[PATCH 1/2] f2fs: clear page's up-to-date if block was deallocated

[PATCH 1/2] f2fs: clear page's up-to-date if block was deallocated

[Jaegeuk Kim]

If page's on-disk block was deallocated, let's remove up-to-date flag to avoid
further access with wrong contents.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/data.c | 6 +++++-
 fs/f2fs/dir.c  | 1 +
 fs/f2fs/node.c | 2 ++
 3 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index ae5c414..62e31b2 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -1259,8 +1259,10 @@ int do_write_data_page(struct page *page, struct f2fs_io_info *fio)
 	fio->blk_addr = dn.data_blkaddr;
 
 	/* This page is already truncated */
-	if (fio->blk_addr == NULL_ADDR)
+	if (fio->blk_addr == NULL_ADDR) {
+		ClearPageUptodate(page);
 		goto out_writepage;
+	}
 
 	set_page_writeback(page);
 
@@ -1355,6 +1357,8 @@ done:
 	clear_cold_data(page);
 out:
 	inode_dec_dirty_pages(inode);
+	if (err)
+		ClearPageUptodate(page);
 	unlock_page(page);
 	if (need_balance_fs)
 		f2fs_balance_fs(sbi);
diff --git a/fs/f2fs/dir.c b/fs/f2fs/dir.c
index 583896c..590aeef 100644
--- a/fs/f2fs/dir.c
+++ b/fs/f2fs/dir.c
@@ -682,6 +682,7 @@ void f2fs_delete_entry(struct f2fs_dir_entry *dentry, struct page *page,
 	if (bit_pos == NR_DENTRY_IN_BLOCK) {
 		truncate_hole(dir, page->index, page->index + 1);
 		clear_page_dirty_for_io(page);
+		ClearPagePrivate(page);
 		ClearPageUptodate(page);
 		inode_dec_dirty_pages(dir);
 	}
diff --git a/fs/f2fs/node.c b/fs/f2fs/node.c
index 2842369..35a9117 100644
--- a/fs/f2fs/node.c
+++ b/fs/f2fs/node.c
@@ -1002,6 +1002,7 @@ static int read_node_page(struct page *page, int rw)
 	get_node_info(sbi, page->index, &ni);
 
 	if (unlikely(ni.blk_addr == NULL_ADDR)) {
+		ClearPageUptodate(page);
 		f2fs_put_page(page, 1);
 		return -ENOENT;
 	}
@@ -1313,6 +1314,7 @@ static int f2fs_write_node_page(struct page *page,
 
 	/* This page is already truncated */
 	if (unlikely(ni.blk_addr == NULL_ADDR)) {
+		ClearPageUptodate(page);
 		dec_page_count(sbi, F2FS_DIRTY_NODES);
 		unlock_page(page);
 		return 0;
-- 
2.1.1


------------------------------------------------------------------------------
Dive into the World of Parallel Programming The Go Parallel Website, sponsored
by Intel and developed in partnership with Slashdot Media, is your hub for all
things parallel software development, from weekly thought leadership blogs to
news, videos, case studies, tutorials and more. Take a look and join the 
conversation now. http://goparallel.sourceforge.net/

[PATCH 2/2] f2fs: check its block allocation to avoid producing wrong dirty pages

[Jaegeuk Kim]

If a page is cached but its block was deallocated, we don't need to make
the page dirty again by gc and truncate_partial_data_page.

In that case, it needs to check its block allocation all the time instead
of giving up-to-date page.

Signed-off-by: Jaegeuk Kim <jaegeuk@kernel.org>
---
 fs/f2fs/data.c | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/fs/f2fs/data.c b/fs/f2fs/data.c
index 62e31b2..389fda7 100644
--- a/fs/f2fs/data.c
+++ b/fs/f2fs/data.c
@@ -813,11 +813,19 @@ struct page *find_data_page(struct inode *inode, pgoff_t index, bool sync)
 		.rw = sync ? READ_SYNC : READA,
 	};
 
+	/*
+	 * If sync is false, it needs to check its block allocation.
+	 * This is need and triggered by two flows:
+	 *   gc and truncate_partial_data_page.
+	 */
+	if (!sync)
+		goto search;
+
 	page = find_get_page(mapping, index);
 	if (page && PageUptodate(page))
 		return page;
 	f2fs_put_page(page, 0);
-
+search:
 	if (f2fs_lookup_extent_cache(inode, index, &ei)) {
 		dn.data_blkaddr = ei.blk + index - ei.fofs;
 		goto got_it;
-- 
2.1.1


------------------------------------------------------------------------------
Dive into the World of Parallel Programming The Go Parallel Website, sponsored
by Intel and developed in partnership with Slashdot Media, is your hub for all
things parallel software development, from weekly thought leadership blogs to
news, videos, case studies, tutorials and more. Take a look and join the 
conversation now. http://goparallel.sourceforge.net/