[PATCH 13/20] intel-iommu: switch from ioremap_cache to memremap

On Mon, Oct 12, 2015 at 3:05 PM, Dan Williams [off-list ref] wrote:

On Mon, Oct 12, 2015 at 2:58 PM, Joerg Roedel [off-list ref] wrote:

quoted

Hi Dan,

On Fri, Oct 09, 2015 at 06:16:46PM -0400, Dan Williams wrote:

quoted

In preparation for deprecating ioremap_cache() convert its usage in
intel-iommu to memremap.  This also eliminates the mishandling of the
__iomem annotation in the implementation.

I appreciate the change, but in the cover letter you write the API is
only implemented for x86 so far? The VT-d driver is also used on ia64,
will it still work there with this patch?

I did convert ia64 in the series [1]  but you're right this patch
can't go in independently of that conversion.  If you ack this one
I'll carry them in order.

[1]: https://lkml.org/lkml/2015/10/9/702

Ah sorry, got ahead of myself ;-p.  You can safely take this patch
through your tree.  The ia64 patch can happen asynchronously because
kernel/memremap.c has:

__weak void *arch_memremap(resource_size_t offset, size_t size,
unsigned long flags)

...as a temporary fallback until the conversion is complete.