I have a class in which I implement the singelton design pattern. I know some people don't think its a good idea, but it helps a lot,
Anyway - I have a memory leak and vlagrind points me to these lines:
_singleton = new Manager(); //Manager::instance() (Manager.cpp:18)
And
Manager::Manager() : _file(new ofstream), _tasks(new map<int, Task *>()),
_idState(new map<int, int>()), _closing(false), _pending(false),
_lock(new pthread_mutex_t), _endLock(new pthread_mutex_t), _cond(new pthread_cond_t),
_flushCond(new map<int, pthread_cond_t *>()), _attr(new pthread_attr_t) {
//The last line is line 25 in Manager::Manager
Now in Manager's destructor I can't explicitly delete it, because it creates a silly loop (as destructor will be called when deleting _singleton resulting in an infinite loop). How do I get rid of this leak? Thanks!
P.s. here is Valgrind's output:
==17823== 512 bytes in 1 blocks are definitely lost in loss record 2 of 2
==17823== at 0x4C27297: operator new(unsigned long) (in /usr/lib/valgrind/vgpreload_memcheck-amd64-linux.so)
==17823== by 0x40151E: Manager::Manager() (Manager.cpp:25)
==17823== by 0x4014DB: Manager::instance() (Manager.cpp:18)
==17823== by 0x406475: initdevice(char*) (outputdevice.cpp:66)
==17823== by 0x4061D5: main (driver.cpp:21)
==17823==
==17823== LEAK SUMMARY:
==17823== definitely lost: 512 bytes in 1 blocks
= =17823== indirectly lost: 0 bytes in 0 blocks
==17823== possibly lost: 288 bytes in 1 blocks
==17823== still reachable: 0 bytes in 0 blocks
==17823== suppressed: 0 bytes in 0 blocks
Addition: here's the code where I create Manager:
Manager.h:
class Manager {
public:
~Manager();
static Manager * instance();
private:
Manager();
static Manager * _singleton;
};
And the implementation:
Manager.cpp:
Manager * Manager::_singleton = 0;
Manager * Manager::instance() {
if (!_singleton)
_singleton = new Manager();
return _singleton;
}