[Bug lto/44196] lto1: ICE: tree check: expected field_decl, have type_decl in gimple_types_compatible_p, at gimple.c:3597

["rguenther at suse dot de" <gcc-bugzilla at gcc dot gnu dot org>]

------- Comment #8 from rguenther at suse dot de  2010-05-19 15:13 -------
Subject: Re:  lto1: ICE: tree check: expected field_decl, have
 type_decl in gimple_types_compatible_p, at gimple.c:3597

On Wed, 19 May 2010, hubicka at ucw dot cz wrote:

> ------- Comment #7 from hubicka at ucw dot cz  2010-05-19 15:11 -------
> Subject: Re:  lto1: ICE: tree check: expected field_decl,
>         have type_decl in gimple_types_compatible_p, at gimple.c:3597
> 
> > I suppose this is an alias but we fail to clear its body/block tree.  We end
> > up refering to some TYPE_DECL in its BLOCK tree from somewhere else
> > (but we don't stream that function decl in the end).
> 
> For aliases we never clear their body/blocks because they are never finalized.
> If this is same_body_alias produced by C++ FE, I guess it should be responsible
> for giving us a decl alone.  Overall C++ FE is leaking quite a few bodies/trees
> in unfinalized decls.   I used to have aptch to clear them out but it was
> rejected
> for PCH reasons.

Ok, so my patch might be even correct as we are reaching the odd
TYPE_DECL via streaming block abstract origin from a different function.

Richard.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=44196