[petsc-users] KSPSolve crash

[Barry Smith]

  Best to run under valgrind to make sure there is not a memory access error that only kills the program sometimes.

  Barry

On Mar 19, 2014, at 4:05 PM, Qin Lu <lu_qin_2000 at yahoo.com> wrote:

> I made a few more tests. The weird thing is that the case runs in 2 win-7 workstations but failed in other 2 win-7 workstations. At the failing workstations it did not print out any call stack when it crashes (I tried with both release and debug PETSc libs), however, win-7 itself popped out a window that says "A problem caused the program to stop working correctly. Windows will close the program and notify you if a solution is available." Does this mean the program was trying to thing such as writing to reserved memory address?
>  
> Unfortunately, I can only run the program from command line in the failing workstations since I am not allowd to install VS debuger in them.
>  
> Thanks for any info,
>  
> Qin
> 
> From: "Balay, Satish" <balay at mcs.anl.gov>
> To: "Smith, Barry F." <bsmith at mcs.anl.gov>; Qin Lu <lu_qin_2000 at yahoo.com> 
> Cc: petsc-users <petsc-users at mcs.anl.gov> 
> Sent: Thursday, March 13, 2014 4:25 PM
> Subject: RE: [petsc-users] KSPSolve crash
> 
> Also run Linux version in valgrind
> From: Barry Smith
> Sent: ‎3/‎13/‎2014 2:22 PM
> To: Qin Lu
> Cc: petsc-users
> Subject: Re: [petsc-users] KSPSolve crash
> 
> 
>    Compile in debug mode and run in the debugger (visual studio has it built in).
> 
> On Mar 13, 2014, at 4:02 PM, Qin Lu <lu_qin_2000 at yahoo.com> wrote:
> 
> > Yes, it crashed without any printout. It always crashes at the first call to KSPSolve. Currently I only has a release version of petsc lib although my program has debug version.
> >  
> > Thanks,
> > Qin
> > 
> > 
> > ----- Original Message -----
> > From: Barry Smith <bsmith at mcs.anl.gov>
> > To: Qin Lu <lu_qin_2000 at yahoo.com>
> > Cc: petsc-users <petsc-users at mcs.anl.gov>
> > Sent: Thursday, March 13, 2014 3:57 PM
> > Subject: Re: [petsc-users] KSPSolve crash
> > 
> > 
> >   How did it crash? Absolutely nothing printed to the screen, the program just ended? Please send any output. Since it worked on linux it is likely something specific to the windows machine like lack of memory, a compiler bug, ….  Does it always crash at the same place, can you run it in the debugger on the windows machine and where does it end up?
> > 
> >   Barry
> > 
> > 
> > On Mar 13, 2014, at 2:45 PM, Qin Lu <lu_qin_2000 at yahoo.com> wrote:
> > 
> >> I forget to mention: it crashed in Win-7 only, while it runs fine in Linux.
> >>   
> >> Qin
> >> 
> >> 
> >> ----- Original Message -----
> >> From: Qin Lu <lu_qin_2000 at yahoo.com>
> >> To: petsc-users <petsc-users at mcs.anl.gov>
> >> Cc: 
> >> Sent: Thursday, March 13, 2014 2:43 PM
> >> Subject: [petsc-users] KSPSolve crash
> >> 
> >> PETSc team,
> >> 
> >> I have a program using PETSc linear solver (using KSPBCG with PCILU level 0). But with one case it crashed inside KSPSolve without any error message. I have tested this program with many other cases successfully.
> >> 
> >> Could you debug PETSc with the linear system? I can send you the matrix and rhs if you let me know where to upload them.
> >> 
> >> Many thanks,
> >> Qin