Thread (6 messages) flat view 6 messages, 4 authors, 1d ago

Re: [PATCH net] veth: manage XDP program pointers during channel resize

From: Jakub Kicinski <kuba@kernel.org>
Date: 2026-09-21 22:41:49
Also in: lkml, netdev, stable
Subsystem: networking drivers, the rest · Maintainers: Andrew Lunn, "David S. Miller", Eric Dumazet, Jakub Kicinski, Paolo Abeni, Linus Torvalds

On Thu, 17 Sep 2026 23:39:25 +0800 Weiming Shi wrote:
quoted hunk ↗ jump to hunk
veth_set_channels() tears down XDP resources for removed RX queues
without clearing rq->xdp_prog.  If the program is then detached or
replaced, those queues keep the old pointer after bpf_prog_put().  A
later channel increase can re-enable NAPI and run the freed program.

Clear the program pointer after taking an RX queue offline, and publish
the current program only after bringing the queue back online succeeds.

  BUG: unable to handle page fault for address: ffffc90000256048
  Oops: Oops: 0000 [#1] SMP KASAN NOPTI
  RIP: veth_xdp_rcv_skb (include/linux/filter.h:779
                         include/net/xdp.h:696 drivers/net/veth.c:820)
  Call Trace:
   veth_xdp_rcv (drivers/net/veth.c:941)
   veth_poll (drivers/net/veth.c:986)
   __napi_poll (net/core/dev.c:7787)
   net_rx_action (net/core/dev.c:7850 net/core/dev.c:8007)
   handle_softirqs (kernel/softirq.c:645)
  Kernel panic - not syncing: Fatal exception in interrupt

Cc: stable@vger.kernel.org
Fixes: 4752eeb3d891 ("veth: implement support for set_channel ethtool op")
Reported-by: <redacted>
Assisted-by: LLM
Signed-off-by: Weiming Shi <redacted>
---
 drivers/net/veth.c | 8 +++++++-
 1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/drivers/net/veth.c b/drivers/net/veth.c
index 6ed3ee81153fb..ef7624bc2c0ec 100644
--- a/drivers/net/veth.c
+++ b/drivers/net/veth.c
@@ -1247,12 +1247,15 @@ static int veth_napi_enable(struct net_device *dev)
 static void veth_disable_range_safe(struct net_device *dev, int start, int end)
 {
 	struct veth_priv *priv = netdev_priv(dev);
+	int i;
 
 	if (start >= end)
 		return;
 
 	if (priv->_xdp_prog) {
 		veth_napi_del_range(dev, start, end);
+		for (i = start; i < end; i++)
+			rcu_assign_pointer(priv->rq[i].xdp_prog, NULL);
 		veth_disable_xdp_range(dev, start, end, false);
 	} else if (veth_gro_requested(dev)) {
 		veth_napi_del_range(dev, start, end);
@@ -1262,7 +1265,7 @@ static void veth_disable_range_safe(struct net_device *dev, int start, int end)
 static int veth_enable_range_safe(struct net_device *dev, int start, int end)
 {
 	struct veth_priv *priv = netdev_priv(dev);
-	int err;
+	int err, i;
 
 	if (start >= end)
 		return 0;
@@ -1281,6 +1284,9 @@ static int veth_enable_range_safe(struct net_device *dev, int start, int end)
 			veth_disable_xdp_range(dev, start, end, true);
 			return err;
 		}
+		for (i = start; i < end; i++)
+			rcu_assign_pointer(priv->rq[i].xdp_prog,
+					   priv->_xdp_prog);
 	} else if (veth_gro_requested(dev)) {
 		return veth_napi_enable_range(dev, start, end);
 	}
This is pure slop, cramming loops into functions which have none.
Shameful, really.

Please test this and submit it if it passes the tests:
diff --git a/drivers/net/veth.c b/drivers/net/veth.c
index 6ed3ee81153f..5d5e8ebce65a 100644
--- a/drivers/net/veth.c
+++ b/drivers/net/veth.c
@@ -1054,6 +1054,8 @@ static int __veth_napi_enable_range(struct net_device *dev, int start, int end)
        for (i = start; i < end; i++) {
                struct veth_rq *rq = &priv->rq[i];
 
+               if (priv->_xdp_prog)
+                       rcu_assign_pointer(rq->xdp_prog, priv->_xdp_prog);
                napi_enable(&rq->xdp_napi);
                rcu_assign_pointer(priv->rq[i].napi, &priv->rq[i].xdp_napi);
        }
@@ -1088,6 +1090,7 @@ static void veth_napi_del_range(struct net_device *dev, int start, int end)
 
                rcu_assign_pointer(priv->rq[i].napi, NULL);
                napi_disable(&rq->xdp_napi);
+               rcu_assign_pointer(rq->xdp_prog, NULL);
                __netif_napi_del(&rq->xdp_napi);
        }
        synchronize_net();
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help