linux-kernel - Forwarded: [PATCH] ext4: fix missing i_data_sem lock in move

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <6936c882.a70a0220.38f243.00a5.GAE@google.com>
Date: Mon, 08 Dec 2025 04:45:54 -0800
From: syzbot <syzbot+4ea6bd8737669b423aae@...kaller.appspotmail.com>
To: linux-kernel@...r.kernel.org, syzkaller-bugs@...glegroups.com
Subject: Forwarded: [PATCH] ext4: fix missing i_data_sem lock in move_extent
 repair path

For archival purposes, forwarding an incoming command email to
linux-kernel@...r.kernel.org, syzkaller-bugs@...glegroups.com.

***

Subject: [PATCH] ext4: fix missing i_data_sem lock in move_extent repair path
Author: kartikey406@...il.com

#syz test: git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git master

ext4_swap_extents() requires both i_data_sem locks to be held, as
enforced by BUG_ON checks at the function entry. However, the
repair_branches error path in mext_move_extent() calls
ext4_swap_extents() without holding these locks.

When mext_folio_mkwrite() fails after a successful extent swap, the
code jumps to repair_branches to restore the original extent mapping.
At this point, i_data_sem has already been released, causing the
BUG_ON in ext4_swap_extents() to trigger.

Fix this by acquiring i_data_sem for both inodes before calling
ext4_swap_extents() in the repair_branches path, matching the
locking pattern used in the normal swap path.

Reported-by: syzbot+4ea6bd8737669b423aae@...kaller.appspotmail.com
Closes: https://syzkaller.appspot.com/bug?extid=4ea6bd8737669b423aae
Signed-off-by: Deepanshu Kartikey <kartikey406@...il.com>
---
 fs/ext4/move_extent.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fs/ext4/move_extent.c b/fs/ext4/move_extent.c
index 0550fd30fd10..635fb8a52e0c 100644
--- a/fs/ext4/move_extent.c
+++ b/fs/ext4/move_extent.c
@@ -393,9 +393,11 @@ static int mext_move_extent(struct mext_data *mext, u64 *m_len)

 repair_branches:
 	ret2 = 0;
+	ext4_double_down_write_data_sem(orig_inode, donor_inode);
 	r_len = ext4_swap_extents(handle, donor_inode, orig_inode,
 				  mext->donor_lblk, orig_map->m_lblk,
 				  *m_len, 0, &ret2);
+	ext4_double_up_write_data_sem(orig_inode, donor_inode);
 	if (ret2 || r_len != *m_len) {
 		ext4_error_inode_block(orig_inode, (sector_t)(orig_map->m_lblk),
 				       EIO, "Unable to copy data block, data will be lost!");
-- 
2.43.0