On Wed, May 13, 2020 at 08:31:35PM +0200, Andrey Konovalov wrote: > On Wed, May 13, 2020 at 8:14 PM Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> wrote: > > > > On Wed, May 13, 2020 at 07:07:20PM +0200, Andrey Konovalov wrote: > > > Hi Alan, > > > > > > I've been looking at this a little more. Do I understand correctly > > > that even though Dummy UDC names endpoints as "ep1in", etc. it > > > actually allows to assign endpoints addresses different from what is > > > specified in the endpoint names (it uses find_endpoint() to find the > > > right endpoint based on ep->desc)? E.g. you can technically assign > > > endpoint with address 2 (| USB_DIR_IN) to "ep1in". > > > > Yes, that's right. In fact, you can do this with any UDC. (But with > > other UDCs it won't work, whereas with dummy-hcd it will.) > > > > > If this is correct, this kind of limits Dummy UDC usage with Raw > > > Gadget the way it is currently implemented, as Raw Gadget assumes that > > > the endpoint address must be fixed when the endpoint is named as > > > ep1in. > > > > Okay. That makes sense, since it is true for most UDCs. > > > > > Would it be acceptable to add another mode to Dummy UDC that names the > > > endpoints as "ep-a"? Perhaps enabled with a module parameter. I'm not > > > sure if this kind of naming would be technically correct, as "ep-a" > > > name assumes that we can assign arbitrary transfer type to the > > > endpoint as well, which isn't possible with Dummy UDC, as it doesn't > > > support ISO transfers. > > > > > > Or do you think there can be another way to expose the fact that Dummy > > > UDC allows arbitrary addresses? I could hardcode this into Raw Gadget, > > > but it doesn't feel like the right approach. > > > > Why do you want to do this? Does anything go wrong if you just continue > > to assume the endpoint numbers are fixed? > > Yes. Some USB drivers require endpoints with certain logical functions > to have certain addresses (e.g. for ath9k: [1]). This limits the > ability to use Raw Gadget + Dummy UDC for fuzzing, as sometimes we > can't emulate such devices unless Dummy UDC endpoint with a particular > address has required capabilities to implement those logical functions > (e.g. for ath9k: we can't emulate USB_REG_IN_PIPE endpoint, as Dummy > UDC only has an OUT endpoint with address 3). > > It's acceptable to be unable to emulate such devices with real UDCs > with fixed address endpoints, but it would be highly desirable to be > able to do that with Dummy UDC, as it technically supports > configurable endpoint addresses. Okay, that's reasonable. > [1] https://elixir.bootlin.com/linux/v5.6.12/source/drivers/net/wireless/ath/ath9k/hif_usb.h#L68 > > > I suppose, if you thought this was really necessary, we could change > > find_endpoint() so that it looks for a match against the endpoint's name > > instead of against the address stored in the descriptor. > > That would make the behaviour of Dummy UDC match what is expected from > it based on the endpoint names, but won't help with the problem > described above. Would you want to do this anyway? It doesn't seem necessary to me. > > Or we could > > change the last thirteen "generic" endpoints in ep_info[] to be > > configurable: "ep-a", ..., "ep-m", or "ep-aout", ..., "ep-min". (The > > fact that the endpoints don't support isochronous is exposed through the > > usb_ep_caps structures.) > > I think this should work. If we put this under a module parameter, > then we can make all endpoints have configurable names. No need for a module parameter; just make it a permanent change to the driver. Would you like to submit a patch? Alan Stern
