wined3d: Only generate the clipplane emulation KIL if a clipplane is used.
[wine] / dlls / wined3d / pixelshader.c
1 /*
2  * shaders implementation
3  *
4  * Copyright 2002-2003 Jason Edmeades
5  * Copyright 2002-2003 Raphael Junqueira
6  * Copyright 2004 Christian Costa
7  * Copyright 2005 Oliver Stieber
8  * Copyright 2006 Ivan Gyurdiev
9  * Copyright 2007-2008 Stefan Dösinger for CodeWeavers
10  *
11  * This library is free software; you can redistribute it and/or
12  * modify it under the terms of the GNU Lesser General Public
13  * License as published by the Free Software Foundation; either
14  * version 2.1 of the License, or (at your option) any later version.
15  *
16  * This library is distributed in the hope that it will be useful,
17  * but WITHOUT ANY WARRANTY; without even the implied warranty of
18  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
19  * Lesser General Public License for more details.
20  *
21  * You should have received a copy of the GNU Lesser General Public
22  * License along with this library; if not, write to the Free Software
23  * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301, USA
24  */
25
26 #include "config.h"
27
28 #include <math.h>
29 #include <stdio.h>
30
31 #include "wined3d_private.h"
32
33 WINE_DEFAULT_DEBUG_CHANNEL(d3d_shader);
34
35 #define GLINFO_LOCATION ((IWineD3DDeviceImpl *) This->baseShader.device)->adapter->gl_info
36
37 static HRESULT  WINAPI IWineD3DPixelShaderImpl_QueryInterface(IWineD3DPixelShader *iface, REFIID riid, LPVOID *ppobj) {
38     TRACE("iface %p, riid %s, ppobj %p\n", iface, debugstr_guid(riid), ppobj);
39
40     if (IsEqualGUID(riid, &IID_IWineD3DPixelShader)
41             || IsEqualGUID(riid, &IID_IWineD3DBaseShader)
42             || IsEqualGUID(riid, &IID_IWineD3DBase)
43             || IsEqualGUID(riid, &IID_IUnknown))
44     {
45         IUnknown_AddRef(iface);
46         *ppobj = iface;
47         return S_OK;
48     }
49
50     WARN("%s not implemented, returning E_NOINTERFACE\n", debugstr_guid(riid));
51
52     *ppobj = NULL;
53     return E_NOINTERFACE;
54 }
55
56 static ULONG  WINAPI IWineD3DPixelShaderImpl_AddRef(IWineD3DPixelShader *iface) {
57     IWineD3DPixelShaderImpl *This = (IWineD3DPixelShaderImpl *)iface;
58     ULONG refcount = InterlockedIncrement(&This->baseShader.ref);
59
60     TRACE("%p increasing refcount to %u\n", This, refcount);
61
62     return refcount;
63 }
64
65 static ULONG  WINAPI IWineD3DPixelShaderImpl_Release(IWineD3DPixelShader *iface) {
66     IWineD3DPixelShaderImpl *This = (IWineD3DPixelShaderImpl *)iface;
67     ULONG refcount = InterlockedDecrement(&This->baseShader.ref);
68
69     TRACE("%p decreasing refcount to %u\n", This, refcount);
70
71     if (!refcount)
72     {
73         shader_cleanup((IWineD3DBaseShader *)iface);
74         HeapFree(GetProcessHeap(), 0, This);
75     }
76
77     return refcount;
78 }
79
80 /* *******************************************
81    IWineD3DPixelShader IWineD3DPixelShader parts follow
82    ******************************************* */
83
84 static HRESULT  WINAPI IWineD3DPixelShaderImpl_GetParent(IWineD3DPixelShader *iface, IUnknown** parent){
85     IWineD3DPixelShaderImpl *This = (IWineD3DPixelShaderImpl *)iface;
86
87     *parent = This->parent;
88     IUnknown_AddRef(*parent);
89     TRACE("(%p) : returning %p\n", This, *parent);
90     return WINED3D_OK;
91 }
92
93 static HRESULT  WINAPI IWineD3DPixelShaderImpl_GetDevice(IWineD3DPixelShader* iface, IWineD3DDevice **pDevice){
94     IWineD3DPixelShaderImpl *This = (IWineD3DPixelShaderImpl *)iface;
95     IWineD3DDevice_AddRef(This->baseShader.device);
96     *pDevice = This->baseShader.device;
97     TRACE("(%p) returning %p\n", This, *pDevice);
98     return WINED3D_OK;
99 }
100
101
102 static HRESULT  WINAPI IWineD3DPixelShaderImpl_GetFunction(IWineD3DPixelShader* impl, VOID* pData, UINT* pSizeOfData) {
103   IWineD3DPixelShaderImpl *This = (IWineD3DPixelShaderImpl *)impl;
104   TRACE("(%p) : pData(%p), pSizeOfData(%p)\n", This, pData, pSizeOfData);
105
106   if (NULL == pData) {
107     *pSizeOfData = This->baseShader.functionLength;
108     return WINED3D_OK;
109   }
110   if (*pSizeOfData < This->baseShader.functionLength) {
111     /* MSDN claims (for d3d8 at least) that if *pSizeOfData is smaller
112      * than the required size we should write the required size and
113      * return D3DERR_MOREDATA. That's not actually true. */
114     return WINED3DERR_INVALIDCALL;
115   }
116
117   TRACE("(%p) : GetFunction copying to %p\n", This, pData);
118   memcpy(pData, This->baseShader.function, This->baseShader.functionLength);
119
120   return WINED3D_OK;
121 }
122
123 static void pshader_set_limits(IWineD3DPixelShaderImpl *This)
124 {
125     DWORD shader_version = WINED3D_SHADER_VERSION(This->baseShader.reg_maps.shader_version.major,
126             This->baseShader.reg_maps.shader_version.minor);
127
128     This->baseShader.limits.attributes = 0;
129     This->baseShader.limits.address = 0;
130     This->baseShader.limits.packed_output = 0;
131
132     switch (shader_version)
133     {
134         case WINED3D_SHADER_VERSION(1,0):
135         case WINED3D_SHADER_VERSION(1,1):
136         case WINED3D_SHADER_VERSION(1,2):
137         case WINED3D_SHADER_VERSION(1,3):
138             This->baseShader.limits.temporary = 2;
139             This->baseShader.limits.constant_float = 8;
140             This->baseShader.limits.constant_int = 0;
141             This->baseShader.limits.constant_bool = 0;
142             This->baseShader.limits.texcoord = 4;
143             This->baseShader.limits.sampler = 4;
144             This->baseShader.limits.packed_input = 0;
145             This->baseShader.limits.label = 0;
146             break;
147
148         case WINED3D_SHADER_VERSION(1,4):
149             This->baseShader.limits.temporary = 6;
150             This->baseShader.limits.constant_float = 8;
151             This->baseShader.limits.constant_int = 0;
152             This->baseShader.limits.constant_bool = 0;
153             This->baseShader.limits.texcoord = 6;
154             This->baseShader.limits.sampler = 6;
155             This->baseShader.limits.packed_input = 0;
156             This->baseShader.limits.label = 0;
157             break;
158
159         /* FIXME: temporaries must match D3DPSHADERCAPS2_0.NumTemps */
160         case WINED3D_SHADER_VERSION(2,0):
161             This->baseShader.limits.temporary = 32;
162             This->baseShader.limits.constant_float = 32;
163             This->baseShader.limits.constant_int = 16;
164             This->baseShader.limits.constant_bool = 16;
165             This->baseShader.limits.texcoord = 8;
166             This->baseShader.limits.sampler = 16;
167             This->baseShader.limits.packed_input = 0;
168             break;
169
170         case WINED3D_SHADER_VERSION(2,1):
171             This->baseShader.limits.temporary = 32;
172             This->baseShader.limits.constant_float = 32;
173             This->baseShader.limits.constant_int = 16;
174             This->baseShader.limits.constant_bool = 16;
175             This->baseShader.limits.texcoord = 8;
176             This->baseShader.limits.sampler = 16;
177             This->baseShader.limits.packed_input = 0;
178             This->baseShader.limits.label = 16;
179             break;
180
181         case WINED3D_SHADER_VERSION(4,0):
182             FIXME("Using 3.0 limits for 4.0 shader\n");
183             /* Fall through */
184
185         case WINED3D_SHADER_VERSION(3,0):
186             This->baseShader.limits.temporary = 32;
187             This->baseShader.limits.constant_float = 224;
188             This->baseShader.limits.constant_int = 16;
189             This->baseShader.limits.constant_bool = 16;
190             This->baseShader.limits.texcoord = 0;
191             This->baseShader.limits.sampler = 16;
192             This->baseShader.limits.packed_input = 12;
193             This->baseShader.limits.label = 16; /* FIXME: 2048 */
194             break;
195
196         default:
197             This->baseShader.limits.temporary = 32;
198             This->baseShader.limits.constant_float = 32;
199             This->baseShader.limits.constant_int = 16;
200             This->baseShader.limits.constant_bool = 16;
201             This->baseShader.limits.texcoord = 8;
202             This->baseShader.limits.sampler = 16;
203             This->baseShader.limits.packed_input = 0;
204             This->baseShader.limits.label = 0;
205             FIXME("Unrecognized pixel shader version %u.%u\n",
206                     This->baseShader.reg_maps.shader_version.major,
207                     This->baseShader.reg_maps.shader_version.minor);
208     }
209 }
210
211 static HRESULT WINAPI IWineD3DPixelShaderImpl_SetFunction(IWineD3DPixelShader *iface,
212         const DWORD *pFunction, const struct wined3d_shader_signature *output_signature)
213 {
214     IWineD3DPixelShaderImpl *This =(IWineD3DPixelShaderImpl *)iface;
215     unsigned int i, highest_reg_used = 0, num_regs_used = 0;
216     shader_reg_maps *reg_maps = &This->baseShader.reg_maps;
217     const struct wined3d_shader_frontend *fe;
218     HRESULT hr;
219
220     TRACE("(%p) : pFunction %p\n", iface, pFunction);
221
222     fe = shader_select_frontend(*pFunction);
223     if (!fe)
224     {
225         FIXME("Unable to find frontend for shader.\n");
226         return WINED3DERR_INVALIDCALL;
227     }
228     This->baseShader.frontend = fe;
229     This->baseShader.frontend_data = fe->shader_init(pFunction, output_signature);
230     if (!This->baseShader.frontend_data)
231     {
232         FIXME("Failed to initialize frontend.\n");
233         return WINED3DERR_INVALIDCALL;
234     }
235
236     /* First pass: trace shader */
237     if (TRACE_ON(d3d_shader)) shader_trace_init(fe, This->baseShader.frontend_data, pFunction);
238
239     /* Initialize immediate constant lists */
240     list_init(&This->baseShader.constantsF);
241     list_init(&This->baseShader.constantsB);
242     list_init(&This->baseShader.constantsI);
243
244     /* Second pass: figure out which registers are used, what the semantics are, etc.. */
245     hr = shader_get_registers_used((IWineD3DBaseShader *)This, fe,
246             reg_maps, NULL, This->input_signature, NULL,
247             pFunction, GL_LIMITS(pshader_constantsF));
248     if (FAILED(hr)) return hr;
249
250     pshader_set_limits(This);
251
252     for (i = 0; i < MAX_REG_INPUT; ++i)
253     {
254         if (This->input_reg_used[i])
255         {
256             ++num_regs_used;
257             highest_reg_used = i;
258         }
259     }
260
261     /* Don't do any register mapping magic if it is not needed, or if we can't
262      * achieve anything anyway */
263     if (highest_reg_used < (GL_LIMITS(glsl_varyings) / 4)
264             || num_regs_used > (GL_LIMITS(glsl_varyings) / 4))
265     {
266         if (num_regs_used > (GL_LIMITS(glsl_varyings) / 4))
267         {
268             /* This happens with relative addressing. The input mapper function
269              * warns about this if the higher registers are declared too, so
270              * don't write a FIXME here */
271             WARN("More varying registers used than supported\n");
272         }
273
274         for (i = 0; i < MAX_REG_INPUT; ++i)
275         {
276             This->input_reg_map[i] = i;
277         }
278
279         This->declared_in_count = highest_reg_used + 1;
280     }
281     else
282     {
283         This->declared_in_count = 0;
284         for (i = 0; i < MAX_REG_INPUT; ++i)
285         {
286             if (This->input_reg_used[i]) This->input_reg_map[i] = This->declared_in_count++;
287             else This->input_reg_map[i] = ~0U;
288         }
289     }
290
291     This->baseShader.load_local_constsF = FALSE;
292
293     TRACE("(%p) : Copying the function\n", This);
294
295     This->baseShader.function = HeapAlloc(GetProcessHeap(), 0, This->baseShader.functionLength);
296     if (!This->baseShader.function) return E_OUTOFMEMORY;
297     memcpy(This->baseShader.function, pFunction, This->baseShader.functionLength);
298
299     return WINED3D_OK;
300 }
301
302 void pixelshader_update_samplers(struct shader_reg_maps *reg_maps, IWineD3DBaseTexture * const *textures)
303 {
304     WINED3DSAMPLER_TEXTURE_TYPE *sampler_type = reg_maps->sampler_type;
305     unsigned int i;
306
307     if (reg_maps->shader_version.major != 1) return;
308
309     for (i = 0; i < max(MAX_FRAGMENT_SAMPLERS, MAX_VERTEX_SAMPLERS); ++i)
310     {
311         /* We don't sample from this sampler */
312         if (!sampler_type[i]) continue;
313
314         if (!textures[i])
315         {
316             ERR("No texture bound to sampler %u, using 2D\n", i);
317             sampler_type[i] = WINED3DSTT_2D;
318             continue;
319         }
320
321         switch (IWineD3DBaseTexture_GetTextureDimensions(textures[i]))
322         {
323             case GL_TEXTURE_RECTANGLE_ARB:
324             case GL_TEXTURE_2D:
325                 /* We have to select between texture rectangles and 2D textures later because 2.0 and
326                  * 3.0 shaders only have WINED3DSTT_2D as well */
327                 sampler_type[i] = WINED3DSTT_2D;
328                 break;
329
330             case GL_TEXTURE_3D:
331                 sampler_type[i] = WINED3DSTT_VOLUME;
332                 break;
333
334             case GL_TEXTURE_CUBE_MAP_ARB:
335                 sampler_type[i] = WINED3DSTT_CUBE;
336                 break;
337
338             default:
339                 FIXME("Unrecognized texture type %#x, using 2D\n",
340                         IWineD3DBaseTexture_GetTextureDimensions(textures[i]));
341                 sampler_type[i] = WINED3DSTT_2D;
342         }
343     }
344 }
345
346 const IWineD3DPixelShaderVtbl IWineD3DPixelShader_Vtbl =
347 {
348     /*** IUnknown methods ***/
349     IWineD3DPixelShaderImpl_QueryInterface,
350     IWineD3DPixelShaderImpl_AddRef,
351     IWineD3DPixelShaderImpl_Release,
352     /*** IWineD3DBase methods ***/
353     IWineD3DPixelShaderImpl_GetParent,
354     /*** IWineD3DBaseShader methods ***/
355     IWineD3DPixelShaderImpl_SetFunction,
356     /*** IWineD3DPixelShader methods ***/
357     IWineD3DPixelShaderImpl_GetDevice,
358     IWineD3DPixelShaderImpl_GetFunction
359 };
360
361 void find_ps_compile_args(IWineD3DPixelShaderImpl *shader, IWineD3DStateBlockImpl *stateblock, struct ps_compile_args *args) {
362     UINT i;
363     IWineD3DBaseTextureImpl *tex;
364
365     memset(args, 0, sizeof(*args)); /* FIXME: Make sure all bits are set */
366     args->srgb_correction = stateblock->renderState[WINED3DRS_SRGBWRITEENABLE] ? 1 : 0;
367     args->np2_fixup = 0;
368
369     for(i = 0; i < MAX_FRAGMENT_SAMPLERS; i++) {
370         if (!shader->baseShader.reg_maps.sampler_type[i]) continue;
371         tex = (IWineD3DBaseTextureImpl *) stateblock->textures[i];
372         if(!tex) {
373             args->color_fixup[i] = COLOR_FIXUP_IDENTITY;
374             continue;
375         }
376         args->color_fixup[i] = tex->resource.format_desc->color_fixup;
377
378         /* Flag samplers that need NP2 texcoord fixup. */
379         if(!tex->baseTexture.pow2Matrix_identity) {
380             args->np2_fixup |= (1 << i);
381         }
382     }
383     if (shader->baseShader.reg_maps.shader_version.major >= 3)
384     {
385         if (((IWineD3DDeviceImpl *)shader->baseShader.device)->strided_streams.position_transformed)
386         {
387             args->vp_mode = pretransformed;
388         }
389         else if (use_vs(stateblock))
390         {
391             args->vp_mode = vertexshader;
392         } else {
393             args->vp_mode = fixedfunction;
394         }
395         args->fog = FOG_OFF;
396     } else {
397         args->vp_mode = vertexshader;
398         if(stateblock->renderState[WINED3DRS_FOGENABLE]) {
399             switch(stateblock->renderState[WINED3DRS_FOGTABLEMODE]) {
400                 case WINED3DFOG_NONE:
401                     if (((IWineD3DDeviceImpl *)shader->baseShader.device)->strided_streams.position_transformed
402                             || use_vs(stateblock))
403                     {
404                         args->fog = FOG_LINEAR;
405                         break;
406                     }
407                     switch(stateblock->renderState[WINED3DRS_FOGVERTEXMODE]) {
408                         case WINED3DFOG_NONE: /* Drop through */
409                         case WINED3DFOG_LINEAR: args->fog = FOG_LINEAR; break;
410                         case WINED3DFOG_EXP:    args->fog = FOG_EXP;    break;
411                         case WINED3DFOG_EXP2:   args->fog = FOG_EXP2;   break;
412                     }
413                     break;
414
415                 case WINED3DFOG_LINEAR: args->fog = FOG_LINEAR; break;
416                 case WINED3DFOG_EXP:    args->fog = FOG_EXP;    break;
417                 case WINED3DFOG_EXP2:   args->fog = FOG_EXP2;   break;
418             }
419         } else {
420             args->fog = FOG_OFF;
421         }
422     }
423 }