Re: machine description porting question.

[Spundun Bhatt <spundun at ISI dot EDU>]

Hi, I am redirecting my message from the gcc-bugs list to here.
I still have many confusions about creating temp registers in machine 
description.

Hans-Peter Nilsson wrote:

> On Fri, 18 Oct 2002, Spundun Bhatt wrote:
>
> > Hi,
> > I am porting gcc to a new platform and it seems this mailing list is the
> > closest match. My apologies if this is not the correct place.
> gcc@ is better.
>
> > The machine that I am porting to has an instruction set similar to IBM
> > rs6000, so we have taken rs6000 config files as the tmplate and are
> > trying to make modifications to it.
> > My question is mainly regarding adding a temporary register in any insn
> > definition in the machine description file.
> It's not really "mainly", but a special case.  (Getting a
> temporary register is done by other solutions in other
> contexts.)
I am still trying to get hold of -one- way to do this. :)

> > the scenario:
> > my machine doesnt have a compare instruction so I want to use subc
> > instruction (subtract and modify cc).
> > The compare instruction on powerpc looks similar to cmp RA,RB while
> > subtract instruction on my architecture looks like SUBC rD,rA,rB.
> > So gcc has to allocate a temporary register for rD, and I dont know how
> > to go about that.
> You make the compare instruction pattern a define_expand which
> expands to the subtract insn.  Or, since you say your machine is
I tried as you but couldnt get it working. Details follow.

> rs6000-like, you use the same expand-compare-insn-in-the-cc-user
> method used there, and expand to a *subtraction* (e.g.
> gen_subsi3) in your equivalent to rs6000_generate_compare.  You
> get the temporary result register from gen_reg_rtx (mode).
I think this expand-compare.... is some mnemonic that you came up with 
and I didnt understand, because I couldnt grep it in the cvs code :). 
Here I need to clarify that the version of gcc that I am working with 
is a branch from gcc 2.95.2 patched by motorola for the altivec 
instruction set. In 2.95.2, the rs6000_generate_compare does not exist 
so I couldnt try this sugession.

> brgds, H-P
the following is the snipped part of my original mail.