Commit 0638eb57 authored by David S. Miller's avatar David S. Miller
Browse files

Merge branch 'ipv6-Another-followup-to-the-fib6_info-change' 



David Ahern says:

====================
net/ipv6: Another followup to the fib6_info change

Last one - for this week.

Patches 1, 2 and 7 are more cleanup patches - removing dead code,
moving code from a header to near its single caller, and updating
function name.

Patches 3-5 do some refactoring leading up to patch 6 which fixes
a NULL dereference. I have only managed to trigger a panic once, so
I can not definitively confirm it addresses the problem but it seems
pretty clear that it is a race on removing a 'from' reference on
an rt6_info and another path using that 'from' value to do
cookie checking.
====================

Signed-off-by: default avatarDavid S. Miller <davem@davemloft.net>
parents 1b80f86e 8ae86971

include/net/ip6_fib.h

+12 −29
Original line number Diff line number Diff line
@@ -174,7 +174,7 @@ struct fib6_info { 


struct rt6_info {

	struct dst_entry		dst;

	struct fib6_info		*from;

	struct fib6_info __rcu		*from;



	struct rt6key			rt6i_dst;

	struct rt6key			rt6i_src;
@@ -223,39 +223,17 @@ static inline bool fib6_check_expired(const struct fib6_info *f6i) 
	return false;

}



static inline void rt6_clean_expires(struct rt6_info *rt)

{

	rt->rt6i_flags &= ~RTF_EXPIRES;

	rt->dst.expires = 0;

}



static inline void rt6_set_expires(struct rt6_info *rt, unsigned long expires)

{

	rt->dst.expires = expires;

	rt->rt6i_flags |= RTF_EXPIRES;

}



static inline void rt6_update_expires(struct rt6_info *rt0, int timeout)

{

	if (!(rt0->rt6i_flags & RTF_EXPIRES) && rt0->from)

		rt0->dst.expires = rt0->from->expires;



	dst_set_expires(&rt0->dst, timeout);

	rt0->rt6i_flags |= RTF_EXPIRES;

}



/* Function to safely get fn->sernum for passed in rt

 * and store result in passed in cookie.

 * Return true if we can get cookie safely

 * Return false if not

 */

static inline bool rt6_get_cookie_safe(const struct fib6_info *f6i,

static inline bool fib6_get_cookie_safe(const struct fib6_info *f6i,

					u32 *cookie)

{

	struct fib6_node *fn;

	bool status = false;



	rcu_read_lock();

	fn = rcu_dereference(f6i->fib6_node);



	if (fn) {
@@ -265,17 +243,22 @@ static inline bool rt6_get_cookie_safe(const struct fib6_info *f6i, 
		status = true;

	}



	rcu_read_unlock();

	return status;

}



static inline u32 rt6_get_cookie(const struct rt6_info *rt)

{

	struct fib6_info *from;

	u32 cookie = 0;



	if (rt->rt6i_flags & RTF_PCPU ||

	    (unlikely(!list_empty(&rt->rt6i_uncached)) && rt->from))

		rt6_get_cookie_safe(rt->from, &cookie);

	rcu_read_lock();



	from = rcu_dereference(rt->from);

	if (from && (rt->rt6i_flags & RTF_PCPU ||

	    unlikely(!list_empty(&rt->rt6i_uncached))))

		fib6_get_cookie_safe(from, &cookie);



	rcu_read_unlock();



	return cookie;

}

net/ipv6/ip6_fib.c

+27 −18
Original line number Diff line number Diff line
@@ -860,6 +860,31 @@ insert_above: 
	return ln;

}



static void fib6_drop_pcpu_from(struct fib6_info *f6i,

				const struct fib6_table *table)

{

	int cpu;



	/* release the reference to this fib entry from

	 * all of its cached pcpu routes

	 */

	for_each_possible_cpu(cpu) {

		struct rt6_info **ppcpu_rt;

		struct rt6_info *pcpu_rt;



		ppcpu_rt = per_cpu_ptr(f6i->rt6i_pcpu, cpu);

		pcpu_rt = *ppcpu_rt;

		if (pcpu_rt) {

			struct fib6_info *from;



			from = rcu_dereference_protected(pcpu_rt->from,

					     lockdep_is_held(&table->tb6_lock));

			rcu_assign_pointer(pcpu_rt->from, NULL);

			fib6_info_release(from);

		}

	}

}



static void fib6_purge_rt(struct fib6_info *rt, struct fib6_node *fn,

			  struct net *net)

{
@@ -887,24 +912,8 @@ static void fib6_purge_rt(struct fib6_info *rt, struct fib6_node *fn, 
				    lockdep_is_held(&table->tb6_lock));

		}



		if (rt->rt6i_pcpu) {

			int cpu;



			/* release the reference to this fib entry from

			 * all of its cached pcpu routes

			 */

			for_each_possible_cpu(cpu) {

				struct rt6_info **ppcpu_rt;

				struct rt6_info *pcpu_rt;



				ppcpu_rt = per_cpu_ptr(rt->rt6i_pcpu, cpu);

				pcpu_rt = *ppcpu_rt;

				if (pcpu_rt) {

					fib6_info_release(pcpu_rt->from);

					pcpu_rt->from = NULL;

				}

			}

		}

		if (rt->rt6i_pcpu)

			fib6_drop_pcpu_from(rt, table);

	}

}

net/ipv6/ip6_output.c

+7 −2
Original line number Diff line number Diff line
@@ -962,16 +962,21 @@ static int ip6_dst_lookup_tail(struct net *net, const struct sock *sk, 
	 * that's why we try it again later.

	 */

	if (ipv6_addr_any(&fl6->saddr) && (!*dst || !(*dst)->error)) {

		struct fib6_info *from;

		struct rt6_info *rt;

		bool had_dst = *dst != NULL;



		if (!had_dst)

			*dst = ip6_route_output(net, sk, fl6);

		rt = (*dst)->error ? NULL : (struct rt6_info *)*dst;

		err = ip6_route_get_saddr(net, rt ? rt->from : NULL,

					  &fl6->daddr,



		rcu_read_lock();

		from = rt ? rcu_dereference(rt->from) : NULL;

		err = ip6_route_get_saddr(net, from, &fl6->daddr,

					  sk ? inet6_sk(sk)->srcprefs : 0,

					  &fl6->saddr);

		rcu_read_unlock();



		if (err)

			goto out_err_release;

net/ipv6/route.c

+90 −34
Original line number Diff line number Diff line
@@ -359,7 +359,7 @@ EXPORT_SYMBOL(ip6_dst_alloc); 
static void ip6_dst_destroy(struct dst_entry *dst)

{

	struct rt6_info *rt = (struct rt6_info *)dst;

	struct fib6_info *from = rt->from;

	struct fib6_info *from;

	struct inet6_dev *idev;



	dst_destroy_metrics_generic(dst);
@@ -371,8 +371,11 @@ static void ip6_dst_destroy(struct dst_entry *dst) 
		in6_dev_put(idev);

	}



	rt->from = NULL;

	rcu_read_lock();

	from = rcu_dereference(rt->from);

	rcu_assign_pointer(rt->from, NULL);

	fib6_info_release(from);

	rcu_read_unlock();

}



static void ip6_dst_ifdown(struct dst_entry *dst, struct net_device *dev,
@@ -402,12 +405,16 @@ static bool __rt6_check_expired(const struct rt6_info *rt) 


static bool rt6_check_expired(const struct rt6_info *rt)

{

	struct fib6_info *from;



	from = rcu_dereference(rt->from);



	if (rt->rt6i_flags & RTF_EXPIRES) {

		if (time_after(jiffies, rt->dst.expires))

			return true;

	} else if (rt->from) {

	} else if (from) {

		return rt->dst.obsolete != DST_OBSOLETE_FORCE_CHK ||

			fib6_check_expired(rt->from);

			fib6_check_expired(from);

	}

	return false;

}
@@ -963,7 +970,7 @@ static void rt6_set_from(struct rt6_info *rt, struct fib6_info *from) 
{

	rt->rt6i_flags &= ~RTF_EXPIRES;

	fib6_info_hold(from);

	rt->from = from;

	rcu_assign_pointer(rt->from, from);

	dst_init_metrics(&rt->dst, from->fib6_metrics->metrics, true);

	if (from->fib6_metrics != &dst_default_metrics) {

		rt->dst._metrics |= DST_METRICS_REFCOUNTED;
@@ -1164,10 +1171,8 @@ static struct rt6_info *ip6_rt_cache_alloc(struct fib6_info *ort, 
	 *	Clone the route.

	 */



	rcu_read_lock();

	dev = ip6_rt_get_dev_rcu(ort);

	rt = ip6_dst_alloc(dev_net(dev), dev, 0);

	rcu_read_unlock();

	if (!rt)

		return NULL;
@@ -1855,14 +1860,11 @@ redo_rt6_select: 
		 * the daddr in the skb during the neighbor look-up is different

		 * from the fl6->daddr used to look-up route here.

		 */



		struct rt6_info *uncached_rt;



		fib6_info_hold(f6i);

		rcu_read_unlock();



		uncached_rt = ip6_rt_cache_alloc(f6i, &fl6->daddr, NULL);

		fib6_info_release(f6i);



		rcu_read_unlock();



		if (uncached_rt) {

			/* Uncached_rt's refcnt is taken during ip6_rt_cache_alloc()
@@ -2128,8 +2130,7 @@ static bool fib6_check(struct fib6_info *f6i, u32 cookie) 
{

	u32 rt_cookie = 0;



	if ((f6i && !rt6_get_cookie_safe(f6i, &rt_cookie)) ||

	     rt_cookie != cookie)

	if (!fib6_get_cookie_safe(f6i, &rt_cookie) || rt_cookie != cookie)

		return false;



	if (fib6_check_expired(f6i))
@@ -2138,11 +2139,13 @@ static bool fib6_check(struct fib6_info *f6i, u32 cookie) 
	return true;

}



static struct dst_entry *rt6_check(struct rt6_info *rt, u32 cookie)

static struct dst_entry *rt6_check(struct rt6_info *rt,

				   struct fib6_info *from,

				   u32 cookie)

{

	u32 rt_cookie = 0;



	if ((rt->from && !rt6_get_cookie_safe(rt->from, &rt_cookie)) ||

	if ((from && !fib6_get_cookie_safe(from, &rt_cookie)) ||

	    rt_cookie != cookie)

		return NULL;
@@ -2152,11 +2155,13 @@ static struct dst_entry *rt6_check(struct rt6_info *rt, u32 cookie) 
	return &rt->dst;

}



static struct dst_entry *rt6_dst_from_check(struct rt6_info *rt, u32 cookie)

static struct dst_entry *rt6_dst_from_check(struct rt6_info *rt,

					    struct fib6_info *from,

					    u32 cookie)

{

	if (!__rt6_check_expired(rt) &&

	    rt->dst.obsolete == DST_OBSOLETE_FORCE_CHK &&

	    fib6_check(rt->from, cookie))

	    fib6_check(from, cookie))

		return &rt->dst;

	else

		return NULL;
@@ -2164,20 +2169,30 @@ static struct dst_entry *rt6_dst_from_check(struct rt6_info *rt, u32 cookie) 


static struct dst_entry *ip6_dst_check(struct dst_entry *dst, u32 cookie)

{

	struct dst_entry *dst_ret;

	struct fib6_info *from;

	struct rt6_info *rt;



	rt = (struct rt6_info *) dst;

	rt = container_of(dst, struct rt6_info, dst);



	rcu_read_lock();



	/* All IPV6 dsts are created with ->obsolete set to the value

	 * DST_OBSOLETE_FORCE_CHK which forces validation calls down

	 * into this function always.

	 */



	if (rt->rt6i_flags & RTF_PCPU ||

	    (unlikely(!list_empty(&rt->rt6i_uncached)) && rt->from))

		return rt6_dst_from_check(rt, cookie);

	from = rcu_dereference(rt->from);



	if (from && (rt->rt6i_flags & RTF_PCPU ||

	    unlikely(!list_empty(&rt->rt6i_uncached))))

		dst_ret = rt6_dst_from_check(rt, from, cookie);

	else

		return rt6_check(rt, cookie);

		dst_ret = rt6_check(rt, from, cookie);



	rcu_read_unlock();



	return dst_ret;

}



static struct dst_entry *ip6_negative_advice(struct dst_entry *dst)
@@ -2209,18 +2224,38 @@ static void ip6_link_failure(struct sk_buff *skb) 
		if (rt->rt6i_flags & RTF_CACHE) {

			if (dst_hold_safe(&rt->dst))

				rt6_remove_exception_rt(rt);

		} else if (rt->from) {

		} else {

			struct fib6_info *from;

			struct fib6_node *fn;



			rcu_read_lock();

			fn = rcu_dereference(rt->from->fib6_node);

			from = rcu_dereference(rt->from);

			if (from) {

				fn = rcu_dereference(from->fib6_node);

				if (fn && (rt->rt6i_flags & RTF_DEFAULT))

					fn->fn_sernum = -1;

			}

			rcu_read_unlock();

		}

	}

}



static void rt6_update_expires(struct rt6_info *rt0, int timeout)

{

	if (!(rt0->rt6i_flags & RTF_EXPIRES)) {

		struct fib6_info *from;



		rcu_read_lock();

		from = rcu_dereference(rt0->from);

		if (from)

			rt0->dst.expires = from->expires;

		rcu_read_unlock();

	}



	dst_set_expires(&rt0->dst, timeout);

	rt0->rt6i_flags |= RTF_EXPIRES;

}



static void rt6_do_update_pmtu(struct rt6_info *rt, u32 mtu)

{

	struct net *net = dev_net(rt->dst.dev);
@@ -2232,8 +2267,14 @@ static void rt6_do_update_pmtu(struct rt6_info *rt, u32 mtu) 


static bool rt6_cache_allowed_for_pmtu(const struct rt6_info *rt)

{

	bool from_set;



	rcu_read_lock();

	from_set = !!rcu_dereference(rt->from);

	rcu_read_unlock();



	return !(rt->rt6i_flags & RTF_CACHE) &&

		(rt->rt6i_flags & RTF_PCPU || rt->from);

		(rt->rt6i_flags & RTF_PCPU || from_set);

}



static void __ip6_rt_update_pmtu(struct dst_entry *dst, const struct sock *sk,
@@ -2269,14 +2310,18 @@ static void __ip6_rt_update_pmtu(struct dst_entry *dst, const struct sock *sk, 
		if (rt6->rt6i_flags & RTF_CACHE)

			rt6_update_exception_stamp_rt(rt6);

	} else if (daddr) {

		struct fib6_info *from;

		struct rt6_info *nrt6;



		nrt6 = ip6_rt_cache_alloc(rt6->from, daddr, saddr);

		rcu_read_lock();

		from = rcu_dereference(rt6->from);

		nrt6 = ip6_rt_cache_alloc(from, daddr, saddr);

		if (nrt6) {

			rt6_do_update_pmtu(nrt6, mtu);

			if (rt6_insert_exception(nrt6, rt6->from))

			if (rt6_insert_exception(nrt6, from))

				dst_release_immediate(&nrt6->dst);

		}

		rcu_read_unlock();

	}

}
@@ -3209,6 +3254,7 @@ static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_bu 
	struct ndisc_options ndopts;

	struct inet6_dev *in6_dev;

	struct neighbour *neigh;

	struct fib6_info *from;

	struct rd_msg *msg;

	int optlen, on_link;

	u8 *lladdr;
@@ -3290,7 +3336,10 @@ static void rt6_do_redirect(struct dst_entry *dst, struct sock *sk, struct sk_bu 
				     NEIGH_UPDATE_F_ISROUTER)),

		     NDISC_REDIRECT, &ndopts);



	nrt = ip6_rt_cache_alloc(rt->from, &msg->dest, NULL);

	rcu_read_lock();

	from = rcu_dereference(rt->from);

	nrt = ip6_rt_cache_alloc(from, &msg->dest, NULL);

	rcu_read_unlock();

	if (!nrt)

		goto out;
@@ -4672,6 +4721,7 @@ static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, 
	struct net *net = sock_net(in_skb->sk);

	struct nlattr *tb[RTA_MAX+1];

	int err, iif = 0, oif = 0;

	struct fib6_info *from;

	struct dst_entry *dst;

	struct rt6_info *rt;

	struct sk_buff *skb;
@@ -4768,15 +4818,21 @@ static int inet6_rtm_getroute(struct sk_buff *in_skb, struct nlmsghdr *nlh, 
	}



	skb_dst_set(skb, &rt->dst);



	rcu_read_lock();

	from = rcu_dereference(rt->from);



	if (fibmatch)

		err = rt6_fill_node(net, skb, rt->from, NULL, NULL, NULL, iif,

		err = rt6_fill_node(net, skb, from, NULL, NULL, NULL, iif,

				    RTM_NEWROUTE, NETLINK_CB(in_skb).portid,

				    nlh->nlmsg_seq, 0);

	else

		err = rt6_fill_node(net, skb, rt->from, dst,

				    &fl6.daddr, &fl6.saddr, iif, RTM_NEWROUTE,

		err = rt6_fill_node(net, skb, from, dst, &fl6.daddr,

				    &fl6.saddr, iif, RTM_NEWROUTE,

				    NETLINK_CB(in_skb).portid, nlh->nlmsg_seq,

				    0);

	rcu_read_unlock();



	if (err < 0) {

		kfree_skb(skb);

		goto errout;

CRA Git | Maintained and supported by SUSTech CRA and CCSE