[PR18457] Don't require rtld lock to compute DTV addr for static TLS

[Carlos O'Donell]

On 06/04/2015 12:36 PM, Andreas Schwab wrote:
> Alexandre Oliva <aoliva@redhat.com> writes:
> 
>>>> +static int running = 1;
>>>> +  while (running)
>>
>>> That's a data race with the modification do_end, or isn't it?
> 
> There is no way this can go wrong.

That doesn't matter, it's still conceptually wrong and continues
to promote the idea that concurrency doesn't matter because the
hardware will just fix it (tm).

The whole point of writing data-race-free code is to get people
to think about concurrency and using the right constructs to
write concurrent code. That way when we have better tooling to
detect serious data races we don't have to fix all of these issues.

Cheers,
Carlos.