Request for Assistance with free(): invalid next size (fast) Error in Material Class

[vikramroybarc]

Hi Moose Team,

I am encountering a free(): invalid next size (fast) error in one of my material classes within the Moose application.

The function computeQpStress() successfully executes for each of the 8 _qps, which I verified using the VSCode GDB debugger. Afterwards, computeQpStress() executes successfully again for all 8 _qps.The output up to that point looks like this:

Time Step 1, time = 5e-06, dt = 5e-06  
Currently Executing  
    Computing Pre-SMO Residual  
      Computing Residual                                                                 [ 24.14 s] [  208 MB]  
Pre-SMO residual: 8e-08  
 Skipping predictor this step  
    Finished Computing Pre-SMO Residual                                                  [ 24.14 s] [  208 MB]  
    Computing Residual.....                                                              [ 39.61 s] [  208 MB]  

However, the issue arises when the code reaches the following section in NonlinearSystem.C (located at /framework/src/systems/NonlinearSystem.C, lines 190–197):

if (time_integrator_solve)  
  mooseAssert(_time_integrators.size() == 1,  
              "If solve is overridden, then there must be only one time integrator");  

if (time_integrator_solve)  
  _time_integrators.front()->solve();  
else  
  system().solve();  

The moment the code executes system().solve();, the application crashes with the following error:

free(): invalid next size (fast) error  

Single stepping until exit from function libmesh_petsc_snes_residual,  
which has no line number information.  
Single stepping until exit from function SNESComputeFunction,  
which has no line number information.  
Single stepping until exit from function SNESSolve_NEWTONLS,  
which has no line number information.  

Thread 1 "rhocp-dbg" received signal SIGABRT, Aborted.  
__pthread_kill_implementation (no_tid=0, signo=6, threadid=<optimized out>) at ./nptl/pthread_kill.c:44  

Notably, other material classes in my Moose application are functioning correctly, and I’m unable to pinpoint the root cause of this issue.

I would greatly appreciate your guidance on troubleshooting and resolving this problem.

Thank you for your help!

Best regards,
Vikram Roy

[GiudGiud]

Hello

If you are already running this using a debugger, can you print a backtrace? You use a breakpoint on the error (if you're lucky b MPI_Abort should suffice), then the execution stops there and you can use bt to get a full backtrace.

You ll want to compile in debug mode to get the line number

[vikramroybarc]

Hi Giud,

Thankyou for reply.

This is the backtrace, I get from the GDB Debugger:

#0  __pthread_kill_implementation (no_tid=0, signo=6, threadid=<optimized out>) at ./nptl/pthread_kill.c:44
#1  __pthread_kill_internal (signo=6, threadid=<optimized out>) at ./nptl/pthread_kill.c:78
#2  __GI___pthread_kill (threadid=<optimized out>, signo=signo@entry=6) at ./nptl/pthread_kill.c:89
#3  0x00007d3506a4526e in __GI_raise (sig=sig@entry=6) at ../sysdeps/posix/raise.c:26
#4  0x00007d3506a288ff in __GI_abort () at ./stdlib/abort.c:79
#5  0x00007d3506a297b6 in __libc_message_impl (fmt=fmt@entry=0x7d3506bce8d7 "%s\n") at ../sysdeps/posix/libc_fatal.c:132
#6  0x00007d3506aa8fe5 in malloc_printerr (str=str@entry=0x7d3506bd1c20 "free(): invalid next size (fast)") at ./malloc/malloc.c:5772
#7  0x00007d3506aab3dc in _int_free (av=0x7d3506c03ac0 <main_arena>, p=<optimized out>, have_lock=0) at ./malloc/malloc.c:4590
#8  0x00007d3506aadd9e in __GI___libc_free (mem=0x64d6df2408c0) at ./malloc/malloc.c:3398
#9  0x00007d350d177fef in PetscFreeAlign () from /home/vikramroy/miniforge/envs/moose/petsc/lib/libpetsc.so.3.21
#10 0x00007d350d1528ea in PetscFunctionListDestroy () from /home/vikramroy/miniforge/envs/moose/petsc/lib/libpetsc.so.3.21
#11 0x00007d350d182eaa in PetscHeaderDestroy_Private () from /home/vikramroy/miniforge/envs/moose/petsc/lib/libpetsc.so.3.21
#12 0x00007d350d1830e8 in PetscHeaderDestroy_Function () from /home/vikramroy/miniforge/envs/moose/petsc/lib/libpetsc.so.3.21
#13 0x00007d350d1cf1c9 in ISDestroy () from /home/vikramroy/miniforge/envs/moose/petsc/lib/libpetsc.so.3.21
#14 0x00007d3514e0d5d2 in libMesh::WrappedPetsc<_p_IS*>::destroy() () from /home/vikramroy/miniforge/envs/moose/libmesh/lib/libmesh_dbg.so.0
#15 0x00007d3514de20ba in libMesh::WrappedPetsc<_p_IS*>::~WrappedPetsc() () from /home/vikramroy/miniforge/envs/moose/libmesh/lib/libmesh_dbg.so.0
#16 0x00007d3514df7807 in libMesh::PetscVector<double>::localize(libMesh::NumericVector<double>&, std::vector<unsigned long, std::allocator<unsigned long> > const&) const () from /home/vikramroy/miniforge/envs/moose/libmesh/lib/libmesh_dbg.so.0
#17 0x00007d351514f3be in libMesh::System::update() () from /home/vikramroy/miniforge/envs/moose/libmesh/lib/libmesh_dbg.so.0
#18 0x00007d35150a0501 in libmesh_petsc_snes_jacobian () from /home/vikramroy/miniforge/envs/moose/libmesh/lib/libmesh_dbg.so.0
#19 0x00007d350da4164e in SNESComputeJacobian () from /home/vikramroy/miniforge/envs/moose/petsc/lib/libpetsc.so.3.21
#20 0x00007d350da1867b in SNESSolve_NEWTONLS () from /home/vikramroy/miniforge/envs/moose/petsc/lib/libpetsc.so.3.21
#21 0x00007d350da46065 in SNESSolve () from /home/vikramroy/miniforge/envs/moose/petsc/lib/libpetsc.so.3.21
#22 0x00007d35150a4d9e in libMesh::PetscNonlinearSolver<double>::solve(libMesh::SparseMatrix<double>&, libMesh::NumericVector<double>&, libMesh::NumericVector<double>&, double, unsigned int) () from /home/vikramroy/miniforge/envs/moose/libmesh/lib/libmesh_dbg.so.0
#23 0x00007d35151426a8 in libMesh::NonlinearImplicitSystem::solve() () from /home/vikramroy/miniforge/envs/moose/libmesh/lib/libmesh_dbg.so.0
#24 0x00007d351b986990 in NonlinearSystem::solve (this=0x64d6deaf30f0) at /home/vikramroy/projects/moose/framework/src/systems/NonlinearSystem.C:197
#25 0x00007d35196378af in FEProblemBase::solve (this=0x64d6dead01a0, nl_sys_num=0) at /home/vikramroy/projects/moose/framework/src/problems/FEProblemBase.C:6150
#26 0x00007d351a52537c in FEProblemSolve::solve (this=0x64d6deb52ed8) at /home/vikramroy/projects/moose/framework/src/executioners/FEProblemSolve.C:291
#27 0x00007d351a52a87e in FixedPointSolve::solveStep (this=0x64d6deb53520, begin_norm=@0x64d6df9a4640: 0, end_norm=@0x64d6df9d30e0: 0, transformed_dofs=...) at /home/vikramroy/projects/moose/framework/src/executioners/FixedPointSolve.C:439
#28 0x00007d351a52970a in FixedPointSolve::solve (this=0x64d6deb53520) at /home/vikramroy/projects/moose/framework/src/executioners/FixedPointSolve.C:291
#29 0x00007d351a56f4dc in TimeStepper::step (this=0x64d6de2c3a30) at /home/vikramroy/projects/moose/framework/src/timesteppers/TimeStepper.C:168
#30 0x00007d351a539b72 in Transient::takeStep (this=0x64d6deb52b20, input_dt=-1) at /home/vikramroy/projects/moose/framework/src/executioners/Transient.C:418
#31 0x00007d351a5385f2 in Transient::execute (this=0x64d6deb52b20) at /home/vikramroy/projects/moose/framework/src/executioners/Transient.C:293
#32 0x00007d351afda3a2 in MooseApp::executeExecutioner (this=0x64d6de300e20) at /home/vikramroy/projects/moose/framework/src/base/MooseApp.C:1179
#33 0x00007d351afe1772 in MooseApp::run (this=0x64d6de300e20) at /home/vikramroy/projects/moose/framework/src/base/MooseApp.C:1561
#34 0x000064d6d6bafe94 in Moose::main<RhocpApp> (argc=4, argv=0x7ffd19261de8) at /home/vikramroy/projects/moose/framework/build/header_symlinks/MooseMain.h:47
#35 0x000064d6d6baf6fd in main (argc=4, argv=0x7ffd19261de8) at /home/vikramroy/projects/rhocp-vr/src/main.C:26

[GiudGiud]

Is it stateful? Does it use old and older values of material properties?

if so try setting snesmf_reuse_base=false in the executioner. It's just a hunch

[vikramroybarc]

Yes, it uses old values of material properties

  // Recover "old" state variables
  std::copy(_state_var_old[_qp].begin(), _state_var_old[_qp].end(), _state_var[_qp].begin());
  std::copy(_properties_old[_qp].begin(), _properties_old[_qp].end(), _properties[_qp].begin());

  // Recover "old" miller indices
  std::copy(_y_old[_qp].begin(), _y_old[_qp].end(), _y[_qp].begin());
  std::copy(_z_old[_qp].begin(), _z_old[_qp].end(), _z[_qp].begin());

Also,

I tried using the option snesmf_reuse_base=false, still it results in the same error:

Time Step 1, time = 5e-06, dt = 5e-06
Pre-SMO residual: 8e-08
 Skipping predictor this step
munmap_chunk(): invalid pointer
Aborted (core dumped)

[GiudGiud]

are these old variable values?

  std::copy(_state_var_old[_qp].begin(), _state_var_old[_qp].end(), _state_var[_qp].begin());

[vikramroybarc]

These are old state variables

[dschwen]

Make sure the current properties are sized correctly before copying. std::copy does not allocate space.