Idea: move grub-emu to a separate platform

Idea: move grub-emu to a separate platform

[Bean]

Hi,

Currently, the grub-emu is in every rmk files, but I don't see the
difference between grub-emu in i386-pc, i386-coreboot or i386-efi. In
fact, grub-emu runs inside the host os, it can't access firmware
facility anyway. One quick fix to move grub-emu to i386.rmk, but we
could also put it in a new platform altogether, something like
i386-host.

The advantage of using a separate platform is that we could now use
modules like the other platform, instead of embedding all source files
in grub-emu. For example, whenever we add a new command, we need to
modify the source list of grub-emu. But if we build i386-host like
other platform, we only need to change common.rmk.

What's more, i386-host is not just a testing tool, it could be a real
boot loader. kboot provides a minimum linux environment. We could use
i386-host as a front end, and use kexec to launch additional boot
files.

-- 
Bean

Re: Idea: move grub-emu to a separate platform

[Vesa Jääskeläinen]

Bean wrote:
> Hi,
> 
> Currently, the grub-emu is in every rmk files, but I don't see the
> difference between grub-emu in i386-pc, i386-coreboot or i386-efi. In
> fact, grub-emu runs inside the host os, it can't access firmware
> facility anyway. One quick fix to move grub-emu to i386.rmk, but we
> could also put it in a new platform altogether, something like
> i386-host.
> 
> The advantage of using a separate platform is that we could now use
> modules like the other platform, instead of embedding all source files
> in grub-emu. For example, whenever we add a new command, we need to
> modify the source list of grub-emu. But if we build i386-host like
> other platform, we only need to change common.rmk.
> 
> What's more, i386-host is not just a testing tool, it could be a real
> boot loader. kboot provides a minimum linux environment. We could use
> i386-host as a front end, and use kexec to launch additional boot
> files.
> 

I think I have from time to time proposed this too :) So I am all for it :)

But my point was more to cleanup makefiles in sensible way and then on
PC to allow more advanced automated (module) testing.

Re: Idea: move grub-emu to a separate platform

[Robert Millan]

On Sat, Mar 28, 2009 at 02:36:00PM +0800, Bean wrote:
> Hi,
> 
> Currently, the grub-emu is in every rmk files, but I don't see the
> difference between grub-emu in i386-pc, i386-coreboot or i386-efi. In
> fact, grub-emu runs inside the host os, it can't access firmware
> facility anyway. One quick fix to move grub-emu to i386.rmk, but we
> could also put it in a new platform altogether, something like
> i386-host.
> 
> The advantage of using a separate platform is that we could now use
> modules like the other platform, instead of embedding all source files
> in grub-emu. For example, whenever we add a new command, we need to
> modify the source list of grub-emu. But if we build i386-host like
> other platform, we only need to change common.rmk.

Can't we just use common.rmk instead?  Adding a separate namespace for
util/ stuff looks like unnecessary complication.  We already have e.g.
grub-probe, grub-setup, etc which don't need this hack.

-- 
Robert Millan

  The DRM opt-in fallacy: "Your data belongs to us. We will decide when (and
  how) you may access your data; but nobody's threatening your freedom: we
  still allow you to remove your data and not access it at all."

Re: Idea: move grub-emu to a separate platform

[David Miller]

From: Robert Millan <rmh@aybabtu.com>
Date: Sat, 28 Mar 2009 14:42:34 +0100

> On Sat, Mar 28, 2009 at 02:36:00PM +0800, Bean wrote:
> > Hi,
> > 
> > Currently, the grub-emu is in every rmk files, but I don't see the
> > difference between grub-emu in i386-pc, i386-coreboot or i386-efi. In
> > fact, grub-emu runs inside the host os, it can't access firmware
> > facility anyway. One quick fix to move grub-emu to i386.rmk, but we
> > could also put it in a new platform altogether, something like
> > i386-host.
> > 
> > The advantage of using a separate platform is that we could now use
> > modules like the other platform, instead of embedding all source files
> > in grub-emu. For example, whenever we add a new command, we need to
> > modify the source list of grub-emu. But if we build i386-host like
> > other platform, we only need to change common.rmk.
> 
> Can't we just use common.rmk instead?  Adding a separate namespace for
> util/ stuff looks like unnecessary complication.  We already have e.g.
> grub-probe, grub-setup, etc which don't need this hack.

I think this would be nicer too.

If there are target specific bits to deal with, we could just
have a target.rmk specific rule that builds a library with
those objects, or use some macro containing those target
specific objects.  And then the common.rmk rule links them
into the common grub-emu rule.

Re: Idea: move grub-emu to a separate platform

[Robert Millan]

Btw, Vladimir's thread "Move normal.mod to conf/common.rmk" seems related
to this, since normal.mod and grub-emu have almost the same problem.

-- 
Robert Millan

  The DRM opt-in fallacy: "Your data belongs to us. We will decide when (and
  how) you may access your data; but nobody's threatening your freedom: we
  still allow you to remove your data and not access it at all."