> On Mon, Jan 21, 2019 at 11:05:03AM -0500, Alan Stern wrote:
> > On Mon, 21 Jan 2019, Peter Zijlstra wrote:
>
> > > Any additional ordering; like the one you have above; are not strictly
> > > required for the proper functioning of the refcount. Rather, you rely on
> > > additional ordering and will need to provide this explicitly:
> > >
> > >
> > > if (refcount_dec_and_text(&x->rc)) {
> > > /*
> > > * Comment that explains what we order against....
> > > */
> > > smp_mb__after_atomic();
> > > BUG_ON(!x->done*);
> > > free(x);
> > > }
> > >
> > >
> > > Also; these patches explicitly mention that refcount_t is weaker,
> > > specifically to make people aware of this difference.
> > >
> > > A full smp_mb() (or two) would also be much more expensive on a number
> > > of platforms and in the vast majority of the cases it is not required.
> >
> > How about adding smp_rmb() into refcount_dec_and_test()? That would
> > give acq+rel semantics, which seems to be what people will expect. And
> > it wouldn't be nearly as expensive as a full smp_mb().
>
> Yes, that's a very good suggestion.
>
> I suppose we can add smp_acquire__after_ctrl_dep() on the true branch.
> Then it reall does become rel_acq.
>
> A wee something like so (I couldn't find an arm64 refcount, even though
> I have distinct memories of talk about it).
>
> This isn't compiled, and obviously needs comment/documentation updates
> to go along with it.
>
> ---
> arch/x86/include/asm/refcount.h | 9 ++++++++-
> lib/refcount.c | 7 ++++++-
> 2 files changed, 14 insertions(+), 2 deletions(-)
>
> diff --git a/arch/x86/include/asm/refcount.h b/arch/x86/include/asm/refcount.h
> index dbaed55c1c24..6f7a1eb345b4 100644
> --- a/arch/x86/include/asm/refcount.h
> +++ b/arch/x86/include/asm/refcount.h
> @@ -74,9 +74,16 @@ bool refcount_sub_and_test(unsigned int i, refcount_t *r)
>
> static __always_inline __must_check bool refcount_dec_and_test(refcount_t *r)
> {
> - return GEN_UNARY_SUFFIXED_RMWcc(LOCK_PREFIX "decl",
> + bool ret = GEN_UNARY_SUFFIXED_RMWcc(LOCK_PREFIX "decl",
>
> REFCOUNT_CHECK_LT_ZERO,
> r-
> >refs.counter, e, "cx");
> +
> + if (ret) {
> + smp_acquire__after_ctrl_dep();
> + return true;
> + }
> +
> + return false;
> }
Actually as I started to do this, any reason why the change here only for
dec_and_test and not
for sub_and _test also? Should not the arch. specific logic follow the generic?
Best Regards,
Elena.
