[PATCH] mm: optimize replace_page_cache_page

[Minchan Kim <minchan.kim@gmail.com>]

This patch optmizes replace_page_cache_page.

1) remove radix_tree_preload
2) single radix_tree_lookup_slot and replace radix tree slot
3) page accounting optimization if both pages are in same zone.

Cc: Miklos Szeredi <mszeredi@suse.cz>
Cc: Rik van Riel <riel@redhat.com>
Cc: KAMEZAWA Hiroyuki <kamezawa.hiroyu@jp.fujitsu.com>
Cc: Mel Gorman <mel@csn.ul.ie>
Signed-off-by: Minchan Kim <minchan.kim@gmail.com>
---
 mm/filemap.c |   61 ++++++++++++++++++++++++++++++++++++++++++++++++---------
 1 files changed, 51 insertions(+), 10 deletions(-)

Hi Miklos,
This patch is totally not tested.
Could you test this patch?

diff --git a/mm/filemap.c b/mm/filemap.c
index a25c898..918ef1e 100644
--- a/mm/filemap.c
+++ b/mm/filemap.c
@@ -148,6 +148,56 @@ void __remove_from_page_cache(struct page *page)
 	}
 }
 
+/*
+ * Replace a page from the page cache with a new one.
+ * Both the old and new page must be locked.
+ * The caller must hold the mapping's tree_lock.
+ */
+void __replace_page_cache_page(struct page *old, struct page *new)
+{
+	void **pslot;
+	struct address_space *mapping = old->mapping;
+	struct zone *old_zone, *new_zone;
+
+	old_zone = page_zone(old);
+	new_zone = page_zone(new);
+	/*
+	 * if we're uptodate, flush out into the cleancache, otherwise
+	 * invalidate any existing cleancache entries.  We can't leave
+	 * stale data around in the cleancache once our page is gone
+	 */
+	if (PageUptodate(old))
+		cleancache_put_page(old);
+	else
+		cleancache_flush_page(mapping, old);
+
+	pslot = radix_tree_lookup_slot(&mapping->page_tree,
+			page_index(old));
+	get_page(new);      /* add cache reference */
+	radix_tree_replace_slot(pslot, new);
+	old->mapping = NULL;
+	if (old_zone != new_zone) {
+		__dec_zone_page_state(old, NR_FILE_PAGES);
+		__inc_zone_page_state(new, NR_FILE_PAGES);
+		if (PageSwapBacked(old)) {
+			__dec_zone_page_state(old, NR_SHMEM);
+			__inc_zone_page_state(new, NR_SHMEM);
+		}
+	}
+	BUG_ON(page_mapped(old));
+	/*
+	 * Some filesystems seem to re-dirty the page even after
+	 * the VM has canceled the dirty bit (eg ext3 journaling).
+	 *
+	 * Fix it up by doing a final dirty accounting check after
+	 * having removed the page entirely.
+	 */
+	if (PageDirty(old) && mapping_cap_account_dirty(mapping)) {
+		dec_zone_page_state(old, NR_FILE_DIRTY);
+		dec_bdi_stat(mapping->backing_dev_info, BDI_RECLAIMABLE);
+	}
+}
+
 void remove_from_page_cache(struct page *page)
 {
 	struct address_space *mapping = page->mapping;
@@ -433,7 +483,6 @@ int replace_page_cache_page(struct page *old, struct page *new, gfp_t gfp_mask)
 	if (error)
 		return error;
 
-	error = radix_tree_preload(gfp_mask & ~__GFP_HIGHMEM);
 	if (!error) {
 		struct address_space *mapping = old->mapping;
 		void (*freepage)(struct page *);
@@ -441,20 +490,12 @@ int replace_page_cache_page(struct page *old, struct page *new, gfp_t gfp_mask)
 		pgoff_t offset = old->index;
 		freepage = mapping->a_ops->freepage;
 
-		page_cache_get(new);
 		new->mapping = mapping;
 		new->index = offset;
 
 		spin_lock_irq(&mapping->tree_lock);
-		__remove_from_page_cache(old);
-		error = radix_tree_insert(&mapping->page_tree, offset, new);
-		BUG_ON(error);
-		mapping->nrpages++;
-		__inc_zone_page_state(new, NR_FILE_PAGES);
-		if (PageSwapBacked(new))
-			__inc_zone_page_state(new, NR_SHMEM);
+		__replace_page_cache_page(old, new);
 		spin_unlock_irq(&mapping->tree_lock);
-		radix_tree_preload_end();
 		if (freepage)
 			freepage(old);
 		page_cache_release(old);
-- 
1.7.1

--
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/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>