[PATCH 2/4] Add atomic operations similar to those provided by C11.

[Torvald Riegel]

On Thu, 2014-10-30 at 20:12 +0000, Joseph S. Myers wrote:
> Comparing code for one architecture, at the time of each conversion, seems 
> appropriate (with additional benchmarking for cases that don't otherwise 
> involve atomics).

So, if you want this code comparison as default, we can as well start
with using relaxed __atomic* loads and stores instead of first using
plain loads and stores, agreed?

Do you have any other feedback on the patch set?