CVE-2025-39682

Quoted from the kernel CNA's own scoring file, word for word. A highlighted metric carries the most severe value CVSS v3.1 defines for it, which is what pushes the score up. source

AV:NAttack vector, network

The bug is triggered entirely by the sequence and content of TLS records arriving from the remote peer on a kTLS-enabled TCP socket (net/tls RX path); no local access is needed. Any kTLS RX consumer — OpenSSL-with-kTLS servers/clients, NFS-over-TLS, SMB/RPC-over-TLS via net/handshake — is remotely reachable over the network.

AC:LAttack complexity, low

The attacker sends three back-to-back records ([DATA][zero-length non-DATA][DATA]) and the corruption occurs deterministically on the victim's ordinary recvmsg() loop, with no race to win and no memory-layout dependency; the kernel selftest added with the fix reproduces it in one shot. TLS 1.2 is zero-copy-capable by default (zc_capable = rx_no_pad || version != TLS_1_3_VERSION), so no unusual kernel config or sockopt is required.

PR:NPrivileges required, none

The attacker is the remote TLS peer, which holds no privileges on the target system — an anonymous client completing a server-authenticated TLS handshake with a kTLS server, or a malicious server that a kTLS client connects to, suffices. This matches the repo's PR:N scoring of the comparable remote-peer net/tls bug CVE-2025-39946.

UI:NUser interaction, none

The victim application only has to run its normal recvmsg() loop on an already-established kTLS socket; no click, mount, file open, or any other victim action is required. The attacker drives the entire trigger from the wire.

S:UScope, unchanged

The corruption is confined to kernel skb/slab state within the same security authority; there is no VM, IOMMU, or sandbox boundary crossed. Standard in-kernel memory corruption is Scope:Unchanged.

C:HConfidentiality impact, high

The strparser anchor is queued on rx_list after tcp_read_done() already released its backing TCP skbs, so the next recvmsg() copies freed kernel heap memory straight into the userspace buffer via skb_copy_datagram_msg(). This is a directly usable, repeatable kernel-memory disclosure primitive, not a bounded pointer leak.

I:HIntegrity impact, high

The live strp->anchor is consume_skb()'d by process_rx_list() while the strparser still uses it, after which incoming attacker-controlled TCP data is written into the freed sk_buff/skb_shared_info (skb_fill_page_desc, tls_strp_load_anchor_with_queue), and socket teardown double-frees the anchor and its frag_list skbs. This yields attacker-controlled write-after-free and freelist corruption suitable for control-flow hijack.

A:HAvailability impact, high

The use-after-free and double-free of the anchor and of already-released TCP receive-queue skbs reliably corrupt slab state, producing oopses/panics and list corruption. Any remote peer can repeat this on every connection, giving a trivial remote kernel-crash DoS.