Re: Memory leaks in common cpp.

[Joshua Moore-Oliva]

I have the exact same problems in 1.1.2.

From the looks of it it seems that pthread_key_delete is not being called at 
the end?

Errors from 1.1.2 pasted below.

==7272== 200 bytes in 1 blocks are definitely lost in loss record 1 of 3
==7272==    at 0x402E9D63: (within /usr/lib/valgrind/libpthread.so)
==7272==    by 0x402EBB21: (within /usr/lib/valgrind/libpthread.so)
==7272==    by 0x402EBC4A: __pthread_key_create (in 
/usr/lib/valgrind/libpthread.so)
==7272==    by 0x402628EE: ost::ThreadKey::ThreadKey(void (*)(void*)) (in 
/usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x40260BB3: (within /usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x40260CFB: (within /usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x40279A34: (within /usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x4025CB30: (within /usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x4000B78D: (within /lib/ld-2.3.2.so)
==7272==    by 0x4000B879: _dl_init (in /lib/ld-2.3.2.so)
==7272==    by 0x40000A24: (within /lib/ld-2.3.2.so)
==7272==
==7272==
==7272== 220 bytes in 1 blocks are still reachable in loss record 2 of 3
==7272==    at 0x4002ADFE: operator new(unsigned) (in 
/usr/lib/valgrind/vgskin_memcheck.so)
==7272==    by 0x4025EBDA: ost::Thread::Thread(bool) (in 
/usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x40260F56: ost::MainThread::MainThread() (in 
/usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x40260BF0: (within /usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x40260CFB: (within /usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x40279A34: (within /usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x4025CB30: (within /usr/lib/libccgnu2-1.1.so.0.0.2)
==7272==    by 0x4000B78D: (within /lib/ld-2.3.2.so)
==7272==    by 0x4000B879: _dl_init (in /lib/ld-2.3.2.so)
==7272==    by 0x40000A24: (within /lib/ld-2.3.2.so)

On April 2, 2004 6:29 pm, David Sugar wrote:
> This looks like a 1.0.11 release.  Does this still happen in 1.1.2?  I know 
> of 
> several memory leaks that were found and fixed in 1.1.0 and 1.1.1.
> 
> On Friday 02 April 2004 06:07 pm, Joshua Moore-Oliva wrote:
> > I am not sure if I need to call a function at the end of my program ala apr
> > style (apr_terminate()) or if this is a memory leak in gnu common c++ but
> > there are two memory leaks when I just link a program to gnucpp...
> >
> > I would really like to get rid of these..
> >
> > Thanks, Joshua Moore-Oliva
> >
> > 200 bytes in 1 blocks are definitely lost in loss record 1 of 3
> > ==7196==    at 0x402E2D63: (within /usr/lib/valgrind/libpthread.so)
> > ==7196==    by 0x402E4B21: (within /usr/lib/valgrind/libpthread.so)
> > ==7196==    by 0x402E4C4A: __pthread_key_create (in
> > /usr/lib/valgrind/libpthread.so) ==7196==    by 0x40261186:
> > ost::ThreadKey::ThreadKey(void (*)(void*)) (in
> > /usr/lib/libccgnu2-1.0.so.0.0.11) ==7196==    by 0x4025F77C: (within
> > /usr/lib/libccgnu2-1.0.so.0.0.11) ==7196==    by 0x4025F87A: (within
> > /usr/lib/libccgnu2-1.0.so.0.0.11) ==7196==    by 0x40276FC4: (within
> > /usr/lib/libccgnu2-1.0.so.0.0.11) ==7196==    by 0x4025C2F0: (within
> > /usr/lib/libccgnu2-1.0.so.0.0.11) ==7196==    by 0x4000B78D: (within
> > /lib/ld-2.3.2.so)
> > ==7196==    by 0x4000B879: _dl_init (in /lib/ld-2.3.2.so)
> > ==7196==    by 0x40000A24: (within /lib/ld-2.3.2.so)
> > ==7196==
> > ==7196==
> > ==7196== 212 bytes in 1 blocks are still reachable in loss record 2 of 3
> > ==7196==    at 0x4002ADFE: operator new(unsigned) (in
> > /usr/lib/valgrind/vgskin_memcheck.so) ==7196==    by 0x4025DDB5:
> > ost::Thread::Thread(bool) (in /usr/lib/libccgnu2-1.0.so.0.0.11) ==7196==   
> > by 0x4025F7B4: (within /usr/lib/libccgnu2-1.0.so.0.0.11) ==7196==    by
> > 0x4025F87A: (within /usr/lib/libccgnu2-1.0.so.0.0.11) ==7196==    by
> > 0x40276FC4: (within /usr/lib/libccgnu2-1.0.so.0.0.11) ==7196==    by
> > 0x4025C2F0: (within /usr/lib/libccgnu2-1.0.so.0.0.11) ==7196==    by
> > 0x4000B78D: (within /lib/ld-2.3.2.so)
> > ==7196==    by 0x4000B879: _dl_init (in /lib/ld-2.3.2.so)
> > ==7196==    by 0x40000A24: (within /lib/ld-2.3.2.so)
> >
> >
> > _______________________________________________
> > Bug-commoncpp mailing list
> > address@hidden
> > http://mail.gnu.org/mailman/listinfo/bug-commoncpp
>