I am seeing code that allocates an std::string on the stack result in the CPU spinning at 100% within TBB's scalable_malloc(). The same issue was described in this post from last year:
http://software.intel.com/en-us/forums/topic/278331
The stack trace looks as follows:
#0 0x00002b6411b861e9 in rml::internal::Block::privatizePublicFreeList() () from ./libtbbmalloc.so.2
#1 0x00002b6411b868e3 in rml::internal::Bin::getPublicFreeListBlock() () from ./libtbbmalloc.so.2
#2 0x00002b6411b87e00 in rml::internal::internalPoolMalloc(rml::internal::MemoryPool*, unsigned long) () from ./libtbbmalloc.so.2
#3 0x00002b6411b880d6 in scalable_malloc () from ./libtbbmalloc.so.2
#4 0x00002b6411da4279 in operator new(unsigned long) () from ./libtbbmalloc_proxy.so.2
#5 0x00002b641500b8f9 in std::basic_string<char, std::char_traits<char>, std::allocator<char> >::_Rep::_S_create(unsigned long, unsigned long, std::allocator<char> const&) () from /opt/spotdev/3rdParty/cpp/gnu/gcc/gcc-4.6.2/lib64/libstdc++.so.6
Any ideas or suggestions to further debug?
Thanks
