LDREX/STREX and pre-emption on SMP hardware

On Mon, 2009-09-14 at 07:43 -0500, Bill Gatliff wrote:

Catalin Marinas wrote:

quoted

And here's an untested patch to clear the exclusive monitor on the
exception return path. If you are OK with the idea, I'll do some testing
before pushing it for upstream:

How are you going to "test" for this one? It's such a corner case, it's 
going to be tricky to validate that your changes are actually getting 
test coverage methinks...

I'm just testing it to make sure it doesn't break the compilation :-). I
can do a few more tests like running LTP in parallel but it may not find
any problems. It's strange that we haven't hit a bug yet (and we run
some intensive testing for CPU validation in ARM) or maybe we blamed
something else.

Apart from the switch_to and atomic_set modifications, the patch cannot
do more harm. As for these functions, I don't see any reason why the
modifications would not be correct but we can delay the merging until we
are sure (anyway, the more people looking at the patch, the better).

-- 
Catalin