Received-SPF: pass (google.com: domain of linux-kernel+bounces-9366-linux.lists.archive=gmail.com@vger.kernel.org designates 147.75.199.223 as permitted sender) client-ip=147.75.199.223;
From: Lizhi Xu <lizhi.xu@windriver.com>
To: <syzbot+83693dbba860b4f2e549@syzkaller.appspotmail.com>
CC: <linux-fsdevel@vger.kernel.org>, <linux-kernel@vger.kernel.org>,
        <reiserfs-devel@vger.kernel.org>, <syzkaller-bugs@googlegroups.com>
Subject: [PATCH arm64] reiserfs: fix deadlock in chmod_common
Date: Fri, 22 Dec 2023 13:26:45 +0800
Message-ID: <20231222052645.2884576-1-lizhi.xu@windriver.com>
In-Reply-To: <000000000000879f66060cfd35e6@google.com>
References: <000000000000879f66060cfd35e6@google.com>
Precedence: bulk
MIME-Version: 1.0
Content-Transfer-Encoding: 8bit
Content-Type: text/plain

[Syz report]
syz-executor240/6153 is trying to acquire lock:
ffff0000dc6a3e80 (&type->i_mutex_dir_key#6){+.+.}-{3:3}, at: inode_lock include/linux/fs.h:802 [inline]
ffff0000dc6a3e80 (&type->i_mutex_dir_key#6){+.+.}-{3:3}, at: chmod_common+0x17c/0x418 fs/open.c:637

but task is already holding lock:
ffff0000d5dda418 (sb_writers#8){.+.+}-{0:0}, at: mnt_want_write+0x44/0x9c fs/namespace.c:404

which lock already depends on the new lock.

the existing dependency chain (in reverse order) is:

-> #2 (sb_writers#8){.+.+}-{0:0}:
       percpu_down_read include/linux/percpu-rwsem.h:51 [inline]
       __sb_start_write include/linux/fs.h:1635 [inline]
       sb_start_write+0x60/0x2ec include/linux/fs.h:1710
       mnt_want_write_file+0x64/0x1e8 fs/namespace.c:448
       reiserfs_ioctl+0x188/0x42c fs/reiserfs/ioctl.c:103
       vfs_ioctl fs/ioctl.c:51 [inline]
       __do_sys_ioctl fs/ioctl.c:871 [inline]
       __se_sys_ioctl fs/ioctl.c:857 [inline]
       __arm64_sys_ioctl+0x14c/0x1c8 fs/ioctl.c:857
       __invoke_syscall arch/arm64/kernel/syscall.c:37 [inline]
       invoke_syscall+0x98/0x2b8 arch/arm64/kernel/syscall.c:51
       el0_svc_common+0x130/0x23c arch/arm64/kernel/syscall.c:136
       do_el0_svc+0x48/0x58 arch/arm64/kernel/syscall.c:155
       el0_svc+0x54/0x158 arch/arm64/kernel/entry-common.c:678
       el0t_64_sync_handler+0x84/0xfc arch/arm64/kernel/entry-common.c:696
       el0t_64_sync+0x190/0x194 arch/arm64/kernel/entry.S:595

-> #1
 (&sbi->lock){+.+.}-{3:3}:
       __mutex_lock_common+0x190/0x21a0 kernel/locking/mutex.c:603
       __mutex_lock kernel/locking/mutex.c:747 [inline]
       mutex_lock_nested+0x2c/0x38 kernel/locking/mutex.c:799
       reiserfs_write_lock+0x7c/0xe8 fs/reiserfs/lock.c:27
       reiserfs_lookup+0x128/0x45c fs/reiserfs/namei.c:364
       __lookup_slow+0x250/0x374 fs/namei.c:1694
       lookup_one_len+0x178/0x28c fs/namei.c:2746
       reiserfs_lookup_privroot+0x8c/0x184 fs/reiserfs/xattr.c:977
       reiserfs_fill_super+0x15b4/0x2028 fs/reiserfs/super.c:2192
       mount_bdev+0x1e8/0x2b4 fs/super.c:1650
       get_super_block+0x44/0x58 fs/reiserfs/super.c:2601
       legacy_get_tree+0xd4/0x16c fs/fs_context.c:662
       vfs_get_tree+0x90/0x288 fs/super.c:1771
       do_new_mount+0x25c/0x8c8 fs/namespace.c:3337
       path_mount+0x590/0xe04 fs/namespace.c:3664
       do_mount fs/namespace.c:3677 [inline]
       __do_sys_mount fs/namespace.c:3886 [inline]
       __se_sys_mount fs/namespace.c:3863 [inline]
       __arm64_sys_mount+0x45c/0x594 fs/namespace.c:3863
       __invoke_syscall arch/arm64/kernel/syscall.c:37 [inline]
       invoke_syscall+0x98/0x2b8 arch/arm64/kernel/syscall.c:51
       el0_svc_common+0x130/0x23c arch/arm64/kernel/syscall.c:136
       do_el0_svc+0x48/0x58 arch/arm64/kernel/syscall.c:155
       el0_svc+0x54/0x158 arch/arm64/kernel/entry-common.c:678
       el0t_64_sync_handler+0x84/0xfc arch/arm64/kernel/entry-common.c:696
       el0t_64_sync+0x190/0x194 arch/arm64/kernel/entry.S:595

-> #0 (&type->i_mutex_dir_key#6){+.+.}-{3:3}:
       check_prev_add kernel/locking/lockdep.c:3134 [inline]
       check_prevs_add kernel/locking/lockdep.c:3253 [inline]
       validate_chain kernel/locking/lockdep.c:3869 [inline]
       __lock_acquire+0x3384/0x763c kernel/locking/lockdep.c:5137
       lock_acquire+0x23c/0x71c kernel/locking/lockdep.c:5754
       down_write+0x50/0xc0 kernel/locking/rwsem.c:1579
       inode_lock include/linux/fs.h:802 [inline]
       chmod_common+0x17c/0x418 fs/open.c:637
       vfs_fchmod fs/open.c:659 [inline]
       __do_sys_fchmod fs/open.c:668 [inline]
       __se_sys_fchmod fs/open.c:662 [inline]
       __arm64_sys_fchmod+0xe0/0x150 fs/open.c:662
       __invoke_syscall arch/arm64/kernel/syscall.c:37 [inline]
       invoke_syscall+0x98/0x2b8 arch/arm64/kernel/syscall.c:51
       el0_svc_common+0x130/0x23c arch/arm64/kernel/syscall.c:136
       do_el0_svc+0x48/0x58 arch/arm64/kernel/syscall.c:155
       el0_svc+0x54/0x158 arch/arm64/kernel/entry-common.c:678
       el0t_64_sync_handler+0x84/0xfc arch/arm64/kernel/entry-common.c:696
       el0t_64_sync+0x190/0x194 arch/arm64/kernel/entry.S:595

other info that might help us debug this:

Chain exists of:
  &type->i_mutex_dir_key#6 --> &sbi->lock --> sb_writers#8

 Possible unsafe locking scenario:

       CPU0                    CPU1
       ----                    ----
  rlock(sb_writers#8);
                               lock(&sbi->lock);
                               lock(sb_writers#8);
  lock(&type->i_mutex_dir_key#6);

 *** DEADLOCK ***

[Analysis]
The deadlock of this issue is caused by the following two paths:
1. reiserfs_ioctl()
   lock(&sbi->lock);
   lock(sb_writers#8);

2. chmod_common()
   rlock(sb_writers#8);
   lock(&type->i_mutex_dir_key#6);

[Fix]
Solution: Reorder lock holding order.

Reported-and-tested-by: syzbot+83693dbba860b4f2e549@syzkaller.appspotmail.com
Signed-off-by: Lizhi Xu <lizhi.xu@windriver.com>
---
 fs/reiserfs/ioctl.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/fs/reiserfs/ioctl.c b/fs/reiserfs/ioctl.c
index dd33f8cc6eda..cf219a998a89 100644
--- a/fs/reiserfs/ioctl.c
+++ b/fs/reiserfs/ioctl.c
@@ -100,7 +100,9 @@ long reiserfs_ioctl(struct file *filp, unsigned int cmd, unsigned long arg)
 			err = -EPERM;
 			break;
 		}
+		reiserfs_write_unlock(inode->i_sb);
 		err = mnt_want_write_file(filp);
+		reiserfs_write_lock(inode->i_sb);
 		if (err)
 			break;
 		if (get_user(inode->i_generation, (int __user *)arg)) {
-- 
2.43.0