Merge branch 'for-linus' of git://linux-nfs.org/~bfields/linux

* 'for-linus' of git://linux-nfs.org/~bfields/linux: (100 commits)
  SUNRPC: RPC program information is stored in unsigned integers
  SUNRPC: Move exported symbol definitions after function declaration part 2
  NLM: tear down RPC clients in nlm_shutdown_hosts
  SUNRPC: spin svc_rqst initialization to its own function
  nfsd: more careful input validation in nfsctl write methods
  lockd: minor log message fix
  knfsd: don't bother mapping putrootfh enoent to eperm
  rdma: makefile
  rdma: ONCRPC RDMA protocol marshalling
  rdma: SVCRDMA sendto
  rdma: SVCRDMA recvfrom
  rdma: SVCRDMA Core Transport Services
  rdma: SVCRDMA Transport Module
  rdma: SVCRMDA Header File
  svc: Add svc_xprt_names service to replace svc_sock_names
  knfsd: Support adding transports by writing portlist file
  svc: Add svc API that queries for a transport instance
  svc: Add /proc/sys/sunrpc/transport files
  svc: Add transport hdr size for defer/revisit
  svc: Move the xprt independent code to the svc_xprt.c file
  ...

MAINTAINERS

@@ -2247,7 +2247,7 @@ P:	J. Bruce Fields
 M:	bfields@fieldses.org
 P:	Neil Brown
 M:	neilb@suse.de
-L:	nfs@lists.sourceforge.net
+L:	linux-nfs@vger.kernel.org
 W:	http://nfs.sourceforge.net/
 S:	Supported
 

fs/Kconfig

@@ -1674,6 +1674,8 @@ config NFSD
 	select CRYPTO_MD5 if NFSD_V4
 	select CRYPTO if NFSD_V4
 	select FS_POSIX_ACL if NFSD_V4
+	select PROC_FS if NFSD_V4
+	select PROC_FS if SUNRPC_GSS
 	help
 	  If you want your Linux box to act as an NFS *server*, so that other
 	  computers on your local network which support NFS can access certain

fs/lockd/host.c

@@ -34,10 +34,10 @@ static DEFINE_MUTEX(nlm_host_mutex);
 
 static void			nlm_gc_hosts(void);
 static struct nsm_handle *	__nsm_find(const struct sockaddr_in *,
-					const char *, int, int);
+					const char *, unsigned int, int);
 static struct nsm_handle *	nsm_find(const struct sockaddr_in *sin,
 					 const char *hostname,
-					 int hostname_len);
+					 unsigned int hostname_len);
 
 /*
  * Common host lookup routine for server & client
@@ -45,7 +45,8 @@ static struct nsm_handle *	nsm_find(const struct sockaddr_in *sin,
 static struct nlm_host *
 nlm_lookup_host(int server, const struct sockaddr_in *sin,
 		int proto, int version, const char *hostname,
-		int hostname_len, const struct sockaddr_in *ssin)
+		unsigned int hostname_len,
+		const struct sockaddr_in *ssin)
 {
 	struct hlist_head *chain;
 	struct hlist_node *pos;
@@ -176,7 +177,7 @@ nlm_destroy_host(struct nlm_host *host)
  */
 struct nlm_host *
 nlmclnt_lookup_host(const struct sockaddr_in *sin, int proto, int version,
-			const char *hostname, int hostname_len)
+			const char *hostname, unsigned int hostname_len)
 {
 	struct sockaddr_in ssin = {0};
 
@@ -189,7 +190,7 @@ nlmclnt_lookup_host(const struct sockaddr_in *sin, int proto, int version,
  */
 struct nlm_host *
 nlmsvc_lookup_host(struct svc_rqst *rqstp,
-			const char *hostname, int hostname_len)
+			const char *hostname, unsigned int hostname_len)
 {
 	struct sockaddr_in ssin = {0};
 
@@ -307,7 +308,8 @@ void nlm_release_host(struct nlm_host *host)
  * Release all resources held by that peer.
  */
 void nlm_host_rebooted(const struct sockaddr_in *sin,
-				const char *hostname, int hostname_len,
+				const char *hostname,
+				unsigned int hostname_len,
 				u32 new_state)
 {
 	struct hlist_head *chain;
@@ -377,8 +379,13 @@ nlm_shutdown_hosts(void)
 	/* First, make all hosts eligible for gc */
 	dprintk("lockd: nuking all hosts...\n");
 	for (chain = nlm_hosts; chain < nlm_hosts + NLM_HOST_NRHASH; ++chain) {
-		hlist_for_each_entry(host, pos, chain, h_hash)
+		hlist_for_each_entry(host, pos, chain, h_hash) {
 			host->h_expires = jiffies - 1;
+			if (host->h_rpcclnt) {
+				rpc_shutdown_client(host->h_rpcclnt);
+				host->h_rpcclnt = NULL;
+			}
+		}
 	}
 
 	/* Then, perform a garbage collection pass */
@@ -449,7 +456,7 @@ static DEFINE_MUTEX(nsm_mutex);
 
 static struct nsm_handle *
 __nsm_find(const struct sockaddr_in *sin,
-		const char *hostname, int hostname_len,
+		const char *hostname, unsigned int hostname_len,
 		int create)
 {
 	struct nsm_handle *nsm = NULL;
@@ -503,7 +510,8 @@ out:
 }
 
 static struct nsm_handle *
-nsm_find(const struct sockaddr_in *sin, const char *hostname, int hostname_len)
+nsm_find(const struct sockaddr_in *sin, const char *hostname,
+	 unsigned int hostname_len)
 {
 	return __nsm_find(sin, hostname, hostname_len, 1);
 }

fs/lockd/svc.c

@@ -219,19 +219,6 @@ lockd(struct svc_rqst *rqstp)
 	module_put_and_exit(0);
 }
 
-
-static int find_socket(struct svc_serv *serv, int proto)
-{
-	struct svc_sock *svsk;
-	int found = 0;
-	list_for_each_entry(svsk, &serv->sv_permsocks, sk_list)
-		if (svsk->sk_sk->sk_protocol == proto) {
-			found = 1;
-			break;
-		}
-	return found;
-}
-
 /*
  * Make any sockets that are needed but not present.
  * If nlm_udpport or nlm_tcpport were set as module
@@ -240,17 +227,25 @@ static int find_socket(struct svc_serv *serv, int proto)
 static int make_socks(struct svc_serv *serv, int proto)
 {
 	static int warned;
+	struct svc_xprt *xprt;
 	int err = 0;
 
-	if (proto == IPPROTO_UDP || nlm_udpport)
+	if (proto == IPPROTO_UDP || nlm_udpport) {
-		if (!find_socket(serv, IPPROTO_UDP))
+		xprt = svc_find_xprt(serv, "udp", 0, 0);
-			err = svc_makesock(serv, IPPROTO_UDP, nlm_udpport,
+		if (!xprt)
+			err = svc_create_xprt(serv, "udp", nlm_udpport,
 					      SVC_SOCK_DEFAULTS);
-	if (err >= 0 && (proto == IPPROTO_TCP || nlm_tcpport))
+		else
-		if (!find_socket(serv, IPPROTO_TCP))
+			svc_xprt_put(xprt);
-			err = svc_makesock(serv, IPPROTO_TCP, nlm_tcpport,
+	}
+	if (err >= 0 && (proto == IPPROTO_TCP || nlm_tcpport)) {
+		xprt = svc_find_xprt(serv, "tcp", 0, 0);
+		if (!xprt)
+			err = svc_create_xprt(serv, "tcp", nlm_tcpport,
 					      SVC_SOCK_DEFAULTS);
-
+		else
+			svc_xprt_put(xprt);
+	}
 	if (err >= 0) {
 		warned = 0;
 		err = 0;

fs/lockd/svc4proc.c

@@ -84,6 +84,7 @@ nlm4svc_proc_test(struct svc_rqst *rqstp, struct nlm_args *argp,
 {
 	struct nlm_host	*host;
 	struct nlm_file	*file;
+	int rc = rpc_success;
 
 	dprintk("lockd: TEST4        called\n");
 	resp->cookie = argp->cookie;
@@ -91,7 +92,7 @@ nlm4svc_proc_test(struct svc_rqst *rqstp, struct nlm_args *argp,
 	/* Don't accept test requests during grace period */
 	if (nlmsvc_grace_period) {
 		resp->status = nlm_lck_denied_grace_period;
-		return rpc_success;
+		return rc;
 	}
 
 	/* Obtain client and file */
@@ -101,12 +102,13 @@ nlm4svc_proc_test(struct svc_rqst *rqstp, struct nlm_args *argp,
 	/* Now check for conflicting locks */
 	resp->status = nlmsvc_testlock(rqstp, file, &argp->lock, &resp->lock, &resp->cookie);
 	if (resp->status == nlm_drop_reply)
-		return rpc_drop_reply;
+		rc = rpc_drop_reply;
-
+	else
 		dprintk("lockd: TEST4        status %d\n", ntohl(resp->status));
+
 	nlm_release_host(host);
 	nlm_release_file(file);
-	return rpc_success;
+	return rc;
 }
 
 static __be32
@@ -115,6 +117,7 @@ nlm4svc_proc_lock(struct svc_rqst *rqstp, struct nlm_args *argp,
 {
 	struct nlm_host	*host;
 	struct nlm_file	*file;
+	int rc = rpc_success;
 
 	dprintk("lockd: LOCK          called\n");
 
@@ -123,7 +126,7 @@ nlm4svc_proc_lock(struct svc_rqst *rqstp, struct nlm_args *argp,
 	/* Don't accept new lock requests during grace period */
 	if (nlmsvc_grace_period && !argp->reclaim) {
 		resp->status = nlm_lck_denied_grace_period;
-		return rpc_success;
+		return rc;
 	}
 
 	/* Obtain client and file */
@@ -146,12 +149,13 @@ nlm4svc_proc_lock(struct svc_rqst *rqstp, struct nlm_args *argp,
 	resp->status = nlmsvc_lock(rqstp, file, &argp->lock,
 					argp->block, &argp->cookie);
 	if (resp->status == nlm_drop_reply)
-		return rpc_drop_reply;
+		rc = rpc_drop_reply;
-
+	else
 		dprintk("lockd: LOCK         status %d\n", ntohl(resp->status));
+
 	nlm_release_host(host);
 	nlm_release_file(file);
-	return rpc_success;
+	return rc;
 }
 
 static __be32

fs/lockd/svclock.c

@@ -501,25 +501,29 @@ nlmsvc_testlock(struct svc_rqst *rqstp, struct nlm_file *file,
 			block, block->b_flags, block->b_fl);
 		if (block->b_flags & B_TIMED_OUT) {
 			nlmsvc_unlink_block(block);
-			return nlm_lck_denied;
+			ret = nlm_lck_denied;
+			goto out;
 		}
 		if (block->b_flags & B_GOT_CALLBACK) {
+			nlmsvc_unlink_block(block);
 			if (block->b_fl != NULL
 					&& block->b_fl->fl_type != F_UNLCK) {
 				lock->fl = *block->b_fl;
 				goto conf_lock;
-			}
+			} else {
-			else {
+				ret = nlm_granted;
-				nlmsvc_unlink_block(block);
+				goto out;
-				return nlm_granted;
 			}
 		}
-		return nlm_drop_reply;
+		ret = nlm_drop_reply;
+		goto out;
 	}
 
 	error = vfs_test_lock(file->f_file, &lock->fl);
-	if (error == -EINPROGRESS)
+	if (error == -EINPROGRESS) {
-		return nlmsvc_defer_lock_rqst(rqstp, block);
+		ret = nlmsvc_defer_lock_rqst(rqstp, block);
+		goto out;
+	}
 	if (error) {
 		ret = nlm_lck_denied_nolocks;
 		goto out;

fs/lockd/svcproc.c

@@ -113,6 +113,7 @@ nlmsvc_proc_test(struct svc_rqst *rqstp, struct nlm_args *argp,
 {
 	struct nlm_host	*host;
 	struct nlm_file	*file;
+	int rc = rpc_success;
 
 	dprintk("lockd: TEST          called\n");
 	resp->cookie = argp->cookie;
@@ -120,7 +121,7 @@ nlmsvc_proc_test(struct svc_rqst *rqstp, struct nlm_args *argp,
 	/* Don't accept test requests during grace period */
 	if (nlmsvc_grace_period) {
 		resp->status = nlm_lck_denied_grace_period;
-		return rpc_success;
+		return rc;
 	}
 
 	/* Obtain client and file */
@@ -130,13 +131,14 @@ nlmsvc_proc_test(struct svc_rqst *rqstp, struct nlm_args *argp,
 	/* Now check for conflicting locks */
 	resp->status = cast_status(nlmsvc_testlock(rqstp, file, &argp->lock, &resp->lock, &resp->cookie));
 	if (resp->status == nlm_drop_reply)
-		return rpc_drop_reply;
+		rc = rpc_drop_reply;
-
+	else
 		dprintk("lockd: TEST          status %d vers %d\n",
 			ntohl(resp->status), rqstp->rq_vers);
+
 	nlm_release_host(host);
 	nlm_release_file(file);
-	return rpc_success;
+	return rc;
 }
 
 static __be32
@@ -145,6 +147,7 @@ nlmsvc_proc_lock(struct svc_rqst *rqstp, struct nlm_args *argp,
 {
 	struct nlm_host	*host;
 	struct nlm_file	*file;
+	int rc = rpc_success;
 
 	dprintk("lockd: LOCK          called\n");
 
@@ -153,7 +156,7 @@ nlmsvc_proc_lock(struct svc_rqst *rqstp, struct nlm_args *argp,
 	/* Don't accept new lock requests during grace period */
 	if (nlmsvc_grace_period && !argp->reclaim) {
 		resp->status = nlm_lck_denied_grace_period;
-		return rpc_success;
+		return rc;
 	}
 
 	/* Obtain client and file */
@@ -176,12 +179,13 @@ nlmsvc_proc_lock(struct svc_rqst *rqstp, struct nlm_args *argp,
 	resp->status = cast_status(nlmsvc_lock(rqstp, file, &argp->lock,
 					       argp->block, &argp->cookie));
 	if (resp->status == nlm_drop_reply)
-		return rpc_drop_reply;
+		rc = rpc_drop_reply;
-
+	else
 		dprintk("lockd: LOCK         status %d\n", ntohl(resp->status));
+
 	nlm_release_host(host);
 	nlm_release_file(file);
-	return rpc_success;
+	return rc;
 }
 
 static __be32

fs/lockd/svcsubs.c

@@ -87,7 +87,7 @@ nlm_lookup_file(struct svc_rqst *rqstp, struct nlm_file **result,
 	unsigned int	hash;
 	__be32		nfserr;
 
-	nlm_debug_print_fh("nlm_file_lookup", f);
+	nlm_debug_print_fh("nlm_lookup_file", f);
 
 	hash = file_hash(f);
 

fs/nfs/callback.c

@@ -119,7 +119,7 @@ int nfs_callback_up(void)
 	if (!serv)
 		goto out_err;
 
-	ret = svc_makesock(serv, IPPROTO_TCP, nfs_callback_set_tcpport,
+	ret = svc_create_xprt(serv, "tcp", nfs_callback_set_tcpport,
 			      SVC_SOCK_ANONYMOUS);
 	if (ret <= 0)
 		goto out_destroy;

fs/nfsd/auth.h

@@ -1,6 +1,4 @@
 /*
- * include/linux/nfsd/auth.h
- *
  * nfsd-specific authentication stuff.
  * uid/gid mapping not yet implemented.
  *
@@ -10,8 +8,6 @@
 #ifndef LINUX_NFSD_AUTH_H
 #define LINUX_NFSD_AUTH_H
 
-#ifdef __KERNEL__
-
 #define nfsd_luid(rq, uid)	((u32)(uid))
 #define nfsd_lgid(rq, gid)	((u32)(gid))
 #define nfsd_ruid(rq, uid)	((u32)(uid))
@@ -23,5 +19,4 @@
  */
 int nfsd_setuser(struct svc_rqst *, struct svc_export *);
 
-#endif /* __KERNEL__ */
 #endif /* LINUX_NFSD_AUTH_H */

fs/nfsd/export.c

@@ -1357,8 +1357,6 @@ exp_pseudoroot(struct svc_rqst *rqstp, struct svc_fh *fhp)
 	mk_fsid(FSID_NUM, fsidv, 0, 0, 0, NULL);
 
 	exp = rqst_exp_find(rqstp, FSID_NUM, fsidv);
-	if (PTR_ERR(exp) == -ENOENT)
-		return nfserr_perm;
 	if (IS_ERR(exp))
 		return nfserrno(PTR_ERR(exp));
 	rv = fh_compose(fhp, exp, exp->ex_dentry, NULL);
@@ -1637,13 +1635,19 @@ exp_verify_string(char *cp, int max)
 /*
  * Initialize the exports module.
  */
-void
+int
 nfsd_export_init(void)
 {
+	int rv;
 	dprintk("nfsd: initializing export module.\n");
 
-	cache_register(&svc_export_cache);
+	rv = cache_register(&svc_export_cache);
-	cache_register(&svc_expkey_cache);
+	if (rv)
+		return rv;
+	rv = cache_register(&svc_expkey_cache);
+	if (rv)
+		cache_unregister(&svc_export_cache);
+	return rv;
 
 }
 
@@ -1670,10 +1674,8 @@ nfsd_export_shutdown(void)
 
 	exp_writelock();
 
-	if (cache_unregister(&svc_expkey_cache))
+	cache_unregister(&svc_expkey_cache);
-		printk(KERN_ERR "nfsd: failed to unregister expkey cache\n");
+	cache_unregister(&svc_export_cache);
-	if (cache_unregister(&svc_export_cache))
-		printk(KERN_ERR "nfsd: failed to unregister export cache\n");
 	svcauth_unix_purge();
 
 	exp_writeunlock();

fs/nfsd/nfs2acl.c

@@ -221,12 +221,17 @@ static int nfsaclsvc_encode_getaclres(struct svc_rqst *rqstp, __be32 *p,
 		struct nfsd3_getaclres *resp)
 {
 	struct dentry *dentry = resp->fh.fh_dentry;
-	struct inode *inode = dentry->d_inode;
+	struct inode *inode;
 	struct kvec *head = rqstp->rq_res.head;
 	unsigned int base;
 	int n;
 	int w;
 
+	/*
+	 * Since this is version 2, the check for nfserr in
+	 * nfsd_dispatch actually ensures the following cannot happen.
+	 * However, it seems fragile to depend on that.
+	 */
 	if (dentry == NULL || dentry->d_inode == NULL)
 		return 0;
 	inode = dentry->d_inode;

fs/nfsd/nfs3xdr.c

@@ -21,6 +21,7 @@
 #include <linux/sunrpc/svc.h>
 #include <linux/nfsd/nfsd.h>
 #include <linux/nfsd/xdr3.h>
+#include "auth.h"
 
 #define NFSDDBG_FACILITY		NFSDDBG_XDR
 
@@ -88,10 +89,10 @@ encode_fh(__be32 *p, struct svc_fh *fhp)
  * no slashes or null bytes.
  */
 static __be32 *
-decode_filename(__be32 *p, char **namp, int *lenp)
+decode_filename(__be32 *p, char **namp, unsigned int *lenp)
 {
 	char		*name;
-	int		i;
+	unsigned int	i;
 
 	if ((p = xdr_decode_string_inplace(p, namp, lenp, NFS3_MAXNAMLEN)) != NULL) {
 		for (i = 0, name = *namp; i < *lenp; i++, name++) {
@@ -452,8 +453,7 @@ int
 nfs3svc_decode_symlinkargs(struct svc_rqst *rqstp, __be32 *p,
 					struct nfsd3_symlinkargs *args)
 {
-	unsigned int len;
+	unsigned int len, avail;
-	int avail;
 	char *old, *new;
 	struct kvec *vec;
 
@@ -486,7 +486,8 @@ nfs3svc_decode_symlinkargs(struct svc_rqst *rqstp, __be32 *p,
 	/* now copy next page if there is one */
 	if (len && !avail && rqstp->rq_arg.page_len) {
 		avail = rqstp->rq_arg.page_len;
-		if (avail > PAGE_SIZE) avail = PAGE_SIZE;
+		if (avail > PAGE_SIZE)
+			avail = PAGE_SIZE;
 		old = page_address(rqstp->rq_arg.pages[0]);
 	}
 	while (len && avail && *old) {

fs/nfsd/nfs4callback.c

@@ -350,30 +350,6 @@ static struct rpc_version *	nfs_cb_version[] = {
 static int do_probe_callback(void *data)
 {
 	struct nfs4_client *clp = data;
-	struct nfs4_callback *cb = &clp->cl_callback;
-	struct rpc_message msg = {
-		.rpc_proc       = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_NULL],
-		.rpc_argp       = clp,
-	};
-	int status;
-
-	status = rpc_call_sync(cb->cb_client, &msg, RPC_TASK_SOFT);
-
-	if (status) {
-		rpc_shutdown_client(cb->cb_client);
-		cb->cb_client = NULL;
-	} else
-		atomic_set(&cb->cb_set, 1);
-	put_nfs4_client(clp);
-	return 0;
-}
-
-/*
- * Set up the callback client and put a NFSPROC4_CB_NULL on the wire...
- */
-void
-nfsd4_probe_callback(struct nfs4_client *clp)
-{
 	struct sockaddr_in	addr;
 	struct nfs4_callback    *cb = &clp->cl_callback;
 	struct rpc_timeout	timeparms = {
@@ -393,10 +369,12 @@ nfsd4_probe_callback(struct nfs4_client *clp)
 		.authflavor	= RPC_AUTH_UNIX, /* XXX: need AUTH_GSS... */
 		.flags		= (RPC_CLNT_CREATE_NOPING),
 	};
-	struct task_struct *t;
+	struct rpc_message msg = {
-
+		.rpc_proc       = &nfs4_cb_procedures[NFSPROC4_CLNT_CB_NULL],
-	if (atomic_read(&cb->cb_set))
+		.rpc_argp       = clp,
-		return;
+	};
+	struct rpc_clnt *client;
+	int status;
 
 	/* Initialize address */
 	memset(&addr, 0, sizeof(addr));
@@ -416,29 +394,50 @@ nfsd4_probe_callback(struct nfs4_client *clp)
 	program->stats->program = program;
 
 	/* Create RPC client */
-	cb->cb_client = rpc_create(&args);
+	client = rpc_create(&args);
-	if (IS_ERR(cb->cb_client)) {
+	if (IS_ERR(client)) {
 		dprintk("NFSD: couldn't create callback client\n");
+		status = PTR_ERR(client);
 		goto out_err;
 	}
 
+	status = rpc_call_sync(client, &msg, RPC_TASK_SOFT);
+
+	if (status)
+		goto out_release_client;
+
+	cb->cb_client = client;
+	atomic_set(&cb->cb_set, 1);
+	put_nfs4_client(clp);
+	return 0;
+out_release_client:
+	rpc_shutdown_client(client);
+out_err:
+	put_nfs4_client(clp);
+	dprintk("NFSD: warning: no callback path to client %.*s\n",
+		(int)clp->cl_name.len, clp->cl_name.data);
+	return status;
+}
+
+/*
+ * Set up the callback client and put a NFSPROC4_CB_NULL on the wire...
+ */
+void
+nfsd4_probe_callback(struct nfs4_client *clp)
+{
+	struct task_struct *t;
+
+	BUG_ON(atomic_read(&clp->cl_callback.cb_set));
+
 	/* the task holds a reference to the nfs4_client struct */
 	atomic_inc(&clp->cl_count);
 
 	t = kthread_run(do_probe_callback, clp, "nfs4_cb_probe");
 
 	if (IS_ERR(t))
-		goto out_release_clp;
+		atomic_dec(&clp->cl_count);
 
 	return;
-
-out_release_clp:
-	atomic_dec(&clp->cl_count);
-	rpc_shutdown_client(cb->cb_client);
-out_err:
-	cb->cb_client = NULL;
-	dprintk("NFSD: warning: no callback path to client %.*s\n",
-		(int)clp->cl_name.len, clp->cl_name.data);
 }
 
 /*
@@ -458,9 +457,6 @@ nfsd4_cb_recall(struct nfs4_delegation *dp)
 	int retries = 1;
 	int status = 0;
 
-	if ((!atomic_read(&clp->cl_callback.cb_set)) || !clnt)
-		return;
-
 	cbr->cbr_trunc = 0; /* XXX need to implement truncate optimization */
 	cbr->cbr_dp = dp;
 
@@ -469,6 +465,7 @@ nfsd4_cb_recall(struct nfs4_delegation *dp)
 		switch (status) {
 			case -EIO:
 				/* Network partition? */
+				atomic_set(&clp->cl_callback.cb_set, 0);
 			case -EBADHANDLE:
 			case -NFS4ERR_BAD_STATEID:
 				/* Race: client probably got cb_recall
@@ -481,11 +478,10 @@ nfsd4_cb_recall(struct nfs4_delegation *dp)
 		status = rpc_call_sync(clnt, &msg, RPC_TASK_SOFT);
 	}
 out_put_cred:
-	if (status == -EIO)
+	/*
-		atomic_set(&clp->cl_callback.cb_set, 0);
+	 * Success or failure, now we're either waiting for lease expiration
-	/* Success or failure, now we're either waiting for lease expiration
+	 * or deleg_return.
-	 * or deleg_return. */
+	 */
-	dprintk("NFSD: nfs4_cb_recall: dp %p dl_flock %p dl_count %d\n",dp, dp->dl_flock, atomic_read(&dp->dl_count));
 	put_nfs4_client(clp);
 	nfs4_put_delegation(dp);
 	return;

fs/nfsd/nfs4idmap.c

@@ -255,13 +255,10 @@ idtoname_parse(struct cache_detail *cd, char *buf, int buflen)
 		goto out;
 	if (len == 0)
 		set_bit(CACHE_NEGATIVE, &ent.h.flags);
-	else {
+	else if (len >= IDMAP_NAMESZ)
-		if (error >= IDMAP_NAMESZ) {
-			error = -EINVAL;
 		goto out;
-		}
+	else
 		memcpy(ent.name, buf1, sizeof(ent.name));
-	}
 	error = -ENOMEM;
 	res = idtoname_update(&ent, res);
 	if (res == NULL)
@@ -467,20 +464,25 @@ nametoid_update(struct ent *new, struct ent *old)
  * Exported API
  */
 
-void
+int
 nfsd_idmap_init(void)
 {
-	cache_register(&idtoname_cache);
+	int rv;
-	cache_register(&nametoid_cache);
+
+	rv = cache_register(&idtoname_cache);
+	if (rv)
+		return rv;
+	rv = cache_register(&nametoid_cache);
+	if (rv)
+		cache_unregister(&idtoname_cache);
+	return rv;
 }
 
 void
 nfsd_idmap_shutdown(void)
 {
-	if (cache_unregister(&idtoname_cache))
+	cache_unregister(&idtoname_cache);
-		printk(KERN_ERR "nfsd: failed to unregister idtoname cache\n");
+	cache_unregister(&nametoid_cache);
-	if (cache_unregister(&nametoid_cache))
-		printk(KERN_ERR "nfsd: failed to unregister nametoid cache\n");
 }
 
 /*

fs/nfsd/nfs4proc.c

@@ -750,7 +750,7 @@ _nfsd4_verify(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 				    cstate->current_fh.fh_export,
 				    cstate->current_fh.fh_dentry, buf,
 				    &count, verify->ve_bmval,
-				    rqstp);
+				    rqstp, 0);
 
 	/* this means that nfsd4_encode_fattr() ran out of space */
 	if (status == nfserr_resource && count == 0)

fs/nfsd/nfs4state.c

@@ -61,7 +61,6 @@ static time_t lease_time = 90;     /* default lease time */
 static time_t user_lease_time = 90;
 static time_t boot_time;
 static int in_grace = 1;
-static u32 current_clientid = 1;
 static u32 current_ownerid = 1;
 static u32 current_fileid = 1;
 static u32 current_delegid = 1;
@@ -340,21 +339,20 @@ STALE_CLIENTID(clientid_t *clid)
  * This type of memory management is somewhat inefficient, but we use it
  * anyway since SETCLIENTID is not a common operation.
  */
-static inline struct nfs4_client *
+static struct nfs4_client *alloc_client(struct xdr_netobj name)
-alloc_client(struct xdr_netobj name)
 {
 	struct nfs4_client *clp;
 
-	if ((clp = kzalloc(sizeof(struct nfs4_client), GFP_KERNEL))!= NULL) {
+	clp = kzalloc(sizeof(struct nfs4_client), GFP_KERNEL);
-		if ((clp->cl_name.data = kmalloc(name.len, GFP_KERNEL)) != NULL) {
+	if (clp == NULL)
+		return NULL;
+	clp->cl_name.data = kmalloc(name.len, GFP_KERNEL);
+	if (clp->cl_name.data == NULL) {
+		kfree(clp);
+		return NULL;
+	}
 	memcpy(clp->cl_name.data, name.data, name.len);
 	clp->cl_name.len = name.len;
-		}
-		else {
-			kfree(clp);
-			clp = NULL;
-		}
-	}
 	return clp;
 }
 
@@ -363,8 +361,11 @@ shutdown_callback_client(struct nfs4_client *clp)
 {
 	struct rpc_clnt *clnt = clp->cl_callback.cb_client;
 
-	/* shutdown rpc client, ending any outstanding recall rpcs */
 	if (clnt) {
+		/*
+		 * Callback threads take a reference on the client, so there
+		 * should be no outstanding callbacks at this point.
+		 */
 		clp->cl_callback.cb_client = NULL;
 		rpc_shutdown_client(clnt);
 	}
@@ -422,12 +423,13 @@ expire_client(struct nfs4_client *clp)
 	put_nfs4_client(clp);
 }
 
-static struct nfs4_client *
+static struct nfs4_client *create_client(struct xdr_netobj name, char *recdir)
-create_client(struct xdr_netobj name, char *recdir) {
+{
 	struct nfs4_client *clp;
 
-	if (!(clp = alloc_client(name)))
+	clp = alloc_client(name);
-		goto out;
+	if (clp == NULL)
+		return NULL;
 	memcpy(clp->cl_recdir, recdir, HEXDIR_LEN);
 	atomic_set(&clp->cl_count, 1);
 	atomic_set(&clp->cl_callback.cb_set, 0);
@@ -436,32 +438,30 @@ create_client(struct xdr_netobj name, char *recdir) {
 	INIT_LIST_HEAD(&clp->cl_openowners);
 	INIT_LIST_HEAD(&clp->cl_delegations);
 	INIT_LIST_HEAD(&clp->cl_lru);
-out:
 	return clp;
 }
 
-static void
+static void copy_verf(struct nfs4_client *target, nfs4_verifier *source)
-copy_verf(struct nfs4_client *target, nfs4_verifier *source) {
+{
-	memcpy(target->cl_verifier.data, source->data, sizeof(target->cl_verifier.data));
+	memcpy(target->cl_verifier.data, source->data,
+			sizeof(target->cl_verifier.data));
 }
 
-static void
+static void copy_clid(struct nfs4_client *target, struct nfs4_client *source)
-copy_clid(struct nfs4_client *target, struct nfs4_client *source) {
+{
 	target->cl_clientid.cl_boot = source->cl_clientid.cl_boot; 
 	target->cl_clientid.cl_id = source->cl_clientid.cl_id; 
 }
 
-static void
+static void copy_cred(struct svc_cred *target, struct svc_cred *source)
-copy_cred(struct svc_cred *target, struct svc_cred *source) {
+{
-
 	target->cr_uid = source->cr_uid;
 	target->cr_gid = source->cr_gid;
 	target->cr_group_info = source->cr_group_info;
 	get_group_info(target->cr_group_info);
 }
 
-static inline int
+static int same_name(const char *n1, const char *n2)
-same_name(const char *n1, const char *n2)
 {
 	return 0 == memcmp(n1, n2, HEXDIR_LEN);
 }
@@ -485,26 +485,26 @@ same_creds(struct svc_cred *cr1, struct svc_cred *cr2)
 	return cr1->cr_uid == cr2->cr_uid;
 }
 
-static void
+static void gen_clid(struct nfs4_client *clp)
-gen_clid(struct nfs4_client *clp) {
+{
+	static u32 current_clientid = 1;
+
 	clp->cl_clientid.cl_boot = boot_time;
 	clp->cl_clientid.cl_id = current_clientid++; 
 }
 
-static void
+static void gen_confirm(struct nfs4_client *clp)
-gen_confirm(struct nfs4_client *clp) {
+{
-	struct timespec 	tv;
+	static u32 i;
-	u32 *			p;
+	u32 *p;
 
-	tv = CURRENT_TIME;
 	p = (u32 *)clp->cl_confirm.data;
-	*p++ = tv.tv_sec;
+	*p++ = get_seconds();
-	*p++ = tv.tv_nsec;
+	*p++ = i++;
 }
 
-static int
+static int check_name(struct xdr_netobj name)
-check_name(struct xdr_netobj name) {
+{
-
 	if (name.len == 0) 
 		return 0;
 	if (name.len > NFS4_OPAQUE_LIMIT) {
@@ -683,39 +683,6 @@ out_err:
 	return;
 }
 
-/*
- * RFC 3010 has a complex implmentation description of processing a 
- * SETCLIENTID request consisting of 5 bullets, labeled as 
- * CASE0 - CASE4 below.
- *
- * NOTES:
- * 	callback information will be processed in a future patch
- *
- *	an unconfirmed record is added when:
- *      NORMAL (part of CASE 4): there is no confirmed nor unconfirmed record.
- *	CASE 1: confirmed record found with matching name, principal,
- *		verifier, and clientid.
- *	CASE 2: confirmed record found with matching name, principal,
- *		and there is no unconfirmed record with matching
- *		name and principal
- *
- *      an unconfirmed record is replaced when:
- *	CASE 3: confirmed record found with matching name, principal,
- *		and an unconfirmed record is found with matching 
- *		name, principal, and with clientid and
- *		confirm that does not match the confirmed record.
- *	CASE 4: there is no confirmed record with matching name and 
- *		principal. there is an unconfirmed record with 
- *		matching name, principal.
- *
- *	an unconfirmed record is deleted when:
- *	CASE 1: an unconfirmed record that matches input name, verifier,
- *		and confirmed clientid.
- *	CASE 4: any unconfirmed records with matching name and principal
- *		that exist after an unconfirmed record has been replaced
- *		as described above.
- *
- */
 __be32
 nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		  struct nfsd4_setclientid *setclid)
@@ -748,11 +715,7 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 	nfs4_lock_state();
 	conf = find_confirmed_client_by_str(dname, strhashval);
 	if (conf) {
-		/* 
+		/* RFC 3530 14.2.33 CASE 0: */
-		 * CASE 0:
-		 * clname match, confirmed, different principal
-		 * or different ip_address
-		 */
 		status = nfserr_clid_inuse;
 		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred)
 				|| conf->cl_addr != sin->sin_addr.s_addr) {
@@ -761,12 +724,17 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 			goto out;
 		}
 	}
+	/*
+	 * section 14.2.33 of RFC 3530 (under the heading "IMPLEMENTATION")
+	 * has a description of SETCLIENTID request processing consisting
+	 * of 5 bullet points, labeled as CASE0 - CASE4 below.
+	 */
 	unconf = find_unconfirmed_client_by_str(dname, strhashval);
 	status = nfserr_resource;
 	if (!conf) {
 		/*
-		 * CASE 4:
+		 * RFC 3530 14.2.33 CASE 4:
-		 * placed first, because it is the normal case.
+		 * placed first, because it is the normal case
 		 */
 		if (unconf)
 			expire_client(unconf);
@@ -776,17 +744,8 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		gen_clid(new);
 	} else if (same_verf(&conf->cl_verifier, &clverifier)) {
 		/*
-		 * CASE 1:
+		 * RFC 3530 14.2.33 CASE 1:
-		 * cl_name match, confirmed, principal match
+		 * probable callback update
-		 * verifier match: probable callback update
-		 *
-		 * remove any unconfirmed nfs4_client with 
-		 * matching cl_name, cl_verifier, and cl_clientid
-		 *
-		 * create and insert an unconfirmed nfs4_client with same 
-		 * cl_name, cl_verifier, and cl_clientid as existing 
-		 * nfs4_client,  but with the new callback info and a 
-		 * new cl_confirm
 		 */
 		if (unconf) {
 			/* Note this is removing unconfirmed {*x***},
@@ -802,43 +761,25 @@ nfsd4_setclientid(struct svc_rqst *rqstp, struct nfsd4_compound_state *cstate,
 		copy_clid(new, conf);
 	} else if (!unconf) {
 		/*
-		 * CASE 2:
+		 * RFC 3530 14.2.33 CASE 2:
-		 * clname match, confirmed, principal match
+		 * probable client reboot; state will be removed if
-		 * verfier does not match
+		 * confirmed.
-		 * no unconfirmed. create a new unconfirmed nfs4_client
-		 * using input clverifier, clname, and callback info
-		 * and generate a new cl_clientid and cl_confirm.
 		 */
 		new = create_client(clname, dname);
 		if (new == NULL)
 			goto out;
 		gen_clid(new);
-	} else if (!same_verf(&conf->cl_confirm, &unconf->cl_confirm)) {
+	} else {
 		/*
-		 * CASE3:
+		 * RFC 3530 14.2.33 CASE 3:
-		 * confirmed found (name, principal match)
+		 * probable client reboot; state will be removed if
-		 * confirmed verifier does not match input clverifier
+		 * confirmed.
-		 *
-		 * unconfirmed found (name match)
-		 * confirmed->cl_confirm != unconfirmed->cl_confirm
-		 *
-		 * remove unconfirmed.
-		 *
-		 * create an unconfirmed nfs4_client 
-		 * with same cl_name as existing confirmed nfs4_client, 
-		 * but with new callback info, new cl_clientid,
-		 * new cl_verifier and a new cl_confirm
 		 */
 		expire_client(unconf);
 		new = create_client(clname, dname);
 		if (new == NULL)
 			goto out;
 		gen_clid(new);
-	} else {
-		/* No cases hit !!! */
-		status = nfserr_inval;
-		goto out;
-
 	}
 	copy_verf(new, &clverifier);
 	new->cl_addr = sin->sin_addr.s_addr;
@@ -857,11 +798,9 @@ out:
 
 
 /*
- * RFC 3010 has a complex implmentation description of processing a 
+ * Section 14.2.34 of RFC 3530 (under the heading "IMPLEMENTATION") has
- * SETCLIENTID_CONFIRM request consisting of 4 bullets describing
+ * a description of SETCLIENTID_CONFIRM request processing consisting of 4
- * processing on a DRC miss, labeled as CASE1 - CASE4 below.
+ * bullets, labeled as CASE1 - CASE4 below.
- *
- * NOTE: callback information will be processed here in a future patch
  */
 __be32
 nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
@@ -892,15 +831,15 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 	if (unconf && unconf->cl_addr != sin->sin_addr.s_addr)
 		goto out;
 
-	if ((conf && unconf) && 
+	/*
-	    (same_verf(&unconf->cl_confirm, &confirm)) &&
+	 * section 14.2.34 of RFC 3530 has a description of
-	    (same_verf(&conf->cl_verifier, &unconf->cl_verifier)) &&
+	 * SETCLIENTID_CONFIRM request processing consisting
-	    (same_name(conf->cl_recdir,unconf->cl_recdir))  &&
+	 * of 4 bullet points, labeled as CASE1 - CASE4 below.
-	    (!same_verf(&conf->cl_confirm, &unconf->cl_confirm))) {
+	 */
-		/* CASE 1:
+	if (conf && unconf && same_verf(&confirm, &unconf->cl_confirm)) {
-		* unconf record that matches input clientid and input confirm.
+		/*
-		* conf record that matches input clientid.
+		 * RFC 3530 14.2.34 CASE 1:
-		* conf and unconf records match names, verifiers
+		 * callback update
 		 */
 		if (!same_creds(&conf->cl_cred, &unconf->cl_cred))
 			status = nfserr_clid_inuse;
@@ -914,15 +853,11 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 			status = nfs_ok;
 
 		}
-	} else if ((conf && !unconf) ||
+	} else if (conf && !unconf) {
-	    ((conf && unconf) && 
+		/*
-	     (!same_verf(&conf->cl_verifier, &unconf->cl_verifier) ||
+		 * RFC 3530 14.2.34 CASE 2:
-	      !same_name(conf->cl_recdir, unconf->cl_recdir)))) {
+		 * probable retransmitted request; play it safe and
-		/* CASE 2:
+		 * do nothing.
-		 * conf record that matches input clientid.
-		 * if unconf record matches input clientid, then
-		 * unconf->cl_name or unconf->cl_verifier don't match the
-		 * conf record.
 		 */
 		if (!same_creds(&conf->cl_cred, &rqstp->rq_cred))
 			status = nfserr_clid_inuse;
@@ -930,10 +865,9 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 			status = nfs_ok;
 	} else if (!conf && unconf
 			&& same_verf(&unconf->cl_confirm, &confirm)) {
-		/* CASE 3:
+		/*
-		 * conf record not found.
+		 * RFC 3530 14.2.34 CASE 3:
-		 * unconf record found.
+		 * Normal case; new or rebooted client:
-		 * unconf->cl_confirm matches input confirm
 		 */
 		if (!same_creds(&unconf->cl_cred, &rqstp->rq_cred)) {
 			status = nfserr_clid_inuse;
@@ -948,16 +882,15 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 			}
 			move_to_confirmed(unconf);
 			conf = unconf;
+			nfsd4_probe_callback(conf);
 			status = nfs_ok;
 		}
 	} else if ((!conf || (conf && !same_verf(&conf->cl_confirm, &confirm)))
 	    && (!unconf || (unconf && !same_verf(&unconf->cl_confirm,
 				    				&confirm)))) {
-		/* CASE 4:
+		/*
-		 * conf record not found, or if conf, conf->cl_confirm does not
+		 * RFC 3530 14.2.34 CASE 4:
-		 * match input confirm.
+		 * Client probably hasn't noticed that we rebooted yet.
-		 * unconf record not found, or if unconf, unconf->cl_confirm
-		 * does not match input confirm.
 		 */
 		status = nfserr_stale_clientid;
 	} else {
@@ -965,8 +898,6 @@ nfsd4_setclientid_confirm(struct svc_rqst *rqstp,
 		status = nfserr_clid_inuse;
 	}
 out:
-	if (!status)
-		nfsd4_probe_callback(conf);
 	nfs4_unlock_state();
 	return status;
 }
@@ -1226,14 +1157,19 @@ find_file(struct inode *ino)
 	return NULL;
 }
 
-static int access_valid(u32 x)
+static inline int access_valid(u32 x)
 {
-	return (x > 0 && x < 4);
+	if (x < NFS4_SHARE_ACCESS_READ)
+		return 0;
+	if (x > NFS4_SHARE_ACCESS_BOTH)
+		return 0;
+	return 1;
 }
 
-static int deny_valid(u32 x)
+static inline int deny_valid(u32 x)
 {
-	return (x >= 0 && x < 5);
+	/* Note: unlike access bits, deny bits may be zero. */
+	return x <= NFS4_SHARE_DENY_BOTH;
 }
 
 static void
@@ -2162,8 +2098,10 @@ nfs4_preprocess_seqid_op(struct svc_fh *current_fh, u32 seqid, stateid_t *statei
 		goto check_replay;
 	}
 
+	*stpp = stp;
+	*sopp = sop = stp->st_stateowner;
+
 	if (lock) {
-		struct nfs4_stateowner *sop = stp->st_stateowner;
 		clientid_t *lockclid = &lock->v.new.clientid;
 		struct nfs4_client *clp = sop->so_client;
 		int lkflg = 0;
@@ -2193,9 +2131,6 @@ nfs4_preprocess_seqid_op(struct svc_fh *current_fh, u32 seqid, stateid_t *statei
 		return nfserr_bad_stateid;
 	}
 
-	*stpp = stp;
-	*sopp = sop = stp->st_stateowner;
-
 	/*
 	*  We now validate the seqid and stateid generation numbers.
 	*  For the moment, we ignore the possibility of 

fs/nfsd/nfs4xdr.c

@@ -148,12 +148,12 @@ xdr_error:					\
 	}					\
 } while (0)
 
-static __be32 *read_buf(struct nfsd4_compoundargs *argp, int nbytes)
+static __be32 *read_buf(struct nfsd4_compoundargs *argp, u32 nbytes)
 {
 	/* We want more bytes than seem to be available.
 	 * Maybe we need a new page, maybe we have just run out
 	 */
-	int avail = (char*)argp->end - (char*)argp->p;
+	unsigned int avail = (char *)argp->end - (char *)argp->p;
 	__be32 *p;
 	if (avail + argp->pagelen < nbytes)
 		return NULL;
@@ -169,6 +169,11 @@ static __be32 *read_buf(struct nfsd4_compoundargs *argp, int nbytes)
 			return NULL;
 		
 	}
+	/*
+	 * The following memcpy is safe because read_buf is always
+	 * called with nbytes > avail, and the two cases above both
+	 * guarantee p points to at least nbytes bytes.
+	 */
 	memcpy(p, argp->p, avail);
 	/* step to next page */
 	argp->p = page_address(argp->pagelist[0]);
@@ -1448,7 +1453,7 @@ static __be32 fattr_handle_absent_fs(u32 *bmval0, u32 *bmval1, u32 *rdattr_err)
 __be32
 nfsd4_encode_fattr(struct svc_fh *fhp, struct svc_export *exp,
 		struct dentry *dentry, __be32 *buffer, int *countp, u32 *bmval,
-		struct svc_rqst *rqstp)
+		struct svc_rqst *rqstp, int ignore_crossmnt)
 {
 	u32 bmval0 = bmval[0];
 	u32 bmval1 = bmval[1];
@@ -1828,7 +1833,12 @@ out_acl:
 	if (bmval1 & FATTR4_WORD1_MOUNTED_ON_FILEID) {
 		if ((buflen -= 8) < 0)
                 	goto out_resource;
-		if (exp->ex_mnt->mnt_root->d_inode == dentry->d_inode) {
+		/*
+		 * Get parent's attributes if not ignoring crossmount
+		 * and this is the root of a cross-mounted filesystem.
+		 */
+		if (ignore_crossmnt == 0 &&
+		    exp->ex_mnt->mnt_root->d_inode == dentry->d_inode) {
 			err = vfs_getattr(exp->ex_mnt->mnt_parent,
 				exp->ex_mnt->mnt_mountpoint, &stat);
 			if (err)
@@ -1864,13 +1874,25 @@ nfsd4_encode_dirent_fattr(struct nfsd4_readdir *cd,
 	struct svc_export *exp = cd->rd_fhp->fh_export;
 	struct dentry *dentry;
 	__be32 nfserr;
+	int ignore_crossmnt = 0;
 
 	dentry = lookup_one_len(name, cd->rd_fhp->fh_dentry, namlen);
 	if (IS_ERR(dentry))
 		return nfserrno(PTR_ERR(dentry));
 
 	exp_get(exp);
-	if (d_mountpoint(dentry)) {
+	/*
+	 * In the case of a mountpoint, the client may be asking for
+	 * attributes that are only properties of the underlying filesystem
+	 * as opposed to the cross-mounted file system. In such a case,
+	 * we will not follow the cross mount and will fill the attribtutes
+	 * directly from the mountpoint dentry.
+	 */
+	if (d_mountpoint(dentry) &&
+	    (cd->rd_bmval[0] & ~FATTR4_WORD0_RDATTR_ERROR) == 0 &&
+	    (cd->rd_bmval[1] & ~FATTR4_WORD1_MOUNTED_ON_FILEID) == 0)
+		ignore_crossmnt = 1;
+	else if (d_mountpoint(dentry)) {
 		int err;
 
 		/*
@@ -1889,7 +1911,7 @@ nfsd4_encode_dirent_fattr(struct nfsd4_readdir *cd,
 
 	}
 	nfserr = nfsd4_encode_fattr(NULL, exp, dentry, p, buflen, cd->rd_bmval,
-					cd->rd_rqstp);
+					cd->rd_rqstp, ignore_crossmnt);
 out_put:
 	dput(dentry);
 	exp_put(exp);
@@ -2043,7 +2065,7 @@ nfsd4_encode_getattr(struct nfsd4_compoundres *resp, __be32 nfserr, struct nfsd4
 	buflen = resp->end - resp->p - (COMPOUND_ERR_SLACK_SPACE >> 2);
 	nfserr = nfsd4_encode_fattr(fhp, fhp->fh_export, fhp->fh_dentry,
 				    resp->p, &buflen, getattr->ga_bmval,
-				    resp->rqstp);
+				    resp->rqstp, 0);
 	if (!nfserr)
 		resp->p += buflen;
 	return nfserr;

fs/nfsd/nfscache.c

@@ -44,17 +44,17 @@ static int	nfsd_cache_append(struct svc_rqst *rqstp, struct kvec *vec);
  */
 static DEFINE_SPINLOCK(cache_lock);
 
-void
+int nfsd_reply_cache_init(void)
-nfsd_cache_init(void)
 {
 	struct svc_cacherep	*rp;
 	int			i;
 
 	INIT_LIST_HEAD(&lru_head);
 	i = CACHESIZE;
-	while(i) {
+	while (i) {
 		rp = kmalloc(sizeof(*rp), GFP_KERNEL);
-		if (!rp) break;
+		if (!rp)
+			goto out_nomem;
 		list_add(&rp->c_lru, &lru_head);
 		rp->c_state = RC_UNUSED;
 		rp->c_type = RC_NOCACHE;
@@ -62,23 +62,19 @@ nfsd_cache_init(void)
 		i--;
 	}
 
-	if (i)
-		printk (KERN_ERR "nfsd: cannot allocate all %d cache entries, only got %d\n",
-			CACHESIZE, CACHESIZE-i);
-
 	hash_list = kcalloc (HASHSIZE, sizeof(struct hlist_head), GFP_KERNEL);
-	if (!hash_list) {
+	if (!hash_list)
-		nfsd_cache_shutdown();
+		goto out_nomem;
-		printk (KERN_ERR "nfsd: cannot allocate %Zd bytes for hash list\n",
-			HASHSIZE * sizeof(struct hlist_head));
-		return;
-	}
 
 	cache_disabled = 0;
+	return 0;
+out_nomem:
+	printk(KERN_ERR "nfsd: failed to allocate reply cache\n");
+	nfsd_reply_cache_shutdown();
+	return -ENOMEM;
 }
 
-void
+void nfsd_reply_cache_shutdown(void)
-nfsd_cache_shutdown(void)
 {
 	struct svc_cacherep	*rp;
 

fs/nfsd/nfsctl.c

@@ -304,6 +304,9 @@ static ssize_t write_filehandle(struct file *file, char *buf, size_t size)
 	struct auth_domain *dom;
 	struct knfsd_fh fh;
 
+	if (size == 0)
+		return -EINVAL;
+
 	if (buf[size-1] != '\n')
 		return -EINVAL;
 	buf[size-1] = 0;
@@ -503,7 +506,7 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
 		int len = 0;
 		lock_kernel();
 		if (nfsd_serv)
-			len = svc_sock_names(buf, nfsd_serv, NULL);
+			len = svc_xprt_names(nfsd_serv, buf, 0);
 		unlock_kernel();
 		return len;
 	}
@@ -540,7 +543,7 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
 		}
 		return err < 0 ? err : 0;
 	}
-	if (buf[0] == '-') {
+	if (buf[0] == '-' && isdigit(buf[1])) {
 		char *toclose = kstrdup(buf+1, GFP_KERNEL);
 		int len = 0;
 		if (!toclose)
@@ -554,6 +557,53 @@ static ssize_t write_ports(struct file *file, char *buf, size_t size)
 		kfree(toclose);
 		return len;
 	}
+	/*
+	 * Add a transport listener by writing it's transport name
+	 */
+	if (isalpha(buf[0])) {
+		int err;
+		char transport[16];
+		int port;
+		if (sscanf(buf, "%15s %4d", transport, &port) == 2) {
+			err = nfsd_create_serv();
+			if (!err) {
+				err = svc_create_xprt(nfsd_serv,
+						      transport, port,
+						      SVC_SOCK_ANONYMOUS);
+				if (err == -ENOENT)
+					/* Give a reasonable perror msg for
+					 * bad transport string */
+					err = -EPROTONOSUPPORT;
+			}
+			return err < 0 ? err : 0;
+		}
+	}
+	/*
+	 * Remove a transport by writing it's transport name and port number
+	 */
+	if (buf[0] == '-' && isalpha(buf[1])) {
+		struct svc_xprt *xprt;
+		int err = -EINVAL;
+		char transport[16];
+		int port;
+		if (sscanf(&buf[1], "%15s %4d", transport, &port) == 2) {
+			if (port == 0)
+				return -EINVAL;
+			lock_kernel();
+			if (nfsd_serv) {
+				xprt = svc_find_xprt(nfsd_serv, transport,
+						     AF_UNSPEC, port);
+				if (xprt) {
+					svc_close_xprt(xprt);
+					svc_xprt_put(xprt);
+					err = 0;
+				} else
+					err = -ENOTCONN;
+			}
+			unlock_kernel();
+			return err < 0 ? err : 0;
+		}
+	}
 	return -EINVAL;
 }
 
@@ -616,7 +666,7 @@ static ssize_t write_recoverydir(struct file *file, char *buf, size_t size)
 	char *recdir;
 	int len, status;
 
-	if (size > PATH_MAX || buf[size-1] != '\n')
+	if (size == 0 || size > PATH_MAX || buf[size-1] != '\n')
 		return -EINVAL;
 	buf[size-1] = 0;
 
@@ -674,6 +724,27 @@ static struct file_system_type nfsd_fs_type = {
 	.kill_sb	= kill_litter_super,
 };
 
+#ifdef CONFIG_PROC_FS
+static int create_proc_exports_entry(void)
+{
+	struct proc_dir_entry *entry;
+
+	entry = proc_mkdir("fs/nfs", NULL);
+	if (!entry)
+		return -ENOMEM;
+	entry = create_proc_entry("fs/nfs/exports", 0, NULL);
+	if (!entry)
+		return -ENOMEM;
+	entry->proc_fops =  &exports_operations;
+	return 0;
+}
+#else /* CONFIG_PROC_FS */
+static int create_proc_exports_entry(void)
+{
+	return 0;
+}
+#endif
+
 static int __init init_nfsd(void)
 {
 	int retval;
@@ -683,32 +754,43 @@ static int __init init_nfsd(void)
 	if (retval)
 		return retval;
 	nfsd_stat_init();	/* Statistics */
-	nfsd_cache_init();	/* RPC reply cache */
+	retval = nfsd_reply_cache_init();
-	nfsd_export_init();	/* Exports table */
+	if (retval)
+		goto out_free_stat;
+	retval = nfsd_export_init();
+	if (retval)
+		goto out_free_cache;
 	nfsd_lockd_init();	/* lockd->nfsd callbacks */
-	nfsd_idmap_init();      /* Name to ID mapping */
+	retval = nfsd_idmap_init();
-	if (proc_mkdir("fs/nfs", NULL)) {
+	if (retval)
-		struct proc_dir_entry *entry;
+		goto out_free_lockd;
-		entry = create_proc_entry("fs/nfs/exports", 0, NULL);
+	retval = create_proc_exports_entry();
-		if (entry)
+	if (retval)
-			entry->proc_fops =  &exports_operations;
+		goto out_free_idmap;
-	}
 	retval = register_filesystem(&nfsd_fs_type);
-	if (retval) {
+	if (retval)
-		nfsd_export_shutdown();
+		goto out_free_all;
-		nfsd_cache_shutdown();
+	return 0;
+out_free_all:
 	remove_proc_entry("fs/nfs/exports", NULL);
 	remove_proc_entry("fs/nfs", NULL);
-		nfsd_stat_shutdown();
+out_free_idmap:
+	nfsd_idmap_shutdown();
+out_free_lockd:
 	nfsd_lockd_shutdown();
-	}
+	nfsd_export_shutdown();
+out_free_cache:
+	nfsd_reply_cache_shutdown();
+out_free_stat:
+	nfsd_stat_shutdown();
+	nfsd4_free_slabs();
 	return retval;
 }
 
 static void __exit exit_nfsd(void)
 {
 	nfsd_export_shutdown();
-	nfsd_cache_shutdown();
+	nfsd_reply_cache_shutdown();
 	remove_proc_entry("fs/nfs/exports", NULL);
 	remove_proc_entry("fs/nfs", NULL);
 	nfsd_stat_shutdown();

fs/nfsd/nfsfh.c

@@ -22,6 +22,7 @@
 #include <linux/sunrpc/svc.h>
 #include <linux/sunrpc/svcauth_gss.h>
 #include <linux/nfsd/nfsd.h>
+#include "auth.h"
 
 #define NFSDDBG_FACILITY		NFSDDBG_FH
 

fs/nfsd/nfssvc.c

@@ -155,8 +155,8 @@ static int killsig;	/* signal that was used to kill last nfsd */
 static void nfsd_last_thread(struct svc_serv *serv)
 {
 	/* When last nfsd thread exits we need to do some clean-up */
-	struct svc_sock *svsk;
+	struct svc_xprt *xprt;
-	list_for_each_entry(svsk, &serv->sv_permsocks, sk_list)
+	list_for_each_entry(xprt, &serv->sv_permsocks, xpt_list)
 		lockd_down();
 	nfsd_serv = NULL;
 	nfsd_racache_shutdown();
@@ -236,7 +236,7 @@ static int nfsd_init_socks(int port)
 
 	error = lockd_up(IPPROTO_UDP);
 	if (error >= 0) {
-		error = svc_makesock(nfsd_serv, IPPROTO_UDP, port,
+		error = svc_create_xprt(nfsd_serv, "udp", port,
 					SVC_SOCK_DEFAULTS);
 		if (error < 0)
 			lockd_down();
@@ -247,7 +247,7 @@ static int nfsd_init_socks(int port)
 #ifdef CONFIG_NFSD_TCP
 	error = lockd_up(IPPROTO_TCP);
 	if (error >= 0) {
-		error = svc_makesock(nfsd_serv, IPPROTO_TCP, port,
+		error = svc_create_xprt(nfsd_serv, "tcp", port,
 					SVC_SOCK_DEFAULTS);
 		if (error < 0)
 			lockd_down();

fs/nfsd/nfsxdr.c

@@ -15,6 +15,7 @@
 #include <linux/nfsd/nfsd.h>
 #include <linux/nfsd/xdr.h>
 #include <linux/mm.h>
+#include "auth.h"
 
 #define NFSDDBG_FACILITY		NFSDDBG_XDR
 
@@ -62,10 +63,10 @@ encode_fh(__be32 *p, struct svc_fh *fhp)
  * no slashes or null bytes.
  */
 static __be32 *
-decode_filename(__be32 *p, char **namp, int *lenp)
+decode_filename(__be32 *p, char **namp, unsigned int *lenp)
 {
 	char		*name;
-	int		i;
+	unsigned int	i;
 
 	if ((p = xdr_decode_string_inplace(p, namp, lenp, NFS_MAXNAMLEN)) != NULL) {
 		for (i = 0, name = *namp; i < *lenp; i++, name++) {
@@ -78,10 +79,10 @@ decode_filename(__be32 *p, char **namp, int *lenp)
 }
 
 static __be32 *
-decode_pathname(__be32 *p, char **namp, int *lenp)
+decode_pathname(__be32 *p, char **namp, unsigned int *lenp)
 {
 	char		*name;
-	int		i;
+	unsigned int	i;
 
 	if ((p = xdr_decode_string_inplace(p, namp, lenp, NFS_MAXPATHLEN)) != NULL) {
 		for (i = 0, name = *namp; i < *lenp; i++, name++) {

fs/nfsd/vfs.c

@@ -132,7 +132,7 @@ out:
 
 __be32
 nfsd_lookup_dentry(struct svc_rqst *rqstp, struct svc_fh *fhp,
-		   const char *name, int len,
+		   const char *name, unsigned int len,
 		   struct svc_export **exp_ret, struct dentry **dentry_ret)
 {
 	struct svc_export	*exp;
@@ -226,7 +226,7 @@ out_nfserr:
  */
 __be32
 nfsd_lookup(struct svc_rqst *rqstp, struct svc_fh *fhp, const char *name,
-					int len, struct svc_fh *resfh)
+				unsigned int len, struct svc_fh *resfh)
 {
 	struct svc_export	*exp;
 	struct dentry		*dentry;
@@ -1151,6 +1151,26 @@ nfsd_commit(struct svc_rqst *rqstp, struct svc_fh *fhp,
 }
 #endif /* CONFIG_NFSD_V3 */
 
+__be32
+nfsd_create_setattr(struct svc_rqst *rqstp, struct svc_fh *resfhp,
+			struct iattr *iap)
+{
+	/*
+	 * Mode has already been set earlier in create:
+	 */
+	iap->ia_valid &= ~ATTR_MODE;
+	/*
+	 * Setting uid/gid works only for root.  Irix appears to
+	 * send along the gid on create when it tries to implement
+	 * setgid directories via NFS:
+	 */
+	if (current->fsuid != 0)
+		iap->ia_valid &= ~(ATTR_UID|ATTR_GID);
+	if (iap->ia_valid)
+		return nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
+	return 0;
+}
+
 /*
  * Create a file (regular, directory, device, fifo); UNIX sockets 
  * not yet implemented.
@@ -1167,6 +1187,7 @@ nfsd_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
 	struct dentry	*dentry, *dchild = NULL;
 	struct inode	*dirp;
 	__be32		err;
+	__be32		err2;
 	int		host_err;
 
 	err = nfserr_perm;
@@ -1257,16 +1278,9 @@ nfsd_create(struct svc_rqst *rqstp, struct svc_fh *fhp,
 	}
 
 
-	/* Set file attributes. Mode has already been set and
+	err2 = nfsd_create_setattr(rqstp, resfhp, iap);
-	 * setting uid/gid works only for root. Irix appears to
-	 * send along the gid when it tries to implement setgid
-	 * directories via NFS.
-	 */
-	if ((iap->ia_valid &= ~(ATTR_UID|ATTR_GID|ATTR_MODE)) != 0) {
-		__be32 err2 = nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
 	if (err2)
 		err = err2;
-	}
 	/*
 	 * Update the file handle to get the new inode info.
 	 */
@@ -1295,6 +1309,7 @@ nfsd_create_v3(struct svc_rqst *rqstp, struct svc_fh *fhp,
 	struct dentry	*dentry, *dchild = NULL;
 	struct inode	*dirp;
 	__be32		err;
+	__be32		err2;
 	int		host_err;
 	__u32		v_mtime=0, v_atime=0;
 
@@ -1399,16 +1414,10 @@ nfsd_create_v3(struct svc_rqst *rqstp, struct svc_fh *fhp,
 		iap->ia_atime.tv_nsec = 0;
 	}
 
-	/* Set file attributes.
-	 * Irix appears to send along the gid when it tries to
-	 * implement setgid directories via NFS. Clear out all that cruft.
-	 */
  set_attr:
-	if ((iap->ia_valid &= ~(ATTR_UID|ATTR_GID|ATTR_MODE)) != 0) {
+	err2 = nfsd_create_setattr(rqstp, resfhp, iap);
- 		__be32 err2 = nfsd_setattr(rqstp, resfhp, iap, 0, (time_t)0);
 	if (err2)
 		err = err2;
-	}
 
 	/*
 	 * Update the filehandle to get the new inode info.

include/linux/lockd/lockd.h

@@ -173,14 +173,17 @@ void		  nlmclnt_next_cookie(struct nlm_cookie *);
 /*
  * Host cache
  */
-struct nlm_host * nlmclnt_lookup_host(const struct sockaddr_in *, int, int, const char *, int);
+struct nlm_host  *nlmclnt_lookup_host(const struct sockaddr_in *, int, int,
-struct nlm_host * nlmsvc_lookup_host(struct svc_rqst *, const char *, int);
+					const char *, unsigned int);
+struct nlm_host  *nlmsvc_lookup_host(struct svc_rqst *, const char *,
+					unsigned int);
 struct rpc_clnt * nlm_bind_host(struct nlm_host *);
 void		  nlm_rebind_host(struct nlm_host *);
 struct nlm_host * nlm_get_host(struct nlm_host *);
 void		  nlm_release_host(struct nlm_host *);
 void		  nlm_shutdown_hosts(void);
-extern void	  nlm_host_rebooted(const struct sockaddr_in *, const char *, int, u32);
+extern void	  nlm_host_rebooted(const struct sockaddr_in *, const char *,
+					unsigned int, u32);
 void		  nsm_release(struct nsm_handle *);
 
 

include/linux/lockd/xdr.h

@@ -29,7 +29,7 @@ struct svc_rqst;
 /* Lock info passed via NLM */
 struct nlm_lock {
 	char *			caller;
-	int			len; 	/* length of "caller" */
+	unsigned int		len; 	/* length of "caller" */
 	struct nfs_fh		fh;
 	struct xdr_netobj	oh;
 	u32			svid;
@@ -78,7 +78,7 @@ struct nlm_res {
  */
 struct nlm_reboot {
 	char *		mon;
-	int		len;
+	unsigned int	len;
 	u32		state;
 	__be32		addr;
 	__be32		vers;

include/linux/nfsd/Kbuild

@@ -4,4 +4,3 @@ unifdef-y += stats.h
 unifdef-y += syscall.h
 unifdef-y += nfsfh.h
 unifdef-y += debug.h
-unifdef-y += auth.h

include/linux/nfsd/cache.h

@@ -72,8 +72,8 @@ enum {
  */
 #define RC_DELAY		(HZ/5)
 
-void	nfsd_cache_init(void);
+int	nfsd_reply_cache_init(void);
-void	nfsd_cache_shutdown(void);
+void	nfsd_reply_cache_shutdown(void);
 int	nfsd_cache_lookup(struct svc_rqst *, int);
 void	nfsd_cache_update(struct svc_rqst *, int, __be32 *);
 

include/linux/nfsd/export.h

@@ -122,7 +122,7 @@ __be32 check_nfsd_access(struct svc_export *exp, struct svc_rqst *rqstp);
 /*
  * Function declarations
  */
-void			nfsd_export_init(void);
+int			nfsd_export_init(void);
 void			nfsd_export_shutdown(void);
 void			nfsd_export_flush(void);
 void			exp_readlock(void);

include/linux/nfsd/nfsd.h

@@ -20,7 +20,6 @@
 #include <linux/nfsd/debug.h>
 #include <linux/nfsd/nfsfh.h>
 #include <linux/nfsd/export.h>
-#include <linux/nfsd/auth.h>
 #include <linux/nfsd/stats.h>
 /*
  * nfsd version
@@ -70,9 +69,9 @@ void		nfsd_racache_shutdown(void);
 int		nfsd_cross_mnt(struct svc_rqst *rqstp, struct dentry **dpp,
 		                struct svc_export **expp);
 __be32		nfsd_lookup(struct svc_rqst *, struct svc_fh *,
-				const char *, int, struct svc_fh *);
+				const char *, unsigned int, struct svc_fh *);
 __be32		 nfsd_lookup_dentry(struct svc_rqst *, struct svc_fh *,
-				const char *, int,
+				const char *, unsigned int,
 				struct svc_export **, struct dentry **);
 __be32		nfsd_setattr(struct svc_rqst *, struct svc_fh *,
 				struct iattr *, int, time_t);

include/linux/nfsd/syscall.h

@@ -18,7 +18,6 @@
 #include <linux/nfsd/const.h>
 #include <linux/nfsd/export.h>
 #include <linux/nfsd/nfsfh.h>
-#include <linux/nfsd/auth.h>
 
 /*
  * Version of the syscall interface

include/linux/nfsd/xdr.h

@@ -23,7 +23,7 @@ struct nfsd_sattrargs {
 struct nfsd_diropargs {
 	struct svc_fh		fh;
 	char *			name;
-	int			len;
+	unsigned int		len;
 };
 
 struct nfsd_readargs {
@@ -43,17 +43,17 @@ struct nfsd_writeargs {
 struct nfsd_createargs {
 	struct svc_fh		fh;
 	char *			name;
-	int			len;
+	unsigned int		len;
 	struct iattr		attrs;
 };
 
 struct nfsd_renameargs {
 	struct svc_fh		ffh;
 	char *			fname;
-	int			flen;
+	unsigned int		flen;
 	struct svc_fh		tfh;
 	char *			tname;
-	int			tlen;
+	unsigned int		tlen;
 };
 
 struct nfsd_readlinkargs {
@@ -65,15 +65,15 @@ struct nfsd_linkargs {
 	struct svc_fh		ffh;
 	struct svc_fh		tfh;
 	char *			tname;
-	int			tlen;
+	unsigned int		tlen;
 };
 
 struct nfsd_symlinkargs {
 	struct svc_fh		ffh;
 	char *			fname;
-	int			flen;
+	unsigned int		flen;
 	char *			tname;
-	int			tlen;
+	unsigned int		tlen;
 	struct iattr		attrs;
 };
 

include/linux/nfsd/xdr3.h

@@ -21,7 +21,7 @@ struct nfsd3_sattrargs {
 struct nfsd3_diropargs {
 	struct svc_fh		fh;
 	char *			name;
-	int			len;
+	unsigned int		len;
 };
 
 struct nfsd3_accessargs {
@@ -48,7 +48,7 @@ struct nfsd3_writeargs {
 struct nfsd3_createargs {
 	struct svc_fh		fh;
 	char *			name;
-	int			len;
+	unsigned int		len;
 	int			createmode;
 	struct iattr		attrs;
 	__be32 *		verf;
@@ -57,7 +57,7 @@ struct nfsd3_createargs {
 struct nfsd3_mknodargs {
 	struct svc_fh		fh;
 	char *			name;
-	int			len;
+	unsigned int		len;
 	__u32			ftype;
 	__u32			major, minor;
 	struct iattr		attrs;
@@ -66,10 +66,10 @@ struct nfsd3_mknodargs {
 struct nfsd3_renameargs {
 	struct svc_fh		ffh;
 	char *			fname;
-	int			flen;
+	unsigned int		flen;
 	struct svc_fh		tfh;
 	char *			tname;
-	int			tlen;
+	unsigned int		tlen;
 };
 
 struct nfsd3_readlinkargs {
@@ -81,15 +81,15 @@ struct nfsd3_linkargs {
 	struct svc_fh		ffh;
 	struct svc_fh		tfh;
 	char *			tname;
-	int			tlen;
+	unsigned int		tlen;
 };
 
 struct nfsd3_symlinkargs {
 	struct svc_fh		ffh;
 	char *			fname;
-	int			flen;
+	unsigned int		flen;
 	char *			tname;
-	int			tlen;
+	unsigned int		tlen;
 	struct iattr		attrs;
 };
 

include/linux/nfsd/xdr4.h

@@ -441,7 +441,7 @@ void nfsd4_encode_operation(struct nfsd4_compoundres *, struct nfsd4_op *);
 void nfsd4_encode_replay(struct nfsd4_compoundres *resp, struct nfsd4_op *op);
 __be32 nfsd4_encode_fattr(struct svc_fh *fhp, struct svc_export *exp,
 		       struct dentry *dentry, __be32 *buffer, int *countp,
-		       u32 *bmval, struct svc_rqst *);
+		       u32 *bmval, struct svc_rqst *, int ignore_crossmnt);
 extern __be32 nfsd4_setclientid(struct svc_rqst *rqstp,
 		struct nfsd4_compound_state *,
 		struct nfsd4_setclientid *setclid);

include/linux/nfsd_idmap.h

@@ -44,11 +44,16 @@
 #define IDMAP_NAMESZ 128
 
 #ifdef CONFIG_NFSD_V4
-void nfsd_idmap_init(void);
+int nfsd_idmap_init(void);
 void nfsd_idmap_shutdown(void);
 #else
-static inline void nfsd_idmap_init(void) {};
+static inline int nfsd_idmap_init(void)
-static inline void nfsd_idmap_shutdown(void) {};
+{
+	return 0;
+}
+static inline void nfsd_idmap_shutdown(void)
+{
+}
 #endif
 
 int nfsd_map_name_to_uid(struct svc_rqst *, const char *, size_t, __u32 *);

include/linux/sunrpc/cache.h

@@ -169,8 +169,8 @@ extern int cache_check(struct cache_detail *detail,
 extern void cache_flush(void);
 extern void cache_purge(struct cache_detail *detail);
 #define NEVER (0x7FFFFFFF)
-extern void cache_register(struct cache_detail *cd);
+extern int cache_register(struct cache_detail *cd);
-extern int cache_unregister(struct cache_detail *cd);
+extern void cache_unregister(struct cache_detail *cd);
 
 extern void qword_add(char **bpp, int *lp, char *str);
 extern void qword_addhex(char **bpp, int *lp, char *buf, int blen);

include/linux/sunrpc/debug.h

@@ -20,7 +20,7 @@
 #define RPCDBG_BIND		0x0020
 #define RPCDBG_SCHED		0x0040
 #define RPCDBG_TRANS		0x0080
-#define RPCDBG_SVCSOCK		0x0100
+#define RPCDBG_SVCXPRT		0x0100
 #define RPCDBG_SVCDSP		0x0200
 #define RPCDBG_MISC		0x0400
 #define RPCDBG_CACHE		0x0800

include/linux/sunrpc/svc.h

@@ -204,7 +204,7 @@ union svc_addr_u {
 struct svc_rqst {
 	struct list_head	rq_list;	/* idle list */
 	struct list_head	rq_all;		/* all threads list */
-	struct svc_sock *	rq_sock;	/* socket */
+	struct svc_xprt *	rq_xprt;	/* transport ptr */
 	struct sockaddr_storage	rq_addr;	/* peer address */
 	size_t			rq_addrlen;
 
@@ -214,9 +214,10 @@ struct svc_rqst {
 	struct auth_ops *	rq_authop;	/* authentication flavour */
 	u32			rq_flavor;	/* pseudoflavor */
 	struct svc_cred		rq_cred;	/* auth info */
-	struct sk_buff *	rq_skbuff;	/* fast recv inet buffer */
+	void *			rq_xprt_ctxt;	/* transport specific context ptr */
 	struct svc_deferred_req*rq_deferred;	/* deferred request we are replaying */
 
+	size_t			rq_xprt_hlen;	/* xprt header len */
 	struct xdr_buf		rq_arg;
 	struct xdr_buf		rq_res;
 	struct page *		rq_pages[RPCSVC_MAXPAGES];
@@ -317,11 +318,12 @@ static inline void svc_free_res_pages(struct svc_rqst *rqstp)
 
 struct svc_deferred_req {
 	u32			prot;	/* protocol (UDP or TCP) */
-	struct svc_sock		*svsk;
+	struct svc_xprt		*xprt;
 	struct sockaddr_storage	addr;	/* where reply must go */
 	size_t			addrlen;
 	union svc_addr_u	daddr;	/* where reply must come from */
 	struct cache_deferred_req handle;
+	size_t			xprt_hlen;
 	int			argslen;
 	__be32			args[0];
 };
@@ -382,6 +384,8 @@ struct svc_procedure {
  */
 struct svc_serv *  svc_create(struct svc_program *, unsigned int,
 			      void (*shutdown)(struct svc_serv*));
+struct svc_rqst *svc_prepare_thread(struct svc_serv *serv,
+					struct svc_pool *pool);
 int		   svc_create_thread(svc_thread_fn, struct svc_serv *);
 void		   svc_exit_thread(struct svc_rqst *);
 struct svc_serv *  svc_create_pooled(struct svc_program *, unsigned int,

include/linux/sunrpc/svc_rdma.h

@@ -0,0 +1,262 @@
+/*
+ * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the BSD-type
+ * license below:
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *      Redistributions of source code must retain the above copyright
+ *      notice, this list of conditions and the following disclaimer.
+ *
+ *      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.
+ *
+ *      Neither the name of the Network Appliance, Inc. 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 BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "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 COPYRIGHT
+ * OWNER 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.
+ *
+ * Author: Tom Tucker <tom@opengridcomputing.com>
+ */
+
+#ifndef SVC_RDMA_H
+#define SVC_RDMA_H
+#include <linux/sunrpc/xdr.h>
+#include <linux/sunrpc/svcsock.h>
+#include <linux/sunrpc/rpc_rdma.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/rdma_cm.h>
+#define SVCRDMA_DEBUG
+
+/* RPC/RDMA parameters and stats */
+extern unsigned int svcrdma_ord;
+extern unsigned int svcrdma_max_requests;
+extern unsigned int svcrdma_max_req_size;
+
+extern atomic_t rdma_stat_recv;
+extern atomic_t rdma_stat_read;
+extern atomic_t rdma_stat_write;
+extern atomic_t rdma_stat_sq_starve;
+extern atomic_t rdma_stat_rq_starve;
+extern atomic_t rdma_stat_rq_poll;
+extern atomic_t rdma_stat_rq_prod;
+extern atomic_t rdma_stat_sq_poll;
+extern atomic_t rdma_stat_sq_prod;
+
+#define RPCRDMA_VERSION 1
+
+/*
+ * Contexts are built when an RDMA request is created and are a
+ * record of the resources that can be recovered when the request
+ * completes.
+ */
+struct svc_rdma_op_ctxt {
+	struct svc_rdma_op_ctxt *next;
+	struct xdr_buf arg;
+	struct list_head dto_q;
+	enum ib_wr_opcode wr_op;
+	enum ib_wc_status wc_status;
+	u32 byte_len;
+	struct svcxprt_rdma *xprt;
+	unsigned long flags;
+	enum dma_data_direction direction;
+	int count;
+	struct ib_sge sge[RPCSVC_MAXPAGES];
+	struct page *pages[RPCSVC_MAXPAGES];
+};
+
+#define RDMACTXT_F_READ_DONE	1
+#define RDMACTXT_F_LAST_CTXT	2
+
+struct svcxprt_rdma {
+	struct svc_xprt      sc_xprt;		/* SVC transport structure */
+	struct rdma_cm_id    *sc_cm_id;		/* RDMA connection id */
+	struct list_head     sc_accept_q;	/* Conn. waiting accept */
+	int		     sc_ord;		/* RDMA read limit */
+	wait_queue_head_t    sc_read_wait;
+	int                  sc_max_sge;
+
+	int                  sc_sq_depth;	/* Depth of SQ */
+	atomic_t             sc_sq_count;	/* Number of SQ WR on queue */
+
+	int                  sc_max_requests;	/* Depth of RQ */
+	int                  sc_max_req_size;	/* Size of each RQ WR buf */
+
+	struct ib_pd         *sc_pd;
+
+	struct svc_rdma_op_ctxt  *sc_ctxt_head;
+	int		     sc_ctxt_cnt;
+	int		     sc_ctxt_bump;
+	int		     sc_ctxt_max;
+	spinlock_t	     sc_ctxt_lock;
+	struct list_head     sc_rq_dto_q;
+	spinlock_t	     sc_rq_dto_lock;
+	struct ib_qp         *sc_qp;
+	struct ib_cq         *sc_rq_cq;
+	struct ib_cq         *sc_sq_cq;
+	struct ib_mr         *sc_phys_mr;	/* MR for server memory */
+
+	spinlock_t	     sc_lock;		/* transport lock */
+
+	wait_queue_head_t    sc_send_wait;	/* SQ exhaustion waitlist */
+	unsigned long	     sc_flags;
+	struct list_head     sc_dto_q;		/* DTO tasklet I/O pending Q */
+	struct list_head     sc_read_complete_q;
+	spinlock_t           sc_read_complete_lock;
+};
+/* sc_flags */
+#define RDMAXPRT_RQ_PENDING	1
+#define RDMAXPRT_SQ_PENDING	2
+#define RDMAXPRT_CONN_PENDING	3
+
+#define RPCRDMA_LISTEN_BACKLOG  10
+/* The default ORD value is based on two outstanding full-size writes with a
+ * page size of 4k, or 32k * 2 ops / 4k = 16 outstanding RDMA_READ.  */
+#define RPCRDMA_ORD             (64/4)
+#define RPCRDMA_SQ_DEPTH_MULT   8
+#define RPCRDMA_MAX_THREADS     16
+#define RPCRDMA_MAX_REQUESTS    16
+#define RPCRDMA_MAX_REQ_SIZE    4096
+
+/* svc_rdma_marshal.c */
+extern void svc_rdma_rcl_chunk_counts(struct rpcrdma_read_chunk *,
+				      int *, int *);
+extern int svc_rdma_xdr_decode_req(struct rpcrdma_msg **, struct svc_rqst *);
+extern int svc_rdma_xdr_decode_deferred_req(struct svc_rqst *);
+extern int svc_rdma_xdr_encode_error(struct svcxprt_rdma *,
+				     struct rpcrdma_msg *,
+				     enum rpcrdma_errcode, u32 *);
+extern void svc_rdma_xdr_encode_write_list(struct rpcrdma_msg *, int);
+extern void svc_rdma_xdr_encode_reply_array(struct rpcrdma_write_array *, int);
+extern void svc_rdma_xdr_encode_array_chunk(struct rpcrdma_write_array *, int,
+					    u32, u64, u32);
+extern void svc_rdma_xdr_encode_reply_header(struct svcxprt_rdma *,
+					     struct rpcrdma_msg *,
+					     struct rpcrdma_msg *,
+					     enum rpcrdma_proc);
+extern int svc_rdma_xdr_get_reply_hdr_len(struct rpcrdma_msg *);
+
+/* svc_rdma_recvfrom.c */
+extern int svc_rdma_recvfrom(struct svc_rqst *);
+
+/* svc_rdma_sendto.c */
+extern int svc_rdma_sendto(struct svc_rqst *);
+
+/* svc_rdma_transport.c */
+extern int svc_rdma_send(struct svcxprt_rdma *, struct ib_send_wr *);
+extern int svc_rdma_send_error(struct svcxprt_rdma *, struct rpcrdma_msg *,
+			       enum rpcrdma_errcode);
+struct page *svc_rdma_get_page(void);
+extern int svc_rdma_post_recv(struct svcxprt_rdma *);
+extern int svc_rdma_create_listen(struct svc_serv *, int, struct sockaddr *);
+extern struct svc_rdma_op_ctxt *svc_rdma_get_context(struct svcxprt_rdma *);
+extern void svc_rdma_put_context(struct svc_rdma_op_ctxt *, int);
+extern void svc_sq_reap(struct svcxprt_rdma *);
+extern void svc_rq_reap(struct svcxprt_rdma *);
+extern struct svc_xprt_class svc_rdma_class;
+extern void svc_rdma_prep_reply_hdr(struct svc_rqst *);
+
+/* svc_rdma.c */
+extern int svc_rdma_init(void);
+extern void svc_rdma_cleanup(void);
+
+/*
+ * Returns the address of the first read chunk or <nul> if no read chunk is
+ * present
+ */
+static inline struct rpcrdma_read_chunk *
+svc_rdma_get_read_chunk(struct rpcrdma_msg *rmsgp)
+{
+	struct rpcrdma_read_chunk *ch =
+		(struct rpcrdma_read_chunk *)&rmsgp->rm_body.rm_chunks[0];
+
+	if (ch->rc_discrim == 0)
+		return NULL;
+
+	return ch;
+}
+
+/*
+ * Returns the address of the first read write array element or <nul> if no
+ * write array list is present
+ */
+static inline struct rpcrdma_write_array *
+svc_rdma_get_write_array(struct rpcrdma_msg *rmsgp)
+{
+	if (rmsgp->rm_body.rm_chunks[0] != 0
+	    || rmsgp->rm_body.rm_chunks[1] == 0)
+		return NULL;
+
+	return (struct rpcrdma_write_array *)&rmsgp->rm_body.rm_chunks[1];
+}
+
+/*
+ * Returns the address of the first reply array element or <nul> if no
+ * reply array is present
+ */
+static inline struct rpcrdma_write_array *
+svc_rdma_get_reply_array(struct rpcrdma_msg *rmsgp)
+{
+	struct rpcrdma_read_chunk *rch;
+	struct rpcrdma_write_array *wr_ary;
+	struct rpcrdma_write_array *rp_ary;
+
+	/* XXX: Need to fix when reply list may occur with read-list and/or
+	 * write list */
+	if (rmsgp->rm_body.rm_chunks[0] != 0 ||
+	    rmsgp->rm_body.rm_chunks[1] != 0)
+		return NULL;
+
+	rch = svc_rdma_get_read_chunk(rmsgp);
+	if (rch) {
+		while (rch->rc_discrim)
+			rch++;
+
+		/* The reply list follows an empty write array located
+		 * at 'rc_position' here. The reply array is at rc_target.
+		 */
+		rp_ary = (struct rpcrdma_write_array *)&rch->rc_target;
+
+		goto found_it;
+	}
+
+	wr_ary = svc_rdma_get_write_array(rmsgp);
+	if (wr_ary) {
+		rp_ary = (struct rpcrdma_write_array *)
+			&wr_ary->
+			wc_array[wr_ary->wc_nchunks].wc_target.rs_length;
+
+		goto found_it;
+	}
+
+	/* No read list, no write list */
+	rp_ary = (struct rpcrdma_write_array *)
+		&rmsgp->rm_body.rm_chunks[2];
+
+ found_it:
+	if (rp_ary->wc_discrim == 0)
+		return NULL;
+
+	return rp_ary;
+}
+#endif

include/linux/sunrpc/svc_xprt.h

@@ -0,0 +1,159 @@
+/*
+ * linux/include/linux/sunrpc/svc_xprt.h
+ *
+ * RPC server transport I/O
+ */
+
+#ifndef SUNRPC_SVC_XPRT_H
+#define SUNRPC_SVC_XPRT_H
+
+#include <linux/sunrpc/svc.h>
+#include <linux/module.h>
+
+struct svc_xprt_ops {
+	struct svc_xprt	*(*xpo_create)(struct svc_serv *,
+				       struct sockaddr *, int,
+				       int);
+	struct svc_xprt	*(*xpo_accept)(struct svc_xprt *);
+	int		(*xpo_has_wspace)(struct svc_xprt *);
+	int		(*xpo_recvfrom)(struct svc_rqst *);
+	void		(*xpo_prep_reply_hdr)(struct svc_rqst *);
+	int		(*xpo_sendto)(struct svc_rqst *);
+	void		(*xpo_release_rqst)(struct svc_rqst *);
+	void		(*xpo_detach)(struct svc_xprt *);
+	void		(*xpo_free)(struct svc_xprt *);
+};
+
+struct svc_xprt_class {
+	const char		*xcl_name;
+	struct module		*xcl_owner;
+	struct svc_xprt_ops	*xcl_ops;
+	struct list_head	xcl_list;
+	u32			xcl_max_payload;
+};
+
+struct svc_xprt {
+	struct svc_xprt_class	*xpt_class;
+	struct svc_xprt_ops	*xpt_ops;
+	struct kref		xpt_ref;
+	struct list_head	xpt_list;
+	struct list_head	xpt_ready;
+	unsigned long		xpt_flags;
+#define	XPT_BUSY	0		/* enqueued/receiving */
+#define	XPT_CONN	1		/* conn pending */
+#define	XPT_CLOSE	2		/* dead or dying */
+#define	XPT_DATA	3		/* data pending */
+#define	XPT_TEMP	4		/* connected transport */
+#define	XPT_DEAD	6		/* transport closed */
+#define	XPT_CHNGBUF	7		/* need to change snd/rcv buf sizes */
+#define	XPT_DEFERRED	8		/* deferred request pending */
+#define	XPT_OLD		9		/* used for xprt aging mark+sweep */
+#define	XPT_DETACHED	10		/* detached from tempsocks list */
+#define XPT_LISTENER	11		/* listening endpoint */
+#define XPT_CACHE_AUTH	12		/* cache auth info */
+
+	struct svc_pool		*xpt_pool;	/* current pool iff queued */
+	struct svc_serv		*xpt_server;	/* service for transport */
+	atomic_t    	    	xpt_reserved;	/* space on outq that is rsvd */
+	struct mutex		xpt_mutex;	/* to serialize sending data */
+	spinlock_t		xpt_lock;	/* protects sk_deferred
+						 * and xpt_auth_cache */
+	void			*xpt_auth_cache;/* auth cache */
+	struct list_head	xpt_deferred;	/* deferred requests that need
+						 * to be revisted */
+	struct sockaddr_storage	xpt_local;	/* local address */
+	size_t			xpt_locallen;	/* length of address */
+	struct sockaddr_storage	xpt_remote;	/* remote peer's address */
+	size_t			xpt_remotelen;	/* length of address */
+};
+
+int	svc_reg_xprt_class(struct svc_xprt_class *);
+void	svc_unreg_xprt_class(struct svc_xprt_class *);
+void	svc_xprt_init(struct svc_xprt_class *, struct svc_xprt *,
+		      struct svc_serv *);
+int	svc_create_xprt(struct svc_serv *, char *, unsigned short, int);
+void	svc_xprt_enqueue(struct svc_xprt *xprt);
+void	svc_xprt_received(struct svc_xprt *);
+void	svc_xprt_put(struct svc_xprt *xprt);
+void	svc_xprt_copy_addrs(struct svc_rqst *rqstp, struct svc_xprt *xprt);
+void	svc_close_xprt(struct svc_xprt *xprt);
+void	svc_delete_xprt(struct svc_xprt *xprt);
+int	svc_port_is_privileged(struct sockaddr *sin);
+int	svc_print_xprts(char *buf, int maxlen);
+struct	svc_xprt *svc_find_xprt(struct svc_serv *, char *, int, int);
+int	svc_xprt_names(struct svc_serv *serv, char *buf, int buflen);
+
+static inline void svc_xprt_get(struct svc_xprt *xprt)
+{
+	kref_get(&xprt->xpt_ref);
+}
+static inline void svc_xprt_set_local(struct svc_xprt *xprt,
+				      struct sockaddr *sa, int salen)
+{
+	memcpy(&xprt->xpt_local, sa, salen);
+	xprt->xpt_locallen = salen;
+}
+static inline void svc_xprt_set_remote(struct svc_xprt *xprt,
+				       struct sockaddr *sa, int salen)
+{
+	memcpy(&xprt->xpt_remote, sa, salen);
+	xprt->xpt_remotelen = salen;
+}
+static inline unsigned short svc_addr_port(struct sockaddr *sa)
+{
+	unsigned short ret = 0;
+	switch (sa->sa_family) {
+	case AF_INET:
+		ret = ntohs(((struct sockaddr_in *)sa)->sin_port);
+		break;
+	case AF_INET6:
+		ret = ntohs(((struct sockaddr_in6 *)sa)->sin6_port);
+		break;
+	}
+	return ret;
+}
+
+static inline size_t svc_addr_len(struct sockaddr *sa)
+{
+	switch (sa->sa_family) {
+	case AF_INET:
+		return sizeof(struct sockaddr_in);
+	case AF_INET6:
+		return sizeof(struct sockaddr_in6);
+	}
+	return -EAFNOSUPPORT;
+}
+
+static inline unsigned short svc_xprt_local_port(struct svc_xprt *xprt)
+{
+	return svc_addr_port((struct sockaddr *)&xprt->xpt_local);
+}
+
+static inline unsigned short svc_xprt_remote_port(struct svc_xprt *xprt)
+{
+	return svc_addr_port((struct sockaddr *)&xprt->xpt_remote);
+}
+
+static inline char *__svc_print_addr(struct sockaddr *addr,
+				     char *buf, size_t len)
+{
+	switch (addr->sa_family) {
+	case AF_INET:
+		snprintf(buf, len, "%u.%u.%u.%u, port=%u",
+			NIPQUAD(((struct sockaddr_in *) addr)->sin_addr),
+			ntohs(((struct sockaddr_in *) addr)->sin_port));
+		break;
+
+	case AF_INET6:
+		snprintf(buf, len, "%x:%x:%x:%x:%x:%x:%x:%x, port=%u",
+			NIP6(((struct sockaddr_in6 *) addr)->sin6_addr),
+			ntohs(((struct sockaddr_in6 *) addr)->sin6_port));
+		break;
+
+	default:
+		snprintf(buf, len, "unknown address type: %d", addr->sa_family);
+		break;
+	}
+	return buf;
+}
+#endif /* SUNRPC_SVC_XPRT_H */

include/linux/sunrpc/svcsock.h

@@ -10,42 +10,16 @@
 #define SUNRPC_SVCSOCK_H
 
 #include <linux/sunrpc/svc.h>
+#include <linux/sunrpc/svc_xprt.h>
 
 /*
  * RPC server socket.
  */
 struct svc_sock {
-	struct list_head	sk_ready;	/* list of ready sockets */
+	struct svc_xprt		sk_xprt;
-	struct list_head	sk_list;	/* list of all sockets */
 	struct socket *		sk_sock;	/* berkeley socket layer */
 	struct sock *		sk_sk;		/* INET layer */
 
-	struct svc_pool *	sk_pool;	/* current pool iff queued */
-	struct svc_serv *	sk_server;	/* service for this socket */
-	atomic_t		sk_inuse;	/* use count */
-	unsigned long		sk_flags;
-#define	SK_BUSY		0			/* enqueued/receiving */
-#define	SK_CONN		1			/* conn pending */
-#define	SK_CLOSE	2			/* dead or dying */
-#define	SK_DATA		3			/* data pending */
-#define	SK_TEMP		4			/* temp (TCP) socket */
-#define	SK_DEAD		6			/* socket closed */
-#define	SK_CHNGBUF	7			/* need to change snd/rcv buffer sizes */
-#define	SK_DEFERRED	8			/* request on sk_deferred */
-#define	SK_OLD		9			/* used for temp socket aging mark+sweep */
-#define	SK_DETACHED	10			/* detached from tempsocks list */
-
-	atomic_t    	    	sk_reserved;	/* space on outq that is reserved */
-
-	spinlock_t		sk_lock;	/* protects sk_deferred and
-						 * sk_info_authunix */
-	struct list_head	sk_deferred;	/* deferred requests that need to
-						 * be revisted */
-	struct mutex		sk_mutex;	/* to serialize sending data */
-
-	int			(*sk_recvfrom)(struct svc_rqst *rqstp);
-	int			(*sk_sendto)(struct svc_rqst *rqstp);
-
 	/* We keep the old state_change and data_ready CB's here */
 	void			(*sk_ostate)(struct sock *);
 	void			(*sk_odata)(struct sock *, int bytes);
@@ -54,21 +28,12 @@ struct svc_sock {
 	/* private TCP part */
 	int			sk_reclen;	/* length of record */
 	int			sk_tcplen;	/* current read length */
-	time_t			sk_lastrecv;	/* time of last received request */
-
-	/* cache of various info for TCP sockets */
-	void			*sk_info_authunix;
-
-	struct sockaddr_storage	sk_local;	/* local address */
-	struct sockaddr_storage	sk_remote;	/* remote peer's address */
-	int			sk_remotelen;	/* length of address */
 };
 
 /*
  * Function prototypes.
  */
-int		svc_makesock(struct svc_serv *, int, unsigned short, int flags);
+void		svc_close_all(struct list_head *);
-void		svc_force_close_socket(struct svc_sock *);
 int		svc_recv(struct svc_rqst *, long);
 int		svc_send(struct svc_rqst *);
 void		svc_drop(struct svc_rqst *);
@@ -78,6 +43,8 @@ int		svc_addsock(struct svc_serv *serv,
 			    int fd,
 			    char *name_return,
 			    int *proto);
+void		svc_init_xprt_sock(void);
+void		svc_cleanup_xprt_sock(void);
 
 /*
  * svc_makesock socket characteristics

include/linux/sunrpc/xdr.h

@@ -112,7 +112,8 @@ struct xdr_buf {
 __be32 *xdr_encode_opaque_fixed(__be32 *p, const void *ptr, unsigned int len);
 __be32 *xdr_encode_opaque(__be32 *p, const void *ptr, unsigned int len);
 __be32 *xdr_encode_string(__be32 *p, const char *s);
-__be32 *xdr_decode_string_inplace(__be32 *p, char **sp, int *lenp, int maxlen);
+__be32 *xdr_decode_string_inplace(__be32 *p, char **sp, unsigned int *lenp,
+			unsigned int maxlen);
 __be32 *xdr_encode_netobj(__be32 *p, const struct xdr_netobj *);
 __be32 *xdr_decode_netobj(__be32 *p, struct xdr_netobj *);
 

net/sunrpc/Makefile

@@ -11,6 +11,7 @@ sunrpc-y := clnt.o xprt.o socklib.o xprtsock.o sched.o \
 	    auth.o auth_null.o auth_unix.o \
 	    svc.o svcsock.o svcauth.o svcauth_unix.o \
 	    rpcb_clnt.o timer.o xdr.o \
-	    sunrpc_syms.o cache.o rpc_pipe.o
+	    sunrpc_syms.o cache.o rpc_pipe.o \
+	    svc_xprt.o
 sunrpc-$(CONFIG_PROC_FS) += stats.o
 sunrpc-$(CONFIG_SYSCTL) += sysctl.o

net/sunrpc/auth_gss/svcauth_gss.c

@@ -224,11 +224,8 @@ static int rsi_parse(struct cache_detail *cd,
 
 	/* major/minor */
 	len = qword_get(&mesg, buf, mlen);
-	if (len < 0)
+	if (len <= 0)
 		goto out;
-	if (len == 0) {
-		goto out;
-	} else {
 	rsii.major_status = simple_strtoul(buf, &ep, 10);
 	if (*ep)
 		goto out;
@@ -255,7 +252,6 @@ static int rsi_parse(struct cache_detail *cd,
 	status = -ENOMEM;
 	if (dup_to_netobj(&rsii.out_token, buf, len))
 		goto out;
-	}
 	rsii.h.expiry_time = expiry;
 	rsip = rsi_update(&rsii, rsip);
 	status = 0;
@@ -975,6 +971,7 @@ static int svcauth_gss_handle_init(struct svc_rqst *rqstp,
 	struct kvec *resv = &rqstp->rq_res.head[0];
 	struct xdr_netobj tmpobj;
 	struct rsi *rsip, rsikey;
+	int ret;
 
 	/* Read the verifier; should be NULL: */
 	*authp = rpc_autherr_badverf;
@@ -1014,23 +1011,27 @@ static int svcauth_gss_handle_init(struct svc_rqst *rqstp,
 		/* No upcall result: */
 		return SVC_DROP;
 	case 0:
+		ret = SVC_DROP;
 		/* Got an answer to the upcall; use it: */
 		if (gss_write_init_verf(rqstp, rsip))
-			return SVC_DROP;
+			goto out;
 		if (resv->iov_len + 4 > PAGE_SIZE)
-			return SVC_DROP;
+			goto out;
 		svc_putnl(resv, RPC_SUCCESS);
 		if (svc_safe_putnetobj(resv, &rsip->out_handle))
-			return SVC_DROP;
+			goto out;
 		if (resv->iov_len + 3 * 4 > PAGE_SIZE)
-			return SVC_DROP;
+			goto out;
 		svc_putnl(resv, rsip->major_status);
 		svc_putnl(resv, rsip->minor_status);
 		svc_putnl(resv, GSS_SEQ_WIN);
 		if (svc_safe_putnetobj(resv, &rsip->out_token))
-			return SVC_DROP;
+			goto out;
 	}
-	return SVC_COMPLETE;
+	ret = SVC_COMPLETE;
+out:
+	cache_put(&rsip->h, &rsi_cache);
+	return ret;
 }
 
 /*
@@ -1125,6 +1126,7 @@ svcauth_gss_accept(struct svc_rqst *rqstp, __be32 *authp)
 	case RPC_GSS_PROC_DESTROY:
 		if (gss_write_verf(rqstp, rsci->mechctx, gc->gc_seq))
 			goto auth_err;
+		rsci->h.expiry_time = get_seconds();
 		set_bit(CACHE_NEGATIVE, &rsci->h.flags);
 		if (resv->iov_len + 4 > PAGE_SIZE)
 			goto drop;
@@ -1386,19 +1388,26 @@ int
 gss_svc_init(void)
 {
 	int rv = svc_auth_register(RPC_AUTH_GSS, &svcauthops_gss);
-	if (rv == 0) {
+	if (rv)
-		cache_register(&rsc_cache);
+		return rv;
-		cache_register(&rsi_cache);
+	rv = cache_register(&rsc_cache);
-	}
+	if (rv)
+		goto out1;
+	rv = cache_register(&rsi_cache);
+	if (rv)
+		goto out2;
+	return 0;
+out2:
+	cache_unregister(&rsc_cache);
+out1:
+	svc_auth_unregister(RPC_AUTH_GSS);
 	return rv;
 }
 
 void
 gss_svc_shutdown(void)
 {
-	if (cache_unregister(&rsc_cache))
+	cache_unregister(&rsc_cache);
-		printk(KERN_ERR "auth_rpcgss: failed to unregister rsc cache\n");
+	cache_unregister(&rsi_cache);
-	if (cache_unregister(&rsi_cache))
-		printk(KERN_ERR "auth_rpcgss: failed to unregister rsi cache\n");
 	svc_auth_unregister(RPC_AUTH_GSS);
 }

net/sunrpc/cache.c

@@ -245,6 +245,7 @@ int cache_check(struct cache_detail *detail,
 		cache_put(h, detail);
 	return rv;
 }
+EXPORT_SYMBOL(cache_check);
 
 /*
  * caches need to be periodically cleaned.
@@ -290,44 +291,78 @@ static const struct file_operations cache_flush_operations;
 static void do_cache_clean(struct work_struct *work);
 static DECLARE_DELAYED_WORK(cache_cleaner, do_cache_clean);
 
-void cache_register(struct cache_detail *cd)
+static void remove_cache_proc_entries(struct cache_detail *cd)
+{
+	if (cd->proc_ent == NULL)
+		return;
+	if (cd->flush_ent)
+		remove_proc_entry("flush", cd->proc_ent);
+	if (cd->channel_ent)
+		remove_proc_entry("channel", cd->proc_ent);
+	if (cd->content_ent)
+		remove_proc_entry("content", cd->proc_ent);
+	cd->proc_ent = NULL;
+	remove_proc_entry(cd->name, proc_net_rpc);
+}
+
+#ifdef CONFIG_PROC_FS
+static int create_cache_proc_entries(struct cache_detail *cd)
 {
-	cd->proc_ent = proc_mkdir(cd->name, proc_net_rpc);
-	if (cd->proc_ent) {
 	struct proc_dir_entry *p;
+
+	cd->proc_ent = proc_mkdir(cd->name, proc_net_rpc);
+	if (cd->proc_ent == NULL)
+		goto out_nomem;
 	cd->proc_ent->owner = cd->owner;
 	cd->channel_ent = cd->content_ent = NULL;
 
-		p = create_proc_entry("flush", S_IFREG|S_IRUSR|S_IWUSR,
+	p = create_proc_entry("flush", S_IFREG|S_IRUSR|S_IWUSR, cd->proc_ent);
-				      cd->proc_ent);
 	cd->flush_ent = p;
-		if (p) {
+	if (p == NULL)
+		goto out_nomem;
 	p->proc_fops = &cache_flush_operations;
 	p->owner = cd->owner;
 	p->data = cd;
-		}
 
 	if (cd->cache_request || cd->cache_parse) {
 		p = create_proc_entry("channel", S_IFREG|S_IRUSR|S_IWUSR,
 				      cd->proc_ent);
 		cd->channel_ent = p;
-			if (p) {
+		if (p == NULL)
+			goto out_nomem;
 		p->proc_fops = &cache_file_operations;
 		p->owner = cd->owner;
 		p->data = cd;
 	}
-		}
 	if (cd->cache_show) {
 		p = create_proc_entry("content", S_IFREG|S_IRUSR|S_IWUSR,
 				      cd->proc_ent);
 		cd->content_ent = p;
-			if (p) {
+		if (p == NULL)
+			goto out_nomem;
 		p->proc_fops = &content_file_operations;
 		p->owner = cd->owner;
 		p->data = cd;
 	}
-		}
+	return 0;
-	}
+out_nomem:
+	remove_cache_proc_entries(cd);
+	return -ENOMEM;
+}
+#else /* CONFIG_PROC_FS */
+static int create_cache_proc_entries(struct cache_detail *cd)
+{
+	return 0;
+}
+#endif
+
+int cache_register(struct cache_detail *cd)
+{
+	int ret;
+
+	ret = create_cache_proc_entries(cd);
+	if (ret)
+		return ret;
 	rwlock_init(&cd->hash_lock);
 	INIT_LIST_HEAD(&cd->queue);
 	spin_lock(&cache_list_lock);
@@ -341,9 +376,11 @@ void cache_register(struct cache_detail *cd)
 
 	/* start the cleaning process */
 	schedule_delayed_work(&cache_cleaner, 0);
+	return 0;
 }
+EXPORT_SYMBOL(cache_register);
 
-int cache_unregister(struct cache_detail *cd)
+void cache_unregister(struct cache_detail *cd)
 {
 	cache_purge(cd);
 	spin_lock(&cache_list_lock);
@@ -351,30 +388,23 @@ int cache_unregister(struct cache_detail *cd)
 	if (cd->entries || atomic_read(&cd->inuse)) {
 		write_unlock(&cd->hash_lock);
 		spin_unlock(&cache_list_lock);
-		return -EBUSY;
+		goto out;
 	}
 	if (current_detail == cd)
 		current_detail = NULL;
 	list_del_init(&cd->others);
 	write_unlock(&cd->hash_lock);
 	spin_unlock(&cache_list_lock);
-	if (cd->proc_ent) {
+	remove_cache_proc_entries(cd);
-		if (cd->flush_ent)
-			remove_proc_entry("flush", cd->proc_ent);
-		if (cd->channel_ent)
-			remove_proc_entry("channel", cd->proc_ent);
-		if (cd->content_ent)
-			remove_proc_entry("content", cd->proc_ent);
-
-		cd->proc_ent = NULL;
-		remove_proc_entry(cd->name, proc_net_rpc);
-	}
 	if (list_empty(&cache_list)) {
 		/* module must be being unloaded so its safe to kill the worker */
 		cancel_delayed_work_sync(&cache_cleaner);
 	}
-	return 0;
+	return;
+out:
+	printk(KERN_ERR "nfsd: failed to unregister %s cache\n", cd->name);
 }
+EXPORT_SYMBOL(cache_unregister);
 
 /* clean cache tries to find something to clean
  * and cleans it.
@@ -489,6 +519,7 @@ void cache_flush(void)
 	while (cache_clean() != -1)
 		cond_resched();
 }
+EXPORT_SYMBOL(cache_flush);
 
 void cache_purge(struct cache_detail *detail)
 {
@@ -497,7 +528,7 @@ void cache_purge(struct cache_detail *detail)
 	cache_flush();
 	detail->flush_time = 1;
 }
-
+EXPORT_SYMBOL(cache_purge);
 
 
 /*
@@ -634,13 +665,13 @@ void cache_clean_deferred(void *owner)
 /*
  * communicate with user-space
  *
- * We have a magic /proc file - /proc/sunrpc/cache
+ * We have a magic /proc file - /proc/sunrpc/<cachename>/channel.
- * On read, you get a full request, or block
+ * On read, you get a full request, or block.
- * On write, an update request is processed
+ * On write, an update request is processed.
- * Poll works if anything to read, and always allows write
+ * Poll works if anything to read, and always allows write.
  *
  * Implemented by linked list of requests.  Each open file has
- * a ->private that also exists in this list.  New request are added
+ * a ->private that also exists in this list.  New requests are added
  * to the end and may wakeup and preceding readers.
  * New readers are added to the head.  If, on read, an item is found with
  * CACHE_UPCALLING clear, we free it from the list.
@@ -963,6 +994,7 @@ void qword_add(char **bpp, int *lp, char *str)
 	*bpp = bp;
 	*lp = len;
 }
+EXPORT_SYMBOL(qword_add);
 
 void qword_addhex(char **bpp, int *lp, char *buf, int blen)
 {
@@ -991,6 +1023,7 @@ void qword_addhex(char **bpp, int *lp, char *buf, int blen)
 	*bpp = bp;
 	*lp = len;
 }
+EXPORT_SYMBOL(qword_addhex);
 
 static void warn_no_listener(struct cache_detail *detail)
 {
@@ -1113,6 +1146,7 @@ int qword_get(char **bpp, char *dest, int bufsize)
 	*dest = '\0';
 	return len;
 }
+EXPORT_SYMBOL(qword_get);
 
 
 /*
@@ -1244,17 +1278,17 @@ static ssize_t read_flush(struct file *file, char __user *buf,
 	struct cache_detail *cd = PDE(file->f_path.dentry->d_inode)->data;
 	char tbuf[20];
 	unsigned long p = *ppos;
-	int len;
+	size_t len;
 
 	sprintf(tbuf, "%lu\n", cd->flush_time);
 	len = strlen(tbuf);
 	if (p >= len)
 		return 0;
 	len -= p;
-	if (len > count) len = count;
+	if (len > count)
+		len = count;
 	if (copy_to_user(buf, (void*)(tbuf+p), len))
-		len = -EFAULT;
+		return -EFAULT;
-	else
 	*ppos += len;
 	return len;
 }

net/sunrpc/stats.c

@@ -33,7 +33,7 @@ struct proc_dir_entry	*proc_net_rpc = NULL;
 static int rpc_proc_show(struct seq_file *seq, void *v) {
 	const struct rpc_stat	*statp = seq->private;
 	const struct rpc_program *prog = statp->program;
-	int		i, j;
+	unsigned int i, j;
 
 	seq_printf(seq,
 		"net %u %u %u %u\n",
@@ -81,7 +81,7 @@ void svc_seq_show(struct seq_file *seq, const struct svc_stat *statp) {
 	const struct svc_program *prog = statp->program;
 	const struct svc_procedure *proc;
 	const struct svc_version *vers;
-	int		i, j;
+	unsigned int i, j;
 
 	seq_printf(seq,
 		"net %u %u %u %u\n",
@@ -106,6 +106,7 @@ void svc_seq_show(struct seq_file *seq, const struct svc_stat *statp) {
 		seq_putc(seq, '\n');
 	}
 }
+EXPORT_SYMBOL(svc_seq_show);
 
 /**
  * rpc_alloc_iostats - allocate an rpc_iostats structure
@@ -255,12 +256,14 @@ svc_proc_register(struct svc_stat *statp, const struct file_operations *fops)
 {
 	return do_register(statp->program->pg_name, statp, fops);
 }
+EXPORT_SYMBOL(svc_proc_register);
 
 void
 svc_proc_unregister(const char *name)
 {
 	remove_proc_entry(name, proc_net_rpc);
 }
+EXPORT_SYMBOL(svc_proc_unregister);
 
 void
 rpc_proc_init(void)

net/sunrpc/sunrpc_syms.c

@@ -22,48 +22,6 @@
 #include <linux/sunrpc/rpc_pipe_fs.h>
 #include <linux/sunrpc/xprtsock.h>
 
-/* RPC server stuff */
-EXPORT_SYMBOL(svc_create);
-EXPORT_SYMBOL(svc_create_thread);
-EXPORT_SYMBOL(svc_create_pooled);
-EXPORT_SYMBOL(svc_set_num_threads);
-EXPORT_SYMBOL(svc_exit_thread);
-EXPORT_SYMBOL(svc_destroy);
-EXPORT_SYMBOL(svc_drop);
-EXPORT_SYMBOL(svc_process);
-EXPORT_SYMBOL(svc_recv);
-EXPORT_SYMBOL(svc_wake_up);
-EXPORT_SYMBOL(svc_makesock);
-EXPORT_SYMBOL(svc_reserve);
-EXPORT_SYMBOL(svc_auth_register);
-EXPORT_SYMBOL(auth_domain_lookup);
-EXPORT_SYMBOL(svc_authenticate);
-EXPORT_SYMBOL(svc_set_client);
-
-/* RPC statistics */
-#ifdef CONFIG_PROC_FS
-EXPORT_SYMBOL(svc_proc_register);
-EXPORT_SYMBOL(svc_proc_unregister);
-EXPORT_SYMBOL(svc_seq_show);
-#endif
-
-/* caching... */
-EXPORT_SYMBOL(auth_domain_find);
-EXPORT_SYMBOL(auth_domain_put);
-EXPORT_SYMBOL(auth_unix_add_addr);
-EXPORT_SYMBOL(auth_unix_forget_old);
-EXPORT_SYMBOL(auth_unix_lookup);
-EXPORT_SYMBOL(cache_check);
-EXPORT_SYMBOL(cache_flush);
-EXPORT_SYMBOL(cache_purge);
-EXPORT_SYMBOL(cache_register);
-EXPORT_SYMBOL(cache_unregister);
-EXPORT_SYMBOL(qword_add);
-EXPORT_SYMBOL(qword_addhex);
-EXPORT_SYMBOL(qword_get);
-EXPORT_SYMBOL(svcauth_unix_purge);
-EXPORT_SYMBOL(unix_domain_find);
-
 extern struct cache_detail ip_map_cache, unix_gid_cache;
 
 static int __init
@@ -85,7 +43,8 @@ init_sunrpc(void)
 #endif
 	cache_register(&ip_map_cache);
 	cache_register(&unix_gid_cache);
-	init_socket_xprt();
+	svc_init_xprt_sock();	/* svc sock transport */
+	init_socket_xprt();	/* clnt sock transport */
 	rpcauth_init_module();
 out:
 	return err;
@@ -96,12 +55,11 @@ cleanup_sunrpc(void)
 {
 	rpcauth_remove_module();
 	cleanup_socket_xprt();
+	svc_cleanup_xprt_sock();
 	unregister_rpc_pipefs();
 	rpc_destroy_mempool();
-	if (cache_unregister(&ip_map_cache))
+	cache_unregister(&ip_map_cache);
-		printk(KERN_ERR "sunrpc: failed to unregister ip_map cache\n");
+	cache_unregister(&unix_gid_cache);
-	if (cache_unregister(&unix_gid_cache))
-	      printk(KERN_ERR "sunrpc: failed to unregister unix_gid cache\n");
 #ifdef RPC_DEBUG
 	rpc_unregister_sysctl();
 #endif

net/sunrpc/svc.c

@@ -364,7 +364,7 @@ __svc_create(struct svc_program *prog, unsigned int bufsize, int npools,
 	   void (*shutdown)(struct svc_serv *serv))
 {
 	struct svc_serv	*serv;
-	int vers;
+	unsigned int vers;
 	unsigned int xdrsize;
 	unsigned int i;
 
@@ -433,6 +433,7 @@ svc_create(struct svc_program *prog, unsigned int bufsize,
 {
 	return __svc_create(prog, bufsize, /*npools*/1, shutdown);
 }
+EXPORT_SYMBOL(svc_create);
 
 struct svc_serv *
 svc_create_pooled(struct svc_program *prog, unsigned int bufsize,
@@ -452,6 +453,7 @@ svc_create_pooled(struct svc_program *prog, unsigned int bufsize,
 
 	return serv;
 }
+EXPORT_SYMBOL(svc_create_pooled);
 
 /*
  * Destroy an RPC service.  Should be called with the BKL held
@@ -459,9 +461,6 @@ svc_create_pooled(struct svc_program *prog, unsigned int bufsize,
 void
 svc_destroy(struct svc_serv *serv)
 {
-	struct svc_sock	*svsk;
-	struct svc_sock *tmp;
-
 	dprintk("svc: svc_destroy(%s, %d)\n",
 				serv->sv_program->pg_name,
 				serv->sv_nrthreads);
@@ -476,14 +475,12 @@ svc_destroy(struct svc_serv *serv)
 
 	del_timer_sync(&serv->sv_temptimer);
 
-	list_for_each_entry_safe(svsk, tmp, &serv->sv_tempsocks, sk_list)
+	svc_close_all(&serv->sv_tempsocks);
-		svc_force_close_socket(svsk);
 
 	if (serv->sv_shutdown)
 		serv->sv_shutdown(serv);
 
-	list_for_each_entry_safe(svsk, tmp, &serv->sv_permsocks, sk_list)
+	svc_close_all(&serv->sv_permsocks);
-		svc_force_close_socket(svsk);
 
 	BUG_ON(!list_empty(&serv->sv_permsocks));
 	BUG_ON(!list_empty(&serv->sv_tempsocks));
@@ -498,6 +495,7 @@ svc_destroy(struct svc_serv *serv)
 	kfree(serv->sv_pools);
 	kfree(serv);
 }
+EXPORT_SYMBOL(svc_destroy);
 
 /*
  * Allocate an RPC server's buffer space.
@@ -536,6 +534,44 @@ svc_release_buffer(struct svc_rqst *rqstp)
 			put_page(rqstp->rq_pages[i]);
 }
 
+struct svc_rqst *
+svc_prepare_thread(struct svc_serv *serv, struct svc_pool *pool)
+{
+	struct svc_rqst	*rqstp;
+
+	rqstp = kzalloc(sizeof(*rqstp), GFP_KERNEL);
+	if (!rqstp)
+		goto out_enomem;
+
+	init_waitqueue_head(&rqstp->rq_wait);
+
+	serv->sv_nrthreads++;
+	spin_lock_bh(&pool->sp_lock);
+	pool->sp_nrthreads++;
+	list_add(&rqstp->rq_all, &pool->sp_all_threads);
+	spin_unlock_bh(&pool->sp_lock);
+	rqstp->rq_server = serv;
+	rqstp->rq_pool = pool;
+
+	rqstp->rq_argp = kmalloc(serv->sv_xdrsize, GFP_KERNEL);
+	if (!rqstp->rq_argp)
+		goto out_thread;
+
+	rqstp->rq_resp = kmalloc(serv->sv_xdrsize, GFP_KERNEL);
+	if (!rqstp->rq_resp)
+		goto out_thread;
+
+	if (!svc_init_buffer(rqstp, serv->sv_max_mesg))
+		goto out_thread;
+
+	return rqstp;
+out_thread:
+	svc_exit_thread(rqstp);
+out_enomem:
+	return ERR_PTR(-ENOMEM);
+}
+EXPORT_SYMBOL(svc_prepare_thread);
+
 /*
  * Create a thread in the given pool.  Caller must hold BKL.
  * On a NUMA or SMP machine, with a multi-pool serv, the thread
@@ -550,24 +586,11 @@ __svc_create_thread(svc_thread_fn func, struct svc_serv *serv,
 	int		have_oldmask = 0;
 	cpumask_t	oldmask;
 
-	rqstp = kzalloc(sizeof(*rqstp), GFP_KERNEL);
+	rqstp = svc_prepare_thread(serv, pool);
-	if (!rqstp)
+	if (IS_ERR(rqstp)) {
+		error = PTR_ERR(rqstp);
 		goto out;
-
+	}
-	init_waitqueue_head(&rqstp->rq_wait);
-
-	if (!(rqstp->rq_argp = kmalloc(serv->sv_xdrsize, GFP_KERNEL))
-	 || !(rqstp->rq_resp = kmalloc(serv->sv_xdrsize, GFP_KERNEL))
-	 || !svc_init_buffer(rqstp, serv->sv_max_mesg))
-		goto out_thread;
-
-	serv->sv_nrthreads++;
-	spin_lock_bh(&pool->sp_lock);
-	pool->sp_nrthreads++;
-	list_add(&rqstp->rq_all, &pool->sp_all_threads);
-	spin_unlock_bh(&pool->sp_lock);
-	rqstp->rq_server = serv;
-	rqstp->rq_pool = pool;
 
 	if (serv->sv_nrpools > 1)
 		have_oldmask = svc_pool_map_set_cpumask(pool->sp_id, &oldmask);
@@ -597,6 +620,7 @@ svc_create_thread(svc_thread_fn func, struct svc_serv *serv)
 {
 	return __svc_create_thread(func, serv, &serv->sv_pools[0]);
 }
+EXPORT_SYMBOL(svc_create_thread);
 
 /*
  * Choose a pool in which to create a new thread, for svc_set_num_threads
@@ -700,6 +724,7 @@ svc_set_num_threads(struct svc_serv *serv, struct svc_pool *pool, int nrservs)
 
 	return error;
 }
+EXPORT_SYMBOL(svc_set_num_threads);
 
 /*
  * Called from a server thread as it's exiting.  Caller must hold BKL.
@@ -726,6 +751,7 @@ svc_exit_thread(struct svc_rqst *rqstp)
 	if (serv)
 		svc_destroy(serv);
 }
+EXPORT_SYMBOL(svc_exit_thread);
 
 /*
  * Register an RPC service with the local portmapper.
@@ -737,7 +763,8 @@ svc_register(struct svc_serv *serv, int proto, unsigned short port)
 {
 	struct svc_program	*progp;
 	unsigned long		flags;
-	int			i, error = 0, dummy;
+	unsigned int		i;
+	int			error = 0, dummy;
 
 	if (!port)
 		clear_thread_flag(TIF_SIGPENDING);
@@ -840,9 +867,9 @@ svc_process(struct svc_rqst *rqstp)
 	rqstp->rq_res.tail[0].iov_len = 0;
 	/* Will be turned off only in gss privacy case: */
 	rqstp->rq_splice_ok = 1;
-	/* tcp needs a space for the record length... */
+
-	if (rqstp->rq_prot == IPPROTO_TCP)
+	/* Setup reply header */
-		svc_putnl(resv, 0);
+	rqstp->rq_xprt->xpt_ops->xpo_prep_reply_hdr(rqstp);
 
 	rqstp->rq_xid = svc_getu32(argv);
 	svc_putu32(resv, rqstp->rq_xid);
@@ -1049,16 +1076,15 @@ err_bad:
 	svc_putnl(resv, ntohl(rpc_stat));
 	goto sendit;
 }
+EXPORT_SYMBOL(svc_process);
 
 /*
  * Return (transport-specific) limit on the rpc payload.
  */
 u32 svc_max_payload(const struct svc_rqst *rqstp)
 {
-	int max = RPCSVC_MAXPAYLOAD_TCP;
+	u32 max = rqstp->rq_xprt->xpt_class->xcl_max_payload;
 
-	if (rqstp->rq_sock->sk_sock->type == SOCK_DGRAM)
-		max = RPCSVC_MAXPAYLOAD_UDP;
 	if (rqstp->rq_server->sv_max_payload < max)
 		max = rqstp->rq_server->sv_max_payload;
 	return max;

net/sunrpc/svcauth.c

@@ -57,11 +57,13 @@ svc_authenticate(struct svc_rqst *rqstp, __be32 *authp)
 	rqstp->rq_authop = aops;
 	return aops->accept(rqstp, authp);
 }
+EXPORT_SYMBOL(svc_authenticate);
 
 int svc_set_client(struct svc_rqst *rqstp)
 {
 	return rqstp->rq_authop->set_client(rqstp);
 }
+EXPORT_SYMBOL(svc_set_client);
 
 /* A request, which was authenticated, has now executed.
  * Time to finalise the credentials and verifier
@@ -93,6 +95,7 @@ svc_auth_register(rpc_authflavor_t flavor, struct auth_ops *aops)
 	spin_unlock(&authtab_lock);
 	return rv;
 }
+EXPORT_SYMBOL(svc_auth_register);
 
 void
 svc_auth_unregister(rpc_authflavor_t flavor)
@@ -129,6 +132,7 @@ void auth_domain_put(struct auth_domain *dom)
 		spin_unlock(&auth_domain_lock);
 	}
 }
+EXPORT_SYMBOL(auth_domain_put);
 
 struct auth_domain *
 auth_domain_lookup(char *name, struct auth_domain *new)
@@ -153,8 +157,10 @@ auth_domain_lookup(char *name, struct auth_domain *new)
 	spin_unlock(&auth_domain_lock);
 	return new;
 }
+EXPORT_SYMBOL(auth_domain_lookup);
 
 struct auth_domain *auth_domain_find(char *name)
 {
 	return auth_domain_lookup(name, NULL);
 }
+EXPORT_SYMBOL(auth_domain_find);

net/sunrpc/svcauth_unix.c

@@ -63,6 +63,7 @@ struct auth_domain *unix_domain_find(char *name)
 		rv = auth_domain_lookup(name, &new->h);
 	}
 }
+EXPORT_SYMBOL(unix_domain_find);
 
 static void svcauth_unix_domain_release(struct auth_domain *dom)
 {
@@ -340,6 +341,7 @@ int auth_unix_add_addr(struct in_addr addr, struct auth_domain *dom)
 	else
 		return -ENOMEM;
 }
+EXPORT_SYMBOL(auth_unix_add_addr);
 
 int auth_unix_forget_old(struct auth_domain *dom)
 {
@@ -351,6 +353,7 @@ int auth_unix_forget_old(struct auth_domain *dom)
 	udom->addr_changes++;
 	return 0;
 }
+EXPORT_SYMBOL(auth_unix_forget_old);
 
 struct auth_domain *auth_unix_lookup(struct in_addr addr)
 {
@@ -375,19 +378,23 @@ struct auth_domain *auth_unix_lookup(struct in_addr addr)
 	cache_put(&ipm->h, &ip_map_cache);
 	return rv;
 }
+EXPORT_SYMBOL(auth_unix_lookup);
 
 void svcauth_unix_purge(void)
 {
 	cache_purge(&ip_map_cache);
 }
+EXPORT_SYMBOL(svcauth_unix_purge);
 
 static inline struct ip_map *
 ip_map_cached_get(struct svc_rqst *rqstp)
 {
-	struct ip_map *ipm;
+	struct ip_map *ipm = NULL;
-	struct svc_sock *svsk = rqstp->rq_sock;
+	struct svc_xprt *xprt = rqstp->rq_xprt;
-	spin_lock(&svsk->sk_lock);
+
-	ipm = svsk->sk_info_authunix;
+	if (test_bit(XPT_CACHE_AUTH, &xprt->xpt_flags)) {
+		spin_lock(&xprt->xpt_lock);
+		ipm = xprt->xpt_auth_cache;
 		if (ipm != NULL) {
 			if (!cache_valid(&ipm->h)) {
 				/*
@@ -395,30 +402,32 @@ ip_map_cached_get(struct svc_rqst *rqstp)
 				 * remembered, e.g. by a second mount from the
 				 * same IP address.
 				 */
-			svsk->sk_info_authunix = NULL;
+				xprt->xpt_auth_cache = NULL;
-			spin_unlock(&svsk->sk_lock);
+				spin_unlock(&xprt->xpt_lock);
 				cache_put(&ipm->h, &ip_map_cache);
 				return NULL;
 			}
 			cache_get(&ipm->h);
 		}
-	spin_unlock(&svsk->sk_lock);
+		spin_unlock(&xprt->xpt_lock);
+	}
 	return ipm;
 }
 
 static inline void
 ip_map_cached_put(struct svc_rqst *rqstp, struct ip_map *ipm)
 {
-	struct svc_sock *svsk = rqstp->rq_sock;
+	struct svc_xprt *xprt = rqstp->rq_xprt;
 
-	spin_lock(&svsk->sk_lock);
+	if (test_bit(XPT_CACHE_AUTH, &xprt->xpt_flags)) {
-	if (svsk->sk_sock->type == SOCK_STREAM &&
+		spin_lock(&xprt->xpt_lock);
-	    svsk->sk_info_authunix == NULL) {
+		if (xprt->xpt_auth_cache == NULL) {
 			/* newly cached, keep the reference */
-		svsk->sk_info_authunix = ipm;
+			xprt->xpt_auth_cache = ipm;
 			ipm = NULL;
 		}
-	spin_unlock(&svsk->sk_lock);
+		spin_unlock(&xprt->xpt_lock);
+	}
 	if (ipm)
 		cache_put(&ipm->h, &ip_map_cache);
 }

net/sunrpc/sysctl.c

@@ -18,6 +18,7 @@
 #include <linux/sunrpc/types.h>
 #include <linux/sunrpc/sched.h>
 #include <linux/sunrpc/stats.h>
+#include <linux/sunrpc/svc_xprt.h>
 
 /*
  * Declare the debug flags here
@@ -55,6 +56,30 @@ rpc_unregister_sysctl(void)
 	}
 }
 
+static int proc_do_xprt(ctl_table *table, int write, struct file *file,
+			void __user *buffer, size_t *lenp, loff_t *ppos)
+{
+	char tmpbuf[256];
+	int len;
+	if ((*ppos && !write) || !*lenp) {
+		*lenp = 0;
+		return 0;
+	}
+	if (write)
+		return -EINVAL;
+	else {
+		len = svc_print_xprts(tmpbuf, sizeof(tmpbuf));
+		if (!access_ok(VERIFY_WRITE, buffer, len))
+			return -EFAULT;
+
+		if (__copy_to_user(buffer, tmpbuf, len))
+			return -EFAULT;
+	}
+	*lenp -= len;
+	*ppos += len;
+	return 0;
+}
+
 static int
 proc_dodebug(ctl_table *table, int write, struct file *file,
 				void __user *buffer, size_t *lenp, loff_t *ppos)
@@ -147,6 +172,12 @@ static ctl_table debug_table[] = {
 		.mode		= 0644,
 		.proc_handler	= &proc_dodebug
 	},
+	{
+		.procname	= "transports",
+		.maxlen		= 256,
+		.mode		= 0444,
+		.proc_handler	= &proc_do_xprt,
+	},
 	{ .ctl_name = 0 }
 };
 

net/sunrpc/xdr.c

@@ -96,11 +96,13 @@ xdr_encode_string(__be32 *p, const char *string)
 EXPORT_SYMBOL(xdr_encode_string);
 
 __be32 *
-xdr_decode_string_inplace(__be32 *p, char **sp, int *lenp, int maxlen)
+xdr_decode_string_inplace(__be32 *p, char **sp,
+			  unsigned int *lenp, unsigned int maxlen)
 {
-	unsigned int	len;
+	u32 len;
 
-	if ((len = ntohl(*p++)) > maxlen)
+	len = ntohl(*p++);
+	if (len > maxlen)
 		return NULL;
 	*lenp = len;
 	*sp = (char *) p;

net/sunrpc/xprtrdma/Makefile

@@ -1,3 +1,8 @@
 obj-$(CONFIG_SUNRPC_XPRT_RDMA) += xprtrdma.o
 
 xprtrdma-y := transport.o rpc_rdma.o verbs.o
+
+obj-$(CONFIG_SUNRPC_XPRT_RDMA) += svcrdma.o
+
+svcrdma-y := svc_rdma.o svc_rdma_transport.o \
+	svc_rdma_marshal.o svc_rdma_sendto.o svc_rdma_recvfrom.o

net/sunrpc/xprtrdma/svc_rdma.c

@@ -0,0 +1,266 @@
+/*
+ * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the BSD-type
+ * license below:
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *      Redistributions of source code must retain the above copyright
+ *      notice, this list of conditions and the following disclaimer.
+ *
+ *      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.
+ *
+ *      Neither the name of the Network Appliance, Inc. 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 BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "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 COPYRIGHT
+ * OWNER 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.
+ *
+ * Author: Tom Tucker <tom@opengridcomputing.com>
+ */
+#include <linux/module.h>
+#include <linux/init.h>
+#include <linux/fs.h>
+#include <linux/sysctl.h>
+#include <linux/sunrpc/clnt.h>
+#include <linux/sunrpc/sched.h>
+#include <linux/sunrpc/svc_rdma.h>
+
+#define RPCDBG_FACILITY	RPCDBG_SVCXPRT
+
+/* RPC/RDMA parameters */
+unsigned int svcrdma_ord = RPCRDMA_ORD;
+static unsigned int min_ord = 1;
+static unsigned int max_ord = 4096;
+unsigned int svcrdma_max_requests = RPCRDMA_MAX_REQUESTS;
+static unsigned int min_max_requests = 4;
+static unsigned int max_max_requests = 16384;
+unsigned int svcrdma_max_req_size = RPCRDMA_MAX_REQ_SIZE;
+static unsigned int min_max_inline = 4096;
+static unsigned int max_max_inline = 65536;
+
+atomic_t rdma_stat_recv;
+atomic_t rdma_stat_read;
+atomic_t rdma_stat_write;
+atomic_t rdma_stat_sq_starve;
+atomic_t rdma_stat_rq_starve;
+atomic_t rdma_stat_rq_poll;
+atomic_t rdma_stat_rq_prod;
+atomic_t rdma_stat_sq_poll;
+atomic_t rdma_stat_sq_prod;
+
+/*
+ * This function implements reading and resetting an atomic_t stat
+ * variable through read/write to a proc file. Any write to the file
+ * resets the associated statistic to zero. Any read returns it's
+ * current value.
+ */
+static int read_reset_stat(ctl_table *table, int write,
+			   struct file *filp, void __user *buffer, size_t *lenp,
+			   loff_t *ppos)
+{
+	atomic_t *stat = (atomic_t *)table->data;
+
+	if (!stat)
+		return -EINVAL;
+
+	if (write)
+		atomic_set(stat, 0);
+	else {
+		char str_buf[32];
+		char *data;
+		int len = snprintf(str_buf, 32, "%d\n", atomic_read(stat));
+		if (len >= 32)
+			return -EFAULT;
+		len = strlen(str_buf);
+		if (*ppos > len) {
+			*lenp = 0;
+			return 0;
+		}
+		data = &str_buf[*ppos];
+		len -= *ppos;
+		if (len > *lenp)
+			len = *lenp;
+		if (len && copy_to_user(buffer, str_buf, len))
+			return -EFAULT;
+		*lenp = len;
+		*ppos += len;
+	}
+	return 0;
+}
+
+static struct ctl_table_header *svcrdma_table_header;
+static ctl_table svcrdma_parm_table[] = {
+	{
+		.procname	= "max_requests",
+		.data		= &svcrdma_max_requests,
+		.maxlen		= sizeof(unsigned int),
+		.mode		= 0644,
+		.proc_handler	= &proc_dointvec_minmax,
+		.strategy	= &sysctl_intvec,
+		.extra1		= &min_max_requests,
+		.extra2		= &max_max_requests
+	},
+	{
+		.procname	= "max_req_size",
+		.data		= &svcrdma_max_req_size,
+		.maxlen		= sizeof(unsigned int),
+		.mode		= 0644,
+		.proc_handler	= &proc_dointvec_minmax,
+		.strategy	= &sysctl_intvec,
+		.extra1		= &min_max_inline,
+		.extra2		= &max_max_inline
+	},
+	{
+		.procname	= "max_outbound_read_requests",
+		.data		= &svcrdma_ord,
+		.maxlen		= sizeof(unsigned int),
+		.mode		= 0644,
+		.proc_handler	= &proc_dointvec_minmax,
+		.strategy	= &sysctl_intvec,
+		.extra1		= &min_ord,
+		.extra2		= &max_ord,
+	},
+
+	{
+		.procname	= "rdma_stat_read",
+		.data		= &rdma_stat_read,
+		.maxlen		= sizeof(atomic_t),
+		.mode		= 0644,
+		.proc_handler	= &read_reset_stat,
+	},
+	{
+		.procname	= "rdma_stat_recv",
+		.data		= &rdma_stat_recv,
+		.maxlen		= sizeof(atomic_t),
+		.mode		= 0644,
+		.proc_handler	= &read_reset_stat,
+	},
+	{
+		.procname	= "rdma_stat_write",
+		.data		= &rdma_stat_write,
+		.maxlen		= sizeof(atomic_t),
+		.mode		= 0644,
+		.proc_handler	= &read_reset_stat,
+	},
+	{
+		.procname	= "rdma_stat_sq_starve",
+		.data		= &rdma_stat_sq_starve,
+		.maxlen		= sizeof(atomic_t),
+		.mode		= 0644,
+		.proc_handler	= &read_reset_stat,
+	},
+	{
+		.procname	= "rdma_stat_rq_starve",
+		.data		= &rdma_stat_rq_starve,
+		.maxlen		= sizeof(atomic_t),
+		.mode		= 0644,
+		.proc_handler	= &read_reset_stat,
+	},
+	{
+		.procname	= "rdma_stat_rq_poll",
+		.data		= &rdma_stat_rq_poll,
+		.maxlen		= sizeof(atomic_t),
+		.mode		= 0644,
+		.proc_handler	= &read_reset_stat,
+	},
+	{
+		.procname	= "rdma_stat_rq_prod",
+		.data		= &rdma_stat_rq_prod,
+		.maxlen		= sizeof(atomic_t),
+		.mode		= 0644,
+		.proc_handler	= &read_reset_stat,
+	},
+	{
+		.procname	= "rdma_stat_sq_poll",
+		.data		= &rdma_stat_sq_poll,
+		.maxlen		= sizeof(atomic_t),
+		.mode		= 0644,
+		.proc_handler	= &read_reset_stat,
+	},
+	{
+		.procname	= "rdma_stat_sq_prod",
+		.data		= &rdma_stat_sq_prod,
+		.maxlen		= sizeof(atomic_t),
+		.mode		= 0644,
+		.proc_handler	= &read_reset_stat,
+	},
+	{
+		.ctl_name = 0,
+	},
+};
+
+static ctl_table svcrdma_table[] = {
+	{
+		.procname	= "svc_rdma",
+		.mode		= 0555,
+		.child		= svcrdma_parm_table
+	},
+	{
+		.ctl_name = 0,
+	},
+};
+
+static ctl_table svcrdma_root_table[] = {
+	{
+		.ctl_name	= CTL_SUNRPC,
+		.procname	= "sunrpc",
+		.mode		= 0555,
+		.child		= svcrdma_table
+	},
+	{
+		.ctl_name = 0,
+	},
+};
+
+void svc_rdma_cleanup(void)
+{
+	dprintk("SVCRDMA Module Removed, deregister RPC RDMA transport\n");
+	if (svcrdma_table_header) {
+		unregister_sysctl_table(svcrdma_table_header);
+		svcrdma_table_header = NULL;
+	}
+	svc_unreg_xprt_class(&svc_rdma_class);
+}
+
+int svc_rdma_init(void)
+{
+	dprintk("SVCRDMA Module Init, register RPC RDMA transport\n");
+	dprintk("\tsvcrdma_ord      : %d\n", svcrdma_ord);
+	dprintk("\tmax_requests     : %d\n", svcrdma_max_requests);
+	dprintk("\tsq_depth         : %d\n",
+		svcrdma_max_requests * RPCRDMA_SQ_DEPTH_MULT);
+	dprintk("\tmax_inline       : %d\n", svcrdma_max_req_size);
+	if (!svcrdma_table_header)
+		svcrdma_table_header =
+			register_sysctl_table(svcrdma_root_table);
+
+	/* Register RDMA with the SVC transport switch */
+	svc_reg_xprt_class(&svc_rdma_class);
+	return 0;
+}
+MODULE_AUTHOR("Tom Tucker <tom@opengridcomputing.com>");
+MODULE_DESCRIPTION("SVC RDMA Transport");
+MODULE_LICENSE("Dual BSD/GPL");
+module_init(svc_rdma_init);
+module_exit(svc_rdma_cleanup);

net/sunrpc/xprtrdma/svc_rdma_marshal.c

@@ -0,0 +1,412 @@
+/*
+ * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the BSD-type
+ * license below:
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *      Redistributions of source code must retain the above copyright
+ *      notice, this list of conditions and the following disclaimer.
+ *
+ *      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.
+ *
+ *      Neither the name of the Network Appliance, Inc. 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 BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "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 COPYRIGHT
+ * OWNER 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.
+ *
+ * Author: Tom Tucker <tom@opengridcomputing.com>
+ */
+
+#include <linux/sunrpc/xdr.h>
+#include <linux/sunrpc/debug.h>
+#include <asm/unaligned.h>
+#include <linux/sunrpc/rpc_rdma.h>
+#include <linux/sunrpc/svc_rdma.h>
+
+#define RPCDBG_FACILITY	RPCDBG_SVCXPRT
+
+/*
+ * Decodes a read chunk list. The expected format is as follows:
+ *    descrim  : xdr_one
+ *    position : u32 offset into XDR stream
+ *    handle   : u32 RKEY
+ *    . . .
+ *  end-of-list: xdr_zero
+ */
+static u32 *decode_read_list(u32 *va, u32 *vaend)
+{
+	struct rpcrdma_read_chunk *ch = (struct rpcrdma_read_chunk *)va;
+
+	while (ch->rc_discrim != xdr_zero) {
+		u64 ch_offset;
+
+		if (((unsigned long)ch + sizeof(struct rpcrdma_read_chunk)) >
+		    (unsigned long)vaend) {
+			dprintk("svcrdma: vaend=%p, ch=%p\n", vaend, ch);
+			return NULL;
+		}
+
+		ch->rc_discrim = ntohl(ch->rc_discrim);
+		ch->rc_position = ntohl(ch->rc_position);
+		ch->rc_target.rs_handle = ntohl(ch->rc_target.rs_handle);
+		ch->rc_target.rs_length = ntohl(ch->rc_target.rs_length);
+		va = (u32 *)&ch->rc_target.rs_offset;
+		xdr_decode_hyper(va, &ch_offset);
+		put_unaligned(ch_offset, (u64 *)va);
+		ch++;
+	}
+	return (u32 *)&ch->rc_position;
+}
+
+/*
+ * Determine number of chunks and total bytes in chunk list. The chunk
+ * list has already been verified to fit within the RPCRDMA header.
+ */
+void svc_rdma_rcl_chunk_counts(struct rpcrdma_read_chunk *ch,
+			       int *ch_count, int *byte_count)
+{
+	/* compute the number of bytes represented by read chunks */
+	*byte_count = 0;
+	*ch_count = 0;
+	for (; ch->rc_discrim != 0; ch++) {
+		*byte_count = *byte_count + ch->rc_target.rs_length;
+		*ch_count = *ch_count + 1;
+	}
+}
+
+/*
+ * Decodes a write chunk list. The expected format is as follows:
+ *    descrim  : xdr_one
+ *    nchunks  : <count>
+ *       handle   : u32 RKEY              ---+
+ *       length   : u32 <len of segment>     |
+ *       offset   : remove va                + <count>
+ *       . . .                               |
+ *                                        ---+
+ */
+static u32 *decode_write_list(u32 *va, u32 *vaend)
+{
+	int ch_no;
+	struct rpcrdma_write_array *ary =
+		(struct rpcrdma_write_array *)va;
+
+	/* Check for not write-array */
+	if (ary->wc_discrim == xdr_zero)
+		return (u32 *)&ary->wc_nchunks;
+
+	if ((unsigned long)ary + sizeof(struct rpcrdma_write_array) >
+	    (unsigned long)vaend) {
+		dprintk("svcrdma: ary=%p, vaend=%p\n", ary, vaend);
+		return NULL;
+	}
+	ary->wc_discrim = ntohl(ary->wc_discrim);
+	ary->wc_nchunks = ntohl(ary->wc_nchunks);
+	if (((unsigned long)&ary->wc_array[0] +
+	     (sizeof(struct rpcrdma_write_chunk) * ary->wc_nchunks)) >
+	    (unsigned long)vaend) {
+		dprintk("svcrdma: ary=%p, wc_nchunks=%d, vaend=%p\n",
+			ary, ary->wc_nchunks, vaend);
+		return NULL;
+	}
+	for (ch_no = 0; ch_no < ary->wc_nchunks; ch_no++) {
+		u64 ch_offset;
+
+		ary->wc_array[ch_no].wc_target.rs_handle =
+			ntohl(ary->wc_array[ch_no].wc_target.rs_handle);
+		ary->wc_array[ch_no].wc_target.rs_length =
+			ntohl(ary->wc_array[ch_no].wc_target.rs_length);
+		va = (u32 *)&ary->wc_array[ch_no].wc_target.rs_offset;
+		xdr_decode_hyper(va, &ch_offset);
+		put_unaligned(ch_offset, (u64 *)va);
+	}
+
+	/*
+	 * rs_length is the 2nd 4B field in wc_target and taking its
+	 * address skips the list terminator
+	 */
+	return (u32 *)&ary->wc_array[ch_no].wc_target.rs_length;
+}
+
+static u32 *decode_reply_array(u32 *va, u32 *vaend)
+{
+	int ch_no;
+	struct rpcrdma_write_array *ary =
+		(struct rpcrdma_write_array *)va;
+
+	/* Check for no reply-array */
+	if (ary->wc_discrim == xdr_zero)
+		return (u32 *)&ary->wc_nchunks;
+
+	if ((unsigned long)ary + sizeof(struct rpcrdma_write_array) >
+	    (unsigned long)vaend) {
+		dprintk("svcrdma: ary=%p, vaend=%p\n", ary, vaend);
+		return NULL;
+	}
+	ary->wc_discrim = ntohl(ary->wc_discrim);
+	ary->wc_nchunks = ntohl(ary->wc_nchunks);
+	if (((unsigned long)&ary->wc_array[0] +
+	     (sizeof(struct rpcrdma_write_chunk) * ary->wc_nchunks)) >
+	    (unsigned long)vaend) {
+		dprintk("svcrdma: ary=%p, wc_nchunks=%d, vaend=%p\n",
+			ary, ary->wc_nchunks, vaend);
+		return NULL;
+	}
+	for (ch_no = 0; ch_no < ary->wc_nchunks; ch_no++) {
+		u64 ch_offset;
+
+		ary->wc_array[ch_no].wc_target.rs_handle =
+			ntohl(ary->wc_array[ch_no].wc_target.rs_handle);
+		ary->wc_array[ch_no].wc_target.rs_length =
+			ntohl(ary->wc_array[ch_no].wc_target.rs_length);
+		va = (u32 *)&ary->wc_array[ch_no].wc_target.rs_offset;
+		xdr_decode_hyper(va, &ch_offset);
+		put_unaligned(ch_offset, (u64 *)va);
+	}
+
+	return (u32 *)&ary->wc_array[ch_no];
+}
+
+int svc_rdma_xdr_decode_req(struct rpcrdma_msg **rdma_req,
+			    struct svc_rqst *rqstp)
+{
+	struct rpcrdma_msg *rmsgp = NULL;
+	u32 *va;
+	u32 *vaend;
+	u32 hdr_len;
+
+	rmsgp = (struct rpcrdma_msg *)rqstp->rq_arg.head[0].iov_base;
+
+	/* Verify that there's enough bytes for header + something */
+	if (rqstp->rq_arg.len <= RPCRDMA_HDRLEN_MIN) {
+		dprintk("svcrdma: header too short = %d\n",
+			rqstp->rq_arg.len);
+		return -EINVAL;
+	}
+
+	/* Decode the header */
+	rmsgp->rm_xid = ntohl(rmsgp->rm_xid);
+	rmsgp->rm_vers = ntohl(rmsgp->rm_vers);
+	rmsgp->rm_credit = ntohl(rmsgp->rm_credit);
+	rmsgp->rm_type = ntohl(rmsgp->rm_type);
+
+	if (rmsgp->rm_vers != RPCRDMA_VERSION)
+		return -ENOSYS;
+
+	/* Pull in the extra for the padded case and bump our pointer */
+	if (rmsgp->rm_type == RDMA_MSGP) {
+		int hdrlen;
+		rmsgp->rm_body.rm_padded.rm_align =
+			ntohl(rmsgp->rm_body.rm_padded.rm_align);
+		rmsgp->rm_body.rm_padded.rm_thresh =
+			ntohl(rmsgp->rm_body.rm_padded.rm_thresh);
+
+		va = &rmsgp->rm_body.rm_padded.rm_pempty[4];
+		rqstp->rq_arg.head[0].iov_base = va;
+		hdrlen = (u32)((unsigned long)va - (unsigned long)rmsgp);
+		rqstp->rq_arg.head[0].iov_len -= hdrlen;
+		if (hdrlen > rqstp->rq_arg.len)
+			return -EINVAL;
+		return hdrlen;
+	}
+
+	/* The chunk list may contain either a read chunk list or a write
+	 * chunk list and a reply chunk list.
+	 */
+	va = &rmsgp->rm_body.rm_chunks[0];
+	vaend = (u32 *)((unsigned long)rmsgp + rqstp->rq_arg.len);
+	va = decode_read_list(va, vaend);
+	if (!va)
+		return -EINVAL;
+	va = decode_write_list(va, vaend);
+	if (!va)
+		return -EINVAL;
+	va = decode_reply_array(va, vaend);
+	if (!va)
+		return -EINVAL;
+
+	rqstp->rq_arg.head[0].iov_base = va;
+	hdr_len = (unsigned long)va - (unsigned long)rmsgp;
+	rqstp->rq_arg.head[0].iov_len -= hdr_len;
+
+	*rdma_req = rmsgp;
+	return hdr_len;
+}
+
+int svc_rdma_xdr_decode_deferred_req(struct svc_rqst *rqstp)
+{
+	struct rpcrdma_msg *rmsgp = NULL;
+	struct rpcrdma_read_chunk *ch;
+	struct rpcrdma_write_array *ary;
+	u32 *va;
+	u32 hdrlen;
+
+	dprintk("svcrdma: processing deferred RDMA header on rqstp=%p\n",
+		rqstp);
+	rmsgp = (struct rpcrdma_msg *)rqstp->rq_arg.head[0].iov_base;
+
+	/* Pull in the extra for the padded case and bump our pointer */
+	if (rmsgp->rm_type == RDMA_MSGP) {
+		va = &rmsgp->rm_body.rm_padded.rm_pempty[4];
+		rqstp->rq_arg.head[0].iov_base = va;
+		hdrlen = (u32)((unsigned long)va - (unsigned long)rmsgp);
+		rqstp->rq_arg.head[0].iov_len -= hdrlen;
+		return hdrlen;
+	}
+
+	/*
+	 * Skip all chunks to find RPC msg. These were previously processed
+	 */
+	va = &rmsgp->rm_body.rm_chunks[0];
+
+	/* Skip read-list */
+	for (ch = (struct rpcrdma_read_chunk *)va;
+	     ch->rc_discrim != xdr_zero; ch++);
+	va = (u32 *)&ch->rc_position;
+
+	/* Skip write-list */
+	ary = (struct rpcrdma_write_array *)va;
+	if (ary->wc_discrim == xdr_zero)
+		va = (u32 *)&ary->wc_nchunks;
+	else
+		/*
+		 * rs_length is the 2nd 4B field in wc_target and taking its
+		 * address skips the list terminator
+		 */
+		va = (u32 *)&ary->wc_array[ary->wc_nchunks].wc_target.rs_length;
+
+	/* Skip reply-array */
+	ary = (struct rpcrdma_write_array *)va;
+	if (ary->wc_discrim == xdr_zero)
+		va = (u32 *)&ary->wc_nchunks;
+	else
+		va = (u32 *)&ary->wc_array[ary->wc_nchunks];
+
+	rqstp->rq_arg.head[0].iov_base = va;
+	hdrlen = (unsigned long)va - (unsigned long)rmsgp;
+	rqstp->rq_arg.head[0].iov_len -= hdrlen;
+
+	return hdrlen;
+}
+
+int svc_rdma_xdr_encode_error(struct svcxprt_rdma *xprt,
+			      struct rpcrdma_msg *rmsgp,
+			      enum rpcrdma_errcode err, u32 *va)
+{
+	u32 *startp = va;
+
+	*va++ = htonl(rmsgp->rm_xid);
+	*va++ = htonl(rmsgp->rm_vers);
+	*va++ = htonl(xprt->sc_max_requests);
+	*va++ = htonl(RDMA_ERROR);
+	*va++ = htonl(err);
+	if (err == ERR_VERS) {
+		*va++ = htonl(RPCRDMA_VERSION);
+		*va++ = htonl(RPCRDMA_VERSION);
+	}
+
+	return (int)((unsigned long)va - (unsigned long)startp);
+}
+
+int svc_rdma_xdr_get_reply_hdr_len(struct rpcrdma_msg *rmsgp)
+{
+	struct rpcrdma_write_array *wr_ary;
+
+	/* There is no read-list in a reply */
+
+	/* skip write list */
+	wr_ary = (struct rpcrdma_write_array *)
+		&rmsgp->rm_body.rm_chunks[1];
+	if (wr_ary->wc_discrim)
+		wr_ary = (struct rpcrdma_write_array *)
+			&wr_ary->wc_array[ntohl(wr_ary->wc_nchunks)].
+			wc_target.rs_length;
+	else
+		wr_ary = (struct rpcrdma_write_array *)
+			&wr_ary->wc_nchunks;
+
+	/* skip reply array */
+	if (wr_ary->wc_discrim)
+		wr_ary = (struct rpcrdma_write_array *)
+			&wr_ary->wc_array[ntohl(wr_ary->wc_nchunks)];
+	else
+		wr_ary = (struct rpcrdma_write_array *)
+			&wr_ary->wc_nchunks;
+
+	return (unsigned long) wr_ary - (unsigned long) rmsgp;
+}
+
+void svc_rdma_xdr_encode_write_list(struct rpcrdma_msg *rmsgp, int chunks)
+{
+	struct rpcrdma_write_array *ary;
+
+	/* no read-list */
+	rmsgp->rm_body.rm_chunks[0] = xdr_zero;
+
+	/* write-array discrim */
+	ary = (struct rpcrdma_write_array *)
+		&rmsgp->rm_body.rm_chunks[1];
+	ary->wc_discrim = xdr_one;
+	ary->wc_nchunks = htonl(chunks);
+
+	/* write-list terminator */
+	ary->wc_array[chunks].wc_target.rs_handle = xdr_zero;
+
+	/* reply-array discriminator */
+	ary->wc_array[chunks].wc_target.rs_length = xdr_zero;
+}
+
+void svc_rdma_xdr_encode_reply_array(struct rpcrdma_write_array *ary,
+				 int chunks)
+{
+	ary->wc_discrim = xdr_one;
+	ary->wc_nchunks = htonl(chunks);
+}
+
+void svc_rdma_xdr_encode_array_chunk(struct rpcrdma_write_array *ary,
+				     int chunk_no,
+				     u32 rs_handle, u64 rs_offset,
+				     u32 write_len)
+{
+	struct rpcrdma_segment *seg = &ary->wc_array[chunk_no].wc_target;
+	seg->rs_handle = htonl(rs_handle);
+	seg->rs_length = htonl(write_len);
+	xdr_encode_hyper((u32 *) &seg->rs_offset, rs_offset);
+}
+
+void svc_rdma_xdr_encode_reply_header(struct svcxprt_rdma *xprt,
+				  struct rpcrdma_msg *rdma_argp,
+				  struct rpcrdma_msg *rdma_resp,
+				  enum rpcrdma_proc rdma_type)
+{
+	rdma_resp->rm_xid = htonl(rdma_argp->rm_xid);
+	rdma_resp->rm_vers = htonl(rdma_argp->rm_vers);
+	rdma_resp->rm_credit = htonl(xprt->sc_max_requests);
+	rdma_resp->rm_type = htonl(rdma_type);
+
+	/* Encode <nul> chunks lists */
+	rdma_resp->rm_body.rm_chunks[0] = xdr_zero;
+	rdma_resp->rm_body.rm_chunks[1] = xdr_zero;
+	rdma_resp->rm_body.rm_chunks[2] = xdr_zero;
+}

net/sunrpc/xprtrdma/svc_rdma_recvfrom.c

@@ -0,0 +1,586 @@
+/*
+ * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the BSD-type
+ * license below:
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *      Redistributions of source code must retain the above copyright
+ *      notice, this list of conditions and the following disclaimer.
+ *
+ *      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.
+ *
+ *      Neither the name of the Network Appliance, Inc. 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 BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "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 COPYRIGHT
+ * OWNER 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.
+ *
+ * Author: Tom Tucker <tom@opengridcomputing.com>
+ */
+
+#include <linux/sunrpc/debug.h>
+#include <linux/sunrpc/rpc_rdma.h>
+#include <linux/spinlock.h>
+#include <asm/unaligned.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/rdma_cm.h>
+#include <linux/sunrpc/svc_rdma.h>
+
+#define RPCDBG_FACILITY	RPCDBG_SVCXPRT
+
+/*
+ * Replace the pages in the rq_argpages array with the pages from the SGE in
+ * the RDMA_RECV completion. The SGL should contain full pages up until the
+ * last one.
+ */
+static void rdma_build_arg_xdr(struct svc_rqst *rqstp,
+			       struct svc_rdma_op_ctxt *ctxt,
+			       u32 byte_count)
+{
+	struct page *page;
+	u32 bc;
+	int sge_no;
+
+	/* Swap the page in the SGE with the page in argpages */
+	page = ctxt->pages[0];
+	put_page(rqstp->rq_pages[0]);
+	rqstp->rq_pages[0] = page;
+
+	/* Set up the XDR head */
+	rqstp->rq_arg.head[0].iov_base = page_address(page);
+	rqstp->rq_arg.head[0].iov_len = min(byte_count, ctxt->sge[0].length);
+	rqstp->rq_arg.len = byte_count;
+	rqstp->rq_arg.buflen = byte_count;
+
+	/* Compute bytes past head in the SGL */
+	bc = byte_count - rqstp->rq_arg.head[0].iov_len;
+
+	/* If data remains, store it in the pagelist */
+	rqstp->rq_arg.page_len = bc;
+	rqstp->rq_arg.page_base = 0;
+	rqstp->rq_arg.pages = &rqstp->rq_pages[1];
+	sge_no = 1;
+	while (bc && sge_no < ctxt->count) {
+		page = ctxt->pages[sge_no];
+		put_page(rqstp->rq_pages[sge_no]);
+		rqstp->rq_pages[sge_no] = page;
+		bc -= min(bc, ctxt->sge[sge_no].length);
+		rqstp->rq_arg.buflen += ctxt->sge[sge_no].length;
+		sge_no++;
+	}
+	rqstp->rq_respages = &rqstp->rq_pages[sge_no];
+
+	/* We should never run out of SGE because the limit is defined to
+	 * support the max allowed RPC data length
+	 */
+	BUG_ON(bc && (sge_no == ctxt->count));
+	BUG_ON((rqstp->rq_arg.head[0].iov_len + rqstp->rq_arg.page_len)
+	       != byte_count);
+	BUG_ON(rqstp->rq_arg.len != byte_count);
+
+	/* If not all pages were used from the SGL, free the remaining ones */
+	bc = sge_no;
+	while (sge_no < ctxt->count) {
+		page = ctxt->pages[sge_no++];
+		put_page(page);
+	}
+	ctxt->count = bc;
+
+	/* Set up tail */
+	rqstp->rq_arg.tail[0].iov_base = NULL;
+	rqstp->rq_arg.tail[0].iov_len = 0;
+}
+
+struct chunk_sge {
+	int start;		/* sge no for this chunk */
+	int count;		/* sge count for this chunk */
+};
+
+/* Encode a read-chunk-list as an array of IB SGE
+ *
+ * Assumptions:
+ * - chunk[0]->position points to pages[0] at an offset of 0
+ * - pages[] is not physically or virtually contigous and consists of
+ *   PAGE_SIZE elements.
+ *
+ * Output:
+ * - sge array pointing into pages[] array.
+ * - chunk_sge array specifying sge index and count for each
+ *   chunk in the read list
+ *
+ */
+static int rdma_rcl_to_sge(struct svcxprt_rdma *xprt,
+			   struct svc_rqst *rqstp,
+			   struct svc_rdma_op_ctxt *head,
+			   struct rpcrdma_msg *rmsgp,
+			   struct ib_sge *sge,
+			   struct chunk_sge *ch_sge_ary,
+			   int ch_count,
+			   int byte_count)
+{
+	int sge_no;
+	int sge_bytes;
+	int page_off;
+	int page_no;
+	int ch_bytes;
+	int ch_no;
+	struct rpcrdma_read_chunk *ch;
+
+	sge_no = 0;
+	page_no = 0;
+	page_off = 0;
+	ch = (struct rpcrdma_read_chunk *)&rmsgp->rm_body.rm_chunks[0];
+	ch_no = 0;
+	ch_bytes = ch->rc_target.rs_length;
+	head->arg.head[0] = rqstp->rq_arg.head[0];
+	head->arg.tail[0] = rqstp->rq_arg.tail[0];
+	head->arg.pages = &head->pages[head->count];
+	head->sge[0].length = head->count; /* save count of hdr pages */
+	head->arg.page_base = 0;
+	head->arg.page_len = ch_bytes;
+	head->arg.len = rqstp->rq_arg.len + ch_bytes;
+	head->arg.buflen = rqstp->rq_arg.buflen + ch_bytes;
+	head->count++;
+	ch_sge_ary[0].start = 0;
+	while (byte_count) {
+		sge_bytes = min_t(int, PAGE_SIZE-page_off, ch_bytes);
+		sge[sge_no].addr =
+			ib_dma_map_page(xprt->sc_cm_id->device,
+					rqstp->rq_arg.pages[page_no],
+					page_off, sge_bytes,
+					DMA_FROM_DEVICE);
+		sge[sge_no].length = sge_bytes;
+		sge[sge_no].lkey = xprt->sc_phys_mr->lkey;
+		/*
+		 * Don't bump head->count here because the same page
+		 * may be used by multiple SGE.
+		 */
+		head->arg.pages[page_no] = rqstp->rq_arg.pages[page_no];
+		rqstp->rq_respages = &rqstp->rq_arg.pages[page_no+1];
+
+		byte_count -= sge_bytes;
+		ch_bytes -= sge_bytes;
+		sge_no++;
+		/*
+		 * If all bytes for this chunk have been mapped to an
+		 * SGE, move to the next SGE
+		 */
+		if (ch_bytes == 0) {
+			ch_sge_ary[ch_no].count =
+				sge_no - ch_sge_ary[ch_no].start;
+			ch_no++;
+			ch++;
+			ch_sge_ary[ch_no].start = sge_no;
+			ch_bytes = ch->rc_target.rs_length;
+			/* If bytes remaining account for next chunk */
+			if (byte_count) {
+				head->arg.page_len += ch_bytes;
+				head->arg.len += ch_bytes;
+				head->arg.buflen += ch_bytes;
+			}
+		}
+		/*
+		 * If this SGE consumed all of the page, move to the
+		 * next page
+		 */
+		if ((sge_bytes + page_off) == PAGE_SIZE) {
+			page_no++;
+			page_off = 0;
+			/*
+			 * If there are still bytes left to map, bump
+			 * the page count
+			 */
+			if (byte_count)
+				head->count++;
+		} else
+			page_off += sge_bytes;
+	}
+	BUG_ON(byte_count != 0);
+	return sge_no;
+}
+
+static void rdma_set_ctxt_sge(struct svc_rdma_op_ctxt *ctxt,
+			      struct ib_sge *sge,
+			      u64 *sgl_offset,
+			      int count)
+{
+	int i;
+
+	ctxt->count = count;
+	for (i = 0; i < count; i++) {
+		ctxt->sge[i].addr = sge[i].addr;
+		ctxt->sge[i].length = sge[i].length;
+		*sgl_offset = *sgl_offset + sge[i].length;
+	}
+}
+
+static int rdma_read_max_sge(struct svcxprt_rdma *xprt, int sge_count)
+{
+#ifdef RDMA_TRANSPORT_IWARP
+	if ((RDMA_TRANSPORT_IWARP ==
+	     rdma_node_get_transport(xprt->sc_cm_id->
+				     device->node_type))
+	    && sge_count > 1)
+		return 1;
+	else
+#endif
+		return min_t(int, sge_count, xprt->sc_max_sge);
+}
+
+/*
+ * Use RDMA_READ to read data from the advertised client buffer into the
+ * XDR stream starting at rq_arg.head[0].iov_base.
+ * Each chunk in the array
+ * contains the following fields:
+ * discrim      - '1', This isn't used for data placement
+ * position     - The xdr stream offset (the same for every chunk)
+ * handle       - RMR for client memory region
+ * length       - data transfer length
+ * offset       - 64 bit tagged offset in remote memory region
+ *
+ * On our side, we need to read into a pagelist. The first page immediately
+ * follows the RPC header.
+ *
+ * This function returns 1 to indicate success. The data is not yet in
+ * the pagelist and therefore the RPC request must be deferred. The
+ * I/O completion will enqueue the transport again and
+ * svc_rdma_recvfrom will complete the request.
+ *
+ * NOTE: The ctxt must not be touched after the last WR has been posted
+ * because the I/O completion processing may occur on another
+ * processor and free / modify the context. Ne touche pas!
+ */
+static int rdma_read_xdr(struct svcxprt_rdma *xprt,
+			 struct rpcrdma_msg *rmsgp,
+			 struct svc_rqst *rqstp,
+			 struct svc_rdma_op_ctxt *hdr_ctxt)
+{
+	struct ib_send_wr read_wr;
+	int err = 0;
+	int ch_no;
+	struct ib_sge *sge;
+	int ch_count;
+	int byte_count;
+	int sge_count;
+	u64 sgl_offset;
+	struct rpcrdma_read_chunk *ch;
+	struct svc_rdma_op_ctxt *ctxt = NULL;
+	struct svc_rdma_op_ctxt *head;
+	struct svc_rdma_op_ctxt *tmp_sge_ctxt;
+	struct svc_rdma_op_ctxt *tmp_ch_ctxt;
+	struct chunk_sge *ch_sge_ary;
+
+	/* If no read list is present, return 0 */
+	ch = svc_rdma_get_read_chunk(rmsgp);
+	if (!ch)
+		return 0;
+
+	/* Allocate temporary contexts to keep SGE */
+	BUG_ON(sizeof(struct ib_sge) < sizeof(struct chunk_sge));
+	tmp_sge_ctxt = svc_rdma_get_context(xprt);
+	sge = tmp_sge_ctxt->sge;
+	tmp_ch_ctxt = svc_rdma_get_context(xprt);
+	ch_sge_ary = (struct chunk_sge *)tmp_ch_ctxt->sge;
+
+	svc_rdma_rcl_chunk_counts(ch, &ch_count, &byte_count);
+	sge_count = rdma_rcl_to_sge(xprt, rqstp, hdr_ctxt, rmsgp,
+				    sge, ch_sge_ary,
+				    ch_count, byte_count);
+	head = svc_rdma_get_context(xprt);
+	sgl_offset = 0;
+	ch_no = 0;
+
+	for (ch = (struct rpcrdma_read_chunk *)&rmsgp->rm_body.rm_chunks[0];
+	     ch->rc_discrim != 0; ch++, ch_no++) {
+next_sge:
+		if (!ctxt)
+			ctxt = head;
+		else {
+			ctxt->next = svc_rdma_get_context(xprt);
+			ctxt = ctxt->next;
+		}
+		ctxt->next = NULL;
+		ctxt->direction = DMA_FROM_DEVICE;
+		clear_bit(RDMACTXT_F_READ_DONE, &ctxt->flags);
+		clear_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags);
+		if ((ch+1)->rc_discrim == 0) {
+			/*
+			 * Checked in sq_cq_reap to see if we need to
+			 * be enqueued
+			 */
+			set_bit(RDMACTXT_F_LAST_CTXT, &ctxt->flags);
+			ctxt->next = hdr_ctxt;
+			hdr_ctxt->next = head;
+		}
+
+		/* Prepare READ WR */
+		memset(&read_wr, 0, sizeof read_wr);
+		ctxt->wr_op = IB_WR_RDMA_READ;
+		read_wr.wr_id = (unsigned long)ctxt;
+		read_wr.opcode = IB_WR_RDMA_READ;
+		read_wr.send_flags = IB_SEND_SIGNALED;
+		read_wr.wr.rdma.rkey = ch->rc_target.rs_handle;
+		read_wr.wr.rdma.remote_addr =
+			get_unaligned(&(ch->rc_target.rs_offset)) +
+			sgl_offset;
+		read_wr.sg_list = &sge[ch_sge_ary[ch_no].start];
+		read_wr.num_sge =
+			rdma_read_max_sge(xprt, ch_sge_ary[ch_no].count);
+		rdma_set_ctxt_sge(ctxt, &sge[ch_sge_ary[ch_no].start],
+				  &sgl_offset,
+				  read_wr.num_sge);
+
+		/* Post the read */
+		err = svc_rdma_send(xprt, &read_wr);
+		if (err) {
+			printk(KERN_ERR "svcrdma: Error posting send = %d\n",
+			       err);
+			/*
+			 * Break the circular list so free knows when
+			 * to stop if the error happened to occur on
+			 * the last read
+			 */
+			ctxt->next = NULL;
+			goto out;
+		}
+		atomic_inc(&rdma_stat_read);
+
+		if (read_wr.num_sge < ch_sge_ary[ch_no].count) {
+			ch_sge_ary[ch_no].count -= read_wr.num_sge;
+			ch_sge_ary[ch_no].start += read_wr.num_sge;
+			goto next_sge;
+		}
+		sgl_offset = 0;
+		err = 0;
+	}
+
+ out:
+	svc_rdma_put_context(tmp_sge_ctxt, 0);
+	svc_rdma_put_context(tmp_ch_ctxt, 0);
+
+	/* Detach arg pages. svc_recv will replenish them */
+	for (ch_no = 0; &rqstp->rq_pages[ch_no] < rqstp->rq_respages; ch_no++)
+		rqstp->rq_pages[ch_no] = NULL;
+
+	/*
+	 * Detach res pages. svc_release must see a resused count of
+	 * zero or it will attempt to put them.
+	 */
+	while (rqstp->rq_resused)
+		rqstp->rq_respages[--rqstp->rq_resused] = NULL;
+
+	if (err) {
+		printk(KERN_ERR "svcrdma : RDMA_READ error = %d\n", err);
+		set_bit(XPT_CLOSE, &xprt->sc_xprt.xpt_flags);
+		/* Free the linked list of read contexts */
+		while (head != NULL) {
+			ctxt = head->next;
+			svc_rdma_put_context(head, 1);
+			head = ctxt;
+		}
+		return 0;
+	}
+
+	return 1;
+}
+
+static int rdma_read_complete(struct svc_rqst *rqstp,
+			      struct svc_rdma_op_ctxt *data)
+{
+	struct svc_rdma_op_ctxt *head = data->next;
+	int page_no;
+	int ret;
+
+	BUG_ON(!head);
+
+	/* Copy RPC pages */
+	for (page_no = 0; page_no < head->count; page_no++) {
+		put_page(rqstp->rq_pages[page_no]);
+		rqstp->rq_pages[page_no] = head->pages[page_no];
+	}
+	/* Point rq_arg.pages past header */
+	rqstp->rq_arg.pages = &rqstp->rq_pages[head->sge[0].length];
+	rqstp->rq_arg.page_len = head->arg.page_len;
+	rqstp->rq_arg.page_base = head->arg.page_base;
+
+	/* rq_respages starts after the last arg page */
+	rqstp->rq_respages = &rqstp->rq_arg.pages[page_no];
+	rqstp->rq_resused = 0;
+
+	/* Rebuild rq_arg head and tail. */
+	rqstp->rq_arg.head[0] = head->arg.head[0];
+	rqstp->rq_arg.tail[0] = head->arg.tail[0];
+	rqstp->rq_arg.len = head->arg.len;
+	rqstp->rq_arg.buflen = head->arg.buflen;
+
+	/* XXX: What should this be? */
+	rqstp->rq_prot = IPPROTO_MAX;
+
+	/*
+	 * Free the contexts we used to build the RDMA_READ. We have
+	 * to be careful here because the context list uses the same
+	 * next pointer used to chain the contexts associated with the
+	 * RDMA_READ
+	 */
+	data->next = NULL;	/* terminate circular list */
+	do {
+		data = head->next;
+		svc_rdma_put_context(head, 0);
+		head = data;
+	} while (head != NULL);
+
+	ret = rqstp->rq_arg.head[0].iov_len
+		+ rqstp->rq_arg.page_len
+		+ rqstp->rq_arg.tail[0].iov_len;
+	dprintk("svcrdma: deferred read ret=%d, rq_arg.len =%d, "
+		"rq_arg.head[0].iov_base=%p, rq_arg.head[0].iov_len = %zd\n",
+		ret, rqstp->rq_arg.len,	rqstp->rq_arg.head[0].iov_base,
+		rqstp->rq_arg.head[0].iov_len);
+
+	/* Indicate that we've consumed an RQ credit */
+	rqstp->rq_xprt_ctxt = rqstp->rq_xprt;
+	svc_xprt_received(rqstp->rq_xprt);
+	return ret;
+}
+
+/*
+ * Set up the rqstp thread context to point to the RQ buffer. If
+ * necessary, pull additional data from the client with an RDMA_READ
+ * request.
+ */
+int svc_rdma_recvfrom(struct svc_rqst *rqstp)
+{
+	struct svc_xprt *xprt = rqstp->rq_xprt;
+	struct svcxprt_rdma *rdma_xprt =
+		container_of(xprt, struct svcxprt_rdma, sc_xprt);
+	struct svc_rdma_op_ctxt *ctxt = NULL;
+	struct rpcrdma_msg *rmsgp;
+	int ret = 0;
+	int len;
+
+	dprintk("svcrdma: rqstp=%p\n", rqstp);
+
+	/*
+	 * The rq_xprt_ctxt indicates if we've consumed an RQ credit
+	 * or not. It is used in the rdma xpo_release_rqst function to
+	 * determine whether or not to return an RQ WQE to the RQ.
+	 */
+	rqstp->rq_xprt_ctxt = NULL;
+
+	spin_lock_bh(&rdma_xprt->sc_read_complete_lock);
+	if (!list_empty(&rdma_xprt->sc_read_complete_q)) {
+		ctxt = list_entry(rdma_xprt->sc_read_complete_q.next,
+				  struct svc_rdma_op_ctxt,
+				  dto_q);
+		list_del_init(&ctxt->dto_q);
+	}
+	spin_unlock_bh(&rdma_xprt->sc_read_complete_lock);
+	if (ctxt)
+		return rdma_read_complete(rqstp, ctxt);
+
+	spin_lock_bh(&rdma_xprt->sc_rq_dto_lock);
+	if (!list_empty(&rdma_xprt->sc_rq_dto_q)) {
+		ctxt = list_entry(rdma_xprt->sc_rq_dto_q.next,
+				  struct svc_rdma_op_ctxt,
+				  dto_q);
+		list_del_init(&ctxt->dto_q);
+	} else {
+		atomic_inc(&rdma_stat_rq_starve);
+		clear_bit(XPT_DATA, &xprt->xpt_flags);
+		ctxt = NULL;
+	}
+	spin_unlock_bh(&rdma_xprt->sc_rq_dto_lock);
+	if (!ctxt) {
+		/* This is the EAGAIN path. The svc_recv routine will
+		 * return -EAGAIN, the nfsd thread will go to call into
+		 * svc_recv again and we shouldn't be on the active
+		 * transport list
+		 */
+		if (test_bit(XPT_CLOSE, &xprt->xpt_flags))
+			goto close_out;
+
+		BUG_ON(ret);
+		goto out;
+	}
+	dprintk("svcrdma: processing ctxt=%p on xprt=%p, rqstp=%p, status=%d\n",
+		ctxt, rdma_xprt, rqstp, ctxt->wc_status);
+	BUG_ON(ctxt->wc_status != IB_WC_SUCCESS);
+	atomic_inc(&rdma_stat_recv);
+
+	/* Build up the XDR from the receive buffers. */
+	rdma_build_arg_xdr(rqstp, ctxt, ctxt->byte_len);
+
+	/* Decode the RDMA header. */
+	len = svc_rdma_xdr_decode_req(&rmsgp, rqstp);
+	rqstp->rq_xprt_hlen = len;
+
+	/* If the request is invalid, reply with an error */
+	if (len < 0) {
+		if (len == -ENOSYS)
+			(void)svc_rdma_send_error(rdma_xprt, rmsgp, ERR_VERS);
+		goto close_out;
+	}
+
+	/* Read read-list data. If we would need to wait, defer
+	 * it. Not that in this case, we don't return the RQ credit
+	 * until after the read completes.
+	 */
+	if (rdma_read_xdr(rdma_xprt, rmsgp, rqstp, ctxt)) {
+		svc_xprt_received(xprt);
+		return 0;
+	}
+
+	/* Indicate we've consumed an RQ credit */
+	rqstp->rq_xprt_ctxt = rqstp->rq_xprt;
+
+	ret = rqstp->rq_arg.head[0].iov_len
+		+ rqstp->rq_arg.page_len
+		+ rqstp->rq_arg.tail[0].iov_len;
+	svc_rdma_put_context(ctxt, 0);
+ out:
+	dprintk("svcrdma: ret = %d, rq_arg.len =%d, "
+		"rq_arg.head[0].iov_base=%p, rq_arg.head[0].iov_len = %zd\n",
+		ret, rqstp->rq_arg.len,
+		rqstp->rq_arg.head[0].iov_base,
+		rqstp->rq_arg.head[0].iov_len);
+	rqstp->rq_prot = IPPROTO_MAX;
+	svc_xprt_copy_addrs(rqstp, xprt);
+	svc_xprt_received(xprt);
+	return ret;
+
+ close_out:
+	if (ctxt) {
+		svc_rdma_put_context(ctxt, 1);
+		/* Indicate we've consumed an RQ credit */
+		rqstp->rq_xprt_ctxt = rqstp->rq_xprt;
+	}
+	dprintk("svcrdma: transport %p is closing\n", xprt);
+	/*
+	 * Set the close bit and enqueue it. svc_recv will see the
+	 * close bit and call svc_xprt_delete
+	 */
+	set_bit(XPT_CLOSE, &xprt->xpt_flags);
+	svc_xprt_received(xprt);
+	return 0;
+}

net/sunrpc/xprtrdma/svc_rdma_sendto.c

@@ -0,0 +1,520 @@
+/*
+ * Copyright (c) 2005-2006 Network Appliance, Inc. All rights reserved.
+ *
+ * This software is available to you under a choice of one of two
+ * licenses.  You may choose to be licensed under the terms of the GNU
+ * General Public License (GPL) Version 2, available from the file
+ * COPYING in the main directory of this source tree, or the BSD-type
+ * license below:
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ *
+ *      Redistributions of source code must retain the above copyright
+ *      notice, this list of conditions and the following disclaimer.
+ *
+ *      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.
+ *
+ *      Neither the name of the Network Appliance, Inc. 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 BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
+ * "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 COPYRIGHT
+ * OWNER 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.
+ *
+ * Author: Tom Tucker <tom@opengridcomputing.com>
+ */
+
+#include <linux/sunrpc/debug.h>
+#include <linux/sunrpc/rpc_rdma.h>
+#include <linux/spinlock.h>
+#include <asm/unaligned.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/rdma_cm.h>
+#include <linux/sunrpc/svc_rdma.h>
+
+#define RPCDBG_FACILITY	RPCDBG_SVCXPRT
+
+/* Encode an XDR as an array of IB SGE
+ *
+ * Assumptions:
+ * - head[0] is physically contiguous.
+ * - tail[0] is physically contiguous.
+ * - pages[] is not physically or virtually contigous and consists of
+ *   PAGE_SIZE elements.
+ *
+ * Output:
+ * SGE[0]              reserved for RCPRDMA header
+ * SGE[1]              data from xdr->head[]
+ * SGE[2..sge_count-2] data from xdr->pages[]
+ * SGE[sge_count-1]    data from xdr->tail.
+ *
+ */
+static struct ib_sge *xdr_to_sge(struct svcxprt_rdma *xprt,
+				 struct xdr_buf *xdr,
+				 struct ib_sge *sge,
+				 int *sge_count)
+{
+	/* Max we need is the length of the XDR / pagesize + one for
+	 * head + one for tail + one for RPCRDMA header
+	 */
+	int sge_max = (xdr->len+PAGE_SIZE-1) / PAGE_SIZE + 3;
+	int sge_no;
+	u32 byte_count = xdr->len;
+	u32 sge_bytes;
+	u32 page_bytes;
+	int page_off;
+	int page_no;
+
+	/* Skip the first sge, this is for the RPCRDMA header */
+	sge_no = 1;
+
+	/* Head SGE */
+	sge[sge_no].addr = ib_dma_map_single(xprt->sc_cm_id->device,
+					     xdr->head[0].iov_base,
+					     xdr->head[0].iov_len,
+					     DMA_TO_DEVICE);
+	sge_bytes = min_t(u32, byte_count, xdr->head[0].iov_len);
+	byte_count -= sge_bytes;
+	sge[sge_no].length = sge_bytes;
+	sge[sge_no].lkey = xprt->sc_phys_mr->lkey;
+	sge_no++;
+
+	/* pages SGE */
+	page_no = 0;
+	page_bytes = xdr->page_len;
+	page_off = xdr->page_base;
+	while (byte_count && page_bytes) {
+		sge_bytes = min_t(u32, byte_count, (PAGE_SIZE-page_off));
+		sge[sge_no].addr =
+			ib_dma_map_page(xprt->sc_cm_id->device,
+					xdr->pages[page_no], page_off,
+					sge_bytes, DMA_TO_DEVICE);
+		sge_bytes = min(sge_bytes, page_bytes);
+		byte_count -= sge_bytes;
+		page_bytes -= sge_bytes;
+		sge[sge_no].length = sge_bytes;
+		sge[sge_no].lkey = xprt->sc_phys_mr->lkey;
+
+		sge_no++;
+		page_no++;
+		page_off = 0; /* reset for next time through loop */
+	}
+
+	/* Tail SGE */
+	if (byte_count && xdr->tail[0].iov_len) {
+		sge[sge_no].addr =
+			ib_dma_map_single(xprt->sc_cm_id->device,
+					  xdr->tail[0].iov_base,
+					  xdr->tail[0].iov_len,
+					  DMA_TO_DEVICE);
+		sge_bytes = min_t(u32, byte_count, xdr->tail[0].iov_len);
+		byte_count -= sge_bytes;
+		sge[sge_no].length = sge_bytes;
+		sge[sge_no].lkey = xprt->sc_phys_mr->lkey;
+		sge_no++;
+	}
+
+	BUG_ON(sge_no > sge_max);
+	BUG_ON(byte_count != 0);
+
+	*sge_count = sge_no;
+	return sge;
+}
+
+
+/* Assumptions:
+ * - The specified write_len can be represented in sc_max_sge * PAGE_SIZE
+ */
+static int send_write(struct svcxprt_rdma *xprt, struct svc_rqst *rqstp,
+		      u32 rmr, u64 to,
+		      u32 xdr_off, int write_len,
+		      struct ib_sge *xdr_sge, int sge_count)
+{
+	struct svc_rdma_op_ctxt *tmp_sge_ctxt;
+	struct ib_send_wr write_wr;
+	struct ib_sge *sge;
+	int xdr_sge_no;
+	int sge_no;
+	int sge_bytes;
+	int sge_off;
+	int bc;
+	struct svc_rdma_op_ctxt *ctxt;
+	int ret = 0;
+
+	BUG_ON(sge_count >= 32);
+	dprintk("svcrdma: RDMA_WRITE rmr=%x, to=%llx, xdr_off=%d, "
+		"write_len=%d, xdr_sge=%p, sge_count=%d\n",
+		rmr, to, xdr_off, write_len, xdr_sge, sge_count);
+
+	ctxt = svc_rdma_get_context(xprt);
+	ctxt->count = 0;
+	tmp_sge_ctxt = svc_rdma_get_context(xprt);
+	sge = tmp_sge_ctxt->sge;
+
+	/* Find the SGE associated with xdr_off */
+	for (bc = xdr_off, xdr_sge_no = 1; bc && xdr_sge_no < sge_count;
+	     xdr_sge_no++) {
+		if (xdr_sge[xdr_sge_no].length > bc)
+			break;
+		bc -= xdr_sge[xdr_sge_no].length;
+	}
+
+	sge_off = bc;
+	bc = write_len;
+	sge_no = 0;
+
+	/* Copy the remaining SGE */
+	while (bc != 0 && xdr_sge_no < sge_count) {
+		sge[sge_no].addr = xdr_sge[xdr_sge_no].addr + sge_off;
+		sge[sge_no].lkey = xdr_sge[xdr_sge_no].lkey;
+		sge_bytes = min((size_t)bc,
+				(size_t)(xdr_sge[xdr_sge_no].length-sge_off));
+		sge[sge_no].length = sge_bytes;
+
+		sge_off = 0;
+		sge_no++;
+		xdr_sge_no++;
+		bc -= sge_bytes;
+	}
+
+	BUG_ON(bc != 0);
+	BUG_ON(xdr_sge_no > sge_count);
+
+	/* Prepare WRITE WR */
+	memset(&write_wr, 0, sizeof write_wr);
+	ctxt->wr_op = IB_WR_RDMA_WRITE;
+	write_wr.wr_id = (unsigned long)ctxt;
+	write_wr.sg_list = &sge[0];
+	write_wr.num_sge = sge_no;
+	write_wr.opcode = IB_WR_RDMA_WRITE;
+	write_wr.send_flags = IB_SEND_SIGNALED;
+	write_wr.wr.rdma.rkey = rmr;
+	write_wr.wr.rdma.remote_addr = to;
+
+	/* Post It */
+	atomic_inc(&rdma_stat_write);
+	if (svc_rdma_send(xprt, &write_wr)) {
+		svc_rdma_put_context(ctxt, 1);
+		/* Fatal error, close transport */
+		ret = -EIO;
+	}
+	svc_rdma_put_context(tmp_sge_ctxt, 0);
+	return ret;
+}
+
+static int send_write_chunks(struct svcxprt_rdma *xprt,
+			     struct rpcrdma_msg *rdma_argp,
+			     struct rpcrdma_msg *rdma_resp,
+			     struct svc_rqst *rqstp,
+			     struct ib_sge *sge,
+			     int sge_count)
+{
+	u32 xfer_len = rqstp->rq_res.page_len + rqstp->rq_res.tail[0].iov_len;
+	int write_len;
+	int max_write;
+	u32 xdr_off;
+	int chunk_off;
+	int chunk_no;
+	struct rpcrdma_write_array *arg_ary;
+	struct rpcrdma_write_array *res_ary;
+	int ret;
+
+	arg_ary = svc_rdma_get_write_array(rdma_argp);
+	if (!arg_ary)
+		return 0;
+	res_ary = (struct rpcrdma_write_array *)
+		&rdma_resp->rm_body.rm_chunks[1];
+
+	max_write = xprt->sc_max_sge * PAGE_SIZE;
+
+	/* Write chunks start at the pagelist */
+	for (xdr_off = rqstp->rq_res.head[0].iov_len, chunk_no = 0;
+	     xfer_len && chunk_no < arg_ary->wc_nchunks;
+	     chunk_no++) {
+		struct rpcrdma_segment *arg_ch;
+		u64 rs_offset;
+
+		arg_ch = &arg_ary->wc_array[chunk_no].wc_target;
+		write_len = min(xfer_len, arg_ch->rs_length);
+
+		/* Prepare the response chunk given the length actually
+		 * written */
+		rs_offset = get_unaligned(&(arg_ch->rs_offset));
+		svc_rdma_xdr_encode_array_chunk(res_ary, chunk_no,
+					    arg_ch->rs_handle,
+					    rs_offset,
+					    write_len);
+		chunk_off = 0;
+		while (write_len) {
+			int this_write;
+			this_write = min(write_len, max_write);
+			ret = send_write(xprt, rqstp,
+					 arg_ch->rs_handle,
+					 rs_offset + chunk_off,
+					 xdr_off,
+					 this_write,
+					 sge,
+					 sge_count);
+			if (ret) {
+				dprintk("svcrdma: RDMA_WRITE failed, ret=%d\n",
+					ret);
+				return -EIO;
+			}
+			chunk_off += this_write;
+			xdr_off += this_write;
+			xfer_len -= this_write;
+			write_len -= this_write;
+		}
+	}
+	/* Update the req with the number of chunks actually used */
+	svc_rdma_xdr_encode_write_list(rdma_resp, chunk_no);
+
+	return rqstp->rq_res.page_len + rqstp->rq_res.tail[0].iov_len;
+}
+
+static int send_reply_chunks(struct svcxprt_rdma *xprt,
+			     struct rpcrdma_msg *rdma_argp,
+			     struct rpcrdma_msg *rdma_resp,
+			     struct svc_rqst *rqstp,
+			     struct ib_sge *sge,
+			     int sge_count)
+{
+	u32 xfer_len = rqstp->rq_res.len;
+	int write_len;
+	int max_write;
+	u32 xdr_off;
+	int chunk_no;
+	int chunk_off;
+	struct rpcrdma_segment *ch;
+	struct rpcrdma_write_array *arg_ary;
+	struct rpcrdma_write_array *res_ary;
+	int ret;
+
+	arg_ary = svc_rdma_get_reply_array(rdma_argp);
+	if (!arg_ary)
+		return 0;
+	/* XXX: need to fix when reply lists occur with read-list and or
+	 * write-list */
+	res_ary = (struct rpcrdma_write_array *)
+		&rdma_resp->rm_body.rm_chunks[2];
+
+	max_write = xprt->sc_max_sge * PAGE_SIZE;
+
+	/* xdr offset starts at RPC message */
+	for (xdr_off = 0, chunk_no = 0;
+	     xfer_len && chunk_no < arg_ary->wc_nchunks;
+	     chunk_no++) {
+		u64 rs_offset;
+		ch = &arg_ary->wc_array[chunk_no].wc_target;
+		write_len = min(xfer_len, ch->rs_length);
+
+
+		/* Prepare the reply chunk given the length actually
+		 * written */
+		rs_offset = get_unaligned(&(ch->rs_offset));
+		svc_rdma_xdr_encode_array_chunk(res_ary, chunk_no,
+					    ch->rs_handle, rs_offset,
+					    write_len);
+		chunk_off = 0;
+		while (write_len) {
+			int this_write;
+
+			this_write = min(write_len, max_write);
+			ret = send_write(xprt, rqstp,
+					 ch->rs_handle,
+					 rs_offset + chunk_off,
+					 xdr_off,
+					 this_write,
+					 sge,
+					 sge_count);
+			if (ret) {
+				dprintk("svcrdma: RDMA_WRITE failed, ret=%d\n",
+					ret);
+				return -EIO;
+			}
+			chunk_off += this_write;
+			xdr_off += this_write;
+			xfer_len -= this_write;
+			write_len -= this_write;
+		}
+	}
+	/* Update the req with the number of chunks actually used */
+	svc_rdma_xdr_encode_reply_array(res_ary, chunk_no);
+
+	return rqstp->rq_res.len;
+}
+
+/* This function prepares the portion of the RPCRDMA message to be
+ * sent in the RDMA_SEND. This function is called after data sent via
+ * RDMA has already been transmitted. There are three cases:
+ * - The RPCRDMA header, RPC header, and payload are all sent in a
+ *   single RDMA_SEND. This is the "inline" case.
+ * - The RPCRDMA header and some portion of the RPC header and data
+ *   are sent via this RDMA_SEND and another portion of the data is
+ *   sent via RDMA.
+ * - The RPCRDMA header [NOMSG] is sent in this RDMA_SEND and the RPC
+ *   header and data are all transmitted via RDMA.
+ * In all three cases, this function prepares the RPCRDMA header in
+ * sge[0], the 'type' parameter indicates the type to place in the
+ * RPCRDMA header, and the 'byte_count' field indicates how much of
+ * the XDR to include in this RDMA_SEND.
+ */
+static int send_reply(struct svcxprt_rdma *rdma,
+		      struct svc_rqst *rqstp,
+		      struct page *page,
+		      struct rpcrdma_msg *rdma_resp,
+		      struct svc_rdma_op_ctxt *ctxt,
+		      int sge_count,
+		      int byte_count)
+{
+	struct ib_send_wr send_wr;
+	int sge_no;
+	int sge_bytes;
+	int page_no;
+	int ret;
+
+	/* Prepare the context */
+	ctxt->pages[0] = page;
+	ctxt->count = 1;
+
+	/* Prepare the SGE for the RPCRDMA Header */
+	ctxt->sge[0].addr =
+		ib_dma_map_page(rdma->sc_cm_id->device,
+				page, 0, PAGE_SIZE, DMA_TO_DEVICE);
+	ctxt->direction = DMA_TO_DEVICE;
+	ctxt->sge[0].length = svc_rdma_xdr_get_reply_hdr_len(rdma_resp);
+	ctxt->sge[0].lkey = rdma->sc_phys_mr->lkey;
+
+	/* Determine how many of our SGE are to be transmitted */
+	for (sge_no = 1; byte_count && sge_no < sge_count; sge_no++) {
+		sge_bytes = min((size_t)ctxt->sge[sge_no].length,
+				(size_t)byte_count);
+		byte_count -= sge_bytes;
+	}
+	BUG_ON(byte_count != 0);
+
+	/* Save all respages in the ctxt and remove them from the
+	 * respages array. They are our pages until the I/O
+	 * completes.
+	 */
+	for (page_no = 0; page_no < rqstp->rq_resused; page_no++) {
+		ctxt->pages[page_no+1] = rqstp->rq_respages[page_no];
+		ctxt->count++;
+		rqstp->rq_respages[page_no] = NULL;
+	}
+
+	BUG_ON(sge_no > rdma->sc_max_sge);
+	memset(&send_wr, 0, sizeof send_wr);
+	ctxt->wr_op = IB_WR_SEND;
+	send_wr.wr_id = (unsigned long)ctxt;
+	send_wr.sg_list = ctxt->sge;
+	send_wr.num_sge = sge_no;
+	send_wr.opcode = IB_WR_SEND;
+	send_wr.send_flags =  IB_SEND_SIGNALED;
+
+	ret = svc_rdma_send(rdma, &send_wr);
+	if (ret)
+		svc_rdma_put_context(ctxt, 1);
+
+	return ret;
+}
+
+void svc_rdma_prep_reply_hdr(struct svc_rqst *rqstp)
+{
+}
+
+/*
+ * Return the start of an xdr buffer.
+ */
+static void *xdr_start(struct xdr_buf *xdr)
+{
+	return xdr->head[0].iov_base -
+		(xdr->len -
+		 xdr->page_len -
+		 xdr->tail[0].iov_len -
+		 xdr->head[0].iov_len);
+}
+
+int svc_rdma_sendto(struct svc_rqst *rqstp)
+{
+	struct svc_xprt *xprt = rqstp->rq_xprt;
+	struct svcxprt_rdma *rdma =
+		container_of(xprt, struct svcxprt_rdma, sc_xprt);
+	struct rpcrdma_msg *rdma_argp;
+	struct rpcrdma_msg *rdma_resp;
+	struct rpcrdma_write_array *reply_ary;
+	enum rpcrdma_proc reply_type;
+	int ret;
+	int inline_bytes;
+	struct ib_sge *sge;
+	int sge_count = 0;
+	struct page *res_page;
+	struct svc_rdma_op_ctxt *ctxt;
+
+	dprintk("svcrdma: sending response for rqstp=%p\n", rqstp);
+
+	/* Get the RDMA request header. */
+	rdma_argp = xdr_start(&rqstp->rq_arg);
+
+	/* Build an SGE for the XDR */
+	ctxt = svc_rdma_get_context(rdma);
+	ctxt->direction = DMA_TO_DEVICE;
+	sge = xdr_to_sge(rdma, &rqstp->rq_res, ctxt->sge, &sge_count);
+
+	inline_bytes = rqstp->rq_res.len;
+
+	/* Create the RDMA response header */
+	res_page = svc_rdma_get_page();
+	rdma_resp = page_address(res_page);
+	reply_ary = svc_rdma_get_reply_array(rdma_argp);
+	if (reply_ary)
+		reply_type = RDMA_NOMSG;
+	else
+		reply_type = RDMA_MSG;
+	svc_rdma_xdr_encode_reply_header(rdma, rdma_argp,
+					 rdma_resp, reply_type);
+
+	/* Send any write-chunk data and build resp write-list */
+	ret = send_write_chunks(rdma, rdma_argp, rdma_resp,
+				rqstp, sge, sge_count);
+	if (ret < 0) {
+		printk(KERN_ERR "svcrdma: failed to send write chunks, rc=%d\n",
+		       ret);
+		goto error;
+	}
+	inline_bytes -= ret;
+
+	/* Send any reply-list data and update resp reply-list */
+	ret = send_reply_chunks(rdma, rdma_argp, rdma_resp,
+				rqstp, sge, sge_count);
+	if (ret < 0) {
+		printk(KERN_ERR "svcrdma: failed to send reply chunks, rc=%d\n",
+		       ret);
+		goto error;
+	}
+	inline_bytes -= ret;
+
+	ret = send_reply(rdma, rqstp, res_page, rdma_resp, ctxt, sge_count,
+			 inline_bytes);
+	dprintk("svcrdma: send_reply returns %d\n", ret);
+	return ret;
+ error:
+	svc_rdma_put_context(ctxt, 0);
+	put_page(res_page);
+	return ret;
+}