Opencyc in Ubuntu?

[borgel]

I have been trying to run Opencyc (http://www.opencyc.org/) on Gutsy, but I keep getting a segmentation fault error ( Error: Stack Overflow (signaled by signal 11 SIGSEGV (Segmentation violation)). I didn't see an answer for this on Opencyc's forums so I thought I would check to see if anyone here has seen a solution for this problem.

[SuperTrooper]

EDIT: Ignore these instructions and look at my next post instead (the next post is much simpler)....

I too was pulling my hair out over OpenCyc for the last several days. It works dandy on Windows. But **I** absolutely do not work dandy under Windows. Here's what I did to get it working in Ubuntu (essentially, I did a chroot install of CentOS underneath Ubuntu - rationale is near the bottom):

1. Download CentOS 3 (http://www.centos.org/) (I got the install DVD image .iso via BitTorrent)

2. Install qemu:
sudo aptitude install qemu

3. Setup a disk image:
qemu-img create -f raw /tmp/centos3_fresh_image.img 4G

4. Run qemu:
qemu -hda /tmp/centos3_fresh_image.img -cdrom CentOS-3.8-i386-binDVD.iso -boot d

5. (follow the steps in the QEMU window to install CentOS)

6. Find the sector offset of the partition by using fdisk. Here is what my setup looked like:

fdisk -ul centos3_fresh_image.img
You must set cylinders.
You can do this from the extra functions menu.

Disk centos3_fresh_image.img: 0 MB, 0 bytes
255 heads, 63 sectors/track, 0 cylinders, total 0 sectors
Units = sectors of 1 * 512 = 512 bytes
Disk identifier: 0x00000000

Device Boot Start End Blocks Id System
centos3_fresh_image.img1 * 63 208844 104391 83 Linux
centos3_fresh_image.img2 208845 7871849 3831502+ 83 Linux
centos3_fresh_image.img3 7871850 8385929 257040 82 Linux swap / Solaris

The sector offset for the root partition is: 208845.
Each sector is 512 bytes, so the byte offset is:
208845 * 512 = 106928640 bytes.


7. Use the byte offset (from above) to do a loopback mount of the partition:

sudo su # you need to be root from this point on...
mkdir /mnt/loop
mount -o loop,offset=106928640 /tmp/centos3_fresh_image.img /mnt/loop/


8. Copy all the data needed for the chroot:

mkdir /home/centos_cyc_chroot
cp -a /mnt/loop/. /home/centos_cyc_chroot/.


9. Untar/unzip the Cyc data (we are almost finished!):

mkdir /home/centos_cyc_chroot/home/cyc
tar xzvf opencyc-1.0.2-linux.tgz -C /home/centos_cyc_chroot/home/cyc


10. Now, the chroot magic happens (this is where we pretend like we are running CentOS while still running Ubuntu/Gentoo/Slackware/whatever):

# WARNING: you still need to be root for this to work!
chroot /home/centos_cyc_chroot
adduser cyc
chown -Rf cyc:users /home/cyc
su cyc
cd /home/cyc/opencyc-1.0/scripts/
echo RH-ES3-x86_32 > platform-override.txt
./run-cyc.sh
# Then you wait a very long time for Cyc to build a world file
# WARNING: you **might** see an error like this:
# ./run-cyc.sh: line 63: 2600 Segmentation fault (core dumped) $CYC_EXE -w $CYC_WORLD
# HOWEVER! That error is **not** fatal - things are okay (at least in my case they were)

# Now, if you did get a "Segmentation fault", just run this command again
# (try it several times until it succeeds ... it might fight a little, but eventually
# it should succeed):
./run-cyc.sh

# You should see an output that looks like this:
# bin/opencyc-RH-ES3-x86_32.bin -w world/kb-5006-RH-ES3-x86_32-local.load
# Loading statics.
# Mapping in the memory image.
# Running initializers ... PROCESS (47032), STREAM (47038), STREAM (47053), PROCESS (47064), done.
# Reconnected to shared object RED Library (auto-loaded).
# Cyc 10 (Linux 32-bit)
# Copyright (C) 1995 - 2006 Cycorp, Inc. All rights reserved.
# RTL (Green Threads/ASM Contexts) initialized.
# CycL Initialized.
# The executable and world objects matched.
# Initializing HL backing store caches from units/5006/.
# Enabling base TCP services to port 3600.
#
# Ready for services.
# Process ID: 2681
# System 1.11058.2.20.2.3 KB 5006.
# CYC(1):


I was able to connect to http://localhost:3602/cgi-bin/cyccgi/cg?cb-start and get the Login page for Cyc. Super!

I tried repeating these steps with vanilla Ubuntu (i.e. by *not* chrooting to CentOS), and kept getting failures that looked like this:

1.
Mapping in the memory image.
./run-cyc.sh: line 63: 14473 Segmentation fault (core dumped) $CYC_EXE -w $CYC_WORLD


2.
Mapping in the memory image.
./run-cyc.sh: line 63: 14853 Illegal instruction (core dumped) $CYC_EXE -w $CYC_WORLD


3. This is half-way failure: I can use the CYC shell (by choosing the "0" option), but the web server (on port 3602) is not available:

Mapping in the memory image.
Running initializers ... PROCESS (47028), STREAM (47034), STREAM (47049), PROCESS (47060), done.
Reconnected to shared object RED Library (auto-loaded).
Cyc 10 (Linux 32-bit)
Copyright (C) 1995 - 2006 Cycorp, Inc. All rights reserved.
RTL (Green Threads/ASM Contexts) initialized.
CycL Initialized.
The executable and world objects matched.
Initializing HL backing store caches from units/5006/.
Enabling base TCP services to port 3600.

Ready for services.
Error: Received signal 7 SIGBUS (Bus error).
[Switching to single-threaded mode ....]

Select a restart:
0: Recursive read loop.
1: Exit program.
?


...well, so there you have it... If you or anyone else is brave enough to try all these steps, I would very much appreciate if at least one person could reply with a success or failure message so that these instructions can be modified if necessary.

EDIT: here is my system info, just in case:

uname -a
# Linux kikiliki 2.6.22-14-generic #1 SMP Tue Dec 18 08:02:57 UTC 2007 i686 GNU/Linux
RAM: 1 GB
Swap: 1 GB (maybe more would be recommended?)


EDIT: I also tried with Wine and CrossOver Linux, but failed, and here is a transcript of the failure:

dbaird@kikiliki (1229 03:32:07) scripts $ wine cmd
CMD Version 0.9.46

Z:\afs\goober.tld\user\dbaird\distfiles\tmp\opency c-win\opencyc-1.0\scripts>run-cyc.bat
<...lots of noisy messages get printed...>
Loading statics.
Mapping in the memory image.

Implementation error: Could not create file mapping

OS-Err: 87: Invalid parameter

[SuperTrooper]

Hey, even better news!!! I think Cyc was not getting along with the AFS filesystem (which contains my home directory).

So I moved off of AFS and then I was able to get Cyc running by doing the "run-cyc.sh" command (several times, because it crashes a lot) and eventually it just ran. Like magic!!

Here is just a handful of the error messages it gives:


Mapping in the memory image.
./run-cyc.sh: line 63: 3224 Illegal instruction (core dumped) $CYC_EXE -w $CYC_WORLD



Mapping in the memory image.
./run-cyc.sh: line 63: 3238 Segmentation fault (core dumped) $CYC_EXE -w $CYC_WORLD



./run-cyc.sh: line 63: 3252 Segmentation fault (core dumped) $CYC_EXE -w $CYC_WORLD



Mapping in the memory image.
bin/opencyc-RH-ES3-x86_32.bin: relocation error: bin/opencyc-RH-ES3-x86_32.bin: symbol munmap, version GLIBC_2.0 not defined in file libc.so.6 with link time reference


But that's okay! It will work if you keep trying (for me, it usually takes like 5-10 tries. Just don't run Cyc out of AFS! (this might apply to NFS also ... maybe?)

Btw, don't forget to do

echo RH-ES3-x86_32 > platform-override.txt


I'll give this one more fresh test just to make sure I am not lying to you.

EDIT: Okay! I just did one more fresh test! It works:

tar xzvf opencyc-1.0.2-linux.tgz -C /tmp
cd /tmp/opencyc-1.0/scripts
echo RH-ES3-x86_32 > platform-override.txt
./run-cyc.sh
# Surprisingly, it worked on *the first try*...
# Confused, I restarted it and tried again ... and it *worked again*
# So, I restarted nearly 10 times and it worked every try
# Weird. It mysteriously works great now (or so it seems.)


EDIT: Can someone else confirm this please?

[tzarabagliu]

I tried to run OpenCyc 1.0.2 on my Ubuntu 7.10 Gutsy Gibbon and got this error:

./run-cyc.sh: 28: [[: not found
Please run ./run-cyc.sh from opencyc-1.0/scripts
exit: 28: Illegal number: -1

it was because of the bad interpreter specified inside every script (/bin/sh which in gutsy points to the DASH shell)... See a solution to this problem here: https://wiki.ubuntu.com/DashAsBinSh

I changed the interpreter in each script to /bin/bash and it worked

[zerwas]

I would very much appreciate if at least one person could reply with a success or failure message so that these instructions can be modified if necessary.

SUCCESS.

Renaming the first line of all scripts from
#!/bin/sh
to
#! /bin/bash
(note the blank after "!") worked.

edit: But the initialization didn't work:
bin/opencyc-RH-ES3-x86_32.bin -w world/kb-5006-RH-ES3-x86_32-local.load
Loading statics.
Mapping in the memory image.
./run-cyc.sh: line 63: 14931 Segmentation fault $CYC_EXE -w $CYC_WORLD

edit2: Simply try to start OpenCYC again and again and it will work after the 5th or 6th time. It's working now, thanks for the help.

-- zerwas

[Big D]

tar -xvf opencyc
cd opencyc/scripts
echo SuSE-9.2-x86_64 > platform-override.txt
./run.sh

should work for 64 bit systems.

[FaceLeg]

tar -xvf opencyc
cd opencyc/scripts

Then renaming the first line of all scripts from
#!/bin/sh
to
SHELL = /bin/bash

Then
echo SuSE-9.2-x86_64 > platform-override.txt
./run.sh


Worked for me in 8.10

It loads first try and hasn't crashed once for me - 2007 Macbook.

[mjsantof]

I had same problem (stack overflow), because I hadn't enabled cgi on tomcat. This link might help.

http://cs.calstatela.edu/wiki/index.php/OpenCyc

[mjsantof]

I've just have it working (although in Debian, but shouldn't be different in Gutsy). Have you configure Tomcat to run cgi? as explained in the link I post above?

[omitt]

[QUOTE=FaceLeg;6128994]tar -xvf opencyc
cd opencyc/scripts

Then renaming the first line of all scripts from
#!/bin/sh
to
SHELL = /bin/bash
That seemed to do the trick. Now if only I can figure out how to open a local host.