MSS bugs ?

[LatinSuD]

I think that dynamic TCP MSS rules for PPPoE interfaces could be enhanced in 2 ways:

• The rule that matches by “out-interface” should be in “postrouting” instead of forward.
  Locally generated TCP packets seem not to have correct MSS. Those don’t go through “forward”, but they could be addressed at “postrouting”.

• I’m not sure about this point, but i think that TCP MSS mangle rules should match by existing MSS.
  For example, if i set MSS to 1440 the rule should have a “tcp-mss=1441-65535” to prevent from increasing mss of a packet that had, for example, mss of 1420.

Thank you

[sten]

@LatinSuD

• The rule that matches by “out-interface” should be in “postrouting” instead of forward.
  Locally generated TCP packets seem not to have correct MSS. Those don’t go through “forward”, but they could be addressed at “postrouting”.

When you think about it, the local router already knows the correct MTU size and will set it’s initial MSS accordingly. But yes, sometimes it would have been very helpful to at least have the option.

@LatinSuD

• I’m not sure about this point, but i think that TCP MSS mangle rules should match by existing MSS.
  For example, if i set MSS to 1440 the rule should have a “tcp-mss=1441-65535” to prevent from increasing mss of a packet that had, for example, mss of 1420.

Yes, this is a bug. Dunno when they are going to fix this. I’ve reported it to support@ and i think you should too. The current behavior is definitely broken big time.

[LatinSuD]

I’m a little confused, i have a MT 2.9.23 that uses wrong MSS for locally generated packets, but a MT 2.9.24 does work. I’ll try upgrade and tell you later.

Thank you.