Merge http://nfsclient.bkbits.net/linux-2.5

into fys.uio.no:/home/linux/bitkeeper/nfsclient-2.5

fs/nfs/dir.c

@@ -557,7 +557,7 @@ static int nfs_lookup_revalidate(struct dentry * dentry, struct nameidata *nd)
 	/* Force a full look up iff the parent directory has changed */
 	if (nfs_check_verifier(dir, dentry)) {
 		if (nfs_lookup_verify_inode(inode, isopen))
-			goto out_bad;
+			goto out_zap_parent;
 		goto out_valid;
 	}
 
@@ -566,7 +566,7 @@ static int nfs_lookup_revalidate(struct dentry * dentry, struct nameidata *nd)
 		if (memcmp(NFS_FH(inode), &fhandle, sizeof(struct nfs_fh))!= 0)
 			goto out_bad;
 		if (nfs_lookup_verify_inode(inode, isopen))
-			goto out_bad;
+			goto out_zap_parent;
 		goto out_valid_renew;
 	}
 
@@ -587,6 +587,8 @@ static int nfs_lookup_revalidate(struct dentry * dentry, struct nameidata *nd)
 	unlock_kernel();
 	dput(parent);
 	return 1;
+out_zap_parent:
+	nfs_zap_caches(dir);
  out_bad:
 	NFS_CACHEINV(dir);
 	if (inode && S_ISDIR(inode->i_mode)) {
@@ -670,36 +672,29 @@ static struct dentry *nfs_lookup(struct inode *dir, struct dentry * dentry, stru
 	error = -ENOMEM;
 	dentry->d_op = &nfs_dentry_operations;
 
+	lock_kernel();
+
 	/* If we're doing an exclusive create, optimize away the lookup */
 	if (nfs_is_exclusive_create(dir, nd))
-		return NULL;
+		goto no_entry;
 
-	lock_kernel();
 	error = nfs_cached_lookup(dir, dentry, &fhandle, &fattr);
-	if (!error) {
-		error = -EACCES;
-		inode = nfs_fhget(dentry, &fhandle, &fattr);
-		if (inode) {
-			d_add(dentry, inode);
-			nfs_renew_times(dentry);
-			error = 0;
-		}
-		goto out_unlock;
-	}
-
-	error = NFS_PROTO(dir)->lookup(dir, &dentry->d_name, &fhandle, &fattr);
+	if (error != 0) {
+		error = NFS_PROTO(dir)->lookup(dir, &dentry->d_name,
+				&fhandle, &fattr);
 		if (error == -ENOENT)
 			goto no_entry;
-	if (!error) {
+		if (error != 0)
+			goto out_unlock;
+	}
 	error = -EACCES;
 	inode = nfs_fhget(dentry, &fhandle, &fattr);
-		if (inode) {
-	    no_entry:
-			d_add(dentry, inode);
+	if (!inode)
+		goto out_unlock;
+no_entry:
 	error = 0;
-		}
+	d_add(dentry, inode);
 	nfs_renew_times(dentry);
-	}
 out_unlock:
 	unlock_kernel();
 out:

fs/nfs/nfsroot.c

@@ -437,6 +437,7 @@ static int __init root_nfs_ports(void)
  */
 static int __init root_nfs_get_handle(void)
 {
+	struct nfs_fh fh;
 	struct sockaddr_in sin;
 	int status;
 	int protocol = (nfs_data.flags & NFS_MOUNT_TCP) ?
@@ -445,11 +446,14 @@ static int __init root_nfs_get_handle(void)
 					NFS_MNT3_VERSION : NFS_MNT_VERSION;
 
 	set_sockaddr(&sin, servaddr, mount_port);
-	status = nfsroot_mount(&sin, nfs_path, &nfs_data.root,
-							version, protocol);
+	status = nfsroot_mount(&sin, nfs_path, &fh, version, protocol);
 	if (status < 0)
 		printk(KERN_ERR "Root-NFS: Server returned error %d "
 				"while mounting %s\n", status, nfs_path);
+	else {
+		nfs_data.root.size = fh.size;
+		memcpy(nfs_data.root.data, fh.data, fh.size);
+	}
 
 	return status;
 }

include/linux/sunrpc/timer.h

@@ -15,7 +15,6 @@ struct rpc_rtt {
 	unsigned long timeo;	/* default timeout value */
 	unsigned long srtt[5];	/* smoothed round trip time << 3 */
 	unsigned long sdrtt[5];	/* smoothed medium deviation of RTT */
-	atomic_t  ntimeouts;	/* Global count of the number of timeouts */
 };
 
 
@@ -23,19 +22,4 @@ extern void rpc_init_rtt(struct rpc_rtt *rt, unsigned long timeo);
 extern void rpc_update_rtt(struct rpc_rtt *rt, unsigned timer, long m);
 extern unsigned long rpc_calc_rto(struct rpc_rtt *rt, unsigned timer);
 
-static inline void rpc_inc_timeo(struct rpc_rtt *rt)
-{
-	atomic_inc(&rt->ntimeouts);
-}
-
-static inline void rpc_clear_timeo(struct rpc_rtt *rt)
-{
-	atomic_set(&rt->ntimeouts, 0);
-}
-
-static inline int rpc_ntimeo(struct rpc_rtt *rt)
-{
-	return atomic_read(&rt->ntimeouts);
-}
-
 #endif /* _LINUX_SUNRPC_TIMER_H */

include/linux/sunrpc/xprt.h

@@ -98,6 +98,10 @@ struct rpc_rqst {
 
 	struct list_head	rq_list;
 
+	struct xdr_buf		rq_private_buf;		/* The receive buffer
+							 * used in the softirq.
+							 */
+
 	/*
 	 * For authentication (e.g. auth_des)
 	 */
@@ -111,7 +115,7 @@ struct rpc_rqst {
 
 	unsigned long		rq_xtime;	/* when transmitted */
 	int			rq_ntimeo;
-	int			rq_nresend;
+	int			rq_ntrans;
 };
 #define rq_svec			rq_snd_buf.head
 #define rq_slen			rq_snd_buf.len

net/sunrpc/clnt.c

@@ -659,7 +659,7 @@ call_transmit(struct rpc_task *task)
 	if (task->tk_status < 0)
 		return;
 	task->tk_status = xprt_prepare_transmit(task);
-	if (task->tk_status < 0)
+	if (task->tk_status != 0)
 		return;
 	/* Encode here so that rpcsec_gss can use correct sequence number. */
 	if (!task->tk_rqstp->rq_bytes_sent)
@@ -685,7 +685,7 @@ call_status(struct rpc_task *task)
 	struct rpc_rqst	*req = task->tk_rqstp;
 	int		status;
 
-	if (req->rq_received != 0)
+	if (req->rq_received > 0 && !req->rq_bytes_sent)
 		task->tk_status = req->rq_received;
 
 	dprintk("RPC: %4d call_status (status %d)\n", 
@@ -744,14 +744,14 @@ call_timeout(struct rpc_task *task)
 
 	dprintk("RPC: %4d call_timeout (major)\n", task->tk_pid);
 	if (clnt->cl_softrtry) {
-		if (clnt->cl_chatty && !task->tk_exit)
+		if (clnt->cl_chatty)
 			printk(KERN_NOTICE "%s: server %s not responding, timed out\n",
 				clnt->cl_protname, clnt->cl_server);
 		rpc_exit(task, -EIO);
 		return;
 	}
 
-	if (clnt->cl_chatty && !(task->tk_flags & RPC_CALL_MAJORSEEN) && rpc_ntimeo(&clnt->cl_rtt) > 7) {
+	if (clnt->cl_chatty && !(task->tk_flags & RPC_CALL_MAJORSEEN)) {
 		task->tk_flags |= RPC_CALL_MAJORSEEN;
 		printk(KERN_NOTICE "%s: server %s not responding, still trying\n",
 			clnt->cl_protname, clnt->cl_server);
@@ -787,19 +787,26 @@ call_decode(struct rpc_task *task)
 
 	if (task->tk_status < 12) {
 		if (!clnt->cl_softrtry) {
-			task->tk_action = call_transmit;
+			task->tk_action = call_bind;
 			clnt->cl_stats->rpcretrans++;
-		} else {
+			goto out_retry;
+		}
 		printk(KERN_WARNING "%s: too small RPC reply size (%d bytes)\n",
 			clnt->cl_protname, task->tk_status);
 		rpc_exit(task, -EIO);
-		}
 		return;
 	}
 
+	/* Check that the softirq receive buffer is valid */
+	WARN_ON(memcmp(&req->rq_rcv_buf, &req->rq_private_buf,
+				sizeof(req->rq_rcv_buf)) != 0);
+
 	/* Verify the RPC header */
-	if (!(p = call_verify(task)))
+	if (!(p = call_verify(task))) {
+		if (task->tk_action == NULL)
 			return;
+		goto out_retry;
+	}
 
 	/*
 	 * The following is an NFS-specific hack to cater for setuid
@@ -812,7 +819,7 @@ call_decode(struct rpc_task *task)
 			task->tk_flags ^= RPC_CALL_REALUID;
 			task->tk_action = call_bind;
 			task->tk_suid_retry--;
-			return;
+			goto out_retry;
 		}
 	}
 
@@ -822,6 +829,10 @@ call_decode(struct rpc_task *task)
 		task->tk_status = decode(req, p, task->tk_msg.rpc_resp);
 	dprintk("RPC: %4d call_decode result %d\n", task->tk_pid,
 					task->tk_status);
+	return;
+out_retry:
+	req->rq_received = 0;
+	task->tk_status = 0;
 }
 
 /*

net/sunrpc/timer.c

@@ -25,7 +25,7 @@
 
 #define RPC_RTO_MAX (60*HZ)
 #define RPC_RTO_INIT (HZ/5)
-#define RPC_RTO_MIN (2)
+#define RPC_RTO_MIN (HZ/10)
 
 void
 rpc_init_rtt(struct rpc_rtt *rt, unsigned long timeo)
@@ -41,8 +41,6 @@ rpc_init_rtt(struct rpc_rtt *rt, unsigned long timeo)
 		rt->srtt[i] = init;
 		rt->sdrtt[i] = RPC_RTO_INIT;
 	}
-
-	atomic_set(&rt->ntimeouts, 0);
 }
 
 /*
@@ -52,7 +50,7 @@ rpc_init_rtt(struct rpc_rtt *rt, unsigned long timeo)
 void
 rpc_update_rtt(struct rpc_rtt *rt, unsigned timer, long m)
 {
-	unsigned long *srtt, *sdrtt;
+	long *srtt, *sdrtt;
 
 	if (timer-- == 0)
 		return;
@@ -64,14 +62,14 @@ rpc_update_rtt(struct rpc_rtt *rt, unsigned timer, long m)
 	if (m == 0)
 		m = 1L;
 
-	srtt = &rt->srtt[timer];
+	srtt = (long *)&rt->srtt[timer];
 	m -= *srtt >> 3;
 	*srtt += m;
 
 	if (m < 0)
 		m = -m;
 
-	sdrtt = &rt->sdrtt[timer];
+	sdrtt = (long *)&rt->sdrtt[timer];
 	m -= *sdrtt >> 2;
 	*sdrtt += m;
 
@@ -101,7 +99,7 @@ rpc_calc_rto(struct rpc_rtt *rt, unsigned timer)
 	if (timer-- == 0)
 		return rt->timeo;
 
-	res = (rt->srtt[timer] >> 3) + rt->sdrtt[timer];
+	res = ((rt->srtt[timer] + 7) >> 3) + rt->sdrtt[timer];
 	if (res > RPC_RTO_MAX)
 		res = RPC_RTO_MAX;
 

net/sunrpc/xprt.c

@@ -138,15 +138,22 @@ xprt_from_sock(struct sock *sk)
 static int
 __xprt_lock_write(struct rpc_xprt *xprt, struct rpc_task *task)
 {
+	struct rpc_rqst *req = task->tk_rqstp;
+
 	if (!xprt->snd_task) {
-		if (xprt->nocong || __xprt_get_cong(xprt, task))
+		if (xprt->nocong || __xprt_get_cong(xprt, task)) {
 			xprt->snd_task = task;
+			if (req) {
+				req->rq_bytes_sent = 0;
+				req->rq_ntrans++;
+			}
+		}
 	}
 	if (xprt->snd_task != task) {
 		dprintk("RPC: %4d TCP write queue full\n", task->tk_pid);
 		task->tk_timeout = 0;
 		task->tk_status = -EAGAIN;
-		if (task->tk_rqstp && task->tk_rqstp->rq_nresend)
+		if (req && req->rq_ntrans)
 			rpc_sleep_on(&xprt->resend, task, NULL, NULL);
 		else
 			rpc_sleep_on(&xprt->sending, task, NULL, NULL);
@@ -181,8 +188,14 @@ __xprt_lock_write_next(struct rpc_xprt *xprt)
 		if (!task)
 			return;
 	}
-	if (xprt->nocong || __xprt_get_cong(xprt, task))
+	if (xprt->nocong || __xprt_get_cong(xprt, task)) {
+		struct rpc_rqst *req = task->tk_rqstp;
 		xprt->snd_task = task;
+		if (req) {
+			req->rq_bytes_sent = 0;
+			req->rq_ntrans++;
+		}
+	}
 }
 
 /*
@@ -422,6 +435,9 @@ xprt_connect(struct rpc_task *task)
 	if (xprt_connected(xprt))
 		goto out_write;
 
+	if (task->tk_rqstp)
+		task->tk_rqstp->rq_bytes_sent = 0;
+
 	/*
 	 * We're here because the xprt was marked disconnected.
 	 * Start by resetting any existing state.
@@ -566,14 +582,13 @@ xprt_complete_rqst(struct rpc_xprt *xprt, struct rpc_rqst *req, int copied)
 	if (!xprt->nocong) {
 		xprt_adjust_cwnd(xprt, copied);
 		__xprt_put_cong(xprt, req);
-	       	if (!req->rq_nresend) {
+	       	if (req->rq_ntrans == 1) {
 			unsigned timer =
 				task->tk_msg.rpc_proc->p_timer;
 			if (timer)
 				rpc_update_rtt(&clnt->cl_rtt, timer,
 						(long)jiffies - req->rq_xtime);
 		}
-		rpc_clear_timeo(&clnt->cl_rtt);
 	}
 
 #ifdef RPC_PROFILE
@@ -714,11 +729,11 @@ udp_data_ready(struct sock *sk, int len)
 
 	dprintk("RPC: %4d received reply\n", task->tk_pid);
 
-	if ((copied = rovr->rq_rlen) > repsize)
+	if ((copied = rovr->rq_private_buf.len) > repsize)
 		copied = repsize;
 
 	/* Suck it into the iovec, verify checksum if not done by hw. */
-	if (csum_partial_copy_to_xdr(&rovr->rq_rcv_buf, skb))
+	if (csum_partial_copy_to_xdr(&rovr->rq_private_buf, skb))
 		goto out_unlock;
 
 	/* Something worked... */
@@ -841,7 +856,7 @@ tcp_read_request(struct rpc_xprt *xprt, skb_reader_t *desc)
 		return;
 	}
 
-	rcvbuf = &req->rq_rcv_buf;
+	rcvbuf = &req->rq_private_buf;
 	len = desc->count;
 	if (len > xprt->tcp_reclen - xprt->tcp_offset) {
 		skb_reader_t my_desc;
@@ -859,7 +874,7 @@ tcp_read_request(struct rpc_xprt *xprt, skb_reader_t *desc)
 	xprt->tcp_copied += len;
 	xprt->tcp_offset += len;
 
-	if (xprt->tcp_copied == req->rq_rlen)
+	if (xprt->tcp_copied == req->rq_private_buf.len)
 		xprt->tcp_flags &= ~XPRT_COPY_DATA;
 	else if (xprt->tcp_offset == xprt->tcp_reclen) {
 		if (xprt->tcp_flags & XPRT_LAST_FRAG)
@@ -1039,21 +1054,6 @@ xprt_write_space(struct sock *sk)
 	read_unlock(&sk->sk_callback_lock);
 }
 
-/*
- * Exponential backoff for UDP retries
- */
-static inline int
-xprt_expbackoff(struct rpc_task *task, struct rpc_rqst *req)
-{
-	int backoff;
-
-	req->rq_ntimeo++;
-	backoff = min(rpc_ntimeo(&task->tk_client->cl_rtt), XPRT_MAX_BACKOFF);
-	if (req->rq_ntimeo < (1 << backoff))
-		return 1;
-	return 0;
-}
-
 /*
  * RPC receive timeout handler.
  */
@@ -1067,15 +1067,8 @@ xprt_timer(struct rpc_task *task)
 	if (req->rq_received)
 		goto out;
 
-	if (!xprt->nocong) {
-		if (xprt_expbackoff(task, req)) {
-			rpc_add_timer(task, xprt_timer);
-			goto out_unlock;
-		}
-		rpc_inc_timeo(&task->tk_client->cl_rtt);
 	xprt_adjust_cwnd(req->rq_xprt, -ETIMEDOUT);
-	}
-	req->rq_nresend++;
+	__xprt_put_cong(xprt, req);
 
 	dprintk("RPC: %4d xprt_timer (%s request)\n",
 		task->tk_pid, req ? "pending" : "backlogged");
@@ -1084,7 +1077,6 @@ xprt_timer(struct rpc_task *task)
 out:
 	task->tk_timeout = 0;
 	rpc_wake_up_task(task);
-out_unlock:
 	spin_unlock(&xprt->sock_lock);
 }
 
@@ -1104,10 +1096,11 @@ xprt_prepare_transmit(struct rpc_task *task)
 	if (xprt->shutdown)
 		return -EIO;
 
-	if (task->tk_rpcwait)
-		rpc_remove_wait_queue(task);
-
 	spin_lock_bh(&xprt->sock_lock);
+	if (req->rq_received && !req->rq_bytes_sent) {
+		err = req->rq_received;
+		goto out_unlock;
+	}
 	if (!__xprt_lock_write(xprt, task)) {
 		err = -EAGAIN;
 		goto out_unlock;
@@ -1117,11 +1110,6 @@ xprt_prepare_transmit(struct rpc_task *task)
 		err = -ENOTCONN;
 		goto out_unlock;
 	}
-
-	if (list_empty(&req->rq_list)) {
-		list_add_tail(&req->rq_list, &xprt->recv);
-		req->rq_received = 0;
-	}
 out_unlock:
 	spin_unlock_bh(&xprt->sock_lock);
 	return err;
@@ -1146,6 +1134,20 @@ xprt_transmit(struct rpc_task *task)
 		*marker = htonl(0x80000000|(req->rq_slen-sizeof(*marker)));
 	}
 
+	smp_rmb();
+	if (!req->rq_received) {
+		if (list_empty(&req->rq_list)) {
+			spin_lock_bh(&xprt->sock_lock);
+			/* Update the softirq receive buffer */
+			memcpy(&req->rq_private_buf, &req->rq_rcv_buf,
+					sizeof(req->rq_private_buf));
+			/* Add request to the receive list */
+			list_add_tail(&req->rq_list, &xprt->recv);
+			spin_unlock_bh(&xprt->sock_lock);
+		}
+	} else if (!req->rq_bytes_sent)
+		return;
+
 	/* Continue transmitting the packet/record. We must be careful
 	 * to cope with writespace callbacks arriving _after_ we have
 	 * called xprt_sendmsg().
@@ -1160,8 +1162,12 @@ xprt_transmit(struct rpc_task *task)
 		if (xprt->stream) {
 			req->rq_bytes_sent += status;
 
-			if (req->rq_bytes_sent >= req->rq_slen)
+			/* If we've sent the entire packet, immediately
+			 * reset the count of bytes sent. */
+			if (req->rq_bytes_sent >= req->rq_slen) {
+				req->rq_bytes_sent = 0;
 				goto out_receive;
+			}
 		} else {
 			if (status >= req->rq_slen)
 				goto out_receive;
@@ -1182,9 +1188,6 @@ xprt_transmit(struct rpc_task *task)
 	 *	 hence there is no danger of the waking up task being put on
 	 *	 schedq, and being picked up by a parallel run of rpciod().
 	 */
-	if (req->rq_received)
-		goto out_release;
-
 	task->tk_status = status;
 
 	switch (status) {
@@ -1214,22 +1217,21 @@ xprt_transmit(struct rpc_task *task)
 		if (xprt->stream)
 			xprt_disconnect(xprt);
 	}
- out_release:
 	xprt_release_write(xprt, task);
-	req->rq_bytes_sent = 0;
 	return;
  out_receive:
 	dprintk("RPC: %4d xmit complete\n", task->tk_pid);
 	/* Set the task's receive timeout value */
+	spin_lock_bh(&xprt->sock_lock);
 	if (!xprt->nocong) {
 		task->tk_timeout = rpc_calc_rto(&clnt->cl_rtt,
 				task->tk_msg.rpc_proc->p_timer);
-		req->rq_ntimeo = 0;
+		task->tk_timeout <<= clnt->cl_timeout.to_retries
+			- req->rq_timeout.to_retries;
 		if (task->tk_timeout > req->rq_timeout.to_maxval)
 			task->tk_timeout = req->rq_timeout.to_maxval;
 	} else
 		task->tk_timeout = req->rq_timeout.to_current;
-	spin_lock_bh(&xprt->sock_lock);
 	/* Don't race with disconnect */
 	if (!xprt_connected(xprt))
 		task->tk_status = -ENOTCONN;
@@ -1237,7 +1239,6 @@ xprt_transmit(struct rpc_task *task)
 		rpc_sleep_on(&xprt->pending, task, NULL, xprt_timer);
 	__xprt_release_write(xprt, task);
 	spin_unlock_bh(&xprt->sock_lock);
-	req->rq_bytes_sent = 0;
 }
 
 /*