Re: A ready to run GRUB2 boot loader for the Genesi ODW ?

[Dennis Clarke]

On 11/15/05, Hollis Blanchard <address@hidden> wrote:
> On Tuesday 15 November 2005 15:07, Dennis Clarke wrote:
> > http://www.blastware.org/grub2/
>
> Looks helpful. :) Is there a particular reason you chose not to use the GRUB
> wiki?

I didn't know it existed.

> It doesn't look like this information is OpenSolaris-specific...

One of the many joys of GRUB2 is that it isn't specific to anything
really.  But I am guessing.  Maybe it could never work on ARM or a Dec
Alpha but I don't see why not.

> That reminds me, how are you booting OpenSolaris using GRUB2 on PPC? With the
> "linux" loader?

No .. booting OpenSolaris with GRUB2.  But that stage is being worked on.

> > I put this page together such that people working on the PowerPC
> > port can focus on more interesting things like the hardware address
> > translation and the memory management layer.
>
> I'm actually not sure what you mean by "hardware address translation and the
> memory management layer"... could you elaborate?

Actually .. not really.  :-(

I wish I could.  Really it comes down to the handling of pages of
memory and how they relate to actual physical memory in the machine
and the address translation that must happen between extended virtual
memory and real physical memory.  This is the venue of the uts code
for the kernel.  We are currently looking at the architecture specific
modules that will be needed and quite frankly, a complete architecture
plan.

    http://www.blastwave.org/dclarke/grub/MMU.png

> From the page:
> > PLEASE TAKE NOTE : You may choose to suspend GRUB2 in order to drop to the
> > firmware however you will need to warm boot your system as you can not yet
> > get back to the GRUB2 prompt.  This may be a flaw in the Pegasos2 firmware
> > or the GRUB2 code.
>
> This is a bug in Pegasos firmware. Genesi was notified a while back but I
> haven't heard anything about a fix. In the meantime, you might be able to run
> "resume" instead of "go" to get back.

ah .. I will give that a try.  Thank you !

> If that does work, you could implement a patch that detects you're running on
> CodeGen firmware (the original author) and sets a quirk/feature bit, then
> output a different message in the suspend command. I believe we already
> detect CodeGen firmware to work around a bug I found on the briQ, so you
> would have an example to work from...

Let me look into that also.

Dennis