Hello,
I just got a segfault with tleap while trying to prepare a new system. It
didn't occur loading a library, strictly, but it was certainly related to
the library being loaded. It was a custom ligand for which I created an OFF
library file (saveoff). The library and the associated PDB file would each
load separately, but trying to load the PDB after the library gave a
segfault.
I'm going to presume it's a tleap bug (perhaps the same one manifesting
slightly differently?). The same commands worked just fine in sleap (which,
to further Scott's pending post, is written in C++ ;-) ). Since I was able
to get around the issue using the newer version of leap (which the website
says will eventually replace tleap...), I put off pursuing it, but I'll
throw it at a debugger and see what I get (I'm not an expert g/idb-er, but
I'll see what I can do).
As a note, the compiler I used was icc 11.1.069 . I've attached the files
that gave me an issue (loading just the lib, then pdb gave me the segfault).
All the best,
Jason
Uggh! gmail will not let me send a gzipped tarball for security reasons.
The file is accessible at
http://crunch.qtp.ufl.edu/~swails/Files/tleap_segfault.tgz
On Fri, May 14, 2010 at 3:58 PM, case <case.biomaps.rutgers.edu> wrote:
> A user (Omair Khan) has reported tleap segfaults in Fedora Core 12, with
> gcc
> version 4.4.3 20100127 (Red Hat 4.4.3-4).
>
> We found a person in the lab with a laptop using Fedora Core 12, and can
> reproduce what looks like the same error (segfault on loading in standard
> off files from the Amber library). Like the original poster, the owner of
> the laptop is not familiar with debugging problems like this, but can loan
> it next week to someone who can investigate more fully. But if any Amber
> guru
> has Fedora Core 12 (x86_64) [or wants to set up a virtual machine]
> you might also look to see what happens there. But now that we can
> reproduce
> the problem, we have a good shot at fixing it, presumably.
>
> [This note also gives Scott time to prepare his usual post about how this
> all could have been avoided if we had just waited a few years (way back
> when)
> for C++ to be developed into a useful language....]
>
> Perhaps related: Ubuntu 10.04 is also using gcc 4.4.3 (in this case
> Ubuntu 4.4.3-4ubuntu5). On a 32-bit virtual machine, this compiler/OS
> combo
> seems fine. So it is not strictly just the compiler version; Fedora Core
> 12
> seems to be exposing problems with our older codes.
>
> [One thing to look for: this may(!?!) be related to the bug found by
> Andreas
> Svrcek-Seiler with the same compiler:
>
> http://dev-archive.ambermd.org/201004/0148.html
>
> In that case, the culprit was strcpy() being used with overlapping memory,
> something that seems only to fail with Fedora, even though the standard
> says
> you can't rely on that.]
>
> cc-ing this to Andreas in case he knows -- do you see any tleap failures
> with your Fedora installation?
>
> ...thanks...dave
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
--
Jason M. Swails
Quantum Theory Project,
University of Florida
Ph.D. Graduate Student
352-392-4032
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri May 14 2010 - 15:00:03 PDT
