[Bug 250499] [tcp] Should we reject the packet with timestamp if no timestamp in SYN and SYN_ACK?

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=250499

--- Comment #6 from Michael Tuexen <tuexen at freebsd.org> ---
(In reply to Roman Bogorodskiy from comment #5)
Hi Roman,

I introduced a regression, which was fixed in
https://cgit.FreeBSD.org/src/commit/?id=cc3c34859eab1b317d0f38731355b53f7d978c97
and MFCed to stable/12 in
https://cgit.FreeBSD.org/src/commit/?id=d05d908d6d3c85479c84c707f931148439ae826b.
This was affecting the handling of received TCP RST segments.
Does this fix your issue?

In bug #252449 it was reported that there are broken TCP stacks, which
negotiate TCP timestamp support but then don't behave as required.
To deal with such broken implementations I added a sysctl variable in
https://cgit.FreeBSD.org/src/commit/?id=d2b3ceddccac60b563f642898e3a314647666a10
, which was MFCed to stable/12 in
https://cgit.FreeBSD.org/src/commit/?id=e82353f84c58da9a5c38bd471a09936c16a5b6ea.
In this case you can set the sysctl variable net.inet.tcp.tolerate_missing_ts
to 1. I would be interested to now which stack is broken and it would be great
if you could provide a .pcap file of the failing TCP connection.
Does this fix your issue?

-- 
You are receiving this mail because:
You are on the CC list for the bug.