myallod commented 10 years ago

My licq 1.9.0-fb75e8e/SSL is crashed 2-3 times per day. I use one icq account and eight jabber accounts. i suspect licq is crashing as a whole when one of jabber accounts disconnects suddenly, but not always - sometimes jabber account goes offline and licq continues to work properly. I made licq with cmake --DCMAKE_BUILD_TYPE=Debug in build-all directory. Then i run licq under gdb and as usual after some hours of working i've got:

Program received signal SIGSEGV, Segmentation fault. [Switching to Thread 0x7ffff3a69700 (LWP 20088)] 0x00007ffff6893503 in std::_Rb_tree_increment(std::_Rb_tree_node_base*) () from /usr/lib/x86_64-linux-gnu/libstdc++.so.6 (gdb) bt

0 0x00007ffff6893503 in std::_Rb_tree_increment(std::_Rb_tree_node_base*) () from /usr/lib/x86_64-linux-gnu/libstdc++.so.6

1 0x0000000000543487 in std::_Rb_tree_iterator<std::pair<int const, Licq::MainLoop::Private::File> >::operator++ (this=0x7ffff3a68c40)

at /usr/include/c++/4.7/bits/stl_tree.h:188

2 0x00000000005426b7 in Licq::MainLoop::run (this=0x9b7a20) at /home/lek/Downloads/src/licq/licq/src/mainloop.cpp:198

3 0x00007ffff305312a in LicqJabber::Plugin::run (this=0x9b7a00) at /home/lek/Downloads/src/licq/jabber/src/plugin.cpp:62

4 0x00000000005c6bc2 in LicqDaemon::PluginInstance::startThreadEntry (plugin=0x9b7b20) at /home/lek/Downloads/src/licq/licq/src/plugin/plugininstance.cpp:175

5 0x00000000005c704b in pluginThreadEntry (arg=0x859ae0) at /home/lek/Downloads/src/licq/licq/src/plugin/pluginthread.cpp:138

6 0x00007ffff7bc7b50 in start_thread () from /lib/x86_64-linux-gnu/libpthread.so.0

7 0x00007ffff62e8a7d in clone () from /lib/x86_64-linux-gnu/libc.so.6

8 0x0000000000000000 in ?? ()

(gdb) quit A debugging session is active.

Inferior 1 [process 20084] will be killed.

Quit anyway? (y or n) y

Linux 3.2.0-4-amd64 #1 SMP Debian 3.2.51-1 x86_64 GNU/Linux

Please help.

Next crash - run without gdb.

licq_backtrace: time: 1386056749 /usr/local/bin/licq() [0x555346] /lib/x86_64-linux-gnu/libc.so.6(+0x324f0) [0x7f2ba25124f0] /lib/x86_64-linux-gnu/libc.so.6(gsignal+0x35) [0x7f2ba2512475] /lib/x86_64-linux-gnu/libc.so.6(abort+0x180) [0x7f2ba25156f0] /usr/local/bin/licq() [0x554f16] /lib/x86_64-linux-gnu/libc.so.6(+0x324f0) [0x7f2ba25124f0] /usr/lib/x86_64-linux-gnu/libstdc++.so.6(_ZSt18_Rb_tree_incrementPSt18_Rb_tree_node_base+0x13) [0x7f2ba2b65503] /usr/local/bin/licq(_ZNSt17_Rb_tree_iteratorISt4pairIKiN4Licq8MainLoop7Private4FileEEEppEv+0x1b) [0x543487] /usr/local/bin/licq(_ZN4Licq8MainLoop3runEv+0x4a7) [0x5426b7] /usr/local/lib/licq/protocol_jabber.so(+0x1512a) [0x7f2b9f32512a] /usr/local/bin/licq(_ZN10LicqDaemon14PluginInstance16startThreadEntryEPv+0x82) [0x5c6bc2] /usr/local/bin/licq() [0x5c704b] /lib/x86_64-linux-gnu/libpthread.so.0(+0x6b50) [0x7f2ba3e99b50] /lib/x86_64-linux-gnu/libc.so.6(clone+0x6d) [0x7f2ba25baa7d]

licq.backtrace.gdb http://pastie.org/pastes/8524756/text

licq.debug_rw_mutex time: 1386057501 Possible deadlock for thread 0x7f2ba08b8700 trying to get a write lock on 'ICQ_277507035' No thread holds the write lock These threads hold the read lock: 0x7f2b9e221700

flynd commented 10 years ago

I see a possible race in the mainloop class. Could you try the following patch and see if it solves your problem?

http://pastebin.com/YwDB4REe