diff --git a/config/kernel-inode-times.m4 b/config/kernel-inode-times.m4
index aae95abf1720..4d861596ed0b 100644
--- a/config/kernel-inode-times.m4
+++ b/config/kernel-inode-times.m4
@@ -1,93 +1,171 @@
AC_DEFUN([ZFS_AC_KERNEL_SRC_INODE_TIMES], [
dnl #
dnl # 5.6 API change
dnl # timespec64_trunc() replaced by timestamp_truncate() interface.
dnl #
ZFS_LINUX_TEST_SRC([timestamp_truncate], [
#include <linux/fs.h>
],[
struct timespec64 ts;
struct inode ip;
memset(&ts, 0, sizeof(ts));
ts = timestamp_truncate(ts, &ip);
])
dnl #
dnl # 4.18 API change
dnl # i_atime, i_mtime, and i_ctime changed from timespec to timespec64.
dnl #
ZFS_LINUX_TEST_SRC([inode_times], [
#include <linux/fs.h>
],[
struct inode ip;
struct timespec ts;
memset(&ip, 0, sizeof(ip));
ts = ip.i_mtime;
])
dnl #
dnl # 6.6 API change
dnl # i_ctime no longer directly accessible, must use
dnl # inode_get_ctime(ip), inode_set_ctime*(ip) to
dnl # read/write.
dnl #
ZFS_LINUX_TEST_SRC([inode_get_ctime], [
#include <linux/fs.h>
],[
struct inode ip;
memset(&ip, 0, sizeof(ip));
inode_get_ctime(&ip);
])
ZFS_LINUX_TEST_SRC([inode_set_ctime_to_ts], [
#include <linux/fs.h>
],[
struct inode ip;
struct timespec64 ts = {0};
memset(&ip, 0, sizeof(ip));
inode_set_ctime_to_ts(&ip, ts);
])
+
+ dnl #
+ dnl # 6.7 API change
+ dnl # i_atime/i_mtime no longer directly accessible, must use
+ dnl # inode_get_mtime(ip), inode_set_mtime*(ip) to
+ dnl # read/write.
+ dnl #
+ ZFS_LINUX_TEST_SRC([inode_get_atime], [
+ #include <linux/fs.h>
+ ],[
+ struct inode ip;
+
+ memset(&ip, 0, sizeof(ip));
+ inode_get_atime(&ip);
+ ])
+ ZFS_LINUX_TEST_SRC([inode_get_mtime], [
+ #include <linux/fs.h>
+ ],[
+ struct inode ip;
+
+ memset(&ip, 0, sizeof(ip));
+ inode_get_mtime(&ip);
+ ])
+
+ ZFS_LINUX_TEST_SRC([inode_set_atime_to_ts], [
+ #include <linux/fs.h>
+ ],[
+ struct inode ip;
+ struct timespec64 ts = {0};
+
+ memset(&ip, 0, sizeof(ip));
+ inode_set_atime_to_ts(&ip, ts);
+ ])
+ ZFS_LINUX_TEST_SRC([inode_set_mtime_to_ts], [
+ #include <linux/fs.h>
+ ],[
+ struct inode ip;
+ struct timespec64 ts = {0};
+
+ memset(&ip, 0, sizeof(ip));
+ inode_set_mtime_to_ts(&ip, ts);
+ ])
])
AC_DEFUN([ZFS_AC_KERNEL_INODE_TIMES], [
AC_MSG_CHECKING([whether timestamp_truncate() exists])
ZFS_LINUX_TEST_RESULT([timestamp_truncate], [
AC_MSG_RESULT(yes)
AC_DEFINE(HAVE_INODE_TIMESTAMP_TRUNCATE, 1,
[timestamp_truncate() exists])
],[
AC_MSG_RESULT(no)
])
AC_MSG_CHECKING([whether inode->i_*time's are timespec64])
ZFS_LINUX_TEST_RESULT([inode_times], [
AC_MSG_RESULT(no)
],[
AC_MSG_RESULT(yes)
AC_DEFINE(HAVE_INODE_TIMESPEC64_TIMES, 1,
[inode->i_*time's are timespec64])
])
AC_MSG_CHECKING([whether inode_get_ctime() exists])
ZFS_LINUX_TEST_RESULT([inode_get_ctime], [
AC_MSG_RESULT(yes)
AC_DEFINE(HAVE_INODE_GET_CTIME, 1,
[inode_get_ctime() exists in linux/fs.h])
],[
AC_MSG_RESULT(no)
])
AC_MSG_CHECKING([whether inode_set_ctime_to_ts() exists])
ZFS_LINUX_TEST_RESULT([inode_set_ctime_to_ts], [
AC_MSG_RESULT(yes)
AC_DEFINE(HAVE_INODE_SET_CTIME_TO_TS, 1,
[inode_set_ctime_to_ts() exists in linux/fs.h])
],[
AC_MSG_RESULT(no)
])
+
+ AC_MSG_CHECKING([whether inode_get_atime() exists])
+ ZFS_LINUX_TEST_RESULT([inode_get_atime], [
+ AC_MSG_RESULT(yes)
+ AC_DEFINE(HAVE_INODE_GET_ATIME, 1,
+ [inode_get_atime() exists in linux/fs.h])
+ ],[
+ AC_MSG_RESULT(no)
+ ])
+
+ AC_MSG_CHECKING([whether inode_set_atime_to_ts() exists])
+ ZFS_LINUX_TEST_RESULT([inode_set_atime_to_ts], [
+ AC_MSG_RESULT(yes)
+ AC_DEFINE(HAVE_INODE_SET_ATIME_TO_TS, 1,
+ [inode_set_atime_to_ts() exists in linux/fs.h])
+ ],[
+ AC_MSG_RESULT(no)
+ ])
+
+ AC_MSG_CHECKING([whether inode_get_mtime() exists])
+ ZFS_LINUX_TEST_RESULT([inode_get_mtime], [
+ AC_MSG_RESULT(yes)
+ AC_DEFINE(HAVE_INODE_GET_MTIME, 1,
+ [inode_get_mtime() exists in linux/fs.h])
+ ],[
+ AC_MSG_RESULT(no)
+ ])
+
+ AC_MSG_CHECKING([whether inode_set_mtime_to_ts() exists])
+ ZFS_LINUX_TEST_RESULT([inode_set_mtime_to_ts], [
+ AC_MSG_RESULT(yes)
+ AC_DEFINE(HAVE_INODE_SET_MTIME_TO_TS, 1,
+ [inode_set_mtime_to_ts() exists in linux/fs.h])
+ ],[
+ AC_MSG_RESULT(no)
+ ])
])
diff --git a/include/os/linux/zfs/sys/zpl.h b/include/os/linux/zfs/sys/zpl.h
index bd1462b16c4b..69b6c1185bb1 100644
--- a/include/os/linux/zfs/sys/zpl.h
+++ b/include/os/linux/zfs/sys/zpl.h
@@ -1,221 +1,241 @@
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2011, Lawrence Livermore National Security, LLC.
*/
#ifndef _SYS_ZPL_H
#define _SYS_ZPL_H
#include <sys/mntent.h>
#include <sys/vfs.h>
#include <linux/aio.h>
#include <linux/dcache_compat.h>
#include <linux/exportfs.h>
#include <linux/falloc.h>
#include <linux/parser.h>
#include <linux/task_io_accounting_ops.h>
#include <linux/vfs_compat.h>
#include <linux/writeback.h>
#include <linux/xattr_compat.h>
/* zpl_inode.c */
extern void zpl_vap_init(vattr_t *vap, struct inode *dir,
umode_t mode, cred_t *cr);
extern const struct inode_operations zpl_inode_operations;
extern const struct inode_operations zpl_dir_inode_operations;
extern const struct inode_operations zpl_symlink_inode_operations;
extern const struct inode_operations zpl_special_inode_operations;
/* zpl_file.c */
extern const struct address_space_operations zpl_address_space_operations;
extern const struct file_operations zpl_file_operations;
extern const struct file_operations zpl_dir_file_operations;
/* zpl_super.c */
extern void zpl_prune_sb(int64_t nr_to_scan, void *arg);
extern const struct super_operations zpl_super_operations;
extern const struct export_operations zpl_export_operations;
extern struct file_system_type zpl_fs_type;
/* zpl_xattr.c */
extern ssize_t zpl_xattr_list(struct dentry *dentry, char *buf, size_t size);
extern int zpl_xattr_security_init(struct inode *ip, struct inode *dip,
const struct qstr *qstr);
#if defined(CONFIG_FS_POSIX_ACL)
#if defined(HAVE_SET_ACL)
#if defined(HAVE_SET_ACL_IDMAP_DENTRY)
extern int zpl_set_acl(struct mnt_idmap *idmap, struct dentry *dentry,
struct posix_acl *acl, int type);
#elif defined(HAVE_SET_ACL_USERNS)
extern int zpl_set_acl(struct user_namespace *userns, struct inode *ip,
struct posix_acl *acl, int type);
#elif defined(HAVE_SET_ACL_USERNS_DENTRY_ARG2)
extern int zpl_set_acl(struct user_namespace *userns, struct dentry *dentry,
struct posix_acl *acl, int type);
#else
extern int zpl_set_acl(struct inode *ip, struct posix_acl *acl, int type);
#endif /* HAVE_SET_ACL_USERNS */
#endif /* HAVE_SET_ACL */
#if defined(HAVE_GET_ACL_RCU) || defined(HAVE_GET_INODE_ACL)
extern struct posix_acl *zpl_get_acl(struct inode *ip, int type, bool rcu);
#elif defined(HAVE_GET_ACL)
extern struct posix_acl *zpl_get_acl(struct inode *ip, int type);
#endif
extern int zpl_init_acl(struct inode *ip, struct inode *dir);
extern int zpl_chmod_acl(struct inode *ip);
#else
static inline int
zpl_init_acl(struct inode *ip, struct inode *dir)
{
return (0);
}
static inline int
zpl_chmod_acl(struct inode *ip)
{
return (0);
}
#endif /* CONFIG_FS_POSIX_ACL */
extern xattr_handler_t *zpl_xattr_handlers[];
/* zpl_ctldir.c */
extern const struct file_operations zpl_fops_root;
extern const struct inode_operations zpl_ops_root;
extern const struct file_operations zpl_fops_snapdir;
extern const struct inode_operations zpl_ops_snapdir;
extern const struct dentry_operations zpl_dops_snapdirs;
extern const struct file_operations zpl_fops_shares;
extern const struct inode_operations zpl_ops_shares;
#if defined(HAVE_VFS_ITERATE) || defined(HAVE_VFS_ITERATE_SHARED)
#define ZPL_DIR_CONTEXT_INIT(_dirent, _actor, _pos) { \
.actor = _actor, \
.pos = _pos, \
}
typedef struct dir_context zpl_dir_context_t;
#define zpl_dir_emit dir_emit
#define zpl_dir_emit_dot dir_emit_dot
#define zpl_dir_emit_dotdot dir_emit_dotdot
#define zpl_dir_emit_dots dir_emit_dots
#else
typedef struct zpl_dir_context {
void *dirent;
const filldir_t actor;
loff_t pos;
} zpl_dir_context_t;
#define ZPL_DIR_CONTEXT_INIT(_dirent, _actor, _pos) { \
.dirent = _dirent, \
.actor = _actor, \
.pos = _pos, \
}
static inline bool
zpl_dir_emit(zpl_dir_context_t *ctx, const char *name, int namelen,
uint64_t ino, unsigned type)
{
return (!ctx->actor(ctx->dirent, name, namelen, ctx->pos, ino, type));
}
static inline bool
zpl_dir_emit_dot(struct file *file, zpl_dir_context_t *ctx)
{
return (ctx->actor(ctx->dirent, ".", 1, ctx->pos,
file_inode(file)->i_ino, DT_DIR) == 0);
}
static inline bool
zpl_dir_emit_dotdot(struct file *file, zpl_dir_context_t *ctx)
{
return (ctx->actor(ctx->dirent, "..", 2, ctx->pos,
parent_ino(file_dentry(file)), DT_DIR) == 0);
}
static inline bool
zpl_dir_emit_dots(struct file *file, zpl_dir_context_t *ctx)
{
if (ctx->pos == 0) {
if (!zpl_dir_emit_dot(file, ctx))
return (false);
ctx->pos = 1;
}
if (ctx->pos == 1) {
if (!zpl_dir_emit_dotdot(file, ctx))
return (false);
ctx->pos = 2;
}
return (true);
}
#endif /* HAVE_VFS_ITERATE */
#if defined(HAVE_INODE_TIMESTAMP_TRUNCATE)
#define zpl_inode_timestamp_truncate(ts, ip) timestamp_truncate(ts, ip)
#elif defined(HAVE_INODE_TIMESPEC64_TIMES)
#define zpl_inode_timestamp_truncate(ts, ip) \
timespec64_trunc(ts, (ip)->i_sb->s_time_gran)
#else
#define zpl_inode_timestamp_truncate(ts, ip) \
timespec_trunc(ts, (ip)->i_sb->s_time_gran)
#endif
#if defined(HAVE_INODE_OWNER_OR_CAPABLE)
#define zpl_inode_owner_or_capable(ns, ip) inode_owner_or_capable(ip)
#elif defined(HAVE_INODE_OWNER_OR_CAPABLE_USERNS)
#define zpl_inode_owner_or_capable(ns, ip) inode_owner_or_capable(ns, ip)
#elif defined(HAVE_INODE_OWNER_OR_CAPABLE_IDMAP)
#define zpl_inode_owner_or_capable(idmap, ip) inode_owner_or_capable(idmap, ip)
#else
#error "Unsupported kernel"
#endif
#if defined(HAVE_SETATTR_PREPARE_USERNS) || defined(HAVE_SETATTR_PREPARE_IDMAP)
#define zpl_setattr_prepare(ns, dentry, ia) setattr_prepare(ns, dentry, ia)
#else
/*
* Use kernel-provided version, or our own from
* linux/vfs_compat.h
*/
#define zpl_setattr_prepare(ns, dentry, ia) setattr_prepare(dentry, ia)
#endif
#ifdef HAVE_INODE_GET_CTIME
#define zpl_inode_get_ctime(ip) inode_get_ctime(ip)
#else
#define zpl_inode_get_ctime(ip) (ip->i_ctime)
#endif
#ifdef HAVE_INODE_SET_CTIME_TO_TS
#define zpl_inode_set_ctime_to_ts(ip, ts) inode_set_ctime_to_ts(ip, ts)
#else
#define zpl_inode_set_ctime_to_ts(ip, ts) (ip->i_ctime = ts)
#endif
+#ifdef HAVE_INODE_GET_ATIME
+#define zpl_inode_get_atime(ip) inode_get_atime(ip)
+#else
+#define zpl_inode_get_atime(ip) (ip->i_atime)
+#endif
+#ifdef HAVE_INODE_SET_ATIME_TO_TS
+#define zpl_inode_set_atime_to_ts(ip, ts) inode_set_atime_to_ts(ip, ts)
+#else
+#define zpl_inode_set_atime_to_ts(ip, ts) (ip->i_atime = ts)
+#endif
+#ifdef HAVE_INODE_GET_MTIME
+#define zpl_inode_get_mtime(ip) inode_get_mtime(ip)
+#else
+#define zpl_inode_get_mtime(ip) (ip->i_mtime)
+#endif
+#ifdef HAVE_INODE_SET_MTIME_TO_TS
+#define zpl_inode_set_mtime_to_ts(ip, ts) inode_set_mtime_to_ts(ip, ts)
+#else
+#define zpl_inode_set_mtime_to_ts(ip, ts) (ip->i_mtime = ts)
+#endif
#endif /* _SYS_ZPL_H */
diff --git a/module/os/linux/zfs/zfs_ctldir.c b/module/os/linux/zfs/zfs_ctldir.c
index 3ce18bbfacc0..b51df6a2ebe5 100644
--- a/module/os/linux/zfs/zfs_ctldir.c
+++ b/module/os/linux/zfs/zfs_ctldir.c
@@ -1,1294 +1,1294 @@
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (C) 2011 Lawrence Livermore National Security, LLC.
* Produced at Lawrence Livermore National Laboratory (cf, DISCLAIMER).
* LLNL-CODE-403049.
* Rewritten for Linux by:
* Rohan Puri <rohan.puri15@gmail.com>
* Brian Behlendorf <behlendorf1@llnl.gov>
* Copyright (c) 2013 by Delphix. All rights reserved.
* Copyright 2015, OmniTI Computer Consulting, Inc. All rights reserved.
* Copyright (c) 2018 George Melikov. All Rights Reserved.
* Copyright (c) 2019 Datto, Inc. All rights reserved.
* Copyright (c) 2020 The MathWorks, Inc. All rights reserved.
*/
/*
* ZFS control directory (a.k.a. ".zfs")
*
* This directory provides a common location for all ZFS meta-objects.
* Currently, this is only the 'snapshot' and 'shares' directory, but this may
* expand in the future. The elements are built dynamically, as the hierarchy
* does not actually exist on disk.
*
* For 'snapshot', we don't want to have all snapshots always mounted, because
* this would take up a huge amount of space in /etc/mnttab. We have three
* types of objects:
*
* ctldir ------> snapshotdir -------> snapshot
* |
* |
* V
* mounted fs
*
* The 'snapshot' node contains just enough information to lookup '..' and act
* as a mountpoint for the snapshot. Whenever we lookup a specific snapshot, we
* perform an automount of the underlying filesystem and return the
* corresponding inode.
*
* All mounts are handled automatically by an user mode helper which invokes
* the mount procedure. Unmounts are handled by allowing the mount
* point to expire so the kernel may automatically unmount it.
*
* The '.zfs', '.zfs/snapshot', and all directories created under
* '.zfs/snapshot' (ie: '.zfs/snapshot/<snapname>') all share the same
* zfsvfs_t as the head filesystem (what '.zfs' lives under).
*
* File systems mounted on top of the '.zfs/snapshot/<snapname>' paths
* (ie: snapshots) are complete ZFS filesystems and have their own unique
* zfsvfs_t. However, the fsid reported by these mounts will be the same
* as that used by the parent zfsvfs_t to make NFS happy.
*/
#include <sys/types.h>
#include <sys/param.h>
#include <sys/time.h>
#include <sys/sysmacros.h>
#include <sys/pathname.h>
#include <sys/vfs.h>
#include <sys/zfs_ctldir.h>
#include <sys/zfs_ioctl.h>
#include <sys/zfs_vfsops.h>
#include <sys/zfs_vnops.h>
#include <sys/stat.h>
#include <sys/dmu.h>
#include <sys/dmu_objset.h>
#include <sys/dsl_destroy.h>
#include <sys/dsl_deleg.h>
#include <sys/zpl.h>
#include <sys/mntent.h>
#include "zfs_namecheck.h"
/*
* Two AVL trees are maintained which contain all currently automounted
* snapshots. Every automounted snapshots maps to a single zfs_snapentry_t
* entry which MUST:
*
* - be attached to both trees, and
* - be unique, no duplicate entries are allowed.
*
* The zfs_snapshots_by_name tree is indexed by the full dataset name
* while the zfs_snapshots_by_objsetid tree is indexed by the unique
* objsetid. This allows for fast lookups either by name or objsetid.
*/
static avl_tree_t zfs_snapshots_by_name;
static avl_tree_t zfs_snapshots_by_objsetid;
static krwlock_t zfs_snapshot_lock;
/*
* Control Directory Tunables (.zfs)
*/
int zfs_expire_snapshot = ZFSCTL_EXPIRE_SNAPSHOT;
int zfs_admin_snapshot = 0;
typedef struct {
char *se_name; /* full snapshot name */
char *se_path; /* full mount path */
spa_t *se_spa; /* pool spa */
uint64_t se_objsetid; /* snapshot objset id */
struct dentry *se_root_dentry; /* snapshot root dentry */
krwlock_t se_taskqid_lock; /* scheduled unmount taskqid lock */
taskqid_t se_taskqid; /* scheduled unmount taskqid */
avl_node_t se_node_name; /* zfs_snapshots_by_name link */
avl_node_t se_node_objsetid; /* zfs_snapshots_by_objsetid link */
zfs_refcount_t se_refcount; /* reference count */
} zfs_snapentry_t;
static void zfsctl_snapshot_unmount_delay_impl(zfs_snapentry_t *se, int delay);
/*
* Allocate a new zfs_snapentry_t being careful to make a copy of the
* the snapshot name and provided mount point. No reference is taken.
*/
static zfs_snapentry_t *
zfsctl_snapshot_alloc(const char *full_name, const char *full_path, spa_t *spa,
uint64_t objsetid, struct dentry *root_dentry)
{
zfs_snapentry_t *se;
se = kmem_zalloc(sizeof (zfs_snapentry_t), KM_SLEEP);
se->se_name = kmem_strdup(full_name);
se->se_path = kmem_strdup(full_path);
se->se_spa = spa;
se->se_objsetid = objsetid;
se->se_root_dentry = root_dentry;
se->se_taskqid = TASKQID_INVALID;
rw_init(&se->se_taskqid_lock, NULL, RW_DEFAULT, NULL);
zfs_refcount_create(&se->se_refcount);
return (se);
}
/*
* Free a zfs_snapentry_t the caller must ensure there are no active
* references.
*/
static void
zfsctl_snapshot_free(zfs_snapentry_t *se)
{
zfs_refcount_destroy(&se->se_refcount);
kmem_strfree(se->se_name);
kmem_strfree(se->se_path);
rw_destroy(se->se_taskqid_lock);
kmem_free(se, sizeof (zfs_snapentry_t));
}
/*
* Hold a reference on the zfs_snapentry_t.
*/
static void
zfsctl_snapshot_hold(zfs_snapentry_t *se)
{
zfs_refcount_add(&se->se_refcount, NULL);
}
/*
* Release a reference on the zfs_snapentry_t. When the number of
* references drops to zero the structure will be freed.
*/
static void
zfsctl_snapshot_rele(zfs_snapentry_t *se)
{
if (zfs_refcount_remove(&se->se_refcount, NULL) == 0)
zfsctl_snapshot_free(se);
}
/*
* Add a zfs_snapentry_t to both the zfs_snapshots_by_name and
* zfs_snapshots_by_objsetid trees. While the zfs_snapentry_t is part
* of the trees a reference is held.
*/
static void
zfsctl_snapshot_add(zfs_snapentry_t *se)
{
ASSERT(RW_WRITE_HELD(&zfs_snapshot_lock));
zfsctl_snapshot_hold(se);
avl_add(&zfs_snapshots_by_name, se);
avl_add(&zfs_snapshots_by_objsetid, se);
}
/*
* Remove a zfs_snapentry_t from both the zfs_snapshots_by_name and
* zfs_snapshots_by_objsetid trees. Upon removal a reference is dropped,
* this can result in the structure being freed if that was the last
* remaining reference.
*/
static void
zfsctl_snapshot_remove(zfs_snapentry_t *se)
{
ASSERT(RW_WRITE_HELD(&zfs_snapshot_lock));
avl_remove(&zfs_snapshots_by_name, se);
avl_remove(&zfs_snapshots_by_objsetid, se);
zfsctl_snapshot_rele(se);
}
/*
* Snapshot name comparison function for the zfs_snapshots_by_name.
*/
static int
snapentry_compare_by_name(const void *a, const void *b)
{
const zfs_snapentry_t *se_a = a;
const zfs_snapentry_t *se_b = b;
int ret;
ret = strcmp(se_a->se_name, se_b->se_name);
if (ret < 0)
return (-1);
else if (ret > 0)
return (1);
else
return (0);
}
/*
* Snapshot name comparison function for the zfs_snapshots_by_objsetid.
*/
static int
snapentry_compare_by_objsetid(const void *a, const void *b)
{
const zfs_snapentry_t *se_a = a;
const zfs_snapentry_t *se_b = b;
if (se_a->se_spa != se_b->se_spa)
return ((ulong_t)se_a->se_spa < (ulong_t)se_b->se_spa ? -1 : 1);
if (se_a->se_objsetid < se_b->se_objsetid)
return (-1);
else if (se_a->se_objsetid > se_b->se_objsetid)
return (1);
else
return (0);
}
/*
* Find a zfs_snapentry_t in zfs_snapshots_by_name. If the snapname
* is found a pointer to the zfs_snapentry_t is returned and a reference
* taken on the structure. The caller is responsible for dropping the
* reference with zfsctl_snapshot_rele(). If the snapname is not found
* NULL will be returned.
*/
static zfs_snapentry_t *
zfsctl_snapshot_find_by_name(const char *snapname)
{
zfs_snapentry_t *se, search;
ASSERT(RW_LOCK_HELD(&zfs_snapshot_lock));
search.se_name = (char *)snapname;
se = avl_find(&zfs_snapshots_by_name, &search, NULL);
if (se)
zfsctl_snapshot_hold(se);
return (se);
}
/*
* Find a zfs_snapentry_t in zfs_snapshots_by_objsetid given the objset id
* rather than the snapname. In all other respects it behaves the same
* as zfsctl_snapshot_find_by_name().
*/
static zfs_snapentry_t *
zfsctl_snapshot_find_by_objsetid(spa_t *spa, uint64_t objsetid)
{
zfs_snapentry_t *se, search;
ASSERT(RW_LOCK_HELD(&zfs_snapshot_lock));
search.se_spa = spa;
search.se_objsetid = objsetid;
se = avl_find(&zfs_snapshots_by_objsetid, &search, NULL);
if (se)
zfsctl_snapshot_hold(se);
return (se);
}
/*
* Rename a zfs_snapentry_t in the zfs_snapshots_by_name. The structure is
* removed, renamed, and added back to the new correct location in the tree.
*/
static int
zfsctl_snapshot_rename(const char *old_snapname, const char *new_snapname)
{
zfs_snapentry_t *se;
ASSERT(RW_WRITE_HELD(&zfs_snapshot_lock));
se = zfsctl_snapshot_find_by_name(old_snapname);
if (se == NULL)
return (SET_ERROR(ENOENT));
zfsctl_snapshot_remove(se);
kmem_strfree(se->se_name);
se->se_name = kmem_strdup(new_snapname);
zfsctl_snapshot_add(se);
zfsctl_snapshot_rele(se);
return (0);
}
/*
* Delayed task responsible for unmounting an expired automounted snapshot.
*/
static void
snapentry_expire(void *data)
{
zfs_snapentry_t *se = (zfs_snapentry_t *)data;
spa_t *spa = se->se_spa;
uint64_t objsetid = se->se_objsetid;
if (zfs_expire_snapshot <= 0) {
zfsctl_snapshot_rele(se);
return;
}
rw_enter(&se->se_taskqid_lock, RW_WRITER);
se->se_taskqid = TASKQID_INVALID;
rw_exit(&se->se_taskqid_lock);
(void) zfsctl_snapshot_unmount(se->se_name, MNT_EXPIRE);
zfsctl_snapshot_rele(se);
/*
* Reschedule the unmount if the zfs_snapentry_t wasn't removed.
* This can occur when the snapshot is busy.
*/
rw_enter(&zfs_snapshot_lock, RW_READER);
if ((se = zfsctl_snapshot_find_by_objsetid(spa, objsetid)) != NULL) {
zfsctl_snapshot_unmount_delay_impl(se, zfs_expire_snapshot);
zfsctl_snapshot_rele(se);
}
rw_exit(&zfs_snapshot_lock);
}
/*
* Cancel an automatic unmount of a snapname. This callback is responsible
* for dropping the reference on the zfs_snapentry_t which was taken when
* during dispatch.
*/
static void
zfsctl_snapshot_unmount_cancel(zfs_snapentry_t *se)
{
int err = 0;
rw_enter(&se->se_taskqid_lock, RW_WRITER);
err = taskq_cancel_id(system_delay_taskq, se->se_taskqid);
/*
* if we get ENOENT, the taskq couldn't be found to be
* canceled, so we can just mark it as invalid because
* it's already gone. If we got EBUSY, then we already
* blocked until it was gone _anyway_, so we don't care.
*/
se->se_taskqid = TASKQID_INVALID;
rw_exit(&se->se_taskqid_lock);
if (err == 0) {
zfsctl_snapshot_rele(se);
}
}
/*
* Dispatch the unmount task for delayed handling with a hold protecting it.
*/
static void
zfsctl_snapshot_unmount_delay_impl(zfs_snapentry_t *se, int delay)
{
if (delay <= 0)
return;
zfsctl_snapshot_hold(se);
rw_enter(&se->se_taskqid_lock, RW_WRITER);
/*
* If this condition happens, we managed to:
* - dispatch once
* - want to dispatch _again_ before it returned
*
* So let's just return - if that task fails at unmounting,
* we'll eventually dispatch again, and if it succeeds,
* no problem.
*/
if (se->se_taskqid != TASKQID_INVALID) {
rw_exit(&se->se_taskqid_lock);
zfsctl_snapshot_rele(se);
return;
}
se->se_taskqid = taskq_dispatch_delay(system_delay_taskq,
snapentry_expire, se, TQ_SLEEP, ddi_get_lbolt() + delay * HZ);
rw_exit(&se->se_taskqid_lock);
}
/*
* Schedule an automatic unmount of objset id to occur in delay seconds from
* now. Any previous delayed unmount will be cancelled in favor of the
* updated deadline. A reference is taken by zfsctl_snapshot_find_by_name()
* and held until the outstanding task is handled or cancelled.
*/
int
zfsctl_snapshot_unmount_delay(spa_t *spa, uint64_t objsetid, int delay)
{
zfs_snapentry_t *se;
int error = ENOENT;
rw_enter(&zfs_snapshot_lock, RW_READER);
if ((se = zfsctl_snapshot_find_by_objsetid(spa, objsetid)) != NULL) {
zfsctl_snapshot_unmount_cancel(se);
zfsctl_snapshot_unmount_delay_impl(se, delay);
zfsctl_snapshot_rele(se);
error = 0;
}
rw_exit(&zfs_snapshot_lock);
return (error);
}
/*
* Check if snapname is currently mounted. Returned non-zero when mounted
* and zero when unmounted.
*/
static boolean_t
zfsctl_snapshot_ismounted(const char *snapname)
{
zfs_snapentry_t *se;
boolean_t ismounted = B_FALSE;
rw_enter(&zfs_snapshot_lock, RW_READER);
if ((se = zfsctl_snapshot_find_by_name(snapname)) != NULL) {
zfsctl_snapshot_rele(se);
ismounted = B_TRUE;
}
rw_exit(&zfs_snapshot_lock);
return (ismounted);
}
/*
* Check if the given inode is a part of the virtual .zfs directory.
*/
boolean_t
zfsctl_is_node(struct inode *ip)
{
return (ITOZ(ip)->z_is_ctldir);
}
/*
* Check if the given inode is a .zfs/snapshots/snapname directory.
*/
boolean_t
zfsctl_is_snapdir(struct inode *ip)
{
return (zfsctl_is_node(ip) && (ip->i_ino <= ZFSCTL_INO_SNAPDIRS));
}
/*
* Allocate a new inode with the passed id and ops.
*/
static struct inode *
zfsctl_inode_alloc(zfsvfs_t *zfsvfs, uint64_t id,
const struct file_operations *fops, const struct inode_operations *ops)
{
inode_timespec_t now;
struct inode *ip;
znode_t *zp;
ip = new_inode(zfsvfs->z_sb);
if (ip == NULL)
return (NULL);
now = current_time(ip);
zp = ITOZ(ip);
ASSERT3P(zp->z_dirlocks, ==, NULL);
ASSERT3P(zp->z_acl_cached, ==, NULL);
ASSERT3P(zp->z_xattr_cached, ==, NULL);
zp->z_id = id;
zp->z_unlinked = B_FALSE;
zp->z_atime_dirty = B_FALSE;
zp->z_zn_prefetch = B_FALSE;
zp->z_is_sa = B_FALSE;
#if !defined(HAVE_FILEMAP_RANGE_HAS_PAGE)
zp->z_is_mapped = B_FALSE;
#endif
zp->z_is_ctldir = B_TRUE;
zp->z_sa_hdl = NULL;
zp->z_blksz = 0;
zp->z_seq = 0;
zp->z_mapcnt = 0;
zp->z_size = 0;
zp->z_pflags = 0;
zp->z_mode = 0;
zp->z_sync_cnt = 0;
zp->z_sync_writes_cnt = 0;
zp->z_async_writes_cnt = 0;
ip->i_generation = 0;
ip->i_ino = id;
ip->i_mode = (S_IFDIR | S_IRWXUGO);
ip->i_uid = SUID_TO_KUID(0);
ip->i_gid = SGID_TO_KGID(0);
ip->i_blkbits = SPA_MINBLOCKSHIFT;
- ip->i_atime = now;
- ip->i_mtime = now;
+ zpl_inode_set_atime_to_ts(ip, now);
+ zpl_inode_set_mtime_to_ts(ip, now);
zpl_inode_set_ctime_to_ts(ip, now);
ip->i_fop = fops;
ip->i_op = ops;
#if defined(IOP_XATTR)
ip->i_opflags &= ~IOP_XATTR;
#endif
if (insert_inode_locked(ip)) {
unlock_new_inode(ip);
iput(ip);
return (NULL);
}
mutex_enter(&zfsvfs->z_znodes_lock);
list_insert_tail(&zfsvfs->z_all_znodes, zp);
zfsvfs->z_nr_znodes++;
membar_producer();
mutex_exit(&zfsvfs->z_znodes_lock);
unlock_new_inode(ip);
return (ip);
}
/*
* Lookup the inode with given id, it will be allocated if needed.
*/
static struct inode *
zfsctl_inode_lookup(zfsvfs_t *zfsvfs, uint64_t id,
const struct file_operations *fops, const struct inode_operations *ops)
{
struct inode *ip = NULL;
while (ip == NULL) {
ip = ilookup(zfsvfs->z_sb, (unsigned long)id);
if (ip)
break;
/* May fail due to concurrent zfsctl_inode_alloc() */
ip = zfsctl_inode_alloc(zfsvfs, id, fops, ops);
}
return (ip);
}
/*
* Create the '.zfs' directory. This directory is cached as part of the VFS
* structure. This results in a hold on the zfsvfs_t. The code in zfs_umount()
* therefore checks against a vfs_count of 2 instead of 1. This reference
* is removed when the ctldir is destroyed in the unmount. All other entities
* under the '.zfs' directory are created dynamically as needed.
*
* Because the dynamically created '.zfs' directory entries assume the use
* of 64-bit inode numbers this support must be disabled on 32-bit systems.
*/
int
zfsctl_create(zfsvfs_t *zfsvfs)
{
ASSERT(zfsvfs->z_ctldir == NULL);
zfsvfs->z_ctldir = zfsctl_inode_alloc(zfsvfs, ZFSCTL_INO_ROOT,
&zpl_fops_root, &zpl_ops_root);
if (zfsvfs->z_ctldir == NULL)
return (SET_ERROR(ENOENT));
return (0);
}
/*
* Destroy the '.zfs' directory or remove a snapshot from zfs_snapshots_by_name.
* Only called when the filesystem is unmounted.
*/
void
zfsctl_destroy(zfsvfs_t *zfsvfs)
{
if (zfsvfs->z_issnap) {
zfs_snapentry_t *se;
spa_t *spa = zfsvfs->z_os->os_spa;
uint64_t objsetid = dmu_objset_id(zfsvfs->z_os);
rw_enter(&zfs_snapshot_lock, RW_WRITER);
se = zfsctl_snapshot_find_by_objsetid(spa, objsetid);
if (se != NULL)
zfsctl_snapshot_remove(se);
rw_exit(&zfs_snapshot_lock);
if (se != NULL) {
zfsctl_snapshot_unmount_cancel(se);
zfsctl_snapshot_rele(se);
}
} else if (zfsvfs->z_ctldir) {
iput(zfsvfs->z_ctldir);
zfsvfs->z_ctldir = NULL;
}
}
/*
* Given a root znode, retrieve the associated .zfs directory.
* Add a hold to the vnode and return it.
*/
struct inode *
zfsctl_root(znode_t *zp)
{
ASSERT(zfs_has_ctldir(zp));
/* Must have an existing ref, so igrab() cannot return NULL */
VERIFY3P(igrab(ZTOZSB(zp)->z_ctldir), !=, NULL);
return (ZTOZSB(zp)->z_ctldir);
}
/*
* Generate a long fid to indicate a snapdir. We encode whether snapdir is
* already mounted in gen field. We do this because nfsd lookup will not
* trigger automount. Next time the nfsd does fh_to_dentry, we will notice
* this and do automount and return ESTALE to force nfsd revalidate and follow
* mount.
*/
static int
zfsctl_snapdir_fid(struct inode *ip, fid_t *fidp)
{
zfid_short_t *zfid = (zfid_short_t *)fidp;
zfid_long_t *zlfid = (zfid_long_t *)fidp;
uint32_t gen = 0;
uint64_t object;
uint64_t objsetid;
int i;
struct dentry *dentry;
if (fidp->fid_len < LONG_FID_LEN) {
fidp->fid_len = LONG_FID_LEN;
return (SET_ERROR(ENOSPC));
}
object = ip->i_ino;
objsetid = ZFSCTL_INO_SNAPDIRS - ip->i_ino;
zfid->zf_len = LONG_FID_LEN;
dentry = d_obtain_alias(igrab(ip));
if (!IS_ERR(dentry)) {
gen = !!d_mountpoint(dentry);
dput(dentry);
}
for (i = 0; i < sizeof (zfid->zf_object); i++)
zfid->zf_object[i] = (uint8_t)(object >> (8 * i));
for (i = 0; i < sizeof (zfid->zf_gen); i++)
zfid->zf_gen[i] = (uint8_t)(gen >> (8 * i));
for (i = 0; i < sizeof (zlfid->zf_setid); i++)
zlfid->zf_setid[i] = (uint8_t)(objsetid >> (8 * i));
for (i = 0; i < sizeof (zlfid->zf_setgen); i++)
zlfid->zf_setgen[i] = 0;
return (0);
}
/*
* Generate an appropriate fid for an entry in the .zfs directory.
*/
int
zfsctl_fid(struct inode *ip, fid_t *fidp)
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
uint64_t object = zp->z_id;
zfid_short_t *zfid;
int i;
ZFS_ENTER(zfsvfs);
if (zfsctl_is_snapdir(ip)) {
ZFS_EXIT(zfsvfs);
return (zfsctl_snapdir_fid(ip, fidp));
}
if (fidp->fid_len < SHORT_FID_LEN) {
fidp->fid_len = SHORT_FID_LEN;
ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENOSPC));
}
zfid = (zfid_short_t *)fidp;
zfid->zf_len = SHORT_FID_LEN;
for (i = 0; i < sizeof (zfid->zf_object); i++)
zfid->zf_object[i] = (uint8_t)(object >> (8 * i));
/* .zfs znodes always have a generation number of 0 */
for (i = 0; i < sizeof (zfid->zf_gen); i++)
zfid->zf_gen[i] = 0;
ZFS_EXIT(zfsvfs);
return (0);
}
/*
* Construct a full dataset name in full_name: "pool/dataset@snap_name"
*/
static int
zfsctl_snapshot_name(zfsvfs_t *zfsvfs, const char *snap_name, int len,
char *full_name)
{
objset_t *os = zfsvfs->z_os;
if (zfs_component_namecheck(snap_name, NULL, NULL) != 0)
return (SET_ERROR(EILSEQ));
dmu_objset_name(os, full_name);
if ((strlen(full_name) + 1 + strlen(snap_name)) >= len)
return (SET_ERROR(ENAMETOOLONG));
(void) strcat(full_name, "@");
(void) strcat(full_name, snap_name);
return (0);
}
/*
* Returns full path in full_path: "/pool/dataset/.zfs/snapshot/snap_name/"
*/
static int
zfsctl_snapshot_path_objset(zfsvfs_t *zfsvfs, uint64_t objsetid,
int path_len, char *full_path)
{
objset_t *os = zfsvfs->z_os;
fstrans_cookie_t cookie;
char *snapname;
boolean_t case_conflict;
uint64_t id, pos = 0;
int error = 0;
if (zfsvfs->z_vfs->vfs_mntpoint == NULL)
return (SET_ERROR(ENOENT));
cookie = spl_fstrans_mark();
snapname = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP);
while (error == 0) {
dsl_pool_config_enter(dmu_objset_pool(os), FTAG);
error = dmu_snapshot_list_next(zfsvfs->z_os,
ZFS_MAX_DATASET_NAME_LEN, snapname, &id, &pos,
&case_conflict);
dsl_pool_config_exit(dmu_objset_pool(os), FTAG);
if (error)
goto out;
if (id == objsetid)
break;
}
snprintf(full_path, path_len, "%s/.zfs/snapshot/%s",
zfsvfs->z_vfs->vfs_mntpoint, snapname);
out:
kmem_free(snapname, ZFS_MAX_DATASET_NAME_LEN);
spl_fstrans_unmark(cookie);
return (error);
}
/*
* Special case the handling of "..".
*/
int
zfsctl_root_lookup(struct inode *dip, const char *name, struct inode **ipp,
int flags, cred_t *cr, int *direntflags, pathname_t *realpnp)
{
zfsvfs_t *zfsvfs = ITOZSB(dip);
int error = 0;
ZFS_ENTER(zfsvfs);
if (strcmp(name, "..") == 0) {
*ipp = dip->i_sb->s_root->d_inode;
} else if (strcmp(name, ZFS_SNAPDIR_NAME) == 0) {
*ipp = zfsctl_inode_lookup(zfsvfs, ZFSCTL_INO_SNAPDIR,
&zpl_fops_snapdir, &zpl_ops_snapdir);
} else if (strcmp(name, ZFS_SHAREDIR_NAME) == 0) {
*ipp = zfsctl_inode_lookup(zfsvfs, ZFSCTL_INO_SHARES,
&zpl_fops_shares, &zpl_ops_shares);
} else {
*ipp = NULL;
}
if (*ipp == NULL)
error = SET_ERROR(ENOENT);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Lookup entry point for the 'snapshot' directory. Try to open the
* snapshot if it exist, creating the pseudo filesystem inode as necessary.
*/
int
zfsctl_snapdir_lookup(struct inode *dip, const char *name, struct inode **ipp,
int flags, cred_t *cr, int *direntflags, pathname_t *realpnp)
{
zfsvfs_t *zfsvfs = ITOZSB(dip);
uint64_t id;
int error;
ZFS_ENTER(zfsvfs);
error = dmu_snapshot_lookup(zfsvfs->z_os, name, &id);
if (error) {
ZFS_EXIT(zfsvfs);
return (error);
}
*ipp = zfsctl_inode_lookup(zfsvfs, ZFSCTL_INO_SNAPDIRS - id,
&simple_dir_operations, &simple_dir_inode_operations);
if (*ipp == NULL)
error = SET_ERROR(ENOENT);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Renaming a directory under '.zfs/snapshot' will automatically trigger
* a rename of the snapshot to the new given name. The rename is confined
* to the '.zfs/snapshot' directory snapshots cannot be moved elsewhere.
*/
int
zfsctl_snapdir_rename(struct inode *sdip, const char *snm,
struct inode *tdip, const char *tnm, cred_t *cr, int flags)
{
zfsvfs_t *zfsvfs = ITOZSB(sdip);
char *to, *from, *real, *fsname;
int error;
if (!zfs_admin_snapshot)
return (SET_ERROR(EACCES));
ZFS_ENTER(zfsvfs);
to = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP);
from = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP);
real = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP);
fsname = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP);
if (zfsvfs->z_case == ZFS_CASE_INSENSITIVE) {
error = dmu_snapshot_realname(zfsvfs->z_os, snm, real,
ZFS_MAX_DATASET_NAME_LEN, NULL);
if (error == 0) {
snm = real;
} else if (error != ENOTSUP) {
goto out;
}
}
dmu_objset_name(zfsvfs->z_os, fsname);
error = zfsctl_snapshot_name(ITOZSB(sdip), snm,
ZFS_MAX_DATASET_NAME_LEN, from);
if (error == 0)
error = zfsctl_snapshot_name(ITOZSB(tdip), tnm,
ZFS_MAX_DATASET_NAME_LEN, to);
if (error == 0)
error = zfs_secpolicy_rename_perms(from, to, cr);
if (error != 0)
goto out;
/*
* Cannot move snapshots out of the snapdir.
*/
if (sdip != tdip) {
error = SET_ERROR(EINVAL);
goto out;
}
/*
* No-op when names are identical.
*/
if (strcmp(snm, tnm) == 0) {
error = 0;
goto out;
}
rw_enter(&zfs_snapshot_lock, RW_WRITER);
error = dsl_dataset_rename_snapshot(fsname, snm, tnm, B_FALSE);
if (error == 0)
(void) zfsctl_snapshot_rename(snm, tnm);
rw_exit(&zfs_snapshot_lock);
out:
kmem_free(from, ZFS_MAX_DATASET_NAME_LEN);
kmem_free(to, ZFS_MAX_DATASET_NAME_LEN);
kmem_free(real, ZFS_MAX_DATASET_NAME_LEN);
kmem_free(fsname, ZFS_MAX_DATASET_NAME_LEN);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Removing a directory under '.zfs/snapshot' will automatically trigger
* the removal of the snapshot with the given name.
*/
int
zfsctl_snapdir_remove(struct inode *dip, const char *name, cred_t *cr,
int flags)
{
zfsvfs_t *zfsvfs = ITOZSB(dip);
char *snapname, *real;
int error;
if (!zfs_admin_snapshot)
return (SET_ERROR(EACCES));
ZFS_ENTER(zfsvfs);
snapname = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP);
real = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP);
if (zfsvfs->z_case == ZFS_CASE_INSENSITIVE) {
error = dmu_snapshot_realname(zfsvfs->z_os, name, real,
ZFS_MAX_DATASET_NAME_LEN, NULL);
if (error == 0) {
name = real;
} else if (error != ENOTSUP) {
goto out;
}
}
error = zfsctl_snapshot_name(ITOZSB(dip), name,
ZFS_MAX_DATASET_NAME_LEN, snapname);
if (error == 0)
error = zfs_secpolicy_destroy_perms(snapname, cr);
if (error != 0)
goto out;
error = zfsctl_snapshot_unmount(snapname, MNT_FORCE);
if ((error == 0) || (error == ENOENT))
error = dsl_destroy_snapshot(snapname, B_FALSE);
out:
kmem_free(snapname, ZFS_MAX_DATASET_NAME_LEN);
kmem_free(real, ZFS_MAX_DATASET_NAME_LEN);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Creating a directory under '.zfs/snapshot' will automatically trigger
* the creation of a new snapshot with the given name.
*/
int
zfsctl_snapdir_mkdir(struct inode *dip, const char *dirname, vattr_t *vap,
struct inode **ipp, cred_t *cr, int flags)
{
zfsvfs_t *zfsvfs = ITOZSB(dip);
char *dsname;
int error;
if (!zfs_admin_snapshot)
return (SET_ERROR(EACCES));
dsname = kmem_alloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP);
if (zfs_component_namecheck(dirname, NULL, NULL) != 0) {
error = SET_ERROR(EILSEQ);
goto out;
}
dmu_objset_name(zfsvfs->z_os, dsname);
error = zfs_secpolicy_snapshot_perms(dsname, cr);
if (error != 0)
goto out;
if (error == 0) {
error = dmu_objset_snapshot_one(dsname, dirname);
if (error != 0)
goto out;
error = zfsctl_snapdir_lookup(dip, dirname, ipp,
0, cr, NULL, NULL);
}
out:
kmem_free(dsname, ZFS_MAX_DATASET_NAME_LEN);
return (error);
}
/*
* Flush everything out of the kernel's export table and such.
* This is needed as once the snapshot is used over NFS, its
* entries in svc_export and svc_expkey caches hold reference
* to the snapshot mount point. There is no known way of flushing
* only the entries related to the snapshot.
*/
static void
exportfs_flush(void)
{
char *argv[] = { "/usr/sbin/exportfs", "-f", NULL };
char *envp[] = { NULL };
(void) call_usermodehelper(argv[0], argv, envp, UMH_WAIT_PROC);
}
/*
* Attempt to unmount a snapshot by making a call to user space.
* There is no assurance that this can or will succeed, is just a
* best effort. In the case where it does fail, perhaps because
* it's in use, the unmount will fail harmlessly.
*/
int
zfsctl_snapshot_unmount(const char *snapname, int flags)
{
char *argv[] = { "/usr/bin/env", "umount", "-t", "zfs", "-n", NULL,
NULL };
char *envp[] = { NULL };
zfs_snapentry_t *se;
int error;
rw_enter(&zfs_snapshot_lock, RW_READER);
if ((se = zfsctl_snapshot_find_by_name(snapname)) == NULL) {
rw_exit(&zfs_snapshot_lock);
return (SET_ERROR(ENOENT));
}
rw_exit(&zfs_snapshot_lock);
exportfs_flush();
if (flags & MNT_FORCE)
argv[4] = "-fn";
argv[5] = se->se_path;
dprintf("unmount; path=%s\n", se->se_path);
error = call_usermodehelper(argv[0], argv, envp, UMH_WAIT_PROC);
zfsctl_snapshot_rele(se);
/*
* The umount system utility will return 256 on error. We must
* assume this error is because the file system is busy so it is
* converted to the more sensible EBUSY.
*/
if (error)
error = SET_ERROR(EBUSY);
return (error);
}
int
zfsctl_snapshot_mount(struct path *path, int flags)
{
struct dentry *dentry = path->dentry;
struct inode *ip = dentry->d_inode;
zfsvfs_t *zfsvfs;
zfsvfs_t *snap_zfsvfs;
zfs_snapentry_t *se;
char *full_name, *full_path;
char *argv[] = { "/usr/bin/env", "mount", "-t", "zfs", "-n", NULL, NULL,
NULL };
char *envp[] = { NULL };
int error;
struct path spath;
if (ip == NULL)
return (SET_ERROR(EISDIR));
zfsvfs = ITOZSB(ip);
ZFS_ENTER(zfsvfs);
full_name = kmem_zalloc(ZFS_MAX_DATASET_NAME_LEN, KM_SLEEP);
full_path = kmem_zalloc(MAXPATHLEN, KM_SLEEP);
error = zfsctl_snapshot_name(zfsvfs, dname(dentry),
ZFS_MAX_DATASET_NAME_LEN, full_name);
if (error)
goto error;
/*
* Construct a mount point path from sb of the ctldir inode and dirent
* name, instead of from d_path(), so that chroot'd process doesn't fail
* on mount.zfs(8).
*/
snprintf(full_path, MAXPATHLEN, "%s/.zfs/snapshot/%s",
zfsvfs->z_vfs->vfs_mntpoint ? zfsvfs->z_vfs->vfs_mntpoint : "",
dname(dentry));
/*
* Multiple concurrent automounts of a snapshot are never allowed.
* The snapshot may be manually mounted as many times as desired.
*/
if (zfsctl_snapshot_ismounted(full_name)) {
error = 0;
goto error;
}
/*
* Attempt to mount the snapshot from user space. Normally this
* would be done using the vfs_kern_mount() function, however that
* function is marked GPL-only and cannot be used. On error we
* careful to log the real error to the console and return EISDIR
* to safely abort the automount. This should be very rare.
*
* If the user mode helper happens to return EBUSY, a concurrent
* mount is already in progress in which case the error is ignored.
* Take note that if the program was executed successfully the return
* value from call_usermodehelper() will be (exitcode << 8 + signal).
*/
dprintf("mount; name=%s path=%s\n", full_name, full_path);
argv[5] = full_name;
argv[6] = full_path;
error = call_usermodehelper(argv[0], argv, envp, UMH_WAIT_PROC);
if (error) {
if (!(error & MOUNT_BUSY << 8)) {
zfs_dbgmsg("Unable to automount %s error=%d",
full_path, error);
error = SET_ERROR(EISDIR);
} else {
/*
* EBUSY, this could mean a concurrent mount, or the
* snapshot has already been mounted at completely
* different place. We return 0 so VFS will retry. For
* the latter case the VFS will retry several times
* and return ELOOP, which is probably not a very good
* behavior.
*/
error = 0;
}
goto error;
}
/*
* Follow down in to the mounted snapshot and set MNT_SHRINKABLE
* to identify this as an automounted filesystem.
*/
spath = *path;
path_get(&spath);
if (follow_down_one(&spath)) {
snap_zfsvfs = ITOZSB(spath.dentry->d_inode);
snap_zfsvfs->z_parent = zfsvfs;
dentry = spath.dentry;
spath.mnt->mnt_flags |= MNT_SHRINKABLE;
rw_enter(&zfs_snapshot_lock, RW_WRITER);
se = zfsctl_snapshot_alloc(full_name, full_path,
snap_zfsvfs->z_os->os_spa, dmu_objset_id(snap_zfsvfs->z_os),
dentry);
zfsctl_snapshot_add(se);
zfsctl_snapshot_unmount_delay_impl(se, zfs_expire_snapshot);
rw_exit(&zfs_snapshot_lock);
}
path_put(&spath);
error:
kmem_free(full_name, ZFS_MAX_DATASET_NAME_LEN);
kmem_free(full_path, MAXPATHLEN);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Get the snapdir inode from fid
*/
int
zfsctl_snapdir_vget(struct super_block *sb, uint64_t objsetid, int gen,
struct inode **ipp)
{
int error;
struct path path;
char *mnt;
struct dentry *dentry;
mnt = kmem_alloc(MAXPATHLEN, KM_SLEEP);
error = zfsctl_snapshot_path_objset(sb->s_fs_info, objsetid,
MAXPATHLEN, mnt);
if (error)
goto out;
/* Trigger automount */
error = -kern_path(mnt, LOOKUP_FOLLOW|LOOKUP_DIRECTORY, &path);
if (error)
goto out;
path_put(&path);
/*
* Get the snapdir inode. Note, we don't want to use the above
* path because it contains the root of the snapshot rather
* than the snapdir.
*/
*ipp = ilookup(sb, ZFSCTL_INO_SNAPDIRS - objsetid);
if (*ipp == NULL) {
error = SET_ERROR(ENOENT);
goto out;
}
/* check gen, see zfsctl_snapdir_fid */
dentry = d_obtain_alias(igrab(*ipp));
if (gen != (!IS_ERR(dentry) && d_mountpoint(dentry))) {
iput(*ipp);
*ipp = NULL;
error = SET_ERROR(ENOENT);
}
if (!IS_ERR(dentry))
dput(dentry);
out:
kmem_free(mnt, MAXPATHLEN);
return (error);
}
int
zfsctl_shares_lookup(struct inode *dip, char *name, struct inode **ipp,
int flags, cred_t *cr, int *direntflags, pathname_t *realpnp)
{
zfsvfs_t *zfsvfs = ITOZSB(dip);
znode_t *zp;
znode_t *dzp;
int error;
ZFS_ENTER(zfsvfs);
if (zfsvfs->z_shares_dir == 0) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENOTSUP));
}
if ((error = zfs_zget(zfsvfs, zfsvfs->z_shares_dir, &dzp)) == 0) {
error = zfs_lookup(dzp, name, &zp, 0, cr, NULL, NULL);
zrele(dzp);
}
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Initialize the various pieces we'll need to create and manipulate .zfs
* directories. Currently this is unused but available.
*/
void
zfsctl_init(void)
{
avl_create(&zfs_snapshots_by_name, snapentry_compare_by_name,
sizeof (zfs_snapentry_t), offsetof(zfs_snapentry_t,
se_node_name));
avl_create(&zfs_snapshots_by_objsetid, snapentry_compare_by_objsetid,
sizeof (zfs_snapentry_t), offsetof(zfs_snapentry_t,
se_node_objsetid));
rw_init(&zfs_snapshot_lock, NULL, RW_DEFAULT, NULL);
}
/*
* Cleanup the various pieces we needed for .zfs directories. In particular
* ensure the expiry timer is canceled safely.
*/
void
zfsctl_fini(void)
{
avl_destroy(&zfs_snapshots_by_name);
avl_destroy(&zfs_snapshots_by_objsetid);
rw_destroy(&zfs_snapshot_lock);
}
module_param(zfs_admin_snapshot, int, 0644);
MODULE_PARM_DESC(zfs_admin_snapshot, "Enable mkdir/rmdir/mv in .zfs/snapshot");
module_param(zfs_expire_snapshot, int, 0644);
MODULE_PARM_DESC(zfs_expire_snapshot, "Seconds to expire .zfs/snapshot");
diff --git a/module/os/linux/zfs/zfs_vnops_os.c b/module/os/linux/zfs/zfs_vnops_os.c
index 991d1aa1c22c..294a8e6314d2 100644
--- a/module/os/linux/zfs/zfs_vnops_os.c
+++ b/module/os/linux/zfs/zfs_vnops_os.c
@@ -1,4089 +1,4096 @@
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2012, 2018 by Delphix. All rights reserved.
* Copyright (c) 2015 by Chunwei Chen. All rights reserved.
* Copyright 2017 Nexenta Systems, Inc.
*/
/* Portions Copyright 2007 Jeremy Teo */
/* Portions Copyright 2010 Robert Milkowski */
#include <sys/types.h>
#include <sys/param.h>
#include <sys/time.h>
#include <sys/sysmacros.h>
#include <sys/vfs.h>
#include <sys/file.h>
#include <sys/stat.h>
#include <sys/kmem.h>
#include <sys/taskq.h>
#include <sys/uio.h>
#include <sys/vmsystm.h>
#include <sys/atomic.h>
#include <sys/pathname.h>
#include <sys/cmn_err.h>
#include <sys/errno.h>
#include <sys/zfs_dir.h>
#include <sys/zfs_acl.h>
#include <sys/zfs_ioctl.h>
#include <sys/fs/zfs.h>
#include <sys/dmu.h>
#include <sys/dmu_objset.h>
#include <sys/spa.h>
#include <sys/txg.h>
#include <sys/dbuf.h>
#include <sys/zap.h>
#include <sys/sa.h>
#include <sys/policy.h>
#include <sys/sunddi.h>
#include <sys/sid.h>
#include <sys/zfs_ctldir.h>
#include <sys/zfs_fuid.h>
#include <sys/zfs_quota.h>
#include <sys/zfs_sa.h>
#include <sys/zfs_vnops.h>
#include <sys/zfs_rlock.h>
#include <sys/cred.h>
#include <sys/zpl.h>
#include <sys/zil.h>
#include <sys/sa_impl.h>
/*
* Programming rules.
*
* Each vnode op performs some logical unit of work. To do this, the ZPL must
* properly lock its in-core state, create a DMU transaction, do the work,
* record this work in the intent log (ZIL), commit the DMU transaction,
* and wait for the intent log to commit if it is a synchronous operation.
* Moreover, the vnode ops must work in both normal and log replay context.
* The ordering of events is important to avoid deadlocks and references
* to freed memory. The example below illustrates the following Big Rules:
*
* (1) A check must be made in each zfs thread for a mounted file system.
* This is done avoiding races using ZFS_ENTER(zfsvfs).
* A ZFS_EXIT(zfsvfs) is needed before all returns. Any znodes
* must be checked with ZFS_VERIFY_ZP(zp). Both of these macros
* can return EIO from the calling function.
*
* (2) zrele() should always be the last thing except for zil_commit() (if
* necessary) and ZFS_EXIT(). This is for 3 reasons: First, if it's the
* last reference, the vnode/znode can be freed, so the zp may point to
* freed memory. Second, the last reference will call zfs_zinactive(),
* which may induce a lot of work -- pushing cached pages (which acquires
* range locks) and syncing out cached atime changes. Third,
* zfs_zinactive() may require a new tx, which could deadlock the system
* if you were already holding one. This deadlock occurs because the tx
* currently being operated on prevents a txg from syncing, which
* prevents the new tx from progressing, resulting in a deadlock. If you
* must call zrele() within a tx, use zfs_zrele_async(). Note that iput()
* is a synonym for zrele().
*
* (3) All range locks must be grabbed before calling dmu_tx_assign(),
* as they can span dmu_tx_assign() calls.
*
* (4) If ZPL locks are held, pass TXG_NOWAIT as the second argument to
* dmu_tx_assign(). This is critical because we don't want to block
* while holding locks.
*
* If no ZPL locks are held (aside from ZFS_ENTER()), use TXG_WAIT. This
* reduces lock contention and CPU usage when we must wait (note that if
* throughput is constrained by the storage, nearly every transaction
* must wait).
*
* Note, in particular, that if a lock is sometimes acquired before
* the tx assigns, and sometimes after (e.g. z_lock), then failing
* to use a non-blocking assign can deadlock the system. The scenario:
*
* Thread A has grabbed a lock before calling dmu_tx_assign().
* Thread B is in an already-assigned tx, and blocks for this lock.
* Thread A calls dmu_tx_assign(TXG_WAIT) and blocks in txg_wait_open()
* forever, because the previous txg can't quiesce until B's tx commits.
*
* If dmu_tx_assign() returns ERESTART and zfsvfs->z_assign is TXG_NOWAIT,
* then drop all locks, call dmu_tx_wait(), and try again. On subsequent
* calls to dmu_tx_assign(), pass TXG_NOTHROTTLE in addition to TXG_NOWAIT,
* to indicate that this operation has already called dmu_tx_wait().
* This will ensure that we don't retry forever, waiting a short bit
* each time.
*
* (5) If the operation succeeded, generate the intent log entry for it
* before dropping locks. This ensures that the ordering of events
* in the intent log matches the order in which they actually occurred.
* During ZIL replay the zfs_log_* functions will update the sequence
* number to indicate the zil transaction has replayed.
*
* (6) At the end of each vnode op, the DMU tx must always commit,
* regardless of whether there were any errors.
*
* (7) After dropping all locks, invoke zil_commit(zilog, foid)
* to ensure that synchronous semantics are provided when necessary.
*
* In general, this is how things should be ordered in each vnode op:
*
* ZFS_ENTER(zfsvfs); // exit if unmounted
* top:
* zfs_dirent_lock(&dl, ...) // lock directory entry (may igrab())
* rw_enter(...); // grab any other locks you need
* tx = dmu_tx_create(...); // get DMU tx
* dmu_tx_hold_*(); // hold each object you might modify
* error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
* if (error) {
* rw_exit(...); // drop locks
* zfs_dirent_unlock(dl); // unlock directory entry
* zrele(...); // release held znodes
* if (error == ERESTART) {
* waited = B_TRUE;
* dmu_tx_wait(tx);
* dmu_tx_abort(tx);
* goto top;
* }
* dmu_tx_abort(tx); // abort DMU tx
* ZFS_EXIT(zfsvfs); // finished in zfs
* return (error); // really out of space
* }
* error = do_real_work(); // do whatever this VOP does
* if (error == 0)
* zfs_log_*(...); // on success, make ZIL entry
* dmu_tx_commit(tx); // commit DMU tx -- error or not
* rw_exit(...); // drop locks
* zfs_dirent_unlock(dl); // unlock directory entry
* zrele(...); // release held znodes
* zil_commit(zilog, foid); // synchronous when necessary
* ZFS_EXIT(zfsvfs); // finished in zfs
* return (error); // done, report error
*/
/*
* Virus scanning is unsupported. It would be possible to add a hook
* here to performance the required virus scan. This could be done
* entirely in the kernel or potentially as an update to invoke a
* scanning utility.
*/
static int
zfs_vscan(struct inode *ip, cred_t *cr, int async)
{
return (0);
}
/* ARGSUSED */
int
zfs_open(struct inode *ip, int mode, int flag, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
/* Honor ZFS_APPENDONLY file attribute */
if (blk_mode_is_open_write(mode) && (zp->z_pflags & ZFS_APPENDONLY) &&
((flag & O_APPEND) == 0)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
/* Virus scan eligible files on open */
if (!zfs_has_ctldir(zp) && zfsvfs->z_vscan && S_ISREG(ip->i_mode) &&
!(zp->z_pflags & ZFS_AV_QUARANTINED) && zp->z_size > 0) {
if (zfs_vscan(ip, cr, 0) != 0) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EACCES));
}
}
/* Keep a count of the synchronous opens in the znode */
if (flag & O_SYNC)
atomic_inc_32(&zp->z_sync_cnt);
ZFS_EXIT(zfsvfs);
return (0);
}
/* ARGSUSED */
int
zfs_close(struct inode *ip, int flag, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
/* Decrement the synchronous opens in the znode */
if (flag & O_SYNC)
atomic_dec_32(&zp->z_sync_cnt);
if (!zfs_has_ctldir(zp) && zfsvfs->z_vscan && S_ISREG(ip->i_mode) &&
!(zp->z_pflags & ZFS_AV_QUARANTINED) && zp->z_size > 0)
VERIFY(zfs_vscan(ip, cr, 1) == 0);
ZFS_EXIT(zfsvfs);
return (0);
}
#if defined(_KERNEL)
static int zfs_fillpage(struct inode *ip, struct page *pp);
/*
* When a file is memory mapped, we must keep the IO data synchronized
* between the DMU cache and the memory mapped pages. Update all mapped
* pages with the contents of the coresponding dmu buffer.
*/
void
update_pages(znode_t *zp, int64_t start, int len, objset_t *os)
{
struct address_space *mp = ZTOI(zp)->i_mapping;
int64_t off = start & (PAGE_SIZE - 1);
for (start &= PAGE_MASK; len > 0; start += PAGE_SIZE) {
uint64_t nbytes = MIN(PAGE_SIZE - off, len);
struct page *pp = find_lock_page(mp, start >> PAGE_SHIFT);
if (pp) {
if (mapping_writably_mapped(mp))
flush_dcache_page(pp);
void *pb = kmap(pp);
int error = dmu_read(os, zp->z_id, start + off,
nbytes, pb + off, DMU_READ_PREFETCH);
kunmap(pp);
if (error) {
SetPageError(pp);
ClearPageUptodate(pp);
} else {
ClearPageError(pp);
SetPageUptodate(pp);
if (mapping_writably_mapped(mp))
flush_dcache_page(pp);
mark_page_accessed(pp);
}
unlock_page(pp);
put_page(pp);
}
len -= nbytes;
off = 0;
}
}
/*
* When a file is memory mapped, we must keep the I/O data synchronized
* between the DMU cache and the memory mapped pages. Preferentially read
* from memory mapped pages, otherwise fallback to reading through the dmu.
*/
int
mappedread(znode_t *zp, int nbytes, zfs_uio_t *uio)
{
struct inode *ip = ZTOI(zp);
struct address_space *mp = ip->i_mapping;
int64_t start = uio->uio_loffset;
int64_t off = start & (PAGE_SIZE - 1);
int len = nbytes;
int error = 0;
for (start &= PAGE_MASK; len > 0; start += PAGE_SIZE) {
uint64_t bytes = MIN(PAGE_SIZE - off, len);
struct page *pp = find_lock_page(mp, start >> PAGE_SHIFT);
if (pp) {
/*
* If filemap_fault() retries there exists a window
* where the page will be unlocked and not up to date.
* In this case we must try and fill the page.
*/
if (unlikely(!PageUptodate(pp))) {
error = zfs_fillpage(ip, pp);
if (error) {
unlock_page(pp);
put_page(pp);
return (error);
}
}
ASSERT(PageUptodate(pp) || PageDirty(pp));
unlock_page(pp);
void *pb = kmap(pp);
error = zfs_uiomove(pb + off, bytes, UIO_READ, uio);
kunmap(pp);
if (mapping_writably_mapped(mp))
flush_dcache_page(pp);
mark_page_accessed(pp);
put_page(pp);
} else {
error = dmu_read_uio_dbuf(sa_get_db(zp->z_sa_hdl),
uio, bytes);
}
len -= bytes;
off = 0;
if (error)
break;
}
return (error);
}
#endif /* _KERNEL */
unsigned long zfs_delete_blocks = DMU_MAX_DELETEBLKCNT;
/*
* Write the bytes to a file.
*
* IN: zp - znode of file to be written to
* data - bytes to write
* len - number of bytes to write
* pos - offset to start writing at
*
* OUT: resid - remaining bytes to write
*
* RETURN: 0 if success
* positive error code if failure. EIO is returned
* for a short write when residp isn't provided.
*
* Timestamps:
* zp - ctime|mtime updated if byte count > 0
*/
int
zfs_write_simple(znode_t *zp, const void *data, size_t len,
loff_t pos, size_t *residp)
{
fstrans_cookie_t cookie;
int error;
struct iovec iov;
iov.iov_base = (void *)data;
iov.iov_len = len;
zfs_uio_t uio;
zfs_uio_iovec_init(&uio, &iov, 1, pos, UIO_SYSSPACE, len, 0);
cookie = spl_fstrans_mark();
error = zfs_write(zp, &uio, 0, kcred);
spl_fstrans_unmark(cookie);
if (error == 0) {
if (residp != NULL)
*residp = zfs_uio_resid(&uio);
else if (zfs_uio_resid(&uio) != 0)
error = SET_ERROR(EIO);
}
return (error);
}
static void
zfs_rele_async_task(void *arg)
{
iput(arg);
}
void
zfs_zrele_async(znode_t *zp)
{
struct inode *ip = ZTOI(zp);
objset_t *os = ITOZSB(ip)->z_os;
ASSERT(atomic_read(&ip->i_count) > 0);
ASSERT(os != NULL);
/*
* If decrementing the count would put us at 0, we can't do it inline
* here, because that would be synchronous. Instead, dispatch an iput
* to run later.
*
* For more information on the dangers of a synchronous iput, see the
* header comment of this file.
*/
if (!atomic_add_unless(&ip->i_count, -1, 1)) {
VERIFY(taskq_dispatch(dsl_pool_zrele_taskq(dmu_objset_pool(os)),
zfs_rele_async_task, ip, TQ_SLEEP) != TASKQID_INVALID);
}
}
/*
* Lookup an entry in a directory, or an extended attribute directory.
* If it exists, return a held inode reference for it.
*
* IN: zdp - znode of directory to search.
* nm - name of entry to lookup.
* flags - LOOKUP_XATTR set if looking for an attribute.
* cr - credentials of caller.
* direntflags - directory lookup flags
* realpnp - returned pathname.
*
* OUT: zpp - znode of located entry, NULL if not found.
*
* RETURN: 0 on success, error code on failure.
*
* Timestamps:
* NA
*/
/* ARGSUSED */
int
zfs_lookup(znode_t *zdp, char *nm, znode_t **zpp, int flags, cred_t *cr,
int *direntflags, pathname_t *realpnp)
{
zfsvfs_t *zfsvfs = ZTOZSB(zdp);
int error = 0;
/*
* Fast path lookup, however we must skip DNLC lookup
* for case folding or normalizing lookups because the
* DNLC code only stores the passed in name. This means
* creating 'a' and removing 'A' on a case insensitive
* file system would work, but DNLC still thinks 'a'
* exists and won't let you create it again on the next
* pass through fast path.
*/
if (!(flags & (LOOKUP_XATTR | FIGNORECASE))) {
if (!S_ISDIR(ZTOI(zdp)->i_mode)) {
return (SET_ERROR(ENOTDIR));
} else if (zdp->z_sa_hdl == NULL) {
return (SET_ERROR(EIO));
}
if (nm[0] == 0 || (nm[0] == '.' && nm[1] == '\0')) {
error = zfs_fastaccesschk_execute(zdp, cr);
if (!error) {
*zpp = zdp;
zhold(*zpp);
return (0);
}
return (error);
}
}
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zdp);
*zpp = NULL;
if (flags & LOOKUP_XATTR) {
/*
* We don't allow recursive attributes..
* Maybe someday we will.
*/
if (zdp->z_pflags & ZFS_XATTR) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
if ((error = zfs_get_xattrdir(zdp, zpp, cr, flags))) {
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Do we have permission to get into attribute directory?
*/
if ((error = zfs_zaccess(*zpp, ACE_EXECUTE, 0,
B_FALSE, cr))) {
zrele(*zpp);
*zpp = NULL;
}
ZFS_EXIT(zfsvfs);
return (error);
}
if (!S_ISDIR(ZTOI(zdp)->i_mode)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENOTDIR));
}
/*
* Check accessibility of directory.
*/
if ((error = zfs_zaccess(zdp, ACE_EXECUTE, 0, B_FALSE, cr))) {
ZFS_EXIT(zfsvfs);
return (error);
}
if (zfsvfs->z_utf8 && u8_validate(nm, strlen(nm),
NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
error = zfs_dirlook(zdp, nm, zpp, flags, direntflags, realpnp);
if ((error == 0) && (*zpp))
zfs_znode_update_vfs(*zpp);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Attempt to create a new entry in a directory. If the entry
* already exists, truncate the file if permissible, else return
* an error. Return the ip of the created or trunc'd file.
*
* IN: dzp - znode of directory to put new file entry in.
* name - name of new file entry.
* vap - attributes of new file.
* excl - flag indicating exclusive or non-exclusive mode.
* mode - mode to open file with.
* cr - credentials of caller.
* flag - file flag.
* vsecp - ACL to be set
*
* OUT: zpp - znode of created or trunc'd entry.
*
* RETURN: 0 on success, error code on failure.
*
* Timestamps:
* dzp - ctime|mtime updated if new entry created
* zp - ctime|mtime always, atime if new
*/
/* ARGSUSED */
int
zfs_create(znode_t *dzp, char *name, vattr_t *vap, int excl,
int mode, znode_t **zpp, cred_t *cr, int flag, vsecattr_t *vsecp)
{
znode_t *zp;
zfsvfs_t *zfsvfs = ZTOZSB(dzp);
zilog_t *zilog;
objset_t *os;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
int error;
uid_t uid;
gid_t gid;
zfs_acl_ids_t acl_ids;
boolean_t fuid_dirtied;
boolean_t have_acl = B_FALSE;
boolean_t waited = B_FALSE;
boolean_t skip_acl = (flag & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
/*
* If we have an ephemeral id, ACL, or XVATTR then
* make sure file system is at proper version
*/
gid = crgetgid(cr);
uid = crgetuid(cr);
if (zfsvfs->z_use_fuids == B_FALSE &&
(vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid)))
return (SET_ERROR(EINVAL));
if (name == NULL)
return (SET_ERROR(EINVAL));
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
os = zfsvfs->z_os;
zilog = zfsvfs->z_log;
if (zfsvfs->z_utf8 && u8_validate(name, strlen(name),
NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
if (vap->va_mask & ATTR_XVATTR) {
if ((error = secpolicy_xvattr((xvattr_t *)vap,
crgetuid(cr), cr, vap->va_mode)) != 0) {
ZFS_EXIT(zfsvfs);
return (error);
}
}
top:
*zpp = NULL;
if (*name == '\0') {
/*
* Null component name refers to the directory itself.
*/
zhold(dzp);
zp = dzp;
dl = NULL;
error = 0;
} else {
/* possible igrab(zp) */
int zflg = 0;
if (flag & FIGNORECASE)
zflg |= ZCILOOK;
error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
NULL, NULL);
if (error) {
if (have_acl)
zfs_acl_ids_free(&acl_ids);
if (strcmp(name, "..") == 0)
error = SET_ERROR(EISDIR);
ZFS_EXIT(zfsvfs);
return (error);
}
}
if (zp == NULL) {
uint64_t txtype;
uint64_t projid = ZFS_DEFAULT_PROJID;
/*
* Create a new file object and update the directory
* to reference it.
*/
if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, skip_acl, cr))) {
if (have_acl)
zfs_acl_ids_free(&acl_ids);
goto out;
}
/*
* We only support the creation of regular files in
* extended attribute directories.
*/
if ((dzp->z_pflags & ZFS_XATTR) && !S_ISREG(vap->va_mode)) {
if (have_acl)
zfs_acl_ids_free(&acl_ids);
error = SET_ERROR(EINVAL);
goto out;
}
if (!have_acl && (error = zfs_acl_ids_create(dzp, 0, vap,
cr, vsecp, &acl_ids)) != 0)
goto out;
have_acl = B_TRUE;
if (S_ISREG(vap->va_mode) || S_ISDIR(vap->va_mode))
projid = zfs_inherit_projid(dzp);
if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, projid)) {
zfs_acl_ids_free(&acl_ids);
error = SET_ERROR(EDQUOT);
goto out;
}
tx = dmu_tx_create(os);
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE);
fuid_dirtied = zfsvfs->z_fuid_dirty;
if (fuid_dirtied)
zfs_fuid_txhold(zfsvfs, tx);
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
if (!zfsvfs->z_use_sa &&
acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
0, acl_ids.z_aclp->z_acl_bytes);
}
error = dmu_tx_assign(tx,
(waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
ZFS_EXIT(zfsvfs);
return (error);
}
zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
error = zfs_link_create(dl, zp, tx, ZNEW);
if (error != 0) {
/*
* Since, we failed to add the directory entry for it,
* delete the newly created dnode.
*/
zfs_znode_delete(zp, tx);
remove_inode_hash(ZTOI(zp));
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
goto out;
}
if (fuid_dirtied)
zfs_fuid_sync(zfsvfs, tx);
txtype = zfs_log_create_txtype(Z_FILE, vsecp, vap);
if (flag & FIGNORECASE)
txtype |= TX_CI;
zfs_log_create(zilog, tx, txtype, dzp, zp, name,
vsecp, acl_ids.z_fuidp, vap);
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
} else {
int aflags = (flag & O_APPEND) ? V_APPEND : 0;
if (have_acl)
zfs_acl_ids_free(&acl_ids);
have_acl = B_FALSE;
/*
* A directory entry already exists for this name.
*/
/*
* Can't truncate an existing file if in exclusive mode.
*/
if (excl) {
error = SET_ERROR(EEXIST);
goto out;
}
/*
* Can't open a directory for writing.
*/
if (S_ISDIR(ZTOI(zp)->i_mode)) {
error = SET_ERROR(EISDIR);
goto out;
}
/*
* Verify requested access to file.
*/
if (mode && (error = zfs_zaccess_rwx(zp, mode, aflags, cr))) {
goto out;
}
mutex_enter(&dzp->z_lock);
dzp->z_seq++;
mutex_exit(&dzp->z_lock);
/*
* Truncate regular files if requested.
*/
if (S_ISREG(ZTOI(zp)->i_mode) &&
(vap->va_mask & ATTR_SIZE) && (vap->va_size == 0)) {
/* we can't hold any locks when calling zfs_freesp() */
if (dl) {
zfs_dirent_unlock(dl);
dl = NULL;
}
error = zfs_freesp(zp, 0, 0, mode, TRUE);
}
}
out:
if (dl)
zfs_dirent_unlock(dl);
if (error) {
if (zp)
zrele(zp);
} else {
zfs_znode_update_vfs(dzp);
zfs_znode_update_vfs(zp);
*zpp = zp;
}
if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
ZFS_EXIT(zfsvfs);
return (error);
}
/* ARGSUSED */
int
zfs_tmpfile(struct inode *dip, vattr_t *vap, int excl,
int mode, struct inode **ipp, cred_t *cr, int flag, vsecattr_t *vsecp)
{
znode_t *zp = NULL, *dzp = ITOZ(dip);
zfsvfs_t *zfsvfs = ITOZSB(dip);
objset_t *os;
dmu_tx_t *tx;
int error;
uid_t uid;
gid_t gid;
zfs_acl_ids_t acl_ids;
uint64_t projid = ZFS_DEFAULT_PROJID;
boolean_t fuid_dirtied;
boolean_t have_acl = B_FALSE;
boolean_t waited = B_FALSE;
/*
* If we have an ephemeral id, ACL, or XVATTR then
* make sure file system is at proper version
*/
gid = crgetgid(cr);
uid = crgetuid(cr);
if (zfsvfs->z_use_fuids == B_FALSE &&
(vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid)))
return (SET_ERROR(EINVAL));
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
os = zfsvfs->z_os;
if (vap->va_mask & ATTR_XVATTR) {
if ((error = secpolicy_xvattr((xvattr_t *)vap,
crgetuid(cr), cr, vap->va_mode)) != 0) {
ZFS_EXIT(zfsvfs);
return (error);
}
}
top:
*ipp = NULL;
/*
* Create a new file object and update the directory
* to reference it.
*/
if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
if (have_acl)
zfs_acl_ids_free(&acl_ids);
goto out;
}
if (!have_acl && (error = zfs_acl_ids_create(dzp, 0, vap,
cr, vsecp, &acl_ids)) != 0)
goto out;
have_acl = B_TRUE;
if (S_ISREG(vap->va_mode) || S_ISDIR(vap->va_mode))
projid = zfs_inherit_projid(dzp);
if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, projid)) {
zfs_acl_ids_free(&acl_ids);
error = SET_ERROR(EDQUOT);
goto out;
}
tx = dmu_tx_create(os);
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
fuid_dirtied = zfsvfs->z_fuid_dirty;
if (fuid_dirtied)
zfs_fuid_txhold(zfsvfs, tx);
if (!zfsvfs->z_use_sa &&
acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
0, acl_ids.z_aclp->z_acl_bytes);
}
error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
ZFS_EXIT(zfsvfs);
return (error);
}
zfs_mknode(dzp, vap, tx, cr, IS_TMPFILE, &zp, &acl_ids);
if (fuid_dirtied)
zfs_fuid_sync(zfsvfs, tx);
/* Add to unlinked set */
zp->z_unlinked = B_TRUE;
zfs_unlinked_add(zp, tx);
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
out:
if (error) {
if (zp)
zrele(zp);
} else {
zfs_znode_update_vfs(dzp);
zfs_znode_update_vfs(zp);
*ipp = ZTOI(zp);
}
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Remove an entry from a directory.
*
* IN: dzp - znode of directory to remove entry from.
* name - name of entry to remove.
* cr - credentials of caller.
* flags - case flags.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
* dzp - ctime|mtime
* ip - ctime (if nlink > 0)
*/
uint64_t null_xattr = 0;
/*ARGSUSED*/
int
zfs_remove(znode_t *dzp, char *name, cred_t *cr, int flags)
{
znode_t *zp;
znode_t *xzp;
zfsvfs_t *zfsvfs = ZTOZSB(dzp);
zilog_t *zilog;
uint64_t acl_obj, xattr_obj;
uint64_t xattr_obj_unlinked = 0;
uint64_t obj = 0;
uint64_t links;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
boolean_t may_delete_now, delete_now = FALSE;
boolean_t unlinked, toobig = FALSE;
uint64_t txtype;
pathname_t *realnmp = NULL;
pathname_t realnm;
int error;
int zflg = ZEXISTS;
boolean_t waited = B_FALSE;
if (name == NULL)
return (SET_ERROR(EINVAL));
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
zilog = zfsvfs->z_log;
if (flags & FIGNORECASE) {
zflg |= ZCILOOK;
pn_alloc(&realnm);
realnmp = &realnm;
}
top:
xattr_obj = 0;
xzp = NULL;
/*
* Attempt to lock directory; fail if entry doesn't exist.
*/
if ((error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
NULL, realnmp))) {
if (realnmp)
pn_free(realnmp);
ZFS_EXIT(zfsvfs);
return (error);
}
if ((error = zfs_zaccess_delete(dzp, zp, cr))) {
goto out;
}
/*
* Need to use rmdir for removing directories.
*/
if (S_ISDIR(ZTOI(zp)->i_mode)) {
error = SET_ERROR(EPERM);
goto out;
}
mutex_enter(&zp->z_lock);
may_delete_now = atomic_read(&ZTOI(zp)->i_count) == 1 &&
!zn_has_cached_data(zp, 0, LLONG_MAX);
mutex_exit(&zp->z_lock);
/*
* We may delete the znode now, or we may put it in the unlinked set;
* it depends on whether we're the last link, and on whether there are
* other holds on the inode. So we dmu_tx_hold() the right things to
* allow for either case.
*/
obj = zp->z_id;
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, zp);
zfs_sa_upgrade_txholds(tx, dzp);
if (may_delete_now) {
toobig = zp->z_size > zp->z_blksz * zfs_delete_blocks;
/* if the file is too big, only hold_free a token amount */
dmu_tx_hold_free(tx, zp->z_id, 0,
(toobig ? DMU_MAX_ACCESS : DMU_OBJECT_END));
}
/* are there any extended attributes? */
error = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
&xattr_obj, sizeof (xattr_obj));
if (error == 0 && xattr_obj) {
error = zfs_zget(zfsvfs, xattr_obj, &xzp);
ASSERT0(error);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
dmu_tx_hold_sa(tx, xzp->z_sa_hdl, B_FALSE);
}
mutex_enter(&zp->z_lock);
if ((acl_obj = zfs_external_acl(zp)) != 0 && may_delete_now)
dmu_tx_hold_free(tx, acl_obj, 0, DMU_OBJECT_END);
mutex_exit(&zp->z_lock);
/* charge as an update -- would be nice not to charge at all */
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
/*
* Mark this transaction as typically resulting in a net free of space
*/
dmu_tx_mark_netfree(tx);
error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
zrele(zp);
if (xzp)
zrele(xzp);
goto top;
}
if (realnmp)
pn_free(realnmp);
dmu_tx_abort(tx);
zrele(zp);
if (xzp)
zrele(xzp);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Remove the directory entry.
*/
error = zfs_link_destroy(dl, zp, tx, zflg, &unlinked);
if (error) {
dmu_tx_commit(tx);
goto out;
}
if (unlinked) {
/*
* Hold z_lock so that we can make sure that the ACL obj
* hasn't changed. Could have been deleted due to
* zfs_sa_upgrade().
*/
mutex_enter(&zp->z_lock);
(void) sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
&xattr_obj_unlinked, sizeof (xattr_obj_unlinked));
delete_now = may_delete_now && !toobig &&
atomic_read(&ZTOI(zp)->i_count) == 1 &&
!zn_has_cached_data(zp, 0, LLONG_MAX) &&
xattr_obj == xattr_obj_unlinked &&
zfs_external_acl(zp) == acl_obj;
}
if (delete_now) {
if (xattr_obj_unlinked) {
ASSERT3U(ZTOI(xzp)->i_nlink, ==, 2);
mutex_enter(&xzp->z_lock);
xzp->z_unlinked = B_TRUE;
clear_nlink(ZTOI(xzp));
links = 0;
error = sa_update(xzp->z_sa_hdl, SA_ZPL_LINKS(zfsvfs),
&links, sizeof (links), tx);
ASSERT3U(error, ==, 0);
mutex_exit(&xzp->z_lock);
zfs_unlinked_add(xzp, tx);
if (zp->z_is_sa)
error = sa_remove(zp->z_sa_hdl,
SA_ZPL_XATTR(zfsvfs), tx);
else
error = sa_update(zp->z_sa_hdl,
SA_ZPL_XATTR(zfsvfs), &null_xattr,
sizeof (uint64_t), tx);
ASSERT0(error);
}
/*
* Add to the unlinked set because a new reference could be
* taken concurrently resulting in a deferred destruction.
*/
zfs_unlinked_add(zp, tx);
mutex_exit(&zp->z_lock);
} else if (unlinked) {
mutex_exit(&zp->z_lock);
zfs_unlinked_add(zp, tx);
}
txtype = TX_REMOVE;
if (flags & FIGNORECASE)
txtype |= TX_CI;
zfs_log_remove(zilog, tx, txtype, dzp, name, obj, unlinked);
dmu_tx_commit(tx);
out:
if (realnmp)
pn_free(realnmp);
zfs_dirent_unlock(dl);
zfs_znode_update_vfs(dzp);
zfs_znode_update_vfs(zp);
if (delete_now)
zrele(zp);
else
zfs_zrele_async(zp);
if (xzp) {
zfs_znode_update_vfs(xzp);
zfs_zrele_async(xzp);
}
if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Create a new directory and insert it into dzp using the name
* provided. Return a pointer to the inserted directory.
*
* IN: dzp - znode of directory to add subdir to.
* dirname - name of new directory.
* vap - attributes of new directory.
* cr - credentials of caller.
* flags - case flags.
* vsecp - ACL to be set
*
* OUT: zpp - znode of created directory.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
* dzp - ctime|mtime updated
* zpp - ctime|mtime|atime updated
*/
/*ARGSUSED*/
int
zfs_mkdir(znode_t *dzp, char *dirname, vattr_t *vap, znode_t **zpp,
cred_t *cr, int flags, vsecattr_t *vsecp)
{
znode_t *zp;
zfsvfs_t *zfsvfs = ZTOZSB(dzp);
zilog_t *zilog;
zfs_dirlock_t *dl;
uint64_t txtype;
dmu_tx_t *tx;
int error;
int zf = ZNEW;
uid_t uid;
gid_t gid = crgetgid(cr);
zfs_acl_ids_t acl_ids;
boolean_t fuid_dirtied;
boolean_t waited = B_FALSE;
ASSERT(S_ISDIR(vap->va_mode));
/*
* If we have an ephemeral id, ACL, or XVATTR then
* make sure file system is at proper version
*/
uid = crgetuid(cr);
if (zfsvfs->z_use_fuids == B_FALSE &&
(vsecp || IS_EPHEMERAL(uid) || IS_EPHEMERAL(gid)))
return (SET_ERROR(EINVAL));
if (dirname == NULL)
return (SET_ERROR(EINVAL));
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
zilog = zfsvfs->z_log;
if (dzp->z_pflags & ZFS_XATTR) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
if (zfsvfs->z_utf8 && u8_validate(dirname,
strlen(dirname), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
if (flags & FIGNORECASE)
zf |= ZCILOOK;
if (vap->va_mask & ATTR_XVATTR) {
if ((error = secpolicy_xvattr((xvattr_t *)vap,
crgetuid(cr), cr, vap->va_mode)) != 0) {
ZFS_EXIT(zfsvfs);
return (error);
}
}
if ((error = zfs_acl_ids_create(dzp, 0, vap, cr,
vsecp, &acl_ids)) != 0) {
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* First make sure the new directory doesn't exist.
*
* Existence is checked first to make sure we don't return
* EACCES instead of EEXIST which can cause some applications
* to fail.
*/
top:
*zpp = NULL;
if ((error = zfs_dirent_lock(&dl, dzp, dirname, &zp, zf,
NULL, NULL))) {
zfs_acl_ids_free(&acl_ids);
ZFS_EXIT(zfsvfs);
return (error);
}
if ((error = zfs_zaccess(dzp, ACE_ADD_SUBDIRECTORY, 0, B_FALSE, cr))) {
zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
ZFS_EXIT(zfsvfs);
return (error);
}
if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, zfs_inherit_projid(dzp))) {
zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EDQUOT));
}
/*
* Add a new entry to the directory.
*/
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, dirname);
dmu_tx_hold_zap(tx, DMU_NEW_OBJECT, FALSE, NULL);
fuid_dirtied = zfsvfs->z_fuid_dirty;
if (fuid_dirtied)
zfs_fuid_txhold(zfsvfs, tx);
if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
acl_ids.z_aclp->z_acl_bytes);
}
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE);
error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Create new node.
*/
zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
/*
* Now put new name in parent dir.
*/
error = zfs_link_create(dl, zp, tx, ZNEW);
if (error != 0) {
zfs_znode_delete(zp, tx);
remove_inode_hash(ZTOI(zp));
goto out;
}
if (fuid_dirtied)
zfs_fuid_sync(zfsvfs, tx);
*zpp = zp;
txtype = zfs_log_create_txtype(Z_DIR, vsecp, vap);
if (flags & FIGNORECASE)
txtype |= TX_CI;
zfs_log_create(zilog, tx, txtype, dzp, zp, dirname, vsecp,
acl_ids.z_fuidp, vap);
out:
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
zfs_dirent_unlock(dl);
if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
if (error != 0) {
zrele(zp);
} else {
zfs_znode_update_vfs(dzp);
zfs_znode_update_vfs(zp);
}
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Remove a directory subdir entry. If the current working
* directory is the same as the subdir to be removed, the
* remove will fail.
*
* IN: dzp - znode of directory to remove from.
* name - name of directory to be removed.
* cwd - inode of current working directory.
* cr - credentials of caller.
* flags - case flags
*
* RETURN: 0 on success, error code on failure.
*
* Timestamps:
* dzp - ctime|mtime updated
*/
/*ARGSUSED*/
int
zfs_rmdir(znode_t *dzp, char *name, znode_t *cwd, cred_t *cr,
int flags)
{
znode_t *zp;
zfsvfs_t *zfsvfs = ZTOZSB(dzp);
zilog_t *zilog;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
int error;
int zflg = ZEXISTS;
boolean_t waited = B_FALSE;
if (name == NULL)
return (SET_ERROR(EINVAL));
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
zilog = zfsvfs->z_log;
if (flags & FIGNORECASE)
zflg |= ZCILOOK;
top:
zp = NULL;
/*
* Attempt to lock directory; fail if entry doesn't exist.
*/
if ((error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg,
NULL, NULL))) {
ZFS_EXIT(zfsvfs);
return (error);
}
if ((error = zfs_zaccess_delete(dzp, zp, cr))) {
goto out;
}
if (!S_ISDIR(ZTOI(zp)->i_mode)) {
error = SET_ERROR(ENOTDIR);
goto out;
}
if (zp == cwd) {
error = SET_ERROR(EINVAL);
goto out;
}
/*
* Grab a lock on the directory to make sure that no one is
* trying to add (or lookup) entries while we are removing it.
*/
rw_enter(&zp->z_name_lock, RW_WRITER);
/*
* Grab a lock on the parent pointer to make sure we play well
* with the treewalk and directory rename code.
*/
rw_enter(&zp->z_parent_lock, RW_WRITER);
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_zap(tx, dzp->z_id, FALSE, name);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
zfs_sa_upgrade_txholds(tx, zp);
zfs_sa_upgrade_txholds(tx, dzp);
dmu_tx_mark_netfree(tx);
error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
rw_exit(&zp->z_parent_lock);
rw_exit(&zp->z_name_lock);
zfs_dirent_unlock(dl);
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
zrele(zp);
goto top;
}
dmu_tx_abort(tx);
zrele(zp);
ZFS_EXIT(zfsvfs);
return (error);
}
error = zfs_link_destroy(dl, zp, tx, zflg, NULL);
if (error == 0) {
uint64_t txtype = TX_RMDIR;
if (flags & FIGNORECASE)
txtype |= TX_CI;
zfs_log_remove(zilog, tx, txtype, dzp, name, ZFS_NO_OBJECT,
B_FALSE);
}
dmu_tx_commit(tx);
rw_exit(&zp->z_parent_lock);
rw_exit(&zp->z_name_lock);
out:
zfs_dirent_unlock(dl);
zfs_znode_update_vfs(dzp);
zfs_znode_update_vfs(zp);
zrele(zp);
if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Read directory entries from the given directory cursor position and emit
* name and position for each entry.
*
* IN: ip - inode of directory to read.
* ctx - directory entry context.
* cr - credentials of caller.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
* ip - atime updated
*
* Note that the low 4 bits of the cookie returned by zap is always zero.
* This allows us to use the low range for "special" directory entries:
* We use 0 for '.', and 1 for '..'. If this is the root of the filesystem,
* we use the offset 2 for the '.zfs' directory.
*/
/* ARGSUSED */
int
zfs_readdir(struct inode *ip, zpl_dir_context_t *ctx, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
objset_t *os;
zap_cursor_t zc;
zap_attribute_t zap;
int error;
uint8_t prefetch;
uint8_t type;
int done = 0;
uint64_t parent;
uint64_t offset; /* must be unsigned; checks for < 1 */
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
&parent, sizeof (parent))) != 0)
goto out;
/*
* Quit if directory has been removed (posix)
*/
if (zp->z_unlinked)
goto out;
error = 0;
os = zfsvfs->z_os;
offset = ctx->pos;
prefetch = zp->z_zn_prefetch;
/*
* Initialize the iterator cursor.
*/
if (offset <= 3) {
/*
* Start iteration from the beginning of the directory.
*/
zap_cursor_init(&zc, os, zp->z_id);
} else {
/*
* The offset is a serialized cursor.
*/
zap_cursor_init_serialized(&zc, os, zp->z_id, offset);
}
/*
* Transform to file-system independent format
*/
while (!done) {
uint64_t objnum;
/*
* Special case `.', `..', and `.zfs'.
*/
if (offset == 0) {
(void) strcpy(zap.za_name, ".");
zap.za_normalization_conflict = 0;
objnum = zp->z_id;
type = DT_DIR;
} else if (offset == 1) {
(void) strcpy(zap.za_name, "..");
zap.za_normalization_conflict = 0;
objnum = parent;
type = DT_DIR;
} else if (offset == 2 && zfs_show_ctldir(zp)) {
(void) strcpy(zap.za_name, ZFS_CTLDIR_NAME);
zap.za_normalization_conflict = 0;
objnum = ZFSCTL_INO_ROOT;
type = DT_DIR;
} else {
/*
* Grab next entry.
*/
if ((error = zap_cursor_retrieve(&zc, &zap))) {
if (error == ENOENT)
break;
else
goto update;
}
/*
* Allow multiple entries provided the first entry is
* the object id. Non-zpl consumers may safely make
* use of the additional space.
*
* XXX: This should be a feature flag for compatibility
*/
if (zap.za_integer_length != 8 ||
zap.za_num_integers == 0) {
cmn_err(CE_WARN, "zap_readdir: bad directory "
"entry, obj = %lld, offset = %lld, "
"length = %d, num = %lld\n",
(u_longlong_t)zp->z_id,
(u_longlong_t)offset,
zap.za_integer_length,
(u_longlong_t)zap.za_num_integers);
error = SET_ERROR(ENXIO);
goto update;
}
objnum = ZFS_DIRENT_OBJ(zap.za_first_integer);
type = ZFS_DIRENT_TYPE(zap.za_first_integer);
}
done = !zpl_dir_emit(ctx, zap.za_name, strlen(zap.za_name),
objnum, type);
if (done)
break;
/* Prefetch znode */
if (prefetch) {
dmu_prefetch(os, objnum, 0, 0, 0,
ZIO_PRIORITY_SYNC_READ);
}
/*
* Move to the next entry, fill in the previous offset.
*/
if (offset > 2 || (offset == 2 && !zfs_show_ctldir(zp))) {
zap_cursor_advance(&zc);
offset = zap_cursor_serialize(&zc);
} else {
offset += 1;
}
ctx->pos = offset;
}
zp->z_zn_prefetch = B_FALSE; /* a lookup will re-enable pre-fetching */
update:
zap_cursor_fini(&zc);
if (error == ENOENT)
error = 0;
out:
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Get the basic file attributes and place them in the provided kstat
* structure. The inode is assumed to be the authoritative source
* for most of the attributes. However, the znode currently has the
* authoritative atime, blksize, and block count.
*
* IN: ip - inode of file.
*
* OUT: sp - kstat values.
*
* RETURN: 0 (always succeeds)
*/
/* ARGSUSED */
int
#ifdef HAVE_GENERIC_FILLATTR_IDMAP_REQMASK
zfs_getattr_fast(zidmap_t *user_ns, u32 request_mask, struct inode *ip,
struct kstat *sp)
#else
zfs_getattr_fast(zidmap_t *user_ns, struct inode *ip, struct kstat *sp)
#endif
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
uint32_t blksize;
u_longlong_t nblocks;
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
mutex_enter(&zp->z_lock);
#ifdef HAVE_GENERIC_FILLATTR_IDMAP_REQMASK
zpl_generic_fillattr(user_ns, request_mask, ip, sp);
#else
zpl_generic_fillattr(user_ns, ip, sp);
#endif
/*
* +1 link count for root inode with visible '.zfs' directory.
*/
if ((zp->z_id == zfsvfs->z_root) && zfs_show_ctldir(zp))
if (sp->nlink < ZFS_LINK_MAX)
sp->nlink++;
sa_object_size(zp->z_sa_hdl, &blksize, &nblocks);
sp->blksize = blksize;
sp->blocks = nblocks;
if (unlikely(zp->z_blksz == 0)) {
/*
* Block size hasn't been set; suggest maximal I/O transfers.
*/
sp->blksize = zfsvfs->z_max_blksz;
}
mutex_exit(&zp->z_lock);
/*
* Required to prevent NFS client from detecting different inode
* numbers of snapshot root dentry before and after snapshot mount.
*/
if (zfsvfs->z_issnap) {
if (ip->i_sb->s_root->d_inode == ip)
sp->ino = ZFSCTL_INO_SNAPDIRS -
dmu_objset_id(zfsvfs->z_os);
}
ZFS_EXIT(zfsvfs);
return (0);
}
/*
* For the operation of changing file's user/group/project, we need to
* handle not only the main object that is assigned to the file directly,
* but also the ones that are used by the file via hidden xattr directory.
*
* Because the xattr directory may contains many EA entries, as to it may
* be impossible to change all of them via the transaction of changing the
* main object's user/group/project attributes. Then we have to change them
* via other multiple independent transactions one by one. It may be not good
* solution, but we have no better idea yet.
*/
static int
zfs_setattr_dir(znode_t *dzp)
{
struct inode *dxip = ZTOI(dzp);
struct inode *xip = NULL;
zfsvfs_t *zfsvfs = ZTOZSB(dzp);
objset_t *os = zfsvfs->z_os;
zap_cursor_t zc;
zap_attribute_t zap;
zfs_dirlock_t *dl;
znode_t *zp = NULL;
dmu_tx_t *tx = NULL;
uint64_t uid, gid;
sa_bulk_attr_t bulk[4];
int count;
int err;
zap_cursor_init(&zc, os, dzp->z_id);
while ((err = zap_cursor_retrieve(&zc, &zap)) == 0) {
count = 0;
if (zap.za_integer_length != 8 || zap.za_num_integers != 1) {
err = ENXIO;
break;
}
err = zfs_dirent_lock(&dl, dzp, (char *)zap.za_name, &zp,
ZEXISTS, NULL, NULL);
if (err == ENOENT)
goto next;
if (err)
break;
xip = ZTOI(zp);
if (KUID_TO_SUID(xip->i_uid) == KUID_TO_SUID(dxip->i_uid) &&
KGID_TO_SGID(xip->i_gid) == KGID_TO_SGID(dxip->i_gid) &&
zp->z_projid == dzp->z_projid)
goto next;
tx = dmu_tx_create(os);
if (!(zp->z_pflags & ZFS_PROJID))
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
else
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
err = dmu_tx_assign(tx, TXG_WAIT);
if (err)
break;
mutex_enter(&dzp->z_lock);
if (KUID_TO_SUID(xip->i_uid) != KUID_TO_SUID(dxip->i_uid)) {
xip->i_uid = dxip->i_uid;
uid = zfs_uid_read(dxip);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
&uid, sizeof (uid));
}
if (KGID_TO_SGID(xip->i_gid) != KGID_TO_SGID(dxip->i_gid)) {
xip->i_gid = dxip->i_gid;
gid = zfs_gid_read(dxip);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs), NULL,
&gid, sizeof (gid));
}
if (zp->z_projid != dzp->z_projid) {
if (!(zp->z_pflags & ZFS_PROJID)) {
zp->z_pflags |= ZFS_PROJID;
SA_ADD_BULK_ATTR(bulk, count,
SA_ZPL_FLAGS(zfsvfs), NULL, &zp->z_pflags,
sizeof (zp->z_pflags));
}
zp->z_projid = dzp->z_projid;
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_PROJID(zfsvfs),
NULL, &zp->z_projid, sizeof (zp->z_projid));
}
mutex_exit(&dzp->z_lock);
if (likely(count > 0)) {
err = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
dmu_tx_commit(tx);
} else {
dmu_tx_abort(tx);
}
tx = NULL;
if (err != 0 && err != ENOENT)
break;
next:
if (zp) {
zrele(zp);
zp = NULL;
zfs_dirent_unlock(dl);
}
zap_cursor_advance(&zc);
}
if (tx)
dmu_tx_abort(tx);
if (zp) {
zrele(zp);
zfs_dirent_unlock(dl);
}
zap_cursor_fini(&zc);
return (err == ENOENT ? 0 : err);
}
/*
* Set the file attributes to the values contained in the
* vattr structure.
*
* IN: zp - znode of file to be modified.
* vap - new attribute values.
* If ATTR_XVATTR set, then optional attrs are being set
* flags - ATTR_UTIME set if non-default time values provided.
* - ATTR_NOACLCHECK (CIFS context only).
* cr - credentials of caller.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
* ip - ctime updated, mtime updated if size changed.
*/
/* ARGSUSED */
int
zfs_setattr(znode_t *zp, vattr_t *vap, int flags, cred_t *cr)
{
struct inode *ip;
zfsvfs_t *zfsvfs = ZTOZSB(zp);
objset_t *os = zfsvfs->z_os;
zilog_t *zilog;
dmu_tx_t *tx;
vattr_t oldva;
xvattr_t *tmpxvattr;
uint_t mask = vap->va_mask;
uint_t saved_mask = 0;
int trim_mask = 0;
uint64_t new_mode;
uint64_t new_kuid = 0, new_kgid = 0, new_uid, new_gid;
uint64_t xattr_obj;
uint64_t mtime[2], ctime[2], atime[2];
uint64_t projid = ZFS_INVALID_PROJID;
znode_t *attrzp;
int need_policy = FALSE;
int err, err2 = 0;
zfs_fuid_info_t *fuidp = NULL;
xvattr_t *xvap = (xvattr_t *)vap; /* vap may be an xvattr_t * */
xoptattr_t *xoap;
zfs_acl_t *aclp;
boolean_t skipaclchk = (flags & ATTR_NOACLCHECK) ? B_TRUE : B_FALSE;
boolean_t fuid_dirtied = B_FALSE;
boolean_t handle_eadir = B_FALSE;
sa_bulk_attr_t *bulk, *xattr_bulk;
int count = 0, xattr_count = 0, bulks = 8;
if (mask == 0)
return (0);
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
ip = ZTOI(zp);
/*
* If this is a xvattr_t, then get a pointer to the structure of
* optional attributes. If this is NULL, then we have a vattr_t.
*/
xoap = xva_getxoptattr(xvap);
if (xoap != NULL && (mask & ATTR_XVATTR)) {
if (XVA_ISSET_REQ(xvap, XAT_PROJID)) {
if (!dmu_objset_projectquota_enabled(os) ||
(!S_ISREG(ip->i_mode) && !S_ISDIR(ip->i_mode))) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENOTSUP));
}
projid = xoap->xoa_projid;
if (unlikely(projid == ZFS_INVALID_PROJID)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
if (projid == zp->z_projid && zp->z_pflags & ZFS_PROJID)
projid = ZFS_INVALID_PROJID;
else
need_policy = TRUE;
}
if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT) &&
(xoap->xoa_projinherit !=
((zp->z_pflags & ZFS_PROJINHERIT) != 0)) &&
(!dmu_objset_projectquota_enabled(os) ||
(!S_ISREG(ip->i_mode) && !S_ISDIR(ip->i_mode)))) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENOTSUP));
}
}
zilog = zfsvfs->z_log;
/*
* Make sure that if we have ephemeral uid/gid or xvattr specified
* that file system is at proper version level
*/
if (zfsvfs->z_use_fuids == B_FALSE &&
(((mask & ATTR_UID) && IS_EPHEMERAL(vap->va_uid)) ||
((mask & ATTR_GID) && IS_EPHEMERAL(vap->va_gid)) ||
(mask & ATTR_XVATTR))) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
if (mask & ATTR_SIZE && S_ISDIR(ip->i_mode)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EISDIR));
}
if (mask & ATTR_SIZE && !S_ISREG(ip->i_mode) && !S_ISFIFO(ip->i_mode)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
tmpxvattr = kmem_alloc(sizeof (xvattr_t), KM_SLEEP);
xva_init(tmpxvattr);
bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * bulks, KM_SLEEP);
xattr_bulk = kmem_alloc(sizeof (sa_bulk_attr_t) * bulks, KM_SLEEP);
/*
* Immutable files can only alter immutable bit and atime
*/
if ((zp->z_pflags & ZFS_IMMUTABLE) &&
((mask & (ATTR_SIZE|ATTR_UID|ATTR_GID|ATTR_MTIME|ATTR_MODE)) ||
((mask & ATTR_XVATTR) && XVA_ISSET_REQ(xvap, XAT_CREATETIME)))) {
err = SET_ERROR(EPERM);
goto out3;
}
if ((mask & ATTR_SIZE) && (zp->z_pflags & ZFS_READONLY)) {
err = SET_ERROR(EPERM);
goto out3;
}
/*
* Verify timestamps doesn't overflow 32 bits.
* ZFS can handle large timestamps, but 32bit syscalls can't
* handle times greater than 2039. This check should be removed
* once large timestamps are fully supported.
*/
if (mask & (ATTR_ATIME | ATTR_MTIME)) {
if (((mask & ATTR_ATIME) &&
TIMESPEC_OVERFLOW(&vap->va_atime)) ||
((mask & ATTR_MTIME) &&
TIMESPEC_OVERFLOW(&vap->va_mtime))) {
err = SET_ERROR(EOVERFLOW);
goto out3;
}
}
top:
attrzp = NULL;
aclp = NULL;
/* Can this be moved to before the top label? */
if (zfs_is_readonly(zfsvfs)) {
err = SET_ERROR(EROFS);
goto out3;
}
/*
* First validate permissions
*/
if (mask & ATTR_SIZE) {
err = zfs_zaccess(zp, ACE_WRITE_DATA, 0, skipaclchk, cr);
if (err)
goto out3;
/*
* XXX - Note, we are not providing any open
* mode flags here (like FNDELAY), so we may
* block if there are locks present... this
* should be addressed in openat().
*/
/* XXX - would it be OK to generate a log record here? */
err = zfs_freesp(zp, vap->va_size, 0, 0, FALSE);
if (err)
goto out3;
}
if (mask & (ATTR_ATIME|ATTR_MTIME) ||
((mask & ATTR_XVATTR) && (XVA_ISSET_REQ(xvap, XAT_HIDDEN) ||
XVA_ISSET_REQ(xvap, XAT_READONLY) ||
XVA_ISSET_REQ(xvap, XAT_ARCHIVE) ||
XVA_ISSET_REQ(xvap, XAT_OFFLINE) ||
XVA_ISSET_REQ(xvap, XAT_SPARSE) ||
XVA_ISSET_REQ(xvap, XAT_CREATETIME) ||
XVA_ISSET_REQ(xvap, XAT_SYSTEM)))) {
need_policy = zfs_zaccess(zp, ACE_WRITE_ATTRIBUTES, 0,
skipaclchk, cr);
}
if (mask & (ATTR_UID|ATTR_GID)) {
int idmask = (mask & (ATTR_UID|ATTR_GID));
int take_owner;
int take_group;
/*
* NOTE: even if a new mode is being set,
* we may clear S_ISUID/S_ISGID bits.
*/
if (!(mask & ATTR_MODE))
vap->va_mode = zp->z_mode;
/*
* Take ownership or chgrp to group we are a member of
*/
take_owner = (mask & ATTR_UID) && (vap->va_uid == crgetuid(cr));
take_group = (mask & ATTR_GID) &&
zfs_groupmember(zfsvfs, vap->va_gid, cr);
/*
* If both ATTR_UID and ATTR_GID are set then take_owner and
* take_group must both be set in order to allow taking
* ownership.
*
* Otherwise, send the check through secpolicy_vnode_setattr()
*
*/
if (((idmask == (ATTR_UID|ATTR_GID)) &&
take_owner && take_group) ||
((idmask == ATTR_UID) && take_owner) ||
((idmask == ATTR_GID) && take_group)) {
if (zfs_zaccess(zp, ACE_WRITE_OWNER, 0,
skipaclchk, cr) == 0) {
/*
* Remove setuid/setgid for non-privileged users
*/
(void) secpolicy_setid_clear(vap, cr);
trim_mask = (mask & (ATTR_UID|ATTR_GID));
} else {
need_policy = TRUE;
}
} else {
need_policy = TRUE;
}
}
mutex_enter(&zp->z_lock);
oldva.va_mode = zp->z_mode;
zfs_fuid_map_ids(zp, cr, &oldva.va_uid, &oldva.va_gid);
if (mask & ATTR_XVATTR) {
/*
* Update xvattr mask to include only those attributes
* that are actually changing.
*
* the bits will be restored prior to actually setting
* the attributes so the caller thinks they were set.
*/
if (XVA_ISSET_REQ(xvap, XAT_APPENDONLY)) {
if (xoap->xoa_appendonly !=
((zp->z_pflags & ZFS_APPENDONLY) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_APPENDONLY);
XVA_SET_REQ(tmpxvattr, XAT_APPENDONLY);
}
}
if (XVA_ISSET_REQ(xvap, XAT_PROJINHERIT)) {
if (xoap->xoa_projinherit !=
((zp->z_pflags & ZFS_PROJINHERIT) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_PROJINHERIT);
XVA_SET_REQ(tmpxvattr, XAT_PROJINHERIT);
}
}
if (XVA_ISSET_REQ(xvap, XAT_NOUNLINK)) {
if (xoap->xoa_nounlink !=
((zp->z_pflags & ZFS_NOUNLINK) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_NOUNLINK);
XVA_SET_REQ(tmpxvattr, XAT_NOUNLINK);
}
}
if (XVA_ISSET_REQ(xvap, XAT_IMMUTABLE)) {
if (xoap->xoa_immutable !=
((zp->z_pflags & ZFS_IMMUTABLE) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_IMMUTABLE);
XVA_SET_REQ(tmpxvattr, XAT_IMMUTABLE);
}
}
if (XVA_ISSET_REQ(xvap, XAT_NODUMP)) {
if (xoap->xoa_nodump !=
((zp->z_pflags & ZFS_NODUMP) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_NODUMP);
XVA_SET_REQ(tmpxvattr, XAT_NODUMP);
}
}
if (XVA_ISSET_REQ(xvap, XAT_AV_MODIFIED)) {
if (xoap->xoa_av_modified !=
((zp->z_pflags & ZFS_AV_MODIFIED) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_AV_MODIFIED);
XVA_SET_REQ(tmpxvattr, XAT_AV_MODIFIED);
}
}
if (XVA_ISSET_REQ(xvap, XAT_AV_QUARANTINED)) {
if ((!S_ISREG(ip->i_mode) &&
xoap->xoa_av_quarantined) ||
xoap->xoa_av_quarantined !=
((zp->z_pflags & ZFS_AV_QUARANTINED) != 0)) {
need_policy = TRUE;
} else {
XVA_CLR_REQ(xvap, XAT_AV_QUARANTINED);
XVA_SET_REQ(tmpxvattr, XAT_AV_QUARANTINED);
}
}
if (XVA_ISSET_REQ(xvap, XAT_REPARSE)) {
mutex_exit(&zp->z_lock);
err = SET_ERROR(EPERM);
goto out3;
}
if (need_policy == FALSE &&
(XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP) ||
XVA_ISSET_REQ(xvap, XAT_OPAQUE))) {
need_policy = TRUE;
}
}
mutex_exit(&zp->z_lock);
if (mask & ATTR_MODE) {
if (zfs_zaccess(zp, ACE_WRITE_ACL, 0, skipaclchk, cr) == 0) {
err = secpolicy_setid_setsticky_clear(ip, vap,
&oldva, cr);
if (err)
goto out3;
trim_mask |= ATTR_MODE;
} else {
need_policy = TRUE;
}
}
if (need_policy) {
/*
* If trim_mask is set then take ownership
* has been granted or write_acl is present and user
* has the ability to modify mode. In that case remove
* UID|GID and or MODE from mask so that
* secpolicy_vnode_setattr() doesn't revoke it.
*/
if (trim_mask) {
saved_mask = vap->va_mask;
vap->va_mask &= ~trim_mask;
}
err = secpolicy_vnode_setattr(cr, ip, vap, &oldva, flags,
(int (*)(void *, int, cred_t *))zfs_zaccess_unix, zp);
if (err)
goto out3;
if (trim_mask)
vap->va_mask |= saved_mask;
}
/*
* secpolicy_vnode_setattr, or take ownership may have
* changed va_mask
*/
mask = vap->va_mask;
if ((mask & (ATTR_UID | ATTR_GID)) || projid != ZFS_INVALID_PROJID) {
handle_eadir = B_TRUE;
err = sa_lookup(zp->z_sa_hdl, SA_ZPL_XATTR(zfsvfs),
&xattr_obj, sizeof (xattr_obj));
if (err == 0 && xattr_obj) {
err = zfs_zget(ZTOZSB(zp), xattr_obj, &attrzp);
if (err)
goto out2;
}
if (mask & ATTR_UID) {
new_kuid = zfs_fuid_create(zfsvfs,
(uint64_t)vap->va_uid, cr, ZFS_OWNER, &fuidp);
if (new_kuid != KUID_TO_SUID(ZTOI(zp)->i_uid) &&
zfs_id_overquota(zfsvfs, DMU_USERUSED_OBJECT,
new_kuid)) {
if (attrzp)
zrele(attrzp);
err = SET_ERROR(EDQUOT);
goto out2;
}
}
if (mask & ATTR_GID) {
new_kgid = zfs_fuid_create(zfsvfs,
(uint64_t)vap->va_gid, cr, ZFS_GROUP, &fuidp);
if (new_kgid != KGID_TO_SGID(ZTOI(zp)->i_gid) &&
zfs_id_overquota(zfsvfs, DMU_GROUPUSED_OBJECT,
new_kgid)) {
if (attrzp)
zrele(attrzp);
err = SET_ERROR(EDQUOT);
goto out2;
}
}
if (projid != ZFS_INVALID_PROJID &&
zfs_id_overquota(zfsvfs, DMU_PROJECTUSED_OBJECT, projid)) {
if (attrzp)
zrele(attrzp);
err = EDQUOT;
goto out2;
}
}
tx = dmu_tx_create(os);
if (mask & ATTR_MODE) {
uint64_t pmode = zp->z_mode;
uint64_t acl_obj;
new_mode = (pmode & S_IFMT) | (vap->va_mode & ~S_IFMT);
if (ZTOZSB(zp)->z_acl_mode == ZFS_ACL_RESTRICTED &&
!(zp->z_pflags & ZFS_ACL_TRIVIAL)) {
err = EPERM;
goto out;
}
if ((err = zfs_acl_chmod_setattr(zp, &aclp, new_mode)))
goto out;
mutex_enter(&zp->z_lock);
if (!zp->z_is_sa && ((acl_obj = zfs_external_acl(zp)) != 0)) {
/*
* Are we upgrading ACL from old V0 format
* to V1 format?
*/
if (zfsvfs->z_version >= ZPL_VERSION_FUID &&
zfs_znode_acl_version(zp) ==
ZFS_ACL_VERSION_INITIAL) {
dmu_tx_hold_free(tx, acl_obj, 0,
DMU_OBJECT_END);
dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
0, aclp->z_acl_bytes);
} else {
dmu_tx_hold_write(tx, acl_obj, 0,
aclp->z_acl_bytes);
}
} else if (!zp->z_is_sa && aclp->z_acl_bytes > ZFS_ACE_SPACE) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT,
0, aclp->z_acl_bytes);
}
mutex_exit(&zp->z_lock);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
} else {
if (((mask & ATTR_XVATTR) &&
XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP)) ||
(projid != ZFS_INVALID_PROJID &&
!(zp->z_pflags & ZFS_PROJID)))
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_TRUE);
else
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
}
if (attrzp) {
dmu_tx_hold_sa(tx, attrzp->z_sa_hdl, B_FALSE);
}
fuid_dirtied = zfsvfs->z_fuid_dirty;
if (fuid_dirtied)
zfs_fuid_txhold(zfsvfs, tx);
zfs_sa_upgrade_txholds(tx, zp);
err = dmu_tx_assign(tx, TXG_WAIT);
if (err)
goto out;
count = 0;
/*
* Set each attribute requested.
* We group settings according to the locks they need to acquire.
*
* Note: you cannot set ctime directly, although it will be
* updated as a side-effect of calling this function.
*/
if (projid != ZFS_INVALID_PROJID && !(zp->z_pflags & ZFS_PROJID)) {
/*
* For the existed object that is upgraded from old system,
* its on-disk layout has no slot for the project ID attribute.
* But quota accounting logic needs to access related slots by
* offset directly. So we need to adjust old objects' layout
* to make the project ID to some unified and fixed offset.
*/
if (attrzp)
err = sa_add_projid(attrzp->z_sa_hdl, tx, projid);
if (err == 0)
err = sa_add_projid(zp->z_sa_hdl, tx, projid);
if (unlikely(err == EEXIST))
err = 0;
else if (err != 0)
goto out;
else
projid = ZFS_INVALID_PROJID;
}
if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE))
mutex_enter(&zp->z_acl_lock);
mutex_enter(&zp->z_lock);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_FLAGS(zfsvfs), NULL,
&zp->z_pflags, sizeof (zp->z_pflags));
if (attrzp) {
if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE))
mutex_enter(&attrzp->z_acl_lock);
mutex_enter(&attrzp->z_lock);
SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
SA_ZPL_FLAGS(zfsvfs), NULL, &attrzp->z_pflags,
sizeof (attrzp->z_pflags));
if (projid != ZFS_INVALID_PROJID) {
attrzp->z_projid = projid;
SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
SA_ZPL_PROJID(zfsvfs), NULL, &attrzp->z_projid,
sizeof (attrzp->z_projid));
}
}
if (mask & (ATTR_UID|ATTR_GID)) {
if (mask & ATTR_UID) {
ZTOI(zp)->i_uid = SUID_TO_KUID(new_kuid);
new_uid = zfs_uid_read(ZTOI(zp));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_UID(zfsvfs), NULL,
&new_uid, sizeof (new_uid));
if (attrzp) {
SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
SA_ZPL_UID(zfsvfs), NULL, &new_uid,
sizeof (new_uid));
ZTOI(attrzp)->i_uid = SUID_TO_KUID(new_uid);
}
}
if (mask & ATTR_GID) {
ZTOI(zp)->i_gid = SGID_TO_KGID(new_kgid);
new_gid = zfs_gid_read(ZTOI(zp));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GID(zfsvfs),
NULL, &new_gid, sizeof (new_gid));
if (attrzp) {
SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
SA_ZPL_GID(zfsvfs), NULL, &new_gid,
sizeof (new_gid));
ZTOI(attrzp)->i_gid = SGID_TO_KGID(new_kgid);
}
}
if (!(mask & ATTR_MODE)) {
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs),
NULL, &new_mode, sizeof (new_mode));
new_mode = zp->z_mode;
}
err = zfs_acl_chown_setattr(zp);
ASSERT(err == 0);
if (attrzp) {
err = zfs_acl_chown_setattr(attrzp);
ASSERT(err == 0);
}
}
if (mask & ATTR_MODE) {
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MODE(zfsvfs), NULL,
&new_mode, sizeof (new_mode));
zp->z_mode = ZTOI(zp)->i_mode = new_mode;
ASSERT3P(aclp, !=, NULL);
err = zfs_aclset_common(zp, aclp, cr, tx);
ASSERT0(err);
if (zp->z_acl_cached)
zfs_acl_free(zp->z_acl_cached);
zp->z_acl_cached = aclp;
aclp = NULL;
}
if ((mask & ATTR_ATIME) || zp->z_atime_dirty) {
zp->z_atime_dirty = B_FALSE;
- ZFS_TIME_ENCODE(&ip->i_atime, atime);
+ inode_timespec_t tmp_atime;
+ ZFS_TIME_ENCODE(&tmp_atime, atime);
+ zpl_inode_set_atime_to_ts(ZTOI(zp), tmp_atime);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_ATIME(zfsvfs), NULL,
&atime, sizeof (atime));
}
if (mask & (ATTR_MTIME | ATTR_SIZE)) {
ZFS_TIME_ENCODE(&vap->va_mtime, mtime);
- ZTOI(zp)->i_mtime = zpl_inode_timestamp_truncate(
- vap->va_mtime, ZTOI(zp));
+ zpl_inode_set_mtime_to_ts(ZTOI(zp),
+ zpl_inode_timestamp_truncate(vap->va_mtime, ZTOI(zp)));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_MTIME(zfsvfs), NULL,
mtime, sizeof (mtime));
}
if (mask & (ATTR_CTIME | ATTR_SIZE)) {
ZFS_TIME_ENCODE(&vap->va_ctime, ctime);
zpl_inode_set_ctime_to_ts(ZTOI(zp),
zpl_inode_timestamp_truncate(vap->va_ctime, ZTOI(zp)));
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CTIME(zfsvfs), NULL,
ctime, sizeof (ctime));
}
if (projid != ZFS_INVALID_PROJID) {
zp->z_projid = projid;
SA_ADD_BULK_ATTR(bulk, count,
SA_ZPL_PROJID(zfsvfs), NULL, &zp->z_projid,
sizeof (zp->z_projid));
}
if (attrzp && mask) {
SA_ADD_BULK_ATTR(xattr_bulk, xattr_count,
SA_ZPL_CTIME(zfsvfs), NULL, &ctime,
sizeof (ctime));
}
/*
* Do this after setting timestamps to prevent timestamp
* update from toggling bit
*/
if (xoap && (mask & ATTR_XVATTR)) {
/*
* restore trimmed off masks
* so that return masks can be set for caller.
*/
if (XVA_ISSET_REQ(tmpxvattr, XAT_APPENDONLY)) {
XVA_SET_REQ(xvap, XAT_APPENDONLY);
}
if (XVA_ISSET_REQ(tmpxvattr, XAT_NOUNLINK)) {
XVA_SET_REQ(xvap, XAT_NOUNLINK);
}
if (XVA_ISSET_REQ(tmpxvattr, XAT_IMMUTABLE)) {
XVA_SET_REQ(xvap, XAT_IMMUTABLE);
}
if (XVA_ISSET_REQ(tmpxvattr, XAT_NODUMP)) {
XVA_SET_REQ(xvap, XAT_NODUMP);
}
if (XVA_ISSET_REQ(tmpxvattr, XAT_AV_MODIFIED)) {
XVA_SET_REQ(xvap, XAT_AV_MODIFIED);
}
if (XVA_ISSET_REQ(tmpxvattr, XAT_AV_QUARANTINED)) {
XVA_SET_REQ(xvap, XAT_AV_QUARANTINED);
}
if (XVA_ISSET_REQ(tmpxvattr, XAT_PROJINHERIT)) {
XVA_SET_REQ(xvap, XAT_PROJINHERIT);
}
if (XVA_ISSET_REQ(xvap, XAT_AV_SCANSTAMP))
ASSERT(S_ISREG(ip->i_mode));
zfs_xvattr_set(zp, xvap, tx);
}
if (fuid_dirtied)
zfs_fuid_sync(zfsvfs, tx);
if (mask != 0)
zfs_log_setattr(zilog, tx, TX_SETATTR, zp, vap, mask, fuidp);
mutex_exit(&zp->z_lock);
if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE))
mutex_exit(&zp->z_acl_lock);
if (attrzp) {
if (mask & (ATTR_UID|ATTR_GID|ATTR_MODE))
mutex_exit(&attrzp->z_acl_lock);
mutex_exit(&attrzp->z_lock);
}
out:
if (err == 0 && xattr_count > 0) {
err2 = sa_bulk_update(attrzp->z_sa_hdl, xattr_bulk,
xattr_count, tx);
ASSERT(err2 == 0);
}
if (aclp)
zfs_acl_free(aclp);
if (fuidp) {
zfs_fuid_info_free(fuidp);
fuidp = NULL;
}
if (err) {
dmu_tx_abort(tx);
if (attrzp)
zrele(attrzp);
if (err == ERESTART)
goto top;
} else {
if (count > 0)
err2 = sa_bulk_update(zp->z_sa_hdl, bulk, count, tx);
dmu_tx_commit(tx);
if (attrzp) {
if (err2 == 0 && handle_eadir)
err2 = zfs_setattr_dir(attrzp);
zrele(attrzp);
}
zfs_znode_update_vfs(zp);
}
out2:
if (os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
out3:
kmem_free(xattr_bulk, sizeof (sa_bulk_attr_t) * bulks);
kmem_free(bulk, sizeof (sa_bulk_attr_t) * bulks);
kmem_free(tmpxvattr, sizeof (xvattr_t));
ZFS_EXIT(zfsvfs);
return (err);
}
typedef struct zfs_zlock {
krwlock_t *zl_rwlock; /* lock we acquired */
znode_t *zl_znode; /* znode we held */
struct zfs_zlock *zl_next; /* next in list */
} zfs_zlock_t;
/*
* Drop locks and release vnodes that were held by zfs_rename_lock().
*/
static void
zfs_rename_unlock(zfs_zlock_t **zlpp)
{
zfs_zlock_t *zl;
while ((zl = *zlpp) != NULL) {
if (zl->zl_znode != NULL)
zfs_zrele_async(zl->zl_znode);
rw_exit(zl->zl_rwlock);
*zlpp = zl->zl_next;
kmem_free(zl, sizeof (*zl));
}
}
/*
* Search back through the directory tree, using the ".." entries.
* Lock each directory in the chain to prevent concurrent renames.
* Fail any attempt to move a directory into one of its own descendants.
* XXX - z_parent_lock can overlap with map or grow locks
*/
static int
zfs_rename_lock(znode_t *szp, znode_t *tdzp, znode_t *sdzp, zfs_zlock_t **zlpp)
{
zfs_zlock_t *zl;
znode_t *zp = tdzp;
uint64_t rootid = ZTOZSB(zp)->z_root;
uint64_t oidp = zp->z_id;
krwlock_t *rwlp = &szp->z_parent_lock;
krw_t rw = RW_WRITER;
/*
* First pass write-locks szp and compares to zp->z_id.
* Later passes read-lock zp and compare to zp->z_parent.
*/
do {
if (!rw_tryenter(rwlp, rw)) {
/*
* Another thread is renaming in this path.
* Note that if we are a WRITER, we don't have any
* parent_locks held yet.
*/
if (rw == RW_READER && zp->z_id > szp->z_id) {
/*
* Drop our locks and restart
*/
zfs_rename_unlock(&zl);
*zlpp = NULL;
zp = tdzp;
oidp = zp->z_id;
rwlp = &szp->z_parent_lock;
rw = RW_WRITER;
continue;
} else {
/*
* Wait for other thread to drop its locks
*/
rw_enter(rwlp, rw);
}
}
zl = kmem_alloc(sizeof (*zl), KM_SLEEP);
zl->zl_rwlock = rwlp;
zl->zl_znode = NULL;
zl->zl_next = *zlpp;
*zlpp = zl;
if (oidp == szp->z_id) /* We're a descendant of szp */
return (SET_ERROR(EINVAL));
if (oidp == rootid) /* We've hit the top */
return (0);
if (rw == RW_READER) { /* i.e. not the first pass */
int error = zfs_zget(ZTOZSB(zp), oidp, &zp);
if (error)
return (error);
zl->zl_znode = zp;
}
(void) sa_lookup(zp->z_sa_hdl, SA_ZPL_PARENT(ZTOZSB(zp)),
&oidp, sizeof (oidp));
rwlp = &zp->z_parent_lock;
rw = RW_READER;
} while (zp->z_id != sdzp->z_id);
return (0);
}
/*
* Move an entry from the provided source directory to the target
* directory. Change the entry name as indicated.
*
* IN: sdzp - Source directory containing the "old entry".
* snm - Old entry name.
* tdzp - Target directory to contain the "new entry".
* tnm - New entry name.
* cr - credentials of caller.
* flags - case flags
*
* RETURN: 0 on success, error code on failure.
*
* Timestamps:
* sdzp,tdzp - ctime|mtime updated
*/
/*ARGSUSED*/
int
zfs_rename(znode_t *sdzp, char *snm, znode_t *tdzp, char *tnm,
cred_t *cr, int flags)
{
znode_t *szp, *tzp;
zfsvfs_t *zfsvfs = ZTOZSB(sdzp);
zilog_t *zilog;
zfs_dirlock_t *sdl, *tdl;
dmu_tx_t *tx;
zfs_zlock_t *zl;
int cmp, serr, terr;
int error = 0;
int zflg = 0;
boolean_t waited = B_FALSE;
if (snm == NULL || tnm == NULL)
return (SET_ERROR(EINVAL));
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(sdzp);
zilog = zfsvfs->z_log;
ZFS_VERIFY_ZP(tdzp);
/*
* We check i_sb because snapshots and the ctldir must have different
* super blocks.
*/
if (ZTOI(tdzp)->i_sb != ZTOI(sdzp)->i_sb ||
zfsctl_is_node(ZTOI(tdzp))) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EXDEV));
}
if (zfsvfs->z_utf8 && u8_validate(tnm,
strlen(tnm), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
if (flags & FIGNORECASE)
zflg |= ZCILOOK;
top:
szp = NULL;
tzp = NULL;
zl = NULL;
/*
* This is to prevent the creation of links into attribute space
* by renaming a linked file into/outof an attribute directory.
* See the comment in zfs_link() for why this is considered bad.
*/
if ((tdzp->z_pflags & ZFS_XATTR) != (sdzp->z_pflags & ZFS_XATTR)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
/*
* Lock source and target directory entries. To prevent deadlock,
* a lock ordering must be defined. We lock the directory with
* the smallest object id first, or if it's a tie, the one with
* the lexically first name.
*/
if (sdzp->z_id < tdzp->z_id) {
cmp = -1;
} else if (sdzp->z_id > tdzp->z_id) {
cmp = 1;
} else {
/*
* First compare the two name arguments without
* considering any case folding.
*/
int nofold = (zfsvfs->z_norm & ~U8_TEXTPREP_TOUPPER);
cmp = u8_strcmp(snm, tnm, 0, nofold, U8_UNICODE_LATEST, &error);
ASSERT(error == 0 || !zfsvfs->z_utf8);
if (cmp == 0) {
/*
* POSIX: "If the old argument and the new argument
* both refer to links to the same existing file,
* the rename() function shall return successfully
* and perform no other action."
*/
ZFS_EXIT(zfsvfs);
return (0);
}
/*
* If the file system is case-folding, then we may
* have some more checking to do. A case-folding file
* system is either supporting mixed case sensitivity
* access or is completely case-insensitive. Note
* that the file system is always case preserving.
*
* In mixed sensitivity mode case sensitive behavior
* is the default. FIGNORECASE must be used to
* explicitly request case insensitive behavior.
*
* If the source and target names provided differ only
* by case (e.g., a request to rename 'tim' to 'Tim'),
* we will treat this as a special case in the
* case-insensitive mode: as long as the source name
* is an exact match, we will allow this to proceed as
* a name-change request.
*/
if ((zfsvfs->z_case == ZFS_CASE_INSENSITIVE ||
(zfsvfs->z_case == ZFS_CASE_MIXED &&
flags & FIGNORECASE)) &&
u8_strcmp(snm, tnm, 0, zfsvfs->z_norm, U8_UNICODE_LATEST,
&error) == 0) {
/*
* case preserving rename request, require exact
* name matches
*/
zflg |= ZCIEXACT;
zflg &= ~ZCILOOK;
}
}
/*
* If the source and destination directories are the same, we should
* grab the z_name_lock of that directory only once.
*/
if (sdzp == tdzp) {
zflg |= ZHAVELOCK;
rw_enter(&sdzp->z_name_lock, RW_READER);
}
if (cmp < 0) {
serr = zfs_dirent_lock(&sdl, sdzp, snm, &szp,
ZEXISTS | zflg, NULL, NULL);
terr = zfs_dirent_lock(&tdl,
tdzp, tnm, &tzp, ZRENAMING | zflg, NULL, NULL);
} else {
terr = zfs_dirent_lock(&tdl,
tdzp, tnm, &tzp, zflg, NULL, NULL);
serr = zfs_dirent_lock(&sdl,
sdzp, snm, &szp, ZEXISTS | ZRENAMING | zflg,
NULL, NULL);
}
if (serr) {
/*
* Source entry invalid or not there.
*/
if (!terr) {
zfs_dirent_unlock(tdl);
if (tzp)
zrele(tzp);
}
if (sdzp == tdzp)
rw_exit(&sdzp->z_name_lock);
if (strcmp(snm, "..") == 0)
serr = EINVAL;
ZFS_EXIT(zfsvfs);
return (serr);
}
if (terr) {
zfs_dirent_unlock(sdl);
zrele(szp);
if (sdzp == tdzp)
rw_exit(&sdzp->z_name_lock);
if (strcmp(tnm, "..") == 0)
terr = EINVAL;
ZFS_EXIT(zfsvfs);
return (terr);
}
/*
* If we are using project inheritance, means if the directory has
* ZFS_PROJINHERIT set, then its descendant directories will inherit
* not only the project ID, but also the ZFS_PROJINHERIT flag. Under
* such case, we only allow renames into our tree when the project
* IDs are the same.
*/
if (tdzp->z_pflags & ZFS_PROJINHERIT &&
tdzp->z_projid != szp->z_projid) {
error = SET_ERROR(EXDEV);
goto out;
}
/*
* Must have write access at the source to remove the old entry
* and write access at the target to create the new entry.
* Note that if target and source are the same, this can be
* done in a single check.
*/
if ((error = zfs_zaccess_rename(sdzp, szp, tdzp, tzp, cr)))
goto out;
if (S_ISDIR(ZTOI(szp)->i_mode)) {
/*
* Check to make sure rename is valid.
* Can't do a move like this: /usr/a/b to /usr/a/b/c/d
*/
if ((error = zfs_rename_lock(szp, tdzp, sdzp, &zl)))
goto out;
}
/*
* Does target exist?
*/
if (tzp) {
/*
* Source and target must be the same type.
*/
if (S_ISDIR(ZTOI(szp)->i_mode)) {
if (!S_ISDIR(ZTOI(tzp)->i_mode)) {
error = SET_ERROR(ENOTDIR);
goto out;
}
} else {
if (S_ISDIR(ZTOI(tzp)->i_mode)) {
error = SET_ERROR(EISDIR);
goto out;
}
}
/*
* POSIX dictates that when the source and target
* entries refer to the same file object, rename
* must do nothing and exit without error.
*/
if (szp->z_id == tzp->z_id) {
error = 0;
goto out;
}
}
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
dmu_tx_hold_sa(tx, sdzp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, sdzp->z_id, FALSE, snm);
dmu_tx_hold_zap(tx, tdzp->z_id, TRUE, tnm);
if (sdzp != tdzp) {
dmu_tx_hold_sa(tx, tdzp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, tdzp);
}
if (tzp) {
dmu_tx_hold_sa(tx, tzp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, tzp);
}
zfs_sa_upgrade_txholds(tx, szp);
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
if (zl != NULL)
zfs_rename_unlock(&zl);
zfs_dirent_unlock(sdl);
zfs_dirent_unlock(tdl);
if (sdzp == tdzp)
rw_exit(&sdzp->z_name_lock);
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
zrele(szp);
if (tzp)
zrele(tzp);
goto top;
}
dmu_tx_abort(tx);
zrele(szp);
if (tzp)
zrele(tzp);
ZFS_EXIT(zfsvfs);
return (error);
}
if (tzp) /* Attempt to remove the existing target */
error = zfs_link_destroy(tdl, tzp, tx, zflg, NULL);
if (error == 0) {
error = zfs_link_create(tdl, szp, tx, ZRENAMING);
if (error == 0) {
szp->z_pflags |= ZFS_AV_MODIFIED;
if (tdzp->z_pflags & ZFS_PROJINHERIT)
szp->z_pflags |= ZFS_PROJINHERIT;
error = sa_update(szp->z_sa_hdl, SA_ZPL_FLAGS(zfsvfs),
(void *)&szp->z_pflags, sizeof (uint64_t), tx);
ASSERT0(error);
error = zfs_link_destroy(sdl, szp, tx, ZRENAMING, NULL);
if (error == 0) {
zfs_log_rename(zilog, tx, TX_RENAME |
(flags & FIGNORECASE ? TX_CI : 0), sdzp,
sdl->dl_name, tdzp, tdl->dl_name, szp);
} else {
/*
* At this point, we have successfully created
* the target name, but have failed to remove
* the source name. Since the create was done
* with the ZRENAMING flag, there are
* complications; for one, the link count is
* wrong. The easiest way to deal with this
* is to remove the newly created target, and
* return the original error. This must
* succeed; fortunately, it is very unlikely to
* fail, since we just created it.
*/
VERIFY3U(zfs_link_destroy(tdl, szp, tx,
ZRENAMING, NULL), ==, 0);
}
} else {
/*
* If we had removed the existing target, subsequent
* call to zfs_link_create() to add back the same entry
* but, the new dnode (szp) should not fail.
*/
ASSERT(tzp == NULL);
}
}
dmu_tx_commit(tx);
out:
if (zl != NULL)
zfs_rename_unlock(&zl);
zfs_dirent_unlock(sdl);
zfs_dirent_unlock(tdl);
zfs_znode_update_vfs(sdzp);
if (sdzp == tdzp)
rw_exit(&sdzp->z_name_lock);
if (sdzp != tdzp)
zfs_znode_update_vfs(tdzp);
zfs_znode_update_vfs(szp);
zrele(szp);
if (tzp) {
zfs_znode_update_vfs(tzp);
zrele(tzp);
}
if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Insert the indicated symbolic reference entry into the directory.
*
* IN: dzp - Directory to contain new symbolic link.
* name - Name of directory entry in dip.
* vap - Attributes of new entry.
* link - Name for new symlink entry.
* cr - credentials of caller.
* flags - case flags
*
* OUT: zpp - Znode for new symbolic link.
*
* RETURN: 0 on success, error code on failure.
*
* Timestamps:
* dip - ctime|mtime updated
*/
/*ARGSUSED*/
int
zfs_symlink(znode_t *dzp, char *name, vattr_t *vap, char *link,
znode_t **zpp, cred_t *cr, int flags)
{
znode_t *zp;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
zfsvfs_t *zfsvfs = ZTOZSB(dzp);
zilog_t *zilog;
uint64_t len = strlen(link);
int error;
int zflg = ZNEW;
zfs_acl_ids_t acl_ids;
boolean_t fuid_dirtied;
uint64_t txtype = TX_SYMLINK;
boolean_t waited = B_FALSE;
ASSERT(S_ISLNK(vap->va_mode));
if (name == NULL)
return (SET_ERROR(EINVAL));
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(dzp);
zilog = zfsvfs->z_log;
if (zfsvfs->z_utf8 && u8_validate(name, strlen(name),
NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
if (flags & FIGNORECASE)
zflg |= ZCILOOK;
if (len > MAXPATHLEN) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENAMETOOLONG));
}
if ((error = zfs_acl_ids_create(dzp, 0,
vap, cr, NULL, &acl_ids)) != 0) {
ZFS_EXIT(zfsvfs);
return (error);
}
top:
*zpp = NULL;
/*
* Attempt to lock directory; fail if entry already exists.
*/
error = zfs_dirent_lock(&dl, dzp, name, &zp, zflg, NULL, NULL);
if (error) {
zfs_acl_ids_free(&acl_ids);
ZFS_EXIT(zfsvfs);
return (error);
}
if ((error = zfs_zaccess(dzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
ZFS_EXIT(zfsvfs);
return (error);
}
if (zfs_acl_ids_overquota(zfsvfs, &acl_ids, ZFS_DEFAULT_PROJID)) {
zfs_acl_ids_free(&acl_ids);
zfs_dirent_unlock(dl);
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EDQUOT));
}
tx = dmu_tx_create(zfsvfs->z_os);
fuid_dirtied = zfsvfs->z_fuid_dirty;
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0, MAX(1, len));
dmu_tx_hold_zap(tx, dzp->z_id, TRUE, name);
dmu_tx_hold_sa_create(tx, acl_ids.z_aclp->z_acl_bytes +
ZFS_SA_BASE_ATTR_SIZE + len);
dmu_tx_hold_sa(tx, dzp->z_sa_hdl, B_FALSE);
if (!zfsvfs->z_use_sa && acl_ids.z_aclp->z_acl_bytes > ZFS_ACE_SPACE) {
dmu_tx_hold_write(tx, DMU_NEW_OBJECT, 0,
acl_ids.z_aclp->z_acl_bytes);
}
if (fuid_dirtied)
zfs_fuid_txhold(zfsvfs, tx);
error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
zfs_acl_ids_free(&acl_ids);
dmu_tx_abort(tx);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Create a new object for the symlink.
* for version 4 ZPL datasets the symlink will be an SA attribute
*/
zfs_mknode(dzp, vap, tx, cr, 0, &zp, &acl_ids);
if (fuid_dirtied)
zfs_fuid_sync(zfsvfs, tx);
mutex_enter(&zp->z_lock);
if (zp->z_is_sa)
error = sa_update(zp->z_sa_hdl, SA_ZPL_SYMLINK(zfsvfs),
link, len, tx);
else
zfs_sa_symlink(zp, link, len, tx);
mutex_exit(&zp->z_lock);
zp->z_size = len;
(void) sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(zfsvfs),
&zp->z_size, sizeof (zp->z_size), tx);
/*
* Insert the new object into the directory.
*/
error = zfs_link_create(dl, zp, tx, ZNEW);
if (error != 0) {
zfs_znode_delete(zp, tx);
remove_inode_hash(ZTOI(zp));
} else {
if (flags & FIGNORECASE)
txtype |= TX_CI;
zfs_log_symlink(zilog, tx, txtype, dzp, zp, name, link);
zfs_znode_update_vfs(dzp);
zfs_znode_update_vfs(zp);
}
zfs_acl_ids_free(&acl_ids);
dmu_tx_commit(tx);
zfs_dirent_unlock(dl);
if (error == 0) {
*zpp = zp;
if (zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
} else {
zrele(zp);
}
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Return, in the buffer contained in the provided uio structure,
* the symbolic path referred to by ip.
*
* IN: ip - inode of symbolic link
* uio - structure to contain the link path.
* cr - credentials of caller.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
* ip - atime updated
*/
/* ARGSUSED */
int
zfs_readlink(struct inode *ip, zfs_uio_t *uio, cred_t *cr)
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
int error;
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
mutex_enter(&zp->z_lock);
if (zp->z_is_sa)
error = sa_lookup_uio(zp->z_sa_hdl,
SA_ZPL_SYMLINK(zfsvfs), uio);
else
error = zfs_sa_readlink(zp, uio);
mutex_exit(&zp->z_lock);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Insert a new entry into directory tdzp referencing szp.
*
* IN: tdzp - Directory to contain new entry.
* szp - znode of new entry.
* name - name of new entry.
* cr - credentials of caller.
* flags - case flags.
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
* tdzp - ctime|mtime updated
* szp - ctime updated
*/
/* ARGSUSED */
int
zfs_link(znode_t *tdzp, znode_t *szp, char *name, cred_t *cr,
int flags)
{
struct inode *sip = ZTOI(szp);
znode_t *tzp;
zfsvfs_t *zfsvfs = ZTOZSB(tdzp);
zilog_t *zilog;
zfs_dirlock_t *dl;
dmu_tx_t *tx;
int error;
int zf = ZNEW;
uint64_t parent;
uid_t owner;
boolean_t waited = B_FALSE;
boolean_t is_tmpfile = 0;
uint64_t txg;
#ifdef HAVE_TMPFILE
is_tmpfile = (sip->i_nlink == 0 && (sip->i_state & I_LINKABLE));
#endif
ASSERT(S_ISDIR(ZTOI(tdzp)->i_mode));
if (name == NULL)
return (SET_ERROR(EINVAL));
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(tdzp);
zilog = zfsvfs->z_log;
/*
* POSIX dictates that we return EPERM here.
* Better choices include ENOTSUP or EISDIR.
*/
if (S_ISDIR(sip->i_mode)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
ZFS_VERIFY_ZP(szp);
/*
* If we are using project inheritance, means if the directory has
* ZFS_PROJINHERIT set, then its descendant directories will inherit
* not only the project ID, but also the ZFS_PROJINHERIT flag. Under
* such case, we only allow hard link creation in our tree when the
* project IDs are the same.
*/
if (tdzp->z_pflags & ZFS_PROJINHERIT &&
tdzp->z_projid != szp->z_projid) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EXDEV));
}
/*
* We check i_sb because snapshots and the ctldir must have different
* super blocks.
*/
if (sip->i_sb != ZTOI(tdzp)->i_sb || zfsctl_is_node(sip)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EXDEV));
}
/* Prevent links to .zfs/shares files */
if ((error = sa_lookup(szp->z_sa_hdl, SA_ZPL_PARENT(zfsvfs),
&parent, sizeof (uint64_t))) != 0) {
ZFS_EXIT(zfsvfs);
return (error);
}
if (parent == zfsvfs->z_shares_dir) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
if (zfsvfs->z_utf8 && u8_validate(name,
strlen(name), NULL, U8_VALIDATE_ENTIRE, &error) < 0) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EILSEQ));
}
if (flags & FIGNORECASE)
zf |= ZCILOOK;
/*
* We do not support links between attributes and non-attributes
* because of the potential security risk of creating links
* into "normal" file space in order to circumvent restrictions
* imposed in attribute space.
*/
if ((szp->z_pflags & ZFS_XATTR) != (tdzp->z_pflags & ZFS_XATTR)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
owner = zfs_fuid_map_id(zfsvfs, KUID_TO_SUID(sip->i_uid),
cr, ZFS_OWNER);
if (owner != crgetuid(cr) && secpolicy_basic_link(cr) != 0) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
if ((error = zfs_zaccess(tdzp, ACE_ADD_FILE, 0, B_FALSE, cr))) {
ZFS_EXIT(zfsvfs);
return (error);
}
top:
/*
* Attempt to lock directory; fail if entry already exists.
*/
error = zfs_dirent_lock(&dl, tdzp, name, &tzp, zf, NULL, NULL);
if (error) {
ZFS_EXIT(zfsvfs);
return (error);
}
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_sa(tx, szp->z_sa_hdl, B_FALSE);
dmu_tx_hold_zap(tx, tdzp->z_id, TRUE, name);
if (is_tmpfile)
dmu_tx_hold_zap(tx, zfsvfs->z_unlinkedobj, FALSE, NULL);
zfs_sa_upgrade_txholds(tx, szp);
zfs_sa_upgrade_txholds(tx, tdzp);
error = dmu_tx_assign(tx, (waited ? TXG_NOTHROTTLE : 0) | TXG_NOWAIT);
if (error) {
zfs_dirent_unlock(dl);
if (error == ERESTART) {
waited = B_TRUE;
dmu_tx_wait(tx);
dmu_tx_abort(tx);
goto top;
}
dmu_tx_abort(tx);
ZFS_EXIT(zfsvfs);
return (error);
}
/* unmark z_unlinked so zfs_link_create will not reject */
if (is_tmpfile)
szp->z_unlinked = B_FALSE;
error = zfs_link_create(dl, szp, tx, 0);
if (error == 0) {
uint64_t txtype = TX_LINK;
/*
* tmpfile is created to be in z_unlinkedobj, so remove it.
* Also, we don't log in ZIL, because all previous file
* operation on the tmpfile are ignored by ZIL. Instead we
* always wait for txg to sync to make sure all previous
* operation are sync safe.
*/
if (is_tmpfile) {
VERIFY(zap_remove_int(zfsvfs->z_os,
zfsvfs->z_unlinkedobj, szp->z_id, tx) == 0);
} else {
if (flags & FIGNORECASE)
txtype |= TX_CI;
zfs_log_link(zilog, tx, txtype, tdzp, szp, name);
}
} else if (is_tmpfile) {
/* restore z_unlinked since when linking failed */
szp->z_unlinked = B_TRUE;
}
txg = dmu_tx_get_txg(tx);
dmu_tx_commit(tx);
zfs_dirent_unlock(dl);
if (!is_tmpfile && zfsvfs->z_os->os_sync == ZFS_SYNC_ALWAYS)
zil_commit(zilog, 0);
if (is_tmpfile && zfsvfs->z_os->os_sync != ZFS_SYNC_DISABLED)
txg_wait_synced(dmu_objset_pool(zfsvfs->z_os), txg);
zfs_znode_update_vfs(tdzp);
zfs_znode_update_vfs(szp);
ZFS_EXIT(zfsvfs);
return (error);
}
static void
zfs_putpage_sync_commit_cb(void *arg)
{
struct page *pp = arg;
ClearPageError(pp);
end_page_writeback(pp);
}
static void
zfs_putpage_async_commit_cb(void *arg)
{
struct page *pp = arg;
znode_t *zp = ITOZ(pp->mapping->host);
ClearPageError(pp);
end_page_writeback(pp);
atomic_dec_32(&zp->z_async_writes_cnt);
}
/*
* Push a page out to disk, once the page is on stable storage the
* registered commit callback will be run as notification of completion.
*
* IN: ip - page mapped for inode.
* pp - page to push (page is locked)
* wbc - writeback control data
* for_sync - does the caller intend to wait synchronously for the
* page writeback to complete?
*
* RETURN: 0 if success
* error code if failure
*
* Timestamps:
* ip - ctime|mtime updated
*/
/* ARGSUSED */
int
zfs_putpage(struct inode *ip, struct page *pp, struct writeback_control *wbc,
boolean_t for_sync)
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
loff_t offset;
loff_t pgoff;
unsigned int pglen;
dmu_tx_t *tx;
caddr_t va;
int err = 0;
uint64_t mtime[2], ctime[2];
- inode_timespec_t tmp_ctime;
+ inode_timespec_t tmp_ts;
sa_bulk_attr_t bulk[3];
int cnt = 0;
struct address_space *mapping;
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
ASSERT(PageLocked(pp));
pgoff = page_offset(pp); /* Page byte-offset in file */
offset = i_size_read(ip); /* File length in bytes */
pglen = MIN(PAGE_SIZE, /* Page length in bytes */
P2ROUNDUP(offset, PAGE_SIZE)-pgoff);
/* Page is beyond end of file */
if (pgoff >= offset) {
unlock_page(pp);
ZFS_EXIT(zfsvfs);
return (0);
}
/* Truncate page length to end of file */
if (pgoff + pglen > offset)
pglen = offset - pgoff;
#if 0
/*
* FIXME: Allow mmap writes past its quota. The correct fix
* is to register a page_mkwrite() handler to count the page
* against its quota when it is about to be dirtied.
*/
if (zfs_id_overblockquota(zfsvfs, DMU_USERUSED_OBJECT,
KUID_TO_SUID(ip->i_uid)) ||
zfs_id_overblockquota(zfsvfs, DMU_GROUPUSED_OBJECT,
KGID_TO_SGID(ip->i_gid)) ||
(zp->z_projid != ZFS_DEFAULT_PROJID &&
zfs_id_overblockquota(zfsvfs, DMU_PROJECTUSED_OBJECT,
zp->z_projid))) {
err = EDQUOT;
}
#endif
/*
* The ordering here is critical and must adhere to the following
* rules in order to avoid deadlocking in either zfs_read() or
* zfs_free_range() due to a lock inversion.
*
* 1) The page must be unlocked prior to acquiring the range lock.
* This is critical because zfs_read() calls find_lock_page()
* which may block on the page lock while holding the range lock.
*
* 2) Before setting or clearing write back on a page the range lock
* must be held in order to prevent a lock inversion with the
* zfs_free_range() function.
*
* This presents a problem because upon entering this function the
* page lock is already held. To safely acquire the range lock the
* page lock must be dropped. This creates a window where another
* process could truncate, invalidate, dirty, or write out the page.
*
* Therefore, after successfully reacquiring the range and page locks
* the current page state is checked. In the common case everything
* will be as is expected and it can be written out. However, if
* the page state has changed it must be handled accordingly.
*/
mapping = pp->mapping;
redirty_page_for_writepage(wbc, pp);
unlock_page(pp);
zfs_locked_range_t *lr = zfs_rangelock_enter(&zp->z_rangelock,
pgoff, pglen, RL_WRITER);
lock_page(pp);
/* Page mapping changed or it was no longer dirty, we're done */
if (unlikely((mapping != pp->mapping) || !PageDirty(pp))) {
unlock_page(pp);
zfs_rangelock_exit(lr);
ZFS_EXIT(zfsvfs);
return (0);
}
/* Another process started write block if required */
if (PageWriteback(pp)) {
unlock_page(pp);
zfs_rangelock_exit(lr);
if (wbc->sync_mode != WB_SYNC_NONE) {
/*
* Speed up any non-sync page writebacks since
* they may take several seconds to complete.
* Refer to the comment in zpl_fsync() (when
* HAVE_FSYNC_RANGE is defined) for details.
*/
if (atomic_load_32(&zp->z_async_writes_cnt) > 0) {
zil_commit(zfsvfs->z_log, zp->z_id);
}
if (PageWriteback(pp))
#ifdef HAVE_PAGEMAP_FOLIO_WAIT_BIT
folio_wait_bit(page_folio(pp), PG_writeback);
#else
wait_on_page_bit(pp, PG_writeback);
#endif
}
ZFS_EXIT(zfsvfs);
return (0);
}
/* Clear the dirty flag the required locks are held */
if (!clear_page_dirty_for_io(pp)) {
unlock_page(pp);
zfs_rangelock_exit(lr);
ZFS_EXIT(zfsvfs);
return (0);
}
/*
* Counterpart for redirty_page_for_writepage() above. This page
* was in fact not skipped and should not be counted as if it were.
*/
wbc->pages_skipped--;
if (!for_sync)
atomic_inc_32(&zp->z_async_writes_cnt);
set_page_writeback(pp);
unlock_page(pp);
tx = dmu_tx_create(zfsvfs->z_os);
dmu_tx_hold_write(tx, zp->z_id, pgoff, pglen);
dmu_tx_hold_sa(tx, zp->z_sa_hdl, B_FALSE);
zfs_sa_upgrade_txholds(tx, zp);
err = dmu_tx_assign(tx, TXG_NOWAIT);
if (err != 0) {
if (err == ERESTART)
dmu_tx_wait(tx);
dmu_tx_abort(tx);
#ifdef HAVE_VFS_FILEMAP_DIRTY_FOLIO
filemap_dirty_folio(page_mapping(pp), page_folio(pp));
#else
__set_page_dirty_nobuffers(pp);
#endif
ClearPageError(pp);
end_page_writeback(pp);
if (!for_sync)
atomic_dec_32(&zp->z_async_writes_cnt);
zfs_rangelock_exit(lr);
ZFS_EXIT(zfsvfs);
return (err);
}
va = kmap(pp);
ASSERT3U(pglen, <=, PAGE_SIZE);
dmu_write(zfsvfs->z_os, zp->z_id, pgoff, pglen, va, tx);
kunmap(pp);
SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_FLAGS(zfsvfs), NULL,
&zp->z_pflags, 8);
/* Preserve the mtime and ctime provided by the inode */
- ZFS_TIME_ENCODE(&ip->i_mtime, mtime);
- tmp_ctime = zpl_inode_get_ctime(ip);
- ZFS_TIME_ENCODE(&tmp_ctime, ctime);
+ tmp_ts = zpl_inode_get_mtime(ip);
+ ZFS_TIME_ENCODE(&tmp_ts, mtime);
+ tmp_ts = zpl_inode_get_ctime(ip);
+ ZFS_TIME_ENCODE(&tmp_ts, ctime);
zp->z_atime_dirty = B_FALSE;
zp->z_seq++;
err = sa_bulk_update(zp->z_sa_hdl, bulk, cnt, tx);
zfs_log_write(zfsvfs->z_log, tx, TX_WRITE, zp, pgoff, pglen, 0,
for_sync ? zfs_putpage_sync_commit_cb :
zfs_putpage_async_commit_cb, pp);
dmu_tx_commit(tx);
zfs_rangelock_exit(lr);
if (wbc->sync_mode != WB_SYNC_NONE) {
/*
* Note that this is rarely called under writepages(), because
* writepages() normally handles the entire commit for
* performance reasons.
*/
zil_commit(zfsvfs->z_log, zp->z_id);
} else if (!for_sync && atomic_load_32(&zp->z_sync_writes_cnt) > 0) {
/*
* If the caller does not intend to wait synchronously
* for this page writeback to complete and there are active
* synchronous calls on this file, do a commit so that
* the latter don't accidentally end up waiting for
* our writeback to complete. Refer to the comment in
* zpl_fsync() (when HAVE_FSYNC_RANGE is defined) for details.
*/
zil_commit(zfsvfs->z_log, zp->z_id);
}
dataset_kstats_update_write_kstats(&zfsvfs->z_kstat, pglen);
ZFS_EXIT(zfsvfs);
return (err);
}
/*
* Update the system attributes when the inode has been dirtied. For the
* moment we only update the mode, atime, mtime, and ctime.
*/
int
zfs_dirty_inode(struct inode *ip, int flags)
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
dmu_tx_t *tx;
uint64_t mode, atime[2], mtime[2], ctime[2];
- inode_timespec_t tmp_ctime;
+ inode_timespec_t tmp_ts;
sa_bulk_attr_t bulk[4];
int error = 0;
int cnt = 0;
if (zfs_is_readonly(zfsvfs) || dmu_objset_is_snapshot(zfsvfs->z_os))
return (0);
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
#ifdef I_DIRTY_TIME
/*
* This is the lazytime semantic introduced in Linux 4.0
* This flag will only be called from update_time when lazytime is set.
* (Note, I_DIRTY_SYNC will also set if not lazytime)
* Fortunately mtime and ctime are managed within ZFS itself, so we
* only need to dirty atime.
*/
if (flags == I_DIRTY_TIME) {
zp->z_atime_dirty = B_TRUE;
goto out;
}
#endif
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);
if (error) {
dmu_tx_abort(tx);
goto out;
}
mutex_enter(&zp->z_lock);
zp->z_atime_dirty = B_FALSE;
SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MODE(zfsvfs), NULL, &mode, 8);
SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_ATIME(zfsvfs), NULL, &atime, 16);
SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_MTIME(zfsvfs), NULL, &mtime, 16);
SA_ADD_BULK_ATTR(bulk, cnt, SA_ZPL_CTIME(zfsvfs), NULL, &ctime, 16);
/* Preserve the mode, mtime and ctime provided by the inode */
- ZFS_TIME_ENCODE(&ip->i_atime, atime);
- ZFS_TIME_ENCODE(&ip->i_mtime, mtime);
- tmp_ctime = zpl_inode_get_ctime(ip);
- ZFS_TIME_ENCODE(&tmp_ctime, ctime);
+ tmp_ts = zpl_inode_get_atime(ip);
+ ZFS_TIME_ENCODE(&tmp_ts, atime);
+ tmp_ts = zpl_inode_get_mtime(ip);
+ ZFS_TIME_ENCODE(&tmp_ts, mtime);
+ tmp_ts = zpl_inode_get_ctime(ip);
+ ZFS_TIME_ENCODE(&tmp_ts, ctime);
mode = ip->i_mode;
zp->z_mode = mode;
error = sa_bulk_update(zp->z_sa_hdl, bulk, cnt, tx);
mutex_exit(&zp->z_lock);
dmu_tx_commit(tx);
out:
ZFS_EXIT(zfsvfs);
return (error);
}
/*ARGSUSED*/
void
zfs_inactive(struct inode *ip)
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
uint64_t atime[2];
int error;
int need_unlock = 0;
/* Only read lock if we haven't already write locked, e.g. rollback */
if (!RW_WRITE_HELD(&zfsvfs->z_teardown_inactive_lock)) {
need_unlock = 1;
rw_enter(&zfsvfs->z_teardown_inactive_lock, RW_READER);
}
if (zp->z_sa_hdl == NULL) {
if (need_unlock)
rw_exit(&zfsvfs->z_teardown_inactive_lock);
return;
}
if (zp->z_atime_dirty && zp->z_unlinked == B_FALSE) {
dmu_tx_t *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);
if (error) {
dmu_tx_abort(tx);
} else {
- ZFS_TIME_ENCODE(&ip->i_atime, atime);
+ inode_timespec_t tmp_atime;
+ tmp_atime = zpl_inode_get_atime(ip);
+ ZFS_TIME_ENCODE(&tmp_atime, atime);
mutex_enter(&zp->z_lock);
(void) sa_update(zp->z_sa_hdl, SA_ZPL_ATIME(zfsvfs),
(void *)&atime, sizeof (atime), tx);
zp->z_atime_dirty = B_FALSE;
mutex_exit(&zp->z_lock);
dmu_tx_commit(tx);
}
}
zfs_zinactive(zp);
if (need_unlock)
rw_exit(&zfsvfs->z_teardown_inactive_lock);
}
/*
* Fill pages with data from the disk.
*/
static int
zfs_fillpage(struct inode *ip, struct page *pp)
{
zfsvfs_t *zfsvfs = ITOZSB(ip);
loff_t i_size = i_size_read(ip);
u_offset_t io_off = page_offset(pp);
size_t io_len = PAGE_SIZE;
ASSERT3U(io_off, <, i_size);
if (io_off + io_len > i_size)
io_len = i_size - io_off;
void *va = kmap(pp);
int error = dmu_read(zfsvfs->z_os, ITOZ(ip)->z_id, io_off,
io_len, va, DMU_READ_PREFETCH);
if (io_len != PAGE_SIZE)
memset((char *)va + io_len, 0, PAGE_SIZE - io_len);
kunmap(pp);
if (error) {
/* convert checksum errors into IO errors */
if (error == ECKSUM)
error = SET_ERROR(EIO);
SetPageError(pp);
ClearPageUptodate(pp);
} else {
ClearPageError(pp);
SetPageUptodate(pp);
}
return (error);
}
/*
* Uses zfs_fillpage to read data from the file and fill the page.
*
* IN: ip - inode of file to get data from.
* pp - page to read
*
* RETURN: 0 on success, error code on failure.
*
* Timestamps:
* vp - atime updated
*/
/* ARGSUSED */
int
zfs_getpage(struct inode *ip, struct page *pp)
{
zfsvfs_t *zfsvfs = ITOZSB(ip);
znode_t *zp = ITOZ(ip);
int error;
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
error = zfs_fillpage(ip, pp);
if (error == 0)
dataset_kstats_update_read_kstats(&zfsvfs->z_kstat, PAGE_SIZE);
ZFS_EXIT(zfsvfs);
return (error);
}
/*
* Check ZFS specific permissions to memory map a section of a file.
*
* IN: ip - inode of the file to mmap
* off - file offset
* addrp - start address in memory region
* len - length of memory region
* vm_flags- address flags
*
* RETURN: 0 if success
* error code if failure
*/
/*ARGSUSED*/
int
zfs_map(struct inode *ip, offset_t off, caddr_t *addrp, size_t len,
unsigned long vm_flags)
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
if ((vm_flags & VM_WRITE) && (zp->z_pflags &
(ZFS_IMMUTABLE | ZFS_READONLY | ZFS_APPENDONLY))) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EPERM));
}
if ((vm_flags & (VM_READ | VM_EXEC)) &&
(zp->z_pflags & ZFS_AV_QUARANTINED)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EACCES));
}
if (off < 0 || len > MAXOFFSET_T - off) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENXIO));
}
ZFS_EXIT(zfsvfs);
return (0);
}
/*
* Free or allocate space in a file. Currently, this function only
* supports the `F_FREESP' command. However, this command is somewhat
* misnamed, as its functionality includes the ability to allocate as
* well as free space.
*
* IN: zp - znode of file to free data in.
* cmd - action to take (only F_FREESP supported).
* bfp - section of file to free/alloc.
* flag - current file open mode flags.
* offset - current file offset.
* cr - credentials of caller.
*
* RETURN: 0 on success, error code on failure.
*
* Timestamps:
* zp - ctime|mtime updated
*/
/* ARGSUSED */
int
zfs_space(znode_t *zp, int cmd, flock64_t *bfp, int flag,
offset_t offset, cred_t *cr)
{
zfsvfs_t *zfsvfs = ZTOZSB(zp);
uint64_t off, len;
int error;
ZFS_ENTER(zfsvfs);
ZFS_VERIFY_ZP(zp);
if (cmd != F_FREESP) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
/*
* Callers might not be able to detect properly that we are read-only,
* so check it explicitly here.
*/
if (zfs_is_readonly(zfsvfs)) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EROFS));
}
if (bfp->l_len < 0) {
ZFS_EXIT(zfsvfs);
return (SET_ERROR(EINVAL));
}
/*
* Permissions aren't checked on Solaris because on this OS
* zfs_space() can only be called with an opened file handle.
* On Linux we can get here through truncate_range() which
* operates directly on inodes, so we need to check access rights.
*/
if ((error = zfs_zaccess(zp, ACE_WRITE_DATA, 0, B_FALSE, cr))) {
ZFS_EXIT(zfsvfs);
return (error);
}
off = bfp->l_start;
len = bfp->l_len; /* 0 means from off to end of file */
error = zfs_freesp(zp, off, len, flag, TRUE);
ZFS_EXIT(zfsvfs);
return (error);
}
/*ARGSUSED*/
int
zfs_fid(struct inode *ip, fid_t *fidp)
{
znode_t *zp = ITOZ(ip);
zfsvfs_t *zfsvfs = ITOZSB(ip);
uint32_t gen;
uint64_t gen64;
uint64_t object = zp->z_id;
zfid_short_t *zfid;
int size, i, error;
ZFS_ENTER(zfsvfs);
if (fidp->fid_len < SHORT_FID_LEN) {
fidp->fid_len = SHORT_FID_LEN;
ZFS_EXIT(zfsvfs);
return (SET_ERROR(ENOSPC));
}
ZFS_VERIFY_ZP(zp);
if ((error = sa_lookup(zp->z_sa_hdl, SA_ZPL_GEN(zfsvfs),
&gen64, sizeof (uint64_t))) != 0) {
ZFS_EXIT(zfsvfs);
return (error);
}
gen = (uint32_t)gen64;
size = SHORT_FID_LEN;
zfid = (zfid_short_t *)fidp;
zfid->zf_len = size;
for (i = 0; i < sizeof (zfid->zf_object); i++)
zfid->zf_object[i] = (uint8_t)(object >> (8 * i));
/* Must have a non-zero generation number to distinguish from .zfs */
if (gen == 0)
gen = 1;
for (i = 0; i < sizeof (zfid->zf_gen); i++)
zfid->zf_gen[i] = (uint8_t)(gen >> (8 * i));
ZFS_EXIT(zfsvfs);
return (0);
}
#if defined(_KERNEL)
EXPORT_SYMBOL(zfs_open);
EXPORT_SYMBOL(zfs_close);
EXPORT_SYMBOL(zfs_lookup);
EXPORT_SYMBOL(zfs_create);
EXPORT_SYMBOL(zfs_tmpfile);
EXPORT_SYMBOL(zfs_remove);
EXPORT_SYMBOL(zfs_mkdir);
EXPORT_SYMBOL(zfs_rmdir);
EXPORT_SYMBOL(zfs_readdir);
EXPORT_SYMBOL(zfs_getattr_fast);
EXPORT_SYMBOL(zfs_setattr);
EXPORT_SYMBOL(zfs_rename);
EXPORT_SYMBOL(zfs_symlink);
EXPORT_SYMBOL(zfs_readlink);
EXPORT_SYMBOL(zfs_link);
EXPORT_SYMBOL(zfs_inactive);
EXPORT_SYMBOL(zfs_space);
EXPORT_SYMBOL(zfs_fid);
EXPORT_SYMBOL(zfs_getpage);
EXPORT_SYMBOL(zfs_putpage);
EXPORT_SYMBOL(zfs_dirty_inode);
EXPORT_SYMBOL(zfs_map);
/* BEGIN CSTYLED */
module_param(zfs_delete_blocks, ulong, 0644);
MODULE_PARM_DESC(zfs_delete_blocks, "Delete files larger than N blocks async");
/* END CSTYLED */
#endif
diff --git a/module/os/linux/zfs/zfs_znode.c b/module/os/linux/zfs/zfs_znode.c
index 41ccae5ff954..d1e64d0a5cbb 100644
--- a/module/os/linux/zfs/zfs_znode.c
+++ b/module/os/linux/zfs/zfs_znode.c
@@ -1,2268 +1,2275 @@
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2005, 2010, Oracle and/or its affiliates. All rights reserved.
* Copyright (c) 2012, 2018 by Delphix. All rights reserved.
*/
/* Portions Copyright 2007 Jeremy Teo */
#ifdef _KERNEL
#include <sys/types.h>
#include <sys/param.h>
#include <sys/time.h>
#include <sys/sysmacros.h>
#include <sys/mntent.h>
#include <sys/u8_textprep.h>
#include <sys/dsl_dataset.h>
#include <sys/vfs.h>
#include <sys/vnode.h>
#include <sys/file.h>
#include <sys/kmem.h>
#include <sys/errno.h>
#include <sys/atomic.h>
#include <sys/zfs_dir.h>
#include <sys/zfs_acl.h>
#include <sys/zfs_ioctl.h>
#include <sys/zfs_rlock.h>
#include <sys/zfs_fuid.h>
#include <sys/zfs_vnops.h>
#include <sys/zfs_ctldir.h>
#include <sys/dnode.h>
#include <sys/fs/zfs.h>
#include <sys/zpl.h>
#endif /* _KERNEL */
#include <sys/dmu.h>
#include <sys/dmu_objset.h>
#include <sys/dmu_tx.h>
#include <sys/zfs_refcount.h>
#include <sys/stat.h>
#include <sys/zap.h>
#include <sys/zfs_znode.h>
#include <sys/sa.h>
#include <sys/zfs_sa.h>
#include <sys/zfs_stat.h>
#include "zfs_prop.h"
#include "zfs_comutil.h"
/*
* Functions needed for userland (ie: libzpool) are not put under
* #ifdef_KERNEL; the rest of the functions have dependencies
* (such as VFS logic) that will not compile easily in userland.
*/
#ifdef _KERNEL
static kmem_cache_t *znode_cache = NULL;
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(zfs_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)
{
znode_t *zp = buf;
inode_init_once(ZTOI(zp));
list_link_init(&zp->z_link_node);
mutex_init(&zp->z_lock, NULL, MUTEX_DEFAULT, NULL);
rw_init(&zp->z_parent_lock, NULL, RW_DEFAULT, NULL);
rw_init(&zp->z_name_lock, NULL, RW_NOLOCKDEP, NULL);
mutex_init(&zp->z_acl_lock, NULL, MUTEX_DEFAULT, NULL);
rw_init(&zp->z_xattr_lock, NULL, RW_DEFAULT, NULL);
zfs_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 = 0;
zp->z_sync_writes_cnt = 0;
zp->z_async_writes_cnt = 0;
return (0);
}
/*ARGSUSED*/
static void
zfs_znode_cache_destructor(void *buf, void *arg)
{
znode_t *zp = buf;
ASSERT(!list_link_active(&zp->z_link_node));
mutex_destroy(&zp->z_lock);
rw_destroy(&zp->z_parent_lock);
rw_destroy(&zp->z_name_lock);
mutex_destroy(&zp->z_acl_lock);
rw_destroy(&zp->z_xattr_lock);
zfs_rangelock_fini(&zp->z_rangelock);
ASSERT3P(zp->z_dirlocks, ==, NULL);
ASSERT3P(zp->z_acl_cached, ==, NULL);
ASSERT3P(zp->z_xattr_cached, ==, NULL);
ASSERT0(atomic_load_32(&zp->z_sync_writes_cnt));
ASSERT0(atomic_load_32(&zp->z_async_writes_cnt));
}
static int
zfs_znode_hold_cache_constructor(void *buf, void *arg, int kmflags)
{
znode_hold_t *zh = buf;
mutex_init(&zh->zh_lock, NULL, MUTEX_DEFAULT, NULL);
zh->zh_refcount = 0;
return (0);
}
static void
zfs_znode_hold_cache_destructor(void *buf, void *arg)
{
znode_hold_t *zh = buf;
mutex_destroy(&zh->zh_lock);
}
void
zfs_znode_init(void)
{
/*
* Initialize zcache. The KMC_SLAB hint is used in order that it be
* backed by kmalloc() when on the Linux slab in order that any
* wait_on_bit() operations on the related inode operate properly.
*/
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, KMC_SLAB);
ASSERT(znode_hold_cache == NULL);
znode_hold_cache = kmem_cache_create("zfs_znode_hold_cache",
sizeof (znode_hold_t), 0, zfs_znode_hold_cache_constructor,
zfs_znode_hold_cache_destructor, NULL, NULL, NULL, 0);
}
void
zfs_znode_fini(void)
{
/*
* Cleanup zcache
*/
if (znode_cache)
kmem_cache_destroy(znode_cache);
znode_cache = NULL;
if (znode_hold_cache)
kmem_cache_destroy(znode_hold_cache);
znode_hold_cache = NULL;
}
/*
* The zfs_znode_hold_enter() / zfs_znode_hold_exit() functions are used to
* serialize access to a znode and its SA buffer while the object is being
* created or destroyed. This kind of locking would normally reside in the
* znode itself but in this case that's impossible because the znode and SA
* buffer may not yet exist. Therefore the locking is handled externally
* with an array of mutexes and AVLs trees which contain per-object locks.
*
* In zfs_znode_hold_enter() a per-object lock is created as needed, inserted
* in to the correct AVL tree and finally the per-object lock is held. In
* zfs_znode_hold_exit() the process is reversed. The per-object lock is
* released, removed from the AVL tree and destroyed if there are no waiters.
*
* This scheme has two important properties:
*
* 1) No memory allocations are performed while holding one of the z_hold_locks.
* This ensures evict(), which can be called from direct memory reclaim, will
* never block waiting on a z_hold_locks which just happens to have hashed
* to the same index.
*
* 2) All locks used to serialize access to an object are per-object and never
* shared. This minimizes lock contention without creating a large number
* of dedicated locks.
*
* On the downside it does require znode_lock_t structures to be frequently
* allocated and freed. However, because these are backed by a kmem cache
* and very short lived this cost is minimal.
*/
int
zfs_znode_hold_compare(const void *a, const void *b)
{
const znode_hold_t *zh_a = (const znode_hold_t *)a;
const znode_hold_t *zh_b = (const znode_hold_t *)b;
return (TREE_CMP(zh_a->zh_obj, zh_b->zh_obj));
}
static boolean_t __maybe_unused
zfs_znode_held(zfsvfs_t *zfsvfs, uint64_t obj)
{
znode_hold_t *zh, search;
int i = ZFS_OBJ_HASH(zfsvfs, obj);
boolean_t held;
search.zh_obj = obj;
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(&zfsvfs->z_hold_locks[i]);
return (held);
}
static znode_hold_t *
zfs_znode_hold_enter(zfsvfs_t *zfsvfs, uint64_t obj)
{
znode_hold_t *zh, *zh_new, search;
int i = ZFS_OBJ_HASH(zfsvfs, obj);
boolean_t found = B_FALSE;
zh_new = kmem_cache_alloc(znode_hold_cache, KM_SLEEP);
search.zh_obj = obj;
mutex_enter(&zfsvfs->z_hold_locks[i]);
zh = avl_find(&zfsvfs->z_hold_trees[i], &search, NULL);
if (likely(zh == NULL)) {
zh = zh_new;
zh->zh_obj = obj;
avl_add(&zfsvfs->z_hold_trees[i], zh);
} else {
ASSERT3U(zh->zh_obj, ==, obj);
found = B_TRUE;
}
zh->zh_refcount++;
ASSERT3S(zh->zh_refcount, >, 0);
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));
mutex_enter(&zh->zh_lock);
return (zh);
}
static void
zfs_znode_hold_exit(zfsvfs_t *zfsvfs, znode_hold_t *zh)
{
int i = ZFS_OBJ_HASH(zfsvfs, zh->zh_obj);
boolean_t remove = B_FALSE;
ASSERT(zfs_znode_held(zfsvfs, zh->zh_obj));
mutex_exit(&zh->zh_lock);
mutex_enter(&zfsvfs->z_hold_locks[i]);
ASSERT3S(zh->zh_refcount, >, 0);
if (--zh->zh_refcount == 0) {
avl_remove(&zfsvfs->z_hold_trees[i], zh);
remove = B_TRUE;
}
mutex_exit(&zfsvfs->z_hold_locks[i]);
if (remove == B_TRUE)
kmem_cache_free(znode_hold_cache, zh);
}
dev_t
zfs_cmpldev(uint64_t dev)
{
return (dev);
}
static void
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(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(zfsvfs->z_os, db, zp,
SA_HDL_SHARED, &zp->z_sa_hdl));
} else {
zp->z_sa_hdl = sa_hdl;
sa_set_userp(sa_hdl, zp);
}
zp->z_is_sa = (obj_type == DMU_OT_SA) ? B_TRUE : B_FALSE;
mutex_exit(&zp->z_lock);
}
void
zfs_znode_dmu_fini(znode_t *zp)
{
ASSERT(zfs_znode_held(ZTOZSB(zp), zp->z_id) || zp->z_unlinked ||
RW_WRITE_HELD(&ZTOZSB(zp)->z_teardown_inactive_lock));
sa_handle_destroy(zp->z_sa_hdl);
zp->z_sa_hdl = NULL;
}
/*
* 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);
zfsvfs_t *zfsvfs = ZTOZSB(zp);
mutex_enter(&zfsvfs->z_znodes_lock);
if (list_link_active(&zp->z_link_node)) {
list_remove(&zfsvfs->z_all_znodes, zp);
zfsvfs->z_nr_znodes--;
}
mutex_exit(&zfsvfs->z_znodes_lock);
if (zp->z_acl_cached) {
zfs_acl_free(zp->z_acl_cached);
zp->z_acl_cached = NULL;
}
if (zp->z_xattr_cached) {
nvlist_free(zp->z_xattr_cached);
zp->z_xattr_cached = NULL;
}
kmem_cache_free(znode_cache, zp);
}
static void
zfs_inode_set_ops(zfsvfs_t *zfsvfs, struct inode *ip)
{
uint64_t rdev = 0;
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;
/*
* rdev is only stored in a SA only for device files.
*/
case S_IFCHR:
case S_IFBLK:
(void) sa_lookup(ITOZ(ip)->z_sa_hdl, SA_ZPL_RDEV(zfsvfs), &rdev,
sizeof (rdev));
fallthrough;
case S_IFIFO:
case S_IFSOCK:
init_special_inode(ip, ip->i_mode, rdev);
ip->i_op = &zpl_special_inode_operations;
break;
default:
zfs_panic_recover("inode %llu has invalid mode: 0x%x\n",
(u_longlong_t)ip->i_ino, ip->i_mode);
/* Assume the inode is a file and attempt to continue */
ip->i_mode = S_IFREG | 0644;
ip->i_op = &zpl_inode_operations;
ip->i_fop = &zpl_file_operations;
ip->i_mapping->a_ops = &zpl_address_space_operations;
break;
}
}
static void
zfs_set_inode_flags(znode_t *zp, struct inode *ip)
{
/*
* 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_IMMUTABLE;
if (zp->z_pflags & ZFS_APPENDONLY)
ip->i_flags |= S_APPEND;
else
ip->i_flags &= ~S_APPEND;
#endif
}
/*
* Update the embedded inode given the znode.
*/
void
zfs_znode_update_vfs(znode_t *zp)
{
zfsvfs_t *zfsvfs;
struct inode *ip;
uint32_t blksize;
u_longlong_t i_blocks;
ASSERT(zp != NULL);
zfsvfs = ZTOZSB(zp);
ip = ZTOI(zp);
/* Skip .zfs control nodes which do not exist on disk. */
if (zfsctl_is_node(ip))
return;
dmu_object_size_from_db(sa_get_db(zp->z_sa_hdl), &blksize, &i_blocks);
spin_lock(&ip->i_lock);
ip->i_mode = zp->z_mode;
ip->i_blocks = i_blocks;
i_size_write(ip, zp->z_size);
spin_unlock(&ip->i_lock);
}
/*
* 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)
{
znode_t *zp;
struct inode *ip;
uint64_t mode;
uint64_t parent;
uint64_t tmp_gen;
uint64_t links;
uint64_t z_uid, z_gid;
uint64_t atime[2], mtime[2], ctime[2], btime[2];
- inode_timespec_t tmp_ctime;
+ inode_timespec_t tmp_ts;
uint64_t projid = ZFS_DEFAULT_PROJID;
sa_bulk_attr_t bulk[12];
int count = 0;
ASSERT(zfsvfs != NULL);
ip = new_inode(zfsvfs->z_sb);
if (ip == NULL)
return (NULL);
zp = ITOZ(ip);
ASSERT(zp->z_dirlocks == NULL);
ASSERT3P(zp->z_acl_cached, ==, NULL);
ASSERT3P(zp->z_xattr_cached, ==, NULL);
zp->z_unlinked = B_FALSE;
zp->z_atime_dirty = B_FALSE;
#if !defined(HAVE_FILEMAP_RANGE_HAS_PAGE)
zp->z_is_mapped = B_FALSE;
#endif
zp->z_is_ctldir = B_FALSE;
zp->z_suspended = B_FALSE;
zp->z_sa_hdl = NULL;
zp->z_mapcnt = 0;
zp->z_id = db->db_object;
zp->z_blksz = blksz;
zp->z_seq = 0x7A4653;
zp->z_sync_cnt = 0;
zp->z_sync_writes_cnt = 0;
zp->z_async_writes_cnt = 0;
zfs_znode_sa_init(zfsvfs, 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, &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(zfsvfs), NULL,
&parent, 8);
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);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CRTIME(zfsvfs), NULL, &btime, 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_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);
/* 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(&tmp_ctime, ctime);
- zpl_inode_set_ctime_to_ts(ip, tmp_ctime);
+ ZFS_TIME_DECODE(&tmp_ts, atime);
+ zpl_inode_set_atime_to_ts(ip, tmp_ts);
+ ZFS_TIME_DECODE(&tmp_ts, mtime);
+ zpl_inode_set_mtime_to_ts(ip, tmp_ts);
+ ZFS_TIME_DECODE(&tmp_ts, ctime);
+ zpl_inode_set_ctime_to_ts(ip, tmp_ts);
ZFS_TIME_DECODE(&zp->z_btime, btime);
ip->i_ino = zp->z_id;
zfs_znode_update_vfs(zp);
zfs_inode_set_ops(zfsvfs, ip);
/*
* The only way insert_inode_locked() can fail is if the ip->i_ino
* number is already hashed for this super block. This can never
* happen because the inode numbers map 1:1 with the object numbers.
*
* Exceptions include rolling back a mounted file system, either
* from the zfs rollback or zfs recv command.
*
* Active inodes are unhashed during the rollback, but since zrele
* can happen asynchronously, we can't guarantee they've been
* unhashed. This can cause hash collisions in unlinked drain
* processing so do not hash unlinked znodes.
*/
if (links > 0)
VERIFY3S(insert_inode_locked(ip), ==, 0);
mutex_enter(&zfsvfs->z_znodes_lock);
list_insert_tail(&zfsvfs->z_all_znodes, zp);
zfsvfs->z_nr_znodes++;
mutex_exit(&zfsvfs->z_znodes_lock);
if (links > 0)
unlock_new_inode(ip);
return (zp);
error:
iput(ip);
return (NULL);
}
/*
* Safely mark an inode dirty. Inodes which are part of a read-only
* file system or snapshot may not be dirtied.
*/
void
zfs_mark_inode_dirty(struct inode *ip)
{
zfsvfs_t *zfsvfs = ITOZSB(ip);
if (zfs_is_readonly(zfsvfs) || dmu_objset_is_snapshot(zfsvfs->z_os))
return;
mark_inode_dirty(ip);
}
static uint64_t empty_xattr;
static uint64_t pad[4];
static zfs_acl_phys_t acl_phys;
/*
* Create a new DMU object to hold a zfs znode.
*
* IN: dzp - parent directory for new znode
* vap - file attributes for new znode
* tx - dmu transaction id for zap operations
* cr - credentials of caller
* flag - flags:
* IS_ROOT_NODE - new object will be root
* IS_TMPFILE - new object is of O_TMPFILE
* IS_XATTR - new object is an attribute
* acl_ids - ACL related attributes
*
* OUT: zpp - allocated znode (set to dzp if IS_ROOT_NODE)
*
*/
void
zfs_mknode(znode_t *dzp, vattr_t *vap, dmu_tx_t *tx, cred_t *cr,
uint_t flag, znode_t **zpp, zfs_acl_ids_t *acl_ids)
{
uint64_t crtime[2], atime[2], mtime[2], ctime[2];
uint64_t mode, size, links, parent, pflags;
uint64_t projid = ZFS_DEFAULT_PROJID;
uint64_t rdev = 0;
zfsvfs_t *zfsvfs = ZTOZSB(dzp);
dmu_buf_t *db;
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;
int cnt = 0;
zfs_acl_locator_cb_t locate = { 0 };
znode_hold_t *zh;
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);
}
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_BONUS_SIZE(dnodesize) : ZFS_OLD_ZNODE_PHYS_SIZE;
/*
* Create a new DMU object.
*/
/*
* There's currently no mechanism for pre-reading the blocks that will
* be needed to allocate a new object, so we accept the small chance
* that there will be an i/o error and we will fail one of the
* assertions below.
*/
if (S_ISDIR(vap->va_mode)) {
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_dnsize(zfsvfs->z_os,
zfsvfs->z_norm, DMU_OT_DIRECTORY_CONTENTS,
obj_type, bonuslen, dnodesize, tx);
}
} else {
if (zfsvfs->z_replay) {
VERIFY0(dmu_object_claim_dnsize(zfsvfs->z_os, obj,
DMU_OT_PLAIN_FILE_CONTENTS, 0,
obj_type, bonuslen, dnodesize, tx));
} else {
obj = dmu_object_alloc_dnsize(zfsvfs->z_os,
DMU_OT_PLAIN_FILE_CONTENTS, 0,
obj_type, bonuslen, dnodesize, tx);
}
}
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
* to reference the just-allocated physical data area.
*/
if (flag & IS_ROOT_NODE) {
dzp->z_id = obj;
}
/*
* If parent is an xattr, so am I.
*/
if (dzp->z_pflags & ZFS_XATTR) {
flag |= IS_XATTR;
}
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 = 2;
} else {
size = 0;
links = (flag & IS_TMPFILE) ? 0 : 1;
}
if (S_ISBLK(vap->va_mode) || S_ISCHR(vap->va_mode))
rdev = vap->va_rdev;
parent = dzp->z_id;
mode = acl_ids->z_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 determined when zfs_mode_compute() is called.
*/
pflags |= acl_ids->z_aclp->z_hints &
(ZFS_ACL_TRIVIAL|ZFS_INHERIT_ACE|ZFS_ACL_AUTO_INHERIT|
ZFS_ACL_DEFAULTED|ZFS_ACL_PROTECTED);
ZFS_TIME_ENCODE(&now, crtime);
ZFS_TIME_ENCODE(&now, ctime);
if (vap->va_mask & ATTR_ATIME) {
ZFS_TIME_ENCODE(&vap->va_atime, atime);
} else {
ZFS_TIME_ENCODE(&now, atime);
}
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,
&sa_hdl));
/*
* Setup the array of attributes to be replaced/set on the new file
*
* 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),
NULL, &atime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zfsvfs),
NULL, &mtime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zfsvfs),
NULL, &ctime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zfsvfs),
NULL, &crtime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zfsvfs),
NULL, &gen, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MODE(zfsvfs),
NULL, &mode, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zfsvfs),
NULL, &size, 8);
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(zfsvfs),
NULL, &mode, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_SIZE(zfsvfs),
NULL, &size, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_GEN(zfsvfs),
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),
NULL, &parent, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zfsvfs),
NULL, &pflags, 8);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_ATIME(zfsvfs),
NULL, &atime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_MTIME(zfsvfs),
NULL, &mtime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CTIME(zfsvfs),
NULL, &ctime, 16);
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_CRTIME(zfsvfs),
NULL, &crtime, 16);
}
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(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(zfsvfs),
NULL, &rdev, 8);
}
if (obj_type == DMU_OT_ZNODE) {
SA_ADD_BULK_ATTR(sa_attrs, cnt, SA_ZPL_FLAGS(zfsvfs),
NULL, &pflags, 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_PAD(zfsvfs), NULL, pad,
sizeof (uint64_t) * 4);
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(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(zfsvfs),
zfs_acl_data_locator, &locate,
acl_ids->z_aclp->z_acl_bytes);
mode = zfs_mode_compute(mode, acl_ids->z_aclp, &pflags,
acl_ids->z_fuid, acl_ids->z_fgid);
}
VERIFY(sa_replace_all_by_template(sa_hdl, sa_attrs, cnt, tx) == 0);
if (!(flag & IS_ROOT_NODE)) {
/*
* 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, sa_hdl);
} while (*zpp == NULL);
VERIFY(*zpp != NULL);
VERIFY(dzp != NULL);
} else {
/*
* If we are creating the root node, the "parent" we
* passed in is the znode for the root.
*/
*zpp = dzp;
(*zpp)->z_sa_hdl = sa_hdl;
}
(*zpp)->z_pflags = pflags;
(*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(zfsvfs, zh);
}
/*
* Update 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;
boolean_t update_inode = B_FALSE;
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(ZTOZSB(zp)),
×, 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);
update_inode = B_TRUE;
}
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);
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_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);
}
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(zfsvfs_t *zfsvfs, uint64_t obj_num, znode_t **zpp)
{
dmu_object_info_t doi;
dmu_buf_t *db;
znode_t *zp;
znode_hold_t *zh;
int err;
sa_handle_t *hdl;
*zpp = NULL;
again:
zh = zfs_znode_hold_enter(zfsvfs, obj_num);
err = sa_buf_hold(zfsvfs->z_os, obj_num, NULL, &db);
if (err) {
zfs_znode_hold_exit(zfsvfs, zh);
return (err);
}
dmu_object_info_from_db(db, &doi);
if (doi.doi_bonus_type != DMU_OT_SA &&
(doi.doi_bonus_type != DMU_OT_ZNODE ||
(doi.doi_bonus_type == DMU_OT_ZNODE &&
doi.doi_bonus_size < sizeof (znode_phys_t)))) {
sa_buf_rele(db, NULL);
zfs_znode_hold_exit(zfsvfs, zh);
return (SET_ERROR(EINVAL));
}
hdl = dmu_buf_get_user(db);
if (hdl != NULL) {
zp = sa_get_userdata(hdl);
/*
* Since "SA" does immediate eviction we
* should never find a sa handle that doesn't
* know about the znode.
*/
ASSERT3P(zp, !=, NULL);
mutex_enter(&zp->z_lock);
ASSERT3U(zp->z_id, ==, obj_num);
/*
* If zp->z_unlinked is set, the znode is already marked
* for deletion and should not be discovered. Check this
* after checking igrab() due to fsetxattr() & O_TMPFILE.
*
* 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) {
if (zp->z_unlinked)
err = SET_ERROR(ENOENT);
else
err = SET_ERROR(EAGAIN);
} else {
*zpp = zp;
err = 0;
}
mutex_exit(&zp->z_lock);
sa_buf_rele(db, NULL);
zfs_znode_hold_exit(zfsvfs, zh);
if (err == EAGAIN) {
/* inode might need this to finish evict */
cond_resched();
goto again;
}
return (err);
}
/*
* 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
* progress. This is checked for in zfs_znode_alloc()
*
* 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);
if (zp == NULL) {
err = SET_ERROR(ENOENT);
} else {
*zpp = zp;
}
zfs_znode_hold_exit(zfsvfs, zh);
return (err);
}
int
zfs_rezget(znode_t *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;
uint64_t links;
sa_bulk_attr_t bulk[11];
int err;
int count = 0;
uint64_t gen;
uint64_t z_uid, z_gid;
uint64_t atime[2], mtime[2], ctime[2], btime[2];
- inode_timespec_t tmp_ctime;
+ inode_timespec_t tmp_ts;
uint64_t projid = ZFS_DEFAULT_PROJID;
znode_hold_t *zh;
/*
* skip ctldir, otherwise they will always get invalidated. This will
* cause funny behaviour for the mounted snapdirs. Especially for
* Linux >= 3.18, d_invalidate will detach the mountpoint and prevent
* anyone automount it again as long as someone is still using the
* detached mount.
*/
if (zp->z_is_ctldir)
return (0);
zh = zfs_znode_hold_enter(zfsvfs, obj_num);
mutex_enter(&zp->z_acl_lock);
if (zp->z_acl_cached) {
zfs_acl_free(zp->z_acl_cached);
zp->z_acl_cached = NULL;
}
mutex_exit(&zp->z_acl_lock);
rw_enter(&zp->z_xattr_lock, RW_WRITER);
if (zp->z_xattr_cached) {
nvlist_free(zp->z_xattr_cached);
zp->z_xattr_cached = NULL;
}
rw_exit(&zp->z_xattr_lock);
ASSERT(zp->z_sa_hdl == NULL);
err = sa_buf_hold(zfsvfs->z_os, obj_num, NULL, &db);
if (err) {
zfs_znode_hold_exit(zfsvfs, zh);
return (err);
}
dmu_object_info_from_db(db, &doi);
if (doi.doi_bonus_type != DMU_OT_SA &&
(doi.doi_bonus_type != DMU_OT_ZNODE ||
(doi.doi_bonus_type == DMU_OT_ZNODE &&
doi.doi_bonus_size < sizeof (znode_phys_t)))) {
sa_buf_rele(db, NULL);
zfs_znode_hold_exit(zfsvfs, zh);
return (SET_ERROR(EINVAL));
}
zfs_znode_sa_init(zfsvfs, zp, db, doi.doi_bonus_type, NULL);
/* reload cached values */
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_GEN(zfsvfs), NULL,
&gen, sizeof (gen));
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(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(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);
SA_ADD_BULK_ATTR(bulk, count, SA_ZPL_CRTIME(zfsvfs), NULL, &btime, 16);
if (sa_bulk_lookup(zp->z_sa_hdl, bulk, count)) {
zfs_znode_dmu_fini(zp);
zfs_znode_hold_exit(zfsvfs, zh);
return (SET_ERROR(EIO));
}
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(&tmp_ctime, ctime);
- zpl_inode_set_ctime_to_ts(ZTOI(zp), tmp_ctime);
+ ZFS_TIME_DECODE(&tmp_ts, atime);
+ zpl_inode_set_atime_to_ts(ZTOI(zp), tmp_ts);
+ ZFS_TIME_DECODE(&tmp_ts, mtime);
+ zpl_inode_set_mtime_to_ts(ZTOI(zp), tmp_ts);
+ ZFS_TIME_DECODE(&tmp_ts, ctime);
+ zpl_inode_set_ctime_to_ts(ZTOI(zp), tmp_ts);
ZFS_TIME_DECODE(&zp->z_btime, btime);
if ((uint32_t)gen != ZTOI(zp)->i_generation) {
zfs_znode_dmu_fini(zp);
zfs_znode_hold_exit(zfsvfs, zh);
return (SET_ERROR(EIO));
}
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 = B_FALSE;
zfs_znode_update_vfs(zp);
/*
* 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 automatic 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)
{
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(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(zfsvfs, zh);
}
void
zfs_zinactive(znode_t *zp)
{
zfsvfs_t *zfsvfs = ZTOZSB(zp);
uint64_t z_id = zp->z_id;
znode_hold_t *zh;
ASSERT(zp->z_sa_hdl);
/*
* Don't allow a zfs_zget() while were trying to release this znode.
*/
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 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) {
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(zfsvfs, zh);
}
#if defined(HAVE_INODE_TIMESPEC64_TIMES)
#define zfs_compare_timespec timespec64_compare
#else
#define zfs_compare_timespec timespec_compare
#endif
/*
* Determine whether the znode's atime must be updated. The logic mostly
* duplicates the Linux kernel's relatime_need_update() functionality.
* This function is only called if the underlying filesystem actually has
* atime updates enabled.
*/
boolean_t
zfs_relatime_need_update(const struct inode *ip)
{
- inode_timespec_t now, tmp_ctime;
+ inode_timespec_t now, tmp_atime, tmp_ts;
gethrestime(&now);
+ tmp_atime = zpl_inode_get_atime(ip);
/*
* In relatime mode, only update the atime if the previous atime
* is earlier than either the ctime or mtime or if at least a day
* has passed since the last update of atime.
*/
- if (zfs_compare_timespec(&ip->i_mtime, &ip->i_atime) >= 0)
+ tmp_ts = zpl_inode_get_mtime(ip);
+ if (zfs_compare_timespec(&tmp_ts, &tmp_atime) >= 0)
return (B_TRUE);
- tmp_ctime = zpl_inode_get_ctime(ip);
- if (zfs_compare_timespec(&tmp_ctime, &ip->i_atime) >= 0)
+ tmp_ts = zpl_inode_get_ctime(ip);
+ if (zfs_compare_timespec(&tmp_ts, &tmp_atime) >= 0)
return (B_TRUE);
- if ((hrtime_t)now.tv_sec - (hrtime_t)ip->i_atime.tv_sec >= 24*60*60)
+ if ((hrtime_t)now.tv_sec - (hrtime_t)tmp_atime.tv_sec >= 24*60*60)
return (B_TRUE);
return (B_FALSE);
}
/*
* Prepare to update znode time stamps.
*
* IN: zp - znode requiring timestamp update
* flag - ATTR_MTIME, ATTR_CTIME flags
*
* OUT: zp - z_seq
* mtime - new mtime
* ctime - new ctime
*
* Note: We don't update atime here, because we rely on Linux VFS to do
* atime updating.
*/
void
zfs_tstamp_update_setup(znode_t *zp, uint_t flag, uint64_t mtime[2],
uint64_t ctime[2])
{
- inode_timespec_t now, tmp_ctime;
+ inode_timespec_t now, tmp_ts;
gethrestime(&now);
zp->z_seq++;
if (flag & ATTR_MTIME) {
ZFS_TIME_ENCODE(&now, mtime);
- ZFS_TIME_DECODE(&(ZTOI(zp)->i_mtime), mtime);
+ ZFS_TIME_DECODE(&tmp_ts, mtime);
+ zpl_inode_set_mtime_to_ts(ZTOI(zp), tmp_ts);
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(&tmp_ctime, ctime);
- zpl_inode_set_ctime_to_ts(ZTOI(zp), tmp_ctime);
+ ZFS_TIME_DECODE(&tmp_ts, ctime);
+ zpl_inode_set_ctime_to_ts(ZTOI(zp), tmp_ts);
if (ZTOZSB(zp)->z_use_fuids)
zp->z_pflags |= ZFS_ARCHIVE;
}
}
/*
* Grow the block size for a file.
*
* IN: zp - znode of file to free data in.
* size - requested block size
* tx - open transaction.
*
* NOTE: this function assumes that the znode is write locked.
*/
void
zfs_grow_blocksize(znode_t *zp, uint64_t size, dmu_tx_t *tx)
{
int error;
u_longlong_t dummy;
if (size <= zp->z_blksz)
return;
/*
* If the file size is already greater than the current blocksize,
* we will not grow. If there is more than one block in a file,
* the blocksize cannot change.
*/
if (zp->z_blksz && zp->z_size > zp->z_blksz)
return;
error = dmu_object_set_blocksize(ZTOZSB(zp)->z_os, zp->z_id,
size, 0, tx);
if (error == ENOTSUP)
return;
ASSERT0(error);
/* What blocksize did we actually get? */
dmu_object_size_from_db(sa_get_db(zp->z_sa_hdl), &zp->z_blksz, &dummy);
}
/*
* Increase the file length
*
* IN: zp - znode of file to free data in.
* end - new end-of-file
*
* RETURN: 0 on success, error code on failure
*/
static int
zfs_extend(znode_t *zp, uint64_t end)
{
zfsvfs_t *zfsvfs = ZTOZSB(zp);
dmu_tx_t *tx;
zfs_locked_range_t *lr;
uint64_t newblksz;
int error;
/*
* We will change zp_size, lock the whole file.
*/
lr = zfs_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_rangelock_exit(lr);
return (0);
}
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 < zfsvfs->z_max_blksz)) {
/*
* We are growing the file past the current block size.
*/
if (zp->z_blksz > ZTOZSB(zp)->z_max_blksz) {
/*
* File's blocksize is already larger than the
* "recordsize" property. Only let it grow to
* the next power of 2.
*/
ASSERT(!ISP2(zp->z_blksz));
newblksz = MIN(end, 1 << highbit64(zp->z_blksz));
} else {
newblksz = MIN(end, ZTOZSB(zp)->z_max_blksz);
}
dmu_tx_hold_write(tx, zp->z_id, 0, newblksz);
} else {
newblksz = 0;
}
error = dmu_tx_assign(tx, TXG_WAIT);
if (error) {
dmu_tx_abort(tx);
zfs_rangelock_exit(lr);
return (error);
}
if (newblksz)
zfs_grow_blocksize(zp, newblksz, tx);
zp->z_size = end;
VERIFY(0 == sa_update(zp->z_sa_hdl, SA_ZPL_SIZE(ZTOZSB(zp)),
&zp->z_size, sizeof (zp->z_size), tx));
zfs_rangelock_exit(lr);
dmu_tx_commit(tx);
return (0);
}
/*
* zfs_zero_partial_page - Modeled after update_pages() but
* with different arguments and semantics for use by zfs_freesp().
*
* Zeroes a piece of a single page cache entry for zp at offset
* start and length len.
*
* Caller must acquire a range lock on the file for the region
* being zeroed in order that the ARC and page cache stay in sync.
*/
static void
zfs_zero_partial_page(znode_t *zp, uint64_t start, uint64_t len)
{
struct address_space *mp = ZTOI(zp)->i_mapping;
struct page *pp;
int64_t off;
void *pb;
ASSERT((start & PAGE_MASK) == ((start + len - 1) & PAGE_MASK));
off = start & (PAGE_SIZE - 1);
start &= PAGE_MASK;
pp = find_lock_page(mp, start >> PAGE_SHIFT);
if (pp) {
if (mapping_writably_mapped(mp))
flush_dcache_page(pp);
pb = kmap(pp);
bzero(pb + off, len);
kunmap(pp);
if (mapping_writably_mapped(mp))
flush_dcache_page(pp);
mark_page_accessed(pp);
SetPageUptodate(pp);
ClearPageError(pp);
unlock_page(pp);
put_page(pp);
}
}
/*
* Free space in a file.
*
* IN: zp - znode of file to free data in.
* off - start of section to free.
* len - length of section to free.
*
* RETURN: 0 on success, error code on failure
*/
static int
zfs_free_range(znode_t *zp, uint64_t off, uint64_t len)
{
zfsvfs_t *zfsvfs = ZTOZSB(zp);
zfs_locked_range_t *lr;
int error;
/*
* Lock the range being freed.
*/
lr = zfs_rangelock_enter(&zp->z_rangelock, off, len, RL_WRITER);
/*
* Nothing to do if file already at desired length.
*/
if (off >= zp->z_size) {
zfs_rangelock_exit(lr);
return (0);
}
if (off + len > zp->z_size)
len = zp->z_size - off;
error = dmu_free_long_range(zfsvfs->z_os, zp->z_id, off, len);
/*
* Zero partial page cache entries. This must be done under a
* range lock in order to keep the ARC and page cache in sync.
*/
if (zn_has_cached_data(zp, off, off + len - 1)) {
loff_t first_page, last_page, page_len;
loff_t first_page_offset, last_page_offset;
/* first possible full page in hole */
first_page = (off + PAGE_SIZE - 1) >> PAGE_SHIFT;
/* last page of hole */
last_page = (off + len) >> PAGE_SHIFT;
/* offset of first_page */
first_page_offset = first_page << PAGE_SHIFT;
/* offset of last_page */
last_page_offset = last_page << PAGE_SHIFT;
/* truncate whole pages */
if (last_page_offset > first_page_offset) {
truncate_inode_pages_range(ZTOI(zp)->i_mapping,
first_page_offset, last_page_offset - 1);
}
/* truncate sub-page ranges */
if (first_page > last_page) {
/* entire punched area within a single page */
zfs_zero_partial_page(zp, off, len);
} else {
/* beginning of punched area at the end of a page */
page_len = first_page_offset - off;
if (page_len > 0)
zfs_zero_partial_page(zp, off, page_len);
/* end of punched area at the beginning of a page */
page_len = off + len - last_page_offset;
if (page_len > 0)
zfs_zero_partial_page(zp, last_page_offset,
page_len);
}
}
zfs_rangelock_exit(lr);
return (error);
}
/*
* Truncate a file
*
* IN: zp - znode of file to free data in.
* end - new end-of-file.
*
* RETURN: 0 on success, error code on failure
*/
static int
zfs_trunc(znode_t *zp, uint64_t end)
{
zfsvfs_t *zfsvfs = ZTOZSB(zp);
dmu_tx_t *tx;
zfs_locked_range_t *lr;
int error;
sa_bulk_attr_t bulk[2];
int count = 0;
/*
* We will change zp_size, lock the whole file.
*/
lr = zfs_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_rangelock_exit(lr);
return (0);
}
error = dmu_free_long_range(zfsvfs->z_os, zp->z_id, end,
DMU_OBJECT_END);
if (error) {
zfs_rangelock_exit(lr);
return (error);
}
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_rangelock_exit(lr);
return (error);
}
zp->z_size = end;
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(zfsvfs),
NULL, &zp->z_pflags, 8);
}
VERIFY(sa_bulk_update(zp->z_sa_hdl, bulk, count, tx) == 0);
dmu_tx_commit(tx);
zfs_rangelock_exit(lr);
return (0);
}
/*
* Free space in a file
*
* IN: zp - znode of file to free data in.
* off - start of range
* len - end of range (0 => EOF)
* flag - current file open mode flags.
* log - TRUE if this action should be logged
*
* RETURN: 0 on success, error code on failure
*/
int
zfs_freesp(znode_t *zp, uint64_t off, uint64_t len, int flag, boolean_t log)
{
dmu_tx_t *tx;
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(zfsvfs), &mode,
sizeof (mode))) != 0)
return (error);
if (off > zp->z_size) {
error = zfs_extend(zp, off+len);
if (error == 0 && log)
goto log;
goto out;
}
if (len == 0) {
error = zfs_trunc(zp, off);
} else {
if ((error = zfs_free_range(zp, off, len)) == 0 &&
off + len > zp->z_size)
error = zfs_extend(zp, off+len);
}
if (error || !log)
goto out;
log:
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);
if (error) {
dmu_tx_abort(tx);
goto out;
}
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);
ASSERT(error == 0);
zfs_log_truncate(zilog, tx, TX_TRUNCATE, zp, off, len);
dmu_tx_commit(tx);
zfs_znode_update_vfs(zp);
error = 0;
out:
/*
* Truncate the page cache - for file truncate operations, use
* the purpose-built API for truncations. For punching operations,
* the truncation is handled under a range lock in zfs_free_range.
*/
if (len == 0)
truncate_setsize(ZTOI(zp), off);
return (error);
}
void
zfs_create_fs(objset_t *os, cred_t *cr, nvlist_t *zplprops, dmu_tx_t *tx)
{
struct super_block *sb;
zfsvfs_t *zfsvfs;
uint64_t moid, obj, sa_obj, version;
uint64_t sense = ZFS_CASE_SENSITIVE;
uint64_t norm = 0;
nvpair_t *elem;
int size;
int error;
int i;
znode_t *rootzp = NULL;
vattr_t vattr;
znode_t *zp;
zfs_acl_ids_t acl_ids;
/*
* First attempt to create master node.
*/
/*
* In an empty objset, there are no blocks to read and thus
* there can be no i/o errors (which we assert below).
*/
moid = MASTER_NODE_OBJ;
error = zap_create_claim(os, moid, DMU_OT_MASTER_NODE,
DMU_OT_NONE, 0, tx);
ASSERT(error == 0);
/*
* Set starting attributes.
*/
version = zfs_zpl_version_map(spa_version(dmu_objset_spa(os)));
elem = NULL;
while ((elem = nvlist_next_nvpair(zplprops, elem)) != NULL) {
/* For the moment we expect all zpl props to be uint64_ts */
uint64_t val;
char *name;
ASSERT(nvpair_type(elem) == DATA_TYPE_UINT64);
VERIFY(nvpair_value_uint64(elem, &val) == 0);
name = nvpair_name(elem);
if (strcmp(name, zfs_prop_to_name(ZFS_PROP_VERSION)) == 0) {
if (val < version)
version = val;
} else {
error = zap_update(os, moid, name, 8, 1, &val, 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);
/*
* Create zap object used for SA attribute registration
*/
if (version >= ZPL_VERSION_SA) {
sa_obj = zap_create(os, DMU_OT_SA_MASTER_NODE,
DMU_OT_NONE, 0, tx);
error = zap_add(os, moid, ZFS_SA_ATTRS, 8, 1, &sa_obj, tx);
ASSERT(error == 0);
} else {
sa_obj = 0;
}
/*
* Create a delete queue.
*/
obj = zap_create(os, DMU_OT_UNLINKED_SET, DMU_OT_NONE, 0, tx);
error = zap_add(os, moid, ZFS_UNLINKED_SET, 8, 1, &obj, tx);
ASSERT(error == 0);
/*
* Create root znode. Create minimal znode/inode/zfsvfs/sb
* to allow zfs_mknode to work.
*/
vattr.va_mask = ATTR_MODE|ATTR_UID|ATTR_GID;
vattr.va_mode = S_IFDIR|0755;
vattr.va_uid = crgetuid(cr);
vattr.va_gid = crgetgid(cr);
rootzp = kmem_cache_alloc(znode_cache, KM_SLEEP);
rootzp->z_unlinked = B_FALSE;
rootzp->z_atime_dirty = B_FALSE;
rootzp->z_is_sa = USE_SA(version, os);
rootzp->z_pflags = 0;
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 = zfsvfs;
ZTOI(rootzp)->i_sb = sb;
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;
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);
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(&zfsvfs->z_hold_trees[i], zfs_znode_hold_compare,
sizeof (znode_hold_t), offsetof(znode_hold_t, zh_node));
mutex_init(&zfsvfs->z_hold_locks[i], NULL, MUTEX_DEFAULT, NULL);
}
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);
error = zap_add(os, moid, ZFS_ROOT_OBJ, 8, 1, &rootzp->z_id, tx);
ASSERT(error == 0);
zfs_acl_ids_free(&acl_ids);
atomic_set(&ZTOI(rootzp)->i_count, 0);
sa_handle_destroy(rootzp->z_sa_hdl);
kmem_cache_free(znode_cache, rootzp);
for (i = 0; i != size; i++) {
avl_destroy(&zfsvfs->z_hold_trees[i]);
mutex_destroy(&zfsvfs->z_hold_locks[i]);
}
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(zfsvfs, sizeof (zfsvfs_t));
}
#endif /* _KERNEL */
static int
zfs_sa_setup(objset_t *osp, sa_attr_type_t **sa_table)
{
uint64_t sa_obj = 0;
int error;
error = zap_lookup(osp, MASTER_NODE_OBJ, ZFS_SA_ATTRS, 8, 1, &sa_obj);
if (error != 0 && error != ENOENT)
return (error);
error = sa_setup(osp, sa_obj, zfs_attr_table, ZPL_END, sa_table);
return (error);
}
static int
zfs_grab_sa_handle(objset_t *osp, uint64_t obj, sa_handle_t **hdlp,
dmu_buf_t **db, void *tag)
{
dmu_object_info_t doi;
int error;
if ((error = sa_buf_hold(osp, obj, tag, db)) != 0)
return (error);
dmu_object_info_from_db(*db, &doi);
if ((doi.doi_bonus_type != DMU_OT_SA &&
doi.doi_bonus_type != DMU_OT_ZNODE) ||
(doi.doi_bonus_type == DMU_OT_ZNODE &&
doi.doi_bonus_size < sizeof (znode_phys_t))) {
sa_buf_rele(*db, tag);
return (SET_ERROR(ENOTSUP));
}
error = sa_handle_get(osp, obj, NULL, SA_HDL_PRIVATE, hdlp);
if (error != 0) {
sa_buf_rele(*db, tag);
return (error);
}
return (0);
}
static void
zfs_release_sa_handle(sa_handle_t *hdl, dmu_buf_t *db, void *tag)
{
sa_handle_destroy(hdl);
sa_buf_rele(db, tag);
}
/*
* Given an object number, return its parent object number and whether
* or not the object is an extended attribute directory.
*/
static int
zfs_obj_to_pobj(objset_t *osp, sa_handle_t *hdl, sa_attr_type_t *sa_table,
uint64_t *pobjp, int *is_xattrdir)
{
uint64_t parent;
uint64_t pflags;
uint64_t mode;
uint64_t parent_mode;
sa_bulk_attr_t bulk[3];
sa_handle_t *sa_hdl;
dmu_buf_t *sa_db;
int count = 0;
int error;
SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_PARENT], NULL,
&parent, sizeof (parent));
SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_FLAGS], NULL,
&pflags, sizeof (pflags));
SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_MODE], NULL,
&mode, sizeof (mode));
if ((error = sa_bulk_lookup(hdl, bulk, count)) != 0)
return (error);
/*
* When a link is removed its parent pointer is not changed and will
* be invalid. There are two cases where a link is removed but the
* file stays around, when it goes to the delete queue and when there
* are additional links.
*/
error = zfs_grab_sa_handle(osp, parent, &sa_hdl, &sa_db, FTAG);
if (error != 0)
return (error);
error = sa_lookup(sa_hdl, ZPL_MODE, &parent_mode, sizeof (parent_mode));
zfs_release_sa_handle(sa_hdl, sa_db, FTAG);
if (error != 0)
return (error);
*is_xattrdir = ((pflags & ZFS_XATTR) != 0) && S_ISDIR(mode);
/*
* Extended attributes can be applied to files, directories, etc.
* Otherwise the parent must be a directory.
*/
if (!*is_xattrdir && !S_ISDIR(parent_mode))
return (SET_ERROR(EINVAL));
*pobjp = parent;
return (0);
}
/*
* Given an object number, return some zpl level statistics
*/
static int
zfs_obj_to_stats_impl(sa_handle_t *hdl, sa_attr_type_t *sa_table,
zfs_stat_t *sb)
{
sa_bulk_attr_t bulk[4];
int count = 0;
SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_MODE], NULL,
&sb->zs_mode, sizeof (sb->zs_mode));
SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_GEN], NULL,
&sb->zs_gen, sizeof (sb->zs_gen));
SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_LINKS], NULL,
&sb->zs_links, sizeof (sb->zs_links));
SA_ADD_BULK_ATTR(bulk, count, sa_table[ZPL_CTIME], NULL,
&sb->zs_ctime, sizeof (sb->zs_ctime));
return (sa_bulk_lookup(hdl, bulk, count));
}
static int
zfs_obj_to_path_impl(objset_t *osp, uint64_t obj, sa_handle_t *hdl,
sa_attr_type_t *sa_table, char *buf, int len)
{
sa_handle_t *sa_hdl;
sa_handle_t *prevhdl = NULL;
dmu_buf_t *prevdb = NULL;
dmu_buf_t *sa_db = NULL;
char *path = buf + len - 1;
int error;
*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];
size_t complen;
int is_xattrdir = 0;
if (prevdb) {
ASSERT(prevhdl != NULL);
zfs_release_sa_handle(prevhdl, prevdb, FTAG);
}
if ((error = zfs_obj_to_pobj(osp, sa_hdl, sa_table, &pobj,
&is_xattrdir)) != 0)
break;
if (pobj == obj) {
if (path[0] != '/')
*--path = '/';
break;
}
component[0] = '/';
if (is_xattrdir) {
(void) sprintf(component + 1, "<xattrdir>");
} else {
error = zap_value_search(osp, pobj, obj,
ZFS_DIRENT_OBJ(-1ULL), component + 1);
if (error != 0)
break;
}
complen = strlen(component);
path -= complen;
ASSERT(path >= buf);
bcopy(component, path, complen);
obj = pobj;
if (sa_hdl != hdl) {
prevhdl = sa_hdl;
prevdb = sa_db;
}
error = zfs_grab_sa_handle(osp, obj, &sa_hdl, &sa_db, FTAG);
if (error != 0) {
sa_hdl = prevhdl;
sa_db = prevdb;
break;
}
}
if (sa_hdl != NULL && sa_hdl != hdl) {
ASSERT(sa_db != NULL);
zfs_release_sa_handle(sa_hdl, sa_db, FTAG);
}
if (error == 0)
(void) memmove(buf, path, buf + len - path);
return (error);
}
int
zfs_obj_to_path(objset_t *osp, uint64_t obj, char *buf, int len)
{
sa_attr_type_t *sa_table;
sa_handle_t *hdl;
dmu_buf_t *db;
int error;
error = zfs_sa_setup(osp, &sa_table);
if (error != 0)
return (error);
error = zfs_grab_sa_handle(osp, obj, &hdl, &db, FTAG);
if (error != 0)
return (error);
error = zfs_obj_to_path_impl(osp, obj, hdl, sa_table, buf, len);
zfs_release_sa_handle(hdl, db, FTAG);
return (error);
}
int
zfs_obj_to_stats(objset_t *osp, uint64_t obj, zfs_stat_t *sb,
char *buf, int len)
{
char *path = buf + len - 1;
sa_attr_type_t *sa_table;
sa_handle_t *hdl;
dmu_buf_t *db;
int error;
*path = '\0';
error = zfs_sa_setup(osp, &sa_table);
if (error != 0)
return (error);
error = zfs_grab_sa_handle(osp, obj, &hdl, &db, FTAG);
if (error != 0)
return (error);
error = zfs_obj_to_stats_impl(hdl, sa_table, sb);
if (error != 0) {
zfs_release_sa_handle(hdl, db, FTAG);
return (error);
}
error = zfs_obj_to_path_impl(osp, obj, hdl, sa_table, buf, len);
zfs_release_sa_handle(hdl, db, FTAG);
return (error);
}
#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
diff --git a/module/os/linux/zfs/zpl_inode.c b/module/os/linux/zfs/zpl_inode.c
index ce1524440ad0..4bd185baeb19 100644
--- a/module/os/linux/zfs/zpl_inode.c
+++ b/module/os/linux/zfs/zpl_inode.c
@@ -1,853 +1,854 @@
/*
* CDDL HEADER START
*
* The contents of this file are subject to the terms of the
* Common Development and Distribution License (the "License").
* You may not use this file except in compliance with the License.
*
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
* or http://www.opensolaris.org/os/licensing.
* See the License for the specific language governing permissions
* and limitations under the License.
*
* When distributing Covered Code, include this CDDL HEADER in each
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
* If applicable, add the following below this CDDL HEADER, with the
* fields enclosed by brackets "[]" replaced with your own identifying
* information: Portions Copyright [yyyy] [name of copyright owner]
*
* CDDL HEADER END
*/
/*
* Copyright (c) 2011, Lawrence Livermore National Security, LLC.
* Copyright (c) 2015 by Chunwei Chen. All rights reserved.
*/
#include <sys/zfs_ctldir.h>
#include <sys/zfs_vfsops.h>
#include <sys/zfs_vnops.h>
#include <sys/zfs_znode.h>
#include <sys/dmu_objset.h>
#include <sys/vfs.h>
#include <sys/zpl.h>
#include <sys/file.h>
static struct dentry *
zpl_lookup(struct inode *dir, struct dentry *dentry, unsigned int flags)
{
cred_t *cr = CRED();
struct inode *ip;
znode_t *zp;
int error;
fstrans_cookie_t cookie;
pathname_t *ppn = NULL;
pathname_t pn;
int zfs_flags = 0;
zfsvfs_t *zfsvfs = dentry->d_sb->s_fs_info;
if (dlen(dentry) >= ZAP_MAXNAMELEN)
return (ERR_PTR(-ENAMETOOLONG));
crhold(cr);
cookie = spl_fstrans_mark();
/* If we are a case insensitive fs, we need the real name */
if (zfsvfs->z_case == ZFS_CASE_INSENSITIVE) {
zfs_flags = FIGNORECASE;
pn_alloc(&pn);
ppn = &pn;
}
error = -zfs_lookup(ITOZ(dir), dname(dentry), &zp,
zfs_flags, cr, NULL, ppn);
spl_fstrans_unmark(cookie);
ASSERT3S(error, <=, 0);
crfree(cr);
spin_lock(&dentry->d_lock);
dentry->d_time = jiffies;
spin_unlock(&dentry->d_lock);
if (error) {
/*
* If we have a case sensitive fs, we do not want to
* insert negative entries, so return NULL for ENOENT.
* Fall through if the error is not ENOENT. Also free memory.
*/
if (ppn) {
pn_free(ppn);
if (error == -ENOENT)
return (NULL);
}
if (error == -ENOENT)
return (d_splice_alias(NULL, dentry));
else
return (ERR_PTR(error));
}
ip = ZTOI(zp);
/*
* If we are case insensitive, call the correct function
* to install the name.
*/
if (ppn) {
struct dentry *new_dentry;
struct qstr ci_name;
if (strcmp(dname(dentry), pn.pn_buf) == 0) {
new_dentry = d_splice_alias(ip, dentry);
} else {
ci_name.name = pn.pn_buf;
ci_name.len = strlen(pn.pn_buf);
new_dentry = d_add_ci(dentry, ip, &ci_name);
}
pn_free(ppn);
return (new_dentry);
} else {
return (d_splice_alias(ip, dentry));
}
}
void
zpl_vap_init(vattr_t *vap, struct inode *dir, umode_t mode, cred_t *cr)
{
vap->va_mask = ATTR_MODE;
vap->va_mode = mode;
vap->va_uid = crgetuid(cr);
if (dir && dir->i_mode & S_ISGID) {
vap->va_gid = KGID_TO_SGID(dir->i_gid);
if (S_ISDIR(mode))
vap->va_mode |= S_ISGID;
} else {
vap->va_gid = crgetgid(cr);
}
}
static int
#ifdef HAVE_IOPS_CREATE_USERNS
zpl_create(struct user_namespace *user_ns, struct inode *dir,
struct dentry *dentry, umode_t mode, bool flag)
#elif defined(HAVE_IOPS_CREATE_IDMAP)
zpl_create(struct mnt_idmap *user_ns, struct inode *dir,
struct dentry *dentry, umode_t mode, bool flag)
#else
zpl_create(struct inode *dir, struct dentry *dentry, umode_t mode, bool flag)
#endif
{
cred_t *cr = CRED();
znode_t *zp;
vattr_t *vap;
int error;
fstrans_cookie_t cookie;
crhold(cr);
vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
zpl_vap_init(vap, dir, mode, cr);
cookie = spl_fstrans_mark();
error = -zfs_create(ITOZ(dir), dname(dentry), vap, 0,
mode, &zp, cr, 0, NULL);
if (error == 0) {
error = zpl_xattr_security_init(ZTOI(zp), dir, &dentry->d_name);
if (error == 0)
error = zpl_init_acl(ZTOI(zp), dir);
if (error) {
(void) zfs_remove(ITOZ(dir), dname(dentry), cr, 0);
remove_inode_hash(ZTOI(zp));
iput(ZTOI(zp));
} else {
d_instantiate(dentry, ZTOI(zp));
}
}
spl_fstrans_unmark(cookie);
kmem_free(vap, sizeof (vattr_t));
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
static int
#ifdef HAVE_IOPS_MKNOD_USERNS
zpl_mknod(struct user_namespace *user_ns, struct inode *dir,
struct dentry *dentry, umode_t mode,
#elif defined(HAVE_IOPS_MKNOD_IDMAP)
zpl_mknod(struct mnt_idmap *user_ns, struct inode *dir,
struct dentry *dentry, umode_t mode,
#else
zpl_mknod(struct inode *dir, struct dentry *dentry, umode_t mode,
#endif
dev_t rdev)
{
cred_t *cr = CRED();
znode_t *zp;
vattr_t *vap;
int error;
fstrans_cookie_t cookie;
/*
* We currently expect Linux to supply rdev=0 for all sockets
* and fifos, but we want to know if this behavior ever changes.
*/
if (S_ISSOCK(mode) || S_ISFIFO(mode))
ASSERT(rdev == 0);
crhold(cr);
vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
zpl_vap_init(vap, dir, mode, cr);
vap->va_rdev = rdev;
cookie = spl_fstrans_mark();
error = -zfs_create(ITOZ(dir), dname(dentry), vap, 0,
mode, &zp, cr, 0, NULL);
if (error == 0) {
error = zpl_xattr_security_init(ZTOI(zp), dir, &dentry->d_name);
if (error == 0)
error = zpl_init_acl(ZTOI(zp), dir);
if (error) {
(void) zfs_remove(ITOZ(dir), dname(dentry), cr, 0);
remove_inode_hash(ZTOI(zp));
iput(ZTOI(zp));
} else {
d_instantiate(dentry, ZTOI(zp));
}
}
spl_fstrans_unmark(cookie);
kmem_free(vap, sizeof (vattr_t));
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
#ifdef HAVE_TMPFILE
static int
#ifdef HAVE_TMPFILE_IDMAP
zpl_tmpfile(struct mnt_idmap *userns, struct inode *dir,
struct file *file, umode_t mode)
#elif !defined(HAVE_TMPFILE_DENTRY)
zpl_tmpfile(struct user_namespace *userns, struct inode *dir,
struct file *file, umode_t mode)
#else
#ifdef HAVE_TMPFILE_USERNS
zpl_tmpfile(struct user_namespace *userns, struct inode *dir,
struct dentry *dentry, umode_t mode)
#else
zpl_tmpfile(struct inode *dir, struct dentry *dentry, umode_t mode)
#endif
#endif
{
cred_t *cr = CRED();
struct inode *ip;
vattr_t *vap;
int error;
fstrans_cookie_t cookie;
crhold(cr);
vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
/*
* The VFS does not apply the umask, therefore it is applied here
* when POSIX ACLs are not enabled.
*/
if (!IS_POSIXACL(dir))
mode &= ~current_umask();
zpl_vap_init(vap, dir, mode, cr);
cookie = spl_fstrans_mark();
error = -zfs_tmpfile(dir, vap, 0, mode, &ip, cr, 0, NULL);
if (error == 0) {
/* d_tmpfile will do drop_nlink, so we should set it first */
set_nlink(ip, 1);
#ifndef HAVE_TMPFILE_DENTRY
d_tmpfile(file, ip);
error = zpl_xattr_security_init(ip, dir,
&file->f_path.dentry->d_name);
#else
d_tmpfile(dentry, ip);
error = zpl_xattr_security_init(ip, dir, &dentry->d_name);
#endif
if (error == 0)
error = zpl_init_acl(ip, dir);
#ifndef HAVE_TMPFILE_DENTRY
error = finish_open_simple(file, error);
#endif
/*
* don't need to handle error here, file is already in
* unlinked set.
*/
}
spl_fstrans_unmark(cookie);
kmem_free(vap, sizeof (vattr_t));
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
#endif
static int
zpl_unlink(struct inode *dir, struct dentry *dentry)
{
cred_t *cr = CRED();
int error;
fstrans_cookie_t cookie;
zfsvfs_t *zfsvfs = dentry->d_sb->s_fs_info;
crhold(cr);
cookie = spl_fstrans_mark();
error = -zfs_remove(ITOZ(dir), dname(dentry), cr, 0);
/*
* For a CI FS we must invalidate the dentry to prevent the
* creation of negative entries.
*/
if (error == 0 && zfsvfs->z_case == ZFS_CASE_INSENSITIVE)
d_invalidate(dentry);
spl_fstrans_unmark(cookie);
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
static int
#ifdef HAVE_IOPS_MKDIR_USERNS
zpl_mkdir(struct user_namespace *user_ns, struct inode *dir,
struct dentry *dentry, umode_t mode)
#elif defined(HAVE_IOPS_MKDIR_IDMAP)
zpl_mkdir(struct mnt_idmap *user_ns, struct inode *dir,
struct dentry *dentry, umode_t mode)
#else
zpl_mkdir(struct inode *dir, struct dentry *dentry, umode_t mode)
#endif
{
cred_t *cr = CRED();
vattr_t *vap;
znode_t *zp;
int error;
fstrans_cookie_t cookie;
crhold(cr);
vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
zpl_vap_init(vap, dir, mode | S_IFDIR, cr);
cookie = spl_fstrans_mark();
error = -zfs_mkdir(ITOZ(dir), dname(dentry), vap, &zp, cr, 0, NULL);
if (error == 0) {
error = zpl_xattr_security_init(ZTOI(zp), dir, &dentry->d_name);
if (error == 0)
error = zpl_init_acl(ZTOI(zp), dir);
if (error) {
(void) zfs_rmdir(ITOZ(dir), dname(dentry), NULL, cr, 0);
remove_inode_hash(ZTOI(zp));
iput(ZTOI(zp));
} else {
d_instantiate(dentry, ZTOI(zp));
}
}
spl_fstrans_unmark(cookie);
kmem_free(vap, sizeof (vattr_t));
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
static int
zpl_rmdir(struct inode *dir, struct dentry *dentry)
{
cred_t *cr = CRED();
int error;
fstrans_cookie_t cookie;
zfsvfs_t *zfsvfs = dentry->d_sb->s_fs_info;
crhold(cr);
cookie = spl_fstrans_mark();
error = -zfs_rmdir(ITOZ(dir), dname(dentry), NULL, cr, 0);
/*
* For a CI FS we must invalidate the dentry to prevent the
* creation of negative entries.
*/
if (error == 0 && zfsvfs->z_case == ZFS_CASE_INSENSITIVE)
d_invalidate(dentry);
spl_fstrans_unmark(cookie);
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
static int
#ifdef HAVE_USERNS_IOPS_GETATTR
zpl_getattr_impl(struct user_namespace *user_ns,
const struct path *path, struct kstat *stat, u32 request_mask,
unsigned int query_flags)
#elif defined(HAVE_IDMAP_IOPS_GETATTR)
zpl_getattr_impl(struct mnt_idmap *user_ns,
const struct path *path, struct kstat *stat, u32 request_mask,
unsigned int query_flags)
#else
zpl_getattr_impl(const struct path *path, struct kstat *stat, u32 request_mask,
unsigned int query_flags)
#endif
{
int error;
fstrans_cookie_t cookie;
struct inode *ip = path->dentry->d_inode;
znode_t *zp __maybe_unused = ITOZ(ip);
cookie = spl_fstrans_mark();
/*
* XXX query_flags currently ignored.
*/
#ifdef HAVE_GENERIC_FILLATTR_IDMAP_REQMASK
error = -zfs_getattr_fast(user_ns, request_mask, ip, stat);
#elif (defined(HAVE_USERNS_IOPS_GETATTR) || defined(HAVE_IDMAP_IOPS_GETATTR))
error = -zfs_getattr_fast(user_ns, ip, stat);
#else
error = -zfs_getattr_fast(kcred->user_ns, ip, stat);
#endif
#ifdef STATX_BTIME
if (request_mask & STATX_BTIME) {
stat->btime = zp->z_btime;
stat->result_mask |= STATX_BTIME;
}
#endif
#ifdef STATX_ATTR_IMMUTABLE
if (zp->z_pflags & ZFS_IMMUTABLE)
stat->attributes |= STATX_ATTR_IMMUTABLE;
stat->attributes_mask |= STATX_ATTR_IMMUTABLE;
#endif
#ifdef STATX_ATTR_APPEND
if (zp->z_pflags & ZFS_APPENDONLY)
stat->attributes |= STATX_ATTR_APPEND;
stat->attributes_mask |= STATX_ATTR_APPEND;
#endif
#ifdef STATX_ATTR_NODUMP
if (zp->z_pflags & ZFS_NODUMP)
stat->attributes |= STATX_ATTR_NODUMP;
stat->attributes_mask |= STATX_ATTR_NODUMP;
#endif
spl_fstrans_unmark(cookie);
ASSERT3S(error, <=, 0);
return (error);
}
ZPL_GETATTR_WRAPPER(zpl_getattr);
static int
#ifdef HAVE_USERNS_IOPS_SETATTR
zpl_setattr(struct user_namespace *user_ns, struct dentry *dentry,
struct iattr *ia)
#elif defined(HAVE_IDMAP_IOPS_SETATTR)
zpl_setattr(struct mnt_idmap *user_ns, struct dentry *dentry,
struct iattr *ia)
#else
zpl_setattr(struct dentry *dentry, struct iattr *ia)
#endif
{
struct inode *ip = dentry->d_inode;
cred_t *cr = CRED();
vattr_t *vap;
int error;
fstrans_cookie_t cookie;
#ifdef HAVE_SETATTR_PREPARE_USERNS
error = zpl_setattr_prepare(user_ns, dentry, ia);
#elif defined(HAVE_SETATTR_PREPARE_IDMAP)
error = zpl_setattr_prepare(user_ns, dentry, ia);
#else
error = zpl_setattr_prepare(zfs_init_idmap, dentry, ia);
#endif
if (error)
return (error);
crhold(cr);
vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
vap->va_mask = ia->ia_valid & ATTR_IATTR_MASK;
vap->va_mode = ia->ia_mode;
vap->va_uid = KUID_TO_SUID(ia->ia_uid);
vap->va_gid = KGID_TO_SGID(ia->ia_gid);
vap->va_size = ia->ia_size;
vap->va_atime = ia->ia_atime;
vap->va_mtime = ia->ia_mtime;
vap->va_ctime = ia->ia_ctime;
if (vap->va_mask & ATTR_ATIME)
- ip->i_atime = zpl_inode_timestamp_truncate(ia->ia_atime, ip);
+ zpl_inode_set_atime_to_ts(ip,
+ zpl_inode_timestamp_truncate(ia->ia_atime, ip));
cookie = spl_fstrans_mark();
error = -zfs_setattr(ITOZ(ip), vap, 0, cr);
if (!error && (ia->ia_valid & ATTR_MODE))
error = zpl_chmod_acl(ip);
spl_fstrans_unmark(cookie);
kmem_free(vap, sizeof (vattr_t));
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
static int
#ifdef HAVE_IOPS_RENAME_USERNS
zpl_rename2(struct user_namespace *user_ns, struct inode *sdip,
struct dentry *sdentry, struct inode *tdip, struct dentry *tdentry,
unsigned int rflags)
#elif defined(HAVE_IOPS_RENAME_IDMAP)
zpl_rename2(struct mnt_idmap *user_ns, struct inode *sdip,
struct dentry *sdentry, struct inode *tdip, struct dentry *tdentry,
unsigned int rflags)
#else
zpl_rename2(struct inode *sdip, struct dentry *sdentry,
struct inode *tdip, struct dentry *tdentry, unsigned int rflags)
#endif
{
cred_t *cr = CRED();
int error;
fstrans_cookie_t cookie;
/* We don't have renameat2(2) support */
if (rflags)
return (-EINVAL);
crhold(cr);
cookie = spl_fstrans_mark();
error = -zfs_rename(ITOZ(sdip), dname(sdentry), ITOZ(tdip),
dname(tdentry), cr, 0);
spl_fstrans_unmark(cookie);
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
#if !defined(HAVE_IOPS_RENAME_USERNS) && \
!defined(HAVE_RENAME_WANTS_FLAGS) && \
!defined(HAVE_IOPS_RENAME_IDMAP)
static int
zpl_rename(struct inode *sdip, struct dentry *sdentry,
struct inode *tdip, struct dentry *tdentry)
{
return (zpl_rename2(sdip, sdentry, tdip, tdentry, 0));
}
#endif
static int
#ifdef HAVE_IOPS_SYMLINK_USERNS
zpl_symlink(struct user_namespace *user_ns, struct inode *dir,
struct dentry *dentry, const char *name)
#elif defined(HAVE_IOPS_SYMLINK_IDMAP)
zpl_symlink(struct mnt_idmap *user_ns, struct inode *dir,
struct dentry *dentry, const char *name)
#else
zpl_symlink(struct inode *dir, struct dentry *dentry, const char *name)
#endif
{
cred_t *cr = CRED();
vattr_t *vap;
znode_t *zp;
int error;
fstrans_cookie_t cookie;
crhold(cr);
vap = kmem_zalloc(sizeof (vattr_t), KM_SLEEP);
zpl_vap_init(vap, dir, S_IFLNK | S_IRWXUGO, cr);
cookie = spl_fstrans_mark();
error = -zfs_symlink(ITOZ(dir), dname(dentry), vap,
(char *)name, &zp, cr, 0);
if (error == 0) {
error = zpl_xattr_security_init(ZTOI(zp), dir, &dentry->d_name);
if (error) {
(void) zfs_remove(ITOZ(dir), dname(dentry), cr, 0);
remove_inode_hash(ZTOI(zp));
iput(ZTOI(zp));
} else {
d_instantiate(dentry, ZTOI(zp));
}
}
spl_fstrans_unmark(cookie);
kmem_free(vap, sizeof (vattr_t));
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
#if defined(HAVE_PUT_LINK_COOKIE)
static void
zpl_put_link(struct inode *unused, void *cookie)
{
kmem_free(cookie, MAXPATHLEN);
}
#elif defined(HAVE_PUT_LINK_NAMEIDATA)
static void
zpl_put_link(struct dentry *dentry, struct nameidata *nd, void *ptr)
{
const char *link = nd_get_link(nd);
if (!IS_ERR(link))
kmem_free(link, MAXPATHLEN);
}
#elif defined(HAVE_PUT_LINK_DELAYED)
static void
zpl_put_link(void *ptr)
{
kmem_free(ptr, MAXPATHLEN);
}
#endif
static int
zpl_get_link_common(struct dentry *dentry, struct inode *ip, char **link)
{
fstrans_cookie_t cookie;
cred_t *cr = CRED();
int error;
crhold(cr);
*link = NULL;
struct iovec iov;
iov.iov_len = MAXPATHLEN;
iov.iov_base = kmem_zalloc(MAXPATHLEN, KM_SLEEP);
zfs_uio_t uio;
zfs_uio_iovec_init(&uio, &iov, 1, 0, UIO_SYSSPACE, MAXPATHLEN - 1, 0);
cookie = spl_fstrans_mark();
error = -zfs_readlink(ip, &uio, cr);
spl_fstrans_unmark(cookie);
crfree(cr);
if (error)
kmem_free(iov.iov_base, MAXPATHLEN);
else
*link = iov.iov_base;
return (error);
}
#if defined(HAVE_GET_LINK_DELAYED)
static const char *
zpl_get_link(struct dentry *dentry, struct inode *inode,
struct delayed_call *done)
{
char *link = NULL;
int error;
if (!dentry)
return (ERR_PTR(-ECHILD));
error = zpl_get_link_common(dentry, inode, &link);
if (error)
return (ERR_PTR(error));
set_delayed_call(done, zpl_put_link, link);
return (link);
}
#elif defined(HAVE_GET_LINK_COOKIE)
static const char *
zpl_get_link(struct dentry *dentry, struct inode *inode, void **cookie)
{
char *link = NULL;
int error;
if (!dentry)
return (ERR_PTR(-ECHILD));
error = zpl_get_link_common(dentry, inode, &link);
if (error)
return (ERR_PTR(error));
return (*cookie = link);
}
#elif defined(HAVE_FOLLOW_LINK_COOKIE)
static const char *
zpl_follow_link(struct dentry *dentry, void **cookie)
{
char *link = NULL;
int error;
error = zpl_get_link_common(dentry, dentry->d_inode, &link);
if (error)
return (ERR_PTR(error));
return (*cookie = link);
}
#elif defined(HAVE_FOLLOW_LINK_NAMEIDATA)
static void *
zpl_follow_link(struct dentry *dentry, struct nameidata *nd)
{
char *link = NULL;
int error;
error = zpl_get_link_common(dentry, dentry->d_inode, &link);
if (error)
nd_set_link(nd, ERR_PTR(error));
else
nd_set_link(nd, link);
return (NULL);
}
#endif
static int
zpl_link(struct dentry *old_dentry, struct inode *dir, struct dentry *dentry)
{
cred_t *cr = CRED();
struct inode *ip = old_dentry->d_inode;
int error;
fstrans_cookie_t cookie;
if (ip->i_nlink >= ZFS_LINK_MAX)
return (-EMLINK);
crhold(cr);
zpl_inode_set_ctime_to_ts(ip, current_time(ip));
/* Must have an existing ref, so igrab() cannot return NULL */
VERIFY3P(igrab(ip), !=, NULL);
cookie = spl_fstrans_mark();
error = -zfs_link(ITOZ(dir), ITOZ(ip), dname(dentry), cr, 0);
if (error) {
iput(ip);
goto out;
}
d_instantiate(dentry, ip);
out:
spl_fstrans_unmark(cookie);
crfree(cr);
ASSERT3S(error, <=, 0);
return (error);
}
const struct inode_operations zpl_inode_operations = {
.setattr = zpl_setattr,
.getattr = zpl_getattr,
#ifdef HAVE_GENERIC_SETXATTR
.setxattr = generic_setxattr,
.getxattr = generic_getxattr,
.removexattr = generic_removexattr,
#endif
.listxattr = zpl_xattr_list,
#if defined(CONFIG_FS_POSIX_ACL)
#if defined(HAVE_SET_ACL)
.set_acl = zpl_set_acl,
#endif /* HAVE_SET_ACL */
#if defined(HAVE_GET_INODE_ACL)
.get_inode_acl = zpl_get_acl,
#else
.get_acl = zpl_get_acl,
#endif /* HAVE_GET_INODE_ACL */
#endif /* CONFIG_FS_POSIX_ACL */
};
const struct inode_operations zpl_dir_inode_operations = {
.create = zpl_create,
.lookup = zpl_lookup,
.link = zpl_link,
.unlink = zpl_unlink,
.symlink = zpl_symlink,
.mkdir = zpl_mkdir,
.rmdir = zpl_rmdir,
.mknod = zpl_mknod,
#if defined(HAVE_RENAME_WANTS_FLAGS) || defined(HAVE_IOPS_RENAME_USERNS)
.rename = zpl_rename2,
#elif defined(HAVE_IOPS_RENAME_IDMAP)
.rename = zpl_rename2,
#else
.rename = zpl_rename,
#endif
#ifdef HAVE_TMPFILE
.tmpfile = zpl_tmpfile,
#endif
.setattr = zpl_setattr,
.getattr = zpl_getattr,
#ifdef HAVE_GENERIC_SETXATTR
.setxattr = generic_setxattr,
.getxattr = generic_getxattr,
.removexattr = generic_removexattr,
#endif
.listxattr = zpl_xattr_list,
#if defined(CONFIG_FS_POSIX_ACL)
#if defined(HAVE_SET_ACL)
.set_acl = zpl_set_acl,
#endif /* HAVE_SET_ACL */
#if defined(HAVE_GET_INODE_ACL)
.get_inode_acl = zpl_get_acl,
#else
.get_acl = zpl_get_acl,
#endif /* HAVE_GET_INODE_ACL */
#endif /* CONFIG_FS_POSIX_ACL */
};
const struct inode_operations zpl_symlink_inode_operations = {
#ifdef HAVE_GENERIC_READLINK
.readlink = generic_readlink,
#endif
#if defined(HAVE_GET_LINK_DELAYED) || defined(HAVE_GET_LINK_COOKIE)
.get_link = zpl_get_link,
#elif defined(HAVE_FOLLOW_LINK_COOKIE) || defined(HAVE_FOLLOW_LINK_NAMEIDATA)
.follow_link = zpl_follow_link,
#endif
#if defined(HAVE_PUT_LINK_COOKIE) || defined(HAVE_PUT_LINK_NAMEIDATA)
.put_link = zpl_put_link,
#endif
.setattr = zpl_setattr,
.getattr = zpl_getattr,
#ifdef HAVE_GENERIC_SETXATTR
.setxattr = generic_setxattr,
.getxattr = generic_getxattr,
.removexattr = generic_removexattr,
#endif
.listxattr = zpl_xattr_list,
};
const struct inode_operations zpl_special_inode_operations = {
.setattr = zpl_setattr,
.getattr = zpl_getattr,
#ifdef HAVE_GENERIC_SETXATTR
.setxattr = generic_setxattr,
.getxattr = generic_getxattr,
.removexattr = generic_removexattr,
#endif
.listxattr = zpl_xattr_list,
#if defined(CONFIG_FS_POSIX_ACL)
#if defined(HAVE_SET_ACL)
.set_acl = zpl_set_acl,
#endif /* HAVE_SET_ACL */
#if defined(HAVE_GET_INODE_ACL)
.get_inode_acl = zpl_get_acl,
#else
.get_acl = zpl_get_acl,
#endif /* HAVE_GET_INODE_ACL */
#endif /* CONFIG_FS_POSIX_ACL */
};