svn commit: r330004 - in head/stand: . arm efi forth i386 mips powerpc sparc64

[John Baldwin]

On Tuesday, February 27, 2018 11:38:40 AM Rodney W. Grimes wrote:
> > On Tuesday, February 27, 2018 09:28:49 AM Kyle Evans wrote:
> > > On Sun, Feb 25, 2018 at 9:16 PM, Warner Losh <imp at freebsd.org> wrote:
> > > > Author: imp
> > > > Date: Mon Feb 26 03:16:04 2018
> > > > New Revision: 330004
> > > > URL: https://svnweb.freebsd.org/changeset/base/330004
> > > >
> > > > Log:
> > > >   Add NO_OBJ to those directories that don't make anything.
> > > >
> > > >   For directories that don't many anything, add NO_OBJ=t just before we
> > > >   include bsd.init.mk. This prevents them from creating an OBJ
> > > >   directory. In addition, prevent defs.mk from creating the machine
> > > >   related links in these cases. They aren't needed and break, at least
> > > >   on stable, the read-only src tree build.
> > > >
> > > 
> > > Any objection to also removing efi from i386 SUBDIR? It seems silly
> > > to be explicitly adding it when we know nothing here is applicable and
> > > it's going to take some amount of work to get there.
> > 
> > Actually, I believe the only thing it needs is for the i386 kernel to
> > understand the EFI memory map.  If you made the i386 EFI loader fake up the
> > SMAP metadata from the EFI memory map you could probably boot an unmodified
> > i386 kernel.  (It would be better to teach the i386 kernel to handle the EFI
> > memory map though.)  It's more a matter of does someone care enough to
> > actually get a test environment going to test it.
> 
> I may have some of the test pieces in place in the form of ipxe32.efi
> and network booting.  I just cant rember which piece of hardware
> triggered that path in my configuration, I had to force it to legacy
> bios mode to get it working.
> 
> > One would also eventually want to teach the i386 kernel about efifb as well
> > as EFI runtime services, but that shouldn't be strictly required for booting
> > an i386 kernel via EFI.
> 
> How much of that could be done in bhyve with a different compile of edk2?

It could be done, but I've never built edk2 for bhyve.  I believe there are
existing EFI32 blobs for qemu though and that is probably a quicker route to
getting this working (it is what I would do if I worked on it).

-- 
John Baldwin