On 9/19/05, Christopher Schultz <chris@christopherschultz.net> wrote:
> Vini,
>=20
> > Of course I am not trying to load a mips kernel on an intel based
> > cobalt, I am just trying to work out a way to make its boot loader load
> > the kernel.
>=20
> You mentioned that it "doesn't seem to work", but that something is
> happening. Can you be a little more specific?
>=20
> Usually there's a hardware bootloader (the one you're calling the
> "Cobalt bootloader"?) that loads the software bootloader (which, on a
> regular PC is usually stored in the boot sector of the hard disk). I
> have forgotton where this is on the Cobalt-mips boxes, but it's pretty
> much the same thing. The Cobalt bootloader is super-simple and does
> nothing but get the disk bootloader into memory and execute it.
>=20
> The disk bootloader is also pretty simple, because it's generally
> limited by the size of the boot sector (often around the 512k mark).
> This bootloader loads the final, more complicated bootloader which knows
> how to do things like find and un-gzip the kernel, etc.
>=20
> Do you know which, if any of these, is being executed? For example, does
> it give you the "loading kernel\|/-\|/-\|/-\|/-\|/-\|/-\|/-..." message
> like the Cobalt(mips) bootloader does?
>=20
> I used to have 16MB of RAM in my mips-based Cobalt and the loader used
> after 1.6.0 wasn't capable of properly loading the kernel without either
> (a) overwriting itself in memory or (b) writing the kernel "off the end"
> of memory -- which often results in it's being written to the beginning
> or memory. In either case, it was bad.
>=20
> Is it possible that the Intel-based bootloader has a similar problem?

It would be really cool if NetBSD worked on the Intel ones. The mips
box would be much more useful if it was just a little faster. Also,
some stuff doesn't build well under mipsel. If I knew NetBSD would
boot or the bootloader can actually find and boot the kernel, I'd
probably buy one just to help out.

Andy