This should be moved into the PROTON project. I don't have the needed perms, so let's wait for an admin to do it. [~robbie]?
> Crash when closing application using Qpid - proton 0.25
> -------------------------------------------------------
>
> Key: QPID-8498
> URL:
https://issues.apache.org/jira/browse/QPID-8498> Project: Qpid
> Issue Type: Bug
> Components: C++ Broker
> Affects Versions: 0.25
> Reporter: Koen
> Priority: Major
>
> Hello,
> I'm having a problem with Qpid-proton-cpp version 0.25.0 and was wondering if this was already solved in a newer version.
> When I shutdown the application, it crashes sometimes (1 time in 40 shutdown's) and generates the following stacktrace:
>
>
> {code:java}
> #0 0x00000000 in ?? ()
> #1 0xf325203e in pn_class_decref (clazz=0xee100fb0, object=0xee100738) at /opt/jenkins_home/workspace/Qpid_release_build/qpid-proton-src/c/src/core/object/object.c:91
> #2 0xf32522ce in pn_decref (object=0xee100738) at /opt/jenkins_home/workspace/Qpid_release_build/qpid-proton-src/c/src/core/object/object.c:253
> #3 0xf3ee0b50 in proton::internal::pn_ptr_base::decref (p=0xee100738) at /opt/jenkins_home/workspace/Qpid_release_build/qpid-proton-src/cpp/src/object.cpp:31
> #4 0x08eca58e in proton::internal::pn_ptr<pn_connection_t>::~pn_ptr (this=0xaf5eeac, __in_chrg=<optimized out>) at /opt/qpid-itr/include/proton/internal/object.hpp:55
> #5 0x08ec7004 in proton::internal::object<pn_connection_t>::~object (this=0xaf5eeac, __in_chrg=<optimized out>) at /opt/qpid-itr/include/proton/internal/object.hpp:86
> #6 0x08ede3be in proton::connection::~connection (this=0xaf5eea8, __in_chrg=<optimized out>) at /opt/qpid-itr/include/proton/connection.hpp:45
> #7 REST OF THE STACKTRACE
> {code}
>
> I looked through my own code and didn't find anything that would cause this issue so I was wondering if the stacktrace is known and fixed in newer version of qpid.
> I ran my application through valgrind, but valgrind didn't find the memory leak. That could be due to not crashing all the time.
> It looks like a double free (pni_head_t) happening inside qpid. Although I'm not entirely sure..
> I was also curious if the number of items in this list needs to be decremented every time when calling the pni_free_children destructor
> {code:java}
> Function:
> static void pni_free_children(pn_list_t *children, pn_list_t *freed)
> while (pn_list_size(children) > 0) { //added printf to print the children pointers and there list size.
> #pn_list_size(children) for list size
> #printing pointer = children
> Output:
> children = 0xd039d4d8, list size = 1
> children = 0xedc00a50, list size = 2568
> children = 0xedc0c938, list size = 1
> children = 0xedc00a50, list size = 2567
> children = 0xedc0f890, list size = 18
> children = 0xedc0f890, list size = 17
> children = 0xedc0f890, list size = 17 <==This would need to be 16 right?
> Received signal: Segmentation fault (11)
> ____________________ STACKTRACE:
> #1 ip=0x0000000008a5f37f sp=0x00000000ef5f0d70 backtrace::SignalHandler(int) + 0x7f
> #2 ip=0x00000000f76e2410 sp=0x00000000ef5f0d80 + 0x7f
> #3 ip=0x00000000edc01858 sp=0x00000000ef5f135c + 0x7f
> #4 ip=0x000000006e657264 sp=0x00000000f34512bc + 0x7f
> {code}
>
Locked
