[Bug target/92361] [8/9 Regression] failing fortran libcgns test on powerpc64le-linux-gnu

[tkoenig at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92361

Thomas Koenig <tkoenig at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
   Last reconfirmed|                            |2019-11-04
     Ever confirmed|0                           |1

--- Comment #1 from Thomas Koenig <tkoenig at gcc dot gnu.org> ---
I am AFK at the moment, so I can not dig deeply right now.

However, the description raises one important flag: The vararg call.

There is no way that a vararg call to a Fortran procedure can be valid.
POWER just happens to have an ABI which exposes this.

The change in question was a fix for an ABI violation on POWER for
valid code, BTW.

If there is anything more, please let me know, otherwise I would resolve
this as INVALID and advise the authirs to fix the code.