Re: [PATCH v1 25/26] mm: memcontrol: eliminate the problem of dying memory cgroup for LRU folios

[Qi Zheng]

On 11/20/25 7:56 PM, Chen Ridong wrote:
> On 2025/10/28 21:58, Qi Zheng wrote:
> >   static void reparent_locks(struct mem_cgroup *src, struct mem_cgroup *dst)
> >   {
> > +	int nid, nest = 0;
> > +
> >   	spin_lock_irq(&objcg_lock);
> > +	for_each_node(nid) {
> > +		spin_lock_nested(&mem_cgroup_lruvec(src,
> > +				 NODE_DATA(nid))->lru_lock, nest++);
> > +		spin_lock_nested(&mem_cgroup_lruvec(dst,
> > +				 NODE_DATA(nid))->lru_lock, nest++);
> > +	}
> >   }
> >   
> >   static void reparent_unlocks(struct mem_cgroup *src, struct mem_cgroup *dst)
> >   {
> > +	int nid;
> > +
> > +	for_each_node(nid) {
> > +		spin_unlock(&mem_cgroup_lruvec(dst, NODE_DATA(nid))->lru_lock);
> > +		spin_unlock(&mem_cgroup_lruvec(src, NODE_DATA(nid))->lru_lock);
> > +	}
> >   	spin_unlock_irq(&objcg_lock);
> >   }
> >   
>
> The lock order follows S0→D0→S1→D1→…, and the correct unlock sequence should be Dn→Sn→…→D1→S0
>
> However, the current unlock implementation uses D0→S0→D1→S1→…
>
> I’m not certain whether this unlock order will cause any issues—could this lead to potential
> problems like deadlocks or lock state inconsistencies?

As long as the order in which the locks are held is consistent, there 
should be no deadlock problem?

>