Re: [Help-gsl] gsl_multimin_fminimizer_free should deallocate gracefully

[Carl Boettiger]

I'm not seeing the attached code, but Id look at what size you are
passing for the parameter space in the multimin argument.  If you are
using dynamically allocated parameter space, you need to provide the
copy-constructor, copy, and destructor yourself (instead of the NULL
pointers given in the examples -- sorry if this is what you are doing,
just my guess from the error).  I got very similar stack trace in an
error in my own codes do to not providing my own copy functions for a
variable sized structure.  Hope this helps.  If not, maybe try
reposting the source code.

-Carl

On Tue, May 5, 2009 at 10:03 PM, kamaraju kusumanchi
<address@hidden> wrote:
> Consider the attached crash.cpp.
>
> It is compiled using
>
> $g++ -g -O0 crash.cpp -lgsl -lgslcblas
>
> When executed, it crashes with
>
> $./a.out
>
> initial x vector is
> 0.1  1.1  2.1  3.1  4.1  5.1  6.1  7.1  8.1  9.1  10.1  11.1  12.1
> 13.1  14.1  15.1  16.1  17.1  18.1  19.1  20.1  21.1  22.1  23.1  24.1
>  25.1  26.1  27.1  28.1  29.1  30.1  31.1  32.1  33.1  34.1  35.1
> 36.1  37.1  38.1  39.1  40.1  41.1  42.1  43.1  44.1  45.1  46.1  47.1
>  48.1  49.1
> initial step size vector is
> 0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2
> 0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2
> 0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2
> 0.2  0.2  0.2  0.2  0.2  0.2  0.2  0.2
> *** glibc detected *** ./a.out: free(): invalid pointer: 0x097b55b0 ***
> ======= Backtrace: =========
> /lib/i686/cmov/libc.so.6[0x403e6564]
> /lib/i686/cmov/libc.so.6(cfree+0x96)[0x403e8766]
> /usr/lib/libgsl.so.0(gsl_block_free+0x25)[0x4007bee5]
> /usr/lib/libgsl.so.0(gsl_vector_free+0x2c)[0x401b88bc]
> /usr/lib/libgsl.so.0(gsl_multimin_fminimizer_free+0x3b)[0x401246eb]
> ./a.out(__gxx_personality_v0+0x457)[0x8048f07]
> /lib/i686/cmov/libc.so.6(__libc_start_main+0xe6)[0x4038e456]
> ./a.out(__gxx_personality_v0+0x31)[0x8048ae1]
> ======= Memory map: ========
> 08048000-0804a000 r-xp 00000000 08:08 2506337    
> /home/rajulocal/debug/gsl/a.out
> 0804a000-0804b000 rw-p 00001000 08:08 2506337    
> /home/rajulocal/debug/gsl/a.out
> 097b5000-097d6000 rw-p 097b5000 00:00 0          [heap]
> 40000000-4001a000 r-xp 00000000 08:07 163575     /lib/ld-2.7.so
> 4001a000-4001c000 rw-p 0001a000 08:07 163575     /lib/ld-2.7.so
> 4001c000-4001d000 r-xp 4001c000 00:00 0          [vdso]
> 4001d000-40020000 rw-p 4001d000 00:00 0
> 40038000-40214000 r-xp 00000000 08:07 1756194    /usr/lib/libgsl.so.0.13.0
> 40214000-4021e000 rw-p 001dc000 08:07 1756194    /usr/lib/libgsl.so.0.13.0
> 4021e000-40255000 r-xp 00000000 08:07 1756196    /usr/lib/libgslcblas.so.0.0.0
> 40255000-40256000 rw-p 00036000 08:07 1756196    /usr/lib/libgslcblas.so.0.0.0
> 40256000-40257000 rw-p 40256000 00:00 0
> 40257000-4033a000 r-xp 00000000 08:07 1748044    /usr/lib/libstdc++.so.6.0.10
> 4033a000-4033d000 r--p 000e2000 08:07 1748044    /usr/lib/libstdc++.so.6.0.10
> 4033d000-4033f000 rw-p 000e5000 08:07 1748044    /usr/lib/libstdc++.so.6.0.10
> 4033f000-40345000 rw-p 4033f000 00:00 0
> 40345000-40369000 r-xp 00000000 08:07 163291     /lib/i686/cmov/libm-2.7.so
> 40369000-4036b000 rw-p 00023000 08:07 163291     /lib/i686/cmov/libm-2.7.so
> 4036b000-40377000 r-xp 00000000 08:07 163613     /lib/libgcc_s.so.1
> 40377000-40378000 rw-p 0000b000 08:07 163613     /lib/libgcc_s.so.1
> 40378000-404cd000 r-xp 00000000 08:07 163287     /lib/i686/cmov/libc-2.7.so
> 404cd000-404ce000 r--p 00155000 08:07 163287     /lib/i686/cmov/libc-2.7.so
> 404ce000-404d0000 rw-p 00156000 08:07 163287     /lib/i686/cmov/libc-2.7.so
> 404d0000-404d4000 rw-p 404d0000 00:00 0
> 40500000-40521000 rw-p 40500000 00:00 0
> 40521000-40600000 ---p 40521000 00:00 0
> bff91000-bffa6000 rw-p bffeb000 00:00 0          [stack]
> Aborted (core dumped)
>
>
>
> I believe this happens because gsl_vector x is already deallocated
> before the corresponding vector in gsl_multimin_fminimizer s is
> deallocated. Is this considered a bug in my code or a bug in gsl? In
> any case, It would be nice if cases like these are handled more
> gracefully.
>
> regards
> raju
>
>
> _______________________________________________
> Help-gsl mailing list
> address@hidden
> http://lists.gnu.org/mailman/listinfo/help-gsl
>



-- 
Carl Boettiger
Population Biology, UC Davis
http://two.ucdavis.edu/~cboettig