[Bug lto/58723] [4.9 Regression] ICE in lto_output_edge, at lto-cgraph.c:300 for OpenMP's simd reduction

[rguenth at gcc dot gnu.org]

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58723

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

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |hubicka at gcc dot gnu.org

--- Comment #3 from Richard Biener <rguenth at gcc dot gnu.org> ---
Looks like cgraph doesn't know about internal functions and treats them as
indirect calls ... I suppose we shouldn't record any call edges for internal
fns which all should expand to non-calls (and should all be 'leaf')?
Runs into interesting fallout though ...

The other variant is to forcefully build non-indirect edges, also with
interesting fallout.

Bah.

Honza?

Internal functions are like builtins just we don't have a decl (and thus
a cgraph node) for them.  They are just special instructions, not really
calls to "functions".