Thread (17 messages) 17 messages, 6 authors, 1d ago

[PATCH v3 net-next 6/6] ipv6: Batch rt6_uncached_list_flush_dev() in netdev_run_todo().

flat view
WARM1d

From: Kuniyuki Iwashima <kuniyu@google.com>
Date: 2026-10-01 20:48:01
Subsystem: networking [general], networking [ipv4/ipv6], the rest · Maintainers: "David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, David Ahern, Ido Schimmel, Linus Torvalds

IPv6 uncached routes are linked to the global per-cpu lists,
rt6_uncached_list.

When unregistering a netdev, rt6_uncached_list_flush_dev() iterates
over the potentially long lists to find uncached routes tied to
the device and swap it with blackhole_netdev.

Since it is called for every device in dying netns under RTNL,
it adds O(N_dev x (N_cpu + N_route)) costs to any batched device
unregistration.

Let's call it once per batched device unregistration without RTNL.

Because it now runs without RTNL, rt->dst.dev and rt->rt6i_idev->dev
may transition to NETREG_UNREGISTERED at different times.  Thus, we
must check that both devices are blackhole_netdev before unlinking
the route; otherwise, the remaining device reference would not be
flushed when the other device is unregistered.

In addition, since netdev_run_todo() runs even when IPv6 is disabled,
rt6_uncached_list_flush_dev() must check ipv6_mod_enabled() before
touching rt6_uncached_list.

Note also that netdev_run_todo() cannot be called before
rt6_uncached_list is initialised because:

  1. unregister_netdevice_many_notify() panics when
     dev_boot_phase == 1 (net_dev_init() is subsys_initcall()).

  2. the only device registered before fs_initcall_sync() is
     loopback and loopback_net_init() panics in case of failure.

Reported-by: Chris J Arges <redacted>
Closes: https://lore.kernel.org/netdev/20260917-hash-bucket-route-lists-v3-0-30493a37b6eb@cloudflare.com/ (local)
Signed-off-by: Kuniyuki Iwashima <kuniyu@google.com>
---
v3:
  * Unlink only when both devices are blackhole_netdev
  * Clarify netdev_run_todo() does not run before
    rt6_uncached_list is initialised
  * Fix CONFIG_IPV6=n build
---
 include/net/ip6_route.h |  7 +++++++
 net/core/dev.c          |  5 ++++-
 net/ipv6/route.c        | 24 ++++++++++++++++--------
 3 files changed, 27 insertions(+), 9 deletions(-)
diff --git a/include/net/ip6_route.h b/include/net/ip6_route.h
index 0f9b7a260d25..b9de335b639a 100644
--- a/include/net/ip6_route.h
+++ b/include/net/ip6_route.h
@@ -231,6 +231,13 @@ void rt6_multipath_rebalance(struct fib6_info *f6i);
 
 void rt6_uncached_list_add(struct rt6_info *rt);
 void rt6_uncached_list_del(struct rt6_info *rt);
+#ifdef CONFIG_IPV6
+void rt6_uncached_list_flush_dev(struct net_device *dev);
+#else
+static inline void rt6_uncached_list_flush_dev(struct net_device *dev)
+{
+}
+#endif
 
 static inline const struct rt6_info *skb_rt6_info(const struct sk_buff *skb)
 {
diff --git a/net/core/dev.c b/net/core/dev.c
index 7dba0292f052..12a0e7d6567b 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -131,6 +131,7 @@
 #include <net/mpls.h>
 #include <net/route.h>
 #include <linux/ipv6.h>
+#include <net/ip6_route.h>
 #include <linux/in.h>
 #include <linux/jhash.h>
 #include <linux/random.h>
@@ -11839,8 +11840,10 @@ void netdev_run_todo(void)
 		linkwatch_sync_dev(dev);
 	}
 
-	if (!list_empty(&list))
+	if (!list_empty(&list)) {
 		rt_flush_dev(NULL);
+		rt6_uncached_list_flush_dev(NULL);
+	}
 
 	cnt = 0;
 	while (!list_empty(&list)) {
diff --git a/net/ipv6/route.c b/net/ipv6/route.c
index a76869ff87cd..b1dcb5e5da88 100644
--- a/net/ipv6/route.c
+++ b/net/ipv6/route.c
@@ -158,10 +158,16 @@ void rt6_uncached_list_del(struct rt6_info *rt)
 	}
 }
 
-static void rt6_uncached_list_flush_dev(struct net_device *dev)
+void rt6_uncached_list_flush_dev(struct net_device *dev)
 {
 	int cpu;
 
+	if (!ipv6_mod_enabled())
+		return;
+
+	if (dev && dev->dismantle)
+		return;
+
 	for_each_possible_cpu(cpu) {
 		struct uncached_list *ul = per_cpu_ptr(&rt6_uncached_list, cpu);
 		struct rt6_info *rt, *safe;
@@ -173,22 +179,24 @@ static void rt6_uncached_list_flush_dev(struct net_device *dev)
 		list_for_each_entry_safe(rt, safe, &ul->head, dst.rt_uncached) {
 			struct inet6_dev *rt_idev = rt->rt6i_idev;
 			struct net_device *rt_dev = rt->dst.dev;
-			bool handled = false;
 
-			if (rt_idev && rt_idev->dev == dev) {
+			if (rt_idev &&
+			    (dev ? rt_idev->dev == dev :
+			     READ_ONCE(rt_idev->dev->reg_state) == NETREG_UNREGISTERED)) {
 				rt->rt6i_idev = in6_dev_get(blackhole_netdev);
 				in6_dev_put(rt_idev);
-				handled = true;
 			}
 
-			if (rt_dev == dev) {
-				rt->dst.dev = blackhole_netdev;
+			if (dev ? rt_dev == dev :
+			    READ_ONCE(rt_dev->reg_state) == NETREG_UNREGISTERED) {
+				rcu_assign_pointer(rt->dst.dev_rcu, blackhole_netdev);
 				netdev_ref_replace(rt_dev, blackhole_netdev,
 						   &rt->dst.dev_tracker,
 						   GFP_ATOMIC);
-				handled = true;
 			}
-			if (handled)
+
+			if (rt->dst.dev == blackhole_netdev &&
+			    (!rt->rt6i_idev || rt->rt6i_idev->dev == blackhole_netdev))
 				list_del_init(&rt->dst.rt_uncached);
 		}
 		spin_unlock_bh(&ul->lock);
-- 
2.56.0.rc1.315.gc6ed9934b7-goog
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help