Subject: re: possible bug in gcc
To: matthew green <email@example.com>
From: M L Riechers <firstname.lastname@example.org>
Date: 03/17/2003 16:54:20
(My apologies to email@example.com, to whom I evidently
sent my last email. Please, just thow it away.)
> "likely use a gcc-3_3-branch compiler successfully" certainly perked
> my interest. Would you or someone else be so kind as to make a short
> outline of what I got to do to cvs or something a 3.3 for compile, and
> what's involved in making NetBSD makes and such happy for the gcc
> % cvs -d :pserver:firstname.lastname@example.org:/cvs/gcc login
> <answer "anoncvs">
> % cvs -d :pserver:email@example.com:/cvs/gcc co -r gcc-3_3-branch gcc
> see http://sources.redhat.com/
Thanks much, Matt. Following your instructions I successfully got the
3.3 branch, and (mostly, maybe?) compiled gcc.
Hooray! We're finally on our way!
However, the gcc build stopped at compiling or using g++ at, oh, what
the heck this:
/usr/local/powerpc-unknown-netbsd1.6/include -msoft- float -fPIC
-mstrict-align -g -O2 -msoft-float -fPIC -mstrict-align -msoft-float
-fPIC -mstrict-a lign -o abi_check abi_check.o
undefined reference to `__fixsfsi'
... and so on: the stuff defined in libc softfloat to link with gcc's
I'm not sure whether to rejoice or despair. I wasn't quite expecting
soft float support out of this build, but I'd be quite happy to get
it, (even optionally) since a full, complete, working soft float
system's where I'd like to be. (mpc860 cpu). I'm even a little
puzzled that I got soft-float, although maybe this is standard with
I'm toying with defining MKSOFTFLOAT and rebuilding libc, but I don't
know what effect that will have.
My system is 1.6:
uname -a gives
NetBSD t982.rse.com 1.6 NetBSD 1.6 (EASTERN-1.6.backside.2) #1: Sun
Oct 6 14:05:52 EDT 2002
Should I upgrade, or fiddle with what I have? I don't really want to
go to current, but I will if I have to.