Re: [RFC for-4.5 12/12] drivers/passthrough: arm: Add support for SMMU drivers

[Julien Grall <julien.grall@linaro.org>]

On 02/19/2014 05:27 PM, Ian Campbell wrote:
> On Wed, 2014-02-19 at 17:21 +0000, Julien Grall wrote:
>>>
>>>> + * This driver currently supports:
>>>> + *  - SMMUv1 and v2 implementations (didn't try v2 SMMU)
>>>
>>> I guess this is Will's original comment, I thought SMMU-400, which
>>> you've tried, was v2?
>>
>> No it's SMMU v2. As I understand:
>> 	- SMMU-400 => SMMU v1
>>         - SMMU-500 => SMMU v2
> 
> Ah yes, that sounds more familiar.
> 
>> The words in parenthesis was added by me because I don't have a test box
>> with SMMU v2.
> 
> OK.
> 
>>>> +#define SZ_4K                               (1 << 12)
>>>> +#define SZ_64K                              (1 << 16)
>>>> +
>>>> +/* Driver options */
>>>> +#define SMMU_OPT_SECURE_CONFIG_ACCESS   (1 << 0)
>>>
>>> Is this just retained to reduce the deviation from the Linux driver?
>>> It's no use to us I think. (I suppose that goes for a bunch of other
>>> stuff, eg.. the PGSZ_4K stuff, which I will avoid commenting on
>>> further).
>>
>> SZ_4K and SZ_64K is used later in the code.
> 
> But they are actually useless to us aren't they?

As we use only 4K page in Xen yes. I kept it because there is few place
where the SMMU configuration is not the same.

If we want to support 64K page in the future it will harder to had
support if this stuff is removed.

The constant was added by himself as it doesn't exists on Xen. I can
move it into the generic code.

Cheers,

-- 
Julien Grall