On Sat, Feb 18, 2023 at 2:21 PM Paul E. McKenney <paulmck@xxxxxxxxxx> wrote:
>
> On Sat, Feb 18, 2023 at 01:13:59AM -0500, Joel Fernandes wrote:
> > Hi Alan,
> >
> > On Sat, Feb 11, 2023 at 9:59 PM Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote:
> > >
> > [...]
> > >
> > > Would you like to post a few examples showing some of the most difficult
> > > points you encountered? Maybe explanation.txt can be improved.
> >
> > One additional feedback I wanted to mention, regarding this paragraph
> > under "WARNING":
> > ===========
> > The protections provided by READ_ONCE(), WRITE_ONCE(), and others are
> > not perfect; and under some circumstances it is possible for the
> > compiler to undermine the memory model. Here is an example. Suppose
> > both branches of an "if" statement store the same value to the same
> > location:
> > r1 = READ_ONCE(x);
> > if (r1) {
> > WRITE_ONCE(y, 2);
> > ... /* do something */
> > } else {
> > WRITE_ONCE(y, 2);
> > ... /* do something else */
> > }
> > ===========
> >
> > I tried lots of different compilers with varying degrees of
> > optimization, in all cases I find that the conditional instruction
> > always appears in program order before the stores inside the body of
> > the conditional. So I am not sure if this is really a valid concern on
> > current compilers, if not - could you provide an example of a compiler
> > and options that cause it?
> >
> > In any case, if it is a theoretical concern, it could be clarified
> > that this is a theoretical possibility in the text. And if it is a
> > real/practical concern, then it could be mentioned the specific
> > compiler/arch this was seen in.
>
> I could be misremembering, but I believe that this reordering has been
> seen in the past.
>
Thank you! And I also confirmed putting a barrier() in the branch
body, also "cures" the optimization... I did not know compilers
optimize so aggressively..
- Joel
