Segfaults with ssmp build

Ondrej Marsalek ondrej.... at gmail.com
Thu Jan 13 13:39:05 UTC 2011


Dear all,

I am getting segfaults with a ssmp build. The arch file and the test
job are here:

http://marge.uochb.cas.cz/~marsalek/tmp/Linux-x86-64-intel.ssmp
http://marge.uochb.cas.cz/~marsalek/tmp/test-smp.tar.gz

I get different results with one or more threads. I run under gdb so
that we can get the stack. With OMP_NUM_THREADS=1 I get, after some
usual output:

  ----------------------------------- OT ---------------------------------------

  Step     Update method      Time    Convergence         Total energy    Change
  ------------------------------------------------------------------------------

Program received signal SIGSEGV, Segmentation fault.
__libc_free (mem=0x8) at malloc.c:3709
3709	malloc.c: No such file or directory.
	in malloc.c
(gdb) backtrace
#0  __libc_free (mem=0x8) at malloc.c:3709
#1  0x000000000220b01c in for_deallocate ()
#2  0x00000000017fddf9 in
PW_SPLINE_UTILS::L_pw_spline_utils_mp_pw_nn_compose_r_work__1247__par_loop0_2_1723
()
    at /home/andy/build/cp2k/cp2k/makefiles/../src/pw_spline_utils.F:1265
#3  0x00000000022b8b33 in L_kmp_invoke_pass_parms ()
#4  0x00007fffffff1158 in ?? ()
#5  0x00007fffffff1134 in ?? ()
#6  0x00007fffffff0a68 in ?? ()
#7  0x00007fffffff0ac8 in ?? ()
#8  0x00007fffffff0c28 in ?? ()
#9  0x00007fffffff0ca0 in ?? ()
#10 0x00007fffffff13c4 in ?? ()
#11 0x00007fffffff13c8 in ?? ()
#12 0x00007fffffff13cc in ?? ()
#13 0x00007fffffff13d0 in ?? ()
#14 0x00007fffffff13d4 in ?? ()
#15 0x00007fffffff13d8 in ?? ()
#16 0x00007fffffff13ac in ?? ()
#17 0x00007fffffff13b0 in ?? ()
#18 0x00007fffffff13b4 in ?? ()
#19 0x00007fffffff13b8 in ?? ()
#20 0x00007fffffff13bc in ?? ()
#21 0x00007fffffff13c0 in ?? ()
#22 0x0000000002994880 in ?? ()
#23 0x0000000000000000 in ?? ()
(gdb)


With OMP_NUM_THREADS=2 the error is this:

 SCF PARAMETERS         Density guess:                                    ATOMIC
                        --------------------------------------------------------
                        max_scf:                                             300
                        max_scf_history:                                       0
                        max_diis:                                              4
                        --------------------------------------------------------
                        eps_scf:                                        1.00E-06
                        eps_scf_history:                                0.00E+00
                        eps_diis:                                       1.00E-01
                        eps_eigval:                                     1.00E-05
                        --------------------------------------------------------
                        level_shift [a.u.]:                                 0.00
                        --------------------------------------------------------
                        No outer SCF

Program received signal SIGSEGV, Segmentation fault.
[Switching to Thread 0x7ffff4fe4700 (LWP 31971)]
0x0000000000aee409 in
REALSPACE_GRID_TYPES::L_realspace_grid_types_mp_rs_grid_zero__1987__par_loop0_2_713
()
    at /home/andy/build/cp2k/cp2k/makefiles/../src/realspace_grid_types.F:1993
1993	       rs % r ( i, j, k ) = 0.0_dp
(gdb) backtrace
#0  0x0000000000aee409 in
REALSPACE_GRID_TYPES::L_realspace_grid_types_mp_rs_grid_zero__1987__par_loop0_2_713
()
    at /home/andy/build/cp2k/cp2k/makefiles/../src/realspace_grid_types.F:1993
#1  0x00000000022b8b33 in L_kmp_invoke_pass_parms ()
#2  0x0000000003fcbf80 in ?? ()
#3  0x0000000000000001 in ?? ()
#4  0x0000000000000000 in ?? ()
(gdb)


My Intel compiler is version 11.1.073. I will be grateful for a check
of whether this is reproducible and for a fix in case it is a bug.

Thanks,
Ondrej



More information about the CP2K-user mailing list