[v2 PATCH] dxgi: Translate DXGI_USAGE_UNORDERED_ACCESS to corresponding wined3d bind flag.

Nikolay Sivov nsivov at codeweavers.com
Fri May 8 13:48:34 CDT 2020



On 5/8/20 9:32 PM, Henri Verbeet wrote:
> On Fri, 8 May 2020 at 22:44, Nikolay Sivov <nsivov at codeweavers.com> wrote:
>> On 5/8/20 9:10 PM, Henri Verbeet wrote:
>>> On Fri, 8 May 2020 at 22:20, Nikolay Sivov <nsivov at codeweavers.com> wrote:
>>>> On 5/8/20 8:40 PM, Henri Verbeet wrote:
>>>>> Does DXGI_USAGE_UNORDERED_ACCESS really need a device created with
>>>>> D3D11CreateDevice()? (As opposed to e.g. just D3D_FEATURE_LEVEL_11_0.)
>>>> D3D10CreateDevice1() doesn't work for me with feature level 11, both
>>>> with 10_1 sdk and 11 sdk flag.How do I create this without d3d11 functions?
>>> Maybe it's not possible, and we should validate the feature level in
>>> the d3d10core and d3d10_1 functions. (I don't think we currently do in
>>> Wine.) Still, if it depends on the feature level, D3D11CreateDevice()
>>> with D3D_FEATURE_LEVEL_10_0/D3D_FEATURE_LEVEL_10_1 would not work
>>> either.
>> I mean I tried on Windows. What we do with in wine is separate from
>> passing flags through.
>>
>> Your point is that I should use strict feature level 11 with d3d11 function,
>> excluding 10.1/10 when creating device? Or is using d3d11 function bad
>> for some reason?
> Well, it depends on how it works. If it only works with feature level
> 11, then yes, we should require that. If it also works with feature
> level 10 or 10.1, it seems odd that it would require the device to be
> created with D3D11CreateDevice().
For device (10.0) used in existing tests it doesn't work, I tried that 
before adding d3d11 one. It returns E_INVALIDARG for UAV flag.

Feature level validation, if it should happen, should already apply to 
CreateTexture2D(D3D11_BIND_UNORDERED_ACCESS), is there a reason to 
believe that CreateSurface() is any different? Internally it looks like 
it takes similar path up to wined3d.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.winehq.org/pipermail/wine-devel/attachments/20200508/8831050d/attachment.htm>


More information about the wine-devel mailing list