On 2/14/25 17:31, Corvin Köhne wrote:
> On Thu, 2025-02-13 at 14:45 -0700, Alex Williamson wrote:
>> On Thu, 13 Feb 2025 14:50:50 +0100
>> Cédric Le Goater <c...@redhat.com> wrote:
>>> + /*
>>> + * IGD
>>> + */
>>> +
>>> + object_class_property_set_description(klass, /* 2.7 */
>>> + "x-igd-opregion",
>>> + "Add IGD OpRegion support for
>>> (headless system)");
>>
>> [Cc Tomita and Corvin have more recent understanding of IGD options]
>>
>> Not necessarily for headless systems, unless others have better
>> suggestions, maybe just "Expose host IGD OpRegion table to guest".
>>
>
> Sounds good. Note that it's required for the Intel GOP (EFI driver) and the
> Windows driver. So, it's mandatory for non headless systems.
"Expose host IGD OpRegion to guest" sounds more accurate I think.
OpRegion is not a table, it contains Video BIOS Table (VBT) for display
output port info.
>>> + object_class_property_set_description(klass, /* 2.7 (See c4c45e943e51)
>>> */
>>> + "x-igd-gms",
>>> + "Add Intel graphics legacy mode
>>> device assignment support. "
>>> + "Assign 00:02.0 from the host to
>>> 00:02.0 in the VM");
>>
>> Not really. Tomita added a useful comment and commit log in
>> 37f05a59e869. Perhaps:
>>
>> "Override DVMT Pre-Allocated value for IGD stolen memory. (32MiB units)"
"Override IGD data stolen memory size. (32MiB units)" might be better.
> Nit: This sets the Graphics Mode Select value of the GMCH Graphics Control
> register. While we currently only support values in the range of 0x00 - 0x40,
> I
> don't see a reason to not support 0xF0 - 0xFE which uses 4 MiB increments.
I skipped the 4MiB values before for keeping the conditions simple and
these 32 MiB values should cover most cases. We can consider supporting
4MiB values.
