Re: [RFC PATCH kernel] block: initialize block_device::bd_bdi for bdev_cache

From: Jan Kara
Date: Wed Jan 06 2021 - 05:41:50 EST


On Wed 06-01-21 20:29:00, Alexey Kardashevskiy wrote:
> This is a workaround to fix a null derefence crash:
>
> [c00000000b01f840] c00000000b01f880 (unreliable)
> [c00000000b01f880] c000000000769a3c bdev_evict_inode+0x21c/0x370
> [c00000000b01f8c0] c00000000070bacc evict+0x11c/0x230
> [c00000000b01f900] c00000000070c138 iput+0x2a8/0x4a0
> [c00000000b01f970] c0000000006ff030 dentry_unlink_inode+0x220/0x250
> [c00000000b01f9b0] c0000000007001c0 __dentry_kill+0x190/0x320
> [c00000000b01fa00] c000000000701fb8 dput+0x5e8/0x860
> [c00000000b01fa80] c000000000705848 shrink_dcache_for_umount+0x58/0x100
> [c00000000b01fb00] c0000000006cf864 generic_shutdown_super+0x54/0x200
> [c00000000b01fb80] c0000000006cfd48 kill_anon_super+0x38/0x60
> [c00000000b01fbc0] c0000000006d12cc deactivate_locked_super+0xbc/0x110
> [c00000000b01fbf0] c0000000006d13bc deactivate_super+0x9c/0xc0
> [c00000000b01fc20] c00000000071a340 cleanup_mnt+0x1b0/0x250
> [c00000000b01fc80] c000000000278fa8 task_work_run+0xf8/0x180
> [c00000000b01fcd0] c00000000002b4ac do_notify_resume+0x4dc/0x5d0
> [c00000000b01fda0] c00000000004ba0c syscall_exit_prepare+0x28c/0x370
> [c00000000b01fe10] c00000000000e06c system_call_common+0xfc/0x27c
> --- Exception: c00 (System Call) at 0000000010034890
>
> Is this fixed properly already somewhere? Thanks,
>
> Fixes: e6cb53827ed6 ("block: initialize struct block_device in bdev_alloc")

I don't think it's fixed anywhere and I've seen the syzbot report and I was
wondering how this can happen when bdev_alloc() initializes bdev->bd_bdi
and it also wasn't clear to me whether bd_bdi is really the only field that
is problematic - if we can get to bdev_evict_inode() without going through
bdev_alloc(), we are probably missing initialization of other fields in
that place as well...

But now I've realized that probably the inode is a root inode for bdev
superblock which is allocated by VFS through new_inode() and thus doesn't
undergo the initialization in bdev_alloc(). And AFAICT the root inode on
bdev superblock can get only to bdev_evict_inode() and bdev_free_inode().
Looking at bdev_evict_inode() the only thing that's used there from struct
block_device is really bd_bdi. bdev_free_inode() will also access
bdev->bd_stats and bdev->bd_meta_info. So we need to at least initialize
these to NULL as well. IMO the most logical place for all these
initializations is in bdev_alloc_inode()...

Honza

> ---
> fs/block_dev.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/fs/block_dev.c b/fs/block_dev.c
> index 3e5b02f6606c..86fdc28d565e 100644
> --- a/fs/block_dev.c
> +++ b/fs/block_dev.c
> @@ -792,8 +792,10 @@ static void bdev_free_inode(struct inode *inode)
> static void init_once(void *data)
> {
> struct bdev_inode *ei = data;
> + struct block_device *bdev = &ei->bdev;
>
> inode_init_once(&ei->vfs_inode);
> + bdev->bd_bdi = &noop_backing_dev_info;
> }
>
> static void bdev_evict_inode(struct inode *inode)
> --
> 2.17.1
>
--
Jan Kara <jack@xxxxxxxx>
SUSE Labs, CR