Re: [Valgrind-users] Unsure about invalid read in destructor

[Stephen T. <st...@to...>]

On Mon, 2005-11-28 at 16:46 +0100, Dennis Lubert wrote:
> Is this all of the error message ? Usually valgrind tells us what it
> knows about the memory, e.g.:
> 
> Address 0x3FA50239 is not stack'd, malloc'd or (recently) free'd
> 
> or that it has been freed already or similar, it should come right after
> the stacktrace of the invalid read.
> 
> greets
> 
> Dennis

No it is not all there is to the message. I am sorry I did not include
it all. What I send seemed to be sufficient but I did not know what is
expected.

        ==11257== Invalid read of size 4
        ==11257==    at 0x804D2F6: guitar::Chord::~Chord() (chord.cpp:32)
        ==11257==    by 0x804EA00: guitar::ChordMap::~ChordMap() (chordmap.cpp:45)
        ==11257==    by 0x804BD44: test_insert() (test_chordmap.cpp:54)
        ==11257==    by 0x804C0A9: main (test_chordmap.cpp:88)
        ==11257==  Address 0x1C878270 is 0 bytes inside a block of size 32 free'd
        ==11257==    at 0x1B900807: operator delete(void*) (in /usr/lib/valgrind/vgpreload_memcheck.so)
        ==11257==    by 0x8057E14: guitar::ChordName::~ChordName() (chordname.cpp:40)
        ==11257==    by 0x804D307: guitar::Chord::~Chord() (chord.cpp:32)
        ==11257==    by 0x804BD0C: test_insert() (test_chordmap.cpp:54)
        ==11257==    by 0x804C0A9: main (test_chordmap.cpp:88)

The message after the "Address ... " line does not make sense because
line 40 of chordname.map is the end of the destructor.

        ChordName::~ChordName ()
        {
            for ( std::vector<ChordName*>::iterator pos = m_aliases.begin();
                    pos != m_aliases.end();
                    ++pos )
            {
                delete *pos;
            }
        }

Stephen