Update: Problem with Alpha OSF1 and mips-tfile

Thu Jul 19 14:41:00 GMT 2001

Antonio Ake writes:
> I tested gcc, g++ and they are working. I saw a very similar problem in
> this URL:
> http://gcc.gnu.org/ml/gcc-patches/1999-08/msg00322.html

Thanks alot for reminding me of this old patch of mine. I had just
stumbled across numerous libjava testsuite failures with mips-tfile errors
on Tru64 UNIX V5.1, but would probably have looked for bugs in mips-tfile
at first.
> I think I am having similar problem, the strange thing is that I have the
> patch in my jvspec.c.

While the patch is in, it got broken later by this one:
	http://gcc.gnu.org/ml/gcc-patches/1999-10n/msg00388.html
The insertion of main after %u and %U breaks all platforms that use
ASM_FINAL_SPEC, which refers to %g.s and knows nothing (and cannot/should
not need to know) about %umain.s, i.e. a suffix `main.s'. There can go
nothing between %[guU] and the suffix, since this breaks the ASM_FINAL_SPEC
`protocol'. Right now I'm bootstrapping with this patch reverted (breaking
-save-temps in the presence of --main again, but I hope this can be fixed
later once we have an idea for a proper solution) to get Tru64 UNIX support
running) and an additional patch to alpha/osf.h's ASM_FINAL_SPEC.
I've a patch for the problem of collect2 passing -fencoding=UTF-8 to cc1 as
well, so I hope Tru64 UNIX support will be in reasonable shape with those
patches. 
	Rainer
-----------------------------------------------------------------------------
Rainer Orth, Faculty of Technology, Bielefeld University
Email: ro@TechFak.Uni-Bielefeld.DE