On Friday 19 August 2011, David Miller wrote:

From: Arnd Bergmann <arnd@arndb.de>
Date: Fri, 19 Aug 2011 14:26:18 +0200

quoted

We could of course change all existing users of of_iounmap on sparc to use
the simpler prototype, because it also just calls iounmap.

Check again, on sparc64 it needs the resource to release it.

Only the 32-bit version on sparc evaluates to just a plain iounmap().

Ok, I see.

On Friday 19 August 2011, David Brown wrote:

What if we left the SPARC calls alone, and changed of_iomap() into
of_dt_iomap() and could then make of_dt_iounmap().  Or, it could just
be of_dt_map(), and of_dt_unmap().

I think at some point in the future, we will have a mass-renaming
of of_* to dt_* or similar, which would result in a silly name.

Also, the of_iomap() name is modeled after pci_iomap/pci_iomap.

How about renaming the sparc of_ioremap/of_iounmap pair to
resource_iomap/resource_iounmap? I believe that it is not
(any more) tied to device tree based probing at all, and also
could be useful for other subsystems on non-sparc architectures.

	Arnd