6.6-stable review patch. If anyone has any objections, please let me know. ------------------ From: Willem de Bruijn [ Upstream commit fc6d80eb504458d6416b75a94188b268c95c6533 ] tcp_write_collapse_fence() sets TCP_SKB_CB(skb)->eor = 1 on tcp_write_queue_tail(sk) to prevent skbs queued after a switch to device encryption from being collapsed into earlier skbs. The fence is a no-op if all earlier data has already been transmitted when the switch happens: sk->sk_write_queue is empty. The not yet acknowledged earlier skbs wait in sk->tcp_rtx_queue with eor 0. On a subsequent retransmit or SACK shift, tcp_retrans_try_collapse() or tcp_shift_skb_data() can then merge an skb queued after the switch into one queued before it. Both users of the fence are affected: - psp: devices only encrypt skbs with skb->decrypted set. The merged skb keeps decrypted = 0 from the earlier skb, so merged data sent after psp_sock_assoc_set_tx() is retransmitted in cleartext. - tls device offload: the merged skb straddles the start marker set in tls_set_device_offload(). The software fallback (fill_sg_in() returns -EINVAL) and the mlx5, nfp and funeth drivers cannot handle such an skb and drop it. Every retransmit rebuilds the same skb, so the connection stalls. Fix this in two places, for defense in depth: 1. Fall back to tcp_rtx_queue_tail(sk) in tcp_write_collapse_fence() when tcp_write_queue_tail(sk) is NULL. 2. Check !skb_cmp_decrypted(to, from) in tcp_skb_can_collapse(), as tcp_skb_can_collapse_rx() does on receive. skb_shift(), which both collapse paths call, already has a DEBUG_NET_WARN_ON_ONCE() for this condition. Fixes: e8f69799810c ("net/tls: Add generic NIC offload infrastructure") Cc: stable@vger.kernel.org Signed-off-by: Willem de Bruijn Reviewed-by: Eric Dumazet Reviewed-by: Daniel Zahka Link: https://patch.msgid.link/20260924154427.953800-1-willemdebruijn.kernel@gmail.com Signed-off-by: Jakub Kicinski [ Applied tcp_write_collapse_fence() changes directly to the existing TLS code blocks because the helper is absent. ] Signed-off-by: Sasha Levin --- include/net/tcp.h | 1 + net/tls/tls_device.c | 4 ++-- 2 files changed, 3 insertions(+), 2 deletions(-) diff --git a/include/net/tcp.h b/include/net/tcp.h index c75d12e5156db..3fe0c1d928407 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -985,6 +985,7 @@ static inline bool tcp_skb_can_collapse(const struct sk_buff *to, { return likely(tcp_skb_can_collapse_to(to) && mptcp_skb_can_collapse(to, from) && + !skb_cmp_decrypted(to, from) && skb_pure_zcopy_same(to, from)); } diff --git a/net/tls/tls_device.c b/net/tls/tls_device.c index 34d321ff11980..bb7057aa828e1 100644 --- a/net/tls/tls_device.c +++ b/net/tls/tls_device.c @@ -239,7 +239,7 @@ static void tls_device_resync_tx(struct sock *sk, struct tls_context *tls_ctx, int err = 0; u8 *rcd_sn; - skb = tcp_write_queue_tail(sk); + skb = tcp_write_queue_tail(sk) ?: tcp_rtx_queue_tail(sk); if (skb) TCP_SKB_CB(skb)->eor = 1; @@ -1157,7 +1157,7 @@ int tls_set_device_offload(struct sock *sk, struct tls_context *ctx) * SKBs where only part of the payload needs to be encrypted. * So mark the last skb in the write queue as end of record. */ - skb = tcp_write_queue_tail(sk); + skb = tcp_write_queue_tail(sk) ?: tcp_rtx_queue_tail(sk); if (skb) TCP_SKB_CB(skb)->eor = 1; -- 2.53.0