On Thu, Jun 13, 2019 at 06:52:44PM +0000, Jason Gunthorpe wrote: > On Thu, Jun 13, 2019 at 11:43:06AM +0200, Christoph Hellwig wrote: > > This function has never been used since it was first added to the kernel > > more than a year and a half ago, and if we ever grow a consumer of the > > MEMORY_DEVICE_PUBLIC infrastructure it can easily use devm_memremap_pages > > directly now that we've simplified the API for it. > > nit: Have we simplified the interface for devm_memremap_pages() at > this point, or are you talking about later patches in this series. After this series. I've just droped that part of the sentence to avoid confusion. > I checked this and all the called functions are exported symbols, so > there is no blocker for a future driver to call devm_memremap_pages(), > maybe even with all this boiler plate, in future. > > If we eventually get many users that need some simplified registration > then we should add a devm_memremap_pages_simplified() interface and > factor out that code when we can see the pattern. After this series devm_memremap_pages already is simpler to use than hmm_device_add_resource was before, so I'm not worried at all. The series actually net removes lines from noveau (if only a few).