Merge bk://linux.bkbits.net/linux-2.5

into plexity.net:/home/dsaxena/src/linux-2.5-bk

fs/Kconfig

@@ -1415,8 +1415,8 @@ config NFS_V3
 	bool "Provide NFSv3 client support"
 	depends on NFS_FS
 	help
-	  Say Y here if you want your NFS client to be able to speak the newer
-	  version 3 of the NFS protocol.
+	  Say Y here if you want your NFS client to be able to speak version
+	  3 of the NFS protocol.
 
 	  If unsure, say Y.
 
@@ -1560,6 +1560,22 @@ config RPCSEC_GSS_KRB5
 
 	  If unsure, say N.
 
+config RPCSEC_GSS_SPKM3
+	tristate "Secure RPC: SPKM3 mechanism (EXPERIMENTAL)"
+	depends on SUNRPC && EXPERIMENTAL
+	select SUNRPC_GSS
+	select CRYPTO
+	select CRYPTO_MD5
+	select CRYPTO_DES
+	help
+	  Provides for secure RPC calls by means of a gss-api
+	  mechanism based on the SPKM3 public-key mechanism.
+
+	  Note: Requires an auxiliary userspace daemon which may be found on
+	  	http://www.citi.umich.edu/projects/nfsv4/
+
+	  If unsure, say N.
+
 config SMB_FS
 	tristate "SMB file system support (to mount Windows shares etc.)"
 	depends on INET

fs/lockd/svclock.c

@@ -237,8 +237,13 @@ nlmsvc_delete_block(struct nlm_block *block, int unlock)
 
 	/* Remove block from list */
 	nlmsvc_remove_block(block);
-	posix_unblock_lock(file->f_file, fl);
-	block->b_granted = 0;
+	if (fl->fl_next)
+		posix_unblock_lock(file->f_file, fl);
+	if (unlock) {
+		fl->fl_type = F_UNLCK;
+		posix_lock_file(file->f_file, fl);
+		block->b_granted = 0;
+	}
 
 	/* If the block is in the middle of a GRANT callback,
 	 * don't kill it yet. */

fs/lockd/svcsubs.c

@@ -67,7 +67,7 @@ nlm_lookup_file(struct svc_rqst *rqstp, struct nlm_file **result,
 	down(&nlm_file_sema);
 
 	for (file = nlm_files[hash]; file; file = file->f_next)
-		if (!memcmp(&file->f_handle, f, sizeof(*f)))
+		if (!nfs_compare_fh(&file->f_handle, f))
 			goto found;
 
 	dprintk("lockd: creating file for (%08x %08x %08x %08x %08x %08x)\n",

fs/nfs/Makefile

@@ -9,6 +9,7 @@ nfs-y 			:= dir.o file.o inode.o nfs2xdr.o pagelist.o \
 nfs-$(CONFIG_ROOT_NFS)	+= nfsroot.o mount_clnt.o      
 nfs-$(CONFIG_NFS_V3)	+= nfs3proc.o nfs3xdr.o
 nfs-$(CONFIG_NFS_V4)	+= nfs4proc.o nfs4xdr.o nfs4state.o nfs4renewd.o \
-			   idmap.o
+			   delegation.o idmap.o \
+			   callback.o callback_xdr.o callback_proc.o
 nfs-$(CONFIG_NFS_DIRECTIO) += direct.o
 nfs-objs		:= $(nfs-y)

fs/nfs/callback.c

+/*
+ * linux/fs/nfs/callback.c
+ *
+ * Copyright (C) 2004 Trond Myklebust
+ *
+ * NFSv4 callback handling
+ */
+
+#include <linux/config.h>
+#include <linux/completion.h>
+#include <linux/ip.h>
+#include <linux/module.h>
+#include <linux/smp_lock.h>
+#include <linux/sunrpc/svc.h>
+#include <linux/sunrpc/svcsock.h>
+#include <linux/nfs_fs.h>
+#include "callback.h"
+
+#define NFSDBG_FACILITY NFSDBG_CALLBACK
+
+struct nfs_callback_data {
+	unsigned int users;
+	struct svc_serv *serv;
+	pid_t pid;
+	struct completion started;
+	struct completion stopped;
+};
+
+static struct nfs_callback_data nfs_callback_info;
+static DECLARE_MUTEX(nfs_callback_sema);
+static struct svc_program nfs4_callback_program;
+
+unsigned short nfs_callback_tcpport;
+
+/*
+ * This is the callback kernel thread.
+ */
+static void nfs_callback_svc(struct svc_rqst *rqstp)
+{
+	struct svc_serv *serv = rqstp->rq_server;
+	int err;
+
+	__module_get(THIS_MODULE);
+	lock_kernel();
+
+	nfs_callback_info.pid = current->pid;
+	daemonize("nfsv4-svc");
+	/* Process request with signals blocked, but allow SIGKILL.  */
+	allow_signal(SIGKILL);
+
+	complete(&nfs_callback_info.started);
+
+	while (nfs_callback_info.users != 0 || !signalled()) {
+		/*
+		 * Listen for a request on the socket
+		 */
+		err = svc_recv(serv, rqstp, MAX_SCHEDULE_TIMEOUT);
+		if (err == -EAGAIN || err == -EINTR)
+			continue;
+		if (err < 0) {
+			printk(KERN_WARNING
+					"%s: terminating on error %d\n",
+					__FUNCTION__, -err);
+			break;
+		}
+		dprintk("%s: request from %u.%u.%u.%u\n", __FUNCTION__,
+				NIPQUAD(rqstp->rq_addr.sin_addr.s_addr));
+		svc_process(serv, rqstp);
+	}
+
+	nfs_callback_info.pid = 0;
+	complete(&nfs_callback_info.stopped);
+	unlock_kernel();
+	module_put_and_exit(0);
+}
+
+/*
+ * Bring up the server process if it is not already up.
+ */
+int nfs_callback_up(void)
+{
+	struct svc_serv *serv;
+	struct svc_sock *svsk;
+	int ret = 0;
+
+	lock_kernel();
+	down(&nfs_callback_sema);
+	if (nfs_callback_info.users++ || nfs_callback_info.pid != 0)
+		goto out;
+	init_completion(&nfs_callback_info.started);
+	init_completion(&nfs_callback_info.stopped);
+	serv = svc_create(&nfs4_callback_program, NFS4_CALLBACK_BUFSIZE);
+	ret = -ENOMEM;
+	if (!serv)
+		goto out_err;
+	/* FIXME: We don't want to register this socket with the portmapper */
+	ret = svc_makesock(serv, IPPROTO_TCP, 0);
+	if (ret < 0)
+		goto out_destroy;
+	if (!list_empty(&serv->sv_permsocks)) {
+		svsk = list_entry(serv->sv_permsocks.next,
+				struct svc_sock, sk_list);
+		nfs_callback_tcpport = ntohs(inet_sk(svsk->sk_sk)->sport);
+		dprintk ("Callback port = 0x%x\n", nfs_callback_tcpport);
+	} else
+		BUG();
+	ret = svc_create_thread(nfs_callback_svc, serv);
+	if (ret < 0)
+		goto out_destroy;
+	nfs_callback_info.serv = serv;
+	wait_for_completion(&nfs_callback_info.started);
+out:
+	up(&nfs_callback_sema);
+	unlock_kernel();
+	return ret;
+out_destroy:
+	svc_destroy(serv);
+out_err:
+	nfs_callback_info.users--;
+	goto out;
+}
+
+/*
+ * Kill the server process if it is not already up.
+ */
+int nfs_callback_down(void)
+{
+	int ret = 0;
+
+	lock_kernel();
+	down(&nfs_callback_sema);
+	if (--nfs_callback_info.users || nfs_callback_info.pid == 0)
+		goto out;
+	kill_proc(nfs_callback_info.pid, SIGKILL, 1);
+	wait_for_completion(&nfs_callback_info.stopped);
+out:
+	up(&nfs_callback_sema);
+	unlock_kernel();
+	return ret;
+}
+
+/*
+ * AUTH_NULL authentication
+ */
+static int nfs_callback_null_accept(struct svc_rqst *rqstp, u32 *authp)
+{
+	struct kvec    *argv = &rqstp->rq_arg.head[0];
+	struct kvec    *resv = &rqstp->rq_res.head[0];
+
+	if (argv->iov_len < 3*4)
+		return SVC_GARBAGE;
+
+	if (svc_getu32(argv) != 0) {
+		dprintk("svc: bad null cred\n");
+		*authp = rpc_autherr_badcred;
+		return SVC_DENIED;
+	}
+	if (svc_getu32(argv) != RPC_AUTH_NULL || svc_getu32(argv) != 0) {
+		dprintk("svc: bad null verf\n");
+		 *authp = rpc_autherr_badverf;
+		 return SVC_DENIED;
+	}
+
+	/* Signal that mapping to nobody uid/gid is required */
+	rqstp->rq_cred.cr_uid = (uid_t) -1;
+	rqstp->rq_cred.cr_gid = (gid_t) -1;
+	rqstp->rq_cred.cr_group_info = groups_alloc(0);
+	if (rqstp->rq_cred.cr_group_info == NULL)
+		return SVC_DROP; /* kmalloc failure - client must retry */
+
+	/* Put NULL verifier */
+	svc_putu32(resv, RPC_AUTH_NULL);
+	svc_putu32(resv, 0);
+	dprintk("%s: success, returning %d!\n", __FUNCTION__, SVC_OK);
+	return SVC_OK;
+}
+
+static int nfs_callback_null_release(struct svc_rqst *rqstp)
+{
+	if (rqstp->rq_cred.cr_group_info)
+		put_group_info(rqstp->rq_cred.cr_group_info);
+	rqstp->rq_cred.cr_group_info = NULL;
+	return 0; /* don't drop */
+}
+
+static struct auth_ops nfs_callback_auth_null = {
+	.name = "null",
+	.flavour = RPC_AUTH_NULL,
+	.accept = nfs_callback_null_accept,
+	.release = nfs_callback_null_release,
+};
+
+/*
+ * AUTH_SYS authentication
+ */
+static int nfs_callback_unix_accept(struct svc_rqst *rqstp, u32 *authp)
+{
+	struct kvec    *argv = &rqstp->rq_arg.head[0];
+	struct kvec    *resv = &rqstp->rq_res.head[0];
+	struct svc_cred *cred = &rqstp->rq_cred;
+	u32 slen, i;
+	int len = argv->iov_len;
+
+	dprintk("%s: start\n", __FUNCTION__);
+	cred->cr_group_info = NULL;
+	rqstp->rq_client = NULL;
+	if ((len -= 3*4) < 0)
+		return SVC_GARBAGE;
+
+	/* Get length, time stamp and machine name */
+	svc_getu32(argv);
+	svc_getu32(argv);
+	slen = XDR_QUADLEN(ntohl(svc_getu32(argv)));
+	if (slen > 64 || (len -= (slen + 3)*4) < 0)
+		goto badcred;
+	argv->iov_base = (void*)((u32*)argv->iov_base + slen);
+	argv->iov_len -= slen*4;
+
+	cred->cr_uid = ntohl(svc_getu32(argv));
+	cred->cr_gid = ntohl(svc_getu32(argv));
+	slen = ntohl(svc_getu32(argv));
+	if (slen > 16 || (len -= (slen + 2)*4) < 0)
+		goto badcred;
+	cred->cr_group_info = groups_alloc(slen);
+	if (cred->cr_group_info == NULL)
+		return SVC_DROP;
+	for (i = 0; i < slen; i++)
+		GROUP_AT(cred->cr_group_info, i) = ntohl(svc_getu32(argv));
+
+	if (svc_getu32(argv) != RPC_AUTH_NULL || svc_getu32(argv) != 0) {
+		*authp = rpc_autherr_badverf;
+		return SVC_DENIED;
+	}
+	/* Put NULL verifier */
+	svc_putu32(resv, RPC_AUTH_NULL);
+	svc_putu32(resv, 0);
+	dprintk("%s: success, returning %d!\n", __FUNCTION__, SVC_OK);
+	return SVC_OK;
+badcred:
+	*authp = rpc_autherr_badcred;
+	return SVC_DENIED;
+}
+
+static int nfs_callback_unix_release(struct svc_rqst *rqstp)
+{
+	if (rqstp->rq_cred.cr_group_info)
+		put_group_info(rqstp->rq_cred.cr_group_info);
+	rqstp->rq_cred.cr_group_info = NULL;
+	return 0;
+}
+
+static struct auth_ops nfs_callback_auth_unix = {
+	.name = "unix",
+	.flavour = RPC_AUTH_UNIX,
+	.accept = nfs_callback_unix_accept,
+	.release = nfs_callback_unix_release,
+};
+
+/*
+ * Hook the authentication protocol
+ */
+static int nfs_callback_auth(struct svc_rqst *rqstp, u32 *authp)
+{
+	struct in_addr *addr = &rqstp->rq_addr.sin_addr;
+	struct nfs4_client *clp;
+	struct kvec *argv = &rqstp->rq_arg.head[0];
+	int flavour;
+	int retval;
+
+	/* Don't talk to strangers */
+	clp = nfs4_find_client(addr);
+	if (clp == NULL)
+		return SVC_DROP;
+	dprintk("%s: %u.%u.%u.%u NFSv4 callback!\n", __FUNCTION__, NIPQUAD(addr));
+	nfs4_put_client(clp);
+	flavour = ntohl(svc_getu32(argv));
+	switch(flavour) {
+		case RPC_AUTH_NULL:
+			if (rqstp->rq_proc != CB_NULL) {
+				*authp = rpc_autherr_tooweak;
+				retval = SVC_DENIED;
+				break;
+			}
+			rqstp->rq_authop = &nfs_callback_auth_null;
+			retval = nfs_callback_null_accept(rqstp, authp);
+			break;
+		case RPC_AUTH_UNIX:
+			/* Eat the authentication flavour */
+			rqstp->rq_authop = &nfs_callback_auth_unix;
+			retval = nfs_callback_unix_accept(rqstp, authp);
+			break;
+		default:
+			/* FIXME: need to add RPCSEC_GSS upcalls */
+#if 0
+			svc_ungetu32(argv);
+			retval = svc_authenticate(rqstp, authp);
+#else
+			*authp = rpc_autherr_rejectedcred;
+			retval = SVC_DENIED;
+#endif
+	}
+	dprintk("%s: flavour %d returning error %d\n", __FUNCTION__, flavour, retval);
+	return retval;
+}
+
+/*
+ * Define NFS4 callback program
+ */
+extern struct svc_version nfs4_callback_version1;
+
+static struct svc_version *nfs4_callback_version[] = {
+	[1] = &nfs4_callback_version1,
+};
+
+static struct svc_stat nfs4_callback_stats;
+
+static struct svc_program nfs4_callback_program = {
+	.pg_prog = NFS4_CALLBACK,			/* RPC service number */
+	.pg_nvers = ARRAY_SIZE(nfs4_callback_version),	/* Number of entries */
+	.pg_vers = nfs4_callback_version,		/* version table */
+	.pg_name = "NFSv4 callback",			/* service name */
+	.pg_class = "nfs",				/* authentication class */
+	.pg_stats = &nfs4_callback_stats,
+	.pg_authenticate = nfs_callback_auth,
+};

fs/nfs/callback.h

+/*
+ * linux/fs/nfs/callback.h
+ *
+ * Copyright (C) 2004 Trond Myklebust
+ *
+ * NFSv4 callback definitions
+ */
+#ifndef __LINUX_FS_NFS_CALLBACK_H
+#define __LINUX_FS_NFS_CALLBACK_H
+
+#define NFS4_CALLBACK 0x40000000
+#define NFS4_CALLBACK_XDRSIZE 2048
+#define NFS4_CALLBACK_BUFSIZE (1024 + NFS4_CALLBACK_XDRSIZE)
+
+enum nfs4_callback_procnum {
+	CB_NULL = 0,
+	CB_COMPOUND = 1,
+};
+
+enum nfs4_callback_opnum {
+	OP_CB_GETATTR = 3,
+	OP_CB_RECALL  = 4,
+	OP_CB_ILLEGAL = 10044,
+};
+
+struct cb_compound_hdr_arg {
+	int taglen;
+	const char *tag;
+	unsigned int callback_ident;
+	unsigned nops;
+};
+
+struct cb_compound_hdr_res {
+	uint32_t *status;
+	int taglen;
+	const char *tag;
+	uint32_t *nops;
+};
+
+struct cb_getattrargs {
+	struct sockaddr_in *addr;
+	struct nfs_fh fh;
+	uint32_t bitmap[2];
+};
+
+struct cb_getattrres {
+	uint32_t status;
+	uint32_t bitmap[2];
+	uint64_t size;
+	uint64_t change_attr;
+	struct timespec ctime;
+	struct timespec mtime;
+};
+
+struct cb_recallargs {
+	struct sockaddr_in *addr;
+	struct nfs_fh fh;
+	nfs4_stateid stateid;
+	uint32_t truncate;
+};
+
+extern unsigned nfs4_callback_getattr(struct cb_getattrargs *args, struct cb_getattrres *res);
+extern unsigned nfs4_callback_recall(struct cb_recallargs *args, void *dummy);
+
+extern int nfs_callback_up(void);
+extern int nfs_callback_down(void);
+
+extern unsigned short nfs_callback_tcpport;
+
+#endif /* __LINUX_FS_NFS_CALLBACK_H */

fs/nfs/callback_proc.c

+/*
+ * linux/fs/nfs/callback_proc.c
+ *
+ * Copyright (C) 2004 Trond Myklebust
+ *
+ * NFSv4 callback procedures
+ */
+#include <linux/config.h>
+#include <linux/nfs4.h>
+#include <linux/nfs_fs.h>
+#include "callback.h"
+#include "delegation.h"
+
+#define NFSDBG_FACILITY NFSDBG_CALLBACK
+ 
+unsigned nfs4_callback_getattr(struct cb_getattrargs *args, struct cb_getattrres *res)
+{
+	struct nfs4_client *clp;
+	struct nfs_delegation *delegation;
+	struct nfs_inode *nfsi;
+	struct inode *inode;
+	
+	res->bitmap[0] = res->bitmap[1] = 0;
+	res->status = htonl(NFS4ERR_BADHANDLE);
+	clp = nfs4_find_client(&args->addr->sin_addr);
+	if (clp == NULL)
+		goto out;
+	inode = nfs_delegation_find_inode(clp, &args->fh);
+	if (inode == NULL)
+		goto out_putclient;
+	nfsi = NFS_I(inode);
+	down_read(&nfsi->rwsem);
+	delegation = nfsi->delegation;
+	if (delegation == NULL || (delegation->type & FMODE_WRITE) == 0)
+		goto out_iput;
+	res->size = i_size_read(inode);
+	res->change_attr = NFS_CHANGE_ATTR(inode);
+	res->ctime = inode->i_ctime;
+	res->mtime = inode->i_mtime;
+	res->bitmap[0] = (FATTR4_WORD0_CHANGE|FATTR4_WORD0_SIZE) &
+		args->bitmap[0];
+	res->bitmap[1] = (FATTR4_WORD1_TIME_METADATA|FATTR4_WORD1_TIME_MODIFY) &
+		args->bitmap[1];
+	res->status = 0;
+out_iput:
+	up_read(&nfsi->rwsem);
+	iput(inode);
+out_putclient:
+	nfs4_put_client(clp);
+out:
+	dprintk("%s: exit with status = %d\n", __FUNCTION__, ntohl(res->status));
+	return res->status;
+}
+
+unsigned nfs4_callback_recall(struct cb_recallargs *args, void *dummy)
+{
+	struct nfs4_client *clp;
+	struct inode *inode;
+	unsigned res;
+	
+	res = htonl(NFS4ERR_BADHANDLE);
+	clp = nfs4_find_client(&args->addr->sin_addr);
+	if (clp == NULL)
+		goto out;
+	inode = nfs_delegation_find_inode(clp, &args->fh);
+	if (inode == NULL)
+		goto out_putclient;
+	/* Set up a helper thread to actually return the delegation */
+	switch(nfs_async_inode_return_delegation(inode, &args->stateid)) {
+		case 0:
+			res = 0;
+			break;
+		case -ENOENT:
+			res = htonl(NFS4ERR_BAD_STATEID);
+			break;
+		default:
+			res = htonl(NFS4ERR_RESOURCE);
+	}
+	iput(inode);
+out_putclient:
+	nfs4_put_client(clp);
+out:
+	dprintk("%s: exit with status = %d\n", __FUNCTION__, ntohl(res));
+	return res;
+}

fs/nfs/callback_xdr.c

+/*
+ * linux/fs/nfs/callback_xdr.c
+ *
+ * Copyright (C) 2004 Trond Myklebust
+ *
+ * NFSv4 callback encode/decode procedures
+ */
+#include <linux/config.h>
+#include <linux/kernel.h>
+#include <linux/sunrpc/svc.h>
+#include <linux/nfs4.h>
+#include <linux/nfs_fs.h>
+#include "callback.h"
+
+#define CB_OP_TAGLEN_MAXSZ	(512)
+#define CB_OP_HDR_RES_MAXSZ	(2 + CB_OP_TAGLEN_MAXSZ)
+#define CB_OP_GETATTR_BITMAP_MAXSZ	(4)
+#define CB_OP_GETATTR_RES_MAXSZ	(CB_OP_HDR_RES_MAXSZ + \
+				CB_OP_GETATTR_BITMAP_MAXSZ + \
+				2 + 2 + 3 + 3)
+#define CB_OP_RECALL_RES_MAXSZ	(CB_OP_HDR_RES_MAXSZ)
+
+#define NFSDBG_FACILITY NFSDBG_CALLBACK
+
+typedef unsigned (*callback_process_op_t)(void *, void *);
+typedef unsigned (*callback_decode_arg_t)(struct svc_rqst *, struct xdr_stream *, void *);
+typedef unsigned (*callback_encode_res_t)(struct svc_rqst *, struct xdr_stream *, void *);
+
+
+struct callback_op {
+	callback_process_op_t process_op;
+	callback_decode_arg_t decode_args;
+	callback_encode_res_t encode_res;
+	long res_maxsize;
+};
+
+static struct callback_op callback_ops[];
+
+static int nfs4_callback_null(struct svc_rqst *rqstp, void *argp, void *resp)
+{
+	return htonl(NFS4_OK);
+}
+
+static int nfs4_decode_void(struct svc_rqst *rqstp, uint32_t *p, void *dummy)
+{
+	return xdr_argsize_check(rqstp, p);
+}
+
+static int nfs4_encode_void(struct svc_rqst *rqstp, uint32_t *p, void *dummy)
+{
+	return xdr_ressize_check(rqstp, p);
+}
+
+static uint32_t *read_buf(struct xdr_stream *xdr, int nbytes)
+{
+	uint32_t *p;
+
+	p = xdr_inline_decode(xdr, nbytes);
+	if (unlikely(p == NULL))
+		printk(KERN_WARNING "NFSv4 callback reply buffer overflowed!\n");
+	return p;
+}
+
+static unsigned decode_string(struct xdr_stream *xdr, unsigned int *len, const char **str)
+{
+	uint32_t *p;
+
+	p = read_buf(xdr, 4);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	*len = ntohl(*p);
+
+	if (*len != 0) {
+		p = read_buf(xdr, *len);
+		if (unlikely(p == NULL))
+			return htonl(NFS4ERR_RESOURCE);
+		*str = (const char *)p;
+	} else
+		*str = NULL;
+
+	return 0;
+}
+
+static unsigned decode_fh(struct xdr_stream *xdr, struct nfs_fh *fh)
+{
+	uint32_t *p;
+
+	p = read_buf(xdr, 4);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	fh->size = ntohl(*p);
+	if (fh->size > NFS4_FHSIZE)
+		return htonl(NFS4ERR_BADHANDLE);
+	p = read_buf(xdr, fh->size);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	memcpy(&fh->data[0], p, fh->size);
+	memset(&fh->data[fh->size], 0, sizeof(fh->data) - fh->size);
+	return 0;
+}
+
+static unsigned decode_bitmap(struct xdr_stream *xdr, uint32_t *bitmap)
+{
+	uint32_t *p;
+	unsigned int attrlen;
+
+	p = read_buf(xdr, 4);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	attrlen = ntohl(*p);
+	p = read_buf(xdr, attrlen << 2);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	if (likely(attrlen > 0))
+		bitmap[0] = ntohl(*p++);
+	if (attrlen > 1)
+		bitmap[1] = ntohl(*p);
+	return 0;
+}
+
+static unsigned decode_stateid(struct xdr_stream *xdr, nfs4_stateid *stateid)
+{
+	uint32_t *p;
+
+	p = read_buf(xdr, 16);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	memcpy(stateid->data, p, 16);
+	return 0;
+}
+
+static unsigned decode_compound_hdr_arg(struct xdr_stream *xdr, struct cb_compound_hdr_arg *hdr)
+{
+	uint32_t *p;
+	unsigned int minor_version;
+	unsigned status;
+
+	status = decode_string(xdr, &hdr->taglen, &hdr->tag);
+	if (unlikely(status != 0))
+		return status;
+	/* We do not like overly long tags! */
+	if (hdr->taglen > CB_OP_TAGLEN_MAXSZ-12 || hdr->taglen < 0) {
+		printk("NFSv4 CALLBACK %s: client sent tag of length %u\n",
+				__FUNCTION__, hdr->taglen);
+		return htonl(NFS4ERR_RESOURCE);
+	}
+	p = read_buf(xdr, 12);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	minor_version = ntohl(*p++);
+	/* Check minor version is zero. */
+	if (minor_version != 0) {
+		printk(KERN_WARNING "%s: NFSv4 server callback with illegal minor version %u!\n",
+				__FUNCTION__, minor_version);
+		return htonl(NFS4ERR_MINOR_VERS_MISMATCH);
+	}
+	hdr->callback_ident = ntohl(*p++);
+	hdr->nops = ntohl(*p);
+	return 0;
+}
+
+static unsigned decode_op_hdr(struct xdr_stream *xdr, unsigned int *op)
+{
+	uint32_t *p;
+	p = read_buf(xdr, 4);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	*op = ntohl(*p);
+	return 0;
+}
+
+static unsigned decode_getattr_args(struct svc_rqst *rqstp, struct xdr_stream *xdr, struct cb_getattrargs *args)
+{
+	unsigned status;
+
+	status = decode_fh(xdr, &args->fh);
+	if (unlikely(status != 0))
+		goto out;
+	args->addr = &rqstp->rq_addr;
+	status = decode_bitmap(xdr, args->bitmap);
+out:
+	dprintk("%s: exit with status = %d\n", __FUNCTION__, status);
+	return status;
+}
+
+static unsigned decode_recall_args(struct svc_rqst *rqstp, struct xdr_stream *xdr, struct cb_recallargs *args)
+{
+	uint32_t *p;
+	unsigned status;
+
+	args->addr = &rqstp->rq_addr;
+	status = decode_stateid(xdr, &args->stateid);
+	if (unlikely(status != 0))
+		goto out;
+	p = read_buf(xdr, 4);
+	if (unlikely(p == NULL)) {
+		status = htonl(NFS4ERR_RESOURCE);
+		goto out;
+	}
+	args->truncate = ntohl(*p);
+	status = decode_fh(xdr, &args->fh);
+out:
+	dprintk("%s: exit with status = %d\n", __FUNCTION__, status);
+	return 0;
+}
+
+static unsigned encode_string(struct xdr_stream *xdr, unsigned int len, const char *str)
+{
+	uint32_t *p;
+
+	p = xdr_reserve_space(xdr, 4 + len);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	xdr_encode_opaque(p, str, len);
+	return 0;
+}
+
+#define CB_SUPPORTED_ATTR0 (FATTR4_WORD0_CHANGE|FATTR4_WORD0_SIZE)
+#define CB_SUPPORTED_ATTR1 (FATTR4_WORD1_TIME_METADATA|FATTR4_WORD1_TIME_MODIFY)
+static unsigned encode_attr_bitmap(struct xdr_stream *xdr, const uint32_t *bitmap, uint32_t **savep)
+{
+	uint32_t bm[2];
+	uint32_t *p;
+
+	bm[0] = htonl(bitmap[0] & CB_SUPPORTED_ATTR0);
+	bm[1] = htonl(bitmap[1] & CB_SUPPORTED_ATTR1);
+	if (bm[1] != 0) {
+		p = xdr_reserve_space(xdr, 16);
+		if (unlikely(p == NULL))
+			return htonl(NFS4ERR_RESOURCE);
+		*p++ = htonl(2);
+		*p++ = bm[0];
+		*p++ = bm[1];
+	} else if (bm[0] != 0) {
+		p = xdr_reserve_space(xdr, 12);
+		if (unlikely(p == NULL))
+			return htonl(NFS4ERR_RESOURCE);
+		*p++ = htonl(1);
+		*p++ = bm[0];
+	} else {
+		p = xdr_reserve_space(xdr, 8);
+		if (unlikely(p == NULL))
+			return htonl(NFS4ERR_RESOURCE);
+		*p++ = htonl(0);
+	}
+	*savep = p;
+	return 0;
+}
+
+static unsigned encode_attr_change(struct xdr_stream *xdr, const uint32_t *bitmap, uint64_t change)
+{
+	uint32_t *p;
+
+	if (!(bitmap[0] & FATTR4_WORD0_CHANGE))
+		return 0;
+	p = xdr_reserve_space(xdr, 8);
+	if (unlikely(p == 0))
+		return htonl(NFS4ERR_RESOURCE);
+	p = xdr_encode_hyper(p, change);
+	return 0;
+}
+
+static unsigned encode_attr_size(struct xdr_stream *xdr, const uint32_t *bitmap, uint64_t size)
+{
+	uint32_t *p;
+
+	if (!(bitmap[0] & FATTR4_WORD0_SIZE))
+		return 0;
+	p = xdr_reserve_space(xdr, 8);
+	if (unlikely(p == 0))
+		return htonl(NFS4ERR_RESOURCE);
+	p = xdr_encode_hyper(p, size);
+	return 0;
+}
+
+static unsigned encode_attr_time(struct xdr_stream *xdr, const struct timespec *time)
+{
+	uint32_t *p;
+
+	p = xdr_reserve_space(xdr, 12);
+	if (unlikely(p == 0))
+		return htonl(NFS4ERR_RESOURCE);
+	p = xdr_encode_hyper(p, time->tv_sec);
+	*p = htonl(time->tv_nsec);
+	return 0;
+}
+
+static unsigned encode_attr_ctime(struct xdr_stream *xdr, const uint32_t *bitmap, const struct timespec *time)
+{
+	if (!(bitmap[1] & FATTR4_WORD1_TIME_METADATA))
+		return 0;
+	return encode_attr_time(xdr,time);
+}
+
+static unsigned encode_attr_mtime(struct xdr_stream *xdr, const uint32_t *bitmap, const struct timespec *time)
+{
+	if (!(bitmap[1] & FATTR4_WORD1_TIME_MODIFY))
+		return 0;
+	return encode_attr_time(xdr,time);
+}
+
+static unsigned encode_compound_hdr_res(struct xdr_stream *xdr, struct cb_compound_hdr_res *hdr)
+{
+	unsigned status;
+
+	hdr->status = xdr_reserve_space(xdr, 4);
+	if (unlikely(hdr->status == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	status = encode_string(xdr, hdr->taglen, hdr->tag);
+	if (unlikely(status != 0))
+		return status;
+	hdr->nops = xdr_reserve_space(xdr, 4);
+	if (unlikely(hdr->nops == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	return 0;
+}
+
+static unsigned encode_op_hdr(struct xdr_stream *xdr, uint32_t op, uint32_t res)
+{
+	uint32_t *p;
+	
+	p = xdr_reserve_space(xdr, 8);
+	if (unlikely(p == NULL))
+		return htonl(NFS4ERR_RESOURCE);
+	*p++ = htonl(op);
+	*p = res;
+	return 0;
+}
+
+static unsigned encode_getattr_res(struct svc_rqst *rqstp, struct xdr_stream *xdr, const struct cb_getattrres *res)
+{
+	uint32_t *savep;
+	unsigned status = res->status;
+	
+	if (unlikely(status != 0))
+		goto out;
+	status = encode_attr_bitmap(xdr, res->bitmap, &savep);
+	if (unlikely(status != 0))
+		goto out;
+	status = encode_attr_change(xdr, res->bitmap, res->change_attr);
+	if (unlikely(status != 0))
+		goto out;
+	status = encode_attr_size(xdr, res->bitmap, res->size);
+	if (unlikely(status != 0))
+		goto out;
+	status = encode_attr_ctime(xdr, res->bitmap, &res->ctime);
+	if (unlikely(status != 0))
+		goto out;
+	status = encode_attr_mtime(xdr, res->bitmap, &res->mtime);
+	*savep = htonl((unsigned int)((char *)xdr->p - (char *)(savep+1)));
+out:
+	dprintk("%s: exit with status = %d\n", __FUNCTION__, status);
+	return status;
+}
+
+static unsigned process_op(struct svc_rqst *rqstp,
+		struct xdr_stream *xdr_in, void *argp,
+		struct xdr_stream *xdr_out, void *resp)
+{
+	struct callback_op *op;
+	unsigned int op_nr;
+	unsigned int status = 0;
+	long maxlen;
+	unsigned res;
+
+	dprintk("%s: start\n", __FUNCTION__);
+	status = decode_op_hdr(xdr_in, &op_nr);
+	if (unlikely(status != 0)) {
+		op_nr = OP_CB_ILLEGAL;
+		op = &callback_ops[0];
+	} else if (unlikely(op_nr != OP_CB_GETATTR && op_nr != OP_CB_RECALL)) {
+		op_nr = OP_CB_ILLEGAL;
+		op = &callback_ops[0];
+		status = htonl(NFS4ERR_OP_ILLEGAL);
+	} else
+		op = &callback_ops[op_nr];
+
+	maxlen = xdr_out->end - xdr_out->p;
+	if (maxlen > 0 && maxlen < PAGE_SIZE) {
+		if (likely(status == 0 && op->decode_args != NULL))
+			status = op->decode_args(rqstp, xdr_in, argp);
+		if (likely(status == 0 && op->process_op != NULL))
+			status = op->process_op(argp, resp);
+	} else
+		status = htonl(NFS4ERR_RESOURCE);
+
+	res = encode_op_hdr(xdr_out, op_nr, status);
+	if (status == 0)
+		status = res;
+	if (op->encode_res != NULL && status == 0)
+		status = op->encode_res(rqstp, xdr_out, resp);
+	dprintk("%s: done, status = %d\n", __FUNCTION__, status);
+	return status;
+}
+
+/*
+ * Decode, process and encode a COMPOUND
+ */
+static int nfs4_callback_compound(struct svc_rqst *rqstp, void *argp, void *resp)
+{
+	struct cb_compound_hdr_arg hdr_arg;
+	struct cb_compound_hdr_res hdr_res;
+	struct xdr_stream xdr_in, xdr_out;
+	uint32_t *p;
+	unsigned int status;
+	unsigned int nops = 1;
+
+	dprintk("%s: start\n", __FUNCTION__);
+
+	xdr_init_decode(&xdr_in, &rqstp->rq_arg, rqstp->rq_arg.head[0].iov_base);
+
+	p = (uint32_t*)((char *)rqstp->rq_res.head[0].iov_base + rqstp->rq_res.head[0].iov_len);
+	rqstp->rq_res.head[0].iov_len = PAGE_SIZE;
+	xdr_init_encode(&xdr_out, &rqstp->rq_res, p);
+
+	decode_compound_hdr_arg(&xdr_in, &hdr_arg);
+	hdr_res.taglen = hdr_arg.taglen;
+	hdr_res.tag = hdr_arg.tag;
+	encode_compound_hdr_res(&xdr_out, &hdr_res);
+
+	for (;;) {
+		status = process_op(rqstp, &xdr_in, argp, &xdr_out, resp);
+		if (status != 0)
+			break;
+		if (nops == hdr_arg.nops)
+			break;
+		nops++;
+	}
+	*hdr_res.status = status;
+	*hdr_res.nops = htonl(nops);
+	dprintk("%s: done, status = %u\n", __FUNCTION__, status);
+	return rpc_success;
+}
+
+/*
+ * Define NFS4 callback COMPOUND ops.
+ */
+static struct callback_op callback_ops[] = {
+	[0] = {
+		.res_maxsize = CB_OP_HDR_RES_MAXSZ,
+	},
+	[OP_CB_GETATTR] = {
+		.process_op = (callback_process_op_t)nfs4_callback_getattr,
+		.decode_args = (callback_decode_arg_t)decode_getattr_args,
+		.encode_res = (callback_encode_res_t)encode_getattr_res,
+		.res_maxsize = CB_OP_GETATTR_RES_MAXSZ,
+	},
+	[OP_CB_RECALL] = {
+		.process_op = (callback_process_op_t)nfs4_callback_recall,
+		.decode_args = (callback_decode_arg_t)decode_recall_args,
+		.res_maxsize = CB_OP_RECALL_RES_MAXSZ,
+	}
+};
+
+/*
+ * Define NFS4 callback procedures
+ */
+static struct svc_procedure nfs4_callback_procedures1[] = {
+	[CB_NULL] = {
+		.pc_func = nfs4_callback_null,
+		.pc_decode = (kxdrproc_t)nfs4_decode_void,
+		.pc_encode = (kxdrproc_t)nfs4_encode_void,
+		.pc_xdrressize = 1,
+	},
+	[CB_COMPOUND] = {
+		.pc_func = nfs4_callback_compound,
+		.pc_encode = (kxdrproc_t)nfs4_encode_void,
+		.pc_argsize = 256,
+		.pc_ressize = 256,
+		.pc_xdrressize = NFS4_CALLBACK_BUFSIZE,
+	}
+};
+
+struct svc_version nfs4_callback_version1 = {
+	.vs_vers = 1,
+	.vs_nproc = ARRAY_SIZE(nfs4_callback_procedures1),
+	.vs_proc = nfs4_callback_procedures1,
+	.vs_xdrsize = NFS4_CALLBACK_XDRSIZE,
+	.vs_dispatch = NULL,
+};
+

fs/nfs/delegation.c

+/*
+ * linux/fs/nfs/delegation.c
+ *
+ * Copyright (C) 2004 Trond Myklebust
+ *
+ * NFS file delegation management
+ *
+ */
+#include <linux/config.h>
+#include <linux/completion.h>
+#include <linux/module.h>
+#include <linux/sched.h>
+#include <linux/spinlock.h>
+
+#include <linux/nfs4.h>
+#include <linux/nfs_fs.h>
+#include <linux/nfs_xdr.h>
+
+#include "delegation.h"
+
+static struct nfs_delegation *nfs_alloc_delegation(void)
+{
+	return (struct nfs_delegation *)kmalloc(sizeof(struct nfs_delegation), GFP_KERNEL);
+}
+
+static void nfs_free_delegation(struct nfs_delegation *delegation)
+{
+	if (delegation->cred)
+		put_rpccred(delegation->cred);
+	kfree(delegation);
+}
+
+static void nfs_delegation_claim_opens(struct inode *inode)
+{
+	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs_open_context *ctx;
+	struct nfs4_state *state;
+
+again:
+	spin_lock(&inode->i_lock);
+	list_for_each_entry(ctx, &nfsi->open_files, list) {
+		state = ctx->state;
+		if (state == NULL)
+			continue;
+		if (!test_bit(NFS_DELEGATED_STATE, &state->flags))
+			continue;
+		get_nfs_open_context(ctx);
+		spin_unlock(&inode->i_lock);
+		if (nfs4_open_delegation_recall(ctx->dentry, state) < 0)
+			return;
+		put_nfs_open_context(ctx);
+		goto again;
+	}
+	spin_unlock(&inode->i_lock);
+}
+
+/*
+ * Set up a delegation on an inode
+ */
+void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res)
+{
+	struct nfs_delegation *delegation = NFS_I(inode)->delegation;
+
+	if (delegation == NULL)
+		return;
+	memcpy(delegation->stateid.data, res->delegation.data,
+			sizeof(delegation->stateid.data));
+	delegation->type = res->delegation_type;
+	delegation->maxsize = res->maxsize;
+	put_rpccred(cred);
+	delegation->cred = get_rpccred(cred);
+	delegation->flags &= ~NFS_DELEGATION_NEED_RECLAIM;
+	NFS_I(inode)->delegation_state = delegation->type;
+	smp_wmb();
+}
+
+/*
+ * Set up a delegation on an inode
+ */
+int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res)
+{
+	struct nfs4_client *clp = NFS_SERVER(inode)->nfs4_state;
+	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs_delegation *delegation;
+	int status = 0;
+
+	delegation = nfs_alloc_delegation();
+	if (delegation == NULL)
+		return -ENOMEM;
+	memcpy(delegation->stateid.data, res->delegation.data,
+			sizeof(delegation->stateid.data));
+	delegation->type = res->delegation_type;
+	delegation->maxsize = res->maxsize;
+	delegation->cred = get_rpccred(cred);
+	delegation->inode = inode;
+
+	spin_lock(&clp->cl_lock);
+	if (nfsi->delegation == NULL) {
+		list_add(&delegation->super_list, &clp->cl_delegations);
+		nfsi->delegation = delegation;
+		nfsi->delegation_state = delegation->type;
+		delegation = NULL;
+	} else {
+		if (memcmp(&delegation->stateid, &nfsi->delegation->stateid,
+					sizeof(delegation->stateid)) != 0 ||
+				delegation->type != nfsi->delegation->type) {
+			printk("%s: server %u.%u.%u.%u, handed out a duplicate delegation!\n",
+					__FUNCTION__, NIPQUAD(clp->cl_addr));
+			status = -EIO;
+		}
+	}
+	spin_unlock(&clp->cl_lock);
+	if (delegation != NULL)
+		kfree(delegation);
+	return status;
+}
+
+static int nfs_do_return_delegation(struct inode *inode, struct nfs_delegation *delegation)
+{
+	int res = 0;
+
+	__nfs_revalidate_inode(NFS_SERVER(inode), inode);
+
+	res = nfs4_proc_delegreturn(inode, delegation->cred, &delegation->stateid);
+	nfs_free_delegation(delegation);
+	return res;
+}
+
+/* Sync all data to disk upon delegation return */
+static void nfs_msync_inode(struct inode *inode)
+{
+	filemap_fdatawrite(inode->i_mapping);
+	nfs_wb_all(inode);
+	filemap_fdatawait(inode->i_mapping);
+}
+
+/*
+ * Basic procedure for returning a delegation to the server
+ */
+int nfs_inode_return_delegation(struct inode *inode)
+{
+	struct nfs4_client *clp = NFS_SERVER(inode)->nfs4_state;
+	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs_delegation *delegation;
+	int res = 0;
+
+	nfs_msync_inode(inode);
+	down_read(&clp->cl_sem);
+	/* Guard against new delegated open calls */
+	down_write(&nfsi->rwsem);
+	spin_lock(&clp->cl_lock);
+	delegation = nfsi->delegation;
+	if (delegation != NULL) {
+		list_del_init(&delegation->super_list);
+		nfsi->delegation = NULL;
+		nfsi->delegation_state = 0;
+	}
+	spin_unlock(&clp->cl_lock);
+	nfs_delegation_claim_opens(inode);
+	up_write(&nfsi->rwsem);
+	up_read(&clp->cl_sem);
+	nfs_msync_inode(inode);
+
+	if (delegation != NULL)
+		res = nfs_do_return_delegation(inode, delegation);
+	return res;
+}
+
+/*
+ * Return all delegations associated to a super block
+ */
+void nfs_return_all_delegations(struct super_block *sb)
+{
+	struct nfs4_client *clp = NFS_SB(sb)->nfs4_state;
+	struct nfs_delegation *delegation;
+	struct inode *inode;
+
+	if (clp == NULL)
+		return;
+restart:
+	spin_lock(&clp->cl_lock);
+	list_for_each_entry(delegation, &clp->cl_delegations, super_list) {
+		if (delegation->inode->i_sb != sb)
+			continue;
+		inode = igrab(delegation->inode);
+		if (inode == NULL)
+			continue;
+		spin_unlock(&clp->cl_lock);
+		nfs_inode_return_delegation(inode);
+		iput(inode);
+		goto restart;
+	}
+	spin_unlock(&clp->cl_lock);
+}
+
+/*
+ * Return all delegations following an NFS4ERR_CB_PATH_DOWN error.
+ */
+void nfs_handle_cb_pathdown(struct nfs4_client *clp)
+{
+	struct nfs_delegation *delegation;
+	struct inode *inode;
+
+	if (clp == NULL)
+		return;
+restart:
+	spin_lock(&clp->cl_lock);
+	list_for_each_entry(delegation, &clp->cl_delegations, super_list) {
+		inode = igrab(delegation->inode);
+		if (inode == NULL)
+			continue;
+		spin_unlock(&clp->cl_lock);
+		nfs_inode_return_delegation(inode);
+		iput(inode);
+		goto restart;
+	}
+	spin_unlock(&clp->cl_lock);
+}
+
+struct recall_threadargs {
+	struct inode *inode;
+	struct nfs4_client *clp;
+	const nfs4_stateid *stateid;
+
+	struct completion started;
+	int result;
+};
+
+static int recall_thread(void *data)
+{
+	struct recall_threadargs *args = (struct recall_threadargs *)data;
+	struct inode *inode = igrab(args->inode);
+	struct nfs4_client *clp = NFS_SERVER(inode)->nfs4_state;
+	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs_delegation *delegation;
+
+	daemonize("nfsv4-delegreturn");
+
+	nfs_msync_inode(inode);
+	down_read(&clp->cl_sem);
+	down_write(&nfsi->rwsem);
+	spin_lock(&clp->cl_lock);
+	delegation = nfsi->delegation;
+	if (delegation != NULL && memcmp(delegation->stateid.data,
+				args->stateid->data,
+				sizeof(delegation->stateid.data)) == 0) {
+		list_del_init(&delegation->super_list);
+		nfsi->delegation = NULL;
+		nfsi->delegation_state = 0;
+		args->result = 0;
+	} else {
+		delegation = NULL;
+		args->result = -ENOENT;
+	}
+	spin_unlock(&clp->cl_lock);
+	complete(&args->started);
+	nfs_delegation_claim_opens(inode);
+	up_write(&nfsi->rwsem);
+	up_read(&clp->cl_sem);
+	nfs_msync_inode(inode);
+
+	if (delegation != NULL)
+		nfs_do_return_delegation(inode, delegation);
+	iput(inode);
+	module_put_and_exit(0);
+}
+
+/*
+ * Asynchronous delegation recall!
+ */
+int nfs_async_inode_return_delegation(struct inode *inode, const nfs4_stateid *stateid)
+{
+	struct recall_threadargs data = {
+		.inode = inode,
+		.stateid = stateid,
+	};
+	int status;
+
+	init_completion(&data.started);
+	__module_get(THIS_MODULE);
+	status = kernel_thread(recall_thread, &data, CLONE_KERNEL);
+	if (status < 0)
+		goto out_module_put;
+	wait_for_completion(&data.started);
+	return data.result;
+out_module_put:
+	module_put(THIS_MODULE);
+	return status;
+}
+
+/*
+ * Retrieve the inode associated with a delegation
+ */
+struct inode *nfs_delegation_find_inode(struct nfs4_client *clp, const struct nfs_fh *fhandle)
+{
+	struct nfs_delegation *delegation;
+	struct inode *res = NULL;
+	spin_lock(&clp->cl_lock);
+	list_for_each_entry(delegation, &clp->cl_delegations, super_list) {
+		if (nfs_compare_fh(fhandle, &NFS_I(delegation->inode)->fh) == 0) {
+			res = igrab(delegation->inode);
+			break;
+		}
+	}
+	spin_unlock(&clp->cl_lock);
+	return res;
+}
+
+/*
+ * Mark all delegations as needing to be reclaimed
+ */
+void nfs_delegation_mark_reclaim(struct nfs4_client *clp)
+{
+	struct nfs_delegation *delegation;
+	spin_lock(&clp->cl_lock);
+	list_for_each_entry(delegation, &clp->cl_delegations, super_list)
+		delegation->flags |= NFS_DELEGATION_NEED_RECLAIM;
+	spin_unlock(&clp->cl_lock);
+}
+
+/*
+ * Reap all unclaimed delegations after reboot recovery is done
+ */
+void nfs_delegation_reap_unclaimed(struct nfs4_client *clp)
+{
+	struct nfs_delegation *delegation, *n;
+	LIST_HEAD(head);
+	spin_lock(&clp->cl_lock);
+	list_for_each_entry_safe(delegation, n, &clp->cl_delegations, super_list) {
+		if ((delegation->flags & NFS_DELEGATION_NEED_RECLAIM) == 0)
+			continue;
+		list_move(&delegation->super_list, &head);
+		NFS_I(delegation->inode)->delegation = NULL;
+		NFS_I(delegation->inode)->delegation_state = 0;
+	}
+	spin_unlock(&clp->cl_lock);
+	while(!list_empty(&head)) {
+		delegation = list_entry(head.next, struct nfs_delegation, super_list);
+		list_del(&delegation->super_list);
+		nfs_free_delegation(delegation);
+	}
+}

fs/nfs/delegation.h

+/*
+ * linux/fs/nfs/delegation.h
+ *
+ * Copyright (c) Trond Myklebust
+ *
+ * Definitions pertaining to NFS delegated files
+ */
+#ifndef FS_NFS_DELEGATION_H
+#define FS_NFS_DELEGATION_H
+
+#if defined(CONFIG_NFS_V4)
+/*
+ * NFSv4 delegation
+ */
+struct nfs_delegation {
+	struct list_head super_list;
+	struct rpc_cred *cred;
+	struct inode *inode;
+	nfs4_stateid stateid;
+	int type;
+#define NFS_DELEGATION_NEED_RECLAIM 1
+	long flags;
+	loff_t maxsize;
+};
+
+int nfs_inode_set_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res);
+void nfs_inode_reclaim_delegation(struct inode *inode, struct rpc_cred *cred, struct nfs_openres *res);
+int nfs_inode_return_delegation(struct inode *inode);
+int nfs_async_inode_return_delegation(struct inode *inode, const nfs4_stateid *stateid);
+
+struct inode *nfs_delegation_find_inode(struct nfs4_client *clp, const struct nfs_fh *fhandle);
+void nfs_return_all_delegations(struct super_block *sb);
+void nfs_handle_cb_pathdown(struct nfs4_client *clp);
+
+void nfs_delegation_mark_reclaim(struct nfs4_client *clp);
+void nfs_delegation_reap_unclaimed(struct nfs4_client *clp);
+
+/* NFSv4 delegation-related procedures */
+int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid);
+int nfs4_open_delegation_recall(struct dentry *dentry, struct nfs4_state *state);
+
+static inline int nfs_have_delegation(struct inode *inode, int flags)
+{
+	flags &= FMODE_READ|FMODE_WRITE;
+	smp_rmb();
+	if ((NFS_I(inode)->delegation_state & flags) == flags)
+		return 1;
+	return 0;
+}
+#else
+static inline int nfs_have_delegation(struct inode *inode, int flags)
+{
+	return 0;
+}
+#endif
+
+#endif

fs/nfs/dir.c

@@ -32,6 +32,8 @@
 #include <linux/smp_lock.h>
 #include <linux/namei.h>
 
+#include "delegation.h"
+
 #define NFS_PARANOIA 1
 /* #define NFS_DEBUG_VERBOSE 1 */
 
@@ -610,7 +612,7 @@ static int nfs_lookup_revalidate(struct dentry * dentry, struct nameidata *nd)
 	verifier = nfs_save_change_attribute(dir);
 	error = nfs_cached_lookup(dir, dentry, &fhandle, &fattr);
 	if (!error) {
-		if (memcmp(NFS_FH(inode), &fhandle, sizeof(struct nfs_fh))!= 0)
+		if (nfs_compare_fh(NFS_FH(inode), &fhandle))
 			goto out_bad;
 		if (nfs_lookup_verify_inode(inode, isopen))
 			goto out_zap_parent;
@@ -623,7 +625,7 @@ static int nfs_lookup_revalidate(struct dentry * dentry, struct nameidata *nd)
 	error = NFS_PROTO(dir)->lookup(dir, &dentry->d_name, &fhandle, &fattr);
 	if (error)
 		goto out_bad;
-	if (memcmp(NFS_FH(inode), &fhandle, sizeof(struct nfs_fh))!= 0)
+	if (nfs_compare_fh(NFS_FH(inode), &fhandle))
 		goto out_bad;
 	if ((error = nfs_refresh_inode(inode, &fattr)) != 0)
 		goto out_bad;
@@ -850,22 +852,22 @@ static int nfs_open_revalidate(struct dentry *dentry, struct nameidata *nd)
 	unsigned long verifier;
 	int openflags, ret = 0;
 
-	/* NFS only supports OPEN for regular files */
-	if (inode && !S_ISREG(inode->i_mode))
-		goto no_open;
 	parent = dget_parent(dentry);
 	dir = parent->d_inode;
 	if (!is_atomic_open(dir, nd))
 		goto no_open;
+	/* We can't create new files in nfs_open_revalidate(), so we
+	 * optimize away revalidation of negative dentries.
+	 */
+	if (inode == NULL)
+		goto out;
+	/* NFS only supports OPEN on regular files */
+	if (!S_ISREG(inode->i_mode))
+		goto no_open;
 	openflags = nd->intent.open.flags;
-	if (openflags & O_CREAT) {
-		/* If this is a negative dentry, just drop it */
-		if (!inode)
-			goto out;
-		/* If this is exclusive open, just revalidate */
-		if (openflags & O_EXCL)
-			goto no_open;
-	}
+	/* We cannot do exclusive creation on a positive dentry */
+	if ((openflags & (O_CREAT|O_EXCL)) == (O_CREAT|O_EXCL))
+		goto no_open;
 	/* We can't create new files, or truncate existing ones here */
 	openflags &= ~(O_CREAT|O_TRUNC);
 
@@ -887,6 +889,8 @@ static int nfs_open_revalidate(struct dentry *dentry, struct nameidata *nd)
 	return ret;
 no_open:
 	dput(parent);
+	if (inode != NULL && nfs_have_delegation(inode, FMODE_READ))
+		return 1;
 	return nfs_lookup_revalidate(dentry, nd);
 }
 #endif /* CONFIG_NFSV4 */
@@ -982,12 +986,18 @@ 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);
@@ -1299,19 +1309,6 @@ nfs_symlink(struct inode *dir, struct dentry *dentry, const char *symname)
 	dfprintk(VFS, "NFS: symlink(%s/%ld, %s, %s)\n", dir->i_sb->s_id,
 		dir->i_ino, dentry->d_name.name, symname);
 
-	error = -ENAMETOOLONG;
-	switch (NFS_PROTO(dir)->version) {
-		case 2:
-			if (strlen(symname) > NFS2_MAXPATHLEN)
-				goto out;
-			break;
-		case 3:
-			if (strlen(symname) > NFS3_MAXPATHLEN)
-				goto out;
-		default:
-			break;
-	}
-
 #ifdef NFS_PARANOIA
 if (dentry->d_inode)
 printk("nfs_proc_symlink: %s/%s not negative!\n",
@@ -1341,8 +1338,6 @@ dentry->d_parent->d_name.name, dentry->d_name.name);
 		d_drop(dentry);
 	}
 	unlock_kernel();
-
-out:
 	return error;
 }
 
@@ -1498,10 +1493,56 @@ static int nfs_rename(struct inode *old_dir, struct dentry *old_dentry,
 	return error;
 }
 
-int
-nfs_permission(struct inode *inode, int mask, struct nameidata *nd)
+int nfs_access_get_cached(struct inode *inode, struct rpc_cred *cred, struct nfs_access_entry *res)
+{
+	struct nfs_access_entry *cache = &NFS_I(inode)->cache_access;
+
+	if (cache->cred != cred
+			|| time_after(jiffies, cache->jiffies + NFS_ATTRTIMEO(inode))
+			|| (NFS_FLAGS(inode) & NFS_INO_INVALID_ATTR))
+		return -ENOENT;
+	memcpy(res, cache, sizeof(*res));
+	return 0;
+}
+
+void nfs_access_add_cache(struct inode *inode, struct nfs_access_entry *set)
+{
+	struct nfs_access_entry *cache = &NFS_I(inode)->cache_access;
+
+	if (cache->cred != set->cred) {
+		if (cache->cred)
+			put_rpccred(cache->cred);
+		cache->cred = get_rpccred(set->cred);
+	}
+	cache->jiffies = set->jiffies;
+	cache->mask = set->mask;
+}
+
+static int nfs_do_access(struct inode *inode, struct rpc_cred *cred, int mask)
+{
+	struct nfs_access_entry cache;
+	int status;
+
+	status = nfs_access_get_cached(inode, cred, &cache);
+	if (status == 0)
+		goto out;
+
+	/* Be clever: ask server to check for all possible rights */
+	cache.mask = MAY_EXEC | MAY_WRITE | MAY_READ;
+	cache.cred = cred;
+	cache.jiffies = jiffies;
+	status = NFS_PROTO(inode)->access(inode, &cache);
+	if (status != 0)
+		return status;
+	nfs_access_add_cache(inode, &cache);
+out:
+	if ((cache.mask & mask) == mask)
+		return 0;
+	return -EACCES;
+}
+
+int nfs_permission(struct inode *inode, int mask, struct nameidata *nd)
 {
-	struct nfs_access_cache *cache = &NFS_I(inode)->cache_access;
 	struct rpc_cred *cred;
 	int mode = inode->i_mode;
 	int res;
@@ -1542,24 +1583,7 @@ nfs_permission(struct inode *inode, int mask, struct nameidata *nd)
 		goto out_notsup;
 
 	cred = rpcauth_lookupcred(NFS_CLIENT(inode)->cl_auth, 0);
-	if (cache->cred == cred
-	    && time_before(jiffies, cache->jiffies + NFS_ATTRTIMEO(inode))
-	    && !(NFS_FLAGS(inode) & NFS_INO_INVALID_ATTR)) {
-		if (!(res = cache->err)) {
-			/* Is the mask a subset of an accepted mask? */
-			if ((cache->mask & mask) == mask)
-				goto out;
-		} else {
-			/* ...or is it a superset of a rejected mask? */
-			if ((cache->mask & mask) == cache->mask)
-				goto out;
-		}
-	}
-
-	res = NFS_PROTO(inode)->access(inode, cred, mask);
-	if (!res || res == -EACCES)
-		goto add_cache;
-out:
+	res = nfs_do_access(inode, cred, mask);
 	put_rpccred(cred);
 	unlock_kernel();
 	return res;
@@ -1568,15 +1592,6 @@ nfs_permission(struct inode *inode, int mask, struct nameidata *nd)
 	res = vfs_permission(inode, mask);
 	unlock_kernel();
 	return res;
-add_cache:
-	cache->jiffies = jiffies;
-	if (cache->cred)
-		put_rpccred(cache->cred);
-	cache->cred = cred;
-	cache->mask = mask;
-	cache->err = res;
-	unlock_kernel();
-	return res;
 }
 
 /*

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

@@ -31,6 +31,8 @@
 #include <asm/uaccess.h>
 #include <asm/system.h>
 
+#include "delegation.h"
+
 #define NFSDBG_FACILITY		NFSDBG_FILE
 
 static int nfs_file_open(struct inode *, struct file *);
@@ -113,6 +115,7 @@ 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;
 
@@ -124,9 +127,9 @@ 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;
-		if (!status)
+		status = ctx->error;
+		ctx->error = 0;
+		if (!status && !nfs_have_delegation(inode, FMODE_READ))
 			__nfs_revalidate_inode(NFS_SERVER(inode), inode);
 	}
 	unlock_kernel();
@@ -197,6 +200,7 @@ 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 +209,8 @@ 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

@@ -39,6 +39,8 @@
 #include <asm/system.h>
 #include <asm/uaccess.h>
 
+#include "delegation.h"
+
 #define NFSDBG_FACILITY		NFSDBG_VFS
 #define NFS_PARANOIA 1
 
@@ -57,7 +59,6 @@ static struct inode *nfs_alloc_inode(struct super_block *sb);
 static void nfs_destroy_inode(struct inode *);
 static void nfs_write_inode(struct inode *,int);
 static void nfs_delete_inode(struct inode *);
-static void nfs_put_super(struct super_block *);
 static void nfs_clear_inode(struct inode *);
 static void nfs_umount_begin(struct super_block *);
 static int  nfs_statfs(struct super_block *, struct kstatfs *);
@@ -68,7 +69,6 @@ static struct super_operations nfs_sops = {
 	.destroy_inode	= nfs_destroy_inode,
 	.write_inode	= nfs_write_inode,
 	.delete_inode	= nfs_delete_inode,
-	.put_super	= nfs_put_super,
 	.statfs		= nfs_statfs,
 	.clear_inode	= nfs_clear_inode,
 	.umount_begin	= nfs_umount_begin,
@@ -123,8 +123,9 @@ 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);
@@ -141,37 +142,16 @@ 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);
 	BUG_ON(atomic_read(&nfsi->data_updates) != 0);
 }
 
-void
-nfs_put_super(struct super_block *sb)
-{
-	struct nfs_server *server = NFS_SB(sb);
-
-	nfs4_renewd_prepare_shutdown(server);
-
-	if (server->client != NULL)
-		rpc_shutdown_client(server->client);
-	if (server->client_sys != NULL)
-		rpc_shutdown_client(server->client_sys);
-
-	if (!(server->flags & NFS_MOUNT_NONLM))
-		lockd_down();	/* release rpc.lockd */
-	rpciod_down();		/* release rpciod */
-
-	destroy_nfsv4_state(server);
-
-	kfree(server->hostname);
-}
-
 void
 nfs_umount_begin(struct super_block *sb)
 {
@@ -293,14 +273,6 @@ nfs_sb_init(struct super_block *sb, rpc_authflavor_t authflavor)
 		server->rsize = nfs_block_size(fsinfo.rtpref, NULL);
 	if (server->wsize == 0)
 		server->wsize = nfs_block_size(fsinfo.wtpref, NULL);
-	if (sb->s_blocksize == 0) {
-		if (fsinfo.wtmult == 0) {
-			sb->s_blocksize = 512;
-			sb->s_blocksize_bits = 9;
-		} else
-			sb->s_blocksize = nfs_block_bits(fsinfo.wtmult,
-							 &sb->s_blocksize_bits);
-	}
 
 	if (fsinfo.rtmax >= 512 && server->rsize > fsinfo.rtmax)
 		server->rsize = nfs_block_size(fsinfo.rtmax, NULL);
@@ -319,6 +291,11 @@ nfs_sb_init(struct super_block *sb, rpc_authflavor_t authflavor)
                 server->wsize = server->wpages << PAGE_CACHE_SHIFT;
 	}
 
+	if (sb->s_blocksize == 0)
+		sb->s_blocksize = nfs_block_bits(server->wsize,
+							 &sb->s_blocksize_bits);
+	server->wtmult = nfs_block_bits(fsinfo.wtmult, NULL);
+
 	server->dtsize = nfs_block_size(fsinfo.dtpref, NULL);
 	if (server->dtsize > PAGE_CACHE_SIZE)
 		server->dtsize = PAGE_CACHE_SIZE;
@@ -405,7 +382,6 @@ static int
 nfs_fill_super(struct super_block *sb, struct nfs_mount_data *data, int silent)
 {
 	struct nfs_server	*server;
-	int			err = -EIO;
 	rpc_authflavor_t	authflavor;
 
 	server           = NFS_SB(sb);
@@ -424,10 +400,14 @@ nfs_fill_super(struct super_block *sb, struct nfs_mount_data *data, int silent)
 	server->acdirmin = data->acdirmin*HZ;
 	server->acdirmax = data->acdirmax*HZ;
 
+	/* Start lockd here, before we might error out */
+	if (!(server->flags & NFS_MOUNT_NONLM))
+		lockd_up();
+
 	server->namelen  = data->namlen;
 	server->hostname = kmalloc(strlen(data->hostname) + 1, GFP_KERNEL);
 	if (!server->hostname)
-		goto out_fail;
+		return -ENOMEM;
 	strcpy(server->hostname, data->hostname);
 
 	/* Check NFS protocol revision and initialize RPC op vector
@@ -438,11 +418,11 @@ nfs_fill_super(struct super_block *sb, struct nfs_mount_data *data, int silent)
 		server->caps |= NFS_CAP_READDIRPLUS;
 		if (data->version < 4) {
 			printk(KERN_NOTICE "NFS: NFSv3 not supported by mount program.\n");
-			goto out_fail;
+			return -EIO;
 		}
 #else
 		printk(KERN_NOTICE "NFS: NFSv3 not supported.\n");
-		goto out_fail;
+		return -EIO;
 #endif
 	} else {
 		server->rpc_ops = &nfs_v2_clientops;
@@ -457,30 +437,19 @@ nfs_fill_super(struct super_block *sb, struct nfs_mount_data *data, int silent)
 	/* Create RPC client handles */
 	server->client = nfs_create_client(server, data);
 	if (IS_ERR(server->client))
-		goto out_fail;
+		return PTR_ERR(server->client);
 	/* RFC 2623, sec 2.3.2 */
 	if (authflavor != RPC_AUTH_UNIX) {
 		server->client_sys = rpc_clone_client(server->client);
-		if (server->client_sys == NULL)
-			goto out_shutdown;
+		if (IS_ERR(server->client_sys))
+			return PTR_ERR(server->client_sys);
 		if (!rpcauth_create(RPC_AUTH_UNIX, server->client_sys))
-			goto out_shutdown;
+			return -ENOMEM;
 	} else {
 		atomic_inc(&server->client->cl_count);
 		server->client_sys = server->client;
 	}
 
-	/* Fire up rpciod if not yet running */
-	if (rpciod_up() != 0) {
-		printk(KERN_WARNING "NFS: couldn't start rpciod!\n");
-		goto out_shutdown;
-	}
-
-	sb->s_op = &nfs_sops;
-	err = nfs_sb_init(sb, authflavor);
-	if (err != 0)
-		goto out_noinit;
-
 	if (server->flags & NFS_MOUNT_VER3) {
 		if (server->namelen == 0 || server->namelen > NFS3_MAXNAMLEN)
 			server->namelen = NFS3_MAXNAMLEN;
@@ -489,21 +458,8 @@ nfs_fill_super(struct super_block *sb, struct nfs_mount_data *data, int silent)
 			server->namelen = NFS2_MAXNAMLEN;
 	}
 
-	/* Check whether to start the lockd process */
-	if (!(server->flags & NFS_MOUNT_NONLM))
-		lockd_up();
-	return 0;
-out_noinit:
-	rpciod_down();
-out_shutdown:
-	if (server->client)
-		rpc_shutdown_client(server->client);
-	if (server->client_sys)
-		rpc_shutdown_client(server->client_sys);
-out_fail:
-	if (server->hostname)
-		kfree(server->hostname);
-	return err;
+	sb->s_op = &nfs_sops;
+	return nfs_sb_init(sb, authflavor);
 }
 
 static int
@@ -526,6 +482,7 @@ nfs_statfs(struct super_block *sb, struct kstatfs *buf)
 	if (error < 0)
 		goto out_err;
 
+	buf->f_frsize = server->wtmult;
 	buf->f_bsize = sb->s_blocksize;
 	blockbits = sb->s_blocksize_bits;
 	blockres = (1 << blockbits) - 1;
@@ -642,7 +599,7 @@ nfs_find_actor(struct inode *inode, void *opaque)
 
 	if (NFS_FILEID(inode) != fattr->fileid)
 		return 0;
-	if (memcmp(NFS_FH(inode), fh, sizeof(struct nfs_fh)) != 0)
+	if (nfs_compare_fh(NFS_FH(inode), fh))
 		return 0;
 	if (is_bad_inode(inode))
 		return 0;
@@ -653,11 +610,10 @@ static int
 nfs_init_locked(struct inode *inode, void *opaque)
 {
 	struct nfs_find_desc	*desc = (struct nfs_find_desc *)opaque;
-	struct nfs_fh		*fh = desc->fh;
 	struct nfs_fattr	*fattr = desc->fattr;
 
 	NFS_FILEID(inode) = fattr->fileid;
-	memcpy(NFS_FH(inode), fh, sizeof(struct nfs_fh));
+	nfs_copy_fh(NFS_FH(inode), desc->fh);
 	return 0;
 }
 
@@ -859,53 +815,114 @@ 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;
 }
 
@@ -946,7 +963,7 @@ __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,
@@ -1002,6 +1019,30 @@ __nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
 	return status;
 }
 
+int nfs_attribute_timeout(struct inode *inode)
+{
+	struct nfs_inode *nfsi = NFS_I(inode);
+
+	if (nfs_have_delegation(inode, FMODE_READ))
+		return 0;
+	return time_after(jiffies, nfsi->read_cache_jiffies+nfsi->attrtimeo);
+}
+
+/**
+ * nfs_revalidate_inode - Revalidate the inode attributes
+ * @server - pointer to nfs_server struct
+ * @inode - pointer to inode struct
+ *
+ * Updates inode attribute information by retrieving the data from the server.
+ */
+int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
+{
+	if (!(NFS_FLAGS(inode) & (NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA))
+			&& !nfs_attribute_timeout(inode))
+		return NFS_STALE(inode) ? -ESTALE : 0;
+	return __nfs_revalidate_inode(server, inode);
+}
+
 /**
  * nfs_begin_data_update
  * @inode - pointer to inode
@@ -1023,11 +1064,13 @@ void nfs_end_data_update(struct inode *inode)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 
-	/* Mark the attribute cache for revalidation */
-	nfsi->flags |= NFS_INO_INVALID_ATTR;
-	/* Directories and symlinks: invalidate page cache too */
-	if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode))
-		nfsi->flags |= NFS_INO_INVALID_DATA;
+	if (!nfs_have_delegation(inode, FMODE_READ)) {
+		/* Mark the attribute cache for revalidation */
+		nfsi->flags |= NFS_INO_INVALID_ATTR;
+		/* Directories and symlinks: invalidate page cache too */
+		if (S_ISDIR(inode->i_mode) || S_ISLNK(inode->i_mode))
+			nfsi->flags |= NFS_INO_INVALID_DATA;
+	}
 	nfsi->cache_change_attribute ++;
 	atomic_dec(&nfsi->data_updates);
 }
@@ -1068,6 +1111,10 @@ int nfs_refresh_inode(struct inode *inode, struct nfs_fattr *fattr)
 	loff_t cur_size, new_isize;
 	int data_unstable;
 
+	/* Do we hold a delegation? */
+	if (nfs_have_delegation(inode, FMODE_READ))
+		return 0;
+
 	/* Are we in the process of updating data on the server? */
 	data_unstable = nfs_caches_unstable(inode);
 
@@ -1265,7 +1312,8 @@ static int nfs_update_inode(struct inode *inode, struct nfs_fattr *fattr, unsign
 	if (!(S_ISREG(inode->i_mode) || S_ISDIR(inode->i_mode)
 				|| S_ISLNK(inode->i_mode)))
 		invalid &= ~NFS_INO_INVALID_DATA;
-	nfsi->flags |= invalid;
+	if (!nfs_have_delegation(inode, FMODE_READ))
+		nfsi->flags |= invalid;
 
 	return 0;
  out_changed:
@@ -1305,7 +1353,7 @@ static int nfs_compare_super(struct super_block *sb, void *data)
 		return 0;
 	if (old->addr.sin_port != server->addr.sin_port)
 		return 0;
-	return !memcmp(&old->fh, &server->fh, sizeof(struct nfs_fh));
+	return !nfs_compare_fh(&old->fh, &server->fh);
 }
 
 static struct super_block *nfs_get_sb(struct file_system_type *fs_type,
@@ -1330,9 +1378,7 @@ static struct super_block *nfs_get_sb(struct file_system_type *fs_type,
 	init_nfsv4_state(server);
 
 	root = &server->fh;
-	memcpy(root, &data->root, sizeof(*root));
-	if (root->size < sizeof(root->data))
-		memset(root->data+root->size, 0, sizeof(root->data)-root->size);
+	nfs_copy_fh(root, (struct nfs_fh *) &data->root);
 
 	if (data->version != NFS_MOUNT_VERSION) {
 		printk("nfs warning: mount version %s than kernel\n",
@@ -1343,7 +1389,6 @@ static struct super_block *nfs_get_sb(struct file_system_type *fs_type,
 			data->bsize  = 0;
 		if (data->version < 4) {
 			data->flags &= ~NFS_MOUNT_VER3;
-			memset(root, 0, sizeof(*root));
 			root->size = NFS2_FHSIZE;
 			memcpy(root->data, data->old_root.data, NFS2_FHSIZE);
 		}
@@ -1373,6 +1418,13 @@ static struct super_block *nfs_get_sb(struct file_system_type *fs_type,
 
 	s->s_flags = flags;
 
+	/* Fire up rpciod if not yet running */
+	if (rpciod_up() != 0) {
+		printk(KERN_WARNING "NFS: couldn't start rpciod!\n");
+		kfree(server);
+		return ERR_PTR(-EIO);
+	}
+
 	error = nfs_fill_super(s, data, flags & MS_VERBOSE ? 1 : 0);
 	if (error) {
 		up_write(&s->s_umount);
@@ -1386,7 +1438,25 @@ static struct super_block *nfs_get_sb(struct file_system_type *fs_type,
 static void nfs_kill_super(struct super_block *s)
 {
 	struct nfs_server *server = NFS_SB(s);
+
 	kill_anon_super(s);
+
+	nfs4_renewd_prepare_shutdown(server);
+
+	if (server->client != NULL && !IS_ERR(server->client))
+		rpc_shutdown_client(server->client);
+	if (server->client_sys != NULL && !IS_ERR(server->client_sys))
+		rpc_shutdown_client(server->client_sys);
+
+	if (!(server->flags & NFS_MOUNT_NONLM))
+		lockd_down();	/* release rpc.lockd */
+
+	rpciod_down();		/* release rpciod */
+
+	destroy_nfsv4_state(server);
+
+	if (server->hostname != NULL)
+		kfree(server->hostname);
 	kfree(server);
 }
 
@@ -1402,12 +1472,12 @@ static struct file_system_type nfs_fs_type = {
 
 static void nfs4_clear_inode(struct inode *);
 
+
 static struct super_operations nfs4_sops = { 
 	.alloc_inode	= nfs_alloc_inode,
 	.destroy_inode	= nfs_destroy_inode,
 	.write_inode	= nfs_write_inode,
 	.delete_inode	= nfs_delete_inode,
-	.put_super	= nfs_put_super,
 	.statfs		= nfs_statfs,
 	.clear_inode	= nfs4_clear_inode,
 	.umount_begin	= nfs_umount_begin,
@@ -1423,6 +1493,12 @@ static void nfs4_clear_inode(struct inode *inode)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 
+	/* If we are holding a delegation, return it! */
+	if (nfsi->delegation != NULL)
+		nfs_inode_return_delegation(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;
 		
@@ -1437,8 +1513,6 @@ 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);
 }
 
 
@@ -1498,7 +1572,7 @@ static int nfs4_fill_super(struct super_block *sb, struct nfs4_mount_data *data,
 	clp = nfs4_get_client(&server->addr.sin_addr);
 	if (!clp) {
 		printk(KERN_WARNING "NFS: failed to create NFS4 client.\n");
-		goto out_fail;
+		return -EIO;
 	}
 
 	/* Now create transport and client */
@@ -1536,8 +1610,13 @@ 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))
@@ -1547,45 +1626,29 @@ static int nfs4_fill_super(struct super_block *sb, struct nfs4_mount_data *data,
 
 	if (IS_ERR(clnt)) {
 		printk(KERN_WARNING "NFS: cannot create RPC client.\n");
-		err = PTR_ERR(clnt);
-		goto out_remove_list;
+		return PTR_ERR(clnt);
 	}
 
 	clnt->cl_intr     = (server->flags & NFS4_MOUNT_INTR) ? 1 : 0;
 	clnt->cl_softrtry = (server->flags & NFS4_MOUNT_SOFT) ? 1 : 0;
 	server->client    = clnt;
 
-	err = -ENOMEM;
 	if (server->nfs4_state->cl_idmap == NULL) {
 		printk(KERN_WARNING "NFS: failed to create idmapper.\n");
-		goto out_shutdown;
+		return -ENOMEM;
 	}
 
 	if (clnt->cl_auth->au_flavor != authflavour) {
 		if (rpcauth_create(authflavour, clnt) == NULL) {
 			printk(KERN_WARNING "NFS: couldn't create credcache!\n");
-			goto out_shutdown;
+			return -ENOMEM;
 		}
 	}
 
-	/* Fire up rpciod if not yet running */
-	if (rpciod_up() != 0) {
-		printk(KERN_WARNING "NFS: couldn't start rpciod!\n");
-		goto out_shutdown;
-	}
-
 	sb->s_op = &nfs4_sops;
 	err = nfs_sb_init(sb, authflavour);
 	if (err == 0)
 		return 0;
-	rpciod_down();
-out_shutdown:
-	rpc_shutdown_client(server->client);
-out_remove_list:
-	down_write(&server->nfs4_state->cl_sem);
-	list_del_init(&server->nfs4_siblings);
-	up_write(&server->nfs4_state->cl_sem);
-	destroy_nfsv4_state(server);
 out_fail:
 	if (clp)
 		nfs4_put_client(clp);
@@ -1691,6 +1754,13 @@ static struct super_block *nfs4_get_sb(struct file_system_type *fs_type,
 
 	s->s_flags = flags;
 
+	/* Fire up rpciod if not yet running */
+	if (rpciod_up() != 0) {
+		printk(KERN_WARNING "NFS: couldn't start rpciod!\n");
+		s = ERR_PTR(-EIO);
+		goto out_free;
+	}
+
 	error = nfs4_fill_super(s, data, flags & MS_VERBOSE ? 1 : 0);
 	if (error) {
 		up_write(&s->s_umount);
@@ -1710,22 +1780,31 @@ static struct super_block *nfs4_get_sb(struct file_system_type *fs_type,
 	return s;
 }
 
+static void nfs4_kill_super(struct super_block *sb)
+{
+	nfs_return_all_delegations(sb);
+	nfs_kill_super(sb);
+}
+
 static struct file_system_type nfs4_fs_type = {
 	.owner		= THIS_MODULE,
 	.name		= "nfs4",
 	.get_sb		= nfs4_get_sb,
-	.kill_sb	= nfs_kill_super,
+	.kill_sb	= nfs4_kill_super,
 	.fs_flags	= FS_ODD_RENAME|FS_REVAL_DOT|FS_BINARY_MOUNTDATA,
 };
 
-#define nfs4_zero_state(nfsi) \
+#define nfs4_init_once(nfsi) \
 	do { \
 		INIT_LIST_HEAD(&(nfsi)->open_states); \
+		nfsi->delegation = NULL; \
+		nfsi->delegation_state = 0; \
+		init_rwsem(&nfsi->rwsem); \
 	} while(0)
 #define register_nfs4fs() register_filesystem(&nfs4_fs_type)
 #define unregister_nfs4fs() unregister_filesystem(&nfs4_fs_type)
 #else
-#define nfs4_zero_state(nfsi) \
+#define nfs4_init_once(nfsi) \
 	do { } while (0)
 #define register_nfs4fs() (0)
 #define unregister_nfs4fs()
@@ -1747,8 +1826,6 @@ 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;
 }
 
@@ -1764,14 +1841,17 @@ static void init_once(void * foo, kmem_cache_t * cachep, unsigned long flags)
 	if ((flags & (SLAB_CTOR_VERIFY|SLAB_CTOR_CONSTRUCTOR)) ==
 	    SLAB_CTOR_CONSTRUCTOR) {
 		inode_init_once(&nfsi->vfs_inode);
+		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;
 		nfsi->ncommit = 0;
 		nfsi->npages = 0;
 		init_waitqueue_head(&nfsi->nfs_i_wait);
+		nfs4_init_once(nfsi);
 	}
 }
  

fs/nfs/mount_clnt.c

@@ -108,7 +108,6 @@ xdr_decode_fhstatus(struct rpc_rqst *req, u32 *p, struct mnt_fhstatus *res)
 {
 	struct nfs_fh *fh = res->fh;
 
-	memset((void *)fh, 0, sizeof(*fh));
 	if ((res->status = ntohl(*p++)) == 0) {
 		fh->size = NFS2_FHSIZE;
 		memcpy(fh->data, p, NFS2_FHSIZE);
@@ -121,7 +120,6 @@ xdr_decode_fhstatus3(struct rpc_rqst *req, u32 *p, struct mnt_fhstatus *res)
 {
 	struct nfs_fh *fh = res->fh;
 
-	memset((void *)fh, 0, sizeof(*fh));
 	if ((res->status = ntohl(*p++)) == 0) {
 		int size = ntohl(*p++);
 		if (size <= NFS3_FHSIZE) {

fs/nfs/nfs2xdr.c

@@ -77,8 +77,6 @@ xdr_encode_fhandle(u32 *p, struct nfs_fh *fhandle)
 static inline u32 *
 xdr_decode_fhandle(u32 *p, struct nfs_fh *fhandle)
 {
-	/* Zero handle first to allow comparisons */
-	memset(fhandle, 0, sizeof(*fhandle));
 	/* NFSv2 handles have a fixed length */
 	fhandle->size = NFS2_FHSIZE;
 	memcpy(fhandle->data, p, NFS2_FHSIZE);
@@ -94,6 +92,23 @@ xdr_encode_time(u32 *p, struct timespec *timep)
 	return p;
 }
 
+static inline u32*
+xdr_encode_current_server_time(u32 *p, struct timespec *timep)
+{
+	/*
+	 * Passing the invalid value useconds=1000000 is a
+	 * Sun convention for "set to current server time".
+	 * It's needed to make permissions checks for the
+	 * "touch" program across v2 mounts to Solaris and
+	 * Irix boxes work correctly. See description of
+	 * sattr in section 6.1 of "NFS Illustrated" by
+	 * Brent Callaghan, Addison-Wesley, ISBN 0-201-32750-5
+	 */
+	*p++ = htonl(timep->tv_sec);
+	*p++ = htonl(1000000);
+	return p;
+}
+
 static inline u32*
 xdr_decode_time(u32 *p, struct timespec *timep)
 {
@@ -142,15 +157,19 @@ xdr_encode_sattr(u32 *p, struct iattr *attr)
 	SATTR(p, attr, ATTR_GID, ia_gid);
 	SATTR(p, attr, ATTR_SIZE, ia_size);
 
-	if (attr->ia_valid & (ATTR_ATIME|ATTR_ATIME_SET)) {
+	if (attr->ia_valid & ATTR_ATIME_SET) {
 		p = xdr_encode_time(p, &attr->ia_atime);
+	} else if (attr->ia_valid & ATTR_ATIME) {
+		p = xdr_encode_current_server_time(p, &attr->ia_atime);
 	} else {
 		*p++ = ~(u32) 0;
 		*p++ = ~(u32) 0;
 	}
 
-	if (attr->ia_valid & (ATTR_MTIME|ATTR_MTIME_SET)) {
+	if (attr->ia_valid & ATTR_MTIME_SET) {
 		p = xdr_encode_time(p, &attr->ia_mtime);
+	} else if (attr->ia_valid & ATTR_MTIME) {
+		p = xdr_encode_current_server_time(p, &attr->ia_mtime);
 	} else {
 		*p++ = ~(u32) 0;	
 		*p++ = ~(u32) 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;
 }
@@ -164,8 +153,7 @@ nfs3_proc_lookup(struct inode *dir, struct qstr *name,
 	return status;
 }
 
-static int
-nfs3_proc_access(struct inode *inode, struct rpc_cred *cred, int mode)
+static int nfs3_proc_access(struct inode *inode, struct nfs_access_entry *entry)
 {
 	struct nfs_fattr	fattr;
 	struct nfs3_accessargs	arg = {
@@ -178,9 +166,10 @@ nfs3_proc_access(struct inode *inode, struct rpc_cred *cred, int mode)
 		.rpc_proc	= &nfs3_procedures[NFS3PROC_ACCESS],
 		.rpc_argp	= &arg,
 		.rpc_resp	= &res,
-		.rpc_cred	= cred
+		.rpc_cred	= entry->cred
 	};
-	int	status;
+	int mode = entry->mask;
+	int status;
 
 	dprintk("NFS call  access\n");
 	fattr.valid = 0;
@@ -200,10 +189,16 @@ nfs3_proc_access(struct inode *inode, struct rpc_cred *cred, int mode)
 	}
 	status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
 	nfs_refresh_inode(inode, &fattr);
-	dprintk("NFS reply access\n");
-
-	if (status == 0 && (arg.access & res.access) != arg.access)
-		status = -EACCES;
+	if (status == 0) {
+		entry->mask = 0;
+		if (res.access & NFS3_ACCESS_READ)
+			entry->mask |= MAY_READ;
+		if (res.access & (NFS3_ACCESS_MODIFY | NFS3_ACCESS_EXTEND | NFS3_ACCESS_DELETE))
+			entry->mask |= MAY_WRITE;
+		if (res.access & (NFS3_ACCESS_LOOKUP|NFS3_ACCESS_EXECUTE))
+			entry->mask |= MAY_EXEC;
+	}
+	dprintk("NFS reply access, status = %d\n", status);
 	return status;
 }
 
@@ -227,8 +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;
@@ -237,13 +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);
@@ -251,8 +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;
@@ -261,13 +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);
@@ -275,8 +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;
@@ -284,13 +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);
@@ -534,6 +526,8 @@ nfs3_proc_symlink(struct inode *dir, struct qstr *name, struct qstr *path,
 	};
 	int			status;
 
+	if (path->len > NFS3_MAXPATHLEN)
+		return -ENAMETOOLONG;
 	dprintk("NFS call  symlink %s -> %s\n", name->name, path->name);
 	dir_attr.valid = 0;
 	fattr->valid = 0;
@@ -832,27 +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)
 {
@@ -892,7 +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/nfs3xdr.c

@@ -109,10 +109,6 @@ xdr_encode_fhandle(u32 *p, struct nfs_fh *fh)
 static inline u32 *
 xdr_decode_fhandle(u32 *p, struct nfs_fh *fh)
 {
-	/*
-	 * Zero all nonused bytes
-	 */
-	memset((u8 *)fh, 0, sizeof(*fh));
 	if ((fh->size = ntohl(*p++)) <= NFS3_FHSIZE) {
 		memcpy(fh->data, p, fh->size);
 		return p + XDR_QUADLEN(fh->size);

fs/nfs/nfs4proc.c

@@ -47,12 +47,16 @@
 #include <linux/smp_lock.h>
 #include <linux/namei.h>
 
+#include "delegation.h"
+
 #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 *);
+static int _nfs4_proc_access(struct inode *inode, struct nfs_access_entry *entry);
 extern u32 *nfs4_decode_dirent(u32 *p, struct nfs_entry *entry, int plus);
 extern struct rpc_procinfo nfs4_procedures[];
 
@@ -189,53 +193,296 @@ 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);
-	struct nfs_fattr fattr = {
-		.valid = 0,
-	};
-	struct nfs_open_reclaimargs o_arg = {
+	struct nfs_delegation *delegation = NFS_I(inode)->delegation;
+	struct nfs_openargs o_arg = {
 		.fh = NFS_FH(inode),
 		.seqid = sp->so_seqid,
 		.id = sp->so_id,
-		.share_access = state->state,
+		.open_flags = state->state,
 		.clientid = server->nfs4_state->cl_clientid,
 		.claim = NFS4_OPEN_CLAIM_PREVIOUS,
 		.bitmask = server->attr_bitmask,
 	};
 	struct nfs_openres o_res = {
-		.f_attr = &fattr,
 		.server = server,	/* Grrr */
 	};
 	struct rpc_message msg = {
-		.rpc_proc       = &nfs4_procedures[NFSPROC4_CLNT_OPEN_RECLAIM],
+		.rpc_proc       = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR],
 		.rpc_argp       = &o_arg,
 		.rpc_resp	= &o_res,
 		.rpc_cred	= sp->so_cred,
 	};
 	int status;
 
-	status = rpc_call_sync(server->client, &msg, 0);
+	if (delegation != NULL) {
+		if (!(delegation->flags & NFS_DELEGATION_NEED_RECLAIM)) {
+			memcpy(&state->stateid, &delegation->stateid,
+					sizeof(state->stateid));
+			set_bit(NFS_DELEGATED_STATE, &state->flags);
+			return 0;
+		}
+		o_arg.u.delegation_type = delegation->type;
+	}
+	status = rpc_call_sync(server->client, &msg, RPC_TASK_NOINTR);
 	nfs4_increment_seqid(status, sp);
-	if (status == 0)
+	if (status == 0) {
 		memcpy(&state->stateid, &o_res.stateid, sizeof(state->stateid));
-	/* Update the inode attributes */
-	nfs_refresh_inode(inode, &fattr);
+		if (o_res.delegation_type != 0) {
+			nfs_inode_reclaim_delegation(inode, sp->so_cred, &o_res);
+			/* Did the server issue an immediate delegation recall? */
+			if (o_res.do_recall)
+				nfs_async_inode_return_delegation(inode, &o_res.stateid);
+		}
+	}
+	clear_bit(NFS_DELEGATED_STATE, &state->flags);
+	/* Ensure we update the inode attributes */
+	NFS_CACHEINV(inode);
 	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;
+}
+
+static int _nfs4_open_delegation_recall(struct dentry *dentry, struct nfs4_state *state)
+{
+	struct nfs4_state_owner  *sp  = state->owner;
+	struct inode *inode = dentry->d_inode;
+	struct nfs_server *server = NFS_SERVER(inode);
+	struct dentry *parent = dget_parent(dentry);
+	struct nfs_openargs arg = {
+		.fh = NFS_FH(parent->d_inode),
+		.clientid = server->nfs4_state->cl_clientid,
+		.name = &dentry->d_name,
+		.id = sp->so_id,
+		.server = server,
+		.bitmask = server->attr_bitmask,
+		.claim = NFS4_OPEN_CLAIM_DELEGATE_CUR,
+	};
+	struct nfs_openres res = {
+		.server = server,
+	};
+	struct 	rpc_message msg = {
+		.rpc_proc       = &nfs4_procedures[NFSPROC4_CLNT_OPEN_NOATTR],
+		.rpc_argp       = &arg,
+		.rpc_resp       = &res,
+		.rpc_cred	= sp->so_cred,
+	};
+	int status = 0;
+
+	down(&sp->so_sema);
+	if (!test_bit(NFS_DELEGATED_STATE, &state->flags))
+		goto out;
+	if (state->state == 0)
+		goto out;
+	arg.seqid = sp->so_seqid;
+	arg.open_flags = state->state;
+	memcpy(arg.u.delegation.data, state->stateid.data, sizeof(arg.u.delegation.data));
+	status = rpc_call_sync(server->client, &msg, RPC_TASK_NOINTR);
+	nfs4_increment_seqid(status, sp);
+	if (status >= 0) {
+		memcpy(state->stateid.data, res.stateid.data,
+				sizeof(state->stateid.data));
+		clear_bit(NFS_DELEGATED_STATE, &state->flags);
+	}
+out:
+	up(&sp->so_sema);
+	dput(parent);
+	return status;
+}
+
+int nfs4_open_delegation_recall(struct dentry *dentry, struct nfs4_state *state)
+{
+	struct nfs4_exception exception = { };
+	struct nfs_server *server = NFS_SERVER(dentry->d_inode);
+	int err;
+	do {
+		err = _nfs4_open_delegation_recall(dentry, state);
+		switch (err) {
+			case 0:
+				return err;
+			case -NFS4ERR_STALE_CLIENTID:
+			case -NFS4ERR_STALE_STATEID:
+			case -NFS4ERR_EXPIRED:
+				/* Don't recall a delegation if it was lost */
+				nfs4_schedule_state_recovery(server->nfs4_state);
+				return err;
+		}
+		err = nfs4_handle_exception(server, err, &exception);
+	} while (exception.retry);
+	return err;
+}
+
+static int _nfs4_proc_open_confirm(struct rpc_clnt *clnt, const struct nfs_fh *fh, struct nfs4_state_owner *sp, nfs4_stateid *stateid)
+{
+	struct nfs_open_confirmargs arg = {
+		.fh             = fh,
+		.seqid          = sp->so_seqid,
+		.stateid	= *stateid,
+	};
+	struct nfs_open_confirmres res;
+	struct 	rpc_message msg = {
+		.rpc_proc       = &nfs4_procedures[NFSPROC4_CLNT_OPEN_CONFIRM],
+		.rpc_argp       = &arg,
+		.rpc_resp       = &res,
+		.rpc_cred	= sp->so_cred,
+	};
+	int status;
+
+	status = rpc_call_sync(clnt, &msg, RPC_TASK_NOINTR);
+	nfs4_increment_seqid(status, sp);
+	if (status >= 0)
+		memcpy(stateid, &res.stateid, sizeof(*stateid));
+	return status;
+}
+
+static int _nfs4_do_access(struct inode *inode, struct rpc_cred *cred, int mask)
+{
+	struct nfs_access_entry cache;
+	int status;
+
+	status = nfs_access_get_cached(inode, cred, &cache);
+	if (status == 0)
+		goto out;
+
+	/* Be clever: ask server to check for all possible rights */
+	cache.mask = MAY_EXEC | MAY_WRITE | MAY_READ;
+	cache.cred = cred;
+	cache.jiffies = jiffies;
+	status = _nfs4_proc_access(inode, &cache);
+	if (status != 0)
+		return status;
+	nfs_access_add_cache(inode, &cache);
+out:
+	if ((cache.mask & mask) == mask)
+		return 0;
+	return -EACCES;
+}
+
+/*
+ * Returns an nfs4_state + an extra reference to the inode
+ */
+int _nfs4_open_delegated(struct inode *inode, int flags, struct rpc_cred *cred, struct nfs4_state **res)
+{
+	struct nfs_delegation *delegation;
+	struct nfs_server *server = NFS_SERVER(inode);
+	struct nfs4_client *clp = server->nfs4_state;
+	struct nfs_inode *nfsi = NFS_I(inode);
+	struct nfs4_state_owner *sp = NULL;
+	struct nfs4_state *state = NULL;
+	int open_flags = flags & (FMODE_READ|FMODE_WRITE);
+	int mask = 0;
+	int err;
+
+	/* Protect against reboot recovery - NOTE ORDER! */
+	down_read(&clp->cl_sem);
+	/* Protect against delegation recall */
+	down_read(&nfsi->rwsem);
+	delegation = NFS_I(inode)->delegation;
+	err = -ENOENT;
+	if (delegation == NULL || (delegation->type & open_flags) != open_flags)
+		goto out_err;
+	err = -ENOMEM;
+	if (!(sp = nfs4_get_state_owner(server, cred))) {
+		dprintk("%s: nfs4_get_state_owner failed!\n", __FUNCTION__);
+		goto out_err;
+	}
+	down(&sp->so_sema);
+	state = nfs4_get_open_state(inode, sp);
+	if (state == NULL)
+		goto out_err;
+
+	err = -ENOENT;
+	if ((state->state & open_flags) == open_flags) {
+		spin_lock(&inode->i_lock);
+		if (open_flags & FMODE_READ)
+			state->nreaders++;
+		if (open_flags & FMODE_WRITE)
+			state->nwriters++;
+		spin_unlock(&inode->i_lock);
+		goto out_ok;
+	} else if (state->state != 0)
+		goto out_err;
+
+	lock_kernel();
+	err = _nfs4_do_access(inode, cred, mask);
+	unlock_kernel();
+	if (err != 0)
+		goto out_err;
+	spin_lock(&inode->i_lock);
+	memcpy(state->stateid.data, delegation->stateid.data,
+			sizeof(state->stateid.data));
+	state->state |= open_flags;
+	if (open_flags & FMODE_READ)
+		state->nreaders++;
+	if (open_flags & FMODE_WRITE)
+		state->nwriters++;
+	set_bit(NFS_DELEGATED_STATE, &state->flags);
+	spin_unlock(&inode->i_lock);
+out_ok:
+	up(&sp->so_sema);
+	nfs4_put_state_owner(sp);
+	up_read(&nfsi->rwsem);
+	up_read(&clp->cl_sem);
+	igrab(inode);
+	*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);
+	}
+	up_read(&nfsi->rwsem);
+	up_read(&clp->cl_sem);
+	return err;
+}
+
+static struct nfs4_state *nfs4_open_delegated(struct inode *inode, int flags, struct rpc_cred *cred)
+{
+	struct nfs4_exception exception = { };
+	struct nfs4_state *res;
+	int err;
+
+	do {
+		err = _nfs4_open_delegated(inode, flags, cred, &res);
+		if (err == 0)
+			break;
+		res = ERR_PTR(nfs4_handle_exception(NFS_SERVER(inode),
+					err, &exception));
+	} while (exception.retry);
+	return res;
+}
+
 /*
  * 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 = {
@@ -243,12 +490,11 @@ nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *satt
 	};
 	struct nfs_openargs o_arg = {
 		.fh             = NFS_FH(dir),
-		.share_access   = flags & (FMODE_READ|FMODE_WRITE),
-		.opentype       = (flags & O_CREAT) ? NFS4_OPEN_CREATE : NFS4_OPEN_NOCREATE,
-		.createmode     = (flags & O_EXCL) ? NFS4_CREATE_EXCLUSIVE : NFS4_CREATE_UNCHECKED,
+		.open_flags	= flags,
 		.name           = name,
 		.server         = server,
 		.bitmask = server->attr_bitmask,
+		.claim = NFS4_OPEN_CLAIM_NULL,
 	};
 	struct nfs_openres o_res = {
 		.f_attr         = &f_attr,
@@ -261,60 +507,50 @@ 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){
+	if (flags & O_EXCL) {
 		u32 *p = (u32 *) o_arg.u.verifier.data;
 		p[0] = jiffies;
 		p[1] = current->pid;
-	} else if (o_arg.createmode == NFS4_CREATE_UNCHECKED) {
+	} else
 		o_arg.u.attrs = sattr;
-	}
 	/* Serialization for the sequence id */
 	down(&sp->so_sema);
 	o_arg.seqid = sp->so_seqid;
 	o_arg.id = sp->so_id;
-	o_arg.clientid = NFS_SERVER(dir)->nfs4_state->cl_clientid,
+	o_arg.clientid = clp->cl_clientid,
 
-	status = rpc_call_sync(server->client, &msg, 0);
+	status = rpc_call_sync(server->client, &msg, RPC_TASK_NOINTR);
 	nfs4_increment_seqid(status, sp);
 	if (status)
-		goto out_up;
+		goto out_err;
 	update_changeattr(dir, &o_res.cinfo);
+	if(o_res.rflags & NFS4_OPEN_RESULT_CONFIRM) {
+		status = _nfs4_proc_open_confirm(server->client, &o_res.fh,
+				sp, &o_res.stateid);
+		if (status != 0)
+			goto out_err;
+	}
+	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;
-
-	if(o_res.rflags & NFS4_OPEN_RESULT_CONFIRM) {
-		struct nfs_open_confirmargs oc_arg = {
-			.fh             = &o_res.fh,
-			.seqid          = sp->so_seqid,
-		};
-		struct nfs_open_confirmres oc_res;
-		struct 	rpc_message msg = {
-			.rpc_proc       = &nfs4_procedures[NFSPROC4_CLNT_OPEN_CONFIRM],
-			.rpc_argp       = &oc_arg,
-			.rpc_resp       = &oc_res,
-			.rpc_cred	= cred,
-		};
-
-		memcpy(&oc_arg.stateid, &o_res.stateid, sizeof(oc_arg.stateid));
-		status = rpc_call_sync(server->client, &msg, 0);
-		nfs4_increment_seqid(status, sp);
-		if (status)
-			goto out_up;
-		memcpy(&state->stateid, &oc_res.stateid, sizeof(state->stateid));
-	} else
-		memcpy(&state->stateid, &o_res.stateid, sizeof(state->stateid));
+		goto out_err;
+	memcpy(&state->stateid, &o_res.stateid, sizeof(state->stateid));
 	spin_lock(&inode->i_lock);
 	if (flags & FMODE_READ)
 		state->nreaders++;
@@ -322,47 +558,62 @@ nfs4_do_open(struct inode *dir, struct qstr *name, int flags, struct iattr *satt
 		state->nwriters++;
 	state->state |= flags & (FMODE_READ|FMODE_WRITE);
 	spin_unlock(&inode->i_lock);
-
+	if (o_res.delegation_type != 0)
+		nfs_inode_set_delegation(inode, cred, &o_res);
 	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 +632,7 @@ 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 +640,22 @@ 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 +669,7 @@ 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;
@@ -426,10 +683,12 @@ nfs4_do_close(struct inode *inode, struct nfs4_state *state)
 		.rpc_resp	= &res,
 	};
 
+	if (test_bit(NFS_DELEGATED_STATE, &state->flags))
+		return 0;
 	memcpy(&arg.stateid, &state->stateid, sizeof(arg.stateid));
 	/* Serialization for the sequence id */
 	arg.seqid = sp->so_seqid,
-	status = rpc_call_sync(NFS_SERVER(inode)->client, &msg, 0);
+	status = rpc_call_sync(NFS_SERVER(inode)->client, &msg, RPC_TASK_NOINTR);
 
         /* hmm. we are done with the inode, and in the process of freeing
 	 * the state_owner. we keep this around to process errors
@@ -441,15 +700,34 @@ 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;
 	struct nfs_closeargs arg = {
 		.fh		= NFS_FH(inode),
 		.seqid		= sp->so_seqid,
-		.share_access	= mode,
+		.open_flags	= mode,
 	};
 	struct nfs_closeres res;
 	struct rpc_message msg = {
@@ -458,8 +736,10 @@ nfs4_do_downgrade(struct inode *inode, struct nfs4_state *state, mode_t mode)
 		.rpc_resp	= &res,
 	};
 
+	if (test_bit(NFS_DELEGATED_STATE, &state->flags))
+		return 0;
 	memcpy(&arg.stateid, &state->stateid, sizeof(arg.stateid));
-	status = rpc_call_sync(NFS_SERVER(inode)->client, &msg, 0);
+	status = rpc_call_sync(NFS_SERVER(inode)->client, &msg, RPC_TASK_NOINTR);
 	nfs4_increment_seqid(status, sp);
 	if (!status)
 		memcpy(&state->stateid, &res.stateid, sizeof(state->stateid));
@@ -467,6 +747,26 @@ 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)
 {
@@ -500,7 +800,9 @@ nfs4_open_revalidate(struct inode *dir, struct dentry *dentry, int openflags)
 	struct inode *inode;
 
 	cred = rpcauth_lookupcred(NFS_SERVER(dir)->client->cl_auth, 0);
-	state = nfs4_do_open(dir, &dentry->d_name, openflags, NULL, cred);
+	state = nfs4_open_delegated(dentry->d_inode, openflags, cred);
+	if (IS_ERR(state))
+		state = nfs4_do_open(dir, &dentry->d_name, openflags, NULL, cred);
 	put_rpccred(cred);
 	if (state == ERR_PTR(-ENOENT) && dentry->d_inode == 0)
 		return 1;
@@ -518,7 +820,7 @@ 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 +844,19 @@ 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 +877,19 @@ 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 +924,8 @@ 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 +935,13 @@ 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 +954,13 @@ 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 +974,19 @@ 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;
 }
 
 /* 
@@ -678,9 +1021,13 @@ nfs4_proc_setattr(struct dentry *dentry, struct nfs_fattr *fattr,
 	if (size_change) {
 		struct rpc_cred *cred = rpcauth_lookupcred(NFS_SERVER(inode)->client->cl_auth, 0);
 		state = nfs4_find_state(inode, cred, FMODE_WRITE);
-		if (!state) {
-			state = nfs4_do_open(dentry->d_parent->d_inode, 
-				&dentry->d_name, FMODE_WRITE, NULL, cred);
+		if (state == NULL) {
+			state = nfs4_open_delegated(dentry->d_inode,
+					FMODE_WRITE, cred);
+			if (IS_ERR(state))
+				state = nfs4_do_open(dentry->d_parent->d_inode,
+						&dentry->d_name, FMODE_WRITE,
+						NULL, cred);
 			need_iput = 1;
 		}
 		put_rpccred(cred);
@@ -705,7 +1052,7 @@ 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;
@@ -731,12 +1078,23 @@ 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 rpc_cred *cred, int mode)
+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)
 {
-	int			status;
 	struct nfs4_accessargs args = {
 		.fh = NFS_FH(inode),
 	};
@@ -745,8 +1103,10 @@ static int nfs4_proc_access(struct inode *inode, struct rpc_cred *cred, int mode
 		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_ACCESS],
 		.rpc_argp = &args,
 		.rpc_resp = &res,
-		.rpc_cred = cred,
+		.rpc_cred = entry->cred,
 	};
+	int mode = entry->mask;
+	int status;
 
 	/*
 	 * Determine which access bits we want to ask for...
@@ -758,8 +1118,7 @@ static int nfs4_proc_access(struct inode *inode, struct rpc_cred *cred, int mode
 			args.access |= NFS4_ACCESS_MODIFY | NFS4_ACCESS_EXTEND | NFS4_ACCESS_DELETE;
 		if (mode & MAY_EXEC)
 			args.access |= NFS4_ACCESS_LOOKUP;
-	}
-	else {
+	} else {
 		if (mode & MAY_WRITE)
 			args.access |= NFS4_ACCESS_MODIFY | NFS4_ACCESS_EXTEND;
 		if (mode & MAY_EXEC)
@@ -767,13 +1126,27 @@ static int nfs4_proc_access(struct inode *inode, struct rpc_cred *cred, int mode
 	}
 	status = rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
 	if (!status) {
-		if (args.access != res.supported) {
-			printk(KERN_NOTICE "NFS: server didn't support all access bits!\n");
-			status = -ENOTSUPP;
-		} else if ((args.access & res.access) != args.access)
-			status = -EACCES;
+		entry->mask = 0;
+		if (res.access & NFS4_ACCESS_READ)
+			entry->mask |= MAY_READ;
+		if (res.access & (NFS4_ACCESS_MODIFY | NFS4_ACCESS_EXTEND | NFS4_ACCESS_DELETE))
+			entry->mask |= MAY_WRITE;
+		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;
 }
 
 /*
@@ -800,7 +1173,7 @@ static int nfs4_proc_access(struct inode *inode, struct rpc_cred *cred, int mode
  * 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),
@@ -813,11 +1186,22 @@ 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;
@@ -827,6 +1211,7 @@ 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;
@@ -834,29 +1219,27 @@ 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;
@@ -866,33 +1249,32 @@ 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;
@@ -901,24 +1283,29 @@ 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;
 }
 
 /*
@@ -965,7 +1352,7 @@ 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),
@@ -982,7 +1369,19 @@ 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 {
@@ -1023,7 +1422,7 @@ 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 = {
@@ -1046,10 +1445,24 @@ 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),
@@ -1068,10 +1481,22 @@ 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)
 {
@@ -1090,22 +1515,39 @@ static int nfs4_proc_symlink(struct inode *dir, struct qstr *name,
 		.fattr = fattr,
 	};
 	struct rpc_message msg = {
-		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_CREATE],
+		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_SYMLINK],
 		.rpc_argp = &arg,
 		.rpc_resp = &res,
 	};
 	int			status;
 
+	if (path->len > NFS4_MAXPATHLEN)
+		return -ENAMETOOLONG;
 	arg.u.symlink = path;
 	fattr->valid = 0;
 	
 	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)
 {
@@ -1135,10 +1577,25 @@ 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;
@@ -1164,10 +1621,24 @@ 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)
 {
@@ -1214,10 +1685,25 @@ 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_statfs(struct nfs_server *server, struct nfs_fh *fhandle,
+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,
 		 struct nfs_fsstat *fsstat)
 {
 	struct nfs4_statfs_arg args = {
@@ -1231,10 +1717,22 @@ 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 = {
@@ -1247,16 +1745,29 @@ 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 = {
@@ -1276,7 +1787,21 @@ 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
@@ -1415,9 +1940,11 @@ renew_done(struct rpc_task *task)
 	if (task->tk_status < 0) {
 		switch (task->tk_status) {
 			case -NFS4ERR_STALE_CLIENTID:
+			case -NFS4ERR_EXPIRED:
+			case -NFS4ERR_CB_PATH_DOWN:
 				nfs4_schedule_state_recovery(clp);
-				return;
 		}
+		return;
 	}
 	spin_lock(&clp->cl_lock);
 	if (time_before(clp->cl_last_renewal,timestamp))
@@ -1450,11 +1977,13 @@ nfs4_proc_renew(struct nfs4_client *clp)
 	int status;
 
 	status = rpc_call_sync(clp->cl_rpcclient, &msg, 0);
+	if (status < 0)
+		return status;
 	spin_lock(&clp->cl_lock);
 	if (time_before(clp->cl_last_renewal,now))
 		clp->cl_last_renewal = now;
 	spin_unlock(&clp->cl_lock);
-	return status;
+	return 0;
 }
 
 /*
@@ -1467,8 +1996,10 @@ 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,
@@ -1478,21 +2009,28 @@ 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;
 }
 
 /*
@@ -1501,37 +2039,12 @@ 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)
 {
@@ -1545,11 +2058,13 @@ 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:
@@ -1560,12 +2075,11 @@ 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();
 
@@ -1573,101 +2087,85 @@ 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,
@@ -1675,15 +2173,24 @@ 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);
 }
@@ -1712,6 +2219,40 @@ nfs4_proc_setclientid_confirm(struct nfs4_client *clp)
 	return status;
 }
 
+static int _nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid)
+{
+	struct nfs4_delegreturnargs args = {
+		.fhandle = NFS_FH(inode),
+		.stateid = stateid,
+	};
+	struct rpc_message msg = {
+		.rpc_proc = &nfs4_procedures[NFSPROC4_CLNT_DELEGRETURN],
+		.rpc_argp = &args,
+		.rpc_cred = cred,
+	};
+
+	return rpc_call_sync(NFS_CLIENT(inode), &msg, 0);
+}
+
+int nfs4_proc_delegreturn(struct inode *inode, struct rpc_cred *cred, const nfs4_stateid *stateid)
+{
+	struct nfs_server *server = NFS_SERVER(inode);
+	struct nfs4_exception exception = { };
+	int err;
+	do {
+		err = _nfs4_proc_delegreturn(inode, cred, stateid);
+		switch (err) {
+			case -NFS4ERR_STALE_STATEID:
+			case -NFS4ERR_EXPIRED:
+				nfs4_schedule_state_recovery(server->nfs4_state);
+			case 0:
+				return 0;
+		}
+		err = nfs4_handle_exception(server, err, &exception);
+	} while (exception.retry);
+	return err;
+}
+
 #define NFS4_LOCK_MINTIMEOUT (1 * HZ)
 #define NFS4_LOCK_MAXTIMEOUT (30 * HZ)
 
@@ -1753,8 +2294,7 @@ 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);
@@ -1778,6 +2318,7 @@ 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);
@@ -1811,14 +2352,28 @@ 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),
@@ -1838,31 +2393,48 @@ 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, RPC_TASK_NOINTR);
+		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);
@@ -1883,23 +2455,22 @@ 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));
@@ -1908,7 +2479,7 @@ nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 		arg.u.lock = &largs;
 		down(&owner->so_sema);
 		otl.open_seqid = owner->so_seqid;
-		status = rpc_call_sync(server->client, &msg, 0);
+		status = rpc_call_sync(server->client, &msg, RPC_TASK_NOINTR);
 		/* increment open_owner seqid on success, and 
 		* seqid mutating errors */
 		nfs4_increment_seqid(status, owner);
@@ -1921,18 +2492,35 @@ nfs4_proc_setlk(struct nfs4_state *state, int cmd, struct file_lock *request)
 		largs.u.exist_lock = &el;
 		largs.new_lock_owner = 0;
 		arg.u.lock = &largs;
-		status = rpc_call_sync(server->client, &msg, 0);
+		status = rpc_call_sync(server->client, &msg, RPC_TASK_NOINTR);
 	}
 	/* increment seqid on success, and * seqid mutating errors*/
 	nfs4_increment_lock_seqid(status, lsp);
 	/* 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_SETLK, 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! */
@@ -1940,19 +2528,34 @@ 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;
@@ -2012,8 +2615,6 @@ 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

@@ -40,11 +40,15 @@
 
 #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>
 #include <linux/bitops.h>
 
+#include "callback.h"
+#include "delegation.h"
+
 #define OPENOWNER_POOL_SIZE	8
 
 static spinlock_t		state_spinlock = SPIN_LOCK_UNLOCKED;
@@ -93,21 +97,26 @@ nfs4_alloc_client(struct in_addr *addr)
 {
 	struct nfs4_client *clp;
 
-	if ((clp = kmalloc(sizeof(*clp), GFP_KERNEL))) {
-		memset(clp, 0, sizeof(*clp));
-		memcpy(&clp->cl_addr, addr, sizeof(clp->cl_addr));
-		init_rwsem(&clp->cl_sem);
-		INIT_LIST_HEAD(&clp->cl_state_owners);
-		INIT_LIST_HEAD(&clp->cl_unused);
-		spin_lock_init(&clp->cl_lock);
-		atomic_set(&clp->cl_count, 1);
-		INIT_WORK(&clp->cl_recoverd, nfs4_recover_state, clp);
-		INIT_WORK(&clp->cl_renewd, nfs4_renew_state, clp);
-		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;
+	if (nfs_callback_up() < 0)
+		return NULL;
+	if ((clp = kmalloc(sizeof(*clp), GFP_KERNEL)) == NULL) {
+		nfs_callback_down();
+		return NULL;
 	}
+	memset(clp, 0, sizeof(*clp));
+	memcpy(&clp->cl_addr, addr, sizeof(clp->cl_addr));
+	init_rwsem(&clp->cl_sem);
+	INIT_LIST_HEAD(&clp->cl_delegations);
+	INIT_LIST_HEAD(&clp->cl_state_owners);
+	INIT_LIST_HEAD(&clp->cl_unused);
+	spin_lock_init(&clp->cl_lock);
+	atomic_set(&clp->cl_count, 1);
+	INIT_WORK(&clp->cl_recoverd, nfs4_recover_state, clp);
+	INIT_WORK(&clp->cl_renewd, nfs4_renew_state, clp);
+	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_OK;
 	return clp;
 }
 
@@ -130,25 +139,52 @@ nfs4_free_client(struct nfs4_client *clp)
 	if (clp->cl_rpcclient)
 		rpc_shutdown_client(clp->cl_rpcclient);
 	kfree(clp);
+	nfs_callback_down();
+}
+
+static struct nfs4_client *__nfs4_find_client(struct in_addr *addr)
+{
+	struct nfs4_client *clp;
+	list_for_each_entry(clp, &nfs4_clientid_list, cl_servers) {
+		if (memcmp(&clp->cl_addr, addr, sizeof(clp->cl_addr)) == 0) {
+			atomic_inc(&clp->cl_count);
+			return clp;
+		}
+	}
+	return NULL;
+}
+
+struct nfs4_client *nfs4_find_client(struct in_addr *addr)
+{
+	struct nfs4_client *clp;
+	spin_lock(&state_spinlock);
+	clp = __nfs4_find_client(addr);
+	spin_unlock(&state_spinlock);
+	return clp;
 }
 
 struct nfs4_client *
 nfs4_get_client(struct in_addr *addr)
 {
-	struct nfs4_client *new, *clp = NULL;
+	struct nfs4_client *clp, *new = NULL;
 
-	new = nfs4_alloc_client(addr);
 	spin_lock(&state_spinlock);
-	list_for_each_entry(clp, &nfs4_clientid_list, cl_servers) {
-		if (memcmp(&clp->cl_addr, addr, sizeof(clp->cl_addr)) == 0)
-			goto found;
+	for (;;) {
+		clp = __nfs4_find_client(addr);
+		if (clp != NULL)
+			break;
+		clp = new;
+		if (clp != NULL) {
+			list_add(&clp->cl_servers, &nfs4_clientid_list);
+			new = NULL;
+			break;
+		}
+		spin_unlock(&state_spinlock);
+		new = nfs4_alloc_client(addr);
+		spin_lock(&state_spinlock);
+		if (new == NULL)
+			break;
 	}
-	if (new)
-		list_add(&new->cl_servers, &nfs4_clientid_list);
-	spin_unlock(&state_spinlock);
-	return new;
-found:
-	atomic_inc(&clp->cl_count);
 	spin_unlock(&state_spinlock);
 	if (new)
 		nfs4_free_client(new);
@@ -169,6 +205,16 @@ 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, NFS4_CALLBACK, nfs_callback_tcpport);
+	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 +231,6 @@ 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;
@@ -224,6 +269,7 @@ nfs4_alloc_state_owner(void)
 	init_MUTEX(&sp->so_sema);
 	sp->so_seqid = 0;                 /* arbitrary */
 	INIT_LIST_HEAD(&sp->so_states);
+	INIT_LIST_HEAD(&sp->so_delegations);
 	atomic_set(&sp->so_count, 1);
 	return sp;
 }
@@ -237,8 +283,11 @@ 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 +303,23 @@ 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 +379,6 @@ __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 +400,6 @@ __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 +456,40 @@ 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 +500,24 @@ 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 +553,7 @@ 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 +561,12 @@ 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 +574,22 @@ 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 +611,9 @@ 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 +640,11 @@ 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 +661,9 @@ 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 +681,7 @@ 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 +693,18 @@ 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 +733,14 @@ 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 +751,66 @@ 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 +837,57 @@ 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;
+	int status = 0;
 
 	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 || status == -NFS4ERR_CB_PATH_DOWN)
+		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);
+	/* Mark all delagations for reclaim */
+	nfs_delegation_mark_reclaim(clp);
+	/* 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);
+	nfs_delegation_reap_unclaimed(clp);
 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);
+	if (status == -NFS4ERR_CB_PATH_DOWN)
+		nfs_handle_cb_pathdown(clp);
+	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

@@ -84,6 +84,7 @@ static int nfs_stat_to_errno(int);
 				((3+NFS4_FHSIZE) >> 2))
 #define encode_getattr_maxsz    (op_encode_hdr_maxsz + 3)
 #define nfs4_name_maxsz		(1 + ((3 + NFS4_MAXNAMLEN) >> 2))
+#define nfs4_path_maxsz		(1 + ((3 + NFS4_MAXPATHLEN) >> 2))
 #define nfs4_fattr_bitmap_maxsz (36 + 2 * nfs4_name_maxsz)
 #define decode_getattr_maxsz    (op_decode_hdr_maxsz + 3 + \
                                 nfs4_fattr_bitmap_maxsz)
@@ -118,10 +119,17 @@ static int nfs_stat_to_errno(int);
 #define encode_link_maxsz	(op_encode_hdr_maxsz + \
 				nfs4_name_maxsz)
 #define decode_link_maxsz	(op_decode_hdr_maxsz + 5)
+#define encode_symlink_maxsz	(op_encode_hdr_maxsz + \
+				1 + nfs4_name_maxsz + \
+				nfs4_path_maxsz + \
+				nfs4_fattr_bitmap_maxsz)
+#define decode_symlink_maxsz	(op_decode_hdr_maxsz + 8)
 #define encode_create_maxsz	(op_encode_hdr_maxsz + \
-				2 + 2 * nfs4_name_maxsz + \
+				2 + nfs4_name_maxsz + \
 				nfs4_fattr_bitmap_maxsz)
 #define decode_create_maxsz	(op_decode_hdr_maxsz + 8)
+#define encode_delegreturn_maxsz (op_encode_hdr_maxsz + 4)
+#define decode_delegreturn_maxsz (op_decode_hdr_maxsz)
 #define NFS4_enc_compound_sz	(1024)  /* XXX: large enough? */
 #define NFS4_dec_compound_sz	(1024)  /* XXX: large enough? */
 #define NFS4_enc_read_sz	(compound_encode_hdr_maxsz + \
@@ -172,16 +180,14 @@ static int nfs_stat_to_errno(int);
 #define NFS4_dec_open_confirm_sz        (compound_decode_hdr_maxsz + \
                                         decode_putfh_maxsz + \
                                         op_decode_hdr_maxsz + 4)
-#define NFS4_enc_open_reclaim_sz	(compound_encode_hdr_maxsz + \
+#define NFS4_enc_open_noattr_sz	(compound_encode_hdr_maxsz + \
 					encode_putfh_maxsz + \
 					op_encode_hdr_maxsz + \
-					11 + \
-					encode_getattr_maxsz)
-#define NFS4_dec_open_reclaim_sz	(compound_decode_hdr_maxsz + \
+					11)
+#define NFS4_dec_open_noattr_sz	(compound_decode_hdr_maxsz + \
 					decode_putfh_maxsz + \
 					op_decode_hdr_maxsz + \
-					4 + 5 + 2 + 3 + \
-					decode_getattr_maxsz)
+					4 + 5 + 2 + 3)
 #define NFS4_enc_open_downgrade_sz \
 				(compound_encode_hdr_maxsz + \
                                 encode_putfh_maxsz + \
@@ -313,6 +319,16 @@ static int nfs_stat_to_errno(int);
 				decode_savefh_maxsz + \
 				decode_putfh_maxsz + \
 				decode_link_maxsz)
+#define NFS4_enc_symlink_sz	(compound_encode_hdr_maxsz + \
+				encode_putfh_maxsz + \
+				encode_symlink_maxsz + \
+				encode_getattr_maxsz + \
+				encode_getfh_maxsz)
+#define NFS4_dec_symlink_sz	(compound_decode_hdr_maxsz + \
+				decode_putfh_maxsz + \
+				decode_symlink_maxsz + \
+				decode_getattr_maxsz + \
+				decode_getfh_maxsz)
 #define NFS4_enc_create_sz	(compound_encode_hdr_maxsz + \
 				encode_putfh_maxsz + \
 				encode_create_maxsz + \
@@ -339,6 +355,11 @@ static int nfs_stat_to_errno(int);
 				encode_getattr_maxsz)
 #define NFS4_dec_server_caps_sz (compound_decode_hdr_maxsz + \
 				decode_getattr_maxsz)
+#define NFS4_enc_delegreturn_sz	(compound_encode_hdr_maxsz + \
+				encode_putfh_maxsz + \
+				encode_delegreturn_maxsz)
+#define NFS4_dec_delegreturn_sz (compound_decode_hdr_maxsz + \
+				decode_delegreturn_maxsz)
 
 static struct {
 	unsigned int	mode;
@@ -388,6 +409,15 @@ 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;
@@ -402,6 +432,15 @@ static int encode_compound_hdr(struct xdr_stream *xdr, struct compound_hdr *hdr)
 	return 0;
 }
 
+static void encode_nfs4_verifier(struct xdr_stream *xdr, const nfs4_verifier *verf)
+{
+	uint32_t *p;
+
+	p = xdr_reserve_space(xdr, NFS4_VERIFIER_SIZE);
+	BUG_ON(p == NULL);
+	xdr_encode_opaque_fixed(p, verf->data, NFS4_VERIFIER_SIZE);
+}
+
 static int encode_attrs(struct xdr_stream *xdr, const struct iattr *iap, const struct nfs_server *server)
 {
 	char owner_name[IDMAP_NAMESZ];
@@ -742,19 +781,12 @@ static int encode_lookup(struct xdr_stream *xdr, const struct qstr *name)
 	return 0;
 }
 
-static int encode_open(struct xdr_stream *xdr, const struct nfs_openargs *arg)
+static void encode_share_access(struct xdr_stream *xdr, int open_flags)
 {
-	int status;
 	uint32_t *p;
 
- /*
- * opcode 4, seqid 4, share_access 4, share_deny 4, clientid 8, ownerlen 4,
- * owner 4, opentype 4 = 36
- */
-	RESERVE_SPACE(36);
-	WRITE32(OP_OPEN);
-	WRITE32(arg->seqid);
-	switch (arg->share_access) {
+	RESERVE_SPACE(8);
+	switch (open_flags & (FMODE_READ|FMODE_WRITE)) {
 		case FMODE_READ:
 			WRITE32(NFS4_SHARE_ACCESS_READ);
 			break;
@@ -767,84 +799,135 @@ static int encode_open(struct xdr_stream *xdr, const struct nfs_openargs *arg)
 		default:
 			BUG();
 	}
-	WRITE32(0);                  /* for linux, share_deny = 0 always */
+	WRITE32(0);		/* for linux, share_deny = 0 always */
+}
+
+static inline void encode_openhdr(struct xdr_stream *xdr, const struct nfs_openargs *arg)
+{
+	uint32_t *p;
+ /*
+ * opcode 4, seqid 4, share_access 4, share_deny 4, clientid 8, ownerlen 4,
+ * owner 4 = 32
+ */
+	RESERVE_SPACE(8);
+	WRITE32(OP_OPEN);
+	WRITE32(arg->seqid);
+	encode_share_access(xdr, arg->open_flags);
+	RESERVE_SPACE(16);
 	WRITE64(arg->clientid);
 	WRITE32(4);
 	WRITE32(arg->id);
-	WRITE32(arg->opentype);
+}
 
-	if (arg->opentype == NFS4_OPEN_CREATE) {
-		if (arg->createmode == NFS4_CREATE_EXCLUSIVE) {
-			RESERVE_SPACE(12);
-			WRITE32(arg->createmode);
-			WRITEMEM(arg->u.verifier.data, sizeof(arg->u.verifier.data));
-		}
-		else if (arg->u.attrs) {
-			RESERVE_SPACE(4);
-			WRITE32(arg->createmode);
-			if ((status = encode_attrs(xdr, arg->u.attrs, arg->server)))
-				return status;
-		}
-		else {
-			RESERVE_SPACE(12);
-			WRITE32(arg->createmode);
-			WRITE32(0);
-			WRITE32(0);
-		}
+static inline void encode_createmode(struct xdr_stream *xdr, const struct nfs_openargs *arg)
+{
+	uint32_t *p;
+
+	RESERVE_SPACE(4);
+	switch(arg->open_flags & O_EXCL) {
+		case 0:
+			WRITE32(NFS4_CREATE_UNCHECKED);
+			encode_attrs(xdr, arg->u.attrs, arg->server);
+			break;
+		default:
+			WRITE32(NFS4_CREATE_EXCLUSIVE);
+			encode_nfs4_verifier(xdr, &arg->u.verifier);
 	}
+}
 
-	RESERVE_SPACE(8 + arg->name->len);
-	WRITE32(NFS4_OPEN_CLAIM_NULL);
-	WRITE32(arg->name->len);
-	WRITEMEM(arg->name->name, arg->name->len);
+static void encode_opentype(struct xdr_stream *xdr, const struct nfs_openargs *arg)
+{
+	uint32_t *p;
 
-	return 0;
+	RESERVE_SPACE(4);
+	switch (arg->open_flags & O_CREAT) {
+		case 0:
+			WRITE32(NFS4_OPEN_NOCREATE);
+			break;
+		default:
+			BUG_ON(arg->claim != NFS4_OPEN_CLAIM_NULL);
+			WRITE32(NFS4_OPEN_CREATE);
+			encode_createmode(xdr, arg);
+	}
 }
 
-static int encode_open_confirm(struct xdr_stream *xdr, const struct nfs_open_confirmargs *arg)
+static inline void encode_delegation_type(struct xdr_stream *xdr, int delegation_type)
 {
 	uint32_t *p;
 
-	RESERVE_SPACE(8+sizeof(arg->stateid.data));
-	WRITE32(OP_OPEN_CONFIRM);
-	WRITEMEM(arg->stateid.data, sizeof(arg->stateid.data));
-	WRITE32(arg->seqid);
+	RESERVE_SPACE(4);
+	switch (delegation_type) {
+		case 0:
+			WRITE32(NFS4_OPEN_DELEGATE_NONE);
+			break;
+		case FMODE_READ:
+			WRITE32(NFS4_OPEN_DELEGATE_READ);
+			break;
+		case FMODE_WRITE|FMODE_READ:
+			WRITE32(NFS4_OPEN_DELEGATE_WRITE);
+			break;
+		default:
+			BUG();
+	}
+}
 
-	return 0;
+static inline void encode_claim_null(struct xdr_stream *xdr, const struct qstr *name)
+{
+	uint32_t *p;
+
+	RESERVE_SPACE(4);
+	WRITE32(NFS4_OPEN_CLAIM_NULL);
+	encode_string(xdr, name->len, name->name);
 }
 
+static inline void encode_claim_previous(struct xdr_stream *xdr, int type)
+{
+	uint32_t *p;
+
+	RESERVE_SPACE(4);
+	WRITE32(NFS4_OPEN_CLAIM_PREVIOUS);
+	encode_delegation_type(xdr, type);
+}
 
-static int encode_open_reclaim(struct xdr_stream *xdr, const struct nfs_open_reclaimargs *arg)
+static inline void encode_claim_delegate_cur(struct xdr_stream *xdr, const struct qstr *name, const nfs4_stateid *stateid)
 {
 	uint32_t *p;
 
- /*
- * opcode 4, seqid 4, share_access 4, share_deny 4, clientid 8, ownerlen 4,
- * owner 4, opentype 4, claim 4, delegation_type 4 = 44
- */
-	RESERVE_SPACE(44);
-	WRITE32(OP_OPEN);
-	WRITE32(arg->seqid);
-	switch (arg->share_access) {
-		case FMODE_READ:
-			WRITE32(NFS4_SHARE_ACCESS_READ);
+	RESERVE_SPACE(4+sizeof(stateid->data));
+	WRITE32(NFS4_OPEN_CLAIM_DELEGATE_CUR);
+	WRITEMEM(stateid->data, sizeof(stateid->data));
+	encode_string(xdr, name->len, name->name);
+}
+
+static int encode_open(struct xdr_stream *xdr, const struct nfs_openargs *arg)
+{
+	encode_openhdr(xdr, arg);
+	encode_opentype(xdr, arg);
+	switch (arg->claim) {
+		case NFS4_OPEN_CLAIM_NULL:
+			encode_claim_null(xdr, arg->name);
 			break;
-		case FMODE_WRITE:
-			WRITE32(NFS4_SHARE_ACCESS_WRITE);
+		case NFS4_OPEN_CLAIM_PREVIOUS:
+			encode_claim_previous(xdr, arg->u.delegation_type);
 			break;
-		case FMODE_READ|FMODE_WRITE:
-			WRITE32(NFS4_SHARE_ACCESS_BOTH);
+		case NFS4_OPEN_CLAIM_DELEGATE_CUR:
+			encode_claim_delegate_cur(xdr, arg->name, &arg->u.delegation);
 			break;
 		default:
 			BUG();
 	}
-	WRITE32(0);                  /* for linux, share_deny = 0 always */
-	WRITE64(arg->clientid);
-	WRITE32(4);
-	WRITE32(arg->id);
-	WRITE32(NFS4_OPEN_NOCREATE);
-	WRITE32(NFS4_OPEN_CLAIM_PREVIOUS);
-	WRITE32(NFS4_OPEN_DELEGATE_NONE);
+	return 0;
+}
+
+static int encode_open_confirm(struct xdr_stream *xdr, const struct nfs_open_confirmargs *arg)
+{
+	uint32_t *p;
+
+	RESERVE_SPACE(8+sizeof(arg->stateid.data));
+	WRITE32(OP_OPEN_CONFIRM);
+	WRITEMEM(arg->stateid.data, sizeof(arg->stateid.data));
+	WRITE32(arg->seqid);
+
 	return 0;
 }
 
@@ -852,14 +935,11 @@ static int encode_open_downgrade(struct xdr_stream *xdr, const struct nfs_closea
 {
 	uint32_t *p;
 
-	RESERVE_SPACE(16+sizeof(arg->stateid.data));
+	RESERVE_SPACE(8+sizeof(arg->stateid.data));
 	WRITE32(OP_OPEN_DOWNGRADE);
 	WRITEMEM(arg->stateid.data, sizeof(arg->stateid.data));
 	WRITE32(arg->seqid);
-	WRITE32(arg->share_access);
-	/* No deny modes */
-	WRITE32(0);
-
+	encode_share_access(xdr, arg->open_flags);
 	return 0;
 }
 
@@ -887,15 +967,15 @@ 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));
@@ -908,7 +988,7 @@ 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);
@@ -927,7 +1007,7 @@ static int encode_readdir(struct xdr_stream *xdr, const struct nfs4_readdir_arg
 	WRITE32(OP_READDIR);
 	WRITE64(readdir->cookie);
 	WRITEMEM(readdir->verifier.data, sizeof(readdir->verifier.data));
-	WRITE32(readdir->count >> 5);  /* meaningless "dircount" field */
+	WRITE32(readdir->count >> 1);  /* We're not doing readdirplus */
 	WRITE32(readdir->count);
 	WRITE32(2);
 	WRITE32(FATTR4_WORD0_FILEID);
@@ -1031,26 +1111,18 @@ 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;
@@ -1075,7 +1147,7 @@ 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);
@@ -1086,6 +1158,18 @@ static int encode_write(struct xdr_stream *xdr, const struct nfs_writeargs *args
 
 	return 0;
 }
+
+static int encode_delegreturn(struct xdr_stream *xdr, const nfs4_stateid *stateid)
+{
+	uint32_t *p;
+
+	RESERVE_SPACE(20);
+
+	WRITE32(OP_DELEGRETURN);
+	WRITEMEM(stateid->data, sizeof(stateid->data));
+	return 0;
+
+}
 /*
  * END OF "GENERIC" ENCODE ROUTINES.
  */
@@ -1236,13 +1320,21 @@ 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;
 }
 
+/*
+ * Encode SYMLINK request
+ */
+static int nfs4_xdr_enc_symlink(struct rpc_rqst *req, uint32_t *p, const struct nfs4_create_arg *args)
+{
+	return nfs4_xdr_enc_create(req, p, args);
+}
+
 /*
  * Encode GETATTR request
  */
@@ -1301,10 +1393,10 @@ 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;
 }
@@ -1331,13 +1423,13 @@ static int nfs4_xdr_enc_open_confirm(struct rpc_rqst *req, uint32_t *p, struct n
 }
 
 /*
- * Encode an OPEN request
+ * Encode an OPEN request with no attributes.
  */
-static int nfs4_xdr_enc_open_reclaim(struct rpc_rqst *req, uint32_t *p, struct nfs_open_reclaimargs *args)
+static int nfs4_xdr_enc_open_noattr(struct rpc_rqst *req, uint32_t *p, struct nfs_openargs *args)
 {
 	struct xdr_stream xdr;
 	struct compound_hdr hdr = {
-		.nops   = 3,
+		.nops   = 2,
 	};
 	int status;
 
@@ -1346,10 +1438,7 @@ static int nfs4_xdr_enc_open_reclaim(struct rpc_rqst *req, uint32_t *p, struct n
 	status = encode_putfh(&xdr, args->fh);
 	if (status)
 		goto out;
-	status = encode_open_reclaim(&xdr, args);
-	if (status)
-		goto out;
-	status = encode_getfattr(&xdr, args->bitmask);
+	status = encode_open(&xdr, args);
 out:
 	return status;
 }
@@ -1715,6 +1804,24 @@ static int nfs4_xdr_enc_setclientid_confirm(struct rpc_rqst *req, uint32_t *p, s
 	return status;
 }
 
+/*
+ * DELEGRETURN request
+ */
+static int nfs4_xdr_enc_delegreturn(struct rpc_rqst *req, uint32_t *p, const struct nfs4_delegreturnargs *args)
+{
+	struct xdr_stream xdr;
+	struct compound_hdr hdr = {
+		.nops = 2,
+	};
+	int status;
+
+	xdr_init_encode(&xdr, &req->rq_snd_buf, p);
+	encode_compound_hdr(&xdr, &hdr);
+	if ((status = encode_putfh(&xdr, args->fhandle)) == 0)
+		status = encode_delegreturn(&xdr, args->stateid);
+	return status;
+}
+
 /*
  * START OF "GENERIC" DECODE ROUTINES.
  *   These may look a little ugly since they are imported from a "generic"
@@ -1749,6 +1856,17 @@ static int nfs4_xdr_enc_setclientid_confirm(struct rpc_rqst *req, uint32_t *p, s
 	} \
 } while (0)
 
+static int decode_opaque_inline(struct xdr_stream *xdr, uint32_t *len, char **string)
+{
+	uint32_t *p;
+
+	READ_BUF(4);
+	READ32(*len);
+	READ_BUF(*len);
+	*string = (char *)p;
+	return 0;
+}
+
 static int decode_compound_hdr(struct xdr_stream *xdr, struct compound_hdr *hdr)
 {
 	uint32_t *p;
@@ -1785,6 +1903,17 @@ static int decode_op_hdr(struct xdr_stream *xdr, enum nfs_opnum4 expected)
 	return 0;
 }
 
+/* Dummy routine */
+static int decode_ace(struct xdr_stream *xdr, void *ace, struct nfs4_client *clp)
+{
+	uint32_t *p;
+	uint32_t strlen;
+	char *str;
+
+	READ_BUF(12);
+	return decode_opaque_inline(xdr, &strlen, &str);
+}
+
 static int decode_attr_bitmap(struct xdr_stream *xdr, uint32_t *bitmap)
 {
 	uint32_t bmlen, *p;
@@ -2717,10 +2846,56 @@ static int decode_lookup(struct xdr_stream *xdr)
 	return decode_op_hdr(xdr, OP_LOOKUP);
 }
 
+/* This is too sick! */
+static int decode_space_limit(struct xdr_stream *xdr, u64 *maxsize)
+{
+        uint32_t *p;
+	uint32_t limit_type, nblocks, blocksize;
+
+	READ_BUF(12);
+	READ32(limit_type);
+	switch (limit_type) {
+		case 1:
+			READ64(*maxsize);
+			break;
+		case 2:
+			READ32(nblocks);
+			READ32(blocksize);
+			*maxsize = (uint64_t)nblocks * (uint64_t)blocksize;
+	}
+	return 0;
+}
+
+static int decode_delegation(struct xdr_stream *xdr, struct nfs_openres *res)
+{
+        uint32_t *p;
+        uint32_t delegation_type;
+
+	READ_BUF(4);
+	READ32(delegation_type);
+	if (delegation_type == NFS4_OPEN_DELEGATE_NONE) {
+		res->delegation_type = 0;
+		return 0;
+	}
+	READ_BUF(20);
+	COPYMEM(res->delegation.data, sizeof(res->delegation.data));
+	READ32(res->do_recall);
+	switch (delegation_type) {
+		case NFS4_OPEN_DELEGATE_READ:
+			res->delegation_type = FMODE_READ;
+			break;
+		case NFS4_OPEN_DELEGATE_WRITE:
+			res->delegation_type = FMODE_WRITE|FMODE_READ;
+			if (decode_space_limit(xdr, &res->maxsize) < 0)
+				return -EIO;
+	}
+	return decode_ace(xdr, NULL, res->server->nfs4_state);
+}
+
 static int decode_open(struct xdr_stream *xdr, struct nfs_openres *res)
 {
         uint32_t *p;
-        uint32_t bmlen, delegation_type;
+        uint32_t bmlen;
         int status;
 
         status = decode_op_hdr(xdr, OP_OPEN);
@@ -2737,11 +2912,9 @@ static int decode_open(struct xdr_stream *xdr, struct nfs_openres *res)
         if (bmlen > 10)
                 goto xdr_error;
 
-        READ_BUF((bmlen << 2) + 4);
+        READ_BUF(bmlen << 2);
         p += bmlen;
-        READ32(delegation_type);
-        if (delegation_type == NFS4_OPEN_DELEGATE_NONE)
-		return 0;
+	return decode_delegation(xdr, res);
 xdr_error:
 	printk(KERN_NOTICE "%s: xdr error!\n", __FUNCTION__);
 	return -EIO;
@@ -2817,8 +2990,8 @@ static int decode_readdir(struct xdr_stream *xdr, struct rpc_rqst *req, struct n
 	struct kvec	*iov = rcvbuf->head;
 	unsigned int	nr, pglen = rcvbuf->page_len;
 	uint32_t	*end, *entry, *p, *kaddr;
-	uint32_t	len, attrlen, word;
-	int 		i, hdrlen, recvd, status;
+	uint32_t	len, attrlen;
+	int 		hdrlen, recvd, status;
 
 	status = decode_op_hdr(xdr, OP_READDIR);
 	if (status)
@@ -2839,42 +3012,24 @@ static int decode_readdir(struct xdr_stream *xdr, struct rpc_rqst *req, struct n
 	for (nr = 0; *p++; nr++) {
 		if (p + 3 > end)
 			goto short_pkt;
-		p += 2;     /* cookie */
-		len = ntohl(*p++);  /* filename length */
+		p += 2;			/* cookie */
+		len = ntohl(*p++);	/* filename length */
 		if (len > NFS4_MAXNAMLEN) {
 			printk(KERN_WARNING "NFS: giant filename in readdir (len 0x%x)\n", len);
 			goto err_unmap;
 		}
-			
 		p += XDR_QUADLEN(len);
 		if (p + 1 > end)
 			goto short_pkt;
-		len = ntohl(*p++);  /* bitmap length */
-		if (len > 10) {
-			printk(KERN_WARNING "NFS: giant bitmap in readdir (len 0x%x)\n", len);
-			goto err_unmap;
-		}
-		if (p + len + 1 > end)
-			goto short_pkt;
-		attrlen = 0;
-		for (i = 0; i < len; i++) {
-			word = ntohl(*p++);
-			if (!word)
-				continue;
-			else if (i == 0 && word == FATTR4_WORD0_FILEID) {
-				attrlen = 8;
-				continue;
-			}
-			printk(KERN_WARNING "NFS: unexpected bitmap word in readdir (0x%x)\n", word);
-			goto err_unmap;
-		}
-		if (ntohl(*p++) != attrlen) {
-			printk(KERN_WARNING "NFS: unexpected attrlen in readdir\n");
-			goto err_unmap;
-		}
-		p += XDR_QUADLEN(attrlen);
+		len = ntohl(*p++);	/* bitmap length */
+		p += len;
 		if (p + 1 > end)
 			goto short_pkt;
+		attrlen = XDR_QUADLEN(ntohl(*p++));
+		p += attrlen;		/* attributes */
+		if (p + 2 > end)
+			goto short_pkt;
+		entry = p;
 	}
 	if (!nr && (entry[0] != 0 || entry[1] == 0))
 		goto short_pkt;
@@ -3048,6 +3203,11 @@ static int decode_write(struct xdr_stream *xdr, struct nfs_writeres *res)
 	return 0;
 }
 
+static int decode_delegreturn(struct xdr_stream *xdr)
+{
+	return decode_op_hdr(xdr, OP_DELEGRETURN);
+}
+
 /*
  * Decode OPEN_DOWNGRADE response
  */
@@ -3214,13 +3374,23 @@ 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;
 }
 
+/*
+ * Decode SYMLINK response
+ */
+static int nfs4_xdr_dec_symlink(struct rpc_rqst *rqstp, uint32_t *p, struct nfs4_create_res *res)
+{
+	return nfs4_xdr_dec_create(rqstp, p, res);
+}
+
 /*
  * Decode GETATTR response
  */
@@ -3282,12 +3452,14 @@ 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;
 }
@@ -3314,9 +3486,9 @@ static int nfs4_xdr_dec_open_confirm(struct rpc_rqst *rqstp, uint32_t *p, struct
 }
 
 /*
- * Decode OPEN_RECLAIM response
+ * Decode OPEN response
  */
-static int nfs4_xdr_dec_open_reclaim(struct rpc_rqst *rqstp, uint32_t *p, struct nfs_openres *res)
+static int nfs4_xdr_dec_open_noattr(struct rpc_rqst *rqstp, uint32_t *p, struct nfs_openres *res)
 {
         struct xdr_stream xdr;
         struct compound_hdr hdr;
@@ -3330,9 +3502,6 @@ static int nfs4_xdr_dec_open_reclaim(struct rpc_rqst *rqstp, uint32_t *p, struct
         if (status)
                 goto out;
         status = decode_open(&xdr, res);
-        if (status)
-                goto out;
-	status = decode_getfattr(&xdr, res->f_attr, res->server);
 out:
         return status;
 }
@@ -3357,6 +3526,8 @@ 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;
 }
@@ -3665,8 +3836,28 @@ static int nfs4_xdr_dec_setclientid_confirm(struct rpc_rqst *req, uint32_t *p, s
 	return status;
 }
 
+/*
+ * DELEGRETURN request
+ */
+static int nfs4_xdr_dec_delegreturn(struct rpc_rqst *rqstp, uint32_t *p, void *dummy)
+{
+	struct xdr_stream xdr;
+	struct compound_hdr hdr;
+	int status;
+
+	xdr_init_decode(&xdr, &rqstp->rq_rcv_buf, p);
+	status = decode_compound_hdr(&xdr, &hdr);
+	if (status == 0) {
+		status = decode_putfh(&xdr);
+		if (status == 0)
+			status = decode_delegreturn(&xdr);
+	}
+	return status;
+}
+
 uint32_t *nfs4_decode_dirent(uint32_t *p, struct nfs_entry *entry, int plus)
 {
+	uint32_t bitmap[1] = {0};
 	uint32_t len;
 
 	if (!*p++) {
@@ -3689,11 +3880,17 @@ uint32_t *nfs4_decode_dirent(uint32_t *p, struct nfs_entry *entry, int plus)
 	 */
 	entry->ino = 1;
 
-	len = ntohl(*p++);             /* bitmap length */
-	p += len;
-	len = ntohl(*p++);             /* attribute buffer length */
-	if (len)
-		p = xdr_decode_hyper(p, &entry->ino);
+	len = ntohl(*p++);		/* bitmap length */
+	if (len > 0) {
+		bitmap[0] = ntohl(*p);
+		p += len;
+	}
+	len = XDR_QUADLEN(ntohl(*p++));	/* attribute buffer length */
+	if (len > 0) {
+		if (bitmap[0] == FATTR4_WORD0_FILEID)
+			xdr_decode_hyper(p, &entry->ino);
+		p += len;
+	}
 
 	entry->eof = !p[0] && p[1];
 	return p;
@@ -3756,7 +3953,7 @@ nfs_stat_to_errno(int stat)
 		if (nfs_errtbl[i].stat == stat)
 			return nfs_errtbl[i].errno;
 	}
-	if (stat < 0) {
+	if (stat <= 10000 || stat > 10100) {
 		/* The server is looney tunes. */
 		return ESERVERFAULT;
 	}
@@ -3786,7 +3983,7 @@ struct rpc_procinfo	nfs4_procedures[] = {
   PROC(COMMIT,		enc_commit,	dec_commit),
   PROC(OPEN,		enc_open,	dec_open),
   PROC(OPEN_CONFIRM,	enc_open_confirm,	dec_open_confirm),
-  PROC(OPEN_RECLAIM,	enc_open_reclaim,	dec_open_reclaim),
+  PROC(OPEN_NOATTR,	enc_open_noattr,	dec_open_noattr),
   PROC(OPEN_DOWNGRADE,	enc_open_downgrade,	dec_open_downgrade),
   PROC(CLOSE,		enc_close,	dec_close),
   PROC(SETATTR,		enc_setattr,	dec_setattr),
@@ -3804,12 +4001,14 @@ struct rpc_procinfo	nfs4_procedures[] = {
   PROC(REMOVE,		enc_remove,	dec_remove),
   PROC(RENAME,		enc_rename,	dec_rename),
   PROC(LINK,		enc_link,	dec_link),
+  PROC(SYMLINK,		enc_symlink,	dec_symlink),
   PROC(CREATE,		enc_create,	dec_create),
   PROC(PATHCONF,	enc_pathconf,	dec_pathconf),
   PROC(STATFS,		enc_statfs,	dec_statfs),
   PROC(READLINK,	enc_readlink,	dec_readlink),
   PROC(READDIR,		enc_readdir,	dec_readdir),
   PROC(SERVER_CAPS,	enc_server_caps, dec_server_caps),
+  PROC(DELEGRETURN,	enc_delegreturn, dec_delegreturn),
 };
 
 struct rpc_version		nfs_version4 = {

fs/nfs/nfsroot.c

@@ -495,10 +495,8 @@ static int __init root_nfs_get_handle(void)
 	if (status < 0)
 		printk(KERN_ERR "Root-NFS: Server returned error %d "
 				"while mounting %s\n", status, nfs_path);
-	else {
-		nfs_data.root.size = fh.size;
-		memcpy(nfs_data.root.data, fh.data, fh.size);
-	}
+	else
+		nfs_copy_fh(nfs_data.root, fh);
 
 	return status;
 }

fs/nfs/pagelist.c

@@ -21,11 +21,6 @@
 
 #define NFS_PARANOIA 1
 
-/*
- * Spinlock
- */
-spinlock_t nfs_wreq_lock = SPIN_LOCK_UNLOCKED;
-
 static kmem_cache_t *nfs_page_cachep;
 
 static inline struct nfs_page *
@@ -36,7 +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;
 }
@@ -62,7 +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)
 {
@@ -94,33 +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;
-	req->wb_count   = 1;
-	server->rpc_ops->request_init(req, file);
+	atomic_set(&req->wb_count, 1);
+	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;
@@ -137,12 +136,8 @@ void nfs_clear_request(struct nfs_page *req)
 void
 nfs_release_request(struct nfs_page *req)
 {
-	spin_lock(&nfs_wreq_lock);
-	if (--req->wb_count) {
-		spin_unlock(&nfs_wreq_lock);
+	if (!atomic_dec_and_test(&req->wb_count))
 		return;
-	}
-	spin_unlock(&nfs_wreq_lock);
 
 #ifdef NFS_PARANOIA
 	BUG_ON (!list_empty(&req->wb_list));
@@ -151,6 +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);
 }
 
@@ -194,12 +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));
 }
 
 /**
@@ -224,7 +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;
@@ -254,7 +254,7 @@ nfs_coalesce_requests(struct list_head *head, struct list_head *dst,
  * If the number of requests is set to 0, the entire address_space
  * starting at index idx_start, is scanned.
  * The requests are *not* checked to ensure that they form a contiguous set.
- * You must be holding the nfs_wreq_lock when calling this function
+ * You must be holding the inode's req_lock when calling this function
  */
 int
 nfs_scan_list(struct list_head *head, struct list_head *dst,

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);
@@ -400,6 +386,8 @@ nfs_proc_symlink(struct inode *dir, struct qstr *name, struct qstr *path,
 	};
 	int			status;
 
+	if (path->len > NFS2_MAXPATHLEN)
+		return -ENAMETOOLONG;
 	dprintk("NFS call  symlink %s -> %s\n", name->name, path->name);
 	fattr->valid = 0;
 	status = rpc_call(NFS_CLIENT(dir), NFSPROC_SYMLINK, &arg, NULL, 0);
@@ -619,27 +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)
 {
@@ -680,7 +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

@@ -63,6 +63,8 @@
 #include <linux/smp_lock.h>
 #include <linux/mempool.h>
 
+#include "delegation.h"
+
 #define NFSDBG_FACILITY		NFSDBG_PAGECACHE
 
 #define MIN_POOL_WRITE		(32)
@@ -71,7 +73,8 @@
 /*
  * 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 +176,7 @@ 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 +190,10 @@ 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 +212,7 @@ 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 +238,19 @@ 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 +277,7 @@ 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 +312,21 @@ 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 +335,7 @@ int nfs_writepage(struct page *page, struct writeback_control *wbc)
 		}
 	}
 	unlock_kernel();
+	put_nfs_open_context(ctx);
 out:
 	unlock_page(page);
 	if (inode_referenced)
@@ -374,8 +384,7 @@ int nfs_writepages(struct address_space *mapping, struct writeback_control *wbc)
 /*
  * Insert a write request into an inode
  */
-static inline int
-nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
+static int nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
 {
 	struct nfs_inode *nfsi = NFS_I(inode);
 	int error;
@@ -387,33 +396,33 @@ nfs_inode_add_request(struct inode *inode, struct nfs_page *req)
 	if (!nfsi->npages) {
 		igrab(inode);
 		nfs_begin_data_update(inode);
+		if (nfs_have_delegation(inode, FMODE_WRITE))
+			nfsi->change_attr++;
 	}
 	nfsi->npages++;
-	req->wb_count++;
+	atomic_inc(&req->wb_count);
 	return 0;
 }
 
 /*
  * 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 nfs_inode *nfsi;
-	struct inode *inode;
+	struct inode *inode = req->wb_context->dentry->d_inode;
+	struct nfs_inode *nfsi = NFS_I(inode);
 
 	BUG_ON (!NFS_WBACK_BUSY(req));
-	spin_lock(&nfs_wreq_lock);
-	inode = req->wb_inode;
-	nfsi = NFS_I(inode);
+
+	spin_lock(&nfsi->req_lock);
 	radix_tree_delete(&nfsi->nfs_page_tree, req->wb_index);
 	nfsi->npages--;
 	if (!nfsi->npages) {
-		spin_unlock(&nfs_wreq_lock);
+		spin_unlock(&nfsi->req_lock);
 		nfs_end_data_update_defer(inode);
 		iput(inode);
 	} else
-		spin_unlock(&nfs_wreq_lock);
+		spin_unlock(&nfsi->req_lock);
 	nfs_clear_request(req);
 	nfs_release_request(req);
 }
@@ -429,7 +438,7 @@ _nfs_find_request(struct inode *inode, unsigned long index)
 
 	req = (struct nfs_page*)radix_tree_lookup(&nfsi->nfs_page_tree, index);
 	if (req)
-		req->wb_count++;
+		atomic_inc(&req->wb_count);
 	return req;
 }
 
@@ -437,10 +446,11 @@ static struct nfs_page *
 nfs_find_request(struct inode *inode, unsigned long index)
 {
 	struct nfs_page		*req;
+	struct nfs_inode	*nfsi = NFS_I(inode);
 
-	spin_lock(&nfs_wreq_lock);
+	spin_lock(&nfsi->req_lock);
 	req = _nfs_find_request(inode, index);
-	spin_unlock(&nfs_wreq_lock);
+	spin_unlock(&nfsi->req_lock);
 	return req;
 }
 
@@ -450,13 +460,13 @@ 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(&nfs_wreq_lock);
+	spin_lock(&nfsi->req_lock);
 	nfs_list_add_request(req, &nfsi->dirty);
 	nfsi->ndirty++;
-	spin_unlock(&nfs_wreq_lock);
+	spin_unlock(&nfsi->req_lock);
 	inc_page_state(nr_dirty);
 	mark_inode_dirty(inode);
 }
@@ -467,7 +477,7 @@ 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;
 }
 
@@ -478,13 +488,13 @@ 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(&nfs_wreq_lock);
+	spin_lock(&nfsi->req_lock);
 	nfs_list_add_request(req, &nfsi->commit);
 	nfsi->ncommit++;
-	spin_unlock(&nfs_wreq_lock);
+	spin_unlock(&nfsi->req_lock);
 	inc_page_state(nr_unstable);
 	mark_inode_dirty(inode);
 }
@@ -509,7 +519,7 @@ nfs_wait_on_requests(struct inode *inode, unsigned long idx_start, unsigned int
 	else
 		idx_end = idx_start + npages - 1;
 
-	spin_lock(&nfs_wreq_lock);
+	spin_lock(&nfsi->req_lock);
 	next = idx_start;
 	while (radix_tree_gang_lookup(&nfsi->nfs_page_tree, (void **)&req, next, 1)) {
 		if (req->wb_index > idx_end)
@@ -519,16 +529,16 @@ nfs_wait_on_requests(struct inode *inode, unsigned long idx_start, unsigned int
 		if (!NFS_WBACK_BUSY(req))
 			continue;
 
-		req->wb_count++;
-		spin_unlock(&nfs_wreq_lock);
+		atomic_inc(&req->wb_count);
+		spin_unlock(&nfsi->req_lock);
 		error = nfs_wait_on_request(req);
 		nfs_release_request(req);
 		if (error < 0)
 			return error;
-		spin_lock(&nfs_wreq_lock);
+		spin_lock(&nfsi->req_lock);
 		res++;
 	}
-	spin_unlock(&nfs_wreq_lock);
+	spin_unlock(&nfsi->req_lock);
 	return res;
 }
 
@@ -619,11 +629,12 @@ 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);
 	struct nfs_page		*req, *new = NULL;
 	unsigned long		rqend, end;
 
@@ -635,19 +646,19 @@ nfs_update_request(struct file* file, struct inode *inode, struct page *page,
 		/* Loop over all inode entries and see if we find
 		 * A request for the page we wish to update
 		 */
-		spin_lock(&nfs_wreq_lock);
+		spin_lock(&nfsi->req_lock);
 		req = _nfs_find_request(inode, page->index);
 		if (req) {
 			if (!nfs_lock_request_dontget(req)) {
 				int error;
-				spin_unlock(&nfs_wreq_lock);
+				spin_unlock(&nfsi->req_lock);
 				error = nfs_wait_on_request(req);
 				nfs_release_request(req);
 				if (error < 0)
 					return ERR_PTR(error);
 				continue;
 			}
-			spin_unlock(&nfs_wreq_lock);
+			spin_unlock(&nfsi->req_lock);
 			if (new)
 				nfs_release_request(new);
 			break;
@@ -658,23 +669,19 @@ nfs_update_request(struct file* file, struct inode *inode, struct page *page,
 			nfs_lock_request_dontget(new);
 			error = nfs_inode_add_request(inode, new);
 			if (error) {
-				spin_unlock(&nfs_wreq_lock);
+				spin_unlock(&nfsi->req_lock);
 				nfs_unlock_request(new);
 				return ERR_PTR(error);
 			}
-			spin_unlock(&nfs_wreq_lock);
+			spin_unlock(&nfsi->req_lock);
 			nfs_mark_request_dirty(new);
 			return new;
 		}
-		spin_unlock(&nfs_wreq_lock);
+		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.
@@ -684,7 +691,7 @@ 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) {
@@ -705,9 +712,9 @@ 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;
@@ -721,7 +728,7 @@ 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);
 	}
@@ -737,6 +744,7 @@ 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;
@@ -747,7 +755,7 @@ 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);
@@ -784,7 +792,7 @@ 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;
@@ -860,16 +868,15 @@ 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;
@@ -1029,7 +1036,7 @@ 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;
 	}
@@ -1054,16 +1061,15 @@ 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)
@@ -1104,16 +1110,15 @@ 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);
@@ -1232,7 +1237,7 @@ 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.
@@ -1246,7 +1251,7 @@ 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;
@@ -1313,13 +1318,12 @@ 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;
@@ -1347,13 +1351,14 @@ nfs_commit_done(struct rpc_task *task)
 int nfs_flush_inode(struct inode *inode, unsigned long idx_start,
 		   unsigned int npages, int how)
 {
+	struct nfs_inode *nfsi = NFS_I(inode);
 	LIST_HEAD(head);
 	int			res,
 				error = 0;
 
-	spin_lock(&nfs_wreq_lock);
+	spin_lock(&nfsi->req_lock);
 	res = nfs_scan_dirty(inode, &head, idx_start, npages);
-	spin_unlock(&nfs_wreq_lock);
+	spin_unlock(&nfsi->req_lock);
 	if (res)
 		error = nfs_flush_list(&head, NFS_SERVER(inode)->wpages, how);
 	if (error < 0)
@@ -1365,18 +1370,19 @@ int nfs_flush_inode(struct inode *inode, unsigned long idx_start,
 int nfs_commit_inode(struct inode *inode, unsigned long idx_start,
 		    unsigned int npages, int how)
 {
+	struct nfs_inode *nfsi = NFS_I(inode);
 	LIST_HEAD(head);
 	int			res,
 				error = 0;
 
-	spin_lock(&nfs_wreq_lock);
+	spin_lock(&nfsi->req_lock);
 	res = nfs_scan_commit(inode, &head, idx_start, npages);
 	if (res) {
 		res += nfs_scan_commit(inode, &head, 0, 0);
-		spin_unlock(&nfs_wreq_lock);
+		spin_unlock(&nfsi->req_lock);
 		error = nfs_commit_list(&head, how);
 	} else
-		spin_unlock(&nfs_wreq_lock);
+		spin_unlock(&nfsi->req_lock);
 	if (error < 0)
 		return error;
 	return res;

include/linux/nfs.h

@@ -8,6 +8,7 @@
 #define _LINUX_NFS_H
 
 #include <linux/sunrpc/msg_prot.h>
+#include <linux/string.h>
 
 #define NFS_PROGRAM	100003
 #define NFS_PORT	2049
@@ -138,6 +139,22 @@ struct nfs_fh {
 	unsigned char		data[NFS_MAXFHSIZE];
 };
 
+/*
+ * Returns a zero iff the size and data fields match.
+ * Checks only "size" bytes in the data field.
+ */
+static inline int nfs_compare_fh(const struct nfs_fh *a, const struct nfs_fh *b)
+{
+	return a->size != b->size || memcmp(a->data, b->data, a->size) != 0;
+}
+
+static inline void nfs_copy_fh(struct nfs_fh *target, const struct nfs_fh *source)
+{
+	target->size = source->size;
+	memcpy(target->data, source->data, source->size);
+}
+
+
 /*
  * This is really a general kernel constant, but since nothing like
  * this is defined in the kernel headers, I have to do it here.

include/linux/nfs4.h

@@ -18,6 +18,7 @@
 
 #define NFS4_VERIFIER_SIZE	8
 #define NFS4_FHSIZE		128
+#define NFS4_MAXPATHLEN		PATH_MAX
 #define NFS4_MAXNAMLEN		NAME_MAX
 
 #define NFS4_ACCESS_READ        0x0001
@@ -354,7 +355,7 @@ enum {
 	NFSPROC4_CLNT_COMMIT,
 	NFSPROC4_CLNT_OPEN,
 	NFSPROC4_CLNT_OPEN_CONFIRM,
-	NFSPROC4_CLNT_OPEN_RECLAIM,
+	NFSPROC4_CLNT_OPEN_NOATTR,
 	NFSPROC4_CLNT_OPEN_DOWNGRADE,
 	NFSPROC4_CLNT_CLOSE,
 	NFSPROC4_CLNT_SETATTR,
@@ -372,12 +373,14 @@ enum {
 	NFSPROC4_CLNT_REMOVE,
 	NFSPROC4_CLNT_RENAME,
 	NFSPROC4_CLNT_LINK,
+	NFSPROC4_CLNT_SYMLINK,
 	NFSPROC4_CLNT_CREATE,
 	NFSPROC4_CLNT_PATHCONF,
 	NFSPROC4_CLNT_STATFS,
 	NFSPROC4_CLNT_READLINK,
 	NFSPROC4_CLNT_READDIR,
 	NFSPROC4_CLNT_SERVER_CAPS,
+	NFSPROC4_CLNT_DELEGRETURN,
 };
 
 #endif

include/linux/nfs_fs.h

@@ -28,6 +28,7 @@
 #include <linux/nfs3.h>
 #include <linux/nfs4.h>
 #include <linux/nfs_xdr.h>
+#include <linux/rwsem.h>
 #include <linux/workqueue.h>
 
 /*
@@ -75,15 +76,33 @@
 #ifdef __KERNEL__
 
 /*
- * NFSv3 Access mode cache
+ * NFSv3/v4 Access mode cache entry
  */
-struct nfs_access_cache {
+struct nfs_access_entry {
 	unsigned long		jiffies;
 	struct rpc_cred *	cred;
 	int			mask;
-	int			err;
+};
+
+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;
 };
 
+/*
+ * NFSv4 delegation
+ */
+struct nfs_delegation;
+
 /*
  * nfs fs inode data in memory
  */
@@ -137,7 +156,7 @@ struct nfs_inode {
 	 */
 	atomic_t		data_updates;
 
-	struct nfs_access_cache	cache_access;
+	struct nfs_access_entry	cache_access;
 
 	/*
 	 * This is the cookie verifier used for NFSv3 readdir
@@ -148,6 +167,7 @@ struct nfs_inode {
 	/*
 	 * This is the list of dirty unwritten pages.
 	 */
+	spinlock_t		req_lock;
 	struct list_head	dirty;
 	struct list_head	commit;
 	struct radix_tree_root	nfs_page_tree;
@@ -156,14 +176,17 @@ 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;
 
 #ifdef CONFIG_NFS_V4
         /* NFSv4 state */
 	struct list_head	open_states;
+	struct nfs_delegation	*delegation;
+	int			 delegation_state;
+	struct rw_semaphore	rwsem;
 #endif /* CONFIG_NFS_V4*/
 
 	struct inode		vfs_inode;
@@ -268,9 +291,12 @@ 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_access_get_cached(struct inode *, struct rpc_cred *, struct nfs_access_entry *);
+extern void nfs_access_add_cache(struct inode *, struct nfs_access_entry *);
 extern int nfs_open(struct inode *, struct file *);
 extern int nfs_release(struct inode *, struct file *);
+extern int nfs_attribute_timeout(struct inode *inode);
+extern int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode);
 extern int __nfs_revalidate_inode(struct nfs_server *, struct inode *);
 extern int nfs_setattr(struct dentry *, struct iattr *);
 extern void nfs_begin_attr_update(struct inode *);
@@ -278,6 +304,12 @@ 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 +321,15 @@ 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;
 }
 
 /*
@@ -418,28 +449,6 @@ extern int  nfsroot_mount(struct sockaddr_in *, char *, struct nfs_fh *,
  * inline functions
  */
 
-static inline int nfs_attribute_timeout(struct inode *inode)
-{
-	struct nfs_inode *nfsi = NFS_I(inode);
-
-	return time_after(jiffies, nfsi->read_cache_jiffies+nfsi->attrtimeo);
-}
-
-/**
- * nfs_revalidate_inode - Revalidate the inode attributes
- * @server - pointer to nfs_server struct
- * @inode - pointer to inode struct
- *
- * Updates inode attribute information by retrieving the data from the server.
- */
-static inline int nfs_revalidate_inode(struct nfs_server *server, struct inode *inode)
-{
-	if (!(NFS_FLAGS(inode) & (NFS_INO_INVALID_ATTR|NFS_INO_INVALID_DATA))
-			&& !nfs_attribute_timeout(inode))
-		return NFS_STALE(inode) ? -ESTALE : 0;
-	return __nfs_revalidate_inode(server, inode);
-}
-
 static inline loff_t
 nfs_size_to_loff_t(__u64 size)
 {
@@ -507,8 +516,6 @@ struct idmap;
 
 enum nfs4_client_state {
 	NFS4CLNT_OK  = 0,
-	NFS4CLNT_NEW,
-	NFS4CLNT_SETUP_STATE,
 };
 
 /*
@@ -520,7 +527,6 @@ struct nfs4_client {
 	u64			cl_clientid;	/* constant */
 	nfs4_verifier		cl_confirm;
 	unsigned long		cl_state;
-	long			cl_generation;
 
 	u32			cl_lockowner_id;
 
@@ -530,6 +536,7 @@ struct nfs4_client {
 	 */
 	struct rw_semaphore	cl_sem;
 
+	struct list_head	cl_delegations;
 	struct list_head	cl_state_owners;
 	struct list_head	cl_unused;
 	int			cl_nunused;
@@ -573,12 +580,11 @@ 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;
+	struct list_head     so_delegations;
 };
 
 /*
@@ -596,7 +602,8 @@ 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;
@@ -606,6 +613,7 @@ struct nfs4_lock_state {
 /* bits for nfs4_state->flags */
 enum {
 	LK_STATE_IN_USE,
+	NFS_DELEGATED_STATE,
 };
 
 struct nfs4_state {
@@ -629,6 +637,11 @@ 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 +652,12 @@ 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 +669,8 @@ 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 struct nfs4_client *nfs4_find_client(struct in_addr *);
 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 +680,13 @@ 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 +697,7 @@ 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
 
@@ -697,6 +714,7 @@ struct nfs4_mount_data;
 #define NFSDBG_XDR		0x0020
 #define NFSDBG_FILE		0x0040
 #define NFSDBG_ROOT		0x0080
+#define NFSDBG_CALLBACK		0x0100
 #define NFSDBG_ALL		0xFFFF
 
 #ifdef __KERNEL__

include/linux/nfs_fs_sb.h

@@ -18,6 +18,7 @@ struct nfs_server {
 	unsigned int		rpages;		/* read size (in pages) */
 	unsigned int		wsize;		/* write size */
 	unsigned int		wpages;		/* write size (in pages) */
+	unsigned int		wtmult;		/* server disk block size */
 	unsigned int		dtsize;		/* readdir size */
 	unsigned int		bsize;		/* server block size */
 	unsigned int		acregmin;	/* attr cache timeouts */

include/linux/nfs_page.h

@@ -29,19 +29,14 @@
 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 */
-				wb_bytes,	/* Length of request */
-				wb_count;	/* reference count */
+				wb_bytes;	/* Length of request */
+	atomic_t		wb_count;	/* reference count */
 	unsigned long		wb_flags;
 	struct nfs_writeverf	wb_verf;	/* Commit cookie */
 };
@@ -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);
 
@@ -64,8 +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	spinlock_t nfs_wreq_lock;
+extern	void nfs_unlock_request(struct nfs_page *req);
 
 /*
  * Lock the page of an asynchronous request without incrementing the wb_count
@@ -86,23 +82,10 @@ nfs_lock_request(struct nfs_page *req)
 {
 	if (test_and_set_bit(PG_BUSY, &req->wb_flags))
 		return 0;
-	req->wb_count++;
+	atomic_inc(&req->wb_count);
 	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

@@ -99,20 +99,21 @@ struct nfs4_change_info {
  * Arguments to the open call.
  */
 struct nfs_openargs {
-	struct nfs_fh *         fh;
+	const struct nfs_fh *	fh;
 	__u32                   seqid;
-	__u32                   share_access;
+	int			open_flags;
 	__u64                   clientid;
 	__u32                   id;
-	__u32                   opentype;
-	__u32                   createmode;
 	union {
 		struct iattr *  attrs;    /* UNCHECKED, GUARDED */
 		nfs4_verifier   verifier; /* EXCLUSIVE */
+		nfs4_stateid	delegation;		/* CLAIM_DELEGATE_CUR */
+		int		delegation_type;	/* CLAIM_PREVIOUS */
 	} u;
 	const struct qstr *	name;
 	const struct nfs_server *server;	 /* Needed for ID mapping */
 	const u32 *		bitmask;
+	__u32			claim;
 };
 
 struct nfs_openres {
@@ -122,13 +123,17 @@ struct nfs_openres {
 	__u32                   rflags;
 	struct nfs_fattr *      f_attr;
 	const struct nfs_server *server;
+	int			delegation_type;
+	nfs4_stateid		delegation;
+	__u32			do_recall;
+	__u64			maxsize;
 };
 
 /*
  * Arguments to the open_confirm call.
  */
 struct nfs_open_confirmargs {
-	struct nfs_fh *         fh;
+	const struct nfs_fh *	fh;
 	nfs4_stateid            stateid;
 	__u32                   seqid;
 };
@@ -137,19 +142,6 @@ struct nfs_open_confirmres {
 	nfs4_stateid            stateid;
 };
 
-/*
- * Arguments to the open_reclaim call.
- */
-struct nfs_open_reclaimargs {
-	struct nfs_fh *		fh;
-	__u64			clientid;
-	__u32			seqid;
-	__u32			id;
-	__u32			share_access;
-	__u32			claim;
-	const __u32 *		bitmask;
-};
-
 /*
  * Arguments to the close call.
  */
@@ -157,7 +149,7 @@ struct nfs_closeargs {
 	struct nfs_fh *         fh;
 	nfs4_stateid            stateid;
 	__u32                   seqid;
-	__u32			share_access;
+	int			open_flags;
 };
 
 struct nfs_closeres {
@@ -224,6 +216,11 @@ struct nfs_lockres {
 	const struct nfs_server *	server;
 };
 
+struct nfs4_delegreturnargs {
+	const struct nfs_fh *fhandle;
+	const nfs4_stateid *stateid;
+};
+
 /*
  * Arguments to the read call.
  */
@@ -235,8 +232,7 @@ 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 +255,7 @@ 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 +592,15 @@ 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 {
@@ -657,6 +654,8 @@ struct nfs_write_data {
 	void (*complete) (struct nfs_write_data *, int);
 };
 
+struct nfs_access_entry;
+
 /*
  * RPC procedure vector for NFSv2/NFSv3 demuxing
  */
@@ -667,16 +666,17 @@ 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 rpc_cred *, int);
+	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 *);
@@ -708,8 +708,6 @@ 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 *);
 };
 

include/linux/sunrpc/gss_asn1.h

@@ -69,7 +69,6 @@ u32 g_verify_token_header(
      struct xdr_netobj *mech,
      int *body_size,
      unsigned char **buf_in,
-     int tok_type,
      int toksize);
 
 u32 g_get_mech_oid(struct xdr_netobj *mech, struct xdr_netobj * in_buf);

include/linux/sunrpc/gss_spkm3.h

+/*
+ *  linux/include/linux/sunrpc/gss_spkm3.h
+ *
+ *  Copyright (c) 2000 The Regents of the University of Michigan.
+ *  All rights reserved.
+ *
+ *  Andy Adamson   <andros@umich.edu>
+ */
+
+#include <linux/sunrpc/auth_gss.h>
+#include <linux/sunrpc/gss_err.h>
+#include <linux/sunrpc/gss_asn1.h>
+
+struct spkm3_ctx {
+	struct xdr_netobj	ctx_id; /* per message context id */
+	int			qop;         /* negotiated qop */
+	struct xdr_netobj	mech_used;
+	unsigned int		ret_flags ;
+	unsigned int		req_flags ;
+	struct xdr_netobj	share_key;
+	int			conf_alg;
+	struct crypto_tfm*	derived_conf_key;
+	int			intg_alg;
+	struct crypto_tfm*	derived_integ_key;
+	int			keyestb_alg;   /* alg used to get share_key */
+	int			owf_alg;   /* one way function */
+};
+
+/* from openssl/objects.h */
+/* XXX need SEAL_ALG_NONE */
+#define NID_md5		4
+#define NID_dhKeyAgreement	28 
+#define NID_des_cbc		31 
+#define NID_sha1		64
+#define NID_cast5_cbc		108
+
+/* SPKM InnerContext Token types */
+
+#define SPKM_ERROR_TOK	3
+#define SPKM_MIC_TOK	4
+#define SPKM_WRAP_TOK	5
+#define SPKM_DEL_TOK	6
+
+u32 spkm3_make_token(struct spkm3_ctx *ctx, int qop_req, struct xdr_buf * text, struct xdr_netobj * token, int toktype);
+
+u32 spkm3_read_token(struct spkm3_ctx *ctx, struct xdr_netobj *read_token, struct xdr_buf *message_buffer, int *qop_state, int toktype);
+
+#define CKSUMTYPE_RSA_MD5            0x0007
+
+s32 make_checksum(s32 cksumtype, char *header, int hdrlen, struct xdr_buf *body,
+                   struct xdr_netobj *cksum);
+void asn1_bitstring_len(struct xdr_netobj *in, int *enclen, int *zerobits);
+int decode_asn1_bitstring(struct xdr_netobj *out, char *in, int enclen, 
+                   int explen);
+void spkm3_mic_header(unsigned char **hdrbuf, unsigned int *hdrlen, 
+                   unsigned char *ctxhdr, int elen, int zbit);
+void spkm3_make_mic_token(unsigned  char **tokp, int toklen, 
+                   struct xdr_netobj *mic_hdr,
+                   struct xdr_netobj *md5cksum, int md5elen, int md5zbit);
+u32 spkm3_verify_mic_token(unsigned char **tokp, int *mic_hdrlen, 
+                   unsigned char **cksum);

include/linux/sunrpc/sched.h

@@ -113,6 +113,7 @@ typedef void			(*rpc_action)(struct rpc_task *);
 #define RPC_TASK_DYNAMIC	0x0080		/* task was kmalloc'ed */
 #define RPC_TASK_KILLED		0x0100		/* task was killed */
 #define RPC_TASK_SOFT		0x0200		/* Use soft timeouts */
+#define RPC_TASK_NOINTR		0x0400		/* uninterruptible task */
 
 #define RPC_IS_ASYNC(t)		((t)->tk_flags & RPC_TASK_ASYNC)
 #define RPC_IS_SETUID(t)	((t)->tk_flags & RPC_TASK_SETUID)
@@ -123,6 +124,7 @@ typedef void			(*rpc_action)(struct rpc_task *);
 #define RPC_IS_ACTIVATED(t)	((t)->tk_active)
 #define RPC_DO_CALLBACK(t)	((t)->tk_callback != NULL)
 #define RPC_IS_SOFT(t)		((t)->tk_flags & RPC_TASK_SOFT)
+#define RPC_TASK_UNINTERRUPTIBLE(t) ((t)->tk_flags & RPC_TASK_NOINTR)
 
 #define RPC_TASK_SLEEPING	0
 #define RPC_TASK_RUNNING	1

include/linux/sunrpc/svc.h

@@ -87,6 +87,14 @@ static inline u32 svc_getu32(struct kvec *iov)
 	iov->iov_len -= sizeof(u32);
 	return val;
 }
+
+static inline void svc_ungetu32(struct kvec *iov)
+{
+	u32 *vp = (u32 *)iov->iov_base;
+	iov->iov_base = (void *)(vp - 1);
+	iov->iov_len += sizeof(*vp);
+}
+
 static inline void svc_putu32(struct kvec *iov, u32 val)
 {
 	u32 *vp = iov->iov_base + iov->iov_len;
@@ -243,6 +251,8 @@ struct svc_program {
 	char *			pg_name;	/* service name */
 	char *			pg_class;	/* class name: services sharing authentication */
 	struct svc_stat *	pg_stats;	/* rpc statistics */
+	/* Override authentication. NULL means use default */
+	int			(*pg_authenticate)(struct svc_rqst *, u32 *);
 };
 
 /*

net/sunrpc/auth_gss/Makefile

@@ -12,3 +12,7 @@ obj-$(CONFIG_RPCSEC_GSS_KRB5) += rpcsec_gss_krb5.o
 rpcsec_gss_krb5-objs := gss_krb5_mech.o gss_krb5_seal.o gss_krb5_unseal.o \
 	gss_krb5_seqnum.o
 
+obj-$(CONFIG_RPCSEC_GSS_SPKM3) += rpcsec_gss_spkm3.o
+
+rpcsec_gss_spkm3-objs := gss_spkm3_mech.o gss_spkm3_seal.o gss_spkm3_unseal.o \
+	gss_spkm3_token.o

net/sunrpc/auth_gss/auth_gss.c

@@ -397,7 +397,7 @@ gss_upcall(struct rpc_clnt *clnt, struct rpc_task *task, struct rpc_cred *cred)
 		spin_unlock(&gss_auth->lock);
 	}
 	gss_release_msg(gss_msg);
-	dprintk("RPC: %4u gss_upcall for uid %u result %d", task->tk_pid,
+	dprintk("RPC: %4u gss_upcall for uid %u result %d\n", task->tk_pid,
 			uid, res);
 	return res;
 out_sleep:

net/sunrpc/auth_gss/gss_generic_token.c

@@ -179,7 +179,7 @@ EXPORT_SYMBOL(g_make_token_header);
  */
 u32
 g_verify_token_header(struct xdr_netobj *mech, int *body_size,
-		      unsigned char **buf_in, int tok_type, int toksize)
+		      unsigned char **buf_in, int toksize)
 {
 	unsigned char *buf = *buf_in;
 	int seqsize;

net/sunrpc/auth_gss/gss_krb5_unseal.c

@@ -96,7 +96,7 @@ krb5_read_token(struct krb5_ctx *ctx,
 
 	dprintk("RPC:      krb5_read_token\n");
 
-	if (g_verify_token_header(&ctx->mech_used, &bodysize, &ptr, toktype,
+	if (g_verify_token_header(&ctx->mech_used, &bodysize, &ptr,
 					read_token->len))
 		goto out;
 

net/sunrpc/auth_gss/gss_spkm3_mech.c

+/*
+ *  linux/net/sunrpc/gss_spkm3_mech.c
+ *
+ *  Copyright (c) 2003 The Regents of the University of Michigan.
+ *  All rights reserved.
+ *
+ *  Andy Adamson <andros@umich.edu>
+ *  J. Bruce Fields <bfields@umich.edu>
+ *
+ *  Redistribution and use in source and binary forms, with or without
+ *  modification, are permitted provided that the following conditions
+ *  are met:
+ *
+ *  1. Redistributions of source code must retain the above copyright
+ *     notice, this list of conditions and the following disclaimer.
+ *  2. Redistributions in binary form must reproduce the above copyright
+ *     notice, this list of conditions and the following disclaimer in the
+ *     documentation and/or other materials provided with the distribution.
+ *  3. Neither the name of the University nor the names of its
+ *     contributors may be used to endorse or promote products derived
+ *     from this software without specific prior written permission.
+ *
+ *  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
+ *  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ *  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ *  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ *  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ *  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ *  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ *  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ *  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+ *  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ *  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/sunrpc/auth.h>
+#include <linux/in.h>
+#include <linux/sunrpc/svcauth_gss.h>
+#include <linux/sunrpc/gss_spkm3.h>
+#include <linux/sunrpc/xdr.h>
+#include <linux/crypto.h>
+
+#ifdef RPC_DEBUG
+# define RPCDBG_FACILITY	RPCDBG_AUTH
+#endif
+
+struct xdr_netobj gss_mech_spkm3_oid =
+   {7, "\053\006\001\005\005\001\003"};
+
+static inline int
+get_bytes(char **ptr, const char *end, void *res, int len)
+{
+	char *p, *q;
+	p = *ptr;
+	q = p + len;
+	if (q > end || q < p)
+		return -1;
+	memcpy(res, p, len);
+	*ptr = q;
+	return 0;
+}
+
+static inline int
+get_netobj(char **ptr, const char *end, struct xdr_netobj *res)
+{
+	char *p, *q;
+	p = *ptr;
+	if (get_bytes(&p, end, &res->len, sizeof(res->len)))
+		return -1;
+	q = p + res->len;
+	if(res->len == 0)
+		goto out_nocopy;
+	if (q > end || q < p)
+		return -1;
+	if (!(res->data = kmalloc(res->len, GFP_KERNEL)))
+		return -1;
+	memcpy(res->data, p, res->len);
+out_nocopy:
+	*ptr = q;
+	return 0;
+}
+
+static inline int
+get_key(char **p, char *end, struct crypto_tfm **res, int *resalg)
+{
+	struct xdr_netobj	key = {
+		.len = 0,
+		.data = NULL,
+	};
+	int			alg_mode,setkey = 0;
+	char			*alg_name;
+
+	if (get_bytes(p, end, resalg, sizeof(int)))
+		goto out_err;
+	if ((get_netobj(p, end, &key)))
+		goto out_err;
+
+	switch (*resalg) {
+		case NID_des_cbc:
+			alg_name = "des";
+			alg_mode = CRYPTO_TFM_MODE_CBC;
+			setkey = 1;
+			break;
+		case NID_md5:
+			if (key.len == 0) {
+				dprintk("RPC: SPKM3 get_key: NID_md5 zero Key length\n");
+			}
+			alg_name = "md5";
+			alg_mode = 0;
+			setkey = 0;
+			break;
+		case NID_cast5_cbc:
+			dprintk("RPC: SPKM3 get_key: case cast5_cbc, UNSUPPORTED \n");
+			goto out_err;
+			break;
+		default:
+			dprintk("RPC: SPKM3 get_key: unsupported algorithm %d", *resalg);
+			goto out_err_free_key;
+	}
+	if (!(*res = crypto_alloc_tfm(alg_name, alg_mode)))
+		goto out_err_free_key;
+	if (setkey) {
+		if (crypto_cipher_setkey(*res, key.data, key.len))
+			goto out_err_free_tfm;
+	}
+
+	if(key.len > 0)
+		kfree(key.data);
+	return 0;
+
+out_err_free_tfm:
+	crypto_free_tfm(*res);
+out_err_free_key:
+	if(key.len > 0)
+		kfree(key.data);
+out_err:
+	return -1;
+}
+
+static u32
+gss_import_sec_context_spkm3(struct xdr_netobj *inbuf,
+				struct gss_ctx *ctx_id)
+{
+	char	*p = inbuf->data;
+	char	*end = inbuf->data + inbuf->len;
+	struct	spkm3_ctx *ctx;
+
+	if (!(ctx = kmalloc(sizeof(*ctx), GFP_KERNEL)))
+		goto out_err;
+	memset(ctx, 0, sizeof(*ctx));
+
+	if (get_netobj(&p, end, &ctx->ctx_id))
+		goto out_err_free_ctx;
+
+	if (get_bytes(&p, end, &ctx->qop, sizeof(ctx->qop)))
+		goto out_err_free_ctx_id;
+
+	if (get_netobj(&p, end, &ctx->mech_used))
+		goto out_err_free_mech;
+
+	if (get_bytes(&p, end, &ctx->ret_flags, sizeof(ctx->ret_flags)))
+		goto out_err_free_mech;
+
+	if (get_bytes(&p, end, &ctx->req_flags, sizeof(ctx->req_flags)))
+		goto out_err_free_mech;
+
+	if (get_netobj(&p, end, &ctx->share_key))
+		goto out_err_free_s_key;
+
+	if (get_key(&p, end, &ctx->derived_conf_key, &ctx->conf_alg)) {
+		dprintk("RPC: SPKM3 confidentiality key will be NULL\n");
+	}
+
+	if (get_key(&p, end, &ctx->derived_integ_key, &ctx->intg_alg)) {
+		dprintk("RPC: SPKM3 integrity key will be NULL\n");
+	}
+
+	if (get_bytes(&p, end, &ctx->owf_alg, sizeof(ctx->owf_alg)))
+		goto out_err_free_s_key;
+
+	if (get_bytes(&p, end, &ctx->owf_alg, sizeof(ctx->owf_alg)))
+		goto out_err_free_s_key;
+
+	if (p != end)
+		goto out_err_free_s_key;
+
+	ctx_id->internal_ctx_id = ctx;
+
+	dprintk("Succesfully imported new spkm context.\n");
+	return 0;
+
+out_err_free_s_key:
+	kfree(ctx->share_key.data);
+out_err_free_mech:
+	kfree(ctx->mech_used.data);
+out_err_free_ctx_id:
+	kfree(ctx->ctx_id.data);
+out_err_free_ctx:
+	kfree(ctx);
+out_err:
+	return GSS_S_FAILURE;
+}
+
+void
+gss_delete_sec_context_spkm3(void *internal_ctx) {
+	struct spkm3_ctx *sctx = internal_ctx;
+
+	if(sctx->derived_integ_key)
+		crypto_free_tfm(sctx->derived_integ_key);
+	if(sctx->derived_conf_key)
+		crypto_free_tfm(sctx->derived_conf_key);
+	if(sctx->share_key.data)
+		kfree(sctx->share_key.data);
+	if(sctx->mech_used.data)
+		kfree(sctx->mech_used.data);
+	kfree(sctx);
+}
+
+u32
+gss_verify_mic_spkm3(struct gss_ctx		*ctx,
+			struct xdr_buf		*signbuf,
+			struct xdr_netobj	*checksum,
+			u32		*qstate) {
+	u32 maj_stat = 0;
+	int qop_state = 0;
+	struct spkm3_ctx *sctx = ctx->internal_ctx_id;
+
+	dprintk("RPC: gss_verify_mic_spkm3 calling spkm3_read_token\n");
+	maj_stat = spkm3_read_token(sctx, checksum, signbuf, &qop_state,
+				   SPKM_MIC_TOK);
+
+	if (!maj_stat && qop_state)
+	    *qstate = qop_state;
+
+	dprintk("RPC: gss_verify_mic_spkm3 returning %d\n", maj_stat);
+	return maj_stat;
+}
+
+u32
+gss_get_mic_spkm3(struct gss_ctx	*ctx,
+		     u32		qop,
+		     struct xdr_buf	*message_buffer,
+		     struct xdr_netobj	*message_token) {
+	u32 err = 0;
+	struct spkm3_ctx *sctx = ctx->internal_ctx_id;
+
+	dprintk("RPC: gss_get_mic_spkm3\n");
+
+	err = spkm3_make_token(sctx, qop, message_buffer,
+			      message_token, SPKM_MIC_TOK);
+	return err;
+}
+
+static struct gss_api_ops gss_spkm3_ops = {
+	.gss_import_sec_context	= gss_import_sec_context_spkm3,
+	.gss_get_mic		= gss_get_mic_spkm3,
+	.gss_verify_mic		= gss_verify_mic_spkm3,
+	.gss_delete_sec_context	= gss_delete_sec_context_spkm3,
+};
+
+static struct pf_desc gss_spkm3_pfs[] = {
+	{RPC_AUTH_GSS_SPKM, 0, RPC_GSS_SVC_NONE, "spkm3"},
+	{RPC_AUTH_GSS_SPKMI, 0, RPC_GSS_SVC_INTEGRITY, "spkm3i"},
+};
+
+static struct gss_api_mech gss_spkm3_mech = {
+	.gm_name	= "spkm3",
+	.gm_owner	= THIS_MODULE,
+	.gm_ops		= &gss_spkm3_ops,
+	.gm_pf_num	= ARRAY_SIZE(gss_spkm3_pfs),
+	.gm_pfs		= gss_spkm3_pfs,
+};
+
+static int __init init_spkm3_module(void)
+{
+	int status;
+
+	status = gss_mech_register(&gss_spkm3_mech);
+	if (status)
+		printk("Failed to register spkm3 gss mechanism!\n");
+	return 0;
+}
+
+static void __exit cleanup_spkm3_module(void)
+{
+	gss_mech_unregister(&gss_spkm3_mech);
+}
+
+MODULE_LICENSE("GPL");
+module_init(init_spkm3_module);
+module_exit(cleanup_spkm3_module);

net/sunrpc/auth_gss/gss_spkm3_seal.c

+/*
+ *  linux/net/sunrpc/gss_spkm3_seal.c
+ *
+ *  Copyright (c) 2003 The Regents of the University of Michigan.
+ *  All rights reserved.
+ *
+ *  Andy Adamson <andros@umich.edu>
+ *
+ *  Redistribution and use in source and binary forms, with or without
+ *  modification, are permitted provided that the following conditions
+ *  are met:
+ *
+ *  1. Redistributions of source code must retain the above copyright
+ *     notice, this list of conditions and the following disclaimer.
+ *  2. Redistributions in binary form must reproduce the above copyright
+ *     notice, this list of conditions and the following disclaimer in the
+ *     documentation and/or other materials provided with the distribution.
+ *  3. Neither the name of the University nor the names of its
+ *     contributors may be used to endorse or promote products derived
+ *     from this software without specific prior written permission.
+ *
+ *  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
+ *  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ *  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ *  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ *  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ *  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ *  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ *  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ *  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+ *  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ *  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/jiffies.h>
+#include <linux/sunrpc/gss_spkm3.h>
+#include <linux/random.h>
+#include <linux/crypto.h>
+
+#ifdef RPC_DEBUG
+# define RPCDBG_FACILITY        RPCDBG_AUTH
+#endif
+
+/*
+ * spkm3_make_token()
+ *
+ * Only SPKM_MIC_TOK with md5 intg-alg is supported
+ */
+
+u32
+spkm3_make_token(struct spkm3_ctx *ctx, int qop_req,
+		   struct xdr_buf * text, struct xdr_netobj * token,
+		   int toktype)
+{
+	s32			checksum_type;
+	char			tokhdrbuf[25];
+	struct xdr_netobj	md5cksum = {.len = 0, .data = NULL};
+	struct xdr_netobj	mic_hdr = {.len = 0, .data = tokhdrbuf};
+	int			tmsglen, tokenlen = 0;
+	unsigned char		*ptr;
+	s32			now;
+	int			ctxelen = 0, ctxzbit = 0;
+	int			md5elen = 0, md5zbit = 0;
+
+	dprintk("RPC: spkm3_make_token\n");
+
+	now = jiffies;
+	if (qop_req != 0)
+		goto out_err;
+
+	if (ctx->ctx_id.len != 16) {
+		dprintk("RPC: spkm3_make_token BAD ctx_id.len %d\n",
+			ctx->ctx_id.len);
+		goto out_err;
+	}
+		
+	switch (ctx->intg_alg) {
+		case NID_md5:
+			checksum_type = CKSUMTYPE_RSA_MD5;
+			break;
+		default:
+			dprintk("RPC: gss_spkm3_seal: ctx->signalg %d not"
+				" supported\n", ctx->intg_alg);
+			goto out_err;
+	}
+	/* XXX since we don't support WRAP, perhaps we don't care... */
+	if (ctx->conf_alg != NID_cast5_cbc) {
+		dprintk("RPC: gss_spkm3_seal: ctx->sealalg %d not supported\n",
+			ctx->conf_alg);
+		goto out_err;
+	}
+
+	if (toktype == SPKM_MIC_TOK) {
+		tmsglen = 0;
+		/* Calculate checksum over the mic-header */
+		asn1_bitstring_len(&ctx->ctx_id, &ctxelen, &ctxzbit);
+		spkm3_mic_header(&mic_hdr.data, &mic_hdr.len, ctx->ctx_id.data,
+		                         ctxelen, ctxzbit);
+
+		if (make_checksum(checksum_type, mic_hdr.data, mic_hdr.len, 
+		                             text, &md5cksum))
+			goto out_err;
+
+		asn1_bitstring_len(&md5cksum, &md5elen, &md5zbit);
+		tokenlen = 10 + ctxelen + 1 + 2 + md5elen + 1;
+
+		/* Create token header using generic routines */
+		token->len = g_token_size(&ctx->mech_used, tokenlen + tmsglen);
+
+		ptr = token->data;
+		g_make_token_header(&ctx->mech_used, tokenlen + tmsglen, &ptr);
+
+		spkm3_make_mic_token(&ptr, tokenlen, &mic_hdr, &md5cksum, md5elen, md5zbit);
+	} else if (toktype == SPKM_WRAP_TOK) { /* Not Supported */
+		dprintk("RPC: gss_spkm3_seal: SPKM_WRAP_TOK not supported\n");
+		goto out_err;
+	}
+	kfree(md5cksum.data);
+
+	/* XXX need to implement sequence numbers, and ctx->expired */
+
+	return  GSS_S_COMPLETE;
+out_err:
+	if (md5cksum.data) 
+		kfree(md5cksum.data);
+	token->data = 0;
+	token->len = 0;
+	return GSS_S_FAILURE;
+}

net/sunrpc/auth_gss/gss_spkm3_token.c

+/*
+ *  linux/net/sunrpc/gss_spkm3_token.c
+ *
+ *  Copyright (c) 2003 The Regents of the University of Michigan.
+ *  All rights reserved.
+ *
+ *  Andy Adamson <andros@umich.edu>
+ *
+ *  Redistribution and use in source and binary forms, with or without
+ *  modification, are permitted provided that the following conditions
+ *  are met:
+ *
+ *  1. Redistributions of source code must retain the above copyright
+ *     notice, this list of conditions and the following disclaimer.
+ *  2. Redistributions in binary form must reproduce the above copyright
+ *     notice, this list of conditions and the following disclaimer in the
+ *     documentation and/or other materials provided with the distribution.
+ *  3. Neither the name of the University nor the names of its
+ *     contributors may be used to endorse or promote products derived
+ *     from this software without specific prior written permission.
+ *
+ *  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
+ *  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ *  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ *  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ *  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ *  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ *  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ *  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ *  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+ *  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ *  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/jiffies.h>
+#include <linux/sunrpc/gss_spkm3.h>
+#include <linux/random.h>
+#include <linux/crypto.h>
+
+#ifdef RPC_DEBUG
+# define RPCDBG_FACILITY        RPCDBG_AUTH
+#endif
+
+/*
+ * asn1_bitstring_len()
+ *
+ * calculate the asn1 bitstring length of the xdr_netobject
+ */
+void
+asn1_bitstring_len(struct xdr_netobj *in, int *enclen, int *zerobits)
+{
+	int i, zbit = 0,elen = in->len;
+	char *ptr;
+
+	ptr = &in->data[in->len -1];
+
+	/* count trailing 0's */
+	for(i = in->len; i > 0; i--) {
+		if (*ptr == 0) { 
+			ptr--;
+			elen--;
+		} else
+			break;
+	}
+
+	/* count number of 0 bits in final octet */
+	ptr = &in->data[elen - 1];
+	for(i = 0; i < 8; i++) {
+		short mask = 0x01;
+
+		if (!((mask << i) & *ptr))
+			zbit++;
+		else
+			break;
+	}
+	*enclen = elen;
+	*zerobits = zbit;
+}
+
+/*
+ * decode_asn1_bitstring()
+ * 
+ * decode a bitstring into a buffer of the expected length.
+ * enclen = bit string length
+ * explen = expected length (define in rfc)
+ */
+int
+decode_asn1_bitstring(struct xdr_netobj *out, char *in, int enclen, int explen)
+{
+	if (!(out->data = kmalloc(explen,GFP_KERNEL)))
+		return 0;
+	out->len = explen;
+	memset(out->data, 0, explen);
+	memcpy(out->data, in, enclen);
+	return 1;
+}
+
+/* 
+ * SPKMInnerContextToken choice SPKM_MIC asn1 token layout
+ * 
+ * contextid is always 16 bytes plain data. max asn1 bitstring len = 17.
+ *
+ * tokenlen = pos[0] to end of token (max pos[45] with MD5 cksum)
+ *
+ * pos  value
+ * ----------
+ * [0]	a4  SPKM-MIC tag
+ * [1]	??  innertoken length  (max 44) 
+ * 
+ * 
+ * tok_hdr piece of checksum data starts here 
+ *
+ * the maximum mic-header len = 9 + 17 = 26 
+ *	mic-header
+ *	----------
+ * [2]	30      SEQUENCE tag  
+ * [3]	??	mic-header length: (max 23) = TokenID + ContextID 
+ *
+ *		TokenID  - all fields constant and can be hardcoded
+ *		-------
+ * [4]	  02	Type 2
+ * [5]	  02	Length 2 
+ * [6][7] 01 01	TokenID (SPKM_MIC_TOK)
+ *
+ *		ContextID  - encoded length not constant, calculated
+ *		---------
+ * [8]	03	Type 3
+ * [9]	??	encoded length
+ * [10]	??	ctxzbit
+ * [11]	 	contextid
+ *
+ * mic_header piece of checksum data ends here. 
+ *
+ *	int-cksum - encoded length not constant, calculated
+ *	---------
+ * [??]	03	Type 3
+ * [??]	??	encoded length 
+ * [??]	??	md5zbit		
+ * [??]	 	int-cksum (NID_md5 = 16)
+ *
+ * maximum SPKM-MIC innercontext token length = 
+ *	 10 + encoded contextid_size(17 max) + 2 + encoded  
+ *       cksum_size (17 maxfor NID_md5) = 46
+ */
+
+/*
+ * spkm3_mic_header()
+ *
+ * Prepare the SPKM_MIC_TOK mic-header for check-sum calculation
+ * elen: 16 byte context id asn1 bitstring encoded length
+ */
+void
+spkm3_mic_header(unsigned char **hdrbuf, unsigned int *hdrlen, unsigned char *ctxdata, int elen, int zbit)
+{
+	char *hptr = *hdrbuf;
+	char *top = *hdrbuf;
+
+	*(u8 *)hptr++ = 0x30;
+	*(u8 *)hptr++ = elen + 7;  /* on the wire header length */
+
+	/* tokenid */
+	*(u8 *)hptr++ = 0x02;
+	*(u8 *)hptr++ = 0x02;
+	*(u8 *)hptr++ = 0x01;
+	*(u8 *)hptr++ = 0x01;
+
+	/* coniextid */
+	*(u8 *)hptr++ = 0x03;
+	*(u8 *)hptr++ = elen + 1; /* add 1 to include zbit */
+	*(u8 *)hptr++ = zbit;
+	memcpy(hptr, ctxdata, elen);
+	hptr += elen;
+	*hdrlen = hptr - top; 
+}
+		
+/* 
+ * spkm3_mic_innercontext_token()
+ *
+ * *tokp points to the beginning of the SPKM_MIC token  described 
+ * in rfc 2025, section 3.2.1: 
+ *
+ */
+void
+spkm3_make_mic_token(unsigned char **tokp, int toklen, struct xdr_netobj *mic_hdr, struct xdr_netobj *md5cksum, int md5elen, int md5zbit)
+{
+	unsigned char *ict = *tokp;
+
+	*(u8 *)ict++ = 0xa4;
+	*(u8 *)ict++ = toklen - 2; 
+	memcpy(ict, mic_hdr->data, mic_hdr->len);
+	ict += mic_hdr->len;
+
+	*(u8 *)ict++ = 0x03;
+	*(u8 *)ict++ = md5elen + 1; /* add 1 to include zbit */
+	*(u8 *)ict++ = md5zbit;
+	memcpy(ict, md5cksum->data, md5elen);
+}
+
+u32
+spkm3_verify_mic_token(unsigned char **tokp, int *mic_hdrlen, unsigned char **cksum)
+{
+	struct xdr_netobj       spkm3_ctx_id = {.len =0, .data = NULL};
+	unsigned char 		*ptr = *tokp;
+	int 			ctxelen;
+	u32     		ret = GSS_S_DEFECTIVE_TOKEN;
+
+	/* spkm3 innercontext token preamble */
+	if ((ptr[0] != 0xa4) || (ptr[2] != 0x30)) {
+		dprintk("RPC: BAD SPKM ictoken preamble\n"); 
+		goto out;
+	}
+
+	*mic_hdrlen = ptr[3];
+
+	/* token type */
+	if ((ptr[4] != 0x02) || (ptr[5] != 0x02)) {
+		dprintk("RPC: BAD asn1 SPKM3 token type\n");
+		goto out;
+	}
+
+	/* only support SPKM_MIC_TOK */
+	if((ptr[6] != 0x01) || (ptr[7] != 0x01)) {
+		dprintk("RPC: ERROR unsupported SPKM3 token \n");
+		goto out;
+	}
+
+	/* contextid */
+	if (ptr[8] != 0x03) {
+		dprintk("RPC: BAD SPKM3 asn1 context-id type\n");
+		goto out;
+	}
+
+	ctxelen = ptr[9];
+	if (ctxelen > 17) {  /* length includes asn1 zbit octet */
+		dprintk("RPC: BAD SPKM3 contextid len %d\n", ctxelen);
+		goto out;
+	}
+
+	/* ignore ptr[10] */
+
+	if(!decode_asn1_bitstring(&spkm3_ctx_id, &ptr[11], ctxelen - 1, 16))
+		goto out;
+
+	/*
+	* in the current implementation: the optional int-alg is not present 
+	* so the default int-alg (md5) is used the optional snd-seq field is 
+	* also not present 
+	*/
+
+	if (*mic_hdrlen != 6 + ctxelen) {
+		dprintk("RPC: BAD SPKM_ MIC_TOK header len %d: we only support default int-alg (should be absent) and do not support snd-seq\n", *mic_hdrlen);
+		goto out;
+	}
+	/* checksum */
+        *cksum = (&ptr[10] + ctxelen); /* ctxelen includes ptr[10] */
+
+	ret = GSS_S_COMPLETE;
+out:
+	if (spkm3_ctx_id.data)
+		kfree(spkm3_ctx_id.data);
+	return ret;
+}
+

net/sunrpc/auth_gss/gss_spkm3_unseal.c

+/*
+ *  linux/net/sunrpc/gss_spkm3_unseal.c
+ *
+ *  Copyright (c) 2003 The Regents of the University of Michigan.
+ *  All rights reserved.
+ *
+ *  Andy Adamson <andros@umich.edu>
+ *
+ *  Redistribution and use in source and binary forms, with or without
+ *  modification, are permitted provided that the following conditions
+ *  are met:
+ *
+ *  1. Redistributions of source code must retain the above copyright
+ *     notice, this list of conditions and the following disclaimer.
+ *  2. Redistributions in binary form must reproduce the above copyright
+ *     notice, this list of conditions and the following disclaimer in the
+ *     documentation and/or other materials provided with the distribution.
+ *  3. Neither the name of the University nor the names of its
+ *     contributors may be used to endorse or promote products derived
+ *     from this software without specific prior written permission.
+ *
+ *  THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESS OR IMPLIED
+ *  WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
+ *  MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+ *  DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ *  FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
+ *  CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
+ *  SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR
+ *  BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
+ *  LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
+ *  NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
+ *  SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ *
+ */
+
+#include <linux/types.h>
+#include <linux/slab.h>
+#include <linux/jiffies.h>
+#include <linux/sunrpc/gss_spkm3.h>
+#include <linux/crypto.h>
+
+#ifdef RPC_DEBUG
+# define RPCDBG_FACILITY        RPCDBG_AUTH
+#endif
+
+/*
+ * spkm3_read_token()
+ * 
+ * only SPKM_MIC_TOK with md5 intg-alg is supported
+ */
+u32
+spkm3_read_token(struct spkm3_ctx *ctx,
+		struct xdr_netobj *read_token,    /* checksum */
+		struct xdr_buf *message_buffer, /* signbuf */
+		int *qop_state, int toktype)
+{
+	s32			code;
+	struct xdr_netobj	wire_cksum = {.len =0, .data = NULL};
+	struct xdr_netobj	md5cksum = {.len = 0, .data = NULL};
+	unsigned char		*ptr = (unsigned char *)read_token->data;
+	unsigned char           *cksum;
+	int			bodysize, md5elen;
+	int			mic_hdrlen;
+	u32			ret = GSS_S_DEFECTIVE_TOKEN;
+
+	dprintk("RPC: spkm3_read_token read_token->len %d\n", read_token->len);
+
+	if (g_verify_token_header((struct xdr_netobj *) &ctx->mech_used,
+					&bodysize, &ptr, read_token->len))
+		goto out;
+
+	/* decode the token */
+
+	if (toktype == SPKM_MIC_TOK) {
+
+		if ((ret = spkm3_verify_mic_token(&ptr, &mic_hdrlen, &cksum))) 
+			goto out;
+
+		if (*cksum++ != 0x03) {
+			dprintk("RPC: spkm3_read_token BAD checksum type\n");
+			goto out;
+		}
+		md5elen = *cksum++; 
+		cksum++; 	/* move past the zbit */
+	
+		if(!decode_asn1_bitstring(&wire_cksum, cksum, md5elen - 1, 16))
+			goto out;
+
+		/* HARD CODED FOR MD5 */
+
+		/* compute the checksum of the message.
+		*  ptr + 2 = start of header piece of checksum
+		*  mic_hdrlen + 2 = length of header piece of checksum
+		*/
+		ret = GSS_S_DEFECTIVE_TOKEN;
+		code = make_checksum(CKSUMTYPE_RSA_MD5, ptr + 2, 
+					mic_hdrlen + 2, 
+		                        message_buffer, &md5cksum);
+
+		if (code)
+			goto out;
+
+		dprintk("RPC: spkm3_read_token: digest wire_cksum.len %d:\n", 
+			wire_cksum.len);
+		dprintk("          md5cksum.data\n");
+		print_hexl((u32 *) md5cksum.data, 16, 0);
+		dprintk("          cksum.data:\n");
+		print_hexl((u32 *) wire_cksum.data, wire_cksum.len, 0);
+
+		ret = GSS_S_BAD_SIG;
+		code = memcmp(md5cksum.data, wire_cksum.data, wire_cksum.len);
+		if (code)
+			goto out;
+
+	} else { 
+		dprintk("RPC: BAD or UNSUPPORTED SPKM3 token type: %d\n",toktype);
+		goto out;
+	}
+
+	/* XXX: need to add expiration and sequencing */
+	ret = GSS_S_COMPLETE;
+out:
+	if (md5cksum.data) 
+		kfree(md5cksum.data);
+	if (wire_cksum.data) 
+		kfree(wire_cksum.data);
+	return ret;
+}

net/sunrpc/clnt.c

@@ -196,7 +196,15 @@ rpc_clone_client(struct rpc_clnt *clnt)
 	memcpy(new, clnt, sizeof(*new));
 	atomic_set(&new->cl_count, 1);
 	atomic_set(&new->cl_users, 0);
-	atomic_inc(&new->cl_parent->cl_count);
+	new->cl_parent = clnt;
+	atomic_inc(&clnt->cl_count);
+	/* Duplicate portmapper */
+	rpc_init_wait_queue(&new->cl_pmap_default.pm_bindwait, "bindwait");
+	/* Turn off autobind on clones */
+	new->cl_autobind = 0;
+	new->cl_oneshot = 0;
+	new->cl_dead = 0;
+	rpc_init_rtt(&new->cl_rtt_default, clnt->cl_xprt->timeout.to_initval);
 	if (new->cl_auth)
 		atomic_inc(&new->cl_auth->au_count);
 	return new;
@@ -335,7 +343,7 @@ void rpc_clnt_sigunmask(struct rpc_clnt *clnt, sigset_t *oldset)
  */
 int rpc_call_sync(struct rpc_clnt *clnt, struct rpc_message *msg, int flags)
 {
-	struct rpc_task	my_task, *task = &my_task;
+	struct rpc_task	*task;
 	sigset_t	oldset;
 	int		status;
 
@@ -343,15 +351,15 @@ int rpc_call_sync(struct rpc_clnt *clnt, struct rpc_message *msg, int flags)
 	if (clnt->cl_dead) 
 		return -EIO;
 
-	if (flags & RPC_TASK_ASYNC) {
-		printk("rpc_call_sync: Illegal flag combination for synchronous task\n");
-		flags &= ~RPC_TASK_ASYNC;
-	}
+	BUG_ON(flags & RPC_TASK_ASYNC);
 
 	rpc_clnt_sigmask(clnt, &oldset);		
 
-	/* Create/initialize a new RPC task */
-	rpc_init_task(task, clnt, NULL, flags);
+	status = -ENOMEM;
+	task = rpc_new_task(clnt, NULL, flags);
+	if (task == NULL)
+		goto out;
+
 	rpc_call_setup(task, msg, 0);
 
 	/* Set up the call info struct and execute the task */
@@ -362,6 +370,7 @@ int rpc_call_sync(struct rpc_clnt *clnt, struct rpc_message *msg, int flags)
 		rpc_release_task(task);
 	}
 
+out:
 	rpc_clnt_sigunmask(clnt, &oldset);		
 
 	return status;
@@ -958,8 +967,12 @@ call_header(struct rpc_task *task)
 static u32 *
 call_verify(struct rpc_task *task)
 {
-	u32	*p = task->tk_rqstp->rq_rcv_buf.head[0].iov_base, n;
+	struct kvec *iov = &task->tk_rqstp->rq_rcv_buf.head[0];
+	int len = task->tk_rqstp->rq_rcv_buf.len >> 2;
+	u32	*p = iov->iov_base, n;
 
+	if ((len -= 3) < 0)
+		goto garbage;
 	p += 1;	/* skip XID */
 
 	if ((n = ntohl(*p++)) != RPC_REPLY) {
@@ -969,9 +982,11 @@ call_verify(struct rpc_task *task)
 	if ((n = ntohl(*p++)) != RPC_MSG_ACCEPTED) {
 		int	error = -EACCES;
 
+		if (--len < 0)
+			goto garbage;
 		if ((n = ntohl(*p++)) != RPC_AUTH_ERROR) {
 			printk(KERN_WARNING "call_verify: RPC call rejected: %x\n", n);
-		} else
+		} else if (--len < 0)
 		switch ((n = ntohl(*p++))) {
 		case RPC_AUTH_REJECTEDCRED:
 		case RPC_AUTH_REJECTEDVERF:
@@ -1002,7 +1017,8 @@ call_verify(struct rpc_task *task)
 		default:
 			printk(KERN_WARNING "call_verify: unknown auth error: %x\n", n);
 			error = -EIO;
-		}
+		} else
+			goto garbage;
 		dprintk("RPC: %4d call_verify: call rejected %d\n",
 						task->tk_pid, n);
 		rpc_exit(task, error);
@@ -1012,6 +1028,9 @@ call_verify(struct rpc_task *task)
 		printk(KERN_WARNING "call_verify: auth check failed\n");
 		goto garbage;		/* bad verifier, retry */
 	}
+	len = p - (u32 *)iov->iov_base - 1;
+	if (len < 0)
+		goto garbage;
 	switch ((n = ntohl(*p++))) {
 	case RPC_SUCCESS:
 		return p;

net/sunrpc/sched.c

@@ -698,7 +698,7 @@ __rpc_execute(struct rpc_task *task)
 		if (current->pid == rpciod_pid)
 			printk(KERN_ERR "RPC: rpciod waiting on sync task!\n");
 
-		if (!task->tk_client->cl_intr) {
+		if (RPC_TASK_UNINTERRUPTIBLE(task)) {
 			__wait_event(task->tk_wait, !RPC_IS_SLEEPING(task));
 		} else {
 			__wait_event_interruptible(task->tk_wait, !RPC_IS_SLEEPING(task), status);
@@ -884,6 +884,8 @@ void rpc_init_task(struct rpc_task *task, struct rpc_clnt *clnt, rpc_action call
 		atomic_inc(&clnt->cl_users);
 		if (clnt->cl_softrtry)
 			task->tk_flags |= RPC_TASK_SOFT;
+		if (!clnt->cl_intr)
+			task->tk_flags |= RPC_TASK_NOINTR;
 	}
 
 #ifdef RPC_DEBUG

net/sunrpc/sunrpc_syms.c

@@ -89,6 +89,7 @@ EXPORT_SYMBOL(svc_makesock);
 EXPORT_SYMBOL(svc_reserve);
 EXPORT_SYMBOL(svc_auth_register);
 EXPORT_SYMBOL(auth_domain_lookup);
+EXPORT_SYMBOL(svc_authenticate);
 
 /* RPC statistics */
 #ifdef CONFIG_PROC_FS

net/sunrpc/svc.c

@@ -263,6 +263,7 @@ svc_process(struct svc_serv *serv, struct svc_rqst *rqstp)
 	u32			*statp;
 	u32			dir, prog, vers, proc,
 				auth_stat, rpc_stat;
+	int			auth_res;
 
 	rpc_stat = rpc_success;
 
@@ -304,12 +305,17 @@ svc_process(struct svc_serv *serv, struct svc_rqst *rqstp)
 	rqstp->rq_vers = vers = ntohl(svc_getu32(argv));	/* version number */
 	rqstp->rq_proc = proc = ntohl(svc_getu32(argv));	/* procedure number */
 
+	progp = serv->sv_program;
 	/*
 	 * Decode auth data, and add verifier to reply buffer.
 	 * We do this before anything else in order to get a decent
 	 * auth verifier.
 	 */
-	switch (svc_authenticate(rqstp, &auth_stat)) {
+	if (progp->pg_authenticate != NULL)
+		auth_res = progp->pg_authenticate(rqstp, &auth_stat);
+	else
+		auth_res = svc_authenticate(rqstp, &auth_stat);
+	switch (auth_res) {
 	case SVC_OK:
 		break;
 	case SVC_GARBAGE:
@@ -326,7 +332,6 @@ svc_process(struct svc_serv *serv, struct svc_rqst *rqstp)
 		goto sendit;
 	}
 		
-	progp = serv->sv_program;
 	if (prog != progp->pg_prog)
 		goto err_bad_prog;