On Wed, 4 Aug 2010, Anand Gadiyar wrote: > The Inventra DMA engine in many instances of the MUSB controller > cannot use DMA for control transfers on EP0, but can use DMA for > all other transfers. > > The USB core maps urbs for DMA if hcd->self.uses_dma is true. > (This is true for MUSB as well). > > One way to solve this issue is to split the uses_dma flag into > two, one for control transfers and one for "ordinary" transfers. > The patch below implements this. > > The original discussion thread for this issue was here [1]. > A suggestion by Oliver Neukum to not use a flag describing > exceptions [2] was not tested, as it would mean touching all > the other HCD drivers. > > An alternative approach Alan suggested is to have the MUSB HCD > driver unmap an urb if it cannot carry out any transfer using DMA. > We're yet to try this out, and I'll try and put this together in a bit. This approach is more efficient because it skips the unnecessary map/unmap steps. > [1] http://marc.info/?t=126477582800002&r=1&w=2 > [2] http://marc.info/?l=linux-kernel&m=126639787526097&w=2 > > NYET-Signed-off-by: Anand Gadiyar <gadiyar@xxxxxx> > Cc: Oliver Neukum <oliver@xxxxxxxxxx> > Cc: Alan Stern <stern@xxxxxxxxxxxxxxxxxxx> > Cc: Praveena NADAHALLY <praveen.nadahally@xxxxxxxxxxxxxx> > --- > drivers/usb/core/hcd.c | 8 +++++--- > drivers/usb/musb/musb_core.c | 3 +++ > drivers/usb/musb/musb_gadget.c | 4 ++++ > include/linux/usb.h | 6 +++++- > 4 files changed, 17 insertions(+), 4 deletions(-) > > Index: linux-2.6/drivers/usb/core/hcd.c > =================================================================== > --- linux-2.6.orig/drivers/usb/core/hcd.c > +++ linux-2.6/drivers/usb/core/hcd.c > @@ -1321,7 +1321,9 @@ static int map_urb_for_dma(struct usb_hc > */ > > if (usb_endpoint_xfer_control(&urb->ep->desc)) { > - if (hcd->self.uses_dma) { > + if (hcd->self.uses_pio_for_control_xfer) > + urb->transfer_flags = URB_NO_TRANSFER_DMA_MAP; You should "or" in the new flag value; don't overwrite other flags. > + if (!hcd->self.uses_pio_for_control_xfer) { This should be else if (hcd->self.uses_dma) { > Index: linux-2.6/include/linux/usb.h > =================================================================== > --- linux-2.6.orig/include/linux/usb.h > +++ linux-2.6/include/linux/usb.h > @@ -310,7 +310,11 @@ struct usb_bus { > struct device *controller; /* host/master side hardware */ > int busnum; /* Bus number (in order of reg) */ > const char *bus_name; /* stable id (PCI slot_name etc) */ > - u8 uses_dma; /* Does the host controller use DMA? */ > + u8 uses_dma_for_ordinary_xfer; /* Does the host controller use DMA? */ I don't see any reason to change the existing name. Especially since your new name is so much more unwieldy. > + u8 uses_pio_for_control_xfer; /* > + * Does the host controller use PIO > + * for control transfers? > + */ And this could be shortened to uses_pio_for_control. The rest of the patch looks okay. 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
