Ollie Lho <[EMAIL PROTECTED]> writes:
> Eric,
> I still have some problem in ldscript/crt0.S with incorrectly
> created ROM_TOP, ROM_BASE or reset_vector. This results in a 512KB
> linuxbios.strip image for USE_DOC_MIL. Can you take a look at this
> issue ??
The code is working correctly. But is not handling the docipl case where
you don't need the reset vector. Architecturally the reset vector
only appears at either 0xFFFF0 or 0xFFFFFFF0, and currently the code assumes
this. So when linuxBIOS is located at 0x80000 and it ends at 0xFFFFF
you get a 512KB image.
To be clear the generated reset vector is correct, it does a far
jump to 0x8000:0 but this is simply not what you want.
There are a number of work arounds and I'll look at them. But the primary
solution we need to find is how to get ipl.S to not be a special case. This
requires a reorganization of crt0.S. And some looking at what kind of
entry points the main linuxBIOS code needs.
My biggest question. Does the docipl case need a 16bit entry point?
Probably.
And I need to see if I can use one linker script or if it will require
two.
Eric
