Re: [PATCH v2] dma-mapping: skip USB devices when configuring DMA during probe

[Johan Hovold]

On Thu, Aug 03, 2017 at 04:49:24PM +0200, Johan Hovold wrote:
> On Thu, Aug 03, 2017 at 04:23:08PM +0200, Johan Hovold wrote:
> > On Thu, Aug 03, 2017 at 10:04:21AM -0400, Alan Stern wrote:
> > > On Thu, 3 Aug 2017, Johan Hovold wrote:
> > > 
> > > > USB devices use the DMA mask and offset of the controller, which have
> > > > already been setup when a device is probed. Note that modifying the
> > > > DMA mask of a USB device would change the mask for the controller (and
> > > > all devices on the bus) as the mask is literally shared.
> > > > 
> > > > Since commit 2bf698671205 ("USB: of: fix root-hub device-tree node
> > > > handling"), of_dma_configure() would be called also for root hubs, which
> > > > use the device node of the controller. A separate, long-standing bug
> > > > that makes of_dma_configure() generate a 30-bit DMA mask from the RPI3's
> > > > "dma-ranges" would thus set a broken mask also for the controller. This
> > > > in turn prevents USB devices from enumerating when control transfers
> > > > fail:
> > > > 
> > > > 	dwc2 3f980000.usb: Cannot do DMA to address 0x000000003a166a00
> > > > 
> > > > Note that the aforementioned DMA-mask bug was benign for the HCD itself
> > > > as the dwc2 driver overwrites the mask previously set by
> > > > of_dma_configure() for the platform device in its probe callback. The
> > > > mask would only later get corrupted when the root-hub child device was
> > > > probed.
> > > > 
> > > > Fix this, and similar future problems, by simply skipping USB devices
> > > > when dma_configure() is called during probe.
> > > > 
> > > > Fixes: 09515ef5ddad ("of/acpi: Configure dma operations at probe time for platform/amba/pci bus devices")
> > > > Cc: stable <stable@xxxxxxxxxxxxxxx>	# 4.12
> > > > Cc: Robin Murphy <robin.murphy@xxxxxxx>
> > > > Cc: Sricharan R <sricharan@xxxxxxxxxxxxxx>
> > > > Cc: Stefan Wahren <stefan.wahren@xxxxxxxx>
> > > > Reported-by: Hans Verkuil <hverkuil@xxxxxxxxx>
> > > > Signed-off-by: Johan Hovold <johan@xxxxxxxxxx>
> > > > ---
> > 
> > > > diff --git a/include/linux/usb.h b/include/linux/usb.h
> > > > index cb9fbd54386e..f86ad9d8c756 100644
> > > > --- a/include/linux/usb.h
> > > > +++ b/include/linux/usb.h
> > > > @@ -1222,6 +1222,11 @@ struct usb_device_driver {
> > > >  
> > > >  extern struct bus_type usb_bus_type;
> > > >  
> > > > +static inline bool dev_is_usb(struct device *dev)
> > > > +{
> > > > +	return dev->bus == &usb_bus_type;
> > > > +}
> > > > +
> > > 
> > > Will this work if the USB subsystem is built as a module?
> > 
> > Nope. :-/
> > 
> > Add another flag (e.g. skip_dma_configure) to struct device for now?
> 
> Or we can make sure that of_dma_configure() is only ever called for
> platform_devices by checking against platform_bus_type instead.
> 
> I guess this was the intention of 09515ef5ddad ("of/acpi: Configure dma
> operations at probe time for platform/amba/pci bus devices") too.

Or perhaps not, as of_dma_configure() should also be called for pci
devices.

I suggest adding a new flag to struct device.

Johan