On Wed, 23 Jun 2021 14:22:29 +0100
Joao Martins wrote:
> On 6/23/21 1:30 PM, Igor Mammedov wrote:
> > On Tue, 22 Jun 2021 16:49:03 +0100
> > Joao Martins wrote:
> >
> >> pci_memory initialized by q35 and i440fx is set to a range
> >> of 0 .. UINT64_MAX, and as a consequence when ACPI and
On 6/23/21 5:33 PM, Laszlo Ersek wrote:
> Adding Marcel and Dave.
>
> Adding Alex (seriously, a vfio- / iommu-related patch set without Alex on CC?)
>
> comments below
>
> On 06/23/21 14:30, Igor Mammedov wrote:
>> On Tue, 22 Jun 2021 16:49:03 +0100
>> Joao Martins wrote:
>>
>>> pci_memory
Adding Marcel and Dave.
Adding Alex (seriously, a vfio- / iommu-related patch set without Alex on CC?)
comments below
On 06/23/21 14:30, Igor Mammedov wrote:
> On Tue, 22 Jun 2021 16:49:03 +0100
> Joao Martins wrote:
>
>> pci_memory initialized by q35 and i440fx is set to a range
>> of 0 ..
On 6/23/21 1:30 PM, Igor Mammedov wrote:
> On Tue, 22 Jun 2021 16:49:03 +0100
> Joao Martins wrote:
>
>> pci_memory initialized by q35 and i440fx is set to a range
>> of 0 .. UINT64_MAX, and as a consequence when ACPI and pci-host
>> pick the hole64_start it does not account for allowed IOVA
On Tue, 22 Jun 2021 16:49:03 +0100
Joao Martins wrote:
> pci_memory initialized by q35 and i440fx is set to a range
> of 0 .. UINT64_MAX, and as a consequence when ACPI and pci-host
> pick the hole64_start it does not account for allowed IOVA ranges.
>
> Rather than blindly returning, round up
pci_memory initialized by q35 and i440fx is set to a range
of 0 .. UINT64_MAX, and as a consequence when ACPI and pci-host
pick the hole64_start it does not account for allowed IOVA ranges.
Rather than blindly returning, round up the hole64_start
value to the allowable IOVA range, such that it