[lttng-dev] [PATCH v2 04/12] urcu/system: Use atomic builtins if configured

[Paul E. McKenney]

On Wed, Jul 05, 2023 at 03:03:21PM -0400, Olivier Dion wrote:
> On Wed, 05 Jul 2023, "Paul E. McKenney" <paulmck at kernel.org> wrote:
> > On Tue, Jul 04, 2023 at 10:43:21AM -0400, Olivier Dion wrote:
> >> On Wed, 21 Jun 2023, "Paul E. McKenney" <paulmck at kernel.org> wrote:
> >> > On Wed, Jun 07, 2023 at 02:53:51PM -0400, Olivier Dion wrote:
> >> >
> >> > Same question here on loss of volatile semantics.
> >> 
> >> This apply to all review on volatile semantics.  I added a
> >> cmm_cast_volatile() macro/template for C/C++ for adding the volatile
> >> qualifier to pointers passed to every atomic builtin calls.
> >
> > Sounds very good, thank you!
> 
> Maybe a case of synchronicity here, but I just stumble upon this
> <https://www.open-std.org/jtc1/sc22/wg21/docs/papers/2016/p0062r1.html>
> where you seem to express the same concerns :-)

Just for completeness, my response to Hans's concern about volatile is
addressed by an empty memory-clobber asm, similar to barrier() in the
Linux kernel.

But yes, this has seen significant discussion over the years.  ;-)

							Thanx, Paul