Re: [PATCH] RSS Container, make page_referenced() container aware

[Balbir Singh <balbir@linux.vnet.ibm.com>]

Pavel Emelianov wrote:
> Balbir Singh wrote:
> > Hi, Pavel,
> >
> > This patch should help with the shared page issue of one container
> > holding shared pages in a another container (the container that
> > brought in the page -- by first touch) hostage.
>
> The shared pages accounting is tricky.
> Actually we planned not to do it right now, but later,
> when (if) this will be accepted and we'll move forward
> to the fractions accounting.

Yes, but this patch ensures that we do not hit bottle necks
when pages are shared

> > ------------------------------------------------------------------------
> >
> >
> >
> > Make page_referenced() container aware. Without this patch, page_referenced()
> > can cause a page to be skipped while reclaiming pages. This patch
> > ensures that other containers do not hold pages in a particular container
> > hostage. It is required to ensure that shared pages are freed from a container
> > when they are not actively referenced from the container that brought
> > them in
> >
> > Signed-off-by: Balbir Singh <balbir@linux.vnet.ibm.com>
> > ---
> >
> >  include/linux/rmap.h |    5 +++--
> >  mm/rmap.c            |   26 ++++++++++++++++++++------
> >  mm/vmscan.c          |    4 ++--
> >  3 files changed, 25 insertions(+), 10 deletions(-)
> >
> > diff -puN mm/vmscan.c~rss-implement-per-container-page-referenced mm/vmscan.c
> > --- linux-2.6.20/mm/vmscan.c~rss-implement-per-container-page-referenced	2007-04-26 23:28:44.000000000 +0530
> > +++ linux-2.6.20-balbir/mm/vmscan.c	2007-04-27 00:04:38.000000000 +0530
> > @@ -489,7 +489,7 @@ static unsigned long shrink_page_list(st
> >  		if (PageWriteback(page))
> >  			goto keep_locked;
> >  
> > -		referenced = page_referenced(page, 1);
> > +		referenced = page_referenced(page, 1, sc->cnt);
> >  		/* In active use or really unfreeable?  Activate it. */
> >  		if (referenced && page_mapping_inuse(page))
> >  			goto activate_locked;
> > @@ -852,7 +852,7 @@ force_reclaim_mapped:
> >  		if (page_mapped(page)) {
> >  			if (!reclaim_mapped ||
> >  			    (total_swap_pages == 0 && PageAnon(page)) ||
> > -			    page_referenced(page, 0)) {
> > +			    page_referenced(page, 0, sc->cnt)) {
> >  				list_add(&page->lru, &l_active);
> >  				continue;
> >  			}
> > diff -puN mm/rmap.c~rss-implement-per-container-page-referenced mm/rmap.c
> > --- linux-2.6.20/mm/rmap.c~rss-implement-per-container-page-referenced	2007-04-26 23:28:44.000000000 +0530
> > +++ linux-2.6.20-balbir/mm/rmap.c	2007-04-26 23:33:41.000000000 +0530
> > @@ -318,7 +318,7 @@ out:
> >  	return referenced;
> >  }
> >  
> > -static int page_referenced_anon(struct page *page)
> > +static int page_referenced_anon(struct page *page, struct rss_container *cnt)
> >  {
> >  	unsigned int mapcount;
> >  	struct anon_vma *anon_vma;
> > @@ -331,6 +331,13 @@ static int page_referenced_anon(struct p
> >  
> >  	mapcount = page_mapcount(page);
> >  	list_for_each_entry(vma, &anon_vma->head, anon_vma_node) {
> > +		/*
> > +		 * If we are reclaiming on behalf of a container, skip
> > +		 * counting on behalf of references from different
> > +		 * containers
> > +		 */
> > +		if (cnt && (vma->vm_mm->rss_container != cnt))
> > +			continue;
>
> An #ifdef missed - mm->rss_container makes sense with 
> the CONFIG_RSS_COUNTAINER set only.

Good catch. I'll fix it!

> >  		referenced += page_referenced_one(page, vma, &mapcount);
> >  		if (!mapcount)
> >  			break;
> > @@ -350,7 +357,7 @@ static int page_referenced_anon(struct p
> >   *
> >   * This function is only called from page_referenced for object-based pages.
> >   */
> > -static int page_referenced_file(struct page *page)
> > +static int page_referenced_file(struct page *page, struct rss_container *cnt)
> >  {
> >  	unsigned int mapcount;
> >  	struct address_space *mapping = page->mapping;
> > @@ -383,6 +390,13 @@ static int page_referenced_file(struct p
> >  	mapcount = page_mapcount(page);
> >  
> >  	vma_prio_tree_foreach(vma, &iter, &mapping->i_mmap, pgoff, pgoff) {
> > +		/*
> > +		 * If we are reclaiming on behalf of a container, skip
> > +		 * counting on behalf of references from different
> > +		 * containers
> > +		 */
> > +		if (cnt && (vma->vm_mm->rss_container != cnt))
> > +			continue;
> >  		if ((vma->vm_flags & (VM_LOCKED|VM_MAYSHARE))
> >  				  == (VM_LOCKED|VM_MAYSHARE)) {
> >  			referenced++;
> > @@ -405,7 +419,7 @@ static int page_referenced_file(struct p
> >   * Quick test_and_clear_referenced for all mappings to a page,
> >   * returns the number of ptes which referenced the page.
> >   */
> > -int page_referenced(struct page *page, int is_locked)
> > +int page_referenced(struct page *page, int is_locked, struct rss_container *cnt)
> >  {
> >  	int referenced = 0;
> >  
> > @@ -417,14 +431,14 @@ int page_referenced(struct page *page, i
> >  
> >  	if (page_mapped(page) && page->mapping) {
> >  		if (PageAnon(page))
> > -			referenced += page_referenced_anon(page);
> > +			referenced += page_referenced_anon(page, cnt);
> >  		else if (is_locked)
> > -			referenced += page_referenced_file(page);
> > +			referenced += page_referenced_file(page, cnt);
> >  		else if (TestSetPageLocked(page))
> >  			referenced++;
> >  		else {
> >  			if (page->mapping)
> > -				referenced += page_referenced_file(page);
> > +				referenced += page_referenced_file(page, cnt);
> >  			unlock_page(page);
> >  		}
> >  	}
> > diff -puN include/linux/rmap.h~rss-implement-per-container-page-referenced include/linux/rmap.h
> > --- linux-2.6.20/include/linux/rmap.h~rss-implement-per-container-page-referenced	2007-04-26 23:28:44.000000000 +0530
> > +++ linux-2.6.20-balbir/include/linux/rmap.h	2007-04-26 23:29:31.000000000 +0530
> > @@ -8,6 +8,7 @@
> >  #include <linux/slab.h>
> >  #include <linux/mm.h>
> >  #include <linux/spinlock.h>
> > +#include <linux/rss_container.h>
> >  
> >  /*
> >   * The anon_vma heads a list of private "related" vmas, to scan if
> > @@ -93,7 +94,7 @@ static inline void page_dup_rmap(struct 
> >  /*
> >   * Called from mm/vmscan.c to handle paging out
> >   */
> > -int page_referenced(struct page *, int is_locked);
> > +int page_referenced(struct page *, int is_locked, struct rss_container *cnt);
> >  int try_to_unmap(struct page *, int ignore_refs);
> >  
> >  /*
> > @@ -121,7 +122,7 @@ int page_mkclean(struct page *);
> >  #define anon_vma_prepare(vma)	(0)
> >  #define anon_vma_link(vma)	do {} while (0)
> >  
> > -#define page_referenced(page,l) TestClearPageReferenced(page)
> > +#define page_referenced(page,l,cnt) TestClearPageReferenced(page)
> >  #define try_to_unmap(page, refs) SWAP_FAIL
> >  
> >  static inline int page_mkclean(struct page *page)
> > _


--
	Thanks,
	Balbir Singh
	Linux Technology Center
	IBM, ISTL
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/