On 24.5.2019 17.52, Nicolas Saenz Julienne wrote:
This was seen on a Dell Precision 5520 using it's WD15 dock. The dock's Ethernet device interfaces with the laptop through one of it's USB3 ports. While idle, the Ethernet device and HCD are suspended by runtime PM, being the only device connected on the bus. Then, both are resumed on behalf of the Ethernet device, which has remote wake-up capabilities. The Ethernet device was observed to randomly disconnect from the USB port shortly after submitting it's remote wake-up request. Probably a weird timing issue yet to be investigated. This causes runtime PM to busyloop causing some tangible CPU load. The reason is the port gets stuck in the middle of a remote wake-up operation, waiting for the device to switch to U0. This never happens, leaving "port_remote_wakeup" enabled, and automatically triggering a failure on any further suspend operation. This patch clears "port_remote_wakeup" upon detecting a device with a wrong resuming port state (see Table 4-9 in 4.15.2.3). Making sure the above mentioned situation doesn't trigger a PM busyloop.
There was a similar case where the USB3 link had transitioned to a lower power U1 or U2 state after resume, before driver read the state, leaving port_remote_wakeup flag uncleared. This was fixed in 5.1 kernel by commit: 6cbcf59 xhci: Fix port resume done detection for SS ports with LPM enable Can you check if you have it? It should be in recent stable releases as well. -Mathias