On Fri, Oct 10, 2008 at 11:40:56AM +0900, FUJITA Tomonori wrote: > IIRC, except for one SGI architecture, coherent_dma_mask is > meaningless, dma_mask is always equal to coherent_dma_mask. Not correct. Several PCI-X devices can only do 32-bit DMA to control data but can do 64-bit DMA for payload data. I don't have the list off the top of my head but that is the origin of coherent DMA mask. > Lots of > IOMMU implementations ignore coherent_dma_mask and use dma_mask for > alloc_coherent(). That sounds like a bug. And I don't think it's "lots". > Some drivers don't set up coherent_dma_mask. They probably should. Current default is 32-bits for PCI devices. > Theoretically, we need to fix this but it doesn't cause any > problem. That's why nobody cares about it, I guess. Agreed - for the "supported" configurations, it works. hth, grant -- To unsubscribe from this list: send the line "unsubscribe linux-pci" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
