Re: [lwip-users] Port to 8086 Architecture.

[Robert Morse]

On Feb 16, 2007, at 9:55 AM, Kieran Mansley wrote:

> On Fri, 2007-02-16 at 09:49 -0500, Robert Morse wrote:
> > Thanks for all the help, but I now have it running.
> >
> > What I had to do was in the one section where the address compare in
> > the pbuf_header() routine.  I had to change the if to call a function
> > that created a linear address from the segment/offset then compare 
> > those
> > address.
>
> It's not entirely clear from this what change you made, and whether it
> is more widely applicable.  Could you supply a diff?
>
> Kieran
Ok,

In the pbuf.c module, here is the diff that I used:

482c482
<     if ((u8_t *)p->payload < (u8_t *)p + sizeof(struct pbuf)) {
---
>     if (MAKELINADDRESS((u8_t *)p->payload) < MAKELINADDRESS( (u8_t 
*)p + sizeof(struct pbuf)) ) {

Then in my sys_arch.c, i created the following function.

u32_t makelinAddress( void *a)
{
    u32_t a_linAddr;
    u32_t a_o = (u32_t)FP_OFF(a);
    u32_t a_s = (u32_t)FP_SEG(a);

    a_linAddr = (a_s << 8) + a_o;
    return( a_linAddr);
}

Then in my cc.h, I did a define of
#define MAKELINADDRESS(x) makelinAddress(x)

For system that already have a linear address, they could define the 
following in cc.h
#define MAKELINADDRESS(x) x

Now, I am not sure this is the only place where something like this 
might need to be done,
as all the pointer manipulations that is done, there might be some more 
offset wraps that
might happen depending on where the data areas are placed.  I have not 
yet looked at
replacing all pointers with 'far pointers'.

Setup.

Runing a Turbo86 processor, which looks like a 8086 but has 256 byte 
pages instead of
16 byte pages of a 8086.  This allows 16Meg memory space instead of the 
8086 1Meg memory
space.  I have to run in Large model as I do not have source to our 
kernel, and that is
only compiled in Large model.

Robert