On Thu, Jun 9, 2011 at 3:34 PM, John Regehr
<regehr@cs.utah.edu> wrote:
It looks like some compilers (for example GCC) are not too far from being ready to try to comply with the C++0x memory model. This will be a big job and they do not yet fully understand how this model will interact with their optimizer. It's fair to say that most other C++ compilers will be running into similar issues in coming years. We can help by adding C++0x memory model constructs to the set of things Csmith can emit.
Note that I'm not suggesting that we emit "interesting" C++, which is a very big job. Rather, we can emit C++ that looks very much like C but that contains locks, atomics, volatiles, etc. I can't imagine there's anything difficult about this, given the infrastructure we already have.
Hans Boehm has an excellent slide deck on what can go wrong when compiling this sort of code:
http://www.hpl.hp.com/personal/Hans_Boehm/misc_slides/c++mm.pdf
The problem of determining the correctness of a compiler's translation of a piece of code containing C++0x concurrency primitives is not trivial, but basically it's not our problem. There exist (or will exist) some smart checking tools for this sort of thing.
John