Re: [PATCH v2] tools/memory-model: Add extra ordering for locks and remove it for ordinary release/acquire

[Andrea Parri <andrea.parri@amarulasolutions.com>]

On Tue, Jul 10, 2018 at 01:17:50PM -0400, Alan Stern wrote:
> On Tue, 10 Jul 2018, Daniel Lustig wrote:
> 
> > > --- usb-4.x.orig/tools/memory-model/linux-kernel.cat
> > > +++ usb-4.x/tools/memory-model/linux-kernel.cat
> > > @@ -38,7 +38,7 @@ let strong-fence = mb | gp
> > >  (* Release Acquire *)
> > >  let acq-po = [Acquire] ; po ; [M]
> > >  let po-rel = [M] ; po ; [Release]
> > > -let rfi-rel-acq = [Release] ; rfi ; [Acquire]
> > > +let unlock-rf-lock-po = [UL] ; rf ; [LKR] ; po
> > 
> > It feels slightly weird that unlock-rf-lock-po is asymmetrical.  And in
> > fact, I think the current RISC-V solution we've been discussing (namely,
> > putting a fence.tso instead of a fence rw,w in front of the release)
> > may not even technically respect that particular sequence.  The
> > fence.tso solution really enforces "po; [UL]; rf; [LKR]", right?
> > 
> > Does something like "po; [UL]; rf; [LKR]; po" fit in with the rest
> > of the model?  If so, maybe that solves the asymmetry and also
> > legalizes the approach of putting fence.tso in front?
> 
> That would work just as well.  For this version of the patch it 
> doesn't make any difference, because nothing that comes po-after the 
> LKR is able to directly read the value stored by the UL.

Consider:

C v2-versus-v3

{}

P0(spinlock_t *s, int *x)
{
	spin_lock(s);   /* A */
	spin_unlock(s);
	spin_lock(s);
	WRITE_ONCE(*x, 1); /* B */
	spin_unlock(s);
}

P1(spinlock_t *s, int *x)
{
	int r0;
	int r1;

	r0 = READ_ONCE(*x); /* C */
	smp_rmb();
	r1 = spin_is_locked(s); /* D */
}

With v3, it's allowed that C reads from B and D reads from (the LKW of) A;
this is not allowed with v2 (unless I mis-applied/mis-read v2).

  Andrea