Re: [RFC]How else could a malicious device sabotage endpoints for usbnet

[Greg KH <gregkh@linuxfoundation.org>]

On Thu, Dec 16, 2021 at 11:16:26AM +0100, Oliver Neukum wrote:
> 
> On 15.12.21 15:57, Greg KH wrote:
> > On Wed, Dec 15, 2021 at 03:47:55PM +0100, Oliver Neukum wrote:
> >> On 09.12.21 16:47, Greg KH wrote:
> >>> Why not use usb_find_common_endpoints() and/or the other helper
> >>> functions instead? that's what they were created for.
> >> Hi,
> >>
> >> which one would I use? In this case I already know the endpoints
> >> to be verified.
> > I have no context here so I have no idea, sorry.
> 
> usbnet has three ways to match the endpoints
> 
> 1) the subdriver provides a method
> 
> 2) a heuristic to find the endpoints is used (that should be converted
> to the new API)
> 
> 3) they are given nummerically by the subdriver
> 
> It turns out that #3 needs to verify the endpoints against malicious
> devices.
> So the following questions
> 
> a) should that verification go into usbcore

the usb_find_common_endpoints() functions are in the usbcore for drivers
to use for this type of problem.

> b) what possible ways for a malicious device to spoof us can you come
> up with

Start with:
	- invalid endpoint sizes and types
	- invalid data being sent on valid endpoint types
and you will catch almost all possible errors.

thanks,

greg k-h