tech-embed archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re(2): Port to 8xx embedded PPCs



On Sat, Jun 17, 2000, Perry E. Metzger <perry%piermont.com@localhost> wrote:

>> I'm working on one, but I haven't got very far yet.  I can cross-build
the  
>> system, and I've figured out what to do about floating-point.  Right
>now I'm 
>> working on boot-up; after that I'll be working on virtual memory support.
>
>I'm not very familiar with the 8xxs, but I'm assuming they have no
>hardware floating point? If so, the soft float support we have in
>several other ports can probably be stolen -- it would be very
>straightforward.

Yep, the 8xx has no floating point. The gcc built-in soft float is
usually the simplest way to go (and more efficient than adding a floating
point emulator to the kernel), but in this case, you lost binary
compatibility with existing PPC stuffs. (This is not a problem for most
embedded anyway).

There a some slight differences between the 8xx and the "classic" PPCs
that can (will) cause trouble with exisiting code (especially with the
libc), mostly the cache line size which is 16 bytes and not 32. (Any
dynamic code linker/loader need to be adapted to flush correctly, any
cache tricks used by the libc has too).

Regarding the kernel, the MMU is different, so all the memory management
stuff need probably a complete rewrite.

Ben.




Home | Main Index | Thread Index | Old Index