views:

88

answers:

1

I've recently been running one of my apps through Valgrind but there's a few MYSQL related leaks I can't fix. I put the offending code in the most basic form and tested that; I got the same leaks. Should I just disregard them or am I doing something wrong?

Code:

#include <stdio.h>
#include <stdlib.h>

#include <mysql/mysql.h>

int main()
{
    MYSQL *MYSQLIns;

    MYSQLIns = mysql_init(NULL);

    mysql_real_connect(MYSQLIns, "localhost", "username", "password", "database", 0, NULL, 0);

    mysql_close(MYSQLIns);

    return EXIT_SUCCESS;
}

Compiled With:

gcc -g -lmysqlclient mysql_mem_test.c -o mysql_mem_test

Valgrind Output:

valgrind --leak-check=full ./mysql_mem_test
==4601== Memcheck, a memory error detector
==4601== Copyright (C) 2002-2009, and GNU GPL'd, by Julian Seward et al.
==4601== Using Valgrind-3.5.0-Debian and LibVEX; rerun with -h for copyright info
==4601== Command: ./mysql_mem_test
==4601== 
==4601== 
==4601== HEAP SUMMARY:
==4601==     in use at exit: 61,336 bytes in 16 blocks
==4601==   total heap usage: 70 allocs, 54 frees, 109,218 bytes allocated
==4601== 
==4601== 24,528 bytes in 6 blocks are possibly lost in loss record 4 of 5
==4601==    at 0x4024C1C: malloc (vg_replace_malloc.c:195)
==4601==    by 0x4079BB2: my_once_alloc (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x407A408: ??? (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x407AED1: ??? (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x407B112: get_charset_by_csname (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x409D55B: mysql_init_character_set (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x409F5A1: mysql_real_connect (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x80485E0: main (mysql_mem_test.c:12)
==4601== 
==4601== 28,616 bytes in 7 blocks are possibly lost in loss record 5 of 5
==4601==    at 0x4024C1C: malloc (vg_replace_malloc.c:195)
==4601==    by 0x4079BB2: my_once_alloc (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x407A3E9: ??? (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x407AED1: ??? (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x407B112: get_charset_by_csname (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x409D55B: mysql_init_character_set (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x409F5A1: mysql_real_connect (in /usr/lib/libmysqlclient.so.16.0.0)
==4601==    by 0x80485E0: main (mysql_mem_test.c:12)
==4601== 
==4601== LEAK SUMMARY:
==4601==    definitely lost: 0 bytes in 0 blocks
==4601==    indirectly lost: 0 bytes in 0 blocks
==4601==      possibly lost: 53,144 bytes in 13 blocks
==4601==    still reachable: 8,192 bytes in 3 blocks
==4601==         suppressed: 0 bytes in 0 blocks
==4601== Reachable blocks (those to which a pointer was found) are not shown.
==4601== To see them, rerun with: --leak-check=full --show-reachable=yes
==4601== 
==4601== For counts of detected and suppressed errors, rerun with: -v
==4601== ERROR SUMMARY: 2 errors from 2 contexts (suppressed: 55 from 38)
+3  A: 

Does adding mysql_library_end(); after the mysql_close help?

Update: installed valgrind and tried it and indeed it does. mysql_init() implicitly does mysql_library_init(), but mysql_close() doesn't similarly do mysql_library_end(), which the documentation implies but doesn't come straight out and say.

ysth
Thanks so much, never heard of that function at all. Thanks again!
Kewley