[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Making akcl on Sun 3/60 with SunOS4.0

> From: kthompso@thoreau.arc.nasa.gov (Kevin Thompson)
> Message-Id: <8901240118.AA17745@thoreau.arc.nasa.gov>
> To: kcl@cli.com
> Subject: Making akcl on Sun 3/60 with SunOS4.0
> I know I've seen other questions like this on this mailing list, , but I could
> find nothing in the archive that would indicate what my problem was here.
> I build akcl with gcc, and everything goes fine up until the second pass
> through Smakefile, when it first tries to invoke unixport/saved_kcl; I get the
> helpful message from UNIX:
>    Unrecoverable error: Someone allocated my memory!.
>    IOT trap
> raw_kcl seems perfectly happy (e.g. I can invoke it from the shell, and it
> works for what it is), and the messages I get when invoking "make saved_kcl"
> are pretty reasonable; a few warnings that things are being redefined.  But the
> created saved_kcl can't be invoked.  I don't know adb enough to fool with it.
> I am using akcl 1.86, copied just this past weekend.
> Help, via e-mail or postings, would be greatly appreciated.
> Kevin Thompson
> kthompso@[pluto|ptolemy].arc.nasa.gov     Sterling Software/Nasa-Ames RC

I built AKCL 1.86 on a VAX 8550 running Ultrix 2.3  using gcc -O.  While I 
didn't have the problem mentioned above, I did have problems... 
(oy, what problems I had....:-)
The problem turned out to be due to the gcc optimizer.   Try using
straight gcc (no optimizations in either the system building makefiles,
or in the internal compiler strings -- this will take some editing to remove
the "-O"s.))  My AKCL built and ran just fine under gcc with no optimizations.

By the way, the system didn't even build under the Ultrix standard cc without
optimization (seg faulted reading in the "initializing" phase of raw_kcl).
I didn't try it with optimization.