module/zfs/zfs_log.c

   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) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
  23  */
  24
  25
  26 #include <sys/types.h>
  27 #include <sys/param.h>
  28 #include <sys/systm.h>
  29 #include <sys/sysmacros.h>
  30 #include <sys/cmn_err.h>
  31 #include <sys/kmem.h>
  32 #include <sys/thread.h>
  33 #include <sys/file.h>
  34 #include <sys/vfs.h>
  35 #include <sys/zfs_znode.h>
  36 #include <sys/zfs_dir.h>
  37 #include <sys/zil.h>
  38 #include <sys/zil_impl.h>
  39 #include <sys/byteorder.h>
  40 #include <sys/policy.h>
  41 #include <sys/stat.h>
  42 #include <sys/mode.h>
  43 #include <sys/acl.h>
  44 #include <sys/dmu.h>
  45 #include <sys/spa.h>
  46 #include <sys/zfs_fuid.h>
  47 #include <sys/ddi.h>
  48 #include <sys/dsl_dataset.h>
  49
  50 /*
  51  * These zfs_log_* functions must be called within a dmu tx, in one
  52  * of 2 contexts depending on zilog->z_replay:
  53  *
  54  * Non replay mode
  55  * ---------------
  56  * We need to record the transaction so that if it is committed to
  57  * the Intent Log then it can be replayed.  An intent log transaction
  58  * structure (itx_t) is allocated and all the information necessary to
  59  * possibly replay the transaction is saved in it. The itx is then assigned
  60  * a sequence number and inserted in the in-memory list anchored in the zilog.
  61  *
  62  * Replay mode
  63  * -----------
  64  * We need to mark the intent log record as replayed in the log header.
  65  * This is done in the same transaction as the replay so that they
  66  * commit atomically.
  67  */
  68
  69 int
  70 zfs_log_create_txtype(zil_create_t type, vsecattr_t *vsecp, vattr_t *vap)
  71 {
  72         int isxvattr = (vap->va_mask & ATTR_XVATTR);
  73         switch (type) {
  74         case Z_FILE:
  75                 if (vsecp == NULL && !isxvattr)
  76                         return (TX_CREATE);
  77                 if (vsecp && isxvattr)
  78                         return (TX_CREATE_ACL_ATTR);
  79                 if (vsecp)
  80                         return (TX_CREATE_ACL);
  81                 else
  82                         return (TX_CREATE_ATTR);
  83                 /*NOTREACHED*/
  84         case Z_DIR:
  85                 if (vsecp == NULL && !isxvattr)
  86                         return (TX_MKDIR);
  87                 if (vsecp && isxvattr)
  88                         return (TX_MKDIR_ACL_ATTR);
  89                 if (vsecp)
  90                         return (TX_MKDIR_ACL);
  91                 else
  92                         return (TX_MKDIR_ATTR);
  93         case Z_XATTRDIR:
  94                 return (TX_MKXATTR);
  95         }
  96         ASSERT(0);
  97         return (TX_MAX_TYPE);
  98 }
  99
 100 /*
 101  * build up the log data necessary for logging xvattr_t
 102  * First lr_attr_t is initialized.  following the lr_attr_t
 103  * is the mapsize and attribute bitmap copied from the xvattr_t.
 104  * Following the bitmap and bitmapsize two 64 bit words are reserved
 105  * for the create time which may be set.  Following the create time
 106  * records a single 64 bit integer which has the bits to set on
 107  * replay for the xvattr.
 108  */
 109 static void
 110 zfs_log_xvattr(lr_attr_t *lrattr, xvattr_t *xvap)
 111 {
 112         uint32_t        *bitmap;
 113         uint64_t        *attrs;
 114         uint64_t        *crtime;
 115         xoptattr_t      *xoap;
 116         void            *scanstamp;
 117         int             i;
 118
 119         xoap = xva_getxoptattr(xvap);
 120         ASSERT(xoap);
 121
 122         lrattr->lr_attr_masksize = xvap->xva_mapsize;
 123         bitmap = &lrattr->lr_attr_bitmap;
 124         for (i = 0; i != xvap->xva_mapsize; i++, bitmap++) {
 125                 *bitmap = xvap->xva_reqattrmap[i];
 126         }
 127
 128         /* Now pack the attributes up in a single uint64_t */
 129         attrs = (uint64_t *)bitmap;
 130         crtime = attrs + 1;
 131         scanstamp = (caddr_t)(crtime + 2);
 132         *attrs = 0;
 133         if (XVA_ISSET_REQ(xvap, XAT_READONLY))
 134                 *attrs |= (xoap->xoa_readonly == 0) ? 0 :
 135                     XAT0_READONLY;
 136         if (XVA_ISSET_REQ(xvap, XAT_HIDDEN))
 137                 *attrs |= (xoap->xoa_hidden == 0) ? 0 :
 138                     XAT0_HIDDEN;
 139         if (XVA_ISSET_REQ(xvap, XAT_SYSTEM))
 140                 *attrs |= (xoap->xoa_system == 0) ? 0 :
 141                     XAT0_SYSTEM;
 142         if (XVA_ISSET_REQ(xvap, XAT_ARCHIVE))
 143                 *attrs |= (xoap->xoa_archive == 0) ? 0 :
 144                     XAT0_ARCHIVE;
 145         if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE))
 146                 *attrs |= (xoap->xoa_immutable == 0) ? 0 :
 147                     XAT0_IMMUTABLE;
 148         if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK))
 149                 *attrs |= (xoap->xoa_nounlink == 0) ? 0 :
 150                     XAT0_NOUNLINK;
 151         if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY))
 152                 *attrs |= (xoap->xoa_appendonly == 0) ? 0 :
 153                     XAT0_APPENDONLY;
 154         if (XVA_ISSET_REQ(xvap, XAT_OPAQUE))
 155                 *attrs |= (xoap->xoa_opaque == 0) ? 0 :
 156                     XAT0_APPENDONLY;
 157         if (XVA_ISSET_REQ(xvap, XAT_NODUMP))
 158                 *attrs |= (xoap->xoa_nodump == 0) ? 0 :
 159                     XAT0_NODUMP;
 160         if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED))
 161                 *attrs |= (xoap->xoa_av_quarantined == 0) ? 0 :
 162                     XAT0_AV_QUARANTINED;
 163         if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED))
 164                 *attrs |= (xoap->xoa_av_modified == 0) ? 0 :
 165                     XAT0_AV_MODIFIED;
 166         if (XVA_ISSET_REQ(xvap, XAT_CREATETIME))
 167                 ZFS_TIME_ENCODE(&xoap->xoa_createtime, crtime);
 168         if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP))
 169                 bcopy(xoap->xoa_av_scanstamp, scanstamp, AV_SCANSTAMP_SZ);
 170         if (XVA_ISSET_REQ(xvap, XAT_REPARSE))
 171                 *attrs |= (xoap->xoa_reparse == 0) ? 0 :
 172                     XAT0_REPARSE;
 173         if (XVA_ISSET_REQ(xvap, XAT_OFFLINE))
 174                 *attrs |= (xoap->xoa_offline == 0) ? 0 :
 175                     XAT0_OFFLINE;
 176         if (XVA_ISSET_REQ(xvap, XAT_SPARSE))
 177                 *attrs |= (xoap->xoa_sparse == 0) ? 0 :
 178                     XAT0_SPARSE;
 179 }
 180
 181 static void *
 182 zfs_log_fuid_ids(zfs_fuid_info_t *fuidp, void *start)
 183 {
 184         zfs_fuid_t *zfuid;
 185         uint64_t *fuidloc = start;
 186
 187         /* First copy in the ACE FUIDs */
 188         for (zfuid = list_head(&fuidp->z_fuids); zfuid;
 189             zfuid = list_next(&fuidp->z_fuids, zfuid)) {
 190                 *fuidloc++ = zfuid->z_logfuid;
 191         }
 192         return (fuidloc);
 193 }
 194
 195
 196 static void *
 197 zfs_log_fuid_domains(zfs_fuid_info_t *fuidp, void *start)
 198 {
 199         zfs_fuid_domain_t *zdomain;
 200
 201         /* now copy in the domain info, if any */
 202         if (fuidp->z_domain_str_sz != 0) {
 203                 for (zdomain = list_head(&fuidp->z_domains); zdomain;
 204                     zdomain = list_next(&fuidp->z_domains, zdomain)) {
 205                         bcopy((void *)zdomain->z_domain, start,
 206                             strlen(zdomain->z_domain) + 1);
 207                         start = (caddr_t)start +
 208                             strlen(zdomain->z_domain) + 1;
 209                 }
 210         }
 211         return (start);
 212 }
 213
 214 /*
 215  * Handles TX_CREATE, TX_CREATE_ATTR, TX_MKDIR, TX_MKDIR_ATTR and
 216  * TK_MKXATTR transactions.
 217  *
 218  * TX_CREATE and TX_MKDIR are standard creates, but they may have FUID
 219  * domain information appended prior to the name.  In this case the
 220  * uid/gid in the log record will be a log centric FUID.
 221  *
 222  * TX_CREATE_ACL_ATTR and TX_MKDIR_ACL_ATTR handle special creates that
 223  * may contain attributes, ACL and optional fuid information.
 224  *
 225  * TX_CREATE_ACL and TX_MKDIR_ACL handle special creates that specify
 226  * and ACL and normal users/groups in the ACEs.
 227  *
 228  * There may be an optional xvattr attribute information similar
 229  * to zfs_log_setattr.
 230  *
 231  * Also, after the file name "domain" strings may be appended.
 232  */
 233 void
 234 zfs_log_create(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
 235     znode_t *dzp, znode_t *zp, char *name, vsecattr_t *vsecp,
 236     zfs_fuid_info_t *fuidp, vattr_t *vap)
 237 {
 238         itx_t *itx;
 239         lr_create_t *lr;
 240         lr_acl_create_t *lracl;
 241         size_t aclsize = 0;
 242         size_t xvatsize = 0;
 243         size_t txsize;
 244         xvattr_t *xvap = (xvattr_t *)vap;
 245         void *end;
 246         size_t lrsize;
 247         size_t namesize = strlen(name) + 1;
 248         size_t fuidsz = 0;
 249
 250         if (zil_replaying(zilog, tx))
 251                 return;
 252
 253         /*
 254          * If we have FUIDs present then add in space for
 255          * domains and ACE fuid's if any.
 256          */
 257         if (fuidp) {
 258                 fuidsz += fuidp->z_domain_str_sz;
 259                 fuidsz += fuidp->z_fuid_cnt * sizeof (uint64_t);
 260         }
 261
 262         if (vap->va_mask & ATTR_XVATTR)
 263                 xvatsize = ZIL_XVAT_SIZE(xvap->xva_mapsize);
 264
 265         if ((int)txtype == TX_CREATE_ATTR || (int)txtype == TX_MKDIR_ATTR ||
 266             (int)txtype == TX_CREATE || (int)txtype == TX_MKDIR ||
 267             (int)txtype == TX_MKXATTR) {
 268                 txsize = sizeof (*lr) + namesize + fuidsz + xvatsize;
 269                 lrsize = sizeof (*lr);
 270         } else {
 271                 txsize =
 272                     sizeof (lr_acl_create_t) + namesize + fuidsz +
 273                     ZIL_ACE_LENGTH(aclsize) + xvatsize;
 274                 lrsize = sizeof (lr_acl_create_t);
 275         }
 276
 277         itx = zil_itx_create(txtype, txsize);
 278
 279         lr = (lr_create_t *)&itx->itx_lr;
 280         lr->lr_doid = dzp->z_id;
 281         lr->lr_foid = zp->z_id;
 282         lr->lr_mode = zp->z_mode;
 283         if (!IS_EPHEMERAL(zp->z_uid)) {
 284                 lr->lr_uid = (uint64_t)zp->z_uid;
 285         } else {
 286                 lr->lr_uid = fuidp->z_fuid_owner;
 287         }
 288         if (!IS_EPHEMERAL(zp->z_gid)) {
 289                 lr->lr_gid = (uint64_t)zp->z_gid;
 290         } else {
 291                 lr->lr_gid = fuidp->z_fuid_group;
 292         }
 293         (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(ZTOZSB(zp)), &lr->lr_gen,
 294             sizeof (uint64_t));
 295         (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_CRTIME(ZTOZSB(zp)),
 296             lr->lr_crtime, sizeof (uint64_t) * 2);
 297
 298         if (sa_lookup(zp->z_sa_hdl, SA_ZPL_RDEV(ZTOZSB(zp)), &lr->lr_rdev,
 299             sizeof (lr->lr_rdev)) != 0)
 300                 lr->lr_rdev = 0;
 301
 302         /*
 303          * Fill in xvattr info if any
 304          */
 305         if (vap->va_mask & ATTR_XVATTR) {
 306                 zfs_log_xvattr((lr_attr_t *)((caddr_t)lr + lrsize), xvap);
 307                 end = (caddr_t)lr + lrsize + xvatsize;
 308         } else {
 309                 end = (caddr_t)lr + lrsize;
 310         }
 311
 312         /* Now fill in any ACL info */
 313
 314         if (vsecp) {
 315                 lracl = (lr_acl_create_t *)&itx->itx_lr;
 316                 lracl->lr_aclcnt = vsecp->vsa_aclcnt;
 317                 lracl->lr_acl_bytes = aclsize;
 318                 lracl->lr_domcnt = fuidp ? fuidp->z_domain_cnt : 0;
 319                 lracl->lr_fuidcnt  = fuidp ? fuidp->z_fuid_cnt : 0;
 320                 if (vsecp->vsa_aclflags & VSA_ACE_ACLFLAGS)
 321                         lracl->lr_acl_flags = (uint64_t)vsecp->vsa_aclflags;
 322                 else
 323                         lracl->lr_acl_flags = 0;
 324
 325                 bcopy(vsecp->vsa_aclentp, end, aclsize);
 326                 end = (caddr_t)end + ZIL_ACE_LENGTH(aclsize);
 327         }
 328
 329         /* drop in FUID info */
 330         if (fuidp) {
 331                 end = zfs_log_fuid_ids(fuidp, end);
 332                 end = zfs_log_fuid_domains(fuidp, end);
 333         }
 334         /*
 335          * Now place file name in log record
 336          */
 337         bcopy(name, end, namesize);
 338
 339         zil_itx_assign(zilog, itx, tx);
 340 }
 341
 342 /*
 343  * Handles both TX_REMOVE and TX_RMDIR transactions.
 344  */
 345 void
 346 zfs_log_remove(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
 347         znode_t *dzp, char *name, uint64_t foid)
 348 {
 349         itx_t *itx;
 350         lr_remove_t *lr;
 351         size_t namesize = strlen(name) + 1;
 352
 353         if (zil_replaying(zilog, tx))
 354                 return;
 355
 356         itx = zil_itx_create(txtype, sizeof (*lr) + namesize);
 357         lr = (lr_remove_t *)&itx->itx_lr;
 358         lr->lr_doid = dzp->z_id;
 359         bcopy(name, (char *)(lr + 1), namesize);
 360
 361         itx->itx_oid = foid;
 362
 363         zil_itx_assign(zilog, itx, tx);
 364 }
 365
 366 /*
 367  * Handles TX_LINK transactions.
 368  */
 369 void
 370 zfs_log_link(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
 371         znode_t *dzp, znode_t *zp, char *name)
 372 {
 373         itx_t *itx;
 374         lr_link_t *lr;
 375         size_t namesize = strlen(name) + 1;
 376
 377         if (zil_replaying(zilog, tx))
 378                 return;
 379
 380         itx = zil_itx_create(txtype, sizeof (*lr) + namesize);
 381         lr = (lr_link_t *)&itx->itx_lr;
 382         lr->lr_doid = dzp->z_id;
 383         lr->lr_link_obj = zp->z_id;
 384         bcopy(name, (char *)(lr + 1), namesize);
 385
 386         zil_itx_assign(zilog, itx, tx);
 387 }
 388
 389 /*
 390  * Handles TX_SYMLINK transactions.
 391  */
 392 void
 393 zfs_log_symlink(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
 394     znode_t *dzp, znode_t *zp, char *name, char *link)
 395 {
 396         itx_t *itx;
 397         lr_create_t *lr;
 398         size_t namesize = strlen(name) + 1;
 399         size_t linksize = strlen(link) + 1;
 400
 401         if (zil_replaying(zilog, tx))
 402                 return;
 403
 404         itx = zil_itx_create(txtype, sizeof (*lr) + namesize + linksize);
 405         lr = (lr_create_t *)&itx->itx_lr;
 406         lr->lr_doid = dzp->z_id;
 407         lr->lr_foid = zp->z_id;
 408         lr->lr_uid = zp->z_uid;
 409         lr->lr_gid = zp->z_gid;
 410         lr->lr_mode = zp->z_mode;
 411         (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(ZTOZSB(zp)), &lr->lr_gen,
 412             sizeof (uint64_t));
 413         (void) sa_lookup(zp->z_sa_hdl, SA_ZPL_CRTIME(ZTOZSB(zp)),
 414             lr->lr_crtime, sizeof (uint64_t) * 2);
 415         bcopy(name, (char *)(lr + 1), namesize);
 416         bcopy(link, (char *)(lr + 1) + namesize, linksize);
 417
 418         zil_itx_assign(zilog, itx, tx);
 419 }
 420
 421 /*
 422  * Handles TX_RENAME transactions.
 423  */
 424 void
 425 zfs_log_rename(zilog_t *zilog, dmu_tx_t *tx, uint64_t txtype,
 426         znode_t *sdzp, char *sname, znode_t *tdzp, char *dname, znode_t *szp)
 427 {
 428         itx_t *itx;
 429         lr_rename_t *lr;
 430         size_t snamesize = strlen(sname) + 1;
 431         size_t dnamesize = strlen(dname) + 1;
 432
 433         if (zil_replaying(zilog, tx))
 434                 return;
 435
 436         itx = zil_itx_create(txtype, sizeof (*lr) + snamesize + dnamesize);
 437         lr = (lr_rename_t *)&itx->itx_lr;
 438         lr->lr_sdoid = sdzp->z_id;
 439         lr->lr_tdoid = tdzp->z_id;
 440         bcopy(sname, (char *)(lr + 1), snamesize);
 441         bcopy(dname, (char *)(lr + 1) + snamesize, dnamesize);
 442         itx->itx_oid = szp->z_id;
 443
 444         zil_itx_assign(zilog, itx, tx);
 445 }
 446
 447 /*
 448  * zfs_log_write() handles TX_WRITE transactions. The specified callback is
 449  * called as soon as the write is on stable storage (be it via a DMU sync or a
 450  * ZIL commit).
 451  */
 452 long zfs_immediate_write_sz = 32768;
 453
 454 void
 455 zfs_log_write(zilog_t *zilog, dmu_tx_t *tx, int txtype,
 456         znode_t *zp, offset_t off, ssize_t resid, int ioflag,
 457         zil_callback_t callback, void *callback_data)
 458 {
 459         itx_wr_state_t write_state;
 460         boolean_t slogging;
 461         uintptr_t fsync_cnt;
 462         ssize_t immediate_write_sz;
 463
 464         if (zil_replaying(zilog, tx) || zp->z_unlinked) {
 465                 if (callback != NULL)
 466                         callback(callback_data);
 467                 return;
 468         }
 469
 470         immediate_write_sz = (zilog->zl_logbias == ZFS_LOGBIAS_THROUGHPUT)
 471             ? 0 : (ssize_t)zfs_immediate_write_sz;
 472
 473         slogging = spa_has_slogs(zilog->zl_spa) &&
 474             (zilog->zl_logbias == ZFS_LOGBIAS_LATENCY);
 475         if (resid > immediate_write_sz && !slogging && resid <= zp->z_blksz)
 476                 write_state = WR_INDIRECT;
 477         else if (ioflag & (FSYNC | FDSYNC))
 478                 write_state = WR_COPIED;
 479         else
 480                 write_state = WR_NEED_COPY;
 481
 482         if ((fsync_cnt = (uintptr_t)tsd_get(zfs_fsyncer_key)) != 0) {
 483                 (void) tsd_set(zfs_fsyncer_key, (void *)(fsync_cnt - 1));
 484         }
 485
 486         while (resid) {
 487                 itx_t *itx;
 488                 lr_write_t *lr;
 489                 ssize_t len;
 490
 491                 /*
 492                  * If the write would overflow the largest block then split it.
 493                  */
 494                 if (write_state != WR_INDIRECT && resid > ZIL_MAX_LOG_DATA)
 495                         len = SPA_OLD_MAXBLOCKSIZE >> 1;
 496                 else
 497                         len = resid;
 498
 499                 itx = zil_itx_create(txtype, sizeof (*lr) +
 500                     (write_state == WR_COPIED ? len : 0));
 501                 lr = (lr_write_t *)&itx->itx_lr;
 502                 if (write_state == WR_COPIED && dmu_read(ZTOZSB(zp)->z_os,
 503                     zp->z_id, off, len, lr + 1, DMU_READ_NO_PREFETCH) != 0) {
 504                         zil_itx_destroy(itx);
 505                         itx = zil_itx_create(txtype, sizeof (*lr));
 506                         lr = (lr_write_t *)&itx->itx_lr;
 507                         write_state = WR_NEED_COPY;
 508                 }
 509
 510                 itx->itx_wr_state = write_state;
 511                 if (write_state == WR_NEED_COPY)
 512                         itx->itx_sod += len;
 513                 lr->lr_foid = zp->z_id;
 514                 lr->lr_offset = off;
 515                 lr->lr_length = len;
 516                 lr->lr_blkoff = 0;
 517                 BP_ZERO(&lr->lr_blkptr);
 518
 519                 itx->itx_private = ZTOZSB(zp);
 520
 521                 if (!(ioflag & (FSYNC | FDSYNC)) && (zp->z_sync_cnt == 0) &&
 522                     (fsync_cnt == 0))
 523                         itx->itx_sync = B_FALSE;
 524
 525                 itx->itx_callback = callback;
 526                 itx->itx_callback_data = callback_data;
 527                 zil_itx_assign(zilog, itx, tx);
 528
 529                 off += len;
 530                 resid -= len;
 531         }
 532 }
 533
 534 /*
 535  * Handles TX_TRUNCATE transactions.
 536  */
 537 void
 538 zfs_log_truncate(zilog_t *zilog, dmu_tx_t *tx, int txtype,
 539         znode_t *zp, uint64_t off, uint64_t len)
 540 {
 541         itx_t *itx;
 542         lr_truncate_t *lr;
 543
 544         if (zil_replaying(zilog, tx) || zp->z_unlinked)
 545                 return;
 546
 547         itx = zil_itx_create(txtype, sizeof (*lr));
 548         lr = (lr_truncate_t *)&itx->itx_lr;
 549         lr->lr_foid = zp->z_id;
 550         lr->lr_offset = off;
 551         lr->lr_length = len;
 552
 553         itx->itx_sync = (zp->z_sync_cnt != 0);
 554         zil_itx_assign(zilog, itx, tx);
 555 }
 556
 557 /*
 558  * Handles TX_SETATTR transactions.
 559  */
 560 void
 561 zfs_log_setattr(zilog_t *zilog, dmu_tx_t *tx, int txtype,
 562     znode_t *zp, vattr_t *vap, uint_t mask_applied, zfs_fuid_info_t *fuidp)
 563 {
 564         itx_t           *itx;
 565         lr_setattr_t    *lr;
 566         xvattr_t        *xvap = (xvattr_t *)vap;
 567         size_t          recsize = sizeof (lr_setattr_t);
 568         void            *start;
 569
 570         if (zil_replaying(zilog, tx) || zp->z_unlinked)
 571                 return;
 572
 573         /*
 574          * If XVATTR set, then log record size needs to allow
 575          * for lr_attr_t + xvattr mask, mapsize and create time
 576          * plus actual attribute values
 577          */
 578         if (vap->va_mask & ATTR_XVATTR)
 579                 recsize = sizeof (*lr) + ZIL_XVAT_SIZE(xvap->xva_mapsize);
 580
 581         if (fuidp)
 582                 recsize += fuidp->z_domain_str_sz;
 583
 584         itx = zil_itx_create(txtype, recsize);
 585         lr = (lr_setattr_t *)&itx->itx_lr;
 586         lr->lr_foid = zp->z_id;
 587         lr->lr_mask = (uint64_t)mask_applied;
 588         lr->lr_mode = (uint64_t)vap->va_mode;
 589         if ((mask_applied & ATTR_UID) && IS_EPHEMERAL(vap->va_uid))
 590                 lr->lr_uid = fuidp->z_fuid_owner;
 591         else
 592                 lr->lr_uid = (uint64_t)vap->va_uid;
 593
 594         if ((mask_applied & ATTR_GID) && IS_EPHEMERAL(vap->va_gid))
 595                 lr->lr_gid = fuidp->z_fuid_group;
 596         else
 597                 lr->lr_gid = (uint64_t)vap->va_gid;
 598
 599         lr->lr_size = (uint64_t)vap->va_size;
 600         ZFS_TIME_ENCODE(&vap->va_atime, lr->lr_atime);
 601         ZFS_TIME_ENCODE(&vap->va_mtime, lr->lr_mtime);
 602         start = (lr_setattr_t *)(lr + 1);
 603         if (vap->va_mask & ATTR_XVATTR) {
 604                 zfs_log_xvattr((lr_attr_t *)start, xvap);
 605                 start = (caddr_t)start + ZIL_XVAT_SIZE(xvap->xva_mapsize);
 606         }
 607
 608         /*
 609          * Now stick on domain information if any on end
 610          */
 611
 612         if (fuidp)
 613                 (void) zfs_log_fuid_domains(fuidp, start);
 614
 615         itx->itx_sync = (zp->z_sync_cnt != 0);
 616         zil_itx_assign(zilog, itx, tx);
 617 }
 618
 619 /*
 620  * Handles TX_ACL transactions.
 621  */
 622 void
 623 zfs_log_acl(zilog_t *zilog, dmu_tx_t *tx, znode_t *zp,
 624     vsecattr_t *vsecp, zfs_fuid_info_t *fuidp)
 625 {
 626         itx_t *itx;
 627         lr_acl_v0_t *lrv0;
 628         lr_acl_t *lr;
 629         int txtype;
 630         int lrsize;
 631         size_t txsize;
 632         size_t aclbytes = vsecp->vsa_aclentsz;
 633
 634         if (zil_replaying(zilog, tx) || zp->z_unlinked)
 635                 return;
 636
 637         txtype = (ZTOZSB(zp)->z_version < ZPL_VERSION_FUID) ?
 638             TX_ACL_V0 : TX_ACL;
 639
 640         if (txtype == TX_ACL)
 641                 lrsize = sizeof (*lr);
 642         else
 643                 lrsize = sizeof (*lrv0);
 644
 645         txsize = lrsize +
 646             ((txtype == TX_ACL) ? ZIL_ACE_LENGTH(aclbytes) : aclbytes) +
 647             (fuidp ? fuidp->z_domain_str_sz : 0) +
 648             sizeof (uint64_t) * (fuidp ? fuidp->z_fuid_cnt : 0);
 649
 650         itx = zil_itx_create(txtype, txsize);
 651
 652         lr = (lr_acl_t *)&itx->itx_lr;
 653         lr->lr_foid = zp->z_id;
 654         if (txtype == TX_ACL) {
 655                 lr->lr_acl_bytes = aclbytes;
 656                 lr->lr_domcnt = fuidp ? fuidp->z_domain_cnt : 0;
 657                 lr->lr_fuidcnt = fuidp ? fuidp->z_fuid_cnt : 0;
 658                 if (vsecp->vsa_mask & VSA_ACE_ACLFLAGS)
 659                         lr->lr_acl_flags = (uint64_t)vsecp->vsa_aclflags;
 660                 else
 661                         lr->lr_acl_flags = 0;
 662         }
 663         lr->lr_aclcnt = (uint64_t)vsecp->vsa_aclcnt;
 664
 665         if (txtype == TX_ACL_V0) {
 666                 lrv0 = (lr_acl_v0_t *)lr;
 667                 bcopy(vsecp->vsa_aclentp, (ace_t *)(lrv0 + 1), aclbytes);
 668         } else {
 669                 void *start = (ace_t *)(lr + 1);
 670
 671                 bcopy(vsecp->vsa_aclentp, start, aclbytes);
 672
 673                 start = (caddr_t)start + ZIL_ACE_LENGTH(aclbytes);
 674
 675                 if (fuidp) {
 676                         start = zfs_log_fuid_ids(fuidp, start);
 677                         (void) zfs_log_fuid_domains(fuidp, start);
 678                 }
 679         }
 680
 681         itx->itx_sync = (zp->z_sync_cnt != 0);
 682         zil_itx_assign(zilog, itx, tx);
 683 }
 684
 685 #if defined(_KERNEL) && defined(HAVE_SPL)
 686 module_param(zfs_immediate_write_sz, long, 0644);
 687 MODULE_PARM_DESC(zfs_immediate_write_sz, "Largest data block to write to zil");
 688 #endif