Hi, On Mon, Jun 25, 2012 at 08:04:08PM +0530, Pratyush Anand wrote: > On 6/25/2012 7:41 PM, Felipe Balbi wrote: > >whenever we want to stall ep0, we always call > >dwc3_ep0_stall_and_restart() which makes sure > > Probably this is not correct. A control ep can also be stalled by > calling set_halt. > > So we can have > dwc3_gadget_ep0_ops.set_halt as dwc3_ep0_stall_and_restart or some > similar function. I don't think we should allow gadget driver to directly stall ep0. I don't see the point because ep0 stall should be done only in case of a bad ctrl request or something which we found to be bogus and that we capture through the return of ->setup(). The call in uvc_v4l2.c seems really bogus to me, as it's trying to stall the endpoint when argument passed to ->ioctl() is screwed up. I don't think that's correct at all. -- balbi
Attachment:
signature.asc
Description: Digital signature
