Re: [PATCH] Fix reload inheritance bug

[bernds_cb1 at t-online dot de (Bernd Schmidt)]

>   http://gcc.gnu.org/ml/gcc-patches/2004-06/msg01647.html

> Bootstrapped/regtested today on amd64-mandrake-linux-gnu.  OK for mainline?

> 2005-01-24  Eric Botcazou  <ebotcazou@libertysurf.fr>
>
> 	Patch from Richard Sandiford <rsandifo@redhat.com>
> 	* reload1.c (choose_reload_regs): Prevent the find_equiv_reg() code
> 	from inheriting a subreg equivalence with a non-spill register.

From the discussion it seems like there's quite a lot of things wrong 
in this area, and the patch doesn't try to address them.  Ideally we 
should find a fix that allows us to actually use these subregs for 
inheritance, rather than disabling them altogether.  This patch feels 
more like a workaround than a solution, but for now I can't think of 
another easy way to fix it.

However, the change does look safe to me, and it should fix the main 
problem.  Mark - any preferences whether you'd like this in gcc-4.0?


Bernd