[lwip-users] [lwip] RS485

[Federico Bianchi]

> There is a bug in the uIP checksum code when compiled for big endian
systems.
> The bug is quite easily fixed: the code in uip_arch.c that looks like
this:
>
>   /* add up any odd byte */
>   if(len == 1) {
>     acc += (u16_t)(*(u8_t *)sdata);
>   }
>
> is changed to:
>
>   /* add up any odd byte */
>   if(len == 1) {
>     acc += htons((u16_t)(*(u8_t *)sdata) << 8);
>   }

Nice to know. I just fixed my copy of uIP sources.

> I have no experience with RS485, but I believe it is a multipoint link
type
> (?). If so, it might need some address resolution protocol such as ARP.
But I
> do not have any experience in this field, so you probably shouldn't listen
to
> me.

It is a master-slave multipoint protocol, closely related to the
single-point RS422A. In control system there are often higher level layers
such as IEC-1131 (for PLCs) or the likes. CANbus or, for that matter, a
FieldBus-like architecture might also be profitably implemented on top of it

As a matter of fact, the main trouble with TCP/IP v4 lies in the addressing.
IPv6 is supposed to solve this, and many other problems as well, but so far
it hasn't caught enough to reach a critical mass.

Best regards, and kudos for your network stack implementation. I also
enjoyed the MiniWeb demonstration code a lot.

[This message was sent through the lwip discussion list.]