On 2021/2/28 0:27, Alan Stern Wrote:

On Sat, Feb 27, 2021 at 11:38:08AM +0800, liulongfang wrote:

quoted

On 2021/2/27 0:30, Alan Stern wrote:

quoted

On Fri, Feb 26, 2021 at 04:21:37PM +0800, Longfang Liu wrote:

quoted

Our current XHCI hardware controller has been customized to only
support USB 2.0 ports. When using the current xhci driver, an xhci
controller device and an ehci controller device will be created
automatically.

That sentence makes no sense at all.  An EHCI controller device is a 
piece of hardware.  How can an xHCI driver, which is a piece of 
software, create a piece of hardware?

Alan Stern
.

The hardware device is a complete USB3.0 controller,
but I hope to support a USB2.0-only mode through software configuration.

Even if it only supports USB-2.0 connections, an xHCI controller is 
still an xHCI controller.  It doesn't magically transform into an EHCI 
controller.

You are not creating an EHCI controller device.  Rather, you are trying 
to restrict an xHCI controller device to make it handle only USB-2.0 
connections.  If you run lsusb on a system that has an xHCI controller, 
you'll see that the controller is bound to two USB buses: a USB-2 bus 
and a USB-3 bus.  But for both buses, the controller is xHCI -- not 
EHCI.

Your patch description is inaccurate.

Alan Stern
.

Yes, you are right.
when I run lsusb on a system that has this xHCI hardware,
the system displays two USB buses: a USB-2 bus and a USB-3 bus.
Both of these are xHCI host controllers, but USB-3 bus's roothub is zero.
Thanks.
Longfang Liu.