Merge branch 'master' of master.kernel.org:/pub/scm/linux/kernel/git/davem/net-2.6
[firefly-linux-kernel-4.4.55.git] / net / ipv6 / addrconf.c
index 3a835578fd1c8adae7d99f1573a16989d0c8d369..8572cb05fc21205db6be341e7aceb8d13f84bae1 100644 (file)
@@ -6,8 +6,6 @@
  *     Pedro Roque             <roque@di.fc.ul.pt>
  *     Alexey Kuznetsov        <kuznet@ms2.inr.ac.ru>
  *
- *     $Id: addrconf.c,v 1.69 2001/10/31 21:55:54 davem Exp $
- *
  *     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
@@ -121,6 +119,7 @@ static void ipv6_regen_rndid(unsigned long data);
 static int desync_factor = MAX_DESYNC_FACTOR * HZ;
 #endif
 
+static int ipv6_generate_eui64(u8 *eui, struct net_device *dev);
 static int ipv6_count_addresses(struct inet6_dev *idev);
 
 /*
@@ -185,6 +184,8 @@ struct ipv6_devconf ipv6_devconf __read_mostly = {
 #endif
        .proxy_ndp              = 0,
        .accept_source_route    = 0,    /* we do not accept RH0 by default. */
+       .disable_ipv6           = 0,
+       .accept_dad             = 1,
 };
 
 static struct ipv6_devconf ipv6_devconf_dflt __read_mostly = {
@@ -217,6 +218,8 @@ static struct ipv6_devconf ipv6_devconf_dflt __read_mostly = {
 #endif
        .proxy_ndp              = 0,
        .accept_source_route    = 0,    /* we do not accept RH0 by default. */
+       .disable_ipv6           = 0,
+       .accept_dad             = 1,
 };
 
 /* IPv6 Wildcard Address and Loopback Address defined by RFC2553 */
@@ -231,6 +234,12 @@ static inline int addrconf_qdisc_ok(struct net_device *dev)
        return (dev->qdisc != &noop_qdisc);
 }
 
+/* Check if a route is valid prefix route */
+static inline int addrconf_is_prefix_route(const struct rt6_info *rt)
+{
+       return ((rt->rt6i_flags & (RTF_GATEWAY | RTF_DEFAULT)) == 0);
+}
+
 static void addrconf_del_timer(struct inet6_ifaddr *ifp)
 {
        if (del_timer(&ifp->timer))
@@ -344,6 +353,8 @@ static struct inet6_dev * ipv6_add_dev(struct net_device *dev)
                kfree(ndev);
                return NULL;
        }
+       if (ndev->cnf.forwarding)
+               dev_disable_lro(dev);
        /* We refer to the device */
        dev_hold(dev);
 
@@ -372,6 +383,9 @@ static struct inet6_dev * ipv6_add_dev(struct net_device *dev)
         */
        in6_dev_hold(ndev);
 
+       if (dev->flags & (IFF_NOARP | IFF_LOOPBACK))
+               ndev->cnf.accept_dad = -1;
+
 #if defined(CONFIG_IPV6_SIT) || defined(CONFIG_IPV6_SIT_MODULE)
        if (dev->type == ARPHRD_SIT && (dev->priv_flags & IFF_ISATAP)) {
                printk(KERN_INFO
@@ -438,6 +452,8 @@ static void dev_forward_change(struct inet6_dev *idev)
        if (!idev)
                return;
        dev = idev->dev;
+       if (idev->cnf.forwarding)
+               dev_disable_lro(dev);
        if (dev && (dev->flags & IFF_MULTICAST)) {
                if (idev->cnf.forwarding)
                        ipv6_dev_mc_inc(dev, &in6addr_linklocal_allrouters);
@@ -483,12 +499,14 @@ static void addrconf_fixup_forwarding(struct ctl_table *table, int *p, int old)
        if (p == &net->ipv6.devconf_dflt->forwarding)
                return;
 
+       rtnl_lock();
        if (p == &net->ipv6.devconf_all->forwarding) {
                __s32 newf = net->ipv6.devconf_all->forwarding;
                net->ipv6.devconf_dflt->forwarding = newf;
                addrconf_forward_change(net, newf);
        } else if ((!*p) ^ (!old))
                dev_forward_change((struct inet6_dev *)table->extra1);
+       rtnl_unlock();
 
        if (*p)
                rt6_purge_dflt_routers(net);
@@ -568,6 +586,13 @@ ipv6_add_addr(struct inet6_dev *idev, const struct in6_addr *addr, int pfxlen,
        struct rt6_info *rt;
        int hash;
        int err = 0;
+       int addr_type = ipv6_addr_type(addr);
+
+       if (addr_type == IPV6_ADDR_ANY ||
+           addr_type & IPV6_ADDR_MULTICAST ||
+           (!(idev->dev->flags & IFF_LOOPBACK) &&
+            addr_type & IPV6_ADDR_LOOPBACK))
+               return ERR_PTR(-EADDRNOTAVAIL);
 
        rcu_read_lock_bh();
        if (idev->dead) {
@@ -731,8 +756,13 @@ static void ipv6_del_addr(struct inet6_ifaddr *ifp)
                                                onlink = -1;
 
                                        spin_lock(&ifa->lock);
-                                       lifetime = min_t(unsigned long,
-                                                        ifa->valid_lft, 0x7fffffffUL/HZ);
+
+                                       lifetime = addrconf_timeout_fixup(ifa->valid_lft, HZ);
+                                       /*
+                                        * Note: Because this address is
+                                        * not permanent, lifetime <
+                                        * LONG_MAX / HZ here.
+                                        */
                                        if (time_before(expires,
                                                        ifa->tstamp + lifetime * HZ))
                                                expires = ifa->tstamp + lifetime * HZ;
@@ -744,12 +774,12 @@ static void ipv6_del_addr(struct inet6_ifaddr *ifp)
        }
        write_unlock_bh(&idev->lock);
 
+       addrconf_del_timer(ifp);
+
        ipv6_ifa_notify(RTM_DELADDR, ifp);
 
        atomic_notifier_call_chain(&inet6addr_chain, NETDEV_DOWN, ifp);
 
-       addrconf_del_timer(ifp);
-
        /*
         * Purge or update corresponding prefix
         *
@@ -772,7 +802,7 @@ static void ipv6_del_addr(struct inet6_ifaddr *ifp)
                ipv6_addr_prefix(&prefix, &ifp->addr, ifp->prefix_len);
                rt = rt6_lookup(net, &prefix, NULL, ifp->idev->dev->ifindex, 1);
 
-               if (rt && ((rt->rt6i_flags & (RTF_GATEWAY | RTF_DEFAULT)) == 0)) {
+               if (rt && addrconf_is_prefix_route(rt)) {
                        if (onlink == 0) {
                                ip6_del_rt(rt);
                                rt = NULL;
@@ -953,7 +983,8 @@ static inline int ipv6_saddr_preferred(int type)
        return 0;
 }
 
-static int ipv6_get_saddr_eval(struct ipv6_saddr_score *score,
+static int ipv6_get_saddr_eval(struct net *net,
+                              struct ipv6_saddr_score *score,
                               struct ipv6_saddr_dst *dst,
                               int i)
 {
@@ -1032,7 +1063,8 @@ static int ipv6_get_saddr_eval(struct ipv6_saddr_score *score,
                break;
        case IPV6_SADDR_RULE_LABEL:
                /* Rule 6: Prefer matching label */
-               ret = ipv6_addr_label(&score->ifa->addr, score->addr_type,
+               ret = ipv6_addr_label(net,
+                                     &score->ifa->addr, score->addr_type,
                                      score->ifa->idev->dev->ifindex) == dst->label;
                break;
 #ifdef CONFIG_IPV6_PRIVACY
@@ -1086,7 +1118,7 @@ int ipv6_dev_get_saddr(struct net_device *dst_dev,
        dst.addr = daddr;
        dst.ifindex = dst_dev ? dst_dev->ifindex : 0;
        dst.scope = __ipv6_addr_src_scope(dst_type);
-       dst.label = ipv6_addr_label(daddr, dst_type, dst.ifindex);
+       dst.label = ipv6_addr_label(net, daddr, dst_type, dst.ifindex);
        dst.prefs = prefs;
 
        hiscore->rule = -1;
@@ -1154,8 +1186,8 @@ int ipv6_dev_get_saddr(struct net_device *dst_dev,
                        for (i = 0; i < IPV6_SADDR_RULE_MAX; i++) {
                                int minihiscore, miniscore;
 
-                               minihiscore = ipv6_get_saddr_eval(hiscore, &dst, i);
-                               miniscore = ipv6_get_saddr_eval(score, &dst, i);
+                               minihiscore = ipv6_get_saddr_eval(net, hiscore, &dst, i);
+                               miniscore = ipv6_get_saddr_eval(net, score, &dst, i);
 
                                if (minihiscore > miniscore) {
                                        if (i == IPV6_SADDR_RULE_SCOPE &&
@@ -1395,6 +1427,20 @@ static void addrconf_dad_stop(struct inet6_ifaddr *ifp)
 
 void addrconf_dad_failure(struct inet6_ifaddr *ifp)
 {
+       struct inet6_dev *idev = ifp->idev;
+       if (idev->cnf.accept_dad > 1 && !idev->cnf.disable_ipv6) {
+               struct in6_addr addr;
+
+               addr.s6_addr32[0] = htonl(0xfe800000);
+               addr.s6_addr32[1] = 0;
+
+               if (!ipv6_generate_eui64(addr.s6_addr + 8, idev->dev) &&
+                   ipv6_addr_equal(&ifp->addr, &addr)) {
+                       /* DAD failed for link-local based on MAC address */
+                       idev->cnf.disable_ipv6 = 1;
+               }
+       }
+
        if (net_ratelimit())
                printk(KERN_INFO "%s: duplicate address detected!\n", ifp->idev->dev->name);
        addrconf_dad_stop(ifp);
@@ -1722,7 +1768,6 @@ void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len)
        __u32 valid_lft;
        __u32 prefered_lft;
        int addr_type;
-       unsigned long rt_expires;
        struct inet6_dev *in6_dev;
 
        pinfo = (struct prefix_info *) opt;
@@ -1764,37 +1809,32 @@ void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len)
         *      2) Configure prefixes with the auto flag set
         */
 
-       if (valid_lft == INFINITY_LIFE_TIME)
-               rt_expires = ~0UL;
-       else if (valid_lft >= 0x7FFFFFFF/HZ) {
+       if (pinfo->onlink) {
+               struct rt6_info *rt;
+               unsigned long rt_expires;
+
                /* Avoid arithmetic overflow. Really, we could
                 * save rt_expires in seconds, likely valid_lft,
                 * but it would require division in fib gc, that it
                 * not good.
                 */
-               rt_expires = 0x7FFFFFFF - (0x7FFFFFFF % HZ);
-       } else
-               rt_expires = valid_lft * HZ;
+               if (HZ > USER_HZ)
+                       rt_expires = addrconf_timeout_fixup(valid_lft, HZ);
+               else
+                       rt_expires = addrconf_timeout_fixup(valid_lft, USER_HZ);
 
-       /*
-        * We convert this (in jiffies) to clock_t later.
-        * Avoid arithmetic overflow there as well.
-        * Overflow can happen only if HZ < USER_HZ.
-        */
-       if (HZ < USER_HZ && ~rt_expires && rt_expires > 0x7FFFFFFF / USER_HZ)
-               rt_expires = 0x7FFFFFFF / USER_HZ;
+               if (addrconf_finite_timeout(rt_expires))
+                       rt_expires *= HZ;
 
-       if (pinfo->onlink) {
-               struct rt6_info *rt;
                rt = rt6_lookup(dev_net(dev), &pinfo->prefix, NULL,
                                dev->ifindex, 1);
 
-               if (rt && ((rt->rt6i_flags & (RTF_GATEWAY | RTF_DEFAULT)) == 0)) {
+               if (rt && addrconf_is_prefix_route(rt)) {
                        /* Autoconf prefix route */
                        if (valid_lft == 0) {
                                ip6_del_rt(rt);
                                rt = NULL;
-                       } else if (~rt_expires) {
+                       } else if (addrconf_finite_timeout(rt_expires)) {
                                /* not infinity */
                                rt->rt6i_expires = jiffies + rt_expires;
                                rt->rt6i_flags |= RTF_EXPIRES;
@@ -1803,9 +1843,9 @@ void addrconf_prefix_rcv(struct net_device *dev, u8 *opt, int len)
                                rt->rt6i_expires = 0;
                        }
                } else if (valid_lft) {
-                       int flags = RTF_ADDRCONF | RTF_PREFIX_RT;
                        clock_t expires = 0;
-                       if (~rt_expires) {
+                       int flags = RTF_ADDRCONF | RTF_PREFIX_RT;
+                       if (addrconf_finite_timeout(rt_expires)) {
                                /* not infinity */
                                flags |= RTF_EXPIRES;
                                expires = jiffies_to_clock_t(rt_expires);
@@ -2027,7 +2067,7 @@ err_exit:
  *     Manual configuration of address on an interface
  */
 static int inet6_addr_add(struct net *net, int ifindex, struct in6_addr *pfx,
-                         int plen, __u8 ifa_flags, __u32 prefered_lft,
+                         unsigned int plen, __u8 ifa_flags, __u32 prefered_lft,
                          __u32 valid_lft)
 {
        struct inet6_ifaddr *ifp;
@@ -2036,9 +2076,13 @@ static int inet6_addr_add(struct net *net, int ifindex, struct in6_addr *pfx,
        int scope;
        u32 flags;
        clock_t expires;
+       unsigned long timeout;
 
        ASSERT_RTNL();
 
+       if (plen > 128)
+               return -EINVAL;
+
        /* check the lifetime */
        if (!valid_lft || prefered_lft > valid_lft)
                return -EINVAL;
@@ -2052,22 +2096,23 @@ static int inet6_addr_add(struct net *net, int ifindex, struct in6_addr *pfx,
 
        scope = ipv6_addr_scope(pfx);
 
-       if (valid_lft == INFINITY_LIFE_TIME) {
-               ifa_flags |= IFA_F_PERMANENT;
-               flags = 0;
-               expires = 0;
-       } else {
-               if (valid_lft >= 0x7FFFFFFF/HZ)
-                       valid_lft = 0x7FFFFFFF/HZ;
+       timeout = addrconf_timeout_fixup(valid_lft, HZ);
+       if (addrconf_finite_timeout(timeout)) {
+               expires = jiffies_to_clock_t(timeout * HZ);
+               valid_lft = timeout;
                flags = RTF_EXPIRES;
-               expires = jiffies_to_clock_t(valid_lft * HZ);
+       } else {
+               expires = 0;
+               flags = 0;
+               ifa_flags |= IFA_F_PERMANENT;
        }
 
-       if (prefered_lft == 0)
-               ifa_flags |= IFA_F_DEPRECATED;
-       else if ((prefered_lft >= 0x7FFFFFFF/HZ) &&
-                (prefered_lft != INFINITY_LIFE_TIME))
-               prefered_lft = 0x7FFFFFFF/HZ;
+       timeout = addrconf_timeout_fixup(prefered_lft, HZ);
+       if (addrconf_finite_timeout(timeout)) {
+               if (timeout == 0)
+                       ifa_flags |= IFA_F_DEPRECATED;
+               prefered_lft = timeout;
+       }
 
        ifp = ipv6_add_addr(idev, pfx, plen, scope, ifa_flags);
 
@@ -2095,12 +2140,15 @@ static int inet6_addr_add(struct net *net, int ifindex, struct in6_addr *pfx,
 }
 
 static int inet6_addr_del(struct net *net, int ifindex, struct in6_addr *pfx,
-                         int plen)
+                         unsigned int plen)
 {
        struct inet6_ifaddr *ifp;
        struct inet6_dev *idev;
        struct net_device *dev;
 
+       if (plen > 128)
+               return -EINVAL;
+
        dev = __dev_get_by_index(net, ifindex);
        if (!dev)
                return -ENODEV;
@@ -2725,6 +2773,7 @@ static void addrconf_dad_start(struct inet6_ifaddr *ifp, u32 flags)
        spin_lock_bh(&ifp->lock);
 
        if (dev->flags&(IFF_NOARP|IFF_LOOPBACK) ||
+           idev->cnf.accept_dad < 1 ||
            !(ifp->flags&IFA_F_TENTATIVE) ||
            ifp->flags & IFA_F_NODAD) {
                ifp->flags &= ~(IFA_F_TENTATIVE|IFA_F_OPTIMISTIC);
@@ -2772,6 +2821,11 @@ static void addrconf_dad_timer(unsigned long data)
                read_unlock_bh(&idev->lock);
                goto out;
        }
+       if (idev->cnf.accept_dad > 1 && idev->cnf.disable_ipv6) {
+               read_unlock_bh(&idev->lock);
+               addrconf_dad_failure(ifp);
+               return;
+       }
        spin_lock_bh(&ifp->lock);
        if (ifp->probes == 0) {
                /*
@@ -3169,26 +3223,28 @@ static int inet6_addr_modify(struct inet6_ifaddr *ifp, u8 ifa_flags,
 {
        u32 flags;
        clock_t expires;
+       unsigned long timeout;
 
        if (!valid_lft || (prefered_lft > valid_lft))
                return -EINVAL;
 
-       if (valid_lft == INFINITY_LIFE_TIME) {
-               ifa_flags |= IFA_F_PERMANENT;
-               flags = 0;
-               expires = 0;
-       } else {
-               if (valid_lft >= 0x7FFFFFFF/HZ)
-                       valid_lft = 0x7FFFFFFF/HZ;
+       timeout = addrconf_timeout_fixup(valid_lft, HZ);
+       if (addrconf_finite_timeout(timeout)) {
+               expires = jiffies_to_clock_t(timeout * HZ);
+               valid_lft = timeout;
                flags = RTF_EXPIRES;
-               expires = jiffies_to_clock_t(valid_lft * HZ);
+       } else {
+               expires = 0;
+               flags = 0;
+               ifa_flags |= IFA_F_PERMANENT;
        }
 
-       if (prefered_lft == 0)
-               ifa_flags |= IFA_F_DEPRECATED;
-       else if ((prefered_lft >= 0x7FFFFFFF/HZ) &&
-                (prefered_lft != INFINITY_LIFE_TIME))
-               prefered_lft = 0x7FFFFFFF/HZ;
+       timeout = addrconf_timeout_fixup(prefered_lft, HZ);
+       if (addrconf_finite_timeout(timeout)) {
+               if (timeout == 0)
+                       ifa_flags |= IFA_F_DEPRECATED;
+               prefered_lft = timeout;
+       }
 
        spin_lock_bh(&ifp->lock);
        ifp->flags = (ifp->flags & ~(IFA_F_DEPRECATED | IFA_F_PERMANENT | IFA_F_NODAD | IFA_F_HOMEADDRESS)) | ifa_flags;
@@ -3629,6 +3685,8 @@ static inline void ipv6_store_devconf(struct ipv6_devconf *cnf,
 #ifdef CONFIG_IPV6_MROUTE
        array[DEVCONF_MC_FORWARDING] = cnf->mc_forwarding;
 #endif
+       array[DEVCONF_DISABLE_IPV6] = cnf->disable_ipv6;
+       array[DEVCONF_ACCEPT_DAD] = cnf->accept_dad;
 }
 
 static inline size_t inet6_if_nlmsg_size(void)
@@ -4187,6 +4245,22 @@ static struct addrconf_sysctl_table
                        .proc_handler   =       &proc_dointvec,
                },
 #endif
+               {
+                       .ctl_name       =       CTL_UNNUMBERED,
+                       .procname       =       "disable_ipv6",
+                       .data           =       &ipv6_devconf.disable_ipv6,
+                       .maxlen         =       sizeof(int),
+                       .mode           =       0644,
+                       .proc_handler   =       &proc_dointvec,
+               },
+               {
+                       .ctl_name       =       CTL_UNNUMBERED,
+                       .procname       =       "accept_dad",
+                       .data           =       &ipv6_devconf.accept_dad,
+                       .maxlen         =       sizeof(int),
+                       .mode           =       0644,
+                       .proc_handler   =       &proc_dointvec,
+               },
                {
                        .ctl_name       =       0,      /* sentinel */
                }