hi,
Greg KH <gregkh@xxxxxxxxxxxxxxxxxxx> writes:
> On Thu, Mar 21, 2019 at 11:50:04AM +0200, Felipe Balbi wrote:
>> > +static struct usb_interface_descriptor ftrace_intf_desc = {
>> > + .bLength = USB_DT_INTERFACE_SIZE,
>> > + .bDescriptorType = USB_DT_INTERFACE,
>> > +
>> > + .bAlternateSetting = 0,
>> > + .bNumEndpoints = 1,
>> > + .bInterfaceClass = USB_CLASS_VENDOR_SPEC,
>> > + .bInterfaceSubClass = USB_SUBCLASS_VENDOR_SPEC,
>>
>> We *may* be able to use the USB_DEBUG_CLASS here with a vendor protocol
>> which we define to be Linux Ftrace. If folks think it would be best,
>> then I can go look at the debug class and check whether it would be
>> feasible.
>
> I didn't think that that class allowed vendor protocols, but if it does,
> sure, that would be nice.
I think it does. That spec is public, actually. Came out as part of usb
3.1. I'll review the spec tomorrow and, hopefully, resend this with
Diagnostic Class for the interface using DvC.Trace subclass and vendor
protocol.
cheers
--
balbi
