Re: [PATCH 1/1] iommu/vt-d: Support asynchronous IOMMU nested capabilities

[Lu Baolu <baolu.lu@linux.intel.com>]

On 5/13/21 10:26 AM, Tian, Kevin wrote:
>> From: Lu Baolu
>> Sent: Wednesday, May 12, 2021 7:31 PM
>>
>> Hi Kevin,
>>
>> On 5/12/21 4:30 PM, Tian, Kevin wrote:
>>>> From: Lu Baolu<baolu.lu@linux.intel.com>
>>>> Sent: Wednesday, May 12, 2021 3:04 PM
>>>>
>>>> Current VT-d implementation supports nested translation only if all
>>>> underlying IOMMUs support the nested capability. This is unnecessary
>>>> as the upper layer is allowed to create different containers and set
>>>> them with different type of iommu backend. The IOMMU driver needs to
>>>> guarantee that devices attached to a nested mode iommu_domain should
>>>> support nested capabilility.
>>> so the consistency check is now applied only to the IOMMUs that are
>>> spanned by a given iommu_domain?
>> Yes.
>>
>>>> Signed-off-by: Lu Baolu<baolu.lu@linux.intel.com>
>>>> ---
>>>>    drivers/iommu/intel/iommu.c | 21 +++++++++++++++++++--
>>>>    1 file changed, 19 insertions(+), 2 deletions(-)
>>>>
>>>> diff --git a/drivers/iommu/intel/iommu.c b/drivers/iommu/intel/iommu.c
>>>> index f1742da42478..1cd4840e6f9f 100644
>>>> --- a/drivers/iommu/intel/iommu.c
>>>> +++ b/drivers/iommu/intel/iommu.c
>>>> @@ -4755,6 +4755,13 @@ static int
>> prepare_domain_attach_device(struct
>>>> iommu_domain *domain,
>>>>    	if (!iommu)
>>>>    		return -ENODEV;
>>>>
>>>> +	if ((dmar_domain->flags & DOMAIN_FLAG_NESTING_MODE) &&
>>>> +	    !ecap_nest(iommu->ecap)) {
>>>> +		dev_err(dev, "%s: iommu not support nested translation\n",
>>>> +			iommu->name);
>>>> +		return -EINVAL;
>>>> +	}
>>>> +
>>>>    	/* check if this iommu agaw is sufficient for max mapped address */
>>>>    	addr_width = agaw_to_width(iommu->agaw);
>>>>    	if (addr_width > cap_mgaw(iommu->cap))
>>>> @@ -5451,11 +5458,21 @@ static int
>>>>    intel_iommu_enable_nesting(struct iommu_domain *domain)
>>>>    {
>>>>    	struct dmar_domain *dmar_domain = to_dmar_domain(domain);
>>>> +	struct dmar_drhd_unit *drhd;
>>>> +	struct intel_iommu *iommu;
>>>> +	bool has_nesting = false;
>>>>    	unsigned long flags;
>>>> -	int ret = -ENODEV;
>>>> +	int ret = -EINVAL;
>>>> +
>>>> +	for_each_active_iommu(iommu, drhd)
>>>> +		if (ecap_nest(iommu->ecap))
>>>> +			has_nesting = true;
>>>> +
>>>> +	if (!has_nesting)
>>>> +		return -ENODEV;
>>> Isn't above still doing global consistency check?
>> The logic is if nested mode is globally unsupported, return false.
> why is this check required? anyway you already have the check when
> prepare device attachment, and only at that point you have accurate
> info for which iommu to be checked. This check looks meaningless
> as even if some IOMMUs support nesting it doesn't mean the IOMMU
> relevant to this domain support it.
> 

Get you. It's really unnecessary. I will drop this check in the new
version.

Best regards,
baolu