Merge fys.uio.no:/home/linux/bitkeeper/nfsclient-2.6

into fys.uio.no:/home/linux/bitkeeper/work/nfsclient-2.6

fs/nfs/dir.c

@@ -982,12 +982,18 @@ static int nfs_instantiate(struct dentry *dentry, struct nfs_fh *fhandle,
 	/* We may have been initialized further down */
 	if (dentry->d_inode)
 		return 0;
-	if (fhandle->size == 0 || !(fattr->valid & NFS_ATTR_FATTR)) {
+	if (fhandle->size == 0) {
 		struct inode *dir = dentry->d_parent->d_inode;
 		error = NFS_PROTO(dir)->lookup(dir, &dentry->d_name, fhandle, fattr);
 		if (error)
 			goto out_err;
 	}
+	if (!(fattr->valid & NFS_ATTR_FATTR)) {
+		struct nfs_server *server = NFS_SB(dentry->d_sb);
+		error = server->rpc_ops->getattr(server, fhandle, fattr);
+		if (error < 0)
+			goto out_err;
+	}
 	inode = nfs_fhget(dentry->d_sb, fhandle, fattr);
 	if (inode) {
 		d_instantiate(dentry, inode);

fs/nfs/direct.c

@@ -110,7 +110,7 @@ nfs_free_user_pages(struct page **pages, int npages, int do_dirty)
  * nfs_direct_read_seg - Read in one iov segment.  Generate separate
  *                        read RPCs for each "rsize" bytes.
  * @inode: target inode
- * @file: target file (may be NULL)
+ * @ctx: target file open context
  * user_addr: starting address of this segment of user's buffer
  * count: size of this segment
  * file_offset: offset in file to begin the operation
@@ -118,7 +118,7 @@ nfs_free_user_pages(struct page **pages, int npages, int do_dirty)
  * nr_pages: size of pages array
  */
 static int
-nfs_direct_read_seg(struct inode *inode, struct file *file,
+nfs_direct_read_seg(struct inode *inode, struct nfs_open_context *ctx,
 		unsigned long user_addr, size_t count, loff_t file_offset,
 		struct page **pages, int nr_pages)
 {
@@ -127,9 +127,10 @@ nfs_direct_read_seg(struct inode *inode, struct file *file,
 	int curpage = 0;
 	struct nfs_read_data	rdata = {
 		.inode		= inode,
+		.cred		= ctx->cred,
 		.args		= {
 			.fh		= NFS_FH(inode),
-			.lockowner	= current->files,
+			.context	= ctx,
 		},
 		.res		= {
 			.fattr		= &rdata.fattr,
@@ -151,7 +152,7 @@ nfs_direct_read_seg(struct inode *inode, struct file *file,
 			user_addr + tot_bytes, rdata.args.pgbase, curpage);
 
 		lock_kernel();
-		result = NFS_PROTO(inode)->read(&rdata, file);
+		result = NFS_PROTO(inode)->read(&rdata);
 		unlock_kernel();
 
 		if (result <= 0) {
@@ -183,7 +184,7 @@ nfs_direct_read_seg(struct inode *inode, struct file *file,
  * nfs_direct_read - For each iov segment, map the user's buffer
  *                   then generate read RPCs.
  * @inode: target inode
- * @file: target file (may be NULL)
+ * @ctx: target file open context
  * @iov: array of vectors that define I/O buffer
  * file_offset: offset in file to begin the operation
  * nr_segs: size of iovec array
@@ -193,7 +194,7 @@ nfs_direct_read_seg(struct inode *inode, struct file *file,
  * server.
  */
 static ssize_t
-nfs_direct_read(struct inode *inode, struct file *file,
+nfs_direct_read(struct inode *inode, struct nfs_open_context *ctx,
 		const struct iovec *iov, loff_t file_offset,
 		unsigned long nr_segs)
 {
@@ -216,7 +217,7 @@ nfs_direct_read(struct inode *inode, struct file *file,
                         return page_count;
                 }
 
-		result = nfs_direct_read_seg(inode, file, user_addr, size,
+		result = nfs_direct_read_seg(inode, ctx, user_addr, size,
 				file_offset, pages, page_count);
 
 		nfs_free_user_pages(pages, page_count, 1);
@@ -239,7 +240,7 @@ nfs_direct_read(struct inode *inode, struct file *file,
  * nfs_direct_write_seg - Write out one iov segment.  Generate separate
  *                        write RPCs for each "wsize" bytes, then commit.
  * @inode: target inode
- * @file: target file (may be NULL)
+ * @ctx: target file open context
  * user_addr: starting address of this segment of user's buffer
  * count: size of this segment
  * file_offset: offset in file to begin the operation
@@ -247,7 +248,7 @@ nfs_direct_read(struct inode *inode, struct file *file,
  * nr_pages: size of pages array
  */
 static int
-nfs_direct_write_seg(struct inode *inode, struct file *file,
+nfs_direct_write_seg(struct inode *inode, struct nfs_open_context *ctx,
 		unsigned long user_addr, size_t count, loff_t file_offset,
 		struct page **pages, int nr_pages)
 {
@@ -257,9 +258,10 @@ nfs_direct_write_seg(struct inode *inode, struct file *file,
 	struct nfs_writeverf first_verf;
 	struct nfs_write_data	wdata = {
 		.inode		= inode,
+		.cred		= ctx->cred,
 		.args		= {
 			.fh		= NFS_FH(inode),
-			.lockowner	= current->files,
+			.context	= ctx,
 		},
 		.res		= {
 			.fattr		= &wdata.fattr,
@@ -290,7 +292,7 @@ nfs_direct_write_seg(struct inode *inode, struct file *file,
 			user_addr + tot_bytes, wdata.args.pgbase, curpage);
 
 		lock_kernel();
-		result = NFS_PROTO(inode)->write(&wdata, file);
+		result = NFS_PROTO(inode)->write(&wdata);
 		unlock_kernel();
 
 		if (result <= 0) {
@@ -325,7 +327,7 @@ nfs_direct_write_seg(struct inode *inode, struct file *file,
 		wdata.args.offset = file_offset;
 
 		lock_kernel();
-		result = NFS_PROTO(inode)->commit(&wdata, file);
+		result = NFS_PROTO(inode)->commit(&wdata);
 		unlock_kernel();
 
 		if (result < 0 || memcmp(&first_verf.verifier,
@@ -349,7 +351,7 @@ nfs_direct_write_seg(struct inode *inode, struct file *file,
  * nfs_direct_write - For each iov segment, map the user's buffer
  *                    then generate write and commit RPCs.
  * @inode: target inode
- * @file: target file (may be NULL)
+ * @ctx: target file open context
  * @iov: array of vectors that define I/O buffer
  * file_offset: offset in file to begin the operation
  * nr_segs: size of iovec array
@@ -358,8 +360,7 @@ nfs_direct_write_seg(struct inode *inode, struct file *file,
  * that non-direct readers might access, so they will pick up these
  * writes immediately.
  */
-static ssize_t
-nfs_direct_write(struct inode *inode, struct file *file,
+static int nfs_direct_write(struct inode *inode, struct nfs_open_context *ctx,
 		const struct iovec *iov, loff_t file_offset,
 		unsigned long nr_segs)
 {
@@ -382,7 +383,7 @@ nfs_direct_write(struct inode *inode, struct file *file,
                         return page_count;
                 }
 
-		result = nfs_direct_write_seg(inode, file, user_addr, size,
+		result = nfs_direct_write_seg(inode, ctx, user_addr, size,
 				file_offset, pages, page_count);
 		nfs_free_user_pages(pages, page_count, 0);
 
@@ -414,6 +415,7 @@ nfs_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov,
 {
 	ssize_t result = -EINVAL;
 	struct file *file = iocb->ki_filp;
+	struct nfs_open_context *ctx;
 	struct dentry *dentry = file->f_dentry;
 	struct inode *inode = dentry->d_inode;
 
@@ -423,19 +425,20 @@ nfs_direct_IO(int rw, struct kiocb *iocb, const struct iovec *iov,
 	if (!is_sync_kiocb(iocb))
 		return result;
 
+	ctx = (struct nfs_open_context *)file->private_data;
 	switch (rw) {
 	case READ:
 		dprintk("NFS: direct_IO(read) (%s) off/no(%Lu/%lu)\n",
 				dentry->d_name.name, file_offset, nr_segs);
 
-		result = nfs_direct_read(inode, file, iov,
+		result = nfs_direct_read(inode, ctx, iov,
 						file_offset, nr_segs);
 		break;
 	case WRITE:
 		dprintk("NFS: direct_IO(write) (%s) off/no(%Lu/%lu)\n",
 				dentry->d_name.name, file_offset, nr_segs);
 
-		result = nfs_direct_write(inode, file, iov,
+		result = nfs_direct_write(inode, ctx, iov,
 						file_offset, nr_segs);
 		break;
 	default:
@@ -471,6 +474,8 @@ nfs_file_direct_read(struct kiocb *iocb, char __user *buf, size_t count, loff_t
 	ssize_t retval = -EINVAL;
 	loff_t *ppos = &iocb->ki_pos;
 	struct file *file = iocb->ki_filp;
+	struct nfs_open_context *ctx =
+			(struct nfs_open_context *) file->private_data;
 	struct dentry *dentry = file->f_dentry;
 	struct address_space *mapping = file->f_mapping;
 	struct inode *inode = mapping->host;
@@ -502,7 +507,7 @@ nfs_file_direct_read(struct kiocb *iocb, char __user *buf, size_t count, loff_t
 			goto out;
 	}
 
-	retval = nfs_direct_read(inode, file, &iov, pos, 1);
+	retval = nfs_direct_read(inode, ctx, &iov, pos, 1);
 	if (retval > 0)
 		*ppos = pos + retval;
 
@@ -542,6 +547,8 @@ nfs_file_direct_write(struct kiocb *iocb, const char __user *buf, size_t count,
 	loff_t *ppos = &iocb->ki_pos;
 	unsigned long limit = current->rlim[RLIMIT_FSIZE].rlim_cur;
 	struct file *file = iocb->ki_filp;
+	struct nfs_open_context *ctx =
+			(struct nfs_open_context *) file->private_data;
 	struct dentry *dentry = file->f_dentry;
 	struct address_space *mapping = file->f_mapping;
 	struct inode *inode = mapping->host;
@@ -589,7 +596,7 @@ nfs_file_direct_write(struct kiocb *iocb, const char __user *buf, size_t count,
 			goto out;
 	}
 
-	retval = nfs_direct_write(inode, file, &iov, pos, 1);
+	retval = nfs_direct_write(inode, ctx, &iov, pos, 1);
 	if (mapping->nrpages)
 		invalidate_inode_pages2(mapping);
 	if (retval > 0)

fs/nfs/file.c

@@ -113,6 +113,7 @@ nfs_file_release(struct inode *inode, struct file *filp)
 static int
 nfs_file_flush(struct file *file)
 {
+	struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data;
 	struct inode	*inode = file->f_dentry->d_inode;
 	int		status;
 
@@ -124,8 +125,8 @@ nfs_file_flush(struct file *file)
 	/* Ensure that data+attribute caches are up to date after close() */
 	status = nfs_wb_all(inode);
 	if (!status) {
-		status = file->f_error;
-		file->f_error = 0;
+		status = ctx->error;
+		ctx->error = 0;
 		if (!status)
 			__nfs_revalidate_inode(NFS_SERVER(inode), inode);
 	}
@@ -197,6 +198,7 @@ nfs_file_mmap(struct file * file, struct vm_area_struct * vma)
 static int
 nfs_fsync(struct file *file, struct dentry *dentry, int datasync)
 {
+	struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data;
 	struct inode *inode = dentry->d_inode;
 	int status;
 
@@ -205,8 +207,8 @@ nfs_fsync(struct file *file, struct dentry *dentry, int datasync)
 	lock_kernel();
 	status = nfs_wb_all(inode);
 	if (!status) {
-		status = file->f_error;
-		file->f_error = 0;
+		status = ctx->error;
+		ctx->error = 0;
 	}
 	unlock_kernel();
 	return status;

fs/nfs/inode.c

@@ -121,8 +121,9 @@ nfs_delete_inode(struct inode * inode)
 {
 	dprintk("NFS: delete_inode(%s/%ld)\n", inode->i_sb->s_id, inode->i_ino);
 
+	nfs_wb_all(inode);
 	/*
-	 * The following can never actually happen...
+	 * The following should never happen...
 	 */
 	if (nfs_have_writebacks(inode)) {
 		printk(KERN_ERR "nfs_delete_inode: inode %ld has pending RPC requests\n", inode->i_ino);
@@ -139,10 +140,10 @@ static void
 nfs_clear_inode(struct inode *inode)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
-	struct rpc_cred *cred = nfsi->mm_cred;
+	struct rpc_cred *cred;
 
-	if (cred)
-		put_rpccred(cred);
+	nfs_wb_all(inode);
+	BUG_ON (!list_empty(&nfsi->open_files));
 	cred = nfsi->cache_access.cred;
 	if (cred)
 		put_rpccred(cred);
@@ -812,53 +813,114 @@ int nfs_getattr(struct vfsmount *mnt, struct dentry *dentry, struct kstat *stat)
 	return err;
 }
 
+struct nfs_open_context *alloc_nfs_open_context(struct dentry *dentry, struct rpc_cred *cred)
+{
+	struct nfs_open_context *ctx;
+
+	ctx = (struct nfs_open_context *)kmalloc(sizeof(*ctx), GFP_KERNEL);
+	if (ctx != NULL) {
+		atomic_set(&ctx->count, 1);
+		ctx->dentry = dget(dentry);
+		ctx->cred = get_rpccred(cred);
+		ctx->state = NULL;
+		ctx->lockowner = current->files;
+		ctx->error = 0;
+		init_waitqueue_head(&ctx->waitq);
+	}
+	return ctx;
+}
+
+struct nfs_open_context *get_nfs_open_context(struct nfs_open_context *ctx)
+{
+	if (ctx != NULL)
+		atomic_inc(&ctx->count);
+	return ctx;
+}
+
+void put_nfs_open_context(struct nfs_open_context *ctx)
+{
+	if (atomic_dec_and_test(&ctx->count)) {
+		if (ctx->state != NULL)
+			nfs4_close_state(ctx->state, ctx->mode);
+		if (ctx->cred != NULL)
+			put_rpccred(ctx->cred);
+		dput(ctx->dentry);
+		kfree(ctx);
+	}
+}
+
 /*
  * Ensure that mmap has a recent RPC credential for use when writing out
  * shared pages
  */
-void
-nfs_set_mmcred(struct inode *inode, struct rpc_cred *cred)
+void nfs_file_set_open_context(struct file *filp, struct nfs_open_context *ctx)
 {
-	struct rpc_cred **p = &NFS_I(inode)->mm_cred,
-			*oldcred = *p;
+	struct inode *inode = filp->f_dentry->d_inode;
+	struct nfs_inode *nfsi = NFS_I(inode);
 
-	*p = get_rpccred(cred);
-	if (oldcred)
-		put_rpccred(oldcred);
+	filp->private_data = get_nfs_open_context(ctx);
+	spin_lock(&inode->i_lock);
+	list_add(&ctx->list, &nfsi->open_files);
+	spin_unlock(&inode->i_lock);
+}
+
+struct nfs_open_context *nfs_find_open_context(struct inode *inode, int mode)
+{
+	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs_open_context *pos, *ctx = NULL;
+
+	spin_lock(&inode->i_lock);
+	list_for_each_entry(pos, &nfsi->open_files, list) {
+		if ((pos->mode & mode) == mode) {
+			ctx = get_nfs_open_context(pos);
+			break;
+		}
+	}
+	spin_unlock(&inode->i_lock);
+	return ctx;
+}
+
+void nfs_file_clear_open_context(struct file *filp)
+{
+	struct inode *inode = filp->f_dentry->d_inode;
+	struct nfs_open_context *ctx = (struct nfs_open_context *)filp->private_data;
+
+	if (ctx) {
+		filp->private_data = NULL;
+		spin_lock(&inode->i_lock);
+		list_del(&ctx->list);
+		spin_unlock(&inode->i_lock);
+		put_nfs_open_context(ctx);
+	}
 }
 
 /*
- * These are probably going to contain hooks for
- * allocating and releasing RPC credentials for
- * the file. I'll have to think about Tronds patch
- * a bit more..
+ * These allocate and release file read/write context information.
  */
 int nfs_open(struct inode *inode, struct file *filp)
 {
-	struct rpc_auth *auth;
+	struct nfs_open_context *ctx;
 	struct rpc_cred *cred;
 
-	auth = NFS_CLIENT(inode)->cl_auth;
-	cred = rpcauth_lookupcred(auth, 0);
-	filp->private_data = cred;
-	if ((filp->f_mode & FMODE_WRITE) != 0) {
-		nfs_set_mmcred(inode, cred);
+	if ((cred = rpcauth_lookupcred(NFS_CLIENT(inode)->cl_auth, 0)) == NULL)
+		return -ENOMEM;
+	ctx = alloc_nfs_open_context(filp->f_dentry, cred);
+	put_rpccred(cred);
+	if (ctx == NULL)
+		return -ENOMEM;
+	ctx->mode = filp->f_mode;
+	nfs_file_set_open_context(filp, ctx);
+	put_nfs_open_context(ctx);
+	if ((filp->f_mode & FMODE_WRITE) != 0)
 		nfs_begin_data_update(inode);
-	}
 	return 0;
 }
 
 int nfs_release(struct inode *inode, struct file *filp)
 {
-	struct rpc_cred *cred;
-
-	lock_kernel();
 	if ((filp->f_mode & FMODE_WRITE) != 0)
 		nfs_end_data_update(inode);
-	cred = nfs_file_cred(filp);
-	if (cred)
-		put_rpccred(cred);
-	unlock_kernel();
+	nfs_file_clear_open_context(filp);
 	return 0;
 }
 
@@ -899,7 +961,7 @@ __nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
 
 	/* Protect against RPC races by saving the change attribute */
 	verifier = nfs_save_change_attribute(inode);
-	status = NFS_PROTO(inode)->getattr(inode, &fattr);
+	status = NFS_PROTO(inode)->getattr(server, NFS_FH(inode), &fattr);
 	if (status) {
 		dfprintk(PAGECACHE, "nfs_revalidate_inode: (%s/%Ld) getattr failed, error=%d\n",
 			 inode->i_sb->s_id,
@@ -1397,6 +1459,9 @@ static void nfs4_clear_inode(struct inode *inode)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 
+	/* First call standard NFS clear_inode() code */
+	nfs_clear_inode(inode);
+	/* Now clear out any remaining state */
 	while (!list_empty(&nfsi->open_states)) {
 		struct nfs4_state *state;
 		
@@ -1411,8 +1476,6 @@ static void nfs4_clear_inode(struct inode *inode)
 		BUG_ON(atomic_read(&state->count) != 1);
 		nfs4_close_state(state, state->state);
 	}
-	/* Now call standard NFS clear_inode() code */
-	nfs_clear_inode(inode);
 }
 
 
@@ -1510,8 +1573,13 @@ static int nfs4_fill_super(struct super_block *sb, struct nfs4_mount_data *data,
 		memcpy(clp->cl_ipaddr, server->ip_addr, sizeof(clp->cl_ipaddr));
 		nfs_idmap_new(clp);
 	}
-	if (list_empty(&clp->cl_superblocks))
-		clear_bit(NFS4CLNT_OK, &clp->cl_state);
+	if (list_empty(&clp->cl_superblocks)) {
+		err = nfs4_init_client(clp);
+		if (err != 0) {
+			up_write(&clp->cl_sem);
+			goto out_fail;
+		}
+	}
 	list_add_tail(&server->nfs4_siblings, &clp->cl_superblocks);
 	clnt = rpc_clone_client(clp->cl_rpcclient);
 	if (!IS_ERR(clnt))
@@ -1712,7 +1780,6 @@ static struct inode *nfs_alloc_inode(struct super_block *sb)
 	if (!nfsi)
 		return NULL;
 	nfsi->flags = 0;
-	nfsi->mm_cred = NULL;
 	nfs4_zero_state(nfsi);
 	return &nfsi->vfs_inode;
 }
@@ -1732,6 +1799,7 @@ static void init_once(void * foo, kmem_cache_t * cachep, unsigned long flags)
 		spin_lock_init(&nfsi->req_lock);
 		INIT_LIST_HEAD(&nfsi->dirty);
 		INIT_LIST_HEAD(&nfsi->commit);
+		INIT_LIST_HEAD(&nfsi->open_files);
 		INIT_RADIX_TREE(&nfsi->nfs_page_tree, GFP_ATOMIC);
 		atomic_set(&nfsi->data_updates, 0);
 		nfsi->ndirty = 0;

fs/nfs/nfs3proc.c

@@ -68,18 +68,6 @@ nfs3_async_handle_jukebox(struct rpc_task *task)
 	return 1;
 }
 
-static struct rpc_cred *
-nfs_cred(struct inode *inode, struct file *filp)
-{
-	struct rpc_cred *cred = NULL;
-
-	if (filp)
-		cred = (struct rpc_cred *)filp->private_data;
-	if (!cred)
-		cred = NFS_I(inode)->mm_cred;
-	return cred;
-}
-
 /*
  * Bare-bones access to getattr: this is for nfs_read_super.
  */
@@ -104,14 +92,15 @@ nfs3_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
  * One function for each procedure in the NFS protocol.
  */
 static int
-nfs3_proc_getattr(struct inode *inode, struct nfs_fattr *fattr)
+nfs3_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
+		struct nfs_fattr *fattr)
 {
 	int	status;
 
 	dprintk("NFS call  getattr\n");
 	fattr->valid = 0;
-	status = rpc_call(NFS_CLIENT(inode), NFS3PROC_GETATTR,
-			  NFS_FH(inode), fattr, 0);
+	status = rpc_call(server->client, NFS3PROC_GETATTR,
+			  fhandle, fattr, 0);
 	dprintk("NFS reply getattr\n");
 	return status;
 }
@@ -233,8 +222,7 @@ nfs3_proc_readlink(struct inode *inode, struct page *page)
 	return status;
 }
 
-static int
-nfs3_proc_read(struct nfs_read_data *rdata, struct file *filp)
+static int nfs3_proc_read(struct nfs_read_data *rdata)
 {
 	int			flags = rdata->flags;
 	struct inode *		inode = rdata->inode;
@@ -243,13 +231,13 @@ nfs3_proc_read(struct nfs_read_data *rdata, struct file *filp)
 		.rpc_proc	= &nfs3_procedures[NFS3PROC_READ],
 		.rpc_argp	= &rdata->args,
 		.rpc_resp	= &rdata->res,
+		.rpc_cred	= rdata->cred,
 	};
 	int			status;
 
 	dprintk("NFS call  read %d @ %Ld\n", rdata->args.count,
 			(long long) rdata->args.offset);
 	fattr->valid = 0;
-	msg.rpc_cred = nfs_cred(inode, filp);
 	status = rpc_call_sync(NFS_CLIENT(inode), &msg, flags);
 	if (status >= 0)
 		nfs_refresh_inode(inode, fattr);
@@ -257,8 +245,7 @@ nfs3_proc_read(struct nfs_read_data *rdata, struct file *filp)
 	return status;
 }
 
-static int
-nfs3_proc_write(struct nfs_write_data *wdata, struct file *filp)
+static int nfs3_proc_write(struct nfs_write_data *wdata)
 {
 	int			rpcflags = wdata->flags;
 	struct inode *		inode = wdata->inode;
@@ -267,13 +254,13 @@ nfs3_proc_write(struct nfs_write_data *wdata, struct file *filp)
 		.rpc_proc	= &nfs3_procedures[NFS3PROC_WRITE],
 		.rpc_argp	= &wdata->args,
 		.rpc_resp	= &wdata->res,
+		.rpc_cred	= wdata->cred,
 	};
 	int			status;
 
 	dprintk("NFS call  write %d @ %Ld\n", wdata->args.count,
 			(long long) wdata->args.offset);
 	fattr->valid = 0;
-	msg.rpc_cred = nfs_cred(inode, filp);
 	status = rpc_call_sync(NFS_CLIENT(inode), &msg, rpcflags);
 	if (status >= 0)
 		nfs_refresh_inode(inode, fattr);
@@ -281,8 +268,7 @@ nfs3_proc_write(struct nfs_write_data *wdata, struct file *filp)
 	return status < 0? status : wdata->res.count;
 }
 
-static int
-nfs3_proc_commit(struct nfs_write_data *cdata, struct file *filp)
+static int nfs3_proc_commit(struct nfs_write_data *cdata)
 {
 	struct inode *		inode = cdata->inode;
 	struct nfs_fattr *	fattr = cdata->res.fattr;
@@ -290,13 +276,13 @@ nfs3_proc_commit(struct nfs_write_data *cdata, struct file *filp)
 		.rpc_proc	= &nfs3_procedures[NFS3PROC_COMMIT],
 		.rpc_argp	= &cdata->args,
 		.rpc_resp	= &cdata->res,
+		.rpc_cred	= cdata->cred,
 	};
 	int			status;
 
 	dprintk("NFS call  commit %d @ %Ld\n", cdata->args.count,
 			(long long) cdata->args.offset);
 	fattr->valid = 0;
-	msg.rpc_cred = nfs_cred(inode, filp);
 	status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
 	if (status >= 0)
 		nfs_refresh_inode(inode, fattr);
@@ -840,27 +826,6 @@ nfs3_proc_commit_setup(struct nfs_write_data *data, int how)
 	rpc_call_setup(task, &msg, 0);
 }
 
-/*
- * Set up the nfspage struct with the right credentials
- */
-void
-nfs3_request_init(struct nfs_page *req, struct file *filp)
-{
-	req->wb_cred = get_rpccred(nfs_cred(req->wb_inode, filp));
-}
-
-static int
-nfs3_request_compatible(struct nfs_page *req, struct file *filp, struct page *page)
-{
-	if (req->wb_file != filp)
-		return 0;
-	if (req->wb_page != page)
-		return 0;
-	if (req->wb_cred != nfs_file_cred(filp))
-		return 0;
-	return 1;
-}
-
 static int
 nfs3_proc_lock(struct file *filp, int cmd, struct file_lock *fl)
 {
@@ -900,7 +865,5 @@ struct nfs_rpc_ops	nfs_v3_clientops = {
 	.commit_setup	= nfs3_proc_commit_setup,
 	.file_open	= nfs_open,
 	.file_release	= nfs_release,
-	.request_init	= nfs3_request_init,
-	.request_compatible = nfs3_request_compatible,
 	.lock		= nfs3_proc_lock,
 };

fs/nfs/nfs4proc.c

@@ -49,7 +49,8 @@
 
 #define NFSDBG_FACILITY		NFSDBG_PROC
 
-#define NFS4_POLL_RETRY_TIME	(15*HZ)
+#define NFS4_POLL_RETRY_MIN	(1*HZ)
+#define NFS4_POLL_RETRY_MAX	(15*HZ)
 
 static int nfs4_do_fsinfo(struct nfs_server *, struct nfs_fh *, struct nfs_fsinfo *);
 static int nfs4_async_handle_error(struct rpc_task *, struct nfs_server *);
@@ -189,8 +190,7 @@ static void update_changeattr(struct inode *inode, struct nfs4_change_info *cinf
  * 	reclaim state on the server after a reboot.
  * 	Assumes caller is holding the sp->so_sem
  */
-int
-nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state)
+static int _nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state)
 {
 	struct inode *inode = state->inode;
 	struct nfs_server *server = NFS_SERVER(inode);
@@ -227,15 +227,34 @@ nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state)
 	return status;
 }
 
+int nfs4_open_reclaim(struct nfs4_state_owner *sp, struct nfs4_state *state)
+{
+	struct nfs_server *server = NFS_SERVER(state->inode);
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = _nfs4_open_reclaim(sp, state);
+		switch (err) {
+			case 0:
+			case -NFS4ERR_STALE_CLIENTID:
+			case -NFS4ERR_STALE_STATEID:
+			case -NFS4ERR_EXPIRED:
+				return err;
+		}
+		err = nfs4_handle_exception(server, err, &exception);
+	} while (exception.retry);
+	return err;
+}
+
 /*
  * Returns an nfs4_state + an referenced inode
  */
-struct nfs4_state *
-nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *sattr, struct rpc_cred *cred)
+static int _nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *sattr, struct rpc_cred *cred, struct nfs4_state **res)
 {
 	struct nfs4_state_owner  *sp;
 	struct nfs4_state     *state = NULL;
 	struct nfs_server       *server = NFS_SERVER(dir);
+	struct nfs4_client *clp = server->nfs4_state;
 	struct inode *inode = NULL;
 	int                     status;
 	struct nfs_fattr        f_attr = {
@@ -261,11 +280,12 @@ nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *satt
 		.rpc_cred	= cred,
 	};
 
-retry:
+	/* Protect against reboot recovery conflicts */
+	down_read(&clp->cl_sem);
 	status = -ENOMEM;
-	if (!(sp = nfs4_get_state_owner(NFS_SERVER(dir), cred))) {
+	if (!(sp = nfs4_get_state_owner(server, cred))) {
 		dprintk("nfs4_do_open: nfs4_get_state_owner failed!\n");
-		goto out;
+		goto out_err;
 	}
 	if (o_arg.createmode & NFS4_CREATE_EXCLUSIVE){
 		u32 *p = (u32 *) o_arg.u.verifier.data;
@@ -283,16 +303,21 @@ nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *satt
 	status = rpc_call_sync(server->client, &msg, 0);
 	nfs4_increment_seqid(status, sp);
 	if (status)
-		goto out_up;
+		goto out_err;
 	update_changeattr(dir, &o_res.cinfo);
+	if (!(f_attr.valid & NFS_ATTR_FATTR)) {
+		status = server->rpc_ops->getattr(server, &o_res.fh, &f_attr);
+		if (status < 0)
+			goto out_err;
+	}
 
 	status = -ENOMEM;
 	inode = nfs_fhget(dir->i_sb, &o_res.fh, &f_attr);
 	if (!inode)
-		goto out_up;
+		goto out_err;
 	state = nfs4_get_open_state(inode, sp);
 	if (!state)
-		goto out_up;
+		goto out_err;
 
 	if(o_res.rflags & NFS4_OPEN_RESULT_CONFIRM) {
 		struct nfs_open_confirmargs oc_arg = {
@@ -311,7 +336,7 @@ nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *satt
 		status = rpc_call_sync(server->client, &msg, 0);
 		nfs4_increment_seqid(status, sp);
 		if (status)
-			goto out_up;
+			goto out_err;
 		memcpy(&state->stateid, &oc_res.stateid, sizeof(state->stateid));
 	} else
 		memcpy(&state->stateid, &o_res.stateid, sizeof(state->stateid));
@@ -325,44 +350,58 @@ nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *satt
 
 	up(&sp->so_sema);
 	nfs4_put_state_owner(sp);
-	return state;
-
-out_up:
-	up(&sp->so_sema);
-	nfs4_put_state_owner(sp);
-	if (state) {
-		nfs4_put_open_state(state);
-		state = NULL;
+	up_read(&clp->cl_sem);
+	*res = state;
+	return 0;
+out_err:
+	if (sp != NULL) {
+		if (state != NULL)
+			nfs4_put_open_state(state);
+		up(&sp->so_sema);
+		nfs4_put_state_owner(sp);
 	}
-	if (inode) {
+	/* Note: clp->cl_sem must be released before nfs4_put_open_state()! */
+	up_read(&clp->cl_sem);
+	if (inode != NULL)
 		iput(inode);
-		inode = NULL;
-	}
-	/* NOTE: BAD_SEQID means the server and client disagree about the
-	 * book-keeping w.r.t. state-changing operations
-	 * (OPEN/CLOSE/LOCK/LOCKU...)
-	 * It is actually a sign of a bug on the client or on the server.
-	 *
-	 * If we receive a BAD_SEQID error in the particular case of
-	 * doing an OPEN, we assume that nfs4_increment_seqid() will
-	 * have unhashed the old state_owner for us, and that we can
-	 * therefore safely retry using a new one. We should still warn
-	 * the user though...
-	 */
-	if (status == -NFS4ERR_BAD_SEQID) {
-		printk(KERN_WARNING "NFS: v4 server returned a bad sequence-id error!\n");
-		goto retry;
-	}
-	status = nfs4_handle_error(server, status);
-	if (!status)
-		goto retry;
-	BUG_ON(status < -1000 || status > 0);
-out:
-	return ERR_PTR(status);
+	*res = NULL;
+	return status;
 }
 
-int
-nfs4_do_setattr(struct nfs_server *server, struct nfs_fattr *fattr,
+
+struct nfs4_state *nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *sattr, struct rpc_cred *cred)
+{
+	struct nfs4_exception exception = { };
+	struct nfs4_state *res;
+	int status;
+
+	do {
+		status = _nfs4_do_open(dir, name, flags, sattr, cred, &res);
+		if (status == 0)
+			break;
+		/* NOTE: BAD_SEQID means the server and client disagree about the
+		 * book-keeping w.r.t. state-changing operations
+		 * (OPEN/CLOSE/LOCK/LOCKU...)
+		 * It is actually a sign of a bug on the client or on the server.
+		 *
+		 * If we receive a BAD_SEQID error in the particular case of
+		 * doing an OPEN, we assume that nfs4_increment_seqid() will
+		 * have unhashed the old state_owner for us, and that we can
+		 * therefore safely retry using a new one. We should still warn
+		 * the user though...
+		 */
+		if (status == -NFS4ERR_BAD_SEQID) {
+			printk(KERN_WARNING "NFS: v4 server returned a bad sequence-id error!\n");
+			exception.retry = 1;
+			continue;
+		}
+		res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(dir),
+					status, &exception));
+	} while (exception.retry);
+	return res;
+}
+
+static int _nfs4_do_setattr(struct nfs_server *server, struct nfs_fattr *fattr,
                 struct nfs_fh *fhandle, struct iattr *sattr,
                 struct nfs4_state *state)
 {
@@ -381,9 +420,7 @@ nfs4_do_setattr(struct nfs_server *server, struct nfs_fattr *fattr,
                 .rpc_argp       = &arg,
                 .rpc_resp       = &res,
         };
-	int status;
 
-retry:
         fattr->valid = 0;
 
 	if (sattr->ia_valid & ATTR_SIZE)
@@ -391,13 +428,22 @@ nfs4_do_setattr(struct nfs_server *server, struct nfs_fattr *fattr,
 	else
 		memcpy(&arg.stateid, &zero_stateid, sizeof(arg.stateid));
 
-	status = rpc_call_sync(server->client, &msg, 0);
-	if (status) {
-		status = nfs4_handle_error(server, status);
-		if (!status)
-			goto retry;
-	}
-	return status;
+	return rpc_call_sync(server->client, &msg, 0);
+}
+
+int nfs4_do_setattr(struct nfs_server *server, struct nfs_fattr *fattr,
+                struct nfs_fh *fhandle, struct iattr *sattr,
+                struct nfs4_state *state)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(server,
+				_nfs4_do_setattr(server, fattr, fhandle, sattr,
+					state),
+				&exception);
+	} while (exception.retry);
+	return err;
 }
 
 /* 
@@ -411,8 +457,7 @@ nfs4_do_setattr(struct nfs_server *server, struct nfs_fattr *fattr,
  *
  * NOTE: Caller must be holding the sp->so_owner semaphore!
  */
-int
-nfs4_do_close(struct inode *inode, struct nfs4_state *state) 
+static int _nfs4_do_close(struct inode *inode, struct nfs4_state *state) 
 {
 	struct nfs4_state_owner *sp = state->owner;
 	int status = 0;
@@ -441,8 +486,27 @@ nfs4_do_close(struct inode *inode, struct nfs4_state *state)
 	return status;
 }
 
-int
-nfs4_do_downgrade(struct inode *inode, struct nfs4_state *state, mode_t mode) 
+int nfs4_do_close(struct inode *inode, struct nfs4_state *state) 
+{
+	struct nfs_server *server = NFS_SERVER(state->inode);
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = _nfs4_do_close(inode, state);
+		switch (err) {
+			case -NFS4ERR_STALE_STATEID:
+			case -NFS4ERR_EXPIRED:
+				nfs4_schedule_state_recovery(server->nfs4_state);
+				err = 0;
+			default:
+				state->state = 0;
+		}
+		err = nfs4_handle_exception(server, err, &exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_do_downgrade(struct inode *inode, struct nfs4_state *state, mode_t mode) 
 {
 	struct nfs4_state_owner *sp = state->owner;
 	int status = 0;
@@ -467,6 +531,26 @@ nfs4_do_downgrade(struct inode *inode, struct nfs4_state *state, mode_t mode)
 	return status;
 }
 
+int nfs4_do_downgrade(struct inode *inode, struct nfs4_state *state, mode_t mode) 
+{
+	struct nfs_server *server = NFS_SERVER(state->inode);
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = _nfs4_do_downgrade(inode, state, mode);
+		switch (err) {
+			case -NFS4ERR_STALE_STATEID:
+			case -NFS4ERR_EXPIRED:
+				nfs4_schedule_state_recovery(server->nfs4_state);
+				err = 0;
+			default:
+				state->state = mode;
+		}
+		err = nfs4_handle_exception(server, err, &exception);
+	} while (exception.retry);
+	return err;
+}
+
 struct inode *
 nfs4_atomic_open(struct inode *dir, struct dentry *dentry, struct nameidata *nd)
 {
@@ -518,7 +602,7 @@ nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags)
 }
 
 
-static int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
+static int _nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
 {
 	struct nfs4_server_caps_res res = {};
 	struct rpc_message msg = {
@@ -542,7 +626,19 @@ static int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fh
 	return status;
 }
 
-static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
+static int nfs4_server_capabilities(struct nfs_server *server, struct nfs_fh *fhandle)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(server,
+				_nfs4_server_capabilities(server, fhandle),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
 		struct nfs_fsinfo *info)
 {
 	struct nfs_fattr *	fattr = info->fattr;
@@ -563,6 +659,19 @@ static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
 	return rpc_call_sync(server->client, &msg, 0);
 }
 
+static int nfs4_lookup_root(struct nfs_server *server, struct nfs_fh *fhandle,
+		struct nfs_fsinfo *info)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(server,
+				_nfs4_lookup_root(server, fhandle, info),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
 static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
 		struct nfs_fsinfo *info)
 {
@@ -597,6 +706,8 @@ static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
 
 	p = server->mnt_path;
 	for (;;) {
+		struct nfs4_exception exception = { };
+
 		while (*p == '/')
 			p++;
 		if (!*p)
@@ -606,9 +717,13 @@ static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
 			p++;
 		q.len = p - q.name;
 
-		fattr->valid = 0;
-		status = rpc_call_sync(server->client, &msg, 0);
-		if (!status)
+		do {
+			fattr->valid = 0;
+			status = nfs4_handle_exception(server,
+					rpc_call_sync(server->client, &msg, 0),
+					&exception);
+		} while (exception.retry);
+		if (status == 0)
 			continue;
 		if (status == -ENOENT) {
 			printk(KERN_NOTICE "NFS: mount path %s does not exist!\n", server->mnt_path);
@@ -621,14 +736,13 @@ static int nfs4_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
 	if (status == 0)
 		status = nfs4_do_fsinfo(server, fhandle, info);
 out:
-	return nfs4_map_errors(status);
+	return status;
 }
 
-static int nfs4_proc_getattr(struct inode *inode, struct nfs_fattr *fattr)
+static int _nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr)
 {
-	struct nfs_server *server = NFS_SERVER(inode);
 	struct nfs4_getattr_arg args = {
-		.fh = NFS_FH(inode),
+		.fh = fhandle,
 		.bitmask = server->attr_bitmask,
 	};
 	struct nfs4_getattr_res res = {
@@ -642,8 +756,19 @@ static int nfs4_proc_getattr(struct inode *inode, struct nfs_fattr *fattr)
 	};
 	
 	fattr->valid = 0;
+	return rpc_call_sync(server->client, &msg, 0);
+}
 
-	return nfs4_map_errors(rpc_call_sync(NFS_CLIENT(inode), &msg, 0));
+static int nfs4_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fattr *fattr)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(server,
+				_nfs4_proc_getattr(server, fhandle, fattr),
+				&exception);
+	} while (exception.retry);
+	return err;
 }
 
 /* 
@@ -705,7 +830,7 @@ nfs4_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr,
 	return status;
 }
 
-static int nfs4_proc_lookup(struct inode *dir, struct qstr *name,
+static int _nfs4_proc_lookup(struct inode *dir, struct qstr *name,
 		struct nfs_fh *fhandle, struct nfs_fattr *fattr)
 {
 	int		       status;
@@ -731,10 +856,22 @@ static int nfs4_proc_lookup(struct inode *dir, struct qstr *name,
 	dprintk("NFS call  lookup %s\n", name->name);
 	status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
 	dprintk("NFS reply lookup: %d\n", status);
-	return nfs4_map_errors(status);
+	return status;
 }
 
-static int nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
+static int nfs4_proc_lookup(struct inode *dir, struct qstr *name, struct nfs_fh *fhandle, struct nfs_fattr *fattr)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(dir),
+				_nfs4_proc_lookup(dir, name, fhandle, fattr),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
 {
 	struct nfs4_accessargs args = {
 		.fh = NFS_FH(inode),
@@ -775,7 +912,19 @@ static int nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
 		if (res.access & (NFS4_ACCESS_LOOKUP|NFS4_ACCESS_EXECUTE))
 			entry->mask |= MAY_EXEC;
 	}
-	return nfs4_map_errors(status);
+	return status;
+}
+
+static int nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(inode),
+				_nfs4_proc_access(inode, entry),
+				&exception);
+	} while (exception.retry);
+	return err;
 }
 
 /*
@@ -802,7 +951,7 @@ static int nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry)
  * Both of these changes to the XDR layer would in fact be quite
  * minor, but I decided to leave them for a subsequent patch.
  */
-static int nfs4_proc_readlink(struct inode *inode, struct page *page)
+static int _nfs4_proc_readlink(struct inode *inode, struct page *page)
 {
 	struct nfs4_readlink args = {
 		.fh       = NFS_FH(inode),
@@ -815,11 +964,22 @@ static int nfs4_proc_readlink(struct inode *inode, struct page *page)
 		.rpc_resp = NULL,
 	};
 
-	return nfs4_map_errors(rpc_call_sync(NFS_CLIENT(inode), &msg, 0));
+	return rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
 }
 
-static int
-nfs4_proc_read(struct nfs_read_data *rdata, struct file *filp)
+static int nfs4_proc_readlink(struct inode *inode, struct page *page)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(inode),
+				_nfs4_proc_readlink(inode, page),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_read(struct nfs_read_data *rdata)
 {
 	int flags = rdata->flags;
 	struct inode *inode = rdata->inode;
@@ -829,6 +989,7 @@ nfs4_proc_read(struct nfs_read_data *rdata, struct file *filp)
 		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_READ],
 		.rpc_argp	= &rdata->args,
 		.rpc_resp	= &rdata->res,
+		.rpc_cred	= rdata->cred,
 	};
 	unsigned long timestamp = jiffies;
 	int status;
@@ -836,29 +997,27 @@ nfs4_proc_read(struct nfs_read_data *rdata, struct file *filp)
 	dprintk("NFS call  read %d @ %Ld\n", rdata->args.count,
 			(long long) rdata->args.offset);
 
-	/*
-	 * Try first to use O_RDONLY, then O_RDWR stateid.
-	 */
-	if (filp) {
-		struct nfs4_state *state;
-		state = (struct nfs4_state *)filp->private_data;
-		rdata->args.state = state;
-		msg.rpc_cred = state->owner->so_cred;
-	} else {
-		rdata->args.state = NULL;
-		msg.rpc_cred = NFS_I(inode)->mm_cred;
-	}
-
 	fattr->valid = 0;
 	status = rpc_call_sync(server->client, &msg, flags);
 	if (!status)
 		renew_lease(server, timestamp);
 	dprintk("NFS reply read: %d\n", status);
-	return nfs4_map_errors(status);
+	return status;
 }
 
-static int
-nfs4_proc_write(struct nfs_write_data *wdata, struct file *filp)
+static int nfs4_proc_read(struct nfs_read_data *rdata)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(rdata->inode),
+				_nfs4_proc_read(rdata),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_write(struct nfs_write_data *wdata)
 {
 	int rpcflags = wdata->flags;
 	struct inode *inode = wdata->inode;
@@ -868,33 +1027,32 @@ nfs4_proc_write(struct nfs_write_data *wdata, struct file *filp)
 		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_WRITE],
 		.rpc_argp	= &wdata->args,
 		.rpc_resp	= &wdata->res,
+		.rpc_cred	= wdata->cred,
 	};
 	int status;
 
 	dprintk("NFS call  write %d @ %Ld\n", wdata->args.count,
 			(long long) wdata->args.offset);
 
-	/*
-	 * Try first to use O_WRONLY, then O_RDWR stateid.
-	 */
-	if (filp) {
-		struct nfs4_state *state;
-		state = (struct nfs4_state *)filp->private_data;
-		wdata->args.state = state;
-		msg.rpc_cred = state->owner->so_cred;
-	} else {
-		wdata->args.state = NULL;
-		msg.rpc_cred = NFS_I(inode)->mm_cred;
-	}
-
 	fattr->valid = 0;
 	status = rpc_call_sync(server->client, &msg, rpcflags);
 	dprintk("NFS reply write: %d\n", status);
-	return nfs4_map_errors(status);
+	return status;
 }
 
-static int
-nfs4_proc_commit(struct nfs_write_data *cdata, struct file *filp)
+static int nfs4_proc_write(struct nfs_write_data *wdata)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(wdata->inode),
+				_nfs4_proc_write(wdata),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_commit(struct nfs_write_data *cdata)
 {
 	struct inode *inode = cdata->inode;
 	struct nfs_fattr *fattr = cdata->res.fattr;
@@ -903,24 +1061,29 @@ nfs4_proc_commit(struct nfs_write_data *cdata, struct file *filp)
 		.rpc_proc	= &nfs4_procedures[NFSPROC4_CLNT_COMMIT],
 		.rpc_argp	= &cdata->args,
 		.rpc_resp	= &cdata->res,
+		.rpc_cred	= cdata->cred,
 	};
 	int status;
 
 	dprintk("NFS call  commit %d @ %Ld\n", cdata->args.count,
 			(long long) cdata->args.offset);
 
-	/*
-	 * Try first to use O_WRONLY, then O_RDWR stateid.
-	 */
-	if (filp)
-		msg.rpc_cred = ((struct nfs4_state *)filp->private_data)->owner->so_cred;
-	else
-		msg.rpc_cred = NFS_I(inode)->mm_cred;
-
 	fattr->valid = 0;
 	status = rpc_call_sync(server->client, &msg, 0);
 	dprintk("NFS reply commit: %d\n", status);
-	return nfs4_map_errors(status);
+	return status;
+}
+
+static int nfs4_proc_commit(struct nfs_write_data *cdata)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(cdata->inode),
+				_nfs4_proc_commit(cdata),
+				&exception);
+	} while (exception.retry);
+	return err;
 }
 
 /*
@@ -967,7 +1130,7 @@ nfs4_proc_create(struct inode *dir, struct qstr *name, struct iattr *sattr,
 	return inode;
 }
 
-static int nfs4_proc_remove(struct inode *dir, struct qstr *name)
+static int _nfs4_proc_remove(struct inode *dir, struct qstr *name)
 {
 	struct nfs4_remove_arg args = {
 		.fh = NFS_FH(dir),
@@ -984,7 +1147,19 @@ static int nfs4_proc_remove(struct inode *dir, struct qstr *name)
 	status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
 	if (status == 0)
 		update_changeattr(dir, &res);
-	return nfs4_map_errors(status);
+	return status;
+}
+
+static int nfs4_proc_remove(struct inode *dir, struct qstr *name)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(dir),
+				_nfs4_proc_remove(dir, name),
+				&exception);
+	} while (exception.retry);
+	return err;
 }
 
 struct unlink_desc {
@@ -1025,7 +1200,7 @@ static int nfs4_proc_unlink_done(struct dentry *dir, struct rpc_task *task)
 	return 0;
 }
 
-static int nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name,
+static int _nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name,
 		struct inode *new_dir, struct qstr *new_name)
 {
 	struct nfs4_rename_arg arg = {
@@ -1048,10 +1223,24 @@ static int nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name,
 		update_changeattr(old_dir, &res.old_cinfo);
 		update_changeattr(new_dir, &res.new_cinfo);
 	}
-	return nfs4_map_errors(status);
+	return status;
 }
 
-static int nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name)
+static int nfs4_proc_rename(struct inode *old_dir, struct qstr *old_name,
+		struct inode *new_dir, struct qstr *new_name)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(old_dir),
+				_nfs4_proc_rename(old_dir, old_name,
+					new_dir, new_name),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name)
 {
 	struct nfs4_link_arg arg = {
 		.fh     = NFS_FH(inode),
@@ -1070,10 +1259,22 @@ static int nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *n
 	if (!status)
 		update_changeattr(dir, &cinfo);
 
-	return nfs4_map_errors(status);
+	return status;
 }
 
-static int nfs4_proc_symlink(struct inode *dir, struct qstr *name,
+static int nfs4_proc_link(struct inode *inode, struct inode *dir, struct qstr *name)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(inode),
+				_nfs4_proc_link(inode, dir, name),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_symlink(struct inode *dir, struct qstr *name,
 		struct qstr *path, struct iattr *sattr, struct nfs_fh *fhandle,
 		struct nfs_fattr *fattr)
 {
@@ -1106,10 +1307,25 @@ static int nfs4_proc_symlink(struct inode *dir, struct qstr *name,
 	status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
 	if (!status)
 		update_changeattr(dir, &res.dir_cinfo);
-	return nfs4_map_errors(status);
+	return status;
 }
 
-static int nfs4_proc_mkdir(struct inode *dir, struct qstr *name,
+static int nfs4_proc_symlink(struct inode *dir, struct qstr *name,
+		struct qstr *path, struct iattr *sattr, struct nfs_fh *fhandle,
+		struct nfs_fattr *fattr)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(dir),
+				_nfs4_proc_symlink(dir, name, path, sattr,
+					fhandle, fattr),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_mkdir(struct inode *dir, struct qstr *name,
 		struct iattr *sattr, struct nfs_fh *fhandle,
 		struct nfs_fattr *fattr)
 {
@@ -1139,10 +1355,25 @@ static int nfs4_proc_mkdir(struct inode *dir, struct qstr *name,
 	status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
 	if (!status)
 		update_changeattr(dir, &res.dir_cinfo);
-	return nfs4_map_errors(status);
+	return status;
 }
 
-static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
+static int nfs4_proc_mkdir(struct inode *dir, struct qstr *name,
+		struct iattr *sattr, struct nfs_fh *fhandle,
+		struct nfs_fattr *fattr)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(dir),
+				_nfs4_proc_mkdir(dir, name, sattr,
+					fhandle, fattr),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
                   u64 cookie, struct page *page, unsigned int count, int plus)
 {
 	struct inode		*dir = dentry->d_inode;
@@ -1168,10 +1399,24 @@ static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
 	if (status == 0)
 		memcpy(NFS_COOKIEVERF(dir), res.verifier.data, NFS4_VERIFIER_SIZE);
 	unlock_kernel();
-	return nfs4_map_errors(status);
+	return status;
 }
 
-static int nfs4_proc_mknod(struct inode *dir, struct qstr *name,
+static int nfs4_proc_readdir(struct dentry *dentry, struct rpc_cred *cred,
+                  u64 cookie, struct page *page, unsigned int count, int plus)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(dentry->d_inode),
+				_nfs4_proc_readdir(dentry, cred, cookie,
+					page, count, plus),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_mknod(struct inode *dir, struct qstr *name,
 		struct iattr *sattr, dev_t rdev, struct nfs_fh *fh,
 		struct nfs_fattr *fattr)
 {
@@ -1218,10 +1463,25 @@ static int nfs4_proc_mknod(struct inode *dir, struct qstr *name,
 	status = rpc_call_sync(NFS_CLIENT(dir), &msg, 0);
 	if (!status)
 		update_changeattr(dir, &res.dir_cinfo);
-	return nfs4_map_errors(status);
+	return status;
+}
+
+static int nfs4_proc_mknod(struct inode *dir, struct qstr *name,
+		struct iattr *sattr, dev_t rdev, struct nfs_fh *fh,
+		struct nfs_fattr *fattr)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(dir),
+				_nfs4_proc_mknod(dir, name, sattr, rdev,
+					fh, fattr),
+				&exception);
+	} while (exception.retry);
+	return err;
 }
 
-static int nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle,
+static int _nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle,
 		 struct nfs_fsstat *fsstat)
 {
 	struct nfs4_statfs_arg args = {
@@ -1235,10 +1495,22 @@ static int nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle,
 	};
 
 	fsstat->fattr->valid = 0;
-	return nfs4_map_errors(rpc_call_sync(server->client, &msg, 0));
+	return rpc_call_sync(server->client, &msg, 0);
+}
+
+static int nfs4_proc_statfs(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsstat *fsstat)
+{
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = nfs4_handle_exception(server,
+				_nfs4_proc_statfs(server, fhandle, fsstat),
+				&exception);
+	} while (exception.retry);
+	return err;
 }
 
-static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle,
+static int _nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle,
 		struct nfs_fsinfo *fsinfo)
 {
 	struct nfs4_fsinfo_arg args = {
@@ -1251,16 +1523,29 @@ static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle,
 		.rpc_resp = fsinfo,
 	};
 
-	return nfs4_map_errors(rpc_call_sync(server->client, &msg, 0));
+	return rpc_call_sync(server->client, &msg, 0);
+}
+
+static int nfs4_do_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
+{
+	struct nfs4_exception exception = { };
+	int err;
+
+	do {
+		err = nfs4_handle_exception(server,
+				_nfs4_do_fsinfo(server, fhandle, fsinfo),
+				&exception);
+	} while (exception.retry);
+	return err;
 }
 
 static int nfs4_proc_fsinfo(struct nfs_server *server, struct nfs_fh *fhandle, struct nfs_fsinfo *fsinfo)
 {
 	fsinfo->fattr->valid = 0;
-	return nfs4_map_errors(nfs4_do_fsinfo(server, fhandle, fsinfo));
+	return nfs4_do_fsinfo(server, fhandle, fsinfo);
 }
 
-static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
+static int _nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
 		struct nfs_pathconf *pathconf)
 {
 	struct nfs4_pathconf_arg args = {
@@ -1280,7 +1565,21 @@ static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
 	}
 
 	pathconf->fattr->valid = 0;
-	return nfs4_map_errors(rpc_call_sync(server->client, &msg, 0));
+	return rpc_call_sync(server->client, &msg, 0);
+}
+
+static int nfs4_proc_pathconf(struct nfs_server *server, struct nfs_fh *fhandle,
+		struct nfs_pathconf *pathconf)
+{
+	struct nfs4_exception exception = { };
+	int err;
+
+	do {
+		err = nfs4_handle_exception(server,
+				_nfs4_proc_pathconf(server, fhandle, pathconf),
+				&exception);
+	} while (exception.retry);
+	return err;
 }
 
 static void
@@ -1471,8 +1770,10 @@ static int
 nfs4_proc_file_open(struct inode *inode, struct file *filp)
 {
 	struct dentry *dentry = filp->f_dentry;
-	struct nfs4_state *state;
+	struct nfs_open_context *ctx;
+	struct nfs4_state *state = NULL;
 	struct rpc_cred *cred;
+	int status = -ENOMEM;
 
 	dprintk("nfs4_proc_file_open: starting on (%.*s/%.*s)\n",
 	                       (int)dentry->d_parent->d_name.len,
@@ -1482,21 +1783,28 @@ nfs4_proc_file_open(struct inode *inode, struct file *filp)
 
 	/* Find our open stateid */
 	cred = rpcauth_lookupcred(NFS_SERVER(inode)->client->cl_auth, 0);
-	state = nfs4_find_state(inode, cred, filp->f_mode);
+	if (unlikely(cred == NULL))
+		return -ENOMEM;
+	ctx = alloc_nfs_open_context(dentry, cred);
 	put_rpccred(cred);
-	if (state == NULL) {
-		printk(KERN_WARNING "NFS: v4 raced in function %s\n", __FUNCTION__);
-		return -EIO; /* ERACE actually */
-	}
+	if (unlikely(ctx == NULL))
+		return -ENOMEM;
+	status = -EIO; /* ERACE actually */
+	state = nfs4_find_state(inode, cred, filp->f_mode);
+	if (unlikely(state == NULL))
+		goto no_state;
+	ctx->state = state;
 	nfs4_close_state(state, filp->f_mode);
-	if (filp->f_mode & FMODE_WRITE) {
-		lock_kernel();
-		nfs_set_mmcred(inode, state->owner->so_cred);
+	ctx->mode = filp->f_mode;
+	nfs_file_set_open_context(filp, ctx);
+	put_nfs_open_context(ctx);
+	if (filp->f_mode & FMODE_WRITE)
 		nfs_begin_data_update(inode);
-		unlock_kernel();
-	}
-	filp->private_data = state;
 	return 0;
+no_state:
+	printk(KERN_WARNING "NFS: v4 raced in function %s\n", __FUNCTION__);
+	put_nfs_open_context(ctx);
+	return status;
 }
 
 /*
@@ -1505,37 +1813,12 @@ nfs4_proc_file_open(struct inode *inode, struct file *filp)
 static int
 nfs4_proc_file_release(struct inode *inode, struct file *filp)
 {
-	struct nfs4_state *state = (struct nfs4_state *)filp->private_data;
-
-	if (state)
-		nfs4_close_state(state, filp->f_mode);
-	if (filp->f_mode & FMODE_WRITE) {
-		lock_kernel();
+	if (filp->f_mode & FMODE_WRITE)
 		nfs_end_data_update(inode);
-		unlock_kernel();
-	}
+	nfs_file_clear_open_context(filp);
 	return 0;
 }
 
-/*
- * Set up the nfspage struct with the right state info and credentials
- */
-static void
-nfs4_request_init(struct nfs_page *req, struct file *filp)
-{
-	struct nfs4_state *state;
-
-	if (!filp) {
-		req->wb_cred = get_rpccred(NFS_I(req->wb_inode)->mm_cred);
-		req->wb_state = NULL;
-		return;
-	}
-	state = (struct nfs4_state *)filp->private_data;
-	req->wb_state = state;
-	req->wb_cred = get_rpccred(state->owner->so_cred);
-	req->wb_lockowner = current->files;
-}
-
 static int
 nfs4_async_handle_error(struct rpc_task *task, struct nfs_server *server)
 {
@@ -1549,11 +1832,13 @@ nfs4_async_handle_error(struct rpc_task *task, struct nfs_server *server)
 		case -NFS4ERR_EXPIRED:
 			rpc_sleep_on(&clp->cl_rpcwaitq, task, NULL, NULL);
 			nfs4_schedule_state_recovery(clp);
+			if (test_bit(NFS4CLNT_OK, &clp->cl_state))
+				rpc_wake_up_task(task);
 			task->tk_status = 0;
 			return -EAGAIN;
 		case -NFS4ERR_GRACE:
 		case -NFS4ERR_DELAY:
-			rpc_delay(task, NFS4_POLL_RETRY_TIME);
+			rpc_delay(task, NFS4_POLL_RETRY_MAX);
 			task->tk_status = 0;
 			return -EAGAIN;
 		case -NFS4ERR_OLD_STATEID:
@@ -1564,12 +1849,11 @@ nfs4_async_handle_error(struct rpc_task *task, struct nfs_server *server)
 	return 0;
 }
 
-int
-nfs4_wait_clnt_recover(struct rpc_clnt *clnt, struct nfs4_client *clp)
+int nfs4_wait_clnt_recover(struct rpc_clnt *clnt, struct nfs4_client *clp)
 {
 	DEFINE_WAIT(wait);
 	sigset_t oldset;
-	int interruptible, res;
+	int interruptible, res = 0;
 
 	might_sleep();
 
@@ -1577,101 +1861,85 @@ nfs4_wait_clnt_recover(struct rpc_clnt *clnt, struct nfs4_client *clp)
 	interruptible = TASK_UNINTERRUPTIBLE;
 	if (clnt->cl_intr)
 		interruptible = TASK_INTERRUPTIBLE;
-	do {
-		res = 0;
-		prepare_to_wait(&clp->cl_waitq, &wait, interruptible);
-		nfs4_schedule_state_recovery(clp);
-		if (test_bit(NFS4CLNT_OK, &clp->cl_state) &&
-				!test_bit(NFS4CLNT_SETUP_STATE, &clp->cl_state))
-			break;
-		if (clnt->cl_intr && signalled()) {
-			res = -ERESTARTSYS;
-			break;
-		}
+	prepare_to_wait(&clp->cl_waitq, &wait, interruptible);
+	nfs4_schedule_state_recovery(clp);
+	if (clnt->cl_intr && signalled())
+		res = -ERESTARTSYS;
+	else if (!test_bit(NFS4CLNT_OK, &clp->cl_state))
 		schedule();
-	} while(!test_bit(NFS4CLNT_OK, &clp->cl_state));
 	finish_wait(&clp->cl_waitq, &wait);
 	rpc_clnt_sigunmask(clnt, &oldset);
 	return res;
 }
 
-static int
-nfs4_delay(struct rpc_clnt *clnt)
+static int nfs4_delay(struct rpc_clnt *clnt, long *timeout)
 {
 	sigset_t oldset;
 	int res = 0;
 
 	might_sleep();
 
+	if (*timeout <= 0)
+		*timeout = NFS4_POLL_RETRY_MIN;
+	if (*timeout > NFS4_POLL_RETRY_MAX)
+		*timeout = NFS4_POLL_RETRY_MAX;
 	rpc_clnt_sigmask(clnt, &oldset);
 	if (clnt->cl_intr) {
 		set_current_state(TASK_INTERRUPTIBLE);
-		schedule_timeout(NFS4_POLL_RETRY_TIME);
+		schedule_timeout(*timeout);
 		if (signalled())
 			res = -ERESTARTSYS;
 	} else {
 		set_current_state(TASK_UNINTERRUPTIBLE);
-		schedule_timeout(NFS4_POLL_RETRY_TIME);
+		schedule_timeout(*timeout);
 	}
 	rpc_clnt_sigunmask(clnt, &oldset);
+	*timeout <<= 1;
 	return res;
 }
 
 /* This is the error handling routine for processes that are allowed
  * to sleep.
  */
-int
-nfs4_handle_error(struct nfs_server *server, int errorcode)
+int nfs4_handle_exception(struct nfs_server *server, int errorcode, struct nfs4_exception *exception)
 {
 	struct nfs4_client *clp = server->nfs4_state;
 	int ret = errorcode;
 
+	exception->retry = 0;
 	switch(errorcode) {
+		case 0:
+			return 0;
 		case -NFS4ERR_STALE_CLIENTID:
 		case -NFS4ERR_STALE_STATEID:
 		case -NFS4ERR_EXPIRED:
 			ret = nfs4_wait_clnt_recover(server->client, clp);
+			if (ret == 0)
+				exception->retry = 1;
 			break;
 		case -NFS4ERR_GRACE:
 		case -NFS4ERR_DELAY:
-			ret = nfs4_delay(server->client);
+			ret = nfs4_delay(server->client, &exception->timeout);
+			if (ret == 0)
+				exception->retry = 1;
 			break;
 		case -NFS4ERR_OLD_STATEID:
-			ret = 0;
+			if (ret == 0)
+				exception->retry = 1;
 	}
 	/* We failed to handle the error */
 	return nfs4_map_errors(ret);
 }
 
-
-static int
-nfs4_request_compatible(struct nfs_page *req, struct file *filp, struct page *page)
-{
-	struct nfs4_state *state = NULL;
-	struct rpc_cred *cred = NULL;
-
-	if (req->wb_file != filp)
-		return 0;
-	if (req->wb_page != page)
-		return 0;
-	state = (struct nfs4_state *)filp->private_data;
-	if (req->wb_state != state)
-		return 0;
-	if (req->wb_lockowner != current->files)
-		return 0;
-	cred = state->owner->so_cred;
-	if (req->wb_cred != cred)
-		return 0;
-	return 1;
-}
-
-int
-nfs4_proc_setclientid(struct nfs4_client *clp,
-		u32 program, unsigned short port)
+int nfs4_proc_setclientid(struct nfs4_client *clp, u32 program, unsigned short port)
 {
-	u32 *p;
-	struct nfs4_setclientid setclientid;
-	struct timespec tv;
+	static nfs4_verifier sc_verifier;
+	static int initialized;
+	
+	struct nfs4_setclientid setclientid = {
+		.sc_verifier = &sc_verifier,
+		.sc_prog = program,
+	};
 	struct rpc_message msg = {
 		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SETCLIENTID],
 		.rpc_argp = &setclientid,
@@ -1679,15 +1947,24 @@ nfs4_proc_setclientid(struct nfs4_client *clp,
 		.rpc_cred = clp->cl_cred,
 	};
 
-	tv = CURRENT_TIME;
-	p = (u32*)setclientid.sc_verifier.data;
-	*p++ = (u32)tv.tv_sec;
-	*p = (u32)tv.tv_nsec;
-	setclientid.sc_name = clp->cl_ipaddr;
-	sprintf(setclientid.sc_netid, "tcp");
-	sprintf(setclientid.sc_uaddr, "%s.%d.%d", clp->cl_ipaddr, port >> 8, port & 255);
-	setclientid.sc_prog = htonl(program);
-	setclientid.sc_cb_ident = 0;
+	if (!initialized) {
+		struct timespec boot_time;
+		u32 *p;
+
+		initialized = 1;
+		boot_time = CURRENT_TIME;
+		p = (u32*)sc_verifier.data;
+		*p++ = htonl((u32)boot_time.tv_sec);
+		*p = htonl((u32)boot_time.tv_nsec);
+	}
+	setclientid.sc_name_len = scnprintf(setclientid.sc_name,
+			sizeof(setclientid.sc_name), "%s/%u.%u.%u.%u",
+			clp->cl_ipaddr, NIPQUAD(clp->cl_addr.s_addr));
+	setclientid.sc_netid_len = scnprintf(setclientid.sc_netid,
+			sizeof(setclientid.sc_netid), "tcp");
+	setclientid.sc_uaddr_len = scnprintf(setclientid.sc_uaddr,
+			sizeof(setclientid.sc_uaddr), "%s.%d.%d",
+			clp->cl_ipaddr, port >> 8, port & 255);
 
 	return rpc_call_sync(clp->cl_rpcclient, &msg, 0);
 }
@@ -1757,8 +2034,7 @@ nfs4_lck_length(struct file_lock *request)
 	return request->fl_end - request->fl_start + 1;
 }
 
-int
-nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request)
+static int _nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 {
 	struct inode *inode = state->inode;
 	struct nfs_server *server = NFS_SERVER(inode);
@@ -1782,6 +2058,7 @@ nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 	struct nfs4_lock_state *lsp;
 	int status;
 
+	down_read(&clp->cl_sem);
 	nlo.clientid = clp->cl_clientid;
 	down(&state->lock_sema);
 	lsp = nfs4_find_lock_state(state, request->fl_owner);
@@ -1815,14 +2092,28 @@ nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 	if (lsp)
 		nfs4_put_lock_state(lsp);
 	up(&state->lock_sema);
-	return nfs4_map_errors(status);
+	up_read(&clp->cl_sem);
+	return status;
 }
 
-int
-nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
+static int nfs4_proc_getlk(struct nfs4_state *state, int cmd, struct file_lock *request)
+{
+	struct nfs4_exception exception = { };
+	int err;
+
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(state->inode),
+				_nfs4_proc_getlk(state, cmd, request),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
 {
 	struct inode *inode = state->inode;
 	struct nfs_server *server = NFS_SERVER(inode);
+	struct nfs4_client *clp = server->nfs4_state;
 	struct nfs_lockargs arg = {
 		.fh = NFS_FH(inode),
 		.type = nfs4_lck_type(cmd, request),
@@ -1842,31 +2133,48 @@ nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
 	struct nfs_locku_opargs luargs;
 	int status = 0;
 			
+	down_read(&clp->cl_sem);
 	down(&state->lock_sema);
 	lsp = nfs4_find_lock_state(state, request->fl_owner);
 	if (!lsp)
 		goto out;
-	luargs.seqid = lsp->ls_seqid;
-	memcpy(&luargs.stateid, &lsp->ls_stateid, sizeof(luargs.stateid));
-	arg.u.locku = &luargs;
-	status = rpc_call_sync(server->client, &msg, 0);
-	nfs4_increment_lock_seqid(status, lsp);
+	/* We might have lost the locks! */
+	if ((lsp->ls_flags & NFS_LOCK_INITIALIZED) != 0) {
+		luargs.seqid = lsp->ls_seqid;
+		memcpy(&luargs.stateid, &lsp->ls_stateid, sizeof(luargs.stateid));
+		arg.u.locku = &luargs;
+		status = rpc_call_sync(server->client, &msg, 0);
+		nfs4_increment_lock_seqid(status, lsp);
+	}
 
 	if (status == 0) {
 		memcpy(&lsp->ls_stateid,  &res.u.stateid, 
 				sizeof(lsp->ls_stateid));
-		nfs4_notify_unlck(inode, request, lsp);
+		nfs4_notify_unlck(state, request, lsp);
 	}
 	nfs4_put_lock_state(lsp);
 out:
 	up(&state->lock_sema);
 	if (status == 0)
 		posix_lock_file(request->fl_file, request);
-	return nfs4_map_errors(status);
+	up_read(&clp->cl_sem);
+	return status;
 }
 
-static int
-nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
+static int nfs4_proc_unlck(struct nfs4_state *state, int cmd, struct file_lock *request)
+{
+	struct nfs4_exception exception = { };
+	int err;
+
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(state->inode),
+				_nfs4_proc_unlck(state, cmd, request),
+				&exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_do_setlk(struct nfs4_state *state, int cmd, struct file_lock *request, int reclaim)
 {
 	struct inode *inode = state->inode;
 	struct nfs_server *server = NFS_SERVER(inode);
@@ -1887,23 +2195,22 @@ nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 		.rpc_cred	= state->owner->so_cred,
 	};
 	struct nfs_lock_opargs largs = {
+		.reclaim = reclaim,
 		.new_lock_owner = 0,
 	};
 	int status;
 
-	down(&state->lock_sema);
-	lsp = nfs4_find_lock_state(state, request->fl_owner);
-	if (lsp == NULL) {
+	lsp = nfs4_get_lock_state(state, request->fl_owner);
+	if (lsp == NULL)
+		return -ENOMEM;
+	if (!(lsp->ls_flags & NFS_LOCK_INITIALIZED)) {
 		struct nfs4_state_owner *owner = state->owner;
 		struct nfs_open_to_lock otl = {
 			.lock_owner = {
 				.clientid = server->nfs4_state->cl_clientid,
 			},
 		};
-		status = -ENOMEM;
-		lsp = nfs4_alloc_lock_state(state, request->fl_owner);
-		if (!lsp)
-			goto out;
+
 		otl.lock_seqid = lsp->ls_seqid;
 		otl.lock_owner.id = lsp->ls_id;
 		memcpy(&otl.open_stateid, &state->stateid, sizeof(otl.open_stateid));
@@ -1932,11 +2239,28 @@ nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 	/* save the returned stateid. */
 	if (status == 0) {
 		memcpy(&lsp->ls_stateid, &res.u.stateid, sizeof(nfs4_stateid));
-		nfs4_notify_setlk(inode, request, lsp);
+		lsp->ls_flags |= NFS_LOCK_INITIALIZED;
+		if (!reclaim)
+			nfs4_notify_setlk(state, request, lsp);
 	} else if (status == -NFS4ERR_DENIED)
 		status = -EAGAIN;
 	nfs4_put_lock_state(lsp);
-out:
+	return status;
+}
+
+int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request)
+{
+	return _nfs4_do_setlk(state, F_SETLK64, request, 1);
+}
+
+static int _nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
+{
+	struct nfs4_client *clp = state->owner->so_client;
+	int status;
+
+	down_read(&clp->cl_sem);
+	down(&state->lock_sema);
+	status = _nfs4_do_setlk(state, cmd, request, 0);
 	up(&state->lock_sema);
 	if (status == 0) {
 		/* Note: we always want to sleep here! */
@@ -1944,19 +2268,34 @@ nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 		if (posix_lock_file_wait(request->fl_file, request) < 0)
 			printk(KERN_WARNING "%s: VFS is out of sync with lock manager!\n", __FUNCTION__);
 	}
-	return nfs4_map_errors(status);
+	up_read(&clp->cl_sem);
+	return status;
+}
+
+static int nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
+{
+	struct nfs4_exception exception = { };
+	int err;
+
+	do {
+		err = nfs4_handle_exception(NFS_SERVER(state->inode),
+				_nfs4_proc_setlk(state, cmd, request),
+				&exception);
+	} while (exception.retry);
+	return err;
 }
 
 static int
 nfs4_proc_lock(struct file *filp, int cmd, struct file_lock *request)
 {
+	struct nfs_open_context *ctx;
 	struct nfs4_state *state;
 	unsigned long timeout = NFS4_LOCK_MINTIMEOUT;
 	int status;
 
 	/* verify open state */
-	state = (struct nfs4_state *)filp->private_data;
-	BUG_ON(!state);
+	ctx = (struct nfs_open_context *)filp->private_data;
+	state = ctx->state;
 
 	if (request->fl_start < 0 || request->fl_end < 0)
 		return -EINVAL;
@@ -2016,8 +2355,6 @@ struct nfs_rpc_ops	nfs_v4_clientops = {
 	.commit_setup	= nfs4_proc_commit_setup,
 	.file_open      = nfs4_proc_file_open,
 	.file_release   = nfs4_proc_file_release,
-	.request_init	= nfs4_request_init,
-	.request_compatible = nfs4_request_compatible,
 	.lock		= nfs4_proc_lock,
 };
 

fs/nfs/nfs4state.c

@@ -40,6 +40,7 @@
 
 #include <linux/config.h>
 #include <linux/slab.h>
+#include <linux/smp_lock.h>
 #include <linux/nfs_fs.h>
 #include <linux/nfs_idmap.h>
 #include <linux/workqueue.h>
@@ -106,7 +107,7 @@ nfs4_alloc_client(struct in_addr *addr)
 		INIT_LIST_HEAD(&clp->cl_superblocks);
 		init_waitqueue_head(&clp->cl_waitq);
 		rpc_init_wait_queue(&clp->cl_rpcwaitq, "NFS4 client");
-		clp->cl_state = 1 << NFS4CLNT_NEW;
+		clp->cl_state = 1 << NFS4CLNT_OK;
 	}
 	return clp;
 }
@@ -169,6 +170,16 @@ nfs4_put_client(struct nfs4_client *clp)
 	nfs4_free_client(clp);
 }
 
+int nfs4_init_client(struct nfs4_client *clp)
+{
+	int status = nfs4_proc_setclientid(clp, 0, 0);
+	if (status == 0)
+		status = nfs4_proc_setclientid_confirm(clp);
+	if (status == 0)
+		nfs4_schedule_state_renewal(clp);
+	return status;
+}
+
 u32
 nfs4_alloc_lockowner_id(struct nfs4_client *clp)
 {
@@ -185,7 +196,6 @@ nfs4_client_grab_unused(struct nfs4_client *clp, struct rpc_cred *cred)
 		atomic_inc(&sp->so_count);
 		sp->so_cred = cred;
 		list_move(&sp->so_list, &clp->cl_state_owners);
-		sp->so_generation = clp->cl_generation;
 		clp->cl_nunused--;
 	}
 	return sp;
@@ -237,8 +247,11 @@ nfs4_unhash_state_owner(struct nfs4_state_owner *sp)
 	spin_unlock(&clp->cl_lock);
 }
 
-struct nfs4_state_owner *
-nfs4_get_state_owner(struct nfs_server *server, struct rpc_cred *cred)
+/*
+ * Note: must be called with clp->cl_sem held in order to prevent races
+ *       with reboot recovery!
+ */
+struct nfs4_state_owner *nfs4_get_state_owner(struct nfs_server *server, struct rpc_cred *cred)
 {
 	struct nfs4_client *clp = server->nfs4_state;
 	struct nfs4_state_owner *sp, *new;
@@ -254,23 +267,23 @@ nfs4_get_state_owner(struct nfs_server *server, struct rpc_cred *cred)
 		new->so_client = clp;
 		new->so_id = nfs4_alloc_lockowner_id(clp);
 		new->so_cred = cred;
-		new->so_generation = clp->cl_generation;
 		sp = new;
 		new = NULL;
 	}
 	spin_unlock(&clp->cl_lock);
 	if (new)
 		kfree(new);
-	if (sp) {
-		if (!test_bit(NFS4CLNT_OK, &clp->cl_state))
-			nfs4_wait_clnt_recover(server->client, clp);
-	} else
-		put_rpccred(cred);
-	return sp;
+	if (sp != NULL)
+		return sp;
+	put_rpccred(cred);
+	return NULL;
 }
 
-void
-nfs4_put_state_owner(struct nfs4_state_owner *sp)
+/*
+ * Must be called with clp->cl_sem held in order to avoid races
+ * with state recovery...
+ */
+void nfs4_put_state_owner(struct nfs4_state_owner *sp)
 {
 	struct nfs4_client *clp = sp->so_client;
 	struct rpc_cred *cred = sp->so_cred;
@@ -330,8 +343,6 @@ __nfs4_find_state(struct inode *inode, struct rpc_cred *cred, mode_t mode)
 			continue;
 		if ((state->state & mode) != mode)
 			continue;
-		/* Add the state to the head of the inode's list */
-		list_move(&state->inode_states, &nfsi->open_states);
 		atomic_inc(&state->count);
 		if (mode & FMODE_READ)
 			state->nreaders++;
@@ -353,8 +364,6 @@ __nfs4_find_state_byowner(struct inode *inode, struct nfs4_state_owner *owner)
 		if (state->nreaders == 0 && state->nwriters == 0)
 			continue;
 		if (state->owner == owner) {
-			/* Add the state to the head of the inode's list */
-			list_move(&state->inode_states, &nfsi->open_states);
 			atomic_inc(&state->count);
 			return state;
 		}
@@ -411,51 +420,40 @@ nfs4_get_open_state(struct inode *inode, struct nfs4_state_owner *owner)
 	return state;
 }
 
-static void
-__nfs4_put_open_state(struct nfs4_state *state)
+/*
+ * Beware! Caller must be holding exactly one
+ * reference to clp->cl_sem and owner->so_sema!
+ */
+void nfs4_put_open_state(struct nfs4_state *state)
 {
 	struct inode *inode = state->inode;
 	struct nfs4_state_owner *owner = state->owner;
-	int status = 0;
 
-	if (!atomic_dec_and_lock(&state->count, &inode->i_lock)) {
-		up(&owner->so_sema);
+	if (!atomic_dec_and_lock(&state->count, &inode->i_lock))
 		return;
-	}
 	if (!list_empty(&state->inode_states))
 		list_del(&state->inode_states);
 	spin_unlock(&inode->i_lock);
 	list_del(&state->open_states);
-	if (state->state != 0) {
-		do {
-			status = nfs4_do_close(inode, state);
-			if (!status)
-				break;
-			up(&owner->so_sema);
-			status = nfs4_handle_error(NFS_SERVER(inode), status);
-			down(&owner->so_sema);
-		} while (!status);
-	}
-	up(&owner->so_sema);
+	BUG_ON (state->state != 0);
 	nfs4_free_open_state(state);
 	nfs4_put_state_owner(owner);
 }
 
-void
-nfs4_put_open_state(struct nfs4_state *state)
-{
-	down(&state->owner->so_sema);
-	__nfs4_put_open_state(state);
-}
-
-void
-nfs4_close_state(struct nfs4_state *state, mode_t mode)
+/*
+ * Beware! Caller must be holding no references to clp->cl_sem!
+ * of owner->so_sema!
+ */
+void nfs4_close_state(struct nfs4_state *state, mode_t mode)
 {
 	struct inode *inode = state->inode;
 	struct nfs4_state_owner *owner = state->owner;
+	struct nfs4_client *clp = owner->so_client;
 	int newstate;
 	int status = 0;
 
+	atomic_inc(&owner->so_count);
+	down_read(&clp->cl_sem);
 	down(&owner->so_sema);
 	/* Protect against nfs4_find_state() */
 	spin_lock(&inode->i_lock);
@@ -466,29 +464,24 @@ nfs4_close_state(struct nfs4_state *state, mode_t mode)
 	if (state->nwriters == 0 && state->nreaders == 0)
 		list_del_init(&state->inode_states);
 	spin_unlock(&inode->i_lock);
-	do {
-		newstate = 0;
-		if (state->state == 0)
-			break;
+	newstate = 0;
+	if (state->state != 0) {
 		if (state->nreaders)
 			newstate |= FMODE_READ;
 		if (state->nwriters)
 			newstate |= FMODE_WRITE;
 		if (state->state == newstate)
-			break;
+			goto out;
 		if (newstate != 0)
 			status = nfs4_do_downgrade(inode, state, newstate);
 		else
 			status = nfs4_do_close(inode, state);
-		if (!status) {
-			state->state = newstate;
-			break;
-		}
-		up(&owner->so_sema);
-		status = nfs4_handle_error(NFS_SERVER(inode), status);
-		down(&owner->so_sema);
-	} while (!status);
-	__nfs4_put_open_state(state);
+	}
+out:
+	nfs4_put_open_state(state);
+	up(&owner->so_sema);
+	nfs4_put_state_owner(owner);
+	up_read(&clp->cl_sem);
 }
 
 /*
@@ -524,8 +517,7 @@ nfs4_find_lock_state(struct nfs4_state *state, fl_owner_t fl_owner)
  *
  * The caller must be holding state->lock_sema
  */
-struct nfs4_lock_state *
-nfs4_alloc_lock_state(struct nfs4_state *state, fl_owner_t fl_owner)
+static struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, fl_owner_t fl_owner)
 {
 	struct nfs4_lock_state *lsp;
 	struct nfs4_client *clp = state->owner->so_client;
@@ -533,12 +525,12 @@ nfs4_alloc_lock_state(struct nfs4_state *state, fl_owner_t fl_owner)
 	lsp = kmalloc(sizeof(*lsp), GFP_KERNEL);
 	if (lsp == NULL)
 		return NULL;
+	lsp->ls_flags = 0;
 	lsp->ls_seqid = 0;	/* arbitrary */
 	lsp->ls_id = -1; 
 	memset(lsp->ls_stateid.data, 0, sizeof(lsp->ls_stateid.data));
 	atomic_set(&lsp->ls_count, 1);
 	lsp->ls_owner = fl_owner;
-	lsp->ls_parent = state;
 	INIT_LIST_HEAD(&lsp->ls_locks);
 	spin_lock(&clp->cl_lock);
 	lsp->ls_id = nfs4_alloc_lockowner_id(clp);
@@ -546,6 +538,22 @@ nfs4_alloc_lock_state(struct nfs4_state *state, fl_owner_t fl_owner)
 	return lsp;
 }
 
+/*
+ * Return a compatible lock_state. If no initialized lock_state structure
+ * exists, return an uninitialized one.
+ *
+ * The caller must be holding state->lock_sema and clp->cl_sem
+ */
+struct nfs4_lock_state *nfs4_get_lock_state(struct nfs4_state *state, fl_owner_t owner)
+{
+	struct nfs4_lock_state * lsp;
+	
+	lsp = nfs4_find_lock_state(state, owner);
+	if (lsp == NULL)
+		lsp = nfs4_alloc_lock_state(state, owner);
+	return lsp;
+}
+
 /*
  * Byte-range lock aware utility to initialize the stateid of read/write
  * requests.
@@ -567,10 +575,9 @@ nfs4_copy_stateid(nfs4_stateid *dst, struct nfs4_state *state, fl_owner_t fl_own
 }
 
 /*
-* Called with state->lock_sema held.
+* Called with state->lock_sema and clp->cl_sem held.
 */
-void
-nfs4_increment_lock_seqid(int status, struct nfs4_lock_state *lsp)
+void nfs4_increment_lock_seqid(int status, struct nfs4_lock_state *lsp)
 {
 	if (status == NFS_OK || seqid_mutating_err(-status))
 		lsp->ls_seqid++;
@@ -597,13 +604,11 @@ nfs4_check_unlock(struct file_lock *fl, struct file_lock *request)
 /*
  * Post an initialized lock_state on the state->lock_states list.
  */
-void
-nfs4_notify_setlk(struct inode *inode, struct file_lock *request, struct nfs4_lock_state *lsp)
+void nfs4_notify_setlk(struct nfs4_state *state, struct file_lock *request, struct nfs4_lock_state *lsp)
 {
-	struct nfs4_state *state = lsp->ls_parent;
-
 	if (!list_empty(&lsp->ls_locks))
 		return;
+	atomic_inc(&lsp->ls_count);
 	write_lock(&state->state_lock);
 	list_add(&lsp->ls_locks, &state->lock_states);
 	set_bit(LK_STATE_IN_USE, &state->flags);
@@ -620,9 +625,9 @@ nfs4_notify_setlk(struct inode *inode, struct file_lock *request, struct nfs4_lo
  *
  */
 void
-nfs4_notify_unlck(struct inode *inode, struct file_lock *request, struct nfs4_lock_state *lsp)
+nfs4_notify_unlck(struct nfs4_state *state, struct file_lock *request, struct nfs4_lock_state *lsp)
 {
-	struct nfs4_state *state = lsp->ls_parent;
+	struct inode *inode = state->inode;
 	struct file_lock *fl;
 
 	for (fl = inode->i_flock; fl != NULL; fl = fl->fl_next) {
@@ -640,6 +645,7 @@ nfs4_notify_unlck(struct inode *inode, struct file_lock *request, struct nfs4_lo
 	if (list_empty(&state->lock_states))
 		clear_bit(LK_STATE_IN_USE, &state->flags);
 	write_unlock(&state->state_lock);
+	nfs4_put_lock_state(lsp);
 }
 
 /*
@@ -651,20 +657,18 @@ nfs4_put_lock_state(struct nfs4_lock_state *lsp)
 {
 	if (!atomic_dec_and_test(&lsp->ls_count))
 		return;
-	if (!list_empty(&lsp->ls_locks))
-		return;
+	BUG_ON (!list_empty(&lsp->ls_locks));
 	kfree(lsp);
 }
 
 /*
-* Called with sp->so_sema held.
+* Called with sp->so_sema and clp->cl_sem held.
 *
 * Increment the seqid if the OPEN/OPEN_DOWNGRADE/CLOSE succeeded, or
 * failed with a seqid incrementing error -
 * see comments nfs_fs.h:seqid_mutating_error()
 */
-void
-nfs4_increment_seqid(int status, struct nfs4_state_owner *sp)
+void nfs4_increment_seqid(int status, struct nfs4_state_owner *sp)
 {
 	if (status == NFS_OK || seqid_mutating_err(-status))
 		sp->so_seqid++;
@@ -693,21 +697,14 @@ nfs4_recover_state(void *data)
 
 	init_completion(&args.complete);
 
-	down_read(&clp->cl_sem);
-	if (test_and_set_bit(NFS4CLNT_SETUP_STATE, &clp->cl_state))
-		goto out_failed;
 	if (kernel_thread(reclaimer, &args, CLONE_KERNEL) < 0)
 		goto out_failed_clear;
 	wait_for_completion(&args.complete);
 	return;
 out_failed_clear:
-	smp_mb__before_clear_bit();
-	clear_bit(NFS4CLNT_SETUP_STATE, &clp->cl_state);
-	smp_mb__after_clear_bit();
+	set_bit(NFS4CLNT_OK, &clp->cl_state);
 	wake_up_all(&clp->cl_waitq);
 	rpc_wake_up(&clp->cl_rpcwaitq);
-out_failed:
-	up_read(&clp->cl_sem);
 }
 
 /*
@@ -718,24 +715,66 @@ nfs4_schedule_state_recovery(struct nfs4_client *clp)
 {
 	if (!clp)
 		return;
-	smp_mb__before_clear_bit();
-	clear_bit(NFS4CLNT_OK, &clp->cl_state);
-	smp_mb__after_clear_bit();
-	schedule_work(&clp->cl_recoverd);
+	if (test_and_clear_bit(NFS4CLNT_OK, &clp->cl_state))
+		schedule_work(&clp->cl_recoverd);
 }
 
-static int
-nfs4_reclaim_open_state(struct nfs4_state_owner *sp)
+static int nfs4_reclaim_locks(struct nfs4_state *state)
+{
+	struct inode *inode = state->inode;
+	struct file_lock *fl;
+	int status = 0;
+
+	for (fl = inode->i_flock; fl != 0; fl = fl->fl_next) {
+		if (!(fl->fl_flags & FL_POSIX))
+			continue;
+		if (((struct nfs_open_context *)fl->fl_file->private_data)->state != state)
+			continue;
+		status = nfs4_lock_reclaim(state, fl);
+		if (status >= 0)
+			continue;
+		switch (status) {
+			default:
+				printk(KERN_ERR "%s: unhandled error %d. Zeroing state\n",
+						__FUNCTION__, status);
+			case -NFS4ERR_EXPIRED:
+			case -NFS4ERR_NO_GRACE:
+			case -NFS4ERR_RECLAIM_BAD:
+			case -NFS4ERR_RECLAIM_CONFLICT:
+				/* kill_proc(fl->fl_owner, SIGLOST, 1); */
+				break;
+			case -NFS4ERR_STALE_CLIENTID:
+				goto out_err;
+		}
+	}
+	return 0;
+out_err:
+	return status;
+}
+
+static int nfs4_reclaim_open_state(struct nfs4_state_owner *sp)
 {
 	struct nfs4_state *state;
+	struct nfs4_lock_state *lock;
 	int status = 0;
 
 	list_for_each_entry(state, &sp->so_states, open_states) {
 		if (state->state == 0)
 			continue;
 		status = nfs4_open_reclaim(sp, state);
-		if (status >= 0)
+		list_for_each_entry(lock, &state->lock_states, ls_locks)
+			lock->ls_flags &= ~NFS_LOCK_INITIALIZED;
+		if (status >= 0) {
+			status = nfs4_reclaim_locks(state);
+			if (status < 0)
+				goto out_err;
+			list_for_each_entry(lock, &state->lock_states, ls_locks) {
+				if (!(lock->ls_flags & NFS_LOCK_INITIALIZED))
+					printk("%s: Lock reclaim failed!\n",
+							__FUNCTION__);
+			}
 			continue;
+		}
 		switch (status) {
 			default:
 				printk(KERN_ERR "%s: unhandled error %d. Zeroing state\n",
@@ -762,75 +801,52 @@ nfs4_reclaim_open_state(struct nfs4_state_owner *sp)
 	return status;
 }
 
-static int
-reclaimer(void *ptr)
+static int reclaimer(void *ptr)
 {
 	struct reclaimer_args *args = (struct reclaimer_args *)ptr;
 	struct nfs4_client *clp = args->clp;
 	struct nfs4_state_owner *sp;
-	int generation;
 	int status;
 
 	daemonize("%u.%u.%u.%u-reclaim", NIPQUAD(clp->cl_addr));
 	allow_signal(SIGKILL);
 
+	atomic_inc(&clp->cl_count);
 	complete(&args->complete);
 
+	/* Ensure exclusive access to NFSv4 state */
+	lock_kernel();
+	down_write(&clp->cl_sem);
 	/* Are there any NFS mounts out there? */
 	if (list_empty(&clp->cl_superblocks))
 		goto out;
-	if (!test_bit(NFS4CLNT_NEW, &clp->cl_state)) {
-		status = nfs4_proc_renew(clp);
-		if (status == 0) {
-			set_bit(NFS4CLNT_OK, &clp->cl_state);
-			goto out;
-		}
-	}
-	status = nfs4_proc_setclientid(clp, 0, 0);
-	if (status)
-		goto out_error;
-	status = nfs4_proc_setclientid_confirm(clp);
+restart_loop:
+	status = nfs4_proc_renew(clp);
+	if (status == 0)
+		goto out;
+	status = nfs4_init_client(clp);
 	if (status)
 		goto out_error;
-	generation = ++(clp->cl_generation);
-	clear_bit(NFS4CLNT_NEW, &clp->cl_state);
-	set_bit(NFS4CLNT_OK, &clp->cl_state);
-	up_read(&clp->cl_sem);
-	nfs4_schedule_state_renewal(clp);
-restart_loop:
-	spin_lock(&clp->cl_lock);
+	/* Note: list is protected by exclusive lock on cl->cl_sem */
 	list_for_each_entry(sp, &clp->cl_state_owners, so_list) {
-		if (sp->so_generation - generation >= 0)
-			continue;
-		atomic_inc(&sp->so_count);
-		spin_unlock(&clp->cl_lock);
-		down(&sp->so_sema);
-		if (sp->so_generation - generation < 0) {
-			smp_rmb();
-			sp->so_generation = clp->cl_generation;
-			status = nfs4_reclaim_open_state(sp);
-		}
-		up(&sp->so_sema);
-		nfs4_put_state_owner(sp);
+		status = nfs4_reclaim_open_state(sp);
 		if (status < 0) {
 			if (status == -NFS4ERR_STALE_CLIENTID)
-				nfs4_schedule_state_recovery(clp);
-			goto out;
+				goto restart_loop;
+			goto out_error;
 		}
-		goto restart_loop;
 	}
-	spin_unlock(&clp->cl_lock);
 out:
-	smp_mb__before_clear_bit();
-	clear_bit(NFS4CLNT_SETUP_STATE, &clp->cl_state);
-	smp_mb__after_clear_bit();
+	set_bit(NFS4CLNT_OK, &clp->cl_state);
+	up_write(&clp->cl_sem);
+	unlock_kernel();
 	wake_up_all(&clp->cl_waitq);
 	rpc_wake_up(&clp->cl_rpcwaitq);
+	nfs4_put_client(clp);
 	return 0;
 out_error:
-	printk(KERN_WARNING "Error: state recovery failed on NFSv4 server %u.%u.%u.%u\n",
-				NIPQUAD(clp->cl_addr.s_addr));
-	up_read(&clp->cl_sem);
+	printk(KERN_WARNING "Error: state recovery failed on NFSv4 server %u.%u.%u.%u with error %d\n",
+				NIPQUAD(clp->cl_addr.s_addr), -status);
 	goto out;
 }
 

fs/nfs/nfs4xdr.c

@@ -404,6 +404,15 @@ struct compound_hdr {
 	BUG_ON(!p);						\
 } while (0)
 
+static void encode_string(struct xdr_stream *xdr, unsigned int len, const char *str)
+{
+	uint32_t *p;
+
+	p = xdr_reserve_space(xdr, 4 + len);
+	BUG_ON(p == NULL);
+	xdr_encode_opaque(p, str, len);
+}
+
 static int encode_compound_hdr(struct xdr_stream *xdr, struct compound_hdr *hdr)
 {
 	uint32_t *p;
@@ -903,15 +912,15 @@ static int encode_putrootfh(struct xdr_stream *xdr)
         return 0;
 }
 
-static void encode_stateid(struct xdr_stream *xdr, struct nfs4_state *state, fl_owner_t lockowner)
+static void encode_stateid(struct xdr_stream *xdr, const struct nfs_open_context *ctx)
 {
 	extern nfs4_stateid zero_stateid;
 	nfs4_stateid stateid;
 	uint32_t *p;
 
 	RESERVE_SPACE(16);
-	if (state != NULL) {
-		nfs4_copy_stateid(&stateid, state, lockowner);
+	if (ctx->state != NULL) {
+		nfs4_copy_stateid(&stateid, ctx->state, ctx->lockowner);
 		WRITEMEM(stateid.data, sizeof(stateid.data));
 	} else
 		WRITEMEM(zero_stateid.data, sizeof(zero_stateid.data));
@@ -924,7 +933,7 @@ static int encode_read(struct xdr_stream *xdr, const struct nfs_readargs *args)
 	RESERVE_SPACE(4);
 	WRITE32(OP_READ);
 
-	encode_stateid(xdr, args->state, args->lockowner);
+	encode_stateid(xdr, args->context);
 
 	RESERVE_SPACE(12);
 	WRITE64(args->offset);
@@ -1047,26 +1056,18 @@ static int encode_setattr(struct xdr_stream *xdr, const struct nfs_setattrargs *
 
 static int encode_setclientid(struct xdr_stream *xdr, const struct nfs4_setclientid *setclientid)
 {
-	uint32_t total_len;
-	uint32_t len1, len2, len3;
 	uint32_t *p;
 
-	len1 = strlen(setclientid->sc_name);
-	len2 = strlen(setclientid->sc_netid);
-	len3 = strlen(setclientid->sc_uaddr);
-	total_len = XDR_QUADLEN(len1) + XDR_QUADLEN(len2) + XDR_QUADLEN(len3);
-	total_len = (total_len << 2) + 24 + sizeof(setclientid->sc_verifier.data);
-
-	RESERVE_SPACE(total_len);
+	RESERVE_SPACE(4 + sizeof(setclientid->sc_verifier->data));
 	WRITE32(OP_SETCLIENTID);
-	WRITEMEM(setclientid->sc_verifier.data, sizeof(setclientid->sc_verifier.data));
-	WRITE32(len1);
-	WRITEMEM(setclientid->sc_name, len1);
+	WRITEMEM(setclientid->sc_verifier->data, sizeof(setclientid->sc_verifier->data));
+
+	encode_string(xdr, setclientid->sc_name_len, setclientid->sc_name);
+	RESERVE_SPACE(4);
 	WRITE32(setclientid->sc_prog);
-	WRITE32(len2);
-	WRITEMEM(setclientid->sc_netid, len2);
-	WRITE32(len3);
-	WRITEMEM(setclientid->sc_uaddr, len3);
+	encode_string(xdr, setclientid->sc_netid_len, setclientid->sc_netid);
+	encode_string(xdr, setclientid->sc_uaddr_len, setclientid->sc_uaddr);
+	RESERVE_SPACE(4);
 	WRITE32(setclientid->sc_cb_ident);
 
 	return 0;
@@ -1091,7 +1092,7 @@ static int encode_write(struct xdr_stream *xdr, const struct nfs_writeargs *args
 	RESERVE_SPACE(4);
 	WRITE32(OP_WRITE);
 
-	encode_stateid(xdr, args->state, args->lockowner);
+	encode_stateid(xdr, args->context);
 
 	RESERVE_SPACE(16);
 	WRITE64(args->offset);
@@ -1252,9 +1253,9 @@ static int nfs4_xdr_enc_create(struct rpc_rqst *req, uint32_t *p, const struct n
 		goto out;
 	if ((status = encode_create(&xdr, args)) != 0)
 		goto out;
-	if ((status = encode_getfattr(&xdr, args->bitmask)) != 0)
+	if ((status = encode_getfh(&xdr)) != 0)
 		goto out;
-	status = encode_getfh(&xdr);
+	status = encode_getfattr(&xdr, args->bitmask);
 out:
 	return status;
 }
@@ -1325,10 +1326,10 @@ static int nfs4_xdr_enc_open(struct rpc_rqst *req, uint32_t *p, struct nfs_opena
 	status = encode_open(&xdr, args);
 	if (status)
 		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
+	status = encode_getfh(&xdr);
 	if (status)
 		goto out;
-	status = encode_getfh(&xdr);
+	status = encode_getfattr(&xdr, args->bitmask);
 out:
 	return status;
 }
@@ -3220,9 +3221,11 @@ static int nfs4_xdr_dec_create(struct rpc_rqst *rqstp, uint32_t *p, struct nfs4_
 		goto out;
 	if ((status = decode_create(&xdr,&res->dir_cinfo)) != 0)
 		goto out;
-	if ((status = decode_getfattr(&xdr, res->fattr, res->server)) != 0)
+	if ((status = decode_getfh(&xdr, res->fh)) != 0)
 		goto out;
-	status = decode_getfh(&xdr, res->fh);
+	status = decode_getfattr(&xdr, res->fattr, res->server);
+	if (status == NFS4ERR_DELAY)
+		status = 0;
 out:
 	return status;
 }
@@ -3296,12 +3299,14 @@ static int nfs4_xdr_dec_open(struct rpc_rqst *rqstp, uint32_t *p, struct nfs_ope
         if (status)
                 goto out;
         status = decode_open(&xdr, res);
-        if (status)
-                goto out;
-	status = decode_getfattr(&xdr, res->f_attr, res->server);
         if (status)
                 goto out;
 	status = decode_getfh(&xdr, &res->fh);
+        if (status)
+		goto out;
+	status = decode_getfattr(&xdr, res->f_attr, res->server);
+	if (status == NFS4ERR_DELAY)
+		status = 0;
 out:
         return status;
 }
@@ -3347,6 +3352,8 @@ static int nfs4_xdr_dec_open_reclaim(struct rpc_rqst *rqstp, uint32_t *p, struct
         if (status)
                 goto out;
 	status = decode_getfattr(&xdr, res->f_attr, res->server);
+	if (status == NFS4ERR_DELAY)
+		status = 0;
 out:
         return status;
 }
@@ -3371,6 +3378,8 @@ static int nfs4_xdr_dec_setattr(struct rpc_rqst *rqstp, uint32_t *p, struct nfs_
         if (status)
                 goto out;
 	status = decode_getfattr(&xdr, res->fattr, res->server);
+	if (status == NFS4ERR_DELAY)
+		status = 0;
 out:
         return status;
 }

fs/nfs/pagelist.c

@@ -31,7 +31,6 @@ nfs_page_alloc(void)
 	if (p) {
 		memset(p, 0, sizeof(*p));
 		INIT_LIST_HEAD(&p->wb_list);
-		init_waitqueue_head(&p->wb_wait);
 	}
 	return p;
 }
@@ -57,7 +56,7 @@ nfs_page_free(struct nfs_page *p)
  * User should ensure it is safe to sleep in this function.
  */
 struct nfs_page *
-nfs_create_request(struct file *file, struct inode *inode,
+nfs_create_request(struct nfs_open_context *ctx, struct inode *inode,
 		   struct page *page,
 		   unsigned int offset, unsigned int count)
 {
@@ -89,33 +88,38 @@ nfs_create_request(struct file *file, struct inode *inode,
 	req->wb_offset  = offset;
 	req->wb_pgbase	= offset;
 	req->wb_bytes   = count;
-	req->wb_inode   = inode;
 	atomic_set(&req->wb_count, 1);
-	server->rpc_ops->request_init(req, file);
+	req->wb_context = get_nfs_open_context(ctx);
 
 	return req;
 }
 
+/**
+ * nfs_unlock_request - Unlock request and wake up sleepers.
+ * @req:
+ */
+void nfs_unlock_request(struct nfs_page *req)
+{
+	if (!NFS_WBACK_BUSY(req)) {
+		printk(KERN_ERR "NFS: Invalid unlock attempted\n");
+		BUG();
+	}
+	smp_mb__before_clear_bit();
+	clear_bit(PG_BUSY, &req->wb_flags);
+	smp_mb__after_clear_bit();
+	wake_up_all(&req->wb_context->waitq);
+	nfs_release_request(req);
+}
+
 /**
  * nfs_clear_request - Free up all resources allocated to the request
  * @req:
  *
- * Release all resources associated with a write request after it
+ * Release page resources associated with a write request after it
  * has completed.
  */
 void nfs_clear_request(struct nfs_page *req)
 {
-	if (req->wb_state)
-		req->wb_state = NULL;
-	/* Release struct file or cached credential */
-	if (req->wb_file) {
-		fput(req->wb_file);
-		req->wb_file = NULL;
-	}
-	if (req->wb_cred) {
-		put_rpccred(req->wb_cred);
-		req->wb_cred = NULL;
-	}
 	if (req->wb_page) {
 		page_cache_release(req->wb_page);
 		req->wb_page = NULL;
@@ -142,6 +146,7 @@ nfs_release_request(struct nfs_page *req)
 
 	/* Release struct file or cached credential */
 	nfs_clear_request(req);
+	put_nfs_open_context(req->wb_context);
 	nfs_page_free(req);
 }
 
@@ -185,12 +190,12 @@ nfs_list_add_request(struct nfs_page *req, struct list_head *head)
 int
 nfs_wait_on_request(struct nfs_page *req)
 {
-	struct inode	*inode = req->wb_inode;
+	struct inode	*inode = req->wb_context->dentry->d_inode;
         struct rpc_clnt	*clnt = NFS_CLIENT(inode);
 
 	if (!NFS_WBACK_BUSY(req))
 		return 0;
-	return nfs_wait_event(clnt, req->wb_wait, !NFS_WBACK_BUSY(req));
+	return nfs_wait_event(clnt, req->wb_context->waitq, !NFS_WBACK_BUSY(req));
 }
 
 /**
@@ -215,7 +220,11 @@ nfs_coalesce_requests(struct list_head *head, struct list_head *dst,
 
 		req = nfs_list_entry(head->next);
 		if (prev) {
-			if (req->wb_cred != prev->wb_cred)
+			if (req->wb_context->cred != prev->wb_context->cred)
+				break;
+			if (req->wb_context->lockowner != prev->wb_context->lockowner)
+				break;
+			if (req->wb_context->state != prev->wb_context->state)
 				break;
 			if (req->wb_index != (prev->wb_index + 1))
 				break;

fs/nfs/proc.c

@@ -49,18 +49,6 @@
 
 extern struct rpc_procinfo nfs_procedures[];
 
-static struct rpc_cred *
-nfs_cred(struct inode *inode, struct file *filp)
-{
-	struct rpc_cred *cred = NULL;
-
-	if (filp)
-		cred = (struct rpc_cred *)filp->private_data;
-	if (!cred)
-		cred = NFS_I(inode)->mm_cred;
-	return cred;
-}
-
 /*
  * Bare-bones access to getattr: this is for nfs_read_super.
  */
@@ -99,14 +87,15 @@ nfs_proc_get_root(struct nfs_server *server, struct nfs_fh *fhandle,
  * One function for each procedure in the NFS protocol.
  */
 static int
-nfs_proc_getattr(struct inode *inode, struct nfs_fattr *fattr)
+nfs_proc_getattr(struct nfs_server *server, struct nfs_fh *fhandle,
+		struct nfs_fattr *fattr)
 {
 	int	status;
 
 	dprintk("NFS call  getattr\n");
 	fattr->valid = 0;
-	status = rpc_call(NFS_CLIENT(inode), NFSPROC_GETATTR,
-				NFS_FH(inode), fattr, 0);
+	status = rpc_call(server->client, NFSPROC_GETATTR,
+				fhandle, fattr, 0);
 	dprintk("NFS reply getattr\n");
 	return status;
 }
@@ -167,8 +156,7 @@ nfs_proc_readlink(struct inode *inode, struct page *page)
 	return status;
 }
 
-static int
-nfs_proc_read(struct nfs_read_data *rdata, struct file *filp)
+static int nfs_proc_read(struct nfs_read_data *rdata)
 {
 	int			flags = rdata->flags;
 	struct inode *		inode = rdata->inode;
@@ -177,15 +165,14 @@ nfs_proc_read(struct nfs_read_data *rdata, struct file *filp)
 		.rpc_proc	= &nfs_procedures[NFSPROC_READ],
 		.rpc_argp	= &rdata->args,
 		.rpc_resp	= &rdata->res,
+		.rpc_cred	= rdata->cred,
 	};
 	int			status;
 
 	dprintk("NFS call  read %d @ %Ld\n", rdata->args.count,
 			(long long) rdata->args.offset);
 	fattr->valid = 0;
-	msg.rpc_cred = nfs_cred(inode, filp);
 	status = rpc_call_sync(NFS_CLIENT(inode), &msg, flags);
-
 	if (status >= 0) {
 		nfs_refresh_inode(inode, fattr);
 		/* Emulate the eof flag, which isn't normally needed in NFSv2
@@ -198,8 +185,7 @@ nfs_proc_read(struct nfs_read_data *rdata, struct file *filp)
 	return status;
 }
 
-static int
-nfs_proc_write(struct nfs_write_data *wdata, struct file *filp)
+static int nfs_proc_write(struct nfs_write_data *wdata)
 {
 	int			flags = wdata->flags;
 	struct inode *		inode = wdata->inode;
@@ -208,13 +194,13 @@ nfs_proc_write(struct nfs_write_data *wdata, struct file *filp)
 		.rpc_proc	= &nfs_procedures[NFSPROC_WRITE],
 		.rpc_argp	= &wdata->args,
 		.rpc_resp	= &wdata->res,
+		.rpc_cred	= wdata->cred,
 	};
 	int			status;
 
 	dprintk("NFS call  write %d @ %Ld\n", wdata->args.count,
 			(long long) wdata->args.offset);
 	fattr->valid = 0;
-	msg.rpc_cred = nfs_cred(inode, filp);
 	status = rpc_call_sync(NFS_CLIENT(inode), &msg, flags);
 	if (status >= 0) {
 		nfs_refresh_inode(inode, fattr);
@@ -621,27 +607,6 @@ nfs_proc_commit_setup(struct nfs_write_data *data, int how)
 	BUG();
 }
 
-/*
- * Set up the nfspage struct with the right credentials
- */
-static void
-nfs_request_init(struct nfs_page *req, struct file *filp)
-{
-	req->wb_cred = get_rpccred(nfs_cred(req->wb_inode, filp));
-}
-
-static int
-nfs_request_compatible(struct nfs_page *req, struct file *filp, struct page *page)
-{
-	if (req->wb_file != filp)
-		return 0;
-	if (req->wb_page != page)
-		return 0;
-	if (req->wb_cred != nfs_file_cred(filp))
-		return 0;
-	return 1;
-}
-
 static int
 nfs_proc_lock(struct file *filp, int cmd, struct file_lock *fl)
 {
@@ -682,7 +647,5 @@ struct nfs_rpc_ops	nfs_v2_clientops = {
 	.commit_setup	= nfs_proc_commit_setup,
 	.file_open	= nfs_open,
 	.file_release	= nfs_release,
-	.request_init	= nfs_request_init,
-	.request_compatible = nfs_request_compatible,
 	.lock		= nfs_proc_lock,
 };

fs/nfs/read.c

@@ -91,8 +91,8 @@ int nfs_return_empty_page(struct page *page)
 /*
  * Read a page synchronously.
  */
-static int
-nfs_readpage_sync(struct file *file, struct inode *inode, struct page *page)
+static int nfs_readpage_sync(struct nfs_open_context *ctx, struct inode *inode,
+		struct page *page)
 {
 	unsigned int	rsize = NFS_SERVER(inode)->rsize;
 	unsigned int	count = PAGE_CACHE_SIZE;
@@ -105,10 +105,11 @@ nfs_readpage_sync(struct file *file, struct inode *inode, struct page *page)
 
 	memset(rdata, 0, sizeof(*rdata));
 	rdata->flags = (IS_SWAPFILE(inode)? NFS_RPC_SWAPFLAGS : 0);
+	rdata->cred = ctx->cred;
 	rdata->inode = inode;
 	INIT_LIST_HEAD(&rdata->pages);
 	rdata->args.fh = NFS_FH(inode);
-	rdata->args.lockowner = current->files;
+	rdata->args.context = ctx;
 	rdata->args.pages = &page;
 	rdata->args.pgbase = 0UL;
 	rdata->args.count = rsize;
@@ -134,7 +135,7 @@ nfs_readpage_sync(struct file *file, struct inode *inode, struct page *page)
 			rdata->args.count);
 
 		lock_kernel();
-		result = NFS_PROTO(inode)->read(rdata, file);
+		result = NFS_PROTO(inode)->read(rdata);
 		unlock_kernel();
 
 		/*
@@ -169,8 +170,8 @@ nfs_readpage_sync(struct file *file, struct inode *inode, struct page *page)
 	return result;
 }
 
-static int
-nfs_readpage_async(struct file *file, struct inode *inode, struct page *page)
+static int nfs_readpage_async(struct nfs_open_context *ctx, struct inode *inode,
+		struct page *page)
 {
 	LIST_HEAD(one_request);
 	struct nfs_page	*new;
@@ -179,7 +180,7 @@ nfs_readpage_async(struct file *file, struct inode *inode, struct page *page)
 	len = nfs_page_length(inode, page);
 	if (len == 0)
 		return nfs_return_empty_page(page);
-	new = nfs_create_request(file, inode, page, 0, len);
+	new = nfs_create_request(ctx, inode, page, 0, len);
 	if (IS_ERR(new)) {
 		unlock_page(page);
 		return PTR_ERR(new);
@@ -202,8 +203,8 @@ static void nfs_readpage_release(struct nfs_page *req)
 	nfs_unlock_request(req);
 
 	dprintk("NFS: read done (%s/%Ld %d@%Ld)\n",
-			req->wb_inode->i_sb->s_id,
-			(long long)NFS_FILEID(req->wb_inode),
+			req->wb_context->dentry->d_inode->i_sb->s_id,
+			(long long)NFS_FILEID(req->wb_context->dentry->d_inode),
 			req->wb_bytes,
 			(long long)req_offset(req));
 }
@@ -217,16 +218,15 @@ static void nfs_read_rpcsetup(struct nfs_page *req, struct nfs_read_data *data,
 	struct inode		*inode;
 
 	data->req	  = req;
-	data->inode	  = inode = req->wb_inode;
-	data->cred	  = req->wb_cred;
+	data->inode	  = inode = req->wb_context->dentry->d_inode;
+	data->cred	  = req->wb_context->cred;
 
 	data->args.fh     = NFS_FH(inode);
 	data->args.offset = req_offset(req) + offset;
 	data->args.pgbase = req->wb_pgbase + offset;
 	data->args.pages  = data->pagevec;
 	data->args.count  = count;
-	data->args.lockowner = req->wb_lockowner;
-	data->args.state  = req->wb_state;
+	data->args.context = req->wb_context;
 
 	data->res.fattr   = &data->fattr;
 	data->res.count   = count;
@@ -396,7 +396,7 @@ nfs_pagein_list(struct list_head *head, int rpages)
 	while (!list_empty(head)) {
 		pages += nfs_coalesce_requests(head, &one_request, rpages);
 		req = nfs_list_entry(one_request.next);
-		error = nfs_pagein_one(&one_request, req->wb_inode);
+		error = nfs_pagein_one(&one_request, req->wb_context->dentry->d_inode);
 		if (error < 0)
 			break;
 	}
@@ -500,9 +500,9 @@ void nfs_readpage_result(struct rpc_task *task)
  *  -	The error flag is set for this page. This happens only when a
  *	previous async read operation failed.
  */
-int
-nfs_readpage(struct file *file, struct page *page)
+int nfs_readpage(struct file *file, struct page *page)
 {
+	struct nfs_open_context *ctx;
 	struct inode *inode = page->mapping->host;
 	int		error;
 
@@ -519,25 +519,33 @@ nfs_readpage(struct file *file, struct page *page)
 	if (error)
 		goto out_error;
 
+	if (file == NULL) {
+		ctx = nfs_find_open_context(inode, FMODE_READ);
+		if (ctx == NULL)
+			return -EBADF;
+	} else
+		ctx = get_nfs_open_context((struct nfs_open_context *)
+				file->private_data);
 	if (!IS_SYNC(inode)) {
-		error = nfs_readpage_async(file, inode, page);
+		error = nfs_readpage_async(ctx, inode, page);
 		goto out;
 	}
 
-	error = nfs_readpage_sync(file, inode, page);
+	error = nfs_readpage_sync(ctx, inode, page);
 	if (error < 0 && IS_SWAPFILE(inode))
 		printk("Aiee.. nfs swap-in of page failed!\n");
 out:
+	put_nfs_open_context(ctx);
 	return error;
 
 out_error:
 	unlock_page(page);
-	goto out;
+	return error;
 }
 
 struct nfs_readdesc {
 	struct list_head *head;
-	struct file *filp;
+	struct nfs_open_context *ctx;
 };
 
 static int
@@ -552,7 +560,7 @@ readpage_async_filler(void *data, struct page *page)
 	len = nfs_page_length(inode, page);
 	if (len == 0)
 		return nfs_return_empty_page(page);
-	new = nfs_create_request(desc->filp, inode, page, 0, len);
+	new = nfs_create_request(desc->ctx, inode, page, 0, len);
 	if (IS_ERR(new)) {
 			SetPageError(page);
 			unlock_page(page);
@@ -565,13 +573,11 @@ readpage_async_filler(void *data, struct page *page)
 	return 0;
 }
 
-int
-nfs_readpages(struct file *filp, struct address_space *mapping,
+int nfs_readpages(struct file *filp, struct address_space *mapping,
 		struct list_head *pages, unsigned nr_pages)
 {
 	LIST_HEAD(head);
 	struct nfs_readdesc desc = {
-		.filp		= filp,
 		.head		= &head,
 	};
 	struct inode *inode = mapping->host;
@@ -583,12 +589,20 @@ nfs_readpages(struct file *filp, struct address_space *mapping,
 			(long long)NFS_FILEID(inode),
 			nr_pages);
 
+	if (filp == NULL) {
+		desc.ctx = nfs_find_open_context(inode, FMODE_READ);
+		if (desc.ctx == NULL)
+			return -EBADF;
+	} else
+		desc.ctx = get_nfs_open_context((struct nfs_open_context *)
+				filp->private_data);
 	ret = read_cache_pages(mapping, pages, readpage_async_filler, &desc);
 	if (!list_empty(&head)) {
 		int err = nfs_pagein_list(&head, server->rpages);
 		if (!ret)
 			ret = err;
 	}
+	put_nfs_open_context(desc.ctx);
 	return ret;
 }
 

fs/nfs/write.c

@@ -71,7 +71,8 @@
 /*
  * Local function declarations
  */
-static struct nfs_page * nfs_update_request(struct file*, struct inode *,
+static struct nfs_page * nfs_update_request(struct nfs_open_context*,
+					    struct inode *,
 					    struct page *,
 					    unsigned int, unsigned int);
 static void nfs_writeback_done_partial(struct nfs_write_data *, int);
@@ -173,7 +174,7 @@ static void nfs_mark_uptodate(struct page *page, unsigned int base, unsigned int
  * Write a page synchronously.
  * Offset is the data offset within the page.
  */
-static int nfs_writepage_sync(struct file *file, struct inode *inode,
+static int nfs_writepage_sync(struct nfs_open_context *ctx, struct inode *inode,
 		struct page *page, unsigned int offset, unsigned int count,
 		int how)
 {
@@ -187,9 +188,10 @@ static int nfs_writepage_sync(struct file *file, struct inode *inode,
 
 	memset(wdata, 0, sizeof(*wdata));
 	wdata->flags = how;
+	wdata->cred = ctx->cred;
 	wdata->inode = inode;
 	wdata->args.fh = NFS_FH(inode);
-	wdata->args.lockowner = current->files;
+	wdata->args.context = ctx;
 	wdata->args.pages = &page;
 	wdata->args.stable = NFS_FILE_SYNC;
 	wdata->args.pgbase = offset;
@@ -208,7 +210,7 @@ static int nfs_writepage_sync(struct file *file, struct inode *inode,
 			wdata->args.count = count;
 		wdata->args.offset = page_offset(page) + wdata->args.pgbase;
 
-		result = NFS_PROTO(inode)->write(wdata, file);
+		result = NFS_PROTO(inode)->write(wdata);
 
 		if (result < 0) {
 			/* Must mark the page invalid after I/O error */
@@ -234,20 +236,19 @@ static int nfs_writepage_sync(struct file *file, struct inode *inode,
 
 io_error:
 	nfs_end_data_update_defer(inode);
-	if (wdata->cred)
-		put_rpccred(wdata->cred);
 
 	kfree(wdata);
 	return written ? written : result;
 }
 
-static int nfs_writepage_async(struct file *file, struct inode *inode,
-		struct page *page, unsigned int offset, unsigned int count)
+static int nfs_writepage_async(struct nfs_open_context *ctx,
+		struct inode *inode, struct page *page,
+		unsigned int offset, unsigned int count)
 {
 	struct nfs_page	*req;
 	int		status;
 
-	req = nfs_update_request(file, inode, page, offset, count);
+	req = nfs_update_request(ctx, inode, page, offset, count);
 	status = (IS_ERR(req)) ? PTR_ERR(req) : 0;
 	if (status < 0)
 		goto out;
@@ -274,6 +275,7 @@ static int wb_priority(struct writeback_control *wbc)
  */
 int nfs_writepage(struct page *page, struct writeback_control *wbc)
 {
+	struct nfs_open_context *ctx;
 	struct inode *inode = page->mapping->host;
 	unsigned long end_index;
 	unsigned offset = PAGE_CACHE_SIZE;
@@ -308,16 +310,21 @@ int nfs_writepage(struct page *page, struct writeback_control *wbc)
 	if (page->index >= end_index+1 || !offset)
 		goto out;
 do_it:
+	ctx = nfs_find_open_context(inode, FMODE_WRITE);
+	if (ctx == NULL) {
+		err = -EBADF;
+		goto out;
+	}
 	lock_kernel();
 	if (!IS_SYNC(inode) && inode_referenced) {
-		err = nfs_writepage_async(NULL, inode, page, 0, offset);
+		err = nfs_writepage_async(ctx, inode, page, 0, offset);
 		if (err >= 0) {
 			err = 0;
 			if (wbc->for_reclaim)
 				nfs_flush_inode(inode, 0, 0, FLUSH_STABLE);
 		}
 	} else {
-		err = nfs_writepage_sync(NULL, inode, page, 0,
+		err = nfs_writepage_sync(ctx, inode, page, 0,
 						offset, priority);
 		if (err >= 0) {
 			if (err != offset)
@@ -326,6 +333,7 @@ int nfs_writepage(struct page *page, struct writeback_control *wbc)
 		}
 	}
 	unlock_kernel();
+	put_nfs_open_context(ctx);
 out:
 	unlock_page(page);
 	if (inode_referenced)
@@ -396,10 +404,9 @@ nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
 /*
  * Insert a write request into an inode
  */
-static void
-nfs_inode_remove_request(struct nfs_page *req)
+static void nfs_inode_remove_request(struct nfs_page *req)
 {
-	struct inode *inode = req->wb_inode;
+	struct inode *inode = req->wb_context->dentry->d_inode;
 	struct nfs_inode *nfsi = NFS_I(inode);
 
 	BUG_ON (!NFS_WBACK_BUSY(req));
@@ -450,7 +457,7 @@ nfs_find_request(struct inode *inode, unsigned long index)
 static void
 nfs_mark_request_dirty(struct nfs_page *req)
 {
-	struct inode *inode = req->wb_inode;
+	struct inode *inode = req->wb_context->dentry->d_inode;
 	struct nfs_inode *nfsi = NFS_I(inode);
 
 	spin_lock(&nfsi->req_lock);
@@ -467,7 +474,7 @@ nfs_mark_request_dirty(struct nfs_page *req)
 static inline int
 nfs_dirty_request(struct nfs_page *req)
 {
-	struct nfs_inode *nfsi = NFS_I(req->wb_inode);
+	struct nfs_inode *nfsi = NFS_I(req->wb_context->dentry->d_inode);
 	return !list_empty(&req->wb_list) && req->wb_list_head == &nfsi->dirty;
 }
 
@@ -478,7 +485,7 @@ nfs_dirty_request(struct nfs_page *req)
 static void
 nfs_mark_request_commit(struct nfs_page *req)
 {
-	struct inode *inode = req->wb_inode;
+	struct inode *inode = req->wb_context->dentry->d_inode;
 	struct nfs_inode *nfsi = NFS_I(inode);
 
 	spin_lock(&nfsi->req_lock);
@@ -619,9 +626,9 @@ static int nfs_wait_on_write_congestion(struct address_space *mapping, int intr)
  *
  * Note: Should always be called with the Page Lock held!
  */
-static struct nfs_page *
-nfs_update_request(struct file* file, struct inode *inode, struct page *page,
-		   unsigned int offset, unsigned int bytes)
+static struct nfs_page * nfs_update_request(struct nfs_open_context* ctx,
+		struct inode *inode, struct page *page,
+		unsigned int offset, unsigned int bytes)
 {
 	struct nfs_server *server = NFS_SERVER(inode);
 	struct nfs_inode *nfsi = NFS_I(inode);
@@ -669,13 +676,9 @@ nfs_update_request(struct file* file, struct inode *inode, struct page *page,
 		}
 		spin_unlock(&nfsi->req_lock);
 
-		new = nfs_create_request(file, inode, page, offset, bytes);
+		new = nfs_create_request(ctx, inode, page, offset, bytes);
 		if (IS_ERR(new))
 			return new;
-		if (file) {
-			new->wb_file = file;
-			get_file(file);
-		}
 	}
 
 	/* We have a request for our page.
@@ -685,7 +688,7 @@ nfs_update_request(struct file* file, struct inode *inode, struct page *page,
 	 * request.
 	 */
 	rqend = req->wb_offset + req->wb_bytes;
-	if (req->wb_file != file
+	if (req->wb_context != ctx
 	    || req->wb_page != page
 	    || !nfs_dirty_request(req)
 	    || offset > rqend || end < req->wb_offset) {
@@ -706,9 +709,9 @@ nfs_update_request(struct file* file, struct inode *inode, struct page *page,
 	return req;
 }
 
-int
-nfs_flush_incompatible(struct file *file, struct page *page)
+int nfs_flush_incompatible(struct file *file, struct page *page)
 {
+	struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data;
 	struct inode	*inode = page->mapping->host;
 	struct nfs_page	*req;
 	int		status = 0;
@@ -722,7 +725,7 @@ nfs_flush_incompatible(struct file *file, struct page *page)
 	 */
 	req = nfs_find_request(inode, page->index);
 	if (req) {
-		if (!NFS_PROTO(inode)->request_compatible(req, file, page))
+		if (req->wb_page != page || ctx != req->wb_context)
 			status = nfs_wb_page(inode, page);
 		nfs_release_request(req);
 	}
@@ -738,6 +741,7 @@ nfs_flush_incompatible(struct file *file, struct page *page)
 int nfs_updatepage(struct file *file, struct page *page,
 		unsigned int offset, unsigned int count)
 {
+	struct nfs_open_context *ctx = (struct nfs_open_context *)file->private_data;
 	struct dentry	*dentry = file->f_dentry;
 	struct inode	*inode = page->mapping->host;
 	struct nfs_page	*req;
@@ -748,7 +752,7 @@ int nfs_updatepage(struct file *file, struct page *page,
 		count, (long long)(page_offset(page) +offset));
 
 	if (IS_SYNC(inode)) {
-		status = nfs_writepage_sync(file, inode, page, offset, count, 0);
+		status = nfs_writepage_sync(ctx, inode, page, offset, count, 0);
 		if (status > 0) {
 			if (offset == 0 && status == PAGE_CACHE_SIZE)
 				SetPageUptodate(page);
@@ -785,7 +789,7 @@ int nfs_updatepage(struct file *file, struct page *page,
 	 * it out now.
 	 */
 	do {
-		req = nfs_update_request(file, inode, page, offset, count);
+		req = nfs_update_request(ctx, inode, page, offset, count);
 		status = (IS_ERR(req)) ? PTR_ERR(req) : 0;
 		if (status != -EBUSY)
 			break;
@@ -861,16 +865,15 @@ static void nfs_write_rpcsetup(struct nfs_page *req,
 	 * NB: take care not to mess about with data->commit et al. */
 
 	data->req = req;
-	data->inode = inode = req->wb_inode;
-	data->cred = req->wb_cred;
+	data->inode = inode = req->wb_context->dentry->d_inode;
+	data->cred = req->wb_context->cred;
 
 	data->args.fh     = NFS_FH(inode);
 	data->args.offset = req_offset(req) + offset;
 	data->args.pgbase = req->wb_pgbase + offset;
 	data->args.pages  = data->pagevec;
 	data->args.count  = count;
-	data->args.lockowner = req->wb_lockowner;
-	data->args.state  = req->wb_state;
+	data->args.context = req->wb_context;
 
 	data->res.fattr   = &data->fattr;
 	data->res.count   = count;
@@ -1030,7 +1033,7 @@ nfs_flush_list(struct list_head *head, int wpages, int how)
 	while (!list_empty(head)) {
 		pages += nfs_coalesce_requests(head, &one_request, wpages);
 		req = nfs_list_entry(one_request.next);
-		error = nfs_flush_one(&one_request, req->wb_inode, how);
+		error = nfs_flush_one(&one_request, req->wb_context->dentry->d_inode, how);
 		if (error < 0)
 			break;
 	}
@@ -1055,16 +1058,15 @@ static void nfs_writeback_done_partial(struct nfs_write_data *data, int status)
 	struct page		*page = req->wb_page;
 
 	dprintk("NFS: write (%s/%Ld %d@%Ld)",
-		req->wb_inode->i_sb->s_id,
-		(long long)NFS_FILEID(req->wb_inode),
+		req->wb_context->dentry->d_inode->i_sb->s_id,
+		(long long)NFS_FILEID(req->wb_context->dentry->d_inode),
 		req->wb_bytes,
 		(long long)req_offset(req));
 
 	if (status < 0) {
 		ClearPageUptodate(page);
 		SetPageError(page);
-		if (req->wb_file)
-			req->wb_file->f_error = status;
+		req->wb_context->error = status;
 		dprintk(", error = %d\n", status);
 	} else {
 #if defined(CONFIG_NFS_V3) || defined(CONFIG_NFS_V4)
@@ -1105,16 +1107,15 @@ static void nfs_writeback_done_full(struct nfs_write_data *data, int status)
 		page = req->wb_page;
 
 		dprintk("NFS: write (%s/%Ld %d@%Ld)",
-			req->wb_inode->i_sb->s_id,
-			(long long)NFS_FILEID(req->wb_inode),
+			req->wb_context->dentry->d_inode->i_sb->s_id,
+			(long long)NFS_FILEID(req->wb_context->dentry->d_inode),
 			req->wb_bytes,
 			(long long)req_offset(req));
 
 		if (status < 0) {
 			ClearPageUptodate(page);
 			SetPageError(page);
-			if (req->wb_file)
-				req->wb_file->f_error = status;
+			req->wb_context->error = status;
 			end_page_writeback(page);
 			nfs_inode_remove_request(req);
 			dprintk(", error = %d\n", status);
@@ -1233,7 +1234,7 @@ static void nfs_commit_rpcsetup(struct list_head *head,
 	list_splice_init(head, &data->pages);
 	first = nfs_list_entry(data->pages.next);
 	last = nfs_list_entry(data->pages.prev);
-	inode = first->wb_inode;
+	inode = first->wb_context->dentry->d_inode;
 
 	/*
 	 * Determine the offset range of requests in the COMMIT call.
@@ -1247,7 +1248,7 @@ static void nfs_commit_rpcsetup(struct list_head *head,
 		len = 0;
 
 	data->inode	  = inode;
-	data->cred	  = first->wb_cred;
+	data->cred	  = first->wb_context->cred;
 
 	data->args.fh     = NFS_FH(data->inode);
 	data->args.offset = start;
@@ -1314,13 +1315,12 @@ nfs_commit_done(struct rpc_task *task)
 		nfs_list_remove_request(req);
 
 		dprintk("NFS: commit (%s/%Ld %d@%Ld)",
-			req->wb_inode->i_sb->s_id,
-			(long long)NFS_FILEID(req->wb_inode),
+			req->wb_context->dentry->d_inode->i_sb->s_id,
+			(long long)NFS_FILEID(req->wb_context->dentry->d_inode),
 			req->wb_bytes,
 			(long long)req_offset(req));
 		if (task->tk_status < 0) {
-			if (req->wb_file)
-				req->wb_file->f_error = task->tk_status;
+			req->wb_context->error = task->tk_status;
 			nfs_inode_remove_request(req);
 			dprintk(", error = %d\n", task->tk_status);
 			goto next;

include/linux/nfs_fs.h

@@ -83,6 +83,20 @@ struct nfs_access_entry {
 	int			mask;
 };
 
+struct nfs4_state;
+struct nfs_open_context {
+	atomic_t count;
+	struct dentry *dentry;
+	struct rpc_cred *cred;
+	struct nfs4_state *state;
+	fl_owner_t lockowner;
+	int mode;
+	int error;
+
+	struct list_head list;
+	wait_queue_head_t waitq;
+};
+
 /*
  * nfs fs inode data in memory
  */
@@ -156,8 +170,8 @@ struct nfs_inode {
 				ncommit,
 				npages;
 
-	/* Credentials for shared mmap */
-	struct rpc_cred		*mm_cred;
+	/* Open contexts for shared mmap writes */
+	struct list_head	open_files;
 
 	wait_queue_head_t	nfs_i_wait;
 
@@ -268,7 +282,6 @@ extern struct inode *nfs_fhget(struct super_block *, struct nfs_fh *,
 extern int nfs_refresh_inode(struct inode *, struct nfs_fattr *);
 extern int nfs_getattr(struct vfsmount *, struct dentry *, struct kstat *);
 extern int nfs_permission(struct inode *, int, struct nameidata *);
-extern void nfs_set_mmcred(struct inode *, struct rpc_cred *);
 extern int nfs_open(struct inode *, struct file *);
 extern int nfs_release(struct inode *, struct file *);
 extern int __nfs_revalidate_inode(struct nfs_server *, struct inode *);
@@ -278,6 +291,12 @@ extern void nfs_end_attr_update(struct inode *);
 extern void nfs_begin_data_update(struct inode *);
 extern void nfs_end_data_update(struct inode *);
 extern void nfs_end_data_update_defer(struct inode *);
+extern struct nfs_open_context *alloc_nfs_open_context(struct dentry *dentry, struct rpc_cred *cred);
+extern struct nfs_open_context *get_nfs_open_context(struct nfs_open_context *ctx);
+extern void put_nfs_open_context(struct nfs_open_context *ctx);
+extern void nfs_file_set_open_context(struct file *filp, struct nfs_open_context *ctx);
+extern struct nfs_open_context *nfs_find_open_context(struct inode *inode, int mode);
+extern void nfs_file_clear_open_context(struct file *filp);
 
 /* linux/net/ipv4/ipconfig.c: trims ip addr off front of name, too. */
 extern u32 root_nfs_parse_addr(char *name); /*__init*/
@@ -289,16 +308,15 @@ extern struct inode_operations nfs_file_inode_operations;
 extern struct file_operations nfs_file_operations;
 extern struct address_space_operations nfs_file_aops;
 
-static __inline__ struct rpc_cred *
-nfs_file_cred(struct file *file)
+static inline struct rpc_cred *nfs_file_cred(struct file *file)
 {
-	struct rpc_cred *cred = NULL;
-	if (file)
-		cred = (struct rpc_cred *)file->private_data;
-#ifdef RPC_DEBUG
-	BUG_ON(cred && cred->cr_magic != RPCAUTH_CRED_MAGIC);
-#endif
-	return cred;
+	if (file != NULL) {
+		struct nfs_open_context *ctx;
+
+		ctx = (struct nfs_open_context*)file->private_data;
+		return ctx->cred;
+	}
+	return NULL;
 }
 
 /*
@@ -507,8 +525,6 @@ struct idmap;
 
 enum nfs4_client_state {
 	NFS4CLNT_OK  = 0,
-	NFS4CLNT_NEW,
-	NFS4CLNT_SETUP_STATE,
 };
 
 /*
@@ -520,7 +536,6 @@ struct nfs4_client {
 	u64			cl_clientid;	/* constant */
 	nfs4_verifier		cl_confirm;
 	unsigned long		cl_state;
-	long			cl_generation;
 
 	u32			cl_lockowner_id;
 
@@ -573,9 +588,7 @@ struct nfs4_state_owner {
 	u32                  so_id;      /* 32-bit identifier, unique */
 	struct semaphore     so_sema;
 	u32                  so_seqid;   /* protected by so_sema */
-	unsigned int         so_flags;   /* protected by so_sema */
 	atomic_t	     so_count;
-	long		     so_generation;
 
 	struct rpc_cred	     *so_cred;	 /* Associated cred */
 	struct list_head     so_states;
@@ -596,7 +609,8 @@ struct nfs4_state_owner {
 struct nfs4_lock_state {
 	struct list_head	ls_locks;	/* Other lock stateids */
 	fl_owner_t		ls_owner;	/* POSIX lock owner */
-	struct nfs4_state *	ls_parent;	/* Parent nfs4_state */
+#define NFS_LOCK_INITIALIZED 1
+	int			ls_flags;
 	u32			ls_seqid;
 	u32			ls_id;
 	nfs4_stateid		ls_stateid;
@@ -629,6 +643,11 @@ struct nfs4_state {
 };
 
 
+struct nfs4_exception {
+	long timeout;
+	int retry;
+};
+
 extern struct dentry_operations nfs4_dentry_operations;
 extern struct inode_operations nfs4_dir_inode_operations;
 
@@ -639,10 +658,12 @@ extern int nfs4_open_reclaim(struct nfs4_state_owner *, struct nfs4_state *);
 extern int nfs4_proc_async_renew(struct nfs4_client *);
 extern int nfs4_proc_renew(struct nfs4_client *);
 extern int nfs4_do_close(struct inode *, struct nfs4_state *);
-int nfs4_do_downgrade(struct inode *inode, struct nfs4_state *state, mode_t mode);
+extern int nfs4_do_downgrade(struct inode *inode, struct nfs4_state *state, mode_t mode);
 extern int nfs4_wait_clnt_recover(struct rpc_clnt *, struct nfs4_client *);
 extern struct inode *nfs4_atomic_open(struct inode *, struct dentry *, struct nameidata *);
 extern int nfs4_open_revalidate(struct inode *, struct dentry *, int);
+extern int nfs4_handle_exception(struct nfs_server *, int, struct nfs4_exception *);
+extern int nfs4_lock_reclaim(struct nfs4_state *state, struct file_lock *request);
 
 /* nfs4renewd.c */
 extern void nfs4_schedule_state_renewal(struct nfs4_client *);
@@ -654,6 +675,7 @@ extern void init_nfsv4_state(struct nfs_server *);
 extern void destroy_nfsv4_state(struct nfs_server *);
 extern struct nfs4_client *nfs4_get_client(struct in_addr *);
 extern void nfs4_put_client(struct nfs4_client *clp);
+extern int nfs4_init_client(struct nfs4_client *clp);
 extern u32 nfs4_alloc_lockowner_id(struct nfs4_client *);
 
 extern struct nfs4_state_owner * nfs4_get_state_owner(struct nfs_server *, struct rpc_cred *);
@@ -663,14 +685,13 @@ extern void nfs4_put_open_state(struct nfs4_state *);
 extern void nfs4_close_state(struct nfs4_state *, mode_t);
 extern struct nfs4_state *nfs4_find_state(struct inode *, struct rpc_cred *, mode_t mode);
 extern void nfs4_increment_seqid(int status, struct nfs4_state_owner *sp);
-extern int nfs4_handle_error(struct nfs_server *, int);
 extern void nfs4_schedule_state_recovery(struct nfs4_client *);
 extern struct nfs4_lock_state *nfs4_find_lock_state(struct nfs4_state *state, fl_owner_t);
-extern struct nfs4_lock_state *nfs4_alloc_lock_state(struct nfs4_state *state, fl_owner_t);
+extern struct nfs4_lock_state *nfs4_get_lock_state(struct nfs4_state *state, fl_owner_t);
 extern void nfs4_put_lock_state(struct nfs4_lock_state *state);
 extern void nfs4_increment_lock_seqid(int status, struct nfs4_lock_state *ls);
-extern void nfs4_notify_setlk(struct inode *, struct file_lock *, struct nfs4_lock_state *);
-extern void nfs4_notify_unlck(struct inode *, struct file_lock *, struct nfs4_lock_state *);
+extern void nfs4_notify_setlk(struct nfs4_state *, struct file_lock *, struct nfs4_lock_state *);
+extern void nfs4_notify_unlck(struct nfs4_state *, struct file_lock *, struct nfs4_lock_state *);
 extern void nfs4_copy_stateid(nfs4_stateid *, struct nfs4_state *, fl_owner_t);
 
 
@@ -681,6 +702,7 @@ struct nfs4_mount_data;
 #define destroy_nfsv4_state(server)       do { } while (0)
 #define nfs4_put_state_owner(inode, owner) do { } while (0)
 #define nfs4_put_open_state(state) do { } while (0)
+#define nfs4_close_state(a, b) do { } while (0)
 #define nfs4_renewd_prepare_shutdown(server) do { } while (0)
 #endif
 

include/linux/nfs_page.h

@@ -29,14 +29,9 @@
 struct nfs_page {
 	struct list_head	wb_list,	/* Defines state of page: */
 				*wb_list_head;	/*      read/write/commit */
-	struct file		*wb_file;
-	fl_owner_t		wb_lockowner;
-	struct inode		*wb_inode;
-	struct rpc_cred		*wb_cred;
-	struct nfs4_state	*wb_state;
 	struct page		*wb_page;	/* page to read in/write out */
+	struct nfs_open_context	*wb_context;	/* File state context info */
 	atomic_t		wb_complete;	/* i/os we're waiting for */
-	wait_queue_head_t	wb_wait;	/* wait queue */
 	unsigned long		wb_index;	/* Offset >> PAGE_CACHE_SHIFT */
 	unsigned int		wb_offset,	/* Offset & ~PAGE_CACHE_MASK */
 				wb_pgbase,	/* Start of page data */
@@ -50,9 +45,11 @@ struct nfs_page {
 #define NFS_NEED_COMMIT(req)	(test_bit(PG_NEED_COMMIT,&(req)->wb_flags))
 #define NFS_NEED_RESCHED(req)	(test_bit(PG_NEED_RESCHED,&(req)->wb_flags))
 
-extern	struct nfs_page *nfs_create_request(struct file *, struct inode *,
-					    struct page *,
-					    unsigned int, unsigned int);
+extern	struct nfs_page *nfs_create_request(struct nfs_open_context *ctx,
+					    struct inode *inode,
+					    struct page *page,
+					    unsigned int offset,
+					    unsigned int count);
 extern	void nfs_clear_request(struct nfs_page *req);
 extern	void nfs_release_request(struct nfs_page *req);
 
@@ -64,6 +61,7 @@ extern	int nfs_scan_list(struct list_head *, struct list_head *,
 extern	int nfs_coalesce_requests(struct list_head *, struct list_head *,
 				  unsigned int);
 extern  int nfs_wait_on_request(struct nfs_page *);
+extern	void nfs_unlock_request(struct nfs_page *req);
 
 /*
  * Lock the page of an asynchronous request without incrementing the wb_count
@@ -88,19 +86,6 @@ nfs_lock_request(struct nfs_page *req)
 	return 1;
 }
 
-static inline void
-nfs_unlock_request(struct nfs_page *req)
-{
-	if (!NFS_WBACK_BUSY(req)) {
-		printk(KERN_ERR "NFS: Invalid unlock attempted\n");
-		BUG();
-	}
-	smp_mb__before_clear_bit();
-	clear_bit(PG_BUSY, &req->wb_flags);
-	smp_mb__after_clear_bit();
-	wake_up_all(&req->wb_wait);
-	nfs_release_request(req);
-}
 
 /**
  * nfs_list_remove_request - Remove a request from its wb_list

include/linux/nfs_xdr.h

@@ -235,8 +235,7 @@ struct nfs_lockres {
 
 struct nfs_readargs {
 	struct nfs_fh *		fh;
-	fl_owner_t		lockowner;
-	struct nfs4_state *	state;
+	struct nfs_open_context *context;
 	__u64			offset;
 	__u32			count;
 	unsigned int		pgbase;
@@ -259,8 +258,7 @@ struct nfs_readres {
 
 struct nfs_writeargs {
 	struct nfs_fh *		fh;
-	fl_owner_t		lockowner;
-	struct nfs4_state *	state;
+	struct nfs_open_context *context;
 	__u64			offset;
 	__u32			count;
 	enum nfs3_stable_how	stable;
@@ -597,13 +595,15 @@ struct nfs4_rename_res {
 };
 
 struct nfs4_setclientid {
-	nfs4_verifier			sc_verifier;      /* request */
-	char *				sc_name;	  /* request */
+	const nfs4_verifier *		sc_verifier;      /* request */
+	unsigned int			sc_name_len;
+	char				sc_name[32];	  /* request */
 	u32				sc_prog;          /* request */
+	unsigned int			sc_netid_len;
 	char				sc_netid[4];	  /* request */
+	unsigned int			sc_uaddr_len;
 	char				sc_uaddr[24];     /* request */
 	u32				sc_cb_ident;      /* request */
-	struct nfs4_client *		sc_state;	  /* response */
 };
 
 struct nfs4_statfs_arg {
@@ -669,16 +669,17 @@ struct nfs_rpc_ops {
 
 	int	(*getroot) (struct nfs_server *, struct nfs_fh *,
 			    struct nfs_fsinfo *);
-	int	(*getattr) (struct inode *, struct nfs_fattr *);
+	int	(*getattr) (struct nfs_server *, struct nfs_fh *,
+			    struct nfs_fattr *);
 	int	(*setattr) (struct dentry *, struct nfs_fattr *,
 			    struct iattr *);
 	int	(*lookup)  (struct inode *, struct qstr *,
 			    struct nfs_fh *, struct nfs_fattr *);
 	int	(*access)  (struct inode *, struct nfs_access_entry *);
 	int	(*readlink)(struct inode *, struct page *);
-	int	(*read)    (struct nfs_read_data *, struct file *);
-	int	(*write)   (struct nfs_write_data *, struct file *);
-	int	(*commit)  (struct nfs_write_data *, struct file *);
+	int	(*read)    (struct nfs_read_data *);
+	int	(*write)   (struct nfs_write_data *);
+	int	(*commit)  (struct nfs_write_data *);
 	struct inode *	(*create)  (struct inode *, struct qstr *,
 			    struct iattr *, int);
 	int	(*remove)  (struct inode *, struct qstr *);
@@ -710,8 +711,6 @@ struct nfs_rpc_ops {
 	void	(*commit_setup) (struct nfs_write_data *, int how);
 	int	(*file_open)   (struct inode *, struct file *);
 	int	(*file_release) (struct inode *, struct file *);
-	void	(*request_init)(struct nfs_page *, struct file *);
-	int	(*request_compatible)(struct nfs_page *, struct file *, struct page *);
 	int	(*lock)(struct file *, int, struct file_lock *);
 };