]>
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" |
d7728c96 | 37 | #include "backref.h" |
606686ee | 38 | #include "rcu-string.h" |
31db9f7c | 39 | #include "send.h" |
3f6bcfbd | 40 | #include "dev-replace.h" |
63541927 | 41 | #include "props.h" |
3b02a68a | 42 | #include "sysfs.h" |
fcebe456 | 43 | #include "qgroup.h" |
1ec9a1ae | 44 | #include "tree-log.h" |
ebb8765b | 45 | #include "compression.h" |
8719aaae | 46 | #include "space-info.h" |
86736342 | 47 | #include "delalloc-space.h" |
aac0023c | 48 | #include "block-group.h" |
f46b5a66 | 49 | |
abccd00f HM |
50 | #ifdef CONFIG_64BIT |
51 | /* If we have a 32-bit userspace and 64-bit kernel, then the UAPI | |
52 | * structures are incorrect, as the timespec structure from userspace | |
53 | * is 4 bytes too small. We define these alternatives here to teach | |
54 | * the kernel about the 32-bit struct packing. | |
55 | */ | |
56 | struct btrfs_ioctl_timespec_32 { | |
57 | __u64 sec; | |
58 | __u32 nsec; | |
59 | } __attribute__ ((__packed__)); | |
60 | ||
61 | struct btrfs_ioctl_received_subvol_args_32 { | |
62 | char uuid[BTRFS_UUID_SIZE]; /* in */ | |
63 | __u64 stransid; /* in */ | |
64 | __u64 rtransid; /* out */ | |
65 | struct btrfs_ioctl_timespec_32 stime; /* in */ | |
66 | struct btrfs_ioctl_timespec_32 rtime; /* out */ | |
67 | __u64 flags; /* in */ | |
68 | __u64 reserved[16]; /* in */ | |
69 | } __attribute__ ((__packed__)); | |
70 | ||
71 | #define BTRFS_IOC_SET_RECEIVED_SUBVOL_32 _IOWR(BTRFS_IOCTL_MAGIC, 37, \ | |
72 | struct btrfs_ioctl_received_subvol_args_32) | |
73 | #endif | |
74 | ||
2351f431 JB |
75 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) |
76 | struct btrfs_ioctl_send_args_32 { | |
77 | __s64 send_fd; /* in */ | |
78 | __u64 clone_sources_count; /* in */ | |
79 | compat_uptr_t clone_sources; /* in */ | |
80 | __u64 parent_root; /* in */ | |
81 | __u64 flags; /* in */ | |
82 | __u64 reserved[4]; /* in */ | |
83 | } __attribute__ ((__packed__)); | |
84 | ||
85 | #define BTRFS_IOC_SEND_32 _IOW(BTRFS_IOCTL_MAGIC, 38, \ | |
86 | struct btrfs_ioctl_send_args_32) | |
87 | #endif | |
abccd00f | 88 | |
6cbff00f | 89 | /* Mask out flags that are inappropriate for the given type of inode. */ |
1905a0f7 DS |
90 | static unsigned int btrfs_mask_fsflags_for_type(struct inode *inode, |
91 | unsigned int flags) | |
6cbff00f | 92 | { |
1905a0f7 | 93 | if (S_ISDIR(inode->i_mode)) |
6cbff00f | 94 | return flags; |
1905a0f7 | 95 | else if (S_ISREG(inode->i_mode)) |
6cbff00f CH |
96 | return flags & ~FS_DIRSYNC_FL; |
97 | else | |
98 | return flags & (FS_NODUMP_FL | FS_NOATIME_FL); | |
99 | } | |
100 | ||
101 | /* | |
a157d4fd DS |
102 | * Export internal inode flags to the format expected by the FS_IOC_GETFLAGS |
103 | * ioctl. | |
6cbff00f | 104 | */ |
a157d4fd | 105 | static unsigned int btrfs_inode_flags_to_fsflags(unsigned int flags) |
6cbff00f CH |
106 | { |
107 | unsigned int iflags = 0; | |
108 | ||
109 | if (flags & BTRFS_INODE_SYNC) | |
110 | iflags |= FS_SYNC_FL; | |
111 | if (flags & BTRFS_INODE_IMMUTABLE) | |
112 | iflags |= FS_IMMUTABLE_FL; | |
113 | if (flags & BTRFS_INODE_APPEND) | |
114 | iflags |= FS_APPEND_FL; | |
115 | if (flags & BTRFS_INODE_NODUMP) | |
116 | iflags |= FS_NODUMP_FL; | |
117 | if (flags & BTRFS_INODE_NOATIME) | |
118 | iflags |= FS_NOATIME_FL; | |
119 | if (flags & BTRFS_INODE_DIRSYNC) | |
120 | iflags |= FS_DIRSYNC_FL; | |
d0092bdd LZ |
121 | if (flags & BTRFS_INODE_NODATACOW) |
122 | iflags |= FS_NOCOW_FL; | |
123 | ||
13f48dc9 | 124 | if (flags & BTRFS_INODE_NOCOMPRESS) |
d0092bdd | 125 | iflags |= FS_NOCOMP_FL; |
13f48dc9 ST |
126 | else if (flags & BTRFS_INODE_COMPRESS) |
127 | iflags |= FS_COMPR_FL; | |
6cbff00f CH |
128 | |
129 | return iflags; | |
130 | } | |
131 | ||
132 | /* | |
133 | * Update inode->i_flags based on the btrfs internal flags. | |
134 | */ | |
7b6a221e | 135 | void btrfs_sync_inode_flags_to_i_flags(struct inode *inode) |
6cbff00f | 136 | { |
5c57b8b6 | 137 | struct btrfs_inode *binode = BTRFS_I(inode); |
3cc79392 | 138 | unsigned int new_fl = 0; |
6cbff00f | 139 | |
5c57b8b6 | 140 | if (binode->flags & BTRFS_INODE_SYNC) |
3cc79392 | 141 | new_fl |= S_SYNC; |
5c57b8b6 | 142 | if (binode->flags & BTRFS_INODE_IMMUTABLE) |
3cc79392 | 143 | new_fl |= S_IMMUTABLE; |
5c57b8b6 | 144 | if (binode->flags & BTRFS_INODE_APPEND) |
3cc79392 | 145 | new_fl |= S_APPEND; |
5c57b8b6 | 146 | if (binode->flags & BTRFS_INODE_NOATIME) |
3cc79392 | 147 | new_fl |= S_NOATIME; |
5c57b8b6 | 148 | if (binode->flags & BTRFS_INODE_DIRSYNC) |
3cc79392 FM |
149 | new_fl |= S_DIRSYNC; |
150 | ||
151 | set_mask_bits(&inode->i_flags, | |
152 | S_SYNC | S_APPEND | S_IMMUTABLE | S_NOATIME | S_DIRSYNC, | |
153 | new_fl); | |
6cbff00f CH |
154 | } |
155 | ||
6cbff00f CH |
156 | static int btrfs_ioctl_getflags(struct file *file, void __user *arg) |
157 | { | |
5c57b8b6 DS |
158 | struct btrfs_inode *binode = BTRFS_I(file_inode(file)); |
159 | unsigned int flags = btrfs_inode_flags_to_fsflags(binode->flags); | |
6cbff00f CH |
160 | |
161 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
162 | return -EFAULT; | |
163 | return 0; | |
164 | } | |
165 | ||
f37c563b DS |
166 | /* |
167 | * Check if @flags are a supported and valid set of FS_*_FL flags and that | |
168 | * the old and new flags are not conflicting | |
169 | */ | |
170 | static int check_fsflags(unsigned int old_flags, unsigned int flags) | |
75e7cb7f LB |
171 | { |
172 | if (flags & ~(FS_IMMUTABLE_FL | FS_APPEND_FL | \ | |
173 | FS_NOATIME_FL | FS_NODUMP_FL | \ | |
174 | FS_SYNC_FL | FS_DIRSYNC_FL | \ | |
e1e8fb6a LZ |
175 | FS_NOCOMP_FL | FS_COMPR_FL | |
176 | FS_NOCOW_FL)) | |
75e7cb7f LB |
177 | return -EOPNOTSUPP; |
178 | ||
f37c563b | 179 | /* COMPR and NOCOMP on new/old are valid */ |
75e7cb7f LB |
180 | if ((flags & FS_NOCOMP_FL) && (flags & FS_COMPR_FL)) |
181 | return -EINVAL; | |
182 | ||
f37c563b DS |
183 | if ((flags & FS_COMPR_FL) && (flags & FS_NOCOW_FL)) |
184 | return -EINVAL; | |
185 | ||
186 | /* NOCOW and compression options are mutually exclusive */ | |
187 | if ((old_flags & FS_NOCOW_FL) && (flags & (FS_COMPR_FL | FS_NOCOMP_FL))) | |
188 | return -EINVAL; | |
189 | if ((flags & FS_NOCOW_FL) && (old_flags & (FS_COMPR_FL | FS_NOCOMP_FL))) | |
190 | return -EINVAL; | |
191 | ||
75e7cb7f LB |
192 | return 0; |
193 | } | |
194 | ||
d206e9c9 NA |
195 | static int check_fsflags_compatible(struct btrfs_fs_info *fs_info, |
196 | unsigned int flags) | |
197 | { | |
198 | if (btrfs_is_zoned(fs_info) && (flags & FS_NOCOW_FL)) | |
199 | return -EPERM; | |
200 | ||
201 | return 0; | |
202 | } | |
203 | ||
6cbff00f CH |
204 | static int btrfs_ioctl_setflags(struct file *file, void __user *arg) |
205 | { | |
496ad9aa | 206 | struct inode *inode = file_inode(file); |
0b246afa | 207 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
5c57b8b6 DS |
208 | struct btrfs_inode *binode = BTRFS_I(inode); |
209 | struct btrfs_root *root = binode->root; | |
6cbff00f | 210 | struct btrfs_trans_handle *trans; |
5aca2842 | 211 | unsigned int fsflags, old_fsflags; |
6cbff00f | 212 | int ret; |
ff9fef55 | 213 | const char *comp = NULL; |
f37c563b | 214 | u32 binode_flags; |
6cbff00f | 215 | |
bd60ea0f DS |
216 | if (!inode_owner_or_capable(inode)) |
217 | return -EPERM; | |
218 | ||
b83cc969 LZ |
219 | if (btrfs_root_readonly(root)) |
220 | return -EROFS; | |
221 | ||
5c57b8b6 | 222 | if (copy_from_user(&fsflags, arg, sizeof(fsflags))) |
6cbff00f CH |
223 | return -EFAULT; |
224 | ||
e7848683 JK |
225 | ret = mnt_want_write_file(file); |
226 | if (ret) | |
227 | return ret; | |
228 | ||
5955102c | 229 | inode_lock(inode); |
5c57b8b6 | 230 | fsflags = btrfs_mask_fsflags_for_type(inode, fsflags); |
5aca2842 | 231 | old_fsflags = btrfs_inode_flags_to_fsflags(binode->flags); |
f37c563b | 232 | |
5aca2842 DW |
233 | ret = vfs_ioc_setflags_prepare(inode, old_fsflags, fsflags); |
234 | if (ret) | |
235 | goto out_unlock; | |
6cbff00f | 236 | |
f37c563b DS |
237 | ret = check_fsflags(old_fsflags, fsflags); |
238 | if (ret) | |
239 | goto out_unlock; | |
240 | ||
d206e9c9 NA |
241 | ret = check_fsflags_compatible(fs_info, fsflags); |
242 | if (ret) | |
243 | goto out_unlock; | |
244 | ||
f37c563b | 245 | binode_flags = binode->flags; |
5c57b8b6 | 246 | if (fsflags & FS_SYNC_FL) |
d2b8fcfe | 247 | binode_flags |= BTRFS_INODE_SYNC; |
6cbff00f | 248 | else |
d2b8fcfe | 249 | binode_flags &= ~BTRFS_INODE_SYNC; |
5c57b8b6 | 250 | if (fsflags & FS_IMMUTABLE_FL) |
d2b8fcfe | 251 | binode_flags |= BTRFS_INODE_IMMUTABLE; |
6cbff00f | 252 | else |
d2b8fcfe | 253 | binode_flags &= ~BTRFS_INODE_IMMUTABLE; |
5c57b8b6 | 254 | if (fsflags & FS_APPEND_FL) |
d2b8fcfe | 255 | binode_flags |= BTRFS_INODE_APPEND; |
6cbff00f | 256 | else |
d2b8fcfe | 257 | binode_flags &= ~BTRFS_INODE_APPEND; |
5c57b8b6 | 258 | if (fsflags & FS_NODUMP_FL) |
d2b8fcfe | 259 | binode_flags |= BTRFS_INODE_NODUMP; |
6cbff00f | 260 | else |
d2b8fcfe | 261 | binode_flags &= ~BTRFS_INODE_NODUMP; |
5c57b8b6 | 262 | if (fsflags & FS_NOATIME_FL) |
d2b8fcfe | 263 | binode_flags |= BTRFS_INODE_NOATIME; |
6cbff00f | 264 | else |
d2b8fcfe | 265 | binode_flags &= ~BTRFS_INODE_NOATIME; |
5c57b8b6 | 266 | if (fsflags & FS_DIRSYNC_FL) |
d2b8fcfe | 267 | binode_flags |= BTRFS_INODE_DIRSYNC; |
6cbff00f | 268 | else |
d2b8fcfe | 269 | binode_flags &= ~BTRFS_INODE_DIRSYNC; |
5c57b8b6 | 270 | if (fsflags & FS_NOCOW_FL) { |
44e5194b | 271 | if (S_ISREG(inode->i_mode)) { |
7e97b8da DS |
272 | /* |
273 | * It's safe to turn csums off here, no extents exist. | |
274 | * Otherwise we want the flag to reflect the real COW | |
275 | * status of the file and will not set it. | |
276 | */ | |
277 | if (inode->i_size == 0) | |
d2b8fcfe AJ |
278 | binode_flags |= BTRFS_INODE_NODATACOW | |
279 | BTRFS_INODE_NODATASUM; | |
7e97b8da | 280 | } else { |
d2b8fcfe | 281 | binode_flags |= BTRFS_INODE_NODATACOW; |
7e97b8da DS |
282 | } |
283 | } else { | |
284 | /* | |
01327610 | 285 | * Revert back under same assumptions as above |
7e97b8da | 286 | */ |
44e5194b | 287 | if (S_ISREG(inode->i_mode)) { |
7e97b8da | 288 | if (inode->i_size == 0) |
d2b8fcfe AJ |
289 | binode_flags &= ~(BTRFS_INODE_NODATACOW | |
290 | BTRFS_INODE_NODATASUM); | |
7e97b8da | 291 | } else { |
d2b8fcfe | 292 | binode_flags &= ~BTRFS_INODE_NODATACOW; |
7e97b8da DS |
293 | } |
294 | } | |
6cbff00f | 295 | |
75e7cb7f LB |
296 | /* |
297 | * The COMPRESS flag can only be changed by users, while the NOCOMPRESS | |
298 | * flag may be changed automatically if compression code won't make | |
299 | * things smaller. | |
300 | */ | |
5c57b8b6 | 301 | if (fsflags & FS_NOCOMP_FL) { |
d2b8fcfe AJ |
302 | binode_flags &= ~BTRFS_INODE_COMPRESS; |
303 | binode_flags |= BTRFS_INODE_NOCOMPRESS; | |
5c57b8b6 | 304 | } else if (fsflags & FS_COMPR_FL) { |
63541927 | 305 | |
eede2bf3 OS |
306 | if (IS_SWAPFILE(inode)) { |
307 | ret = -ETXTBSY; | |
308 | goto out_unlock; | |
309 | } | |
310 | ||
d2b8fcfe AJ |
311 | binode_flags |= BTRFS_INODE_COMPRESS; |
312 | binode_flags &= ~BTRFS_INODE_NOCOMPRESS; | |
63541927 | 313 | |
93370509 DS |
314 | comp = btrfs_compress_type2str(fs_info->compress_type); |
315 | if (!comp || comp[0] == 0) | |
316 | comp = btrfs_compress_type2str(BTRFS_COMPRESS_ZLIB); | |
ebcb904d | 317 | } else { |
d2b8fcfe | 318 | binode_flags &= ~(BTRFS_INODE_COMPRESS | BTRFS_INODE_NOCOMPRESS); |
75e7cb7f | 319 | } |
6cbff00f | 320 | |
ff9fef55 AJ |
321 | /* |
322 | * 1 for inode item | |
323 | * 2 for properties | |
324 | */ | |
325 | trans = btrfs_start_transaction(root, 3); | |
f062abf0 LZ |
326 | if (IS_ERR(trans)) { |
327 | ret = PTR_ERR(trans); | |
d2b8fcfe | 328 | goto out_unlock; |
f062abf0 | 329 | } |
6cbff00f | 330 | |
ff9fef55 AJ |
331 | if (comp) { |
332 | ret = btrfs_set_prop(trans, inode, "btrfs.compression", comp, | |
333 | strlen(comp), 0); | |
334 | if (ret) { | |
335 | btrfs_abort_transaction(trans, ret); | |
336 | goto out_end_trans; | |
337 | } | |
ff9fef55 AJ |
338 | } else { |
339 | ret = btrfs_set_prop(trans, inode, "btrfs.compression", NULL, | |
340 | 0, 0); | |
341 | if (ret && ret != -ENODATA) { | |
342 | btrfs_abort_transaction(trans, ret); | |
343 | goto out_end_trans; | |
344 | } | |
345 | } | |
346 | ||
d2b8fcfe | 347 | binode->flags = binode_flags; |
7b6a221e | 348 | btrfs_sync_inode_flags_to_i_flags(inode); |
0c4d2d95 | 349 | inode_inc_iversion(inode); |
c2050a45 | 350 | inode->i_ctime = current_time(inode); |
9a56fcd1 | 351 | ret = btrfs_update_inode(trans, root, BTRFS_I(inode)); |
6cbff00f | 352 | |
ff9fef55 | 353 | out_end_trans: |
3a45bb20 | 354 | btrfs_end_transaction(trans); |
6cbff00f | 355 | out_unlock: |
5955102c | 356 | inode_unlock(inode); |
e7848683 | 357 | mnt_drop_write_file(file); |
2d4e6f6a | 358 | return ret; |
6cbff00f CH |
359 | } |
360 | ||
19f93b3c DS |
361 | /* |
362 | * Translate btrfs internal inode flags to xflags as expected by the | |
363 | * FS_IOC_FSGETXATT ioctl. Filter only the supported ones, unknown flags are | |
364 | * silently dropped. | |
365 | */ | |
366 | static unsigned int btrfs_inode_flags_to_xflags(unsigned int flags) | |
367 | { | |
368 | unsigned int xflags = 0; | |
369 | ||
370 | if (flags & BTRFS_INODE_APPEND) | |
371 | xflags |= FS_XFLAG_APPEND; | |
372 | if (flags & BTRFS_INODE_IMMUTABLE) | |
373 | xflags |= FS_XFLAG_IMMUTABLE; | |
374 | if (flags & BTRFS_INODE_NOATIME) | |
375 | xflags |= FS_XFLAG_NOATIME; | |
376 | if (flags & BTRFS_INODE_NODUMP) | |
377 | xflags |= FS_XFLAG_NODUMP; | |
378 | if (flags & BTRFS_INODE_SYNC) | |
379 | xflags |= FS_XFLAG_SYNC; | |
380 | ||
381 | return xflags; | |
382 | } | |
383 | ||
384 | /* Check if @flags are a supported and valid set of FS_XFLAGS_* flags */ | |
385 | static int check_xflags(unsigned int flags) | |
386 | { | |
387 | if (flags & ~(FS_XFLAG_APPEND | FS_XFLAG_IMMUTABLE | FS_XFLAG_NOATIME | | |
388 | FS_XFLAG_NODUMP | FS_XFLAG_SYNC)) | |
389 | return -EOPNOTSUPP; | |
390 | return 0; | |
391 | } | |
392 | ||
c3e1f96c GR |
393 | bool btrfs_exclop_start(struct btrfs_fs_info *fs_info, |
394 | enum btrfs_exclusive_operation type) | |
395 | { | |
396 | return !cmpxchg(&fs_info->exclusive_operation, BTRFS_EXCLOP_NONE, type); | |
397 | } | |
398 | ||
399 | void btrfs_exclop_finish(struct btrfs_fs_info *fs_info) | |
400 | { | |
401 | WRITE_ONCE(fs_info->exclusive_operation, BTRFS_EXCLOP_NONE); | |
66a2823c | 402 | sysfs_notify(&fs_info->fs_devices->fsid_kobj, NULL, "exclusive_operation"); |
c3e1f96c GR |
403 | } |
404 | ||
e4202ac9 DS |
405 | /* |
406 | * Set the xflags from the internal inode flags. The remaining items of fsxattr | |
407 | * are zeroed. | |
408 | */ | |
409 | static int btrfs_ioctl_fsgetxattr(struct file *file, void __user *arg) | |
410 | { | |
411 | struct btrfs_inode *binode = BTRFS_I(file_inode(file)); | |
412 | struct fsxattr fa; | |
413 | ||
7b0e492e | 414 | simple_fill_fsxattr(&fa, btrfs_inode_flags_to_xflags(binode->flags)); |
e4202ac9 DS |
415 | if (copy_to_user(arg, &fa, sizeof(fa))) |
416 | return -EFAULT; | |
417 | ||
418 | return 0; | |
419 | } | |
420 | ||
025f2121 DS |
421 | static int btrfs_ioctl_fssetxattr(struct file *file, void __user *arg) |
422 | { | |
423 | struct inode *inode = file_inode(file); | |
424 | struct btrfs_inode *binode = BTRFS_I(inode); | |
425 | struct btrfs_root *root = binode->root; | |
426 | struct btrfs_trans_handle *trans; | |
7b0e492e | 427 | struct fsxattr fa, old_fa; |
025f2121 DS |
428 | unsigned old_flags; |
429 | unsigned old_i_flags; | |
430 | int ret = 0; | |
431 | ||
432 | if (!inode_owner_or_capable(inode)) | |
433 | return -EPERM; | |
434 | ||
435 | if (btrfs_root_readonly(root)) | |
436 | return -EROFS; | |
437 | ||
025f2121 DS |
438 | if (copy_from_user(&fa, arg, sizeof(fa))) |
439 | return -EFAULT; | |
440 | ||
441 | ret = check_xflags(fa.fsx_xflags); | |
442 | if (ret) | |
443 | return ret; | |
444 | ||
445 | if (fa.fsx_extsize != 0 || fa.fsx_projid != 0 || fa.fsx_cowextsize != 0) | |
446 | return -EOPNOTSUPP; | |
447 | ||
448 | ret = mnt_want_write_file(file); | |
449 | if (ret) | |
450 | return ret; | |
451 | ||
452 | inode_lock(inode); | |
453 | ||
454 | old_flags = binode->flags; | |
455 | old_i_flags = inode->i_flags; | |
456 | ||
7b0e492e DW |
457 | simple_fill_fsxattr(&old_fa, |
458 | btrfs_inode_flags_to_xflags(binode->flags)); | |
459 | ret = vfs_ioc_fssetxattr_check(inode, &old_fa, &fa); | |
460 | if (ret) | |
025f2121 | 461 | goto out_unlock; |
025f2121 DS |
462 | |
463 | if (fa.fsx_xflags & FS_XFLAG_SYNC) | |
464 | binode->flags |= BTRFS_INODE_SYNC; | |
465 | else | |
466 | binode->flags &= ~BTRFS_INODE_SYNC; | |
467 | if (fa.fsx_xflags & FS_XFLAG_IMMUTABLE) | |
468 | binode->flags |= BTRFS_INODE_IMMUTABLE; | |
469 | else | |
470 | binode->flags &= ~BTRFS_INODE_IMMUTABLE; | |
471 | if (fa.fsx_xflags & FS_XFLAG_APPEND) | |
472 | binode->flags |= BTRFS_INODE_APPEND; | |
473 | else | |
474 | binode->flags &= ~BTRFS_INODE_APPEND; | |
475 | if (fa.fsx_xflags & FS_XFLAG_NODUMP) | |
476 | binode->flags |= BTRFS_INODE_NODUMP; | |
477 | else | |
478 | binode->flags &= ~BTRFS_INODE_NODUMP; | |
479 | if (fa.fsx_xflags & FS_XFLAG_NOATIME) | |
480 | binode->flags |= BTRFS_INODE_NOATIME; | |
481 | else | |
482 | binode->flags &= ~BTRFS_INODE_NOATIME; | |
483 | ||
484 | /* 1 item for the inode */ | |
485 | trans = btrfs_start_transaction(root, 1); | |
486 | if (IS_ERR(trans)) { | |
487 | ret = PTR_ERR(trans); | |
488 | goto out_unlock; | |
489 | } | |
490 | ||
491 | btrfs_sync_inode_flags_to_i_flags(inode); | |
492 | inode_inc_iversion(inode); | |
493 | inode->i_ctime = current_time(inode); | |
9a56fcd1 | 494 | ret = btrfs_update_inode(trans, root, BTRFS_I(inode)); |
025f2121 DS |
495 | |
496 | btrfs_end_transaction(trans); | |
497 | ||
498 | out_unlock: | |
499 | if (ret) { | |
500 | binode->flags = old_flags; | |
501 | inode->i_flags = old_i_flags; | |
502 | } | |
503 | ||
504 | inode_unlock(inode); | |
505 | mnt_drop_write_file(file); | |
506 | ||
507 | return ret; | |
508 | } | |
509 | ||
6cbff00f CH |
510 | static int btrfs_ioctl_getversion(struct file *file, int __user *arg) |
511 | { | |
496ad9aa | 512 | struct inode *inode = file_inode(file); |
6cbff00f CH |
513 | |
514 | return put_user(inode->i_generation, arg); | |
515 | } | |
f46b5a66 | 516 | |
b929c1d8 MPS |
517 | static noinline int btrfs_ioctl_fitrim(struct btrfs_fs_info *fs_info, |
518 | void __user *arg) | |
f7039b1d | 519 | { |
f7039b1d LD |
520 | struct btrfs_device *device; |
521 | struct request_queue *q; | |
522 | struct fstrim_range range; | |
523 | u64 minlen = ULLONG_MAX; | |
524 | u64 num_devices = 0; | |
525 | int ret; | |
526 | ||
527 | if (!capable(CAP_SYS_ADMIN)) | |
528 | return -EPERM; | |
529 | ||
f35f06c3 FM |
530 | /* |
531 | * If the fs is mounted with nologreplay, which requires it to be | |
532 | * mounted in RO mode as well, we can not allow discard on free space | |
533 | * inside block groups, because log trees refer to extents that are not | |
534 | * pinned in a block group's free space cache (pinning the extents is | |
535 | * precisely the first phase of replaying a log tree). | |
536 | */ | |
537 | if (btrfs_test_opt(fs_info, NOLOGREPLAY)) | |
538 | return -EROFS; | |
539 | ||
1f78160c XG |
540 | rcu_read_lock(); |
541 | list_for_each_entry_rcu(device, &fs_info->fs_devices->devices, | |
542 | dev_list) { | |
f7039b1d LD |
543 | if (!device->bdev) |
544 | continue; | |
545 | q = bdev_get_queue(device->bdev); | |
546 | if (blk_queue_discard(q)) { | |
547 | num_devices++; | |
50d0446e | 548 | minlen = min_t(u64, q->limits.discard_granularity, |
f7039b1d LD |
549 | minlen); |
550 | } | |
551 | } | |
1f78160c | 552 | rcu_read_unlock(); |
f4c697e6 | 553 | |
f7039b1d LD |
554 | if (!num_devices) |
555 | return -EOPNOTSUPP; | |
f7039b1d LD |
556 | if (copy_from_user(&range, arg, sizeof(range))) |
557 | return -EFAULT; | |
6ba9fc8e QW |
558 | |
559 | /* | |
560 | * NOTE: Don't truncate the range using super->total_bytes. Bytenr of | |
561 | * block group is in the logical address space, which can be any | |
562 | * sectorsize aligned bytenr in the range [0, U64_MAX]. | |
563 | */ | |
564 | if (range.len < fs_info->sb->s_blocksize) | |
f4c697e6 | 565 | return -EINVAL; |
f7039b1d LD |
566 | |
567 | range.minlen = max(range.minlen, minlen); | |
2ff7e61e | 568 | ret = btrfs_trim_fs(fs_info, &range); |
f7039b1d LD |
569 | if (ret < 0) |
570 | return ret; | |
571 | ||
572 | if (copy_to_user(arg, &range, sizeof(range))) | |
573 | return -EFAULT; | |
574 | ||
575 | return 0; | |
576 | } | |
577 | ||
e1f60a65 | 578 | int __pure btrfs_is_empty_uuid(u8 *uuid) |
dd5f9615 | 579 | { |
46e0f66a CM |
580 | int i; |
581 | ||
582 | for (i = 0; i < BTRFS_UUID_SIZE; i++) { | |
583 | if (uuid[i]) | |
584 | return 0; | |
585 | } | |
586 | return 1; | |
dd5f9615 SB |
587 | } |
588 | ||
d5c12070 | 589 | static noinline int create_subvol(struct inode *dir, |
cb8e7090 | 590 | struct dentry *dentry, |
52f75f4f | 591 | const char *name, int namelen, |
8696c533 | 592 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 593 | { |
0b246afa | 594 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); |
f46b5a66 CH |
595 | struct btrfs_trans_handle *trans; |
596 | struct btrfs_key key; | |
49a3c4d9 | 597 | struct btrfs_root_item *root_item; |
f46b5a66 CH |
598 | struct btrfs_inode_item *inode_item; |
599 | struct extent_buffer *leaf; | |
d5c12070 | 600 | struct btrfs_root *root = BTRFS_I(dir)->root; |
76dda93c | 601 | struct btrfs_root *new_root; |
d5c12070 | 602 | struct btrfs_block_rsv block_rsv; |
95582b00 | 603 | struct timespec64 cur_time = current_time(dir); |
5662344b | 604 | struct inode *inode; |
f46b5a66 CH |
605 | int ret; |
606 | int err; | |
2dfb1e43 | 607 | dev_t anon_dev = 0; |
f46b5a66 CH |
608 | u64 objectid; |
609 | u64 new_dirid = BTRFS_FIRST_FREE_OBJECTID; | |
3de4586c | 610 | u64 index = 0; |
f46b5a66 | 611 | |
49a3c4d9 DS |
612 | root_item = kzalloc(sizeof(*root_item), GFP_KERNEL); |
613 | if (!root_item) | |
614 | return -ENOMEM; | |
615 | ||
0b246afa | 616 | ret = btrfs_find_free_objectid(fs_info->tree_root, &objectid); |
2fbe8c8a | 617 | if (ret) |
49a3c4d9 | 618 | goto fail_free; |
6a912213 | 619 | |
2dfb1e43 QW |
620 | ret = get_anon_bdev(&anon_dev); |
621 | if (ret < 0) | |
622 | goto fail_free; | |
623 | ||
e09fe2d2 QW |
624 | /* |
625 | * Don't create subvolume whose level is not zero. Or qgroup will be | |
01327610 | 626 | * screwed up since it assumes subvolume qgroup's level to be 0. |
e09fe2d2 | 627 | */ |
49a3c4d9 DS |
628 | if (btrfs_qgroup_level(objectid)) { |
629 | ret = -ENOSPC; | |
630 | goto fail_free; | |
631 | } | |
e09fe2d2 | 632 | |
d5c12070 | 633 | btrfs_init_block_rsv(&block_rsv, BTRFS_BLOCK_RSV_TEMP); |
9ed74f2d | 634 | /* |
d5c12070 MX |
635 | * The same as the snapshot creation, please see the comment |
636 | * of create_snapshot(). | |
9ed74f2d | 637 | */ |
c4c129db | 638 | ret = btrfs_subvolume_reserve_metadata(root, &block_rsv, 8, false); |
d5c12070 | 639 | if (ret) |
49a3c4d9 | 640 | goto fail_free; |
d5c12070 MX |
641 | |
642 | trans = btrfs_start_transaction(root, 0); | |
643 | if (IS_ERR(trans)) { | |
644 | ret = PTR_ERR(trans); | |
e85fde51 | 645 | btrfs_subvolume_release_metadata(root, &block_rsv); |
49a3c4d9 | 646 | goto fail_free; |
d5c12070 MX |
647 | } |
648 | trans->block_rsv = &block_rsv; | |
649 | trans->bytes_reserved = block_rsv.size; | |
f46b5a66 | 650 | |
a9377422 | 651 | ret = btrfs_qgroup_inherit(trans, 0, objectid, inherit); |
6f72c7e2 AJ |
652 | if (ret) |
653 | goto fail; | |
654 | ||
9631e4cc JB |
655 | leaf = btrfs_alloc_tree_block(trans, root, 0, objectid, NULL, 0, 0, 0, |
656 | BTRFS_NESTING_NORMAL); | |
8e8a1e31 JB |
657 | if (IS_ERR(leaf)) { |
658 | ret = PTR_ERR(leaf); | |
659 | goto fail; | |
660 | } | |
f46b5a66 | 661 | |
f46b5a66 CH |
662 | btrfs_mark_buffer_dirty(leaf); |
663 | ||
49a3c4d9 | 664 | inode_item = &root_item->inode; |
3cae210f QW |
665 | btrfs_set_stack_inode_generation(inode_item, 1); |
666 | btrfs_set_stack_inode_size(inode_item, 3); | |
667 | btrfs_set_stack_inode_nlink(inode_item, 1); | |
da17066c | 668 | btrfs_set_stack_inode_nbytes(inode_item, |
0b246afa | 669 | fs_info->nodesize); |
3cae210f | 670 | btrfs_set_stack_inode_mode(inode_item, S_IFDIR | 0755); |
f46b5a66 | 671 | |
49a3c4d9 DS |
672 | btrfs_set_root_flags(root_item, 0); |
673 | btrfs_set_root_limit(root_item, 0); | |
3cae210f | 674 | btrfs_set_stack_inode_flags(inode_item, BTRFS_INODE_ROOT_ITEM_INIT); |
08fe4db1 | 675 | |
49a3c4d9 DS |
676 | btrfs_set_root_bytenr(root_item, leaf->start); |
677 | btrfs_set_root_generation(root_item, trans->transid); | |
678 | btrfs_set_root_level(root_item, 0); | |
679 | btrfs_set_root_refs(root_item, 1); | |
680 | btrfs_set_root_used(root_item, leaf->len); | |
681 | btrfs_set_root_last_snapshot(root_item, 0); | |
f46b5a66 | 682 | |
49a3c4d9 DS |
683 | btrfs_set_root_generation_v2(root_item, |
684 | btrfs_root_generation(root_item)); | |
807fc790 | 685 | generate_random_guid(root_item->uuid); |
49a3c4d9 DS |
686 | btrfs_set_stack_timespec_sec(&root_item->otime, cur_time.tv_sec); |
687 | btrfs_set_stack_timespec_nsec(&root_item->otime, cur_time.tv_nsec); | |
688 | root_item->ctime = root_item->otime; | |
689 | btrfs_set_root_ctransid(root_item, trans->transid); | |
690 | btrfs_set_root_otransid(root_item, trans->transid); | |
f46b5a66 | 691 | |
925baedd | 692 | btrfs_tree_unlock(leaf); |
f46b5a66 CH |
693 | free_extent_buffer(leaf); |
694 | leaf = NULL; | |
695 | ||
49a3c4d9 | 696 | btrfs_set_root_dirid(root_item, new_dirid); |
f46b5a66 CH |
697 | |
698 | key.objectid = objectid; | |
5d4f98a2 | 699 | key.offset = 0; |
962a298f | 700 | key.type = BTRFS_ROOT_ITEM_KEY; |
0b246afa | 701 | ret = btrfs_insert_root(trans, fs_info->tree_root, &key, |
49a3c4d9 | 702 | root_item); |
f46b5a66 CH |
703 | if (ret) |
704 | goto fail; | |
705 | ||
76dda93c | 706 | key.offset = (u64)-1; |
2dfb1e43 | 707 | new_root = btrfs_get_new_fs_root(fs_info, objectid, anon_dev); |
79787eaa | 708 | if (IS_ERR(new_root)) { |
2dfb1e43 | 709 | free_anon_bdev(anon_dev); |
79787eaa | 710 | ret = PTR_ERR(new_root); |
66642832 | 711 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
712 | goto fail; |
713 | } | |
2dfb1e43 QW |
714 | /* Freeing will be done in btrfs_put_root() of new_root */ |
715 | anon_dev = 0; | |
76dda93c YZ |
716 | |
717 | btrfs_record_root_in_trans(trans, new_root); | |
718 | ||
63541927 | 719 | ret = btrfs_create_subvol_root(trans, new_root, root, new_dirid); |
00246528 | 720 | btrfs_put_root(new_root); |
ce598979 MF |
721 | if (ret) { |
722 | /* We potentially lose an unused inode item here */ | |
66642832 | 723 | btrfs_abort_transaction(trans, ret); |
ce598979 MF |
724 | goto fail; |
725 | } | |
726 | ||
f32e48e9 CR |
727 | mutex_lock(&new_root->objectid_mutex); |
728 | new_root->highest_objectid = new_dirid; | |
729 | mutex_unlock(&new_root->objectid_mutex); | |
730 | ||
f46b5a66 CH |
731 | /* |
732 | * insert the directory item | |
733 | */ | |
877574e2 | 734 | ret = btrfs_set_inode_index(BTRFS_I(dir), &index); |
79787eaa | 735 | if (ret) { |
66642832 | 736 | btrfs_abort_transaction(trans, ret); |
79787eaa JM |
737 | goto fail; |
738 | } | |
3de4586c | 739 | |
684572df | 740 | ret = btrfs_insert_dir_item(trans, name, namelen, BTRFS_I(dir), &key, |
3de4586c | 741 | BTRFS_FT_DIR, index); |
79787eaa | 742 | if (ret) { |
66642832 | 743 | btrfs_abort_transaction(trans, ret); |
f46b5a66 | 744 | goto fail; |
79787eaa | 745 | } |
0660b5af | 746 | |
6ef06d27 | 747 | btrfs_i_size_write(BTRFS_I(dir), dir->i_size + namelen * 2); |
9a56fcd1 | 748 | ret = btrfs_update_inode(trans, root, BTRFS_I(dir)); |
c7e54b51 JB |
749 | if (ret) { |
750 | btrfs_abort_transaction(trans, ret); | |
751 | goto fail; | |
752 | } | |
52c26179 | 753 | |
6025c19f | 754 | ret = btrfs_add_root_ref(trans, objectid, root->root_key.objectid, |
4a0cc7ca | 755 | btrfs_ino(BTRFS_I(dir)), index, name, namelen); |
c7e54b51 JB |
756 | if (ret) { |
757 | btrfs_abort_transaction(trans, ret); | |
758 | goto fail; | |
759 | } | |
f46b5a66 | 760 | |
cdb345a8 | 761 | ret = btrfs_uuid_tree_add(trans, root_item->uuid, |
6bccf3ab | 762 | BTRFS_UUID_KEY_SUBVOL, objectid); |
dd5f9615 | 763 | if (ret) |
66642832 | 764 | btrfs_abort_transaction(trans, ret); |
dd5f9615 | 765 | |
f46b5a66 | 766 | fail: |
49a3c4d9 | 767 | kfree(root_item); |
d5c12070 MX |
768 | trans->block_rsv = NULL; |
769 | trans->bytes_reserved = 0; | |
e85fde51 | 770 | btrfs_subvolume_release_metadata(root, &block_rsv); |
de6e8200 | 771 | |
9babda9f | 772 | err = btrfs_commit_transaction(trans); |
f46b5a66 CH |
773 | if (err && !ret) |
774 | ret = err; | |
1a65e24b | 775 | |
5662344b TI |
776 | if (!ret) { |
777 | inode = btrfs_lookup_dentry(dir, dentry); | |
de6e8200 LB |
778 | if (IS_ERR(inode)) |
779 | return PTR_ERR(inode); | |
5662344b TI |
780 | d_instantiate(dentry, inode); |
781 | } | |
f46b5a66 | 782 | return ret; |
49a3c4d9 DS |
783 | |
784 | fail_free: | |
2dfb1e43 QW |
785 | if (anon_dev) |
786 | free_anon_bdev(anon_dev); | |
49a3c4d9 DS |
787 | kfree(root_item); |
788 | return ret; | |
f46b5a66 CH |
789 | } |
790 | ||
e9662f70 | 791 | static int create_snapshot(struct btrfs_root *root, struct inode *dir, |
9babda9f | 792 | struct dentry *dentry, bool readonly, |
e9662f70 | 793 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 794 | { |
0b246afa | 795 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); |
2e4bfab9 | 796 | struct inode *inode; |
f46b5a66 CH |
797 | struct btrfs_pending_snapshot *pending_snapshot; |
798 | struct btrfs_trans_handle *trans; | |
2e4bfab9 | 799 | int ret; |
f46b5a66 | 800 | |
92a7cc42 | 801 | if (!test_bit(BTRFS_ROOT_SHAREABLE, &root->state)) |
f46b5a66 CH |
802 | return -EINVAL; |
803 | ||
eede2bf3 OS |
804 | if (atomic_read(&root->nr_swapfiles)) { |
805 | btrfs_warn(fs_info, | |
806 | "cannot snapshot subvolume with active swapfile"); | |
807 | return -ETXTBSY; | |
808 | } | |
809 | ||
23269bf5 | 810 | pending_snapshot = kzalloc(sizeof(*pending_snapshot), GFP_KERNEL); |
a1ee7362 DS |
811 | if (!pending_snapshot) |
812 | return -ENOMEM; | |
813 | ||
2dfb1e43 QW |
814 | ret = get_anon_bdev(&pending_snapshot->anon_dev); |
815 | if (ret < 0) | |
816 | goto free_pending; | |
b0c0ea63 | 817 | pending_snapshot->root_item = kzalloc(sizeof(struct btrfs_root_item), |
23269bf5 | 818 | GFP_KERNEL); |
8546b570 DS |
819 | pending_snapshot->path = btrfs_alloc_path(); |
820 | if (!pending_snapshot->root_item || !pending_snapshot->path) { | |
b0c0ea63 DS |
821 | ret = -ENOMEM; |
822 | goto free_pending; | |
823 | } | |
824 | ||
66d8f3dd MX |
825 | btrfs_init_block_rsv(&pending_snapshot->block_rsv, |
826 | BTRFS_BLOCK_RSV_TEMP); | |
d5c12070 MX |
827 | /* |
828 | * 1 - parent dir inode | |
829 | * 2 - dir entries | |
830 | * 1 - root item | |
831 | * 2 - root ref/backref | |
832 | * 1 - root of snapshot | |
dd5f9615 | 833 | * 1 - UUID item |
d5c12070 MX |
834 | */ |
835 | ret = btrfs_subvolume_reserve_metadata(BTRFS_I(dir)->root, | |
dd5f9615 | 836 | &pending_snapshot->block_rsv, 8, |
ee3441b4 | 837 | false); |
d5c12070 | 838 | if (ret) |
c11fbb6e | 839 | goto free_pending; |
d5c12070 | 840 | |
3de4586c | 841 | pending_snapshot->dentry = dentry; |
f46b5a66 | 842 | pending_snapshot->root = root; |
b83cc969 | 843 | pending_snapshot->readonly = readonly; |
e9662f70 | 844 | pending_snapshot->dir = dir; |
8696c533 | 845 | pending_snapshot->inherit = inherit; |
a22285a6 | 846 | |
d5c12070 | 847 | trans = btrfs_start_transaction(root, 0); |
a22285a6 YZ |
848 | if (IS_ERR(trans)) { |
849 | ret = PTR_ERR(trans); | |
850 | goto fail; | |
851 | } | |
852 | ||
0b246afa | 853 | spin_lock(&fs_info->trans_lock); |
f46b5a66 CH |
854 | list_add(&pending_snapshot->list, |
855 | &trans->transaction->pending_snapshots); | |
0b246afa | 856 | spin_unlock(&fs_info->trans_lock); |
9babda9f NB |
857 | |
858 | ret = btrfs_commit_transaction(trans); | |
aec8030a | 859 | if (ret) |
c37b2b62 | 860 | goto fail; |
a22285a6 YZ |
861 | |
862 | ret = pending_snapshot->error; | |
863 | if (ret) | |
864 | goto fail; | |
865 | ||
d3797308 CM |
866 | ret = btrfs_orphan_cleanup(pending_snapshot->snap); |
867 | if (ret) | |
868 | goto fail; | |
869 | ||
2b0143b5 | 870 | inode = btrfs_lookup_dentry(d_inode(dentry->d_parent), dentry); |
2e4bfab9 YZ |
871 | if (IS_ERR(inode)) { |
872 | ret = PTR_ERR(inode); | |
873 | goto fail; | |
874 | } | |
5662344b | 875 | |
2e4bfab9 YZ |
876 | d_instantiate(dentry, inode); |
877 | ret = 0; | |
2dfb1e43 | 878 | pending_snapshot->anon_dev = 0; |
2e4bfab9 | 879 | fail: |
2dfb1e43 QW |
880 | /* Prevent double freeing of anon_dev */ |
881 | if (ret && pending_snapshot->snap) | |
882 | pending_snapshot->snap->anon_dev = 0; | |
00246528 | 883 | btrfs_put_root(pending_snapshot->snap); |
e85fde51 | 884 | btrfs_subvolume_release_metadata(root, &pending_snapshot->block_rsv); |
b0c0ea63 | 885 | free_pending: |
2dfb1e43 QW |
886 | if (pending_snapshot->anon_dev) |
887 | free_anon_bdev(pending_snapshot->anon_dev); | |
b0c0ea63 | 888 | kfree(pending_snapshot->root_item); |
8546b570 | 889 | btrfs_free_path(pending_snapshot->path); |
a1ee7362 DS |
890 | kfree(pending_snapshot); |
891 | ||
f46b5a66 CH |
892 | return ret; |
893 | } | |
894 | ||
4260f7c7 SW |
895 | /* copy of may_delete in fs/namei.c() |
896 | * Check whether we can remove a link victim from directory dir, check | |
897 | * whether the type of victim is right. | |
898 | * 1. We can't do it if dir is read-only (done in permission()) | |
899 | * 2. We should have write and exec permissions on dir | |
900 | * 3. We can't remove anything from append-only dir | |
901 | * 4. We can't do anything with immutable dir (done in permission()) | |
902 | * 5. If the sticky bit on dir is set we should either | |
903 | * a. be owner of dir, or | |
904 | * b. be owner of victim, or | |
905 | * c. have CAP_FOWNER capability | |
01327610 | 906 | * 6. If the victim is append-only or immutable we can't do anything with |
4260f7c7 SW |
907 | * links pointing to it. |
908 | * 7. If we were asked to remove a directory and victim isn't one - ENOTDIR. | |
909 | * 8. If we were asked to remove a non-directory and victim isn't one - EISDIR. | |
910 | * 9. We can't remove a root or mountpoint. | |
911 | * 10. We don't allow removal of NFS sillyrenamed files; it's handled by | |
912 | * nfs_async_unlink(). | |
913 | */ | |
914 | ||
67871254 | 915 | static int btrfs_may_delete(struct inode *dir, struct dentry *victim, int isdir) |
4260f7c7 SW |
916 | { |
917 | int error; | |
918 | ||
2b0143b5 | 919 | if (d_really_is_negative(victim)) |
4260f7c7 SW |
920 | return -ENOENT; |
921 | ||
2b0143b5 | 922 | BUG_ON(d_inode(victim->d_parent) != dir); |
4fa6b5ec | 923 | audit_inode_child(dir, victim, AUDIT_TYPE_CHILD_DELETE); |
4260f7c7 SW |
924 | |
925 | error = inode_permission(dir, MAY_WRITE | MAY_EXEC); | |
926 | if (error) | |
927 | return error; | |
928 | if (IS_APPEND(dir)) | |
929 | return -EPERM; | |
2b0143b5 DH |
930 | if (check_sticky(dir, d_inode(victim)) || IS_APPEND(d_inode(victim)) || |
931 | IS_IMMUTABLE(d_inode(victim)) || IS_SWAPFILE(d_inode(victim))) | |
4260f7c7 SW |
932 | return -EPERM; |
933 | if (isdir) { | |
e36cb0b8 | 934 | if (!d_is_dir(victim)) |
4260f7c7 SW |
935 | return -ENOTDIR; |
936 | if (IS_ROOT(victim)) | |
937 | return -EBUSY; | |
e36cb0b8 | 938 | } else if (d_is_dir(victim)) |
4260f7c7 SW |
939 | return -EISDIR; |
940 | if (IS_DEADDIR(dir)) | |
941 | return -ENOENT; | |
942 | if (victim->d_flags & DCACHE_NFSFS_RENAMED) | |
943 | return -EBUSY; | |
944 | return 0; | |
945 | } | |
946 | ||
cb8e7090 CH |
947 | /* copy of may_create in fs/namei.c() */ |
948 | static inline int btrfs_may_create(struct inode *dir, struct dentry *child) | |
949 | { | |
2b0143b5 | 950 | if (d_really_is_positive(child)) |
cb8e7090 CH |
951 | return -EEXIST; |
952 | if (IS_DEADDIR(dir)) | |
953 | return -ENOENT; | |
954 | return inode_permission(dir, MAY_WRITE | MAY_EXEC); | |
955 | } | |
956 | ||
957 | /* | |
958 | * Create a new subvolume below @parent. This is largely modeled after | |
959 | * sys_mkdirat and vfs_mkdir, but we only do a single component lookup | |
960 | * inside this filesystem so it's quite a bit simpler. | |
961 | */ | |
92872094 | 962 | static noinline int btrfs_mksubvol(const struct path *parent, |
52f75f4f | 963 | const char *name, int namelen, |
72fd032e | 964 | struct btrfs_root *snap_src, |
9babda9f | 965 | bool readonly, |
8696c533 | 966 | struct btrfs_qgroup_inherit *inherit) |
cb8e7090 | 967 | { |
0b246afa JM |
968 | struct inode *dir = d_inode(parent->dentry); |
969 | struct btrfs_fs_info *fs_info = btrfs_sb(dir->i_sb); | |
cb8e7090 CH |
970 | struct dentry *dentry; |
971 | int error; | |
972 | ||
00235411 AV |
973 | error = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT); |
974 | if (error == -EINTR) | |
975 | return error; | |
cb8e7090 CH |
976 | |
977 | dentry = lookup_one_len(name, parent->dentry, namelen); | |
978 | error = PTR_ERR(dentry); | |
979 | if (IS_ERR(dentry)) | |
980 | goto out_unlock; | |
981 | ||
76dda93c | 982 | error = btrfs_may_create(dir, dentry); |
cb8e7090 | 983 | if (error) |
a874a63e | 984 | goto out_dput; |
cb8e7090 | 985 | |
9c52057c CM |
986 | /* |
987 | * even if this name doesn't exist, we may get hash collisions. | |
988 | * check for them now when we can safely fail | |
989 | */ | |
990 | error = btrfs_check_dir_item_collision(BTRFS_I(dir)->root, | |
991 | dir->i_ino, name, | |
992 | namelen); | |
993 | if (error) | |
994 | goto out_dput; | |
995 | ||
0b246afa | 996 | down_read(&fs_info->subvol_sem); |
76dda93c YZ |
997 | |
998 | if (btrfs_root_refs(&BTRFS_I(dir)->root->root_item) == 0) | |
999 | goto out_up_read; | |
1000 | ||
9babda9f NB |
1001 | if (snap_src) |
1002 | error = create_snapshot(snap_src, dir, dentry, readonly, inherit); | |
1003 | else | |
1004 | error = create_subvol(dir, dentry, name, namelen, inherit); | |
1005 | ||
76dda93c YZ |
1006 | if (!error) |
1007 | fsnotify_mkdir(dir, dentry); | |
1008 | out_up_read: | |
0b246afa | 1009 | up_read(&fs_info->subvol_sem); |
cb8e7090 CH |
1010 | out_dput: |
1011 | dput(dentry); | |
1012 | out_unlock: | |
5955102c | 1013 | inode_unlock(dir); |
cb8e7090 CH |
1014 | return error; |
1015 | } | |
1016 | ||
c11fbb6e RK |
1017 | static noinline int btrfs_mksnapshot(const struct path *parent, |
1018 | const char *name, int namelen, | |
1019 | struct btrfs_root *root, | |
1020 | bool readonly, | |
1021 | struct btrfs_qgroup_inherit *inherit) | |
1022 | { | |
1023 | int ret; | |
1024 | bool snapshot_force_cow = false; | |
1025 | ||
1026 | /* | |
1027 | * Force new buffered writes to reserve space even when NOCOW is | |
1028 | * possible. This is to avoid later writeback (running dealloc) to | |
1029 | * fallback to COW mode and unexpectedly fail with ENOSPC. | |
1030 | */ | |
1031 | btrfs_drew_read_lock(&root->snapshot_lock); | |
1032 | ||
1033 | ret = btrfs_start_delalloc_snapshot(root); | |
1034 | if (ret) | |
1035 | goto out; | |
1036 | ||
1037 | /* | |
1038 | * All previous writes have started writeback in NOCOW mode, so now | |
1039 | * we force future writes to fallback to COW mode during snapshot | |
1040 | * creation. | |
1041 | */ | |
1042 | atomic_inc(&root->snapshot_force_cow); | |
1043 | snapshot_force_cow = true; | |
1044 | ||
1045 | btrfs_wait_ordered_extents(root, U64_MAX, 0, (u64)-1); | |
1046 | ||
1047 | ret = btrfs_mksubvol(parent, name, namelen, | |
1048 | root, readonly, inherit); | |
1049 | out: | |
1050 | if (snapshot_force_cow) | |
1051 | atomic_dec(&root->snapshot_force_cow); | |
1052 | btrfs_drew_read_unlock(&root->snapshot_lock); | |
1053 | return ret; | |
1054 | } | |
1055 | ||
4cb5300b CM |
1056 | /* |
1057 | * When we're defragging a range, we don't want to kick it off again | |
1058 | * if it is really just waiting for delalloc to send it down. | |
1059 | * If we find a nice big extent or delalloc range for the bytes in the | |
1060 | * file you want to defrag, we return 0 to let you know to skip this | |
1061 | * part of the file | |
1062 | */ | |
aab110ab | 1063 | static int check_defrag_in_cache(struct inode *inode, u64 offset, u32 thresh) |
4cb5300b CM |
1064 | { |
1065 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; | |
1066 | struct extent_map *em = NULL; | |
1067 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
1068 | u64 end; | |
1069 | ||
1070 | read_lock(&em_tree->lock); | |
09cbfeaf | 1071 | em = lookup_extent_mapping(em_tree, offset, PAGE_SIZE); |
4cb5300b CM |
1072 | read_unlock(&em_tree->lock); |
1073 | ||
1074 | if (em) { | |
1075 | end = extent_map_end(em); | |
1076 | free_extent_map(em); | |
1077 | if (end - offset > thresh) | |
1078 | return 0; | |
1079 | } | |
1080 | /* if we already have a nice delalloc here, just stop */ | |
1081 | thresh /= 2; | |
1082 | end = count_range_bits(io_tree, &offset, offset + thresh, | |
1083 | thresh, EXTENT_DELALLOC, 1); | |
1084 | if (end >= thresh) | |
1085 | return 0; | |
1086 | return 1; | |
1087 | } | |
1088 | ||
1089 | /* | |
1090 | * helper function to walk through a file and find extents | |
1091 | * newer than a specific transid, and smaller than thresh. | |
1092 | * | |
1093 | * This is used by the defragging code to find new and small | |
1094 | * extents | |
1095 | */ | |
1096 | static int find_new_extents(struct btrfs_root *root, | |
1097 | struct inode *inode, u64 newer_than, | |
aab110ab | 1098 | u64 *off, u32 thresh) |
4cb5300b CM |
1099 | { |
1100 | struct btrfs_path *path; | |
1101 | struct btrfs_key min_key; | |
4cb5300b CM |
1102 | struct extent_buffer *leaf; |
1103 | struct btrfs_file_extent_item *extent; | |
1104 | int type; | |
1105 | int ret; | |
4a0cc7ca | 1106 | u64 ino = btrfs_ino(BTRFS_I(inode)); |
4cb5300b CM |
1107 | |
1108 | path = btrfs_alloc_path(); | |
1109 | if (!path) | |
1110 | return -ENOMEM; | |
1111 | ||
a4689d2b | 1112 | min_key.objectid = ino; |
4cb5300b CM |
1113 | min_key.type = BTRFS_EXTENT_DATA_KEY; |
1114 | min_key.offset = *off; | |
1115 | ||
67871254 | 1116 | while (1) { |
6174d3cb | 1117 | ret = btrfs_search_forward(root, &min_key, path, newer_than); |
4cb5300b CM |
1118 | if (ret != 0) |
1119 | goto none; | |
f094c9bd | 1120 | process_slot: |
a4689d2b | 1121 | if (min_key.objectid != ino) |
4cb5300b CM |
1122 | goto none; |
1123 | if (min_key.type != BTRFS_EXTENT_DATA_KEY) | |
1124 | goto none; | |
1125 | ||
1126 | leaf = path->nodes[0]; | |
1127 | extent = btrfs_item_ptr(leaf, path->slots[0], | |
1128 | struct btrfs_file_extent_item); | |
1129 | ||
1130 | type = btrfs_file_extent_type(leaf, extent); | |
1131 | if (type == BTRFS_FILE_EXTENT_REG && | |
1132 | btrfs_file_extent_num_bytes(leaf, extent) < thresh && | |
1133 | check_defrag_in_cache(inode, min_key.offset, thresh)) { | |
1134 | *off = min_key.offset; | |
1135 | btrfs_free_path(path); | |
1136 | return 0; | |
1137 | } | |
1138 | ||
f094c9bd FM |
1139 | path->slots[0]++; |
1140 | if (path->slots[0] < btrfs_header_nritems(leaf)) { | |
1141 | btrfs_item_key_to_cpu(leaf, &min_key, path->slots[0]); | |
1142 | goto process_slot; | |
1143 | } | |
1144 | ||
4cb5300b CM |
1145 | if (min_key.offset == (u64)-1) |
1146 | goto none; | |
1147 | ||
1148 | min_key.offset++; | |
1149 | btrfs_release_path(path); | |
1150 | } | |
1151 | none: | |
1152 | btrfs_free_path(path); | |
1153 | return -ENOENT; | |
1154 | } | |
1155 | ||
6c282eb4 | 1156 | static struct extent_map *defrag_lookup_extent(struct inode *inode, u64 start) |
17ce6ef8 LB |
1157 | { |
1158 | struct extent_map_tree *em_tree = &BTRFS_I(inode)->extent_tree; | |
6c282eb4 LZ |
1159 | struct extent_io_tree *io_tree = &BTRFS_I(inode)->io_tree; |
1160 | struct extent_map *em; | |
09cbfeaf | 1161 | u64 len = PAGE_SIZE; |
17ce6ef8 | 1162 | |
6c282eb4 LZ |
1163 | /* |
1164 | * hopefully we have this extent in the tree already, try without | |
1165 | * the full extent lock | |
1166 | */ | |
17ce6ef8 | 1167 | read_lock(&em_tree->lock); |
6c282eb4 | 1168 | em = lookup_extent_mapping(em_tree, start, len); |
17ce6ef8 LB |
1169 | read_unlock(&em_tree->lock); |
1170 | ||
6c282eb4 | 1171 | if (!em) { |
308d9800 FM |
1172 | struct extent_state *cached = NULL; |
1173 | u64 end = start + len - 1; | |
1174 | ||
6c282eb4 | 1175 | /* get the big lock and read metadata off disk */ |
ff13db41 | 1176 | lock_extent_bits(io_tree, start, end, &cached); |
39b07b5d | 1177 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, start, len); |
e43bbe5e | 1178 | unlock_extent_cached(io_tree, start, end, &cached); |
6c282eb4 LZ |
1179 | |
1180 | if (IS_ERR(em)) | |
1181 | return NULL; | |
1182 | } | |
1183 | ||
1184 | return em; | |
1185 | } | |
17ce6ef8 | 1186 | |
6c282eb4 LZ |
1187 | static bool defrag_check_next_extent(struct inode *inode, struct extent_map *em) |
1188 | { | |
1189 | struct extent_map *next; | |
1190 | bool ret = true; | |
1191 | ||
1192 | /* this is the last extent */ | |
1193 | if (em->start + em->len >= i_size_read(inode)) | |
1194 | return false; | |
1195 | ||
1196 | next = defrag_lookup_extent(inode, em->start + em->len); | |
e9512d72 CM |
1197 | if (!next || next->block_start >= EXTENT_MAP_LAST_BYTE) |
1198 | ret = false; | |
1199 | else if ((em->block_start + em->block_len == next->block_start) && | |
ee22184b | 1200 | (em->block_len > SZ_128K && next->block_len > SZ_128K)) |
6c282eb4 LZ |
1201 | ret = false; |
1202 | ||
1203 | free_extent_map(next); | |
17ce6ef8 LB |
1204 | return ret; |
1205 | } | |
1206 | ||
aab110ab | 1207 | static int should_defrag_range(struct inode *inode, u64 start, u32 thresh, |
a43a2111 AM |
1208 | u64 *last_len, u64 *skip, u64 *defrag_end, |
1209 | int compress) | |
940100a4 | 1210 | { |
6c282eb4 | 1211 | struct extent_map *em; |
940100a4 | 1212 | int ret = 1; |
6c282eb4 | 1213 | bool next_mergeable = true; |
4a3560c4 | 1214 | bool prev_mergeable = true; |
940100a4 CM |
1215 | |
1216 | /* | |
008873ea | 1217 | * make sure that once we start defragging an extent, we keep on |
940100a4 CM |
1218 | * defragging it |
1219 | */ | |
1220 | if (start < *defrag_end) | |
1221 | return 1; | |
1222 | ||
1223 | *skip = 0; | |
1224 | ||
6c282eb4 LZ |
1225 | em = defrag_lookup_extent(inode, start); |
1226 | if (!em) | |
1227 | return 0; | |
940100a4 CM |
1228 | |
1229 | /* this will cover holes, and inline extents */ | |
17ce6ef8 | 1230 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { |
940100a4 | 1231 | ret = 0; |
17ce6ef8 LB |
1232 | goto out; |
1233 | } | |
1234 | ||
4a3560c4 LB |
1235 | if (!*defrag_end) |
1236 | prev_mergeable = false; | |
1237 | ||
6c282eb4 | 1238 | next_mergeable = defrag_check_next_extent(inode, em); |
940100a4 | 1239 | /* |
6c282eb4 LZ |
1240 | * we hit a real extent, if it is big or the next extent is not a |
1241 | * real extent, don't bother defragging it | |
940100a4 | 1242 | */ |
a43a2111 | 1243 | if (!compress && (*last_len == 0 || *last_len >= thresh) && |
4a3560c4 | 1244 | (em->len >= thresh || (!next_mergeable && !prev_mergeable))) |
940100a4 | 1245 | ret = 0; |
17ce6ef8 | 1246 | out: |
940100a4 CM |
1247 | /* |
1248 | * last_len ends up being a counter of how many bytes we've defragged. | |
1249 | * every time we choose not to defrag an extent, we reset *last_len | |
1250 | * so that the next tiny extent will force a defrag. | |
1251 | * | |
1252 | * The end result of this is that tiny extents before a single big | |
1253 | * extent will force at least part of that big extent to be defragged. | |
1254 | */ | |
1255 | if (ret) { | |
940100a4 CM |
1256 | *defrag_end = extent_map_end(em); |
1257 | } else { | |
1258 | *last_len = 0; | |
1259 | *skip = extent_map_end(em); | |
1260 | *defrag_end = 0; | |
1261 | } | |
1262 | ||
1263 | free_extent_map(em); | |
1264 | return ret; | |
1265 | } | |
1266 | ||
4cb5300b CM |
1267 | /* |
1268 | * it doesn't do much good to defrag one or two pages | |
1269 | * at a time. This pulls in a nice chunk of pages | |
1270 | * to COW and defrag. | |
1271 | * | |
1272 | * It also makes sure the delalloc code has enough | |
1273 | * dirty data to avoid making new small extents as part | |
1274 | * of the defrag | |
1275 | * | |
1276 | * It's a good idea to start RA on this range | |
1277 | * before calling this. | |
1278 | */ | |
1279 | static int cluster_pages_for_defrag(struct inode *inode, | |
1280 | struct page **pages, | |
1281 | unsigned long start_index, | |
c41570c9 | 1282 | unsigned long num_pages) |
f46b5a66 | 1283 | { |
4cb5300b CM |
1284 | unsigned long file_end; |
1285 | u64 isize = i_size_read(inode); | |
1286 | u64 page_start; | |
1287 | u64 page_end; | |
1f12bd06 | 1288 | u64 page_cnt; |
a1fbc675 | 1289 | u64 start = (u64)start_index << PAGE_SHIFT; |
7f458a38 | 1290 | u64 search_start; |
4cb5300b CM |
1291 | int ret; |
1292 | int i; | |
1293 | int i_done; | |
3eaa2885 | 1294 | struct btrfs_ordered_extent *ordered; |
4cb5300b | 1295 | struct extent_state *cached_state = NULL; |
600a45e1 | 1296 | struct extent_io_tree *tree; |
364ecf36 | 1297 | struct extent_changeset *data_reserved = NULL; |
3b16a4e3 | 1298 | gfp_t mask = btrfs_alloc_write_mask(inode->i_mapping); |
4cb5300b | 1299 | |
09cbfeaf | 1300 | file_end = (isize - 1) >> PAGE_SHIFT; |
1f12bd06 LB |
1301 | if (!isize || start_index > file_end) |
1302 | return 0; | |
1303 | ||
1304 | page_cnt = min_t(u64, (u64)num_pages, (u64)file_end - start_index + 1); | |
4cb5300b | 1305 | |
e5b7231e | 1306 | ret = btrfs_delalloc_reserve_space(BTRFS_I(inode), &data_reserved, |
a1fbc675 | 1307 | start, page_cnt << PAGE_SHIFT); |
4cb5300b CM |
1308 | if (ret) |
1309 | return ret; | |
4cb5300b | 1310 | i_done = 0; |
600a45e1 | 1311 | tree = &BTRFS_I(inode)->io_tree; |
4cb5300b CM |
1312 | |
1313 | /* step one, lock all the pages */ | |
1f12bd06 | 1314 | for (i = 0; i < page_cnt; i++) { |
4cb5300b | 1315 | struct page *page; |
600a45e1 | 1316 | again: |
a94733d0 | 1317 | page = find_or_create_page(inode->i_mapping, |
600a45e1 | 1318 | start_index + i, mask); |
4cb5300b CM |
1319 | if (!page) |
1320 | break; | |
1321 | ||
600a45e1 | 1322 | page_start = page_offset(page); |
09cbfeaf | 1323 | page_end = page_start + PAGE_SIZE - 1; |
600a45e1 | 1324 | while (1) { |
308d9800 | 1325 | lock_extent_bits(tree, page_start, page_end, |
ff13db41 | 1326 | &cached_state); |
c3504372 | 1327 | ordered = btrfs_lookup_ordered_extent(BTRFS_I(inode), |
600a45e1 | 1328 | page_start); |
308d9800 | 1329 | unlock_extent_cached(tree, page_start, page_end, |
e43bbe5e | 1330 | &cached_state); |
600a45e1 MX |
1331 | if (!ordered) |
1332 | break; | |
1333 | ||
1334 | unlock_page(page); | |
c0a43603 | 1335 | btrfs_start_ordered_extent(ordered, 1); |
600a45e1 MX |
1336 | btrfs_put_ordered_extent(ordered); |
1337 | lock_page(page); | |
1f12bd06 LB |
1338 | /* |
1339 | * we unlocked the page above, so we need check if | |
1340 | * it was released or not. | |
1341 | */ | |
1342 | if (page->mapping != inode->i_mapping) { | |
1343 | unlock_page(page); | |
09cbfeaf | 1344 | put_page(page); |
1f12bd06 LB |
1345 | goto again; |
1346 | } | |
600a45e1 MX |
1347 | } |
1348 | ||
4cb5300b CM |
1349 | if (!PageUptodate(page)) { |
1350 | btrfs_readpage(NULL, page); | |
1351 | lock_page(page); | |
1352 | if (!PageUptodate(page)) { | |
1353 | unlock_page(page); | |
09cbfeaf | 1354 | put_page(page); |
4cb5300b CM |
1355 | ret = -EIO; |
1356 | break; | |
1357 | } | |
1358 | } | |
600a45e1 | 1359 | |
600a45e1 MX |
1360 | if (page->mapping != inode->i_mapping) { |
1361 | unlock_page(page); | |
09cbfeaf | 1362 | put_page(page); |
600a45e1 MX |
1363 | goto again; |
1364 | } | |
1365 | ||
4cb5300b CM |
1366 | pages[i] = page; |
1367 | i_done++; | |
1368 | } | |
1369 | if (!i_done || ret) | |
1370 | goto out; | |
1371 | ||
1751e8a6 | 1372 | if (!(inode->i_sb->s_flags & SB_ACTIVE)) |
4cb5300b CM |
1373 | goto out; |
1374 | ||
1375 | /* | |
1376 | * so now we have a nice long stream of locked | |
1377 | * and up to date pages, lets wait on them | |
1378 | */ | |
1379 | for (i = 0; i < i_done; i++) | |
1380 | wait_on_page_writeback(pages[i]); | |
1381 | ||
1382 | page_start = page_offset(pages[0]); | |
09cbfeaf | 1383 | page_end = page_offset(pages[i_done - 1]) + PAGE_SIZE; |
4cb5300b CM |
1384 | |
1385 | lock_extent_bits(&BTRFS_I(inode)->io_tree, | |
ff13db41 | 1386 | page_start, page_end - 1, &cached_state); |
7f458a38 FM |
1387 | |
1388 | /* | |
1389 | * When defragmenting we skip ranges that have holes or inline extents, | |
1390 | * (check should_defrag_range()), to avoid unnecessary IO and wasting | |
1391 | * space. At btrfs_defrag_file(), we check if a range should be defragged | |
1392 | * before locking the inode and then, if it should, we trigger a sync | |
1393 | * page cache readahead - we lock the inode only after that to avoid | |
1394 | * blocking for too long other tasks that possibly want to operate on | |
1395 | * other file ranges. But before we were able to get the inode lock, | |
1396 | * some other task may have punched a hole in the range, or we may have | |
1397 | * now an inline extent, in which case we should not defrag. So check | |
1398 | * for that here, where we have the inode and the range locked, and bail | |
1399 | * out if that happened. | |
1400 | */ | |
1401 | search_start = page_start; | |
1402 | while (search_start < page_end) { | |
1403 | struct extent_map *em; | |
1404 | ||
1405 | em = btrfs_get_extent(BTRFS_I(inode), NULL, 0, search_start, | |
1406 | page_end - search_start); | |
1407 | if (IS_ERR(em)) { | |
1408 | ret = PTR_ERR(em); | |
1409 | goto out_unlock_range; | |
1410 | } | |
1411 | if (em->block_start >= EXTENT_MAP_LAST_BYTE) { | |
1412 | free_extent_map(em); | |
1413 | /* Ok, 0 means we did not defrag anything */ | |
1414 | ret = 0; | |
1415 | goto out_unlock_range; | |
1416 | } | |
1417 | search_start = extent_map_end(em); | |
1418 | free_extent_map(em); | |
1419 | } | |
1420 | ||
4cb5300b | 1421 | clear_extent_bit(&BTRFS_I(inode)->io_tree, page_start, |
e182163d OS |
1422 | page_end - 1, EXTENT_DELALLOC | EXTENT_DO_ACCOUNTING | |
1423 | EXTENT_DEFRAG, 0, 0, &cached_state); | |
4cb5300b | 1424 | |
1f12bd06 | 1425 | if (i_done != page_cnt) { |
9e0baf60 | 1426 | spin_lock(&BTRFS_I(inode)->lock); |
28c4a3e2 | 1427 | btrfs_mod_outstanding_extents(BTRFS_I(inode), 1); |
9e0baf60 | 1428 | spin_unlock(&BTRFS_I(inode)->lock); |
86d52921 | 1429 | btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, |
a1fbc675 | 1430 | start, (page_cnt - i_done) << PAGE_SHIFT, true); |
4cb5300b CM |
1431 | } |
1432 | ||
1433 | ||
9e8a4a8b | 1434 | set_extent_defrag(&BTRFS_I(inode)->io_tree, page_start, page_end - 1, |
018ed4f7 | 1435 | &cached_state); |
4cb5300b CM |
1436 | |
1437 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
e43bbe5e | 1438 | page_start, page_end - 1, &cached_state); |
4cb5300b CM |
1439 | |
1440 | for (i = 0; i < i_done; i++) { | |
1441 | clear_page_dirty_for_io(pages[i]); | |
1442 | ClearPageChecked(pages[i]); | |
1443 | set_page_extent_mapped(pages[i]); | |
1444 | set_page_dirty(pages[i]); | |
1445 | unlock_page(pages[i]); | |
09cbfeaf | 1446 | put_page(pages[i]); |
4cb5300b | 1447 | } |
8702ba93 | 1448 | btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT); |
364ecf36 | 1449 | extent_changeset_free(data_reserved); |
4cb5300b | 1450 | return i_done; |
7f458a38 FM |
1451 | |
1452 | out_unlock_range: | |
1453 | unlock_extent_cached(&BTRFS_I(inode)->io_tree, | |
1454 | page_start, page_end - 1, &cached_state); | |
4cb5300b CM |
1455 | out: |
1456 | for (i = 0; i < i_done; i++) { | |
1457 | unlock_page(pages[i]); | |
09cbfeaf | 1458 | put_page(pages[i]); |
4cb5300b | 1459 | } |
86d52921 | 1460 | btrfs_delalloc_release_space(BTRFS_I(inode), data_reserved, |
a1fbc675 | 1461 | start, page_cnt << PAGE_SHIFT, true); |
8702ba93 | 1462 | btrfs_delalloc_release_extents(BTRFS_I(inode), page_cnt << PAGE_SHIFT); |
364ecf36 | 1463 | extent_changeset_free(data_reserved); |
4cb5300b CM |
1464 | return ret; |
1465 | ||
1466 | } | |
1467 | ||
1468 | int btrfs_defrag_file(struct inode *inode, struct file *file, | |
1469 | struct btrfs_ioctl_defrag_range_args *range, | |
1470 | u64 newer_than, unsigned long max_to_defrag) | |
1471 | { | |
0b246afa | 1472 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
4cb5300b | 1473 | struct btrfs_root *root = BTRFS_I(inode)->root; |
4cb5300b | 1474 | struct file_ra_state *ra = NULL; |
f46b5a66 | 1475 | unsigned long last_index; |
151a31b2 | 1476 | u64 isize = i_size_read(inode); |
940100a4 CM |
1477 | u64 last_len = 0; |
1478 | u64 skip = 0; | |
1479 | u64 defrag_end = 0; | |
4cb5300b | 1480 | u64 newer_off = range->start; |
f46b5a66 | 1481 | unsigned long i; |
008873ea | 1482 | unsigned long ra_index = 0; |
f46b5a66 | 1483 | int ret; |
4cb5300b | 1484 | int defrag_count = 0; |
1a419d85 | 1485 | int compress_type = BTRFS_COMPRESS_ZLIB; |
aab110ab | 1486 | u32 extent_thresh = range->extent_thresh; |
09cbfeaf | 1487 | unsigned long max_cluster = SZ_256K >> PAGE_SHIFT; |
c41570c9 | 1488 | unsigned long cluster = max_cluster; |
ee22184b | 1489 | u64 new_align = ~((u64)SZ_128K - 1); |
4cb5300b | 1490 | struct page **pages = NULL; |
1e2ef46d | 1491 | bool do_compress = range->flags & BTRFS_DEFRAG_RANGE_COMPRESS; |
4cb5300b | 1492 | |
0abd5b17 LB |
1493 | if (isize == 0) |
1494 | return 0; | |
1495 | ||
1496 | if (range->start >= isize) | |
1497 | return -EINVAL; | |
1a419d85 | 1498 | |
1e2ef46d | 1499 | if (do_compress) { |
ce96b7ff | 1500 | if (range->compress_type >= BTRFS_NR_COMPRESS_TYPES) |
1a419d85 LZ |
1501 | return -EINVAL; |
1502 | if (range->compress_type) | |
1503 | compress_type = range->compress_type; | |
1504 | } | |
f46b5a66 | 1505 | |
0abd5b17 | 1506 | if (extent_thresh == 0) |
ee22184b | 1507 | extent_thresh = SZ_256K; |
940100a4 | 1508 | |
4cb5300b | 1509 | /* |
0a52d108 DS |
1510 | * If we were not given a file, allocate a readahead context. As |
1511 | * readahead is just an optimization, defrag will work without it so | |
1512 | * we don't error out. | |
4cb5300b CM |
1513 | */ |
1514 | if (!file) { | |
63e727ec | 1515 | ra = kzalloc(sizeof(*ra), GFP_KERNEL); |
0a52d108 DS |
1516 | if (ra) |
1517 | file_ra_state_init(ra, inode->i_mapping); | |
4cb5300b CM |
1518 | } else { |
1519 | ra = &file->f_ra; | |
1520 | } | |
1521 | ||
63e727ec | 1522 | pages = kmalloc_array(max_cluster, sizeof(struct page *), GFP_KERNEL); |
4cb5300b CM |
1523 | if (!pages) { |
1524 | ret = -ENOMEM; | |
1525 | goto out_ra; | |
1526 | } | |
1527 | ||
1528 | /* find the last page to defrag */ | |
1e701a32 | 1529 | if (range->start + range->len > range->start) { |
151a31b2 | 1530 | last_index = min_t(u64, isize - 1, |
09cbfeaf | 1531 | range->start + range->len - 1) >> PAGE_SHIFT; |
1e701a32 | 1532 | } else { |
09cbfeaf | 1533 | last_index = (isize - 1) >> PAGE_SHIFT; |
1e701a32 CM |
1534 | } |
1535 | ||
4cb5300b CM |
1536 | if (newer_than) { |
1537 | ret = find_new_extents(root, inode, newer_than, | |
ee22184b | 1538 | &newer_off, SZ_64K); |
4cb5300b CM |
1539 | if (!ret) { |
1540 | range->start = newer_off; | |
1541 | /* | |
1542 | * we always align our defrag to help keep | |
1543 | * the extents in the file evenly spaced | |
1544 | */ | |
09cbfeaf | 1545 | i = (newer_off & new_align) >> PAGE_SHIFT; |
4cb5300b CM |
1546 | } else |
1547 | goto out_ra; | |
1548 | } else { | |
09cbfeaf | 1549 | i = range->start >> PAGE_SHIFT; |
4cb5300b CM |
1550 | } |
1551 | if (!max_to_defrag) | |
070034bd | 1552 | max_to_defrag = last_index - i + 1; |
4cb5300b | 1553 | |
2a0f7f57 LZ |
1554 | /* |
1555 | * make writeback starts from i, so the defrag range can be | |
1556 | * written sequentially. | |
1557 | */ | |
1558 | if (i < inode->i_mapping->writeback_index) | |
1559 | inode->i_mapping->writeback_index = i; | |
1560 | ||
f7f43cc8 | 1561 | while (i <= last_index && defrag_count < max_to_defrag && |
09cbfeaf | 1562 | (i < DIV_ROUND_UP(i_size_read(inode), PAGE_SIZE))) { |
4cb5300b CM |
1563 | /* |
1564 | * make sure we stop running if someone unmounts | |
1565 | * the FS | |
1566 | */ | |
1751e8a6 | 1567 | if (!(inode->i_sb->s_flags & SB_ACTIVE)) |
4cb5300b CM |
1568 | break; |
1569 | ||
0b246afa JM |
1570 | if (btrfs_defrag_cancelled(fs_info)) { |
1571 | btrfs_debug(fs_info, "defrag_file cancelled"); | |
210549eb DS |
1572 | ret = -EAGAIN; |
1573 | break; | |
1574 | } | |
1575 | ||
09cbfeaf | 1576 | if (!should_defrag_range(inode, (u64)i << PAGE_SHIFT, |
6c282eb4 | 1577 | extent_thresh, &last_len, &skip, |
1e2ef46d | 1578 | &defrag_end, do_compress)){ |
940100a4 CM |
1579 | unsigned long next; |
1580 | /* | |
1581 | * the should_defrag function tells us how much to skip | |
1582 | * bump our counter by the suggested amount | |
1583 | */ | |
09cbfeaf | 1584 | next = DIV_ROUND_UP(skip, PAGE_SIZE); |
940100a4 CM |
1585 | i = max(i + 1, next); |
1586 | continue; | |
1587 | } | |
008873ea LZ |
1588 | |
1589 | if (!newer_than) { | |
09cbfeaf KS |
1590 | cluster = (PAGE_ALIGN(defrag_end) >> |
1591 | PAGE_SHIFT) - i; | |
008873ea LZ |
1592 | cluster = min(cluster, max_cluster); |
1593 | } else { | |
1594 | cluster = max_cluster; | |
1595 | } | |
1596 | ||
008873ea LZ |
1597 | if (i + cluster > ra_index) { |
1598 | ra_index = max(i, ra_index); | |
0a52d108 | 1599 | if (ra) |
d3c0bab5 DS |
1600 | page_cache_sync_readahead(inode->i_mapping, ra, |
1601 | file, ra_index, cluster); | |
e4826a5b | 1602 | ra_index += cluster; |
008873ea | 1603 | } |
940100a4 | 1604 | |
5955102c | 1605 | inode_lock(inode); |
eede2bf3 OS |
1606 | if (IS_SWAPFILE(inode)) { |
1607 | ret = -ETXTBSY; | |
1608 | } else { | |
1609 | if (do_compress) | |
1610 | BTRFS_I(inode)->defrag_compress = compress_type; | |
1611 | ret = cluster_pages_for_defrag(inode, pages, i, cluster); | |
1612 | } | |
ecb8bea8 | 1613 | if (ret < 0) { |
5955102c | 1614 | inode_unlock(inode); |
4cb5300b | 1615 | goto out_ra; |
ecb8bea8 | 1616 | } |
4cb5300b CM |
1617 | |
1618 | defrag_count += ret; | |
d0e1d66b | 1619 | balance_dirty_pages_ratelimited(inode->i_mapping); |
5955102c | 1620 | inode_unlock(inode); |
4cb5300b CM |
1621 | |
1622 | if (newer_than) { | |
1623 | if (newer_off == (u64)-1) | |
1624 | break; | |
1625 | ||
e1f041e1 LB |
1626 | if (ret > 0) |
1627 | i += ret; | |
1628 | ||
4cb5300b | 1629 | newer_off = max(newer_off + 1, |
09cbfeaf | 1630 | (u64)i << PAGE_SHIFT); |
4cb5300b | 1631 | |
ee22184b BL |
1632 | ret = find_new_extents(root, inode, newer_than, |
1633 | &newer_off, SZ_64K); | |
4cb5300b CM |
1634 | if (!ret) { |
1635 | range->start = newer_off; | |
09cbfeaf | 1636 | i = (newer_off & new_align) >> PAGE_SHIFT; |
4cb5300b CM |
1637 | } else { |
1638 | break; | |
f46b5a66 | 1639 | } |
4cb5300b | 1640 | } else { |
008873ea | 1641 | if (ret > 0) { |
cbcc8326 | 1642 | i += ret; |
09cbfeaf | 1643 | last_len += ret << PAGE_SHIFT; |
008873ea | 1644 | } else { |
cbcc8326 | 1645 | i++; |
008873ea LZ |
1646 | last_len = 0; |
1647 | } | |
f46b5a66 | 1648 | } |
f46b5a66 CH |
1649 | } |
1650 | ||
dec8ef90 | 1651 | if ((range->flags & BTRFS_DEFRAG_RANGE_START_IO)) { |
1e701a32 | 1652 | filemap_flush(inode->i_mapping); |
dec8ef90 FM |
1653 | if (test_bit(BTRFS_INODE_HAS_ASYNC_EXTENT, |
1654 | &BTRFS_I(inode)->runtime_flags)) | |
1655 | filemap_flush(inode->i_mapping); | |
1656 | } | |
1e701a32 | 1657 | |
1a419d85 | 1658 | if (range->compress_type == BTRFS_COMPRESS_LZO) { |
0b246afa | 1659 | btrfs_set_fs_incompat(fs_info, COMPRESS_LZO); |
5c1aab1d NT |
1660 | } else if (range->compress_type == BTRFS_COMPRESS_ZSTD) { |
1661 | btrfs_set_fs_incompat(fs_info, COMPRESS_ZSTD); | |
1a419d85 LZ |
1662 | } |
1663 | ||
60ccf82f | 1664 | ret = defrag_count; |
940100a4 | 1665 | |
4cb5300b | 1666 | out_ra: |
1e2ef46d | 1667 | if (do_compress) { |
5955102c | 1668 | inode_lock(inode); |
eec63c65 | 1669 | BTRFS_I(inode)->defrag_compress = BTRFS_COMPRESS_NONE; |
5955102c | 1670 | inode_unlock(inode); |
633085c7 | 1671 | } |
4cb5300b CM |
1672 | if (!file) |
1673 | kfree(ra); | |
1674 | kfree(pages); | |
940100a4 | 1675 | return ret; |
f46b5a66 CH |
1676 | } |
1677 | ||
198605a8 | 1678 | static noinline int btrfs_ioctl_resize(struct file *file, |
76dda93c | 1679 | void __user *arg) |
f46b5a66 | 1680 | { |
0b246afa JM |
1681 | struct inode *inode = file_inode(file); |
1682 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
f46b5a66 CH |
1683 | u64 new_size; |
1684 | u64 old_size; | |
1685 | u64 devid = 1; | |
0b246afa | 1686 | struct btrfs_root *root = BTRFS_I(inode)->root; |
f46b5a66 CH |
1687 | struct btrfs_ioctl_vol_args *vol_args; |
1688 | struct btrfs_trans_handle *trans; | |
1689 | struct btrfs_device *device = NULL; | |
1690 | char *sizestr; | |
9a40f122 | 1691 | char *retptr; |
f46b5a66 CH |
1692 | char *devstr = NULL; |
1693 | int ret = 0; | |
f46b5a66 CH |
1694 | int mod = 0; |
1695 | ||
e441d54d CM |
1696 | if (!capable(CAP_SYS_ADMIN)) |
1697 | return -EPERM; | |
1698 | ||
198605a8 MX |
1699 | ret = mnt_want_write_file(file); |
1700 | if (ret) | |
1701 | return ret; | |
1702 | ||
c3e1f96c | 1703 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_RESIZE)) { |
97547676 | 1704 | mnt_drop_write_file(file); |
e57138b3 | 1705 | return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
c9e9f97b ID |
1706 | } |
1707 | ||
dae7b665 | 1708 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
1709 | if (IS_ERR(vol_args)) { |
1710 | ret = PTR_ERR(vol_args); | |
1711 | goto out; | |
1712 | } | |
5516e595 MF |
1713 | |
1714 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
f46b5a66 | 1715 | |
f46b5a66 CH |
1716 | sizestr = vol_args->name; |
1717 | devstr = strchr(sizestr, ':'); | |
1718 | if (devstr) { | |
f46b5a66 CH |
1719 | sizestr = devstr + 1; |
1720 | *devstr = '\0'; | |
1721 | devstr = vol_args->name; | |
58dfae63 Z |
1722 | ret = kstrtoull(devstr, 10, &devid); |
1723 | if (ret) | |
1724 | goto out_free; | |
dfd79829 MX |
1725 | if (!devid) { |
1726 | ret = -EINVAL; | |
1727 | goto out_free; | |
1728 | } | |
0b246afa | 1729 | btrfs_info(fs_info, "resizing devid %llu", devid); |
f46b5a66 | 1730 | } |
dba60f3f | 1731 | |
b2598edf | 1732 | device = btrfs_find_device(fs_info->fs_devices, devid, NULL, NULL); |
f46b5a66 | 1733 | if (!device) { |
0b246afa JM |
1734 | btrfs_info(fs_info, "resizer unable to find device %llu", |
1735 | devid); | |
dfd79829 | 1736 | ret = -ENODEV; |
c9e9f97b | 1737 | goto out_free; |
f46b5a66 | 1738 | } |
dba60f3f | 1739 | |
ebbede42 | 1740 | if (!test_bit(BTRFS_DEV_STATE_WRITEABLE, &device->dev_state)) { |
0b246afa | 1741 | btrfs_info(fs_info, |
efe120a0 | 1742 | "resizer unable to apply on readonly device %llu", |
c1c9ff7c | 1743 | devid); |
dfd79829 | 1744 | ret = -EPERM; |
4e42ae1b LB |
1745 | goto out_free; |
1746 | } | |
1747 | ||
f46b5a66 CH |
1748 | if (!strcmp(sizestr, "max")) |
1749 | new_size = device->bdev->bd_inode->i_size; | |
1750 | else { | |
1751 | if (sizestr[0] == '-') { | |
1752 | mod = -1; | |
1753 | sizestr++; | |
1754 | } else if (sizestr[0] == '+') { | |
1755 | mod = 1; | |
1756 | sizestr++; | |
1757 | } | |
9a40f122 GH |
1758 | new_size = memparse(sizestr, &retptr); |
1759 | if (*retptr != '\0' || new_size == 0) { | |
f46b5a66 | 1760 | ret = -EINVAL; |
c9e9f97b | 1761 | goto out_free; |
f46b5a66 CH |
1762 | } |
1763 | } | |
1764 | ||
401e29c1 | 1765 | if (test_bit(BTRFS_DEV_STATE_REPLACE_TGT, &device->dev_state)) { |
dfd79829 | 1766 | ret = -EPERM; |
63a212ab SB |
1767 | goto out_free; |
1768 | } | |
1769 | ||
7cc8e58d | 1770 | old_size = btrfs_device_get_total_bytes(device); |
f46b5a66 CH |
1771 | |
1772 | if (mod < 0) { | |
1773 | if (new_size > old_size) { | |
1774 | ret = -EINVAL; | |
c9e9f97b | 1775 | goto out_free; |
f46b5a66 CH |
1776 | } |
1777 | new_size = old_size - new_size; | |
1778 | } else if (mod > 0) { | |
eb8052e0 | 1779 | if (new_size > ULLONG_MAX - old_size) { |
902c68a4 | 1780 | ret = -ERANGE; |
eb8052e0 WF |
1781 | goto out_free; |
1782 | } | |
f46b5a66 CH |
1783 | new_size = old_size + new_size; |
1784 | } | |
1785 | ||
ee22184b | 1786 | if (new_size < SZ_256M) { |
f46b5a66 | 1787 | ret = -EINVAL; |
c9e9f97b | 1788 | goto out_free; |
f46b5a66 CH |
1789 | } |
1790 | if (new_size > device->bdev->bd_inode->i_size) { | |
1791 | ret = -EFBIG; | |
c9e9f97b | 1792 | goto out_free; |
f46b5a66 CH |
1793 | } |
1794 | ||
47f08b96 | 1795 | new_size = round_down(new_size, fs_info->sectorsize); |
f46b5a66 | 1796 | |
f46b5a66 | 1797 | if (new_size > old_size) { |
a22285a6 | 1798 | trans = btrfs_start_transaction(root, 0); |
98d5dc13 TI |
1799 | if (IS_ERR(trans)) { |
1800 | ret = PTR_ERR(trans); | |
c9e9f97b | 1801 | goto out_free; |
98d5dc13 | 1802 | } |
f46b5a66 | 1803 | ret = btrfs_grow_device(trans, device, new_size); |
3a45bb20 | 1804 | btrfs_commit_transaction(trans); |
ece7d20e | 1805 | } else if (new_size < old_size) { |
f46b5a66 | 1806 | ret = btrfs_shrink_device(device, new_size); |
0253f40e | 1807 | } /* equal, nothing need to do */ |
f46b5a66 | 1808 | |
faf8f7b9 MPS |
1809 | if (ret == 0 && new_size != old_size) |
1810 | btrfs_info_in_rcu(fs_info, | |
1811 | "resize device %s (devid %llu) from %llu to %llu", | |
1812 | rcu_str_deref(device->name), device->devid, | |
1813 | old_size, new_size); | |
c9e9f97b | 1814 | out_free: |
f46b5a66 | 1815 | kfree(vol_args); |
c9e9f97b | 1816 | out: |
c3e1f96c | 1817 | btrfs_exclop_finish(fs_info); |
18f39c41 | 1818 | mnt_drop_write_file(file); |
f46b5a66 CH |
1819 | return ret; |
1820 | } | |
1821 | ||
5d54c67e | 1822 | static noinline int __btrfs_ioctl_snap_create(struct file *file, |
52f75f4f | 1823 | const char *name, unsigned long fd, int subvol, |
5d54c67e | 1824 | bool readonly, |
8696c533 | 1825 | struct btrfs_qgroup_inherit *inherit) |
f46b5a66 | 1826 | { |
f46b5a66 | 1827 | int namelen; |
3de4586c | 1828 | int ret = 0; |
f46b5a66 | 1829 | |
325c50e3 JM |
1830 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1831 | return -ENOTDIR; | |
1832 | ||
a874a63e LB |
1833 | ret = mnt_want_write_file(file); |
1834 | if (ret) | |
1835 | goto out; | |
1836 | ||
72fd032e SW |
1837 | namelen = strlen(name); |
1838 | if (strchr(name, '/')) { | |
f46b5a66 | 1839 | ret = -EINVAL; |
a874a63e | 1840 | goto out_drop_write; |
f46b5a66 CH |
1841 | } |
1842 | ||
16780cab CM |
1843 | if (name[0] == '.' && |
1844 | (namelen == 1 || (name[1] == '.' && namelen == 2))) { | |
1845 | ret = -EEXIST; | |
a874a63e | 1846 | goto out_drop_write; |
16780cab CM |
1847 | } |
1848 | ||
3de4586c | 1849 | if (subvol) { |
72fd032e | 1850 | ret = btrfs_mksubvol(&file->f_path, name, namelen, |
5d54c67e | 1851 | NULL, readonly, inherit); |
cb8e7090 | 1852 | } else { |
2903ff01 | 1853 | struct fd src = fdget(fd); |
3de4586c | 1854 | struct inode *src_inode; |
2903ff01 | 1855 | if (!src.file) { |
3de4586c | 1856 | ret = -EINVAL; |
a874a63e | 1857 | goto out_drop_write; |
3de4586c CM |
1858 | } |
1859 | ||
496ad9aa AV |
1860 | src_inode = file_inode(src.file); |
1861 | if (src_inode->i_sb != file_inode(file)->i_sb) { | |
c79b4713 | 1862 | btrfs_info(BTRFS_I(file_inode(file))->root->fs_info, |
efe120a0 | 1863 | "Snapshot src from another FS"); |
23ad5b17 | 1864 | ret = -EXDEV; |
d0242061 DS |
1865 | } else if (!inode_owner_or_capable(src_inode)) { |
1866 | /* | |
1867 | * Subvolume creation is not restricted, but snapshots | |
1868 | * are limited to own subvolumes only | |
1869 | */ | |
1870 | ret = -EPERM; | |
ecd18815 | 1871 | } else { |
c11fbb6e | 1872 | ret = btrfs_mksnapshot(&file->f_path, name, namelen, |
ecd18815 | 1873 | BTRFS_I(src_inode)->root, |
5d54c67e | 1874 | readonly, inherit); |
3de4586c | 1875 | } |
2903ff01 | 1876 | fdput(src); |
cb8e7090 | 1877 | } |
a874a63e LB |
1878 | out_drop_write: |
1879 | mnt_drop_write_file(file); | |
f46b5a66 | 1880 | out: |
72fd032e SW |
1881 | return ret; |
1882 | } | |
1883 | ||
1884 | static noinline int btrfs_ioctl_snap_create(struct file *file, | |
fa0d2b9b | 1885 | void __user *arg, int subvol) |
72fd032e | 1886 | { |
fa0d2b9b | 1887 | struct btrfs_ioctl_vol_args *vol_args; |
72fd032e SW |
1888 | int ret; |
1889 | ||
325c50e3 JM |
1890 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1891 | return -ENOTDIR; | |
1892 | ||
fa0d2b9b LZ |
1893 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1894 | if (IS_ERR(vol_args)) | |
1895 | return PTR_ERR(vol_args); | |
1896 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; | |
72fd032e | 1897 | |
5d54c67e NB |
1898 | ret = __btrfs_ioctl_snap_create(file, vol_args->name, vol_args->fd, |
1899 | subvol, false, NULL); | |
fdfb1e4f | 1900 | |
fa0d2b9b LZ |
1901 | kfree(vol_args); |
1902 | return ret; | |
1903 | } | |
fdfb1e4f | 1904 | |
fa0d2b9b LZ |
1905 | static noinline int btrfs_ioctl_snap_create_v2(struct file *file, |
1906 | void __user *arg, int subvol) | |
1907 | { | |
1908 | struct btrfs_ioctl_vol_args_v2 *vol_args; | |
1909 | int ret; | |
b83cc969 | 1910 | bool readonly = false; |
6f72c7e2 | 1911 | struct btrfs_qgroup_inherit *inherit = NULL; |
75eaa0e2 | 1912 | |
325c50e3 JM |
1913 | if (!S_ISDIR(file_inode(file)->i_mode)) |
1914 | return -ENOTDIR; | |
1915 | ||
fa0d2b9b LZ |
1916 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
1917 | if (IS_ERR(vol_args)) | |
1918 | return PTR_ERR(vol_args); | |
1919 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
75eaa0e2 | 1920 | |
673990db | 1921 | if (vol_args->flags & ~BTRFS_SUBVOL_CREATE_ARGS_MASK) { |
b83cc969 | 1922 | ret = -EOPNOTSUPP; |
c47ca32d | 1923 | goto free_args; |
72fd032e | 1924 | } |
fa0d2b9b | 1925 | |
b83cc969 LZ |
1926 | if (vol_args->flags & BTRFS_SUBVOL_RDONLY) |
1927 | readonly = true; | |
6f72c7e2 | 1928 | if (vol_args->flags & BTRFS_SUBVOL_QGROUP_INHERIT) { |
09cbfeaf | 1929 | if (vol_args->size > PAGE_SIZE) { |
6f72c7e2 | 1930 | ret = -EINVAL; |
c47ca32d | 1931 | goto free_args; |
6f72c7e2 AJ |
1932 | } |
1933 | inherit = memdup_user(vol_args->qgroup_inherit, vol_args->size); | |
1934 | if (IS_ERR(inherit)) { | |
1935 | ret = PTR_ERR(inherit); | |
c47ca32d | 1936 | goto free_args; |
6f72c7e2 AJ |
1937 | } |
1938 | } | |
fa0d2b9b | 1939 | |
5d54c67e NB |
1940 | ret = __btrfs_ioctl_snap_create(file, vol_args->name, vol_args->fd, |
1941 | subvol, readonly, inherit); | |
c47ca32d DC |
1942 | if (ret) |
1943 | goto free_inherit; | |
c47ca32d | 1944 | free_inherit: |
6f72c7e2 | 1945 | kfree(inherit); |
c47ca32d DC |
1946 | free_args: |
1947 | kfree(vol_args); | |
f46b5a66 CH |
1948 | return ret; |
1949 | } | |
1950 | ||
0caa102d LZ |
1951 | static noinline int btrfs_ioctl_subvol_getflags(struct file *file, |
1952 | void __user *arg) | |
1953 | { | |
496ad9aa | 1954 | struct inode *inode = file_inode(file); |
0b246afa | 1955 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
0caa102d LZ |
1956 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1957 | int ret = 0; | |
1958 | u64 flags = 0; | |
1959 | ||
4a0cc7ca | 1960 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) |
0caa102d LZ |
1961 | return -EINVAL; |
1962 | ||
0b246afa | 1963 | down_read(&fs_info->subvol_sem); |
0caa102d LZ |
1964 | if (btrfs_root_readonly(root)) |
1965 | flags |= BTRFS_SUBVOL_RDONLY; | |
0b246afa | 1966 | up_read(&fs_info->subvol_sem); |
0caa102d LZ |
1967 | |
1968 | if (copy_to_user(arg, &flags, sizeof(flags))) | |
1969 | ret = -EFAULT; | |
1970 | ||
1971 | return ret; | |
1972 | } | |
1973 | ||
1974 | static noinline int btrfs_ioctl_subvol_setflags(struct file *file, | |
1975 | void __user *arg) | |
1976 | { | |
496ad9aa | 1977 | struct inode *inode = file_inode(file); |
0b246afa | 1978 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
0caa102d LZ |
1979 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1980 | struct btrfs_trans_handle *trans; | |
1981 | u64 root_flags; | |
1982 | u64 flags; | |
1983 | int ret = 0; | |
1984 | ||
bd60ea0f DS |
1985 | if (!inode_owner_or_capable(inode)) |
1986 | return -EPERM; | |
1987 | ||
b9ca0664 LB |
1988 | ret = mnt_want_write_file(file); |
1989 | if (ret) | |
1990 | goto out; | |
0caa102d | 1991 | |
4a0cc7ca | 1992 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
b9ca0664 LB |
1993 | ret = -EINVAL; |
1994 | goto out_drop_write; | |
1995 | } | |
0caa102d | 1996 | |
b9ca0664 LB |
1997 | if (copy_from_user(&flags, arg, sizeof(flags))) { |
1998 | ret = -EFAULT; | |
1999 | goto out_drop_write; | |
2000 | } | |
0caa102d | 2001 | |
b9ca0664 LB |
2002 | if (flags & ~BTRFS_SUBVOL_RDONLY) { |
2003 | ret = -EOPNOTSUPP; | |
2004 | goto out_drop_write; | |
2005 | } | |
0caa102d | 2006 | |
0b246afa | 2007 | down_write(&fs_info->subvol_sem); |
0caa102d LZ |
2008 | |
2009 | /* nothing to do */ | |
2010 | if (!!(flags & BTRFS_SUBVOL_RDONLY) == btrfs_root_readonly(root)) | |
b9ca0664 | 2011 | goto out_drop_sem; |
0caa102d LZ |
2012 | |
2013 | root_flags = btrfs_root_flags(&root->root_item); | |
2c686537 | 2014 | if (flags & BTRFS_SUBVOL_RDONLY) { |
0caa102d LZ |
2015 | btrfs_set_root_flags(&root->root_item, |
2016 | root_flags | BTRFS_ROOT_SUBVOL_RDONLY); | |
2c686537 DS |
2017 | } else { |
2018 | /* | |
2019 | * Block RO -> RW transition if this subvolume is involved in | |
2020 | * send | |
2021 | */ | |
2022 | spin_lock(&root->root_item_lock); | |
2023 | if (root->send_in_progress == 0) { | |
2024 | btrfs_set_root_flags(&root->root_item, | |
0caa102d | 2025 | root_flags & ~BTRFS_ROOT_SUBVOL_RDONLY); |
2c686537 DS |
2026 | spin_unlock(&root->root_item_lock); |
2027 | } else { | |
2028 | spin_unlock(&root->root_item_lock); | |
0b246afa JM |
2029 | btrfs_warn(fs_info, |
2030 | "Attempt to set subvolume %llu read-write during send", | |
2031 | root->root_key.objectid); | |
2c686537 DS |
2032 | ret = -EPERM; |
2033 | goto out_drop_sem; | |
2034 | } | |
2035 | } | |
0caa102d LZ |
2036 | |
2037 | trans = btrfs_start_transaction(root, 1); | |
2038 | if (IS_ERR(trans)) { | |
2039 | ret = PTR_ERR(trans); | |
2040 | goto out_reset; | |
2041 | } | |
2042 | ||
0b246afa | 2043 | ret = btrfs_update_root(trans, fs_info->tree_root, |
0caa102d | 2044 | &root->root_key, &root->root_item); |
9417ebc8 NB |
2045 | if (ret < 0) { |
2046 | btrfs_end_transaction(trans); | |
2047 | goto out_reset; | |
2048 | } | |
2049 | ||
2050 | ret = btrfs_commit_transaction(trans); | |
0caa102d | 2051 | |
0caa102d LZ |
2052 | out_reset: |
2053 | if (ret) | |
2054 | btrfs_set_root_flags(&root->root_item, root_flags); | |
b9ca0664 | 2055 | out_drop_sem: |
0b246afa | 2056 | up_write(&fs_info->subvol_sem); |
b9ca0664 LB |
2057 | out_drop_write: |
2058 | mnt_drop_write_file(file); | |
2059 | out: | |
0caa102d LZ |
2060 | return ret; |
2061 | } | |
2062 | ||
ac8e9819 CM |
2063 | static noinline int key_in_sk(struct btrfs_key *key, |
2064 | struct btrfs_ioctl_search_key *sk) | |
2065 | { | |
abc6e134 CM |
2066 | struct btrfs_key test; |
2067 | int ret; | |
2068 | ||
2069 | test.objectid = sk->min_objectid; | |
2070 | test.type = sk->min_type; | |
2071 | test.offset = sk->min_offset; | |
2072 | ||
2073 | ret = btrfs_comp_cpu_keys(key, &test); | |
2074 | if (ret < 0) | |
ac8e9819 | 2075 | return 0; |
abc6e134 CM |
2076 | |
2077 | test.objectid = sk->max_objectid; | |
2078 | test.type = sk->max_type; | |
2079 | test.offset = sk->max_offset; | |
2080 | ||
2081 | ret = btrfs_comp_cpu_keys(key, &test); | |
2082 | if (ret > 0) | |
ac8e9819 CM |
2083 | return 0; |
2084 | return 1; | |
2085 | } | |
2086 | ||
df397565 | 2087 | static noinline int copy_to_sk(struct btrfs_path *path, |
ac8e9819 CM |
2088 | struct btrfs_key *key, |
2089 | struct btrfs_ioctl_search_key *sk, | |
9b6e817d | 2090 | size_t *buf_size, |
ba346b35 | 2091 | char __user *ubuf, |
ac8e9819 CM |
2092 | unsigned long *sk_offset, |
2093 | int *num_found) | |
2094 | { | |
2095 | u64 found_transid; | |
2096 | struct extent_buffer *leaf; | |
2097 | struct btrfs_ioctl_search_header sh; | |
dd81d459 | 2098 | struct btrfs_key test; |
ac8e9819 CM |
2099 | unsigned long item_off; |
2100 | unsigned long item_len; | |
2101 | int nritems; | |
2102 | int i; | |
2103 | int slot; | |
ac8e9819 CM |
2104 | int ret = 0; |
2105 | ||
2106 | leaf = path->nodes[0]; | |
2107 | slot = path->slots[0]; | |
2108 | nritems = btrfs_header_nritems(leaf); | |
2109 | ||
2110 | if (btrfs_header_generation(leaf) > sk->max_transid) { | |
2111 | i = nritems; | |
2112 | goto advance_key; | |
2113 | } | |
2114 | found_transid = btrfs_header_generation(leaf); | |
2115 | ||
2116 | for (i = slot; i < nritems; i++) { | |
2117 | item_off = btrfs_item_ptr_offset(leaf, i); | |
2118 | item_len = btrfs_item_size_nr(leaf, i); | |
2119 | ||
03b71c6c GP |
2120 | btrfs_item_key_to_cpu(leaf, key, i); |
2121 | if (!key_in_sk(key, sk)) | |
2122 | continue; | |
2123 | ||
9b6e817d | 2124 | if (sizeof(sh) + item_len > *buf_size) { |
8f5f6178 GH |
2125 | if (*num_found) { |
2126 | ret = 1; | |
2127 | goto out; | |
2128 | } | |
2129 | ||
2130 | /* | |
2131 | * return one empty item back for v1, which does not | |
2132 | * handle -EOVERFLOW | |
2133 | */ | |
2134 | ||
9b6e817d | 2135 | *buf_size = sizeof(sh) + item_len; |
ac8e9819 | 2136 | item_len = 0; |
8f5f6178 GH |
2137 | ret = -EOVERFLOW; |
2138 | } | |
ac8e9819 | 2139 | |
9b6e817d | 2140 | if (sizeof(sh) + item_len + *sk_offset > *buf_size) { |
ac8e9819 | 2141 | ret = 1; |
25c9bc2e | 2142 | goto out; |
ac8e9819 CM |
2143 | } |
2144 | ||
ac8e9819 CM |
2145 | sh.objectid = key->objectid; |
2146 | sh.offset = key->offset; | |
2147 | sh.type = key->type; | |
2148 | sh.len = item_len; | |
2149 | sh.transid = found_transid; | |
2150 | ||
a48b73ec JB |
2151 | /* |
2152 | * Copy search result header. If we fault then loop again so we | |
2153 | * can fault in the pages and -EFAULT there if there's a | |
2154 | * problem. Otherwise we'll fault and then copy the buffer in | |
2155 | * properly this next time through | |
2156 | */ | |
2157 | if (copy_to_user_nofault(ubuf + *sk_offset, &sh, sizeof(sh))) { | |
2158 | ret = 0; | |
ba346b35 GH |
2159 | goto out; |
2160 | } | |
2161 | ||
ac8e9819 CM |
2162 | *sk_offset += sizeof(sh); |
2163 | ||
2164 | if (item_len) { | |
ba346b35 | 2165 | char __user *up = ubuf + *sk_offset; |
a48b73ec JB |
2166 | /* |
2167 | * Copy the item, same behavior as above, but reset the | |
2168 | * * sk_offset so we copy the full thing again. | |
2169 | */ | |
2170 | if (read_extent_buffer_to_user_nofault(leaf, up, | |
2171 | item_off, item_len)) { | |
2172 | ret = 0; | |
2173 | *sk_offset -= sizeof(sh); | |
ba346b35 GH |
2174 | goto out; |
2175 | } | |
2176 | ||
ac8e9819 | 2177 | *sk_offset += item_len; |
ac8e9819 | 2178 | } |
e2156867 | 2179 | (*num_found)++; |
ac8e9819 | 2180 | |
8f5f6178 GH |
2181 | if (ret) /* -EOVERFLOW from above */ |
2182 | goto out; | |
2183 | ||
25c9bc2e GH |
2184 | if (*num_found >= sk->nr_items) { |
2185 | ret = 1; | |
2186 | goto out; | |
2187 | } | |
ac8e9819 CM |
2188 | } |
2189 | advance_key: | |
abc6e134 | 2190 | ret = 0; |
dd81d459 NA |
2191 | test.objectid = sk->max_objectid; |
2192 | test.type = sk->max_type; | |
2193 | test.offset = sk->max_offset; | |
2194 | if (btrfs_comp_cpu_keys(key, &test) >= 0) | |
2195 | ret = 1; | |
2196 | else if (key->offset < (u64)-1) | |
ac8e9819 | 2197 | key->offset++; |
dd81d459 | 2198 | else if (key->type < (u8)-1) { |
abc6e134 | 2199 | key->offset = 0; |
ac8e9819 | 2200 | key->type++; |
dd81d459 | 2201 | } else if (key->objectid < (u64)-1) { |
abc6e134 CM |
2202 | key->offset = 0; |
2203 | key->type = 0; | |
ac8e9819 | 2204 | key->objectid++; |
abc6e134 CM |
2205 | } else |
2206 | ret = 1; | |
25c9bc2e | 2207 | out: |
ba346b35 GH |
2208 | /* |
2209 | * 0: all items from this leaf copied, continue with next | |
2210 | * 1: * more items can be copied, but unused buffer is too small | |
2211 | * * all items were found | |
2212 | * Either way, it will stops the loop which iterates to the next | |
2213 | * leaf | |
2214 | * -EOVERFLOW: item was to large for buffer | |
2215 | * -EFAULT: could not copy extent buffer back to userspace | |
2216 | */ | |
ac8e9819 CM |
2217 | return ret; |
2218 | } | |
2219 | ||
2220 | static noinline int search_ioctl(struct inode *inode, | |
12544442 | 2221 | struct btrfs_ioctl_search_key *sk, |
9b6e817d | 2222 | size_t *buf_size, |
ba346b35 | 2223 | char __user *ubuf) |
ac8e9819 | 2224 | { |
0b246afa | 2225 | struct btrfs_fs_info *info = btrfs_sb(inode->i_sb); |
ac8e9819 CM |
2226 | struct btrfs_root *root; |
2227 | struct btrfs_key key; | |
ac8e9819 | 2228 | struct btrfs_path *path; |
ac8e9819 CM |
2229 | int ret; |
2230 | int num_found = 0; | |
2231 | unsigned long sk_offset = 0; | |
2232 | ||
9b6e817d GH |
2233 | if (*buf_size < sizeof(struct btrfs_ioctl_search_header)) { |
2234 | *buf_size = sizeof(struct btrfs_ioctl_search_header); | |
12544442 | 2235 | return -EOVERFLOW; |
9b6e817d | 2236 | } |
12544442 | 2237 | |
ac8e9819 CM |
2238 | path = btrfs_alloc_path(); |
2239 | if (!path) | |
2240 | return -ENOMEM; | |
2241 | ||
2242 | if (sk->tree_id == 0) { | |
2243 | /* search the root of the inode that was passed */ | |
00246528 | 2244 | root = btrfs_grab_root(BTRFS_I(inode)->root); |
ac8e9819 | 2245 | } else { |
56e9357a | 2246 | root = btrfs_get_fs_root(info, sk->tree_id, true); |
ac8e9819 | 2247 | if (IS_ERR(root)) { |
ac8e9819 | 2248 | btrfs_free_path(path); |
ad1e3d56 | 2249 | return PTR_ERR(root); |
ac8e9819 CM |
2250 | } |
2251 | } | |
2252 | ||
2253 | key.objectid = sk->min_objectid; | |
2254 | key.type = sk->min_type; | |
2255 | key.offset = sk->min_offset; | |
2256 | ||
67871254 | 2257 | while (1) { |
1c78544e FM |
2258 | ret = fault_in_pages_writeable(ubuf + sk_offset, |
2259 | *buf_size - sk_offset); | |
a48b73ec JB |
2260 | if (ret) |
2261 | break; | |
2262 | ||
6174d3cb | 2263 | ret = btrfs_search_forward(root, &key, path, sk->min_transid); |
ac8e9819 CM |
2264 | if (ret != 0) { |
2265 | if (ret > 0) | |
2266 | ret = 0; | |
2267 | goto err; | |
2268 | } | |
df397565 | 2269 | ret = copy_to_sk(path, &key, sk, buf_size, ubuf, |
ac8e9819 | 2270 | &sk_offset, &num_found); |
b3b4aa74 | 2271 | btrfs_release_path(path); |
25c9bc2e | 2272 | if (ret) |
ac8e9819 CM |
2273 | break; |
2274 | ||
2275 | } | |
8f5f6178 GH |
2276 | if (ret > 0) |
2277 | ret = 0; | |
ac8e9819 CM |
2278 | err: |
2279 | sk->nr_items = num_found; | |
00246528 | 2280 | btrfs_put_root(root); |
ac8e9819 CM |
2281 | btrfs_free_path(path); |
2282 | return ret; | |
2283 | } | |
2284 | ||
2285 | static noinline int btrfs_ioctl_tree_search(struct file *file, | |
2286 | void __user *argp) | |
2287 | { | |
ba346b35 GH |
2288 | struct btrfs_ioctl_search_args __user *uargs; |
2289 | struct btrfs_ioctl_search_key sk; | |
9b6e817d GH |
2290 | struct inode *inode; |
2291 | int ret; | |
2292 | size_t buf_size; | |
ac8e9819 CM |
2293 | |
2294 | if (!capable(CAP_SYS_ADMIN)) | |
2295 | return -EPERM; | |
2296 | ||
ba346b35 GH |
2297 | uargs = (struct btrfs_ioctl_search_args __user *)argp; |
2298 | ||
2299 | if (copy_from_user(&sk, &uargs->key, sizeof(sk))) | |
2300 | return -EFAULT; | |
ac8e9819 | 2301 | |
ba346b35 | 2302 | buf_size = sizeof(uargs->buf); |
ac8e9819 | 2303 | |
496ad9aa | 2304 | inode = file_inode(file); |
ba346b35 | 2305 | ret = search_ioctl(inode, &sk, &buf_size, uargs->buf); |
8f5f6178 GH |
2306 | |
2307 | /* | |
2308 | * In the origin implementation an overflow is handled by returning a | |
2309 | * search header with a len of zero, so reset ret. | |
2310 | */ | |
2311 | if (ret == -EOVERFLOW) | |
2312 | ret = 0; | |
2313 | ||
ba346b35 | 2314 | if (ret == 0 && copy_to_user(&uargs->key, &sk, sizeof(sk))) |
ac8e9819 | 2315 | ret = -EFAULT; |
ac8e9819 CM |
2316 | return ret; |
2317 | } | |
2318 | ||
cc68a8a5 GH |
2319 | static noinline int btrfs_ioctl_tree_search_v2(struct file *file, |
2320 | void __user *argp) | |
2321 | { | |
2322 | struct btrfs_ioctl_search_args_v2 __user *uarg; | |
2323 | struct btrfs_ioctl_search_args_v2 args; | |
2324 | struct inode *inode; | |
2325 | int ret; | |
2326 | size_t buf_size; | |
ee22184b | 2327 | const size_t buf_limit = SZ_16M; |
cc68a8a5 GH |
2328 | |
2329 | if (!capable(CAP_SYS_ADMIN)) | |
2330 | return -EPERM; | |
2331 | ||
2332 | /* copy search header and buffer size */ | |
2333 | uarg = (struct btrfs_ioctl_search_args_v2 __user *)argp; | |
2334 | if (copy_from_user(&args, uarg, sizeof(args))) | |
2335 | return -EFAULT; | |
2336 | ||
2337 | buf_size = args.buf_size; | |
2338 | ||
cc68a8a5 GH |
2339 | /* limit result size to 16MB */ |
2340 | if (buf_size > buf_limit) | |
2341 | buf_size = buf_limit; | |
2342 | ||
2343 | inode = file_inode(file); | |
2344 | ret = search_ioctl(inode, &args.key, &buf_size, | |
718dc5fa | 2345 | (char __user *)(&uarg->buf[0])); |
cc68a8a5 GH |
2346 | if (ret == 0 && copy_to_user(&uarg->key, &args.key, sizeof(args.key))) |
2347 | ret = -EFAULT; | |
2348 | else if (ret == -EOVERFLOW && | |
2349 | copy_to_user(&uarg->buf_size, &buf_size, sizeof(buf_size))) | |
2350 | ret = -EFAULT; | |
2351 | ||
ac8e9819 CM |
2352 | return ret; |
2353 | } | |
2354 | ||
98d377a0 | 2355 | /* |
ac8e9819 CM |
2356 | * Search INODE_REFs to identify path name of 'dirid' directory |
2357 | * in a 'tree_id' tree. and sets path name to 'name'. | |
2358 | */ | |
98d377a0 TH |
2359 | static noinline int btrfs_search_path_in_tree(struct btrfs_fs_info *info, |
2360 | u64 tree_id, u64 dirid, char *name) | |
2361 | { | |
2362 | struct btrfs_root *root; | |
2363 | struct btrfs_key key; | |
ac8e9819 | 2364 | char *ptr; |
98d377a0 TH |
2365 | int ret = -1; |
2366 | int slot; | |
2367 | int len; | |
2368 | int total_len = 0; | |
2369 | struct btrfs_inode_ref *iref; | |
2370 | struct extent_buffer *l; | |
2371 | struct btrfs_path *path; | |
2372 | ||
2373 | if (dirid == BTRFS_FIRST_FREE_OBJECTID) { | |
2374 | name[0]='\0'; | |
2375 | return 0; | |
2376 | } | |
2377 | ||
2378 | path = btrfs_alloc_path(); | |
2379 | if (!path) | |
2380 | return -ENOMEM; | |
2381 | ||
c8bcbfbd | 2382 | ptr = &name[BTRFS_INO_LOOKUP_PATH_MAX - 1]; |
98d377a0 | 2383 | |
56e9357a | 2384 | root = btrfs_get_fs_root(info, tree_id, true); |
98d377a0 | 2385 | if (IS_ERR(root)) { |
ad1e3d56 | 2386 | ret = PTR_ERR(root); |
88234012 JB |
2387 | root = NULL; |
2388 | goto out; | |
2389 | } | |
98d377a0 TH |
2390 | |
2391 | key.objectid = dirid; | |
2392 | key.type = BTRFS_INODE_REF_KEY; | |
8ad6fcab | 2393 | key.offset = (u64)-1; |
98d377a0 | 2394 | |
67871254 | 2395 | while (1) { |
98d377a0 TH |
2396 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); |
2397 | if (ret < 0) | |
2398 | goto out; | |
18674c6c FDBM |
2399 | else if (ret > 0) { |
2400 | ret = btrfs_previous_item(root, path, dirid, | |
2401 | BTRFS_INODE_REF_KEY); | |
2402 | if (ret < 0) | |
2403 | goto out; | |
2404 | else if (ret > 0) { | |
2405 | ret = -ENOENT; | |
2406 | goto out; | |
2407 | } | |
2408 | } | |
98d377a0 TH |
2409 | |
2410 | l = path->nodes[0]; | |
2411 | slot = path->slots[0]; | |
2412 | btrfs_item_key_to_cpu(l, &key, slot); | |
2413 | ||
98d377a0 TH |
2414 | iref = btrfs_item_ptr(l, slot, struct btrfs_inode_ref); |
2415 | len = btrfs_inode_ref_name_len(l, iref); | |
2416 | ptr -= len + 1; | |
2417 | total_len += len + 1; | |
a696cf35 FDBM |
2418 | if (ptr < name) { |
2419 | ret = -ENAMETOOLONG; | |
98d377a0 | 2420 | goto out; |
a696cf35 | 2421 | } |
98d377a0 TH |
2422 | |
2423 | *(ptr + len) = '/'; | |
67871254 | 2424 | read_extent_buffer(l, ptr, (unsigned long)(iref + 1), len); |
98d377a0 TH |
2425 | |
2426 | if (key.offset == BTRFS_FIRST_FREE_OBJECTID) | |
2427 | break; | |
2428 | ||
b3b4aa74 | 2429 | btrfs_release_path(path); |
98d377a0 | 2430 | key.objectid = key.offset; |
8ad6fcab | 2431 | key.offset = (u64)-1; |
98d377a0 | 2432 | dirid = key.objectid; |
98d377a0 | 2433 | } |
77906a50 | 2434 | memmove(name, ptr, total_len); |
67871254 | 2435 | name[total_len] = '\0'; |
98d377a0 TH |
2436 | ret = 0; |
2437 | out: | |
00246528 | 2438 | btrfs_put_root(root); |
98d377a0 | 2439 | btrfs_free_path(path); |
ac8e9819 CM |
2440 | return ret; |
2441 | } | |
2442 | ||
23d0b79d TM |
2443 | static int btrfs_search_path_in_tree_user(struct inode *inode, |
2444 | struct btrfs_ioctl_ino_lookup_user_args *args) | |
2445 | { | |
2446 | struct btrfs_fs_info *fs_info = BTRFS_I(inode)->root->fs_info; | |
2447 | struct super_block *sb = inode->i_sb; | |
2448 | struct btrfs_key upper_limit = BTRFS_I(inode)->location; | |
2449 | u64 treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2450 | u64 dirid = args->dirid; | |
2451 | unsigned long item_off; | |
2452 | unsigned long item_len; | |
2453 | struct btrfs_inode_ref *iref; | |
2454 | struct btrfs_root_ref *rref; | |
b8a49ae1 | 2455 | struct btrfs_root *root = NULL; |
23d0b79d TM |
2456 | struct btrfs_path *path; |
2457 | struct btrfs_key key, key2; | |
2458 | struct extent_buffer *leaf; | |
2459 | struct inode *temp_inode; | |
2460 | char *ptr; | |
2461 | int slot; | |
2462 | int len; | |
2463 | int total_len = 0; | |
2464 | int ret; | |
2465 | ||
2466 | path = btrfs_alloc_path(); | |
2467 | if (!path) | |
2468 | return -ENOMEM; | |
2469 | ||
2470 | /* | |
2471 | * If the bottom subvolume does not exist directly under upper_limit, | |
2472 | * construct the path in from the bottom up. | |
2473 | */ | |
2474 | if (dirid != upper_limit.objectid) { | |
2475 | ptr = &args->path[BTRFS_INO_LOOKUP_USER_PATH_MAX - 1]; | |
2476 | ||
56e9357a | 2477 | root = btrfs_get_fs_root(fs_info, treeid, true); |
23d0b79d TM |
2478 | if (IS_ERR(root)) { |
2479 | ret = PTR_ERR(root); | |
2480 | goto out; | |
2481 | } | |
2482 | ||
2483 | key.objectid = dirid; | |
2484 | key.type = BTRFS_INODE_REF_KEY; | |
2485 | key.offset = (u64)-1; | |
2486 | while (1) { | |
2487 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2488 | if (ret < 0) { | |
b8a49ae1 | 2489 | goto out_put; |
23d0b79d TM |
2490 | } else if (ret > 0) { |
2491 | ret = btrfs_previous_item(root, path, dirid, | |
2492 | BTRFS_INODE_REF_KEY); | |
2493 | if (ret < 0) { | |
b8a49ae1 | 2494 | goto out_put; |
23d0b79d TM |
2495 | } else if (ret > 0) { |
2496 | ret = -ENOENT; | |
b8a49ae1 | 2497 | goto out_put; |
23d0b79d TM |
2498 | } |
2499 | } | |
2500 | ||
2501 | leaf = path->nodes[0]; | |
2502 | slot = path->slots[0]; | |
2503 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2504 | ||
2505 | iref = btrfs_item_ptr(leaf, slot, struct btrfs_inode_ref); | |
2506 | len = btrfs_inode_ref_name_len(leaf, iref); | |
2507 | ptr -= len + 1; | |
2508 | total_len += len + 1; | |
2509 | if (ptr < args->path) { | |
2510 | ret = -ENAMETOOLONG; | |
b8a49ae1 | 2511 | goto out_put; |
23d0b79d TM |
2512 | } |
2513 | ||
2514 | *(ptr + len) = '/'; | |
2515 | read_extent_buffer(leaf, ptr, | |
2516 | (unsigned long)(iref + 1), len); | |
2517 | ||
2518 | /* Check the read+exec permission of this directory */ | |
2519 | ret = btrfs_previous_item(root, path, dirid, | |
2520 | BTRFS_INODE_ITEM_KEY); | |
2521 | if (ret < 0) { | |
b8a49ae1 | 2522 | goto out_put; |
23d0b79d TM |
2523 | } else if (ret > 0) { |
2524 | ret = -ENOENT; | |
b8a49ae1 | 2525 | goto out_put; |
23d0b79d TM |
2526 | } |
2527 | ||
2528 | leaf = path->nodes[0]; | |
2529 | slot = path->slots[0]; | |
2530 | btrfs_item_key_to_cpu(leaf, &key2, slot); | |
2531 | if (key2.objectid != dirid) { | |
2532 | ret = -ENOENT; | |
b8a49ae1 | 2533 | goto out_put; |
23d0b79d TM |
2534 | } |
2535 | ||
0202e83f | 2536 | temp_inode = btrfs_iget(sb, key2.objectid, root); |
3ca57bd6 MT |
2537 | if (IS_ERR(temp_inode)) { |
2538 | ret = PTR_ERR(temp_inode); | |
b8a49ae1 | 2539 | goto out_put; |
3ca57bd6 | 2540 | } |
23d0b79d TM |
2541 | ret = inode_permission(temp_inode, MAY_READ | MAY_EXEC); |
2542 | iput(temp_inode); | |
2543 | if (ret) { | |
2544 | ret = -EACCES; | |
b8a49ae1 | 2545 | goto out_put; |
23d0b79d TM |
2546 | } |
2547 | ||
2548 | if (key.offset == upper_limit.objectid) | |
2549 | break; | |
2550 | if (key.objectid == BTRFS_FIRST_FREE_OBJECTID) { | |
2551 | ret = -EACCES; | |
b8a49ae1 | 2552 | goto out_put; |
23d0b79d TM |
2553 | } |
2554 | ||
2555 | btrfs_release_path(path); | |
2556 | key.objectid = key.offset; | |
2557 | key.offset = (u64)-1; | |
2558 | dirid = key.objectid; | |
2559 | } | |
2560 | ||
2561 | memmove(args->path, ptr, total_len); | |
2562 | args->path[total_len] = '\0'; | |
00246528 | 2563 | btrfs_put_root(root); |
b8a49ae1 | 2564 | root = NULL; |
23d0b79d TM |
2565 | btrfs_release_path(path); |
2566 | } | |
2567 | ||
2568 | /* Get the bottom subvolume's name from ROOT_REF */ | |
23d0b79d TM |
2569 | key.objectid = treeid; |
2570 | key.type = BTRFS_ROOT_REF_KEY; | |
2571 | key.offset = args->treeid; | |
b8a49ae1 | 2572 | ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); |
23d0b79d TM |
2573 | if (ret < 0) { |
2574 | goto out; | |
2575 | } else if (ret > 0) { | |
2576 | ret = -ENOENT; | |
2577 | goto out; | |
2578 | } | |
2579 | ||
2580 | leaf = path->nodes[0]; | |
2581 | slot = path->slots[0]; | |
2582 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2583 | ||
2584 | item_off = btrfs_item_ptr_offset(leaf, slot); | |
2585 | item_len = btrfs_item_size_nr(leaf, slot); | |
2586 | /* Check if dirid in ROOT_REF corresponds to passed dirid */ | |
2587 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2588 | if (args->dirid != btrfs_root_ref_dirid(leaf, rref)) { | |
2589 | ret = -EINVAL; | |
2590 | goto out; | |
2591 | } | |
2592 | ||
2593 | /* Copy subvolume's name */ | |
2594 | item_off += sizeof(struct btrfs_root_ref); | |
2595 | item_len -= sizeof(struct btrfs_root_ref); | |
2596 | read_extent_buffer(leaf, args->name, item_off, item_len); | |
2597 | args->name[item_len] = 0; | |
2598 | ||
b8a49ae1 | 2599 | out_put: |
00246528 | 2600 | btrfs_put_root(root); |
23d0b79d TM |
2601 | out: |
2602 | btrfs_free_path(path); | |
2603 | return ret; | |
2604 | } | |
2605 | ||
ac8e9819 CM |
2606 | static noinline int btrfs_ioctl_ino_lookup(struct file *file, |
2607 | void __user *argp) | |
2608 | { | |
bece2e82 BVA |
2609 | struct btrfs_ioctl_ino_lookup_args *args; |
2610 | struct inode *inode; | |
01b810b8 | 2611 | int ret = 0; |
ac8e9819 | 2612 | |
2354d08f JL |
2613 | args = memdup_user(argp, sizeof(*args)); |
2614 | if (IS_ERR(args)) | |
2615 | return PTR_ERR(args); | |
c2b96929 | 2616 | |
496ad9aa | 2617 | inode = file_inode(file); |
ac8e9819 | 2618 | |
01b810b8 DS |
2619 | /* |
2620 | * Unprivileged query to obtain the containing subvolume root id. The | |
2621 | * path is reset so it's consistent with btrfs_search_path_in_tree. | |
2622 | */ | |
1b53ac4d CM |
2623 | if (args->treeid == 0) |
2624 | args->treeid = BTRFS_I(inode)->root->root_key.objectid; | |
2625 | ||
01b810b8 DS |
2626 | if (args->objectid == BTRFS_FIRST_FREE_OBJECTID) { |
2627 | args->name[0] = 0; | |
2628 | goto out; | |
2629 | } | |
2630 | ||
2631 | if (!capable(CAP_SYS_ADMIN)) { | |
2632 | ret = -EPERM; | |
2633 | goto out; | |
2634 | } | |
2635 | ||
ac8e9819 CM |
2636 | ret = btrfs_search_path_in_tree(BTRFS_I(inode)->root->fs_info, |
2637 | args->treeid, args->objectid, | |
2638 | args->name); | |
2639 | ||
01b810b8 | 2640 | out: |
ac8e9819 CM |
2641 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) |
2642 | ret = -EFAULT; | |
2643 | ||
2644 | kfree(args); | |
98d377a0 TH |
2645 | return ret; |
2646 | } | |
2647 | ||
23d0b79d TM |
2648 | /* |
2649 | * Version of ino_lookup ioctl (unprivileged) | |
2650 | * | |
2651 | * The main differences from ino_lookup ioctl are: | |
2652 | * | |
2653 | * 1. Read + Exec permission will be checked using inode_permission() during | |
2654 | * path construction. -EACCES will be returned in case of failure. | |
2655 | * 2. Path construction will be stopped at the inode number which corresponds | |
2656 | * to the fd with which this ioctl is called. If constructed path does not | |
2657 | * exist under fd's inode, -EACCES will be returned. | |
2658 | * 3. The name of bottom subvolume is also searched and filled. | |
2659 | */ | |
2660 | static int btrfs_ioctl_ino_lookup_user(struct file *file, void __user *argp) | |
2661 | { | |
2662 | struct btrfs_ioctl_ino_lookup_user_args *args; | |
2663 | struct inode *inode; | |
2664 | int ret; | |
2665 | ||
2666 | args = memdup_user(argp, sizeof(*args)); | |
2667 | if (IS_ERR(args)) | |
2668 | return PTR_ERR(args); | |
2669 | ||
2670 | inode = file_inode(file); | |
2671 | ||
2672 | if (args->dirid == BTRFS_FIRST_FREE_OBJECTID && | |
2673 | BTRFS_I(inode)->location.objectid != BTRFS_FIRST_FREE_OBJECTID) { | |
2674 | /* | |
2675 | * The subvolume does not exist under fd with which this is | |
2676 | * called | |
2677 | */ | |
2678 | kfree(args); | |
2679 | return -EACCES; | |
2680 | } | |
2681 | ||
2682 | ret = btrfs_search_path_in_tree_user(inode, args); | |
2683 | ||
2684 | if (ret == 0 && copy_to_user(argp, args, sizeof(*args))) | |
2685 | ret = -EFAULT; | |
2686 | ||
2687 | kfree(args); | |
2688 | return ret; | |
2689 | } | |
2690 | ||
b64ec075 TM |
2691 | /* Get the subvolume information in BTRFS_ROOT_ITEM and BTRFS_ROOT_BACKREF */ |
2692 | static int btrfs_ioctl_get_subvol_info(struct file *file, void __user *argp) | |
2693 | { | |
2694 | struct btrfs_ioctl_get_subvol_info_args *subvol_info; | |
2695 | struct btrfs_fs_info *fs_info; | |
2696 | struct btrfs_root *root; | |
2697 | struct btrfs_path *path; | |
2698 | struct btrfs_key key; | |
2699 | struct btrfs_root_item *root_item; | |
2700 | struct btrfs_root_ref *rref; | |
2701 | struct extent_buffer *leaf; | |
2702 | unsigned long item_off; | |
2703 | unsigned long item_len; | |
2704 | struct inode *inode; | |
2705 | int slot; | |
2706 | int ret = 0; | |
2707 | ||
2708 | path = btrfs_alloc_path(); | |
2709 | if (!path) | |
2710 | return -ENOMEM; | |
2711 | ||
2712 | subvol_info = kzalloc(sizeof(*subvol_info), GFP_KERNEL); | |
2713 | if (!subvol_info) { | |
2714 | btrfs_free_path(path); | |
2715 | return -ENOMEM; | |
2716 | } | |
2717 | ||
2718 | inode = file_inode(file); | |
2719 | fs_info = BTRFS_I(inode)->root->fs_info; | |
2720 | ||
2721 | /* Get root_item of inode's subvolume */ | |
2722 | key.objectid = BTRFS_I(inode)->root->root_key.objectid; | |
56e9357a | 2723 | root = btrfs_get_fs_root(fs_info, key.objectid, true); |
b64ec075 TM |
2724 | if (IS_ERR(root)) { |
2725 | ret = PTR_ERR(root); | |
04734e84 JB |
2726 | goto out_free; |
2727 | } | |
b64ec075 TM |
2728 | root_item = &root->root_item; |
2729 | ||
2730 | subvol_info->treeid = key.objectid; | |
2731 | ||
2732 | subvol_info->generation = btrfs_root_generation(root_item); | |
2733 | subvol_info->flags = btrfs_root_flags(root_item); | |
2734 | ||
2735 | memcpy(subvol_info->uuid, root_item->uuid, BTRFS_UUID_SIZE); | |
2736 | memcpy(subvol_info->parent_uuid, root_item->parent_uuid, | |
2737 | BTRFS_UUID_SIZE); | |
2738 | memcpy(subvol_info->received_uuid, root_item->received_uuid, | |
2739 | BTRFS_UUID_SIZE); | |
2740 | ||
2741 | subvol_info->ctransid = btrfs_root_ctransid(root_item); | |
2742 | subvol_info->ctime.sec = btrfs_stack_timespec_sec(&root_item->ctime); | |
2743 | subvol_info->ctime.nsec = btrfs_stack_timespec_nsec(&root_item->ctime); | |
2744 | ||
2745 | subvol_info->otransid = btrfs_root_otransid(root_item); | |
2746 | subvol_info->otime.sec = btrfs_stack_timespec_sec(&root_item->otime); | |
2747 | subvol_info->otime.nsec = btrfs_stack_timespec_nsec(&root_item->otime); | |
2748 | ||
2749 | subvol_info->stransid = btrfs_root_stransid(root_item); | |
2750 | subvol_info->stime.sec = btrfs_stack_timespec_sec(&root_item->stime); | |
2751 | subvol_info->stime.nsec = btrfs_stack_timespec_nsec(&root_item->stime); | |
2752 | ||
2753 | subvol_info->rtransid = btrfs_root_rtransid(root_item); | |
2754 | subvol_info->rtime.sec = btrfs_stack_timespec_sec(&root_item->rtime); | |
2755 | subvol_info->rtime.nsec = btrfs_stack_timespec_nsec(&root_item->rtime); | |
2756 | ||
2757 | if (key.objectid != BTRFS_FS_TREE_OBJECTID) { | |
2758 | /* Search root tree for ROOT_BACKREF of this subvolume */ | |
b64ec075 TM |
2759 | key.type = BTRFS_ROOT_BACKREF_KEY; |
2760 | key.offset = 0; | |
04734e84 | 2761 | ret = btrfs_search_slot(NULL, fs_info->tree_root, &key, path, 0, 0); |
b64ec075 TM |
2762 | if (ret < 0) { |
2763 | goto out; | |
2764 | } else if (path->slots[0] >= | |
2765 | btrfs_header_nritems(path->nodes[0])) { | |
04734e84 | 2766 | ret = btrfs_next_leaf(fs_info->tree_root, path); |
b64ec075 TM |
2767 | if (ret < 0) { |
2768 | goto out; | |
2769 | } else if (ret > 0) { | |
2770 | ret = -EUCLEAN; | |
2771 | goto out; | |
2772 | } | |
2773 | } | |
2774 | ||
2775 | leaf = path->nodes[0]; | |
2776 | slot = path->slots[0]; | |
2777 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2778 | if (key.objectid == subvol_info->treeid && | |
2779 | key.type == BTRFS_ROOT_BACKREF_KEY) { | |
2780 | subvol_info->parent_id = key.offset; | |
2781 | ||
2782 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2783 | subvol_info->dirid = btrfs_root_ref_dirid(leaf, rref); | |
2784 | ||
2785 | item_off = btrfs_item_ptr_offset(leaf, slot) | |
2786 | + sizeof(struct btrfs_root_ref); | |
2787 | item_len = btrfs_item_size_nr(leaf, slot) | |
2788 | - sizeof(struct btrfs_root_ref); | |
2789 | read_extent_buffer(leaf, subvol_info->name, | |
2790 | item_off, item_len); | |
2791 | } else { | |
2792 | ret = -ENOENT; | |
2793 | goto out; | |
2794 | } | |
2795 | } | |
2796 | ||
2797 | if (copy_to_user(argp, subvol_info, sizeof(*subvol_info))) | |
2798 | ret = -EFAULT; | |
2799 | ||
2800 | out: | |
00246528 | 2801 | btrfs_put_root(root); |
04734e84 | 2802 | out_free: |
b64ec075 | 2803 | btrfs_free_path(path); |
b091f7fe | 2804 | kfree(subvol_info); |
b64ec075 TM |
2805 | return ret; |
2806 | } | |
2807 | ||
42e4b520 TM |
2808 | /* |
2809 | * Return ROOT_REF information of the subvolume containing this inode | |
2810 | * except the subvolume name. | |
2811 | */ | |
2812 | static int btrfs_ioctl_get_subvol_rootref(struct file *file, void __user *argp) | |
2813 | { | |
2814 | struct btrfs_ioctl_get_subvol_rootref_args *rootrefs; | |
2815 | struct btrfs_root_ref *rref; | |
2816 | struct btrfs_root *root; | |
2817 | struct btrfs_path *path; | |
2818 | struct btrfs_key key; | |
2819 | struct extent_buffer *leaf; | |
2820 | struct inode *inode; | |
2821 | u64 objectid; | |
2822 | int slot; | |
2823 | int ret; | |
2824 | u8 found; | |
2825 | ||
2826 | path = btrfs_alloc_path(); | |
2827 | if (!path) | |
2828 | return -ENOMEM; | |
2829 | ||
2830 | rootrefs = memdup_user(argp, sizeof(*rootrefs)); | |
2831 | if (IS_ERR(rootrefs)) { | |
2832 | btrfs_free_path(path); | |
2833 | return PTR_ERR(rootrefs); | |
2834 | } | |
2835 | ||
2836 | inode = file_inode(file); | |
2837 | root = BTRFS_I(inode)->root->fs_info->tree_root; | |
2838 | objectid = BTRFS_I(inode)->root->root_key.objectid; | |
2839 | ||
2840 | key.objectid = objectid; | |
2841 | key.type = BTRFS_ROOT_REF_KEY; | |
2842 | key.offset = rootrefs->min_treeid; | |
2843 | found = 0; | |
2844 | ||
2845 | ret = btrfs_search_slot(NULL, root, &key, path, 0, 0); | |
2846 | if (ret < 0) { | |
2847 | goto out; | |
2848 | } else if (path->slots[0] >= | |
2849 | btrfs_header_nritems(path->nodes[0])) { | |
2850 | ret = btrfs_next_leaf(root, path); | |
2851 | if (ret < 0) { | |
2852 | goto out; | |
2853 | } else if (ret > 0) { | |
2854 | ret = -EUCLEAN; | |
2855 | goto out; | |
2856 | } | |
2857 | } | |
2858 | while (1) { | |
2859 | leaf = path->nodes[0]; | |
2860 | slot = path->slots[0]; | |
2861 | ||
2862 | btrfs_item_key_to_cpu(leaf, &key, slot); | |
2863 | if (key.objectid != objectid || key.type != BTRFS_ROOT_REF_KEY) { | |
2864 | ret = 0; | |
2865 | goto out; | |
2866 | } | |
2867 | ||
2868 | if (found == BTRFS_MAX_ROOTREF_BUFFER_NUM) { | |
2869 | ret = -EOVERFLOW; | |
2870 | goto out; | |
2871 | } | |
2872 | ||
2873 | rref = btrfs_item_ptr(leaf, slot, struct btrfs_root_ref); | |
2874 | rootrefs->rootref[found].treeid = key.offset; | |
2875 | rootrefs->rootref[found].dirid = | |
2876 | btrfs_root_ref_dirid(leaf, rref); | |
2877 | found++; | |
2878 | ||
2879 | ret = btrfs_next_item(root, path); | |
2880 | if (ret < 0) { | |
2881 | goto out; | |
2882 | } else if (ret > 0) { | |
2883 | ret = -EUCLEAN; | |
2884 | goto out; | |
2885 | } | |
2886 | } | |
2887 | ||
2888 | out: | |
2889 | if (!ret || ret == -EOVERFLOW) { | |
2890 | rootrefs->num_items = found; | |
2891 | /* update min_treeid for next search */ | |
2892 | if (found) | |
2893 | rootrefs->min_treeid = | |
2894 | rootrefs->rootref[found - 1].treeid + 1; | |
2895 | if (copy_to_user(argp, rootrefs, sizeof(*rootrefs))) | |
2896 | ret = -EFAULT; | |
2897 | } | |
2898 | ||
2899 | kfree(rootrefs); | |
2900 | btrfs_free_path(path); | |
2901 | ||
2902 | return ret; | |
2903 | } | |
2904 | ||
76dda93c | 2905 | static noinline int btrfs_ioctl_snap_destroy(struct file *file, |
949964c9 MPS |
2906 | void __user *arg, |
2907 | bool destroy_v2) | |
76dda93c | 2908 | { |
54563d41 | 2909 | struct dentry *parent = file->f_path.dentry; |
0b246afa | 2910 | struct btrfs_fs_info *fs_info = btrfs_sb(parent->d_sb); |
76dda93c | 2911 | struct dentry *dentry; |
2b0143b5 | 2912 | struct inode *dir = d_inode(parent); |
76dda93c YZ |
2913 | struct inode *inode; |
2914 | struct btrfs_root *root = BTRFS_I(dir)->root; | |
2915 | struct btrfs_root *dest = NULL; | |
949964c9 MPS |
2916 | struct btrfs_ioctl_vol_args *vol_args = NULL; |
2917 | struct btrfs_ioctl_vol_args_v2 *vol_args2 = NULL; | |
2918 | char *subvol_name, *subvol_name_ptr = NULL; | |
2919 | int subvol_namelen; | |
76dda93c | 2920 | int err = 0; |
949964c9 | 2921 | bool destroy_parent = false; |
76dda93c | 2922 | |
949964c9 MPS |
2923 | if (destroy_v2) { |
2924 | vol_args2 = memdup_user(arg, sizeof(*vol_args2)); | |
2925 | if (IS_ERR(vol_args2)) | |
2926 | return PTR_ERR(vol_args2); | |
325c50e3 | 2927 | |
949964c9 MPS |
2928 | if (vol_args2->flags & ~BTRFS_SUBVOL_DELETE_ARGS_MASK) { |
2929 | err = -EOPNOTSUPP; | |
2930 | goto out; | |
2931 | } | |
76dda93c | 2932 | |
949964c9 MPS |
2933 | /* |
2934 | * If SPEC_BY_ID is not set, we are looking for the subvolume by | |
2935 | * name, same as v1 currently does. | |
2936 | */ | |
2937 | if (!(vol_args2->flags & BTRFS_SUBVOL_SPEC_BY_ID)) { | |
2938 | vol_args2->name[BTRFS_SUBVOL_NAME_MAX] = 0; | |
2939 | subvol_name = vol_args2->name; | |
2940 | ||
2941 | err = mnt_want_write_file(file); | |
2942 | if (err) | |
2943 | goto out; | |
2944 | } else { | |
2945 | if (vol_args2->subvolid < BTRFS_FIRST_FREE_OBJECTID) { | |
2946 | err = -EINVAL; | |
2947 | goto out; | |
2948 | } | |
2949 | ||
2950 | err = mnt_want_write_file(file); | |
2951 | if (err) | |
2952 | goto out; | |
2953 | ||
2954 | dentry = btrfs_get_dentry(fs_info->sb, | |
2955 | BTRFS_FIRST_FREE_OBJECTID, | |
2956 | vol_args2->subvolid, 0, 0); | |
2957 | if (IS_ERR(dentry)) { | |
2958 | err = PTR_ERR(dentry); | |
2959 | goto out_drop_write; | |
2960 | } | |
2961 | ||
2962 | /* | |
2963 | * Change the default parent since the subvolume being | |
2964 | * deleted can be outside of the current mount point. | |
2965 | */ | |
2966 | parent = btrfs_get_parent(dentry); | |
2967 | ||
2968 | /* | |
2969 | * At this point dentry->d_name can point to '/' if the | |
2970 | * subvolume we want to destroy is outsite of the | |
2971 | * current mount point, so we need to release the | |
2972 | * current dentry and execute the lookup to return a new | |
2973 | * one with ->d_name pointing to the | |
2974 | * <mount point>/subvol_name. | |
2975 | */ | |
2976 | dput(dentry); | |
2977 | if (IS_ERR(parent)) { | |
2978 | err = PTR_ERR(parent); | |
2979 | goto out_drop_write; | |
2980 | } | |
2981 | dir = d_inode(parent); | |
2982 | ||
2983 | /* | |
2984 | * If v2 was used with SPEC_BY_ID, a new parent was | |
2985 | * allocated since the subvolume can be outside of the | |
2986 | * current mount point. Later on we need to release this | |
2987 | * new parent dentry. | |
2988 | */ | |
2989 | destroy_parent = true; | |
2990 | ||
2991 | subvol_name_ptr = btrfs_get_subvol_name_from_objectid( | |
2992 | fs_info, vol_args2->subvolid); | |
2993 | if (IS_ERR(subvol_name_ptr)) { | |
2994 | err = PTR_ERR(subvol_name_ptr); | |
2995 | goto free_parent; | |
2996 | } | |
2997 | /* subvol_name_ptr is already NULL termined */ | |
2998 | subvol_name = (char *)kbasename(subvol_name_ptr); | |
2999 | } | |
3000 | } else { | |
3001 | vol_args = memdup_user(arg, sizeof(*vol_args)); | |
3002 | if (IS_ERR(vol_args)) | |
3003 | return PTR_ERR(vol_args); | |
3004 | ||
3005 | vol_args->name[BTRFS_PATH_NAME_MAX] = 0; | |
3006 | subvol_name = vol_args->name; | |
3007 | ||
3008 | err = mnt_want_write_file(file); | |
3009 | if (err) | |
3010 | goto out; | |
76dda93c YZ |
3011 | } |
3012 | ||
949964c9 | 3013 | subvol_namelen = strlen(subvol_name); |
76dda93c | 3014 | |
949964c9 MPS |
3015 | if (strchr(subvol_name, '/') || |
3016 | strncmp(subvol_name, "..", subvol_namelen) == 0) { | |
3017 | err = -EINVAL; | |
3018 | goto free_subvol_name; | |
3019 | } | |
3020 | ||
3021 | if (!S_ISDIR(dir->i_mode)) { | |
3022 | err = -ENOTDIR; | |
3023 | goto free_subvol_name; | |
3024 | } | |
521e0546 | 3025 | |
00235411 AV |
3026 | err = down_write_killable_nested(&dir->i_rwsem, I_MUTEX_PARENT); |
3027 | if (err == -EINTR) | |
949964c9 MPS |
3028 | goto free_subvol_name; |
3029 | dentry = lookup_one_len(subvol_name, parent, subvol_namelen); | |
76dda93c YZ |
3030 | if (IS_ERR(dentry)) { |
3031 | err = PTR_ERR(dentry); | |
3032 | goto out_unlock_dir; | |
3033 | } | |
3034 | ||
2b0143b5 | 3035 | if (d_really_is_negative(dentry)) { |
76dda93c YZ |
3036 | err = -ENOENT; |
3037 | goto out_dput; | |
3038 | } | |
3039 | ||
2b0143b5 | 3040 | inode = d_inode(dentry); |
4260f7c7 | 3041 | dest = BTRFS_I(inode)->root; |
67871254 | 3042 | if (!capable(CAP_SYS_ADMIN)) { |
4260f7c7 SW |
3043 | /* |
3044 | * Regular user. Only allow this with a special mount | |
3045 | * option, when the user has write+exec access to the | |
3046 | * subvol root, and when rmdir(2) would have been | |
3047 | * allowed. | |
3048 | * | |
3049 | * Note that this is _not_ check that the subvol is | |
3050 | * empty or doesn't contain data that we wouldn't | |
3051 | * otherwise be able to delete. | |
3052 | * | |
3053 | * Users who want to delete empty subvols should try | |
3054 | * rmdir(2). | |
3055 | */ | |
3056 | err = -EPERM; | |
0b246afa | 3057 | if (!btrfs_test_opt(fs_info, USER_SUBVOL_RM_ALLOWED)) |
4260f7c7 SW |
3058 | goto out_dput; |
3059 | ||
3060 | /* | |
3061 | * Do not allow deletion if the parent dir is the same | |
3062 | * as the dir to be deleted. That means the ioctl | |
3063 | * must be called on the dentry referencing the root | |
3064 | * of the subvol, not a random directory contained | |
3065 | * within it. | |
3066 | */ | |
3067 | err = -EINVAL; | |
3068 | if (root == dest) | |
3069 | goto out_dput; | |
3070 | ||
3071 | err = inode_permission(inode, MAY_WRITE | MAY_EXEC); | |
3072 | if (err) | |
3073 | goto out_dput; | |
4260f7c7 SW |
3074 | } |
3075 | ||
5c39da5b MX |
3076 | /* check if subvolume may be deleted by a user */ |
3077 | err = btrfs_may_delete(dir, dentry, 1); | |
3078 | if (err) | |
3079 | goto out_dput; | |
3080 | ||
4a0cc7ca | 3081 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
76dda93c YZ |
3082 | err = -EINVAL; |
3083 | goto out_dput; | |
3084 | } | |
3085 | ||
5955102c | 3086 | inode_lock(inode); |
f60a2364 | 3087 | err = btrfs_delete_subvolume(dir, dentry); |
5955102c | 3088 | inode_unlock(inode); |
46008d9d AG |
3089 | if (!err) { |
3090 | fsnotify_rmdir(dir, dentry); | |
76dda93c | 3091 | d_delete(dentry); |
46008d9d | 3092 | } |
fa6ac876 | 3093 | |
76dda93c YZ |
3094 | out_dput: |
3095 | dput(dentry); | |
3096 | out_unlock_dir: | |
5955102c | 3097 | inode_unlock(dir); |
949964c9 MPS |
3098 | free_subvol_name: |
3099 | kfree(subvol_name_ptr); | |
3100 | free_parent: | |
3101 | if (destroy_parent) | |
3102 | dput(parent); | |
00235411 | 3103 | out_drop_write: |
2a79f17e | 3104 | mnt_drop_write_file(file); |
76dda93c | 3105 | out: |
949964c9 | 3106 | kfree(vol_args2); |
76dda93c YZ |
3107 | kfree(vol_args); |
3108 | return err; | |
3109 | } | |
3110 | ||
1e701a32 | 3111 | static int btrfs_ioctl_defrag(struct file *file, void __user *argp) |
f46b5a66 | 3112 | { |
496ad9aa | 3113 | struct inode *inode = file_inode(file); |
f46b5a66 | 3114 | struct btrfs_root *root = BTRFS_I(inode)->root; |
1e701a32 | 3115 | struct btrfs_ioctl_defrag_range_args *range; |
c146afad YZ |
3116 | int ret; |
3117 | ||
25122d15 ID |
3118 | ret = mnt_want_write_file(file); |
3119 | if (ret) | |
3120 | return ret; | |
b83cc969 | 3121 | |
25122d15 ID |
3122 | if (btrfs_root_readonly(root)) { |
3123 | ret = -EROFS; | |
3124 | goto out; | |
5ac00add | 3125 | } |
f46b5a66 CH |
3126 | |
3127 | switch (inode->i_mode & S_IFMT) { | |
3128 | case S_IFDIR: | |
e441d54d CM |
3129 | if (!capable(CAP_SYS_ADMIN)) { |
3130 | ret = -EPERM; | |
3131 | goto out; | |
3132 | } | |
de78b51a | 3133 | ret = btrfs_defrag_root(root); |
f46b5a66 CH |
3134 | break; |
3135 | case S_IFREG: | |
616d374e AB |
3136 | /* |
3137 | * Note that this does not check the file descriptor for write | |
3138 | * access. This prevents defragmenting executables that are | |
3139 | * running and allows defrag on files open in read-only mode. | |
3140 | */ | |
3141 | if (!capable(CAP_SYS_ADMIN) && | |
3142 | inode_permission(inode, MAY_WRITE)) { | |
3143 | ret = -EPERM; | |
e441d54d CM |
3144 | goto out; |
3145 | } | |
1e701a32 CM |
3146 | |
3147 | range = kzalloc(sizeof(*range), GFP_KERNEL); | |
3148 | if (!range) { | |
3149 | ret = -ENOMEM; | |
3150 | goto out; | |
3151 | } | |
3152 | ||
3153 | if (argp) { | |
3154 | if (copy_from_user(range, argp, | |
3155 | sizeof(*range))) { | |
3156 | ret = -EFAULT; | |
3157 | kfree(range); | |
683be16e | 3158 | goto out; |
1e701a32 CM |
3159 | } |
3160 | /* compression requires us to start the IO */ | |
3161 | if ((range->flags & BTRFS_DEFRAG_RANGE_COMPRESS)) { | |
3162 | range->flags |= BTRFS_DEFRAG_RANGE_START_IO; | |
3163 | range->extent_thresh = (u32)-1; | |
3164 | } | |
3165 | } else { | |
3166 | /* the rest are all set to zero by kzalloc */ | |
3167 | range->len = (u64)-1; | |
3168 | } | |
496ad9aa | 3169 | ret = btrfs_defrag_file(file_inode(file), file, |
7c829b72 | 3170 | range, BTRFS_OLDEST_GENERATION, 0); |
4cb5300b CM |
3171 | if (ret > 0) |
3172 | ret = 0; | |
1e701a32 | 3173 | kfree(range); |
f46b5a66 | 3174 | break; |
8929ecfa YZ |
3175 | default: |
3176 | ret = -EINVAL; | |
f46b5a66 | 3177 | } |
e441d54d | 3178 | out: |
25122d15 | 3179 | mnt_drop_write_file(file); |
e441d54d | 3180 | return ret; |
f46b5a66 CH |
3181 | } |
3182 | ||
2ff7e61e | 3183 | static long btrfs_ioctl_add_dev(struct btrfs_fs_info *fs_info, void __user *arg) |
f46b5a66 CH |
3184 | { |
3185 | struct btrfs_ioctl_vol_args *vol_args; | |
3186 | int ret; | |
3187 | ||
e441d54d CM |
3188 | if (!capable(CAP_SYS_ADMIN)) |
3189 | return -EPERM; | |
3190 | ||
c3e1f96c | 3191 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_ADD)) |
e57138b3 | 3192 | return BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
c9e9f97b | 3193 | |
dae7b665 | 3194 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
3195 | if (IS_ERR(vol_args)) { |
3196 | ret = PTR_ERR(vol_args); | |
3197 | goto out; | |
3198 | } | |
f46b5a66 | 3199 | |
5516e595 | 3200 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
2ff7e61e | 3201 | ret = btrfs_init_new_device(fs_info, vol_args->name); |
f46b5a66 | 3202 | |
43d20761 | 3203 | if (!ret) |
0b246afa | 3204 | btrfs_info(fs_info, "disk added %s", vol_args->name); |
43d20761 | 3205 | |
f46b5a66 | 3206 | kfree(vol_args); |
c9e9f97b | 3207 | out: |
c3e1f96c | 3208 | btrfs_exclop_finish(fs_info); |
f46b5a66 CH |
3209 | return ret; |
3210 | } | |
3211 | ||
6b526ed7 | 3212 | static long btrfs_ioctl_rm_dev_v2(struct file *file, void __user *arg) |
f46b5a66 | 3213 | { |
0b246afa JM |
3214 | struct inode *inode = file_inode(file); |
3215 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
6b526ed7 | 3216 | struct btrfs_ioctl_vol_args_v2 *vol_args; |
f46b5a66 CH |
3217 | int ret; |
3218 | ||
e441d54d CM |
3219 | if (!capable(CAP_SYS_ADMIN)) |
3220 | return -EPERM; | |
3221 | ||
da24927b MX |
3222 | ret = mnt_want_write_file(file); |
3223 | if (ret) | |
3224 | return ret; | |
c146afad | 3225 | |
dae7b665 | 3226 | vol_args = memdup_user(arg, sizeof(*vol_args)); |
c9e9f97b ID |
3227 | if (IS_ERR(vol_args)) { |
3228 | ret = PTR_ERR(vol_args); | |
c47ca32d | 3229 | goto err_drop; |
c9e9f97b | 3230 | } |
f46b5a66 | 3231 | |
748449cd | 3232 | if (vol_args->flags & ~BTRFS_DEVICE_REMOVE_ARGS_MASK) { |
fd4e994b OS |
3233 | ret = -EOPNOTSUPP; |
3234 | goto out; | |
3235 | } | |
f46b5a66 | 3236 | |
c3e1f96c | 3237 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REMOVE)) { |
183860f6 AJ |
3238 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
3239 | goto out; | |
3240 | } | |
3241 | ||
735654ea | 3242 | if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) { |
2ff7e61e | 3243 | ret = btrfs_rm_device(fs_info, NULL, vol_args->devid); |
6b526ed7 AJ |
3244 | } else { |
3245 | vol_args->name[BTRFS_SUBVOL_NAME_MAX] = '\0'; | |
2ff7e61e | 3246 | ret = btrfs_rm_device(fs_info, vol_args->name, 0); |
6b526ed7 | 3247 | } |
c3e1f96c | 3248 | btrfs_exclop_finish(fs_info); |
183860f6 | 3249 | |
6b526ed7 | 3250 | if (!ret) { |
735654ea | 3251 | if (vol_args->flags & BTRFS_DEVICE_SPEC_BY_ID) |
0b246afa | 3252 | btrfs_info(fs_info, "device deleted: id %llu", |
6b526ed7 AJ |
3253 | vol_args->devid); |
3254 | else | |
0b246afa | 3255 | btrfs_info(fs_info, "device deleted: %s", |
6b526ed7 AJ |
3256 | vol_args->name); |
3257 | } | |
183860f6 AJ |
3258 | out: |
3259 | kfree(vol_args); | |
c47ca32d | 3260 | err_drop: |
4ac20c70 | 3261 | mnt_drop_write_file(file); |
f46b5a66 CH |
3262 | return ret; |
3263 | } | |
3264 | ||
da24927b | 3265 | static long btrfs_ioctl_rm_dev(struct file *file, void __user *arg) |
f46b5a66 | 3266 | { |
0b246afa JM |
3267 | struct inode *inode = file_inode(file); |
3268 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
f46b5a66 CH |
3269 | struct btrfs_ioctl_vol_args *vol_args; |
3270 | int ret; | |
3271 | ||
e441d54d CM |
3272 | if (!capable(CAP_SYS_ADMIN)) |
3273 | return -EPERM; | |
3274 | ||
da24927b MX |
3275 | ret = mnt_want_write_file(file); |
3276 | if (ret) | |
3277 | return ret; | |
c146afad | 3278 | |
c3e1f96c | 3279 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REMOVE)) { |
183860f6 | 3280 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
58d7bbf8 DS |
3281 | goto out_drop_write; |
3282 | } | |
3283 | ||
3284 | vol_args = memdup_user(arg, sizeof(*vol_args)); | |
3285 | if (IS_ERR(vol_args)) { | |
3286 | ret = PTR_ERR(vol_args); | |
183860f6 AJ |
3287 | goto out; |
3288 | } | |
3289 | ||
58d7bbf8 | 3290 | vol_args->name[BTRFS_PATH_NAME_MAX] = '\0'; |
2ff7e61e | 3291 | ret = btrfs_rm_device(fs_info, vol_args->name, 0); |
183860f6 | 3292 | |
ec95d491 | 3293 | if (!ret) |
0b246afa | 3294 | btrfs_info(fs_info, "disk deleted %s", vol_args->name); |
183860f6 | 3295 | kfree(vol_args); |
58d7bbf8 | 3296 | out: |
c3e1f96c | 3297 | btrfs_exclop_finish(fs_info); |
58d7bbf8 | 3298 | out_drop_write: |
4ac20c70 | 3299 | mnt_drop_write_file(file); |
58d7bbf8 | 3300 | |
f46b5a66 CH |
3301 | return ret; |
3302 | } | |
3303 | ||
2ff7e61e JM |
3304 | static long btrfs_ioctl_fs_info(struct btrfs_fs_info *fs_info, |
3305 | void __user *arg) | |
475f6387 | 3306 | { |
027ed2f0 | 3307 | struct btrfs_ioctl_fs_info_args *fi_args; |
475f6387 | 3308 | struct btrfs_device *device; |
0b246afa | 3309 | struct btrfs_fs_devices *fs_devices = fs_info->fs_devices; |
137c5418 | 3310 | u64 flags_in; |
027ed2f0 | 3311 | int ret = 0; |
475f6387 | 3312 | |
137c5418 JT |
3313 | fi_args = memdup_user(arg, sizeof(*fi_args)); |
3314 | if (IS_ERR(fi_args)) | |
3315 | return PTR_ERR(fi_args); | |
3316 | ||
3317 | flags_in = fi_args->flags; | |
3318 | memset(fi_args, 0, sizeof(*fi_args)); | |
027ed2f0 | 3319 | |
d03262c7 | 3320 | rcu_read_lock(); |
027ed2f0 | 3321 | fi_args->num_devices = fs_devices->num_devices; |
475f6387 | 3322 | |
d03262c7 | 3323 | list_for_each_entry_rcu(device, &fs_devices->devices, dev_list) { |
027ed2f0 LZ |
3324 | if (device->devid > fi_args->max_id) |
3325 | fi_args->max_id = device->devid; | |
475f6387 | 3326 | } |
d03262c7 | 3327 | rcu_read_unlock(); |
475f6387 | 3328 | |
de37aa51 | 3329 | memcpy(&fi_args->fsid, fs_devices->fsid, sizeof(fi_args->fsid)); |
bea7eafd OS |
3330 | fi_args->nodesize = fs_info->nodesize; |
3331 | fi_args->sectorsize = fs_info->sectorsize; | |
3332 | fi_args->clone_alignment = fs_info->sectorsize; | |
80a773fb | 3333 | |
137c5418 JT |
3334 | if (flags_in & BTRFS_FS_INFO_FLAG_CSUM_INFO) { |
3335 | fi_args->csum_type = btrfs_super_csum_type(fs_info->super_copy); | |
3336 | fi_args->csum_size = btrfs_super_csum_size(fs_info->super_copy); | |
3337 | fi_args->flags |= BTRFS_FS_INFO_FLAG_CSUM_INFO; | |
3338 | } | |
3339 | ||
0fb408a5 JT |
3340 | if (flags_in & BTRFS_FS_INFO_FLAG_GENERATION) { |
3341 | fi_args->generation = fs_info->generation; | |
3342 | fi_args->flags |= BTRFS_FS_INFO_FLAG_GENERATION; | |
3343 | } | |
3344 | ||
49bac897 JT |
3345 | if (flags_in & BTRFS_FS_INFO_FLAG_METADATA_UUID) { |
3346 | memcpy(&fi_args->metadata_uuid, fs_devices->metadata_uuid, | |
3347 | sizeof(fi_args->metadata_uuid)); | |
3348 | fi_args->flags |= BTRFS_FS_INFO_FLAG_METADATA_UUID; | |
3349 | } | |
3350 | ||
027ed2f0 LZ |
3351 | if (copy_to_user(arg, fi_args, sizeof(*fi_args))) |
3352 | ret = -EFAULT; | |
475f6387 | 3353 | |
027ed2f0 LZ |
3354 | kfree(fi_args); |
3355 | return ret; | |
475f6387 JS |
3356 | } |
3357 | ||
2ff7e61e JM |
3358 | static long btrfs_ioctl_dev_info(struct btrfs_fs_info *fs_info, |
3359 | void __user *arg) | |
475f6387 JS |
3360 | { |
3361 | struct btrfs_ioctl_dev_info_args *di_args; | |
3362 | struct btrfs_device *dev; | |
475f6387 JS |
3363 | int ret = 0; |
3364 | char *s_uuid = NULL; | |
475f6387 | 3365 | |
475f6387 JS |
3366 | di_args = memdup_user(arg, sizeof(*di_args)); |
3367 | if (IS_ERR(di_args)) | |
3368 | return PTR_ERR(di_args); | |
3369 | ||
dd5f9615 | 3370 | if (!btrfs_is_empty_uuid(di_args->uuid)) |
475f6387 JS |
3371 | s_uuid = di_args->uuid; |
3372 | ||
c5593ca3 | 3373 | rcu_read_lock(); |
e4319cd9 | 3374 | dev = btrfs_find_device(fs_info->fs_devices, di_args->devid, s_uuid, |
b2598edf | 3375 | NULL); |
475f6387 JS |
3376 | |
3377 | if (!dev) { | |
3378 | ret = -ENODEV; | |
3379 | goto out; | |
3380 | } | |
3381 | ||
3382 | di_args->devid = dev->devid; | |
7cc8e58d MX |
3383 | di_args->bytes_used = btrfs_device_get_bytes_used(dev); |
3384 | di_args->total_bytes = btrfs_device_get_total_bytes(dev); | |
475f6387 | 3385 | memcpy(di_args->uuid, dev->uuid, sizeof(di_args->uuid)); |
a27202fb | 3386 | if (dev->name) { |
672d5990 MT |
3387 | strncpy(di_args->path, rcu_str_deref(dev->name), |
3388 | sizeof(di_args->path) - 1); | |
a27202fb JM |
3389 | di_args->path[sizeof(di_args->path) - 1] = 0; |
3390 | } else { | |
99ba55ad | 3391 | di_args->path[0] = '\0'; |
a27202fb | 3392 | } |
475f6387 JS |
3393 | |
3394 | out: | |
c5593ca3 | 3395 | rcu_read_unlock(); |
475f6387 JS |
3396 | if (ret == 0 && copy_to_user(arg, di_args, sizeof(*di_args))) |
3397 | ret = -EFAULT; | |
3398 | ||
3399 | kfree(di_args); | |
3400 | return ret; | |
3401 | } | |
3402 | ||
6ef5ed0d JB |
3403 | static long btrfs_ioctl_default_subvol(struct file *file, void __user *argp) |
3404 | { | |
496ad9aa | 3405 | struct inode *inode = file_inode(file); |
0b246afa | 3406 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
6ef5ed0d JB |
3407 | struct btrfs_root *root = BTRFS_I(inode)->root; |
3408 | struct btrfs_root *new_root; | |
3409 | struct btrfs_dir_item *di; | |
3410 | struct btrfs_trans_handle *trans; | |
2a2b5d62 | 3411 | struct btrfs_path *path = NULL; |
6ef5ed0d | 3412 | struct btrfs_disk_key disk_key; |
6ef5ed0d JB |
3413 | u64 objectid = 0; |
3414 | u64 dir_id; | |
3c04ce01 | 3415 | int ret; |
6ef5ed0d JB |
3416 | |
3417 | if (!capable(CAP_SYS_ADMIN)) | |
3418 | return -EPERM; | |
3419 | ||
3c04ce01 MX |
3420 | ret = mnt_want_write_file(file); |
3421 | if (ret) | |
3422 | return ret; | |
3423 | ||
3424 | if (copy_from_user(&objectid, argp, sizeof(objectid))) { | |
3425 | ret = -EFAULT; | |
3426 | goto out; | |
3427 | } | |
6ef5ed0d JB |
3428 | |
3429 | if (!objectid) | |
1cecf579 | 3430 | objectid = BTRFS_FS_TREE_OBJECTID; |
6ef5ed0d | 3431 | |
56e9357a | 3432 | new_root = btrfs_get_fs_root(fs_info, objectid, true); |
3c04ce01 MX |
3433 | if (IS_ERR(new_root)) { |
3434 | ret = PTR_ERR(new_root); | |
3435 | goto out; | |
3436 | } | |
2a2b5d62 JB |
3437 | if (!is_fstree(new_root->root_key.objectid)) { |
3438 | ret = -ENOENT; | |
3439 | goto out_free; | |
3440 | } | |
6ef5ed0d | 3441 | |
6ef5ed0d | 3442 | path = btrfs_alloc_path(); |
3c04ce01 MX |
3443 | if (!path) { |
3444 | ret = -ENOMEM; | |
2a2b5d62 | 3445 | goto out_free; |
3c04ce01 | 3446 | } |
6ef5ed0d JB |
3447 | |
3448 | trans = btrfs_start_transaction(root, 1); | |
98d5dc13 | 3449 | if (IS_ERR(trans)) { |
3c04ce01 | 3450 | ret = PTR_ERR(trans); |
2a2b5d62 | 3451 | goto out_free; |
6ef5ed0d JB |
3452 | } |
3453 | ||
0b246afa JM |
3454 | dir_id = btrfs_super_root_dir(fs_info->super_copy); |
3455 | di = btrfs_lookup_dir_item(trans, fs_info->tree_root, path, | |
6ef5ed0d | 3456 | dir_id, "default", 7, 1); |
cf1e99a4 | 3457 | if (IS_ERR_OR_NULL(di)) { |
2a2b5d62 | 3458 | btrfs_release_path(path); |
3a45bb20 | 3459 | btrfs_end_transaction(trans); |
0b246afa | 3460 | btrfs_err(fs_info, |
5d163e0e | 3461 | "Umm, you don't have the default diritem, this isn't going to work"); |
3c04ce01 | 3462 | ret = -ENOENT; |
2a2b5d62 | 3463 | goto out_free; |
6ef5ed0d JB |
3464 | } |
3465 | ||
3466 | btrfs_cpu_key_to_disk(&disk_key, &new_root->root_key); | |
3467 | btrfs_set_dir_item_key(path->nodes[0], di, &disk_key); | |
3468 | btrfs_mark_buffer_dirty(path->nodes[0]); | |
2a2b5d62 | 3469 | btrfs_release_path(path); |
6ef5ed0d | 3470 | |
0b246afa | 3471 | btrfs_set_fs_incompat(fs_info, DEFAULT_SUBVOL); |
3a45bb20 | 3472 | btrfs_end_transaction(trans); |
2a2b5d62 | 3473 | out_free: |
00246528 | 3474 | btrfs_put_root(new_root); |
2a2b5d62 | 3475 | btrfs_free_path(path); |
3c04ce01 MX |
3476 | out: |
3477 | mnt_drop_write_file(file); | |
3478 | return ret; | |
6ef5ed0d JB |
3479 | } |
3480 | ||
c065f5b1 SY |
3481 | static void get_block_group_info(struct list_head *groups_list, |
3482 | struct btrfs_ioctl_space_info *space) | |
bf5fc093 | 3483 | { |
32da5386 | 3484 | struct btrfs_block_group *block_group; |
bf5fc093 JB |
3485 | |
3486 | space->total_bytes = 0; | |
3487 | space->used_bytes = 0; | |
3488 | space->flags = 0; | |
3489 | list_for_each_entry(block_group, groups_list, list) { | |
3490 | space->flags = block_group->flags; | |
b3470b5d | 3491 | space->total_bytes += block_group->length; |
bf38be65 | 3492 | space->used_bytes += block_group->used; |
bf5fc093 JB |
3493 | } |
3494 | } | |
3495 | ||
2ff7e61e JM |
3496 | static long btrfs_ioctl_space_info(struct btrfs_fs_info *fs_info, |
3497 | void __user *arg) | |
1406e432 JB |
3498 | { |
3499 | struct btrfs_ioctl_space_args space_args; | |
3500 | struct btrfs_ioctl_space_info space; | |
3501 | struct btrfs_ioctl_space_info *dest; | |
7fde62bf | 3502 | struct btrfs_ioctl_space_info *dest_orig; |
13f2696f | 3503 | struct btrfs_ioctl_space_info __user *user_dest; |
1406e432 | 3504 | struct btrfs_space_info *info; |
315d8e98 CIK |
3505 | static const u64 types[] = { |
3506 | BTRFS_BLOCK_GROUP_DATA, | |
3507 | BTRFS_BLOCK_GROUP_SYSTEM, | |
3508 | BTRFS_BLOCK_GROUP_METADATA, | |
3509 | BTRFS_BLOCK_GROUP_DATA | BTRFS_BLOCK_GROUP_METADATA | |
3510 | }; | |
bf5fc093 | 3511 | int num_types = 4; |
7fde62bf | 3512 | int alloc_size; |
1406e432 | 3513 | int ret = 0; |
51788b1b | 3514 | u64 slot_count = 0; |
bf5fc093 | 3515 | int i, c; |
1406e432 JB |
3516 | |
3517 | if (copy_from_user(&space_args, | |
3518 | (struct btrfs_ioctl_space_args __user *)arg, | |
3519 | sizeof(space_args))) | |
3520 | return -EFAULT; | |
3521 | ||
bf5fc093 JB |
3522 | for (i = 0; i < num_types; i++) { |
3523 | struct btrfs_space_info *tmp; | |
3524 | ||
3525 | info = NULL; | |
72804905 | 3526 | list_for_each_entry(tmp, &fs_info->space_info, list) { |
bf5fc093 JB |
3527 | if (tmp->flags == types[i]) { |
3528 | info = tmp; | |
3529 | break; | |
3530 | } | |
3531 | } | |
bf5fc093 JB |
3532 | |
3533 | if (!info) | |
3534 | continue; | |
3535 | ||
3536 | down_read(&info->groups_sem); | |
3537 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3538 | if (!list_empty(&info->block_groups[c])) | |
3539 | slot_count++; | |
3540 | } | |
3541 | up_read(&info->groups_sem); | |
3542 | } | |
7fde62bf | 3543 | |
36523e95 DS |
3544 | /* |
3545 | * Global block reserve, exported as a space_info | |
3546 | */ | |
3547 | slot_count++; | |
3548 | ||
7fde62bf CM |
3549 | /* space_slots == 0 means they are asking for a count */ |
3550 | if (space_args.space_slots == 0) { | |
3551 | space_args.total_spaces = slot_count; | |
3552 | goto out; | |
3553 | } | |
bf5fc093 | 3554 | |
51788b1b | 3555 | slot_count = min_t(u64, space_args.space_slots, slot_count); |
bf5fc093 | 3556 | |
7fde62bf | 3557 | alloc_size = sizeof(*dest) * slot_count; |
bf5fc093 | 3558 | |
7fde62bf CM |
3559 | /* we generally have at most 6 or so space infos, one for each raid |
3560 | * level. So, a whole page should be more than enough for everyone | |
3561 | */ | |
09cbfeaf | 3562 | if (alloc_size > PAGE_SIZE) |
7fde62bf CM |
3563 | return -ENOMEM; |
3564 | ||
1406e432 | 3565 | space_args.total_spaces = 0; |
8d2db785 | 3566 | dest = kmalloc(alloc_size, GFP_KERNEL); |
7fde62bf CM |
3567 | if (!dest) |
3568 | return -ENOMEM; | |
3569 | dest_orig = dest; | |
1406e432 | 3570 | |
7fde62bf | 3571 | /* now we have a buffer to copy into */ |
bf5fc093 JB |
3572 | for (i = 0; i < num_types; i++) { |
3573 | struct btrfs_space_info *tmp; | |
3574 | ||
51788b1b DR |
3575 | if (!slot_count) |
3576 | break; | |
3577 | ||
bf5fc093 | 3578 | info = NULL; |
72804905 | 3579 | list_for_each_entry(tmp, &fs_info->space_info, list) { |
bf5fc093 JB |
3580 | if (tmp->flags == types[i]) { |
3581 | info = tmp; | |
3582 | break; | |
3583 | } | |
3584 | } | |
7fde62bf | 3585 | |
bf5fc093 JB |
3586 | if (!info) |
3587 | continue; | |
3588 | down_read(&info->groups_sem); | |
3589 | for (c = 0; c < BTRFS_NR_RAID_TYPES; c++) { | |
3590 | if (!list_empty(&info->block_groups[c])) { | |
c065f5b1 SY |
3591 | get_block_group_info(&info->block_groups[c], |
3592 | &space); | |
bf5fc093 JB |
3593 | memcpy(dest, &space, sizeof(space)); |
3594 | dest++; | |
3595 | space_args.total_spaces++; | |
51788b1b | 3596 | slot_count--; |
bf5fc093 | 3597 | } |
51788b1b DR |
3598 | if (!slot_count) |
3599 | break; | |
bf5fc093 JB |
3600 | } |
3601 | up_read(&info->groups_sem); | |
1406e432 | 3602 | } |
1406e432 | 3603 | |
36523e95 DS |
3604 | /* |
3605 | * Add global block reserve | |
3606 | */ | |
3607 | if (slot_count) { | |
0b246afa | 3608 | struct btrfs_block_rsv *block_rsv = &fs_info->global_block_rsv; |
36523e95 DS |
3609 | |
3610 | spin_lock(&block_rsv->lock); | |
3611 | space.total_bytes = block_rsv->size; | |
3612 | space.used_bytes = block_rsv->size - block_rsv->reserved; | |
3613 | spin_unlock(&block_rsv->lock); | |
3614 | space.flags = BTRFS_SPACE_INFO_GLOBAL_RSV; | |
3615 | memcpy(dest, &space, sizeof(space)); | |
3616 | space_args.total_spaces++; | |
3617 | } | |
3618 | ||
2eec6c81 | 3619 | user_dest = (struct btrfs_ioctl_space_info __user *) |
7fde62bf CM |
3620 | (arg + sizeof(struct btrfs_ioctl_space_args)); |
3621 | ||
3622 | if (copy_to_user(user_dest, dest_orig, alloc_size)) | |
3623 | ret = -EFAULT; | |
3624 | ||
3625 | kfree(dest_orig); | |
3626 | out: | |
3627 | if (ret == 0 && copy_to_user(arg, &space_args, sizeof(space_args))) | |
1406e432 JB |
3628 | ret = -EFAULT; |
3629 | ||
3630 | return ret; | |
3631 | } | |
3632 | ||
9a8c28be MX |
3633 | static noinline long btrfs_ioctl_start_sync(struct btrfs_root *root, |
3634 | void __user *argp) | |
46204592 | 3635 | { |
46204592 SW |
3636 | struct btrfs_trans_handle *trans; |
3637 | u64 transid; | |
db5b493a | 3638 | int ret; |
46204592 | 3639 | |
d4edf39b | 3640 | trans = btrfs_attach_transaction_barrier(root); |
ff7c1d33 MX |
3641 | if (IS_ERR(trans)) { |
3642 | if (PTR_ERR(trans) != -ENOENT) | |
3643 | return PTR_ERR(trans); | |
3644 | ||
3645 | /* No running transaction, don't bother */ | |
3646 | transid = root->fs_info->last_trans_committed; | |
3647 | goto out; | |
3648 | } | |
46204592 | 3649 | transid = trans->transid; |
3a45bb20 | 3650 | ret = btrfs_commit_transaction_async(trans, 0); |
8b2b2d3c | 3651 | if (ret) { |
3a45bb20 | 3652 | btrfs_end_transaction(trans); |
db5b493a | 3653 | return ret; |
8b2b2d3c | 3654 | } |
ff7c1d33 | 3655 | out: |
46204592 SW |
3656 | if (argp) |
3657 | if (copy_to_user(argp, &transid, sizeof(transid))) | |
3658 | return -EFAULT; | |
3659 | return 0; | |
3660 | } | |
3661 | ||
2ff7e61e | 3662 | static noinline long btrfs_ioctl_wait_sync(struct btrfs_fs_info *fs_info, |
9a8c28be | 3663 | void __user *argp) |
46204592 | 3664 | { |
46204592 SW |
3665 | u64 transid; |
3666 | ||
3667 | if (argp) { | |
3668 | if (copy_from_user(&transid, argp, sizeof(transid))) | |
3669 | return -EFAULT; | |
3670 | } else { | |
3671 | transid = 0; /* current trans */ | |
3672 | } | |
2ff7e61e | 3673 | return btrfs_wait_for_commit(fs_info, transid); |
46204592 SW |
3674 | } |
3675 | ||
b8e95489 | 3676 | static long btrfs_ioctl_scrub(struct file *file, void __user *arg) |
475f6387 | 3677 | { |
0b246afa | 3678 | struct btrfs_fs_info *fs_info = btrfs_sb(file_inode(file)->i_sb); |
475f6387 | 3679 | struct btrfs_ioctl_scrub_args *sa; |
b8e95489 | 3680 | int ret; |
475f6387 JS |
3681 | |
3682 | if (!capable(CAP_SYS_ADMIN)) | |
3683 | return -EPERM; | |
3684 | ||
3685 | sa = memdup_user(arg, sizeof(*sa)); | |
3686 | if (IS_ERR(sa)) | |
3687 | return PTR_ERR(sa); | |
3688 | ||
b8e95489 MX |
3689 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) { |
3690 | ret = mnt_want_write_file(file); | |
3691 | if (ret) | |
3692 | goto out; | |
3693 | } | |
3694 | ||
0b246afa | 3695 | ret = btrfs_scrub_dev(fs_info, sa->devid, sa->start, sa->end, |
63a212ab SB |
3696 | &sa->progress, sa->flags & BTRFS_SCRUB_READONLY, |
3697 | 0); | |
475f6387 | 3698 | |
5afe6ce7 FM |
3699 | /* |
3700 | * Copy scrub args to user space even if btrfs_scrub_dev() returned an | |
3701 | * error. This is important as it allows user space to know how much | |
3702 | * progress scrub has done. For example, if scrub is canceled we get | |
3703 | * -ECANCELED from btrfs_scrub_dev() and return that error back to user | |
3704 | * space. Later user space can inspect the progress from the structure | |
3705 | * btrfs_ioctl_scrub_args and resume scrub from where it left off | |
3706 | * previously (btrfs-progs does this). | |
3707 | * If we fail to copy the btrfs_ioctl_scrub_args structure to user space | |
3708 | * then return -EFAULT to signal the structure was not copied or it may | |
3709 | * be corrupt and unreliable due to a partial copy. | |
3710 | */ | |
3711 | if (copy_to_user(arg, sa, sizeof(*sa))) | |
475f6387 JS |
3712 | ret = -EFAULT; |
3713 | ||
b8e95489 MX |
3714 | if (!(sa->flags & BTRFS_SCRUB_READONLY)) |
3715 | mnt_drop_write_file(file); | |
3716 | out: | |
475f6387 JS |
3717 | kfree(sa); |
3718 | return ret; | |
3719 | } | |
3720 | ||
2ff7e61e | 3721 | static long btrfs_ioctl_scrub_cancel(struct btrfs_fs_info *fs_info) |
475f6387 JS |
3722 | { |
3723 | if (!capable(CAP_SYS_ADMIN)) | |
3724 | return -EPERM; | |
3725 | ||
2ff7e61e | 3726 | return btrfs_scrub_cancel(fs_info); |
475f6387 JS |
3727 | } |
3728 | ||
2ff7e61e | 3729 | static long btrfs_ioctl_scrub_progress(struct btrfs_fs_info *fs_info, |
475f6387 JS |
3730 | void __user *arg) |
3731 | { | |
3732 | struct btrfs_ioctl_scrub_args *sa; | |
3733 | int ret; | |
3734 | ||
3735 | if (!capable(CAP_SYS_ADMIN)) | |
3736 | return -EPERM; | |
3737 | ||
3738 | sa = memdup_user(arg, sizeof(*sa)); | |
3739 | if (IS_ERR(sa)) | |
3740 | return PTR_ERR(sa); | |
3741 | ||
2ff7e61e | 3742 | ret = btrfs_scrub_progress(fs_info, sa->devid, &sa->progress); |
475f6387 | 3743 | |
4fa99b00 | 3744 | if (ret == 0 && copy_to_user(arg, sa, sizeof(*sa))) |
475f6387 JS |
3745 | ret = -EFAULT; |
3746 | ||
3747 | kfree(sa); | |
3748 | return ret; | |
3749 | } | |
3750 | ||
2ff7e61e | 3751 | static long btrfs_ioctl_get_dev_stats(struct btrfs_fs_info *fs_info, |
b27f7c0c | 3752 | void __user *arg) |
c11d2c23 SB |
3753 | { |
3754 | struct btrfs_ioctl_get_dev_stats *sa; | |
3755 | int ret; | |
3756 | ||
c11d2c23 SB |
3757 | sa = memdup_user(arg, sizeof(*sa)); |
3758 | if (IS_ERR(sa)) | |
3759 | return PTR_ERR(sa); | |
3760 | ||
b27f7c0c DS |
3761 | if ((sa->flags & BTRFS_DEV_STATS_RESET) && !capable(CAP_SYS_ADMIN)) { |
3762 | kfree(sa); | |
3763 | return -EPERM; | |
3764 | } | |
3765 | ||
2ff7e61e | 3766 | ret = btrfs_get_dev_stats(fs_info, sa); |
c11d2c23 | 3767 | |
eee99577 | 3768 | if (ret == 0 && copy_to_user(arg, sa, sizeof(*sa))) |
c11d2c23 SB |
3769 | ret = -EFAULT; |
3770 | ||
3771 | kfree(sa); | |
3772 | return ret; | |
3773 | } | |
3774 | ||
2ff7e61e JM |
3775 | static long btrfs_ioctl_dev_replace(struct btrfs_fs_info *fs_info, |
3776 | void __user *arg) | |
3f6bcfbd SB |
3777 | { |
3778 | struct btrfs_ioctl_dev_replace_args *p; | |
3779 | int ret; | |
3780 | ||
3781 | if (!capable(CAP_SYS_ADMIN)) | |
3782 | return -EPERM; | |
3783 | ||
3784 | p = memdup_user(arg, sizeof(*p)); | |
3785 | if (IS_ERR(p)) | |
3786 | return PTR_ERR(p); | |
3787 | ||
3788 | switch (p->cmd) { | |
3789 | case BTRFS_IOCTL_DEV_REPLACE_CMD_START: | |
bc98a42c | 3790 | if (sb_rdonly(fs_info->sb)) { |
adfa97cb ID |
3791 | ret = -EROFS; |
3792 | goto out; | |
3793 | } | |
c3e1f96c | 3794 | if (!btrfs_exclop_start(fs_info, BTRFS_EXCLOP_DEV_REPLACE)) { |
e57138b3 | 3795 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
3f6bcfbd | 3796 | } else { |
2ff7e61e | 3797 | ret = btrfs_dev_replace_by_ioctl(fs_info, p); |
c3e1f96c | 3798 | btrfs_exclop_finish(fs_info); |
3f6bcfbd SB |
3799 | } |
3800 | break; | |
3801 | case BTRFS_IOCTL_DEV_REPLACE_CMD_STATUS: | |
0b246afa | 3802 | btrfs_dev_replace_status(fs_info, p); |
3f6bcfbd SB |
3803 | ret = 0; |
3804 | break; | |
3805 | case BTRFS_IOCTL_DEV_REPLACE_CMD_CANCEL: | |
17d202b9 | 3806 | p->result = btrfs_dev_replace_cancel(fs_info); |
97282031 | 3807 | ret = 0; |
3f6bcfbd SB |
3808 | break; |
3809 | default: | |
3810 | ret = -EINVAL; | |
3811 | break; | |
3812 | } | |
3813 | ||
d3a53286 | 3814 | if ((ret == 0 || ret == -ECANCELED) && copy_to_user(arg, p, sizeof(*p))) |
3f6bcfbd | 3815 | ret = -EFAULT; |
adfa97cb | 3816 | out: |
3f6bcfbd SB |
3817 | kfree(p); |
3818 | return ret; | |
3819 | } | |
3820 | ||
d7728c96 JS |
3821 | static long btrfs_ioctl_ino_to_path(struct btrfs_root *root, void __user *arg) |
3822 | { | |
3823 | int ret = 0; | |
3824 | int i; | |
740c3d22 | 3825 | u64 rel_ptr; |
d7728c96 | 3826 | int size; |
806468f8 | 3827 | struct btrfs_ioctl_ino_path_args *ipa = NULL; |
d7728c96 JS |
3828 | struct inode_fs_paths *ipath = NULL; |
3829 | struct btrfs_path *path; | |
3830 | ||
82b22ac8 | 3831 | if (!capable(CAP_DAC_READ_SEARCH)) |
d7728c96 JS |
3832 | return -EPERM; |
3833 | ||
3834 | path = btrfs_alloc_path(); | |
3835 | if (!path) { | |
3836 | ret = -ENOMEM; | |
3837 | goto out; | |
3838 | } | |
3839 | ||
3840 | ipa = memdup_user(arg, sizeof(*ipa)); | |
3841 | if (IS_ERR(ipa)) { | |
3842 | ret = PTR_ERR(ipa); | |
3843 | ipa = NULL; | |
3844 | goto out; | |
3845 | } | |
3846 | ||
3847 | size = min_t(u32, ipa->size, 4096); | |
3848 | ipath = init_ipath(size, root, path); | |
3849 | if (IS_ERR(ipath)) { | |
3850 | ret = PTR_ERR(ipath); | |
3851 | ipath = NULL; | |
3852 | goto out; | |
3853 | } | |
3854 | ||
3855 | ret = paths_from_inode(ipa->inum, ipath); | |
3856 | if (ret < 0) | |
3857 | goto out; | |
3858 | ||
3859 | for (i = 0; i < ipath->fspath->elem_cnt; ++i) { | |
745c4d8e JM |
3860 | rel_ptr = ipath->fspath->val[i] - |
3861 | (u64)(unsigned long)ipath->fspath->val; | |
740c3d22 | 3862 | ipath->fspath->val[i] = rel_ptr; |
d7728c96 JS |
3863 | } |
3864 | ||
718dc5fa OS |
3865 | ret = copy_to_user((void __user *)(unsigned long)ipa->fspath, |
3866 | ipath->fspath, size); | |
d7728c96 JS |
3867 | if (ret) { |
3868 | ret = -EFAULT; | |
3869 | goto out; | |
3870 | } | |
3871 | ||
3872 | out: | |
3873 | btrfs_free_path(path); | |
3874 | free_ipath(ipath); | |
3875 | kfree(ipa); | |
3876 | ||
3877 | return ret; | |
3878 | } | |
3879 | ||
3880 | static int build_ino_list(u64 inum, u64 offset, u64 root, void *ctx) | |
3881 | { | |
3882 | struct btrfs_data_container *inodes = ctx; | |
3883 | const size_t c = 3 * sizeof(u64); | |
3884 | ||
3885 | if (inodes->bytes_left >= c) { | |
3886 | inodes->bytes_left -= c; | |
3887 | inodes->val[inodes->elem_cnt] = inum; | |
3888 | inodes->val[inodes->elem_cnt + 1] = offset; | |
3889 | inodes->val[inodes->elem_cnt + 2] = root; | |
3890 | inodes->elem_cnt += 3; | |
3891 | } else { | |
3892 | inodes->bytes_missing += c - inodes->bytes_left; | |
3893 | inodes->bytes_left = 0; | |
3894 | inodes->elem_missed += 3; | |
3895 | } | |
3896 | ||
3897 | return 0; | |
3898 | } | |
3899 | ||
2ff7e61e | 3900 | static long btrfs_ioctl_logical_to_ino(struct btrfs_fs_info *fs_info, |
d24a67b2 | 3901 | void __user *arg, int version) |
d7728c96 JS |
3902 | { |
3903 | int ret = 0; | |
3904 | int size; | |
d7728c96 JS |
3905 | struct btrfs_ioctl_logical_ino_args *loi; |
3906 | struct btrfs_data_container *inodes = NULL; | |
3907 | struct btrfs_path *path = NULL; | |
d24a67b2 | 3908 | bool ignore_offset; |
d7728c96 JS |
3909 | |
3910 | if (!capable(CAP_SYS_ADMIN)) | |
3911 | return -EPERM; | |
3912 | ||
3913 | loi = memdup_user(arg, sizeof(*loi)); | |
7b9ea627 SV |
3914 | if (IS_ERR(loi)) |
3915 | return PTR_ERR(loi); | |
d7728c96 | 3916 | |
d24a67b2 ZB |
3917 | if (version == 1) { |
3918 | ignore_offset = false; | |
b115e3bc | 3919 | size = min_t(u32, loi->size, SZ_64K); |
d24a67b2 ZB |
3920 | } else { |
3921 | /* All reserved bits must be 0 for now */ | |
3922 | if (memchr_inv(loi->reserved, 0, sizeof(loi->reserved))) { | |
3923 | ret = -EINVAL; | |
3924 | goto out_loi; | |
3925 | } | |
3926 | /* Only accept flags we have defined so far */ | |
3927 | if (loi->flags & ~(BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET)) { | |
3928 | ret = -EINVAL; | |
3929 | goto out_loi; | |
3930 | } | |
3931 | ignore_offset = loi->flags & BTRFS_LOGICAL_INO_ARGS_IGNORE_OFFSET; | |
b115e3bc | 3932 | size = min_t(u32, loi->size, SZ_16M); |
d24a67b2 ZB |
3933 | } |
3934 | ||
d7728c96 JS |
3935 | path = btrfs_alloc_path(); |
3936 | if (!path) { | |
3937 | ret = -ENOMEM; | |
3938 | goto out; | |
3939 | } | |
3940 | ||
d7728c96 JS |
3941 | inodes = init_data_container(size); |
3942 | if (IS_ERR(inodes)) { | |
3943 | ret = PTR_ERR(inodes); | |
3944 | inodes = NULL; | |
3945 | goto out; | |
3946 | } | |
3947 | ||
2ff7e61e | 3948 | ret = iterate_inodes_from_logical(loi->logical, fs_info, path, |
d24a67b2 | 3949 | build_ino_list, inodes, ignore_offset); |
df031f07 | 3950 | if (ret == -EINVAL) |
d7728c96 JS |
3951 | ret = -ENOENT; |
3952 | if (ret < 0) | |
3953 | goto out; | |
3954 | ||
718dc5fa OS |
3955 | ret = copy_to_user((void __user *)(unsigned long)loi->inodes, inodes, |
3956 | size); | |
d7728c96 JS |
3957 | if (ret) |
3958 | ret = -EFAULT; | |
3959 | ||
3960 | out: | |
3961 | btrfs_free_path(path); | |
f54de068 | 3962 | kvfree(inodes); |
d24a67b2 | 3963 | out_loi: |
d7728c96 JS |
3964 | kfree(loi); |
3965 | ||
3966 | return ret; | |
3967 | } | |
3968 | ||
008ef096 | 3969 | void btrfs_update_ioctl_balance_args(struct btrfs_fs_info *fs_info, |
c9e9f97b ID |
3970 | struct btrfs_ioctl_balance_args *bargs) |
3971 | { | |
3972 | struct btrfs_balance_control *bctl = fs_info->balance_ctl; | |
3973 | ||
3974 | bargs->flags = bctl->flags; | |
3975 | ||
3009a62f | 3976 | if (test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) |
837d5b6e ID |
3977 | bargs->state |= BTRFS_BALANCE_STATE_RUNNING; |
3978 | if (atomic_read(&fs_info->balance_pause_req)) | |
3979 | bargs->state |= BTRFS_BALANCE_STATE_PAUSE_REQ; | |
a7e99c69 ID |
3980 | if (atomic_read(&fs_info->balance_cancel_req)) |
3981 | bargs->state |= BTRFS_BALANCE_STATE_CANCEL_REQ; | |
837d5b6e | 3982 | |
c9e9f97b ID |
3983 | memcpy(&bargs->data, &bctl->data, sizeof(bargs->data)); |
3984 | memcpy(&bargs->meta, &bctl->meta, sizeof(bargs->meta)); | |
3985 | memcpy(&bargs->sys, &bctl->sys, sizeof(bargs->sys)); | |
19a39dce | 3986 | |
008ef096 DS |
3987 | spin_lock(&fs_info->balance_lock); |
3988 | memcpy(&bargs->stat, &bctl->stat, sizeof(bargs->stat)); | |
3989 | spin_unlock(&fs_info->balance_lock); | |
c9e9f97b ID |
3990 | } |
3991 | ||
9ba1f6e4 | 3992 | static long btrfs_ioctl_balance(struct file *file, void __user *arg) |
c9e9f97b | 3993 | { |
496ad9aa | 3994 | struct btrfs_root *root = BTRFS_I(file_inode(file))->root; |
c9e9f97b ID |
3995 | struct btrfs_fs_info *fs_info = root->fs_info; |
3996 | struct btrfs_ioctl_balance_args *bargs; | |
3997 | struct btrfs_balance_control *bctl; | |
ed0fb78f | 3998 | bool need_unlock; /* for mut. excl. ops lock */ |
c9e9f97b ID |
3999 | int ret; |
4000 | ||
4001 | if (!capable(CAP_SYS_ADMIN)) | |
4002 | return -EPERM; | |
4003 | ||
e54bfa31 | 4004 | ret = mnt_want_write_file(file); |
9ba1f6e4 LB |
4005 | if (ret) |
4006 | return ret; | |
4007 | ||
ed0fb78f | 4008 | again: |
c3e1f96c | 4009 | if (btrfs_exclop_start(fs_info, BTRFS_EXCLOP_BALANCE)) { |
ed0fb78f ID |
4010 | mutex_lock(&fs_info->balance_mutex); |
4011 | need_unlock = true; | |
4012 | goto locked; | |
4013 | } | |
4014 | ||
4015 | /* | |
01327610 | 4016 | * mut. excl. ops lock is locked. Three possibilities: |
ed0fb78f ID |
4017 | * (1) some other op is running |
4018 | * (2) balance is running | |
4019 | * (3) balance is paused -- special case (think resume) | |
4020 | */ | |
c9e9f97b | 4021 | mutex_lock(&fs_info->balance_mutex); |
ed0fb78f ID |
4022 | if (fs_info->balance_ctl) { |
4023 | /* this is either (2) or (3) */ | |
3009a62f | 4024 | if (!test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) { |
ed0fb78f | 4025 | mutex_unlock(&fs_info->balance_mutex); |
dccdb07b DS |
4026 | /* |
4027 | * Lock released to allow other waiters to continue, | |
4028 | * we'll reexamine the status again. | |
4029 | */ | |
ed0fb78f ID |
4030 | mutex_lock(&fs_info->balance_mutex); |
4031 | ||
4032 | if (fs_info->balance_ctl && | |
3009a62f | 4033 | !test_bit(BTRFS_FS_BALANCE_RUNNING, &fs_info->flags)) { |
ed0fb78f ID |
4034 | /* this is (3) */ |
4035 | need_unlock = false; | |
4036 | goto locked; | |
4037 | } | |
4038 | ||
4039 | mutex_unlock(&fs_info->balance_mutex); | |
ed0fb78f ID |
4040 | goto again; |
4041 | } else { | |
4042 | /* this is (2) */ | |
4043 | mutex_unlock(&fs_info->balance_mutex); | |
4044 | ret = -EINPROGRESS; | |
4045 | goto out; | |
4046 | } | |
4047 | } else { | |
4048 | /* this is (1) */ | |
4049 | mutex_unlock(&fs_info->balance_mutex); | |
e57138b3 | 4050 | ret = BTRFS_ERROR_DEV_EXCL_RUN_IN_PROGRESS; |
ed0fb78f ID |
4051 | goto out; |
4052 | } | |
4053 | ||
4054 | locked: | |
c9e9f97b ID |
4055 | |
4056 | if (arg) { | |
4057 | bargs = memdup_user(arg, sizeof(*bargs)); | |
4058 | if (IS_ERR(bargs)) { | |
4059 | ret = PTR_ERR(bargs); | |
ed0fb78f | 4060 | goto out_unlock; |
c9e9f97b | 4061 | } |
de322263 ID |
4062 | |
4063 | if (bargs->flags & BTRFS_BALANCE_RESUME) { | |
4064 | if (!fs_info->balance_ctl) { | |
4065 | ret = -ENOTCONN; | |
4066 | goto out_bargs; | |
4067 | } | |
4068 | ||
4069 | bctl = fs_info->balance_ctl; | |
4070 | spin_lock(&fs_info->balance_lock); | |
4071 | bctl->flags |= BTRFS_BALANCE_RESUME; | |
4072 | spin_unlock(&fs_info->balance_lock); | |
4073 | ||
4074 | goto do_balance; | |
4075 | } | |
c9e9f97b ID |
4076 | } else { |
4077 | bargs = NULL; | |
4078 | } | |
4079 | ||
ed0fb78f | 4080 | if (fs_info->balance_ctl) { |
837d5b6e ID |
4081 | ret = -EINPROGRESS; |
4082 | goto out_bargs; | |
4083 | } | |
4084 | ||
8d2db785 | 4085 | bctl = kzalloc(sizeof(*bctl), GFP_KERNEL); |
c9e9f97b ID |
4086 | if (!bctl) { |
4087 | ret = -ENOMEM; | |
4088 | goto out_bargs; | |
4089 | } | |
4090 | ||
c9e9f97b ID |
4091 | if (arg) { |
4092 | memcpy(&bctl->data, &bargs->data, sizeof(bctl->data)); | |
4093 | memcpy(&bctl->meta, &bargs->meta, sizeof(bctl->meta)); | |
4094 | memcpy(&bctl->sys, &bargs->sys, sizeof(bctl->sys)); | |
4095 | ||
4096 | bctl->flags = bargs->flags; | |
f43ffb60 ID |
4097 | } else { |
4098 | /* balance everything - no filters */ | |
4099 | bctl->flags |= BTRFS_BALANCE_TYPE_MASK; | |
c9e9f97b ID |
4100 | } |
4101 | ||
8eb93459 DS |
4102 | if (bctl->flags & ~(BTRFS_BALANCE_ARGS_MASK | BTRFS_BALANCE_TYPE_MASK)) { |
4103 | ret = -EINVAL; | |
0f89abf5 | 4104 | goto out_bctl; |
8eb93459 DS |
4105 | } |
4106 | ||
de322263 | 4107 | do_balance: |
c9e9f97b | 4108 | /* |
c3e1f96c GR |
4109 | * Ownership of bctl and exclusive operation goes to btrfs_balance. |
4110 | * bctl is freed in reset_balance_state, or, if restriper was paused | |
4111 | * all the way until unmount, in free_fs_info. The flag should be | |
4112 | * cleared after reset_balance_state. | |
c9e9f97b | 4113 | */ |
ed0fb78f ID |
4114 | need_unlock = false; |
4115 | ||
6fcf6e2b | 4116 | ret = btrfs_balance(fs_info, bctl, bargs); |
0f89abf5 | 4117 | bctl = NULL; |
ed0fb78f | 4118 | |
d00c2d9c | 4119 | if ((ret == 0 || ret == -ECANCELED) && arg) { |
c9e9f97b ID |
4120 | if (copy_to_user(arg, bargs, sizeof(*bargs))) |
4121 | ret = -EFAULT; | |
4122 | } | |
4123 | ||
0f89abf5 CE |
4124 | out_bctl: |
4125 | kfree(bctl); | |
c9e9f97b ID |
4126 | out_bargs: |
4127 | kfree(bargs); | |
ed0fb78f | 4128 | out_unlock: |
c9e9f97b | 4129 | mutex_unlock(&fs_info->balance_mutex); |
ed0fb78f | 4130 | if (need_unlock) |
c3e1f96c | 4131 | btrfs_exclop_finish(fs_info); |
ed0fb78f | 4132 | out: |
e54bfa31 | 4133 | mnt_drop_write_file(file); |
c9e9f97b ID |
4134 | return ret; |
4135 | } | |
4136 | ||
2ff7e61e | 4137 | static long btrfs_ioctl_balance_ctl(struct btrfs_fs_info *fs_info, int cmd) |
837d5b6e ID |
4138 | { |
4139 | if (!capable(CAP_SYS_ADMIN)) | |
4140 | return -EPERM; | |
4141 | ||
4142 | switch (cmd) { | |
4143 | case BTRFS_BALANCE_CTL_PAUSE: | |
0b246afa | 4144 | return btrfs_pause_balance(fs_info); |
a7e99c69 | 4145 | case BTRFS_BALANCE_CTL_CANCEL: |
0b246afa | 4146 | return btrfs_cancel_balance(fs_info); |
837d5b6e ID |
4147 | } |
4148 | ||
4149 | return -EINVAL; | |
4150 | } | |
4151 | ||
2ff7e61e | 4152 | static long btrfs_ioctl_balance_progress(struct btrfs_fs_info *fs_info, |
19a39dce ID |
4153 | void __user *arg) |
4154 | { | |
19a39dce ID |
4155 | struct btrfs_ioctl_balance_args *bargs; |
4156 | int ret = 0; | |
4157 | ||
4158 | if (!capable(CAP_SYS_ADMIN)) | |
4159 | return -EPERM; | |
4160 | ||
4161 | mutex_lock(&fs_info->balance_mutex); | |
4162 | if (!fs_info->balance_ctl) { | |
4163 | ret = -ENOTCONN; | |
4164 | goto out; | |
4165 | } | |
4166 | ||
8d2db785 | 4167 | bargs = kzalloc(sizeof(*bargs), GFP_KERNEL); |
19a39dce ID |
4168 | if (!bargs) { |
4169 | ret = -ENOMEM; | |
4170 | goto out; | |
4171 | } | |
4172 | ||
008ef096 | 4173 | btrfs_update_ioctl_balance_args(fs_info, bargs); |
19a39dce ID |
4174 | |
4175 | if (copy_to_user(arg, bargs, sizeof(*bargs))) | |
4176 | ret = -EFAULT; | |
4177 | ||
4178 | kfree(bargs); | |
4179 | out: | |
4180 | mutex_unlock(&fs_info->balance_mutex); | |
4181 | return ret; | |
4182 | } | |
4183 | ||
905b0dda | 4184 | static long btrfs_ioctl_quota_ctl(struct file *file, void __user *arg) |
5d13a37b | 4185 | { |
0b246afa JM |
4186 | struct inode *inode = file_inode(file); |
4187 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
5d13a37b | 4188 | struct btrfs_ioctl_quota_ctl_args *sa; |
5d13a37b | 4189 | int ret; |
5d13a37b AJ |
4190 | |
4191 | if (!capable(CAP_SYS_ADMIN)) | |
4192 | return -EPERM; | |
4193 | ||
905b0dda MX |
4194 | ret = mnt_want_write_file(file); |
4195 | if (ret) | |
4196 | return ret; | |
5d13a37b AJ |
4197 | |
4198 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4199 | if (IS_ERR(sa)) { |
4200 | ret = PTR_ERR(sa); | |
4201 | goto drop_write; | |
4202 | } | |
5d13a37b | 4203 | |
0b246afa | 4204 | down_write(&fs_info->subvol_sem); |
5d13a37b AJ |
4205 | |
4206 | switch (sa->cmd) { | |
4207 | case BTRFS_QUOTA_CTL_ENABLE: | |
340f1aa2 | 4208 | ret = btrfs_quota_enable(fs_info); |
5d13a37b AJ |
4209 | break; |
4210 | case BTRFS_QUOTA_CTL_DISABLE: | |
340f1aa2 | 4211 | ret = btrfs_quota_disable(fs_info); |
5d13a37b | 4212 | break; |
5d13a37b AJ |
4213 | default: |
4214 | ret = -EINVAL; | |
4215 | break; | |
4216 | } | |
4217 | ||
5d13a37b | 4218 | kfree(sa); |
0b246afa | 4219 | up_write(&fs_info->subvol_sem); |
905b0dda MX |
4220 | drop_write: |
4221 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4222 | return ret; |
4223 | } | |
4224 | ||
905b0dda | 4225 | static long btrfs_ioctl_qgroup_assign(struct file *file, void __user *arg) |
5d13a37b | 4226 | { |
0b246afa JM |
4227 | struct inode *inode = file_inode(file); |
4228 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4229 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
5d13a37b AJ |
4230 | struct btrfs_ioctl_qgroup_assign_args *sa; |
4231 | struct btrfs_trans_handle *trans; | |
4232 | int ret; | |
4233 | int err; | |
4234 | ||
4235 | if (!capable(CAP_SYS_ADMIN)) | |
4236 | return -EPERM; | |
4237 | ||
905b0dda MX |
4238 | ret = mnt_want_write_file(file); |
4239 | if (ret) | |
4240 | return ret; | |
5d13a37b AJ |
4241 | |
4242 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4243 | if (IS_ERR(sa)) { |
4244 | ret = PTR_ERR(sa); | |
4245 | goto drop_write; | |
4246 | } | |
5d13a37b AJ |
4247 | |
4248 | trans = btrfs_join_transaction(root); | |
4249 | if (IS_ERR(trans)) { | |
4250 | ret = PTR_ERR(trans); | |
4251 | goto out; | |
4252 | } | |
4253 | ||
5d13a37b | 4254 | if (sa->assign) { |
9f8a6ce6 | 4255 | ret = btrfs_add_qgroup_relation(trans, sa->src, sa->dst); |
5d13a37b | 4256 | } else { |
39616c27 | 4257 | ret = btrfs_del_qgroup_relation(trans, sa->src, sa->dst); |
5d13a37b AJ |
4258 | } |
4259 | ||
e082f563 | 4260 | /* update qgroup status and info */ |
280f8bd2 | 4261 | err = btrfs_run_qgroups(trans); |
e082f563 | 4262 | if (err < 0) |
0b246afa JM |
4263 | btrfs_handle_fs_error(fs_info, err, |
4264 | "failed to update qgroup status and info"); | |
3a45bb20 | 4265 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4266 | if (err && !ret) |
4267 | ret = err; | |
4268 | ||
4269 | out: | |
4270 | kfree(sa); | |
905b0dda MX |
4271 | drop_write: |
4272 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4273 | return ret; |
4274 | } | |
4275 | ||
905b0dda | 4276 | static long btrfs_ioctl_qgroup_create(struct file *file, void __user *arg) |
5d13a37b | 4277 | { |
0b246afa | 4278 | struct inode *inode = file_inode(file); |
0b246afa | 4279 | struct btrfs_root *root = BTRFS_I(inode)->root; |
5d13a37b AJ |
4280 | struct btrfs_ioctl_qgroup_create_args *sa; |
4281 | struct btrfs_trans_handle *trans; | |
4282 | int ret; | |
4283 | int err; | |
4284 | ||
4285 | if (!capable(CAP_SYS_ADMIN)) | |
4286 | return -EPERM; | |
4287 | ||
905b0dda MX |
4288 | ret = mnt_want_write_file(file); |
4289 | if (ret) | |
4290 | return ret; | |
5d13a37b AJ |
4291 | |
4292 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4293 | if (IS_ERR(sa)) { |
4294 | ret = PTR_ERR(sa); | |
4295 | goto drop_write; | |
4296 | } | |
5d13a37b | 4297 | |
d86e56cf MX |
4298 | if (!sa->qgroupid) { |
4299 | ret = -EINVAL; | |
4300 | goto out; | |
4301 | } | |
4302 | ||
5d13a37b AJ |
4303 | trans = btrfs_join_transaction(root); |
4304 | if (IS_ERR(trans)) { | |
4305 | ret = PTR_ERR(trans); | |
4306 | goto out; | |
4307 | } | |
4308 | ||
5d13a37b | 4309 | if (sa->create) { |
49a05ecd | 4310 | ret = btrfs_create_qgroup(trans, sa->qgroupid); |
5d13a37b | 4311 | } else { |
3efbee1d | 4312 | ret = btrfs_remove_qgroup(trans, sa->qgroupid); |
5d13a37b AJ |
4313 | } |
4314 | ||
3a45bb20 | 4315 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4316 | if (err && !ret) |
4317 | ret = err; | |
4318 | ||
4319 | out: | |
4320 | kfree(sa); | |
905b0dda MX |
4321 | drop_write: |
4322 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4323 | return ret; |
4324 | } | |
4325 | ||
905b0dda | 4326 | static long btrfs_ioctl_qgroup_limit(struct file *file, void __user *arg) |
5d13a37b | 4327 | { |
0b246afa | 4328 | struct inode *inode = file_inode(file); |
0b246afa | 4329 | struct btrfs_root *root = BTRFS_I(inode)->root; |
5d13a37b AJ |
4330 | struct btrfs_ioctl_qgroup_limit_args *sa; |
4331 | struct btrfs_trans_handle *trans; | |
4332 | int ret; | |
4333 | int err; | |
4334 | u64 qgroupid; | |
4335 | ||
4336 | if (!capable(CAP_SYS_ADMIN)) | |
4337 | return -EPERM; | |
4338 | ||
905b0dda MX |
4339 | ret = mnt_want_write_file(file); |
4340 | if (ret) | |
4341 | return ret; | |
5d13a37b AJ |
4342 | |
4343 | sa = memdup_user(arg, sizeof(*sa)); | |
905b0dda MX |
4344 | if (IS_ERR(sa)) { |
4345 | ret = PTR_ERR(sa); | |
4346 | goto drop_write; | |
4347 | } | |
5d13a37b AJ |
4348 | |
4349 | trans = btrfs_join_transaction(root); | |
4350 | if (IS_ERR(trans)) { | |
4351 | ret = PTR_ERR(trans); | |
4352 | goto out; | |
4353 | } | |
4354 | ||
4355 | qgroupid = sa->qgroupid; | |
4356 | if (!qgroupid) { | |
4357 | /* take the current subvol as qgroup */ | |
4358 | qgroupid = root->root_key.objectid; | |
4359 | } | |
4360 | ||
f0042d5e | 4361 | ret = btrfs_limit_qgroup(trans, qgroupid, &sa->lim); |
5d13a37b | 4362 | |
3a45bb20 | 4363 | err = btrfs_end_transaction(trans); |
5d13a37b AJ |
4364 | if (err && !ret) |
4365 | ret = err; | |
4366 | ||
4367 | out: | |
4368 | kfree(sa); | |
905b0dda MX |
4369 | drop_write: |
4370 | mnt_drop_write_file(file); | |
5d13a37b AJ |
4371 | return ret; |
4372 | } | |
4373 | ||
2f232036 JS |
4374 | static long btrfs_ioctl_quota_rescan(struct file *file, void __user *arg) |
4375 | { | |
0b246afa JM |
4376 | struct inode *inode = file_inode(file); |
4377 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
2f232036 JS |
4378 | struct btrfs_ioctl_quota_rescan_args *qsa; |
4379 | int ret; | |
4380 | ||
4381 | if (!capable(CAP_SYS_ADMIN)) | |
4382 | return -EPERM; | |
4383 | ||
4384 | ret = mnt_want_write_file(file); | |
4385 | if (ret) | |
4386 | return ret; | |
4387 | ||
4388 | qsa = memdup_user(arg, sizeof(*qsa)); | |
4389 | if (IS_ERR(qsa)) { | |
4390 | ret = PTR_ERR(qsa); | |
4391 | goto drop_write; | |
4392 | } | |
4393 | ||
4394 | if (qsa->flags) { | |
4395 | ret = -EINVAL; | |
4396 | goto out; | |
4397 | } | |
4398 | ||
0b246afa | 4399 | ret = btrfs_qgroup_rescan(fs_info); |
2f232036 JS |
4400 | |
4401 | out: | |
4402 | kfree(qsa); | |
4403 | drop_write: | |
4404 | mnt_drop_write_file(file); | |
4405 | return ret; | |
4406 | } | |
4407 | ||
b929c1d8 MPS |
4408 | static long btrfs_ioctl_quota_rescan_status(struct btrfs_fs_info *fs_info, |
4409 | void __user *arg) | |
2f232036 | 4410 | { |
2f232036 JS |
4411 | struct btrfs_ioctl_quota_rescan_args *qsa; |
4412 | int ret = 0; | |
4413 | ||
4414 | if (!capable(CAP_SYS_ADMIN)) | |
4415 | return -EPERM; | |
4416 | ||
8d2db785 | 4417 | qsa = kzalloc(sizeof(*qsa), GFP_KERNEL); |
2f232036 JS |
4418 | if (!qsa) |
4419 | return -ENOMEM; | |
4420 | ||
0b246afa | 4421 | if (fs_info->qgroup_flags & BTRFS_QGROUP_STATUS_FLAG_RESCAN) { |
2f232036 | 4422 | qsa->flags = 1; |
0b246afa | 4423 | qsa->progress = fs_info->qgroup_rescan_progress.objectid; |
2f232036 JS |
4424 | } |
4425 | ||
4426 | if (copy_to_user(arg, qsa, sizeof(*qsa))) | |
4427 | ret = -EFAULT; | |
4428 | ||
4429 | kfree(qsa); | |
4430 | return ret; | |
4431 | } | |
4432 | ||
b929c1d8 MPS |
4433 | static long btrfs_ioctl_quota_rescan_wait(struct btrfs_fs_info *fs_info, |
4434 | void __user *arg) | |
57254b6e | 4435 | { |
57254b6e JS |
4436 | if (!capable(CAP_SYS_ADMIN)) |
4437 | return -EPERM; | |
4438 | ||
0b246afa | 4439 | return btrfs_qgroup_wait_for_completion(fs_info, true); |
57254b6e JS |
4440 | } |
4441 | ||
abccd00f HM |
4442 | static long _btrfs_ioctl_set_received_subvol(struct file *file, |
4443 | struct btrfs_ioctl_received_subvol_args *sa) | |
8ea05e3a | 4444 | { |
496ad9aa | 4445 | struct inode *inode = file_inode(file); |
0b246afa | 4446 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); |
8ea05e3a AB |
4447 | struct btrfs_root *root = BTRFS_I(inode)->root; |
4448 | struct btrfs_root_item *root_item = &root->root_item; | |
4449 | struct btrfs_trans_handle *trans; | |
95582b00 | 4450 | struct timespec64 ct = current_time(inode); |
8ea05e3a | 4451 | int ret = 0; |
dd5f9615 | 4452 | int received_uuid_changed; |
8ea05e3a | 4453 | |
bd60ea0f DS |
4454 | if (!inode_owner_or_capable(inode)) |
4455 | return -EPERM; | |
4456 | ||
8ea05e3a AB |
4457 | ret = mnt_want_write_file(file); |
4458 | if (ret < 0) | |
4459 | return ret; | |
4460 | ||
0b246afa | 4461 | down_write(&fs_info->subvol_sem); |
8ea05e3a | 4462 | |
4a0cc7ca | 4463 | if (btrfs_ino(BTRFS_I(inode)) != BTRFS_FIRST_FREE_OBJECTID) { |
8ea05e3a AB |
4464 | ret = -EINVAL; |
4465 | goto out; | |
4466 | } | |
4467 | ||
4468 | if (btrfs_root_readonly(root)) { | |
4469 | ret = -EROFS; | |
4470 | goto out; | |
4471 | } | |
4472 | ||
dd5f9615 SB |
4473 | /* |
4474 | * 1 - root item | |
4475 | * 2 - uuid items (received uuid + subvol uuid) | |
4476 | */ | |
4477 | trans = btrfs_start_transaction(root, 3); | |
8ea05e3a AB |
4478 | if (IS_ERR(trans)) { |
4479 | ret = PTR_ERR(trans); | |
4480 | trans = NULL; | |
4481 | goto out; | |
4482 | } | |
4483 | ||
4484 | sa->rtransid = trans->transid; | |
4485 | sa->rtime.sec = ct.tv_sec; | |
4486 | sa->rtime.nsec = ct.tv_nsec; | |
4487 | ||
dd5f9615 SB |
4488 | received_uuid_changed = memcmp(root_item->received_uuid, sa->uuid, |
4489 | BTRFS_UUID_SIZE); | |
4490 | if (received_uuid_changed && | |
d87ff758 | 4491 | !btrfs_is_empty_uuid(root_item->received_uuid)) { |
d1957791 | 4492 | ret = btrfs_uuid_tree_remove(trans, root_item->received_uuid, |
d87ff758 NB |
4493 | BTRFS_UUID_KEY_RECEIVED_SUBVOL, |
4494 | root->root_key.objectid); | |
4495 | if (ret && ret != -ENOENT) { | |
4496 | btrfs_abort_transaction(trans, ret); | |
4497 | btrfs_end_transaction(trans); | |
4498 | goto out; | |
4499 | } | |
4500 | } | |
8ea05e3a AB |
4501 | memcpy(root_item->received_uuid, sa->uuid, BTRFS_UUID_SIZE); |
4502 | btrfs_set_root_stransid(root_item, sa->stransid); | |
4503 | btrfs_set_root_rtransid(root_item, sa->rtransid); | |
3cae210f QW |
4504 | btrfs_set_stack_timespec_sec(&root_item->stime, sa->stime.sec); |
4505 | btrfs_set_stack_timespec_nsec(&root_item->stime, sa->stime.nsec); | |
4506 | btrfs_set_stack_timespec_sec(&root_item->rtime, sa->rtime.sec); | |
4507 | btrfs_set_stack_timespec_nsec(&root_item->rtime, sa->rtime.nsec); | |
8ea05e3a | 4508 | |
0b246afa | 4509 | ret = btrfs_update_root(trans, fs_info->tree_root, |
8ea05e3a AB |
4510 | &root->root_key, &root->root_item); |
4511 | if (ret < 0) { | |
3a45bb20 | 4512 | btrfs_end_transaction(trans); |
8ea05e3a | 4513 | goto out; |
dd5f9615 SB |
4514 | } |
4515 | if (received_uuid_changed && !btrfs_is_empty_uuid(sa->uuid)) { | |
cdb345a8 | 4516 | ret = btrfs_uuid_tree_add(trans, sa->uuid, |
dd5f9615 SB |
4517 | BTRFS_UUID_KEY_RECEIVED_SUBVOL, |
4518 | root->root_key.objectid); | |
4519 | if (ret < 0 && ret != -EEXIST) { | |
66642832 | 4520 | btrfs_abort_transaction(trans, ret); |
efd38150 | 4521 | btrfs_end_transaction(trans); |
8ea05e3a | 4522 | goto out; |
dd5f9615 SB |
4523 | } |
4524 | } | |
3a45bb20 | 4525 | ret = btrfs_commit_transaction(trans); |
abccd00f | 4526 | out: |
0b246afa | 4527 | up_write(&fs_info->subvol_sem); |
abccd00f HM |
4528 | mnt_drop_write_file(file); |
4529 | return ret; | |
4530 | } | |
4531 | ||
4532 | #ifdef CONFIG_64BIT | |
4533 | static long btrfs_ioctl_set_received_subvol_32(struct file *file, | |
4534 | void __user *arg) | |
4535 | { | |
4536 | struct btrfs_ioctl_received_subvol_args_32 *args32 = NULL; | |
4537 | struct btrfs_ioctl_received_subvol_args *args64 = NULL; | |
4538 | int ret = 0; | |
4539 | ||
4540 | args32 = memdup_user(arg, sizeof(*args32)); | |
7b9ea627 SV |
4541 | if (IS_ERR(args32)) |
4542 | return PTR_ERR(args32); | |
abccd00f | 4543 | |
8d2db785 | 4544 | args64 = kmalloc(sizeof(*args64), GFP_KERNEL); |
84dbeb87 DC |
4545 | if (!args64) { |
4546 | ret = -ENOMEM; | |
abccd00f HM |
4547 | goto out; |
4548 | } | |
4549 | ||
4550 | memcpy(args64->uuid, args32->uuid, BTRFS_UUID_SIZE); | |
4551 | args64->stransid = args32->stransid; | |
4552 | args64->rtransid = args32->rtransid; | |
4553 | args64->stime.sec = args32->stime.sec; | |
4554 | args64->stime.nsec = args32->stime.nsec; | |
4555 | args64->rtime.sec = args32->rtime.sec; | |
4556 | args64->rtime.nsec = args32->rtime.nsec; | |
4557 | args64->flags = args32->flags; | |
4558 | ||
4559 | ret = _btrfs_ioctl_set_received_subvol(file, args64); | |
4560 | if (ret) | |
4561 | goto out; | |
4562 | ||
4563 | memcpy(args32->uuid, args64->uuid, BTRFS_UUID_SIZE); | |
4564 | args32->stransid = args64->stransid; | |
4565 | args32->rtransid = args64->rtransid; | |
4566 | args32->stime.sec = args64->stime.sec; | |
4567 | args32->stime.nsec = args64->stime.nsec; | |
4568 | args32->rtime.sec = args64->rtime.sec; | |
4569 | args32->rtime.nsec = args64->rtime.nsec; | |
4570 | args32->flags = args64->flags; | |
4571 | ||
4572 | ret = copy_to_user(arg, args32, sizeof(*args32)); | |
4573 | if (ret) | |
4574 | ret = -EFAULT; | |
4575 | ||
4576 | out: | |
4577 | kfree(args32); | |
4578 | kfree(args64); | |
4579 | return ret; | |
4580 | } | |
4581 | #endif | |
4582 | ||
4583 | static long btrfs_ioctl_set_received_subvol(struct file *file, | |
4584 | void __user *arg) | |
4585 | { | |
4586 | struct btrfs_ioctl_received_subvol_args *sa = NULL; | |
4587 | int ret = 0; | |
4588 | ||
4589 | sa = memdup_user(arg, sizeof(*sa)); | |
7b9ea627 SV |
4590 | if (IS_ERR(sa)) |
4591 | return PTR_ERR(sa); | |
abccd00f HM |
4592 | |
4593 | ret = _btrfs_ioctl_set_received_subvol(file, sa); | |
4594 | ||
4595 | if (ret) | |
4596 | goto out; | |
4597 | ||
8ea05e3a AB |
4598 | ret = copy_to_user(arg, sa, sizeof(*sa)); |
4599 | if (ret) | |
4600 | ret = -EFAULT; | |
4601 | ||
4602 | out: | |
4603 | kfree(sa); | |
8ea05e3a AB |
4604 | return ret; |
4605 | } | |
4606 | ||
b929c1d8 MPS |
4607 | static int btrfs_ioctl_get_fslabel(struct btrfs_fs_info *fs_info, |
4608 | void __user *arg) | |
867ab667 | 4609 | { |
a1b83ac5 | 4610 | size_t len; |
867ab667 | 4611 | int ret; |
a1b83ac5 AJ |
4612 | char label[BTRFS_LABEL_SIZE]; |
4613 | ||
0b246afa JM |
4614 | spin_lock(&fs_info->super_lock); |
4615 | memcpy(label, fs_info->super_copy->label, BTRFS_LABEL_SIZE); | |
4616 | spin_unlock(&fs_info->super_lock); | |
a1b83ac5 AJ |
4617 | |
4618 | len = strnlen(label, BTRFS_LABEL_SIZE); | |
867ab667 | 4619 | |
4620 | if (len == BTRFS_LABEL_SIZE) { | |
0b246afa JM |
4621 | btrfs_warn(fs_info, |
4622 | "label is too long, return the first %zu bytes", | |
4623 | --len); | |
867ab667 | 4624 | } |
4625 | ||
867ab667 | 4626 | ret = copy_to_user(arg, label, len); |
867ab667 | 4627 | |
4628 | return ret ? -EFAULT : 0; | |
4629 | } | |
4630 | ||
a8bfd4ab | 4631 | static int btrfs_ioctl_set_fslabel(struct file *file, void __user *arg) |
4632 | { | |
0b246afa JM |
4633 | struct inode *inode = file_inode(file); |
4634 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4635 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4636 | struct btrfs_super_block *super_block = fs_info->super_copy; | |
a8bfd4ab | 4637 | struct btrfs_trans_handle *trans; |
4638 | char label[BTRFS_LABEL_SIZE]; | |
4639 | int ret; | |
4640 | ||
4641 | if (!capable(CAP_SYS_ADMIN)) | |
4642 | return -EPERM; | |
4643 | ||
4644 | if (copy_from_user(label, arg, sizeof(label))) | |
4645 | return -EFAULT; | |
4646 | ||
4647 | if (strnlen(label, BTRFS_LABEL_SIZE) == BTRFS_LABEL_SIZE) { | |
0b246afa | 4648 | btrfs_err(fs_info, |
5d163e0e JM |
4649 | "unable to set label with more than %d bytes", |
4650 | BTRFS_LABEL_SIZE - 1); | |
a8bfd4ab | 4651 | return -EINVAL; |
4652 | } | |
4653 | ||
4654 | ret = mnt_want_write_file(file); | |
4655 | if (ret) | |
4656 | return ret; | |
4657 | ||
a8bfd4ab | 4658 | trans = btrfs_start_transaction(root, 0); |
4659 | if (IS_ERR(trans)) { | |
4660 | ret = PTR_ERR(trans); | |
4661 | goto out_unlock; | |
4662 | } | |
4663 | ||
0b246afa | 4664 | spin_lock(&fs_info->super_lock); |
a8bfd4ab | 4665 | strcpy(super_block->label, label); |
0b246afa | 4666 | spin_unlock(&fs_info->super_lock); |
3a45bb20 | 4667 | ret = btrfs_commit_transaction(trans); |
a8bfd4ab | 4668 | |
4669 | out_unlock: | |
a8bfd4ab | 4670 | mnt_drop_write_file(file); |
4671 | return ret; | |
4672 | } | |
4673 | ||
2eaa055f JM |
4674 | #define INIT_FEATURE_FLAGS(suffix) \ |
4675 | { .compat_flags = BTRFS_FEATURE_COMPAT_##suffix, \ | |
4676 | .compat_ro_flags = BTRFS_FEATURE_COMPAT_RO_##suffix, \ | |
4677 | .incompat_flags = BTRFS_FEATURE_INCOMPAT_##suffix } | |
4678 | ||
d5131b65 | 4679 | int btrfs_ioctl_get_supported_features(void __user *arg) |
2eaa055f | 4680 | { |
4d4ab6d6 | 4681 | static const struct btrfs_ioctl_feature_flags features[3] = { |
2eaa055f JM |
4682 | INIT_FEATURE_FLAGS(SUPP), |
4683 | INIT_FEATURE_FLAGS(SAFE_SET), | |
4684 | INIT_FEATURE_FLAGS(SAFE_CLEAR) | |
4685 | }; | |
4686 | ||
4687 | if (copy_to_user(arg, &features, sizeof(features))) | |
4688 | return -EFAULT; | |
4689 | ||
4690 | return 0; | |
4691 | } | |
4692 | ||
b929c1d8 MPS |
4693 | static int btrfs_ioctl_get_features(struct btrfs_fs_info *fs_info, |
4694 | void __user *arg) | |
2eaa055f | 4695 | { |
0b246afa | 4696 | struct btrfs_super_block *super_block = fs_info->super_copy; |
2eaa055f JM |
4697 | struct btrfs_ioctl_feature_flags features; |
4698 | ||
4699 | features.compat_flags = btrfs_super_compat_flags(super_block); | |
4700 | features.compat_ro_flags = btrfs_super_compat_ro_flags(super_block); | |
4701 | features.incompat_flags = btrfs_super_incompat_flags(super_block); | |
4702 | ||
4703 | if (copy_to_user(arg, &features, sizeof(features))) | |
4704 | return -EFAULT; | |
4705 | ||
4706 | return 0; | |
4707 | } | |
4708 | ||
2ff7e61e | 4709 | static int check_feature_bits(struct btrfs_fs_info *fs_info, |
3b02a68a | 4710 | enum btrfs_feature_set set, |
2eaa055f JM |
4711 | u64 change_mask, u64 flags, u64 supported_flags, |
4712 | u64 safe_set, u64 safe_clear) | |
4713 | { | |
f10152bc | 4714 | const char *type = btrfs_feature_set_name(set); |
3b02a68a | 4715 | char *names; |
2eaa055f JM |
4716 | u64 disallowed, unsupported; |
4717 | u64 set_mask = flags & change_mask; | |
4718 | u64 clear_mask = ~flags & change_mask; | |
4719 | ||
4720 | unsupported = set_mask & ~supported_flags; | |
4721 | if (unsupported) { | |
3b02a68a JM |
4722 | names = btrfs_printable_features(set, unsupported); |
4723 | if (names) { | |
0b246afa JM |
4724 | btrfs_warn(fs_info, |
4725 | "this kernel does not support the %s feature bit%s", | |
4726 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4727 | kfree(names); |
4728 | } else | |
0b246afa JM |
4729 | btrfs_warn(fs_info, |
4730 | "this kernel does not support %s bits 0x%llx", | |
4731 | type, unsupported); | |
2eaa055f JM |
4732 | return -EOPNOTSUPP; |
4733 | } | |
4734 | ||
4735 | disallowed = set_mask & ~safe_set; | |
4736 | if (disallowed) { | |
3b02a68a JM |
4737 | names = btrfs_printable_features(set, disallowed); |
4738 | if (names) { | |
0b246afa JM |
4739 | btrfs_warn(fs_info, |
4740 | "can't set the %s feature bit%s while mounted", | |
4741 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4742 | kfree(names); |
4743 | } else | |
0b246afa JM |
4744 | btrfs_warn(fs_info, |
4745 | "can't set %s bits 0x%llx while mounted", | |
4746 | type, disallowed); | |
2eaa055f JM |
4747 | return -EPERM; |
4748 | } | |
4749 | ||
4750 | disallowed = clear_mask & ~safe_clear; | |
4751 | if (disallowed) { | |
3b02a68a JM |
4752 | names = btrfs_printable_features(set, disallowed); |
4753 | if (names) { | |
0b246afa JM |
4754 | btrfs_warn(fs_info, |
4755 | "can't clear the %s feature bit%s while mounted", | |
4756 | names, strchr(names, ',') ? "s" : ""); | |
3b02a68a JM |
4757 | kfree(names); |
4758 | } else | |
0b246afa JM |
4759 | btrfs_warn(fs_info, |
4760 | "can't clear %s bits 0x%llx while mounted", | |
4761 | type, disallowed); | |
2eaa055f JM |
4762 | return -EPERM; |
4763 | } | |
4764 | ||
4765 | return 0; | |
4766 | } | |
4767 | ||
2ff7e61e JM |
4768 | #define check_feature(fs_info, change_mask, flags, mask_base) \ |
4769 | check_feature_bits(fs_info, FEAT_##mask_base, change_mask, flags, \ | |
2eaa055f JM |
4770 | BTRFS_FEATURE_ ## mask_base ## _SUPP, \ |
4771 | BTRFS_FEATURE_ ## mask_base ## _SAFE_SET, \ | |
4772 | BTRFS_FEATURE_ ## mask_base ## _SAFE_CLEAR) | |
4773 | ||
4774 | static int btrfs_ioctl_set_features(struct file *file, void __user *arg) | |
4775 | { | |
0b246afa JM |
4776 | struct inode *inode = file_inode(file); |
4777 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4778 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4779 | struct btrfs_super_block *super_block = fs_info->super_copy; | |
2eaa055f JM |
4780 | struct btrfs_ioctl_feature_flags flags[2]; |
4781 | struct btrfs_trans_handle *trans; | |
4782 | u64 newflags; | |
4783 | int ret; | |
4784 | ||
4785 | if (!capable(CAP_SYS_ADMIN)) | |
4786 | return -EPERM; | |
4787 | ||
4788 | if (copy_from_user(flags, arg, sizeof(flags))) | |
4789 | return -EFAULT; | |
4790 | ||
4791 | /* Nothing to do */ | |
4792 | if (!flags[0].compat_flags && !flags[0].compat_ro_flags && | |
4793 | !flags[0].incompat_flags) | |
4794 | return 0; | |
4795 | ||
2ff7e61e | 4796 | ret = check_feature(fs_info, flags[0].compat_flags, |
2eaa055f JM |
4797 | flags[1].compat_flags, COMPAT); |
4798 | if (ret) | |
4799 | return ret; | |
4800 | ||
2ff7e61e | 4801 | ret = check_feature(fs_info, flags[0].compat_ro_flags, |
2eaa055f JM |
4802 | flags[1].compat_ro_flags, COMPAT_RO); |
4803 | if (ret) | |
4804 | return ret; | |
4805 | ||
2ff7e61e | 4806 | ret = check_feature(fs_info, flags[0].incompat_flags, |
2eaa055f JM |
4807 | flags[1].incompat_flags, INCOMPAT); |
4808 | if (ret) | |
4809 | return ret; | |
4810 | ||
7ab19625 DS |
4811 | ret = mnt_want_write_file(file); |
4812 | if (ret) | |
4813 | return ret; | |
4814 | ||
8051aa1a | 4815 | trans = btrfs_start_transaction(root, 0); |
7ab19625 DS |
4816 | if (IS_ERR(trans)) { |
4817 | ret = PTR_ERR(trans); | |
4818 | goto out_drop_write; | |
4819 | } | |
2eaa055f | 4820 | |
0b246afa | 4821 | spin_lock(&fs_info->super_lock); |
2eaa055f JM |
4822 | newflags = btrfs_super_compat_flags(super_block); |
4823 | newflags |= flags[0].compat_flags & flags[1].compat_flags; | |
4824 | newflags &= ~(flags[0].compat_flags & ~flags[1].compat_flags); | |
4825 | btrfs_set_super_compat_flags(super_block, newflags); | |
4826 | ||
4827 | newflags = btrfs_super_compat_ro_flags(super_block); | |
4828 | newflags |= flags[0].compat_ro_flags & flags[1].compat_ro_flags; | |
4829 | newflags &= ~(flags[0].compat_ro_flags & ~flags[1].compat_ro_flags); | |
4830 | btrfs_set_super_compat_ro_flags(super_block, newflags); | |
4831 | ||
4832 | newflags = btrfs_super_incompat_flags(super_block); | |
4833 | newflags |= flags[0].incompat_flags & flags[1].incompat_flags; | |
4834 | newflags &= ~(flags[0].incompat_flags & ~flags[1].incompat_flags); | |
4835 | btrfs_set_super_incompat_flags(super_block, newflags); | |
0b246afa | 4836 | spin_unlock(&fs_info->super_lock); |
2eaa055f | 4837 | |
3a45bb20 | 4838 | ret = btrfs_commit_transaction(trans); |
7ab19625 DS |
4839 | out_drop_write: |
4840 | mnt_drop_write_file(file); | |
4841 | ||
4842 | return ret; | |
2eaa055f JM |
4843 | } |
4844 | ||
2351f431 JB |
4845 | static int _btrfs_ioctl_send(struct file *file, void __user *argp, bool compat) |
4846 | { | |
4847 | struct btrfs_ioctl_send_args *arg; | |
4848 | int ret; | |
4849 | ||
4850 | if (compat) { | |
4851 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) | |
4852 | struct btrfs_ioctl_send_args_32 args32; | |
4853 | ||
4854 | ret = copy_from_user(&args32, argp, sizeof(args32)); | |
4855 | if (ret) | |
4856 | return -EFAULT; | |
4857 | arg = kzalloc(sizeof(*arg), GFP_KERNEL); | |
4858 | if (!arg) | |
4859 | return -ENOMEM; | |
4860 | arg->send_fd = args32.send_fd; | |
4861 | arg->clone_sources_count = args32.clone_sources_count; | |
4862 | arg->clone_sources = compat_ptr(args32.clone_sources); | |
4863 | arg->parent_root = args32.parent_root; | |
4864 | arg->flags = args32.flags; | |
4865 | memcpy(arg->reserved, args32.reserved, | |
4866 | sizeof(args32.reserved)); | |
4867 | #else | |
4868 | return -ENOTTY; | |
4869 | #endif | |
4870 | } else { | |
4871 | arg = memdup_user(argp, sizeof(*arg)); | |
4872 | if (IS_ERR(arg)) | |
4873 | return PTR_ERR(arg); | |
4874 | } | |
4875 | ret = btrfs_ioctl_send(file, arg); | |
4876 | kfree(arg); | |
4877 | return ret; | |
4878 | } | |
4879 | ||
f46b5a66 CH |
4880 | long btrfs_ioctl(struct file *file, unsigned int |
4881 | cmd, unsigned long arg) | |
4882 | { | |
0b246afa JM |
4883 | struct inode *inode = file_inode(file); |
4884 | struct btrfs_fs_info *fs_info = btrfs_sb(inode->i_sb); | |
4885 | struct btrfs_root *root = BTRFS_I(inode)->root; | |
4bcabaa3 | 4886 | void __user *argp = (void __user *)arg; |
f46b5a66 CH |
4887 | |
4888 | switch (cmd) { | |
6cbff00f CH |
4889 | case FS_IOC_GETFLAGS: |
4890 | return btrfs_ioctl_getflags(file, argp); | |
4891 | case FS_IOC_SETFLAGS: | |
4892 | return btrfs_ioctl_setflags(file, argp); | |
4893 | case FS_IOC_GETVERSION: | |
4894 | return btrfs_ioctl_getversion(file, argp); | |
40cf931f | 4895 | case FS_IOC_GETFSLABEL: |
b929c1d8 | 4896 | return btrfs_ioctl_get_fslabel(fs_info, argp); |
40cf931f ES |
4897 | case FS_IOC_SETFSLABEL: |
4898 | return btrfs_ioctl_set_fslabel(file, argp); | |
f7039b1d | 4899 | case FITRIM: |
b929c1d8 | 4900 | return btrfs_ioctl_fitrim(fs_info, argp); |
f46b5a66 | 4901 | case BTRFS_IOC_SNAP_CREATE: |
fa0d2b9b | 4902 | return btrfs_ioctl_snap_create(file, argp, 0); |
fdfb1e4f | 4903 | case BTRFS_IOC_SNAP_CREATE_V2: |
fa0d2b9b | 4904 | return btrfs_ioctl_snap_create_v2(file, argp, 0); |
3de4586c | 4905 | case BTRFS_IOC_SUBVOL_CREATE: |
fa0d2b9b | 4906 | return btrfs_ioctl_snap_create(file, argp, 1); |
6f72c7e2 AJ |
4907 | case BTRFS_IOC_SUBVOL_CREATE_V2: |
4908 | return btrfs_ioctl_snap_create_v2(file, argp, 1); | |
76dda93c | 4909 | case BTRFS_IOC_SNAP_DESTROY: |
949964c9 MPS |
4910 | return btrfs_ioctl_snap_destroy(file, argp, false); |
4911 | case BTRFS_IOC_SNAP_DESTROY_V2: | |
4912 | return btrfs_ioctl_snap_destroy(file, argp, true); | |
0caa102d LZ |
4913 | case BTRFS_IOC_SUBVOL_GETFLAGS: |
4914 | return btrfs_ioctl_subvol_getflags(file, argp); | |
4915 | case BTRFS_IOC_SUBVOL_SETFLAGS: | |
4916 | return btrfs_ioctl_subvol_setflags(file, argp); | |
6ef5ed0d JB |
4917 | case BTRFS_IOC_DEFAULT_SUBVOL: |
4918 | return btrfs_ioctl_default_subvol(file, argp); | |
f46b5a66 | 4919 | case BTRFS_IOC_DEFRAG: |
1e701a32 CM |
4920 | return btrfs_ioctl_defrag(file, NULL); |
4921 | case BTRFS_IOC_DEFRAG_RANGE: | |
4922 | return btrfs_ioctl_defrag(file, argp); | |
f46b5a66 | 4923 | case BTRFS_IOC_RESIZE: |
198605a8 | 4924 | return btrfs_ioctl_resize(file, argp); |
f46b5a66 | 4925 | case BTRFS_IOC_ADD_DEV: |
2ff7e61e | 4926 | return btrfs_ioctl_add_dev(fs_info, argp); |
f46b5a66 | 4927 | case BTRFS_IOC_RM_DEV: |
da24927b | 4928 | return btrfs_ioctl_rm_dev(file, argp); |
6b526ed7 AJ |
4929 | case BTRFS_IOC_RM_DEV_V2: |
4930 | return btrfs_ioctl_rm_dev_v2(file, argp); | |
475f6387 | 4931 | case BTRFS_IOC_FS_INFO: |
2ff7e61e | 4932 | return btrfs_ioctl_fs_info(fs_info, argp); |
475f6387 | 4933 | case BTRFS_IOC_DEV_INFO: |
2ff7e61e | 4934 | return btrfs_ioctl_dev_info(fs_info, argp); |
f46b5a66 | 4935 | case BTRFS_IOC_BALANCE: |
9ba1f6e4 | 4936 | return btrfs_ioctl_balance(file, NULL); |
ac8e9819 CM |
4937 | case BTRFS_IOC_TREE_SEARCH: |
4938 | return btrfs_ioctl_tree_search(file, argp); | |
cc68a8a5 GH |
4939 | case BTRFS_IOC_TREE_SEARCH_V2: |
4940 | return btrfs_ioctl_tree_search_v2(file, argp); | |
ac8e9819 CM |
4941 | case BTRFS_IOC_INO_LOOKUP: |
4942 | return btrfs_ioctl_ino_lookup(file, argp); | |
d7728c96 JS |
4943 | case BTRFS_IOC_INO_PATHS: |
4944 | return btrfs_ioctl_ino_to_path(root, argp); | |
4945 | case BTRFS_IOC_LOGICAL_INO: | |
d24a67b2 ZB |
4946 | return btrfs_ioctl_logical_to_ino(fs_info, argp, 1); |
4947 | case BTRFS_IOC_LOGICAL_INO_V2: | |
4948 | return btrfs_ioctl_logical_to_ino(fs_info, argp, 2); | |
1406e432 | 4949 | case BTRFS_IOC_SPACE_INFO: |
2ff7e61e | 4950 | return btrfs_ioctl_space_info(fs_info, argp); |
9b199859 FDBM |
4951 | case BTRFS_IOC_SYNC: { |
4952 | int ret; | |
4953 | ||
3d45f221 | 4954 | ret = btrfs_start_delalloc_roots(fs_info, U64_MAX, false); |
9b199859 FDBM |
4955 | if (ret) |
4956 | return ret; | |
0b246afa | 4957 | ret = btrfs_sync_fs(inode->i_sb, 1); |
2fad4e83 DS |
4958 | /* |
4959 | * The transaction thread may want to do more work, | |
01327610 | 4960 | * namely it pokes the cleaner kthread that will start |
2fad4e83 DS |
4961 | * processing uncleaned subvols. |
4962 | */ | |
0b246afa | 4963 | wake_up_process(fs_info->transaction_kthread); |
9b199859 FDBM |
4964 | return ret; |
4965 | } | |
46204592 | 4966 | case BTRFS_IOC_START_SYNC: |
9a8c28be | 4967 | return btrfs_ioctl_start_sync(root, argp); |
46204592 | 4968 | case BTRFS_IOC_WAIT_SYNC: |
2ff7e61e | 4969 | return btrfs_ioctl_wait_sync(fs_info, argp); |
475f6387 | 4970 | case BTRFS_IOC_SCRUB: |
b8e95489 | 4971 | return btrfs_ioctl_scrub(file, argp); |
475f6387 | 4972 | case BTRFS_IOC_SCRUB_CANCEL: |
2ff7e61e | 4973 | return btrfs_ioctl_scrub_cancel(fs_info); |
475f6387 | 4974 | case BTRFS_IOC_SCRUB_PROGRESS: |
2ff7e61e | 4975 | return btrfs_ioctl_scrub_progress(fs_info, argp); |
c9e9f97b | 4976 | case BTRFS_IOC_BALANCE_V2: |
9ba1f6e4 | 4977 | return btrfs_ioctl_balance(file, argp); |
837d5b6e | 4978 | case BTRFS_IOC_BALANCE_CTL: |
2ff7e61e | 4979 | return btrfs_ioctl_balance_ctl(fs_info, arg); |
19a39dce | 4980 | case BTRFS_IOC_BALANCE_PROGRESS: |
2ff7e61e | 4981 | return btrfs_ioctl_balance_progress(fs_info, argp); |
8ea05e3a AB |
4982 | case BTRFS_IOC_SET_RECEIVED_SUBVOL: |
4983 | return btrfs_ioctl_set_received_subvol(file, argp); | |
abccd00f HM |
4984 | #ifdef CONFIG_64BIT |
4985 | case BTRFS_IOC_SET_RECEIVED_SUBVOL_32: | |
4986 | return btrfs_ioctl_set_received_subvol_32(file, argp); | |
4987 | #endif | |
31db9f7c | 4988 | case BTRFS_IOC_SEND: |
2351f431 JB |
4989 | return _btrfs_ioctl_send(file, argp, false); |
4990 | #if defined(CONFIG_64BIT) && defined(CONFIG_COMPAT) | |
4991 | case BTRFS_IOC_SEND_32: | |
4992 | return _btrfs_ioctl_send(file, argp, true); | |
4993 | #endif | |
c11d2c23 | 4994 | case BTRFS_IOC_GET_DEV_STATS: |
2ff7e61e | 4995 | return btrfs_ioctl_get_dev_stats(fs_info, argp); |
5d13a37b | 4996 | case BTRFS_IOC_QUOTA_CTL: |
905b0dda | 4997 | return btrfs_ioctl_quota_ctl(file, argp); |
5d13a37b | 4998 | case BTRFS_IOC_QGROUP_ASSIGN: |
905b0dda | 4999 | return btrfs_ioctl_qgroup_assign(file, argp); |
5d13a37b | 5000 | case BTRFS_IOC_QGROUP_CREATE: |
905b0dda | 5001 | return btrfs_ioctl_qgroup_create(file, argp); |
5d13a37b | 5002 | case BTRFS_IOC_QGROUP_LIMIT: |
905b0dda | 5003 | return btrfs_ioctl_qgroup_limit(file, argp); |
2f232036 JS |
5004 | case BTRFS_IOC_QUOTA_RESCAN: |
5005 | return btrfs_ioctl_quota_rescan(file, argp); | |
5006 | case BTRFS_IOC_QUOTA_RESCAN_STATUS: | |
b929c1d8 | 5007 | return btrfs_ioctl_quota_rescan_status(fs_info, argp); |
57254b6e | 5008 | case BTRFS_IOC_QUOTA_RESCAN_WAIT: |
b929c1d8 | 5009 | return btrfs_ioctl_quota_rescan_wait(fs_info, argp); |
3f6bcfbd | 5010 | case BTRFS_IOC_DEV_REPLACE: |
2ff7e61e | 5011 | return btrfs_ioctl_dev_replace(fs_info, argp); |
2eaa055f | 5012 | case BTRFS_IOC_GET_SUPPORTED_FEATURES: |
d5131b65 | 5013 | return btrfs_ioctl_get_supported_features(argp); |
2eaa055f | 5014 | case BTRFS_IOC_GET_FEATURES: |
b929c1d8 | 5015 | return btrfs_ioctl_get_features(fs_info, argp); |
2eaa055f JM |
5016 | case BTRFS_IOC_SET_FEATURES: |
5017 | return btrfs_ioctl_set_features(file, argp); | |
e4202ac9 DS |
5018 | case FS_IOC_FSGETXATTR: |
5019 | return btrfs_ioctl_fsgetxattr(file, argp); | |
025f2121 DS |
5020 | case FS_IOC_FSSETXATTR: |
5021 | return btrfs_ioctl_fssetxattr(file, argp); | |
b64ec075 TM |
5022 | case BTRFS_IOC_GET_SUBVOL_INFO: |
5023 | return btrfs_ioctl_get_subvol_info(file, argp); | |
42e4b520 TM |
5024 | case BTRFS_IOC_GET_SUBVOL_ROOTREF: |
5025 | return btrfs_ioctl_get_subvol_rootref(file, argp); | |
23d0b79d TM |
5026 | case BTRFS_IOC_INO_LOOKUP_USER: |
5027 | return btrfs_ioctl_ino_lookup_user(file, argp); | |
f46b5a66 CH |
5028 | } |
5029 | ||
5030 | return -ENOTTY; | |
5031 | } | |
4c63c245 LD |
5032 | |
5033 | #ifdef CONFIG_COMPAT | |
5034 | long btrfs_compat_ioctl(struct file *file, unsigned int cmd, unsigned long arg) | |
5035 | { | |
2a362249 JM |
5036 | /* |
5037 | * These all access 32-bit values anyway so no further | |
5038 | * handling is necessary. | |
5039 | */ | |
4c63c245 LD |
5040 | switch (cmd) { |
5041 | case FS_IOC32_GETFLAGS: | |
5042 | cmd = FS_IOC_GETFLAGS; | |
5043 | break; | |
5044 | case FS_IOC32_SETFLAGS: | |
5045 | cmd = FS_IOC_SETFLAGS; | |
5046 | break; | |
5047 | case FS_IOC32_GETVERSION: | |
5048 | cmd = FS_IOC_GETVERSION; | |
5049 | break; | |
4c63c245 LD |
5050 | } |
5051 | ||
5052 | return btrfs_ioctl(file, cmd, (unsigned long) compat_ptr(arg)); | |
5053 | } | |
5054 | #endif |