[PATCH net v2] ipv6: fix fib6 walker UAF on NLM_F_REPLACE
DORMANTno replies
From: Cen Zhang (Microsoft) <hidden>
Date: 2026-10-28 16:00:35
Also in:
lkml, stable
Subsystem:
networking [general], networking [ipv4/ipv6], the rest · Maintainers:
"David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, David Ahern, Ido Schimmel, Linus Torvalds
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: <redacted>
Signed-off-by: Cen Zhang (Microsoft) <redacted>
---
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/ (local)
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