valgrind: blocks are definitely lost in loss record... in new thread

Question

I runing this code and I getting this message from Valgrind, I new in c++ and Linux but I know that I have to fix it. please can you advise what is wrong?. I am using Ubuntu.

The message is this:

==29304== Thread 1:
==29304== 72 bytes in 18 blocks are definitely lost in loss record 55 of 89
==29304==    at 0x402641D: operator new(unsigned int) (vg_replace_malloc.c:255)
==29304==    by 0x4032BBC: GmpPipePlayer::GmpPipePlayer(IOBase*, Referee*, unsigned char, int, DataBoard const*, int, char const*, int) (unixgmppipe.cpp:126)
==29304==    by 0x40329F9: GmpPipePlayer::CreateFunc(IOBase*, Referee*, unsigned char, int, DataBoard const*, void*) (unixgmppipe.cpp:55)

the code is: int down[2], up[2];

pipe(down);   
pipe(up);

_pid = fork();

if (_pid < 0)
    exit(1);


if (_pid == 0)
{
    close(down[1]);
    close(up[0]);

    dup2(down[0], 0);
    dup2(up[1], 1);

    execl("/bin/sh", "sh", "-c", cmd_line, NULL);

    _exit(1);
}

close(down[0]);
close(up[1]);
_down = down[1];
_up = up[0];

_reader_thd = new Thread(reader_wrapper, this); //here is the error happening.

the function readre_wrapper is:

THREAD_Return GmpPipePlayer::reader_wrapper(void *p)

{
    GmpPipePlayer *t = (GmpPipePlayer *)p;

    t->reader_fn();

    return NULL;
}

this new Thread is called many times. before I call it again i do the following:

if (_pid > 0)
{
    kill(_pid, SIGTERM);
    _pid = 0;
}

if (_up)
{
    close(_up);
    _up = 0;
}

if (_down)
{
    close(_down);
    _down = 0;
}   

Any ideas what is wrong?

Answer 1

Well, you have a new that's never matched by a delete. You'll need a delete _reader_thd; at some point in your code. Or not use dynamic objects at all.