Re: [PATCH v2] iommu/rockchip: Add missing set_platform_dma_ops callback

[Steven Price <steven.price@arm.com>]

On 27/03/2023 16:26, Jason Gunthorpe wrote:
> On Mon, Mar 27, 2023 at 03:35:04PM +0100, John Keeping wrote:
>> On Fri, Mar 24, 2023 at 11:11:27AM +0000, Steven Price wrote:
>>> Similar to exynos, we need a set_platform_dma_ops() callback for proper
>>> operation on ARM 32 bit after recent changes in the IOMMU framework
>>> (detach ops removal). But also the use of a NULL domain is confusing.
>>>
>>> Rework the code to have a singleton rk_identity_domain which is assigned
>>> to domain when using an identity mapping rather than "detaching". This
>>> makes the code easier to reason about.
>>>
>>> Signed-off-by: Steven Price <steven.price@arm.com>
>>> ---
>>> Changes since v1[1]:
>>>
>>>  * Reworked the code to avoid a NULL domain, instead a singleton
>>>    rk_identity_domain is used instead. The 'detach' language is no
>>>    longer used.
>>>
>>> [1] https://lore.kernel.org/r/20230315164152.333251-1-steven.price%40arm.com
>>>
>>>  drivers/iommu/rockchip-iommu.c | 50 ++++++++++++++++++++++++++--------
>>>  1 file changed, 39 insertions(+), 11 deletions(-)
>>>
>>> diff --git a/drivers/iommu/rockchip-iommu.c b/drivers/iommu/rockchip-iommu.c
>>> index f30db22ea5d7..437541004994 100644
>>> --- a/drivers/iommu/rockchip-iommu.c
>>> +++ b/drivers/iommu/rockchip-iommu.c
>> [snip]
>>> +static struct iommu_domain rk_identity_domain = {
>>> +	.type = IOMMU_DOMAIN_IDENTITY,
>>> +	.ops = &rk_identity_ops,
>>> +};
>>> +
>>> +#ifdef CONFIG_ARM
>>
>> Is this #ifdef needed?  I can't see anything ARM-specific about this
>> function or .set_platform_dma_ops.
> 
> set_platform_dma_ops is never called on ARM64.

I have to admit this was somewhat cargo-culted from the exynos-iommu
change. This is only needed for Arm, but I'm not sure if there's any
real harm including it for other arches.

>> Not shown in the patch are the pm_runtime hooks.  Do they need to
>> change like this?
> 
> Most likely yes

Good spot - I'll send a v3 with this fixed.

Thanks,

Steve