On Wed, Jan 29, 2020 at 06:19:56PM +0100, Christian Borntraeger wrote: > There is not necessarily a device for that. It is a hypervisor interface (an > instruction that is interpreted by z/VM). We do have the netiucv driver that > creates a virtual nic, but there is also AF_IUCV which works without a device. > > But back to the original question: If we mark kmalloc caches as usercopy caches, > we should do the same for DMA kmalloc caches. As outlined by Christoph, this has > nothing to do with device DMA. Oh well, s/390 with its weird mix of cpu and I/O again. Everywhere else where we have addressing limits we do treat that as a DMA address. We've also had a bit of a lose plan to force ZONE_DMA as a public interface out, as it is generally the wrong thing to do for drivers. A ZONE_32 and/or ZONE_31 makes some sense as the backing for the dma allocator, but it mostly shouldn't be exposed, especially not to the slab allocator.
