-fsanitize=nonnull-attribute failing on empty container

[Lars Gullik Bjønnes]

On 9 June 2015 at 01:09, Jonathan Wakely <jwakely.gcc@gmail.com> wrote:
> On 8 June 2015 at 19:22, Lars Gullik Bjønnes wrote:
>> I wonder if this is another in the same vein:
>>
>> #0  __ubsan::__ubsan_handle_nonnull_arg_abort (Data=0x555555edbfe0)
>>    at ../../../../gcc/libsanitizer/ubsan/ubsan_handlers.cc:408
>> #1  0x000055555593e397 in void std::__cxx11::basic_string<char,
>> std::char_traits<char>, std::allocator<char>
>>>::_M_construct<char*>(char*, char*, std::forward_iterator_tag) [clone
>> .isra.75] ()
>>
>> I don't have a proper test case for it yet, I get it via some
>> boost::regex stuff.
>>
>> Will look harder at it.
>>
>> This is the sanitizer message:
>>
>> host-gcc-5.1.1-2ed068d-x86_64-linux-1/include/c++/5.1.1/bits/char_traits.h:275:73:
>> runtime error: null pointer passed as argument 2, which is declared to
>> never be null
>>
>> I'll continue investigation.
>
> It's probably https://gcc.gnu.org/bugzilla/show_bug.cgi?id=65049

Yes, looks like it.

I am surprised that it does not trigger more often, but I guess
not too many are using the sanitizers yet.



-- 
        Lgb