X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fmesa%2Fmain%2Ftexenvprogram.c;h=3736138b9ed8b6f911163f1adc59a133b3410fa8;hb=56c4226fcc54158eb7fe54eeb13539a979ec155c;hp=3cb103f51f8494b0db48954f454e0a9d8774b6d5;hpb=ee0735fa2d4b2d2a9eae1f55e8f93f1f57feaf6d;p=mesa.git diff --git a/src/mesa/main/texenvprogram.c b/src/mesa/main/texenvprogram.c index 3cb103f51f8..3736138b9ed 100644 --- a/src/mesa/main/texenvprogram.c +++ b/src/mesa/main/texenvprogram.c @@ -2,6 +2,7 @@ * * Copyright 2007 Tungsten Graphics, Inc., Cedar Park, Texas. * All Rights Reserved. + * Copyright 2009 VMware, Inc. All Rights Reserved. * * Permission is hereby granted, free of charge, to any person obtaining a * copy of this software and associated documentation files (the @@ -38,6 +39,21 @@ #include "texenvprogram.h" +/* + * Note on texture units: + * + * The number of texture units supported by fixed-function fragment + * processing is MAX_TEXTURE_COORD_UNITS, not MAX_TEXTURE_IMAGE_UNITS. + * That's because there's a one-to-one correspondence between texture + * coordinates and samplers in fixed-function processing. + * + * Since fixed-function vertex processing is limited to MAX_TEXTURE_COORD_UNITS + * sets of texcoords, so is fixed-function fragment processing. + * + * We can safely use ctx->Const.MaxTextureUnits for loop bounds. + */ + + struct texenvprog_cache_item { GLuint hash; @@ -46,17 +62,29 @@ struct texenvprog_cache_item struct texenvprog_cache_item *next; }; +static GLboolean +texenv_doing_secondary_color(GLcontext *ctx) +{ + if (ctx->Light.Enabled && + (ctx->Light.Model.ColorControl == GL_SEPARATE_SPECULAR_COLOR)) + return GL_TRUE; + + if (ctx->Fog.ColorSumEnabled) + return GL_TRUE; + + return GL_FALSE; +} /** * Up to nine instructions per tex unit, plus fog, specular color. */ -#define MAX_INSTRUCTIONS ((MAX_TEXTURE_UNITS * 9) + 12) +#define MAX_INSTRUCTIONS ((MAX_TEXTURE_COORD_UNITS * 9) + 12) #define DISASSEM (MESA_VERBOSE & VERBOSE_DISASSEM) struct mode_opt { - GLubyte Source:4; - GLubyte Operand:3; + GLuint Source:4; + GLuint Operand:3; }; struct state_key { @@ -74,13 +102,13 @@ struct state_key { GLuint ScaleShiftRGB:2; GLuint ScaleShiftA:2; - GLuint NumArgsRGB:2; - GLuint ModeRGB:4; - GLuint NumArgsA:2; - GLuint ModeA:4; + GLuint NumArgsRGB:3; + GLuint ModeRGB:5; + struct mode_opt OptRGB[MAX_COMBINER_TERMS]; - struct mode_opt OptRGB[3]; - struct mode_opt OptA[3]; + GLuint NumArgsA:3; + GLuint ModeA:5; + struct mode_opt OptA[MAX_COMBINER_TERMS]; } unit[8]; }; @@ -116,7 +144,9 @@ static GLuint translate_operand( GLenum operand ) case GL_ONE_MINUS_SRC_ALPHA: return OPR_ONE_MINUS_SRC_ALPHA; case GL_ZERO: return OPR_ZERO; case GL_ONE: return OPR_ONE; - default: return OPR_UNKNOWN; + default: + assert(0); + return OPR_UNKNOWN; } } @@ -132,6 +162,7 @@ static GLuint translate_operand( GLenum operand ) #define SRC_CONSTANT 9 #define SRC_PRIMARY_COLOR 10 #define SRC_PREVIOUS 11 +#define SRC_ZERO 12 #define SRC_UNKNOWN 15 static GLuint translate_source( GLenum src ) @@ -149,32 +180,50 @@ static GLuint translate_source( GLenum src ) case GL_CONSTANT: return SRC_CONSTANT; case GL_PRIMARY_COLOR: return SRC_PRIMARY_COLOR; case GL_PREVIOUS: return SRC_PREVIOUS; - default: return SRC_UNKNOWN; + case GL_ZERO: + return SRC_ZERO; + default: + assert(0); + return SRC_UNKNOWN; } } -#define MODE_REPLACE 0 -#define MODE_MODULATE 1 -#define MODE_ADD 2 -#define MODE_ADD_SIGNED 3 -#define MODE_INTERPOLATE 4 -#define MODE_SUBTRACT 5 -#define MODE_DOT3_RGB 6 -#define MODE_DOT3_RGB_EXT 7 -#define MODE_DOT3_RGBA 8 -#define MODE_DOT3_RGBA_EXT 9 -#define MODE_MODULATE_ADD_ATI 10 -#define MODE_MODULATE_SIGNED_ADD_ATI 11 -#define MODE_MODULATE_SUBTRACT_ATI 12 -#define MODE_UNKNOWN 15 - -static GLuint translate_mode( GLenum mode ) +#define MODE_REPLACE 0 /* r = a0 */ +#define MODE_MODULATE 1 /* r = a0 * a1 */ +#define MODE_ADD 2 /* r = a0 + a1 */ +#define MODE_ADD_SIGNED 3 /* r = a0 + a1 - 0.5 */ +#define MODE_INTERPOLATE 4 /* r = a0 * a2 + a1 * (1 - a2) */ +#define MODE_SUBTRACT 5 /* r = a0 - a1 */ +#define MODE_DOT3_RGB 6 /* r = a0 . a1 */ +#define MODE_DOT3_RGB_EXT 7 /* r = a0 . a1 */ +#define MODE_DOT3_RGBA 8 /* r = a0 . a1 */ +#define MODE_DOT3_RGBA_EXT 9 /* r = a0 . a1 */ +#define MODE_MODULATE_ADD_ATI 10 /* r = a0 * a2 + a1 */ +#define MODE_MODULATE_SIGNED_ADD_ATI 11 /* r = a0 * a2 + a1 - 0.5 */ +#define MODE_MODULATE_SUBTRACT_ATI 12 /* r = a0 * a2 - a1 */ +#define MODE_ADD_PRODUCTS 13 /* r = a0 * a1 + a2 * a3 */ +#define MODE_ADD_PRODUCTS_SIGNED 14 /* r = a0 * a1 + a2 * a3 - 0.5 */ +#define MODE_BUMP_ENVMAP_ATI 15 /* special */ +#define MODE_UNKNOWN 16 + +/** + * Translate GL combiner state into a MODE_x value + */ +static GLuint translate_mode( GLenum envMode, GLenum mode ) { switch (mode) { case GL_REPLACE: return MODE_REPLACE; case GL_MODULATE: return MODE_MODULATE; - case GL_ADD: return MODE_ADD; - case GL_ADD_SIGNED: return MODE_ADD_SIGNED; + case GL_ADD: + if (envMode == GL_COMBINE4_NV) + return MODE_ADD_PRODUCTS; + else + return MODE_ADD; + case GL_ADD_SIGNED: + if (envMode == GL_COMBINE4_NV) + return MODE_ADD_PRODUCTS_SIGNED; + else + return MODE_ADD_SIGNED; case GL_INTERPOLATE: return MODE_INTERPOLATE; case GL_SUBTRACT: return MODE_SUBTRACT; case GL_DOT3_RGB: return MODE_DOT3_RGB; @@ -184,20 +233,29 @@ static GLuint translate_mode( GLenum mode ) case GL_MODULATE_ADD_ATI: return MODE_MODULATE_ADD_ATI; case GL_MODULATE_SIGNED_ADD_ATI: return MODE_MODULATE_SIGNED_ADD_ATI; case GL_MODULATE_SUBTRACT_ATI: return MODE_MODULATE_SUBTRACT_ATI; - default: return MODE_UNKNOWN; + case GL_BUMP_ENVMAP_ATI: return MODE_BUMP_ENVMAP_ATI; + default: + assert(0); + return MODE_UNKNOWN; } } #define TEXTURE_UNKNOWN_INDEX 7 static GLuint translate_tex_src_bit( GLbitfield bit ) { + /* make sure number of switch cases is correct */ + assert(NUM_TEXTURE_TARGETS == 7); switch (bit) { case TEXTURE_1D_BIT: return TEXTURE_1D_INDEX; case TEXTURE_2D_BIT: return TEXTURE_2D_INDEX; case TEXTURE_RECT_BIT: return TEXTURE_RECT_INDEX; case TEXTURE_3D_BIT: return TEXTURE_3D_INDEX; case TEXTURE_CUBE_BIT: return TEXTURE_CUBE_INDEX; - default: return TEXTURE_UNKNOWN_INDEX; + case TEXTURE_1D_ARRAY_BIT: return TEXTURE_1D_ARRAY_INDEX; + case TEXTURE_2D_ARRAY_BIT: return TEXTURE_2D_ARRAY_INDEX; + default: + assert(0); + return TEXTURE_UNKNOWN_INDEX; } } @@ -215,6 +273,11 @@ static GLuint translate_tex_src_bit( GLbitfield bit ) */ static GLbitfield get_fp_input_mask( GLcontext *ctx ) { + /* _NEW_PROGRAM */ + const GLboolean vertexShader = (ctx->Shader.CurrentProgram && + ctx->Shader.CurrentProgram->LinkStatus && + ctx->Shader.CurrentProgram->VertexProgram); + const GLboolean vertexProgram = ctx->VertexProgram._Enabled; GLbitfield fp_inputs = 0x0; if (ctx->VertexProgram._Overriden) { @@ -225,38 +288,44 @@ static GLbitfield get_fp_input_mask( GLcontext *ctx ) fp_inputs = ~0; } else if (ctx->RenderMode == GL_FEEDBACK) { + /* _NEW_RENDERMODE */ fp_inputs = (FRAG_BIT_COL0 | FRAG_BIT_TEX0); } - else if (!ctx->VertexProgram._Enabled || + else if (!(vertexProgram || vertexShader) || !ctx->VertexProgram._Current) { - - /* Fixed function logic */ + /* Fixed function vertex logic */ + /* _NEW_ARRAY */ GLbitfield varying_inputs = ctx->varying_vp_inputs; /* These get generated in the setup routine regardless of the * vertex program: */ + /* _NEW_POINT */ if (ctx->Point.PointSprite) varying_inputs |= FRAG_BITS_TEX_ANY; /* First look at what values may be computed by the generated * vertex program: */ + /* _NEW_LIGHT */ if (ctx->Light.Enabled) { fp_inputs |= FRAG_BIT_COL0; - if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR) + if (texenv_doing_secondary_color(ctx)) fp_inputs |= FRAG_BIT_COL1; } + /* _NEW_TEXTURE */ fp_inputs |= (ctx->Texture._TexGenEnabled | ctx->Texture._TexMatEnabled) << FRAG_ATTRIB_TEX0; /* Then look at what might be varying as a result of enabled * arrays, etc: */ - if (varying_inputs & VERT_BIT_COLOR0) fp_inputs |= FRAG_BIT_COL0; - if (varying_inputs & VERT_BIT_COLOR1) fp_inputs |= FRAG_BIT_COL1; + if (varying_inputs & VERT_BIT_COLOR0) + fp_inputs |= FRAG_BIT_COL0; + if (varying_inputs & VERT_BIT_COLOR1) + fp_inputs |= FRAG_BIT_COL1; fp_inputs |= (((varying_inputs & VERT_BIT_TEX_ANY) >> VERT_ATTRIB_TEX0) << FRAG_ATTRIB_TEX0); @@ -264,16 +333,31 @@ static GLbitfield get_fp_input_mask( GLcontext *ctx ) } else { /* calculate from vp->outputs */ - GLbitfield vp_outputs = ctx->VertexProgram._Current->Base.OutputsWritten; + struct gl_vertex_program *vprog; + GLbitfield vp_outputs; + + /* Choose GLSL vertex shader over ARB vertex program. Need this + * since vertex shader state validation comes after fragment state + * validation (see additional comments in state.c). + */ + if (vertexShader) + vprog = ctx->Shader.CurrentProgram->VertexProgram; + else + vprog = ctx->VertexProgram.Current; + + vp_outputs = vprog->Base.OutputsWritten; /* These get generated in the setup routine regardless of the * vertex program: */ + /* _NEW_POINT */ if (ctx->Point.PointSprite) vp_outputs |= FRAG_BITS_TEX_ANY; - if (vp_outputs & (1 << VERT_RESULT_COL0)) fp_inputs |= FRAG_BIT_COL0; - if (vp_outputs & (1 << VERT_RESULT_COL1)) fp_inputs |= FRAG_BIT_COL1; + if (vp_outputs & (1 << VERT_RESULT_COL0)) + fp_inputs |= FRAG_BIT_COL0; + if (vp_outputs & (1 << VERT_RESULT_COL1)) + fp_inputs |= FRAG_BIT_COL1; fp_inputs |= (((vp_outputs & VERT_RESULT_TEX_ANY) >> VERT_RESULT_TEX0) << FRAG_ATTRIB_TEX0); @@ -295,7 +379,8 @@ static void make_state_key( GLcontext *ctx, struct state_key *key ) memset(key, 0, sizeof(*key)); - for (i=0;iConst.MaxTextureUnits; i++) { const struct gl_texture_unit *texUnit = &ctx->Texture.Unit[i]; GLenum format; @@ -319,14 +404,14 @@ static void make_state_key( GLcontext *ctx, struct state_key *key ) key->unit[i].NumArgsA = texUnit->_CurrentCombine->_NumArgsA; key->unit[i].ModeRGB = - translate_mode(texUnit->_CurrentCombine->ModeRGB); + translate_mode(texUnit->EnvMode, texUnit->_CurrentCombine->ModeRGB); key->unit[i].ModeA = - translate_mode(texUnit->_CurrentCombine->ModeA); - + translate_mode(texUnit->EnvMode, texUnit->_CurrentCombine->ModeA); + key->unit[i].ScaleShiftRGB = texUnit->_CurrentCombine->ScaleShiftRGB; key->unit[i].ScaleShiftA = texUnit->_CurrentCombine->ScaleShiftA; - for (j=0;j<3;j++) { + for (j = 0; j < MAX_COMBINER_TERMS; j++) { key->unit[i].OptRGB[j].Operand = translate_operand(texUnit->_CurrentCombine->OperandRGB[j]); key->unit[i].OptA[j].Operand = @@ -336,13 +421,25 @@ static void make_state_key( GLcontext *ctx, struct state_key *key ) key->unit[i].OptA[j].Source = translate_source(texUnit->_CurrentCombine->SourceA[j]); } + + if (key->unit[i].ModeRGB == MODE_BUMP_ENVMAP_ATI) { + /* requires some special translation */ + key->unit[i].NumArgsRGB = 2; + key->unit[i].ScaleShiftRGB = 0; + key->unit[i].OptRGB[0].Operand = OPR_SRC_COLOR; + key->unit[i].OptRGB[0].Source = SRC_TEXTURE; + key->unit[i].OptRGB[1].Operand = OPR_SRC_COLOR; + key->unit[i].OptRGB[1].Source = texUnit->BumpTarget - GL_TEXTURE0 + SRC_TEXTURE0; + } } - - if (ctx->_TriangleCaps & DD_SEPARATE_SPECULAR) { + + /* _NEW_LIGHT | _NEW_FOG */ + if (texenv_doing_secondary_color(ctx)) { key->separate_specular = 1; inputs_referenced |= FRAG_BIT_COL1; } + /* _NEW_FOG */ if (ctx->Fog.Enabled) { key->fog_enabled = 1; key->fog_mode = translate_fog_mode(ctx->Fog.Mode); @@ -398,11 +495,16 @@ struct texenv_fragment_program { GLbitfield temp_in_use; /**< Tracks temporary regs which are in use. */ GLboolean error; - struct ureg src_texture[MAX_TEXTURE_UNITS]; + struct ureg src_texture[MAX_TEXTURE_COORD_UNITS]; /* Reg containing each texture unit's sampled texture color, * else undef. */ + struct ureg texcoord_tex[MAX_TEXTURE_COORD_UNITS]; + /* Reg containing texcoord for a texture unit, + * needed for bump mapping, else undef. + */ + struct ureg src_previous; /**< Reg containing color from previous * stage. May need to be decl'd. */ @@ -588,9 +690,8 @@ static void emit_arg( struct prog_src_register *reg, reg->File = ureg.file; reg->Index = ureg.idx; reg->Swizzle = ureg.swz; - reg->NegateBase = ureg.negatebase ? 0xf : 0x0; + reg->Negate = ureg.negatebase ? NEGATE_XYZW : NEGATE_NONE; reg->Abs = ureg.abs; - reg->NegateAbs = ureg.negateabs; } static void emit_dst( struct prog_dst_register *dst, @@ -675,6 +776,7 @@ static struct ureg emit_texld( struct texenv_fragment_program *p, GLuint destmask, GLuint tex_unit, GLuint tex_idx, + GLuint tex_shadow, struct ureg coord ) { struct prog_instruction *inst = emit_op( p, op, @@ -686,6 +788,7 @@ static struct ureg emit_texld( struct texenv_fragment_program *p, inst->TexSrcTarget = tex_idx; inst->TexSrcUnit = tex_unit; + inst->TexShadow = tex_shadow; p->program->Base.NumTexInstructions++; @@ -693,6 +796,7 @@ static struct ureg emit_texld( struct texenv_fragment_program *p, */ reserve_temp(p, dest); +#if 0 /* Is this a texture indirection? */ if ((coord.file == PROGRAM_TEMPORARY && @@ -704,6 +808,7 @@ static struct ureg emit_texld( struct texenv_fragment_program *p, p->alu_temps = 0; assert(0); /* KW: texture env crossbar */ } +#endif return dest; } @@ -788,12 +893,18 @@ static struct ureg get_source( struct texenv_fragment_program *p, case SRC_PRIMARY_COLOR: return register_input(p, FRAG_ATTRIB_COL0); + case SRC_ZERO: + return get_zero(p); + case SRC_PREVIOUS: - default: if (is_undef(p->src_previous)) return register_input(p, FRAG_ATTRIB_COL0); else return p->src_previous; + + default: + assert(0); + return undef; } } @@ -834,7 +945,9 @@ static struct ureg emit_combine_source( struct texenv_fragment_program *p, case OPR_ONE: return get_one(p); case OPR_SRC_COLOR: + return src; default: + assert(0); return src; } } @@ -883,10 +996,12 @@ static struct ureg emit_combine( struct texenv_fragment_program *p, GLuint mode, const struct mode_opt *opt) { - struct ureg src[3]; + struct ureg src[MAX_COMBINER_TERMS]; struct ureg tmp, half; GLuint i; + assert(nr <= MAX_COMBINER_TERMS); + tmp = undef; /* silence warning (bug 5318) */ for (i = 0; i < nr; i++) @@ -962,7 +1077,30 @@ static struct ureg emit_combine( struct texenv_fragment_program *p, /* Arg0 * Arg2 - Arg1 */ emit_arith( p, OPCODE_MAD, dest, mask, 0, src[0], src[2], negate(src[1]) ); return dest; + case MODE_ADD_PRODUCTS: + /* Arg0 * Arg1 + Arg2 * Arg3 */ + { + struct ureg tmp0 = get_temp(p); + emit_arith( p, OPCODE_MUL, tmp0, mask, 0, src[0], src[1], undef ); + emit_arith( p, OPCODE_MAD, dest, mask, saturate, src[2], src[3], tmp0 ); + } + return dest; + case MODE_ADD_PRODUCTS_SIGNED: + /* Arg0 * Arg1 + Arg2 * Arg3 - 0.5 */ + { + struct ureg tmp0 = get_temp(p); + half = get_half(p); + emit_arith( p, OPCODE_MUL, tmp0, mask, 0, src[0], src[1], undef ); + emit_arith( p, OPCODE_MAD, tmp0, mask, 0, src[2], src[3], tmp0 ); + emit_arith( p, OPCODE_SUB, dest, mask, saturate, tmp0, half, undef ); + } + return dest; + case MODE_BUMP_ENVMAP_ATI: + /* special - not handled here */ + assert(0); + return src[0]; default: + assert(0); return src[0]; } } @@ -983,6 +1121,10 @@ emit_texenv(struct texenv_fragment_program *p, GLuint unit) if (!key->unit[unit].enabled) { return get_source(p, SRC_PREVIOUS, 0); } + if (key->unit[unit].ModeRGB == MODE_BUMP_ENVMAP_ATI) { + /* this isn't really a env stage delivering a color and handled elsewhere */ + return get_source(p, SRC_PREVIOUS, 0); + } switch (key->unit[unit].ModeRGB) { case MODE_DOT3_RGB_EXT: @@ -1007,7 +1149,7 @@ emit_texenv(struct texenv_fragment_program *p, GLuint unit) rgb_shift) dest = get_temp( p ); else - dest = make_ureg(PROGRAM_OUTPUT, FRAG_RESULT_COLR); + dest = make_ureg(PROGRAM_OUTPUT, FRAG_RESULT_COLOR); /* Emit the RGB and A combine ops */ @@ -1071,22 +1213,35 @@ emit_texenv(struct texenv_fragment_program *p, GLuint unit) static void load_texture( struct texenv_fragment_program *p, GLuint unit ) { if (is_undef(p->src_texture[unit])) { - GLuint dim = p->state->unit[unit].source_index; - struct ureg texcoord = register_input(p, FRAG_ATTRIB_TEX0+unit); + GLuint texTarget = p->state->unit[unit].source_index; + struct ureg texcoord; struct ureg tmp = get_tex_temp( p ); - if (dim == TEXTURE_UNKNOWN_INDEX) + if (is_undef(p->texcoord_tex[unit])) { + texcoord = register_input(p, FRAG_ATTRIB_TEX0+unit); + } + else { + /* might want to reuse this reg for tex output actually */ + texcoord = p->texcoord_tex[unit]; + } + + if (texTarget == TEXTURE_UNKNOWN_INDEX) program_error(p, "TexSrcBit"); /* TODO: Use D0_MASK_XY where possible. */ if (p->state->unit[unit].enabled) { - p->src_texture[unit] = emit_texld( p, OPCODE_TXP, - tmp, WRITEMASK_XYZW, - unit, dim, texcoord ); + GLboolean shadow = GL_FALSE; - if (p->state->unit[unit].shadow) + if (p->state->unit[unit].shadow) { p->program->Base.ShadowSamplers |= 1 << unit; + shadow = GL_TRUE; + } + + p->src_texture[unit] = emit_texld( p, OPCODE_TXP, + tmp, WRITEMASK_XYZW, + unit, texTarget, shadow, + texcoord ); p->program->Base.SamplersUsed |= (1 << unit); /* This identity mapping should already be in place @@ -1119,6 +1274,7 @@ static GLboolean load_texenv_source( struct texenv_fragment_program *p, break; default: + /* not a texture src - do nothing */ break; } @@ -1136,7 +1292,7 @@ load_texunit_sources( struct texenv_fragment_program *p, int unit ) GLuint i; for (i = 0; i < key->unit[unit].NumArgsRGB; i++) { - load_texenv_source( p, key->unit[unit].OptRGB[i].Source, unit); + load_texenv_source( p, key->unit[unit].OptRGB[i].Source, unit ); } for (i = 0; i < key->unit[unit].NumArgsA; i++) { @@ -1146,6 +1302,40 @@ load_texunit_sources( struct texenv_fragment_program *p, int unit ) return GL_TRUE; } +/** + * Generate instructions for loading bump map textures. + */ +static GLboolean +load_texunit_bumpmap( struct texenv_fragment_program *p, int unit ) +{ + struct state_key *key = p->state; + GLuint bumpedUnitNr = key->unit[unit].OptRGB[1].Source - SRC_TEXTURE0; + struct ureg texcDst, bumpMapRes; + struct ureg constdudvcolor = register_const4f(p, 0.0, 0.0, 0.0, 1.0); + struct ureg texcSrc = register_input(p, FRAG_ATTRIB_TEX0 + bumpedUnitNr); + struct ureg rotMat0 = register_param3( p, STATE_INTERNAL, STATE_ROT_MATRIX_0, unit ); + struct ureg rotMat1 = register_param3( p, STATE_INTERNAL, STATE_ROT_MATRIX_1, unit ); + + load_texenv_source( p, unit + SRC_TEXTURE0, unit ); + + bumpMapRes = get_source(p, key->unit[unit].OptRGB[0].Source, unit); + texcDst = get_tex_temp( p ); + p->texcoord_tex[bumpedUnitNr] = texcDst; + + /* apply rot matrix and add coords to be available in next phase */ + /* dest = (Arg0.xxxx * rotMat0 + Arg1) + (Arg0.yyyy * rotMat1) */ + /* note only 2 coords are affected the rest are left unchanged (mul by 0) */ + emit_arith( p, OPCODE_MAD, texcDst, WRITEMASK_XYZW, 0, + swizzle1(bumpMapRes, SWIZZLE_X), rotMat0, texcSrc ); + emit_arith( p, OPCODE_MAD, texcDst, WRITEMASK_XYZW, 0, + swizzle1(bumpMapRes, SWIZZLE_Y), rotMat1, texcDst ); + + /* move 0,0,0,1 into bumpmap src if someone (crossbar) is foolish + enough to access this later, should optimize away */ + emit_arith( p, OPCODE_MOV, bumpMapRes, WRITEMASK_XYZW, 0, constdudvcolor, undef, undef ); + + return GL_TRUE; +} /** * Generate a new fragment program which implements the context's @@ -1170,7 +1360,7 @@ create_new_program(GLcontext *ctx, struct state_key *key, */ p.program->Base.Instructions = instBuffer; p.program->Base.Target = GL_FRAGMENT_PROGRAM_ARB; - p.program->Base.NumTexIndirections = 1; /* correct? */ + p.program->Base.NumTexIndirections = 1; p.program->Base.NumTexInstructions = 0; p.program->Base.NumAluInstructions = 0; p.program->Base.String = NULL; @@ -1181,10 +1371,12 @@ create_new_program(GLcontext *ctx, struct state_key *key, p.program->Base.Parameters = _mesa_new_parameter_list(); p.program->Base.InputsRead = 0; - p.program->Base.OutputsWritten = 1 << FRAG_RESULT_COLR; + p.program->Base.OutputsWritten = 1 << FRAG_RESULT_COLOR; - for (unit = 0; unit < MAX_TEXTURE_UNITS; unit++) + for (unit = 0; unit < ctx->Const.MaxTextureUnits; unit++) { p.src_texture[unit] = undef; + p.texcoord_tex[unit] = undef; + } p.src_previous = undef; p.half = undef; @@ -1195,6 +1387,16 @@ create_new_program(GLcontext *ctx, struct state_key *key, release_temps(ctx, &p); if (key->enabled_units) { + GLboolean needbumpstage = GL_FALSE; + /* Zeroth pass - bump map textures first */ + for (unit = 0 ; unit < ctx->Const.MaxTextureUnits ; unit++) + if (key->unit[unit].enabled && key->unit[unit].ModeRGB == MODE_BUMP_ENVMAP_ATI) { + needbumpstage = GL_TRUE; + load_texunit_bumpmap( &p, unit ); + } + if (needbumpstage) + p.program->Base.NumTexIndirections++; + /* First pass - to support texture_env_crossbar, first identify * all referenced texture sources and emit texld instructions * for each: @@ -1216,7 +1418,7 @@ create_new_program(GLcontext *ctx, struct state_key *key, } cf = get_source( &p, SRC_PREVIOUS, 0 ); - out = make_ureg( PROGRAM_OUTPUT, FRAG_RESULT_COLR ); + out = make_ureg( PROGRAM_OUTPUT, FRAG_RESULT_COLOR ); if (key->separate_specular) { /* Emit specular add. @@ -1315,42 +1517,3 @@ _mesa_get_fixed_func_fragment_program(GLcontext *ctx) return prog; } - - - -/** - * If _MaintainTexEnvProgram is set we'll generate a fragment program that - * implements the current texture env/combine mode. - * This function generates that program and puts it into effect. - * - * XXX: remove this function. currently only called by some drivers, - * not by mesa core. We now handle this properly from inside mesa. - */ -void -_mesa_UpdateTexEnvProgram( GLcontext *ctx ) -{ - const struct gl_fragment_program *prev = ctx->FragmentProgram._Current; - - ASSERT(ctx->FragmentProgram._MaintainTexEnvProgram); - - /* If a conventional fragment program/shader isn't in effect... */ - if (!ctx->FragmentProgram._Enabled && - (!ctx->Shader.CurrentProgram || - !ctx->Shader.CurrentProgram->FragmentProgram) ) - { - struct gl_fragment_program *newProg; - - newProg = _mesa_get_fixed_func_fragment_program(ctx); - - _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._Current, newProg); - _mesa_reference_fragprog(ctx, &ctx->FragmentProgram._TexEnvProgram, newProg); - } - - /* Tell the driver about the change. Could define a new target for - * this? - */ - if (ctx->FragmentProgram._Current != prev && ctx->Driver.BindProgram) { - ctx->Driver.BindProgram(ctx, GL_FRAGMENT_PROGRAM_ARB, - (struct gl_program *) ctx->FragmentProgram._Current); - } -}