Re: [PATCH v2] writeback: use exact memcg dirty counts

[Andrew Morton <akpm@linux-foundation.org>]

On Mon, 1 Apr 2019 14:20:44 -0400 Johannes Weiner <hannes@cmpxchg.org> wrote:

> On Fri, Mar 29, 2019 at 10:46:09AM -0700, Greg Thelen wrote:
> > @@ -3907,10 +3923,10 @@ void mem_cgroup_wb_stats(struct bdi_writeback *wb, unsigned long *pfilepages,
> >  	struct mem_cgroup *memcg = mem_cgroup_from_css(wb->memcg_css);
> >  	struct mem_cgroup *parent;
> >  
> > -	*pdirty = memcg_page_state(memcg, NR_FILE_DIRTY);
> > +	*pdirty = memcg_exact_page_state(memcg, NR_FILE_DIRTY);
> >  
> >  	/* this should eventually include NR_UNSTABLE_NFS */
> > -	*pwriteback = memcg_page_state(memcg, NR_WRITEBACK);
> > +	*pwriteback = memcg_exact_page_state(memcg, NR_WRITEBACK);
> >  	*pfilepages = mem_cgroup_nr_lru_pages(memcg, (1 << LRU_INACTIVE_FILE) |
> >  						     (1 << LRU_ACTIVE_FILE));
> 
> Andrew,
> 
> just a head-up: -mm has that LRU stat cleanup series queued ("mm:
> memcontrol: clean up the LRU counts tracking") that changes the
> mem_cgroup_nr_lru_pages() call here to two memcg_page_state().
> 
> I'm assuming Greg's fix here will get merged before the cleanup. When
> it gets picked up, it'll conflict with "mm: memcontrol: push down
> mem_cgroup_nr_lru_pages()".
> 
> "mm: memcontrol: push down mem_cgroup_nr_lru_pages()" will need to be
> changed to use memcg_exact_page_state() calls instead of the plain
> memcg_page_state() for *pfilepages.
> 

Thanks.  Like this?

void mem_cgroup_wb_stats(struct bdi_writeback *wb, unsigned long *pfilepages,
			 unsigned long *pheadroom, unsigned long *pdirty,
			 unsigned long *pwriteback)
{
	struct mem_cgroup *memcg = mem_cgroup_from_css(wb->memcg_css);
	struct mem_cgroup *parent;

	*pdirty = memcg_exact_page_state(memcg, NR_FILE_DIRTY);

	/* this should eventually include NR_UNSTABLE_NFS */
	*pwriteback = memcg_exact_page_state(memcg, NR_WRITEBACK);
	*pfilepages = memcg_exact_page_state(memcg, NR_INACTIVE_FILE) +
			memcg_exact_page_state(memcg, NR_ACTIVE_FILE);
	*pheadroom = PAGE_COUNTER_MAX;

	while ((parent = parent_mem_cgroup(memcg))) {
		unsigned long ceiling = min(memcg->memory.max, memcg->high);
		unsigned long used = page_counter_read(&memcg->memory);

		*pheadroom = min(*pheadroom, ceiling - min(ceiling, used));
		memcg = parent;
	}
}