f2fs: drop obsolete node page when it is truncated
authorJaegeuk Kim <jaegeuk.kim@samsung.com>
Wed, 22 Jan 2014 11:41:57 +0000 (20:41 +0900)
committerJaegeuk Kim <jaegeuk.kim@samsung.com>
Wed, 22 Jan 2014 23:04:21 +0000 (08:04 +0900)
If a node page is trucated, we'd better drop the page in the node_inode's page
cache for better memory footprint.

Signed-off-by: Jaegeuk Kim <jaegeuk.kim@samsung.com>
fs/f2fs/node.c

index bbfc655493dcc76bd0b5ae09ead1a1a6ffdf45fc..b0649b76eb4f390bea393344af07a9e65dd32211 100644 (file)
@@ -518,6 +518,10 @@ invalidate:
        F2FS_SET_SB_DIRT(sbi);
 
        f2fs_put_page(dn->node_page, 1);
+
+       invalidate_mapping_pages(NODE_MAPPING(sbi),
+                       dn->node_page->index, dn->node_page->index);
+
        dn->node_page = NULL;
        trace_f2fs_truncate_node(dn->inode, dn->nid, ni.blk_addr);
 }