NFSv4: Convert open() into an asynchronous RPC call

 OPEN is a stateful operation, so we must ensure that it always
 completes. In order to allow users to interrupt the operation,
 we need to make the RPC call asynchronous, and then wait on
 completion (or cancel).

 Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>

2 files changed, 130 insertions, 66 deletions

diff --git a/fs/nfs/nfs4proc.c b/fs/nfs/nfs4proc.c
index 4a5cc840..aed8701 100644
--- a/fs/nfs/nfs4proc.c
+++ b/fs/nfs/nfs4proc.c
@@ -195,6 +195,7 @@ static void update_changeattr(struct inode *inode, struct nfs4_change_info *cinf
 }
 
 struct nfs4_opendata {
+	atomic_t count;
 	struct nfs_openargs o_arg;
 	struct nfs_openres o_res;
 	struct nfs_fattr f_attr;
@@ -203,6 +204,8 @@ struct nfs4_opendata {
 	struct dentry *dir;
 	struct nfs4_state_owner *owner;
 	struct iattr attrs;
+	int rpc_status;
+	int cancelled;
 };
 
 static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
@@ -220,6 +223,7 @@ static struct nfs4_opendata *nfs4_opendata_alloc(struct dentry *dentry,
 	p->o_arg.seqid = nfs_alloc_seqid(&sp->so_seqid);
 	if (p->o_arg.seqid == NULL)
 		goto err_free;
+	atomic_set(&p->count, 1);
 	p->dentry = dget(dentry);
 	p->dir = parent;
 	p->owner = sp;
@@ -255,7 +259,7 @@ err:
 
 static void nfs4_opendata_free(struct nfs4_opendata *p)
 {
-	if (p != NULL) {
+	if (p != NULL && atomic_dec_and_test(&p->count)) {
 		nfs_free_seqid(p->o_arg.seqid);
 		nfs4_put_state_owner(p->owner);
 		dput(p->dir);
@@ -305,6 +309,26 @@ static void update_open_stateid(struct nfs4_state *state, nfs4_stateid *stateid,
 	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;
+
+	if (!(data->f_attr.valid & NFS_ATTR_FATTR))
+		goto out;
+	inode = nfs_fhget(data->dir->d_sb, &data->o_res.fh, &data->f_attr);
+	if (inode == NULL)
+		goto out;
+	state = nfs4_get_open_state(inode, data->owner);
+	if (state == NULL)
+		goto put_inode;
+	update_open_stateid(state, &data->o_res.stateid, data->o_arg.open_flags);
+put_inode:
+	iput(inode);
+out:
+	return state;
+}
+
 /*
  * OPEN_RECLAIM:
  * 	reclaim state on the server after a reboot.
@@ -473,41 +497,105 @@ static int _nfs4_proc_open_confirm(struct rpc_clnt *clnt, const struct nfs_fh *f
 	return status;
 }
 
-static int _nfs4_proc_open(struct inode *dir, struct nfs4_state_owner  *sp, struct nfs_openargs *o_arg, struct nfs_openres *o_res)
+static void nfs4_open_prepare(struct rpc_task *task, void *calldata)
 {
-	struct nfs_server *server = NFS_SERVER(dir);
+	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 = o_arg,
-		.rpc_resp = o_res,
+		.rpc_argp = &data->o_arg,
+		.rpc_resp = &data->o_res,
 		.rpc_cred = sp->so_cred,
 	};
-	int status;
+	
+	if (nfs_wait_on_sequence(data->o_arg.seqid, task) != 0)
+		return;
+	/* Update sequence id. */
+	data->o_arg.id = sp->so_id;
+	data->o_arg.clientid = sp->so_client->cl_clientid;
+	rpc_call_setup(task, &msg, 0);
+}
 
-	/* Update sequence id. The caller must serialize! */
-	o_arg->id = sp->so_id;
-	o_arg->clientid = sp->so_client->cl_clientid;
+static void nfs4_open_done(struct rpc_task *task, void *calldata)
+{
+	struct nfs4_opendata *data = calldata;
 
-	status = rpc_call_sync(server->client, &msg, RPC_TASK_NOINTR);
-	if (status == 0) {
-		/* OPEN on anything except a regular file is disallowed in NFSv4 */
-		switch (o_res->f_attr->mode & S_IFMT) {
+	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:
-				status = -ELOOP;
+				data->rpc_status = -ELOOP;
 				break;
 			case S_IFDIR:
-				status = -EISDIR;
+				data->rpc_status = -EISDIR;
 				break;
 			default:
-				status = -ENOTDIR;
+				data->rpc_status = -ENOTDIR;
 		}
 	}
+	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 (state != NULL)
+		nfs4_close_state(state, data->o_arg.open_flags);
+out_free:
+	nfs4_opendata_free(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,
+};
 
-	nfs_increment_open_seqid(status, o_arg->seqid);
+/*
+ * 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;
+
+	atomic_inc(&data->count);
+	task = rpc_run_task(server->client, RPC_TASK_ASYNC, &nfs4_open_ops, data);
+	if (IS_ERR(task)) {
+		nfs4_opendata_free(data);
+		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_release_task(task);
 	if (status != 0)
-		goto out;
+		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);
@@ -515,15 +603,14 @@ static int _nfs4_proc_open(struct inode *dir, struct nfs4_state_owner  *sp, stru
 		nfs_refresh_inode(dir, o_res->dir_attr);
 	if(o_res->rflags & NFS4_OPEN_RESULT_CONFIRM) {
 		status = _nfs4_proc_open_confirm(server->client, &o_res->fh,
-				sp, &o_res->stateid, o_arg->seqid);
+				data->owner, &o_res->stateid, o_arg->seqid);
 		if (status != 0)
-			goto out;
+			return status;
 	}
-	nfs_confirm_seqid(&sp->so_seqid, 0);
+	nfs_confirm_seqid(&data->owner->so_seqid, 0);
 	if (!(o_res->f_attr->valid & NFS_ATTR_FATTR))
-		status = server->rpc_ops->getattr(server, &o_res->fh, o_res->f_attr);
-out:
-	return status;
+		return server->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)
@@ -562,14 +649,15 @@ out:
 static int _nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state, struct dentry *dentry)
 {
 	struct dentry *parent = dget_parent(dentry);
-	struct inode *dir = parent->d_inode;
 	struct inode *inode = state->inode;
 	struct nfs_delegation *delegation = NFS_I(inode)->delegation;
 	struct nfs4_opendata *opendata;
+	struct nfs4_state *newstate;
+	int openflags = state->state & (FMODE_READ|FMODE_WRITE);
 	int status = 0;
 
 	if (delegation != NULL && !(delegation->flags & NFS_DELEGATION_NEED_RECLAIM)) {
-		status = _nfs4_do_access(inode, sp->so_cred, state->state);
+		status = _nfs4_do_access(inode, sp->so_cred, openflags);
 		if (status < 0)
 			goto out;
 		memcpy(&state->stateid, &delegation->stateid, sizeof(state->stateid));
@@ -577,27 +665,15 @@ static int _nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *st
 		goto out;
 	}
 	status = -ENOMEM;
-	opendata = nfs4_opendata_alloc(dentry, sp, state->state, NULL);
+	opendata = nfs4_opendata_alloc(dentry, sp, openflags, NULL);
 	if (opendata == NULL)
 		goto out;
-	status = _nfs4_proc_open(dir, sp, &opendata->o_arg, &opendata->o_res);
+	status = _nfs4_proc_open(opendata);
 	if (status != 0)
 		goto out_nodeleg;
-	/* Check if files differ */
-	if ((opendata->f_attr.mode & S_IFMT) != (inode->i_mode & S_IFMT))
+	newstate = nfs4_opendata_to_nfs4_state(opendata);
+	if (newstate != state)
 		goto out_stale;
-	/* Has the file handle changed? */
-	if (nfs_compare_fh(&opendata->o_res.fh, NFS_FH(inode)) != 0) {
-		/* Verify if the change attributes are the same */
-		if (opendata->f_attr.change_attr != NFS_I(inode)->change_attr)
-			goto out_stale;
-		if (nfs_size_to_loff_t(opendata->f_attr.size) != inode->i_size)
-			goto out_stale;
-		/* Lets just pretend that this is the same file */
-		nfs_copy_fh(NFS_FH(inode), &opendata->o_res.fh);
-		NFS_I(inode)->fileid = opendata->f_attr.fileid;
-	}
-	memcpy(&state->stateid, &opendata->o_res.stateid, sizeof(state->stateid));
 	if (opendata->o_res.delegation_type != 0) {
 		if (!(delegation->flags & NFS_DELEGATION_NEED_RECLAIM))
 			nfs_inode_set_delegation(inode, sp->so_cred,
@@ -606,6 +682,8 @@ static int _nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *st
 			nfs_inode_reclaim_delegation(inode, sp->so_cred,
 					&opendata->o_res);
 	}
+out_close_state:
+	nfs4_close_state(newstate, openflags);
 out_nodeleg:
 	nfs4_opendata_free(opendata);
 	clear_bit(NFS_DELEGATED_STATE, &state->flags);
@@ -618,7 +696,7 @@ out_stale:
 	nfs4_drop_state_owner(sp);
 	d_drop(dentry);
 	/* Should we be trying to close that stateid? */
-	goto out_nodeleg;
+	goto out_close_state;
 }
 
 static inline int nfs4_do_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *state, struct dentry *dentry)
@@ -656,7 +734,7 @@ static int nfs4_open_expired(struct nfs4_state_owner *sp, struct nfs4_state *sta
 }
 
 /*
- * Returns an nfs4_state + an extra reference to the inode
+ * 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)
 {
@@ -709,7 +787,6 @@ out_ok:
 	nfs4_put_state_owner(sp);
 	up_read(&nfsi->rwsem);
 	up_read(&clp->cl_sem);
-	igrab(inode);
 	*res = state;
 	return 0; 
 out_err:
@@ -742,7 +819,7 @@ static struct nfs4_state *nfs4_open_delegated(struct inode *inode, int flags, st
 }
 
 /*
- * Returns an nfs4_state + an referenced inode
+ * Returns a referenced nfs4_state
  */
 static int _nfs4_do_open(struct inode *dir, struct dentry *dentry, int flags, struct iattr *sattr, struct rpc_cred *cred, struct nfs4_state **res)
 {
@@ -750,7 +827,6 @@ static int _nfs4_do_open(struct inode *dir, struct dentry *dentry, int flags, st
 	struct nfs4_state     *state = NULL;
 	struct nfs_server       *server = NFS_SERVER(dir);
 	struct nfs4_client *clp = server->nfs4_state;
-	struct inode *inode = NULL;
 	struct nfs4_opendata *opendata;
 	int                     status;
 
@@ -765,20 +841,16 @@ static int _nfs4_do_open(struct inode *dir, struct dentry *dentry, int flags, st
 	if (opendata == NULL)
 		goto err_put_state_owner;
 
-	status = _nfs4_proc_open(dir, sp, &opendata->o_arg, &opendata->o_res);
+	status = _nfs4_proc_open(opendata);
 	if (status != 0)
 		goto err_opendata_free;
 
 	status = -ENOMEM;
-	inode = nfs_fhget(dir->i_sb, &opendata->o_res.fh, &opendata->f_attr);
-	if (!inode)
-		goto err_opendata_free;
-	state = nfs4_get_open_state(inode, sp);
-	if (!state)
+	state = nfs4_opendata_to_nfs4_state(opendata);
+	if (state == NULL)
 		goto err_opendata_free;
-	update_open_stateid(state, &opendata->o_res.stateid, flags);
 	if (opendata->o_res.delegation_type != 0)
-		nfs_inode_set_delegation(inode, cred, &opendata->o_res);
+		nfs_inode_set_delegation(state->inode, cred, &opendata->o_res);
 	nfs4_opendata_free(opendata);
 	nfs4_put_state_owner(sp);
 	up_read(&clp->cl_sem);
@@ -791,8 +863,6 @@ err_put_state_owner:
 out_err:
 	/* Note: clp->cl_sem must be released before nfs4_put_open_state()! */
 	up_read(&clp->cl_sem);
-	if (inode != NULL)
-		iput(inode);
 	*res = NULL;
 	return status;
 }
@@ -1066,7 +1136,7 @@ nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 			d_add(dentry, NULL);
 		return (struct dentry *)state;
 	}
-	res = d_add_unique(dentry, state->inode);
+	res = d_add_unique(dentry, igrab(state->inode));
 	if (res != NULL)
 		dentry = res;
 	nfs4_intent_set_file(nd, dentry, state);
@@ -1078,7 +1148,6 @@ nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags, st
 {
 	struct rpc_cred *cred;
 	struct nfs4_state *state;
-	struct inode *inode;
 
 	cred = rpcauth_lookupcred(NFS_SERVER(dir)->client->cl_auth, 0);
 	if (IS_ERR(cred))
@@ -1102,9 +1171,7 @@ nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags, st
 		}
 		goto out_drop;
 	}
-	inode = state->inode;
-	iput(inode);
-	if (inode == dentry->d_inode) {
+	if (state->inode == dentry->d_inode) {
 		nfs4_intent_set_file(nd, dentry, state);
 		return 1;
 	}
@@ -1633,7 +1700,7 @@ nfs4_proc_create(struct inode *dir, struct dentry *dentry, struct iattr *sattr,
 		status = PTR_ERR(state);
 		goto out;
 	}
-	d_instantiate(dentry, state->inode);
+	d_instantiate(dentry, igrab(state->inode));
 	if (flags & O_EXCL) {
 		struct nfs_fattr fattr;
 		status = nfs4_do_setattr(NFS_SERVER(dir), &fattr,
diff --git a/fs/nfs/nfs4xdr.c b/fs/nfs/nfs4xdr.c
index fbbace8..db2bcf7 100644
--- a/fs/nfs/nfs4xdr.c
+++ b/fs/nfs/nfs4xdr.c
@@ -1499,9 +1499,6 @@ static int nfs4_xdr_enc_open(struct rpc_rqst *req, uint32_t *p, struct nfs_opena
 	};
 	int status;
 
-	status = nfs_wait_on_sequence(args->seqid, req->rq_task);
-	if (status != 0)
-		goto out;
 	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
 	encode_compound_hdr(&xdr, &hdr);
 	status = encode_putfh(&xdr, args->fh);