Skip to contentSkip to conversation

f2fs: fix stale nids in an indirect node block

PATCH New Jaegeuk Kim 1 participant 1 message Mar 28, 2016

Subsystems (2)
VFS F2FS file system
lore
Share & tools
lore

Jaegeuk Kim

author maintainer
#0 lore raw

PATCH 1/1 f2fs: fix stale nids in an indirect node block

In truncate_inode_blocks, we can get some stale node block addresses in the following scenario.

1. ipage = get_node_page(ino)
2. unlock_page(ipage)
3. trundate_dnode()
- invalidates all the data indices
- truncate node block
4. lock_page(ipage) and then failed, goto restart
5. -ENOMEM or -EIO between #1 and #3

In this case, we already invalidate node block, but its indirect block still has its nid.

Signed-off-by Jaegeuk Kim <[email protected]>
---
 fs/f2fs/node.c | 12 +++---------
 1 file changed, 3 insertions(+), 9 deletions(-)
fs/f2fs/node.c +3 -9
index 118321b..780ce68 100644
@@ -832,7 +832,7 @@ int truncate_inode_blocks(struct inode *inode, pgoff_t from)
 	trace_f2fs_truncate_inode_blocks_enter(inode, from);
 
 	level = get_node_path(inode, from, offset, noffset);
-restart:
+
 	page = get_node_page(sbi, inode->i_ino);
 	if (IS_ERR(page)) {
 		trace_f2fs_truncate_inode_blocks_exit(inode, PTR_ERR(page));

@@ -840,7 +840,7 @@ restart:
 	}
 
 	set_new_dnode(&dn, inode, page, NULL, 0);
-	unlock_page(page);
+	dn.inode_page_locked = true;
 
 	ri = F2FS_INODE(page);
 	switch (level) {

@@ -895,22 +895,16 @@ skip_partial:
 			goto fail;
 		if (offset[1] == 0 &&
 				ri->i_nid[offset[0] - NODE_DIR1_BLOCK]) {
-			lock_page(page);
-			if (unlikely(page->mapping != NODE_MAPPING(sbi))) {
-				f2fs_put_page(page, 1);
-				goto restart;
-			}
 			f2fs_wait_on_page_writeback(page, NODE, true);
 			ri->i_nid[offset[0] - NODE_DIR1_BLOCK] = 0;
 			set_page_dirty(page);
-			unlock_page(page);
 		}
 		offset[1] = 0;
 		offset[0]++;
 		nofs += err;
 	}
 fail:
-	f2fs_put_page(page, 0);
+	f2fs_put_page(page, 1);
 	trace_f2fs_truncate_inode_blocks_exit(inode, err);
 	return err > 0 ? 0 : err;
 }
Signature · 1 line
2.6.3