[Bug lto/88112] [9 regression] ICE in lto1: TYPE_FIELDS defined in incomplete type

[rguenther at suse dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88112

--- Comment #12 from rguenther at suse dot de <rguenther at suse dot de> ---
On Wed, 21 Nov 2018, ebotcazou at gcc dot gnu.org wrote:

> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88112
> 
> --- Comment #11 from Eric Botcazou <ebotcazou at gcc dot gnu.org> ---
> (> So short of avoiding the streaming of trees that are only targets
> > of abstract origins (the idea of streaming those as references to
> > DIEs rather than references to trees) a pragmatic "fix" is to
> > simply allow CALL_EXPRs to be streamed which fixes this testcase
> 
> This sounds heavy though.  If the C++ front-end cannot be changed to resetting
> the problematic size expressions, then maybe restrict your PR lto/87229 change
> to local types.

The type verifier will still trip over I guess.  The patch to stream
CALL_EXPRs isn't big, the code is mostly in place, a bit bitrotten
though.  I think it's probably the way to go for GCC 8 at least.
Will post it in a second.