Netdev List
 help / color / mirror / Atom feed
From: Pavel Begunkov <asml.silence@gmail.com>
To: netdev@vger.kernel.org
Cc: "David S . Miller" <davem@davemloft.net>,
	Eric Dumazet <edumazet@google.com>,
	Jakub Kicinski <kuba@kernel.org>, Paolo Abeni <pabeni@redhat.com>,
	Simon Horman <horms@kernel.org>,
	Jamal Hadi Salim <jhs@mojatatu.com>,
	io-uring@vger.kernel.org, asml.silence@gmail.com
Subject: [RFC 02/10] net: reject zcrx skbs to not registered devices
Date: Sat, 11 Jul 2026 11:48:31 +0100	[thread overview]
Message-ID: <005a62b553c8b3ecb0ab5492e3842429e4dd1439.1783614400.git.asml.silence@gmail.com> (raw)
In-Reply-To: <cover.1783614400.git.asml.silence@gmail.com>

Tx of netmems that weren't created for the targeted device should be
rejected. Devmem TCP does it by looking up the net device in devmem TCP
private strucures. Introduce a netdev pointer in struct net_iov_area and
set it for zcrx so that we can check it in a more generic way. Keep the
existing devmem TCP path for the RFC version of the patch.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 include/net/netmem.h |  1 +
 io_uring/zcrx.c      |  3 +++
 net/core/dev.c       | 10 ++++++++--
 3 files changed, 12 insertions(+), 2 deletions(-)

diff --git a/include/net/netmem.h b/include/net/netmem.h
index bccacd21b6c3..71024c7ce884 100644
--- a/include/net/netmem.h
+++ b/include/net/netmem.h
@@ -103,6 +103,7 @@ struct net_iov_area {
 	struct net_iov *niovs;
 	size_t num_niovs;
 
+	struct net_device *netdev;
 	/* Offset into the dma-buf where this chunk starts.  */
 	unsigned long base_virtual;
 };
diff --git a/io_uring/zcrx.c b/io_uring/zcrx.c
index 7ad52f499f87..ef82e064e796 100644
--- a/io_uring/zcrx.c
+++ b/io_uring/zcrx.c
@@ -516,6 +516,7 @@ static int io_zcrx_create_area(struct io_zcrx_ifq *ifq,
 			goto err;
 	}
 
+	area->nia.netdev = ifq->netdev;
 	area->free_count = nr_iovs;
 	/* we're only supporting one area per ifq for now */
 	area->area_id = 0;
@@ -554,6 +555,7 @@ static void io_zcrx_drop_netdev(struct io_zcrx_ifq *ifq)
 
 	if (!ifq->netdev)
 		return;
+	WRITE_ONCE(ifq->area->nia.netdev, NULL);
 	netdev_put(ifq->netdev, &ifq->netdev_tracker);
 	ifq->netdev = NULL;
 }
@@ -568,6 +570,7 @@ static void io_close_queue(struct io_zcrx_ifq *ifq)
 	};
 
 	scoped_guard(mutex, &ifq->pp_lock) {
+		WRITE_ONCE(ifq->area->nia.netdev, NULL);
 		netdev = ifq->netdev;
 		netdev_tracker = ifq->netdev_tracker;
 		ifq->netdev = NULL;
diff --git a/net/core/dev.c b/net/core/dev.c
index 4b3d5cfdf6e0..703b55778c32 100644
--- a/net/core/dev.c
+++ b/net/core/dev.c
@@ -4004,9 +4004,15 @@ static struct sk_buff *validate_xmit_unreadable_skb(struct sk_buff *skb,
 	shinfo = skb_shinfo(skb);
 
 	if (shinfo->nr_frags > 0) {
+		struct net_device *trgt_dev;
+
 		niov = netmem_to_net_iov(skb_frag_netmem(&shinfo->frags[0]));
-		if (net_is_devmem_iov(niov) &&
-		    READ_ONCE(net_devmem_iov_binding(niov)->dev) != dev)
+		if (net_is_devmem_iov(niov))
+			trgt_dev = READ_ONCE(net_devmem_iov_binding(niov)->dev);
+		else
+			trgt_dev = READ_ONCE(net_iov_owner(niov)->netdev);
+
+		if (trgt_dev != dev)
 			goto out_free;
 	}
 
-- 
2.54.0


  parent reply	other threads:[~2026-07-11 10:49 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-07-11 10:48 [RFC 00/10] io_uring: prototype for device memory tx Pavel Begunkov
2026-07-11 10:48 ` [RFC 01/10] net: pass ubuf to custom sg_from_iter callbacks Pavel Begunkov
2026-07-11 10:48 ` Pavel Begunkov [this message]
2026-07-11 10:48 ` [RFC 03/10] io_uring/zcrx: switch to pcpu refcounting Pavel Begunkov
2026-07-11 10:48 ` [RFC 04/10] io_uring/zcrx: prepare areas to be exported for tx Pavel Begunkov
2026-07-11 10:48 ` [RFC 05/10] io_uring/rsrc: introduce buf registration structure Pavel Begunkov
2026-07-11 10:48 ` [RFC 06/10] io_uring/rsrc: extend buffer update Pavel Begunkov
2026-07-11 10:48 ` [RFC 07/10] io_uring/rsrc: add uncloneable regbuf flag Pavel Begunkov
2026-07-11 10:48 ` [RFC 08/10] io_uring/rsrc: add regbuf import flags Pavel Begunkov
2026-07-11 10:48 ` [RFC 09/10] io_uring/rsrc: add zcrx backed registered buffers Pavel Begunkov
2026-07-11 10:48 ` [RFC 10/10] io_uring/net: implement device memory send Pavel Begunkov

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=005a62b553c8b3ecb0ab5492e3842429e4dd1439.1783614400.git.asml.silence@gmail.com \
    --to=asml.silence@gmail.com \
    --cc=davem@davemloft.net \
    --cc=edumazet@google.com \
    --cc=horms@kernel.org \
    --cc=io-uring@vger.kernel.org \
    --cc=jhs@mojatatu.com \
    --cc=kuba@kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=pabeni@redhat.com \
    /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