[PATCH v2 0/3] fix memremap on ARM

[Ard Biesheuvel]

On 26 February 2016 at 18:34, Ard Biesheuvel <ard.biesheuvel at linaro.org> wrote:
> This is something I ran into while working on support for the UEFI
> memory attributes and ESRT tables. In both cases, these tables are
> passed to the kernel in memory which is guaranteed to be below 4 GB,
> but may be outside of the kernel direct mapping. (UEFI typically
> attempts to allocate from the top down, which means such tables are
> highly likely to be in highmem for any system with more than 760 MB
> of system RAM)
>
> The recently introduced memremap() is a very useful abstraction for
> accessing such tables, because it is generic, and already attempts to
> do the right thing with respect to regions that may already have been
> mapped directly. However, it falls back to ioremap_cache() for mapping
> high memory, which is not allowed on ARM for system RAM, and also results
> in the region to be mapped with different attributes depending on whether
> it is covered by lowmem or not.
>
> So instead, create an arch specific hook 'arch_memremap_wb(), and
> implement it for ARM using the same memory attributes used for the
> linear mapping. Note that memremap will only call this hook for regions
> that are not already mapped permanently.
>
> Since this change results in memremap() to use attributes different from
> the ones used by ioremap_cache(), revert the change to pxa2xx-flash that
> moved it to memremap.
>
> Changes since v1/rfc:
> - new patch #1 that reverts the ioremap_cache->memremap conversion for the
>   pxa2xx-flash driver
> - added Dan's ack to patch #2
>

If there are no remaining objections, may I suggest that Dan takes
patch #1 and #2, and that I put the third patch into Russell's patch
system? The only strict ordering requirement is that #1 is merged
before #2 and #3 both hit mainline, and that is solved by taking #1
and #2 in order.

Thanks,
Ard.