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 2006 Sun Microsystems, Inc. All rights reserved. 23 * Use is subject to license terms. 24 */ 25 26 #pragma ident "%Z%%M% %I% %E% SMI" 27 28 #include <sys/types.h> 29 #include <sys/param.h> 30 #include <sys/systm.h> 31 #include <sys/sysmacros.h> 32 #include <sys/cmn_err.h> 33 #include <sys/kmem.h> 34 #include <sys/thread.h> 35 #include <sys/file.h> 36 #include <sys/fcntl.h> 37 #include <sys/vfs.h> 38 #include <sys/fs/zfs.h> 39 #include <sys/zfs_znode.h> 40 #include <sys/zfs_dir.h> 41 #include <sys/zfs_acl.h> 42 #include <sys/spa.h> 43 #include <sys/zil.h> 44 #include <sys/byteorder.h> 45 #include <sys/stat.h> 46 #include <sys/mode.h> 47 #include <sys/acl.h> 48 #include <sys/atomic.h> 49 #include <sys/cred.h> 50 51 /* 52 * Functions to replay ZFS intent log (ZIL) records 53 * The functions are called through a function vector (zfs_replay_vector) 54 * which is indexed by the transaction type. 55 */ 56 57 static void 58 zfs_init_vattr(vattr_t *vap, uint64_t mask, uint64_t mode, 59 uint64_t uid, uint64_t gid, uint64_t rdev, uint64_t nodeid) 60 { 61 bzero(vap, sizeof (*vap)); 62 vap->va_mask = (uint_t)mask; 63 vap->va_type = IFTOVT(mode); 64 vap->va_mode = mode & MODEMASK; 65 vap->va_uid = (uid_t)uid; 66 vap->va_gid = (gid_t)gid; 67 vap->va_rdev = zfs_cmpldev(rdev); 68 vap->va_nodeid = nodeid; 69 } 70 71 /* ARGSUSED */ 72 static int 73 zfs_replay_error(zfsvfs_t *zfsvfs, lr_t *lr, boolean_t byteswap) 74 { 75 return (ENOTSUP); 76 } 77 78 static int 79 zfs_replay_create(zfsvfs_t *zfsvfs, lr_create_t *lr, boolean_t byteswap) 80 { 81 char *name = (char *)(lr + 1); /* name follows lr_create_t */ 82 char *link; /* symlink content follows name */ 83 znode_t *dzp; 84 vnode_t *vp = NULL; 85 vattr_t va; 86 int error; 87 88 if (byteswap) 89 byteswap_uint64_array(lr, sizeof (*lr)); 90 91 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0) 92 return (error); 93 94 zfs_init_vattr(&va, AT_TYPE | AT_MODE | AT_UID | AT_GID, 95 lr->lr_mode, lr->lr_uid, lr->lr_gid, lr->lr_rdev, lr->lr_foid); 96 97 /* 98 * All forms of zfs create (create, mkdir, mkxattrdir, symlink) 99 * eventually end up in zfs_mknode(), which assigns the object's 100 * creation time and generation number. The generic VOP_CREATE() 101 * doesn't have either concept, so we smuggle the values inside 102 * the vattr's otherwise unused va_ctime and va_nblocks fields. 103 */ 104 ZFS_TIME_DECODE(&va.va_ctime, lr->lr_crtime); 105 va.va_nblocks = lr->lr_gen; 106 107 switch ((int)lr->lr_common.lrc_txtype) { 108 case TX_CREATE: 109 error = VOP_CREATE(ZTOV(dzp), name, &va, 0, 0, &vp, kcred, 0); 110 break; 111 case TX_MKDIR: 112 error = VOP_MKDIR(ZTOV(dzp), name, &va, &vp, kcred); 113 break; 114 case TX_MKXATTR: 115 error = zfs_make_xattrdir(dzp, &va, &vp, kcred); 116 break; 117 case TX_SYMLINK: 118 link = name + strlen(name) + 1; 119 error = VOP_SYMLINK(ZTOV(dzp), name, &va, link, kcred); 120 break; 121 default: 122 error = ENOTSUP; 123 } 124 125 if (error == 0 && vp != NULL) 126 VN_RELE(vp); 127 128 VN_RELE(ZTOV(dzp)); 129 130 return (error); 131 } 132 133 static int 134 zfs_replay_remove(zfsvfs_t *zfsvfs, lr_remove_t *lr, boolean_t byteswap) 135 { 136 char *name = (char *)(lr + 1); /* name follows lr_remove_t */ 137 znode_t *dzp; 138 int error; 139 140 if (byteswap) 141 byteswap_uint64_array(lr, sizeof (*lr)); 142 143 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0) 144 return (error); 145 146 switch ((int)lr->lr_common.lrc_txtype) { 147 case TX_REMOVE: 148 error = VOP_REMOVE(ZTOV(dzp), name, kcred); 149 break; 150 case TX_RMDIR: 151 error = VOP_RMDIR(ZTOV(dzp), name, NULL, kcred); 152 break; 153 default: 154 error = ENOTSUP; 155 } 156 157 VN_RELE(ZTOV(dzp)); 158 159 return (error); 160 } 161 162 static int 163 zfs_replay_link(zfsvfs_t *zfsvfs, lr_link_t *lr, boolean_t byteswap) 164 { 165 char *name = (char *)(lr + 1); /* name follows lr_link_t */ 166 znode_t *dzp, *zp; 167 int error; 168 169 if (byteswap) 170 byteswap_uint64_array(lr, sizeof (*lr)); 171 172 if ((error = zfs_zget(zfsvfs, lr->lr_doid, &dzp)) != 0) 173 return (error); 174 175 if ((error = zfs_zget(zfsvfs, lr->lr_link_obj, &zp)) != 0) { 176 VN_RELE(ZTOV(dzp)); 177 return (error); 178 } 179 180 error = VOP_LINK(ZTOV(dzp), ZTOV(zp), name, kcred); 181 182 VN_RELE(ZTOV(zp)); 183 VN_RELE(ZTOV(dzp)); 184 185 return (error); 186 } 187 188 static int 189 zfs_replay_rename(zfsvfs_t *zfsvfs, lr_rename_t *lr, boolean_t byteswap) 190 { 191 char *sname = (char *)(lr + 1); /* sname and tname follow lr_rename_t */ 192 char *tname = sname + strlen(sname) + 1; 193 znode_t *sdzp, *tdzp; 194 int error; 195 196 if (byteswap) 197 byteswap_uint64_array(lr, sizeof (*lr)); 198 199 if ((error = zfs_zget(zfsvfs, lr->lr_sdoid, &sdzp)) != 0) 200 return (error); 201 202 if ((error = zfs_zget(zfsvfs, lr->lr_tdoid, &tdzp)) != 0) { 203 VN_RELE(ZTOV(sdzp)); 204 return (error); 205 } 206 207 error = VOP_RENAME(ZTOV(sdzp), sname, ZTOV(tdzp), tname, kcred); 208 209 VN_RELE(ZTOV(tdzp)); 210 VN_RELE(ZTOV(sdzp)); 211 212 return (error); 213 } 214 215 static int 216 zfs_replay_write(zfsvfs_t *zfsvfs, lr_write_t *lr, boolean_t byteswap) 217 { 218 char *data = (char *)(lr + 1); /* data follows lr_write_t */ 219 znode_t *zp; 220 int error; 221 ssize_t resid; 222 223 if (byteswap) 224 byteswap_uint64_array(lr, sizeof (*lr)); 225 226 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) { 227 /* 228 * As we can log writes out of order, it's possible the 229 * file has been removed. In this case just drop the write 230 * and return success. 231 */ 232 if (error == ENOENT) 233 error = 0; 234 return (error); 235 } 236 237 error = vn_rdwr(UIO_WRITE, ZTOV(zp), data, lr->lr_length, 238 lr->lr_offset, UIO_SYSSPACE, 0, RLIM64_INFINITY, kcred, &resid); 239 240 VN_RELE(ZTOV(zp)); 241 242 return (error); 243 } 244 245 static int 246 zfs_replay_truncate(zfsvfs_t *zfsvfs, lr_truncate_t *lr, boolean_t byteswap) 247 { 248 znode_t *zp; 249 flock64_t fl; 250 int error; 251 252 if (byteswap) 253 byteswap_uint64_array(lr, sizeof (*lr)); 254 255 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) { 256 /* 257 * As we can log truncates out of order, it's possible the 258 * file has been removed. In this case just drop the truncate 259 * and return success. 260 */ 261 if (error == ENOENT) 262 error = 0; 263 return (error); 264 } 265 266 bzero(&fl, sizeof (fl)); 267 fl.l_type = F_WRLCK; 268 fl.l_whence = 0; 269 fl.l_start = lr->lr_offset; 270 fl.l_len = lr->lr_length; 271 272 error = VOP_SPACE(ZTOV(zp), F_FREESP, &fl, FWRITE | FOFFMAX, 273 lr->lr_offset, kcred, NULL); 274 275 VN_RELE(ZTOV(zp)); 276 277 return (error); 278 } 279 280 static int 281 zfs_replay_setattr(zfsvfs_t *zfsvfs, lr_setattr_t *lr, boolean_t byteswap) 282 { 283 znode_t *zp; 284 vattr_t va; 285 int error; 286 287 if (byteswap) 288 byteswap_uint64_array(lr, sizeof (*lr)); 289 290 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) { 291 /* 292 * As we can log setattrs out of order, it's possible the 293 * file has been removed. In this case just drop the setattr 294 * and return success. 295 */ 296 if (error == ENOENT) 297 error = 0; 298 return (error); 299 } 300 301 zfs_init_vattr(&va, lr->lr_mask, lr->lr_mode, 302 lr->lr_uid, lr->lr_gid, 0, lr->lr_foid); 303 304 va.va_size = lr->lr_size; 305 ZFS_TIME_DECODE(&va.va_atime, lr->lr_atime); 306 ZFS_TIME_DECODE(&va.va_mtime, lr->lr_mtime); 307 308 error = VOP_SETATTR(ZTOV(zp), &va, 0, kcred, NULL); 309 310 VN_RELE(ZTOV(zp)); 311 312 return (error); 313 } 314 315 static int 316 zfs_replay_acl(zfsvfs_t *zfsvfs, lr_acl_t *lr, boolean_t byteswap) 317 { 318 ace_t *ace = (ace_t *)(lr + 1); /* ace array follows lr_acl_t */ 319 vsecattr_t vsa; 320 znode_t *zp; 321 int error; 322 323 if (byteswap) { 324 byteswap_uint64_array(lr, sizeof (*lr)); 325 zfs_ace_byteswap(ace, lr->lr_aclcnt); 326 } 327 328 if ((error = zfs_zget(zfsvfs, lr->lr_foid, &zp)) != 0) { 329 /* 330 * As we can log acls out of order, it's possible the 331 * file has been removed. In this case just drop the acl 332 * and return success. 333 */ 334 if (error == ENOENT) 335 error = 0; 336 return (error); 337 } 338 339 bzero(&vsa, sizeof (vsa)); 340 vsa.vsa_mask = VSA_ACE | VSA_ACECNT; 341 vsa.vsa_aclcnt = lr->lr_aclcnt; 342 vsa.vsa_aclentp = ace; 343 344 error = VOP_SETSECATTR(ZTOV(zp), &vsa, 0, kcred); 345 346 VN_RELE(ZTOV(zp)); 347 348 return (error); 349 } 350 351 /* 352 * Callback vectors for replaying records 353 */ 354 zil_replay_func_t *zfs_replay_vector[TX_MAX_TYPE] = { 355 zfs_replay_error, /* 0 no such transaction type */ 356 zfs_replay_create, /* TX_CREATE */ 357 zfs_replay_create, /* TX_MKDIR */ 358 zfs_replay_create, /* TX_MKXATTR */ 359 zfs_replay_create, /* TX_SYMLINK */ 360 zfs_replay_remove, /* TX_REMOVE */ 361 zfs_replay_remove, /* TX_RMDIR */ 362 zfs_replay_link, /* TX_LINK */ 363 zfs_replay_rename, /* TX_RENAME */ 364 zfs_replay_write, /* TX_WRITE */ 365 zfs_replay_truncate, /* TX_TRUNCATE */ 366 zfs_replay_setattr, /* TX_SETATTR */ 367 zfs_replay_acl, /* TX_ACL */ 368 }; 369