Re: How to clean up i386 machine description?

[Jan Hubicka <hubicka at horac dot ta dot jcu dot cz>]

> 
> special patterns for inc/dec are actually a problem; they should not be
> enabled before reload.
Interesting.. I was asking for this Jeff and he told me that it is not
problem.
Why this happends?
> 
> If you need to order instruction alternatives differently for different
> CPUs or use different constraint modifiers so that the alternative
> selection is changed, currently the only way is to have different
> patterns for the different CPUs.
What do you mean by constraint modifier?
> I'm afraid this doesn't sound like much of a cleanup...
:(
I am using constraint for selection in case selection is indepenent on CPU
(like in add case, where add with 8 bit immediate is always better
as well as short opcode for add eax)
> 
> The cc0_probably_useless hack is probably useless.  It violates the
> assumtions of gcc about targets that use cc0.
> I think you have to re-write the i386 target to explicitly use
> condition code registers (AFAIK there are actually separate integer
> and fp condition code registers?).  These should probably be fixed
> registers.
If someone will help me to understand problematic closely, I should try
to do that (it seems to be mostly monotonous work of adding clobbers
and rewirting condition patterns that I should handle (uff))
But I want to be sure that it will work once I will finish it.

Honza

-- 
------------------------------------------------------------------------------
                   Have you browsed my www pages? Look at:
                       http://www.paru.cas.cz/~hubicka
      Koules-the game for Svgalib,X11 and OS/2,  Xonix-the game for X11
      czech documentation for linux index, original 2D computer art and
              funny 100 years old photos and articles are there!