Hi all,
I've been working on lguest64 and in order to do this, I had to move
a lot of the i386 specific out of the way. Well, the lguest64 port
is still not ready to display, but before Rusty makes too many changes
I would like this in upstream so I don't have to keep repeating my
changes :-)
So this patch series moves lguest32 out of the way for other archs.
-- Steve
On Wed, 2007-08-08 at 20:32 -0400, Steven Rostedt wrote:
> Hi all,
>
> I've been working on lguest64 and in order to do this, I had to move
> a lot of the i386 specific out of the way. Well, the lguest64 port
> is still not ready to display, but before Rusty makes too many changes
> I would like this in upstream so I don't have to keep repeating my
> changes :-)
>
>
> So this patch series moves lguest32 out of the way for other archs.
Yeah, after some more thought I've not applied most of this. We really
don't want to move everything then move it back; I prefer Jes' more
cautious approach of moving a little bit at a time.
We really have three parts: (1) bits that are generic, (2) bits that
should be generic but my implementation is naive, (3) bits that really
are i386-specific.
Patches which move 2 to 1 are gratefully accepted: I realize a mass move
is easier and this requires thought, but that's what we need.
Since I can't build a module over two directories, that seems to destroy
the idea of an i386/ subdir. Instead I've done a patch which renames
the *clearly* i386-specific things to i386_<name>, which at least works.
I've pushed it into the repository http://lguest.ozlabs.org/patches/
Cheers,
Rusty.
Rusty Russell wrote:
> Yeah, after some more thought I've not applied most of this. We really
> don't want to move everything then move it back; I prefer Jes' more
> cautious approach of moving a little bit at a time.
>
> We really have three parts: (1) bits that are generic, (2) bits that
> should be generic but my implementation is naive, (3) bits that really
> are i386-specific.
I have actually been using Steven's patchset to do my work, but in a
way it's a tool. Moving things out of the way in bulk and see whats
missing kinda helps for that :)
> Patches which move 2 to 1 are gratefully accepted: I realize a mass move
> is easier and this requires thought, but that's what we need.
>
> Since I can't build a module over two directories, that seems to destroy
> the idea of an i386/ subdir. Instead I've done a patch which renames
> the *clearly* i386-specific things to i386_<name>, which at least works.
> I've pushed it into the repository http://lguest.ozlabs.org/patches/
That works - alternatively we could build two modules, lg and lg_<arch>
and just have lg pull in the arch one as well? I'm not really biased,
but I think it will get messy later once we add ia64 and x86_64 to the
directory.
Cheers,
Jes