[RFC PATCH v4 28/34] ext4: writeback partial blocks before zeroing out range
From: Zhang Yi
Date: Wed Apr 10 2024 - 09:44:58 EST
From: Zhang Yi <yi.zhang@xxxxxxxxxx>
If we zero partial blocks, iomap_zero_iter() will skip zeroing out the
IOMAP_UNWRITTEN srcmap, it works fine in xfs because this type means the
block is pure unwritten and doesn't contain any delayed data. But it
doesn't work in ext4, because IOMAP_UNWRITTEN may contain delayed data
in ext4. For now it's hard to unify the meaning of this flag, so just
fix it by writeback partial blocks before zeroing out.
Signed-off-by: Zhang Yi <yi.zhang@xxxxxxxxxx>
---
fs/ext4/extents.c | 9 +++++++++
1 file changed, 9 insertions(+)
diff --git a/fs/ext4/extents.c b/fs/ext4/extents.c
index 9849947cec56..c4c38a323ff7 100644
--- a/fs/ext4/extents.c
+++ b/fs/ext4/extents.c
@@ -4580,6 +4580,15 @@ static long ext4_zero_range(struct file *file, loff_t offset,
if (ret)
goto out_mutex;
+ ret = filemap_write_and_wait_range(mapping,
+ round_down(offset, 1 << blkbits), offset);
+ if (ret)
+ goto out_mutex;
+
+ ret = filemap_write_and_wait_range(mapping, offset + len,
+ round_up((offset + len), 1 << blkbits));
+ if (ret)
+ goto out_mutex;
}
/* Zero range excluding the unaligned edges */
--
2.39.2