[PATCH] ntfs3: sync uptodate with iomap folio state in read_folio

[Adarsh Das]

ntfs3 registers iomap page-cache helpers for its regular files.  When
the filesystem block size is smaller than the folio size, folios are
backed by iomap_folio_state with a per-block uptodate bitmap in addition
to PG_uptodate.

The i_valid <= vbo early exit in ntfs_read_folio() zeroes folios past
valid data and marks them uptodate with folio_mark_uptodate(), updating
only the folio flag.  That leaves the iomap bitmap out of sync and
ifs_free() WARNs during evict when the folio is torn down.

Use iomap_folio_mark_uptodate() to update both the folio uptodate state
and the iomap uptodate bitmap.

Reported-by: syzbot+3db654d91b471c4d7a7d@xxxxxxxxxxxxxxxxxxxxxxxxx
Link: https://lore.kernel.org/all/6aab8055.71f81b7d.278072.0016.GAE@xxxxxxxxxx/
Tested-by: syzbot+3db654d91b471c4d7a7d@xxxxxxxxxxxxxxxxxxxxxxxxx
Assisted-by: LLM
Signed-off-by: Adarsh Das <adarshdas950@xxxxxxxxx>
---
 fs/ntfs3/inode.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/fs/ntfs3/inode.c b/fs/ntfs3/inode.c
index 56b4f6469a28..a9b4e288ca20 100644
--- a/fs/ntfs3/inode.c
+++ b/fs/ntfs3/inode.c
@@ -842,7 +842,7 @@ static int ntfs_read_folio(struct file *file, struct folio *folio)
 
 	if (ni->i_valid <= vbo) {
 		folio_zero_range(folio, 0, folio_size(folio));
-		folio_mark_uptodate(folio);
+		iomap_folio_mark_uptodate(folio);
 		folio_unlock(folio);
 		return 0;
 	}
-- 
2.55.0