On Thu, Apr 17, 2025 at 10:19 AM Eric Botcazou <botca...@adacore.com> wrote: > > > LGTM. I do wonder if this could be documented somewhere if not > > already. Because I suspect there are other places which miss that if > > TYPE_PRECISION are equal, you might still need a cast for boolean > > types. Maybe a helper function might be useful too. > > The knowledge is indeed encoded in useless_type_conversion_p, int_fits_type_p, > fits_to_tree_p (which ought to be called fits_to_type_p AFAICS) but I'm not > sure how to factor it out in a predicate. But we can certainly add a blurb in > tree.def for BOOLEAN_TYPE, for example:
That looks like a good place - can we possibly say sth about signedness a well? IIRC all languages have unsigned bools but vector mask components are signed bools? > diff --git a/gcc/tree.def b/gcc/tree.def > index c4ad8d08f10..19533cfe77c 100644 > --- a/gcc/tree.def > +++ b/gcc/tree.def > @@ -135,7 +135,9 @@ DEFTREECODE (OFFSET_TYPE, "offset_type", tcc_type, 0) > DEFTREECODE (ENUMERAL_TYPE, "enumeral_type", tcc_type, 0) > > /* Boolean type (true or false are the only values). Looks like an > - INTEGRAL_TYPE. */ > + INTEGER_TYPE, but must be dealt with specially because TYPE_PRECISION > + may be arbitrary despite the restricted set of valid values (in other > + words, boolean types with TYPE_PRECISION > 1 exist in some languages). > DEFTREECODE (BOOLEAN_TYPE, "boolean_type", tcc_type, 0) > > /* Integer types in all languages, including char in C. > > -- > Eric Botcazou > >
