diff mbox series

[008/262] ocfs2: do not zero pages beyond i_size

Message ID 20211105203458.Q4tA6BQfK%akpm@linux-foundation.org (mailing list archive)
State New
Headers show
Series [001/262] scripts/spelling.txt: add more spellings to spelling.txt | expand

Commit Message

Andrew Morton Nov. 5, 2021, 8:34 p.m. UTC 
From: Jan Kara <jack@suse.cz>
Subject: ocfs2: do not zero pages beyond i_size

ocfs2_zero_range_for_truncate() can try to zero pages beyond current inode
size despite the fact that underlying blocks should be already zeroed out
and writeback will skip writing such pages anyway.  Avoid the pointless
work.

Link: https://lkml.kernel.org/r/20211025151332.11301-2-jack@suse.cz
Signed-off-by: Jan Kara <jack@suse.cz>
Reviewed-by: Joseph Qi <joseph.qi@linux.alibaba.com>
Cc: Changwei Ge <gechangwei@live.cn>
Cc: Gang He <ghe@suse.com>
Cc: Joel Becker <jlbec@evilplan.org>
Cc: Jun Piao <piaojun@huawei.com>
Cc: Junxiao Bi <junxiao.bi@oracle.com>
Cc: Mark Fasheh <mark@fasheh.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
---

 fs/ocfs2/alloc.c |   19 ++++++++++++-------
 1 file changed, 12 insertions(+), 7 deletions(-)
diff mbox series

Patch

--- a/fs/ocfs2/alloc.c~ocfs2-do-not-zero-pages-beyond-i_size
+++ a/fs/ocfs2/alloc.c
@@ -6923,13 +6923,12 @@  static int ocfs2_grab_eof_pages(struct i
 }
 
 /*
- * Zero the area past i_size but still within an allocated
- * cluster. This avoids exposing nonzero data on subsequent file
- * extends.
+ * Zero partial cluster for a hole punch or truncate. This avoids exposing
+ * nonzero data on subsequent file extends.
  *
  * We need to call this before i_size is updated on the inode because
  * otherwise block_write_full_page() will skip writeout of pages past
- * i_size. The new_i_size parameter is passed for this reason.
+ * i_size.
  */
 int ocfs2_zero_range_for_truncate(struct inode *inode, handle_t *handle,
 				  u64 range_start, u64 range_end)
@@ -6947,6 +6946,15 @@  int ocfs2_zero_range_for_truncate(struct
 	if (!ocfs2_sparse_alloc(OCFS2_SB(sb)))
 		return 0;
 
+	/*
+	 * Avoid zeroing pages fully beyond current i_size. It is pointless as
+	 * underlying blocks of those pages should be already zeroed out and
+	 * page writeback will skip them anyway.
+	 */
+	range_end = min_t(u64, range_end, i_size_read(inode));
+	if (range_start >= range_end)
+		return 0;
+
 	pages = kcalloc(ocfs2_pages_per_cluster(sb),
 			sizeof(struct page *), GFP_NOFS);
 	if (pages == NULL) {
@@ -6955,9 +6963,6 @@  int ocfs2_zero_range_for_truncate(struct
 		goto out;
 	}
 
-	if (range_start == range_end)
-		goto out;
-
 	ret = ocfs2_extent_map_get_blocks(inode,
 					  range_start >> sb->s_blocksize_bits,
 					  &phys, NULL, &ext_flags);