netdev - Re: [PATCH] net: devmem: Remove dst (ENODEV) check in net_devmem_get

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <aQEHQReSmbXeIw15@devvm11784.nha0.facebook.com>
Date: Tue, 28 Oct 2025 11:11:13 -0700
From: Bobby Eshleman <bobbyeshleman@...il.com>
To: Shivaji Kant <shivajikant@...gle.com>
Cc: netdev@...r.kernel.org, "David S . Miller" <davem@...emloft.net>,
	Eric Dumazet <edumazet@...gle.com>,
	Jakub Kicinski <kuba@...nel.org>, Paolo Abeni <pabeni@...hat.com>,
	Simon Horman <horms@...nel.org>,
	Mina Almasry <almasrymina@...gle.com>,
	Stanislav Fomichev <sdf@...ichev.me>,
	Pavel Begunkov <asml.silence@...il.com>,
	Pranjal Shrivastava <praan@...gle.com>,
	Vedant Mathur <vedantmathur@...gle.com>
Subject: Re: [PATCH] net: devmem: Remove dst (ENODEV) check in
 net_devmem_get_binding

On Tue, Oct 28, 2025 at 06:07:14AM +0000, Shivaji Kant wrote:
> The Devmem TX binding lookup function, performs a strict
> check against the socket's destination cache (`dst`) to
> ensure the bound `dmabuf_id` corresponds to the correct
> network device (`dst->dev->ifindex == binding->dev->ifindex`).
> 
> However, this check incorrectly fails and returns `-ENODEV`
> if the socket's route cache entry (`dst`) is merely missing
> or expired (`dst == NULL`). This scenario is observed during
> network events, such as when flow steering rules are deleted,
> leading to a temporary route cache invalidation.
> 
> The parent caller, `tcp_sendmsg_locked()`, is already
> responsible for acquiring or validating the route (`dst_entry`).
> If `dst` is `NULL`, `tcp_sendmsg_locked()` will correctly
> derive the route before transmission.
> 
> This patch removes the `dst` validation from
> `net_devmem_get_binding()`. The function now only validates
> the existence of the binding and its TX vector, relying on the
> calling context for device/route correctness. This allows
> temporary route cache misses to be handled gracefully by the
> TCP/IP stack without ENODEV error on the Devmem TX path.
> 
> Reported-by: Eric Dumazet <edumazet@...gle.com>
> Reported-by: Vedant Mathur <vedantmathur@...gle.com>
> Suggested-by: Eric Dumazet <edumazet@...gle.com>
> Fixes: bd61848900bf ("net: devmem: Implement TX path")
> Signed-off-by: Shivaji Kant <shivajikant@...gle.com>
> ---
>  net/core/devmem.c | 27 ++++++++++++++++++++++++---
>  1 file changed, 24 insertions(+), 3 deletions(-)
> 
> diff --git a/net/core/devmem.c b/net/core/devmem.c
> index d9de31a6cc7f..1d04754bc756 100644
> --- a/net/core/devmem.c
> +++ b/net/core/devmem.c
> @@ -17,6 +17,7 @@
>  #include <net/page_pool/helpers.h>
>  #include <net/page_pool/memory_provider.h>
>  #include <net/sock.h>
> +#include <net/tcp.h>
>  #include <trace/events/page_pool.h>
>  
>  #include "devmem.h"
> @@ -357,7 +358,8 @@ struct net_devmem_dmabuf_binding *net_devmem_get_binding(struct sock *sk,
>  							 unsigned int dmabuf_id)
>  {
>  	struct net_devmem_dmabuf_binding *binding;
> -	struct dst_entry *dst = __sk_dst_get(sk);
> +	struct net_device *dst_dev;
> +	struct dst_entry *dst;
>  	int err = 0;
>  
>  	binding = net_devmem_lookup_dmabuf(dmabuf_id);
> @@ -366,16 +368,35 @@ struct net_devmem_dmabuf_binding *net_devmem_get_binding(struct sock *sk,
>  		goto out_err;
>  	}
>  
> +	rcu_read_lock();
> +	dst = __sk_dst_get(sk);
> +	/* If dst is NULL (route expired), attempt to rebuild it. */
> +	if (unlikely(!dst)) {
> +		if (inet_csk(sk)->icsk_af_ops->rebuild_header(sk)) {
> +			err = -EHOSTUNREACH;
> +			goto out_unlock;
> +		}

Echoing your discussion with Eric, I think the message might want to
call out this part. Besides that, all looks good!

Pending that nit:

Reviewed-by: Bobby Eshleman <bobbyeshleman@...a.com>

Best,
Bobby