On Tue, Aug 23, 2016 at 10:03 PM, Chris Wilson <ch...@chris-wilson.co.uk> wrote: > On Tue, Aug 16, 2016 at 10:14:57AM +0200, Linus Walleij wrote: >> On Mon, Aug 15, 2016 at 5:08 PM, kbuild test robot
>> I think I've finally fixed this now by making OF_GPIO depend on !NO_IOMEM. > > I've stumbled over this patch (...) > Fix this by including <linux/io-mapping.h> explicitly. > > having touched io-mapping.h > > The original patch is incorrect as io-mapping.h is not the right header > for iounmap() - that is #include <linux/io.h> (pulling in <asm/io.h>) > which is (and was) already present. You're right, I'll revert this patch. > So the root cause here is arch/um not supplying enough stubs? arch/um is explicitly not supporting IOMEM so anyone code using iomap()/iounmap() have to make sure they do not get compiled on UM. > But in the > meantime, do you mind removing the incorrect include now that you have a > w/a to keep 0day quiet? :) OK Yours, Linus Walleij