I've tried a few other things (building without javadx, building with SGI's
compilers instead of gcc) and still get the same result. The errors all
seem to be memory related (or at least libc-related). I got segmentation
violations in
strcmp in src/exec/dpexec/d.c, line 299
malloc in src/exec/libdx/memory.c, line 1943
and bus errors (several times) from src/exec/dpexec/d.c, line 260. The
dictionary search and purge routines (and the memory allocation/freeing
associated with them) seem to do nasty things to my SGI systems. The CVS
version seemed to work fine until about 2 weeks ago. Since then I've
rebuilt opendx at least 30 times (using different compilers, -o32 and -n32,
Motif-1.2 and Motif-2.1, etc.) and haven't been able to find anything
stable. Until then, the only sample program that routinely caused trouble
was the AutoColor.net (with the water molecule). My systems have dumped
core on that from the first time I downloaded the source back in May!
I'll try rebuilding it with optimization turned off and see if that makes a
difference. Any other suggestions are welcome.
****************************************************************************
***
Alan M. Ferrenberg ([EMAIL PROTECTED])
Manager, Research and Computational Science Support
UGA Computing and Networking Services
Get the latest information from UCNS! Subscribe to our weekly e-mail
publication "UCNS Weekly News". For more information, check out:
http://www.uga.edu/ucns/consult/casnews/
****************************************************************************
***
----- Original Message -----
From: "David L. Thompson" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Wednesday, February 09, 2000 12:04 PM
Subject: Re: [opendx-dev] Bug in 4.0.10
On LinuxPPC, I'm having coredumps as well with the dxexec. All of my
crashes seem to be coming from the memory routines. That is why I
started to look at the memory source code. I did notice that if I
turned the optimization off, the dumps were less frequent--you might
try this. If you can try and document the locations where it dumps
when it happens, it might be useful to me in tracking down what is
going on.
As for the Motif 2.1 stuff. Rick Scott, our lesstif god, and Suhaib,
the cygwin god, had commented that there were going to probably be
problems with Motif 2.1. A lot of the Motif stuff had been coded by
someone with knowledge of the internals of Motif. Please file a bug
report about the control panel crashing a system with Motif 2.1.
Maybe somebody will have time to look at it.
David
>I had a similar problem and resolved it by uninstalling the 2.1 version of
>motif_dev. Unfortunately, I'm having other bad problems with 4.0.10 and
>6.5.6f. If I run some of the samples, I eventually get dxexec crashing
with
>a segmentation violation (sometimes in strcmp in src/exec/dpexec/d.c and
>sometimes in other places). The common thread between them is that they're
>all happening in routines in libc.so. Some info on the build: I used gcc
>2.95.2, mips4 ISA and n32 ABI.
>
>I've also built an o32, mips2 version for some older systems (using an
older
>o32-enabled gcc). Some of the samples crash it, but I've had more luck
with
>it.
>
>Anyone have an idea what could be causing these crashes? I've been beating
>on this for a week or so with no luck.
>
>***************************************************************************
*
>***
>Alan M. Ferrenberg ([EMAIL PROTECTED])
>Manager, Research and Computational Science Support
>UGA Computing and Networking Services
>
>Get the latest information from UCNS! Subscribe to our weekly e-mail
>publication "UCNS Weekly News". For more information, check out:
>
>http://www.uga.edu/ucns/consult/casnews/
>
>***************************************************************************
*
>***
>
--
............................................................................
.
David L. Thompson The University of Montana
mailto:[EMAIL PROTECTED] Computer Science Department
http://www.cs.umt.edu/u/dthompsn Missoula, MT 59812
Work Phone : (406)257-8530
