]>
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> |
f46b5a66 CH |
29 | #include "ctree.h" |
30 | #include "disk-io.h" | |
949964c9 | 31 | #include "export.h" |
f46b5a66 CH |
32 | #include "transaction.h" |
33 | #include "btrfs_inode.h" | |
f46b5a66 CH |
34 | #include "print-tree.h" |
35 | #include "volumes.h" | |
925baedd | 36 | #include "locking.h" |
581bb050 | 37 | #include "inode-map.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 | ||
6cbff00f CH |
157 | static int btrfs_ioctl_getflags(struct file *file, void __user *arg) |
158 | { | |
5c57b8b6 DS |
159 | struct btrfs_inode *binode = BTRFS_I(file_inode(file)); |
160 | unsigned int flags = btrfs_inode_flags_to_fsflags(binode->flags); | |
6cbff00f CH |
161 | |
162 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
163 | return -EFAULT; | |
164 | return 0; | |
165 | } | |
166 | ||
f37c563b DS |
167 | /* |
168 | * Check if @flags are a supported and valid set of FS_*_FL flags and that | |
169 | * the old and new flags are not conflicting | |
170 | */ | |
171 | static int check_fsflags(unsigned int old_flags, unsigned int flags) | |
75e7cb7f LB |
172 | { |
173 | if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \ | |
174 | FS_NOATIME_FL | FS_NODUMP_FL | \ | |
175 | FS_SYNC_FL | FS_DIRSYNC_FL | \ | |
e1e8fb6a LZ |
176 | FS_NOCOMP_FL | FS_COMPR_FL | |
177 | FS_NOCOW_FL)) | |
75e7cb7f LB |
178 | return -EOPNOTSUPP; |
179 | ||
f37c563b | 180 | /* COMPR and NOCOMP on new/old are valid */ |
75e7cb7f LB |
181 | if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL)) |
182 | return -EINVAL; | |
183 | ||
f37c563b DS |
184 | if ((flags & FS_COMPR_FL) && (flags & FS_NOCOW_FL)) |
185 | return -EINVAL; | |
186 | ||
187 | /* NOCOW and compression options are mutually exclusive */ | |
188 | if ((old_flags & FS_NOCOW_FL) && (flags & (FS_COMPR_FL | FS_NOCOMP_FL))) | |
189 | return -EINVAL; | |
190 | if ((flags & FS_NOCOW_FL) && (old_flags & (FS_COMPR_FL | FS_NOCOMP_FL))) | |
191 | return -EINVAL; | |
192 | ||
75e7cb7f LB |
193 | return 0; |
194 | } | |
195 | ||
6cbff00f CH |
196 | static int btrfs_ioctl_setflags(struct file *file, void __user *arg) |
197 | { | |
496ad9aa | 198 | struct inode *inode = file_inode(file); |
0b246afa | 199 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
5c57b8b6 DS |
200 | struct btrfs_inode *binode = BTRFS_I(inode); |
201 | struct btrfs_root *root = binode->root; | |
6cbff00f | 202 | struct btrfs_trans_handle *trans; |
5aca2842 | 203 | unsigned int fsflags, old_fsflags; |
6cbff00f | 204 | int ret; |
ff9fef55 | 205 | const char *comp = NULL; |
f37c563b | 206 | u32 binode_flags; |
6cbff00f | 207 | |
bd60ea0f DS |
208 | if (!inode_owner_or_capable(inode)) |
209 | return -EPERM; | |
210 | ||
b83cc969 LZ |
211 | if (btrfs_root_readonly(root)) |
212 | return -EROFS; | |
213 | ||
5c57b8b6 | 214 | if (copy_from_user(&fsflags, arg, sizeof(fsflags))) |
6cbff00f CH |
215 | return -EFAULT; |
216 | ||
e7848683 JK |
217 | ret = mnt_want_write_file(file); |
218 | if (ret) | |
219 | return ret; | |
220 | ||
5955102c | 221 | inode_lock(inode); |
5c57b8b6 | 222 | fsflags = btrfs_mask_fsflags_for_type(inode, fsflags); |
5aca2842 | 223 | old_fsflags = btrfs_inode_flags_to_fsflags(binode->flags); |
f37c563b | 224 | |
5aca2842 DW |
225 | ret = vfs_ioc_setflags_prepare(inode, old_fsflags, fsflags); |
226 | if (ret) | |
227 | goto out_unlock; | |
6cbff00f | 228 | |
f37c563b DS |
229 | ret = check_fsflags(old_fsflags, fsflags); |
230 | if (ret) | |
231 | goto out_unlock; | |
232 | ||
233 | binode_flags = binode->flags; | |
5c57b8b6 | 234 | if (fsflags & FS_SYNC_FL) |
d2b8fcfe | 235 | binode_flags |= BTRFS_INODE_SYNC; |
6cbff00f | 236 | else |
d2b8fcfe | 237 | binode_flags &= ~BTRFS_INODE_SYNC; |
5c57b8b6 | 238 | if (fsflags & FS_IMMUTABLE_FL) |
d2b8fcfe | 239 | binode_flags |= BTRFS_INODE_IMMUTABLE; |
6cbff00f | 240 | else |
d2b8fcfe | 241 | binode_flags &= ~BTRFS_INODE_IMMUTABLE; |
5c57b8b6 | 242 | if (fsflags & FS_APPEND_FL) |
d2b8fcfe | 243 | binode_flags |= BTRFS_INODE_APPEND; |
6cbff00f | 244 | else |
d2b8fcfe | 245 | binode_flags &= ~BTRFS_INODE_APPEND; |
5c57b8b6 | 246 | if (fsflags & FS_NODUMP_FL) |
d2b8fcfe | 247 | binode_flags |= BTRFS_INODE_NODUMP; |
6cbff00f | 248 | else |
d2b8fcfe | 249 | binode_flags &= ~BTRFS_INODE_NODUMP; |
5c57b8b6 | 250 | if (fsflags & FS_NOATIME_FL) |
d2b8fcfe | 251 | binode_flags |= BTRFS_INODE_NOATIME; |
6cbff00f | 252 | else |
d2b8fcfe | 253 | binode_flags &= ~BTRFS_INODE_NOATIME; |
5c57b8b6 | 254 | if (fsflags & FS_DIRSYNC_FL) |
d2b8fcfe | 255 | binode_flags |= BTRFS_INODE_DIRSYNC; |
6cbff00f | 256 | else |
d2b8fcfe | 257 | binode_flags &= ~BTRFS_INODE_DIRSYNC; |
5c57b8b6 | 258 | if (fsflags & FS_NOCOW_FL) { |
44e5194b | 259 | if (S_ISREG(inode->i_mode)) { |
7e97b8da DS |
260 | /* |
261 | * It's safe to turn csums off here, no extents exist. | |
262 | * Otherwise we want the flag to reflect the real COW | |
263 | * status of the file and will not set it. | |
264 | */ | |
265 | if (inode->i_size == 0) | |
d2b8fcfe AJ |
266 | binode_flags |= BTRFS_INODE_NODATACOW | |
267 | BTRFS_INODE_NODATASUM; | |
7e97b8da | 268 | } else { |
d2b8fcfe | 269 | binode_flags |= BTRFS_INODE_NODATACOW; |
7e97b8da DS |
270 | } |
271 | } else { | |
272 | /* | |
01327610 | 273 | * Revert back under same assumptions as above |
7e97b8da | 274 | */ |
44e5194b | 275 | if (S_ISREG(inode->i_mode)) { |
7e97b8da | 276 | if (inode->i_size == 0) |
d2b8fcfe AJ |
277 | binode_flags &= ~(BTRFS_INODE_NODATACOW | |
278 | BTRFS_INODE_NODATASUM); | |
7e97b8da | 279 | } else { |
d2b8fcfe | 280 | binode_flags &= ~BTRFS_INODE_NODATACOW; |
7e97b8da DS |
281 | } |
282 | } | |
6cbff00f | 283 | |
75e7cb7f LB |
284 | /* |
285 | * The COMPRESS flag can only be changed by users, while the NOCOMPRESS | |
286 | * flag may be changed automatically if compression code won't make | |
287 | * things smaller. | |
288 | */ | |
5c57b8b6 | 289 | if (fsflags & FS_NOCOMP_FL) { |
d2b8fcfe AJ |
290 | binode_flags &= ~BTRFS_INODE_COMPRESS; |
291 | binode_flags |= BTRFS_INODE_NOCOMPRESS; | |
5c57b8b6 | 292 | } else if (fsflags & FS_COMPR_FL) { |
63541927 | 293 | |
eede2bf3 OS |
294 | if (IS_SWAPFILE(inode)) { |
295 | ret = -ETXTBSY; | |
296 | goto out_unlock; | |
297 | } | |
298 | ||
d2b8fcfe AJ |
299 | binode_flags |= BTRFS_INODE_COMPRESS; |
300 | binode_flags &= ~BTRFS_INODE_NOCOMPRESS; | |
63541927 | 301 | |
93370509 DS |
302 | comp = btrfs_compress_type2str(fs_info->compress_type); |
303 | if (!comp || comp[0] == 0) | |
304 | comp = btrfs_compress_type2str(BTRFS_COMPRESS_ZLIB); | |
ebcb904d | 305 | } else { |
d2b8fcfe | 306 | binode_flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS); |
75e7cb7f | 307 | } |
6cbff00f | 308 | |
ff9fef55 AJ |
309 | /* |
310 | * 1 for inode item | |
311 | * 2 for properties | |
312 | */ | |
313 | trans = btrfs_start_transaction(root, 3); | |
f062abf0 LZ |
314 | if (IS_ERR(trans)) { |
315 | ret = PTR_ERR(trans); | |
d2b8fcfe | 316 | goto out_unlock; |
f062abf0 | 317 | } |
6cbff00f | 318 | |
ff9fef55 AJ |
319 | if (comp) { |
320 | ret = btrfs_set_prop(trans, inode, "btrfs.compression", comp, | |
321 | strlen(comp), 0); | |
322 | if (ret) { | |
323 | btrfs_abort_transaction(trans, ret); | |
324 | goto out_end_trans; | |
325 | } | |
ff9fef55 AJ |
326 | } else { |
327 | ret = btrfs_set_prop(trans, inode, "btrfs.compression", NULL, | |
328 | 0, 0); | |
329 | if (ret && ret != -ENODATA) { | |
330 | btrfs_abort_transaction(trans, ret); | |
331 | goto out_end_trans; | |
332 | } | |
333 | } | |
334 | ||
d2b8fcfe | 335 | binode->flags = binode_flags; |
7b6a221e | 336 | btrfs_sync_inode_flags_to_i_flags(inode); |
0c4d2d95 | 337 | inode_inc_iversion(inode); |
c2050a45 | 338 | inode->i_ctime = current_time(inode); |
6cbff00f | 339 | ret = btrfs_update_inode(trans, root, inode); |
6cbff00f | 340 | |
ff9fef55 | 341 | out_end_trans: |
3a45bb20 | 342 | btrfs_end_transaction(trans); |
6cbff00f | 343 | out_unlock: |
5955102c | 344 | inode_unlock(inode); |
e7848683 | 345 | mnt_drop_write_file(file); |
2d4e6f6a | 346 | return ret; |
6cbff00f CH |
347 | } |
348 | ||
19f93b3c DS |
349 | /* |
350 | * Translate btrfs internal inode flags to xflags as expected by the | |
351 | * FS_IOC_FSGETXATT ioctl. Filter only the supported ones, unknown flags are | |
352 | * silently dropped. | |
353 | */ | |
354 | static unsigned int btrfs_inode_flags_to_xflags(unsigned int flags) | |
355 | { | |
356 | unsigned int xflags = 0; | |
357 | ||
358 | if (flags & BTRFS_INODE_APPEND) | |
359 | xflags |= FS_XFLAG_APPEND; | |
360 | if (flags & BTRFS_INODE_IMMUTABLE) | |
361 | xflags |= FS_XFLAG_IMMUTABLE; | |
362 | if (flags & BTRFS_INODE_NOATIME) | |
363 | xflags |= FS_XFLAG_NOATIME; | |
364 | if (flags & BTRFS_INODE_NODUMP) | |
365 | xflags |= FS_XFLAG_NODUMP; | |
366 | if (flags & BTRFS_INODE_SYNC) | |
367 | xflags |= FS_XFLAG_SYNC; | |
368 | ||
369 | return xflags; | |
370 | } | |
371 | ||
372 | /* Check if @flags are a supported and valid set of FS_XFLAGS_* flags */ | |
373 | static int check_xflags(unsigned int flags) | |
374 | { | |
375 | if (flags & ~(FS_XFLAG_APPEND | FS_XFLAG_IMMUTABLE | FS_XFLAG_NOATIME | | |
376 | FS_XFLAG_NODUMP | FS_XFLAG_SYNC)) | |
377 | return -EOPNOTSUPP; | |
378 | return 0; | |
379 | } | |
380 | ||
c3e1f96c GR |
381 | bool btrfs_exclop_start(struct btrfs_fs_info *fs_info, |
382 | enum btrfs_exclusive_operation type) | |
383 | { | |
384 | return !cmpxchg(&fs_info->exclusive_operation, BTRFS_EXCLOP_NONE, type); | |
385 | } | |
386 | ||
387 | void btrfs_exclop_finish(struct btrfs_fs_info *fs_info) | |
388 | { | |
389 | WRITE_ONCE(fs_info->exclusive_operation, BTRFS_EXCLOP_NONE); | |
390 | } | |
391 | ||
e4202ac9 DS |
392 | /* |
393 | * Set the xflags from the internal inode flags. The remaining items of fsxattr | |
394 | * are zeroed. | |
395 | */ | |
396 | static int btrfs_ioctl_fsgetxattr(struct file *file, void __user *arg) | |
397 | { | |
398 | struct btrfs_inode *binode = BTRFS_I(file_inode(file)); | |
399 | struct fsxattr fa; | |
400 | ||
7b0e492e | 401 | simple_fill_fsxattr(&fa, btrfs_inode_flags_to_xflags(binode->flags)); |
e4202ac9 DS |
402 | if (copy_to_user(arg, &fa, sizeof(fa))) |
403 | return -EFAULT; | |
404 | ||
405 | return 0; | |
406 | } | |
407 | ||
025f2121 DS |
408 | static int btrfs_ioctl_fssetxattr(struct file *file, void __user *arg) |
409 | { | |
410 | struct inode *inode = file_inode(file); | |
411 | struct btrfs_inode *binode = BTRFS_I(inode); | |
412 | struct btrfs_root *root = binode->root; | |
413 | struct btrfs_trans_handle *trans; | |
7b0e492e | 414 | struct fsxattr fa, old_fa; |
025f2121 DS |
415 | unsigned old_flags; |
416 | unsigned old_i_flags; | |
417 | int ret = 0; | |
418 | ||
419 | if (!inode_owner_or_capable(inode)) | |
420 | return -EPERM; | |
421 | ||
422 | if (btrfs_root_readonly(root)) | |
423 | return -EROFS; | |
424 | ||
025f2121 DS |
425 | if (copy_from_user(&fa, arg, sizeof(fa))) |
426 | return -EFAULT; | |
427 | ||
428 | ret = check_xflags(fa.fsx_xflags); | |
429 | if (ret) | |
430 | return ret; | |
431 | ||
432 | if (fa.fsx_extsize != 0 || fa.fsx_projid != 0 || fa.fsx_cowextsize != 0) | |
433 | return -EOPNOTSUPP; | |
434 | ||
435 | ret = mnt_want_write_file(file); | |
436 | if (ret) | |
437 | return ret; | |
438 | ||
439 | inode_lock(inode); | |
440 | ||
441 | old_flags = binode->flags; | |
442 | old_i_flags = inode->i_flags; | |
443 | ||
7b0e492e DW |
444 | simple_fill_fsxattr(&old_fa, |
445 | btrfs_inode_flags_to_xflags(binode->flags)); | |
446 | ret = vfs_ioc_fssetxattr_check(inode, &old_fa, &fa); | |
447 | if (ret) | |
025f2121 | 448 | goto out_unlock; |
025f2121 DS |
449 | |
450 | if (fa.fsx_xflags & FS_XFLAG_SYNC) | |
451 | binode->flags |= BTRFS_INODE_SYNC; | |
452 | else | |
453 | binode->flags &= ~BTRFS_INODE_SYNC; | |
454 | if (fa.fsx_xflags & FS_XFLAG_IMMUTABLE) | |
455 | binode->flags |= BTRFS_INODE_IMMUTABLE; | |
456 | else | |
457 | binode->flags &= ~BTRFS_INODE_IMMUTABLE; | |
458 | if (fa.fsx_xflags & FS_XFLAG_APPEND) | |
459 | binode->flags |= BTRFS_INODE_APPEND; | |
460 | else | |
461 | binode->flags &= ~BTRFS_INODE_APPEND; | |
462 | if (fa.fsx_xflags & FS_XFLAG_NODUMP) | |
463 | binode->flags |= BTRFS_INODE_NODUMP; | |
464 | else | |
465 | binode->flags &= ~BTRFS_INODE_NODUMP; | |
466 | if (fa.fsx_xflags & FS_XFLAG_NOATIME) | |
467 | binode->flags |= BTRFS_INODE_NOATIME; | |
468 | else | |
469 | binode->flags &= ~BTRFS_INODE_NOATIME; | |
470 | ||
471 | /* 1 item for the inode */ | |
472 | trans = btrfs_start_transaction(root, 1); | |
473 | if (IS_ERR(trans)) { | |
474 | ret = PTR_ERR(trans); | |
475 | goto out_unlock; | |
476 | } | |
477 | ||
478 | btrfs_sync_inode_flags_to_i_flags(inode); | |
479 | inode_inc_iversion(inode); | |
480 | inode->i_ctime = current_time(inode); | |
481 | ret = btrfs_update_inode(trans, root, inode); | |
482 | ||
483 | btrfs_end_transaction(trans); | |
484 | ||
485 | out_unlock: | |
486 | if (ret) { | |
487 | binode->flags = old_flags; | |
488 | inode->i_flags = old_i_flags; | |
489 | } | |
490 | ||
491 | inode_unlock(inode); | |
492 | mnt_drop_write_file(file); | |
493 | ||
494 | return ret; | |
495 | } | |
496 | ||
6cbff00f CH |
497 | static int btrfs_ioctl_getversion(struct file *file, int __user *arg) |
498 | { | |
496ad9aa | 499 | struct inode *inode = file_inode(file); |
6cbff00f CH |
500 | |
501 | return put_user(inode->i_generation, arg); | |
502 | } | |
f46b5a66 | 503 | |
b929c1d8 MPS |
504 | static noinline int btrfs_ioctl_fitrim(struct btrfs_fs_info *fs_info, |
505 | void __user *arg) | |
f7039b1d | 506 | { |
f7039b1d LD |
507 | struct btrfs_device *device; |
508 | struct request_queue *q; | |
509 | struct fstrim_range range; | |
510 | u64 minlen = ULLONG_MAX; | |
511 | u64 num_devices = 0; | |
512 | int ret; | |
513 | ||
514 | if (!capable(CAP_SYS_ADMIN)) | |
515 | return -EPERM; | |
516 | ||
f35f06c3 FM |
517 | /* |
518 | * If the fs is mounted with nologreplay, which requires it to be | |
519 | * mounted in RO mode as well, we can not allow discard on free space | |
520 | * inside block groups, because log trees refer to extents that are not | |
521 | * pinned in a block group's free space cache (pinning the extents is | |
522 | * precisely the first phase of replaying a log tree). | |
523 | */ | |
524 | if (btrfs_test_opt(fs_info, NOLOGREPLAY)) | |
525 | return -EROFS; | |
526 | ||
1f78160c XG |
527 | rcu_read_lock(); |
528 | list_for_each_entry_rcu(device, &fs_info->fs_devices->devices, | |
529 | dev_list) { | |
f7039b1d LD |
530 | if (!device->bdev) |
531 | continue; | |
532 | q = bdev_get_queue(device->bdev); | |
533 | if (blk_queue_discard(q)) { | |
534 | num_devices++; | |
50d0446e | 535 | minlen = min_t(u64, q->limits.discard_granularity, |
f7039b1d LD |
536 | minlen); |
537 | } | |
538 | } | |
1f78160c | 539 | rcu_read_unlock(); |
f4c697e6 | 540 | |
f7039b1d LD |
541 | if (!num_devices) |
542 | return -EOPNOTSUPP; | |
f7039b1d LD |
543 | if (copy_from_user(&range, arg, sizeof(range))) |
544 | return -EFAULT; | |
6ba9fc8e QW |
545 | |
546 | /* | |
547 | * NOTE: Don't truncate the range using super->total_bytes. Bytenr of | |
548 | * block group is in the logical address space, which can be any | |
549 | * sectorsize aligned bytenr in the range [0, U64_MAX]. | |
550 | */ | |
551 | if (range.len < fs_info->sb->s_blocksize) | |
f4c697e6 | 552 | return -EINVAL; |
f7039b1d LD |
553 | |
554 | range.minlen = max(range.minlen, minlen); | |
2ff7e61e | 555 | ret = btrfs_trim_fs(fs_info, &range); |
f7039b1d LD |
556 | if (ret < 0) |
557 | return ret; | |
558 | ||
559 | if (copy_to_user(arg, &range, sizeof(range))) | |
560 | return -EFAULT; | |
561 | ||
562 | return 0; | |
563 | } | |
564 | ||
e1f60a65 | 565 | int __pure btrfs_is_empty_uuid(u8 *uuid) |
dd5f9615 | 566 | { |
46e0f66a CM |
567 | int i; |
568 | ||
569 | for (i = 0; i < BTRFS_UUID_SIZE; i++) { | |
570 | if (uuid[i]) | |
571 | return 0; | |
572 | } | |
573 | return 1; | |
dd5f9615 SB |
574 | } |
575 | ||
d5c12070 | 576 | static noinline int create_subvol(struct inode *dir, |
cb8e7090 | 577 | struct dentry *dentry, |
52f75f4f | 578 | const char *name, int namelen, |
8696c533 | 579 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 580 | { |
0b246afa | 581 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); |
f46b5a66 CH |
582 | struct btrfs_trans_handle *trans; |
583 | struct btrfs_key key; | |
49a3c4d9 | 584 | struct btrfs_root_item *root_item; |
f46b5a66 CH |
585 | struct btrfs_inode_item *inode_item; |
586 | struct extent_buffer *leaf; | |
d5c12070 | 587 | struct btrfs_root *root = BTRFS_I(dir)->root; |
76dda93c | 588 | struct btrfs_root *new_root; |
d5c12070 | 589 | struct btrfs_block_rsv block_rsv; |
95582b00 | 590 | struct timespec64 cur_time = current_time(dir); |
5662344b | 591 | struct inode *inode; |
f46b5a66 CH |
592 | int ret; |
593 | int err; | |
2dfb1e43 | 594 | dev_t anon_dev = 0; |
f46b5a66 CH |
595 | u64 objectid; |
596 | u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID; | |
3de4586c | 597 | u64 index = 0; |
f46b5a66 | 598 | |
49a3c4d9 DS |
599 | root_item = kzalloc(sizeof(*root_item), GFP_KERNEL); |
600 | if (!root_item) | |
601 | return -ENOMEM; | |
602 | ||
0b246afa | 603 | ret = btrfs_find_free_objectid(fs_info->tree_root, &objectid); |
2fbe8c8a | 604 | if (ret) |
49a3c4d9 | 605 | goto fail_free; |
6a912213 | 606 | |
2dfb1e43 QW |
607 | ret = get_anon_bdev(&anon_dev); |
608 | if (ret < 0) | |
609 | goto fail_free; | |
610 | ||
e09fe2d2 QW |
611 | /* |
612 | * Don't create subvolume whose level is not zero. Or qgroup will be | |
01327610 | 613 | * screwed up since it assumes subvolume qgroup's level to be 0. |
e09fe2d2 | 614 | */ |
49a3c4d9 DS |
615 | if (btrfs_qgroup_level(objectid)) { |
616 | ret = -ENOSPC; | |
617 | goto fail_free; | |
618 | } | |
e09fe2d2 | 619 | |
d5c12070 | 620 | btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP); |
9ed74f2d | 621 | /* |
d5c12070 MX |
622 | * The same as the snapshot creation, please see the comment |
623 | * of create_snapshot(). | |
9ed74f2d | 624 | */ |
c4c129db | 625 | ret = btrfs_subvolume_reserve_metadata(root, &block_rsv, 8, false); |
d5c12070 | 626 | if (ret) |
49a3c4d9 | 627 | goto fail_free; |
d5c12070 MX |
628 | |
629 | trans = btrfs_start_transaction(root, 0); | |
630 | if (IS_ERR(trans)) { | |
631 | ret = PTR_ERR(trans); | |
e85fde51 | 632 | btrfs_subvolume_release_metadata(root, &block_rsv); |
49a3c4d9 | 633 | goto fail_free; |
d5c12070 MX |
634 | } |
635 | trans->block_rsv = &block_rsv; | |
636 | trans->bytes_reserved = block_rsv.size; | |
f46b5a66 | 637 | |
a9377422 | 638 | ret = btrfs_qgroup_inherit(trans, 0, objectid, inherit); |
6f72c7e2 AJ |
639 | if (ret) |
640 | goto fail; | |
641 | ||
9631e4cc JB |
642 | leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0, |
643 | BTRFS_NESTING_NORMAL); | |
8e8a1e31 JB |
644 | if (IS_ERR(leaf)) { |
645 | ret = PTR_ERR(leaf); | |
646 | goto fail; | |
647 | } | |
f46b5a66 | 648 | |
f46b5a66 CH |
649 | btrfs_mark_buffer_dirty(leaf); |
650 | ||
49a3c4d9 | 651 | inode_item = &root_item->inode; |
3cae210f QW |
652 | btrfs_set_stack_inode_generation(inode_item, 1); |
653 | btrfs_set_stack_inode_size(inode_item, 3); | |
654 | btrfs_set_stack_inode_nlink(inode_item, 1); | |
da17066c | 655 | btrfs_set_stack_inode_nbytes(inode_item, |
0b246afa | 656 | fs_info->nodesize); |
3cae210f | 657 | btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755); |
f46b5a66 | 658 | |
49a3c4d9 DS |
659 | btrfs_set_root_flags(root_item, 0); |
660 | btrfs_set_root_limit(root_item, 0); | |
3cae210f | 661 | btrfs_set_stack_inode_flags(inode_item, BTRFS_INODE_ROOT_ITEM_INIT); |
08fe4db1 | 662 | |
49a3c4d9 DS |
663 | btrfs_set_root_bytenr(root_item, leaf->start); |
664 | btrfs_set_root_generation(root_item, trans->transid); | |
665 | btrfs_set_root_level(root_item, 0); | |
666 | btrfs_set_root_refs(root_item, 1); | |
667 | btrfs_set_root_used(root_item, leaf->len); | |
668 | btrfs_set_root_last_snapshot(root_item, 0); | |
f46b5a66 | 669 | |
49a3c4d9 DS |
670 | btrfs_set_root_generation_v2(root_item, |
671 | btrfs_root_generation(root_item)); | |
807fc790 | 672 | generate_random_guid(root_item->uuid); |
49a3c4d9 DS |
673 | btrfs_set_stack_timespec_sec(&root_item->otime, cur_time.tv_sec); |
674 | btrfs_set_stack_timespec_nsec(&root_item->otime, cur_time.tv_nsec); | |
675 | root_item->ctime = root_item->otime; | |
676 | btrfs_set_root_ctransid(root_item, trans->transid); | |
677 | btrfs_set_root_otransid(root_item, trans->transid); | |
f46b5a66 | 678 | |
925baedd | 679 | btrfs_tree_unlock(leaf); |
f46b5a66 CH |
680 | free_extent_buffer(leaf); |
681 | leaf = NULL; | |
682 | ||
49a3c4d9 | 683 | btrfs_set_root_dirid(root_item, new_dirid); |
f46b5a66 CH |
684 | |
685 | key.objectid = objectid; | |
5d4f98a2 | 686 | key.offset = 0; |
962a298f | 687 | key.type = BTRFS_ROOT_ITEM_KEY; |
0b246afa | 688 | ret = btrfs_insert_root(trans, fs_info->tree_root, &key, |
49a3c4d9 | 689 | root_item); |
f46b5a66 CH |
690 | if (ret) |
691 | goto fail; | |
692 | ||
76dda93c | 693 | key.offset = (u64)-1; |
2dfb1e43 | 694 | new_root = btrfs_get_new_fs_root(fs_info, objectid, anon_dev); |
79787eaa | 695 | if (IS_ERR(new_root)) { |
2dfb1e43 | 696 | free_anon_bdev(anon_dev); |
79787eaa | 697 | ret = PTR_ERR(new_root); |
66642832 | 698 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
699 | goto fail; |
700 | } | |
2dfb1e43 QW |
701 | /* Freeing will be done in btrfs_put_root() of new_root */ |
702 | anon_dev = 0; | |
76dda93c YZ |
703 | |
704 | btrfs_record_root_in_trans(trans, new_root); | |
705 | ||
63541927 | 706 | ret = btrfs_create_subvol_root(trans, new_root, root, new_dirid); |
00246528 | 707 | btrfs_put_root(new_root); |
ce598979 MF |
708 | if (ret) { |
709 | /* We potentially lose an unused inode item here */ | |
66642832 | 710 | btrfs_abort_transaction(trans, ret); |
ce598979 MF |
711 | goto fail; |
712 | } | |
713 | ||
f32e48e9 CR |
714 | mutex_lock(&new_root->objectid_mutex); |
715 | new_root->highest_objectid = new_dirid; | |
716 | mutex_unlock(&new_root->objectid_mutex); | |
717 | ||
f46b5a66 CH |
718 | /* |
719 | * insert the directory item | |
720 | */ | |
877574e2 | 721 | ret = btrfs_set_inode_index(BTRFS_I(dir), &index); |
79787eaa | 722 | if (ret) { |
66642832 | 723 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
724 | goto fail; |
725 | } | |
3de4586c | 726 | |
684572df | 727 | ret = btrfs_insert_dir_item(trans, name, namelen, BTRFS_I(dir), &key, |
3de4586c | 728 | BTRFS_FT_DIR, index); |
79787eaa | 729 | if (ret) { |
66642832 | 730 | btrfs_abort_transaction(trans, ret); |
f46b5a66 | 731 | goto fail; |
79787eaa | 732 | } |
0660b5af | 733 | |
6ef06d27 | 734 | btrfs_i_size_write(BTRFS_I(dir), dir->i_size + namelen * 2); |
52c26179 | 735 | ret = btrfs_update_inode(trans, root, dir); |
c7e54b51 JB |
736 | if (ret) { |
737 | btrfs_abort_transaction(trans, ret); | |
738 | goto fail; | |
739 | } | |
52c26179 | 740 | |
6025c19f | 741 | ret = btrfs_add_root_ref(trans, objectid, root->root_key.objectid, |
4a0cc7ca | 742 | btrfs_ino(BTRFS_I(dir)), index, name, namelen); |
c7e54b51 JB |
743 | if (ret) { |
744 | btrfs_abort_transaction(trans, ret); | |
745 | goto fail; | |
746 | } | |
f46b5a66 | 747 | |
cdb345a8 | 748 | ret = btrfs_uuid_tree_add(trans, root_item->uuid, |
6bccf3ab | 749 | BTRFS_UUID_KEY_SUBVOL, objectid); |
dd5f9615 | 750 | if (ret) |
66642832 | 751 | btrfs_abort_transaction(trans, ret); |
dd5f9615 | 752 | |
f46b5a66 | 753 | fail: |
49a3c4d9 | 754 | kfree(root_item); |
d5c12070 MX |
755 | trans->block_rsv = NULL; |
756 | trans->bytes_reserved = 0; | |
e85fde51 | 757 | btrfs_subvolume_release_metadata(root, &block_rsv); |
de6e8200 | 758 | |
9babda9f | 759 | err = btrfs_commit_transaction(trans); |
f46b5a66 CH |
760 | if (err && !ret) |
761 | ret = err; | |
1a65e24b | 762 | |
5662344b TI |
763 | if (!ret) { |
764 | inode = btrfs_lookup_dentry(dir, dentry); | |
de6e8200 LB |
765 | if (IS_ERR(inode)) |
766 | return PTR_ERR(inode); | |
5662344b TI |
767 | d_instantiate(dentry, inode); |
768 | } | |
f46b5a66 | 769 | return ret; |
49a3c4d9 DS |
770 | |
771 | fail_free: | |
2dfb1e43 QW |
772 | if (anon_dev) |
773 | free_anon_bdev(anon_dev); | |
49a3c4d9 DS |
774 | kfree(root_item); |
775 | return ret; | |
f46b5a66 CH |
776 | } |
777 | ||
e9662f70 | 778 | static int create_snapshot(struct btrfs_root *root, struct inode *dir, |
9babda9f | 779 | struct dentry *dentry, bool readonly, |
e9662f70 | 780 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 781 | { |
0b246afa | 782 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); |
2e4bfab9 | 783 | struct inode *inode; |
f46b5a66 CH |
784 | struct btrfs_pending_snapshot *pending_snapshot; |
785 | struct btrfs_trans_handle *trans; | |
2e4bfab9 | 786 | int ret; |
f46b5a66 | 787 | |
92a7cc42 | 788 | if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) |
f46b5a66 CH |
789 | return -EINVAL; |
790 | ||
eede2bf3 OS |
791 | if (atomic_read(&root->nr_swapfiles)) { |
792 | btrfs_warn(fs_info, | |
793 | "cannot snapshot subvolume with active swapfile"); | |
794 | return -ETXTBSY; | |
795 | } | |
796 | ||
23269bf5 | 797 | pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_KERNEL); |
a1ee7362 DS |
798 | if (!pending_snapshot) |
799 | return -ENOMEM; | |
800 | ||
2dfb1e43 QW |
801 | ret = get_anon_bdev(&pending_snapshot->anon_dev); |
802 | if (ret < 0) | |
803 | goto free_pending; | |
b0c0ea63 | 804 | pending_snapshot->root_item = kzalloc(sizeof(struct btrfs_root_item), |
23269bf5 | 805 | GFP_KERNEL); |
8546b570 DS |
806 | pending_snapshot->path = btrfs_alloc_path(); |
807 | if (!pending_snapshot->root_item || !pending_snapshot->path) { | |
b0c0ea63 DS |
808 | ret = -ENOMEM; |
809 | goto free_pending; | |
810 | } | |
811 | ||
66d8f3dd MX |
812 | btrfs_init_block_rsv(&pending_snapshot->block_rsv, |
813 | BTRFS_BLOCK_RSV_TEMP); | |
d5c12070 MX |
814 | /* |
815 | * 1 - parent dir inode | |
816 | * 2 - dir entries | |
817 | * 1 - root item | |
818 | * 2 - root ref/backref | |
819 | * 1 - root of snapshot | |
dd5f9615 | 820 | * 1 - UUID item |
d5c12070 MX |
821 | */ |
822 | ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root, | |
dd5f9615 | 823 | &pending_snapshot->block_rsv, 8, |
ee3441b4 | 824 | false); |
d5c12070 | 825 | if (ret) |
c11fbb6e | 826 | goto free_pending; |
d5c12070 | 827 | |
3de4586c | 828 | pending_snapshot->dentry = dentry; |
f46b5a66 | 829 | pending_snapshot->root = root; |
b83cc969 | 830 | pending_snapshot->readonly = readonly; |
e9662f70 | 831 | pending_snapshot->dir = dir; |
8696c533 | 832 | pending_snapshot->inherit = inherit; |
a22285a6 | 833 | |
d5c12070 | 834 | trans = btrfs_start_transaction(root, 0); |
a22285a6 YZ |
835 | if (IS_ERR(trans)) { |
836 | ret = PTR_ERR(trans); | |
837 | goto fail; | |
838 | } | |
839 | ||
0b246afa | 840 | spin_lock(&fs_info->trans_lock); |
f46b5a66 CH |
841 | list_add(&pending_snapshot->list, |
842 | &trans->transaction->pending_snapshots); | |
0b246afa | 843 | spin_unlock(&fs_info->trans_lock); |
9babda9f NB |
844 | |
845 | ret = btrfs_commit_transaction(trans); | |
aec8030a | 846 | if (ret) |
c37b2b62 | 847 | goto fail; |
a22285a6 YZ |
848 | |
849 | ret = pending_snapshot->error; | |
850 | if (ret) | |
851 | goto fail; | |
852 | ||
d3797308 CM |
853 | ret = btrfs_orphan_cleanup(pending_snapshot->snap); |
854 | if (ret) | |
855 | goto fail; | |
856 | ||
2b0143b5 | 857 | inode = btrfs_lookup_dentry(d_inode(dentry->d_parent), dentry); |
2e4bfab9 YZ |
858 | if (IS_ERR(inode)) { |
859 | ret = PTR_ERR(inode); | |
860 | goto fail; | |
861 | } | |
5662344b | 862 | |
2e4bfab9 YZ |
863 | d_instantiate(dentry, inode); |
864 | ret = 0; | |
2dfb1e43 | 865 | pending_snapshot->anon_dev = 0; |
2e4bfab9 | 866 | fail: |
2dfb1e43 QW |
867 | /* Prevent double freeing of anon_dev */ |
868 | if (ret && pending_snapshot->snap) | |
869 | pending_snapshot->snap->anon_dev = 0; | |
00246528 | 870 | btrfs_put_root(pending_snapshot->snap); |
e85fde51 | 871 | btrfs_subvolume_release_metadata(root, &pending_snapshot->block_rsv); |
b0c0ea63 | 872 | free_pending: |
2dfb1e43 QW |
873 | if (pending_snapshot->anon_dev) |
874 | free_anon_bdev(pending_snapshot->anon_dev); | |
b0c0ea63 | 875 | kfree(pending_snapshot->root_item); |
8546b570 | 876 | btrfs_free_path(pending_snapshot->path); |
a1ee7362 DS |
877 | kfree(pending_snapshot); |
878 | ||
f46b5a66 CH |
879 | return ret; |
880 | } | |
881 | ||
4260f7c7 SW |
882 | /* copy of may_delete in fs/namei.c() |
883 | * Check whether we can remove a link victim from directory dir, check | |
884 | * whether the type of victim is right. | |
885 | * 1. We can't do it if dir is read-only (done in permission()) | |
886 | * 2. We should have write and exec permissions on dir | |
887 | * 3. We can't remove anything from append-only dir | |
888 | * 4. We can't do anything with immutable dir (done in permission()) | |
889 | * 5. If the sticky bit on dir is set we should either | |
890 | * a. be owner of dir, or | |
891 | * b. be owner of victim, or | |
892 | * c. have CAP_FOWNER capability | |
01327610 | 893 | * 6. If the victim is append-only or immutable we can't do anything with |
4260f7c7 SW |
894 | * links pointing to it. |
895 | * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR. | |
896 | * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR. | |
897 | * 9. We can't remove a root or mountpoint. | |
898 | * 10. We don't allow removal of NFS sillyrenamed files; it's handled by | |
899 | * nfs_async_unlink(). | |
900 | */ | |
901 | ||
67871254 | 902 | static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir) |
4260f7c7 SW |
903 | { |
904 | int error; | |
905 | ||
2b0143b5 | 906 | if (d_really_is_negative(victim)) |
4260f7c7 SW |
907 | return -ENOENT; |
908 | ||
2b0143b5 | 909 | BUG_ON(d_inode(victim->d_parent) != dir); |
4fa6b5ec | 910 | audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE); |
4260f7c7 SW |
911 | |
912 | error = inode_permission(dir, MAY_WRITE | MAY_EXEC); | |
913 | if (error) | |
914 | return error; | |
915 | if (IS_APPEND(dir)) | |
916 | return -EPERM; | |
2b0143b5 DH |
917 | if (check_sticky(dir, d_inode(victim)) || IS_APPEND(d_inode(victim)) || |
918 | IS_IMMUTABLE(d_inode(victim)) || IS_SWAPFILE(d_inode(victim))) | |
4260f7c7 SW |
919 | return -EPERM; |
920 | if (isdir) { | |
e36cb0b8 | 921 | if (!d_is_dir(victim)) |
4260f7c7 SW |
922 | return -ENOTDIR; |
923 | if (IS_ROOT(victim)) | |
924 | return -EBUSY; | |
e36cb0b8 | 925 | } else if (d_is_dir(victim)) |
4260f7c7 SW |
926 | return -EISDIR; |
927 | if (IS_DEADDIR(dir)) | |
928 | return -ENOENT; | |
929 | if (victim->d_flags & DCACHE_NFSFS_RENAMED) | |
930 | return -EBUSY; | |
931 | return 0; | |
932 | } | |
933 | ||
cb8e7090 CH |
934 | /* copy of may_create in fs/namei.c() */ |
935 | static inline int btrfs_may_create(struct inode *dir, struct dentry *child) | |
936 | { | |
2b0143b5 | 937 | if (d_really_is_positive(child)) |
cb8e7090 CH |
938 | return -EEXIST; |
939 | if (IS_DEADDIR(dir)) | |
940 | return -ENOENT; | |
941 | return inode_permission(dir, MAY_WRITE | MAY_EXEC); | |
942 | } | |
943 | ||
944 | /* | |
945 | * Create a new subvolume below @parent. This is largely modeled after | |
946 | * sys_mkdirat and vfs_mkdir, but we only do a single component lookup | |
947 | * inside this filesystem so it's quite a bit simpler. | |
948 | */ | |
92872094 | 949 | static noinline int btrfs_mksubvol(const struct path *parent, |
52f75f4f | 950 | const char *name, int namelen, |
72fd032e | 951 | struct btrfs_root *snap_src, |
9babda9f | 952 | bool readonly, |
8696c533 | 953 | struct btrfs_qgroup_inherit *inherit) |
cb8e7090 | 954 | { |
0b246afa JM |
955 | struct inode *dir = d_inode(parent->dentry); |
956 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); | |
cb8e7090 CH |
957 | struct dentry *dentry; |
958 | int error; | |
959 | ||
00235411 AV |
960 | error = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT); |
961 | if (error == -EINTR) | |
962 | return error; | |
cb8e7090 CH |
963 | |
964 | dentry = lookup_one_len(name, parent->dentry, namelen); | |
965 | error = PTR_ERR(dentry); | |
966 | if (IS_ERR(dentry)) | |
967 | goto out_unlock; | |
968 | ||
76dda93c | 969 | error = btrfs_may_create(dir, dentry); |
cb8e7090 | 970 | if (error) |
a874a63e | 971 | goto out_dput; |
cb8e7090 | 972 | |
9c52057c CM |
973 | /* |
974 | * even if this name doesn't exist, we may get hash collisions. | |
975 | * check for them now when we can safely fail | |
976 | */ | |
977 | error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root, | |
978 | dir->i_ino, name, | |
979 | namelen); | |
980 | if (error) | |
981 | goto out_dput; | |
982 | ||
0b246afa | 983 | down_read(&fs_info->subvol_sem); |
76dda93c YZ |
984 | |
985 | if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0) | |
986 | goto out_up_read; | |
987 | ||
9babda9f NB |
988 | if (snap_src) |
989 | error = create_snapshot(snap_src, dir, dentry, readonly, inherit); | |
990 | else | |
991 | error = create_subvol(dir, dentry, name, namelen, inherit); | |
992 | ||
76dda93c YZ |
993 | if (!error) |
994 | fsnotify_mkdir(dir, dentry); | |
995 | out_up_read: | |
0b246afa | 996 | up_read(&fs_info->subvol_sem); |
cb8e7090 CH |
997 | out_dput: |
998 | dput(dentry); | |
999 | out_unlock: | |
5955102c | 1000 | inode_unlock(dir); |
cb8e7090 CH |
1001 | return error; |
1002 | } | |
1003 | ||
c11fbb6e RK |
1004 | static noinline int btrfs_mksnapshot(const struct path *parent, |
1005 | const char *name, int namelen, | |
1006 | struct btrfs_root *root, | |
1007 | bool readonly, | |
1008 | struct btrfs_qgroup_inherit *inherit) | |
1009 | { | |
1010 | int ret; | |
1011 | bool snapshot_force_cow = false; | |
1012 | ||
1013 | /* | |
1014 | * Force new buffered writes to reserve space even when NOCOW is | |
1015 | * possible. This is to avoid later writeback (running dealloc) to | |
1016 | * fallback to COW mode and unexpectedly fail with ENOSPC. | |
1017 | */ | |
1018 | btrfs_drew_read_lock(&root->snapshot_lock); | |
1019 | ||
1020 | ret = btrfs_start_delalloc_snapshot(root); | |
1021 | if (ret) | |
1022 | goto out; | |
1023 | ||
1024 | /* | |
1025 | * All previous writes have started writeback in NOCOW mode, so now | |
1026 | * we force future writes to fallback to COW mode during snapshot | |
1027 | * creation. | |
1028 | */ | |
1029 | atomic_inc(&root->snapshot_force_cow); | |
1030 | snapshot_force_cow = true; | |
1031 | ||
1032 | btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1); | |
1033 | ||
1034 | ret = btrfs_mksubvol(parent, name, namelen, | |
1035 | root, readonly, inherit); | |
1036 | out: | |
1037 | if (snapshot_force_cow) | |
1038 | atomic_dec(&root->snapshot_force_cow); | |
1039 | btrfs_drew_read_unlock(&root->snapshot_lock); | |
1040 | return ret; | |
1041 | } | |
1042 | ||
4cb5300b CM |
1043 | /* |
1044 | * When we're defragging a range, we don't want to kick it off again | |
1045 | * if it is really just waiting for delalloc to send it down. | |
1046 | * If we find a nice big extent or delalloc range for the bytes in the | |
1047 | * file you want to defrag, we return 0 to let you know to skip this | |
1048 | * part of the file | |
1049 | */ | |
aab110ab | 1050 | static int check_defrag_in_cache(struct inode *inode, u64 offset, u32 thresh) |
4cb5300b CM |
1051 | { |
1052 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; | |
1053 | struct extent_map *em = NULL; | |
1054 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
1055 | u64 end; | |
1056 | ||
1057 | read_lock(&em_tree->lock); | |
09cbfeaf | 1058 | em = lookup_extent_mapping(em_tree, offset, PAGE_SIZE); |
4cb5300b CM |
1059 | read_unlock(&em_tree->lock); |
1060 | ||
1061 | if (em) { | |
1062 | end = extent_map_end(em); | |
1063 | free_extent_map(em); | |
1064 | if (end - offset > thresh) | |
1065 | return 0; | |
1066 | } | |
1067 | /* if we already have a nice delalloc here, just stop */ | |
1068 | thresh /= 2; | |
1069 | end = count_range_bits(io_tree, &offset, offset + thresh, | |
1070 | thresh, EXTENT_DELALLOC, 1); | |
1071 | if (end >= thresh) | |
1072 | return 0; | |
1073 | return 1; | |
1074 | } | |
1075 | ||
1076 | /* | |
1077 | * helper function to walk through a file and find extents | |
1078 | * newer than a specific transid, and smaller than thresh. | |
1079 | * | |
1080 | * This is used by the defragging code to find new and small | |
1081 | * extents | |
1082 | */ | |
1083 | static int find_new_extents(struct btrfs_root *root, | |
1084 | struct inode *inode, u64 newer_than, | |
aab110ab | 1085 | u64 *off, u32 thresh) |
4cb5300b CM |
1086 | { |
1087 | struct btrfs_path *path; | |
1088 | struct btrfs_key min_key; | |
4cb5300b CM |
1089 | struct extent_buffer *leaf; |
1090 | struct btrfs_file_extent_item *extent; | |
1091 | int type; | |
1092 | int ret; | |
4a0cc7ca | 1093 | u64 ino = btrfs_ino(BTRFS_I(inode)); |
4cb5300b CM |
1094 | |
1095 | path = btrfs_alloc_path(); | |
1096 | if (!path) | |
1097 | return -ENOMEM; | |
1098 | ||
a4689d2b | 1099 | min_key.objectid = ino; |
4cb5300b CM |
1100 | min_key.type = BTRFS_EXTENT_DATA_KEY; |
1101 | min_key.offset = *off; | |
1102 | ||
67871254 | 1103 | while (1) { |
6174d3cb | 1104 | ret = btrfs_search_forward(root, &min_key, path, newer_than); |
4cb5300b CM |
1105 | if (ret != 0) |
1106 | goto none; | |
f094c9bd | 1107 | process_slot: |
a4689d2b | 1108 | if (min_key.objectid != ino) |
4cb5300b CM |
1109 | goto none; |
1110 | if (min_key.type != BTRFS_EXTENT_DATA_KEY) | |
1111 | goto none; | |
1112 | ||
1113 | leaf = path->nodes[0]; | |
1114 | extent = btrfs_item_ptr(leaf, path->slots[0], | |
1115 | struct btrfs_file_extent_item); | |
1116 | ||
1117 | type = btrfs_file_extent_type(leaf, extent); | |
1118 | if (type == BTRFS_FILE_EXTENT_REG && | |
1119 | btrfs_file_extent_num_bytes(leaf, extent) < thresh && | |
1120 | check_defrag_in_cache(inode, min_key.offset, thresh)) { | |
1121 | *off = min_key.offset; | |
1122 | btrfs_free_path(path); | |
1123 | return 0; | |
1124 | } | |
1125 | ||
f094c9bd FM |
1126 | path->slots[0]++; |
1127 | if (path->slots[0] < btrfs_header_nritems(leaf)) { | |
1128 | btrfs_item_key_to_cpu(leaf, &min_key, path->slots[0]); | |
1129 | goto process_slot; | |
1130 | } | |
1131 | ||
4cb5300b CM |
1132 | if (min_key.offset == (u64)-1) |
1133 | goto none; | |
1134 | ||
1135 | min_key.offset++; | |
1136 | btrfs_release_path(path); | |
1137 | } | |
1138 | none: | |
1139 | btrfs_free_path(path); | |
1140 | return -ENOENT; | |
1141 | } | |
1142 | ||
6c282eb4 | 1143 | static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start) |
17ce6ef8 LB |
1144 | { |
1145 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
6c282eb4 LZ |
1146 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; |
1147 | struct extent_map *em; | |
09cbfeaf | 1148 | u64 len = PAGE_SIZE; |
17ce6ef8 | 1149 | |
6c282eb4 LZ |
1150 | /* |
1151 | * hopefully we have this extent in the tree already, try without | |
1152 | * the full extent lock | |
1153 | */ | |
17ce6ef8 | 1154 | read_lock(&em_tree->lock); |
6c282eb4 | 1155 | em = lookup_extent_mapping(em_tree, start, len); |
17ce6ef8 LB |
1156 | read_unlock(&em_tree->lock); |
1157 | ||
6c282eb4 | 1158 | if (!em) { |
308d9800 FM |
1159 | struct extent_state *cached = NULL; |
1160 | u64 end = start + len - 1; | |
1161 | ||
6c282eb4 | 1162 | /* get the big lock and read metadata off disk */ |
ff13db41 | 1163 | lock_extent_bits(io_tree, start, end, &cached); |
39b07b5d | 1164 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start, len); |
e43bbe5e | 1165 | unlock_extent_cached(io_tree, start, end, &cached); |
6c282eb4 LZ |
1166 | |
1167 | if (IS_ERR(em)) | |
1168 | return NULL; | |
1169 | } | |
1170 | ||
1171 | return em; | |
1172 | } | |
17ce6ef8 | 1173 | |
6c282eb4 LZ |
1174 | static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em) |
1175 | { | |
1176 | struct extent_map *next; | |
1177 | bool ret = true; | |
1178 | ||
1179 | /* this is the last extent */ | |
1180 | if (em->start + em->len >= i_size_read(inode)) | |
1181 | return false; | |
1182 | ||
1183 | next = defrag_lookup_extent(inode, em->start + em->len); | |
e9512d72 CM |
1184 | if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE) |
1185 | ret = false; | |
1186 | else if ((em->block_start + em->block_len == next->block_start) && | |
ee22184b | 1187 | (em->block_len > SZ_128K && next->block_len > SZ_128K)) |
6c282eb4 LZ |
1188 | ret = false; |
1189 | ||
1190 | free_extent_map(next); | |
17ce6ef8 LB |
1191 | return ret; |
1192 | } | |
1193 | ||
aab110ab | 1194 | static int should_defrag_range(struct inode *inode, u64 start, u32 thresh, |
a43a2111 AM |
1195 | u64 *last_len, u64 *skip, u64 *defrag_end, |
1196 | int compress) | |
940100a4 | 1197 | { |
6c282eb4 | 1198 | struct extent_map *em; |
940100a4 | 1199 | int ret = 1; |
6c282eb4 | 1200 | bool next_mergeable = true; |
4a3560c4 | 1201 | bool prev_mergeable = true; |
940100a4 CM |
1202 | |
1203 | /* | |
008873ea | 1204 | * make sure that once we start defragging an extent, we keep on |
940100a4 CM |
1205 | * defragging it |
1206 | */ | |
1207 | if (start < *defrag_end) | |
1208 | return 1; | |
1209 | ||
1210 | *skip = 0; | |
1211 | ||
6c282eb4 LZ |
1212 | em = defrag_lookup_extent(inode, start); |
1213 | if (!em) | |
1214 | return 0; | |
940100a4 CM |
1215 | |
1216 | /* this will cover holes, and inline extents */ | |
17ce6ef8 | 1217 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { |
940100a4 | 1218 | ret = 0; |
17ce6ef8 LB |
1219 | goto out; |
1220 | } | |
1221 | ||
4a3560c4 LB |
1222 | if (!*defrag_end) |
1223 | prev_mergeable = false; | |
1224 | ||
6c282eb4 | 1225 | next_mergeable = defrag_check_next_extent(inode, em); |
940100a4 | 1226 | /* |
6c282eb4 LZ |
1227 | * we hit a real extent, if it is big or the next extent is not a |
1228 | * real extent, don't bother defragging it | |
940100a4 | 1229 | */ |
a43a2111 | 1230 | if (!compress && (*last_len == 0 || *last_len >= thresh) && |
4a3560c4 | 1231 | (em->len >= thresh || (!next_mergeable && !prev_mergeable))) |
940100a4 | 1232 | ret = 0; |
17ce6ef8 | 1233 | out: |
940100a4 CM |
1234 | /* |
1235 | * last_len ends up being a counter of how many bytes we've defragged. | |
1236 | * every time we choose not to defrag an extent, we reset *last_len | |
1237 | * so that the next tiny extent will force a defrag. | |
1238 | * | |
1239 | * The end result of this is that tiny extents before a single big | |
1240 | * extent will force at least part of that big extent to be defragged. | |
1241 | */ | |
1242 | if (ret) { | |
940100a4 CM |
1243 | *defrag_end = extent_map_end(em); |
1244 | } else { | |
1245 | *last_len = 0; | |
1246 | *skip = extent_map_end(em); | |
1247 | *defrag_end = 0; | |
1248 | } | |
1249 | ||
1250 | free_extent_map(em); | |
1251 | return ret; | |
1252 | } | |
1253 | ||
4cb5300b CM |
1254 | /* |
1255 | * it doesn't do much good to defrag one or two pages | |
1256 | * at a time. This pulls in a nice chunk of pages | |
1257 | * to COW and defrag. | |
1258 | * | |
1259 | * It also makes sure the delalloc code has enough | |
1260 | * dirty data to avoid making new small extents as part | |
1261 | * of the defrag | |
1262 | * | |
1263 | * It's a good idea to start RA on this range | |
1264 | * before calling this. | |
1265 | */ | |
1266 | static int cluster_pages_for_defrag(struct inode *inode, | |
1267 | struct page **pages, | |
1268 | unsigned long start_index, | |
c41570c9 | 1269 | unsigned long num_pages) |
f46b5a66 | 1270 | { |
4cb5300b CM |
1271 | unsigned long file_end; |
1272 | u64 isize = i_size_read(inode); | |
1273 | u64 page_start; | |
1274 | u64 page_end; | |
1f12bd06 | 1275 | u64 page_cnt; |
4cb5300b CM |
1276 | int ret; |
1277 | int i; | |
1278 | int i_done; | |
3eaa2885 | 1279 | struct btrfs_ordered_extent *ordered; |
4cb5300b | 1280 | struct extent_state *cached_state = NULL; |
600a45e1 | 1281 | struct extent_io_tree *tree; |
364ecf36 | 1282 | struct extent_changeset *data_reserved = NULL; |
3b16a4e3 | 1283 | gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); |
4cb5300b | 1284 | |
09cbfeaf | 1285 | file_end = (isize - 1) >> PAGE_SHIFT; |
1f12bd06 LB |
1286 | if (!isize || start_index > file_end) |
1287 | return 0; | |
1288 | ||
1289 | page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1); | |
4cb5300b | 1290 | |
e5b7231e | 1291 | ret = btrfs_delalloc_reserve_space(BTRFS_I(inode), &data_reserved, |
09cbfeaf KS |
1292 | start_index << PAGE_SHIFT, |
1293 | page_cnt << PAGE_SHIFT); | |
4cb5300b CM |
1294 | if (ret) |
1295 | return ret; | |
4cb5300b | 1296 | i_done = 0; |
600a45e1 | 1297 | tree = &BTRFS_I(inode)->io_tree; |
4cb5300b CM |
1298 | |
1299 | /* step one, lock all the pages */ | |
1f12bd06 | 1300 | for (i = 0; i < page_cnt; i++) { |
4cb5300b | 1301 | struct page *page; |
600a45e1 | 1302 | again: |
a94733d0 | 1303 | page = find_or_create_page(inode->i_mapping, |
600a45e1 | 1304 | start_index + i, mask); |
4cb5300b CM |
1305 | if (!page) |
1306 | break; | |
1307 | ||
600a45e1 | 1308 | page_start = page_offset(page); |
09cbfeaf | 1309 | page_end = page_start + PAGE_SIZE - 1; |
600a45e1 | 1310 | while (1) { |
308d9800 | 1311 | lock_extent_bits(tree, page_start, page_end, |
ff13db41 | 1312 | &cached_state); |
c3504372 | 1313 | ordered = btrfs_lookup_ordered_extent(BTRFS_I(inode), |
600a45e1 | 1314 | page_start); |
308d9800 | 1315 | unlock_extent_cached(tree, page_start, page_end, |
e43bbe5e | 1316 | &cached_state); |
600a45e1 MX |
1317 | if (!ordered) |
1318 | break; | |
1319 | ||
1320 | unlock_page(page); | |
1321 | btrfs_start_ordered_extent(inode, ordered, 1); | |
1322 | btrfs_put_ordered_extent(ordered); | |
1323 | lock_page(page); | |
1f12bd06 LB |
1324 | /* |
1325 | * we unlocked the page above, so we need check if | |
1326 | * it was released or not. | |
1327 | */ | |
1328 | if (page->mapping != inode->i_mapping) { | |
1329 | unlock_page(page); | |
09cbfeaf | 1330 | put_page(page); |
1f12bd06 LB |
1331 | goto again; |
1332 | } | |
600a45e1 MX |
1333 | } |
1334 | ||
4cb5300b CM |
1335 | if (!PageUptodate(page)) { |
1336 | btrfs_readpage(NULL, page); | |
1337 | lock_page(page); | |
1338 | if (!PageUptodate(page)) { | |
1339 | unlock_page(page); | |
09cbfeaf | 1340 | put_page(page); |
4cb5300b CM |
1341 | ret = -EIO; |
1342 | break; | |
1343 | } | |
1344 | } | |
600a45e1 | 1345 | |
600a45e1 MX |
1346 | if (page->mapping != inode->i_mapping) { |
1347 | unlock_page(page); | |
09cbfeaf | 1348 | put_page(page); |
600a45e1 MX |
1349 | goto again; |
1350 | } | |
1351 | ||
4cb5300b CM |
1352 | pages[i] = page; |
1353 | i_done++; | |
1354 | } | |
1355 | if (!i_done || ret) | |
1356 | goto out; | |
1357 | ||
1751e8a6 | 1358 | if (!(inode->i_sb->s_flags & SB_ACTIVE)) |
4cb5300b CM |
1359 | goto out; |
1360 | ||
1361 | /* | |
1362 | * so now we have a nice long stream of locked | |
1363 | * and up to date pages, lets wait on them | |
1364 | */ | |
1365 | for (i = 0; i < i_done; i++) | |
1366 | wait_on_page_writeback(pages[i]); | |
1367 | ||
1368 | page_start = page_offset(pages[0]); | |
09cbfeaf | 1369 | page_end = page_offset(pages[i_done - 1]) + PAGE_SIZE; |
4cb5300b CM |
1370 | |
1371 | lock_extent_bits(&BTRFS_I(inode)->io_tree, | |
ff13db41 | 1372 | page_start, page_end - 1, &cached_state); |
4cb5300b | 1373 | clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, |
e182163d OS |
1374 | page_end - 1, EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | |
1375 | EXTENT_DEFRAG, 0, 0, &cached_state); | |
4cb5300b | 1376 | |
1f12bd06 | 1377 | if (i_done != page_cnt) { |
9e0baf60 | 1378 | spin_lock(&BTRFS_I(inode)->lock); |
28c4a3e2 | 1379 | btrfs_mod_outstanding_extents(BTRFS_I(inode), 1); |
9e0baf60 | 1380 | spin_unlock(&BTRFS_I(inode)->lock); |
86d52921 | 1381 | btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, |
09cbfeaf | 1382 | start_index << PAGE_SHIFT, |
43b18595 | 1383 | (page_cnt - i_done) << PAGE_SHIFT, true); |
4cb5300b CM |
1384 | } |
1385 | ||
1386 | ||
9e8a4a8b | 1387 | set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1, |
018ed4f7 | 1388 | &cached_state); |
4cb5300b CM |
1389 | |
1390 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
e43bbe5e | 1391 | page_start, page_end - 1, &cached_state); |
4cb5300b CM |
1392 | |
1393 | for (i = 0; i < i_done; i++) { | |
1394 | clear_page_dirty_for_io(pages[i]); | |
1395 | ClearPageChecked(pages[i]); | |
1396 | set_page_extent_mapped(pages[i]); | |
1397 | set_page_dirty(pages[i]); | |
1398 | unlock_page(pages[i]); | |
09cbfeaf | 1399 | put_page(pages[i]); |
4cb5300b | 1400 | } |
8702ba93 | 1401 | btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT); |
364ecf36 | 1402 | extent_changeset_free(data_reserved); |
4cb5300b CM |
1403 | return i_done; |
1404 | out: | |
1405 | for (i = 0; i < i_done; i++) { | |
1406 | unlock_page(pages[i]); | |
09cbfeaf | 1407 | put_page(pages[i]); |
4cb5300b | 1408 | } |
86d52921 | 1409 | btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, |
09cbfeaf | 1410 | start_index << PAGE_SHIFT, |
43b18595 | 1411 | page_cnt << PAGE_SHIFT, true); |
8702ba93 | 1412 | btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT); |
364ecf36 | 1413 | extent_changeset_free(data_reserved); |
4cb5300b CM |
1414 | return ret; |
1415 | ||
1416 | } | |
1417 | ||
1418 | int btrfs_defrag_file(struct inode *inode, struct file *file, | |
1419 | struct btrfs_ioctl_defrag_range_args *range, | |
1420 | u64 newer_than, unsigned long max_to_defrag) | |
1421 | { | |
0b246afa | 1422 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
4cb5300b | 1423 | struct btrfs_root *root = BTRFS_I(inode)->root; |
4cb5300b | 1424 | struct file_ra_state *ra = NULL; |
f46b5a66 | 1425 | unsigned long last_index; |
151a31b2 | 1426 | u64 isize = i_size_read(inode); |
940100a4 CM |
1427 | u64 last_len = 0; |
1428 | u64 skip = 0; | |
1429 | u64 defrag_end = 0; | |
4cb5300b | 1430 | u64 newer_off = range->start; |
f46b5a66 | 1431 | unsigned long i; |
008873ea | 1432 | unsigned long ra_index = 0; |
f46b5a66 | 1433 | int ret; |
4cb5300b | 1434 | int defrag_count = 0; |
1a419d85 | 1435 | int compress_type = BTRFS_COMPRESS_ZLIB; |
aab110ab | 1436 | u32 extent_thresh = range->extent_thresh; |
09cbfeaf | 1437 | unsigned long max_cluster = SZ_256K >> PAGE_SHIFT; |
c41570c9 | 1438 | unsigned long cluster = max_cluster; |
ee22184b | 1439 | u64 new_align = ~((u64)SZ_128K - 1); |
4cb5300b | 1440 | struct page **pages = NULL; |
1e2ef46d | 1441 | bool do_compress = range->flags & BTRFS_DEFRAG_RANGE_COMPRESS; |
4cb5300b | 1442 | |
0abd5b17 LB |
1443 | if (isize == 0) |
1444 | return 0; | |
1445 | ||
1446 | if (range->start >= isize) | |
1447 | return -EINVAL; | |
1a419d85 | 1448 | |
1e2ef46d | 1449 | if (do_compress) { |
ce96b7ff | 1450 | if (range->compress_type >= BTRFS_NR_COMPRESS_TYPES) |
1a419d85 LZ |
1451 | return -EINVAL; |
1452 | if (range->compress_type) | |
1453 | compress_type = range->compress_type; | |
1454 | } | |
f46b5a66 | 1455 | |
0abd5b17 | 1456 | if (extent_thresh == 0) |
ee22184b | 1457 | extent_thresh = SZ_256K; |
940100a4 | 1458 | |
4cb5300b | 1459 | /* |
0a52d108 DS |
1460 | * If we were not given a file, allocate a readahead context. As |
1461 | * readahead is just an optimization, defrag will work without it so | |
1462 | * we don't error out. | |
4cb5300b CM |
1463 | */ |
1464 | if (!file) { | |
63e727ec | 1465 | ra = kzalloc(sizeof(*ra), GFP_KERNEL); |
0a52d108 DS |
1466 | if (ra) |
1467 | file_ra_state_init(ra, inode->i_mapping); | |
4cb5300b CM |
1468 | } else { |
1469 | ra = &file->f_ra; | |
1470 | } | |
1471 | ||
63e727ec | 1472 | pages = kmalloc_array(max_cluster, sizeof(struct page *), GFP_KERNEL); |
4cb5300b CM |
1473 | if (!pages) { |
1474 | ret = -ENOMEM; | |
1475 | goto out_ra; | |
1476 | } | |
1477 | ||
1478 | /* find the last page to defrag */ | |
1e701a32 | 1479 | if (range->start + range->len > range->start) { |
151a31b2 | 1480 | last_index = min_t(u64, isize - 1, |
09cbfeaf | 1481 | range->start + range->len - 1) >> PAGE_SHIFT; |
1e701a32 | 1482 | } else { |
09cbfeaf | 1483 | last_index = (isize - 1) >> PAGE_SHIFT; |
1e701a32 CM |
1484 | } |
1485 | ||
4cb5300b CM |
1486 | if (newer_than) { |
1487 | ret = find_new_extents(root, inode, newer_than, | |
ee22184b | 1488 | &newer_off, SZ_64K); |
4cb5300b CM |
1489 | if (!ret) { |
1490 | range->start = newer_off; | |
1491 | /* | |
1492 | * we always align our defrag to help keep | |
1493 | * the extents in the file evenly spaced | |
1494 | */ | |
09cbfeaf | 1495 | i = (newer_off & new_align) >> PAGE_SHIFT; |
4cb5300b CM |
1496 | } else |
1497 | goto out_ra; | |
1498 | } else { | |
09cbfeaf | 1499 | i = range->start >> PAGE_SHIFT; |
4cb5300b CM |
1500 | } |
1501 | if (!max_to_defrag) | |
070034bd | 1502 | max_to_defrag = last_index - i + 1; |
4cb5300b | 1503 | |
2a0f7f57 LZ |
1504 | /* |
1505 | * make writeback starts from i, so the defrag range can be | |
1506 | * written sequentially. | |
1507 | */ | |
1508 | if (i < inode->i_mapping->writeback_index) | |
1509 | inode->i_mapping->writeback_index = i; | |
1510 | ||
f7f43cc8 | 1511 | while (i <= last_index && defrag_count < max_to_defrag && |
09cbfeaf | 1512 | (i < DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE))) { |
4cb5300b CM |
1513 | /* |
1514 | * make sure we stop running if someone unmounts | |
1515 | * the FS | |
1516 | */ | |
1751e8a6 | 1517 | if (!(inode->i_sb->s_flags & SB_ACTIVE)) |
4cb5300b CM |
1518 | break; |
1519 | ||
0b246afa JM |
1520 | if (btrfs_defrag_cancelled(fs_info)) { |
1521 | btrfs_debug(fs_info, "defrag_file cancelled"); | |
210549eb DS |
1522 | ret = -EAGAIN; |
1523 | break; | |
1524 | } | |
1525 | ||
09cbfeaf | 1526 | if (!should_defrag_range(inode, (u64)i << PAGE_SHIFT, |
6c282eb4 | 1527 | extent_thresh, &last_len, &skip, |
1e2ef46d | 1528 | &defrag_end, do_compress)){ |
940100a4 CM |
1529 | unsigned long next; |
1530 | /* | |
1531 | * the should_defrag function tells us how much to skip | |
1532 | * bump our counter by the suggested amount | |
1533 | */ | |
09cbfeaf | 1534 | next = DIV_ROUND_UP(skip, PAGE_SIZE); |
940100a4 CM |
1535 | i = max(i + 1, next); |
1536 | continue; | |
1537 | } | |
008873ea LZ |
1538 | |
1539 | if (!newer_than) { | |
09cbfeaf KS |
1540 | cluster = (PAGE_ALIGN(defrag_end) >> |
1541 | PAGE_SHIFT) - i; | |
008873ea LZ |
1542 | cluster = min(cluster, max_cluster); |
1543 | } else { | |
1544 | cluster = max_cluster; | |
1545 | } | |
1546 | ||
008873ea LZ |
1547 | if (i + cluster > ra_index) { |
1548 | ra_index = max(i, ra_index); | |
0a52d108 | 1549 | if (ra) |
d3c0bab5 DS |
1550 | page_cache_sync_readahead(inode->i_mapping, ra, |
1551 | file, ra_index, cluster); | |
e4826a5b | 1552 | ra_index += cluster; |
008873ea | 1553 | } |
940100a4 | 1554 | |
5955102c | 1555 | inode_lock(inode); |
eede2bf3 OS |
1556 | if (IS_SWAPFILE(inode)) { |
1557 | ret = -ETXTBSY; | |
1558 | } else { | |
1559 | if (do_compress) | |
1560 | BTRFS_I(inode)->defrag_compress = compress_type; | |
1561 | ret = cluster_pages_for_defrag(inode, pages, i, cluster); | |
1562 | } | |
ecb8bea8 | 1563 | if (ret < 0) { |
5955102c | 1564 | inode_unlock(inode); |
4cb5300b | 1565 | goto out_ra; |
ecb8bea8 | 1566 | } |
4cb5300b CM |
1567 | |
1568 | defrag_count += ret; | |
d0e1d66b | 1569 | balance_dirty_pages_ratelimited(inode->i_mapping); |
5955102c | 1570 | inode_unlock(inode); |
4cb5300b CM |
1571 | |
1572 | if (newer_than) { | |
1573 | if (newer_off == (u64)-1) | |
1574 | break; | |
1575 | ||
e1f041e1 LB |
1576 | if (ret > 0) |
1577 | i += ret; | |
1578 | ||
4cb5300b | 1579 | newer_off = max(newer_off + 1, |
09cbfeaf | 1580 | (u64)i << PAGE_SHIFT); |
4cb5300b | 1581 | |
ee22184b BL |
1582 | ret = find_new_extents(root, inode, newer_than, |
1583 | &newer_off, SZ_64K); | |
4cb5300b CM |
1584 | if (!ret) { |
1585 | range->start = newer_off; | |
09cbfeaf | 1586 | i = (newer_off & new_align) >> PAGE_SHIFT; |
4cb5300b CM |
1587 | } else { |
1588 | break; | |
f46b5a66 | 1589 | } |
4cb5300b | 1590 | } else { |
008873ea | 1591 | if (ret > 0) { |
cbcc8326 | 1592 | i += ret; |
09cbfeaf | 1593 | last_len += ret << PAGE_SHIFT; |
008873ea | 1594 | } else { |
cbcc8326 | 1595 | i++; |
008873ea LZ |
1596 | last_len = 0; |
1597 | } | |
f46b5a66 | 1598 | } |
f46b5a66 CH |
1599 | } |
1600 | ||
dec8ef90 | 1601 | if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) { |
1e701a32 | 1602 | filemap_flush(inode->i_mapping); |
dec8ef90 FM |
1603 | if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, |
1604 | &BTRFS_I(inode)->runtime_flags)) | |
1605 | filemap_flush(inode->i_mapping); | |
1606 | } | |
1e701a32 | 1607 | |
1a419d85 | 1608 | if (range->compress_type == BTRFS_COMPRESS_LZO) { |
0b246afa | 1609 | btrfs_set_fs_incompat(fs_info, COMPRESS_LZO); |
5c1aab1d NT |
1610 | } else if (range->compress_type == BTRFS_COMPRESS_ZSTD) { |
1611 | btrfs_set_fs_incompat(fs_info, COMPRESS_ZSTD); | |
1a419d85 LZ |
1612 | } |
1613 | ||
60ccf82f | 1614 | ret = defrag_count; |
940100a4 | 1615 | |
4cb5300b | 1616 | out_ra: |
1e2ef46d | 1617 | if (do_compress) { |
5955102c | 1618 | inode_lock(inode); |
eec63c65 | 1619 | BTRFS_I(inode)->defrag_compress = BTRFS_COMPRESS_NONE; |
5955102c | 1620 | inode_unlock(inode); |
633085c7 | 1621 | } |
4cb5300b CM |
1622 | if (!file) |
1623 | kfree(ra); | |
1624 | kfree(pages); | |
940100a4 | 1625 | return ret; |
f46b5a66 CH |
1626 | } |
1627 | ||
198605a8 | 1628 | static noinline int btrfs_ioctl_resize(struct file *file, |
76dda93c | 1629 | void __user *arg) |
f46b5a66 | 1630 | { |
0b246afa JM |
1631 | struct inode *inode = file_inode(file); |
1632 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
f46b5a66 CH |
1633 | u64 new_size; |
1634 | u64 old_size; | |
1635 | u64 devid = 1; | |
0b246afa | 1636 | struct btrfs_root *root = BTRFS_I(inode)->root; |
f46b5a66 CH |
1637 | struct btrfs_ioctl_vol_args *vol_args; |
1638 | struct btrfs_trans_handle *trans; | |
1639 | struct btrfs_device *device = NULL; | |
1640 | char *sizestr; | |
9a40f122 | 1641 | char *retptr; |
f46b5a66 CH |
1642 | char *devstr = NULL; |
1643 | int ret = 0; | |
f46b5a66 CH |
1644 | int mod = 0; |
1645 | ||
e441d54d CM |
1646 | if (!capable(CAP_SYS_ADMIN)) |
1647 | return -EPERM; | |
1648 | ||
198605a8 MX |
1649 | ret = mnt_want_write_file(file); |
1650 | if (ret) | |
1651 | return ret; | |
1652 | ||
c3e1f96c | 1653 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_RESIZE)) { |
97547676 | 1654 | mnt_drop_write_file(file); |
e57138b3 | 1655 | return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
c9e9f97b ID |
1656 | } |
1657 | ||
dae7b665 | 1658 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
1659 | if (IS_ERR(vol_args)) { |
1660 | ret = PTR_ERR(vol_args); | |
1661 | goto out; | |
1662 | } | |
5516e595 MF |
1663 | |
1664 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
f46b5a66 | 1665 | |
f46b5a66 CH |
1666 | sizestr = vol_args->name; |
1667 | devstr = strchr(sizestr, ':'); | |
1668 | if (devstr) { | |
f46b5a66 CH |
1669 | sizestr = devstr + 1; |
1670 | *devstr = '\0'; | |
1671 | devstr = vol_args->name; | |
58dfae63 Z |
1672 | ret = kstrtoull(devstr, 10, &devid); |
1673 | if (ret) | |
1674 | goto out_free; | |
dfd79829 MX |
1675 | if (!devid) { |
1676 | ret = -EINVAL; | |
1677 | goto out_free; | |
1678 | } | |
0b246afa | 1679 | btrfs_info(fs_info, "resizing devid %llu", devid); |
f46b5a66 | 1680 | } |
dba60f3f | 1681 | |
09ba3bc9 | 1682 | device = btrfs_find_device(fs_info->fs_devices, devid, NULL, NULL, true); |
f46b5a66 | 1683 | if (!device) { |
0b246afa JM |
1684 | btrfs_info(fs_info, "resizer unable to find device %llu", |
1685 | devid); | |
dfd79829 | 1686 | ret = -ENODEV; |
c9e9f97b | 1687 | goto out_free; |
f46b5a66 | 1688 | } |
dba60f3f | 1689 | |
ebbede42 | 1690 | if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { |
0b246afa | 1691 | btrfs_info(fs_info, |
efe120a0 | 1692 | "resizer unable to apply on readonly device %llu", |
c1c9ff7c | 1693 | devid); |
dfd79829 | 1694 | ret = -EPERM; |
4e42ae1b LB |
1695 | goto out_free; |
1696 | } | |
1697 | ||
f46b5a66 CH |
1698 | if (!strcmp(sizestr, "max")) |
1699 | new_size = device->bdev->bd_inode->i_size; | |
1700 | else { | |
1701 | if (sizestr[0] == '-') { | |
1702 | mod = -1; | |
1703 | sizestr++; | |
1704 | } else if (sizestr[0] == '+') { | |
1705 | mod = 1; | |
1706 | sizestr++; | |
1707 | } | |
9a40f122 GH |
1708 | new_size = memparse(sizestr, &retptr); |
1709 | if (*retptr != '\0' || new_size == 0) { | |
f46b5a66 | 1710 | ret = -EINVAL; |
c9e9f97b | 1711 | goto out_free; |
f46b5a66 CH |
1712 | } |
1713 | } | |
1714 | ||
401e29c1 | 1715 | if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) { |
dfd79829 | 1716 | ret = -EPERM; |
63a212ab SB |
1717 | goto out_free; |
1718 | } | |
1719 | ||
7cc8e58d | 1720 | old_size = btrfs_device_get_total_bytes(device); |
f46b5a66 CH |
1721 | |
1722 | if (mod < 0) { | |
1723 | if (new_size > old_size) { | |
1724 | ret = -EINVAL; | |
c9e9f97b | 1725 | goto out_free; |
f46b5a66 CH |
1726 | } |
1727 | new_size = old_size - new_size; | |
1728 | } else if (mod > 0) { | |
eb8052e0 | 1729 | if (new_size > ULLONG_MAX - old_size) { |
902c68a4 | 1730 | ret = -ERANGE; |
eb8052e0 WF |
1731 | goto out_free; |
1732 | } | |
f46b5a66 CH |
1733 | new_size = old_size + new_size; |
1734 | } | |
1735 | ||
ee22184b | 1736 | if (new_size < SZ_256M) { |
f46b5a66 | 1737 | ret = -EINVAL; |
c9e9f97b | 1738 | goto out_free; |
f46b5a66 CH |
1739 | } |
1740 | if (new_size > device->bdev->bd_inode->i_size) { | |
1741 | ret = -EFBIG; | |
c9e9f97b | 1742 | goto out_free; |
f46b5a66 CH |
1743 | } |
1744 | ||
47f08b96 | 1745 | new_size = round_down(new_size, fs_info->sectorsize); |
f46b5a66 | 1746 | |
f46b5a66 | 1747 | if (new_size > old_size) { |
a22285a6 | 1748 | trans = btrfs_start_transaction(root, 0); |
98d5dc13 TI |
1749 | if (IS_ERR(trans)) { |
1750 | ret = PTR_ERR(trans); | |
c9e9f97b | 1751 | goto out_free; |
98d5dc13 | 1752 | } |
f46b5a66 | 1753 | ret = btrfs_grow_device(trans, device, new_size); |
3a45bb20 | 1754 | btrfs_commit_transaction(trans); |
ece7d20e | 1755 | } else if (new_size < old_size) { |
f46b5a66 | 1756 | ret = btrfs_shrink_device(device, new_size); |
0253f40e | 1757 | } /* equal, nothing need to do */ |
f46b5a66 | 1758 | |
faf8f7b9 MPS |
1759 | if (ret == 0 && new_size != old_size) |
1760 | btrfs_info_in_rcu(fs_info, | |
1761 | "resize device %s (devid %llu) from %llu to %llu", | |
1762 | rcu_str_deref(device->name), device->devid, | |
1763 | old_size, new_size); | |
c9e9f97b | 1764 | out_free: |
f46b5a66 | 1765 | kfree(vol_args); |
c9e9f97b | 1766 | out: |
c3e1f96c | 1767 | btrfs_exclop_finish(fs_info); |
18f39c41 | 1768 | mnt_drop_write_file(file); |
f46b5a66 CH |
1769 | return ret; |
1770 | } | |
1771 | ||
5d54c67e | 1772 | static noinline int __btrfs_ioctl_snap_create(struct file *file, |
52f75f4f | 1773 | const char *name, unsigned long fd, int subvol, |
5d54c67e | 1774 | bool readonly, |
8696c533 | 1775 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 1776 | { |
f46b5a66 | 1777 | int namelen; |
3de4586c | 1778 | int ret = 0; |
f46b5a66 | 1779 | |
325c50e3 JM |
1780 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1781 | return -ENOTDIR; | |
1782 | ||
a874a63e LB |
1783 | ret = mnt_want_write_file(file); |
1784 | if (ret) | |
1785 | goto out; | |
1786 | ||
72fd032e SW |
1787 | namelen = strlen(name); |
1788 | if (strchr(name, '/')) { | |
f46b5a66 | 1789 | ret = -EINVAL; |
a874a63e | 1790 | goto out_drop_write; |
f46b5a66 CH |
1791 | } |
1792 | ||
16780cab CM |
1793 | if (name[0] == '.' && |
1794 | (namelen == 1 || (name[1] == '.' && namelen == 2))) { | |
1795 | ret = -EEXIST; | |
a874a63e | 1796 | goto out_drop_write; |
16780cab CM |
1797 | } |
1798 | ||
3de4586c | 1799 | if (subvol) { |
72fd032e | 1800 | ret = btrfs_mksubvol(&file->f_path, name, namelen, |
5d54c67e | 1801 | NULL, readonly, inherit); |
cb8e7090 | 1802 | } else { |
2903ff01 | 1803 | struct fd src = fdget(fd); |
3de4586c | 1804 | struct inode *src_inode; |
2903ff01 | 1805 | if (!src.file) { |
3de4586c | 1806 | ret = -EINVAL; |
a874a63e | 1807 | goto out_drop_write; |
3de4586c CM |
1808 | } |
1809 | ||
496ad9aa AV |
1810 | src_inode = file_inode(src.file); |
1811 | if (src_inode->i_sb != file_inode(file)->i_sb) { | |
c79b4713 | 1812 | btrfs_info(BTRFS_I(file_inode(file))->root->fs_info, |
efe120a0 | 1813 | "Snapshot src from another FS"); |
23ad5b17 | 1814 | ret = -EXDEV; |
d0242061 DS |
1815 | } else if (!inode_owner_or_capable(src_inode)) { |
1816 | /* | |
1817 | * Subvolume creation is not restricted, but snapshots | |
1818 | * are limited to own subvolumes only | |
1819 | */ | |
1820 | ret = -EPERM; | |
ecd18815 | 1821 | } else { |
c11fbb6e | 1822 | ret = btrfs_mksnapshot(&file->f_path, name, namelen, |
ecd18815 | 1823 | BTRFS_I(src_inode)->root, |
5d54c67e | 1824 | readonly, inherit); |
3de4586c | 1825 | } |
2903ff01 | 1826 | fdput(src); |
cb8e7090 | 1827 | } |
a874a63e LB |
1828 | out_drop_write: |
1829 | mnt_drop_write_file(file); | |
f46b5a66 | 1830 | out: |
72fd032e SW |
1831 | return ret; |
1832 | } | |
1833 | ||
1834 | static noinline int btrfs_ioctl_snap_create(struct file *file, | |
fa0d2b9b | 1835 | void __user *arg, int subvol) |
72fd032e | 1836 | { |
fa0d2b9b | 1837 | struct btrfs_ioctl_vol_args *vol_args; |
72fd032e SW |
1838 | int ret; |
1839 | ||
325c50e3 JM |
1840 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1841 | return -ENOTDIR; | |
1842 | ||
fa0d2b9b LZ |
1843 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1844 | if (IS_ERR(vol_args)) | |
1845 | return PTR_ERR(vol_args); | |
1846 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
72fd032e | 1847 | |
5d54c67e NB |
1848 | ret = __btrfs_ioctl_snap_create(file, vol_args->name, vol_args->fd, |
1849 | subvol, false, NULL); | |
fdfb1e4f | 1850 | |
fa0d2b9b LZ |
1851 | kfree(vol_args); |
1852 | return ret; | |
1853 | } | |
fdfb1e4f | 1854 | |
fa0d2b9b LZ |
1855 | static noinline int btrfs_ioctl_snap_create_v2(struct file *file, |
1856 | void __user *arg, int subvol) | |
1857 | { | |
1858 | struct btrfs_ioctl_vol_args_v2 *vol_args; | |
1859 | int ret; | |
b83cc969 | 1860 | bool readonly = false; |
6f72c7e2 | 1861 | struct btrfs_qgroup_inherit *inherit = NULL; |
75eaa0e2 | 1862 | |
325c50e3 JM |
1863 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1864 | return -ENOTDIR; | |
1865 | ||
fa0d2b9b LZ |
1866 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1867 | if (IS_ERR(vol_args)) | |
1868 | return PTR_ERR(vol_args); | |
1869 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
75eaa0e2 | 1870 | |
673990db | 1871 | if (vol_args->flags & ~BTRFS_SUBVOL_CREATE_ARGS_MASK) { |
b83cc969 | 1872 | ret = -EOPNOTSUPP; |
c47ca32d | 1873 | goto free_args; |
72fd032e | 1874 | } |
fa0d2b9b | 1875 | |
b83cc969 LZ |
1876 | if (vol_args->flags & BTRFS_SUBVOL_RDONLY) |
1877 | readonly = true; | |
6f72c7e2 | 1878 | if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) { |
09cbfeaf | 1879 | if (vol_args->size > PAGE_SIZE) { |
6f72c7e2 | 1880 | ret = -EINVAL; |
c47ca32d | 1881 | goto free_args; |
6f72c7e2 AJ |
1882 | } |
1883 | inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size); | |
1884 | if (IS_ERR(inherit)) { | |
1885 | ret = PTR_ERR(inherit); | |
c47ca32d | 1886 | goto free_args; |
6f72c7e2 AJ |
1887 | } |
1888 | } | |
fa0d2b9b | 1889 | |
5d54c67e NB |
1890 | ret = __btrfs_ioctl_snap_create(file, vol_args->name, vol_args->fd, |
1891 | subvol, readonly, inherit); | |
c47ca32d DC |
1892 | if (ret) |
1893 | goto free_inherit; | |
c47ca32d | 1894 | free_inherit: |
6f72c7e2 | 1895 | kfree(inherit); |
c47ca32d DC |
1896 | free_args: |
1897 | kfree(vol_args); | |
f46b5a66 CH |
1898 | return ret; |
1899 | } | |
1900 | ||
0caa102d LZ |
1901 | static noinline int btrfs_ioctl_subvol_getflags(struct file *file, |
1902 | void __user *arg) | |
1903 | { | |
496ad9aa | 1904 | struct inode *inode = file_inode(file); |
0b246afa | 1905 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
0caa102d LZ |
1906 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1907 | int ret = 0; | |
1908 | u64 flags = 0; | |
1909 | ||
4a0cc7ca | 1910 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) |
0caa102d LZ |
1911 | return -EINVAL; |
1912 | ||
0b246afa | 1913 | down_read(&fs_info->subvol_sem); |
0caa102d LZ |
1914 | if (btrfs_root_readonly(root)) |
1915 | flags |= BTRFS_SUBVOL_RDONLY; | |
0b246afa | 1916 | up_read(&fs_info->subvol_sem); |
0caa102d LZ |
1917 | |
1918 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
1919 | ret = -EFAULT; | |
1920 | ||
1921 | return ret; | |
1922 | } | |
1923 | ||
1924 | static noinline int btrfs_ioctl_subvol_setflags(struct file *file, | |
1925 | void __user *arg) | |
1926 | { | |
496ad9aa | 1927 | struct inode *inode = file_inode(file); |
0b246afa | 1928 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
0caa102d LZ |
1929 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1930 | struct btrfs_trans_handle *trans; | |
1931 | u64 root_flags; | |
1932 | u64 flags; | |
1933 | int ret = 0; | |
1934 | ||
bd60ea0f DS |
1935 | if (!inode_owner_or_capable(inode)) |
1936 | return -EPERM; | |
1937 | ||
b9ca0664 LB |
1938 | ret = mnt_want_write_file(file); |
1939 | if (ret) | |
1940 | goto out; | |
0caa102d | 1941 | |
4a0cc7ca | 1942 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
b9ca0664 LB |
1943 | ret = -EINVAL; |
1944 | goto out_drop_write; | |
1945 | } | |
0caa102d | 1946 | |
b9ca0664 LB |
1947 | if (copy_from_user(&flags, arg, sizeof(flags))) { |
1948 | ret = -EFAULT; | |
1949 | goto out_drop_write; | |
1950 | } | |
0caa102d | 1951 | |
b9ca0664 LB |
1952 | if (flags & ~BTRFS_SUBVOL_RDONLY) { |
1953 | ret = -EOPNOTSUPP; | |
1954 | goto out_drop_write; | |
1955 | } | |
0caa102d | 1956 | |
0b246afa | 1957 | down_write(&fs_info->subvol_sem); |
0caa102d LZ |
1958 | |
1959 | /* nothing to do */ | |
1960 | if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root)) | |
b9ca0664 | 1961 | goto out_drop_sem; |
0caa102d LZ |
1962 | |
1963 | root_flags = btrfs_root_flags(&root->root_item); | |
2c686537 | 1964 | if (flags & BTRFS_SUBVOL_RDONLY) { |
0caa102d LZ |
1965 | btrfs_set_root_flags(&root->root_item, |
1966 | root_flags | BTRFS_ROOT_SUBVOL_RDONLY); | |
2c686537 DS |
1967 | } else { |
1968 | /* | |
1969 | * Block RO -> RW transition if this subvolume is involved in | |
1970 | * send | |
1971 | */ | |
1972 | spin_lock(&root->root_item_lock); | |
1973 | if (root->send_in_progress == 0) { | |
1974 | btrfs_set_root_flags(&root->root_item, | |
0caa102d | 1975 | root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY); |
2c686537 DS |
1976 | spin_unlock(&root->root_item_lock); |
1977 | } else { | |
1978 | spin_unlock(&root->root_item_lock); | |
0b246afa JM |
1979 | btrfs_warn(fs_info, |
1980 | "Attempt to set subvolume %llu read-write during send", | |
1981 | root->root_key.objectid); | |
2c686537 DS |
1982 | ret = -EPERM; |
1983 | goto out_drop_sem; | |
1984 | } | |
1985 | } | |
0caa102d LZ |
1986 | |
1987 | trans = btrfs_start_transaction(root, 1); | |
1988 | if (IS_ERR(trans)) { | |
1989 | ret = PTR_ERR(trans); | |
1990 | goto out_reset; | |
1991 | } | |
1992 | ||
0b246afa | 1993 | ret = btrfs_update_root(trans, fs_info->tree_root, |
0caa102d | 1994 | &root->root_key, &root->root_item); |
9417ebc8 NB |
1995 | if (ret < 0) { |
1996 | btrfs_end_transaction(trans); | |
1997 | goto out_reset; | |
1998 | } | |
1999 | ||
2000 | ret = btrfs_commit_transaction(trans); | |
0caa102d | 2001 | |
0caa102d LZ |
2002 | out_reset: |
2003 | if (ret) | |
2004 | btrfs_set_root_flags(&root->root_item, root_flags); | |
b9ca0664 | 2005 | out_drop_sem: |
0b246afa | 2006 | up_write(&fs_info->subvol_sem); |
b9ca0664 LB |
2007 | out_drop_write: |
2008 | mnt_drop_write_file(file); | |
2009 | out: | |
0caa102d LZ |
2010 | return ret; |
2011 | } | |
2012 | ||
ac8e9819 CM |
2013 | static noinline int key_in_sk(struct btrfs_key *key, |
2014 | struct btrfs_ioctl_search_key *sk) | |
2015 | { | |
abc6e134 CM |
2016 | struct btrfs_key test; |
2017 | int ret; | |
2018 | ||
2019 | test.objectid = sk->min_objectid; | |
2020 | test.type = sk->min_type; | |
2021 | test.offset = sk->min_offset; | |
2022 | ||
2023 | ret = btrfs_comp_cpu_keys(key, &test); | |
2024 | if (ret < 0) | |
ac8e9819 | 2025 | return 0; |
abc6e134 CM |
2026 | |
2027 | test.objectid = sk->max_objectid; | |
2028 | test.type = sk->max_type; | |
2029 | test.offset = sk->max_offset; | |
2030 | ||
2031 | ret = btrfs_comp_cpu_keys(key, &test); | |
2032 | if (ret > 0) | |
ac8e9819 CM |
2033 | return 0; |
2034 | return 1; | |
2035 | } | |
2036 | ||
df397565 | 2037 | static noinline int copy_to_sk(struct btrfs_path *path, |
ac8e9819 CM |
2038 | struct btrfs_key *key, |
2039 | struct btrfs_ioctl_search_key *sk, | |
9b6e817d | 2040 | size_t *buf_size, |
ba346b35 | 2041 | char __user *ubuf, |
ac8e9819 CM |
2042 | unsigned long *sk_offset, |
2043 | int *num_found) | |
2044 | { | |
2045 | u64 found_transid; | |
2046 | struct extent_buffer *leaf; | |
2047 | struct btrfs_ioctl_search_header sh; | |
dd81d459 | 2048 | struct btrfs_key test; |
ac8e9819 CM |
2049 | unsigned long item_off; |
2050 | unsigned long item_len; | |
2051 | int nritems; | |
2052 | int i; | |
2053 | int slot; | |
ac8e9819 CM |
2054 | int ret = 0; |
2055 | ||
2056 | leaf = path->nodes[0]; | |
2057 | slot = path->slots[0]; | |
2058 | nritems = btrfs_header_nritems(leaf); | |
2059 | ||
2060 | if (btrfs_header_generation(leaf) > sk->max_transid) { | |
2061 | i = nritems; | |
2062 | goto advance_key; | |
2063 | } | |
2064 | found_transid = btrfs_header_generation(leaf); | |
2065 | ||
2066 | for (i = slot; i < nritems; i++) { | |
2067 | item_off = btrfs_item_ptr_offset(leaf, i); | |
2068 | item_len = btrfs_item_size_nr(leaf, i); | |
2069 | ||
03b71c6c GP |
2070 | btrfs_item_key_to_cpu(leaf, key, i); |
2071 | if (!key_in_sk(key, sk)) | |
2072 | continue; | |
2073 | ||
9b6e817d | 2074 | if (sizeof(sh) + item_len > *buf_size) { |
8f5f6178 GH |
2075 | if (*num_found) { |
2076 | ret = 1; | |
2077 | goto out; | |
2078 | } | |
2079 | ||
2080 | /* | |
2081 | * return one empty item back for v1, which does not | |
2082 | * handle -EOVERFLOW | |
2083 | */ | |
2084 | ||
9b6e817d | 2085 | *buf_size = sizeof(sh) + item_len; |
ac8e9819 | 2086 | item_len = 0; |
8f5f6178 GH |
2087 | ret = -EOVERFLOW; |
2088 | } | |
ac8e9819 | 2089 | |
9b6e817d | 2090 | if (sizeof(sh) + item_len + *sk_offset > *buf_size) { |
ac8e9819 | 2091 | ret = 1; |
25c9bc2e | 2092 | goto out; |
ac8e9819 CM |
2093 | } |
2094 | ||
ac8e9819 CM |
2095 | sh.objectid = key->objectid; |
2096 | sh.offset = key->offset; | |
2097 | sh.type = key->type; | |
2098 | sh.len = item_len; | |
2099 | sh.transid = found_transid; | |
2100 | ||
a48b73ec JB |
2101 | /* |
2102 | * Copy search result header. If we fault then loop again so we | |
2103 | * can fault in the pages and -EFAULT there if there's a | |
2104 | * problem. Otherwise we'll fault and then copy the buffer in | |
2105 | * properly this next time through | |
2106 | */ | |
2107 | if (copy_to_user_nofault(ubuf + *sk_offset, &sh, sizeof(sh))) { | |
2108 | ret = 0; | |
ba346b35 GH |
2109 | goto out; |
2110 | } | |
2111 | ||
ac8e9819 CM |
2112 | *sk_offset += sizeof(sh); |
2113 | ||
2114 | if (item_len) { | |
ba346b35 | 2115 | char __user *up = ubuf + *sk_offset; |
a48b73ec JB |
2116 | /* |
2117 | * Copy the item, same behavior as above, but reset the | |
2118 | * * sk_offset so we copy the full thing again. | |
2119 | */ | |
2120 | if (read_extent_buffer_to_user_nofault(leaf, up, | |
2121 | item_off, item_len)) { | |
2122 | ret = 0; | |
2123 | *sk_offset -= sizeof(sh); | |
ba346b35 GH |
2124 | goto out; |
2125 | } | |
2126 | ||
ac8e9819 | 2127 | *sk_offset += item_len; |
ac8e9819 | 2128 | } |
e2156867 | 2129 | (*num_found)++; |
ac8e9819 | 2130 | |
8f5f6178 GH |
2131 | if (ret) /* -EOVERFLOW from above */ |
2132 | goto out; | |
2133 | ||
25c9bc2e GH |
2134 | if (*num_found >= sk->nr_items) { |
2135 | ret = 1; | |
2136 | goto out; | |
2137 | } | |
ac8e9819 CM |
2138 | } |
2139 | advance_key: | |
abc6e134 | 2140 | ret = 0; |
dd81d459 NA |
2141 | test.objectid = sk->max_objectid; |
2142 | test.type = sk->max_type; | |
2143 | test.offset = sk->max_offset; | |
2144 | if (btrfs_comp_cpu_keys(key, &test) >= 0) | |
2145 | ret = 1; | |
2146 | else if (key->offset < (u64)-1) | |
ac8e9819 | 2147 | key->offset++; |
dd81d459 | 2148 | else if (key->type < (u8)-1) { |
abc6e134 | 2149 | key->offset = 0; |
ac8e9819 | 2150 | key->type++; |
dd81d459 | 2151 | } else if (key->objectid < (u64)-1) { |
abc6e134 CM |
2152 | key->offset = 0; |
2153 | key->type = 0; | |
ac8e9819 | 2154 | key->objectid++; |
abc6e134 CM |
2155 | } else |
2156 | ret = 1; | |
25c9bc2e | 2157 | out: |
ba346b35 GH |
2158 | /* |
2159 | * 0: all items from this leaf copied, continue with next | |
2160 | * 1: * more items can be copied, but unused buffer is too small | |
2161 | * * all items were found | |
2162 | * Either way, it will stops the loop which iterates to the next | |
2163 | * leaf | |
2164 | * -EOVERFLOW: item was to large for buffer | |
2165 | * -EFAULT: could not copy extent buffer back to userspace | |
2166 | */ | |
ac8e9819 CM |
2167 | return ret; |
2168 | } | |
2169 | ||
2170 | static noinline int search_ioctl(struct inode *inode, | |
12544442 | 2171 | struct btrfs_ioctl_search_key *sk, |
9b6e817d | 2172 | size_t *buf_size, |
ba346b35 | 2173 | char __user *ubuf) |
ac8e9819 | 2174 | { |
0b246afa | 2175 | struct btrfs_fs_info *info = btrfs_sb(inode->i_sb); |
ac8e9819 CM |
2176 | struct btrfs_root *root; |
2177 | struct btrfs_key key; | |
ac8e9819 | 2178 | struct btrfs_path *path; |
ac8e9819 CM |
2179 | int ret; |
2180 | int num_found = 0; | |
2181 | unsigned long sk_offset = 0; | |
2182 | ||
9b6e817d GH |
2183 | if (*buf_size < sizeof(struct btrfs_ioctl_search_header)) { |
2184 | *buf_size = sizeof(struct btrfs_ioctl_search_header); | |
12544442 | 2185 | return -EOVERFLOW; |
9b6e817d | 2186 | } |
12544442 | 2187 | |
ac8e9819 CM |
2188 | path = btrfs_alloc_path(); |
2189 | if (!path) | |
2190 | return -ENOMEM; | |
2191 | ||
2192 | if (sk->tree_id == 0) { | |
2193 | /* search the root of the inode that was passed */ | |
00246528 | 2194 | root = btrfs_grab_root(BTRFS_I(inode)->root); |
ac8e9819 | 2195 | } else { |
56e9357a | 2196 | root = btrfs_get_fs_root(info, sk->tree_id, true); |
ac8e9819 | 2197 | if (IS_ERR(root)) { |
ac8e9819 | 2198 | btrfs_free_path(path); |
ad1e3d56 | 2199 | return PTR_ERR(root); |
ac8e9819 CM |
2200 | } |
2201 | } | |
2202 | ||
2203 | key.objectid = sk->min_objectid; | |
2204 | key.type = sk->min_type; | |
2205 | key.offset = sk->min_offset; | |
2206 | ||
67871254 | 2207 | while (1) { |
1c78544e FM |
2208 | ret = fault_in_pages_writeable(ubuf + sk_offset, |
2209 | *buf_size - sk_offset); | |
a48b73ec JB |
2210 | if (ret) |
2211 | break; | |
2212 | ||
6174d3cb | 2213 | ret = btrfs_search_forward(root, &key, path, sk->min_transid); |
ac8e9819 CM |
2214 | if (ret != 0) { |
2215 | if (ret > 0) | |
2216 | ret = 0; | |
2217 | goto err; | |
2218 | } | |
df397565 | 2219 | ret = copy_to_sk(path, &key, sk, buf_size, ubuf, |
ac8e9819 | 2220 | &sk_offset, &num_found); |
b3b4aa74 | 2221 | btrfs_release_path(path); |
25c9bc2e | 2222 | if (ret) |
ac8e9819 CM |
2223 | break; |
2224 | ||
2225 | } | |
8f5f6178 GH |
2226 | if (ret > 0) |
2227 | ret = 0; | |
ac8e9819 CM |
2228 | err: |
2229 | sk->nr_items = num_found; | |
00246528 | 2230 | btrfs_put_root(root); |
ac8e9819 CM |
2231 | btrfs_free_path(path); |
2232 | return ret; | |
2233 | } | |
2234 | ||
2235 | static noinline int btrfs_ioctl_tree_search(struct file *file, | |
2236 | void __user *argp) | |
2237 | { | |
ba346b35 GH |
2238 | struct btrfs_ioctl_search_args __user *uargs; |
2239 | struct btrfs_ioctl_search_key sk; | |
9b6e817d GH |
2240 | struct inode *inode; |
2241 | int ret; | |
2242 | size_t buf_size; | |
ac8e9819 CM |
2243 | |
2244 | if (!capable(CAP_SYS_ADMIN)) | |
2245 | return -EPERM; | |
2246 | ||
ba346b35 GH |
2247 | uargs = (struct btrfs_ioctl_search_args __user *)argp; |
2248 | ||
2249 | if (copy_from_user(&sk, &uargs->key, sizeof(sk))) | |
2250 | return -EFAULT; | |
ac8e9819 | 2251 | |
ba346b35 | 2252 | buf_size = sizeof(uargs->buf); |
ac8e9819 | 2253 | |
496ad9aa | 2254 | inode = file_inode(file); |
ba346b35 | 2255 | ret = search_ioctl(inode, &sk, &buf_size, uargs->buf); |
8f5f6178 GH |
2256 | |
2257 | /* | |
2258 | * In the origin implementation an overflow is handled by returning a | |
2259 | * search header with a len of zero, so reset ret. | |
2260 | */ | |
2261 | if (ret == -EOVERFLOW) | |
2262 | ret = 0; | |
2263 | ||
ba346b35 | 2264 | if (ret == 0 && copy_to_user(&uargs->key, &sk, sizeof(sk))) |
ac8e9819 | 2265 | ret = -EFAULT; |
ac8e9819 CM |
2266 | return ret; |
2267 | } | |
2268 | ||
cc68a8a5 GH |
2269 | static noinline int btrfs_ioctl_tree_search_v2(struct file *file, |
2270 | void __user *argp) | |
2271 | { | |
2272 | struct btrfs_ioctl_search_args_v2 __user *uarg; | |
2273 | struct btrfs_ioctl_search_args_v2 args; | |
2274 | struct inode *inode; | |
2275 | int ret; | |
2276 | size_t buf_size; | |
ee22184b | 2277 | const size_t buf_limit = SZ_16M; |
cc68a8a5 GH |
2278 | |
2279 | if (!capable(CAP_SYS_ADMIN)) | |
2280 | return -EPERM; | |
2281 | ||
2282 | /* copy search header and buffer size */ | |
2283 | uarg = (struct btrfs_ioctl_search_args_v2 __user *)argp; | |
2284 | if (copy_from_user(&args, uarg, sizeof(args))) | |
2285 | return -EFAULT; | |
2286 | ||
2287 | buf_size = args.buf_size; | |
2288 | ||
cc68a8a5 GH |
2289 | /* limit result size to 16MB */ |
2290 | if (buf_size > buf_limit) | |
2291 | buf_size = buf_limit; | |
2292 | ||
2293 | inode = file_inode(file); | |
2294 | ret = search_ioctl(inode, &args.key, &buf_size, | |
718dc5fa | 2295 | (char __user *)(&uarg->buf[0])); |
cc68a8a5 GH |
2296 | if (ret == 0 && copy_to_user(&uarg->key, &args.key, sizeof(args.key))) |
2297 | ret = -EFAULT; | |
2298 | else if (ret == -EOVERFLOW && | |
2299 | copy_to_user(&uarg->buf_size, &buf_size, sizeof(buf_size))) | |
2300 | ret = -EFAULT; | |
2301 | ||
ac8e9819 CM |
2302 | return ret; |
2303 | } | |
2304 | ||
98d377a0 | 2305 | /* |
ac8e9819 CM |
2306 | * Search INODE_REFs to identify path name of 'dirid' directory |
2307 | * in a 'tree_id' tree. and sets path name to 'name'. | |
2308 | */ | |
98d377a0 TH |
2309 | static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, |
2310 | u64 tree_id, u64 dirid, char *name) | |
2311 | { | |
2312 | struct btrfs_root *root; | |
2313 | struct btrfs_key key; | |
ac8e9819 | 2314 | char *ptr; |
98d377a0 TH |
2315 | int ret = -1; |
2316 | int slot; | |
2317 | int len; | |
2318 | int total_len = 0; | |
2319 | struct btrfs_inode_ref *iref; | |
2320 | struct extent_buffer *l; | |
2321 | struct btrfs_path *path; | |
2322 | ||
2323 | if (dirid == BTRFS_FIRST_FREE_OBJECTID) { | |
2324 | name[0]='\0'; | |
2325 | return 0; | |
2326 | } | |
2327 | ||
2328 | path = btrfs_alloc_path(); | |
2329 | if (!path) | |
2330 | return -ENOMEM; | |
2331 | ||
c8bcbfbd | 2332 | ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX - 1]; |
98d377a0 | 2333 | |
56e9357a | 2334 | root = btrfs_get_fs_root(info, tree_id, true); |
98d377a0 | 2335 | if (IS_ERR(root)) { |
ad1e3d56 | 2336 | ret = PTR_ERR(root); |
88234012 JB |
2337 | root = NULL; |
2338 | goto out; | |
2339 | } | |
98d377a0 TH |
2340 | |
2341 | key.objectid = dirid; | |
2342 | key.type = BTRFS_INODE_REF_KEY; | |
8ad6fcab | 2343 | key.offset = (u64)-1; |
98d377a0 | 2344 | |
67871254 | 2345 | while (1) { |
98d377a0 TH |
2346 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); |
2347 | if (ret < 0) | |
2348 | goto out; | |
18674c6c FDBM |
2349 | else if (ret > 0) { |
2350 | ret = btrfs_previous_item(root, path, dirid, | |
2351 | BTRFS_INODE_REF_KEY); | |
2352 | if (ret < 0) | |
2353 | goto out; | |
2354 | else if (ret > 0) { | |
2355 | ret = -ENOENT; | |
2356 | goto out; | |
2357 | } | |
2358 | } | |
98d377a0 TH |
2359 | |
2360 | l = path->nodes[0]; | |
2361 | slot = path->slots[0]; | |
2362 | btrfs_item_key_to_cpu(l, &key, slot); | |
2363 | ||
98d377a0 TH |
2364 | iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); |
2365 | len = btrfs_inode_ref_name_len(l, iref); | |
2366 | ptr -= len + 1; | |
2367 | total_len += len + 1; | |
a696cf35 FDBM |
2368 | if (ptr < name) { |
2369 | ret = -ENAMETOOLONG; | |
98d377a0 | 2370 | goto out; |
a696cf35 | 2371 | } |
98d377a0 TH |
2372 | |
2373 | *(ptr + len) = '/'; | |
67871254 | 2374 | read_extent_buffer(l, ptr, (unsigned long)(iref + 1), len); |
98d377a0 TH |
2375 | |
2376 | if (key.offset == BTRFS_FIRST_FREE_OBJECTID) | |
2377 | break; | |
2378 | ||
b3b4aa74 | 2379 | btrfs_release_path(path); |
98d377a0 | 2380 | key.objectid = key.offset; |
8ad6fcab | 2381 | key.offset = (u64)-1; |
98d377a0 | 2382 | dirid = key.objectid; |
98d377a0 | 2383 | } |
77906a50 | 2384 | memmove(name, ptr, total_len); |
67871254 | 2385 | name[total_len] = '\0'; |
98d377a0 TH |
2386 | ret = 0; |
2387 | out: | |
00246528 | 2388 | btrfs_put_root(root); |
98d377a0 | 2389 | btrfs_free_path(path); |
ac8e9819 CM |
2390 | return ret; |
2391 | } | |
2392 | ||
23d0b79d TM |
2393 | static int btrfs_search_path_in_tree_user(struct inode *inode, |
2394 | struct btrfs_ioctl_ino_lookup_user_args *args) | |
2395 | { | |
2396 | struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info; | |
2397 | struct super_block *sb = inode->i_sb; | |
2398 | struct btrfs_key upper_limit = BTRFS_I(inode)->location; | |
2399 | u64 treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2400 | u64 dirid = args->dirid; | |
2401 | unsigned long item_off; | |
2402 | unsigned long item_len; | |
2403 | struct btrfs_inode_ref *iref; | |
2404 | struct btrfs_root_ref *rref; | |
b8a49ae1 | 2405 | struct btrfs_root *root = NULL; |
23d0b79d TM |
2406 | struct btrfs_path *path; |
2407 | struct btrfs_key key, key2; | |
2408 | struct extent_buffer *leaf; | |
2409 | struct inode *temp_inode; | |
2410 | char *ptr; | |
2411 | int slot; | |
2412 | int len; | |
2413 | int total_len = 0; | |
2414 | int ret; | |
2415 | ||
2416 | path = btrfs_alloc_path(); | |
2417 | if (!path) | |
2418 | return -ENOMEM; | |
2419 | ||
2420 | /* | |
2421 | * If the bottom subvolume does not exist directly under upper_limit, | |
2422 | * construct the path in from the bottom up. | |
2423 | */ | |
2424 | if (dirid != upper_limit.objectid) { | |
2425 | ptr = &args->path[BTRFS_INO_LOOKUP_USER_PATH_MAX - 1]; | |
2426 | ||
56e9357a | 2427 | root = btrfs_get_fs_root(fs_info, treeid, true); |
23d0b79d TM |
2428 | if (IS_ERR(root)) { |
2429 | ret = PTR_ERR(root); | |
2430 | goto out; | |
2431 | } | |
2432 | ||
2433 | key.objectid = dirid; | |
2434 | key.type = BTRFS_INODE_REF_KEY; | |
2435 | key.offset = (u64)-1; | |
2436 | while (1) { | |
2437 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2438 | if (ret < 0) { | |
b8a49ae1 | 2439 | goto out_put; |
23d0b79d TM |
2440 | } else if (ret > 0) { |
2441 | ret = btrfs_previous_item(root, path, dirid, | |
2442 | BTRFS_INODE_REF_KEY); | |
2443 | if (ret < 0) { | |
b8a49ae1 | 2444 | goto out_put; |
23d0b79d TM |
2445 | } else if (ret > 0) { |
2446 | ret = -ENOENT; | |
b8a49ae1 | 2447 | goto out_put; |
23d0b79d TM |
2448 | } |
2449 | } | |
2450 | ||
2451 | leaf = path->nodes[0]; | |
2452 | slot = path->slots[0]; | |
2453 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2454 | ||
2455 | iref = btrfs_item_ptr(leaf, slot, struct btrfs_inode_ref); | |
2456 | len = btrfs_inode_ref_name_len(leaf, iref); | |
2457 | ptr -= len + 1; | |
2458 | total_len += len + 1; | |
2459 | if (ptr < args->path) { | |
2460 | ret = -ENAMETOOLONG; | |
b8a49ae1 | 2461 | goto out_put; |
23d0b79d TM |
2462 | } |
2463 | ||
2464 | *(ptr + len) = '/'; | |
2465 | read_extent_buffer(leaf, ptr, | |
2466 | (unsigned long)(iref + 1), len); | |
2467 | ||
2468 | /* Check the read+exec permission of this directory */ | |
2469 | ret = btrfs_previous_item(root, path, dirid, | |
2470 | BTRFS_INODE_ITEM_KEY); | |
2471 | if (ret < 0) { | |
b8a49ae1 | 2472 | goto out_put; |
23d0b79d TM |
2473 | } else if (ret > 0) { |
2474 | ret = -ENOENT; | |
b8a49ae1 | 2475 | goto out_put; |
23d0b79d TM |
2476 | } |
2477 | ||
2478 | leaf = path->nodes[0]; | |
2479 | slot = path->slots[0]; | |
2480 | btrfs_item_key_to_cpu(leaf, &key2, slot); | |
2481 | if (key2.objectid != dirid) { | |
2482 | ret = -ENOENT; | |
b8a49ae1 | 2483 | goto out_put; |
23d0b79d TM |
2484 | } |
2485 | ||
0202e83f | 2486 | temp_inode = btrfs_iget(sb, key2.objectid, root); |
3ca57bd6 MT |
2487 | if (IS_ERR(temp_inode)) { |
2488 | ret = PTR_ERR(temp_inode); | |
b8a49ae1 | 2489 | goto out_put; |
3ca57bd6 | 2490 | } |
23d0b79d TM |
2491 | ret = inode_permission(temp_inode, MAY_READ | MAY_EXEC); |
2492 | iput(temp_inode); | |
2493 | if (ret) { | |
2494 | ret = -EACCES; | |
b8a49ae1 | 2495 | goto out_put; |
23d0b79d TM |
2496 | } |
2497 | ||
2498 | if (key.offset == upper_limit.objectid) | |
2499 | break; | |
2500 | if (key.objectid == BTRFS_FIRST_FREE_OBJECTID) { | |
2501 | ret = -EACCES; | |
b8a49ae1 | 2502 | goto out_put; |
23d0b79d TM |
2503 | } |
2504 | ||
2505 | btrfs_release_path(path); | |
2506 | key.objectid = key.offset; | |
2507 | key.offset = (u64)-1; | |
2508 | dirid = key.objectid; | |
2509 | } | |
2510 | ||
2511 | memmove(args->path, ptr, total_len); | |
2512 | args->path[total_len] = '\0'; | |
00246528 | 2513 | btrfs_put_root(root); |
b8a49ae1 | 2514 | root = NULL; |
23d0b79d TM |
2515 | btrfs_release_path(path); |
2516 | } | |
2517 | ||
2518 | /* Get the bottom subvolume's name from ROOT_REF */ | |
23d0b79d TM |
2519 | key.objectid = treeid; |
2520 | key.type = BTRFS_ROOT_REF_KEY; | |
2521 | key.offset = args->treeid; | |
b8a49ae1 | 2522 | ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); |
23d0b79d TM |
2523 | if (ret < 0) { |
2524 | goto out; | |
2525 | } else if (ret > 0) { | |
2526 | ret = -ENOENT; | |
2527 | goto out; | |
2528 | } | |
2529 | ||
2530 | leaf = path->nodes[0]; | |
2531 | slot = path->slots[0]; | |
2532 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2533 | ||
2534 | item_off = btrfs_item_ptr_offset(leaf, slot); | |
2535 | item_len = btrfs_item_size_nr(leaf, slot); | |
2536 | /* Check if dirid in ROOT_REF corresponds to passed dirid */ | |
2537 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2538 | if (args->dirid != btrfs_root_ref_dirid(leaf, rref)) { | |
2539 | ret = -EINVAL; | |
2540 | goto out; | |
2541 | } | |
2542 | ||
2543 | /* Copy subvolume's name */ | |
2544 | item_off += sizeof(struct btrfs_root_ref); | |
2545 | item_len -= sizeof(struct btrfs_root_ref); | |
2546 | read_extent_buffer(leaf, args->name, item_off, item_len); | |
2547 | args->name[item_len] = 0; | |
2548 | ||
b8a49ae1 | 2549 | out_put: |
00246528 | 2550 | btrfs_put_root(root); |
23d0b79d TM |
2551 | out: |
2552 | btrfs_free_path(path); | |
2553 | return ret; | |
2554 | } | |
2555 | ||
ac8e9819 CM |
2556 | static noinline int btrfs_ioctl_ino_lookup(struct file *file, |
2557 | void __user *argp) | |
2558 | { | |
bece2e82 BVA |
2559 | struct btrfs_ioctl_ino_lookup_args *args; |
2560 | struct inode *inode; | |
01b810b8 | 2561 | int ret = 0; |
ac8e9819 | 2562 | |
2354d08f JL |
2563 | args = memdup_user(argp, sizeof(*args)); |
2564 | if (IS_ERR(args)) | |
2565 | return PTR_ERR(args); | |
c2b96929 | 2566 | |
496ad9aa | 2567 | inode = file_inode(file); |
ac8e9819 | 2568 | |
01b810b8 DS |
2569 | /* |
2570 | * Unprivileged query to obtain the containing subvolume root id. The | |
2571 | * path is reset so it's consistent with btrfs_search_path_in_tree. | |
2572 | */ | |
1b53ac4d CM |
2573 | if (args->treeid == 0) |
2574 | args->treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2575 | ||
01b810b8 DS |
2576 | if (args->objectid == BTRFS_FIRST_FREE_OBJECTID) { |
2577 | args->name[0] = 0; | |
2578 | goto out; | |
2579 | } | |
2580 | ||
2581 | if (!capable(CAP_SYS_ADMIN)) { | |
2582 | ret = -EPERM; | |
2583 | goto out; | |
2584 | } | |
2585 | ||
ac8e9819 CM |
2586 | ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, |
2587 | args->treeid, args->objectid, | |
2588 | args->name); | |
2589 | ||
01b810b8 | 2590 | out: |
ac8e9819 CM |
2591 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) |
2592 | ret = -EFAULT; | |
2593 | ||
2594 | kfree(args); | |
98d377a0 TH |
2595 | return ret; |
2596 | } | |
2597 | ||
23d0b79d TM |
2598 | /* |
2599 | * Version of ino_lookup ioctl (unprivileged) | |
2600 | * | |
2601 | * The main differences from ino_lookup ioctl are: | |
2602 | * | |
2603 | * 1. Read + Exec permission will be checked using inode_permission() during | |
2604 | * path construction. -EACCES will be returned in case of failure. | |
2605 | * 2. Path construction will be stopped at the inode number which corresponds | |
2606 | * to the fd with which this ioctl is called. If constructed path does not | |
2607 | * exist under fd's inode, -EACCES will be returned. | |
2608 | * 3. The name of bottom subvolume is also searched and filled. | |
2609 | */ | |
2610 | static int btrfs_ioctl_ino_lookup_user(struct file *file, void __user *argp) | |
2611 | { | |
2612 | struct btrfs_ioctl_ino_lookup_user_args *args; | |
2613 | struct inode *inode; | |
2614 | int ret; | |
2615 | ||
2616 | args = memdup_user(argp, sizeof(*args)); | |
2617 | if (IS_ERR(args)) | |
2618 | return PTR_ERR(args); | |
2619 | ||
2620 | inode = file_inode(file); | |
2621 | ||
2622 | if (args->dirid == BTRFS_FIRST_FREE_OBJECTID && | |
2623 | BTRFS_I(inode)->location.objectid != BTRFS_FIRST_FREE_OBJECTID) { | |
2624 | /* | |
2625 | * The subvolume does not exist under fd with which this is | |
2626 | * called | |
2627 | */ | |
2628 | kfree(args); | |
2629 | return -EACCES; | |
2630 | } | |
2631 | ||
2632 | ret = btrfs_search_path_in_tree_user(inode, args); | |
2633 | ||
2634 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
2635 | ret = -EFAULT; | |
2636 | ||
2637 | kfree(args); | |
2638 | return ret; | |
2639 | } | |
2640 | ||
b64ec075 TM |
2641 | /* Get the subvolume information in BTRFS_ROOT_ITEM and BTRFS_ROOT_BACKREF */ |
2642 | static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp) | |
2643 | { | |
2644 | struct btrfs_ioctl_get_subvol_info_args *subvol_info; | |
2645 | struct btrfs_fs_info *fs_info; | |
2646 | struct btrfs_root *root; | |
2647 | struct btrfs_path *path; | |
2648 | struct btrfs_key key; | |
2649 | struct btrfs_root_item *root_item; | |
2650 | struct btrfs_root_ref *rref; | |
2651 | struct extent_buffer *leaf; | |
2652 | unsigned long item_off; | |
2653 | unsigned long item_len; | |
2654 | struct inode *inode; | |
2655 | int slot; | |
2656 | int ret = 0; | |
2657 | ||
2658 | path = btrfs_alloc_path(); | |
2659 | if (!path) | |
2660 | return -ENOMEM; | |
2661 | ||
2662 | subvol_info = kzalloc(sizeof(*subvol_info), GFP_KERNEL); | |
2663 | if (!subvol_info) { | |
2664 | btrfs_free_path(path); | |
2665 | return -ENOMEM; | |
2666 | } | |
2667 | ||
2668 | inode = file_inode(file); | |
2669 | fs_info = BTRFS_I(inode)->root->fs_info; | |
2670 | ||
2671 | /* Get root_item of inode's subvolume */ | |
2672 | key.objectid = BTRFS_I(inode)->root->root_key.objectid; | |
56e9357a | 2673 | root = btrfs_get_fs_root(fs_info, key.objectid, true); |
b64ec075 TM |
2674 | if (IS_ERR(root)) { |
2675 | ret = PTR_ERR(root); | |
04734e84 JB |
2676 | goto out_free; |
2677 | } | |
b64ec075 TM |
2678 | root_item = &root->root_item; |
2679 | ||
2680 | subvol_info->treeid = key.objectid; | |
2681 | ||
2682 | subvol_info->generation = btrfs_root_generation(root_item); | |
2683 | subvol_info->flags = btrfs_root_flags(root_item); | |
2684 | ||
2685 | memcpy(subvol_info->uuid, root_item->uuid, BTRFS_UUID_SIZE); | |
2686 | memcpy(subvol_info->parent_uuid, root_item->parent_uuid, | |
2687 | BTRFS_UUID_SIZE); | |
2688 | memcpy(subvol_info->received_uuid, root_item->received_uuid, | |
2689 | BTRFS_UUID_SIZE); | |
2690 | ||
2691 | subvol_info->ctransid = btrfs_root_ctransid(root_item); | |
2692 | subvol_info->ctime.sec = btrfs_stack_timespec_sec(&root_item->ctime); | |
2693 | subvol_info->ctime.nsec = btrfs_stack_timespec_nsec(&root_item->ctime); | |
2694 | ||
2695 | subvol_info->otransid = btrfs_root_otransid(root_item); | |
2696 | subvol_info->otime.sec = btrfs_stack_timespec_sec(&root_item->otime); | |
2697 | subvol_info->otime.nsec = btrfs_stack_timespec_nsec(&root_item->otime); | |
2698 | ||
2699 | subvol_info->stransid = btrfs_root_stransid(root_item); | |
2700 | subvol_info->stime.sec = btrfs_stack_timespec_sec(&root_item->stime); | |
2701 | subvol_info->stime.nsec = btrfs_stack_timespec_nsec(&root_item->stime); | |
2702 | ||
2703 | subvol_info->rtransid = btrfs_root_rtransid(root_item); | |
2704 | subvol_info->rtime.sec = btrfs_stack_timespec_sec(&root_item->rtime); | |
2705 | subvol_info->rtime.nsec = btrfs_stack_timespec_nsec(&root_item->rtime); | |
2706 | ||
2707 | if (key.objectid != BTRFS_FS_TREE_OBJECTID) { | |
2708 | /* Search root tree for ROOT_BACKREF of this subvolume */ | |
b64ec075 TM |
2709 | key.type = BTRFS_ROOT_BACKREF_KEY; |
2710 | key.offset = 0; | |
04734e84 | 2711 | ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); |
b64ec075 TM |
2712 | if (ret < 0) { |
2713 | goto out; | |
2714 | } else if (path->slots[0] >= | |
2715 | btrfs_header_nritems(path->nodes[0])) { | |
04734e84 | 2716 | ret = btrfs_next_leaf(fs_info->tree_root, path); |
b64ec075 TM |
2717 | if (ret < 0) { |
2718 | goto out; | |
2719 | } else if (ret > 0) { | |
2720 | ret = -EUCLEAN; | |
2721 | goto out; | |
2722 | } | |
2723 | } | |
2724 | ||
2725 | leaf = path->nodes[0]; | |
2726 | slot = path->slots[0]; | |
2727 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2728 | if (key.objectid == subvol_info->treeid && | |
2729 | key.type == BTRFS_ROOT_BACKREF_KEY) { | |
2730 | subvol_info->parent_id = key.offset; | |
2731 | ||
2732 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2733 | subvol_info->dirid = btrfs_root_ref_dirid(leaf, rref); | |
2734 | ||
2735 | item_off = btrfs_item_ptr_offset(leaf, slot) | |
2736 | + sizeof(struct btrfs_root_ref); | |
2737 | item_len = btrfs_item_size_nr(leaf, slot) | |
2738 | - sizeof(struct btrfs_root_ref); | |
2739 | read_extent_buffer(leaf, subvol_info->name, | |
2740 | item_off, item_len); | |
2741 | } else { | |
2742 | ret = -ENOENT; | |
2743 | goto out; | |
2744 | } | |
2745 | } | |
2746 | ||
2747 | if (copy_to_user(argp, subvol_info, sizeof(*subvol_info))) | |
2748 | ret = -EFAULT; | |
2749 | ||
2750 | out: | |
00246528 | 2751 | btrfs_put_root(root); |
04734e84 | 2752 | out_free: |
b64ec075 | 2753 | btrfs_free_path(path); |
b091f7fe | 2754 | kfree(subvol_info); |
b64ec075 TM |
2755 | return ret; |
2756 | } | |
2757 | ||
42e4b520 TM |
2758 | /* |
2759 | * Return ROOT_REF information of the subvolume containing this inode | |
2760 | * except the subvolume name. | |
2761 | */ | |
2762 | static int btrfs_ioctl_get_subvol_rootref(struct file *file, void __user *argp) | |
2763 | { | |
2764 | struct btrfs_ioctl_get_subvol_rootref_args *rootrefs; | |
2765 | struct btrfs_root_ref *rref; | |
2766 | struct btrfs_root *root; | |
2767 | struct btrfs_path *path; | |
2768 | struct btrfs_key key; | |
2769 | struct extent_buffer *leaf; | |
2770 | struct inode *inode; | |
2771 | u64 objectid; | |
2772 | int slot; | |
2773 | int ret; | |
2774 | u8 found; | |
2775 | ||
2776 | path = btrfs_alloc_path(); | |
2777 | if (!path) | |
2778 | return -ENOMEM; | |
2779 | ||
2780 | rootrefs = memdup_user(argp, sizeof(*rootrefs)); | |
2781 | if (IS_ERR(rootrefs)) { | |
2782 | btrfs_free_path(path); | |
2783 | return PTR_ERR(rootrefs); | |
2784 | } | |
2785 | ||
2786 | inode = file_inode(file); | |
2787 | root = BTRFS_I(inode)->root->fs_info->tree_root; | |
2788 | objectid = BTRFS_I(inode)->root->root_key.objectid; | |
2789 | ||
2790 | key.objectid = objectid; | |
2791 | key.type = BTRFS_ROOT_REF_KEY; | |
2792 | key.offset = rootrefs->min_treeid; | |
2793 | found = 0; | |
2794 | ||
2795 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2796 | if (ret < 0) { | |
2797 | goto out; | |
2798 | } else if (path->slots[0] >= | |
2799 | btrfs_header_nritems(path->nodes[0])) { | |
2800 | ret = btrfs_next_leaf(root, path); | |
2801 | if (ret < 0) { | |
2802 | goto out; | |
2803 | } else if (ret > 0) { | |
2804 | ret = -EUCLEAN; | |
2805 | goto out; | |
2806 | } | |
2807 | } | |
2808 | while (1) { | |
2809 | leaf = path->nodes[0]; | |
2810 | slot = path->slots[0]; | |
2811 | ||
2812 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2813 | if (key.objectid != objectid || key.type != BTRFS_ROOT_REF_KEY) { | |
2814 | ret = 0; | |
2815 | goto out; | |
2816 | } | |
2817 | ||
2818 | if (found == BTRFS_MAX_ROOTREF_BUFFER_NUM) { | |
2819 | ret = -EOVERFLOW; | |
2820 | goto out; | |
2821 | } | |
2822 | ||
2823 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2824 | rootrefs->rootref[found].treeid = key.offset; | |
2825 | rootrefs->rootref[found].dirid = | |
2826 | btrfs_root_ref_dirid(leaf, rref); | |
2827 | found++; | |
2828 | ||
2829 | ret = btrfs_next_item(root, path); | |
2830 | if (ret < 0) { | |
2831 | goto out; | |
2832 | } else if (ret > 0) { | |
2833 | ret = -EUCLEAN; | |
2834 | goto out; | |
2835 | } | |
2836 | } | |
2837 | ||
2838 | out: | |
2839 | if (!ret || ret == -EOVERFLOW) { | |
2840 | rootrefs->num_items = found; | |
2841 | /* update min_treeid for next search */ | |
2842 | if (found) | |
2843 | rootrefs->min_treeid = | |
2844 | rootrefs->rootref[found - 1].treeid + 1; | |
2845 | if (copy_to_user(argp, rootrefs, sizeof(*rootrefs))) | |
2846 | ret = -EFAULT; | |
2847 | } | |
2848 | ||
2849 | kfree(rootrefs); | |
2850 | btrfs_free_path(path); | |
2851 | ||
2852 | return ret; | |
2853 | } | |
2854 | ||
76dda93c | 2855 | static noinline int btrfs_ioctl_snap_destroy(struct file *file, |
949964c9 MPS |
2856 | void __user *arg, |
2857 | bool destroy_v2) | |
76dda93c | 2858 | { |
54563d41 | 2859 | struct dentry *parent = file->f_path.dentry; |
0b246afa | 2860 | struct btrfs_fs_info *fs_info = btrfs_sb(parent->d_sb); |
76dda93c | 2861 | struct dentry *dentry; |
2b0143b5 | 2862 | struct inode *dir = d_inode(parent); |
76dda93c YZ |
2863 | struct inode *inode; |
2864 | struct btrfs_root *root = BTRFS_I(dir)->root; | |
2865 | struct btrfs_root *dest = NULL; | |
949964c9 MPS |
2866 | struct btrfs_ioctl_vol_args *vol_args = NULL; |
2867 | struct btrfs_ioctl_vol_args_v2 *vol_args2 = NULL; | |
2868 | char *subvol_name, *subvol_name_ptr = NULL; | |
2869 | int subvol_namelen; | |
76dda93c | 2870 | int err = 0; |
949964c9 | 2871 | bool destroy_parent = false; |
76dda93c | 2872 | |
949964c9 MPS |
2873 | if (destroy_v2) { |
2874 | vol_args2 = memdup_user(arg, sizeof(*vol_args2)); | |
2875 | if (IS_ERR(vol_args2)) | |
2876 | return PTR_ERR(vol_args2); | |
325c50e3 | 2877 | |
949964c9 MPS |
2878 | if (vol_args2->flags & ~BTRFS_SUBVOL_DELETE_ARGS_MASK) { |
2879 | err = -EOPNOTSUPP; | |
2880 | goto out; | |
2881 | } | |
76dda93c | 2882 | |
949964c9 MPS |
2883 | /* |
2884 | * If SPEC_BY_ID is not set, we are looking for the subvolume by | |
2885 | * name, same as v1 currently does. | |
2886 | */ | |
2887 | if (!(vol_args2->flags & BTRFS_SUBVOL_SPEC_BY_ID)) { | |
2888 | vol_args2->name[BTRFS_SUBVOL_NAME_MAX] = 0; | |
2889 | subvol_name = vol_args2->name; | |
2890 | ||
2891 | err = mnt_want_write_file(file); | |
2892 | if (err) | |
2893 | goto out; | |
2894 | } else { | |
2895 | if (vol_args2->subvolid < BTRFS_FIRST_FREE_OBJECTID) { | |
2896 | err = -EINVAL; | |
2897 | goto out; | |
2898 | } | |
2899 | ||
2900 | err = mnt_want_write_file(file); | |
2901 | if (err) | |
2902 | goto out; | |
2903 | ||
2904 | dentry = btrfs_get_dentry(fs_info->sb, | |
2905 | BTRFS_FIRST_FREE_OBJECTID, | |
2906 | vol_args2->subvolid, 0, 0); | |
2907 | if (IS_ERR(dentry)) { | |
2908 | err = PTR_ERR(dentry); | |
2909 | goto out_drop_write; | |
2910 | } | |
2911 | ||
2912 | /* | |
2913 | * Change the default parent since the subvolume being | |
2914 | * deleted can be outside of the current mount point. | |
2915 | */ | |
2916 | parent = btrfs_get_parent(dentry); | |
2917 | ||
2918 | /* | |
2919 | * At this point dentry->d_name can point to '/' if the | |
2920 | * subvolume we want to destroy is outsite of the | |
2921 | * current mount point, so we need to release the | |
2922 | * current dentry and execute the lookup to return a new | |
2923 | * one with ->d_name pointing to the | |
2924 | * <mount point>/subvol_name. | |
2925 | */ | |
2926 | dput(dentry); | |
2927 | if (IS_ERR(parent)) { | |
2928 | err = PTR_ERR(parent); | |
2929 | goto out_drop_write; | |
2930 | } | |
2931 | dir = d_inode(parent); | |
2932 | ||
2933 | /* | |
2934 | * If v2 was used with SPEC_BY_ID, a new parent was | |
2935 | * allocated since the subvolume can be outside of the | |
2936 | * current mount point. Later on we need to release this | |
2937 | * new parent dentry. | |
2938 | */ | |
2939 | destroy_parent = true; | |
2940 | ||
2941 | subvol_name_ptr = btrfs_get_subvol_name_from_objectid( | |
2942 | fs_info, vol_args2->subvolid); | |
2943 | if (IS_ERR(subvol_name_ptr)) { | |
2944 | err = PTR_ERR(subvol_name_ptr); | |
2945 | goto free_parent; | |
2946 | } | |
2947 | /* subvol_name_ptr is already NULL termined */ | |
2948 | subvol_name = (char *)kbasename(subvol_name_ptr); | |
2949 | } | |
2950 | } else { | |
2951 | vol_args = memdup_user(arg, sizeof(*vol_args)); | |
2952 | if (IS_ERR(vol_args)) | |
2953 | return PTR_ERR(vol_args); | |
2954 | ||
2955 | vol_args->name[BTRFS_PATH_NAME_MAX] = 0; | |
2956 | subvol_name = vol_args->name; | |
2957 | ||
2958 | err = mnt_want_write_file(file); | |
2959 | if (err) | |
2960 | goto out; | |
76dda93c YZ |
2961 | } |
2962 | ||
949964c9 | 2963 | subvol_namelen = strlen(subvol_name); |
76dda93c | 2964 | |
949964c9 MPS |
2965 | if (strchr(subvol_name, '/') || |
2966 | strncmp(subvol_name, "..", subvol_namelen) == 0) { | |
2967 | err = -EINVAL; | |
2968 | goto free_subvol_name; | |
2969 | } | |
2970 | ||
2971 | if (!S_ISDIR(dir->i_mode)) { | |
2972 | err = -ENOTDIR; | |
2973 | goto free_subvol_name; | |
2974 | } | |
521e0546 | 2975 | |
00235411 AV |
2976 | err = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT); |
2977 | if (err == -EINTR) | |
949964c9 MPS |
2978 | goto free_subvol_name; |
2979 | dentry = lookup_one_len(subvol_name, parent, subvol_namelen); | |
76dda93c YZ |
2980 | if (IS_ERR(dentry)) { |
2981 | err = PTR_ERR(dentry); | |
2982 | goto out_unlock_dir; | |
2983 | } | |
2984 | ||
2b0143b5 | 2985 | if (d_really_is_negative(dentry)) { |
76dda93c YZ |
2986 | err = -ENOENT; |
2987 | goto out_dput; | |
2988 | } | |
2989 | ||
2b0143b5 | 2990 | inode = d_inode(dentry); |
4260f7c7 | 2991 | dest = BTRFS_I(inode)->root; |
67871254 | 2992 | if (!capable(CAP_SYS_ADMIN)) { |
4260f7c7 SW |
2993 | /* |
2994 | * Regular user. Only allow this with a special mount | |
2995 | * option, when the user has write+exec access to the | |
2996 | * subvol root, and when rmdir(2) would have been | |
2997 | * allowed. | |
2998 | * | |
2999 | * Note that this is _not_ check that the subvol is | |
3000 | * empty or doesn't contain data that we wouldn't | |
3001 | * otherwise be able to delete. | |
3002 | * | |
3003 | * Users who want to delete empty subvols should try | |
3004 | * rmdir(2). | |
3005 | */ | |
3006 | err = -EPERM; | |
0b246afa | 3007 | if (!btrfs_test_opt(fs_info, USER_SUBVOL_RM_ALLOWED)) |
4260f7c7 SW |
3008 | goto out_dput; |
3009 | ||
3010 | /* | |
3011 | * Do not allow deletion if the parent dir is the same | |
3012 | * as the dir to be deleted. That means the ioctl | |
3013 | * must be called on the dentry referencing the root | |
3014 | * of the subvol, not a random directory contained | |
3015 | * within it. | |
3016 | */ | |
3017 | err = -EINVAL; | |
3018 | if (root == dest) | |
3019 | goto out_dput; | |
3020 | ||
3021 | err = inode_permission(inode, MAY_WRITE | MAY_EXEC); | |
3022 | if (err) | |
3023 | goto out_dput; | |
4260f7c7 SW |
3024 | } |
3025 | ||
5c39da5b MX |
3026 | /* check if subvolume may be deleted by a user */ |
3027 | err = btrfs_may_delete(dir, dentry, 1); | |
3028 | if (err) | |
3029 | goto out_dput; | |
3030 | ||
4a0cc7ca | 3031 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
76dda93c YZ |
3032 | err = -EINVAL; |
3033 | goto out_dput; | |
3034 | } | |
3035 | ||
5955102c | 3036 | inode_lock(inode); |
f60a2364 | 3037 | err = btrfs_delete_subvolume(dir, dentry); |
5955102c | 3038 | inode_unlock(inode); |
46008d9d AG |
3039 | if (!err) { |
3040 | fsnotify_rmdir(dir, dentry); | |
76dda93c | 3041 | d_delete(dentry); |
46008d9d | 3042 | } |
fa6ac876 | 3043 | |
76dda93c YZ |
3044 | out_dput: |
3045 | dput(dentry); | |
3046 | out_unlock_dir: | |
5955102c | 3047 | inode_unlock(dir); |
949964c9 MPS |
3048 | free_subvol_name: |
3049 | kfree(subvol_name_ptr); | |
3050 | free_parent: | |
3051 | if (destroy_parent) | |
3052 | dput(parent); | |
00235411 | 3053 | out_drop_write: |
2a79f17e | 3054 | mnt_drop_write_file(file); |
76dda93c | 3055 | out: |
949964c9 | 3056 | kfree(vol_args2); |
76dda93c YZ |
3057 | kfree(vol_args); |
3058 | return err; | |
3059 | } | |
3060 | ||
1e701a32 | 3061 | static int btrfs_ioctl_defrag(struct file *file, void __user *argp) |
f46b5a66 | 3062 | { |
496ad9aa | 3063 | struct inode *inode = file_inode(file); |
f46b5a66 | 3064 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1e701a32 | 3065 | struct btrfs_ioctl_defrag_range_args *range; |
c146afad YZ |
3066 | int ret; |
3067 | ||
25122d15 ID |
3068 | ret = mnt_want_write_file(file); |
3069 | if (ret) | |
3070 | return ret; | |
b83cc969 | 3071 | |
25122d15 ID |
3072 | if (btrfs_root_readonly(root)) { |
3073 | ret = -EROFS; | |
3074 | goto out; | |
5ac00add | 3075 | } |
f46b5a66 CH |
3076 | |
3077 | switch (inode->i_mode & S_IFMT) { | |
3078 | case S_IFDIR: | |
e441d54d CM |
3079 | if (!capable(CAP_SYS_ADMIN)) { |
3080 | ret = -EPERM; | |
3081 | goto out; | |
3082 | } | |
de78b51a | 3083 | ret = btrfs_defrag_root(root); |
f46b5a66 CH |
3084 | break; |
3085 | case S_IFREG: | |
616d374e AB |
3086 | /* |
3087 | * Note that this does not check the file descriptor for write | |
3088 | * access. This prevents defragmenting executables that are | |
3089 | * running and allows defrag on files open in read-only mode. | |
3090 | */ | |
3091 | if (!capable(CAP_SYS_ADMIN) && | |
3092 | inode_permission(inode, MAY_WRITE)) { | |
3093 | ret = -EPERM; | |
e441d54d CM |
3094 | goto out; |
3095 | } | |
1e701a32 CM |
3096 | |
3097 | range = kzalloc(sizeof(*range), GFP_KERNEL); | |
3098 | if (!range) { | |
3099 | ret = -ENOMEM; | |
3100 | goto out; | |
3101 | } | |
3102 | ||
3103 | if (argp) { | |
3104 | if (copy_from_user(range, argp, | |
3105 | sizeof(*range))) { | |
3106 | ret = -EFAULT; | |
3107 | kfree(range); | |
683be16e | 3108 | goto out; |
1e701a32 CM |
3109 | } |
3110 | /* compression requires us to start the IO */ | |
3111 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
3112 | range->flags |= BTRFS_DEFRAG_RANGE_START_IO; | |
3113 | range->extent_thresh = (u32)-1; | |
3114 | } | |
3115 | } else { | |
3116 | /* the rest are all set to zero by kzalloc */ | |
3117 | range->len = (u64)-1; | |
3118 | } | |
496ad9aa | 3119 | ret = btrfs_defrag_file(file_inode(file), file, |
7c829b72 | 3120 | range, BTRFS_OLDEST_GENERATION, 0); |
4cb5300b CM |
3121 | if (ret > 0) |
3122 | ret = 0; | |
1e701a32 | 3123 | kfree(range); |
f46b5a66 | 3124 | break; |
8929ecfa YZ |
3125 | default: |
3126 | ret = -EINVAL; | |
f46b5a66 | 3127 | } |
e441d54d | 3128 | out: |
25122d15 | 3129 | mnt_drop_write_file(file); |
e441d54d | 3130 | return ret; |
f46b5a66 CH |
3131 | } |
3132 | ||
2ff7e61e | 3133 | static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg) |
f46b5a66 CH |
3134 | { |
3135 | struct btrfs_ioctl_vol_args *vol_args; | |
3136 | int ret; | |
3137 | ||
e441d54d CM |
3138 | if (!capable(CAP_SYS_ADMIN)) |
3139 | return -EPERM; | |
3140 | ||
c3e1f96c | 3141 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_ADD)) |
e57138b3 | 3142 | return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
c9e9f97b | 3143 | |
dae7b665 | 3144 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
3145 | if (IS_ERR(vol_args)) { |
3146 | ret = PTR_ERR(vol_args); | |
3147 | goto out; | |
3148 | } | |
f46b5a66 | 3149 | |
5516e595 | 3150 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
2ff7e61e | 3151 | ret = btrfs_init_new_device(fs_info, vol_args->name); |
f46b5a66 | 3152 | |
43d20761 | 3153 | if (!ret) |
0b246afa | 3154 | btrfs_info(fs_info, "disk added %s", vol_args->name); |
43d20761 | 3155 | |
f46b5a66 | 3156 | kfree(vol_args); |
c9e9f97b | 3157 | out: |
c3e1f96c | 3158 | btrfs_exclop_finish(fs_info); |
f46b5a66 CH |
3159 | return ret; |
3160 | } | |
3161 | ||
6b526ed7 | 3162 | static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg) |
f46b5a66 | 3163 | { |
0b246afa JM |
3164 | struct inode *inode = file_inode(file); |
3165 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
6b526ed7 | 3166 | struct btrfs_ioctl_vol_args_v2 *vol_args; |
f46b5a66 CH |
3167 | int ret; |
3168 | ||
e441d54d CM |
3169 | if (!capable(CAP_SYS_ADMIN)) |
3170 | return -EPERM; | |
3171 | ||
da24927b MX |
3172 | ret = mnt_want_write_file(file); |
3173 | if (ret) | |
3174 | return ret; | |
c146afad | 3175 | |
dae7b665 | 3176 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
3177 | if (IS_ERR(vol_args)) { |
3178 | ret = PTR_ERR(vol_args); | |
c47ca32d | 3179 | goto err_drop; |
c9e9f97b | 3180 | } |
f46b5a66 | 3181 | |
748449cd | 3182 | if (vol_args->flags & ~BTRFS_DEVICE_REMOVE_ARGS_MASK) { |
fd4e994b OS |
3183 | ret = -EOPNOTSUPP; |
3184 | goto out; | |
3185 | } | |
f46b5a66 | 3186 | |
c3e1f96c | 3187 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REMOVE)) { |
183860f6 AJ |
3188 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
3189 | goto out; | |
3190 | } | |
3191 | ||
735654ea | 3192 | if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) { |
2ff7e61e | 3193 | ret = btrfs_rm_device(fs_info, NULL, vol_args->devid); |
6b526ed7 AJ |
3194 | } else { |
3195 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
2ff7e61e | 3196 | ret = btrfs_rm_device(fs_info, vol_args->name, 0); |
6b526ed7 | 3197 | } |
c3e1f96c | 3198 | btrfs_exclop_finish(fs_info); |
183860f6 | 3199 | |
6b526ed7 | 3200 | if (!ret) { |
735654ea | 3201 | if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) |
0b246afa | 3202 | btrfs_info(fs_info, "device deleted: id %llu", |
6b526ed7 AJ |
3203 | vol_args->devid); |
3204 | else | |
0b246afa | 3205 | btrfs_info(fs_info, "device deleted: %s", |
6b526ed7 AJ |
3206 | vol_args->name); |
3207 | } | |
183860f6 AJ |
3208 | out: |
3209 | kfree(vol_args); | |
c47ca32d | 3210 | err_drop: |
4ac20c70 | 3211 | mnt_drop_write_file(file); |
f46b5a66 CH |
3212 | return ret; |
3213 | } | |
3214 | ||
da24927b | 3215 | static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg) |
f46b5a66 | 3216 | { |
0b246afa JM |
3217 | struct inode *inode = file_inode(file); |
3218 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
f46b5a66 CH |
3219 | struct btrfs_ioctl_vol_args *vol_args; |
3220 | int ret; | |
3221 | ||
e441d54d CM |
3222 | if (!capable(CAP_SYS_ADMIN)) |
3223 | return -EPERM; | |
3224 | ||
da24927b MX |
3225 | ret = mnt_want_write_file(file); |
3226 | if (ret) | |
3227 | return ret; | |
c146afad | 3228 | |
c3e1f96c | 3229 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REMOVE)) { |
183860f6 | 3230 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
58d7bbf8 DS |
3231 | goto out_drop_write; |
3232 | } | |
3233 | ||
3234 | vol_args = memdup_user(arg, sizeof(*vol_args)); | |
3235 | if (IS_ERR(vol_args)) { | |
3236 | ret = PTR_ERR(vol_args); | |
183860f6 AJ |
3237 | goto out; |
3238 | } | |
3239 | ||
58d7bbf8 | 3240 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
2ff7e61e | 3241 | ret = btrfs_rm_device(fs_info, vol_args->name, 0); |
183860f6 | 3242 | |
ec95d491 | 3243 | if (!ret) |
0b246afa | 3244 | btrfs_info(fs_info, "disk deleted %s", vol_args->name); |
183860f6 | 3245 | kfree(vol_args); |
58d7bbf8 | 3246 | out: |
c3e1f96c | 3247 | btrfs_exclop_finish(fs_info); |
58d7bbf8 | 3248 | out_drop_write: |
4ac20c70 | 3249 | mnt_drop_write_file(file); |
58d7bbf8 | 3250 | |
f46b5a66 CH |
3251 | return ret; |
3252 | } | |
3253 | ||
2ff7e61e JM |
3254 | static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info, |
3255 | void __user *arg) | |
475f6387 | 3256 | { |
027ed2f0 | 3257 | struct btrfs_ioctl_fs_info_args *fi_args; |
475f6387 | 3258 | struct btrfs_device *device; |
0b246afa | 3259 | struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; |
137c5418 | 3260 | u64 flags_in; |
027ed2f0 | 3261 | int ret = 0; |
475f6387 | 3262 | |
137c5418 JT |
3263 | fi_args = memdup_user(arg, sizeof(*fi_args)); |
3264 | if (IS_ERR(fi_args)) | |
3265 | return PTR_ERR(fi_args); | |
3266 | ||
3267 | flags_in = fi_args->flags; | |
3268 | memset(fi_args, 0, sizeof(*fi_args)); | |
027ed2f0 | 3269 | |
d03262c7 | 3270 | rcu_read_lock(); |
027ed2f0 | 3271 | fi_args->num_devices = fs_devices->num_devices; |
475f6387 | 3272 | |
d03262c7 | 3273 | list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) { |
027ed2f0 LZ |
3274 | if (device->devid > fi_args->max_id) |
3275 | fi_args->max_id = device->devid; | |
475f6387 | 3276 | } |
d03262c7 | 3277 | rcu_read_unlock(); |
475f6387 | 3278 | |
de37aa51 | 3279 | memcpy(&fi_args->fsid, fs_devices->fsid, sizeof(fi_args->fsid)); |
bea7eafd OS |
3280 | fi_args->nodesize = fs_info->nodesize; |
3281 | fi_args->sectorsize = fs_info->sectorsize; | |
3282 | fi_args->clone_alignment = fs_info->sectorsize; | |
80a773fb | 3283 | |
137c5418 JT |
3284 | if (flags_in & BTRFS_FS_INFO_FLAG_CSUM_INFO) { |
3285 | fi_args->csum_type = btrfs_super_csum_type(fs_info->super_copy); | |
3286 | fi_args->csum_size = btrfs_super_csum_size(fs_info->super_copy); | |
3287 | fi_args->flags |= BTRFS_FS_INFO_FLAG_CSUM_INFO; | |
3288 | } | |
3289 | ||
0fb408a5 JT |
3290 | if (flags_in & BTRFS_FS_INFO_FLAG_GENERATION) { |
3291 | fi_args->generation = fs_info->generation; | |
3292 | fi_args->flags |= BTRFS_FS_INFO_FLAG_GENERATION; | |
3293 | } | |
3294 | ||
49bac897 JT |
3295 | if (flags_in & BTRFS_FS_INFO_FLAG_METADATA_UUID) { |
3296 | memcpy(&fi_args->metadata_uuid, fs_devices->metadata_uuid, | |
3297 | sizeof(fi_args->metadata_uuid)); | |
3298 | fi_args->flags |= BTRFS_FS_INFO_FLAG_METADATA_UUID; | |
3299 | } | |
3300 | ||
027ed2f0 LZ |
3301 | if (copy_to_user(arg, fi_args, sizeof(*fi_args))) |
3302 | ret = -EFAULT; | |
475f6387 | 3303 | |
027ed2f0 LZ |
3304 | kfree(fi_args); |
3305 | return ret; | |
475f6387 JS |
3306 | } |
3307 | ||
2ff7e61e JM |
3308 | static long btrfs_ioctl_dev_info(struct btrfs_fs_info *fs_info, |
3309 | void __user *arg) | |
475f6387 JS |
3310 | { |
3311 | struct btrfs_ioctl_dev_info_args *di_args; | |
3312 | struct btrfs_device *dev; | |
475f6387 JS |
3313 | int ret = 0; |
3314 | char *s_uuid = NULL; | |
475f6387 | 3315 | |
475f6387 JS |
3316 | di_args = memdup_user(arg, sizeof(*di_args)); |
3317 | if (IS_ERR(di_args)) | |
3318 | return PTR_ERR(di_args); | |
3319 | ||
dd5f9615 | 3320 | if (!btrfs_is_empty_uuid(di_args->uuid)) |
475f6387 JS |
3321 | s_uuid = di_args->uuid; |
3322 | ||
c5593ca3 | 3323 | rcu_read_lock(); |
e4319cd9 | 3324 | dev = btrfs_find_device(fs_info->fs_devices, di_args->devid, s_uuid, |
09ba3bc9 | 3325 | NULL, true); |
475f6387 JS |
3326 | |
3327 | if (!dev) { | |
3328 | ret = -ENODEV; | |
3329 | goto out; | |
3330 | } | |
3331 | ||
3332 | di_args->devid = dev->devid; | |
7cc8e58d MX |
3333 | di_args->bytes_used = btrfs_device_get_bytes_used(dev); |
3334 | di_args->total_bytes = btrfs_device_get_total_bytes(dev); | |
475f6387 | 3335 | memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid)); |
a27202fb | 3336 | if (dev->name) { |
672d5990 MT |
3337 | strncpy(di_args->path, rcu_str_deref(dev->name), |
3338 | sizeof(di_args->path) - 1); | |
a27202fb JM |
3339 | di_args->path[sizeof(di_args->path) - 1] = 0; |
3340 | } else { | |
99ba55ad | 3341 | di_args->path[0] = '\0'; |
a27202fb | 3342 | } |
475f6387 JS |
3343 | |
3344 | out: | |
c5593ca3 | 3345 | rcu_read_unlock(); |
475f6387 JS |
3346 | if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args))) |
3347 | ret = -EFAULT; | |
3348 | ||
3349 | kfree(di_args); | |
3350 | return ret; | |
3351 | } | |
3352 | ||
6ef5ed0d JB |
3353 | static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) |
3354 | { | |
496ad9aa | 3355 | struct inode *inode = file_inode(file); |
0b246afa | 3356 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
6ef5ed0d JB |
3357 | struct btrfs_root *root = BTRFS_I(inode)->root; |
3358 | struct btrfs_root *new_root; | |
3359 | struct btrfs_dir_item *di; | |
3360 | struct btrfs_trans_handle *trans; | |
2a2b5d62 | 3361 | struct btrfs_path *path = NULL; |
6ef5ed0d | 3362 | struct btrfs_disk_key disk_key; |
6ef5ed0d JB |
3363 | u64 objectid = 0; |
3364 | u64 dir_id; | |
3c04ce01 | 3365 | int ret; |
6ef5ed0d JB |
3366 | |
3367 | if (!capable(CAP_SYS_ADMIN)) | |
3368 | return -EPERM; | |
3369 | ||
3c04ce01 MX |
3370 | ret = mnt_want_write_file(file); |
3371 | if (ret) | |
3372 | return ret; | |
3373 | ||
3374 | if (copy_from_user(&objectid, argp, sizeof(objectid))) { | |
3375 | ret = -EFAULT; | |
3376 | goto out; | |
3377 | } | |
6ef5ed0d JB |
3378 | |
3379 | if (!objectid) | |
1cecf579 | 3380 | objectid = BTRFS_FS_TREE_OBJECTID; |
6ef5ed0d | 3381 | |
56e9357a | 3382 | new_root = btrfs_get_fs_root(fs_info, objectid, true); |
3c04ce01 MX |
3383 | if (IS_ERR(new_root)) { |
3384 | ret = PTR_ERR(new_root); | |
3385 | goto out; | |
3386 | } | |
2a2b5d62 JB |
3387 | if (!is_fstree(new_root->root_key.objectid)) { |
3388 | ret = -ENOENT; | |
3389 | goto out_free; | |
3390 | } | |
6ef5ed0d | 3391 | |
6ef5ed0d | 3392 | path = btrfs_alloc_path(); |
3c04ce01 MX |
3393 | if (!path) { |
3394 | ret = -ENOMEM; | |
2a2b5d62 | 3395 | goto out_free; |
3c04ce01 | 3396 | } |
6ef5ed0d JB |
3397 | path->leave_spinning = 1; |
3398 | ||
3399 | trans = btrfs_start_transaction(root, 1); | |
98d5dc13 | 3400 | if (IS_ERR(trans)) { |
3c04ce01 | 3401 | ret = PTR_ERR(trans); |
2a2b5d62 | 3402 | goto out_free; |
6ef5ed0d JB |
3403 | } |
3404 | ||
0b246afa JM |
3405 | dir_id = btrfs_super_root_dir(fs_info->super_copy); |
3406 | di = btrfs_lookup_dir_item(trans, fs_info->tree_root, path, | |
6ef5ed0d | 3407 | dir_id, "default", 7, 1); |
cf1e99a4 | 3408 | if (IS_ERR_OR_NULL(di)) { |
2a2b5d62 | 3409 | btrfs_release_path(path); |
3a45bb20 | 3410 | btrfs_end_transaction(trans); |
0b246afa | 3411 | btrfs_err(fs_info, |
5d163e0e | 3412 | "Umm, you don't have the default diritem, this isn't going to work"); |
3c04ce01 | 3413 | ret = -ENOENT; |
2a2b5d62 | 3414 | goto out_free; |
6ef5ed0d JB |
3415 | } |
3416 | ||
3417 | btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key); | |
3418 | btrfs_set_dir_item_key(path->nodes[0], di, &disk_key); | |
3419 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
2a2b5d62 | 3420 | btrfs_release_path(path); |
6ef5ed0d | 3421 | |
0b246afa | 3422 | btrfs_set_fs_incompat(fs_info, DEFAULT_SUBVOL); |
3a45bb20 | 3423 | btrfs_end_transaction(trans); |
2a2b5d62 | 3424 | out_free: |
00246528 | 3425 | btrfs_put_root(new_root); |
2a2b5d62 | 3426 | btrfs_free_path(path); |
3c04ce01 MX |
3427 | out: |
3428 | mnt_drop_write_file(file); | |
3429 | return ret; | |
6ef5ed0d JB |
3430 | } |
3431 | ||
c065f5b1 SY |
3432 | static void get_block_group_info(struct list_head *groups_list, |
3433 | struct btrfs_ioctl_space_info *space) | |
bf5fc093 | 3434 | { |
32da5386 | 3435 | struct btrfs_block_group *block_group; |
bf5fc093 JB |
3436 | |
3437 | space->total_bytes = 0; | |
3438 | space->used_bytes = 0; | |
3439 | space->flags = 0; | |
3440 | list_for_each_entry(block_group, groups_list, list) { | |
3441 | space->flags = block_group->flags; | |
b3470b5d | 3442 | space->total_bytes += block_group->length; |
bf38be65 | 3443 | space->used_bytes += block_group->used; |
bf5fc093 JB |
3444 | } |
3445 | } | |
3446 | ||
2ff7e61e JM |
3447 | static long btrfs_ioctl_space_info(struct btrfs_fs_info *fs_info, |
3448 | void __user *arg) | |
1406e432 JB |
3449 | { |
3450 | struct btrfs_ioctl_space_args space_args; | |
3451 | struct btrfs_ioctl_space_info space; | |
3452 | struct btrfs_ioctl_space_info *dest; | |
7fde62bf | 3453 | struct btrfs_ioctl_space_info *dest_orig; |
13f2696f | 3454 | struct btrfs_ioctl_space_info __user *user_dest; |
1406e432 | 3455 | struct btrfs_space_info *info; |
315d8e98 CIK |
3456 | static const u64 types[] = { |
3457 | BTRFS_BLOCK_GROUP_DATA, | |
3458 | BTRFS_BLOCK_GROUP_SYSTEM, | |
3459 | BTRFS_BLOCK_GROUP_METADATA, | |
3460 | BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA | |
3461 | }; | |
bf5fc093 | 3462 | int num_types = 4; |
7fde62bf | 3463 | int alloc_size; |
1406e432 | 3464 | int ret = 0; |
51788b1b | 3465 | u64 slot_count = 0; |
bf5fc093 | 3466 | int i, c; |
1406e432 JB |
3467 | |
3468 | if (copy_from_user(&space_args, | |
3469 | (struct btrfs_ioctl_space_args __user *)arg, | |
3470 | sizeof(space_args))) | |
3471 | return -EFAULT; | |
3472 | ||
bf5fc093 JB |
3473 | for (i = 0; i < num_types; i++) { |
3474 | struct btrfs_space_info *tmp; | |
3475 | ||
3476 | info = NULL; | |
3477 | rcu_read_lock(); | |
0b246afa | 3478 | list_for_each_entry_rcu(tmp, &fs_info->space_info, |
bf5fc093 JB |
3479 | list) { |
3480 | if (tmp->flags == types[i]) { | |
3481 | info = tmp; | |
3482 | break; | |
3483 | } | |
3484 | } | |
3485 | rcu_read_unlock(); | |
3486 | ||
3487 | if (!info) | |
3488 | continue; | |
3489 | ||
3490 | down_read(&info->groups_sem); | |
3491 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3492 | if (!list_empty(&info->block_groups[c])) | |
3493 | slot_count++; | |
3494 | } | |
3495 | up_read(&info->groups_sem); | |
3496 | } | |
7fde62bf | 3497 | |
36523e95 DS |
3498 | /* |
3499 | * Global block reserve, exported as a space_info | |
3500 | */ | |
3501 | slot_count++; | |
3502 | ||
7fde62bf CM |
3503 | /* space_slots == 0 means they are asking for a count */ |
3504 | if (space_args.space_slots == 0) { | |
3505 | space_args.total_spaces = slot_count; | |
3506 | goto out; | |
3507 | } | |
bf5fc093 | 3508 | |
51788b1b | 3509 | slot_count = min_t(u64, space_args.space_slots, slot_count); |
bf5fc093 | 3510 | |
7fde62bf | 3511 | alloc_size = sizeof(*dest) * slot_count; |
bf5fc093 | 3512 | |
7fde62bf CM |
3513 | /* we generally have at most 6 or so space infos, one for each raid |
3514 | * level. So, a whole page should be more than enough for everyone | |
3515 | */ | |
09cbfeaf | 3516 | if (alloc_size > PAGE_SIZE) |
7fde62bf CM |
3517 | return -ENOMEM; |
3518 | ||
1406e432 | 3519 | space_args.total_spaces = 0; |
8d2db785 | 3520 | dest = kmalloc(alloc_size, GFP_KERNEL); |
7fde62bf CM |
3521 | if (!dest) |
3522 | return -ENOMEM; | |
3523 | dest_orig = dest; | |
1406e432 | 3524 | |
7fde62bf | 3525 | /* now we have a buffer to copy into */ |
bf5fc093 JB |
3526 | for (i = 0; i < num_types; i++) { |
3527 | struct btrfs_space_info *tmp; | |
3528 | ||
51788b1b DR |
3529 | if (!slot_count) |
3530 | break; | |
3531 | ||
bf5fc093 JB |
3532 | info = NULL; |
3533 | rcu_read_lock(); | |
0b246afa | 3534 | list_for_each_entry_rcu(tmp, &fs_info->space_info, |
bf5fc093 JB |
3535 | list) { |
3536 | if (tmp->flags == types[i]) { | |
3537 | info = tmp; | |
3538 | break; | |
3539 | } | |
3540 | } | |
3541 | rcu_read_unlock(); | |
7fde62bf | 3542 | |
bf5fc093 JB |
3543 | if (!info) |
3544 | continue; | |
3545 | down_read(&info->groups_sem); | |
3546 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3547 | if (!list_empty(&info->block_groups[c])) { | |
c065f5b1 SY |
3548 | get_block_group_info(&info->block_groups[c], |
3549 | &space); | |
bf5fc093 JB |
3550 | memcpy(dest, &space, sizeof(space)); |
3551 | dest++; | |
3552 | space_args.total_spaces++; | |
51788b1b | 3553 | slot_count--; |
bf5fc093 | 3554 | } |
51788b1b DR |
3555 | if (!slot_count) |
3556 | break; | |
bf5fc093 JB |
3557 | } |
3558 | up_read(&info->groups_sem); | |
1406e432 | 3559 | } |
1406e432 | 3560 | |
36523e95 DS |
3561 | /* |
3562 | * Add global block reserve | |
3563 | */ | |
3564 | if (slot_count) { | |
0b246afa | 3565 | struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv; |
36523e95 DS |
3566 | |
3567 | spin_lock(&block_rsv->lock); | |
3568 | space.total_bytes = block_rsv->size; | |
3569 | space.used_bytes = block_rsv->size - block_rsv->reserved; | |
3570 | spin_unlock(&block_rsv->lock); | |
3571 | space.flags = BTRFS_SPACE_INFO_GLOBAL_RSV; | |
3572 | memcpy(dest, &space, sizeof(space)); | |
3573 | space_args.total_spaces++; | |
3574 | } | |
3575 | ||
2eec6c81 | 3576 | user_dest = (struct btrfs_ioctl_space_info __user *) |
7fde62bf CM |
3577 | (arg + sizeof(struct btrfs_ioctl_space_args)); |
3578 | ||
3579 | if (copy_to_user(user_dest, dest_orig, alloc_size)) | |
3580 | ret = -EFAULT; | |
3581 | ||
3582 | kfree(dest_orig); | |
3583 | out: | |
3584 | if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args))) | |
1406e432 JB |
3585 | ret = -EFAULT; |
3586 | ||
3587 | return ret; | |
3588 | } | |
3589 | ||
9a8c28be MX |
3590 | static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root, |
3591 | void __user *argp) | |
46204592 | 3592 | { |
46204592 SW |
3593 | struct btrfs_trans_handle *trans; |
3594 | u64 transid; | |
db5b493a | 3595 | int ret; |
46204592 | 3596 | |
d4edf39b | 3597 | trans = btrfs_attach_transaction_barrier(root); |
ff7c1d33 MX |
3598 | if (IS_ERR(trans)) { |
3599 | if (PTR_ERR(trans) != -ENOENT) | |
3600 | return PTR_ERR(trans); | |
3601 | ||
3602 | /* No running transaction, don't bother */ | |
3603 | transid = root->fs_info->last_trans_committed; | |
3604 | goto out; | |
3605 | } | |
46204592 | 3606 | transid = trans->transid; |
3a45bb20 | 3607 | ret = btrfs_commit_transaction_async(trans, 0); |
8b2b2d3c | 3608 | if (ret) { |
3a45bb20 | 3609 | btrfs_end_transaction(trans); |
db5b493a | 3610 | return ret; |
8b2b2d3c | 3611 | } |
ff7c1d33 | 3612 | out: |
46204592 SW |
3613 | if (argp) |
3614 | if (copy_to_user(argp, &transid, sizeof(transid))) | |
3615 | return -EFAULT; | |
3616 | return 0; | |
3617 | } | |
3618 | ||
2ff7e61e | 3619 | static noinline long btrfs_ioctl_wait_sync(struct btrfs_fs_info *fs_info, |
9a8c28be | 3620 | void __user *argp) |
46204592 | 3621 | { |
46204592 SW |
3622 | u64 transid; |
3623 | ||
3624 | if (argp) { | |
3625 | if (copy_from_user(&transid, argp, sizeof(transid))) | |
3626 | return -EFAULT; | |
3627 | } else { | |
3628 | transid = 0; /* current trans */ | |
3629 | } | |
2ff7e61e | 3630 | return btrfs_wait_for_commit(fs_info, transid); |
46204592 SW |
3631 | } |
3632 | ||
b8e95489 | 3633 | static long btrfs_ioctl_scrub(struct file *file, void __user *arg) |
475f6387 | 3634 | { |
0b246afa | 3635 | struct btrfs_fs_info *fs_info = btrfs_sb(file_inode(file)->i_sb); |
475f6387 | 3636 | struct btrfs_ioctl_scrub_args *sa; |
b8e95489 | 3637 | int ret; |
475f6387 JS |
3638 | |
3639 | if (!capable(CAP_SYS_ADMIN)) | |
3640 | return -EPERM; | |
3641 | ||
3642 | sa = memdup_user(arg, sizeof(*sa)); | |
3643 | if (IS_ERR(sa)) | |
3644 | return PTR_ERR(sa); | |
3645 | ||
b8e95489 MX |
3646 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) { |
3647 | ret = mnt_want_write_file(file); | |
3648 | if (ret) | |
3649 | goto out; | |
3650 | } | |
3651 | ||
0b246afa | 3652 | ret = btrfs_scrub_dev(fs_info, sa->devid, sa->start, sa->end, |
63a212ab SB |
3653 | &sa->progress, sa->flags & BTRFS_SCRUB_READONLY, |
3654 | 0); | |
475f6387 | 3655 | |
5afe6ce7 FM |
3656 | /* |
3657 | * Copy scrub args to user space even if btrfs_scrub_dev() returned an | |
3658 | * error. This is important as it allows user space to know how much | |
3659 | * progress scrub has done. For example, if scrub is canceled we get | |
3660 | * -ECANCELED from btrfs_scrub_dev() and return that error back to user | |
3661 | * space. Later user space can inspect the progress from the structure | |
3662 | * btrfs_ioctl_scrub_args and resume scrub from where it left off | |
3663 | * previously (btrfs-progs does this). | |
3664 | * If we fail to copy the btrfs_ioctl_scrub_args structure to user space | |
3665 | * then return -EFAULT to signal the structure was not copied or it may | |
3666 | * be corrupt and unreliable due to a partial copy. | |
3667 | */ | |
3668 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
475f6387 JS |
3669 | ret = -EFAULT; |
3670 | ||
b8e95489 MX |
3671 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) |
3672 | mnt_drop_write_file(file); | |
3673 | out: | |
475f6387 JS |
3674 | kfree(sa); |
3675 | return ret; | |
3676 | } | |
3677 | ||
2ff7e61e | 3678 | static long btrfs_ioctl_scrub_cancel(struct btrfs_fs_info *fs_info) |
475f6387 JS |
3679 | { |
3680 | if (!capable(CAP_SYS_ADMIN)) | |
3681 | return -EPERM; | |
3682 | ||
2ff7e61e | 3683 | return btrfs_scrub_cancel(fs_info); |
475f6387 JS |
3684 | } |
3685 | ||
2ff7e61e | 3686 | static long btrfs_ioctl_scrub_progress(struct btrfs_fs_info *fs_info, |
475f6387 JS |
3687 | void __user *arg) |
3688 | { | |
3689 | struct btrfs_ioctl_scrub_args *sa; | |
3690 | int ret; | |
3691 | ||
3692 | if (!capable(CAP_SYS_ADMIN)) | |
3693 | return -EPERM; | |
3694 | ||
3695 | sa = memdup_user(arg, sizeof(*sa)); | |
3696 | if (IS_ERR(sa)) | |
3697 | return PTR_ERR(sa); | |
3698 | ||
2ff7e61e | 3699 | ret = btrfs_scrub_progress(fs_info, sa->devid, &sa->progress); |
475f6387 | 3700 | |
4fa99b00 | 3701 | if (ret == 0 && copy_to_user(arg, sa, sizeof(*sa))) |
475f6387 JS |
3702 | ret = -EFAULT; |
3703 | ||
3704 | kfree(sa); | |
3705 | return ret; | |
3706 | } | |
3707 | ||
2ff7e61e | 3708 | static long btrfs_ioctl_get_dev_stats(struct btrfs_fs_info *fs_info, |
b27f7c0c | 3709 | void __user *arg) |
c11d2c23 SB |
3710 | { |
3711 | struct btrfs_ioctl_get_dev_stats *sa; | |
3712 | int ret; | |
3713 | ||
c11d2c23 SB |
3714 | sa = memdup_user(arg, sizeof(*sa)); |
3715 | if (IS_ERR(sa)) | |
3716 | return PTR_ERR(sa); | |
3717 | ||
b27f7c0c DS |
3718 | if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) { |
3719 | kfree(sa); | |
3720 | return -EPERM; | |
3721 | } | |
3722 | ||
2ff7e61e | 3723 | ret = btrfs_get_dev_stats(fs_info, sa); |
c11d2c23 | 3724 | |
eee99577 | 3725 | if (ret == 0 && copy_to_user(arg, sa, sizeof(*sa))) |
c11d2c23 SB |
3726 | ret = -EFAULT; |
3727 | ||
3728 | kfree(sa); | |
3729 | return ret; | |
3730 | } | |
3731 | ||
2ff7e61e JM |
3732 | static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info, |
3733 | void __user *arg) | |
3f6bcfbd SB |
3734 | { |
3735 | struct btrfs_ioctl_dev_replace_args *p; | |
3736 | int ret; | |
3737 | ||
3738 | if (!capable(CAP_SYS_ADMIN)) | |
3739 | return -EPERM; | |
3740 | ||
3741 | p = memdup_user(arg, sizeof(*p)); | |
3742 | if (IS_ERR(p)) | |
3743 | return PTR_ERR(p); | |
3744 | ||
3745 | switch (p->cmd) { | |
3746 | case BTRFS_IOCTL_DEV_REPLACE_CMD_START: | |
bc98a42c | 3747 | if (sb_rdonly(fs_info->sb)) { |
adfa97cb ID |
3748 | ret = -EROFS; |
3749 | goto out; | |
3750 | } | |
c3e1f96c | 3751 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REPLACE)) { |
e57138b3 | 3752 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
3f6bcfbd | 3753 | } else { |
2ff7e61e | 3754 | ret = btrfs_dev_replace_by_ioctl(fs_info, p); |
c3e1f96c | 3755 | btrfs_exclop_finish(fs_info); |
3f6bcfbd SB |
3756 | } |
3757 | break; | |
3758 | case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS: | |
0b246afa | 3759 | btrfs_dev_replace_status(fs_info, p); |
3f6bcfbd SB |
3760 | ret = 0; |
3761 | break; | |
3762 | case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL: | |
17d202b9 | 3763 | p->result = btrfs_dev_replace_cancel(fs_info); |
97282031 | 3764 | ret = 0; |
3f6bcfbd SB |
3765 | break; |
3766 | default: | |
3767 | ret = -EINVAL; | |
3768 | break; | |
3769 | } | |
3770 | ||
d3a53286 | 3771 | if ((ret == 0 || ret == -ECANCELED) && copy_to_user(arg, p, sizeof(*p))) |
3f6bcfbd | 3772 | ret = -EFAULT; |
adfa97cb | 3773 | out: |
3f6bcfbd SB |
3774 | kfree(p); |
3775 | return ret; | |
3776 | } | |
3777 | ||
d7728c96 JS |
3778 | static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg) |
3779 | { | |
3780 | int ret = 0; | |
3781 | int i; | |
740c3d22 | 3782 | u64 rel_ptr; |
d7728c96 | 3783 | int size; |
806468f8 | 3784 | struct btrfs_ioctl_ino_path_args *ipa = NULL; |
d7728c96 JS |
3785 | struct inode_fs_paths *ipath = NULL; |
3786 | struct btrfs_path *path; | |
3787 | ||
82b22ac8 | 3788 | if (!capable(CAP_DAC_READ_SEARCH)) |
d7728c96 JS |
3789 | return -EPERM; |
3790 | ||
3791 | path = btrfs_alloc_path(); | |
3792 | if (!path) { | |
3793 | ret = -ENOMEM; | |
3794 | goto out; | |
3795 | } | |
3796 | ||
3797 | ipa = memdup_user(arg, sizeof(*ipa)); | |
3798 | if (IS_ERR(ipa)) { | |
3799 | ret = PTR_ERR(ipa); | |
3800 | ipa = NULL; | |
3801 | goto out; | |
3802 | } | |
3803 | ||
3804 | size = min_t(u32, ipa->size, 4096); | |
3805 | ipath = init_ipath(size, root, path); | |
3806 | if (IS_ERR(ipath)) { | |
3807 | ret = PTR_ERR(ipath); | |
3808 | ipath = NULL; | |
3809 | goto out; | |
3810 | } | |
3811 | ||
3812 | ret = paths_from_inode(ipa->inum, ipath); | |
3813 | if (ret < 0) | |
3814 | goto out; | |
3815 | ||
3816 | for (i = 0; i < ipath->fspath->elem_cnt; ++i) { | |
745c4d8e JM |
3817 | rel_ptr = ipath->fspath->val[i] - |
3818 | (u64)(unsigned long)ipath->fspath->val; | |
740c3d22 | 3819 | ipath->fspath->val[i] = rel_ptr; |
d7728c96 JS |
3820 | } |
3821 | ||
718dc5fa OS |
3822 | ret = copy_to_user((void __user *)(unsigned long)ipa->fspath, |
3823 | ipath->fspath, size); | |
d7728c96 JS |
3824 | if (ret) { |
3825 | ret = -EFAULT; | |
3826 | goto out; | |
3827 | } | |
3828 | ||
3829 | out: | |
3830 | btrfs_free_path(path); | |
3831 | free_ipath(ipath); | |
3832 | kfree(ipa); | |
3833 | ||
3834 | return ret; | |
3835 | } | |
3836 | ||
3837 | static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx) | |
3838 | { | |
3839 | struct btrfs_data_container *inodes = ctx; | |
3840 | const size_t c = 3 * sizeof(u64); | |
3841 | ||
3842 | if (inodes->bytes_left >= c) { | |
3843 | inodes->bytes_left -= c; | |
3844 | inodes->val[inodes->elem_cnt] = inum; | |
3845 | inodes->val[inodes->elem_cnt + 1] = offset; | |
3846 | inodes->val[inodes->elem_cnt + 2] = root; | |
3847 | inodes->elem_cnt += 3; | |
3848 | } else { | |
3849 | inodes->bytes_missing += c - inodes->bytes_left; | |
3850 | inodes->bytes_left = 0; | |
3851 | inodes->elem_missed += 3; | |
3852 | } | |
3853 | ||
3854 | return 0; | |
3855 | } | |
3856 | ||
2ff7e61e | 3857 | static long btrfs_ioctl_logical_to_ino(struct btrfs_fs_info *fs_info, |
d24a67b2 | 3858 | void __user *arg, int version) |
d7728c96 JS |
3859 | { |
3860 | int ret = 0; | |
3861 | int size; | |
d7728c96 JS |
3862 | struct btrfs_ioctl_logical_ino_args *loi; |
3863 | struct btrfs_data_container *inodes = NULL; | |
3864 | struct btrfs_path *path = NULL; | |
d24a67b2 | 3865 | bool ignore_offset; |
d7728c96 JS |
3866 | |
3867 | if (!capable(CAP_SYS_ADMIN)) | |
3868 | return -EPERM; | |
3869 | ||
3870 | loi = memdup_user(arg, sizeof(*loi)); | |
7b9ea627 SV |
3871 | if (IS_ERR(loi)) |
3872 | return PTR_ERR(loi); | |
d7728c96 | 3873 | |
d24a67b2 ZB |
3874 | if (version == 1) { |
3875 | ignore_offset = false; | |
b115e3bc | 3876 | size = min_t(u32, loi->size, SZ_64K); |
d24a67b2 ZB |
3877 | } else { |
3878 | /* All reserved bits must be 0 for now */ | |
3879 | if (memchr_inv(loi->reserved, 0, sizeof(loi->reserved))) { | |
3880 | ret = -EINVAL; | |
3881 | goto out_loi; | |
3882 | } | |
3883 | /* Only accept flags we have defined so far */ | |
3884 | if (loi->flags & ~(BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET)) { | |
3885 | ret = -EINVAL; | |
3886 | goto out_loi; | |
3887 | } | |
3888 | ignore_offset = loi->flags & BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET; | |
b115e3bc | 3889 | size = min_t(u32, loi->size, SZ_16M); |
d24a67b2 ZB |
3890 | } |
3891 | ||
d7728c96 JS |
3892 | path = btrfs_alloc_path(); |
3893 | if (!path) { | |
3894 | ret = -ENOMEM; | |
3895 | goto out; | |
3896 | } | |
3897 | ||
d7728c96 JS |
3898 | inodes = init_data_container(size); |
3899 | if (IS_ERR(inodes)) { | |
3900 | ret = PTR_ERR(inodes); | |
3901 | inodes = NULL; | |
3902 | goto out; | |
3903 | } | |
3904 | ||
2ff7e61e | 3905 | ret = iterate_inodes_from_logical(loi->logical, fs_info, path, |
d24a67b2 | 3906 | build_ino_list, inodes, ignore_offset); |
df031f07 | 3907 | if (ret == -EINVAL) |
d7728c96 JS |
3908 | ret = -ENOENT; |
3909 | if (ret < 0) | |
3910 | goto out; | |
3911 | ||
718dc5fa OS |
3912 | ret = copy_to_user((void __user *)(unsigned long)loi->inodes, inodes, |
3913 | size); | |
d7728c96 JS |
3914 | if (ret) |
3915 | ret = -EFAULT; | |
3916 | ||
3917 | out: | |
3918 | btrfs_free_path(path); | |
f54de068 | 3919 | kvfree(inodes); |
d24a67b2 | 3920 | out_loi: |
d7728c96 JS |
3921 | kfree(loi); |
3922 | ||
3923 | return ret; | |
3924 | } | |
3925 | ||
008ef096 | 3926 | void btrfs_update_ioctl_balance_args(struct btrfs_fs_info *fs_info, |
c9e9f97b ID |
3927 | struct btrfs_ioctl_balance_args *bargs) |
3928 | { | |
3929 | struct btrfs_balance_control *bctl = fs_info->balance_ctl; | |
3930 | ||
3931 | bargs->flags = bctl->flags; | |
3932 | ||
3009a62f | 3933 | if (test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) |
837d5b6e ID |
3934 | bargs->state |= BTRFS_BALANCE_STATE_RUNNING; |
3935 | if (atomic_read(&fs_info->balance_pause_req)) | |
3936 | bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ; | |
a7e99c69 ID |
3937 | if (atomic_read(&fs_info->balance_cancel_req)) |
3938 | bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ; | |
837d5b6e | 3939 | |
c9e9f97b ID |
3940 | memcpy(&bargs->data, &bctl->data, sizeof(bargs->data)); |
3941 | memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta)); | |
3942 | memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys)); | |
19a39dce | 3943 | |
008ef096 DS |
3944 | spin_lock(&fs_info->balance_lock); |
3945 | memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat)); | |
3946 | spin_unlock(&fs_info->balance_lock); | |
c9e9f97b ID |
3947 | } |
3948 | ||
9ba1f6e4 | 3949 | static long btrfs_ioctl_balance(struct file *file, void __user *arg) |
c9e9f97b | 3950 | { |
496ad9aa | 3951 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
c9e9f97b ID |
3952 | struct btrfs_fs_info *fs_info = root->fs_info; |
3953 | struct btrfs_ioctl_balance_args *bargs; | |
3954 | struct btrfs_balance_control *bctl; | |
ed0fb78f | 3955 | bool need_unlock; /* for mut. excl. ops lock */ |
c9e9f97b ID |
3956 | int ret; |
3957 | ||
3958 | if (!capable(CAP_SYS_ADMIN)) | |
3959 | return -EPERM; | |
3960 | ||
e54bfa31 | 3961 | ret = mnt_want_write_file(file); |
9ba1f6e4 LB |
3962 | if (ret) |
3963 | return ret; | |
3964 | ||
ed0fb78f | 3965 | again: |
c3e1f96c | 3966 | if (btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) { |
ed0fb78f ID |
3967 | mutex_lock(&fs_info->balance_mutex); |
3968 | need_unlock = true; | |
3969 | goto locked; | |
3970 | } | |
3971 | ||
3972 | /* | |
01327610 | 3973 | * mut. excl. ops lock is locked. Three possibilities: |
ed0fb78f ID |
3974 | * (1) some other op is running |
3975 | * (2) balance is running | |
3976 | * (3) balance is paused -- special case (think resume) | |
3977 | */ | |
c9e9f97b | 3978 | mutex_lock(&fs_info->balance_mutex); |
ed0fb78f ID |
3979 | if (fs_info->balance_ctl) { |
3980 | /* this is either (2) or (3) */ | |
3009a62f | 3981 | if (!test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) { |
ed0fb78f | 3982 | mutex_unlock(&fs_info->balance_mutex); |
dccdb07b DS |
3983 | /* |
3984 | * Lock released to allow other waiters to continue, | |
3985 | * we'll reexamine the status again. | |
3986 | */ | |
ed0fb78f ID |
3987 | mutex_lock(&fs_info->balance_mutex); |
3988 | ||
3989 | if (fs_info->balance_ctl && | |
3009a62f | 3990 | !test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) { |
ed0fb78f ID |
3991 | /* this is (3) */ |
3992 | need_unlock = false; | |
3993 | goto locked; | |
3994 | } | |
3995 | ||
3996 | mutex_unlock(&fs_info->balance_mutex); | |
ed0fb78f ID |
3997 | goto again; |
3998 | } else { | |
3999 | /* this is (2) */ | |
4000 | mutex_unlock(&fs_info->balance_mutex); | |
4001 | ret = -EINPROGRESS; | |
4002 | goto out; | |
4003 | } | |
4004 | } else { | |
4005 | /* this is (1) */ | |
4006 | mutex_unlock(&fs_info->balance_mutex); | |
e57138b3 | 4007 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
ed0fb78f ID |
4008 | goto out; |
4009 | } | |
4010 | ||
4011 | locked: | |
c9e9f97b ID |
4012 | |
4013 | if (arg) { | |
4014 | bargs = memdup_user(arg, sizeof(*bargs)); | |
4015 | if (IS_ERR(bargs)) { | |
4016 | ret = PTR_ERR(bargs); | |
ed0fb78f | 4017 | goto out_unlock; |
c9e9f97b | 4018 | } |
de322263 ID |
4019 | |
4020 | if (bargs->flags & BTRFS_BALANCE_RESUME) { | |
4021 | if (!fs_info->balance_ctl) { | |
4022 | ret = -ENOTCONN; | |
4023 | goto out_bargs; | |
4024 | } | |
4025 | ||
4026 | bctl = fs_info->balance_ctl; | |
4027 | spin_lock(&fs_info->balance_lock); | |
4028 | bctl->flags |= BTRFS_BALANCE_RESUME; | |
4029 | spin_unlock(&fs_info->balance_lock); | |
4030 | ||
4031 | goto do_balance; | |
4032 | } | |
c9e9f97b ID |
4033 | } else { |
4034 | bargs = NULL; | |
4035 | } | |
4036 | ||
ed0fb78f | 4037 | if (fs_info->balance_ctl) { |
837d5b6e ID |
4038 | ret = -EINPROGRESS; |
4039 | goto out_bargs; | |
4040 | } | |
4041 | ||
8d2db785 | 4042 | bctl = kzalloc(sizeof(*bctl), GFP_KERNEL); |
c9e9f97b ID |
4043 | if (!bctl) { |
4044 | ret = -ENOMEM; | |
4045 | goto out_bargs; | |
4046 | } | |
4047 | ||
c9e9f97b ID |
4048 | if (arg) { |
4049 | memcpy(&bctl->data, &bargs->data, sizeof(bctl->data)); | |
4050 | memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta)); | |
4051 | memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys)); | |
4052 | ||
4053 | bctl->flags = bargs->flags; | |
f43ffb60 ID |
4054 | } else { |
4055 | /* balance everything - no filters */ | |
4056 | bctl->flags |= BTRFS_BALANCE_TYPE_MASK; | |
c9e9f97b ID |
4057 | } |
4058 | ||
8eb93459 DS |
4059 | if (bctl->flags & ~(BTRFS_BALANCE_ARGS_MASK | BTRFS_BALANCE_TYPE_MASK)) { |
4060 | ret = -EINVAL; | |
0f89abf5 | 4061 | goto out_bctl; |
8eb93459 DS |
4062 | } |
4063 | ||
de322263 | 4064 | do_balance: |
c9e9f97b | 4065 | /* |
c3e1f96c GR |
4066 | * Ownership of bctl and exclusive operation goes to btrfs_balance. |
4067 | * bctl is freed in reset_balance_state, or, if restriper was paused | |
4068 | * all the way until unmount, in free_fs_info. The flag should be | |
4069 | * cleared after reset_balance_state. | |
c9e9f97b | 4070 | */ |
ed0fb78f ID |
4071 | need_unlock = false; |
4072 | ||
6fcf6e2b | 4073 | ret = btrfs_balance(fs_info, bctl, bargs); |
0f89abf5 | 4074 | bctl = NULL; |
ed0fb78f | 4075 | |
d00c2d9c | 4076 | if ((ret == 0 || ret == -ECANCELED) && arg) { |
c9e9f97b ID |
4077 | if (copy_to_user(arg, bargs, sizeof(*bargs))) |
4078 | ret = -EFAULT; | |
4079 | } | |
4080 | ||
0f89abf5 CE |
4081 | out_bctl: |
4082 | kfree(bctl); | |
c9e9f97b ID |
4083 | out_bargs: |
4084 | kfree(bargs); | |
ed0fb78f | 4085 | out_unlock: |
c9e9f97b | 4086 | mutex_unlock(&fs_info->balance_mutex); |
ed0fb78f | 4087 | if (need_unlock) |
c3e1f96c | 4088 | btrfs_exclop_finish(fs_info); |
ed0fb78f | 4089 | out: |
e54bfa31 | 4090 | mnt_drop_write_file(file); |
c9e9f97b ID |
4091 | return ret; |
4092 | } | |
4093 | ||
2ff7e61e | 4094 | static long btrfs_ioctl_balance_ctl(struct btrfs_fs_info *fs_info, int cmd) |
837d5b6e ID |
4095 | { |
4096 | if (!capable(CAP_SYS_ADMIN)) | |
4097 | return -EPERM; | |
4098 | ||
4099 | switch (cmd) { | |
4100 | case BTRFS_BALANCE_CTL_PAUSE: | |
0b246afa | 4101 | return btrfs_pause_balance(fs_info); |
a7e99c69 | 4102 | case BTRFS_BALANCE_CTL_CANCEL: |
0b246afa | 4103 | return btrfs_cancel_balance(fs_info); |
837d5b6e ID |
4104 | } |
4105 | ||
4106 | return -EINVAL; | |
4107 | } | |
4108 | ||
2ff7e61e | 4109 | static long btrfs_ioctl_balance_progress(struct btrfs_fs_info *fs_info, |
19a39dce ID |
4110 | void __user *arg) |
4111 | { | |
19a39dce ID |
4112 | struct btrfs_ioctl_balance_args *bargs; |
4113 | int ret = 0; | |
4114 | ||
4115 | if (!capable(CAP_SYS_ADMIN)) | |
4116 | return -EPERM; | |
4117 | ||
4118 | mutex_lock(&fs_info->balance_mutex); | |
4119 | if (!fs_info->balance_ctl) { | |
4120 | ret = -ENOTCONN; | |
4121 | goto out; | |
4122 | } | |
4123 | ||
8d2db785 | 4124 | bargs = kzalloc(sizeof(*bargs), GFP_KERNEL); |
19a39dce ID |
4125 | if (!bargs) { |
4126 | ret = -ENOMEM; | |
4127 | goto out; | |
4128 | } | |
4129 | ||
008ef096 | 4130 | btrfs_update_ioctl_balance_args(fs_info, bargs); |
19a39dce ID |
4131 | |
4132 | if (copy_to_user(arg, bargs, sizeof(*bargs))) | |
4133 | ret = -EFAULT; | |
4134 | ||
4135 | kfree(bargs); | |
4136 | out: | |
4137 | mutex_unlock(&fs_info->balance_mutex); | |
4138 | return ret; | |
4139 | } | |
4140 | ||
905b0dda | 4141 | static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg) |
5d13a37b | 4142 | { |
0b246afa JM |
4143 | struct inode *inode = file_inode(file); |
4144 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
5d13a37b | 4145 | struct btrfs_ioctl_quota_ctl_args *sa; |
5d13a37b | 4146 | int ret; |
5d13a37b AJ |
4147 | |
4148 | if (!capable(CAP_SYS_ADMIN)) | |
4149 | return -EPERM; | |
4150 | ||
905b0dda MX |
4151 | ret = mnt_want_write_file(file); |
4152 | if (ret) | |
4153 | return ret; | |
5d13a37b AJ |
4154 | |
4155 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4156 | if (IS_ERR(sa)) { |
4157 | ret = PTR_ERR(sa); | |
4158 | goto drop_write; | |
4159 | } | |
5d13a37b | 4160 | |
0b246afa | 4161 | down_write(&fs_info->subvol_sem); |
5d13a37b AJ |
4162 | |
4163 | switch (sa->cmd) { | |
4164 | case BTRFS_QUOTA_CTL_ENABLE: | |
340f1aa2 | 4165 | ret = btrfs_quota_enable(fs_info); |
5d13a37b AJ |
4166 | break; |
4167 | case BTRFS_QUOTA_CTL_DISABLE: | |
340f1aa2 | 4168 | ret = btrfs_quota_disable(fs_info); |
5d13a37b | 4169 | break; |
5d13a37b AJ |
4170 | default: |
4171 | ret = -EINVAL; | |
4172 | break; | |
4173 | } | |
4174 | ||
5d13a37b | 4175 | kfree(sa); |
0b246afa | 4176 | up_write(&fs_info->subvol_sem); |
905b0dda MX |
4177 | drop_write: |
4178 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4179 | return ret; |
4180 | } | |
4181 | ||
905b0dda | 4182 | static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg) |
5d13a37b | 4183 | { |
0b246afa JM |
4184 | struct inode *inode = file_inode(file); |
4185 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4186 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
5d13a37b AJ |
4187 | struct btrfs_ioctl_qgroup_assign_args *sa; |
4188 | struct btrfs_trans_handle *trans; | |
4189 | int ret; | |
4190 | int err; | |
4191 | ||
4192 | if (!capable(CAP_SYS_ADMIN)) | |
4193 | return -EPERM; | |
4194 | ||
905b0dda MX |
4195 | ret = mnt_want_write_file(file); |
4196 | if (ret) | |
4197 | return ret; | |
5d13a37b AJ |
4198 | |
4199 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4200 | if (IS_ERR(sa)) { |
4201 | ret = PTR_ERR(sa); | |
4202 | goto drop_write; | |
4203 | } | |
5d13a37b AJ |
4204 | |
4205 | trans = btrfs_join_transaction(root); | |
4206 | if (IS_ERR(trans)) { | |
4207 | ret = PTR_ERR(trans); | |
4208 | goto out; | |
4209 | } | |
4210 | ||
5d13a37b | 4211 | if (sa->assign) { |
9f8a6ce6 | 4212 | ret = btrfs_add_qgroup_relation(trans, sa->src, sa->dst); |
5d13a37b | 4213 | } else { |
39616c27 | 4214 | ret = btrfs_del_qgroup_relation(trans, sa->src, sa->dst); |
5d13a37b AJ |
4215 | } |
4216 | ||
e082f563 | 4217 | /* update qgroup status and info */ |
280f8bd2 | 4218 | err = btrfs_run_qgroups(trans); |
e082f563 | 4219 | if (err < 0) |
0b246afa JM |
4220 | btrfs_handle_fs_error(fs_info, err, |
4221 | "failed to update qgroup status and info"); | |
3a45bb20 | 4222 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4223 | if (err && !ret) |
4224 | ret = err; | |
4225 | ||
4226 | out: | |
4227 | kfree(sa); | |
905b0dda MX |
4228 | drop_write: |
4229 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4230 | return ret; |
4231 | } | |
4232 | ||
905b0dda | 4233 | static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg) |
5d13a37b | 4234 | { |
0b246afa | 4235 | struct inode *inode = file_inode(file); |
0b246afa | 4236 | struct btrfs_root *root = BTRFS_I(inode)->root; |
5d13a37b AJ |
4237 | struct btrfs_ioctl_qgroup_create_args *sa; |
4238 | struct btrfs_trans_handle *trans; | |
4239 | int ret; | |
4240 | int err; | |
4241 | ||
4242 | if (!capable(CAP_SYS_ADMIN)) | |
4243 | return -EPERM; | |
4244 | ||
905b0dda MX |
4245 | ret = mnt_want_write_file(file); |
4246 | if (ret) | |
4247 | return ret; | |
5d13a37b AJ |
4248 | |
4249 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4250 | if (IS_ERR(sa)) { |
4251 | ret = PTR_ERR(sa); | |
4252 | goto drop_write; | |
4253 | } | |
5d13a37b | 4254 | |
d86e56cf MX |
4255 | if (!sa->qgroupid) { |
4256 | ret = -EINVAL; | |
4257 | goto out; | |
4258 | } | |
4259 | ||
5d13a37b AJ |
4260 | trans = btrfs_join_transaction(root); |
4261 | if (IS_ERR(trans)) { | |
4262 | ret = PTR_ERR(trans); | |
4263 | goto out; | |
4264 | } | |
4265 | ||
5d13a37b | 4266 | if (sa->create) { |
49a05ecd | 4267 | ret = btrfs_create_qgroup(trans, sa->qgroupid); |
5d13a37b | 4268 | } else { |
3efbee1d | 4269 | ret = btrfs_remove_qgroup(trans, sa->qgroupid); |
5d13a37b AJ |
4270 | } |
4271 | ||
3a45bb20 | 4272 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4273 | if (err && !ret) |
4274 | ret = err; | |
4275 | ||
4276 | out: | |
4277 | kfree(sa); | |
905b0dda MX |
4278 | drop_write: |
4279 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4280 | return ret; |
4281 | } | |
4282 | ||
905b0dda | 4283 | static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg) |
5d13a37b | 4284 | { |
0b246afa | 4285 | struct inode *inode = file_inode(file); |
0b246afa | 4286 | struct btrfs_root *root = BTRFS_I(inode)->root; |
5d13a37b AJ |
4287 | struct btrfs_ioctl_qgroup_limit_args *sa; |
4288 | struct btrfs_trans_handle *trans; | |
4289 | int ret; | |
4290 | int err; | |
4291 | u64 qgroupid; | |
4292 | ||
4293 | if (!capable(CAP_SYS_ADMIN)) | |
4294 | return -EPERM; | |
4295 | ||
905b0dda MX |
4296 | ret = mnt_want_write_file(file); |
4297 | if (ret) | |
4298 | return ret; | |
5d13a37b AJ |
4299 | |
4300 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4301 | if (IS_ERR(sa)) { |
4302 | ret = PTR_ERR(sa); | |
4303 | goto drop_write; | |
4304 | } | |
5d13a37b AJ |
4305 | |
4306 | trans = btrfs_join_transaction(root); | |
4307 | if (IS_ERR(trans)) { | |
4308 | ret = PTR_ERR(trans); | |
4309 | goto out; | |
4310 | } | |
4311 | ||
4312 | qgroupid = sa->qgroupid; | |
4313 | if (!qgroupid) { | |
4314 | /* take the current subvol as qgroup */ | |
4315 | qgroupid = root->root_key.objectid; | |
4316 | } | |
4317 | ||
f0042d5e | 4318 | ret = btrfs_limit_qgroup(trans, qgroupid, &sa->lim); |
5d13a37b | 4319 | |
3a45bb20 | 4320 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4321 | if (err && !ret) |
4322 | ret = err; | |
4323 | ||
4324 | out: | |
4325 | kfree(sa); | |
905b0dda MX |
4326 | drop_write: |
4327 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4328 | return ret; |
4329 | } | |
4330 | ||
2f232036 JS |
4331 | static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg) |
4332 | { | |
0b246afa JM |
4333 | struct inode *inode = file_inode(file); |
4334 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
2f232036 JS |
4335 | struct btrfs_ioctl_quota_rescan_args *qsa; |
4336 | int ret; | |
4337 | ||
4338 | if (!capable(CAP_SYS_ADMIN)) | |
4339 | return -EPERM; | |
4340 | ||
4341 | ret = mnt_want_write_file(file); | |
4342 | if (ret) | |
4343 | return ret; | |
4344 | ||
4345 | qsa = memdup_user(arg, sizeof(*qsa)); | |
4346 | if (IS_ERR(qsa)) { | |
4347 | ret = PTR_ERR(qsa); | |
4348 | goto drop_write; | |
4349 | } | |
4350 | ||
4351 | if (qsa->flags) { | |
4352 | ret = -EINVAL; | |
4353 | goto out; | |
4354 | } | |
4355 | ||
0b246afa | 4356 | ret = btrfs_qgroup_rescan(fs_info); |
2f232036 JS |
4357 | |
4358 | out: | |
4359 | kfree(qsa); | |
4360 | drop_write: | |
4361 | mnt_drop_write_file(file); | |
4362 | return ret; | |
4363 | } | |
4364 | ||
b929c1d8 MPS |
4365 | static long btrfs_ioctl_quota_rescan_status(struct btrfs_fs_info *fs_info, |
4366 | void __user *arg) | |
2f232036 | 4367 | { |
2f232036 JS |
4368 | struct btrfs_ioctl_quota_rescan_args *qsa; |
4369 | int ret = 0; | |
4370 | ||
4371 | if (!capable(CAP_SYS_ADMIN)) | |
4372 | return -EPERM; | |
4373 | ||
8d2db785 | 4374 | qsa = kzalloc(sizeof(*qsa), GFP_KERNEL); |
2f232036 JS |
4375 | if (!qsa) |
4376 | return -ENOMEM; | |
4377 | ||
0b246afa | 4378 | if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) { |
2f232036 | 4379 | qsa->flags = 1; |
0b246afa | 4380 | qsa->progress = fs_info->qgroup_rescan_progress.objectid; |
2f232036 JS |
4381 | } |
4382 | ||
4383 | if (copy_to_user(arg, qsa, sizeof(*qsa))) | |
4384 | ret = -EFAULT; | |
4385 | ||
4386 | kfree(qsa); | |
4387 | return ret; | |
4388 | } | |
4389 | ||
b929c1d8 MPS |
4390 | static long btrfs_ioctl_quota_rescan_wait(struct btrfs_fs_info *fs_info, |
4391 | void __user *arg) | |
57254b6e | 4392 | { |
57254b6e JS |
4393 | if (!capable(CAP_SYS_ADMIN)) |
4394 | return -EPERM; | |
4395 | ||
0b246afa | 4396 | return btrfs_qgroup_wait_for_completion(fs_info, true); |
57254b6e JS |
4397 | } |
4398 | ||
abccd00f HM |
4399 | static long _btrfs_ioctl_set_received_subvol(struct file *file, |
4400 | struct btrfs_ioctl_received_subvol_args *sa) | |
8ea05e3a | 4401 | { |
496ad9aa | 4402 | struct inode *inode = file_inode(file); |
0b246afa | 4403 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
8ea05e3a AB |
4404 | struct btrfs_root *root = BTRFS_I(inode)->root; |
4405 | struct btrfs_root_item *root_item = &root->root_item; | |
4406 | struct btrfs_trans_handle *trans; | |
95582b00 | 4407 | struct timespec64 ct = current_time(inode); |
8ea05e3a | 4408 | int ret = 0; |
dd5f9615 | 4409 | int received_uuid_changed; |
8ea05e3a | 4410 | |
bd60ea0f DS |
4411 | if (!inode_owner_or_capable(inode)) |
4412 | return -EPERM; | |
4413 | ||
8ea05e3a AB |
4414 | ret = mnt_want_write_file(file); |
4415 | if (ret < 0) | |
4416 | return ret; | |
4417 | ||
0b246afa | 4418 | down_write(&fs_info->subvol_sem); |
8ea05e3a | 4419 | |
4a0cc7ca | 4420 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
8ea05e3a AB |
4421 | ret = -EINVAL; |
4422 | goto out; | |
4423 | } | |
4424 | ||
4425 | if (btrfs_root_readonly(root)) { | |
4426 | ret = -EROFS; | |
4427 | goto out; | |
4428 | } | |
4429 | ||
dd5f9615 SB |
4430 | /* |
4431 | * 1 - root item | |
4432 | * 2 - uuid items (received uuid + subvol uuid) | |
4433 | */ | |
4434 | trans = btrfs_start_transaction(root, 3); | |
8ea05e3a AB |
4435 | if (IS_ERR(trans)) { |
4436 | ret = PTR_ERR(trans); | |
4437 | trans = NULL; | |
4438 | goto out; | |
4439 | } | |
4440 | ||
4441 | sa->rtransid = trans->transid; | |
4442 | sa->rtime.sec = ct.tv_sec; | |
4443 | sa->rtime.nsec = ct.tv_nsec; | |
4444 | ||
dd5f9615 SB |
4445 | received_uuid_changed = memcmp(root_item->received_uuid, sa->uuid, |
4446 | BTRFS_UUID_SIZE); | |
4447 | if (received_uuid_changed && | |
d87ff758 | 4448 | !btrfs_is_empty_uuid(root_item->received_uuid)) { |
d1957791 | 4449 | ret = btrfs_uuid_tree_remove(trans, root_item->received_uuid, |
d87ff758 NB |
4450 | BTRFS_UUID_KEY_RECEIVED_SUBVOL, |
4451 | root->root_key.objectid); | |
4452 | if (ret && ret != -ENOENT) { | |
4453 | btrfs_abort_transaction(trans, ret); | |
4454 | btrfs_end_transaction(trans); | |
4455 | goto out; | |
4456 | } | |
4457 | } | |
8ea05e3a AB |
4458 | memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE); |
4459 | btrfs_set_root_stransid(root_item, sa->stransid); | |
4460 | btrfs_set_root_rtransid(root_item, sa->rtransid); | |
3cae210f QW |
4461 | btrfs_set_stack_timespec_sec(&root_item->stime, sa->stime.sec); |
4462 | btrfs_set_stack_timespec_nsec(&root_item->stime, sa->stime.nsec); | |
4463 | btrfs_set_stack_timespec_sec(&root_item->rtime, sa->rtime.sec); | |
4464 | btrfs_set_stack_timespec_nsec(&root_item->rtime, sa->rtime.nsec); | |
8ea05e3a | 4465 | |
0b246afa | 4466 | ret = btrfs_update_root(trans, fs_info->tree_root, |
8ea05e3a AB |
4467 | &root->root_key, &root->root_item); |
4468 | if (ret < 0) { | |
3a45bb20 | 4469 | btrfs_end_transaction(trans); |
8ea05e3a | 4470 | goto out; |
dd5f9615 SB |
4471 | } |
4472 | if (received_uuid_changed && !btrfs_is_empty_uuid(sa->uuid)) { | |
cdb345a8 | 4473 | ret = btrfs_uuid_tree_add(trans, sa->uuid, |
dd5f9615 SB |
4474 | BTRFS_UUID_KEY_RECEIVED_SUBVOL, |
4475 | root->root_key.objectid); | |
4476 | if (ret < 0 && ret != -EEXIST) { | |
66642832 | 4477 | btrfs_abort_transaction(trans, ret); |
efd38150 | 4478 | btrfs_end_transaction(trans); |
8ea05e3a | 4479 | goto out; |
dd5f9615 SB |
4480 | } |
4481 | } | |
3a45bb20 | 4482 | ret = btrfs_commit_transaction(trans); |
abccd00f | 4483 | out: |
0b246afa | 4484 | up_write(&fs_info->subvol_sem); |
abccd00f HM |
4485 | mnt_drop_write_file(file); |
4486 | return ret; | |
4487 | } | |
4488 | ||
4489 | #ifdef CONFIG_64BIT | |
4490 | static long btrfs_ioctl_set_received_subvol_32(struct file *file, | |
4491 | void __user *arg) | |
4492 | { | |
4493 | struct btrfs_ioctl_received_subvol_args_32 *args32 = NULL; | |
4494 | struct btrfs_ioctl_received_subvol_args *args64 = NULL; | |
4495 | int ret = 0; | |
4496 | ||
4497 | args32 = memdup_user(arg, sizeof(*args32)); | |
7b9ea627 SV |
4498 | if (IS_ERR(args32)) |
4499 | return PTR_ERR(args32); | |
abccd00f | 4500 | |
8d2db785 | 4501 | args64 = kmalloc(sizeof(*args64), GFP_KERNEL); |
84dbeb87 DC |
4502 | if (!args64) { |
4503 | ret = -ENOMEM; | |
abccd00f HM |
4504 | goto out; |
4505 | } | |
4506 | ||
4507 | memcpy(args64->uuid, args32->uuid, BTRFS_UUID_SIZE); | |
4508 | args64->stransid = args32->stransid; | |
4509 | args64->rtransid = args32->rtransid; | |
4510 | args64->stime.sec = args32->stime.sec; | |
4511 | args64->stime.nsec = args32->stime.nsec; | |
4512 | args64->rtime.sec = args32->rtime.sec; | |
4513 | args64->rtime.nsec = args32->rtime.nsec; | |
4514 | args64->flags = args32->flags; | |
4515 | ||
4516 | ret = _btrfs_ioctl_set_received_subvol(file, args64); | |
4517 | if (ret) | |
4518 | goto out; | |
4519 | ||
4520 | memcpy(args32->uuid, args64->uuid, BTRFS_UUID_SIZE); | |
4521 | args32->stransid = args64->stransid; | |
4522 | args32->rtransid = args64->rtransid; | |
4523 | args32->stime.sec = args64->stime.sec; | |
4524 | args32->stime.nsec = args64->stime.nsec; | |
4525 | args32->rtime.sec = args64->rtime.sec; | |
4526 | args32->rtime.nsec = args64->rtime.nsec; | |
4527 | args32->flags = args64->flags; | |
4528 | ||
4529 | ret = copy_to_user(arg, args32, sizeof(*args32)); | |
4530 | if (ret) | |
4531 | ret = -EFAULT; | |
4532 | ||
4533 | out: | |
4534 | kfree(args32); | |
4535 | kfree(args64); | |
4536 | return ret; | |
4537 | } | |
4538 | #endif | |
4539 | ||
4540 | static long btrfs_ioctl_set_received_subvol(struct file *file, | |
4541 | void __user *arg) | |
4542 | { | |
4543 | struct btrfs_ioctl_received_subvol_args *sa = NULL; | |
4544 | int ret = 0; | |
4545 | ||
4546 | sa = memdup_user(arg, sizeof(*sa)); | |
7b9ea627 SV |
4547 | if (IS_ERR(sa)) |
4548 | return PTR_ERR(sa); | |
abccd00f HM |
4549 | |
4550 | ret = _btrfs_ioctl_set_received_subvol(file, sa); | |
4551 | ||
4552 | if (ret) | |
4553 | goto out; | |
4554 | ||
8ea05e3a AB |
4555 | ret = copy_to_user(arg, sa, sizeof(*sa)); |
4556 | if (ret) | |
4557 | ret = -EFAULT; | |
4558 | ||
4559 | out: | |
4560 | kfree(sa); | |
8ea05e3a AB |
4561 | return ret; |
4562 | } | |
4563 | ||
b929c1d8 MPS |
4564 | static int btrfs_ioctl_get_fslabel(struct btrfs_fs_info *fs_info, |
4565 | void __user *arg) | |
867ab667 | 4566 | { |
a1b83ac5 | 4567 | size_t len; |
867ab667 | 4568 | int ret; |
a1b83ac5 AJ |
4569 | char label[BTRFS_LABEL_SIZE]; |
4570 | ||
0b246afa JM |
4571 | spin_lock(&fs_info->super_lock); |
4572 | memcpy(label, fs_info->super_copy->label, BTRFS_LABEL_SIZE); | |
4573 | spin_unlock(&fs_info->super_lock); | |
a1b83ac5 AJ |
4574 | |
4575 | len = strnlen(label, BTRFS_LABEL_SIZE); | |
867ab667 | 4576 | |
4577 | if (len == BTRFS_LABEL_SIZE) { | |
0b246afa JM |
4578 | btrfs_warn(fs_info, |
4579 | "label is too long, return the first %zu bytes", | |
4580 | --len); | |
867ab667 | 4581 | } |
4582 | ||
867ab667 | 4583 | ret = copy_to_user(arg, label, len); |
867ab667 | 4584 | |
4585 | return ret ? -EFAULT : 0; | |
4586 | } | |
4587 | ||
a8bfd4ab | 4588 | static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg) |
4589 | { | |
0b246afa JM |
4590 | struct inode *inode = file_inode(file); |
4591 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4592 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4593 | struct btrfs_super_block *super_block = fs_info->super_copy; | |
a8bfd4ab | 4594 | struct btrfs_trans_handle *trans; |
4595 | char label[BTRFS_LABEL_SIZE]; | |
4596 | int ret; | |
4597 | ||
4598 | if (!capable(CAP_SYS_ADMIN)) | |
4599 | return -EPERM; | |
4600 | ||
4601 | if (copy_from_user(label, arg, sizeof(label))) | |
4602 | return -EFAULT; | |
4603 | ||
4604 | if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) { | |
0b246afa | 4605 | btrfs_err(fs_info, |
5d163e0e JM |
4606 | "unable to set label with more than %d bytes", |
4607 | BTRFS_LABEL_SIZE - 1); | |
a8bfd4ab | 4608 | return -EINVAL; |
4609 | } | |
4610 | ||
4611 | ret = mnt_want_write_file(file); | |
4612 | if (ret) | |
4613 | return ret; | |
4614 | ||
a8bfd4ab | 4615 | trans = btrfs_start_transaction(root, 0); |
4616 | if (IS_ERR(trans)) { | |
4617 | ret = PTR_ERR(trans); | |
4618 | goto out_unlock; | |
4619 | } | |
4620 | ||
0b246afa | 4621 | spin_lock(&fs_info->super_lock); |
a8bfd4ab | 4622 | strcpy(super_block->label, label); |
0b246afa | 4623 | spin_unlock(&fs_info->super_lock); |
3a45bb20 | 4624 | ret = btrfs_commit_transaction(trans); |
a8bfd4ab | 4625 | |
4626 | out_unlock: | |
a8bfd4ab | 4627 | mnt_drop_write_file(file); |
4628 | return ret; | |
4629 | } | |
4630 | ||
2eaa055f JM |
4631 | #define INIT_FEATURE_FLAGS(suffix) \ |
4632 | { .compat_flags = BTRFS_FEATURE_COMPAT_##suffix, \ | |
4633 | .compat_ro_flags = BTRFS_FEATURE_COMPAT_RO_##suffix, \ | |
4634 | .incompat_flags = BTRFS_FEATURE_INCOMPAT_##suffix } | |
4635 | ||
d5131b65 | 4636 | int btrfs_ioctl_get_supported_features(void __user *arg) |
2eaa055f | 4637 | { |
4d4ab6d6 | 4638 | static const struct btrfs_ioctl_feature_flags features[3] = { |
2eaa055f JM |
4639 | INIT_FEATURE_FLAGS(SUPP), |
4640 | INIT_FEATURE_FLAGS(SAFE_SET), | |
4641 | INIT_FEATURE_FLAGS(SAFE_CLEAR) | |
4642 | }; | |
4643 | ||
4644 | if (copy_to_user(arg, &features, sizeof(features))) | |
4645 | return -EFAULT; | |
4646 | ||
4647 | return 0; | |
4648 | } | |
4649 | ||
b929c1d8 MPS |
4650 | static int btrfs_ioctl_get_features(struct btrfs_fs_info *fs_info, |
4651 | void __user *arg) | |
2eaa055f | 4652 | { |
0b246afa | 4653 | struct btrfs_super_block *super_block = fs_info->super_copy; |
2eaa055f JM |
4654 | struct btrfs_ioctl_feature_flags features; |
4655 | ||
4656 | features.compat_flags = btrfs_super_compat_flags(super_block); | |
4657 | features.compat_ro_flags = btrfs_super_compat_ro_flags(super_block); | |
4658 | features.incompat_flags = btrfs_super_incompat_flags(super_block); | |
4659 | ||
4660 | if (copy_to_user(arg, &features, sizeof(features))) | |
4661 | return -EFAULT; | |
4662 | ||
4663 | return 0; | |
4664 | } | |
4665 | ||
2ff7e61e | 4666 | static int check_feature_bits(struct btrfs_fs_info *fs_info, |
3b02a68a | 4667 | enum btrfs_feature_set set, |
2eaa055f JM |
4668 | u64 change_mask, u64 flags, u64 supported_flags, |
4669 | u64 safe_set, u64 safe_clear) | |
4670 | { | |
f10152bc | 4671 | const char *type = btrfs_feature_set_name(set); |
3b02a68a | 4672 | char *names; |
2eaa055f JM |
4673 | u64 disallowed, unsupported; |
4674 | u64 set_mask = flags & change_mask; | |
4675 | u64 clear_mask = ~flags & change_mask; | |
4676 | ||
4677 | unsupported = set_mask & ~supported_flags; | |
4678 | if (unsupported) { | |
3b02a68a JM |
4679 | names = btrfs_printable_features(set, unsupported); |
4680 | if (names) { | |
0b246afa JM |
4681 | btrfs_warn(fs_info, |
4682 | "this kernel does not support the %s feature bit%s", | |
4683 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4684 | kfree(names); |
4685 | } else | |
0b246afa JM |
4686 | btrfs_warn(fs_info, |
4687 | "this kernel does not support %s bits 0x%llx", | |
4688 | type, unsupported); | |
2eaa055f JM |
4689 | return -EOPNOTSUPP; |
4690 | } | |
4691 | ||
4692 | disallowed = set_mask & ~safe_set; | |
4693 | if (disallowed) { | |
3b02a68a JM |
4694 | names = btrfs_printable_features(set, disallowed); |
4695 | if (names) { | |
0b246afa JM |
4696 | btrfs_warn(fs_info, |
4697 | "can't set the %s feature bit%s while mounted", | |
4698 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4699 | kfree(names); |
4700 | } else | |
0b246afa JM |
4701 | btrfs_warn(fs_info, |
4702 | "can't set %s bits 0x%llx while mounted", | |
4703 | type, disallowed); | |
2eaa055f JM |
4704 | return -EPERM; |
4705 | } | |
4706 | ||
4707 | disallowed = clear_mask & ~safe_clear; | |
4708 | if (disallowed) { | |
3b02a68a JM |
4709 | names = btrfs_printable_features(set, disallowed); |
4710 | if (names) { | |
0b246afa JM |
4711 | btrfs_warn(fs_info, |
4712 | "can't clear the %s feature bit%s while mounted", | |
4713 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4714 | kfree(names); |
4715 | } else | |
0b246afa JM |
4716 | btrfs_warn(fs_info, |
4717 | "can't clear %s bits 0x%llx while mounted", | |
4718 | type, disallowed); | |
2eaa055f JM |
4719 | return -EPERM; |
4720 | } | |
4721 | ||
4722 | return 0; | |
4723 | } | |
4724 | ||
2ff7e61e JM |
4725 | #define check_feature(fs_info, change_mask, flags, mask_base) \ |
4726 | check_feature_bits(fs_info, FEAT_##mask_base, change_mask, flags, \ | |
2eaa055f JM |
4727 | BTRFS_FEATURE_ ## mask_base ## _SUPP, \ |
4728 | BTRFS_FEATURE_ ## mask_base ## _SAFE_SET, \ | |
4729 | BTRFS_FEATURE_ ## mask_base ## _SAFE_CLEAR) | |
4730 | ||
4731 | static int btrfs_ioctl_set_features(struct file *file, void __user *arg) | |
4732 | { | |
0b246afa JM |
4733 | struct inode *inode = file_inode(file); |
4734 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4735 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4736 | struct btrfs_super_block *super_block = fs_info->super_copy; | |
2eaa055f JM |
4737 | struct btrfs_ioctl_feature_flags flags[2]; |
4738 | struct btrfs_trans_handle *trans; | |
4739 | u64 newflags; | |
4740 | int ret; | |
4741 | ||
4742 | if (!capable(CAP_SYS_ADMIN)) | |
4743 | return -EPERM; | |
4744 | ||
4745 | if (copy_from_user(flags, arg, sizeof(flags))) | |
4746 | return -EFAULT; | |
4747 | ||
4748 | /* Nothing to do */ | |
4749 | if (!flags[0].compat_flags && !flags[0].compat_ro_flags && | |
4750 | !flags[0].incompat_flags) | |
4751 | return 0; | |
4752 | ||
2ff7e61e | 4753 | ret = check_feature(fs_info, flags[0].compat_flags, |
2eaa055f JM |
4754 | flags[1].compat_flags, COMPAT); |
4755 | if (ret) | |
4756 | return ret; | |
4757 | ||
2ff7e61e | 4758 | ret = check_feature(fs_info, flags[0].compat_ro_flags, |
2eaa055f JM |
4759 | flags[1].compat_ro_flags, COMPAT_RO); |
4760 | if (ret) | |
4761 | return ret; | |
4762 | ||
2ff7e61e | 4763 | ret = check_feature(fs_info, flags[0].incompat_flags, |
2eaa055f JM |
4764 | flags[1].incompat_flags, INCOMPAT); |
4765 | if (ret) | |
4766 | return ret; | |
4767 | ||
7ab19625 DS |
4768 | ret = mnt_want_write_file(file); |
4769 | if (ret) | |
4770 | return ret; | |
4771 | ||
8051aa1a | 4772 | trans = btrfs_start_transaction(root, 0); |
7ab19625 DS |
4773 | if (IS_ERR(trans)) { |
4774 | ret = PTR_ERR(trans); | |
4775 | goto out_drop_write; | |
4776 | } | |
2eaa055f | 4777 | |
0b246afa | 4778 | spin_lock(&fs_info->super_lock); |
2eaa055f JM |
4779 | newflags = btrfs_super_compat_flags(super_block); |
4780 | newflags |= flags[0].compat_flags & flags[1].compat_flags; | |
4781 | newflags &= ~(flags[0].compat_flags & ~flags[1].compat_flags); | |
4782 | btrfs_set_super_compat_flags(super_block, newflags); | |
4783 | ||
4784 | newflags = btrfs_super_compat_ro_flags(super_block); | |
4785 | newflags |= flags[0].compat_ro_flags & flags[1].compat_ro_flags; | |
4786 | newflags &= ~(flags[0].compat_ro_flags & ~flags[1].compat_ro_flags); | |
4787 | btrfs_set_super_compat_ro_flags(super_block, newflags); | |
4788 | ||
4789 | newflags = btrfs_super_incompat_flags(super_block); | |
4790 | newflags |= flags[0].incompat_flags & flags[1].incompat_flags; | |
4791 | newflags &= ~(flags[0].incompat_flags & ~flags[1].incompat_flags); | |
4792 | btrfs_set_super_incompat_flags(super_block, newflags); | |
0b246afa | 4793 | spin_unlock(&fs_info->super_lock); |
2eaa055f | 4794 | |
3a45bb20 | 4795 | ret = btrfs_commit_transaction(trans); |
7ab19625 DS |
4796 | out_drop_write: |
4797 | mnt_drop_write_file(file); | |
4798 | ||
4799 | return ret; | |
2eaa055f JM |
4800 | } |
4801 | ||
2351f431 JB |
4802 | static int _btrfs_ioctl_send(struct file *file, void __user *argp, bool compat) |
4803 | { | |
4804 | struct btrfs_ioctl_send_args *arg; | |
4805 | int ret; | |
4806 | ||
4807 | if (compat) { | |
4808 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) | |
4809 | struct btrfs_ioctl_send_args_32 args32; | |
4810 | ||
4811 | ret = copy_from_user(&args32, argp, sizeof(args32)); | |
4812 | if (ret) | |
4813 | return -EFAULT; | |
4814 | arg = kzalloc(sizeof(*arg), GFP_KERNEL); | |
4815 | if (!arg) | |
4816 | return -ENOMEM; | |
4817 | arg->send_fd = args32.send_fd; | |
4818 | arg->clone_sources_count = args32.clone_sources_count; | |
4819 | arg->clone_sources = compat_ptr(args32.clone_sources); | |
4820 | arg->parent_root = args32.parent_root; | |
4821 | arg->flags = args32.flags; | |
4822 | memcpy(arg->reserved, args32.reserved, | |
4823 | sizeof(args32.reserved)); | |
4824 | #else | |
4825 | return -ENOTTY; | |
4826 | #endif | |
4827 | } else { | |
4828 | arg = memdup_user(argp, sizeof(*arg)); | |
4829 | if (IS_ERR(arg)) | |
4830 | return PTR_ERR(arg); | |
4831 | } | |
4832 | ret = btrfs_ioctl_send(file, arg); | |
4833 | kfree(arg); | |
4834 | return ret; | |
4835 | } | |
4836 | ||
f46b5a66 CH |
4837 | long btrfs_ioctl(struct file *file, unsigned int |
4838 | cmd, unsigned long arg) | |
4839 | { | |
0b246afa JM |
4840 | struct inode *inode = file_inode(file); |
4841 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4842 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4bcabaa3 | 4843 | void __user *argp = (void __user *)arg; |
f46b5a66 CH |
4844 | |
4845 | switch (cmd) { | |
6cbff00f CH |
4846 | case FS_IOC_GETFLAGS: |
4847 | return btrfs_ioctl_getflags(file, argp); | |
4848 | case FS_IOC_SETFLAGS: | |
4849 | return btrfs_ioctl_setflags(file, argp); | |
4850 | case FS_IOC_GETVERSION: | |
4851 | return btrfs_ioctl_getversion(file, argp); | |
40cf931f | 4852 | case FS_IOC_GETFSLABEL: |
b929c1d8 | 4853 | return btrfs_ioctl_get_fslabel(fs_info, argp); |
40cf931f ES |
4854 | case FS_IOC_SETFSLABEL: |
4855 | return btrfs_ioctl_set_fslabel(file, argp); | |
f7039b1d | 4856 | case FITRIM: |
b929c1d8 | 4857 | return btrfs_ioctl_fitrim(fs_info, argp); |
f46b5a66 | 4858 | case BTRFS_IOC_SNAP_CREATE: |
fa0d2b9b | 4859 | return btrfs_ioctl_snap_create(file, argp, 0); |
fdfb1e4f | 4860 | case BTRFS_IOC_SNAP_CREATE_V2: |
fa0d2b9b | 4861 | return btrfs_ioctl_snap_create_v2(file, argp, 0); |
3de4586c | 4862 | case BTRFS_IOC_SUBVOL_CREATE: |
fa0d2b9b | 4863 | return btrfs_ioctl_snap_create(file, argp, 1); |
6f72c7e2 AJ |
4864 | case BTRFS_IOC_SUBVOL_CREATE_V2: |
4865 | return btrfs_ioctl_snap_create_v2(file, argp, 1); | |
76dda93c | 4866 | case BTRFS_IOC_SNAP_DESTROY: |
949964c9 MPS |
4867 | return btrfs_ioctl_snap_destroy(file, argp, false); |
4868 | case BTRFS_IOC_SNAP_DESTROY_V2: | |
4869 | return btrfs_ioctl_snap_destroy(file, argp, true); | |
0caa102d LZ |
4870 | case BTRFS_IOC_SUBVOL_GETFLAGS: |
4871 | return btrfs_ioctl_subvol_getflags(file, argp); | |
4872 | case BTRFS_IOC_SUBVOL_SETFLAGS: | |
4873 | return btrfs_ioctl_subvol_setflags(file, argp); | |
6ef5ed0d JB |
4874 | case BTRFS_IOC_DEFAULT_SUBVOL: |
4875 | return btrfs_ioctl_default_subvol(file, argp); | |
f46b5a66 | 4876 | case BTRFS_IOC_DEFRAG: |
1e701a32 CM |
4877 | return btrfs_ioctl_defrag(file, NULL); |
4878 | case BTRFS_IOC_DEFRAG_RANGE: | |
4879 | return btrfs_ioctl_defrag(file, argp); | |
f46b5a66 | 4880 | case BTRFS_IOC_RESIZE: |
198605a8 | 4881 | return btrfs_ioctl_resize(file, argp); |
f46b5a66 | 4882 | case BTRFS_IOC_ADD_DEV: |
2ff7e61e | 4883 | return btrfs_ioctl_add_dev(fs_info, argp); |
f46b5a66 | 4884 | case BTRFS_IOC_RM_DEV: |
da24927b | 4885 | return btrfs_ioctl_rm_dev(file, argp); |
6b526ed7 AJ |
4886 | case BTRFS_IOC_RM_DEV_V2: |
4887 | return btrfs_ioctl_rm_dev_v2(file, argp); | |
475f6387 | 4888 | case BTRFS_IOC_FS_INFO: |
2ff7e61e | 4889 | return btrfs_ioctl_fs_info(fs_info, argp); |
475f6387 | 4890 | case BTRFS_IOC_DEV_INFO: |
2ff7e61e | 4891 | return btrfs_ioctl_dev_info(fs_info, argp); |
f46b5a66 | 4892 | case BTRFS_IOC_BALANCE: |
9ba1f6e4 | 4893 | return btrfs_ioctl_balance(file, NULL); |
ac8e9819 CM |
4894 | case BTRFS_IOC_TREE_SEARCH: |
4895 | return btrfs_ioctl_tree_search(file, argp); | |
cc68a8a5 GH |
4896 | case BTRFS_IOC_TREE_SEARCH_V2: |
4897 | return btrfs_ioctl_tree_search_v2(file, argp); | |
ac8e9819 CM |
4898 | case BTRFS_IOC_INO_LOOKUP: |
4899 | return btrfs_ioctl_ino_lookup(file, argp); | |
d7728c96 JS |
4900 | case BTRFS_IOC_INO_PATHS: |
4901 | return btrfs_ioctl_ino_to_path(root, argp); | |
4902 | case BTRFS_IOC_LOGICAL_INO: | |
d24a67b2 ZB |
4903 | return btrfs_ioctl_logical_to_ino(fs_info, argp, 1); |
4904 | case BTRFS_IOC_LOGICAL_INO_V2: | |
4905 | return btrfs_ioctl_logical_to_ino(fs_info, argp, 2); | |
1406e432 | 4906 | case BTRFS_IOC_SPACE_INFO: |
2ff7e61e | 4907 | return btrfs_ioctl_space_info(fs_info, argp); |
9b199859 FDBM |
4908 | case BTRFS_IOC_SYNC: { |
4909 | int ret; | |
4910 | ||
b4912139 | 4911 | ret = btrfs_start_delalloc_roots(fs_info, U64_MAX); |
9b199859 FDBM |
4912 | if (ret) |
4913 | return ret; | |
0b246afa | 4914 | ret = btrfs_sync_fs(inode->i_sb, 1); |
2fad4e83 DS |
4915 | /* |
4916 | * The transaction thread may want to do more work, | |
01327610 | 4917 | * namely it pokes the cleaner kthread that will start |
2fad4e83 DS |
4918 | * processing uncleaned subvols. |
4919 | */ | |
0b246afa | 4920 | wake_up_process(fs_info->transaction_kthread); |
9b199859 FDBM |
4921 | return ret; |
4922 | } | |
46204592 | 4923 | case BTRFS_IOC_START_SYNC: |
9a8c28be | 4924 | return btrfs_ioctl_start_sync(root, argp); |
46204592 | 4925 | case BTRFS_IOC_WAIT_SYNC: |
2ff7e61e | 4926 | return btrfs_ioctl_wait_sync(fs_info, argp); |
475f6387 | 4927 | case BTRFS_IOC_SCRUB: |
b8e95489 | 4928 | return btrfs_ioctl_scrub(file, argp); |
475f6387 | 4929 | case BTRFS_IOC_SCRUB_CANCEL: |
2ff7e61e | 4930 | return btrfs_ioctl_scrub_cancel(fs_info); |
475f6387 | 4931 | case BTRFS_IOC_SCRUB_PROGRESS: |
2ff7e61e | 4932 | return btrfs_ioctl_scrub_progress(fs_info, argp); |
c9e9f97b | 4933 | case BTRFS_IOC_BALANCE_V2: |
9ba1f6e4 | 4934 | return btrfs_ioctl_balance(file, argp); |
837d5b6e | 4935 | case BTRFS_IOC_BALANCE_CTL: |
2ff7e61e | 4936 | return btrfs_ioctl_balance_ctl(fs_info, arg); |
19a39dce | 4937 | case BTRFS_IOC_BALANCE_PROGRESS: |
2ff7e61e | 4938 | return btrfs_ioctl_balance_progress(fs_info, argp); |
8ea05e3a AB |
4939 | case BTRFS_IOC_SET_RECEIVED_SUBVOL: |
4940 | return btrfs_ioctl_set_received_subvol(file, argp); | |
abccd00f HM |
4941 | #ifdef CONFIG_64BIT |
4942 | case BTRFS_IOC_SET_RECEIVED_SUBVOL_32: | |
4943 | return btrfs_ioctl_set_received_subvol_32(file, argp); | |
4944 | #endif | |
31db9f7c | 4945 | case BTRFS_IOC_SEND: |
2351f431 JB |
4946 | return _btrfs_ioctl_send(file, argp, false); |
4947 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) | |
4948 | case BTRFS_IOC_SEND_32: | |
4949 | return _btrfs_ioctl_send(file, argp, true); | |
4950 | #endif | |
c11d2c23 | 4951 | case BTRFS_IOC_GET_DEV_STATS: |
2ff7e61e | 4952 | return btrfs_ioctl_get_dev_stats(fs_info, argp); |
5d13a37b | 4953 | case BTRFS_IOC_QUOTA_CTL: |
905b0dda | 4954 | return btrfs_ioctl_quota_ctl(file, argp); |
5d13a37b | 4955 | case BTRFS_IOC_QGROUP_ASSIGN: |
905b0dda | 4956 | return btrfs_ioctl_qgroup_assign(file, argp); |
5d13a37b | 4957 | case BTRFS_IOC_QGROUP_CREATE: |
905b0dda | 4958 | return btrfs_ioctl_qgroup_create(file, argp); |
5d13a37b | 4959 | case BTRFS_IOC_QGROUP_LIMIT: |
905b0dda | 4960 | return btrfs_ioctl_qgroup_limit(file, argp); |
2f232036 JS |
4961 | case BTRFS_IOC_QUOTA_RESCAN: |
4962 | return btrfs_ioctl_quota_rescan(file, argp); | |
4963 | case BTRFS_IOC_QUOTA_RESCAN_STATUS: | |
b929c1d8 | 4964 | return btrfs_ioctl_quota_rescan_status(fs_info, argp); |
57254b6e | 4965 | case BTRFS_IOC_QUOTA_RESCAN_WAIT: |
b929c1d8 | 4966 | return btrfs_ioctl_quota_rescan_wait(fs_info, argp); |
3f6bcfbd | 4967 | case BTRFS_IOC_DEV_REPLACE: |
2ff7e61e | 4968 | return btrfs_ioctl_dev_replace(fs_info, argp); |
2eaa055f | 4969 | case BTRFS_IOC_GET_SUPPORTED_FEATURES: |
d5131b65 | 4970 | return btrfs_ioctl_get_supported_features(argp); |
2eaa055f | 4971 | case BTRFS_IOC_GET_FEATURES: |
b929c1d8 | 4972 | return btrfs_ioctl_get_features(fs_info, argp); |
2eaa055f JM |
4973 | case BTRFS_IOC_SET_FEATURES: |
4974 | return btrfs_ioctl_set_features(file, argp); | |
e4202ac9 DS |
4975 | case FS_IOC_FSGETXATTR: |
4976 | return btrfs_ioctl_fsgetxattr(file, argp); | |
025f2121 DS |
4977 | case FS_IOC_FSSETXATTR: |
4978 | return btrfs_ioctl_fssetxattr(file, argp); | |
b64ec075 TM |
4979 | case BTRFS_IOC_GET_SUBVOL_INFO: |
4980 | return btrfs_ioctl_get_subvol_info(file, argp); | |
42e4b520 TM |
4981 | case BTRFS_IOC_GET_SUBVOL_ROOTREF: |
4982 | return btrfs_ioctl_get_subvol_rootref(file, argp); | |
23d0b79d TM |
4983 | case BTRFS_IOC_INO_LOOKUP_USER: |
4984 | return btrfs_ioctl_ino_lookup_user(file, argp); | |
f46b5a66 CH |
4985 | } |
4986 | ||
4987 | return -ENOTTY; | |
4988 | } | |
4c63c245 LD |
4989 | |
4990 | #ifdef CONFIG_COMPAT | |
4991 | long btrfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg) | |
4992 | { | |
2a362249 JM |
4993 | /* |
4994 | * These all access 32-bit values anyway so no further | |
4995 | * handling is necessary. | |
4996 | */ | |
4c63c245 LD |
4997 | switch (cmd) { |
4998 | case FS_IOC32_GETFLAGS: | |
4999 | cmd = FS_IOC_GETFLAGS; | |
5000 | break; | |
5001 | case FS_IOC32_SETFLAGS: | |
5002 | cmd = FS_IOC_SETFLAGS; | |
5003 | break; | |
5004 | case FS_IOC32_GETVERSION: | |
5005 | cmd = FS_IOC_GETVERSION; | |
5006 | break; | |
4c63c245 LD |
5007 | } |
5008 | ||
5009 | return btrfs_ioctl(file, cmd, (unsigned long) compat_ptr(arg)); | |
5010 | } | |
5011 | #endif |