Hi Paul, > > + /* > > + * Use reasonable defaults so platforms don't have to provide these. > > + */ > > + if (!dev->dev.dma_mask) > > + dev->dev.dma_mask = &dev->dev.coherent_dma_mask; > > + if (!dev->dev.coherent_dma_mask) > > + dev->dev.coherent_dma_mask = DMA_BIT_MASK(32); > > Hmm. Is it kosher to override these in a driver and force DMA to be > enabled? Apparently, since a lot of drivers do it like this. This particular code was taken from the ehci-platform driver. See also: http://thread.gmane.org/gmane.linux.usb.general/86066 The discussion in that thread suggests that even though this is not quite the optimal way to do this, it is the accepted way for now and changing it needs a bit more complicated changes and more discussion, apparently. > What if it has been disabled earlier on purpose, say because the > platform does not have DMA support? You say "This still allows any > platform code to set any more specific mask if needed", but how would > that be done exactly? Platform code could go over the list of platform devices and (based on the device-tree compatible value, for example) change the mask of a device. Admittedly, this could set a more specific mask, but not exactly disable dma entirely. I guess this is not a usecase for the other drivers? For the dwc2 driver, I guess the dwc2 hardware would not have dma enabled if the system does not support it anyway? Note that this code only runs for platform devices, so not for pci devices, which can disable dma by not setting a dma_mask when combined with the next patch. Gr. Matthijs -- 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
