Re: [PATCH v3 8/8] iommu/arm-smmu-v3: Make default domain type of HiSilicon PTT device to identity

[Yicong Yang]

On 2022/2/8 19:56, John Garry wrote:
> On 08/02/2022 11:21, Yicong Yang wrote:
>>> This patch should be earlier in the series, before the PTT driver, and the comment on hisi_ptt_check_iommu_mapping() should mention what is going on here.
>>>
>> ok I'll reorder the serives and modify the comments of hisi_ptt_check_iommu_mapping() like:
>>
>> /*
>>   * The DMA of PTT trace can only use direct mapping, due to some
>>   * hardware restriction. Check whether there is an iommu or the
>>   * policy of the iommu domain is passthrough, otherwise the trace
>>   * cannot work.
> 
> IOMMU, capitalize acronyms
> 

ok.

>>   *
>>   * The PTT device is supposed to behind the arm smmu v3, which
>>   * should have passthrough the device by a quirk. Otherwise user
>>   * should manually set the iommu domain type to identity through
>>   * sysfs.
> 
> Sorry, but I don't really understand your meaning here.
> 
> I did not think that if we have a default domain then we can change via sysfs to anything else.
> 

ok I think the last sentence maybe misleading and better drop it.

>>   */
>>
>>>> Signed-off-by: Yicong Yang<yangyicong@xxxxxxxxxxxxx>
>>>> ---
>>>>    drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c | 16 ++++++++++++++++
>>>>    1 file changed, 16 insertions(+)
>>>>
>>>> diff --git a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
>>>> index 6dc6d8b6b368..6f67a2b1dd27 100644
>>>> --- a/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
>>>> +++ b/drivers/iommu/arm/arm-smmu-v3/arm-smmu-v3.c
>>>> @@ -2838,6 +2838,21 @@ static int arm_smmu_dev_disable_feature(struct device *dev,
>>>>        }
>>>>    }
>>>>    +#define IS_HISI_PTT_DEVICE(pdev)    ((pdev)->vendor == PCI_VENDOR_ID_HUAWEI && \
>>>> +                     (pdev)->device == 0xa12e)
>>> I assume that not all revisions will require this check, right?
> 
> So if you are very confident that the next revision will be fixed then I would add a check for this current broken revision.
> 
>>>
>> For current revisions it's necessary.
>>
> 
> .