Patchwork [RFA,middle-end] Fix PR target/41993

login
register
mail settings
Submitter Uros Bizjak
Date Nov. 12, 2012, 8:36 a.m.
Message ID <CAFULd4bNJounwyhgT3DgAJzPZ_iH-yZG-Z-rbRoJytD4M0UJJQ@mail.gmail.com>
Download mbox | patch
Permalink /patch/198330/
State New
Headers show

Comments

Uros Bizjak - Nov. 12, 2012, 8:36 a.m.
On Tue, Nov 6, 2012 at 9:12 AM, Eric Botcazou <ebotcazou@adacore.com> wrote:
>> 2012-11-05  Uros Bizjak  <ubizjak@gmail.com>
>>           Kaz Kojima  <kkojima@gcc.gnu.org>
>>
>>       PR target/41993
>>       * mode-switching.c (create_pre_exit): Set return_copy to
>>       last_insn when copy_start is a pseudo reg.

It looks to me, that we in fact want:

--cut here--
--cut here--

If we find an unrelated HARD register, we will fail in the same way as
described in the PR. This was found by post-reload vzeroupper
insertion pass that tripped on unrelated hard reg assignment. At this
point, we are interested only in hard registers that are also used for
function return value. Actually, even in pre-reload pass, there are no
other assignments to hard registers.

Uros.
Eric Botcazou - Nov. 12, 2012, 8:40 a.m.
> It looks to me, that we in fact want:
> 
> --cut here--
> Index: mode-switching.c
> ===================================================================
> --- mode-switching.c    (revision 193407)
> +++ mode-switching.c    (working copy)
> @@ -330,7 +330,7 @@
>                           short_block = 1;
>                         break;
>                       }
> -                   if (copy_start >= FIRST_PSEUDO_REGISTER)
> +                   if (!targetm.calls.function_value_regno_p (copy_start))
>                       {
>                         last_insn = return_copy;
>                         continue;
> --cut here--
> 
> If we find an unrelated HARD register, we will fail in the same way as
> described in the PR. This was found by post-reload vzeroupper
> insertion pass that tripped on unrelated hard reg assignment. At this
> point, we are interested only in hard registers that are also used for
> function return value. Actually, even in pre-reload pass, there are no
> other assignments to hard registers.

Fine with me if this passes testing on x86-avx and SH4.

Patch

Index: mode-switching.c
===================================================================
--- mode-switching.c    (revision 193407)
+++ mode-switching.c    (working copy)
@@ -330,7 +330,7 @@ 
                          short_block = 1;
                        break;
                      }
-                   if (copy_start >= FIRST_PSEUDO_REGISTER)
+                   if (!targetm.calls.function_value_regno_p (copy_start))
                      {
                        last_insn = return_copy;
                        continue;