Fixed MTP to work with TWRP

net/decnet/Kconfig

@@ -0,0 +1,43 @@
+#
+# DECnet configuration
+#
+config DECNET
+	tristate "DECnet Support"
+	---help---
+	  The DECnet networking protocol was used in many products made by
+	  Digital (now Compaq).  It provides reliable stream and sequenced
+	  packet communications over which run a variety of services similar
+	  to those which run over TCP/IP.
+
+	  To find some tools to use with the kernel layer support, please
+	  look at Patrick Caulfield's web site:
+	  <http://linux-decnet.sourceforge.net/>.
+
+	  More detailed documentation is available in
+	  <file:Documentation/networking/decnet.txt>.
+
+	  Be sure to say Y to "/proc file system support" and "Sysctl support"
+	  below when using DECnet, since you will need sysctl support to aid
+	  in configuration at run time.
+
+	  The DECnet code is also available as a module ( = code which can be
+	  inserted in and removed from the running kernel whenever you want).
+	  The module is called decnet.
+
+config DECNET_ROUTER
+	bool "DECnet: router support"
+	depends on DECNET
+	select FIB_RULES
+	---help---
+	  Add support for turning your DECnet Endnode into a level 1 or 2
+	  router.  This is an experimental, but functional option.  If you
+	  do say Y here, then make sure that you also say Y to "Kernel/User
+	  network link driver", "Routing messages" and "Network packet
+	  filtering".  The first two are required to allow configuration via
+	  rtnetlink (you will need Alexey Kuznetsov's iproute2 package
+	  from <ftp://ftp.tux.org/pub/net/ip-routing/>). The "Network packet
+	  filtering" option will be required for the forthcoming routing daemon
+	  to work.
+
+	  See <file:Documentation/networking/decnet.txt> for more information.
+

net/decnet/Makefile

@@ -0,0 +1,10 @@
+
+obj-$(CONFIG_DECNET) += decnet.o
+
+decnet-y := af_decnet.o dn_nsp_in.o dn_nsp_out.o \
+	    dn_route.o dn_dev.o dn_neigh.o dn_timer.o
+decnet-$(CONFIG_DECNET_ROUTER) += dn_fib.o dn_rules.o dn_table.o
+decnet-y += sysctl_net_decnet.o
+
+obj-$(CONFIG_NETFILTER) += netfilter/
+

net/decnet/README

@@ -0,0 +1,8 @@
+                       Linux DECnet Project
+                      ======================
+
+The documentation for this kernel subsystem is available in the
+Documentation/networking subdirectory of this distribution and also
+on line at http://www.chygwyn.com/DECnet/
+
+Steve Whitehouse <SteveW@ACM.org>

net/decnet/TODO

@@ -0,0 +1,41 @@
+Steve's quick list of things that need finishing off:
+[they are in no particular order and range from the trivial to the long winded]
+
+ o Proper timeouts on each neighbour (in routing mode) rather than
+   just the 60 second On-Ethernet cache value.
+
+ o Support for X.25 linklayer
+
+ o Support for DDCMP link layer
+
+ o The DDCMP device itself
+
+ o PPP support (rfc1762)
+
+ o Lots of testing with real applications
+
+ o Verify errors etc. against POSIX 1003.1g (draft)
+
+ o Using send/recvmsg() to get at connect/disconnect data (POSIX 1003.1g) 
+   [maybe this should be done at socket level... the control data in the
+    send/recvmsg() calls should simply be a vector of set/getsockopt()
+    calls]
+
+ o check MSG_CTRUNC is set where it should be.
+
+ o Find all the commonality between DECnet and IPv4 routing code and extract 
+   it into a small library of routines. [probably a project for 2.7.xx]
+
+ o Add perfect socket hashing - an idea suggested by Paul Koning. Currently
+   we have a half-way house scheme which seems to work reasonably well, but
+   the full scheme is still worth implementing, its not not top of my list
+   right now.
+
+ o Add session control message flow control
+
+ o Add NSP message flow control
+
+ o DECnet sendpages() function
+
+ o AIO for DECnet
+

net/decnet/dn_fib.c

@@ -0,0 +1,790 @@
+/*
+ * DECnet       An implementation of the DECnet protocol suite for the LINUX
+ *              operating system.  DECnet is implemented using the  BSD Socket
+ *              interface as the means of communication with the user level.
+ *
+ *              DECnet Routing Forwarding Information Base (Glue/Info List)
+ *
+ * Author:      Steve Whitehouse <SteveW@ACM.org>
+ *
+ *
+ * Changes:
+ *              Alexey Kuznetsov : SMP locking changes
+ *              Steve Whitehouse : Rewrote it... Well to be more correct, I
+ *                                 copied most of it from the ipv4 fib code.
+ *              Steve Whitehouse : Updated it in style and fixed a few bugs
+ *                                 which were fixed in the ipv4 code since
+ *                                 this code was copied from it.
+ *
+ */
+#include <linux/string.h>
+#include <linux/net.h>
+#include <linux/socket.h>
+#include <linux/slab.h>
+#include <linux/sockios.h>
+#include <linux/init.h>
+#include <linux/skbuff.h>
+#include <linux/netlink.h>
+#include <linux/rtnetlink.h>
+#include <linux/proc_fs.h>
+#include <linux/netdevice.h>
+#include <linux/timer.h>
+#include <linux/spinlock.h>
+#include <linux/atomic.h>
+#include <asm/uaccess.h>
+#include <net/neighbour.h>
+#include <net/dst.h>
+#include <net/flow.h>
+#include <net/fib_rules.h>
+#include <net/dn.h>
+#include <net/dn_route.h>
+#include <net/dn_fib.h>
+#include <net/dn_neigh.h>
+#include <net/dn_dev.h>
+
+#define RT_MIN_TABLE 1
+
+#define for_fib_info() { struct dn_fib_info *fi;\
+	for(fi = dn_fib_info_list; fi; fi = fi->fib_next)
+#define endfor_fib_info() }
+
+#define for_nexthops(fi) { int nhsel; const struct dn_fib_nh *nh;\
+	for(nhsel = 0, nh = (fi)->fib_nh; nhsel < (fi)->fib_nhs; nh++, nhsel++)
+
+#define change_nexthops(fi) { int nhsel; struct dn_fib_nh *nh;\
+	for(nhsel = 0, nh = (struct dn_fib_nh *)((fi)->fib_nh); nhsel < (fi)->fib_nhs; nh++, nhsel++)
+
+#define endfor_nexthops(fi) }
+
+static DEFINE_SPINLOCK(dn_fib_multipath_lock);
+static struct dn_fib_info *dn_fib_info_list;
+static DEFINE_SPINLOCK(dn_fib_info_lock);
+
+static struct
+{
+	int error;
+	u8 scope;
+} dn_fib_props[RTN_MAX+1] = {
+	[RTN_UNSPEC] =      { .error = 0,       .scope = RT_SCOPE_NOWHERE },
+	[RTN_UNICAST] =     { .error = 0,       .scope = RT_SCOPE_UNIVERSE },
+	[RTN_LOCAL] =       { .error = 0,       .scope = RT_SCOPE_HOST },
+	[RTN_BROADCAST] =   { .error = -EINVAL, .scope = RT_SCOPE_NOWHERE },
+	[RTN_ANYCAST] =     { .error = -EINVAL, .scope = RT_SCOPE_NOWHERE },
+	[RTN_MULTICAST] =   { .error = -EINVAL, .scope = RT_SCOPE_NOWHERE },
+	[RTN_BLACKHOLE] =   { .error = -EINVAL, .scope = RT_SCOPE_UNIVERSE },
+	[RTN_UNREACHABLE] = { .error = -EHOSTUNREACH, .scope = RT_SCOPE_UNIVERSE },
+	[RTN_PROHIBIT] =    { .error = -EACCES, .scope = RT_SCOPE_UNIVERSE },
+	[RTN_THROW] =       { .error = -EAGAIN, .scope = RT_SCOPE_UNIVERSE },
+	[RTN_NAT] =         { .error = 0,       .scope = RT_SCOPE_NOWHERE },
+	[RTN_XRESOLVE] =    { .error = -EINVAL, .scope = RT_SCOPE_NOWHERE },
+};
+
+static int dn_fib_sync_down(__le16 local, struct net_device *dev, int force);
+static int dn_fib_sync_up(struct net_device *dev);
+
+void dn_fib_free_info(struct dn_fib_info *fi)
+{
+	if (fi->fib_dead == 0) {
+		printk(KERN_DEBUG "DECnet: BUG! Attempt to free alive dn_fib_info\n");
+		return;
+	}
+
+	change_nexthops(fi) {
+		if (nh->nh_dev)
+			dev_put(nh->nh_dev);
+		nh->nh_dev = NULL;
+	} endfor_nexthops(fi);
+	kfree(fi);
+}
+
+void dn_fib_release_info(struct dn_fib_info *fi)
+{
+	spin_lock(&dn_fib_info_lock);
+	if (fi && --fi->fib_treeref == 0) {
+		if (fi->fib_next)
+			fi->fib_next->fib_prev = fi->fib_prev;
+		if (fi->fib_prev)
+			fi->fib_prev->fib_next = fi->fib_next;
+		if (fi == dn_fib_info_list)
+			dn_fib_info_list = fi->fib_next;
+		fi->fib_dead = 1;
+		dn_fib_info_put(fi);
+	}
+	spin_unlock(&dn_fib_info_lock);
+}
+
+static inline int dn_fib_nh_comp(const struct dn_fib_info *fi, const struct dn_fib_info *ofi)
+{
+	const struct dn_fib_nh *onh = ofi->fib_nh;
+
+	for_nexthops(fi) {
+		if (nh->nh_oif != onh->nh_oif ||
+			nh->nh_gw != onh->nh_gw ||
+			nh->nh_scope != onh->nh_scope ||
+			nh->nh_weight != onh->nh_weight ||
+			((nh->nh_flags^onh->nh_flags)&~RTNH_F_DEAD))
+				return -1;
+		onh++;
+	} endfor_nexthops(fi);
+	return 0;
+}
+
+static inline struct dn_fib_info *dn_fib_find_info(const struct dn_fib_info *nfi)
+{
+	for_fib_info() {
+		if (fi->fib_nhs != nfi->fib_nhs)
+			continue;
+		if (nfi->fib_protocol == fi->fib_protocol &&
+			nfi->fib_prefsrc == fi->fib_prefsrc &&
+			nfi->fib_priority == fi->fib_priority &&
+			memcmp(nfi->fib_metrics, fi->fib_metrics, sizeof(fi->fib_metrics)) == 0 &&
+			((nfi->fib_flags^fi->fib_flags)&~RTNH_F_DEAD) == 0 &&
+			(nfi->fib_nhs == 0 || dn_fib_nh_comp(fi, nfi) == 0))
+				return fi;
+	} endfor_fib_info();
+	return NULL;
+}
+
+static int dn_fib_count_nhs(const struct nlattr *attr)
+{
+	struct rtnexthop *nhp = nla_data(attr);
+	int nhs = 0, nhlen = nla_len(attr);
+
+	while(nhlen >= (int)sizeof(struct rtnexthop)) {
+		if ((nhlen -= nhp->rtnh_len) < 0)
+			return 0;
+		nhs++;
+		nhp = RTNH_NEXT(nhp);
+	}
+
+	return nhs;
+}
+
+static int dn_fib_get_nhs(struct dn_fib_info *fi, const struct nlattr *attr,
+			  const struct rtmsg *r)
+{
+	struct rtnexthop *nhp = nla_data(attr);
+	int nhlen = nla_len(attr);
+
+	change_nexthops(fi) {
+		int attrlen = nhlen - sizeof(struct rtnexthop);
+		if (attrlen < 0 || (nhlen -= nhp->rtnh_len) < 0)
+			return -EINVAL;
+
+		nh->nh_flags  = (r->rtm_flags&~0xFF) | nhp->rtnh_flags;
+		nh->nh_oif    = nhp->rtnh_ifindex;
+		nh->nh_weight = nhp->rtnh_hops + 1;
+
+		if (attrlen) {
+			struct nlattr *gw_attr;
+
+			gw_attr = nla_find((struct nlattr *) (nhp + 1), attrlen, RTA_GATEWAY);
+			nh->nh_gw = gw_attr ? nla_get_le16(gw_attr) : 0;
+		}
+		nhp = RTNH_NEXT(nhp);
+	} endfor_nexthops(fi);
+
+	return 0;
+}
+
+
+static int dn_fib_check_nh(const struct rtmsg *r, struct dn_fib_info *fi, struct dn_fib_nh *nh)
+{
+	int err;
+
+	if (nh->nh_gw) {
+		struct flowidn fld;
+		struct dn_fib_res res;
+
+		if (nh->nh_flags&RTNH_F_ONLINK) {
+			struct net_device *dev;
+
+			if (r->rtm_scope >= RT_SCOPE_LINK)
+				return -EINVAL;
+			if (dnet_addr_type(nh->nh_gw) != RTN_UNICAST)
+				return -EINVAL;
+			if ((dev = __dev_get_by_index(&init_net, nh->nh_oif)) == NULL)
+				return -ENODEV;
+			if (!(dev->flags&IFF_UP))
+				return -ENETDOWN;
+			nh->nh_dev = dev;
+			dev_hold(dev);
+			nh->nh_scope = RT_SCOPE_LINK;
+			return 0;
+		}
+
+		memset(&fld, 0, sizeof(fld));
+		fld.daddr = nh->nh_gw;
+		fld.flowidn_oif = nh->nh_oif;
+		fld.flowidn_scope = r->rtm_scope + 1;
+
+		if (fld.flowidn_scope < RT_SCOPE_LINK)
+			fld.flowidn_scope = RT_SCOPE_LINK;
+
+		if ((err = dn_fib_lookup(&fld, &res)) != 0)
+			return err;
+
+		err = -EINVAL;
+		if (res.type != RTN_UNICAST && res.type != RTN_LOCAL)
+			goto out;
+		nh->nh_scope = res.scope;
+		nh->nh_oif = DN_FIB_RES_OIF(res);
+		nh->nh_dev = DN_FIB_RES_DEV(res);
+		if (nh->nh_dev == NULL)
+			goto out;
+		dev_hold(nh->nh_dev);
+		err = -ENETDOWN;
+		if (!(nh->nh_dev->flags & IFF_UP))
+			goto out;
+		err = 0;
+out:
+		dn_fib_res_put(&res);
+		return err;
+	} else {
+		struct net_device *dev;
+
+		if (nh->nh_flags&(RTNH_F_PERVASIVE|RTNH_F_ONLINK))
+			return -EINVAL;
+
+		dev = __dev_get_by_index(&init_net, nh->nh_oif);
+		if (dev == NULL || dev->dn_ptr == NULL)
+			return -ENODEV;
+		if (!(dev->flags&IFF_UP))
+			return -ENETDOWN;
+		nh->nh_dev = dev;
+		dev_hold(nh->nh_dev);
+		nh->nh_scope = RT_SCOPE_HOST;
+	}
+
+	return 0;
+}
+
+
+struct dn_fib_info *dn_fib_create_info(const struct rtmsg *r, struct nlattr *attrs[],
+				       const struct nlmsghdr *nlh, int *errp)
+{
+	int err;
+	struct dn_fib_info *fi = NULL;
+	struct dn_fib_info *ofi;
+	int nhs = 1;
+
+	if (r->rtm_type > RTN_MAX)
+		goto err_inval;
+
+	if (dn_fib_props[r->rtm_type].scope > r->rtm_scope)
+		goto err_inval;
+
+	if (attrs[RTA_MULTIPATH] &&
+	    (nhs = dn_fib_count_nhs(attrs[RTA_MULTIPATH])) == 0)
+		goto err_inval;
+
+	fi = kzalloc(sizeof(*fi)+nhs*sizeof(struct dn_fib_nh), GFP_KERNEL);
+	err = -ENOBUFS;
+	if (fi == NULL)
+		goto failure;
+
+	fi->fib_protocol = r->rtm_protocol;
+	fi->fib_nhs = nhs;
+	fi->fib_flags = r->rtm_flags;
+
+	if (attrs[RTA_PRIORITY])
+		fi->fib_priority = nla_get_u32(attrs[RTA_PRIORITY]);
+
+	if (attrs[RTA_METRICS]) {
+		struct nlattr *attr;
+		int rem;
+
+		nla_for_each_nested(attr, attrs[RTA_METRICS], rem) {
+			int type = nla_type(attr);
+
+			if (type) {
+				if (type > RTAX_MAX || nla_len(attr) < 4)
+					goto err_inval;
+
+				fi->fib_metrics[type-1] = nla_get_u32(attr);
+			}
+		}
+	}
+
+	if (attrs[RTA_PREFSRC])
+		fi->fib_prefsrc = nla_get_le16(attrs[RTA_PREFSRC]);
+
+	if (attrs[RTA_MULTIPATH]) {
+		if ((err = dn_fib_get_nhs(fi, attrs[RTA_MULTIPATH], r)) != 0)
+			goto failure;
+
+		if (attrs[RTA_OIF] &&
+		    fi->fib_nh->nh_oif != nla_get_u32(attrs[RTA_OIF]))
+			goto err_inval;
+
+		if (attrs[RTA_GATEWAY] &&
+		    fi->fib_nh->nh_gw != nla_get_le16(attrs[RTA_GATEWAY]))
+			goto err_inval;
+	} else {
+		struct dn_fib_nh *nh = fi->fib_nh;
+
+		if (attrs[RTA_OIF])
+			nh->nh_oif = nla_get_u32(attrs[RTA_OIF]);
+
+		if (attrs[RTA_GATEWAY])
+			nh->nh_gw = nla_get_le16(attrs[RTA_GATEWAY]);
+
+		nh->nh_flags = r->rtm_flags;
+		nh->nh_weight = 1;
+	}
+
+	if (r->rtm_type == RTN_NAT) {
+		if (!attrs[RTA_GATEWAY] || nhs != 1 || attrs[RTA_OIF])
+			goto err_inval;
+
+		fi->fib_nh->nh_gw = nla_get_le16(attrs[RTA_GATEWAY]);
+		goto link_it;
+	}
+
+	if (dn_fib_props[r->rtm_type].error) {
+		if (attrs[RTA_GATEWAY] || attrs[RTA_OIF] || attrs[RTA_MULTIPATH])
+			goto err_inval;
+
+		goto link_it;
+	}
+
+	if (r->rtm_scope > RT_SCOPE_HOST)
+		goto err_inval;
+
+	if (r->rtm_scope == RT_SCOPE_HOST) {
+		struct dn_fib_nh *nh = fi->fib_nh;
+
+		/* Local address is added */
+		if (nhs != 1 || nh->nh_gw)
+			goto err_inval;
+		nh->nh_scope = RT_SCOPE_NOWHERE;
+		nh->nh_dev = dev_get_by_index(&init_net, fi->fib_nh->nh_oif);
+		err = -ENODEV;
+		if (nh->nh_dev == NULL)
+			goto failure;
+	} else {
+		change_nexthops(fi) {
+			if ((err = dn_fib_check_nh(r, fi, nh)) != 0)
+				goto failure;
+		} endfor_nexthops(fi)
+	}
+
+	if (fi->fib_prefsrc) {
+		if (r->rtm_type != RTN_LOCAL || !attrs[RTA_DST] ||
+		    fi->fib_prefsrc != nla_get_le16(attrs[RTA_DST]))
+			if (dnet_addr_type(fi->fib_prefsrc) != RTN_LOCAL)
+				goto err_inval;
+	}
+
+link_it:
+	if ((ofi = dn_fib_find_info(fi)) != NULL) {
+		fi->fib_dead = 1;
+		dn_fib_free_info(fi);
+		ofi->fib_treeref++;
+		return ofi;
+	}
+
+	fi->fib_treeref++;
+	atomic_inc(&fi->fib_clntref);
+	spin_lock(&dn_fib_info_lock);
+	fi->fib_next = dn_fib_info_list;
+	fi->fib_prev = NULL;
+	if (dn_fib_info_list)
+		dn_fib_info_list->fib_prev = fi;
+	dn_fib_info_list = fi;
+	spin_unlock(&dn_fib_info_lock);
+	return fi;
+
+err_inval:
+	err = -EINVAL;
+
+failure:
+	*errp = err;
+	if (fi) {
+		fi->fib_dead = 1;
+		dn_fib_free_info(fi);
+	}
+
+	return NULL;
+}
+
+int dn_fib_semantic_match(int type, struct dn_fib_info *fi, const struct flowidn *fld, struct dn_fib_res *res)
+{
+	int err = dn_fib_props[type].error;
+
+	if (err == 0) {
+		if (fi->fib_flags & RTNH_F_DEAD)
+			return 1;
+
+		res->fi = fi;
+
+		switch (type) {
+		case RTN_NAT:
+			DN_FIB_RES_RESET(*res);
+			atomic_inc(&fi->fib_clntref);
+			return 0;
+		case RTN_UNICAST:
+		case RTN_LOCAL:
+			for_nexthops(fi) {
+				if (nh->nh_flags & RTNH_F_DEAD)
+					continue;
+				if (!fld->flowidn_oif ||
+				    fld->flowidn_oif == nh->nh_oif)
+					break;
+			}
+			if (nhsel < fi->fib_nhs) {
+				res->nh_sel = nhsel;
+				atomic_inc(&fi->fib_clntref);
+				return 0;
+			}
+			endfor_nexthops(fi);
+			res->fi = NULL;
+			return 1;
+		default:
+			net_err_ratelimited("DECnet: impossible routing event : dn_fib_semantic_match type=%d\n",
+					    type);
+			res->fi = NULL;
+			return -EINVAL;
+		}
+	}
+	return err;
+}
+
+void dn_fib_select_multipath(const struct flowidn *fld, struct dn_fib_res *res)
+{
+	struct dn_fib_info *fi = res->fi;
+	int w;
+
+	spin_lock_bh(&dn_fib_multipath_lock);
+	if (fi->fib_power <= 0) {
+		int power = 0;
+		change_nexthops(fi) {
+			if (!(nh->nh_flags&RTNH_F_DEAD)) {
+				power += nh->nh_weight;
+				nh->nh_power = nh->nh_weight;
+			}
+		} endfor_nexthops(fi);
+		fi->fib_power = power;
+		if (power < 0) {
+			spin_unlock_bh(&dn_fib_multipath_lock);
+			res->nh_sel = 0;
+			return;
+		}
+	}
+
+	w = jiffies % fi->fib_power;
+
+	change_nexthops(fi) {
+		if (!(nh->nh_flags&RTNH_F_DEAD) && nh->nh_power) {
+			if ((w -= nh->nh_power) <= 0) {
+				nh->nh_power--;
+				fi->fib_power--;
+				res->nh_sel = nhsel;
+				spin_unlock_bh(&dn_fib_multipath_lock);
+				return;
+			}
+		}
+	} endfor_nexthops(fi);
+	res->nh_sel = 0;
+	spin_unlock_bh(&dn_fib_multipath_lock);
+}
+
+static inline u32 rtm_get_table(struct nlattr *attrs[], u8 table)
+{
+	if (attrs[RTA_TABLE])
+		table = nla_get_u32(attrs[RTA_TABLE]);
+
+	return table;
+}
+
+static int dn_fib_rtm_delroute(struct sk_buff *skb, struct nlmsghdr *nlh)
+{
+	struct net *net = sock_net(skb->sk);
+	struct dn_fib_table *tb;
+	struct rtmsg *r = nlmsg_data(nlh);
+	struct nlattr *attrs[RTA_MAX+1];
+	int err;
+
+	if (!netlink_capable(skb, CAP_NET_ADMIN))
+		return -EPERM;
+
+	if (!net_eq(net, &init_net))
+		return -EINVAL;
+
+	err = nlmsg_parse(nlh, sizeof(*r), attrs, RTA_MAX, rtm_dn_policy);
+	if (err < 0)
+		return err;
+
+	tb = dn_fib_get_table(rtm_get_table(attrs, r->rtm_table), 0);
+	if (!tb)
+		return -ESRCH;
+
+	return tb->delete(tb, r, attrs, nlh, &NETLINK_CB(skb));
+}
+
+static int dn_fib_rtm_newroute(struct sk_buff *skb, struct nlmsghdr *nlh)
+{
+	struct net *net = sock_net(skb->sk);
+	struct dn_fib_table *tb;
+	struct rtmsg *r = nlmsg_data(nlh);
+	struct nlattr *attrs[RTA_MAX+1];
+	int err;
+
+	if (!netlink_capable(skb, CAP_NET_ADMIN))
+		return -EPERM;
+
+	if (!net_eq(net, &init_net))
+		return -EINVAL;
+
+	err = nlmsg_parse(nlh, sizeof(*r), attrs, RTA_MAX, rtm_dn_policy);
+	if (err < 0)
+		return err;
+
+	tb = dn_fib_get_table(rtm_get_table(attrs, r->rtm_table), 1);
+	if (!tb)
+		return -ENOBUFS;
+
+	return tb->insert(tb, r, attrs, nlh, &NETLINK_CB(skb));
+}
+
+static void fib_magic(int cmd, int type, __le16 dst, int dst_len, struct dn_ifaddr *ifa)
+{
+	struct dn_fib_table *tb;
+	struct {
+		struct nlmsghdr nlh;
+		struct rtmsg rtm;
+	} req;
+	struct {
+		struct nlattr hdr;
+		__le16 dst;
+	} dst_attr = {
+		.dst = dst,
+	};
+	struct {
+		struct nlattr hdr;
+		__le16 prefsrc;
+	} prefsrc_attr = {
+		.prefsrc = ifa->ifa_local,
+	};
+	struct {
+		struct nlattr hdr;
+		u32 oif;
+	} oif_attr = {
+		.oif = ifa->ifa_dev->dev->ifindex,
+	};
+	struct nlattr *attrs[RTA_MAX+1] = {
+		[RTA_DST] = (struct nlattr *) &dst_attr,
+		[RTA_PREFSRC] = (struct nlattr * ) &prefsrc_attr,
+		[RTA_OIF] = (struct nlattr *) &oif_attr,
+	};
+
+	memset(&req.rtm, 0, sizeof(req.rtm));
+
+	if (type == RTN_UNICAST)
+		tb = dn_fib_get_table(RT_MIN_TABLE, 1);
+	else
+		tb = dn_fib_get_table(RT_TABLE_LOCAL, 1);
+
+	if (tb == NULL)
+		return;
+
+	req.nlh.nlmsg_len = sizeof(req);
+	req.nlh.nlmsg_type = cmd;
+	req.nlh.nlmsg_flags = NLM_F_REQUEST|NLM_F_CREATE|NLM_F_APPEND;
+	req.nlh.nlmsg_pid = 0;
+	req.nlh.nlmsg_seq = 0;
+
+	req.rtm.rtm_dst_len = dst_len;
+	req.rtm.rtm_table = tb->n;
+	req.rtm.rtm_protocol = RTPROT_KERNEL;
+	req.rtm.rtm_scope = (type != RTN_LOCAL ? RT_SCOPE_LINK : RT_SCOPE_HOST);
+	req.rtm.rtm_type = type;
+
+	if (cmd == RTM_NEWROUTE)
+		tb->insert(tb, &req.rtm, attrs, &req.nlh, NULL);
+	else
+		tb->delete(tb, &req.rtm, attrs, &req.nlh, NULL);
+}
+
+static void dn_fib_add_ifaddr(struct dn_ifaddr *ifa)
+{
+
+	fib_magic(RTM_NEWROUTE, RTN_LOCAL, ifa->ifa_local, 16, ifa);
+
+#if 0
+	if (!(dev->flags&IFF_UP))
+		return;
+	/* In the future, we will want to add default routes here */
+
+#endif
+}
+
+static void dn_fib_del_ifaddr(struct dn_ifaddr *ifa)
+{
+	int found_it = 0;
+	struct net_device *dev;
+	struct dn_dev *dn_db;
+	struct dn_ifaddr *ifa2;
+
+	ASSERT_RTNL();
+
+	/* Scan device list */
+	rcu_read_lock();
+	for_each_netdev_rcu(&init_net, dev) {
+		dn_db = rcu_dereference(dev->dn_ptr);
+		if (dn_db == NULL)
+			continue;
+		for (ifa2 = rcu_dereference(dn_db->ifa_list);
+		     ifa2 != NULL;
+		     ifa2 = rcu_dereference(ifa2->ifa_next)) {
+			if (ifa2->ifa_local == ifa->ifa_local) {
+				found_it = 1;
+				break;
+			}
+		}
+	}
+	rcu_read_unlock();
+
+	if (found_it == 0) {
+		fib_magic(RTM_DELROUTE, RTN_LOCAL, ifa->ifa_local, 16, ifa);
+
+		if (dnet_addr_type(ifa->ifa_local) != RTN_LOCAL) {
+			if (dn_fib_sync_down(ifa->ifa_local, NULL, 0))
+				dn_fib_flush();
+		}
+	}
+}
+
+static void dn_fib_disable_addr(struct net_device *dev, int force)
+{
+	if (dn_fib_sync_down(0, dev, force))
+		dn_fib_flush();
+	dn_rt_cache_flush(0);
+	neigh_ifdown(&dn_neigh_table, dev);
+}
+
+static int dn_fib_dnaddr_event(struct notifier_block *this, unsigned long event, void *ptr)
+{
+	struct dn_ifaddr *ifa = (struct dn_ifaddr *)ptr;
+
+	switch (event) {
+	case NETDEV_UP:
+		dn_fib_add_ifaddr(ifa);
+		dn_fib_sync_up(ifa->ifa_dev->dev);
+		dn_rt_cache_flush(-1);
+		break;
+	case NETDEV_DOWN:
+		dn_fib_del_ifaddr(ifa);
+		if (ifa->ifa_dev && ifa->ifa_dev->ifa_list == NULL) {
+			dn_fib_disable_addr(ifa->ifa_dev->dev, 1);
+		} else {
+			dn_rt_cache_flush(-1);
+		}
+		break;
+	}
+	return NOTIFY_DONE;
+}
+
+static int dn_fib_sync_down(__le16 local, struct net_device *dev, int force)
+{
+	int ret = 0;
+	int scope = RT_SCOPE_NOWHERE;
+
+	if (force)
+		scope = -1;
+
+	for_fib_info() {
+		/*
+		 * This makes no sense for DECnet.... we will almost
+		 * certainly have more than one local address the same
+		 * over all our interfaces. It needs thinking about
+		 * some more.
+		 */
+		if (local && fi->fib_prefsrc == local) {
+			fi->fib_flags |= RTNH_F_DEAD;
+			ret++;
+		} else if (dev && fi->fib_nhs) {
+			int dead = 0;
+
+			change_nexthops(fi) {
+				if (nh->nh_flags&RTNH_F_DEAD)
+					dead++;
+				else if (nh->nh_dev == dev &&
+						nh->nh_scope != scope) {
+					spin_lock_bh(&dn_fib_multipath_lock);
+					nh->nh_flags |= RTNH_F_DEAD;
+					fi->fib_power -= nh->nh_power;
+					nh->nh_power = 0;
+					spin_unlock_bh(&dn_fib_multipath_lock);
+					dead++;
+				}
+			} endfor_nexthops(fi)
+			if (dead == fi->fib_nhs) {
+				fi->fib_flags |= RTNH_F_DEAD;
+				ret++;
+			}
+		}
+	} endfor_fib_info();
+	return ret;
+}
+
+
+static int dn_fib_sync_up(struct net_device *dev)
+{
+	int ret = 0;
+
+	if (!(dev->flags&IFF_UP))
+		return 0;
+
+	for_fib_info() {
+		int alive = 0;
+
+		change_nexthops(fi) {
+			if (!(nh->nh_flags&RTNH_F_DEAD)) {
+				alive++;
+				continue;
+			}
+			if (nh->nh_dev == NULL || !(nh->nh_dev->flags&IFF_UP))
+				continue;
+			if (nh->nh_dev != dev || dev->dn_ptr == NULL)
+				continue;
+			alive++;
+			spin_lock_bh(&dn_fib_multipath_lock);
+			nh->nh_power = 0;
+			nh->nh_flags &= ~RTNH_F_DEAD;
+			spin_unlock_bh(&dn_fib_multipath_lock);
+		} endfor_nexthops(fi);
+
+		if (alive > 0) {
+			fi->fib_flags &= ~RTNH_F_DEAD;
+			ret++;
+		}
+	} endfor_fib_info();
+	return ret;
+}
+
+static struct notifier_block dn_fib_dnaddr_notifier = {
+	.notifier_call = dn_fib_dnaddr_event,
+};
+
+void __exit dn_fib_cleanup(void)
+{
+	dn_fib_table_cleanup();
+	dn_fib_rules_cleanup();
+
+	unregister_dnaddr_notifier(&dn_fib_dnaddr_notifier);
+}
+
+
+void __init dn_fib_init(void)
+{
+	dn_fib_table_init();
+	dn_fib_rules_init();
+
+	register_dnaddr_notifier(&dn_fib_dnaddr_notifier);
+
+	rtnl_register(PF_DECnet, RTM_NEWROUTE, dn_fib_rtm_newroute, NULL, NULL);
+	rtnl_register(PF_DECnet, RTM_DELROUTE, dn_fib_rtm_delroute, NULL, NULL);
+}
+
+

net/decnet/dn_neigh.c

@@ -0,0 +1,603 @@
+/*
+ * DECnet       An implementation of the DECnet protocol suite for the LINUX
+ *              operating system.  DECnet is implemented using the  BSD Socket
+ *              interface as the means of communication with the user level.
+ *
+ *              DECnet Neighbour Functions (Adjacency Database and
+ *                                                        On-Ethernet Cache)
+ *
+ * Author:      Steve Whitehouse <SteveW@ACM.org>
+ *
+ *
+ * Changes:
+ *     Steve Whitehouse     : Fixed router listing routine
+ *     Steve Whitehouse     : Added error_report functions
+ *     Steve Whitehouse     : Added default router detection
+ *     Steve Whitehouse     : Hop counts in outgoing messages
+ *     Steve Whitehouse     : Fixed src/dst in outgoing messages so
+ *                            forwarding now stands a good chance of
+ *                            working.
+ *     Steve Whitehouse     : Fixed neighbour states (for now anyway).
+ *     Steve Whitehouse     : Made error_report functions dummies. This
+ *                            is not the right place to return skbs.
+ *     Steve Whitehouse     : Convert to seq_file
+ *
+ */
+
+#include <linux/net.h>
+#include <linux/module.h>
+#include <linux/socket.h>
+#include <linux/if_arp.h>
+#include <linux/slab.h>
+#include <linux/if_ether.h>
+#include <linux/init.h>
+#include <linux/proc_fs.h>
+#include <linux/string.h>
+#include <linux/netfilter_decnet.h>
+#include <linux/spinlock.h>
+#include <linux/seq_file.h>
+#include <linux/rcupdate.h>
+#include <linux/jhash.h>
+#include <linux/atomic.h>
+#include <net/net_namespace.h>
+#include <net/neighbour.h>
+#include <net/dst.h>
+#include <net/flow.h>
+#include <net/dn.h>
+#include <net/dn_dev.h>
+#include <net/dn_neigh.h>
+#include <net/dn_route.h>
+
+static int dn_neigh_construct(struct neighbour *);
+static void dn_long_error_report(struct neighbour *, struct sk_buff *);
+static void dn_short_error_report(struct neighbour *, struct sk_buff *);
+static int dn_long_output(struct neighbour *, struct sk_buff *);
+static int dn_short_output(struct neighbour *, struct sk_buff *);
+static int dn_phase3_output(struct neighbour *, struct sk_buff *);
+
+
+/*
+ * For talking to broadcast devices: Ethernet & PPP
+ */
+static const struct neigh_ops dn_long_ops = {
+	.family =		AF_DECnet,
+	.error_report =		dn_long_error_report,
+	.output =		dn_long_output,
+	.connected_output =	dn_long_output,
+};
+
+/*
+ * For talking to pointopoint and multidrop devices: DDCMP and X.25
+ */
+static const struct neigh_ops dn_short_ops = {
+	.family =		AF_DECnet,
+	.error_report =		dn_short_error_report,
+	.output =		dn_short_output,
+	.connected_output =	dn_short_output,
+};
+
+/*
+ * For talking to DECnet phase III nodes
+ */
+static const struct neigh_ops dn_phase3_ops = {
+	.family =		AF_DECnet,
+	.error_report =		dn_short_error_report, /* Can use short version here */
+	.output =		dn_phase3_output,
+	.connected_output =	dn_phase3_output,
+};
+
+static u32 dn_neigh_hash(const void *pkey,
+			 const struct net_device *dev,
+			 __u32 *hash_rnd)
+{
+	return jhash_2words(*(__u16 *)pkey, 0, hash_rnd[0]);
+}
+
+struct neigh_table dn_neigh_table = {
+	.family =			PF_DECnet,
+	.entry_size =			NEIGH_ENTRY_SIZE(sizeof(struct dn_neigh)),
+	.key_len =			sizeof(__le16),
+	.hash =				dn_neigh_hash,
+	.constructor =			dn_neigh_construct,
+	.id =				"dn_neigh_cache",
+	.parms ={
+		.tbl =			&dn_neigh_table,
+		.reachable_time =	30 * HZ,
+		.data = {
+			[NEIGH_VAR_MCAST_PROBES] = 0,
+			[NEIGH_VAR_UCAST_PROBES] = 0,
+			[NEIGH_VAR_APP_PROBES] = 0,
+			[NEIGH_VAR_RETRANS_TIME] = 1 * HZ,
+			[NEIGH_VAR_BASE_REACHABLE_TIME] = 30 * HZ,
+			[NEIGH_VAR_DELAY_PROBE_TIME] = 5 * HZ,
+			[NEIGH_VAR_GC_STALETIME] = 60 * HZ,
+			[NEIGH_VAR_QUEUE_LEN_BYTES] = 64*1024,
+			[NEIGH_VAR_PROXY_QLEN] = 0,
+			[NEIGH_VAR_ANYCAST_DELAY] = 0,
+			[NEIGH_VAR_PROXY_DELAY] = 0,
+			[NEIGH_VAR_LOCKTIME] = 1 * HZ,
+		},
+	},
+	.gc_interval =			30 * HZ,
+	.gc_thresh1 =			128,
+	.gc_thresh2 =			512,
+	.gc_thresh3 =			1024,
+};
+
+static int dn_neigh_construct(struct neighbour *neigh)
+{
+	struct net_device *dev = neigh->dev;
+	struct dn_neigh *dn = (struct dn_neigh *)neigh;
+	struct dn_dev *dn_db;
+	struct neigh_parms *parms;
+
+	rcu_read_lock();
+	dn_db = rcu_dereference(dev->dn_ptr);
+	if (dn_db == NULL) {
+		rcu_read_unlock();
+		return -EINVAL;
+	}
+
+	parms = dn_db->neigh_parms;
+	if (!parms) {
+		rcu_read_unlock();
+		return -EINVAL;
+	}
+
+	__neigh_parms_put(neigh->parms);
+	neigh->parms = neigh_parms_clone(parms);
+
+	if (dn_db->use_long)
+		neigh->ops = &dn_long_ops;
+	else
+		neigh->ops = &dn_short_ops;
+	rcu_read_unlock();
+
+	if (dn->flags & DN_NDFLAG_P3)
+		neigh->ops = &dn_phase3_ops;
+
+	neigh->nud_state = NUD_NOARP;
+	neigh->output = neigh->ops->connected_output;
+
+	if ((dev->type == ARPHRD_IPGRE) || (dev->flags & IFF_POINTOPOINT))
+		memcpy(neigh->ha, dev->broadcast, dev->addr_len);
+	else if ((dev->type == ARPHRD_ETHER) || (dev->type == ARPHRD_LOOPBACK))
+		dn_dn2eth(neigh->ha, dn->addr);
+	else {
+		net_dbg_ratelimited("Trying to create neigh for hw %d\n",
+				    dev->type);
+		return -EINVAL;
+	}
+
+	/*
+	 * Make an estimate of the remote block size by assuming that its
+	 * two less then the device mtu, which it true for ethernet (and
+	 * other things which support long format headers) since there is
+	 * an extra length field (of 16 bits) which isn't part of the
+	 * ethernet headers and which the DECnet specs won't admit is part
+	 * of the DECnet routing headers either.
+	 *
+	 * If we over estimate here its no big deal, the NSP negotiations
+	 * will prevent us from sending packets which are too large for the
+	 * remote node to handle. In any case this figure is normally updated
+	 * by a hello message in most cases.
+	 */
+	dn->blksize = dev->mtu - 2;
+
+	return 0;
+}
+
+static void dn_long_error_report(struct neighbour *neigh, struct sk_buff *skb)
+{
+	printk(KERN_DEBUG "dn_long_error_report: called\n");
+	kfree_skb(skb);
+}
+
+
+static void dn_short_error_report(struct neighbour *neigh, struct sk_buff *skb)
+{
+	printk(KERN_DEBUG "dn_short_error_report: called\n");
+	kfree_skb(skb);
+}
+
+static int dn_neigh_output_packet(struct sk_buff *skb)
+{
+	struct dst_entry *dst = skb_dst(skb);
+	struct dn_route *rt = (struct dn_route *)dst;
+	struct neighbour *neigh = rt->n;
+	struct net_device *dev = neigh->dev;
+	char mac_addr[ETH_ALEN];
+	unsigned int seq;
+	int err;
+
+	dn_dn2eth(mac_addr, rt->rt_local_src);
+	do {
+		seq = read_seqbegin(&neigh->ha_lock);
+		err = dev_hard_header(skb, dev, ntohs(skb->protocol),
+				      neigh->ha, mac_addr, skb->len);
+	} while (read_seqretry(&neigh->ha_lock, seq));
+
+	if (err >= 0)
+		err = dev_queue_xmit(skb);
+	else {
+		kfree_skb(skb);
+		err = -EINVAL;
+	}
+	return err;
+}
+
+static int dn_long_output(struct neighbour *neigh, struct sk_buff *skb)
+{
+	struct net_device *dev = neigh->dev;
+	int headroom = dev->hard_header_len + sizeof(struct dn_long_packet) + 3;
+	unsigned char *data;
+	struct dn_long_packet *lp;
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+
+
+	if (skb_headroom(skb) < headroom) {
+		struct sk_buff *skb2 = skb_realloc_headroom(skb, headroom);
+		if (skb2 == NULL) {
+			net_crit_ratelimited("dn_long_output: no memory\n");
+			kfree_skb(skb);
+			return -ENOBUFS;
+		}
+		consume_skb(skb);
+		skb = skb2;
+		net_info_ratelimited("dn_long_output: Increasing headroom\n");
+	}
+
+	data = skb_push(skb, sizeof(struct dn_long_packet) + 3);
+	lp = (struct dn_long_packet *)(data+3);
+
+	*((__le16 *)data) = cpu_to_le16(skb->len - 2);
+	*(data + 2) = 1 | DN_RT_F_PF; /* Padding */
+
+	lp->msgflg   = DN_RT_PKT_LONG|(cb->rt_flags&(DN_RT_F_IE|DN_RT_F_RQR|DN_RT_F_RTS));
+	lp->d_area   = lp->d_subarea = 0;
+	dn_dn2eth(lp->d_id, cb->dst);
+	lp->s_area   = lp->s_subarea = 0;
+	dn_dn2eth(lp->s_id, cb->src);
+	lp->nl2      = 0;
+	lp->visit_ct = cb->hops & 0x3f;
+	lp->s_class  = 0;
+	lp->pt       = 0;
+
+	skb_reset_network_header(skb);
+
+	return NF_HOOK(NFPROTO_DECNET, NF_DN_POST_ROUTING, skb, NULL,
+		       neigh->dev, dn_neigh_output_packet);
+}
+
+static int dn_short_output(struct neighbour *neigh, struct sk_buff *skb)
+{
+	struct net_device *dev = neigh->dev;
+	int headroom = dev->hard_header_len + sizeof(struct dn_short_packet) + 2;
+	struct dn_short_packet *sp;
+	unsigned char *data;
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+
+
+	if (skb_headroom(skb) < headroom) {
+		struct sk_buff *skb2 = skb_realloc_headroom(skb, headroom);
+		if (skb2 == NULL) {
+			net_crit_ratelimited("dn_short_output: no memory\n");
+			kfree_skb(skb);
+			return -ENOBUFS;
+		}
+		consume_skb(skb);
+		skb = skb2;
+		net_info_ratelimited("dn_short_output: Increasing headroom\n");
+	}
+
+	data = skb_push(skb, sizeof(struct dn_short_packet) + 2);
+	*((__le16 *)data) = cpu_to_le16(skb->len - 2);
+	sp = (struct dn_short_packet *)(data+2);
+
+	sp->msgflg     = DN_RT_PKT_SHORT|(cb->rt_flags&(DN_RT_F_RQR|DN_RT_F_RTS));
+	sp->dstnode    = cb->dst;
+	sp->srcnode    = cb->src;
+	sp->forward    = cb->hops & 0x3f;
+
+	skb_reset_network_header(skb);
+
+	return NF_HOOK(NFPROTO_DECNET, NF_DN_POST_ROUTING, skb, NULL,
+		       neigh->dev, dn_neigh_output_packet);
+}
+
+/*
+ * Phase 3 output is the same is short output, execpt that
+ * it clears the area bits before transmission.
+ */
+static int dn_phase3_output(struct neighbour *neigh, struct sk_buff *skb)
+{
+	struct net_device *dev = neigh->dev;
+	int headroom = dev->hard_header_len + sizeof(struct dn_short_packet) + 2;
+	struct dn_short_packet *sp;
+	unsigned char *data;
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+
+	if (skb_headroom(skb) < headroom) {
+		struct sk_buff *skb2 = skb_realloc_headroom(skb, headroom);
+		if (skb2 == NULL) {
+			net_crit_ratelimited("dn_phase3_output: no memory\n");
+			kfree_skb(skb);
+			return -ENOBUFS;
+		}
+		consume_skb(skb);
+		skb = skb2;
+		net_info_ratelimited("dn_phase3_output: Increasing headroom\n");
+	}
+
+	data = skb_push(skb, sizeof(struct dn_short_packet) + 2);
+	*((__le16 *)data) = cpu_to_le16(skb->len - 2);
+	sp = (struct dn_short_packet *)(data + 2);
+
+	sp->msgflg   = DN_RT_PKT_SHORT|(cb->rt_flags&(DN_RT_F_RQR|DN_RT_F_RTS));
+	sp->dstnode  = cb->dst & cpu_to_le16(0x03ff);
+	sp->srcnode  = cb->src & cpu_to_le16(0x03ff);
+	sp->forward  = cb->hops & 0x3f;
+
+	skb_reset_network_header(skb);
+
+	return NF_HOOK(NFPROTO_DECNET, NF_DN_POST_ROUTING, skb, NULL,
+		       neigh->dev, dn_neigh_output_packet);
+}
+
+/*
+ * Unfortunately, the neighbour code uses the device in its hash
+ * function, so we don't get any advantage from it. This function
+ * basically does a neigh_lookup(), but without comparing the device
+ * field. This is required for the On-Ethernet cache
+ */
+
+/*
+ * Pointopoint link receives a hello message
+ */
+void dn_neigh_pointopoint_hello(struct sk_buff *skb)
+{
+	kfree_skb(skb);
+}
+
+/*
+ * Ethernet router hello message received
+ */
+int dn_neigh_router_hello(struct sk_buff *skb)
+{
+	struct rtnode_hello_message *msg = (struct rtnode_hello_message *)skb->data;
+
+	struct neighbour *neigh;
+	struct dn_neigh *dn;
+	struct dn_dev *dn_db;
+	__le16 src;
+
+	src = dn_eth2dn(msg->id);
+
+	neigh = __neigh_lookup(&dn_neigh_table, &src, skb->dev, 1);
+
+	dn = (struct dn_neigh *)neigh;
+
+	if (neigh) {
+		write_lock(&neigh->lock);
+
+		neigh->used = jiffies;
+		dn_db = rcu_dereference(neigh->dev->dn_ptr);
+
+		if (!(neigh->nud_state & NUD_PERMANENT)) {
+			neigh->updated = jiffies;
+
+			if (neigh->dev->type == ARPHRD_ETHER)
+				memcpy(neigh->ha, &eth_hdr(skb)->h_source, ETH_ALEN);
+
+			dn->blksize  = le16_to_cpu(msg->blksize);
+			dn->priority = msg->priority;
+
+			dn->flags &= ~DN_NDFLAG_P3;
+
+			switch (msg->iinfo & DN_RT_INFO_TYPE) {
+			case DN_RT_INFO_L1RT:
+				dn->flags &=~DN_NDFLAG_R2;
+				dn->flags |= DN_NDFLAG_R1;
+				break;
+			case DN_RT_INFO_L2RT:
+				dn->flags |= DN_NDFLAG_R2;
+			}
+		}
+
+		/* Only use routers in our area */
+		if ((le16_to_cpu(src)>>10) == (le16_to_cpu((decnet_address))>>10)) {
+			if (!dn_db->router) {
+				dn_db->router = neigh_clone(neigh);
+			} else {
+				if (msg->priority > ((struct dn_neigh *)dn_db->router)->priority)
+					neigh_release(xchg(&dn_db->router, neigh_clone(neigh)));
+			}
+		}
+		write_unlock(&neigh->lock);
+		neigh_release(neigh);
+	}
+
+	kfree_skb(skb);
+	return 0;
+}
+
+/*
+ * Endnode hello message received
+ */
+int dn_neigh_endnode_hello(struct sk_buff *skb)
+{
+	struct endnode_hello_message *msg = (struct endnode_hello_message *)skb->data;
+	struct neighbour *neigh;
+	struct dn_neigh *dn;
+	__le16 src;
+
+	src = dn_eth2dn(msg->id);
+
+	neigh = __neigh_lookup(&dn_neigh_table, &src, skb->dev, 1);
+
+	dn = (struct dn_neigh *)neigh;
+
+	if (neigh) {
+		write_lock(&neigh->lock);
+
+		neigh->used = jiffies;
+
+		if (!(neigh->nud_state & NUD_PERMANENT)) {
+			neigh->updated = jiffies;
+
+			if (neigh->dev->type == ARPHRD_ETHER)
+				memcpy(neigh->ha, &eth_hdr(skb)->h_source, ETH_ALEN);
+			dn->flags   &= ~(DN_NDFLAG_R1 | DN_NDFLAG_R2);
+			dn->blksize  = le16_to_cpu(msg->blksize);
+			dn->priority = 0;
+		}
+
+		write_unlock(&neigh->lock);
+		neigh_release(neigh);
+	}
+
+	kfree_skb(skb);
+	return 0;
+}
+
+static char *dn_find_slot(char *base, int max, int priority)
+{
+	int i;
+	unsigned char *min = NULL;
+
+	base += 6; /* skip first id */
+
+	for(i = 0; i < max; i++) {
+		if (!min || (*base < *min))
+			min = base;
+		base += 7; /* find next priority */
+	}
+
+	if (!min)
+		return NULL;
+
+	return (*min < priority) ? (min - 6) : NULL;
+}
+
+struct elist_cb_state {
+	struct net_device *dev;
+	unsigned char *ptr;
+	unsigned char *rs;
+	int t, n;
+};
+
+static void neigh_elist_cb(struct neighbour *neigh, void *_info)
+{
+	struct elist_cb_state *s = _info;
+	struct dn_neigh *dn;
+
+	if (neigh->dev != s->dev)
+		return;
+
+	dn = (struct dn_neigh *) neigh;
+	if (!(dn->flags & (DN_NDFLAG_R1|DN_NDFLAG_R2)))
+		return;
+
+	if (s->t == s->n)
+		s->rs = dn_find_slot(s->ptr, s->n, dn->priority);
+	else
+		s->t++;
+	if (s->rs == NULL)
+		return;
+
+	dn_dn2eth(s->rs, dn->addr);
+	s->rs += 6;
+	*(s->rs) = neigh->nud_state & NUD_CONNECTED ? 0x80 : 0x0;
+	*(s->rs) |= dn->priority;
+	s->rs++;
+}
+
+int dn_neigh_elist(struct net_device *dev, unsigned char *ptr, int n)
+{
+	struct elist_cb_state state;
+
+	state.dev = dev;
+	state.t = 0;
+	state.n = n;
+	state.ptr = ptr;
+	state.rs = ptr;
+
+	neigh_for_each(&dn_neigh_table, neigh_elist_cb, &state);
+
+	return state.t;
+}
+
+
+#ifdef CONFIG_PROC_FS
+
+static inline void dn_neigh_format_entry(struct seq_file *seq,
+					 struct neighbour *n)
+{
+	struct dn_neigh *dn = (struct dn_neigh *) n;
+	char buf[DN_ASCBUF_LEN];
+
+	read_lock(&n->lock);
+	seq_printf(seq, "%-7s %s%s%s   %02x    %02d  %07ld %-8s\n",
+		   dn_addr2asc(le16_to_cpu(dn->addr), buf),
+		   (dn->flags&DN_NDFLAG_R1) ? "1" : "-",
+		   (dn->flags&DN_NDFLAG_R2) ? "2" : "-",
+		   (dn->flags&DN_NDFLAG_P3) ? "3" : "-",
+		   dn->n.nud_state,
+		   atomic_read(&dn->n.refcnt),
+		   dn->blksize,
+		   (dn->n.dev) ? dn->n.dev->name : "?");
+	read_unlock(&n->lock);
+}
+
+static int dn_neigh_seq_show(struct seq_file *seq, void *v)
+{
+	if (v == SEQ_START_TOKEN) {
+		seq_puts(seq, "Addr    Flags State Use Blksize Dev\n");
+	} else {
+		dn_neigh_format_entry(seq, v);
+	}
+
+	return 0;
+}
+
+static void *dn_neigh_seq_start(struct seq_file *seq, loff_t *pos)
+{
+	return neigh_seq_start(seq, pos, &dn_neigh_table,
+			       NEIGH_SEQ_NEIGH_ONLY);
+}
+
+static const struct seq_operations dn_neigh_seq_ops = {
+	.start = dn_neigh_seq_start,
+	.next  = neigh_seq_next,
+	.stop  = neigh_seq_stop,
+	.show  = dn_neigh_seq_show,
+};
+
+static int dn_neigh_seq_open(struct inode *inode, struct file *file)
+{
+	return seq_open_net(inode, file, &dn_neigh_seq_ops,
+			    sizeof(struct neigh_seq_state));
+}
+
+static const struct file_operations dn_neigh_seq_fops = {
+	.owner		= THIS_MODULE,
+	.open		= dn_neigh_seq_open,
+	.read		= seq_read,
+	.llseek		= seq_lseek,
+	.release	= seq_release_net,
+};
+
+#endif
+
+void __init dn_neigh_init(void)
+{
+	neigh_table_init(&dn_neigh_table);
+	proc_create("decnet_neigh", S_IRUGO, init_net.proc_net,
+		    &dn_neigh_seq_fops);
+}
+
+void __exit dn_neigh_cleanup(void)
+{
+	remove_proc_entry("decnet_neigh", init_net.proc_net);
+	neigh_table_clear(&dn_neigh_table);
+}

net/decnet/dn_nsp_in.c

@@ -0,0 +1,916 @@
+/*
+ * DECnet       An implementation of the DECnet protocol suite for the LINUX
+ *              operating system.  DECnet is implemented using the  BSD Socket
+ *              interface as the means of communication with the user level.
+ *
+ *              DECnet Network Services Protocol (Input)
+ *
+ * Author:      Eduardo Marcelo Serrat <emserrat@geocities.com>
+ *
+ * Changes:
+ *
+ *    Steve Whitehouse:  Split into dn_nsp_in.c and dn_nsp_out.c from
+ *                       original dn_nsp.c.
+ *    Steve Whitehouse:  Updated to work with my new routing architecture.
+ *    Steve Whitehouse:  Add changes from Eduardo Serrat's patches.
+ *    Steve Whitehouse:  Put all ack handling code in a common routine.
+ *    Steve Whitehouse:  Put other common bits into dn_nsp_rx()
+ *    Steve Whitehouse:  More checks on skb->len to catch bogus packets
+ *                       Fixed various race conditions and possible nasties.
+ *    Steve Whitehouse:  Now handles returned conninit frames.
+ *     David S. Miller:  New socket locking
+ *    Steve Whitehouse:  Fixed lockup when socket filtering was enabled.
+ *         Paul Koning:  Fix to push CC sockets into RUN when acks are
+ *                       received.
+ *    Steve Whitehouse:
+ *   Patrick Caulfield:  Checking conninits for correctness & sending of error
+ *                       responses.
+ *    Steve Whitehouse:  Added backlog congestion level return codes.
+ *   Patrick Caulfield:
+ *    Steve Whitehouse:  Added flow control support (outbound)
+ *    Steve Whitehouse:  Prepare for nonlinear skbs
+ */
+
+/******************************************************************************
+    (c) 1995-1998 E.M. Serrat		emserrat@geocities.com
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+*******************************************************************************/
+
+#include <linux/errno.h>
+#include <linux/types.h>
+#include <linux/socket.h>
+#include <linux/in.h>
+#include <linux/kernel.h>
+#include <linux/timer.h>
+#include <linux/string.h>
+#include <linux/sockios.h>
+#include <linux/net.h>
+#include <linux/netdevice.h>
+#include <linux/inet.h>
+#include <linux/route.h>
+#include <linux/slab.h>
+#include <net/sock.h>
+#include <net/tcp_states.h>
+#include <linux/fcntl.h>
+#include <linux/mm.h>
+#include <linux/termios.h>
+#include <linux/interrupt.h>
+#include <linux/proc_fs.h>
+#include <linux/stat.h>
+#include <linux/init.h>
+#include <linux/poll.h>
+#include <linux/netfilter_decnet.h>
+#include <net/neighbour.h>
+#include <net/dst.h>
+#include <net/dn.h>
+#include <net/dn_nsp.h>
+#include <net/dn_dev.h>
+#include <net/dn_route.h>
+
+extern int decnet_log_martians;
+
+static void dn_log_martian(struct sk_buff *skb, const char *msg)
+{
+	if (decnet_log_martians) {
+		char *devname = skb->dev ? skb->dev->name : "???";
+		struct dn_skb_cb *cb = DN_SKB_CB(skb);
+		net_info_ratelimited("DECnet: Martian packet (%s) dev=%s src=0x%04hx dst=0x%04hx srcport=0x%04hx dstport=0x%04hx\n",
+				     msg, devname,
+				     le16_to_cpu(cb->src),
+				     le16_to_cpu(cb->dst),
+				     le16_to_cpu(cb->src_port),
+				     le16_to_cpu(cb->dst_port));
+	}
+}
+
+/*
+ * For this function we've flipped the cross-subchannel bit
+ * if the message is an otherdata or linkservice message. Thus
+ * we can use it to work out what to update.
+ */
+static void dn_ack(struct sock *sk, struct sk_buff *skb, unsigned short ack)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	unsigned short type = ((ack >> 12) & 0x0003);
+	int wakeup = 0;
+
+	switch (type) {
+	case 0: /* ACK - Data */
+		if (dn_after(ack, scp->ackrcv_dat)) {
+			scp->ackrcv_dat = ack & 0x0fff;
+			wakeup |= dn_nsp_check_xmit_queue(sk, skb,
+							  &scp->data_xmit_queue,
+							  ack);
+		}
+		break;
+	case 1: /* NAK - Data */
+		break;
+	case 2: /* ACK - OtherData */
+		if (dn_after(ack, scp->ackrcv_oth)) {
+			scp->ackrcv_oth = ack & 0x0fff;
+			wakeup |= dn_nsp_check_xmit_queue(sk, skb,
+							  &scp->other_xmit_queue,
+							  ack);
+		}
+		break;
+	case 3: /* NAK - OtherData */
+		break;
+	}
+
+	if (wakeup && !sock_flag(sk, SOCK_DEAD))
+		sk->sk_state_change(sk);
+}
+
+/*
+ * This function is a universal ack processor.
+ */
+static int dn_process_ack(struct sock *sk, struct sk_buff *skb, int oth)
+{
+	__le16 *ptr = (__le16 *)skb->data;
+	int len = 0;
+	unsigned short ack;
+
+	if (skb->len < 2)
+		return len;
+
+	if ((ack = le16_to_cpu(*ptr)) & 0x8000) {
+		skb_pull(skb, 2);
+		ptr++;
+		len += 2;
+		if ((ack & 0x4000) == 0) {
+			if (oth)
+				ack ^= 0x2000;
+			dn_ack(sk, skb, ack);
+		}
+	}
+
+	if (skb->len < 2)
+		return len;
+
+	if ((ack = le16_to_cpu(*ptr)) & 0x8000) {
+		skb_pull(skb, 2);
+		len += 2;
+		if ((ack & 0x4000) == 0) {
+			if (oth)
+				ack ^= 0x2000;
+			dn_ack(sk, skb, ack);
+		}
+	}
+
+	return len;
+}
+
+
+/**
+ * dn_check_idf - Check an image data field format is correct.
+ * @pptr: Pointer to pointer to image data
+ * @len: Pointer to length of image data
+ * @max: The maximum allowed length of the data in the image data field
+ * @follow_on: Check that this many bytes exist beyond the end of the image data
+ *
+ * Returns: 0 if ok, -1 on error
+ */
+static inline int dn_check_idf(unsigned char **pptr, int *len, unsigned char max, unsigned char follow_on)
+{
+	unsigned char *ptr = *pptr;
+	unsigned char flen = *ptr++;
+
+	(*len)--;
+	if (flen > max)
+		return -1;
+	if ((flen + follow_on) > *len)
+		return -1;
+
+	*len -= flen;
+	*pptr = ptr + flen;
+	return 0;
+}
+
+/*
+ * Table of reason codes to pass back to node which sent us a badly
+ * formed message, plus text messages for the log. A zero entry in
+ * the reason field means "don't reply" otherwise a disc init is sent with
+ * the specified reason code.
+ */
+static struct {
+	unsigned short reason;
+	const char *text;
+} ci_err_table[] = {
+ { 0,             "CI: Truncated message" },
+ { NSP_REASON_ID, "CI: Destination username error" },
+ { NSP_REASON_ID, "CI: Destination username type" },
+ { NSP_REASON_US, "CI: Source username error" },
+ { 0,             "CI: Truncated at menuver" },
+ { 0,             "CI: Truncated before access or user data" },
+ { NSP_REASON_IO, "CI: Access data format error" },
+ { NSP_REASON_IO, "CI: User data format error" }
+};
+
+/*
+ * This function uses a slightly different lookup method
+ * to find its sockets, since it searches on object name/number
+ * rather than port numbers. Various tests are done to ensure that
+ * the incoming data is in the correct format before it is queued to
+ * a socket.
+ */
+static struct sock *dn_find_listener(struct sk_buff *skb, unsigned short *reason)
+{
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	struct nsp_conn_init_msg *msg = (struct nsp_conn_init_msg *)skb->data;
+	struct sockaddr_dn dstaddr;
+	struct sockaddr_dn srcaddr;
+	unsigned char type = 0;
+	int dstlen;
+	int srclen;
+	unsigned char *ptr;
+	int len;
+	int err = 0;
+	unsigned char menuver;
+
+	memset(&dstaddr, 0, sizeof(struct sockaddr_dn));
+	memset(&srcaddr, 0, sizeof(struct sockaddr_dn));
+
+	/*
+	 * 1. Decode & remove message header
+	 */
+	cb->src_port = msg->srcaddr;
+	cb->dst_port = msg->dstaddr;
+	cb->services = msg->services;
+	cb->info     = msg->info;
+	cb->segsize  = le16_to_cpu(msg->segsize);
+
+	if (!pskb_may_pull(skb, sizeof(*msg)))
+		goto err_out;
+
+	skb_pull(skb, sizeof(*msg));
+
+	len = skb->len;
+	ptr = skb->data;
+
+	/*
+	 * 2. Check destination end username format
+	 */
+	dstlen = dn_username2sockaddr(ptr, len, &dstaddr, &type);
+	err++;
+	if (dstlen < 0)
+		goto err_out;
+
+	err++;
+	if (type > 1)
+		goto err_out;
+
+	len -= dstlen;
+	ptr += dstlen;
+
+	/*
+	 * 3. Check source end username format
+	 */
+	srclen = dn_username2sockaddr(ptr, len, &srcaddr, &type);
+	err++;
+	if (srclen < 0)
+		goto err_out;
+
+	len -= srclen;
+	ptr += srclen;
+	err++;
+	if (len < 1)
+		goto err_out;
+
+	menuver = *ptr;
+	ptr++;
+	len--;
+
+	/*
+	 * 4. Check that optional data actually exists if menuver says it does
+	 */
+	err++;
+	if ((menuver & (DN_MENUVER_ACC | DN_MENUVER_USR)) && (len < 1))
+		goto err_out;
+
+	/*
+	 * 5. Check optional access data format
+	 */
+	err++;
+	if (menuver & DN_MENUVER_ACC) {
+		if (dn_check_idf(&ptr, &len, 39, 1))
+			goto err_out;
+		if (dn_check_idf(&ptr, &len, 39, 1))
+			goto err_out;
+		if (dn_check_idf(&ptr, &len, 39, (menuver & DN_MENUVER_USR) ? 1 : 0))
+			goto err_out;
+	}
+
+	/*
+	 * 6. Check optional user data format
+	 */
+	err++;
+	if (menuver & DN_MENUVER_USR) {
+		if (dn_check_idf(&ptr, &len, 16, 0))
+			goto err_out;
+	}
+
+	/*
+	 * 7. Look up socket based on destination end username
+	 */
+	return dn_sklist_find_listener(&dstaddr);
+err_out:
+	dn_log_martian(skb, ci_err_table[err].text);
+	*reason = ci_err_table[err].reason;
+	return NULL;
+}
+
+
+static void dn_nsp_conn_init(struct sock *sk, struct sk_buff *skb)
+{
+	if (sk_acceptq_is_full(sk)) {
+		kfree_skb(skb);
+		return;
+	}
+
+	sk->sk_ack_backlog++;
+	skb_queue_tail(&sk->sk_receive_queue, skb);
+	sk->sk_state_change(sk);
+}
+
+static void dn_nsp_conn_conf(struct sock *sk, struct sk_buff *skb)
+{
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	struct dn_scp *scp = DN_SK(sk);
+	unsigned char *ptr;
+
+	if (skb->len < 4)
+		goto out;
+
+	ptr = skb->data;
+	cb->services = *ptr++;
+	cb->info = *ptr++;
+	cb->segsize = le16_to_cpu(*(__le16 *)ptr);
+
+	if ((scp->state == DN_CI) || (scp->state == DN_CD)) {
+		scp->persist = 0;
+		scp->addrrem = cb->src_port;
+		sk->sk_state = TCP_ESTABLISHED;
+		scp->state = DN_RUN;
+		scp->services_rem = cb->services;
+		scp->info_rem = cb->info;
+		scp->segsize_rem = cb->segsize;
+
+		if ((scp->services_rem & NSP_FC_MASK) == NSP_FC_NONE)
+			scp->max_window = decnet_no_fc_max_cwnd;
+
+		if (skb->len > 0) {
+			u16 dlen = *skb->data;
+			if ((dlen <= 16) && (dlen <= skb->len)) {
+				scp->conndata_in.opt_optl = cpu_to_le16(dlen);
+				skb_copy_from_linear_data_offset(skb, 1,
+					      scp->conndata_in.opt_data, dlen);
+			}
+		}
+		dn_nsp_send_link(sk, DN_NOCHANGE, 0);
+		if (!sock_flag(sk, SOCK_DEAD))
+			sk->sk_state_change(sk);
+	}
+
+out:
+	kfree_skb(skb);
+}
+
+static void dn_nsp_conn_ack(struct sock *sk, struct sk_buff *skb)
+{
+	struct dn_scp *scp = DN_SK(sk);
+
+	if (scp->state == DN_CI) {
+		scp->state = DN_CD;
+		scp->persist = 0;
+	}
+
+	kfree_skb(skb);
+}
+
+static void dn_nsp_disc_init(struct sock *sk, struct sk_buff *skb)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	unsigned short reason;
+
+	if (skb->len < 2)
+		goto out;
+
+	reason = le16_to_cpu(*(__le16 *)skb->data);
+	skb_pull(skb, 2);
+
+	scp->discdata_in.opt_status = cpu_to_le16(reason);
+	scp->discdata_in.opt_optl   = 0;
+	memset(scp->discdata_in.opt_data, 0, 16);
+
+	if (skb->len > 0) {
+		u16 dlen = *skb->data;
+		if ((dlen <= 16) && (dlen <= skb->len)) {
+			scp->discdata_in.opt_optl = cpu_to_le16(dlen);
+			skb_copy_from_linear_data_offset(skb, 1, scp->discdata_in.opt_data, dlen);
+		}
+	}
+
+	scp->addrrem = cb->src_port;
+	sk->sk_state = TCP_CLOSE;
+
+	switch (scp->state) {
+	case DN_CI:
+	case DN_CD:
+		scp->state = DN_RJ;
+		sk->sk_err = ECONNREFUSED;
+		break;
+	case DN_RUN:
+		sk->sk_shutdown |= SHUTDOWN_MASK;
+		scp->state = DN_DN;
+		break;
+	case DN_DI:
+		scp->state = DN_DIC;
+		break;
+	}
+
+	if (!sock_flag(sk, SOCK_DEAD)) {
+		if (sk->sk_socket->state != SS_UNCONNECTED)
+			sk->sk_socket->state = SS_DISCONNECTING;
+		sk->sk_state_change(sk);
+	}
+
+	/*
+	 * It appears that its possible for remote machines to send disc
+	 * init messages with no port identifier if we are in the CI and
+	 * possibly also the CD state. Obviously we shouldn't reply with
+	 * a message if we don't know what the end point is.
+	 */
+	if (scp->addrrem) {
+		dn_nsp_send_disc(sk, NSP_DISCCONF, NSP_REASON_DC, GFP_ATOMIC);
+	}
+	scp->persist_fxn = dn_destroy_timer;
+	scp->persist = dn_nsp_persist(sk);
+
+out:
+	kfree_skb(skb);
+}
+
+/*
+ * disc_conf messages are also called no_resources or no_link
+ * messages depending upon the "reason" field.
+ */
+static void dn_nsp_disc_conf(struct sock *sk, struct sk_buff *skb)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	unsigned short reason;
+
+	if (skb->len != 2)
+		goto out;
+
+	reason = le16_to_cpu(*(__le16 *)skb->data);
+
+	sk->sk_state = TCP_CLOSE;
+
+	switch (scp->state) {
+	case DN_CI:
+		scp->state = DN_NR;
+		break;
+	case DN_DR:
+		if (reason == NSP_REASON_DC)
+			scp->state = DN_DRC;
+		if (reason == NSP_REASON_NL)
+			scp->state = DN_CN;
+		break;
+	case DN_DI:
+		scp->state = DN_DIC;
+		break;
+	case DN_RUN:
+		sk->sk_shutdown |= SHUTDOWN_MASK;
+	case DN_CC:
+		scp->state = DN_CN;
+	}
+
+	if (!sock_flag(sk, SOCK_DEAD)) {
+		if (sk->sk_socket->state != SS_UNCONNECTED)
+			sk->sk_socket->state = SS_DISCONNECTING;
+		sk->sk_state_change(sk);
+	}
+
+	scp->persist_fxn = dn_destroy_timer;
+	scp->persist = dn_nsp_persist(sk);
+
+out:
+	kfree_skb(skb);
+}
+
+static void dn_nsp_linkservice(struct sock *sk, struct sk_buff *skb)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	unsigned short segnum;
+	unsigned char lsflags;
+	signed char fcval;
+	int wake_up = 0;
+	char *ptr = skb->data;
+	unsigned char fctype = scp->services_rem & NSP_FC_MASK;
+
+	if (skb->len != 4)
+		goto out;
+
+	segnum = le16_to_cpu(*(__le16 *)ptr);
+	ptr += 2;
+	lsflags = *(unsigned char *)ptr++;
+	fcval = *ptr;
+
+	/*
+	 * Here we ignore erronous packets which should really
+	 * should cause a connection abort. It is not critical
+	 * for now though.
+	 */
+	if (lsflags & 0xf8)
+		goto out;
+
+	if (seq_next(scp->numoth_rcv, segnum)) {
+		seq_add(&scp->numoth_rcv, 1);
+		switch(lsflags & 0x04) { /* FCVAL INT */
+		case 0x00: /* Normal Request */
+			switch(lsflags & 0x03) { /* FCVAL MOD */
+			case 0x00: /* Request count */
+				if (fcval < 0) {
+					unsigned char p_fcval = -fcval;
+					if ((scp->flowrem_dat > p_fcval) &&
+					    (fctype == NSP_FC_SCMC)) {
+						scp->flowrem_dat -= p_fcval;
+					}
+				} else if (fcval > 0) {
+					scp->flowrem_dat += fcval;
+					wake_up = 1;
+				}
+				break;
+			case 0x01: /* Stop outgoing data */
+				scp->flowrem_sw = DN_DONTSEND;
+				break;
+			case 0x02: /* Ok to start again */
+				scp->flowrem_sw = DN_SEND;
+				dn_nsp_output(sk);
+				wake_up = 1;
+			}
+			break;
+		case 0x04: /* Interrupt Request */
+			if (fcval > 0) {
+				scp->flowrem_oth += fcval;
+				wake_up = 1;
+			}
+			break;
+		}
+		if (wake_up && !sock_flag(sk, SOCK_DEAD))
+			sk->sk_state_change(sk);
+	}
+
+	dn_nsp_send_oth_ack(sk);
+
+out:
+	kfree_skb(skb);
+}
+
+/*
+ * Copy of sock_queue_rcv_skb (from sock.h) without
+ * bh_lock_sock() (its already held when this is called) which
+ * also allows data and other data to be queued to a socket.
+ */
+static __inline__ int dn_queue_skb(struct sock *sk, struct sk_buff *skb, int sig, struct sk_buff_head *queue)
+{
+	int err;
+
+	/* Cast skb->rcvbuf to unsigned... It's pointless, but reduces
+	   number of warnings when compiling with -W --ANK
+	 */
+	if (atomic_read(&sk->sk_rmem_alloc) + skb->truesize >=
+	    (unsigned int)sk->sk_rcvbuf) {
+		err = -ENOMEM;
+		goto out;
+	}
+
+	err = sk_filter(sk, skb);
+	if (err)
+		goto out;
+
+	skb_set_owner_r(skb, sk);
+	skb_queue_tail(queue, skb);
+
+	if (!sock_flag(sk, SOCK_DEAD))
+		sk->sk_data_ready(sk);
+out:
+	return err;
+}
+
+static void dn_nsp_otherdata(struct sock *sk, struct sk_buff *skb)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	unsigned short segnum;
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	int queued = 0;
+
+	if (skb->len < 2)
+		goto out;
+
+	cb->segnum = segnum = le16_to_cpu(*(__le16 *)skb->data);
+	skb_pull(skb, 2);
+
+	if (seq_next(scp->numoth_rcv, segnum)) {
+
+		if (dn_queue_skb(sk, skb, SIGURG, &scp->other_receive_queue) == 0) {
+			seq_add(&scp->numoth_rcv, 1);
+			scp->other_report = 0;
+			queued = 1;
+		}
+	}
+
+	dn_nsp_send_oth_ack(sk);
+out:
+	if (!queued)
+		kfree_skb(skb);
+}
+
+static void dn_nsp_data(struct sock *sk, struct sk_buff *skb)
+{
+	int queued = 0;
+	unsigned short segnum;
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	struct dn_scp *scp = DN_SK(sk);
+
+	if (skb->len < 2)
+		goto out;
+
+	cb->segnum = segnum = le16_to_cpu(*(__le16 *)skb->data);
+	skb_pull(skb, 2);
+
+	if (seq_next(scp->numdat_rcv, segnum)) {
+		if (dn_queue_skb(sk, skb, SIGIO, &sk->sk_receive_queue) == 0) {
+			seq_add(&scp->numdat_rcv, 1);
+			queued = 1;
+		}
+
+		if ((scp->flowloc_sw == DN_SEND) && dn_congested(sk)) {
+			scp->flowloc_sw = DN_DONTSEND;
+			dn_nsp_send_link(sk, DN_DONTSEND, 0);
+		}
+	}
+
+	dn_nsp_send_data_ack(sk);
+out:
+	if (!queued)
+		kfree_skb(skb);
+}
+
+/*
+ * If one of our conninit messages is returned, this function
+ * deals with it. It puts the socket into the NO_COMMUNICATION
+ * state.
+ */
+static void dn_returned_conn_init(struct sock *sk, struct sk_buff *skb)
+{
+	struct dn_scp *scp = DN_SK(sk);
+
+	if (scp->state == DN_CI) {
+		scp->state = DN_NC;
+		sk->sk_state = TCP_CLOSE;
+		if (!sock_flag(sk, SOCK_DEAD))
+			sk->sk_state_change(sk);
+	}
+
+	kfree_skb(skb);
+}
+
+static int dn_nsp_no_socket(struct sk_buff *skb, unsigned short reason)
+{
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	int ret = NET_RX_DROP;
+
+	/* Must not reply to returned packets */
+	if (cb->rt_flags & DN_RT_F_RTS)
+		goto out;
+
+	if ((reason != NSP_REASON_OK) && ((cb->nsp_flags & 0x0c) == 0x08)) {
+		switch (cb->nsp_flags & 0x70) {
+		case 0x10:
+		case 0x60: /* (Retransmitted) Connect Init */
+			dn_nsp_return_disc(skb, NSP_DISCINIT, reason);
+			ret = NET_RX_SUCCESS;
+			break;
+		case 0x20: /* Connect Confirm */
+			dn_nsp_return_disc(skb, NSP_DISCCONF, reason);
+			ret = NET_RX_SUCCESS;
+			break;
+		}
+	}
+
+out:
+	kfree_skb(skb);
+	return ret;
+}
+
+static int dn_nsp_rx_packet(struct sk_buff *skb)
+{
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	struct sock *sk = NULL;
+	unsigned char *ptr = (unsigned char *)skb->data;
+	unsigned short reason = NSP_REASON_NL;
+
+	if (!pskb_may_pull(skb, 2))
+		goto free_out;
+
+	skb_reset_transport_header(skb);
+	cb->nsp_flags = *ptr++;
+
+	if (decnet_debug_level & 2)
+		printk(KERN_DEBUG "dn_nsp_rx: Message type 0x%02x\n", (int)cb->nsp_flags);
+
+	if (cb->nsp_flags & 0x83)
+		goto free_out;
+
+	/*
+	 * Filter out conninits and useless packet types
+	 */
+	if ((cb->nsp_flags & 0x0c) == 0x08) {
+		switch (cb->nsp_flags & 0x70) {
+		case 0x00: /* NOP */
+		case 0x70: /* Reserved */
+		case 0x50: /* Reserved, Phase II node init */
+			goto free_out;
+		case 0x10:
+		case 0x60:
+			if (unlikely(cb->rt_flags & DN_RT_F_RTS))
+				goto free_out;
+			sk = dn_find_listener(skb, &reason);
+			goto got_it;
+		}
+	}
+
+	if (!pskb_may_pull(skb, 3))
+		goto free_out;
+
+	/*
+	 * Grab the destination address.
+	 */
+	cb->dst_port = *(__le16 *)ptr;
+	cb->src_port = 0;
+	ptr += 2;
+
+	/*
+	 * If not a connack, grab the source address too.
+	 */
+	if (pskb_may_pull(skb, 5)) {
+		cb->src_port = *(__le16 *)ptr;
+		ptr += 2;
+		skb_pull(skb, 5);
+	}
+
+	/*
+	 * Returned packets...
+	 * Swap src & dst and look up in the normal way.
+	 */
+	if (unlikely(cb->rt_flags & DN_RT_F_RTS)) {
+		__le16 tmp = cb->dst_port;
+		cb->dst_port = cb->src_port;
+		cb->src_port = tmp;
+		tmp = cb->dst;
+		cb->dst = cb->src;
+		cb->src = tmp;
+	}
+
+	/*
+	 * Find the socket to which this skb is destined.
+	 */
+	sk = dn_find_by_skb(skb);
+got_it:
+	if (sk != NULL) {
+		struct dn_scp *scp = DN_SK(sk);
+
+		/* Reset backoff */
+		scp->nsp_rxtshift = 0;
+
+		/*
+		 * We linearize everything except data segments here.
+		 */
+		if (cb->nsp_flags & ~0x60) {
+			if (unlikely(skb_linearize(skb)))
+				goto free_out;
+		}
+
+		return sk_receive_skb(sk, skb, 0);
+	}
+
+	return dn_nsp_no_socket(skb, reason);
+
+free_out:
+	kfree_skb(skb);
+	return NET_RX_DROP;
+}
+
+int dn_nsp_rx(struct sk_buff *skb)
+{
+	return NF_HOOK(NFPROTO_DECNET, NF_DN_LOCAL_IN, skb, skb->dev, NULL,
+		       dn_nsp_rx_packet);
+}
+
+/*
+ * This is the main receive routine for sockets. It is called
+ * from the above when the socket is not busy, and also from
+ * sock_release() when there is a backlog queued up.
+ */
+int dn_nsp_backlog_rcv(struct sock *sk, struct sk_buff *skb)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+
+	if (cb->rt_flags & DN_RT_F_RTS) {
+		if (cb->nsp_flags == 0x18 || cb->nsp_flags == 0x68)
+			dn_returned_conn_init(sk, skb);
+		else
+			kfree_skb(skb);
+		return NET_RX_SUCCESS;
+	}
+
+	/*
+	 * Control packet.
+	 */
+	if ((cb->nsp_flags & 0x0c) == 0x08) {
+		switch (cb->nsp_flags & 0x70) {
+		case 0x10:
+		case 0x60:
+			dn_nsp_conn_init(sk, skb);
+			break;
+		case 0x20:
+			dn_nsp_conn_conf(sk, skb);
+			break;
+		case 0x30:
+			dn_nsp_disc_init(sk, skb);
+			break;
+		case 0x40:
+			dn_nsp_disc_conf(sk, skb);
+			break;
+		}
+
+	} else if (cb->nsp_flags == 0x24) {
+		/*
+		 * Special for connacks, 'cos they don't have
+		 * ack data or ack otherdata info.
+		 */
+		dn_nsp_conn_ack(sk, skb);
+	} else {
+		int other = 1;
+
+		/* both data and ack frames can kick a CC socket into RUN */
+		if ((scp->state == DN_CC) && !sock_flag(sk, SOCK_DEAD)) {
+			scp->state = DN_RUN;
+			sk->sk_state = TCP_ESTABLISHED;
+			sk->sk_state_change(sk);
+		}
+
+		if ((cb->nsp_flags & 0x1c) == 0)
+			other = 0;
+		if (cb->nsp_flags == 0x04)
+			other = 0;
+
+		/*
+		 * Read out ack data here, this applies equally
+		 * to data, other data, link serivce and both
+		 * ack data and ack otherdata.
+		 */
+		dn_process_ack(sk, skb, other);
+
+		/*
+		 * If we've some sort of data here then call a
+		 * suitable routine for dealing with it, otherwise
+		 * the packet is an ack and can be discarded.
+		 */
+		if ((cb->nsp_flags & 0x0c) == 0) {
+
+			if (scp->state != DN_RUN)
+				goto free_out;
+
+			switch (cb->nsp_flags) {
+			case 0x10: /* LS */
+				dn_nsp_linkservice(sk, skb);
+				break;
+			case 0x30: /* OD */
+				dn_nsp_otherdata(sk, skb);
+				break;
+			default:
+				dn_nsp_data(sk, skb);
+			}
+
+		} else { /* Ack, chuck it out here */
+free_out:
+			kfree_skb(skb);
+		}
+	}
+
+	return NET_RX_SUCCESS;
+}
+

net/decnet/dn_nsp_out.c

@@ -0,0 +1,718 @@
+/*
+ * DECnet       An implementation of the DECnet protocol suite for the LINUX
+ *              operating system.  DECnet is implemented using the  BSD Socket
+ *              interface as the means of communication with the user level.
+ *
+ *              DECnet Network Services Protocol (Output)
+ *
+ * Author:      Eduardo Marcelo Serrat <emserrat@geocities.com>
+ *
+ * Changes:
+ *
+ *    Steve Whitehouse:  Split into dn_nsp_in.c and dn_nsp_out.c from
+ *                       original dn_nsp.c.
+ *    Steve Whitehouse:  Updated to work with my new routing architecture.
+ *    Steve Whitehouse:  Added changes from Eduardo Serrat's patches.
+ *    Steve Whitehouse:  Now conninits have the "return" bit set.
+ *    Steve Whitehouse:  Fixes to check alloc'd skbs are non NULL!
+ *                       Moved output state machine into one function
+ *    Steve Whitehouse:  New output state machine
+ *         Paul Koning:  Connect Confirm message fix.
+ *      Eduardo Serrat:  Fix to stop dn_nsp_do_disc() sending malformed packets.
+ *    Steve Whitehouse:  dn_nsp_output() and friends needed a spring clean
+ *    Steve Whitehouse:  Moved dn_nsp_send() in here from route.h
+ */
+
+/******************************************************************************
+    (c) 1995-1998 E.M. Serrat		emserrat@geocities.com
+
+    This program is free software; you can redistribute it and/or modify
+    it under the terms of the GNU General Public License as published by
+    the Free Software Foundation; either version 2 of the License, or
+    any later version.
+
+    This program is distributed in the hope that it will be useful,
+    but WITHOUT ANY WARRANTY; without even the implied warranty of
+    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+    GNU General Public License for more details.
+*******************************************************************************/
+
+#include <linux/errno.h>
+#include <linux/types.h>
+#include <linux/socket.h>
+#include <linux/in.h>
+#include <linux/kernel.h>
+#include <linux/timer.h>
+#include <linux/string.h>
+#include <linux/sockios.h>
+#include <linux/net.h>
+#include <linux/netdevice.h>
+#include <linux/inet.h>
+#include <linux/route.h>
+#include <linux/slab.h>
+#include <net/sock.h>
+#include <linux/fcntl.h>
+#include <linux/mm.h>
+#include <linux/termios.h>
+#include <linux/interrupt.h>
+#include <linux/proc_fs.h>
+#include <linux/stat.h>
+#include <linux/init.h>
+#include <linux/poll.h>
+#include <linux/if_packet.h>
+#include <net/neighbour.h>
+#include <net/dst.h>
+#include <net/flow.h>
+#include <net/dn.h>
+#include <net/dn_nsp.h>
+#include <net/dn_dev.h>
+#include <net/dn_route.h>
+
+
+static int nsp_backoff[NSP_MAXRXTSHIFT + 1] = { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
+
+static void dn_nsp_send(struct sk_buff *skb)
+{
+	struct sock *sk = skb->sk;
+	struct dn_scp *scp = DN_SK(sk);
+	struct dst_entry *dst;
+	struct flowidn fld;
+
+	skb_reset_transport_header(skb);
+	scp->stamp = jiffies;
+
+	dst = sk_dst_check(sk, 0);
+	if (dst) {
+try_again:
+		skb_dst_set(skb, dst);
+		dst_output(skb);
+		return;
+	}
+
+	memset(&fld, 0, sizeof(fld));
+	fld.flowidn_oif = sk->sk_bound_dev_if;
+	fld.saddr = dn_saddr2dn(&scp->addr);
+	fld.daddr = dn_saddr2dn(&scp->peer);
+	dn_sk_ports_copy(&fld, scp);
+	fld.flowidn_proto = DNPROTO_NSP;
+	if (dn_route_output_sock(&sk->sk_dst_cache, &fld, sk, 0) == 0) {
+		dst = sk_dst_get(sk);
+		sk->sk_route_caps = dst->dev->features;
+		goto try_again;
+	}
+
+	sk->sk_err = EHOSTUNREACH;
+	if (!sock_flag(sk, SOCK_DEAD))
+		sk->sk_state_change(sk);
+}
+
+
+/*
+ * If sk == NULL, then we assume that we are supposed to be making
+ * a routing layer skb. If sk != NULL, then we are supposed to be
+ * creating an skb for the NSP layer.
+ *
+ * The eventual aim is for each socket to have a cached header size
+ * for its outgoing packets, and to set hdr from this when sk != NULL.
+ */
+struct sk_buff *dn_alloc_skb(struct sock *sk, int size, gfp_t pri)
+{
+	struct sk_buff *skb;
+	int hdr = 64;
+
+	if ((skb = alloc_skb(size + hdr, pri)) == NULL)
+		return NULL;
+
+	skb->protocol = htons(ETH_P_DNA_RT);
+	skb->pkt_type = PACKET_OUTGOING;
+
+	if (sk)
+		skb_set_owner_w(skb, sk);
+
+	skb_reserve(skb, hdr);
+
+	return skb;
+}
+
+/*
+ * Calculate persist timer based upon the smoothed round
+ * trip time and the variance. Backoff according to the
+ * nsp_backoff[] array.
+ */
+unsigned long dn_nsp_persist(struct sock *sk)
+{
+	struct dn_scp *scp = DN_SK(sk);
+
+	unsigned long t = ((scp->nsp_srtt >> 2) + scp->nsp_rttvar) >> 1;
+
+	t *= nsp_backoff[scp->nsp_rxtshift];
+
+	if (t < HZ) t = HZ;
+	if (t > (600*HZ)) t = (600*HZ);
+
+	if (scp->nsp_rxtshift < NSP_MAXRXTSHIFT)
+		scp->nsp_rxtshift++;
+
+	/* printk(KERN_DEBUG "rxtshift %lu, t=%lu\n", scp->nsp_rxtshift, t); */
+
+	return t;
+}
+
+/*
+ * This is called each time we get an estimate for the rtt
+ * on the link.
+ */
+static void dn_nsp_rtt(struct sock *sk, long rtt)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	long srtt = (long)scp->nsp_srtt;
+	long rttvar = (long)scp->nsp_rttvar;
+	long delta;
+
+	/*
+	 * If the jiffies clock flips over in the middle of timestamp
+	 * gathering this value might turn out negative, so we make sure
+	 * that is it always positive here.
+	 */
+	if (rtt < 0)
+		rtt = -rtt;
+	/*
+	 * Add new rtt to smoothed average
+	 */
+	delta = ((rtt << 3) - srtt);
+	srtt += (delta >> 3);
+	if (srtt >= 1)
+		scp->nsp_srtt = (unsigned long)srtt;
+	else
+		scp->nsp_srtt = 1;
+
+	/*
+	 * Add new rtt varience to smoothed varience
+	 */
+	delta >>= 1;
+	rttvar += ((((delta>0)?(delta):(-delta)) - rttvar) >> 2);
+	if (rttvar >= 1)
+		scp->nsp_rttvar = (unsigned long)rttvar;
+	else
+		scp->nsp_rttvar = 1;
+
+	/* printk(KERN_DEBUG "srtt=%lu rttvar=%lu\n", scp->nsp_srtt, scp->nsp_rttvar); */
+}
+
+/**
+ * dn_nsp_clone_and_send - Send a data packet by cloning it
+ * @skb: The packet to clone and transmit
+ * @gfp: memory allocation flag
+ *
+ * Clone a queued data or other data packet and transmit it.
+ *
+ * Returns: The number of times the packet has been sent previously
+ */
+static inline unsigned int dn_nsp_clone_and_send(struct sk_buff *skb,
+					     gfp_t gfp)
+{
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	struct sk_buff *skb2;
+	int ret = 0;
+
+	if ((skb2 = skb_clone(skb, gfp)) != NULL) {
+		ret = cb->xmit_count;
+		cb->xmit_count++;
+		cb->stamp = jiffies;
+		skb2->sk = skb->sk;
+		dn_nsp_send(skb2);
+	}
+
+	return ret;
+}
+
+/**
+ * dn_nsp_output - Try and send something from socket queues
+ * @sk: The socket whose queues are to be investigated
+ *
+ * Try and send the packet on the end of the data and other data queues.
+ * Other data gets priority over data, and if we retransmit a packet we
+ * reduce the window by dividing it in two.
+ *
+ */
+void dn_nsp_output(struct sock *sk)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	struct sk_buff *skb;
+	unsigned int reduce_win = 0;
+
+	/*
+	 * First we check for otherdata/linkservice messages
+	 */
+	if ((skb = skb_peek(&scp->other_xmit_queue)) != NULL)
+		reduce_win = dn_nsp_clone_and_send(skb, GFP_ATOMIC);
+
+	/*
+	 * If we may not send any data, we don't.
+	 * If we are still trying to get some other data down the
+	 * channel, we don't try and send any data.
+	 */
+	if (reduce_win || (scp->flowrem_sw != DN_SEND))
+		goto recalc_window;
+
+	if ((skb = skb_peek(&scp->data_xmit_queue)) != NULL)
+		reduce_win = dn_nsp_clone_and_send(skb, GFP_ATOMIC);
+
+	/*
+	 * If we've sent any frame more than once, we cut the
+	 * send window size in half. There is always a minimum
+	 * window size of one available.
+	 */
+recalc_window:
+	if (reduce_win) {
+		scp->snd_window >>= 1;
+		if (scp->snd_window < NSP_MIN_WINDOW)
+			scp->snd_window = NSP_MIN_WINDOW;
+	}
+}
+
+int dn_nsp_xmit_timeout(struct sock *sk)
+{
+	struct dn_scp *scp = DN_SK(sk);
+
+	dn_nsp_output(sk);
+
+	if (!skb_queue_empty(&scp->data_xmit_queue) ||
+	    !skb_queue_empty(&scp->other_xmit_queue))
+		scp->persist = dn_nsp_persist(sk);
+
+	return 0;
+}
+
+static inline __le16 *dn_mk_common_header(struct dn_scp *scp, struct sk_buff *skb, unsigned char msgflag, int len)
+{
+	unsigned char *ptr = skb_push(skb, len);
+
+	BUG_ON(len < 5);
+
+	*ptr++ = msgflag;
+	*((__le16 *)ptr) = scp->addrrem;
+	ptr += 2;
+	*((__le16 *)ptr) = scp->addrloc;
+	ptr += 2;
+	return (__le16 __force *)ptr;
+}
+
+static __le16 *dn_mk_ack_header(struct sock *sk, struct sk_buff *skb, unsigned char msgflag, int hlen, int other)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	unsigned short acknum = scp->numdat_rcv & 0x0FFF;
+	unsigned short ackcrs = scp->numoth_rcv & 0x0FFF;
+	__le16 *ptr;
+
+	BUG_ON(hlen < 9);
+
+	scp->ackxmt_dat = acknum;
+	scp->ackxmt_oth = ackcrs;
+	acknum |= 0x8000;
+	ackcrs |= 0x8000;
+
+	/* If this is an "other data/ack" message, swap acknum and ackcrs */
+	if (other) {
+		unsigned short tmp = acknum;
+		acknum = ackcrs;
+		ackcrs = tmp;
+	}
+
+	/* Set "cross subchannel" bit in ackcrs */
+	ackcrs |= 0x2000;
+
+	ptr = dn_mk_common_header(scp, skb, msgflag, hlen);
+
+	*ptr++ = cpu_to_le16(acknum);
+	*ptr++ = cpu_to_le16(ackcrs);
+
+	return ptr;
+}
+
+static __le16 *dn_nsp_mk_data_header(struct sock *sk, struct sk_buff *skb, int oth)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	__le16 *ptr = dn_mk_ack_header(sk, skb, cb->nsp_flags, 11, oth);
+
+	if (unlikely(oth)) {
+		cb->segnum = scp->numoth;
+		seq_add(&scp->numoth, 1);
+	} else {
+		cb->segnum = scp->numdat;
+		seq_add(&scp->numdat, 1);
+	}
+	*(ptr++) = cpu_to_le16(cb->segnum);
+
+	return ptr;
+}
+
+void dn_nsp_queue_xmit(struct sock *sk, struct sk_buff *skb,
+			gfp_t gfp, int oth)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	unsigned long t = ((scp->nsp_srtt >> 2) + scp->nsp_rttvar) >> 1;
+
+	cb->xmit_count = 0;
+	dn_nsp_mk_data_header(sk, skb, oth);
+
+	/*
+	 * Slow start: If we have been idle for more than
+	 * one RTT, then reset window to min size.
+	 */
+	if ((jiffies - scp->stamp) > t)
+		scp->snd_window = NSP_MIN_WINDOW;
+
+	if (oth)
+		skb_queue_tail(&scp->other_xmit_queue, skb);
+	else
+		skb_queue_tail(&scp->data_xmit_queue, skb);
+
+	if (scp->flowrem_sw != DN_SEND)
+		return;
+
+	dn_nsp_clone_and_send(skb, gfp);
+}
+
+
+int dn_nsp_check_xmit_queue(struct sock *sk, struct sk_buff *skb, struct sk_buff_head *q, unsigned short acknum)
+{
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	struct dn_scp *scp = DN_SK(sk);
+	struct sk_buff *skb2, *n, *ack = NULL;
+	int wakeup = 0;
+	int try_retrans = 0;
+	unsigned long reftime = cb->stamp;
+	unsigned long pkttime;
+	unsigned short xmit_count;
+	unsigned short segnum;
+
+	skb_queue_walk_safe(q, skb2, n) {
+		struct dn_skb_cb *cb2 = DN_SKB_CB(skb2);
+
+		if (dn_before_or_equal(cb2->segnum, acknum))
+			ack = skb2;
+
+		/* printk(KERN_DEBUG "ack: %s %04x %04x\n", ack ? "ACK" : "SKIP", (int)cb2->segnum, (int)acknum); */
+
+		if (ack == NULL)
+			continue;
+
+		/* printk(KERN_DEBUG "check_xmit_queue: %04x, %d\n", acknum, cb2->xmit_count); */
+
+		/* Does _last_ packet acked have xmit_count > 1 */
+		try_retrans = 0;
+		/* Remember to wake up the sending process */
+		wakeup = 1;
+		/* Keep various statistics */
+		pkttime = cb2->stamp;
+		xmit_count = cb2->xmit_count;
+		segnum = cb2->segnum;
+		/* Remove and drop ack'ed packet */
+		skb_unlink(ack, q);
+		kfree_skb(ack);
+		ack = NULL;
+
+		/*
+		 * We don't expect to see acknowledgements for packets we
+		 * haven't sent yet.
+		 */
+		WARN_ON(xmit_count == 0);
+
+		/*
+		 * If the packet has only been sent once, we can use it
+		 * to calculate the RTT and also open the window a little
+		 * further.
+		 */
+		if (xmit_count == 1) {
+			if (dn_equal(segnum, acknum))
+				dn_nsp_rtt(sk, (long)(pkttime - reftime));
+
+			if (scp->snd_window < scp->max_window)
+				scp->snd_window++;
+		}
+
+		/*
+		 * Packet has been sent more than once. If this is the last
+		 * packet to be acknowledged then we want to send the next
+		 * packet in the send queue again (assumes the remote host does
+		 * go-back-N error control).
+		 */
+		if (xmit_count > 1)
+			try_retrans = 1;
+	}
+
+	if (try_retrans)
+		dn_nsp_output(sk);
+
+	return wakeup;
+}
+
+void dn_nsp_send_data_ack(struct sock *sk)
+{
+	struct sk_buff *skb = NULL;
+
+	if ((skb = dn_alloc_skb(sk, 9, GFP_ATOMIC)) == NULL)
+		return;
+
+	skb_reserve(skb, 9);
+	dn_mk_ack_header(sk, skb, 0x04, 9, 0);
+	dn_nsp_send(skb);
+}
+
+void dn_nsp_send_oth_ack(struct sock *sk)
+{
+	struct sk_buff *skb = NULL;
+
+	if ((skb = dn_alloc_skb(sk, 9, GFP_ATOMIC)) == NULL)
+		return;
+
+	skb_reserve(skb, 9);
+	dn_mk_ack_header(sk, skb, 0x14, 9, 1);
+	dn_nsp_send(skb);
+}
+
+
+void dn_send_conn_ack (struct sock *sk)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	struct sk_buff *skb = NULL;
+	struct nsp_conn_ack_msg *msg;
+
+	if ((skb = dn_alloc_skb(sk, 3, sk->sk_allocation)) == NULL)
+		return;
+
+	msg = (struct nsp_conn_ack_msg *)skb_put(skb, 3);
+	msg->msgflg = 0x24;
+	msg->dstaddr = scp->addrrem;
+
+	dn_nsp_send(skb);
+}
+
+void dn_nsp_delayed_ack(struct sock *sk)
+{
+	struct dn_scp *scp = DN_SK(sk);
+
+	if (scp->ackxmt_oth != scp->numoth_rcv)
+		dn_nsp_send_oth_ack(sk);
+
+	if (scp->ackxmt_dat != scp->numdat_rcv)
+		dn_nsp_send_data_ack(sk);
+}
+
+static int dn_nsp_retrans_conn_conf(struct sock *sk)
+{
+	struct dn_scp *scp = DN_SK(sk);
+
+	if (scp->state == DN_CC)
+		dn_send_conn_conf(sk, GFP_ATOMIC);
+
+	return 0;
+}
+
+void dn_send_conn_conf(struct sock *sk, gfp_t gfp)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	struct sk_buff *skb = NULL;
+	struct nsp_conn_init_msg *msg;
+	__u8 len = (__u8)le16_to_cpu(scp->conndata_out.opt_optl);
+
+	if ((skb = dn_alloc_skb(sk, 50 + len, gfp)) == NULL)
+		return;
+
+	msg = (struct nsp_conn_init_msg *)skb_put(skb, sizeof(*msg));
+	msg->msgflg = 0x28;
+	msg->dstaddr = scp->addrrem;
+	msg->srcaddr = scp->addrloc;
+	msg->services = scp->services_loc;
+	msg->info = scp->info_loc;
+	msg->segsize = cpu_to_le16(scp->segsize_loc);
+
+	*skb_put(skb,1) = len;
+
+	if (len > 0)
+		memcpy(skb_put(skb, len), scp->conndata_out.opt_data, len);
+
+
+	dn_nsp_send(skb);
+
+	scp->persist = dn_nsp_persist(sk);
+	scp->persist_fxn = dn_nsp_retrans_conn_conf;
+}
+
+
+static __inline__ void dn_nsp_do_disc(struct sock *sk, unsigned char msgflg,
+			unsigned short reason, gfp_t gfp,
+			struct dst_entry *dst,
+			int ddl, unsigned char *dd, __le16 rem, __le16 loc)
+{
+	struct sk_buff *skb = NULL;
+	int size = 7 + ddl + ((msgflg == NSP_DISCINIT) ? 1 : 0);
+	unsigned char *msg;
+
+	if ((dst == NULL) || (rem == 0)) {
+		net_dbg_ratelimited("DECnet: dn_nsp_do_disc: BUG! Please report this to SteveW@ACM.org rem=%u dst=%p\n",
+				    le16_to_cpu(rem), dst);
+		return;
+	}
+
+	if ((skb = dn_alloc_skb(sk, size, gfp)) == NULL)
+		return;
+
+	msg = skb_put(skb, size);
+	*msg++ = msgflg;
+	*(__le16 *)msg = rem;
+	msg += 2;
+	*(__le16 *)msg = loc;
+	msg += 2;
+	*(__le16 *)msg = cpu_to_le16(reason);
+	msg += 2;
+	if (msgflg == NSP_DISCINIT)
+		*msg++ = ddl;
+
+	if (ddl) {
+		memcpy(msg, dd, ddl);
+	}
+
+	/*
+	 * This doesn't go via the dn_nsp_send() function since we need
+	 * to be able to send disc packets out which have no socket
+	 * associations.
+	 */
+	skb_dst_set(skb, dst_clone(dst));
+	dst_output(skb);
+}
+
+
+void dn_nsp_send_disc(struct sock *sk, unsigned char msgflg,
+			unsigned short reason, gfp_t gfp)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	int ddl = 0;
+
+	if (msgflg == NSP_DISCINIT)
+		ddl = le16_to_cpu(scp->discdata_out.opt_optl);
+
+	if (reason == 0)
+		reason = le16_to_cpu(scp->discdata_out.opt_status);
+
+	dn_nsp_do_disc(sk, msgflg, reason, gfp, __sk_dst_get(sk), ddl,
+		scp->discdata_out.opt_data, scp->addrrem, scp->addrloc);
+}
+
+
+void dn_nsp_return_disc(struct sk_buff *skb, unsigned char msgflg,
+			unsigned short reason)
+{
+	struct dn_skb_cb *cb = DN_SKB_CB(skb);
+	int ddl = 0;
+	gfp_t gfp = GFP_ATOMIC;
+
+	dn_nsp_do_disc(NULL, msgflg, reason, gfp, skb_dst(skb), ddl,
+			NULL, cb->src_port, cb->dst_port);
+}
+
+
+void dn_nsp_send_link(struct sock *sk, unsigned char lsflags, char fcval)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	struct sk_buff *skb;
+	unsigned char *ptr;
+	gfp_t gfp = GFP_ATOMIC;
+
+	if ((skb = dn_alloc_skb(sk, DN_MAX_NSP_DATA_HEADER + 2, gfp)) == NULL)
+		return;
+
+	skb_reserve(skb, DN_MAX_NSP_DATA_HEADER);
+	ptr = skb_put(skb, 2);
+	DN_SKB_CB(skb)->nsp_flags = 0x10;
+	*ptr++ = lsflags;
+	*ptr = fcval;
+
+	dn_nsp_queue_xmit(sk, skb, gfp, 1);
+
+	scp->persist = dn_nsp_persist(sk);
+	scp->persist_fxn = dn_nsp_xmit_timeout;
+}
+
+static int dn_nsp_retrans_conninit(struct sock *sk)
+{
+	struct dn_scp *scp = DN_SK(sk);
+
+	if (scp->state == DN_CI)
+		dn_nsp_send_conninit(sk, NSP_RCI);
+
+	return 0;
+}
+
+void dn_nsp_send_conninit(struct sock *sk, unsigned char msgflg)
+{
+	struct dn_scp *scp = DN_SK(sk);
+	struct nsp_conn_init_msg *msg;
+	unsigned char aux;
+	unsigned char menuver;
+	struct dn_skb_cb *cb;
+	unsigned char type = 1;
+	gfp_t allocation = (msgflg == NSP_CI) ? sk->sk_allocation : GFP_ATOMIC;
+	struct sk_buff *skb = dn_alloc_skb(sk, 200, allocation);
+
+	if (!skb)
+		return;
+
+	cb  = DN_SKB_CB(skb);
+	msg = (struct nsp_conn_init_msg *)skb_put(skb,sizeof(*msg));
+
+	msg->msgflg	= msgflg;
+	msg->dstaddr	= 0x0000;		/* Remote Node will assign it*/
+
+	msg->srcaddr	= scp->addrloc;
+	msg->services	= scp->services_loc;	/* Requested flow control    */
+	msg->info	= scp->info_loc;	/* Version Number            */
+	msg->segsize	= cpu_to_le16(scp->segsize_loc);	/* Max segment size  */
+
+	if (scp->peer.sdn_objnum)
+		type = 0;
+
+	skb_put(skb, dn_sockaddr2username(&scp->peer,
+					  skb_tail_pointer(skb), type));
+	skb_put(skb, dn_sockaddr2username(&scp->addr,
+					  skb_tail_pointer(skb), 2));
+
+	menuver = DN_MENUVER_ACC | DN_MENUVER_USR;
+	if (scp->peer.sdn_flags & SDF_PROXY)
+		menuver |= DN_MENUVER_PRX;
+	if (scp->peer.sdn_flags & SDF_UICPROXY)
+		menuver |= DN_MENUVER_UIC;
+
+	*skb_put(skb, 1) = menuver;	/* Menu Version		*/
+
+	aux = scp->accessdata.acc_userl;
+	*skb_put(skb, 1) = aux;
+	if (aux > 0)
+		memcpy(skb_put(skb, aux), scp->accessdata.acc_user, aux);
+
+	aux = scp->accessdata.acc_passl;
+	*skb_put(skb, 1) = aux;
+	if (aux > 0)
+		memcpy(skb_put(skb, aux), scp->accessdata.acc_pass, aux);
+
+	aux = scp->accessdata.acc_accl;
+	*skb_put(skb, 1) = aux;
+	if (aux > 0)
+		memcpy(skb_put(skb, aux), scp->accessdata.acc_acc, aux);
+
+	aux = (__u8)le16_to_cpu(scp->conndata_out.opt_optl);
+	*skb_put(skb, 1) = aux;
+	if (aux > 0)
+		memcpy(skb_put(skb, aux), scp->conndata_out.opt_data, aux);
+
+	scp->persist = dn_nsp_persist(sk);
+	scp->persist_fxn = dn_nsp_retrans_conninit;
+
+	cb->rt_flags = DN_RT_F_RQR;
+
+	dn_nsp_send(skb);
+}
+

net/decnet/dn_rules.c

@@ -0,0 +1,255 @@
+
+/*
+ * DECnet       An implementation of the DECnet protocol suite for the LINUX
+ *              operating system.  DECnet is implemented using the  BSD Socket
+ *              interface as the means of communication with the user level.
+ *
+ *              DECnet Routing Forwarding Information Base (Rules)
+ *
+ * Author:      Steve Whitehouse <SteveW@ACM.org>
+ *              Mostly copied from Alexey Kuznetsov's ipv4/fib_rules.c
+ *
+ *
+ * Changes:
+ *              Steve Whitehouse <steve@chygwyn.com>
+ *              Updated for Thomas Graf's generic rules
+ *
+ */
+#include <linux/net.h>
+#include <linux/init.h>
+#include <linux/netlink.h>
+#include <linux/rtnetlink.h>
+#include <linux/netdevice.h>
+#include <linux/spinlock.h>
+#include <linux/list.h>
+#include <linux/rcupdate.h>
+#include <linux/export.h>
+#include <net/neighbour.h>
+#include <net/dst.h>
+#include <net/flow.h>
+#include <net/fib_rules.h>
+#include <net/dn.h>
+#include <net/dn_fib.h>
+#include <net/dn_neigh.h>
+#include <net/dn_dev.h>
+#include <net/dn_route.h>
+
+static struct fib_rules_ops *dn_fib_rules_ops;
+
+struct dn_fib_rule
+{
+	struct fib_rule		common;
+	unsigned char		dst_len;
+	unsigned char		src_len;
+	__le16			src;
+	__le16			srcmask;
+	__le16			dst;
+	__le16			dstmask;
+	__le16			srcmap;
+	u8			flags;
+};
+
+
+int dn_fib_lookup(struct flowidn *flp, struct dn_fib_res *res)
+{
+	struct fib_lookup_arg arg = {
+		.result = res,
+	};
+	int err;
+
+	err = fib_rules_lookup(dn_fib_rules_ops,
+			       flowidn_to_flowi(flp), 0, &arg);
+	res->r = arg.rule;
+
+	return err;
+}
+
+static int dn_fib_rule_action(struct fib_rule *rule, struct flowi *flp,
+			      int flags, struct fib_lookup_arg *arg)
+{
+	struct flowidn *fld = &flp->u.dn;
+	int err = -EAGAIN;
+	struct dn_fib_table *tbl;
+
+	switch(rule->action) {
+	case FR_ACT_TO_TBL:
+		break;
+
+	case FR_ACT_UNREACHABLE:
+		err = -ENETUNREACH;
+		goto errout;
+
+	case FR_ACT_PROHIBIT:
+		err = -EACCES;
+		goto errout;
+
+	case FR_ACT_BLACKHOLE:
+	default:
+		err = -EINVAL;
+		goto errout;
+	}
+
+	tbl = dn_fib_get_table(rule->table, 0);
+	if (tbl == NULL)
+		goto errout;
+
+	err = tbl->lookup(tbl, fld, (struct dn_fib_res *)arg->result);
+	if (err > 0)
+		err = -EAGAIN;
+errout:
+	return err;
+}
+
+static const struct nla_policy dn_fib_rule_policy[FRA_MAX+1] = {
+	FRA_GENERIC_POLICY,
+};
+
+static int dn_fib_rule_match(struct fib_rule *rule, struct flowi *fl, int flags)
+{
+	struct dn_fib_rule *r = (struct dn_fib_rule *)rule;
+	struct flowidn *fld = &fl->u.dn;
+	__le16 daddr = fld->daddr;
+	__le16 saddr = fld->saddr;
+
+	if (((saddr ^ r->src) & r->srcmask) ||
+	    ((daddr ^ r->dst) & r->dstmask))
+		return 0;
+
+	return 1;
+}
+
+static int dn_fib_rule_configure(struct fib_rule *rule, struct sk_buff *skb,
+				 struct fib_rule_hdr *frh,
+				 struct nlattr **tb)
+{
+	int err = -EINVAL;
+	struct dn_fib_rule *r = (struct dn_fib_rule *)rule;
+
+	if (frh->tos)
+		goto  errout;
+
+	if (rule->table == RT_TABLE_UNSPEC) {
+		if (rule->action == FR_ACT_TO_TBL) {
+			struct dn_fib_table *table;
+
+			table = dn_fib_empty_table();
+			if (table == NULL) {
+				err = -ENOBUFS;
+				goto errout;
+			}
+
+			rule->table = table->n;
+		}
+	}
+
+	if (frh->src_len)
+		r->src = nla_get_le16(tb[FRA_SRC]);
+
+	if (frh->dst_len)
+		r->dst = nla_get_le16(tb[FRA_DST]);
+
+	r->src_len = frh->src_len;
+	r->srcmask = dnet_make_mask(r->src_len);
+	r->dst_len = frh->dst_len;
+	r->dstmask = dnet_make_mask(r->dst_len);
+	err = 0;
+errout:
+	return err;
+}
+
+static int dn_fib_rule_compare(struct fib_rule *rule, struct fib_rule_hdr *frh,
+			       struct nlattr **tb)
+{
+	struct dn_fib_rule *r = (struct dn_fib_rule *)rule;
+
+	if (frh->src_len && (r->src_len != frh->src_len))
+		return 0;
+
+	if (frh->dst_len && (r->dst_len != frh->dst_len))
+		return 0;
+
+	if (frh->src_len && (r->src != nla_get_le16(tb[FRA_SRC])))
+		return 0;
+
+	if (frh->dst_len && (r->dst != nla_get_le16(tb[FRA_DST])))
+		return 0;
+
+	return 1;
+}
+
+unsigned int dnet_addr_type(__le16 addr)
+{
+	struct flowidn fld = { .daddr = addr };
+	struct dn_fib_res res;
+	unsigned int ret = RTN_UNICAST;
+	struct dn_fib_table *tb = dn_fib_get_table(RT_TABLE_LOCAL, 0);
+
+	res.r = NULL;
+
+	if (tb) {
+		if (!tb->lookup(tb, &fld, &res)) {
+			ret = res.type;
+			dn_fib_res_put(&res);
+		}
+	}
+	return ret;
+}
+
+static int dn_fib_rule_fill(struct fib_rule *rule, struct sk_buff *skb,
+			    struct fib_rule_hdr *frh)
+{
+	struct dn_fib_rule *r = (struct dn_fib_rule *)rule;
+
+	frh->dst_len = r->dst_len;
+	frh->src_len = r->src_len;
+	frh->tos = 0;
+
+	if ((r->dst_len &&
+	     nla_put_le16(skb, FRA_DST, r->dst)) ||
+	    (r->src_len &&
+	     nla_put_le16(skb, FRA_SRC, r->src)))
+		goto nla_put_failure;
+	return 0;
+
+nla_put_failure:
+	return -ENOBUFS;
+}
+
+static void dn_fib_rule_flush_cache(struct fib_rules_ops *ops)
+{
+	dn_rt_cache_flush(-1);
+}
+
+static const struct fib_rules_ops __net_initconst dn_fib_rules_ops_template = {
+	.family		= AF_DECnet,
+	.rule_size	= sizeof(struct dn_fib_rule),
+	.addr_size	= sizeof(u16),
+	.action		= dn_fib_rule_action,
+	.match		= dn_fib_rule_match,
+	.configure	= dn_fib_rule_configure,
+	.compare	= dn_fib_rule_compare,
+	.fill		= dn_fib_rule_fill,
+	.default_pref	= fib_default_rule_pref,
+	.flush_cache	= dn_fib_rule_flush_cache,
+	.nlgroup	= RTNLGRP_DECnet_RULE,
+	.policy		= dn_fib_rule_policy,
+	.owner		= THIS_MODULE,
+	.fro_net	= &init_net,
+};
+
+void __init dn_fib_rules_init(void)
+{
+	dn_fib_rules_ops =
+		fib_rules_register(&dn_fib_rules_ops_template, &init_net);
+	BUG_ON(IS_ERR(dn_fib_rules_ops));
+	BUG_ON(fib_default_rule_add(dn_fib_rules_ops, 0x7fff,
+			            RT_TABLE_MAIN, 0));
+}
+
+void __exit dn_fib_rules_cleanup(void)
+{
+	fib_rules_unregister(dn_fib_rules_ops);
+	rcu_barrier();
+}
+
+

net/decnet/dn_table.c

@@ -0,0 +1,923 @@
+/*
+ * DECnet       An implementation of the DECnet protocol suite for the LINUX
+ *              operating system.  DECnet is implemented using the  BSD Socket
+ *              interface as the means of communication with the user level.
+ *
+ *              DECnet Routing Forwarding Information Base (Routing Tables)
+ *
+ * Author:      Steve Whitehouse <SteveW@ACM.org>
+ *              Mostly copied from the IPv4 routing code
+ *
+ *
+ * Changes:
+ *
+ */
+#include <linux/string.h>
+#include <linux/net.h>
+#include <linux/socket.h>
+#include <linux/slab.h>
+#include <linux/sockios.h>
+#include <linux/init.h>
+#include <linux/skbuff.h>
+#include <linux/rtnetlink.h>
+#include <linux/proc_fs.h>
+#include <linux/netdevice.h>
+#include <linux/timer.h>
+#include <linux/spinlock.h>
+#include <linux/atomic.h>
+#include <asm/uaccess.h>
+#include <linux/route.h> /* RTF_xxx */
+#include <net/neighbour.h>
+#include <net/netlink.h>
+#include <net/dst.h>
+#include <net/flow.h>
+#include <net/fib_rules.h>
+#include <net/dn.h>
+#include <net/dn_route.h>
+#include <net/dn_fib.h>
+#include <net/dn_neigh.h>
+#include <net/dn_dev.h>
+
+struct dn_zone
+{
+	struct dn_zone		*dz_next;
+	struct dn_fib_node 	**dz_hash;
+	int			dz_nent;
+	int			dz_divisor;
+	u32			dz_hashmask;
+#define DZ_HASHMASK(dz)	((dz)->dz_hashmask)
+	int			dz_order;
+	__le16			dz_mask;
+#define DZ_MASK(dz)	((dz)->dz_mask)
+};
+
+struct dn_hash
+{
+	struct dn_zone	*dh_zones[17];
+	struct dn_zone	*dh_zone_list;
+};
+
+#define dz_key_0(key)		((key).datum = 0)
+
+#define for_nexthops(fi) { int nhsel; const struct dn_fib_nh *nh;\
+	for(nhsel = 0, nh = (fi)->fib_nh; nhsel < (fi)->fib_nhs; nh++, nhsel++)
+
+#define endfor_nexthops(fi) }
+
+#define DN_MAX_DIVISOR 1024
+#define DN_S_ZOMBIE 1
+#define DN_S_ACCESSED 2
+
+#define DN_FIB_SCAN(f, fp) \
+for( ; ((f) = *(fp)) != NULL; (fp) = &(f)->fn_next)
+
+#define DN_FIB_SCAN_KEY(f, fp, key) \
+for( ; ((f) = *(fp)) != NULL && dn_key_eq((f)->fn_key, (key)); (fp) = &(f)->fn_next)
+
+#define RT_TABLE_MIN 1
+#define DN_FIB_TABLE_HASHSZ 256
+static struct hlist_head dn_fib_table_hash[DN_FIB_TABLE_HASHSZ];
+static DEFINE_RWLOCK(dn_fib_tables_lock);
+
+static struct kmem_cache *dn_hash_kmem __read_mostly;
+static int dn_fib_hash_zombies;
+
+static inline dn_fib_idx_t dn_hash(dn_fib_key_t key, struct dn_zone *dz)
+{
+	u16 h = le16_to_cpu(key.datum)>>(16 - dz->dz_order);
+	h ^= (h >> 10);
+	h ^= (h >> 6);
+	h &= DZ_HASHMASK(dz);
+	return *(dn_fib_idx_t *)&h;
+}
+
+static inline dn_fib_key_t dz_key(__le16 dst, struct dn_zone *dz)
+{
+	dn_fib_key_t k;
+	k.datum = dst & DZ_MASK(dz);
+	return k;
+}
+
+static inline struct dn_fib_node **dn_chain_p(dn_fib_key_t key, struct dn_zone *dz)
+{
+	return &dz->dz_hash[dn_hash(key, dz).datum];
+}
+
+static inline struct dn_fib_node *dz_chain(dn_fib_key_t key, struct dn_zone *dz)
+{
+	return dz->dz_hash[dn_hash(key, dz).datum];
+}
+
+static inline int dn_key_eq(dn_fib_key_t a, dn_fib_key_t b)
+{
+	return a.datum == b.datum;
+}
+
+static inline int dn_key_leq(dn_fib_key_t a, dn_fib_key_t b)
+{
+	return a.datum <= b.datum;
+}
+
+static inline void dn_rebuild_zone(struct dn_zone *dz,
+				   struct dn_fib_node **old_ht,
+				   int old_divisor)
+{
+	struct dn_fib_node *f, **fp, *next;
+	int i;
+
+	for(i = 0; i < old_divisor; i++) {
+		for(f = old_ht[i]; f; f = next) {
+			next = f->fn_next;
+			for(fp = dn_chain_p(f->fn_key, dz);
+				*fp && dn_key_leq((*fp)->fn_key, f->fn_key);
+				fp = &(*fp)->fn_next)
+				/* NOTHING */;
+			f->fn_next = *fp;
+			*fp = f;
+		}
+	}
+}
+
+static void dn_rehash_zone(struct dn_zone *dz)
+{
+	struct dn_fib_node **ht, **old_ht;
+	int old_divisor, new_divisor;
+	u32 new_hashmask;
+
+	old_divisor = dz->dz_divisor;
+
+	switch (old_divisor) {
+	case 16:
+		new_divisor = 256;
+		new_hashmask = 0xFF;
+		break;
+	default:
+		printk(KERN_DEBUG "DECnet: dn_rehash_zone: BUG! %d\n",
+		       old_divisor);
+	case 256:
+		new_divisor = 1024;
+		new_hashmask = 0x3FF;
+		break;
+	}
+
+	ht = kcalloc(new_divisor, sizeof(struct dn_fib_node*), GFP_KERNEL);
+	if (ht == NULL)
+		return;
+
+	write_lock_bh(&dn_fib_tables_lock);
+	old_ht = dz->dz_hash;
+	dz->dz_hash = ht;
+	dz->dz_hashmask = new_hashmask;
+	dz->dz_divisor = new_divisor;
+	dn_rebuild_zone(dz, old_ht, old_divisor);
+	write_unlock_bh(&dn_fib_tables_lock);
+	kfree(old_ht);
+}
+
+static void dn_free_node(struct dn_fib_node *f)
+{
+	dn_fib_release_info(DN_FIB_INFO(f));
+	kmem_cache_free(dn_hash_kmem, f);
+}
+
+
+static struct dn_zone *dn_new_zone(struct dn_hash *table, int z)
+{
+	int i;
+	struct dn_zone *dz = kzalloc(sizeof(struct dn_zone), GFP_KERNEL);
+	if (!dz)
+		return NULL;
+
+	if (z) {
+		dz->dz_divisor = 16;
+		dz->dz_hashmask = 0x0F;
+	} else {
+		dz->dz_divisor = 1;
+		dz->dz_hashmask = 0;
+	}
+
+	dz->dz_hash = kcalloc(dz->dz_divisor, sizeof(struct dn_fib_node *), GFP_KERNEL);
+	if (!dz->dz_hash) {
+		kfree(dz);
+		return NULL;
+	}
+
+	dz->dz_order = z;
+	dz->dz_mask = dnet_make_mask(z);
+
+	for(i = z + 1; i <= 16; i++)
+		if (table->dh_zones[i])
+			break;
+
+	write_lock_bh(&dn_fib_tables_lock);
+	if (i>16) {
+		dz->dz_next = table->dh_zone_list;
+		table->dh_zone_list = dz;
+	} else {
+		dz->dz_next = table->dh_zones[i]->dz_next;
+		table->dh_zones[i]->dz_next = dz;
+	}
+	table->dh_zones[z] = dz;
+	write_unlock_bh(&dn_fib_tables_lock);
+	return dz;
+}
+
+
+static int dn_fib_nh_match(struct rtmsg *r, struct nlmsghdr *nlh, struct nlattr *attrs[], struct dn_fib_info *fi)
+{
+	struct rtnexthop *nhp;
+	int nhlen;
+
+	if (attrs[RTA_PRIORITY] &&
+	    nla_get_u32(attrs[RTA_PRIORITY]) != fi->fib_priority)
+		return 1;
+
+	if (attrs[RTA_OIF] || attrs[RTA_GATEWAY]) {
+		if ((!attrs[RTA_OIF] || nla_get_u32(attrs[RTA_OIF]) == fi->fib_nh->nh_oif) &&
+		    (!attrs[RTA_GATEWAY]  || nla_get_le16(attrs[RTA_GATEWAY]) != fi->fib_nh->nh_gw))
+			return 0;
+		return 1;
+	}
+
+	if (!attrs[RTA_MULTIPATH])
+		return 0;
+
+	nhp = nla_data(attrs[RTA_MULTIPATH]);
+	nhlen = nla_len(attrs[RTA_MULTIPATH]);
+
+	for_nexthops(fi) {
+		int attrlen = nhlen - sizeof(struct rtnexthop);
+		__le16 gw;
+
+		if (attrlen < 0 || (nhlen -= nhp->rtnh_len) < 0)
+			return -EINVAL;
+		if (nhp->rtnh_ifindex && nhp->rtnh_ifindex != nh->nh_oif)
+			return 1;
+		if (attrlen) {
+			struct nlattr *gw_attr;
+
+			gw_attr = nla_find((struct nlattr *) (nhp + 1), attrlen, RTA_GATEWAY);
+			gw = gw_attr ? nla_get_le16(gw_attr) : 0;
+
+			if (gw && gw != nh->nh_gw)
+				return 1;
+		}
+		nhp = RTNH_NEXT(nhp);
+	} endfor_nexthops(fi);
+
+	return 0;
+}
+
+static inline size_t dn_fib_nlmsg_size(struct dn_fib_info *fi)
+{
+	size_t payload = NLMSG_ALIGN(sizeof(struct rtmsg))
+			 + nla_total_size(4) /* RTA_TABLE */
+			 + nla_total_size(2) /* RTA_DST */
+			 + nla_total_size(4); /* RTA_PRIORITY */
+
+	/* space for nested metrics */
+	payload += nla_total_size((RTAX_MAX * nla_total_size(4)));
+
+	if (fi->fib_nhs) {
+		/* Also handles the special case fib_nhs == 1 */
+
+		/* each nexthop is packed in an attribute */
+		size_t nhsize = nla_total_size(sizeof(struct rtnexthop));
+
+		/* may contain a gateway attribute */
+		nhsize += nla_total_size(4);
+
+		/* all nexthops are packed in a nested attribute */
+		payload += nla_total_size(fi->fib_nhs * nhsize);
+	}
+
+	return payload;
+}
+
+static int dn_fib_dump_info(struct sk_buff *skb, u32 portid, u32 seq, int event,
+			u32 tb_id, u8 type, u8 scope, void *dst, int dst_len,
+			struct dn_fib_info *fi, unsigned int flags)
+{
+	struct rtmsg *rtm;
+	struct nlmsghdr *nlh;
+
+	nlh = nlmsg_put(skb, portid, seq, event, sizeof(*rtm), flags);
+	if (!nlh)
+		return -EMSGSIZE;
+
+	rtm = nlmsg_data(nlh);
+	rtm->rtm_family = AF_DECnet;
+	rtm->rtm_dst_len = dst_len;
+	rtm->rtm_src_len = 0;
+	rtm->rtm_tos = 0;
+	rtm->rtm_table = tb_id;
+	rtm->rtm_flags = fi->fib_flags;
+	rtm->rtm_scope = scope;
+	rtm->rtm_type  = type;
+	rtm->rtm_protocol = fi->fib_protocol;
+
+	if (nla_put_u32(skb, RTA_TABLE, tb_id) < 0)
+		goto errout;
+
+	if (rtm->rtm_dst_len &&
+	    nla_put(skb, RTA_DST, 2, dst) < 0)
+		goto errout;
+
+	if (fi->fib_priority &&
+	    nla_put_u32(skb, RTA_PRIORITY, fi->fib_priority) < 0)
+		goto errout;
+
+	if (rtnetlink_put_metrics(skb, fi->fib_metrics) < 0)
+		goto errout;
+
+	if (fi->fib_nhs == 1) {
+		if (fi->fib_nh->nh_gw &&
+		    nla_put_le16(skb, RTA_GATEWAY, fi->fib_nh->nh_gw) < 0)
+			goto errout;
+
+		if (fi->fib_nh->nh_oif &&
+		    nla_put_u32(skb, RTA_OIF, fi->fib_nh->nh_oif) < 0)
+			goto errout;
+	}
+
+	if (fi->fib_nhs > 1) {
+		struct rtnexthop *nhp;
+		struct nlattr *mp_head;
+
+		if (!(mp_head = nla_nest_start(skb, RTA_MULTIPATH)))
+			goto errout;
+
+		for_nexthops(fi) {
+			if (!(nhp = nla_reserve_nohdr(skb, sizeof(*nhp))))
+				goto errout;
+
+			nhp->rtnh_flags = nh->nh_flags & 0xFF;
+			nhp->rtnh_hops = nh->nh_weight - 1;
+			nhp->rtnh_ifindex = nh->nh_oif;
+
+			if (nh->nh_gw &&
+			    nla_put_le16(skb, RTA_GATEWAY, nh->nh_gw) < 0)
+				goto errout;
+
+			nhp->rtnh_len = skb_tail_pointer(skb) - (unsigned char *)nhp;
+		} endfor_nexthops(fi);
+
+		nla_nest_end(skb, mp_head);
+	}
+
+	return nlmsg_end(skb, nlh);
+
+errout:
+	nlmsg_cancel(skb, nlh);
+	return -EMSGSIZE;
+}
+
+
+static void dn_rtmsg_fib(int event, struct dn_fib_node *f, int z, u32 tb_id,
+			struct nlmsghdr *nlh, struct netlink_skb_parms *req)
+{
+	struct sk_buff *skb;
+	u32 portid = req ? req->portid : 0;
+	int err = -ENOBUFS;
+
+	skb = nlmsg_new(dn_fib_nlmsg_size(DN_FIB_INFO(f)), GFP_KERNEL);
+	if (skb == NULL)
+		goto errout;
+
+	err = dn_fib_dump_info(skb, portid, nlh->nlmsg_seq, event, tb_id,
+			       f->fn_type, f->fn_scope, &f->fn_key, z,
+			       DN_FIB_INFO(f), 0);
+	if (err < 0) {
+		/* -EMSGSIZE implies BUG in dn_fib_nlmsg_size() */
+		WARN_ON(err == -EMSGSIZE);
+		kfree_skb(skb);
+		goto errout;
+	}
+	rtnl_notify(skb, &init_net, portid, RTNLGRP_DECnet_ROUTE, nlh, GFP_KERNEL);
+	return;
+errout:
+	if (err < 0)
+		rtnl_set_sk_err(&init_net, RTNLGRP_DECnet_ROUTE, err);
+}
+
+static __inline__ int dn_hash_dump_bucket(struct sk_buff *skb,
+				struct netlink_callback *cb,
+				struct dn_fib_table *tb,
+				struct dn_zone *dz,
+				struct dn_fib_node *f)
+{
+	int i, s_i;
+
+	s_i = cb->args[4];
+	for(i = 0; f; i++, f = f->fn_next) {
+		if (i < s_i)
+			continue;
+		if (f->fn_state & DN_S_ZOMBIE)
+			continue;
+		if (dn_fib_dump_info(skb, NETLINK_CB(cb->skb).portid,
+				cb->nlh->nlmsg_seq,
+				RTM_NEWROUTE,
+				tb->n,
+				(f->fn_state & DN_S_ZOMBIE) ? 0 : f->fn_type,
+				f->fn_scope, &f->fn_key, dz->dz_order,
+				f->fn_info, NLM_F_MULTI) < 0) {
+			cb->args[4] = i;
+			return -1;
+		}
+	}
+	cb->args[4] = i;
+	return skb->len;
+}
+
+static __inline__ int dn_hash_dump_zone(struct sk_buff *skb,
+				struct netlink_callback *cb,
+				struct dn_fib_table *tb,
+				struct dn_zone *dz)
+{
+	int h, s_h;
+
+	s_h = cb->args[3];
+	for(h = 0; h < dz->dz_divisor; h++) {
+		if (h < s_h)
+			continue;
+		if (h > s_h)
+			memset(&cb->args[4], 0, sizeof(cb->args) - 4*sizeof(cb->args[0]));
+		if (dz->dz_hash == NULL || dz->dz_hash[h] == NULL)
+			continue;
+		if (dn_hash_dump_bucket(skb, cb, tb, dz, dz->dz_hash[h]) < 0) {
+			cb->args[3] = h;
+			return -1;
+		}
+	}
+	cb->args[3] = h;
+	return skb->len;
+}
+
+static int dn_fib_table_dump(struct dn_fib_table *tb, struct sk_buff *skb,
+				struct netlink_callback *cb)
+{
+	int m, s_m;
+	struct dn_zone *dz;
+	struct dn_hash *table = (struct dn_hash *)tb->data;
+
+	s_m = cb->args[2];
+	read_lock(&dn_fib_tables_lock);
+	for(dz = table->dh_zone_list, m = 0; dz; dz = dz->dz_next, m++) {
+		if (m < s_m)
+			continue;
+		if (m > s_m)
+			memset(&cb->args[3], 0, sizeof(cb->args) - 3*sizeof(cb->args[0]));
+
+		if (dn_hash_dump_zone(skb, cb, tb, dz) < 0) {
+			cb->args[2] = m;
+			read_unlock(&dn_fib_tables_lock);
+			return -1;
+		}
+	}
+	read_unlock(&dn_fib_tables_lock);
+	cb->args[2] = m;
+
+	return skb->len;
+}
+
+int dn_fib_dump(struct sk_buff *skb, struct netlink_callback *cb)
+{
+	struct net *net = sock_net(skb->sk);
+	unsigned int h, s_h;
+	unsigned int e = 0, s_e;
+	struct dn_fib_table *tb;
+	int dumped = 0;
+
+	if (!net_eq(net, &init_net))
+		return 0;
+
+	if (nlmsg_len(cb->nlh) >= sizeof(struct rtmsg) &&
+		((struct rtmsg *)nlmsg_data(cb->nlh))->rtm_flags&RTM_F_CLONED)
+			return dn_cache_dump(skb, cb);
+
+	s_h = cb->args[0];
+	s_e = cb->args[1];
+
+	for (h = s_h; h < DN_FIB_TABLE_HASHSZ; h++, s_h = 0) {
+		e = 0;
+		hlist_for_each_entry(tb, &dn_fib_table_hash[h], hlist) {
+			if (e < s_e)
+				goto next;
+			if (dumped)
+				memset(&cb->args[2], 0, sizeof(cb->args) -
+						 2 * sizeof(cb->args[0]));
+			if (tb->dump(tb, skb, cb) < 0)
+				goto out;
+			dumped = 1;
+next:
+			e++;
+		}
+	}
+out:
+	cb->args[1] = e;
+	cb->args[0] = h;
+
+	return skb->len;
+}
+
+static int dn_fib_table_insert(struct dn_fib_table *tb, struct rtmsg *r, struct nlattr *attrs[],
+			       struct nlmsghdr *n, struct netlink_skb_parms *req)
+{
+	struct dn_hash *table = (struct dn_hash *)tb->data;
+	struct dn_fib_node *new_f, *f, **fp, **del_fp;
+	struct dn_zone *dz;
+	struct dn_fib_info *fi;
+	int z = r->rtm_dst_len;
+	int type = r->rtm_type;
+	dn_fib_key_t key;
+	int err;
+
+	if (z > 16)
+		return -EINVAL;
+
+	dz = table->dh_zones[z];
+	if (!dz && !(dz = dn_new_zone(table, z)))
+		return -ENOBUFS;
+
+	dz_key_0(key);
+	if (attrs[RTA_DST]) {
+		__le16 dst = nla_get_le16(attrs[RTA_DST]);
+		if (dst & ~DZ_MASK(dz))
+			return -EINVAL;
+		key = dz_key(dst, dz);
+	}
+
+	if ((fi = dn_fib_create_info(r, attrs, n, &err)) == NULL)
+		return err;
+
+	if (dz->dz_nent > (dz->dz_divisor << 2) &&
+			dz->dz_divisor > DN_MAX_DIVISOR &&
+			(z==16 || (1<<z) > dz->dz_divisor))
+		dn_rehash_zone(dz);
+
+	fp = dn_chain_p(key, dz);
+
+	DN_FIB_SCAN(f, fp) {
+		if (dn_key_leq(key, f->fn_key))
+			break;
+	}
+
+	del_fp = NULL;
+
+	if (f && (f->fn_state & DN_S_ZOMBIE) &&
+			dn_key_eq(f->fn_key, key)) {
+		del_fp = fp;
+		fp = &f->fn_next;
+		f = *fp;
+		goto create;
+	}
+
+	DN_FIB_SCAN_KEY(f, fp, key) {
+		if (fi->fib_priority <= DN_FIB_INFO(f)->fib_priority)
+			break;
+	}
+
+	if (f && dn_key_eq(f->fn_key, key) &&
+			fi->fib_priority == DN_FIB_INFO(f)->fib_priority) {
+		struct dn_fib_node **ins_fp;
+
+		err = -EEXIST;
+		if (n->nlmsg_flags & NLM_F_EXCL)
+			goto out;
+
+		if (n->nlmsg_flags & NLM_F_REPLACE) {
+			del_fp = fp;
+			fp = &f->fn_next;
+			f = *fp;
+			goto replace;
+		}
+
+		ins_fp = fp;
+		err = -EEXIST;
+
+		DN_FIB_SCAN_KEY(f, fp, key) {
+			if (fi->fib_priority != DN_FIB_INFO(f)->fib_priority)
+				break;
+			if (f->fn_type == type &&
+			    f->fn_scope == r->rtm_scope &&
+			    DN_FIB_INFO(f) == fi)
+				goto out;
+		}
+
+		if (!(n->nlmsg_flags & NLM_F_APPEND)) {
+			fp = ins_fp;
+			f = *fp;
+		}
+	}
+
+create:
+	err = -ENOENT;
+	if (!(n->nlmsg_flags & NLM_F_CREATE))
+		goto out;
+
+replace:
+	err = -ENOBUFS;
+	new_f = kmem_cache_zalloc(dn_hash_kmem, GFP_KERNEL);
+	if (new_f == NULL)
+		goto out;
+
+	new_f->fn_key = key;
+	new_f->fn_type = type;
+	new_f->fn_scope = r->rtm_scope;
+	DN_FIB_INFO(new_f) = fi;
+
+	new_f->fn_next = f;
+	write_lock_bh(&dn_fib_tables_lock);
+	*fp = new_f;
+	write_unlock_bh(&dn_fib_tables_lock);
+	dz->dz_nent++;
+
+	if (del_fp) {
+		f = *del_fp;
+		write_lock_bh(&dn_fib_tables_lock);
+		*del_fp = f->fn_next;
+		write_unlock_bh(&dn_fib_tables_lock);
+
+		if (!(f->fn_state & DN_S_ZOMBIE))
+			dn_rtmsg_fib(RTM_DELROUTE, f, z, tb->n, n, req);
+		if (f->fn_state & DN_S_ACCESSED)
+			dn_rt_cache_flush(-1);
+		dn_free_node(f);
+		dz->dz_nent--;
+	} else {
+		dn_rt_cache_flush(-1);
+	}
+
+	dn_rtmsg_fib(RTM_NEWROUTE, new_f, z, tb->n, n, req);
+
+	return 0;
+out:
+	dn_fib_release_info(fi);
+	return err;
+}
+
+
+static int dn_fib_table_delete(struct dn_fib_table *tb, struct rtmsg *r, struct nlattr *attrs[],
+			       struct nlmsghdr *n, struct netlink_skb_parms *req)
+{
+	struct dn_hash *table = (struct dn_hash*)tb->data;
+	struct dn_fib_node **fp, **del_fp, *f;
+	int z = r->rtm_dst_len;
+	struct dn_zone *dz;
+	dn_fib_key_t key;
+	int matched;
+
+
+	if (z > 16)
+		return -EINVAL;
+
+	if ((dz = table->dh_zones[z]) == NULL)
+		return -ESRCH;
+
+	dz_key_0(key);
+	if (attrs[RTA_DST]) {
+		__le16 dst = nla_get_le16(attrs[RTA_DST]);
+		if (dst & ~DZ_MASK(dz))
+			return -EINVAL;
+		key = dz_key(dst, dz);
+	}
+
+	fp = dn_chain_p(key, dz);
+
+	DN_FIB_SCAN(f, fp) {
+		if (dn_key_eq(f->fn_key, key))
+			break;
+		if (dn_key_leq(key, f->fn_key))
+			return -ESRCH;
+	}
+
+	matched = 0;
+	del_fp = NULL;
+	DN_FIB_SCAN_KEY(f, fp, key) {
+		struct dn_fib_info *fi = DN_FIB_INFO(f);
+
+		if (f->fn_state & DN_S_ZOMBIE)
+			return -ESRCH;
+
+		matched++;
+
+		if (del_fp == NULL &&
+				(!r->rtm_type || f->fn_type == r->rtm_type) &&
+				(r->rtm_scope == RT_SCOPE_NOWHERE || f->fn_scope == r->rtm_scope) &&
+				(!r->rtm_protocol ||
+					fi->fib_protocol == r->rtm_protocol) &&
+				dn_fib_nh_match(r, n, attrs, fi) == 0)
+			del_fp = fp;
+	}
+
+	if (del_fp) {
+		f = *del_fp;
+		dn_rtmsg_fib(RTM_DELROUTE, f, z, tb->n, n, req);
+
+		if (matched != 1) {
+			write_lock_bh(&dn_fib_tables_lock);
+			*del_fp = f->fn_next;
+			write_unlock_bh(&dn_fib_tables_lock);
+
+			if (f->fn_state & DN_S_ACCESSED)
+				dn_rt_cache_flush(-1);
+			dn_free_node(f);
+			dz->dz_nent--;
+		} else {
+			f->fn_state |= DN_S_ZOMBIE;
+			if (f->fn_state & DN_S_ACCESSED) {
+				f->fn_state &= ~DN_S_ACCESSED;
+				dn_rt_cache_flush(-1);
+			}
+			if (++dn_fib_hash_zombies > 128)
+				dn_fib_flush();
+		}
+
+		return 0;
+	}
+
+	return -ESRCH;
+}
+
+static inline int dn_flush_list(struct dn_fib_node **fp, int z, struct dn_hash *table)
+{
+	int found = 0;
+	struct dn_fib_node *f;
+
+	while((f = *fp) != NULL) {
+		struct dn_fib_info *fi = DN_FIB_INFO(f);
+
+		if (fi && ((f->fn_state & DN_S_ZOMBIE) || (fi->fib_flags & RTNH_F_DEAD))) {
+			write_lock_bh(&dn_fib_tables_lock);
+			*fp = f->fn_next;
+			write_unlock_bh(&dn_fib_tables_lock);
+
+			dn_free_node(f);
+			found++;
+			continue;
+		}
+		fp = &f->fn_next;
+	}
+
+	return found;
+}
+
+static int dn_fib_table_flush(struct dn_fib_table *tb)
+{
+	struct dn_hash *table = (struct dn_hash *)tb->data;
+	struct dn_zone *dz;
+	int found = 0;
+
+	dn_fib_hash_zombies = 0;
+	for(dz = table->dh_zone_list; dz; dz = dz->dz_next) {
+		int i;
+		int tmp = 0;
+		for(i = dz->dz_divisor-1; i >= 0; i--)
+			tmp += dn_flush_list(&dz->dz_hash[i], dz->dz_order, table);
+		dz->dz_nent -= tmp;
+		found += tmp;
+	}
+
+	return found;
+}
+
+static int dn_fib_table_lookup(struct dn_fib_table *tb, const struct flowidn *flp, struct dn_fib_res *res)
+{
+	int err;
+	struct dn_zone *dz;
+	struct dn_hash *t = (struct dn_hash *)tb->data;
+
+	read_lock(&dn_fib_tables_lock);
+	for(dz = t->dh_zone_list; dz; dz = dz->dz_next) {
+		struct dn_fib_node *f;
+		dn_fib_key_t k = dz_key(flp->daddr, dz);
+
+		for(f = dz_chain(k, dz); f; f = f->fn_next) {
+			if (!dn_key_eq(k, f->fn_key)) {
+				if (dn_key_leq(k, f->fn_key))
+					break;
+				else
+					continue;
+			}
+
+			f->fn_state |= DN_S_ACCESSED;
+
+			if (f->fn_state&DN_S_ZOMBIE)
+				continue;
+
+			if (f->fn_scope < flp->flowidn_scope)
+				continue;
+
+			err = dn_fib_semantic_match(f->fn_type, DN_FIB_INFO(f), flp, res);
+
+			if (err == 0) {
+				res->type = f->fn_type;
+				res->scope = f->fn_scope;
+				res->prefixlen = dz->dz_order;
+				goto out;
+			}
+			if (err < 0)
+				goto out;
+		}
+	}
+	err = 1;
+out:
+	read_unlock(&dn_fib_tables_lock);
+	return err;
+}
+
+
+struct dn_fib_table *dn_fib_get_table(u32 n, int create)
+{
+	struct dn_fib_table *t;
+	unsigned int h;
+
+	if (n < RT_TABLE_MIN)
+		return NULL;
+
+	if (n > RT_TABLE_MAX)
+		return NULL;
+
+	h = n & (DN_FIB_TABLE_HASHSZ - 1);
+	rcu_read_lock();
+	hlist_for_each_entry_rcu(t, &dn_fib_table_hash[h], hlist) {
+		if (t->n == n) {
+			rcu_read_unlock();
+			return t;
+		}
+	}
+	rcu_read_unlock();
+
+	if (!create)
+		return NULL;
+
+	if (in_interrupt()) {
+		net_dbg_ratelimited("DECnet: BUG! Attempt to create routing table from interrupt\n");
+		return NULL;
+	}
+
+	t = kzalloc(sizeof(struct dn_fib_table) + sizeof(struct dn_hash),
+		    GFP_KERNEL);
+	if (t == NULL)
+		return NULL;
+
+	t->n = n;
+	t->insert = dn_fib_table_insert;
+	t->delete = dn_fib_table_delete;
+	t->lookup = dn_fib_table_lookup;
+	t->flush  = dn_fib_table_flush;
+	t->dump = dn_fib_table_dump;
+	hlist_add_head_rcu(&t->hlist, &dn_fib_table_hash[h]);
+
+	return t;
+}
+
+struct dn_fib_table *dn_fib_empty_table(void)
+{
+	u32 id;
+
+	for(id = RT_TABLE_MIN; id <= RT_TABLE_MAX; id++)
+		if (dn_fib_get_table(id, 0) == NULL)
+			return dn_fib_get_table(id, 1);
+	return NULL;
+}
+
+void dn_fib_flush(void)
+{
+	int flushed = 0;
+	struct dn_fib_table *tb;
+	unsigned int h;
+
+	for (h = 0; h < DN_FIB_TABLE_HASHSZ; h++) {
+		hlist_for_each_entry(tb, &dn_fib_table_hash[h], hlist)
+			flushed += tb->flush(tb);
+	}
+
+	if (flushed)
+		dn_rt_cache_flush(-1);
+}
+
+void __init dn_fib_table_init(void)
+{
+	dn_hash_kmem = kmem_cache_create("dn_fib_info_cache",
+					sizeof(struct dn_fib_info),
+					0, SLAB_HWCACHE_ALIGN,
+					NULL);
+}
+
+void __exit dn_fib_table_cleanup(void)
+{
+	struct dn_fib_table *t;
+	struct hlist_node *next;
+	unsigned int h;
+
+	write_lock(&dn_fib_tables_lock);
+	for (h = 0; h < DN_FIB_TABLE_HASHSZ; h++) {
+		hlist_for_each_entry_safe(t, next, &dn_fib_table_hash[h],
+					  hlist) {
+			hlist_del(&t->hlist);
+			kfree(t);
+		}
+	}
+	write_unlock(&dn_fib_tables_lock);
+}

net/decnet/dn_timer.c

@@ -0,0 +1,103 @@
+/*
+ * DECnet       An implementation of the DECnet protocol suite for the LINUX
+ *              operating system.  DECnet is implemented using the  BSD Socket
+ *              interface as the means of communication with the user level.
+ *
+ *              DECnet Socket Timer Functions
+ *
+ * Author:      Steve Whitehouse <SteveW@ACM.org>
+ *
+ *
+ * Changes:
+ *       Steve Whitehouse      : Made keepalive timer part of the same
+ *                               timer idea.
+ *       Steve Whitehouse      : Added checks for sk->sock_readers
+ *       David S. Miller       : New socket locking
+ *       Steve Whitehouse      : Timer grabs socket ref.
+ */
+#include <linux/net.h>
+#include <linux/socket.h>
+#include <linux/skbuff.h>
+#include <linux/netdevice.h>
+#include <linux/timer.h>
+#include <linux/spinlock.h>
+#include <net/sock.h>
+#include <linux/atomic.h>
+#include <linux/jiffies.h>
+#include <net/flow.h>
+#include <net/dn.h>
+
+/*
+ * Slow timer is for everything else (n * 500mS)
+ */
+
+#define SLOW_INTERVAL (HZ/2)
+
+static void dn_slow_timer(unsigned long arg);
+
+void dn_start_slow_timer(struct sock *sk)
+{
+	setup_timer(&sk->sk_timer, dn_slow_timer, (unsigned long)sk);
+	sk_reset_timer(sk, &sk->sk_timer, jiffies + SLOW_INTERVAL);
+}
+
+void dn_stop_slow_timer(struct sock *sk)
+{
+	sk_stop_timer(sk, &sk->sk_timer);
+}
+
+static void dn_slow_timer(unsigned long arg)
+{
+	struct sock *sk = (struct sock *)arg;
+	struct dn_scp *scp = DN_SK(sk);
+
+	bh_lock_sock(sk);
+
+	if (sock_owned_by_user(sk)) {
+		sk_reset_timer(sk, &sk->sk_timer, jiffies + HZ / 10);
+		goto out;
+	}
+
+	/*
+	 * The persist timer is the standard slow timer used for retransmits
+	 * in both connection establishment and disconnection as well as
+	 * in the RUN state. The different states are catered for by changing
+	 * the function pointer in the socket. Setting the timer to a value
+	 * of zero turns it off. We allow the persist_fxn to turn the
+	 * timer off in a permant way by returning non-zero, so that
+	 * timer based routines may remove sockets. This is why we have a
+	 * sock_hold()/sock_put() around the timer to prevent the socket
+	 * going away in the middle.
+	 */
+	if (scp->persist && scp->persist_fxn) {
+		if (scp->persist <= SLOW_INTERVAL) {
+			scp->persist = 0;
+
+			if (scp->persist_fxn(sk))
+				goto out;
+		} else {
+			scp->persist -= SLOW_INTERVAL;
+		}
+	}
+
+	/*
+	 * Check for keepalive timeout. After the other timer 'cos if
+	 * the previous timer caused a retransmit, we don't need to
+	 * do this. scp->stamp is the last time that we sent a packet.
+	 * The keepalive function sends a link service packet to the
+	 * other end. If it remains unacknowledged, the standard
+	 * socket timers will eventually shut the socket down. Each
+	 * time we do this, scp->stamp will be updated, thus
+	 * we won't try and send another until scp->keepalive has passed
+	 * since the last successful transmission.
+	 */
+	if (scp->keepalive && scp->keepalive_fxn && (scp->state == DN_RUN)) {
+		if (time_after_eq(jiffies, scp->stamp + scp->keepalive))
+			scp->keepalive_fxn(sk);
+	}
+
+	sk_reset_timer(sk, &sk->sk_timer, jiffies + SLOW_INTERVAL);
+out:
+	bh_unlock_sock(sk);
+	sock_put(sk);
+}

net/decnet/netfilter/Kconfig

@@ -0,0 +1,16 @@
+#
+# DECnet netfilter configuration
+#
+
+menu "DECnet: Netfilter Configuration"
+	depends on DECNET && NETFILTER
+	depends on NETFILTER_ADVANCED
+
+config DECNET_NF_GRABULATOR
+	tristate "Routing message grabulator (for userland routing daemon)"
+	help
+	  Enable this module if you want to use the userland DECnet routing
+	  daemon. You will also need to enable routing support for DECnet
+	  unless you just want to monitor routing messages from other nodes.
+
+endmenu

net/decnet/netfilter/Makefile

@@ -0,0 +1,6 @@
+#
+# Makefile for DECnet netfilter modules
+#
+
+obj-$(CONFIG_DECNET_NF_GRABULATOR) += dn_rtmsg.o
+

net/decnet/netfilter/dn_rtmsg.c

@@ -0,0 +1,161 @@
+/*
+ * DECnet       An implementation of the DECnet protocol suite for the LINUX
+ *              operating system.  DECnet is implemented using the  BSD Socket
+ *              interface as the means of communication with the user level.
+ *
+ *              DECnet Routing Message Grabulator
+ *
+ *              (C) 2000 ChyGwyn Limited  -  http://www.chygwyn.com/
+ *              This code may be copied under the GPL v.2 or at your option
+ *              any later version.
+ *
+ * Author:      Steven Whitehouse <steve@chygwyn.com>
+ *
+ */
+#include <linux/module.h>
+#include <linux/skbuff.h>
+#include <linux/slab.h>
+#include <linux/init.h>
+#include <linux/netdevice.h>
+#include <linux/netfilter.h>
+#include <linux/spinlock.h>
+#include <net/netlink.h>
+#include <linux/netfilter_decnet.h>
+
+#include <net/sock.h>
+#include <net/flow.h>
+#include <net/dn.h>
+#include <net/dn_route.h>
+
+static struct sock *dnrmg = NULL;
+
+
+static struct sk_buff *dnrmg_build_message(struct sk_buff *rt_skb, int *errp)
+{
+	struct sk_buff *skb = NULL;
+	size_t size;
+	sk_buff_data_t old_tail;
+	struct nlmsghdr *nlh;
+	unsigned char *ptr;
+	struct nf_dn_rtmsg *rtm;
+
+	size = NLMSG_ALIGN(rt_skb->len) +
+	       NLMSG_ALIGN(sizeof(struct nf_dn_rtmsg));
+	skb = nlmsg_new(size, GFP_ATOMIC);
+	if (!skb) {
+		*errp = -ENOMEM;
+		return NULL;
+	}
+	old_tail = skb->tail;
+	nlh = nlmsg_put(skb, 0, 0, 0, size, 0);
+	if (!nlh) {
+		kfree_skb(skb);
+		*errp = -ENOMEM;
+		return NULL;
+	}
+	rtm = (struct nf_dn_rtmsg *)nlmsg_data(nlh);
+	rtm->nfdn_ifindex = rt_skb->dev->ifindex;
+	ptr = NFDN_RTMSG(rtm);
+	skb_copy_from_linear_data(rt_skb, ptr, rt_skb->len);
+	nlh->nlmsg_len = skb->tail - old_tail;
+	return skb;
+}
+
+static void dnrmg_send_peer(struct sk_buff *skb)
+{
+	struct sk_buff *skb2;
+	int status = 0;
+	int group = 0;
+	unsigned char flags = *skb->data;
+
+	switch (flags & DN_RT_CNTL_MSK) {
+	case DN_RT_PKT_L1RT:
+		group = DNRNG_NLGRP_L1;
+		break;
+	case DN_RT_PKT_L2RT:
+		group = DNRNG_NLGRP_L2;
+		break;
+	default:
+		return;
+	}
+
+	skb2 = dnrmg_build_message(skb, &status);
+	if (skb2 == NULL)
+		return;
+	NETLINK_CB(skb2).dst_group = group;
+	netlink_broadcast(dnrmg, skb2, 0, group, GFP_ATOMIC);
+}
+
+
+static unsigned int dnrmg_hook(const struct nf_hook_ops *ops,
+			struct sk_buff *skb,
+			const struct net_device *in,
+			const struct net_device *out,
+			int (*okfn)(struct sk_buff *))
+{
+	dnrmg_send_peer(skb);
+	return NF_ACCEPT;
+}
+
+
+#define RCV_SKB_FAIL(err) do { netlink_ack(skb, nlh, (err)); return; } while (0)
+
+static inline void dnrmg_receive_user_skb(struct sk_buff *skb)
+{
+	struct nlmsghdr *nlh = nlmsg_hdr(skb);
+
+	if (nlh->nlmsg_len < sizeof(*nlh) || skb->len < nlh->nlmsg_len)
+		return;
+
+	if (!netlink_capable(skb, CAP_NET_ADMIN))
+		RCV_SKB_FAIL(-EPERM);
+
+	/* Eventually we might send routing messages too */
+
+	RCV_SKB_FAIL(-EINVAL);
+}
+
+static struct nf_hook_ops dnrmg_ops __read_mostly = {
+	.hook		= dnrmg_hook,
+	.pf		= NFPROTO_DECNET,
+	.hooknum	= NF_DN_ROUTE,
+	.priority	= NF_DN_PRI_DNRTMSG,
+};
+
+static int __init dn_rtmsg_init(void)
+{
+	int rv = 0;
+	struct netlink_kernel_cfg cfg = {
+		.groups	= DNRNG_NLGRP_MAX,
+		.input	= dnrmg_receive_user_skb,
+	};
+
+	dnrmg = netlink_kernel_create(&init_net, NETLINK_DNRTMSG, &cfg);
+	if (dnrmg == NULL) {
+		printk(KERN_ERR "dn_rtmsg: Cannot create netlink socket");
+		return -ENOMEM;
+	}
+
+	rv = nf_register_hook(&dnrmg_ops);
+	if (rv) {
+		netlink_kernel_release(dnrmg);
+	}
+
+	return rv;
+}
+
+static void __exit dn_rtmsg_fini(void)
+{
+	nf_unregister_hook(&dnrmg_ops);
+	netlink_kernel_release(dnrmg);
+}
+
+
+MODULE_DESCRIPTION("DECnet Routing Message Grabulator");
+MODULE_AUTHOR("Steven Whitehouse <steve@chygwyn.com>");
+MODULE_LICENSE("GPL");
+MODULE_ALIAS_NET_PF_PROTO(PF_NETLINK, NETLINK_DNRTMSG);
+
+module_init(dn_rtmsg_init);
+module_exit(dn_rtmsg_fini);
+

net/decnet/sysctl_net_decnet.c

@@ -0,0 +1,372 @@
+/*
+ * DECnet       An implementation of the DECnet protocol suite for the LINUX
+ *              operating system.  DECnet is implemented using the  BSD Socket
+ *              interface as the means of communication with the user level.
+ *
+ *              DECnet sysctl support functions
+ *
+ * Author:      Steve Whitehouse <SteveW@ACM.org>
+ *
+ *
+ * Changes:
+ * Steve Whitehouse - C99 changes and default device handling
+ * Steve Whitehouse - Memory buffer settings, like the tcp ones
+ *
+ */
+#include <linux/mm.h>
+#include <linux/sysctl.h>
+#include <linux/fs.h>
+#include <linux/netdevice.h>
+#include <linux/string.h>
+#include <net/neighbour.h>
+#include <net/dst.h>
+#include <net/flow.h>
+
+#include <asm/uaccess.h>
+
+#include <net/dn.h>
+#include <net/dn_dev.h>
+#include <net/dn_route.h>
+
+
+int decnet_debug_level;
+int decnet_time_wait = 30;
+int decnet_dn_count = 1;
+int decnet_di_count = 3;
+int decnet_dr_count = 3;
+int decnet_log_martians = 1;
+int decnet_no_fc_max_cwnd = NSP_MIN_WINDOW;
+
+/* Reasonable defaults, I hope, based on tcp's defaults */
+long sysctl_decnet_mem[3] = { 768 << 3, 1024 << 3, 1536 << 3 };
+int sysctl_decnet_wmem[3] = { 4 * 1024, 16 * 1024, 128 * 1024 };
+int sysctl_decnet_rmem[3] = { 4 * 1024, 87380, 87380 * 2 };
+
+#ifdef CONFIG_SYSCTL
+extern int decnet_dst_gc_interval;
+static int min_decnet_time_wait[] = { 5 };
+static int max_decnet_time_wait[] = { 600 };
+static int min_state_count[] = { 1 };
+static int max_state_count[] = { NSP_MAXRXTSHIFT };
+static int min_decnet_dst_gc_interval[] = { 1 };
+static int max_decnet_dst_gc_interval[] = { 60 };
+static int min_decnet_no_fc_max_cwnd[] = { NSP_MIN_WINDOW };
+static int max_decnet_no_fc_max_cwnd[] = { NSP_MAX_WINDOW };
+static char node_name[7] = "???";
+
+static struct ctl_table_header *dn_table_header = NULL;
+
+/*
+ * ctype.h :-)
+ */
+#define ISNUM(x) (((x) >= '0') && ((x) <= '9'))
+#define ISLOWER(x) (((x) >= 'a') && ((x) <= 'z'))
+#define ISUPPER(x) (((x) >= 'A') && ((x) <= 'Z'))
+#define ISALPHA(x) (ISLOWER(x) || ISUPPER(x))
+#define INVALID_END_CHAR(x) (ISNUM(x) || ISALPHA(x))
+
+static void strip_it(char *str)
+{
+	for(;;) {
+		switch (*str) {
+		case ' ':
+		case '\n':
+		case '\r':
+		case ':':
+			*str = 0;
+			/* Fallthrough */
+		case 0:
+			return;
+		}
+		str++;
+	}
+}
+
+/*
+ * Simple routine to parse an ascii DECnet address
+ * into a network order address.
+ */
+static int parse_addr(__le16 *addr, char *str)
+{
+	__u16 area, node;
+
+	while(*str && !ISNUM(*str)) str++;
+
+	if (*str == 0)
+		return -1;
+
+	area = (*str++ - '0');
+	if (ISNUM(*str)) {
+		area *= 10;
+		area += (*str++ - '0');
+	}
+
+	if (*str++ != '.')
+		return -1;
+
+	if (!ISNUM(*str))
+		return -1;
+
+	node = *str++ - '0';
+	if (ISNUM(*str)) {
+		node *= 10;
+		node += (*str++ - '0');
+	}
+	if (ISNUM(*str)) {
+		node *= 10;
+		node += (*str++ - '0');
+	}
+	if (ISNUM(*str)) {
+		node *= 10;
+		node += (*str++ - '0');
+	}
+
+	if ((node > 1023) || (area > 63))
+		return -1;
+
+	if (INVALID_END_CHAR(*str))
+		return -1;
+
+	*addr = cpu_to_le16((area << 10) | node);
+
+	return 0;
+}
+
+static int dn_node_address_handler(struct ctl_table *table, int write,
+				void __user *buffer,
+				size_t *lenp, loff_t *ppos)
+{
+	char addr[DN_ASCBUF_LEN];
+	size_t len;
+	__le16 dnaddr;
+
+	if (!*lenp || (*ppos && !write)) {
+		*lenp = 0;
+		return 0;
+	}
+
+	if (write) {
+		len = (*lenp < DN_ASCBUF_LEN) ? *lenp : (DN_ASCBUF_LEN-1);
+
+		if (copy_from_user(addr, buffer, len))
+			return -EFAULT;
+
+		addr[len] = 0;
+		strip_it(addr);
+
+		if (parse_addr(&dnaddr, addr))
+			return -EINVAL;
+
+		dn_dev_devices_off();
+
+		decnet_address = dnaddr;
+
+		dn_dev_devices_on();
+
+		*ppos += len;
+
+		return 0;
+	}
+
+	dn_addr2asc(le16_to_cpu(decnet_address), addr);
+	len = strlen(addr);
+	addr[len++] = '\n';
+
+	if (len > *lenp) len = *lenp;
+
+	if (copy_to_user(buffer, addr, len))
+		return -EFAULT;
+
+	*lenp = len;
+	*ppos += len;
+
+	return 0;
+}
+
+static int dn_def_dev_handler(struct ctl_table *table, int write,
+				void __user *buffer,
+				size_t *lenp, loff_t *ppos)
+{
+	size_t len;
+	struct net_device *dev;
+	char devname[17];
+
+	if (!*lenp || (*ppos && !write)) {
+		*lenp = 0;
+		return 0;
+	}
+
+	if (write) {
+		if (*lenp > 16)
+			return -E2BIG;
+
+		if (copy_from_user(devname, buffer, *lenp))
+			return -EFAULT;
+
+		devname[*lenp] = 0;
+		strip_it(devname);
+
+		dev = dev_get_by_name(&init_net, devname);
+		if (dev == NULL)
+			return -ENODEV;
+
+		if (dev->dn_ptr == NULL) {
+			dev_put(dev);
+			return -ENODEV;
+		}
+
+		if (dn_dev_set_default(dev, 1)) {
+			dev_put(dev);
+			return -ENODEV;
+		}
+		*ppos += *lenp;
+
+		return 0;
+	}
+
+	dev = dn_dev_get_default();
+	if (dev == NULL) {
+		*lenp = 0;
+		return 0;
+	}
+
+	strcpy(devname, dev->name);
+	dev_put(dev);
+	len = strlen(devname);
+	devname[len++] = '\n';
+
+	if (len > *lenp) len = *lenp;
+
+	if (copy_to_user(buffer, devname, len))
+		return -EFAULT;
+
+	*lenp = len;
+	*ppos += len;
+
+	return 0;
+}
+
+static struct ctl_table dn_table[] = {
+	{
+		.procname = "node_address",
+		.maxlen = 7,
+		.mode = 0644,
+		.proc_handler = dn_node_address_handler,
+	},
+	{
+		.procname = "node_name",
+		.data = node_name,
+		.maxlen = 7,
+		.mode = 0644,
+		.proc_handler = proc_dostring,
+	},
+	{
+		.procname = "default_device",
+		.maxlen = 16,
+		.mode = 0644,
+		.proc_handler = dn_def_dev_handler,
+	},
+	{
+		.procname = "time_wait",
+		.data = &decnet_time_wait,
+		.maxlen = sizeof(int),
+		.mode = 0644,
+		.proc_handler = proc_dointvec_minmax,
+		.extra1 = &min_decnet_time_wait,
+		.extra2 = &max_decnet_time_wait
+	},
+	{
+		.procname = "dn_count",
+		.data = &decnet_dn_count,
+		.maxlen = sizeof(int),
+		.mode = 0644,
+		.proc_handler = proc_dointvec_minmax,
+		.extra1 = &min_state_count,
+		.extra2 = &max_state_count
+	},
+	{
+		.procname = "di_count",
+		.data = &decnet_di_count,
+		.maxlen = sizeof(int),
+		.mode = 0644,
+		.proc_handler = proc_dointvec_minmax,
+		.extra1 = &min_state_count,
+		.extra2 = &max_state_count
+	},
+	{
+		.procname = "dr_count",
+		.data = &decnet_dr_count,
+		.maxlen = sizeof(int),
+		.mode = 0644,
+		.proc_handler = proc_dointvec_minmax,
+		.extra1 = &min_state_count,
+		.extra2 = &max_state_count
+	},
+	{
+		.procname = "dst_gc_interval",
+		.data = &decnet_dst_gc_interval,
+		.maxlen = sizeof(int),
+		.mode = 0644,
+		.proc_handler = proc_dointvec_minmax,
+		.extra1 = &min_decnet_dst_gc_interval,
+		.extra2 = &max_decnet_dst_gc_interval
+	},
+	{
+		.procname = "no_fc_max_cwnd",
+		.data = &decnet_no_fc_max_cwnd,
+		.maxlen = sizeof(int),
+		.mode = 0644,
+		.proc_handler = proc_dointvec_minmax,
+		.extra1 = &min_decnet_no_fc_max_cwnd,
+		.extra2 = &max_decnet_no_fc_max_cwnd
+	},
+       {
+		.procname = "decnet_mem",
+		.data = &sysctl_decnet_mem,
+		.maxlen = sizeof(sysctl_decnet_mem),
+		.mode = 0644,
+		.proc_handler = proc_doulongvec_minmax
+	},
+	{
+		.procname = "decnet_rmem",
+		.data = &sysctl_decnet_rmem,
+		.maxlen = sizeof(sysctl_decnet_rmem),
+		.mode = 0644,
+		.proc_handler = proc_dointvec,
+	},
+	{
+		.procname = "decnet_wmem",
+		.data = &sysctl_decnet_wmem,
+		.maxlen = sizeof(sysctl_decnet_wmem),
+		.mode = 0644,
+		.proc_handler = proc_dointvec,
+	},
+	{
+		.procname = "debug",
+		.data = &decnet_debug_level,
+		.maxlen = sizeof(int),
+		.mode = 0644,
+		.proc_handler = proc_dointvec,
+	},
+	{ }
+};
+
+void dn_register_sysctl(void)
+{
+	dn_table_header = register_net_sysctl(&init_net, "net/decnet", dn_table);
+}
+
+void dn_unregister_sysctl(void)
+{
+	unregister_net_sysctl_table(dn_table_header);
+}
+
+#else  /* CONFIG_SYSCTL */
+void dn_unregister_sysctl(void)
+{
+}
+void dn_register_sysctl(void)
+{
+}
+
+#endif