]>
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 | ||
0d7ed32c DS |
356 | /* |
357 | * Start exclusive operation @type, return true on success | |
358 | */ | |
c3e1f96c GR |
359 | bool btrfs_exclop_start(struct btrfs_fs_info *fs_info, |
360 | enum btrfs_exclusive_operation type) | |
361 | { | |
0d7ed32c DS |
362 | bool ret = false; |
363 | ||
364 | spin_lock(&fs_info->super_lock); | |
365 | if (fs_info->exclusive_operation == BTRFS_EXCLOP_NONE) { | |
366 | fs_info->exclusive_operation = type; | |
367 | ret = true; | |
368 | } | |
369 | spin_unlock(&fs_info->super_lock); | |
370 | ||
371 | return ret; | |
c3e1f96c GR |
372 | } |
373 | ||
374 | void btrfs_exclop_finish(struct btrfs_fs_info *fs_info) | |
375 | { | |
0d7ed32c | 376 | spin_lock(&fs_info->super_lock); |
c3e1f96c | 377 | WRITE_ONCE(fs_info->exclusive_operation, BTRFS_EXCLOP_NONE); |
0d7ed32c | 378 | spin_unlock(&fs_info->super_lock); |
66a2823c | 379 | sysfs_notify(&fs_info->fs_devices->fsid_kobj, NULL, "exclusive_operation"); |
c3e1f96c GR |
380 | } |
381 | ||
6cbff00f CH |
382 | static int btrfs_ioctl_getversion(struct file *file, int __user *arg) |
383 | { | |
496ad9aa | 384 | struct inode *inode = file_inode(file); |
6cbff00f CH |
385 | |
386 | return put_user(inode->i_generation, arg); | |
387 | } | |
f46b5a66 | 388 | |
b929c1d8 MPS |
389 | static noinline int btrfs_ioctl_fitrim(struct btrfs_fs_info *fs_info, |
390 | void __user *arg) | |
f7039b1d | 391 | { |
f7039b1d LD |
392 | struct btrfs_device *device; |
393 | struct request_queue *q; | |
394 | struct fstrim_range range; | |
395 | u64 minlen = ULLONG_MAX; | |
396 | u64 num_devices = 0; | |
397 | int ret; | |
398 | ||
399 | if (!capable(CAP_SYS_ADMIN)) | |
400 | return -EPERM; | |
401 | ||
1cb3dc3f NA |
402 | /* |
403 | * btrfs_trim_block_group() depends on space cache, which is not | |
404 | * available in zoned filesystem. So, disallow fitrim on a zoned | |
405 | * filesystem for now. | |
406 | */ | |
407 | if (btrfs_is_zoned(fs_info)) | |
408 | return -EOPNOTSUPP; | |
409 | ||
f35f06c3 FM |
410 | /* |
411 | * If the fs is mounted with nologreplay, which requires it to be | |
412 | * mounted in RO mode as well, we can not allow discard on free space | |
413 | * inside block groups, because log trees refer to extents that are not | |
414 | * pinned in a block group's free space cache (pinning the extents is | |
415 | * precisely the first phase of replaying a log tree). | |
416 | */ | |
417 | if (btrfs_test_opt(fs_info, NOLOGREPLAY)) | |
418 | return -EROFS; | |
419 | ||
1f78160c XG |
420 | rcu_read_lock(); |
421 | list_for_each_entry_rcu(device, &fs_info->fs_devices->devices, | |
422 | dev_list) { | |
f7039b1d LD |
423 | if (!device->bdev) |
424 | continue; | |
425 | q = bdev_get_queue(device->bdev); | |
426 | if (blk_queue_discard(q)) { | |
427 | num_devices++; | |
50d0446e | 428 | minlen = min_t(u64, q->limits.discard_granularity, |
f7039b1d LD |
429 | minlen); |
430 | } | |
431 | } | |
1f78160c | 432 | rcu_read_unlock(); |
f4c697e6 | 433 | |
f7039b1d LD |
434 | if (!num_devices) |
435 | return -EOPNOTSUPP; | |
f7039b1d LD |
436 | if (copy_from_user(&range, arg, sizeof(range))) |
437 | return -EFAULT; | |
6ba9fc8e QW |
438 | |
439 | /* | |
440 | * NOTE: Don't truncate the range using super->total_bytes. Bytenr of | |
441 | * block group is in the logical address space, which can be any | |
442 | * sectorsize aligned bytenr in the range [0, U64_MAX]. | |
443 | */ | |
444 | if (range.len < fs_info->sb->s_blocksize) | |
f4c697e6 | 445 | return -EINVAL; |
f7039b1d LD |
446 | |
447 | range.minlen = max(range.minlen, minlen); | |
2ff7e61e | 448 | ret = btrfs_trim_fs(fs_info, &range); |
f7039b1d LD |
449 | if (ret < 0) |
450 | return ret; | |
451 | ||
452 | if (copy_to_user(arg, &range, sizeof(range))) | |
453 | return -EFAULT; | |
454 | ||
455 | return 0; | |
456 | } | |
457 | ||
e1f60a65 | 458 | int __pure btrfs_is_empty_uuid(u8 *uuid) |
dd5f9615 | 459 | { |
46e0f66a CM |
460 | int i; |
461 | ||
462 | for (i = 0; i < BTRFS_UUID_SIZE; i++) { | |
463 | if (uuid[i]) | |
464 | return 0; | |
465 | } | |
466 | return 1; | |
dd5f9615 SB |
467 | } |
468 | ||
d5c12070 | 469 | static noinline int create_subvol(struct inode *dir, |
cb8e7090 | 470 | struct dentry *dentry, |
52f75f4f | 471 | const char *name, int namelen, |
8696c533 | 472 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 473 | { |
0b246afa | 474 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); |
f46b5a66 CH |
475 | struct btrfs_trans_handle *trans; |
476 | struct btrfs_key key; | |
49a3c4d9 | 477 | struct btrfs_root_item *root_item; |
f46b5a66 CH |
478 | struct btrfs_inode_item *inode_item; |
479 | struct extent_buffer *leaf; | |
d5c12070 | 480 | struct btrfs_root *root = BTRFS_I(dir)->root; |
76dda93c | 481 | struct btrfs_root *new_root; |
d5c12070 | 482 | struct btrfs_block_rsv block_rsv; |
95582b00 | 483 | struct timespec64 cur_time = current_time(dir); |
5662344b | 484 | struct inode *inode; |
f46b5a66 CH |
485 | int ret; |
486 | int err; | |
2dfb1e43 | 487 | dev_t anon_dev = 0; |
f46b5a66 | 488 | u64 objectid; |
3de4586c | 489 | u64 index = 0; |
f46b5a66 | 490 | |
49a3c4d9 DS |
491 | root_item = kzalloc(sizeof(*root_item), GFP_KERNEL); |
492 | if (!root_item) | |
493 | return -ENOMEM; | |
494 | ||
543068a2 | 495 | ret = btrfs_get_free_objectid(fs_info->tree_root, &objectid); |
2fbe8c8a | 496 | if (ret) |
49a3c4d9 | 497 | goto fail_free; |
6a912213 | 498 | |
2dfb1e43 QW |
499 | ret = get_anon_bdev(&anon_dev); |
500 | if (ret < 0) | |
501 | goto fail_free; | |
502 | ||
e09fe2d2 QW |
503 | /* |
504 | * Don't create subvolume whose level is not zero. Or qgroup will be | |
01327610 | 505 | * screwed up since it assumes subvolume qgroup's level to be 0. |
e09fe2d2 | 506 | */ |
49a3c4d9 DS |
507 | if (btrfs_qgroup_level(objectid)) { |
508 | ret = -ENOSPC; | |
509 | goto fail_free; | |
510 | } | |
e09fe2d2 | 511 | |
d5c12070 | 512 | btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP); |
9ed74f2d | 513 | /* |
d5c12070 MX |
514 | * The same as the snapshot creation, please see the comment |
515 | * of create_snapshot(). | |
9ed74f2d | 516 | */ |
c4c129db | 517 | ret = btrfs_subvolume_reserve_metadata(root, &block_rsv, 8, false); |
d5c12070 | 518 | if (ret) |
49a3c4d9 | 519 | goto fail_free; |
d5c12070 MX |
520 | |
521 | trans = btrfs_start_transaction(root, 0); | |
522 | if (IS_ERR(trans)) { | |
523 | ret = PTR_ERR(trans); | |
e85fde51 | 524 | btrfs_subvolume_release_metadata(root, &block_rsv); |
49a3c4d9 | 525 | goto fail_free; |
d5c12070 MX |
526 | } |
527 | trans->block_rsv = &block_rsv; | |
528 | trans->bytes_reserved = block_rsv.size; | |
f46b5a66 | 529 | |
a9377422 | 530 | ret = btrfs_qgroup_inherit(trans, 0, objectid, inherit); |
6f72c7e2 AJ |
531 | if (ret) |
532 | goto fail; | |
533 | ||
9631e4cc JB |
534 | leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0, |
535 | BTRFS_NESTING_NORMAL); | |
8e8a1e31 JB |
536 | if (IS_ERR(leaf)) { |
537 | ret = PTR_ERR(leaf); | |
538 | goto fail; | |
539 | } | |
f46b5a66 | 540 | |
f46b5a66 CH |
541 | btrfs_mark_buffer_dirty(leaf); |
542 | ||
49a3c4d9 | 543 | inode_item = &root_item->inode; |
3cae210f QW |
544 | btrfs_set_stack_inode_generation(inode_item, 1); |
545 | btrfs_set_stack_inode_size(inode_item, 3); | |
546 | btrfs_set_stack_inode_nlink(inode_item, 1); | |
da17066c | 547 | btrfs_set_stack_inode_nbytes(inode_item, |
0b246afa | 548 | fs_info->nodesize); |
3cae210f | 549 | btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755); |
f46b5a66 | 550 | |
49a3c4d9 DS |
551 | btrfs_set_root_flags(root_item, 0); |
552 | btrfs_set_root_limit(root_item, 0); | |
3cae210f | 553 | btrfs_set_stack_inode_flags(inode_item, BTRFS_INODE_ROOT_ITEM_INIT); |
08fe4db1 | 554 | |
49a3c4d9 DS |
555 | btrfs_set_root_bytenr(root_item, leaf->start); |
556 | btrfs_set_root_generation(root_item, trans->transid); | |
557 | btrfs_set_root_level(root_item, 0); | |
558 | btrfs_set_root_refs(root_item, 1); | |
559 | btrfs_set_root_used(root_item, leaf->len); | |
560 | btrfs_set_root_last_snapshot(root_item, 0); | |
f46b5a66 | 561 | |
49a3c4d9 DS |
562 | btrfs_set_root_generation_v2(root_item, |
563 | btrfs_root_generation(root_item)); | |
807fc790 | 564 | generate_random_guid(root_item->uuid); |
49a3c4d9 DS |
565 | btrfs_set_stack_timespec_sec(&root_item->otime, cur_time.tv_sec); |
566 | btrfs_set_stack_timespec_nsec(&root_item->otime, cur_time.tv_nsec); | |
567 | root_item->ctime = root_item->otime; | |
568 | btrfs_set_root_ctransid(root_item, trans->transid); | |
569 | btrfs_set_root_otransid(root_item, trans->transid); | |
f46b5a66 | 570 | |
925baedd | 571 | btrfs_tree_unlock(leaf); |
f46b5a66 | 572 | |
69948022 | 573 | btrfs_set_root_dirid(root_item, BTRFS_FIRST_FREE_OBJECTID); |
f46b5a66 CH |
574 | |
575 | key.objectid = objectid; | |
5d4f98a2 | 576 | key.offset = 0; |
962a298f | 577 | key.type = BTRFS_ROOT_ITEM_KEY; |
0b246afa | 578 | ret = btrfs_insert_root(trans, fs_info->tree_root, &key, |
49a3c4d9 | 579 | root_item); |
67addf29 FM |
580 | if (ret) { |
581 | /* | |
582 | * Since we don't abort the transaction in this case, free the | |
583 | * tree block so that we don't leak space and leave the | |
584 | * filesystem in an inconsistent state (an extent item in the | |
585 | * extent tree without backreferences). Also no need to have | |
586 | * the tree block locked since it is not in any tree at this | |
587 | * point, so no other task can find it and use it. | |
588 | */ | |
589 | btrfs_free_tree_block(trans, root, leaf, 0, 1); | |
590 | free_extent_buffer(leaf); | |
f46b5a66 | 591 | goto fail; |
67addf29 FM |
592 | } |
593 | ||
594 | free_extent_buffer(leaf); | |
595 | leaf = NULL; | |
f46b5a66 | 596 | |
76dda93c | 597 | key.offset = (u64)-1; |
2dfb1e43 | 598 | new_root = btrfs_get_new_fs_root(fs_info, objectid, anon_dev); |
79787eaa | 599 | if (IS_ERR(new_root)) { |
2dfb1e43 | 600 | free_anon_bdev(anon_dev); |
79787eaa | 601 | ret = PTR_ERR(new_root); |
66642832 | 602 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
603 | goto fail; |
604 | } | |
2dfb1e43 QW |
605 | /* Freeing will be done in btrfs_put_root() of new_root */ |
606 | anon_dev = 0; | |
76dda93c | 607 | |
221581e4 JB |
608 | ret = btrfs_record_root_in_trans(trans, new_root); |
609 | if (ret) { | |
610 | btrfs_put_root(new_root); | |
611 | btrfs_abort_transaction(trans, ret); | |
612 | goto fail; | |
613 | } | |
76dda93c | 614 | |
69948022 | 615 | ret = btrfs_create_subvol_root(trans, new_root, root); |
00246528 | 616 | btrfs_put_root(new_root); |
ce598979 MF |
617 | if (ret) { |
618 | /* We potentially lose an unused inode item here */ | |
66642832 | 619 | btrfs_abort_transaction(trans, ret); |
ce598979 MF |
620 | goto fail; |
621 | } | |
622 | ||
f46b5a66 CH |
623 | /* |
624 | * insert the directory item | |
625 | */ | |
877574e2 | 626 | ret = btrfs_set_inode_index(BTRFS_I(dir), &index); |
79787eaa | 627 | if (ret) { |
66642832 | 628 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
629 | goto fail; |
630 | } | |
3de4586c | 631 | |
684572df | 632 | ret = btrfs_insert_dir_item(trans, name, namelen, BTRFS_I(dir), &key, |
3de4586c | 633 | BTRFS_FT_DIR, index); |
79787eaa | 634 | if (ret) { |
66642832 | 635 | btrfs_abort_transaction(trans, ret); |
f46b5a66 | 636 | goto fail; |
79787eaa | 637 | } |
0660b5af | 638 | |
6ef06d27 | 639 | btrfs_i_size_write(BTRFS_I(dir), dir->i_size + namelen * 2); |
9a56fcd1 | 640 | ret = btrfs_update_inode(trans, root, BTRFS_I(dir)); |
c7e54b51 JB |
641 | if (ret) { |
642 | btrfs_abort_transaction(trans, ret); | |
643 | goto fail; | |
644 | } | |
52c26179 | 645 | |
6025c19f | 646 | ret = btrfs_add_root_ref(trans, objectid, root->root_key.objectid, |
4a0cc7ca | 647 | btrfs_ino(BTRFS_I(dir)), index, name, namelen); |
c7e54b51 JB |
648 | if (ret) { |
649 | btrfs_abort_transaction(trans, ret); | |
650 | goto fail; | |
651 | } | |
f46b5a66 | 652 | |
cdb345a8 | 653 | ret = btrfs_uuid_tree_add(trans, root_item->uuid, |
6bccf3ab | 654 | BTRFS_UUID_KEY_SUBVOL, objectid); |
dd5f9615 | 655 | if (ret) |
66642832 | 656 | btrfs_abort_transaction(trans, ret); |
dd5f9615 | 657 | |
f46b5a66 | 658 | fail: |
49a3c4d9 | 659 | kfree(root_item); |
d5c12070 MX |
660 | trans->block_rsv = NULL; |
661 | trans->bytes_reserved = 0; | |
e85fde51 | 662 | btrfs_subvolume_release_metadata(root, &block_rsv); |
de6e8200 | 663 | |
9babda9f | 664 | err = btrfs_commit_transaction(trans); |
f46b5a66 CH |
665 | if (err && !ret) |
666 | ret = err; | |
1a65e24b | 667 | |
5662344b TI |
668 | if (!ret) { |
669 | inode = btrfs_lookup_dentry(dir, dentry); | |
de6e8200 LB |
670 | if (IS_ERR(inode)) |
671 | return PTR_ERR(inode); | |
5662344b TI |
672 | d_instantiate(dentry, inode); |
673 | } | |
f46b5a66 | 674 | return ret; |
49a3c4d9 DS |
675 | |
676 | fail_free: | |
2dfb1e43 QW |
677 | if (anon_dev) |
678 | free_anon_bdev(anon_dev); | |
49a3c4d9 DS |
679 | kfree(root_item); |
680 | return ret; | |
f46b5a66 CH |
681 | } |
682 | ||
e9662f70 | 683 | static int create_snapshot(struct btrfs_root *root, struct inode *dir, |
9babda9f | 684 | struct dentry *dentry, bool readonly, |
e9662f70 | 685 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 686 | { |
0b246afa | 687 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); |
2e4bfab9 | 688 | struct inode *inode; |
f46b5a66 CH |
689 | struct btrfs_pending_snapshot *pending_snapshot; |
690 | struct btrfs_trans_handle *trans; | |
2e4bfab9 | 691 | int ret; |
f46b5a66 | 692 | |
92a7cc42 | 693 | if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) |
f46b5a66 CH |
694 | return -EINVAL; |
695 | ||
eede2bf3 OS |
696 | if (atomic_read(&root->nr_swapfiles)) { |
697 | btrfs_warn(fs_info, | |
698 | "cannot snapshot subvolume with active swapfile"); | |
699 | return -ETXTBSY; | |
700 | } | |
701 | ||
23269bf5 | 702 | pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_KERNEL); |
a1ee7362 DS |
703 | if (!pending_snapshot) |
704 | return -ENOMEM; | |
705 | ||
2dfb1e43 QW |
706 | ret = get_anon_bdev(&pending_snapshot->anon_dev); |
707 | if (ret < 0) | |
708 | goto free_pending; | |
b0c0ea63 | 709 | pending_snapshot->root_item = kzalloc(sizeof(struct btrfs_root_item), |
23269bf5 | 710 | GFP_KERNEL); |
8546b570 DS |
711 | pending_snapshot->path = btrfs_alloc_path(); |
712 | if (!pending_snapshot->root_item || !pending_snapshot->path) { | |
b0c0ea63 DS |
713 | ret = -ENOMEM; |
714 | goto free_pending; | |
715 | } | |
716 | ||
66d8f3dd MX |
717 | btrfs_init_block_rsv(&pending_snapshot->block_rsv, |
718 | BTRFS_BLOCK_RSV_TEMP); | |
d5c12070 MX |
719 | /* |
720 | * 1 - parent dir inode | |
721 | * 2 - dir entries | |
722 | * 1 - root item | |
723 | * 2 - root ref/backref | |
724 | * 1 - root of snapshot | |
dd5f9615 | 725 | * 1 - UUID item |
d5c12070 MX |
726 | */ |
727 | ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root, | |
dd5f9615 | 728 | &pending_snapshot->block_rsv, 8, |
ee3441b4 | 729 | false); |
d5c12070 | 730 | if (ret) |
c11fbb6e | 731 | goto free_pending; |
d5c12070 | 732 | |
3de4586c | 733 | pending_snapshot->dentry = dentry; |
f46b5a66 | 734 | pending_snapshot->root = root; |
b83cc969 | 735 | pending_snapshot->readonly = readonly; |
e9662f70 | 736 | pending_snapshot->dir = dir; |
8696c533 | 737 | pending_snapshot->inherit = inherit; |
a22285a6 | 738 | |
d5c12070 | 739 | trans = btrfs_start_transaction(root, 0); |
a22285a6 YZ |
740 | if (IS_ERR(trans)) { |
741 | ret = PTR_ERR(trans); | |
742 | goto fail; | |
743 | } | |
744 | ||
0b246afa | 745 | spin_lock(&fs_info->trans_lock); |
f46b5a66 CH |
746 | list_add(&pending_snapshot->list, |
747 | &trans->transaction->pending_snapshots); | |
0b246afa | 748 | spin_unlock(&fs_info->trans_lock); |
9babda9f NB |
749 | |
750 | ret = btrfs_commit_transaction(trans); | |
aec8030a | 751 | if (ret) |
c37b2b62 | 752 | goto fail; |
a22285a6 YZ |
753 | |
754 | ret = pending_snapshot->error; | |
755 | if (ret) | |
756 | goto fail; | |
757 | ||
d3797308 CM |
758 | ret = btrfs_orphan_cleanup(pending_snapshot->snap); |
759 | if (ret) | |
760 | goto fail; | |
761 | ||
2b0143b5 | 762 | inode = btrfs_lookup_dentry(d_inode(dentry->d_parent), dentry); |
2e4bfab9 YZ |
763 | if (IS_ERR(inode)) { |
764 | ret = PTR_ERR(inode); | |
765 | goto fail; | |
766 | } | |
5662344b | 767 | |
2e4bfab9 YZ |
768 | d_instantiate(dentry, inode); |
769 | ret = 0; | |
2dfb1e43 | 770 | pending_snapshot->anon_dev = 0; |
2e4bfab9 | 771 | fail: |
2dfb1e43 QW |
772 | /* Prevent double freeing of anon_dev */ |
773 | if (ret && pending_snapshot->snap) | |
774 | pending_snapshot->snap->anon_dev = 0; | |
00246528 | 775 | btrfs_put_root(pending_snapshot->snap); |
e85fde51 | 776 | btrfs_subvolume_release_metadata(root, &pending_snapshot->block_rsv); |
b0c0ea63 | 777 | free_pending: |
2dfb1e43 QW |
778 | if (pending_snapshot->anon_dev) |
779 | free_anon_bdev(pending_snapshot->anon_dev); | |
b0c0ea63 | 780 | kfree(pending_snapshot->root_item); |
8546b570 | 781 | btrfs_free_path(pending_snapshot->path); |
a1ee7362 DS |
782 | kfree(pending_snapshot); |
783 | ||
f46b5a66 CH |
784 | return ret; |
785 | } | |
786 | ||
4260f7c7 SW |
787 | /* copy of may_delete in fs/namei.c() |
788 | * Check whether we can remove a link victim from directory dir, check | |
789 | * whether the type of victim is right. | |
790 | * 1. We can't do it if dir is read-only (done in permission()) | |
791 | * 2. We should have write and exec permissions on dir | |
792 | * 3. We can't remove anything from append-only dir | |
793 | * 4. We can't do anything with immutable dir (done in permission()) | |
794 | * 5. If the sticky bit on dir is set we should either | |
795 | * a. be owner of dir, or | |
796 | * b. be owner of victim, or | |
797 | * c. have CAP_FOWNER capability | |
01327610 | 798 | * 6. If the victim is append-only or immutable we can't do anything with |
4260f7c7 SW |
799 | * links pointing to it. |
800 | * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR. | |
801 | * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR. | |
802 | * 9. We can't remove a root or mountpoint. | |
803 | * 10. We don't allow removal of NFS sillyrenamed files; it's handled by | |
804 | * nfs_async_unlink(). | |
805 | */ | |
806 | ||
67871254 | 807 | static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir) |
4260f7c7 SW |
808 | { |
809 | int error; | |
810 | ||
2b0143b5 | 811 | if (d_really_is_negative(victim)) |
4260f7c7 SW |
812 | return -ENOENT; |
813 | ||
2b0143b5 | 814 | BUG_ON(d_inode(victim->d_parent) != dir); |
4fa6b5ec | 815 | audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE); |
4260f7c7 | 816 | |
47291baa | 817 | error = inode_permission(&init_user_ns, dir, MAY_WRITE | MAY_EXEC); |
4260f7c7 SW |
818 | if (error) |
819 | return error; | |
820 | if (IS_APPEND(dir)) | |
821 | return -EPERM; | |
ba73d987 CB |
822 | if (check_sticky(&init_user_ns, dir, d_inode(victim)) || |
823 | IS_APPEND(d_inode(victim)) || IS_IMMUTABLE(d_inode(victim)) || | |
824 | IS_SWAPFILE(d_inode(victim))) | |
4260f7c7 SW |
825 | return -EPERM; |
826 | if (isdir) { | |
e36cb0b8 | 827 | if (!d_is_dir(victim)) |
4260f7c7 SW |
828 | return -ENOTDIR; |
829 | if (IS_ROOT(victim)) | |
830 | return -EBUSY; | |
e36cb0b8 | 831 | } else if (d_is_dir(victim)) |
4260f7c7 SW |
832 | return -EISDIR; |
833 | if (IS_DEADDIR(dir)) | |
834 | return -ENOENT; | |
835 | if (victim->d_flags & DCACHE_NFSFS_RENAMED) | |
836 | return -EBUSY; | |
837 | return 0; | |
838 | } | |
839 | ||
cb8e7090 CH |
840 | /* copy of may_create in fs/namei.c() */ |
841 | static inline int btrfs_may_create(struct inode *dir, struct dentry *child) | |
842 | { | |
2b0143b5 | 843 | if (d_really_is_positive(child)) |
cb8e7090 CH |
844 | return -EEXIST; |
845 | if (IS_DEADDIR(dir)) | |
846 | return -ENOENT; | |
47291baa | 847 | return inode_permission(&init_user_ns, dir, MAY_WRITE | MAY_EXEC); |
cb8e7090 CH |
848 | } |
849 | ||
850 | /* | |
851 | * Create a new subvolume below @parent. This is largely modeled after | |
852 | * sys_mkdirat and vfs_mkdir, but we only do a single component lookup | |
853 | * inside this filesystem so it's quite a bit simpler. | |
854 | */ | |
92872094 | 855 | static noinline int btrfs_mksubvol(const struct path *parent, |
52f75f4f | 856 | const char *name, int namelen, |
72fd032e | 857 | struct btrfs_root *snap_src, |
9babda9f | 858 | bool readonly, |
8696c533 | 859 | struct btrfs_qgroup_inherit *inherit) |
cb8e7090 | 860 | { |
0b246afa JM |
861 | struct inode *dir = d_inode(parent->dentry); |
862 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); | |
cb8e7090 CH |
863 | struct dentry *dentry; |
864 | int error; | |
865 | ||
00235411 AV |
866 | error = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT); |
867 | if (error == -EINTR) | |
868 | return error; | |
cb8e7090 CH |
869 | |
870 | dentry = lookup_one_len(name, parent->dentry, namelen); | |
871 | error = PTR_ERR(dentry); | |
872 | if (IS_ERR(dentry)) | |
873 | goto out_unlock; | |
874 | ||
76dda93c | 875 | error = btrfs_may_create(dir, dentry); |
cb8e7090 | 876 | if (error) |
a874a63e | 877 | goto out_dput; |
cb8e7090 | 878 | |
9c52057c CM |
879 | /* |
880 | * even if this name doesn't exist, we may get hash collisions. | |
881 | * check for them now when we can safely fail | |
882 | */ | |
883 | error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root, | |
884 | dir->i_ino, name, | |
885 | namelen); | |
886 | if (error) | |
887 | goto out_dput; | |
888 | ||
0b246afa | 889 | down_read(&fs_info->subvol_sem); |
76dda93c YZ |
890 | |
891 | if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0) | |
892 | goto out_up_read; | |
893 | ||
9babda9f NB |
894 | if (snap_src) |
895 | error = create_snapshot(snap_src, dir, dentry, readonly, inherit); | |
896 | else | |
897 | error = create_subvol(dir, dentry, name, namelen, inherit); | |
898 | ||
76dda93c YZ |
899 | if (!error) |
900 | fsnotify_mkdir(dir, dentry); | |
901 | out_up_read: | |
0b246afa | 902 | up_read(&fs_info->subvol_sem); |
cb8e7090 CH |
903 | out_dput: |
904 | dput(dentry); | |
905 | out_unlock: | |
64708539 | 906 | btrfs_inode_unlock(dir, 0); |
cb8e7090 CH |
907 | return error; |
908 | } | |
909 | ||
c11fbb6e RK |
910 | static noinline int btrfs_mksnapshot(const struct path *parent, |
911 | const char *name, int namelen, | |
912 | struct btrfs_root *root, | |
913 | bool readonly, | |
914 | struct btrfs_qgroup_inherit *inherit) | |
915 | { | |
916 | int ret; | |
917 | bool snapshot_force_cow = false; | |
918 | ||
919 | /* | |
920 | * Force new buffered writes to reserve space even when NOCOW is | |
921 | * possible. This is to avoid later writeback (running dealloc) to | |
922 | * fallback to COW mode and unexpectedly fail with ENOSPC. | |
923 | */ | |
924 | btrfs_drew_read_lock(&root->snapshot_lock); | |
925 | ||
f9baa501 | 926 | ret = btrfs_start_delalloc_snapshot(root, false); |
c11fbb6e RK |
927 | if (ret) |
928 | goto out; | |
929 | ||
930 | /* | |
931 | * All previous writes have started writeback in NOCOW mode, so now | |
932 | * we force future writes to fallback to COW mode during snapshot | |
933 | * creation. | |
934 | */ | |
935 | atomic_inc(&root->snapshot_force_cow); | |
936 | snapshot_force_cow = true; | |
937 | ||
938 | btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1); | |
939 | ||
940 | ret = btrfs_mksubvol(parent, name, namelen, | |
941 | root, readonly, inherit); | |
942 | out: | |
943 | if (snapshot_force_cow) | |
944 | atomic_dec(&root->snapshot_force_cow); | |
945 | btrfs_drew_read_unlock(&root->snapshot_lock); | |
946 | return ret; | |
947 | } | |
948 | ||
4cb5300b CM |
949 | /* |
950 | * When we're defragging a range, we don't want to kick it off again | |
951 | * if it is really just waiting for delalloc to send it down. | |
952 | * If we find a nice big extent or delalloc range for the bytes in the | |
953 | * file you want to defrag, we return 0 to let you know to skip this | |
954 | * part of the file | |
955 | */ | |
aab110ab | 956 | static int check_defrag_in_cache(struct inode *inode, u64 offset, u32 thresh) |
4cb5300b CM |
957 | { |
958 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; | |
959 | struct extent_map *em = NULL; | |
960 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
961 | u64 end; | |
962 | ||
963 | read_lock(&em_tree->lock); | |
09cbfeaf | 964 | em = lookup_extent_mapping(em_tree, offset, PAGE_SIZE); |
4cb5300b CM |
965 | read_unlock(&em_tree->lock); |
966 | ||
967 | if (em) { | |
968 | end = extent_map_end(em); | |
969 | free_extent_map(em); | |
970 | if (end - offset > thresh) | |
971 | return 0; | |
972 | } | |
973 | /* if we already have a nice delalloc here, just stop */ | |
974 | thresh /= 2; | |
975 | end = count_range_bits(io_tree, &offset, offset + thresh, | |
976 | thresh, EXTENT_DELALLOC, 1); | |
977 | if (end >= thresh) | |
978 | return 0; | |
979 | return 1; | |
980 | } | |
981 | ||
982 | /* | |
983 | * helper function to walk through a file and find extents | |
984 | * newer than a specific transid, and smaller than thresh. | |
985 | * | |
986 | * This is used by the defragging code to find new and small | |
987 | * extents | |
988 | */ | |
989 | static int find_new_extents(struct btrfs_root *root, | |
990 | struct inode *inode, u64 newer_than, | |
aab110ab | 991 | u64 *off, u32 thresh) |
4cb5300b CM |
992 | { |
993 | struct btrfs_path *path; | |
994 | struct btrfs_key min_key; | |
4cb5300b CM |
995 | struct extent_buffer *leaf; |
996 | struct btrfs_file_extent_item *extent; | |
997 | int type; | |
998 | int ret; | |
4a0cc7ca | 999 | u64 ino = btrfs_ino(BTRFS_I(inode)); |
4cb5300b CM |
1000 | |
1001 | path = btrfs_alloc_path(); | |
1002 | if (!path) | |
1003 | return -ENOMEM; | |
1004 | ||
a4689d2b | 1005 | min_key.objectid = ino; |
4cb5300b CM |
1006 | min_key.type = BTRFS_EXTENT_DATA_KEY; |
1007 | min_key.offset = *off; | |
1008 | ||
67871254 | 1009 | while (1) { |
6174d3cb | 1010 | ret = btrfs_search_forward(root, &min_key, path, newer_than); |
4cb5300b CM |
1011 | if (ret != 0) |
1012 | goto none; | |
f094c9bd | 1013 | process_slot: |
a4689d2b | 1014 | if (min_key.objectid != ino) |
4cb5300b CM |
1015 | goto none; |
1016 | if (min_key.type != BTRFS_EXTENT_DATA_KEY) | |
1017 | goto none; | |
1018 | ||
1019 | leaf = path->nodes[0]; | |
1020 | extent = btrfs_item_ptr(leaf, path->slots[0], | |
1021 | struct btrfs_file_extent_item); | |
1022 | ||
1023 | type = btrfs_file_extent_type(leaf, extent); | |
1024 | if (type == BTRFS_FILE_EXTENT_REG && | |
1025 | btrfs_file_extent_num_bytes(leaf, extent) < thresh && | |
1026 | check_defrag_in_cache(inode, min_key.offset, thresh)) { | |
1027 | *off = min_key.offset; | |
1028 | btrfs_free_path(path); | |
1029 | return 0; | |
1030 | } | |
1031 | ||
f094c9bd FM |
1032 | path->slots[0]++; |
1033 | if (path->slots[0] < btrfs_header_nritems(leaf)) { | |
1034 | btrfs_item_key_to_cpu(leaf, &min_key, path->slots[0]); | |
1035 | goto process_slot; | |
1036 | } | |
1037 | ||
4cb5300b CM |
1038 | if (min_key.offset == (u64)-1) |
1039 | goto none; | |
1040 | ||
1041 | min_key.offset++; | |
1042 | btrfs_release_path(path); | |
1043 | } | |
1044 | none: | |
1045 | btrfs_free_path(path); | |
1046 | return -ENOENT; | |
1047 | } | |
1048 | ||
6c282eb4 | 1049 | static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start) |
17ce6ef8 LB |
1050 | { |
1051 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
6c282eb4 LZ |
1052 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; |
1053 | struct extent_map *em; | |
09cbfeaf | 1054 | u64 len = PAGE_SIZE; |
17ce6ef8 | 1055 | |
6c282eb4 LZ |
1056 | /* |
1057 | * hopefully we have this extent in the tree already, try without | |
1058 | * the full extent lock | |
1059 | */ | |
17ce6ef8 | 1060 | read_lock(&em_tree->lock); |
6c282eb4 | 1061 | em = lookup_extent_mapping(em_tree, start, len); |
17ce6ef8 LB |
1062 | read_unlock(&em_tree->lock); |
1063 | ||
6c282eb4 | 1064 | if (!em) { |
308d9800 FM |
1065 | struct extent_state *cached = NULL; |
1066 | u64 end = start + len - 1; | |
1067 | ||
6c282eb4 | 1068 | /* get the big lock and read metadata off disk */ |
ff13db41 | 1069 | lock_extent_bits(io_tree, start, end, &cached); |
39b07b5d | 1070 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start, len); |
e43bbe5e | 1071 | unlock_extent_cached(io_tree, start, end, &cached); |
6c282eb4 LZ |
1072 | |
1073 | if (IS_ERR(em)) | |
1074 | return NULL; | |
1075 | } | |
1076 | ||
1077 | return em; | |
1078 | } | |
17ce6ef8 | 1079 | |
6c282eb4 LZ |
1080 | static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em) |
1081 | { | |
1082 | struct extent_map *next; | |
1083 | bool ret = true; | |
1084 | ||
1085 | /* this is the last extent */ | |
1086 | if (em->start + em->len >= i_size_read(inode)) | |
1087 | return false; | |
1088 | ||
1089 | next = defrag_lookup_extent(inode, em->start + em->len); | |
e9512d72 CM |
1090 | if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE) |
1091 | ret = false; | |
1092 | else if ((em->block_start + em->block_len == next->block_start) && | |
ee22184b | 1093 | (em->block_len > SZ_128K && next->block_len > SZ_128K)) |
6c282eb4 LZ |
1094 | ret = false; |
1095 | ||
1096 | free_extent_map(next); | |
17ce6ef8 LB |
1097 | return ret; |
1098 | } | |
1099 | ||
aab110ab | 1100 | static int should_defrag_range(struct inode *inode, u64 start, u32 thresh, |
a43a2111 AM |
1101 | u64 *last_len, u64 *skip, u64 *defrag_end, |
1102 | int compress) | |
940100a4 | 1103 | { |
6c282eb4 | 1104 | struct extent_map *em; |
940100a4 | 1105 | int ret = 1; |
6c282eb4 | 1106 | bool next_mergeable = true; |
4a3560c4 | 1107 | bool prev_mergeable = true; |
940100a4 CM |
1108 | |
1109 | /* | |
008873ea | 1110 | * make sure that once we start defragging an extent, we keep on |
940100a4 CM |
1111 | * defragging it |
1112 | */ | |
1113 | if (start < *defrag_end) | |
1114 | return 1; | |
1115 | ||
1116 | *skip = 0; | |
1117 | ||
6c282eb4 LZ |
1118 | em = defrag_lookup_extent(inode, start); |
1119 | if (!em) | |
1120 | return 0; | |
940100a4 CM |
1121 | |
1122 | /* this will cover holes, and inline extents */ | |
17ce6ef8 | 1123 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { |
940100a4 | 1124 | ret = 0; |
17ce6ef8 LB |
1125 | goto out; |
1126 | } | |
1127 | ||
4a3560c4 LB |
1128 | if (!*defrag_end) |
1129 | prev_mergeable = false; | |
1130 | ||
6c282eb4 | 1131 | next_mergeable = defrag_check_next_extent(inode, em); |
940100a4 | 1132 | /* |
6c282eb4 LZ |
1133 | * we hit a real extent, if it is big or the next extent is not a |
1134 | * real extent, don't bother defragging it | |
940100a4 | 1135 | */ |
a43a2111 | 1136 | if (!compress && (*last_len == 0 || *last_len >= thresh) && |
4a3560c4 | 1137 | (em->len >= thresh || (!next_mergeable && !prev_mergeable))) |
940100a4 | 1138 | ret = 0; |
17ce6ef8 | 1139 | out: |
940100a4 CM |
1140 | /* |
1141 | * last_len ends up being a counter of how many bytes we've defragged. | |
1142 | * every time we choose not to defrag an extent, we reset *last_len | |
1143 | * so that the next tiny extent will force a defrag. | |
1144 | * | |
1145 | * The end result of this is that tiny extents before a single big | |
1146 | * extent will force at least part of that big extent to be defragged. | |
1147 | */ | |
1148 | if (ret) { | |
940100a4 CM |
1149 | *defrag_end = extent_map_end(em); |
1150 | } else { | |
1151 | *last_len = 0; | |
1152 | *skip = extent_map_end(em); | |
1153 | *defrag_end = 0; | |
1154 | } | |
1155 | ||
1156 | free_extent_map(em); | |
1157 | return ret; | |
1158 | } | |
1159 | ||
4cb5300b CM |
1160 | /* |
1161 | * it doesn't do much good to defrag one or two pages | |
1162 | * at a time. This pulls in a nice chunk of pages | |
1163 | * to COW and defrag. | |
1164 | * | |
1165 | * It also makes sure the delalloc code has enough | |
1166 | * dirty data to avoid making new small extents as part | |
1167 | * of the defrag | |
1168 | * | |
1169 | * It's a good idea to start RA on this range | |
1170 | * before calling this. | |
1171 | */ | |
1172 | static int cluster_pages_for_defrag(struct inode *inode, | |
1173 | struct page **pages, | |
1174 | unsigned long start_index, | |
c41570c9 | 1175 | unsigned long num_pages) |
f46b5a66 | 1176 | { |
4cb5300b CM |
1177 | unsigned long file_end; |
1178 | u64 isize = i_size_read(inode); | |
1179 | u64 page_start; | |
1180 | u64 page_end; | |
1f12bd06 | 1181 | u64 page_cnt; |
a1fbc675 | 1182 | u64 start = (u64)start_index << PAGE_SHIFT; |
7f458a38 | 1183 | u64 search_start; |
4cb5300b CM |
1184 | int ret; |
1185 | int i; | |
1186 | int i_done; | |
3eaa2885 | 1187 | struct btrfs_ordered_extent *ordered; |
4cb5300b | 1188 | struct extent_state *cached_state = NULL; |
600a45e1 | 1189 | struct extent_io_tree *tree; |
364ecf36 | 1190 | struct extent_changeset *data_reserved = NULL; |
3b16a4e3 | 1191 | gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); |
4cb5300b | 1192 | |
09cbfeaf | 1193 | file_end = (isize - 1) >> PAGE_SHIFT; |
1f12bd06 LB |
1194 | if (!isize || start_index > file_end) |
1195 | return 0; | |
1196 | ||
1197 | page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1); | |
4cb5300b | 1198 | |
e5b7231e | 1199 | ret = btrfs_delalloc_reserve_space(BTRFS_I(inode), &data_reserved, |
a1fbc675 | 1200 | start, page_cnt << PAGE_SHIFT); |
4cb5300b CM |
1201 | if (ret) |
1202 | return ret; | |
4cb5300b | 1203 | i_done = 0; |
600a45e1 | 1204 | tree = &BTRFS_I(inode)->io_tree; |
4cb5300b CM |
1205 | |
1206 | /* step one, lock all the pages */ | |
1f12bd06 | 1207 | for (i = 0; i < page_cnt; i++) { |
4cb5300b | 1208 | struct page *page; |
600a45e1 | 1209 | again: |
a94733d0 | 1210 | page = find_or_create_page(inode->i_mapping, |
600a45e1 | 1211 | start_index + i, mask); |
4cb5300b CM |
1212 | if (!page) |
1213 | break; | |
1214 | ||
32443de3 QW |
1215 | ret = set_page_extent_mapped(page); |
1216 | if (ret < 0) { | |
1217 | unlock_page(page); | |
1218 | put_page(page); | |
1219 | break; | |
1220 | } | |
1221 | ||
600a45e1 | 1222 | page_start = page_offset(page); |
09cbfeaf | 1223 | page_end = page_start + PAGE_SIZE - 1; |
600a45e1 | 1224 | while (1) { |
308d9800 | 1225 | lock_extent_bits(tree, page_start, page_end, |
ff13db41 | 1226 | &cached_state); |
c3504372 | 1227 | ordered = btrfs_lookup_ordered_extent(BTRFS_I(inode), |
600a45e1 | 1228 | page_start); |
308d9800 | 1229 | unlock_extent_cached(tree, page_start, page_end, |
e43bbe5e | 1230 | &cached_state); |
600a45e1 MX |
1231 | if (!ordered) |
1232 | break; | |
1233 | ||
1234 | unlock_page(page); | |
c0a43603 | 1235 | btrfs_start_ordered_extent(ordered, 1); |
600a45e1 MX |
1236 | btrfs_put_ordered_extent(ordered); |
1237 | lock_page(page); | |
1f12bd06 LB |
1238 | /* |
1239 | * we unlocked the page above, so we need check if | |
1240 | * it was released or not. | |
1241 | */ | |
1242 | if (page->mapping != inode->i_mapping) { | |
1243 | unlock_page(page); | |
09cbfeaf | 1244 | put_page(page); |
1f12bd06 LB |
1245 | goto again; |
1246 | } | |
600a45e1 MX |
1247 | } |
1248 | ||
4cb5300b CM |
1249 | if (!PageUptodate(page)) { |
1250 | btrfs_readpage(NULL, page); | |
1251 | lock_page(page); | |
1252 | if (!PageUptodate(page)) { | |
1253 | unlock_page(page); | |
09cbfeaf | 1254 | put_page(page); |
4cb5300b CM |
1255 | ret = -EIO; |
1256 | break; | |
1257 | } | |
1258 | } | |
600a45e1 | 1259 | |
600a45e1 MX |
1260 | if (page->mapping != inode->i_mapping) { |
1261 | unlock_page(page); | |
09cbfeaf | 1262 | put_page(page); |
600a45e1 MX |
1263 | goto again; |
1264 | } | |
1265 | ||
4cb5300b CM |
1266 | pages[i] = page; |
1267 | i_done++; | |
1268 | } | |
1269 | if (!i_done || ret) | |
1270 | goto out; | |
1271 | ||
1751e8a6 | 1272 | if (!(inode->i_sb->s_flags & SB_ACTIVE)) |
4cb5300b CM |
1273 | goto out; |
1274 | ||
1275 | /* | |
1276 | * so now we have a nice long stream of locked | |
1277 | * and up to date pages, lets wait on them | |
1278 | */ | |
1279 | for (i = 0; i < i_done; i++) | |
1280 | wait_on_page_writeback(pages[i]); | |
1281 | ||
1282 | page_start = page_offset(pages[0]); | |
09cbfeaf | 1283 | page_end = page_offset(pages[i_done - 1]) + PAGE_SIZE; |
4cb5300b CM |
1284 | |
1285 | lock_extent_bits(&BTRFS_I(inode)->io_tree, | |
ff13db41 | 1286 | page_start, page_end - 1, &cached_state); |
7f458a38 FM |
1287 | |
1288 | /* | |
1289 | * When defragmenting we skip ranges that have holes or inline extents, | |
1290 | * (check should_defrag_range()), to avoid unnecessary IO and wasting | |
1291 | * space. At btrfs_defrag_file(), we check if a range should be defragged | |
1292 | * before locking the inode and then, if it should, we trigger a sync | |
1293 | * page cache readahead - we lock the inode only after that to avoid | |
1294 | * blocking for too long other tasks that possibly want to operate on | |
1295 | * other file ranges. But before we were able to get the inode lock, | |
1296 | * some other task may have punched a hole in the range, or we may have | |
1297 | * now an inline extent, in which case we should not defrag. So check | |
1298 | * for that here, where we have the inode and the range locked, and bail | |
1299 | * out if that happened. | |
1300 | */ | |
1301 | search_start = page_start; | |
1302 | while (search_start < page_end) { | |
1303 | struct extent_map *em; | |
1304 | ||
1305 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, search_start, | |
1306 | page_end - search_start); | |
1307 | if (IS_ERR(em)) { | |
1308 | ret = PTR_ERR(em); | |
1309 | goto out_unlock_range; | |
1310 | } | |
1311 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { | |
1312 | free_extent_map(em); | |
1313 | /* Ok, 0 means we did not defrag anything */ | |
1314 | ret = 0; | |
1315 | goto out_unlock_range; | |
1316 | } | |
1317 | search_start = extent_map_end(em); | |
1318 | free_extent_map(em); | |
1319 | } | |
1320 | ||
4cb5300b | 1321 | clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, |
e182163d OS |
1322 | page_end - 1, EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | |
1323 | EXTENT_DEFRAG, 0, 0, &cached_state); | |
4cb5300b | 1324 | |
1f12bd06 | 1325 | if (i_done != page_cnt) { |
9e0baf60 | 1326 | spin_lock(&BTRFS_I(inode)->lock); |
28c4a3e2 | 1327 | btrfs_mod_outstanding_extents(BTRFS_I(inode), 1); |
9e0baf60 | 1328 | spin_unlock(&BTRFS_I(inode)->lock); |
86d52921 | 1329 | btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, |
a1fbc675 | 1330 | start, (page_cnt - i_done) << PAGE_SHIFT, true); |
4cb5300b CM |
1331 | } |
1332 | ||
1333 | ||
9e8a4a8b | 1334 | set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1, |
018ed4f7 | 1335 | &cached_state); |
4cb5300b CM |
1336 | |
1337 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
e43bbe5e | 1338 | page_start, page_end - 1, &cached_state); |
4cb5300b CM |
1339 | |
1340 | for (i = 0; i < i_done; i++) { | |
1341 | clear_page_dirty_for_io(pages[i]); | |
1342 | ClearPageChecked(pages[i]); | |
4cb5300b CM |
1343 | set_page_dirty(pages[i]); |
1344 | unlock_page(pages[i]); | |
09cbfeaf | 1345 | put_page(pages[i]); |
4cb5300b | 1346 | } |
8702ba93 | 1347 | btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT); |
364ecf36 | 1348 | extent_changeset_free(data_reserved); |
4cb5300b | 1349 | return i_done; |
7f458a38 FM |
1350 | |
1351 | out_unlock_range: | |
1352 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
1353 | page_start, page_end - 1, &cached_state); | |
4cb5300b CM |
1354 | out: |
1355 | for (i = 0; i < i_done; i++) { | |
1356 | unlock_page(pages[i]); | |
09cbfeaf | 1357 | put_page(pages[i]); |
4cb5300b | 1358 | } |
86d52921 | 1359 | btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, |
a1fbc675 | 1360 | start, page_cnt << PAGE_SHIFT, true); |
8702ba93 | 1361 | btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT); |
364ecf36 | 1362 | extent_changeset_free(data_reserved); |
4cb5300b CM |
1363 | return ret; |
1364 | ||
1365 | } | |
1366 | ||
1367 | int btrfs_defrag_file(struct inode *inode, struct file *file, | |
1368 | struct btrfs_ioctl_defrag_range_args *range, | |
1369 | u64 newer_than, unsigned long max_to_defrag) | |
1370 | { | |
0b246afa | 1371 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
4cb5300b | 1372 | struct btrfs_root *root = BTRFS_I(inode)->root; |
4cb5300b | 1373 | struct file_ra_state *ra = NULL; |
f46b5a66 | 1374 | unsigned long last_index; |
151a31b2 | 1375 | u64 isize = i_size_read(inode); |
940100a4 CM |
1376 | u64 last_len = 0; |
1377 | u64 skip = 0; | |
1378 | u64 defrag_end = 0; | |
4cb5300b | 1379 | u64 newer_off = range->start; |
f46b5a66 | 1380 | unsigned long i; |
008873ea | 1381 | unsigned long ra_index = 0; |
f46b5a66 | 1382 | int ret; |
4cb5300b | 1383 | int defrag_count = 0; |
1a419d85 | 1384 | int compress_type = BTRFS_COMPRESS_ZLIB; |
aab110ab | 1385 | u32 extent_thresh = range->extent_thresh; |
09cbfeaf | 1386 | unsigned long max_cluster = SZ_256K >> PAGE_SHIFT; |
c41570c9 | 1387 | unsigned long cluster = max_cluster; |
ee22184b | 1388 | u64 new_align = ~((u64)SZ_128K - 1); |
4cb5300b | 1389 | struct page **pages = NULL; |
1e2ef46d | 1390 | bool do_compress = range->flags & BTRFS_DEFRAG_RANGE_COMPRESS; |
4cb5300b | 1391 | |
0abd5b17 LB |
1392 | if (isize == 0) |
1393 | return 0; | |
1394 | ||
1395 | if (range->start >= isize) | |
1396 | return -EINVAL; | |
1a419d85 | 1397 | |
1e2ef46d | 1398 | if (do_compress) { |
ce96b7ff | 1399 | if (range->compress_type >= BTRFS_NR_COMPRESS_TYPES) |
1a419d85 LZ |
1400 | return -EINVAL; |
1401 | if (range->compress_type) | |
1402 | compress_type = range->compress_type; | |
1403 | } | |
f46b5a66 | 1404 | |
0abd5b17 | 1405 | if (extent_thresh == 0) |
ee22184b | 1406 | extent_thresh = SZ_256K; |
940100a4 | 1407 | |
4cb5300b | 1408 | /* |
0a52d108 DS |
1409 | * If we were not given a file, allocate a readahead context. As |
1410 | * readahead is just an optimization, defrag will work without it so | |
1411 | * we don't error out. | |
4cb5300b CM |
1412 | */ |
1413 | if (!file) { | |
63e727ec | 1414 | ra = kzalloc(sizeof(*ra), GFP_KERNEL); |
0a52d108 DS |
1415 | if (ra) |
1416 | file_ra_state_init(ra, inode->i_mapping); | |
4cb5300b CM |
1417 | } else { |
1418 | ra = &file->f_ra; | |
1419 | } | |
1420 | ||
63e727ec | 1421 | pages = kmalloc_array(max_cluster, sizeof(struct page *), GFP_KERNEL); |
4cb5300b CM |
1422 | if (!pages) { |
1423 | ret = -ENOMEM; | |
1424 | goto out_ra; | |
1425 | } | |
1426 | ||
1427 | /* find the last page to defrag */ | |
1e701a32 | 1428 | if (range->start + range->len > range->start) { |
151a31b2 | 1429 | last_index = min_t(u64, isize - 1, |
09cbfeaf | 1430 | range->start + range->len - 1) >> PAGE_SHIFT; |
1e701a32 | 1431 | } else { |
09cbfeaf | 1432 | last_index = (isize - 1) >> PAGE_SHIFT; |
1e701a32 CM |
1433 | } |
1434 | ||
4cb5300b CM |
1435 | if (newer_than) { |
1436 | ret = find_new_extents(root, inode, newer_than, | |
ee22184b | 1437 | &newer_off, SZ_64K); |
4cb5300b CM |
1438 | if (!ret) { |
1439 | range->start = newer_off; | |
1440 | /* | |
1441 | * we always align our defrag to help keep | |
1442 | * the extents in the file evenly spaced | |
1443 | */ | |
09cbfeaf | 1444 | i = (newer_off & new_align) >> PAGE_SHIFT; |
4cb5300b CM |
1445 | } else |
1446 | goto out_ra; | |
1447 | } else { | |
09cbfeaf | 1448 | i = range->start >> PAGE_SHIFT; |
4cb5300b CM |
1449 | } |
1450 | if (!max_to_defrag) | |
070034bd | 1451 | max_to_defrag = last_index - i + 1; |
4cb5300b | 1452 | |
2a0f7f57 LZ |
1453 | /* |
1454 | * make writeback starts from i, so the defrag range can be | |
1455 | * written sequentially. | |
1456 | */ | |
1457 | if (i < inode->i_mapping->writeback_index) | |
1458 | inode->i_mapping->writeback_index = i; | |
1459 | ||
f7f43cc8 | 1460 | while (i <= last_index && defrag_count < max_to_defrag && |
09cbfeaf | 1461 | (i < DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE))) { |
4cb5300b CM |
1462 | /* |
1463 | * make sure we stop running if someone unmounts | |
1464 | * the FS | |
1465 | */ | |
1751e8a6 | 1466 | if (!(inode->i_sb->s_flags & SB_ACTIVE)) |
4cb5300b CM |
1467 | break; |
1468 | ||
0b246afa JM |
1469 | if (btrfs_defrag_cancelled(fs_info)) { |
1470 | btrfs_debug(fs_info, "defrag_file cancelled"); | |
210549eb | 1471 | ret = -EAGAIN; |
50535db8 | 1472 | goto error; |
210549eb DS |
1473 | } |
1474 | ||
09cbfeaf | 1475 | if (!should_defrag_range(inode, (u64)i << PAGE_SHIFT, |
6c282eb4 | 1476 | extent_thresh, &last_len, &skip, |
1e2ef46d | 1477 | &defrag_end, do_compress)){ |
940100a4 CM |
1478 | unsigned long next; |
1479 | /* | |
1480 | * the should_defrag function tells us how much to skip | |
1481 | * bump our counter by the suggested amount | |
1482 | */ | |
09cbfeaf | 1483 | next = DIV_ROUND_UP(skip, PAGE_SIZE); |
940100a4 CM |
1484 | i = max(i + 1, next); |
1485 | continue; | |
1486 | } | |
008873ea LZ |
1487 | |
1488 | if (!newer_than) { | |
09cbfeaf KS |
1489 | cluster = (PAGE_ALIGN(defrag_end) >> |
1490 | PAGE_SHIFT) - i; | |
008873ea LZ |
1491 | cluster = min(cluster, max_cluster); |
1492 | } else { | |
1493 | cluster = max_cluster; | |
1494 | } | |
1495 | ||
008873ea LZ |
1496 | if (i + cluster > ra_index) { |
1497 | ra_index = max(i, ra_index); | |
0a52d108 | 1498 | if (ra) |
d3c0bab5 DS |
1499 | page_cache_sync_readahead(inode->i_mapping, ra, |
1500 | file, ra_index, cluster); | |
e4826a5b | 1501 | ra_index += cluster; |
008873ea | 1502 | } |
940100a4 | 1503 | |
64708539 | 1504 | btrfs_inode_lock(inode, 0); |
eede2bf3 OS |
1505 | if (IS_SWAPFILE(inode)) { |
1506 | ret = -ETXTBSY; | |
1507 | } else { | |
1508 | if (do_compress) | |
1509 | BTRFS_I(inode)->defrag_compress = compress_type; | |
1510 | ret = cluster_pages_for_defrag(inode, pages, i, cluster); | |
1511 | } | |
ecb8bea8 | 1512 | if (ret < 0) { |
64708539 | 1513 | btrfs_inode_unlock(inode, 0); |
4cb5300b | 1514 | goto out_ra; |
ecb8bea8 | 1515 | } |
4cb5300b CM |
1516 | |
1517 | defrag_count += ret; | |
d0e1d66b | 1518 | balance_dirty_pages_ratelimited(inode->i_mapping); |
64708539 | 1519 | btrfs_inode_unlock(inode, 0); |
4cb5300b CM |
1520 | |
1521 | if (newer_than) { | |
1522 | if (newer_off == (u64)-1) | |
1523 | break; | |
1524 | ||
e1f041e1 LB |
1525 | if (ret > 0) |
1526 | i += ret; | |
1527 | ||
4cb5300b | 1528 | newer_off = max(newer_off + 1, |
09cbfeaf | 1529 | (u64)i << PAGE_SHIFT); |
4cb5300b | 1530 | |
ee22184b BL |
1531 | ret = find_new_extents(root, inode, newer_than, |
1532 | &newer_off, SZ_64K); | |
4cb5300b CM |
1533 | if (!ret) { |
1534 | range->start = newer_off; | |
09cbfeaf | 1535 | i = (newer_off & new_align) >> PAGE_SHIFT; |
4cb5300b CM |
1536 | } else { |
1537 | break; | |
f46b5a66 | 1538 | } |
4cb5300b | 1539 | } else { |
008873ea | 1540 | if (ret > 0) { |
cbcc8326 | 1541 | i += ret; |
09cbfeaf | 1542 | last_len += ret << PAGE_SHIFT; |
008873ea | 1543 | } else { |
cbcc8326 | 1544 | i++; |
008873ea LZ |
1545 | last_len = 0; |
1546 | } | |
f46b5a66 | 1547 | } |
f46b5a66 CH |
1548 | } |
1549 | ||
50535db8 TT |
1550 | ret = defrag_count; |
1551 | error: | |
dec8ef90 | 1552 | if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) { |
1e701a32 | 1553 | filemap_flush(inode->i_mapping); |
dec8ef90 FM |
1554 | if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, |
1555 | &BTRFS_I(inode)->runtime_flags)) | |
1556 | filemap_flush(inode->i_mapping); | |
1557 | } | |
1e701a32 | 1558 | |
1a419d85 | 1559 | if (range->compress_type == BTRFS_COMPRESS_LZO) { |
0b246afa | 1560 | btrfs_set_fs_incompat(fs_info, COMPRESS_LZO); |
5c1aab1d NT |
1561 | } else if (range->compress_type == BTRFS_COMPRESS_ZSTD) { |
1562 | btrfs_set_fs_incompat(fs_info, COMPRESS_ZSTD); | |
1a419d85 LZ |
1563 | } |
1564 | ||
4cb5300b | 1565 | out_ra: |
1e2ef46d | 1566 | if (do_compress) { |
64708539 | 1567 | btrfs_inode_lock(inode, 0); |
eec63c65 | 1568 | BTRFS_I(inode)->defrag_compress = BTRFS_COMPRESS_NONE; |
64708539 | 1569 | btrfs_inode_unlock(inode, 0); |
633085c7 | 1570 | } |
4cb5300b CM |
1571 | if (!file) |
1572 | kfree(ra); | |
1573 | kfree(pages); | |
940100a4 | 1574 | return ret; |
f46b5a66 CH |
1575 | } |
1576 | ||
198605a8 | 1577 | static noinline int btrfs_ioctl_resize(struct file *file, |
76dda93c | 1578 | void __user *arg) |
f46b5a66 | 1579 | { |
0b246afa JM |
1580 | struct inode *inode = file_inode(file); |
1581 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
f46b5a66 CH |
1582 | u64 new_size; |
1583 | u64 old_size; | |
1584 | u64 devid = 1; | |
0b246afa | 1585 | struct btrfs_root *root = BTRFS_I(inode)->root; |
f46b5a66 CH |
1586 | struct btrfs_ioctl_vol_args *vol_args; |
1587 | struct btrfs_trans_handle *trans; | |
1588 | struct btrfs_device *device = NULL; | |
1589 | char *sizestr; | |
9a40f122 | 1590 | char *retptr; |
f46b5a66 CH |
1591 | char *devstr = NULL; |
1592 | int ret = 0; | |
f46b5a66 CH |
1593 | int mod = 0; |
1594 | ||
e441d54d CM |
1595 | if (!capable(CAP_SYS_ADMIN)) |
1596 | return -EPERM; | |
1597 | ||
198605a8 MX |
1598 | ret = mnt_want_write_file(file); |
1599 | if (ret) | |
1600 | return ret; | |
1601 | ||
c3e1f96c | 1602 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_RESIZE)) { |
97547676 | 1603 | mnt_drop_write_file(file); |
e57138b3 | 1604 | return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
c9e9f97b ID |
1605 | } |
1606 | ||
dae7b665 | 1607 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
1608 | if (IS_ERR(vol_args)) { |
1609 | ret = PTR_ERR(vol_args); | |
1610 | goto out; | |
1611 | } | |
5516e595 MF |
1612 | |
1613 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
f46b5a66 | 1614 | |
f46b5a66 CH |
1615 | sizestr = vol_args->name; |
1616 | devstr = strchr(sizestr, ':'); | |
1617 | if (devstr) { | |
f46b5a66 CH |
1618 | sizestr = devstr + 1; |
1619 | *devstr = '\0'; | |
1620 | devstr = vol_args->name; | |
58dfae63 Z |
1621 | ret = kstrtoull(devstr, 10, &devid); |
1622 | if (ret) | |
1623 | goto out_free; | |
dfd79829 MX |
1624 | if (!devid) { |
1625 | ret = -EINVAL; | |
1626 | goto out_free; | |
1627 | } | |
0b246afa | 1628 | btrfs_info(fs_info, "resizing devid %llu", devid); |
f46b5a66 | 1629 | } |
dba60f3f | 1630 | |
b2598edf | 1631 | device = btrfs_find_device(fs_info->fs_devices, devid, NULL, NULL); |
f46b5a66 | 1632 | if (!device) { |
0b246afa JM |
1633 | btrfs_info(fs_info, "resizer unable to find device %llu", |
1634 | devid); | |
dfd79829 | 1635 | ret = -ENODEV; |
c9e9f97b | 1636 | goto out_free; |
f46b5a66 | 1637 | } |
dba60f3f | 1638 | |
ebbede42 | 1639 | if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { |
0b246afa | 1640 | btrfs_info(fs_info, |
efe120a0 | 1641 | "resizer unable to apply on readonly device %llu", |
c1c9ff7c | 1642 | devid); |
dfd79829 | 1643 | ret = -EPERM; |
4e42ae1b LB |
1644 | goto out_free; |
1645 | } | |
1646 | ||
f46b5a66 CH |
1647 | if (!strcmp(sizestr, "max")) |
1648 | new_size = device->bdev->bd_inode->i_size; | |
1649 | else { | |
1650 | if (sizestr[0] == '-') { | |
1651 | mod = -1; | |
1652 | sizestr++; | |
1653 | } else if (sizestr[0] == '+') { | |
1654 | mod = 1; | |
1655 | sizestr++; | |
1656 | } | |
9a40f122 GH |
1657 | new_size = memparse(sizestr, &retptr); |
1658 | if (*retptr != '\0' || new_size == 0) { | |
f46b5a66 | 1659 | ret = -EINVAL; |
c9e9f97b | 1660 | goto out_free; |
f46b5a66 CH |
1661 | } |
1662 | } | |
1663 | ||
401e29c1 | 1664 | if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) { |
dfd79829 | 1665 | ret = -EPERM; |
63a212ab SB |
1666 | goto out_free; |
1667 | } | |
1668 | ||
7cc8e58d | 1669 | old_size = btrfs_device_get_total_bytes(device); |
f46b5a66 CH |
1670 | |
1671 | if (mod < 0) { | |
1672 | if (new_size > old_size) { | |
1673 | ret = -EINVAL; | |
c9e9f97b | 1674 | goto out_free; |
f46b5a66 CH |
1675 | } |
1676 | new_size = old_size - new_size; | |
1677 | } else if (mod > 0) { | |
eb8052e0 | 1678 | if (new_size > ULLONG_MAX - old_size) { |
902c68a4 | 1679 | ret = -ERANGE; |
eb8052e0 WF |
1680 | goto out_free; |
1681 | } | |
f46b5a66 CH |
1682 | new_size = old_size + new_size; |
1683 | } | |
1684 | ||
ee22184b | 1685 | if (new_size < SZ_256M) { |
f46b5a66 | 1686 | ret = -EINVAL; |
c9e9f97b | 1687 | goto out_free; |
f46b5a66 CH |
1688 | } |
1689 | if (new_size > device->bdev->bd_inode->i_size) { | |
1690 | ret = -EFBIG; | |
c9e9f97b | 1691 | goto out_free; |
f46b5a66 CH |
1692 | } |
1693 | ||
47f08b96 | 1694 | new_size = round_down(new_size, fs_info->sectorsize); |
f46b5a66 | 1695 | |
f46b5a66 | 1696 | if (new_size > old_size) { |
a22285a6 | 1697 | trans = btrfs_start_transaction(root, 0); |
98d5dc13 TI |
1698 | if (IS_ERR(trans)) { |
1699 | ret = PTR_ERR(trans); | |
c9e9f97b | 1700 | goto out_free; |
98d5dc13 | 1701 | } |
f46b5a66 | 1702 | ret = btrfs_grow_device(trans, device, new_size); |
3a45bb20 | 1703 | btrfs_commit_transaction(trans); |
ece7d20e | 1704 | } else if (new_size < old_size) { |
f46b5a66 | 1705 | ret = btrfs_shrink_device(device, new_size); |
0253f40e | 1706 | } /* equal, nothing need to do */ |
f46b5a66 | 1707 | |
faf8f7b9 MPS |
1708 | if (ret == 0 && new_size != old_size) |
1709 | btrfs_info_in_rcu(fs_info, | |
1710 | "resize device %s (devid %llu) from %llu to %llu", | |
1711 | rcu_str_deref(device->name), device->devid, | |
1712 | old_size, new_size); | |
c9e9f97b | 1713 | out_free: |
f46b5a66 | 1714 | kfree(vol_args); |
c9e9f97b | 1715 | out: |
c3e1f96c | 1716 | btrfs_exclop_finish(fs_info); |
18f39c41 | 1717 | mnt_drop_write_file(file); |
f46b5a66 CH |
1718 | return ret; |
1719 | } | |
1720 | ||
5d54c67e | 1721 | static noinline int __btrfs_ioctl_snap_create(struct file *file, |
52f75f4f | 1722 | const char *name, unsigned long fd, int subvol, |
5d54c67e | 1723 | bool readonly, |
8696c533 | 1724 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 1725 | { |
f46b5a66 | 1726 | int namelen; |
3de4586c | 1727 | int ret = 0; |
f46b5a66 | 1728 | |
325c50e3 JM |
1729 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1730 | return -ENOTDIR; | |
1731 | ||
a874a63e LB |
1732 | ret = mnt_want_write_file(file); |
1733 | if (ret) | |
1734 | goto out; | |
1735 | ||
72fd032e SW |
1736 | namelen = strlen(name); |
1737 | if (strchr(name, '/')) { | |
f46b5a66 | 1738 | ret = -EINVAL; |
a874a63e | 1739 | goto out_drop_write; |
f46b5a66 CH |
1740 | } |
1741 | ||
16780cab CM |
1742 | if (name[0] == '.' && |
1743 | (namelen == 1 || (name[1] == '.' && namelen == 2))) { | |
1744 | ret = -EEXIST; | |
a874a63e | 1745 | goto out_drop_write; |
16780cab CM |
1746 | } |
1747 | ||
3de4586c | 1748 | if (subvol) { |
72fd032e | 1749 | ret = btrfs_mksubvol(&file->f_path, name, namelen, |
5d54c67e | 1750 | NULL, readonly, inherit); |
cb8e7090 | 1751 | } else { |
2903ff01 | 1752 | struct fd src = fdget(fd); |
3de4586c | 1753 | struct inode *src_inode; |
2903ff01 | 1754 | if (!src.file) { |
3de4586c | 1755 | ret = -EINVAL; |
a874a63e | 1756 | goto out_drop_write; |
3de4586c CM |
1757 | } |
1758 | ||
496ad9aa AV |
1759 | src_inode = file_inode(src.file); |
1760 | if (src_inode->i_sb != file_inode(file)->i_sb) { | |
c79b4713 | 1761 | btrfs_info(BTRFS_I(file_inode(file))->root->fs_info, |
efe120a0 | 1762 | "Snapshot src from another FS"); |
23ad5b17 | 1763 | ret = -EXDEV; |
21cb47be | 1764 | } else if (!inode_owner_or_capable(&init_user_ns, src_inode)) { |
d0242061 DS |
1765 | /* |
1766 | * Subvolume creation is not restricted, but snapshots | |
1767 | * are limited to own subvolumes only | |
1768 | */ | |
1769 | ret = -EPERM; | |
ecd18815 | 1770 | } else { |
c11fbb6e | 1771 | ret = btrfs_mksnapshot(&file->f_path, name, namelen, |
ecd18815 | 1772 | BTRFS_I(src_inode)->root, |
5d54c67e | 1773 | readonly, inherit); |
3de4586c | 1774 | } |
2903ff01 | 1775 | fdput(src); |
cb8e7090 | 1776 | } |
a874a63e LB |
1777 | out_drop_write: |
1778 | mnt_drop_write_file(file); | |
f46b5a66 | 1779 | out: |
72fd032e SW |
1780 | return ret; |
1781 | } | |
1782 | ||
1783 | static noinline int btrfs_ioctl_snap_create(struct file *file, | |
fa0d2b9b | 1784 | void __user *arg, int subvol) |
72fd032e | 1785 | { |
fa0d2b9b | 1786 | struct btrfs_ioctl_vol_args *vol_args; |
72fd032e SW |
1787 | int ret; |
1788 | ||
325c50e3 JM |
1789 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1790 | return -ENOTDIR; | |
1791 | ||
fa0d2b9b LZ |
1792 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1793 | if (IS_ERR(vol_args)) | |
1794 | return PTR_ERR(vol_args); | |
1795 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
72fd032e | 1796 | |
5d54c67e NB |
1797 | ret = __btrfs_ioctl_snap_create(file, vol_args->name, vol_args->fd, |
1798 | subvol, false, NULL); | |
fdfb1e4f | 1799 | |
fa0d2b9b LZ |
1800 | kfree(vol_args); |
1801 | return ret; | |
1802 | } | |
fdfb1e4f | 1803 | |
fa0d2b9b LZ |
1804 | static noinline int btrfs_ioctl_snap_create_v2(struct file *file, |
1805 | void __user *arg, int subvol) | |
1806 | { | |
1807 | struct btrfs_ioctl_vol_args_v2 *vol_args; | |
1808 | int ret; | |
b83cc969 | 1809 | bool readonly = false; |
6f72c7e2 | 1810 | struct btrfs_qgroup_inherit *inherit = NULL; |
75eaa0e2 | 1811 | |
325c50e3 JM |
1812 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1813 | return -ENOTDIR; | |
1814 | ||
fa0d2b9b LZ |
1815 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1816 | if (IS_ERR(vol_args)) | |
1817 | return PTR_ERR(vol_args); | |
1818 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
75eaa0e2 | 1819 | |
673990db | 1820 | if (vol_args->flags & ~BTRFS_SUBVOL_CREATE_ARGS_MASK) { |
b83cc969 | 1821 | ret = -EOPNOTSUPP; |
c47ca32d | 1822 | goto free_args; |
72fd032e | 1823 | } |
fa0d2b9b | 1824 | |
b83cc969 LZ |
1825 | if (vol_args->flags & BTRFS_SUBVOL_RDONLY) |
1826 | readonly = true; | |
6f72c7e2 | 1827 | if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) { |
5011c5a6 DC |
1828 | u64 nums; |
1829 | ||
1830 | if (vol_args->size < sizeof(*inherit) || | |
1831 | vol_args->size > PAGE_SIZE) { | |
6f72c7e2 | 1832 | ret = -EINVAL; |
c47ca32d | 1833 | goto free_args; |
6f72c7e2 AJ |
1834 | } |
1835 | inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size); | |
1836 | if (IS_ERR(inherit)) { | |
1837 | ret = PTR_ERR(inherit); | |
c47ca32d | 1838 | goto free_args; |
6f72c7e2 | 1839 | } |
5011c5a6 DC |
1840 | |
1841 | if (inherit->num_qgroups > PAGE_SIZE || | |
1842 | inherit->num_ref_copies > PAGE_SIZE || | |
1843 | inherit->num_excl_copies > PAGE_SIZE) { | |
1844 | ret = -EINVAL; | |
1845 | goto free_inherit; | |
1846 | } | |
1847 | ||
1848 | nums = inherit->num_qgroups + 2 * inherit->num_ref_copies + | |
1849 | 2 * inherit->num_excl_copies; | |
1850 | if (vol_args->size != struct_size(inherit, qgroups, nums)) { | |
1851 | ret = -EINVAL; | |
1852 | goto free_inherit; | |
1853 | } | |
6f72c7e2 | 1854 | } |
fa0d2b9b | 1855 | |
5d54c67e NB |
1856 | ret = __btrfs_ioctl_snap_create(file, vol_args->name, vol_args->fd, |
1857 | subvol, readonly, inherit); | |
c47ca32d DC |
1858 | if (ret) |
1859 | goto free_inherit; | |
c47ca32d | 1860 | free_inherit: |
6f72c7e2 | 1861 | kfree(inherit); |
c47ca32d DC |
1862 | free_args: |
1863 | kfree(vol_args); | |
f46b5a66 CH |
1864 | return ret; |
1865 | } | |
1866 | ||
0caa102d LZ |
1867 | static noinline int btrfs_ioctl_subvol_getflags(struct file *file, |
1868 | void __user *arg) | |
1869 | { | |
496ad9aa | 1870 | struct inode *inode = file_inode(file); |
0b246afa | 1871 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
0caa102d LZ |
1872 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1873 | int ret = 0; | |
1874 | u64 flags = 0; | |
1875 | ||
4a0cc7ca | 1876 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) |
0caa102d LZ |
1877 | return -EINVAL; |
1878 | ||
0b246afa | 1879 | down_read(&fs_info->subvol_sem); |
0caa102d LZ |
1880 | if (btrfs_root_readonly(root)) |
1881 | flags |= BTRFS_SUBVOL_RDONLY; | |
0b246afa | 1882 | up_read(&fs_info->subvol_sem); |
0caa102d LZ |
1883 | |
1884 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
1885 | ret = -EFAULT; | |
1886 | ||
1887 | return ret; | |
1888 | } | |
1889 | ||
1890 | static noinline int btrfs_ioctl_subvol_setflags(struct file *file, | |
1891 | void __user *arg) | |
1892 | { | |
496ad9aa | 1893 | struct inode *inode = file_inode(file); |
0b246afa | 1894 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
0caa102d LZ |
1895 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1896 | struct btrfs_trans_handle *trans; | |
1897 | u64 root_flags; | |
1898 | u64 flags; | |
1899 | int ret = 0; | |
1900 | ||
21cb47be | 1901 | if (!inode_owner_or_capable(&init_user_ns, inode)) |
bd60ea0f DS |
1902 | return -EPERM; |
1903 | ||
b9ca0664 LB |
1904 | ret = mnt_want_write_file(file); |
1905 | if (ret) | |
1906 | goto out; | |
0caa102d | 1907 | |
4a0cc7ca | 1908 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
b9ca0664 LB |
1909 | ret = -EINVAL; |
1910 | goto out_drop_write; | |
1911 | } | |
0caa102d | 1912 | |
b9ca0664 LB |
1913 | if (copy_from_user(&flags, arg, sizeof(flags))) { |
1914 | ret = -EFAULT; | |
1915 | goto out_drop_write; | |
1916 | } | |
0caa102d | 1917 | |
b9ca0664 LB |
1918 | if (flags & ~BTRFS_SUBVOL_RDONLY) { |
1919 | ret = -EOPNOTSUPP; | |
1920 | goto out_drop_write; | |
1921 | } | |
0caa102d | 1922 | |
0b246afa | 1923 | down_write(&fs_info->subvol_sem); |
0caa102d LZ |
1924 | |
1925 | /* nothing to do */ | |
1926 | if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root)) | |
b9ca0664 | 1927 | goto out_drop_sem; |
0caa102d LZ |
1928 | |
1929 | root_flags = btrfs_root_flags(&root->root_item); | |
2c686537 | 1930 | if (flags & BTRFS_SUBVOL_RDONLY) { |
0caa102d LZ |
1931 | btrfs_set_root_flags(&root->root_item, |
1932 | root_flags | BTRFS_ROOT_SUBVOL_RDONLY); | |
2c686537 DS |
1933 | } else { |
1934 | /* | |
1935 | * Block RO -> RW transition if this subvolume is involved in | |
1936 | * send | |
1937 | */ | |
1938 | spin_lock(&root->root_item_lock); | |
1939 | if (root->send_in_progress == 0) { | |
1940 | btrfs_set_root_flags(&root->root_item, | |
0caa102d | 1941 | root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY); |
2c686537 DS |
1942 | spin_unlock(&root->root_item_lock); |
1943 | } else { | |
1944 | spin_unlock(&root->root_item_lock); | |
0b246afa JM |
1945 | btrfs_warn(fs_info, |
1946 | "Attempt to set subvolume %llu read-write during send", | |
1947 | root->root_key.objectid); | |
2c686537 DS |
1948 | ret = -EPERM; |
1949 | goto out_drop_sem; | |
1950 | } | |
1951 | } | |
0caa102d LZ |
1952 | |
1953 | trans = btrfs_start_transaction(root, 1); | |
1954 | if (IS_ERR(trans)) { | |
1955 | ret = PTR_ERR(trans); | |
1956 | goto out_reset; | |
1957 | } | |
1958 | ||
0b246afa | 1959 | ret = btrfs_update_root(trans, fs_info->tree_root, |
0caa102d | 1960 | &root->root_key, &root->root_item); |
9417ebc8 NB |
1961 | if (ret < 0) { |
1962 | btrfs_end_transaction(trans); | |
1963 | goto out_reset; | |
1964 | } | |
1965 | ||
1966 | ret = btrfs_commit_transaction(trans); | |
0caa102d | 1967 | |
0caa102d LZ |
1968 | out_reset: |
1969 | if (ret) | |
1970 | btrfs_set_root_flags(&root->root_item, root_flags); | |
b9ca0664 | 1971 | out_drop_sem: |
0b246afa | 1972 | up_write(&fs_info->subvol_sem); |
b9ca0664 LB |
1973 | out_drop_write: |
1974 | mnt_drop_write_file(file); | |
1975 | out: | |
0caa102d LZ |
1976 | return ret; |
1977 | } | |
1978 | ||
ac8e9819 CM |
1979 | static noinline int key_in_sk(struct btrfs_key *key, |
1980 | struct btrfs_ioctl_search_key *sk) | |
1981 | { | |
abc6e134 CM |
1982 | struct btrfs_key test; |
1983 | int ret; | |
1984 | ||
1985 | test.objectid = sk->min_objectid; | |
1986 | test.type = sk->min_type; | |
1987 | test.offset = sk->min_offset; | |
1988 | ||
1989 | ret = btrfs_comp_cpu_keys(key, &test); | |
1990 | if (ret < 0) | |
ac8e9819 | 1991 | return 0; |
abc6e134 CM |
1992 | |
1993 | test.objectid = sk->max_objectid; | |
1994 | test.type = sk->max_type; | |
1995 | test.offset = sk->max_offset; | |
1996 | ||
1997 | ret = btrfs_comp_cpu_keys(key, &test); | |
1998 | if (ret > 0) | |
ac8e9819 CM |
1999 | return 0; |
2000 | return 1; | |
2001 | } | |
2002 | ||
df397565 | 2003 | static noinline int copy_to_sk(struct btrfs_path *path, |
ac8e9819 CM |
2004 | struct btrfs_key *key, |
2005 | struct btrfs_ioctl_search_key *sk, | |
9b6e817d | 2006 | size_t *buf_size, |
ba346b35 | 2007 | char __user *ubuf, |
ac8e9819 CM |
2008 | unsigned long *sk_offset, |
2009 | int *num_found) | |
2010 | { | |
2011 | u64 found_transid; | |
2012 | struct extent_buffer *leaf; | |
2013 | struct btrfs_ioctl_search_header sh; | |
dd81d459 | 2014 | struct btrfs_key test; |
ac8e9819 CM |
2015 | unsigned long item_off; |
2016 | unsigned long item_len; | |
2017 | int nritems; | |
2018 | int i; | |
2019 | int slot; | |
ac8e9819 CM |
2020 | int ret = 0; |
2021 | ||
2022 | leaf = path->nodes[0]; | |
2023 | slot = path->slots[0]; | |
2024 | nritems = btrfs_header_nritems(leaf); | |
2025 | ||
2026 | if (btrfs_header_generation(leaf) > sk->max_transid) { | |
2027 | i = nritems; | |
2028 | goto advance_key; | |
2029 | } | |
2030 | found_transid = btrfs_header_generation(leaf); | |
2031 | ||
2032 | for (i = slot; i < nritems; i++) { | |
2033 | item_off = btrfs_item_ptr_offset(leaf, i); | |
2034 | item_len = btrfs_item_size_nr(leaf, i); | |
2035 | ||
03b71c6c GP |
2036 | btrfs_item_key_to_cpu(leaf, key, i); |
2037 | if (!key_in_sk(key, sk)) | |
2038 | continue; | |
2039 | ||
9b6e817d | 2040 | if (sizeof(sh) + item_len > *buf_size) { |
8f5f6178 GH |
2041 | if (*num_found) { |
2042 | ret = 1; | |
2043 | goto out; | |
2044 | } | |
2045 | ||
2046 | /* | |
2047 | * return one empty item back for v1, which does not | |
2048 | * handle -EOVERFLOW | |
2049 | */ | |
2050 | ||
9b6e817d | 2051 | *buf_size = sizeof(sh) + item_len; |
ac8e9819 | 2052 | item_len = 0; |
8f5f6178 GH |
2053 | ret = -EOVERFLOW; |
2054 | } | |
ac8e9819 | 2055 | |
9b6e817d | 2056 | if (sizeof(sh) + item_len + *sk_offset > *buf_size) { |
ac8e9819 | 2057 | ret = 1; |
25c9bc2e | 2058 | goto out; |
ac8e9819 CM |
2059 | } |
2060 | ||
ac8e9819 CM |
2061 | sh.objectid = key->objectid; |
2062 | sh.offset = key->offset; | |
2063 | sh.type = key->type; | |
2064 | sh.len = item_len; | |
2065 | sh.transid = found_transid; | |
2066 | ||
a48b73ec JB |
2067 | /* |
2068 | * Copy search result header. If we fault then loop again so we | |
2069 | * can fault in the pages and -EFAULT there if there's a | |
2070 | * problem. Otherwise we'll fault and then copy the buffer in | |
2071 | * properly this next time through | |
2072 | */ | |
2073 | if (copy_to_user_nofault(ubuf + *sk_offset, &sh, sizeof(sh))) { | |
2074 | ret = 0; | |
ba346b35 GH |
2075 | goto out; |
2076 | } | |
2077 | ||
ac8e9819 CM |
2078 | *sk_offset += sizeof(sh); |
2079 | ||
2080 | if (item_len) { | |
ba346b35 | 2081 | char __user *up = ubuf + *sk_offset; |
a48b73ec JB |
2082 | /* |
2083 | * Copy the item, same behavior as above, but reset the | |
2084 | * * sk_offset so we copy the full thing again. | |
2085 | */ | |
2086 | if (read_extent_buffer_to_user_nofault(leaf, up, | |
2087 | item_off, item_len)) { | |
2088 | ret = 0; | |
2089 | *sk_offset -= sizeof(sh); | |
ba346b35 GH |
2090 | goto out; |
2091 | } | |
2092 | ||
ac8e9819 | 2093 | *sk_offset += item_len; |
ac8e9819 | 2094 | } |
e2156867 | 2095 | (*num_found)++; |
ac8e9819 | 2096 | |
8f5f6178 GH |
2097 | if (ret) /* -EOVERFLOW from above */ |
2098 | goto out; | |
2099 | ||
25c9bc2e GH |
2100 | if (*num_found >= sk->nr_items) { |
2101 | ret = 1; | |
2102 | goto out; | |
2103 | } | |
ac8e9819 CM |
2104 | } |
2105 | advance_key: | |
abc6e134 | 2106 | ret = 0; |
dd81d459 NA |
2107 | test.objectid = sk->max_objectid; |
2108 | test.type = sk->max_type; | |
2109 | test.offset = sk->max_offset; | |
2110 | if (btrfs_comp_cpu_keys(key, &test) >= 0) | |
2111 | ret = 1; | |
2112 | else if (key->offset < (u64)-1) | |
ac8e9819 | 2113 | key->offset++; |
dd81d459 | 2114 | else if (key->type < (u8)-1) { |
abc6e134 | 2115 | key->offset = 0; |
ac8e9819 | 2116 | key->type++; |
dd81d459 | 2117 | } else if (key->objectid < (u64)-1) { |
abc6e134 CM |
2118 | key->offset = 0; |
2119 | key->type = 0; | |
ac8e9819 | 2120 | key->objectid++; |
abc6e134 CM |
2121 | } else |
2122 | ret = 1; | |
25c9bc2e | 2123 | out: |
ba346b35 GH |
2124 | /* |
2125 | * 0: all items from this leaf copied, continue with next | |
2126 | * 1: * more items can be copied, but unused buffer is too small | |
2127 | * * all items were found | |
2128 | * Either way, it will stops the loop which iterates to the next | |
2129 | * leaf | |
2130 | * -EOVERFLOW: item was to large for buffer | |
2131 | * -EFAULT: could not copy extent buffer back to userspace | |
2132 | */ | |
ac8e9819 CM |
2133 | return ret; |
2134 | } | |
2135 | ||
2136 | static noinline int search_ioctl(struct inode *inode, | |
12544442 | 2137 | struct btrfs_ioctl_search_key *sk, |
9b6e817d | 2138 | size_t *buf_size, |
ba346b35 | 2139 | char __user *ubuf) |
ac8e9819 | 2140 | { |
0b246afa | 2141 | struct btrfs_fs_info *info = btrfs_sb(inode->i_sb); |
ac8e9819 CM |
2142 | struct btrfs_root *root; |
2143 | struct btrfs_key key; | |
ac8e9819 | 2144 | struct btrfs_path *path; |
ac8e9819 CM |
2145 | int ret; |
2146 | int num_found = 0; | |
2147 | unsigned long sk_offset = 0; | |
2148 | ||
9b6e817d GH |
2149 | if (*buf_size < sizeof(struct btrfs_ioctl_search_header)) { |
2150 | *buf_size = sizeof(struct btrfs_ioctl_search_header); | |
12544442 | 2151 | return -EOVERFLOW; |
9b6e817d | 2152 | } |
12544442 | 2153 | |
ac8e9819 CM |
2154 | path = btrfs_alloc_path(); |
2155 | if (!path) | |
2156 | return -ENOMEM; | |
2157 | ||
2158 | if (sk->tree_id == 0) { | |
2159 | /* search the root of the inode that was passed */ | |
00246528 | 2160 | root = btrfs_grab_root(BTRFS_I(inode)->root); |
ac8e9819 | 2161 | } else { |
56e9357a | 2162 | root = btrfs_get_fs_root(info, sk->tree_id, true); |
ac8e9819 | 2163 | if (IS_ERR(root)) { |
ac8e9819 | 2164 | btrfs_free_path(path); |
ad1e3d56 | 2165 | return PTR_ERR(root); |
ac8e9819 CM |
2166 | } |
2167 | } | |
2168 | ||
2169 | key.objectid = sk->min_objectid; | |
2170 | key.type = sk->min_type; | |
2171 | key.offset = sk->min_offset; | |
2172 | ||
67871254 | 2173 | while (1) { |
1c78544e FM |
2174 | ret = fault_in_pages_writeable(ubuf + sk_offset, |
2175 | *buf_size - sk_offset); | |
a48b73ec JB |
2176 | if (ret) |
2177 | break; | |
2178 | ||
6174d3cb | 2179 | ret = btrfs_search_forward(root, &key, path, sk->min_transid); |
ac8e9819 CM |
2180 | if (ret != 0) { |
2181 | if (ret > 0) | |
2182 | ret = 0; | |
2183 | goto err; | |
2184 | } | |
df397565 | 2185 | ret = copy_to_sk(path, &key, sk, buf_size, ubuf, |
ac8e9819 | 2186 | &sk_offset, &num_found); |
b3b4aa74 | 2187 | btrfs_release_path(path); |
25c9bc2e | 2188 | if (ret) |
ac8e9819 CM |
2189 | break; |
2190 | ||
2191 | } | |
8f5f6178 GH |
2192 | if (ret > 0) |
2193 | ret = 0; | |
ac8e9819 CM |
2194 | err: |
2195 | sk->nr_items = num_found; | |
00246528 | 2196 | btrfs_put_root(root); |
ac8e9819 CM |
2197 | btrfs_free_path(path); |
2198 | return ret; | |
2199 | } | |
2200 | ||
2201 | static noinline int btrfs_ioctl_tree_search(struct file *file, | |
2202 | void __user *argp) | |
2203 | { | |
ba346b35 GH |
2204 | struct btrfs_ioctl_search_args __user *uargs; |
2205 | struct btrfs_ioctl_search_key sk; | |
9b6e817d GH |
2206 | struct inode *inode; |
2207 | int ret; | |
2208 | size_t buf_size; | |
ac8e9819 CM |
2209 | |
2210 | if (!capable(CAP_SYS_ADMIN)) | |
2211 | return -EPERM; | |
2212 | ||
ba346b35 GH |
2213 | uargs = (struct btrfs_ioctl_search_args __user *)argp; |
2214 | ||
2215 | if (copy_from_user(&sk, &uargs->key, sizeof(sk))) | |
2216 | return -EFAULT; | |
ac8e9819 | 2217 | |
ba346b35 | 2218 | buf_size = sizeof(uargs->buf); |
ac8e9819 | 2219 | |
496ad9aa | 2220 | inode = file_inode(file); |
ba346b35 | 2221 | ret = search_ioctl(inode, &sk, &buf_size, uargs->buf); |
8f5f6178 GH |
2222 | |
2223 | /* | |
2224 | * In the origin implementation an overflow is handled by returning a | |
2225 | * search header with a len of zero, so reset ret. | |
2226 | */ | |
2227 | if (ret == -EOVERFLOW) | |
2228 | ret = 0; | |
2229 | ||
ba346b35 | 2230 | if (ret == 0 && copy_to_user(&uargs->key, &sk, sizeof(sk))) |
ac8e9819 | 2231 | ret = -EFAULT; |
ac8e9819 CM |
2232 | return ret; |
2233 | } | |
2234 | ||
cc68a8a5 GH |
2235 | static noinline int btrfs_ioctl_tree_search_v2(struct file *file, |
2236 | void __user *argp) | |
2237 | { | |
2238 | struct btrfs_ioctl_search_args_v2 __user *uarg; | |
2239 | struct btrfs_ioctl_search_args_v2 args; | |
2240 | struct inode *inode; | |
2241 | int ret; | |
2242 | size_t buf_size; | |
ee22184b | 2243 | const size_t buf_limit = SZ_16M; |
cc68a8a5 GH |
2244 | |
2245 | if (!capable(CAP_SYS_ADMIN)) | |
2246 | return -EPERM; | |
2247 | ||
2248 | /* copy search header and buffer size */ | |
2249 | uarg = (struct btrfs_ioctl_search_args_v2 __user *)argp; | |
2250 | if (copy_from_user(&args, uarg, sizeof(args))) | |
2251 | return -EFAULT; | |
2252 | ||
2253 | buf_size = args.buf_size; | |
2254 | ||
cc68a8a5 GH |
2255 | /* limit result size to 16MB */ |
2256 | if (buf_size > buf_limit) | |
2257 | buf_size = buf_limit; | |
2258 | ||
2259 | inode = file_inode(file); | |
2260 | ret = search_ioctl(inode, &args.key, &buf_size, | |
718dc5fa | 2261 | (char __user *)(&uarg->buf[0])); |
cc68a8a5 GH |
2262 | if (ret == 0 && copy_to_user(&uarg->key, &args.key, sizeof(args.key))) |
2263 | ret = -EFAULT; | |
2264 | else if (ret == -EOVERFLOW && | |
2265 | copy_to_user(&uarg->buf_size, &buf_size, sizeof(buf_size))) | |
2266 | ret = -EFAULT; | |
2267 | ||
ac8e9819 CM |
2268 | return ret; |
2269 | } | |
2270 | ||
98d377a0 | 2271 | /* |
ac8e9819 CM |
2272 | * Search INODE_REFs to identify path name of 'dirid' directory |
2273 | * in a 'tree_id' tree. and sets path name to 'name'. | |
2274 | */ | |
98d377a0 TH |
2275 | static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, |
2276 | u64 tree_id, u64 dirid, char *name) | |
2277 | { | |
2278 | struct btrfs_root *root; | |
2279 | struct btrfs_key key; | |
ac8e9819 | 2280 | char *ptr; |
98d377a0 TH |
2281 | int ret = -1; |
2282 | int slot; | |
2283 | int len; | |
2284 | int total_len = 0; | |
2285 | struct btrfs_inode_ref *iref; | |
2286 | struct extent_buffer *l; | |
2287 | struct btrfs_path *path; | |
2288 | ||
2289 | if (dirid == BTRFS_FIRST_FREE_OBJECTID) { | |
2290 | name[0]='\0'; | |
2291 | return 0; | |
2292 | } | |
2293 | ||
2294 | path = btrfs_alloc_path(); | |
2295 | if (!path) | |
2296 | return -ENOMEM; | |
2297 | ||
c8bcbfbd | 2298 | ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX - 1]; |
98d377a0 | 2299 | |
56e9357a | 2300 | root = btrfs_get_fs_root(info, tree_id, true); |
98d377a0 | 2301 | if (IS_ERR(root)) { |
ad1e3d56 | 2302 | ret = PTR_ERR(root); |
88234012 JB |
2303 | root = NULL; |
2304 | goto out; | |
2305 | } | |
98d377a0 TH |
2306 | |
2307 | key.objectid = dirid; | |
2308 | key.type = BTRFS_INODE_REF_KEY; | |
8ad6fcab | 2309 | key.offset = (u64)-1; |
98d377a0 | 2310 | |
67871254 | 2311 | while (1) { |
98d377a0 TH |
2312 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); |
2313 | if (ret < 0) | |
2314 | goto out; | |
18674c6c FDBM |
2315 | else if (ret > 0) { |
2316 | ret = btrfs_previous_item(root, path, dirid, | |
2317 | BTRFS_INODE_REF_KEY); | |
2318 | if (ret < 0) | |
2319 | goto out; | |
2320 | else if (ret > 0) { | |
2321 | ret = -ENOENT; | |
2322 | goto out; | |
2323 | } | |
2324 | } | |
98d377a0 TH |
2325 | |
2326 | l = path->nodes[0]; | |
2327 | slot = path->slots[0]; | |
2328 | btrfs_item_key_to_cpu(l, &key, slot); | |
2329 | ||
98d377a0 TH |
2330 | iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); |
2331 | len = btrfs_inode_ref_name_len(l, iref); | |
2332 | ptr -= len + 1; | |
2333 | total_len += len + 1; | |
a696cf35 FDBM |
2334 | if (ptr < name) { |
2335 | ret = -ENAMETOOLONG; | |
98d377a0 | 2336 | goto out; |
a696cf35 | 2337 | } |
98d377a0 TH |
2338 | |
2339 | *(ptr + len) = '/'; | |
67871254 | 2340 | read_extent_buffer(l, ptr, (unsigned long)(iref + 1), len); |
98d377a0 TH |
2341 | |
2342 | if (key.offset == BTRFS_FIRST_FREE_OBJECTID) | |
2343 | break; | |
2344 | ||
b3b4aa74 | 2345 | btrfs_release_path(path); |
98d377a0 | 2346 | key.objectid = key.offset; |
8ad6fcab | 2347 | key.offset = (u64)-1; |
98d377a0 | 2348 | dirid = key.objectid; |
98d377a0 | 2349 | } |
77906a50 | 2350 | memmove(name, ptr, total_len); |
67871254 | 2351 | name[total_len] = '\0'; |
98d377a0 TH |
2352 | ret = 0; |
2353 | out: | |
00246528 | 2354 | btrfs_put_root(root); |
98d377a0 | 2355 | btrfs_free_path(path); |
ac8e9819 CM |
2356 | return ret; |
2357 | } | |
2358 | ||
23d0b79d TM |
2359 | static int btrfs_search_path_in_tree_user(struct inode *inode, |
2360 | struct btrfs_ioctl_ino_lookup_user_args *args) | |
2361 | { | |
2362 | struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info; | |
2363 | struct super_block *sb = inode->i_sb; | |
2364 | struct btrfs_key upper_limit = BTRFS_I(inode)->location; | |
2365 | u64 treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2366 | u64 dirid = args->dirid; | |
2367 | unsigned long item_off; | |
2368 | unsigned long item_len; | |
2369 | struct btrfs_inode_ref *iref; | |
2370 | struct btrfs_root_ref *rref; | |
b8a49ae1 | 2371 | struct btrfs_root *root = NULL; |
23d0b79d TM |
2372 | struct btrfs_path *path; |
2373 | struct btrfs_key key, key2; | |
2374 | struct extent_buffer *leaf; | |
2375 | struct inode *temp_inode; | |
2376 | char *ptr; | |
2377 | int slot; | |
2378 | int len; | |
2379 | int total_len = 0; | |
2380 | int ret; | |
2381 | ||
2382 | path = btrfs_alloc_path(); | |
2383 | if (!path) | |
2384 | return -ENOMEM; | |
2385 | ||
2386 | /* | |
2387 | * If the bottom subvolume does not exist directly under upper_limit, | |
2388 | * construct the path in from the bottom up. | |
2389 | */ | |
2390 | if (dirid != upper_limit.objectid) { | |
2391 | ptr = &args->path[BTRFS_INO_LOOKUP_USER_PATH_MAX - 1]; | |
2392 | ||
56e9357a | 2393 | root = btrfs_get_fs_root(fs_info, treeid, true); |
23d0b79d TM |
2394 | if (IS_ERR(root)) { |
2395 | ret = PTR_ERR(root); | |
2396 | goto out; | |
2397 | } | |
2398 | ||
2399 | key.objectid = dirid; | |
2400 | key.type = BTRFS_INODE_REF_KEY; | |
2401 | key.offset = (u64)-1; | |
2402 | while (1) { | |
2403 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2404 | if (ret < 0) { | |
b8a49ae1 | 2405 | goto out_put; |
23d0b79d TM |
2406 | } else if (ret > 0) { |
2407 | ret = btrfs_previous_item(root, path, dirid, | |
2408 | BTRFS_INODE_REF_KEY); | |
2409 | if (ret < 0) { | |
b8a49ae1 | 2410 | goto out_put; |
23d0b79d TM |
2411 | } else if (ret > 0) { |
2412 | ret = -ENOENT; | |
b8a49ae1 | 2413 | goto out_put; |
23d0b79d TM |
2414 | } |
2415 | } | |
2416 | ||
2417 | leaf = path->nodes[0]; | |
2418 | slot = path->slots[0]; | |
2419 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2420 | ||
2421 | iref = btrfs_item_ptr(leaf, slot, struct btrfs_inode_ref); | |
2422 | len = btrfs_inode_ref_name_len(leaf, iref); | |
2423 | ptr -= len + 1; | |
2424 | total_len += len + 1; | |
2425 | if (ptr < args->path) { | |
2426 | ret = -ENAMETOOLONG; | |
b8a49ae1 | 2427 | goto out_put; |
23d0b79d TM |
2428 | } |
2429 | ||
2430 | *(ptr + len) = '/'; | |
2431 | read_extent_buffer(leaf, ptr, | |
2432 | (unsigned long)(iref + 1), len); | |
2433 | ||
2434 | /* Check the read+exec permission of this directory */ | |
2435 | ret = btrfs_previous_item(root, path, dirid, | |
2436 | BTRFS_INODE_ITEM_KEY); | |
2437 | if (ret < 0) { | |
b8a49ae1 | 2438 | goto out_put; |
23d0b79d TM |
2439 | } else if (ret > 0) { |
2440 | ret = -ENOENT; | |
b8a49ae1 | 2441 | goto out_put; |
23d0b79d TM |
2442 | } |
2443 | ||
2444 | leaf = path->nodes[0]; | |
2445 | slot = path->slots[0]; | |
2446 | btrfs_item_key_to_cpu(leaf, &key2, slot); | |
2447 | if (key2.objectid != dirid) { | |
2448 | ret = -ENOENT; | |
b8a49ae1 | 2449 | goto out_put; |
23d0b79d TM |
2450 | } |
2451 | ||
0202e83f | 2452 | temp_inode = btrfs_iget(sb, key2.objectid, root); |
3ca57bd6 MT |
2453 | if (IS_ERR(temp_inode)) { |
2454 | ret = PTR_ERR(temp_inode); | |
b8a49ae1 | 2455 | goto out_put; |
3ca57bd6 | 2456 | } |
47291baa CB |
2457 | ret = inode_permission(&init_user_ns, temp_inode, |
2458 | MAY_READ | MAY_EXEC); | |
23d0b79d TM |
2459 | iput(temp_inode); |
2460 | if (ret) { | |
2461 | ret = -EACCES; | |
b8a49ae1 | 2462 | goto out_put; |
23d0b79d TM |
2463 | } |
2464 | ||
2465 | if (key.offset == upper_limit.objectid) | |
2466 | break; | |
2467 | if (key.objectid == BTRFS_FIRST_FREE_OBJECTID) { | |
2468 | ret = -EACCES; | |
b8a49ae1 | 2469 | goto out_put; |
23d0b79d TM |
2470 | } |
2471 | ||
2472 | btrfs_release_path(path); | |
2473 | key.objectid = key.offset; | |
2474 | key.offset = (u64)-1; | |
2475 | dirid = key.objectid; | |
2476 | } | |
2477 | ||
2478 | memmove(args->path, ptr, total_len); | |
2479 | args->path[total_len] = '\0'; | |
00246528 | 2480 | btrfs_put_root(root); |
b8a49ae1 | 2481 | root = NULL; |
23d0b79d TM |
2482 | btrfs_release_path(path); |
2483 | } | |
2484 | ||
2485 | /* Get the bottom subvolume's name from ROOT_REF */ | |
23d0b79d TM |
2486 | key.objectid = treeid; |
2487 | key.type = BTRFS_ROOT_REF_KEY; | |
2488 | key.offset = args->treeid; | |
b8a49ae1 | 2489 | ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); |
23d0b79d TM |
2490 | if (ret < 0) { |
2491 | goto out; | |
2492 | } else if (ret > 0) { | |
2493 | ret = -ENOENT; | |
2494 | goto out; | |
2495 | } | |
2496 | ||
2497 | leaf = path->nodes[0]; | |
2498 | slot = path->slots[0]; | |
2499 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2500 | ||
2501 | item_off = btrfs_item_ptr_offset(leaf, slot); | |
2502 | item_len = btrfs_item_size_nr(leaf, slot); | |
2503 | /* Check if dirid in ROOT_REF corresponds to passed dirid */ | |
2504 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2505 | if (args->dirid != btrfs_root_ref_dirid(leaf, rref)) { | |
2506 | ret = -EINVAL; | |
2507 | goto out; | |
2508 | } | |
2509 | ||
2510 | /* Copy subvolume's name */ | |
2511 | item_off += sizeof(struct btrfs_root_ref); | |
2512 | item_len -= sizeof(struct btrfs_root_ref); | |
2513 | read_extent_buffer(leaf, args->name, item_off, item_len); | |
2514 | args->name[item_len] = 0; | |
2515 | ||
b8a49ae1 | 2516 | out_put: |
00246528 | 2517 | btrfs_put_root(root); |
23d0b79d TM |
2518 | out: |
2519 | btrfs_free_path(path); | |
2520 | return ret; | |
2521 | } | |
2522 | ||
ac8e9819 CM |
2523 | static noinline int btrfs_ioctl_ino_lookup(struct file *file, |
2524 | void __user *argp) | |
2525 | { | |
bece2e82 BVA |
2526 | struct btrfs_ioctl_ino_lookup_args *args; |
2527 | struct inode *inode; | |
01b810b8 | 2528 | int ret = 0; |
ac8e9819 | 2529 | |
2354d08f JL |
2530 | args = memdup_user(argp, sizeof(*args)); |
2531 | if (IS_ERR(args)) | |
2532 | return PTR_ERR(args); | |
c2b96929 | 2533 | |
496ad9aa | 2534 | inode = file_inode(file); |
ac8e9819 | 2535 | |
01b810b8 DS |
2536 | /* |
2537 | * Unprivileged query to obtain the containing subvolume root id. The | |
2538 | * path is reset so it's consistent with btrfs_search_path_in_tree. | |
2539 | */ | |
1b53ac4d CM |
2540 | if (args->treeid == 0) |
2541 | args->treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2542 | ||
01b810b8 DS |
2543 | if (args->objectid == BTRFS_FIRST_FREE_OBJECTID) { |
2544 | args->name[0] = 0; | |
2545 | goto out; | |
2546 | } | |
2547 | ||
2548 | if (!capable(CAP_SYS_ADMIN)) { | |
2549 | ret = -EPERM; | |
2550 | goto out; | |
2551 | } | |
2552 | ||
ac8e9819 CM |
2553 | ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, |
2554 | args->treeid, args->objectid, | |
2555 | args->name); | |
2556 | ||
01b810b8 | 2557 | out: |
ac8e9819 CM |
2558 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) |
2559 | ret = -EFAULT; | |
2560 | ||
2561 | kfree(args); | |
98d377a0 TH |
2562 | return ret; |
2563 | } | |
2564 | ||
23d0b79d TM |
2565 | /* |
2566 | * Version of ino_lookup ioctl (unprivileged) | |
2567 | * | |
2568 | * The main differences from ino_lookup ioctl are: | |
2569 | * | |
2570 | * 1. Read + Exec permission will be checked using inode_permission() during | |
2571 | * path construction. -EACCES will be returned in case of failure. | |
2572 | * 2. Path construction will be stopped at the inode number which corresponds | |
2573 | * to the fd with which this ioctl is called. If constructed path does not | |
2574 | * exist under fd's inode, -EACCES will be returned. | |
2575 | * 3. The name of bottom subvolume is also searched and filled. | |
2576 | */ | |
2577 | static int btrfs_ioctl_ino_lookup_user(struct file *file, void __user *argp) | |
2578 | { | |
2579 | struct btrfs_ioctl_ino_lookup_user_args *args; | |
2580 | struct inode *inode; | |
2581 | int ret; | |
2582 | ||
2583 | args = memdup_user(argp, sizeof(*args)); | |
2584 | if (IS_ERR(args)) | |
2585 | return PTR_ERR(args); | |
2586 | ||
2587 | inode = file_inode(file); | |
2588 | ||
2589 | if (args->dirid == BTRFS_FIRST_FREE_OBJECTID && | |
2590 | BTRFS_I(inode)->location.objectid != BTRFS_FIRST_FREE_OBJECTID) { | |
2591 | /* | |
2592 | * The subvolume does not exist under fd with which this is | |
2593 | * called | |
2594 | */ | |
2595 | kfree(args); | |
2596 | return -EACCES; | |
2597 | } | |
2598 | ||
2599 | ret = btrfs_search_path_in_tree_user(inode, args); | |
2600 | ||
2601 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
2602 | ret = -EFAULT; | |
2603 | ||
2604 | kfree(args); | |
2605 | return ret; | |
2606 | } | |
2607 | ||
b64ec075 TM |
2608 | /* Get the subvolume information in BTRFS_ROOT_ITEM and BTRFS_ROOT_BACKREF */ |
2609 | static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp) | |
2610 | { | |
2611 | struct btrfs_ioctl_get_subvol_info_args *subvol_info; | |
2612 | struct btrfs_fs_info *fs_info; | |
2613 | struct btrfs_root *root; | |
2614 | struct btrfs_path *path; | |
2615 | struct btrfs_key key; | |
2616 | struct btrfs_root_item *root_item; | |
2617 | struct btrfs_root_ref *rref; | |
2618 | struct extent_buffer *leaf; | |
2619 | unsigned long item_off; | |
2620 | unsigned long item_len; | |
2621 | struct inode *inode; | |
2622 | int slot; | |
2623 | int ret = 0; | |
2624 | ||
2625 | path = btrfs_alloc_path(); | |
2626 | if (!path) | |
2627 | return -ENOMEM; | |
2628 | ||
2629 | subvol_info = kzalloc(sizeof(*subvol_info), GFP_KERNEL); | |
2630 | if (!subvol_info) { | |
2631 | btrfs_free_path(path); | |
2632 | return -ENOMEM; | |
2633 | } | |
2634 | ||
2635 | inode = file_inode(file); | |
2636 | fs_info = BTRFS_I(inode)->root->fs_info; | |
2637 | ||
2638 | /* Get root_item of inode's subvolume */ | |
2639 | key.objectid = BTRFS_I(inode)->root->root_key.objectid; | |
56e9357a | 2640 | root = btrfs_get_fs_root(fs_info, key.objectid, true); |
b64ec075 TM |
2641 | if (IS_ERR(root)) { |
2642 | ret = PTR_ERR(root); | |
04734e84 JB |
2643 | goto out_free; |
2644 | } | |
b64ec075 TM |
2645 | root_item = &root->root_item; |
2646 | ||
2647 | subvol_info->treeid = key.objectid; | |
2648 | ||
2649 | subvol_info->generation = btrfs_root_generation(root_item); | |
2650 | subvol_info->flags = btrfs_root_flags(root_item); | |
2651 | ||
2652 | memcpy(subvol_info->uuid, root_item->uuid, BTRFS_UUID_SIZE); | |
2653 | memcpy(subvol_info->parent_uuid, root_item->parent_uuid, | |
2654 | BTRFS_UUID_SIZE); | |
2655 | memcpy(subvol_info->received_uuid, root_item->received_uuid, | |
2656 | BTRFS_UUID_SIZE); | |
2657 | ||
2658 | subvol_info->ctransid = btrfs_root_ctransid(root_item); | |
2659 | subvol_info->ctime.sec = btrfs_stack_timespec_sec(&root_item->ctime); | |
2660 | subvol_info->ctime.nsec = btrfs_stack_timespec_nsec(&root_item->ctime); | |
2661 | ||
2662 | subvol_info->otransid = btrfs_root_otransid(root_item); | |
2663 | subvol_info->otime.sec = btrfs_stack_timespec_sec(&root_item->otime); | |
2664 | subvol_info->otime.nsec = btrfs_stack_timespec_nsec(&root_item->otime); | |
2665 | ||
2666 | subvol_info->stransid = btrfs_root_stransid(root_item); | |
2667 | subvol_info->stime.sec = btrfs_stack_timespec_sec(&root_item->stime); | |
2668 | subvol_info->stime.nsec = btrfs_stack_timespec_nsec(&root_item->stime); | |
2669 | ||
2670 | subvol_info->rtransid = btrfs_root_rtransid(root_item); | |
2671 | subvol_info->rtime.sec = btrfs_stack_timespec_sec(&root_item->rtime); | |
2672 | subvol_info->rtime.nsec = btrfs_stack_timespec_nsec(&root_item->rtime); | |
2673 | ||
2674 | if (key.objectid != BTRFS_FS_TREE_OBJECTID) { | |
2675 | /* Search root tree for ROOT_BACKREF of this subvolume */ | |
b64ec075 TM |
2676 | key.type = BTRFS_ROOT_BACKREF_KEY; |
2677 | key.offset = 0; | |
04734e84 | 2678 | ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); |
b64ec075 TM |
2679 | if (ret < 0) { |
2680 | goto out; | |
2681 | } else if (path->slots[0] >= | |
2682 | btrfs_header_nritems(path->nodes[0])) { | |
04734e84 | 2683 | ret = btrfs_next_leaf(fs_info->tree_root, path); |
b64ec075 TM |
2684 | if (ret < 0) { |
2685 | goto out; | |
2686 | } else if (ret > 0) { | |
2687 | ret = -EUCLEAN; | |
2688 | goto out; | |
2689 | } | |
2690 | } | |
2691 | ||
2692 | leaf = path->nodes[0]; | |
2693 | slot = path->slots[0]; | |
2694 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2695 | if (key.objectid == subvol_info->treeid && | |
2696 | key.type == BTRFS_ROOT_BACKREF_KEY) { | |
2697 | subvol_info->parent_id = key.offset; | |
2698 | ||
2699 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2700 | subvol_info->dirid = btrfs_root_ref_dirid(leaf, rref); | |
2701 | ||
2702 | item_off = btrfs_item_ptr_offset(leaf, slot) | |
2703 | + sizeof(struct btrfs_root_ref); | |
2704 | item_len = btrfs_item_size_nr(leaf, slot) | |
2705 | - sizeof(struct btrfs_root_ref); | |
2706 | read_extent_buffer(leaf, subvol_info->name, | |
2707 | item_off, item_len); | |
2708 | } else { | |
2709 | ret = -ENOENT; | |
2710 | goto out; | |
2711 | } | |
2712 | } | |
2713 | ||
2714 | if (copy_to_user(argp, subvol_info, sizeof(*subvol_info))) | |
2715 | ret = -EFAULT; | |
2716 | ||
2717 | out: | |
00246528 | 2718 | btrfs_put_root(root); |
04734e84 | 2719 | out_free: |
b64ec075 | 2720 | btrfs_free_path(path); |
b091f7fe | 2721 | kfree(subvol_info); |
b64ec075 TM |
2722 | return ret; |
2723 | } | |
2724 | ||
42e4b520 TM |
2725 | /* |
2726 | * Return ROOT_REF information of the subvolume containing this inode | |
2727 | * except the subvolume name. | |
2728 | */ | |
2729 | static int btrfs_ioctl_get_subvol_rootref(struct file *file, void __user *argp) | |
2730 | { | |
2731 | struct btrfs_ioctl_get_subvol_rootref_args *rootrefs; | |
2732 | struct btrfs_root_ref *rref; | |
2733 | struct btrfs_root *root; | |
2734 | struct btrfs_path *path; | |
2735 | struct btrfs_key key; | |
2736 | struct extent_buffer *leaf; | |
2737 | struct inode *inode; | |
2738 | u64 objectid; | |
2739 | int slot; | |
2740 | int ret; | |
2741 | u8 found; | |
2742 | ||
2743 | path = btrfs_alloc_path(); | |
2744 | if (!path) | |
2745 | return -ENOMEM; | |
2746 | ||
2747 | rootrefs = memdup_user(argp, sizeof(*rootrefs)); | |
2748 | if (IS_ERR(rootrefs)) { | |
2749 | btrfs_free_path(path); | |
2750 | return PTR_ERR(rootrefs); | |
2751 | } | |
2752 | ||
2753 | inode = file_inode(file); | |
2754 | root = BTRFS_I(inode)->root->fs_info->tree_root; | |
2755 | objectid = BTRFS_I(inode)->root->root_key.objectid; | |
2756 | ||
2757 | key.objectid = objectid; | |
2758 | key.type = BTRFS_ROOT_REF_KEY; | |
2759 | key.offset = rootrefs->min_treeid; | |
2760 | found = 0; | |
2761 | ||
2762 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2763 | if (ret < 0) { | |
2764 | goto out; | |
2765 | } else if (path->slots[0] >= | |
2766 | btrfs_header_nritems(path->nodes[0])) { | |
2767 | ret = btrfs_next_leaf(root, path); | |
2768 | if (ret < 0) { | |
2769 | goto out; | |
2770 | } else if (ret > 0) { | |
2771 | ret = -EUCLEAN; | |
2772 | goto out; | |
2773 | } | |
2774 | } | |
2775 | while (1) { | |
2776 | leaf = path->nodes[0]; | |
2777 | slot = path->slots[0]; | |
2778 | ||
2779 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2780 | if (key.objectid != objectid || key.type != BTRFS_ROOT_REF_KEY) { | |
2781 | ret = 0; | |
2782 | goto out; | |
2783 | } | |
2784 | ||
2785 | if (found == BTRFS_MAX_ROOTREF_BUFFER_NUM) { | |
2786 | ret = -EOVERFLOW; | |
2787 | goto out; | |
2788 | } | |
2789 | ||
2790 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2791 | rootrefs->rootref[found].treeid = key.offset; | |
2792 | rootrefs->rootref[found].dirid = | |
2793 | btrfs_root_ref_dirid(leaf, rref); | |
2794 | found++; | |
2795 | ||
2796 | ret = btrfs_next_item(root, path); | |
2797 | if (ret < 0) { | |
2798 | goto out; | |
2799 | } else if (ret > 0) { | |
2800 | ret = -EUCLEAN; | |
2801 | goto out; | |
2802 | } | |
2803 | } | |
2804 | ||
2805 | out: | |
2806 | if (!ret || ret == -EOVERFLOW) { | |
2807 | rootrefs->num_items = found; | |
2808 | /* update min_treeid for next search */ | |
2809 | if (found) | |
2810 | rootrefs->min_treeid = | |
2811 | rootrefs->rootref[found - 1].treeid + 1; | |
2812 | if (copy_to_user(argp, rootrefs, sizeof(*rootrefs))) | |
2813 | ret = -EFAULT; | |
2814 | } | |
2815 | ||
2816 | kfree(rootrefs); | |
2817 | btrfs_free_path(path); | |
2818 | ||
2819 | return ret; | |
2820 | } | |
2821 | ||
76dda93c | 2822 | static noinline int btrfs_ioctl_snap_destroy(struct file *file, |
949964c9 MPS |
2823 | void __user *arg, |
2824 | bool destroy_v2) | |
76dda93c | 2825 | { |
54563d41 | 2826 | struct dentry *parent = file->f_path.dentry; |
0b246afa | 2827 | struct btrfs_fs_info *fs_info = btrfs_sb(parent->d_sb); |
76dda93c | 2828 | struct dentry *dentry; |
2b0143b5 | 2829 | struct inode *dir = d_inode(parent); |
76dda93c YZ |
2830 | struct inode *inode; |
2831 | struct btrfs_root *root = BTRFS_I(dir)->root; | |
2832 | struct btrfs_root *dest = NULL; | |
949964c9 MPS |
2833 | struct btrfs_ioctl_vol_args *vol_args = NULL; |
2834 | struct btrfs_ioctl_vol_args_v2 *vol_args2 = NULL; | |
2835 | char *subvol_name, *subvol_name_ptr = NULL; | |
2836 | int subvol_namelen; | |
76dda93c | 2837 | int err = 0; |
949964c9 | 2838 | bool destroy_parent = false; |
76dda93c | 2839 | |
949964c9 MPS |
2840 | if (destroy_v2) { |
2841 | vol_args2 = memdup_user(arg, sizeof(*vol_args2)); | |
2842 | if (IS_ERR(vol_args2)) | |
2843 | return PTR_ERR(vol_args2); | |
325c50e3 | 2844 | |
949964c9 MPS |
2845 | if (vol_args2->flags & ~BTRFS_SUBVOL_DELETE_ARGS_MASK) { |
2846 | err = -EOPNOTSUPP; | |
2847 | goto out; | |
2848 | } | |
76dda93c | 2849 | |
949964c9 MPS |
2850 | /* |
2851 | * If SPEC_BY_ID is not set, we are looking for the subvolume by | |
2852 | * name, same as v1 currently does. | |
2853 | */ | |
2854 | if (!(vol_args2->flags & BTRFS_SUBVOL_SPEC_BY_ID)) { | |
2855 | vol_args2->name[BTRFS_SUBVOL_NAME_MAX] = 0; | |
2856 | subvol_name = vol_args2->name; | |
2857 | ||
2858 | err = mnt_want_write_file(file); | |
2859 | if (err) | |
2860 | goto out; | |
2861 | } else { | |
2862 | if (vol_args2->subvolid < BTRFS_FIRST_FREE_OBJECTID) { | |
2863 | err = -EINVAL; | |
2864 | goto out; | |
2865 | } | |
2866 | ||
2867 | err = mnt_want_write_file(file); | |
2868 | if (err) | |
2869 | goto out; | |
2870 | ||
2871 | dentry = btrfs_get_dentry(fs_info->sb, | |
2872 | BTRFS_FIRST_FREE_OBJECTID, | |
2873 | vol_args2->subvolid, 0, 0); | |
2874 | if (IS_ERR(dentry)) { | |
2875 | err = PTR_ERR(dentry); | |
2876 | goto out_drop_write; | |
2877 | } | |
2878 | ||
2879 | /* | |
2880 | * Change the default parent since the subvolume being | |
2881 | * deleted can be outside of the current mount point. | |
2882 | */ | |
2883 | parent = btrfs_get_parent(dentry); | |
2884 | ||
2885 | /* | |
2886 | * At this point dentry->d_name can point to '/' if the | |
2887 | * subvolume we want to destroy is outsite of the | |
2888 | * current mount point, so we need to release the | |
2889 | * current dentry and execute the lookup to return a new | |
2890 | * one with ->d_name pointing to the | |
2891 | * <mount point>/subvol_name. | |
2892 | */ | |
2893 | dput(dentry); | |
2894 | if (IS_ERR(parent)) { | |
2895 | err = PTR_ERR(parent); | |
2896 | goto out_drop_write; | |
2897 | } | |
2898 | dir = d_inode(parent); | |
2899 | ||
2900 | /* | |
2901 | * If v2 was used with SPEC_BY_ID, a new parent was | |
2902 | * allocated since the subvolume can be outside of the | |
2903 | * current mount point. Later on we need to release this | |
2904 | * new parent dentry. | |
2905 | */ | |
2906 | destroy_parent = true; | |
2907 | ||
2908 | subvol_name_ptr = btrfs_get_subvol_name_from_objectid( | |
2909 | fs_info, vol_args2->subvolid); | |
2910 | if (IS_ERR(subvol_name_ptr)) { | |
2911 | err = PTR_ERR(subvol_name_ptr); | |
2912 | goto free_parent; | |
2913 | } | |
2914 | /* subvol_name_ptr is already NULL termined */ | |
2915 | subvol_name = (char *)kbasename(subvol_name_ptr); | |
2916 | } | |
2917 | } else { | |
2918 | vol_args = memdup_user(arg, sizeof(*vol_args)); | |
2919 | if (IS_ERR(vol_args)) | |
2920 | return PTR_ERR(vol_args); | |
2921 | ||
2922 | vol_args->name[BTRFS_PATH_NAME_MAX] = 0; | |
2923 | subvol_name = vol_args->name; | |
2924 | ||
2925 | err = mnt_want_write_file(file); | |
2926 | if (err) | |
2927 | goto out; | |
76dda93c YZ |
2928 | } |
2929 | ||
949964c9 | 2930 | subvol_namelen = strlen(subvol_name); |
76dda93c | 2931 | |
949964c9 MPS |
2932 | if (strchr(subvol_name, '/') || |
2933 | strncmp(subvol_name, "..", subvol_namelen) == 0) { | |
2934 | err = -EINVAL; | |
2935 | goto free_subvol_name; | |
2936 | } | |
2937 | ||
2938 | if (!S_ISDIR(dir->i_mode)) { | |
2939 | err = -ENOTDIR; | |
2940 | goto free_subvol_name; | |
2941 | } | |
521e0546 | 2942 | |
00235411 AV |
2943 | err = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT); |
2944 | if (err == -EINTR) | |
949964c9 MPS |
2945 | goto free_subvol_name; |
2946 | dentry = lookup_one_len(subvol_name, parent, subvol_namelen); | |
76dda93c YZ |
2947 | if (IS_ERR(dentry)) { |
2948 | err = PTR_ERR(dentry); | |
2949 | goto out_unlock_dir; | |
2950 | } | |
2951 | ||
2b0143b5 | 2952 | if (d_really_is_negative(dentry)) { |
76dda93c YZ |
2953 | err = -ENOENT; |
2954 | goto out_dput; | |
2955 | } | |
2956 | ||
2b0143b5 | 2957 | inode = d_inode(dentry); |
4260f7c7 | 2958 | dest = BTRFS_I(inode)->root; |
67871254 | 2959 | if (!capable(CAP_SYS_ADMIN)) { |
4260f7c7 SW |
2960 | /* |
2961 | * Regular user. Only allow this with a special mount | |
2962 | * option, when the user has write+exec access to the | |
2963 | * subvol root, and when rmdir(2) would have been | |
2964 | * allowed. | |
2965 | * | |
2966 | * Note that this is _not_ check that the subvol is | |
2967 | * empty or doesn't contain data that we wouldn't | |
2968 | * otherwise be able to delete. | |
2969 | * | |
2970 | * Users who want to delete empty subvols should try | |
2971 | * rmdir(2). | |
2972 | */ | |
2973 | err = -EPERM; | |
0b246afa | 2974 | if (!btrfs_test_opt(fs_info, USER_SUBVOL_RM_ALLOWED)) |
4260f7c7 SW |
2975 | goto out_dput; |
2976 | ||
2977 | /* | |
2978 | * Do not allow deletion if the parent dir is the same | |
2979 | * as the dir to be deleted. That means the ioctl | |
2980 | * must be called on the dentry referencing the root | |
2981 | * of the subvol, not a random directory contained | |
2982 | * within it. | |
2983 | */ | |
2984 | err = -EINVAL; | |
2985 | if (root == dest) | |
2986 | goto out_dput; | |
2987 | ||
47291baa CB |
2988 | err = inode_permission(&init_user_ns, inode, |
2989 | MAY_WRITE | MAY_EXEC); | |
4260f7c7 SW |
2990 | if (err) |
2991 | goto out_dput; | |
4260f7c7 SW |
2992 | } |
2993 | ||
5c39da5b MX |
2994 | /* check if subvolume may be deleted by a user */ |
2995 | err = btrfs_may_delete(dir, dentry, 1); | |
2996 | if (err) | |
2997 | goto out_dput; | |
2998 | ||
4a0cc7ca | 2999 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
76dda93c YZ |
3000 | err = -EINVAL; |
3001 | goto out_dput; | |
3002 | } | |
3003 | ||
64708539 | 3004 | btrfs_inode_lock(inode, 0); |
f60a2364 | 3005 | err = btrfs_delete_subvolume(dir, dentry); |
64708539 | 3006 | btrfs_inode_unlock(inode, 0); |
46008d9d AG |
3007 | if (!err) { |
3008 | fsnotify_rmdir(dir, dentry); | |
76dda93c | 3009 | d_delete(dentry); |
46008d9d | 3010 | } |
fa6ac876 | 3011 | |
76dda93c YZ |
3012 | out_dput: |
3013 | dput(dentry); | |
3014 | out_unlock_dir: | |
64708539 | 3015 | btrfs_inode_unlock(dir, 0); |
949964c9 MPS |
3016 | free_subvol_name: |
3017 | kfree(subvol_name_ptr); | |
3018 | free_parent: | |
3019 | if (destroy_parent) | |
3020 | dput(parent); | |
00235411 | 3021 | out_drop_write: |
2a79f17e | 3022 | mnt_drop_write_file(file); |
76dda93c | 3023 | out: |
949964c9 | 3024 | kfree(vol_args2); |
76dda93c YZ |
3025 | kfree(vol_args); |
3026 | return err; | |
3027 | } | |
3028 | ||
1e701a32 | 3029 | static int btrfs_ioctl_defrag(struct file *file, void __user *argp) |
f46b5a66 | 3030 | { |
496ad9aa | 3031 | struct inode *inode = file_inode(file); |
f46b5a66 | 3032 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1e701a32 | 3033 | struct btrfs_ioctl_defrag_range_args *range; |
c146afad YZ |
3034 | int ret; |
3035 | ||
25122d15 ID |
3036 | ret = mnt_want_write_file(file); |
3037 | if (ret) | |
3038 | return ret; | |
b83cc969 | 3039 | |
25122d15 ID |
3040 | if (btrfs_root_readonly(root)) { |
3041 | ret = -EROFS; | |
3042 | goto out; | |
5ac00add | 3043 | } |
f46b5a66 CH |
3044 | |
3045 | switch (inode->i_mode & S_IFMT) { | |
3046 | case S_IFDIR: | |
e441d54d CM |
3047 | if (!capable(CAP_SYS_ADMIN)) { |
3048 | ret = -EPERM; | |
3049 | goto out; | |
3050 | } | |
de78b51a | 3051 | ret = btrfs_defrag_root(root); |
f46b5a66 CH |
3052 | break; |
3053 | case S_IFREG: | |
616d374e AB |
3054 | /* |
3055 | * Note that this does not check the file descriptor for write | |
3056 | * access. This prevents defragmenting executables that are | |
3057 | * running and allows defrag on files open in read-only mode. | |
3058 | */ | |
3059 | if (!capable(CAP_SYS_ADMIN) && | |
47291baa | 3060 | inode_permission(&init_user_ns, inode, MAY_WRITE)) { |
616d374e | 3061 | ret = -EPERM; |
e441d54d CM |
3062 | goto out; |
3063 | } | |
1e701a32 CM |
3064 | |
3065 | range = kzalloc(sizeof(*range), GFP_KERNEL); | |
3066 | if (!range) { | |
3067 | ret = -ENOMEM; | |
3068 | goto out; | |
3069 | } | |
3070 | ||
3071 | if (argp) { | |
3072 | if (copy_from_user(range, argp, | |
3073 | sizeof(*range))) { | |
3074 | ret = -EFAULT; | |
3075 | kfree(range); | |
683be16e | 3076 | goto out; |
1e701a32 CM |
3077 | } |
3078 | /* compression requires us to start the IO */ | |
3079 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
3080 | range->flags |= BTRFS_DEFRAG_RANGE_START_IO; | |
3081 | range->extent_thresh = (u32)-1; | |
3082 | } | |
3083 | } else { | |
3084 | /* the rest are all set to zero by kzalloc */ | |
3085 | range->len = (u64)-1; | |
3086 | } | |
496ad9aa | 3087 | ret = btrfs_defrag_file(file_inode(file), file, |
7c829b72 | 3088 | range, BTRFS_OLDEST_GENERATION, 0); |
4cb5300b CM |
3089 | if (ret > 0) |
3090 | ret = 0; | |
1e701a32 | 3091 | kfree(range); |
f46b5a66 | 3092 | break; |
8929ecfa YZ |
3093 | default: |
3094 | ret = -EINVAL; | |
f46b5a66 | 3095 | } |
e441d54d | 3096 | out: |
25122d15 | 3097 | mnt_drop_write_file(file); |
e441d54d | 3098 | return ret; |
f46b5a66 CH |
3099 | } |
3100 | ||
2ff7e61e | 3101 | static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg) |
f46b5a66 CH |
3102 | { |
3103 | struct btrfs_ioctl_vol_args *vol_args; | |
3104 | int ret; | |
3105 | ||
e441d54d CM |
3106 | if (!capable(CAP_SYS_ADMIN)) |
3107 | return -EPERM; | |
3108 | ||
c3e1f96c | 3109 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_ADD)) |
e57138b3 | 3110 | return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
c9e9f97b | 3111 | |
dae7b665 | 3112 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
3113 | if (IS_ERR(vol_args)) { |
3114 | ret = PTR_ERR(vol_args); | |
3115 | goto out; | |
3116 | } | |
f46b5a66 | 3117 | |
5516e595 | 3118 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
2ff7e61e | 3119 | ret = btrfs_init_new_device(fs_info, vol_args->name); |
f46b5a66 | 3120 | |
43d20761 | 3121 | if (!ret) |
0b246afa | 3122 | btrfs_info(fs_info, "disk added %s", vol_args->name); |
43d20761 | 3123 | |
f46b5a66 | 3124 | kfree(vol_args); |
c9e9f97b | 3125 | out: |
c3e1f96c | 3126 | btrfs_exclop_finish(fs_info); |
f46b5a66 CH |
3127 | return ret; |
3128 | } | |
3129 | ||
6b526ed7 | 3130 | static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg) |
f46b5a66 | 3131 | { |
0b246afa JM |
3132 | struct inode *inode = file_inode(file); |
3133 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
6b526ed7 | 3134 | struct btrfs_ioctl_vol_args_v2 *vol_args; |
f46b5a66 CH |
3135 | int ret; |
3136 | ||
e441d54d CM |
3137 | if (!capable(CAP_SYS_ADMIN)) |
3138 | return -EPERM; | |
3139 | ||
da24927b MX |
3140 | ret = mnt_want_write_file(file); |
3141 | if (ret) | |
3142 | return ret; | |
c146afad | 3143 | |
dae7b665 | 3144 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
3145 | if (IS_ERR(vol_args)) { |
3146 | ret = PTR_ERR(vol_args); | |
c47ca32d | 3147 | goto err_drop; |
c9e9f97b | 3148 | } |
f46b5a66 | 3149 | |
748449cd | 3150 | if (vol_args->flags & ~BTRFS_DEVICE_REMOVE_ARGS_MASK) { |
fd4e994b OS |
3151 | ret = -EOPNOTSUPP; |
3152 | goto out; | |
3153 | } | |
f46b5a66 | 3154 | |
c3e1f96c | 3155 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REMOVE)) { |
183860f6 AJ |
3156 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
3157 | goto out; | |
3158 | } | |
3159 | ||
735654ea | 3160 | if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) { |
2ff7e61e | 3161 | ret = btrfs_rm_device(fs_info, NULL, vol_args->devid); |
6b526ed7 AJ |
3162 | } else { |
3163 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
2ff7e61e | 3164 | ret = btrfs_rm_device(fs_info, vol_args->name, 0); |
6b526ed7 | 3165 | } |
c3e1f96c | 3166 | btrfs_exclop_finish(fs_info); |
183860f6 | 3167 | |
6b526ed7 | 3168 | if (!ret) { |
735654ea | 3169 | if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) |
0b246afa | 3170 | btrfs_info(fs_info, "device deleted: id %llu", |
6b526ed7 AJ |
3171 | vol_args->devid); |
3172 | else | |
0b246afa | 3173 | btrfs_info(fs_info, "device deleted: %s", |
6b526ed7 AJ |
3174 | vol_args->name); |
3175 | } | |
183860f6 AJ |
3176 | out: |
3177 | kfree(vol_args); | |
c47ca32d | 3178 | err_drop: |
4ac20c70 | 3179 | mnt_drop_write_file(file); |
f46b5a66 CH |
3180 | return ret; |
3181 | } | |
3182 | ||
da24927b | 3183 | static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg) |
f46b5a66 | 3184 | { |
0b246afa JM |
3185 | struct inode *inode = file_inode(file); |
3186 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
f46b5a66 CH |
3187 | struct btrfs_ioctl_vol_args *vol_args; |
3188 | int ret; | |
3189 | ||
e441d54d CM |
3190 | if (!capable(CAP_SYS_ADMIN)) |
3191 | return -EPERM; | |
3192 | ||
da24927b MX |
3193 | ret = mnt_want_write_file(file); |
3194 | if (ret) | |
3195 | return ret; | |
c146afad | 3196 | |
c3e1f96c | 3197 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REMOVE)) { |
183860f6 | 3198 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
58d7bbf8 DS |
3199 | goto out_drop_write; |
3200 | } | |
3201 | ||
3202 | vol_args = memdup_user(arg, sizeof(*vol_args)); | |
3203 | if (IS_ERR(vol_args)) { | |
3204 | ret = PTR_ERR(vol_args); | |
183860f6 AJ |
3205 | goto out; |
3206 | } | |
3207 | ||
58d7bbf8 | 3208 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
2ff7e61e | 3209 | ret = btrfs_rm_device(fs_info, vol_args->name, 0); |
183860f6 | 3210 | |
ec95d491 | 3211 | if (!ret) |
0b246afa | 3212 | btrfs_info(fs_info, "disk deleted %s", vol_args->name); |
183860f6 | 3213 | kfree(vol_args); |
58d7bbf8 | 3214 | out: |
c3e1f96c | 3215 | btrfs_exclop_finish(fs_info); |
58d7bbf8 | 3216 | out_drop_write: |
4ac20c70 | 3217 | mnt_drop_write_file(file); |
58d7bbf8 | 3218 | |
f46b5a66 CH |
3219 | return ret; |
3220 | } | |
3221 | ||
2ff7e61e JM |
3222 | static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info, |
3223 | void __user *arg) | |
475f6387 | 3224 | { |
027ed2f0 | 3225 | struct btrfs_ioctl_fs_info_args *fi_args; |
475f6387 | 3226 | struct btrfs_device *device; |
0b246afa | 3227 | struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; |
137c5418 | 3228 | u64 flags_in; |
027ed2f0 | 3229 | int ret = 0; |
475f6387 | 3230 | |
137c5418 JT |
3231 | fi_args = memdup_user(arg, sizeof(*fi_args)); |
3232 | if (IS_ERR(fi_args)) | |
3233 | return PTR_ERR(fi_args); | |
3234 | ||
3235 | flags_in = fi_args->flags; | |
3236 | memset(fi_args, 0, sizeof(*fi_args)); | |
027ed2f0 | 3237 | |
d03262c7 | 3238 | rcu_read_lock(); |
027ed2f0 | 3239 | fi_args->num_devices = fs_devices->num_devices; |
475f6387 | 3240 | |
d03262c7 | 3241 | list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) { |
027ed2f0 LZ |
3242 | if (device->devid > fi_args->max_id) |
3243 | fi_args->max_id = device->devid; | |
475f6387 | 3244 | } |
d03262c7 | 3245 | rcu_read_unlock(); |
475f6387 | 3246 | |
de37aa51 | 3247 | memcpy(&fi_args->fsid, fs_devices->fsid, sizeof(fi_args->fsid)); |
bea7eafd OS |
3248 | fi_args->nodesize = fs_info->nodesize; |
3249 | fi_args->sectorsize = fs_info->sectorsize; | |
3250 | fi_args->clone_alignment = fs_info->sectorsize; | |
80a773fb | 3251 | |
137c5418 JT |
3252 | if (flags_in & BTRFS_FS_INFO_FLAG_CSUM_INFO) { |
3253 | fi_args->csum_type = btrfs_super_csum_type(fs_info->super_copy); | |
3254 | fi_args->csum_size = btrfs_super_csum_size(fs_info->super_copy); | |
3255 | fi_args->flags |= BTRFS_FS_INFO_FLAG_CSUM_INFO; | |
3256 | } | |
3257 | ||
0fb408a5 JT |
3258 | if (flags_in & BTRFS_FS_INFO_FLAG_GENERATION) { |
3259 | fi_args->generation = fs_info->generation; | |
3260 | fi_args->flags |= BTRFS_FS_INFO_FLAG_GENERATION; | |
3261 | } | |
3262 | ||
49bac897 JT |
3263 | if (flags_in & BTRFS_FS_INFO_FLAG_METADATA_UUID) { |
3264 | memcpy(&fi_args->metadata_uuid, fs_devices->metadata_uuid, | |
3265 | sizeof(fi_args->metadata_uuid)); | |
3266 | fi_args->flags |= BTRFS_FS_INFO_FLAG_METADATA_UUID; | |
3267 | } | |
3268 | ||
027ed2f0 LZ |
3269 | if (copy_to_user(arg, fi_args, sizeof(*fi_args))) |
3270 | ret = -EFAULT; | |
475f6387 | 3271 | |
027ed2f0 LZ |
3272 | kfree(fi_args); |
3273 | return ret; | |
475f6387 JS |
3274 | } |
3275 | ||
2ff7e61e JM |
3276 | static long btrfs_ioctl_dev_info(struct btrfs_fs_info *fs_info, |
3277 | void __user *arg) | |
475f6387 JS |
3278 | { |
3279 | struct btrfs_ioctl_dev_info_args *di_args; | |
3280 | struct btrfs_device *dev; | |
475f6387 JS |
3281 | int ret = 0; |
3282 | char *s_uuid = NULL; | |
475f6387 | 3283 | |
475f6387 JS |
3284 | di_args = memdup_user(arg, sizeof(*di_args)); |
3285 | if (IS_ERR(di_args)) | |
3286 | return PTR_ERR(di_args); | |
3287 | ||
dd5f9615 | 3288 | if (!btrfs_is_empty_uuid(di_args->uuid)) |
475f6387 JS |
3289 | s_uuid = di_args->uuid; |
3290 | ||
c5593ca3 | 3291 | rcu_read_lock(); |
e4319cd9 | 3292 | dev = btrfs_find_device(fs_info->fs_devices, di_args->devid, s_uuid, |
b2598edf | 3293 | NULL); |
475f6387 JS |
3294 | |
3295 | if (!dev) { | |
3296 | ret = -ENODEV; | |
3297 | goto out; | |
3298 | } | |
3299 | ||
3300 | di_args->devid = dev->devid; | |
7cc8e58d MX |
3301 | di_args->bytes_used = btrfs_device_get_bytes_used(dev); |
3302 | di_args->total_bytes = btrfs_device_get_total_bytes(dev); | |
475f6387 | 3303 | memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid)); |
a27202fb | 3304 | if (dev->name) { |
672d5990 MT |
3305 | strncpy(di_args->path, rcu_str_deref(dev->name), |
3306 | sizeof(di_args->path) - 1); | |
a27202fb JM |
3307 | di_args->path[sizeof(di_args->path) - 1] = 0; |
3308 | } else { | |
99ba55ad | 3309 | di_args->path[0] = '\0'; |
a27202fb | 3310 | } |
475f6387 JS |
3311 | |
3312 | out: | |
c5593ca3 | 3313 | rcu_read_unlock(); |
475f6387 JS |
3314 | if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args))) |
3315 | ret = -EFAULT; | |
3316 | ||
3317 | kfree(di_args); | |
3318 | return ret; | |
3319 | } | |
3320 | ||
6ef5ed0d JB |
3321 | static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) |
3322 | { | |
496ad9aa | 3323 | struct inode *inode = file_inode(file); |
0b246afa | 3324 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
6ef5ed0d JB |
3325 | struct btrfs_root *root = BTRFS_I(inode)->root; |
3326 | struct btrfs_root *new_root; | |
3327 | struct btrfs_dir_item *di; | |
3328 | struct btrfs_trans_handle *trans; | |
2a2b5d62 | 3329 | struct btrfs_path *path = NULL; |
6ef5ed0d | 3330 | struct btrfs_disk_key disk_key; |
6ef5ed0d JB |
3331 | u64 objectid = 0; |
3332 | u64 dir_id; | |
3c04ce01 | 3333 | int ret; |
6ef5ed0d JB |
3334 | |
3335 | if (!capable(CAP_SYS_ADMIN)) | |
3336 | return -EPERM; | |
3337 | ||
3c04ce01 MX |
3338 | ret = mnt_want_write_file(file); |
3339 | if (ret) | |
3340 | return ret; | |
3341 | ||
3342 | if (copy_from_user(&objectid, argp, sizeof(objectid))) { | |
3343 | ret = -EFAULT; | |
3344 | goto out; | |
3345 | } | |
6ef5ed0d JB |
3346 | |
3347 | if (!objectid) | |
1cecf579 | 3348 | objectid = BTRFS_FS_TREE_OBJECTID; |
6ef5ed0d | 3349 | |
56e9357a | 3350 | new_root = btrfs_get_fs_root(fs_info, objectid, true); |
3c04ce01 MX |
3351 | if (IS_ERR(new_root)) { |
3352 | ret = PTR_ERR(new_root); | |
3353 | goto out; | |
3354 | } | |
2a2b5d62 JB |
3355 | if (!is_fstree(new_root->root_key.objectid)) { |
3356 | ret = -ENOENT; | |
3357 | goto out_free; | |
3358 | } | |
6ef5ed0d | 3359 | |
6ef5ed0d | 3360 | path = btrfs_alloc_path(); |
3c04ce01 MX |
3361 | if (!path) { |
3362 | ret = -ENOMEM; | |
2a2b5d62 | 3363 | goto out_free; |
3c04ce01 | 3364 | } |
6ef5ed0d JB |
3365 | |
3366 | trans = btrfs_start_transaction(root, 1); | |
98d5dc13 | 3367 | if (IS_ERR(trans)) { |
3c04ce01 | 3368 | ret = PTR_ERR(trans); |
2a2b5d62 | 3369 | goto out_free; |
6ef5ed0d JB |
3370 | } |
3371 | ||
0b246afa JM |
3372 | dir_id = btrfs_super_root_dir(fs_info->super_copy); |
3373 | di = btrfs_lookup_dir_item(trans, fs_info->tree_root, path, | |
6ef5ed0d | 3374 | dir_id, "default", 7, 1); |
cf1e99a4 | 3375 | if (IS_ERR_OR_NULL(di)) { |
2a2b5d62 | 3376 | btrfs_release_path(path); |
3a45bb20 | 3377 | btrfs_end_transaction(trans); |
0b246afa | 3378 | btrfs_err(fs_info, |
5d163e0e | 3379 | "Umm, you don't have the default diritem, this isn't going to work"); |
3c04ce01 | 3380 | ret = -ENOENT; |
2a2b5d62 | 3381 | goto out_free; |
6ef5ed0d JB |
3382 | } |
3383 | ||
3384 | btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key); | |
3385 | btrfs_set_dir_item_key(path->nodes[0], di, &disk_key); | |
3386 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
2a2b5d62 | 3387 | btrfs_release_path(path); |
6ef5ed0d | 3388 | |
0b246afa | 3389 | btrfs_set_fs_incompat(fs_info, DEFAULT_SUBVOL); |
3a45bb20 | 3390 | btrfs_end_transaction(trans); |
2a2b5d62 | 3391 | out_free: |
00246528 | 3392 | btrfs_put_root(new_root); |
2a2b5d62 | 3393 | btrfs_free_path(path); |
3c04ce01 MX |
3394 | out: |
3395 | mnt_drop_write_file(file); | |
3396 | return ret; | |
6ef5ed0d JB |
3397 | } |
3398 | ||
c065f5b1 SY |
3399 | static void get_block_group_info(struct list_head *groups_list, |
3400 | struct btrfs_ioctl_space_info *space) | |
bf5fc093 | 3401 | { |
32da5386 | 3402 | struct btrfs_block_group *block_group; |
bf5fc093 JB |
3403 | |
3404 | space->total_bytes = 0; | |
3405 | space->used_bytes = 0; | |
3406 | space->flags = 0; | |
3407 | list_for_each_entry(block_group, groups_list, list) { | |
3408 | space->flags = block_group->flags; | |
b3470b5d | 3409 | space->total_bytes += block_group->length; |
bf38be65 | 3410 | space->used_bytes += block_group->used; |
bf5fc093 JB |
3411 | } |
3412 | } | |
3413 | ||
2ff7e61e JM |
3414 | static long btrfs_ioctl_space_info(struct btrfs_fs_info *fs_info, |
3415 | void __user *arg) | |
1406e432 JB |
3416 | { |
3417 | struct btrfs_ioctl_space_args space_args; | |
3418 | struct btrfs_ioctl_space_info space; | |
3419 | struct btrfs_ioctl_space_info *dest; | |
7fde62bf | 3420 | struct btrfs_ioctl_space_info *dest_orig; |
13f2696f | 3421 | struct btrfs_ioctl_space_info __user *user_dest; |
1406e432 | 3422 | struct btrfs_space_info *info; |
315d8e98 CIK |
3423 | static const u64 types[] = { |
3424 | BTRFS_BLOCK_GROUP_DATA, | |
3425 | BTRFS_BLOCK_GROUP_SYSTEM, | |
3426 | BTRFS_BLOCK_GROUP_METADATA, | |
3427 | BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA | |
3428 | }; | |
bf5fc093 | 3429 | int num_types = 4; |
7fde62bf | 3430 | int alloc_size; |
1406e432 | 3431 | int ret = 0; |
51788b1b | 3432 | u64 slot_count = 0; |
bf5fc093 | 3433 | int i, c; |
1406e432 JB |
3434 | |
3435 | if (copy_from_user(&space_args, | |
3436 | (struct btrfs_ioctl_space_args __user *)arg, | |
3437 | sizeof(space_args))) | |
3438 | return -EFAULT; | |
3439 | ||
bf5fc093 JB |
3440 | for (i = 0; i < num_types; i++) { |
3441 | struct btrfs_space_info *tmp; | |
3442 | ||
3443 | info = NULL; | |
72804905 | 3444 | list_for_each_entry(tmp, &fs_info->space_info, list) { |
bf5fc093 JB |
3445 | if (tmp->flags == types[i]) { |
3446 | info = tmp; | |
3447 | break; | |
3448 | } | |
3449 | } | |
bf5fc093 JB |
3450 | |
3451 | if (!info) | |
3452 | continue; | |
3453 | ||
3454 | down_read(&info->groups_sem); | |
3455 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3456 | if (!list_empty(&info->block_groups[c])) | |
3457 | slot_count++; | |
3458 | } | |
3459 | up_read(&info->groups_sem); | |
3460 | } | |
7fde62bf | 3461 | |
36523e95 DS |
3462 | /* |
3463 | * Global block reserve, exported as a space_info | |
3464 | */ | |
3465 | slot_count++; | |
3466 | ||
7fde62bf CM |
3467 | /* space_slots == 0 means they are asking for a count */ |
3468 | if (space_args.space_slots == 0) { | |
3469 | space_args.total_spaces = slot_count; | |
3470 | goto out; | |
3471 | } | |
bf5fc093 | 3472 | |
51788b1b | 3473 | slot_count = min_t(u64, space_args.space_slots, slot_count); |
bf5fc093 | 3474 | |
7fde62bf | 3475 | alloc_size = sizeof(*dest) * slot_count; |
bf5fc093 | 3476 | |
7fde62bf CM |
3477 | /* we generally have at most 6 or so space infos, one for each raid |
3478 | * level. So, a whole page should be more than enough for everyone | |
3479 | */ | |
09cbfeaf | 3480 | if (alloc_size > PAGE_SIZE) |
7fde62bf CM |
3481 | return -ENOMEM; |
3482 | ||
1406e432 | 3483 | space_args.total_spaces = 0; |
8d2db785 | 3484 | dest = kmalloc(alloc_size, GFP_KERNEL); |
7fde62bf CM |
3485 | if (!dest) |
3486 | return -ENOMEM; | |
3487 | dest_orig = dest; | |
1406e432 | 3488 | |
7fde62bf | 3489 | /* now we have a buffer to copy into */ |
bf5fc093 JB |
3490 | for (i = 0; i < num_types; i++) { |
3491 | struct btrfs_space_info *tmp; | |
3492 | ||
51788b1b DR |
3493 | if (!slot_count) |
3494 | break; | |
3495 | ||
bf5fc093 | 3496 | info = NULL; |
72804905 | 3497 | list_for_each_entry(tmp, &fs_info->space_info, list) { |
bf5fc093 JB |
3498 | if (tmp->flags == types[i]) { |
3499 | info = tmp; | |
3500 | break; | |
3501 | } | |
3502 | } | |
7fde62bf | 3503 | |
bf5fc093 JB |
3504 | if (!info) |
3505 | continue; | |
3506 | down_read(&info->groups_sem); | |
3507 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3508 | if (!list_empty(&info->block_groups[c])) { | |
c065f5b1 SY |
3509 | get_block_group_info(&info->block_groups[c], |
3510 | &space); | |
bf5fc093 JB |
3511 | memcpy(dest, &space, sizeof(space)); |
3512 | dest++; | |
3513 | space_args.total_spaces++; | |
51788b1b | 3514 | slot_count--; |
bf5fc093 | 3515 | } |
51788b1b DR |
3516 | if (!slot_count) |
3517 | break; | |
bf5fc093 JB |
3518 | } |
3519 | up_read(&info->groups_sem); | |
1406e432 | 3520 | } |
1406e432 | 3521 | |
36523e95 DS |
3522 | /* |
3523 | * Add global block reserve | |
3524 | */ | |
3525 | if (slot_count) { | |
0b246afa | 3526 | struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv; |
36523e95 DS |
3527 | |
3528 | spin_lock(&block_rsv->lock); | |
3529 | space.total_bytes = block_rsv->size; | |
3530 | space.used_bytes = block_rsv->size - block_rsv->reserved; | |
3531 | spin_unlock(&block_rsv->lock); | |
3532 | space.flags = BTRFS_SPACE_INFO_GLOBAL_RSV; | |
3533 | memcpy(dest, &space, sizeof(space)); | |
3534 | space_args.total_spaces++; | |
3535 | } | |
3536 | ||
2eec6c81 | 3537 | user_dest = (struct btrfs_ioctl_space_info __user *) |
7fde62bf CM |
3538 | (arg + sizeof(struct btrfs_ioctl_space_args)); |
3539 | ||
3540 | if (copy_to_user(user_dest, dest_orig, alloc_size)) | |
3541 | ret = -EFAULT; | |
3542 | ||
3543 | kfree(dest_orig); | |
3544 | out: | |
3545 | if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args))) | |
1406e432 JB |
3546 | ret = -EFAULT; |
3547 | ||
3548 | return ret; | |
3549 | } | |
3550 | ||
9a8c28be MX |
3551 | static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root, |
3552 | void __user *argp) | |
46204592 | 3553 | { |
46204592 SW |
3554 | struct btrfs_trans_handle *trans; |
3555 | u64 transid; | |
db5b493a | 3556 | int ret; |
46204592 | 3557 | |
d4edf39b | 3558 | trans = btrfs_attach_transaction_barrier(root); |
ff7c1d33 MX |
3559 | if (IS_ERR(trans)) { |
3560 | if (PTR_ERR(trans) != -ENOENT) | |
3561 | return PTR_ERR(trans); | |
3562 | ||
3563 | /* No running transaction, don't bother */ | |
3564 | transid = root->fs_info->last_trans_committed; | |
3565 | goto out; | |
3566 | } | |
46204592 | 3567 | transid = trans->transid; |
3a45bb20 | 3568 | ret = btrfs_commit_transaction_async(trans, 0); |
8b2b2d3c | 3569 | if (ret) { |
3a45bb20 | 3570 | btrfs_end_transaction(trans); |
db5b493a | 3571 | return ret; |
8b2b2d3c | 3572 | } |
ff7c1d33 | 3573 | out: |
46204592 SW |
3574 | if (argp) |
3575 | if (copy_to_user(argp, &transid, sizeof(transid))) | |
3576 | return -EFAULT; | |
3577 | return 0; | |
3578 | } | |
3579 | ||
2ff7e61e | 3580 | static noinline long btrfs_ioctl_wait_sync(struct btrfs_fs_info *fs_info, |
9a8c28be | 3581 | void __user *argp) |
46204592 | 3582 | { |
46204592 SW |
3583 | u64 transid; |
3584 | ||
3585 | if (argp) { | |
3586 | if (copy_from_user(&transid, argp, sizeof(transid))) | |
3587 | return -EFAULT; | |
3588 | } else { | |
3589 | transid = 0; /* current trans */ | |
3590 | } | |
2ff7e61e | 3591 | return btrfs_wait_for_commit(fs_info, transid); |
46204592 SW |
3592 | } |
3593 | ||
b8e95489 | 3594 | static long btrfs_ioctl_scrub(struct file *file, void __user *arg) |
475f6387 | 3595 | { |
0b246afa | 3596 | struct btrfs_fs_info *fs_info = btrfs_sb(file_inode(file)->i_sb); |
475f6387 | 3597 | struct btrfs_ioctl_scrub_args *sa; |
b8e95489 | 3598 | int ret; |
475f6387 JS |
3599 | |
3600 | if (!capable(CAP_SYS_ADMIN)) | |
3601 | return -EPERM; | |
3602 | ||
3603 | sa = memdup_user(arg, sizeof(*sa)); | |
3604 | if (IS_ERR(sa)) | |
3605 | return PTR_ERR(sa); | |
3606 | ||
b8e95489 MX |
3607 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) { |
3608 | ret = mnt_want_write_file(file); | |
3609 | if (ret) | |
3610 | goto out; | |
3611 | } | |
3612 | ||
0b246afa | 3613 | ret = btrfs_scrub_dev(fs_info, sa->devid, sa->start, sa->end, |
63a212ab SB |
3614 | &sa->progress, sa->flags & BTRFS_SCRUB_READONLY, |
3615 | 0); | |
475f6387 | 3616 | |
5afe6ce7 FM |
3617 | /* |
3618 | * Copy scrub args to user space even if btrfs_scrub_dev() returned an | |
3619 | * error. This is important as it allows user space to know how much | |
3620 | * progress scrub has done. For example, if scrub is canceled we get | |
3621 | * -ECANCELED from btrfs_scrub_dev() and return that error back to user | |
3622 | * space. Later user space can inspect the progress from the structure | |
3623 | * btrfs_ioctl_scrub_args and resume scrub from where it left off | |
3624 | * previously (btrfs-progs does this). | |
3625 | * If we fail to copy the btrfs_ioctl_scrub_args structure to user space | |
3626 | * then return -EFAULT to signal the structure was not copied or it may | |
3627 | * be corrupt and unreliable due to a partial copy. | |
3628 | */ | |
3629 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
475f6387 JS |
3630 | ret = -EFAULT; |
3631 | ||
b8e95489 MX |
3632 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) |
3633 | mnt_drop_write_file(file); | |
3634 | out: | |
475f6387 JS |
3635 | kfree(sa); |
3636 | return ret; | |
3637 | } | |
3638 | ||
2ff7e61e | 3639 | static long btrfs_ioctl_scrub_cancel(struct btrfs_fs_info *fs_info) |
475f6387 JS |
3640 | { |
3641 | if (!capable(CAP_SYS_ADMIN)) | |
3642 | return -EPERM; | |
3643 | ||
2ff7e61e | 3644 | return btrfs_scrub_cancel(fs_info); |
475f6387 JS |
3645 | } |
3646 | ||
2ff7e61e | 3647 | static long btrfs_ioctl_scrub_progress(struct btrfs_fs_info *fs_info, |
475f6387 JS |
3648 | void __user *arg) |
3649 | { | |
3650 | struct btrfs_ioctl_scrub_args *sa; | |
3651 | int ret; | |
3652 | ||
3653 | if (!capable(CAP_SYS_ADMIN)) | |
3654 | return -EPERM; | |
3655 | ||
3656 | sa = memdup_user(arg, sizeof(*sa)); | |
3657 | if (IS_ERR(sa)) | |
3658 | return PTR_ERR(sa); | |
3659 | ||
2ff7e61e | 3660 | ret = btrfs_scrub_progress(fs_info, sa->devid, &sa->progress); |
475f6387 | 3661 | |
4fa99b00 | 3662 | if (ret == 0 && copy_to_user(arg, sa, sizeof(*sa))) |
475f6387 JS |
3663 | ret = -EFAULT; |
3664 | ||
3665 | kfree(sa); | |
3666 | return ret; | |
3667 | } | |
3668 | ||
2ff7e61e | 3669 | static long btrfs_ioctl_get_dev_stats(struct btrfs_fs_info *fs_info, |
b27f7c0c | 3670 | void __user *arg) |
c11d2c23 SB |
3671 | { |
3672 | struct btrfs_ioctl_get_dev_stats *sa; | |
3673 | int ret; | |
3674 | ||
c11d2c23 SB |
3675 | sa = memdup_user(arg, sizeof(*sa)); |
3676 | if (IS_ERR(sa)) | |
3677 | return PTR_ERR(sa); | |
3678 | ||
b27f7c0c DS |
3679 | if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) { |
3680 | kfree(sa); | |
3681 | return -EPERM; | |
3682 | } | |
3683 | ||
2ff7e61e | 3684 | ret = btrfs_get_dev_stats(fs_info, sa); |
c11d2c23 | 3685 | |
eee99577 | 3686 | if (ret == 0 && copy_to_user(arg, sa, sizeof(*sa))) |
c11d2c23 SB |
3687 | ret = -EFAULT; |
3688 | ||
3689 | kfree(sa); | |
3690 | return ret; | |
3691 | } | |
3692 | ||
2ff7e61e JM |
3693 | static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info, |
3694 | void __user *arg) | |
3f6bcfbd SB |
3695 | { |
3696 | struct btrfs_ioctl_dev_replace_args *p; | |
3697 | int ret; | |
3698 | ||
3699 | if (!capable(CAP_SYS_ADMIN)) | |
3700 | return -EPERM; | |
3701 | ||
3702 | p = memdup_user(arg, sizeof(*p)); | |
3703 | if (IS_ERR(p)) | |
3704 | return PTR_ERR(p); | |
3705 | ||
3706 | switch (p->cmd) { | |
3707 | case BTRFS_IOCTL_DEV_REPLACE_CMD_START: | |
bc98a42c | 3708 | if (sb_rdonly(fs_info->sb)) { |
adfa97cb ID |
3709 | ret = -EROFS; |
3710 | goto out; | |
3711 | } | |
c3e1f96c | 3712 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REPLACE)) { |
e57138b3 | 3713 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
3f6bcfbd | 3714 | } else { |
2ff7e61e | 3715 | ret = btrfs_dev_replace_by_ioctl(fs_info, p); |
c3e1f96c | 3716 | btrfs_exclop_finish(fs_info); |
3f6bcfbd SB |
3717 | } |
3718 | break; | |
3719 | case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS: | |
0b246afa | 3720 | btrfs_dev_replace_status(fs_info, p); |
3f6bcfbd SB |
3721 | ret = 0; |
3722 | break; | |
3723 | case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL: | |
17d202b9 | 3724 | p->result = btrfs_dev_replace_cancel(fs_info); |
97282031 | 3725 | ret = 0; |
3f6bcfbd SB |
3726 | break; |
3727 | default: | |
3728 | ret = -EINVAL; | |
3729 | break; | |
3730 | } | |
3731 | ||
d3a53286 | 3732 | if ((ret == 0 || ret == -ECANCELED) && copy_to_user(arg, p, sizeof(*p))) |
3f6bcfbd | 3733 | ret = -EFAULT; |
adfa97cb | 3734 | out: |
3f6bcfbd SB |
3735 | kfree(p); |
3736 | return ret; | |
3737 | } | |
3738 | ||
d7728c96 JS |
3739 | static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg) |
3740 | { | |
3741 | int ret = 0; | |
3742 | int i; | |
740c3d22 | 3743 | u64 rel_ptr; |
d7728c96 | 3744 | int size; |
806468f8 | 3745 | struct btrfs_ioctl_ino_path_args *ipa = NULL; |
d7728c96 JS |
3746 | struct inode_fs_paths *ipath = NULL; |
3747 | struct btrfs_path *path; | |
3748 | ||
82b22ac8 | 3749 | if (!capable(CAP_DAC_READ_SEARCH)) |
d7728c96 JS |
3750 | return -EPERM; |
3751 | ||
3752 | path = btrfs_alloc_path(); | |
3753 | if (!path) { | |
3754 | ret = -ENOMEM; | |
3755 | goto out; | |
3756 | } | |
3757 | ||
3758 | ipa = memdup_user(arg, sizeof(*ipa)); | |
3759 | if (IS_ERR(ipa)) { | |
3760 | ret = PTR_ERR(ipa); | |
3761 | ipa = NULL; | |
3762 | goto out; | |
3763 | } | |
3764 | ||
3765 | size = min_t(u32, ipa->size, 4096); | |
3766 | ipath = init_ipath(size, root, path); | |
3767 | if (IS_ERR(ipath)) { | |
3768 | ret = PTR_ERR(ipath); | |
3769 | ipath = NULL; | |
3770 | goto out; | |
3771 | } | |
3772 | ||
3773 | ret = paths_from_inode(ipa->inum, ipath); | |
3774 | if (ret < 0) | |
3775 | goto out; | |
3776 | ||
3777 | for (i = 0; i < ipath->fspath->elem_cnt; ++i) { | |
745c4d8e JM |
3778 | rel_ptr = ipath->fspath->val[i] - |
3779 | (u64)(unsigned long)ipath->fspath->val; | |
740c3d22 | 3780 | ipath->fspath->val[i] = rel_ptr; |
d7728c96 JS |
3781 | } |
3782 | ||
718dc5fa OS |
3783 | ret = copy_to_user((void __user *)(unsigned long)ipa->fspath, |
3784 | ipath->fspath, size); | |
d7728c96 JS |
3785 | if (ret) { |
3786 | ret = -EFAULT; | |
3787 | goto out; | |
3788 | } | |
3789 | ||
3790 | out: | |
3791 | btrfs_free_path(path); | |
3792 | free_ipath(ipath); | |
3793 | kfree(ipa); | |
3794 | ||
3795 | return ret; | |
3796 | } | |
3797 | ||
3798 | static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx) | |
3799 | { | |
3800 | struct btrfs_data_container *inodes = ctx; | |
3801 | const size_t c = 3 * sizeof(u64); | |
3802 | ||
3803 | if (inodes->bytes_left >= c) { | |
3804 | inodes->bytes_left -= c; | |
3805 | inodes->val[inodes->elem_cnt] = inum; | |
3806 | inodes->val[inodes->elem_cnt + 1] = offset; | |
3807 | inodes->val[inodes->elem_cnt + 2] = root; | |
3808 | inodes->elem_cnt += 3; | |
3809 | } else { | |
3810 | inodes->bytes_missing += c - inodes->bytes_left; | |
3811 | inodes->bytes_left = 0; | |
3812 | inodes->elem_missed += 3; | |
3813 | } | |
3814 | ||
3815 | return 0; | |
3816 | } | |
3817 | ||
2ff7e61e | 3818 | static long btrfs_ioctl_logical_to_ino(struct btrfs_fs_info *fs_info, |
d24a67b2 | 3819 | void __user *arg, int version) |
d7728c96 JS |
3820 | { |
3821 | int ret = 0; | |
3822 | int size; | |
d7728c96 JS |
3823 | struct btrfs_ioctl_logical_ino_args *loi; |
3824 | struct btrfs_data_container *inodes = NULL; | |
3825 | struct btrfs_path *path = NULL; | |
d24a67b2 | 3826 | bool ignore_offset; |
d7728c96 JS |
3827 | |
3828 | if (!capable(CAP_SYS_ADMIN)) | |
3829 | return -EPERM; | |
3830 | ||
3831 | loi = memdup_user(arg, sizeof(*loi)); | |
7b9ea627 SV |
3832 | if (IS_ERR(loi)) |
3833 | return PTR_ERR(loi); | |
d7728c96 | 3834 | |
d24a67b2 ZB |
3835 | if (version == 1) { |
3836 | ignore_offset = false; | |
b115e3bc | 3837 | size = min_t(u32, loi->size, SZ_64K); |
d24a67b2 ZB |
3838 | } else { |
3839 | /* All reserved bits must be 0 for now */ | |
3840 | if (memchr_inv(loi->reserved, 0, sizeof(loi->reserved))) { | |
3841 | ret = -EINVAL; | |
3842 | goto out_loi; | |
3843 | } | |
3844 | /* Only accept flags we have defined so far */ | |
3845 | if (loi->flags & ~(BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET)) { | |
3846 | ret = -EINVAL; | |
3847 | goto out_loi; | |
3848 | } | |
3849 | ignore_offset = loi->flags & BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET; | |
b115e3bc | 3850 | size = min_t(u32, loi->size, SZ_16M); |
d24a67b2 ZB |
3851 | } |
3852 | ||
d7728c96 JS |
3853 | path = btrfs_alloc_path(); |
3854 | if (!path) { | |
3855 | ret = -ENOMEM; | |
3856 | goto out; | |
3857 | } | |
3858 | ||
d7728c96 JS |
3859 | inodes = init_data_container(size); |
3860 | if (IS_ERR(inodes)) { | |
3861 | ret = PTR_ERR(inodes); | |
3862 | inodes = NULL; | |
3863 | goto out; | |
3864 | } | |
3865 | ||
2ff7e61e | 3866 | ret = iterate_inodes_from_logical(loi->logical, fs_info, path, |
d24a67b2 | 3867 | build_ino_list, inodes, ignore_offset); |
df031f07 | 3868 | if (ret == -EINVAL) |
d7728c96 JS |
3869 | ret = -ENOENT; |
3870 | if (ret < 0) | |
3871 | goto out; | |
3872 | ||
718dc5fa OS |
3873 | ret = copy_to_user((void __user *)(unsigned long)loi->inodes, inodes, |
3874 | size); | |
d7728c96 JS |
3875 | if (ret) |
3876 | ret = -EFAULT; | |
3877 | ||
3878 | out: | |
3879 | btrfs_free_path(path); | |
f54de068 | 3880 | kvfree(inodes); |
d24a67b2 | 3881 | out_loi: |
d7728c96 JS |
3882 | kfree(loi); |
3883 | ||
3884 | return ret; | |
3885 | } | |
3886 | ||
008ef096 | 3887 | void btrfs_update_ioctl_balance_args(struct btrfs_fs_info *fs_info, |
c9e9f97b ID |
3888 | struct btrfs_ioctl_balance_args *bargs) |
3889 | { | |
3890 | struct btrfs_balance_control *bctl = fs_info->balance_ctl; | |
3891 | ||
3892 | bargs->flags = bctl->flags; | |
3893 | ||
3009a62f | 3894 | if (test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) |
837d5b6e ID |
3895 | bargs->state |= BTRFS_BALANCE_STATE_RUNNING; |
3896 | if (atomic_read(&fs_info->balance_pause_req)) | |
3897 | bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ; | |
a7e99c69 ID |
3898 | if (atomic_read(&fs_info->balance_cancel_req)) |
3899 | bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ; | |
837d5b6e | 3900 | |
c9e9f97b ID |
3901 | memcpy(&bargs->data, &bctl->data, sizeof(bargs->data)); |
3902 | memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta)); | |
3903 | memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys)); | |
19a39dce | 3904 | |
008ef096 DS |
3905 | spin_lock(&fs_info->balance_lock); |
3906 | memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat)); | |
3907 | spin_unlock(&fs_info->balance_lock); | |
c9e9f97b ID |
3908 | } |
3909 | ||
9ba1f6e4 | 3910 | static long btrfs_ioctl_balance(struct file *file, void __user *arg) |
c9e9f97b | 3911 | { |
496ad9aa | 3912 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
c9e9f97b ID |
3913 | struct btrfs_fs_info *fs_info = root->fs_info; |
3914 | struct btrfs_ioctl_balance_args *bargs; | |
3915 | struct btrfs_balance_control *bctl; | |
ed0fb78f | 3916 | bool need_unlock; /* for mut. excl. ops lock */ |
c9e9f97b ID |
3917 | int ret; |
3918 | ||
3919 | if (!capable(CAP_SYS_ADMIN)) | |
3920 | return -EPERM; | |
3921 | ||
e54bfa31 | 3922 | ret = mnt_want_write_file(file); |
9ba1f6e4 LB |
3923 | if (ret) |
3924 | return ret; | |
3925 | ||
ed0fb78f | 3926 | again: |
c3e1f96c | 3927 | if (btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) { |
ed0fb78f ID |
3928 | mutex_lock(&fs_info->balance_mutex); |
3929 | need_unlock = true; | |
3930 | goto locked; | |
3931 | } | |
3932 | ||
3933 | /* | |
01327610 | 3934 | * mut. excl. ops lock is locked. Three possibilities: |
ed0fb78f ID |
3935 | * (1) some other op is running |
3936 | * (2) balance is running | |
3937 | * (3) balance is paused -- special case (think resume) | |
3938 | */ | |
c9e9f97b | 3939 | mutex_lock(&fs_info->balance_mutex); |
ed0fb78f ID |
3940 | if (fs_info->balance_ctl) { |
3941 | /* this is either (2) or (3) */ | |
3009a62f | 3942 | if (!test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) { |
ed0fb78f | 3943 | mutex_unlock(&fs_info->balance_mutex); |
dccdb07b DS |
3944 | /* |
3945 | * Lock released to allow other waiters to continue, | |
3946 | * we'll reexamine the status again. | |
3947 | */ | |
ed0fb78f ID |
3948 | mutex_lock(&fs_info->balance_mutex); |
3949 | ||
3950 | if (fs_info->balance_ctl && | |
3009a62f | 3951 | !test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) { |
ed0fb78f ID |
3952 | /* this is (3) */ |
3953 | need_unlock = false; | |
3954 | goto locked; | |
3955 | } | |
3956 | ||
3957 | mutex_unlock(&fs_info->balance_mutex); | |
ed0fb78f ID |
3958 | goto again; |
3959 | } else { | |
3960 | /* this is (2) */ | |
3961 | mutex_unlock(&fs_info->balance_mutex); | |
3962 | ret = -EINPROGRESS; | |
3963 | goto out; | |
3964 | } | |
3965 | } else { | |
3966 | /* this is (1) */ | |
3967 | mutex_unlock(&fs_info->balance_mutex); | |
e57138b3 | 3968 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
ed0fb78f ID |
3969 | goto out; |
3970 | } | |
3971 | ||
3972 | locked: | |
c9e9f97b ID |
3973 | |
3974 | if (arg) { | |
3975 | bargs = memdup_user(arg, sizeof(*bargs)); | |
3976 | if (IS_ERR(bargs)) { | |
3977 | ret = PTR_ERR(bargs); | |
ed0fb78f | 3978 | goto out_unlock; |
c9e9f97b | 3979 | } |
de322263 ID |
3980 | |
3981 | if (bargs->flags & BTRFS_BALANCE_RESUME) { | |
3982 | if (!fs_info->balance_ctl) { | |
3983 | ret = -ENOTCONN; | |
3984 | goto out_bargs; | |
3985 | } | |
3986 | ||
3987 | bctl = fs_info->balance_ctl; | |
3988 | spin_lock(&fs_info->balance_lock); | |
3989 | bctl->flags |= BTRFS_BALANCE_RESUME; | |
3990 | spin_unlock(&fs_info->balance_lock); | |
3991 | ||
3992 | goto do_balance; | |
3993 | } | |
c9e9f97b ID |
3994 | } else { |
3995 | bargs = NULL; | |
3996 | } | |
3997 | ||
ed0fb78f | 3998 | if (fs_info->balance_ctl) { |
837d5b6e ID |
3999 | ret = -EINPROGRESS; |
4000 | goto out_bargs; | |
4001 | } | |
4002 | ||
8d2db785 | 4003 | bctl = kzalloc(sizeof(*bctl), GFP_KERNEL); |
c9e9f97b ID |
4004 | if (!bctl) { |
4005 | ret = -ENOMEM; | |
4006 | goto out_bargs; | |
4007 | } | |
4008 | ||
c9e9f97b ID |
4009 | if (arg) { |
4010 | memcpy(&bctl->data, &bargs->data, sizeof(bctl->data)); | |
4011 | memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta)); | |
4012 | memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys)); | |
4013 | ||
4014 | bctl->flags = bargs->flags; | |
f43ffb60 ID |
4015 | } else { |
4016 | /* balance everything - no filters */ | |
4017 | bctl->flags |= BTRFS_BALANCE_TYPE_MASK; | |
c9e9f97b ID |
4018 | } |
4019 | ||
8eb93459 DS |
4020 | if (bctl->flags & ~(BTRFS_BALANCE_ARGS_MASK | BTRFS_BALANCE_TYPE_MASK)) { |
4021 | ret = -EINVAL; | |
0f89abf5 | 4022 | goto out_bctl; |
8eb93459 DS |
4023 | } |
4024 | ||
de322263 | 4025 | do_balance: |
c9e9f97b | 4026 | /* |
c3e1f96c GR |
4027 | * Ownership of bctl and exclusive operation goes to btrfs_balance. |
4028 | * bctl is freed in reset_balance_state, or, if restriper was paused | |
4029 | * all the way until unmount, in free_fs_info. The flag should be | |
4030 | * cleared after reset_balance_state. | |
c9e9f97b | 4031 | */ |
ed0fb78f ID |
4032 | need_unlock = false; |
4033 | ||
6fcf6e2b | 4034 | ret = btrfs_balance(fs_info, bctl, bargs); |
0f89abf5 | 4035 | bctl = NULL; |
ed0fb78f | 4036 | |
d00c2d9c | 4037 | if ((ret == 0 || ret == -ECANCELED) && arg) { |
c9e9f97b ID |
4038 | if (copy_to_user(arg, bargs, sizeof(*bargs))) |
4039 | ret = -EFAULT; | |
4040 | } | |
4041 | ||
0f89abf5 CE |
4042 | out_bctl: |
4043 | kfree(bctl); | |
c9e9f97b ID |
4044 | out_bargs: |
4045 | kfree(bargs); | |
ed0fb78f | 4046 | out_unlock: |
c9e9f97b | 4047 | mutex_unlock(&fs_info->balance_mutex); |
ed0fb78f | 4048 | if (need_unlock) |
c3e1f96c | 4049 | btrfs_exclop_finish(fs_info); |
ed0fb78f | 4050 | out: |
e54bfa31 | 4051 | mnt_drop_write_file(file); |
c9e9f97b ID |
4052 | return ret; |
4053 | } | |
4054 | ||
2ff7e61e | 4055 | static long btrfs_ioctl_balance_ctl(struct btrfs_fs_info *fs_info, int cmd) |
837d5b6e ID |
4056 | { |
4057 | if (!capable(CAP_SYS_ADMIN)) | |
4058 | return -EPERM; | |
4059 | ||
4060 | switch (cmd) { | |
4061 | case BTRFS_BALANCE_CTL_PAUSE: | |
0b246afa | 4062 | return btrfs_pause_balance(fs_info); |
a7e99c69 | 4063 | case BTRFS_BALANCE_CTL_CANCEL: |
0b246afa | 4064 | return btrfs_cancel_balance(fs_info); |
837d5b6e ID |
4065 | } |
4066 | ||
4067 | return -EINVAL; | |
4068 | } | |
4069 | ||
2ff7e61e | 4070 | static long btrfs_ioctl_balance_progress(struct btrfs_fs_info *fs_info, |
19a39dce ID |
4071 | void __user *arg) |
4072 | { | |
19a39dce ID |
4073 | struct btrfs_ioctl_balance_args *bargs; |
4074 | int ret = 0; | |
4075 | ||
4076 | if (!capable(CAP_SYS_ADMIN)) | |
4077 | return -EPERM; | |
4078 | ||
4079 | mutex_lock(&fs_info->balance_mutex); | |
4080 | if (!fs_info->balance_ctl) { | |
4081 | ret = -ENOTCONN; | |
4082 | goto out; | |
4083 | } | |
4084 | ||
8d2db785 | 4085 | bargs = kzalloc(sizeof(*bargs), GFP_KERNEL); |
19a39dce ID |
4086 | if (!bargs) { |
4087 | ret = -ENOMEM; | |
4088 | goto out; | |
4089 | } | |
4090 | ||
008ef096 | 4091 | btrfs_update_ioctl_balance_args(fs_info, bargs); |
19a39dce ID |
4092 | |
4093 | if (copy_to_user(arg, bargs, sizeof(*bargs))) | |
4094 | ret = -EFAULT; | |
4095 | ||
4096 | kfree(bargs); | |
4097 | out: | |
4098 | mutex_unlock(&fs_info->balance_mutex); | |
4099 | return ret; | |
4100 | } | |
4101 | ||
905b0dda | 4102 | static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg) |
5d13a37b | 4103 | { |
0b246afa JM |
4104 | struct inode *inode = file_inode(file); |
4105 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
5d13a37b | 4106 | struct btrfs_ioctl_quota_ctl_args *sa; |
5d13a37b | 4107 | int ret; |
5d13a37b AJ |
4108 | |
4109 | if (!capable(CAP_SYS_ADMIN)) | |
4110 | return -EPERM; | |
4111 | ||
905b0dda MX |
4112 | ret = mnt_want_write_file(file); |
4113 | if (ret) | |
4114 | return ret; | |
5d13a37b AJ |
4115 | |
4116 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4117 | if (IS_ERR(sa)) { |
4118 | ret = PTR_ERR(sa); | |
4119 | goto drop_write; | |
4120 | } | |
5d13a37b | 4121 | |
0b246afa | 4122 | down_write(&fs_info->subvol_sem); |
5d13a37b AJ |
4123 | |
4124 | switch (sa->cmd) { | |
4125 | case BTRFS_QUOTA_CTL_ENABLE: | |
340f1aa2 | 4126 | ret = btrfs_quota_enable(fs_info); |
5d13a37b AJ |
4127 | break; |
4128 | case BTRFS_QUOTA_CTL_DISABLE: | |
340f1aa2 | 4129 | ret = btrfs_quota_disable(fs_info); |
5d13a37b | 4130 | break; |
5d13a37b AJ |
4131 | default: |
4132 | ret = -EINVAL; | |
4133 | break; | |
4134 | } | |
4135 | ||
5d13a37b | 4136 | kfree(sa); |
0b246afa | 4137 | up_write(&fs_info->subvol_sem); |
905b0dda MX |
4138 | drop_write: |
4139 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4140 | return ret; |
4141 | } | |
4142 | ||
905b0dda | 4143 | static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg) |
5d13a37b | 4144 | { |
0b246afa JM |
4145 | struct inode *inode = file_inode(file); |
4146 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4147 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
5d13a37b AJ |
4148 | struct btrfs_ioctl_qgroup_assign_args *sa; |
4149 | struct btrfs_trans_handle *trans; | |
4150 | int ret; | |
4151 | int err; | |
4152 | ||
4153 | if (!capable(CAP_SYS_ADMIN)) | |
4154 | return -EPERM; | |
4155 | ||
905b0dda MX |
4156 | ret = mnt_want_write_file(file); |
4157 | if (ret) | |
4158 | return ret; | |
5d13a37b AJ |
4159 | |
4160 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4161 | if (IS_ERR(sa)) { |
4162 | ret = PTR_ERR(sa); | |
4163 | goto drop_write; | |
4164 | } | |
5d13a37b AJ |
4165 | |
4166 | trans = btrfs_join_transaction(root); | |
4167 | if (IS_ERR(trans)) { | |
4168 | ret = PTR_ERR(trans); | |
4169 | goto out; | |
4170 | } | |
4171 | ||
5d13a37b | 4172 | if (sa->assign) { |
9f8a6ce6 | 4173 | ret = btrfs_add_qgroup_relation(trans, sa->src, sa->dst); |
5d13a37b | 4174 | } else { |
39616c27 | 4175 | ret = btrfs_del_qgroup_relation(trans, sa->src, sa->dst); |
5d13a37b AJ |
4176 | } |
4177 | ||
e082f563 | 4178 | /* update qgroup status and info */ |
280f8bd2 | 4179 | err = btrfs_run_qgroups(trans); |
e082f563 | 4180 | if (err < 0) |
0b246afa JM |
4181 | btrfs_handle_fs_error(fs_info, err, |
4182 | "failed to update qgroup status and info"); | |
3a45bb20 | 4183 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4184 | if (err && !ret) |
4185 | ret = err; | |
4186 | ||
4187 | out: | |
4188 | kfree(sa); | |
905b0dda MX |
4189 | drop_write: |
4190 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4191 | return ret; |
4192 | } | |
4193 | ||
905b0dda | 4194 | static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg) |
5d13a37b | 4195 | { |
0b246afa | 4196 | struct inode *inode = file_inode(file); |
0b246afa | 4197 | struct btrfs_root *root = BTRFS_I(inode)->root; |
5d13a37b AJ |
4198 | struct btrfs_ioctl_qgroup_create_args *sa; |
4199 | struct btrfs_trans_handle *trans; | |
4200 | int ret; | |
4201 | int err; | |
4202 | ||
4203 | if (!capable(CAP_SYS_ADMIN)) | |
4204 | return -EPERM; | |
4205 | ||
905b0dda MX |
4206 | ret = mnt_want_write_file(file); |
4207 | if (ret) | |
4208 | return ret; | |
5d13a37b AJ |
4209 | |
4210 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4211 | if (IS_ERR(sa)) { |
4212 | ret = PTR_ERR(sa); | |
4213 | goto drop_write; | |
4214 | } | |
5d13a37b | 4215 | |
d86e56cf MX |
4216 | if (!sa->qgroupid) { |
4217 | ret = -EINVAL; | |
4218 | goto out; | |
4219 | } | |
4220 | ||
5d13a37b AJ |
4221 | trans = btrfs_join_transaction(root); |
4222 | if (IS_ERR(trans)) { | |
4223 | ret = PTR_ERR(trans); | |
4224 | goto out; | |
4225 | } | |
4226 | ||
5d13a37b | 4227 | if (sa->create) { |
49a05ecd | 4228 | ret = btrfs_create_qgroup(trans, sa->qgroupid); |
5d13a37b | 4229 | } else { |
3efbee1d | 4230 | ret = btrfs_remove_qgroup(trans, sa->qgroupid); |
5d13a37b AJ |
4231 | } |
4232 | ||
3a45bb20 | 4233 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4234 | if (err && !ret) |
4235 | ret = err; | |
4236 | ||
4237 | out: | |
4238 | kfree(sa); | |
905b0dda MX |
4239 | drop_write: |
4240 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4241 | return ret; |
4242 | } | |
4243 | ||
905b0dda | 4244 | static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg) |
5d13a37b | 4245 | { |
0b246afa | 4246 | struct inode *inode = file_inode(file); |
0b246afa | 4247 | struct btrfs_root *root = BTRFS_I(inode)->root; |
5d13a37b AJ |
4248 | struct btrfs_ioctl_qgroup_limit_args *sa; |
4249 | struct btrfs_trans_handle *trans; | |
4250 | int ret; | |
4251 | int err; | |
4252 | u64 qgroupid; | |
4253 | ||
4254 | if (!capable(CAP_SYS_ADMIN)) | |
4255 | return -EPERM; | |
4256 | ||
905b0dda MX |
4257 | ret = mnt_want_write_file(file); |
4258 | if (ret) | |
4259 | return ret; | |
5d13a37b AJ |
4260 | |
4261 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4262 | if (IS_ERR(sa)) { |
4263 | ret = PTR_ERR(sa); | |
4264 | goto drop_write; | |
4265 | } | |
5d13a37b AJ |
4266 | |
4267 | trans = btrfs_join_transaction(root); | |
4268 | if (IS_ERR(trans)) { | |
4269 | ret = PTR_ERR(trans); | |
4270 | goto out; | |
4271 | } | |
4272 | ||
4273 | qgroupid = sa->qgroupid; | |
4274 | if (!qgroupid) { | |
4275 | /* take the current subvol as qgroup */ | |
4276 | qgroupid = root->root_key.objectid; | |
4277 | } | |
4278 | ||
f0042d5e | 4279 | ret = btrfs_limit_qgroup(trans, qgroupid, &sa->lim); |
5d13a37b | 4280 | |
3a45bb20 | 4281 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4282 | if (err && !ret) |
4283 | ret = err; | |
4284 | ||
4285 | out: | |
4286 | kfree(sa); | |
905b0dda MX |
4287 | drop_write: |
4288 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4289 | return ret; |
4290 | } | |
4291 | ||
2f232036 JS |
4292 | static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg) |
4293 | { | |
0b246afa JM |
4294 | struct inode *inode = file_inode(file); |
4295 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
2f232036 JS |
4296 | struct btrfs_ioctl_quota_rescan_args *qsa; |
4297 | int ret; | |
4298 | ||
4299 | if (!capable(CAP_SYS_ADMIN)) | |
4300 | return -EPERM; | |
4301 | ||
4302 | ret = mnt_want_write_file(file); | |
4303 | if (ret) | |
4304 | return ret; | |
4305 | ||
4306 | qsa = memdup_user(arg, sizeof(*qsa)); | |
4307 | if (IS_ERR(qsa)) { | |
4308 | ret = PTR_ERR(qsa); | |
4309 | goto drop_write; | |
4310 | } | |
4311 | ||
4312 | if (qsa->flags) { | |
4313 | ret = -EINVAL; | |
4314 | goto out; | |
4315 | } | |
4316 | ||
0b246afa | 4317 | ret = btrfs_qgroup_rescan(fs_info); |
2f232036 JS |
4318 | |
4319 | out: | |
4320 | kfree(qsa); | |
4321 | drop_write: | |
4322 | mnt_drop_write_file(file); | |
4323 | return ret; | |
4324 | } | |
4325 | ||
b929c1d8 MPS |
4326 | static long btrfs_ioctl_quota_rescan_status(struct btrfs_fs_info *fs_info, |
4327 | void __user *arg) | |
2f232036 | 4328 | { |
2f232036 JS |
4329 | struct btrfs_ioctl_quota_rescan_args *qsa; |
4330 | int ret = 0; | |
4331 | ||
4332 | if (!capable(CAP_SYS_ADMIN)) | |
4333 | return -EPERM; | |
4334 | ||
8d2db785 | 4335 | qsa = kzalloc(sizeof(*qsa), GFP_KERNEL); |
2f232036 JS |
4336 | if (!qsa) |
4337 | return -ENOMEM; | |
4338 | ||
0b246afa | 4339 | if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) { |
2f232036 | 4340 | qsa->flags = 1; |
0b246afa | 4341 | qsa->progress = fs_info->qgroup_rescan_progress.objectid; |
2f232036 JS |
4342 | } |
4343 | ||
4344 | if (copy_to_user(arg, qsa, sizeof(*qsa))) | |
4345 | ret = -EFAULT; | |
4346 | ||
4347 | kfree(qsa); | |
4348 | return ret; | |
4349 | } | |
4350 | ||
b929c1d8 MPS |
4351 | static long btrfs_ioctl_quota_rescan_wait(struct btrfs_fs_info *fs_info, |
4352 | void __user *arg) | |
57254b6e | 4353 | { |
57254b6e JS |
4354 | if (!capable(CAP_SYS_ADMIN)) |
4355 | return -EPERM; | |
4356 | ||
0b246afa | 4357 | return btrfs_qgroup_wait_for_completion(fs_info, true); |
57254b6e JS |
4358 | } |
4359 | ||
abccd00f HM |
4360 | static long _btrfs_ioctl_set_received_subvol(struct file *file, |
4361 | struct btrfs_ioctl_received_subvol_args *sa) | |
8ea05e3a | 4362 | { |
496ad9aa | 4363 | struct inode *inode = file_inode(file); |
0b246afa | 4364 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
8ea05e3a AB |
4365 | struct btrfs_root *root = BTRFS_I(inode)->root; |
4366 | struct btrfs_root_item *root_item = &root->root_item; | |
4367 | struct btrfs_trans_handle *trans; | |
95582b00 | 4368 | struct timespec64 ct = current_time(inode); |
8ea05e3a | 4369 | int ret = 0; |
dd5f9615 | 4370 | int received_uuid_changed; |
8ea05e3a | 4371 | |
21cb47be | 4372 | if (!inode_owner_or_capable(&init_user_ns, inode)) |
bd60ea0f DS |
4373 | return -EPERM; |
4374 | ||
8ea05e3a AB |
4375 | ret = mnt_want_write_file(file); |
4376 | if (ret < 0) | |
4377 | return ret; | |
4378 | ||
0b246afa | 4379 | down_write(&fs_info->subvol_sem); |
8ea05e3a | 4380 | |
4a0cc7ca | 4381 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
8ea05e3a AB |
4382 | ret = -EINVAL; |
4383 | goto out; | |
4384 | } | |
4385 | ||
4386 | if (btrfs_root_readonly(root)) { | |
4387 | ret = -EROFS; | |
4388 | goto out; | |
4389 | } | |
4390 | ||
dd5f9615 SB |
4391 | /* |
4392 | * 1 - root item | |
4393 | * 2 - uuid items (received uuid + subvol uuid) | |
4394 | */ | |
4395 | trans = btrfs_start_transaction(root, 3); | |
8ea05e3a AB |
4396 | if (IS_ERR(trans)) { |
4397 | ret = PTR_ERR(trans); | |
4398 | trans = NULL; | |
4399 | goto out; | |
4400 | } | |
4401 | ||
4402 | sa->rtransid = trans->transid; | |
4403 | sa->rtime.sec = ct.tv_sec; | |
4404 | sa->rtime.nsec = ct.tv_nsec; | |
4405 | ||
dd5f9615 SB |
4406 | received_uuid_changed = memcmp(root_item->received_uuid, sa->uuid, |
4407 | BTRFS_UUID_SIZE); | |
4408 | if (received_uuid_changed && | |
d87ff758 | 4409 | !btrfs_is_empty_uuid(root_item->received_uuid)) { |
d1957791 | 4410 | ret = btrfs_uuid_tree_remove(trans, root_item->received_uuid, |
d87ff758 NB |
4411 | BTRFS_UUID_KEY_RECEIVED_SUBVOL, |
4412 | root->root_key.objectid); | |
4413 | if (ret && ret != -ENOENT) { | |
4414 | btrfs_abort_transaction(trans, ret); | |
4415 | btrfs_end_transaction(trans); | |
4416 | goto out; | |
4417 | } | |
4418 | } | |
8ea05e3a AB |
4419 | memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE); |
4420 | btrfs_set_root_stransid(root_item, sa->stransid); | |
4421 | btrfs_set_root_rtransid(root_item, sa->rtransid); | |
3cae210f QW |
4422 | btrfs_set_stack_timespec_sec(&root_item->stime, sa->stime.sec); |
4423 | btrfs_set_stack_timespec_nsec(&root_item->stime, sa->stime.nsec); | |
4424 | btrfs_set_stack_timespec_sec(&root_item->rtime, sa->rtime.sec); | |
4425 | btrfs_set_stack_timespec_nsec(&root_item->rtime, sa->rtime.nsec); | |
8ea05e3a | 4426 | |
0b246afa | 4427 | ret = btrfs_update_root(trans, fs_info->tree_root, |
8ea05e3a AB |
4428 | &root->root_key, &root->root_item); |
4429 | if (ret < 0) { | |
3a45bb20 | 4430 | btrfs_end_transaction(trans); |
8ea05e3a | 4431 | goto out; |
dd5f9615 SB |
4432 | } |
4433 | if (received_uuid_changed && !btrfs_is_empty_uuid(sa->uuid)) { | |
cdb345a8 | 4434 | ret = btrfs_uuid_tree_add(trans, sa->uuid, |
dd5f9615 SB |
4435 | BTRFS_UUID_KEY_RECEIVED_SUBVOL, |
4436 | root->root_key.objectid); | |
4437 | if (ret < 0 && ret != -EEXIST) { | |
66642832 | 4438 | btrfs_abort_transaction(trans, ret); |
efd38150 | 4439 | btrfs_end_transaction(trans); |
8ea05e3a | 4440 | goto out; |
dd5f9615 SB |
4441 | } |
4442 | } | |
3a45bb20 | 4443 | ret = btrfs_commit_transaction(trans); |
abccd00f | 4444 | out: |
0b246afa | 4445 | up_write(&fs_info->subvol_sem); |
abccd00f HM |
4446 | mnt_drop_write_file(file); |
4447 | return ret; | |
4448 | } | |
4449 | ||
4450 | #ifdef CONFIG_64BIT | |
4451 | static long btrfs_ioctl_set_received_subvol_32(struct file *file, | |
4452 | void __user *arg) | |
4453 | { | |
4454 | struct btrfs_ioctl_received_subvol_args_32 *args32 = NULL; | |
4455 | struct btrfs_ioctl_received_subvol_args *args64 = NULL; | |
4456 | int ret = 0; | |
4457 | ||
4458 | args32 = memdup_user(arg, sizeof(*args32)); | |
7b9ea627 SV |
4459 | if (IS_ERR(args32)) |
4460 | return PTR_ERR(args32); | |
abccd00f | 4461 | |
8d2db785 | 4462 | args64 = kmalloc(sizeof(*args64), GFP_KERNEL); |
84dbeb87 DC |
4463 | if (!args64) { |
4464 | ret = -ENOMEM; | |
abccd00f HM |
4465 | goto out; |
4466 | } | |
4467 | ||
4468 | memcpy(args64->uuid, args32->uuid, BTRFS_UUID_SIZE); | |
4469 | args64->stransid = args32->stransid; | |
4470 | args64->rtransid = args32->rtransid; | |
4471 | args64->stime.sec = args32->stime.sec; | |
4472 | args64->stime.nsec = args32->stime.nsec; | |
4473 | args64->rtime.sec = args32->rtime.sec; | |
4474 | args64->rtime.nsec = args32->rtime.nsec; | |
4475 | args64->flags = args32->flags; | |
4476 | ||
4477 | ret = _btrfs_ioctl_set_received_subvol(file, args64); | |
4478 | if (ret) | |
4479 | goto out; | |
4480 | ||
4481 | memcpy(args32->uuid, args64->uuid, BTRFS_UUID_SIZE); | |
4482 | args32->stransid = args64->stransid; | |
4483 | args32->rtransid = args64->rtransid; | |
4484 | args32->stime.sec = args64->stime.sec; | |
4485 | args32->stime.nsec = args64->stime.nsec; | |
4486 | args32->rtime.sec = args64->rtime.sec; | |
4487 | args32->rtime.nsec = args64->rtime.nsec; | |
4488 | args32->flags = args64->flags; | |
4489 | ||
4490 | ret = copy_to_user(arg, args32, sizeof(*args32)); | |
4491 | if (ret) | |
4492 | ret = -EFAULT; | |
4493 | ||
4494 | out: | |
4495 | kfree(args32); | |
4496 | kfree(args64); | |
4497 | return ret; | |
4498 | } | |
4499 | #endif | |
4500 | ||
4501 | static long btrfs_ioctl_set_received_subvol(struct file *file, | |
4502 | void __user *arg) | |
4503 | { | |
4504 | struct btrfs_ioctl_received_subvol_args *sa = NULL; | |
4505 | int ret = 0; | |
4506 | ||
4507 | sa = memdup_user(arg, sizeof(*sa)); | |
7b9ea627 SV |
4508 | if (IS_ERR(sa)) |
4509 | return PTR_ERR(sa); | |
abccd00f HM |
4510 | |
4511 | ret = _btrfs_ioctl_set_received_subvol(file, sa); | |
4512 | ||
4513 | if (ret) | |
4514 | goto out; | |
4515 | ||
8ea05e3a AB |
4516 | ret = copy_to_user(arg, sa, sizeof(*sa)); |
4517 | if (ret) | |
4518 | ret = -EFAULT; | |
4519 | ||
4520 | out: | |
4521 | kfree(sa); | |
8ea05e3a AB |
4522 | return ret; |
4523 | } | |
4524 | ||
b929c1d8 MPS |
4525 | static int btrfs_ioctl_get_fslabel(struct btrfs_fs_info *fs_info, |
4526 | void __user *arg) | |
867ab667 | 4527 | { |
a1b83ac5 | 4528 | size_t len; |
867ab667 | 4529 | int ret; |
a1b83ac5 AJ |
4530 | char label[BTRFS_LABEL_SIZE]; |
4531 | ||
0b246afa JM |
4532 | spin_lock(&fs_info->super_lock); |
4533 | memcpy(label, fs_info->super_copy->label, BTRFS_LABEL_SIZE); | |
4534 | spin_unlock(&fs_info->super_lock); | |
a1b83ac5 AJ |
4535 | |
4536 | len = strnlen(label, BTRFS_LABEL_SIZE); | |
867ab667 | 4537 | |
4538 | if (len == BTRFS_LABEL_SIZE) { | |
0b246afa JM |
4539 | btrfs_warn(fs_info, |
4540 | "label is too long, return the first %zu bytes", | |
4541 | --len); | |
867ab667 | 4542 | } |
4543 | ||
867ab667 | 4544 | ret = copy_to_user(arg, label, len); |
867ab667 | 4545 | |
4546 | return ret ? -EFAULT : 0; | |
4547 | } | |
4548 | ||
a8bfd4ab | 4549 | static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg) |
4550 | { | |
0b246afa JM |
4551 | struct inode *inode = file_inode(file); |
4552 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4553 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4554 | struct btrfs_super_block *super_block = fs_info->super_copy; | |
a8bfd4ab | 4555 | struct btrfs_trans_handle *trans; |
4556 | char label[BTRFS_LABEL_SIZE]; | |
4557 | int ret; | |
4558 | ||
4559 | if (!capable(CAP_SYS_ADMIN)) | |
4560 | return -EPERM; | |
4561 | ||
4562 | if (copy_from_user(label, arg, sizeof(label))) | |
4563 | return -EFAULT; | |
4564 | ||
4565 | if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) { | |
0b246afa | 4566 | btrfs_err(fs_info, |
5d163e0e JM |
4567 | "unable to set label with more than %d bytes", |
4568 | BTRFS_LABEL_SIZE - 1); | |
a8bfd4ab | 4569 | return -EINVAL; |
4570 | } | |
4571 | ||
4572 | ret = mnt_want_write_file(file); | |
4573 | if (ret) | |
4574 | return ret; | |
4575 | ||
a8bfd4ab | 4576 | trans = btrfs_start_transaction(root, 0); |
4577 | if (IS_ERR(trans)) { | |
4578 | ret = PTR_ERR(trans); | |
4579 | goto out_unlock; | |
4580 | } | |
4581 | ||
0b246afa | 4582 | spin_lock(&fs_info->super_lock); |
a8bfd4ab | 4583 | strcpy(super_block->label, label); |
0b246afa | 4584 | spin_unlock(&fs_info->super_lock); |
3a45bb20 | 4585 | ret = btrfs_commit_transaction(trans); |
a8bfd4ab | 4586 | |
4587 | out_unlock: | |
a8bfd4ab | 4588 | mnt_drop_write_file(file); |
4589 | return ret; | |
4590 | } | |
4591 | ||
2eaa055f JM |
4592 | #define INIT_FEATURE_FLAGS(suffix) \ |
4593 | { .compat_flags = BTRFS_FEATURE_COMPAT_##suffix, \ | |
4594 | .compat_ro_flags = BTRFS_FEATURE_COMPAT_RO_##suffix, \ | |
4595 | .incompat_flags = BTRFS_FEATURE_INCOMPAT_##suffix } | |
4596 | ||
d5131b65 | 4597 | int btrfs_ioctl_get_supported_features(void __user *arg) |
2eaa055f | 4598 | { |
4d4ab6d6 | 4599 | static const struct btrfs_ioctl_feature_flags features[3] = { |
2eaa055f JM |
4600 | INIT_FEATURE_FLAGS(SUPP), |
4601 | INIT_FEATURE_FLAGS(SAFE_SET), | |
4602 | INIT_FEATURE_FLAGS(SAFE_CLEAR) | |
4603 | }; | |
4604 | ||
4605 | if (copy_to_user(arg, &features, sizeof(features))) | |
4606 | return -EFAULT; | |
4607 | ||
4608 | return 0; | |
4609 | } | |
4610 | ||
b929c1d8 MPS |
4611 | static int btrfs_ioctl_get_features(struct btrfs_fs_info *fs_info, |
4612 | void __user *arg) | |
2eaa055f | 4613 | { |
0b246afa | 4614 | struct btrfs_super_block *super_block = fs_info->super_copy; |
2eaa055f JM |
4615 | struct btrfs_ioctl_feature_flags features; |
4616 | ||
4617 | features.compat_flags = btrfs_super_compat_flags(super_block); | |
4618 | features.compat_ro_flags = btrfs_super_compat_ro_flags(super_block); | |
4619 | features.incompat_flags = btrfs_super_incompat_flags(super_block); | |
4620 | ||
4621 | if (copy_to_user(arg, &features, sizeof(features))) | |
4622 | return -EFAULT; | |
4623 | ||
4624 | return 0; | |
4625 | } | |
4626 | ||
2ff7e61e | 4627 | static int check_feature_bits(struct btrfs_fs_info *fs_info, |
3b02a68a | 4628 | enum btrfs_feature_set set, |
2eaa055f JM |
4629 | u64 change_mask, u64 flags, u64 supported_flags, |
4630 | u64 safe_set, u64 safe_clear) | |
4631 | { | |
f10152bc | 4632 | const char *type = btrfs_feature_set_name(set); |
3b02a68a | 4633 | char *names; |
2eaa055f JM |
4634 | u64 disallowed, unsupported; |
4635 | u64 set_mask = flags & change_mask; | |
4636 | u64 clear_mask = ~flags & change_mask; | |
4637 | ||
4638 | unsupported = set_mask & ~supported_flags; | |
4639 | if (unsupported) { | |
3b02a68a JM |
4640 | names = btrfs_printable_features(set, unsupported); |
4641 | if (names) { | |
0b246afa JM |
4642 | btrfs_warn(fs_info, |
4643 | "this kernel does not support the %s feature bit%s", | |
4644 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4645 | kfree(names); |
4646 | } else | |
0b246afa JM |
4647 | btrfs_warn(fs_info, |
4648 | "this kernel does not support %s bits 0x%llx", | |
4649 | type, unsupported); | |
2eaa055f JM |
4650 | return -EOPNOTSUPP; |
4651 | } | |
4652 | ||
4653 | disallowed = set_mask & ~safe_set; | |
4654 | if (disallowed) { | |
3b02a68a JM |
4655 | names = btrfs_printable_features(set, disallowed); |
4656 | if (names) { | |
0b246afa JM |
4657 | btrfs_warn(fs_info, |
4658 | "can't set the %s feature bit%s while mounted", | |
4659 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4660 | kfree(names); |
4661 | } else | |
0b246afa JM |
4662 | btrfs_warn(fs_info, |
4663 | "can't set %s bits 0x%llx while mounted", | |
4664 | type, disallowed); | |
2eaa055f JM |
4665 | return -EPERM; |
4666 | } | |
4667 | ||
4668 | disallowed = clear_mask & ~safe_clear; | |
4669 | if (disallowed) { | |
3b02a68a JM |
4670 | names = btrfs_printable_features(set, disallowed); |
4671 | if (names) { | |
0b246afa JM |
4672 | btrfs_warn(fs_info, |
4673 | "can't clear the %s feature bit%s while mounted", | |
4674 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4675 | kfree(names); |
4676 | } else | |
0b246afa JM |
4677 | btrfs_warn(fs_info, |
4678 | "can't clear %s bits 0x%llx while mounted", | |
4679 | type, disallowed); | |
2eaa055f JM |
4680 | return -EPERM; |
4681 | } | |
4682 | ||
4683 | return 0; | |
4684 | } | |
4685 | ||
2ff7e61e JM |
4686 | #define check_feature(fs_info, change_mask, flags, mask_base) \ |
4687 | check_feature_bits(fs_info, FEAT_##mask_base, change_mask, flags, \ | |
2eaa055f JM |
4688 | BTRFS_FEATURE_ ## mask_base ## _SUPP, \ |
4689 | BTRFS_FEATURE_ ## mask_base ## _SAFE_SET, \ | |
4690 | BTRFS_FEATURE_ ## mask_base ## _SAFE_CLEAR) | |
4691 | ||
4692 | static int btrfs_ioctl_set_features(struct file *file, void __user *arg) | |
4693 | { | |
0b246afa JM |
4694 | struct inode *inode = file_inode(file); |
4695 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4696 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4697 | struct btrfs_super_block *super_block = fs_info->super_copy; | |
2eaa055f JM |
4698 | struct btrfs_ioctl_feature_flags flags[2]; |
4699 | struct btrfs_trans_handle *trans; | |
4700 | u64 newflags; | |
4701 | int ret; | |
4702 | ||
4703 | if (!capable(CAP_SYS_ADMIN)) | |
4704 | return -EPERM; | |
4705 | ||
4706 | if (copy_from_user(flags, arg, sizeof(flags))) | |
4707 | return -EFAULT; | |
4708 | ||
4709 | /* Nothing to do */ | |
4710 | if (!flags[0].compat_flags && !flags[0].compat_ro_flags && | |
4711 | !flags[0].incompat_flags) | |
4712 | return 0; | |
4713 | ||
2ff7e61e | 4714 | ret = check_feature(fs_info, flags[0].compat_flags, |
2eaa055f JM |
4715 | flags[1].compat_flags, COMPAT); |
4716 | if (ret) | |
4717 | return ret; | |
4718 | ||
2ff7e61e | 4719 | ret = check_feature(fs_info, flags[0].compat_ro_flags, |
2eaa055f JM |
4720 | flags[1].compat_ro_flags, COMPAT_RO); |
4721 | if (ret) | |
4722 | return ret; | |
4723 | ||
2ff7e61e | 4724 | ret = check_feature(fs_info, flags[0].incompat_flags, |
2eaa055f JM |
4725 | flags[1].incompat_flags, INCOMPAT); |
4726 | if (ret) | |
4727 | return ret; | |
4728 | ||
7ab19625 DS |
4729 | ret = mnt_want_write_file(file); |
4730 | if (ret) | |
4731 | return ret; | |
4732 | ||
8051aa1a | 4733 | trans = btrfs_start_transaction(root, 0); |
7ab19625 DS |
4734 | if (IS_ERR(trans)) { |
4735 | ret = PTR_ERR(trans); | |
4736 | goto out_drop_write; | |
4737 | } | |
2eaa055f | 4738 | |
0b246afa | 4739 | spin_lock(&fs_info->super_lock); |
2eaa055f JM |
4740 | newflags = btrfs_super_compat_flags(super_block); |
4741 | newflags |= flags[0].compat_flags & flags[1].compat_flags; | |
4742 | newflags &= ~(flags[0].compat_flags & ~flags[1].compat_flags); | |
4743 | btrfs_set_super_compat_flags(super_block, newflags); | |
4744 | ||
4745 | newflags = btrfs_super_compat_ro_flags(super_block); | |
4746 | newflags |= flags[0].compat_ro_flags & flags[1].compat_ro_flags; | |
4747 | newflags &= ~(flags[0].compat_ro_flags & ~flags[1].compat_ro_flags); | |
4748 | btrfs_set_super_compat_ro_flags(super_block, newflags); | |
4749 | ||
4750 | newflags = btrfs_super_incompat_flags(super_block); | |
4751 | newflags |= flags[0].incompat_flags & flags[1].incompat_flags; | |
4752 | newflags &= ~(flags[0].incompat_flags & ~flags[1].incompat_flags); | |
4753 | btrfs_set_super_incompat_flags(super_block, newflags); | |
0b246afa | 4754 | spin_unlock(&fs_info->super_lock); |
2eaa055f | 4755 | |
3a45bb20 | 4756 | ret = btrfs_commit_transaction(trans); |
7ab19625 DS |
4757 | out_drop_write: |
4758 | mnt_drop_write_file(file); | |
4759 | ||
4760 | return ret; | |
2eaa055f JM |
4761 | } |
4762 | ||
2351f431 JB |
4763 | static int _btrfs_ioctl_send(struct file *file, void __user *argp, bool compat) |
4764 | { | |
4765 | struct btrfs_ioctl_send_args *arg; | |
4766 | int ret; | |
4767 | ||
4768 | if (compat) { | |
4769 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) | |
4770 | struct btrfs_ioctl_send_args_32 args32; | |
4771 | ||
4772 | ret = copy_from_user(&args32, argp, sizeof(args32)); | |
4773 | if (ret) | |
4774 | return -EFAULT; | |
4775 | arg = kzalloc(sizeof(*arg), GFP_KERNEL); | |
4776 | if (!arg) | |
4777 | return -ENOMEM; | |
4778 | arg->send_fd = args32.send_fd; | |
4779 | arg->clone_sources_count = args32.clone_sources_count; | |
4780 | arg->clone_sources = compat_ptr(args32.clone_sources); | |
4781 | arg->parent_root = args32.parent_root; | |
4782 | arg->flags = args32.flags; | |
4783 | memcpy(arg->reserved, args32.reserved, | |
4784 | sizeof(args32.reserved)); | |
4785 | #else | |
4786 | return -ENOTTY; | |
4787 | #endif | |
4788 | } else { | |
4789 | arg = memdup_user(argp, sizeof(*arg)); | |
4790 | if (IS_ERR(arg)) | |
4791 | return PTR_ERR(arg); | |
4792 | } | |
4793 | ret = btrfs_ioctl_send(file, arg); | |
4794 | kfree(arg); | |
4795 | return ret; | |
4796 | } | |
4797 | ||
f46b5a66 CH |
4798 | long btrfs_ioctl(struct file *file, unsigned int |
4799 | cmd, unsigned long arg) | |
4800 | { | |
0b246afa JM |
4801 | struct inode *inode = file_inode(file); |
4802 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4803 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4bcabaa3 | 4804 | void __user *argp = (void __user *)arg; |
f46b5a66 CH |
4805 | |
4806 | switch (cmd) { | |
6cbff00f CH |
4807 | case FS_IOC_GETVERSION: |
4808 | return btrfs_ioctl_getversion(file, argp); | |
40cf931f | 4809 | case FS_IOC_GETFSLABEL: |
b929c1d8 | 4810 | return btrfs_ioctl_get_fslabel(fs_info, argp); |
40cf931f ES |
4811 | case FS_IOC_SETFSLABEL: |
4812 | return btrfs_ioctl_set_fslabel(file, argp); | |
f7039b1d | 4813 | case FITRIM: |
b929c1d8 | 4814 | return btrfs_ioctl_fitrim(fs_info, argp); |
f46b5a66 | 4815 | case BTRFS_IOC_SNAP_CREATE: |
fa0d2b9b | 4816 | return btrfs_ioctl_snap_create(file, argp, 0); |
fdfb1e4f | 4817 | case BTRFS_IOC_SNAP_CREATE_V2: |
fa0d2b9b | 4818 | return btrfs_ioctl_snap_create_v2(file, argp, 0); |
3de4586c | 4819 | case BTRFS_IOC_SUBVOL_CREATE: |
fa0d2b9b | 4820 | return btrfs_ioctl_snap_create(file, argp, 1); |
6f72c7e2 AJ |
4821 | case BTRFS_IOC_SUBVOL_CREATE_V2: |
4822 | return btrfs_ioctl_snap_create_v2(file, argp, 1); | |
76dda93c | 4823 | case BTRFS_IOC_SNAP_DESTROY: |
949964c9 MPS |
4824 | return btrfs_ioctl_snap_destroy(file, argp, false); |
4825 | case BTRFS_IOC_SNAP_DESTROY_V2: | |
4826 | return btrfs_ioctl_snap_destroy(file, argp, true); | |
0caa102d LZ |
4827 | case BTRFS_IOC_SUBVOL_GETFLAGS: |
4828 | return btrfs_ioctl_subvol_getflags(file, argp); | |
4829 | case BTRFS_IOC_SUBVOL_SETFLAGS: | |
4830 | return btrfs_ioctl_subvol_setflags(file, argp); | |
6ef5ed0d JB |
4831 | case BTRFS_IOC_DEFAULT_SUBVOL: |
4832 | return btrfs_ioctl_default_subvol(file, argp); | |
f46b5a66 | 4833 | case BTRFS_IOC_DEFRAG: |
1e701a32 CM |
4834 | return btrfs_ioctl_defrag(file, NULL); |
4835 | case BTRFS_IOC_DEFRAG_RANGE: | |
4836 | return btrfs_ioctl_defrag(file, argp); | |
f46b5a66 | 4837 | case BTRFS_IOC_RESIZE: |
198605a8 | 4838 | return btrfs_ioctl_resize(file, argp); |
f46b5a66 | 4839 | case BTRFS_IOC_ADD_DEV: |
2ff7e61e | 4840 | return btrfs_ioctl_add_dev(fs_info, argp); |
f46b5a66 | 4841 | case BTRFS_IOC_RM_DEV: |
da24927b | 4842 | return btrfs_ioctl_rm_dev(file, argp); |
6b526ed7 AJ |
4843 | case BTRFS_IOC_RM_DEV_V2: |
4844 | return btrfs_ioctl_rm_dev_v2(file, argp); | |
475f6387 | 4845 | case BTRFS_IOC_FS_INFO: |
2ff7e61e | 4846 | return btrfs_ioctl_fs_info(fs_info, argp); |
475f6387 | 4847 | case BTRFS_IOC_DEV_INFO: |
2ff7e61e | 4848 | return btrfs_ioctl_dev_info(fs_info, argp); |
f46b5a66 | 4849 | case BTRFS_IOC_BALANCE: |
9ba1f6e4 | 4850 | return btrfs_ioctl_balance(file, NULL); |
ac8e9819 CM |
4851 | case BTRFS_IOC_TREE_SEARCH: |
4852 | return btrfs_ioctl_tree_search(file, argp); | |
cc68a8a5 GH |
4853 | case BTRFS_IOC_TREE_SEARCH_V2: |
4854 | return btrfs_ioctl_tree_search_v2(file, argp); | |
ac8e9819 CM |
4855 | case BTRFS_IOC_INO_LOOKUP: |
4856 | return btrfs_ioctl_ino_lookup(file, argp); | |
d7728c96 JS |
4857 | case BTRFS_IOC_INO_PATHS: |
4858 | return btrfs_ioctl_ino_to_path(root, argp); | |
4859 | case BTRFS_IOC_LOGICAL_INO: | |
d24a67b2 ZB |
4860 | return btrfs_ioctl_logical_to_ino(fs_info, argp, 1); |
4861 | case BTRFS_IOC_LOGICAL_INO_V2: | |
4862 | return btrfs_ioctl_logical_to_ino(fs_info, argp, 2); | |
1406e432 | 4863 | case BTRFS_IOC_SPACE_INFO: |
2ff7e61e | 4864 | return btrfs_ioctl_space_info(fs_info, argp); |
9b199859 FDBM |
4865 | case BTRFS_IOC_SYNC: { |
4866 | int ret; | |
4867 | ||
9db4dc24 | 4868 | ret = btrfs_start_delalloc_roots(fs_info, LONG_MAX, false); |
9b199859 FDBM |
4869 | if (ret) |
4870 | return ret; | |
0b246afa | 4871 | ret = btrfs_sync_fs(inode->i_sb, 1); |
2fad4e83 DS |
4872 | /* |
4873 | * The transaction thread may want to do more work, | |
01327610 | 4874 | * namely it pokes the cleaner kthread that will start |
2fad4e83 DS |
4875 | * processing uncleaned subvols. |
4876 | */ | |
0b246afa | 4877 | wake_up_process(fs_info->transaction_kthread); |
9b199859 FDBM |
4878 | return ret; |
4879 | } | |
46204592 | 4880 | case BTRFS_IOC_START_SYNC: |
9a8c28be | 4881 | return btrfs_ioctl_start_sync(root, argp); |
46204592 | 4882 | case BTRFS_IOC_WAIT_SYNC: |
2ff7e61e | 4883 | return btrfs_ioctl_wait_sync(fs_info, argp); |
475f6387 | 4884 | case BTRFS_IOC_SCRUB: |
b8e95489 | 4885 | return btrfs_ioctl_scrub(file, argp); |
475f6387 | 4886 | case BTRFS_IOC_SCRUB_CANCEL: |
2ff7e61e | 4887 | return btrfs_ioctl_scrub_cancel(fs_info); |
475f6387 | 4888 | case BTRFS_IOC_SCRUB_PROGRESS: |
2ff7e61e | 4889 | return btrfs_ioctl_scrub_progress(fs_info, argp); |
c9e9f97b | 4890 | case BTRFS_IOC_BALANCE_V2: |
9ba1f6e4 | 4891 | return btrfs_ioctl_balance(file, argp); |
837d5b6e | 4892 | case BTRFS_IOC_BALANCE_CTL: |
2ff7e61e | 4893 | return btrfs_ioctl_balance_ctl(fs_info, arg); |
19a39dce | 4894 | case BTRFS_IOC_BALANCE_PROGRESS: |
2ff7e61e | 4895 | return btrfs_ioctl_balance_progress(fs_info, argp); |
8ea05e3a AB |
4896 | case BTRFS_IOC_SET_RECEIVED_SUBVOL: |
4897 | return btrfs_ioctl_set_received_subvol(file, argp); | |
abccd00f HM |
4898 | #ifdef CONFIG_64BIT |
4899 | case BTRFS_IOC_SET_RECEIVED_SUBVOL_32: | |
4900 | return btrfs_ioctl_set_received_subvol_32(file, argp); | |
4901 | #endif | |
31db9f7c | 4902 | case BTRFS_IOC_SEND: |
2351f431 JB |
4903 | return _btrfs_ioctl_send(file, argp, false); |
4904 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) | |
4905 | case BTRFS_IOC_SEND_32: | |
4906 | return _btrfs_ioctl_send(file, argp, true); | |
4907 | #endif | |
c11d2c23 | 4908 | case BTRFS_IOC_GET_DEV_STATS: |
2ff7e61e | 4909 | return btrfs_ioctl_get_dev_stats(fs_info, argp); |
5d13a37b | 4910 | case BTRFS_IOC_QUOTA_CTL: |
905b0dda | 4911 | return btrfs_ioctl_quota_ctl(file, argp); |
5d13a37b | 4912 | case BTRFS_IOC_QGROUP_ASSIGN: |
905b0dda | 4913 | return btrfs_ioctl_qgroup_assign(file, argp); |
5d13a37b | 4914 | case BTRFS_IOC_QGROUP_CREATE: |
905b0dda | 4915 | return btrfs_ioctl_qgroup_create(file, argp); |
5d13a37b | 4916 | case BTRFS_IOC_QGROUP_LIMIT: |
905b0dda | 4917 | return btrfs_ioctl_qgroup_limit(file, argp); |
2f232036 JS |
4918 | case BTRFS_IOC_QUOTA_RESCAN: |
4919 | return btrfs_ioctl_quota_rescan(file, argp); | |
4920 | case BTRFS_IOC_QUOTA_RESCAN_STATUS: | |
b929c1d8 | 4921 | return btrfs_ioctl_quota_rescan_status(fs_info, argp); |
57254b6e | 4922 | case BTRFS_IOC_QUOTA_RESCAN_WAIT: |
b929c1d8 | 4923 | return btrfs_ioctl_quota_rescan_wait(fs_info, argp); |
3f6bcfbd | 4924 | case BTRFS_IOC_DEV_REPLACE: |
2ff7e61e | 4925 | return btrfs_ioctl_dev_replace(fs_info, argp); |
2eaa055f | 4926 | case BTRFS_IOC_GET_SUPPORTED_FEATURES: |
d5131b65 | 4927 | return btrfs_ioctl_get_supported_features(argp); |
2eaa055f | 4928 | case BTRFS_IOC_GET_FEATURES: |
b929c1d8 | 4929 | return btrfs_ioctl_get_features(fs_info, argp); |
2eaa055f JM |
4930 | case BTRFS_IOC_SET_FEATURES: |
4931 | return btrfs_ioctl_set_features(file, argp); | |
b64ec075 TM |
4932 | case BTRFS_IOC_GET_SUBVOL_INFO: |
4933 | return btrfs_ioctl_get_subvol_info(file, argp); | |
42e4b520 TM |
4934 | case BTRFS_IOC_GET_SUBVOL_ROOTREF: |
4935 | return btrfs_ioctl_get_subvol_rootref(file, argp); | |
23d0b79d TM |
4936 | case BTRFS_IOC_INO_LOOKUP_USER: |
4937 | return btrfs_ioctl_ino_lookup_user(file, argp); | |
f46b5a66 CH |
4938 | } |
4939 | ||
4940 | return -ENOTTY; | |
4941 | } | |
4c63c245 LD |
4942 | |
4943 | #ifdef CONFIG_COMPAT | |
4944 | long btrfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg) | |
4945 | { | |
2a362249 JM |
4946 | /* |
4947 | * These all access 32-bit values anyway so no further | |
4948 | * handling is necessary. | |
4949 | */ | |
4c63c245 | 4950 | switch (cmd) { |
4c63c245 LD |
4951 | case FS_IOC32_GETVERSION: |
4952 | cmd = FS_IOC_GETVERSION; | |
4953 | break; | |
4c63c245 LD |
4954 | } |
4955 | ||
4956 | return btrfs_ioctl(file, cmd, (unsigned long) compat_ptr(arg)); | |
4957 | } | |
4958 | #endif |