On Sun, Jan 18, 2015 at 06:08:12PM +0100, SF Markus Elfring wrote:

From: Markus Elfring <redacted>
Date: Sun, 18 Jan 2015 17:30:23 +0100

The iounmap() function performs also input parameter validation.
Thus the test around the call is not needed.

Is this guaranteed for all arch'es? I expect that it would be, but I see
that, for instance, ARM allows replaceable iounmap() for
subarchitectures. Also, I see checks for various sorts of static
mappings in ARM and x86; these likely (always?) cover the NULL case, but
they're not always straightforward.

Anyway, I'm essentially saying that I'd like to be 100% sure we have a
guarantee before dropping all these.

This issue was detected by using the Coccinelle software.

What script? Hand-rolled I guess?

Brian