[Valgrind-users] Two stacktraces printed: what is the relationship between them?

[Graham L. <mi...@sh...>]

Hi all,

I have a very strange case of a piece of code that crashes when run under valgrind, but doesn’t crash when run normally or under gdb. Obviously something is wrong, but I don’t understand the information valgrind is trying to tell me.

I get an error "Invalid read of size 1”, a stacktrace, a message about overlapping memory then a second stacktrace, and that’s it.

Can anyone explain what the relationship is between these two stacktraces, what is this message trying to tell me?

==29256== Invalid read of size 1
==29256==    at 0xA0148F0: pn_selectable_is_terminal (selectable.c:247)
==29256==    by 0x997599B: run_messenger_thread (mod_amqp.c:1900)
==29256==    by 0x5F21DF4: start_thread (in /usr/lib64/libpthread-2.17.so)
==29256==    by 0x64301AC: clone (in /usr/lib64/libc-2.17.so)
==29256==  Address 0x698e343 is 99 bytes inside a block of size 104 free'd
==29256==    at 0x4C2AD17: free (in /usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
==29256==    by 0x9FF6BE5: pn_class_decref (object.c:103)
==29256==    by 0xA01067B: pn_messenger_free (messenger.c:822)
==29256==    by 0x9971583: messenger_cleanup (mod_amqp.c:314)
==29256==    by 0x58C3F0D: run_cleanups (apr_pools.c:2352)
==29256==    by 0x58C3F0D: apr_pool_destroy (apr_pools.c:814)
==29256==    by 0x58C3EE4: apr_pool_destroy (apr_pools.c:811)
==29256==    by 0x58C4144: apr_pool_clear (apr_pools.c:769)
==29256==    by 0x13E70A: main (in /usr/sbin/httpd)

Regards,
Graham
—