From 554cfb7ee5d4f2d0edb280e66e4a2db1906a8300 Mon Sep 17 00:00:00 2001 From: YOSHIFUJI Hideaki Date: Mon, 20 Mar 2006 17:00:26 -0800 Subject: [IPV6]: ROUTE: Eliminate lock for default route pointer. And prepare for more advanced router selection. Signed-off-by: YOSHIFUJI Hideaki Signed-off-by: David S. Miller --- net/ipv6/ip6_fib.c | 1 - net/ipv6/route.c | 197 +++++++++++++++++++---------------------------------- 2 files changed, 69 insertions(+), 129 deletions(-) (limited to 'net') diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c index 1bf6d9a769e..2cb6149349b 100644 --- a/net/ipv6/ip6_fib.c +++ b/net/ipv6/ip6_fib.c @@ -1105,7 +1105,6 @@ static int fib6_age(struct rt6_info *rt, void *arg) if (rt->rt6i_flags&RTF_EXPIRES && rt->rt6i_expires) { if (time_after(now, rt->rt6i_expires)) { RT6_TRACE("expiring %p\n", rt); - rt6_reset_dflt_pointer(rt); return -1; } gc_args.more++; diff --git a/net/ipv6/route.c b/net/ipv6/route.c index 6a4019a4ca8..f71e2365b43 100644 --- a/net/ipv6/route.c +++ b/net/ipv6/route.c @@ -74,6 +74,9 @@ #define CLONE_OFFLINK_ROUTE 0 +#define RT6_SELECT_F_IFACE 0x1 +#define RT6_SELECT_F_REACHABLE 0x2 + static int ip6_rt_max_size = 4096; static int ip6_rt_gc_min_interval = HZ / 2; static int ip6_rt_gc_timeout = 60*HZ; @@ -216,148 +219,89 @@ static __inline__ struct rt6_info *rt6_device_match(struct rt6_info *rt, } /* - * pointer to the last default router chosen. BH is disabled locally. + * Default Router Selection (RFC 2461 6.3.6) */ -static struct rt6_info *rt6_dflt_pointer; -static DEFINE_SPINLOCK(rt6_dflt_lock); +static int inline rt6_check_dev(struct rt6_info *rt, int oif) +{ + struct net_device *dev = rt->rt6i_dev; + if (!oif || dev->ifindex == oif) + return 2; + if ((dev->flags & IFF_LOOPBACK) && + rt->rt6i_idev && rt->rt6i_idev->dev->ifindex == oif) + return 1; + return 0; +} -void rt6_reset_dflt_pointer(struct rt6_info *rt) +static int inline rt6_check_neigh(struct rt6_info *rt) { - spin_lock_bh(&rt6_dflt_lock); - if (rt == NULL || rt == rt6_dflt_pointer) { - RT6_TRACE("reset default router: %p->NULL\n", rt6_dflt_pointer); - rt6_dflt_pointer = NULL; + struct neighbour *neigh = rt->rt6i_nexthop; + int m = 0; + if (neigh) { + read_lock_bh(&neigh->lock); + if (neigh->nud_state & NUD_VALID) + m = 1; + read_unlock_bh(&neigh->lock); } - spin_unlock_bh(&rt6_dflt_lock); + return m; } -/* Default Router Selection (RFC 2461 6.3.6) */ -static struct rt6_info *rt6_best_dflt(struct rt6_info *rt, int oif) +static int rt6_score_route(struct rt6_info *rt, int oif, + int strict) { - struct rt6_info *match = NULL; - struct rt6_info *sprt; - int mpri = 0; + int m = rt6_check_dev(rt, oif); + if (!m && (strict & RT6_SELECT_F_IFACE)) + return -1; + if (rt6_check_neigh(rt)) + m |= 4; + else if (strict & RT6_SELECT_F_REACHABLE) + return -1; + return m; +} - for (sprt = rt; sprt; sprt = sprt->u.next) { - struct neighbour *neigh; - int m = 0; +static struct rt6_info *rt6_select(struct rt6_info **head, int oif, + int strict) +{ + struct rt6_info *match = NULL, *last = NULL; + struct rt6_info *rt, *rt0 = *head; + u32 metric; + int mpri = -1; - if (!oif || - (sprt->rt6i_dev && - sprt->rt6i_dev->ifindex == oif)) - m += 8; + RT6_TRACE("%s(head=%p(*head=%p), oif=%d)\n", + __FUNCTION__, head, head ? *head : NULL, oif); - if (rt6_check_expired(sprt)) + for (rt = rt0, metric = rt0->rt6i_metric; + rt && rt->rt6i_metric == metric; + rt = rt->u.next) { + int m; + + if (rt6_check_expired(rt)) continue; - if (sprt == rt6_dflt_pointer) - m += 4; - - if ((neigh = sprt->rt6i_nexthop) != NULL) { - read_lock_bh(&neigh->lock); - switch (neigh->nud_state) { - case NUD_REACHABLE: - m += 3; - break; - - case NUD_STALE: - case NUD_DELAY: - case NUD_PROBE: - m += 2; - break; - - case NUD_NOARP: - case NUD_PERMANENT: - m += 1; - break; - - case NUD_INCOMPLETE: - default: - read_unlock_bh(&neigh->lock); - continue; - } - read_unlock_bh(&neigh->lock); - } else { + last = rt; + + m = rt6_score_route(rt, oif, strict); + if (m < 0) continue; - } - if (m > mpri || m >= 12) { - match = sprt; + if (m > mpri) { + match = rt; mpri = m; - if (m >= 12) { - /* we choose the last default router if it - * is in (probably) reachable state. - * If route changed, we should do pmtu - * discovery. --yoshfuji - */ - break; - } } } - spin_lock(&rt6_dflt_lock); - if (!match) { - /* - * No default routers are known to be reachable. - * SHOULD round robin - */ - if (rt6_dflt_pointer) { - for (sprt = rt6_dflt_pointer->u.next; - sprt; sprt = sprt->u.next) { - if (sprt->u.dst.obsolete <= 0 && - sprt->u.dst.error == 0 && - !rt6_check_expired(sprt)) { - match = sprt; - break; - } - } - for (sprt = rt; - !match && sprt; - sprt = sprt->u.next) { - if (sprt->u.dst.obsolete <= 0 && - sprt->u.dst.error == 0 && - !rt6_check_expired(sprt)) { - match = sprt; - break; - } - if (sprt == rt6_dflt_pointer) - break; - } - } - } - - if (match) { - if (rt6_dflt_pointer != match) - RT6_TRACE("changed default router: %p->%p\n", - rt6_dflt_pointer, match); - rt6_dflt_pointer = match; + if (!match && + (strict & RT6_SELECT_F_REACHABLE) && + last && last != rt0) { + /* no entries matched; do round-robin */ + *head = rt0->u.next; + rt0->u.next = last->u.next; + last->u.next = rt0; } - spin_unlock(&rt6_dflt_lock); - if (!match) { - /* - * Last Resort: if no default routers found, - * use addrconf default route. - * We don't record this route. - */ - for (sprt = ip6_routing_table.leaf; - sprt; sprt = sprt->u.next) { - if (!rt6_check_expired(sprt) && - (sprt->rt6i_flags & RTF_DEFAULT) && - (!oif || - (sprt->rt6i_dev && - sprt->rt6i_dev->ifindex == oif))) { - match = sprt; - break; - } - } - if (!match) { - /* no default route. give up. */ - match = &ip6_null_entry; - } - } + RT6_TRACE("%s() => %p, score=%d\n", + __FUNCTION__, match, mpri); - return match; + return (match ? match : &ip6_null_entry); } struct rt6_info *rt6_lookup(struct in6_addr *daddr, struct in6_addr *saddr, @@ -542,7 +486,7 @@ struct dst_entry * ip6_route_output(struct sock *sk, struct flowi *fl) int attempts = 3; int err; - strict = ipv6_addr_type(&fl->fl6_dst) & (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL); + strict = ipv6_addr_type(&fl->fl6_dst) & (IPV6_ADDR_MULTICAST|IPV6_ADDR_LINKLOCAL) ? RT6_SELECT_F_IFACE : 0; relookup: read_lock_bh(&rt6_lock); @@ -558,8 +502,9 @@ restart: goto out; } if (rt->rt6i_flags & RTF_DEFAULT) { - if (rt->rt6i_metric >= IP6_RT_PRIO_ADDRCONF) - rt = rt6_best_dflt(rt, fl->oif); + rt = rt6_select(&fn->leaf, fl->oif, strict | RT6_SELECT_F_REACHABLE); + if (rt == &ip6_null_entry) + rt = rt6_select(&fn->leaf, fl->oif, strict); } else { rt = rt6_device_match(rt, fl->oif, strict); BACKTRACK(); @@ -1025,8 +970,6 @@ int ip6_del_rt(struct rt6_info *rt, struct nlmsghdr *nlh, void *_rtattr, struct write_lock_bh(&rt6_lock); - rt6_reset_dflt_pointer(NULL); - err = fib6_del(rt, nlh, _rtattr, req); dst_release(&rt->u.dst); @@ -1341,8 +1284,6 @@ restart: if (rt->rt6i_flags & (RTF_DEFAULT | RTF_ADDRCONF)) { dst_hold(&rt->u.dst); - rt6_reset_dflt_pointer(NULL); - read_unlock_bh(&rt6_lock); ip6_del_rt(rt, NULL, NULL, NULL); -- cgit v1.2.3-70-g09d2