visualpython-users

[Visualpython-users] Linux : segmentation faults

[Lew R. <lr...@ur...>]

Hi Folks,

I'm running Fedora Core 2 (gcc version 3.3.3). I get my openGL libraries
from openmotif-2.2.3-4.1. According to my xorg.conf file, I have an "ATI
Radeon Mobility M9" video card.

I just installed Vpython-3.0, and the following demos work fine:

   convex.py
   crossproduct.py
   dipole.py
   faces_heightfield.py
   glinfo.py
   lathe.py
   texttest.py

However, any attempt to instantiate a box object leads to a segmentation 
fault with no prior error messages. I also see strange segmentation 
faults with the following demos:

   orbit.py

     ...dies via segmentation fault when either

     (a) appending a point to one of the curves. The curve to which it is
         appending (giant or dwarf) and the number of points at which it
         dies both vary.

     or

     (b) at the 'for a in [giant, dwarf]' before it executes any code in
         the block

   graphtest.py

     ...dies vie segmentation fault. Since most of the work is done in
     the C++ code, I don't know what's happening here.

   lorenz.py

     ...dies via segmentation fault apparently in the
     'for t in arrange(0,10,dt)'. Here again, the number of iterations it
     successfully completes varies. This is the most disturbing case,
     because it seems to point to problems with python itself(?).

   stars.py

     ...dies via segmentation fault. I haven't tried to track down where,
     but print statements before the various for loops inside the while
     show that it can die in different places.

Does any of this sound familiar? Any ideas?

Thanks,

Lew
-- 
___________________________________________________
Lew Riley        http://webpages.ursinus.edu/lriley
Ursinus College Department of Physics and Astronomy
(610) 409-3000 ext. 2293       (610) 409-3660 (FAX)

Re: [Visualpython-users] Linux : segmentation faults

[Jonathan B. <jbr...@ea...>]

On Sun, 2004-08-15 at 11:18, Lew Riley wrote:
> Hi Folks,
> 
> I'm running Fedora Core 2 (gcc version 3.3.3). I get my openGL libraries
> from openmotif-2.2.3-4.1. According to my xorg.conf file, I have an "ATI
> Radeon Mobility M9" video card.
> 
> I just installed Vpython-3.0, and the following demos work fine:

(snipped ... and other stuff crashes half-randomly)
> Does any of this sound familiar? Any ideas?

Uh-oh.  Yea, it sounds familiar.  I'm assuming that you have a laptop. 
Do you by chance have a Pentium M processor in it (not a Pentium 3-M or
Pentium 4-M)?

Please run one of the VPython programs that crashes in GDB (the GNU
Debugger) by following these instructions, and send me the console
output from the entire session.  I'm hoping that setting MESA_DEBUG=1 in
the environment will produce something useful, but it might not.

Start gdb:
`MESA_DEBUG=1 gdb python2.3`
At the "(gdb) " prompt type 'run'.
(gdb) run
At the Python ">>> " prompt, do the following:
>>> import sys
>>> sys.path.append(
'/usr/local/lib/python2.3/site-packages/visual/demos')

If you installed VPython somewhere other than /usr/local, then change
that part of the string above.
Run the demo program that crashes by typing "import programname"
_without_ the .py at the end.  ex:
>>> import orbit

When the program crashes, control will be returned to the (gdb) prompt. 
Run the "bt" command:
(gdb) bt

This produces the backtrace.  I have a bad feeling that the crash will
be in _mesa_test_os_sse_exception_support.


Thanks,
-Jonathan

P.S.  FYI, the graph module is written entirely in Python and rides on
top of Visual's basic functionality.