>> I'm not a TCP/IP or HTTP expert but I think it is possible to response
>> for SYN packet of second connection just after first one drops. Then
>> it would work as fast as my microcontroller can.
>
> It would make the behaviour of the stack very specialised for your particular 
> use case.  It would require that the stack knew that you wanted to do it that 
> way, and for it to know that the SYN was for your PCB, and that you already 
> had an active connection accepted via the same PCB.  None of these are 
> usually of concern to the stack when a SYN arrives.
>
> Going back to the start, can you explain why you need to limit it to a single 
> connection?  Is it just that there is little memory and so you don't want 
> extra connections using it up?
>
> It sounds like what you want is for the stack to accept the incoming 
> connection, but for your server to not have to do anything with it until it 
> is ready.  Perhaps you could allow more than one connection, and just process 
> them serially in your web server?  E.g. whenever the accept callback is 
> called, put the new PCB in a list but do nothing with it, and whenever a 
> connection is closed pull the oldest PCB off the list and start using that 
> one, and so on.

When I set MEMP_NUM_TCP_PCB to 1 I have almost what I want - it's just
too slow! Is it possible to speed up TCP processes by calling some
function few times, for example tcp_tmr?

And Yes, I want to limit connection to only one because I don't have
space in RAM for 2 or more HTTP requests

I tried to do want you are writting about, but maybe I do something
wrong, because it doesn't work. I tried to accept second connection
and save its pcb, arg and err, but do not bind any callbacks to this
pcb until first connection drops. When first connection drops I bind
callbacks but data received callback isn't called. When I send new
data in second connection they are received as normal, but browser
will send data at once after connection when callbacks aren't bound so
I wan't received it.

I use lwIP 1.3.0.

Tomasz

_______________________________________________
lwip-users mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/lwip-users

Reply via email to