BFT memory problem

[Alexandre Lebouvier]

Hi,
 
I use saturne v2. I ran a simple case and I got this error message :

Rsum de l'allocation mmoire
------------------------------

Mmoire dynamique instrumente thorique : 147976 ko
Mmoire dynamique instrumente thorique : 173147 ko
Mmoire programme mesure maximale:        333524 ko
Mmoire programme mesure courante:        326172 ko

/usr/local/saturne/bft-1.1.1/src/bft_mem.c:402: Erreur fatale.

Adress [0x7fff2834dcfc] does not correspond to the beginning of an allocated block.

Pile d'appels :
   1: 0x2b3fd47350c5 ?                                (?)
   2: 0x2b3fd785b6ba ?                                (?)
   3: 0x2b3fd785b764 ?                                (?)
   4: 0x2b3fd785b7c7 ?                                (?)
   5: 0x2b3fd785c11f <bft_mem_free+0x42>              (libbft.so.1)
   6: 0x2b3fd47e82ce <pstvar_+0x167b>                 (libsaturne.so.0)
Fin de la pile

Is anyone know where does it could be from ?

Thanks

Alexandre

[Yvan Fournier]

Hello,
Are you using user subroutines ? This is probably a memory corruption issue. If you have user subroutine, chances are there is a bug in them. Otherwise, the bug would seem to be in the code itself, or due to some build issue.
You should try to run under Valgrind (if Valgrind is installed on you system, just set VALGRIND=valgrind in the runcase script) and check the output. Valgrind is usally installable as a package on just about any Linux distribution. The code will run 20x on more slowly and will require 10x as much memory, but if there is a memory corruption issue, it will probably help locate it.
Best regards,
  Yvan
 

[Alexandre Lebouvier]

Hello,

Thanks for the answer. I don't use the GUI and my case is only composed of usini1.f90 and usclim.f90.

Here is the results after using valgrind command :

Single processor Code_Saturne simulation
                      Code_Saturne is running
                      ***********************

 Working directory (to be periodically cleaned) :
     /b1/alexa/tmp_Saturne/TEST_CS2.DEBUG.01110922
 Kernel version:           /usr/local/saturne/cs-2.0-beta2
 Preprocessor version:     /usr/local/saturne/cs-2.0-beta2

  ***************************************************************
   Compilation of user subroutines and linking of Code_Saturne
  ***************************************************************

  ********************************************
             Preparing calculation
  ********************************************

  ********************************************
             Starting calculation
  ********************************************

==16261== Memcheck, a memory error detector
==16261== Copyright (C) 2002-2009, and GNU GPL'd, by Julian Seward et al.
==16261== Using Valgrind-3.5.0 and LibVEX; rerun with -h for copyright info
==16261== Command: /b1/alexa/tmp_Saturne/TEST_CS2.DEBUG.01110922/cs_solver
==16261==
==16261== Conditional jump or move depends on uninitialised value(s)
==16261==    at 0x3204407607: ??? (in /lib64/libgcc_s-4.1.2-20080825.so.1)
==16261==    by 0x32044088BA: _Unwind_Backtrace (in /lib64/libgcc_s-4.1.2-20080825.so.1) ==16261==    by 0x31FD8E61F7: backtrace (in /lib64/libc-2.5.so)
==16261==    by 0x7E1A6F9: bft_backtrace_create (bft_backtrace.c:141)
==16261==    by 0x4CF70E4: _cs_base_backtrace_print (cs_base.c:499)
==16261==    by 0x7E1AEFF: bft_backtrace_print (bft_backtrace.c:458)
==16261==    by 0x4CF6F9A: _cs_base_gestion_erreur (cs_base.c:411)
==16261==    by 0x4CF70C4: _cs_mem_error_handler (cs_base.c:486)
==16261==    by 0x7E1D6B9: _bft_mem_error (bft_mem.c:373)
==16261==    by 0x7E1D763: _bft_mem_block_info (bft_mem.c:402)
==16261==    by 0x7E1D7C6: _bft_mem_block_size (bft_mem.c:429)
==16261==    by 0x7E1E11E: bft_mem_free (bft_mem.c:907)
==16261== ==16261== Conditional jump or move depends on uninitialised value(s)
==16261==    at 0x31FD8E6213: backtrace (in /lib64/libc-2.5.so)
==16261==    by 0x7E1A6F9: bft_backtrace_create (bft_backtrace.c:141)
==16261==    by 0x4CF70E4: _cs_base_backtrace_print (cs_base.c:499)
==16261==    by 0x7E1AEFF: bft_backtrace_print (bft_backtrace.c:458)
==16261==    by 0x4CF6F9A: _cs_base_gestion_erreur (cs_base.c:411)
==16261==    by 0x4CF70C4: _cs_mem_error_handler (cs_base.c:486)
==16261==    by 0x7E1D6B9: _bft_mem_error (bft_mem.c:373)
==16261==    by 0x7E1D763: _bft_mem_block_info (bft_mem.c:402)
==16261==    by 0x7E1D7C6: _bft_mem_block_size (bft_mem.c:429)
==16261==    by 0x7E1E11E: bft_mem_free (bft_mem.c:907)
==16261==    by 0x4DAA2CD: pstvar_ (cs_post.c:1815)
==16261==
==16261==
==16261== HEAP SUMMARY:
==16261==     in use at exit: 151,583,670 bytes in 284 blocks
==16261==   total heap usage: 3,185 allocs, 2,901 frees, 1,178,319,265 bytes allocated ==16261==
==16261== LEAK SUMMARY:
==16261==    definitely lost: 0 bytes in 0 blocks
==16261==    indirectly lost: 0 bytes in 0 blocks
==16261==      possibly lost: 0 bytes in 0 blocks
==16261==    still reachable: 151,583,670 bytes in 284 blocks
==16261==         suppressed: 0 bytes in 0 blocks
==16261== Rerun with --leak-check=full to see details of leaked memory
==16261==
==16261== For counts of detected and suppressed errors, rerun with: -v
==16261== Use --track-origins=yes to see where uninitialised values come from
==16261== ERROR SUMMARY: 2 errors from 2 contexts (suppressed: 4 from 4)
Error running the calculation.

Check Kernel log (listing) and error* files for details


  ********************************************
         Error in calculation stage.
  ********************************************

Regards,

Alexandre

[Alexandre Lebouvier]

We finally reinstall entirely CS2-rc2 properly and it's now working.