4 * The contents of this file are subject to the terms of the
5 * Common Development and Distribution License (the "License").
6 * You may not use this file except in compliance with the License.
8 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
9 * or http://www.opensolaris.org/os/licensing.
10 * See the License for the specific language governing permissions
11 * and limitations under the License.
13 * When distributing Covered Code, include this CDDL HEADER in each
14 * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
15 * If applicable, add the following below this CDDL HEADER, with the
16 * fields enclosed by brackets "[]" replaced with your own identifying
17 * information: Portions Copyright [yyyy] [name of copyright owner]
22 * Copyright (c) 2011, Lawrence Livermore National Security, LLC.
26 #include <sys/zfs_vfsops.h>
27 #include <sys/zfs_vnops.h>
28 #include <sys/zfs_znode.h>
29 #include <sys/zfs_ctldir.h>
34 zpl_inode_alloc(struct super_block *sb)
38 VERIFY3S(zfs_inode_alloc(sb, &ip), ==, 0);
45 zpl_inode_destroy(struct inode *ip)
47 ASSERT(atomic_read(&ip->i_count) == 0);
48 zfs_inode_destroy(ip);
52 * When ->drop_inode() is called its return value indicates if the
53 * inode should be evicted from the inode cache. If the inode is
54 * unhashed and has no links the default policy is to evict it
57 * Prior to 2.6.36 this eviction was accomplished by the vfs calling
58 * ->delete_inode(). It was ->delete_inode()'s responsibility to
59 * truncate the inode pages and call clear_inode(). The call to
60 * clear_inode() synchronously invalidates all the buffers and
61 * calls ->clear_inode(). It was ->clear_inode()'s responsibility
62 * to cleanup and filesystem specific data before freeing the inode.
64 * This elaborate mechanism was replaced by ->evict_inode() which
65 * does the job of both ->delete_inode() and ->clear_inode(). It
66 * will be called exactly once, and when it returns the inode must
67 * be in a state where it can simply be freed. The ->evict_inode()
68 * callback must minimally truncate the inode pages, and call
69 * end_writeback() to complete all outstanding writeback for the
70 * inode. After this is complete evict inode can cleanup any
71 * remaining filesystem specific data.
73 #ifdef HAVE_EVICT_INODE
75 zpl_evict_inode(struct inode *ip)
77 truncate_setsize(ip, 0);
85 zpl_clear_inode(struct inode *ip)
91 zpl_inode_delete(struct inode *ip)
93 truncate_setsize(ip, 0);
97 #endif /* HAVE_EVICT_INODE */
100 zpl_put_super(struct super_block *sb)
104 error = -zfs_umount(sb);
105 ASSERT3S(error, <=, 0);
109 zpl_sync_fs(struct super_block *sb, int wait)
115 error = -zfs_sync(sb, wait, cr);
117 ASSERT3S(error, <=, 0);
123 zpl_statfs(struct dentry *dentry, struct kstatfs *statp)
127 error = -zfs_statvfs(dentry, statp);
128 ASSERT3S(error, <=, 0);
134 zpl_remount_fs(struct super_block *sb, int *flags, char *data)
137 error = -zfs_remount(sb, flags, data);
138 ASSERT3S(error, <=, 0);
144 zpl_umount_begin(struct super_block *sb)
146 zfs_sb_t *zsb = sb->s_fs_info;
150 * Best effort to unmount snapshots in .zfs/snapshot/. Normally this
151 * isn't required because snapshots have the MNT_SHRINKABLE flag set.
154 (void) zfsctl_unmount_snapshots(zsb, MNT_FORCE, &count);
158 * The Linux VFS automatically handles the following flags:
159 * MNT_NOSUID, MNT_NODEV, MNT_NOEXEC, MNT_NOATIME, MNT_READONLY
161 #ifdef HAVE_SHOW_OPTIONS_WITH_DENTRY
163 zpl_show_options(struct seq_file *seq, struct dentry *root)
165 zfs_sb_t *zsb = root->d_sb->s_fs_info;
167 seq_printf(seq, ",%s", zsb->z_flags & ZSB_XATTR ? "xattr" : "noxattr");
173 zpl_show_options(struct seq_file *seq, struct vfsmount *vfsp)
175 zfs_sb_t *zsb = vfsp->mnt_sb->s_fs_info;
177 seq_printf(seq, ",%s", zsb->z_flags & ZSB_XATTR ? "xattr" : "noxattr");
181 #endif /* HAVE_SHOW_OPTIONS_WITH_DENTRY */
184 zpl_fill_super(struct super_block *sb, void *data, int silent)
188 error = -zfs_domount(sb, data, silent);
189 ASSERT3S(error, <=, 0);
194 #ifdef HAVE_MOUNT_NODEV
195 static struct dentry *
196 zpl_mount(struct file_system_type *fs_type, int flags,
197 const char *osname, void *data)
199 zpl_mount_data_t zmd = { osname, data };
201 return mount_nodev(fs_type, flags, &zmd, zpl_fill_super);
205 zpl_get_sb(struct file_system_type *fs_type, int flags,
206 const char *osname, void *data, struct vfsmount *mnt)
208 zpl_mount_data_t zmd = { osname, data };
210 return get_sb_nodev(fs_type, flags, &zmd, zpl_fill_super, mnt);
212 #endif /* HAVE_MOUNT_NODEV */
215 zpl_kill_sb(struct super_block *sb)
223 * Linux 3.1 - 3.x API
225 * The Linux 3.1 API introduced per-sb cache shrinkers to replace the
226 * global ones. This allows us a mechanism to cleanly target a specific
227 * zfs file system when the dnode and inode caches grow too large.
229 * In addition, the 3.0 kernel added the iterate_supers_type() helper
230 * function which is used to safely walk all of the zfs file systems.
233 zpl_prune_sb(struct super_block *sb, void *arg)
238 error = -zfs_sb_prune(sb, *(unsigned long *)arg, &objects);
239 ASSERT3S(error, <=, 0);
245 zpl_prune_sbs(int64_t bytes_to_scan, void *private)
247 unsigned long nr_to_scan = (bytes_to_scan / sizeof(znode_t));
249 iterate_supers_type(&zpl_fs_type, zpl_prune_sb, &nr_to_scan);
254 * Linux 2.6.x - 3.0 API
256 * These are best effort interfaces are provided by the SPL to induce
257 * the Linux VM subsystem to reclaim a fraction of the both dnode and
258 * inode caches. Ideally, we want to just target the zfs file systems
259 * however our only option is to reclaim from them all.
262 zpl_prune_sbs(int64_t bytes_to_scan, void *private)
264 unsigned long nr_to_scan = (bytes_to_scan / sizeof(znode_t));
266 shrink_dcache_memory(nr_to_scan, GFP_KERNEL);
267 shrink_icache_memory(nr_to_scan, GFP_KERNEL);
270 #endif /* HAVE_SHRINK */
272 #ifdef HAVE_NR_CACHED_OBJECTS
274 zpl_nr_cached_objects(struct super_block *sb)
276 zfs_sb_t *zsb = sb->s_fs_info;
279 mutex_enter(&zsb->z_znodes_lock);
280 nr = zsb->z_nr_znodes;
281 mutex_exit(&zsb->z_znodes_lock);
285 #endif /* HAVE_NR_CACHED_OBJECTS */
287 #ifdef HAVE_FREE_CACHED_OBJECTS
289 * Attempt to evict some meta data from the cache. The ARC operates in
290 * terms of bytes while the Linux VFS uses objects. Now because this is
291 * just a best effort eviction and the exact values aren't critical so we
292 * extrapolate from an object count to a byte size using the znode_t size.
295 zpl_free_cached_objects(struct super_block *sb, int nr_to_scan)
297 arc_adjust_meta(nr_to_scan * sizeof(znode_t), B_FALSE);
299 #endif /* HAVE_FREE_CACHED_OBJECTS */
301 const struct super_operations zpl_super_operations = {
302 .alloc_inode = zpl_inode_alloc,
303 .destroy_inode = zpl_inode_destroy,
307 #ifdef HAVE_EVICT_INODE
308 .evict_inode = zpl_evict_inode,
310 .clear_inode = zpl_clear_inode,
311 .delete_inode = zpl_inode_delete,
312 #endif /* HAVE_EVICT_INODE */
313 .put_super = zpl_put_super,
315 .sync_fs = zpl_sync_fs,
316 .statfs = zpl_statfs,
317 .remount_fs = zpl_remount_fs,
318 .umount_begin = zpl_umount_begin,
319 .show_options = zpl_show_options,
321 #ifdef HAVE_NR_CACHED_OBJECTS
322 .nr_cached_objects = zpl_nr_cached_objects,
323 #endif /* HAVE_NR_CACHED_OBJECTS */
324 #ifdef HAVE_FREE_CACHED_OBJECTS
325 .free_cached_objects = zpl_free_cached_objects,
326 #endif /* HAVE_FREE_CACHED_OBJECTS */
329 struct file_system_type zpl_fs_type = {
330 .owner = THIS_MODULE,
332 #ifdef HAVE_MOUNT_NODEV
335 .get_sb = zpl_get_sb,
336 #endif /* HAVE_MOUNT_NODEV */
337 .kill_sb = zpl_kill_sb,