On Friday 30 October 2015 16:55:16 Timur Tabi wrote: > > On 10/30/2015 04:42 PM, Sinan Kaya wrote: > >> > >> if (WARN_ON(!pdev->dev.dma_mask)) > >> return -ENXIO; > >> > >> The dma mask has to always be set by the platform code before probe() > >> is called. If it is not set, you are not allowed to perform DMA. > > > > I tested this on an ACPI platform BTW when I was working on the initial > > implementation. > > PowerPC sets the mask to 32 bits by default: > > http://lxr.free-electrons.com/ident?i=arch_setup_pdev_archdata > > Should we do something similar in ARM64? Today, we have to manually set > the DMA mask in all drivers. We set the dma mask from the 'dma-ranges' property of the parent device, but fall back to 32-bit because we did not manage to mandate this property in time for all arm64 machines to use. Arnd -- To unsubscribe from this list: send the line "unsubscribe devicetree" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html