This is the mail archive of the
mailing list for the GCC project.
Re: PATCH, rs6000 (alpha?) long const
Richard Henderson wrote:
> On Sat, Dec 29, 2001 at 09:22:23AM -0600, Tom Rix wrote:
> > Looks like alpha also uses this logic. Can someone familar with alpha
> > comment on the old logic's correctness?
> The 'L' constraint is for exactly this sort of constant. This
> is supported by the "ldah" insn on alpha, and "addis" on ppc64.
Yes. The internal errors complained about not meeting the insn's 'L'
contraint. In the case I looked at, rs6000_emit_set_long_const was
called from rs6000_emit_allocate_stack, via try_split. I believe the
usual contraint checking is bypassed and was worried alpha would have the