Re: [PATCH 2 1/2] vfs: check dentry is still valid in get_link()

From: Brian Foster
Date: Mon Nov 15 2021 - 09:26:24 EST


On Mon, Nov 15, 2021 at 10:24:22AM +0800, Ian Kent wrote:
> When following a trailing symlink in rcu-walk mode it's possible for
> the dentry to become invalid between the last dentry seq lock check
> and getting the link (eg. an unlink) leading to a backtrace similar
> to this:
>
> crash> bt
> PID: 10964 TASK: ffff951c8aa92f80 CPU: 3 COMMAND: "TaniumCX"
>
> #7 [ffffae44d0a6fbe0] page_fault at ffffffff8d6010fe
> [exception RIP: unknown or invalid address]
> RIP: 0000000000000000 RSP: ffffae44d0a6fc90 RFLAGS: 00010246
> RAX: ffffffff8da3cc80 RBX: ffffae44d0a6fd30 RCX: 0000000000000000
> RDX: ffffae44d0a6fd98 RSI: ffff951aa9af3008 RDI: 0000000000000000
> RBP: 0000000000000000 R8: ffffae44d0a6fb94 R9: 0000000000000000
> R10: ffff951c95d8c318 R11: 0000000000080000 R12: ffffae44d0a6fd98
> R13: ffff951aa9af3008 R14: ffff951c8c9eb840 R15: 0000000000000000
> ORIG_RAX: ffffffffffffffff CS: 0010 SS: 0018
> #8 [ffffae44d0a6fc90] trailing_symlink at ffffffff8cf24e61
> #9 [ffffae44d0a6fcc8] path_lookupat at ffffffff8cf261d1
> #10 [ffffae44d0a6fd28] filename_lookup at ffffffff8cf2a700
> #11 [ffffae44d0a6fe40] vfs_statx at ffffffff8cf1dbc4
> #12 [ffffae44d0a6fe98] __do_sys_newstat at ffffffff8cf1e1f9
> #13 [ffffae44d0a6ff38] do_syscall_64 at ffffffff8cc0420b
>
> Most of the time this is not a problem because the inode is unchanged
> while the rcu read lock is held.
>
> But xfs can re-use inodes which can result in the inode ->get_link()
> method becoming invalid (or NULL).
>
> This case needs to be checked for in fs/namei.c:get_link() and if
> detected the walk re-started.
>
> Signed-off-by: Ian Kent <raven@xxxxxxxxxx>
> ---

Reviewed-by: Brian Foster <bfoster@xxxxxxxxxx>

> 0 files changed
>
> diff --git a/fs/namei.c b/fs/namei.c
> index 1946d9667790..9a48a6106516 100644
> --- a/fs/namei.c
> +++ b/fs/namei.c
> @@ -1760,8 +1760,11 @@ static const char *pick_link(struct nameidata *nd, struct path *link,
> if (!res) {
> const char * (*get)(struct dentry *, struct inode *,
> struct delayed_call *);
> - get = inode->i_op->get_link;
> + get = READ_ONCE(inode->i_op->get_link);
> if (nd->flags & LOOKUP_RCU) {
> + /* Does the inode still match the associated dentry? */
> + if (unlikely(read_seqcount_retry(&link->dentry->d_seq, last->seq)))
> + return ERR_PTR(-ECHILD);
> res = get(NULL, inode, &last->done);
> if (res == ERR_PTR(-ECHILD) && try_to_unlazy(nd))
> res = get(link->dentry, inode, &last->done);
>
>