Boost logo

Threads-Devel :

Subject: [Threads-devel] leak in win32 detail::basic_condition_variable
From: Kenneth Porter (shiva.blacklist_at_[hidden])
Date: 2011-11-09 12:26:08


I'm finding that an element in the CV's generation_list is left orphaned
sometimes when the CV is destroyed. I'm running Visual Leak Detector and it
gives this stack dump of the allocation point of the leaked object:

    c:\devel\boost\boost_1_46_1\boost\thread\win32\condition_variable.hpp
(172): boost::detail::basic_condition_variable::get_wait_entry

    c:\devel\boost\boost_1_46_1\boost\thread\win32\condition_variable.hpp
(211): boost::detail::basic_condition_variable::do_wait<boost::unique_lock
<boost::mutex> > + 0xC bytes

    c:\devel\boost\boost_1_46_1\boost\thread\win32\condition_variable.hpp
(369): boost::condition_variable_any::wait<boost::unique_lock<boost::mutex>
>

    c:\devel\boost\boost_1_46_1\boost\thread\win32\condition_variable.hpp
(374): boost::condition_variable_any::wait<boost::unique_lock
<boost::mutex>,boost::_bi::bind_t<bool,boost::_mfi::cmf0
<bool,bounded_buffer<Axis::Report> >,boost::_bi::list1<boost::_bi::value
<bounded_buffer<Axis: + 0x1B bytes

I compared the source file against trunk and it hasn't changed.


Threads-Devel list run by bdawes at acm.org, david.abrahams at rcn.com, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk