Re: [PATCH v1 04/26] mm: vmscan: refactor move_folios_to_lru()

[Harry Yoo]

On Wed, Nov 12, 2025 at 08:49:30AM +0100, Sebastian Andrzej Siewior wrote:
> On 2025-11-11 08:44:14 [-0800], Shakeel Butt wrote:
> > On Tue, Nov 11, 2025 at 09:49:00AM +0100, Sebastian Andrzej Siewior wrote:
> > > On 2025-11-11 12:16:43 [+0900], Harry Yoo wrote:
> > > > > However, in the !CONFIG_HAVE_CMPXCHG_LOCAL case, mod_node_page_state()
> > > > > still calls local_irq_save(). Is this feasible in the PREEMPT_RT kernel?
> > > > 
> > > > Hmm I was going to say it's necessary, but AFAICT we don't allocate
> > > > or free memory in hardirq context on PREEMPT_RT (that's the policy)
> > > > and so I'd say it's not necessary to disable IRQs.
> > > > 
> > > > Sounds like we still want to disable IRQs only on !PREEMPT_RT on
> > > > such architectures?
> > > > 
> > > > Not sure how seriously do PREEMPT_RT folks care about architectures
> > > > without HAVE_CMPXCHG_LOCAL. (riscv and loongarch have ARCH_SUPPORTS_RT
> > > > but doesn't have HAVE_CMPXCHG_LOCAL).
> > > 
> > > We take things seriously and you shouldn't make assumption based on
> > > implementation. Either the API can be used as such or not.
> > > In case of mod_node_page_state(), the non-IRQ off version
> > > (__mod_node_page_state()) has a preempt_disable_nested() to ensure
> > > atomic update on PREEMPT_RT without disabling interrupts.
> > 
> > Harry is talking about mod_node_page_state() on
> > !CONFIG_HAVE_CMPXCHG_LOCAL which is disabling irqs.
> > 
> > void mod_node_page_state(struct pglist_data *pgdat, enum node_stat_item item,
> > 					long delta)
> > {
> > 	unsigned long flags;
> > 
> > 	local_irq_save(flags);
> > 	__mod_node_page_state(pgdat, item, delta);
> > 	local_irq_restore(flags);
> > }
> > 
> > Is PREEMPT_RT fine with this?
> 
> Yes.
> The local_irq_save() is not strictly needed but I am fine with it to
> keep it simple. The inner part is just counting.

Yeah I was wondering about this... and thanks for confirming!

-- 
Cheers,
Harry / Hyeonggon