This looks like a case where some conformance test is testing a corner case in the standard where the Linux TCP stack is not following the standard for valid reasons. Linux behavior of silently dropping the packet would reduce DoS changes and information leakage for MiTM attacks.
Begin forwarded message: Date: Fri, 08 Jan 2021 08:17:40 +0000 From: bugzilla-dae...@bugzilla.kernel.org To: step...@networkplumber.org Subject: [Bug 211085] New: No response from TCP connection in ESTALISHED state if sending data segment with unacceptable ACK https://bugzilla.kernel.org/show_bug.cgi?id=211085 Bug ID: 211085 Summary: No response from TCP connection in ESTALISHED state if sending data segment with unacceptable ACK Product: Networking Version: 2.5 Kernel Version: 5.4.24 and upstream Hardware: All OS: Linux Tree: Mainline Status: NEW Severity: normal Priority: P1 Component: IPV4 Assignee: step...@networkplumber.org Reporter: alex.thres...@googlemail.com Regression: No We're currently failing to pass the TCP_UNACCEPTABLE_04 test from TCP/IP tests of TC8 ECU and Network Test specification (Open Alliance). The Test specification is freely available on the Internet. TCP_UNACCEPTABLE_04 validates the behavior of RFC793 'Chapter 3.4. Establishing a connection / Reset Generation (3)' (Page 36) and 'Chapter 9 - Event Processing / SEGMENT ARRIVES / Otherwise / Check ACK field' (Page 71) which pointed out that sending an unacceptable ACK in ESTALISHED state for example ACK acks a data segment not yet sent should be answered with an ACK (from RFC793 - '...must elicit only an empty acknowledgment segment containing the current send-sequence number and an acknowledgment indicating the next sequence number expected to be received...'). After that the segment needs to be discarded. Unfortunately the segment is only discarded and not confirmed by ACK. I did some research in the TCP/IP stack and found the code which is responsible for discarding the ACK. It's in 'net/ipv4/tcp_input.c' in function tcp_ack currently on line 3727 of upstream kernel. >>/* If the ack includes data we haven't sent yet, discard >> * this segment (RFC793 Section 3.9). >> */ >>if (after(ack, tp->snd_nxt)) >> return -1; Validation was done on DUT target with Kernel 5.4.24. Let me if you need further informations. -- You may reply to this email to add a comment. You are receiving this mail because: You are the assignee for the bug.
