MFP r287070,r287073: split radix implementation and route table structure.

There are number of radix consumers in kernel land (pf,ipfw,nfs,route)
  with different requirements. In fact, first 3 don't have _any_ requirements
  and first 2 does not use radix locking. On the other hand, routing
  structure do have these requirements (rnh_gen, multipath, custom
  to-be-added control plane functions, different locking).
Additionally, radix should not known anything about its consumers internals.

So, radix code now uses tiny 'struct radix_head' structure along with
  internal 'struct radix_mask_head' instead of 'struct radix_node_head'.
  Existing consumers still uses the same 'struct radix_node_head' with
  slight modifications: they need to pass pointer to (embedded)
  'struct radix_head' to all radix callbacks.

Routing code now uses new 'struct rib_head' with different locking macro:
  RADIX_NODE_HEAD prefix was renamed to RIB_ (which stands for routing
  information base).

New net/route_var.h header was added to hold routing subsystem internal
  data. 'struct rib_head' was placed there. 'struct rtentry' will also
  be moved there soon.

2 files changed, 55 insertions, 54 deletions

diff --git a/sys/netpfil/ipfw/ip_fw_table_algo.c b/sys/netpfil/ipfw/ip_fw_table_algo.c
index 2ce550e..c04ee86 100644
--- a/sys/netpfil/ipfw/ip_fw_table_algo.c
+++ b/sys/netpfil/ipfw/ip_fw_table_algo.c
@@ -51,6 +51,7 @@ __FBSDID("$FreeBSD$");
 #include <net/if.h>	/* ip_fw.h requires IFNAMSIZ */
 #include <net/radix.h>
 #include <net/route.h>
+#include <net/route_var.h>
 
 #include <netinet/in.h>
 #include <netinet/in_fib.h>
@@ -409,7 +410,7 @@ ta_lookup_radix(struct table_info *ti, void *key, uint32_t keylen,
 		KEY_LEN(sa) = KEY_LEN_INET;
 		sa.sin_addr.s_addr = *((in_addr_t *)key);
 		rnh = (struct radix_node_head *)ti->state;
-		ent = (struct radix_addr_entry *)(rnh->rnh_matchaddr(&sa, rnh));
+		ent = (struct radix_addr_entry *)(rnh->rnh_matchaddr(&sa, &rnh->rh));
 		if (ent != NULL) {
 			*val = ent->value;
 			return (1);
@@ -420,7 +421,7 @@ ta_lookup_radix(struct table_info *ti, void *key, uint32_t keylen,
 		KEY_LEN(sa6) = KEY_LEN_INET6;
 		memcpy(&sa6.sin6_addr, key, sizeof(struct in6_addr));
 		rnh = (struct radix_node_head *)ti->xstate;
-		xent = (struct radix_addr_xentry *)(rnh->rnh_matchaddr(&sa6, rnh));
+		xent = (struct radix_addr_xentry *)(rnh->rnh_matchaddr(&sa6, &rnh->rh));
 		if (xent != NULL) {
 			*val = xent->value;
 			return (1);
@@ -461,7 +462,7 @@ flush_radix_entry(struct radix_node *rn, void *arg)
 	struct radix_addr_entry *ent;
 
 	ent = (struct radix_addr_entry *)
-	    rnh->rnh_deladdr(rn->rn_key, rn->rn_mask, rnh);
+	    rnh->rnh_deladdr(rn->rn_key, rn->rn_mask, &rnh->rh);
 	if (ent != NULL)
 		free(ent, M_IPFW_TBL);
 	return (0);
@@ -476,11 +477,11 @@ ta_destroy_radix(void *ta_state, struct table_info *ti)
 	cfg = (struct radix_cfg *)ta_state;
 
 	rnh = (struct radix_node_head *)(ti->state);
-	rnh->rnh_walktree(rnh, flush_radix_entry, rnh);
+	rnh->rnh_walktree(&rnh->rh, flush_radix_entry, rnh);
 	rn_detachhead(&ti->state);
 
 	rnh = (struct radix_node_head *)(ti->xstate);
-	rnh->rnh_walktree(rnh, flush_radix_entry, rnh);
+	rnh->rnh_walktree(&rnh->rh, flush_radix_entry, rnh);
 	rn_detachhead(&ti->xstate);
 
 	free(cfg, M_IPFW);
@@ -548,13 +549,13 @@ ta_find_radix_tentry(void *ta_state, struct table_info *ti,
 		KEY_LEN(sa) = KEY_LEN_INET;
 		sa.sin_addr.s_addr = tent->k.addr.s_addr;
 		rnh = (struct radix_node_head *)ti->state;
-		e = rnh->rnh_matchaddr(&sa, rnh);
+		e = rnh->rnh_matchaddr(&sa, &rnh->rh);
 	} else {
 		struct sa_in6 sa6;
 		KEY_LEN(sa6) = KEY_LEN_INET6;
 		memcpy(&sa6.sin6_addr, &tent->k.addr6, sizeof(struct in6_addr));
 		rnh = (struct radix_node_head *)ti->xstate;
-		e = rnh->rnh_matchaddr(&sa6, rnh);
+		e = rnh->rnh_matchaddr(&sa6, &rnh->rh);
 	}
 
 	if (e != NULL) {
@@ -572,10 +573,10 @@ ta_foreach_radix(void *ta_state, struct table_info *ti, ta_foreach_f *f,
 	struct radix_node_head *rnh;
 
 	rnh = (struct radix_node_head *)(ti->state);
-	rnh->rnh_walktree(rnh, (walktree_f_t *)f, arg);
+	rnh->rnh_walktree(&rnh->rh, (walktree_f_t *)f, arg);
 
 	rnh = (struct radix_node_head *)(ti->xstate);
-	rnh->rnh_walktree(rnh, (walktree_f_t *)f, arg);
+	rnh->rnh_walktree(&rnh->rh, (walktree_f_t *)f, arg);
 }
 
 
@@ -722,7 +723,7 @@ ta_add_radix(void *ta_state, struct table_info *ti, struct tentry_info *tei,
 	}
 
 	/* Search for an entry first */
-	rn = rnh->rnh_lookup(tb->addr_ptr, tb->mask_ptr, rnh);
+	rn = rnh->rnh_lookup(tb->addr_ptr, tb->mask_ptr, &rnh->rh);
 	if (rn != NULL) {
 		if ((tei->flags & TEI_FLAGS_UPDATE) == 0)
 			return (EEXIST);
@@ -746,7 +747,7 @@ ta_add_radix(void *ta_state, struct table_info *ti, struct tentry_info *tei,
 	if ((tei->flags & TEI_FLAGS_DONTADD) != 0)
 		return (EFBIG);
 
-	rn = rnh->rnh_addaddr(tb->addr_ptr, tb->mask_ptr, rnh, tb->ent_ptr);
+	rn = rnh->rnh_addaddr(tb->addr_ptr, tb->mask_ptr, &rnh->rh,tb->ent_ptr);
 	if (rn == NULL) {
 		/* Unknown error */
 		return (EINVAL);
@@ -817,7 +818,7 @@ ta_del_radix(void *ta_state, struct table_info *ti, struct tentry_info *tei,
 	else
 		rnh = ti->xstate;
 
-	rn = rnh->rnh_deladdr(tb->addr_ptr, tb->mask_ptr, rnh);
+	rn = rnh->rnh_deladdr(tb->addr_ptr, tb->mask_ptr, &rnh->rh);
 
 	if (rn == NULL)
 		return (ENOENT);
@@ -4042,21 +4043,21 @@ static void
 ta_foreach_kfib(void *ta_state, struct table_info *ti, ta_foreach_f *f,
     void *arg)
 {
-	struct radix_node_head *rnh;
+	struct rib_head *rh;
 	int error;
 
-	rnh = rt_tables_get_rnh(ti->data, AF_INET);
-	if (rnh != NULL) {
-		RADIX_NODE_HEAD_RLOCK(rnh); 
-		error = rnh->rnh_walktree(rnh, (walktree_f_t *)f, arg);
-		RADIX_NODE_HEAD_RUNLOCK(rnh);
+	rh = rt_tables_get_rnh(ti->data, AF_INET);
+	if (rh != NULL) {
+		RIB_RLOCK(rh); 
+		error = rh->rnh_walktree(&rh->head, (walktree_f_t *)f, arg);
+		RIB_RUNLOCK(rh);
 	}
 
-	rnh = rt_tables_get_rnh(ti->data, AF_INET6);
-	if (rnh != NULL) {
-		RADIX_NODE_HEAD_RLOCK(rnh); 
-		error = rnh->rnh_walktree(rnh, (walktree_f_t *)f, arg);
-		RADIX_NODE_HEAD_RUNLOCK(rnh);
+	rh = rt_tables_get_rnh(ti->data, AF_INET6);
+	if (rh != NULL) {
+		RIB_RLOCK(rh); 
+		error = rh->rnh_walktree(&rh->head, (walktree_f_t *)f, arg);
+		RIB_RUNLOCK(rh); 
 	}
 }
 
diff --git a/sys/netpfil/pf/pf_table.c b/sys/netpfil/pf/pf_table.c
index b9f13b9..b9889b9 100644
--- a/sys/netpfil/pf/pf_table.c
+++ b/sys/netpfil/pf/pf_table.c
@@ -559,10 +559,10 @@ pfr_get_addrs(struct pfr_table *tbl, struct pfr_addr *addr, int *size,
 	w.pfrw_op = PFRW_GET_ADDRS;
 	w.pfrw_addr = addr;
 	w.pfrw_free = kt->pfrkt_cnt;
-	rv = kt->pfrkt_ip4->rnh_walktree(kt->pfrkt_ip4, pfr_walktree, &w);
+	rv = kt->pfrkt_ip4->rnh_walktree(&kt->pfrkt_ip4->rh, pfr_walktree, &w);
 	if (!rv)
-		rv = kt->pfrkt_ip6->rnh_walktree(kt->pfrkt_ip6, pfr_walktree,
-		    &w);
+		rv = kt->pfrkt_ip6->rnh_walktree(&kt->pfrkt_ip6->rh,
+		    pfr_walktree, &w);
 	if (rv)
 		return (rv);
 
@@ -601,10 +601,10 @@ pfr_get_astats(struct pfr_table *tbl, struct pfr_astats *addr, int *size,
 	w.pfrw_op = PFRW_GET_ASTATS;
 	w.pfrw_astats = addr;
 	w.pfrw_free = kt->pfrkt_cnt;
-	rv = kt->pfrkt_ip4->rnh_walktree(kt->pfrkt_ip4, pfr_walktree, &w);
+	rv = kt->pfrkt_ip4->rnh_walktree(&kt->pfrkt_ip4->rh, pfr_walktree, &w);
 	if (!rv)
-		rv = kt->pfrkt_ip6->rnh_walktree(kt->pfrkt_ip6, pfr_walktree,
-		    &w);
+		rv = kt->pfrkt_ip6->rnh_walktree(&kt->pfrkt_ip6->rh,
+		    pfr_walktree, &w);
 	if (!rv && (flags & PFR_FLAG_CLSTATS)) {
 		pfr_enqueue_addrs(kt, &workq, NULL, 0);
 		pfr_clstats_kentries(&workq, tzero, 0);
@@ -710,12 +710,12 @@ pfr_enqueue_addrs(struct pfr_ktable *kt, struct pfr_kentryworkq *workq,
 	w.pfrw_op = sweep ? PFRW_SWEEP : PFRW_ENQUEUE;
 	w.pfrw_workq = workq;
 	if (kt->pfrkt_ip4 != NULL)
-		if (kt->pfrkt_ip4->rnh_walktree(kt->pfrkt_ip4, pfr_walktree,
-		    &w))
+		if (kt->pfrkt_ip4->rnh_walktree(&kt->pfrkt_ip4->rh,
+		    pfr_walktree, &w))
 			printf("pfr_enqueue_addrs: IPv4 walktree failed.\n");
 	if (kt->pfrkt_ip6 != NULL)
-		if (kt->pfrkt_ip6->rnh_walktree(kt->pfrkt_ip6, pfr_walktree,
-		    &w))
+		if (kt->pfrkt_ip6->rnh_walktree(&kt->pfrkt_ip6->rh,
+		    pfr_walktree, &w))
 			printf("pfr_enqueue_addrs: IPv6 walktree failed.\n");
 	if (naddr != NULL)
 		*naddr = w.pfrw_cnt;
@@ -728,9 +728,9 @@ pfr_mark_addrs(struct pfr_ktable *kt)
 
 	bzero(&w, sizeof(w));
 	w.pfrw_op = PFRW_MARK;
-	if (kt->pfrkt_ip4->rnh_walktree(kt->pfrkt_ip4, pfr_walktree, &w))
+	if (kt->pfrkt_ip4->rnh_walktree(&kt->pfrkt_ip4->rh, pfr_walktree, &w))
 		printf("pfr_mark_addrs: IPv4 walktree failed.\n");
-	if (kt->pfrkt_ip6->rnh_walktree(kt->pfrkt_ip6, pfr_walktree, &w))
+	if (kt->pfrkt_ip6->rnh_walktree(&kt->pfrkt_ip6->rh, pfr_walktree, &w))
 		printf("pfr_mark_addrs: IPv6 walktree failed.\n");
 }
 
@@ -739,7 +739,7 @@ static struct pfr_kentry *
 pfr_lookup_addr(struct pfr_ktable *kt, struct pfr_addr *ad, int exact)
 {
 	union sockaddr_union	 sa, mask;
-	struct radix_node_head	*head = NULL;
+	struct radix_head	*head = NULL;
 	struct pfr_kentry	*ke;
 
 	PF_RULES_ASSERT();
@@ -747,10 +747,10 @@ pfr_lookup_addr(struct pfr_ktable *kt, struct pfr_addr *ad, int exact)
 	bzero(&sa, sizeof(sa));
 	if (ad->pfra_af == AF_INET) {
 		FILLIN_SIN(sa.sin, ad->pfra_ip4addr);
-		head = kt->pfrkt_ip4;
+		head = &kt->pfrkt_ip4->rh;
 	} else if ( ad->pfra_af == AF_INET6 ) {
 		FILLIN_SIN6(sa.sin6, ad->pfra_ip6addr);
-		head = kt->pfrkt_ip6;
+		head = &kt->pfrkt_ip6->rh;
 	}
 	if (ADDR_NETWORK(ad)) {
 		pfr_prepare_network(&mask, ad->pfra_af, ad->pfra_net);
@@ -929,15 +929,15 @@ pfr_route_kentry(struct pfr_ktable *kt, struct pfr_kentry *ke)
 {
 	union sockaddr_union	 mask;
 	struct radix_node	*rn;
-	struct radix_node_head	*head = NULL;
+	struct radix_head	*head = NULL;
 
 	PF_RULES_WASSERT();
 
 	bzero(ke->pfrke_node, sizeof(ke->pfrke_node));
 	if (ke->pfrke_af == AF_INET)
-		head = kt->pfrkt_ip4;
+		head = &kt->pfrkt_ip4->rh;
 	else if (ke->pfrke_af == AF_INET6)
-		head = kt->pfrkt_ip6;
+		head = &kt->pfrkt_ip6->rh;
 
 	if (KENTRY_NETWORK(ke)) {
 		pfr_prepare_network(&mask, ke->pfrke_af, ke->pfrke_net);
@@ -953,12 +953,12 @@ pfr_unroute_kentry(struct pfr_ktable *kt, struct pfr_kentry *ke)
 {
 	union sockaddr_union	 mask;
 	struct radix_node	*rn;
-	struct radix_node_head	*head = NULL;
+	struct radix_head	*head = NULL;
 
 	if (ke->pfrke_af == AF_INET)
-		head = kt->pfrkt_ip4;
+		head = &kt->pfrkt_ip4->rh;
 	else if (ke->pfrke_af == AF_INET6)
-		head = kt->pfrkt_ip6;
+		head = &kt->pfrkt_ip6->rh;
 
 	if (KENTRY_NETWORK(ke)) {
 		pfr_prepare_network(&mask, ke->pfrke_af, ke->pfrke_net);
@@ -1907,7 +1907,7 @@ pfr_match_addr(struct pfr_ktable *kt, struct pf_addr *a, sa_family_t af)
 		sin.sin_len = sizeof(sin);
 		sin.sin_family = AF_INET;
 		sin.sin_addr.s_addr = a->addr32[0];
-		ke = (struct pfr_kentry *)rn_match(&sin, kt->pfrkt_ip4);
+		ke = (struct pfr_kentry *)rn_match(&sin, &kt->pfrkt_ip4->rh);
 		if (ke && KENTRY_RNF_ROOT(ke))
 			ke = NULL;
 		break;
@@ -1922,7 +1922,7 @@ pfr_match_addr(struct pfr_ktable *kt, struct pf_addr *a, sa_family_t af)
 		sin6.sin6_len = sizeof(sin6);
 		sin6.sin6_family = AF_INET6;
 		bcopy(a, &sin6.sin6_addr, sizeof(sin6.sin6_addr));
-		ke = (struct pfr_kentry *)rn_match(&sin6, kt->pfrkt_ip6);
+		ke = (struct pfr_kentry *)rn_match(&sin6, &kt->pfrkt_ip6->rh);
 		if (ke && KENTRY_RNF_ROOT(ke))
 			ke = NULL;
 		break;
@@ -1958,7 +1958,7 @@ pfr_update_stats(struct pfr_ktable *kt, struct pf_addr *a, sa_family_t af,
 		sin.sin_len = sizeof(sin);
 		sin.sin_family = AF_INET;
 		sin.sin_addr.s_addr = a->addr32[0];
-		ke = (struct pfr_kentry *)rn_match(&sin, kt->pfrkt_ip4);
+		ke = (struct pfr_kentry *)rn_match(&sin, &kt->pfrkt_ip4->rh);
 		if (ke && KENTRY_RNF_ROOT(ke))
 			ke = NULL;
 		break;
@@ -1973,7 +1973,7 @@ pfr_update_stats(struct pfr_ktable *kt, struct pf_addr *a, sa_family_t af,
 		sin6.sin6_len = sizeof(sin6);
 		sin6.sin6_family = AF_INET6;
 		bcopy(a, &sin6.sin6_addr, sizeof(sin6.sin6_addr));
-		ke = (struct pfr_kentry *)rn_match(&sin6, kt->pfrkt_ip6);
+		ke = (struct pfr_kentry *)rn_match(&sin6, &kt->pfrkt_ip6->rh);
 		if (ke && KENTRY_RNF_ROOT(ke))
 			ke = NULL;
 		break;
@@ -2120,11 +2120,11 @@ _next_block:
 		switch (af) {
 		case AF_INET:
 			ke2 = (struct pfr_kentry *)rn_match(&uaddr,
-			    kt->pfrkt_ip4);
+			    &kt->pfrkt_ip4->rh);
 			break;
 		case AF_INET6:
 			ke2 = (struct pfr_kentry *)rn_match(&uaddr,
-			    kt->pfrkt_ip6);
+			    &kt->pfrkt_ip6->rh);
 			break;
 		}
 		/* no need to check KENTRY_RNF_ROOT() here */
@@ -2162,12 +2162,12 @@ pfr_kentry_byidx(struct pfr_ktable *kt, int idx, int af)
 	switch (af) {
 #ifdef INET
 	case AF_INET:
-		kt->pfrkt_ip4->rnh_walktree(kt->pfrkt_ip4, pfr_walktree, &w);
+		kt->pfrkt_ip4->rnh_walktree(&kt->pfrkt_ip4->rh, pfr_walktree, &w);
 		return (w.pfrw_kentry);
 #endif /* INET */
 #ifdef INET6
 	case AF_INET6:
-		kt->pfrkt_ip6->rnh_walktree(kt->pfrkt_ip6, pfr_walktree, &w);
+		kt->pfrkt_ip6->rnh_walktree(&kt->pfrkt_ip6->rh, pfr_walktree, &w);
 		return (w.pfrw_kentry);
 #endif /* INET6 */
 	default:
@@ -2187,7 +2187,7 @@ pfr_dynaddr_update(struct pfr_ktable *kt, struct pfi_dynaddr *dyn)
 	dyn->pfid_acnt4 = 0;
 	dyn->pfid_acnt6 = 0;
 	if (!dyn->pfid_af || dyn->pfid_af == AF_INET)
-		kt->pfrkt_ip4->rnh_walktree(kt->pfrkt_ip4, pfr_walktree, &w);
+		kt->pfrkt_ip4->rnh_walktree(&kt->pfrkt_ip4->rh, pfr_walktree, &w);
 	if (!dyn->pfid_af || dyn->pfid_af == AF_INET6)
-		kt->pfrkt_ip6->rnh_walktree(kt->pfrkt_ip6, pfr_walktree, &w);
+		kt->pfrkt_ip6->rnh_walktree(&kt->pfrkt_ip6->rh, pfr_walktree, &w);
 }