Re: [PATCH v2 15/18] IOMMU/x86: prefill newly allocate page tables

[Jan Beulich <jbeulich@xxxxxxxx>]

On 14.12.2021 16:17, Roger Pau Monné wrote:
> On Tue, Dec 14, 2021 at 04:10:28PM +0100, Jan Beulich wrote:
>> On 14.12.2021 16:06, Roger Pau Monné wrote:
>>> Forgot to comment.
>>>
>>> On Fri, Sep 24, 2021 at 11:54:58AM +0200, Jan Beulich wrote:
>>>> --- a/xen/drivers/passthrough/amd/pci_amd_iommu.c
>>>> +++ b/xen/drivers/passthrough/amd/pci_amd_iommu.c
>>>> @@ -238,7 +238,7 @@ int amd_iommu_alloc_root(struct domain *
>>>>  
>>>>      if ( unlikely(!hd->arch.amd.root_table) )
>>>>      {
>>>> -        hd->arch.amd.root_table = iommu_alloc_pgtable(d);
>>>> +        hd->arch.amd.root_table = iommu_alloc_pgtable(d, 0);
>>>
>>> So root tables don't get markers setup...
>>>
>>>
>>>>          if ( !hd->arch.amd.root_table )
>>>>              return -ENOMEM;
>>>>      }
>>>> --- a/xen/drivers/passthrough/vtd/iommu.c
>>>> +++ b/xen/drivers/passthrough/vtd/iommu.c
>>>> @@ -297,7 +297,7 @@ static uint64_t addr_to_dma_page_maddr(s
>>>>              goto out;
>>>>  
>>>>          pte_maddr = level;
>>>> -        if ( !(pg = iommu_alloc_pgtable(domain)) )
>>>> +        if ( !(pg = iommu_alloc_pgtable(domain, 0)) )
>>>
>>> ...likewise here.
>>
>> Yes. Plus quarantine domain's page tables also don't. Neither root
>> tables nor quarantine domain's are ever eligible for re-coalescing,
>> so there's no point having markers there.
> 
> Quarantine won't be coalesced anyway as the same mfn is repeated over
> all the entries, so it will never be a suitable candidate for
> coalescing?

Correct.

Jan