This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: 3.0 PATCH: IRIX 5/6 O32 fixes
Mark Mitchell writes:
> Do your patches allow people to build a working IRIX5 compiler? What
I cannot really check that, since I don't have any IRIX 5 system around.
But since the toolchain from IRIX 5 to O32 IRIX 6 is unchanged, they should
be the same. Especially, the MIPSpro Assembly Language Programmer's Guide
explicitly states that .2byte and .4byte are only supported by the N32/N64
assembler, so the iris5.h patch will be necessary to make that
configuration at least compile.
> about a working O32 IRIX6 compiler? If they only get people part-way
> there, there's no point for 3.0. If they completely work, you're
I've checked the testsuite failures some more: except for many EH related
failures (even with sjlj), many g77 tests fail with an internal linker
error.
> right that they look very safe; it's hard to imagine them making
> anything worse. Please let me know.
Given the above, and the fact that it's unlikely that those problems are
fixed in the next two days, I'd agree that it's much better to document
that GCC 3.0 does not work on IRIX 5 (at least without gas, I may try this
separately, time permitting) or O32 IRIX 6. I suppose those problems can
be resolved either for 3.0.1 or 3.1, though.
What about mainline?
> Also, please test on IRIX6 with the N32/N64 compiler to make absolutely
> sure it is not affected by your change. That will be necessary before
> we can consider the patch.
Will do, together with the final form (irix6-hacks.c renamed to
irix6-libc-compat.c) of the N32/N64 structure passing workaround. I'm in
contact with rms about the legal status of that patch.
Rainer
-----------------------------------------------------------------------------
Rainer Orth, Faculty of Technology, Bielefeld University
Email: ro@TechFak.Uni-Bielefeld.DE