Crashes/segfaults with GCC 8.1

[jtebert]

Per @bcornille, fails with gcc 8.1.0 on Ubuntu 16.04

[r-barnes]

I can reproduce this with g++-8 (Ubuntu 8.2.0-1ubuntu2~18.04) 8.2.0

Using the address sanitizer reveals that the bug arises at Line 30 of Logger.cpp:

m_trial_group_name = "trial_" + std::to_string(trial_num);

WARNING: Overwrote trial data
=================================================================
==14158==ERROR: AddressSanitizer: attempting free on address which was not malloc()-ed: 0x7fff65934f20 in thread T0
    #0 0x7faf4e4aa640 in operator delete(void*) (/usr/lib/x86_64-linux-gnu/libasan.so.5+0xf0640)
    #1 0x56025309e3b6 in __gnu_cxx::new_allocator<char>::deallocate(char*, unsigned long) /usr/include/c++/8/ext/new_allocator.h:125
    #2 0x56025309e3b6 in std::allocator_traits<std::allocator<char> >::deallocate(std::allocator<char>&, char*, unsigned long) /usr/include/c++/8/bits/alloc_traits.h:462
    #3 0x56025309e3b6 in std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >::_M_destroy(unsigned long) /usr/include/c++/8/bits/basic_string.h:226
    #4 0x56025309e3b6 in std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >::_M_dispose() /usr/include/c++/8/bits/basic_string.h:221
    #5 0x56025309e3b6 in std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >::~basic_string() /usr/include/c++/8/bits/basic_string.h:657
    #6 0x56025309e3b6 in KiloSim::Logger::set_trial(unsigned int) src/Logger.cpp:30
    #7 0x5602530fc839 in KiloSim::Logger::Logger(KiloSim::World*, std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >, int, bool) src/Logger.cpp:18
    #8 0x5602530a31e7 in main src/test.cpp:68
    #9 0x7faf4c22cb96 in __libc_start_main (/lib/x86_64-linux-gnu/libc.so.6+0x21b96)
    #10 0x5602530a5609 in _start (/home/rick/projects/contributing/kilosim/bin/kilosim+0x15609)

Address 0x7fff65934f20 is located in stack of thread T0
SUMMARY: AddressSanitizer: bad-free (/usr/lib/x86_64-linux-gnu/libasan.so.5+0xf0640) in operator delete(void*)
==14158==ABORTING
Aborted (core dumped)

I can't see any reason why this would be a problem, so I'm guessing it's a compiler regression and the best way to fix it is to try to build a small reproducer and forward it to the GCC folks.

[r-barnes]

I've just tried with g++-8 (Ubuntu 8.3.0-6ubuntu1~18.04.1) 8.3.0 and there is no crash. Either it was a compiler bug in earlier versions of the code (I believe this), or it was a bug that got fixed amid other changes (I think this is less likely). We can probably declare victory here, or we could be more thorough and test earlier GCC8 series compilers to see if we can rove this more rigorously.

[r-barnes]

One way to do this would be godbolt plus this.