]>
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 | ||
578bda9e DS |
374 | /* |
375 | * Conditionally allow to enter the exclusive operation in case it's compatible | |
376 | * with the running one. This must be paired with btrfs_exclop_start_unlock and | |
377 | * btrfs_exclop_finish. | |
378 | * | |
379 | * Compatibility: | |
380 | * - the same type is already running | |
381 | * - not BTRFS_EXCLOP_NONE - this is intentionally incompatible and the caller | |
382 | * must check the condition first that would allow none -> @type | |
383 | */ | |
384 | bool btrfs_exclop_start_try_lock(struct btrfs_fs_info *fs_info, | |
385 | enum btrfs_exclusive_operation type) | |
386 | { | |
387 | spin_lock(&fs_info->super_lock); | |
388 | if (fs_info->exclusive_operation == type) | |
389 | return true; | |
390 | ||
391 | spin_unlock(&fs_info->super_lock); | |
392 | return false; | |
393 | } | |
394 | ||
395 | void btrfs_exclop_start_unlock(struct btrfs_fs_info *fs_info) | |
396 | { | |
397 | spin_unlock(&fs_info->super_lock); | |
398 | } | |
399 | ||
c3e1f96c GR |
400 | void btrfs_exclop_finish(struct btrfs_fs_info *fs_info) |
401 | { | |
0d7ed32c | 402 | spin_lock(&fs_info->super_lock); |
c3e1f96c | 403 | WRITE_ONCE(fs_info->exclusive_operation, BTRFS_EXCLOP_NONE); |
0d7ed32c | 404 | spin_unlock(&fs_info->super_lock); |
66a2823c | 405 | sysfs_notify(&fs_info->fs_devices->fsid_kobj, NULL, "exclusive_operation"); |
c3e1f96c GR |
406 | } |
407 | ||
6cbff00f CH |
408 | static int btrfs_ioctl_getversion(struct file *file, int __user *arg) |
409 | { | |
496ad9aa | 410 | struct inode *inode = file_inode(file); |
6cbff00f CH |
411 | |
412 | return put_user(inode->i_generation, arg); | |
413 | } | |
f46b5a66 | 414 | |
b929c1d8 MPS |
415 | static noinline int btrfs_ioctl_fitrim(struct btrfs_fs_info *fs_info, |
416 | void __user *arg) | |
f7039b1d | 417 | { |
f7039b1d LD |
418 | struct btrfs_device *device; |
419 | struct request_queue *q; | |
420 | struct fstrim_range range; | |
421 | u64 minlen = ULLONG_MAX; | |
422 | u64 num_devices = 0; | |
423 | int ret; | |
424 | ||
425 | if (!capable(CAP_SYS_ADMIN)) | |
426 | return -EPERM; | |
427 | ||
1cb3dc3f NA |
428 | /* |
429 | * btrfs_trim_block_group() depends on space cache, which is not | |
430 | * available in zoned filesystem. So, disallow fitrim on a zoned | |
431 | * filesystem for now. | |
432 | */ | |
433 | if (btrfs_is_zoned(fs_info)) | |
434 | return -EOPNOTSUPP; | |
435 | ||
f35f06c3 FM |
436 | /* |
437 | * If the fs is mounted with nologreplay, which requires it to be | |
438 | * mounted in RO mode as well, we can not allow discard on free space | |
439 | * inside block groups, because log trees refer to extents that are not | |
440 | * pinned in a block group's free space cache (pinning the extents is | |
441 | * precisely the first phase of replaying a log tree). | |
442 | */ | |
443 | if (btrfs_test_opt(fs_info, NOLOGREPLAY)) | |
444 | return -EROFS; | |
445 | ||
1f78160c XG |
446 | rcu_read_lock(); |
447 | list_for_each_entry_rcu(device, &fs_info->fs_devices->devices, | |
448 | dev_list) { | |
f7039b1d LD |
449 | if (!device->bdev) |
450 | continue; | |
451 | q = bdev_get_queue(device->bdev); | |
452 | if (blk_queue_discard(q)) { | |
453 | num_devices++; | |
50d0446e | 454 | minlen = min_t(u64, q->limits.discard_granularity, |
f7039b1d LD |
455 | minlen); |
456 | } | |
457 | } | |
1f78160c | 458 | rcu_read_unlock(); |
f4c697e6 | 459 | |
f7039b1d LD |
460 | if (!num_devices) |
461 | return -EOPNOTSUPP; | |
f7039b1d LD |
462 | if (copy_from_user(&range, arg, sizeof(range))) |
463 | return -EFAULT; | |
6ba9fc8e QW |
464 | |
465 | /* | |
466 | * NOTE: Don't truncate the range using super->total_bytes. Bytenr of | |
467 | * block group is in the logical address space, which can be any | |
468 | * sectorsize aligned bytenr in the range [0, U64_MAX]. | |
469 | */ | |
470 | if (range.len < fs_info->sb->s_blocksize) | |
f4c697e6 | 471 | return -EINVAL; |
f7039b1d LD |
472 | |
473 | range.minlen = max(range.minlen, minlen); | |
2ff7e61e | 474 | ret = btrfs_trim_fs(fs_info, &range); |
f7039b1d LD |
475 | if (ret < 0) |
476 | return ret; | |
477 | ||
478 | if (copy_to_user(arg, &range, sizeof(range))) | |
479 | return -EFAULT; | |
480 | ||
481 | return 0; | |
482 | } | |
483 | ||
e1f60a65 | 484 | int __pure btrfs_is_empty_uuid(u8 *uuid) |
dd5f9615 | 485 | { |
46e0f66a CM |
486 | int i; |
487 | ||
488 | for (i = 0; i < BTRFS_UUID_SIZE; i++) { | |
489 | if (uuid[i]) | |
490 | return 0; | |
491 | } | |
492 | return 1; | |
dd5f9615 SB |
493 | } |
494 | ||
d5c12070 | 495 | static noinline int create_subvol(struct inode *dir, |
cb8e7090 | 496 | struct dentry *dentry, |
52f75f4f | 497 | const char *name, int namelen, |
8696c533 | 498 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 499 | { |
0b246afa | 500 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); |
f46b5a66 CH |
501 | struct btrfs_trans_handle *trans; |
502 | struct btrfs_key key; | |
49a3c4d9 | 503 | struct btrfs_root_item *root_item; |
f46b5a66 CH |
504 | struct btrfs_inode_item *inode_item; |
505 | struct extent_buffer *leaf; | |
d5c12070 | 506 | struct btrfs_root *root = BTRFS_I(dir)->root; |
76dda93c | 507 | struct btrfs_root *new_root; |
d5c12070 | 508 | struct btrfs_block_rsv block_rsv; |
95582b00 | 509 | struct timespec64 cur_time = current_time(dir); |
5662344b | 510 | struct inode *inode; |
f46b5a66 CH |
511 | int ret; |
512 | int err; | |
2dfb1e43 | 513 | dev_t anon_dev = 0; |
f46b5a66 | 514 | u64 objectid; |
3de4586c | 515 | u64 index = 0; |
f46b5a66 | 516 | |
49a3c4d9 DS |
517 | root_item = kzalloc(sizeof(*root_item), GFP_KERNEL); |
518 | if (!root_item) | |
519 | return -ENOMEM; | |
520 | ||
543068a2 | 521 | ret = btrfs_get_free_objectid(fs_info->tree_root, &objectid); |
2fbe8c8a | 522 | if (ret) |
49a3c4d9 | 523 | goto fail_free; |
6a912213 | 524 | |
2dfb1e43 QW |
525 | ret = get_anon_bdev(&anon_dev); |
526 | if (ret < 0) | |
527 | goto fail_free; | |
528 | ||
e09fe2d2 QW |
529 | /* |
530 | * Don't create subvolume whose level is not zero. Or qgroup will be | |
01327610 | 531 | * screwed up since it assumes subvolume qgroup's level to be 0. |
e09fe2d2 | 532 | */ |
49a3c4d9 DS |
533 | if (btrfs_qgroup_level(objectid)) { |
534 | ret = -ENOSPC; | |
535 | goto fail_free; | |
536 | } | |
e09fe2d2 | 537 | |
d5c12070 | 538 | btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP); |
9ed74f2d | 539 | /* |
d5c12070 MX |
540 | * The same as the snapshot creation, please see the comment |
541 | * of create_snapshot(). | |
9ed74f2d | 542 | */ |
c4c129db | 543 | ret = btrfs_subvolume_reserve_metadata(root, &block_rsv, 8, false); |
d5c12070 | 544 | if (ret) |
49a3c4d9 | 545 | goto fail_free; |
d5c12070 MX |
546 | |
547 | trans = btrfs_start_transaction(root, 0); | |
548 | if (IS_ERR(trans)) { | |
549 | ret = PTR_ERR(trans); | |
e85fde51 | 550 | btrfs_subvolume_release_metadata(root, &block_rsv); |
49a3c4d9 | 551 | goto fail_free; |
d5c12070 MX |
552 | } |
553 | trans->block_rsv = &block_rsv; | |
554 | trans->bytes_reserved = block_rsv.size; | |
f46b5a66 | 555 | |
a9377422 | 556 | ret = btrfs_qgroup_inherit(trans, 0, objectid, inherit); |
6f72c7e2 AJ |
557 | if (ret) |
558 | goto fail; | |
559 | ||
9631e4cc JB |
560 | leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0, |
561 | BTRFS_NESTING_NORMAL); | |
8e8a1e31 JB |
562 | if (IS_ERR(leaf)) { |
563 | ret = PTR_ERR(leaf); | |
564 | goto fail; | |
565 | } | |
f46b5a66 | 566 | |
f46b5a66 CH |
567 | btrfs_mark_buffer_dirty(leaf); |
568 | ||
49a3c4d9 | 569 | inode_item = &root_item->inode; |
3cae210f QW |
570 | btrfs_set_stack_inode_generation(inode_item, 1); |
571 | btrfs_set_stack_inode_size(inode_item, 3); | |
572 | btrfs_set_stack_inode_nlink(inode_item, 1); | |
da17066c | 573 | btrfs_set_stack_inode_nbytes(inode_item, |
0b246afa | 574 | fs_info->nodesize); |
3cae210f | 575 | btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755); |
f46b5a66 | 576 | |
49a3c4d9 DS |
577 | btrfs_set_root_flags(root_item, 0); |
578 | btrfs_set_root_limit(root_item, 0); | |
3cae210f | 579 | btrfs_set_stack_inode_flags(inode_item, BTRFS_INODE_ROOT_ITEM_INIT); |
08fe4db1 | 580 | |
49a3c4d9 DS |
581 | btrfs_set_root_bytenr(root_item, leaf->start); |
582 | btrfs_set_root_generation(root_item, trans->transid); | |
583 | btrfs_set_root_level(root_item, 0); | |
584 | btrfs_set_root_refs(root_item, 1); | |
585 | btrfs_set_root_used(root_item, leaf->len); | |
586 | btrfs_set_root_last_snapshot(root_item, 0); | |
f46b5a66 | 587 | |
49a3c4d9 DS |
588 | btrfs_set_root_generation_v2(root_item, |
589 | btrfs_root_generation(root_item)); | |
807fc790 | 590 | generate_random_guid(root_item->uuid); |
49a3c4d9 DS |
591 | btrfs_set_stack_timespec_sec(&root_item->otime, cur_time.tv_sec); |
592 | btrfs_set_stack_timespec_nsec(&root_item->otime, cur_time.tv_nsec); | |
593 | root_item->ctime = root_item->otime; | |
594 | btrfs_set_root_ctransid(root_item, trans->transid); | |
595 | btrfs_set_root_otransid(root_item, trans->transid); | |
f46b5a66 | 596 | |
925baedd | 597 | btrfs_tree_unlock(leaf); |
f46b5a66 | 598 | |
69948022 | 599 | btrfs_set_root_dirid(root_item, BTRFS_FIRST_FREE_OBJECTID); |
f46b5a66 CH |
600 | |
601 | key.objectid = objectid; | |
5d4f98a2 | 602 | key.offset = 0; |
962a298f | 603 | key.type = BTRFS_ROOT_ITEM_KEY; |
0b246afa | 604 | ret = btrfs_insert_root(trans, fs_info->tree_root, &key, |
49a3c4d9 | 605 | root_item); |
67addf29 FM |
606 | if (ret) { |
607 | /* | |
608 | * Since we don't abort the transaction in this case, free the | |
609 | * tree block so that we don't leak space and leave the | |
610 | * filesystem in an inconsistent state (an extent item in the | |
611 | * extent tree without backreferences). Also no need to have | |
612 | * the tree block locked since it is not in any tree at this | |
613 | * point, so no other task can find it and use it. | |
614 | */ | |
615 | btrfs_free_tree_block(trans, root, leaf, 0, 1); | |
616 | free_extent_buffer(leaf); | |
f46b5a66 | 617 | goto fail; |
67addf29 FM |
618 | } |
619 | ||
620 | free_extent_buffer(leaf); | |
621 | leaf = NULL; | |
f46b5a66 | 622 | |
76dda93c | 623 | key.offset = (u64)-1; |
2dfb1e43 | 624 | new_root = btrfs_get_new_fs_root(fs_info, objectid, anon_dev); |
79787eaa | 625 | if (IS_ERR(new_root)) { |
2dfb1e43 | 626 | free_anon_bdev(anon_dev); |
79787eaa | 627 | ret = PTR_ERR(new_root); |
66642832 | 628 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
629 | goto fail; |
630 | } | |
2dfb1e43 QW |
631 | /* Freeing will be done in btrfs_put_root() of new_root */ |
632 | anon_dev = 0; | |
76dda93c | 633 | |
221581e4 JB |
634 | ret = btrfs_record_root_in_trans(trans, new_root); |
635 | if (ret) { | |
636 | btrfs_put_root(new_root); | |
637 | btrfs_abort_transaction(trans, ret); | |
638 | goto fail; | |
639 | } | |
76dda93c | 640 | |
69948022 | 641 | ret = btrfs_create_subvol_root(trans, new_root, root); |
00246528 | 642 | btrfs_put_root(new_root); |
ce598979 MF |
643 | if (ret) { |
644 | /* We potentially lose an unused inode item here */ | |
66642832 | 645 | btrfs_abort_transaction(trans, ret); |
ce598979 MF |
646 | goto fail; |
647 | } | |
648 | ||
f46b5a66 CH |
649 | /* |
650 | * insert the directory item | |
651 | */ | |
877574e2 | 652 | ret = btrfs_set_inode_index(BTRFS_I(dir), &index); |
79787eaa | 653 | if (ret) { |
66642832 | 654 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
655 | goto fail; |
656 | } | |
3de4586c | 657 | |
684572df | 658 | ret = btrfs_insert_dir_item(trans, name, namelen, BTRFS_I(dir), &key, |
3de4586c | 659 | BTRFS_FT_DIR, index); |
79787eaa | 660 | if (ret) { |
66642832 | 661 | btrfs_abort_transaction(trans, ret); |
f46b5a66 | 662 | goto fail; |
79787eaa | 663 | } |
0660b5af | 664 | |
6ef06d27 | 665 | btrfs_i_size_write(BTRFS_I(dir), dir->i_size + namelen * 2); |
9a56fcd1 | 666 | ret = btrfs_update_inode(trans, root, BTRFS_I(dir)); |
c7e54b51 JB |
667 | if (ret) { |
668 | btrfs_abort_transaction(trans, ret); | |
669 | goto fail; | |
670 | } | |
52c26179 | 671 | |
6025c19f | 672 | ret = btrfs_add_root_ref(trans, objectid, root->root_key.objectid, |
4a0cc7ca | 673 | btrfs_ino(BTRFS_I(dir)), index, name, namelen); |
c7e54b51 JB |
674 | if (ret) { |
675 | btrfs_abort_transaction(trans, ret); | |
676 | goto fail; | |
677 | } | |
f46b5a66 | 678 | |
cdb345a8 | 679 | ret = btrfs_uuid_tree_add(trans, root_item->uuid, |
6bccf3ab | 680 | BTRFS_UUID_KEY_SUBVOL, objectid); |
dd5f9615 | 681 | if (ret) |
66642832 | 682 | btrfs_abort_transaction(trans, ret); |
dd5f9615 | 683 | |
f46b5a66 | 684 | fail: |
49a3c4d9 | 685 | kfree(root_item); |
d5c12070 MX |
686 | trans->block_rsv = NULL; |
687 | trans->bytes_reserved = 0; | |
e85fde51 | 688 | btrfs_subvolume_release_metadata(root, &block_rsv); |
de6e8200 | 689 | |
9babda9f | 690 | err = btrfs_commit_transaction(trans); |
f46b5a66 CH |
691 | if (err && !ret) |
692 | ret = err; | |
1a65e24b | 693 | |
5662344b TI |
694 | if (!ret) { |
695 | inode = btrfs_lookup_dentry(dir, dentry); | |
de6e8200 LB |
696 | if (IS_ERR(inode)) |
697 | return PTR_ERR(inode); | |
5662344b TI |
698 | d_instantiate(dentry, inode); |
699 | } | |
f46b5a66 | 700 | return ret; |
49a3c4d9 DS |
701 | |
702 | fail_free: | |
2dfb1e43 QW |
703 | if (anon_dev) |
704 | free_anon_bdev(anon_dev); | |
49a3c4d9 DS |
705 | kfree(root_item); |
706 | return ret; | |
f46b5a66 CH |
707 | } |
708 | ||
e9662f70 | 709 | static int create_snapshot(struct btrfs_root *root, struct inode *dir, |
9babda9f | 710 | struct dentry *dentry, bool readonly, |
e9662f70 | 711 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 712 | { |
0b246afa | 713 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); |
2e4bfab9 | 714 | struct inode *inode; |
f46b5a66 CH |
715 | struct btrfs_pending_snapshot *pending_snapshot; |
716 | struct btrfs_trans_handle *trans; | |
2e4bfab9 | 717 | int ret; |
f46b5a66 | 718 | |
92a7cc42 | 719 | if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) |
f46b5a66 CH |
720 | return -EINVAL; |
721 | ||
eede2bf3 OS |
722 | if (atomic_read(&root->nr_swapfiles)) { |
723 | btrfs_warn(fs_info, | |
724 | "cannot snapshot subvolume with active swapfile"); | |
725 | return -ETXTBSY; | |
726 | } | |
727 | ||
23269bf5 | 728 | pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_KERNEL); |
a1ee7362 DS |
729 | if (!pending_snapshot) |
730 | return -ENOMEM; | |
731 | ||
2dfb1e43 QW |
732 | ret = get_anon_bdev(&pending_snapshot->anon_dev); |
733 | if (ret < 0) | |
734 | goto free_pending; | |
b0c0ea63 | 735 | pending_snapshot->root_item = kzalloc(sizeof(struct btrfs_root_item), |
23269bf5 | 736 | GFP_KERNEL); |
8546b570 DS |
737 | pending_snapshot->path = btrfs_alloc_path(); |
738 | if (!pending_snapshot->root_item || !pending_snapshot->path) { | |
b0c0ea63 DS |
739 | ret = -ENOMEM; |
740 | goto free_pending; | |
741 | } | |
742 | ||
66d8f3dd MX |
743 | btrfs_init_block_rsv(&pending_snapshot->block_rsv, |
744 | BTRFS_BLOCK_RSV_TEMP); | |
d5c12070 MX |
745 | /* |
746 | * 1 - parent dir inode | |
747 | * 2 - dir entries | |
748 | * 1 - root item | |
749 | * 2 - root ref/backref | |
750 | * 1 - root of snapshot | |
dd5f9615 | 751 | * 1 - UUID item |
d5c12070 MX |
752 | */ |
753 | ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root, | |
dd5f9615 | 754 | &pending_snapshot->block_rsv, 8, |
ee3441b4 | 755 | false); |
d5c12070 | 756 | if (ret) |
c11fbb6e | 757 | goto free_pending; |
d5c12070 | 758 | |
3de4586c | 759 | pending_snapshot->dentry = dentry; |
f46b5a66 | 760 | pending_snapshot->root = root; |
b83cc969 | 761 | pending_snapshot->readonly = readonly; |
e9662f70 | 762 | pending_snapshot->dir = dir; |
8696c533 | 763 | pending_snapshot->inherit = inherit; |
a22285a6 | 764 | |
d5c12070 | 765 | trans = btrfs_start_transaction(root, 0); |
a22285a6 YZ |
766 | if (IS_ERR(trans)) { |
767 | ret = PTR_ERR(trans); | |
768 | goto fail; | |
769 | } | |
770 | ||
0b246afa | 771 | spin_lock(&fs_info->trans_lock); |
f46b5a66 CH |
772 | list_add(&pending_snapshot->list, |
773 | &trans->transaction->pending_snapshots); | |
0b246afa | 774 | spin_unlock(&fs_info->trans_lock); |
9babda9f NB |
775 | |
776 | ret = btrfs_commit_transaction(trans); | |
aec8030a | 777 | if (ret) |
c37b2b62 | 778 | goto fail; |
a22285a6 YZ |
779 | |
780 | ret = pending_snapshot->error; | |
781 | if (ret) | |
782 | goto fail; | |
783 | ||
d3797308 CM |
784 | ret = btrfs_orphan_cleanup(pending_snapshot->snap); |
785 | if (ret) | |
786 | goto fail; | |
787 | ||
2b0143b5 | 788 | inode = btrfs_lookup_dentry(d_inode(dentry->d_parent), dentry); |
2e4bfab9 YZ |
789 | if (IS_ERR(inode)) { |
790 | ret = PTR_ERR(inode); | |
791 | goto fail; | |
792 | } | |
5662344b | 793 | |
2e4bfab9 YZ |
794 | d_instantiate(dentry, inode); |
795 | ret = 0; | |
2dfb1e43 | 796 | pending_snapshot->anon_dev = 0; |
2e4bfab9 | 797 | fail: |
2dfb1e43 QW |
798 | /* Prevent double freeing of anon_dev */ |
799 | if (ret && pending_snapshot->snap) | |
800 | pending_snapshot->snap->anon_dev = 0; | |
00246528 | 801 | btrfs_put_root(pending_snapshot->snap); |
e85fde51 | 802 | btrfs_subvolume_release_metadata(root, &pending_snapshot->block_rsv); |
b0c0ea63 | 803 | free_pending: |
2dfb1e43 QW |
804 | if (pending_snapshot->anon_dev) |
805 | free_anon_bdev(pending_snapshot->anon_dev); | |
b0c0ea63 | 806 | kfree(pending_snapshot->root_item); |
8546b570 | 807 | btrfs_free_path(pending_snapshot->path); |
a1ee7362 DS |
808 | kfree(pending_snapshot); |
809 | ||
f46b5a66 CH |
810 | return ret; |
811 | } | |
812 | ||
4260f7c7 SW |
813 | /* copy of may_delete in fs/namei.c() |
814 | * Check whether we can remove a link victim from directory dir, check | |
815 | * whether the type of victim is right. | |
816 | * 1. We can't do it if dir is read-only (done in permission()) | |
817 | * 2. We should have write and exec permissions on dir | |
818 | * 3. We can't remove anything from append-only dir | |
819 | * 4. We can't do anything with immutable dir (done in permission()) | |
820 | * 5. If the sticky bit on dir is set we should either | |
821 | * a. be owner of dir, or | |
822 | * b. be owner of victim, or | |
823 | * c. have CAP_FOWNER capability | |
01327610 | 824 | * 6. If the victim is append-only or immutable we can't do anything with |
4260f7c7 SW |
825 | * links pointing to it. |
826 | * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR. | |
827 | * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR. | |
828 | * 9. We can't remove a root or mountpoint. | |
829 | * 10. We don't allow removal of NFS sillyrenamed files; it's handled by | |
830 | * nfs_async_unlink(). | |
831 | */ | |
832 | ||
67871254 | 833 | static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir) |
4260f7c7 SW |
834 | { |
835 | int error; | |
836 | ||
2b0143b5 | 837 | if (d_really_is_negative(victim)) |
4260f7c7 SW |
838 | return -ENOENT; |
839 | ||
2b0143b5 | 840 | BUG_ON(d_inode(victim->d_parent) != dir); |
4fa6b5ec | 841 | audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE); |
4260f7c7 | 842 | |
47291baa | 843 | error = inode_permission(&init_user_ns, dir, MAY_WRITE | MAY_EXEC); |
4260f7c7 SW |
844 | if (error) |
845 | return error; | |
846 | if (IS_APPEND(dir)) | |
847 | return -EPERM; | |
ba73d987 CB |
848 | if (check_sticky(&init_user_ns, dir, d_inode(victim)) || |
849 | IS_APPEND(d_inode(victim)) || IS_IMMUTABLE(d_inode(victim)) || | |
850 | IS_SWAPFILE(d_inode(victim))) | |
4260f7c7 SW |
851 | return -EPERM; |
852 | if (isdir) { | |
e36cb0b8 | 853 | if (!d_is_dir(victim)) |
4260f7c7 SW |
854 | return -ENOTDIR; |
855 | if (IS_ROOT(victim)) | |
856 | return -EBUSY; | |
e36cb0b8 | 857 | } else if (d_is_dir(victim)) |
4260f7c7 SW |
858 | return -EISDIR; |
859 | if (IS_DEADDIR(dir)) | |
860 | return -ENOENT; | |
861 | if (victim->d_flags & DCACHE_NFSFS_RENAMED) | |
862 | return -EBUSY; | |
863 | return 0; | |
864 | } | |
865 | ||
cb8e7090 CH |
866 | /* copy of may_create in fs/namei.c() */ |
867 | static inline int btrfs_may_create(struct inode *dir, struct dentry *child) | |
868 | { | |
2b0143b5 | 869 | if (d_really_is_positive(child)) |
cb8e7090 CH |
870 | return -EEXIST; |
871 | if (IS_DEADDIR(dir)) | |
872 | return -ENOENT; | |
47291baa | 873 | return inode_permission(&init_user_ns, dir, MAY_WRITE | MAY_EXEC); |
cb8e7090 CH |
874 | } |
875 | ||
876 | /* | |
877 | * Create a new subvolume below @parent. This is largely modeled after | |
878 | * sys_mkdirat and vfs_mkdir, but we only do a single component lookup | |
879 | * inside this filesystem so it's quite a bit simpler. | |
880 | */ | |
92872094 | 881 | static noinline int btrfs_mksubvol(const struct path *parent, |
52f75f4f | 882 | const char *name, int namelen, |
72fd032e | 883 | struct btrfs_root *snap_src, |
9babda9f | 884 | bool readonly, |
8696c533 | 885 | struct btrfs_qgroup_inherit *inherit) |
cb8e7090 | 886 | { |
0b246afa JM |
887 | struct inode *dir = d_inode(parent->dentry); |
888 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); | |
cb8e7090 CH |
889 | struct dentry *dentry; |
890 | int error; | |
891 | ||
00235411 AV |
892 | error = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT); |
893 | if (error == -EINTR) | |
894 | return error; | |
cb8e7090 CH |
895 | |
896 | dentry = lookup_one_len(name, parent->dentry, namelen); | |
897 | error = PTR_ERR(dentry); | |
898 | if (IS_ERR(dentry)) | |
899 | goto out_unlock; | |
900 | ||
76dda93c | 901 | error = btrfs_may_create(dir, dentry); |
cb8e7090 | 902 | if (error) |
a874a63e | 903 | goto out_dput; |
cb8e7090 | 904 | |
9c52057c CM |
905 | /* |
906 | * even if this name doesn't exist, we may get hash collisions. | |
907 | * check for them now when we can safely fail | |
908 | */ | |
909 | error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root, | |
910 | dir->i_ino, name, | |
911 | namelen); | |
912 | if (error) | |
913 | goto out_dput; | |
914 | ||
0b246afa | 915 | down_read(&fs_info->subvol_sem); |
76dda93c YZ |
916 | |
917 | if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0) | |
918 | goto out_up_read; | |
919 | ||
9babda9f NB |
920 | if (snap_src) |
921 | error = create_snapshot(snap_src, dir, dentry, readonly, inherit); | |
922 | else | |
923 | error = create_subvol(dir, dentry, name, namelen, inherit); | |
924 | ||
76dda93c YZ |
925 | if (!error) |
926 | fsnotify_mkdir(dir, dentry); | |
927 | out_up_read: | |
0b246afa | 928 | up_read(&fs_info->subvol_sem); |
cb8e7090 CH |
929 | out_dput: |
930 | dput(dentry); | |
931 | out_unlock: | |
64708539 | 932 | btrfs_inode_unlock(dir, 0); |
cb8e7090 CH |
933 | return error; |
934 | } | |
935 | ||
c11fbb6e RK |
936 | static noinline int btrfs_mksnapshot(const struct path *parent, |
937 | const char *name, int namelen, | |
938 | struct btrfs_root *root, | |
939 | bool readonly, | |
940 | struct btrfs_qgroup_inherit *inherit) | |
941 | { | |
942 | int ret; | |
943 | bool snapshot_force_cow = false; | |
944 | ||
945 | /* | |
946 | * Force new buffered writes to reserve space even when NOCOW is | |
947 | * possible. This is to avoid later writeback (running dealloc) to | |
948 | * fallback to COW mode and unexpectedly fail with ENOSPC. | |
949 | */ | |
950 | btrfs_drew_read_lock(&root->snapshot_lock); | |
951 | ||
f9baa501 | 952 | ret = btrfs_start_delalloc_snapshot(root, false); |
c11fbb6e RK |
953 | if (ret) |
954 | goto out; | |
955 | ||
956 | /* | |
957 | * All previous writes have started writeback in NOCOW mode, so now | |
958 | * we force future writes to fallback to COW mode during snapshot | |
959 | * creation. | |
960 | */ | |
961 | atomic_inc(&root->snapshot_force_cow); | |
962 | snapshot_force_cow = true; | |
963 | ||
964 | btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1); | |
965 | ||
966 | ret = btrfs_mksubvol(parent, name, namelen, | |
967 | root, readonly, inherit); | |
968 | out: | |
969 | if (snapshot_force_cow) | |
970 | atomic_dec(&root->snapshot_force_cow); | |
971 | btrfs_drew_read_unlock(&root->snapshot_lock); | |
972 | return ret; | |
973 | } | |
974 | ||
4cb5300b CM |
975 | /* |
976 | * When we're defragging a range, we don't want to kick it off again | |
977 | * if it is really just waiting for delalloc to send it down. | |
978 | * If we find a nice big extent or delalloc range for the bytes in the | |
979 | * file you want to defrag, we return 0 to let you know to skip this | |
980 | * part of the file | |
981 | */ | |
aab110ab | 982 | static int check_defrag_in_cache(struct inode *inode, u64 offset, u32 thresh) |
4cb5300b CM |
983 | { |
984 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; | |
985 | struct extent_map *em = NULL; | |
986 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
987 | u64 end; | |
988 | ||
989 | read_lock(&em_tree->lock); | |
09cbfeaf | 990 | em = lookup_extent_mapping(em_tree, offset, PAGE_SIZE); |
4cb5300b CM |
991 | read_unlock(&em_tree->lock); |
992 | ||
993 | if (em) { | |
994 | end = extent_map_end(em); | |
995 | free_extent_map(em); | |
996 | if (end - offset > thresh) | |
997 | return 0; | |
998 | } | |
999 | /* if we already have a nice delalloc here, just stop */ | |
1000 | thresh /= 2; | |
1001 | end = count_range_bits(io_tree, &offset, offset + thresh, | |
1002 | thresh, EXTENT_DELALLOC, 1); | |
1003 | if (end >= thresh) | |
1004 | return 0; | |
1005 | return 1; | |
1006 | } | |
1007 | ||
1008 | /* | |
1009 | * helper function to walk through a file and find extents | |
1010 | * newer than a specific transid, and smaller than thresh. | |
1011 | * | |
1012 | * This is used by the defragging code to find new and small | |
1013 | * extents | |
1014 | */ | |
1015 | static int find_new_extents(struct btrfs_root *root, | |
1016 | struct inode *inode, u64 newer_than, | |
aab110ab | 1017 | u64 *off, u32 thresh) |
4cb5300b CM |
1018 | { |
1019 | struct btrfs_path *path; | |
1020 | struct btrfs_key min_key; | |
4cb5300b CM |
1021 | struct extent_buffer *leaf; |
1022 | struct btrfs_file_extent_item *extent; | |
1023 | int type; | |
1024 | int ret; | |
4a0cc7ca | 1025 | u64 ino = btrfs_ino(BTRFS_I(inode)); |
4cb5300b CM |
1026 | |
1027 | path = btrfs_alloc_path(); | |
1028 | if (!path) | |
1029 | return -ENOMEM; | |
1030 | ||
a4689d2b | 1031 | min_key.objectid = ino; |
4cb5300b CM |
1032 | min_key.type = BTRFS_EXTENT_DATA_KEY; |
1033 | min_key.offset = *off; | |
1034 | ||
67871254 | 1035 | while (1) { |
6174d3cb | 1036 | ret = btrfs_search_forward(root, &min_key, path, newer_than); |
4cb5300b CM |
1037 | if (ret != 0) |
1038 | goto none; | |
f094c9bd | 1039 | process_slot: |
a4689d2b | 1040 | if (min_key.objectid != ino) |
4cb5300b CM |
1041 | goto none; |
1042 | if (min_key.type != BTRFS_EXTENT_DATA_KEY) | |
1043 | goto none; | |
1044 | ||
1045 | leaf = path->nodes[0]; | |
1046 | extent = btrfs_item_ptr(leaf, path->slots[0], | |
1047 | struct btrfs_file_extent_item); | |
1048 | ||
1049 | type = btrfs_file_extent_type(leaf, extent); | |
1050 | if (type == BTRFS_FILE_EXTENT_REG && | |
1051 | btrfs_file_extent_num_bytes(leaf, extent) < thresh && | |
1052 | check_defrag_in_cache(inode, min_key.offset, thresh)) { | |
1053 | *off = min_key.offset; | |
1054 | btrfs_free_path(path); | |
1055 | return 0; | |
1056 | } | |
1057 | ||
f094c9bd FM |
1058 | path->slots[0]++; |
1059 | if (path->slots[0] < btrfs_header_nritems(leaf)) { | |
1060 | btrfs_item_key_to_cpu(leaf, &min_key, path->slots[0]); | |
1061 | goto process_slot; | |
1062 | } | |
1063 | ||
4cb5300b CM |
1064 | if (min_key.offset == (u64)-1) |
1065 | goto none; | |
1066 | ||
1067 | min_key.offset++; | |
1068 | btrfs_release_path(path); | |
1069 | } | |
1070 | none: | |
1071 | btrfs_free_path(path); | |
1072 | return -ENOENT; | |
1073 | } | |
1074 | ||
6c282eb4 | 1075 | static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start) |
17ce6ef8 LB |
1076 | { |
1077 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
6c282eb4 LZ |
1078 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; |
1079 | struct extent_map *em; | |
09cbfeaf | 1080 | u64 len = PAGE_SIZE; |
17ce6ef8 | 1081 | |
6c282eb4 LZ |
1082 | /* |
1083 | * hopefully we have this extent in the tree already, try without | |
1084 | * the full extent lock | |
1085 | */ | |
17ce6ef8 | 1086 | read_lock(&em_tree->lock); |
6c282eb4 | 1087 | em = lookup_extent_mapping(em_tree, start, len); |
17ce6ef8 LB |
1088 | read_unlock(&em_tree->lock); |
1089 | ||
6c282eb4 | 1090 | if (!em) { |
308d9800 FM |
1091 | struct extent_state *cached = NULL; |
1092 | u64 end = start + len - 1; | |
1093 | ||
6c282eb4 | 1094 | /* get the big lock and read metadata off disk */ |
ff13db41 | 1095 | lock_extent_bits(io_tree, start, end, &cached); |
39b07b5d | 1096 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start, len); |
e43bbe5e | 1097 | unlock_extent_cached(io_tree, start, end, &cached); |
6c282eb4 LZ |
1098 | |
1099 | if (IS_ERR(em)) | |
1100 | return NULL; | |
1101 | } | |
1102 | ||
1103 | return em; | |
1104 | } | |
17ce6ef8 | 1105 | |
6c282eb4 LZ |
1106 | static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em) |
1107 | { | |
1108 | struct extent_map *next; | |
1109 | bool ret = true; | |
1110 | ||
1111 | /* this is the last extent */ | |
1112 | if (em->start + em->len >= i_size_read(inode)) | |
1113 | return false; | |
1114 | ||
1115 | next = defrag_lookup_extent(inode, em->start + em->len); | |
e9512d72 CM |
1116 | if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE) |
1117 | ret = false; | |
1118 | else if ((em->block_start + em->block_len == next->block_start) && | |
ee22184b | 1119 | (em->block_len > SZ_128K && next->block_len > SZ_128K)) |
6c282eb4 LZ |
1120 | ret = false; |
1121 | ||
1122 | free_extent_map(next); | |
17ce6ef8 LB |
1123 | return ret; |
1124 | } | |
1125 | ||
aab110ab | 1126 | static int should_defrag_range(struct inode *inode, u64 start, u32 thresh, |
a43a2111 AM |
1127 | u64 *last_len, u64 *skip, u64 *defrag_end, |
1128 | int compress) | |
940100a4 | 1129 | { |
6c282eb4 | 1130 | struct extent_map *em; |
940100a4 | 1131 | int ret = 1; |
6c282eb4 | 1132 | bool next_mergeable = true; |
4a3560c4 | 1133 | bool prev_mergeable = true; |
940100a4 CM |
1134 | |
1135 | /* | |
008873ea | 1136 | * make sure that once we start defragging an extent, we keep on |
940100a4 CM |
1137 | * defragging it |
1138 | */ | |
1139 | if (start < *defrag_end) | |
1140 | return 1; | |
1141 | ||
1142 | *skip = 0; | |
1143 | ||
6c282eb4 LZ |
1144 | em = defrag_lookup_extent(inode, start); |
1145 | if (!em) | |
1146 | return 0; | |
940100a4 CM |
1147 | |
1148 | /* this will cover holes, and inline extents */ | |
17ce6ef8 | 1149 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { |
940100a4 | 1150 | ret = 0; |
17ce6ef8 LB |
1151 | goto out; |
1152 | } | |
1153 | ||
4a3560c4 LB |
1154 | if (!*defrag_end) |
1155 | prev_mergeable = false; | |
1156 | ||
6c282eb4 | 1157 | next_mergeable = defrag_check_next_extent(inode, em); |
940100a4 | 1158 | /* |
6c282eb4 LZ |
1159 | * we hit a real extent, if it is big or the next extent is not a |
1160 | * real extent, don't bother defragging it | |
940100a4 | 1161 | */ |
a43a2111 | 1162 | if (!compress && (*last_len == 0 || *last_len >= thresh) && |
4a3560c4 | 1163 | (em->len >= thresh || (!next_mergeable && !prev_mergeable))) |
940100a4 | 1164 | ret = 0; |
17ce6ef8 | 1165 | out: |
940100a4 CM |
1166 | /* |
1167 | * last_len ends up being a counter of how many bytes we've defragged. | |
1168 | * every time we choose not to defrag an extent, we reset *last_len | |
1169 | * so that the next tiny extent will force a defrag. | |
1170 | * | |
1171 | * The end result of this is that tiny extents before a single big | |
1172 | * extent will force at least part of that big extent to be defragged. | |
1173 | */ | |
1174 | if (ret) { | |
940100a4 CM |
1175 | *defrag_end = extent_map_end(em); |
1176 | } else { | |
1177 | *last_len = 0; | |
1178 | *skip = extent_map_end(em); | |
1179 | *defrag_end = 0; | |
1180 | } | |
1181 | ||
1182 | free_extent_map(em); | |
1183 | return ret; | |
1184 | } | |
1185 | ||
4cb5300b CM |
1186 | /* |
1187 | * it doesn't do much good to defrag one or two pages | |
1188 | * at a time. This pulls in a nice chunk of pages | |
1189 | * to COW and defrag. | |
1190 | * | |
1191 | * It also makes sure the delalloc code has enough | |
1192 | * dirty data to avoid making new small extents as part | |
1193 | * of the defrag | |
1194 | * | |
1195 | * It's a good idea to start RA on this range | |
1196 | * before calling this. | |
1197 | */ | |
1198 | static int cluster_pages_for_defrag(struct inode *inode, | |
1199 | struct page **pages, | |
1200 | unsigned long start_index, | |
c41570c9 | 1201 | unsigned long num_pages) |
f46b5a66 | 1202 | { |
4cb5300b CM |
1203 | unsigned long file_end; |
1204 | u64 isize = i_size_read(inode); | |
1205 | u64 page_start; | |
1206 | u64 page_end; | |
1f12bd06 | 1207 | u64 page_cnt; |
a1fbc675 | 1208 | u64 start = (u64)start_index << PAGE_SHIFT; |
7f458a38 | 1209 | u64 search_start; |
4cb5300b CM |
1210 | int ret; |
1211 | int i; | |
1212 | int i_done; | |
3eaa2885 | 1213 | struct btrfs_ordered_extent *ordered; |
4cb5300b | 1214 | struct extent_state *cached_state = NULL; |
600a45e1 | 1215 | struct extent_io_tree *tree; |
364ecf36 | 1216 | struct extent_changeset *data_reserved = NULL; |
3b16a4e3 | 1217 | gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); |
4cb5300b | 1218 | |
09cbfeaf | 1219 | file_end = (isize - 1) >> PAGE_SHIFT; |
1f12bd06 LB |
1220 | if (!isize || start_index > file_end) |
1221 | return 0; | |
1222 | ||
1223 | page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1); | |
4cb5300b | 1224 | |
e5b7231e | 1225 | ret = btrfs_delalloc_reserve_space(BTRFS_I(inode), &data_reserved, |
a1fbc675 | 1226 | start, page_cnt << PAGE_SHIFT); |
4cb5300b CM |
1227 | if (ret) |
1228 | return ret; | |
4cb5300b | 1229 | i_done = 0; |
600a45e1 | 1230 | tree = &BTRFS_I(inode)->io_tree; |
4cb5300b CM |
1231 | |
1232 | /* step one, lock all the pages */ | |
1f12bd06 | 1233 | for (i = 0; i < page_cnt; i++) { |
4cb5300b | 1234 | struct page *page; |
600a45e1 | 1235 | again: |
a94733d0 | 1236 | page = find_or_create_page(inode->i_mapping, |
600a45e1 | 1237 | start_index + i, mask); |
4cb5300b CM |
1238 | if (!page) |
1239 | break; | |
1240 | ||
32443de3 QW |
1241 | ret = set_page_extent_mapped(page); |
1242 | if (ret < 0) { | |
1243 | unlock_page(page); | |
1244 | put_page(page); | |
1245 | break; | |
1246 | } | |
1247 | ||
600a45e1 | 1248 | page_start = page_offset(page); |
09cbfeaf | 1249 | page_end = page_start + PAGE_SIZE - 1; |
600a45e1 | 1250 | while (1) { |
308d9800 | 1251 | lock_extent_bits(tree, page_start, page_end, |
ff13db41 | 1252 | &cached_state); |
c3504372 | 1253 | ordered = btrfs_lookup_ordered_extent(BTRFS_I(inode), |
600a45e1 | 1254 | page_start); |
308d9800 | 1255 | unlock_extent_cached(tree, page_start, page_end, |
e43bbe5e | 1256 | &cached_state); |
600a45e1 MX |
1257 | if (!ordered) |
1258 | break; | |
1259 | ||
1260 | unlock_page(page); | |
c0a43603 | 1261 | btrfs_start_ordered_extent(ordered, 1); |
600a45e1 MX |
1262 | btrfs_put_ordered_extent(ordered); |
1263 | lock_page(page); | |
1f12bd06 LB |
1264 | /* |
1265 | * we unlocked the page above, so we need check if | |
1266 | * it was released or not. | |
1267 | */ | |
1268 | if (page->mapping != inode->i_mapping) { | |
1269 | unlock_page(page); | |
09cbfeaf | 1270 | put_page(page); |
1f12bd06 LB |
1271 | goto again; |
1272 | } | |
600a45e1 MX |
1273 | } |
1274 | ||
4cb5300b CM |
1275 | if (!PageUptodate(page)) { |
1276 | btrfs_readpage(NULL, page); | |
1277 | lock_page(page); | |
1278 | if (!PageUptodate(page)) { | |
1279 | unlock_page(page); | |
09cbfeaf | 1280 | put_page(page); |
4cb5300b CM |
1281 | ret = -EIO; |
1282 | break; | |
1283 | } | |
1284 | } | |
600a45e1 | 1285 | |
600a45e1 MX |
1286 | if (page->mapping != inode->i_mapping) { |
1287 | unlock_page(page); | |
09cbfeaf | 1288 | put_page(page); |
600a45e1 MX |
1289 | goto again; |
1290 | } | |
1291 | ||
4cb5300b CM |
1292 | pages[i] = page; |
1293 | i_done++; | |
1294 | } | |
1295 | if (!i_done || ret) | |
1296 | goto out; | |
1297 | ||
1751e8a6 | 1298 | if (!(inode->i_sb->s_flags & SB_ACTIVE)) |
4cb5300b CM |
1299 | goto out; |
1300 | ||
1301 | /* | |
1302 | * so now we have a nice long stream of locked | |
1303 | * and up to date pages, lets wait on them | |
1304 | */ | |
1305 | for (i = 0; i < i_done; i++) | |
1306 | wait_on_page_writeback(pages[i]); | |
1307 | ||
1308 | page_start = page_offset(pages[0]); | |
09cbfeaf | 1309 | page_end = page_offset(pages[i_done - 1]) + PAGE_SIZE; |
4cb5300b CM |
1310 | |
1311 | lock_extent_bits(&BTRFS_I(inode)->io_tree, | |
ff13db41 | 1312 | page_start, page_end - 1, &cached_state); |
7f458a38 FM |
1313 | |
1314 | /* | |
1315 | * When defragmenting we skip ranges that have holes or inline extents, | |
1316 | * (check should_defrag_range()), to avoid unnecessary IO and wasting | |
1317 | * space. At btrfs_defrag_file(), we check if a range should be defragged | |
1318 | * before locking the inode and then, if it should, we trigger a sync | |
1319 | * page cache readahead - we lock the inode only after that to avoid | |
1320 | * blocking for too long other tasks that possibly want to operate on | |
1321 | * other file ranges. But before we were able to get the inode lock, | |
1322 | * some other task may have punched a hole in the range, or we may have | |
1323 | * now an inline extent, in which case we should not defrag. So check | |
1324 | * for that here, where we have the inode and the range locked, and bail | |
1325 | * out if that happened. | |
1326 | */ | |
1327 | search_start = page_start; | |
1328 | while (search_start < page_end) { | |
1329 | struct extent_map *em; | |
1330 | ||
1331 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, search_start, | |
1332 | page_end - search_start); | |
1333 | if (IS_ERR(em)) { | |
1334 | ret = PTR_ERR(em); | |
1335 | goto out_unlock_range; | |
1336 | } | |
1337 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { | |
1338 | free_extent_map(em); | |
1339 | /* Ok, 0 means we did not defrag anything */ | |
1340 | ret = 0; | |
1341 | goto out_unlock_range; | |
1342 | } | |
1343 | search_start = extent_map_end(em); | |
1344 | free_extent_map(em); | |
1345 | } | |
1346 | ||
4cb5300b | 1347 | clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, |
e182163d OS |
1348 | page_end - 1, EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | |
1349 | EXTENT_DEFRAG, 0, 0, &cached_state); | |
4cb5300b | 1350 | |
1f12bd06 | 1351 | if (i_done != page_cnt) { |
9e0baf60 | 1352 | spin_lock(&BTRFS_I(inode)->lock); |
28c4a3e2 | 1353 | btrfs_mod_outstanding_extents(BTRFS_I(inode), 1); |
9e0baf60 | 1354 | spin_unlock(&BTRFS_I(inode)->lock); |
86d52921 | 1355 | btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, |
a1fbc675 | 1356 | start, (page_cnt - i_done) << PAGE_SHIFT, true); |
4cb5300b CM |
1357 | } |
1358 | ||
1359 | ||
9e8a4a8b | 1360 | set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1, |
018ed4f7 | 1361 | &cached_state); |
4cb5300b CM |
1362 | |
1363 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
e43bbe5e | 1364 | page_start, page_end - 1, &cached_state); |
4cb5300b CM |
1365 | |
1366 | for (i = 0; i < i_done; i++) { | |
1367 | clear_page_dirty_for_io(pages[i]); | |
1368 | ClearPageChecked(pages[i]); | |
4cb5300b CM |
1369 | set_page_dirty(pages[i]); |
1370 | unlock_page(pages[i]); | |
09cbfeaf | 1371 | put_page(pages[i]); |
4cb5300b | 1372 | } |
8702ba93 | 1373 | btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT); |
364ecf36 | 1374 | extent_changeset_free(data_reserved); |
4cb5300b | 1375 | return i_done; |
7f458a38 FM |
1376 | |
1377 | out_unlock_range: | |
1378 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
1379 | page_start, page_end - 1, &cached_state); | |
4cb5300b CM |
1380 | out: |
1381 | for (i = 0; i < i_done; i++) { | |
1382 | unlock_page(pages[i]); | |
09cbfeaf | 1383 | put_page(pages[i]); |
4cb5300b | 1384 | } |
86d52921 | 1385 | btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, |
a1fbc675 | 1386 | start, page_cnt << PAGE_SHIFT, true); |
8702ba93 | 1387 | btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT); |
364ecf36 | 1388 | extent_changeset_free(data_reserved); |
4cb5300b CM |
1389 | return ret; |
1390 | ||
1391 | } | |
1392 | ||
1393 | int btrfs_defrag_file(struct inode *inode, struct file *file, | |
1394 | struct btrfs_ioctl_defrag_range_args *range, | |
1395 | u64 newer_than, unsigned long max_to_defrag) | |
1396 | { | |
0b246afa | 1397 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
4cb5300b | 1398 | struct btrfs_root *root = BTRFS_I(inode)->root; |
4cb5300b | 1399 | struct file_ra_state *ra = NULL; |
f46b5a66 | 1400 | unsigned long last_index; |
151a31b2 | 1401 | u64 isize = i_size_read(inode); |
940100a4 CM |
1402 | u64 last_len = 0; |
1403 | u64 skip = 0; | |
1404 | u64 defrag_end = 0; | |
4cb5300b | 1405 | u64 newer_off = range->start; |
f46b5a66 | 1406 | unsigned long i; |
008873ea | 1407 | unsigned long ra_index = 0; |
f46b5a66 | 1408 | int ret; |
4cb5300b | 1409 | int defrag_count = 0; |
1a419d85 | 1410 | int compress_type = BTRFS_COMPRESS_ZLIB; |
aab110ab | 1411 | u32 extent_thresh = range->extent_thresh; |
09cbfeaf | 1412 | unsigned long max_cluster = SZ_256K >> PAGE_SHIFT; |
c41570c9 | 1413 | unsigned long cluster = max_cluster; |
ee22184b | 1414 | u64 new_align = ~((u64)SZ_128K - 1); |
4cb5300b | 1415 | struct page **pages = NULL; |
1e2ef46d | 1416 | bool do_compress = range->flags & BTRFS_DEFRAG_RANGE_COMPRESS; |
4cb5300b | 1417 | |
0abd5b17 LB |
1418 | if (isize == 0) |
1419 | return 0; | |
1420 | ||
1421 | if (range->start >= isize) | |
1422 | return -EINVAL; | |
1a419d85 | 1423 | |
1e2ef46d | 1424 | if (do_compress) { |
ce96b7ff | 1425 | if (range->compress_type >= BTRFS_NR_COMPRESS_TYPES) |
1a419d85 LZ |
1426 | return -EINVAL; |
1427 | if (range->compress_type) | |
1428 | compress_type = range->compress_type; | |
1429 | } | |
f46b5a66 | 1430 | |
0abd5b17 | 1431 | if (extent_thresh == 0) |
ee22184b | 1432 | extent_thresh = SZ_256K; |
940100a4 | 1433 | |
4cb5300b | 1434 | /* |
0a52d108 DS |
1435 | * If we were not given a file, allocate a readahead context. As |
1436 | * readahead is just an optimization, defrag will work without it so | |
1437 | * we don't error out. | |
4cb5300b CM |
1438 | */ |
1439 | if (!file) { | |
63e727ec | 1440 | ra = kzalloc(sizeof(*ra), GFP_KERNEL); |
0a52d108 DS |
1441 | if (ra) |
1442 | file_ra_state_init(ra, inode->i_mapping); | |
4cb5300b CM |
1443 | } else { |
1444 | ra = &file->f_ra; | |
1445 | } | |
1446 | ||
63e727ec | 1447 | pages = kmalloc_array(max_cluster, sizeof(struct page *), GFP_KERNEL); |
4cb5300b CM |
1448 | if (!pages) { |
1449 | ret = -ENOMEM; | |
1450 | goto out_ra; | |
1451 | } | |
1452 | ||
1453 | /* find the last page to defrag */ | |
1e701a32 | 1454 | if (range->start + range->len > range->start) { |
151a31b2 | 1455 | last_index = min_t(u64, isize - 1, |
09cbfeaf | 1456 | range->start + range->len - 1) >> PAGE_SHIFT; |
1e701a32 | 1457 | } else { |
09cbfeaf | 1458 | last_index = (isize - 1) >> PAGE_SHIFT; |
1e701a32 CM |
1459 | } |
1460 | ||
4cb5300b CM |
1461 | if (newer_than) { |
1462 | ret = find_new_extents(root, inode, newer_than, | |
ee22184b | 1463 | &newer_off, SZ_64K); |
4cb5300b CM |
1464 | if (!ret) { |
1465 | range->start = newer_off; | |
1466 | /* | |
1467 | * we always align our defrag to help keep | |
1468 | * the extents in the file evenly spaced | |
1469 | */ | |
09cbfeaf | 1470 | i = (newer_off & new_align) >> PAGE_SHIFT; |
4cb5300b CM |
1471 | } else |
1472 | goto out_ra; | |
1473 | } else { | |
09cbfeaf | 1474 | i = range->start >> PAGE_SHIFT; |
4cb5300b CM |
1475 | } |
1476 | if (!max_to_defrag) | |
070034bd | 1477 | max_to_defrag = last_index - i + 1; |
4cb5300b | 1478 | |
2a0f7f57 LZ |
1479 | /* |
1480 | * make writeback starts from i, so the defrag range can be | |
1481 | * written sequentially. | |
1482 | */ | |
1483 | if (i < inode->i_mapping->writeback_index) | |
1484 | inode->i_mapping->writeback_index = i; | |
1485 | ||
f7f43cc8 | 1486 | while (i <= last_index && defrag_count < max_to_defrag && |
09cbfeaf | 1487 | (i < DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE))) { |
4cb5300b CM |
1488 | /* |
1489 | * make sure we stop running if someone unmounts | |
1490 | * the FS | |
1491 | */ | |
1751e8a6 | 1492 | if (!(inode->i_sb->s_flags & SB_ACTIVE)) |
4cb5300b CM |
1493 | break; |
1494 | ||
0b246afa JM |
1495 | if (btrfs_defrag_cancelled(fs_info)) { |
1496 | btrfs_debug(fs_info, "defrag_file cancelled"); | |
210549eb | 1497 | ret = -EAGAIN; |
50535db8 | 1498 | goto error; |
210549eb DS |
1499 | } |
1500 | ||
09cbfeaf | 1501 | if (!should_defrag_range(inode, (u64)i << PAGE_SHIFT, |
6c282eb4 | 1502 | extent_thresh, &last_len, &skip, |
1e2ef46d | 1503 | &defrag_end, do_compress)){ |
940100a4 CM |
1504 | unsigned long next; |
1505 | /* | |
1506 | * the should_defrag function tells us how much to skip | |
1507 | * bump our counter by the suggested amount | |
1508 | */ | |
09cbfeaf | 1509 | next = DIV_ROUND_UP(skip, PAGE_SIZE); |
940100a4 CM |
1510 | i = max(i + 1, next); |
1511 | continue; | |
1512 | } | |
008873ea LZ |
1513 | |
1514 | if (!newer_than) { | |
09cbfeaf KS |
1515 | cluster = (PAGE_ALIGN(defrag_end) >> |
1516 | PAGE_SHIFT) - i; | |
008873ea LZ |
1517 | cluster = min(cluster, max_cluster); |
1518 | } else { | |
1519 | cluster = max_cluster; | |
1520 | } | |
1521 | ||
008873ea LZ |
1522 | if (i + cluster > ra_index) { |
1523 | ra_index = max(i, ra_index); | |
0a52d108 | 1524 | if (ra) |
d3c0bab5 DS |
1525 | page_cache_sync_readahead(inode->i_mapping, ra, |
1526 | file, ra_index, cluster); | |
e4826a5b | 1527 | ra_index += cluster; |
008873ea | 1528 | } |
940100a4 | 1529 | |
64708539 | 1530 | btrfs_inode_lock(inode, 0); |
eede2bf3 OS |
1531 | if (IS_SWAPFILE(inode)) { |
1532 | ret = -ETXTBSY; | |
1533 | } else { | |
1534 | if (do_compress) | |
1535 | BTRFS_I(inode)->defrag_compress = compress_type; | |
1536 | ret = cluster_pages_for_defrag(inode, pages, i, cluster); | |
1537 | } | |
ecb8bea8 | 1538 | if (ret < 0) { |
64708539 | 1539 | btrfs_inode_unlock(inode, 0); |
4cb5300b | 1540 | goto out_ra; |
ecb8bea8 | 1541 | } |
4cb5300b CM |
1542 | |
1543 | defrag_count += ret; | |
d0e1d66b | 1544 | balance_dirty_pages_ratelimited(inode->i_mapping); |
64708539 | 1545 | btrfs_inode_unlock(inode, 0); |
4cb5300b CM |
1546 | |
1547 | if (newer_than) { | |
1548 | if (newer_off == (u64)-1) | |
1549 | break; | |
1550 | ||
e1f041e1 LB |
1551 | if (ret > 0) |
1552 | i += ret; | |
1553 | ||
4cb5300b | 1554 | newer_off = max(newer_off + 1, |
09cbfeaf | 1555 | (u64)i << PAGE_SHIFT); |
4cb5300b | 1556 | |
ee22184b BL |
1557 | ret = find_new_extents(root, inode, newer_than, |
1558 | &newer_off, SZ_64K); | |
4cb5300b CM |
1559 | if (!ret) { |
1560 | range->start = newer_off; | |
09cbfeaf | 1561 | i = (newer_off & new_align) >> PAGE_SHIFT; |
4cb5300b CM |
1562 | } else { |
1563 | break; | |
f46b5a66 | 1564 | } |
4cb5300b | 1565 | } else { |
008873ea | 1566 | if (ret > 0) { |
cbcc8326 | 1567 | i += ret; |
09cbfeaf | 1568 | last_len += ret << PAGE_SHIFT; |
008873ea | 1569 | } else { |
cbcc8326 | 1570 | i++; |
008873ea LZ |
1571 | last_len = 0; |
1572 | } | |
f46b5a66 | 1573 | } |
f46b5a66 CH |
1574 | } |
1575 | ||
50535db8 TT |
1576 | ret = defrag_count; |
1577 | error: | |
dec8ef90 | 1578 | if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) { |
1e701a32 | 1579 | filemap_flush(inode->i_mapping); |
dec8ef90 FM |
1580 | if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, |
1581 | &BTRFS_I(inode)->runtime_flags)) | |
1582 | filemap_flush(inode->i_mapping); | |
1583 | } | |
1e701a32 | 1584 | |
1a419d85 | 1585 | if (range->compress_type == BTRFS_COMPRESS_LZO) { |
0b246afa | 1586 | btrfs_set_fs_incompat(fs_info, COMPRESS_LZO); |
5c1aab1d NT |
1587 | } else if (range->compress_type == BTRFS_COMPRESS_ZSTD) { |
1588 | btrfs_set_fs_incompat(fs_info, COMPRESS_ZSTD); | |
1a419d85 LZ |
1589 | } |
1590 | ||
4cb5300b | 1591 | out_ra: |
1e2ef46d | 1592 | if (do_compress) { |
64708539 | 1593 | btrfs_inode_lock(inode, 0); |
eec63c65 | 1594 | BTRFS_I(inode)->defrag_compress = BTRFS_COMPRESS_NONE; |
64708539 | 1595 | btrfs_inode_unlock(inode, 0); |
633085c7 | 1596 | } |
4cb5300b CM |
1597 | if (!file) |
1598 | kfree(ra); | |
1599 | kfree(pages); | |
940100a4 | 1600 | return ret; |
f46b5a66 CH |
1601 | } |
1602 | ||
198605a8 | 1603 | static noinline int btrfs_ioctl_resize(struct file *file, |
76dda93c | 1604 | void __user *arg) |
f46b5a66 | 1605 | { |
0b246afa JM |
1606 | struct inode *inode = file_inode(file); |
1607 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
f46b5a66 CH |
1608 | u64 new_size; |
1609 | u64 old_size; | |
1610 | u64 devid = 1; | |
0b246afa | 1611 | struct btrfs_root *root = BTRFS_I(inode)->root; |
f46b5a66 CH |
1612 | struct btrfs_ioctl_vol_args *vol_args; |
1613 | struct btrfs_trans_handle *trans; | |
1614 | struct btrfs_device *device = NULL; | |
1615 | char *sizestr; | |
9a40f122 | 1616 | char *retptr; |
f46b5a66 CH |
1617 | char *devstr = NULL; |
1618 | int ret = 0; | |
f46b5a66 CH |
1619 | int mod = 0; |
1620 | ||
e441d54d CM |
1621 | if (!capable(CAP_SYS_ADMIN)) |
1622 | return -EPERM; | |
1623 | ||
198605a8 MX |
1624 | ret = mnt_want_write_file(file); |
1625 | if (ret) | |
1626 | return ret; | |
1627 | ||
c3e1f96c | 1628 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_RESIZE)) { |
97547676 | 1629 | mnt_drop_write_file(file); |
e57138b3 | 1630 | return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
c9e9f97b ID |
1631 | } |
1632 | ||
dae7b665 | 1633 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
1634 | if (IS_ERR(vol_args)) { |
1635 | ret = PTR_ERR(vol_args); | |
1636 | goto out; | |
1637 | } | |
5516e595 MF |
1638 | |
1639 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
f46b5a66 | 1640 | |
f46b5a66 CH |
1641 | sizestr = vol_args->name; |
1642 | devstr = strchr(sizestr, ':'); | |
1643 | if (devstr) { | |
f46b5a66 CH |
1644 | sizestr = devstr + 1; |
1645 | *devstr = '\0'; | |
1646 | devstr = vol_args->name; | |
58dfae63 Z |
1647 | ret = kstrtoull(devstr, 10, &devid); |
1648 | if (ret) | |
1649 | goto out_free; | |
dfd79829 MX |
1650 | if (!devid) { |
1651 | ret = -EINVAL; | |
1652 | goto out_free; | |
1653 | } | |
0b246afa | 1654 | btrfs_info(fs_info, "resizing devid %llu", devid); |
f46b5a66 | 1655 | } |
dba60f3f | 1656 | |
b2598edf | 1657 | device = btrfs_find_device(fs_info->fs_devices, devid, NULL, NULL); |
f46b5a66 | 1658 | if (!device) { |
0b246afa JM |
1659 | btrfs_info(fs_info, "resizer unable to find device %llu", |
1660 | devid); | |
dfd79829 | 1661 | ret = -ENODEV; |
c9e9f97b | 1662 | goto out_free; |
f46b5a66 | 1663 | } |
dba60f3f | 1664 | |
ebbede42 | 1665 | if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { |
0b246afa | 1666 | btrfs_info(fs_info, |
efe120a0 | 1667 | "resizer unable to apply on readonly device %llu", |
c1c9ff7c | 1668 | devid); |
dfd79829 | 1669 | ret = -EPERM; |
4e42ae1b LB |
1670 | goto out_free; |
1671 | } | |
1672 | ||
f46b5a66 CH |
1673 | if (!strcmp(sizestr, "max")) |
1674 | new_size = device->bdev->bd_inode->i_size; | |
1675 | else { | |
1676 | if (sizestr[0] == '-') { | |
1677 | mod = -1; | |
1678 | sizestr++; | |
1679 | } else if (sizestr[0] == '+') { | |
1680 | mod = 1; | |
1681 | sizestr++; | |
1682 | } | |
9a40f122 GH |
1683 | new_size = memparse(sizestr, &retptr); |
1684 | if (*retptr != '\0' || new_size == 0) { | |
f46b5a66 | 1685 | ret = -EINVAL; |
c9e9f97b | 1686 | goto out_free; |
f46b5a66 CH |
1687 | } |
1688 | } | |
1689 | ||
401e29c1 | 1690 | if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) { |
dfd79829 | 1691 | ret = -EPERM; |
63a212ab SB |
1692 | goto out_free; |
1693 | } | |
1694 | ||
7cc8e58d | 1695 | old_size = btrfs_device_get_total_bytes(device); |
f46b5a66 CH |
1696 | |
1697 | if (mod < 0) { | |
1698 | if (new_size > old_size) { | |
1699 | ret = -EINVAL; | |
c9e9f97b | 1700 | goto out_free; |
f46b5a66 CH |
1701 | } |
1702 | new_size = old_size - new_size; | |
1703 | } else if (mod > 0) { | |
eb8052e0 | 1704 | if (new_size > ULLONG_MAX - old_size) { |
902c68a4 | 1705 | ret = -ERANGE; |
eb8052e0 WF |
1706 | goto out_free; |
1707 | } | |
f46b5a66 CH |
1708 | new_size = old_size + new_size; |
1709 | } | |
1710 | ||
ee22184b | 1711 | if (new_size < SZ_256M) { |
f46b5a66 | 1712 | ret = -EINVAL; |
c9e9f97b | 1713 | goto out_free; |
f46b5a66 CH |
1714 | } |
1715 | if (new_size > device->bdev->bd_inode->i_size) { | |
1716 | ret = -EFBIG; | |
c9e9f97b | 1717 | goto out_free; |
f46b5a66 CH |
1718 | } |
1719 | ||
47f08b96 | 1720 | new_size = round_down(new_size, fs_info->sectorsize); |
f46b5a66 | 1721 | |
f46b5a66 | 1722 | if (new_size > old_size) { |
a22285a6 | 1723 | trans = btrfs_start_transaction(root, 0); |
98d5dc13 TI |
1724 | if (IS_ERR(trans)) { |
1725 | ret = PTR_ERR(trans); | |
c9e9f97b | 1726 | goto out_free; |
98d5dc13 | 1727 | } |
f46b5a66 | 1728 | ret = btrfs_grow_device(trans, device, new_size); |
3a45bb20 | 1729 | btrfs_commit_transaction(trans); |
ece7d20e | 1730 | } else if (new_size < old_size) { |
f46b5a66 | 1731 | ret = btrfs_shrink_device(device, new_size); |
0253f40e | 1732 | } /* equal, nothing need to do */ |
f46b5a66 | 1733 | |
faf8f7b9 MPS |
1734 | if (ret == 0 && new_size != old_size) |
1735 | btrfs_info_in_rcu(fs_info, | |
1736 | "resize device %s (devid %llu) from %llu to %llu", | |
1737 | rcu_str_deref(device->name), device->devid, | |
1738 | old_size, new_size); | |
c9e9f97b | 1739 | out_free: |
f46b5a66 | 1740 | kfree(vol_args); |
c9e9f97b | 1741 | out: |
c3e1f96c | 1742 | btrfs_exclop_finish(fs_info); |
18f39c41 | 1743 | mnt_drop_write_file(file); |
f46b5a66 CH |
1744 | return ret; |
1745 | } | |
1746 | ||
5d54c67e | 1747 | static noinline int __btrfs_ioctl_snap_create(struct file *file, |
52f75f4f | 1748 | const char *name, unsigned long fd, int subvol, |
5d54c67e | 1749 | bool readonly, |
8696c533 | 1750 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 1751 | { |
f46b5a66 | 1752 | int namelen; |
3de4586c | 1753 | int ret = 0; |
f46b5a66 | 1754 | |
325c50e3 JM |
1755 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1756 | return -ENOTDIR; | |
1757 | ||
a874a63e LB |
1758 | ret = mnt_want_write_file(file); |
1759 | if (ret) | |
1760 | goto out; | |
1761 | ||
72fd032e SW |
1762 | namelen = strlen(name); |
1763 | if (strchr(name, '/')) { | |
f46b5a66 | 1764 | ret = -EINVAL; |
a874a63e | 1765 | goto out_drop_write; |
f46b5a66 CH |
1766 | } |
1767 | ||
16780cab CM |
1768 | if (name[0] == '.' && |
1769 | (namelen == 1 || (name[1] == '.' && namelen == 2))) { | |
1770 | ret = -EEXIST; | |
a874a63e | 1771 | goto out_drop_write; |
16780cab CM |
1772 | } |
1773 | ||
3de4586c | 1774 | if (subvol) { |
72fd032e | 1775 | ret = btrfs_mksubvol(&file->f_path, name, namelen, |
5d54c67e | 1776 | NULL, readonly, inherit); |
cb8e7090 | 1777 | } else { |
2903ff01 | 1778 | struct fd src = fdget(fd); |
3de4586c | 1779 | struct inode *src_inode; |
2903ff01 | 1780 | if (!src.file) { |
3de4586c | 1781 | ret = -EINVAL; |
a874a63e | 1782 | goto out_drop_write; |
3de4586c CM |
1783 | } |
1784 | ||
496ad9aa AV |
1785 | src_inode = file_inode(src.file); |
1786 | if (src_inode->i_sb != file_inode(file)->i_sb) { | |
c79b4713 | 1787 | btrfs_info(BTRFS_I(file_inode(file))->root->fs_info, |
efe120a0 | 1788 | "Snapshot src from another FS"); |
23ad5b17 | 1789 | ret = -EXDEV; |
21cb47be | 1790 | } else if (!inode_owner_or_capable(&init_user_ns, src_inode)) { |
d0242061 DS |
1791 | /* |
1792 | * Subvolume creation is not restricted, but snapshots | |
1793 | * are limited to own subvolumes only | |
1794 | */ | |
1795 | ret = -EPERM; | |
ecd18815 | 1796 | } else { |
c11fbb6e | 1797 | ret = btrfs_mksnapshot(&file->f_path, name, namelen, |
ecd18815 | 1798 | BTRFS_I(src_inode)->root, |
5d54c67e | 1799 | readonly, inherit); |
3de4586c | 1800 | } |
2903ff01 | 1801 | fdput(src); |
cb8e7090 | 1802 | } |
a874a63e LB |
1803 | out_drop_write: |
1804 | mnt_drop_write_file(file); | |
f46b5a66 | 1805 | out: |
72fd032e SW |
1806 | return ret; |
1807 | } | |
1808 | ||
1809 | static noinline int btrfs_ioctl_snap_create(struct file *file, | |
fa0d2b9b | 1810 | void __user *arg, int subvol) |
72fd032e | 1811 | { |
fa0d2b9b | 1812 | struct btrfs_ioctl_vol_args *vol_args; |
72fd032e SW |
1813 | int ret; |
1814 | ||
325c50e3 JM |
1815 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1816 | return -ENOTDIR; | |
1817 | ||
fa0d2b9b LZ |
1818 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1819 | if (IS_ERR(vol_args)) | |
1820 | return PTR_ERR(vol_args); | |
1821 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
72fd032e | 1822 | |
5d54c67e NB |
1823 | ret = __btrfs_ioctl_snap_create(file, vol_args->name, vol_args->fd, |
1824 | subvol, false, NULL); | |
fdfb1e4f | 1825 | |
fa0d2b9b LZ |
1826 | kfree(vol_args); |
1827 | return ret; | |
1828 | } | |
fdfb1e4f | 1829 | |
fa0d2b9b LZ |
1830 | static noinline int btrfs_ioctl_snap_create_v2(struct file *file, |
1831 | void __user *arg, int subvol) | |
1832 | { | |
1833 | struct btrfs_ioctl_vol_args_v2 *vol_args; | |
1834 | int ret; | |
b83cc969 | 1835 | bool readonly = false; |
6f72c7e2 | 1836 | struct btrfs_qgroup_inherit *inherit = NULL; |
75eaa0e2 | 1837 | |
325c50e3 JM |
1838 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1839 | return -ENOTDIR; | |
1840 | ||
fa0d2b9b LZ |
1841 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1842 | if (IS_ERR(vol_args)) | |
1843 | return PTR_ERR(vol_args); | |
1844 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
75eaa0e2 | 1845 | |
673990db | 1846 | if (vol_args->flags & ~BTRFS_SUBVOL_CREATE_ARGS_MASK) { |
b83cc969 | 1847 | ret = -EOPNOTSUPP; |
c47ca32d | 1848 | goto free_args; |
72fd032e | 1849 | } |
fa0d2b9b | 1850 | |
b83cc969 LZ |
1851 | if (vol_args->flags & BTRFS_SUBVOL_RDONLY) |
1852 | readonly = true; | |
6f72c7e2 | 1853 | if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) { |
5011c5a6 DC |
1854 | u64 nums; |
1855 | ||
1856 | if (vol_args->size < sizeof(*inherit) || | |
1857 | vol_args->size > PAGE_SIZE) { | |
6f72c7e2 | 1858 | ret = -EINVAL; |
c47ca32d | 1859 | goto free_args; |
6f72c7e2 AJ |
1860 | } |
1861 | inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size); | |
1862 | if (IS_ERR(inherit)) { | |
1863 | ret = PTR_ERR(inherit); | |
c47ca32d | 1864 | goto free_args; |
6f72c7e2 | 1865 | } |
5011c5a6 DC |
1866 | |
1867 | if (inherit->num_qgroups > PAGE_SIZE || | |
1868 | inherit->num_ref_copies > PAGE_SIZE || | |
1869 | inherit->num_excl_copies > PAGE_SIZE) { | |
1870 | ret = -EINVAL; | |
1871 | goto free_inherit; | |
1872 | } | |
1873 | ||
1874 | nums = inherit->num_qgroups + 2 * inherit->num_ref_copies + | |
1875 | 2 * inherit->num_excl_copies; | |
1876 | if (vol_args->size != struct_size(inherit, qgroups, nums)) { | |
1877 | ret = -EINVAL; | |
1878 | goto free_inherit; | |
1879 | } | |
6f72c7e2 | 1880 | } |
fa0d2b9b | 1881 | |
5d54c67e NB |
1882 | ret = __btrfs_ioctl_snap_create(file, vol_args->name, vol_args->fd, |
1883 | subvol, readonly, inherit); | |
c47ca32d DC |
1884 | if (ret) |
1885 | goto free_inherit; | |
c47ca32d | 1886 | free_inherit: |
6f72c7e2 | 1887 | kfree(inherit); |
c47ca32d DC |
1888 | free_args: |
1889 | kfree(vol_args); | |
f46b5a66 CH |
1890 | return ret; |
1891 | } | |
1892 | ||
0caa102d LZ |
1893 | static noinline int btrfs_ioctl_subvol_getflags(struct file *file, |
1894 | void __user *arg) | |
1895 | { | |
496ad9aa | 1896 | struct inode *inode = file_inode(file); |
0b246afa | 1897 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
0caa102d LZ |
1898 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1899 | int ret = 0; | |
1900 | u64 flags = 0; | |
1901 | ||
4a0cc7ca | 1902 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) |
0caa102d LZ |
1903 | return -EINVAL; |
1904 | ||
0b246afa | 1905 | down_read(&fs_info->subvol_sem); |
0caa102d LZ |
1906 | if (btrfs_root_readonly(root)) |
1907 | flags |= BTRFS_SUBVOL_RDONLY; | |
0b246afa | 1908 | up_read(&fs_info->subvol_sem); |
0caa102d LZ |
1909 | |
1910 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
1911 | ret = -EFAULT; | |
1912 | ||
1913 | return ret; | |
1914 | } | |
1915 | ||
1916 | static noinline int btrfs_ioctl_subvol_setflags(struct file *file, | |
1917 | void __user *arg) | |
1918 | { | |
496ad9aa | 1919 | struct inode *inode = file_inode(file); |
0b246afa | 1920 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
0caa102d LZ |
1921 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1922 | struct btrfs_trans_handle *trans; | |
1923 | u64 root_flags; | |
1924 | u64 flags; | |
1925 | int ret = 0; | |
1926 | ||
21cb47be | 1927 | if (!inode_owner_or_capable(&init_user_ns, inode)) |
bd60ea0f DS |
1928 | return -EPERM; |
1929 | ||
b9ca0664 LB |
1930 | ret = mnt_want_write_file(file); |
1931 | if (ret) | |
1932 | goto out; | |
0caa102d | 1933 | |
4a0cc7ca | 1934 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
b9ca0664 LB |
1935 | ret = -EINVAL; |
1936 | goto out_drop_write; | |
1937 | } | |
0caa102d | 1938 | |
b9ca0664 LB |
1939 | if (copy_from_user(&flags, arg, sizeof(flags))) { |
1940 | ret = -EFAULT; | |
1941 | goto out_drop_write; | |
1942 | } | |
0caa102d | 1943 | |
b9ca0664 LB |
1944 | if (flags & ~BTRFS_SUBVOL_RDONLY) { |
1945 | ret = -EOPNOTSUPP; | |
1946 | goto out_drop_write; | |
1947 | } | |
0caa102d | 1948 | |
0b246afa | 1949 | down_write(&fs_info->subvol_sem); |
0caa102d LZ |
1950 | |
1951 | /* nothing to do */ | |
1952 | if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root)) | |
b9ca0664 | 1953 | goto out_drop_sem; |
0caa102d LZ |
1954 | |
1955 | root_flags = btrfs_root_flags(&root->root_item); | |
2c686537 | 1956 | if (flags & BTRFS_SUBVOL_RDONLY) { |
0caa102d LZ |
1957 | btrfs_set_root_flags(&root->root_item, |
1958 | root_flags | BTRFS_ROOT_SUBVOL_RDONLY); | |
2c686537 DS |
1959 | } else { |
1960 | /* | |
1961 | * Block RO -> RW transition if this subvolume is involved in | |
1962 | * send | |
1963 | */ | |
1964 | spin_lock(&root->root_item_lock); | |
1965 | if (root->send_in_progress == 0) { | |
1966 | btrfs_set_root_flags(&root->root_item, | |
0caa102d | 1967 | root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY); |
2c686537 DS |
1968 | spin_unlock(&root->root_item_lock); |
1969 | } else { | |
1970 | spin_unlock(&root->root_item_lock); | |
0b246afa JM |
1971 | btrfs_warn(fs_info, |
1972 | "Attempt to set subvolume %llu read-write during send", | |
1973 | root->root_key.objectid); | |
2c686537 DS |
1974 | ret = -EPERM; |
1975 | goto out_drop_sem; | |
1976 | } | |
1977 | } | |
0caa102d LZ |
1978 | |
1979 | trans = btrfs_start_transaction(root, 1); | |
1980 | if (IS_ERR(trans)) { | |
1981 | ret = PTR_ERR(trans); | |
1982 | goto out_reset; | |
1983 | } | |
1984 | ||
0b246afa | 1985 | ret = btrfs_update_root(trans, fs_info->tree_root, |
0caa102d | 1986 | &root->root_key, &root->root_item); |
9417ebc8 NB |
1987 | if (ret < 0) { |
1988 | btrfs_end_transaction(trans); | |
1989 | goto out_reset; | |
1990 | } | |
1991 | ||
1992 | ret = btrfs_commit_transaction(trans); | |
0caa102d | 1993 | |
0caa102d LZ |
1994 | out_reset: |
1995 | if (ret) | |
1996 | btrfs_set_root_flags(&root->root_item, root_flags); | |
b9ca0664 | 1997 | out_drop_sem: |
0b246afa | 1998 | up_write(&fs_info->subvol_sem); |
b9ca0664 LB |
1999 | out_drop_write: |
2000 | mnt_drop_write_file(file); | |
2001 | out: | |
0caa102d LZ |
2002 | return ret; |
2003 | } | |
2004 | ||
ac8e9819 CM |
2005 | static noinline int key_in_sk(struct btrfs_key *key, |
2006 | struct btrfs_ioctl_search_key *sk) | |
2007 | { | |
abc6e134 CM |
2008 | struct btrfs_key test; |
2009 | int ret; | |
2010 | ||
2011 | test.objectid = sk->min_objectid; | |
2012 | test.type = sk->min_type; | |
2013 | test.offset = sk->min_offset; | |
2014 | ||
2015 | ret = btrfs_comp_cpu_keys(key, &test); | |
2016 | if (ret < 0) | |
ac8e9819 | 2017 | return 0; |
abc6e134 CM |
2018 | |
2019 | test.objectid = sk->max_objectid; | |
2020 | test.type = sk->max_type; | |
2021 | test.offset = sk->max_offset; | |
2022 | ||
2023 | ret = btrfs_comp_cpu_keys(key, &test); | |
2024 | if (ret > 0) | |
ac8e9819 CM |
2025 | return 0; |
2026 | return 1; | |
2027 | } | |
2028 | ||
df397565 | 2029 | static noinline int copy_to_sk(struct btrfs_path *path, |
ac8e9819 CM |
2030 | struct btrfs_key *key, |
2031 | struct btrfs_ioctl_search_key *sk, | |
9b6e817d | 2032 | size_t *buf_size, |
ba346b35 | 2033 | char __user *ubuf, |
ac8e9819 CM |
2034 | unsigned long *sk_offset, |
2035 | int *num_found) | |
2036 | { | |
2037 | u64 found_transid; | |
2038 | struct extent_buffer *leaf; | |
2039 | struct btrfs_ioctl_search_header sh; | |
dd81d459 | 2040 | struct btrfs_key test; |
ac8e9819 CM |
2041 | unsigned long item_off; |
2042 | unsigned long item_len; | |
2043 | int nritems; | |
2044 | int i; | |
2045 | int slot; | |
ac8e9819 CM |
2046 | int ret = 0; |
2047 | ||
2048 | leaf = path->nodes[0]; | |
2049 | slot = path->slots[0]; | |
2050 | nritems = btrfs_header_nritems(leaf); | |
2051 | ||
2052 | if (btrfs_header_generation(leaf) > sk->max_transid) { | |
2053 | i = nritems; | |
2054 | goto advance_key; | |
2055 | } | |
2056 | found_transid = btrfs_header_generation(leaf); | |
2057 | ||
2058 | for (i = slot; i < nritems; i++) { | |
2059 | item_off = btrfs_item_ptr_offset(leaf, i); | |
2060 | item_len = btrfs_item_size_nr(leaf, i); | |
2061 | ||
03b71c6c GP |
2062 | btrfs_item_key_to_cpu(leaf, key, i); |
2063 | if (!key_in_sk(key, sk)) | |
2064 | continue; | |
2065 | ||
9b6e817d | 2066 | if (sizeof(sh) + item_len > *buf_size) { |
8f5f6178 GH |
2067 | if (*num_found) { |
2068 | ret = 1; | |
2069 | goto out; | |
2070 | } | |
2071 | ||
2072 | /* | |
2073 | * return one empty item back for v1, which does not | |
2074 | * handle -EOVERFLOW | |
2075 | */ | |
2076 | ||
9b6e817d | 2077 | *buf_size = sizeof(sh) + item_len; |
ac8e9819 | 2078 | item_len = 0; |
8f5f6178 GH |
2079 | ret = -EOVERFLOW; |
2080 | } | |
ac8e9819 | 2081 | |
9b6e817d | 2082 | if (sizeof(sh) + item_len + *sk_offset > *buf_size) { |
ac8e9819 | 2083 | ret = 1; |
25c9bc2e | 2084 | goto out; |
ac8e9819 CM |
2085 | } |
2086 | ||
ac8e9819 CM |
2087 | sh.objectid = key->objectid; |
2088 | sh.offset = key->offset; | |
2089 | sh.type = key->type; | |
2090 | sh.len = item_len; | |
2091 | sh.transid = found_transid; | |
2092 | ||
a48b73ec JB |
2093 | /* |
2094 | * Copy search result header. If we fault then loop again so we | |
2095 | * can fault in the pages and -EFAULT there if there's a | |
2096 | * problem. Otherwise we'll fault and then copy the buffer in | |
2097 | * properly this next time through | |
2098 | */ | |
2099 | if (copy_to_user_nofault(ubuf + *sk_offset, &sh, sizeof(sh))) { | |
2100 | ret = 0; | |
ba346b35 GH |
2101 | goto out; |
2102 | } | |
2103 | ||
ac8e9819 CM |
2104 | *sk_offset += sizeof(sh); |
2105 | ||
2106 | if (item_len) { | |
ba346b35 | 2107 | char __user *up = ubuf + *sk_offset; |
a48b73ec JB |
2108 | /* |
2109 | * Copy the item, same behavior as above, but reset the | |
2110 | * * sk_offset so we copy the full thing again. | |
2111 | */ | |
2112 | if (read_extent_buffer_to_user_nofault(leaf, up, | |
2113 | item_off, item_len)) { | |
2114 | ret = 0; | |
2115 | *sk_offset -= sizeof(sh); | |
ba346b35 GH |
2116 | goto out; |
2117 | } | |
2118 | ||
ac8e9819 | 2119 | *sk_offset += item_len; |
ac8e9819 | 2120 | } |
e2156867 | 2121 | (*num_found)++; |
ac8e9819 | 2122 | |
8f5f6178 GH |
2123 | if (ret) /* -EOVERFLOW from above */ |
2124 | goto out; | |
2125 | ||
25c9bc2e GH |
2126 | if (*num_found >= sk->nr_items) { |
2127 | ret = 1; | |
2128 | goto out; | |
2129 | } | |
ac8e9819 CM |
2130 | } |
2131 | advance_key: | |
abc6e134 | 2132 | ret = 0; |
dd81d459 NA |
2133 | test.objectid = sk->max_objectid; |
2134 | test.type = sk->max_type; | |
2135 | test.offset = sk->max_offset; | |
2136 | if (btrfs_comp_cpu_keys(key, &test) >= 0) | |
2137 | ret = 1; | |
2138 | else if (key->offset < (u64)-1) | |
ac8e9819 | 2139 | key->offset++; |
dd81d459 | 2140 | else if (key->type < (u8)-1) { |
abc6e134 | 2141 | key->offset = 0; |
ac8e9819 | 2142 | key->type++; |
dd81d459 | 2143 | } else if (key->objectid < (u64)-1) { |
abc6e134 CM |
2144 | key->offset = 0; |
2145 | key->type = 0; | |
ac8e9819 | 2146 | key->objectid++; |
abc6e134 CM |
2147 | } else |
2148 | ret = 1; | |
25c9bc2e | 2149 | out: |
ba346b35 GH |
2150 | /* |
2151 | * 0: all items from this leaf copied, continue with next | |
2152 | * 1: * more items can be copied, but unused buffer is too small | |
2153 | * * all items were found | |
2154 | * Either way, it will stops the loop which iterates to the next | |
2155 | * leaf | |
2156 | * -EOVERFLOW: item was to large for buffer | |
2157 | * -EFAULT: could not copy extent buffer back to userspace | |
2158 | */ | |
ac8e9819 CM |
2159 | return ret; |
2160 | } | |
2161 | ||
2162 | static noinline int search_ioctl(struct inode *inode, | |
12544442 | 2163 | struct btrfs_ioctl_search_key *sk, |
9b6e817d | 2164 | size_t *buf_size, |
ba346b35 | 2165 | char __user *ubuf) |
ac8e9819 | 2166 | { |
0b246afa | 2167 | struct btrfs_fs_info *info = btrfs_sb(inode->i_sb); |
ac8e9819 CM |
2168 | struct btrfs_root *root; |
2169 | struct btrfs_key key; | |
ac8e9819 | 2170 | struct btrfs_path *path; |
ac8e9819 CM |
2171 | int ret; |
2172 | int num_found = 0; | |
2173 | unsigned long sk_offset = 0; | |
2174 | ||
9b6e817d GH |
2175 | if (*buf_size < sizeof(struct btrfs_ioctl_search_header)) { |
2176 | *buf_size = sizeof(struct btrfs_ioctl_search_header); | |
12544442 | 2177 | return -EOVERFLOW; |
9b6e817d | 2178 | } |
12544442 | 2179 | |
ac8e9819 CM |
2180 | path = btrfs_alloc_path(); |
2181 | if (!path) | |
2182 | return -ENOMEM; | |
2183 | ||
2184 | if (sk->tree_id == 0) { | |
2185 | /* search the root of the inode that was passed */ | |
00246528 | 2186 | root = btrfs_grab_root(BTRFS_I(inode)->root); |
ac8e9819 | 2187 | } else { |
56e9357a | 2188 | root = btrfs_get_fs_root(info, sk->tree_id, true); |
ac8e9819 | 2189 | if (IS_ERR(root)) { |
ac8e9819 | 2190 | btrfs_free_path(path); |
ad1e3d56 | 2191 | return PTR_ERR(root); |
ac8e9819 CM |
2192 | } |
2193 | } | |
2194 | ||
2195 | key.objectid = sk->min_objectid; | |
2196 | key.type = sk->min_type; | |
2197 | key.offset = sk->min_offset; | |
2198 | ||
67871254 | 2199 | while (1) { |
1c78544e FM |
2200 | ret = fault_in_pages_writeable(ubuf + sk_offset, |
2201 | *buf_size - sk_offset); | |
a48b73ec JB |
2202 | if (ret) |
2203 | break; | |
2204 | ||
6174d3cb | 2205 | ret = btrfs_search_forward(root, &key, path, sk->min_transid); |
ac8e9819 CM |
2206 | if (ret != 0) { |
2207 | if (ret > 0) | |
2208 | ret = 0; | |
2209 | goto err; | |
2210 | } | |
df397565 | 2211 | ret = copy_to_sk(path, &key, sk, buf_size, ubuf, |
ac8e9819 | 2212 | &sk_offset, &num_found); |
b3b4aa74 | 2213 | btrfs_release_path(path); |
25c9bc2e | 2214 | if (ret) |
ac8e9819 CM |
2215 | break; |
2216 | ||
2217 | } | |
8f5f6178 GH |
2218 | if (ret > 0) |
2219 | ret = 0; | |
ac8e9819 CM |
2220 | err: |
2221 | sk->nr_items = num_found; | |
00246528 | 2222 | btrfs_put_root(root); |
ac8e9819 CM |
2223 | btrfs_free_path(path); |
2224 | return ret; | |
2225 | } | |
2226 | ||
2227 | static noinline int btrfs_ioctl_tree_search(struct file *file, | |
2228 | void __user *argp) | |
2229 | { | |
ba346b35 GH |
2230 | struct btrfs_ioctl_search_args __user *uargs; |
2231 | struct btrfs_ioctl_search_key sk; | |
9b6e817d GH |
2232 | struct inode *inode; |
2233 | int ret; | |
2234 | size_t buf_size; | |
ac8e9819 CM |
2235 | |
2236 | if (!capable(CAP_SYS_ADMIN)) | |
2237 | return -EPERM; | |
2238 | ||
ba346b35 GH |
2239 | uargs = (struct btrfs_ioctl_search_args __user *)argp; |
2240 | ||
2241 | if (copy_from_user(&sk, &uargs->key, sizeof(sk))) | |
2242 | return -EFAULT; | |
ac8e9819 | 2243 | |
ba346b35 | 2244 | buf_size = sizeof(uargs->buf); |
ac8e9819 | 2245 | |
496ad9aa | 2246 | inode = file_inode(file); |
ba346b35 | 2247 | ret = search_ioctl(inode, &sk, &buf_size, uargs->buf); |
8f5f6178 GH |
2248 | |
2249 | /* | |
2250 | * In the origin implementation an overflow is handled by returning a | |
2251 | * search header with a len of zero, so reset ret. | |
2252 | */ | |
2253 | if (ret == -EOVERFLOW) | |
2254 | ret = 0; | |
2255 | ||
ba346b35 | 2256 | if (ret == 0 && copy_to_user(&uargs->key, &sk, sizeof(sk))) |
ac8e9819 | 2257 | ret = -EFAULT; |
ac8e9819 CM |
2258 | return ret; |
2259 | } | |
2260 | ||
cc68a8a5 GH |
2261 | static noinline int btrfs_ioctl_tree_search_v2(struct file *file, |
2262 | void __user *argp) | |
2263 | { | |
2264 | struct btrfs_ioctl_search_args_v2 __user *uarg; | |
2265 | struct btrfs_ioctl_search_args_v2 args; | |
2266 | struct inode *inode; | |
2267 | int ret; | |
2268 | size_t buf_size; | |
ee22184b | 2269 | const size_t buf_limit = SZ_16M; |
cc68a8a5 GH |
2270 | |
2271 | if (!capable(CAP_SYS_ADMIN)) | |
2272 | return -EPERM; | |
2273 | ||
2274 | /* copy search header and buffer size */ | |
2275 | uarg = (struct btrfs_ioctl_search_args_v2 __user *)argp; | |
2276 | if (copy_from_user(&args, uarg, sizeof(args))) | |
2277 | return -EFAULT; | |
2278 | ||
2279 | buf_size = args.buf_size; | |
2280 | ||
cc68a8a5 GH |
2281 | /* limit result size to 16MB */ |
2282 | if (buf_size > buf_limit) | |
2283 | buf_size = buf_limit; | |
2284 | ||
2285 | inode = file_inode(file); | |
2286 | ret = search_ioctl(inode, &args.key, &buf_size, | |
718dc5fa | 2287 | (char __user *)(&uarg->buf[0])); |
cc68a8a5 GH |
2288 | if (ret == 0 && copy_to_user(&uarg->key, &args.key, sizeof(args.key))) |
2289 | ret = -EFAULT; | |
2290 | else if (ret == -EOVERFLOW && | |
2291 | copy_to_user(&uarg->buf_size, &buf_size, sizeof(buf_size))) | |
2292 | ret = -EFAULT; | |
2293 | ||
ac8e9819 CM |
2294 | return ret; |
2295 | } | |
2296 | ||
98d377a0 | 2297 | /* |
ac8e9819 CM |
2298 | * Search INODE_REFs to identify path name of 'dirid' directory |
2299 | * in a 'tree_id' tree. and sets path name to 'name'. | |
2300 | */ | |
98d377a0 TH |
2301 | static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, |
2302 | u64 tree_id, u64 dirid, char *name) | |
2303 | { | |
2304 | struct btrfs_root *root; | |
2305 | struct btrfs_key key; | |
ac8e9819 | 2306 | char *ptr; |
98d377a0 TH |
2307 | int ret = -1; |
2308 | int slot; | |
2309 | int len; | |
2310 | int total_len = 0; | |
2311 | struct btrfs_inode_ref *iref; | |
2312 | struct extent_buffer *l; | |
2313 | struct btrfs_path *path; | |
2314 | ||
2315 | if (dirid == BTRFS_FIRST_FREE_OBJECTID) { | |
2316 | name[0]='\0'; | |
2317 | return 0; | |
2318 | } | |
2319 | ||
2320 | path = btrfs_alloc_path(); | |
2321 | if (!path) | |
2322 | return -ENOMEM; | |
2323 | ||
c8bcbfbd | 2324 | ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX - 1]; |
98d377a0 | 2325 | |
56e9357a | 2326 | root = btrfs_get_fs_root(info, tree_id, true); |
98d377a0 | 2327 | if (IS_ERR(root)) { |
ad1e3d56 | 2328 | ret = PTR_ERR(root); |
88234012 JB |
2329 | root = NULL; |
2330 | goto out; | |
2331 | } | |
98d377a0 TH |
2332 | |
2333 | key.objectid = dirid; | |
2334 | key.type = BTRFS_INODE_REF_KEY; | |
8ad6fcab | 2335 | key.offset = (u64)-1; |
98d377a0 | 2336 | |
67871254 | 2337 | while (1) { |
98d377a0 TH |
2338 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); |
2339 | if (ret < 0) | |
2340 | goto out; | |
18674c6c FDBM |
2341 | else if (ret > 0) { |
2342 | ret = btrfs_previous_item(root, path, dirid, | |
2343 | BTRFS_INODE_REF_KEY); | |
2344 | if (ret < 0) | |
2345 | goto out; | |
2346 | else if (ret > 0) { | |
2347 | ret = -ENOENT; | |
2348 | goto out; | |
2349 | } | |
2350 | } | |
98d377a0 TH |
2351 | |
2352 | l = path->nodes[0]; | |
2353 | slot = path->slots[0]; | |
2354 | btrfs_item_key_to_cpu(l, &key, slot); | |
2355 | ||
98d377a0 TH |
2356 | iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); |
2357 | len = btrfs_inode_ref_name_len(l, iref); | |
2358 | ptr -= len + 1; | |
2359 | total_len += len + 1; | |
a696cf35 FDBM |
2360 | if (ptr < name) { |
2361 | ret = -ENAMETOOLONG; | |
98d377a0 | 2362 | goto out; |
a696cf35 | 2363 | } |
98d377a0 TH |
2364 | |
2365 | *(ptr + len) = '/'; | |
67871254 | 2366 | read_extent_buffer(l, ptr, (unsigned long)(iref + 1), len); |
98d377a0 TH |
2367 | |
2368 | if (key.offset == BTRFS_FIRST_FREE_OBJECTID) | |
2369 | break; | |
2370 | ||
b3b4aa74 | 2371 | btrfs_release_path(path); |
98d377a0 | 2372 | key.objectid = key.offset; |
8ad6fcab | 2373 | key.offset = (u64)-1; |
98d377a0 | 2374 | dirid = key.objectid; |
98d377a0 | 2375 | } |
77906a50 | 2376 | memmove(name, ptr, total_len); |
67871254 | 2377 | name[total_len] = '\0'; |
98d377a0 TH |
2378 | ret = 0; |
2379 | out: | |
00246528 | 2380 | btrfs_put_root(root); |
98d377a0 | 2381 | btrfs_free_path(path); |
ac8e9819 CM |
2382 | return ret; |
2383 | } | |
2384 | ||
23d0b79d TM |
2385 | static int btrfs_search_path_in_tree_user(struct inode *inode, |
2386 | struct btrfs_ioctl_ino_lookup_user_args *args) | |
2387 | { | |
2388 | struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info; | |
2389 | struct super_block *sb = inode->i_sb; | |
2390 | struct btrfs_key upper_limit = BTRFS_I(inode)->location; | |
2391 | u64 treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2392 | u64 dirid = args->dirid; | |
2393 | unsigned long item_off; | |
2394 | unsigned long item_len; | |
2395 | struct btrfs_inode_ref *iref; | |
2396 | struct btrfs_root_ref *rref; | |
b8a49ae1 | 2397 | struct btrfs_root *root = NULL; |
23d0b79d TM |
2398 | struct btrfs_path *path; |
2399 | struct btrfs_key key, key2; | |
2400 | struct extent_buffer *leaf; | |
2401 | struct inode *temp_inode; | |
2402 | char *ptr; | |
2403 | int slot; | |
2404 | int len; | |
2405 | int total_len = 0; | |
2406 | int ret; | |
2407 | ||
2408 | path = btrfs_alloc_path(); | |
2409 | if (!path) | |
2410 | return -ENOMEM; | |
2411 | ||
2412 | /* | |
2413 | * If the bottom subvolume does not exist directly under upper_limit, | |
2414 | * construct the path in from the bottom up. | |
2415 | */ | |
2416 | if (dirid != upper_limit.objectid) { | |
2417 | ptr = &args->path[BTRFS_INO_LOOKUP_USER_PATH_MAX - 1]; | |
2418 | ||
56e9357a | 2419 | root = btrfs_get_fs_root(fs_info, treeid, true); |
23d0b79d TM |
2420 | if (IS_ERR(root)) { |
2421 | ret = PTR_ERR(root); | |
2422 | goto out; | |
2423 | } | |
2424 | ||
2425 | key.objectid = dirid; | |
2426 | key.type = BTRFS_INODE_REF_KEY; | |
2427 | key.offset = (u64)-1; | |
2428 | while (1) { | |
2429 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2430 | if (ret < 0) { | |
b8a49ae1 | 2431 | goto out_put; |
23d0b79d TM |
2432 | } else if (ret > 0) { |
2433 | ret = btrfs_previous_item(root, path, dirid, | |
2434 | BTRFS_INODE_REF_KEY); | |
2435 | if (ret < 0) { | |
b8a49ae1 | 2436 | goto out_put; |
23d0b79d TM |
2437 | } else if (ret > 0) { |
2438 | ret = -ENOENT; | |
b8a49ae1 | 2439 | goto out_put; |
23d0b79d TM |
2440 | } |
2441 | } | |
2442 | ||
2443 | leaf = path->nodes[0]; | |
2444 | slot = path->slots[0]; | |
2445 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2446 | ||
2447 | iref = btrfs_item_ptr(leaf, slot, struct btrfs_inode_ref); | |
2448 | len = btrfs_inode_ref_name_len(leaf, iref); | |
2449 | ptr -= len + 1; | |
2450 | total_len += len + 1; | |
2451 | if (ptr < args->path) { | |
2452 | ret = -ENAMETOOLONG; | |
b8a49ae1 | 2453 | goto out_put; |
23d0b79d TM |
2454 | } |
2455 | ||
2456 | *(ptr + len) = '/'; | |
2457 | read_extent_buffer(leaf, ptr, | |
2458 | (unsigned long)(iref + 1), len); | |
2459 | ||
2460 | /* Check the read+exec permission of this directory */ | |
2461 | ret = btrfs_previous_item(root, path, dirid, | |
2462 | BTRFS_INODE_ITEM_KEY); | |
2463 | if (ret < 0) { | |
b8a49ae1 | 2464 | goto out_put; |
23d0b79d TM |
2465 | } else if (ret > 0) { |
2466 | ret = -ENOENT; | |
b8a49ae1 | 2467 | goto out_put; |
23d0b79d TM |
2468 | } |
2469 | ||
2470 | leaf = path->nodes[0]; | |
2471 | slot = path->slots[0]; | |
2472 | btrfs_item_key_to_cpu(leaf, &key2, slot); | |
2473 | if (key2.objectid != dirid) { | |
2474 | ret = -ENOENT; | |
b8a49ae1 | 2475 | goto out_put; |
23d0b79d TM |
2476 | } |
2477 | ||
0202e83f | 2478 | temp_inode = btrfs_iget(sb, key2.objectid, root); |
3ca57bd6 MT |
2479 | if (IS_ERR(temp_inode)) { |
2480 | ret = PTR_ERR(temp_inode); | |
b8a49ae1 | 2481 | goto out_put; |
3ca57bd6 | 2482 | } |
47291baa CB |
2483 | ret = inode_permission(&init_user_ns, temp_inode, |
2484 | MAY_READ | MAY_EXEC); | |
23d0b79d TM |
2485 | iput(temp_inode); |
2486 | if (ret) { | |
2487 | ret = -EACCES; | |
b8a49ae1 | 2488 | goto out_put; |
23d0b79d TM |
2489 | } |
2490 | ||
2491 | if (key.offset == upper_limit.objectid) | |
2492 | break; | |
2493 | if (key.objectid == BTRFS_FIRST_FREE_OBJECTID) { | |
2494 | ret = -EACCES; | |
b8a49ae1 | 2495 | goto out_put; |
23d0b79d TM |
2496 | } |
2497 | ||
2498 | btrfs_release_path(path); | |
2499 | key.objectid = key.offset; | |
2500 | key.offset = (u64)-1; | |
2501 | dirid = key.objectid; | |
2502 | } | |
2503 | ||
2504 | memmove(args->path, ptr, total_len); | |
2505 | args->path[total_len] = '\0'; | |
00246528 | 2506 | btrfs_put_root(root); |
b8a49ae1 | 2507 | root = NULL; |
23d0b79d TM |
2508 | btrfs_release_path(path); |
2509 | } | |
2510 | ||
2511 | /* Get the bottom subvolume's name from ROOT_REF */ | |
23d0b79d TM |
2512 | key.objectid = treeid; |
2513 | key.type = BTRFS_ROOT_REF_KEY; | |
2514 | key.offset = args->treeid; | |
b8a49ae1 | 2515 | ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); |
23d0b79d TM |
2516 | if (ret < 0) { |
2517 | goto out; | |
2518 | } else if (ret > 0) { | |
2519 | ret = -ENOENT; | |
2520 | goto out; | |
2521 | } | |
2522 | ||
2523 | leaf = path->nodes[0]; | |
2524 | slot = path->slots[0]; | |
2525 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2526 | ||
2527 | item_off = btrfs_item_ptr_offset(leaf, slot); | |
2528 | item_len = btrfs_item_size_nr(leaf, slot); | |
2529 | /* Check if dirid in ROOT_REF corresponds to passed dirid */ | |
2530 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2531 | if (args->dirid != btrfs_root_ref_dirid(leaf, rref)) { | |
2532 | ret = -EINVAL; | |
2533 | goto out; | |
2534 | } | |
2535 | ||
2536 | /* Copy subvolume's name */ | |
2537 | item_off += sizeof(struct btrfs_root_ref); | |
2538 | item_len -= sizeof(struct btrfs_root_ref); | |
2539 | read_extent_buffer(leaf, args->name, item_off, item_len); | |
2540 | args->name[item_len] = 0; | |
2541 | ||
b8a49ae1 | 2542 | out_put: |
00246528 | 2543 | btrfs_put_root(root); |
23d0b79d TM |
2544 | out: |
2545 | btrfs_free_path(path); | |
2546 | return ret; | |
2547 | } | |
2548 | ||
ac8e9819 CM |
2549 | static noinline int btrfs_ioctl_ino_lookup(struct file *file, |
2550 | void __user *argp) | |
2551 | { | |
bece2e82 BVA |
2552 | struct btrfs_ioctl_ino_lookup_args *args; |
2553 | struct inode *inode; | |
01b810b8 | 2554 | int ret = 0; |
ac8e9819 | 2555 | |
2354d08f JL |
2556 | args = memdup_user(argp, sizeof(*args)); |
2557 | if (IS_ERR(args)) | |
2558 | return PTR_ERR(args); | |
c2b96929 | 2559 | |
496ad9aa | 2560 | inode = file_inode(file); |
ac8e9819 | 2561 | |
01b810b8 DS |
2562 | /* |
2563 | * Unprivileged query to obtain the containing subvolume root id. The | |
2564 | * path is reset so it's consistent with btrfs_search_path_in_tree. | |
2565 | */ | |
1b53ac4d CM |
2566 | if (args->treeid == 0) |
2567 | args->treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2568 | ||
01b810b8 DS |
2569 | if (args->objectid == BTRFS_FIRST_FREE_OBJECTID) { |
2570 | args->name[0] = 0; | |
2571 | goto out; | |
2572 | } | |
2573 | ||
2574 | if (!capable(CAP_SYS_ADMIN)) { | |
2575 | ret = -EPERM; | |
2576 | goto out; | |
2577 | } | |
2578 | ||
ac8e9819 CM |
2579 | ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, |
2580 | args->treeid, args->objectid, | |
2581 | args->name); | |
2582 | ||
01b810b8 | 2583 | out: |
ac8e9819 CM |
2584 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) |
2585 | ret = -EFAULT; | |
2586 | ||
2587 | kfree(args); | |
98d377a0 TH |
2588 | return ret; |
2589 | } | |
2590 | ||
23d0b79d TM |
2591 | /* |
2592 | * Version of ino_lookup ioctl (unprivileged) | |
2593 | * | |
2594 | * The main differences from ino_lookup ioctl are: | |
2595 | * | |
2596 | * 1. Read + Exec permission will be checked using inode_permission() during | |
2597 | * path construction. -EACCES will be returned in case of failure. | |
2598 | * 2. Path construction will be stopped at the inode number which corresponds | |
2599 | * to the fd with which this ioctl is called. If constructed path does not | |
2600 | * exist under fd's inode, -EACCES will be returned. | |
2601 | * 3. The name of bottom subvolume is also searched and filled. | |
2602 | */ | |
2603 | static int btrfs_ioctl_ino_lookup_user(struct file *file, void __user *argp) | |
2604 | { | |
2605 | struct btrfs_ioctl_ino_lookup_user_args *args; | |
2606 | struct inode *inode; | |
2607 | int ret; | |
2608 | ||
2609 | args = memdup_user(argp, sizeof(*args)); | |
2610 | if (IS_ERR(args)) | |
2611 | return PTR_ERR(args); | |
2612 | ||
2613 | inode = file_inode(file); | |
2614 | ||
2615 | if (args->dirid == BTRFS_FIRST_FREE_OBJECTID && | |
2616 | BTRFS_I(inode)->location.objectid != BTRFS_FIRST_FREE_OBJECTID) { | |
2617 | /* | |
2618 | * The subvolume does not exist under fd with which this is | |
2619 | * called | |
2620 | */ | |
2621 | kfree(args); | |
2622 | return -EACCES; | |
2623 | } | |
2624 | ||
2625 | ret = btrfs_search_path_in_tree_user(inode, args); | |
2626 | ||
2627 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
2628 | ret = -EFAULT; | |
2629 | ||
2630 | kfree(args); | |
2631 | return ret; | |
2632 | } | |
2633 | ||
b64ec075 TM |
2634 | /* Get the subvolume information in BTRFS_ROOT_ITEM and BTRFS_ROOT_BACKREF */ |
2635 | static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp) | |
2636 | { | |
2637 | struct btrfs_ioctl_get_subvol_info_args *subvol_info; | |
2638 | struct btrfs_fs_info *fs_info; | |
2639 | struct btrfs_root *root; | |
2640 | struct btrfs_path *path; | |
2641 | struct btrfs_key key; | |
2642 | struct btrfs_root_item *root_item; | |
2643 | struct btrfs_root_ref *rref; | |
2644 | struct extent_buffer *leaf; | |
2645 | unsigned long item_off; | |
2646 | unsigned long item_len; | |
2647 | struct inode *inode; | |
2648 | int slot; | |
2649 | int ret = 0; | |
2650 | ||
2651 | path = btrfs_alloc_path(); | |
2652 | if (!path) | |
2653 | return -ENOMEM; | |
2654 | ||
2655 | subvol_info = kzalloc(sizeof(*subvol_info), GFP_KERNEL); | |
2656 | if (!subvol_info) { | |
2657 | btrfs_free_path(path); | |
2658 | return -ENOMEM; | |
2659 | } | |
2660 | ||
2661 | inode = file_inode(file); | |
2662 | fs_info = BTRFS_I(inode)->root->fs_info; | |
2663 | ||
2664 | /* Get root_item of inode's subvolume */ | |
2665 | key.objectid = BTRFS_I(inode)->root->root_key.objectid; | |
56e9357a | 2666 | root = btrfs_get_fs_root(fs_info, key.objectid, true); |
b64ec075 TM |
2667 | if (IS_ERR(root)) { |
2668 | ret = PTR_ERR(root); | |
04734e84 JB |
2669 | goto out_free; |
2670 | } | |
b64ec075 TM |
2671 | root_item = &root->root_item; |
2672 | ||
2673 | subvol_info->treeid = key.objectid; | |
2674 | ||
2675 | subvol_info->generation = btrfs_root_generation(root_item); | |
2676 | subvol_info->flags = btrfs_root_flags(root_item); | |
2677 | ||
2678 | memcpy(subvol_info->uuid, root_item->uuid, BTRFS_UUID_SIZE); | |
2679 | memcpy(subvol_info->parent_uuid, root_item->parent_uuid, | |
2680 | BTRFS_UUID_SIZE); | |
2681 | memcpy(subvol_info->received_uuid, root_item->received_uuid, | |
2682 | BTRFS_UUID_SIZE); | |
2683 | ||
2684 | subvol_info->ctransid = btrfs_root_ctransid(root_item); | |
2685 | subvol_info->ctime.sec = btrfs_stack_timespec_sec(&root_item->ctime); | |
2686 | subvol_info->ctime.nsec = btrfs_stack_timespec_nsec(&root_item->ctime); | |
2687 | ||
2688 | subvol_info->otransid = btrfs_root_otransid(root_item); | |
2689 | subvol_info->otime.sec = btrfs_stack_timespec_sec(&root_item->otime); | |
2690 | subvol_info->otime.nsec = btrfs_stack_timespec_nsec(&root_item->otime); | |
2691 | ||
2692 | subvol_info->stransid = btrfs_root_stransid(root_item); | |
2693 | subvol_info->stime.sec = btrfs_stack_timespec_sec(&root_item->stime); | |
2694 | subvol_info->stime.nsec = btrfs_stack_timespec_nsec(&root_item->stime); | |
2695 | ||
2696 | subvol_info->rtransid = btrfs_root_rtransid(root_item); | |
2697 | subvol_info->rtime.sec = btrfs_stack_timespec_sec(&root_item->rtime); | |
2698 | subvol_info->rtime.nsec = btrfs_stack_timespec_nsec(&root_item->rtime); | |
2699 | ||
2700 | if (key.objectid != BTRFS_FS_TREE_OBJECTID) { | |
2701 | /* Search root tree for ROOT_BACKREF of this subvolume */ | |
b64ec075 TM |
2702 | key.type = BTRFS_ROOT_BACKREF_KEY; |
2703 | key.offset = 0; | |
04734e84 | 2704 | ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); |
b64ec075 TM |
2705 | if (ret < 0) { |
2706 | goto out; | |
2707 | } else if (path->slots[0] >= | |
2708 | btrfs_header_nritems(path->nodes[0])) { | |
04734e84 | 2709 | ret = btrfs_next_leaf(fs_info->tree_root, path); |
b64ec075 TM |
2710 | if (ret < 0) { |
2711 | goto out; | |
2712 | } else if (ret > 0) { | |
2713 | ret = -EUCLEAN; | |
2714 | goto out; | |
2715 | } | |
2716 | } | |
2717 | ||
2718 | leaf = path->nodes[0]; | |
2719 | slot = path->slots[0]; | |
2720 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2721 | if (key.objectid == subvol_info->treeid && | |
2722 | key.type == BTRFS_ROOT_BACKREF_KEY) { | |
2723 | subvol_info->parent_id = key.offset; | |
2724 | ||
2725 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2726 | subvol_info->dirid = btrfs_root_ref_dirid(leaf, rref); | |
2727 | ||
2728 | item_off = btrfs_item_ptr_offset(leaf, slot) | |
2729 | + sizeof(struct btrfs_root_ref); | |
2730 | item_len = btrfs_item_size_nr(leaf, slot) | |
2731 | - sizeof(struct btrfs_root_ref); | |
2732 | read_extent_buffer(leaf, subvol_info->name, | |
2733 | item_off, item_len); | |
2734 | } else { | |
2735 | ret = -ENOENT; | |
2736 | goto out; | |
2737 | } | |
2738 | } | |
2739 | ||
2740 | if (copy_to_user(argp, subvol_info, sizeof(*subvol_info))) | |
2741 | ret = -EFAULT; | |
2742 | ||
2743 | out: | |
00246528 | 2744 | btrfs_put_root(root); |
04734e84 | 2745 | out_free: |
b64ec075 | 2746 | btrfs_free_path(path); |
b091f7fe | 2747 | kfree(subvol_info); |
b64ec075 TM |
2748 | return ret; |
2749 | } | |
2750 | ||
42e4b520 TM |
2751 | /* |
2752 | * Return ROOT_REF information of the subvolume containing this inode | |
2753 | * except the subvolume name. | |
2754 | */ | |
2755 | static int btrfs_ioctl_get_subvol_rootref(struct file *file, void __user *argp) | |
2756 | { | |
2757 | struct btrfs_ioctl_get_subvol_rootref_args *rootrefs; | |
2758 | struct btrfs_root_ref *rref; | |
2759 | struct btrfs_root *root; | |
2760 | struct btrfs_path *path; | |
2761 | struct btrfs_key key; | |
2762 | struct extent_buffer *leaf; | |
2763 | struct inode *inode; | |
2764 | u64 objectid; | |
2765 | int slot; | |
2766 | int ret; | |
2767 | u8 found; | |
2768 | ||
2769 | path = btrfs_alloc_path(); | |
2770 | if (!path) | |
2771 | return -ENOMEM; | |
2772 | ||
2773 | rootrefs = memdup_user(argp, sizeof(*rootrefs)); | |
2774 | if (IS_ERR(rootrefs)) { | |
2775 | btrfs_free_path(path); | |
2776 | return PTR_ERR(rootrefs); | |
2777 | } | |
2778 | ||
2779 | inode = file_inode(file); | |
2780 | root = BTRFS_I(inode)->root->fs_info->tree_root; | |
2781 | objectid = BTRFS_I(inode)->root->root_key.objectid; | |
2782 | ||
2783 | key.objectid = objectid; | |
2784 | key.type = BTRFS_ROOT_REF_KEY; | |
2785 | key.offset = rootrefs->min_treeid; | |
2786 | found = 0; | |
2787 | ||
2788 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2789 | if (ret < 0) { | |
2790 | goto out; | |
2791 | } else if (path->slots[0] >= | |
2792 | btrfs_header_nritems(path->nodes[0])) { | |
2793 | ret = btrfs_next_leaf(root, path); | |
2794 | if (ret < 0) { | |
2795 | goto out; | |
2796 | } else if (ret > 0) { | |
2797 | ret = -EUCLEAN; | |
2798 | goto out; | |
2799 | } | |
2800 | } | |
2801 | while (1) { | |
2802 | leaf = path->nodes[0]; | |
2803 | slot = path->slots[0]; | |
2804 | ||
2805 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2806 | if (key.objectid != objectid || key.type != BTRFS_ROOT_REF_KEY) { | |
2807 | ret = 0; | |
2808 | goto out; | |
2809 | } | |
2810 | ||
2811 | if (found == BTRFS_MAX_ROOTREF_BUFFER_NUM) { | |
2812 | ret = -EOVERFLOW; | |
2813 | goto out; | |
2814 | } | |
2815 | ||
2816 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2817 | rootrefs->rootref[found].treeid = key.offset; | |
2818 | rootrefs->rootref[found].dirid = | |
2819 | btrfs_root_ref_dirid(leaf, rref); | |
2820 | found++; | |
2821 | ||
2822 | ret = btrfs_next_item(root, path); | |
2823 | if (ret < 0) { | |
2824 | goto out; | |
2825 | } else if (ret > 0) { | |
2826 | ret = -EUCLEAN; | |
2827 | goto out; | |
2828 | } | |
2829 | } | |
2830 | ||
2831 | out: | |
2832 | if (!ret || ret == -EOVERFLOW) { | |
2833 | rootrefs->num_items = found; | |
2834 | /* update min_treeid for next search */ | |
2835 | if (found) | |
2836 | rootrefs->min_treeid = | |
2837 | rootrefs->rootref[found - 1].treeid + 1; | |
2838 | if (copy_to_user(argp, rootrefs, sizeof(*rootrefs))) | |
2839 | ret = -EFAULT; | |
2840 | } | |
2841 | ||
2842 | kfree(rootrefs); | |
2843 | btrfs_free_path(path); | |
2844 | ||
2845 | return ret; | |
2846 | } | |
2847 | ||
76dda93c | 2848 | static noinline int btrfs_ioctl_snap_destroy(struct file *file, |
949964c9 MPS |
2849 | void __user *arg, |
2850 | bool destroy_v2) | |
76dda93c | 2851 | { |
54563d41 | 2852 | struct dentry *parent = file->f_path.dentry; |
0b246afa | 2853 | struct btrfs_fs_info *fs_info = btrfs_sb(parent->d_sb); |
76dda93c | 2854 | struct dentry *dentry; |
2b0143b5 | 2855 | struct inode *dir = d_inode(parent); |
76dda93c YZ |
2856 | struct inode *inode; |
2857 | struct btrfs_root *root = BTRFS_I(dir)->root; | |
2858 | struct btrfs_root *dest = NULL; | |
949964c9 MPS |
2859 | struct btrfs_ioctl_vol_args *vol_args = NULL; |
2860 | struct btrfs_ioctl_vol_args_v2 *vol_args2 = NULL; | |
2861 | char *subvol_name, *subvol_name_ptr = NULL; | |
2862 | int subvol_namelen; | |
76dda93c | 2863 | int err = 0; |
949964c9 | 2864 | bool destroy_parent = false; |
76dda93c | 2865 | |
949964c9 MPS |
2866 | if (destroy_v2) { |
2867 | vol_args2 = memdup_user(arg, sizeof(*vol_args2)); | |
2868 | if (IS_ERR(vol_args2)) | |
2869 | return PTR_ERR(vol_args2); | |
325c50e3 | 2870 | |
949964c9 MPS |
2871 | if (vol_args2->flags & ~BTRFS_SUBVOL_DELETE_ARGS_MASK) { |
2872 | err = -EOPNOTSUPP; | |
2873 | goto out; | |
2874 | } | |
76dda93c | 2875 | |
949964c9 MPS |
2876 | /* |
2877 | * If SPEC_BY_ID is not set, we are looking for the subvolume by | |
2878 | * name, same as v1 currently does. | |
2879 | */ | |
2880 | if (!(vol_args2->flags & BTRFS_SUBVOL_SPEC_BY_ID)) { | |
2881 | vol_args2->name[BTRFS_SUBVOL_NAME_MAX] = 0; | |
2882 | subvol_name = vol_args2->name; | |
2883 | ||
2884 | err = mnt_want_write_file(file); | |
2885 | if (err) | |
2886 | goto out; | |
2887 | } else { | |
2888 | if (vol_args2->subvolid < BTRFS_FIRST_FREE_OBJECTID) { | |
2889 | err = -EINVAL; | |
2890 | goto out; | |
2891 | } | |
2892 | ||
2893 | err = mnt_want_write_file(file); | |
2894 | if (err) | |
2895 | goto out; | |
2896 | ||
2897 | dentry = btrfs_get_dentry(fs_info->sb, | |
2898 | BTRFS_FIRST_FREE_OBJECTID, | |
2899 | vol_args2->subvolid, 0, 0); | |
2900 | if (IS_ERR(dentry)) { | |
2901 | err = PTR_ERR(dentry); | |
2902 | goto out_drop_write; | |
2903 | } | |
2904 | ||
2905 | /* | |
2906 | * Change the default parent since the subvolume being | |
2907 | * deleted can be outside of the current mount point. | |
2908 | */ | |
2909 | parent = btrfs_get_parent(dentry); | |
2910 | ||
2911 | /* | |
2912 | * At this point dentry->d_name can point to '/' if the | |
2913 | * subvolume we want to destroy is outsite of the | |
2914 | * current mount point, so we need to release the | |
2915 | * current dentry and execute the lookup to return a new | |
2916 | * one with ->d_name pointing to the | |
2917 | * <mount point>/subvol_name. | |
2918 | */ | |
2919 | dput(dentry); | |
2920 | if (IS_ERR(parent)) { | |
2921 | err = PTR_ERR(parent); | |
2922 | goto out_drop_write; | |
2923 | } | |
2924 | dir = d_inode(parent); | |
2925 | ||
2926 | /* | |
2927 | * If v2 was used with SPEC_BY_ID, a new parent was | |
2928 | * allocated since the subvolume can be outside of the | |
2929 | * current mount point. Later on we need to release this | |
2930 | * new parent dentry. | |
2931 | */ | |
2932 | destroy_parent = true; | |
2933 | ||
2934 | subvol_name_ptr = btrfs_get_subvol_name_from_objectid( | |
2935 | fs_info, vol_args2->subvolid); | |
2936 | if (IS_ERR(subvol_name_ptr)) { | |
2937 | err = PTR_ERR(subvol_name_ptr); | |
2938 | goto free_parent; | |
2939 | } | |
2940 | /* subvol_name_ptr is already NULL termined */ | |
2941 | subvol_name = (char *)kbasename(subvol_name_ptr); | |
2942 | } | |
2943 | } else { | |
2944 | vol_args = memdup_user(arg, sizeof(*vol_args)); | |
2945 | if (IS_ERR(vol_args)) | |
2946 | return PTR_ERR(vol_args); | |
2947 | ||
2948 | vol_args->name[BTRFS_PATH_NAME_MAX] = 0; | |
2949 | subvol_name = vol_args->name; | |
2950 | ||
2951 | err = mnt_want_write_file(file); | |
2952 | if (err) | |
2953 | goto out; | |
76dda93c YZ |
2954 | } |
2955 | ||
949964c9 | 2956 | subvol_namelen = strlen(subvol_name); |
76dda93c | 2957 | |
949964c9 MPS |
2958 | if (strchr(subvol_name, '/') || |
2959 | strncmp(subvol_name, "..", subvol_namelen) == 0) { | |
2960 | err = -EINVAL; | |
2961 | goto free_subvol_name; | |
2962 | } | |
2963 | ||
2964 | if (!S_ISDIR(dir->i_mode)) { | |
2965 | err = -ENOTDIR; | |
2966 | goto free_subvol_name; | |
2967 | } | |
521e0546 | 2968 | |
00235411 AV |
2969 | err = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT); |
2970 | if (err == -EINTR) | |
949964c9 MPS |
2971 | goto free_subvol_name; |
2972 | dentry = lookup_one_len(subvol_name, parent, subvol_namelen); | |
76dda93c YZ |
2973 | if (IS_ERR(dentry)) { |
2974 | err = PTR_ERR(dentry); | |
2975 | goto out_unlock_dir; | |
2976 | } | |
2977 | ||
2b0143b5 | 2978 | if (d_really_is_negative(dentry)) { |
76dda93c YZ |
2979 | err = -ENOENT; |
2980 | goto out_dput; | |
2981 | } | |
2982 | ||
2b0143b5 | 2983 | inode = d_inode(dentry); |
4260f7c7 | 2984 | dest = BTRFS_I(inode)->root; |
67871254 | 2985 | if (!capable(CAP_SYS_ADMIN)) { |
4260f7c7 SW |
2986 | /* |
2987 | * Regular user. Only allow this with a special mount | |
2988 | * option, when the user has write+exec access to the | |
2989 | * subvol root, and when rmdir(2) would have been | |
2990 | * allowed. | |
2991 | * | |
2992 | * Note that this is _not_ check that the subvol is | |
2993 | * empty or doesn't contain data that we wouldn't | |
2994 | * otherwise be able to delete. | |
2995 | * | |
2996 | * Users who want to delete empty subvols should try | |
2997 | * rmdir(2). | |
2998 | */ | |
2999 | err = -EPERM; | |
0b246afa | 3000 | if (!btrfs_test_opt(fs_info, USER_SUBVOL_RM_ALLOWED)) |
4260f7c7 SW |
3001 | goto out_dput; |
3002 | ||
3003 | /* | |
3004 | * Do not allow deletion if the parent dir is the same | |
3005 | * as the dir to be deleted. That means the ioctl | |
3006 | * must be called on the dentry referencing the root | |
3007 | * of the subvol, not a random directory contained | |
3008 | * within it. | |
3009 | */ | |
3010 | err = -EINVAL; | |
3011 | if (root == dest) | |
3012 | goto out_dput; | |
3013 | ||
47291baa CB |
3014 | err = inode_permission(&init_user_ns, inode, |
3015 | MAY_WRITE | MAY_EXEC); | |
4260f7c7 SW |
3016 | if (err) |
3017 | goto out_dput; | |
4260f7c7 SW |
3018 | } |
3019 | ||
5c39da5b MX |
3020 | /* check if subvolume may be deleted by a user */ |
3021 | err = btrfs_may_delete(dir, dentry, 1); | |
3022 | if (err) | |
3023 | goto out_dput; | |
3024 | ||
4a0cc7ca | 3025 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
76dda93c YZ |
3026 | err = -EINVAL; |
3027 | goto out_dput; | |
3028 | } | |
3029 | ||
64708539 | 3030 | btrfs_inode_lock(inode, 0); |
f60a2364 | 3031 | err = btrfs_delete_subvolume(dir, dentry); |
64708539 | 3032 | btrfs_inode_unlock(inode, 0); |
46008d9d AG |
3033 | if (!err) { |
3034 | fsnotify_rmdir(dir, dentry); | |
76dda93c | 3035 | d_delete(dentry); |
46008d9d | 3036 | } |
fa6ac876 | 3037 | |
76dda93c YZ |
3038 | out_dput: |
3039 | dput(dentry); | |
3040 | out_unlock_dir: | |
64708539 | 3041 | btrfs_inode_unlock(dir, 0); |
949964c9 MPS |
3042 | free_subvol_name: |
3043 | kfree(subvol_name_ptr); | |
3044 | free_parent: | |
3045 | if (destroy_parent) | |
3046 | dput(parent); | |
00235411 | 3047 | out_drop_write: |
2a79f17e | 3048 | mnt_drop_write_file(file); |
76dda93c | 3049 | out: |
949964c9 | 3050 | kfree(vol_args2); |
76dda93c YZ |
3051 | kfree(vol_args); |
3052 | return err; | |
3053 | } | |
3054 | ||
1e701a32 | 3055 | static int btrfs_ioctl_defrag(struct file *file, void __user *argp) |
f46b5a66 | 3056 | { |
496ad9aa | 3057 | struct inode *inode = file_inode(file); |
f46b5a66 | 3058 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1e701a32 | 3059 | struct btrfs_ioctl_defrag_range_args *range; |
c146afad YZ |
3060 | int ret; |
3061 | ||
25122d15 ID |
3062 | ret = mnt_want_write_file(file); |
3063 | if (ret) | |
3064 | return ret; | |
b83cc969 | 3065 | |
25122d15 ID |
3066 | if (btrfs_root_readonly(root)) { |
3067 | ret = -EROFS; | |
3068 | goto out; | |
5ac00add | 3069 | } |
f46b5a66 CH |
3070 | |
3071 | switch (inode->i_mode & S_IFMT) { | |
3072 | case S_IFDIR: | |
e441d54d CM |
3073 | if (!capable(CAP_SYS_ADMIN)) { |
3074 | ret = -EPERM; | |
3075 | goto out; | |
3076 | } | |
de78b51a | 3077 | ret = btrfs_defrag_root(root); |
f46b5a66 CH |
3078 | break; |
3079 | case S_IFREG: | |
616d374e AB |
3080 | /* |
3081 | * Note that this does not check the file descriptor for write | |
3082 | * access. This prevents defragmenting executables that are | |
3083 | * running and allows defrag on files open in read-only mode. | |
3084 | */ | |
3085 | if (!capable(CAP_SYS_ADMIN) && | |
47291baa | 3086 | inode_permission(&init_user_ns, inode, MAY_WRITE)) { |
616d374e | 3087 | ret = -EPERM; |
e441d54d CM |
3088 | goto out; |
3089 | } | |
1e701a32 CM |
3090 | |
3091 | range = kzalloc(sizeof(*range), GFP_KERNEL); | |
3092 | if (!range) { | |
3093 | ret = -ENOMEM; | |
3094 | goto out; | |
3095 | } | |
3096 | ||
3097 | if (argp) { | |
3098 | if (copy_from_user(range, argp, | |
3099 | sizeof(*range))) { | |
3100 | ret = -EFAULT; | |
3101 | kfree(range); | |
683be16e | 3102 | goto out; |
1e701a32 CM |
3103 | } |
3104 | /* compression requires us to start the IO */ | |
3105 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
3106 | range->flags |= BTRFS_DEFRAG_RANGE_START_IO; | |
3107 | range->extent_thresh = (u32)-1; | |
3108 | } | |
3109 | } else { | |
3110 | /* the rest are all set to zero by kzalloc */ | |
3111 | range->len = (u64)-1; | |
3112 | } | |
496ad9aa | 3113 | ret = btrfs_defrag_file(file_inode(file), file, |
7c829b72 | 3114 | range, BTRFS_OLDEST_GENERATION, 0); |
4cb5300b CM |
3115 | if (ret > 0) |
3116 | ret = 0; | |
1e701a32 | 3117 | kfree(range); |
f46b5a66 | 3118 | break; |
8929ecfa YZ |
3119 | default: |
3120 | ret = -EINVAL; | |
f46b5a66 | 3121 | } |
e441d54d | 3122 | out: |
25122d15 | 3123 | mnt_drop_write_file(file); |
e441d54d | 3124 | return ret; |
f46b5a66 CH |
3125 | } |
3126 | ||
2ff7e61e | 3127 | static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg) |
f46b5a66 CH |
3128 | { |
3129 | struct btrfs_ioctl_vol_args *vol_args; | |
3130 | int ret; | |
3131 | ||
e441d54d CM |
3132 | if (!capable(CAP_SYS_ADMIN)) |
3133 | return -EPERM; | |
3134 | ||
c3e1f96c | 3135 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_ADD)) |
e57138b3 | 3136 | return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
c9e9f97b | 3137 | |
dae7b665 | 3138 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
3139 | if (IS_ERR(vol_args)) { |
3140 | ret = PTR_ERR(vol_args); | |
3141 | goto out; | |
3142 | } | |
f46b5a66 | 3143 | |
5516e595 | 3144 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
2ff7e61e | 3145 | ret = btrfs_init_new_device(fs_info, vol_args->name); |
f46b5a66 | 3146 | |
43d20761 | 3147 | if (!ret) |
0b246afa | 3148 | btrfs_info(fs_info, "disk added %s", vol_args->name); |
43d20761 | 3149 | |
f46b5a66 | 3150 | kfree(vol_args); |
c9e9f97b | 3151 | out: |
c3e1f96c | 3152 | btrfs_exclop_finish(fs_info); |
f46b5a66 CH |
3153 | return ret; |
3154 | } | |
3155 | ||
6b526ed7 | 3156 | static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg) |
f46b5a66 | 3157 | { |
0b246afa JM |
3158 | struct inode *inode = file_inode(file); |
3159 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
6b526ed7 | 3160 | struct btrfs_ioctl_vol_args_v2 *vol_args; |
f46b5a66 CH |
3161 | int ret; |
3162 | ||
e441d54d CM |
3163 | if (!capable(CAP_SYS_ADMIN)) |
3164 | return -EPERM; | |
3165 | ||
da24927b MX |
3166 | ret = mnt_want_write_file(file); |
3167 | if (ret) | |
3168 | return ret; | |
c146afad | 3169 | |
dae7b665 | 3170 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
3171 | if (IS_ERR(vol_args)) { |
3172 | ret = PTR_ERR(vol_args); | |
c47ca32d | 3173 | goto err_drop; |
c9e9f97b | 3174 | } |
f46b5a66 | 3175 | |
748449cd | 3176 | if (vol_args->flags & ~BTRFS_DEVICE_REMOVE_ARGS_MASK) { |
fd4e994b OS |
3177 | ret = -EOPNOTSUPP; |
3178 | goto out; | |
3179 | } | |
f46b5a66 | 3180 | |
c3e1f96c | 3181 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REMOVE)) { |
183860f6 AJ |
3182 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
3183 | goto out; | |
3184 | } | |
3185 | ||
735654ea | 3186 | if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) { |
2ff7e61e | 3187 | ret = btrfs_rm_device(fs_info, NULL, vol_args->devid); |
6b526ed7 AJ |
3188 | } else { |
3189 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
2ff7e61e | 3190 | ret = btrfs_rm_device(fs_info, vol_args->name, 0); |
6b526ed7 | 3191 | } |
c3e1f96c | 3192 | btrfs_exclop_finish(fs_info); |
183860f6 | 3193 | |
6b526ed7 | 3194 | if (!ret) { |
735654ea | 3195 | if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) |
0b246afa | 3196 | btrfs_info(fs_info, "device deleted: id %llu", |
6b526ed7 AJ |
3197 | vol_args->devid); |
3198 | else | |
0b246afa | 3199 | btrfs_info(fs_info, "device deleted: %s", |
6b526ed7 AJ |
3200 | vol_args->name); |
3201 | } | |
183860f6 AJ |
3202 | out: |
3203 | kfree(vol_args); | |
c47ca32d | 3204 | err_drop: |
4ac20c70 | 3205 | mnt_drop_write_file(file); |
f46b5a66 CH |
3206 | return ret; |
3207 | } | |
3208 | ||
da24927b | 3209 | static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg) |
f46b5a66 | 3210 | { |
0b246afa JM |
3211 | struct inode *inode = file_inode(file); |
3212 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
f46b5a66 CH |
3213 | struct btrfs_ioctl_vol_args *vol_args; |
3214 | int ret; | |
3215 | ||
e441d54d CM |
3216 | if (!capable(CAP_SYS_ADMIN)) |
3217 | return -EPERM; | |
3218 | ||
da24927b MX |
3219 | ret = mnt_want_write_file(file); |
3220 | if (ret) | |
3221 | return ret; | |
c146afad | 3222 | |
c3e1f96c | 3223 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REMOVE)) { |
183860f6 | 3224 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
58d7bbf8 DS |
3225 | goto out_drop_write; |
3226 | } | |
3227 | ||
3228 | vol_args = memdup_user(arg, sizeof(*vol_args)); | |
3229 | if (IS_ERR(vol_args)) { | |
3230 | ret = PTR_ERR(vol_args); | |
183860f6 AJ |
3231 | goto out; |
3232 | } | |
3233 | ||
58d7bbf8 | 3234 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
2ff7e61e | 3235 | ret = btrfs_rm_device(fs_info, vol_args->name, 0); |
183860f6 | 3236 | |
ec95d491 | 3237 | if (!ret) |
0b246afa | 3238 | btrfs_info(fs_info, "disk deleted %s", vol_args->name); |
183860f6 | 3239 | kfree(vol_args); |
58d7bbf8 | 3240 | out: |
c3e1f96c | 3241 | btrfs_exclop_finish(fs_info); |
58d7bbf8 | 3242 | out_drop_write: |
4ac20c70 | 3243 | mnt_drop_write_file(file); |
58d7bbf8 | 3244 | |
f46b5a66 CH |
3245 | return ret; |
3246 | } | |
3247 | ||
2ff7e61e JM |
3248 | static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info, |
3249 | void __user *arg) | |
475f6387 | 3250 | { |
027ed2f0 | 3251 | struct btrfs_ioctl_fs_info_args *fi_args; |
475f6387 | 3252 | struct btrfs_device *device; |
0b246afa | 3253 | struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; |
137c5418 | 3254 | u64 flags_in; |
027ed2f0 | 3255 | int ret = 0; |
475f6387 | 3256 | |
137c5418 JT |
3257 | fi_args = memdup_user(arg, sizeof(*fi_args)); |
3258 | if (IS_ERR(fi_args)) | |
3259 | return PTR_ERR(fi_args); | |
3260 | ||
3261 | flags_in = fi_args->flags; | |
3262 | memset(fi_args, 0, sizeof(*fi_args)); | |
027ed2f0 | 3263 | |
d03262c7 | 3264 | rcu_read_lock(); |
027ed2f0 | 3265 | fi_args->num_devices = fs_devices->num_devices; |
475f6387 | 3266 | |
d03262c7 | 3267 | list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) { |
027ed2f0 LZ |
3268 | if (device->devid > fi_args->max_id) |
3269 | fi_args->max_id = device->devid; | |
475f6387 | 3270 | } |
d03262c7 | 3271 | rcu_read_unlock(); |
475f6387 | 3272 | |
de37aa51 | 3273 | memcpy(&fi_args->fsid, fs_devices->fsid, sizeof(fi_args->fsid)); |
bea7eafd OS |
3274 | fi_args->nodesize = fs_info->nodesize; |
3275 | fi_args->sectorsize = fs_info->sectorsize; | |
3276 | fi_args->clone_alignment = fs_info->sectorsize; | |
80a773fb | 3277 | |
137c5418 JT |
3278 | if (flags_in & BTRFS_FS_INFO_FLAG_CSUM_INFO) { |
3279 | fi_args->csum_type = btrfs_super_csum_type(fs_info->super_copy); | |
3280 | fi_args->csum_size = btrfs_super_csum_size(fs_info->super_copy); | |
3281 | fi_args->flags |= BTRFS_FS_INFO_FLAG_CSUM_INFO; | |
3282 | } | |
3283 | ||
0fb408a5 JT |
3284 | if (flags_in & BTRFS_FS_INFO_FLAG_GENERATION) { |
3285 | fi_args->generation = fs_info->generation; | |
3286 | fi_args->flags |= BTRFS_FS_INFO_FLAG_GENERATION; | |
3287 | } | |
3288 | ||
49bac897 JT |
3289 | if (flags_in & BTRFS_FS_INFO_FLAG_METADATA_UUID) { |
3290 | memcpy(&fi_args->metadata_uuid, fs_devices->metadata_uuid, | |
3291 | sizeof(fi_args->metadata_uuid)); | |
3292 | fi_args->flags |= BTRFS_FS_INFO_FLAG_METADATA_UUID; | |
3293 | } | |
3294 | ||
027ed2f0 LZ |
3295 | if (copy_to_user(arg, fi_args, sizeof(*fi_args))) |
3296 | ret = -EFAULT; | |
475f6387 | 3297 | |
027ed2f0 LZ |
3298 | kfree(fi_args); |
3299 | return ret; | |
475f6387 JS |
3300 | } |
3301 | ||
2ff7e61e JM |
3302 | static long btrfs_ioctl_dev_info(struct btrfs_fs_info *fs_info, |
3303 | void __user *arg) | |
475f6387 JS |
3304 | { |
3305 | struct btrfs_ioctl_dev_info_args *di_args; | |
3306 | struct btrfs_device *dev; | |
475f6387 JS |
3307 | int ret = 0; |
3308 | char *s_uuid = NULL; | |
475f6387 | 3309 | |
475f6387 JS |
3310 | di_args = memdup_user(arg, sizeof(*di_args)); |
3311 | if (IS_ERR(di_args)) | |
3312 | return PTR_ERR(di_args); | |
3313 | ||
dd5f9615 | 3314 | if (!btrfs_is_empty_uuid(di_args->uuid)) |
475f6387 JS |
3315 | s_uuid = di_args->uuid; |
3316 | ||
c5593ca3 | 3317 | rcu_read_lock(); |
e4319cd9 | 3318 | dev = btrfs_find_device(fs_info->fs_devices, di_args->devid, s_uuid, |
b2598edf | 3319 | NULL); |
475f6387 JS |
3320 | |
3321 | if (!dev) { | |
3322 | ret = -ENODEV; | |
3323 | goto out; | |
3324 | } | |
3325 | ||
3326 | di_args->devid = dev->devid; | |
7cc8e58d MX |
3327 | di_args->bytes_used = btrfs_device_get_bytes_used(dev); |
3328 | di_args->total_bytes = btrfs_device_get_total_bytes(dev); | |
475f6387 | 3329 | memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid)); |
a27202fb | 3330 | if (dev->name) { |
672d5990 MT |
3331 | strncpy(di_args->path, rcu_str_deref(dev->name), |
3332 | sizeof(di_args->path) - 1); | |
a27202fb JM |
3333 | di_args->path[sizeof(di_args->path) - 1] = 0; |
3334 | } else { | |
99ba55ad | 3335 | di_args->path[0] = '\0'; |
a27202fb | 3336 | } |
475f6387 JS |
3337 | |
3338 | out: | |
c5593ca3 | 3339 | rcu_read_unlock(); |
475f6387 JS |
3340 | if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args))) |
3341 | ret = -EFAULT; | |
3342 | ||
3343 | kfree(di_args); | |
3344 | return ret; | |
3345 | } | |
3346 | ||
6ef5ed0d JB |
3347 | static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) |
3348 | { | |
496ad9aa | 3349 | struct inode *inode = file_inode(file); |
0b246afa | 3350 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
6ef5ed0d JB |
3351 | struct btrfs_root *root = BTRFS_I(inode)->root; |
3352 | struct btrfs_root *new_root; | |
3353 | struct btrfs_dir_item *di; | |
3354 | struct btrfs_trans_handle *trans; | |
2a2b5d62 | 3355 | struct btrfs_path *path = NULL; |
6ef5ed0d | 3356 | struct btrfs_disk_key disk_key; |
6ef5ed0d JB |
3357 | u64 objectid = 0; |
3358 | u64 dir_id; | |
3c04ce01 | 3359 | int ret; |
6ef5ed0d JB |
3360 | |
3361 | if (!capable(CAP_SYS_ADMIN)) | |
3362 | return -EPERM; | |
3363 | ||
3c04ce01 MX |
3364 | ret = mnt_want_write_file(file); |
3365 | if (ret) | |
3366 | return ret; | |
3367 | ||
3368 | if (copy_from_user(&objectid, argp, sizeof(objectid))) { | |
3369 | ret = -EFAULT; | |
3370 | goto out; | |
3371 | } | |
6ef5ed0d JB |
3372 | |
3373 | if (!objectid) | |
1cecf579 | 3374 | objectid = BTRFS_FS_TREE_OBJECTID; |
6ef5ed0d | 3375 | |
56e9357a | 3376 | new_root = btrfs_get_fs_root(fs_info, objectid, true); |
3c04ce01 MX |
3377 | if (IS_ERR(new_root)) { |
3378 | ret = PTR_ERR(new_root); | |
3379 | goto out; | |
3380 | } | |
2a2b5d62 JB |
3381 | if (!is_fstree(new_root->root_key.objectid)) { |
3382 | ret = -ENOENT; | |
3383 | goto out_free; | |
3384 | } | |
6ef5ed0d | 3385 | |
6ef5ed0d | 3386 | path = btrfs_alloc_path(); |
3c04ce01 MX |
3387 | if (!path) { |
3388 | ret = -ENOMEM; | |
2a2b5d62 | 3389 | goto out_free; |
3c04ce01 | 3390 | } |
6ef5ed0d JB |
3391 | |
3392 | trans = btrfs_start_transaction(root, 1); | |
98d5dc13 | 3393 | if (IS_ERR(trans)) { |
3c04ce01 | 3394 | ret = PTR_ERR(trans); |
2a2b5d62 | 3395 | goto out_free; |
6ef5ed0d JB |
3396 | } |
3397 | ||
0b246afa JM |
3398 | dir_id = btrfs_super_root_dir(fs_info->super_copy); |
3399 | di = btrfs_lookup_dir_item(trans, fs_info->tree_root, path, | |
6ef5ed0d | 3400 | dir_id, "default", 7, 1); |
cf1e99a4 | 3401 | if (IS_ERR_OR_NULL(di)) { |
2a2b5d62 | 3402 | btrfs_release_path(path); |
3a45bb20 | 3403 | btrfs_end_transaction(trans); |
0b246afa | 3404 | btrfs_err(fs_info, |
5d163e0e | 3405 | "Umm, you don't have the default diritem, this isn't going to work"); |
3c04ce01 | 3406 | ret = -ENOENT; |
2a2b5d62 | 3407 | goto out_free; |
6ef5ed0d JB |
3408 | } |
3409 | ||
3410 | btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key); | |
3411 | btrfs_set_dir_item_key(path->nodes[0], di, &disk_key); | |
3412 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
2a2b5d62 | 3413 | btrfs_release_path(path); |
6ef5ed0d | 3414 | |
0b246afa | 3415 | btrfs_set_fs_incompat(fs_info, DEFAULT_SUBVOL); |
3a45bb20 | 3416 | btrfs_end_transaction(trans); |
2a2b5d62 | 3417 | out_free: |
00246528 | 3418 | btrfs_put_root(new_root); |
2a2b5d62 | 3419 | btrfs_free_path(path); |
3c04ce01 MX |
3420 | out: |
3421 | mnt_drop_write_file(file); | |
3422 | return ret; | |
6ef5ed0d JB |
3423 | } |
3424 | ||
c065f5b1 SY |
3425 | static void get_block_group_info(struct list_head *groups_list, |
3426 | struct btrfs_ioctl_space_info *space) | |
bf5fc093 | 3427 | { |
32da5386 | 3428 | struct btrfs_block_group *block_group; |
bf5fc093 JB |
3429 | |
3430 | space->total_bytes = 0; | |
3431 | space->used_bytes = 0; | |
3432 | space->flags = 0; | |
3433 | list_for_each_entry(block_group, groups_list, list) { | |
3434 | space->flags = block_group->flags; | |
b3470b5d | 3435 | space->total_bytes += block_group->length; |
bf38be65 | 3436 | space->used_bytes += block_group->used; |
bf5fc093 JB |
3437 | } |
3438 | } | |
3439 | ||
2ff7e61e JM |
3440 | static long btrfs_ioctl_space_info(struct btrfs_fs_info *fs_info, |
3441 | void __user *arg) | |
1406e432 JB |
3442 | { |
3443 | struct btrfs_ioctl_space_args space_args; | |
3444 | struct btrfs_ioctl_space_info space; | |
3445 | struct btrfs_ioctl_space_info *dest; | |
7fde62bf | 3446 | struct btrfs_ioctl_space_info *dest_orig; |
13f2696f | 3447 | struct btrfs_ioctl_space_info __user *user_dest; |
1406e432 | 3448 | struct btrfs_space_info *info; |
315d8e98 CIK |
3449 | static const u64 types[] = { |
3450 | BTRFS_BLOCK_GROUP_DATA, | |
3451 | BTRFS_BLOCK_GROUP_SYSTEM, | |
3452 | BTRFS_BLOCK_GROUP_METADATA, | |
3453 | BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA | |
3454 | }; | |
bf5fc093 | 3455 | int num_types = 4; |
7fde62bf | 3456 | int alloc_size; |
1406e432 | 3457 | int ret = 0; |
51788b1b | 3458 | u64 slot_count = 0; |
bf5fc093 | 3459 | int i, c; |
1406e432 JB |
3460 | |
3461 | if (copy_from_user(&space_args, | |
3462 | (struct btrfs_ioctl_space_args __user *)arg, | |
3463 | sizeof(space_args))) | |
3464 | return -EFAULT; | |
3465 | ||
bf5fc093 JB |
3466 | for (i = 0; i < num_types; i++) { |
3467 | struct btrfs_space_info *tmp; | |
3468 | ||
3469 | info = NULL; | |
72804905 | 3470 | list_for_each_entry(tmp, &fs_info->space_info, list) { |
bf5fc093 JB |
3471 | if (tmp->flags == types[i]) { |
3472 | info = tmp; | |
3473 | break; | |
3474 | } | |
3475 | } | |
bf5fc093 JB |
3476 | |
3477 | if (!info) | |
3478 | continue; | |
3479 | ||
3480 | down_read(&info->groups_sem); | |
3481 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3482 | if (!list_empty(&info->block_groups[c])) | |
3483 | slot_count++; | |
3484 | } | |
3485 | up_read(&info->groups_sem); | |
3486 | } | |
7fde62bf | 3487 | |
36523e95 DS |
3488 | /* |
3489 | * Global block reserve, exported as a space_info | |
3490 | */ | |
3491 | slot_count++; | |
3492 | ||
7fde62bf CM |
3493 | /* space_slots == 0 means they are asking for a count */ |
3494 | if (space_args.space_slots == 0) { | |
3495 | space_args.total_spaces = slot_count; | |
3496 | goto out; | |
3497 | } | |
bf5fc093 | 3498 | |
51788b1b | 3499 | slot_count = min_t(u64, space_args.space_slots, slot_count); |
bf5fc093 | 3500 | |
7fde62bf | 3501 | alloc_size = sizeof(*dest) * slot_count; |
bf5fc093 | 3502 | |
7fde62bf CM |
3503 | /* we generally have at most 6 or so space infos, one for each raid |
3504 | * level. So, a whole page should be more than enough for everyone | |
3505 | */ | |
09cbfeaf | 3506 | if (alloc_size > PAGE_SIZE) |
7fde62bf CM |
3507 | return -ENOMEM; |
3508 | ||
1406e432 | 3509 | space_args.total_spaces = 0; |
8d2db785 | 3510 | dest = kmalloc(alloc_size, GFP_KERNEL); |
7fde62bf CM |
3511 | if (!dest) |
3512 | return -ENOMEM; | |
3513 | dest_orig = dest; | |
1406e432 | 3514 | |
7fde62bf | 3515 | /* now we have a buffer to copy into */ |
bf5fc093 JB |
3516 | for (i = 0; i < num_types; i++) { |
3517 | struct btrfs_space_info *tmp; | |
3518 | ||
51788b1b DR |
3519 | if (!slot_count) |
3520 | break; | |
3521 | ||
bf5fc093 | 3522 | info = NULL; |
72804905 | 3523 | list_for_each_entry(tmp, &fs_info->space_info, list) { |
bf5fc093 JB |
3524 | if (tmp->flags == types[i]) { |
3525 | info = tmp; | |
3526 | break; | |
3527 | } | |
3528 | } | |
7fde62bf | 3529 | |
bf5fc093 JB |
3530 | if (!info) |
3531 | continue; | |
3532 | down_read(&info->groups_sem); | |
3533 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3534 | if (!list_empty(&info->block_groups[c])) { | |
c065f5b1 SY |
3535 | get_block_group_info(&info->block_groups[c], |
3536 | &space); | |
bf5fc093 JB |
3537 | memcpy(dest, &space, sizeof(space)); |
3538 | dest++; | |
3539 | space_args.total_spaces++; | |
51788b1b | 3540 | slot_count--; |
bf5fc093 | 3541 | } |
51788b1b DR |
3542 | if (!slot_count) |
3543 | break; | |
bf5fc093 JB |
3544 | } |
3545 | up_read(&info->groups_sem); | |
1406e432 | 3546 | } |
1406e432 | 3547 | |
36523e95 DS |
3548 | /* |
3549 | * Add global block reserve | |
3550 | */ | |
3551 | if (slot_count) { | |
0b246afa | 3552 | struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv; |
36523e95 DS |
3553 | |
3554 | spin_lock(&block_rsv->lock); | |
3555 | space.total_bytes = block_rsv->size; | |
3556 | space.used_bytes = block_rsv->size - block_rsv->reserved; | |
3557 | spin_unlock(&block_rsv->lock); | |
3558 | space.flags = BTRFS_SPACE_INFO_GLOBAL_RSV; | |
3559 | memcpy(dest, &space, sizeof(space)); | |
3560 | space_args.total_spaces++; | |
3561 | } | |
3562 | ||
2eec6c81 | 3563 | user_dest = (struct btrfs_ioctl_space_info __user *) |
7fde62bf CM |
3564 | (arg + sizeof(struct btrfs_ioctl_space_args)); |
3565 | ||
3566 | if (copy_to_user(user_dest, dest_orig, alloc_size)) | |
3567 | ret = -EFAULT; | |
3568 | ||
3569 | kfree(dest_orig); | |
3570 | out: | |
3571 | if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args))) | |
1406e432 JB |
3572 | ret = -EFAULT; |
3573 | ||
3574 | return ret; | |
3575 | } | |
3576 | ||
9a8c28be MX |
3577 | static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root, |
3578 | void __user *argp) | |
46204592 | 3579 | { |
46204592 SW |
3580 | struct btrfs_trans_handle *trans; |
3581 | u64 transid; | |
db5b493a | 3582 | int ret; |
46204592 | 3583 | |
d4edf39b | 3584 | trans = btrfs_attach_transaction_barrier(root); |
ff7c1d33 MX |
3585 | if (IS_ERR(trans)) { |
3586 | if (PTR_ERR(trans) != -ENOENT) | |
3587 | return PTR_ERR(trans); | |
3588 | ||
3589 | /* No running transaction, don't bother */ | |
3590 | transid = root->fs_info->last_trans_committed; | |
3591 | goto out; | |
3592 | } | |
46204592 | 3593 | transid = trans->transid; |
3a45bb20 | 3594 | ret = btrfs_commit_transaction_async(trans, 0); |
8b2b2d3c | 3595 | if (ret) { |
3a45bb20 | 3596 | btrfs_end_transaction(trans); |
db5b493a | 3597 | return ret; |
8b2b2d3c | 3598 | } |
ff7c1d33 | 3599 | out: |
46204592 SW |
3600 | if (argp) |
3601 | if (copy_to_user(argp, &transid, sizeof(transid))) | |
3602 | return -EFAULT; | |
3603 | return 0; | |
3604 | } | |
3605 | ||
2ff7e61e | 3606 | static noinline long btrfs_ioctl_wait_sync(struct btrfs_fs_info *fs_info, |
9a8c28be | 3607 | void __user *argp) |
46204592 | 3608 | { |
46204592 SW |
3609 | u64 transid; |
3610 | ||
3611 | if (argp) { | |
3612 | if (copy_from_user(&transid, argp, sizeof(transid))) | |
3613 | return -EFAULT; | |
3614 | } else { | |
3615 | transid = 0; /* current trans */ | |
3616 | } | |
2ff7e61e | 3617 | return btrfs_wait_for_commit(fs_info, transid); |
46204592 SW |
3618 | } |
3619 | ||
b8e95489 | 3620 | static long btrfs_ioctl_scrub(struct file *file, void __user *arg) |
475f6387 | 3621 | { |
0b246afa | 3622 | struct btrfs_fs_info *fs_info = btrfs_sb(file_inode(file)->i_sb); |
475f6387 | 3623 | struct btrfs_ioctl_scrub_args *sa; |
b8e95489 | 3624 | int ret; |
475f6387 JS |
3625 | |
3626 | if (!capable(CAP_SYS_ADMIN)) | |
3627 | return -EPERM; | |
3628 | ||
3629 | sa = memdup_user(arg, sizeof(*sa)); | |
3630 | if (IS_ERR(sa)) | |
3631 | return PTR_ERR(sa); | |
3632 | ||
b8e95489 MX |
3633 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) { |
3634 | ret = mnt_want_write_file(file); | |
3635 | if (ret) | |
3636 | goto out; | |
3637 | } | |
3638 | ||
0b246afa | 3639 | ret = btrfs_scrub_dev(fs_info, sa->devid, sa->start, sa->end, |
63a212ab SB |
3640 | &sa->progress, sa->flags & BTRFS_SCRUB_READONLY, |
3641 | 0); | |
475f6387 | 3642 | |
5afe6ce7 FM |
3643 | /* |
3644 | * Copy scrub args to user space even if btrfs_scrub_dev() returned an | |
3645 | * error. This is important as it allows user space to know how much | |
3646 | * progress scrub has done. For example, if scrub is canceled we get | |
3647 | * -ECANCELED from btrfs_scrub_dev() and return that error back to user | |
3648 | * space. Later user space can inspect the progress from the structure | |
3649 | * btrfs_ioctl_scrub_args and resume scrub from where it left off | |
3650 | * previously (btrfs-progs does this). | |
3651 | * If we fail to copy the btrfs_ioctl_scrub_args structure to user space | |
3652 | * then return -EFAULT to signal the structure was not copied or it may | |
3653 | * be corrupt and unreliable due to a partial copy. | |
3654 | */ | |
3655 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
475f6387 JS |
3656 | ret = -EFAULT; |
3657 | ||
b8e95489 MX |
3658 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) |
3659 | mnt_drop_write_file(file); | |
3660 | out: | |
475f6387 JS |
3661 | kfree(sa); |
3662 | return ret; | |
3663 | } | |
3664 | ||
2ff7e61e | 3665 | static long btrfs_ioctl_scrub_cancel(struct btrfs_fs_info *fs_info) |
475f6387 JS |
3666 | { |
3667 | if (!capable(CAP_SYS_ADMIN)) | |
3668 | return -EPERM; | |
3669 | ||
2ff7e61e | 3670 | return btrfs_scrub_cancel(fs_info); |
475f6387 JS |
3671 | } |
3672 | ||
2ff7e61e | 3673 | static long btrfs_ioctl_scrub_progress(struct btrfs_fs_info *fs_info, |
475f6387 JS |
3674 | void __user *arg) |
3675 | { | |
3676 | struct btrfs_ioctl_scrub_args *sa; | |
3677 | int ret; | |
3678 | ||
3679 | if (!capable(CAP_SYS_ADMIN)) | |
3680 | return -EPERM; | |
3681 | ||
3682 | sa = memdup_user(arg, sizeof(*sa)); | |
3683 | if (IS_ERR(sa)) | |
3684 | return PTR_ERR(sa); | |
3685 | ||
2ff7e61e | 3686 | ret = btrfs_scrub_progress(fs_info, sa->devid, &sa->progress); |
475f6387 | 3687 | |
4fa99b00 | 3688 | if (ret == 0 && copy_to_user(arg, sa, sizeof(*sa))) |
475f6387 JS |
3689 | ret = -EFAULT; |
3690 | ||
3691 | kfree(sa); | |
3692 | return ret; | |
3693 | } | |
3694 | ||
2ff7e61e | 3695 | static long btrfs_ioctl_get_dev_stats(struct btrfs_fs_info *fs_info, |
b27f7c0c | 3696 | void __user *arg) |
c11d2c23 SB |
3697 | { |
3698 | struct btrfs_ioctl_get_dev_stats *sa; | |
3699 | int ret; | |
3700 | ||
c11d2c23 SB |
3701 | sa = memdup_user(arg, sizeof(*sa)); |
3702 | if (IS_ERR(sa)) | |
3703 | return PTR_ERR(sa); | |
3704 | ||
b27f7c0c DS |
3705 | if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) { |
3706 | kfree(sa); | |
3707 | return -EPERM; | |
3708 | } | |
3709 | ||
2ff7e61e | 3710 | ret = btrfs_get_dev_stats(fs_info, sa); |
c11d2c23 | 3711 | |
eee99577 | 3712 | if (ret == 0 && copy_to_user(arg, sa, sizeof(*sa))) |
c11d2c23 SB |
3713 | ret = -EFAULT; |
3714 | ||
3715 | kfree(sa); | |
3716 | return ret; | |
3717 | } | |
3718 | ||
2ff7e61e JM |
3719 | static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info, |
3720 | void __user *arg) | |
3f6bcfbd SB |
3721 | { |
3722 | struct btrfs_ioctl_dev_replace_args *p; | |
3723 | int ret; | |
3724 | ||
3725 | if (!capable(CAP_SYS_ADMIN)) | |
3726 | return -EPERM; | |
3727 | ||
3728 | p = memdup_user(arg, sizeof(*p)); | |
3729 | if (IS_ERR(p)) | |
3730 | return PTR_ERR(p); | |
3731 | ||
3732 | switch (p->cmd) { | |
3733 | case BTRFS_IOCTL_DEV_REPLACE_CMD_START: | |
bc98a42c | 3734 | if (sb_rdonly(fs_info->sb)) { |
adfa97cb ID |
3735 | ret = -EROFS; |
3736 | goto out; | |
3737 | } | |
c3e1f96c | 3738 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REPLACE)) { |
e57138b3 | 3739 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
3f6bcfbd | 3740 | } else { |
2ff7e61e | 3741 | ret = btrfs_dev_replace_by_ioctl(fs_info, p); |
c3e1f96c | 3742 | btrfs_exclop_finish(fs_info); |
3f6bcfbd SB |
3743 | } |
3744 | break; | |
3745 | case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS: | |
0b246afa | 3746 | btrfs_dev_replace_status(fs_info, p); |
3f6bcfbd SB |
3747 | ret = 0; |
3748 | break; | |
3749 | case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL: | |
17d202b9 | 3750 | p->result = btrfs_dev_replace_cancel(fs_info); |
97282031 | 3751 | ret = 0; |
3f6bcfbd SB |
3752 | break; |
3753 | default: | |
3754 | ret = -EINVAL; | |
3755 | break; | |
3756 | } | |
3757 | ||
d3a53286 | 3758 | if ((ret == 0 || ret == -ECANCELED) && copy_to_user(arg, p, sizeof(*p))) |
3f6bcfbd | 3759 | ret = -EFAULT; |
adfa97cb | 3760 | out: |
3f6bcfbd SB |
3761 | kfree(p); |
3762 | return ret; | |
3763 | } | |
3764 | ||
d7728c96 JS |
3765 | static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg) |
3766 | { | |
3767 | int ret = 0; | |
3768 | int i; | |
740c3d22 | 3769 | u64 rel_ptr; |
d7728c96 | 3770 | int size; |
806468f8 | 3771 | struct btrfs_ioctl_ino_path_args *ipa = NULL; |
d7728c96 JS |
3772 | struct inode_fs_paths *ipath = NULL; |
3773 | struct btrfs_path *path; | |
3774 | ||
82b22ac8 | 3775 | if (!capable(CAP_DAC_READ_SEARCH)) |
d7728c96 JS |
3776 | return -EPERM; |
3777 | ||
3778 | path = btrfs_alloc_path(); | |
3779 | if (!path) { | |
3780 | ret = -ENOMEM; | |
3781 | goto out; | |
3782 | } | |
3783 | ||
3784 | ipa = memdup_user(arg, sizeof(*ipa)); | |
3785 | if (IS_ERR(ipa)) { | |
3786 | ret = PTR_ERR(ipa); | |
3787 | ipa = NULL; | |
3788 | goto out; | |
3789 | } | |
3790 | ||
3791 | size = min_t(u32, ipa->size, 4096); | |
3792 | ipath = init_ipath(size, root, path); | |
3793 | if (IS_ERR(ipath)) { | |
3794 | ret = PTR_ERR(ipath); | |
3795 | ipath = NULL; | |
3796 | goto out; | |
3797 | } | |
3798 | ||
3799 | ret = paths_from_inode(ipa->inum, ipath); | |
3800 | if (ret < 0) | |
3801 | goto out; | |
3802 | ||
3803 | for (i = 0; i < ipath->fspath->elem_cnt; ++i) { | |
745c4d8e JM |
3804 | rel_ptr = ipath->fspath->val[i] - |
3805 | (u64)(unsigned long)ipath->fspath->val; | |
740c3d22 | 3806 | ipath->fspath->val[i] = rel_ptr; |
d7728c96 JS |
3807 | } |
3808 | ||
718dc5fa OS |
3809 | ret = copy_to_user((void __user *)(unsigned long)ipa->fspath, |
3810 | ipath->fspath, size); | |
d7728c96 JS |
3811 | if (ret) { |
3812 | ret = -EFAULT; | |
3813 | goto out; | |
3814 | } | |
3815 | ||
3816 | out: | |
3817 | btrfs_free_path(path); | |
3818 | free_ipath(ipath); | |
3819 | kfree(ipa); | |
3820 | ||
3821 | return ret; | |
3822 | } | |
3823 | ||
3824 | static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx) | |
3825 | { | |
3826 | struct btrfs_data_container *inodes = ctx; | |
3827 | const size_t c = 3 * sizeof(u64); | |
3828 | ||
3829 | if (inodes->bytes_left >= c) { | |
3830 | inodes->bytes_left -= c; | |
3831 | inodes->val[inodes->elem_cnt] = inum; | |
3832 | inodes->val[inodes->elem_cnt + 1] = offset; | |
3833 | inodes->val[inodes->elem_cnt + 2] = root; | |
3834 | inodes->elem_cnt += 3; | |
3835 | } else { | |
3836 | inodes->bytes_missing += c - inodes->bytes_left; | |
3837 | inodes->bytes_left = 0; | |
3838 | inodes->elem_missed += 3; | |
3839 | } | |
3840 | ||
3841 | return 0; | |
3842 | } | |
3843 | ||
2ff7e61e | 3844 | static long btrfs_ioctl_logical_to_ino(struct btrfs_fs_info *fs_info, |
d24a67b2 | 3845 | void __user *arg, int version) |
d7728c96 JS |
3846 | { |
3847 | int ret = 0; | |
3848 | int size; | |
d7728c96 JS |
3849 | struct btrfs_ioctl_logical_ino_args *loi; |
3850 | struct btrfs_data_container *inodes = NULL; | |
3851 | struct btrfs_path *path = NULL; | |
d24a67b2 | 3852 | bool ignore_offset; |
d7728c96 JS |
3853 | |
3854 | if (!capable(CAP_SYS_ADMIN)) | |
3855 | return -EPERM; | |
3856 | ||
3857 | loi = memdup_user(arg, sizeof(*loi)); | |
7b9ea627 SV |
3858 | if (IS_ERR(loi)) |
3859 | return PTR_ERR(loi); | |
d7728c96 | 3860 | |
d24a67b2 ZB |
3861 | if (version == 1) { |
3862 | ignore_offset = false; | |
b115e3bc | 3863 | size = min_t(u32, loi->size, SZ_64K); |
d24a67b2 ZB |
3864 | } else { |
3865 | /* All reserved bits must be 0 for now */ | |
3866 | if (memchr_inv(loi->reserved, 0, sizeof(loi->reserved))) { | |
3867 | ret = -EINVAL; | |
3868 | goto out_loi; | |
3869 | } | |
3870 | /* Only accept flags we have defined so far */ | |
3871 | if (loi->flags & ~(BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET)) { | |
3872 | ret = -EINVAL; | |
3873 | goto out_loi; | |
3874 | } | |
3875 | ignore_offset = loi->flags & BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET; | |
b115e3bc | 3876 | size = min_t(u32, loi->size, SZ_16M); |
d24a67b2 ZB |
3877 | } |
3878 | ||
d7728c96 JS |
3879 | path = btrfs_alloc_path(); |
3880 | if (!path) { | |
3881 | ret = -ENOMEM; | |
3882 | goto out; | |
3883 | } | |
3884 | ||
d7728c96 JS |
3885 | inodes = init_data_container(size); |
3886 | if (IS_ERR(inodes)) { | |
3887 | ret = PTR_ERR(inodes); | |
3888 | inodes = NULL; | |
3889 | goto out; | |
3890 | } | |
3891 | ||
2ff7e61e | 3892 | ret = iterate_inodes_from_logical(loi->logical, fs_info, path, |
d24a67b2 | 3893 | build_ino_list, inodes, ignore_offset); |
df031f07 | 3894 | if (ret == -EINVAL) |
d7728c96 JS |
3895 | ret = -ENOENT; |
3896 | if (ret < 0) | |
3897 | goto out; | |
3898 | ||
718dc5fa OS |
3899 | ret = copy_to_user((void __user *)(unsigned long)loi->inodes, inodes, |
3900 | size); | |
d7728c96 JS |
3901 | if (ret) |
3902 | ret = -EFAULT; | |
3903 | ||
3904 | out: | |
3905 | btrfs_free_path(path); | |
f54de068 | 3906 | kvfree(inodes); |
d24a67b2 | 3907 | out_loi: |
d7728c96 JS |
3908 | kfree(loi); |
3909 | ||
3910 | return ret; | |
3911 | } | |
3912 | ||
008ef096 | 3913 | void btrfs_update_ioctl_balance_args(struct btrfs_fs_info *fs_info, |
c9e9f97b ID |
3914 | struct btrfs_ioctl_balance_args *bargs) |
3915 | { | |
3916 | struct btrfs_balance_control *bctl = fs_info->balance_ctl; | |
3917 | ||
3918 | bargs->flags = bctl->flags; | |
3919 | ||
3009a62f | 3920 | if (test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) |
837d5b6e ID |
3921 | bargs->state |= BTRFS_BALANCE_STATE_RUNNING; |
3922 | if (atomic_read(&fs_info->balance_pause_req)) | |
3923 | bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ; | |
a7e99c69 ID |
3924 | if (atomic_read(&fs_info->balance_cancel_req)) |
3925 | bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ; | |
837d5b6e | 3926 | |
c9e9f97b ID |
3927 | memcpy(&bargs->data, &bctl->data, sizeof(bargs->data)); |
3928 | memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta)); | |
3929 | memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys)); | |
19a39dce | 3930 | |
008ef096 DS |
3931 | spin_lock(&fs_info->balance_lock); |
3932 | memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat)); | |
3933 | spin_unlock(&fs_info->balance_lock); | |
c9e9f97b ID |
3934 | } |
3935 | ||
9ba1f6e4 | 3936 | static long btrfs_ioctl_balance(struct file *file, void __user *arg) |
c9e9f97b | 3937 | { |
496ad9aa | 3938 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
c9e9f97b ID |
3939 | struct btrfs_fs_info *fs_info = root->fs_info; |
3940 | struct btrfs_ioctl_balance_args *bargs; | |
3941 | struct btrfs_balance_control *bctl; | |
ed0fb78f | 3942 | bool need_unlock; /* for mut. excl. ops lock */ |
c9e9f97b ID |
3943 | int ret; |
3944 | ||
3945 | if (!capable(CAP_SYS_ADMIN)) | |
3946 | return -EPERM; | |
3947 | ||
e54bfa31 | 3948 | ret = mnt_want_write_file(file); |
9ba1f6e4 LB |
3949 | if (ret) |
3950 | return ret; | |
3951 | ||
ed0fb78f | 3952 | again: |
c3e1f96c | 3953 | if (btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) { |
ed0fb78f ID |
3954 | mutex_lock(&fs_info->balance_mutex); |
3955 | need_unlock = true; | |
3956 | goto locked; | |
3957 | } | |
3958 | ||
3959 | /* | |
01327610 | 3960 | * mut. excl. ops lock is locked. Three possibilities: |
ed0fb78f ID |
3961 | * (1) some other op is running |
3962 | * (2) balance is running | |
3963 | * (3) balance is paused -- special case (think resume) | |
3964 | */ | |
c9e9f97b | 3965 | mutex_lock(&fs_info->balance_mutex); |
ed0fb78f ID |
3966 | if (fs_info->balance_ctl) { |
3967 | /* this is either (2) or (3) */ | |
3009a62f | 3968 | if (!test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) { |
ed0fb78f | 3969 | mutex_unlock(&fs_info->balance_mutex); |
dccdb07b DS |
3970 | /* |
3971 | * Lock released to allow other waiters to continue, | |
3972 | * we'll reexamine the status again. | |
3973 | */ | |
ed0fb78f ID |
3974 | mutex_lock(&fs_info->balance_mutex); |
3975 | ||
3976 | if (fs_info->balance_ctl && | |
3009a62f | 3977 | !test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) { |
ed0fb78f ID |
3978 | /* this is (3) */ |
3979 | need_unlock = false; | |
3980 | goto locked; | |
3981 | } | |
3982 | ||
3983 | mutex_unlock(&fs_info->balance_mutex); | |
ed0fb78f ID |
3984 | goto again; |
3985 | } else { | |
3986 | /* this is (2) */ | |
3987 | mutex_unlock(&fs_info->balance_mutex); | |
3988 | ret = -EINPROGRESS; | |
3989 | goto out; | |
3990 | } | |
3991 | } else { | |
3992 | /* this is (1) */ | |
3993 | mutex_unlock(&fs_info->balance_mutex); | |
e57138b3 | 3994 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
ed0fb78f ID |
3995 | goto out; |
3996 | } | |
3997 | ||
3998 | locked: | |
c9e9f97b ID |
3999 | |
4000 | if (arg) { | |
4001 | bargs = memdup_user(arg, sizeof(*bargs)); | |
4002 | if (IS_ERR(bargs)) { | |
4003 | ret = PTR_ERR(bargs); | |
ed0fb78f | 4004 | goto out_unlock; |
c9e9f97b | 4005 | } |
de322263 ID |
4006 | |
4007 | if (bargs->flags & BTRFS_BALANCE_RESUME) { | |
4008 | if (!fs_info->balance_ctl) { | |
4009 | ret = -ENOTCONN; | |
4010 | goto out_bargs; | |
4011 | } | |
4012 | ||
4013 | bctl = fs_info->balance_ctl; | |
4014 | spin_lock(&fs_info->balance_lock); | |
4015 | bctl->flags |= BTRFS_BALANCE_RESUME; | |
4016 | spin_unlock(&fs_info->balance_lock); | |
4017 | ||
4018 | goto do_balance; | |
4019 | } | |
c9e9f97b ID |
4020 | } else { |
4021 | bargs = NULL; | |
4022 | } | |
4023 | ||
ed0fb78f | 4024 | if (fs_info->balance_ctl) { |
837d5b6e ID |
4025 | ret = -EINPROGRESS; |
4026 | goto out_bargs; | |
4027 | } | |
4028 | ||
8d2db785 | 4029 | bctl = kzalloc(sizeof(*bctl), GFP_KERNEL); |
c9e9f97b ID |
4030 | if (!bctl) { |
4031 | ret = -ENOMEM; | |
4032 | goto out_bargs; | |
4033 | } | |
4034 | ||
c9e9f97b ID |
4035 | if (arg) { |
4036 | memcpy(&bctl->data, &bargs->data, sizeof(bctl->data)); | |
4037 | memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta)); | |
4038 | memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys)); | |
4039 | ||
4040 | bctl->flags = bargs->flags; | |
f43ffb60 ID |
4041 | } else { |
4042 | /* balance everything - no filters */ | |
4043 | bctl->flags |= BTRFS_BALANCE_TYPE_MASK; | |
c9e9f97b ID |
4044 | } |
4045 | ||
8eb93459 DS |
4046 | if (bctl->flags & ~(BTRFS_BALANCE_ARGS_MASK | BTRFS_BALANCE_TYPE_MASK)) { |
4047 | ret = -EINVAL; | |
0f89abf5 | 4048 | goto out_bctl; |
8eb93459 DS |
4049 | } |
4050 | ||
de322263 | 4051 | do_balance: |
c9e9f97b | 4052 | /* |
c3e1f96c GR |
4053 | * Ownership of bctl and exclusive operation goes to btrfs_balance. |
4054 | * bctl is freed in reset_balance_state, or, if restriper was paused | |
4055 | * all the way until unmount, in free_fs_info. The flag should be | |
4056 | * cleared after reset_balance_state. | |
c9e9f97b | 4057 | */ |
ed0fb78f ID |
4058 | need_unlock = false; |
4059 | ||
6fcf6e2b | 4060 | ret = btrfs_balance(fs_info, bctl, bargs); |
0f89abf5 | 4061 | bctl = NULL; |
ed0fb78f | 4062 | |
d00c2d9c | 4063 | if ((ret == 0 || ret == -ECANCELED) && arg) { |
c9e9f97b ID |
4064 | if (copy_to_user(arg, bargs, sizeof(*bargs))) |
4065 | ret = -EFAULT; | |
4066 | } | |
4067 | ||
0f89abf5 CE |
4068 | out_bctl: |
4069 | kfree(bctl); | |
c9e9f97b ID |
4070 | out_bargs: |
4071 | kfree(bargs); | |
ed0fb78f | 4072 | out_unlock: |
c9e9f97b | 4073 | mutex_unlock(&fs_info->balance_mutex); |
ed0fb78f | 4074 | if (need_unlock) |
c3e1f96c | 4075 | btrfs_exclop_finish(fs_info); |
ed0fb78f | 4076 | out: |
e54bfa31 | 4077 | mnt_drop_write_file(file); |
c9e9f97b ID |
4078 | return ret; |
4079 | } | |
4080 | ||
2ff7e61e | 4081 | static long btrfs_ioctl_balance_ctl(struct btrfs_fs_info *fs_info, int cmd) |
837d5b6e ID |
4082 | { |
4083 | if (!capable(CAP_SYS_ADMIN)) | |
4084 | return -EPERM; | |
4085 | ||
4086 | switch (cmd) { | |
4087 | case BTRFS_BALANCE_CTL_PAUSE: | |
0b246afa | 4088 | return btrfs_pause_balance(fs_info); |
a7e99c69 | 4089 | case BTRFS_BALANCE_CTL_CANCEL: |
0b246afa | 4090 | return btrfs_cancel_balance(fs_info); |
837d5b6e ID |
4091 | } |
4092 | ||
4093 | return -EINVAL; | |
4094 | } | |
4095 | ||
2ff7e61e | 4096 | static long btrfs_ioctl_balance_progress(struct btrfs_fs_info *fs_info, |
19a39dce ID |
4097 | void __user *arg) |
4098 | { | |
19a39dce ID |
4099 | struct btrfs_ioctl_balance_args *bargs; |
4100 | int ret = 0; | |
4101 | ||
4102 | if (!capable(CAP_SYS_ADMIN)) | |
4103 | return -EPERM; | |
4104 | ||
4105 | mutex_lock(&fs_info->balance_mutex); | |
4106 | if (!fs_info->balance_ctl) { | |
4107 | ret = -ENOTCONN; | |
4108 | goto out; | |
4109 | } | |
4110 | ||
8d2db785 | 4111 | bargs = kzalloc(sizeof(*bargs), GFP_KERNEL); |
19a39dce ID |
4112 | if (!bargs) { |
4113 | ret = -ENOMEM; | |
4114 | goto out; | |
4115 | } | |
4116 | ||
008ef096 | 4117 | btrfs_update_ioctl_balance_args(fs_info, bargs); |
19a39dce ID |
4118 | |
4119 | if (copy_to_user(arg, bargs, sizeof(*bargs))) | |
4120 | ret = -EFAULT; | |
4121 | ||
4122 | kfree(bargs); | |
4123 | out: | |
4124 | mutex_unlock(&fs_info->balance_mutex); | |
4125 | return ret; | |
4126 | } | |
4127 | ||
905b0dda | 4128 | static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg) |
5d13a37b | 4129 | { |
0b246afa JM |
4130 | struct inode *inode = file_inode(file); |
4131 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
5d13a37b | 4132 | struct btrfs_ioctl_quota_ctl_args *sa; |
5d13a37b | 4133 | int ret; |
5d13a37b AJ |
4134 | |
4135 | if (!capable(CAP_SYS_ADMIN)) | |
4136 | return -EPERM; | |
4137 | ||
905b0dda MX |
4138 | ret = mnt_want_write_file(file); |
4139 | if (ret) | |
4140 | return ret; | |
5d13a37b AJ |
4141 | |
4142 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4143 | if (IS_ERR(sa)) { |
4144 | ret = PTR_ERR(sa); | |
4145 | goto drop_write; | |
4146 | } | |
5d13a37b | 4147 | |
0b246afa | 4148 | down_write(&fs_info->subvol_sem); |
5d13a37b AJ |
4149 | |
4150 | switch (sa->cmd) { | |
4151 | case BTRFS_QUOTA_CTL_ENABLE: | |
340f1aa2 | 4152 | ret = btrfs_quota_enable(fs_info); |
5d13a37b AJ |
4153 | break; |
4154 | case BTRFS_QUOTA_CTL_DISABLE: | |
340f1aa2 | 4155 | ret = btrfs_quota_disable(fs_info); |
5d13a37b | 4156 | break; |
5d13a37b AJ |
4157 | default: |
4158 | ret = -EINVAL; | |
4159 | break; | |
4160 | } | |
4161 | ||
5d13a37b | 4162 | kfree(sa); |
0b246afa | 4163 | up_write(&fs_info->subvol_sem); |
905b0dda MX |
4164 | drop_write: |
4165 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4166 | return ret; |
4167 | } | |
4168 | ||
905b0dda | 4169 | static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg) |
5d13a37b | 4170 | { |
0b246afa JM |
4171 | struct inode *inode = file_inode(file); |
4172 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4173 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
5d13a37b AJ |
4174 | struct btrfs_ioctl_qgroup_assign_args *sa; |
4175 | struct btrfs_trans_handle *trans; | |
4176 | int ret; | |
4177 | int err; | |
4178 | ||
4179 | if (!capable(CAP_SYS_ADMIN)) | |
4180 | return -EPERM; | |
4181 | ||
905b0dda MX |
4182 | ret = mnt_want_write_file(file); |
4183 | if (ret) | |
4184 | return ret; | |
5d13a37b AJ |
4185 | |
4186 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4187 | if (IS_ERR(sa)) { |
4188 | ret = PTR_ERR(sa); | |
4189 | goto drop_write; | |
4190 | } | |
5d13a37b AJ |
4191 | |
4192 | trans = btrfs_join_transaction(root); | |
4193 | if (IS_ERR(trans)) { | |
4194 | ret = PTR_ERR(trans); | |
4195 | goto out; | |
4196 | } | |
4197 | ||
5d13a37b | 4198 | if (sa->assign) { |
9f8a6ce6 | 4199 | ret = btrfs_add_qgroup_relation(trans, sa->src, sa->dst); |
5d13a37b | 4200 | } else { |
39616c27 | 4201 | ret = btrfs_del_qgroup_relation(trans, sa->src, sa->dst); |
5d13a37b AJ |
4202 | } |
4203 | ||
e082f563 | 4204 | /* update qgroup status and info */ |
280f8bd2 | 4205 | err = btrfs_run_qgroups(trans); |
e082f563 | 4206 | if (err < 0) |
0b246afa JM |
4207 | btrfs_handle_fs_error(fs_info, err, |
4208 | "failed to update qgroup status and info"); | |
3a45bb20 | 4209 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4210 | if (err && !ret) |
4211 | ret = err; | |
4212 | ||
4213 | out: | |
4214 | kfree(sa); | |
905b0dda MX |
4215 | drop_write: |
4216 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4217 | return ret; |
4218 | } | |
4219 | ||
905b0dda | 4220 | static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg) |
5d13a37b | 4221 | { |
0b246afa | 4222 | struct inode *inode = file_inode(file); |
0b246afa | 4223 | struct btrfs_root *root = BTRFS_I(inode)->root; |
5d13a37b AJ |
4224 | struct btrfs_ioctl_qgroup_create_args *sa; |
4225 | struct btrfs_trans_handle *trans; | |
4226 | int ret; | |
4227 | int err; | |
4228 | ||
4229 | if (!capable(CAP_SYS_ADMIN)) | |
4230 | return -EPERM; | |
4231 | ||
905b0dda MX |
4232 | ret = mnt_want_write_file(file); |
4233 | if (ret) | |
4234 | return ret; | |
5d13a37b AJ |
4235 | |
4236 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4237 | if (IS_ERR(sa)) { |
4238 | ret = PTR_ERR(sa); | |
4239 | goto drop_write; | |
4240 | } | |
5d13a37b | 4241 | |
d86e56cf MX |
4242 | if (!sa->qgroupid) { |
4243 | ret = -EINVAL; | |
4244 | goto out; | |
4245 | } | |
4246 | ||
5d13a37b AJ |
4247 | trans = btrfs_join_transaction(root); |
4248 | if (IS_ERR(trans)) { | |
4249 | ret = PTR_ERR(trans); | |
4250 | goto out; | |
4251 | } | |
4252 | ||
5d13a37b | 4253 | if (sa->create) { |
49a05ecd | 4254 | ret = btrfs_create_qgroup(trans, sa->qgroupid); |
5d13a37b | 4255 | } else { |
3efbee1d | 4256 | ret = btrfs_remove_qgroup(trans, sa->qgroupid); |
5d13a37b AJ |
4257 | } |
4258 | ||
3a45bb20 | 4259 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4260 | if (err && !ret) |
4261 | ret = err; | |
4262 | ||
4263 | out: | |
4264 | kfree(sa); | |
905b0dda MX |
4265 | drop_write: |
4266 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4267 | return ret; |
4268 | } | |
4269 | ||
905b0dda | 4270 | static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg) |
5d13a37b | 4271 | { |
0b246afa | 4272 | struct inode *inode = file_inode(file); |
0b246afa | 4273 | struct btrfs_root *root = BTRFS_I(inode)->root; |
5d13a37b AJ |
4274 | struct btrfs_ioctl_qgroup_limit_args *sa; |
4275 | struct btrfs_trans_handle *trans; | |
4276 | int ret; | |
4277 | int err; | |
4278 | u64 qgroupid; | |
4279 | ||
4280 | if (!capable(CAP_SYS_ADMIN)) | |
4281 | return -EPERM; | |
4282 | ||
905b0dda MX |
4283 | ret = mnt_want_write_file(file); |
4284 | if (ret) | |
4285 | return ret; | |
5d13a37b AJ |
4286 | |
4287 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4288 | if (IS_ERR(sa)) { |
4289 | ret = PTR_ERR(sa); | |
4290 | goto drop_write; | |
4291 | } | |
5d13a37b AJ |
4292 | |
4293 | trans = btrfs_join_transaction(root); | |
4294 | if (IS_ERR(trans)) { | |
4295 | ret = PTR_ERR(trans); | |
4296 | goto out; | |
4297 | } | |
4298 | ||
4299 | qgroupid = sa->qgroupid; | |
4300 | if (!qgroupid) { | |
4301 | /* take the current subvol as qgroup */ | |
4302 | qgroupid = root->root_key.objectid; | |
4303 | } | |
4304 | ||
f0042d5e | 4305 | ret = btrfs_limit_qgroup(trans, qgroupid, &sa->lim); |
5d13a37b | 4306 | |
3a45bb20 | 4307 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4308 | if (err && !ret) |
4309 | ret = err; | |
4310 | ||
4311 | out: | |
4312 | kfree(sa); | |
905b0dda MX |
4313 | drop_write: |
4314 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4315 | return ret; |
4316 | } | |
4317 | ||
2f232036 JS |
4318 | static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg) |
4319 | { | |
0b246afa JM |
4320 | struct inode *inode = file_inode(file); |
4321 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
2f232036 JS |
4322 | struct btrfs_ioctl_quota_rescan_args *qsa; |
4323 | int ret; | |
4324 | ||
4325 | if (!capable(CAP_SYS_ADMIN)) | |
4326 | return -EPERM; | |
4327 | ||
4328 | ret = mnt_want_write_file(file); | |
4329 | if (ret) | |
4330 | return ret; | |
4331 | ||
4332 | qsa = memdup_user(arg, sizeof(*qsa)); | |
4333 | if (IS_ERR(qsa)) { | |
4334 | ret = PTR_ERR(qsa); | |
4335 | goto drop_write; | |
4336 | } | |
4337 | ||
4338 | if (qsa->flags) { | |
4339 | ret = -EINVAL; | |
4340 | goto out; | |
4341 | } | |
4342 | ||
0b246afa | 4343 | ret = btrfs_qgroup_rescan(fs_info); |
2f232036 JS |
4344 | |
4345 | out: | |
4346 | kfree(qsa); | |
4347 | drop_write: | |
4348 | mnt_drop_write_file(file); | |
4349 | return ret; | |
4350 | } | |
4351 | ||
b929c1d8 MPS |
4352 | static long btrfs_ioctl_quota_rescan_status(struct btrfs_fs_info *fs_info, |
4353 | void __user *arg) | |
2f232036 | 4354 | { |
2f232036 JS |
4355 | struct btrfs_ioctl_quota_rescan_args *qsa; |
4356 | int ret = 0; | |
4357 | ||
4358 | if (!capable(CAP_SYS_ADMIN)) | |
4359 | return -EPERM; | |
4360 | ||
8d2db785 | 4361 | qsa = kzalloc(sizeof(*qsa), GFP_KERNEL); |
2f232036 JS |
4362 | if (!qsa) |
4363 | return -ENOMEM; | |
4364 | ||
0b246afa | 4365 | if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) { |
2f232036 | 4366 | qsa->flags = 1; |
0b246afa | 4367 | qsa->progress = fs_info->qgroup_rescan_progress.objectid; |
2f232036 JS |
4368 | } |
4369 | ||
4370 | if (copy_to_user(arg, qsa, sizeof(*qsa))) | |
4371 | ret = -EFAULT; | |
4372 | ||
4373 | kfree(qsa); | |
4374 | return ret; | |
4375 | } | |
4376 | ||
b929c1d8 MPS |
4377 | static long btrfs_ioctl_quota_rescan_wait(struct btrfs_fs_info *fs_info, |
4378 | void __user *arg) | |
57254b6e | 4379 | { |
57254b6e JS |
4380 | if (!capable(CAP_SYS_ADMIN)) |
4381 | return -EPERM; | |
4382 | ||
0b246afa | 4383 | return btrfs_qgroup_wait_for_completion(fs_info, true); |
57254b6e JS |
4384 | } |
4385 | ||
abccd00f HM |
4386 | static long _btrfs_ioctl_set_received_subvol(struct file *file, |
4387 | struct btrfs_ioctl_received_subvol_args *sa) | |
8ea05e3a | 4388 | { |
496ad9aa | 4389 | struct inode *inode = file_inode(file); |
0b246afa | 4390 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
8ea05e3a AB |
4391 | struct btrfs_root *root = BTRFS_I(inode)->root; |
4392 | struct btrfs_root_item *root_item = &root->root_item; | |
4393 | struct btrfs_trans_handle *trans; | |
95582b00 | 4394 | struct timespec64 ct = current_time(inode); |
8ea05e3a | 4395 | int ret = 0; |
dd5f9615 | 4396 | int received_uuid_changed; |
8ea05e3a | 4397 | |
21cb47be | 4398 | if (!inode_owner_or_capable(&init_user_ns, inode)) |
bd60ea0f DS |
4399 | return -EPERM; |
4400 | ||
8ea05e3a AB |
4401 | ret = mnt_want_write_file(file); |
4402 | if (ret < 0) | |
4403 | return ret; | |
4404 | ||
0b246afa | 4405 | down_write(&fs_info->subvol_sem); |
8ea05e3a | 4406 | |
4a0cc7ca | 4407 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
8ea05e3a AB |
4408 | ret = -EINVAL; |
4409 | goto out; | |
4410 | } | |
4411 | ||
4412 | if (btrfs_root_readonly(root)) { | |
4413 | ret = -EROFS; | |
4414 | goto out; | |
4415 | } | |
4416 | ||
dd5f9615 SB |
4417 | /* |
4418 | * 1 - root item | |
4419 | * 2 - uuid items (received uuid + subvol uuid) | |
4420 | */ | |
4421 | trans = btrfs_start_transaction(root, 3); | |
8ea05e3a AB |
4422 | if (IS_ERR(trans)) { |
4423 | ret = PTR_ERR(trans); | |
4424 | trans = NULL; | |
4425 | goto out; | |
4426 | } | |
4427 | ||
4428 | sa->rtransid = trans->transid; | |
4429 | sa->rtime.sec = ct.tv_sec; | |
4430 | sa->rtime.nsec = ct.tv_nsec; | |
4431 | ||
dd5f9615 SB |
4432 | received_uuid_changed = memcmp(root_item->received_uuid, sa->uuid, |
4433 | BTRFS_UUID_SIZE); | |
4434 | if (received_uuid_changed && | |
d87ff758 | 4435 | !btrfs_is_empty_uuid(root_item->received_uuid)) { |
d1957791 | 4436 | ret = btrfs_uuid_tree_remove(trans, root_item->received_uuid, |
d87ff758 NB |
4437 | BTRFS_UUID_KEY_RECEIVED_SUBVOL, |
4438 | root->root_key.objectid); | |
4439 | if (ret && ret != -ENOENT) { | |
4440 | btrfs_abort_transaction(trans, ret); | |
4441 | btrfs_end_transaction(trans); | |
4442 | goto out; | |
4443 | } | |
4444 | } | |
8ea05e3a AB |
4445 | memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE); |
4446 | btrfs_set_root_stransid(root_item, sa->stransid); | |
4447 | btrfs_set_root_rtransid(root_item, sa->rtransid); | |
3cae210f QW |
4448 | btrfs_set_stack_timespec_sec(&root_item->stime, sa->stime.sec); |
4449 | btrfs_set_stack_timespec_nsec(&root_item->stime, sa->stime.nsec); | |
4450 | btrfs_set_stack_timespec_sec(&root_item->rtime, sa->rtime.sec); | |
4451 | btrfs_set_stack_timespec_nsec(&root_item->rtime, sa->rtime.nsec); | |
8ea05e3a | 4452 | |
0b246afa | 4453 | ret = btrfs_update_root(trans, fs_info->tree_root, |
8ea05e3a AB |
4454 | &root->root_key, &root->root_item); |
4455 | if (ret < 0) { | |
3a45bb20 | 4456 | btrfs_end_transaction(trans); |
8ea05e3a | 4457 | goto out; |
dd5f9615 SB |
4458 | } |
4459 | if (received_uuid_changed && !btrfs_is_empty_uuid(sa->uuid)) { | |
cdb345a8 | 4460 | ret = btrfs_uuid_tree_add(trans, sa->uuid, |
dd5f9615 SB |
4461 | BTRFS_UUID_KEY_RECEIVED_SUBVOL, |
4462 | root->root_key.objectid); | |
4463 | if (ret < 0 && ret != -EEXIST) { | |
66642832 | 4464 | btrfs_abort_transaction(trans, ret); |
efd38150 | 4465 | btrfs_end_transaction(trans); |
8ea05e3a | 4466 | goto out; |
dd5f9615 SB |
4467 | } |
4468 | } | |
3a45bb20 | 4469 | ret = btrfs_commit_transaction(trans); |
abccd00f | 4470 | out: |
0b246afa | 4471 | up_write(&fs_info->subvol_sem); |
abccd00f HM |
4472 | mnt_drop_write_file(file); |
4473 | return ret; | |
4474 | } | |
4475 | ||
4476 | #ifdef CONFIG_64BIT | |
4477 | static long btrfs_ioctl_set_received_subvol_32(struct file *file, | |
4478 | void __user *arg) | |
4479 | { | |
4480 | struct btrfs_ioctl_received_subvol_args_32 *args32 = NULL; | |
4481 | struct btrfs_ioctl_received_subvol_args *args64 = NULL; | |
4482 | int ret = 0; | |
4483 | ||
4484 | args32 = memdup_user(arg, sizeof(*args32)); | |
7b9ea627 SV |
4485 | if (IS_ERR(args32)) |
4486 | return PTR_ERR(args32); | |
abccd00f | 4487 | |
8d2db785 | 4488 | args64 = kmalloc(sizeof(*args64), GFP_KERNEL); |
84dbeb87 DC |
4489 | if (!args64) { |
4490 | ret = -ENOMEM; | |
abccd00f HM |
4491 | goto out; |
4492 | } | |
4493 | ||
4494 | memcpy(args64->uuid, args32->uuid, BTRFS_UUID_SIZE); | |
4495 | args64->stransid = args32->stransid; | |
4496 | args64->rtransid = args32->rtransid; | |
4497 | args64->stime.sec = args32->stime.sec; | |
4498 | args64->stime.nsec = args32->stime.nsec; | |
4499 | args64->rtime.sec = args32->rtime.sec; | |
4500 | args64->rtime.nsec = args32->rtime.nsec; | |
4501 | args64->flags = args32->flags; | |
4502 | ||
4503 | ret = _btrfs_ioctl_set_received_subvol(file, args64); | |
4504 | if (ret) | |
4505 | goto out; | |
4506 | ||
4507 | memcpy(args32->uuid, args64->uuid, BTRFS_UUID_SIZE); | |
4508 | args32->stransid = args64->stransid; | |
4509 | args32->rtransid = args64->rtransid; | |
4510 | args32->stime.sec = args64->stime.sec; | |
4511 | args32->stime.nsec = args64->stime.nsec; | |
4512 | args32->rtime.sec = args64->rtime.sec; | |
4513 | args32->rtime.nsec = args64->rtime.nsec; | |
4514 | args32->flags = args64->flags; | |
4515 | ||
4516 | ret = copy_to_user(arg, args32, sizeof(*args32)); | |
4517 | if (ret) | |
4518 | ret = -EFAULT; | |
4519 | ||
4520 | out: | |
4521 | kfree(args32); | |
4522 | kfree(args64); | |
4523 | return ret; | |
4524 | } | |
4525 | #endif | |
4526 | ||
4527 | static long btrfs_ioctl_set_received_subvol(struct file *file, | |
4528 | void __user *arg) | |
4529 | { | |
4530 | struct btrfs_ioctl_received_subvol_args *sa = NULL; | |
4531 | int ret = 0; | |
4532 | ||
4533 | sa = memdup_user(arg, sizeof(*sa)); | |
7b9ea627 SV |
4534 | if (IS_ERR(sa)) |
4535 | return PTR_ERR(sa); | |
abccd00f HM |
4536 | |
4537 | ret = _btrfs_ioctl_set_received_subvol(file, sa); | |
4538 | ||
4539 | if (ret) | |
4540 | goto out; | |
4541 | ||
8ea05e3a AB |
4542 | ret = copy_to_user(arg, sa, sizeof(*sa)); |
4543 | if (ret) | |
4544 | ret = -EFAULT; | |
4545 | ||
4546 | out: | |
4547 | kfree(sa); | |
8ea05e3a AB |
4548 | return ret; |
4549 | } | |
4550 | ||
b929c1d8 MPS |
4551 | static int btrfs_ioctl_get_fslabel(struct btrfs_fs_info *fs_info, |
4552 | void __user *arg) | |
867ab667 | 4553 | { |
a1b83ac5 | 4554 | size_t len; |
867ab667 | 4555 | int ret; |
a1b83ac5 AJ |
4556 | char label[BTRFS_LABEL_SIZE]; |
4557 | ||
0b246afa JM |
4558 | spin_lock(&fs_info->super_lock); |
4559 | memcpy(label, fs_info->super_copy->label, BTRFS_LABEL_SIZE); | |
4560 | spin_unlock(&fs_info->super_lock); | |
a1b83ac5 AJ |
4561 | |
4562 | len = strnlen(label, BTRFS_LABEL_SIZE); | |
867ab667 | 4563 | |
4564 | if (len == BTRFS_LABEL_SIZE) { | |
0b246afa JM |
4565 | btrfs_warn(fs_info, |
4566 | "label is too long, return the first %zu bytes", | |
4567 | --len); | |
867ab667 | 4568 | } |
4569 | ||
867ab667 | 4570 | ret = copy_to_user(arg, label, len); |
867ab667 | 4571 | |
4572 | return ret ? -EFAULT : 0; | |
4573 | } | |
4574 | ||
a8bfd4ab | 4575 | static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg) |
4576 | { | |
0b246afa JM |
4577 | struct inode *inode = file_inode(file); |
4578 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4579 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4580 | struct btrfs_super_block *super_block = fs_info->super_copy; | |
a8bfd4ab | 4581 | struct btrfs_trans_handle *trans; |
4582 | char label[BTRFS_LABEL_SIZE]; | |
4583 | int ret; | |
4584 | ||
4585 | if (!capable(CAP_SYS_ADMIN)) | |
4586 | return -EPERM; | |
4587 | ||
4588 | if (copy_from_user(label, arg, sizeof(label))) | |
4589 | return -EFAULT; | |
4590 | ||
4591 | if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) { | |
0b246afa | 4592 | btrfs_err(fs_info, |
5d163e0e JM |
4593 | "unable to set label with more than %d bytes", |
4594 | BTRFS_LABEL_SIZE - 1); | |
a8bfd4ab | 4595 | return -EINVAL; |
4596 | } | |
4597 | ||
4598 | ret = mnt_want_write_file(file); | |
4599 | if (ret) | |
4600 | return ret; | |
4601 | ||
a8bfd4ab | 4602 | trans = btrfs_start_transaction(root, 0); |
4603 | if (IS_ERR(trans)) { | |
4604 | ret = PTR_ERR(trans); | |
4605 | goto out_unlock; | |
4606 | } | |
4607 | ||
0b246afa | 4608 | spin_lock(&fs_info->super_lock); |
a8bfd4ab | 4609 | strcpy(super_block->label, label); |
0b246afa | 4610 | spin_unlock(&fs_info->super_lock); |
3a45bb20 | 4611 | ret = btrfs_commit_transaction(trans); |
a8bfd4ab | 4612 | |
4613 | out_unlock: | |
a8bfd4ab | 4614 | mnt_drop_write_file(file); |
4615 | return ret; | |
4616 | } | |
4617 | ||
2eaa055f JM |
4618 | #define INIT_FEATURE_FLAGS(suffix) \ |
4619 | { .compat_flags = BTRFS_FEATURE_COMPAT_##suffix, \ | |
4620 | .compat_ro_flags = BTRFS_FEATURE_COMPAT_RO_##suffix, \ | |
4621 | .incompat_flags = BTRFS_FEATURE_INCOMPAT_##suffix } | |
4622 | ||
d5131b65 | 4623 | int btrfs_ioctl_get_supported_features(void __user *arg) |
2eaa055f | 4624 | { |
4d4ab6d6 | 4625 | static const struct btrfs_ioctl_feature_flags features[3] = { |
2eaa055f JM |
4626 | INIT_FEATURE_FLAGS(SUPP), |
4627 | INIT_FEATURE_FLAGS(SAFE_SET), | |
4628 | INIT_FEATURE_FLAGS(SAFE_CLEAR) | |
4629 | }; | |
4630 | ||
4631 | if (copy_to_user(arg, &features, sizeof(features))) | |
4632 | return -EFAULT; | |
4633 | ||
4634 | return 0; | |
4635 | } | |
4636 | ||
b929c1d8 MPS |
4637 | static int btrfs_ioctl_get_features(struct btrfs_fs_info *fs_info, |
4638 | void __user *arg) | |
2eaa055f | 4639 | { |
0b246afa | 4640 | struct btrfs_super_block *super_block = fs_info->super_copy; |
2eaa055f JM |
4641 | struct btrfs_ioctl_feature_flags features; |
4642 | ||
4643 | features.compat_flags = btrfs_super_compat_flags(super_block); | |
4644 | features.compat_ro_flags = btrfs_super_compat_ro_flags(super_block); | |
4645 | features.incompat_flags = btrfs_super_incompat_flags(super_block); | |
4646 | ||
4647 | if (copy_to_user(arg, &features, sizeof(features))) | |
4648 | return -EFAULT; | |
4649 | ||
4650 | return 0; | |
4651 | } | |
4652 | ||
2ff7e61e | 4653 | static int check_feature_bits(struct btrfs_fs_info *fs_info, |
3b02a68a | 4654 | enum btrfs_feature_set set, |
2eaa055f JM |
4655 | u64 change_mask, u64 flags, u64 supported_flags, |
4656 | u64 safe_set, u64 safe_clear) | |
4657 | { | |
f10152bc | 4658 | const char *type = btrfs_feature_set_name(set); |
3b02a68a | 4659 | char *names; |
2eaa055f JM |
4660 | u64 disallowed, unsupported; |
4661 | u64 set_mask = flags & change_mask; | |
4662 | u64 clear_mask = ~flags & change_mask; | |
4663 | ||
4664 | unsupported = set_mask & ~supported_flags; | |
4665 | if (unsupported) { | |
3b02a68a JM |
4666 | names = btrfs_printable_features(set, unsupported); |
4667 | if (names) { | |
0b246afa JM |
4668 | btrfs_warn(fs_info, |
4669 | "this kernel does not support the %s feature bit%s", | |
4670 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4671 | kfree(names); |
4672 | } else | |
0b246afa JM |
4673 | btrfs_warn(fs_info, |
4674 | "this kernel does not support %s bits 0x%llx", | |
4675 | type, unsupported); | |
2eaa055f JM |
4676 | return -EOPNOTSUPP; |
4677 | } | |
4678 | ||
4679 | disallowed = set_mask & ~safe_set; | |
4680 | if (disallowed) { | |
3b02a68a JM |
4681 | names = btrfs_printable_features(set, disallowed); |
4682 | if (names) { | |
0b246afa JM |
4683 | btrfs_warn(fs_info, |
4684 | "can't set the %s feature bit%s while mounted", | |
4685 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4686 | kfree(names); |
4687 | } else | |
0b246afa JM |
4688 | btrfs_warn(fs_info, |
4689 | "can't set %s bits 0x%llx while mounted", | |
4690 | type, disallowed); | |
2eaa055f JM |
4691 | return -EPERM; |
4692 | } | |
4693 | ||
4694 | disallowed = clear_mask & ~safe_clear; | |
4695 | if (disallowed) { | |
3b02a68a JM |
4696 | names = btrfs_printable_features(set, disallowed); |
4697 | if (names) { | |
0b246afa JM |
4698 | btrfs_warn(fs_info, |
4699 | "can't clear the %s feature bit%s while mounted", | |
4700 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4701 | kfree(names); |
4702 | } else | |
0b246afa JM |
4703 | btrfs_warn(fs_info, |
4704 | "can't clear %s bits 0x%llx while mounted", | |
4705 | type, disallowed); | |
2eaa055f JM |
4706 | return -EPERM; |
4707 | } | |
4708 | ||
4709 | return 0; | |
4710 | } | |
4711 | ||
2ff7e61e JM |
4712 | #define check_feature(fs_info, change_mask, flags, mask_base) \ |
4713 | check_feature_bits(fs_info, FEAT_##mask_base, change_mask, flags, \ | |
2eaa055f JM |
4714 | BTRFS_FEATURE_ ## mask_base ## _SUPP, \ |
4715 | BTRFS_FEATURE_ ## mask_base ## _SAFE_SET, \ | |
4716 | BTRFS_FEATURE_ ## mask_base ## _SAFE_CLEAR) | |
4717 | ||
4718 | static int btrfs_ioctl_set_features(struct file *file, void __user *arg) | |
4719 | { | |
0b246afa JM |
4720 | struct inode *inode = file_inode(file); |
4721 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4722 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4723 | struct btrfs_super_block *super_block = fs_info->super_copy; | |
2eaa055f JM |
4724 | struct btrfs_ioctl_feature_flags flags[2]; |
4725 | struct btrfs_trans_handle *trans; | |
4726 | u64 newflags; | |
4727 | int ret; | |
4728 | ||
4729 | if (!capable(CAP_SYS_ADMIN)) | |
4730 | return -EPERM; | |
4731 | ||
4732 | if (copy_from_user(flags, arg, sizeof(flags))) | |
4733 | return -EFAULT; | |
4734 | ||
4735 | /* Nothing to do */ | |
4736 | if (!flags[0].compat_flags && !flags[0].compat_ro_flags && | |
4737 | !flags[0].incompat_flags) | |
4738 | return 0; | |
4739 | ||
2ff7e61e | 4740 | ret = check_feature(fs_info, flags[0].compat_flags, |
2eaa055f JM |
4741 | flags[1].compat_flags, COMPAT); |
4742 | if (ret) | |
4743 | return ret; | |
4744 | ||
2ff7e61e | 4745 | ret = check_feature(fs_info, flags[0].compat_ro_flags, |
2eaa055f JM |
4746 | flags[1].compat_ro_flags, COMPAT_RO); |
4747 | if (ret) | |
4748 | return ret; | |
4749 | ||
2ff7e61e | 4750 | ret = check_feature(fs_info, flags[0].incompat_flags, |
2eaa055f JM |
4751 | flags[1].incompat_flags, INCOMPAT); |
4752 | if (ret) | |
4753 | return ret; | |
4754 | ||
7ab19625 DS |
4755 | ret = mnt_want_write_file(file); |
4756 | if (ret) | |
4757 | return ret; | |
4758 | ||
8051aa1a | 4759 | trans = btrfs_start_transaction(root, 0); |
7ab19625 DS |
4760 | if (IS_ERR(trans)) { |
4761 | ret = PTR_ERR(trans); | |
4762 | goto out_drop_write; | |
4763 | } | |
2eaa055f | 4764 | |
0b246afa | 4765 | spin_lock(&fs_info->super_lock); |
2eaa055f JM |
4766 | newflags = btrfs_super_compat_flags(super_block); |
4767 | newflags |= flags[0].compat_flags & flags[1].compat_flags; | |
4768 | newflags &= ~(flags[0].compat_flags & ~flags[1].compat_flags); | |
4769 | btrfs_set_super_compat_flags(super_block, newflags); | |
4770 | ||
4771 | newflags = btrfs_super_compat_ro_flags(super_block); | |
4772 | newflags |= flags[0].compat_ro_flags & flags[1].compat_ro_flags; | |
4773 | newflags &= ~(flags[0].compat_ro_flags & ~flags[1].compat_ro_flags); | |
4774 | btrfs_set_super_compat_ro_flags(super_block, newflags); | |
4775 | ||
4776 | newflags = btrfs_super_incompat_flags(super_block); | |
4777 | newflags |= flags[0].incompat_flags & flags[1].incompat_flags; | |
4778 | newflags &= ~(flags[0].incompat_flags & ~flags[1].incompat_flags); | |
4779 | btrfs_set_super_incompat_flags(super_block, newflags); | |
0b246afa | 4780 | spin_unlock(&fs_info->super_lock); |
2eaa055f | 4781 | |
3a45bb20 | 4782 | ret = btrfs_commit_transaction(trans); |
7ab19625 DS |
4783 | out_drop_write: |
4784 | mnt_drop_write_file(file); | |
4785 | ||
4786 | return ret; | |
2eaa055f JM |
4787 | } |
4788 | ||
2351f431 JB |
4789 | static int _btrfs_ioctl_send(struct file *file, void __user *argp, bool compat) |
4790 | { | |
4791 | struct btrfs_ioctl_send_args *arg; | |
4792 | int ret; | |
4793 | ||
4794 | if (compat) { | |
4795 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) | |
4796 | struct btrfs_ioctl_send_args_32 args32; | |
4797 | ||
4798 | ret = copy_from_user(&args32, argp, sizeof(args32)); | |
4799 | if (ret) | |
4800 | return -EFAULT; | |
4801 | arg = kzalloc(sizeof(*arg), GFP_KERNEL); | |
4802 | if (!arg) | |
4803 | return -ENOMEM; | |
4804 | arg->send_fd = args32.send_fd; | |
4805 | arg->clone_sources_count = args32.clone_sources_count; | |
4806 | arg->clone_sources = compat_ptr(args32.clone_sources); | |
4807 | arg->parent_root = args32.parent_root; | |
4808 | arg->flags = args32.flags; | |
4809 | memcpy(arg->reserved, args32.reserved, | |
4810 | sizeof(args32.reserved)); | |
4811 | #else | |
4812 | return -ENOTTY; | |
4813 | #endif | |
4814 | } else { | |
4815 | arg = memdup_user(argp, sizeof(*arg)); | |
4816 | if (IS_ERR(arg)) | |
4817 | return PTR_ERR(arg); | |
4818 | } | |
4819 | ret = btrfs_ioctl_send(file, arg); | |
4820 | kfree(arg); | |
4821 | return ret; | |
4822 | } | |
4823 | ||
f46b5a66 CH |
4824 | long btrfs_ioctl(struct file *file, unsigned int |
4825 | cmd, unsigned long arg) | |
4826 | { | |
0b246afa JM |
4827 | struct inode *inode = file_inode(file); |
4828 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4829 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4bcabaa3 | 4830 | void __user *argp = (void __user *)arg; |
f46b5a66 CH |
4831 | |
4832 | switch (cmd) { | |
6cbff00f CH |
4833 | case FS_IOC_GETVERSION: |
4834 | return btrfs_ioctl_getversion(file, argp); | |
40cf931f | 4835 | case FS_IOC_GETFSLABEL: |
b929c1d8 | 4836 | return btrfs_ioctl_get_fslabel(fs_info, argp); |
40cf931f ES |
4837 | case FS_IOC_SETFSLABEL: |
4838 | return btrfs_ioctl_set_fslabel(file, argp); | |
f7039b1d | 4839 | case FITRIM: |
b929c1d8 | 4840 | return btrfs_ioctl_fitrim(fs_info, argp); |
f46b5a66 | 4841 | case BTRFS_IOC_SNAP_CREATE: |
fa0d2b9b | 4842 | return btrfs_ioctl_snap_create(file, argp, 0); |
fdfb1e4f | 4843 | case BTRFS_IOC_SNAP_CREATE_V2: |
fa0d2b9b | 4844 | return btrfs_ioctl_snap_create_v2(file, argp, 0); |
3de4586c | 4845 | case BTRFS_IOC_SUBVOL_CREATE: |
fa0d2b9b | 4846 | return btrfs_ioctl_snap_create(file, argp, 1); |
6f72c7e2 AJ |
4847 | case BTRFS_IOC_SUBVOL_CREATE_V2: |
4848 | return btrfs_ioctl_snap_create_v2(file, argp, 1); | |
76dda93c | 4849 | case BTRFS_IOC_SNAP_DESTROY: |
949964c9 MPS |
4850 | return btrfs_ioctl_snap_destroy(file, argp, false); |
4851 | case BTRFS_IOC_SNAP_DESTROY_V2: | |
4852 | return btrfs_ioctl_snap_destroy(file, argp, true); | |
0caa102d LZ |
4853 | case BTRFS_IOC_SUBVOL_GETFLAGS: |
4854 | return btrfs_ioctl_subvol_getflags(file, argp); | |
4855 | case BTRFS_IOC_SUBVOL_SETFLAGS: | |
4856 | return btrfs_ioctl_subvol_setflags(file, argp); | |
6ef5ed0d JB |
4857 | case BTRFS_IOC_DEFAULT_SUBVOL: |
4858 | return btrfs_ioctl_default_subvol(file, argp); | |
f46b5a66 | 4859 | case BTRFS_IOC_DEFRAG: |
1e701a32 CM |
4860 | return btrfs_ioctl_defrag(file, NULL); |
4861 | case BTRFS_IOC_DEFRAG_RANGE: | |
4862 | return btrfs_ioctl_defrag(file, argp); | |
f46b5a66 | 4863 | case BTRFS_IOC_RESIZE: |
198605a8 | 4864 | return btrfs_ioctl_resize(file, argp); |
f46b5a66 | 4865 | case BTRFS_IOC_ADD_DEV: |
2ff7e61e | 4866 | return btrfs_ioctl_add_dev(fs_info, argp); |
f46b5a66 | 4867 | case BTRFS_IOC_RM_DEV: |
da24927b | 4868 | return btrfs_ioctl_rm_dev(file, argp); |
6b526ed7 AJ |
4869 | case BTRFS_IOC_RM_DEV_V2: |
4870 | return btrfs_ioctl_rm_dev_v2(file, argp); | |
475f6387 | 4871 | case BTRFS_IOC_FS_INFO: |
2ff7e61e | 4872 | return btrfs_ioctl_fs_info(fs_info, argp); |
475f6387 | 4873 | case BTRFS_IOC_DEV_INFO: |
2ff7e61e | 4874 | return btrfs_ioctl_dev_info(fs_info, argp); |
f46b5a66 | 4875 | case BTRFS_IOC_BALANCE: |
9ba1f6e4 | 4876 | return btrfs_ioctl_balance(file, NULL); |
ac8e9819 CM |
4877 | case BTRFS_IOC_TREE_SEARCH: |
4878 | return btrfs_ioctl_tree_search(file, argp); | |
cc68a8a5 GH |
4879 | case BTRFS_IOC_TREE_SEARCH_V2: |
4880 | return btrfs_ioctl_tree_search_v2(file, argp); | |
ac8e9819 CM |
4881 | case BTRFS_IOC_INO_LOOKUP: |
4882 | return btrfs_ioctl_ino_lookup(file, argp); | |
d7728c96 JS |
4883 | case BTRFS_IOC_INO_PATHS: |
4884 | return btrfs_ioctl_ino_to_path(root, argp); | |
4885 | case BTRFS_IOC_LOGICAL_INO: | |
d24a67b2 ZB |
4886 | return btrfs_ioctl_logical_to_ino(fs_info, argp, 1); |
4887 | case BTRFS_IOC_LOGICAL_INO_V2: | |
4888 | return btrfs_ioctl_logical_to_ino(fs_info, argp, 2); | |
1406e432 | 4889 | case BTRFS_IOC_SPACE_INFO: |
2ff7e61e | 4890 | return btrfs_ioctl_space_info(fs_info, argp); |
9b199859 FDBM |
4891 | case BTRFS_IOC_SYNC: { |
4892 | int ret; | |
4893 | ||
9db4dc24 | 4894 | ret = btrfs_start_delalloc_roots(fs_info, LONG_MAX, false); |
9b199859 FDBM |
4895 | if (ret) |
4896 | return ret; | |
0b246afa | 4897 | ret = btrfs_sync_fs(inode->i_sb, 1); |
2fad4e83 DS |
4898 | /* |
4899 | * The transaction thread may want to do more work, | |
01327610 | 4900 | * namely it pokes the cleaner kthread that will start |
2fad4e83 DS |
4901 | * processing uncleaned subvols. |
4902 | */ | |
0b246afa | 4903 | wake_up_process(fs_info->transaction_kthread); |
9b199859 FDBM |
4904 | return ret; |
4905 | } | |
46204592 | 4906 | case BTRFS_IOC_START_SYNC: |
9a8c28be | 4907 | return btrfs_ioctl_start_sync(root, argp); |
46204592 | 4908 | case BTRFS_IOC_WAIT_SYNC: |
2ff7e61e | 4909 | return btrfs_ioctl_wait_sync(fs_info, argp); |
475f6387 | 4910 | case BTRFS_IOC_SCRUB: |
b8e95489 | 4911 | return btrfs_ioctl_scrub(file, argp); |
475f6387 | 4912 | case BTRFS_IOC_SCRUB_CANCEL: |
2ff7e61e | 4913 | return btrfs_ioctl_scrub_cancel(fs_info); |
475f6387 | 4914 | case BTRFS_IOC_SCRUB_PROGRESS: |
2ff7e61e | 4915 | return btrfs_ioctl_scrub_progress(fs_info, argp); |
c9e9f97b | 4916 | case BTRFS_IOC_BALANCE_V2: |
9ba1f6e4 | 4917 | return btrfs_ioctl_balance(file, argp); |
837d5b6e | 4918 | case BTRFS_IOC_BALANCE_CTL: |
2ff7e61e | 4919 | return btrfs_ioctl_balance_ctl(fs_info, arg); |
19a39dce | 4920 | case BTRFS_IOC_BALANCE_PROGRESS: |
2ff7e61e | 4921 | return btrfs_ioctl_balance_progress(fs_info, argp); |
8ea05e3a AB |
4922 | case BTRFS_IOC_SET_RECEIVED_SUBVOL: |
4923 | return btrfs_ioctl_set_received_subvol(file, argp); | |
abccd00f HM |
4924 | #ifdef CONFIG_64BIT |
4925 | case BTRFS_IOC_SET_RECEIVED_SUBVOL_32: | |
4926 | return btrfs_ioctl_set_received_subvol_32(file, argp); | |
4927 | #endif | |
31db9f7c | 4928 | case BTRFS_IOC_SEND: |
2351f431 JB |
4929 | return _btrfs_ioctl_send(file, argp, false); |
4930 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) | |
4931 | case BTRFS_IOC_SEND_32: | |
4932 | return _btrfs_ioctl_send(file, argp, true); | |
4933 | #endif | |
c11d2c23 | 4934 | case BTRFS_IOC_GET_DEV_STATS: |
2ff7e61e | 4935 | return btrfs_ioctl_get_dev_stats(fs_info, argp); |
5d13a37b | 4936 | case BTRFS_IOC_QUOTA_CTL: |
905b0dda | 4937 | return btrfs_ioctl_quota_ctl(file, argp); |
5d13a37b | 4938 | case BTRFS_IOC_QGROUP_ASSIGN: |
905b0dda | 4939 | return btrfs_ioctl_qgroup_assign(file, argp); |
5d13a37b | 4940 | case BTRFS_IOC_QGROUP_CREATE: |
905b0dda | 4941 | return btrfs_ioctl_qgroup_create(file, argp); |
5d13a37b | 4942 | case BTRFS_IOC_QGROUP_LIMIT: |
905b0dda | 4943 | return btrfs_ioctl_qgroup_limit(file, argp); |
2f232036 JS |
4944 | case BTRFS_IOC_QUOTA_RESCAN: |
4945 | return btrfs_ioctl_quota_rescan(file, argp); | |
4946 | case BTRFS_IOC_QUOTA_RESCAN_STATUS: | |
b929c1d8 | 4947 | return btrfs_ioctl_quota_rescan_status(fs_info, argp); |
57254b6e | 4948 | case BTRFS_IOC_QUOTA_RESCAN_WAIT: |
b929c1d8 | 4949 | return btrfs_ioctl_quota_rescan_wait(fs_info, argp); |
3f6bcfbd | 4950 | case BTRFS_IOC_DEV_REPLACE: |
2ff7e61e | 4951 | return btrfs_ioctl_dev_replace(fs_info, argp); |
2eaa055f | 4952 | case BTRFS_IOC_GET_SUPPORTED_FEATURES: |
d5131b65 | 4953 | return btrfs_ioctl_get_supported_features(argp); |
2eaa055f | 4954 | case BTRFS_IOC_GET_FEATURES: |
b929c1d8 | 4955 | return btrfs_ioctl_get_features(fs_info, argp); |
2eaa055f JM |
4956 | case BTRFS_IOC_SET_FEATURES: |
4957 | return btrfs_ioctl_set_features(file, argp); | |
b64ec075 TM |
4958 | case BTRFS_IOC_GET_SUBVOL_INFO: |
4959 | return btrfs_ioctl_get_subvol_info(file, argp); | |
42e4b520 TM |
4960 | case BTRFS_IOC_GET_SUBVOL_ROOTREF: |
4961 | return btrfs_ioctl_get_subvol_rootref(file, argp); | |
23d0b79d TM |
4962 | case BTRFS_IOC_INO_LOOKUP_USER: |
4963 | return btrfs_ioctl_ino_lookup_user(file, argp); | |
f46b5a66 CH |
4964 | } |
4965 | ||
4966 | return -ENOTTY; | |
4967 | } | |
4c63c245 LD |
4968 | |
4969 | #ifdef CONFIG_COMPAT | |
4970 | long btrfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg) | |
4971 | { | |
2a362249 JM |
4972 | /* |
4973 | * These all access 32-bit values anyway so no further | |
4974 | * handling is necessary. | |
4975 | */ | |
4c63c245 | 4976 | switch (cmd) { |
4c63c245 LD |
4977 | case FS_IOC32_GETVERSION: |
4978 | cmd = FS_IOC_GETVERSION; | |
4979 | break; | |
4c63c245 LD |
4980 | } |
4981 | ||
4982 | return btrfs_ioctl(file, cmd, (unsigned long) compat_ptr(arg)); | |
4983 | } | |
4984 | #endif |