]>
Commit | Line | Data |
---|---|---|
51f0bbe4 BB |
1 | /* |
2 | * CDDL HEADER START | |
3 | * | |
4 | * The contents of this file are subject to the terms of the | |
5 | * Common Development and Distribution License (the "License"). | |
6 | * You may not use this file except in compliance with the License. | |
7 | * | |
8 | * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE | |
9 | * or http://www.opensolaris.org/os/licensing. | |
10 | * See the License for the specific language governing permissions | |
11 | * and limitations under the License. | |
12 | * | |
13 | * When distributing Covered Code, include this CDDL HEADER in each | |
14 | * file and include the License file at usr/src/OPENSOLARIS.LICENSE. | |
15 | * If applicable, add the following below this CDDL HEADER, with the | |
16 | * fields enclosed by brackets "[]" replaced with your own identifying | |
17 | * information: Portions Copyright [yyyy] [name of copyright owner] | |
18 | * | |
19 | * CDDL HEADER END | |
20 | */ | |
21 | /* | |
22 | * Copyright (c) 2011, Lawrence Livermore National Security, LLC. | |
23 | */ | |
24 | ||
25 | ||
26 | #include <sys/zfs_vfsops.h> | |
27 | #include <sys/zfs_vnops.h> | |
28 | #include <sys/zfs_znode.h> | |
ebe7e575 | 29 | #include <sys/zfs_ctldir.h> |
51f0bbe4 BB |
30 | #include <sys/zpl.h> |
31 | ||
32 | ||
33 | static struct inode * | |
34 | zpl_inode_alloc(struct super_block *sb) | |
35 | { | |
36 | struct inode *ip; | |
37 | ||
38 | VERIFY3S(zfs_inode_alloc(sb, &ip), ==, 0); | |
18f57327 | 39 | inode_set_iversion(ip, 1); |
51f0bbe4 BB |
40 | |
41 | return (ip); | |
42 | } | |
43 | ||
44 | static void | |
45 | zpl_inode_destroy(struct inode *ip) | |
46 | { | |
d1d7e268 | 47 | ASSERT(atomic_read(&ip->i_count) == 0); |
51f0bbe4 BB |
48 | zfs_inode_destroy(ip); |
49 | } | |
50 | ||
8780c539 BB |
51 | /* |
52 | * Called from __mark_inode_dirty() to reflect that something in the | |
53 | * inode has changed. We use it to ensure the znode system attributes | |
54 | * are always strictly update to date with respect to the inode. | |
55 | */ | |
56 | #ifdef HAVE_DIRTY_INODE_WITH_FLAGS | |
57 | static void | |
58 | zpl_dirty_inode(struct inode *ip, int flags) | |
59 | { | |
a438ff0e BB |
60 | fstrans_cookie_t cookie; |
61 | ||
62 | cookie = spl_fstrans_mark(); | |
8780c539 | 63 | zfs_dirty_inode(ip, flags); |
a438ff0e | 64 | spl_fstrans_unmark(cookie); |
8780c539 BB |
65 | } |
66 | #else | |
67 | static void | |
68 | zpl_dirty_inode(struct inode *ip) | |
69 | { | |
a438ff0e BB |
70 | fstrans_cookie_t cookie; |
71 | ||
72 | cookie = spl_fstrans_mark(); | |
8780c539 | 73 | zfs_dirty_inode(ip, 0); |
a438ff0e | 74 | spl_fstrans_unmark(cookie); |
8780c539 BB |
75 | } |
76 | #endif /* HAVE_DIRTY_INODE_WITH_FLAGS */ | |
77 | ||
2c395def BB |
78 | /* |
79 | * When ->drop_inode() is called its return value indicates if the | |
80 | * inode should be evicted from the inode cache. If the inode is | |
81 | * unhashed and has no links the default policy is to evict it | |
82 | * immediately. | |
83 | * | |
84 | * Prior to 2.6.36 this eviction was accomplished by the vfs calling | |
85 | * ->delete_inode(). It was ->delete_inode()'s responsibility to | |
86 | * truncate the inode pages and call clear_inode(). The call to | |
87 | * clear_inode() synchronously invalidates all the buffers and | |
88 | * calls ->clear_inode(). It was ->clear_inode()'s responsibility | |
89 | * to cleanup and filesystem specific data before freeing the inode. | |
90 | * | |
91 | * This elaborate mechanism was replaced by ->evict_inode() which | |
92 | * does the job of both ->delete_inode() and ->clear_inode(). It | |
93 | * will be called exactly once, and when it returns the inode must | |
739a1a82 RY |
94 | * be in a state where it can simply be freed.i |
95 | * | |
96 | * The ->evict_inode() callback must minimally truncate the inode pages, | |
97 | * and call clear_inode(). For 2.6.35 and later kernels this will | |
98 | * simply update the inode state, with the sync occurring before the | |
99 | * truncate in evict(). For earlier kernels clear_inode() maps to | |
100 | * end_writeback() which is responsible for completing all outstanding | |
101 | * write back. In either case, once this is done it is safe to cleanup | |
102 | * any remaining inode specific data via zfs_inactive(). | |
2c395def BB |
103 | * remaining filesystem specific data. |
104 | */ | |
105 | #ifdef HAVE_EVICT_INODE | |
51f0bbe4 | 106 | static void |
2c395def | 107 | zpl_evict_inode(struct inode *ip) |
51f0bbe4 | 108 | { |
7f3e4662 BB |
109 | fstrans_cookie_t cookie; |
110 | ||
111 | cookie = spl_fstrans_mark(); | |
b3129792 | 112 | truncate_setsize(ip, 0); |
739a1a82 | 113 | clear_inode(ip); |
2c395def | 114 | zfs_inactive(ip); |
7f3e4662 | 115 | spl_fstrans_unmark(cookie); |
51f0bbe4 BB |
116 | } |
117 | ||
2c395def BB |
118 | #else |
119 | ||
2cbb06b5 BB |
120 | static void |
121 | zpl_drop_inode(struct inode *ip) | |
122 | { | |
123 | generic_delete_inode(ip); | |
124 | } | |
125 | ||
51f0bbe4 | 126 | static void |
2c395def | 127 | zpl_clear_inode(struct inode *ip) |
51f0bbe4 | 128 | { |
7f3e4662 BB |
129 | fstrans_cookie_t cookie; |
130 | ||
131 | cookie = spl_fstrans_mark(); | |
51f0bbe4 | 132 | zfs_inactive(ip); |
7f3e4662 | 133 | spl_fstrans_unmark(cookie); |
51f0bbe4 BB |
134 | } |
135 | ||
2c395def BB |
136 | static void |
137 | zpl_inode_delete(struct inode *ip) | |
138 | { | |
b3129792 | 139 | truncate_setsize(ip, 0); |
2c395def BB |
140 | clear_inode(ip); |
141 | } | |
2c395def BB |
142 | #endif /* HAVE_EVICT_INODE */ |
143 | ||
51f0bbe4 BB |
144 | static void |
145 | zpl_put_super(struct super_block *sb) | |
146 | { | |
7fad6290 | 147 | fstrans_cookie_t cookie; |
51f0bbe4 BB |
148 | int error; |
149 | ||
7fad6290 | 150 | cookie = spl_fstrans_mark(); |
51f0bbe4 | 151 | error = -zfs_umount(sb); |
7fad6290 | 152 | spl_fstrans_unmark(cookie); |
51f0bbe4 BB |
153 | ASSERT3S(error, <=, 0); |
154 | } | |
155 | ||
03f9ba9d BB |
156 | static int |
157 | zpl_sync_fs(struct super_block *sb, int wait) | |
158 | { | |
7fad6290 | 159 | fstrans_cookie_t cookie; |
0d3ac5e7 | 160 | cred_t *cr = CRED(); |
03f9ba9d BB |
161 | int error; |
162 | ||
0d3ac5e7 | 163 | crhold(cr); |
7fad6290 | 164 | cookie = spl_fstrans_mark(); |
03f9ba9d | 165 | error = -zfs_sync(sb, wait, cr); |
7fad6290 | 166 | spl_fstrans_unmark(cookie); |
0d3ac5e7 | 167 | crfree(cr); |
03f9ba9d BB |
168 | ASSERT3S(error, <=, 0); |
169 | ||
170 | return (error); | |
171 | } | |
172 | ||
51f0bbe4 BB |
173 | static int |
174 | zpl_statfs(struct dentry *dentry, struct kstatfs *statp) | |
175 | { | |
7fad6290 | 176 | fstrans_cookie_t cookie; |
51f0bbe4 BB |
177 | int error; |
178 | ||
7fad6290 | 179 | cookie = spl_fstrans_mark(); |
51f0bbe4 | 180 | error = -zfs_statvfs(dentry, statp); |
7fad6290 | 181 | spl_fstrans_unmark(cookie); |
51f0bbe4 BB |
182 | ASSERT3S(error, <=, 0); |
183 | ||
e897a23e BB |
184 | /* |
185 | * If required by a 32-bit system call, dynamically scale the | |
186 | * block size up to 16MiB and decrease the block counts. This | |
187 | * allows for a maximum size of 64EiB to be reported. The file | |
188 | * counts must be artificially capped at 2^32-1. | |
189 | */ | |
190 | if (unlikely(zpl_is_32bit_api())) { | |
191 | while (statp->f_blocks > UINT32_MAX && | |
192 | statp->f_bsize < SPA_MAXBLOCKSIZE) { | |
193 | statp->f_frsize <<= 1; | |
194 | statp->f_bsize <<= 1; | |
195 | ||
196 | statp->f_blocks >>= 1; | |
197 | statp->f_bfree >>= 1; | |
198 | statp->f_bavail >>= 1; | |
199 | } | |
200 | ||
201 | uint64_t usedobjs = statp->f_files - statp->f_ffree; | |
202 | statp->f_ffree = MIN(statp->f_ffree, UINT32_MAX - usedobjs); | |
203 | statp->f_files = statp->f_ffree + usedobjs; | |
204 | } | |
205 | ||
51f0bbe4 BB |
206 | return (error); |
207 | } | |
208 | ||
0de19dad BB |
209 | static int |
210 | zpl_remount_fs(struct super_block *sb, int *flags, char *data) | |
211 | { | |
1c2555ef | 212 | zfs_mnt_t zm = { .mnt_osname = NULL, .mnt_data = data }; |
7fad6290 | 213 | fstrans_cookie_t cookie; |
0de19dad | 214 | int error; |
7fad6290 BB |
215 | |
216 | cookie = spl_fstrans_mark(); | |
1c2555ef | 217 | error = -zfs_remount(sb, flags, &zm); |
7fad6290 | 218 | spl_fstrans_unmark(cookie); |
0de19dad BB |
219 | ASSERT3S(error, <=, 0); |
220 | ||
221 | return (error); | |
222 | } | |
223 | ||
51f0bbe4 | 224 | static int |
0037b49e | 225 | __zpl_show_options(struct seq_file *seq, zfsvfs_t *zfsvfs) |
51f0bbe4 | 226 | { |
0037b49e BB |
227 | seq_printf(seq, ",%s", |
228 | zfsvfs->z_flags & ZSB_XATTR ? "xattr" : "noxattr"); | |
47621f3d | 229 | |
b695c34e | 230 | #ifdef CONFIG_FS_POSIX_ACL |
0037b49e | 231 | switch (zfsvfs->z_acl_type) { |
023699cd MM |
232 | case ZFS_ACLTYPE_POSIXACL: |
233 | seq_puts(seq, ",posixacl"); | |
234 | break; | |
235 | default: | |
236 | seq_puts(seq, ",noacl"); | |
237 | break; | |
238 | } | |
b695c34e | 239 | #endif /* CONFIG_FS_POSIX_ACL */ |
023699cd | 240 | |
47621f3d BB |
241 | return (0); |
242 | } | |
023699cd MM |
243 | |
244 | #ifdef HAVE_SHOW_OPTIONS_WITH_DENTRY | |
245 | static int | |
246 | zpl_show_options(struct seq_file *seq, struct dentry *root) | |
247 | { | |
d1d7e268 | 248 | return (__zpl_show_options(seq, root->d_sb->s_fs_info)); |
023699cd | 249 | } |
47621f3d BB |
250 | #else |
251 | static int | |
252 | zpl_show_options(struct seq_file *seq, struct vfsmount *vfsp) | |
253 | { | |
d1d7e268 | 254 | return (__zpl_show_options(seq, vfsp->mnt_sb->s_fs_info)); |
51f0bbe4 | 255 | } |
47621f3d | 256 | #endif /* HAVE_SHOW_OPTIONS_WITH_DENTRY */ |
51f0bbe4 BB |
257 | |
258 | static int | |
259 | zpl_fill_super(struct super_block *sb, void *data, int silent) | |
260 | { | |
1c2555ef | 261 | zfs_mnt_t *zm = (zfs_mnt_t *)data; |
7fad6290 | 262 | fstrans_cookie_t cookie; |
51f0bbe4 BB |
263 | int error; |
264 | ||
7fad6290 | 265 | cookie = spl_fstrans_mark(); |
1c2555ef | 266 | error = -zfs_domount(sb, zm, silent); |
7fad6290 | 267 | spl_fstrans_unmark(cookie); |
51f0bbe4 BB |
268 | ASSERT3S(error, <=, 0); |
269 | ||
270 | return (error); | |
271 | } | |
272 | ||
93b43af1 SF |
273 | static int |
274 | zpl_test_super(struct super_block *s, void *data) | |
275 | { | |
276 | zfsvfs_t *zfsvfs = s->s_fs_info; | |
277 | objset_t *os = data; | |
278 | ||
279 | if (zfsvfs == NULL) | |
280 | return (0); | |
281 | ||
282 | return (os == zfsvfs->z_os); | |
283 | } | |
284 | ||
285 | static struct super_block * | |
286 | zpl_mount_impl(struct file_system_type *fs_type, int flags, zfs_mnt_t *zm) | |
287 | { | |
288 | struct super_block *s; | |
289 | objset_t *os; | |
290 | int err; | |
291 | ||
292 | err = dmu_objset_hold(zm->mnt_osname, FTAG, &os); | |
293 | if (err) | |
294 | return (ERR_PTR(-err)); | |
295 | ||
ac09630d BB |
296 | /* |
297 | * The dsl pool lock must be released prior to calling sget(). | |
298 | * It is possible sget() may block on the lock in grab_super() | |
299 | * while deactivate_super() holds that same lock and waits for | |
e1cfd73f | 300 | * a txg sync. If the dsl_pool lock is held over sget() |
ac09630d BB |
301 | * this can prevent the pool sync and cause a deadlock. |
302 | */ | |
303 | dsl_pool_rele(dmu_objset_pool(os), FTAG); | |
93b43af1 | 304 | s = zpl_sget(fs_type, zpl_test_super, set_anon_super, flags, os); |
ac09630d BB |
305 | dsl_dataset_rele(dmu_objset_ds(os), FTAG); |
306 | ||
93b43af1 SF |
307 | if (IS_ERR(s)) |
308 | return (ERR_CAST(s)); | |
309 | ||
310 | if (s->s_root == NULL) { | |
311 | err = zpl_fill_super(s, zm, flags & SB_SILENT ? 1 : 0); | |
312 | if (err) { | |
313 | deactivate_locked_super(s); | |
314 | return (ERR_PTR(err)); | |
315 | } | |
316 | s->s_flags |= SB_ACTIVE; | |
317 | } else if ((flags ^ s->s_flags) & SB_RDONLY) { | |
318 | deactivate_locked_super(s); | |
319 | return (ERR_PTR(-EBUSY)); | |
320 | } | |
321 | ||
322 | return (s); | |
323 | } | |
324 | ||
325 | #ifdef HAVE_FST_MOUNT | |
2cf7f52b BB |
326 | static struct dentry * |
327 | zpl_mount(struct file_system_type *fs_type, int flags, | |
328 | const char *osname, void *data) | |
329 | { | |
1c2555ef | 330 | zfs_mnt_t zm = { .mnt_osname = osname, .mnt_data = data }; |
2cf7f52b | 331 | |
93b43af1 SF |
332 | struct super_block *sb = zpl_mount_impl(fs_type, flags, &zm); |
333 | if (IS_ERR(sb)) | |
334 | return (ERR_CAST(sb)); | |
335 | ||
336 | return (dget(sb->s_root)); | |
2cf7f52b BB |
337 | } |
338 | #else | |
51f0bbe4 BB |
339 | static int |
340 | zpl_get_sb(struct file_system_type *fs_type, int flags, | |
341 | const char *osname, void *data, struct vfsmount *mnt) | |
342 | { | |
1c2555ef | 343 | zfs_mnt_t zm = { .mnt_osname = osname, .mnt_data = data }; |
51f0bbe4 | 344 | |
93b43af1 SF |
345 | struct super_block *sb = zpl_mount_impl(fs_type, flags, &zm); |
346 | if (IS_ERR(sb)) | |
347 | return (PTR_ERR(sb)); | |
348 | ||
349 | (void) simple_set_mnt(mnt, sb); | |
350 | ||
351 | return (0); | |
51f0bbe4 | 352 | } |
93b43af1 | 353 | #endif /* HAVE_FST_MOUNT */ |
51f0bbe4 BB |
354 | |
355 | static void | |
356 | zpl_kill_sb(struct super_block *sb) | |
357 | { | |
ebe7e575 | 358 | zfs_preumount(sb); |
51f0bbe4 | 359 | kill_anon_super(sb); |
dba1d705 BB |
360 | |
361 | #ifdef HAVE_S_INSTANCES_LIST_HEAD | |
362 | sb->s_instances.next = &(zpl_fs_type.fs_supers); | |
363 | #endif /* HAVE_S_INSTANCES_LIST_HEAD */ | |
51f0bbe4 BB |
364 | } |
365 | ||
ab26409d | 366 | void |
2cbb06b5 | 367 | zpl_prune_sb(int64_t nr_to_scan, void *arg) |
ab26409d | 368 | { |
2cbb06b5 BB |
369 | struct super_block *sb = (struct super_block *)arg; |
370 | int objects = 0; | |
ab26409d | 371 | |
f298b24d | 372 | (void) -zfs_prune(sb, nr_to_scan, &objects); |
050d22b0 | 373 | } |
ab26409d BB |
374 | |
375 | #ifdef HAVE_NR_CACHED_OBJECTS | |
376 | static int | |
377 | zpl_nr_cached_objects(struct super_block *sb) | |
378 | { | |
04870568 | 379 | return (0); |
ab26409d BB |
380 | } |
381 | #endif /* HAVE_NR_CACHED_OBJECTS */ | |
382 | ||
383 | #ifdef HAVE_FREE_CACHED_OBJECTS | |
ab26409d BB |
384 | static void |
385 | zpl_free_cached_objects(struct super_block *sb, int nr_to_scan) | |
386 | { | |
94520ca4 | 387 | /* noop */ |
ab26409d BB |
388 | } |
389 | #endif /* HAVE_FREE_CACHED_OBJECTS */ | |
390 | ||
51f0bbe4 | 391 | const struct super_operations zpl_super_operations = { |
ab26409d BB |
392 | .alloc_inode = zpl_inode_alloc, |
393 | .destroy_inode = zpl_inode_destroy, | |
8780c539 | 394 | .dirty_inode = zpl_dirty_inode, |
ab26409d | 395 | .write_inode = NULL, |
2c395def | 396 | #ifdef HAVE_EVICT_INODE |
ab26409d | 397 | .evict_inode = zpl_evict_inode, |
2c395def | 398 | #else |
2cbb06b5 | 399 | .drop_inode = zpl_drop_inode, |
ab26409d BB |
400 | .clear_inode = zpl_clear_inode, |
401 | .delete_inode = zpl_inode_delete, | |
2c395def | 402 | #endif /* HAVE_EVICT_INODE */ |
ab26409d | 403 | .put_super = zpl_put_super, |
ab26409d BB |
404 | .sync_fs = zpl_sync_fs, |
405 | .statfs = zpl_statfs, | |
406 | .remount_fs = zpl_remount_fs, | |
407 | .show_options = zpl_show_options, | |
408 | .show_stats = NULL, | |
409 | #ifdef HAVE_NR_CACHED_OBJECTS | |
410 | .nr_cached_objects = zpl_nr_cached_objects, | |
411 | #endif /* HAVE_NR_CACHED_OBJECTS */ | |
412 | #ifdef HAVE_FREE_CACHED_OBJECTS | |
413 | .free_cached_objects = zpl_free_cached_objects, | |
414 | #endif /* HAVE_FREE_CACHED_OBJECTS */ | |
51f0bbe4 BB |
415 | }; |
416 | ||
51f0bbe4 | 417 | struct file_system_type zpl_fs_type = { |
ab26409d BB |
418 | .owner = THIS_MODULE, |
419 | .name = ZFS_DRIVER, | |
93b43af1 | 420 | #ifdef HAVE_FST_MOUNT |
ab26409d | 421 | .mount = zpl_mount, |
2cf7f52b | 422 | #else |
ab26409d | 423 | .get_sb = zpl_get_sb, |
93b43af1 | 424 | #endif /* HAVE_FST_MOUNT */ |
ab26409d | 425 | .kill_sb = zpl_kill_sb, |
51f0bbe4 | 426 | }; |