When rtl8150_start_xmit() fails to submit the tx URB, the URB is never
handed to the USB core and write_bulk_callback() will not run.  The
driver returns NETDEV_TX_OK, which tells the networking stack that the
skb has been consumed, but nothing actually frees the skb on this
error path:

  dev->tx_skb = skb;
  ...
  if ((res = usb_submit_urb(dev->tx_urb, GFP_ATOMIC))) {
          ...
          /* no kfree_skb here */
  }
  return NETDEV_TX_OK;

This leaks the skb on every submit failure and also leaves dev->tx_skb
pointing at memory that the driver itself may later free, which is
fragile.

Free the skb with dev_kfree_skb_any() in the error path and clear
dev->tx_skb so no stale pointer is left behind.

Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Reviewed-by: Andrew Lunn <andrew@lunn.ch>
Signed-off-by: Morduan Zang <zhangdandan@uniontech.com>
---
Changes in v2:
 - Rebase on net/main as requested (Jakub Kicinski).  v1 was based on
   a tree that also carried the pending UAF fix ("net: usb: rtl8150:
   fix use-after-free in rtl8150_start_xmit()"), so v1 did not apply
   on net/main.  v2 is an independent fix that applies cleanly to
   net/main on its own; it does not depend on the UAF fix being
   applied first.
 - No code change besides the rebase context.

v1: https://lore.kernel.org/all/678BC10BB9E39322+20260421111025.15833-1-zhangdandan@uniontech.com/
---
 drivers/net/usb/rtl8150.c | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/drivers/net/usb/rtl8150.c b/drivers/net/usb/rtl8150.c
index 4cda0643afb6..9999484d2c5e 100644
--- a/drivers/net/usb/rtl8150.c
+++ b/drivers/net/usb/rtl8150.c
@@ -707,6 +707,13 @@ static netdev_tx_t rtl8150_start_xmit(struct sk_buff *skb,
 			netdev->stats.tx_errors++;
 			netif_start_queue(netdev);
 		}
+		/*
+		 * The URB was not submitted, so write_bulk_callback() will
+		 * never run to free dev->tx_skb.  Drop the skb here and
+		 * clear tx_skb to avoid leaving a stale pointer.
+		 */
+		dev->tx_skb = NULL;
+		dev_kfree_skb_any(skb);
 	} else {
 		netdev->stats.tx_packets++;
 		netdev->stats.tx_bytes += skb->len;
-- 
2.50.1

On 4/24/26 3:55 AM, Morduan Zang wrote:
> When rtl8150_start_xmit() fails to submit the tx URB, the URB is never
> handed to the USB core and write_bulk_callback() will not run.  The
> driver returns NETDEV_TX_OK, which tells the networking stack that the
> skb has been consumed, but nothing actually frees the skb on this
> error path:
> 
>   dev->tx_skb = skb;
>   ...
>   if ((res = usb_submit_urb(dev->tx_urb, GFP_ATOMIC))) {
>           ...
>           /* no kfree_skb here */
>   }
>   return NETDEV_TX_OK;
> 
> This leaks the skb on every submit failure and also leaves dev->tx_skb
> pointing at memory that the driver itself may later free, which is
> fragile.
> 
> Free the skb with dev_kfree_skb_any() in the error path and clear
> dev->tx_skb so no stale pointer is left behind.
> 
> Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> Reviewed-by: Andrew Lunn <andrew@lunn.ch>
> Signed-off-by: Morduan Zang <zhangdandan@uniontech.com>

Applied (by Jakub, AFAICS the bot did not send out the notification for
some reason), thanks!

/P

On Tue, 28 Apr 2026 13:32:13 +0200 Paolo Abeni wrote:
> > Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
> > Reviewed-by: Andrew Lunn <andrew@lunn.ch>
> > Signed-off-by: Morduan Zang <zhangdandan@uniontech.com>  
> 
> Applied (by Jakub, AFAICS the bot did not send out the notification for
> some reason), thanks!

my bad, yes, the patch was conflicting with another one, I had to tweak.
Morduan, please use patch series if you need to submit patches which
overlap in context.