On Wed, Apr 22, 2009 at 06:38:50PM +0300, Pekka Enberg wrote:
> Hi,
>
> On Wed, 22 Apr 2009 12:21:01 +0300
> Eduard - Gabriel Munteanu <[email protected]> wrote:
> > > Some code relating to x86 and x86-64 differs only in regards to register
> > > width. This change makes it possible to specify registers by word-size
> > > independent names, thus allowing for generic code while keeping the
> > > compatibilities visible.
>
> On Wed, 2009-04-22 at 17:25 +0200, Arthur Huillet wrote:
> > I do not love this change.
>
> <snip>
>
> > Eduard: I suggested aliasing REG_EAX and REG_RAX instead of using a new
> > symbolic name.
> > Is there anything wrong with this approach?
>
> I think Arthur's suggestion is fine as REG_EAX is just an alias for
> (part of) REG_RAX on 64-bit. So you can never allocate REG_EAX and
> REG_RAX at the same time anyway.
>
> Pekka
My intention was to provide some sort of semantic difference between
EAX, RAX and either-EAX-or-RAX, i.e. have REG_EAX and REG_RAX be different
constants but mutually-exclusive. As in "some things do well with 32-bit
regs, others with 64-bit and others just don't care". I'm not sure this is
actually needed or useful to implement proper instruction selection and
register allocation, although it makes intentions a bit more explicit.
I'll fix it up as suggested and resubmit.
(Then I'll have to do some research on monoburg, the insn selector and
register allocation and get up to speed with these things.)
Eduard
------------------------------------------------------------------------------
Stay on top of everything new and different, both inside and
around Java (TM) technology - register by April 22, and save
$200 on the JavaOne (SM) conference, June 2-5, 2009, San Francisco.
300 plus technical and hands-on sessions. Register today.
Use priority code J9JMT32. http://p.sf.net/sfu/p
_______________________________________________
Jatovm-devel mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/jatovm-devel
