Re: [lwip-users] MSS maximum

[Art R.]

Greetings, Simon,

I have rerun our tests and cannot reproduce the situation I saw - where the
SYN and SYN ACK specified MSS options of 536. I can only assume that we had
a build problem and the updated TCP_MSS value did not get used until I did a
complete rebuild. I have set TCP_MSS to 1460 and that is what I am now
seeing in the MSS options going both ways. (The test runs 2 instances of the
lwIP stack 'back-to-back' - one lwIP talks to another.)

Thank you for your reply - I'm sorry to have wasted time on what appears to
be just our build problem.

Art R.




address@hidden wrote:
> 
> Hi Art,
> 
> Can you tell us which version of lwIP you are using? There has been a 
> change in the MSS handling in 1.3.0.
> 
> Apart from that, TCP_MSS is the mss advertised to the remote host (the 
> remote host will use this to calculate the mss used to send to lwIP), 
> while pcb->mss contains the mss as advertised by the remote host. lwIP 
> will use this for sending.
> 
> So if you see lwIP sending with 536, it might be the other side 
> advertises this.
> 
> In my tests, I'm sure to have seen mss's of 1460, so your problem might 
> probably have another reason. The value of 536 is a historical value as 
> described in on of the TCP RFCs and has probably nothing to do with PPP.
> 
> Simon
> 
> 
> Art R. schrieb:
>> The lwIP code seems to enforce an absolute maximum value for MSS of 536.
>> If
>> you set the TCP_MSS tag larger, the lwIP code limits the actual mss to
>> 536.
>> This value is hard-coded in tcp.c. I do not see any way for pcb->mss to
>> become larger than 536, implying that the largest MSS that can be
>> advertised
>> by lwIP to be larger than 536.
>>
>> The result is that a tcp write of more than 536 bytes creates multiple
>> outbound TCP packets.
>>
>> I would like to send 'full size' packets of 1460 bytes. Is there a way to
>> do
>> so using lwIP?
>>
>> (I realize that the 536 value is probably in place to limit packets
>> destined
>> for PPP or etc. I assume there is nothng in lwIP currently to discover
>> packet delivery failures and reduce MSS in response? Is this the reason
>> why
>> lwIP clamps MSS to 536? Or ...?)
>>
>> Thanks,
>> Art R.
>>
>>
>>
>>   
> 
> 
> 
> _______________________________________________
> lwip-users mailing list
> address@hidden
> http://lists.nongnu.org/mailman/listinfo/lwip-users
> 
> 

-- 
View this message in context: 
http://www.nabble.com/MSS-maximum-tp16288107p16291047.html
Sent from the lwip-users mailing list archive at Nabble.com.