[Bug ipa/87957] [9 Regression] ICE tree check: expected tree that contains ‘decl minimal’ structure, have ‘identifier_node’ in warn_odr, at ipa-devirt.c:1051 since r265519

[rguenth at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=87957

Richard Biener <rguenth at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Priority|P3                          |P1

--- Comment #21 from Richard Biener <rguenth at gcc dot gnu.org> ---
(In reply to Eric Botcazou from comment #20)
> > I guess the middle-end relies on TYPE_NAME to have the TYPE_MAIN_VARIANT
> > type rather than be qualified.  The question if just here and it is possible
> > to cope with that, or elsewhere too.
> 
> Yes, but only here and only for a couple of weeks, i.e. it's a novelty.

Enforced as novelty, yes, but it was my understanding all the time?

OTOH typedef variants can have different qualifiers so I guess it's either
TREE_TYPE (TYPE_NAME (t)) == t [or DECL_ORIGINAL_TYPE?] or
== TYPE_MAIN_VARIANT (t).  But I guess even that restriction is too strict
for variants of the typedef variants...

Still the bug morphed into sth else, the gnat regression.