[PATCHv4 7/7] arm64: add better page protections to arm64

[Steve Capper]

On 28 October 2014 11:44, Ard Biesheuvel <ard.biesheuvel at linaro.org> wrote:
> On 28 October 2014 12:29, Steve Capper <steve.capper at linaro.org> wrote:
>> On Mon, Oct 27, 2014 at 01:12:32PM -0700, Laura Abbott wrote:

[...]

>>
>> I think we should wipe out map_io from all these functions as it's
>> causing too much complexity. It's enough to have just sect_prot and
>> pte_prot. I posted some similar feedback to Ard:
>> http://lists.infradead.org/pipermail/linux-arm-kernel/2014-October/296918.html
>>
>
> I do agree. But could we have a single enum that maps onto {sect_prot,
> pte_prot} tuples rather than having to pass both values everywhere we
> call any of these functions?
> I.e., { MMU_PROT_DEFAULT, MMU_PROT_READONLY, MMU_PROT_READWRITE,
> MMU_PROT_READEXEC, MMU_PROT_DEVICE }, with the mapping local to mmu.c?
>

arch/arm has a mem_types array, I'm not sure how applicable that would
be for arm64
One could also be (slightly) filthy and promote a pte_prot into a
sect_prot via manipulation of the lower 2 bits (see for example
pte_mkhuge).

Basically as long as the mapping logic doesn't make decisions about
which pgprots to use, then I'm happy :-).

Cheers,
--
Steve