{"id":"CVE-2026-90111","published":"2026-09-17T17:17:03.130","lastModified":"2026-09-18T18:17:42.500","description":"In the Linux kernel, the following vulnerability has been resolved:\n\nip6mr: do not clone dst in ip6mr_cache_report()\n\nIPv6 input attaches a non-refcounted (NOREF) dst to skbs under RCU.\nWhen an ingress multicast packet misses MFC lookup,\nip6mr_cache_unresolved() places the skb onto the unresolved queue,\nescaping the receive-side RCU grace period.\n\nIf the underlying route is deleted and freed, and the MFC queue is later\nresolved with a wrong parent interface, ip6_mr_forward() invokes\nip6mr_cache_report(..., MRT6MSG_WRONGMIF), which executes\ndst_clone(skb_dst(pkt)) on the freed dst entry, triggering a slab\nuse-after-free.\n\nReport packets queued to mroute6_sk (a raw socket) and netlink\nnotifications do not require an attached dst entry.\n\nFix this by:\n1. Removing dst_clone() in ip6mr_cache_report() and ensuring report skbs\n   do not hold a dst.\n2. Dropping skb_dst before queuing unresolved skbs in\n   ip6mr_cache_unresolved(), matching the fact that multicast\n   forwarding resolves outgoing routes anew via ip6_route_output().","cvssScore":7.8,"cvssSeverity":"HIGH","cvssVector":"CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H","cwes":[],"vendors":[],"products":[],"references":[{"url":"https://git.kernel.org/stable/c/235b42b5860189eb8c27c36435ad932cae65a734","tags":[]},{"url":"https://git.kernel.org/stable/c/4a674afaae4136c71943aa3e35d433676fd40e8f","tags":[]}],"exploitRefs":[],"hasPoc":false,"ai":null}