D36074.diff
No OneTemporary
Actions

Size

28 KB

Referenced Files

None

Subscribers

None

D36074.diff
View Options

	diff --git a/sys/conf/files b/sys/conf/files
	--- a/sys/conf/files
	+++ b/sys/conf/files
	@@ -4176,6 +4176,8 @@
	net/route/route_ddb.c optional ddb
	net/route/route_helpers.c standard
	net/route/route_ifaddrs.c standard
	+net/route/route_rtentry.c standard
	+net/route/route_subscription.c standard
	net/route/route_tables.c standard
	net/route/route_temporal.c standard
	net/rss_config.c optional inet rss \| inet6 rss
	diff --git a/sys/net/route/route_ctl.h b/sys/net/route/route_ctl.h
	--- a/sys/net/route/route_ctl.h
	+++ b/sys/net/route/route_ctl.h
	@@ -178,5 +178,7 @@
	rib_subscription_cb_t f, void arg, enum rib_subscription_type type);
	void rib_unsubscribe(struct rib_subscription *rs);
	void rib_unsubscribe_locked(struct rib_subscription *rs);
	+void rib_notify(struct rib_head *rnh, enum rib_subscription_type type,
	+ struct rib_cmd_info *rc);

	#endif
	diff --git a/sys/net/route/route_ctl.c b/sys/net/route/route_ctl.c
	--- a/sys/net/route/route_ctl.c
	+++ b/sys/net/route/route_ctl.c
	@@ -56,8 +56,6 @@
	#include <netinet6/scope6_var.h>
	#include <netinet6/in6_var.h>

	-#include <vm/uma.h>
	-
	#define DEBUG_MOD_NAME route_ctl
	#define DEBUG_MAX_LEVEL LOG_DEBUG
	#include <net/route/route_debug.h>
	@@ -69,15 +67,6 @@
	* All functions assumes they are called in net epoch.
	*/

	-struct rib_subscription {
	- CK_STAILQ_ENTRY(rib_subscription) next;
	- rib_subscription_cb_t *func;
	- void *arg;
	- struct rib_head *rnh;
	- enum rib_subscription_type type;
	- struct epoch_context epoch_ctx;
	-};
	-
	union sockaddr_union {
	struct sockaddr sa;
	struct sockaddr_in sin;
	@@ -104,16 +93,9 @@
	static int rt_delete_conditional(struct rib_head rnh, struct rtentry rt,
	int prio, rib_filter_f_t cb, void cbdata, struct rib_cmd_info *rc);

	-static struct rtentry create_rtentry(struct rib_head rnh,
	- const struct sockaddr dst, struct sockaddr netmask);
	-
	-static void rib_notify(struct rib_head *rnh, enum rib_subscription_type type,
	- struct rib_cmd_info *rc);
	-
	static int get_prio_from_info(const struct rt_addrinfo *info);
	static int nhop_get_prio(const struct nhop_object *nh);

	-static void destroy_subscription_epoch(epoch_context_t ctx);
	#ifdef ROUTE_MPATH
	static bool rib_can_multipath(struct rib_head *rh);
	#endif
	@@ -155,86 +137,9 @@
	&VNET_NAME(rib_route_ipv6_nexthop), 0, "Enable IPv4 route via IPv6 Next Hop address");
	#endif

	-/* Routing table UMA zone */
	-VNET_DEFINE_STATIC(uma_zone_t, rtzone);
	-#define V_rtzone VNET(rtzone)
	-
	/* Debug bits */
	SYSCTL_NODE(_net_route, OID_AUTO, debug, CTLFLAG_RD \| CTLFLAG_MPSAFE, 0, "");

	-void
	-vnet_rtzone_init(void)
	-{
	-
	- V_rtzone = uma_zcreate("rtentry", sizeof(struct rtentry),
	- NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0);
	-}
	-
	-#ifdef VIMAGE
	-void
	-vnet_rtzone_destroy(void)
	-{
	-
	- uma_zdestroy(V_rtzone);
	-}
	-#endif
	-
	-static void
	-destroy_rtentry(struct rtentry *rt)
	-{
	-#ifdef VIMAGE
	- struct nhop_object *nh = rt->rt_nhop;
	-
	- /*
	- * At this moment rnh, nh_control may be already freed.
	- * nhop interface may have been migrated to a different vnet.
	- * Use vnet stored in the nexthop to delete the entry.
	- */
	-#ifdef ROUTE_MPATH
	- if (NH_IS_NHGRP(nh)) {
	- const struct weightened_nhop *wn;
	- uint32_t num_nhops;
	- wn = nhgrp_get_nhops((struct nhgrp_object *)nh, &num_nhops);
	- nh = wn[0].nh;
	- }
	-#endif
	- CURVNET_SET(nhop_get_vnet(nh));
	-#endif
	-
	- /* Unreference nexthop */
	- nhop_free_any(rt->rt_nhop);
	-
	- uma_zfree(V_rtzone, rt);
	-
	- CURVNET_RESTORE();
	-}
	-
	-/*
	- * Epoch callback indicating rtentry is safe to destroy
	- */
	-static void
	-destroy_rtentry_epoch(epoch_context_t ctx)
	-{
	- struct rtentry *rt;
	-
	- rt = __containerof(ctx, struct rtentry, rt_epoch_ctx);
	-
	- destroy_rtentry(rt);
	-}
	-
	-/*
	- * Schedule rtentry deletion
	- */
	-static void
	-rtfree(struct rtentry *rt)
	-{
	-
	- KASSERT(rt != NULL, ("%s: NULL rt", __func__));
	-
	- epoch_call(net_epoch_preempt, destroy_rtentry_epoch,
	- &rt->rt_epoch_ctx);
	-}
	-
	static struct rib_head *
	get_rnh(uint32_t fibnum, const struct rt_addrinfo *info)
	{
	@@ -316,142 +221,6 @@
	return (weight);
	}

	-bool
	-rt_is_host(const struct rtentry *rt)
	-{
	-
	- return (rt->rte_flags & RTF_HOST);
	-}
	-
	-sa_family_t
	-rt_get_family(const struct rtentry *rt)
	-{
	- const struct sockaddr *dst;
	-
	- dst = (const struct sockaddr *)rt_key_const(rt);
	-
	- return (dst->sa_family);
	-}
	-
	-/*
	- * Returns pointer to nexthop or nexthop group
	- * associated with @rt
	- */
	-struct nhop_object *
	-rt_get_raw_nhop(const struct rtentry *rt)
	-{
	-
	- return (rt->rt_nhop);
	-}
	-
	-#ifdef INET
	-/*
	- * Stores IPv4 address and prefix length of @rt inside
	- * @paddr and @plen.
	- * @pscopeid is currently always set to 0.
	- */
	-void
	-rt_get_inet_prefix_plen(const struct rtentry rt, struct in_addr paddr,
	- int plen, uint32_t pscopeid)
	-{
	- const struct sockaddr_in *dst;
	-
	- dst = (const struct sockaddr_in *)rt_key_const(rt);
	- KASSERT((dst->sin_family == AF_INET),
	- ("rt family is %d, not inet", dst->sin_family));
	- *paddr = dst->sin_addr;
	- dst = (const struct sockaddr_in *)rt_mask_const(rt);
	- if (dst == NULL)
	- *plen = 32;
	- else
	- *plen = bitcount32(dst->sin_addr.s_addr);
	- *pscopeid = 0;
	-}
	-
	-/*
	- * Stores IPv4 address and prefix mask of @rt inside
	- * @paddr and @pmask. Sets mask to INADDR_ANY for host routes.
	- * @pscopeid is currently always set to 0.
	- */
	-void
	-rt_get_inet_prefix_pmask(const struct rtentry rt, struct in_addr paddr,
	- struct in_addr pmask, uint32_t pscopeid)
	-{
	- const struct sockaddr_in *dst;
	-
	- dst = (const struct sockaddr_in *)rt_key_const(rt);
	- KASSERT((dst->sin_family == AF_INET),
	- ("rt family is %d, not inet", dst->sin_family));
	- *paddr = dst->sin_addr;
	- dst = (const struct sockaddr_in *)rt_mask_const(rt);
	- if (dst == NULL)
	- pmask->s_addr = INADDR_BROADCAST;
	- else
	- *pmask = dst->sin_addr;
	- *pscopeid = 0;
	-}
	-#endif
	-
	-#ifdef INET6
	-static int
	-inet6_get_plen(const struct in6_addr *addr)
	-{
	-
	- return (bitcount32(addr->s6_addr32[0]) + bitcount32(addr->s6_addr32[1]) +
	- bitcount32(addr->s6_addr32[2]) + bitcount32(addr->s6_addr32[3]));
	-}
	-
	-/*
	- * Stores IPv6 address and prefix length of @rt inside
	- * @paddr and @plen. Addresses are returned in de-embedded form.
	- * Scopeid is set to 0 for non-LL addresses.
	- */
	-void
	-rt_get_inet6_prefix_plen(const struct rtentry rt, struct in6_addr paddr,
	- int plen, uint32_t pscopeid)
	-{
	- const struct sockaddr_in6 *dst;
	-
	- dst = (const struct sockaddr_in6 *)rt_key_const(rt);
	- KASSERT((dst->sin6_family == AF_INET6),
	- ("rt family is %d, not inet6", dst->sin6_family));
	- if (IN6_IS_SCOPE_LINKLOCAL(&dst->sin6_addr))
	- in6_splitscope(&dst->sin6_addr, paddr, pscopeid);
	- else
	- *paddr = dst->sin6_addr;
	- dst = (const struct sockaddr_in6 *)rt_mask_const(rt);
	- if (dst == NULL)
	- *plen = 128;
	- else
	- *plen = inet6_get_plen(&dst->sin6_addr);
	-}
	-
	-/*
	- * Stores IPv6 address and prefix mask of @rt inside
	- * @paddr and @pmask. Addresses are returned in de-embedded form.
	- * Scopeid is set to 0 for non-LL addresses.
	- */
	-void
	-rt_get_inet6_prefix_pmask(const struct rtentry rt, struct in6_addr paddr,
	- struct in6_addr pmask, uint32_t pscopeid)
	-{
	- const struct sockaddr_in6 *dst;
	-
	- dst = (const struct sockaddr_in6 *)rt_key_const(rt);
	- KASSERT((dst->sin6_family == AF_INET6),
	- ("rt family is %d, not inet", dst->sin6_family));
	- if (IN6_IS_SCOPE_LINKLOCAL(&dst->sin6_addr))
	- in6_splitscope(&dst->sin6_addr, paddr, pscopeid);
	- else
	- *paddr = dst->sin6_addr;
	- dst = (const struct sockaddr_in6 *)rt_mask_const(rt);
	- if (dst == NULL)
	- memset(pmask, 0xFF, sizeof(struct in6_addr));
	- else
	- *pmask = dst->sin6_addr;
	-}
	-#endif
	-
	/*
	* File-local concept for distingushing between the normal and
	* RTF_PINNED routes tha can override the "normal" one.
	@@ -700,7 +469,7 @@
	}

	if (op_flags & RTM_F_CREATE) {
	- if ((rt = create_rtentry(rnh, dst, netmask)) == NULL)
	+ if ((rt = rt_alloc(rnh, dst, netmask)) == NULL)
	return (ENOMEM);
	} else {
	struct route_nhop_data rnd_tmp;
	@@ -795,7 +564,7 @@
	rib_notify(rnh, RIB_NOTIFY_DELAYED, rc);

	if (rc->rc_cmd == RTM_DELETE)
	- rtfree(rc->rc_rt);
	+ rt_free(rc->rc_rt);
	#ifdef ROUTE_MPATH
	else {
	/*
	@@ -874,36 +643,6 @@
	return (false);
	}

	-/*
	- * Creates rtentry and nexthop based on @info data.
	- * Return 0 and fills in rtentry into @prt on success,
	- * Note: rtentry mask will be set to RTAX_NETMASK, thus its pointer is required
	- * to be stable till the end of the operation (radix rt insertion/change/removal).
	- * return errno otherwise.
	- */
	-static struct rtentry *
	-create_rtentry(struct rib_head rnh, const struct sockaddr dst,
	- struct sockaddr *netmask)
	-{
	- MPASS(dst->sa_len <= sizeof(((struct rtentry *)NULL)->rt_dstb));
	-
	- struct rtentry *rt = uma_zalloc(V_rtzone, M_NOWAIT \| M_ZERO);
	- if (rt == NULL)
	- return (NULL);
	- rt->rte_flags = RTF_UP \| (netmask == NULL ? RTF_HOST : 0);
	-
	- /* Fill in dst, ensuring it's masked if needed. */
	- if (netmask != NULL) {
	- rt_maskedcopy(dst, &rt->rt_dst, netmask);
	- } else
	- bcopy(dst, &rt->rt_dst, dst->sa_len);
	- rt_key(rt) = &rt->rt_dst;
	- /* Set netmask to the storage from info. It will be updated upon insertion */
	- rt_mask(rt) = netmask;
	-
	- return (rt);
	-}
	-
	static int
	add_route_byinfo(struct rib_head rnh, struct rt_addrinfo info,
	struct rib_cmd_info *rc)
	@@ -941,12 +680,12 @@
	return (error);
	}

	- if ((rt = create_rtentry(rnh, dst, netmask)) == NULL)
	+ if ((rt = rt_alloc(rnh, dst, netmask)) == NULL)
	return (ENOBUFS);

	error = nhop_create_from_info(rnh, info, &nh);
	if (error != 0) {
	- uma_zfree(V_rtzone, rt);
	+ rt_free_immediate(rt);
	return (error);
	}

	@@ -1029,7 +768,7 @@
	nhop_free_any(rnd_add->rnd_nhop);
	if (op_flags & RTM_F_CREATE) {
	if (error != 0 \|\| rc->rc_cmd != RTM_ADD)
	- uma_zfree(V_rtzone, rt);
	+ rt_free_immediate(rt);
	}
	return (error);
	}
	@@ -1038,7 +777,7 @@
	error = EEXIST;
	out:
	if (op_flags & RTM_F_CREATE)
	- uma_zfree(V_rtzone, rt);
	+ rt_free_immediate(rt);
	nhop_free_any(nh);

	return (error);
	@@ -1154,7 +893,7 @@
	rib_notify(rnh, RIB_NOTIFY_DELAYED, rc);

	if (rc->rc_cmd == RTM_DELETE)
	- rtfree(rc->rc_rt);
	+ rt_free(rc->rc_rt);
	#ifdef ROUTE_MPATH
	else {
	/*
	@@ -1699,7 +1438,7 @@
	#endif
	rt_routemsg(RTM_DELETE, rt, nh, fibnum);
	}
	- rtfree(rt);
	+ rt_free(rt);
	}

	NET_EPOCH_EXIT(et);
	@@ -1713,7 +1452,7 @@

	rn = rnh->rnh_deladdr(rt_key(rt), rt_mask(rt), &rnh->head);
	if (RNTORT(rn) == rt)
	- rtfree(rt);
	+ rt_free(rt);

	return (0);
	}
	@@ -1754,162 +1493,3 @@
	}
	return ("unknown");
	}
	-
	-static void
	-rib_notify(struct rib_head *rnh, enum rib_subscription_type type,
	- struct rib_cmd_info *rc)
	-{
	- struct rib_subscription *rs;
	-
	- CK_STAILQ_FOREACH(rs, &rnh->rnh_subscribers, next) {
	- if (rs->type == type)
	- rs->func(rnh, rc, rs->arg);
	- }
	-}
	-
	-static struct rib_subscription *
	-allocate_subscription(rib_subscription_cb_t f, void arg,
	- enum rib_subscription_type type, bool waitok)
	-{
	- struct rib_subscription *rs;
	- int flags = M_ZERO \| (waitok ? M_WAITOK : M_NOWAIT);
	-
	- rs = malloc(sizeof(struct rib_subscription), M_RTABLE, flags);
	- if (rs == NULL)
	- return (NULL);
	-
	- rs->func = f;
	- rs->arg = arg;
	- rs->type = type;
	-
	- return (rs);
	-}
	-
	-/*
	- * Subscribe for the changes in the routing table specified by @fibnum and
	- * @family.
	- *
	- * Returns pointer to the subscription structure on success.
	- */
	-struct rib_subscription *
	-rib_subscribe(uint32_t fibnum, int family, rib_subscription_cb_t f, void arg,
	- enum rib_subscription_type type, bool waitok)
	-{
	- struct rib_head *rnh;
	- struct epoch_tracker et;
	-
	- NET_EPOCH_ENTER(et);
	- KASSERT((fibnum < rt_numfibs), ("%s: bad fibnum", __func__));
	- rnh = rt_tables_get_rnh(fibnum, family);
	- NET_EPOCH_EXIT(et);
	-
	- return (rib_subscribe_internal(rnh, f, arg, type, waitok));
	-}
	-
	-struct rib_subscription *
	-rib_subscribe_internal(struct rib_head rnh, rib_subscription_cb_t f, void *arg,
	- enum rib_subscription_type type, bool waitok)
	-{
	- struct rib_subscription *rs;
	- struct epoch_tracker et;
	-
	- if ((rs = allocate_subscription(f, arg, type, waitok)) == NULL)
	- return (NULL);
	- rs->rnh = rnh;
	-
	- NET_EPOCH_ENTER(et);
	- RIB_WLOCK(rnh);
	- CK_STAILQ_INSERT_HEAD(&rnh->rnh_subscribers, rs, next);
	- RIB_WUNLOCK(rnh);
	- NET_EPOCH_EXIT(et);
	-
	- return (rs);
	-}
	-
	-struct rib_subscription *
	-rib_subscribe_locked(struct rib_head rnh, rib_subscription_cb_t f, void *arg,
	- enum rib_subscription_type type)
	-{
	- struct rib_subscription *rs;
	-
	- NET_EPOCH_ASSERT();
	- RIB_WLOCK_ASSERT(rnh);
	-
	- if ((rs = allocate_subscription(f, arg, type, false)) == NULL)
	- return (NULL);
	- rs->rnh = rnh;
	-
	- CK_STAILQ_INSERT_HEAD(&rnh->rnh_subscribers, rs, next);
	-
	- return (rs);
	-}
	-
	-/*
	- * Remove rtable subscription @rs from the routing table.
	- * Needs to be run in network epoch.
	- */
	-void
	-rib_unsubscribe(struct rib_subscription *rs)
	-{
	- struct rib_head *rnh = rs->rnh;
	-
	- NET_EPOCH_ASSERT();
	-
	- RIB_WLOCK(rnh);
	- CK_STAILQ_REMOVE(&rnh->rnh_subscribers, rs, rib_subscription, next);
	- RIB_WUNLOCK(rnh);
	-
	- epoch_call(net_epoch_preempt, destroy_subscription_epoch,
	- &rs->epoch_ctx);
	-}
	-
	-void
	-rib_unsubscribe_locked(struct rib_subscription *rs)
	-{
	- struct rib_head *rnh = rs->rnh;
	-
	- NET_EPOCH_ASSERT();
	- RIB_WLOCK_ASSERT(rnh);
	-
	- CK_STAILQ_REMOVE(&rnh->rnh_subscribers, rs, rib_subscription, next);
	-
	- epoch_call(net_epoch_preempt, destroy_subscription_epoch,
	- &rs->epoch_ctx);
	-}
	-
	-/*
	- * Epoch callback indicating subscription is safe to destroy
	- */
	-static void
	-destroy_subscription_epoch(epoch_context_t ctx)
	-{
	- struct rib_subscription *rs;
	-
	- rs = __containerof(ctx, struct rib_subscription, epoch_ctx);
	-
	- free(rs, M_RTABLE);
	-}
	-
	-void
	-rib_init_subscriptions(struct rib_head *rnh)
	-{
	-
	- CK_STAILQ_INIT(&rnh->rnh_subscribers);
	-}
	-
	-void
	-rib_destroy_subscriptions(struct rib_head *rnh)
	-{
	- struct rib_subscription *rs;
	- struct epoch_tracker et;
	-
	- NET_EPOCH_ENTER(et);
	- RIB_WLOCK(rnh);
	- while ((rs = CK_STAILQ_FIRST(&rnh->rnh_subscribers)) != NULL) {
	- CK_STAILQ_REMOVE_HEAD(&rnh->rnh_subscribers, next);
	- epoch_call(net_epoch_preempt, destroy_subscription_epoch,
	- &rs->epoch_ctx);
	- }
	- RIB_WUNLOCK(rnh);
	- NET_EPOCH_EXIT(et);
	-}
	diff --git a/sys/net/route/route_rtentry.c b/sys/net/route/route_rtentry.c
	new file mode 100644
	--- /dev/null
	+++ b/sys/net/route/route_rtentry.c
	@@ -0,0 +1,303 @@
	+/*-
	+ * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
	+ *
	+ * Copyright (c) 2021-2022 Alexander V. Chernikov
	+ *
	+ * Redistribution and use in source and binary forms, with or without
	+ * modification, are permitted provided that the following conditions
	+ * are met:
	+ * 1. Redistributions of source code must retain the above copyright
	+ * notice, this list of conditions and the following disclaimer.
	+ * 2. Redistributions in binary form must reproduce the above copyright
	+ * notice, this list of conditions and the following disclaimer in the
	+ * documentation and/or other materials provided with the distribution.
	+ *
	+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
	+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	+ * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
	+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
	+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
	+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
	+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
	+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
	+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
	+ * SUCH DAMAGE.
	+ */
	+
	+#include <sys/cdefs.h>
	+__FBSDID("$FreeBSD$");
	+#include "opt_inet.h"
	+#include "opt_inet6.h"
	+#include "opt_route.h"
	+
	+#include <sys/param.h>
	+#include <sys/systm.h>
	+#include <sys/malloc.h>
	+#include <sys/socket.h>
	+#include <sys/kernel.h>
	+#include <sys/lock.h>
	+#include <sys/rmlock.h>
	+
	+#include <net/if.h>
	+#include <net/if_var.h>
	+#include <net/vnet.h>
	+#include <net/route.h>
	+#include <net/route/route_ctl.h>
	+#include <net/route/route_var.h>
	+#include <net/route/nhop.h>
	+#include <netinet/in.h>
	+#include <netinet6/scope6_var.h>
	+#include <netinet6/in6_var.h>
	+
	+#include <vm/uma.h>
	+
	+/* Routing table UMA zone */
	+VNET_DEFINE_STATIC(uma_zone_t, rtzone);
	+#define V_rtzone VNET(rtzone)
	+
	+void
	+vnet_rtzone_init(void)
	+{
	+
	+ V_rtzone = uma_zcreate("rtentry", sizeof(struct rtentry),
	+ NULL, NULL, NULL, NULL, UMA_ALIGN_PTR, 0);
	+}
	+
	+#ifdef VIMAGE
	+void
	+vnet_rtzone_destroy(void)
	+{
	+
	+ uma_zdestroy(V_rtzone);
	+}
	+#endif
	+
	+/*
	+ * Creates rtentry and based on @dst/@netmask data.
	+ * Return 0 and fills in rtentry into @prt on success,
	+ * Note: rtentry mask ptr will be set to @netmask , thus its pointer is required
	+ * to be stable till the end of the operation (radix rt insertion/change/removal).
	+ */
	+struct rtentry *
	+rt_alloc(struct rib_head rnh, const struct sockaddr dst,
	+ struct sockaddr *netmask)
	+{
	+ MPASS(dst->sa_len <= sizeof(((struct rtentry *)NULL)->rt_dstb));
	+
	+ struct rtentry *rt = uma_zalloc(V_rtzone, M_NOWAIT \| M_ZERO);
	+ if (rt == NULL)
	+ return (NULL);
	+ rt->rte_flags = RTF_UP \| (netmask == NULL ? RTF_HOST : 0);
	+
	+ /* Fill in dst, ensuring it's masked if needed. */
	+ if (netmask != NULL) {
	+ rt_maskedcopy(dst, &rt->rt_dst, netmask);
	+ } else
	+ bcopy(dst, &rt->rt_dst, dst->sa_len);
	+ rt_key(rt) = &rt->rt_dst;
	+ /* Set netmask to the storage from info. It will be updated upon insertion */
	+ rt_mask(rt) = netmask;
	+
	+ return (rt);
	+}
	+
	+static void
	+destroy_rtentry(struct rtentry *rt)
	+{
	+#ifdef VIMAGE
	+ struct nhop_object *nh = rt->rt_nhop;
	+
	+ /*
	+ * At this moment rnh, nh_control may be already freed.
	+ * nhop interface may have been migrated to a different vnet.
	+ * Use vnet stored in the nexthop to delete the entry.
	+ */
	+#ifdef ROUTE_MPATH
	+ if (NH_IS_NHGRP(nh)) {
	+ const struct weightened_nhop *wn;
	+ uint32_t num_nhops;
	+ wn = nhgrp_get_nhops((struct nhgrp_object *)nh, &num_nhops);
	+ nh = wn[0].nh;
	+ }
	+#endif
	+ CURVNET_SET(nhop_get_vnet(nh));
	+#endif
	+
	+ /* Unreference nexthop */
	+ nhop_free_any(rt->rt_nhop);
	+
	+ rt_free_immediate(rt);
	+
	+ CURVNET_RESTORE();
	+}
	+
	+/*
	+ * Epoch callback indicating rtentry is safe to destroy
	+ */
	+static void
	+destroy_rtentry_epoch(epoch_context_t ctx)
	+{
	+ struct rtentry *rt;
	+
	+ rt = __containerof(ctx, struct rtentry, rt_epoch_ctx);
	+
	+ destroy_rtentry(rt);
	+}
	+
	+/*
	+ * Schedule rtentry deletion
	+ */
	+void
	+rt_free(struct rtentry *rt)
	+{
	+
	+ KASSERT(rt != NULL, ("%s: NULL rt", __func__));
	+
	+ epoch_call(net_epoch_preempt, destroy_rtentry_epoch,
	+ &rt->rt_epoch_ctx);
	+}
	+
	+void
	+rt_free_immediate(struct rtentry *rt)
	+{
	+ uma_zfree(V_rtzone, rt);
	+}
	+
	+bool
	+rt_is_host(const struct rtentry *rt)
	+{
	+
	+ return (rt->rte_flags & RTF_HOST);
	+}
	+
	+sa_family_t
	+rt_get_family(const struct rtentry *rt)
	+{
	+ const struct sockaddr *dst;
	+
	+ dst = (const struct sockaddr *)rt_key_const(rt);
	+
	+ return (dst->sa_family);
	+}
	+
	+/*
	+ * Returns pointer to nexthop or nexthop group
	+ * associated with @rt
	+ */
	+struct nhop_object *
	+rt_get_raw_nhop(const struct rtentry *rt)
	+{
	+
	+ return (rt->rt_nhop);
	+}
	+
	+#ifdef INET
	+/*
	+ * Stores IPv4 address and prefix length of @rt inside
	+ * @paddr and @plen.
	+ * @pscopeid is currently always set to 0.
	+ */
	+void
	+rt_get_inet_prefix_plen(const struct rtentry rt, struct in_addr paddr,
	+ int plen, uint32_t pscopeid)
	+{
	+ const struct sockaddr_in *dst;
	+
	+ dst = (const struct sockaddr_in *)rt_key_const(rt);
	+ KASSERT((dst->sin_family == AF_INET),
	+ ("rt family is %d, not inet", dst->sin_family));
	+ *paddr = dst->sin_addr;
	+ dst = (const struct sockaddr_in *)rt_mask_const(rt);
	+ if (dst == NULL)
	+ *plen = 32;
	+ else
	+ *plen = bitcount32(dst->sin_addr.s_addr);
	+ *pscopeid = 0;
	+}
	+
	+/*
	+ * Stores IPv4 address and prefix mask of @rt inside
	+ * @paddr and @pmask. Sets mask to INADDR_ANY for host routes.
	+ * @pscopeid is currently always set to 0.
	+ */
	+void
	+rt_get_inet_prefix_pmask(const struct rtentry rt, struct in_addr paddr,
	+ struct in_addr pmask, uint32_t pscopeid)
	+{
	+ const struct sockaddr_in *dst;
	+
	+ dst = (const struct sockaddr_in *)rt_key_const(rt);
	+ KASSERT((dst->sin_family == AF_INET),
	+ ("rt family is %d, not inet", dst->sin_family));
	+ *paddr = dst->sin_addr;
	+ dst = (const struct sockaddr_in *)rt_mask_const(rt);
	+ if (dst == NULL)
	+ pmask->s_addr = INADDR_BROADCAST;
	+ else
	+ *pmask = dst->sin_addr;
	+ *pscopeid = 0;
	+}
	+#endif
	+
	+#ifdef INET6
	+static int
	+inet6_get_plen(const struct in6_addr *addr)
	+{
	+
	+ return (bitcount32(addr->s6_addr32[0]) + bitcount32(addr->s6_addr32[1]) +
	+ bitcount32(addr->s6_addr32[2]) + bitcount32(addr->s6_addr32[3]));
	+}
	+
	+/*
	+ * Stores IPv6 address and prefix length of @rt inside
	+ * @paddr and @plen. Addresses are returned in de-embedded form.
	+ * Scopeid is set to 0 for non-LL addresses.
	+ */
	+void
	+rt_get_inet6_prefix_plen(const struct rtentry rt, struct in6_addr paddr,
	+ int plen, uint32_t pscopeid)
	+{
	+ const struct sockaddr_in6 *dst;
	+
	+ dst = (const struct sockaddr_in6 *)rt_key_const(rt);
	+ KASSERT((dst->sin6_family == AF_INET6),
	+ ("rt family is %d, not inet6", dst->sin6_family));
	+ if (IN6_IS_SCOPE_LINKLOCAL(&dst->sin6_addr))
	+ in6_splitscope(&dst->sin6_addr, paddr, pscopeid);
	+ else
	+ *paddr = dst->sin6_addr;
	+ dst = (const struct sockaddr_in6 *)rt_mask_const(rt);
	+ if (dst == NULL)
	+ *plen = 128;
	+ else
	+ *plen = inet6_get_plen(&dst->sin6_addr);
	+}
	+
	+/*
	+ * Stores IPv6 address and prefix mask of @rt inside
	+ * @paddr and @pmask. Addresses are returned in de-embedded form.
	+ * Scopeid is set to 0 for non-LL addresses.
	+ */
	+void
	+rt_get_inet6_prefix_pmask(const struct rtentry rt, struct in6_addr paddr,
	+ struct in6_addr pmask, uint32_t pscopeid)
	+{
	+ const struct sockaddr_in6 *dst;
	+
	+ dst = (const struct sockaddr_in6 *)rt_key_const(rt);
	+ KASSERT((dst->sin6_family == AF_INET6),
	+ ("rt family is %d, not inet", dst->sin6_family));
	+ if (IN6_IS_SCOPE_LINKLOCAL(&dst->sin6_addr))
	+ in6_splitscope(&dst->sin6_addr, paddr, pscopeid);
	+ else
	+ *paddr = dst->sin6_addr;
	+ dst = (const struct sockaddr_in6 *)rt_mask_const(rt);
	+ if (dst == NULL)
	+ memset(pmask, 0xFF, sizeof(struct in6_addr));
	+ else
	+ *pmask = dst->sin6_addr;
	+}
	+#endif
	+
	+
	diff --git a/sys/net/route/route_subscription.c b/sys/net/route/route_subscription.c
	new file mode 100644
	--- /dev/null
	+++ b/sys/net/route/route_subscription.c
	@@ -0,0 +1,213 @@
	+/*-
	+ * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
	+ *
	+ * Copyright (c) 2021-2022 Alexander V. Chernikov
	+ *
	+ * Redistribution and use in source and binary forms, with or without
	+ * modification, are permitted provided that the following conditions
	+ * are met:
	+ * 1. Redistributions of source code must retain the above copyright
	+ * notice, this list of conditions and the following disclaimer.
	+ * 2. Redistributions in binary form must reproduce the above copyright
	+ * notice, this list of conditions and the following disclaimer in the
	+ * documentation and/or other materials provided with the distribution.
	+ *
	+ * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
	+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
	+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
	+ * ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
	+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
	+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
	+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
	+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
	+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
	+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
	+ * SUCH DAMAGE.
	+ */
	+
	+#include <sys/cdefs.h>
	+__FBSDID("$FreeBSD$");
	+#include "opt_route.h"
	+
	+#include <sys/param.h>
	+#include <sys/systm.h>
	+#include <sys/malloc.h>
	+#include <sys/socket.h>
	+#include <sys/kernel.h>
	+#include <sys/lock.h>
	+#include <sys/rmlock.h>
	+
	+#include <net/route.h>
	+#include <net/route/route_ctl.h>
	+#include <net/route/route_var.h>
	+#include <net/route/nhop.h>
	+
	+struct rib_subscription {
	+ CK_STAILQ_ENTRY(rib_subscription) next;
	+ rib_subscription_cb_t *func;
	+ void *arg;
	+ struct rib_head *rnh;
	+ enum rib_subscription_type type;
	+ struct epoch_context epoch_ctx;
	+};
	+
	+static void destroy_subscription_epoch(epoch_context_t ctx);
	+
	+void
	+rib_notify(struct rib_head *rnh, enum rib_subscription_type type,
	+ struct rib_cmd_info *rc)
	+{
	+ struct rib_subscription *rs;
	+
	+ CK_STAILQ_FOREACH(rs, &rnh->rnh_subscribers, next) {
	+ if (rs->type == type)
	+ rs->func(rnh, rc, rs->arg);
	+ }
	+}
	+
	+static struct rib_subscription *
	+allocate_subscription(rib_subscription_cb_t f, void arg,
	+ enum rib_subscription_type type, bool waitok)
	+{
	+ struct rib_subscription *rs;
	+ int flags = M_ZERO \| (waitok ? M_WAITOK : M_NOWAIT);
	+
	+ rs = malloc(sizeof(struct rib_subscription), M_RTABLE, flags);
	+ if (rs == NULL)
	+ return (NULL);
	+
	+ rs->func = f;
	+ rs->arg = arg;
	+ rs->type = type;
	+
	+ return (rs);
	+}
	+
	+/*
	+ * Subscribe for the changes in the routing table specified by @fibnum and
	+ * @family.
	+ *
	+ * Returns pointer to the subscription structure on success.
	+ */
	+struct rib_subscription *
	+rib_subscribe(uint32_t fibnum, int family, rib_subscription_cb_t f, void arg,
	+ enum rib_subscription_type type, bool waitok)
	+{
	+ struct rib_head *rnh;
	+ struct epoch_tracker et;
	+
	+ NET_EPOCH_ENTER(et);
	+ KASSERT((fibnum < rt_numfibs), ("%s: bad fibnum", __func__));
	+ rnh = rt_tables_get_rnh(fibnum, family);
	+ NET_EPOCH_EXIT(et);
	+
	+ return (rib_subscribe_internal(rnh, f, arg, type, waitok));
	+}
	+
	+struct rib_subscription *
	+rib_subscribe_internal(struct rib_head rnh, rib_subscription_cb_t f, void *arg,
	+ enum rib_subscription_type type, bool waitok)
	+{
	+ struct rib_subscription *rs;
	+ struct epoch_tracker et;
	+
	+ if ((rs = allocate_subscription(f, arg, type, waitok)) == NULL)
	+ return (NULL);
	+ rs->rnh = rnh;
	+
	+ NET_EPOCH_ENTER(et);
	+ RIB_WLOCK(rnh);
	+ CK_STAILQ_INSERT_HEAD(&rnh->rnh_subscribers, rs, next);
	+ RIB_WUNLOCK(rnh);
	+ NET_EPOCH_EXIT(et);
	+
	+ return (rs);
	+}
	+
	+struct rib_subscription *
	+rib_subscribe_locked(struct rib_head rnh, rib_subscription_cb_t f, void *arg,
	+ enum rib_subscription_type type)
	+{
	+ struct rib_subscription *rs;
	+
	+ NET_EPOCH_ASSERT();
	+ RIB_WLOCK_ASSERT(rnh);
	+
	+ if ((rs = allocate_subscription(f, arg, type, false)) == NULL)
	+ return (NULL);
	+ rs->rnh = rnh;
	+
	+ CK_STAILQ_INSERT_HEAD(&rnh->rnh_subscribers, rs, next);
	+
	+ return (rs);
	+}
	+
	+/*
	+ * Remove rtable subscription @rs from the routing table.
	+ * Needs to be run in network epoch.
	+ */
	+void
	+rib_unsubscribe(struct rib_subscription *rs)
	+{
	+ struct rib_head *rnh = rs->rnh;
	+
	+ NET_EPOCH_ASSERT();
	+
	+ RIB_WLOCK(rnh);
	+ CK_STAILQ_REMOVE(&rnh->rnh_subscribers, rs, rib_subscription, next);
	+ RIB_WUNLOCK(rnh);
	+
	+ epoch_call(net_epoch_preempt, destroy_subscription_epoch,
	+ &rs->epoch_ctx);
	+}
	+
	+void
	+rib_unsubscribe_locked(struct rib_subscription *rs)
	+{
	+ struct rib_head *rnh = rs->rnh;
	+
	+ NET_EPOCH_ASSERT();
	+ RIB_WLOCK_ASSERT(rnh);
	+
	+ CK_STAILQ_REMOVE(&rnh->rnh_subscribers, rs, rib_subscription, next);
	+
	+ epoch_call(net_epoch_preempt, destroy_subscription_epoch,
	+ &rs->epoch_ctx);
	+}
	+
	+/*
	+ * Epoch callback indicating subscription is safe to destroy
	+ */
	+static void
	+destroy_subscription_epoch(epoch_context_t ctx)
	+{
	+ struct rib_subscription *rs;
	+
	+ rs = __containerof(ctx, struct rib_subscription, epoch_ctx);
	+
	+ free(rs, M_RTABLE);
	+}
	+
	+void
	+rib_init_subscriptions(struct rib_head *rnh)
	+{
	+
	+ CK_STAILQ_INIT(&rnh->rnh_subscribers);
	+}
	+
	+void
	+rib_destroy_subscriptions(struct rib_head *rnh)
	+{
	+ struct rib_subscription *rs;
	+ struct epoch_tracker et;
	+
	+ NET_EPOCH_ENTER(et);
	+ RIB_WLOCK(rnh);
	+ while ((rs = CK_STAILQ_FIRST(&rnh->rnh_subscribers)) != NULL) {
	+ CK_STAILQ_REMOVE_HEAD(&rnh->rnh_subscribers, next);
	+ epoch_call(net_epoch_preempt, destroy_subscription_epoch,
	+ &rs->epoch_ctx);
	+ }
	+ RIB_WUNLOCK(rnh);
	+ NET_EPOCH_EXIT(et);
	+}
	diff --git a/sys/net/route/route_var.h b/sys/net/route/route_var.h
	--- a/sys/net/route/route_var.h
	+++ b/sys/net/route/route_var.h
	@@ -227,8 +227,13 @@
	int check_info_match_nhop(const struct rt_addrinfo *info,
	const struct rtentry rt, const struct nhop_object nh);

	+/* route_rtentry.c */
	void vnet_rtzone_init(void);
	void vnet_rtzone_destroy(void);
	+void rt_free(struct rtentry *rt);
	+void rt_free_immediate(struct rtentry *rt);
	+struct rtentry rt_alloc(struct rib_head rnh, const struct sockaddr *dst,
	+ struct sockaddr *netmask);

	/* subscriptions */
	void rib_init_subscriptions(struct rib_head *rnh);

File Metadata

Mime Type: text/plain
Expires: Mon, Feb 23, 2:14 AM (7 h, 15 m)
Storage Engine: blob
Storage Format: Raw Data
Storage Handle: 28948963
Default Alt Text: D36074.diff (28 KB)

D36074.diffNo OneTemporaryActions

D36074.diffView Options

File Metadata

Event Timeline

D36074.diff
No OneTemporary
Actions

D36074.diff
View Options