https://gcc.gnu.org/bugzilla/show_bug.cgi?id=122021
Andrew Pinski <pinskia at gcc dot gnu.org> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |pan2.li at intel dot com
--- Comment #4 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
(In reply to Andrew Pinski from comment #3)
> (In reply to Andrew Pinski from comment #2)
> > I am 80% sure it was r16-3890-gd0526e8ef0d92e which caused this.
>
> Yes now I am 99% sure. It looks like it missed the unsigned change between
> the casts.
That is:
```
if (is_gimple_assign (def)
&& CONVERT_EXPR_CODE_P (gimple_assign_rhs_code (def)))
{
tree cast_rhs = gimple_assign_rhs1 (def);
unsigned rhs_prec = TYPE_PRECISION (TREE_TYPE (cast_rhs));
unsigned type_prec = TYPE_PRECISION (type);
unsigned val_prec = TYPE_PRECISION (TREE_TYPE (val));
if (type_prec >= rhs_prec && val_prec >= rhs_prec)
rhs = cast_rhs;
}
}
```
Does not check TREE_UNSIGNED between the different casts.
_3 is `unsigned short`
_4 = (short int) _3;
_15 = (signed int) _4;
is different from:
_15 = (signed int) _3;
because the first case sign extends from 16bit to 32bit while the second one
zero extends from 16bit.
