February 2014 - Wine-Devel - list.winehq.org

Re: [PATCH 1/2] d3dx9: Implement D3DXCreateBox()

by Henri Verbeet

I'm not really in charge of d3dx9, and I would've expected Matteo or Rico to write this instead, but here are some comments. On 27 February 2014 11:38, Gediminas Jakutis <gediminas(a)varciai.lt> wrote: > + static const float sign_x[24] = {-1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, 1.0f, 1.0f, > + 1.0f, 1.0f, 1.0f, 1.0f, -1.0f, -1.0f, 1.0f, 1.0f, > + -1.0f, 1.0f, 1.0f, -1.0f, -1.0f, -1.0f, 1.0f, 1.0f}; The explicit array size isn't needed, and I don't think it adds much for clarity. The indentation looks pretty arbitrary, the usual style for tables like these is to put the brace on a new line, and then indent the elements once. I.e.: static const float sign_x[] = { -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, -1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, 1.0f, -1.0f, -1.0f, 1.0f, 1.0f, -1.0f, 1.0f, 1.0f, -1.0f, -1.0f, -1.0f, 1.0f, 1.0f, }; > + TRACE("(%p, %f, %f, %f, %p, %p)\n", device, width, height, depth, mesh, adjacency); I know d3dx9 isn't terribly consistent about this, but: TRACE("device %p, width %.8e, height %.8e, depth %.8e, mesh %p, adjacency %p.\n", device, width, height, depth, mesh, adjacency); > + hr = D3DXCreateMeshFVF(12, 24, D3DXMESH_MANAGED, D3DFVF_XYZ | D3DFVF_NORMAL, device, &box); > + > + if (FAILED(hr)) > + { > + return hr; > + } > + > + if (FAILED(hr = box->lpVtbl->LockVertexBuffer(box, 0, (void **)&vertices))) > + { > + box->lpVtbl->Release(box); > + return hr; > + } These are inconsistent. I'd prefer the second. > + for (i = 0; i < 24; i++) > + { > + vertices[i].position.x = width * sign_x[i]; > + vertices[i].position.y = height * sign_y[i]; > + vertices[i].position.z = depth * sign_z[i]; I suppose this works, but I think it would be clearer to store a unit box as an array of D3DXVECTOR3s, and then call D3DXVec3Scale() on the vectors. The "24" would then become "ARRAY_SIZE(unit_box)" as well. > + hr = D3DXCreateBuffer(36 * sizeof(DWORD), adjacency); > + > + if (FAILED(hr)) if (FAILED(hr = D3DXCreateBuffer(sizeof(adjacency_table), adjacency)))

Feb. 27, 2014

Re: ws2_32/tests: Mark a SIOCATMARK test result as broken

by Bruno Jesus

On Thu, Feb 27, 2014 at 11:33 AM, André Hentschel <nerv(a)dawncrow.de> wrote: > --- > dlls/ws2_32/tests/sock.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/dlls/ws2_32/tests/sock.c b/dlls/ws2_32/tests/sock.c > index 2252245..3f27b02 100644 > --- a/dlls/ws2_32/tests/sock.c > +++ b/dlls/ws2_32/tests/sock.c > @@ -610,7 +610,7 @@ static VOID WINAPI oob_server ( server_params *par ) > ok ( pos == -1, "simple_server (%x): test pattern error: %d\n", id, pos); > > ioctlsocket ( mem->sock[0].s, SIOCATMARK, &atmark ); > - ok ( atmark == 1, "oob_server (%x): unexpectedly at the OOB mark: %i\n", id, atmark ); > + ok ( atmark == 1 || broken(atmark == 0), "oob_server (%x): unexpectedly at the OOB mark: %i\n", id, atmark ); There is other test failing too, right? At least for me. So this would only "solve" half of the problem. What we need to do IMO is simplify the test using a more sequential way to ensure normal and OOB data does not get mixed. Regards, Bruno

Feb. 27, 2014

Re: ws2_32/tests: Mark a SIOCATMARK test result as broken

by Marvin

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at https://newtestbot.winehq.org/JobDetails.pl?Key=5238 Your paranoid android. === wxppro (32 bit sock) === sock: unhandled exception c0000005 at 00000000 === wvistau64 (32 bit sock) === sock: unhandled exception c0000005 at 00000000 === w2008s64 (64 bit sock) === sock: unhandled exception c0000005 at 000007FEFE4220C0 === w7pro64 (64 bit sock) === sock.c:622: Test failed: oob_server (98c): not at the OOB mark: 1

Feb. 27, 2014

Re: ddraw/tests: Don't crash after a failed CreateSurface() call.

by Henri Verbeet

On 27 February 2014 02:46, Francois Gouget <fgouget(a)codeweavers.com> wrote: > Is there a way to check what surface types are supported? In theory, but in practice the most reliable way seems to be calling CreateSurface() and seeing if it works. Unfortunately the correct way to fix these depends a bit on the specific case. E.g.: > @@ -4817,6 +4819,8 @@ static void test_set_surface_desc(void) > > hr = IDirectDraw2_CreateSurface(ddraw, &ddsd, &surface, NULL); > ok(SUCCEEDED(hr), "Failed to create surface, hr %#x.\n", hr); > + if (FAILED(hr)) > + goto done; > > hr = IDirectDrawSurface_QueryInterface(surface, &IID_IDirectDrawSurface3, (void **)&surface3); > ok(SUCCEEDED(hr), "Failed to get IDirectDrawSurface3 interface, hr %#x.\n", hr); > @@ -5113,6 +5117,12 @@ static void test_user_memory_getdc(void) > ddsd.ddsCaps.dwCaps = DDSCAPS_SYSTEMMEMORY; > hr = IDirectDraw2_CreateSurface(ddraw, &ddsd, &surface, NULL); > ok(SUCCEEDED(hr), "Failed to create surface, hr %#x.\n", hr); > + if (FAILED(hr)) > + { > + IDirectDraw2_Release(ddraw); > + DestroyWindow(window); > + return; > + } > > hr = IDirectDrawSurface_QueryInterface(surface, &IID_IDirectDrawSurface3, (void **)&surface3); > ok(SUCCEEDED(hr), "Failed to get IDirectDrawSurface3 interface, hr %#x.\n", hr); These can be fixed by adding DDSCAPS_OFFSCREENPLAIN to ddsd.ddsCaps.dwCaps.

Feb. 27, 2014

Re: d3d9/tests: Fix some test failures on Windows 2000.

by Henri Verbeet

On 27 February 2014 03:27, Francois Gouget <fgouget(a)codeweavers.com> wrote: > The remaining CreateDevice() calls seem to be using pretty different > parameters. > > For instance in test_mipmap_levels() it uses a different backbuffer > format, does not set its size, sets EnableAutoDepthStencil to FALSE and > passes D3DDEVTYPE_NULLREF. Will the test really still make sense if all > that is changed? > Yeah. > Another example is test_depthstenciltest() which sets > EnableAutoDepthStencil to FALSE but still sets AutoDepthStencilFormat. > That's different from create_device() and intentionally so. > > (test_reset() does so once too without a comment which looks more > questionable) > Yeah, both of those intentionally don't set EnableAutoDepthStencil. > The problem is there's about 20 remaining CreateDevice() calls and I > would be hard pressed to tell if they are crucially different from what > create_device() does. > Ok, I'll look into it a bit.

Feb. 27, 2014

Re: ddraw/tests: Probe the supported display modes and skip tests based on the result.

by Henri Verbeet

On 27 February 2014 02:24, Francois Gouget <fgouget(a)codeweavers.com> wrote: > * QEmu can emulate an old Cirrus 5446 graphics card that tops out at > 24bpp and thus does not support the 32bpp modes we want. For Windows > 2000 that's all that works. > For most of these tests, perhaps all of them, we don't care about the difference between 24 and 32 bpp modes. I'm going to look into writing a small set_display_mode() helper that just tries both, which would allow these tests to be run on the Windows 2000 testbot as well. That doesn't help for 8 bpp modes, but there are only a few tests that need those, and I think the way we handle those currently is fine.

Feb. 27, 2014

Re: [PATCH 1/5] wined3d: Decouple private data from the resource structure (try 2).

by Henri Verbeet

> diff --git a/dlls/wined3d/wined3d_private.h b/dlls/wined3d/wined3d_private.h > index 0ceb6f1..c7c420a 100644 > --- a/dlls/wined3d/wined3d_private.h > +++ b/dlls/wined3d/wined3d_private.h > @@ -1976,6 +1976,19 @@ static inline void context_invalidate_active_texture(struct wined3d_context *con > context_invalidate_state(context, STATE_SAMPLER(sampler)); > } > > +struct wined3d_privstore > +{ > + struct list content; > +}; > + > +void wined3d_privstore_init(struct wined3d_privstore *store) DECLSPEC_HIDDEN; > +void wined3d_privstore_cleanup(struct wined3d_privstore *store) DECLSPEC_HIDDEN; > +HRESULT wined3d_privstore_free_private_data(struct wined3d_privstore *store, const GUID *guid); > +HRESULT wined3d_privstore_get_private_data(const struct wined3d_privstore *store, > + const GUID *guid, void *data, DWORD *data_size, DWORD *flags) DECLSPEC_HIDDEN; > +HRESULT wined3d_privstore_set_private_data(struct wined3d_privstore *store, > + const GUID *guid, const void *data, DWORD data_size, DWORD flags) DECLSPEC_HIDDEN; > + > #define WINED3D_RESOURCE_ACCESS_GPU 0x1 > #define WINED3D_RESOURCE_ACCESS_CPU 0x2 > Perhaps this should just be a set of inline helpers in something like include/wine/wined3d_util.h, or possibly wined3d.h, similar to e.g. include/wine/list.h. I'm not really a fan of the name wined3d_privstore. I suppose wined3d_private_store would be slightly preferable, but I don't think that's particularly great either.

Feb. 27, 2014

Re: [PATCH 2/2] d3dx9/tests: Add a mesh test for D3DXCreateBox()

by Marvin

Hi, While running your changed tests on Windows, I think I found new failures. Being a bot and all I'm not very good at pattern recognition, so I might be wrong, but could you please double-check? Full results can be found at https://newtestbot.winehq.org/JobDetails.pl?Key=5232 Your paranoid android. === w8 (32 bit mesh) === mesh.c:8156: Test failed: Mesh 27: new_num_vertices == 6, expected 4. mesh.c:8170: Test failed: Mesh 27: indices[3] == 3, expected 1 mesh.c:8170: Test failed: Mesh 27: indices[4] == 4, expected 3 mesh.c:8170: Test failed: Mesh 27: indices[5] == 5, expected 2 mesh.c:8211: Test failed: Mesh 27: vertex_remap_ptr[3] == 3, expected 4 mesh.c:8211: Test failed: Mesh 27: vertex_remap_ptr[4] == 4, expected -1 mesh.c:8211: Test failed: Mesh 27: vertex_remap_ptr[5] == 5, expected -1 mesh.c:8226: Test failed: Mesh 27: Got (2.000000, 3.000000, 0.000000) for vertex 3 position, expected (3.000000, 0.000000, 0.000000).

Feb. 27, 2014

Re: ddraw/tests: Probe the supported display modes and skip tests based on the result.

by Stefan Dösinger

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Am 2014-02-27 02:24, schrieb Francois Gouget: > * HD4000 : Intel they decided that 256 color modes are obsolete > (they are) and not worth supporting. Out of curiosity, does that system run old game classics like Age of Empires 2 or old StarCraft 1? If so, how? If not, I'd consider that a pretty bad non-feature. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.22 (GNU/Linux) Comment: Using GnuPG with Thunderbird - http://www.enigmail.net/ iQIcBAEBAgAGBQJTDvTdAAoJEN0/YqbEcdMw+lcP/j5K7N52A6J6WhLqpSZrP7Ha AbptP8Z9L+O0G33rZlUbZAsT8w64aTDXjKL9ahIua9ia+lxl55JJaD9ectiV8clo 22PNXK1CdubY3pIrwKdlax0f+hM5ZqTOBN6gaG1th/CTau8SqXSI3scJZDoimnJO 9UQpwYj6smghh3uQhYx62KJKPDe3Ek9zgAMKRdkWRGc5P4zL/tk4uwGYm8CGM8Oy eAg9tthRUwDeYUaQ/MPdAOO82T42Z9WLEvJO+zk/VNr23P7E+H4WoYAuM9JMM1dP 8/x6CM7hmVrz8sE45wgpxrD/IDzZtIIzEBndjojhBEp2bctEIUJADKk23zGnIxHy 87OotuP7BFgrtZG0vXWO2eOV1xmcdXWdhZaHYzG8MDHLsUwLBgzL49gC5wA+UHL3 ad/IhIr+ok9wx1dXJYMrLfrPG05x93f1CmeNnPXU7gv1z/pQ+VoLbIvhsvlLUdap 9mmJC9ogbBju7whb+aNF1SAi/JrE4IkohiDkGvii2vWB4eGpOUCxKfhriEyJTKRQ iEjcWzxZ0d9CzDA0Lhe18ia8LoUafDvY9U/Q6c6fWhA/xt8ysTvgU+CmfiFNg9xN cYGkZSEaFMrq9mWfTRAdi/Kl7ZJ2DgOZGq3hKHm5XSd4rKkLlAWOpELiycpl8tqQ gaOLXi9E/iShT2ECIEu5 =AGOH -----END PGP SIGNATURE-----

Feb. 27, 2014

Re: [PATCH] taskschd: Create dummy task definition for GetTask()

by Dmitry Timoshkov

Nikolay Sivov <nsivov(a)codeweavers.com> wrote: > >> RegisteredTask_create() doesn't handle NULL definition, > >> neither does IRegisteredTask code. If it's meant to be fixed differently > >> please let me know. > > This still needs a lot of work, there is no need to pitch in. > > > Oh yeah, I need it, it crashes for me. I have a patch which implements most of task management on the client side, and that would fix this particular problem. But I'm not sure yet whether to send it for inclusuion since it should be moved into the service at some point. -- Dmitry.

Feb. 27, 2014