CVE-2026-98122 PUBLISHED

vxlan: mdb: Fix use-after-free in vxlan_mdb_remote_src_del()

Assigner: Linux
Reserved: 25.09.2026 Published: 25.09.2026 Updated: 25.09.2026

In the Linux kernel, the following vulnerability has been resolved:

vxlan: mdb: Fix use-after-free in vxlan_mdb_remote_src_del()

vxlan_mdb_is_valid_source(), which validates MDBE_ATTR_SOURCE and every MDBE_ATTR_SRC_LIST member, accepts the all-zeros address.

A source list is only accepted on a (, G) entry, whose source is the all-zeros address, and for each member of the list an (S, G) entry is derived from it by substituting the source. Entries are keyed by a plain memcmp() of struct vxlan_mdb_entry_key, so if MDBE_ATTR_SOURCE is present and holds the all-zeros address and the source list holds it as well, the derived (S, G) key is byte-identical to the (, G) key and resolves to the same entry. Omitting MDBE_ATTR_SOURCE is not equivalent, as the key is then left with a zero address family.

vxlan_mdb_remote_src_del() removes the forwarding entry of a source before freeing the source entry:

<pre>vxlan_mdb_remote_src_fwd_del(vxlan, group, remote, &ent->addr); vxlan_mdb_remote_src_entry_del(ent); </pre>

With the keys aliased, the first call deletes the remote of the entry that owns 'ent' instead of a separate (S, G) entry, and frees 'ent'. The second call then runs on the freed entry, and its hlist_del() reads ->pprev and ->next out of it and writes through them.

Adding the (*, G) entry with NLM_F_REPLACE and no source list marks the all-zeros source for deletion and reaches this from the sweep at the end of vxlan_mdb_remote_srcs_replace().

BUG: KASAN: slab-use-after-free in __vxlan_mdb_add+0x1cd/0xd70 Read of size 8 at addr ffff888102852500 by task poc/84 __vxlan_mdb_add+0x1cd/0xd70 vxlan_mdb_add+0xc0/0x140 rtnl_mdb_add+0x157/0x2a0 rtnetlink_rcv_msg+0x207/0x5a0 Allocated by task 84: __kmalloc_cache_noprof+0x153/0x360 vxlan_mdb_remote_srcs_add+0x2eb/0x440 __vxlan_mdb_add+0x803/0xd70 Freed by task 84: kfree+0x14c/0x3b0 vxlan_mdb_remote_del+0x129/0x1a0 __vxlan_mdb_del+0x4f/0xe0 vxlan_mdb_remote_src_fwd_del.isra.0+0x162/0x1b0 __vxlan_mdb_add+0x1c5/0xd70

The MDB operations are netns-scoped, so an unprivileged user can perform them in a new user and network namespace.

Reject the all-zeros address in vxlan_mdb_is_valid_source(), which covers both call sites. A (*, G) entry is expressed by omitting the source, so nothing legitimate is refused.

Discovered by XBOW, triaged by Baul Lee baul.lee@xbow.com

Metrics

CVSS Vector: CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H
CVSS Score: 7.8

AV:L - The malformed input is an RTM_NEWMDB rtnetlink message sent from a local socket (rtnl_mdb_add -> vxlan_mdb_add -> __vxlan_mdb_add -> vxlan_mdb_remote_srcs_replace -> vxlan_mdb_remote_src_del); no packet received from the network reaches the MDB control path. AC:L - It is deterministic, with no race: add a (*, G) entry with MDBE_ATTR_SOURCE 0.0.0.0 and 0.0.0.0 in MDBE_ATTR_SRC_LIST, then re-add it with NLM_F_REPLACE and no source list. The attacker creates every piece of the required state. PR:L - The only check is netlink_net_capable(CAP_NET_ADMIN) in rtnetlink_rcv_msg, which is namespace-scoped. An unprivileged user can get it with unshare -Urn, create their own vxlan device and send RTM_NEWMDB to it. UI:N - The attacker's own netlink messages, on a vxlan device they created in their own namespace, trigger the bug; no other user has to do anything. S:U - Corrupting kernel heap memory gives kernel-level impact within the same security authority, which is ordinary local privilege escalation with no VM or hardware boundary crossed. C:H - A heap spray can reclaim the freed vxlan_mdb_src_entry before vxlan_mdb_remote_src_entry_del reads it, so the attacker controls the pointers the kernel then uses. This is a slab use-after-free that can be built into a kernel memory disclosure. I:H - hlist_del() in vxlan_mdb_remote_src_entry_del reads ->next and ->pprev from the freed entry and writes through them. Once the slot is reclaimed with controlled data, that is an attacker-directed pointer write (an unlink write primitive). A:H - Unlinking a freed hlist node corrupts kernel memory; KASAN reports it in __vxlan_mdb_add, and without KASAN the corrupted list pointers can oops or panic the kernel. Any user who can open a user namespace can trigger it repeatedly.

Product Status

Vendor Linux
Product Linux
Versions Default: unaffected
  • affected from a3a48de5eade770e911d35291217bdd69ce04ef1 to e12c903fd13c6261b98ca18d8f70ac26e739e2f8 (excl.)
  • affected from a3a48de5eade770e911d35291217bdd69ce04ef1 to 74e2a56c82209b0335b53e25f19f9b0590e2483d (excl.)
  • affected from a3a48de5eade770e911d35291217bdd69ce04ef1 to 71203a41d6fa2fa0ea2f3a7541987958bd3694fd (excl.)
  • affected from a3a48de5eade770e911d35291217bdd69ce04ef1 to 4aa61c88b4e292e10abdfd791334b8272108d68a (excl.)
Vendor Linux
Product Linux
Versions Default: affected
  • Version 6.4 is affected
  • unaffected from 0 to 6.4 (excl.)
  • unaffected from 6.12.111 to 6.12.* (incl.)
  • unaffected from 6.18.53 to 6.18.* (incl.)
  • unaffected from 7.2.7 to 7.2.* (incl.)
  • unaffected from 7.3-rc2 to * (incl.)

References