2 * Copyright (C) 2008-2009 Advanced Micro Devices, Inc.
4 * Permission is hereby granted, free of charge, to any person obtaining a
5 * copy of this software and associated documentation files (the "Software"),
6 * to deal in the Software without restriction, including without limitation
7 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
8 * and/or sell copies of the Software, and to permit persons to whom the
9 * Software is furnished to do so, subject to the following conditions:
11 * The above copyright notice and this permission notice shall be included
12 * in all copies or substantial portions of the Software.
14 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
15 * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
16 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
17 * THE COPYRIGHT HOLDER(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
18 * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
19 * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
24 * Richard Li <RichardZ.Li@amd.com>, <richardradeon@gmail.com>
25 * CooperYuan <cooper.yuan@amd.com>, <cooperyuan@gmail.com>
34 #include "main/imports.h"
35 #include "shader/prog_parameter.h"
36 #include "shader/prog_statevars.h"
37 #include "shader/program.h"
39 #include "r600_context.h"
40 #include "r600_cmdbuf.h"
42 #include "r700_fragprog.h"
44 #include "r700_debug.h"
46 void insert_wpos_code(GLcontext
*ctx
, struct gl_fragment_program
*fprog
)
48 static const gl_state_index winstate
[STATE_LENGTH
]
49 = { STATE_INTERNAL
, STATE_FB_SIZE
, 0, 0, 0};
50 struct prog_instruction
*newInst
, *inst
;
51 GLint win_size
; /* state reference */
52 GLuint wpos_temp
; /* temp register */
55 /* PARAM win_size = STATE_FB_SIZE */
56 win_size
= _mesa_add_state_reference(fprog
->Base
.Parameters
, winstate
);
58 wpos_temp
= fprog
->Base
.NumTemporaries
++;
60 /* scan program where WPOS is used and replace with wpos_temp */
61 inst
= fprog
->Base
.Instructions
;
62 for (i
= 0; i
< fprog
->Base
.NumInstructions
; i
++) {
63 for (j
=0; j
< 3; j
++) {
64 if(inst
->SrcReg
[j
].File
== PROGRAM_INPUT
&&
65 inst
->SrcReg
[j
].Index
== FRAG_ATTRIB_WPOS
) {
66 inst
->SrcReg
[j
].File
= PROGRAM_TEMPORARY
;
67 inst
->SrcReg
[j
].Index
= wpos_temp
;
73 _mesa_insert_instructions(&(fprog
->Base
), 0, 1);
75 newInst
= fprog
->Base
.Instructions
;
77 * wpos_temp.xyzw = wpos.x-yzw + winsize.0y00 */
78 newInst
[0].Opcode
= OPCODE_ADD
;
79 newInst
[0].DstReg
.File
= PROGRAM_TEMPORARY
;
80 newInst
[0].DstReg
.Index
= wpos_temp
;
81 newInst
[0].DstReg
.WriteMask
= WRITEMASK_XYZW
;
83 newInst
[0].SrcReg
[0].File
= PROGRAM_INPUT
;
84 newInst
[0].SrcReg
[0].Index
= FRAG_ATTRIB_WPOS
;
85 newInst
[0].SrcReg
[0].Swizzle
= SWIZZLE_XYZW
;
86 newInst
[0].SrcReg
[0].Negate
= NEGATE_Y
;
88 newInst
[0].SrcReg
[1].File
= PROGRAM_STATE_VAR
;
89 newInst
[0].SrcReg
[1].Index
= win_size
;
90 newInst
[0].SrcReg
[1].Swizzle
= MAKE_SWIZZLE4(SWIZZLE_ZERO
, SWIZZLE_Y
, SWIZZLE_ZERO
, SWIZZLE_ZERO
);
94 //TODO : Validate FP input with VP output.
95 void Map_Fragment_Program(r700_AssemblerBase
*pAsm
,
96 struct gl_fragment_program
*mesa_fp
,
103 /* match fp inputs with vp exports. */
104 struct r700_vertex_program_cont
*vpc
=
105 (struct r700_vertex_program_cont
*)ctx
->VertexProgram
._Current
;
106 GLbitfield OutputsWritten
= vpc
->mesa_program
.Base
.OutputsWritten
;
108 pAsm
->number_used_registers
= 0;
110 //Input mapping : mesa_fp->Base.InputsRead set the flag, set in
111 //The flags parsed in parse_attrib_binding. FRAG_ATTRIB_COLx, FRAG_ATTRIB_TEXx, ...
112 //MUST match order in Map_Vertex_Output
113 unBit
= 1 << FRAG_ATTRIB_WPOS
;
114 if(mesa_fp
->Base
.InputsRead
& unBit
)
116 pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_WPOS
] = pAsm
->number_used_registers
++;
119 unBit
= 1 << VERT_RESULT_COL0
;
120 if(OutputsWritten
& unBit
)
122 pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_COL0
] = pAsm
->number_used_registers
++;
125 unBit
= 1 << VERT_RESULT_COL1
;
126 if(OutputsWritten
& unBit
)
128 pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_COL1
] = pAsm
->number_used_registers
++;
131 unBit
= 1 << VERT_RESULT_FOGC
;
132 if(OutputsWritten
& unBit
)
134 pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_FOGC
] = pAsm
->number_used_registers
++;
139 unBit
= 1 << (VERT_RESULT_TEX0
+ i
);
140 if(OutputsWritten
& unBit
)
142 pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_TEX0
+ i
] = pAsm
->number_used_registers
++;
146 /* order has been taken care of */
148 for(i
=VERT_RESULT_VAR0
; i
<VERT_RESULT_MAX
; i
++)
151 if(OutputsWritten
& unBit
)
153 pAsm
->uiFP_AttributeMap
[i
-VERT_RESULT_VAR0
+FRAG_ATTRIB_VAR0
] = pAsm
->number_used_registers
++;
157 if( (mesa_fp
->Base
.InputsRead
>> FRAG_ATTRIB_VAR0
) > 0 )
159 struct r700_vertex_program_cont
*vpc
=
160 (struct r700_vertex_program_cont
*)ctx
->VertexProgram
._Current
;
161 struct gl_program_parameter_list
* VsVarying
= vpc
->mesa_program
.Base
.Varying
;
162 struct gl_program_parameter_list
* PsVarying
= mesa_fp
->Base
.Varying
;
163 struct gl_program_parameter
* pVsParam
;
164 struct gl_program_parameter
* pPsParam
;
166 GLuint unMaxVarying
= 0;
168 for(i
=0; i
<VsVarying
->NumParameters
; i
++)
170 pAsm
->uiFP_AttributeMap
[i
+ FRAG_ATTRIB_VAR0
] = 0;
173 for(i
=FRAG_ATTRIB_VAR0
; i
<FRAG_ATTRIB_MAX
; i
++)
176 if(mesa_fp
->Base
.InputsRead
& unBit
)
178 j
= i
- FRAG_ATTRIB_VAR0
;
179 pPsParam
= PsVarying
->Parameters
+ j
;
181 for(k
=0; k
<VsVarying
->NumParameters
; k
++)
183 pVsParam
= VsVarying
->Parameters
+ k
;
185 if( strcmp(pPsParam
->Name
, pVsParam
->Name
) == 0)
187 pAsm
->uiFP_AttributeMap
[i
] = pAsm
->number_used_registers
+ k
;
198 pAsm
->number_used_registers
+= unMaxVarying
+ 1;
201 unBit
= 1 << FRAG_ATTRIB_FACE
;
202 if(mesa_fp
->Base
.InputsRead
& unBit
)
204 pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_FACE
] = pAsm
->number_used_registers
++;
207 unBit
= 1 << FRAG_ATTRIB_PNTC
;
208 if(mesa_fp
->Base
.InputsRead
& unBit
)
210 pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_PNTC
] = pAsm
->number_used_registers
++;
213 /* Map temporary registers (GPRs) */
214 pAsm
->starting_temp_register_number
= pAsm
->number_used_registers
;
216 if(mesa_fp
->Base
.NumNativeTemporaries
>= mesa_fp
->Base
.NumTemporaries
)
218 pAsm
->number_used_registers
+= mesa_fp
->Base
.NumNativeTemporaries
;
222 pAsm
->number_used_registers
+= mesa_fp
->Base
.NumTemporaries
;
226 pAsm
->number_of_exports
= 0;
227 pAsm
->number_of_colorandz_exports
= 0; /* don't include stencil and mask out. */
228 pAsm
->starting_export_register_number
= pAsm
->number_used_registers
;
229 unBit
= 1 << FRAG_RESULT_COLOR
;
230 if(mesa_fp
->Base
.OutputsWritten
& unBit
)
232 pAsm
->uiFP_OutputMap
[FRAG_RESULT_COLOR
] = pAsm
->number_used_registers
++;
233 pAsm
->number_of_exports
++;
234 pAsm
->number_of_colorandz_exports
++;
236 unBit
= 1 << FRAG_RESULT_DEPTH
;
237 if(mesa_fp
->Base
.OutputsWritten
& unBit
)
239 pAsm
->depth_export_register_number
= pAsm
->number_used_registers
;
240 pAsm
->uiFP_OutputMap
[FRAG_RESULT_DEPTH
] = pAsm
->number_used_registers
++;
241 pAsm
->number_of_exports
++;
242 pAsm
->number_of_colorandz_exports
++;
243 pAsm
->pR700Shader
->depthIsExported
= 1;
246 pAsm
->pucOutMask
= (unsigned char*) MALLOC(pAsm
->number_of_exports
);
247 for(ui
=0; ui
<pAsm
->number_of_exports
; ui
++)
249 pAsm
->pucOutMask
[ui
] = 0x0;
252 pAsm
->flag_reg_index
= pAsm
->number_used_registers
++;
254 pAsm
->uFirstHelpReg
= pAsm
->number_used_registers
;
257 GLboolean
Find_Instruction_Dependencies_fp(struct r700_fragment_program
*fp
,
258 struct gl_fragment_program
*mesa_fp
)
261 GLint
* puiTEMPwrites
;
262 GLint
* puiTEMPreads
;
263 struct prog_instruction
* pILInst
;
265 struct prog_instruction
* texcoord_DepInst
;
268 puiTEMPwrites
= (GLint
*) MALLOC(sizeof(GLuint
)*mesa_fp
->Base
.NumTemporaries
);
269 puiTEMPreads
= (GLint
*) MALLOC(sizeof(GLuint
)*mesa_fp
->Base
.NumTemporaries
);
271 for(i
=0; i
<mesa_fp
->Base
.NumTemporaries
; i
++)
273 puiTEMPwrites
[i
] = -1;
274 puiTEMPreads
[i
] = -1;
277 pInstDeps
= (InstDeps
*)MALLOC(sizeof(InstDeps
)*mesa_fp
->Base
.NumInstructions
);
279 for(i
=0; i
<mesa_fp
->Base
.NumInstructions
; i
++)
281 pInstDeps
[i
].nDstDep
= -1;
282 pILInst
= &(mesa_fp
->Base
.Instructions
[i
]);
285 if(pILInst
->DstReg
.File
== PROGRAM_TEMPORARY
)
287 //Set lastwrite for the temp
288 puiTEMPwrites
[pILInst
->DstReg
.Index
] = i
;
294 if(pILInst
->SrcReg
[j
].File
== PROGRAM_TEMPORARY
)
297 pInstDeps
[i
].nSrcDeps
[j
] = puiTEMPwrites
[pILInst
->SrcReg
[j
].Index
];
299 if(puiTEMPreads
[pILInst
->SrcReg
[j
].Index
] < 0 )
301 puiTEMPreads
[pILInst
->SrcReg
[j
].Index
] = i
;
306 pInstDeps
[i
].nSrcDeps
[j
] = -1;
311 fp
->r700AsmCode
.pInstDeps
= pInstDeps
;
313 //Find dep for tex inst
314 for(i
=0; i
<mesa_fp
->Base
.NumInstructions
; i
++)
316 pILInst
= &(mesa_fp
->Base
.Instructions
[i
]);
318 if(GL_TRUE
== IsTex(pILInst
->Opcode
))
319 { //src0 is the tex coord register, src1 is texunit, src2 is textype
320 nDepInstID
= pInstDeps
[i
].nSrcDeps
[0];
323 texcoord_DepInst
= &(mesa_fp
->Base
.Instructions
[nDepInstID
]);
324 if(GL_TRUE
== IsAlu(texcoord_DepInst
->Opcode
) )
326 pInstDeps
[nDepInstID
].nDstDep
= i
;
327 pInstDeps
[i
].nDstDep
= i
;
329 else if(GL_TRUE
== IsTex(texcoord_DepInst
->Opcode
) )
331 pInstDeps
[i
].nDstDep
= i
;
337 // make sure that we dont overwrite src used earlier
338 nDepInstID
= puiTEMPreads
[pILInst
->DstReg
.Index
];
341 pInstDeps
[i
].nDstDep
= puiTEMPreads
[pILInst
->DstReg
.Index
];
342 texcoord_DepInst
= &(mesa_fp
->Base
.Instructions
[nDepInstID
]);
343 if(GL_TRUE
== IsAlu(texcoord_DepInst
->Opcode
) )
345 pInstDeps
[nDepInstID
].nDstDep
= i
;
359 GLboolean
r700TranslateFragmentShader(struct r700_fragment_program
*fp
,
360 struct gl_fragment_program
*mesa_fp
,
363 GLuint number_of_colors_exported
;
364 GLboolean z_enabled
= GL_FALSE
;
365 GLuint unBit
, shadow_unit
;
367 struct prog_instruction
*inst
;
368 gl_state_index shadow_ambient
[STATE_LENGTH
]
369 = { STATE_INTERNAL
, STATE_SHADOW_AMBIENT
, 0, 0, 0};
372 Init_r700_AssemblerBase( SPT_FP
, &(fp
->r700AsmCode
), &(fp
->r700Shader
) );
374 if(mesa_fp
->Base
.InputsRead
& FRAG_BIT_WPOS
)
376 insert_wpos_code(ctx
, mesa_fp
);
379 /* add/map consts for ARB_shadow_ambient */
380 if(mesa_fp
->Base
.ShadowSamplers
)
382 inst
= mesa_fp
->Base
.Instructions
;
383 for (i
= 0; i
< mesa_fp
->Base
.NumInstructions
; i
++)
385 if(inst
->TexShadow
== 1)
387 shadow_unit
= inst
->TexSrcUnit
;
388 shadow_ambient
[2] = shadow_unit
;
389 fp
->r700AsmCode
.shadow_regs
[shadow_unit
] =
390 _mesa_add_state_reference(mesa_fp
->Base
.Parameters
, shadow_ambient
);
396 Map_Fragment_Program(&(fp
->r700AsmCode
), mesa_fp
, ctx
);
398 if( GL_FALSE
== Find_Instruction_Dependencies_fp(fp
, mesa_fp
) )
403 InitShaderProgram(&(fp
->r700AsmCode
));
405 for(i
=0; i
< MAX_SAMPLERS
; i
++)
407 fp
->r700AsmCode
.SamplerUnits
[i
] = fp
->mesa_program
.Base
.SamplerUnits
[i
];
410 fp
->r700AsmCode
.unCurNumILInsts
= mesa_fp
->Base
.NumInstructions
;
412 if( GL_FALSE
== AssembleInstr(0,
414 mesa_fp
->Base
.NumInstructions
,
415 &(mesa_fp
->Base
.Instructions
[0]),
416 &(fp
->r700AsmCode
)) )
421 if(GL_FALSE
== Process_Fragment_Exports(&(fp
->r700AsmCode
), mesa_fp
->Base
.OutputsWritten
) )
426 if( GL_FALSE
== RelocProgram(&(fp
->r700AsmCode
), &(mesa_fp
->Base
)) )
431 fp
->r700Shader
.nRegs
= (fp
->r700AsmCode
.number_used_registers
== 0) ? 0
432 : (fp
->r700AsmCode
.number_used_registers
- 1);
434 fp
->r700Shader
.nParamExports
= fp
->r700AsmCode
.number_of_exports
;
436 number_of_colors_exported
= fp
->r700AsmCode
.number_of_colorandz_exports
;
438 unBit
= 1 << FRAG_RESULT_DEPTH
;
439 if(mesa_fp
->Base
.OutputsWritten
& unBit
)
442 number_of_colors_exported
--;
445 /* illegal to set this to 0 */
446 if(number_of_colors_exported
|| z_enabled
)
448 fp
->r700Shader
.exportMode
= number_of_colors_exported
<< 1 | z_enabled
;
452 fp
->r700Shader
.exportMode
= (1 << 1);
455 fp
->translated
= GL_TRUE
;
460 void r700SelectFragmentShader(GLcontext
*ctx
)
462 context_t
*context
= R700_CONTEXT(ctx
);
463 struct r700_fragment_program
*fp
= (struct r700_fragment_program
*)
464 (ctx
->FragmentProgram
._Current
);
465 if (context
->radeon
.radeonScreen
->chip_family
< CHIP_FAMILY_RV770
)
467 fp
->r700AsmCode
.bR6xx
= 1;
470 if (GL_FALSE
== fp
->translated
)
471 r700TranslateFragmentShader(fp
, &(fp
->mesa_program
), ctx
);
474 void * r700GetActiveFpShaderBo(GLcontext
* ctx
)
476 struct r700_fragment_program
*fp
= (struct r700_fragment_program
*)
477 (ctx
->FragmentProgram
._Current
);
482 GLboolean
r700SetupFragmentProgram(GLcontext
* ctx
)
484 context_t
*context
= R700_CONTEXT(ctx
);
485 R700_CHIP_CONTEXT
*r700
= (R700_CHIP_CONTEXT
*)(&context
->hw
);
486 struct r700_fragment_program
*fp
= (struct r700_fragment_program
*)
487 (ctx
->FragmentProgram
._Current
);
488 r700_AssemblerBase
*pAsm
= &(fp
->r700AsmCode
);
489 struct gl_fragment_program
*mesa_fp
= &(fp
->mesa_program
);
490 struct gl_program_parameter_list
*paramList
;
491 unsigned int unNumParamData
;
493 unsigned int unNumOfReg
;
496 GLboolean point_sprite
= GL_FALSE
;
498 if(GL_FALSE
== fp
->loaded
)
500 if(fp
->r700Shader
.bNeedsAssembly
== GL_TRUE
)
502 Assemble( &(fp
->r700Shader
) );
508 (GLvoid
*)(fp
->r700Shader
.pProgram
),
509 fp
->r700Shader
.uShaderBinaryDWORDSize
,
512 fp
->loaded
= GL_TRUE
;
515 DumpHwBinary(DUMP_PIXEL_SHADER
, (GLvoid
*)(fp
->r700Shader
.pProgram
),
516 fp
->r700Shader
.uShaderBinaryDWORDSize
);
518 /* TODO : enable this after MemUse fixed *=
519 (context->chipobj.MemUse)(context, fp->shadercode.buf->id);
522 R600_STATECHANGE(context
, ps
);
524 r700
->ps
.SQ_PGM_RESOURCES_PS
.u32All
= 0;
525 SETbit(r700
->ps
.SQ_PGM_RESOURCES_PS
.u32All
, PGM_RESOURCES__PRIME_CACHE_ON_DRAW_bit
);
527 r700
->ps
.SQ_PGM_START_PS
.u32All
= 0; /* set from buffer obj */
529 R600_STATECHANGE(context
, spi
);
531 unNumOfReg
= fp
->r700Shader
.nRegs
+ 1;
533 ui
= (r700
->SPI_PS_IN_CONTROL_0
.u32All
& NUM_INTERP_mask
) / (1 << NUM_INTERP_shift
);
535 /* PS uses fragment.position */
536 if (mesa_fp
->Base
.InputsRead
& (1 << FRAG_ATTRIB_WPOS
))
539 SETfield(r700
->SPI_PS_IN_CONTROL_0
.u32All
, ui
, NUM_INTERP_shift
, NUM_INTERP_mask
);
540 SETfield(r700
->SPI_PS_IN_CONTROL_0
.u32All
, CENTERS_ONLY
, BARYC_SAMPLE_CNTL_shift
, BARYC_SAMPLE_CNTL_mask
);
541 SETbit(r700
->SPI_PS_IN_CONTROL_0
.u32All
, POSITION_ENA_bit
);
542 SETbit(r700
->SPI_INPUT_Z
.u32All
, PROVIDE_Z_TO_SPI_bit
);
546 CLEARbit(r700
->SPI_PS_IN_CONTROL_0
.u32All
, POSITION_ENA_bit
);
547 CLEARbit(r700
->SPI_INPUT_Z
.u32All
, PROVIDE_Z_TO_SPI_bit
);
550 if (mesa_fp
->Base
.InputsRead
& (1 << FRAG_ATTRIB_FACE
))
553 SETfield(r700
->SPI_PS_IN_CONTROL_0
.u32All
, ui
, NUM_INTERP_shift
, NUM_INTERP_mask
);
554 SETbit(r700
->SPI_PS_IN_CONTROL_1
.u32All
, FRONT_FACE_ENA_bit
);
555 SETbit(r700
->SPI_PS_IN_CONTROL_1
.u32All
, FRONT_FACE_ALL_BITS_bit
);
556 SETfield(r700
->SPI_PS_IN_CONTROL_1
.u32All
, pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_FACE
], FRONT_FACE_ADDR_shift
, FRONT_FACE_ADDR_mask
);
560 CLEARbit(r700
->SPI_PS_IN_CONTROL_1
.u32All
, FRONT_FACE_ENA_bit
);
563 /* see if we need any point_sprite replacements */
564 for (i
= VERT_RESULT_TEX0
; i
<= VERT_RESULT_TEX7
; i
++)
566 if(ctx
->Point
.CoordReplace
[i
- VERT_RESULT_TEX0
] == GL_TRUE
)
567 point_sprite
= GL_TRUE
;
570 if ((mesa_fp
->Base
.InputsRead
& (1 << FRAG_ATTRIB_PNTC
)) || point_sprite
)
572 /* for FRAG_ATTRIB_PNTC we need to increase num_interp */
573 if(mesa_fp
->Base
.InputsRead
& (1 << FRAG_ATTRIB_PNTC
))
576 SETfield(r700
->SPI_PS_IN_CONTROL_0
.u32All
, ui
, NUM_INTERP_shift
, NUM_INTERP_mask
);
578 SETbit(r700
->SPI_INTERP_CONTROL_0
.u32All
, PNT_SPRITE_ENA_bit
);
579 SETfield(r700
->SPI_INTERP_CONTROL_0
.u32All
, SPI_PNT_SPRITE_SEL_S
, PNT_SPRITE_OVRD_X_shift
, PNT_SPRITE_OVRD_X_mask
);
580 SETfield(r700
->SPI_INTERP_CONTROL_0
.u32All
, SPI_PNT_SPRITE_SEL_T
, PNT_SPRITE_OVRD_Y_shift
, PNT_SPRITE_OVRD_Y_mask
);
581 SETfield(r700
->SPI_INTERP_CONTROL_0
.u32All
, SPI_PNT_SPRITE_SEL_0
, PNT_SPRITE_OVRD_Z_shift
, PNT_SPRITE_OVRD_Z_mask
);
582 SETfield(r700
->SPI_INTERP_CONTROL_0
.u32All
, SPI_PNT_SPRITE_SEL_1
, PNT_SPRITE_OVRD_W_shift
, PNT_SPRITE_OVRD_W_mask
);
583 if(ctx
->Point
.SpriteOrigin
== GL_LOWER_LEFT
)
584 SETbit(r700
->SPI_INTERP_CONTROL_0
.u32All
, PNT_SPRITE_TOP_1_bit
);
586 CLEARbit(r700
->SPI_INTERP_CONTROL_0
.u32All
, PNT_SPRITE_TOP_1_bit
);
590 CLEARbit(r700
->SPI_INTERP_CONTROL_0
.u32All
, PNT_SPRITE_ENA_bit
);
594 ui
= (unNumOfReg
< ui
) ? ui
: unNumOfReg
;
596 SETfield(r700
->ps
.SQ_PGM_RESOURCES_PS
.u32All
, ui
, NUM_GPRS_shift
, NUM_GPRS_mask
);
598 CLEARbit(r700
->ps
.SQ_PGM_RESOURCES_PS
.u32All
, UNCACHED_FIRST_INST_bit
);
600 if(fp
->r700Shader
.uStackSize
) /* we don't use branch for now, it should be zero. */
602 SETfield(r700
->ps
.SQ_PGM_RESOURCES_PS
.u32All
, fp
->r700Shader
.uStackSize
,
603 STACK_SIZE_shift
, STACK_SIZE_mask
);
606 SETfield(r700
->ps
.SQ_PGM_EXPORTS_PS
.u32All
, fp
->r700Shader
.exportMode
,
607 EXPORT_MODE_shift
, EXPORT_MODE_mask
);
610 struct r700_vertex_program_cont
*vpc
=
611 (struct r700_vertex_program_cont
*)ctx
->VertexProgram
._Current
;
612 GLbitfield OutputsWritten
= vpc
->mesa_program
.Base
.OutputsWritten
;
614 for(ui
= 0; ui
< R700_MAX_SHADER_EXPORTS
; ui
++)
615 r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
= 0;
617 unBit
= 1 << FRAG_ATTRIB_WPOS
;
618 if(mesa_fp
->Base
.InputsRead
& unBit
)
620 ui
= pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_WPOS
];
621 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, SEL_CENTROID_bit
);
622 SETfield(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, ui
,
623 SEMANTIC_shift
, SEMANTIC_mask
);
624 if (r700
->SPI_INTERP_CONTROL_0
.u32All
& FLAT_SHADE_ENA_bit
)
625 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
627 CLEARbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
630 unBit
= 1 << VERT_RESULT_COL0
;
631 if(OutputsWritten
& unBit
)
633 ui
= pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_COL0
];
634 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, SEL_CENTROID_bit
);
635 SETfield(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, ui
,
636 SEMANTIC_shift
, SEMANTIC_mask
);
637 if (r700
->SPI_INTERP_CONTROL_0
.u32All
& FLAT_SHADE_ENA_bit
)
638 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
640 CLEARbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
643 unBit
= 1 << VERT_RESULT_COL1
;
644 if(OutputsWritten
& unBit
)
646 ui
= pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_COL1
];
647 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, SEL_CENTROID_bit
);
648 SETfield(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, ui
,
649 SEMANTIC_shift
, SEMANTIC_mask
);
650 if (r700
->SPI_INTERP_CONTROL_0
.u32All
& FLAT_SHADE_ENA_bit
)
651 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
653 CLEARbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
656 unBit
= 1 << VERT_RESULT_FOGC
;
657 if(OutputsWritten
& unBit
)
659 ui
= pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_FOGC
];
660 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, SEL_CENTROID_bit
);
661 SETfield(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, ui
,
662 SEMANTIC_shift
, SEMANTIC_mask
);
663 if (r700
->SPI_INTERP_CONTROL_0
.u32All
& FLAT_SHADE_ENA_bit
)
664 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
666 CLEARbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
671 unBit
= 1 << (VERT_RESULT_TEX0
+ i
);
672 if(OutputsWritten
& unBit
)
674 ui
= pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_TEX0
+ i
];
675 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, SEL_CENTROID_bit
);
676 SETfield(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, ui
,
677 SEMANTIC_shift
, SEMANTIC_mask
);
678 CLEARbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
679 /* ARB_point_sprite */
680 if(ctx
->Point
.CoordReplace
[i
] == GL_TRUE
)
682 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, PT_SPRITE_TEX_bit
);
687 unBit
= 1 << FRAG_ATTRIB_FACE
;
688 if(mesa_fp
->Base
.InputsRead
& unBit
)
690 ui
= pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_FACE
];
691 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, SEL_CENTROID_bit
);
692 SETfield(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, ui
,
693 SEMANTIC_shift
, SEMANTIC_mask
);
694 if (r700
->SPI_INTERP_CONTROL_0
.u32All
& FLAT_SHADE_ENA_bit
)
695 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
697 CLEARbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
699 unBit
= 1 << FRAG_ATTRIB_PNTC
;
700 if(mesa_fp
->Base
.InputsRead
& unBit
)
702 ui
= pAsm
->uiFP_AttributeMap
[FRAG_ATTRIB_PNTC
];
703 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, SEL_CENTROID_bit
);
704 SETfield(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, ui
,
705 SEMANTIC_shift
, SEMANTIC_mask
);
706 if (r700
->SPI_INTERP_CONTROL_0
.u32All
& FLAT_SHADE_ENA_bit
)
707 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
709 CLEARbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
710 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, PT_SPRITE_TEX_bit
);
716 for(i
=VERT_RESULT_VAR0
; i
<VERT_RESULT_MAX
; i
++)
719 if(OutputsWritten
& unBit
)
721 ui
= pAsm
->uiFP_AttributeMap
[i
-VERT_RESULT_VAR0
+FRAG_ATTRIB_VAR0
];
722 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, SEL_CENTROID_bit
);
723 SETfield(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, ui
,
724 SEMANTIC_shift
, SEMANTIC_mask
);
725 if (r700
->SPI_INTERP_CONTROL_0
.u32All
& FLAT_SHADE_ENA_bit
)
726 SETbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
728 CLEARbit(r700
->SPI_PS_INPUT_CNTL
[ui
].u32All
, FLAT_SHADE_bit
);
732 exportCount
= (r700
->ps
.SQ_PGM_EXPORTS_PS
.u32All
& EXPORT_MODE_mask
) / (1 << EXPORT_MODE_shift
);
733 if (r700
->CB_SHADER_CONTROL
.u32All
!= ((1 << exportCount
) - 1))
735 R600_STATECHANGE(context
, cb
);
736 r700
->CB_SHADER_CONTROL
.u32All
= (1 << exportCount
) - 1;
739 /* sent out shader constants. */
740 paramList
= fp
->mesa_program
.Base
.Parameters
;
742 if(NULL
!= paramList
)
744 _mesa_load_state_parameters(ctx
, paramList
);
746 if (paramList
->NumParameters
> R700_MAX_DX9_CONSTS
)
749 R600_STATECHANGE(context
, ps_consts
);
751 r700
->ps
.num_consts
= paramList
->NumParameters
;
753 unNumParamData
= paramList
->NumParameters
;
755 for(ui
=0; ui
<unNumParamData
; ui
++) {
756 r700
->ps
.consts
[ui
][0].f32All
= paramList
->ParameterValues
[ui
][0];
757 r700
->ps
.consts
[ui
][1].f32All
= paramList
->ParameterValues
[ui
][1];
758 r700
->ps
.consts
[ui
][2].f32All
= paramList
->ParameterValues
[ui
][2];
759 r700
->ps
.consts
[ui
][3].f32All
= paramList
->ParameterValues
[ui
][3];
762 r700
->ps
.num_consts
= 0;
764 COMPILED_SUB
* pCompiledSub
;
766 GLuint unConstOffset
= r700
->ps
.num_consts
;
767 for(ui
=0; ui
<pAsm
->unNumPresub
; ui
++)
769 pCompiledSub
= pAsm
->presubs
[ui
].pCompiledSub
;
771 r700
->ps
.num_consts
+= pCompiledSub
->NumParameters
;
773 for(uj
=0; uj
<pCompiledSub
->NumParameters
; uj
++)
775 r700
->ps
.consts
[uj
+ unConstOffset
][0].f32All
= pCompiledSub
->ParameterValues
[uj
][0];
776 r700
->ps
.consts
[uj
+ unConstOffset
][1].f32All
= pCompiledSub
->ParameterValues
[uj
][1];
777 r700
->ps
.consts
[uj
+ unConstOffset
][2].f32All
= pCompiledSub
->ParameterValues
[uj
][2];
778 r700
->ps
.consts
[uj
+ unConstOffset
][3].f32All
= pCompiledSub
->ParameterValues
[uj
][3];
780 unConstOffset
+= pCompiledSub
->NumParameters
;