[PATCH net v2] ipv6: fix fib6 walker UAF on NLM_F_REPLACE

[PATCH net v2] ipv6: fix fib6 walker UAF on NLM_F_REPLACE

[Cen Zhang (Microsoft)]

fib6_walker uses w->leaf to remember the next fib6_info to dump, so an
RTM_GETROUTE dump can continue there on the next recvmsg().  This
assumes that whoever removes a route from the tree also updates the
walkers pointing at it.  fib6_del_route() does: it moves w->leaf to
the next route.  The NLM_F_REPLACE branch of fib6_add_rt2node() does
not: it removes the old route, fib6_add() frees it, and w->leaf keeps
pointing at the freed object.

The fn_sernum version check in fib6_dump_table() normally protects
against resuming on a stale w->leaf, but it can be bypassed:
ip6_link_failure() resets fn_sernum to -1 when the default route
fails, which an unprivileged user in a user+net namespace can cause:

  1) link failure            fn_sernum = -1
  2) dump pauses on route R
  3) replace R               fn_sernum changed, R freed
  4) link failure            fn_sernum = -1
  5) dump resumes            -1 == -1, reads freed R  <- UAF

  BUG: KASAN: slab-use-after-free in rt6_fill_node+0x1c22/0x2cd0
  Read of size 4 at addr ff11000011fa0064 by task poc/101
   rt6_fill_node <- rt6_dump_route

Fix by adjusting walkers separately for replacement and deletion.
For replacement, add fib6_walkers_replace_leaf() points w->leaf
at the new route. For deletion and ECMP sibling removal,
fib6_walkers_skip_rt() calls fib6_walkers_replace_leaf() with the
next route. Reset w->skip_in_node when updating w->leaf, and set
w->state to FWS_U when no next route remains.

Fixes: 4a287eba2de3 ("IPv6 routing, NLM_F_* flag support: REPLACE and EXCL flags support, warn about missing CREATE flag")
Cc: AutonomousCodeSecurity@microsoft.com
Signed-off-by: Cen Zhang (Microsoft) <cenzhang@linux.microsoft.com>
---
Developed with LLM assistance.

v2:
 - Move replacement walkers to the new route instead of skipping it
   (Sashiko).
 - Add fib6_walkers_replace_leaf() for the shared adjustment; make
   fib6_walkers_skip_rt() a deletion wrapper around it.
 - Reset w->skip_in_node when updating w->leaf (Sashiko).
 - Rebase on net.
v1: https://lore.kernel.org/netdev/20261009225142.47005-1-cenzhang@linux.microsoft.com/

 net/ipv6/ip6_fib.c | 50 ++++++++++++++++++++++++++++++++++------------
 1 file changed, 37 insertions(+), 13 deletions(-)

diff --git a/net/ipv6/ip6_fib.c b/net/ipv6/ip6_fib.c
index 0dc6ebee6e56..fb19670521c6 100644
--- a/net/ipv6/ip6_fib.c
+++ b/net/ipv6/ip6_fib.c
@@ -89,6 +89,40 @@ static void fib6_walker_unlink(struct net *net, struct fib6_walker *w)
 	write_unlock_bh(&net->ipv6.fib6_walker_lock);
 }
 
+/* @old is leaving the tree.  Walkers resting on it continue from @new, or
+ * climb to the parent node when there is no @new.  Called with tb6_lock
+ * held.
+ */
+static void fib6_walkers_replace_leaf(struct net *net, struct fib6_info *old,
+				      struct fib6_info *new)
+{
+	struct fib6_walker *w;
+
+	read_lock(&net->ipv6.fib6_walker_lock);
+	FOR_WALKERS(net, w) {
+		if (w->state == FWS_C && w->leaf == old) {
+			pr_debug("walker %p adjusted by unlink of %p\n", w, old);
+			w->leaf = new;
+			w->skip_in_node = 0;
+			if (!new)
+				w->state = FWS_U;
+		}
+	}
+	read_unlock(&net->ipv6.fib6_walker_lock);
+}
+
+/* @rt is removed from its node: walkers resting on it continue from the
+ * next route of the node.
+ */
+static void fib6_walkers_skip_rt(struct net *net, struct fib6_info *rt)
+{
+	struct fib6_info *next;
+
+	next = rcu_dereference_protected(rt->fib6_next,
+					 lockdep_is_held(&rt->fib6_table->tb6_lock));
+	fib6_walkers_replace_leaf(net, rt, next);
+}
+
 static int fib6_new_sernum(struct net *net)
 {
 	int new, old = atomic_read(&net->ipv6.fib6_sernum);
@@ -1316,6 +1350,7 @@ static int fib6_add_rt2node(struct fib6_node *fn, struct fib6_info *rt,
 		rcu_assign_pointer(rt->fib6_node, fn);
 		rt->fib6_next = iter->fib6_next;
 		rcu_assign_pointer(*ins, rt);
+		fib6_walkers_replace_leaf(info->nl_net, iter, rt);
 		if (!info->skip_notify)
 			inet6_rt_notify(RTM_NEWROUTE, rt, info, NLM_F_REPLACE);
 		if (!(fn->fn_flags & RTN_RTINFO)) {
@@ -1337,6 +1372,7 @@ static int fib6_add_rt2node(struct fib6_node *fn, struct fib6_info *rt,
 				if (iter->fib6_metric > rt->fib6_metric)
 					break;
 				if (rt6_qualify_for_ecmp(iter)) {
+					fib6_walkers_skip_rt(info->nl_net, iter);
 					*ins = iter->fib6_next;
 					iter->fib6_node = NULL;
 					list_add(&iter->purge_link, purge_list);
@@ -1971,7 +2007,6 @@ static void fib6_del_route(struct fib6_table *table, struct fib6_node *fn,
 			   enum rt_del_reason del_reason)
 {
 	struct fib6_info *leaf, *replace_rt = NULL;
-	struct fib6_walker *w;
 	struct fib6_info *rt = rcu_dereference_protected(*rtp,
 				    lockdep_is_held(&table->tb6_lock));
 	struct net *net = info->nl_net;
@@ -2022,18 +2057,7 @@ static void fib6_del_route(struct fib6_table *table, struct fib6_node *fn,
 		rt6_multipath_rebalance(next_sibling);
 	}
 
-	/* Adjust walkers */
-	read_lock(&net->ipv6.fib6_walker_lock);
-	FOR_WALKERS(net, w) {
-		if (w->state == FWS_C && w->leaf == rt) {
-			pr_debug("walker %p adjusted by delroute\n", w);
-			w->leaf = rcu_dereference_protected(rt->fib6_next,
-					    lockdep_is_held(&table->tb6_lock));
-			if (!w->leaf)
-				w->state = FWS_U;
-		}
-	}
-	read_unlock(&net->ipv6.fib6_walker_lock);
+	fib6_walkers_skip_rt(net, rt);
 
 	/* If it was last route, call fib6_repair_tree() to:
 	 * 1. For root node, put back null_entry as how the table was created.
-- 
2.55.0