Re: [PATCH] locking/osq_lock: fix a data race in osq_wait_next

[Peter Zijlstra]

On Thu, Jan 30, 2020 at 02:39:38PM +0100, Marco Elver wrote:
> On Wed, 29 Jan 2020 at 19:40, Peter Zijlstra <peterz@xxxxxxxxxxxxx> wrote:

> > It's probably not terrible to put a READ_ONCE() there; we just need to
> > make sure the compiler doesn't do something stupid (it is known to do
> > stupid when 'volatile' is present).
> 
> Maybe we need to optimize READ_ONCE().

I think recent compilers have gotten better at volatile. In part because
of our complaints.

> 'if (data_race(..))' would also work here and has no cost.

Right, that might be the best option.