3.0.3 PATCH: Properly fix stdio.h on Tru64 UNIX V5.1A

[Rainer Orth]

Rainer Orth writes:

> Bruce Korb writes:
> 
> > It looks safe to me, but at least one person with a Tru64
> > platform ought to sanity check it.

Unfortunately nobody seems to have done so yet.  On the other hand, we have
now two PRs about this bootstrap failure (libf2c/4826 target/4962) and the
issue comes up on several mailing lists and in private mail.  Since the GCC
3.1 release is still several months away, and probably no 3.0.4 release
planned, we can expect many more in the meantime.

Several different users have meanwhile reported successfull bootstrap with
my patch.  So I'd really like to get this into GCC 3.0.3, especially
considering the difficulty some users reported to get autogen running on
Tru64 UNIX.

> Indeed, although I'm more worried that the patch might break other
> platforms than causing problems on Tru64 UNIX: __VA_LIST__ doesn't occur in
> stdio.h in V4.0F and V5.1, nor on Solaris 2.4, 2.5.1, 2.6, 8, or 9 Beta or
> IRIX 6.2 or 6.5.  That's all I can check here.
> 
> Wrt. Tru64 UNIX, I've bootstrapped on V5.1 (where the patch isn't needed)
> without regressions.  I'm awaiting more success reports (and hopefully
> testsuite results to prove there are no regressions) from V5.1A users.

My testsuite results are at

	http://gcc.gnu.org/ml/gcc-testresults/2001-12/msg00101.html

now.  One of my testers ran the testsuite on V5.1A; his results prove no
regressions compared to V5.1.  All I have is a make check output, so I had
to check this manually and cannot easily get it to gcc-testresults for
comparision.

Ok for mainline now?

Mark, ok for the branch as well?

FYI:
	http://gcc.gnu.org/ml/gcc-patches/2001-11/msg01735.html

	Rainer

-----------------------------------------------------------------------------
Rainer Orth, Faculty of Technology, Bielefeld University

Email: ro@TechFak.Uni-Bielefeld.DE