From: Benjamin Coddington <ben.coddington@hammerspace.com>
To: Sean Chang <seanwascoding@gmail.com>
Cc: Benjamin Coddington <ben.coddington@hammerspace.com>,
Jeff Layton <jlayton@kernel.org>,
trondmy@kernel.org, anna@kernel.org, linux-nfs@vger.kernel.org,
linux-kernel@vger.kernel.org
Subject: Re: [PATCH v2 2/2] NFS: Fix RCU dereference of cl_xprt in nfs_compare_super_address
Date: Sun, 19 Apr 2026 06:52:28 -0700 [thread overview]
Message-ID: <35E1BD18-1B61-4F71-B50A-D1C830760562@hammerspace.com> (raw)
In-Reply-To: <20260419100128.20546-3-seanwascoding@gmail.com>
On 19 Apr 2026, at 3:01, Sean Chang wrote:
> The cl_xprt pointer in struct rpc_clnt is marked as __rcu. Accessing
> it directly in nfs_compare_super_address() is unsafe and triggers
> Sparse warnings.
>
> Fix this by wrapping the access with rcu_read_lock() and using
> rcu_dereference() to safely retrieve the transport pointer. This
> ensures the xprt structure remains memory-safe during the comparison
> of network namespaces and addresses.
>
> Additionally, add a check for the XPRT_CONNECTED state bit. While RCU
> guarantees the memory remains valid, checking XPRT_CONNECTED ensures
> the transport is still logically active, preventing operations on a
> transport that is already undergoing teardown.
>
> Fixes: 7e3fcf61abde ("nfs: don't share mounts between network namespaces")
> Signed-off-by: Sean Chang <seanwascoding@gmail.com>
> ---
> fs/nfs/super.c | 21 ++++++++++++++++++---
> 1 file changed, 18 insertions(+), 3 deletions(-)
>
> diff --git a/fs/nfs/super.c b/fs/nfs/super.c
> index 7a318581f85b..c9044d9d64cc 100644
> --- a/fs/nfs/super.c
> +++ b/fs/nfs/super.c
> @@ -1166,12 +1166,23 @@ static int nfs_set_super(struct super_block *s, struct fs_context *fc)
> static int nfs_compare_super_address(struct nfs_server *server1,
> struct nfs_server *server2)
> {
> + struct rpc_xprt *xprt1, *xprt2;
> struct sockaddr *sap1, *sap2;
> - struct rpc_xprt *xprt1 = server1->client->cl_xprt;
> - struct rpc_xprt *xprt2 = server2->client->cl_xprt;
> +
> + rcu_read_lock();
> +
> + xprt1 = rcu_dereference(server1->client->cl_xprt);
> + xprt2 = rcu_dereference(server2->client->cl_xprt);
This is a good fix for the sparse warning, but..
> +
> + if (!xprt1 || !xprt2 ||
> + !test_bit(XPRT_CONNECTED, &xprt1->state) ||
> + !test_bit(XPRT_CONNECTED, &xprt2->state))
> + goto out_unlock;
^^ I really don't think this check is necessary. Aren't we only ever
comparing with one freshly created, and the other looked up holding sb_lock?
I'm doubtful this hunk is fixing a real problem.
Ben
next prev parent reply other threads:[~2026-04-19 13:52 UTC|newest]
Thread overview: 6+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-04-19 10:01 [PATCH v2 0/2] NFS: fix RCU and tracing pointer safety Sean Chang
2026-04-19 10:01 ` [PATCH v2 1/2] NFS: remove redundant __private attribute from nfs_page_class Sean Chang
2026-04-19 13:42 ` Benjamin Coddington
2026-04-19 10:01 ` [PATCH v2 2/2] NFS: Fix RCU dereference of cl_xprt in nfs_compare_super_address Sean Chang
2026-04-19 13:52 ` Benjamin Coddington [this message]
2026-04-19 16:06 ` Sean Chang
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=35E1BD18-1B61-4F71-B50A-D1C830760562@hammerspace.com \
--to=ben.coddington@hammerspace.com \
--cc=anna@kernel.org \
--cc=jlayton@kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-nfs@vger.kernel.org \
--cc=seanwascoding@gmail.com \
--cc=trondmy@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).