I finally figured out the why "fparser" is crashing on our Linux machines.
It's been maddening, tracking this bug down since the behavior only
manifested itself in certain situations and in optimized mode only. It was
a "Heisenbug" since any attempt to observe the bug would modify it's
behavior. You know you have an issue when even Valgrind crashes with a
message "The impossible has happened - Valgrind has received a seg fault!"
and then continues to crash too along with the program it is grinding -
nice touch!
At any rate, it turns out there is a problem with the fparser Makefile
which I haven't identified yet but will keep looking. The problem is easy
to repeat on all of our Linux systems. If you "make clean" in fparser and
build, you can immediately type make again and it'll compile
"bytecodesynth.cc" once more and relink. Some part of the build process is
either modifying that file or not waiting for that file to compile before
linking which is causing the memory corruption issue. I haven't been able
to get the fparser to crash with pure libMesh code either which might be
way others aren't seeing it but it's definitely there.
Hopefully I can track it down tomorrow with a little more
Makefile sleuthing.
Cody
On Fri, Feb 10, 2012 at 3:59 PM, Cody Permann <codyperm...@gmail.com> wrote:
>
>
> On Fri, Feb 10, 2012 at 3:54 PM, Roy Stogner <royst...@ices.utexas.edu>wrote:
>
>>
>> On Tue, 7 Feb 2012, Roy Stogner wrote:
>>
>> > On Mon, 6 Feb 2012, Roy Stogner wrote:
>> >
>> >> On Mon, 6 Feb 2012, Kirk, Benjamin (JSC-EG311) wrote:
>> >>
>> >>> Anyone got a snippet showing how to use the contributed fparser?
>> >>>
>> >>> We'd love to add that capability to some application code here...
>> >>
>> >> I'm going to put it into an example, but probably not until after the
>> >> System::project_vector changes are committed. That latter should be
>> >> this afternoon or tonight unless anyone pipes up with a problem or
>> >> objection.
>> >
>> > A bit more delay on this - fparser isn't thread-safe by default, and
>> > the new project_vector functor interface doesn't play with it in a
>> > thread-safe way. For this and other reasons I'm going to break down
>> > and add a FunctionBase::clone() method.
>>
>> project_vector (and ExactSolution and ExactErrorEstimator) changes are
>> done, FunctionBase cloning's done, and I've put a snippet
>> demonstrating use (for a single-variable system) into adaptivity_ex5.
>>
>> It's working correctly for me, but now Cody's got me worried. Adding
>> an Optimize() call doesn't break anything on this end, but maybe
>> there's something about my compiler/system/trivial-example that isn't
>> triggering whatever problem they're seeing?
>>
>
> I wouldn't worry about it - we are using fparser in a wrapper class and we
> could easily be doing something wrong there. I'll check it out soon and
> let you know.
>
>
>> ---
>> Roy
>>
>>
>> ------------------------------------------------------------------------------
>> Virtualization & Cloud Management Using Capacity Planning
>> Cloud computing makes use of virtualization - but cloud computing
>> also focuses on allowing computing to be delivered as a service.
>> http://www.accelacomm.com/jaw/sfnl/114/51521223/
>> _______________________________________________
>> Libmesh-devel mailing list
>> Libmesh-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/libmesh-devel
>>
>
>
------------------------------------------------------------------------------
Keep Your Developer Skills Current with LearnDevNow!
The most comprehensive online learning library for Microsoft developers
is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3,
Metro Style Apps, more. Free future releases when you subscribe now!
http://p.sf.net/sfu/learndevnow-d2d
_______________________________________________
Libmesh-devel mailing list
Libmesh-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/libmesh-devel
