_start_transaction(root, 3); | |
97fc2977 MS |
325 | if (IS_ERR(trans)) |
326 | return PTR_ERR(trans); | |
6cbff00f | 327 | |
ff9fef55 AJ |
328 | if (comp) { |
329 | ret = btrfs_set_prop(trans, inode, "btrfs.compression", comp, | |
330 | strlen(comp), 0); | |
331 | if (ret) { | |
332 | btrfs_abort_transaction(trans, ret); | |
333 | goto out_end_trans; | |
334 | } | |
ff9fef55 AJ |
335 | } else { |
336 | ret = btrfs_set_prop(trans, inode, "btrfs.compression", NULL, | |
337 | 0, 0); | |
338 | if (ret && ret != -ENODATA) { | |
339 | btrfs_abort_transaction(trans, ret); | |
340 | goto out_end_trans; | |
341 | } | |
342 | } | |
343 | ||
97fc2977 | 344 | update_flags: |
d2b8fcfe | 345 | binode->flags = binode_flags; |
7b6a221e | 346 | btrfs_sync_inode_flags_to_i_flags(inode); |
0c4d2d95 | 347 | inode_inc_iversion(inode); |
c2050a45 | 348 | inode->i_ctime = current_time(inode); |
9a56fcd1 | 349 | ret = btrfs_update_inode(trans, root, BTRFS_I(inode)); |
6cbff00f | 350 | |
ff9fef55 | 351 | out_end_trans: |
3a45bb20 | 352 | btrfs_end_transaction(trans); |
2d4e6f6a | 353 | return ret; |
6cbff00f CH |
354 | } |
355 | ||
c3e1f96c GR |
356 | bool btrfs_exclop_start(struct btrfs_fs_info *fs_info, |
357 | enum btrfs_exclusive_operation type) | |
358 | { | |
359 | return !cmpxchg(&fs_info->exclusive_operation, BTRFS_EXCLOP_NONE, type); | |
360 | } | |
361 | ||
362 | void btrfs_exclop_finish(struct btrfs_fs_info *fs_info) | |
363 | { | |
364 | WRITE_ONCE(fs_info->exclusive_operation, BTRFS_EXCLOP_NONE); | |
66a2823c | 365 | sysfs_notify(&fs_info->fs_devices->fsid_kobj, NULL, "exclusive_operation"); |
c3e1f96c GR |
366 | } |
367 | ||
6cbff00f CH |
368 | static int btrfs_ioctl_getversion(struct file *file, int __user *arg) |
369 | { | |
496ad9aa | 370 | struct inode *inode = file_inode(file); |
6cbff00f CH |
371 | |
372 | return put_user(inode->i_generation, arg); | |
373 | } | |
f46b5a66 | 374 | |
b929c1d8 MPS |
375 | static noinline int btrfs_ioctl_fitrim(struct btrfs_fs_info *fs_info, |
376 | void __user *arg) | |
f7039b1d | 377 | { |
f7039b1d LD |
378 | struct btrfs_device *device; |
379 | struct request_queue *q; | |
380 | struct fstrim_range range; | |
381 | u64 minlen = ULLONG_MAX; | |
382 | u64 num_devices = 0; | |
383 | int ret; | |
384 | ||
385 | if (!capable(CAP_SYS_ADMIN)) | |
386 | return -EPERM; | |
387 | ||
1cb3dc3f NA |
388 | /* |
389 | * btrfs_trim_block_group() depends on space cache, which is not | |
390 | * available in zoned filesystem. So, disallow fitrim on a zoned | |
391 | * filesystem for now. | |
392 | */ | |
393 | if (btrfs_is_zoned(fs_info)) | |
394 | return -EOPNOTSUPP; | |
395 | ||
f35f06c3 FM |
396 | /* |
397 | * If the fs is mounted with nologreplay, which requires it to be | |
398 | * mounted in RO mode as well, we can not allow discard on free space | |
399 | * inside block groups, because log trees refer to extents that are not | |
400 | * pinned in a block group's free space cache (pinning the extents is | |
401 | * precisely the first phase of replaying a log tree). | |
402 | */ | |
403 | if (btrfs_test_opt(fs_info, NOLOGREPLAY)) | |
404 | return -EROFS; | |
405 | ||
1f78160c XG |
406 | rcu_read_lock(); |
407 | list_for_each_entry_rcu(device, &fs_info->fs_devices->devices, | |
408 | dev_list) { | |
f7039b1d LD |
409 | if (!device->bdev) |
410 | continue; | |
411 | q = bdev_get_queue(device->bdev); | |
412 | if (blk_queue_discard(q)) { | |
413 | num_devices++; | |
50d0446e | 414 | minlen = min_t(u64, q->limits.discard_granularity, |
f7039b1d LD |
415 | minlen); |
416 | } | |
417 | } | |
1f78160c | 418 | rcu_read_unlock(); |
f4c697e6 | 419 | |
f7039b1d LD |
420 | if (!num_devices) |
421 | return -EOPNOTSUPP; | |
f7039b1d LD |
422 | if (copy_from_user(&range, arg, sizeof(range))) |
423 | return -EFAULT; | |
6ba9fc8e QW |
424 | |
425 | /* | |
426 | * NOTE: Don't truncate the range using super->total_bytes. Bytenr of | |
427 | * block group is in the logical address space, which can be any | |
428 | * sectorsize aligned bytenr in the range [0, U64_MAX]. | |
429 | */ | |
430 | if (range.len < fs_info->sb->s_blocksize) | |
f4c697e6 | 431 | return -EINVAL; |
f7039b1d LD |
432 | |
433 | range.minlen = max(range.minlen, minlen); | |
2ff7e61e | 434 | ret = btrfs_trim_fs(fs_info, &range); |
f7039b1d LD |
435 | if (ret < 0) |
436 | return ret; | |
437 | ||
438 | if (copy_to_user(arg, &range, sizeof(range))) | |
439 | return -EFAULT; | |
440 | ||
441 | return 0; | |
442 | } | |
443 | ||
e1f60a65 | 444 | int __pure btrfs_is_empty_uuid(u8 *uuid) |
dd5f9615 | 445 | { |
46e0f66a CM |
446 | int i; |
447 | ||
448 | for (i = 0; i < BTRFS_UUID_SIZE; i++) { | |
449 | if (uuid[i]) | |
450 | return 0; | |
451 | } | |
452 | return 1; | |
dd5f9615 SB |
453 | } |
454 | ||
d5c12070 | 455 | static noinline int create_subvol(struct inode *dir, |
cb8e7090 | 456 | struct dentry *dentry, |
52f75f4f | 457 | const char *name, int namelen, |
8696c533 | 458 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 459 | { |
0b246afa | 460 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); |
f46b5a66 CH |
461 | struct btrfs_trans_handle *trans; |
462 | struct btrfs_key key; | |
49a3c4d9 | 463 | struct btrfs_root_item *root_item; |
f46b5a66 CH |
464 | struct btrfs_inode_item *inode_item; |
465 | struct extent_buffer *leaf; | |
d5c12070 | 466 | struct btrfs_root *root = BTRFS_I(dir)->root; |
76dda93c | 467 | struct btrfs_root *new_root; |
d5c12070 | 468 | struct btrfs_block_rsv block_rsv; |
95582b00 | 469 | struct timespec64 cur_time = current_time(dir); |
5662344b | 470 | struct inode *inode; |
f46b5a66 CH |
471 | int ret; |
472 | int err; | |
2dfb1e43 | 473 | dev_t anon_dev = 0; |
f46b5a66 | 474 | u64 objectid; |
3de4586c | 475 | u64 index = 0; |
f46b5a66 | 476 | |
49a3c4d9 DS |
477 | root_item = kzalloc(sizeof(*root_item), GFP_KERNEL); |
478 | if (!root_item) | |
479 | return -ENOMEM; | |
480 | ||
543068a2 | 481 | ret = btrfs_get_free_objectid(fs_info->tree_root, &objectid); |
2fbe8c8a | 482 | if (ret) |
49a3c4d9 | 483 | goto fail_free; |
6a912213 | 484 | |
2dfb1e43 QW |
485 | ret = get_anon_bdev(&anon_dev); |
486 | if (ret < 0) | |
487 | goto fail_free; | |
488 | ||
e09fe2d2 QW |
489 | /* |
490 | * Don't create subvolume whose level is not zero. Or qgroup will be | |
01327610 | 491 | * screwed up since it assumes subvolume qgroup's level to be 0. |
e09fe2d2 | 492 | */ |
49a3c4d9 DS |
493 | if (btrfs_qgroup_level(objectid)) { |
494 | ret = -ENOSPC; | |
495 | goto fail_free; | |
496 | } | |
e09fe2d2 | 497 | |
d5c12070 | 498 | btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP); |
9ed74f2d | 499 | /* |
d5c12070 MX |
500 | * The same as the snapshot creation, please see the comment |
501 | * of create_snapshot(). | |
9ed74f2d | 502 | */ |
c4c129db | 503 | ret = btrfs_subvolume_reserve_metadata(root, &block_rsv, 8, false); |
d5c12070 | 504 | if (ret) |
49a3c4d9 | 505 | goto fail_free; |
d5c12070 MX |
506 | |
507 | trans = btrfs_start_transaction(root, 0); | |
508 | if (IS_ERR(trans)) { | |
509 | ret = PTR_ERR(trans); | |
e85fde51 | 510 | btrfs_subvolume_release_metadata(root, &block_rsv); |
49a3c4d9 | 511 | goto fail_free; |
d5c12070 MX |
512 | } |
513 | trans->block_rsv = &block_rsv; | |
514 | trans->bytes_reserved = block_rsv.size; | |
f46b5a66 | 515 | |
a9377422 | 516 | ret = btrfs_qgroup_inherit(trans, 0, objectid, inherit); |
6f72c7e2 AJ |
517 | if (ret) |
518 | goto fail; | |
519 | ||
9631e4cc JB |
520 | leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0, |
521 | BTRFS_NESTING_NORMAL); | |
8e8a1e31 JB |
522 | if (IS_ERR(leaf)) { |
523 | ret = PTR_ERR(leaf); | |
524 | goto fail; | |
525 | } | |
f46b5a66 | 526 | |
f46b5a66 CH |
527 | btrfs_mark_buffer_dirty(leaf); |
528 | ||
49a3c4d9 | 529 | inode_item = &root_item->inode; |
3cae210f QW |
530 | btrfs_set_stack_inode_generation(inode_item, 1); |
531 | btrfs_set_stack_inode_size(inode_item, 3); | |
532 | btrfs_set_stack_inode_nlink(inode_item, 1); | |
da17066c | 533 | btrfs_set_stack_inode_nbytes(inode_item, |
0b246afa | 534 | fs_info->nodesize); |
3cae210f | 535 | btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755); |
f46b5a66 | 536 | |
49a3c4d9 DS |
537 | btrfs_set_root_flags(root_item, 0); |
538 | btrfs_set_root_limit(root_item, 0); | |
3cae210f | 539 | btrfs_set_stack_inode_flags(inode_item, BTRFS_INODE_ROOT_ITEM_INIT); |
08fe4db1 | 540 | |
49a3c4d9 DS |
541 | btrfs_set_root_bytenr(root_item, leaf->start); |
542 | btrfs_set_root_generation(root_item, trans->transid); | |
543 | btrfs_set_root_level(root_item, 0); | |
544 | btrfs_set_root_refs(root_item, 1); | |
545 | btrfs_set_root_used(root_item, leaf->len); | |
546 | btrfs_set_root_last_snapshot(root_item, 0); | |
f46b5a66 | 547 | |
49a3c4d9 DS |
548 | btrfs_set_root_generation_v2(root_item, |
549 | btrfs_root_generation(root_item)); | |
807fc790 | 550 | generate_random_guid(root_item->uuid); |
49a3c4d9 DS |
551 | btrfs_set_stack_timespec_sec(&root_item->otime, cur_time.tv_sec); |
552 | btrfs_set_stack_timespec_nsec(&root_item->otime, cur_time.tv_nsec); | |
553 | root_item->ctime = root_item->otime; | |
554 | btrfs_set_root_ctransid(root_item, trans->transid); | |
555 | btrfs_set_root_otransid(root_item, trans->transid); | |
f46b5a66 | 556 | |
925baedd | 557 | btrfs_tree_unlock(leaf); |
f46b5a66 | 558 | |
69948022 | 559 | btrfs_set_root_dirid(root_item, BTRFS_FIRST_FREE_OBJECTID); |
f46b5a66 CH |
560 | |
561 | key.objectid = objectid; | |
5d4f98a2 | 562 | key.offset = 0; |
962a298f | 563 | key.type = BTRFS_ROOT_ITEM_KEY; |
0b246afa | 564 | ret = btrfs_insert_root(trans, fs_info->tree_root, &key, |
49a3c4d9 | 565 | root_item); |
67addf29 FM |
566 | if (ret) { |
567 | /* | |
568 | * Since we don't abort the transaction in this case, free the | |
569 | * tree block so that we don't leak space and leave the | |
570 | * filesystem in an inconsistent state (an extent item in the | |
571 | * extent tree without backreferences). Also no need to have | |
572 | * the tree block locked since it is not in any tree at this | |
573 | * point, so no other task can find it and use it. | |
574 | */ | |
575 | btrfs_free_tree_block(trans, root, leaf, 0, 1); | |
576 | free_extent_buffer(leaf); | |
f46b5a66 | 577 | goto fail; |
67addf29 FM |
578 | } |
579 | ||
580 | free_extent_buffer(leaf); | |
581 | leaf = NULL; | |
f46b5a66 | 582 | |
76dda93c | 583 | key.offset = (u64)-1; |
2dfb1e43 | 584 | new_root = btrfs_get_new_fs_root(fs_info, objectid, anon_dev); |
79787eaa | 585 | if (IS_ERR(new_root)) { |
2dfb1e43 | 586 | free_anon_bdev(anon_dev); |
79787eaa | 587 | ret = PTR_ERR(new_root); |
66642832 | 588 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
589 | goto fail; |
590 | } | |
2dfb1e43 QW |
591 | /* Freeing will be done in btrfs_put_root() of new_root */ |
592 | anon_dev = 0; | |
76dda93c | 593 | |
221581e4 JB |
594 | ret = btrfs_record_root_in_trans(trans, new_root); |
595 | if (ret) { | |
596 | btrfs_put_root(new_root); | |
597 | btrfs_abort_transaction(trans, ret); | |
598 | goto fail; | |
599 | } | |
76dda93c | 600 | |
69948022 | 601 | ret = btrfs_create_subvol_root(trans, new_root, root); |
00246528 | 602 | btrfs_put_root(new_root); |
ce598979 MF |
603 | if (ret) { |
604 | /* We potentially lose an unused inode item here */ | |
66642832 | 605 | btrfs_abort_transaction(trans, ret); |
ce598979 MF |
606 | goto fail; |
607 | } | |
608 | ||
f46b5a66 CH |
609 | /* |
610 | * insert the directory item | |
611 | */ | |
877574e2 | 612 | ret = btrfs_set_inode_index(BTRFS_I(dir), &index); |
79787eaa | 613 | if (ret) { |
66642832 | 614 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
615 | goto fail; |
616 | } | |
3de4586c | 617 | |
684572df | 618 | ret = btrfs_insert_dir_item(trans, name, namelen, BTRFS_I(dir), &key, |
3de4586c | 619 | BTRFS_FT_DIR, index); |
79787eaa | 620 | if (ret) { |
66642832 | 621 | btrfs_abort_transaction(trans, ret); |
f46b5a66 | 622 | goto fail; |
79787eaa | 623 | } |
0660b5af | 624 | |
6ef06d27 | 625 | btrfs_i_size_write(BTRFS_I(dir), dir->i_size + namelen * 2); |
9a56fcd1 | 626 | ret = btrfs_update_inode(trans, root, BTRFS_I(dir)); |
c7e54b51 JB |
627 | if (ret) { |
628 | btrfs_abort_transaction(trans, ret); | |
629 | goto fail; | |
630 | } | |
52c26179 | 631 | |
6025c19f | 632 | ret = btrfs_add_root_ref(trans, objectid, root->root_key.objectid, |
4a0cc7ca | 633 | btrfs_ino(BTRFS_I(dir)), index, name, namelen); |
c7e54b51 JB |
634 | if (ret) { |
635 | btrfs_abort_transaction(trans, ret); | |
636 | goto fail; | |
637 | } | |
f46b5a66 | 638 | |
cdb345a8 | 639 | ret = btrfs_uuid_tree_add(trans, root_item->uuid, |
6bccf3ab | 640 | BTRFS_UUID_KEY_SUBVOL, objectid); |
dd5f9615 | 641 | if (ret) |
66642832 | 642 | btrfs_abort_transaction(trans, ret); |
dd5f9615 | 643 | |
f46b5a66 | 644 | fail: |
49a3c4d9 | 645 | kfree(root_item); |
d5c12070 MX |
646 | trans->block_rsv = NULL; |
647 | trans->bytes_reserved = 0; | |
e85fde51 | 648 | btrfs_subvolume_release_metadata(root, &block_rsv); |
de6e8200 | 649 | |
9babda9f | 650 | err = btrfs_commit_transaction(trans); |
f46b5a66 CH |
651 | if (err && !ret) |
652 | ret = err; | |
1a65e24b | 653 | |
5662344b TI |
654 | if (!ret) { |
655 | inode = btrfs_lookup_dentry(dir, dentry); | |
de6e8200 LB |
656 | if (IS_ERR(inode)) |
657 | return PTR_ERR(inode); | |
5662344b TI |
658 | d_instantiate(dentry, inode); |
659 | } | |
f46b5a66 | 660 | return ret; |
49a3c4d9 DS |
661 | |
662 | fail_free: | |
2dfb1e43 QW |
663 | if (anon_dev) |
664 | free_anon_bdev(anon_dev); | |
49a3c4d9 DS |
665 | kfree(root_item); |
666 | return ret; | |
f46b5a66 CH |
667 | } |
668 | ||
e9662f70 | 669 | static int create_snapshot(struct btrfs_root *root, struct inode *dir, |
9babda9f | 670 | struct dentry *dentry, bool readonly, |
e9662f70 | 671 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 672 | { |
0b246afa | 673 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); |
2e4bfab9 | 674 | struct inode *inode; |
f46b5a66 CH |
675 | struct btrfs_pending_snapshot *pending_snapshot; |
676 | struct btrfs_trans_handle *trans; | |
2e4bfab9 | 677 | int ret; |
f46b5a66 | 678 | |
92a7cc42 | 679 | if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) |
f46b5a66 CH |
680 | return -EINVAL; |
681 | ||
eede2bf3 OS |
682 | if (atomic_read(&root->nr_swapfiles)) { |
683 | btrfs_warn(fs_info, | |
684 | "cannot snapshot subvolume with active swapfile"); | |
685 | return -ETXTBSY; | |
686 | } | |
687 | ||
23269bf5 | 688 | pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_KERNEL); |
a1ee7362 DS |
689 | if (!pending_snapshot) |
690 | return -ENOMEM; | |
691 | ||
2dfb1e43 QW |
692 | ret = get_anon_bdev(&pending_snapshot->anon_dev); |
693 | if (ret < 0) | |
694 | goto free_pending; | |
b0c0ea63 | 695 | pending_snapshot->root_item = kzalloc(sizeof(struct btrfs_root_item), |
23269bf5 | 696 | GFP_KERNEL); |
8546b570 DS |
697 | pending_snapshot->path = btrfs_alloc_path(); |
698 | if (!pending_snapshot->root_item || !pending_snapshot->path) { | |
b0c0ea63 DS |
699 | ret = -ENOMEM; |
700 | goto free_pending; | |
701 | } | |
702 | ||
66d8f3dd MX |
703 | btrfs_init_block_rsv(&pending_snapshot->block_rsv, |
704 | BTRFS_BLOCK_RSV_TEMP); | |
d5c12070 MX |
705 | /* |
706 | * 1 - parent dir inode | |
707 | * 2 - dir entries | |
708 | * 1 - root item | |
709 | * 2 - root ref/backref | |
710 | * 1 - root of snapshot | |
dd5f9615 | 711 | * 1 - UUID item |
d5c12070 MX |
712 | */ |
713 | ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root, | |
dd5f9615 | 714 | &pending_snapshot->block_rsv, 8, |
ee3441b4 | 715 | false); |
d5c12070 | 716 | if (ret) |
c11fbb6e | 717 | goto free_pending; |
d5c12070 | 718 | |
3de4586c | 719 | pending_snapshot->dentry = dentry; |
f46b5a66 | 720 | pending_snapshot->root = root; |
b83cc969 | 721 | pending_snapshot->readonly = readonly; |
e9662f70 | 722 | pending_snapshot->dir = dir; |
8696c533 | 723 | pending_snapshot->inherit = inherit; |
a22285a6 | 724 | |
d5c12070 | 725 | trans = btrfs_start_transaction(root, 0); |
a22285a6 YZ |
726 | if (IS_ERR(trans)) { |
727 | ret = PTR_ERR(trans); | |
728 | goto fail; | |
729 | } | |
730 | ||
0b246afa | 731 | spin_lock(&fs_info->trans_lock); |
f46b5a66 CH |
732 | list_add(&pending_snapshot->list, |
733 | &trans->transaction->pending_snapshots); | |
0b246afa | 734 | spin_unlock(&fs_info->trans_lock); |
9babda9f NB |
735 | |
736 | ret = btrfs_commit_transaction(trans); | |
aec8030a | 737 | if (ret) |
c37b2b62 | 738 | goto fail; |
a22285a6 YZ |
739 | |
740 | ret = pending_snapshot->error; | |
741 | if (ret) | |
742 | goto fail; | |
743 | ||
d3797308 CM |
744 | ret = btrfs_orphan_cleanup(pending_snapshot->snap); |
745 | if (ret) | |
746 | goto fail; | |
747 | ||
2b0143b5 | 748 | inode = btrfs_lookup_dentry(d_inode(dentry->d_parent), dentry); |
2e4bfab9 YZ |
749 | if (IS_ERR(inode)) { |
750 | ret = PTR_ERR(inode); | |
751 | goto fail; | |
752 | } | |
5662344b | 753 | |
2e4bfab9 YZ |
754 | d_instantiate(dentry, inode); |
755 | ret = 0; | |
2dfb1e43 | 756 | pending_snapshot->anon_dev = 0; |
2e4bfab9 | 757 | fail: |
2dfb1e43 QW |
758 | /* Prevent double freeing of anon_dev */ |
759 | if (ret && pending_snapshot->snap) | |
760 | pending_snapshot->snap->anon_dev = 0; | |
00246528 | 761 | btrfs_put_root(pending_snapshot->snap); |
e85fde51 | 762 | btrfs_subvolume_release_metadata(root, &pending_snapshot->block_rsv); |
b0c0ea63 | 763 | free_pending: |
2dfb1e43 QW |
764 | if (pending_snapshot->anon_dev) |
765 | free_anon_bdev(pending_snapshot->anon_dev); | |
b0c0ea63 | 766 | kfree(pending_snapshot->root_item); |
8546b570 | 767 | btrfs_free_path(pending_snapshot->path); |
a1ee7362 DS |
768 | kfree(pending_snapshot); |
769 | ||
f46b5a66 CH |
770 | return ret; |
771 | } | |
772 | ||
4260f7c7 SW |
773 | /* copy of may_delete in fs/namei.c() |
774 | * Check whether we can remove a link victim from directory dir, check | |
775 | * whether the type of victim is right. | |
776 | * 1. We can't do it if dir is read-only (done in permission()) | |
777 | * 2. We should have write and exec permissions on dir | |
778 | * 3. We can't remove anything from append-only dir | |
779 | * 4. We can't do anything with immutable dir (done in permission()) | |
780 | * 5. If the sticky bit on dir is set we should either | |
781 | * a. be owner of dir, or | |
782 | * b. be owner of victim, or | |
783 | * c. have CAP_FOWNER capability | |
01327610 | 784 | * 6. If the victim is append-only or immutable we can't do anything with |
4260f7c7 SW |
785 | * links pointing to it. |
786 | * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR. | |
787 | * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR. | |
788 | * 9. We can't remove a root or mountpoint. | |
789 | * 10. We don't allow removal of NFS sillyrenamed files; it's handled by | |
790 | * nfs_async_unlink(). | |
791 | */ | |
792 | ||
67871254 | 793 | static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir) |
4260f7c7 SW |
794 | { |
795 | int error; | |
796 | ||
2b0143b5 | 797 | if (d_really_is_negative(victim)) |
4260f7c7 SW |
798 | return -ENOENT; |
799 | ||
2b0143b5 | 800 | BUG_ON(d_inode(victim->d_parent) != dir); |
4fa6b5ec | 801 | audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE); |
4260f7c7 | 802 | |
47291baa | 803 | error = inode_permission(&init_user_ns, dir, MAY_WRITE | MAY_EXEC); |
4260f7c7 SW |
804 | if (error) |
805 | return error; | |
806 | if (IS_APPEND(dir)) | |
807 | return -EPERM; | |
ba73d987 CB |
808 | if (check_sticky(&init_user_ns, dir, d_inode(victim)) || |
809 | IS_APPEND(d_inode(victim)) || IS_IMMUTABLE(d_inode(victim)) || | |
810 | IS_SWAPFILE(d_inode(victim))) | |
4260f7c7 SW |
811 | return -EPERM; |
812 | if (isdir) { | |
e36cb0b8 | 813 | if (!d_is_dir(victim)) |
4260f7c7 SW |
814 | return -ENOTDIR; |
815 | if (IS_ROOT(victim)) | |
816 | return -EBUSY; | |
e36cb0b8 | 817 | } else if (d_is_dir(victim)) |
4260f7c7 SW |
818 | return -EISDIR; |
819 | if (IS_DEADDIR(dir)) | |
820 | return -ENOENT; | |
821 | if (victim->d_flags & DCACHE_NFSFS_RENAMED) | |
822 | return -EBUSY; | |
823 | return 0; | |
824 | } | |
825 | ||
cb8e7090 CH |
826 | /* copy of may_create in fs/namei.c() */ |
827 | static inline int btrfs_may_create(struct inode *dir, struct dentry *child) | |
828 | { | |
2b0143b5 | 829 | if (d_really_is_positive(child)) |
cb8e7090 CH |
830 | return -EEXIST; |
831 | if (IS_DEADDIR(dir)) | |
832 | return -ENOENT; | |
47291baa | 833 | return inode_permission(&init_user_ns, dir, MAY_WRITE | MAY_EXEC); |
cb8e7090 CH |
834 | } |
835 | ||
836 | /* | |
837 | * Create a new subvolume below @parent. This is largely modeled after | |
838 | * sys_mkdirat and vfs_mkdir, but we only do a single component lookup | |
839 | * inside this filesystem so it's quite a bit simpler. | |
840 | */ | |
92872094 | 841 | static noinline int btrfs_mksubvol(const struct path *parent, |
52f75f4f | 842 | const char *name, int namelen, |
72fd032e | 843 | struct btrfs_root *snap_src, |
9babda9f | 844 | bool readonly, |
8696c533 | 845 | struct btrfs_qgroup_inherit *inherit) |
cb8e7090 | 846 | { |
0b246afa JM |
847 | struct inode *dir = d_inode(parent->dentry); |
848 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); | |
cb8e7090 CH |
849 | struct dentry *dentry; |
850 | int error; | |
851 | ||
00235411 AV |
852 | error = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT); |
853 | if (error == -EINTR) | |
854 | return error; | |
cb8e7090 CH |
855 | |
856 | dentry = lookup_one_len(name, parent->dentry, namelen); | |
857 | error = PTR_ERR(dentry); | |
858 | if (IS_ERR(dentry)) | |
859 | goto out_unlock; | |
860 | ||
76dda93c | 861 | error = btrfs_may_create(dir, dentry); |
cb8e7090 | 862 | if (error) |
a874a63e | 863 | goto out_dput; |
cb8e7090 | 864 | |
9c52057c CM |
865 | /* |
866 | * even if this name doesn't exist, we may get hash collisions. | |
867 | * check for them now when we can safely fail | |
868 | */ | |
869 | error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root, | |
870 | dir->i_ino, name, | |
871 | namelen); | |
872 | if (error) | |
873 | goto out_dput; | |
874 | ||
0b246afa | 875 | down_read(&fs_info->subvol_sem); |
76dda93c YZ |
876 | |
877 | if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0) | |
878 | goto out_up_read; | |
879 | ||
9babda9f NB |
880 | if (snap_src) |
881 | error = create_snapshot(snap_src, dir, dentry, readonly, inherit); | |
882 | else | |
883 | error = create_subvol(dir, dentry, name, namelen, inherit); | |
884 | ||
76dda93c YZ |
885 | if (!error) |
886 | fsnotify_mkdir(dir, dentry); | |
887 | out_up_read: | |
0b246afa | 888 | up_read(&fs_info->subvol_sem); |
cb8e7090 CH |
889 | out_dput: |
890 | dput(dentry); | |
891 | out_unlock: | |
64708539 | 892 | btrfs_inode_unlock(dir, 0); |
cb8e7090 CH |
893 | return error; |
894 | } | |
895 | ||
c11fbb6e RK |
896 | static noinline int btrfs_mksnapshot(const struct path *parent, |
897 | const char *name, int namelen, | |
898 | struct btrfs_root *root, | |
899 | bool readonly, | |
900 | struct btrfs_qgroup_inherit *inherit) | |
901 | { | |
902 | int ret; | |
903 | bool snapshot_force_cow = false; | |
904 | ||
905 | /* | |
906 | * Force new buffered writes to reserve space even when NOCOW is | |
907 | * possible. This is to avoid later writeback (running dealloc) to | |
908 | * fallback to COW mode and unexpectedly fail with ENOSPC. | |
909 | */ | |
910 | btrfs_drew_read_lock(&root->snapshot_lock); | |
911 | ||
f9baa501 | 912 | ret = btrfs_start_delalloc_snapshot(root, false); |
c11fbb6e RK |
913 | if (ret) |
914 | goto out; | |
915 | ||
916 | /* | |
917 | * All previous writes have started writeback in NOCOW mode, so now | |
918 | * we force future writes to fallback to COW mode during snapshot | |
919 | * creation. | |
920 | */ | |
921 | atomic_inc(&root->snapshot_force_cow); | |
922 | snapshot_force_cow = true; | |
923 | ||
924 | btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1); | |
925 | ||
926 | ret = btrfs_mksubvol(parent, name, namelen, | |
927 | root, readonly, inherit); | |
928 | out: | |
929 | if (snapshot_force_cow) | |
930 | atomic_dec(&root->snapshot_force_cow); | |
931 | btrfs_drew_read_unlock(&root->snapshot_lock); | |
932 | return ret; | |
933 | } | |
934 | ||
4cb5300b CM |
935 | /* |
936 | * When we're defragging a range, we don't want to kick it off again | |
937 | * if it is really just waiting for delalloc to send it down. | |
938 | * If we find a nice big extent or delalloc range for the bytes in the | |
939 | * file you want to defrag, we return 0 to let you know to skip this | |
940 | * part of the file | |
941 | */ | |
aab110ab | 942 | static int check_defrag_in_cache(struct inode *inode, u64 offset, u32 thresh) |
4cb5300b CM |
943 | { |
944 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; | |
945 | struct extent_map *em = NULL; | |
946 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
947 | u64 end; | |
948 | ||
949 | read_lock(&em_tree->lock); | |
09cbfeaf | 950 | em = lookup_extent_mapping(em_tree, offset, PAGE_SIZE); |
4cb5300b CM |
951 | read_unlock(&em_tree->lock); |
952 | ||
953 | if (em) { | |
954 | end = extent_map_end(em); | |
955 | free_extent_map(em); | |
956 | if (end - offset > thresh) | |
957 | return 0; | |
958 | } | |
959 | /* if we already have a nice delalloc here, just stop */ | |
960 | thresh /= 2; | |
961 | end = count_range_bits(io_tree, &offset, offset + thresh, | |
962 | thresh, EXTENT_DELALLOC, 1); | |
963 | if (end >= thresh) | |
964 | return 0; | |
965 | return 1; | |
966 | } | |
967 | ||
968 | /* | |
969 | * helper function to walk through a file and find extents | |
970 | * newer than a specific transid, and smaller than thresh. | |
971 | * | |
972 | * This is used by the defragging code to find new and small | |
973 | * extents | |
974 | */ | |
975 | static int find_new_extents(struct btrfs_root *root, | |
976 | struct inode *inode, u64 newer_than, | |
aab110ab | 977 | u64 *off, u32 thresh) |
4cb5300b CM |
978 | { |
979 | struct btrfs_path *path; | |
980 | struct btrfs_key min_key; | |
4cb5300b CM |
981 | struct extent_buffer *leaf; |
982 | struct btrfs_file_extent_item *extent; | |
983 | int type; | |
984 | int ret; | |
4a0cc7ca | 985 | u64 ino = btrfs_ino(BTRFS_I(inode)); |
4cb5300b CM |
986 | |
987 | path = btrfs_alloc_path(); | |
988 | if (!path) | |
989 | return -ENOMEM; | |
990 | ||
a4689d2b | 991 | min_key.objectid = ino; |
4cb5300b CM |
992 | min_key.type = BTRFS_EXTENT_DATA_KEY; |
993 | min_key.offset = *off; | |
994 | ||
67871254 | 995 | while (1) { |
6174d3cb | 996 | ret = btrfs_search_forward(root, &min_key, path, newer_than); |
4cb5300b CM |
997 | if (ret != 0) |
998 | goto none; | |
f094c9bd | 999 | process_slot: |
a4689d2b | 1000 | if (min_key.objectid != ino) |
4cb5300b CM |
1001 | goto none; |
1002 | if (min_key.type != BTRFS_EXTENT_DATA_KEY) | |
1003 | goto none; | |
1004 | ||
1005 | leaf = path->nodes[0]; | |
1006 | extent = btrfs_item_ptr(leaf, path->slots[0], | |
1007 | struct btrfs_file_extent_item); | |
1008 | ||
1009 | type = btrfs_file_extent_type(leaf, extent); | |
1010 | if (type == BTRFS_FILE_EXTENT_REG && | |
1011 | btrfs_file_extent_num_bytes(leaf, extent) < thresh && | |
1012 | check_defrag_in_cache(inode, min_key.offset, thresh)) { | |
1013 | *off = min_key.offset; | |
1014 | btrfs_free_path(path); | |
1015 | return 0; | |
1016 | } | |
1017 | ||
f094c9bd FM |
1018 | path->slots[0]++; |
1019 | if (path->slots[0] < btrfs_header_nritems(leaf)) { | |
1020 | btrfs_item_key_to_cpu(leaf, &min_key, path->slots[0]); | |
1021 | goto process_slot; | |
1022 | } | |
1023 | ||
4cb5300b CM |
1024 | if (min_key.offset == (u64)-1) |
1025 | goto none; | |
1026 | ||
1027 | min_key.offset++; | |
1028 | btrfs_release_path(path); | |
1029 | } | |
1030 | none: | |
1031 | btrfs_free_path(path); | |
1032 | return -ENOENT; | |
1033 | } | |
1034 | ||
6c282eb4 | 1035 | static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start) |
17ce6ef8 LB |
1036 | { |
1037 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
6c282eb4 LZ |
1038 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; |
1039 | struct extent_map *em; | |
09cbfeaf | 1040 | u64 len = PAGE_SIZE; |
17ce6ef8 | 1041 | |
6c282eb4 LZ |
1042 | /* |
1043 | * hopefully we have this extent in the tree already, try without | |
1044 | * the full extent lock | |
1045 | */ | |
17ce6ef8 | 1046 | read_lock(&em_tree->lock); |
6c282eb4 | 1047 | em = lookup_extent_mapping(em_tree, start, len); |
17ce6ef8 LB |
1048 | read_unlock(&em_tree->lock); |
1049 | ||
6c282eb4 | 1050 | if (!em) { |
308d9800 FM |
1051 | struct extent_state *cached = NULL; |
1052 | u64 end = start + len - 1; | |
1053 | ||
6c282eb4 | 1054 | /* get the big lock and read metadata off disk */ |
ff13db41 | 1055 | lock_extent_bits(io_tree, start, end, &cached); |
39b07b5d | 1056 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start, len); |
e43bbe5e | 1057 | unlock_extent_cached(io_tree, start, end, &cached); |
6c282eb4 LZ |
1058 | |
1059 | if (IS_ERR(em)) | |
1060 | return NULL; | |
1061 | } | |
1062 | ||
1063 | return em; | |
1064 | } | |
17ce6ef8 | 1065 | |
6c282eb4 LZ |
1066 | static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em) |
1067 | { | |
1068 | struct extent_map *next; | |
1069 | bool ret = true; | |
1070 | ||
1071 | /* this is the last extent */ | |
1072 | if (em->start + em->len >= i_size_read(inode)) | |
1073 | return false; | |
1074 | ||
1075 | next = defrag_lookup_extent(inode, em->start + em->len); | |
e9512d72 CM |
1076 | if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE) |
1077 | ret = false; | |
1078 | else if ((em->block_start + em->block_len == next->block_start) && | |
ee22184b | 1079 | (em->block_len > SZ_128K && next->block_len > SZ_128K)) |
6c282eb4 LZ |
1080 | ret = false; |
1081 | ||
1082 | free_extent_map(next); | |
17ce6ef8 LB |
1083 | return ret; |
1084 | } | |
1085 | ||
aab110ab | 1086 | static int should_defrag_range(struct inode *inode, u64 start, u32 thresh, |
a43a2111 AM |
1087 | u64 *last_len, u64 *skip, u64 *defrag_end, |
1088 | int compress) | |
940100a4 | 1089 | { |
6c282eb4 | 1090 | struct extent_map *em; |
940100a4 | 1091 | int ret = 1; |
6c282eb4 | 1092 | bool next_mergeable = true; |
4a3560c4 | 1093 | bool prev_mergeable = true; |
940100a4 CM |
1094 | |
1095 | /* | |
008873ea | 1096 | * make sure that once we start defragging an extent, we keep on |
940100a4 CM |
1097 | * defragging it |
1098 | */ | |
1099 | if (start < *defrag_end) | |
1100 | return 1; | |
1101 | ||
1102 | *skip = 0; | |
1103 | ||
6c282eb4 LZ |
1104 | em = defrag_lookup_extent(inode, start); |
1105 | if (!em) | |
1106 | return 0; | |
940100a4 CM |
1107 | |
1108 | /* this will cover holes, and inline extents */ | |
17ce6ef8 | 1109 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { |
940100a4 | 1110 | ret = 0; |
17ce6ef8 LB |
1111 | goto out; |
1112 | } | |
1113 | ||
4a3560c4 LB |
1114 | if (!*defrag_end) |
1115 | prev_mergeable = false; | |
1116 | ||
6c282eb4 | 1117 | next_mergeable = defrag_check_next_extent(inode, em); |
940100a4 | 1118 | /* |
6c282eb4 LZ |
1119 | * we hit a real extent, if it is big or the next extent is not a |
1120 | * real extent, don't bother defragging it | |
940100a4 | 1121 | */ |
a43a2111 | 1122 | if (!compress && (*last_len == 0 || *last_len >= thresh) && |
4a3560c4 | 1123 | (em->len >= thresh || (!next_mergeable && !prev_mergeable))) |
940100a4 | 1124 | ret = 0; |
17ce6ef8 | 1125 | out: |
940100a4 CM |
1126 | /* |
1127 | * last_len ends up being a counter of how many bytes we've defragged. | |
1128 | * every time we choose not to defrag an extent, we reset *last_len | |
1129 | * so that the next tiny extent will force a defrag. | |
1130 | * | |
1131 | * The end result of this is that tiny extents before a single big | |
1132 | * extent will force at least part of that big extent to be defragged. | |
1133 | */ | |
1134 | if (ret) { | |
940100a4 CM |
1135 | *defrag_end = extent_map_end(em); |
1136 | } else { | |
1137 | *last_len = 0; | |
1138 | *skip = extent_map_end(em); | |
1139 | *defrag_end = 0; | |
1140 | } | |
1141 | ||
1142 | free_extent_map(em); | |
1143 | return ret; | |
1144 | } | |
1145 | ||
4cb5300b CM |
1146 | /* |
1147 | * it doesn't do much good to defrag one or two pages | |
1148 | * at a time. This pulls in a nice chunk of pages | |
1149 | * to COW and defrag. | |
1150 | * | |
1151 | * It also makes sure the delalloc code has enough | |
1152 | * dirty data to avoid making new small extents as part | |
1153 | * of the defrag | |
1154 | * | |
1155 | * It's a good idea to start RA on this range | |
1156 | * before calling this. | |
1157 | */ | |
1158 | static int cluster_pages_for_defrag(struct inode *inode, | |
1159 | struct page **pages, | |
1160 | unsigned long start_index, | |
c41570c9 | 1161 | unsigned long num_pages) |
f46b5a66 | 1162 | { |
4cb5300b CM |
1163 | unsigned long file_end; |
1164 | u64 isize = i_size_read(inode); | |
1165 | u64 page_start; | |
1166 | u64 page_end; | |
1f12bd06 | 1167 | u64 page_cnt; |
a1fbc675 | 1168 | u64 start = (u64)start_index << PAGE_SHIFT; |
7f458a38 | 1169 | u64 search_start; |
4cb5300b CM |
1170 | int ret; |
1171 | int i; | |
1172 | int i_done; | |
3eaa2885 | 1173 | struct btrfs_ordered_extent *ordered; |
4cb5300b | 1174 | struct extent_state *cached_state = NULL; |
600a45e1 | 1175 | struct extent_io_tree *tree; |
364ecf36 | 1176 | struct extent_changeset *data_reserved = NULL; |
3b16a4e3 | 1177 | gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); |
4cb5300b | 1178 | |
09cbfeaf | 1179 | file_end = (isize - 1) >> PAGE_SHIFT; |
1f12bd06 LB |
1180 | if (!isize || start_index > file_end) |
1181 | return 0; | |
1182 | ||
1183 | page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1); | |
4cb5300b | 1184 | |
e5b7231e | 1185 | ret = btrfs_delalloc_reserve_space(BTRFS_I(inode), &data_reserved, |
a1fbc675 | 1186 | start, page_cnt << PAGE_SHIFT); |
4cb5300b CM |
1187 | if (ret) |
1188 | return ret; | |
4cb5300b | 1189 | i_done = 0; |
600a45e1 | 1190 | tree = &BTRFS_I(inode)->io_tree; |
4cb5300b CM |
1191 | |
1192 | /* step one, lock all the pages */ | |
1f12bd06 | 1193 | for (i = 0; i < page_cnt; i++) { |
4cb5300b | 1194 | struct page *page; |
600a45e1 | 1195 | again: |
a94733d0 | 1196 | page = find_or_create_page(inode->i_mapping, |
600a45e1 | 1197 | start_index + i, mask); |
4cb5300b CM |
1198 | if (!page) |
1199 | break; | |
1200 | ||
32443de3 QW |
1201 | ret = set_page_extent_mapped(page); |
1202 | if (ret < 0) { | |
1203 | unlock_page(page); | |
1204 | put_page(page); | |
1205 | break; | |
1206 | } | |
1207 | ||
600a45e1 | 1208 | page_start = page_offset(page); |
09cbfeaf | 1209 | page_end = page_start + PAGE_SIZE - 1; |
600a45e1 | 1210 | while (1) { |
308d9800 | 1211 | lock_extent_bits(tree, page_start, page_end, |
ff13db41 | 1212 | &cached_state); |
c3504372 | 1213 | ordered = btrfs_lookup_ordered_extent(BTRFS_I(inode), |
600a45e1 | 1214 | page_start); |
308d9800 | 1215 | unlock_extent_cached(tree, page_start, page_end, |
e43bbe5e | 1216 | &cached_state); |
600a45e1 MX |
1217 | if (!ordered) |
1218 | break; | |
1219 | ||
1220 | unlock_page(page); | |
c0a43603 | 1221 | btrfs_start_ordered_extent(ordered, 1); |
600a45e1 MX |
1222 | btrfs_put_ordered_extent(ordered); |
1223 | lock_page(page); | |
1f12bd06 LB |
1224 | /* |
1225 | * we unlocked the page above, so we need check if | |
1226 | * it was released or not. | |
1227 | */ | |
1228 | if (page->mapping != inode->i_mapping) { | |
1229 | unlock_page(page); | |
09cbfeaf | 1230 | put_page(page); |
1f12bd06 LB |
1231 | goto again; |
1232 | } | |
600a45e1 MX |
1233 | } |
1234 | ||
4cb5300b CM |
1235 | if (!PageUptodate(page)) { |
1236 | btrfs_readpage(NULL, page); | |
1237 | lock_page(page); | |
1238 | if (!PageUptodate(page)) { | |
1239 | unlock_page(page); | |
09cbfeaf | 1240 | put_page(page); |
4cb5300b CM |
1241 | ret = -EIO; |
1242 | break; | |
1243 | } | |
1244 | } | |
600a45e1 | 1245 | |
600a45e1 MX |
1246 | if (page->mapping != inode->i_mapping) { |
1247 | unlock_page(page); | |
09cbfeaf | 1248 | put_page(page); |
600a45e1 MX |
1249 | goto again; |
1250 | } | |
1251 | ||
4cb5300b CM |
1252 | pages[i] = page; |
1253 | i_done++; | |
1254 | } | |
1255 | if (!i_done || ret) | |
1256 | goto out; | |
1257 | ||
1751e8a6 | 1258 | if (!(inode->i_sb->s_flags & SB_ACTIVE)) |
4cb5300b CM |
1259 | goto out; |
1260 | ||
1261 | /* | |
1262 | * so now we have a nice long stream of locked | |
1263 | * and up to date pages, lets wait on them | |
1264 | */ | |
1265 | for (i = 0; i < i_done; i++) | |
1266 | wait_on_page_writeback(pages[i]); | |
1267 | ||
1268 | page_start = page_offset(pages[0]); | |
09cbfeaf | 1269 | page_end = page_offset(pages[i_done - 1]) + PAGE_SIZE; |
4cb5300b CM |
1270 | |
1271 | lock_extent_bits(&BTRFS_I(inode)->io_tree, | |
ff13db41 | 1272 | page_start, page_end - 1, &cached_state); |
7f458a38 FM |
1273 | |
1274 | /* | |
1275 | * When defragmenting we skip ranges that have holes or inline extents, | |
1276 | * (check should_defrag_range()), to avoid unnecessary IO and wasting | |
1277 | * space. At btrfs_defrag_file(), we check if a range should be defragged | |
1278 | * before locking the inode and then, if it should, we trigger a sync | |
1279 | * page cache readahead - we lock the inode only after that to avoid | |
1280 | * blocking for too long other tasks that possibly want to operate on | |
1281 | * other file ranges. But before we were able to get the inode lock, | |
1282 | * some other task may have punched a hole in the range, or we may have | |
1283 | * now an inline extent, in which case we should not defrag. So check | |
1284 | * for that here, where we have the inode and the range locked, and bail | |
1285 | * out if that happened. | |
1286 | */ | |
1287 | search_start = page_start; | |
1288 | while (search_start < page_end) { | |
1289 | struct extent_map *em; | |
1290 | ||
1291 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, search_start, | |
1292 | page_end - search_start); | |
1293 | if (IS_ERR(em)) { | |
1294 | ret = PTR_ERR(em); | |
1295 | goto out_unlock_range; | |
1296 | } | |
1297 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { | |
1298 | free_extent_map(em); | |
1299 | /* Ok, 0 means we did not defrag anything */ | |
1300 | ret = 0; | |
1301 | goto out_unlock_range; | |
1302 | } | |
1303 | search_start = extent_map_end(em); | |
1304 | free_extent_map(em); | |
1305 | } | |
1306 | ||
4cb5300b | 1307 | clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, |
e182163d OS |
1308 | page_end - 1, EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | |
1309 | EXTENT_DEFRAG, 0, 0, &cached_state); | |
4cb5300b | 1310 | |
1f12bd06 | 1311 | if (i_done != page_cnt) { |
9e0baf60 | 1312 | spin_lock(&BTRFS_I(inode)->lock); |
28c4a3e2 | 1313 | btrfs_mod_outstanding_extents(BTRFS_I(inode), 1); |
9e0baf60 | 1314 | spin_unlock(&BTRFS_I(inode)->lock); |
86d52921 | 1315 | btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, |
a1fbc675 | 1316 | start, (page_cnt - i_done) << PAGE_SHIFT, true); |
4cb5300b CM |
1317 | } |
1318 | ||
1319 | ||
9e8a4a8b | 1320 | set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1, |
018ed4f7 | 1321 | &cached_state); |
4cb5300b CM |
1322 | |
1323 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
e43bbe5e | 1324 | page_start, page_end - 1, &cached_state); |
4cb5300b CM |
1325 | |
1326 | for (i = 0; i < i_done; i++) { | |
1327 | clear_page_dirty_for_io(pages[i]); | |
1328 | ClearPageChecked(pages[i]); | |
4cb5300b CM |
1329 | set_page_dirty(pages[i]); |
1330 | unlock_page(pages[i]); | |
09cbfeaf | 1331 | put_page(pages[i]); |
4cb5300b | 1332 | } |
8702ba93 | 1333 | btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT); |
364ecf36 | 1334 | extent_changeset_free(data_reserved); |
4cb5300b | 1335 | return i_done; |
7f458a38 FM |
1336 | |
1337 | out_unlock_range: | |
1338 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
1339 | page_start, page_end - 1, &cached_state); | |
4cb5300b CM |
1340 | out: |
1341 | for (i = 0; i < i_done; i++) { | |
1342 | unlock_page(pages[i]); | |
09cbfeaf | 1343 | put_page(pages[i]); |
4cb5300b | 1344 | } |
86d52921 | 1345 | btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, |
a1fbc675 | 1346 | start, page_cnt << PAGE_SHIFT, true); |
8702ba93 | 1347 | btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT); |
364ecf36 | 1348 | extent_changeset_free(data_reserved); |
4cb5300b CM |
1349 | return ret; |
1350 | ||
1351 | } | |
1352 | ||
1353 | int btrfs_defrag_file(struct inode *inode, struct file *file, | |
1354 | struct btrfs_ioctl_defrag_range_args *range, | |
1355 | u64 newer_than, unsigned long max_to_defrag) | |
1356 | { | |
0b246afa | 1357 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
4cb5300b | 1358 | struct btrfs_root *root = BTRFS_I(inode)->root; |
4cb5300b | 1359 | struct file_ra_state *ra = NULL; |
f46b5a66 | 1360 | unsigned long last_index; |
151a31b2 | 1361 | u64 isize = i_size_read(inode); |
940100a4 CM |
1362 | u64 last_len = 0; |
1363 | u64 skip = 0; | |
1364 | u64 defrag_end = 0; | |
4cb5300b | 1365 | u64 newer_off = range->start; |
f46b5a66 | 1366 | unsigned long i; |
008873ea | 1367 | unsigned long ra_index = 0; |
f46b5a66 | 1368 | int ret; |
4cb5300b | 1369 | int defrag_count = 0; |
1a419d85 | 1370 | int compress_type = BTRFS_COMPRESS_ZLIB; |
aab110ab | 1371 | u32 extent_thresh = range->extent_thresh; |
09cbfeaf | 1372 | unsigned long max_cluster = SZ_256K >> PAGE_SHIFT; |
c41570c9 | 1373 | unsigned long cluster = max_cluster; |
ee22184b | 1374 | u64 new_align = ~((u64)SZ_128K - 1); |
4cb5300b | 1375 | struct page **pages = NULL; |
1e2ef46d | 1376 | bool do_compress = range->flags & BTRFS_DEFRAG_RANGE_COMPRESS; |
4cb5300b | 1377 | |
0abd5b17 LB |
1378 | if (isize == 0) |
1379 | return 0; | |
1380 | ||
1381 | if (range->start >= isize) | |
1382 | return -EINVAL; | |
1a419d85 | 1383 | |
1e2ef46d | 1384 | if (do_compress) { |
ce96b7ff | 1385 | if (range->compress_type >= BTRFS_NR_COMPRESS_TYPES) |
1a419d85 LZ |
1386 | return -EINVAL; |
1387 | if (range->compress_type) | |
1388 | compress_type = range->compress_type; | |
1389 | } | |
f46b5a66 | 1390 | |
0abd5b17 | 1391 | if (extent_thresh == 0) |
ee22184b | 1392 | extent_thresh = SZ_256K; |
940100a4 | 1393 | |
4cb5300b | 1394 | /* |
0a52d108 DS |
1395 | * If we were not given a file, allocate a readahead context. As |
1396 | * readahead is just an optimization, defrag will work without it so | |
1397 | * we don't error out. | |
4cb5300b CM |
1398 | */ |
1399 | if (!file) { | |
63e727ec | 1400 | ra = kzalloc(sizeof(*ra), GFP_KERNEL); |
0a52d108 DS |
1401 | if (ra) |
1402 | file_ra_state_init(ra, inode->i_mapping); | |
4cb5300b CM |
1403 | } else { |
1404 | ra = &file->f_ra; | |
1405 | } | |
1406 | ||
63e727ec | 1407 | pages = kmalloc_array(max_cluster, sizeof(struct page *), GFP_KERNEL); |
4cb5300b CM |
1408 | if (!pages) { |
1409 | ret = -ENOMEM; | |
1410 | goto out_ra; | |
1411 | } | |
1412 | ||
1413 | /* find the last page to defrag */ | |
1e701a32 | 1414 | if (range->start + range->len > range->start) { |
151a31b2 | 1415 | last_index = min_t(u64, isize - 1, |
09cbfeaf | 1416 | range->start + range->len - 1) >> PAGE_SHIFT; |
1e701a32 | 1417 | } else { |
09cbfeaf | 1418 | last_index = (isize - 1) >> PAGE_SHIFT; |
1e701a32 CM |
1419 | } |
1420 | ||
4cb5300b CM |
1421 | if (newer_than) { |
1422 | ret = find_new_extents(root, inode, newer_than, | |
ee22184b | 1423 | &newer_off, SZ_64K); |
4cb5300b CM |
1424 | if (!ret) { |
1425 | range->start = newer_off; | |
1426 | /* | |
1427 | * we always align our defrag to help keep | |
1428 | * the extents in the file evenly spaced | |
1429 | */ | |
09cbfeaf | 1430 | i = (newer_off & new_align) >> PAGE_SHIFT; |
4cb5300b CM |
1431 | } else |
1432 | goto out_ra; | |
1433 | } else { | |
09cbfeaf | 1434 | i = range->start >> PAGE_SHIFT; |
4cb5300b CM |
1435 | } |
1436 | if (!max_to_defrag) | |
070034bd | 1437 | max_to_defrag = last_index - i + 1; |
4cb5300b | 1438 | |
2a0f7f57 LZ |
1439 | /* |
1440 | * make writeback starts from i, so the defrag range can be | |
1441 | * written sequentially. | |
1442 | */ | |
1443 | if (i < inode->i_mapping->writeback_index) | |
1444 | inode->i_mapping->writeback_index = i; | |
1445 | ||
f7f43cc8 | 1446 | while (i <= last_index && defrag_count < max_to_defrag && |
09cbfeaf | 1447 | (i < DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE))) { |
4cb5300b CM |
1448 | /* |
1449 | * make sure we stop running if someone unmounts | |
1450 | * the FS | |
1451 | */ | |
1751e8a6 | 1452 | if (!(inode->i_sb->s_flags & SB_ACTIVE)) |
4cb5300b CM |
1453 | break; |
1454 | ||
0b246afa JM |
1455 | if (btrfs_defrag_cancelled(fs_info)) { |
1456 | btrfs_debug(fs_info, "defrag_file cancelled"); | |
210549eb | 1457 | ret = -EAGAIN; |
50535db8 | 1458 | goto error; |
210549eb DS |
1459 | } |
1460 | ||
09cbfeaf | 1461 | if (!should_defrag_range(inode, (u64)i << PAGE_SHIFT, |
6c282eb4 | 1462 | extent_thresh, &last_len, &skip, |
1e2ef46d | 1463 | &defrag_end, do_compress)){ |
940100a4 CM |
1464 | unsigned long next; |
1465 | /* | |
1466 | * the should_defrag function tells us how much to skip | |
1467 | * bump our counter by the suggested amount | |
1468 | */ | |
09cbfeaf | 1469 | next = DIV_ROUND_UP(skip, PAGE_SIZE); |
940100a4 CM |
1470 | i = max(i + 1, next); |
1471 | continue; | |
1472 | } | |
008873ea LZ |
1473 | |
1474 | if (!newer_than) { | |
09cbfeaf KS |
1475 | cluster = (PAGE_ALIGN(defrag_end) >> |
1476 | PAGE_SHIFT) - i; | |
008873ea LZ |
1477 | cluster = min(cluster, max_cluster); |
1478 | } else { | |
1479 | cluster = max_cluster; | |
1480 | } | |
1481 | ||
008873ea LZ |
1482 | if (i + cluster > ra_index) { |
1483 | ra_index = max(i, ra_index); | |
0a52d108 | 1484 | if (ra) |
d3c0bab5 DS |
1485 | page_cache_sync_readahead(inode->i_mapping, ra, |
1486 | file, ra_index, cluster); | |
e4826a5b | 1487 | ra_index += cluster; |
008873ea | 1488 | } |
940100a4 | 1489 | |
64708539 | 1490 | btrfs_inode_lock(inode, 0); |
eede2bf3 OS |
1491 | if (IS_SWAPFILE(inode)) { |
1492 | ret = -ETXTBSY; | |
1493 | } else { | |
1494 | if (do_compress) | |
1495 | BTRFS_I(inode)->defrag_compress = compress_type; | |
1496 | ret = cluster_pages_for_defrag(inode, pages, i, cluster); | |
1497 | } | |
ecb8bea8 | 1498 | if (ret < 0) { |
64708539 | 1499 | btrfs_inode_unlock(inode, 0); |
4cb5300b | 1500 | goto out_ra; |
ecb8bea8 | 1501 | } |
4cb5300b CM |
1502 | |
1503 | defrag_count += ret; | |
d0e1d66b | 1504 | balance_dirty_pages_ratelimited(inode->i_mapping); |
64708539 | 1505 | btrfs_inode_unlock(inode, 0); |
4cb5300b CM |
1506 | |
1507 | if (newer_than) { | |
1508 | if (newer_off == (u64)-1) | |
1509 | break; | |
1510 | ||
e1f041e1 LB |
1511 | if (ret > 0) |
1512 | i += ret; | |
1513 | ||
4cb5300b | 1514 | newer_off = max(newer_off + 1, |
09cbfeaf | 1515 | (u64)i << PAGE_SHIFT); |
4cb5300b | 1516 | |
ee22184b BL |
1517 | ret = find_new_extents(root, inode, newer_than, |
1518 | &newer_off, SZ_64K); | |
4cb5300b CM |
1519 | if (!ret) { |
1520 | range->start = newer_off; | |
09cbfeaf | 1521 | i = (newer_off & new_align) >> PAGE_SHIFT; |
4cb5300b CM |
1522 | } else { |
1523 | break; | |
f46b5a66 | 1524 | } |
4cb5300b | 1525 | } else { |
008873ea | 1526 | if (ret > 0) { |
cbcc8326 | 1527 | i += ret; |
09cbfeaf | 1528 | last_len += ret << PAGE_SHIFT; |
008873ea | 1529 | } else { |
cbcc8326 | 1530 | i++; |
008873ea LZ |
1531 | last_len = 0; |
1532 | } | |
f46b5a66 | 1533 | } |
f46b5a66 CH |
1534 | } |
1535 | ||
50535db8 TT |
1536 | ret = defrag_count; |
1537 | error: | |
dec8ef90 | 1538 | if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) { |
1e701a32 | 1539 | filemap_flush(inode->i_mapping); |
dec8ef90 FM |
1540 | if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, |
1541 | &BTRFS_I(inode)->runtime_flags)) | |
1542 | filemap_flush(inode->i_mapping); | |
1543 | } | |
1e701a32 | 1544 | |
1a419d85 | 1545 | if (range->compress_type == BTRFS_COMPRESS_LZO) { |
0b246afa | 1546 | btrfs_set_fs_incompat(fs_info, COMPRESS_LZO); |
5c1aab1d NT |
1547 | } else if (range->compress_type == BTRFS_COMPRESS_ZSTD) { |
1548 | btrfs_set_fs_incompat(fs_info, COMPRESS_ZSTD); | |
1a419d85 LZ |
1549 | } |
1550 | ||
4cb5300b | 1551 | out_ra: |
1e2ef46d | 1552 | if (do_compress) { |
64708539 | 1553 | btrfs_inode_lock(inode, 0); |
eec63c65 | 1554 | BTRFS_I(inode)->defrag_compress = BTRFS_COMPRESS_NONE; |
64708539 | 1555 | btrfs_inode_unlock(inode, 0); |
633085c7 | 1556 | } |
4cb5300b CM |
1557 | if (!file) |
1558 | kfree(ra); | |
1559 | kfree(pages); | |
940100a4 | 1560 | return ret; |
f46b5a66 CH |
1561 | } |
1562 | ||
198605a8 | 1563 | static noinline int btrfs_ioctl_resize(struct file *file, |
76dda93c | 1564 | void __user *arg) |
f46b5a66 | 1565 | { |
0b246afa JM |
1566 | struct inode *inode = file_inode(file); |
1567 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
f46b5a66 CH |
1568 | u64 new_size; |
1569 | u64 old_size; | |
1570 | u64 devid = 1; | |
0b246afa | 1571 | struct btrfs_root *root = BTRFS_I(inode)->root; |
f46b5a66 CH |
1572 | struct btrfs_ioctl_vol_args *vol_args; |
1573 | struct btrfs_trans_handle *trans; | |
1574 | struct btrfs_device *device = NULL; | |
1575 | char *sizestr; | |
9a40f122 | 1576 | char *retptr; |
f46b5a66 CH |
1577 | char *devstr = NULL; |
1578 | int ret = 0; | |
f46b5a66 CH |
1579 | int mod = 0; |
1580 | ||
e441d54d CM |
1581 | if (!capable(CAP_SYS_ADMIN)) |
1582 | return -EPERM; | |
1583 | ||
198605a8 MX |
1584 | ret = mnt_want_write_file(file); |
1585 | if (ret) | |
1586 | return ret; | |
1587 | ||
c3e1f96c | 1588 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_RESIZE)) { |
97547676 | 1589 | mnt_drop_write_file(file); |
e57138b3 | 1590 | return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
c9e9f97b ID |
1591 | } |
1592 | ||
dae7b665 | 1593 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
1594 | if (IS_ERR(vol_args)) { |
1595 | ret = PTR_ERR(vol_args); | |
1596 | goto out; | |
1597 | } | |
5516e595 MF |
1598 | |
1599 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
f46b5a66 | 1600 | |
f46b5a66 CH |
1601 | sizestr = vol_args->name; |
1602 | devstr = strchr(sizestr, ':'); | |
1603 | if (devstr) { | |
f46b5a66 CH |
1604 | sizestr = devstr + 1; |
1605 | *devstr = '\0'; | |
1606 | devstr = vol_args->name; | |
58dfae63 Z |
1607 | ret = kstrtoull(devstr, 10, &devid); |
1608 | if (ret) | |
1609 | goto out_free; | |
dfd79829 MX |
1610 | if (!devid) { |
1611 | ret = -EINVAL; | |
1612 | goto out_free; | |
1613 | } | |
0b246afa | 1614 | btrfs_info(fs_info, "resizing devid %llu", devid); |
f46b5a66 | 1615 | } |
dba60f3f | 1616 | |
b2598edf | 1617 | device = btrfs_find_device(fs_info->fs_devices, devid, NULL, NULL); |
f46b5a66 | 1618 | if (!device) { |
0b246afa JM |
1619 | btrfs_info(fs_info, "resizer unable to find device %llu", |
1620 | devid); | |
dfd79829 | 1621 | ret = -ENODEV; |
c9e9f97b | 1622 | goto out_free; |
f46b5a66 | 1623 | } |
dba60f3f | 1624 | |
ebbede42 | 1625 | if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { |
0b246afa | 1626 | btrfs_info(fs_info, |
efe120a0 | 1627 | "resizer unable to apply on readonly device %llu", |
c1c9ff7c | 1628 | devid); |
dfd79829 | 1629 | ret = -EPERM; |
4e42ae1b LB |
1630 | goto out_free; |
1631 | } | |
1632 | ||
f46b5a66 CH |
1633 | if (!strcmp(sizestr, "max")) |
1634 | new_size = device->bdev->bd_inode->i_size; | |
1635 | else { | |
1636 | if (sizestr[0] == '-') { | |
1637 | mod = -1; | |
1638 | sizestr++; | |
1639 | } else if (sizestr[0] == '+') { | |
1640 | mod = 1; | |
1641 | sizestr++; | |
1642 | } | |
9a40f122 GH |
1643 | new_size = memparse(sizestr, &retptr); |
1644 | if (*retptr != '\0' || new_size == 0) { | |
f46b5a66 | 1645 | ret = -EINVAL; |
c9e9f97b | 1646 | goto out_free; |
f46b5a66 CH |
1647 | } |
1648 | } | |
1649 | ||
401e29c1 | 1650 | if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) { |
dfd79829 | 1651 | ret = -EPERM; |
63a212ab SB |
1652 | goto out_free; |
1653 | } | |
1654 | ||
7cc8e58d | 1655 | old_size = btrfs_device_get_total_bytes(device); |
f46b5a66 CH |
1656 | |
1657 | if (mod < 0) { | |
1658 | if (new_size > old_size) { | |
1659 | ret = -EINVAL; | |
c9e9f97b | 1660 | goto out_free; |
f46b5a66 CH |
1661 | } |
1662 | new_size = old_size - new_size; | |
1663 | } else if (mod > 0) { | |
eb8052e0 | 1664 | if (new_size > ULLONG_MAX - old_size) { |
902c68a4 | 1665 | ret = -ERANGE; |
eb8052e0 WF |
1666 | goto out_free; |
1667 | } | |
f46b5a66 CH |
1668 | new_size = old_size + new_size; |
1669 | } | |
1670 | ||
ee22184b | 1671 | if (new_size < SZ_256M) { |
f46b5a66 | 1672 | ret = -EINVAL; |
c9e9f97b | 1673 | goto out_free; |
f46b5a66 CH |
1674 | } |
1675 | if (new_size > device->bdev->bd_inode->i_size) { | |
1676 | ret = -EFBIG; | |
c9e9f97b | 1677 | goto out_free; |
f46b5a66 CH |
1678 | } |
1679 | ||
47f08b96 | 1680 | new_size = round_down(new_size, fs_info->sectorsize); |
f46b5a66 | 1681 | |
f46b5a66 | 1682 | if (new_size > old_size) { |
a22285a6 | 1683 | trans = btrfs_start_transaction(root, 0); |
98d5dc13 TI |
1684 | if (IS_ERR(trans)) { |
1685 | ret = PTR_ERR(trans); | |
c9e9f97b | 1686 | goto out_free; |
98d5dc13 | 1687 | } |
f46b5a66 | 1688 | ret = btrfs_grow_device(trans, device, new_size); |
3a45bb20 | 1689 | btrfs_commit_transaction(trans); |
ece7d20e | 1690 | } else if (new_size < old_size) { |
f46b5a66 | 1691 | ret = btrfs_shrink_device(device, new_size); |
0253f40e | 1692 | } /* equal, nothing need to do */ |
f46b5a66 | 1693 | |
faf8f7b9 MPS |
1694 | if (ret == 0 && new_size != old_size) |
1695 | btrfs_info_in_rcu(fs_info, | |
1696 | "resize device %s (devid %llu) from %llu to %llu", | |
1697 | rcu_str_deref(device->name), device->devid, | |
1698 | old_size, new_size); | |
c9e9f97b | 1699 | out_free: |
f46b5a66 | 1700 | kfree(vol_args); |
c9e9f97b | 1701 | out: |
c3e1f96c | 1702 | btrfs_exclop_finish(fs_info); |
18f39c41 | 1703 | mnt_drop_write_file(file); |
f46b5a66 CH |
1704 | return ret; |
1705 | } | |
1706 | ||
5d54c67e | 1707 | static noinline int __btrfs_ioctl_snap_create(struct file *file, |
52f75f4f | 1708 | const char *name, unsigned long fd, int subvol, |
5d54c67e | 1709 | bool readonly, |
8696c533 | 1710 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 1711 | { |
f46b5a66 | 1712 | int namelen; |
3de4586c | 1713 | int ret = 0; |
f46b5a66 | 1714 | |
325c50e3 JM |
1715 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1716 | return -ENOTDIR; | |
1717 | ||
a874a63e LB |
1718 | ret = mnt_want_write_file(file); |
1719 | if (ret) | |
1720 | goto out; | |
1721 | ||
72fd032e SW |
1722 | namelen = strlen(name); |
1723 | if (strchr(name, '/')) { | |
f46b5a66 | 1724 | ret = -EINVAL; |
a874a63e | 1725 | goto out_drop_write; |
f46b5a66 CH |
1726 | } |
1727 | ||
16780cab CM |
1728 | if (name[0] == '.' && |
1729 | (namelen == 1 || (name[1] == '.' && namelen == 2))) { | |
1730 | ret = -EEXIST; | |
a874a63e | 1731 | goto out_drop_write; |
16780cab CM |
1732 | } |
1733 | ||
3de4586c | 1734 | if (subvol) { |
72fd032e | 1735 | ret = btrfs_mksubvol(&file->f_path, name, namelen, |
5d54c67e | 1736 | NULL, readonly, inherit); |
cb8e7090 | 1737 | } else { |
2903ff01 | 1738 | struct fd src = fdget(fd); |
3de4586c | 1739 | struct inode *src_inode; |
2903ff01 | 1740 | if (!src.file) { |
3de4586c | 1741 | ret = -EINVAL; |
a874a63e | 1742 | goto out_drop_write; |
3de4586c CM |
1743 | } |
1744 | ||
496ad9aa AV |
1745 | src_inode = file_inode(src.file); |
1746 | if (src_inode->i_sb != file_inode(file)->i_sb) { | |
c79b4713 | 1747 | btrfs_info(BTRFS_I(file_inode(file))->root->fs_info, |
efe120a0 | 1748 | "Snapshot src from another FS"); |
23ad5b17 | 1749 | ret = -EXDEV; |
21cb47be | 1750 | } else if (!inode_owner_or_capable(&init_user_ns, src_inode)) { |
d0242061 DS |
1751 | /* |
1752 | * Subvolume creation is not restricted, but snapshots | |
1753 | * are limited to own subvolumes only | |
1754 | */ | |
1755 | ret = -EPERM; | |
ecd18815 | 1756 | } else { |
c11fbb6e | 1757 | ret = btrfs_mksnapshot(&file->f_path, name, namelen, |
ecd18815 | 1758 | BTRFS_I(src_inode)->root, |
5d54c67e | 1759 | readonly, inherit); |
3de4586c | 1760 | } |
2903ff01 | 1761 | fdput(src); |
cb8e7090 | 1762 | } |
a874a63e LB |
1763 | out_drop_write: |
1764 | mnt_drop_write_file(file); | |
f46b5a66 | 1765 | out: |
72fd032e SW |
1766 | return ret; |
1767 | } | |
1768 | ||
1769 | static noinline int btrfs_ioctl_snap_create(struct file *file, | |
fa0d2b9b | 1770 | void __user *arg, int subvol) |
72fd032e | 1771 | { |
fa0d2b9b | 1772 | struct btrfs_ioctl_vol_args *vol_args; |
72fd032e SW |
1773 | int ret; |
1774 | ||
325c50e3 JM |
1775 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1776 | return -ENOTDIR; | |
1777 | ||
fa0d2b9b LZ |
1778 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1779 | if (IS_ERR(vol_args)) | |
1780 | return PTR_ERR(vol_args); | |
1781 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
72fd032e | 1782 | |
5d54c67e NB |
1783 | ret = __btrfs_ioctl_snap_create(file, vol_args->name, vol_args->fd, |
1784 | subvol, false, NULL); | |
fdfb1e4f | 1785 | |
fa0d2b9b LZ |
1786 | kfree(vol_args); |
1787 | return ret; | |
1788 | } | |
fdfb1e4f | 1789 | |
fa0d2b9b LZ |
1790 | static noinline int btrfs_ioctl_snap_create_v2(struct file *file, |
1791 | void __user *arg, int subvol) | |
1792 | { | |
1793 | struct btrfs_ioctl_vol_args_v2 *vol_args; | |
1794 | int ret; | |
b83cc969 | 1795 | bool readonly = false; |
6f72c7e2 | 1796 | struct btrfs_qgroup_inherit *inherit = NULL; |
75eaa0e2 | 1797 | |
325c50e3 JM |
1798 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1799 | return -ENOTDIR; | |
1800 | ||
fa0d2b9b LZ |
1801 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1802 | if (IS_ERR(vol_args)) | |
1803 | return PTR_ERR(vol_args); | |
1804 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
75eaa0e2 | 1805 | |
673990db | 1806 | if (vol_args->flags & ~BTRFS_SUBVOL_CREATE_ARGS_MASK) { |
b83cc969 | 1807 | ret = -EOPNOTSUPP; |
c47ca32d | 1808 | goto free_args; |
72fd032e | 1809 | } |
fa0d2b9b | 1810 | |
b83cc969 LZ |
1811 | if (vol_args->flags & BTRFS_SUBVOL_RDONLY) |
1812 | readonly = true; | |
6f72c7e2 | 1813 | if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) { |
5011c5a6 DC |
1814 | u64 nums; |
1815 | ||
1816 | if (vol_args->size < sizeof(*inherit) || | |
1817 | vol_args->size > PAGE_SIZE) { | |
6f72c7e2 | 1818 | ret = -EINVAL; |
c47ca32d | 1819 | goto free_args; |
6f72c7e2 AJ |
1820 | } |
1821 | inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size); | |
1822 | if (IS_ERR(inherit)) { | |
1823 | ret = PTR_ERR(inherit); | |
c47ca32d | 1824 | goto free_args; |
6f72c7e2 | 1825 | } |
5011c5a6 DC |
1826 | |
1827 | if (inherit->num_qgroups > PAGE_SIZE || | |
1828 | inherit->num_ref_copies > PAGE_SIZE || | |
1829 | inherit->num_excl_copies > PAGE_SIZE) { | |
1830 | ret = -EINVAL; | |
1831 | goto free_inherit; | |
1832 | } | |
1833 | ||
1834 | nums = inherit->num_qgroups + 2 * inherit->num_ref_copies + | |
1835 | 2 * inherit->num_excl_copies; | |
1836 | if (vol_args->size != struct_size(inherit, qgroups, nums)) { | |
1837 | ret = -EINVAL; | |
1838 | goto free_inherit; | |
1839 | } | |
6f72c7e2 | 1840 | } |
fa0d2b9b | 1841 | |
5d54c67e NB |
1842 | ret = __btrfs_ioctl_snap_create(file, vol_args->name, vol_args->fd, |
1843 | subvol, readonly, inherit); | |
c47ca32d DC |
1844 | if (ret) |
1845 | goto free_inherit; | |
c47ca32d | 1846 | free_inherit: |
6f72c7e2 | 1847 | kfree(inherit); |
c47ca32d DC |
1848 | free_args: |
1849 | kfree(vol_args); | |
f46b5a66 CH |
1850 | return ret; |
1851 | } | |
1852 | ||
0caa102d LZ |
1853 | static noinline int btrfs_ioctl_subvol_getflags(struct file *file, |
1854 | void __user *arg) | |
1855 | { | |
496ad9aa | 1856 | struct inode *inode = file_inode(file); |
0b246afa | 1857 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
0caa102d LZ |
1858 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1859 | int ret = 0; | |
1860 | u64 flags = 0; | |
1861 | ||
4a0cc7ca | 1862 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) |
0caa102d LZ |
1863 | return -EINVAL; |
1864 | ||
0b246afa | 1865 | down_read(&fs_info->subvol_sem); |
0caa102d LZ |
1866 | if (btrfs_root_readonly(root)) |
1867 | flags |= BTRFS_SUBVOL_RDONLY; | |
0b246afa | 1868 | up_read(&fs_info->subvol_sem); |
0caa102d LZ |
1869 | |
1870 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
1871 | ret = -EFAULT; | |
1872 | ||
1873 | return ret; | |
1874 | } | |
1875 | ||
1876 | static noinline int btrfs_ioctl_subvol_setflags(struct file *file, | |
1877 | void __user *arg) | |
1878 | { | |
496ad9aa | 1879 | struct inode *inode = file_inode(file); |
0b246afa | 1880 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
0caa102d LZ |
1881 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1882 | struct btrfs_trans_handle *trans; | |
1883 | u64 root_flags; | |
1884 | u64 flags; | |
1885 | int ret = 0; | |
1886 | ||
21cb47be | 1887 | if (!inode_owner_or_capable(&init_user_ns, inode)) |
bd60ea0f DS |
1888 | return -EPERM; |
1889 | ||
b9ca0664 LB |
1890 | ret = mnt_want_write_file(file); |
1891 | if (ret) | |
1892 | goto out; | |
0caa102d | 1893 | |
4a0cc7ca | 1894 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
b9ca0664 LB |
1895 | ret = -EINVAL; |
1896 | goto out_drop_write; | |
1897 | } | |
0caa102d | 1898 | |
b9ca0664 LB |
1899 | if (copy_from_user(&flags, arg, sizeof(flags))) { |
1900 | ret = -EFAULT; | |
1901 | goto out_drop_write; | |
1902 | } | |
0caa102d | 1903 | |
b9ca0664 LB |
1904 | if (flags & ~BTRFS_SUBVOL_RDONLY) { |
1905 | ret = -EOPNOTSUPP; | |
1906 | goto out_drop_write; | |
1907 | } | |
0caa102d | 1908 | |
0b246afa | 1909 | down_write(&fs_info->subvol_sem); |
0caa102d LZ |
1910 | |
1911 | /* nothing to do */ | |
1912 | if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root)) | |
b9ca0664 | 1913 | goto out_drop_sem; |
0caa102d LZ |
1914 | |
1915 | root_flags = btrfs_root_flags(&root->root_item); | |
2c686537 | 1916 | if (flags & BTRFS_SUBVOL_RDONLY) { |
0caa102d LZ |
1917 | btrfs_set_root_flags(&root->root_item, |
1918 | root_flags | BTRFS_ROOT_SUBVOL_RDONLY); | |
2c686537 DS |
1919 | } else { |
1920 | /* | |
1921 | * Block RO -> RW transition if this subvolume is involved in | |
1922 | * send | |
1923 | */ | |
1924 | spin_lock(&root->root_item_lock); | |
1925 | if (root->send_in_progress == 0) { | |
1926 | btrfs_set_root_flags(&root->root_item, | |
0caa102d | 1927 | root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY); |
2c686537 DS |
1928 | spin_unlock(&root->root_item_lock); |
1929 | } else { | |
1930 | spin_unlock(&root->root_item_lock); | |
0b246afa JM |
1931 | btrfs_warn(fs_info, |
1932 | "Attempt to set subvolume %llu read-write during send", | |
1933 | root->root_key.objectid); | |
2c686537 DS |
1934 | ret = -EPERM; |
1935 | goto out_drop_sem; | |
1936 | } | |
1937 | } | |
0caa102d LZ |
1938 | |
1939 | trans = btrfs_start_transaction(root, 1); | |
1940 | if (IS_ERR(trans)) { | |
1941 | ret = PTR_ERR(trans); | |
1942 | goto out_reset; | |
1943 | } | |
1944 | ||
0b246afa | 1945 | ret = btrfs_update_root(trans, fs_info->tree_root, |
0caa102d | 1946 | &root->root_key, &root->root_item); |
9417ebc8 NB |
1947 | if (ret < 0) { |
1948 | btrfs_end_transaction(trans); | |
1949 | goto out_reset; | |
1950 | } | |
1951 | ||
1952 | ret = btrfs_commit_transaction(trans); | |
0caa102d | 1953 | |
0caa102d LZ |
1954 | out_reset: |
1955 | if (ret) | |
1956 | btrfs_set_root_flags(&root->root_item, root_flags); | |
b9ca0664 | 1957 | out_drop_sem: |
0b246afa | 1958 | up_write(&fs_info->subvol_sem); |
b9ca0664 LB |
1959 | out_drop_write: |
1960 | mnt_drop_write_file(file); | |
1961 | out: | |
0caa102d LZ |
1962 | return ret; |
1963 | } | |
1964 | ||
ac8e9819 CM |
1965 | static noinline int key_in_sk(struct btrfs_key *key, |
1966 | struct btrfs_ioctl_search_key *sk) | |
1967 | { | |
abc6e134 CM |
1968 | struct btrfs_key test; |
1969 | int ret; | |
1970 | ||
1971 | test.objectid = sk->min_objectid; | |
1972 | test.type = sk->min_type; | |
1973 | test.offset = sk->min_offset; | |
1974 | ||
1975 | ret = btrfs_comp_cpu_keys(key, &test); | |
1976 | if (ret < 0) | |
ac8e9819 | 1977 | return 0; |
abc6e134 CM |
1978 | |
1979 | test.objectid = sk->max_objectid; | |
1980 | test.type = sk->max_type; | |
1981 | test.offset = sk->max_offset; | |
1982 | ||
1983 | ret = btrfs_comp_cpu_keys(key, &test); | |
1984 | if (ret > 0) | |
ac8e9819 CM |
1985 | return 0; |
1986 | return 1; | |
1987 | } | |
1988 | ||
df397565 | 1989 | static noinline int copy_to_sk(struct btrfs_path *path, |
ac8e9819 CM |
1990 | struct btrfs_key *key, |
1991 | struct btrfs_ioctl_search_key *sk, | |
9b6e817d | 1992 | size_t *buf_size, |
ba346b35 | 1993 | char __user *ubuf, |
ac8e9819 CM |
1994 | unsigned long *sk_offset, |
1995 | int *num_found) | |
1996 | { | |
1997 | u64 found_transid; | |
1998 | struct extent_buffer *leaf; | |
1999 | struct btrfs_ioctl_search_header sh; | |
dd81d459 | 2000 | struct btrfs_key test; |
ac8e9819 CM |
2001 | unsigned long item_off; |
2002 | unsigned long item_len; | |
2003 | int nritems; | |
2004 | int i; | |
2005 | int slot; | |
ac8e9819 CM |
2006 | int ret = 0; |
2007 | ||
2008 | leaf = path->nodes[0]; | |
2009 | slot = path->slots[0]; | |
2010 | nritems = btrfs_header_nritems(leaf); | |
2011 | ||
2012 | if (btrfs_header_generation(leaf) > sk->max_transid) { | |
2013 | i = nritems; | |
2014 | goto advance_key; | |
2015 | } | |
2016 | found_transid = btrfs_header_generation(leaf); | |
2017 | ||
2018 | for (i = slot; i < nritems; i++) { | |
2019 | item_off = btrfs_item_ptr_offset(leaf, i); | |
2020 | item_len = btrfs_item_size_nr(leaf, i); | |
2021 | ||
03b71c6c GP |
2022 | btrfs_item_key_to_cpu(leaf, key, i); |
2023 | if (!key_in_sk(key, sk)) | |
2024 | continue; | |
2025 | ||
9b6e817d | 2026 | if (sizeof(sh) + item_len > *buf_size) { |
8f5f6178 GH |
2027 | if (*num_found) { |
2028 | ret = 1; | |
2029 | goto out; | |
2030 | } | |
2031 | ||
2032 | /* | |
2033 | * return one empty item back for v1, which does not | |
2034 | * handle -EOVERFLOW | |
2035 | */ | |
2036 | ||
9b6e817d | 2037 | *buf_size = sizeof(sh) + item_len; |
ac8e9819 | 2038 | item_len = 0; |
8f5f6178 GH |
2039 | ret = -EOVERFLOW; |
2040 | } | |
ac8e9819 | 2041 | |
9b6e817d | 2042 | if (sizeof(sh) + item_len + *sk_offset > *buf_size) { |
ac8e9819 | 2043 | ret = 1; |
25c9bc2e | 2044 | goto out; |
ac8e9819 CM |
2045 | } |
2046 | ||
ac8e9819 CM |
2047 | sh.objectid = key->objectid; |
2048 | sh.offset = key->offset; | |
2049 | sh.type = key->type; | |
2050 | sh.len = item_len; | |
2051 | sh.transid = found_transid; | |
2052 | ||
a48b73ec JB |
2053 | /* |
2054 | * Copy search result header. If we fault then loop again so we | |
2055 | * can fault in the pages and -EFAULT there if there's a | |
2056 | * problem. Otherwise we'll fault and then copy the buffer in | |
2057 | * properly this next time through | |
2058 | */ | |
2059 | if (copy_to_user_nofault(ubuf + *sk_offset, &sh, sizeof(sh))) { | |
2060 | ret = 0; | |
ba346b35 GH |
2061 | goto out; |
2062 | } | |
2063 | ||
ac8e9819 CM |
2064 | *sk_offset += sizeof(sh); |
2065 | ||
2066 | if (item_len) { | |
ba346b35 | 2067 | char __user *up = ubuf + *sk_offset; |
a48b73ec JB |
2068 | /* |
2069 | * Copy the item, same behavior as above, but reset the | |
2070 | * * sk_offset so we copy the full thing again. | |
2071 | */ | |
2072 | if (read_extent_buffer_to_user_nofault(leaf, up, | |
2073 | item_off, item_len)) { | |
2074 | ret = 0; | |
2075 | *sk_offset -= sizeof(sh); | |
ba346b35 GH |
2076 | goto out; |
2077 | } | |
2078 | ||
ac8e9819 | 2079 | *sk_offset += item_len; |
ac8e9819 | 2080 | } |
e2156867 | 2081 | (*num_found)++; |
ac8e9819 | 2082 | |
8f5f6178 GH |
2083 | if (ret) /* -EOVERFLOW from above */ |
2084 | goto out; | |
2085 | ||
25c9bc2e GH |
2086 | if (*num_found >= sk->nr_items) { |
2087 | ret = 1; | |
2088 | goto out; | |
2089 | } | |
ac8e9819 CM |
2090 | } |
2091 | advance_key: | |
abc6e134 | 2092 | ret = 0; |
dd81d459 NA |
2093 | test.objectid = sk->max_objectid; |
2094 | test.type = sk->max_type; | |
2095 | test.offset = sk->max_offset; | |
2096 | if (btrfs_comp_cpu_keys(key, &test) >= 0) | |
2097 | ret = 1; | |
2098 | else if (key->offset < (u64)-1) | |
ac8e9819 | 2099 | key->offset++; |
dd81d459 | 2100 | else if (key->type < (u8)-1) { |
abc6e134 | 2101 | key->offset = 0; |
ac8e9819 | 2102 | key->type++; |
dd81d459 | 2103 | } else if (key->objectid < (u64)-1) { |
abc6e134 CM |
2104 | key->offset = 0; |
2105 | key->type = 0; | |
ac8e9819 | 2106 | key->objectid++; |
abc6e134 CM |
2107 | } else |
2108 | ret = 1; | |
25c9bc2e | 2109 | out: |
ba346b35 GH |
2110 | /* |
2111 | * 0: all items from this leaf copied, continue with next | |
2112 | * 1: * more items can be copied, but unused buffer is too small | |
2113 | * * all items were found | |
2114 | * Either way, it will stops the loop which iterates to the next | |
2115 | * leaf | |
2116 | * -EOVERFLOW: item was to large for buffer | |
2117 | * -EFAULT: could not copy extent buffer back to userspace | |
2118 | */ | |
ac8e9819 CM |
2119 | return ret; |
2120 | } | |
2121 | ||
2122 | static noinline int search_ioctl(struct inode *inode, | |
12544442 | 2123 | struct btrfs_ioctl_search_key *sk, |
9b6e817d | 2124 | size_t *buf_size, |
ba346b35 | 2125 | char __user *ubuf) |
ac8e9819 | 2126 | { |
0b246afa | 2127 | struct btrfs_fs_info *info = btrfs_sb(inode->i_sb); |
ac8e9819 CM |
2128 | struct btrfs_root *root; |
2129 | struct btrfs_key key; | |
ac8e9819 | 2130 | struct btrfs_path *path; |
ac8e9819 CM |
2131 | int ret; |
2132 | int num_found = 0; | |
2133 | unsigned long sk_offset = 0; | |
2134 | ||
9b6e817d GH |
2135 | if (*buf_size < sizeof(struct btrfs_ioctl_search_header)) { |
2136 | *buf_size = sizeof(struct btrfs_ioctl_search_header); | |
12544442 | 2137 | return -EOVERFLOW; |
9b6e817d | 2138 | } |
12544442 | 2139 | |
ac8e9819 CM |
2140 | path = btrfs_alloc_path(); |
2141 | if (!path) | |
2142 | return -ENOMEM; | |
2143 | ||
2144 | if (sk->tree_id == 0) { | |
2145 | /* search the root of the inode that was passed */ | |
00246528 | 2146 | root = btrfs_grab_root(BTRFS_I(inode)->root); |
ac8e9819 | 2147 | } else { |
56e9357a | 2148 | root = btrfs_get_fs_root(info, sk->tree_id, true); |
ac8e9819 | 2149 | if (IS_ERR(root)) { |
ac8e9819 | 2150 | btrfs_free_path(path); |
ad1e3d56 | 2151 | return PTR_ERR(root); |
ac8e9819 CM |
2152 | } |
2153 | } | |
2154 | ||
2155 | key.objectid = sk->min_objectid; | |
2156 | key.type = sk->min_type; | |
2157 | key.offset = sk->min_offset; | |
2158 | ||
67871254 | 2159 | while (1) { |
1c78544e FM |
2160 | ret = fault_in_pages_writeable(ubuf + sk_offset, |
2161 | *buf_size - sk_offset); | |
a48b73ec JB |
2162 | if (ret) |
2163 | break; | |
2164 | ||
6174d3cb | 2165 | ret = btrfs_search_forward(root, &key, path, sk->min_transid); |
ac8e9819 CM |
2166 | if (ret != 0) { |
2167 | if (ret > 0) | |
2168 | ret = 0; | |
2169 | goto err; | |
2170 | } | |
df397565 | 2171 | ret = copy_to_sk(path, &key, sk, buf_size, ubuf, |
ac8e9819 | 2172 | &sk_offset, &num_found); |
b3b4aa74 | 2173 | btrfs_release_path(path); |
25c9bc2e | 2174 | if (ret) |
ac8e9819 CM |
2175 | break; |
2176 | ||
2177 | } | |
8f5f6178 GH |
2178 | if (ret > 0) |
2179 | ret = 0; | |
ac8e9819 CM |
2180 | err: |
2181 | sk->nr_items = num_found; | |
00246528 | 2182 | btrfs_put_root(root); |
ac8e9819 CM |
2183 | btrfs_free_path(path); |
2184 | return ret; | |
2185 | } | |
2186 | ||
2187 | static noinline int btrfs_ioctl_tree_search(struct file *file, | |
2188 | void __user *argp) | |
2189 | { | |
ba346b35 GH |
2190 | struct btrfs_ioctl_search_args __user *uargs; |
2191 | struct btrfs_ioctl_search_key sk; | |
9b6e817d GH |
2192 | struct inode *inode; |
2193 | int ret; | |
2194 | size_t buf_size; | |
ac8e9819 CM |
2195 | |
2196 | if (!capable(CAP_SYS_ADMIN)) | |
2197 | return -EPERM; | |
2198 | ||
ba346b35 GH |
2199 | uargs = (struct btrfs_ioctl_search_args __user *)argp; |
2200 | ||
2201 | if (copy_from_user(&sk, &uargs->key, sizeof(sk))) | |
2202 | return -EFAULT; | |
ac8e9819 | 2203 | |
ba346b35 | 2204 | buf_size = sizeof(uargs->buf); |
ac8e9819 | 2205 | |
496ad9aa | 2206 | inode = file_inode(file); |
ba346b35 | 2207 | ret = search_ioctl(inode, &sk, &buf_size, uargs->buf); |
8f5f6178 GH |
2208 | |
2209 | /* | |
2210 | * In the origin implementation an overflow is handled by returning a | |
2211 | * search header with a len of zero, so reset ret. | |
2212 | */ | |
2213 | if (ret == -EOVERFLOW) | |
2214 | ret = 0; | |
2215 | ||
ba346b35 | 2216 | if (ret == 0 && copy_to_user(&uargs->key, &sk, sizeof(sk))) |
ac8e9819 | 2217 | ret = -EFAULT; |
ac8e9819 CM |
2218 | return ret; |
2219 | } | |
2220 | ||
cc68a8a5 GH |
2221 | static noinline int btrfs_ioctl_tree_search_v2(struct file *file, |
2222 | void __user *argp) | |
2223 | { | |
2224 | struct btrfs_ioctl_search_args_v2 __user *uarg; | |
2225 | struct btrfs_ioctl_search_args_v2 args; | |
2226 | struct inode *inode; | |
2227 | int ret; | |
2228 | size_t buf_size; | |
ee22184b | 2229 | const size_t buf_limit = SZ_16M; |
cc68a8a5 GH |
2230 | |
2231 | if (!capable(CAP_SYS_ADMIN)) | |
2232 | return -EPERM; | |
2233 | ||
2234 | /* copy search header and buffer size */ | |
2235 | uarg = (struct btrfs_ioctl_search_args_v2 __user *)argp; | |
2236 | if (copy_from_user(&args, uarg, sizeof(args))) | |
2237 | return -EFAULT; | |
2238 | ||
2239 | buf_size = args.buf_size; | |
2240 | ||
cc68a8a5 GH |
2241 | /* limit result size to 16MB */ |
2242 | if (buf_size > buf_limit) | |
2243 | buf_size = buf_limit; | |
2244 | ||
2245 | inode = file_inode(file); | |
2246 | ret = search_ioctl(inode, &args.key, &buf_size, | |
718dc5fa | 2247 | (char __user *)(&uarg->buf[0])); |
cc68a8a5 GH |
2248 | if (ret == 0 && copy_to_user(&uarg->key, &args.key, sizeof(args.key))) |
2249 | ret = -EFAULT; | |
2250 | else if (ret == -EOVERFLOW && | |
2251 | copy_to_user(&uarg->buf_size, &buf_size, sizeof(buf_size))) | |
2252 | ret = -EFAULT; | |
2253 | ||
ac8e9819 CM |
2254 | return ret; |
2255 | } | |
2256 | ||
98d377a0 | 2257 | /* |
ac8e9819 CM |
2258 | * Search INODE_REFs to identify path name of 'dirid' directory |
2259 | * in a 'tree_id' tree. and sets path name to 'name'. | |
2260 | */ | |
98d377a0 TH |
2261 | static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, |
2262 | u64 tree_id, u64 dirid, char *name) | |
2263 | { | |
2264 | struct btrfs_root *root; | |
2265 | struct btrfs_key key; | |
ac8e9819 | 2266 | char *ptr; |
98d377a0 TH |
2267 | int ret = -1; |
2268 | int slot; | |
2269 | int len; | |
2270 | int total_len = 0; | |
2271 | struct btrfs_inode_ref *iref; | |
2272 | struct extent_buffer *l; | |
2273 | struct btrfs_path *path; | |
2274 | ||
2275 | if (dirid == BTRFS_FIRST_FREE_OBJECTID) { | |
2276 | name[0]='\0'; | |
2277 | return 0; | |
2278 | } | |
2279 | ||
2280 | path = btrfs_alloc_path(); | |
2281 | if (!path) | |
2282 | return -ENOMEM; | |
2283 | ||
c8bcbfbd | 2284 | ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX - 1]; |
98d377a0 | 2285 | |
56e9357a | 2286 | root = btrfs_get_fs_root(info, tree_id, true); |
98d377a0 | 2287 | if (IS_ERR(root)) { |
ad1e3d56 | 2288 | ret = PTR_ERR(root); |
88234012 JB |
2289 | root = NULL; |
2290 | goto out; | |
2291 | } | |
98d377a0 TH |
2292 | |
2293 | key.objectid = dirid; | |
2294 | key.type = BTRFS_INODE_REF_KEY; | |
8ad6fcab | 2295 | key.offset = (u64)-1; |
98d377a0 | 2296 | |
67871254 | 2297 | while (1) { |
98d377a0 TH |
2298 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); |
2299 | if (ret < 0) | |
2300 | goto out; | |
18674c6c FDBM |
2301 | else if (ret > 0) { |
2302 | ret = btrfs_previous_item(root, path, dirid, | |
2303 | BTRFS_INODE_REF_KEY); | |
2304 | if (ret < 0) | |
2305 | goto out; | |
2306 | else if (ret > 0) { | |
2307 | ret = -ENOENT; | |
2308 | goto out; | |
2309 | } | |
2310 | } | |
98d377a0 TH |
2311 | |
2312 | l = path->nodes[0]; | |
2313 | slot = path->slots[0]; | |
2314 | btrfs_item_key_to_cpu(l, &key, slot); | |
2315 | ||
98d377a0 TH |
2316 | iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); |
2317 | len = btrfs_inode_ref_name_len(l, iref); | |
2318 | ptr -= len + 1; | |
2319 | total_len += len + 1; | |
a696cf35 FDBM |
2320 | if (ptr < name) { |
2321 | ret = -ENAMETOOLONG; | |
98d377a0 | 2322 | goto out; |
a696cf35 | 2323 | } |
98d377a0 TH |
2324 | |
2325 | *(ptr + len) = '/'; | |
67871254 | 2326 | read_extent_buffer(l, ptr, (unsigned long)(iref + 1), len); |
98d377a0 TH |
2327 | |
2328 | if (key.offset == BTRFS_FIRST_FREE_OBJECTID) | |
2329 | break; | |
2330 | ||
b3b4aa74 | 2331 | btrfs_release_path(path); |
98d377a0 | 2332 | key.objectid = key.offset; |
8ad6fcab | 2333 | key.offset = (u64)-1; |
98d377a0 | 2334 | dirid = key.objectid; |
98d377a0 | 2335 | } |
77906a50 | 2336 | memmove(name, ptr, total_len); |
67871254 | 2337 | name[total_len] = '\0'; |
98d377a0 TH |
2338 | ret = 0; |
2339 | out: | |
00246528 | 2340 | btrfs_put_root(root); |
98d377a0 | 2341 | btrfs_free_path(path); |
ac8e9819 CM |
2342 | return ret; |
2343 | } | |
2344 | ||
23d0b79d TM |
2345 | static int btrfs_search_path_in_tree_user(struct inode *inode, |
2346 | struct btrfs_ioctl_ino_lookup_user_args *args) | |
2347 | { | |
2348 | struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info; | |
2349 | struct super_block *sb = inode->i_sb; | |
2350 | struct btrfs_key upper_limit = BTRFS_I(inode)->location; | |
2351 | u64 treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2352 | u64 dirid = args->dirid; | |
2353 | unsigned long item_off; | |
2354 | unsigned long item_len; | |
2355 | struct btrfs_inode_ref *iref; | |
2356 | struct btrfs_root_ref *rref; | |
b8a49ae1 | 2357 | struct btrfs_root *root = NULL; |
23d0b79d TM |
2358 | struct btrfs_path *path; |
2359 | struct btrfs_key key, key2; | |
2360 | struct extent_buffer *leaf; | |
2361 | struct inode *temp_inode; | |
2362 | char *ptr; | |
2363 | int slot; | |
2364 | int len; | |
2365 | int total_len = 0; | |
2366 | int ret; | |
2367 | ||
2368 | path = btrfs_alloc_path(); | |
2369 | if (!path) | |
2370 | return -ENOMEM; | |
2371 | ||
2372 | /* | |
2373 | * If the bottom subvolume does not exist directly under upper_limit, | |
2374 | * construct the path in from the bottom up. | |
2375 | */ | |
2376 | if (dirid != upper_limit.objectid) { | |
2377 | ptr = &args->path[BTRFS_INO_LOOKUP_USER_PATH_MAX - 1]; | |
2378 | ||
56e9357a | 2379 | root = btrfs_get_fs_root(fs_info, treeid, true); |
23d0b79d TM |
2380 | if (IS_ERR(root)) { |
2381 | ret = PTR_ERR(root); | |
2382 | goto out; | |
2383 | } | |
2384 | ||
2385 | key.objectid = dirid; | |
2386 | key.type = BTRFS_INODE_REF_KEY; | |
2387 | key.offset = (u64)-1; | |
2388 | while (1) { | |
2389 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2390 | if (ret < 0) { | |
b8a49ae1 | 2391 | goto out_put; |
23d0b79d TM |
2392 | } else if (ret > 0) { |
2393 | ret = btrfs_previous_item(root, path, dirid, | |
2394 | BTRFS_INODE_REF_KEY); | |
2395 | if (ret < 0) { | |
b8a49ae1 | 2396 | goto out_put; |
23d0b79d TM |
2397 | } else if (ret > 0) { |
2398 | ret = -ENOENT; | |
b8a49ae1 | 2399 | goto out_put; |
23d0b79d TM |
2400 | } |
2401 | } | |
2402 | ||
2403 | leaf = path->nodes[0]; | |
2404 | slot = path->slots[0]; | |
2405 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2406 | ||
2407 | iref = btrfs_item_ptr(leaf, slot, struct btrfs_inode_ref); | |
2408 | len = btrfs_inode_ref_name_len(leaf, iref); | |
2409 | ptr -= len + 1; | |
2410 | total_len += len + 1; | |
2411 | if (ptr < args->path) { | |
2412 | ret = -ENAMETOOLONG; | |
b8a49ae1 | 2413 | goto out_put; |
23d0b79d TM |
2414 | } |
2415 | ||
2416 | *(ptr + len) = '/'; | |
2417 | read_extent_buffer(leaf, ptr, | |
2418 | (unsigned long)(iref + 1), len); | |
2419 | ||
2420 | /* Check the read+exec permission of this directory */ | |
2421 | ret = btrfs_previous_item(root, path, dirid, | |
2422 | BTRFS_INODE_ITEM_KEY); | |
2423 | if (ret < 0) { | |
b8a49ae1 | 2424 | goto out_put; |
23d0b79d TM |
2425 | } else if (ret > 0) { |
2426 | ret = -ENOENT; | |
b8a49ae1 | 2427 | goto out_put; |
23d0b79d TM |
2428 | } |
2429 | ||
2430 | leaf = path->nodes[0]; | |
2431 | slot = path->slots[0]; | |
2432 | btrfs_item_key_to_cpu(leaf, &key2, slot); | |
2433 | if (key2.objectid != dirid) { | |
2434 | ret = -ENOENT; | |
b8a49ae1 | 2435 | goto out_put; |
23d0b79d TM |
2436 | } |
2437 | ||
0202e83f | 2438 | temp_inode = btrfs_iget(sb, key2.objectid, root); |
3ca57bd6 MT |
2439 | if (IS_ERR(temp_inode)) { |
2440 | ret = PTR_ERR(temp_inode); | |
b8a49ae1 | 2441 | goto out_put; |
3ca57bd6 | 2442 | } |
47291baa CB |
2443 | ret = inode_permission(&init_user_ns, temp_inode, |
2444 | MAY_READ | MAY_EXEC); | |
23d0b79d TM |
2445 | iput(temp_inode); |
2446 | if (ret) { | |
2447 | ret = -EACCES; | |
b8a49ae1 | 2448 | goto out_put; |
23d0b79d TM |
2449 | } |
2450 | ||
2451 | if (key.offset == upper_limit.objectid) | |
2452 | break; | |
2453 | if (key.objectid == BTRFS_FIRST_FREE_OBJECTID) { | |
2454 | ret = -EACCES; | |
b8a49ae1 | 2455 | goto out_put; |
23d0b79d TM |
2456 | } |
2457 | ||
2458 | btrfs_release_path(path); | |
2459 | key.objectid = key.offset; | |
2460 | key.offset = (u64)-1; | |
2461 | dirid = key.objectid; | |
2462 | } | |
2463 | ||
2464 | memmove(args->path, ptr, total_len); | |
2465 | args->path[total_len] = '\0'; | |
00246528 | 2466 | btrfs_put_root(root); |
b8a49ae1 | 2467 | root = NULL; |
23d0b79d TM |
2468 | btrfs_release_path(path); |
2469 | } | |
2470 | ||
2471 | /* Get the bottom subvolume's name from ROOT_REF */ | |
23d0b79d TM |
2472 | key.objectid = treeid; |
2473 | key.type = BTRFS_ROOT_REF_KEY; | |
2474 | key.offset = args->treeid; | |
b8a49ae1 | 2475 | ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); |
23d0b79d TM |
2476 | if (ret < 0) { |
2477 | goto out; | |
2478 | } else if (ret > 0) { | |
2479 | ret = -ENOENT; | |
2480 | goto out; | |
2481 | } | |
2482 | ||
2483 | leaf = path->nodes[0]; | |
2484 | slot = path->slots[0]; | |
2485 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2486 | ||
2487 | item_off = btrfs_item_ptr_offset(leaf, slot); | |
2488 | item_len = btrfs_item_size_nr(leaf, slot); | |
2489 | /* Check if dirid in ROOT_REF corresponds to passed dirid */ | |
2490 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2491 | if (args->dirid != btrfs_root_ref_dirid(leaf, rref)) { | |
2492 | ret = -EINVAL; | |
2493 | goto out; | |
2494 | } | |
2495 | ||
2496 | /* Copy subvolume's name */ | |
2497 | item_off += sizeof(struct btrfs_root_ref); | |
2498 | item_len -= sizeof(struct btrfs_root_ref); | |
2499 | read_extent_buffer(leaf, args->name, item_off, item_len); | |
2500 | args->name[item_len] = 0; | |
2501 | ||
b8a49ae1 | 2502 | out_put: |
00246528 | 2503 | btrfs_put_root(root); |
23d0b79d TM |
2504 | out: |
2505 | btrfs_free_path(path); | |
2506 | return ret; | |
2507 | } | |
2508 | ||
ac8e9819 CM |
2509 | static noinline int btrfs_ioctl_ino_lookup(struct file *file, |
2510 | void __user *argp) | |
2511 | { | |
bece2e82 BVA |
2512 | struct btrfs_ioctl_ino_lookup_args *args; |
2513 | struct inode *inode; | |
01b810b8 | 2514 | int ret = 0; |
ac8e9819 | 2515 | |
2354d08f JL |
2516 | args = memdup_user(argp, sizeof(*args)); |
2517 | if (IS_ERR(args)) | |
2518 | return PTR_ERR(args); | |
c2b96929 | 2519 | |
496ad9aa | 2520 | inode = file_inode(file); |
ac8e9819 | 2521 | |
01b810b8 DS |
2522 | /* |
2523 | * Unprivileged query to obtain the containing subvolume root id. The | |
2524 | * path is reset so it's consistent with btrfs_search_path_in_tree. | |
2525 | */ | |
1b53ac4d CM |
2526 | if (args->treeid == 0) |
2527 | args->treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2528 | ||
01b810b8 DS |
2529 | if (args->objectid == BTRFS_FIRST_FREE_OBJECTID) { |
2530 | args->name[0] = 0; | |
2531 | goto out; | |
2532 | } | |
2533 | ||
2534 | if (!capable(CAP_SYS_ADMIN)) { | |
2535 | ret = -EPERM; | |
2536 | goto out; | |
2537 | } | |
2538 | ||
ac8e9819 CM |
2539 | ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, |
2540 | args->treeid, args->objectid, | |
2541 | args->name); | |
2542 | ||
01b810b8 | 2543 | out: |
ac8e9819 CM |
2544 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) |
2545 | ret = -EFAULT; | |
2546 | ||
2547 | kfree(args); | |
98d377a0 TH |
2548 | return ret; |
2549 | } | |
2550 | ||
23d0b79d TM |
2551 | /* |
2552 | * Version of ino_lookup ioctl (unprivileged) | |
2553 | * | |
2554 | * The main differences from ino_lookup ioctl are: | |
2555 | * | |
2556 | * 1. Read + Exec permission will be checked using inode_permission() during | |
2557 | * path construction. -EACCES will be returned in case of failure. | |
2558 | * 2. Path construction will be stopped at the inode number which corresponds | |
2559 | * to the fd with which this ioctl is called. If constructed path does not | |
2560 | * exist under fd's inode, -EACCES will be returned. | |
2561 | * 3. The name of bottom subvolume is also searched and filled. | |
2562 | */ | |
2563 | static int btrfs_ioctl_ino_lookup_user(struct file *file, void __user *argp) | |
2564 | { | |
2565 | struct btrfs_ioctl_ino_lookup_user_args *args; | |
2566 | struct inode *inode; | |
2567 | int ret; | |
2568 | ||
2569 | args = memdup_user(argp, sizeof(*args)); | |
2570 | if (IS_ERR(args)) | |
2571 | return PTR_ERR(args); | |
2572 | ||
2573 | inode = file_inode(file); | |
2574 | ||
2575 | if (args->dirid == BTRFS_FIRST_FREE_OBJECTID && | |
2576 | BTRFS_I(inode)->location.objectid != BTRFS_FIRST_FREE_OBJECTID) { | |
2577 | /* | |
2578 | * The subvolume does not exist under fd with which this is | |
2579 | * called | |
2580 | */ | |
2581 | kfree(args); | |
2582 | return -EACCES; | |
2583 | } | |
2584 | ||
2585 | ret = btrfs_search_path_in_tree_user(inode, args); | |
2586 | ||
2587 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
2588 | ret = -EFAULT; | |
2589 | ||
2590 | kfree(args); | |
2591 | return ret; | |
2592 | } | |
2593 | ||
b64ec075 TM |
2594 | /* Get the subvolume information in BTRFS_ROOT_ITEM and BTRFS_ROOT_BACKREF */ |
2595 | static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp) | |
2596 | { | |
2597 | struct btrfs_ioctl_get_subvol_info_args *subvol_info; | |
2598 | struct btrfs_fs_info *fs_info; | |
2599 | struct btrfs_root *root; | |
2600 | struct btrfs_path *path; | |
2601 | struct btrfs_key key; | |
2602 | struct btrfs_root_item *root_item; | |
2603 | struct btrfs_root_ref *rref; | |
2604 | struct extent_buffer *leaf; | |
2605 | unsigned long item_off; | |
2606 | unsigned long item_len; | |
2607 | struct inode *inode; | |
2608 | int slot; | |
2609 | int ret = 0; | |
2610 | ||
2611 | path = btrfs_alloc_path(); | |
2612 | if (!path) | |
2613 | return -ENOMEM; | |
2614 | ||
2615 | subvol_info = kzalloc(sizeof(*subvol_info), GFP_KERNEL); | |
2616 | if (!subvol_info) { | |
2617 | btrfs_free_path(path); | |
2618 | return -ENOMEM; | |
2619 | } | |
2620 | ||
2621 | inode = file_inode(file); | |
2622 | fs_info = BTRFS_I(inode)->root->fs_info; | |
2623 | ||
2624 | /* Get root_item of inode's subvolume */ | |
2625 | key.objectid = BTRFS_I(inode)->root->root_key.objectid; | |
56e9357a | 2626 | root = btrfs_get_fs_root(fs_info, key.objectid, true); |
b64ec075 TM |
2627 | if (IS_ERR(root)) { |
2628 | ret = PTR_ERR(root); | |
04734e84 JB |
2629 | goto out_free; |
2630 | } | |
b64ec075 TM |
2631 | root_item = &root->root_item; |
2632 | ||
2633 | subvol_info->treeid = key.objectid; | |
2634 | ||
2635 | subvol_info->generation = btrfs_root_generation(root_item); | |
2636 | subvol_info->flags = btrfs_root_flags(root_item); | |
2637 | ||
2638 | memcpy(subvol_info->uuid, root_item->uuid, BTRFS_UUID_SIZE); | |
2639 | memcpy(subvol_info->parent_uuid, root_item->parent_uuid, | |
2640 | BTRFS_UUID_SIZE); | |
2641 | memcpy(subvol_info->received_uuid, root_item->received_uuid, | |
2642 | BTRFS_UUID_SIZE); | |
2643 | ||
2644 | subvol_info->ctransid = btrfs_root_ctransid(root_item); | |
2645 | subvol_info->ctime.sec = btrfs_stack_timespec_sec(&root_item->ctime); | |
2646 | subvol_info->ctime.nsec = btrfs_stack_timespec_nsec(&root_item->ctime); | |
2647 | ||
2648 | subvol_info->otransid = btrfs_root_otransid(root_item); | |
2649 | subvol_info->otime.sec = btrfs_stack_timespec_sec(&root_item->otime); | |
2650 | subvol_info->otime.nsec = btrfs_stack_timespec_nsec(&root_item->otime); | |
2651 | ||
2652 | subvol_info->stransid = btrfs_root_stransid(root_item); | |
2653 | subvol_info->stime.sec = btrfs_stack_timespec_sec(&root_item->stime); | |
2654 | subvol_info->stime.nsec = btrfs_stack_timespec_nsec(&root_item->stime); | |
2655 | ||
2656 | subvol_info->rtransid = btrfs_root_rtransid(root_item); | |
2657 | subvol_info->rtime.sec = btrfs_stack_timespec_sec(&root_item->rtime); | |
2658 | subvol_info->rtime.nsec = btrfs_stack_timespec_nsec(&root_item->rtime); | |
2659 | ||
2660 | if (key.objectid != BTRFS_FS_TREE_OBJECTID) { | |
2661 | /* Search root tree for ROOT_BACKREF of this subvolume */ | |
b64ec075 TM |
2662 | key.type = BTRFS_ROOT_BACKREF_KEY; |
2663 | key.offset = 0; | |
04734e84 | 2664 | ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); |
b64ec075 TM |
2665 | if (ret < 0) { |
2666 | goto out; | |
2667 | } else if (path->slots[0] >= | |
2668 | btrfs_header_nritems(path->nodes[0])) { | |
04734e84 | 2669 | ret = btrfs_next_leaf(fs_info->tree_root, path); |
b64ec075 TM |
2670 | if (ret < 0) { |
2671 | goto out; | |
2672 | } else if (ret > 0) { | |
2673 | ret = -EUCLEAN; | |
2674 | goto out; | |
2675 | } | |
2676 | } | |
2677 | ||
2678 | leaf = path->nodes[0]; | |
2679 | slot = path->slots[0]; | |
2680 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2681 | if (key.objectid == subvol_info->treeid && | |
2682 | key.type == BTRFS_ROOT_BACKREF_KEY) { | |
2683 | subvol_info->parent_id = key.offset; | |
2684 | ||
2685 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2686 | subvol_info->dirid = btrfs_root_ref_dirid(leaf, rref); | |
2687 | ||
2688 | item_off = btrfs_item_ptr_offset(leaf, slot) | |
2689 | + sizeof(struct btrfs_root_ref); | |
2690 | item_len = btrfs_item_size_nr(leaf, slot) | |
2691 | - sizeof(struct btrfs_root_ref); | |
2692 | read_extent_buffer(leaf, subvol_info->name, | |
2693 | item_off, item_len); | |
2694 | } else { | |
2695 | ret = -ENOENT; | |
2696 | goto out; | |
2697 | } | |
2698 | } | |
2699 | ||
2700 | if (copy_to_user(argp, subvol_info, sizeof(*subvol_info))) | |
2701 | ret = -EFAULT; | |
2702 | ||
2703 | out: | |
00246528 | 2704 | btrfs_put_root(root); |
04734e84 | 2705 | out_free: |
b64ec075 | 2706 | btrfs_free_path(path); |
b091f7fe | 2707 | kfree(subvol_info); |
b64ec075 TM |
2708 | return ret; |
2709 | } | |
2710 | ||
42e4b520 TM |
2711 | /* |
2712 | * Return ROOT_REF information of the subvolume containing this inode | |
2713 | * except the subvolume name. | |
2714 | */ | |
2715 | static int btrfs_ioctl_get_subvol_rootref(struct file *file, void __user *argp) | |
2716 | { | |
2717 | struct btrfs_ioctl_get_subvol_rootref_args *rootrefs; | |
2718 | struct btrfs_root_ref *rref; | |
2719 | struct btrfs_root *root; | |
2720 | struct btrfs_path *path; | |
2721 | struct btrfs_key key; | |
2722 | struct extent_buffer *leaf; | |
2723 | struct inode *inode; | |
2724 | u64 objectid; | |
2725 | int slot; | |
2726 | int ret; | |
2727 | u8 found; | |
2728 | ||
2729 | path = btrfs_alloc_path(); | |
2730 | if (!path) | |
2731 | return -ENOMEM; | |
2732 | ||
2733 | rootrefs = memdup_user(argp, sizeof(*rootrefs)); | |
2734 | if (IS_ERR(rootrefs)) { | |
2735 | btrfs_free_path(path); | |
2736 | return PTR_ERR(rootrefs); | |
2737 | } | |
2738 | ||
2739 | inode = file_inode(file); | |
2740 | root = BTRFS_I(inode)->root->fs_info->tree_root; | |
2741 | objectid = BTRFS_I(inode)->root->root_key.objectid; | |
2742 | ||
2743 | key.objectid = objectid; | |
2744 | key.type = BTRFS_ROOT_REF_KEY; | |
2745 | key.offset = rootrefs->min_treeid; | |
2746 | found = 0; | |
2747 | ||
2748 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2749 | if (ret < 0) { | |
2750 | goto out; | |
2751 | } else if (path->slots[0] >= | |
2752 | btrfs_header_nritems(path->nodes[0])) { | |
2753 | ret = btrfs_next_leaf(root, path); | |
2754 | if (ret < 0) { | |
2755 | goto out; | |
2756 | } else if (ret > 0) { | |
2757 | ret = -EUCLEAN; | |
2758 | goto out; | |
2759 | } | |
2760 | } | |
2761 | while (1) { | |
2762 | leaf = path->nodes[0]; | |
2763 | slot = path->slots[0]; | |
2764 | ||
2765 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2766 | if (key.objectid != objectid || key.type != BTRFS_ROOT_REF_KEY) { | |
2767 | ret = 0; | |
2768 | goto out; | |
2769 | } | |
2770 | ||
2771 | if (found == BTRFS_MAX_ROOTREF_BUFFER_NUM) { | |
2772 | ret = -EOVERFLOW; | |
2773 | goto out; | |
2774 | } | |
2775 | ||
2776 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2777 | rootrefs->rootref[found].treeid = key.offset; | |
2778 | rootrefs->rootref[found].dirid = | |
2779 | btrfs_root_ref_dirid(leaf, rref); | |
2780 | found++; | |
2781 | ||
2782 | ret = btrfs_next_item(root, path); | |
2783 | if (ret < 0) { | |
2784 | goto out; | |
2785 | } else if (ret > 0) { | |
2786 | ret = -EUCLEAN; | |
2787 | goto out; | |
2788 | } | |
2789 | } | |
2790 | ||
2791 | out: | |
2792 | if (!ret || ret == -EOVERFLOW) { | |
2793 | rootrefs->num_items = found; | |
2794 | /* update min_treeid for next search */ | |
2795 | if (found) | |
2796 | rootrefs->min_treeid = | |
2797 | rootrefs->rootref[found - 1].treeid + 1; | |
2798 | if (copy_to_user(argp, rootrefs, sizeof(*rootrefs))) | |
2799 | ret = -EFAULT; | |
2800 | } | |
2801 | ||
2802 | kfree(rootrefs); | |
2803 | btrfs_free_path(path); | |
2804 | ||
2805 | return ret; | |
2806 | } | |
2807 | ||
76dda93c | 2808 | static noinline int btrfs_ioctl_snap_destroy(struct file *file, |
949964c9 MPS |
2809 | void __user *arg, |
2810 | bool destroy_v2) | |
76dda93c | 2811 | { |
54563d41 | 2812 | struct dentry *parent = file->f_path.dentry; |
0b246afa | 2813 | struct btrfs_fs_info *fs_info = btrfs_sb(parent->d_sb); |
76dda93c | 2814 | struct dentry *dentry; |
2b0143b5 | 2815 | struct inode *dir = d_inode(parent); |
76dda93c YZ |
2816 | struct inode *inode; |
2817 | struct btrfs_root *root = BTRFS_I(dir)->root; | |
2818 | struct btrfs_root *dest = NULL; | |
949964c9 MPS |
2819 | struct btrfs_ioctl_vol_args *vol_args = NULL; |
2820 | struct btrfs_ioctl_vol_args_v2 *vol_args2 = NULL; | |
2821 | char *subvol_name, *subvol_name_ptr = NULL; | |
2822 | int subvol_namelen; | |
76dda93c | 2823 | int err = 0; |
949964c9 | 2824 | bool destroy_parent = false; |
76dda93c | 2825 | |
949964c9 MPS |
2826 | if (destroy_v2) { |
2827 | vol_args2 = memdup_user(arg, sizeof(*vol_args2)); | |
2828 | if (IS_ERR(vol_args2)) | |
2829 | return PTR_ERR(vol_args2); | |
325c50e3 | 2830 | |
949964c9 MPS |
2831 | if (vol_args2->flags & ~BTRFS_SUBVOL_DELETE_ARGS_MASK) { |
2832 | err = -EOPNOTSUPP; | |
2833 | goto out; | |
2834 | } | |
76dda93c | 2835 | |
949964c9 MPS |
2836 | /* |
2837 | * If SPEC_BY_ID is not set, we are looking for the subvolume by | |
2838 | * name, same as v1 currently does. | |
2839 | */ | |
2840 | if (!(vol_args2->flags & BTRFS_SUBVOL_SPEC_BY_ID)) { | |
2841 | vol_args2->name[BTRFS_SUBVOL_NAME_MAX] = 0; | |
2842 | subvol_name = vol_args2->name; | |
2843 | ||
2844 | err = mnt_want_write_file(file); | |
2845 | if (err) | |
2846 | goto out; | |
2847 | } else { | |
2848 | if (vol_args2->subvolid < BTRFS_FIRST_FREE_OBJECTID) { | |
2849 | err = -EINVAL; | |
2850 | goto out; | |
2851 | } | |
2852 | ||
2853 | err = mnt_want_write_file(file); | |
2854 | if (err) | |
2855 | goto out; | |
2856 | ||
2857 | dentry = btrfs_get_dentry(fs_info->sb, | |
2858 | BTRFS_FIRST_FREE_OBJECTID, | |
2859 | vol_args2->subvolid, 0, 0); | |
2860 | if (IS_ERR(dentry)) { | |
2861 | err = PTR_ERR(dentry); | |
2862 | goto out_drop_write; | |
2863 | } | |
2864 | ||
2865 | /* | |
2866 | * Change the default parent since the subvolume being | |
2867 | * deleted can be outside of the current mount point. | |
2868 | */ | |
2869 | parent = btrfs_get_parent(dentry); | |
2870 | ||
2871 | /* | |
2872 | * At this point dentry->d_name can point to '/' if the | |
2873 | * subvolume we want to destroy is outsite of the | |
2874 | * current mount point, so we need to release the | |
2875 | * current dentry and execute the lookup to return a new | |
2876 | * one with ->d_name pointing to the | |
2877 | * <mount point>/subvol_name. | |
2878 | */ | |
2879 | dput(dentry); | |
2880 | if (IS_ERR(parent)) { | |
2881 | err = PTR_ERR(parent); | |
2882 | goto out_drop_write; | |
2883 | } | |
2884 | dir = d_inode(parent); | |
2885 | ||
2886 | /* | |
2887 | * If v2 was used with SPEC_BY_ID, a new parent was | |
2888 | * allocated since the subvolume can be outside of the | |
2889 | * current mount point. Later on we need to release this | |
2890 | * new parent dentry. | |
2891 | */ | |
2892 | destroy_parent = true; | |
2893 | ||
2894 | subvol_name_ptr = btrfs_get_subvol_name_from_objectid( | |
2895 | fs_info, vol_args2->subvolid); | |
2896 | if (IS_ERR(subvol_name_ptr)) { | |
2897 | err = PTR_ERR(subvol_name_ptr); | |
2898 | goto free_parent; | |
2899 | } | |
2900 | /* subvol_name_ptr is already NULL termined */ | |
2901 | subvol_name = (char *)kbasename(subvol_name_ptr); | |
2902 | } | |
2903 | } else { | |
2904 | vol_args = memdup_user(arg, sizeof(*vol_args)); | |
2905 | if (IS_ERR(vol_args)) | |
2906 | return PTR_ERR(vol_args); | |
2907 | ||
2908 | vol_args->name[BTRFS_PATH_NAME_MAX] = 0; | |
2909 | subvol_name = vol_args->name; | |
2910 | ||
2911 | err = mnt_want_write_file(file); | |
2912 | if (err) | |
2913 | goto out; | |
76dda93c YZ |
2914 | } |
2915 | ||
949964c9 | 2916 | subvol_namelen = strlen(subvol_name); |
76dda93c | 2917 | |
949964c9 MPS |
2918 | if (strchr(subvol_name, '/') || |
2919 | strncmp(subvol_name, "..", subvol_namelen) == 0) { | |
2920 | err = -EINVAL; | |
2921 | goto free_subvol_name; | |
2922 | } | |
2923 | ||
2924 | if (!S_ISDIR(dir->i_mode)) { | |
2925 | err = -ENOTDIR; | |
2926 | goto free_subvol_name; | |
2927 | } | |
521e0546 | 2928 | |
00235411 AV |
2929 | err = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT); |
2930 | if (err == -EINTR) | |
949964c9 MPS |
2931 | goto free_subvol_name; |
2932 | dentry = lookup_one_len(subvol_name, parent, subvol_namelen); | |
76dda93c YZ |
2933 | if (IS_ERR(dentry)) { |
2934 | err = PTR_ERR(dentry); | |
2935 | goto out_unlock_dir; | |
2936 | } | |
2937 | ||
2b0143b5 | 2938 | if (d_really_is_negative(dentry)) { |
76dda93c YZ |
2939 | err = -ENOENT; |
2940 | goto out_dput; | |
2941 | } | |
2942 | ||
2b0143b5 | 2943 | inode = d_inode(dentry); |
4260f7c7 | 2944 | dest = BTRFS_I(inode)->root; |
67871254 | 2945 | if (!capable(CAP_SYS_ADMIN)) { |
4260f7c7 SW |
2946 | /* |
2947 | * Regular user. Only allow this with a special mount | |
2948 | * option, when the user has write+exec access to the | |
2949 | * subvol root, and when rmdir(2) would have been | |
2950 | * allowed. | |
2951 | * | |
2952 | * Note that this is _not_ check that the subvol is | |
2953 | * empty or doesn't contain data that we wouldn't | |
2954 | * otherwise be able to delete. | |
2955 | * | |
2956 | * Users who want to delete empty subvols should try | |
2957 | * rmdir(2). | |
2958 | */ | |
2959 | err = -EPERM; | |
0b246afa | 2960 | if (!btrfs_test_opt(fs_info, USER_SUBVOL_RM_ALLOWED)) |
4260f7c7 SW |
2961 | goto out_dput; |
2962 | ||
2963 | /* | |
2964 | * Do not allow deletion if the parent dir is the same | |
2965 | * as the dir to be deleted. That means the ioctl | |
2966 | * must be called on the dentry referencing the root | |
2967 | * of the subvol, not a random directory contained | |
2968 | * within it. | |
2969 | */ | |
2970 | err = -EINVAL; | |
2971 | if (root == dest) | |
2972 | goto out_dput; | |
2973 | ||
47291baa CB |
2974 | err = inode_permission(&init_user_ns, inode, |
2975 | MAY_WRITE | MAY_EXEC); | |
4260f7c7 SW |
2976 | if (err) |
2977 | goto out_dput; | |
4260f7c7 SW |
2978 | } |
2979 | ||
5c39da5b MX |
2980 | /* check if subvolume may be deleted by a user */ |
2981 | err = btrfs_may_delete(dir, dentry, 1); | |
2982 | if (err) | |
2983 | goto out_dput; | |
2984 | ||
4a0cc7ca | 2985 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
76dda93c YZ |
2986 | err = -EINVAL; |
2987 | goto out_dput; | |
2988 | } | |
2989 | ||
64708539 | 2990 | btrfs_inode_lock(inode, 0); |
f60a2364 | 2991 | err = btrfs_delete_subvolume(dir, dentry); |
64708539 | 2992 | btrfs_inode_unlock(inode, 0); |
46008d9d AG |
2993 | if (!err) { |
2994 | fsnotify_rmdir(dir, dentry); | |
76dda93c | 2995 | d_delete(dentry); |
46008d9d | 2996 | } |
fa6ac876 | 2997 | |
76dda93c YZ |
2998 | out_dput: |
2999 | dput(dentry); | |
3000 | out_unlock_dir: | |
64708539 | 3001 | btrfs_inode_unlock(dir, 0); |
949964c9 MPS |
3002 | free_subvol_name: |
3003 | kfree(subvol_name_ptr); | |
3004 | free_parent: | |
3005 | if (destroy_parent) | |
3006 | dput(parent); | |
00235411 | 3007 | out_drop_write: |
2a79f17e | 3008 | mnt_drop_write_file(file); |
76dda93c | 3009 | out: |
949964c9 | 3010 | kfree(vol_args2); |
76dda93c YZ |
3011 | kfree(vol_args); |
3012 | return err; | |
3013 | } | |
3014 | ||
1e701a32 | 3015 | static int btrfs_ioctl_defrag(struct file *file, void __user *argp) |
f46b5a66 | 3016 | { |
496ad9aa | 3017 | struct inode *inode = file_inode(file); |
f46b5a66 | 3018 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1e701a32 | 3019 | struct btrfs_ioctl_defrag_range_args *range; |
c146afad YZ |
3020 | int ret; |
3021 | ||
25122d15 ID |
3022 | ret = mnt_want_write_file(file); |
3023 | if (ret) | |
3024 | return ret; | |
b83cc969 | 3025 | |
25122d15 ID |
3026 | if (btrfs_root_readonly(root)) { |
3027 | ret = -EROFS; | |
3028 | goto out; | |
5ac00add | 3029 | } |
f46b5a66 CH |
3030 | |
3031 | switch (inode->i_mode & S_IFMT) { | |
3032 | case S_IFDIR: | |
e441d54d CM |
3033 | if (!capable(CAP_SYS_ADMIN)) { |
3034 | ret = -EPERM; | |
3035 | goto out; | |
3036 | } | |
de78b51a | 3037 | ret = btrfs_defrag_root(root); |
f46b5a66 CH |
3038 | break; |
3039 | case S_IFREG: | |
616d374e AB |
3040 | /* |
3041 | * Note that this does not check the file descriptor for write | |
3042 | * access. This prevents defragmenting executables that are | |
3043 | * running and allows defrag on files open in read-only mode. | |
3044 | */ | |
3045 | if (!capable(CAP_SYS_ADMIN) && | |
47291baa | 3046 | inode_permission(&init_user_ns, inode, MAY_WRITE)) { |
616d374e | 3047 | ret = -EPERM; |
e441d54d CM |
3048 | goto out; |
3049 | } | |
1e701a32 CM |
3050 | |
3051 | range = kzalloc(sizeof(*range), GFP_KERNEL); | |
3052 | if (!range) { | |
3053 | ret = -ENOMEM; | |
3054 | goto out; | |
3055 | } | |
3056 | ||
3057 | if (argp) { | |
3058 | if (copy_from_user(range, argp, | |
3059 | sizeof(*range))) { | |
3060 | ret = -EFAULT; | |
3061 | kfree(range); | |
683be16e | 3062 | goto out; |
1e701a32 CM |
3063 | } |
3064 | /* compression requires us to start the IO */ | |
3065 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
3066 | range->flags |= BTRFS_DEFRAG_RANGE_START_IO; | |
3067 | range->extent_thresh = (u32)-1; | |
3068 | } | |
3069 | } else { | |
3070 | /* the rest are all set to zero by kzalloc */ | |
3071 | range->len = (u64)-1; | |
3072 | } | |
496ad9aa | 3073 | ret = btrfs_defrag_file(file_inode(file), file, |
7c829b72 | 3074 | range, BTRFS_OLDEST_GENERATION, 0); |
4cb5300b CM |
3075 | if (ret > 0) |
3076 | ret = 0; | |
1e701a32 | 3077 | kfree(range); |
f46b5a66 | 3078 | break; |
8929ecfa YZ |
3079 | default: |
3080 | ret = -EINVAL; | |
f46b5a66 | 3081 | } |
e441d54d | 3082 | out: |
25122d15 | 3083 | mnt_drop_write_file(file); |
e441d54d | 3084 | return ret; |
f46b5a66 CH |
3085 | } |
3086 | ||
2ff7e61e | 3087 | static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg) |
f46b5a66 CH |
3088 | { |
3089 | struct btrfs_ioctl_vol_args *vol_args; | |
3090 | int ret; | |
3091 | ||
e441d54d CM |
3092 | if (!capable(CAP_SYS_ADMIN)) |
3093 | return -EPERM; | |
3094 | ||
c3e1f96c | 3095 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_ADD)) |
e57138b3 | 3096 | return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
c9e9f97b | 3097 | |
dae7b665 | 3098 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
3099 | if (IS_ERR(vol_args)) { |
3100 | ret = PTR_ERR(vol_args); | |
3101 | goto out; | |
3102 | } | |
f46b5a66 | 3103 | |
5516e595 | 3104 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
2ff7e61e | 3105 | ret = btrfs_init_new_device(fs_info, vol_args->name); |
f46b5a66 | 3106 | |
43d20761 | 3107 | if (!ret) |
0b246afa | 3108 | btrfs_info(fs_info, "disk added %s", vol_args->name); |
43d20761 | 3109 | |
f46b5a66 | 3110 | kfree(vol_args); |
c9e9f97b | 3111 | out: |
c3e1f96c | 3112 | btrfs_exclop_finish(fs_info); |
f46b5a66 CH |
3113 | return ret; |
3114 | } | |
3115 | ||
6b526ed7 | 3116 | static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg) |
f46b5a66 | 3117 | { |
0b246afa JM |
3118 | struct inode *inode = file_inode(file); |
3119 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
6b526ed7 | 3120 | struct btrfs_ioctl_vol_args_v2 *vol_args; |
f46b5a66 CH |
3121 | int ret; |
3122 | ||
e441d54d CM |
3123 | if (!capable(CAP_SYS_ADMIN)) |
3124 | return -EPERM; | |
3125 | ||
da24927b MX |
3126 | ret = mnt_want_write_file(file); |
3127 | if (ret) | |
3128 | return ret; | |
c146afad | 3129 | |
dae7b665 | 3130 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
3131 | if (IS_ERR(vol_args)) { |
3132 | ret = PTR_ERR(vol_args); | |
c47ca32d | 3133 | goto err_drop; |
c9e9f97b | 3134 | } |
f46b5a66 | 3135 | |
748449cd | 3136 | if (vol_args->flags & ~BTRFS_DEVICE_REMOVE_ARGS_MASK) { |
fd4e994b OS |
3137 | ret = -EOPNOTSUPP; |
3138 | goto out; | |
3139 | } | |
f46b5a66 | 3140 | |
c3e1f96c | 3141 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REMOVE)) { |
183860f6 AJ |
3142 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
3143 | goto out; | |
3144 | } | |
3145 | ||
735654ea | 3146 | if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) { |
2ff7e61e | 3147 | ret = btrfs_rm_device(fs_info, NULL, vol_args->devid); |
6b526ed7 AJ |
3148 | } else { |
3149 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
2ff7e61e | 3150 | ret = btrfs_rm_device(fs_info, vol_args->name, 0); |
6b526ed7 | 3151 | } |
c3e1f96c | 3152 | btrfs_exclop_finish(fs_info); |
183860f6 | 3153 | |
6b526ed7 | 3154 | if (!ret) { |
735654ea | 3155 | if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) |
0b246afa | 3156 | btrfs_info(fs_info, "device deleted: id %llu", |
6b526ed7 AJ |
3157 | vol_args->devid); |
3158 | else | |
0b246afa | 3159 | btrfs_info(fs_info, "device deleted: %s", |
6b526ed7 AJ |
3160 | vol_args->name); |
3161 | } | |
183860f6 AJ |
3162 | out: |
3163 | kfree(vol_args); | |
c47ca32d | 3164 | err_drop: |
4ac20c70 | 3165 | mnt_drop_write_file(file); |
f46b5a66 CH |
3166 | return ret; |
3167 | } | |
3168 | ||
da24927b | 3169 | static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg) |
f46b5a66 | 3170 | { |
0b246afa JM |
3171 | struct inode *inode = file_inode(file); |
3172 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
f46b5a66 CH |
3173 | struct btrfs_ioctl_vol_args *vol_args; |
3174 | int ret; | |
3175 | ||
e441d54d CM |
3176 | if (!capable(CAP_SYS_ADMIN)) |
3177 | return -EPERM; | |
3178 | ||
da24927b MX |
3179 | ret = mnt_want_write_file(file); |
3180 | if (ret) | |
3181 | return ret; | |
c146afad | 3182 | |
c3e1f96c | 3183 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REMOVE)) { |
183860f6 | 3184 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
58d7bbf8 DS |
3185 | goto out_drop_write; |
3186 | } | |
3187 | ||
3188 | vol_args = memdup_user(arg, sizeof(*vol_args)); | |
3189 | if (IS_ERR(vol_args)) { | |
3190 | ret = PTR_ERR(vol_args); | |
183860f6 AJ |
3191 | goto out; |
3192 | } | |
3193 | ||
58d7bbf8 | 3194 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
2ff7e61e | 3195 | ret = btrfs_rm_device(fs_info, vol_args->name, 0); |
183860f6 | 3196 | |
ec95d491 | 3197 | if (!ret) |
0b246afa | 3198 | btrfs_info(fs_info, "disk deleted %s", vol_args->name); |
183860f6 | 3199 | kfree(vol_args); |
58d7bbf8 | 3200 | out: |
c3e1f96c | 3201 | btrfs_exclop_finish(fs_info); |
58d7bbf8 | 3202 | out_drop_write: |
4ac20c70 | 3203 | mnt_drop_write_file(file); |
58d7bbf8 | 3204 | |
f46b5a66 CH |
3205 | return ret; |
3206 | } | |
3207 | ||
2ff7e61e JM |
3208 | static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info, |
3209 | void __user *arg) | |
475f6387 | 3210 | { |
027ed2f0 | 3211 | struct btrfs_ioctl_fs_info_args *fi_args; |
475f6387 | 3212 | struct btrfs_device *device; |
0b246afa | 3213 | struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; |
137c5418 | 3214 | u64 flags_in; |
027ed2f0 | 3215 | int ret = 0; |
475f6387 | 3216 | |
137c5418 JT |
3217 | fi_args = memdup_user(arg, sizeof(*fi_args)); |
3218 | if (IS_ERR(fi_args)) | |
3219 | return PTR_ERR(fi_args); | |
3220 | ||
3221 | flags_in = fi_args->flags; | |
3222 | memset(fi_args, 0, sizeof(*fi_args)); | |
027ed2f0 | 3223 | |
d03262c7 | 3224 | rcu_read_lock(); |
027ed2f0 | 3225 | fi_args->num_devices = fs_devices->num_devices; |
475f6387 | 3226 | |
d03262c7 | 3227 | list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) { |
027ed2f0 LZ |
3228 | if (device->devid > fi_args->max_id) |
3229 | fi_args->max_id = device->devid; | |
475f6387 | 3230 | } |
d03262c7 | 3231 | rcu_read_unlock(); |
475f6387 | 3232 | |
de37aa51 | 3233 | memcpy(&fi_args->fsid, fs_devices->fsid, sizeof(fi_args->fsid)); |
bea7eafd OS |
3234 | fi_args->nodesize = fs_info->nodesize; |
3235 | fi_args->sectorsize = fs_info->sectorsize; | |
3236 | fi_args->clone_alignment = fs_info->sectorsize; | |
80a773fb | 3237 | |
137c5418 JT |
3238 | if (flags_in & BTRFS_FS_INFO_FLAG_CSUM_INFO) { |
3239 | fi_args->csum_type = btrfs_super_csum_type(fs_info->super_copy); | |
3240 | fi_args->csum_size = btrfs_super_csum_size(fs_info->super_copy); | |
3241 | fi_args->flags |= BTRFS_FS_INFO_FLAG_CSUM_INFO; | |
3242 | } | |
3243 | ||
0fb408a5 JT |
3244 | if (flags_in & BTRFS_FS_INFO_FLAG_GENERATION) { |
3245 | fi_args->generation = fs_info->generation; | |
3246 | fi_args->flags |= BTRFS_FS_INFO_FLAG_GENERATION; | |
3247 | } | |
3248 | ||
49bac897 JT |
3249 | if (flags_in & BTRFS_FS_INFO_FLAG_METADATA_UUID) { |
3250 | memcpy(&fi_args->metadata_uuid, fs_devices->metadata_uuid, | |
3251 | sizeof(fi_args->metadata_uuid)); | |
3252 | fi_args->flags |= BTRFS_FS_INFO_FLAG_METADATA_UUID; | |
3253 | } | |
3254 | ||
027ed2f0 LZ |
3255 | if (copy_to_user(arg, fi_args, sizeof(*fi_args))) |
3256 | ret = -EFAULT; | |
475f6387 | 3257 | |
027ed2f0 LZ |
3258 | kfree(fi_args); |
3259 | return ret; | |
475f6387 JS |
3260 | } |
3261 | ||
2ff7e61e JM |
3262 | static long btrfs_ioctl_dev_info(struct btrfs_fs_info *fs_info, |
3263 | void __user *arg) | |
475f6387 JS |
3264 | { |
3265 | struct btrfs_ioctl_dev_info_args *di_args; | |
3266 | struct btrfs_device *dev; | |
475f6387 JS |
3267 | int ret = 0; |
3268 | char *s_uuid = NULL; | |
475f6387 | 3269 | |
475f6387 JS |
3270 | di_args = memdup_user(arg, sizeof(*di_args)); |
3271 | if (IS_ERR(di_args)) | |
3272 | return PTR_ERR(di_args); | |
3273 | ||
dd5f9615 | 3274 | if (!btrfs_is_empty_uuid(di_args->uuid)) |
475f6387 JS |
3275 | s_uuid = di_args->uuid; |
3276 | ||
c5593ca3 | 3277 | rcu_read_lock(); |
e4319cd9 | 3278 | dev = btrfs_find_device(fs_info->fs_devices, di_args->devid, s_uuid, |
b2598edf | 3279 | NULL); |
475f6387 JS |
3280 | |
3281 | if (!dev) { | |
3282 | ret = -ENODEV; | |
3283 | goto out; | |
3284 | } | |
3285 | ||
3286 | di_args->devid = dev->devid; | |
7cc8e58d MX |
3287 | di_args->bytes_used = btrfs_device_get_bytes_used(dev); |
3288 | di_args->total_bytes = btrfs_device_get_total_bytes(dev); | |
475f6387 | 3289 | memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid)); |
a27202fb | 3290 | if (dev->name) { |
672d5990 MT |
3291 | strncpy(di_args->path, rcu_str_deref(dev->name), |
3292 | sizeof(di_args->path) - 1); | |
a27202fb JM |
3293 | di_args->path[sizeof(di_args->path) - 1] = 0; |
3294 | } else { | |
99ba55ad | 3295 | di_args->path[0] = '\0'; |
a27202fb | 3296 | } |
475f6387 JS |
3297 | |
3298 | out: | |
c5593ca3 | 3299 | rcu_read_unlock(); |
475f6387 JS |
3300 | if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args))) |
3301 | ret = -EFAULT; | |
3302 | ||
3303 | kfree(di_args); | |
3304 | return ret; | |
3305 | } | |
3306 | ||
6ef5ed0d JB |
3307 | static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) |
3308 | { | |
496ad9aa | 3309 | struct inode *inode = file_inode(file); |
0b246afa | 3310 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
6ef5ed0d JB |
3311 | struct btrfs_root *root = BTRFS_I(inode)->root; |
3312 | struct btrfs_root *new_root; | |
3313 | struct btrfs_dir_item *di; | |
3314 | struct btrfs_trans_handle *trans; | |
2a2b5d62 | 3315 | struct btrfs_path *path = NULL; |
6ef5ed0d | 3316 | struct btrfs_disk_key disk_key; |
6ef5ed0d JB |
3317 | u64 objectid = 0; |
3318 | u64 dir_id; | |
3c04ce01 | 3319 | int ret; |
6ef5ed0d JB |
3320 | |
3321 | if (!capable(CAP_SYS_ADMIN)) | |
3322 | return -EPERM; | |
3323 | ||
3c04ce01 MX |
3324 | ret = mnt_want_write_file(file); |
3325 | if (ret) | |
3326 | return ret; | |
3327 | ||
3328 | if (copy_from_user(&objectid, argp, sizeof(objectid))) { | |
3329 | ret = -EFAULT; | |
3330 | goto out; | |
3331 | } | |
6ef5ed0d JB |
3332 | |
3333 | if (!objectid) | |
1cecf579 | 3334 | objectid = BTRFS_FS_TREE_OBJECTID; |
6ef5ed0d | 3335 | |
56e9357a | 3336 | new_root = btrfs_get_fs_root(fs_info, objectid, true); |
3c04ce01 MX |
3337 | if (IS_ERR(new_root)) { |
3338 | ret = PTR_ERR(new_root); | |
3339 | goto out; | |
3340 | } | |
2a2b5d62 JB |
3341 | if (!is_fstree(new_root->root_key.objectid)) { |
3342 | ret = -ENOENT; | |
3343 | goto out_free; | |
3344 | } | |
6ef5ed0d | 3345 | |
6ef5ed0d | 3346 | path = btrfs_alloc_path(); |
3c04ce01 MX |
3347 | if (!path) { |
3348 | ret = -ENOMEM; | |
2a2b5d62 | 3349 | goto out_free; |
3c04ce01 | 3350 | } |
6ef5ed0d JB |
3351 | |
3352 | trans = btrfs_start_transaction(root, 1); | |
98d5dc13 | 3353 | if (IS_ERR(trans)) { |
3c04ce01 | 3354 | ret = PTR_ERR(trans); |
2a2b5d62 | 3355 | goto out_free; |
6ef5ed0d JB |
3356 | } |
3357 | ||
0b246afa JM |
3358 | dir_id = btrfs_super_root_dir(fs_info->super_copy); |
3359 | di = btrfs_lookup_dir_item(trans, fs_info->tree_root, path, | |
6ef5ed0d | 3360 | dir_id, "default", 7, 1); |
cf1e99a4 | 3361 | if (IS_ERR_OR_NULL(di)) { |
2a2b5d62 | 3362 | btrfs_release_path(path); |
3a45bb20 | 3363 | btrfs_end_transaction(trans); |
0b246afa | 3364 | btrfs_err(fs_info, |
5d163e0e | 3365 | "Umm, you don't have the default diritem, this isn't going to work"); |
3c04ce01 | 3366 | ret = -ENOENT; |
2a2b5d62 | 3367 | goto out_free; |
6ef5ed0d JB |
3368 | } |
3369 | ||
3370 | btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key); | |
3371 | btrfs_set_dir_item_key(path->nodes[0], di, &disk_key); | |
3372 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
2a2b5d62 | 3373 | btrfs_release_path(path); |
6ef5ed0d | 3374 | |
0b246afa | 3375 | btrfs_set_fs_incompat(fs_info, DEFAULT_SUBVOL); |
3a45bb20 | 3376 | btrfs_end_transaction(trans); |
2a2b5d62 | 3377 | out_free: |
00246528 | 3378 | btrfs_put_root(new_root); |
2a2b5d62 | 3379 | btrfs_free_path(path); |
3c04ce01 MX |
3380 | out: |
3381 | mnt_drop_write_file(file); | |
3382 | return ret; | |
6ef5ed0d JB |
3383 | } |
3384 | ||
c065f5b1 SY |
3385 | static void get_block_group_info(struct list_head *groups_list, |
3386 | struct btrfs_ioctl_space_info *space) | |
bf5fc093 | 3387 | { |
32da5386 | 3388 | struct btrfs_block_group *block_group; |
bf5fc093 JB |
3389 | |
3390 | space->total_bytes = 0; | |
3391 | space->used_bytes = 0; | |
3392 | space->flags = 0; | |
3393 | list_for_each_entry(block_group, groups_list, list) { | |
3394 | space->flags = block_group->flags; | |
b3470b5d | 3395 | space->total_bytes += block_group->length; |
bf38be65 | 3396 | space->used_bytes += block_group->used; |
bf5fc093 JB |
3397 | } |
3398 | } | |
3399 | ||
2ff7e61e JM |
3400 | static long btrfs_ioctl_space_info(struct btrfs_fs_info *fs_info, |
3401 | void __user *arg) | |
1406e432 JB |
3402 | { |
3403 | struct btrfs_ioctl_space_args space_args; | |
3404 | struct btrfs_ioctl_space_info space; | |
3405 | struct btrfs_ioctl_space_info *dest; | |
7fde62bf | 3406 | struct btrfs_ioctl_space_info *dest_orig; |
13f2696f | 3407 | struct btrfs_ioctl_space_info __user *user_dest; |
1406e432 | 3408 | struct btrfs_space_info *info; |
315d8e98 CIK |
3409 | static const u64 types[] = { |
3410 | BTRFS_BLOCK_GROUP_DATA, | |
3411 | BTRFS_BLOCK_GROUP_SYSTEM, | |
3412 | BTRFS_BLOCK_GROUP_METADATA, | |
3413 | BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA | |
3414 | }; | |
bf5fc093 | 3415 | int num_types = 4; |
7fde62bf | 3416 | int alloc_size; |
1406e432 | 3417 | int ret = 0; |
51788b1b | 3418 | u64 slot_count = 0; |
bf5fc093 | 3419 | int i, c; |
1406e432 JB |
3420 | |
3421 | if (copy_from_user(&space_args, | |
3422 | (struct btrfs_ioctl_space_args __user *)arg, | |
3423 | sizeof(space_args))) | |
3424 | return -EFAULT; | |
3425 | ||
bf5fc093 JB |
3426 | for (i = 0; i < num_types; i++) { |
3427 | struct btrfs_space_info *tmp; | |
3428 | ||
3429 | info = NULL; | |
72804905 | 3430 | list_for_each_entry(tmp, &fs_info->space_info, list) { |
bf5fc093 JB |
3431 | if (tmp->flags == types[i]) { |
3432 | info = tmp; | |
3433 | break; | |
3434 | } | |
3435 | } | |
bf5fc093 JB |
3436 | |
3437 | if (!info) | |
3438 | continue; | |
3439 | ||
3440 | down_read(&info->groups_sem); | |
3441 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3442 | if (!list_empty(&info->block_groups[c])) | |
3443 | slot_count++; | |
3444 | } | |
3445 | up_read(&info->groups_sem); | |
3446 | } | |
7fde62bf | 3447 | |
36523e95 DS |
3448 | /* |
3449 | * Global block reserve, exported as a space_info | |
3450 | */ | |
3451 | slot_count++; | |
3452 | ||
7fde62bf CM |
3453 | /* space_slots == 0 means they are asking for a count */ |
3454 | if (space_args.space_slots == 0) { | |
3455 | space_args.total_spaces = slot_count; | |
3456 | goto out; | |
3457 | } | |
bf5fc093 | 3458 | |
51788b1b | 3459 | slot_count = min_t(u64, space_args.space_slots, slot_count); |
bf5fc093 | 3460 | |
7fde62bf | 3461 | alloc_size = sizeof(*dest) * slot_count; |
bf5fc093 | 3462 | |
7fde62bf CM |
3463 | /* we generally have at most 6 or so space infos, one for each raid |
3464 | * level. So, a whole page should be more than enough for everyone | |
3465 | */ | |
09cbfeaf | 3466 | if (alloc_size > PAGE_SIZE) |
7fde62bf CM |
3467 | return -ENOMEM; |
3468 | ||
1406e432 | 3469 | space_args.total_spaces = 0; |
8d2db785 | 3470 | dest = kmalloc(alloc_size, GFP_KERNEL); |
7fde62bf CM |
3471 | if (!dest) |
3472 | return -ENOMEM; | |
3473 | dest_orig = dest; | |
1406e432 | 3474 | |
7fde62bf | 3475 | /* now we have a buffer to copy into */ |
bf5fc093 JB |
3476 | for (i = 0; i < num_types; i++) { |
3477 | struct btrfs_space_info *tmp; | |
3478 | ||
51788b1b DR |
3479 | if (!slot_count) |
3480 | break; | |
3481 | ||
bf5fc093 | 3482 | info = NULL; |
72804905 | 3483 | list_for_each_entry(tmp, &fs_info->space_info, list) { |
bf5fc093 JB |
3484 | if (tmp->flags == types[i]) { |
3485 | info = tmp; | |
3486 | break; | |
3487 | } | |
3488 | } | |
7fde62bf | 3489 | |
bf5fc093 JB |
3490 | if (!info) |
3491 | continue; | |
3492 | down_read(&info->groups_sem); | |
3493 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3494 | if (!list_empty(&info->block_groups[c])) { | |
c065f5b1 SY |
3495 | get_block_group_info(&info->block_groups[c], |
3496 | &space); | |
bf5fc093 JB |
3497 | memcpy(dest, &space, sizeof(space)); |
3498 | dest++; | |
3499 | space_args.total_spaces++; | |
51788b1b | 3500 | slot_count--; |
bf5fc093 | 3501 | } |
51788b1b DR |
3502 | if (!slot_count) |
3503 | break; | |
bf5fc093 JB |
3504 | } |
3505 | up_read(&info->groups_sem); | |
1406e432 | 3506 | } |
1406e432 | 3507 | |
36523e95 DS |
3508 | /* |
3509 | * Add global block reserve | |
3510 | */ | |
3511 | if (slot_count) { | |
0b246afa | 3512 | struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv; |
36523e95 DS |
3513 | |
3514 | spin_lock(&block_rsv->lock); | |
3515 | space.total_bytes = block_rsv->size; | |
3516 | space.used_bytes = block_rsv->size - block_rsv->reserved; | |
3517 | spin_unlock(&block_rsv->lock); | |
3518 | space.flags = BTRFS_SPACE_INFO_GLOBAL_RSV; | |
3519 | memcpy(dest, &space, sizeof(space)); | |
3520 | space_args.total_spaces++; | |
3521 | } | |
3522 | ||
2eec6c81 | 3523 | user_dest = (struct btrfs_ioctl_space_info __user *) |
7fde62bf CM |
3524 | (arg + sizeof(struct btrfs_ioctl_space_args)); |
3525 | ||
3526 | if (copy_to_user(user_dest, dest_orig, alloc_size)) | |
3527 | ret = -EFAULT; | |
3528 | ||
3529 | kfree(dest_orig); | |
3530 | out: | |
3531 | if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args))) | |
1406e432 JB |
3532 | ret = -EFAULT; |
3533 | ||
3534 | return ret; | |
3535 | } | |
3536 | ||
9a8c28be MX |
3537 | static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root, |
3538 | void __user *argp) | |
46204592 | 3539 | { |
46204592 SW |
3540 | struct btrfs_trans_handle *trans; |
3541 | u64 transid; | |
db5b493a | 3542 | int ret; |
46204592 | 3543 | |
d4edf39b | 3544 | trans = btrfs_attach_transaction_barrier(root); |
ff7c1d33 MX |
3545 | if (IS_ERR(trans)) { |
3546 | if (PTR_ERR(trans) != -ENOENT) | |
3547 | return PTR_ERR(trans); | |
3548 | ||
3549 | /* No running transaction, don't bother */ | |
3550 | transid = root->fs_info->last_trans_committed; | |
3551 | goto out; | |
3552 | } | |
46204592 | 3553 | transid = trans->transid; |
3a45bb20 | 3554 | ret = btrfs_commit_transaction_async(trans, 0); |
8b2b2d3c | 3555 | if (ret) { |
3a45bb20 | 3556 | btrfs_end_transaction(trans); |
db5b493a | 3557 | return ret; |
8b2b2d3c | 3558 | } |
ff7c1d33 | 3559 | out: |
46204592 SW |
3560 | if (argp) |
3561 | if (copy_to_user(argp, &transid, sizeof(transid))) | |
3562 | return -EFAULT; | |
3563 | return 0; | |
3564 | } | |
3565 | ||
2ff7e61e | 3566 | static noinline long btrfs_ioctl_wait_sync(struct btrfs_fs_info *fs_info, |
9a8c28be | 3567 | void __user *argp) |
46204592 | 3568 | { |
46204592 SW |
3569 | u64 transid; |
3570 | ||
3571 | if (argp) { | |
3572 | if (copy_from_user(&transid, argp, sizeof(transid))) | |
3573 | return -EFAULT; | |
3574 | } else { | |
3575 | transid = 0; /* current trans */ | |
3576 | } | |
2ff7e61e | 3577 | return btrfs_wait_for_commit(fs_info, transid); |
46204592 SW |
3578 | } |
3579 | ||
b8e95489 | 3580 | static long btrfs_ioctl_scrub(struct file *file, void __user *arg) |
475f6387 | 3581 | { |
0b246afa | 3582 | struct btrfs_fs_info *fs_info = btrfs_sb(file_inode(file)->i_sb); |
475f6387 | 3583 | struct btrfs_ioctl_scrub_args *sa; |
b8e95489 | 3584 | int ret; |
475f6387 JS |
3585 | |
3586 | if (!capable(CAP_SYS_ADMIN)) | |
3587 | return -EPERM; | |
3588 | ||
3589 | sa = memdup_user(arg, sizeof(*sa)); | |
3590 | if (IS_ERR(sa)) | |
3591 | return PTR_ERR(sa); | |
3592 | ||
b8e95489 MX |
3593 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) { |
3594 | ret = mnt_want_write_file(file); | |
3595 | if (ret) | |
3596 | goto out; | |
3597 | } | |
3598 | ||
0b246afa | 3599 | ret = btrfs_scrub_dev(fs_info, sa->devid, sa->start, sa->end, |
63a212ab SB |
3600 | &sa->progress, sa->flags & BTRFS_SCRUB_READONLY, |
3601 | 0); | |
475f6387 | 3602 | |
5afe6ce7 FM |
3603 | /* |
3604 | * Copy scrub args to user space even if btrfs_scrub_dev() returned an | |
3605 | * error. This is important as it allows user space to know how much | |
3606 | * progress scrub has done. For example, if scrub is canceled we get | |
3607 | * -ECANCELED from btrfs_scrub_dev() and return that error back to user | |
3608 | * space. Later user space can inspect the progress from the structure | |
3609 | * btrfs_ioctl_scrub_args and resume scrub from where it left off | |
3610 | * previously (btrfs-progs does this). | |
3611 | * If we fail to copy the btrfs_ioctl_scrub_args structure to user space | |
3612 | * then return -EFAULT to signal the structure was not copied or it may | |
3613 | * be corrupt and unreliable due to a partial copy. | |
3614 | */ | |
3615 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
475f6387 JS |
3616 | ret = -EFAULT; |
3617 | ||
b8e95489 MX |
3618 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) |
3619 | mnt_drop_write_file(file); | |
3620 | out: | |
475f6387 JS |
3621 | kfree(sa); |
3622 | return ret; | |
3623 | } | |
3624 | ||
2ff7e61e | 3625 | static long btrfs_ioctl_scrub_cancel(struct btrfs_fs_info *fs_info) |
475f6387 JS |
3626 | { |
3627 | if (!capable(CAP_SYS_ADMIN)) | |
3628 | return -EPERM; | |
3629 | ||
2ff7e61e | 3630 | return btrfs_scrub_cancel(fs_info); |
475f6387 JS |
3631 | } |
3632 | ||
2ff7e61e | 3633 | static long btrfs_ioctl_scrub_progress(struct btrfs_fs_info *fs_info, |
475f6387 JS |
3634 | void __user *arg) |
3635 | { | |
3636 | struct btrfs_ioctl_scrub_args *sa; | |
3637 | int ret; | |
3638 | ||
3639 | if (!capable(CAP_SYS_ADMIN)) | |
3640 | return -EPERM; | |
3641 | ||
3642 | sa = memdup_user(arg, sizeof(*sa)); | |
3643 | if (IS_ERR(sa)) | |
3644 | return PTR_ERR(sa); | |
3645 | ||
2ff7e61e | 3646 | ret = btrfs_scrub_progress(fs_info, sa->devid, &sa->progress); |
475f6387 | 3647 | |
4fa99b00 | 3648 | if (ret == 0 && copy_to_user(arg, sa, sizeof(*sa))) |
475f6387 JS |
3649 | ret = -EFAULT; |
3650 | ||
3651 | kfree(sa); | |
3652 | return ret; | |
3653 | } | |
3654 | ||
2ff7e61e | 3655 | static long btrfs_ioctl_get_dev_stats(struct btrfs_fs_info *fs_info, |
b27f7c0c | 3656 | void __user *arg) |
c11d2c23 SB |
3657 | { |
3658 | struct btrfs_ioctl_get_dev_stats *sa; | |
3659 | int ret; | |
3660 | ||
c11d2c23 SB |
3661 | sa = memdup_user(arg, sizeof(*sa)); |
3662 | if (IS_ERR(sa)) | |
3663 | return PTR_ERR(sa); | |
3664 | ||
b27f7c0c DS |
3665 | if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) { |
3666 | kfree(sa); | |
3667 | return -EPERM; | |
3668 | } | |
3669 | ||
2ff7e61e | 3670 | ret = btrfs_get_dev_stats(fs_info, sa); |
c11d2c23 | 3671 | |
eee99577 | 3672 | if (ret == 0 && copy_to_user(arg, sa, sizeof(*sa))) |
c11d2c23 SB |
3673 | ret = -EFAULT; |
3674 | ||
3675 | kfree(sa); | |
3676 | return ret; | |
3677 | } | |
3678 | ||
2ff7e61e JM |
3679 | static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info, |
3680 | void __user *arg) | |
3f6bcfbd SB |
3681 | { |
3682 | struct btrfs_ioctl_dev_replace_args *p; | |
3683 | int ret; | |
3684 | ||
3685 | if (!capable(CAP_SYS_ADMIN)) | |
3686 | return -EPERM; | |
3687 | ||
3688 | p = memdup_user(arg, sizeof(*p)); | |
3689 | if (IS_ERR(p)) | |
3690 | return PTR_ERR(p); | |
3691 | ||
3692 | switch (p->cmd) { | |
3693 | case BTRFS_IOCTL_DEV_REPLACE_CMD_START: | |
bc98a42c | 3694 | if (sb_rdonly(fs_info->sb)) { |
adfa97cb ID |
3695 | ret = -EROFS; |
3696 | goto out; | |
3697 | } | |
c3e1f96c | 3698 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REPLACE)) { |
e57138b3 | 3699 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
3f6bcfbd | 3700 | } else { |
2ff7e61e | 3701 | ret = btrfs_dev_replace_by_ioctl(fs_info, p); |
c3e1f96c | 3702 | btrfs_exclop_finish(fs_info); |
3f6bcfbd SB |
3703 | } |
3704 | break; | |
3705 | case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS: | |
0b246afa | 3706 | btrfs_dev_replace_status(fs_info, p); |
3f6bcfbd SB |
3707 | ret = 0; |
3708 | break; | |
3709 | case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL: | |
17d202b9 | 3710 | p->result = btrfs_dev_replace_cancel(fs_info); |
97282031 | 3711 | ret = 0; |
3f6bcfbd SB |
3712 | break; |
3713 | default: | |
3714 | ret = -EINVAL; | |
3715 | break; | |
3716 | } | |
3717 | ||
d3a53286 | 3718 | if ((ret == 0 || ret == -ECANCELED) && copy_to_user(arg, p, sizeof(*p))) |
3f6bcfbd | 3719 | ret = -EFAULT; |
adfa97cb | 3720 | out: |
3f6bcfbd SB |
3721 | kfree(p); |
3722 | return ret; | |
3723 | } | |
3724 | ||
d7728c96 JS |
3725 | static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg) |
3726 | { | |
3727 | int ret = 0; | |
3728 | int i; | |
740c3d22 | 3729 | u64 rel_ptr; |
d7728c96 | 3730 | int size; |
806468f8 | 3731 | struct btrfs_ioctl_ino_path_args *ipa = NULL; |
d7728c96 JS |
3732 | struct inode_fs_paths *ipath = NULL; |
3733 | struct btrfs_path *path; | |
3734 | ||
82b22ac8 | 3735 | if (!capable(CAP_DAC_READ_SEARCH)) |
d7728c96 JS |
3736 | return -EPERM; |
3737 | ||
3738 | path = btrfs_alloc_path(); | |
3739 | if (!path) { | |
3740 | ret = -ENOMEM; | |
3741 | goto out; | |
3742 | } | |
3743 | ||
3744 | ipa = memdup_user(arg, sizeof(*ipa)); | |
3745 | if (IS_ERR(ipa)) { | |
3746 | ret = PTR_ERR(ipa); | |
3747 | ipa = NULL; | |
3748 | goto out; | |
3749 | } | |
3750 | ||
3751 | size = min_t(u32, ipa->size, 4096); | |
3752 | ipath = init_ipath(size, root, path); | |
3753 | if (IS_ERR(ipath)) { | |
3754 | ret = PTR_ERR(ipath); | |
3755 | ipath = NULL; | |
3756 | goto out; | |
3757 | } | |
3758 | ||
3759 | ret = paths_from_inode(ipa->inum, ipath); | |
3760 | if (ret < 0) | |
3761 | goto out; | |
3762 | ||
3763 | for (i = 0; i < ipath->fspath->elem_cnt; ++i) { | |
745c4d8e JM |
3764 | rel_ptr = ipath->fspath->val[i] - |
3765 | (u64)(unsigned long)ipath->fspath->val; | |
740c3d22 | 3766 | ipath->fspath->val[i] = rel_ptr; |
d7728c96 JS |
3767 | } |
3768 | ||
718dc5fa OS |
3769 | ret = copy_to_user((void __user *)(unsigned long)ipa->fspath, |
3770 | ipath->fspath, size); | |
d7728c96 JS |
3771 | if (ret) { |
3772 | ret = -EFAULT; | |
3773 | goto out; | |
3774 | } | |
3775 | ||
3776 | out: | |
3777 | btrfs_free_path(path); | |
3778 | free_ipath(ipath); | |
3779 | kfree(ipa); | |
3780 | ||
3781 | return ret; | |
3782 | } | |
3783 | ||
3784 | static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx) | |
3785 | { | |
3786 | struct btrfs_data_container *inodes = ctx; | |
3787 | const size_t c = 3 * sizeof(u64); | |
3788 | ||
3789 | if (inodes->bytes_left >= c) { | |
3790 | inodes->bytes_left -= c; | |
3791 | inodes->val[inodes->elem_cnt] = inum; | |
3792 | inodes->val[inodes->elem_cnt + 1] = offset; | |
3793 | inodes->val[inodes->elem_cnt + 2] = root; | |
3794 | inodes->elem_cnt += 3; | |
3795 | } else { | |
3796 | inodes->bytes_missing += c - inodes->bytes_left; | |
3797 | inodes->bytes_left = 0; | |
3798 | inodes->elem_missed += 3; | |
3799 | } | |
3800 | ||
3801 | return 0; | |
3802 | } | |
3803 | ||
2ff7e61e | 3804 | static long btrfs_ioctl_logical_to_ino(struct btrfs_fs_info *fs_info, |
d24a67b2 | 3805 | void __user *arg, int version) |
d7728c96 JS |
3806 | { |
3807 | int ret = 0; | |
3808 | int size; | |
d7728c96 JS |
3809 | struct btrfs_ioctl_logical_ino_args *loi; |
3810 | struct btrfs_data_container *inodes = NULL; | |
3811 | struct btrfs_path *path = NULL; | |
d24a67b2 | 3812 | bool ignore_offset; |
d7728c96 JS |
3813 | |
3814 | if (!capable(CAP_SYS_ADMIN)) | |
3815 | return -EPERM; | |
3816 | ||
3817 | loi = memdup_user(arg, sizeof(*loi)); | |
7b9ea627 SV |
3818 | if (IS_ERR(loi)) |
3819 | return PTR_ERR(loi); | |
d7728c96 | 3820 | |
d24a67b2 ZB |
3821 | if (version == 1) { |
3822 | ignore_offset = false; | |
b115e3bc | 3823 | size = min_t(u32, loi->size, SZ_64K); |
d24a67b2 ZB |
3824 | } else { |
3825 | /* All reserved bits must be 0 for now */ | |
3826 | if (memchr_inv(loi->reserved, 0, sizeof(loi->reserved))) { | |
3827 | ret = -EINVAL; | |
3828 | goto out_loi; | |
3829 | } | |
3830 | /* Only accept flags we have defined so far */ | |
3831 | if (loi->flags & ~(BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET)) { | |
3832 | ret = -EINVAL; | |
3833 | goto out_loi; | |
3834 | } | |
3835 | ignore_offset = loi->flags & BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET; | |
b115e3bc | 3836 | size = min_t(u32, loi->size, SZ_16M); |
d24a67b2 ZB |
3837 | } |
3838 | ||
d7728c96 JS |
3839 | path = btrfs_alloc_path(); |
3840 | if (!path) { | |
3841 | ret = -ENOMEM; | |
3842 | goto out; | |
3843 | } | |
3844 | ||
d7728c96 JS |
3845 | inodes = init_data_container(size); |
3846 | if (IS_ERR(inodes)) { | |
3847 | ret = PTR_ERR(inodes); | |
3848 | inodes = NULL; | |
3849 | goto out; | |
3850 | } | |
3851 | ||
2ff7e61e | 3852 | ret = iterate_inodes_from_logical(loi->logical, fs_info, path, |
d24a67b2 | 3853 | build_ino_list, inodes, ignore_offset); |
df031f07 | 3854 | if (ret == -EINVAL) |
d7728c96 JS |
3855 | ret = -ENOENT; |
3856 | if (ret < 0) | |
3857 | goto out; | |
3858 | ||
718dc5fa OS |
3859 | ret = copy_to_user((void __user *)(unsigned long)loi->inodes, inodes, |
3860 | size); | |
d7728c96 JS |
3861 | if (ret) |
3862 | ret = -EFAULT; | |
3863 | ||
3864 | out: | |
3865 | btrfs_free_path(path); | |
f54de068 | 3866 | kvfree(inodes); |
d24a67b2 | 3867 | out_loi: |
d7728c96 JS |
3868 | kfree(loi); |
3869 | ||
3870 | return ret; | |
3871 | } | |
3872 | ||
008ef096 | 3873 | void btrfs_update_ioctl_balance_args(struct btrfs_fs_info *fs_info, |
c9e9f97b ID |
3874 | struct btrfs_ioctl_balance_args *bargs) |
3875 | { | |
3876 | struct btrfs_balance_control *bctl = fs_info->balance_ctl; | |
3877 | ||
3878 | bargs->flags = bctl->flags; | |
3879 | ||
3009a62f | 3880 | if (test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) |
837d5b6e ID |
3881 | bargs->state |= BTRFS_BALANCE_STATE_RUNNING; |
3882 | if (atomic_read(&fs_info->balance_pause_req)) | |
3883 | bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ; | |
a7e99c69 ID |
3884 | if (atomic_read(&fs_info->balance_cancel_req)) |
3885 | bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ; | |
837d5b6e | 3886 | |
c9e9f97b ID |
3887 | memcpy(&bargs->data, &bctl->data, sizeof(bargs->data)); |
3888 | memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta)); | |
3889 | memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys)); | |
19a39dce | 3890 | |
008ef096 DS |
3891 | spin_lock(&fs_info->balance_lock); |
3892 | memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat)); | |
3893 | spin_unlock(&fs_info->balance_lock); | |
c9e9f97b ID |
3894 | } |
3895 | ||
9ba1f6e4 | 3896 | static long btrfs_ioctl_balance(struct file *file, void __user *arg) |
c9e9f97b | 3897 | { |
496ad9aa | 3898 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
c9e9f97b ID |
3899 | struct btrfs_fs_info *fs_info = root->fs_info; |
3900 | struct btrfs_ioctl_balance_args *bargs; | |
3901 | struct btrfs_balance_control *bctl; | |
ed0fb78f | 3902 | bool need_unlock; /* for mut. excl. ops lock */ |
c9e9f97b ID |
3903 | int ret; |
3904 | ||
3905 | if (!capable(CAP_SYS_ADMIN)) | |
3906 | return -EPERM; | |
3907 | ||
e54bfa31 | 3908 | ret = mnt_want_write_file(file); |
9ba1f6e4 LB |
3909 | if (ret) |
3910 | return ret; | |
3911 | ||
ed0fb78f | 3912 | again: |
c3e1f96c | 3913 | if (btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) { |
ed0fb78f ID |
3914 | mutex_lock(&fs_info->balance_mutex); |
3915 | need_unlock = true; | |
3916 | goto locked; | |
3917 | } | |
3918 | ||
3919 | /* | |
01327610 | 3920 | * mut. excl. ops lock is locked. Three possibilities: |
ed0fb78f ID |
3921 | * (1) some other op is running |
3922 | * (2) balance is running | |
3923 | * (3) balance is paused -- special case (think resume) | |
3924 | */ | |
c9e9f97b | 3925 | mutex_lock(&fs_info->balance_mutex); |
ed0fb78f ID |
3926 | if (fs_info->balance_ctl) { |
3927 | /* this is either (2) or (3) */ | |
3009a62f | 3928 | if (!test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) { |
ed0fb78f | 3929 | mutex_unlock(&fs_info->balance_mutex); |
dccdb07b DS |
3930 | /* |
3931 | * Lock released to allow other waiters to continue, | |
3932 | * we'll reexamine the status again. | |
3933 | */ | |
ed0fb78f ID |
3934 | mutex_lock(&fs_info->balance_mutex); |
3935 | ||
3936 | if (fs_info->balance_ctl && | |
3009a62f | 3937 | !test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) { |
ed0fb78f ID |
3938 | /* this is (3) */ |
3939 | need_unlock = false; | |
3940 | goto locked; | |
3941 | } | |
3942 | ||
3943 | mutex_unlock(&fs_info->balance_mutex); | |
ed0fb78f ID |
3944 | goto again; |
3945 | } else { | |
3946 | /* this is (2) */ | |
3947 | mutex_unlock(&fs_info->balance_mutex); | |
3948 | ret = -EINPROGRESS; | |
3949 | goto out; | |
3950 | } | |
3951 | } else { | |
3952 | /* this is (1) */ | |
3953 | mutex_unlock(&fs_info->balance_mutex); | |
e57138b3 | 3954 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
ed0fb78f ID |
3955 | goto out; |
3956 | } | |
3957 | ||
3958 | locked: | |
c9e9f97b ID |
3959 | |
3960 | if (arg) { | |
3961 | bargs = memdup_user(arg, sizeof(*bargs)); | |
3962 | if (IS_ERR(bargs)) { | |
3963 | ret = PTR_ERR(bargs); | |
ed0fb78f | 3964 | goto out_unlock; |
c9e9f97b | 3965 | } |
de322263 ID |
3966 | |
3967 | if (bargs->flags & BTRFS_BALANCE_RESUME) { | |
3968 | if (!fs_info->balance_ctl) { | |
3969 | ret = -ENOTCONN; | |
3970 | goto out_bargs; | |
3971 | } | |
3972 | ||
3973 | bctl = fs_info->balance_ctl; | |
3974 | spin_lock(&fs_info->balance_lock); | |
3975 | bctl->flags |= BTRFS_BALANCE_RESUME; | |
3976 | spin_unlock(&fs_info->balance_lock); | |
3977 | ||
3978 | goto do_balance; | |
3979 | } | |
c9e9f97b ID |
3980 | } else { |
3981 | bargs = NULL; | |
3982 | } | |
3983 | ||
ed0fb78f | 3984 | if (fs_info->balance_ctl) { |
837d5b6e ID |
3985 | ret = -EINPROGRESS; |
3986 | goto out_bargs; | |
3987 | } | |
3988 | ||
8d2db785 | 3989 | bctl = kzalloc(sizeof(*bctl), GFP_KERNEL); |
c9e9f97b ID |
3990 | if (!bctl) { |
3991 | ret = -ENOMEM; | |
3992 | goto out_bargs; | |
3993 | } | |
3994 | ||
c9e9f97b ID |
3995 | if (arg) { |
3996 | memcpy(&bctl->data, &bargs->data, sizeof(bctl->data)); | |
3997 | memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta)); | |
3998 | memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys)); | |
3999 | ||
4000 | bctl->flags = bargs->flags; | |
f43ffb60 ID |
4001 | } else { |
4002 | /* balance everything - no filters */ | |
4003 | bctl->flags |= BTRFS_BALANCE_TYPE_MASK; | |
c9e9f97b ID |
4004 | } |
4005 | ||
8eb93459 DS |
4006 | if (bctl->flags & ~(BTRFS_BALANCE_ARGS_MASK | BTRFS_BALANCE_TYPE_MASK)) { |
4007 | ret = -EINVAL; | |
0f89abf5 | 4008 | goto out_bctl; |
8eb93459 DS |
4009 | } |
4010 | ||
de322263 | 4011 | do_balance: |
c9e9f97b | 4012 | /* |
c3e1f96c GR |
4013 | * Ownership of bctl and exclusive operation goes to btrfs_balance. |
4014 | * bctl is freed in reset_balance_state, or, if restriper was paused | |
4015 | * all the way until unmount, in free_fs_info. The flag should be | |
4016 | * cleared after reset_balance_state. | |
c9e9f97b | 4017 | */ |
ed0fb78f ID |
4018 | need_unlock = false; |
4019 | ||
6fcf6e2b | 4020 | ret = btrfs_balance(fs_info, bctl, bargs); |
0f89abf5 | 4021 | bctl = NULL; |
ed0fb78f | 4022 | |
d00c2d9c | 4023 | if ((ret == 0 || ret == -ECANCELED) && arg) { |
c9e9f97b ID |
4024 | if (copy_to_user(arg, bargs, sizeof(*bargs))) |
4025 | ret = -EFAULT; | |
4026 | } | |
4027 | ||
0f89abf5 CE |
4028 | out_bctl: |
4029 | kfree(bctl); | |
c9e9f97b ID |
4030 | out_bargs: |
4031 | kfree(bargs); | |
ed0fb78f | 4032 | out_unlock: |
c9e9f97b | 4033 | mutex_unlock(&fs_info->balance_mutex); |
ed0fb78f | 4034 | if (need_unlock) |
c3e1f96c | 4035 | btrfs_exclop_finish(fs_info); |
ed0fb78f | 4036 | out: |
e54bfa31 | 4037 | mnt_drop_write_file(file); |
c9e9f97b ID |
4038 | return ret; |
4039 | } | |
4040 | ||
2ff7e61e | 4041 | static long btrfs_ioctl_balance_ctl(struct btrfs_fs_info *fs_info, int cmd) |
837d5b6e ID |
4042 | { |
4043 | if (!capable(CAP_SYS_ADMIN)) | |
4044 | return -EPERM; | |
4045 | ||
4046 | switch (cmd) { | |
4047 | case BTRFS_BALANCE_CTL_PAUSE: | |
0b246afa | 4048 | return btrfs_pause_balance(fs_info); |
a7e99c69 | 4049 | case BTRFS_BALANCE_CTL_CANCEL: |
0b246afa | 4050 | return btrfs_cancel_balance(fs_info); |
837d5b6e ID |
4051 | } |
4052 | ||
4053 | return -EINVAL; | |
4054 | } | |
4055 | ||
2ff7e61e | 4056 | static long btrfs_ioctl_balance_progress(struct btrfs_fs_info *fs_info, |
19a39dce ID |
4057 | void __user *arg) |
4058 | { | |
19a39dce ID |
4059 | struct btrfs_ioctl_balance_args *bargs; |
4060 | int ret = 0; | |
4061 | ||
4062 | if (!capable(CAP_SYS_ADMIN)) | |
4063 | return -EPERM; | |
4064 | ||
4065 | mutex_lock(&fs_info->balance_mutex); | |
4066 | if (!fs_info->balance_ctl) { | |
4067 | ret = -ENOTCONN; | |
4068 | goto out; | |
4069 | } | |
4070 | ||
8d2db785 | 4071 | bargs = kzalloc(sizeof(*bargs), GFP_KERNEL); |
19a39dce ID |
4072 | if (!bargs) { |
4073 | ret = -ENOMEM; | |
4074 | goto out; | |
4075 | } | |
4076 | ||
008ef096 | 4077 | btrfs_update_ioctl_balance_args(fs_info, bargs); |
19a39dce ID |
4078 | |
4079 | if (copy_to_user(arg, bargs, sizeof(*bargs))) | |
4080 | ret = -EFAULT; | |
4081 | ||
4082 | kfree(bargs); | |
4083 | out: | |
4084 | mutex_unlock(&fs_info->balance_mutex); | |
4085 | return ret; | |
4086 | } | |
4087 | ||
905b0dda | 4088 | static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg) |
5d13a37b | 4089 | { |
0b246afa JM |
4090 | struct inode *inode = file_inode(file); |
4091 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
5d13a37b | 4092 | struct btrfs_ioctl_quota_ctl_args *sa; |
5d13a37b | 4093 | int ret; |
5d13a37b AJ |
4094 | |
4095 | if (!capable(CAP_SYS_ADMIN)) | |
4096 | return -EPERM; | |
4097 | ||
905b0dda MX |
4098 | ret = mnt_want_write_file(file); |
4099 | if (ret) | |
4100 | return ret; | |
5d13a37b AJ |
4101 | |
4102 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4103 | if (IS_ERR(sa)) { |
4104 | ret = PTR_ERR(sa); | |
4105 | goto drop_write; | |
4106 | } | |
5d13a37b | 4107 | |
0b246afa | 4108 | down_write(&fs_info->subvol_sem); |
5d13a37b AJ |
4109 | |
4110 | switch (sa->cmd) { | |
4111 | case BTRFS_QUOTA_CTL_ENABLE: | |
340f1aa2 | 4112 | ret = btrfs_quota_enable(fs_info); |
5d13a37b AJ |
4113 | break; |
4114 | case BTRFS_QUOTA_CTL_DISABLE: | |
340f1aa2 | 4115 | ret = btrfs_quota_disable(fs_info); |
5d13a37b | 4116 | break; |
5d13a37b AJ |
4117 | default: |
4118 | ret = -EINVAL; | |
4119 | break; | |
4120 | } | |
4121 | ||
5d13a37b | 4122 | kfree(sa); |
0b246afa | 4123 | up_write(&fs_info->subvol_sem); |
905b0dda MX |
4124 | drop_write: |
4125 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4126 | return ret; |
4127 | } | |
4128 | ||
905b0dda | 4129 | static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg) |
5d13a37b | 4130 | { |
0b246afa JM |
4131 | struct inode *inode = file_inode(file); |
4132 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4133 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
5d13a37b AJ |
4134 | struct btrfs_ioctl_qgroup_assign_args *sa; |
4135 | struct btrfs_trans_handle *trans; | |
4136 | int ret; | |
4137 | int err; | |
4138 | ||
4139 | if (!capable(CAP_SYS_ADMIN)) | |
4140 | return -EPERM; | |
4141 | ||
905b0dda MX |
4142 | ret = mnt_want_write_file(file); |
4143 | if (ret) | |
4144 | return ret; | |
5d13a37b AJ |
4145 | |
4146 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4147 | if (IS_ERR(sa)) { |
4148 | ret = PTR_ERR(sa); | |
4149 | goto drop_write; | |
4150 | } | |
5d13a37b AJ |
4151 | |
4152 | trans = btrfs_join_transaction(root); | |
4153 | if (IS_ERR(trans)) { | |
4154 | ret = PTR_ERR(trans); | |
4155 | goto out; | |
4156 | } | |
4157 | ||
5d13a37b | 4158 | if (sa->assign) { |
9f8a6ce6 | 4159 | ret = btrfs_add_qgroup_relation(trans, sa->src, sa->dst); |
5d13a37b | 4160 | } else { |
39616c27 | 4161 | ret = btrfs_del_qgroup_relation(trans, sa->src, sa->dst); |
5d13a37b AJ |
4162 | } |
4163 | ||
e082f563 | 4164 | /* update qgroup status and info */ |
280f8bd2 | 4165 | err = btrfs_run_qgroups(trans); |
e082f563 | 4166 | if (err < 0) |
0b246afa JM |
4167 | btrfs_handle_fs_error(fs_info, err, |
4168 | "failed to update qgroup status and info"); | |
3a45bb20 | 4169 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4170 | if (err && !ret) |
4171 | ret = err; | |
4172 | ||
4173 | out: | |
4174 | kfree(sa); | |
905b0dda MX |
4175 | drop_write: |
4176 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4177 | return ret; |
4178 | } | |
4179 | ||
905b0dda | 4180 | static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg) |
5d13a37b | 4181 | { |
0b246afa | 4182 | struct inode *inode = file_inode(file); |
0b246afa | 4183 | struct btrfs_root *root = BTRFS_I(inode)->root; |
5d13a37b AJ |
4184 | struct btrfs_ioctl_qgroup_create_args *sa; |
4185 | struct btrfs_trans_handle *trans; | |
4186 | int ret; | |
4187 | int err; | |
4188 | ||
4189 | if (!capable(CAP_SYS_ADMIN)) | |
4190 | return -EPERM; | |
4191 | ||
905b0dda MX |
4192 | ret = mnt_want_write_file(file); |
4193 | if (ret) | |
4194 | return ret; | |
5d13a37b AJ |
4195 | |
4196 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4197 | if (IS_ERR(sa)) { |
4198 | ret = PTR_ERR(sa); | |
4199 | goto drop_write; | |
4200 | } | |
5d13a37b | 4201 | |
d86e56cf MX |
4202 | if (!sa->qgroupid) { |
4203 | ret = -EINVAL; | |
4204 | goto out; | |
4205 | } | |
4206 | ||
5d13a37b AJ |
4207 | trans = btrfs_join_transaction(root); |
4208 | if (IS_ERR(trans)) { | |
4209 | ret = PTR_ERR(trans); | |
4210 | goto out; | |
4211 | } | |
4212 | ||
5d13a37b | 4213 | if (sa->create) { |
49a05ecd | 4214 | ret = btrfs_create_qgroup(trans, sa->qgroupid); |
5d13a37b | 4215 | } else { |
3efbee1d | 4216 | ret = btrfs_remove_qgroup(trans, sa->qgroupid); |
5d13a37b AJ |
4217 | } |
4218 | ||
3a45bb20 | 4219 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4220 | if (err && !ret) |
4221 | ret = err; | |
4222 | ||
4223 | out: | |
4224 | kfree(sa); | |
905b0dda MX |
4225 | drop_write: |
4226 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4227 | return ret; |
4228 | } | |
4229 | ||
905b0dda | 4230 | static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg) |
5d13a37b | 4231 | { |
0b246afa | 4232 | struct inode *inode = file_inode(file); |
0b246afa | 4233 | struct btrfs_root *root = BTRFS_I(inode)->root; |
5d13a37b AJ |
4234 | struct btrfs_ioctl_qgroup_limit_args *sa; |
4235 | struct btrfs_trans_handle *trans; | |
4236 | int ret; | |
4237 | int err; | |
4238 | u64 qgroupid; | |
4239 | ||
4240 | if (!capable(CAP_SYS_ADMIN)) | |
4241 | return -EPERM; | |
4242 | ||
905b0dda MX |
4243 | ret = mnt_want_write_file(file); |
4244 | if (ret) | |
4245 | return ret; | |
5d13a37b AJ |
4246 | |
4247 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4248 | if (IS_ERR(sa)) { |
4249 | ret = PTR_ERR(sa); | |
4250 | goto drop_write; | |
4251 | } | |
5d13a37b AJ |
4252 | |
4253 | trans = btrfs_join_transaction(root); | |
4254 | if (IS_ERR(trans)) { | |
4255 | ret = PTR_ERR(trans); | |
4256 | goto out; | |
4257 | } | |
4258 | ||
4259 | qgroupid = sa->qgroupid; | |
4260 | if (!qgroupid) { | |
4261 | /* take the current subvol as qgroup */ | |
4262 | qgroupid = root->root_key.objectid; | |
4263 | } | |
4264 | ||
f0042d5e | 4265 | ret = btrfs_limit_qgroup(trans, qgroupid, &sa->lim); |
5d13a37b | 4266 | |
3a45bb20 | 4267 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4268 | if (err && !ret) |
4269 | ret = err; | |
4270 | ||
4271 | out: | |
4272 | kfree(sa); | |
905b0dda MX |
4273 | drop_write: |
4274 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4275 | return ret; |
4276 | } | |
4277 | ||
2f232036 JS |
4278 | static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg) |
4279 | { | |
0b246afa JM |
4280 | struct inode *inode = file_inode(file); |
4281 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
2f232036 JS |
4282 | struct btrfs_ioctl_quota_rescan_args *qsa; |
4283 | int ret; | |
4284 | ||
4285 | if (!capable(CAP_SYS_ADMIN)) | |
4286 | return -EPERM; | |
4287 | ||
4288 | ret = mnt_want_write_file(file); | |
4289 | if (ret) | |
4290 | return ret; | |
4291 | ||
4292 | qsa = memdup_user(arg, sizeof(*qsa)); | |
4293 | if (IS_ERR(qsa)) { | |
4294 | ret = PTR_ERR(qsa); | |
4295 | goto drop_write; | |
4296 | } | |
4297 | ||
4298 | if (qsa->flags) { | |
4299 | ret = -EINVAL; | |
4300 | goto out; | |
4301 | } | |
4302 | ||
0b246afa | 4303 | ret = btrfs_qgroup_rescan(fs_info); |
2f232036 JS |
4304 | |
4305 | out: | |
4306 | kfree(qsa); | |
4307 | drop_write: | |
4308 | mnt_drop_write_file(file); | |
4309 | return ret; | |
4310 | } | |
4311 | ||
b929c1d8 MPS |
4312 | static long btrfs_ioctl_quota_rescan_status(struct btrfs_fs_info *fs_info, |
4313 | void __user *arg) | |
2f232036 | 4314 | { |
2f232036 JS |
4315 | struct btrfs_ioctl_quota_rescan_args *qsa; |
4316 | int ret = 0; | |
4317 | ||
4318 | if (!capable(CAP_SYS_ADMIN)) | |
4319 | return -EPERM; | |
4320 | ||
8d2db785 | 4321 | qsa = kzalloc(sizeof(*qsa), GFP_KERNEL); |
2f232036 JS |
4322 | if (!qsa) |
4323 | return -ENOMEM; | |
4324 | ||
0b246afa | 4325 | if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) { |
2f232036 | 4326 | qsa->flags = 1; |
0b246afa | 4327 | qsa->progress = fs_info->qgroup_rescan_progress.objectid; |
2f232036 JS |
4328 | } |
4329 | ||
4330 | if (copy_to_user(arg, qsa, sizeof(*qsa))) | |
4331 | ret = -EFAULT; | |
4332 | ||
4333 | kfree(qsa); | |
4334 | return ret; | |
4335 | } | |
4336 | ||
b929c1d8 MPS |
4337 | static long btrfs_ioctl_quota_rescan_wait(struct btrfs_fs_info *fs_info, |
4338 | void __user *arg) | |
57254b6e | 4339 | { |
57254b6e JS |
4340 | if (!capable(CAP_SYS_ADMIN)) |
4341 | return -EPERM; | |
4342 | ||
0b246afa | 4343 | return btrfs_qgroup_wait_for_completion(fs_info, true); |
57254b6e JS |
4344 | } |
4345 | ||
abccd00f HM |
4346 | static long _btrfs_ioctl_set_received_subvol(struct file *file, |
4347 | struct btrfs_ioctl_received_subvol_args *sa) | |
8ea05e3a | 4348 | { |
496ad9aa | 4349 | struct inode *inode = file_inode(file); |
0b246afa | 4350 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
8ea05e3a AB |
4351 | struct btrfs_root *root = BTRFS_I(inode)->root; |
4352 | struct btrfs_root_item *root_item = &root->root_item; | |
4353 | struct btrfs_trans_handle *trans; | |
95582b00 | 4354 | struct timespec64 ct = current_time(inode); |
8ea05e3a | 4355 | int ret = 0; |
dd5f9615 | 4356 | int received_uuid_changed; |
8ea05e3a | 4357 | |
21cb47be | 4358 | if (!inode_owner_or_capable(&init_user_ns, inode)) |
bd60ea0f DS |
4359 | return -EPERM; |
4360 | ||
8ea05e3a AB |
4361 | ret = mnt_want_write_file(file); |
4362 | if (ret < 0) | |
4363 | return ret; | |
4364 | ||
0b246afa | 4365 | down_write(&fs_info->subvol_sem); |
8ea05e3a | 4366 | |
4a0cc7ca | 4367 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
8ea05e3a AB |
4368 | ret = -EINVAL; |
4369 | goto out; | |
4370 | } | |
4371 | ||
4372 | if (btrfs_root_readonly(root)) { | |
4373 | ret = -EROFS; | |
4374 | goto out; | |
4375 | } | |
4376 | ||
dd5f9615 SB |
4377 | /* |
4378 | * 1 - root item | |
4379 | * 2 - uuid items (received uuid + subvol uuid) | |
4380 | */ | |
4381 | trans = btrfs_start_transaction(root, 3); | |
8ea05e3a AB |
4382 | if (IS_ERR(trans)) { |
4383 | ret = PTR_ERR(trans); | |
4384 | trans = NULL; | |
4385 | goto out; | |
4386 | } | |
4387 | ||
4388 | sa->rtransid = trans->transid; | |
4389 | sa->rtime.sec = ct.tv_sec; | |
4390 | sa->rtime.nsec = ct.tv_nsec; | |
4391 | ||
dd5f9615 SB |
4392 | received_uuid_changed = memcmp(root_item->received_uuid, sa->uuid, |
4393 | BTRFS_UUID_SIZE); | |
4394 | if (received_uuid_changed && | |
d87ff758 | 4395 | !btrfs_is_empty_uuid(root_item->received_uuid)) { |
d1957791 | 4396 | ret = btrfs_uuid_tree_remove(trans, root_item->received_uuid, |
d87ff758 NB |
4397 | BTRFS_UUID_KEY_RECEIVED_SUBVOL, |
4398 | root->root_key.objectid); | |
4399 | if (ret && ret != -ENOENT) { | |
4400 | btrfs_abort_transaction(trans, ret); | |
4401 | btrfs_end_transaction(trans); | |
4402 | goto out; | |
4403 | } | |
4404 | } | |
8ea05e3a AB |
4405 | memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE); |
4406 | btrfs_set_root_stransid(root_item, sa->stransid); | |
4407 | btrfs_set_root_rtransid(root_item, sa->rtransid); | |
3cae210f QW |
4408 | btrfs_set_stack_timespec_sec(&root_item->stime, sa->stime.sec); |
4409 | btrfs_set_stack_timespec_nsec(&root_item->stime, sa->stime.nsec); | |
4410 | btrfs_set_stack_timespec_sec(&root_item->rtime, sa->rtime.sec); | |
4411 | btrfs_set_stack_timespec_nsec(&root_item->rtime, sa->rtime.nsec); | |
8ea05e3a | 4412 | |
0b246afa | 4413 | ret = btrfs_update_root(trans, fs_info->tree_root, |
8ea05e3a AB |
4414 | &root->root_key, &root->root_item); |
4415 | if (ret < 0) { | |
3a45bb20 | 4416 | btrfs_end_transaction(trans); |
8ea05e3a | 4417 | goto out; |
dd5f9615 SB |
4418 | } |
4419 | if (received_uuid_changed && !btrfs_is_empty_uuid(sa->uuid)) { | |
cdb345a8 | 4420 | ret = btrfs_uuid_tree_add(trans, sa->uuid, |
dd5f9615 SB |
4421 | BTRFS_UUID_KEY_RECEIVED_SUBVOL, |
4422 | root->root_key.objectid); | |
4423 | if (ret < 0 && ret != -EEXIST) { | |
66642832 | 4424 | btrfs_abort_transaction(trans, ret); |
efd38150 | 4425 | btrfs_end_transaction(trans); |
8ea05e3a | 4426 | goto out; |
dd5f9615 SB |
4427 | } |
4428 | } | |
3a45bb20 | 4429 | ret = btrfs_commit_transaction(trans); |
abccd00f | 4430 | out: |
0b246afa | 4431 | up_write(&fs_info->subvol_sem); |
abccd00f HM |
4432 | mnt_drop_write_file(file); |
4433 | return ret; | |
4434 | } | |
4435 | ||
4436 | #ifdef CONFIG_64BIT | |
4437 | static long btrfs_ioctl_set_received_subvol_32(struct file *file, | |
4438 | void __user *arg) | |
4439 | { | |
4440 | struct btrfs_ioctl_received_subvol_args_32 *args32 = NULL; | |
4441 | struct btrfs_ioctl_received_subvol_args *args64 = NULL; | |
4442 | int ret = 0; | |
4443 | ||
4444 | args32 = memdup_user(arg, sizeof(*args32)); | |
7b9ea627 SV |
4445 | if (IS_ERR(args32)) |
4446 | return PTR_ERR(args32); | |
abccd00f | 4447 | |
8d2db785 | 4448 | args64 = kmalloc(sizeof(*args64), GFP_KERNEL); |
84dbeb87 DC |
4449 | if (!args64) { |
4450 | ret = -ENOMEM; | |
abccd00f HM |
4451 | goto out; |
4452 | } | |
4453 | ||
4454 | memcpy(args64->uuid, args32->uuid, BTRFS_UUID_SIZE); | |
4455 | args64->stransid = args32->stransid; | |
4456 | args64->rtransid = args32->rtransid; | |
4457 | args64->stime.sec = args32->stime.sec; | |
4458 | args64->stime.nsec = args32->stime.nsec; | |
4459 | args64->rtime.sec = args32->rtime.sec; | |
4460 | args64->rtime.nsec = args32->rtime.nsec; | |
4461 | args64->flags = args32->flags; | |
4462 | ||
4463 | ret = _btrfs_ioctl_set_received_subvol(file, args64); | |
4464 | if (ret) | |
4465 | goto out; | |
4466 | ||
4467 | memcpy(args32->uuid, args64->uuid, BTRFS_UUID_SIZE); | |
4468 | args32->stransid = args64->stransid; | |
4469 | args32->rtransid = args64->rtransid; | |
4470 | args32->stime.sec = args64->stime.sec; | |
4471 | args32->stime.nsec = args64->stime.nsec; | |
4472 | args32->rtime.sec = args64->rtime.sec; | |
4473 | args32->rtime.nsec = args64->rtime.nsec; | |
4474 | args32->flags = args64->flags; | |
4475 | ||
4476 | ret = copy_to_user(arg, args32, sizeof(*args32)); | |
4477 | if (ret) | |
4478 | ret = -EFAULT; | |
4479 | ||
4480 | out: | |
4481 | kfree(args32); | |
4482 | kfree(args64); | |
4483 | return ret; | |
4484 | } | |
4485 | #endif | |
4486 | ||
4487 | static long btrfs_ioctl_set_received_subvol(struct file *file, | |
4488 | void __user *arg) | |
4489 | { | |
4490 | struct btrfs_ioctl_received_subvol_args *sa = NULL; | |
4491 | int ret = 0; | |
4492 | ||
4493 | sa = memdup_user(arg, sizeof(*sa)); | |
7b9ea627 SV |
4494 | if (IS_ERR(sa)) |
4495 | return PTR_ERR(sa); | |
abccd00f HM |
4496 | |
4497 | ret = _btrfs_ioctl_set_received_subvol(file, sa); | |
4498 | ||
4499 | if (ret) | |
4500 | goto out; | |
4501 | ||
8ea05e3a AB |
4502 | ret = copy_to_user(arg, sa, sizeof(*sa)); |
4503 | if (ret) | |
4504 | ret = -EFAULT; | |
4505 | ||
4506 | out: | |
4507 | kfree(sa); | |
8ea05e3a AB |
4508 | return ret; |
4509 | } | |
4510 | ||
b929c1d8 MPS |
4511 | static int btrfs_ioctl_get_fslabel(struct btrfs_fs_info *fs_info, |
4512 | void __user *arg) | |
867ab667 | 4513 | { |
a1b83ac5 | 4514 | size_t len; |
867ab667 | 4515 | int ret; |
a1b83ac5 AJ |
4516 | char label[BTRFS_LABEL_SIZE]; |
4517 | ||
0b246afa JM |
4518 | spin_lock(&fs_info->super_lock); |
4519 | memcpy(label, fs_info->super_copy->label, BTRFS_LABEL_SIZE); | |
4520 | spin_unlock(&fs_info->super_lock); | |
a1b83ac5 AJ |
4521 | |
4522 | len = strnlen(label, BTRFS_LABEL_SIZE); | |
867ab667 | 4523 | |
4524 | if (len == BTRFS_LABEL_SIZE) { | |
0b246afa JM |
4525 | btrfs_warn(fs_info, |
4526 | "label is too long, return the first %zu bytes", | |
4527 | --len); | |
867ab667 | 4528 | } |
4529 | ||
867ab667 | 4530 | ret = copy_to_user(arg, label, len); |
867ab667 | 4531 | |
4532 | return ret ? -EFAULT : 0; | |
4533 | } | |
4534 | ||
a8bfd4ab | 4535 | static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg) |
4536 | { | |
0b246afa JM |
4537 | struct inode *inode = file_inode(file); |
4538 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4539 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4540 | struct btrfs_super_block *super_block = fs_info->super_copy; | |
a8bfd4ab | 4541 | struct btrfs_trans_handle *trans; |
4542 | char label[BTRFS_LABEL_SIZE]; | |
4543 | int ret; | |
4544 | ||
4545 | if (!capable(CAP_SYS_ADMIN)) | |
4546 | return -EPERM; | |
4547 | ||
4548 | if (copy_from_user(label, arg, sizeof(label))) | |
4549 | return -EFAULT; | |
4550 | ||
4551 | if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) { | |
0b246afa | 4552 | btrfs_err(fs_info, |
5d163e0e JM |
4553 | "unable to set label with more than %d bytes", |
4554 | BTRFS_LABEL_SIZE - 1); | |
a8bfd4ab | 4555 | return -EINVAL; |
4556 | } | |
4557 | ||
4558 | ret = mnt_want_write_file(file); | |
4559 | if (ret) | |
4560 | return ret; | |
4561 | ||
a8bfd4ab | 4562 | trans = btrfs_start_transaction(root, 0); |
4563 | if (IS_ERR(trans)) { | |
4564 | ret = PTR_ERR(trans); | |
4565 | goto out_unlock; | |
4566 | } | |
4567 | ||
0b246afa | 4568 | spin_lock(&fs_info->super_lock); |
a8bfd4ab | 4569 | strcpy(super_block->label, label); |
0b246afa | 4570 | spin_unlock(&fs_info->super_lock); |
3a45bb20 | 4571 | ret = btrfs_commit_transaction(trans); |
a8bfd4ab | 4572 | |
4573 | out_unlock: | |
a8bfd4ab | 4574 | mnt_drop_write_file(file); |
4575 | return ret; | |
4576 | } | |
4577 | ||
2eaa055f JM |
4578 | #define INIT_FEATURE_FLAGS(suffix) \ |
4579 | { .compat_flags = BTRFS_FEATURE_COMPAT_##suffix, \ | |
4580 | .compat_ro_flags = BTRFS_FEATURE_COMPAT_RO_##suffix, \ | |
4581 | .incompat_flags = BTRFS_FEATURE_INCOMPAT_##suffix } | |
4582 | ||
d5131b65 | 4583 | int btrfs_ioctl_get_supported_features(void __user *arg) |
2eaa055f | 4584 | { |
4d4ab6d6 | 4585 | static const struct btrfs_ioctl_feature_flags features[3] = { |
2eaa055f JM |
4586 | INIT_FEATURE_FLAGS(SUPP), |
4587 | INIT_FEATURE_FLAGS(SAFE_SET), | |
4588 | INIT_FEATURE_FLAGS(SAFE_CLEAR) | |
4589 | }; | |
4590 | ||
4591 | if (copy_to_user(arg, &features, sizeof(features))) | |
4592 | return -EFAULT; | |
4593 | ||
4594 | return 0; | |
4595 | } | |
4596 | ||
b929c1d8 MPS |
4597 | static int btrfs_ioctl_get_features(struct btrfs_fs_info *fs_info, |
4598 | void __user *arg) | |
2eaa055f | 4599 | { |
0b246afa | 4600 | struct btrfs_super_block *super_block = fs_info->super_copy; |
2eaa055f JM |
4601 | struct btrfs_ioctl_feature_flags features; |
4602 | ||
4603 | features.compat_flags = btrfs_super_compat_flags(super_block); | |
4604 | features.compat_ro_flags = btrfs_super_compat_ro_flags(super_block); | |
4605 | features.incompat_flags = btrfs_super_incompat_flags(super_block); | |
4606 | ||
4607 | if (copy_to_user(arg, &features, sizeof(features))) | |
4608 | return -EFAULT; | |
4609 | ||
4610 | return 0; | |
4611 | } | |
4612 | ||
2ff7e61e | 4613 | static int check_feature_bits(struct btrfs_fs_info *fs_info, |
3b02a68a | 4614 | enum btrfs_feature_set set, |
2eaa055f JM |
4615 | u64 change_mask, u64 flags, u64 supported_flags, |
4616 | u64 safe_set, u64 safe_clear) | |
4617 | { | |
f10152bc | 4618 | const char *type = btrfs_feature_set_name(set); |
3b02a68a | 4619 | char *names; |
2eaa055f JM |
4620 | u64 disallowed, unsupported; |
4621 | u64 set_mask = flags & change_mask; | |
4622 | u64 clear_mask = ~flags & change_mask; | |
4623 | ||
4624 | unsupported = set_mask & ~supported_flags; | |
4625 | if (unsupported) { | |
3b02a68a JM |
4626 | names = btrfs_printable_features(set, unsupported); |
4627 | if (names) { | |
0b246afa JM |
4628 | btrfs_warn(fs_info, |
4629 | "this kernel does not support the %s feature bit%s", | |
4630 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4631 | kfree(names); |
4632 | } else | |
0b246afa JM |
4633 | btrfs_warn(fs_info, |
4634 | "this kernel does not support %s bits 0x%llx", | |
4635 | type, unsupported); | |
2eaa055f JM |
4636 | return -EOPNOTSUPP; |
4637 | } | |
4638 | ||
4639 | disallowed = set_mask & ~safe_set; | |
4640 | if (disallowed) { | |
3b02a68a JM |
4641 | names = btrfs_printable_features(set, disallowed); |
4642 | if (names) { | |
0b246afa JM |
4643 | btrfs_warn(fs_info, |
4644 | "can't set the %s feature bit%s while mounted", | |
4645 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4646 | kfree(names); |
4647 | } else | |
0b246afa JM |
4648 | btrfs_warn(fs_info, |
4649 | "can't set %s bits 0x%llx while mounted", | |
4650 | type, disallowed); | |
2eaa055f JM |
4651 | return -EPERM; |
4652 | } | |
4653 | ||
4654 | disallowed = clear_mask & ~safe_clear; | |
4655 | if (disallowed) { | |
3b02a68a JM |
4656 | names = btrfs_printable_features(set, disallowed); |
4657 | if (names) { | |
0b246afa JM |
4658 | btrfs_warn(fs_info, |
4659 | "can't clear the %s feature bit%s while mounted", | |
4660 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4661 | kfree(names); |
4662 | } else | |
0b246afa JM |
4663 | btrfs_warn(fs_info, |
4664 | "can't clear %s bits 0x%llx while mounted", | |
4665 | type, disallowed); | |
2eaa055f JM |
4666 | return -EPERM; |
4667 | } | |
4668 | ||
4669 | return 0; | |
4670 | } | |
4671 | ||
2ff7e61e JM |
4672 | #define check_feature(fs_info, change_mask, flags, mask_base) \ |
4673 | check_feature_bits(fs_info, FEAT_##mask_base, change_mask, flags, \ | |
2eaa055f JM |
4674 | BTRFS_FEATURE_ ## mask_base ## _SUPP, \ |
4675 | BTRFS_FEATURE_ ## mask_base ## _SAFE_SET, \ | |
4676 | BTRFS_FEATURE_ ## mask_base ## _SAFE_CLEAR) | |
4677 | ||
4678 | static int btrfs_ioctl_set_features(struct file *file, void __user *arg) | |
4679 | { | |
0b246afa JM |
4680 | struct inode *inode = file_inode(file); |
4681 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4682 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4683 | struct btrfs_super_block *super_block = fs_info->super_copy; | |
2eaa055f JM |
4684 | struct btrfs_ioctl_feature_flags flags[2]; |
4685 | struct btrfs_trans_handle *trans; | |
4686 | u64 newflags; | |
4687 | int ret; | |
4688 | ||
4689 | if (!capable(CAP_SYS_ADMIN)) | |
4690 | return -EPERM; | |
4691 | ||
4692 | if (copy_from_user(flags, arg, sizeof(flags))) | |
4693 | return -EFAULT; | |
4694 | ||
4695 | /* Nothing to do */ | |
4696 | if (!flags[0].compat_flags && !flags[0].compat_ro_flags && | |
4697 | !flags[0].incompat_flags) | |
4698 | return 0; | |
4699 | ||
2ff7e61e | 4700 | ret = check_feature(fs_info, flags[0].compat_flags, |
2eaa055f JM |
4701 | flags[1].compat_flags, COMPAT); |
4702 | if (ret) | |
4703 | return ret; | |
4704 | ||
2ff7e61e | 4705 | ret = check_feature(fs_info, flags[0].compat_ro_flags, |
2eaa055f JM |
4706 | flags[1].compat_ro_flags, COMPAT_RO); |
4707 | if (ret) | |
4708 | return ret; | |
4709 | ||
2ff7e61e | 4710 | ret = check_feature(fs_info, flags[0].incompat_flags, |
2eaa055f JM |
4711 | flags[1].incompat_flags, INCOMPAT); |
4712 | if (ret) | |
4713 | return ret; | |
4714 | ||
7ab19625 DS |
4715 | ret = mnt_want_write_file(file); |
4716 | if (ret) | |
4717 | return ret; | |
4718 | ||
8051aa1a | 4719 | trans = btrfs_start_transaction(root, 0); |
7ab19625 DS |
4720 | if (IS_ERR(trans)) { |
4721 | ret = PTR_ERR(trans); | |
4722 | goto out_drop_write; | |
4723 | } | |
2eaa055f | 4724 | |
0b246afa | 4725 | spin_lock(&fs_info->super_lock); |
2eaa055f JM |
4726 | newflags = btrfs_super_compat_flags(super_block); |
4727 | newflags |= flags[0].compat_flags & flags[1].compat_flags; | |
4728 | newflags &= ~(flags[0].compat_flags & ~flags[1].compat_flags); | |
4729 | btrfs_set_super_compat_flags(super_block, newflags); | |
4730 | ||
4731 | newflags = btrfs_super_compat_ro_flags(super_block); | |
4732 | newflags |= flags[0].compat_ro_flags & flags[1].compat_ro_flags; | |
4733 | newflags &= ~(flags[0].compat_ro_flags & ~flags[1].compat_ro_flags); | |
4734 | btrfs_set_super_compat_ro_flags(super_block, newflags); | |
4735 | ||
4736 | newflags = btrfs_super_incompat_flags(super_block); | |
4737 | newflags |= flags[0].incompat_flags & flags[1].incompat_flags; | |
4738 | newflags &= ~(flags[0].incompat_flags & ~flags[1].incompat_flags); | |
4739 | btrfs_set_super_incompat_flags(super_block, newflags); | |
0b246afa | 4740 | spin_unlock(&fs_info->super_lock); |
2eaa055f | 4741 | |
3a45bb20 | 4742 | ret = btrfs_commit_transaction(trans); |
7ab19625 DS |
4743 | out_drop_write: |
4744 | mnt_drop_write_file(file); | |
4745 | ||
4746 | return ret; | |
2eaa055f JM |
4747 | } |
4748 | ||
2351f431 JB |
4749 | static int _btrfs_ioctl_send(struct file *file, void __user *argp, bool compat) |
4750 | { | |
4751 | struct btrfs_ioctl_send_args *arg; | |
4752 | int ret; | |
4753 | ||
4754 | if (compat) { | |
4755 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) | |
4756 | struct btrfs_ioctl_send_args_32 args32; | |
4757 | ||
4758 | ret = copy_from_user(&args32, argp, sizeof(args32)); | |
4759 | if (ret) | |
4760 | return -EFAULT; | |
4761 | arg = kzalloc(sizeof(*arg), GFP_KERNEL); | |
4762 | if (!arg) | |
4763 | return -ENOMEM; | |
4764 | arg->send_fd = args32.send_fd; | |
4765 | arg->clone_sources_count = args32.clone_sources_count; | |
4766 | arg->clone_sources = compat_ptr(args32.clone_sources); | |
4767 | arg->parent_root = args32.parent_root; | |
4768 | arg->flags = args32.flags; | |
4769 | memcpy(arg->reserved, args32.reserved, | |
4770 | sizeof(args32.reserved)); | |
4771 | #else | |
4772 | return -ENOTTY; | |
4773 | #endif | |
4774 | } else { | |
4775 | arg = memdup_user(argp, sizeof(*arg)); | |
4776 | if (IS_ERR(arg)) | |
4777 | return PTR_ERR(arg); | |
4778 | } | |
4779 | ret = btrfs_ioctl_send(file, arg); | |
4780 | kfree(arg); | |
4781 | return ret; | |
4782 | } | |
4783 | ||
f46b5a66 CH |
4784 | long btrfs_ioctl(struct file *file, unsigned int |
4785 | cmd, unsigned long arg) | |
4786 | { | |
0b246afa JM |
4787 | struct inode *inode = file_inode(file); |
4788 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4789 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4bcabaa3 | 4790 | void __user *argp = (void __user *)arg; |
f46b5a66 CH |
4791 | |
4792 | switch (cmd) { | |
6cbff00f CH |
4793 | case FS_IOC_GETVERSION: |
4794 | return btrfs_ioctl_getversion(file, argp); | |
40cf931f | 4795 | case FS_IOC_GETFSLABEL: |
b929c1d8 | 4796 | return btrfs_ioctl_get_fslabel(fs_info, argp); |
40cf931f ES |
4797 | case FS_IOC_SETFSLABEL: |
4798 | return btrfs_ioctl_set_fslabel(file, argp); | |
f7039b1d | 4799 | case FITRIM: |
b929c1d8 | 4800 | return btrfs_ioctl_fitrim(fs_info, argp); |
f46b5a66 | 4801 | case BTRFS_IOC_SNAP_CREATE: |
fa0d2b9b | 4802 | return btrfs_ioctl_snap_create(file, argp, 0); |
fdfb1e4f | 4803 | case BTRFS_IOC_SNAP_CREATE_V2: |
fa0d2b9b | 4804 | return btrfs_ioctl_snap_create_v2(file, argp, 0); |
3de4586c | 4805 | case BTRFS_IOC_SUBVOL_CREATE: |
fa0d2b9b | 4806 | return btrfs_ioctl_snap_create(file, argp, 1); |
6f72c7e2 AJ |
4807 | case BTRFS_IOC_SUBVOL_CREATE_V2: |
4808 | return btrfs_ioctl_snap_create_v2(file, argp, 1); | |
76dda93c | 4809 | case BTRFS_IOC_SNAP_DESTROY: |
949964c9 MPS |
4810 | return btrfs_ioctl_snap_destroy(file, argp, false); |
4811 | case BTRFS_IOC_SNAP_DESTROY_V2: | |
4812 | return btrfs_ioctl_snap_destroy(file, argp, true); | |
0caa102d LZ |
4813 | case BTRFS_IOC_SUBVOL_GETFLAGS: |
4814 | return btrfs_ioctl_subvol_getflags(file, argp); | |
4815 | case BTRFS_IOC_SUBVOL_SETFLAGS: | |
4816 | return btrfs_ioctl_subvol_setflags(file, argp); | |
6ef5ed0d JB |
4817 | case BTRFS_IOC_DEFAULT_SUBVOL: |
4818 | return btrfs_ioctl_default_subvol(file, argp); | |
f46b5a66 | 4819 | case BTRFS_IOC_DEFRAG: |
1e701a32 CM |
4820 | return btrfs_ioctl_defrag(file, NULL); |
4821 | case BTRFS_IOC_DEFRAG_RANGE: | |
4822 | return btrfs_ioctl_defrag(file, argp); | |
f46b5a66 | 4823 | case BTRFS_IOC_RESIZE: |
198605a8 | 4824 | return btrfs_ioctl_resize(file, argp); |
f46b5a66 | 4825 | case BTRFS_IOC_ADD_DEV: |
2ff7e61e | 4826 | return btrfs_ioctl_add_dev(fs_info, argp); |
f46b5a66 | 4827 | case BTRFS_IOC_RM_DEV: |
da24927b | 4828 | return btrfs_ioctl_rm_dev(file, argp); |
6b526ed7 AJ |
4829 | case BTRFS_IOC_RM_DEV_V2: |
4830 | return btrfs_ioctl_rm_dev_v2(file, argp); | |
475f6387 | 4831 | case BTRFS_IOC_FS_INFO: |
2ff7e61e | 4832 | return btrfs_ioctl_fs_info(fs_info, argp); |
475f6387 | 4833 | case BTRFS_IOC_DEV_INFO: |
2ff7e61e | 4834 | return btrfs_ioctl_dev_info(fs_info, argp); |
f46b5a66 | 4835 | case BTRFS_IOC_BALANCE: |
9ba1f6e4 | 4836 | return btrfs_ioctl_balance(file, NULL); |
ac8e9819 CM |
4837 | case BTRFS_IOC_TREE_SEARCH: |
4838 | return btrfs_ioctl_tree_search(file, argp); | |
cc68a8a5 GH |
4839 | case BTRFS_IOC_TREE_SEARCH_V2: |
4840 | return btrfs_ioctl_tree_search_v2(file, argp); | |
ac8e9819 CM |
4841 | case BTRFS_IOC_INO_LOOKUP: |
4842 | return btrfs_ioctl_ino_lookup(file, argp); | |
d7728c96 JS |
4843 | case BTRFS_IOC_INO_PATHS: |
4844 | return btrfs_ioctl_ino_to_path(root, argp); | |
4845 | case BTRFS_IOC_LOGICAL_INO: | |
d24a67b2 ZB |
4846 | return btrfs_ioctl_logical_to_ino(fs_info, argp, 1); |
4847 | case BTRFS_IOC_LOGICAL_INO_V2: | |
4848 | return btrfs_ioctl_logical_to_ino(fs_info, argp, 2); | |
1406e432 | 4849 | case BTRFS_IOC_SPACE_INFO: |
2ff7e61e | 4850 | return btrfs_ioctl_space_info(fs_info, argp); |
9b199859 FDBM |
4851 | case BTRFS_IOC_SYNC: { |
4852 | int ret; | |
4853 | ||
9db4dc24 | 4854 | ret = btrfs_start_delalloc_roots(fs_info, LONG_MAX, false); |
9b199859 FDBM |
4855 | if (ret) |
4856 | return ret; | |
0b246afa | 4857 | ret = btrfs_sync_fs(inode->i_sb, 1); |
2fad4e83 DS |
4858 | /* |
4859 | * The transaction thread may want to do more work, | |
01327610 | 4860 | * namely it pokes the cleaner kthread that will start |
2fad4e83 DS |
4861 | * processing uncleaned subvols. |
4862 | */ | |
0b246afa | 4863 | wake_up_process(fs_info->transaction_kthread); |
9b199859 FDBM |
4864 | return ret; |
4865 | } | |
46204592 | 4866 | case BTRFS_IOC_START_SYNC: |
9a8c28be | 4867 | return btrfs_ioctl_start_sync(root, argp); |
46204592 | 4868 | case BTRFS_IOC_WAIT_SYNC: |
2ff7e61e | 4869 | return btrfs_ioctl_wait_sync(fs_info, argp); |
475f6387 | 4870 | case BTRFS_IOC_SCRUB: |
b8e95489 | 4871 | return btrfs_ioctl_scrub(file, argp); |
475f6387 | 4872 | case BTRFS_IOC_SCRUB_CANCEL: |
2ff7e61e | 4873 | return btrfs_ioctl_scrub_cancel(fs_info); |
475f6387 | 4874 | case BTRFS_IOC_SCRUB_PROGRESS: |
2ff7e61e | 4875 | return btrfs_ioctl_scrub_progress(fs_info, argp); |
c9e9f97b | 4876 | case BTRFS_IOC_BALANCE_V2: |
9ba1f6e4 | 4877 | return btrfs_ioctl_balance(file, argp); |
837d5b6e | 4878 | case BTRFS_IOC_BALANCE_CTL: |
2ff7e61e | 4879 | return btrfs_ioctl_balance_ctl(fs_info, arg); |
19a39dce | 4880 | case BTRFS_IOC_BALANCE_PROGRESS: |
2ff7e61e | 4881 | return btrfs_ioctl_balance_progress(fs_info, argp); |
8ea05e3a AB |
4882 | case BTRFS_IOC_SET_RECEIVED_SUBVOL: |
4883 | return btrfs_ioctl_set_received_subvol(file, argp); | |
abccd00f HM |
4884 | #ifdef CONFIG_64BIT |
4885 | case BTRFS_IOC_SET_RECEIVED_SUBVOL_32: | |
4886 | return btrfs_ioctl_set_received_subvol_32(file, argp); | |
4887 | #endif | |
31db9f7c | 4888 | case BTRFS_IOC_SEND: |
2351f431 JB |
4889 | return _btrfs_ioctl_send(file, argp, false); |
4890 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) | |
4891 | case BTRFS_IOC_SEND_32: | |
4892 | return _btrfs_ioctl_send(file, argp, true); | |
4893 | #endif | |
c11d2c23 | 4894 | case BTRFS_IOC_GET_DEV_STATS: |
2ff7e61e | 4895 | return btrfs_ioctl_get_dev_stats(fs_info, argp); |
5d13a37b | 4896 | case BTRFS_IOC_QUOTA_CTL: |
905b0dda | 4897 | return btrfs_ioctl_quota_ctl(file, argp); |
5d13a37b | 4898 | case BTRFS_IOC_QGROUP_ASSIGN: |
905b0dda | 4899 | return btrfs_ioctl_qgroup_assign(file, argp); |
5d13a37b | 4900 | case BTRFS_IOC_QGROUP_CREATE: |
905b0dda | 4901 | return btrfs_ioctl_qgroup_create(file, argp); |
5d13a37b | 4902 | case BTRFS_IOC_QGROUP_LIMIT: |
905b0dda | 4903 | return btrfs_ioctl_qgroup_limit(file, argp); |
2f232036 JS |
4904 | case BTRFS_IOC_QUOTA_RESCAN: |
4905 | return btrfs_ioctl_quota_rescan(file, argp); | |
4906 | case BTRFS_IOC_QUOTA_RESCAN_STATUS: | |
b929c1d8 | 4907 | return btrfs_ioctl_quota_rescan_status(fs_info, argp); |
57254b6e | 4908 | case BTRFS_IOC_QUOTA_RESCAN_WAIT: |
b929c1d8 | 4909 | return btrfs_ioctl_quota_rescan_wait(fs_info, argp); |
3f6bcfbd | 4910 | case BTRFS_IOC_DEV_REPLACE: |
2ff7e61e | 4911 | return btrfs_ioctl_dev_replace(fs_info, argp); |
2eaa055f | 4912 | case BTRFS_IOC_GET_SUPPORTED_FEATURES: |
d5131b65 | 4913 | return btrfs_ioctl_get_supported_features(argp); |
2eaa055f | 4914 | case BTRFS_IOC_GET_FEATURES: |
b929c1d8 | 4915 | return btrfs_ioctl_get_features(fs_info, argp); |
2eaa055f JM |
4916 | case BTRFS_IOC_SET_FEATURES: |
4917 | return btrfs_ioctl_set_features(file, argp); | |
b64ec075 TM |
4918 | case BTRFS_IOC_GET_SUBVOL_INFO: |
4919 | return btrfs_ioctl_get_subvol_info(file, argp); | |
42e4b520 TM |
4920 | case BTRFS_IOC_GET_SUBVOL_ROOTREF: |
4921 | return btrfs_ioctl_get_subvol_rootref(file, argp); | |
23d0b79d TM |
4922 | case BTRFS_IOC_INO_LOOKUP_USER: |
4923 | return btrfs_ioctl_ino_lookup_user(file, argp); | |
f46b5a66 CH |
4924 | } |
4925 | ||
4926 | return -ENOTTY; | |
4927 | } | |
4c63c245 LD |
4928 | |
4929 | #ifdef CONFIG_COMPAT | |
4930 | long btrfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg) | |
4931 | { | |
2a362249 JM |
4932 | /* |
4933 | * These all access 32-bit values anyway so no further | |
4934 | * handling is necessary. | |
4935 | */ | |
4c63c245 | 4936 | switch (cmd) { |
4c63c245 LD |
4937 | case FS_IOC32_GETVERSION: |
4938 | cmd = FS_IOC_GETVERSION; | |
4939 | break; | |
4c63c245 LD |
4940 | } |
4941 | ||
4942 | return btrfs_ioctl(file, cmd, (unsigned long) compat_ptr(arg)); | |
4943 | } | |
4944 | #endif |