[Bug rtl-optimization/62208] [5 Regression] ICE with -fwhole-program on valid code at -O3 on x86_64-linux-gnu in trunc_int_for_mode, at explow.c:56

[rsandifo at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=62208

rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
           Assignee|unassigned at gcc dot gnu.org      |rsandifo at gcc dot gnu.org

--- Comment #5 from rsandifo at gcc dot gnu.org <rsandifo at gcc dot gnu.org> ---
(In reply to Richard Biener from comment #4)
> As it's a single file, does it reproduce with -fwhole-program instead of
> -flto?
> Still can't reproduce in my dev tree (but as usual this may have too many
> patches...).  Confirmed on match-and-simplify branch with -fwhole-program.
> 
> Needs -march=x86-64 as well.
> 
> 55        /* You want to truncate to a _what_?  */
> 56        gcc_assert (SCALAR_INT_MODE_P (mode));
> 
> (gdb) p mode
> $1 = V4SImode
> 
> #4  0x0000000000c1683e in simplify_const_unary_operation (code=NOT, 
>     mode=V4SImode, op=0x7ffff66ca470, op_mode=V4SImode)
>     at /space/rguenther/src/svn/match-and-simplify/gcc/simplify-rtx.c:1735
> 1735          return immed_wide_int_const (result, mode);
> 
> 
> Likely caused by the wide-int merge.  Richard?

Seems to be an unrelated latent bug.  Testing a fix.