[linux-2.6-block.git] / fs / afs / inode.c

/*
 * Copyright (c) 2002 Red Hat, Inc. All rights reserved.
 *
 * This software may be freely redistributed under the terms of the
 * GNU General Public License.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
 *
 * Authors: David Woodhouse <dwmw2@infradead.org>
 *          David Howells <dhowells@redhat.com>
 *
 */

#include <linux/kernel.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/fs.h>
#include <linux/pagemap.h>
#include <linux/sched.h>
#include <linux/mount.h>
#include <linux/namei.h>
#include <linux/iversion.h>
#include "internal.h"

static const struct inode_operations afs_symlink_inode_operations = {
	.get_link	= page_get_link,
	.listxattr	= afs_listxattr,
};

/*
 * map the AFS file status to the inode member variables
 */
static int afs_inode_map_status(struct afs_vnode *vnode, struct key *key)
{
	struct inode *inode = AFS_VNODE_TO_I(vnode);
	bool changed;

	_debug("FS: ft=%d lk=%d sz=%llu ver=%Lu mod=%hu",
	       vnode->status.type,
	       vnode->status.nlink,
	       (unsigned long long) vnode->status.size,
	       vnode->status.data_version,
	       vnode->status.mode);

	read_seqlock_excl(&vnode->cb_lock);

	switch (vnode->status.type) {
	case AFS_FTYPE_FILE:
		inode->i_mode	= S_IFREG | vnode->status.mode;
		inode->i_op	= &afs_file_inode_operations;
		inode->i_fop	= &afs_file_operations;
		break;
	case AFS_FTYPE_DIR:
		inode->i_mode	= S_IFDIR | vnode->status.mode;
		inode->i_op	= &afs_dir_inode_operations;
		inode->i_fop	= &afs_dir_file_operations;
		break;
	case AFS_FTYPE_SYMLINK:
		/* Symlinks with a mode of 0644 are actually mountpoints. */
		if ((vnode->status.mode & 0777) == 0644) {
			inode->i_flags |= S_AUTOMOUNT;

			set_bit(AFS_VNODE_MOUNTPOINT, &vnode->flags);

			inode->i_mode	= S_IFDIR | 0555;
			inode->i_op	= &afs_mntpt_inode_operations;
			inode->i_fop	= &afs_mntpt_file_operations;
		} else {
			inode->i_mode	= S_IFLNK | vnode->status.mode;
			inode->i_op	= &afs_symlink_inode_operations;
		}
		inode_nohighmem(inode);
		break;
	default:
		printk("kAFS: AFS vnode with undefined type\n");
		read_sequnlock_excl(&vnode->cb_lock);
		return -EBADMSG;
	}

	changed = (vnode->status.size != inode->i_size);

	set_nlink(inode, vnode->status.nlink);
	inode->i_uid		= vnode->status.owner;
	inode->i_gid            = vnode->status.group;
	inode->i_size		= vnode->status.size;
	inode->i_ctime.tv_sec	= vnode->status.mtime_client;
	inode->i_ctime.tv_nsec	= 0;
	inode->i_atime		= inode->i_mtime = inode->i_ctime;
	inode->i_blocks		= 0;
	inode->i_generation	= vnode->fid.unique;
	inode_set_iversion_raw(inode, vnode->status.data_version);
	inode->i_mapping->a_ops	= &afs_fs_aops;

	read_sequnlock_excl(&vnode->cb_lock);

#ifdef CONFIG_AFS_FSCACHE
	if (changed)
		fscache_attr_changed(vnode->cache);
#endif
	return 0;
}

/*
 * Fetch file status from the volume.
 */
int afs_fetch_status(struct afs_vnode *vnode, struct key *key)
{
	struct afs_fs_cursor fc;
	int ret;

	_enter("%s,{%x:%u.%u,S=%lx}",
	       vnode->volume->name,
	       vnode->fid.vid, vnode->fid.vnode, vnode->fid.unique,
	       vnode->flags);

	ret = -ERESTARTSYS;
	if (afs_begin_vnode_operation(&fc, vnode, key)) {
		while (afs_select_fileserver(&fc)) {
			fc.cb_break = vnode->cb_break + vnode->cb_s_break;
			afs_fs_fetch_file_status(&fc, NULL);
		}

		afs_check_for_remote_deletion(&fc, fc.vnode);
		afs_vnode_commit_status(&fc, vnode, fc.cb_break);
		ret = afs_end_vnode_operation(&fc);
	}

	_leave(" = %d", ret);
	return ret;
}

/*
 * iget5() comparator
 */
int afs_iget5_test(struct inode *inode, void *opaque)
{
	struct afs_iget_data *data = opaque;

	return inode->i_ino == data->fid.vnode &&
		inode->i_generation == data->fid.unique;
}

/*
 * iget5() comparator for inode created by autocell operations
 *
 * These pseudo inodes don't match anything.
 */
static int afs_iget5_pseudo_dir_test(struct inode *inode, void *opaque)
{
	return 0;
}

/*
 * iget5() inode initialiser
 */
static int afs_iget5_set(struct inode *inode, void *opaque)
{
	struct afs_iget_data *data = opaque;
	struct afs_vnode *vnode = AFS_FS_I(inode);

	inode->i_ino = data->fid.vnode;
	inode->i_generation = data->fid.unique;
	vnode->fid = data->fid;
	vnode->volume = data->volume;

	return 0;
}

/*
 * Create an inode for a dynamic root directory or an autocell dynamic
 * automount dir.
 */
struct inode *afs_iget_pseudo_dir(struct super_block *sb, bool root)
{
	struct afs_iget_data data;
	struct afs_super_info *as;
	struct afs_vnode *vnode;
	struct inode *inode;
	static atomic_t afs_autocell_ino;

	_enter("");

	as = sb->s_fs_info;
	if (as->volume) {
		data.volume = as->volume;
		data.fid.vid = as->volume->vid;
	}
	if (root) {
		data.fid.vnode = 1;
		data.fid.unique = 1;
	} else {
		data.fid.vnode = atomic_inc_return(&afs_autocell_ino);
		data.fid.unique = 0;
	}

	inode = iget5_locked(sb, data.fid.vnode,
			     afs_iget5_pseudo_dir_test, afs_iget5_set,
			     &data);
	if (!inode) {
		_leave(" = -ENOMEM");
		return ERR_PTR(-ENOMEM);
	}

	_debug("GOT INODE %p { ino=%lu, vl=%x, vn=%x, u=%x }",
	       inode, inode->i_ino, data.fid.vid, data.fid.vnode,
	       data.fid.unique);

	vnode = AFS_FS_I(inode);

	/* there shouldn't be an existing inode */
	BUG_ON(!(inode->i_state & I_NEW));

	inode->i_size		= 0;
	inode->i_mode		= S_IFDIR | S_IRUGO | S_IXUGO;
	if (root) {
		inode->i_op	= &afs_dynroot_inode_operations;
		inode->i_fop	= &afs_dynroot_file_operations;
	} else {
		inode->i_op	= &afs_autocell_inode_operations;
	}
	set_nlink(inode, 2);
	inode->i_uid		= GLOBAL_ROOT_UID;
	inode->i_gid		= GLOBAL_ROOT_GID;
	inode->i_ctime.tv_sec	= get_seconds();
	inode->i_ctime.tv_nsec	= 0;
	inode->i_atime		= inode->i_mtime = inode->i_ctime;
	inode->i_blocks		= 0;
	inode_set_iversion_raw(inode, 0);
	inode->i_generation	= 0;

	set_bit(AFS_VNODE_PSEUDODIR, &vnode->flags);
	if (!root) {
		set_bit(AFS_VNODE_MOUNTPOINT, &vnode->flags);
		inode->i_flags |= S_AUTOMOUNT;
	}

	inode->i_flags |= S_NOATIME;
	unlock_new_inode(inode);
	_leave(" = %p", inode);
	return inode;
}

/*
 * inode retrieval
 */
struct inode *afs_iget(struct super_block *sb, struct key *key,
		       struct afs_fid *fid, struct afs_file_status *status,
		       struct afs_callback *cb, struct afs_cb_interest *cbi)
{
	struct afs_iget_data data = { .fid = *fid };
	struct afs_super_info *as;
	struct afs_vnode *vnode;
	struct inode *inode;
	int ret;

	_enter(",{%x:%u.%u},,", fid->vid, fid->vnode, fid->unique);

	as = sb->s_fs_info;
	data.volume = as->volume;

	inode = iget5_locked(sb, fid->vnode, afs_iget5_test, afs_iget5_set,
			     &data);
	if (!inode) {
		_leave(" = -ENOMEM");
		return ERR_PTR(-ENOMEM);
	}

	_debug("GOT INODE %p { vl=%x vn=%x, u=%x }",
	       inode, fid->vid, fid->vnode, fid->unique);

	vnode = AFS_FS_I(inode);

	/* deal with an existing inode */
	if (!(inode->i_state & I_NEW)) {
		_leave(" = %p", inode);
		return inode;
	}

	if (!status) {
		/* it's a remotely extant inode */
		ret = afs_fetch_status(vnode, key);
		if (ret < 0)
			goto bad_inode;
	} else {
		/* it's an inode we just created */
		memcpy(&vnode->status, status, sizeof(vnode->status));

		if (!cb) {
			/* it's a symlink we just created (the fileserver
			 * didn't give us a callback) */
			vnode->cb_version = 0;
			vnode->cb_type = 0;
			vnode->cb_expires_at = 0;
		} else {
			vnode->cb_version = cb->version;
			vnode->cb_type = cb->type;
			vnode->cb_expires_at = cb->expiry;
			vnode->cb_interest = afs_get_cb_interest(cbi);
			set_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);
		}

		vnode->cb_expires_at += ktime_get_real_seconds();
	}

	/* set up caching before mapping the status, as map-status reads the
	 * first page of symlinks to see if they're really mountpoints */
	inode->i_size = vnode->status.size;
#ifdef CONFIG_AFS_FSCACHE
	vnode->cache = fscache_acquire_cookie(vnode->volume->cache,
					      &afs_vnode_cache_index_def,
					      vnode, true);
#endif

	ret = afs_inode_map_status(vnode, key);
	if (ret < 0)
		goto bad_inode;

	/* success */
	clear_bit(AFS_VNODE_UNSET, &vnode->flags);
	inode->i_flags |= S_NOATIME;
	unlock_new_inode(inode);
	_leave(" = %p [CB { v=%u t=%u }]", inode, vnode->cb_version, vnode->cb_type);
	return inode;

	/* failure */
bad_inode:
#ifdef CONFIG_AFS_FSCACHE
	fscache_relinquish_cookie(vnode->cache, 0);
	vnode->cache = NULL;
#endif
	iget_failed(inode);
	_leave(" = %d [bad]", ret);
	return ERR_PTR(ret);
}

/*
 * mark the data attached to an inode as obsolete due to a write on the server
 * - might also want to ditch all the outstanding writes and dirty pages
 */
void afs_zap_data(struct afs_vnode *vnode)
{
	_enter("{%x:%u}", vnode->fid.vid, vnode->fid.vnode);

	/* nuke all the non-dirty pages that aren't locked, mapped or being
	 * written back in a regular file and completely discard the pages in a
	 * directory or symlink */
	if (S_ISREG(vnode->vfs_inode.i_mode))
		invalidate_remote_inode(&vnode->vfs_inode);
	else
		invalidate_inode_pages2(vnode->vfs_inode.i_mapping);
}

/*
 * validate a vnode/inode
 * - there are several things we need to check
 *   - parent dir data changes (rm, rmdir, rename, mkdir, create, link,
 *     symlink)
 *   - parent dir metadata changed (security changes)
 *   - dentry data changed (write, truncate)
 *   - dentry metadata changed (security changes)
 */
int afs_validate(struct afs_vnode *vnode, struct key *key)
{
	time64_t now = ktime_get_real_seconds();
	bool valid = false;
	int ret;

	_enter("{v={%x:%u} fl=%lx},%x",
	       vnode->fid.vid, vnode->fid.vnode, vnode->flags,
	       key_serial(key));

	/* Quickly check the callback state.  Ideally, we'd use read_seqbegin
	 * here, but we have no way to pass the net namespace to the RCU
	 * cleanup for the server record.
	 */
	read_seqlock_excl(&vnode->cb_lock);

	if (test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
		if (vnode->cb_s_break != vnode->cb_interest->server->cb_s_break) {
			vnode->cb_s_break = vnode->cb_interest->server->cb_s_break;
		} else if (!test_bit(AFS_VNODE_DIR_MODIFIED, &vnode->flags) &&
			   !test_bit(AFS_VNODE_ZAP_DATA, &vnode->flags) &&
			   vnode->cb_expires_at - 10 > now) {
				valid = true;
		}
	} else if (test_bit(AFS_VNODE_DELETED, &vnode->flags)) {
		valid = true;
	}

	read_sequnlock_excl(&vnode->cb_lock);

	if (test_bit(AFS_VNODE_DELETED, &vnode->flags))
		clear_nlink(&vnode->vfs_inode);

	if (valid)
		goto valid;

	mutex_lock(&vnode->validate_lock);

	/* if the promise has expired, we need to check the server again to get
	 * a new promise - note that if the (parent) directory's metadata was
	 * changed then the security may be different and we may no longer have
	 * access */
	if (!test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
		_debug("not promised");
		ret = afs_fetch_status(vnode, key);
		if (ret < 0) {
			if (ret == -ENOENT) {
				set_bit(AFS_VNODE_DELETED, &vnode->flags);
				ret = -ESTALE;
			}
			goto error_unlock;
		}
		_debug("new promise [fl=%lx]", vnode->flags);
	}

	if (test_bit(AFS_VNODE_DELETED, &vnode->flags)) {
		_debug("file already deleted");
		ret = -ESTALE;
		goto error_unlock;
	}

	/* if the vnode's data version number changed then its contents are
	 * different */
	if (test_and_clear_bit(AFS_VNODE_ZAP_DATA, &vnode->flags))
		afs_zap_data(vnode);

	clear_bit(AFS_VNODE_DIR_MODIFIED, &vnode->flags);
	mutex_unlock(&vnode->validate_lock);
valid:
	_leave(" = 0");
	return 0;

error_unlock:
	mutex_unlock(&vnode->validate_lock);
	_leave(" = %d", ret);
	return ret;
}

/*
 * read the attributes of an inode
 */
int afs_getattr(const struct path *path, struct kstat *stat,
		u32 request_mask, unsigned int query_flags)
{
	struct inode *inode = d_inode(path->dentry);
	struct afs_vnode *vnode = AFS_FS_I(inode);
	int seq = 0;

	_enter("{ ino=%lu v=%u }", inode->i_ino, inode->i_generation);

	do {
		read_seqbegin_or_lock(&vnode->cb_lock, &seq);
		generic_fillattr(inode, stat);
	} while (need_seqretry(&vnode->cb_lock, seq));

	done_seqretry(&vnode->cb_lock, seq);
	return 0;
}

/*
 * discard an AFS inode
 */
int afs_drop_inode(struct inode *inode)
{
	_enter("");

	if (test_bit(AFS_VNODE_PSEUDODIR, &AFS_FS_I(inode)->flags))
		return generic_delete_inode(inode);
	else
		return generic_drop_inode(inode);
}

/*
 * clear an AFS inode
 */
void afs_evict_inode(struct inode *inode)
{
	struct afs_vnode *vnode;

	vnode = AFS_FS_I(inode);

	_enter("{%x:%u.%d}",
	       vnode->fid.vid,
	       vnode->fid.vnode,
	       vnode->fid.unique);

	_debug("CLEAR INODE %p", inode);

	ASSERTCMP(inode->i_ino, ==, vnode->fid.vnode);

	truncate_inode_pages_final(&inode->i_data);
	clear_inode(inode);

	if (vnode->cb_interest) {
		afs_put_cb_interest(afs_i2net(inode), vnode->cb_interest);
		vnode->cb_interest = NULL;
	}

	while (!list_empty(&vnode->wb_keys)) {
		struct afs_wb_key *wbk = list_entry(vnode->wb_keys.next,
						    struct afs_wb_key, vnode_link);
		list_del(&wbk->vnode_link);
		afs_put_wb_key(wbk);
	}

#ifdef CONFIG_AFS_FSCACHE
	fscache_relinquish_cookie(vnode->cache, 0);
	vnode->cache = NULL;
#endif

	afs_put_permits(vnode->permit_cache);
	_leave("");
}

/*
 * set the attributes of an inode
 */
int afs_setattr(struct dentry *dentry, struct iattr *attr)
{
	struct afs_fs_cursor fc;
	struct afs_vnode *vnode = AFS_FS_I(d_inode(dentry));
	struct key *key;
	int ret;

	_enter("{%x:%u},{n=%pd},%x",
	       vnode->fid.vid, vnode->fid.vnode, dentry,
	       attr->ia_valid);

	if (!(attr->ia_valid & (ATTR_SIZE | ATTR_MODE | ATTR_UID | ATTR_GID |
				ATTR_MTIME))) {
		_leave(" = 0 [unsupported]");
		return 0;
	}

	/* flush any dirty data outstanding on a regular file */
	if (S_ISREG(vnode->vfs_inode.i_mode))
		filemap_write_and_wait(vnode->vfs_inode.i_mapping);

	if (attr->ia_valid & ATTR_FILE) {
		key = afs_file_key(attr->ia_file);
	} else {
		key = afs_request_key(vnode->volume->cell);
		if (IS_ERR(key)) {
			ret = PTR_ERR(key);
			goto error;
		}
	}

	ret = -ERESTARTSYS;
	if (afs_begin_vnode_operation(&fc, vnode, key)) {
		while (afs_select_fileserver(&fc)) {
			fc.cb_break = vnode->cb_break + vnode->cb_s_break;
			afs_fs_setattr(&fc, attr);
		}

		afs_check_for_remote_deletion(&fc, fc.vnode);
		afs_vnode_commit_status(&fc, vnode, fc.cb_break);
		ret = afs_end_vnode_operation(&fc);
	}

	if (!(attr->ia_valid & ATTR_FILE))
		key_put(key);

error:
	_leave(" = %d", ret);
	return ret;
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* Copyright (c) 2002 Red Hat, Inc. All rights reserved.
	3	*
	4	* This software may be freely redistributed under the terms of the
	5	* GNU General Public License.
	6	*
	7	* You should have received a copy of the GNU General Public License
	8	* along with this program; if not, write to the Free Software
	9	* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
	10	*
44d1b980	11	* Authors: David Woodhouse <dwmw2@infradead.org>
1da177e4 LT	12	* David Howells <dhowells@redhat.com>
	13	*
	14	*/
	15
	16	#include <linux/kernel.h>
	17	#include <linux/module.h>
	18	#include <linux/init.h>
1da177e4 LT	19	#include <linux/fs.h>
1da177e4 LT	20	#include <linux/pagemap.h>
e8edc6e0	21	#include <linux/sched.h>
bec5eb61	22	#include <linux/mount.h>
bec5eb61	23	#include <linux/namei.h>
a01179e6	24	#include <linux/iversion.h>
1da177e4 LT	25	#include "internal.h"
1da177e4 LT	26
d3e3b7ea DH	27	static const struct inode_operations afs_symlink_inode_operations = {
	28	.get_link = page_get_link,
	29	.listxattr = afs_listxattr,
	30	};
	31
1da177e4 LT	32	/*
	33	* map the AFS file status to the inode member variables
	34	*/
00d3b7a4	35	static int afs_inode_map_status(struct afs_vnode vnode, struct key key)
1da177e4 LT	36	{
1da177e4 LT	37	struct inode *inode = AFS_VNODE_TO_I(vnode);
c435ee34	38	bool changed;
1da177e4	39
260a9803	40	_debug("FS: ft=%d lk=%d sz=%llu ver=%Lu mod=%hu",
1da177e4 LT	41	vnode->status.type,
1da177e4 LT	42	vnode->status.nlink,
ba3e0e1a	43	(unsigned long long) vnode->status.size,
08e0e7c8	44	vnode->status.data_version,
1da177e4 LT	45	vnode->status.mode);
1da177e4 LT	46
c435ee34 DH	47	read_seqlock_excl(&vnode->cb_lock);
c435ee34 DH	48
1da177e4 LT	49	switch (vnode->status.type) {
	50	case AFS_FTYPE_FILE:
	51	inode->i_mode = S_IFREG \| vnode->status.mode;
	52	inode->i_op = &afs_file_inode_operations;
00d3b7a4	53	inode->i_fop = &afs_file_operations;
1da177e4 LT	54	break;
	55	case AFS_FTYPE_DIR:
	56	inode->i_mode = S_IFDIR \| vnode->status.mode;
	57	inode->i_op = &afs_dir_inode_operations;
	58	inode->i_fop = &afs_dir_file_operations;
	59	break;
	60	case AFS_FTYPE_SYMLINK:
944c74f4 DH	61	/* Symlinks with a mode of 0644 are actually mountpoints. */
	62	if ((vnode->status.mode & 0777) == 0644) {
	63	inode->i_flags \|= S_AUTOMOUNT;
	64
944c74f4	65	set_bit(AFS_VNODE_MOUNTPOINT, &vnode->flags);
944c74f4 DH	66
	67	inode->i_mode = S_IFDIR \| 0555;
	68	inode->i_op = &afs_mntpt_inode_operations;
	69	inode->i_fop = &afs_mntpt_file_operations;
	70	} else {
	71	inode->i_mode = S_IFLNK \| vnode->status.mode;
d3e3b7ea	72	inode->i_op = &afs_symlink_inode_operations;
944c74f4	73	}
21fc61c7	74	inode_nohighmem(inode);
1da177e4 LT	75	break;
	76	default:
	77	printk("kAFS: AFS vnode with undefined type\n");
c435ee34	78	read_sequnlock_excl(&vnode->cb_lock);
1da177e4 LT	79	return -EBADMSG;
	80	}
	81
c435ee34	82	changed = (vnode->status.size != inode->i_size);
9b3f26c9	83
bfe86848	84	set_nlink(inode, vnode->status.nlink);
1da177e4	85	inode->i_uid = vnode->status.owner;
6186f078	86	inode->i_gid = vnode->status.group;
1da177e4	87	inode->i_size = vnode->status.size;
ab94f5d0	88	inode->i_ctime.tv_sec = vnode->status.mtime_client;
1da177e4 LT	89	inode->i_ctime.tv_nsec = 0;
1da177e4 LT	90	inode->i_atime = inode->i_mtime = inode->i_ctime;
1da177e4	91	inode->i_blocks = 0;
d6e43f75	92	inode->i_generation = vnode->fid.unique;
a01179e6	93	inode_set_iversion_raw(inode, vnode->status.data_version);
1da177e4	94	inode->i_mapping->a_ops = &afs_fs_aops;
c435ee34 DH	95
	96	read_sequnlock_excl(&vnode->cb_lock);
	97
	98	#ifdef CONFIG_AFS_FSCACHE
	99	if (changed)
	100	fscache_attr_changed(vnode->cache);
	101	#endif
1da177e4	102	return 0;
ec26815a	103	}
1da177e4	104
d2ddc776 DH	105	/*
	106	* Fetch file status from the volume.
	107	*/
	108	int afs_fetch_status(struct afs_vnode vnode, struct key key)
	109	{
	110	struct afs_fs_cursor fc;
	111	int ret;
	112
	113	_enter("%s,{%x:%u.%u,S=%lx}",
	114	vnode->volume->name,
	115	vnode->fid.vid, vnode->fid.vnode, vnode->fid.unique,
	116	vnode->flags);
	117
	118	ret = -ERESTARTSYS;
	119	if (afs_begin_vnode_operation(&fc, vnode, key)) {
	120	while (afs_select_fileserver(&fc)) {
	121	fc.cb_break = vnode->cb_break + vnode->cb_s_break;
	122	afs_fs_fetch_file_status(&fc, NULL);
	123	}
	124
	125	afs_check_for_remote_deletion(&fc, fc.vnode);
	126	afs_vnode_commit_status(&fc, vnode, fc.cb_break);
	127	ret = afs_end_vnode_operation(&fc);
	128	}
	129
	130	_leave(" = %d", ret);
	131	return ret;
	132	}
	133
1da177e4 LT	134	/*
	135	* iget5() comparator
	136	*/
c435ee34	137	int afs_iget5_test(struct inode inode, void opaque)
1da177e4 LT	138	{
	139	struct afs_iget_data *data = opaque;
	140
	141	return inode->i_ino == data->fid.vnode &&
d6e43f75	142	inode->i_generation == data->fid.unique;
ec26815a	143	}
1da177e4	144
bec5eb61	145	/*
	146	* iget5() comparator for inode created by autocell operations
	147	*
	148	* These pseudo inodes don't match anything.
	149	*/
4d673da1	150	static int afs_iget5_pseudo_dir_test(struct inode inode, void opaque)
bec5eb61	151	{
	152	return 0;
	153	}
	154
1da177e4 LT	155	/*
	156	* iget5() inode initialiser
	157	*/
	158	static int afs_iget5_set(struct inode inode, void opaque)
	159	{
	160	struct afs_iget_data *data = opaque;
	161	struct afs_vnode *vnode = AFS_FS_I(inode);
	162
	163	inode->i_ino = data->fid.vnode;
d6e43f75	164	inode->i_generation = data->fid.unique;
1da177e4 LT	165	vnode->fid = data->fid;
	166	vnode->volume = data->volume;
	167
	168	return 0;
ec26815a	169	}
1da177e4	170
bec5eb61	171	/*
4d673da1 DH	172	* Create an inode for a dynamic root directory or an autocell dynamic
4d673da1 DH	173	* automount dir.
bec5eb61	174	*/
4d673da1	175	struct inode afs_iget_pseudo_dir(struct super_block sb, bool root)
bec5eb61	176	{
	177	struct afs_iget_data data;
	178	struct afs_super_info *as;
	179	struct afs_vnode *vnode;
bec5eb61	180	struct inode *inode;
	181	static atomic_t afs_autocell_ino;
	182
4d673da1	183	_enter("");
bec5eb61	184
bec5eb61	185	as = sb->s_fs_info;
4d673da1 DH	186	if (as->volume) {
	187	data.volume = as->volume;
	188	data.fid.vid = as->volume->vid;
	189	}
	190	if (root) {
	191	data.fid.vnode = 1;
	192	data.fid.unique = 1;
	193	} else {
	194	data.fid.vnode = atomic_inc_return(&afs_autocell_ino);
	195	data.fid.unique = 0;
	196	}
bec5eb61	197
4d673da1 DH	198	inode = iget5_locked(sb, data.fid.vnode,
4d673da1 DH	199	afs_iget5_pseudo_dir_test, afs_iget5_set,
bec5eb61	200	&data);
	201	if (!inode) {
	202	_leave(" = -ENOMEM");
	203	return ERR_PTR(-ENOMEM);
	204	}
	205
	206	_debug("GOT INODE %p { ino=%lu, vl=%x, vn=%x, u=%x }",
	207	inode, inode->i_ino, data.fid.vid, data.fid.vnode,
	208	data.fid.unique);
	209
	210	vnode = AFS_FS_I(inode);
	211
	212	/* there shouldn't be an existing inode */
	213	BUG_ON(!(inode->i_state & I_NEW));
	214
	215	inode->i_size = 0;
	216	inode->i_mode = S_IFDIR \| S_IRUGO \| S_IXUGO;
4d673da1 DH	217	if (root) {
	218	inode->i_op = &afs_dynroot_inode_operations;
	219	inode->i_fop = &afs_dynroot_file_operations;
	220	} else {
	221	inode->i_op = &afs_autocell_inode_operations;
	222	}
bfe86848	223	set_nlink(inode, 2);
a0a5386a EB	224	inode->i_uid = GLOBAL_ROOT_UID;
a0a5386a EB	225	inode->i_gid = GLOBAL_ROOT_GID;
bec5eb61	226	inode->i_ctime.tv_sec = get_seconds();
	227	inode->i_ctime.tv_nsec = 0;
	228	inode->i_atime = inode->i_mtime = inode->i_ctime;
	229	inode->i_blocks = 0;
a01179e6	230	inode_set_iversion_raw(inode, 0);
bec5eb61	231	inode->i_generation = 0;
	232
	233	set_bit(AFS_VNODE_PSEUDODIR, &vnode->flags);
4d673da1 DH	234	if (!root) {
	235	set_bit(AFS_VNODE_MOUNTPOINT, &vnode->flags);
	236	inode->i_flags \|= S_AUTOMOUNT;
	237	}
	238
	239	inode->i_flags \|= S_NOATIME;
bec5eb61	240	unlock_new_inode(inode);
	241	_leave(" = %p", inode);
	242	return inode;
	243	}
	244
1da177e4 LT	245	/*
	246	* inode retrieval
	247	*/
260a9803 DH	248	struct inode afs_iget(struct super_block sb, struct key *key,
260a9803 DH	249	struct afs_fid fid, struct afs_file_status status,
d2ddc776	250	struct afs_callback cb, struct afs_cb_interest cbi)
1da177e4 LT	251	{
	252	struct afs_iget_data data = { .fid = *fid };
	253	struct afs_super_info *as;
	254	struct afs_vnode *vnode;
	255	struct inode *inode;
	256	int ret;
	257
416351f2	258	_enter(",{%x:%u.%u},,", fid->vid, fid->vnode, fid->unique);
1da177e4 LT	259
	260	as = sb->s_fs_info;
	261	data.volume = as->volume;
	262
	263	inode = iget5_locked(sb, fid->vnode, afs_iget5_test, afs_iget5_set,
	264	&data);
	265	if (!inode) {
	266	_leave(" = -ENOMEM");
08e0e7c8	267	return ERR_PTR(-ENOMEM);
1da177e4 LT	268	}
1da177e4 LT	269
08e0e7c8 DH	270	_debug("GOT INODE %p { vl=%x vn=%x, u=%x }",
	271	inode, fid->vid, fid->vnode, fid->unique);
	272
1da177e4 LT	273	vnode = AFS_FS_I(inode);
	274
	275	/* deal with an existing inode */
	276	if (!(inode->i_state & I_NEW)) {
08e0e7c8 DH	277	_leave(" = %p", inode);
08e0e7c8 DH	278	return inode;
1da177e4 LT	279	}
1da177e4 LT	280
260a9803 DH	281	if (!status) {
260a9803 DH	282	/* it's a remotely extant inode */
d2ddc776	283	ret = afs_fetch_status(vnode, key);
260a9803 DH	284	if (ret < 0)
	285	goto bad_inode;
	286	} else {
	287	/* it's an inode we just created */
	288	memcpy(&vnode->status, status, sizeof(vnode->status));
	289
	290	if (!cb) {
	291	/* it's a symlink we just created (the fileserver
	292	* didn't give us a callback) */
	293	vnode->cb_version = 0;
260a9803	294	vnode->cb_type = 0;
c435ee34	295	vnode->cb_expires_at = 0;
260a9803 DH	296	} else {
260a9803 DH	297	vnode->cb_version = cb->version;
260a9803	298	vnode->cb_type = cb->type;
c435ee34	299	vnode->cb_expires_at = cb->expiry;
d2ddc776	300	vnode->cb_interest = afs_get_cb_interest(cbi);
c435ee34	301	set_bit(AFS_VNODE_CB_PROMISED, &vnode->flags);
260a9803	302	}
c435ee34 DH	303
c435ee34 DH	304	vnode->cb_expires_at += ktime_get_real_seconds();
260a9803 DH	305	}
260a9803 DH	306
9b3f26c9 DH	307	/* set up caching before mapping the status, as map-status reads the
	308	* first page of symlinks to see if they're really mountpoints */
	309	inode->i_size = vnode->status.size;
	310	#ifdef CONFIG_AFS_FSCACHE
	311	vnode->cache = fscache_acquire_cookie(vnode->volume->cache,
	312	&afs_vnode_cache_index_def,
94d30ae9	313	vnode, true);
9b3f26c9 DH	314	#endif
9b3f26c9 DH	315
00d3b7a4	316	ret = afs_inode_map_status(vnode, key);
08e0e7c8	317	if (ret < 0)
1da177e4 LT	318	goto bad_inode;
	319
	320	/* success */
260a9803	321	clear_bit(AFS_VNODE_UNSET, &vnode->flags);
08e0e7c8	322	inode->i_flags \|= S_NOATIME;
1da177e4	323	unlock_new_inode(inode);
08e0e7c8 DH	324	_leave(" = %p [CB { v=%u t=%u }]", inode, vnode->cb_version, vnode->cb_type);
08e0e7c8 DH	325	return inode;
1da177e4 LT	326
1da177e4 LT	327	/* failure */
ec26815a	328	bad_inode:
9b3f26c9 DH	329	#ifdef CONFIG_AFS_FSCACHE
	330	fscache_relinquish_cookie(vnode->cache, 0);
	331	vnode->cache = NULL;
	332	#endif
aa7fa240	333	iget_failed(inode);
1da177e4	334	_leave(" = %d [bad]", ret);
08e0e7c8	335	return ERR_PTR(ret);
ec26815a	336	}
1da177e4	337
416351f2 DH	338	/*
	339	* mark the data attached to an inode as obsolete due to a write on the server
	340	* - might also want to ditch all the outstanding writes and dirty pages
	341	*/
	342	void afs_zap_data(struct afs_vnode *vnode)
	343	{
0f300ca9	344	_enter("{%x:%u}", vnode->fid.vid, vnode->fid.vnode);
416351f2 DH	345
416351f2 DH	346	/* nuke all the non-dirty pages that aren't locked, mapped or being
0f300ca9 DH	347	* written back in a regular file and completely discard the pages in a
	348	* directory or symlink */
	349	if (S_ISREG(vnode->vfs_inode.i_mode))
	350	invalidate_remote_inode(&vnode->vfs_inode);
	351	else
	352	invalidate_inode_pages2(vnode->vfs_inode.i_mapping);
416351f2 DH	353	}
416351f2 DH	354
260a9803 DH	355	/*
	356	* validate a vnode/inode
	357	* - there are several things we need to check
	358	* - parent dir data changes (rm, rmdir, rename, mkdir, create, link,
	359	* symlink)
	360	* - parent dir metadata changed (security changes)
	361	* - dentry data changed (write, truncate)
	362	* - dentry metadata changed (security changes)
	363	*/
	364	int afs_validate(struct afs_vnode vnode, struct key key)
	365	{
c435ee34 DH	366	time64_t now = ktime_get_real_seconds();
c435ee34 DH	367	bool valid = false;
260a9803 DH	368	int ret;
	369
	370	_enter("{v={%x:%u} fl=%lx},%x",
	371	vnode->fid.vid, vnode->fid.vnode, vnode->flags,
	372	key_serial(key));
	373
c435ee34 DH	374	/* Quickly check the callback state. Ideally, we'd use read_seqbegin
	375	* here, but we have no way to pass the net namespace to the RCU
	376	* cleanup for the server record.
	377	*/
	378	read_seqlock_excl(&vnode->cb_lock);
	379
	380	if (test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
	381	if (vnode->cb_s_break != vnode->cb_interest->server->cb_s_break) {
	382	vnode->cb_s_break = vnode->cb_interest->server->cb_s_break;
	383	} else if (!test_bit(AFS_VNODE_DIR_MODIFIED, &vnode->flags) &&
	384	!test_bit(AFS_VNODE_ZAP_DATA, &vnode->flags) &&
	385	vnode->cb_expires_at - 10 > now) {
	386	valid = true;
260a9803	387	}
c435ee34 DH	388	} else if (test_bit(AFS_VNODE_DELETED, &vnode->flags)) {
c435ee34 DH	389	valid = true;
260a9803 DH	390	}
260a9803 DH	391
c435ee34	392	read_sequnlock_excl(&vnode->cb_lock);
440fbc3a DH	393
	394	if (test_bit(AFS_VNODE_DELETED, &vnode->flags))
	395	clear_nlink(&vnode->vfs_inode);
	396
c435ee34	397	if (valid)
260a9803 DH	398	goto valid;
	399
	400	mutex_lock(&vnode->validate_lock);
	401
	402	/* if the promise has expired, we need to check the server again to get
	403	* a new promise - note that if the (parent) directory's metadata was
	404	* changed then the security may be different and we may no longer have
	405	* access */
c435ee34	406	if (!test_bit(AFS_VNODE_CB_PROMISED, &vnode->flags)) {
260a9803	407	_debug("not promised");
d2ddc776	408	ret = afs_fetch_status(vnode, key);
c435ee34 DH	409	if (ret < 0) {
	410	if (ret == -ENOENT) {
	411	set_bit(AFS_VNODE_DELETED, &vnode->flags);
	412	ret = -ESTALE;
	413	}
260a9803	414	goto error_unlock;
c435ee34	415	}
260a9803 DH	416	_debug("new promise [fl=%lx]", vnode->flags);
	417	}
	418
	419	if (test_bit(AFS_VNODE_DELETED, &vnode->flags)) {
	420	_debug("file already deleted");
	421	ret = -ESTALE;
	422	goto error_unlock;
	423	}
	424
	425	/* if the vnode's data version number changed then its contents are
	426	* different */
416351f2 DH	427	if (test_and_clear_bit(AFS_VNODE_ZAP_DATA, &vnode->flags))
416351f2 DH	428	afs_zap_data(vnode);
260a9803	429
c435ee34	430	clear_bit(AFS_VNODE_DIR_MODIFIED, &vnode->flags);
260a9803 DH	431	mutex_unlock(&vnode->validate_lock);
	432	valid:
	433	_leave(" = 0");
	434	return 0;
	435
	436	error_unlock:
	437	mutex_unlock(&vnode->validate_lock);
	438	_leave(" = %d", ret);
	439	return ret;
	440	}
	441
1da177e4 LT	442	/*
	443	* read the attributes of an inode
	444	*/
a528d35e DH	445	int afs_getattr(const struct path path, struct kstat stat,
a528d35e DH	446	u32 request_mask, unsigned int query_flags)
1da177e4	447	{
a528d35e	448	struct inode *inode = d_inode(path->dentry);
c435ee34 DH	449	struct afs_vnode *vnode = AFS_FS_I(inode);
c435ee34 DH	450	int seq = 0;
1da177e4	451
d6e43f75	452	_enter("{ ino=%lu v=%u }", inode->i_ino, inode->i_generation);
1da177e4	453
c435ee34 DH	454	do {
	455	read_seqbegin_or_lock(&vnode->cb_lock, &seq);
	456	generic_fillattr(inode, stat);
	457	} while (need_seqretry(&vnode->cb_lock, seq));
	458
	459	done_seqretry(&vnode->cb_lock, seq);
1da177e4	460	return 0;
ec26815a	461	}
1da177e4	462
bec5eb61	463	/*
	464	* discard an AFS inode
	465	*/
	466	int afs_drop_inode(struct inode *inode)
	467	{
	468	_enter("");
	469
	470	if (test_bit(AFS_VNODE_PSEUDODIR, &AFS_FS_I(inode)->flags))
	471	return generic_delete_inode(inode);
	472	else
	473	return generic_drop_inode(inode);
	474	}
	475
1da177e4 LT	476	/*
	477	* clear an AFS inode
	478	*/
b57922d9	479	void afs_evict_inode(struct inode *inode)
1da177e4 LT	480	{
	481	struct afs_vnode *vnode;
	482
	483	vnode = AFS_FS_I(inode);
	484
c435ee34	485	_enter("{%x:%u.%d}",
260a9803	486	vnode->fid.vid,
1da177e4	487	vnode->fid.vnode,
c435ee34	488	vnode->fid.unique);
1da177e4	489
08e0e7c8 DH	490	_debug("CLEAR INODE %p", inode);
	491
	492	ASSERTCMP(inode->i_ino, ==, vnode->fid.vnode);
	493
91b0abe3	494	truncate_inode_pages_final(&inode->i_data);
dbd5768f	495	clear_inode(inode);
b57922d9	496
c435ee34 DH	497	if (vnode->cb_interest) {
	498	afs_put_cb_interest(afs_i2net(inode), vnode->cb_interest);
	499	vnode->cb_interest = NULL;
08e0e7c8	500	}
1da177e4	501
4343d008 DH	502	while (!list_empty(&vnode->wb_keys)) {
	503	struct afs_wb_key *wbk = list_entry(vnode->wb_keys.next,
	504	struct afs_wb_key, vnode_link);
	505	list_del(&wbk->vnode_link);
	506	afs_put_wb_key(wbk);
	507	}
1da177e4	508
9b3f26c9 DH	509	#ifdef CONFIG_AFS_FSCACHE
9b3f26c9 DH	510	fscache_relinquish_cookie(vnode->cache, 0);
1da177e4 LT	511	vnode->cache = NULL;
	512	#endif
	513
be080a6f	514	afs_put_permits(vnode->permit_cache);
1da177e4	515	_leave("");
ec26815a	516	}
31143d5d DH	517
	518	/*
	519	* set the attributes of an inode
	520	*/
	521	int afs_setattr(struct dentry dentry, struct iattr attr)
	522	{
d2ddc776	523	struct afs_fs_cursor fc;
2b0143b5	524	struct afs_vnode *vnode = AFS_FS_I(d_inode(dentry));
31143d5d DH	525	struct key *key;
	526	int ret;
	527
a455589f AV	528	_enter("{%x:%u},{n=%pd},%x",
a455589f AV	529	vnode->fid.vid, vnode->fid.vnode, dentry,
31143d5d DH	530	attr->ia_valid);
	531
	532	if (!(attr->ia_valid & (ATTR_SIZE \| ATTR_MODE \| ATTR_UID \| ATTR_GID \|
	533	ATTR_MTIME))) {
	534	_leave(" = 0 [unsupported]");
	535	return 0;
	536	}
	537
	538	/* flush any dirty data outstanding on a regular file */
4343d008	539	if (S_ISREG(vnode->vfs_inode.i_mode))
31143d5d	540	filemap_write_and_wait(vnode->vfs_inode.i_mapping);
31143d5d DH	541
31143d5d DH	542	if (attr->ia_valid & ATTR_FILE) {
215804a9	543	key = afs_file_key(attr->ia_file);
31143d5d DH	544	} else {
	545	key = afs_request_key(vnode->volume->cell);
	546	if (IS_ERR(key)) {
	547	ret = PTR_ERR(key);
	548	goto error;
	549	}
	550	}
	551
d2ddc776 DH	552	ret = -ERESTARTSYS;
	553	if (afs_begin_vnode_operation(&fc, vnode, key)) {
	554	while (afs_select_fileserver(&fc)) {
	555	fc.cb_break = vnode->cb_break + vnode->cb_s_break;
	556	afs_fs_setattr(&fc, attr);
	557	}
	558
	559	afs_check_for_remote_deletion(&fc, fc.vnode);
	560	afs_vnode_commit_status(&fc, vnode, fc.cb_break);
	561	ret = afs_end_vnode_operation(&fc);
	562	}
	563
31143d5d DH	564	if (!(attr->ia_valid & ATTR_FILE))
	565	key_put(key);
	566
	567	error:
	568	_leave(" = %d", ret);
	569	return ret;
	570	}