Single branch specified, using that and the previous one:
   remotes/brancher/net-next-hw-2026-01-03--16-00 (v6.19-rc3-208-g8d79429ee12f)
   remotes/brancher/net-next-hw-2026-01-04--00-00 (v6.19-rc3-222-gcc7b6af9bd19)

==== BASE IDENTICAL ====
==== COMMIT DIFF ====
diff --git a/tmp/tmp.cD7CUzLmCn b/tmp/tmp.ETOFHgFx81
index 240c3373eb5b..4ecd53b7462c 100644
--- a/tmp/tmp.cD7CUzLmCn
+++ b/tmp/tmp.ETOFHgFx81
@@ -13,6 +13,20 @@ nipa: config: x86: use periodic HZ tick
 nipa: profile (time) test output
 nipa: config: disable CPU_MITIGATIONS
 nipa: disable random kunit tests
+tcp: accecn: enable AccECN
+tcp: accecn: detect loss ACK w/ AccECN option and add TCP_ACCECN_OPTION_PERSIST
+tcp: accecn: fallback outgoing half link to non-AccECN
+tcp: accecn: unset ECT if receive or send ACE=0 in AccECN negotiaion
+tcp: accecn: retransmit SYN/ACK without AccECN option or non-AccECN SYN/ACK
+tcp: add TCP_SYNACK_RETRANS synack_type
+tcp: accecn: retransmit downgraded SYN in AccECN negotiation
+tcp: accecn: handle unexpected AccECN negotiation feedback
+tcp: disable RFC3168 fallback identifier for CC modules
+tcp: ECT_1_NEGOTIATION and NEEDS_ACCECN identifiers
+selftests/net: gro: add self-test for TCP CWR flag
+gro: flushing when CWR is set negatively affects AccECN
+tcp: try to avoid safer when ACKs are thinned
+net: dlink: mask rx_coalesce/rx_timeout before writing RxDMAIntCtrl
 udp: add drop count for packets in udp_prod_queue
 net: mana: Implement ndo_tx_timeout and serialize queue resets per port.
 net: mediatek: add null pointer check for hardware offloading