Re: [PATCH 3/5] wined3d: Get rid of wined3d_device_color_fill().

[Stefan Dösinger]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Am 2014-08-22 12:32, schrieb Henri Verbeet:
> - if (desc.pool != WINED3D_POOL_DEFAULT && desc.pool != WINED3D_POOL_SYSTEM_MEM)
> - {
> -
>    WARN("Color-fill not allowed on surfaces in pool %#x.\n", desc.pool);
> -
>    return D3DERR_INVALIDCALL;
> - }

I realize that this just keeps existing code in place, but as far as I can see this check is redundant. There's already a somewhat strange check right above this one that tests if the surface is a render target or in WINED3D_POOL_DEFAULT. Unless I missed a way to create a non-default pool render target testing only for WINED3D_POOL_DEFAULT should be enough.

The patch also works only as long as we keep the D3DUSAGE_RENDERTARGET check in wined3d_device_set_rendertarget_view and not move it to wined3d_rendertarget_view_create. Is that your intention?
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v2
Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/

iQIcBAEBAgAGBQJT9yX4AAoJEN0/YqbEcdMw5i4P/Ao5NQVk+CMZtE6wAenF/YIx
V/NuOHczkC1qnXt74CkZe96o/HukbpoyyQ3uKlDnEzgwN8OrJUfwL3kVq6N9llqu
MaYGH0a4P/Ib+ZROouuolNcOFSsgjyHvTnZmHQUONhpzURnmLNdUzoKyfrZTKrwe
XtIk4prgEMny5W9Oae0eniYHeVv3Gyvg1jtSoviO/KT/fys5Bb+3xW1Voi7ARPWj
Ylm+D61JIQotblAcJVon4JElegbS4ENnziRSJM2774VaCaEpFOkCJ+0e3ZDzgqBQ
SjGbuLURL48HHYHKzCBEy9zpTiWr40cLjwm/h6f35koHl3IjHofFuxQPxOLqVDFt
L/BsdTUNKFL6EkvalC5ObdFUvgkX+1tx60h4vllue1sFbBIZRkISARoc1Q+HAwJE
4B2Ktwlwy0w2fuy8Ml7j3X8g+wqyPkCpaoFcKtauELFBkY/eqn+HFmoRAJTa2oMb
ps3d79EbeRPEVhh9NOOqAAkMuA+ztr52uEbKgq7PS0PzcFIvyECTBWxSnAEjw+d8
CKRk6HGC/v8Jo4Vqw/SzDTYRM/bO5sdvgMulux0zSAJuBZe+YGIcl2d0akT6MYQf
615tmoVbSfR3FY2e7SaMXqfXMLMEZzkQWjRy0A+Zs1+sPycobS3/Q7YjOmhW6goC
KQGqfd79P2SuiuAbvlue
=h6dU
-----END PGP SIGNATURE-----

[PATCH 3/5] wined3d: Get rid of wined3d_device_color_fill().

[Henri Verbeet]

On 22 August 2014 13:14, Stefan Dösinger  wrote:
> Am 2014-08-22 12:32, schrieb Henri Verbeet:
> > - if (desc.pool != WINED3D_POOL_DEFAULT && desc.pool != WINED3D_POOL_SYSTEM_MEM)
> > - {
> > -
> >    WARN("Color-fill not allowed on surfaces in pool %#x.\n", desc.pool);
> > -
> >    return D3DERR_INVALIDCALL;
> > - }
>
> I realize that this just keeps existing code in place, but as far as I can see this check is redundant. There's already a somewhat strange check right above this one that tests if the surface is a render target or in WINED3D_POOL_DEFAULT. Unless I missed a way to create a non-default pool render target testing only for WINED3D_POOL_DEFAULT should be enough.

It's probably a bug that we don't enforce rendertargets to be in the
default pool, but there's currently nothing to prevent you from
creating e.g. a texture with D3DUSAGE_RENDERTARGET and
D3DPOOL_MANAGED. You'll get a FIXME from surface_init(), but creation
will succeed.
> The patch also works only as long as we keep the D3DUSAGE_RENDERTARGET check in wined3d_device_set_rendertarget_view and not move it to wined3d_rendertarget_view_create. Is that your intention?

Yes, at least for the time being. Moving that check wouldn't work well
for depth/stencil views, and I haven't quite decided yet if the
blitter interface should use views or just a resource + index,
although I'm leaning towards the latter.