*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
- * Copyright (c) 2012, 2014 by Delphix. All rights reserved.
+ * Copyright (c) 2012, 2018 by Delphix. All rights reserved.
*/
/* Portions Copyright 2007 Jeremy Teo */
#include <sys/types.h>
#include <sys/param.h>
#include <sys/time.h>
-#include <sys/systm.h>
#include <sys/sysmacros.h>
-#include <sys/resource.h>
#include <sys/mntent.h>
-#include <sys/mkdev.h>
#include <sys/u8_textprep.h>
#include <sys/dsl_dataset.h>
#include <sys/vfs.h>
-#include <sys/vfs_opreg.h>
#include <sys/vnode.h>
#include <sys/file.h>
#include <sys/kmem.h>
#include <sys/errno.h>
-#include <sys/unistd.h>
#include <sys/mode.h>
#include <sys/atomic.h>
-#include <vm/pvn.h>
-#include "fs/fs_subr.h"
#include <sys/zfs_dir.h>
#include <sys/zfs_acl.h>
#include <sys/zfs_ioctl.h>
#include <sys/zfs_ctldir.h>
#include <sys/dnode.h>
#include <sys/fs/zfs.h>
-#include <sys/kidmap.h>
#include <sys/zpl.h>
#endif /* _KERNEL */
#include <sys/dmu.h>
#include <sys/dmu_objset.h>
+#include <sys/dmu_tx.h>
#include <sys/refcount.h>
#include <sys/stat.h>
#include <sys/zap.h>
#include "zfs_prop.h"
#include "zfs_comutil.h"
-/*
- * Define ZNODE_STATS to turn on statistic gathering. By default, it is only
- * turned on when DEBUG is also defined.
- */
-#ifdef DEBUG
-#define ZNODE_STATS
-#endif /* DEBUG */
-
-#ifdef ZNODE_STATS
-#define ZNODE_STAT_ADD(stat) ((stat)++)
-#else
-#define ZNODE_STAT_ADD(stat) /* nothing */
-#endif /* ZNODE_STATS */
-
/*
* Functions needed for userland (ie: libzpool) are not put under
* #ifdef_KERNEL; the rest of the functions have dependencies
static kmem_cache_t *znode_hold_cache = NULL;
unsigned int zfs_object_mutex_size = ZFS_OBJ_MTX_SZ;
+/*
+ * This is used by the test suite so that it can delay znodes from being
+ * freed in order to inspect the unlinked set.
+ */
+int zfs_unlink_suspend_progress = 0;
+
+/*
+ * This callback is invoked when acquiring a RL_WRITER or RL_APPEND lock on
+ * z_rangelock. It will modify the offset and length of the lock to reflect
+ * znode-specific information, and convert RL_APPEND to RL_WRITER. This is
+ * called with the rangelock_t's rl_lock held, which avoids races.
+ */
+static void
+zfs_rangelock_cb(locked_range_t *new, void *arg)
+{
+ znode_t *zp = arg;
+
+ /*
+ * If in append mode, convert to writer and lock starting at the
+ * current end of file.
+ */
+ if (new->lr_type == RL_APPEND) {
+ new->lr_offset = zp->z_size;
+ new->lr_type = RL_WRITER;
+ }
+
+ /*
+ * If we need to grow the block size then lock the whole file range.
+ */
+ uint64_t end_size = MAX(zp->z_size, new->lr_offset + new->lr_length);
+ if (end_size > zp->z_blksz && (!ISP2(zp->z_blksz) ||
+ zp->z_blksz < ZTOZSB(zp)->z_max_blksz)) {
+ new->lr_offset = 0;
+ new->lr_length = UINT64_MAX;
+ }
+}
+
/*ARGSUSED*/
static int
zfs_znode_cache_constructor(void *buf, void *arg, int kmflags)
mutex_init(&zp->z_acl_lock, NULL, MUTEX_DEFAULT, NULL);
rw_init(&zp->z_xattr_lock, NULL, RW_DEFAULT, NULL);
- zfs_rlock_init(&zp->z_range_lock);
+ rangelock_init(&zp->z_rangelock, zfs_rangelock_cb, zp);
zp->z_dirlocks = NULL;
zp->z_acl_cached = NULL;
zp->z_xattr_cached = NULL;
- zp->z_xattr_parent = NULL;
+ zp->z_xattr_parent = 0;
zp->z_moved = 0;
return (0);
}
rw_destroy(&zp->z_name_lock);
mutex_destroy(&zp->z_acl_lock);
rw_destroy(&zp->z_xattr_lock);
- zfs_rlock_destroy(&zp->z_range_lock);
+ rangelock_fini(&zp->z_rangelock);
ASSERT(zp->z_dirlocks == NULL);
ASSERT(zp->z_acl_cached == NULL);
ASSERT(zp->z_xattr_cached == NULL);
- ASSERT(zp->z_xattr_parent == NULL);
}
static int
znode_hold_t *zh = buf;
mutex_init(&zh->zh_lock, NULL, MUTEX_DEFAULT, NULL);
- refcount_create(&zh->zh_refcount);
+ zfs_refcount_create(&zh->zh_refcount);
zh->zh_obj = ZFS_NO_OBJECT;
return (0);
znode_hold_t *zh = buf;
mutex_destroy(&zh->zh_lock);
- refcount_destroy(&zh->zh_refcount);
+ zfs_refcount_destroy(&zh->zh_refcount);
}
void
int
zfs_znode_hold_compare(const void *a, const void *b)
{
- const znode_hold_t *zh_a = a;
- const znode_hold_t *zh_b = b;
+ const znode_hold_t *zh_a = (const znode_hold_t *)a;
+ const znode_hold_t *zh_b = (const znode_hold_t *)b;
- if (zh_a->zh_obj < zh_b->zh_obj)
- return (-1);
- else if (zh_a->zh_obj > zh_b->zh_obj)
- return (1);
- else
- return (0);
+ return (AVL_CMP(zh_a->zh_obj, zh_b->zh_obj));
}
boolean_t
-zfs_znode_held(zfs_sb_t *zsb, uint64_t obj)
+zfs_znode_held(zfsvfs_t *zfsvfs, uint64_t obj)
{
znode_hold_t *zh, search;
- int i = ZFS_OBJ_HASH(zsb, obj);
+ int i = ZFS_OBJ_HASH(zfsvfs, obj);
boolean_t held;
search.zh_obj = obj;
- mutex_enter(&zsb->z_hold_locks[i]);
- zh = avl_find(&zsb->z_hold_trees[i], &search, NULL);
+ mutex_enter(&zfsvfs->z_hold_locks[i]);
+ zh = avl_find(&zfsvfs->z_hold_trees[i], &search, NULL);
held = (zh && MUTEX_HELD(&zh->zh_lock)) ? B_TRUE : B_FALSE;
- mutex_exit(&zsb->z_hold_locks[i]);
+ mutex_exit(&zfsvfs->z_hold_locks[i]);
return (held);
}
static znode_hold_t *
-zfs_znode_hold_enter(zfs_sb_t *zsb, uint64_t obj)
+zfs_znode_hold_enter(zfsvfs_t *zfsvfs, uint64_t obj)
{
znode_hold_t *zh, *zh_new, search;
- int i = ZFS_OBJ_HASH(zsb, obj);
+ int i = ZFS_OBJ_HASH(zfsvfs, obj);
boolean_t found = B_FALSE;
zh_new = kmem_cache_alloc(znode_hold_cache, KM_SLEEP);
zh_new->zh_obj = obj;
search.zh_obj = obj;
- mutex_enter(&zsb->z_hold_locks[i]);
- zh = avl_find(&zsb->z_hold_trees[i], &search, NULL);
+ mutex_enter(&zfsvfs->z_hold_locks[i]);
+ zh = avl_find(&zfsvfs->z_hold_trees[i], &search, NULL);
if (likely(zh == NULL)) {
zh = zh_new;
- avl_add(&zsb->z_hold_trees[i], zh);
+ avl_add(&zfsvfs->z_hold_trees[i], zh);
} else {
ASSERT3U(zh->zh_obj, ==, obj);
found = B_TRUE;
}
- refcount_add(&zh->zh_refcount, NULL);
- mutex_exit(&zsb->z_hold_locks[i]);
+ zfs_refcount_add(&zh->zh_refcount, NULL);
+ mutex_exit(&zfsvfs->z_hold_locks[i]);
if (found == B_TRUE)
kmem_cache_free(znode_hold_cache, zh_new);
ASSERT(MUTEX_NOT_HELD(&zh->zh_lock));
- ASSERT3S(refcount_count(&zh->zh_refcount), >, 0);
+ ASSERT3S(zfs_refcount_count(&zh->zh_refcount), >, 0);
mutex_enter(&zh->zh_lock);
return (zh);
}
static void
-zfs_znode_hold_exit(zfs_sb_t *zsb, znode_hold_t *zh)
+zfs_znode_hold_exit(zfsvfs_t *zfsvfs, znode_hold_t *zh)
{
- int i = ZFS_OBJ_HASH(zsb, zh->zh_obj);
+ int i = ZFS_OBJ_HASH(zfsvfs, zh->zh_obj);
boolean_t remove = B_FALSE;
- ASSERT(zfs_znode_held(zsb, zh->zh_obj));
- ASSERT3S(refcount_count(&zh->zh_refcount), >, 0);
+ ASSERT(zfs_znode_held(zfsvfs, zh->zh_obj));
+ ASSERT3S(zfs_refcount_count(&zh->zh_refcount), >, 0);
mutex_exit(&zh->zh_lock);
- mutex_enter(&zsb->z_hold_locks[i]);
- if (refcount_remove(&zh->zh_refcount, NULL) == 0) {
- avl_remove(&zsb->z_hold_trees[i], zh);
+ mutex_enter(&zfsvfs->z_hold_locks[i]);
+ if (zfs_refcount_remove(&zh->zh_refcount, NULL) == 0) {
+ avl_remove(&zfsvfs->z_hold_trees[i], zh);
remove = B_TRUE;
}
- mutex_exit(&zsb->z_hold_locks[i]);
+ mutex_exit(&zfsvfs->z_hold_locks[i]);
if (remove == B_TRUE)
kmem_cache_free(znode_hold_cache, zh);
}
-int
-zfs_create_share_dir(zfs_sb_t *zsb, dmu_tx_t *tx)
-{
-#ifdef HAVE_SMB_SHARE
- zfs_acl_ids_t acl_ids;
- vattr_t vattr;
- znode_t *sharezp;
- vnode_t *vp;
- znode_t *zp;
- int error;
-
- vattr.va_mask = AT_MODE|AT_UID|AT_GID|AT_TYPE;
- vattr.va_mode = S_IFDIR | 0555;
- vattr.va_uid = crgetuid(kcred);
- vattr.va_gid = crgetgid(kcred);
-
- sharezp = kmem_cache_alloc(znode_cache, KM_SLEEP);
- sharezp->z_moved = 0;
- sharezp->z_unlinked = 0;
- sharezp->z_atime_dirty = 0;
- sharezp->z_zfsvfs = zfsvfs;
- sharezp->z_is_sa = zfsvfs->z_use_sa;
-
- vp = ZTOV(sharezp);
- vn_reinit(vp);
- vp->v_type = VDIR;
-
- VERIFY(0 == zfs_acl_ids_create(sharezp, IS_ROOT_NODE, &vattr,
- kcred, NULL, &acl_ids));
- zfs_mknode(sharezp, &vattr, tx, kcred, IS_ROOT_NODE, &zp, &acl_ids);
- ASSERT3P(zp, ==, sharezp);
- ASSERT(!vn_in_dnlc(ZTOV(sharezp))); /* not valid to move */
- POINTER_INVALIDATE(&sharezp->z_zfsvfs);
- error = zap_add(zfsvfs->z_os, MASTER_NODE_OBJ,
- ZFS_SHARES_DIR, 8, 1, &sharezp->z_id, tx);
- zfsvfs->z_shares_dir = sharezp->z_id;
-
- zfs_acl_ids_free(&acl_ids);
- // 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_SMB_SHARE */
-}
-
static void
-zfs_znode_sa_init(zfs_sb_t *zsb, znode_t *zp,
+zfs_znode_sa_init(zfsvfs_t *zfsvfs, znode_t *zp,
dmu_buf_t *db, dmu_object_type_t obj_type, sa_handle_t *sa_hdl)
{
- ASSERT(zfs_znode_held(zsb, zp->z_id));
+ ASSERT(zfs_znode_held(zfsvfs, 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(zsb->z_os, db, zp,
+ VERIFY(0 == sa_handle_get_from_db(zfsvfs->z_os, db, zp,
SA_HDL_SHARED, &zp->z_sa_hdl));
} else {
zp->z_sa_hdl = sa_hdl;
zfs_inode_destroy(struct inode *ip)
{
znode_t *zp = ITOZ(ip);
- zfs_sb_t *zsb = ZTOZSB(zp);
+ zfsvfs_t *zfsvfs = ZTOZSB(zp);
- mutex_enter(&zsb->z_znodes_lock);
+ mutex_enter(&zfsvfs->z_znodes_lock);
if (list_link_active(&zp->z_link_node)) {
- list_remove(&zsb->z_all_znodes, zp);
- zsb->z_nr_znodes--;
+ list_remove(&zfsvfs->z_all_znodes, zp);
+ zfsvfs->z_nr_znodes--;
}
- mutex_exit(&zsb->z_znodes_lock);
+ mutex_exit(&zfsvfs->z_znodes_lock);
if (zp->z_acl_cached) {
zfs_acl_free(zp->z_acl_cached);
zp->z_xattr_cached = NULL;
}
- if (zp->z_xattr_parent) {
- zfs_iput_async(ZTOI(zp->z_xattr_parent));
- zp->z_xattr_parent = NULL;
- }
-
kmem_cache_free(znode_cache, zp);
}
static void
-zfs_inode_set_ops(zfs_sb_t *zsb, struct inode *ip)
+zfs_inode_set_ops(zfsvfs_t *zfsvfs, struct inode *ip)
{
uint64_t rdev = 0;
*/
case S_IFCHR:
case S_IFBLK:
- sa_lookup(ITOZ(ip)->z_sa_hdl, SA_ZPL_RDEV(zsb), &rdev,
+ (void) sa_lookup(ITOZ(ip)->z_sa_hdl, SA_ZPL_RDEV(zfsvfs), &rdev,
sizeof (rdev));
/*FALLTHROUGH*/
case S_IFIFO:
* Linux and Solaris have different sets of file attributes, so we
* restrict this conversion to the intersection of the two.
*/
+#ifdef HAVE_INODE_SET_FLAGS
+ unsigned int flags = 0;
+ if (zp->z_pflags & ZFS_IMMUTABLE)
+ flags |= S_IMMUTABLE;
+ if (zp->z_pflags & ZFS_APPENDONLY)
+ flags |= S_APPEND;
+ inode_set_flags(ip, flags, S_IMMUTABLE|S_APPEND);
+#else
if (zp->z_pflags & ZFS_IMMUTABLE)
ip->i_flags |= S_IMMUTABLE;
else
ip->i_flags |= S_APPEND;
else
ip->i_flags &= ~S_APPEND;
+#endif
}
/*
* inode has the correct field it should be used, and the ZFS code
* updated to access the inode. This can be done incrementally.
*/
-static void
-zfs_inode_update_impl(znode_t *zp, boolean_t new)
+void
+zfs_inode_update(znode_t *zp)
{
- zfs_sb_t *zsb;
+ zfsvfs_t *zfsvfs;
struct inode *ip;
uint32_t blksize;
u_longlong_t i_blocks;
- uint64_t atime[2], mtime[2], ctime[2];
ASSERT(zp != NULL);
- zsb = ZTOZSB(zp);
+ zfsvfs = ZTOZSB(zp);
ip = ZTOI(zp);
/* Skip .zfs control nodes which do not exist on disk. */
if (zfsctl_is_node(ip))
return;
- 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);
-
dmu_object_size_from_db(sa_get_db(zp->z_sa_hdl), &blksize, &i_blocks);
spin_lock(&ip->i_lock);
- ip->i_uid = SUID_TO_KUID(zp->z_uid);
- ip->i_gid = SGID_TO_KGID(zp->z_gid);
- set_nlink(ip, zp->z_links);
- ip->i_mode = zp->z_mode;
- zfs_set_inode_flags(zp, ip);
- ip->i_blkbits = SPA_MINBLOCKSHIFT;
ip->i_blocks = i_blocks;
-
- /*
- * Only read atime from SA if we are newly created inode (or rezget),
- * otherwise i_atime might be dirty.
- */
- if (new)
- 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 void
-zfs_inode_update_new(znode_t *zp)
-{
- zfs_inode_update_impl(zp, B_TRUE);
-}
-
-void
-zfs_inode_update(znode_t *zp)
-{
- zfs_inode_update_impl(zp, B_FALSE);
-}
/*
* Construct a znode+inode and initialize.
* return the znode
*/
static znode_t *
-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,
- struct inode *dip)
+zfs_znode_alloc(zfsvfs_t *zfsvfs, dmu_buf_t *db, int blksz,
+ dmu_object_type_t obj_type, uint64_t obj, sa_handle_t *hdl)
{
znode_t *zp;
struct inode *ip;
uint64_t mode;
uint64_t parent;
uint64_t tmp_gen;
- sa_bulk_attr_t bulk[8];
+ uint64_t links;
+ uint64_t z_uid, z_gid;
+ uint64_t atime[2], mtime[2], ctime[2];
+ uint64_t projid = ZFS_DEFAULT_PROJID;
+ sa_bulk_attr_t bulk[11];
int count = 0;
- ASSERT(zsb != NULL);
+ ASSERT(zfsvfs != NULL);
- ip = new_inode(zsb->z_sb);
+ ip = new_inode(zfsvfs->z_sb);
if (ip == NULL)
return (NULL);
ASSERT(zp->z_dirlocks == NULL);
ASSERT3P(zp->z_acl_cached, ==, NULL);
ASSERT3P(zp->z_xattr_cached, ==, NULL);
- ASSERT3P(zp->z_xattr_parent, ==, NULL);
zp->z_moved = 0;
zp->z_sa_hdl = NULL;
zp->z_unlinked = 0;
zp->z_is_mapped = B_FALSE;
zp->z_is_ctldir = B_FALSE;
zp->z_is_stale = B_FALSE;
- zp->z_range_lock.zr_size = &zp->z_size;
- zp->z_range_lock.zr_blksz = &zp->z_blksz;
- zp->z_range_lock.zr_max_blksz = &ZTOZSB(zp)->z_max_blksz;
- zfs_znode_sa_init(zsb, zp, db, obj_type, hdl);
+ zfs_znode_sa_init(zfsvfs, zp, db, obj_type, hdl);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb), NULL, &mode, 8);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zsb), NULL, &tmp_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,
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL, &mode, 8);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zfsvfs), NULL, &tmp_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, &links, 8);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
&zp->z_pflags, 8);
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zsb), NULL,
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PARENT(zfsvfs), NULL,
&parent, 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 ||
- tmp_gen == 0) {
-
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL, &z_uid, 8);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL, &z_gid, 8);
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL, &atime, 16);
+ 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);
+
+ if (sa_bulk_lookup(zp->z_sa_hdl, bulk, count) != 0 || tmp_gen == 0 ||
+ (dmu_objset_projectquota_enabled(zfsvfs->z_os) &&
+ (zp->z_pflags & ZFS_PROJID) &&
+ sa_lookup(zp->z_sa_hdl, SA_ZPL_PROJID(zfsvfs), &projid, 8) != 0)) {
if (hdl == NULL)
sa_handle_destroy(zp->z_sa_hdl);
zp->z_sa_hdl = NULL;
goto error;
}
- zp->z_mode = mode;
+ zp->z_projid = projid;
+ zp->z_mode = ip->i_mode = mode;
ip->i_generation = (uint32_t)tmp_gen;
+ ip->i_blkbits = SPA_MINBLOCKSHIFT;
+ set_nlink(ip, (uint32_t)links);
+ zfs_uid_write(ip, z_uid);
+ zfs_gid_write(ip, z_gid);
+ zfs_set_inode_flags(zp, ip);
- /*
- * xattr znodes hold a reference on their unique parent
- */
- if (dip && zp->z_pflags & ZFS_XATTR) {
- igrab(dip);
- zp->z_xattr_parent = ITOZ(dip);
- }
+ /* Cache the xattr parent id */
+ if (zp->z_pflags & ZFS_XATTR)
+ zp->z_xattr_parent = parent;
+
+ ZFS_TIME_DECODE(&ip->i_atime, atime);
+ ZFS_TIME_DECODE(&ip->i_mtime, mtime);
+ ZFS_TIME_DECODE(&ip->i_ctime, ctime);
ip->i_ino = obj;
- zfs_inode_update_new(zp);
- zfs_inode_set_ops(zsb, ip);
+ zfs_inode_update(zp);
+ zfs_inode_set_ops(zfsvfs, ip);
/*
* The only way insert_inode_locked() can fail is if the ip->i_ino
*/
VERIFY3S(insert_inode_locked(ip), ==, 0);
- mutex_enter(&zsb->z_znodes_lock);
- list_insert_tail(&zsb->z_all_znodes, zp);
- zsb->z_nr_znodes++;
+ mutex_enter(&zfsvfs->z_znodes_lock);
+ list_insert_tail(&zfsvfs->z_all_znodes, zp);
+ zfsvfs->z_nr_znodes++;
membar_producer();
- mutex_exit(&zsb->z_znodes_lock);
+ mutex_exit(&zfsvfs->z_znodes_lock);
unlock_new_inode(ip);
return (zp);
void
zfs_mark_inode_dirty(struct inode *ip)
{
- zfs_sb_t *zsb = ITOZSB(ip);
+ zfsvfs_t *zfsvfs = ITOZSB(ip);
- if (zfs_is_readonly(zsb) || dmu_objset_is_snapshot(zsb->z_os))
+ if (zfs_is_readonly(zfsvfs) || dmu_objset_is_snapshot(zfsvfs->z_os))
return;
mark_inode_dirty(ip);
{
uint64_t crtime[2], atime[2], mtime[2], ctime[2];
uint64_t mode, size, links, parent, pflags;
- uint64_t dzp_pflags = 0;
+ uint64_t projid = ZFS_DEFAULT_PROJID;
uint64_t rdev = 0;
- zfs_sb_t *zsb = ZTOZSB(dzp);
+ zfsvfs_t *zfsvfs = ZTOZSB(dzp);
dmu_buf_t *db;
- timestruc_t now;
+ inode_timespec_t now;
uint64_t gen, obj;
int bonuslen;
+ int dnodesize;
sa_handle_t *sa_hdl;
dmu_object_type_t obj_type;
sa_bulk_attr_t *sa_attrs;
zfs_acl_locator_cb_t locate = { 0 };
znode_hold_t *zh;
- if (zsb->z_replay) {
+ if (zfsvfs->z_replay) {
obj = vap->va_nodeid;
now = vap->va_ctime; /* see zfs_replay_create() */
gen = vap->va_nblocks; /* ditto */
+ dnodesize = vap->va_fsid; /* ditto */
} else {
obj = 0;
gethrestime(&now);
gen = dmu_tx_get_txg(tx);
+ dnodesize = dmu_objset_dnodesize(zfsvfs->z_os);
}
- obj_type = zsb->z_use_sa ? DMU_OT_SA : DMU_OT_ZNODE;
+ if (dnodesize == 0)
+ dnodesize = DNODE_MIN_SIZE;
+
+ obj_type = zfsvfs->z_use_sa ? DMU_OT_SA : DMU_OT_ZNODE;
+
bonuslen = (obj_type == DMU_OT_SA) ?
- DN_MAX_BONUSLEN : ZFS_OLD_ZNODE_PHYS_SIZE;
+ DN_BONUS_SIZE(dnodesize) : ZFS_OLD_ZNODE_PHYS_SIZE;
/*
* Create a new DMU object.
* assertions below.
*/
if (S_ISDIR(vap->va_mode)) {
- if (zsb->z_replay) {
- VERIFY0(zap_create_claim_norm(zsb->z_os, obj,
- zsb->z_norm, DMU_OT_DIRECTORY_CONTENTS,
- obj_type, bonuslen, tx));
+ if (zfsvfs->z_replay) {
+ VERIFY0(zap_create_claim_norm_dnsize(zfsvfs->z_os, obj,
+ zfsvfs->z_norm, DMU_OT_DIRECTORY_CONTENTS,
+ obj_type, bonuslen, dnodesize, tx));
} else {
- obj = zap_create_norm(zsb->z_os,
- zsb->z_norm, DMU_OT_DIRECTORY_CONTENTS,
- obj_type, bonuslen, tx);
+ obj = zap_create_norm_dnsize(zfsvfs->z_os,
+ zfsvfs->z_norm, DMU_OT_DIRECTORY_CONTENTS,
+ obj_type, bonuslen, dnodesize, tx);
}
} else {
- if (zsb->z_replay) {
- VERIFY0(dmu_object_claim(zsb->z_os, obj,
+ if (zfsvfs->z_replay) {
+ VERIFY0(dmu_object_claim_dnsize(zfsvfs->z_os, obj,
DMU_OT_PLAIN_FILE_CONTENTS, 0,
- obj_type, bonuslen, tx));
+ obj_type, bonuslen, dnodesize, tx));
} else {
- obj = dmu_object_alloc(zsb->z_os,
+ obj = dmu_object_alloc_dnsize(zfsvfs->z_os,
DMU_OT_PLAIN_FILE_CONTENTS, 0,
- obj_type, bonuslen, tx);
+ obj_type, bonuslen, dnodesize, tx);
}
}
- zh = zfs_znode_hold_enter(zsb, obj);
- VERIFY(0 == sa_buf_hold(zsb->z_os, obj, NULL, &db));
+ zh = zfs_znode_hold_enter(zfsvfs, obj);
+ VERIFY0(sa_buf_hold(zfsvfs->z_os, obj, NULL, &db));
/*
* If this is the root, fix up the half-initialized parent pointer
*/
if (flag & IS_ROOT_NODE) {
dzp->z_id = obj;
- } else {
- dzp_pflags = dzp->z_pflags;
}
/*
* If parent is an xattr, so am I.
*/
- if (dzp_pflags & ZFS_XATTR) {
+ if (dzp->z_pflags & ZFS_XATTR) {
flag |= IS_XATTR;
}
- if (zsb->z_use_fuids)
+ if (zfsvfs->z_use_fuids)
pflags = ZFS_ARCHIVE | ZFS_AV_MODIFIED;
else
pflags = 0;
if (S_ISDIR(vap->va_mode)) {
size = 2; /* contents ("." and "..") */
- links = (flag & (IS_ROOT_NODE | IS_XATTR)) ? 2 : 1;
+ links = 2;
} else {
- size = links = 0;
+ size = 0;
+ links = (flag & IS_TMPFILE) ? 0 : 1;
}
if (S_ISBLK(vap->va_mode) || S_ISCHR(vap->va_mode))
if (flag & IS_XATTR)
pflags |= ZFS_XATTR;
+ if (S_ISREG(vap->va_mode) || S_ISDIR(vap->va_mode)) {
+ /*
+ * With ZFS_PROJID flag, we can easily know whether there is
+ * project ID stored on disk or not. See zfs_space_delta_cb().
+ */
+ if (obj_type != DMU_OT_ZNODE &&
+ dmu_objset_projectquota_enabled(zfsvfs->z_os))
+ pflags |= ZFS_PROJID;
+
+ /*
+ * Inherit project ID from parent if required.
+ */
+ projid = zfs_inherit_projid(dzp);
+ if (dzp->z_pflags & ZFS_PROJINHERIT)
+ pflags |= ZFS_PROJINHERIT;
+ }
+
/*
* No execs denied will be deterimed when zfs_mode_compute() is called.
*/
}
/* Now add in all of the "SA" attributes */
- VERIFY(0 == sa_handle_get_from_db(zsb->z_os, db, NULL, SA_HDL_SHARED,
+ VERIFY(0 == sa_handle_get_from_db(zfsvfs->z_os, db, NULL, SA_HDL_SHARED,
&sa_hdl));
/*
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(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zfsvfs),
NULL, &atime, 16);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zfsvfs),
NULL, &mtime, 16);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zfsvfs),
NULL, &ctime, 16);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zfsvfs),
NULL, &crtime, 16);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zfsvfs),
NULL, &gen, 8);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zfsvfs),
NULL, &mode, 8);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zfsvfs),
NULL, &size, 8);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zfsvfs),
NULL, &parent, 8);
} else {
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zfsvfs),
NULL, &mode, 8);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zfsvfs),
NULL, &size, 8);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zfsvfs),
NULL, &gen, 8);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_UID(zsb),
+ 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(zsb),
+ 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(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PARENT(zfsvfs),
NULL, &parent, 8);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zfsvfs),
NULL, &pflags, 8);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zfsvfs),
NULL, &atime, 16);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zfsvfs),
NULL, &mtime, 16);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zfsvfs),
NULL, &ctime, 16);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zfsvfs),
NULL, &crtime, 16);
}
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_LINKS(zsb), NULL, &links, 8);
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_LINKS(zfsvfs), NULL, &links, 8);
if (obj_type == DMU_OT_ZNODE) {
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_XATTR(zsb), NULL,
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_XATTR(zfsvfs), NULL,
&empty_xattr, 8);
+ } else if (dmu_objset_projectquota_enabled(zfsvfs->z_os) &&
+ pflags & ZFS_PROJID) {
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PROJID(zfsvfs),
+ NULL, &projid, 8);
}
if (obj_type == DMU_OT_ZNODE ||
(S_ISBLK(vap->va_mode) || S_ISCHR(vap->va_mode))) {
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_RDEV(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_RDEV(zfsvfs),
NULL, &rdev, 8);
}
if (obj_type == DMU_OT_ZNODE) {
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zfsvfs),
NULL, &pflags, 8);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_UID(zsb), NULL,
+ 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(zsb), NULL,
+ 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_PAD(zsb), NULL, pad,
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_PAD(zfsvfs), NULL, pad,
sizeof (uint64_t) * 4);
- SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ZNODE_ACL(zsb), NULL,
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ZNODE_ACL(zfsvfs), 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(zsb), NULL,
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_DACL_COUNT(zfsvfs), 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(zsb),
+ SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_DACL_ACES(zfsvfs),
zfs_acl_data_locator, &locate,
acl_ids->z_aclp->z_acl_bytes);
mode = zfs_mode_compute(mode, acl_ids->z_aclp, &pflags,
VERIFY(sa_replace_all_by_template(sa_hdl, sa_attrs, cnt, tx) == 0);
if (!(flag & IS_ROOT_NODE)) {
- *zpp = zfs_znode_alloc(zsb, db, 0, obj_type, obj, sa_hdl,
- ZTOI(dzp));
+ /*
+ * The call to zfs_znode_alloc() may fail if memory is low
+ * via the call path: alloc_inode() -> inode_init_always() ->
+ * security_inode_alloc() -> inode_alloc_security(). Since
+ * the existing code is written such that zfs_mknode() can
+ * not fail retry until sufficient memory has been reclaimed.
+ */
+ do {
+ *zpp = zfs_znode_alloc(zfsvfs, db, 0, obj_type, obj,
+ sa_hdl);
+ } while (*zpp == NULL);
+
VERIFY(*zpp != NULL);
VERIFY(dzp != NULL);
} else {
}
(*zpp)->z_pflags = pflags;
- (*zpp)->z_mode = mode;
+ (*zpp)->z_mode = ZTOI(*zpp)->i_mode = mode;
+ (*zpp)->z_dnodesize = dnodesize;
+ (*zpp)->z_projid = projid;
if (obj_type == DMU_OT_ZNODE ||
acl_ids->z_aclp->z_version < ZFS_ACL_VERSION_FUID) {
VERIFY0(zfs_aclset_common(*zpp, acl_ids->z_aclp, cr, tx));
}
kmem_free(sa_attrs, sizeof (sa_bulk_attr_t) * ZPL_END);
- zfs_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
}
/*
zfs_xvattr_set(znode_t *zp, xvattr_t *xvap, dmu_tx_t *tx)
{
xoptattr_t *xoap;
+ boolean_t update_inode = B_FALSE;
xoap = xva_getxoptattr(xvap);
ASSERT(xoap);
ZFS_ATTR_SET(zp, ZFS_IMMUTABLE, xoap->xoa_immutable,
zp->z_pflags, tx);
XVA_SET_RTN(xvap, XAT_IMMUTABLE);
+
+ update_inode = B_TRUE;
}
if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) {
ZFS_ATTR_SET(zp, ZFS_NOUNLINK, xoap->xoa_nounlink,
ZFS_ATTR_SET(zp, ZFS_APPENDONLY, xoap->xoa_appendonly,
zp->z_pflags, tx);
XVA_SET_RTN(xvap, XAT_APPENDONLY);
+
+ update_inode = B_TRUE;
}
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_SPARSE);
}
+ if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT)) {
+ ZFS_ATTR_SET(zp, ZFS_PROJINHERIT, xoap->xoa_projinherit,
+ zp->z_pflags, tx);
+ XVA_SET_RTN(xvap, XAT_PROJINHERIT);
+ }
+
+ if (update_inode)
+ zfs_set_inode_flags(zp, ZTOI(zp));
}
int
-zfs_zget(zfs_sb_t *zsb, uint64_t obj_num, znode_t **zpp)
+zfs_zget(zfsvfs_t *zfsvfs, uint64_t obj_num, znode_t **zpp)
{
dmu_object_info_t doi;
dmu_buf_t *db;
*zpp = NULL;
again:
- zh = zfs_znode_hold_enter(zsb, obj_num);
+ zh = zfs_znode_hold_enter(zfsvfs, obj_num);
- err = sa_buf_hold(zsb->z_os, obj_num, NULL, &db);
+ err = sa_buf_hold(zfsvfs->z_os, obj_num, NULL, &db);
if (err) {
- zfs_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
return (err);
}
(doi.doi_bonus_type == DMU_OT_ZNODE &&
doi.doi_bonus_size < sizeof (znode_phys_t)))) {
sa_buf_rele(db, NULL);
- zfs_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
return (SET_ERROR(EINVAL));
}
mutex_enter(&zp->z_lock);
ASSERT3U(zp->z_id, ==, obj_num);
- if (zp->z_unlinked) {
- err = SET_ERROR(ENOENT);
- } else {
- /*
- * If igrab() returns NULL the VFS has independently
- * determined the inode should be evicted and has
- * called iput_final() to start the eviction process.
- * The SA handle is still valid but because the VFS
- * requires that the eviction succeed we must drop
- * our locks and references to allow the eviction to
- * complete. The zfs_zget() may then be retried.
- *
- * This unlikely case could be optimized by registering
- * a sops->drop_inode() callback. The callback would
- * need to detect the active SA hold thereby informing
- * the VFS that this inode should not be evicted.
- */
- if (igrab(ZTOI(zp)) == NULL) {
- mutex_exit(&zp->z_lock);
- sa_buf_rele(db, NULL);
- zfs_znode_hold_exit(zsb, zh);
- /* inode might need this to finish evict */
- cond_resched();
- goto again;
- }
- *zpp = zp;
- err = 0;
+ /*
+ * If igrab() returns NULL the VFS has independently
+ * determined the inode should be evicted and has
+ * called iput_final() to start the eviction process.
+ * The SA handle is still valid but because the VFS
+ * requires that the eviction succeed we must drop
+ * our locks and references to allow the eviction to
+ * complete. The zfs_zget() may then be retried.
+ *
+ * This unlikely case could be optimized by registering
+ * a sops->drop_inode() callback. The callback would
+ * need to detect the active SA hold thereby informing
+ * the VFS that this inode should not be evicted.
+ */
+ if (igrab(ZTOI(zp)) == NULL) {
+ mutex_exit(&zp->z_lock);
+ sa_buf_rele(db, NULL);
+ zfs_znode_hold_exit(zfsvfs, zh);
+ /* inode might need this to finish evict */
+ cond_resched();
+ goto again;
}
+ *zpp = zp;
+ err = 0;
mutex_exit(&zp->z_lock);
sa_buf_rele(db, NULL);
- zfs_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
return (err);
}
* if zfs_znode_alloc() fails it will drop the hold on the
* bonus buffer.
*/
- zp = zfs_znode_alloc(zsb, db, doi.doi_data_block_size,
- doi.doi_bonus_type, obj_num, NULL, NULL);
+ zp = zfs_znode_alloc(zfsvfs, db, doi.doi_data_block_size,
+ doi.doi_bonus_type, obj_num, NULL);
if (zp == NULL) {
err = SET_ERROR(ENOENT);
} else {
*zpp = zp;
}
- zfs_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
return (err);
}
int
zfs_rezget(znode_t *zp)
{
- zfs_sb_t *zsb = ZTOZSB(zp);
+ zfsvfs_t *zfsvfs = ZTOZSB(zp);
dmu_object_info_t doi;
dmu_buf_t *db;
uint64_t obj_num = zp->z_id;
uint64_t mode;
- sa_bulk_attr_t bulk[7];
+ uint64_t links;
+ sa_bulk_attr_t bulk[10];
int err;
int count = 0;
uint64_t gen;
+ uint64_t z_uid, z_gid;
+ uint64_t atime[2], mtime[2], ctime[2];
+ uint64_t projid = ZFS_DEFAULT_PROJID;
znode_hold_t *zh;
/*
if (zp->z_is_ctldir)
return (0);
- zh = zfs_znode_hold_enter(zsb, obj_num);
+ zh = zfs_znode_hold_enter(zfsvfs, obj_num);
mutex_enter(&zp->z_acl_lock);
if (zp->z_acl_cached) {
nvlist_free(zp->z_xattr_cached);
zp->z_xattr_cached = NULL;
}
-
- if (zp->z_xattr_parent) {
- zfs_iput_async(ZTOI(zp->z_xattr_parent));
- zp->z_xattr_parent = NULL;
- }
rw_exit(&zp->z_xattr_lock);
ASSERT(zp->z_sa_hdl == NULL);
- err = sa_buf_hold(zsb->z_os, obj_num, NULL, &db);
+ err = sa_buf_hold(zfsvfs->z_os, obj_num, NULL, &db);
if (err) {
- zfs_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
return (err);
}
(doi.doi_bonus_type == DMU_OT_ZNODE &&
doi.doi_bonus_size < sizeof (znode_phys_t)))) {
sa_buf_rele(db, NULL);
- zfs_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
return (SET_ERROR(EINVAL));
}
- zfs_znode_sa_init(zsb, zp, db, doi.doi_bonus_type, NULL);
+ zfs_znode_sa_init(zfsvfs, zp, db, doi.doi_bonus_type, NULL);
/* reload cached values */
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zsb), NULL,
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zfsvfs), NULL,
&gen, sizeof (gen));
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb), NULL,
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs), NULL,
&zp->z_size, sizeof (zp->z_size));
- 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(zsb), NULL,
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_LINKS(zfsvfs), NULL,
+ &links, sizeof (links));
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
&zp->z_pflags, sizeof (zp->z_pflags));
- 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(zsb), NULL,
- &zp->z_gid, sizeof (zp->z_gid));
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zsb), NULL,
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
+ &z_uid, sizeof (z_uid));
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL,
+ &z_gid, sizeof (z_gid));
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL,
&mode, sizeof (mode));
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL,
+ &atime, 16);
+ 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);
if (sa_bulk_lookup(zp->z_sa_hdl, bulk, count)) {
zfs_znode_dmu_fini(zp);
- zfs_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
return (SET_ERROR(EIO));
}
- zp->z_mode = mode;
+ if (dmu_objset_projectquota_enabled(zfsvfs->z_os)) {
+ err = sa_lookup(zp->z_sa_hdl, SA_ZPL_PROJID(zfsvfs),
+ &projid, 8);
+ if (err != 0 && err != ENOENT) {
+ zfs_znode_dmu_fini(zp);
+ zfs_znode_hold_exit(zfsvfs, zh);
+ return (SET_ERROR(err));
+ }
+ }
+
+ zp->z_projid = projid;
+ zp->z_mode = ZTOI(zp)->i_mode = mode;
+ zfs_uid_write(ZTOI(zp), z_uid);
+ zfs_gid_write(ZTOI(zp), z_gid);
+
+ ZFS_TIME_DECODE(&ZTOI(zp)->i_atime, atime);
+ ZFS_TIME_DECODE(&ZTOI(zp)->i_mtime, mtime);
+ ZFS_TIME_DECODE(&ZTOI(zp)->i_ctime, ctime);
if (gen != ZTOI(zp)->i_generation) {
zfs_znode_dmu_fini(zp);
- zfs_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
return (SET_ERROR(EIO));
}
- zp->z_unlinked = (zp->z_links == 0);
+ set_nlink(ZTOI(zp), (uint32_t)links);
+ zfs_set_inode_flags(zp, ZTOI(zp));
+
zp->z_blksz = doi.doi_data_block_size;
zp->z_atime_dirty = 0;
- zfs_inode_update_new(zp);
+ zfs_inode_update(zp);
- zfs_znode_hold_exit(zsb, zh);
+ /*
+ * If the file has zero links, then it has been unlinked on the send
+ * side and it must be in the received unlinked set.
+ * We call zfs_znode_dmu_fini() now to prevent any accesses to the
+ * stale data and to prevent automatical removal of the file in
+ * zfs_zinactive(). The file will be removed either when it is removed
+ * on the send side and the next incremental stream is received or
+ * when the unlinked set gets processed.
+ */
+ zp->z_unlinked = (ZTOI(zp)->i_nlink == 0);
+ if (zp->z_unlinked)
+ zfs_znode_dmu_fini(zp);
+
+ zfs_znode_hold_exit(zfsvfs, zh);
return (0);
}
void
zfs_znode_delete(znode_t *zp, dmu_tx_t *tx)
{
- zfs_sb_t *zsb = ZTOZSB(zp);
- objset_t *os = zsb->z_os;
+ zfsvfs_t *zfsvfs = ZTOZSB(zp);
+ objset_t *os = zfsvfs->z_os;
uint64_t obj = zp->z_id;
uint64_t acl_obj = zfs_external_acl(zp);
znode_hold_t *zh;
- zh = zfs_znode_hold_enter(zsb, obj);
+ zh = zfs_znode_hold_enter(zfsvfs, 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_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
}
void
zfs_zinactive(znode_t *zp)
{
- zfs_sb_t *zsb = ZTOZSB(zp);
+ zfsvfs_t *zfsvfs = ZTOZSB(zp);
uint64_t z_id = zp->z_id;
znode_hold_t *zh;
/*
* Don't allow a zfs_zget() while were trying to release this znode.
*/
- zh = zfs_znode_hold_enter(zsb, z_id);
+ zh = zfs_znode_hold_enter(zfsvfs, z_id);
mutex_enter(&zp->z_lock);
/*
- * If this was the last reference to a file with no links,
- * remove the file from the file system.
+ * If this was the last reference to a file with no links, remove
+ * the file from the file system unless the file system is mounted
+ * read-only. That can happen, for example, if the file system was
+ * originally read-write, the file was opened, then unlinked and
+ * the file system was made read-only before the file was finally
+ * closed. The file will remain in the unlinked set.
*/
if (zp->z_unlinked) {
- mutex_exit(&zp->z_lock);
- zfs_znode_hold_exit(zsb, zh);
- zfs_rmnode(zp);
- return;
+ ASSERT(!zfsvfs->z_issnap);
+ if (!zfs_is_readonly(zfsvfs) && !zfs_unlink_suspend_progress) {
+ mutex_exit(&zp->z_lock);
+ zfs_znode_hold_exit(zfsvfs, zh);
+ zfs_rmnode(zp);
+ return;
+ }
}
mutex_exit(&zp->z_lock);
zfs_znode_dmu_fini(zp);
- zfs_znode_hold_exit(zsb, zh);
+ zfs_znode_hold_exit(zfsvfs, zh);
}
static inline int
zfs_tstamp_update_setup(znode_t *zp, uint_t flag, uint64_t mtime[2],
uint64_t ctime[2])
{
- timestruc_t now;
+ inode_timespec_t now;
gethrestime(&now);
if (flag & ATTR_MTIME) {
ZFS_TIME_ENCODE(&now, mtime);
+ ZFS_TIME_DECODE(&(ZTOI(zp)->i_mtime), mtime);
if (ZTOZSB(zp)->z_use_fuids) {
zp->z_pflags |= (ZFS_ARCHIVE |
ZFS_AV_MODIFIED);
if (flag & ATTR_CTIME) {
ZFS_TIME_ENCODE(&now, ctime);
+ ZFS_TIME_DECODE(&(ZTOI(zp)->i_ctime), ctime);
if (ZTOZSB(zp)->z_use_fuids)
zp->z_pflags |= ZFS_ARCHIVE;
}
static int
zfs_extend(znode_t *zp, uint64_t end)
{
- zfs_sb_t *zsb = ZTOZSB(zp);
+ zfsvfs_t *zfsvfs = ZTOZSB(zp);
dmu_tx_t *tx;
- rl_t *rl;
+ locked_range_t *lr;
uint64_t newblksz;
int error;
/*
* We will change zp_size, lock the whole file.
*/
- rl = zfs_range_lock(&zp->z_range_lock, 0, UINT64_MAX, RL_WRITER);
+ lr = rangelock_enter(&zp->z_rangelock, 0, UINT64_MAX, RL_WRITER);
/*
* Nothing to do if file already at desired length.
*/
if (end <= zp->z_size) {
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
return (0);
}
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(zfsvfs->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 < zsb->z_max_blksz)) {
+ (!ISP2(zp->z_blksz) || zp->z_blksz < zfsvfs->z_max_blksz)) {
/*
* We are growing the file past the current block size.
*/
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
return (error);
}
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);
+ rangelock_exit(lr);
dmu_tx_commit(tx);
static int
zfs_free_range(znode_t *zp, uint64_t off, uint64_t len)
{
- zfs_sb_t *zsb = ZTOZSB(zp);
- rl_t *rl;
+ zfsvfs_t *zfsvfs = ZTOZSB(zp);
+ locked_range_t *lr;
int error;
/*
* Lock the range being freed.
*/
- rl = zfs_range_lock(&zp->z_range_lock, off, len, RL_WRITER);
+ lr = rangelock_enter(&zp->z_rangelock, off, len, RL_WRITER);
/*
* Nothing to do if file already at desired length.
*/
if (off >= zp->z_size) {
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
return (0);
}
if (off + len > zp->z_size)
len = zp->z_size - off;
- error = dmu_free_long_range(zsb->z_os, zp->z_id, off, len);
+ error = dmu_free_long_range(zfsvfs->z_os, zp->z_id, off, len);
/*
* Zero partial page cache entries. This must be done under a
page_len);
}
}
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
return (error);
}
static int
zfs_trunc(znode_t *zp, uint64_t end)
{
- zfs_sb_t *zsb = ZTOZSB(zp);
+ zfsvfs_t *zfsvfs = ZTOZSB(zp);
dmu_tx_t *tx;
- rl_t *rl;
+ locked_range_t *lr;
int error;
sa_bulk_attr_t bulk[2];
int count = 0;
/*
* We will change zp_size, lock the whole file.
*/
- rl = zfs_range_lock(&zp->z_range_lock, 0, UINT64_MAX, RL_WRITER);
+ lr = rangelock_enter(&zp->z_rangelock, 0, UINT64_MAX, RL_WRITER);
/*
* Nothing to do if file already at desired length.
*/
if (end >= zp->z_size) {
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
return (0);
}
- error = dmu_free_long_range(zsb->z_os, zp->z_id, end, -1);
+ error = dmu_free_long_range(zfsvfs->z_os, zp->z_id, end,
+ DMU_OBJECT_END);
if (error) {
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
return (error);
}
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, zp);
dmu_tx_mark_netfree(tx);
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
return (error);
}
zp->z_size = end;
- SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zsb),
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_SIZE(zfsvfs),
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(zsb),
+ SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs),
NULL, &zp->z_pflags, 8);
}
VERIFY(sa_bulk_update(zp->z_sa_hdl, bulk, count, tx) == 0);
dmu_tx_commit(tx);
-
- zfs_range_unlock(rl);
+ rangelock_exit(lr);
return (0);
}
zfs_freesp(znode_t *zp, uint64_t off, uint64_t len, int flag, boolean_t log)
{
dmu_tx_t *tx;
- zfs_sb_t *zsb = ZTOZSB(zp);
- zilog_t *zilog = zsb->z_log;
+ zfsvfs_t *zfsvfs = ZTOZSB(zp);
+ zilog_t *zilog = zfsvfs->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(zsb), &mode,
+ if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_MODE(zfsvfs), &mode,
sizeof (mode))) != 0)
return (error);
if (error || !log)
goto out;
log:
- tx = dmu_tx_create(zsb->z_os);
+ tx = dmu_tx_create(zfsvfs->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_WAIT);
goto out;
}
- 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),
+ 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),
NULL, &zp->z_pflags, 8);
zfs_tstamp_update_setup(zp, CONTENT_MODIFIED, mtime, ctime);
error = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *zplprops, dmu_tx_t *tx)
{
struct super_block *sb;
- zfs_sb_t *zsb;
+ zfsvfs_t *zfsvfs;
uint64_t moid, obj, sa_obj, version;
uint64_t sense = ZFS_CASE_SENSITIVE;
uint64_t norm = 0;
ASSERT(error == 0);
/*
- * Create root znode. Create minimal znode/inode/zsb/sb
+ * Create root znode. Create minimal znode/inode/zfsvfs/sb
* to allow zfs_mknode to work.
*/
vattr.va_mask = ATTR_MODE|ATTR_UID|ATTR_GID;
rootzp->z_unlinked = 0;
rootzp->z_atime_dirty = 0;
rootzp->z_is_sa = USE_SA(version, os);
+ rootzp->z_pflags = 0;
- zsb = kmem_zalloc(sizeof (zfs_sb_t), KM_SLEEP);
- zsb->z_os = os;
- zsb->z_parent = zsb;
- zsb->z_version = version;
- zsb->z_use_fuids = USE_FUIDS(version, os);
- zsb->z_use_sa = USE_SA(version, os);
- zsb->z_norm = norm;
+ zfsvfs = kmem_zalloc(sizeof (zfsvfs_t), KM_SLEEP);
+ 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;
sb = kmem_zalloc(sizeof (struct super_block), KM_SLEEP);
- sb->s_fs_info = zsb;
+ sb->s_fs_info = zfsvfs;
ZTOI(rootzp)->i_sb = sb;
error = sa_setup(os, sa_obj, zfs_attr_table, ZPL_END,
- &zsb->z_attr_table);
+ &zfsvfs->z_attr_table);
ASSERT(error == 0);
* insensitive.
*/
if (sense == ZFS_CASE_INSENSITIVE || sense == ZFS_CASE_MIXED)
- zsb->z_norm |= U8_TEXTPREP_TOUPPER;
+ zfsvfs->z_norm |= U8_TEXTPREP_TOUPPER;
- mutex_init(&zsb->z_znodes_lock, NULL, MUTEX_DEFAULT, NULL);
- list_create(&zsb->z_all_znodes, sizeof (znode_t),
+ 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));
size = MIN(1 << (highbit64(zfs_object_mutex_size)-1), ZFS_OBJ_MTX_MAX);
- zsb->z_hold_size = size;
- zsb->z_hold_trees = vmem_zalloc(sizeof (avl_tree_t) * size, KM_SLEEP);
- zsb->z_hold_locks = vmem_zalloc(sizeof (kmutex_t) * size, KM_SLEEP);
+ zfsvfs->z_hold_size = size;
+ zfsvfs->z_hold_trees = vmem_zalloc(sizeof (avl_tree_t) * size,
+ KM_SLEEP);
+ zfsvfs->z_hold_locks = vmem_zalloc(sizeof (kmutex_t) * size, KM_SLEEP);
for (i = 0; i != size; i++) {
- avl_create(&zsb->z_hold_trees[i], zfs_znode_hold_compare,
+ avl_create(&zfsvfs->z_hold_trees[i], zfs_znode_hold_compare,
sizeof (znode_hold_t), offsetof(znode_hold_t, zh_node));
- mutex_init(&zsb->z_hold_locks[i], NULL, MUTEX_DEFAULT, NULL);
+ mutex_init(&zfsvfs->z_hold_locks[i], NULL, MUTEX_DEFAULT, NULL);
}
VERIFY(0 == zfs_acl_ids_create(rootzp, IS_ROOT_NODE, &vattr,
sa_handle_destroy(rootzp->z_sa_hdl);
kmem_cache_free(znode_cache, rootzp);
- /*
- * Create shares directory
- */
- error = zfs_create_share_dir(zsb, tx);
- ASSERT(error == 0);
-
for (i = 0; i != size; i++) {
- avl_destroy(&zsb->z_hold_trees[i]);
- mutex_destroy(&zsb->z_hold_locks[i]);
+ avl_destroy(&zfsvfs->z_hold_trees[i]);
+ mutex_destroy(&zfsvfs->z_hold_locks[i]);
}
- vmem_free(zsb->z_hold_trees, sizeof (avl_tree_t) * size);
- vmem_free(zsb->z_hold_locks, sizeof (kmutex_t) * size);
+ mutex_destroy(&zfsvfs->z_znodes_lock);
+
+ vmem_free(zfsvfs->z_hold_trees, sizeof (avl_tree_t) * size);
+ vmem_free(zfsvfs->z_hold_locks, sizeof (kmutex_t) * size);
kmem_free(sb, sizeof (struct super_block));
- kmem_free(zsb, sizeof (zfs_sb_t));
+ kmem_free(zfsvfs, sizeof (zfsvfs_t));
}
#endif /* _KERNEL */
* Otherwise the parent must be a directory.
*/
if (!*is_xattrdir && !S_ISDIR(parent_mode))
- return (EINVAL);
+ return (SET_ERROR(EINVAL));
*pobjp = parent;
*path = '\0';
sa_hdl = hdl;
+ uint64_t deleteq_obj;
+ VERIFY0(zap_lookup(osp, MASTER_NODE_OBJ,
+ ZFS_UNLINKED_SET, sizeof (uint64_t), 1, &deleteq_obj));
+ error = zap_lookup_int(osp, deleteq_obj, obj);
+ if (error == 0) {
+ return (ESTALE);
+ } else if (error != ENOENT) {
+ return (error);
+ }
+ error = 0;
+
for (;;) {
uint64_t pobj = 0;
char component[MAXNAMELEN + 2];
return (error);
}
-#if defined(_KERNEL) && defined(HAVE_SPL)
+#if defined(_KERNEL)
EXPORT_SYMBOL(zfs_create_fs);
EXPORT_SYMBOL(zfs_obj_to_path);
+/* CSTYLED */
module_param(zfs_object_mutex_size, uint, 0644);
MODULE_PARM_DESC(zfs_object_mutex_size, "Size of znode hold array");
+module_param(zfs_unlink_suspend_progress, int, 0644);
+MODULE_PARM_DESC(zfs_unlink_suspend_progress, "Set to prevent async unlinks "
+"(debug - leaks space into the unlinked set)");
#endif