<Resend due to ML issues, apologies for duplicates>
Hi,
To avoid duplication of effort, I just wanted to share that I am already
leading the project inside Oracle to (1) replace the Legacy GRUB 0.97 bootloader
with GRUB2 (the codebase is mostly 1.98, with additional things from the trunk,
though it may be a later release) and (2) Add UEFI Firmware support to Solaris
(using GRUB2 as the boot loader, obviously). Yes, this will all be tied into
the installer (and those folks are also working on those changes as well) as
well as the boot administration commands. There will be some kernel impact as I
plan to adopt tagged multiboot2 in lieu of the existing multiboot1 support.
BIOS calls will not be made on UEFI platforms. One of my goals is to minimize
forking, which is why I''ve been working closely with Vladimir to ensure
that
Solaris''s requirements are met as well as possible (and with maximum
potential
for reuse on other platforms). The forthcoming ARC cases that I will be
submitting will make this crystal clear; feel free to comment on these cases
after they are submitted. If you would like to talk about this further, please
contact me off-list.
Thanks,
--S
Quoting Vladimir ''?-coder/phcoder'' Serbinenko, who wrote the
following on...:
> Bayard Bell wrote:
> > [corrected recipients/subscription settings]
> >
> > I was wondering where things stood on official support for grub2 in
> > EFI environments. As far as I''m aware the status is:
> >
> > initial patches were provided by "phcoder" about a year ago
> grub2 efi support was made by Bean Lee. I coded only multiboot support
> > these haven''t been merged into the grub2 distribution
> Both were a part of grub 1.98 release. AFAIK grub 1.98 + grub-extras has
> everything solaris needs except:
> 1) Configuration routines
> 2) PXE DHCP ACK passing
> 3) ZFS support in grub-extras isn''t in sync.
> > these haven''t been merged with subsequent changes to
OpenSolaris ZFS
> > or the grub 0.97 fork maintained by the Sun and the OpenSolaris
community
> 0.97 is a completely different codebase.
> > there is some work that may be in progress or pending for Caiman that
> > would include not just the grub2 updates but beadmin support and other
> > touch points in the installer
> > it''s not clear yet whether the kernel still needs
"fakebios" support
> > for bootstrapping in EFI environments, as there appeared to be some
> > issues here to remove legacy BIOS dependencies in the x86 code
> > the grub2 build toolchain hasn''t been extended beyond
Linux/GNU tools
> > and will require some adaptation to support native builds on
OpenSolaris
> GRUB2 builds fine and is supported on FreeBSD.
> >
> > I''ve trying to sort elements of this out for my own benefit
and was
> > looking at merging phcoder''s patches
> > (http://lists.gnu.org/archive/html/grub-devel/2009-04/msg00512.html)
> > and the latest updates to the OpenSolaris grub-0.97 fork to
> > grub-1.97~beta4, which is on my understanding the last stable release
> > of grub2 (I may be mistaken, as there are later builds of 1.97, but
> > 1.98 doesn''t go anywhere with the build, even on Linux). I
was hoping
> > I might make this more than a personal project and see if I could work
> > with someone with commit rights to get this code to a point where it
> > could be accepted into at least the OpenSolaris base. As I believe all
> > the code for grub-0.97 is supposed to be released under GPL
> > (http://blogs.sun.com/darren/entry/zfs_under_gplv2_already_exists), I
> > was also hoping that the patches might be also be written with a view
> > to adding them to the GNU mainline or at least a recognised branch in
> > their bazaar repo, thus allowing other OS distros to provide support
> > for OpenSolaris install detection and boot in their build and reducing
> > requirements for chainloading. I''ve seen some mails
indicating that
> > this requires an ARC review, which I assume means that it''s
considered
> > an architectural change and requires more extensive review and
> > approval, but I''m not clear on how that would impact my
ability to
> > contribute.
> >
> My patches are already in. For code originating from sun/oracle
we''re
> sorting out legal problems.
> > Alternatively, I may be missing out on work already pending or
> > completed to resolve this, in which case I''m happy to test
anything
> > that''s considered suitable for community access.
> >
> > Cheers,
> > Bayard
>
>
> --
> Regards
> Vladimir ''?-coder/phcoder'' Serbinenko
>
>