RE: blob/src/blob main.c,1.46,1.47

[Christopher H. <ch...@mu...>]

It seems to me we need a concept above and beyond a partition (= a chunk
of flash).

I'll call this an "image" for lack of a better word.  

-- We have a list of "images" to be loaded on boot.
-- No reason for "kernel" or "ramdisk" image to be special.
-- Any image can:
	-- either "execute" in place or copied/loaded into ram from
somewhere
	-- somewhere == raw partition, download, file on a filesystem in
a partition
-- Certain commands should require certain images to be present (i.e.
exist for xip else loaded  from somewhere)

-- Image loading would have a post-success hook so that loading certain
flavor images could update the param table(ugh) -- we would need this
for initrd's, maybe root fs images.

I'd like something along these lines so that I can stuff two kernels and
two jffs roots into flash and chose which one to boot.

-ch


> -----Original Message-----
> From: Russ Dill [mailto:Rus...@as...] 
> Sent: Wednesday, May 01, 2002 4:52 PM
> To: Tim Riker
> Cc: Christopher Hoover; blo...@li...
> Subject: Re: blob/src/blob main.c,1.46,1.47
> 
> 
> On Wed, 2002-05-01 at 16:39, Tim Riker wrote:
> > are you sure you do not have jffs2, cramfs or zimage turned on?
> > 
> > "load kernel" should still load a kernel from your flash 
> partition no?
> > This is exactly what the boot is doing.
> > 
> > if you define --enable-{jffs2|cramfs|zImage} then "load 
> kernel" will try
> > to find a kernel in your ramdisk.
> > 
> 
> random semirelated ramblings...instead of all this hardcoded 
> stuff, why
> not have a pointer for blob, kernel, paramater block, and 
> root (pointer
> to partitions). if two pointers point to the same partition, 
> its assumed
> that there is an fs involved, otherwise, its (likely) raw flash.
> Somehow, CF, ide, netboot, and paramater block needs to get thrown in
> here too.
>