Re: Sparc bootblock [Was: SFS breaks PPC build ]

[Bart Grantham <bart-grubdevel@bartgrantham.com>]

Vincent Pelletier wrote:

> Bart Grantham wrote:
> 
>>- have it verify that the file at the configured sector(s) is an ELF file
> 
> Quite easy if we stick to the header signature.
> 

Well, I was referring to the ELF magic number rather than the multiboot 
magic number.  I'll probably check the first unconditionally and the 
second can be overridden with a boot parameter.

>>- follow through and make sure it properly boots an ELF image (right now
>>it only gets it into memory)
> 
> I think there are 2 ways :
> -write your own Forth ELF relocating code (might be fun, but beside that...)
> -have the "init-program" command do this work for you, which require you
> to load the image where "boot net" loads it (or maybe other commands,
> but for now I only know about boot net as it is the only way to
> currently boot grub2 on sparc64 ;) ).

Heh.  I looked into this and I'm definitely going the init-program 
route.  I don't think that there's a platform independent way of doing 
relocation and I don't really feel that it's in the spirit of OF 
anyways.  Besides, writing F-Code is mind-bending enough as it is.  :)

>>- verify that the code works on PPC, and if not, try to make it work so
>>that there's a unified bootblock [not sure if this is reasonable as I'm
>>not as familiar with how PPC OF boots... how similiar is it to Sparc?]
> 
> I think this one can't be done without making the FCode grow quite big.

Perhaps.  But I'll try for it anyways because if it's possible, I'd like 
to have it as a workable alternative for PPC users who aren't using a 
filesystem their PROM understands.  For example, someone porting ReactOS 
(NTFS) or HaikuOS (BFS) to PPC.  (Hey!  Don't laugh... it could happen! ;) )

>>- same for sparc32
> 
> I would be interested if the port could be unified in 32bits.

Me too.  Or at least try to get the two platforms as close as possible. 
  I'll keep you abreast of what I discover.

Bart