Strange application crash - c++

Question

I have this C++ application running on a Ubuntu Lucid 10.04.3 LTS which is crashed and the reason really escapes me.
The method which exhibits failure is this one:
void
IoLogikCommunicator::processPacket(char const* data, WORD wSize)
{
std::string message(data, wSize);
std::stringstream ss(message);
std::string token;
std::vector<std::string> tokens;
while (std::getline(ss, token, '#')) // <- crash
tokens.push_back(token);
if (tokens[0] == "SENSORS")
processSensorsPacket(tokens);
else if (tokens[0] == "SELECTOR")
processSelectorPacket(tokens);
}
According to the core dump, data content is valid and it is:
p data
$1 = 0xb7520214 "SENSORS#192.168.107.62#DI:00#ON#DI:01#ON#DI:02#ON#DI:03#OFF#DI:04#OFF#DI:05#OFF"
p wSize
$2 = 79
The content of tokens, at crash time, is ["SENSORS"], so the first element was parsed correctly.
What happens then is:
Program terminated with signal 6, Aborted.
#0 0x009de422 in __kernel_vsyscall ()
(gdb) bt
#0 0x009de422 in __kernel_vsyscall ()
#1 0x0766a651 in raise () from /lib/tls/i686/cmov/libc.so.6
#2 0x0766da82 in abort () from /lib/tls/i686/cmov/libc.so.6
#3 0x076a149d in ?? () from /lib/tls/i686/cmov/libc.so.6
#4 0x076ab591 in ?? () from /lib/tls/i686/cmov/libc.so.6
#5 0x076ae710 in ?? () from /lib/tls/i686/cmov/libc.so.6
#6 0x076aff9c in malloc () from /lib/tls/i686/cmov/libc.so.6
#7 0x0070dc07 in operator new(unsigned int) () from /usr/lib/libstdc++.so.6
#8 0x006e7d06 in std::string::_Rep::_S_create(unsigned int, unsigned int, std::allocator<char> const&) () from /usr/lib/libstdc++.so.6
#9 0x006e9f70 in std::string::_M_mutate(unsigned int, unsigned int, unsigned int) () from /usr/lib/libstdc++.so.6
#10 0x006c4274 in std::basic_istream<char, std::char_traits<char> >& std::getline<char, std::char_traits<char>, std::allocator<char> >(std::basic_istream<char, std::char_traits<char> >&, std::basic_string<char, std::char_traits<char>, std::allocator<char> >&, char) () from /usr/lib/libstdc++.so.6
given the SIGABRT it seems that an assert() fails inside the malloc invocation, but what could be the reason? Of course, it was impossible for me to reproduce the bug: this method is invoked several times per second and the application crashed after 30 and more days of continuous running.
The very same data, then, is processed by another identical application which is hosted on another machine: that one didn't crash.
Do you have any suggestion/hint/tips/pointer?