Re: [PATCH v4 03/11] mm: memcontrol: make lruvec lock safe when LRU pages are reparented

[Muchun Song <songmuchun@bytedance.com>]

On Wed, May 25, 2022 at 10:48:54AM -0400, Johannes Weiner wrote:
> On Wed, May 25, 2022 at 09:03:59PM +0800, Muchun Song wrote:
> > On Wed, May 25, 2022 at 08:30:15AM -0400, Johannes Weiner wrote:
> > > On Wed, May 25, 2022 at 05:53:30PM +0800, Muchun Song wrote:
> > > > On Tue, May 24, 2022 at 03:27:20PM -0400, Johannes Weiner wrote:
> > > > > On Tue, May 24, 2022 at 02:05:43PM +0800, Muchun Song wrote:
> > > > > > @@ -1230,10 +1213,23 @@ void lruvec_memcg_debug(struct lruvec *lruvec, struct folio *folio)
> > > > > >   */
> > > > > >  struct lruvec *folio_lruvec_lock(struct folio *folio)
> > > > > >  {
> > > > > > -	struct lruvec *lruvec = folio_lruvec(folio);
> > > > > > +	struct lruvec *lruvec;
> > > > > >  
> > > > > > +	rcu_read_lock();
> > > > > > +retry:
> > > > > > +	lruvec = folio_lruvec(folio);
> > > > > >  	spin_lock(&lruvec->lru_lock);
> > > > > > -	lruvec_memcg_debug(lruvec, folio);
> > > > > > +
> > > > > > +	if (unlikely(lruvec_memcg(lruvec) != folio_memcg(folio))) {
> > > > > > +		spin_unlock(&lruvec->lru_lock);
> > > > > > +		goto retry;
> > > > > > +	}
> > > > > > +
> > > > > > +	/*
> > > > > > +	 * Preemption is disabled in the internal of spin_lock, which can serve
> > > > > > +	 * as RCU read-side critical sections.
> > > > > > +	 */
> > > > > > +	rcu_read_unlock();
> > > > > 
> > > > > The code looks right to me, but I don't understand the comment: why do
> > > > > we care that the rcu read-side continues? With the lru_lock held,
> > > > > reparenting is on hold and the lruvec cannot be rcu-freed anyway, no?
> > > > >
> > > > 
> > > > Right. We could hold rcu read lock until end of reparting.  So you mean
> > > > we do rcu_read_unlock in folio_lruvec_lock()?
> > > 
> > > The comment seems to suggest that disabling preemption is what keeps
> > > the lruvec alive. But it's the lru_lock that keeps it alive. The
> > > cgroup destruction path tries to take the lru_lock long before it even
> > > gets to synchronize_rcu(). Once you hold the lru_lock, having an
> > > implied read-side critical section as well doesn't seem to matter.
> > >
> > 
> > Well, I thought that spinlocks have implicit read-side critical sections
> > because it disables preemption (I learned from the comments above
> > synchronize_rcu() that says interrupts, preemption, or softirqs have been
> > disabled also serve as RCU read-side critical sections).  So I have a
> > question: is it still true in a PREEMPT_RT kernel (I am not familiar with
> > this)?
> 
> Yes, but you're missing my point.
> 
> > > Should the comment be deleted?
> > 
> > I think we could remove the comments. If the above question is false, seems
> > like we should continue holding rcu read lock.
> 
> It's true.
>

Thanks for your answer.

> But assume it's false for a second. Why would you need to continue
> holding it? What would it protect? The lruvec would be pinned by the
> spinlock even if it DIDN'T imply an RCU lock, right?
> 
> So I don't understand the point of the comment. If the implied RCU
> lock is protecting something not covered by the bare spinlock itself,
> it should be added to the comment. Otherwise, the comment should go.
>

Got it. Thanks for your nice explanation. I'll remove
the comment here.