Re: [RFC][PATCH] mips: Fix arch_spin_unlock()

[Will Deacon]

On Tue, Feb 02, 2016 at 08:12:30PM +0800, Boqun Feng wrote:
> On Tue, Feb 02, 2016 at 11:45:59AM +0000, Will Deacon wrote:
> > On Tue, Feb 02, 2016 at 01:19:04PM +0800, Boqun Feng wrote:
> > > On Mon, Feb 01, 2016 at 07:54:58PM -0800, Paul E. McKenney wrote:
> > > > On Mon, Feb 01, 2016 at 01:56:22PM +0000, Will Deacon wrote:
> > > > > On Fri, Jan 29, 2016 at 02:22:53AM -0800, Paul E. McKenney wrote:
> > > > > > On Fri, Jan 29, 2016 at 09:59:59AM +0000, Will Deacon wrote:
> > > > > Locally transitive chain termination:
> > > > > 
> > > > > (i.e. these can't be used to extend a chain)
> > > > 
> > > > Agreed.
> > > > 
> > > > > > o	smp_store_release() -> lockless_dereference() (???)
> > > > > > o	rcu_assign_pointer() -> rcu_dereference()
> > > > > > o	smp_store_release() -> READ_ONCE(); if
> > > 
> > > Just want to make sure, this one is actually:
> > > 
> > > o	smp_store_release() -> READ_ONCE(); if ;<WRITE_ONCE()>
> > > 
> > > right? Because control dependency only orders READ->WRITE.
> > > 
> > > If so, do we also need to take the following pairing into consideration?
> > > 
> > > o	smp_store_release() -> READ_ONCE(); if ;smp_rmb(); <ACCESS_ONCE()>
> > > 
> > > > 
> > > > I am OK with the first and last, but I believe that the middle one
> > > > has real use cases.  So the rcu_assign_pointer() -> rcu_dereference()
> > > > case needs to be locally transitive.
> > > > 
> > > 
> > > Hmm... I don't think we should differ rcu_dereference() and
> > > lockless_dereference(). One reason: list_for_each_entry_rcu() are using
> > > lockless_dereference() right now, which means we used to think
> > > rcu_dereference() and lockless_dereference() are interchangeable, right?
> > > 
> > > Besides, Will, what's the reason of having a locally transitive chain
> > > termination? Because on some architectures RELEASE->DEPENDENCY pairs may
> > > not be locally transitive?
> > 
> > Well, the following ISA2 test is permitted on ARM:
> > 
> > 
> > P0:
> > Wx=1
> > WyRel=1	// rcu_assign_pointer
> > 
> > P1:
> > Ry=1	// rcu_dereference
> 
> What if a <addr> dependency is added here? Same result?

Right, that fixes it. So if we're only considering things like:

  rcu_dereference
  <addr>
  RELEASE

then local transitivity should be preserved.

I think the same applies to <ctrl>, which seems to match your later
example.

Will