This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

RE: Problems building egcs-1.1.2 on NCR MP-RAS 3.02


Hiya Robert,

Thnx for the response.

> > regarding problems I'm having building egcs-1.1.2 on an NCR
> 4300 running NCR
> > MP-RAS 3.02.  Apparently this might be the best place to post
> my SOS's (thnx
>
> Is the MP-RAS that's NCR's own port, NCR-ified UnixWare, or
> NCR-ified Solaris?

Apparently neither...this is NCR's v.own port...

> I lost track of NCRisms long ago.

..I wish I had ;-)

> > ....egcs-1.1.2 still reliably core dumps during the build whilst running
> > cc1obj.  Tracing cc1obj whilst it runs I believe its trying to
> run something
> > at an illegal address.
>
> It will be more productive to stick the offending program (cc1obj) in a
> debugger and see what's going on.

I have gdb available (NCR seem to bundle this as a standard utility within
their Source Code Development Environment (SCDE) package) and I did give it
a go.  However, gdb appears to me to be hanging when trying to run against
cc1obj, 'cause when I 'run' the program from within gdb, I never get the
prompt back:

Reading symbols from cc1obj...done.

Breakpoint 1 at 0x8048d10
Breakpoint 2 at 0x8048bf0
Breakpoint 3 at 0x80d041b
(gdb) r -print-objc-runtime-info tmp-runtime
Starting program:
/tooldir/src/egcs-1.1.2/gcc/cc1obj -print-objc-runtime-info tmp-runtime

...and looking at the process table, I seem to get a 'defunct' child
process:

root  1886 18125  0 16:32:54 pts/15   0:01 gdb cc1obj
root  1921  1886  0                   0:00 <defunct>

...uhhhmmm.  I must profess to using gdb only occassionally, but I'm pretty
certain (unless you can enlighten me otherwise) that things are a bit
skewiff.

> > So my question is now be what's changed from gcc-2.7.2 to
> egcs-1.1.2 that's
> > stopping me completing the build?
>
> Billions of lines of code.  Looking at the diffs is unlikely to be
> productive.  Analyze the bug before you.

Okay.

> You're the first MP-RAS user I recall hearing from in this group.

...and v.probably the last ;-)

> It isn't a system well tested.  We have had reports of success on
> Solaris/x86 and on UnixWare so other SVR4/x86's are known to work.
> http://egcs.cygnus.com/egcs-1.1/buildstat.html lists the systems.  (It
> doesn't explictly cite UnixWare 2, but I'm pretty sure that Rodney Brown
> reports that works.)

..except I don't think that we're dealing with Unixware2 here....I hope that
doesn't mean I've just blown away your help?

> > However, I'm still hitting problems.  After running CC=gcc CXX=g++
> > MAKE=gmake
> > ../configure --host=i586-ncr-sysv4.3.03, and then gmake bootstrap (as
> > suggested by the distribution), The build fails here:
> >
> > ../cc1obj -print-objc-runtime-info tmp-runtime >>objc/runtime-info.h
> > Segmentation Fault - core dumped
> > gmake[2]: *** [objc/runtime-info.h] Error 139
>
> Which stage is this in?  My (admittedly ancient) experience with MP-RAS
> is that it shipped with an, errr, less than robust system compiler.
> If it's miscompiling stage1 during the bootstrap your life will be
> hard. Maybe they've fixed the native tools in recent years.

I'm in stage1.

I've tried with both NCR's native compilers (which nowadays are produced by
Metaware Incorporated), and gcc-2.7.2 & gmake3.77 (both of which build
okay).  Currrently I'm not using the GNU linker, but I've just downloaded
binutils2.9.1, so I'm going to try that next....although I can't see it
helping much....straws & clutching spring to mind.

> The rest of your message focused on a different set of problems.  Please
> pick the first problem during a bootstrap and let's work from there.

Yeah...soz about that, I simply cut & paste a whole thread of postings.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]