"Convert the AMD iommu driver to the dma-iommu api" is buggy

[Julien Grall]

Hi Robin,

On 16/10/2019 17:09, Robin Murphy wrote:
> On 16/10/2019 17:03, Joerg Roedel wrote:
>> On Wed, Oct 16, 2019 at 11:53:33AM -0400, Qian Cai wrote:
>>> On Wed, 2019-10-16 at 17:31 +0200, Joerg Roedel wrote:
>>
>>> The x86 one might just be a mistake.
>>>
>>> diff --git a/drivers/iommu/amd_iommu.c b/drivers/iommu/amd_iommu.c
>>> index ad05484d0c80..63c4b894751d 100644
>>> --- a/drivers/iommu/amd_iommu.c
>>> +++ b/drivers/iommu/amd_iommu.c
>>> @@ -2542,7 +2542,7 @@ static int amd_iommu_map(struct iommu_domain *dom,
>>> unsigned long iova,
>>>          if (iommu_prot & IOMMU_WRITE)
>>>                  prot |= IOMMU_PROT_IW;
>>> -       ret = iommu_map_page(domain, iova, paddr, page_size, prot, GFP_KERNEL);
>>> +       ret = iommu_map_page(domain, iova, paddr, page_size, prot, gfp);
>>
>> Yeah, that is a bug, I spotted that too.
>>
>>> @@ -1185,7 +1185,7 @@ static struct iommu_dma_msi_page
>>> *iommu_dma_get_msi_page(struct device *dev,
>>>          if (!iova)
>>>                  goto out_free_page;
>>> -       if (iommu_map(domain, iova, msi_addr, size, prot))
>>> +       if (iommu_map_atomic(domain, iova, msi_addr, size, prot))
>>>                  goto out_free_iova;
>>
>> Not so sure this is a bug, this code is only about setting up MSIs on
>> ARM. It probably doesn't need to be atomic.
> Right, since the iommu_dma_prepare_msi() operation was broken out to happen at 
> MSI domain setup time, I don't think the comment in there applies any more, so 
> we can probably stop disabling irqs around the iommu_dma_get_msi_page() call.

Yes, I agree that it does not need to be atomic anymore.

Cheers,

-- 
Julien Grall