Re: [C PATCH]: allow aliasing of compatible types derived from enumeral types [PR115157]

[Martin Uecker <uecker@tugraz.at>]

Am Donnerstag, dem 23.05.2024 um 14:30 -0700 schrieb Ian Lance Taylor:
> On Thu, May 23, 2024 at 2:00 PM Joseph Myers <josmyers@redhat.com> wrote:
> > 
> > On Tue, 21 May 2024, Martin Uecker wrote:
> > > 
> > >     C: allow aliasing of compatible types derived from enumeral types [PR115157]
> > > 
> > >     Aliasing of enumeral types with the underlying integer is now allowed
> > >     by setting the aliasing set to zero.  But this does not allow aliasing
> > >     of derived types which are compatible as required by ISO C.  Instead,
> > >     initially set structural equality.  Then set TYPE_CANONICAL and update
> > >     pointers and main variants when the type is completed (as done for
> > >     structures and unions in C23).
> > > 
> > >     PR 115157
> > > 
> > >     gcc/c/
> > >             * c-decl.cc (shadow_tag-warned,parse_xref_tag,start_enum,
> > >             finish_enum): Set SET_TYPE_STRUCTURAL_EQUALITY / TYPE_CANONICAL.
> > >             * c-obj-common.cc (get_alias_set): Remove special case.
> > >             (get_aka_type): Add special case.
> > > 
> > >     gcc/
> > >             * godump.cc (go_output_typedef): use TYPE_MAIN_VARIANT instead
> > >             of TYPE_CANONICAL.
> > > 
> > >     gcc/testsuite/
> > >             * gcc.dg/enum-alias-1.c: New test.
> > >             * gcc.dg/enum-alias-2.c: New test.
> > >             * gcc.dg/enum-alias-3.c: New test.
> > 
> > OK, in the absence of objections on middle-end or Go grounds within the
> > next week.
> 
> The godump.cc patch is
> 
>        && (TYPE_CANONICAL (TREE_TYPE (decl)) == NULL_TREE
>           || !container->decls_seen.contains
> -                                   (TYPE_CANONICAL (TREE_TYPE (decl)))))
> +                                   (TYPE_MAIN_VARIANT (TREE_TYPE (decl)))))
>      {
> 
> What is the problem you are seeing?

Test failures in godump-1.c

> 
> This patch isn't right:
> 
> 1) The code is saying if "X == NULL_TREE || !already_seen(X)".  This
> patch is changing the latter X but not the former.  They should be
> consistent.

Maybe the X == NULL_TREE can be removed if we
add TYPE_MAIN_VARIANTs instead?

> 
> 2) At the bottom of that conditional block is code that adds a value
> to container->decls_seen.  Today that code is adding TYPE_CANONICAL.
> If we change the condition to test TYPE_MAIN_VARIANT, then we need to
> add TYPE_MAIN_VARIANT to decls_seen.

Yes, obviously this is wrong. Thanks!

Martin
> 
> Hope that makes sense.
> 
> I don't know why the patch is required, but it's fine with those
> changes as long as the libgo tests continue to pass.


> 
> Ian