This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
building gcc 2.95.2 fails on R4000 SGIs
- To: gcc-bugs at gcc dot gnu dot org
- Subject: building gcc 2.95.2 fails on R4000 SGIs
- From: Jesse Caldwell <Jesse dot Caldwell at Colorado dot EDU>
- Date: Sat, 29 Jan 2000 14:02:35 -0700
I guess your mail daemon doesn't like empty subject headers. Here
it is again.
Jesse
------- Forwarded Message
To: gcc-bugs@gcc.gnu.org
Date: Sat, 29 Jan 2000 11:26:32 -0700
From: Jesse Caldwell <Jesse.Caldwell@Colorado.EDU>
Hello,
I have to say I already feel guilty reporting a problem with a great
piece of software that's free. :)
I cannot successfully build gcc 2.95.2 on an old sgi architecture. I
need to have an older sgi running the latest version of irix with
the latest compilers so that I can build all of my software and have it
be binary compatible with all of the systems we support (obviously).
gcc is always the first thing to be built, and when it fails, I'm
really up the creek, and the problems I'm having are way over my head.
Here's what it spits out on an R4000 indigo2:
cardiff:~ > /usr/local/gnu/bin/make bootstrap
...
make[2]: Entering directory `/scratch/build-gcc/gcc/intl'
make[2]: Nothing to be done for `all'.
make[2]: Leaving directory `/scratch/build-gcc/gcc/intl'
./genrecog /scratch/gcc-2.95.2/gcc/config/mips/mips.md > tmp-recog.c
make[1]: *** [s-recog] Bus error (core dumped)
make[1]: Leaving directory `/scratch/build-gcc/gcc'
make: *** [all-gcc] Error 2
I'm using 'configure --prefix=/usr/local/gnu --disable-nls' to configure
gcc and I'm not building in the source directory (can't imagine any of
that is relevant...).
Now what really burns me is that I've built gcc successfully on a lot of
our newer sgi systems running the same os/compilers (all of this is done
with irix 6.5.6, cc 7.30). I can build it without problems on R5000 machines,
R10000, R12000, even the R4400 works fine. I've now tried to build on two
R4000 machines (what I need to have work) and it fails. One reports the
error above, the other one gave a fatal signal 10 error during the stage
two bootstrap.
If I can't get gcc built on one of these older systems, it means I'm going
to have to do a lot more local software builds than I'd like to do. Getting
gcc working on an R4000 machine would really save me. Are there any known
problems building on the R4000 architecture? Would it help if I provided
a local account on a system for you guys to muck with? I can't imagine
you guys love irix any more than anyone else, but I can provide an
account on just about any architecture you please if that would help you
(IP12 indigo running 5.3 up to a 32 cpu onyx).
Any assistance would be greatly appreciated,
Jesse
------- End of Forwarded Message