diff options
Diffstat (limited to 'module/zfs/zfs_znode.c')
| -rw-r--r-- | module/zfs/zfs_znode.c | 1136 |
1 files changed, 310 insertions, 826 deletions
diff --git a/module/zfs/zfs_znode.c b/module/zfs/zfs_znode.c index 4f6185fb5..56ac2ab8c 100644 --- a/module/zfs/zfs_znode.c +++ b/module/zfs/zfs_znode.c @@ -51,9 +51,11 @@ #include <sys/zfs_ioctl.h> #include <sys/zfs_rlock.h> #include <sys/zfs_fuid.h> +#include <sys/zfs_vnops.h> #include <sys/dnode.h> #include <sys/fs/zfs.h> #include <sys/kidmap.h> +#include <sys/zpl.h> #endif /* _KERNEL */ #include <sys/dmu.h> @@ -88,40 +90,16 @@ * (such as VFS logic) that will not compile easily in userland. */ #ifdef _KERNEL -#ifdef HAVE_ZPL -/* - * Needed to close a small window in zfs_znode_move() that allows the zfsvfs to - * be freed before it can be safely accessed. - */ -krwlock_t zfsvfs_lock; static kmem_cache_t *znode_cache = NULL; /*ARGSUSED*/ -static void -znode_evict_error(dmu_buf_t *dbuf, void *user_ptr) -{ - /* - * We should never drop all dbuf refs without first clearing - * the eviction callback. - */ - panic("evicting znode %p\n", user_ptr); -} - -/*ARGSUSED*/ static int zfs_znode_cache_constructor(void *buf, void *arg, int kmflags) { znode_t *zp = buf; - ASSERT(!POINTER_IS_VALID(zp->z_zfsvfs)); - - zp->z_vnode = vn_alloc(kmflags); - if (zp->z_vnode == NULL) { - return (-1); - } - ZTOV(zp)->v_data = zp; - + inode_init_once(ZTOI(zp)); list_link_init(&zp->z_link_node); mutex_init(&zp->z_lock, NULL, MUTEX_DEFAULT, NULL); @@ -145,9 +123,6 @@ zfs_znode_cache_destructor(void *buf, void *arg) { znode_t *zp = buf; - ASSERT(!POINTER_IS_VALID(zp->z_zfsvfs)); - ASSERT(ZTOV(zp)->v_data == zp); - vn_free(ZTOV(zp)); ASSERT(!list_link_active(&zp->z_link_node)); mutex_destroy(&zp->z_lock); rw_destroy(&zp->z_parent_lock); @@ -160,315 +135,33 @@ zfs_znode_cache_destructor(void *buf, void *arg) ASSERT(zp->z_acl_cached == NULL); } -#ifdef ZNODE_STATS -static struct { - uint64_t zms_zfsvfs_invalid; - uint64_t zms_zfsvfs_recheck1; - uint64_t zms_zfsvfs_unmounted; - uint64_t zms_zfsvfs_recheck2; - uint64_t zms_obj_held; - uint64_t zms_vnode_locked; - uint64_t zms_not_only_dnlc; -} znode_move_stats; -#endif /* ZNODE_STATS */ - -static void -zfs_znode_move_impl(znode_t *ozp, znode_t *nzp) -{ - vnode_t *vp; - - /* Copy fields. */ - nzp->z_zfsvfs = ozp->z_zfsvfs; - - /* Swap vnodes. */ - vp = nzp->z_vnode; - nzp->z_vnode = ozp->z_vnode; - ozp->z_vnode = vp; /* let destructor free the overwritten vnode */ - ZTOV(ozp)->v_data = ozp; - ZTOV(nzp)->v_data = nzp; - - nzp->z_id = ozp->z_id; - ASSERT(ozp->z_dirlocks == NULL); /* znode not in use */ - ASSERT(avl_numnodes(&ozp->z_range_avl) == 0); - nzp->z_unlinked = ozp->z_unlinked; - nzp->z_atime_dirty = ozp->z_atime_dirty; - nzp->z_zn_prefetch = ozp->z_zn_prefetch; - nzp->z_blksz = ozp->z_blksz; - nzp->z_seq = ozp->z_seq; - nzp->z_mapcnt = ozp->z_mapcnt; - nzp->z_gen = ozp->z_gen; - nzp->z_sync_cnt = ozp->z_sync_cnt; - nzp->z_is_sa = ozp->z_is_sa; - nzp->z_sa_hdl = ozp->z_sa_hdl; - bcopy(ozp->z_atime, nzp->z_atime, sizeof (uint64_t) * 2); - nzp->z_links = ozp->z_links; - nzp->z_size = ozp->z_size; - nzp->z_pflags = ozp->z_pflags; - nzp->z_uid = ozp->z_uid; - nzp->z_gid = ozp->z_gid; - nzp->z_mode = ozp->z_mode; - - /* - * Since this is just an idle znode and kmem is already dealing with - * memory pressure, release any cached ACL. - */ - if (ozp->z_acl_cached) { - zfs_acl_free(ozp->z_acl_cached); - ozp->z_acl_cached = NULL; - } - - sa_set_userp(nzp->z_sa_hdl, nzp); - - /* - * Invalidate the original znode by clearing fields that provide a - * pointer back to the znode. Set the low bit of the vfs pointer to - * ensure that zfs_znode_move() recognizes the znode as invalid in any - * subsequent callback. - */ - ozp->z_sa_hdl = NULL; - POINTER_INVALIDATE(&ozp->z_zfsvfs); - - /* - * Mark the znode. - */ - nzp->z_moved = 1; - ozp->z_moved = (uint8_t)-1; -} - -/*ARGSUSED*/ -static kmem_cbrc_t -zfs_znode_move(void *buf, void *newbuf, size_t size, void *arg) -{ - znode_t *ozp = buf, *nzp = newbuf; - zfsvfs_t *zfsvfs; - vnode_t *vp; - - /* - * The znode is on the file system's list of known znodes if the vfs - * pointer is valid. We set the low bit of the vfs pointer when freeing - * the znode to invalidate it, and the memory patterns written by kmem - * (baddcafe and deadbeef) set at least one of the two low bits. A newly - * created znode sets the vfs pointer last of all to indicate that the - * znode is known and in a valid state to be moved by this function. - */ - zfsvfs = ozp->z_zfsvfs; - if (!POINTER_IS_VALID(zfsvfs)) { - ZNODE_STAT_ADD(znode_move_stats.zms_zfsvfs_invalid); - return (KMEM_CBRC_DONT_KNOW); - } - - /* - * Close a small window in which it's possible that the filesystem could - * be unmounted and freed, and zfsvfs, though valid in the previous - * statement, could point to unrelated memory by the time we try to - * prevent the filesystem from being unmounted. - */ - rw_enter(&zfsvfs_lock, RW_WRITER); - if (zfsvfs != ozp->z_zfsvfs) { - rw_exit(&zfsvfs_lock); - ZNODE_STAT_ADD(znode_move_stats.zms_zfsvfs_recheck1); - return (KMEM_CBRC_DONT_KNOW); - } - - /* - * If the znode is still valid, then so is the file system. We know that - * no valid file system can be freed while we hold zfsvfs_lock, so we - * can safely ensure that the filesystem is not and will not be - * unmounted. The next statement is equivalent to ZFS_ENTER(). - */ - rrw_enter(&zfsvfs->z_teardown_lock, RW_READER, FTAG); - if (zfsvfs->z_unmounted) { - ZFS_EXIT(zfsvfs); - rw_exit(&zfsvfs_lock); - ZNODE_STAT_ADD(znode_move_stats.zms_zfsvfs_unmounted); - return (KMEM_CBRC_DONT_KNOW); - } - rw_exit(&zfsvfs_lock); - - mutex_enter(&zfsvfs->z_znodes_lock); - /* - * Recheck the vfs pointer in case the znode was removed just before - * acquiring the lock. - */ - if (zfsvfs != ozp->z_zfsvfs) { - mutex_exit(&zfsvfs->z_znodes_lock); - ZFS_EXIT(zfsvfs); - ZNODE_STAT_ADD(znode_move_stats.zms_zfsvfs_recheck2); - return (KMEM_CBRC_DONT_KNOW); - } - - /* - * At this point we know that as long as we hold z_znodes_lock, the - * znode cannot be freed and fields within the znode can be safely - * accessed. Now, prevent a race with zfs_zget(). - */ - if (ZFS_OBJ_HOLD_TRYENTER(zfsvfs, ozp->z_id) == 0) { - mutex_exit(&zfsvfs->z_znodes_lock); - ZFS_EXIT(zfsvfs); - ZNODE_STAT_ADD(znode_move_stats.zms_obj_held); - return (KMEM_CBRC_LATER); - } - - vp = ZTOV(ozp); - if (mutex_tryenter(&vp->v_lock) == 0) { - ZFS_OBJ_HOLD_EXIT(zfsvfs, ozp->z_id); - mutex_exit(&zfsvfs->z_znodes_lock); - ZFS_EXIT(zfsvfs); - ZNODE_STAT_ADD(znode_move_stats.zms_vnode_locked); - return (KMEM_CBRC_LATER); - } - - /* Only move znodes that are referenced _only_ by the DNLC. */ - if (vp->v_count != 1 || !vn_in_dnlc(vp)) { - mutex_exit(&vp->v_lock); - ZFS_OBJ_HOLD_EXIT(zfsvfs, ozp->z_id); - mutex_exit(&zfsvfs->z_znodes_lock); - ZFS_EXIT(zfsvfs); - ZNODE_STAT_ADD(znode_move_stats.zms_not_only_dnlc); - return (KMEM_CBRC_LATER); - } - - /* - * The znode is known and in a valid state to move. We're holding the - * locks needed to execute the critical section. - */ - zfs_znode_move_impl(ozp, nzp); - mutex_exit(&vp->v_lock); - ZFS_OBJ_HOLD_EXIT(zfsvfs, ozp->z_id); - - list_link_replace(&ozp->z_link_node, &nzp->z_link_node); - mutex_exit(&zfsvfs->z_znodes_lock); - ZFS_EXIT(zfsvfs); - - return (KMEM_CBRC_YES); -} - void zfs_znode_init(void) { /* * Initialize zcache */ - rw_init(&zfsvfs_lock, NULL, RW_DEFAULT, NULL); ASSERT(znode_cache == NULL); znode_cache = kmem_cache_create("zfs_znode_cache", sizeof (znode_t), 0, zfs_znode_cache_constructor, - zfs_znode_cache_destructor, NULL, NULL, NULL, 0); - kmem_cache_set_move(znode_cache, zfs_znode_move); + zfs_znode_cache_destructor, NULL, NULL, NULL, KMC_KMEM); } void zfs_znode_fini(void) { /* - * Cleanup vfs & vnode ops - */ - zfs_remove_op_tables(); - - /* * Cleanup zcache */ if (znode_cache) kmem_cache_destroy(znode_cache); znode_cache = NULL; - rw_destroy(&zfsvfs_lock); -} - -struct vnodeops *zfs_dvnodeops; -struct vnodeops *zfs_fvnodeops; -struct vnodeops *zfs_symvnodeops; -struct vnodeops *zfs_xdvnodeops; -struct vnodeops *zfs_evnodeops; -struct vnodeops *zfs_sharevnodeops; - -void -zfs_remove_op_tables() -{ - /* - * Remove vfs ops - */ - ASSERT(zfsfstype); - (void) vfs_freevfsops_by_type(zfsfstype); - zfsfstype = 0; - - /* - * Remove vnode ops - */ - if (zfs_dvnodeops) - vn_freevnodeops(zfs_dvnodeops); - if (zfs_fvnodeops) - vn_freevnodeops(zfs_fvnodeops); - if (zfs_symvnodeops) - vn_freevnodeops(zfs_symvnodeops); - if (zfs_xdvnodeops) - vn_freevnodeops(zfs_xdvnodeops); - if (zfs_evnodeops) - vn_freevnodeops(zfs_evnodeops); - if (zfs_sharevnodeops) - vn_freevnodeops(zfs_sharevnodeops); - - zfs_dvnodeops = NULL; - zfs_fvnodeops = NULL; - zfs_symvnodeops = NULL; - zfs_xdvnodeops = NULL; - zfs_evnodeops = NULL; - zfs_sharevnodeops = NULL; -} - -extern const fs_operation_def_t zfs_dvnodeops_template[]; -extern const fs_operation_def_t zfs_fvnodeops_template[]; -extern const fs_operation_def_t zfs_xdvnodeops_template[]; -extern const fs_operation_def_t zfs_symvnodeops_template[]; -extern const fs_operation_def_t zfs_evnodeops_template[]; -extern const fs_operation_def_t zfs_sharevnodeops_template[]; - -int -zfs_create_op_tables() -{ - int error; - - /* - * zfs_dvnodeops can be set if mod_remove() calls mod_installfs() - * due to a failure to remove the the 2nd modlinkage (zfs_modldrv). - * In this case we just return as the ops vectors are already set up. - */ - if (zfs_dvnodeops) - return (0); - - error = vn_make_ops(MNTTYPE_ZFS, zfs_dvnodeops_template, - &zfs_dvnodeops); - if (error) - return (error); - - error = vn_make_ops(MNTTYPE_ZFS, zfs_fvnodeops_template, - &zfs_fvnodeops); - if (error) - return (error); - - error = vn_make_ops(MNTTYPE_ZFS, zfs_symvnodeops_template, - &zfs_symvnodeops); - if (error) - return (error); - - error = vn_make_ops(MNTTYPE_ZFS, zfs_xdvnodeops_template, - &zfs_xdvnodeops); - if (error) - return (error); - - error = vn_make_ops(MNTTYPE_ZFS, zfs_evnodeops_template, - &zfs_evnodeops); - if (error) - return (error); - - error = vn_make_ops(MNTTYPE_ZFS, zfs_sharevnodeops_template, - &zfs_sharevnodeops); - - return (error); } int -zfs_create_share_dir(zfsvfs_t *zfsvfs, dmu_tx_t *tx) +zfs_create_share_dir(zfs_sb_t *zsb, dmu_tx_t *tx) { +#ifdef HAVE_SHARE zfs_acl_ids_t acl_ids; vattr_t vattr; znode_t *sharezp; @@ -477,13 +170,11 @@ zfs_create_share_dir(zfsvfs_t *zfsvfs, dmu_tx_t *tx) int error; vattr.va_mask = AT_MODE|AT_UID|AT_GID|AT_TYPE; - vattr.va_type = VDIR; - vattr.va_mode = S_IFDIR|0555; + vattr.va_mode = S_IFDIR | 0555; vattr.va_uid = crgetuid(kcred); vattr.va_gid = crgetgid(kcred); sharezp = kmem_cache_alloc(znode_cache, KM_SLEEP); - ASSERT(!POINTER_IS_VALID(sharezp->z_zfsvfs)); sharezp->z_moved = 0; sharezp->z_unlinked = 0; sharezp->z_atime_dirty = 0; @@ -505,11 +196,14 @@ zfs_create_share_dir(zfsvfs_t *zfsvfs, dmu_tx_t *tx) zfsvfs->z_shares_dir = sharezp->z_id; zfs_acl_ids_free(&acl_ids); - ZTOV(sharezp)->v_count = 0; + // ZTOV(sharezp)->v_count = 0; sa_handle_destroy(sharezp->z_sa_hdl); kmem_cache_free(znode_cache, sharezp); return (error); +#else + return (0); +#endif /* HAVE_SHARE */ } /* @@ -546,42 +240,18 @@ zfs_expldev(dev_t dev) #endif } -/* - * Special cmpldev for ZFS private use. - * Can't use standard cmpldev since it takes - * a long dev_t and compresses it to dev32_t in - * LP64. We need to do a compaction of a long dev_t - * to a dev32_t in ILP32. - */ -dev_t -zfs_cmpldev(uint64_t dev) -{ -#ifndef _LP64 - minor_t minor = (minor_t)dev & MAXMIN64; - major_t major = (major_t)(dev >> NBITSMINOR64) & MAXMAJ64; - - if (major > MAXMAJ32 || minor > MAXMIN32) - return (NODEV32); - - return (((dev32_t)major << NBITSMINOR32) | minor); -#else - return (dev); -#endif -} - static void -zfs_znode_sa_init(zfsvfs_t *zfsvfs, znode_t *zp, +zfs_znode_sa_init(zfs_sb_t *zsb, znode_t *zp, dmu_buf_t *db, dmu_object_type_t obj_type, sa_handle_t *sa_hdl) { - ASSERT(!POINTER_IS_VALID(zp->z_zfsvfs) || (zfsvfs == zp->z_zfsvfs)); - ASSERT(MUTEX_HELD(ZFS_OBJ_MUTEX(zfsvfs, zp->z_id))); + ASSERT(MUTEX_HELD(ZFS_OBJ_MUTEX(zsb, zp->z_id))); mutex_enter(&zp->z_lock); ASSERT(zp->z_sa_hdl == NULL); ASSERT(zp->z_acl_cached == NULL); if (sa_hdl == NULL) { - VERIFY(0 == sa_handle_get_from_db(zfsvfs->z_os, db, zp, + VERIFY(0 == sa_handle_get_from_db(zsb->z_os, db, zp, SA_HDL_SHARED, &zp->z_sa_hdl)); } else { zp->z_sa_hdl = sa_hdl; @@ -590,55 +260,119 @@ zfs_znode_sa_init(zfsvfs_t *zfsvfs, znode_t *zp, zp->z_is_sa = (obj_type == DMU_OT_SA) ? B_TRUE : B_FALSE; - /* - * Slap on VROOT if we are the root znode - */ - if (zp->z_id == zfsvfs->z_root) - ZTOV(zp)->v_flag |= VROOT; - mutex_exit(&zp->z_lock); - vn_exists(ZTOV(zp)); } void zfs_znode_dmu_fini(znode_t *zp) { - ASSERT(MUTEX_HELD(ZFS_OBJ_MUTEX(zp->z_zfsvfs, zp->z_id)) || + ASSERT(MUTEX_HELD(ZFS_OBJ_MUTEX(ZTOZSB(zp), zp->z_id)) || zp->z_unlinked || - RW_WRITE_HELD(&zp->z_zfsvfs->z_teardown_inactive_lock)); + RW_WRITE_HELD(&ZTOZSB(zp)->z_teardown_inactive_lock)); sa_handle_destroy(zp->z_sa_hdl); zp->z_sa_hdl = NULL; } /* - * Construct a new znode/vnode and intialize. + * Called by new_inode() to allocate a new inode. + */ +int +zfs_inode_alloc(struct super_block *sb, struct inode **ip) +{ + znode_t *zp; + + zp = kmem_cache_alloc(znode_cache, KM_SLEEP); + *ip = ZTOI(zp); + + return (0); +} + +/* + * Called in multiple places when an inode should be destroyed. + */ +void +zfs_inode_destroy(struct inode *ip) +{ + znode_t *zp = ITOZ(ip); + zfs_sb_t *zsb = ZTOZSB(zp); + + mutex_enter(&zsb->z_znodes_lock); + list_remove(&zsb->z_all_znodes, zp); + mutex_exit(&zsb->z_znodes_lock); + + if (zp->z_acl_cached) { + zfs_acl_free(zp->z_acl_cached); + zp->z_acl_cached = NULL; + } + + kmem_cache_free(znode_cache, zp); +} + +static void +zfs_inode_set_ops(zfs_sb_t *zsb, struct inode *ip) +{ + uint64_t rdev; + + switch (ip->i_mode & S_IFMT) { + case S_IFREG: + ip->i_op = &zpl_inode_operations; + ip->i_fop = &zpl_file_operations; + ip->i_mapping->a_ops = &zpl_address_space_operations; + break; + + case S_IFDIR: + ip->i_op = &zpl_dir_inode_operations; + ip->i_fop = &zpl_dir_file_operations; + ITOZ(ip)->z_zn_prefetch = B_TRUE; + break; + + case S_IFLNK: + ip->i_op = &zpl_symlink_inode_operations; + break; + + case S_IFCHR: + case S_IFBLK: + case S_IFIFO: + case S_IFSOCK: + VERIFY(sa_lookup(ITOZ(ip)->z_sa_hdl, SA_ZPL_RDEV(zsb), + &rdev, sizeof (rdev)) == 0); + init_special_inode(ip, ip->i_mode, rdev); + ip->i_op = &zpl_special_inode_operations; + break; + + default: + printk("ZFS: Invalid mode: 0x%x\n", ip->i_mode); + VERIFY(0); + } +} + +/* + * Construct a znode+inode and initialize. * * This does not do a call to dmu_set_user() that is * up to the caller to do, in case you don't want to * return the znode */ static znode_t * -zfs_znode_alloc(zfsvfs_t *zfsvfs, dmu_buf_t *db, int blksz, - dmu_object_type_t obj_type, sa_handle_t *hdl) +zfs_znode_alloc(zfs_sb_t *zsb, dmu_buf_t *db, int blksz, + dmu_object_type_t obj_type, uint64_t obj, sa_handle_t *hdl) { znode_t *zp; - vnode_t *vp; - uint64_t mode; + struct inode *ip; uint64_t parent; sa_bulk_attr_t bulk[9]; int count = 0; - zp = kmem_cache_alloc(znode_cache, KM_SLEEP); + ASSERT(zsb != NULL); + + ip = new_inode(zsb->z_sb); + if (ip == NULL) + return (NULL); + zp = ITOZ(ip); ASSERT(zp->z_dirlocks == NULL); - ASSERT(!POINTER_IS_VALID(zp->z_zfsvfs)); zp->z_moved = 0; - - /* - * Defer setting z_zfsvfs until the znode is ready to be a candidate for - * the zfs_znode_move() callback. - */ zp->z_sa_hdl = NULL; zp->z_unlinked = 0; zp->z_atime_dirty = 0; @@ -648,93 +382,89 @@ zfs_znode_alloc(zfsvfs_t *zfsvfs, dmu_buf_t *db, int blksz, zp->z_seq = 0x7A4653; zp->z_sync_cnt = 0; - vp = ZTOV(zp); - vn_reinit(vp); - - zfs_znode_sa_init(zfsvfs, zp, db, obj_type, hdl); + zfs_znode_sa_init(zsb, zp, db, obj_type, hdl); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL, &mode, 8); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zfsvfs), NULL, &zp->z_gen, 8); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), NULL, - &zp->z_size, 8); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zfsvfs), NULL, - &zp->z_links, 8); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL, + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb), NULL, &zp->z_mode, 8); + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zsb), NULL, &zp->z_gen, 8); + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb), NULL, &zp->z_size, 8); + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zsb), NULL, &zp->z_links, 8); + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL, &zp->z_pflags, 8); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zfsvfs), NULL, &parent, 8); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL, + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zsb), NULL, + &parent, 8); + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zsb), NULL, &zp->z_atime, 16); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL, - &zp->z_uid, 8); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL, - &zp->z_gid, 8); + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zsb), NULL, &zp->z_uid, 8); + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zsb), NULL, &zp->z_gid, 8); if (sa_bulk_lookup(zp->z_sa_hdl, bulk, count) != 0 || zp->z_gen == 0) { if (hdl == NULL) sa_handle_destroy(zp->z_sa_hdl); - kmem_cache_free(znode_cache, zp); - return (NULL); - } - zp->z_mode = mode; - vp->v_vfsp = zfsvfs->z_parent->z_vfs; + goto error; + } - vp->v_type = IFTOVT((mode_t)mode); + ip->i_ino = obj; + ip->i_mode = zp->z_mode; + ip->i_mtime = ip->i_atime = ip->i_ctime = CURRENT_TIME_SEC; + zfs_inode_set_ops(zsb, ip); - switch (vp->v_type) { - case VDIR: - if (zp->z_pflags & ZFS_XATTR) { - vn_setops(vp, zfs_xdvnodeops); - vp->v_flag |= V_XATTRDIR; - } else { - vn_setops(vp, zfs_dvnodeops); - } - zp->z_zn_prefetch = B_TRUE; /* z_prefetch default is enabled */ - break; - case VBLK: - case VCHR: - { - uint64_t rdev; - VERIFY(sa_lookup(zp->z_sa_hdl, SA_ZPL_RDEV(zfsvfs), - &rdev, sizeof (rdev)) == 0); - - vp->v_rdev = zfs_cmpldev(rdev); - } - /*FALLTHROUGH*/ - case VFIFO: - case VSOCK: - case VDOOR: - vn_setops(vp, zfs_fvnodeops); - break; - case VREG: - vp->v_flag |= VMODSORT; - if (parent == zfsvfs->z_shares_dir) { - ASSERT(zp->z_uid == 0 && zp->z_gid == 0); - vn_setops(vp, zfs_sharevnodeops); - } else { - vn_setops(vp, zfs_fvnodeops); - } - break; - case VLNK: - vn_setops(vp, zfs_symvnodeops); - break; - default: - vn_setops(vp, zfs_evnodeops); - break; - } + if (insert_inode_locked(ip)) + goto error; - mutex_enter(&zfsvfs->z_znodes_lock); - list_insert_tail(&zfsvfs->z_all_znodes, zp); + mutex_enter(&zsb->z_znodes_lock); + list_insert_tail(&zsb->z_all_znodes, zp); membar_producer(); - /* - * Everything else must be valid before assigning z_zfsvfs makes the - * znode eligible for zfs_znode_move(). - */ - zp->z_zfsvfs = zfsvfs; - mutex_exit(&zfsvfs->z_znodes_lock); + mutex_exit(&zsb->z_znodes_lock); - VFS_HOLD(zfsvfs->z_vfs); + unlock_new_inode(ip); return (zp); + +error: + unlock_new_inode(ip); + iput(ip); + return NULL; +} + +/* + * Update the embedded inode given the znode. We should work toward + * eliminating this function as soon as possible by removing values + * which are duplicated between the znode and inode. If the generic + * inode has the correct field it should be used, and the ZFS code + * updated to access the inode. This can be done incrementally. + */ +void +zfs_inode_update(znode_t *zp) +{ + zfs_sb_t *zsb; + struct inode *ip; + uint32_t blksize; + uint64_t atime[2], mtime[2], ctime[2]; + + ASSERT(zp != NULL); + zsb = ZTOZSB(zp); + ip = ZTOI(zp); + + sa_lookup(zp->z_sa_hdl, SA_ZPL_ATIME(zsb), &atime, 16); + sa_lookup(zp->z_sa_hdl, SA_ZPL_MTIME(zsb), &mtime, 16); + sa_lookup(zp->z_sa_hdl, SA_ZPL_CTIME(zsb), &ctime, 16); + + spin_lock(&ip->i_lock); + ip->i_generation = zp->z_gen; + ip->i_uid = zp->z_uid; + ip->i_gid = zp->z_gid; + ip->i_nlink = zp->z_links; + ip->i_mode = zp->z_mode; + ip->i_blkbits = SPA_MINBLOCKSHIFT; + dmu_object_size_from_db(sa_get_db(zp->z_sa_hdl), &blksize, + (u_longlong_t *)&ip->i_blocks); + + ZFS_TIME_DECODE(&ip->i_atime, atime); + ZFS_TIME_DECODE(&ip->i_mtime, mtime); + ZFS_TIME_DECODE(&ip->i_ctime, ctime); + + i_size_write(ip, zp->z_size); + spin_unlock(&ip->i_lock); } static uint64_t empty_xattr; @@ -765,7 +495,7 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr, uint64_t mode, size, links, parent, pflags; uint64_t dzp_pflags = 0; uint64_t rdev = 0; - zfsvfs_t *zfsvfs = dzp->z_zfsvfs; + zfs_sb_t *zsb = ZTOZSB(dzp); dmu_buf_t *db; timestruc_t now; uint64_t gen, obj; @@ -773,13 +503,11 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr, int bonuslen; sa_handle_t *sa_hdl; dmu_object_type_t obj_type; - sa_bulk_attr_t sa_attrs[ZPL_END]; + sa_bulk_attr_t *sa_attrs; int cnt = 0; zfs_acl_locator_cb_t locate = { 0 }; - ASSERT(vap && (vap->va_mask & (AT_TYPE|AT_MODE)) == (AT_TYPE|AT_MODE)); - - if (zfsvfs->z_replay) { + if (zsb->z_replay) { obj = vap->va_nodeid; now = vap->va_ctime; /* see zfs_replay_create() */ gen = vap->va_nblocks; /* ditto */ @@ -789,7 +517,7 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr, gen = dmu_tx_get_txg(tx); } - obj_type = zfsvfs->z_use_sa ? DMU_OT_SA : DMU_OT_ZNODE; + obj_type = zsb->z_use_sa ? DMU_OT_SA : DMU_OT_ZNODE; bonuslen = (obj_type == DMU_OT_SA) ? DN_MAX_BONUSLEN : ZFS_OLD_ZNODE_PHYS_SIZE; @@ -802,32 +530,32 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr, * that there will be an i/o error and we will fail one of the * assertions below. */ - if (vap->va_type == VDIR) { - if (zfsvfs->z_replay) { - err = zap_create_claim_norm(zfsvfs->z_os, obj, - zfsvfs->z_norm, DMU_OT_DIRECTORY_CONTENTS, + if (S_ISDIR(vap->va_mode)) { + if (zsb->z_replay) { + err = zap_create_claim_norm(zsb->z_os, obj, + zsb->z_norm, DMU_OT_DIRECTORY_CONTENTS, obj_type, bonuslen, tx); ASSERT3U(err, ==, 0); } else { - obj = zap_create_norm(zfsvfs->z_os, - zfsvfs->z_norm, DMU_OT_DIRECTORY_CONTENTS, + obj = zap_create_norm(zsb->z_os, + zsb->z_norm, DMU_OT_DIRECTORY_CONTENTS, obj_type, bonuslen, tx); } } else { - if (zfsvfs->z_replay) { - err = dmu_object_claim(zfsvfs->z_os, obj, + if (zsb->z_replay) { + err = dmu_object_claim(zsb->z_os, obj, DMU_OT_PLAIN_FILE_CONTENTS, 0, obj_type, bonuslen, tx); ASSERT3U(err, ==, 0); } else { - obj = dmu_object_alloc(zfsvfs->z_os, + obj = dmu_object_alloc(zsb->z_os, DMU_OT_PLAIN_FILE_CONTENTS, 0, obj_type, bonuslen, tx); } } - ZFS_OBJ_HOLD_ENTER(zfsvfs, obj); - VERIFY(0 == sa_buf_hold(zfsvfs->z_os, obj, NULL, &db)); + ZFS_OBJ_HOLD_ENTER(zsb, obj); + VERIFY(0 == sa_buf_hold(zsb->z_os, obj, NULL, &db)); /* * If this is the root, fix up the half-initialized parent pointer @@ -846,21 +574,21 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr, flag |= IS_XATTR; } - if (zfsvfs->z_use_fuids) + if (zsb->z_use_fuids) pflags = ZFS_ARCHIVE | ZFS_AV_MODIFIED; else pflags = 0; - if (vap->va_type == VDIR) { + if (S_ISDIR(vap->va_mode)) { size = 2; /* contents ("." and "..") */ links = (flag & (IS_ROOT_NODE | IS_XATTR)) ? 2 : 1; } else { size = links = 0; } - if (vap->va_type == VBLK || vap->va_type == VCHR) { + if (S_ISBLK(vap->va_mode) || S_ISCHR(vap->va_mode) || + S_ISFIFO(vap->va_mode) || S_ISSOCK(vap->va_mode)) rdev = zfs_expldev(vap->va_rdev); - } parent = dzp->z_id; mode = acl_ids->z_mode; @@ -877,20 +605,20 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr, ZFS_TIME_ENCODE(&now, crtime); ZFS_TIME_ENCODE(&now, ctime); - if (vap->va_mask & AT_ATIME) { + if (vap->va_mask & ATTR_ATIME) { ZFS_TIME_ENCODE(&vap->va_atime, atime); } else { ZFS_TIME_ENCODE(&now, atime); } - if (vap->va_mask & AT_MTIME) { + if (vap->va_mask & ATTR_MTIME) { ZFS_TIME_ENCODE(&vap->va_mtime, mtime); } else { ZFS_TIME_ENCODE(&now, mtime); } /* Now add in all of the "SA" attributes */ - VERIFY(0 == sa_handle_get_from_db(zfsvfs->z_os, db, NULL, SA_HDL_SHARED, + VERIFY(0 == sa_handle_get_from_db(zsb->z_os, db, NULL, SA_HDL_SHARED, &sa_hdl)); /* @@ -899,77 +627,78 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr, * order for DMU_OT_ZNODE is critical since it needs to be constructed * in the old znode_phys_t format. Don't change this ordering */ + sa_attrs = kmem_alloc(sizeof(sa_bulk_attr_t) * ZPL_END, KM_SLEEP); if (obj_type == DMU_OT_ZNODE) { - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zsb), NULL, &atime, 16); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zsb), NULL, &mtime, 16); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zsb), NULL, &ctime, 16); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zsb), NULL, &crtime, 16); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zsb), NULL, &gen, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zsb), NULL, &mode, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zsb), NULL, &size, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zsb), NULL, &parent, 8); } else { - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zsb), NULL, &mode, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zsb), NULL, &size, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zsb), NULL, &gen, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_UID(zfsvfs), NULL, - &acl_ids->z_fuid, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GID(zfsvfs), NULL, - &acl_ids->z_fgid, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_UID(zsb), + NULL, &acl_ids->z_fuid, 8); + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GID(zsb), + NULL, &acl_ids->z_fgid, 8); + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zsb), NULL, &parent, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zsb), NULL, &pflags, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zsb), NULL, &atime, 16); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zsb), NULL, &mtime, 16); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zsb), NULL, &ctime, 16); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zsb), NULL, &crtime, 16); } - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_LINKS(zfsvfs), NULL, &links, 8); + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_LINKS(zsb), NULL, &links, 8); if (obj_type == DMU_OT_ZNODE) { - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_XATTR(zfsvfs), NULL, + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_XATTR(zsb), NULL, &empty_xattr, 8); } if (obj_type == DMU_OT_ZNODE || - (vap->va_type == VBLK || vap->va_type == VCHR)) { - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_RDEV(zfsvfs), + (S_ISBLK(vap->va_mode) || S_ISCHR(vap->va_mode) || + S_ISFIFO(vap->va_mode) || S_ISSOCK(vap->va_mode))) { + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_RDEV(zsb), NULL, &rdev, 8); - } if (obj_type == DMU_OT_ZNODE) { - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zsb), NULL, &pflags, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_UID(zfsvfs), NULL, + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_UID(zsb), NULL, &acl_ids->z_fuid, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GID(zfsvfs), NULL, + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GID(zsb), NULL, &acl_ids->z_fgid, 8); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PAD(zfsvfs), NULL, pad, + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PAD(zsb), NULL, pad, sizeof (uint64_t) * 4); - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ZNODE_ACL(zfsvfs), NULL, + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ZNODE_ACL(zsb), NULL, &acl_phys, sizeof (zfs_acl_phys_t)); } else if (acl_ids->z_aclp->z_version >= ZFS_ACL_VERSION_FUID) { - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_DACL_COUNT(zfsvfs), NULL, + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_DACL_COUNT(zsb), NULL, &acl_ids->z_aclp->z_acl_count, 8); locate.cb_aclp = acl_ids->z_aclp; - SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_DACL_ACES(zfsvfs), + SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_DACL_ACES(zsb), zfs_acl_data_locator, &locate, acl_ids->z_aclp->z_acl_bytes); mode = zfs_mode_compute(mode, acl_ids->z_aclp, &pflags, @@ -979,8 +708,11 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr, VERIFY(sa_replace_all_by_template(sa_hdl, sa_attrs, cnt, tx) == 0); if (!(flag & IS_ROOT_NODE)) { - *zpp = zfs_znode_alloc(zfsvfs, db, 0, obj_type, sa_hdl); + *zpp = zfs_znode_alloc(zsb, db, 0, obj_type, obj, sa_hdl); ASSERT(*zpp != NULL); + ASSERT(dzp != NULL); + err = zpl_xattr_security_init(ZTOI(*zpp), ZTOI(dzp)); + ASSERT3S(err, ==, 0); } else { /* * If we are creating the root node, the "parent" we @@ -994,115 +726,17 @@ zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr, (*zpp)->z_pflags = pflags; (*zpp)->z_mode = mode; - if (vap->va_mask & AT_XVATTR) - zfs_xvattr_set(*zpp, (xvattr_t *)vap, tx); - if (obj_type == DMU_OT_ZNODE || acl_ids->z_aclp->z_version < ZFS_ACL_VERSION_FUID) { err = zfs_aclset_common(*zpp, acl_ids->z_aclp, cr, tx); - ASSERT3P(err, ==, 0); - } - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj); -} - -/* - * zfs_xvattr_set only updates the in-core attributes - * it is assumed the caller will be doing an sa_bulk_update - * to push the changes out - */ -void -zfs_xvattr_set(znode_t *zp, xvattr_t *xvap, dmu_tx_t *tx) -{ - xoptattr_t *xoap; - - xoap = xva_getxoptattr(xvap); - ASSERT(xoap); - - if (XVA_ISSET_REQ(xvap, XAT_CREATETIME)) { - uint64_t times[2]; - ZFS_TIME_ENCODE(&xoap->xoa_createtime, times); - (void) sa_update(zp->z_sa_hdl, SA_ZPL_CRTIME(zp->z_zfsvfs), - ×, sizeof (times), tx); - XVA_SET_RTN(xvap, XAT_CREATETIME); - } - if (XVA_ISSET_REQ(xvap, XAT_READONLY)) { - ZFS_ATTR_SET(zp, ZFS_READONLY, xoap->xoa_readonly, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_READONLY); - } - if (XVA_ISSET_REQ(xvap, XAT_HIDDEN)) { - ZFS_ATTR_SET(zp, ZFS_HIDDEN, xoap->xoa_hidden, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_HIDDEN); - } - if (XVA_ISSET_REQ(xvap, XAT_SYSTEM)) { - ZFS_ATTR_SET(zp, ZFS_SYSTEM, xoap->xoa_system, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_SYSTEM); - } - if (XVA_ISSET_REQ(xvap, XAT_ARCHIVE)) { - ZFS_ATTR_SET(zp, ZFS_ARCHIVE, xoap->xoa_archive, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_ARCHIVE); - } - if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) { - ZFS_ATTR_SET(zp, ZFS_IMMUTABLE, xoap->xoa_immutable, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_IMMUTABLE); - } - if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) { - ZFS_ATTR_SET(zp, ZFS_NOUNLINK, xoap->xoa_nounlink, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_NOUNLINK); - } - if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) { - ZFS_ATTR_SET(zp, ZFS_APPENDONLY, xoap->xoa_appendonly, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_APPENDONLY); - } - if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) { - ZFS_ATTR_SET(zp, ZFS_NODUMP, xoap->xoa_nodump, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_NODUMP); - } - if (XVA_ISSET_REQ(xvap, XAT_OPAQUE)) { - ZFS_ATTR_SET(zp, ZFS_OPAQUE, xoap->xoa_opaque, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_OPAQUE); - } - if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) { - ZFS_ATTR_SET(zp, ZFS_AV_QUARANTINED, - xoap->xoa_av_quarantined, zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_AV_QUARANTINED); - } - if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) { - ZFS_ATTR_SET(zp, ZFS_AV_MODIFIED, xoap->xoa_av_modified, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_AV_MODIFIED); - } - if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) { - zfs_sa_set_scanstamp(zp, xvap, tx); - XVA_SET_RTN(xvap, XAT_AV_SCANSTAMP); - } - if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) { - ZFS_ATTR_SET(zp, ZFS_REPARSE, xoap->xoa_reparse, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_REPARSE); - } - if (XVA_ISSET_REQ(xvap, XAT_OFFLINE)) { - ZFS_ATTR_SET(zp, ZFS_OFFLINE, xoap->xoa_offline, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_OFFLINE); - } - if (XVA_ISSET_REQ(xvap, XAT_SPARSE)) { - ZFS_ATTR_SET(zp, ZFS_SPARSE, xoap->xoa_sparse, - zp->z_pflags, tx); - XVA_SET_RTN(xvap, XAT_SPARSE); + ASSERT3S(err, ==, 0); } + kmem_free(sa_attrs, sizeof(sa_bulk_attr_t) * ZPL_END); + ZFS_OBJ_HOLD_EXIT(zsb, obj); } int -zfs_zget(zfsvfs_t *zfsvfs, uint64_t obj_num, znode_t **zpp) +zfs_zget(zfs_sb_t *zsb, uint64_t obj_num, znode_t **zpp) { dmu_object_info_t doi; dmu_buf_t *db; @@ -1112,11 +746,11 @@ zfs_zget(zfsvfs_t *zfsvfs, uint64_t obj_num, znode_t **zpp) *zpp = NULL; - ZFS_OBJ_HOLD_ENTER(zfsvfs, obj_num); + ZFS_OBJ_HOLD_ENTER(zsb, obj_num); - err = sa_buf_hold(zfsvfs->z_os, obj_num, NULL, &db); + err = sa_buf_hold(zsb->z_os, obj_num, NULL, &db); if (err) { - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num); + ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (err); } @@ -1126,7 +760,7 @@ zfs_zget(zfsvfs_t *zfsvfs, uint64_t obj_num, znode_t **zpp) (doi.doi_bonus_type == DMU_OT_ZNODE && doi.doi_bonus_size < sizeof (znode_phys_t)))) { sa_buf_rele(db, NULL); - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num); + ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (EINVAL); } @@ -1148,19 +782,18 @@ zfs_zget(zfsvfs_t *zfsvfs, uint64_t obj_num, znode_t **zpp) if (zp->z_unlinked) { err = ENOENT; } else { - VN_HOLD(ZTOV(zp)); + igrab(ZTOI(zp)); *zpp = zp; err = 0; } sa_buf_rele(db, NULL); mutex_exit(&zp->z_lock); - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num); + ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (err); } /* - * Not found create new znode/vnode - * but only if file exists. + * Not found create new znode/vnode but only if file exists. * * There is a small window where zfs_vget() could * find this object while a file create is still in @@ -1169,21 +802,21 @@ zfs_zget(zfsvfs_t *zfsvfs, uint64_t obj_num, znode_t **zpp) * if zfs_znode_alloc() fails it will drop the hold on the * bonus buffer. */ - zp = zfs_znode_alloc(zfsvfs, db, doi.doi_data_block_size, - doi.doi_bonus_type, NULL); + zp = zfs_znode_alloc(zsb, db, doi.doi_data_block_size, + doi.doi_bonus_type, obj_num, NULL); if (zp == NULL) { err = ENOENT; } else { *zpp = zp; } - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num); + ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (err); } int zfs_rezget(znode_t *zp) { - zfsvfs_t *zfsvfs = zp->z_zfsvfs; + zfs_sb_t *zsb = ZTOZSB(zp); dmu_object_info_t doi; dmu_buf_t *db; uint64_t obj_num = zp->z_id; @@ -1193,7 +826,7 @@ zfs_rezget(znode_t *zp) int count = 0; uint64_t gen; - ZFS_OBJ_HOLD_ENTER(zfsvfs, obj_num); + ZFS_OBJ_HOLD_ENTER(zsb, obj_num); mutex_enter(&zp->z_acl_lock); if (zp->z_acl_cached) { @@ -1203,9 +836,9 @@ zfs_rezget(znode_t *zp) mutex_exit(&zp->z_acl_lock); ASSERT(zp->z_sa_hdl == NULL); - err = sa_buf_hold(zfsvfs->z_os, obj_num, NULL, &db); + err = sa_buf_hold(zsb->z_os, obj_num, NULL, &db); if (err) { - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num); + ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (err); } @@ -1215,33 +848,33 @@ zfs_rezget(znode_t *zp) (doi.doi_bonus_type == DMU_OT_ZNODE && doi.doi_bonus_size < sizeof (znode_phys_t)))) { sa_buf_rele(db, NULL); - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num); + ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (EINVAL); } - zfs_znode_sa_init(zfsvfs, zp, db, doi.doi_bonus_type, NULL); + zfs_znode_sa_init(zsb, zp, db, doi.doi_bonus_type, NULL); /* reload cached values */ - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zfsvfs), NULL, + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zsb), NULL, &gen, sizeof (gen)); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), NULL, + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb), NULL, &zp->z_size, sizeof (zp->z_size)); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zfsvfs), NULL, + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zsb), NULL, &zp->z_links, sizeof (zp->z_links)); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL, + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL, &zp->z_pflags, sizeof (zp->z_pflags)); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL, + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zsb), NULL, &zp->z_atime, sizeof (zp->z_atime)); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL, + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zsb), NULL, &zp->z_uid, sizeof (zp->z_uid)); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL, + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zsb), NULL, &zp->z_gid, sizeof (zp->z_gid)); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL, + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb), NULL, &mode, sizeof (mode)); if (sa_bulk_lookup(zp->z_sa_hdl, bulk, count)) { zfs_znode_dmu_fini(zp); - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num); + ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (EIO); } @@ -1249,14 +882,14 @@ zfs_rezget(znode_t *zp) if (gen != zp->z_gen) { zfs_znode_dmu_fini(zp); - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num); + ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (EIO); } zp->z_unlinked = (zp->z_links == 0); zp->z_blksz = doi.doi_data_block_size; - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj_num); + ZFS_OBJ_HOLD_EXIT(zsb, obj_num); return (0); } @@ -1264,27 +897,25 @@ zfs_rezget(znode_t *zp) void zfs_znode_delete(znode_t *zp, dmu_tx_t *tx) { - zfsvfs_t *zfsvfs = zp->z_zfsvfs; - objset_t *os = zfsvfs->z_os; + zfs_sb_t *zsb = ZTOZSB(zp); + objset_t *os = zsb->z_os; uint64_t obj = zp->z_id; uint64_t acl_obj = zfs_external_acl(zp); - ZFS_OBJ_HOLD_ENTER(zfsvfs, obj); + ZFS_OBJ_HOLD_ENTER(zsb, obj); if (acl_obj) { VERIFY(!zp->z_is_sa); VERIFY(0 == dmu_object_free(os, acl_obj, tx)); } VERIFY(0 == dmu_object_free(os, obj, tx)); zfs_znode_dmu_fini(zp); - ZFS_OBJ_HOLD_EXIT(zfsvfs, obj); - zfs_znode_free(zp); + ZFS_OBJ_HOLD_EXIT(zsb, obj); } void zfs_zinactive(znode_t *zp) { - vnode_t *vp = ZTOV(zp); - zfsvfs_t *zfsvfs = zp->z_zfsvfs; + zfs_sb_t *zsb = ZTOZSB(zp); uint64_t z_id = zp->z_id; ASSERT(zp->z_sa_hdl); @@ -1292,29 +923,8 @@ zfs_zinactive(znode_t *zp) /* * Don't allow a zfs_zget() while were trying to release this znode */ - ZFS_OBJ_HOLD_ENTER(zfsvfs, z_id); - + ZFS_OBJ_HOLD_ENTER(zsb, z_id); mutex_enter(&zp->z_lock); - mutex_enter(&vp->v_lock); - vp->v_count--; - if (vp->v_count > 0 || vn_has_cached_data(vp)) { - /* - * If the hold count is greater than zero, somebody has - * obtained a new reference on this znode while we were - * processing it here, so we are done. If we still have - * mapped pages then we are also done, since we don't - * want to inactivate the znode until the pages get pushed. - * - * XXX - if vn_has_cached_data(vp) is true, but count == 0, - * this seems like it would leave the znode hanging with - * no chance to go inactive... - */ - mutex_exit(&vp->v_lock); - mutex_exit(&zp->z_lock); - ZFS_OBJ_HOLD_EXIT(zfsvfs, z_id); - return; - } - mutex_exit(&vp->v_lock); /* * If this was the last reference to a file with no links, @@ -1322,39 +932,14 @@ zfs_zinactive(znode_t *zp) */ if (zp->z_unlinked) { mutex_exit(&zp->z_lock); - ZFS_OBJ_HOLD_EXIT(zfsvfs, z_id); + ZFS_OBJ_HOLD_EXIT(zsb, z_id); zfs_rmnode(zp); return; } mutex_exit(&zp->z_lock); zfs_znode_dmu_fini(zp); - ZFS_OBJ_HOLD_EXIT(zfsvfs, z_id); - zfs_znode_free(zp); -} - -void -zfs_znode_free(znode_t *zp) -{ - zfsvfs_t *zfsvfs = zp->z_zfsvfs; - - vn_invalid(ZTOV(zp)); - - ASSERT(ZTOV(zp)->v_count == 0); - - mutex_enter(&zfsvfs->z_znodes_lock); - POINTER_INVALIDATE(&zp->z_zfsvfs); - list_remove(&zfsvfs->z_all_znodes, zp); - mutex_exit(&zfsvfs->z_znodes_lock); - - if (zp->z_acl_cached) { - zfs_acl_free(zp->z_acl_cached); - zp->z_acl_cached = NULL; - } - - kmem_cache_free(znode_cache, zp); - - VFS_RELE(zfsvfs->z_vfs); + ZFS_OBJ_HOLD_EXIT(zsb, z_id); } void @@ -1372,21 +957,21 @@ zfs_tstamp_update_setup(znode_t *zp, uint_t flag, uint64_t mtime[2], zp->z_atime_dirty = 1; } - if (flag & AT_ATIME) { + if (flag & ATTR_ATIME) { ZFS_TIME_ENCODE(&now, zp->z_atime); } - if (flag & AT_MTIME) { + if (flag & ATTR_MTIME) { ZFS_TIME_ENCODE(&now, mtime); - if (zp->z_zfsvfs->z_use_fuids) { + if (ZTOZSB(zp)->z_use_fuids) { zp->z_pflags |= (ZFS_ARCHIVE | ZFS_AV_MODIFIED); } } - if (flag & AT_CTIME) { + if (flag & ATTR_CTIME) { ZFS_TIME_ENCODE(&now, ctime); - if (zp->z_zfsvfs->z_use_fuids) + if (ZTOZSB(zp)->z_use_fuids) zp->z_pflags |= ZFS_ARCHIVE; } } @@ -1416,7 +1001,7 @@ zfs_grow_blocksize(znode_t *zp, uint64_t size, dmu_tx_t *tx) if (zp->z_blksz && zp->z_size > zp->z_blksz) return; - error = dmu_object_set_blocksize(zp->z_zfsvfs->z_os, zp->z_id, + error = dmu_object_set_blocksize(ZTOZSB(zp)->z_os, zp->z_id, size, 0, tx); if (error == ENOTSUP) @@ -1427,6 +1012,7 @@ zfs_grow_blocksize(znode_t *zp, uint64_t size, dmu_tx_t *tx) dmu_object_size_from_db(sa_get_db(zp->z_sa_hdl), &zp->z_blksz, &dummy); } +#ifdef HAVE_MMAP /* * This is a dummy interface used when pvn_vplist_dirty() should *not* * be calling back into the fs for a putpage(). E.g.: when truncating @@ -1440,6 +1026,7 @@ zfs_no_putpage(vnode_t *vp, page_t *pp, u_offset_t *offp, size_t *lenp, ASSERT(0); return (0); } +#endif /* HAVE_MMAP */ /* * Increase the file length @@ -1453,7 +1040,7 @@ zfs_no_putpage(vnode_t *vp, page_t *pp, u_offset_t *offp, size_t *lenp, static int zfs_extend(znode_t *zp, uint64_t end) { - zfsvfs_t *zfsvfs = zp->z_zfsvfs; + zfs_sb_t *zsb = ZTOZSB(zp); dmu_tx_t *tx; rl_t *rl; uint64_t newblksz; @@ -1472,19 +1059,19 @@ zfs_extend(znode_t *zp, uint64_t end) return (0); } top: - tx = dmu_tx_create(zfsvfs->z_os); + tx = dmu_tx_create(zsb->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); if (end > zp->z_blksz && - (!ISP2(zp->z_blksz) || zp->z_blksz < zfsvfs->z_max_blksz)) { + (!ISP2(zp->z_blksz) || zp->z_blksz < zsb->z_max_blksz)) { /* * We are growing the file past the current block size. */ - if (zp->z_blksz > zp->z_zfsvfs->z_max_blksz) { + if (zp->z_blksz > ZTOZSB(zp)->z_max_blksz) { ASSERT(!ISP2(zp->z_blksz)); newblksz = MIN(end, SPA_MAXBLOCKSIZE); } else { - newblksz = MIN(end, zp->z_zfsvfs->z_max_blksz); + newblksz = MIN(end, ZTOZSB(zp)->z_max_blksz); } dmu_tx_hold_write(tx, zp->z_id, 0, newblksz); } else { @@ -1508,7 +1095,7 @@ top: zp->z_size = end; - VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zp->z_zfsvfs), + VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(ZTOZSB(zp)), &zp->z_size, sizeof (zp->z_size), tx)); zfs_range_unlock(rl); @@ -1525,13 +1112,13 @@ top: * off - start of section to free. * len - length of section to free. * - * RETURN: 0 if success + * RETURN: 0 if success * error code if failure */ static int zfs_free_range(znode_t *zp, uint64_t off, uint64_t len) { - zfsvfs_t *zfsvfs = zp->z_zfsvfs; + zfs_sb_t *zsb = ZTOZSB(zp); rl_t *rl; int error; @@ -1551,7 +1138,7 @@ zfs_free_range(znode_t *zp, uint64_t off, uint64_t len) if (off + len > zp->z_size) len = zp->z_size - off; - error = dmu_free_long_range(zfsvfs->z_os, zp->z_id, off, len); + error = dmu_free_long_range(zsb->z_os, zp->z_id, off, len); zfs_range_unlock(rl); @@ -1564,14 +1151,13 @@ zfs_free_range(znode_t *zp, uint64_t off, uint64_t len) * IN: zp - znode of file to free data in. * end - new end-of-file. * - * RETURN: 0 if success + * RETURN: 0 if success * error code if failure */ static int zfs_trunc(znode_t *zp, uint64_t end) { - zfsvfs_t *zfsvfs = zp->z_zfsvfs; - vnode_t *vp = ZTOV(zp); + zfs_sb_t *zsb = ZTOZSB(zp); dmu_tx_t *tx; rl_t *rl; int error; @@ -1591,13 +1177,13 @@ zfs_trunc(znode_t *zp, uint64_t end) return (0); } - error = dmu_free_long_range(zfsvfs->z_os, zp->z_id, end, -1); + error = dmu_free_long_range(zsb->z_os, zp->z_id, end, -1); if (error) { zfs_range_unlock(rl); return (error); } top: - tx = dmu_tx_create(zfsvfs->z_os); + tx = dmu_tx_create(zsb->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, TXG_NOWAIT); @@ -1613,42 +1199,18 @@ top: } zp->z_size = end; - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb), NULL, &zp->z_size, sizeof (zp->z_size)); if (end == 0) { zp->z_pflags &= ~ZFS_SPARSE; - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL, &zp->z_pflags, 8); } VERIFY(sa_bulk_update(zp->z_sa_hdl, bulk, count, tx) == 0); dmu_tx_commit(tx); - /* - * Clear any mapped pages in the truncated region. This has to - * happen outside of the transaction to avoid the possibility of - * a deadlock with someone trying to push a page that we are - * about to invalidate. - */ - if (vn_has_cached_data(vp)) { - page_t *pp; - uint64_t start = end & PAGEMASK; - int poff = end & PAGEOFFSET; - - if (poff != 0 && (pp = page_lookup(vp, start, SE_SHARED))) { - /* - * We need to zero a partial page. - */ - pagezero(pp, poff, PAGESIZE - poff); - start += PAGESIZE; - page_unlock(pp); - } - error = pvn_vplist_dirty(vp, start, zfs_no_putpage, - B_INVAL | B_TRUNC, NULL); - ASSERT(error == 0); - } - zfs_range_unlock(rl); return (0); @@ -1663,23 +1225,25 @@ top: * flag - current file open mode flags. * log - TRUE if this action should be logged * - * RETURN: 0 if success + * RETURN: 0 if success * error code if failure */ int zfs_freesp(znode_t *zp, uint64_t off, uint64_t len, int flag, boolean_t log) { - vnode_t *vp = ZTOV(zp); +#ifdef HAVE_MANDLOCKS + struct inode *ip = ZTOI(zp); +#endif /* HAVE_MANDLOCKS */ dmu_tx_t *tx; - zfsvfs_t *zfsvfs = zp->z_zfsvfs; - zilog_t *zilog = zfsvfs->z_log; + zfs_sb_t *zsb = ZTOZSB(zp); + zilog_t *zilog = zsb->z_log; uint64_t mode; uint64_t mtime[2], ctime[2]; sa_bulk_attr_t bulk[3]; int count = 0; int error; - if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_MODE(zfsvfs), &mode, + if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_MODE(zsb), &mode, sizeof (mode))) != 0) return (error); @@ -1691,15 +1255,17 @@ zfs_freesp(znode_t *zp, uint64_t off, uint64_t len, int flag, boolean_t log) return (error); } +#ifdef HAVE_MANDLOCKS /* * Check for any locks in the region to be freed. */ - if (MANDLOCK(vp, (mode_t)mode)) { + if (MANDLOCK(ip, (mode_t)mode)) { uint64_t length = (len ? len : zp->z_size - off); - if (error = chklock(vp, FWRITE, off, length, flag, NULL)) + if (error = chklock(ip, FWRITE, off, length, flag, NULL)) return (error); } +#endif /* HAVE_MANDLOCKS */ if (len == 0) { error = zfs_trunc(zp, off); @@ -1711,7 +1277,7 @@ zfs_freesp(znode_t *zp, uint64_t off, uint64_t len, int flag, boolean_t log) if (error || !log) return (error); log: - tx = dmu_tx_create(zfsvfs->z_os); + tx = dmu_tx_create(zsb->z_os); dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE); zfs_sa_upgrade_txholds(tx, zp); error = dmu_tx_assign(tx, TXG_NOWAIT); @@ -1725,9 +1291,9 @@ log: return (error); } - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL, mtime, 16); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL, ctime, 16); - SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zsb), NULL, mtime, 16); + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zsb), NULL, ctime, 16); + SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zsb), NULL, &zp->z_pflags, 8); zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime, B_TRUE); error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx); @@ -1736,31 +1302,20 @@ log: zfs_log_truncate(zilog, tx, TX_TRUNCATE, zp, off, len); dmu_tx_commit(tx); + zfs_inode_update(zp); return (0); } -#endif /* HAVE_ZPL */ void zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *zplprops, dmu_tx_t *tx) { uint64_t moid, obj, sa_obj, version; - uint64_t sense = ZFS_CASE_SENSITIVE; uint64_t norm = 0; nvpair_t *elem; int error; -#ifdef HAVE_ZPL - zfsvfs_t zfsvfs; - int i; - znode_t *rootzp = NULL; - vnode_t *vp; - vattr_t vattr; - znode_t *zp; - zfs_acl_ids_t acl_ids; -#else timestruc_t now; dmu_buf_t *db; znode_phys_t *pzp; -#endif /* HAVE_ZPL */ /* * First attempt to create master node. @@ -1796,8 +1351,6 @@ zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *zplprops, dmu_tx_t *tx) ASSERT(error == 0); if (strcmp(name, zfs_prop_to_name(ZFS_PROP_NORMALIZE)) == 0) norm = val; - else if (strcmp(name, zfs_prop_to_name(ZFS_PROP_CASE)) == 0) - sense = val; } ASSERT(version != 0); error = zap_update(os, moid, ZPL_VERSION_STR, 8, 1, &version, tx); @@ -1822,79 +1375,11 @@ zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *zplprops, dmu_tx_t *tx) error = zap_add(os, moid, ZFS_UNLINKED_SET, 8, 1, &obj, tx); ASSERT(error == 0); -#ifdef HAVE_ZPL - /* - * Create root znode. Create minimal znode/vnode/zfsvfs - * to allow zfs_mknode to work. - */ - vattr.va_mask = AT_MODE|AT_UID|AT_GID|AT_TYPE; - vattr.va_type = VDIR; - vattr.va_mode = S_IFDIR|0755; - vattr.va_uid = crgetuid(cr); - vattr.va_gid = crgetgid(cr); - - rootzp = kmem_cache_alloc(znode_cache, KM_SLEEP); - ASSERT(!POINTER_IS_VALID(rootzp->z_zfsvfs)); - rootzp->z_moved = 0; - rootzp->z_unlinked = 0; - rootzp->z_atime_dirty = 0; - rootzp->z_is_sa = USE_SA(version, os); - - vp = ZTOV(rootzp); - vn_reinit(vp); - vp->v_type = VDIR; - - bzero(&zfsvfs, sizeof (zfsvfs_t)); - - zfsvfs.z_os = os; - zfsvfs.z_parent = &zfsvfs; - zfsvfs.z_version = version; - zfsvfs.z_use_fuids = USE_FUIDS(version, os); - zfsvfs.z_use_sa = USE_SA(version, os); - zfsvfs.z_norm = norm; - - error = sa_setup(os, sa_obj, zfs_attr_table, ZPL_END, - &zfsvfs.z_attr_table); - - ASSERT(error == 0); - - /* - * Fold case on file systems that are always or sometimes case - * insensitive. - */ - if (sense == ZFS_CASE_INSENSITIVE || sense == ZFS_CASE_MIXED) - zfsvfs.z_norm |= U8_TEXTPREP_TOUPPER; - - /* XXX - This must be destroyed but I'm not quite sure yet so - * I'm just annotating that fact when it's an issue. -Brian */ - mutex_init(&zfsvfs.z_znodes_lock, NULL, MUTEX_DEFAULT, NULL); - list_create(&zfsvfs.z_all_znodes, sizeof (znode_t), - offsetof(znode_t, z_link_node)); - - for (i = 0; i != ZFS_OBJ_MTX_SZ; i++) - mutex_init(&zfsvfs.z_hold_mtx[i], NULL, MUTEX_DEFAULT, NULL); - - rootzp->z_zfsvfs = &zfsvfs; - VERIFY(0 == zfs_acl_ids_create(rootzp, IS_ROOT_NODE, &vattr, - cr, NULL, &acl_ids)); - zfs_mknode(rootzp, &vattr, tx, cr, IS_ROOT_NODE, &zp, &acl_ids); - ASSERT3P(zp, ==, rootzp); - ASSERT(!vn_in_dnlc(ZTOV(rootzp))); /* not valid to move */ - error = zap_add(os, moid, ZFS_ROOT_OBJ, 8, 1, &rootzp->z_id, tx); - ASSERT(error == 0); - zfs_acl_ids_free(&acl_ids); - POINTER_INVALIDATE(&rootzp->z_zfsvfs); - - ZTOV(rootzp)->v_count = 0; - sa_handle_destroy(rootzp->z_sa_hdl); - kmem_cache_free(znode_cache, rootzp); - error = zfs_create_share_dir(&zfsvfs, tx); - - for (i = 0; i != ZFS_OBJ_MTX_SZ; i++) - mutex_destroy(&zfsvfs.z_hold_mtx[i]); -#else /* - * Create root znode with code free of VFS dependencies + * Create root znode with code free of VFS dependencies. This + * is important because without a registered filesystem and super + * block all the required VFS hooks will be missing. The critical + * thing is to just crete the required root znode. */ obj = zap_create_norm(os, norm, DMU_OT_DIRECTORY_CONTENTS, DMU_OT_ZNODE, sizeof (znode_phys_t), tx); @@ -1930,7 +1415,6 @@ zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *zplprops, dmu_tx_t *tx) ASSERT(error == 0); dmu_buf_rele(db, FTAG); -#endif /* HAVE_ZPL */ } #endif /* _KERNEL */ |