Re: [Bug other/43562] GCC ICE on optimize attribute

[Andrew Pinski <pinskia at gmail dot com>]

Sent from my iPhone

On Apr 19, 2010, at 12:02 AM, "jakub at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org 
> wrote:

> ------- Comment #11 from jakub at gcc dot gnu dot org  2010-04-19  
> 07:02 -------
> This change broke building wine on x86-64.


There was already a bug filed for this and a patch was committed today  
or yesterday to the trunk to fix this ice.



> Distilled testcase:
> __attribute__ ((ms_abi)) fn2 (int, int);
> struct S { int s; };
> struct T { int t; struct S *u; };
>
> void
> foo (struct T *x, void *y)
> {
>  int a, b, c, d;
>  fn1 (0, &a, &b, &c, &d);
>  x->u[x->t].s = fn2 (fn3 (), a);
>  fn1 (y, &a, &b, &c, &d);
> }
>
> __attribute__ ((ms_abi)) void
> bar (int x)
> {
> }
>
> void
> baz (int x)
> {
> }
>
> now ICEs at -O2 on x86_64-linux with:
> wineice.i: In function 'foo':
> wineice.i:12: internal compiler error: in insert_save, at caller- 
> save.c:1171
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <http://gcc.gnu.org/bugs.html> for instructions.
>
> Apparently setup_save_areas
>                  saved_reg->slot
>                    = assign_stack_local_1
>                      (regno_save_mode[regno][1],
>                       GET_MODE_SIZE (regno_save_mode[regno][1]), 0,  
> true);
> creates a slot where regno_save_mode[regno][1] for regno 4 is  
> VOIDmode, and the
> VOIDmode MEM which has MEM_ALIGN 0 then confuses everything.
>
>
> --
>
> jakub at gcc dot gnu dot org changed:
>
>           What    |Removed                     |Added
> --- 
> --- 
> ----------------------------------------------------------------------
>             Status|RESOLVED                    |UNCONFIRMED
>         Resolution|FIXED                       |
>
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43562