This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: [PATCH] Fix PR26439
On Thu, 23 Feb 2006, Diego Novillo wrote:
> Richard Guenther wrote:
>
> > Pinskia changed complex types to be considered candidates for decomposing
> > with SFTs. I vaguely remember him having problems including them in
> > AGGREGATE_TYPE_P - partly because sometimes complex variables are
> > SSA_NAMEs I guess.
> >
> If we decompose complex types with SFTs, then they cannot be gimple
> registers.
>
> Oh, but we have DECL_COMPLEX_GIMPLE_REG_P. We probably registerize them
> if they're always accessed whole (not sure I like that too much).
I guess with complex lowering and SFTing we can maybe drop
DECL_COMPLEX_GIMPLE_REG_P and make complex types aggregates always? Did
we have testcases that show(ed) improvements with the
DECL_COMPLEX_GIMPLE_REG_P stuff?
Richard.