Re: [Libmesh-users] Runtime error in dbg mode on Mac OS X

[John P. <pet...@cf...>]

On Sat, Sep 4, 2010 at 11:29 AM, Anna-Lena Gerner
<ge...@ai...> wrote:
>
> Hi everybody,
>
> I installed libMesh on my Mac OS X, and in opt mode everything is just fine. There also seem to be no compilation problems in dbg mode. However, in dbg mode I get the following runtime error: (./ex4-dbg -d 1 -n 20)
>
> ex4-dbg(73819) malloc: *** error for object 0x1032e6fa0: pointer being freed was not allocated
> *** set a breakpoint in malloc_error_break to debug
>
> Program received signal SIGABRT, Aborted.
> 0x00007fff853623d6 in __kill ()
> (gdb) bt
> #0  0x00007fff853623d6 in __kill ()
> #1  0x00007fff85402972 in abort ()
> #2  0x00007fff8531a615 in free ()
> #3  0x0000000100026e0a in __gnu_cxx::new_allocator<char>::deallocate (this=0x7fff5fbfb5b7, __p=0x1032e6fa0 "") at ext/new_allocator.h:95
> #4  0x0000000100020a61 in std::string::_Rep::_M_destroy (this=0x1032e6fa0, __a=@0x7fff5fbfb61f) at basic_string.tcc:449
> #5  0x000000010001b0e0 in std::string::_Rep::_M_dispose (this=0x1032e6fa0, __a=@0x7fff5fbfb61f) at basic_string.h:236
> #6  0x000000010001351b in ~basic_string (this=0x7fff5fbfcad0) at basic_string.h:503
> #7  0x00000001016199c4 in libMesh::PerfLog::get_info_header (this=0x7fff5fbfd2f0) at src/utils/perf_log.C:171
> #8  0x000000010161b356 in libMesh::PerfLog::get_log (this=0x7fff5fbfd2f0) at src/utils/perf_log.C:452
> #9  0x000000010161b450 in libMesh::PerfLog::print_log (this=0x7fff5fbfd2f0) at src/utils/perf_log.C:469
> #10 0x0000000101619481 in ~PerfLog (this=0x7fff5fbfd2f0) at src/utils/perf_log.C:61
> #11 0x000000010000568b in assemble_poisson (es=@0x7fff5fbff570, system_name=@0x10350fc18) at ex4.C:290
> #12 0x00000001015bbb46 in libMesh::System::user_assembly (this=0x10350fb60) at src/systems/system.C:1451
> #13 0x00000001015b30a0 in libMesh::System::assemble (this=0x10350fb60) at src/systems/system.C:376
> #14 0x000000010151c0a8 in libMesh::ImplicitSystem::assemble (this=0x10350fb60) at src/systems/implicit_system.C:173
> #15 0x000000010152a394 in libMesh::LinearImplicitSystem::assemble (this=0x10350fb60) at linear_implicit_system.h:100
> #16 0x0000000101529e55 in libMesh::LinearImplicitSystem::solve (this=0x10350fb60) at src/systems/linear_implicit_system.C:86
> #17 0x000000010000392f in main (argc=6, argv=0x7fff5fbff880) at ex4.C:246

Whenever I see a segfault coming from a weird place, in this case,
from the destructor of std::string, I usually suspect an incomplete
recompile of some sort.  For example, object files from an older
compiler version getting linked with object files from a new one, etc.
 Can you see if 'make clobber' and remaking in dbg mode doesn't fix
the problem?

-- 
John