Re: [PATCH v13 2/6] mm/vmstat: Use vmstat_dirty to track CPU-specific vmstat discrepancies

[Frederic Weisbecker]

On Tue, Jan 10, 2023 at 01:06:37PM +0100, Christoph Lameter wrote:
> On Thu, 5 Jan 2023, Marcelo Tosatti wrote:
> > @@ -577,6 +579,9 @@ static inline void mod_zone_state(struct
> >  	s8 __percpu *p = pcp->vm_stat_diff + item;
> >  	long o, n, t, z;
> >
> > +	/* cmpxchg and vmstat_mark_dirty should happen on the same CPU */
> > +	preempt_disable();
> 
> If you are disabling preemption then why do we still need cmpxchg?
> Same again below.

Note I'm absolutely clueless with vmstat. But I was wondering about it as well
while reviewing Marcelo's series, so git blame pointed me to:

7c83912062c801738d7d19acaf8f7fec25ea663c ("vmstat: User per cpu atomics to avoid
interrupt disable / enable")

And this seem to mention that this can race with IRQs as well, hence the local
cmpxchg operation.