nfs4proc.c

/*
 *  fs/nfs/nfs4proc.c
 *
 *  Client-side procedure declarations for NFSv4.
 *
 *  Copyright (c) 2002 The Regents of the University of Michigan.
 *  All rights reserved.
 *
 *  Kendrick Smith <kmsmith@umich.edu>
 *  Andy Adamson   <andros@umich.edu>
 *
 *  Redistribution and use in source and binary forms, with or without
 *  modification, are permitted provided that the following conditions
 *  are met:
 *
 *  1. Redistributions of source code must retain the above copyright
 *     notice, this list of conditions and the following disclaimer.
 *  2. Redistributions in binary form must reproduce the above copyright
 *     notice, this list of conditions and the following disclaimer in the
 *     documentation and/or other materials provided with the distribution.
 *  3. Neither the name of the University nor the names of its
 *     contributors may be used to endorse or promote products derived
 *     from this software without specific prior written permission.
 *
 *  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
 *  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
 *  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
 *  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
 *  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
 *  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
 *  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
 *  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
 *  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
 *  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
 *  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
 */

#include <linux/mm.h>
#include <linux/utsname.h>
#include <linux/delay.h>
#include <linux/errno.h>
#include <linux/string.h>
#include <linux/sunrpc/clnt.h>
#include <linux/nfs.h>
#include <linux/nfs4.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_page.h>
#include <linux/smp_lock.h>
#include <linux/namei.h>
#include <linux/mount.h>

#include "nfs4_fs.h"
#include "delegation.h"
#include "iostat.h"

#define NFSDBG_FACILITY		NFSDBG_PROC

#define NFS4_POLL_RETRY_MIN	(HZ/10)
#define NFS4_POLL_RETRY_MAX	(15*HZ)

struct nfs4_opendata;
static int _nfs4_proc_open(struct nfs4_opendata *data);
static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
static int nfs4_async_handle_error(struct rpc_task *, const struct nfs_server *);
static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry);
static int nfs4_handle_exception(const struct nfs_server *server, int errorcode, struct nfs4_exception *exception);
static int nfs4_wait_clnt_recover(struct rpc_clnt *clnt, struct nfs_client *clp);

/* Prevent leaks of NFSv4 errors into userland */
int nfs4_map_errors(int err)
{
	if (err < -1000) {
		dprintk("%s could not handle NFSv4 error %d\n",
				__FUNCTION__, -err);
		return -EIO;
	}
	return err;
}

/*
 * This is our standard bitmap for GETATTR requests.
 */
const u32 nfs4_fattr_bitmap[2] = {
	FATTR4_WORD0_TYPE
	| FATTR4_WORD0_CHANGE
	| FATTR4_WORD0_SIZE
	| FATTR4_WORD0_FSID
	| FATTR4_WORD0_FILEID,
	FATTR4_WORD1_MODE
	| FATTR4_WORD1_NUMLINKS
	| FATTR4_WORD1_OWNER
	| FATTR4_WORD1_OWNER_GROUP
	| FATTR4_WORD1_RAWDEV
	| FATTR4_WORD1_SPACE_USED
	| FATTR4_WORD1_TIME_ACCESS
	| FATTR4_WORD1_TIME_METADATA
	| FATTR4_WORD1_TIME_MODIFY
};

const u32 nfs4_statfs_bitmap[2] = {
	FATTR4_WORD0_FILES_AVAIL
	| FATTR4_WORD0_FILES_FREE
	| FATTR4_WORD0_FILES_TOTAL,
	FATTR4_WORD1_SPACE_AVAIL
	| FATTR4_WORD1_SPACE_FREE
	| FATTR4_WORD1_SPACE_TOTAL
};

const u32 nfs4_pathconf_bitmap[2] = {
	FATTR4_WORD0_MAXLINK
	| FATTR4_WORD0_MAXNAME,
	0
};

const u32 nfs4_fsinfo_bitmap[2] = { FATTR4_WORD0_MAXFILESIZE
			| FATTR4_WORD0_MAXREAD
			| FATTR4_WORD0_MAXWRITE
			| FATTR4_WORD0_LEASE_TIME,
			0
};

const u32 nfs4_fs_locations_bitmap[2] = {
	FATTR4_WORD0_TYPE
	| FATTR4_WORD0_CHANGE
	| FATTR4_WORD0_SIZE
	| FATTR4_WORD0_FSID
	| FATTR4_WORD0_FILEID
	| FATTR4_WORD0_FS_LOCATIONS,
	FATTR4_WORD1_MODE
	| FATTR4_WORD1_NUMLINKS
	| FATTR4_WORD1_OWNER
	| FATTR4_WORD1_OWNER_GROUP
	| FATTR4_WORD1_RAWDEV
	| FATTR4_WORD1_SPACE_USED
	| FATTR4_WORD1_TIME_ACCESS
	| FATTR4_WORD1_TIME_METADATA
	| FATTR4_WORD1_TIME_MODIFY
	| FATTR4_WORD1_MOUNTED_ON_FILEID
};

static void nfs4_setup_readdir(u64 cookie, __be32 *verifier, struct dentry *dentry,
		struct nfs4_readdir_arg *readdir)
{
	__be32 *start, *p;

	BUG_ON(readdir->count < 80);
	if (cookie > 2) {
		readdir->cookie = cookie;
		memcpy(&readdir->verifier, verifier, sizeof(readdir->verifier));
		return;
	}

	readdir->cookie = 0;
	memset(&readdir->verifier, 0, sizeof(readdir->verifier));
	if (cookie == 2)
		return;
	
	/*
	 * NFSv4 servers do not return entries for '.' and '..'
	 * Therefore, we fake these entries here.  We let '.'
	 * have cookie 0 and '..' have cookie 1.  Note that
	 * when talking to the server, we always send cookie 0
	 * instead of 1 or 2.
	 */
	start = p = kmap_atomic(*readdir->pages, KM_USER0);
	
	if (cookie == 0) {
		*p++ = xdr_one;                                  /* next */
		*p++ = xdr_zero;                   /* cookie, first word */
		*p++ = xdr_one;                   /* cookie, second word */
		*p++ = xdr_one;                             /* entry len */
		memcpy(p, ".\0\0\0", 4);                        /* entry */
		p++;
		*p++ = xdr_one;                         /* bitmap length */
		*p++ = htonl(FATTR4_WORD0_FILEID);             /* bitmap */
		*p++ = htonl(8);              /* attribute buffer length */
		p = xdr_encode_hyper(p, dentry->d_inode->i_ino);
	}
	
	*p++ = xdr_one;                                  /* next */
	*p++ = xdr_zero;                   /* cookie, first word */
	*p++ = xdr_two;                   /* cookie, second word */
	*p++ = xdr_two;                             /* entry len */
	memcpy(p, "..\0\0", 4);                         /* entry */
	p++;
	*p++ = xdr_one;                         /* bitmap length */
	*p++ = htonl(FATTR4_WORD0_FILEID);             /* bitmap */
	*p++ = htonl(8);              /* attribute buffer length */
	p = xdr_encode_hyper(p, dentry->d_parent->d_inode->i_ino);

	readdir->pgbase = (char *)p - (char *)start;
	readdir->count -= readdir->pgbase;
	kunmap_atomic(start, KM_USER0);
}

static void renew_lease(const struct nfs_server *server, unsigned long timestamp)
{
	struct nfs_client *clp = server->nfs_client;
	spin_lock(&clp->cl_lock);
	if (time_before(clp->cl_last_renewal,timestamp))
		clp->cl_last_renewal = timestamp;
	spin_unlock(&clp->cl_lock);
}

static void update_changeattr(struct inode *dir, struct nfs4_change_info *cinfo)
{
	struct nfs_inode *nfsi = NFS_I(dir);

	spin_lock(&dir->i_lock);
	nfsi->cache_validity |= NFS_INO_INVALID_ATTR|NFS_INO_REVAL_PAGECACHE|NFS_INO_INVALID_DATA;
	if (cinfo->before == nfsi->change_attr && cinfo->atomic)
		nfsi->change_attr = cinfo->after;
	spin_unlock(&dir->i_lock);
}

struct nfs4_opendata {
	struct kref kref;
	struct nfs_openargs o_arg;
	struct nfs_openres o_res;
	struct nfs_open_confirmargs c_arg;
	struct nfs_open_confirmres c_res;
	struct nfs_fattr f_attr;
	struct nfs_fattr dir_attr;
	struct path path;
	struct dentry *dir;
	struct nfs4_state_owner *owner;
	struct iattr attrs;
	unsigned long timestamp;
	int rpc_status;
	int cancelled;
};


static void nfs4_init_opendata_res(struct nfs4_opendata *p)
{
	p->o_res.f_attr = &p->f_attr;
	p->o_res.dir_attr = &p->dir_attr;
	p->o_res.server = p->o_arg.server;
	nfs_fattr_init(&p->f_attr);
	nfs_fattr_init(&p->dir_attr);
}

static struct nfs4_opendata *nfs4_opendata_alloc(struct path *path,
		struct nfs4_state_owner *sp, int flags,
		const struct iattr *attrs)
{
	struct dentry *parent = dget_parent(path->dentry);
	struct inode *dir = parent->d_inode;
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs4_opendata *p;

	p = kzalloc(sizeof(*p), GFP_KERNEL);
	if (p == NULL)
		goto err;
	p->o_arg.seqid = nfs_alloc_seqid(&sp->so_seqid);
	if (p->o_arg.seqid == NULL)
		goto err_free;
	p->path.mnt = mntget(path->mnt);
	p->path.dentry = dget(path->dentry);
	p->dir = parent;
	p->owner = sp;
	atomic_inc(&sp->so_count);
	p->o_arg.fh = NFS_FH(dir);
	p->o_arg.open_flags = flags,
	p->o_arg.clientid = server->nfs_client->cl_clientid;
	p->o_arg.id = sp->so_owner_id.id;
	p->o_arg.name = &p->path.dentry->d_name;
	p->o_arg.server = server;
	p->o_arg.bitmask = server->attr_bitmask;
	p->o_arg.claim = NFS4_OPEN_CLAIM_NULL;
	if (flags & O_EXCL) {
		u32 *s = (u32 *) p->o_arg.u.verifier.data;
		s[0] = jiffies;
		s[1] = current->pid;
	} else if (flags & O_CREAT) {
		p->o_arg.u.attrs = &p->attrs;
		memcpy(&p->attrs, attrs, sizeof(p->attrs));
	}
	p->c_arg.fh = &p->o_res.fh;
	p->c_arg.stateid = &p->o_res.stateid;
	p->c_arg.seqid = p->o_arg.seqid;
	nfs4_init_opendata_res(p);
	kref_init(&p->kref);
	return p;
err_free:
	kfree(p);
err:
	dput(parent);
	return NULL;
}

static void nfs4_opendata_free(struct kref *kref)
{
	struct nfs4_opendata *p = container_of(kref,
			struct nfs4_opendata, kref);

	nfs_free_seqid(p->o_arg.seqid);
	nfs4_put_state_owner(p->owner);
	dput(p->dir);
	dput(p->path.dentry);
	mntput(p->path.mnt);
	kfree(p);
}

static void nfs4_opendata_put(struct nfs4_opendata *p)
{
	if (p != NULL)
		kref_put(&p->kref, nfs4_opendata_free);
}

static int nfs4_wait_for_completion_rpc_task(struct rpc_task *task)
{
	sigset_t oldset;
	int ret;

	rpc_clnt_sigmask(task->tk_client, &oldset);
	ret = rpc_wait_for_completion_task(task);
	rpc_clnt_sigunmask(task->tk_client, &oldset);
	return ret;
}

static void update_open_stateflags(struct nfs4_state *state, mode_t open_flags)
{
	switch (open_flags) {
		case FMODE_WRITE:
			state->n_wronly++;
			break;
		case FMODE_READ:
			state->n_rdonly++;
			break;
		case FMODE_READ|FMODE_WRITE:
			state->n_rdwr++;
	}
	nfs4_state_set_mode_locked(state, state->state | open_flags);
}

static void nfs_set_open_stateid_locked(struct nfs4_state *state, nfs4_stateid *stateid, int open_flags)
{
	if (test_bit(NFS_DELEGATED_STATE, &state->flags) == 0)
		memcpy(state->stateid.data, stateid->data, sizeof(state->stateid.data));
	memcpy(state->open_stateid.data, stateid->data, sizeof(state->open_stateid.data));
	switch (open_flags) {
		case FMODE_READ:
			set_bit(NFS_O_RDONLY_STATE, &state->flags);
			break;
		case FMODE_WRITE:
			set_bit(NFS_O_WRONLY_STATE, &state->flags);
			break;
		case FMODE_READ|FMODE_WRITE:
			set_bit(NFS_O_RDWR_STATE, &state->flags);
	}
}

static void nfs_set_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid, int open_flags)
{
	spin_lock(&state->owner->so_lock);
	spin_lock(&state->inode->i_lock);
	nfs_set_open_stateid_locked(state, stateid, open_flags);
	spin_unlock(&state->inode->i_lock);
	spin_unlock(&state->owner->so_lock);
}

static void update_open_stateid(struct nfs4_state *state, nfs4_stateid *open_stateid, nfs4_stateid *deleg_stateid, int open_flags)
{
	struct inode *inode = state->inode;

	open_flags &= (FMODE_READ|FMODE_WRITE);
	/* Protect against nfs4_find_state_byowner() */
	spin_lock(&state->owner->so_lock);
	spin_lock(&inode->i_lock);
	if (deleg_stateid != NULL) {
		memcpy(state->stateid.data, deleg_stateid->data, sizeof(state->stateid.data));
		set_bit(NFS_DELEGATED_STATE, &state->flags);
	}
	if (open_stateid != NULL)
		nfs_set_open_stateid_locked(state, open_stateid, open_flags);
	update_open_stateflags(state, open_flags);
	spin_unlock(&inode->i_lock);
	spin_unlock(&state->owner->so_lock);
}

static struct nfs4_state *nfs4_opendata_to_nfs4_state(struct nfs4_opendata *data)
{
	struct inode *inode;
	struct nfs4_state *state = NULL;
	struct nfs_delegation *delegation;
	nfs4_stateid *deleg_stateid = NULL;
	int ret;

	ret = -EAGAIN;
	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
		goto err;
	inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh, &data->f_attr);
	ret = PTR_ERR(inode);
	if (IS_ERR(inode))
		goto err;
	ret = -ENOMEM;
	state = nfs4_get_open_state(inode, data->owner);
	if (state == NULL)
		goto err_put_inode;
	if (data->o_res.delegation_type != 0) {
		int delegation_flags = 0;

		rcu_read_lock();
		delegation = rcu_dereference(NFS_I(inode)->delegation);
		if (delegation)
			delegation_flags = delegation->flags;
		rcu_read_unlock();
		if (!(delegation_flags & NFS_DELEGATION_NEED_RECLAIM))
			nfs_inode_set_delegation(state->inode,
					data->owner->so_cred,
					&data->o_res);
		else
			nfs_inode_reclaim_delegation(state->inode,
					data->owner->so_cred,
					&data->o_res);
	}
	rcu_read_lock();
	delegation = rcu_dereference(NFS_I(inode)->delegation);
	if (delegation != NULL)
		deleg_stateid = &delegation->stateid;
	update_open_stateid(state, &data->o_res.stateid, deleg_stateid, data->o_arg.open_flags);
	rcu_read_unlock();
	iput(inode);
	return state;
err_put_inode:
	iput(inode);
err:
	return ERR_PTR(ret);
}

static struct nfs_open_context *nfs4_state_find_open_context(struct nfs4_state *state)
{
	struct nfs_inode *nfsi = NFS_I(state->inode);
	struct nfs_open_context *ctx;

	spin_lock(&state->inode->i_lock);
	list_for_each_entry(ctx, &nfsi->open_files, list) {
		if (ctx->state != state)
			continue;
		get_nfs_open_context(ctx);
		spin_unlock(&state->inode->i_lock);
		return ctx;
	}
	spin_unlock(&state->inode->i_lock);
	return ERR_PTR(-ENOENT);
}

static int nfs4_open_recover_helper(struct nfs4_opendata *opendata, mode_t openflags, struct nfs4_state **res)
{
	struct nfs4_state *newstate;
	int ret;

	opendata->o_arg.open_flags = openflags;
	memset(&opendata->o_res, 0, sizeof(opendata->o_res));
	memset(&opendata->c_res, 0, sizeof(opendata->c_res));
	nfs4_init_opendata_res(opendata);
	ret = _nfs4_proc_open(opendata);
	if (ret != 0)
		return ret; 
	newstate = nfs4_opendata_to_nfs4_state(opendata);
	if (IS_ERR(newstate))
		return PTR_ERR(newstate);
	nfs4_close_state(&opendata->path, newstate, openflags);
	*res = newstate;
	return 0;
}

static int nfs4_open_recover(struct nfs4_opendata *opendata, struct nfs4_state *state)
{
	struct nfs4_state *newstate;
	int ret;

	/* memory barrier prior to reading state->n_* */
	clear_bit(NFS_DELEGATED_STATE, &state->flags);
	smp_rmb();
	if (state->n_rdwr != 0) {
		ret = nfs4_open_recover_helper(opendata, FMODE_READ|FMODE_WRITE, &newstate);
		if (ret != 0)
			return ret;
		if (newstate != state)
			return -ESTALE;
	}
	if (state->n_wronly != 0) {
		ret = nfs4_open_recover_helper(opendata, FMODE_WRITE, &newstate);
		if (ret != 0)
			return ret;
		if (newstate != state)
			return -ESTALE;
	}
	if (state->n_rdonly != 0) {
		ret = nfs4_open_recover_helper(opendata, FMODE_READ, &newstate);
		if (ret != 0)
			return ret;
		if (newstate != state)
			return -ESTALE;
	}
	return 0;
}

/*
 * OPEN_RECLAIM:
 * 	reclaim state on the server after a reboot.
 */
static int _nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
{
	struct nfs_delegation *delegation = NFS_I(state->inode)->delegation;
	struct nfs4_opendata *opendata;
	int delegation_type = 0;
	int status;

	if (delegation != NULL) {
		if (!(delegation->flags & NFS_DELEGATION_NEED_RECLAIM)) {
			memcpy(&state->stateid, &delegation->stateid,
					sizeof(state->stateid));
			set_bit(NFS_DELEGATED_STATE, &state->flags);
			return 0;
		}
		delegation_type = delegation->type;
	}
	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, 0, NULL);
	if (opendata == NULL)
		return -ENOMEM;
	opendata->o_arg.claim = NFS4_OPEN_CLAIM_PREVIOUS;
	opendata->o_arg.fh = NFS_FH(state->inode);
	nfs_copy_fh(&opendata->o_res.fh, opendata->o_arg.fh);
	opendata->o_arg.u.delegation_type = delegation_type;
	status = nfs4_open_recover(opendata, state);
	nfs4_opendata_put(opendata);
	return status;
}

static int nfs4_do_open_reclaim(struct nfs_open_context *ctx, struct nfs4_state *state)
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;
	do {
		err = _nfs4_do_open_reclaim(ctx, state);
		if (err != -NFS4ERR_DELAY)
			break;
		nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
}

static int nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	struct nfs_open_context *ctx;
	int ret;

	ctx = nfs4_state_find_open_context(state);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);
	ret = nfs4_do_open_reclaim(ctx, state);
	put_nfs_open_context(ctx);
	return ret;
}

static int _nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid)
{
	struct nfs4_state_owner  *sp  = state->owner;
	struct nfs4_opendata *opendata;
	int ret;

	opendata = nfs4_opendata_alloc(&ctx->path, sp, 0, NULL);
	if (opendata == NULL)
		return -ENOMEM;
	opendata->o_arg.claim = NFS4_OPEN_CLAIM_DELEGATE_CUR;
	memcpy(opendata->o_arg.u.delegation.data, stateid->data,
			sizeof(opendata->o_arg.u.delegation.data));
	ret = nfs4_open_recover(opendata, state);
	nfs4_opendata_put(opendata);
	return ret;
}

int nfs4_open_delegation_recall(struct nfs_open_context *ctx, struct nfs4_state *state, const nfs4_stateid *stateid)
{
	struct nfs4_exception exception = { };
	struct nfs_server *server = NFS_SERVER(state->inode);
	int err;
	do {
		err = _nfs4_open_delegation_recall(ctx, state, stateid);
		switch (err) {
			case 0:
				return err;
			case -NFS4ERR_STALE_CLIENTID:
			case -NFS4ERR_STALE_STATEID:
			case -NFS4ERR_EXPIRED:
				/* Don't recall a delegation if it was lost */
				nfs4_schedule_state_recovery(server->nfs_client);
				return err;
		}
		err = nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
}

static void nfs4_open_confirm_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;
	struct  rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_CONFIRM],
		.rpc_argp = &data->c_arg,
		.rpc_resp = &data->c_res,
		.rpc_cred = data->owner->so_cred,
	};
	data->timestamp = jiffies;
	rpc_call_setup(task, &msg, 0);
}

static void nfs4_open_confirm_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

	data->rpc_status = task->tk_status;
	if (RPC_ASSASSINATED(task))
		return;
	if (data->rpc_status == 0) {
		memcpy(data->o_res.stateid.data, data->c_res.stateid.data,
				sizeof(data->o_res.stateid.data));
		renew_lease(data->o_res.server, data->timestamp);
	}
	nfs_confirm_seqid(&data->owner->so_seqid, data->rpc_status);
	nfs_increment_open_seqid(data->rpc_status, data->c_arg.seqid);
}

static void nfs4_open_confirm_release(void *calldata)
{
	struct nfs4_opendata *data = calldata;
	struct nfs4_state *state = NULL;

	/* If this request hasn't been cancelled, do nothing */
	if (data->cancelled == 0)
		goto out_free;
	/* In case of error, no cleanup! */
	if (data->rpc_status != 0)
		goto out_free;
	nfs_confirm_seqid(&data->owner->so_seqid, 0);
	state = nfs4_opendata_to_nfs4_state(data);
	if (!IS_ERR(state))
		nfs4_close_state(&data->path, state, data->o_arg.open_flags);
out_free:
	nfs4_opendata_put(data);
}

static const struct rpc_call_ops nfs4_open_confirm_ops = {
	.rpc_call_prepare = nfs4_open_confirm_prepare,
	.rpc_call_done = nfs4_open_confirm_done,
	.rpc_release = nfs4_open_confirm_release,
};

/*
 * Note: On error, nfs4_proc_open_confirm will free the struct nfs4_opendata
 */
static int _nfs4_proc_open_confirm(struct nfs4_opendata *data)
{
	struct nfs_server *server = NFS_SERVER(data->dir->d_inode);
	struct rpc_task *task;
	int status;

	kref_get(&data->kref);
	/*
	 * If rpc_run_task() ends up calling ->rpc_release(), we
	 * want to ensure that it takes the 'error' code path.
	 */
	data->rpc_status = -ENOMEM;
	task = rpc_run_task(server->client, RPC_TASK_ASYNC, &nfs4_open_confirm_ops, data);
	if (IS_ERR(task))
		return PTR_ERR(task);
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status != 0) {
		data->cancelled = 1;
		smp_wmb();
	} else
		status = data->rpc_status;
	rpc_put_task(task);
	return status;
}

static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;
	struct nfs4_state_owner *sp = data->owner;
	struct rpc_message msg = {
		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN],
		.rpc_argp = &data->o_arg,
		.rpc_resp = &data->o_res,
		.rpc_cred = sp->so_cred,
	};
	
	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
		return;
	/* Update sequence id. */
	data->o_arg.id = sp->so_owner_id.id;
	data->o_arg.clientid = sp->so_client->cl_clientid;
	if (data->o_arg.claim == NFS4_OPEN_CLAIM_PREVIOUS)
		msg.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR];
	data->timestamp = jiffies;
	rpc_call_setup(task, &msg, 0);
}

static void nfs4_open_done(struct rpc_task *task, void *calldata)
{
	struct nfs4_opendata *data = calldata;

	data->rpc_status = task->tk_status;
	if (RPC_ASSASSINATED(task))
		return;
	if (task->tk_status == 0) {
		switch (data->o_res.f_attr->mode & S_IFMT) {
			case S_IFREG:
				break;
			case S_IFLNK:
				data->rpc_status = -ELOOP;
				break;
			case S_IFDIR:
				data->rpc_status = -EISDIR;
				break;
			default:
				data->rpc_status = -ENOTDIR;
		}
		renew_lease(data->o_res.server, data->timestamp);
		if (!(data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM))
			nfs_confirm_seqid(&data->owner->so_seqid, 0);
	}
	nfs_increment_open_seqid(data->rpc_status, data->o_arg.seqid);
}

static void nfs4_open_release(void *calldata)
{
	struct nfs4_opendata *data = calldata;
	struct nfs4_state *state = NULL;

	/* If this request hasn't been cancelled, do nothing */
	if (data->cancelled == 0)
		goto out_free;
	/* In case of error, no cleanup! */
	if (data->rpc_status != 0)
		goto out_free;
	/* In case we need an open_confirm, no cleanup! */
	if (data->o_res.rflags & NFS4_OPEN_RESULT_CONFIRM)
		goto out_free;
	nfs_confirm_seqid(&data->owner->so_seqid, 0);
	state = nfs4_opendata_to_nfs4_state(data);
	if (!IS_ERR(state))
		nfs4_close_state(&data->path, state, data->o_arg.open_flags);
out_free:
	nfs4_opendata_put(data);
}

static const struct rpc_call_ops nfs4_open_ops = {
	.rpc_call_prepare = nfs4_open_prepare,
	.rpc_call_done = nfs4_open_done,
	.rpc_release = nfs4_open_release,
};

/*
 * Note: On error, nfs4_proc_open will free the struct nfs4_opendata
 */
static int _nfs4_proc_open(struct nfs4_opendata *data)
{
	struct inode *dir = data->dir->d_inode;
	struct nfs_server *server = NFS_SERVER(dir);
	struct nfs_openargs *o_arg = &data->o_arg;
	struct nfs_openres *o_res = &data->o_res;
	struct rpc_task *task;
	int status;

	kref_get(&data->kref);
	/*
	 * If rpc_run_task() ends up calling ->rpc_release(), we
	 * want to ensure that it takes the 'error' code path.
	 */
	data->rpc_status = -ENOMEM;
	data->cancelled = 0;
	task = rpc_run_task(server->client, RPC_TASK_ASYNC, &nfs4_open_ops, data);
	if (IS_ERR(task))
		return PTR_ERR(task);
	status = nfs4_wait_for_completion_rpc_task(task);
	if (status != 0) {
		data->cancelled = 1;
		smp_wmb();
	} else
		status = data->rpc_status;
	rpc_put_task(task);
	if (status != 0)
		return status;

	if (o_arg->open_flags & O_CREAT) {
		update_changeattr(dir, &o_res->cinfo);
		nfs_post_op_update_inode(dir, o_res->dir_attr);
	} else
		nfs_refresh_inode(dir, o_res->dir_attr);
	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
		status = _nfs4_proc_open_confirm(data);
		if (status != 0)
			return status;
	}
	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR))
		return server->nfs_client->rpc_ops->getattr(server, &o_res->fh, o_res->f_attr);
	return 0;
}

static int _nfs4_do_access(struct inode *inode, struct rpc_cred *cred, int openflags)
{
	struct nfs_access_entry cache;
	int mask = 0;
	int status;

	if (openflags & FMODE_READ)
		mask |= MAY_READ;
	if (openflags & FMODE_WRITE)
		mask |= MAY_WRITE;
	if (openflags & FMODE_EXEC)
		mask |= MAY_EXEC;
	status = nfs_access_get_cached(inode, cred, &cache);
	if (status == 0)
		goto out;

	/* Be clever: ask server to check for all possible rights */
	cache.mask = MAY_EXEC | MAY_WRITE | MAY_READ;
	cache.cred = cred;
	cache.jiffies = jiffies;
	status = _nfs4_proc_access(inode, &cache);
	if (status != 0)
		return status;
	nfs_access_add_cache(inode, &cache);
out:
	if ((cache.mask & mask) == mask)
		return 0;
	return -EACCES;
}

static int nfs4_recover_expired_lease(struct nfs_server *server)
{
	struct nfs_client *clp = server->nfs_client;
	int ret;

	for (;;) {
		ret = nfs4_wait_clnt_recover(server->client, clp);
		if (ret != 0)
			return ret;
		if (!test_and_clear_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state))
			break;
		nfs4_schedule_state_recovery(clp);
	}
	return 0;
}

/*
 * OPEN_EXPIRED:
 * 	reclaim state on the server after a network partition.
 * 	Assumes caller holds the appropriate lock
 */
static int _nfs4_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
{
	struct inode *inode = state->inode;
	struct nfs_delegation *delegation = NFS_I(inode)->delegation;
	struct nfs4_opendata *opendata;
	int openflags = state->state & (FMODE_READ|FMODE_WRITE);
	int ret;

	if (delegation != NULL && !(delegation->flags & NFS_DELEGATION_NEED_RECLAIM)) {
		ret = _nfs4_do_access(inode, ctx->cred, openflags);
		if (ret < 0)
			return ret;
		memcpy(&state->stateid, &delegation->stateid, sizeof(state->stateid));
		set_bit(NFS_DELEGATED_STATE, &state->flags);
		return 0;
	}
	opendata = nfs4_opendata_alloc(&ctx->path, state->owner, openflags, NULL);
	if (opendata == NULL)
		return -ENOMEM;
	ret = nfs4_open_recover(opendata, state);
	if (ret == -ESTALE) {
		/* Invalidate the state owner so we don't ever use it again */
		nfs4_drop_state_owner(state->owner);
		d_drop(ctx->path.dentry);
	}
	nfs4_opendata_put(opendata);
	return ret;
}

static inline int nfs4_do_open_expired(struct nfs_open_context *ctx, struct nfs4_state *state)
{
	struct nfs_server *server = NFS_SERVER(state->inode);
	struct nfs4_exception exception = { };
	int err;

	do {
		err = _nfs4_open_expired(ctx, state);
		if (err == -NFS4ERR_DELAY)
			nfs4_handle_exception(server, err, &exception);
	} while (exception.retry);
	return err;
}

static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state)
{
	struct nfs_open_context *ctx;
	int ret;

	ctx = nfs4_state_find_open_context(state);
	if (IS_ERR(ctx))
		return PTR_ERR(ctx);
	ret = nfs4_do_open_expired(ctx, state);
	put_nfs_open_context(ctx);
	return ret;
}

/*
 * Returns a referenced nfs4_state if there is an open delegation on the file
 */
static int _nfs4_open_delegated(struct inode *inode, int flags, struct rpc_cred *cred, struct nfs4_state **res)
{
	struct nfs_delegation *delegation;
	struct nfs_server *server = NFS_SERVER(inode);
	struct nfs_client *clp = server->nfs_client;
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs4_state_owner *sp = NULL;
	struct nfs4_state *state = NULL;
	int open_flags = flags & (FMODE_READ|FMODE_WRITE);
	int err;

	err = -ENOMEM;
	if (!(sp = nfs4_get_state_owner(server, cred))) {
		dprintk("%s: nfs4_get_state_owner failed!\n", __FUNCTION__);
		return err;
	}
	err = nfs4_recover_expired_lease(server);
	if (err != 0)
		goto out_put_state_owner;
	/* Protect against reboot recovery - NOTE ORDER! */
	down_read(&clp->cl_sem);
	/* Protect against delegation recall */
	down_read(&nfsi->rwsem);
	delegation = NFS_I(inode)->delegation;
	err = -ENOENT;
	if (delegation == NULL || (delegation->type & open_flags) != open_flags)
		goto out_err;
	err = -ENOMEM;
	state = nfs4_get_open_state(inode, sp);
	if (state == NULL)
		goto out_err;

	err = -ENOENT;
	if ((state->state & open_flags) == open_flags) {
		spin_lock(&inode->i_lock);
		update_open_stateflags(state, open_flags);
		spin_unlock(&inode->i_lock);
		goto out_ok;
	} else if (state->state != 0)
		goto out_put_open_state;

	lock_kernel();
	err = _nfs4_do_access(inode, cred, open_flags);
	unlock_kernel();
	if (err != 0)
		goto out_put_open_state;
	update_open_stateid(state, NULL, &delegation->stateid, open_flags);
out_ok:
	nfs4_put_state_owner(sp);
	up_read(&nfsi->rwsem);
	up_read(&clp->cl_sem);
	*res = state;
	return 0;
out_put_open_state:
	nfs4_put_open_state(state);
out_err:
	up_read(&nfsi->rwsem);
	up_read(&clp->cl_sem);
	if (err != -EACCES)
		nfs_inode_return_delegation(inode);
out_put_state_owner:
	nfs4_put_state_owner(sp);
	return err;
}

static struct nfs4_state *nfs4_open_delegated(struct inode *inode, int flags, struct rpc_cred *cred)
{
	struct nfs4_exception exception = { };
	struct nfs4_state *res = ERR_PTR(-EIO);
	int err;

	do {
		err = _nfs4_open_delegated(inode, flags, cred, &res);
		if (err == 0)
			break;
		res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(inode),
					err, &exception));
	} while (exception.retry);
	return res;
}

/*
 * on an EXCLUSIVE create, the server should send back a bitmask with FATTR4-*
 * fields corresponding to attributes that were used to store the verifier.
 * Make sure we clobber those fields in the later setattr call
 */