]> git.ipfire.org Git - thirdparty/kernel/linux.git/commitdiff
usbnet: fix crash due to missing BQL accounting after resume
authorSimon Schippers <simon.schippers@tu-dortmund.de>
Tue, 13 Jan 2026 07:51:38 +0000 (08:51 +0100)
committerJakub Kicinski <kuba@kernel.org>
Fri, 16 Jan 2026 03:40:59 +0000 (19:40 -0800)
In commit 7ff14c52049e ("usbnet: Add support for Byte Queue Limits
(BQL)"), it was missed that usbnet_resume() may enqueue SKBs using
__skb_queue_tail() without reporting them to BQL. As a result, the next
call to netdev_completed_queue() triggers a BUG_ON() in dql_completed(),
since the SKBs queued during resume were never accounted for.

This patch fixes the issue by adding a corresponding netdev_sent_queue()
call in usbnet_resume() when SKBs are queued after suspend. Because
dev->txq.lock is held at this point, no concurrent calls to
netdev_sent_queue() from usbnet_start_xmit() can occur.

The crash can be reproduced by generating network traffic
(e.g. iperf3 -c ... -t 0), suspending the system, and then waking it up
(e.g. rtcwake -m mem -s 5).

When testing USB2 Android tethering (cdc_ncm), the system crashed within
three suspend/resume cycles without this patch. With the patch applied,
no crashes were observed after 90 cycles. Testing with an AX88179 USB
Ethernet adapter also showed no crashes.

Fixes: 7ff14c52049e ("usbnet: Add support for Byte Queue Limits (BQL)")
Reported-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Tested-by: Bard Liao <yung-chuan.liao@linux.intel.com>
Tested-by: Simon Schippers <simon.schippers@tu-dortmund.de>
Signed-off-by: Simon Schippers <simon.schippers@tu-dortmund.de>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://patch.msgid.link/20260113075139.6735-1-simon.schippers@tu-dortmund.de
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
drivers/net/usb/usbnet.c

index 36742e64cff750cacc2dbb72cf836ff0958b4895..35789ff4dd5510c1aed2528d04d877401ae166d4 100644 (file)
@@ -1984,6 +1984,7 @@ int usbnet_resume(struct usb_interface *intf)
                        } else {
                                netif_trans_update(dev->net);
                                __skb_queue_tail(&dev->txq, skb);
+                               netdev_sent_queue(dev->net, skb->len);
                        }
                }