Possible memory leak in ostei_s_s_s_s.c

[colleeneb]

Hello,

It looks like there might be a memory leak in ostei_s_s_s_s.c, based on the output of valgrind on one of the test cases:

[bertoni@jlselogin2 test]$ OMP_NUM_THREADS=1 valgrind  --leak-check=full --show-leak-kinds=all ./test_eri ../../test/dat/water.sto-3g.mol
==21988== Memcheck, a memory error detector
==21988== Copyright (C) 2002-2017, and GNU GPL'd, by Julian Seward et al.
==21988== Using Valgrind-3.13.0 and LibVEX; rerun with -h for copyright info
==21988== Command: ./test_eri ../../test/dat/water.sto-3g.mol
==21988==

          Quartet      MaxErr     MaxRelErr
(  0  0 |  0  0 )     8.882e-16   7.166e-16
(  0  0 |  0  1 )     1.665e-16   8.373e-16
(  0  0 |  1  0 )     1.665e-16   8.373e-16
(  0  0 |  1  1 )     2.220e-16   5.199e-16
(  0  1 |  0  0 )     1.110e-16   7.467e-16
(  0  1 |  0  1 )     2.776e-17   9.111e-16
(  0  1 |  1  0 )     2.776e-17   8.167e-16
(  0  1 |  1  1 )     1.665e-16   7.012e-16
(  1  0 |  0  0 )     1.665e-16   8.192e-16
(  1  0 |  0  1 )     5.551e-17   9.111e-16
(  1  0 |  1  0 )     5.551e-17   9.111e-16
(  1  0 |  1  1 )     1.110e-16   7.149e-16
(  1  1 |  0  0 )     3.886e-16   7.990e-16
(  1  1 |  0  1 )     1.110e-16   5.623e-16
(  1  1 |  1  0 )     1.388e-16   5.843e-16
(  1  1 |  1  1 )     4.441e-16   5.655e-16

Calculated 2401 contracted integrals

==21988==
==21988== HEAP SUMMARY:
==21988==     in use at exit: 4,616 bytes in 34 blocks
==21988==   total heap usage: 1,362,122 allocs, 1,362,088 frees, 33,479,348 bytes allocated
==21988==
==21988== 8 bytes in 1 blocks are still reachable in loss record 1 of 4
==21988==    at 0x4C29BC3: malloc (vg_replace_malloc.c:299)
==21988==    by 0x5B769C8: gomp_malloc (alloc.c:37)
==21988==    by 0x5B85686: gomp_init_num_threads (proc.c:91)
==21988==    by 0x5B74FFA: initialize_env (env.c:1244)
==21988==    by 0x400FAC2: _dl_init (in /usr/lib64/ld-2.17.so)
==21988==    by 0x4001029: ??? (in /usr/lib64/ld-2.17.so)
==21988==    by 0x1: ???
==21988==    by 0x1FFEFFCC7A: ???
==21988==    by 0x1FFEFFCC85: ???
==21988==
==21988== 192 bytes in 1 blocks are still reachable in loss record 2 of 4
==21988==    at 0x4C29BC3: malloc (vg_replace_malloc.c:299)
==21988==    by 0x5B769C8: gomp_malloc (alloc.c:37)
==21988==    by 0x5B837A5: gomp_get_thread_pool (pool.h:42)
==21988==    by 0x5B837A5: get_last_team (team.c:146)
==21988==    by 0x5B837A5: gomp_new_team (team.c:165)
==21988==    by 0x5B7AC75: GOMP_parallel (parallel.c:167)
==21988==    by 0x4075D7: main (test_eri.cpp:121)
==21988==
**==21988== 1,536 bytes in 16 blocks are definitely lost in loss record 3 of 4
==21988==    at 0x4C29BC3: malloc (vg_replace_malloc.c:299)
==21988==    by 0x4153CE: ostei_s_s_s_s (ostei_s_s_s_s.c:51)
==21988==    by 0x4129D8: simint_compute_ostei (ostei.c:22)
==21988==    by 0x4083C4: main._omp_fn.0 (test_eri.cpp:168)
==21988==    by 0x5B7AC8E: GOMP_parallel (parallel.c:168)
==21988==    by 0x4075D7: main (test_eri.cpp:121)
==21988==
==21988== 2,880 bytes in 16 blocks are definitely lost in loss record 4 of 4
==21988==    at 0x4C29BC3: malloc (vg_replace_malloc.c:299)
==21988==    by 0x415B76: ostei_s_s_s_s (ostei_s_s_s_s.c:83)
==21988==    by 0x4129D8: simint_compute_ostei (ostei.c:22)
==21988==    by 0x4083C4: main._omp_fn.0 (test_eri.cpp:168)
==21988==    by 0x5B7AC8E: GOMP_parallel (parallel.c:168)
==21988==    by 0x4075D7: main (test_eri.cpp:121)**
==21988==
==21988== LEAK SUMMARY:
==21988==    definitely lost: 4,416 bytes in 32 blocks
==21988==    indirectly lost: 0 bytes in 0 blocks
==21988==      possibly lost: 0 bytes in 0 blocks
==21988==    still reachable: 200 bytes in 2 blocks
==21988==         suppressed: 0 bytes in 0 blocks
==21988==
==21988== For counts of detected and suppressed errors, rerun with: -v
==21988== ERROR SUMMARY: 2 errors from 2 contexts (suppressed: 0 from 0)

Based on the message, it looks like there's a pointer that isn't being freed in ostei_s_s_s_s.c:, and I submitted a PR #13 to fix it.

-                if (!offset_info) free(offset_info);
+                if (offset_info != NULL) free(offset_info);

After making the change, valgrind no longer reports that memory is "definitely lost":

[bertoni@jlselogin2 test]$ OMP_NUM_THREADS=1 valgrind  --leak-check=full --show-leak-kinds=all ./test_eri ../../test/dat/water.sto-3g.mol
==24531== Memcheck, a memory error detector
==24531== Copyright (C) 2002-2017, and GNU GPL'd, by Julian Seward et al.
==24531== Using Valgrind-3.13.0 and LibVEX; rerun with -h for copyright info
==24531== Command: ./test_eri ../../test/dat/water.sto-3g.mol
==24531==

          Quartet      MaxErr     MaxRelErr
(  0  0 |  0  0 )     8.882e-16   7.166e-16
(  0  0 |  0  1 )     1.665e-16   8.373e-16
(  0  0 |  1  0 )     1.665e-16   8.373e-16
(  0  0 |  1  1 )     2.220e-16   5.199e-16
(  0  1 |  0  0 )     1.110e-16   7.467e-16
(  0  1 |  0  1 )     2.776e-17   9.111e-16
(  0  1 |  1  0 )     2.776e-17   8.167e-16
(  0  1 |  1  1 )     1.665e-16   7.012e-16
(  1  0 |  0  0 )     1.665e-16   8.192e-16
(  1  0 |  0  1 )     5.551e-17   9.111e-16
(  1  0 |  1  0 )     5.551e-17   9.111e-16
(  1  0 |  1  1 )     1.110e-16   7.149e-16
(  1  1 |  0  0 )     3.886e-16   7.990e-16
(  1  1 |  0  1 )     1.110e-16   5.623e-16
(  1  1 |  1  0 )     1.388e-16   5.843e-16
(  1  1 |  1  1 )     4.441e-16   5.655e-16

Calculated 2401 contracted integrals

==24531==
==24531== HEAP SUMMARY:
==24531==     in use at exit: 200 bytes in 2 blocks
==24531==   total heap usage: 1,362,122 allocs, 1,362,120 frees, 33,479,348 bytes allocated
==24531==
==24531== 8 bytes in 1 blocks are still reachable in loss record 1 of 2
==24531==    at 0x4C29BC3: malloc (vg_replace_malloc.c:299)
==24531==    by 0x5B769C8: gomp_malloc (alloc.c:37)
==24531==    by 0x5B85686: gomp_init_num_threads (proc.c:91)
==24531==    by 0x5B74FFA: initialize_env (env.c:1244)
==24531==    by 0x400FAC2: _dl_init (in /usr/lib64/ld-2.17.so)
==24531==    by 0x4001029: ??? (in /usr/lib64/ld-2.17.so)
==24531==    by 0x1: ???
==24531==    by 0x1FFEFFCC7A: ???
==24531==    by 0x1FFEFFCC85: ???
==24531==
==24531== 192 bytes in 1 blocks are still reachable in loss record 2 of 2
==24531==    at 0x4C29BC3: malloc (vg_replace_malloc.c:299)
==24531==    by 0x5B769C8: gomp_malloc (alloc.c:37)
==24531==    by 0x5B837A5: gomp_get_thread_pool (pool.h:42)
==24531==    by 0x5B837A5: get_last_team (team.c:146)
==24531==    by 0x5B837A5: gomp_new_team (team.c:165)
==24531==    by 0x5B7AC75: GOMP_parallel (parallel.c:167)
==24531==    by 0x4075D7: main (test_eri.cpp:121)
==24531==
==24531== LEAK SUMMARY:
==24531==    definitely lost: 0 bytes in 0 blocks
==24531==    indirectly lost: 0 bytes in 0 blocks
==24531==      possibly lost: 0 bytes in 0 blocks
==24531==    still reachable: 200 bytes in 2 blocks
==24531==         suppressed: 0 bytes in 0 blocks
==24531==
==24531== For counts of detected and suppressed errors, rerun with: -v
==24531== ERROR SUMMARY: 0 errors from 0 contexts (suppressed: 0 from 0)

I think this is the correct fix, but let me know if I'm missing something. Thanks!

[bennybp]

Fixed in PR #13