Re: Low overhead patches for the memory cgroup controller (v4)

[KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>]

On Mon, 15 Jun 2009 00:07:40 +0530
Balbir Singh <balbir@linux.vnet.ibm.com> wrote:

> Here is v4 of the patches, please review and comment
> 
> Feature: Remove the overhead associated with the root cgroup
> 
> From: Balbir Singh <balbir@linux.vnet.ibm.com>
> 
> changelog v4 -> v3
> 1. Rebase to mmotm 9th june 2009
> 2. Remove PageCgroupRoot, we have account LRU flags to indicate that
>    we do only accounting and no reclaim.
> 3. pcg_default_flags has been used again, since PCGF_ROOT is gone,
>    we set PCGF_ACCT_LRU only in mem_cgroup_add_lru_list
> 4. More LRU functions are aware of PageCgroupAcctLRU
> 
> Changelog v3 -> v2
> 
> 1. Rebase to mmotm 2nd June 2009
> 2. Test with some of the test cases recommended by Daisuke-San
> 
> Changelog v2 -> v1
> 1. Rebase to latest mmotm
> 
> This patch changes the memory cgroup and removes the overhead associated
> with accounting all pages in the root cgroup. As a side-effect, we can
> no longer set a memory hard limit in the root cgroup.
> 
> A new flag to track whether the page has been accounted or not
> has been added as well. Flags are now set atomically for page_cgroup,
> 
> Tests:
> 
> Results (for v2)
> 
> Obtained by
> 
> 1. Using tmpfs for mounting filesystem
> 2. Changing sync to be /bin/true (so that sync is not the bottleneck)
> 3. Used -s #cpus*40 -e #cpus*40
> 
> Reaim
> 		withoutpatch	patch
> AIM9		9532.48		9807.59
> dbase		19344.60	19285.71
> new_dbase	20101.65	20163.13
> shared		11827.77	11886.65
> compute		17317.38	17420.05
> 

Hmm, how much overhead this patch adds for non-root cgroup ?
It seems getting better in general. But I have a few suggestions.


> Signed-off-by: Balbir Singh <balbir@linux.vnet.ibm.com>
> ---
> 
>  include/linux/page_cgroup.h |    5 ++++
>  mm/memcontrol.c             |   59 ++++++++++++++++++++++++++++++++++++-------
>  2 files changed, 54 insertions(+), 10 deletions(-)
> 
> 
> diff --git a/include/linux/page_cgroup.h b/include/linux/page_cgroup.h
> index 7339c7b..57c4d50 100644
> --- a/include/linux/page_cgroup.h
> +++ b/include/linux/page_cgroup.h
> @@ -26,6 +26,7 @@ enum {
>  	PCG_LOCK,  /* page cgroup is locked */
>  	PCG_CACHE, /* charged as cache */
>  	PCG_USED, /* this object is in use. */
> +	PCG_ACCT_LRU, /* page has been accounted for */
>  };
>  
>  #define TESTPCGFLAG(uname, lname)			\
> @@ -46,6 +47,10 @@ TESTPCGFLAG(Cache, CACHE)
>  TESTPCGFLAG(Used, USED)
>  CLEARPCGFLAG(Used, USED)
>  
> +SETPCGFLAG(AcctLRU, ACCT_LRU)
> +CLEARPCGFLAG(AcctLRU, ACCT_LRU)
> +TESTPCGFLAG(AcctLRU, ACCT_LRU)
> +
>  static inline int page_cgroup_nid(struct page_cgroup *pc)
>  {
>  	return page_to_nid(pc->page);
> diff --git a/mm/memcontrol.c b/mm/memcontrol.c
> index 6ceb6f2..399d416 100644
> --- a/mm/memcontrol.c
> +++ b/mm/memcontrol.c
> @@ -43,6 +43,7 @@
>  
>  struct cgroup_subsys mem_cgroup_subsys __read_mostly;
>  #define MEM_CGROUP_RECLAIM_RETRIES	5
> +struct mem_cgroup *root_mem_cgroup __read_mostly;
>  
>  #ifdef CONFIG_CGROUP_MEM_RES_CTLR_SWAP
>  /* Turned on only when memory cgroup is enabled && really_do_swap_account = 1 */
> @@ -219,6 +220,11 @@ static void mem_cgroup_get(struct mem_cgroup *mem);
>  static void mem_cgroup_put(struct mem_cgroup *mem);
>  static struct mem_cgroup *parent_mem_cgroup(struct mem_cgroup *mem);
>  
> +static inline bool mem_cgroup_is_root(struct mem_cgroup *mem)
> +{
> +	return (mem == root_mem_cgroup);
> +}
> +
>  static void mem_cgroup_charge_statistics(struct mem_cgroup *mem,
>  					 struct page_cgroup *pc,
>  					 bool charge)
> @@ -378,15 +384,25 @@ void mem_cgroup_del_lru_list(struct page *page, enum lru_list lru)
>  		return;
>  	pc = lookup_page_cgroup(page);
>  	/* can happen while we handle swapcache. */
> -	if (list_empty(&pc->lru) || !pc->mem_cgroup)
> +	mem = pc->mem_cgroup;
> +	if (!mem)
> +		return;
> +	if (mem_cgroup_is_root(mem)) {
> +		if (!PageCgroupAcctLRU(pc))
> +			return;
> +	} else if (list_empty(&pc->lru))
>  		return;
> +
>  	/*
>  	 * We don't check PCG_USED bit. It's cleared when the "page" is finally
>  	 * removed from global LRU.
>  	 */
>  	mz = page_cgroup_zoneinfo(pc);
> -	mem = pc->mem_cgroup;
>  	MEM_CGROUP_ZSTAT(mz, lru) -= 1;
> +	if (PageCgroupAcctLRU(pc)) {
> +		ClearPageCgroupAcctLRU(pc);
> +		return;
> +	}
>  	list_del_init(&pc->lru);
>  	return;
>  }
Looking through the whole code, PageCgroupAcctLRU() is meaningful only when
pc->mem_cgroup == root_mem_cgroup.  Right ?

I wonder making PageCgroupAcctLRU() be always meaningful and remove all
!list_empty(&pc->lru) check is a way to go.

If do so, this function can be written as

==
	if (!PageCgroupAcctLRU(pc))
		return;
	mem = pc->mem_cgroup;
	mz = page_cgroup_zoneinfo(pc);
	MEM_CGROUP_ZSTAT(mz, lru) -= 1;
	ClearPageCgroupAcctLRU(pc);
	/* We don't maintain LRU for root cgroup. Global LRU works for us. */
	if (!mem_cgroup_is_root(mem))
		list_del_init(&pc->lru);
==
This seems much straightforward. 

> @@ -410,8 +426,8 @@ void mem_cgroup_rotate_lru_list(struct page *page, enum lru_list lru)
>  	 * For making pc->mem_cgroup visible, insert smp_rmb() here.
>  	 */
>  	smp_rmb();
> -	/* unused page is not rotated. */
> -	if (!PageCgroupUsed(pc))
> +	/* unused or root page is not rotated. */
> +	if (!PageCgroupUsed(pc) || PageCgroupAcctLRU(pc))
>  		return;
>  	mz = page_cgroup_zoneinfo(pc);
>  	list_move(&pc->lru, &mz->lists[lru]);
> @@ -435,6 +451,10 @@ void mem_cgroup_add_lru_list(struct page *page, enum lru_list lru)
>  
>  	mz = page_cgroup_zoneinfo(pc);
>  	MEM_CGROUP_ZSTAT(mz, lru) += 1;
> +	if (mem_cgroup_is_root(pc->mem_cgroup)) {
> +		SetPageCgroupAcctLRU(pc);
> +		return;
> +	}
>  	list_add(&pc->lru, &mz->lists[lru]);
>  }
With above (my) rule.   Here will be
	SetPageCgroupAcctLRU(pc);
	if (!mem_cgroup_is_root(pc->mem_cgroup))
		list_add(&pc->lru, &mz->lists[lru]);

> @@ -445,12 +465,15 @@ void mem_cgroup_add_lru_list(struct page *page, enum lru_list lru)
>   * it again. This function is only used to charge SwapCache. It's done under
>   * lock_page and expected that zone->lru_lock is never held.
>   */
> -static void mem_cgroup_lru_del_before_commit_swapcache(struct page *page)
> +static void mem_cgroup_lru_del_before_commit_swapcache(struct page *page,
> +							struct page_cgroup *pc)
>  {
>  	unsigned long flags;
>  	struct zone *zone = page_zone(page);
> -	struct page_cgroup *pc = lookup_page_cgroup(page);
>  
> +	if (!pc->mem_cgroup ||
> +		(!PageCgroupAcctLRU(pc) && mem_cgroup_is_root(pc->mem_cgroup)))
> +		return;
PageCgroupAcctLRU() check is done without zone->lock and this is racy if you check
flag. Considering how "pagevec" works, this race tend to be big.


>  	spin_lock_irqsave(&zone->lru_lock, flags);
>  	/*
>  	 * Forget old LRU when this page_cgroup is *not* used. This Used bit
> @@ -461,12 +484,15 @@ static void mem_cgroup_lru_del_before_commit_swapcache(struct page *page)
>  	spin_unlock_irqrestore(&zone->lru_lock, flags);
>  }
>  
> -static void mem_cgroup_lru_add_after_commit_swapcache(struct page *page)
> +static void mem_cgroup_lru_add_after_commit_swapcache(struct page *page,
> +							struct page_cgroup *pc)
>  {
>  	unsigned long flags;
>  	struct zone *zone = page_zone(page);
> -	struct page_cgroup *pc = lookup_page_cgroup(page);
>  
> +	if (!pc->mem_cgroup ||
> +		(!PageCgroupAcctLRU(pc) && mem_cgroup_is_root(pc->mem_cgroup)))
> +		return;

The same comment as above.


>  	spin_lock_irqsave(&zone->lru_lock, flags);
>  	/* link when the page is linked to LRU but page_cgroup isn't */
>  	if (PageLRU(page) && list_empty(&pc->lru))
> @@ -478,8 +504,13 @@ static void mem_cgroup_lru_add_after_commit_swapcache(struct page *page)
>  void mem_cgroup_move_lists(struct page *page,
>  			   enum lru_list from, enum lru_list to)
>  {
> +	struct page_cgroup *pc = lookup_page_cgroup(page);
>  	if (mem_cgroup_disabled())
>  		return;
> +	smp_rmb();
> +	if (!pc->mem_cgroup ||
> +		(!PageCgroupAcctLRU(pc) && mem_cgroup_is_root(pc->mem_cgroup)))
> +		return;
>  	mem_cgroup_del_lru_list(page, from);
>  	mem_cgroup_add_lru_list(page, to);
>  }
Here, too.


> @@ -1114,6 +1145,7 @@ static void __mem_cgroup_commit_charge(struct mem_cgroup *mem,
>  		css_put(&mem->css);
>  		return;
>  	}
> +
>  	pc->mem_cgroup = mem;
>  	smp_wmb();
>  	pc->flags = pcg_default_flags[ctype];
> @@ -1418,9 +1450,10 @@ __mem_cgroup_commit_charge_swapin(struct page *page, struct mem_cgroup *ptr,
>  	if (!ptr)
>  		return;
>  	pc = lookup_page_cgroup(page);
> -	mem_cgroup_lru_del_before_commit_swapcache(page);
> +	smp_rmb();
> +	mem_cgroup_lru_del_before_commit_swapcache(page, pc);
>  	__mem_cgroup_commit_charge(ptr, pc, ctype);
> -	mem_cgroup_lru_add_after_commit_swapcache(page);
> +	mem_cgroup_lru_add_after_commit_swapcache(page, pc);

Why this change ? When you adds memory barrier, plz add comments.


>  	/*
>  	 * Now swap is on-memory. This means this page may be
>  	 * counted both as mem and swap....double count.
> @@ -2055,6 +2088,10 @@ static int mem_cgroup_write(struct cgroup *cont, struct cftype *cft,
>  	name = MEMFILE_ATTR(cft->private);
>  	switch (name) {
>  	case RES_LIMIT:
> +		if (mem_cgroup_is_root(memcg)) { /* Can't set limit on root */
> +			ret = -EINVAL;
> +			break;
> +		}

Could you add modification to Documentation in the next post ?


>  		/* This function does all necessary parse...reuse it */
>  		ret = res_counter_memparse_write_strategy(buffer, &val);
>  		if (ret)
> @@ -2521,6 +2558,7 @@ mem_cgroup_create(struct cgroup_subsys *ss, struct cgroup *cont)
>  	if (cont->parent == NULL) {
>  		enable_swap_cgroup();
>  		parent = NULL;
> +		root_mem_cgroup = mem;
>  	} else {
>  		parent = mem_cgroup_from_cont(cont->parent);
>  		mem->use_hierarchy = parent->use_hierarchy;
> @@ -2549,6 +2587,7 @@ mem_cgroup_create(struct cgroup_subsys *ss, struct cgroup *cont)
>  	return &mem->css;
>  free_out:
>  	__mem_cgroup_free(mem);
> +	root_mem_cgroup = NULL;
>  	return ERR_PTR(error);
>  }
>  

Could you start next thread in the next post ? Once I read and make this from
unread to read, this goes far deep of old mail tree ;)


Regards,
-Kame

--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org.  For more info on Linux MM,
see: http://www.linux-mm.org/ .
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>