TCP fragmentation issue

[Ian Caddy]

Chris Johns wrote:

> Ian Caddy wrote:
> 
>>
>> I thought I would look into 4.6.1 to see if it was still handled the 
>> same way and it appears to be, although I can't find the param.h file, 
>> in fact I can't find a definition anywhere in the tree for MSIZE! (I 
>> even greped the whole tree *.*)
>>
> 
> It is in newlib:
> 
>  newlib/newlib/libc/sys/rtems/sys/param.h

Thanks for that, now I can rest easy. ;-)

> 
>>
>> I have two choices.  The first is to modify sosend to link in the 
>> second mbuf to send the whole frame, or the other option is manually 
>> modify the minimum cluster size to it matches up with the mbuf size 
>> with no gap.  I  will let you all know how I go.
>>
> 
> What happens if a router fragments the packets due to varying MTU sizes 
> in the path ?


I have explained to the project people that there is no guarantee that 
TCP stream will not get broken up.  As I stated in the first email, the 
problem lies in the PC based applications that our projects people are 
talking to.  There are two applications, one of them has been fixed by 
the vendor, but the other one seems to be more of a problem getting fixed.

The nextworks that they will run on are simple Ethernet networks that 
should not (famous last words) contain routers, but at least for the 
short term I have made their lives easier.

I went for the aproach of modifying sosend that it will try to link 
together 2 mbufs when the packet size falls between the 100 and 208 
bytes.  In the rudimentary testing I have done so far, everything looks 
good.

Talk to you soon,

Ian Caddy