[PATCH] Docfix for Unsafe regarding fence methods

[Ian Graves]

Hans,

You raise a good point! I was attempting to re-adapt the existing language with the changes I made. 

I’m going to update this post to an RFR after the Skara update next week and repost (got mixed up on my process). There will be some simplified language that hopefully addresses your concerns.

Thanks for the input!

Ian

> On Aug 28, 2020, at 4:59 PM, Hans Boehm <hboehm at google.com> wrote:
> 
> 
> 
> On Fri, Aug 28, 2020 at 1:21 PM Ian Graves <ian.graves at oracle.com <mailto:ian.graves at oracle.com>> wrote:
> >
> > Looks like the patch attachment was stripped. I’ll include it inline:
> > —Ian
> >
> > ...
> > @@ -3458,6 +3462,12 @@
> >      /**
> >       * Ensures that stores before the fence will not be reordered with
> >       * stores after the fence.
> > +     *
> > +     * This method is operationally equivalent to {@link #storeFence()}
> > +     * as most hardware instructions that provide a StoreStore barrier
> > +     * provide a LoadStore barrier for free.
> > +     *
> > +     * @since 9
> >       */
> >      public final void storeStoreFence() {
> >          storeFence();
> >
> >
> This seems a little misleading, since ARM has "dmb ishst", which orders only stores?