This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [ColdFire 36/63] Use RTL for prologues and epilogues
On Tue, 2007-01-23 at 17:50 +0100, Roman Zippel wrote:
> Hi,
>
> On Tue, 23 Jan 2007, Richard Sandiford wrote:
>
> > > Considering the possibility we generate a movem pattern before reload, the
> > > pattern would be readonly anyway and would only have informational value
> > > and could be optional. The predicate check would then simply verify that
> > > this extra information matches the information in the unspec.
> > >
> > > OTOH considering that there is currently no use for this detailed
> > > information, I'd rather keep it simple (and the unspec doesn't preclude
> > > adding this information later).
> >
> > I simply don't understand this argument. The patch I posted
> > describes what the pattern does in ordinary RTL, and verifies
> > that the pattern is a valid h/w instruction. That seems exactly
> > the right thing to do. Why go out of your way to introduce an
> > unspec when a patch that doesn't need it already exists?
>
> The point is that I think that all the verification/extraction code is
> wasteful for something we know is static.
I don't see that it's wasteful -- having the RTL mirror what the h/w
instruction does is generally the right thing to do unless there is
some significant reason to do otherwise.
Jeff