[PATCH] Fix PR optimization/11637 (x86)

[Zack Weinberg]

Eric Botcazou <ebotcazou@libertysurf.fr> writes:

> Hi,
>
> This is again a problem with -fnon-call-exceptions, a regression present on 
> the 3.3 branch (and latent on mainline).
>
> The compiler emits
>
>   movl    5.0000000745058059692383e-2, (%eax)
>    
> in the assembly file, which is not quite a pure x86 idiom.

Decimal floating point literals should never appear in assembly
output.  The real bug is that print_operand tried to do that.  There's
nothing wrong with the movsf_1 pattern.  I *think* what happened is
that the CONST_DOUBLE in question was marked as one of [DXT]Fmode for
some reason, thus hitting the 

  /* These float cases don't actually occur as immediate operands.  */
  else if (GET_CODE (x) == CONST_DOUBLE && GET_MODE (x) == DFmode)
   ...
  else if (GET_CODE (x) == CONST_DOUBLE
           && (GET_MODE (x) == XFmode || GET_MODE (x) == TFmode))
   ...

cases in i386.c:print_operand.  So there are actually two bugs:
(1) that print_operand uses real_to_decimal at all, (2) that we
a non-SFmode CONST_DOUBLE at this point.  Please look into those.

zw