Re: [PATCH V2 7/25] tools/libacpi: Add new fields in acpi_config for DMAR table

[Lan Tianyu <tianyu.lan@intel.com>]

On 2017年08月24日 14:54, Jan Beulich wrote:
>>>> On 24.08.17 at 04:33, <tianyu.lan@intel.com> wrote:
>> On 2017年08月23日 16:04, Roger Pau Monné wrote:
>>> On Wed, Aug 23, 2017 at 03:52:01PM +0800, Lan Tianyu wrote:
>>>> On 2017年08月23日 00:41, Roger Pau Monné wrote:
>>>>>>> +    drhd = (struct acpi_dmar_hardware_unit *)((void*)dmar + sizeof(*dmar));
>>>>>>> +    drhd->type = ACPI_DMAR_TYPE_HARDWARE_UNIT;
>>>>>>> +    drhd->length = sizeof(*drhd) + ioapic_scope_size;
>>>>>>> +    drhd->flags = ACPI_DMAR_INCLUDE_PCI_ALL;
>>>>>>> +    drhd->pci_segment = 0;
>>>>>>> +    drhd->base_address = config->iommu_base_addr;
>>>>>>> +
>>>>>>> +    scope = &drhd->scope[0];
>>>>>>> +    scope->type = ACPI_DMAR_DEVICE_SCOPE_IOAPIC;
>>>>>>> +    scope->length = ioapic_scope_size;
>>>>>>> +    scope->enumeration_id = config->ioapic_id;
>>>>>>> +    scope->bus = I440_PSEUDO_BUS_PLATFORM;
>>>>>>> +    scope->path[0] = I440_PSEUDO_DEVFN_IOAPIC;
>>>>> I'm not sure whether this constants should instead be fields in the
>>>>> acpi_config struct passed down from libxl. libxc shouldn't really need
>>>>> to know anything about which chipset a VM is using.
>>>>
>>>> How about rename I440_PSEUDO_XXX to VIOMMU_PSEUDO_XXX?
>>>
>>> I'm not really complaining about the naming, I'm just saying that I'm
>>> not sure whether this constants should live in libxl. It would be
>>> better IMHO if they where defined in some libxl x86 specific header,
>>> and passed to libxc inside of the acpi_config struct.
>>>
>>> At the end it is libxl which decides which chipset the VM is going to
>>> use, not libxc.
>>
>> We can do that but the bdf is reserved for IOAPIC and should be same for
>> different chipset. Do we still need to pass it via acpi_config?
> 
> Well, which value is the right (reserved) one surely can - at least
> in theory - depend on the chipset. Which means that it should
> come from the same place which determines the chipset to be
> emulated for the guest.
> 

OK. Will update.


-- 
Best regards
Tianyu Lan

_______________________________________________
Xen-devel mailing list
Xen-devel@lists.xen.org
https://lists.xen.org/xen-devel