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 https://opensource.org/licenses/CDDL-1.0. 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_znode.h> 27 #include <sys/zfs_vfsops.h> 28 #include <sys/zfs_vnops.h> 29 #include <sys/zfs_ctldir.h> 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); 39 inode_set_iversion(ip, 1); 40 41 return (ip); 42 } 43 44 static void 45 zpl_inode_destroy(struct inode *ip) 46 { 47 ASSERT(atomic_read(&ip->i_count) == 0); 48 zfs_inode_destroy(ip); 49 } 50 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 { 60 fstrans_cookie_t cookie; 61 62 cookie = spl_fstrans_mark(); 63 zfs_dirty_inode(ip, flags); 64 spl_fstrans_unmark(cookie); 65 } 66 #else 67 static void 68 zpl_dirty_inode(struct inode *ip) 69 { 70 fstrans_cookie_t cookie; 71 72 cookie = spl_fstrans_mark(); 73 zfs_dirty_inode(ip, 0); 74 spl_fstrans_unmark(cookie); 75 } 76 #endif /* HAVE_DIRTY_INODE_WITH_FLAGS */ 77 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 * The ->evict_inode() callback must minimally truncate the inode pages, 85 * and call clear_inode(). For 2.6.35 and later kernels this will 86 * simply update the inode state, with the sync occurring before the 87 * truncate in evict(). For earlier kernels clear_inode() maps to 88 * end_writeback() which is responsible for completing all outstanding 89 * write back. In either case, once this is done it is safe to cleanup 90 * any remaining inode specific data via zfs_inactive(). 91 * remaining filesystem specific data. 92 */ 93 static void 94 zpl_evict_inode(struct inode *ip) 95 { 96 fstrans_cookie_t cookie; 97 98 cookie = spl_fstrans_mark(); 99 truncate_setsize(ip, 0); 100 clear_inode(ip); 101 zfs_inactive(ip); 102 spl_fstrans_unmark(cookie); 103 } 104 105 static void 106 zpl_put_super(struct super_block *sb) 107 { 108 fstrans_cookie_t cookie; 109 int error; 110 111 cookie = spl_fstrans_mark(); 112 error = -zfs_umount(sb); 113 spl_fstrans_unmark(cookie); 114 ASSERT3S(error, <=, 0); 115 } 116 117 static int 118 zpl_sync_fs(struct super_block *sb, int wait) 119 { 120 fstrans_cookie_t cookie; 121 cred_t *cr = CRED(); 122 int error; 123 124 crhold(cr); 125 cookie = spl_fstrans_mark(); 126 error = -zfs_sync(sb, wait, cr); 127 spl_fstrans_unmark(cookie); 128 crfree(cr); 129 ASSERT3S(error, <=, 0); 130 131 return (error); 132 } 133 134 static int 135 zpl_statfs(struct dentry *dentry, struct kstatfs *statp) 136 { 137 fstrans_cookie_t cookie; 138 int error; 139 140 cookie = spl_fstrans_mark(); 141 error = -zfs_statvfs(dentry->d_inode, statp); 142 spl_fstrans_unmark(cookie); 143 ASSERT3S(error, <=, 0); 144 145 /* 146 * If required by a 32-bit system call, dynamically scale the 147 * block size up to 16MiB and decrease the block counts. This 148 * allows for a maximum size of 64EiB to be reported. The file 149 * counts must be artificially capped at 2^32-1. 150 */ 151 if (unlikely(zpl_is_32bit_api())) { 152 while (statp->f_blocks > UINT32_MAX && 153 statp->f_bsize < SPA_MAXBLOCKSIZE) { 154 statp->f_frsize <<= 1; 155 statp->f_bsize <<= 1; 156 157 statp->f_blocks >>= 1; 158 statp->f_bfree >>= 1; 159 statp->f_bavail >>= 1; 160 } 161 162 uint64_t usedobjs = statp->f_files - statp->f_ffree; 163 statp->f_ffree = MIN(statp->f_ffree, UINT32_MAX - usedobjs); 164 statp->f_files = statp->f_ffree + usedobjs; 165 } 166 167 return (error); 168 } 169 170 static int 171 zpl_remount_fs(struct super_block *sb, int *flags, char *data) 172 { 173 zfs_mnt_t zm = { .mnt_osname = NULL, .mnt_data = data }; 174 fstrans_cookie_t cookie; 175 int error; 176 177 cookie = spl_fstrans_mark(); 178 error = -zfs_remount(sb, flags, &zm); 179 spl_fstrans_unmark(cookie); 180 ASSERT3S(error, <=, 0); 181 182 return (error); 183 } 184 185 static int 186 __zpl_show_devname(struct seq_file *seq, zfsvfs_t *zfsvfs) 187 { 188 int error; 189 if ((error = zpl_enter(zfsvfs, FTAG)) != 0) 190 return (error); 191 192 char *fsname = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP); 193 dmu_objset_name(zfsvfs->z_os, fsname); 194 195 for (int i = 0; fsname[i] != 0; i++) { 196 /* 197 * Spaces in the dataset name must be converted to their 198 * octal escape sequence for getmntent(3) to correctly 199 * parse then fsname portion of /proc/self/mounts. 200 */ 201 if (fsname[i] == ' ') { 202 seq_puts(seq, "\\040"); 203 } else { 204 seq_putc(seq, fsname[i]); 205 } 206 } 207 208 kmem_free(fsname, ZFS_MAX_DATASET_NAME_LEN); 209 210 zpl_exit(zfsvfs, FTAG); 211 212 return (0); 213 } 214 215 static int 216 zpl_show_devname(struct seq_file *seq, struct dentry *root) 217 { 218 return (__zpl_show_devname(seq, root->d_sb->s_fs_info)); 219 } 220 221 static int 222 __zpl_show_options(struct seq_file *seq, zfsvfs_t *zfsvfs) 223 { 224 seq_printf(seq, ",%s", 225 zfsvfs->z_flags & ZSB_XATTR ? "xattr" : "noxattr"); 226 227 #ifdef CONFIG_FS_POSIX_ACL 228 switch (zfsvfs->z_acl_type) { 229 case ZFS_ACLTYPE_POSIX: 230 seq_puts(seq, ",posixacl"); 231 break; 232 default: 233 seq_puts(seq, ",noacl"); 234 break; 235 } 236 #endif /* CONFIG_FS_POSIX_ACL */ 237 238 switch (zfsvfs->z_case) { 239 case ZFS_CASE_SENSITIVE: 240 seq_puts(seq, ",casesensitive"); 241 break; 242 case ZFS_CASE_INSENSITIVE: 243 seq_puts(seq, ",caseinsensitive"); 244 break; 245 default: 246 seq_puts(seq, ",casemixed"); 247 break; 248 } 249 250 return (0); 251 } 252 253 static int 254 zpl_show_options(struct seq_file *seq, struct dentry *root) 255 { 256 return (__zpl_show_options(seq, root->d_sb->s_fs_info)); 257 } 258 259 static int 260 zpl_fill_super(struct super_block *sb, void *data, int silent) 261 { 262 zfs_mnt_t *zm = (zfs_mnt_t *)data; 263 fstrans_cookie_t cookie; 264 int error; 265 266 cookie = spl_fstrans_mark(); 267 error = -zfs_domount(sb, zm, silent); 268 spl_fstrans_unmark(cookie); 269 ASSERT3S(error, <=, 0); 270 271 return (error); 272 } 273 274 static int 275 zpl_test_super(struct super_block *s, void *data) 276 { 277 zfsvfs_t *zfsvfs = s->s_fs_info; 278 objset_t *os = data; 279 280 if (zfsvfs == NULL) 281 return (0); 282 283 return (os == zfsvfs->z_os); 284 } 285 286 static struct super_block * 287 zpl_mount_impl(struct file_system_type *fs_type, int flags, zfs_mnt_t *zm) 288 { 289 struct super_block *s; 290 objset_t *os; 291 int err; 292 293 err = dmu_objset_hold(zm->mnt_osname, FTAG, &os); 294 if (err) 295 return (ERR_PTR(-err)); 296 297 /* 298 * The dsl pool lock must be released prior to calling sget(). 299 * It is possible sget() may block on the lock in grab_super() 300 * while deactivate_super() holds that same lock and waits for 301 * a txg sync. If the dsl_pool lock is held over sget() 302 * this can prevent the pool sync and cause a deadlock. 303 */ 304 dsl_dataset_long_hold(dmu_objset_ds(os), FTAG); 305 dsl_pool_rele(dmu_objset_pool(os), FTAG); 306 307 s = sget(fs_type, zpl_test_super, set_anon_super, flags, os); 308 309 dsl_dataset_long_rele(dmu_objset_ds(os), FTAG); 310 dsl_dataset_rele(dmu_objset_ds(os), FTAG); 311 312 if (IS_ERR(s)) 313 return (ERR_CAST(s)); 314 315 if (s->s_root == NULL) { 316 err = zpl_fill_super(s, zm, flags & SB_SILENT ? 1 : 0); 317 if (err) { 318 deactivate_locked_super(s); 319 return (ERR_PTR(err)); 320 } 321 s->s_flags |= SB_ACTIVE; 322 } else if ((flags ^ s->s_flags) & SB_RDONLY) { 323 deactivate_locked_super(s); 324 return (ERR_PTR(-EBUSY)); 325 } 326 327 return (s); 328 } 329 330 static struct dentry * 331 zpl_mount(struct file_system_type *fs_type, int flags, 332 const char *osname, void *data) 333 { 334 zfs_mnt_t zm = { .mnt_osname = osname, .mnt_data = data }; 335 336 struct super_block *sb = zpl_mount_impl(fs_type, flags, &zm); 337 if (IS_ERR(sb)) 338 return (ERR_CAST(sb)); 339 340 return (dget(sb->s_root)); 341 } 342 343 static void 344 zpl_kill_sb(struct super_block *sb) 345 { 346 zfs_preumount(sb); 347 kill_anon_super(sb); 348 } 349 350 void 351 zpl_prune_sb(int64_t nr_to_scan, void *arg) 352 { 353 struct super_block *sb = (struct super_block *)arg; 354 int objects = 0; 355 356 (void) -zfs_prune(sb, nr_to_scan, &objects); 357 } 358 359 const struct super_operations zpl_super_operations = { 360 .alloc_inode = zpl_inode_alloc, 361 .destroy_inode = zpl_inode_destroy, 362 .dirty_inode = zpl_dirty_inode, 363 .write_inode = NULL, 364 .evict_inode = zpl_evict_inode, 365 .put_super = zpl_put_super, 366 .sync_fs = zpl_sync_fs, 367 .statfs = zpl_statfs, 368 .remount_fs = zpl_remount_fs, 369 .show_devname = zpl_show_devname, 370 .show_options = zpl_show_options, 371 .show_stats = NULL, 372 }; 373 374 struct file_system_type zpl_fs_type = { 375 .owner = THIS_MODULE, 376 .name = ZFS_DRIVER, 377 #if defined(HAVE_IDMAP_MNT_API) 378 .fs_flags = FS_USERNS_MOUNT | FS_ALLOW_IDMAP, 379 #else 380 .fs_flags = FS_USERNS_MOUNT, 381 #endif 382 .mount = zpl_mount, 383 .kill_sb = zpl_kill_sb, 384 }; 385