Re: [PATCH v5 04/11] iommu/vt-d: Add helper to setup pasid nested translation

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

On 2023/10/13 20:40, Yi Liu wrote:
>>
>>> +        return -EINVAL;
>>> +    }
>>> +
>>> +    if ((s1_cfg->flags & IOMMU_VTD_S1_SRE) && !ecap_srs(iommu-
>>>> ecap)) {
>>> +        pr_err_ratelimited("No supervisor request support on %s\n",
>>> +                   iommu->name);
>>> +        return -EINVAL;
>>> +    }
>>> +
>>> +    if ((s1_cfg->flags & IOMMU_VTD_S1_EAFE) && !ecap_eafs(iommu-
>>>> ecap)) {
>>> +        pr_err_ratelimited("No extended access flag support
>>> on %s\n",
>>> +                   iommu->name);
>>> +        return -EINVAL;
>>> +    }
>>> +
>>> +    spin_lock(&iommu->lock);
>>> +    pte = intel_pasid_get_entry(dev, pasid);
>>> +    if (!pte) {
>>> +        spin_unlock(&iommu->lock);
>>> +        return -ENODEV;
>>> +    }
>>> +    if (pasid_pte_is_present(pte)) {
>>> +        spin_unlock(&iommu->lock);
>>> +        return -EBUSY;
>>> +    }
>>> +
>>> +    pasid_clear_entry(pte);
>>> +
>>> +    if (s1_cfg->addr_width == ADDR_WIDTH_5LEVEL)
>>> +        pasid_set_flpm(pte, 1);
>>> +
>>> +    pasid_set_flptr(pte, (uintptr_t)s1_gpgd);
>>> +
>>> +    if (s1_cfg->flags & IOMMU_VTD_S1_SRE) {
>>> +        pasid_set_sre(pte);
>>> +        if (s1_cfg->flags & IOMMU_VTD_S1_WPE)
>>> +            pasid_set_wpe(pte);
>>> +    }
>>> +
>>> +    if (s1_cfg->flags & IOMMU_VTD_S1_EAFE)
>>> +        pasid_set_eafe(pte);
>>> +
>>> +    if (s2_domain->force_snooping)
>>> +        pasid_set_pgsnp(pte);
>>> +
>>> +    pasid_set_slptr(pte, virt_to_phys(pgd));
>>> +    pasid_set_fault_enable(pte);
>>> +    pasid_set_domain_id(pte, did);
>>> +    pasid_set_address_width(pte, s2_domain->agaw);
>>> +    pasid_set_page_snoop(pte, !!ecap_smpwc(iommu->ecap));
>>> +    pasid_set_translation_type(pte, PASID_ENTRY_PGTT_NESTED);
>>> +    pasid_set_present(pte);
>>> +    spin_unlock(&iommu->lock);
>>
>> All changes within iommu->lock are specific to the device specific
>> PASID entry. Probably this is one potential cleanup TODO to
>> use a per-device lock instead.
> 
> yeah, a separate cleanup. is it, @Baolu?

Sure. I'd like to take some time to consider this further. Please keep
it as-is for the time being.

Best regards,
baolu