Re: Bug#927892: grub-ieee1275-bin: Please add bootinfo.txt for sparc and sparc64
On 24/04/2019 23:05, John Paul Adrian Glaubitz wrote:
> Okay, so I have managed to build an image now, that's the good news:
>
>> https://people.debian.org/~glaubitz/debian-10.0-sparc64-grub-NETINST-1.iso
>
> The bad news is, it doesn't boot yet:
>
> {0} ok boot cdrom
> Boot device: /virtual-devices@100/channel-devices@200/disk@1 File and args:
> ERROR: /virtual-devices@100/channel-devices@200/disk@1: Can't open disk label package
>
>
> Can't open boot device
>
> {0} ok
>
> With the attached d-i patch, debian-installer builds with GRUB instead of SILO
> and produces a sparc64.elf boot image plus a grub.cfg configuration file.
>
> The other patch shows the current modifications to debian-cd.
>
> I assume the only thing that is missing is the correct parameter to mkisofs
> so that it builds a bootable image, but I haven't looked that up yet. And we
> don't need to create the directory "sparc64-ieee1275", it stays empty anyway
> as opposed to ppc64el.
FWIW I think you're pretty close here: as you say, you just need to fix the bits
which set up a sun partition label and inject the bootloader. So the bit you need to
re-enable to get this to work should be this:
add_mkisofs_opt $CDDIR/../$N.mkisofs_opts "-G boot1/boot/isofs.b -B ..."
>From looking at your previous grub patches and the genisoimage man page then you need
to point this towards the a.out binary made from
grub/grub-core/boot/sparc64/ieee1275/boot.S which I think reading through
https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=921249 should be cdboot.img?
The mkisofs above should create a sun partition layout and inject the a.out binary
into all slices so it is executed with "boot cdrom". This will then use the CIF to
locate the grub partition via the PROM bootpath property and the open the ISO9660
device in the PROM to locate the grub kernel.
Note: this assumes that the PROM can read ISO9660 devices directly rather than
requiring a UFS filesystem - can all 64-bit SPARC PROMs do this? The above diff
suggests that this was the case for SILO.
ATB,
Mark.
Reply to: