On 3 March 2016 at 19:01, Dan Williams [off-list ref] wrote:

On Thu, Mar 3, 2016 at 9:57 AM, Ard Biesheuvel
[off-list ref] wrote:

quoted

On 3 March 2016 at 18:55, Dan Williams [off-list ref] wrote:

quoted

On Thu, Mar 3, 2016 at 9:41 AM, Ard Biesheuvel

[..]

quoted

quoted

Hmm, then what about Russell's suggestoin, if I'm not
mischaracterizing, that ioremap_cache() move to its old
ioremap_cached() name.  The latter has less cross-arch confusion.

That would be appropriate as soon as we get rid of all the abuses of
ioremap_cache() (for some of which I am responsible myself, in the ARM
UEFI code). If we change it now, we're likely to break stuff.

Ok, sounds good.  I'll take care of the x86 ioremap_cache() removal
and leave the eventual ARM conversion to you.

OK. But I'd still like Russell to indicate whether he agrees with all
of this, i.e., wire up memremap() to perform MT_MEMORY_RW mappings in
the vmalloc area of memory regions that may be potentially be covered
by highmem as well.