Re: [PATCH next] btrfs: fix deadlock in btrfs_read_chunk_tree

[Qu Wenruo <quwenruo.btrfs@gmx.com>]

在 2025/6/25 09:26, Hillf Danton 写道:
> On Wed, 25 Jun 2025 06:00:09 +0930 Qu Wenruo wrote:
>> =E5=9C=A8 2025/6/25 00:00, Edward Adam Davis =E5=86=99=E9=81=93:
>>> Remove the lock uuid_mutex outside of sget_fc() to avoid the deadlock
>>> reported by [1].
>>> =20
>>> [1]
>>> -> #1 (&type->s_umount_key#41/1){+.+.}-{4:4}:
>>>          lock_acquire+0x120/0x360 kernel/locking/lockdep.c:5871
>>>          down_write_nested+0x9d/0x200 kernel/locking/rwsem.c:1693
>>>          alloc_super+0x204/0x970 fs/super.c:345
> 
> Given kzalloc [3], the syzbot report is false positive (a known lockdep
> issue) as nobody else should acquire s->s_umount lock.
> 
> [3] https://web.git.kernel.org/pub/scm/linux/kernel/git/next/linux-next.git/tree/fs/super.c?id=7aacdf6feed1#n319

Not a false alert either.

sget_fc() can return an existing super block, we can race between a 
mount and an unmount on the same super block.

In that case it's going to cause problem.

This is already fixed in the v4 (and later v5) patchset:

https://lore.kernel.org/linux-btrfs/cover.1750724841.git.wqu@suse.com/

Thanks,
Qu

> 
>>>          sget_fc+0x329/0xa40 fs/super.c:761
>>>          btrfs_get_tree_super fs/btrfs/super.c:1867 [inline]
>>>          btrfs_get_tree_subvol fs/btrfs/super.c:2059 [inline]
>>>          btrfs_get_tree+0x4c6/0x12d0 fs/btrfs/super.c:2093
>>>          vfs_get_tree+0x8f/0x2b0 fs/super.c:1804
>>>          do_new_mount+0x24a/0xa40 fs/namespace.c:3902
>>>          do_mount fs/namespace.c:4239 [inline]
>>>          __do_sys_mount fs/namespace.c:4450 [inline]
>>>          __se_sys_mount+0x317/0x410 fs/namespace.c:4427
>>>          do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
>>>          do_syscall_64+0xfa/0x3b0 arch/x86/entry/syscall_64.c:94
>>>          entry_SYSCALL_64_after_hwframe+0x77/0x7f
>>> =20
>>> -> #0 (uuid_mutex){+.+.}-{4:4}:
>>>          check_prev_add kernel/locking/lockdep.c:3168 [inline]
>>>          check_prevs_add kernel/locking/lockdep.c:3287 [inline]
>>>          validate_chain+0xb9b/0x2140 kernel/locking/lockdep.c:3911
>>>          __lock_acquire+0xab9/0xd20 kernel/locking/lockdep.c:5240
>>>          lock_acquire+0x120/0x360 kernel/locking/lockdep.c:5871
>>>          __mutex_lock_common kernel/locking/mutex.c:602 [inline]
>>>          __mutex_lock+0x182/0xe80 kernel/locking/mutex.c:747
>>>          btrfs_read_chunk_tree+0xef/0x2170 fs/btrfs/volumes.c:7462
>>>          open_ctree+0x17f2/0x3a10 fs/btrfs/disk-io.c:3458
>>>          btrfs_fill_super fs/btrfs/super.c:984 [inline]
>>>          btrfs_get_tree_super fs/btrfs/super.c:1922 [inline]
>>>          btrfs_get_tree_subvol fs/btrfs/super.c:2059 [inline]
>>>          btrfs_get_tree+0xc6f/0x12d0 fs/btrfs/super.c:2093
>>>          vfs_get_tree+0x8f/0x2b0 fs/super.c:1804
>>>          do_new_mount+0x24a/0xa40 fs/namespace.c:3902
>>>          do_mount fs/namespace.c:4239 [inline]
>>>          __do_sys_mount fs/namespace.c:4450 [inline]
>>>          __se_sys_mount+0x317/0x410 fs/namespace.c:4427
>>>          do_syscall_x64 arch/x86/entry/syscall_64.c:63 [inline]
>>>          do_syscall_64+0xfa/0x3b0 arch/x86/entry/syscall_64.c:94
>>>          entry_SYSCALL_64_after_hwframe+0x77/0x7f
>>> =20
>>> other info that might help us debug this:
>>> =20
>>>    Possible unsafe locking scenario:
>>> =20
>>>          CPU0                    CPU1
>>>          ----                    ----
>>>     lock(&type->s_umount_key#41/1);
>>>                                  lock(uuid_mutex);
>>>                                  lock(&type->s_umount_key#41/1);
>>>     lock(uuid_mutex);
>>> =20
>>>    *** DEADLOCK ***
>>> =20
>>> Fixes: 7aacdf6feed1 ("btrfs: delay btrfs_open_devices() until super bloc=
>> k is created")
>>> Reported-by: syzbot+fa90fcaa28f5cd4b1fc1@syzkaller.appspotmail.com
>>> Closes: https://syzkaller.appspot.com/bug?extid=3Dfa90fcaa28f5cd4b1fc1
>>> Tested-by: syzbot+fa90fcaa28f5cd4b1fc1@syzkaller.appspotmail.com
>>> Signed-off-by: Edward Adam Davis <eadavis@qq.com>
>>> ---
>>>    fs/btrfs/super.c | 7 ++++---
>>>    1 file changed, 4 insertions(+), 3 deletions(-)
>>> =20
>>> diff --git a/fs/btrfs/super.c b/fs/btrfs/super.c
>>> index 237e60b53192..c2ce1eb53ad7 100644
>>> --- a/fs/btrfs/super.c
>>> +++ b/fs/btrfs/super.c
>>> @@ -1864,11 +1864,10 @@ static int btrfs_get_tree_super(struct fs_contex=
>> t *fc)
>>>    	fs_devices =3D device->fs_devices;
>>>    	fs_info->fs_devices =3D fs_devices;
>>>   =20
>>> +	mutex_unlock(&uuid_mutex);
>>
>> No, you can not unlock uuid_mutex without opening the devices.
>>
>> Just run the test case generic/604.
>>
>>>    	sb =3D sget_fc(fc, btrfs_fc_test_super, set_anon_super_fc);
>>> -	if (IS_ERR(sb)) {
>>> -		mutex_unlock(&uuid_mutex);
>>> +	if (IS_ERR(sb))
>>>    		return PTR_ERR(sb);
>>> -	}
>>>   =20
>>>    	set_device_specific_options(fs_info);
>>>   =20
>>> @@ -1887,6 +1886,7 @@ static int btrfs_get_tree_super(struct fs_context =
>> *fc)
>>>    		 * But the fs_info->fs_devices is not opened, we should not let
>>>    		 * btrfs_free_fs_context() to close them.
>>>    		 */
>>> +		mutex_lock(&uuid_mutex);
>>>    		fs_info->fs_devices =3D NULL;
>>>    		mutex_unlock(&uuid_mutex);
>>>   =20
>>> @@ -1906,6 +1906,7 @@ static int btrfs_get_tree_super(struct fs_context =
>> *fc)
>>>    		 */
>>>    		ASSERT(fc->s_fs_info =3D=3D NULL);
>>>   =20
>>> +		mutex_lock(&uuid_mutex);
>>>    		ret =3D btrfs_open_devices(fs_devices, mode, sb);
>>>    		mutex_unlock(&uuid_mutex);
>>>    		if (ret < 0) {
>