On Wednesday, June 29, 2016 10:10:59 AM CEST Timur Tabi wrote: > Arnd Bergmann wrote: > > That's also not how it works: each device starts out with a 32-bit mask, > > because that's what historically all PCI devices can do. If a device > > is 64-bit DMA capable, it can extend the mask by passing DMA_BIT_MASK(64) > > (or whatever it can support), and the platform code checks if that's > > possible. > > So if it's not possible, then dma_set_mask returns an error, and the > driver should try a smaller mask? Doesn't that mean that every driver > for a 64-bit device should do this: > > for (i = 64; i >=32; i--) { > ret = dma_set_mask_and_coherent(dev, DMA_BIT_MASK(i)); > if (!ret) > break; > } > > if (ret) > return ret; > > Sure, this is overkill, but it seems to me that the driver does not > really know what mask is actually valid, so it has to find the largest > mask that works. >
Usually drivers try 64-bit mask and 32-bit masks, and the 32 bit mask is practically guaranteed to succeed. Platforms will also allow allow the driver to set a mask that is larger than what the bus supports, as long as all RAM is reachable by the bus. Arnd