[linux.git] / fs / nfs / delegation.c

/*
 * linux/fs/nfs/delegation.c
 *
 * Copyright (C) 2004 Trond Myklebust
 *
 * NFS file delegation management
 *
 */
#include <linux/completion.h>
#include <linux/kthread.h>
#include <linux/module.h>
#include <linux/sched.h>
#include <linux/spinlock.h>

#include <linux/nfs4.h>
#include <linux/nfs_fs.h>
#include <linux/nfs_xdr.h>

#include "nfs4_fs.h"
#include "delegation.h"
#include "internal.h"

static void nfs_do_free_delegation(struct nfs_delegation *delegation)
{
	kfree(delegation);
}

static void nfs_free_delegation_callback(struct rcu_head *head)
{
	struct nfs_delegation *delegation = container_of(head, struct nfs_delegation, rcu);

	nfs_do_free_delegation(delegation);
}

static void nfs_free_delegation(struct nfs_delegation *delegation)
{
	struct rpc_cred *cred;

	cred = rcu_dereference(delegation->cred);
	rcu_assign_pointer(delegation->cred, NULL);
	call_rcu(&delegation->rcu, nfs_free_delegation_callback);
	if (cred)
		put_rpccred(cred);
}

static int nfs_delegation_claim_locks(struct nfs_open_context *ctx, struct nfs4_state *state)
{
	struct inode *inode = state->inode;
	struct file_lock *fl;
	int status;

	for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) {
		if (!(fl->fl_flags & (FL_POSIX|FL_FLOCK)))
			continue;
		if (nfs_file_open_context(fl->fl_file) != ctx)
			continue;
		status = nfs4_lock_delegation_recall(state, fl);
		if (status >= 0)
			continue;
		switch (status) {
			default:
				printk(KERN_ERR "%s: unhandled error %d.\n",
						__FUNCTION__, status);
			case -NFS4ERR_EXPIRED:
				/* kill_proc(fl->fl_pid, SIGLOST, 1); */
			case -NFS4ERR_STALE_CLIENTID:
				nfs4_schedule_state_recovery(NFS_SERVER(inode)->nfs_client);
				goto out_err;
		}
	}
	return 0;
out_err:
	return status;
}

static void nfs_delegation_claim_opens(struct inode *inode, const nfs4_stateid *stateid)
{
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_open_context *ctx;
	struct nfs4_state *state;
	int err;

again:
	spin_lock(&inode->i_lock);
	list_for_each_entry(ctx, &nfsi->open_files, list) {
		state = ctx->state;
		if (state == NULL)
			continue;
		if (!test_bit(NFS_DELEGATED_STATE, &state->flags))
			continue;
		if (memcmp(state->stateid.data, stateid->data, sizeof(state->stateid.data)) != 0)
			continue;
		get_nfs_open_context(ctx);
		spin_unlock(&inode->i_lock);
		err = nfs4_open_delegation_recall(ctx, state, stateid);
		if (err >= 0)
			err = nfs_delegation_claim_locks(ctx, state);
		put_nfs_open_context(ctx);
		if (err != 0)
			return;
		goto again;
	}
	spin_unlock(&inode->i_lock);
}

/*
 * Set up a delegation on an inode
 */
void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res)
{
	struct nfs_delegation *delegation = NFS_I(inode)->delegation;
	struct rpc_cred *oldcred;

	if (delegation == NULL)
		return;
	memcpy(delegation->stateid.data, res->delegation.data,
			sizeof(delegation->stateid.data));
	delegation->type = res->delegation_type;
	delegation->maxsize = res->maxsize;
	oldcred = delegation->cred;
	delegation->cred = get_rpccred(cred);
	delegation->flags &= ~NFS_DELEGATION_NEED_RECLAIM;
	NFS_I(inode)->delegation_state = delegation->type;
	smp_wmb();
	put_rpccred(oldcred);
}

static int nfs_do_return_delegation(struct inode *inode, struct nfs_delegation *delegation, int issync)
{
	int res = 0;

	res = nfs4_proc_delegreturn(inode, delegation->cred, &delegation->stateid, issync);
	nfs_free_delegation(delegation);
	return res;
}

static struct nfs_delegation *nfs_detach_delegation_locked(struct nfs_inode *nfsi, const nfs4_stateid *stateid)
{
	struct nfs_delegation *delegation = rcu_dereference(nfsi->delegation);

	if (delegation == NULL)
		goto nomatch;
	if (stateid != NULL && memcmp(delegation->stateid.data, stateid->data,
				sizeof(delegation->stateid.data)) != 0)
		goto nomatch;
	list_del_rcu(&delegation->super_list);
	nfsi->delegation_state = 0;
	rcu_assign_pointer(nfsi->delegation, NULL);
	return delegation;
nomatch:
	return NULL;
}

/*
 * Set up a delegation on an inode
 */
int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res)
{
	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_delegation *delegation;
	struct nfs_delegation *freeme = NULL;
	int status = 0;

	delegation = kmalloc(sizeof(*delegation), GFP_KERNEL);
	if (delegation == NULL)
		return -ENOMEM;
	memcpy(delegation->stateid.data, res->delegation.data,
			sizeof(delegation->stateid.data));
	delegation->type = res->delegation_type;
	delegation->maxsize = res->maxsize;
	delegation->change_attr = nfsi->change_attr;
	delegation->cred = get_rpccred(cred);
	delegation->inode = inode;

	spin_lock(&clp->cl_lock);
	if (rcu_dereference(nfsi->delegation) != NULL) {
		if (memcmp(&delegation->stateid, &nfsi->delegation->stateid,
					sizeof(delegation->stateid)) == 0 &&
				delegation->type == nfsi->delegation->type) {
			goto out;
		}
		/*
		 * Deal with broken servers that hand out two
		 * delegations for the same file.
		 */
		dfprintk(FILE, "%s: server %s handed out "
				"a duplicate delegation!\n",
				__FUNCTION__, clp->cl_hostname);
		if (delegation->type <= nfsi->delegation->type) {
			freeme = delegation;
			delegation = NULL;
			goto out;
		}
		freeme = nfs_detach_delegation_locked(nfsi, NULL);
	}
	list_add_rcu(&delegation->super_list, &clp->cl_delegations);
	nfsi->delegation_state = delegation->type;
	rcu_assign_pointer(nfsi->delegation, delegation);
	delegation = NULL;

	/* Ensure we revalidate the attributes and page cache! */
	spin_lock(&inode->i_lock);
	nfsi->cache_validity |= NFS_INO_REVAL_FORCED;
	spin_unlock(&inode->i_lock);

out:
	spin_unlock(&clp->cl_lock);
	if (delegation != NULL)
		nfs_free_delegation(delegation);
	if (freeme != NULL)
		nfs_do_return_delegation(inode, freeme, 0);
	return status;
}

/* Sync all data to disk upon delegation return */
static void nfs_msync_inode(struct inode *inode)
{
	filemap_fdatawrite(inode->i_mapping);
	nfs_wb_all(inode);
	filemap_fdatawait(inode->i_mapping);
}

/*
 * Basic procedure for returning a delegation to the server
 */
static int __nfs_inode_return_delegation(struct inode *inode, struct nfs_delegation *delegation)
{
	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
	struct nfs_inode *nfsi = NFS_I(inode);

	nfs_msync_inode(inode);
	down_read(&clp->cl_sem);
	/* Guard against new delegated open calls */
	down_write(&nfsi->rwsem);
	nfs_delegation_claim_opens(inode, &delegation->stateid);
	up_write(&nfsi->rwsem);
	up_read(&clp->cl_sem);
	nfs_msync_inode(inode);

	return nfs_do_return_delegation(inode, delegation, 1);
}

/*
 * This function returns the delegation without reclaiming opens
 * or protecting against delegation reclaims.
 * It is therefore really only safe to be called from
 * nfs4_clear_inode()
 */
void nfs_inode_return_delegation_noreclaim(struct inode *inode)
{
	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_delegation *delegation;

	if (rcu_dereference(nfsi->delegation) != NULL) {
		spin_lock(&clp->cl_lock);
		delegation = nfs_detach_delegation_locked(nfsi, NULL);
		spin_unlock(&clp->cl_lock);
		if (delegation != NULL)
			nfs_do_return_delegation(inode, delegation, 0);
	}
}

int nfs_inode_return_delegation(struct inode *inode)
{
	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_delegation *delegation;
	int err = 0;

	if (rcu_dereference(nfsi->delegation) != NULL) {
		spin_lock(&clp->cl_lock);
		delegation = nfs_detach_delegation_locked(nfsi, NULL);
		spin_unlock(&clp->cl_lock);
		if (delegation != NULL)
			err = __nfs_inode_return_delegation(inode, delegation);
	}
	return err;
}

/*
 * Return all delegations associated to a super block
 */
void nfs_return_all_delegations(struct super_block *sb)
{
	struct nfs_client *clp = NFS_SB(sb)->nfs_client;
	struct nfs_delegation *delegation;
	struct inode *inode;

	if (clp == NULL)
		return;
restart:
	rcu_read_lock();
	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
		if (delegation->inode->i_sb != sb)
			continue;
		inode = igrab(delegation->inode);
		if (inode == NULL)
			continue;
		spin_lock(&clp->cl_lock);
		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
		spin_unlock(&clp->cl_lock);
		rcu_read_unlock();
		if (delegation != NULL)
			__nfs_inode_return_delegation(inode, delegation);
		iput(inode);
		goto restart;
	}
	rcu_read_unlock();
}

static int nfs_do_expire_all_delegations(void *ptr)
{
	struct nfs_client *clp = ptr;
	struct nfs_delegation *delegation;
	struct inode *inode;

	allow_signal(SIGKILL);
restart:
	if (test_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) != 0)
		goto out;
	if (test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) == 0)
		goto out;
	rcu_read_lock();
	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
		inode = igrab(delegation->inode);
		if (inode == NULL)
			continue;
		spin_lock(&clp->cl_lock);
		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
		spin_unlock(&clp->cl_lock);
		rcu_read_unlock();
		if (delegation)
			__nfs_inode_return_delegation(inode, delegation);
		iput(inode);
		goto restart;
	}
	rcu_read_unlock();
out:
	nfs_put_client(clp);
	module_put_and_exit(0);
}

void nfs_expire_all_delegations(struct nfs_client *clp)
{
	struct task_struct *task;

	__module_get(THIS_MODULE);
	atomic_inc(&clp->cl_count);
	task = kthread_run(nfs_do_expire_all_delegations, clp,
				"%s-delegreturn",
				rpc_peeraddr2str(clp->cl_rpcclient,
							RPC_DISPLAY_ADDR));
	if (!IS_ERR(task))
		return;
	nfs_put_client(clp);
	module_put(THIS_MODULE);
}

/*
 * Return all delegations following an NFS4ERR_CB_PATH_DOWN error.
 */
void nfs_handle_cb_pathdown(struct nfs_client *clp)
{
	struct nfs_delegation *delegation;
	struct inode *inode;

	if (clp == NULL)
		return;
restart:
	rcu_read_lock();
	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
		inode = igrab(delegation->inode);
		if (inode == NULL)
			continue;
		spin_lock(&clp->cl_lock);
		delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
		spin_unlock(&clp->cl_lock);
		rcu_read_unlock();
		if (delegation != NULL)
			__nfs_inode_return_delegation(inode, delegation);
		iput(inode);
		goto restart;
	}
	rcu_read_unlock();
}

struct recall_threadargs {
	struct inode *inode;
	struct nfs_client *clp;
	const nfs4_stateid *stateid;

	struct completion started;
	int result;
};

static int recall_thread(void *data)
{
	struct recall_threadargs *args = (struct recall_threadargs *)data;
	struct inode *inode = igrab(args->inode);
	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_delegation *delegation;

	daemonize("nfsv4-delegreturn");

	nfs_msync_inode(inode);
	down_read(&clp->cl_sem);
	down_write(&nfsi->rwsem);
	spin_lock(&clp->cl_lock);
	delegation = nfs_detach_delegation_locked(nfsi, args->stateid);
	if (delegation != NULL)
		args->result = 0;
	else
		args->result = -ENOENT;
	spin_unlock(&clp->cl_lock);
	complete(&args->started);
	nfs_delegation_claim_opens(inode, args->stateid);
	up_write(&nfsi->rwsem);
	up_read(&clp->cl_sem);
	nfs_msync_inode(inode);

	if (delegation != NULL)
		nfs_do_return_delegation(inode, delegation, 1);
	iput(inode);
	module_put_and_exit(0);
}

/*
 * Asynchronous delegation recall!
 */
int nfs_async_inode_return_delegation(struct inode *inode, const nfs4_stateid *stateid)
{
	struct recall_threadargs data = {
		.inode = inode,
		.stateid = stateid,
	};
	int status;

	init_completion(&data.started);
	__module_get(THIS_MODULE);
	status = kernel_thread(recall_thread, &data, CLONE_KERNEL);
	if (status < 0)
		goto out_module_put;
	wait_for_completion(&data.started);
	return data.result;
out_module_put:
	module_put(THIS_MODULE);
	return status;
}

/*
 * Retrieve the inode associated with a delegation
 */
struct inode *nfs_delegation_find_inode(struct nfs_client *clp, const struct nfs_fh *fhandle)
{
	struct nfs_delegation *delegation;
	struct inode *res = NULL;
	rcu_read_lock();
	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
		if (nfs_compare_fh(fhandle, &NFS_I(delegation->inode)->fh) == 0) {
			res = igrab(delegation->inode);
			break;
		}
	}
	rcu_read_unlock();
	return res;
}

/*
 * Mark all delegations as needing to be reclaimed
 */
void nfs_delegation_mark_reclaim(struct nfs_client *clp)
{
	struct nfs_delegation *delegation;
	rcu_read_lock();
	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list)
		delegation->flags |= NFS_DELEGATION_NEED_RECLAIM;
	rcu_read_unlock();
}

/*
 * Reap all unclaimed delegations after reboot recovery is done
 */
void nfs_delegation_reap_unclaimed(struct nfs_client *clp)
{
	struct nfs_delegation *delegation;
restart:
	rcu_read_lock();
	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
		if ((delegation->flags & NFS_DELEGATION_NEED_RECLAIM) == 0)
			continue;
		spin_lock(&clp->cl_lock);
		delegation = nfs_detach_delegation_locked(NFS_I(delegation->inode), NULL);
		spin_unlock(&clp->cl_lock);
		rcu_read_unlock();
		if (delegation != NULL)
			nfs_free_delegation(delegation);
		goto restart;
	}
	rcu_read_unlock();
}

int nfs4_copy_delegation_stateid(nfs4_stateid *dst, struct inode *inode)
{
	struct nfs_inode *nfsi = NFS_I(inode);
	struct nfs_delegation *delegation;
	int ret = 0;

	rcu_read_lock();
	delegation = rcu_dereference(nfsi->delegation);
	if (delegation != NULL) {
		memcpy(dst->data, delegation->stateid.data, sizeof(dst->data));
		ret = 1;
	}
	rcu_read_unlock();
	return ret;
}
Commit	Line	Data
1da177e4 LT	1	/*
	2	* linux/fs/nfs/delegation.c
	3	*
	4	* Copyright (C) 2004 Trond Myklebust
	5	*
	6	* NFS file delegation management
	7	*
	8	*/
1da177e4	9	#include <linux/completion.h>
58d9714a	10	#include <linux/kthread.h>
1da177e4 LT	11	#include <linux/module.h>
	12	#include <linux/sched.h>
	13	#include <linux/spinlock.h>
	14
	15	#include <linux/nfs4.h>
	16	#include <linux/nfs_fs.h>
	17	#include <linux/nfs_xdr.h>
	18
4ce79717	19	#include "nfs4_fs.h"
1da177e4	20	#include "delegation.h"
24c8dbbb	21	#include "internal.h"
1da177e4	22
905f8d16	23	static void nfs_do_free_delegation(struct nfs_delegation *delegation)
1da177e4	24	{
1da177e4 LT	25	kfree(delegation);
	26	}
	27
8383e460 TM	28	static void nfs_free_delegation_callback(struct rcu_head *head)
	29	{
	30	struct nfs_delegation *delegation = container_of(head, struct nfs_delegation, rcu);
	31
905f8d16 TM	32	nfs_do_free_delegation(delegation);
	33	}
	34
	35	static void nfs_free_delegation(struct nfs_delegation *delegation)
	36	{
	37	struct rpc_cred *cred;
	38
	39	cred = rcu_dereference(delegation->cred);
	40	rcu_assign_pointer(delegation->cred, NULL);
	41	call_rcu(&delegation->rcu, nfs_free_delegation_callback);
	42	if (cred)
	43	put_rpccred(cred);
8383e460 TM	44	}
8383e460 TM	45
888e694c TM	46	static int nfs_delegation_claim_locks(struct nfs_open_context ctx, struct nfs4_state state)
	47	{
	48	struct inode *inode = state->inode;
	49	struct file_lock *fl;
	50	int status;
	51
90dc7d27	52	for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) {
888e694c TM	53	if (!(fl->fl_flags & (FL_POSIX\|FL_FLOCK)))
888e694c TM	54	continue;
cd3758e3	55	if (nfs_file_open_context(fl->fl_file) != ctx)
888e694c TM	56	continue;
	57	status = nfs4_lock_delegation_recall(state, fl);
	58	if (status >= 0)
	59	continue;
	60	switch (status) {
	61	default:
	62	printk(KERN_ERR "%s: unhandled error %d.\n",
	63	__FUNCTION__, status);
	64	case -NFS4ERR_EXPIRED:
	65	/* kill_proc(fl->fl_pid, SIGLOST, 1); */
	66	case -NFS4ERR_STALE_CLIENTID:
7539bbab	67	nfs4_schedule_state_recovery(NFS_SERVER(inode)->nfs_client);
888e694c TM	68	goto out_err;
	69	}
	70	}
	71	return 0;
	72	out_err:
	73	return status;
	74	}
	75
90163027	76	static void nfs_delegation_claim_opens(struct inode inode, const nfs4_stateid stateid)
1da177e4 LT	77	{
	78	struct nfs_inode *nfsi = NFS_I(inode);
	79	struct nfs_open_context *ctx;
	80	struct nfs4_state *state;
888e694c	81	int err;
1da177e4 LT	82
	83	again:
	84	spin_lock(&inode->i_lock);
	85	list_for_each_entry(ctx, &nfsi->open_files, list) {
	86	state = ctx->state;
	87	if (state == NULL)
	88	continue;
	89	if (!test_bit(NFS_DELEGATED_STATE, &state->flags))
	90	continue;
90163027 TM	91	if (memcmp(state->stateid.data, stateid->data, sizeof(state->stateid.data)) != 0)
90163027 TM	92	continue;
1da177e4 LT	93	get_nfs_open_context(ctx);
1da177e4 LT	94	spin_unlock(&inode->i_lock);
13437e12	95	err = nfs4_open_delegation_recall(ctx, state, stateid);
888e694c TM	96	if (err >= 0)
888e694c TM	97	err = nfs_delegation_claim_locks(ctx, state);
1da177e4	98	put_nfs_open_context(ctx);
888e694c TM	99	if (err != 0)
888e694c TM	100	return;
1da177e4 LT	101	goto again;
	102	}
	103	spin_unlock(&inode->i_lock);
	104	}
	105
	106	/*
	107	* Set up a delegation on an inode
	108	*/
	109	void nfs_inode_reclaim_delegation(struct inode inode, struct rpc_cred cred, struct nfs_openres *res)
	110	{
	111	struct nfs_delegation *delegation = NFS_I(inode)->delegation;
05c88bab	112	struct rpc_cred *oldcred;
1da177e4 LT	113
	114	if (delegation == NULL)
	115	return;
	116	memcpy(delegation->stateid.data, res->delegation.data,
	117	sizeof(delegation->stateid.data));
	118	delegation->type = res->delegation_type;
	119	delegation->maxsize = res->maxsize;
05c88bab	120	oldcred = delegation->cred;
1da177e4 LT	121	delegation->cred = get_rpccred(cred);
	122	delegation->flags &= ~NFS_DELEGATION_NEED_RECLAIM;
	123	NFS_I(inode)->delegation_state = delegation->type;
	124	smp_wmb();
05c88bab	125	put_rpccred(oldcred);
1da177e4 LT	126	}
1da177e4 LT	127
57bfa891 TM	128	static int nfs_do_return_delegation(struct inode inode, struct nfs_delegation delegation, int issync)
	129	{
	130	int res = 0;
	131
	132	res = nfs4_proc_delegreturn(inode, delegation->cred, &delegation->stateid, issync);
	133	nfs_free_delegation(delegation);
	134	return res;
	135	}
	136
	137	static struct nfs_delegation nfs_detach_delegation_locked(struct nfs_inode nfsi, const nfs4_stateid *stateid)
	138	{
	139	struct nfs_delegation *delegation = rcu_dereference(nfsi->delegation);
	140
	141	if (delegation == NULL)
	142	goto nomatch;
	143	if (stateid != NULL && memcmp(delegation->stateid.data, stateid->data,
	144	sizeof(delegation->stateid.data)) != 0)
	145	goto nomatch;
	146	list_del_rcu(&delegation->super_list);
	147	nfsi->delegation_state = 0;
	148	rcu_assign_pointer(nfsi->delegation, NULL);
	149	return delegation;
	150	nomatch:
	151	return NULL;
	152	}
	153
1da177e4 LT	154	/*
	155	* Set up a delegation on an inode
	156	*/
	157	int nfs_inode_set_delegation(struct inode inode, struct rpc_cred cred, struct nfs_openres *res)
	158	{
7539bbab	159	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
1da177e4 LT	160	struct nfs_inode *nfsi = NFS_I(inode);
1da177e4 LT	161	struct nfs_delegation *delegation;
57bfa891	162	struct nfs_delegation *freeme = NULL;
1da177e4 LT	163	int status = 0;
1da177e4 LT	164
f52720ca	165	delegation = kmalloc(sizeof(*delegation), GFP_KERNEL);
1da177e4 LT	166	if (delegation == NULL)
	167	return -ENOMEM;
	168	memcpy(delegation->stateid.data, res->delegation.data,
	169	sizeof(delegation->stateid.data));
	170	delegation->type = res->delegation_type;
	171	delegation->maxsize = res->maxsize;
beb2a5ec	172	delegation->change_attr = nfsi->change_attr;
1da177e4 LT	173	delegation->cred = get_rpccred(cred);
	174	delegation->inode = inode;
	175
	176	spin_lock(&clp->cl_lock);
57bfa891	177	if (rcu_dereference(nfsi->delegation) != NULL) {
1da177e4	178	if (memcmp(&delegation->stateid, &nfsi->delegation->stateid,
57bfa891 TM	179	sizeof(delegation->stateid)) == 0 &&
	180	delegation->type == nfsi->delegation->type) {
	181	goto out;
1da177e4	182	}
57bfa891 TM	183	/*
	184	* Deal with broken servers that hand out two
	185	* delegations for the same file.
	186	*/
	187	dfprintk(FILE, "%s: server %s handed out "
	188	"a duplicate delegation!\n",
	189	__FUNCTION__, clp->cl_hostname);
	190	if (delegation->type <= nfsi->delegation->type) {
	191	freeme = delegation;
	192	delegation = NULL;
	193	goto out;
	194	}
	195	freeme = nfs_detach_delegation_locked(nfsi, NULL);
1da177e4	196	}
57bfa891 TM	197	list_add_rcu(&delegation->super_list, &clp->cl_delegations);
	198	nfsi->delegation_state = delegation->type;
	199	rcu_assign_pointer(nfsi->delegation, delegation);
	200	delegation = NULL;
412c77ce TM	201
	202	/* Ensure we revalidate the attributes and page cache! */
	203	spin_lock(&inode->i_lock);
	204	nfsi->cache_validity \|= NFS_INO_REVAL_FORCED;
	205	spin_unlock(&inode->i_lock);
	206
57bfa891	207	out:
1da177e4	208	spin_unlock(&clp->cl_lock);
603c83da TM	209	if (delegation != NULL)
603c83da TM	210	nfs_free_delegation(delegation);
57bfa891 TM	211	if (freeme != NULL)
57bfa891 TM	212	nfs_do_return_delegation(inode, freeme, 0);
1da177e4 LT	213	return status;
	214	}
	215
1da177e4 LT	216	/* Sync all data to disk upon delegation return */
	217	static void nfs_msync_inode(struct inode *inode)
	218	{
	219	filemap_fdatawrite(inode->i_mapping);
	220	nfs_wb_all(inode);
	221	filemap_fdatawait(inode->i_mapping);
	222	}
	223
	224	/*
	225	* Basic procedure for returning a delegation to the server
	226	*/
90163027	227	static int __nfs_inode_return_delegation(struct inode inode, struct nfs_delegation delegation)
1da177e4	228	{
7539bbab	229	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
1da177e4	230	struct nfs_inode *nfsi = NFS_I(inode);
1da177e4 LT	231
	232	nfs_msync_inode(inode);
	233	down_read(&clp->cl_sem);
	234	/* Guard against new delegated open calls */
	235	down_write(&nfsi->rwsem);
90163027	236	nfs_delegation_claim_opens(inode, &delegation->stateid);
1da177e4 LT	237	up_write(&nfsi->rwsem);
	238	up_read(&clp->cl_sem);
	239	nfs_msync_inode(inode);
	240
e6f81075	241	return nfs_do_return_delegation(inode, delegation, 1);
90163027 TM	242	}
90163027 TM	243
e6f81075 TM	244	/*
	245	* This function returns the delegation without reclaiming opens
	246	* or protecting against delegation reclaims.
	247	* It is therefore really only safe to be called from
	248	* nfs4_clear_inode()
	249	*/
	250	void nfs_inode_return_delegation_noreclaim(struct inode *inode)
	251	{
	252	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
	253	struct nfs_inode *nfsi = NFS_I(inode);
	254	struct nfs_delegation *delegation;
	255
	256	if (rcu_dereference(nfsi->delegation) != NULL) {
	257	spin_lock(&clp->cl_lock);
	258	delegation = nfs_detach_delegation_locked(nfsi, NULL);
	259	spin_unlock(&clp->cl_lock);
	260	if (delegation != NULL)
	261	nfs_do_return_delegation(inode, delegation, 0);
	262	}
	263	}
	264
90163027 TM	265	int nfs_inode_return_delegation(struct inode *inode)
	266	{
	267	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
	268	struct nfs_inode *nfsi = NFS_I(inode);
	269	struct nfs_delegation *delegation;
	270	int err = 0;
	271
8383e460	272	if (rcu_dereference(nfsi->delegation) != NULL) {
90163027 TM	273	spin_lock(&clp->cl_lock);
	274	delegation = nfs_detach_delegation_locked(nfsi, NULL);
	275	spin_unlock(&clp->cl_lock);
	276	if (delegation != NULL)
	277	err = __nfs_inode_return_delegation(inode, delegation);
	278	}
	279	return err;
1da177e4 LT	280	}
	281
	282	/*
	283	* Return all delegations associated to a super block
	284	*/
	285	void nfs_return_all_delegations(struct super_block *sb)
	286	{
7539bbab	287	struct nfs_client *clp = NFS_SB(sb)->nfs_client;
1da177e4 LT	288	struct nfs_delegation *delegation;
	289	struct inode *inode;
	290
	291	if (clp == NULL)
	292	return;
	293	restart:
8383e460 TM	294	rcu_read_lock();
8383e460 TM	295	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
1da177e4 LT	296	if (delegation->inode->i_sb != sb)
	297	continue;
	298	inode = igrab(delegation->inode);
	299	if (inode == NULL)
	300	continue;
8383e460 TM	301	spin_lock(&clp->cl_lock);
8383e460 TM	302	delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
1da177e4	303	spin_unlock(&clp->cl_lock);
8383e460 TM	304	rcu_read_unlock();
	305	if (delegation != NULL)
	306	__nfs_inode_return_delegation(inode, delegation);
1da177e4 LT	307	iput(inode);
	308	goto restart;
	309	}
8383e460	310	rcu_read_unlock();
1da177e4 LT	311	}
1da177e4 LT	312
10afec90	313	static int nfs_do_expire_all_delegations(void *ptr)
58d9714a	314	{
adfa6f98	315	struct nfs_client *clp = ptr;
58d9714a TM	316	struct nfs_delegation *delegation;
58d9714a TM	317	struct inode *inode;
58d9714a TM	318
	319	allow_signal(SIGKILL);
	320	restart:
58d9714a TM	321	if (test_bit(NFS4CLNT_STATE_RECOVER, &clp->cl_state) != 0)
	322	goto out;
	323	if (test_bit(NFS4CLNT_LEASE_EXPIRED, &clp->cl_state) == 0)
	324	goto out;
8383e460 TM	325	rcu_read_lock();
8383e460 TM	326	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
58d9714a TM	327	inode = igrab(delegation->inode);
	328	if (inode == NULL)
	329	continue;
8383e460 TM	330	spin_lock(&clp->cl_lock);
8383e460 TM	331	delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
58d9714a	332	spin_unlock(&clp->cl_lock);
8383e460 TM	333	rcu_read_unlock();
	334	if (delegation)
	335	__nfs_inode_return_delegation(inode, delegation);
58d9714a	336	iput(inode);
26c78e15	337	goto restart;
58d9714a	338	}
8383e460	339	rcu_read_unlock();
58d9714a	340	out:
24c8dbbb	341	nfs_put_client(clp);
58d9714a TM	342	module_put_and_exit(0);
	343	}
	344
adfa6f98	345	void nfs_expire_all_delegations(struct nfs_client *clp)
58d9714a TM	346	{
	347	struct task_struct *task;
	348
	349	__module_get(THIS_MODULE);
	350	atomic_inc(&clp->cl_count);
	351	task = kthread_run(nfs_do_expire_all_delegations, clp,
5d8515ca CL	352	"%s-delegreturn",
	353	rpc_peeraddr2str(clp->cl_rpcclient,
	354	RPC_DISPLAY_ADDR));
58d9714a TM	355	if (!IS_ERR(task))
58d9714a TM	356	return;
24c8dbbb	357	nfs_put_client(clp);
58d9714a TM	358	module_put(THIS_MODULE);
	359	}
	360
1da177e4 LT	361	/*
	362	* Return all delegations following an NFS4ERR_CB_PATH_DOWN error.
	363	*/
adfa6f98	364	void nfs_handle_cb_pathdown(struct nfs_client *clp)
1da177e4 LT	365	{
	366	struct nfs_delegation *delegation;
	367	struct inode *inode;
	368
	369	if (clp == NULL)
	370	return;
	371	restart:
8383e460 TM	372	rcu_read_lock();
8383e460 TM	373	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
1da177e4 LT	374	inode = igrab(delegation->inode);
	375	if (inode == NULL)
	376	continue;
8383e460 TM	377	spin_lock(&clp->cl_lock);
8383e460 TM	378	delegation = nfs_detach_delegation_locked(NFS_I(inode), NULL);
1da177e4	379	spin_unlock(&clp->cl_lock);
8383e460 TM	380	rcu_read_unlock();
	381	if (delegation != NULL)
	382	__nfs_inode_return_delegation(inode, delegation);
1da177e4 LT	383	iput(inode);
	384	goto restart;
	385	}
8383e460	386	rcu_read_unlock();
1da177e4 LT	387	}
	388
	389	struct recall_threadargs {
	390	struct inode *inode;
adfa6f98	391	struct nfs_client *clp;
1da177e4 LT	392	const nfs4_stateid *stateid;
	393
	394	struct completion started;
	395	int result;
	396	};
	397
	398	static int recall_thread(void *data)
	399	{
	400	struct recall_threadargs args = (struct recall_threadargs )data;
	401	struct inode *inode = igrab(args->inode);
7539bbab	402	struct nfs_client *clp = NFS_SERVER(inode)->nfs_client;
1da177e4 LT	403	struct nfs_inode *nfsi = NFS_I(inode);
	404	struct nfs_delegation *delegation;
	405
	406	daemonize("nfsv4-delegreturn");
	407
	408	nfs_msync_inode(inode);
	409	down_read(&clp->cl_sem);
	410	down_write(&nfsi->rwsem);
	411	spin_lock(&clp->cl_lock);
90163027 TM	412	delegation = nfs_detach_delegation_locked(nfsi, args->stateid);
90163027 TM	413	if (delegation != NULL)
1da177e4	414	args->result = 0;
90163027	415	else
1da177e4	416	args->result = -ENOENT;
1da177e4 LT	417	spin_unlock(&clp->cl_lock);
1da177e4 LT	418	complete(&args->started);
90163027	419	nfs_delegation_claim_opens(inode, args->stateid);
1da177e4 LT	420	up_write(&nfsi->rwsem);
	421	up_read(&clp->cl_sem);
	422	nfs_msync_inode(inode);
	423
	424	if (delegation != NULL)
e6f81075	425	nfs_do_return_delegation(inode, delegation, 1);
1da177e4 LT	426	iput(inode);
	427	module_put_and_exit(0);
	428	}
	429
	430	/*
	431	* Asynchronous delegation recall!
	432	*/
	433	int nfs_async_inode_return_delegation(struct inode inode, const nfs4_stateid stateid)
	434	{
	435	struct recall_threadargs data = {
	436	.inode = inode,
	437	.stateid = stateid,
	438	};
	439	int status;
	440
	441	init_completion(&data.started);
	442	__module_get(THIS_MODULE);
	443	status = kernel_thread(recall_thread, &data, CLONE_KERNEL);
	444	if (status < 0)
	445	goto out_module_put;
	446	wait_for_completion(&data.started);
	447	return data.result;
	448	out_module_put:
	449	module_put(THIS_MODULE);
	450	return status;
	451	}
	452
	453	/*
	454	* Retrieve the inode associated with a delegation
	455	*/
adfa6f98	456	struct inode nfs_delegation_find_inode(struct nfs_client clp, const struct nfs_fh *fhandle)
1da177e4 LT	457	{
	458	struct nfs_delegation *delegation;
	459	struct inode *res = NULL;
8383e460 TM	460	rcu_read_lock();
8383e460 TM	461	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
1da177e4 LT	462	if (nfs_compare_fh(fhandle, &NFS_I(delegation->inode)->fh) == 0) {
	463	res = igrab(delegation->inode);
	464	break;
	465	}
	466	}
8383e460	467	rcu_read_unlock();
1da177e4 LT	468	return res;
	469	}
	470
	471	/*
	472	* Mark all delegations as needing to be reclaimed
	473	*/
adfa6f98	474	void nfs_delegation_mark_reclaim(struct nfs_client *clp)
1da177e4 LT	475	{
1da177e4 LT	476	struct nfs_delegation *delegation;
8383e460 TM	477	rcu_read_lock();
8383e460 TM	478	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list)
1da177e4	479	delegation->flags \|= NFS_DELEGATION_NEED_RECLAIM;
8383e460	480	rcu_read_unlock();
1da177e4 LT	481	}
	482
	483	/*
	484	* Reap all unclaimed delegations after reboot recovery is done
	485	*/
adfa6f98	486	void nfs_delegation_reap_unclaimed(struct nfs_client *clp)
1da177e4	487	{
8383e460 TM	488	struct nfs_delegation *delegation;
	489	restart:
	490	rcu_read_lock();
	491	list_for_each_entry_rcu(delegation, &clp->cl_delegations, super_list) {
1da177e4 LT	492	if ((delegation->flags & NFS_DELEGATION_NEED_RECLAIM) == 0)
1da177e4 LT	493	continue;
8383e460 TM	494	spin_lock(&clp->cl_lock);
	495	delegation = nfs_detach_delegation_locked(NFS_I(delegation->inode), NULL);
	496	spin_unlock(&clp->cl_lock);
	497	rcu_read_unlock();
	498	if (delegation != NULL)
905f8d16	499	nfs_free_delegation(delegation);
8383e460	500	goto restart;
1da177e4	501	}
8383e460	502	rcu_read_unlock();
1da177e4	503	}
3e4f6290 TM	504
	505	int nfs4_copy_delegation_stateid(nfs4_stateid dst, struct inode inode)
	506	{
3e4f6290 TM	507	struct nfs_inode *nfsi = NFS_I(inode);
3e4f6290 TM	508	struct nfs_delegation *delegation;
8383e460	509	int ret = 0;
3e4f6290	510
8383e460 TM	511	rcu_read_lock();
8383e460 TM	512	delegation = rcu_dereference(nfsi->delegation);
3e4f6290 TM	513	if (delegation != NULL) {
3e4f6290 TM	514	memcpy(dst->data, delegation->stateid.data, sizeof(dst->data));
8383e460	515	ret = 1;
3e4f6290	516	}
8383e460 TM	517	rcu_read_unlock();
8383e460 TM	518	return ret;
3e4f6290	519	}