On Wed, 23 Jun 2010, Andrea Righi wrote: > On Wed, Jun 23, 2010 at 11:36:52AM -0400, Alan Stern wrote: > > On Wed, 23 Jun 2010, Andrea Righi wrote: > > > > > I'm trying to write a driver for the OXU210 PCI evaluation board, > > > integrating it into the EHCI subsystem (see the patch below). It's still > > > WIP, for now only the SPH host controller is supported (no OTG). > > > > > > Because this host controller can only access its own small SRAM area I'm > > > using the HCD_LOCAL_MEM flag and dma_declare_coherent_memory() to > > > enforce "local" memory utilization. > > > > > > The first issue I've found is that the controller seems to have problems > > > with unaliged memory accesses, so I applied a patch to the core HCD code > > > to remove the put/get_unaligned() stuff. > > > > > > After that something started to work, devices (mass storage, > > > usbs-serial, keyboard devices, etc.) are correctly recognized, but > > > there're still some issues. > > > > > > More exactly, bulk transfers don't seem to complete successfully very > > > often. For example when I connect a USB drive I get this from dmesg: > > > > ... > > > > > Any idea or suggestion that could help me to debug this problem is > > > welcome... > > > > Maybe the problem is in the device and not in your driver. You should > > try testing a different device, preferably one that runs at high speed > > instead of full speed. > > Almost same using a high speed device: The main difference being that this time the first bulk-IN transfer failed, whereas before the second transfer failed. I'm about out of ideas. You could look at the qh status values in the "async" debugging file while waiting for the URB to time out, but they probably won't tell you anything new. Alan Stern -- To unsubscribe from this list: send the line "unsubscribe linux-usb" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
