Netdev List
 help / color / mirror / Atom feed
* [PATCH net v3 0/2] bonding: keep non-Ethernet slaves out of 802.3ad bonds
@ 2026-11-22 14:59 Kyuwon Cho
  2026-11-22 14:59 ` [PATCH net v3 1/2] bonding: reject a non-Ethernet 802.3ad slave even when the type matches Kyuwon Cho
  2026-11-22 14:59 ` [PATCH net v3 2/2] bonding: don't let an 802.3ad slave change its type Kyuwon Cho
  0 siblings, 2 replies; 3+ messages in thread
From: Kyuwon Cho @ 2026-11-22 14:59 UTC (permalink / raw)
  To: netdev
  Cc: Jay Vosburgh, Andrew Lunn, David S. Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Hangbin Liu, linux-kernel

bond_open() adds the 8-byte lacpdu_mcast_addr to every slave of an
802.3ad bond, so a slave whose addr_len is 16, such as an ip6tnl device,
makes dev_mc_add() read past it. Patch 1 rejects such a slave at enslave
time in every case; patch 2 stops an enslaved bond from changing its
type afterwards.

The issue was found by an LLM-guided test generator that we are
developing as a research prototype.

v3:
- Add patch 2: as the Sashiko review of v1 pointed out, a bond enslaved
  to an 802.3ad bond could still change its type.
- Patch 1 is unchanged; its tags are kept.
- Hangbin asked on v2 whether the bond type should be reset when the
  mode is set to 802.3ad; as answered there, the first Ethernet slave
  resets it, so this series leaves it alone.
v2: https://lore.kernel.org/netdev/179505868590.3571070.12381528230330714812@gmail.com/
- Move the check above the type handling, as Jay suggested.
v1: https://lore.kernel.org/netdev/179500849159.1877154.14715369660816392826@gmail.com/

The patches were written with LLM assistance; we reviewed and tested
them. With them, neither the direct nor the nested reproducer triggers
the report. Enslaving Ethernet devices into an 802.3ad bond, a
non-Ethernet device into an active-backup bond, and the type change of
a bond enslaved to a balance-rr bond, as in bond-eth-type-change.sh,
work as before.

Kyuwon Cho (2):
  bonding: reject a non-Ethernet 802.3ad slave even when the type
    matches
  bonding: don't let an 802.3ad slave change its type

 drivers/net/bonding/bond_main.c | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

-- 
2.43.0


^ permalink raw reply	[flat|nested] 3+ messages in thread

* [PATCH net v3 1/2] bonding: reject a non-Ethernet 802.3ad slave even when the type matches
  2026-11-22 14:59 [PATCH net v3 0/2] bonding: keep non-Ethernet slaves out of 802.3ad bonds Kyuwon Cho
@ 2026-11-22 14:59 ` Kyuwon Cho
  2026-11-22 14:59 ` [PATCH net v3 2/2] bonding: don't let an 802.3ad slave change its type Kyuwon Cho
  1 sibling, 0 replies; 3+ messages in thread
From: Kyuwon Cho @ 2026-11-22 14:59 UTC (permalink / raw)
  To: netdev
  Cc: Jay Vosburgh, Andrew Lunn, David S. Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Hangbin Liu, linux-kernel

Commit c84fcb79e5db ("bonding: limit BOND_MODE_8023AD to Ethernet
devices") rejects a non-Ethernet first slave in 802.3ad mode, but only
when enslaving it changes the bond's type. A bond keeps the type of its
last slave after that slave is released, so a bond that once had an
ARPHRD_TUNNEL6 slave can be set to 802.3ad and take another TUNNEL6
device, such as a second bond prepared the same way, without the check.
bond_open() then adds the 8-byte lacpdu_mcast_addr to that slave with
dev_mc_add(), which copies the slave's 16-byte addr_len:

  BUG: KASAN: global-out-of-bounds in __hw_addr_add_ex (net/core/dev_addr_lists.c:102 net/core/dev_addr_lists.c:157)
  Read of size 16 at addr ffffffff878a2940 by task test/114
  Call Trace:
   <TASK>
   __hw_addr_add_ex (net/core/dev_addr_lists.c:102 net/core/dev_addr_lists.c:157)
   dev_mc_add (net/core/dev_addr_lists.c:1036 net/core/dev_addr_lists.c:1054)
   bond_open (drivers/net/bonding/bond_main.c:4405)
   __dev_open (net/core/dev.c:1762)
   __dev_change_flags (net/core/dev.c:10022)
   netif_change_flags (net/core/dev.c:10087)
   do_setlink.isra.0 (net/core/rtnetlink.c:3294)
   rtnl_setlink (net/core/rtnetlink.c:3633)

An unprivileged user can set this up in its own user and network
namespace.

Check that an 802.3ad slave is an Ethernet device before the type
handling, for every slave.

Fixes: c84fcb79e5db ("bonding: limit BOND_MODE_8023AD to Ethernet devices")
Suggested-by: Jay Vosburgh <jv@jvosburgh.net>
Signed-off-by: Kyuwon Cho <kyuwoncho18@gmail.com>
Reviewed-by: Eric Dumazet <edumazet@kernel.org>
---
 drivers/net/bonding/bond_main.c | 13 +++++++------
 1 file changed, 7 insertions(+), 6 deletions(-)

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index 042d7c75548f..bcc12ef34d9a 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -1951,6 +1951,13 @@ int bond_enslave(struct net_device *bond_dev, struct net_device *slave_dev,
 		return -EPERM;
 	}
 
+	if (slave_dev->type != ARPHRD_ETHER &&
+	    BOND_MODE(bond) == BOND_MODE_8023AD) {
+		SLAVE_NL_ERR(bond_dev, slave_dev, extack,
+			     "8023AD mode requires Ethernet devices");
+		return -EINVAL;
+	}
+
 	/* set bonding device ether type by slave - bonding netdevices are
 	 * created with ether_setup, so when the slave type is not ARPHRD_ETHER
 	 * there is a need to override some of the type dependent attribs/funcs.
@@ -1960,12 +1967,6 @@ int bond_enslave(struct net_device *bond_dev, struct net_device *slave_dev,
 	 */
 	if (!bond_has_slaves(bond)) {
 		if (bond_dev->type != slave_dev->type) {
-			if (slave_dev->type != ARPHRD_ETHER &&
-			    BOND_MODE(bond) == BOND_MODE_8023AD) {
-				SLAVE_NL_ERR(bond_dev, slave_dev, extack,
-					     "8023AD mode requires Ethernet devices");
-				return -EINVAL;
-			}
 			slave_dbg(bond_dev, slave_dev, "change device type from %d to %d\n",
 				  bond_dev->type, slave_dev->type);
 
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 3+ messages in thread

* [PATCH net v3 2/2] bonding: don't let an 802.3ad slave change its type
  2026-11-22 14:59 [PATCH net v3 0/2] bonding: keep non-Ethernet slaves out of 802.3ad bonds Kyuwon Cho
  2026-11-22 14:59 ` [PATCH net v3 1/2] bonding: reject a non-Ethernet 802.3ad slave even when the type matches Kyuwon Cho
@ 2026-11-22 14:59 ` Kyuwon Cho
  1 sibling, 0 replies; 3+ messages in thread
From: Kyuwon Cho @ 2026-11-22 14:59 UTC (permalink / raw)
  To: netdev
  Cc: Jay Vosburgh, Andrew Lunn, David S. Miller, Eric Dumazet,
	Jakub Kicinski, Paolo Abeni, Hangbin Liu, linux-kernel

A bond can be a slave of another bond. Enslaving a non-Ethernet device to
an empty inner bond changes the inner bond's type in bond_setup_by_slave()
after NETDEV_PRE_TYPE_CHANGE, and nothing refuses that while the inner
bond is a slave of an 802.3ad bond. The outer bond then adds the 8-byte
lacpdu_mcast_addr to a slave whose addr_len is 16, the same global
out-of-bounds read as when such a slave is enslaved directly:

  BUG: KASAN: global-out-of-bounds in __hw_addr_add_ex (net/core/dev_addr_lists.c:102 net/core/dev_addr_lists.c:157)
  Read of size 16
  Call Trace:
   __hw_addr_add_ex (net/core/dev_addr_lists.c:102 net/core/dev_addr_lists.c:157)
   dev_mc_add (net/core/dev_addr_lists.c:1036 net/core/dev_addr_lists.c:1054)
   bond_open (drivers/net/bonding/bond_main.c:4405)

Refuse NETDEV_PRE_TYPE_CHANGE for a slave of an 802.3ad bond, as team
does for its ports. Other modes still allow it, as
bond-eth-type-change.sh expects.

Fixes: c84fcb79e5db ("bonding: limit BOND_MODE_8023AD to Ethernet devices")
Signed-off-by: Kyuwon Cho <kyuwoncho18@gmail.com>
---
 drivers/net/bonding/bond_main.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/drivers/net/bonding/bond_main.c b/drivers/net/bonding/bond_main.c
index bcc12ef34d9a..1a926b8d28b0 100644
--- a/drivers/net/bonding/bond_main.c
+++ b/drivers/net/bonding/bond_main.c
@@ -3981,6 +3981,10 @@ static int bond_slave_netdev_event(unsigned long event,
 	slave_dbg(bond_dev, slave_dev, "%s called\n", __func__);
 
 	switch (event) {
+	case NETDEV_PRE_TYPE_CHANGE:
+		if (BOND_MODE(bond) == BOND_MODE_8023AD)
+			return NOTIFY_BAD;
+		break;
 	case NETDEV_UNREGISTER:
 		if (bond_dev->type != ARPHRD_ETHER)
 			bond_release_and_destroy(bond_dev, slave_dev);
-- 
2.43.0


^ permalink raw reply related	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2026-11-22 14:59 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-11-22 14:59 [PATCH net v3 0/2] bonding: keep non-Ethernet slaves out of 802.3ad bonds Kyuwon Cho
2026-11-22 14:59 ` [PATCH net v3 1/2] bonding: reject a non-Ethernet 802.3ad slave even when the type matches Kyuwon Cho
2026-11-22 14:59 ` [PATCH net v3 2/2] bonding: don't let an 802.3ad slave change its type Kyuwon Cho

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox