<H1>Mesa 6.5.3 Release Notes / (in progress)</H1>
<p>
-Mesa 6.5.3 is a 6.5 follow-on development release mostly consisting of
-bug fixes</a>.
+Mesa 6.5.3 is a 6.5 follow-on development release with many internal changes.
</p>
<h2>New features</h2>
<ul>
+<li>OpenGL 2.0 support.
+<li>Entirely new Shading Language code generator.
+<li>Much faster software execution of vertex, fragment shaders.
+<li>New vertex buffer object infrastructure (replaces old array_cache code).
<li>Updated glext.h file (version 39)
<li>Updated glxext.h file (version 18)
++<li>GL_MAX_DRAWBUFFERS is now 4 (software rendering) so
++ "multiple render targets" are really supported.
</ul>
<h2>Bug fixes</h2>
<li>Fog was errantly applied when a fragment shader was enabled (bug 9346)
<li>glPush/PopClientAttrib didn't handle VBO bindings correctly (bug 9445)
<li>With 32-bit Z buffer, the fragment Z of lines and points was sometimes wrong.
- <li>GL_MAX_DRAWBUFFERS is now 4 (software rendering) so
- "multiple render targets" are really supported.
+<li>GL_POST_CONVOLUTION_ALPHA_BIAS/SCALE was broken.
+ <li>1D convolution state could effect 2D image transfers
</ul>
<h2>Internal code changes</h2>
<ul>
+<li>Massive changes to the Shading Language compiler.
<li>The _MaintainTnlProgram, _MaintainTexEnvProgram, _TexEnvProgram and
_TnlProgram fields have been moved.
<li>The ctx->FragmentProgram._Active field has been removed.
<h2>To Do (someday) items</h2>
<ul>
<li>Switch to freeglut
-<li>Increase MAX_DRAWBUFFERS
<li>Fix linux-glide target/driver.
<li>Improved lambda and derivative calculation for frag progs.
</ul>
Driver Status
---------------------- ----------------------
DRI drivers varies with the driver
-XMesa/GLX (on Xlib) implements OpenGL 1.5
-OSMesa (off-screen) implements OpenGL 1.5
+XMesa/GLX (on Xlib) implements OpenGL 2.0
+OSMesa (off-screen) implements OpenGL 2.0
Glide (3dfx Voodoo1/2) implements OpenGL 1.3
SVGA implements OpenGL 1.3
Wind River UGL implements OpenGL 1.3
GLboolean enabled;
GLboolean try_pixel_fog;
- if (ctx->FragmentProgram._Active) {
+ if (ctx->FragmentProgram._Enabled) {
/* Pull in static fog state from program */
mode = ctx->FragmentProgram._Current->FogOption;
I1_LOAD_S(4) |
I1_LOAD_S(5) |
I1_LOAD_S(6) |
- (4));
+ (3));
i915->state.Ctx[I915_CTXREG_LIS2] = 0;
i915->state.Ctx[I915_CTXREG_LIS4] = 0;
i915->state.Ctx[I915_CTXREG_LIS5] = 0;
GLcontext *ctx = &intel->ctx;
i915ContextPtr i915 = I915_CONTEXT(intel);
- if (ctx->FragmentProgram._Active)
+ if (ctx->FragmentProgram._Current)
i915ValidateFragmentProgram( i915 );
else {
- assert(!ctx->_MaintainTexEnvProgram);
+ assert(!ctx->FragmentProgram._MaintainTexEnvProgram);
i915ValidateTextureProgram( i915 );
}
}
*/
OUT_BATCH(_3DSTATE_LOAD_STATE_IMMEDIATE_1 |
I1_LOAD_S(3) |
- (1));
+ (0));
OUT_BATCH(0);
/* XXX: Use this */
GLboolean enabled;
GLboolean try_pixel_fog;
- if (ctx->FragmentProgram._Active) {
+ if (ctx->FragmentProgram._Enabled) {
/* Pull in static fog state from program */
mode = ctx->FragmentProgram._Current->FogOption;
i915->state.Ctx[I915_CTXREG_LI] = (_3DSTATE_LOAD_STATE_IMMEDIATE_1 |
I1_LOAD_S(2) |
I1_LOAD_S(4) |
- I1_LOAD_S(5) | I1_LOAD_S(6) | (4));
+ I1_LOAD_S(5) | I1_LOAD_S(6) | (3));
i915->state.Ctx[I915_CTXREG_LIS2] = 0;
i915->state.Ctx[I915_CTXREG_LIS4] = 0;
i915->state.Ctx[I915_CTXREG_LIS5] = 0;
#include "glheader.h"
#include "macros.h"
#include "enums.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_print.h"
#include "brw_vs.h"
#include "brw_state.h"
-#include "shader/program.h"
-#include "shader/program_instruction.h"
-#include "shader/arbprogparse.h"
struct state_key {
unsigned light_global_enabled:1;
{
GLfloat values[4];
GLint idx;
+ GLuint swizzle;
values[0] = s0;
values[1] = s1;
values[2] = s2;
values[3] = s3;
- idx = _mesa_add_unnamed_constant( p->program->Base.Parameters, values, 4 );
+ idx = _mesa_add_unnamed_constant( p->program->Base.Parameters, values, 4,
+ &swizzle);
+ /* XXX what about swizzle? */
return make_ureg(PROGRAM_STATE_VAR, idx);
}
return p->identity;
}
-static struct ureg register_param6( struct tnl_program *p,
- GLint s0,
- GLint s1,
- GLint s2,
- GLint s3,
- GLint s4,
- GLint s5)
+static struct ureg register_param5( struct tnl_program *p,
+ GLint s0,
+ GLint s1,
+ GLint s2,
+ GLint s3,
+ GLint s4)
{
- GLint tokens[6];
+ gl_state_index tokens[STATE_LENGTH];
GLint idx;
tokens[0] = s0;
tokens[1] = s1;
tokens[2] = s2;
tokens[3] = s3;
tokens[4] = s4;
- tokens[5] = s5;
idx = _mesa_add_state_reference( p->program->Base.Parameters, tokens );
return make_ureg(PROGRAM_STATE_VAR, idx);
}
-#define register_param1(p,s0) register_param6(p,s0,0,0,0,0,0)
-#define register_param2(p,s0,s1) register_param6(p,s0,s1,0,0,0,0)
-#define register_param3(p,s0,s1,s2) register_param6(p,s0,s1,s2,0,0,0)
-#define register_param4(p,s0,s1,s2,s3) register_param6(p,s0,s1,s2,s3,0,0)
+#define register_param1(p,s0) register_param5(p,s0,0,0,0,0)
+#define register_param2(p,s0,s1) register_param5(p,s0,s1,0,0,0)
+#define register_param3(p,s0,s1,s2) register_param5(p,s0,s1,s2,0,0)
+#define register_param4(p,s0,s1,s2,s3) register_param5(p,s0,s1,s2,s3,0)
-static void register_matrix_param6( struct tnl_program *p,
- GLint s0,
- GLint s1,
- GLint s2,
- GLint s3,
- GLint s4,
- GLint s5,
+static void register_matrix_param5( struct tnl_program *p,
+ GLint s0, /* matrix name */
+ GLint s1, /* texture matrix number */
+ GLint s2, /* first row */
+ GLint s3, /* last row */
+ GLint s4, /* modifier */
struct ureg *matrix )
{
GLint i;
/* This is a bit sad as the support is there to pull the whole
* matrix out in one go:
*/
- for (i = 0; i <= s4 - s3; i++)
- matrix[i] = register_param6( p, s0, s1, s2, i, i, s5 );
+ for (i = 0; i <= s3 - s2; i++)
+ matrix[i] = register_param5( p, s0, s1, i, i, s4 );
}
p->eye_position = reserve_temp(p);
if (PREFER_DP4) {
- register_matrix_param6( p, STATE_MATRIX, STATE_MODELVIEW, 0, 0, 3,
- STATE_MATRIX, modelview );
+ register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3,
+ 0, modelview );
emit_matrix_transform_vec4(p, p->eye_position, modelview, pos);
}
else {
- register_matrix_param6( p, STATE_MATRIX, STATE_MODELVIEW, 0, 0, 3,
+ register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3,
STATE_MATRIX_TRANSPOSE, modelview );
emit_transpose_matrix_transform_vec4(p, p->eye_position, modelview, pos);
struct ureg normal = register_input(p, VERT_ATTRIB_NORMAL );
struct ureg mvinv[3];
- register_matrix_param6( p, STATE_MATRIX, STATE_MODELVIEW, 0, 0, 2,
+ register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 2,
STATE_MATRIX_INVTRANS, mvinv );
p->eye_normal = reserve_temp(p);
struct ureg mvp[4];
if (PREFER_DP4) {
- register_matrix_param6( p, STATE_MATRIX, STATE_MVP, 0, 0, 3,
- STATE_MATRIX, mvp );
+ register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3,
+ 0, mvp );
emit_matrix_transform_vec4( p, hpos, mvp, pos );
}
else {
- register_matrix_param6( p, STATE_MATRIX, STATE_MVP, 0, 0, 3,
+ register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3,
STATE_MATRIX_TRANSPOSE, mvp );
emit_transpose_matrix_transform_vec4( p, hpos, mvp, pos );
}
/* Calculate spot attenuation:
*/
if (!p->state->unit[i].light_spotcutoff_is_180) {
- struct ureg spot_dir = register_param3(p, STATE_LIGHT, i,
- STATE_SPOT_DIRECTION);
+ struct ureg spot_dir_norm = register_param3(p, STATE_INTERNAL,
+ STATE_SPOT_DIR_NORMALIZED, i);
struct ureg spot = get_temp(p);
struct ureg slt = get_temp(p);
-
- emit_normalize_vec3( p, spot, spot_dir ); /* XXX: precompute! */
- emit_op2(p, OPCODE_DP3, spot, 0, ureg_negate(VPpli), spot);
- emit_op2(p, OPCODE_SLT, slt, 0, swizzle1(spot_dir,W), spot);
+
+ emit_op2(p, OPCODE_DP3, spot, 0, ureg_negate(VPpli), spot_dir_norm);
+ emit_op2(p, OPCODE_SLT, slt, 0, swizzle1(spot_dir_norm,W), spot);
emit_op2(p, OPCODE_POW, spot, 0, spot, swizzle1(attenuation, W));
emit_op2(p, OPCODE_MUL, att, 0, slt, spot);
/* Need to add some addtional parameters to allow lighting in object
-- * space - STATE_SPOT_DIRECTION and STATE_HALF implicitly assume eye
++ * space - STATE_SPOT_DIRECTION and STATE_HALF_VECTOR implicitly assume eye
* space lighting.
*/
static void build_lighting( struct tnl_program *p )
*/
VPpli = register_param3(p, STATE_LIGHT, i,
STATE_POSITION_NORMALIZED);
- half = register_param3(p, STATE_LIGHT, i, STATE_HALF_VECTOR);
+ if (p->state->light_local_viewer) {
+ struct ureg eye_hat = get_eye_position_normalized(p);
+ half = get_temp(p);
+ emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
+ emit_normalize_vec3(p, half, half);
+ } else {
- half = register_param3(p, STATE_LIGHT, i, STATE_HALF);
++ half = register_param3(p, STATE_LIGHT, i, STATE_HALF_VECTOR);
+ }
}
else {
struct ureg Ppli = register_param3(p, STATE_LIGHT, i,
if (p->state->fog_option &&
p->state->tnl_do_vertex_fog) {
- struct ureg params = register_param1(p, STATE_FOG_PARAMS);
+ struct ureg params = register_param2(p, STATE_INTERNAL,
+ STATE_FOG_PARAMS_OPTIMIZED);
struct ureg tmp = get_temp(p);
struct ureg id = get_identity_param(p);
switch (p->state->fog_option) {
case FOG_LINEAR: {
- emit_op1(p, OPCODE_ABS, tmp, 0, input);
- emit_op2(p, OPCODE_SUB, tmp, 0, swizzle1(params,Z), tmp);
- emit_op2(p, OPCODE_MUL, tmp, 0, tmp, swizzle1(params,W));
+ emit_op1(p, OPCODE_ABS, tmp, 0, input);
+ emit_op3(p, OPCODE_MAD, tmp, 0, tmp, swizzle1(params,X), swizzle1(params,Y));
emit_op2(p, OPCODE_MAX, tmp, 0, tmp, swizzle1(id,X)); /* saturate */
emit_op2(p, OPCODE_MIN, fog, WRITEMASK_X, tmp, swizzle1(id,W));
break;
}
case FOG_EXP:
emit_op1(p, OPCODE_ABS, tmp, 0, input);
- emit_op2(p, OPCODE_MUL, tmp, 0, tmp, swizzle1(params,X));
- emit_op2(p, OPCODE_POW, fog, WRITEMASK_X,
- register_const1f(p, M_E), ureg_negate(tmp));
+ emit_op2(p, OPCODE_MUL, tmp, 0, tmp, swizzle1(params,Z));
+ emit_op1(p, OPCODE_EX2, fog, WRITEMASK_X, ureg_negate(tmp));
break;
case FOG_EXP2:
- emit_op2(p, OPCODE_MUL, tmp, 0, input, swizzle1(params,X));
+ emit_op2(p, OPCODE_MUL, tmp, 0, input, swizzle1(params,W));
emit_op2(p, OPCODE_MUL, tmp, 0, tmp, tmp);
- emit_op2(p, OPCODE_POW, fog, WRITEMASK_X,
- register_const1f(p, M_E), ureg_negate(tmp));
+ emit_op1(p, OPCODE_EX2, fog, WRITEMASK_X, ureg_negate(tmp));
break;
}
out_texgen :
register_input(p, VERT_ATTRIB_TEX0+i));
if (PREFER_DP4) {
- register_matrix_param6( p, STATE_MATRIX, STATE_TEXTURE, i,
- 0, 3, STATE_MATRIX, texmat );
+ register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
+ 0, texmat );
emit_matrix_transform_vec4( p, out, texmat, in );
}
else {
- register_matrix_param6( p, STATE_MATRIX, STATE_TEXTURE, i,
- 0, 3, STATE_MATRIX_TRANSPOSE, texmat );
+ register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
+ STATE_MATRIX_TRANSPOSE, texmat );
emit_transpose_matrix_transform_vec4( p, out, texmat, in );
}
}
nouveauShaderInitFuncs(ctx);
/* Install Mesa's fixed-function texenv shader support */
if (nmesa->screen->card->type >= NV_40)
- ctx->_MaintainTexEnvProgram = GL_TRUE;
+ ctx->FragmentProgram._MaintainTexEnvProgram = GL_TRUE;
/* Initialize the swrast */
_swrast_CreateContext( ctx );
OUT_RING (((box->y2 - box->y1) << 16) |
(box->x2 - box->x1));
}
+ FIRE_RING();
UNLOCK_HARDWARE(nmesa);
#endif
#include "enums.h"
#include "extensions.h"
-#include "program.h"
+#include "shader/program.h"
+#include "shader/prog_instruction.h"
+/*#include "shader/arbprogparse.h"*/
#include "tnl/tnl.h"
-#include "shader/arbprogparse.h"
#include "nouveau_context.h"
#include "nouveau_shader.h"
static void
nouveauBindProgram(GLcontext *ctx, GLenum target, struct gl_program *prog)
{
+ NVSDBG("target=%s, prog=%p\n", _mesa_lookup_enum_by_nr(target), prog);
}
static struct gl_program *
{
nouveauShader *nvs;
+ NVSDBG("target=%s, id=%d\n", _mesa_lookup_enum_by_nr(target), id);
+
nvs = CALLOC_STRUCT(_nouveauShader);
+ NVSDBG("prog=%p\n", nvs);
switch (target) {
case GL_VERTEX_PROGRAM_ARB:
return _mesa_init_vertex_program(ctx, &nvs->mesa.vp, target, id);
{
nouveauShader *nvs = (nouveauShader *)prog;
+ NVSDBG("prog=%p\n", prog);
+
if (nvs->translated)
FREE(nvs->program);
_mesa_delete_program(ctx, prog);
{
nouveauShader *nvs = (nouveauShader *)prog;
+ NVSDBG("target=%s, prog=%p\n", _mesa_lookup_enum_by_nr(target), prog);
+
if (nvs->translated)
FREE(nvs->program);
- nvs->translated = 0;
+
+ nvs->error = GL_FALSE;
+ nvs->translated = GL_FALSE;
_tnl_program_string(ctx, target, prog);
}
{
nouveauShader *nvs = (nouveauShader *)prog;
+ NVSDBG("target=%s, prog=%p\n", _mesa_lookup_enum_by_nr(target), prog);
+
return nvs->translated;
}
struct gl_program_parameter_list *plist;
int i;
+ NVSDBG("prog=%p\n", nvs);
+
/* Translate to HW format now if necessary */
if (!nvs->translated) {
/* Mesa ASM shader -> nouveauShader */
#include "macros.h"
#include "enums.h"
-#include "program.h"
-#include "programopt.h"
-#include "program_instruction.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_statevars.h"
+#include "shader/programopt.h"
#include "nouveau_context.h"
#include "nouveau_shader.h"
return mask;
}
+ static GLboolean
+ pass0_opcode_is_tex(enum prog_opcode op)
+ {
+ switch (op) {
+ case OPCODE_TEX:
+ case OPCODE_TXB:
+ case OPCODE_TXD:
+ case OPCODE_TXL:
+ case OPCODE_TXP:
+ return GL_TRUE;
+ default:
+ break;
+ }
+
+ return GL_FALSE;
+ }
+
static nvsTexTarget
pass0_make_tex_target(GLuint mesa)
{
if (!rec->swzconst_done) {
struct gl_program *prog = &nvs->mesa.vp.Base;
+ GLuint swizzle;
rec->swzconst_id = _mesa_add_unnamed_constant(prog->Parameters,
- sc, 4);
+ sc, 4, &swizzle);
+ /* XXX what about swizzle? */
rec->swzconst_done = 1;
COPY_4V(nvs->params[rec->swzconst_id].val, sc);
}
(inst->SaturateMode != SATURATE_OFF),
src[0], src[1], src[2]);
nvsinst->tex_unit = inst->TexSrcUnit;
- nvsinst->tex_target = pass0_make_tex_target(inst->TexSrcTarget);
+ if (pass0_opcode_is_tex(inst->Opcode))
+ nvsinst->tex_target =
+ pass0_make_tex_target(inst->TexSrcTarget);
+ else
+ nvsinst->tex_target = NVS_TEX_TARGET_UNKNOWN;
ret = GL_TRUE;
} else
nvsInstruction *nvsinst;
GLuint fpos = 0;
nvsRegister opos, epos, eqn, mv[4];
- GLint tokens[6] = { STATE_MATRIX, STATE_MODELVIEW, 0, 0, 0, 0 };
+ gl_state_index tokens[STATE_LENGTH]
+ = { STATE_MODELVIEW_MATRIX, 0, 0, 0, 0 };
GLint id;
int i;
pass0_make_reg(nvs, &opos, NVS_FILE_ATTRIB, NVS_FR_POSITION);
pass0_make_reg(nvs, &epos, NVS_FILE_TEMP , -1);
for (i=0; i<4; i++) {
- tokens[3] = tokens[4] = i;
+ tokens[2] = tokens[3] = i;
id = _mesa_add_state_reference(prog->Parameters, tokens);
pass0_make_reg(nvs, &mv[i], NVS_FILE_CONST, id);
}
}
}
- static void
+ static GLboolean
pass0_resolve_mesa_consts(nouveauShader *nvs)
{
struct pass0_rec *rec = nvs->pass_rec;
for (i=0; i<plist->NumParameters; i++) {
int hw = rec->mesa_const_base + i;
+ if (hw > NVS_MAX_CONSTS) {
+ nvsProgramError(nvs, "hw = %d > NVS_MAX_CONSTS!\n", hw);
+ return GL_FALSE;
+ }
+
switch (plist->Parameters[i].Type) {
case PROGRAM_NAMED_PARAM:
case PROGRAM_STATE_VAR:
COPY_4V(nvs->params[hw].val, plist->ParameterValues[i]);
break;
default:
- assert(0);
- break;
+ nvsProgramError(nvs, "hit bad type=%d on param %d\n",
+ plist->Parameters[i].Type, i);
+ return GL_FALSE;
}
}
+
+ return GL_TRUE;
}
GLboolean
struct pass0_rec *rec;
int ret = GL_FALSE;
+ NVSDBG("start: nvs=%p\n", nvs);
+
+ /* Previously detected an error, and haven't recieved new program
+ * string, so fail immediately.
+ */
+ if (nvs->error) {
+ NVSDBG("failed previous compile attempt, not retrying\n");
+ return GL_FALSE;
+ }
+
rec = CALLOC_STRUCT(pass0_rec);
if (!rec)
return GL_FALSE;
ret = pass0_translate_instructions(nvs, 0, 0, nvs->program_tree);
if (ret)
- pass0_resolve_mesa_consts(nvs);
+ ret = pass0_resolve_mesa_consts(nvs);
+
/*XXX: if (!ret) DESTROY TREE!!! */
FREE(rec);
#include "macros.h"
#include "enums.h"
-#include "program.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_print.h"
#include "nouveau_context.h"
#include "nouveau_shader.h"
struct pass2_rec *rec;
int i;
+ NVSDBG("start: nvs=%p\n", nvs);
+
rec = calloc(1, sizeof(struct pass2_rec));
for (i=0; i<NVS_MAX_TEMPS; i++)
rec->temps[i] = -1;
#include "glheader.h"
#include "macros.h"
#include "enums.h"
+#include "shader/prog_instruction.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_print.h"
-#include "program.h"
-#include "program_instruction.h"
#include "r300_context.h"
#include "r300_fragprog.h"
#include "r300_reg.h"
+ #include "r300_state.h"
/*
* Usefull macros and values
#define REG_NEGV_SHIFT 18
#define REG_NEGS_SHIFT 19
#define REG_ABS_SHIFT 20
- #define REG_NO_USE_SHIFT 21
- #define REG_VALID_SHIFT 22
+ #define REG_NO_USE_SHIFT 21 // Hack for refcounting
+ #define REG_VALID_SHIFT 22 // Does the register contain a defined value?
+ #define REG_BUILTIN_SHIFT 23 // Is it a builtin (like all zero/all one)?
#define REG_TYPE_MASK (0x03 << REG_TYPE_SHIFT)
#define REG_INDEX_MASK (0x3F << REG_INDEX_SHIFT)
#define REG_ABS_MASK (0x01 << REG_ABS_SHIFT)
#define REG_NO_USE_MASK (0x01 << REG_NO_USE_SHIFT)
#define REG_VALID_MASK (0x01 << REG_VALID_SHIFT)
+ #define REG_BUILTIN_MASK (0x01 << REG_BUILTIN_SHIFT)
- #define REG(type, index, vswz, sswz, nouse, valid) \
+ #define REG(type, index, vswz, sswz, nouse, valid, builtin) \
(((type << REG_TYPE_SHIFT) & REG_TYPE_MASK) | \
((index << REG_INDEX_SHIFT) & REG_INDEX_MASK) | \
((nouse << REG_NO_USE_SHIFT) & REG_NO_USE_MASK) | \
((valid << REG_VALID_SHIFT) & REG_VALID_MASK) | \
+ ((builtin << REG_BUILTIN_SHIFT) & REG_BUILTIN_MASK) | \
((vswz << REG_VSWZ_SHIFT) & REG_VSWZ_MASK) | \
((sswz << REG_SSWZ_SHIFT) & REG_SSWZ_MASK))
#define REG_GET_TYPE(reg) \
((reg & REG_NO_USE_MASK) >> REG_NO_USE_SHIFT)
#define REG_GET_VALID(reg) \
((reg & REG_VALID_MASK) >> REG_VALID_SHIFT)
+ #define REG_GET_BUILTIN(reg) \
+ ((reg & REG_BUILTIN_MASK) >> REG_BUILTIN_SHIFT)
#define REG_SET_TYPE(reg, type) \
reg = ((reg & ~REG_TYPE_MASK) | \
((type << REG_TYPE_SHIFT) & REG_TYPE_MASK))
#define REG_SET_VALID(reg, valid) \
reg = ((reg & ~REG_VALID_MASK) | \
((valid << REG_VALID_SHIFT) & REG_VALID_MASK))
+ #define REG_SET_BUILTIN(reg, builtin) \
+ reg = ((reg & ~REG_BUILTIN_MASK) | \
+ ((builtin << REG_BUILTIN_SHIFT) & REG_BUILTIN_MASK))
#define REG_ABS(reg) \
reg = (reg | REG_ABS_MASK)
#define REG_NEGV(reg) \
*
* REG_VSWZ/REG_SSWZ is an index into this table
*/
- #define SLOT_VECTOR (1<<0)
- #define SLOT_SCALAR (1<<3)
- #define SLOT_BOTH (SLOT_VECTOR | SLOT_SCALAR)
/* mapping from SWIZZLE_* to r300 native values for scalar insns */
#define SWIZZLE_HALF 6
GLuint flags;
} v_swiz[] = {
/* native swizzles */
- { MAKE_SWZ3(X, Y, Z), R300_FPI0_ARGC_SRC0C_XYZ, 4, SLOT_VECTOR },
- { MAKE_SWZ3(X, X, X), R300_FPI0_ARGC_SRC0C_XXX, 4, SLOT_VECTOR },
- { MAKE_SWZ3(Y, Y, Y), R300_FPI0_ARGC_SRC0C_YYY, 4, SLOT_VECTOR },
- { MAKE_SWZ3(Z, Z, Z), R300_FPI0_ARGC_SRC0C_ZZZ, 4, SLOT_VECTOR },
- { MAKE_SWZ3(W, W, W), R300_FPI0_ARGC_SRC0A, 1, SLOT_SCALAR },
- { MAKE_SWZ3(Y, Z, X), R300_FPI0_ARGC_SRC0C_YZX, 1, SLOT_VECTOR },
- { MAKE_SWZ3(Z, X, Y), R300_FPI0_ARGC_SRC0C_ZXY, 1, SLOT_VECTOR },
- { MAKE_SWZ3(W, Z, Y), R300_FPI0_ARGC_SRC0CA_WZY, 1, SLOT_BOTH },
+ { MAKE_SWZ3(X, Y, Z), R300_FPI0_ARGC_SRC0C_XYZ, 4, SLOT_SRC_VECTOR },
+ { MAKE_SWZ3(X, X, X), R300_FPI0_ARGC_SRC0C_XXX, 4, SLOT_SRC_VECTOR },
+ { MAKE_SWZ3(Y, Y, Y), R300_FPI0_ARGC_SRC0C_YYY, 4, SLOT_SRC_VECTOR },
+ { MAKE_SWZ3(Z, Z, Z), R300_FPI0_ARGC_SRC0C_ZZZ, 4, SLOT_SRC_VECTOR },
+ { MAKE_SWZ3(W, W, W), R300_FPI0_ARGC_SRC0A, 1, SLOT_SRC_SCALAR },
+ { MAKE_SWZ3(Y, Z, X), R300_FPI0_ARGC_SRC0C_YZX, 1, SLOT_SRC_VECTOR },
+ { MAKE_SWZ3(Z, X, Y), R300_FPI0_ARGC_SRC0C_ZXY, 1, SLOT_SRC_VECTOR },
+ { MAKE_SWZ3(W, Z, Y), R300_FPI0_ARGC_SRC0CA_WZY, 1, SLOT_SRC_BOTH },
{ MAKE_SWZ3(ONE, ONE, ONE), R300_FPI0_ARGC_ONE, 0, 0},
{ MAKE_SWZ3(ZERO, ZERO, ZERO), R300_FPI0_ARGC_ZERO, 0, 0},
{ MAKE_SWZ3(HALF, HALF, HALF), R300_FPI0_ARGC_HALF, 0, 0},
int stride; /* difference between SRC0/1/2 */
GLuint flags;
} s_swiz[] = {
- { R300_FPI2_ARGA_SRC0C_X, 3, SLOT_VECTOR },
- { R300_FPI2_ARGA_SRC0C_Y, 3, SLOT_VECTOR },
- { R300_FPI2_ARGA_SRC0C_Z, 3, SLOT_VECTOR },
- { R300_FPI2_ARGA_SRC0A , 1, SLOT_SCALAR },
+ { R300_FPI2_ARGA_SRC0C_X, 3, SLOT_SRC_VECTOR },
+ { R300_FPI2_ARGA_SRC0C_Y, 3, SLOT_SRC_VECTOR },
+ { R300_FPI2_ARGA_SRC0C_Z, 3, SLOT_SRC_VECTOR },
+ { R300_FPI2_ARGA_SRC0A , 1, SLOT_SRC_SCALAR },
{ R300_FPI2_ARGA_ZERO , 0, 0 },
{ R300_FPI2_ARGA_ONE , 0, 0 },
{ R300_FPI2_ARGA_HALF , 0, 0 }
SWIZZLE_XYZ,
SWIZZLE_W,
GL_FALSE,
+ GL_FALSE,
GL_FALSE);
/* constant one source */
SWIZZLE_111,
SWIZZLE_ONE,
GL_FALSE,
+ GL_TRUE,
GL_TRUE);
/* constant half source */
SWIZZLE_HHH,
SWIZZLE_HALF,
GL_FALSE,
+ GL_TRUE,
GL_TRUE);
/* constant zero source */
SWIZZLE_000,
SWIZZLE_ZERO,
GL_FALSE,
+ GL_TRUE,
GL_TRUE);
/*
GLuint src0, GLuint src1, GLuint src2,
int flags);
- /*
- * Helper functions prototypes
+ /**
+ * Get an R300 temporary that can be written to in the given slot.
*/
- static int get_hw_temp(struct r300_fragment_program *rp)
+ static int get_hw_temp(struct r300_fragment_program *rp, int slot)
{
COMPILE_STATE;
- int r = ffs(~cs->hwreg_in_use);
- if (!r) {
+ int r;
+
+ for(r = 0; r < PFS_NUM_TEMP_REGS; ++r) {
+ if (cs->hwtemps[r].free >= 0 && cs->hwtemps[r].free <= slot)
+ break;
+ }
+
+ if (r >= PFS_NUM_TEMP_REGS) {
ERROR("Out of hardware temps\n");
return 0;
}
- cs->hwreg_in_use |= (1 << --r);
+ // Reserved is used to avoid the following scenario:
+ // R300 temporary X is first assigned to Mesa temporary Y during vector ops
+ // R300 temporary X is then assigned to Mesa temporary Z for further vector ops
+ // Then scalar ops on Mesa temporary Z are emitted and move back in time
+ // to overwrite the value of temporary Y.
+ // End scenario.
+ cs->hwtemps[r].reserved = cs->hwtemps[r].free;
+ cs->hwtemps[r].free = -1;
+
+ // Reset to some value that won't mess things up when the user
+ // tries to read from a temporary that hasn't been assigned a value yet.
+ // In the normal case, vector_valid and scalar_valid should be set to
+ // a sane value by the first emit that writes to this temporary.
+ cs->hwtemps[r].vector_valid = 0;
+ cs->hwtemps[r].scalar_valid = 0;
+
if (r > rp->max_temp_idx)
rp->max_temp_idx = r;
return r;
}
+ /**
+ * Get an R300 temporary that will act as a TEX destination register.
+ */
static int get_hw_temp_tex(struct r300_fragment_program *rp)
{
COMPILE_STATE;
int r;
- r = ffs(~(cs->hwreg_in_use | cs->used_in_node));
- if (!r)
- return get_hw_temp(rp); /* Will cause an indirection */
+ for(r = 0; r < PFS_NUM_TEMP_REGS; ++r) {
+ if (cs->used_in_node & (1 << r))
+ continue;
+
+ // Note: Be very careful here
+ if (cs->hwtemps[r].free >= 0 && cs->hwtemps[r].free <= 0)
+ break;
+ }
+
+ if (r >= PFS_NUM_TEMP_REGS)
+ return get_hw_temp(rp, 0); /* Will cause an indirection */
+
+ cs->hwtemps[r].reserved = cs->hwtemps[r].free;
+ cs->hwtemps[r].free = -1;
+
+ // Reset to some value that won't mess things up when the user
+ // tries to read from a temporary that hasn't been assigned a value yet.
+ // In the normal case, vector_valid and scalar_valid should be set to
+ // a sane value by the first emit that writes to this temporary.
+ cs->hwtemps[r].vector_valid = cs->nrslots;
+ cs->hwtemps[r].scalar_valid = cs->nrslots;
- cs->hwreg_in_use |= (1 << --r);
if (r > rp->max_temp_idx)
rp->max_temp_idx = r;
return r;
}
+ /**
+ * Mark the given hardware register as free.
+ */
static void free_hw_temp(struct r300_fragment_program *rp, int idx)
{
COMPILE_STATE;
- cs->hwreg_in_use &= ~(1<<idx);
+
+ // Be very careful here. Consider sequences like
+ // MAD r0, r1,r2,r3
+ // TEX r4, ...
+ // The TEX instruction may be moved in front of the MAD instruction
+ // due to the way nodes work. We don't want to alias r1 and r4 in
+ // this case.
+ // I'm certain the register allocation could be further sanitized,
+ // but it's tricky because of stuff that can happen inside emit_tex
+ // and emit_arith.
+ cs->hwtemps[idx].free = cs->nrslots+1;
}
+
+ /**
+ * Create a new Mesa temporary register.
+ */
static GLuint get_temp_reg(struct r300_fragment_program *rp)
{
COMPILE_STATE;
return r;
}
+ /**
+ * Create a new Mesa temporary register that will act as the destination
+ * register for a texture read.
+ */
static GLuint get_temp_reg_tex(struct r300_fragment_program *rp)
{
COMPILE_STATE;
return r;
}
+ /**
+ * Free a Mesa temporary and the associated R300 temporary.
+ */
static void free_temp(struct r300_fragment_program *rp, GLuint r)
{
COMPILE_STATE;
if (!(cs->temp_in_use & (1 << index)))
return;
-
+
if (REG_GET_TYPE(r) == REG_TYPE_TEMP) {
free_hw_temp(rp, cs->temps[index].reg);
cs->temps[index].reg = -1;
}
}
- static GLuint emit_param4fv(struct r300_fragment_program *rp,
- GLfloat *values)
+ /**
+ * Emit a hardware constant/parameter.
+ *
+ * \p cp Stable pointer to an array of 4 floats.
+ * The pointer must be stable in the sense that it remains to be valid
+ * and hold the contents of the constant/parameter throughout the lifetime
+ * of the fragment program (actually, up until the next time the fragment
+ * program is translated).
+ */
+ static GLuint emit_const4fv(struct r300_fragment_program *rp, const GLfloat* cp)
{
- GLuint r = undef;
- GLuint index;
- int pidx;
+ GLuint reg = undef;
+ int index;
- pidx = rp->param_nr++;
- index = rp->const_nr++;
- if (pidx >= PFS_NUM_CONST_REGS || index >= PFS_NUM_CONST_REGS) {
- ERROR("Out of const/param slots!\n");
- return r;
+ for(index = 0; index < rp->const_nr; ++index) {
+ if (rp->constant[index] == cp)
+ break;
}
- rp->param[pidx].idx = index;
- rp->param[pidx].values = values;
- rp->params_uptodate = GL_FALSE;
-
- REG_SET_TYPE(r, REG_TYPE_CONST);
- REG_SET_INDEX(r, index);
- REG_SET_VALID(r, GL_TRUE);
- return r;
- }
-
- static GLuint emit_const4fv(struct r300_fragment_program *rp, GLfloat *cp)
- {
- GLuint r = undef;
- GLuint index;
+ if (index >= rp->const_nr) {
+ if (index >= PFS_NUM_CONST_REGS) {
+ ERROR("Out of hw constants!\n");
+ return reg;
+ }
- index = rp->const_nr++;
- if (index >= PFS_NUM_CONST_REGS) {
- ERROR("Out of hw constants!\n");
- return r;
+ rp->const_nr++;
+ rp->constant[index] = cp;
}
- COPY_4V(rp->constant[index], cp);
-
- REG_SET_TYPE(r, REG_TYPE_CONST);
- REG_SET_INDEX(r, index);
- REG_SET_VALID(r, GL_TRUE);
- return r;
+ REG_SET_TYPE(reg, REG_TYPE_CONST);
+ REG_SET_INDEX(reg, index);
+ REG_SET_VALID(reg, GL_TRUE);
+ return reg;
}
static inline GLuint negate(GLuint r)
GLuint offset;
for(i=0; i < 4; ++i){
offset = GET_SWZ(arbswz, i);
-
+
newswz |= (offset <= 3)?GET_SWZ(vsrcswz, offset) << i*3:offset << i*3;
}
REG_SET_TYPE(r, REG_TYPE_INPUT);
break;
case PROGRAM_LOCAL_PARAM:
- r = emit_param4fv(rp,
+ r = emit_const4fv(rp,
rp->mesa_program.Base.LocalParams[fpsrc.Index]);
break;
case PROGRAM_ENV_PARAM:
- r = emit_param4fv(rp,
+ r = emit_const4fv(rp,
rp->ctx->FragmentProgram.Parameters[fpsrc.Index]);
break;
case PROGRAM_STATE_VAR:
case PROGRAM_NAMED_PARAM:
- r = emit_param4fv(rp,
+ r = emit_const4fv(rp,
rp->mesa_program.Base.Parameters->ParameterValues[fpsrc.Index]);
break;
default:
struct prog_dst_register dest)
{
GLuint r = undef;
-
+
switch (dest.File) {
case PROGRAM_TEMPORARY:
REG_SET_INDEX(r, dest.Index);
switch(REG_GET_TYPE(src)) {
case REG_TYPE_TEMP:
/* NOTE: if reg==-1 here, a source is being read that
- * hasn't been written to. Undefined results
+ * hasn't been written to. Undefined results.
*/
if (cs->temps[index].reg == -1)
- cs->temps[index].reg = get_hw_temp(rp);
+ cs->temps[index].reg = get_hw_temp(rp, cs->nrslots);
idx = cs->temps[index].reg;
static int t_hw_dst(struct r300_fragment_program *rp,
GLuint dest,
- GLboolean tex)
+ GLboolean tex,
+ int slot)
{
COMPILE_STATE;
int idx;
case REG_TYPE_TEMP:
if (cs->temps[REG_GET_INDEX(dest)].reg == -1) {
if (!tex) {
- cs->temps[index].reg = get_hw_temp(rp);
+ cs->temps[index].reg = get_hw_temp(rp, slot);
} else {
cs->temps[index].reg = get_hw_temp_tex(rp);
}
ERROR("invalid dest reg type %d\n", REG_GET_TYPE(dest));
return 0;
}
-
+
return idx;
}
- static void emit_nop(struct r300_fragment_program *rp,
- GLuint mask,
- GLboolean sync)
+ static void emit_nop(struct r300_fragment_program *rp)
{
COMPILE_STATE;
-
- if (sync)
- cs->v_pos = cs->s_pos = MAX2(cs->v_pos, cs->s_pos);
- if (mask & WRITEMASK_XYZ) {
- rp->alu.inst[cs->v_pos].inst0 = NOP_INST0;
- rp->alu.inst[cs->v_pos].inst1 = NOP_INST1;
- cs->v_pos++;
+ if (cs->nrslots >= PFS_MAX_ALU_INST) {
+ ERROR("Out of ALU instruction slots\n");
+ return;
}
- if (mask & WRITEMASK_W) {
- rp->alu.inst[cs->s_pos].inst2 = NOP_INST2;
- rp->alu.inst[cs->s_pos].inst3 = NOP_INST3;
- cs->s_pos++;
- }
+ rp->alu.inst[cs->nrslots].inst0 = NOP_INST0;
+ rp->alu.inst[cs->nrslots].inst1 = NOP_INST1;
+ rp->alu.inst[cs->nrslots].inst2 = NOP_INST2;
+ rp->alu.inst[cs->nrslots].inst3 = NOP_INST3;
+ cs->nrslots++;
}
static void emit_tex(struct r300_fragment_program *rp,
GLuint din = cs->dest_in_node, uin = cs->used_in_node;
int unit = fpi->TexSrcUnit;
int hwsrc, hwdest;
-
+
/* Resolve source/dest to hardware registers */
hwsrc = t_hw_src(rp, coord, GL_TRUE);
if (opcode != R300_FPITX_OP_KIL) {
rdest = dest;
dest = get_temp_reg_tex(rp);
}
- hwdest = t_hw_dst(rp, dest, GL_TRUE);
-
+ hwdest = t_hw_dst(rp, dest, GL_TRUE, rp->node[rp->cur_node].alu_offset);
+
/* Use a temp that hasn't been used in this node, rather
* than causing an indirection
*/
hwdest = 0;
unit = 0;
}
-
+
/* Indirection if source has been written in this node, or if the
* dest has been read/written in this node
*/
if ((REG_GET_TYPE(coord) != REG_TYPE_CONST &&
(din & (1<<hwsrc))) || (uin & (1<<hwdest))) {
-
+
/* Finish off current node */
- cs->v_pos = cs->s_pos = MAX2(cs->v_pos, cs->s_pos);
- if (rp->node[rp->cur_node].alu_offset == cs->v_pos) {
- /* No alu instructions in the node? Emit a NOP. */
- emit_nop(rp, WRITEMASK_XYZW, GL_TRUE);
- cs->v_pos = cs->s_pos = MAX2(cs->v_pos, cs->s_pos);
- }
-
+ if (rp->node[rp->cur_node].alu_offset == cs->nrslots)
+ emit_nop(rp);
+
rp->node[rp->cur_node].alu_end =
- cs->v_pos - rp->node[rp->cur_node].alu_offset - 1;
+ cs->nrslots - rp->node[rp->cur_node].alu_offset - 1;
assert(rp->node[rp->cur_node].alu_end >= 0);
if (++rp->cur_node >= PFS_MAX_TEX_INDIRECT) {
/* Start new node */
rp->node[rp->cur_node].tex_offset = rp->tex.length;
- rp->node[rp->cur_node].alu_offset = cs->v_pos;
+ rp->node[rp->cur_node].alu_offset = cs->nrslots;
rp->node[rp->cur_node].tex_end = -1;
- rp->node[rp->cur_node].alu_end = -1;
+ rp->node[rp->cur_node].alu_end = -1;
rp->node[rp->cur_node].flags = 0;
cs->used_in_node = 0;
cs->dest_in_node = 0;
}
-
+
if (rp->cur_node == 0)
rp->first_node_has_tex = 1;
/* not entirely sure about this */
| (opcode << R300_FPITX_OPCODE_SHIFT);
- cs->dest_in_node |= (1 << hwdest);
+ cs->dest_in_node |= (1 << hwdest);
if (REG_GET_TYPE(coord) != REG_TYPE_CONST)
cs->used_in_node |= (1 << hwsrc);
}
}
- /* Add sources to FPI1/FPI3 lists. If source is already on list,
- * reuse the index instead of wasting a source.
+
+ /**
+ * Returns the first slot where we could possibly allow writing to dest,
+ * according to register allocation.
*/
- static int add_src(struct r300_fragment_program *rp,
- int reg,
- int pos,
- int srcmask)
+ static int get_earliest_allowed_write(
+ struct r300_fragment_program* rp,
+ GLuint dest, int mask)
{
COMPILE_STATE;
- int csm, i;
-
- /* Look for matches */
- for (i=0,csm=srcmask; i<3; i++,csm=csm<<1) {
- /* If sources have been allocated in this position(s)... */
- if ((cs->slot[pos].umask & csm) == csm) {
- /* ... and the register number(s) match, re-use the
- source */
- if (srcmask == SLOT_VECTOR &&
- cs->slot[pos].vsrc[i] == reg)
- return i;
- if (srcmask == SLOT_SCALAR &&
- cs->slot[pos].ssrc[i] == reg)
- return i;
- if (srcmask == SLOT_BOTH &&
- cs->slot[pos].vsrc[i] == reg &&
- cs->slot[pos].ssrc[i] == reg)
- return i;
- }
+ int idx;
+ int pos;
+ GLuint index = REG_GET_INDEX(dest);
+ assert(REG_GET_VALID(dest));
+
+ switch(REG_GET_TYPE(dest)) {
+ case REG_TYPE_TEMP:
+ if (cs->temps[index].reg == -1)
+ return 0;
+
+ idx = cs->temps[index].reg;
+ break;
+ case REG_TYPE_OUTPUT:
+ return 0;
+ default:
+ ERROR("invalid dest reg type %d\n", REG_GET_TYPE(dest));
+ return 0;
}
- /* Look for free spaces */
- for (i=0,csm=srcmask; i<3; i++,csm=csm<<1) {
- /* If the position(s) haven't been allocated */
- if ((cs->slot[pos].umask & csm) == 0) {
- cs->slot[pos].umask |= csm;
-
- if (srcmask & SLOT_VECTOR)
- cs->slot[pos].vsrc[i] = reg;
- if (srcmask & SLOT_SCALAR)
- cs->slot[pos].ssrc[i] = reg;
- return i;
- }
+ pos = cs->hwtemps[idx].reserved;
+ if (mask & WRITEMASK_XYZ) {
+ if (pos < cs->hwtemps[idx].vector_lastread)
+ pos = cs->hwtemps[idx].vector_lastread;
}
-
- //ERROR("Failed to allocate sources in FPI1/FPI3!\n");
- return 0;
+ if (mask & WRITEMASK_W) {
+ if (pos < cs->hwtemps[idx].scalar_lastread)
+ pos = cs->hwtemps[idx].scalar_lastread;
+ }
+
+ return pos;
}
- /* Determine whether or not to position opcode in the same ALU slot for both
- * vector and scalar portions of an instruction.
+
+ /**
+ * Allocates a slot for an ALU instruction that can consist of
+ * a vertex part or a scalar part or both.
+ *
+ * Sources from src (src[0] to src[argc-1]) are added to the slot in the
+ * appropriate position (vector and/or scalar), and their positions are
+ * recorded in the srcpos array.
*
- * It's not necessary to force the first case, but it makes disassembled
- * shaders easier to read.
+ * This function emits instruction code for the source fetch and the
+ * argument selection. It does not emit instruction code for the
+ * opcode or the destination selection.
+ *
+ * @return the index of the slot
*/
- static GLboolean force_same_slot(int vop,
- int sop,
- GLboolean emit_vop,
- GLboolean emit_sop,
- int argc,
- GLuint *src)
+ static int find_and_prepare_slot(struct r300_fragment_program* rp,
+ GLboolean emit_vop,
+ GLboolean emit_sop,
+ int argc,
+ GLuint* src,
+ GLuint dest,
+ int mask)
{
- int i;
+ COMPILE_STATE;
+ int hwsrc[3];
+ int srcpos[3];
+ unsigned int used;
+ int tempused;
+ int tempvsrc[3];
+ int tempssrc[3];
+ int pos;
+ int regnr;
+ int i,j;
+
+ // Determine instruction slots, whether sources are required on
+ // vector or scalar side, and the smallest slot number where
+ // all source registers are available
+ used = 0;
+ if (emit_vop)
+ used |= SLOT_OP_VECTOR;
+ if (emit_sop)
+ used |= SLOT_OP_SCALAR;
+
+ pos = get_earliest_allowed_write(rp, dest, mask);
+
+ if (rp->node[rp->cur_node].alu_offset > pos)
+ pos = rp->node[rp->cur_node].alu_offset;
+ for(i = 0; i < argc; ++i) {
+ if (!REG_GET_BUILTIN(src[i])) {
+ if (emit_vop)
+ used |= v_swiz[REG_GET_VSWZ(src[i])].flags << i;
+ if (emit_sop)
+ used |= s_swiz[REG_GET_SSWZ(src[i])].flags << i;
+ }
+
+ hwsrc[i] = t_hw_src(rp, src[i], GL_FALSE); /* Note: sideeffects wrt refcounting! */
+ regnr = hwsrc[i] & 31;
+
+ if (REG_GET_TYPE(src[i]) == REG_TYPE_TEMP) {
+ if (used & (SLOT_SRC_VECTOR << i)) {
+ if (cs->hwtemps[regnr].vector_valid > pos)
+ pos = cs->hwtemps[regnr].vector_valid;
+ }
+ if (used & (SLOT_SRC_SCALAR << i)) {
+ if (cs->hwtemps[regnr].scalar_valid > pos)
+ pos = cs->hwtemps[regnr].scalar_valid;
+ }
+ }
+ }
+
+ // Find a slot that fits
+ for(; ; ++pos) {
+ if (cs->slot[pos].used & used & SLOT_OP_BOTH)
+ continue;
+
+ if (pos >= cs->nrslots) {
+ if (cs->nrslots >= PFS_MAX_ALU_INST) {
+ ERROR("Out of ALU instruction slots\n");
+ return -1;
+ }
- if (emit_vop && emit_sop)
- return GL_TRUE;
+ rp->alu.inst[pos].inst0 = NOP_INST0;
+ rp->alu.inst[pos].inst1 = NOP_INST1;
+ rp->alu.inst[pos].inst2 = NOP_INST2;
+ rp->alu.inst[pos].inst3 = NOP_INST3;
+
+ cs->nrslots++;
+ }
+
+ // Note: When we need both parts (vector and scalar) of a source,
+ // we always try to put them into the same position. This makes the
+ // code easier to read, and it is optimal (i.e. one doesn't gain
+ // anything by splitting the parts).
+ // It also avoids headaches with swizzles that access both parts (i.e WXY)
+ tempused = cs->slot[pos].used;
+ for(i = 0; i < 3; ++i) {
+ tempvsrc[i] = cs->slot[pos].vsrc[i];
+ tempssrc[i] = cs->slot[pos].ssrc[i];
+ }
- if (emit_vop && vop == R300_FPI0_OUTC_REPL_ALPHA)
- return GL_TRUE;
+ for(i = 0; i < argc; ++i) {
+ int flags = (used >> i) & SLOT_SRC_BOTH;
+ if (!flags) {
+ srcpos[i] = 0;
+ continue;
+ }
+
+ for(j = 0; j < 3; ++j) {
+ if ((tempused >> j) & flags & SLOT_SRC_VECTOR) {
+ if (tempvsrc[j] != hwsrc[i])
+ continue;
+ }
+
+ if ((tempused >> j) & flags & SLOT_SRC_SCALAR) {
+ if (tempssrc[j] != hwsrc[i])
+ continue;
+ }
+
+ break;
+ }
+
+ if (j == 3)
+ break;
+
+ srcpos[i] = j;
+ tempused |= flags << j;
+ if (flags & SLOT_SRC_VECTOR)
+ tempvsrc[j] = hwsrc[i];
+ if (flags & SLOT_SRC_SCALAR)
+ tempssrc[j] = hwsrc[i];
+ }
+
+ if (i == argc)
+ break;
+ }
+
+ // Found a slot, reserve it
+ cs->slot[pos].used = tempused | (used & SLOT_OP_BOTH);
+ for(i = 0; i < 3; ++i) {
+ cs->slot[pos].vsrc[i] = tempvsrc[i];
+ cs->slot[pos].ssrc[i] = tempssrc[i];
+ }
+
+ for(i = 0; i < argc; ++i) {
+ if (REG_GET_TYPE(src[i]) == REG_TYPE_TEMP) {
+ int regnr = hwsrc[i] & 31;
+
+ if (used & (SLOT_SRC_VECTOR << i)) {
+ if (cs->hwtemps[regnr].vector_lastread < pos)
+ cs->hwtemps[regnr].vector_lastread = pos;
+ }
+ if (used & (SLOT_SRC_SCALAR << i)) {
+ if (cs->hwtemps[regnr].scalar_lastread < pos)
+ cs->hwtemps[regnr].scalar_lastread = pos;
+ }
+ }
+ }
+
+ // Emit the source fetch code
+ rp->alu.inst[pos].inst1 &= ~R300_FPI1_SRC_MASK;
+ rp->alu.inst[pos].inst1 |=
+ ((cs->slot[pos].vsrc[0] << R300_FPI1_SRC0C_SHIFT) |
+ (cs->slot[pos].vsrc[1] << R300_FPI1_SRC1C_SHIFT) |
+ (cs->slot[pos].vsrc[2] << R300_FPI1_SRC2C_SHIFT));
+
+ rp->alu.inst[pos].inst3 &= ~R300_FPI3_SRC_MASK;
+ rp->alu.inst[pos].inst3 |=
+ ((cs->slot[pos].ssrc[0] << R300_FPI3_SRC0A_SHIFT) |
+ (cs->slot[pos].ssrc[1] << R300_FPI3_SRC1A_SHIFT) |
+ (cs->slot[pos].ssrc[2] << R300_FPI3_SRC2A_SHIFT));
+
+ // Emit the argument selection code
if (emit_vop) {
- for (i=0;i<argc;i++)
- if (REG_GET_VSWZ(src[i]) == SWIZZLE_WZY)
- return GL_TRUE;
+ int swz[3];
+
+ for(i = 0; i < 3; ++i) {
+ if (i < argc) {
+ swz[i] = (v_swiz[REG_GET_VSWZ(src[i])].base +
+ (srcpos[i] * v_swiz[REG_GET_VSWZ(src[i])].stride)) |
+ ((src[i] & REG_NEGV_MASK) ? ARG_NEG : 0) |
+ ((src[i] & REG_ABS_MASK) ? ARG_ABS : 0);
+ } else {
+ swz[i] = R300_FPI0_ARGC_ZERO;
+ }
+ }
+
+ rp->alu.inst[pos].inst0 &=
+ ~(R300_FPI0_ARG0C_MASK|R300_FPI0_ARG1C_MASK|R300_FPI0_ARG2C_MASK);
+ rp->alu.inst[pos].inst0 |=
+ (swz[0] << R300_FPI0_ARG0C_SHIFT) |
+ (swz[1] << R300_FPI0_ARG1C_SHIFT) |
+ (swz[2] << R300_FPI0_ARG2C_SHIFT);
+ }
+
+ if (emit_sop) {
+ int swz[3];
+
+ for(i = 0; i < 3; ++i) {
+ if (i < argc) {
+ swz[i] = (s_swiz[REG_GET_SSWZ(src[i])].base +
+ (srcpos[i] * s_swiz[REG_GET_SSWZ(src[i])].stride)) |
+ ((src[i] & REG_NEGV_MASK) ? ARG_NEG : 0) |
+ ((src[i] & REG_ABS_MASK) ? ARG_ABS : 0);
+ } else {
+ swz[i] = R300_FPI2_ARGA_ZERO;
+ }
+ }
+
+ rp->alu.inst[pos].inst2 &=
+ ~(R300_FPI2_ARG0A_MASK|R300_FPI2_ARG1A_MASK|R300_FPI2_ARG2A_MASK);
+ rp->alu.inst[pos].inst2 |=
+ (swz[0] << R300_FPI2_ARG0A_SHIFT) |
+ (swz[1] << R300_FPI2_ARG1A_SHIFT) |
+ (swz[2] << R300_FPI2_ARG2A_SHIFT);
}
- return GL_FALSE;
+ return pos;
}
+
+ /**
+ * Append an ALU instruction to the instruction list.
+ */
static void emit_arith(struct r300_fragment_program *rp,
int op,
GLuint dest,
{
COMPILE_STATE;
GLuint src[3] = { src0, src1, src2 };
- int hwsrc[3], sswz[3], vswz[3];
int hwdest;
- GLboolean emit_vop = GL_FALSE, emit_sop = GL_FALSE;
+ GLboolean emit_vop, emit_sop;
int vop, sop, argc;
- int vpos, spos;
- int i;
+ int pos;
vop = r300_fpop[op].v_op;
sop = r300_fpop[op].s_op;
argc = r300_fpop[op].argc;
+ if (REG_GET_TYPE(dest) == REG_TYPE_OUTPUT &&
+ REG_GET_INDEX(dest) == FRAG_RESULT_DEPR) {
+ if (mask & WRITEMASK_Z) {
+ mask = WRITEMASK_W;
+ } else {
+ return;
+ }
+ }
+
+ emit_vop = GL_FALSE;
+ emit_sop = GL_FALSE;
if ((mask & WRITEMASK_XYZ) || vop == R300_FPI0_OUTC_DP3)
emit_vop = GL_TRUE;
if ((mask & WRITEMASK_W) || vop == R300_FPI0_OUTC_REPL_ALPHA)
emit_sop = GL_TRUE;
- if (REG_GET_TYPE(dest) == REG_TYPE_OUTPUT &&
- REG_GET_INDEX(dest) == FRAG_RESULT_DEPR)
- emit_vop = GL_FALSE;
-
- if (force_same_slot(vop, sop, emit_vop, emit_sop, argc, src)) {
- vpos = spos = MAX2(cs->v_pos, cs->s_pos);
- } else {
- vpos = cs->v_pos;
- spos = cs->s_pos;
- /* Here is where we'd decide on where a safe place is to
- * combine this instruction with a previous one.
- *
- * This is extremely simple for now.. if a source depends
- * on the opposite stream, force the same instruction.
- */
- for (i=0;i<3;i++) {
- if (emit_vop &&
- (v_swiz[REG_GET_VSWZ(src[i])].flags & SLOT_SCALAR)) {
- vpos = spos = MAX2(vpos, spos);
- break;
- }
- if (emit_sop &&
- (s_swiz[REG_GET_SSWZ(src[i])].flags & SLOT_VECTOR)) {
- vpos = spos = MAX2(vpos, spos);
- break;
- }
- }
- }
-
- /* - Convert src->hwsrc, record for FPI1/FPI3
- * - Determine ARG parts of FPI0/FPI2, unused args are filled
- * with ARG_ZERO.
- */
- for (i=0;i<3;i++) {
- int srcpos;
-
- if (i >= argc) {
- vswz[i] = R300_FPI0_ARGC_ZERO;
- sswz[i] = R300_FPI2_ARGA_ZERO;
- continue;
- }
-
- hwsrc[i] = t_hw_src(rp, src[i], GL_FALSE);
-
- if (emit_vop && vop != R300_FPI0_OUTC_REPL_ALPHA) {
- srcpos = add_src(rp, hwsrc[i], vpos,
- v_swiz[REG_GET_VSWZ(src[i])].flags);
- vswz[i] = (v_swiz[REG_GET_VSWZ(src[i])].base +
- (srcpos *
- v_swiz[REG_GET_VSWZ(src[i])].stride)) |
- ((src[i] & REG_NEGV_MASK) ? ARG_NEG : 0) |
- ((src[i] & REG_ABS_MASK) ? ARG_ABS : 0);
- } else vswz[i] = R300_FPI0_ARGC_ZERO;
-
- if (emit_sop) {
- srcpos = add_src(rp, hwsrc[i], spos,
- s_swiz[REG_GET_SSWZ(src[i])].flags);
- sswz[i] = (s_swiz[REG_GET_SSWZ(src[i])].base +
- (srcpos *
- s_swiz[REG_GET_SSWZ(src[i])].stride)) |
- ((src[i] & REG_NEGS_MASK) ? ARG_NEG : 0) |
- ((src[i] & REG_ABS_MASK) ? ARG_ABS : 0);
- } else sswz[i] = R300_FPI2_ARGA_ZERO;
- }
- hwdest = t_hw_dst(rp, dest, GL_FALSE);
-
+ pos = find_and_prepare_slot(rp, emit_vop, emit_sop, argc, src, dest, mask);
+ if (pos < 0)
+ return;
+
+ hwdest = t_hw_dst(rp, dest, GL_FALSE, pos); /* Note: Side effects wrt register allocation */
+
if (flags & PFS_FLAG_SAT) {
vop |= R300_FPI0_OUTC_SAT;
sop |= R300_FPI2_OUTA_SAT;
}
/* Throw the pieces together and get FPI0/1 */
- rp->alu.inst[vpos].inst1 =
- ((cs->slot[vpos].vsrc[0] << R300_FPI1_SRC0C_SHIFT) |
- (cs->slot[vpos].vsrc[1] << R300_FPI1_SRC1C_SHIFT) |
- (cs->slot[vpos].vsrc[2] << R300_FPI1_SRC2C_SHIFT));
if (emit_vop) {
- rp->alu.inst[vpos].inst0 = vop |
- (vswz[0] << R300_FPI0_ARG0C_SHIFT) |
- (vswz[1] << R300_FPI0_ARG1C_SHIFT) |
- (vswz[2] << R300_FPI0_ARG2C_SHIFT);
+ rp->alu.inst[pos].inst0 |= vop;
+
+ rp->alu.inst[pos].inst1 |= hwdest << R300_FPI1_DSTC_SHIFT;
- rp->alu.inst[vpos].inst1 |= hwdest << R300_FPI1_DSTC_SHIFT;
if (REG_GET_TYPE(dest) == REG_TYPE_OUTPUT) {
if (REG_GET_INDEX(dest) == FRAG_RESULT_COLR) {
- rp->alu.inst[vpos].inst1 |=
+ rp->alu.inst[pos].inst1 |=
(mask & WRITEMASK_XYZ) << R300_FPI1_DSTC_OUTPUT_MASK_SHIFT;
} else assert(0);
} else {
- rp->alu.inst[vpos].inst1 |=
+ rp->alu.inst[pos].inst1 |=
(mask & WRITEMASK_XYZ) << R300_FPI1_DSTC_REG_MASK_SHIFT;
+
+ cs->hwtemps[hwdest].vector_valid = pos+1;
}
- cs->v_pos = vpos+1;
- } else if (spos >= vpos)
- rp->alu.inst[spos].inst0 = NOP_INST0;
+ }
/* And now FPI2/3 */
- rp->alu.inst[spos].inst3 =
- ((cs->slot[spos].ssrc[0] << R300_FPI3_SRC0A_SHIFT) |
- (cs->slot[spos].ssrc[1] << R300_FPI3_SRC1A_SHIFT) |
- (cs->slot[spos].ssrc[2] << R300_FPI3_SRC2A_SHIFT));
if (emit_sop) {
- rp->alu.inst[spos].inst2 = sop |
- sswz[0] << R300_FPI2_ARG0A_SHIFT |
- sswz[1] << R300_FPI2_ARG1A_SHIFT |
- sswz[2] << R300_FPI2_ARG2A_SHIFT;
+ rp->alu.inst[pos].inst2 |= sop;
if (mask & WRITEMASK_W) {
if (REG_GET_TYPE(dest) == REG_TYPE_OUTPUT) {
if (REG_GET_INDEX(dest) == FRAG_RESULT_COLR) {
- rp->alu.inst[spos].inst3 |=
+ rp->alu.inst[pos].inst3 |=
(hwdest << R300_FPI3_DSTA_SHIFT) | R300_FPI3_DSTA_OUTPUT;
} else if (REG_GET_INDEX(dest) == FRAG_RESULT_DEPR) {
- rp->alu.inst[spos].inst3 |= R300_FPI3_DSTA_DEPTH;
+ rp->alu.inst[pos].inst3 |= R300_FPI3_DSTA_DEPTH;
} else assert(0);
} else {
- rp->alu.inst[spos].inst3 |=
+ rp->alu.inst[pos].inst3 |=
(hwdest << R300_FPI3_DSTA_SHIFT) | R300_FPI3_DSTA_REG;
+
+ cs->hwtemps[hwdest].scalar_valid = pos+1;
}
}
- cs->s_pos = spos+1;
- } else if (vpos >= spos)
- rp->alu.inst[vpos].inst2 = NOP_INST2;
+ }
return;
}
}
#endif
- static void make_sin_const(struct r300_fragment_program *rp)
+ static GLfloat SinCosConsts[2][4] = {
+ {
+ 1.273239545, // 4/PI
+ -0.405284735, // -4/(PI*PI)
+ 3.141592654, // PI
+ 0.2225 // weight
+ },
+ {
+ 0.75,
+ 0.0,
+ 0.159154943, // 1/(2*PI)
+ 6.283185307 // 2*PI
+ }
+ };
+
+
+ /**
+ * Emit a LIT instruction.
+ * \p flags may be PFS_FLAG_SAT
+ *
+ * Definition of LIT (from ARB_fragment_program):
+ * tmp = VectorLoad(op0);
+ * if (tmp.x < 0) tmp.x = 0;
+ * if (tmp.y < 0) tmp.y = 0;
+ * if (tmp.w < -(128.0-epsilon)) tmp.w = -(128.0-epsilon);
+ * else if (tmp.w > 128-epsilon) tmp.w = 128-epsilon;
+ * result.x = 1.0;
+ * result.y = tmp.x;
+ * result.z = (tmp.x > 0) ? RoughApproxPower(tmp.y, tmp.w) : 0.0;
+ * result.w = 1.0;
+ *
+ * The longest path of computation is the one leading to result.z,
+ * consisting of 5 operations. This implementation of LIT takes
+ * 5 slots. So unless there's some special undocumented opcode,
+ * this implementation is potentially optimal. Unfortunately,
+ * emit_arith is a bit too conservative because it doesn't understand
+ * partial writes to the vector component.
+ */
+ static const GLfloat LitConst[4] = { 127.999999, 127.999999, 127.999999, -127.999999 };
+
+ static void emit_lit(struct r300_fragment_program *rp,
+ GLuint dest,
+ int mask,
+ GLuint src,
+ int flags)
{
- if(rp->const_sin[0] == -1){
- GLfloat cnstv[4];
-
- cnstv[0] = 1.273239545; // 4/PI
- cnstv[1] =-0.405284735; // -4/(PI*PI)
- cnstv[2] = 3.141592654; // PI
- cnstv[3] = 0.2225; // weight
- rp->const_sin[0] = emit_const4fv(rp, cnstv);
-
- cnstv[0] = 0.75;
- cnstv[1] = 0.0;
- cnstv[2] = 0.159154943; // 1/(2*PI)
- cnstv[3] = 6.283185307; // 2*PI
- rp->const_sin[1] = emit_const4fv(rp, cnstv);
+ COMPILE_STATE;
+ GLuint cnst;
+ int needTemporary;
+ GLuint temp;
+
+ cnst = emit_const4fv(rp, LitConst);
+
+ needTemporary = 0;
+ if ((mask & WRITEMASK_XYZW) != WRITEMASK_XYZW) {
+ needTemporary = 1;
+ } else if (REG_GET_TYPE(dest) == REG_TYPE_OUTPUT) {
+ // LIT is typically followed by DP3/DP4, so there's no point
+ // in creating special code for this case
+ needTemporary = 1;
+ }
+
+ if (needTemporary) {
+ temp = keep(get_temp_reg(rp));
+ } else {
+ temp = keep(dest);
+ }
+
+ // Note: The order of emit_arith inside the slots is relevant,
+ // because emit_arith only looks at scalar vs. vector when resolving
+ // dependencies, and it does not consider individual vector components,
+ // so swizzling between the two parts can create fake dependencies.
+
+ // First slot
+ emit_arith(rp, PFS_OP_MAX, temp, WRITEMASK_XY,
+ keep(src), pfs_zero, undef, 0);
+ emit_arith(rp, PFS_OP_MAX, temp, WRITEMASK_W,
+ src, cnst, undef, 0);
+
+ // Second slot
+ emit_arith(rp, PFS_OP_MIN, temp, WRITEMASK_Z,
+ swizzle(temp, W, W, W, W), cnst, undef, 0);
+ emit_arith(rp, PFS_OP_LG2, temp, WRITEMASK_W,
+ swizzle(temp, Y, Y, Y, Y), undef, undef, 0);
+
+ // Third slot
+ // If desired, we saturate the y result here.
+ // This does not affect the use as a condition variable in the CMP later
+ emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_W,
+ temp, swizzle(temp, Z, Z, Z, Z), pfs_zero, 0);
+ emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_Y,
+ swizzle(temp, X, X, X, X), pfs_one, pfs_zero, flags);
+
+ // Fourth slot
+ emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_X,
+ pfs_one, pfs_one, pfs_zero, 0);
+ emit_arith(rp, PFS_OP_EX2, temp, WRITEMASK_W,
+ temp, undef, undef, 0);
+
+ // Fifth slot
+ emit_arith(rp, PFS_OP_CMP, temp, WRITEMASK_Z,
+ pfs_zero, swizzle(temp, W, W, W, W), negate(swizzle(temp, Y, Y, Y, Y)), flags);
+ emit_arith(rp, PFS_OP_MAD, temp, WRITEMASK_W,
+ pfs_one, pfs_one, pfs_zero, 0);
+
+ if (needTemporary) {
+ emit_arith(rp, PFS_OP_MAD, dest, mask,
+ temp, pfs_one, pfs_zero, flags);
+ free_temp(rp, temp);
+ } else {
+ // Decrease refcount of the destination
+ t_hw_dst(rp, dest, GL_FALSE, cs->nrslots);
}
}
+
static GLboolean parse_program(struct r300_fragment_program *rp)
- {
+ {
struct gl_fragment_program *mp = &rp->mesa_program;
const struct prog_instruction *inst = mp->Base.Instructions;
struct prog_instruction *fpi;
GLuint src[3], dest, temp[2];
- GLuint cnst;
int flags, mask = 0;
- GLfloat cnstv[4] = {0.0, 0.0, 0.0, 0.0};
+ int const_sin[2];
if (!inst || inst[0].Opcode == OPCODE_END) {
ERROR("empty program?\n");
* result = sin(x)
*/
temp[0] = get_temp_reg(rp);
- make_sin_const(rp);
+ const_sin[0] = emit_const4fv(rp, SinCosConsts[0]);
+ const_sin[1] = emit_const4fv(rp, SinCosConsts[1]);
src[0] = t_scalar_src(rp, fpi->SrcReg[0]);
/* add 0.5*PI and do range reduction */
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_X,
swizzle(src[0], X, X, X, X),
- swizzle(rp->const_sin[1], Z, Z, Z, Z),
- swizzle(rp->const_sin[1], X, X, X, X),
+ swizzle(const_sin[1], Z, Z, Z, Z),
+ swizzle(const_sin[1], X, X, X, X),
0);
emit_arith(rp, PFS_OP_FRC, temp[0], WRITEMASK_X,
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_Z,
swizzle(temp[0], X, X, X, X),
- swizzle(rp->const_sin[1], W, W, W, W), //2*PI
- negate(swizzle(rp->const_sin[0], Z, Z, Z, Z)), //-PI
+ swizzle(const_sin[1], W, W, W, W), //2*PI
+ negate(swizzle(const_sin[0], Z, Z, Z, Z)), //-PI
0);
/* SIN */
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_X | WRITEMASK_Y,
swizzle(temp[0], Z, Z, Z, Z),
- rp->const_sin[0],
+ const_sin[0],
pfs_zero,
0);
absolute(swizzle(temp[0], Z, Z, Z, Z)),
swizzle(temp[0], X, X, X, X),
0);
-
+
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_Y,
swizzle(temp[0], X, X, X, X),
absolute(swizzle(temp[0], X, X, X, X)),
emit_arith(rp, PFS_OP_MAD, dest, mask,
swizzle(temp[0], Y, Y, Y, Y),
- swizzle(rp->const_sin[0], W, W, W, W),
+ swizzle(const_sin[0], W, W, W, W),
swizzle(temp[0], X, X, X, X),
flags);
0);
emit_arith(rp, PFS_OP_DP4, dest, mask,
temp[0], src[1], undef,
- flags);
+ flags);
free_temp(rp, temp[0]);
#else
emit_arith(rp, PFS_OP_DP4, dest, mask,
swizzle(src[0], X, Y, Z, ONE), src[1],
- undef, flags);
+ undef, flags);
#endif
break;
case OPCODE_DST:
src[0], undef, undef,
flags);
break;
- case OPCODE_FLR:
+ case OPCODE_FLR:
src[0] = t_src(rp, fpi->SrcReg[0]);
temp[0] = get_temp_reg(rp);
/* FRC temp, src0
flags);
break;
case OPCODE_LIT:
- /* LIT
- * if (s.x < 0) t.x = 0; else t.x = s.x;
- * if (s.y < 0) t.y = 0; else t.y = s.y;
- * if (s.w > 128.0) t.w = 128.0; else t.w = s.w;
- * if (s.w < -128.0) t.w = -128.0; else t.w = s.w;
- * r.x = 1.0
- * if (t.x > 0) r.y = pow(t.y, t.w); else r.y = 0;
- * Also r.y = 0 if t.y < 0
- * For the t.x > 0 FGLRX use the CMPH opcode which
- * change the compare to (t.x + 0.5) > 0.5 we may
- * save one instruction by doing CMP -t.x
- */
- cnstv[0] = cnstv[1] = cnstv[2] = cnstv[3] = 0.50001;
src[0] = t_src(rp, fpi->SrcReg[0]);
- temp[0] = get_temp_reg(rp);
- cnst = emit_const4fv(rp, cnstv);
- emit_arith(rp, PFS_OP_CMP, temp[0],
- WRITEMASK_X | WRITEMASK_Y,
- src[0], pfs_zero, src[0], flags);
- emit_arith(rp, PFS_OP_MIN, temp[0], WRITEMASK_Z,
- swizzle(keep(src[0]), W, W, W, W),
- cnst, undef, flags);
- emit_arith(rp, PFS_OP_LG2, temp[0], WRITEMASK_W,
- swizzle(temp[0], Y, Y, Y, Y),
- undef, undef, flags);
- emit_arith(rp, PFS_OP_MAX, temp[0], WRITEMASK_Z,
- temp[0], negate(cnst), undef, flags);
- emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_W,
- temp[0], swizzle(temp[0], Z, Z, Z, Z),
- pfs_zero, flags);
- emit_arith(rp, PFS_OP_EX2, temp[0], WRITEMASK_W,
- temp[0], undef, undef, flags);
- emit_arith(rp, PFS_OP_MAD, dest, WRITEMASK_Y,
- swizzle(keep(temp[0]), X, X, X, X),
- pfs_one, pfs_zero, flags);
- #if 0
- emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_X,
- temp[0], pfs_one, pfs_half, flags);
- emit_arith(rp, PFS_OP_CMPH, temp[0], WRITEMASK_Z,
- swizzle(keep(temp[0]), W, W, W, W),
- pfs_zero, swizzle(keep(temp[0]), X, X, X, X),
- flags);
- #else
- emit_arith(rp, PFS_OP_CMP, temp[0], WRITEMASK_Z,
- pfs_zero,
- swizzle(keep(temp[0]), W, W, W, W),
- negate(swizzle(keep(temp[0]), X, X, X, X)),
- flags);
- #endif
- emit_arith(rp, PFS_OP_CMP, dest, WRITEMASK_Z,
- pfs_zero, temp[0],
- negate(swizzle(keep(temp[0]), Y, Y, Y, Y)),
- flags);
- emit_arith(rp, PFS_OP_MAD, dest,
- WRITEMASK_X | WRITEMASK_W,
- pfs_one,
- pfs_one,
- pfs_zero,
- flags);
- free_temp(rp, temp[0]);
+ emit_lit(rp, dest, mask, src[0], flags);
break;
case OPCODE_LRP:
src[0] = t_src(rp, fpi->SrcReg[0]);
src[0], src[1], temp[0],
flags);
free_temp(rp, temp[0]);
- break;
+ break;
case OPCODE_MAD:
src[0] = t_src(rp, fpi->SrcReg[0]);
src[1] = t_src(rp, fpi->SrcReg[1]);
case OPCODE_SWZ:
src[0] = t_src(rp, fpi->SrcReg[0]);
emit_arith(rp, PFS_OP_MAD, dest, mask,
- src[0], pfs_one, pfs_zero,
+ src[0], pfs_one, pfs_zero,
flags);
break;
case OPCODE_MUL:
case OPCODE_POW:
src[0] = t_scalar_src(rp, fpi->SrcReg[0]);
src[1] = t_scalar_src(rp, fpi->SrcReg[1]);
- temp[0] = get_temp_reg(rp);
+ temp[0] = get_temp_reg(rp);
emit_arith(rp, PFS_OP_LG2, temp[0], WRITEMASK_W,
src[0], undef, undef,
0);
*/
temp[0] = get_temp_reg(rp);
temp[1] = get_temp_reg(rp);
- make_sin_const(rp);
+ const_sin[0] = emit_const4fv(rp, SinCosConsts[0]);
+ const_sin[1] = emit_const4fv(rp, SinCosConsts[1]);
src[0] = t_scalar_src(rp, fpi->SrcReg[0]);
/* x = -abs(x)+0.5*PI */
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_Z,
- swizzle(rp->const_sin[0], Z, Z, Z, Z), //PI
+ swizzle(const_sin[0], Z, Z, Z, Z), //PI
pfs_half,
negate(abs(swizzle(keep(src[0]), X, X, X, X))),
0);
/* C*x (sin) */
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_W,
- swizzle(rp->const_sin[0], Y, Y, Y, Y),
+ swizzle(const_sin[0], Y, Y, Y, Y),
swizzle(keep(src[0]), X, X, X, X),
pfs_zero,
0);
/* B*x, C*x (cos) */
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_X | WRITEMASK_Y,
swizzle(temp[0], Z, Z, Z, Z),
- rp->const_sin[0],
+ const_sin[0],
pfs_zero,
0);
/* B*x (sin) */
emit_arith(rp, PFS_OP_MAD, temp[1], WRITEMASK_W,
- swizzle(rp->const_sin[0], X, X, X, X),
+ swizzle(const_sin[0], X, X, X, X),
keep(src[0]),
pfs_zero,
0);
/* dest.xy = mad(temp.xy, P, temp2.wz) */
emit_arith(rp, PFS_OP_MAD, dest, mask & (WRITEMASK_X | WRITEMASK_Y),
temp[0],
- swizzle(rp->const_sin[0], W, W, W, W),
+ swizzle(const_sin[0], W, W, W, W),
swizzle(temp[1], W, Z, Y, X),
flags);
*/
temp[0] = get_temp_reg(rp);
- make_sin_const(rp);
+ const_sin[0] = emit_const4fv(rp, SinCosConsts[0]);
+ const_sin[1] = emit_const4fv(rp, SinCosConsts[1]);
src[0] = t_scalar_src(rp, fpi->SrcReg[0]);
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_X,
swizzle(keep(src[0]), X, X, X, X),
- swizzle(rp->const_sin[1], Z, Z, Z, Z),
+ swizzle(const_sin[1], Z, Z, Z, Z),
pfs_half,
0);
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_Z,
swizzle(temp[0], X, X, X, X),
- swizzle(rp->const_sin[1], W, W, W, W), //2*PI
- negate(swizzle(rp->const_sin[0], Z, Z, Z, Z)), //PI
+ swizzle(const_sin[1], W, W, W, W), //2*PI
+ negate(swizzle(const_sin[0], Z, Z, Z, Z)), //PI
0);
/* SIN */
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_X | WRITEMASK_Y,
swizzle(temp[0], Z, Z, Z, Z),
- rp->const_sin[0],
+ const_sin[0],
pfs_zero,
0);
absolute(swizzle(temp[0], Z, Z, Z, Z)),
swizzle(temp[0], X, X, X, X),
0);
-
+
emit_arith(rp, PFS_OP_MAD, temp[0], WRITEMASK_Y,
swizzle(temp[0], X, X, X, X),
absolute(swizzle(temp[0], X, X, X, X)),
emit_arith(rp, PFS_OP_MAD, dest, mask,
swizzle(temp[0], Y, Y, Y, Y),
- swizzle(rp->const_sin[0], W, W, W, W),
+ swizzle(const_sin[0], W, W, W, W),
swizzle(temp[0], X, X, X, X),
flags);
swizzle(keep(src[1]), Y, Z, X, W),
pfs_zero,
0);
- /* dest.xyz = src0.yzx * src1.zxy - temp
+ /* dest.xyz = src0.yzx * src1.zxy - temp
* dest.w = undefined
* */
emit_arith(rp, PFS_OP_MAD, dest, mask & WRITEMASK_XYZ,
return GL_TRUE;
}
+ static void insert_wpos(struct gl_program *prog)
+ {
+ GLint tokens[6] = { STATE_INTERNAL, STATE_R300_WINDOW_DIMENSION, 0, 0, 0, 0 };
+ struct prog_instruction *fpi;
+ GLuint window_index;
+ int i = 0;
+ GLuint tempregi = prog->NumTemporaries;
+ /* should do something else if no temps left... */
+ prog->NumTemporaries++;
+
+ fpi = _mesa_alloc_instructions (prog->NumInstructions + 3);
+ _mesa_init_instructions (fpi, prog->NumInstructions + 3);
+
+ /* perspective divide */
+ fpi[i].Opcode = OPCODE_RCP;
+
+ fpi[i].DstReg.File = PROGRAM_TEMPORARY;
+ fpi[i].DstReg.Index = tempregi;
+ fpi[i].DstReg.WriteMask = WRITEMASK_W;
+ fpi[i].DstReg.CondMask = COND_TR;
+
+ fpi[i].SrcReg[0].File = PROGRAM_INPUT;
+ fpi[i].SrcReg[0].Index = FRAG_ATTRIB_WPOS;
+ fpi[i].SrcReg[0].Swizzle = SWIZZLE_WWWW;
+ i++;
+
+ fpi[i].Opcode = OPCODE_MUL;
+
+ fpi[i].DstReg.File = PROGRAM_TEMPORARY;
+ fpi[i].DstReg.Index = tempregi;
+ fpi[i].DstReg.WriteMask = WRITEMASK_XYZ;
+ fpi[i].DstReg.CondMask = COND_TR;
+
+ fpi[i].SrcReg[0].File = PROGRAM_INPUT;
+ fpi[i].SrcReg[0].Index = FRAG_ATTRIB_WPOS;
+ fpi[i].SrcReg[0].Swizzle = SWIZZLE_XYZW;
+
+ fpi[i].SrcReg[1].File = PROGRAM_TEMPORARY;
+ fpi[i].SrcReg[1].Index = tempregi;
+ fpi[i].SrcReg[1].Swizzle = SWIZZLE_WWWW;
+ i++;
+
+ /* viewport transformation */
+ window_index = _mesa_add_state_reference(prog->Parameters, tokens);
+
+ fpi[i].Opcode = OPCODE_MAD;
+
+ fpi[i].DstReg.File = PROGRAM_TEMPORARY;
+ fpi[i].DstReg.Index = tempregi;
+ fpi[i].DstReg.WriteMask = WRITEMASK_XYZ;
+ fpi[i].DstReg.CondMask = COND_TR;
+
+ fpi[i].SrcReg[0].File = PROGRAM_TEMPORARY;
+ fpi[i].SrcReg[0].Index = tempregi;
+ fpi[i].SrcReg[0].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ZERO);
+
+ fpi[i].SrcReg[1].File = PROGRAM_STATE_VAR;
+ fpi[i].SrcReg[1].Index = window_index;
+ fpi[i].SrcReg[1].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ZERO);
+
+ fpi[i].SrcReg[2].File = PROGRAM_STATE_VAR;
+ fpi[i].SrcReg[2].Index = window_index;
+ fpi[i].SrcReg[2].Swizzle = MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_Y, SWIZZLE_Z, SWIZZLE_ZERO);
+ i++;
+
+ _mesa_copy_instructions (&fpi[i], prog->Instructions, prog->NumInstructions);
+
+ free(prog->Instructions);
+
+ prog->Instructions = fpi;
+
+ prog->NumInstructions += i;
+ fpi = &prog->Instructions[prog->NumInstructions-1];
+
+ assert(fpi->Opcode == OPCODE_END);
+
+ for(fpi = &prog->Instructions[3]; fpi->Opcode != OPCODE_END; fpi++){
+ for(i=0; i<3; i++)
+ if( fpi->SrcReg[i].File == PROGRAM_INPUT &&
+ fpi->SrcReg[i].Index == FRAG_ATTRIB_WPOS ){
+ fpi->SrcReg[i].File = PROGRAM_TEMPORARY;
+ fpi->SrcReg[i].Index = tempregi;
+ }
+ }
+ }
+
/* - Init structures
* - Determine what hwregs each input corresponds to
*/
static void init_program(r300ContextPtr r300, struct r300_fragment_program *rp)
{
struct r300_pfs_compile_state *cs = NULL;
- struct gl_fragment_program *mp = &rp->mesa_program;
+ struct gl_fragment_program *mp = &rp->mesa_program;
struct prog_instruction *fpi;
GLuint InputsRead = mp->Base.InputsRead;
GLuint temps_used = 0; /* for rp->temps[] */
rp->cur_node = 0;
rp->first_node_has_tex = 0;
rp->const_nr = 0;
- rp->param_nr = 0;
- rp->params_uptodate = GL_FALSE;
rp->max_temp_idx = 0;
rp->node[0].alu_end = -1;
rp->node[0].tex_end = -1;
- rp->const_sin[0] = -1;
-
+
_mesa_memset(cs, 0, sizeof(*rp->cs));
for (i=0;i<PFS_MAX_ALU_INST;i++) {
for (j=0;j<3;j++) {
cs->slot[i].ssrc[j] = SRC_CONST;
}
}
-
+
/* Work out what temps the Mesa inputs correspond to, this must match
* what setup_rs_unit does, which shouldn't be a problem as rs_unit
* configures itself based on the fragprog's InputsRead
for (i=0;i<rp->ctx->Const.MaxTextureUnits;i++) {
if (InputsRead & (FRAG_BIT_TEX0 << i)) {
cs->inputs[FRAG_ATTRIB_TEX0+i].refcount = 0;
- cs->inputs[FRAG_ATTRIB_TEX0+i].reg = get_hw_temp(rp);
+ cs->inputs[FRAG_ATTRIB_TEX0+i].reg = get_hw_temp(rp, 0);
}
}
InputsRead &= ~FRAG_BITS_TEX_ANY;
/* fragment position treated as a texcoord */
if (InputsRead & FRAG_BIT_WPOS) {
cs->inputs[FRAG_ATTRIB_WPOS].refcount = 0;
- cs->inputs[FRAG_ATTRIB_WPOS].reg = get_hw_temp(rp);
+ cs->inputs[FRAG_ATTRIB_WPOS].reg = get_hw_temp(rp, 0);
+ insert_wpos(&mp->Base);
}
InputsRead &= ~FRAG_BIT_WPOS;
/* Then primary colour */
if (InputsRead & FRAG_BIT_COL0) {
cs->inputs[FRAG_ATTRIB_COL0].refcount = 0;
- cs->inputs[FRAG_ATTRIB_COL0].reg = get_hw_temp(rp);
+ cs->inputs[FRAG_ATTRIB_COL0].reg = get_hw_temp(rp, 0);
}
InputsRead &= ~FRAG_BIT_COL0;
-
+
/* Secondary color */
if (InputsRead & FRAG_BIT_COL1) {
cs->inputs[FRAG_ATTRIB_COL1].refcount = 0;
- cs->inputs[FRAG_ATTRIB_COL1].reg = get_hw_temp(rp);
+ cs->inputs[FRAG_ATTRIB_COL1].reg = get_hw_temp(rp, 0);
}
InputsRead &= ~FRAG_BIT_COL1;
for (fpi=mp->Base.Instructions;fpi->Opcode != OPCODE_END; fpi++) {
int idx;
-
+
for (i=0;i<3;i++) {
idx = fpi->SrcReg[i].Index;
switch (fpi->SrcReg[i].File) {
static void update_params(struct r300_fragment_program *rp)
{
struct gl_fragment_program *mp = &rp->mesa_program;
- int i;
/* Ask Mesa nicely to fill in ParameterValues for us */
- if (rp->param_nr)
+ if (mp->Base.Parameters)
_mesa_load_state_parameters(rp->ctx, mp->Base.Parameters);
-
- for (i=0;i<rp->param_nr;i++)
- COPY_4V(rp->constant[rp->param[i].idx], rp->param[i].values);
-
- rp->params_uptodate = GL_TRUE;
}
void r300_translate_fragment_shader(r300ContextPtr r300, struct r300_fragment_program *rp)
struct r300_pfs_compile_state *cs = NULL;
if (!rp->translated) {
-
+
init_program(r300, rp);
cs = rp->cs;
dump_program(rp);
return;
}
-
+
/* Finish off */
- cs->v_pos = cs->s_pos = MAX2(cs->v_pos, cs->s_pos);
rp->node[rp->cur_node].alu_end =
- cs->v_pos - rp->node[rp->cur_node].alu_offset - 1;
+ cs->nrslots - rp->node[rp->cur_node].alu_offset - 1;
if (rp->node[rp->cur_node].tex_end < 0)
rp->node[rp->cur_node].tex_end = 0;
rp->alu_offset = 0;
- rp->alu_end = cs->v_pos - 1;
+ rp->alu_end = cs->nrslots - 1;
rp->tex_offset = 0;
rp->tex_end = rp->tex.length ? rp->tex.length - 1 : 0;
assert(rp->node[rp->cur_node].alu_end >= 0);
assert(rp->alu_end >= 0);
-
+
rp->translated = GL_TRUE;
- if (0) dump_program(rp);
+ if (RADEON_DEBUG & DEBUG_PIXEL) dump_program(rp);
+ r300UpdateStateParameters(rp->ctx, _NEW_PROGRAM);
}
update_params(rp);
/* just some random things... */
static void dump_program(struct r300_fragment_program *rp)
{
- int i;
+ int n, i, j;
static int pc = 0;
fprintf(stderr, "pc=%d*************************************\n", pc++);
-
+
fprintf(stderr, "Mesa program:\n");
fprintf(stderr, "-------------\n");
_mesa_print_program(&rp->mesa_program.Base);
fprintf(stderr, "Hardware program\n");
fprintf(stderr, "----------------\n");
-
- fprintf(stderr, "tex:\n");
-
- for(i=0;i<rp->tex.length;i++) {
- fprintf(stderr, "%08x\n", rp->tex.inst[i]);
- }
-
- for (i=0;i<(rp->cur_node+1);i++) {
+
+ for (n = 0; n < (rp->cur_node+1); n++) {
fprintf(stderr, "NODE %d: alu_offset: %d, tex_offset: %d, "\
- "alu_end: %d, tex_end: %d\n", i,
- rp->node[i].alu_offset,
- rp->node[i].tex_offset,
- rp->node[i].alu_end,
- rp->node[i].tex_end);
+ "alu_end: %d, tex_end: %d\n", n,
+ rp->node[n].alu_offset,
+ rp->node[n].tex_offset,
+ rp->node[n].alu_end,
+ rp->node[n].tex_end);
+
+ if (rp->tex.length) {
+ fprintf(stderr, " TEX:\n");
+ for(i = rp->node[n].tex_offset; i <= rp->node[n].tex_offset+rp->node[n].tex_end; ++i) {
+ const char* instr;
+
+ switch((rp->tex.inst[i] >> R300_FPITX_OPCODE_SHIFT) & 15) {
+ case R300_FPITX_OP_TEX:
+ instr = "TEX";
+ break;
+ case R300_FPITX_OP_KIL:
+ instr = "KIL";
+ break;
+ case R300_FPITX_OP_TXP:
+ instr = "TXP";
+ break;
+ case R300_FPITX_OP_TXB:
+ instr = "TXB";
+ break;
+ default:
+ instr = "UNKNOWN";
+ }
+
+ fprintf(stderr, " %s t%i, %c%i, texture[%i] (%08x)\n",
+ instr,
+ (rp->tex.inst[i] >> R300_FPITX_DST_SHIFT) & 31,
+ (rp->tex.inst[i] & R300_FPITX_SRC_CONST) ? 'c': 't',
+ (rp->tex.inst[i] >> R300_FPITX_SRC_SHIFT) & 31,
+ (rp->tex.inst[i] & R300_FPITX_IMAGE_MASK) >> R300_FPITX_IMAGE_SHIFT,
+ rp->tex.inst[i]);
+ }
+ }
+
+ for(i = rp->node[n].alu_offset; i <= rp->node[n].alu_offset+rp->node[n].alu_end; ++i) {
+ char srcc[3][10], dstc[20];
+ char srca[3][10], dsta[20];
+ char argc[3][20];
+ char arga[3][20];
+ char flags[5], tmp[10];
+
+ for(j = 0; j < 3; ++j) {
+ int regc = rp->alu.inst[i].inst1 >> (j*6);
+ int rega = rp->alu.inst[i].inst3 >> (j*6);
+
+ sprintf(srcc[j], "%c%i", (regc & 32) ? 'c' : 't', regc & 31);
+ sprintf(srca[j], "%c%i", (rega & 32) ? 'c' : 't', rega & 31);
+ }
+
+ dstc[0] = 0;
+ sprintf(flags, "%s%s%s",
+ (rp->alu.inst[i].inst1 & R300_FPI1_DSTC_REG_X) ? "x" : "",
+ (rp->alu.inst[i].inst1 & R300_FPI1_DSTC_REG_Y) ? "y" : "",
+ (rp->alu.inst[i].inst1 & R300_FPI1_DSTC_REG_Z) ? "z" : "");
+ if (flags[0] != 0) {
+ sprintf(dstc, "t%i.%s ",
+ (rp->alu.inst[i].inst1 >> R300_FPI1_DSTC_SHIFT) & 31,
+ flags);
+ }
+ sprintf(flags, "%s%s%s",
+ (rp->alu.inst[i].inst1 & R300_FPI1_DSTC_OUTPUT_X) ? "x" : "",
+ (rp->alu.inst[i].inst1 & R300_FPI1_DSTC_OUTPUT_Y) ? "y" : "",
+ (rp->alu.inst[i].inst1 & R300_FPI1_DSTC_OUTPUT_Z) ? "z" : "");
+ if (flags[0] != 0) {
+ sprintf(tmp, "o%i.%s",
+ (rp->alu.inst[i].inst1 >> R300_FPI1_DSTC_SHIFT) & 31,
+ flags);
+ strcat(dstc, tmp);
+ }
+
+ dsta[0] = 0;
+ if (rp->alu.inst[i].inst3 & R300_FPI3_DSTA_REG) {
+ sprintf(dsta, "t%i.w ", (rp->alu.inst[i].inst3 >> R300_FPI3_DSTA_SHIFT) & 31);
+ }
+ if (rp->alu.inst[i].inst3 & R300_FPI3_DSTA_OUTPUT) {
+ sprintf(tmp, "o%i.w ", (rp->alu.inst[i].inst3 >> R300_FPI3_DSTA_SHIFT) & 31);
+ strcat(dsta, tmp);
+ }
+ if (rp->alu.inst[i].inst3 & R300_FPI3_DSTA_DEPTH) {
+ strcat(dsta, "Z");
+ }
+
+ fprintf(stderr, "%3i: xyz: %3s %3s %3s -> %-20s (%08x)\n"
+ " w: %3s %3s %3s -> %-20s (%08x)\n",
+ i,
+ srcc[0], srcc[1], srcc[2], dstc, rp->alu.inst[i].inst1,
+ srca[0], srca[1], srca[2], dsta, rp->alu.inst[i].inst3);
+
+ for(j = 0; j < 3; ++j) {
+ int regc = rp->alu.inst[i].inst0 >> (j*7);
+ int rega = rp->alu.inst[i].inst2 >> (j*7);
+ int d;
+ char buf[20];
+
+ d = regc & 31;
+ if (d < 12) {
+ switch(d % 4) {
+ case R300_FPI0_ARGC_SRC0C_XYZ:
+ sprintf(buf, "%s.xyz", srcc[d / 4]);
+ break;
+ case R300_FPI0_ARGC_SRC0C_XXX:
+ sprintf(buf, "%s.xxx", srcc[d / 4]);
+ break;
+ case R300_FPI0_ARGC_SRC0C_YYY:
+ sprintf(buf, "%s.yyy", srcc[d / 4]);
+ break;
+ case R300_FPI0_ARGC_SRC0C_ZZZ:
+ sprintf(buf, "%s.zzz", srcc[d / 4]);
+ break;
+ }
+ } else if (d < 15) {
+ sprintf(buf, "%s.www", srca[d-12]);
+ } else if (d == 20) {
+ sprintf(buf, "0.0");
+ } else if (d == 21) {
+ sprintf(buf, "1.0");
+ } else if (d == 22) {
+ sprintf(buf, "0.5");
+ } else if (d >= 23 && d < 32) {
+ d -= 23;
+ switch(d/3) {
+ case 0:
+ sprintf(buf, "%s.yzx", srcc[d % 3]);
+ break;
+ case 1:
+ sprintf(buf, "%s.zxy", srcc[d % 3]);
+ break;
+ case 2:
+ sprintf(buf, "%s.Wzy", srcc[d % 3]);
+ break;
+ }
+ } else {
+ sprintf(buf, "%i", d);
+ }
+
+ sprintf(argc[j], "%s%s%s%s",
+ (regc & 32) ? "-" : "",
+ (regc & 64) ? "|" : "",
+ buf,
+ (regc & 64) ? "|" : "");
+
+ d = rega & 31;
+ if (d < 9) {
+ sprintf(buf, "%s.%c", srcc[d / 3], 'x' + (char)(d%3));
+ } else if (d < 12) {
+ sprintf(buf, "%s.w", srca[d-9]);
+ } else if (d == 16) {
+ sprintf(buf, "0.0");
+ } else if (d == 17) {
+ sprintf(buf, "1.0");
+ } else if (d == 18) {
+ sprintf(buf, "0.5");
+ } else {
+ sprintf(buf, "%i", d);
+ }
+
+ sprintf(arga[j], "%s%s%s%s",
+ (rega & 32) ? "-" : "",
+ (rega & 64) ? "|" : "",
+ buf,
+ (rega & 64) ? "|" : "");
+ }
+
+ fprintf(stderr, " xyz: %8s %8s %8s op: %08x\n"
+ " w: %8s %8s %8s op: %08x\n",
+ argc[0], argc[1], argc[2], rp->alu.inst[i].inst0,
+ arga[0], arga[1], arga[2], rp->alu.inst[i].inst2);
+ }
}
-
- fprintf(stderr, "%08x\n",
- ((rp->tex_end << 16) | (R300_PFS_TEXI_0 >> 2)));
- for (i=0;i<=rp->tex_end;i++)
- fprintf(stderr, "%08x\n", rp->tex.inst[i]);
-
- /* dump program in pretty_print_command_stream.tcl-readable format */
- fprintf(stderr, "%08x\n",
- ((rp->alu_end << 16) | (R300_PFS_INSTR0_0 >> 2)));
- for (i=0;i<=rp->alu_end;i++)
- fprintf(stderr, "%08x\n", rp->alu.inst[i].inst0);
-
- fprintf(stderr, "%08x\n",
- ((rp->alu_end << 16) | (R300_PFS_INSTR1_0 >> 2)));
- for (i=0;i<=rp->alu_end;i++)
- fprintf(stderr, "%08x\n", rp->alu.inst[i].inst1);
-
- fprintf(stderr, "%08x\n",
- ((rp->alu_end << 16) | (R300_PFS_INSTR2_0 >> 2)));
- for (i=0;i<=rp->alu_end;i++)
- fprintf(stderr, "%08x\n", rp->alu.inst[i].inst2);
-
- fprintf(stderr, "%08x\n",
- ((rp->alu_end << 16) | (R300_PFS_INSTR3_0 >> 2)));
- for (i=0;i<=rp->alu_end;i++)
- fprintf(stderr, "%08x\n", rp->alu.inst[i].inst3);
-
- fprintf(stderr, "00000000\n");
}
#include "api_arrayelt.h"
#include "swrast/swrast.h"
#include "swrast_setup/swrast_setup.h"
+#include "shader/prog_parameter.h"
+#include "shader/prog_statevars.h"
#include "vbo/vbo.h"
#include "tnl/tnl.h"
#include "texformat.h"
GLubyte color[4];
r300ContextPtr rmesa = R300_CONTEXT(ctx);
- R300_STATECHANGE(rmesa, unk4E10);
+ R300_STATECHANGE(rmesa, blend_color);
CLAMPED_FLOAT_TO_UBYTE(color[0], cf[0]);
CLAMPED_FLOAT_TO_UBYTE(color[1], cf[1]);
CLAMPED_FLOAT_TO_UBYTE(color[2], cf[2]);
CLAMPED_FLOAT_TO_UBYTE(color[3], cf[3]);
- rmesa->hw.unk4E10.cmd[1] = r300PackColor(4, color[3], color[0],
+ rmesa->hw.blend_color.cmd[1] = r300PackColor(4, color[3], color[0],
color[1], color[2]);
}
*/
r300ContextPtr r300 = R300_CONTEXT(ctx);
- R300_STATECHANGE(r300, unk4F10);
+ R300_STATECHANGE(r300, zstencil_format);
if (ctx->Color.AlphaEnabled && ctx->Color.AlphaFunc != GL_ALWAYS)
/* disable early Z */
- r300->hw.unk4F10.cmd[2] = R300_EARLY_Z_DISABLE;
+ r300->hw.zstencil_format.cmd[2] = R300_EARLY_Z_DISABLE;
else {
if (ctx->Depth.Test && ctx->Depth.Func != GL_NEVER)
/* enable early Z */
- r300->hw.unk4F10.cmd[2] = R300_EARLY_Z_ENABLE;
+ r300->hw.zstencil_format.cmd[2] = R300_EARLY_Z_ENABLE;
else
/* disable early Z */
- r300->hw.unk4F10.cmd[2] = R300_EARLY_Z_DISABLE;
+ r300->hw.zstencil_format.cmd[2] = R300_EARLY_Z_DISABLE;
}
}
break;
case GL_POLYGON_OFFSET_FILL:
- R300_STATECHANGE(r300, unk42B4);
+ R300_STATECHANGE(r300, occlusion_cntl);
if(state){
- r300->hw.unk42B4.cmd[1] |= (3<<0);
+ r300->hw.occlusion_cntl.cmd[1] |= (3<<0);
} else {
- r300->hw.unk42B4.cmd[1] &= ~(3<<0);
+ r300->hw.occlusion_cntl.cmd[1] &= ~(3<<0);
}
break;
default:
}
}
- if (r300->hw.unk4288.cmd[1] != hw_mode) {
- R300_STATECHANGE(r300, unk4288);
- r300->hw.unk4288.cmd[1] = hw_mode;
+ if (r300->hw.polygon_mode.cmd[1] != hw_mode) {
+ R300_STATECHANGE(r300, polygon_mode);
+ r300->hw.polygon_mode.cmd[1] = hw_mode;
}
}
{
r300ContextPtr rmesa = R300_CONTEXT(ctx);
- R300_STATECHANGE(rmesa, unk4274);
+ R300_STATECHANGE(rmesa, shade);
switch (mode) {
case GL_FLAT:
- rmesa->hw.unk4274.cmd[2] = R300_RE_SHADE_MODEL_FLAT;
+ rmesa->hw.shade.cmd[2] = R300_RE_SHADE_MODEL_FLAT;
break;
case GL_SMOOTH:
- rmesa->hw.unk4274.cmd[2] = R300_RE_SHADE_MODEL_SMOOTH;
+ rmesa->hw.shade.cmd[2] = R300_RE_SHADE_MODEL_SMOOTH;
break;
default:
return;
#endif
}
-static void r300FetchStateParameter(GLcontext *ctx, const enum state_index state[],
- GLfloat *value)
+static void
+r300FetchStateParameter(GLcontext *ctx,
+ const gl_state_index state[STATE_LENGTH],
+ GLfloat *value)
{
r300ContextPtr r300 = R300_CONTEXT(ctx);
switch(state[1])
{
case STATE_R300_WINDOW_DIMENSION:
- value[0] = r300->radeon.dri.drawable->w; /* width */
- value[1] = r300->radeon.dri.drawable->h; /* height */
+ value[0] = r300->radeon.dri.drawable->w*0.5f;/* width*0.5 */
+ value[1] = r300->radeon.dri.drawable->h*0.5f;/* height*0.5 */
value[2] = 0.5F; /* for moving range [-1 1] -> [0 1] */
value[3] = 1.0F; /* not used */
break;
* Update R300's own internal state parameters.
* For now just STATE_R300_WINDOW_DIMENSION
*/
- static void r300UpdateStateParameters(GLcontext * ctx, GLuint new_state)
+ void r300UpdateStateParameters(GLcontext * ctx, GLuint new_state)
{
- struct r300_vertex_program_cont *vpc;
+ struct r300_fragment_program *fp;
struct gl_program_parameter_list *paramList;
GLuint i;
if(!(new_state & (_NEW_BUFFERS|_NEW_PROGRAM)))
return;
- vpc = (struct r300_vertex_program_cont *)ctx->VertexProgram._Current;
- if (!vpc)
+ fp = (struct r300_fragment_program *)ctx->FragmentProgram._Current;
+ if (!fp)
return;
- paramList = vpc->mesa_program.Base.Parameters;
+ paramList = fp->mesa_program.Base.Parameters;
if (!paramList)
return;
R300_STATECHANGE(r300, txe);
R300_STATECHANGE(r300, tex.filter);
- R300_STATECHANGE(r300, tex.unknown1);
+ R300_STATECHANGE(r300, tex.filter_1);
R300_STATECHANGE(r300, tex.size);
R300_STATECHANGE(r300, tex.format);
R300_STATECHANGE(r300, tex.pitch);
R300_STATECHANGE(r300, tex.offset);
- R300_STATECHANGE(r300, tex.unknown4);
+ R300_STATECHANGE(r300, tex.chroma_key);
R300_STATECHANGE(r300, tex.border_color);
r300->hw.txe.cmd[R300_TXE_ENABLE]=0x0;
r300->hw.tex.filter.cmd[R300_TEX_VALUE_0 + hw_tmu] = gen_fixed_filter(t->filter) | (hw_tmu << 28);
/* Currently disabled! */
- r300->hw.tex.unknown1.cmd[R300_TEX_VALUE_0 + hw_tmu] = 0x0; //0x20501f80;
+ r300->hw.tex.filter_1.cmd[R300_TEX_VALUE_0 + hw_tmu] = 0x0; //0x20501f80;
r300->hw.tex.size.cmd[R300_TEX_VALUE_0 + hw_tmu] = t->size;
r300->hw.tex.format.cmd[R300_TEX_VALUE_0 + hw_tmu] = t->format;
r300->hw.tex.pitch.cmd[R300_TEX_VALUE_0 + hw_tmu] = t->pitch_reg;
WARN_ONCE("micro tiling enabled!\n");
}
- r300->hw.tex.unknown4.cmd[R300_TEX_VALUE_0 + hw_tmu] = 0x0;
+ r300->hw.tex.chroma_key.cmd[R300_TEX_VALUE_0 + hw_tmu] = 0x0;
r300->hw.tex.border_color.cmd[R300_TEX_VALUE_0 + hw_tmu] = t->pp_border_color;
last_hw_tmu = hw_tmu;
}
r300->hw.tex.filter.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_FILTER_0, last_hw_tmu + 1);
- r300->hw.tex.unknown1.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_FILTER1_0, last_hw_tmu + 1);
+ r300->hw.tex.filter_1.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_FILTER1_0, last_hw_tmu + 1);
r300->hw.tex.size.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_SIZE_0, last_hw_tmu + 1);
r300->hw.tex.format.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_FORMAT_0, last_hw_tmu + 1);
r300->hw.tex.pitch.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_PITCH_0, last_hw_tmu + 1);
r300->hw.tex.offset.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_OFFSET_0, last_hw_tmu + 1);
- r300->hw.tex.unknown4.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_CHROMA_KEY_0, last_hw_tmu + 1);
+ r300->hw.tex.chroma_key.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_CHROMA_KEY_0, last_hw_tmu + 1);
r300->hw.tex.border_color.cmd[R300_TEX_CMD_0] = cmdpacket0(R300_TX_BORDER_COLOR_0, last_hw_tmu + 1);
/* Initialize magic registers
TODO : learn what they really do, or get rid of
those we don't have to touch */
- r300->hw.unk2080.cmd[1] = 0x0030045A; //0x0030065a /* Dangerous */
+ r300->hw.vap_cntl.cmd[1] = 0x0030045A; //0x0030065a /* Dangerous */
r300->hw.vte.cmd[1] = R300_VPORT_X_SCALE_ENA
| R300_VPORT_X_OFFSET_ENA
r300->hw.unk2134.cmd[1] = 0x00FFFFFF;
r300->hw.unk2134.cmd[2] = 0x00000000;
if (_mesa_little_endian())
- r300->hw.unk2140.cmd[1] = 0x00000000;
+ r300->hw.vap_cntl_status.cmd[1] = 0x00000000;
else
- r300->hw.unk2140.cmd[1] = 0x00000002;
+ r300->hw.vap_cntl_status.cmd[1] = 0x00000002;
#if 0 /* Done in setup routing */
((drm_r300_cmd_header_t*)r300->hw.vir[0].cmd)->packet0.count = 1;
r300->hw.unk4260.cmd[2] = r300PackFloat32(0.0);
r300->hw.unk4260.cmd[3] = r300PackFloat32(1.0);
- r300->hw.unk4274.cmd[1] = 0x00000002;
+ r300->hw.shade.cmd[1] = 0x00000002;
r300ShadeModel(ctx, ctx->Light.ShadeModel);
- r300->hw.unk4274.cmd[3] = 0x00000000;
- r300->hw.unk4274.cmd[4] = 0x00000000;
+ r300->hw.shade.cmd[3] = 0x00000000;
+ r300->hw.shade.cmd[4] = 0x00000000;
r300PolygonMode(ctx, GL_FRONT, ctx->Polygon.FrontMode);
r300PolygonMode(ctx, GL_BACK, ctx->Polygon.BackMode);
- r300->hw.unk4288.cmd[2] = 0x00000001;
- r300->hw.unk4288.cmd[3] = 0x00000000;
- r300->hw.unk42A0.cmd[1] = 0x00000000;
+ r300->hw.polygon_mode.cmd[2] = 0x00000001;
+ r300->hw.polygon_mode.cmd[3] = 0x00000000;
+ r300->hw.zbias_cntl.cmd[1] = 0x00000000;
r300PolygonOffset(ctx, ctx->Polygon.OffsetFactor, ctx->Polygon.OffsetUnits);
r300Enable(ctx, GL_POLYGON_OFFSET_FILL, ctx->Polygon.OffsetFill);
#endif
r300BlendColor(ctx, ctx->Color.BlendColor);
- r300->hw.unk4E10.cmd[2] = 0;
- r300->hw.unk4E10.cmd[3] = 0;
+ r300->hw.blend_color.cmd[2] = 0;
+ r300->hw.blend_color.cmd[3] = 0;
/* Again, r300ClearBuffer uses this */
r300->hw.cb.cmd[R300_CB_OFFSET] = r300->radeon.state.color.drawOffset +
switch (ctx->Visual.depthBits) {
case 16:
- r300->hw.unk4F10.cmd[1] = R300_DEPTH_FORMAT_16BIT_INT_Z;
+ r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_16BIT_INT_Z;
break;
case 24:
- r300->hw.unk4F10.cmd[1] = R300_DEPTH_FORMAT_24BIT_INT_Z;
+ r300->hw.zstencil_format.cmd[1] = R300_DEPTH_FORMAT_24BIT_INT_Z;
break;
default:
fprintf(stderr, "Error: Unsupported depth %d... exiting\n",
}
/* z compress? */
- //r300->hw.unk4F10.cmd[1] |= R300_DEPTH_FORMAT_UNK32;
+ //r300->hw.zstencil_format.cmd[1] |= R300_DEPTH_FORMAT_UNK32;
- r300->hw.unk4F10.cmd[3] = 0x00000003;
- r300->hw.unk4F10.cmd[4] = 0x00000000;
+ r300->hw.zstencil_format.cmd[3] = 0x00000003;
+ r300->hw.zstencil_format.cmd[4] = 0x00000000;
r300->hw.zb.cmd[R300_ZB_OFFSET] =
r300->radeon.radeonScreen->depthOffset +
--- /dev/null
-#include "nvvertexec.h"
+ /**************************************************************************
+
+ Copyright (C) 2005 Aapo Tahkola.
+
+ All Rights Reserved.
+
+ Permission is hereby granted, free of charge, to any person obtaining a
+ copy of this software and associated documentation files (the "Software"),
+ to deal in the Software without restriction, including without limitation
+ on the rights to use, copy, modify, merge, publish, distribute, sub
+ license, and/or sell copies of the Software, and to permit persons to whom
+ the Software is furnished to do so, subject to the following conditions:
+
+ The above copyright notice and this permission notice (including the next
+ paragraph) shall be included in all copies or substantial portions of the
+ Software.
+
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ FITNESS FOR A PARTICULAR PURPOSE AND NON-INFRINGEMENT. IN NO EVENT SHALL
+ THE COPYRIGHT OWNER(S) AND/OR ITS SUPPLIERS BE LIABLE FOR ANY CLAIM,
+ DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR
+ OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE
+ USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+ **************************************************************************/
+
+ /*
+ * Authors:
+ * Aapo Tahkola <aet@rasterburn.org>
+ */
+ #include "glheader.h"
+ #include "macros.h"
+ #include "enums.h"
+ #include "program.h"
-#include "program_instruction.h"
++#include "shader/prog_instruction.h"
++#include "shader/prog_parameter.h"
++#include "shader/prog_statevars.h"
++#include "tnl/tnl.h"
+
+ #include "r300_context.h"
+ #include "r300_program.h"
- _mesa_init_vp_per_primitive_registers(ctx);
+
+ #if SWIZZLE_X != VSF_IN_COMPONENT_X || \
+ SWIZZLE_Y != VSF_IN_COMPONENT_Y || \
+ SWIZZLE_Z != VSF_IN_COMPONENT_Z || \
+ SWIZZLE_W != VSF_IN_COMPONENT_W || \
+ SWIZZLE_ZERO != VSF_IN_COMPONENT_ZERO || \
+ SWIZZLE_ONE != VSF_IN_COMPONENT_ONE || \
+ WRITEMASK_X != VSF_FLAG_X || \
+ WRITEMASK_Y != VSF_FLAG_Y || \
+ WRITEMASK_Z != VSF_FLAG_Z || \
+ WRITEMASK_W != VSF_FLAG_W
+ #error Cannot change these!
+ #endif
+
+ #define SCALAR_FLAG (1<<31)
+ #define FLAG_MASK (1<<31)
+ #define OP_MASK (0xf) /* we are unlikely to have more than 15 */
+ #define OPN(operator, ip) {#operator, OPCODE_##operator, ip}
+
+ static struct{
+ char *name;
+ int opcode;
+ unsigned long ip; /* number of input operands and flags */
+ }op_names[]={
+ OPN(ABS, 1),
+ OPN(ADD, 2),
+ OPN(ARL, 1|SCALAR_FLAG),
+ OPN(DP3, 2),
+ OPN(DP4, 2),
+ OPN(DPH, 2),
+ OPN(DST, 2),
+ OPN(EX2, 1|SCALAR_FLAG),
+ OPN(EXP, 1|SCALAR_FLAG),
+ OPN(FLR, 1),
+ OPN(FRC, 1),
+ OPN(LG2, 1|SCALAR_FLAG),
+ OPN(LIT, 1),
+ OPN(LOG, 1|SCALAR_FLAG),
+ OPN(MAD, 3),
+ OPN(MAX, 2),
+ OPN(MIN, 2),
+ OPN(MOV, 1),
+ OPN(MUL, 2),
+ OPN(POW, 2|SCALAR_FLAG),
+ OPN(RCP, 1|SCALAR_FLAG),
+ OPN(RSQ, 1|SCALAR_FLAG),
+ OPN(SGE, 2),
+ OPN(SLT, 2),
+ OPN(SUB, 2),
+ OPN(SWZ, 1),
+ OPN(XPD, 2),
+ OPN(RCC, 0), //extra
+ OPN(PRINT, 0),
+ OPN(END, 0),
+ };
+ #undef OPN
+
+ int r300VertexProgUpdateParams(GLcontext *ctx, struct r300_vertex_program_cont *vp, float *dst)
+ {
+ int pi;
+ struct gl_vertex_program *mesa_vp = &vp->mesa_program;
+ float *dst_o=dst;
+ struct gl_program_parameter_list *paramList;
+
+ if (mesa_vp->IsNVProgram) {
- GLint tokens[6] = { STATE_MATRIX, STATE_MVP, 0, 0, 0, STATE_MATRIX };
++ _mesa_load_tracked_matrices(ctx);
+
+ for (pi=0; pi < MAX_NV_VERTEX_PROGRAM_PARAMS; pi++) {
+ *dst++=ctx->VertexProgram.Parameters[pi][0];
+ *dst++=ctx->VertexProgram.Parameters[pi][1];
+ *dst++=ctx->VertexProgram.Parameters[pi][2];
+ *dst++=ctx->VertexProgram.Parameters[pi][3];
+ }
+ return dst - dst_o;
+ }
+
+ assert(mesa_vp->Base.Parameters);
+ _mesa_load_state_parameters(ctx, mesa_vp->Base.Parameters);
+
+ if(mesa_vp->Base.Parameters->NumParameters * 4 > VSF_MAX_FRAGMENT_LENGTH){
+ fprintf(stderr, "%s:Params exhausted\n", __FUNCTION__);
+ exit(-1);
+ }
+
+ paramList = mesa_vp->Base.Parameters;
+ for(pi=0; pi < paramList->NumParameters; pi++){
+ switch(paramList->Parameters[pi].Type){
+
+ case PROGRAM_STATE_VAR:
+ case PROGRAM_NAMED_PARAM:
+ //fprintf(stderr, "%s", vp->Parameters->Parameters[pi].Name);
+ case PROGRAM_CONSTANT:
+ *dst++=paramList->ParameterValues[pi][0];
+ *dst++=paramList->ParameterValues[pi][1];
+ *dst++=paramList->ParameterValues[pi][2];
+ *dst++=paramList->ParameterValues[pi][3];
+ break;
+
+ default: _mesa_problem(NULL, "Bad param type in %s", __FUNCTION__);
+ }
+
+ }
+
+ return dst - dst_o;
+ }
+
+ static unsigned long t_dst_mask(GLuint mask)
+ {
+ /* WRITEMASK_* is equivalent to VSF_FLAG_* */
+ return mask & VSF_FLAG_ALL;
+ }
+
+ static unsigned long t_dst_class(enum register_file file)
+ {
+
+ switch(file){
+ case PROGRAM_TEMPORARY:
+ return VSF_OUT_CLASS_TMP;
+ case PROGRAM_OUTPUT:
+ return VSF_OUT_CLASS_RESULT;
+ case PROGRAM_ADDRESS:
+ return VSF_OUT_CLASS_ADDR;
+ /*
+ case PROGRAM_INPUT:
+ case PROGRAM_LOCAL_PARAM:
+ case PROGRAM_ENV_PARAM:
+ case PROGRAM_NAMED_PARAM:
+ case PROGRAM_STATE_VAR:
+ case PROGRAM_WRITE_ONLY:
+ case PROGRAM_ADDRESS:
+ */
+ default:
+ fprintf(stderr, "problem in %s", __FUNCTION__);
+ exit(0);
+ }
+ }
+
+ static unsigned long t_dst_index(struct r300_vertex_program *vp, struct prog_dst_register *dst)
+ {
+ if(dst->File == PROGRAM_OUTPUT)
+ return vp->outputs[dst->Index];
+
+ return dst->Index;
+ }
+
+ static unsigned long t_src_class(enum register_file file)
+ {
+
+ switch(file){
+ case PROGRAM_TEMPORARY:
+ return VSF_IN_CLASS_TMP;
+
+ case PROGRAM_INPUT:
+ return VSF_IN_CLASS_ATTR;
+
+ case PROGRAM_LOCAL_PARAM:
+ case PROGRAM_ENV_PARAM:
+ case PROGRAM_NAMED_PARAM:
+ case PROGRAM_STATE_VAR:
+ return VSF_IN_CLASS_PARAM;
+ /*
+ case PROGRAM_OUTPUT:
+ case PROGRAM_WRITE_ONLY:
+ case PROGRAM_ADDRESS:
+ */
+ default:
+ fprintf(stderr, "problem in %s", __FUNCTION__);
+ exit(0);
+ }
+ }
+
+ static __inline unsigned long t_swizzle(GLubyte swizzle)
+ {
+ /* this is in fact a NOP as the Mesa SWIZZLE_* are all identical to VSF_IN_COMPONENT_* */
+ return swizzle;
+ }
+
+ #if 0
+ static void vp_dump_inputs(struct r300_vertex_program *vp, char *caller)
+ {
+ int i;
+
+ if(vp == NULL){
+ fprintf(stderr, "vp null in call to %s from %s\n", __FUNCTION__, caller);
+ return ;
+ }
+
+ fprintf(stderr, "%s:<", caller);
+ for(i=0; i < VERT_ATTRIB_MAX; i++)
+ fprintf(stderr, "%d ", vp->inputs[i]);
+ fprintf(stderr, ">\n");
+
+ }
+ #endif
+
+ static unsigned long t_src_index(struct r300_vertex_program *vp, struct prog_src_register *src)
+ {
+ int i;
+ int max_reg=-1;
+
+ if(src->File == PROGRAM_INPUT){
+ if(vp->inputs[src->Index] != -1)
+ return vp->inputs[src->Index];
+
+ for(i=0; i < VERT_ATTRIB_MAX; i++)
+ if(vp->inputs[i] > max_reg)
+ max_reg=vp->inputs[i];
+
+ vp->inputs[src->Index]=max_reg+1;
+
+ //vp_dump_inputs(vp, __FUNCTION__);
+
+ return vp->inputs[src->Index];
+ }else{
+ if (src->Index < 0) {
+ fprintf(stderr, "WARNING negative offsets for indirect addressing do not work\n");
+ return 0;
+ }
+ return src->Index;
+ }
+ }
+
+ static unsigned long t_src(struct r300_vertex_program *vp, struct prog_src_register *src)
+ {
+ /* src->NegateBase uses the NEGATE_ flags from program_instruction.h,
+ * which equal our VSF_FLAGS_ values, so it's safe to just pass it here.
+ */
+ return MAKE_VSF_SOURCE(t_src_index(vp, src),
+ t_swizzle(GET_SWZ(src->Swizzle, 0)),
+ t_swizzle(GET_SWZ(src->Swizzle, 1)),
+ t_swizzle(GET_SWZ(src->Swizzle, 2)),
+ t_swizzle(GET_SWZ(src->Swizzle, 3)),
+ t_src_class(src->File),
+ src->NegateBase) | (src->RelAddr << 4);
+ }
+
+ static unsigned long t_src_scalar(struct r300_vertex_program *vp, struct prog_src_register *src)
+ {
+
+ return MAKE_VSF_SOURCE(t_src_index(vp, src),
+ t_swizzle(GET_SWZ(src->Swizzle, 0)),
+ t_swizzle(GET_SWZ(src->Swizzle, 0)),
+ t_swizzle(GET_SWZ(src->Swizzle, 0)),
+ t_swizzle(GET_SWZ(src->Swizzle, 0)),
+ t_src_class(src->File),
+ src->NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src->RelAddr << 4);
+ }
+
+ static unsigned long t_opcode(enum prog_opcode opcode)
+ {
+
+ switch(opcode){
+ case OPCODE_ARL: return R300_VPI_OUT_OP_ARL;
+ case OPCODE_DST: return R300_VPI_OUT_OP_DST;
+ case OPCODE_EX2: return R300_VPI_OUT_OP_EX2;
+ case OPCODE_EXP: return R300_VPI_OUT_OP_EXP;
+ case OPCODE_FRC: return R300_VPI_OUT_OP_FRC;
+ case OPCODE_LG2: return R300_VPI_OUT_OP_LG2;
+ case OPCODE_LOG: return R300_VPI_OUT_OP_LOG;
+ case OPCODE_MAX: return R300_VPI_OUT_OP_MAX;
+ case OPCODE_MIN: return R300_VPI_OUT_OP_MIN;
+ case OPCODE_MUL: return R300_VPI_OUT_OP_MUL;
+ case OPCODE_RCP: return R300_VPI_OUT_OP_RCP;
+ case OPCODE_RSQ: return R300_VPI_OUT_OP_RSQ;
+ case OPCODE_SGE: return R300_VPI_OUT_OP_SGE;
+ case OPCODE_SLT: return R300_VPI_OUT_OP_SLT;
+ case OPCODE_DP4: return R300_VPI_OUT_OP_DOT;
+
+ default:
+ fprintf(stderr, "%s: Should not be called with opcode %d!", __FUNCTION__, opcode);
+ }
+ exit(-1);
+ return 0;
+ }
+
+ static unsigned long op_operands(enum prog_opcode opcode)
+ {
+ int i;
+
+ /* Can we trust mesas opcodes to be in order ? */
+ for(i=0; i < sizeof(op_names) / sizeof(*op_names); i++)
+ if(op_names[i].opcode == opcode)
+ return op_names[i].ip;
+
+ fprintf(stderr, "op %d not found in op_names\n", opcode);
+ exit(-1);
+ return 0;
+ }
+
+ static GLboolean valid_dst(struct r300_vertex_program *vp, struct prog_dst_register *dst)
+ {
+ if(dst->File == PROGRAM_OUTPUT && vp->outputs[dst->Index] == -1){
+ WARN_ONCE("Output %d not used by fragment program\n", dst->Index);
+ return GL_FALSE;
+ }else if(dst->File == PROGRAM_ADDRESS) {
+ assert(dst->Index == 0);
+ }
+
+ return GL_TRUE;
+ }
+
+ /* TODO: Get rid of t_src_class call */
+ #define CMP_SRCS(a, b) ((a.RelAddr != b.RelAddr) || (a.Index != b.Index && \
+ ((t_src_class(a.File) == VSF_IN_CLASS_PARAM && \
+ t_src_class(b.File) == VSF_IN_CLASS_PARAM) || \
+ (t_src_class(a.File) == VSF_IN_CLASS_ATTR && \
+ t_src_class(b.File) == VSF_IN_CLASS_ATTR)))) \
+
+ #define ZERO_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \
+ SWIZZLE_ZERO, SWIZZLE_ZERO, \
+ SWIZZLE_ZERO, SWIZZLE_ZERO, \
+ t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4))
+
+ #define ZERO_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \
+ SWIZZLE_ZERO, SWIZZLE_ZERO, \
+ SWIZZLE_ZERO, SWIZZLE_ZERO, \
+ t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4))
+
+ #define ZERO_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \
+ SWIZZLE_ZERO, SWIZZLE_ZERO, \
+ SWIZZLE_ZERO, SWIZZLE_ZERO, \
+ t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4))
+
+ #define ONE_SRC_0 (MAKE_VSF_SOURCE(t_src_index(vp, &src[0]), \
+ SWIZZLE_ONE, SWIZZLE_ONE, \
+ SWIZZLE_ONE, SWIZZLE_ONE, \
+ t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4))
+
+ #define ONE_SRC_1 (MAKE_VSF_SOURCE(t_src_index(vp, &src[1]), \
+ SWIZZLE_ONE, SWIZZLE_ONE, \
+ SWIZZLE_ONE, SWIZZLE_ONE, \
+ t_src_class(src[1].File), VSF_FLAG_NONE) | (src[1].RelAddr << 4))
+
+ #define ONE_SRC_2 (MAKE_VSF_SOURCE(t_src_index(vp, &src[2]), \
+ SWIZZLE_ONE, SWIZZLE_ONE, \
+ SWIZZLE_ONE, SWIZZLE_ONE, \
+ t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4))
+
+ /* DP4 version seems to trigger some hw peculiarity */
+ //#define PREFER_DP4
+
+ #define FREE_TEMPS() \
+ do { \
+ if(u_temp_i < vp->num_temporaries) { \
+ WARN_ONCE("Ran out of temps, num temps %d, us %d\n", vp->num_temporaries, u_temp_i); \
+ vp->native = GL_FALSE; \
+ } \
+ u_temp_i=VSF_MAX_FRAGMENT_TEMPS-1; \
+ } while (0)
+
+ static void r300_translate_vertex_shader(struct r300_vertex_program *vp, struct prog_instruction *vpi)
+ {
+ int i, cur_reg=0;
+ VERTEX_SHADER_INSTRUCTION *o_inst;
+ unsigned long operands;
+ int are_srcs_scalar;
+ unsigned long hw_op;
+ /* Initial value should be last tmp reg that hw supports.
+ Strangely enough r300 doesnt mind even though these would be out of range.
+ Smart enough to realize that it doesnt need it? */
+ int u_temp_i=VSF_MAX_FRAGMENT_TEMPS-1;
+ struct prog_src_register src[3];
+
+ vp->pos_end=0; /* Not supported yet */
+ vp->program.length=0;
+ /*vp->num_temporaries=mesa_vp->Base.NumTemporaries;*/
+
+ for(i=0; i < VERT_ATTRIB_MAX; i++)
+ vp->inputs[i] = -1;
+
+ for(i=0; i < VERT_RESULT_MAX; i++)
+ vp->outputs[i] = -1;
+
+ assert(vp->key.OutputsWritten & (1 << VERT_RESULT_HPOS));
+
+ /* Assign outputs */
+ if(vp->key.OutputsWritten & (1 << VERT_RESULT_HPOS))
+ vp->outputs[VERT_RESULT_HPOS] = cur_reg++;
+
+ if(vp->key.OutputsWritten & (1 << VERT_RESULT_PSIZ))
+ vp->outputs[VERT_RESULT_PSIZ] = cur_reg++;
+
+ if(vp->key.OutputsWritten & (1 << VERT_RESULT_COL0))
+ vp->outputs[VERT_RESULT_COL0] = cur_reg++;
+
+ if(vp->key.OutputsWritten & (1 << VERT_RESULT_COL1))
+ vp->outputs[VERT_RESULT_COL1] = cur_reg++;
+
+ #if 0 /* Not supported yet */
+ if(vp->key.OutputsWritten & (1 << VERT_RESULT_BFC0))
+ vp->outputs[VERT_RESULT_BFC0] = cur_reg++;
+
+ if(vp->key.OutputsWritten & (1 << VERT_RESULT_BFC1))
+ vp->outputs[VERT_RESULT_BFC1] = cur_reg++;
+
+ if(vp->key.OutputsWritten & (1 << VERT_RESULT_FOGC))
+ vp->outputs[VERT_RESULT_FOGC] = cur_reg++;
+ #endif
+
+ for(i=VERT_RESULT_TEX0; i <= VERT_RESULT_TEX7; i++)
+ if(vp->key.OutputsWritten & (1 << i))
+ vp->outputs[i] = cur_reg++;
+
+ vp->translated = GL_TRUE;
+ vp->native = GL_TRUE;
+
+ o_inst=vp->program.body.i;
+ for(; vpi->Opcode != OPCODE_END; vpi++, o_inst++){
+ FREE_TEMPS();
+
+ if(!valid_dst(vp, &vpi->DstReg))
+ {
+ /* redirect result to unused temp */
+ vpi->DstReg.File = PROGRAM_TEMPORARY;
+ vpi->DstReg.Index = u_temp_i;
+ }
+
+ operands=op_operands(vpi->Opcode);
+ are_srcs_scalar=operands & SCALAR_FLAG;
+ operands &= OP_MASK;
+
+ for(i=0; i < operands; i++)
+ src[i]=vpi->SrcReg[i];
+
+ if(operands == 3){ /* TODO: scalars */
+ if( CMP_SRCS(src[1], src[2]) || CMP_SRCS(src[0], src[2]) ){
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, u_temp_i,
+ VSF_FLAG_ALL, VSF_OUT_CLASS_TMP);
+
+ o_inst->src1=MAKE_VSF_SOURCE(t_src_index(vp, &src[2]),
+ SWIZZLE_X, SWIZZLE_Y,
+ SWIZZLE_Z, SWIZZLE_W,
+ t_src_class(src[2].File), VSF_FLAG_NONE) | (src[2].RelAddr << 4);
+
+ o_inst->src2=ZERO_SRC_2;
+ o_inst->src3=ZERO_SRC_2;
+ o_inst++;
+
+ src[2].File=PROGRAM_TEMPORARY;
+ src[2].Index=u_temp_i;
+ src[2].RelAddr=0;
+ u_temp_i--;
+ }
+
+ }
+
+ if(operands >= 2){
+ if( CMP_SRCS(src[1], src[0]) ){
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, u_temp_i,
+ VSF_FLAG_ALL, VSF_OUT_CLASS_TMP);
+
+ o_inst->src1=MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+ SWIZZLE_X, SWIZZLE_Y,
+ SWIZZLE_Z, SWIZZLE_W,
+ t_src_class(src[0].File), VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+
+ o_inst->src2=ZERO_SRC_0;
+ o_inst->src3=ZERO_SRC_0;
+ o_inst++;
+
+ src[0].File=PROGRAM_TEMPORARY;
+ src[0].Index=u_temp_i;
+ src[0].RelAddr=0;
+ u_temp_i--;
+ }
+ }
+
+ /* These ops need special handling. */
+ switch(vpi->Opcode){
+ case OPCODE_POW:
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_POW, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+ o_inst->src1=t_src_scalar(vp, &src[0]);
+ o_inst->src2=ZERO_SRC_0;
+ o_inst->src3=t_src_scalar(vp, &src[1]);
+ goto next;
+
+ case OPCODE_MOV://ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{} {ZERO ZERO ZERO ZERO}
+ case OPCODE_SWZ:
+ #if 1
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=ZERO_SRC_0;
+ o_inst->src3=ZERO_SRC_0;
+ #else
+ hw_op=(src[0].File == PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : R300_VPI_OUT_OP_MAD;
+
+ o_inst->op=MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=ONE_SRC_0;
+ o_inst->src3=ZERO_SRC_0;
+ #endif
+
+ goto next;
+
+ case OPCODE_ADD:
+ #if 1
+ hw_op=(src[0].File == PROGRAM_TEMPORARY &&
+ src[1].File == PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : R300_VPI_OUT_OP_MAD;
+
+ o_inst->op=MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+ o_inst->src1=ONE_SRC_0;
+ o_inst->src2=t_src(vp, &src[0]);
+ o_inst->src3=t_src(vp, &src[1]);
+ #else
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=t_src(vp, &src[1]);
+ o_inst->src3=ZERO_SRC_1;
+
+ #endif
+ goto next;
+
+ case OPCODE_MAD:
+ hw_op=(src[0].File == PROGRAM_TEMPORARY &&
+ src[1].File == PROGRAM_TEMPORARY &&
+ src[2].File == PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : R300_VPI_OUT_OP_MAD;
+
+ o_inst->op=MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=t_src(vp, &src[1]);
+ o_inst->src3=t_src(vp, &src[2]);
+ goto next;
+
+ case OPCODE_MUL: /* HW mul can take third arg but appears to have some other limitations. */
+ hw_op=(src[0].File == PROGRAM_TEMPORARY &&
+ src[1].File == PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : R300_VPI_OUT_OP_MAD;
+
+ o_inst->op=MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=t_src(vp, &src[1]);
+
+ o_inst->src3=ZERO_SRC_1;
+ goto next;
+
+ case OPCODE_DP3://DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ZERO} PARAM 0{} {X Y Z ZERO}
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+
+ o_inst->src1=MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
+ SWIZZLE_ZERO,
+ t_src_class(src[0].File),
+ src[0].NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+
+ o_inst->src2=MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
+ SWIZZLE_ZERO,
+ t_src_class(src[1].File),
+ src[1].NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+
+ o_inst->src3=ZERO_SRC_1;
+ goto next;
+
+ case OPCODE_SUB://ADD RESULT 1.X Y Z W TMP 0{} {X Y Z W} PARAM 1{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W
+ #if 1
+ hw_op=(src[0].File == PROGRAM_TEMPORARY &&
+ src[1].File == PROGRAM_TEMPORARY) ? R300_VPI_OUT_OP_MAD_2 : R300_VPI_OUT_OP_MAD;
+
+ o_inst->op=MAKE_VSF_OP(hw_op, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=ONE_SRC_0;
+ o_inst->src3=MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
+ t_src_class(src[1].File),
+ (!src[1].NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+ #else
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 0)),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 1)),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 2)),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 3)),
+ t_src_class(src[1].File),
+ (!src[1].NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+ o_inst->src3=0;
+ #endif
+ goto next;
+
+ case OPCODE_ABS://MAX RESULT 1.X Y Z W PARAM 0{} {X Y Z W} PARAM 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_MAX, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 3)),
+ t_src_class(src[0].File),
+ (!src[0].NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+ o_inst->src3=0;
+ goto next;
+
+ case OPCODE_FLR:
+ /* FRC TMP 0.X Y Z W PARAM 0{} {X Y Z W}
+ ADD RESULT 1.X Y Z W PARAM 0{} {X Y Z W} TMP 0{X Y Z W } {X Y Z W} neg Xneg Yneg Zneg W */
+
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_FRC, u_temp_i,
+ t_dst_mask(vpi->DstReg.WriteMask), VSF_OUT_CLASS_TMP);
+
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=ZERO_SRC_0;
+ o_inst->src3=ZERO_SRC_0;
+ o_inst++;
+
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_ADD, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=MAKE_VSF_SOURCE(u_temp_i,
+ VSF_IN_COMPONENT_X,
+ VSF_IN_COMPONENT_Y,
+ VSF_IN_COMPONENT_Z,
+ VSF_IN_COMPONENT_W,
+ VSF_IN_CLASS_TMP,
+ /* Not 100% sure about this */
+ (!src[0].NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE/*VSF_FLAG_ALL*/);
+
+ o_inst->src3=ZERO_SRC_0;
+ u_temp_i--;
+ goto next;
+
+ case OPCODE_LG2:// LG2 RESULT 1.X Y Z W PARAM 0{} {X X X X}
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_LG2, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+
+ o_inst->src1=MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+ t_src_class(src[0].File),
+ src[0].NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+ o_inst->src2=ZERO_SRC_0;
+ o_inst->src3=ZERO_SRC_0;
+ goto next;
+
+ case OPCODE_LIT://LIT TMP 1.Y Z TMP 1{} {X W Z Y} TMP 1{} {Y W Z X} TMP 1{} {Y X Z W}
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_LIT, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+ /* NOTE: Users swizzling might not work. */
+ o_inst->src1=MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x
+ t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w
+ VSF_IN_COMPONENT_ZERO, // z
+ t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y
+ t_src_class(src[0].File),
+ src[0].NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+ o_inst->src2=MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y
+ t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w
+ VSF_IN_COMPONENT_ZERO, // z
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x
+ t_src_class(src[0].File),
+ src[0].NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+ o_inst->src3=MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x
+ VSF_IN_COMPONENT_ZERO, // z
+ t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w
+ t_src_class(src[0].File),
+ src[0].NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+ goto next;
+
+ case OPCODE_DPH://DOT RESULT 1.X Y Z W PARAM 0{} {X Y Z ONE} PARAM 0{} {X Y Z W}
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_DOT, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+
+ o_inst->src1=MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 1)),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 2)),
+ VSF_IN_COMPONENT_ONE,
+ t_src_class(src[0].File),
+ src[0].NegateBase ? VSF_FLAG_XYZ : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+ o_inst->src2=t_src(vp, &src[1]);
+ o_inst->src3=ZERO_SRC_1;
+ goto next;
+
+ case OPCODE_XPD:
+ /* mul r0, r1.yzxw, r2.zxyw
+ mad r0, -r2.yzxw, r1.zxyw, r0
+ NOTE: might need MAD_2
+ */
+
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, u_temp_i,
+ t_dst_mask(vpi->DstReg.WriteMask), VSF_OUT_CLASS_TMP);
+
+ o_inst->src1=MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y
+ t_swizzle(GET_SWZ(src[0].Swizzle, 2)), // z
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x
+ t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w
+ t_src_class(src[0].File),
+ src[0].NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+
+ o_inst->src2=MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 2)), // z
+ t_swizzle(GET_SWZ(src[1].Swizzle, 0)), // x
+ t_swizzle(GET_SWZ(src[1].Swizzle, 1)), // y
+ t_swizzle(GET_SWZ(src[1].Swizzle, 3)), // w
+ t_src_class(src[1].File),
+ src[1].NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+
+ o_inst->src3=ZERO_SRC_1;
+ o_inst++;
+ u_temp_i--;
+
+ o_inst->op=MAKE_VSF_OP(R300_VPI_OUT_OP_MAD, t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+
+ o_inst->src1=MAKE_VSF_SOURCE(t_src_index(vp, &src[1]),
+ t_swizzle(GET_SWZ(src[1].Swizzle, 1)), // y
+ t_swizzle(GET_SWZ(src[1].Swizzle, 2)), // z
+ t_swizzle(GET_SWZ(src[1].Swizzle, 0)), // x
+ t_swizzle(GET_SWZ(src[1].Swizzle, 3)), // w
+ t_src_class(src[1].File),
+ (!src[1].NegateBase) ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[1].RelAddr << 4);
+
+ o_inst->src2=MAKE_VSF_SOURCE(t_src_index(vp, &src[0]),
+ t_swizzle(GET_SWZ(src[0].Swizzle, 2)), // z
+ t_swizzle(GET_SWZ(src[0].Swizzle, 0)), // x
+ t_swizzle(GET_SWZ(src[0].Swizzle, 1)), // y
+ t_swizzle(GET_SWZ(src[0].Swizzle, 3)), // w
+ t_src_class(src[0].File),
+ src[0].NegateBase ? VSF_FLAG_ALL : VSF_FLAG_NONE) | (src[0].RelAddr << 4);
+
+ o_inst->src3=MAKE_VSF_SOURCE(u_temp_i+1,
+ VSF_IN_COMPONENT_X,
+ VSF_IN_COMPONENT_Y,
+ VSF_IN_COMPONENT_Z,
+ VSF_IN_COMPONENT_W,
+ VSF_IN_CLASS_TMP,
+ VSF_FLAG_NONE);
+
+ goto next;
+
+ case OPCODE_RCC:
+ fprintf(stderr, "Dont know how to handle op %d yet\n", vpi->Opcode);
+ exit(-1);
+ break;
+ case OPCODE_END:
+ break;
+ default:
+ break;
+ }
+
+ o_inst->op=MAKE_VSF_OP(t_opcode(vpi->Opcode), t_dst_index(vp, &vpi->DstReg),
+ t_dst_mask(vpi->DstReg.WriteMask), t_dst_class(vpi->DstReg.File));
+
+ if(are_srcs_scalar){
+ switch(operands){
+ case 1:
+ o_inst->src1=t_src_scalar(vp, &src[0]);
+ o_inst->src2=ZERO_SRC_0;
+ o_inst->src3=ZERO_SRC_0;
+ break;
+
+ case 2:
+ o_inst->src1=t_src_scalar(vp, &src[0]);
+ o_inst->src2=t_src_scalar(vp, &src[1]);
+ o_inst->src3=ZERO_SRC_1;
+ break;
+
+ case 3:
+ o_inst->src1=t_src_scalar(vp, &src[0]);
+ o_inst->src2=t_src_scalar(vp, &src[1]);
+ o_inst->src3=t_src_scalar(vp, &src[2]);
+ break;
+
+ default:
+ fprintf(stderr, "scalars and op RCC not handled yet");
+ exit(-1);
+ break;
+ }
+ }else{
+ switch(operands){
+ case 1:
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=ZERO_SRC_0;
+ o_inst->src3=ZERO_SRC_0;
+ break;
+
+ case 2:
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=t_src(vp, &src[1]);
+ o_inst->src3=ZERO_SRC_1;
+ break;
+
+ case 3:
+ o_inst->src1=t_src(vp, &src[0]);
+ o_inst->src2=t_src(vp, &src[1]);
+ o_inst->src3=t_src(vp, &src[2]);
+ break;
+
+ default:
+ fprintf(stderr, "scalars and op RCC not handled yet");
+ exit(-1);
+ break;
+ }
+ }
+ next: ;
+ }
+
+ /* Will most likely segfault before we get here... fix later. */
+ if(o_inst - vp->program.body.i >= VSF_MAX_FRAGMENT_LENGTH/4) {
+ vp->program.length = 0;
+ vp->native = GL_FALSE;
+ return ;
+ }
+ vp->program.length=(o_inst - vp->program.body.i) * 4;
+ #if 0
+ fprintf(stderr, "hw program:\n");
+ for(i=0; i < vp->program.length; i++)
+ fprintf(stderr, "%08x\n", vp->program.body.d[i]);
+ #endif
+ }
+
+ static void position_invariant(struct gl_program *prog)
+ {
+ struct prog_instruction *vpi;
+ struct gl_program_parameter_list *paramList;
+ int i;
+
++ gl_state_index tokens[STATE_LENGTH] = { STATE_MVP_MATRIX, 0, 0, 0, 0 };
+
+ #ifdef PREFER_DP4
+ tokens[5] = STATE_MATRIX;
+ #else
+ tokens[5] = STATE_MATRIX_TRANSPOSE;
+ #endif
+ paramList = prog->Parameters;
+
+ vpi = _mesa_alloc_instructions (prog->NumInstructions + 4);
+ _mesa_init_instructions (vpi, prog->NumInstructions + 4);
+
+ for (i=0; i < 4; i++) {
+ GLint idx;
+ tokens[3] = tokens[4] = i;
+ idx = _mesa_add_state_reference(paramList, tokens);
+ #ifdef PREFER_DP4
+ vpi[i].Opcode = OPCODE_DP4;
+ vpi[i].StringPos = 0;
+ vpi[i].Data = 0;
+
+ vpi[i].DstReg.File = PROGRAM_OUTPUT;
+ vpi[i].DstReg.Index = VERT_RESULT_HPOS;
+ vpi[i].DstReg.WriteMask = 1 << i;
+ vpi[i].DstReg.CondMask = COND_TR;
+
+ vpi[i].SrcReg[0].File = PROGRAM_STATE_VAR;
+ vpi[i].SrcReg[0].Index = idx;
+ vpi[i].SrcReg[0].Swizzle = SWIZZLE_XYZW;
+
+ vpi[i].SrcReg[1].File = PROGRAM_INPUT;
+ vpi[i].SrcReg[1].Index = VERT_ATTRIB_POS;
+ vpi[i].SrcReg[1].Swizzle = SWIZZLE_XYZW;
+ #else
+ if (i == 0)
+ vpi[i].Opcode = OPCODE_MUL;
+ else
+ vpi[i].Opcode = OPCODE_MAD;
+
+ vpi[i].StringPos = 0;
+ vpi[i].Data = 0;
+
+ if (i == 3)
+ vpi[i].DstReg.File = PROGRAM_OUTPUT;
+ else
+ vpi[i].DstReg.File = PROGRAM_TEMPORARY;
+ vpi[i].DstReg.Index = 0;
+ vpi[i].DstReg.WriteMask = 0xf;
+ vpi[i].DstReg.CondMask = COND_TR;
+
+ vpi[i].SrcReg[0].File = PROGRAM_STATE_VAR;
+ vpi[i].SrcReg[0].Index = idx;
+ vpi[i].SrcReg[0].Swizzle = SWIZZLE_XYZW;
+
+ vpi[i].SrcReg[1].File = PROGRAM_INPUT;
+ vpi[i].SrcReg[1].Index = VERT_ATTRIB_POS;
+ vpi[i].SrcReg[1].Swizzle = MAKE_SWIZZLE4(i, i, i, i);
+
+ if (i > 0) {
+ vpi[i].SrcReg[2].File = PROGRAM_TEMPORARY;
+ vpi[i].SrcReg[2].Index = 0;
+ vpi[i].SrcReg[2].Swizzle = SWIZZLE_XYZW;
+ }
+ #endif
+ }
+
+ _mesa_copy_instructions (&vpi[i], prog->Instructions, prog->NumInstructions);
+
+ free(prog->Instructions);
+
+ prog->Instructions = vpi;
+
+ prog->NumInstructions += 4;
+ vpi = &prog->Instructions[prog->NumInstructions-1];
+
+ assert(vpi->Opcode == OPCODE_END);
+ }
+
+ static void insert_wpos(struct r300_vertex_program *vp,
+ struct gl_program *prog,
+ GLuint temp_index)
+ {
+ struct prog_instruction *vpi;
+ struct prog_instruction *vpi_insert;
+ int i = 0;
+
+ vpi = _mesa_alloc_instructions (prog->NumInstructions + 2);
+ _mesa_init_instructions (vpi, prog->NumInstructions + 2);
+ /* all but END */
+ _mesa_copy_instructions (vpi, prog->Instructions, prog->NumInstructions - 1);
+ /* END */
+ _mesa_copy_instructions (&vpi[prog->NumInstructions + 1],
+ &prog->Instructions[prog->NumInstructions - 1],
+ 1);
+ vpi_insert = &vpi[prog->NumInstructions - 1];
+
+ vpi_insert[i].Opcode = OPCODE_MOV;
+
+ vpi_insert[i].DstReg.File = PROGRAM_OUTPUT;
+ vpi_insert[i].DstReg.Index = VERT_RESULT_HPOS;
+ vpi_insert[i].DstReg.WriteMask = WRITEMASK_XYZW;
+ vpi_insert[i].DstReg.CondMask = COND_TR;
+
+ vpi_insert[i].SrcReg[0].File = PROGRAM_TEMPORARY;
+ vpi_insert[i].SrcReg[0].Index = temp_index;
+ vpi_insert[i].SrcReg[0].Swizzle = SWIZZLE_XYZW;
+ i++;
+
+ vpi_insert[i].Opcode = OPCODE_MOV;
+
+ vpi_insert[i].DstReg.File = PROGRAM_OUTPUT;
+ vpi_insert[i].DstReg.Index = VERT_RESULT_TEX0+vp->wpos_idx;
+ vpi_insert[i].DstReg.WriteMask = WRITEMASK_XYZW;
+ vpi_insert[i].DstReg.CondMask = COND_TR;
+
+ vpi_insert[i].SrcReg[0].File = PROGRAM_TEMPORARY;
+ vpi_insert[i].SrcReg[0].Index = temp_index;
+ vpi_insert[i].SrcReg[0].Swizzle = SWIZZLE_XYZW;
+ i++;
+
+ free(prog->Instructions);
+
+ prog->Instructions = vpi;
+
+ prog->NumInstructions += i;
+ vpi = &prog->Instructions[prog->NumInstructions-1];
+
+ assert(vpi->Opcode == OPCODE_END);
+ }
+
+ static void pos_as_texcoord(struct r300_vertex_program *vp,
+ struct gl_program *prog)
+ {
+ struct prog_instruction *vpi;
+ GLuint tempregi = prog->NumTemporaries;
+ /* should do something else if no temps left... */
+ prog->NumTemporaries++;
+
+ for(vpi = prog->Instructions; vpi->Opcode != OPCODE_END; vpi++){
+ if( vpi->DstReg.File == PROGRAM_OUTPUT &&
+ vpi->DstReg.Index == VERT_RESULT_HPOS ){
+ vpi->DstReg.File = PROGRAM_TEMPORARY;
+ vpi->DstReg.Index = tempregi;
+ }
+ }
+ insert_wpos(vp, prog, tempregi);
+ }
+
+ static struct r300_vertex_program *build_program(struct r300_vertex_program_key *wanted_key,
+ struct gl_vertex_program *mesa_vp,
+ GLint wpos_idx)
+ {
+ struct r300_vertex_program *vp;
+
+ vp = _mesa_calloc(sizeof(*vp));
+ _mesa_memcpy(&vp->key, wanted_key, sizeof(vp->key));
+
+ vp->wpos_idx = wpos_idx;
+
+ if(mesa_vp->IsPositionInvariant) {
+ position_invariant(&mesa_vp->Base);
+ }
+
+ if(wpos_idx > -1)
+ pos_as_texcoord(vp, &mesa_vp->Base);
+
+ assert(mesa_vp->Base.NumInstructions);
+
+ vp->num_temporaries=mesa_vp->Base.NumTemporaries;
+
+ r300_translate_vertex_shader(vp, mesa_vp->Base.Instructions);
+
+ return vp;
+ }
+
+ void r300_select_vertex_shader(r300ContextPtr r300)
+ {
+ GLcontext *ctx = ctx = r300->radeon.glCtx;
+ GLuint InputsRead;
+ struct r300_vertex_program_key wanted_key = { 0 };
+ GLint i;
+ struct r300_vertex_program_cont *vpc;
+ struct r300_vertex_program *vp;
+ GLint wpos_idx;
+
+ vpc = (struct r300_vertex_program_cont *)ctx->VertexProgram._Current;
+ InputsRead = ctx->FragmentProgram._Current->Base.InputsRead;
+
+ wanted_key.OutputsWritten |= 1 << VERT_RESULT_HPOS;
+
+ wpos_idx = -1;
+ if (InputsRead & FRAG_BIT_WPOS){
+ for (i = 0; i < ctx->Const.MaxTextureUnits; i++)
+ if (!(InputsRead & (FRAG_BIT_TEX0 << i)))
+ break;
+
+ if(i == ctx->Const.MaxTextureUnits){
+ fprintf(stderr, "\tno free texcoord found\n");
+ exit(0);
+ }
+
+ InputsRead |= (FRAG_BIT_TEX0 << i);
+ wpos_idx = i;
+ }
+
+ if (InputsRead & FRAG_BIT_COL0)
+ wanted_key.OutputsWritten |= 1 << VERT_RESULT_COL0;
+
+ if ((InputsRead & FRAG_BIT_COL1) /*||
+ (InputsRead & FRAG_BIT_FOGC)*/)
+ wanted_key.OutputsWritten |= 1 << VERT_RESULT_COL1;
+
+ for (i = 0; i < ctx->Const.MaxTextureUnits; i++)
+ if (InputsRead & (FRAG_BIT_TEX0 << i))
+ wanted_key.OutputsWritten |= 1 << (VERT_RESULT_TEX0 + i);
+
+ wanted_key.InputsRead = vpc->mesa_program.Base.InputsRead;
+ if(vpc->mesa_program.IsPositionInvariant) {
+ /* we wan't position don't we ? */
+ wanted_key.InputsRead |= (1 << VERT_ATTRIB_POS);
+ }
+
+ for (vp = vpc->progs; vp; vp = vp->next)
+ if (_mesa_memcmp(&vp->key, &wanted_key, sizeof(wanted_key)) == 0) {
+ r300->selected_vp = vp;
+ return ;
+ }
+
+ //_mesa_print_program(&vpc->mesa_program.Base);
+
+ vp = build_program(&wanted_key, &vpc->mesa_program, wpos_idx);
+ vp->next = vpc->progs;
+ vpc->progs = vp;
+ r300->selected_vp = vp;
+ }
/*
* Mesa 3-D graphics library
- * Version: 6.5.2
+ * Version: 6.5.3
*
- * Copyright (C) 1999-2006 Brian Paul All Rights Reserved.
+ * Copyright (C) 1999-2007 Brian Paul All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
GLsizei width, GLenum format, GLenum type,
const GLvoid *data )
{
+ static const GLfloat one[4] = { 1.0, 1.0, 1.0, 1.0 };
+ static const GLfloat zero[4] = { 0.0, 0.0, 0.0, 0.0 };
GET_CURRENT_CONTEXT(ctx);
struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
struct gl_texture_object *texObj = NULL;
struct gl_color_table *table = NULL;
GLboolean proxy = GL_FALSE;
GLint baseFormat;
- GLfloat rScale = 1.0, gScale = 1.0, bScale = 1.0, aScale = 1.0;
- GLfloat rBias = 0.0, gBias = 0.0, bBias = 0.0, aBias = 0.0;
+ const GLfloat *scale = one, *bias = zero;
GLint comps;
+
ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx); /* too complex */
switch (target) {
table = &ctx->Texture.Palette;
break;
case GL_COLOR_TABLE:
- table = &ctx->ColorTable;
- rScale = ctx->Pixel.ColorTableScale[0];
- gScale = ctx->Pixel.ColorTableScale[1];
- bScale = ctx->Pixel.ColorTableScale[2];
- aScale = ctx->Pixel.ColorTableScale[3];
- rBias = ctx->Pixel.ColorTableBias[0];
- gBias = ctx->Pixel.ColorTableBias[1];
- bBias = ctx->Pixel.ColorTableBias[2];
- aBias = ctx->Pixel.ColorTableBias[3];
+ table = &ctx->ColorTable[COLORTABLE_PRECONVOLUTION];
+ scale = ctx->Pixel.ColorTableScale[COLORTABLE_PRECONVOLUTION];
+ bias = ctx->Pixel.ColorTableBias[COLORTABLE_PRECONVOLUTION];
break;
case GL_PROXY_COLOR_TABLE:
- table = &ctx->ProxyColorTable;
+ table = &ctx->ProxyColorTable[COLORTABLE_PRECONVOLUTION];
proxy = GL_TRUE;
break;
case GL_TEXTURE_COLOR_TABLE_SGI:
return;
}
table = &(texUnit->ColorTable);
- rScale = ctx->Pixel.TextureColorTableScale[0];
- gScale = ctx->Pixel.TextureColorTableScale[1];
- bScale = ctx->Pixel.TextureColorTableScale[2];
- aScale = ctx->Pixel.TextureColorTableScale[3];
- rBias = ctx->Pixel.TextureColorTableBias[0];
- gBias = ctx->Pixel.TextureColorTableBias[1];
- bBias = ctx->Pixel.TextureColorTableBias[2];
- aBias = ctx->Pixel.TextureColorTableBias[3];
+ scale = ctx->Pixel.TextureColorTableScale;
+ bias = ctx->Pixel.TextureColorTableBias;
break;
case GL_PROXY_TEXTURE_COLOR_TABLE_SGI:
if (!ctx->Extensions.SGI_texture_color_table) {
proxy = GL_TRUE;
break;
case GL_POST_CONVOLUTION_COLOR_TABLE:
- table = &ctx->PostConvolutionColorTable;
- rScale = ctx->Pixel.PCCTscale[0];
- gScale = ctx->Pixel.PCCTscale[1];
- bScale = ctx->Pixel.PCCTscale[2];
- aScale = ctx->Pixel.PCCTscale[3];
- rBias = ctx->Pixel.PCCTbias[0];
- gBias = ctx->Pixel.PCCTbias[1];
- bBias = ctx->Pixel.PCCTbias[2];
- aBias = ctx->Pixel.PCCTbias[3];
+ table = &ctx->ColorTable[COLORTABLE_POSTCONVOLUTION];
+ scale = ctx->Pixel.ColorTableScale[COLORTABLE_POSTCONVOLUTION];
+ bias = ctx->Pixel.ColorTableBias[COLORTABLE_POSTCONVOLUTION];
break;
case GL_PROXY_POST_CONVOLUTION_COLOR_TABLE:
- table = &ctx->ProxyPostConvolutionColorTable;
+ table = &ctx->ProxyColorTable[COLORTABLE_POSTCONVOLUTION];
proxy = GL_TRUE;
break;
case GL_POST_COLOR_MATRIX_COLOR_TABLE:
- table = &ctx->PostColorMatrixColorTable;
- rScale = ctx->Pixel.PCMCTscale[0];
- gScale = ctx->Pixel.PCMCTscale[1];
- bScale = ctx->Pixel.PCMCTscale[2];
- aScale = ctx->Pixel.PCMCTscale[3];
- rBias = ctx->Pixel.PCMCTbias[0];
- gBias = ctx->Pixel.PCMCTbias[1];
- bBias = ctx->Pixel.PCMCTbias[2];
- aBias = ctx->Pixel.PCMCTbias[3];
+ table = &ctx->ColorTable[COLORTABLE_POSTCOLORMATRIX];
+ scale = ctx->Pixel.ColorTableScale[COLORTABLE_POSTCOLORMATRIX];
+ bias = ctx->Pixel.ColorTableBias[COLORTABLE_POSTCOLORMATRIX];
break;
case GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE:
- table = &ctx->ProxyPostColorMatrixColorTable;
+ table = &ctx->ProxyColorTable[COLORTABLE_POSTCOLORMATRIX];
proxy = GL_TRUE;
break;
default:
store_colortable_entries(ctx, table,
0, width, /* start, count */
format, type, data,
- rScale, rBias,
- gScale, gBias,
- bScale, bBias,
- aScale, aBias);
+ scale[0], bias[0],
+ scale[1], bias[1],
+ scale[2], bias[2],
+ scale[3], bias[3]);
}
} /* proxy */
GLsizei count, GLenum format, GLenum type,
const GLvoid *data )
{
+ static const GLfloat one[4] = { 1.0, 1.0, 1.0, 1.0 };
+ static const GLfloat zero[4] = { 0.0, 0.0, 0.0, 0.0 };
GET_CURRENT_CONTEXT(ctx);
struct gl_texture_unit *texUnit = &ctx->Texture.Unit[ctx->Texture.CurrentUnit];
struct gl_texture_object *texObj = NULL;
struct gl_color_table *table = NULL;
- GLfloat rScale = 1.0, gScale = 1.0, bScale = 1.0, aScale = 1.0;
- GLfloat rBias = 0.0, gBias = 0.0, bBias = 0.0, aBias = 0.0;
+ const GLfloat *scale = one, *bias = zero;
+
ASSERT_OUTSIDE_BEGIN_END_AND_FLUSH(ctx);
switch (target) {
table = &ctx->Texture.Palette;
break;
case GL_COLOR_TABLE:
- table = &ctx->ColorTable;
- rScale = ctx->Pixel.ColorTableScale[0];
- gScale = ctx->Pixel.ColorTableScale[1];
- bScale = ctx->Pixel.ColorTableScale[2];
- aScale = ctx->Pixel.ColorTableScale[3];
- rBias = ctx->Pixel.ColorTableBias[0];
- gBias = ctx->Pixel.ColorTableBias[1];
- bBias = ctx->Pixel.ColorTableBias[2];
- aBias = ctx->Pixel.ColorTableBias[3];
+ table = &ctx->ColorTable[COLORTABLE_PRECONVOLUTION];
+ scale = ctx->Pixel.ColorTableScale[COLORTABLE_PRECONVOLUTION];
+ bias = ctx->Pixel.ColorTableBias[COLORTABLE_PRECONVOLUTION];
break;
case GL_TEXTURE_COLOR_TABLE_SGI:
if (!ctx->Extensions.SGI_texture_color_table) {
return;
}
table = &(texUnit->ColorTable);
- rScale = ctx->Pixel.TextureColorTableScale[0];
- gScale = ctx->Pixel.TextureColorTableScale[1];
- bScale = ctx->Pixel.TextureColorTableScale[2];
- aScale = ctx->Pixel.TextureColorTableScale[3];
- rBias = ctx->Pixel.TextureColorTableBias[0];
- gBias = ctx->Pixel.TextureColorTableBias[1];
- bBias = ctx->Pixel.TextureColorTableBias[2];
- aBias = ctx->Pixel.TextureColorTableBias[3];
+ scale = ctx->Pixel.TextureColorTableScale;
+ bias = ctx->Pixel.TextureColorTableBias;
break;
case GL_POST_CONVOLUTION_COLOR_TABLE:
- table = &ctx->PostConvolutionColorTable;
- rScale = ctx->Pixel.PCCTscale[0];
- gScale = ctx->Pixel.PCCTscale[1];
- bScale = ctx->Pixel.PCCTscale[2];
- aScale = ctx->Pixel.PCCTscale[3];
- rBias = ctx->Pixel.PCCTbias[0];
- gBias = ctx->Pixel.PCCTbias[1];
- bBias = ctx->Pixel.PCCTbias[2];
- aBias = ctx->Pixel.PCCTbias[3];
+ table = &ctx->ColorTable[COLORTABLE_POSTCONVOLUTION];
+ scale = ctx->Pixel.ColorTableScale[COLORTABLE_POSTCONVOLUTION];
+ bias = ctx->Pixel.ColorTableBias[COLORTABLE_POSTCONVOLUTION];
break;
case GL_POST_COLOR_MATRIX_COLOR_TABLE:
- table = &ctx->PostColorMatrixColorTable;
- rScale = ctx->Pixel.PCMCTscale[0];
- gScale = ctx->Pixel.PCMCTscale[1];
- bScale = ctx->Pixel.PCMCTscale[2];
- aScale = ctx->Pixel.PCMCTscale[3];
- rBias = ctx->Pixel.PCMCTbias[0];
- gBias = ctx->Pixel.PCMCTbias[1];
- bBias = ctx->Pixel.PCMCTbias[2];
- aBias = ctx->Pixel.PCMCTbias[3];
+ table = &ctx->ColorTable[COLORTABLE_POSTCOLORMATRIX];
+ scale = ctx->Pixel.ColorTableScale[COLORTABLE_POSTCOLORMATRIX];
+ bias = ctx->Pixel.ColorTableBias[COLORTABLE_POSTCOLORMATRIX];
break;
default:
_mesa_error(ctx, GL_INVALID_ENUM, "glColorSubTable(target)");
store_colortable_entries(ctx, table, start, count,
format, type, data,
- rScale, rBias,
- gScale, gBias,
- bScale, bBias,
- aScale, aBias);
+ scale[0], bias[0],
+ scale[1], bias[1],
+ scale[2], bias[2],
+ scale[3], bias[3]);
if (texObj || target == GL_SHARED_TEXTURE_PALETTE_EXT) {
/* per-texture object palette */
table = &ctx->Texture.Palette;
break;
case GL_COLOR_TABLE:
- table = &ctx->ColorTable;
+ table = &ctx->ColorTable[COLORTABLE_PRECONVOLUTION];
break;
case GL_TEXTURE_COLOR_TABLE_SGI:
if (!ctx->Extensions.SGI_texture_color_table) {
table = &(texUnit->ColorTable);
break;
case GL_POST_CONVOLUTION_COLOR_TABLE:
- table = &ctx->PostConvolutionColorTable;
+ table = &ctx->ColorTable[COLORTABLE_POSTCONVOLUTION];
break;
case GL_POST_COLOR_MATRIX_COLOR_TABLE:
- table = &ctx->PostColorMatrixColorTable;
+ table = &ctx->ColorTable[COLORTABLE_POSTCOLORMATRIX];
break;
default:
_mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTable(target)");
ASSERT(table);
+ if (table->Size <= 0) {
+ return;
+ }
+
switch (table->_BaseFormat) {
case GL_ALPHA:
{
switch (target) {
case GL_COLOR_TABLE_SGI:
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- ctx->Pixel.ColorTableScale[0] = params[0];
- ctx->Pixel.ColorTableScale[1] = params[1];
- ctx->Pixel.ColorTableScale[2] = params[2];
- ctx->Pixel.ColorTableScale[3] = params[3];
+ COPY_4V(ctx->Pixel.ColorTableScale[COLORTABLE_PRECONVOLUTION], params);
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- ctx->Pixel.ColorTableBias[0] = params[0];
- ctx->Pixel.ColorTableBias[1] = params[1];
- ctx->Pixel.ColorTableBias[2] = params[2];
- ctx->Pixel.ColorTableBias[3] = params[3];
+ COPY_4V(ctx->Pixel.ColorTableBias[COLORTABLE_PRECONVOLUTION], params);
}
else {
_mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameterfv(pname)");
return;
}
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- ctx->Pixel.TextureColorTableScale[0] = params[0];
- ctx->Pixel.TextureColorTableScale[1] = params[1];
- ctx->Pixel.TextureColorTableScale[2] = params[2];
- ctx->Pixel.TextureColorTableScale[3] = params[3];
+ COPY_4V(ctx->Pixel.TextureColorTableScale, params);
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- ctx->Pixel.TextureColorTableBias[0] = params[0];
- ctx->Pixel.TextureColorTableBias[1] = params[1];
- ctx->Pixel.TextureColorTableBias[2] = params[2];
- ctx->Pixel.TextureColorTableBias[3] = params[3];
+ COPY_4V(ctx->Pixel.TextureColorTableBias, params);
}
else {
_mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameterfv(pname)");
break;
case GL_POST_CONVOLUTION_COLOR_TABLE_SGI:
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- ctx->Pixel.PCCTscale[0] = params[0];
- ctx->Pixel.PCCTscale[1] = params[1];
- ctx->Pixel.PCCTscale[2] = params[2];
- ctx->Pixel.PCCTscale[3] = params[3];
+ COPY_4V(ctx->Pixel.ColorTableScale[COLORTABLE_POSTCONVOLUTION], params);
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- ctx->Pixel.PCCTbias[0] = params[0];
- ctx->Pixel.PCCTbias[1] = params[1];
- ctx->Pixel.PCCTbias[2] = params[2];
- ctx->Pixel.PCCTbias[3] = params[3];
+ COPY_4V(ctx->Pixel.ColorTableBias[COLORTABLE_POSTCONVOLUTION], params);
}
else {
_mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameterfv(pname)");
break;
case GL_POST_COLOR_MATRIX_COLOR_TABLE_SGI:
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- ctx->Pixel.PCMCTscale[0] = params[0];
- ctx->Pixel.PCMCTscale[1] = params[1];
- ctx->Pixel.PCMCTscale[2] = params[2];
- ctx->Pixel.PCMCTscale[3] = params[3];
+ COPY_4V(ctx->Pixel.ColorTableScale[COLORTABLE_POSTCOLORMATRIX], params);
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- ctx->Pixel.PCMCTbias[0] = params[0];
- ctx->Pixel.PCMCTbias[1] = params[1];
- ctx->Pixel.PCMCTbias[2] = params[2];
- ctx->Pixel.PCMCTbias[3] = params[3];
+ COPY_4V(ctx->Pixel.ColorTableBias[COLORTABLE_POSTCOLORMATRIX], params);
}
else {
_mesa_error(ctx, GL_INVALID_ENUM, "glColorTableParameterfv(pname)");
table = &ctx->Texture.Palette;
break;
case GL_COLOR_TABLE:
- table = &ctx->ColorTable;
+ table = &ctx->ColorTable[COLORTABLE_PRECONVOLUTION];
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- params[0] = ctx->Pixel.ColorTableScale[0];
- params[1] = ctx->Pixel.ColorTableScale[1];
- params[2] = ctx->Pixel.ColorTableScale[2];
- params[3] = ctx->Pixel.ColorTableScale[3];
+ COPY_4V(params, ctx->Pixel.ColorTableScale[COLORTABLE_PRECONVOLUTION]);
return;
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- params[0] = ctx->Pixel.ColorTableBias[0];
- params[1] = ctx->Pixel.ColorTableBias[1];
- params[2] = ctx->Pixel.ColorTableBias[2];
- params[3] = ctx->Pixel.ColorTableBias[3];
+ COPY_4V(params, ctx->Pixel.ColorTableBias[COLORTABLE_PRECONVOLUTION]);
return;
}
break;
case GL_PROXY_COLOR_TABLE:
- table = &ctx->ProxyColorTable;
+ table = &ctx->ProxyColorTable[COLORTABLE_PRECONVOLUTION];
break;
case GL_TEXTURE_COLOR_TABLE_SGI:
if (!ctx->Extensions.SGI_texture_color_table) {
}
table = &(texUnit->ColorTable);
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- params[0] = ctx->Pixel.TextureColorTableScale[0];
- params[1] = ctx->Pixel.TextureColorTableScale[1];
- params[2] = ctx->Pixel.TextureColorTableScale[2];
- params[3] = ctx->Pixel.TextureColorTableScale[3];
+ COPY_4V(params, ctx->Pixel.TextureColorTableScale);
return;
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- params[0] = ctx->Pixel.TextureColorTableBias[0];
- params[1] = ctx->Pixel.TextureColorTableBias[1];
- params[2] = ctx->Pixel.TextureColorTableBias[2];
- params[3] = ctx->Pixel.TextureColorTableBias[3];
+ COPY_4V(params, ctx->Pixel.TextureColorTableBias);
return;
}
break;
table = &(texUnit->ProxyColorTable);
break;
case GL_POST_CONVOLUTION_COLOR_TABLE:
- table = &ctx->PostConvolutionColorTable;
+ table = &ctx->ColorTable[COLORTABLE_POSTCONVOLUTION];
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- params[0] = ctx->Pixel.PCCTscale[0];
- params[1] = ctx->Pixel.PCCTscale[1];
- params[2] = ctx->Pixel.PCCTscale[2];
- params[3] = ctx->Pixel.PCCTscale[3];
+ COPY_4V(params, ctx->Pixel.ColorTableScale[COLORTABLE_POSTCONVOLUTION]);
return;
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- params[0] = ctx->Pixel.PCCTbias[0];
- params[1] = ctx->Pixel.PCCTbias[1];
- params[2] = ctx->Pixel.PCCTbias[2];
- params[3] = ctx->Pixel.PCCTbias[3];
+ COPY_4V(params, ctx->Pixel.ColorTableBias[COLORTABLE_POSTCONVOLUTION]);
return;
}
break;
case GL_PROXY_POST_CONVOLUTION_COLOR_TABLE:
- table = &ctx->ProxyPostConvolutionColorTable;
+ table = &ctx->ProxyColorTable[COLORTABLE_POSTCONVOLUTION];
break;
case GL_POST_COLOR_MATRIX_COLOR_TABLE:
- table = &ctx->PostColorMatrixColorTable;
+ table = &ctx->ColorTable[COLORTABLE_POSTCOLORMATRIX];
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- params[0] = ctx->Pixel.PCMCTscale[0];
- params[1] = ctx->Pixel.PCMCTscale[1];
- params[2] = ctx->Pixel.PCMCTscale[2];
- params[3] = ctx->Pixel.PCMCTscale[3];
+ COPY_4V(params, ctx->Pixel.ColorTableScale[COLORTABLE_POSTCOLORMATRIX]);
return;
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- params[0] = ctx->Pixel.PCMCTbias[0];
- params[1] = ctx->Pixel.PCMCTbias[1];
- params[2] = ctx->Pixel.PCMCTbias[2];
- params[3] = ctx->Pixel.PCMCTbias[3];
+ COPY_4V(params, ctx->Pixel.ColorTableBias[COLORTABLE_POSTCOLORMATRIX]);
return;
}
break;
case GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE:
- table = &ctx->ProxyPostColorMatrixColorTable;
+ table = &ctx->ProxyColorTable[COLORTABLE_POSTCOLORMATRIX];
break;
default:
_mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTableParameterfv(target)");
table = &ctx->Texture.Palette;
break;
case GL_COLOR_TABLE:
- table = &ctx->ColorTable;
+ table = &ctx->ColorTable[COLORTABLE_PRECONVOLUTION];
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- params[0] = (GLint) ctx->Pixel.ColorTableScale[0];
- params[1] = (GLint) ctx->Pixel.ColorTableScale[1];
- params[2] = (GLint) ctx->Pixel.ColorTableScale[2];
- params[3] = (GLint) ctx->Pixel.ColorTableScale[3];
+ GLfloat *scale = ctx->Pixel.ColorTableScale[COLORTABLE_PRECONVOLUTION];
+ params[0] = (GLint) scale[0];
+ params[1] = (GLint) scale[1];
+ params[2] = (GLint) scale[2];
+ params[3] = (GLint) scale[3];
return;
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- params[0] = (GLint) ctx->Pixel.ColorTableBias[0];
- params[1] = (GLint) ctx->Pixel.ColorTableBias[1];
- params[2] = (GLint) ctx->Pixel.ColorTableBias[2];
- params[3] = (GLint) ctx->Pixel.ColorTableBias[3];
+ GLfloat *bias = ctx->Pixel.ColorTableBias[COLORTABLE_PRECONVOLUTION];
+ params[0] = (GLint) bias[0];
+ params[1] = (GLint) bias[1];
+ params[2] = (GLint) bias[2];
+ params[3] = (GLint) bias[3];
return;
}
break;
case GL_PROXY_COLOR_TABLE:
- table = &ctx->ProxyColorTable;
+ table = &ctx->ProxyColorTable[COLORTABLE_PRECONVOLUTION];
break;
case GL_TEXTURE_COLOR_TABLE_SGI:
if (!ctx->Extensions.SGI_texture_color_table) {
table = &(texUnit->ProxyColorTable);
break;
case GL_POST_CONVOLUTION_COLOR_TABLE:
- table = &ctx->PostConvolutionColorTable;
+ table = &ctx->ColorTable[COLORTABLE_POSTCONVOLUTION];
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- params[0] = (GLint) ctx->Pixel.PCCTscale[0];
- params[1] = (GLint) ctx->Pixel.PCCTscale[1];
- params[2] = (GLint) ctx->Pixel.PCCTscale[2];
- params[3] = (GLint) ctx->Pixel.PCCTscale[3];
+ GLfloat *scale = ctx->Pixel.ColorTableScale[COLORTABLE_POSTCONVOLUTION];
+ params[0] = (GLint) scale[0];
+ params[1] = (GLint) scale[1];
+ params[2] = (GLint) scale[2];
+ params[3] = (GLint) scale[3];
return;
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- params[0] = (GLint) ctx->Pixel.PCCTbias[0];
- params[1] = (GLint) ctx->Pixel.PCCTbias[1];
- params[2] = (GLint) ctx->Pixel.PCCTbias[2];
- params[3] = (GLint) ctx->Pixel.PCCTbias[3];
+ GLfloat *bias = ctx->Pixel.ColorTableBias[COLORTABLE_POSTCONVOLUTION];
+ params[0] = (GLint) bias[0];
+ params[1] = (GLint) bias[1];
+ params[2] = (GLint) bias[2];
+ params[3] = (GLint) bias[3];
return;
}
break;
case GL_PROXY_POST_CONVOLUTION_COLOR_TABLE:
- table = &ctx->ProxyPostConvolutionColorTable;
+ table = &ctx->ProxyColorTable[COLORTABLE_POSTCONVOLUTION];
break;
case GL_POST_COLOR_MATRIX_COLOR_TABLE:
- table = &ctx->PostColorMatrixColorTable;
+ table = &ctx->ColorTable[COLORTABLE_POSTCOLORMATRIX];
if (pname == GL_COLOR_TABLE_SCALE_SGI) {
- params[0] = (GLint) ctx->Pixel.PCMCTscale[0];
- params[1] = (GLint) ctx->Pixel.PCMCTscale[1];
- params[2] = (GLint) ctx->Pixel.PCMCTscale[2];
- params[3] = (GLint) ctx->Pixel.PCMCTscale[3];
+ GLfloat *scale = ctx->Pixel.ColorTableScale[COLORTABLE_POSTCOLORMATRIX];
+ params[0] = (GLint) scale[0];
+ params[0] = (GLint) scale[1];
+ params[0] = (GLint) scale[2];
+ params[0] = (GLint) scale[3];
return;
}
else if (pname == GL_COLOR_TABLE_BIAS_SGI) {
- params[0] = (GLint) ctx->Pixel.PCMCTbias[0];
- params[1] = (GLint) ctx->Pixel.PCMCTbias[1];
- params[2] = (GLint) ctx->Pixel.PCMCTbias[2];
- params[3] = (GLint) ctx->Pixel.PCMCTbias[3];
+ GLfloat *bias = ctx->Pixel.ColorTableScale[COLORTABLE_POSTCOLORMATRIX];
+ params[0] = (GLint) bias[0];
+ params[1] = (GLint) bias[1];
+ params[2] = (GLint) bias[2];
+ params[3] = (GLint) bias[3];
return;
}
break;
case GL_PROXY_POST_COLOR_MATRIX_COLOR_TABLE:
- table = &ctx->ProxyPostColorMatrixColorTable;
+ table = &ctx->ProxyColorTable[COLORTABLE_POSTCOLORMATRIX];
break;
default:
_mesa_error(ctx, GL_INVALID_ENUM, "glGetColorTableParameteriv(target)");
void
_mesa_init_colortables( GLcontext * ctx )
{
- /* Color tables */
- _mesa_init_colortable(&ctx->ColorTable);
- _mesa_init_colortable(&ctx->ProxyColorTable);
- _mesa_init_colortable(&ctx->PostConvolutionColorTable);
- _mesa_init_colortable(&ctx->ProxyPostConvolutionColorTable);
- _mesa_init_colortable(&ctx->PostColorMatrixColorTable);
- _mesa_init_colortable(&ctx->ProxyPostColorMatrixColorTable);
+ GLuint i;
+ for (i = 0; i < COLORTABLE_MAX; i++) {
+ _mesa_init_colortable(&ctx->ColorTable[i]);
+ _mesa_init_colortable(&ctx->ProxyColorTable[i]);
+ }
}
void
_mesa_free_colortables_data( GLcontext *ctx )
{
- _mesa_free_colortable_data(&ctx->ColorTable);
- _mesa_free_colortable_data(&ctx->ProxyColorTable);
- _mesa_free_colortable_data(&ctx->PostConvolutionColorTable);
- _mesa_free_colortable_data(&ctx->ProxyPostConvolutionColorTable);
- _mesa_free_colortable_data(&ctx->PostColorMatrixColorTable);
- _mesa_free_colortable_data(&ctx->ProxyPostColorMatrixColorTable);
+ GLuint i;
+ for (i = 0; i < COLORTABLE_MAX; i++) {
+ _mesa_free_colortable_data(&ctx->ColorTable[i]);
+ _mesa_free_colortable_data(&ctx->ProxyColorTable[i]);
+ }
}
params[0] = ENUM_TO_BOOLEAN(ctx->Hint.PerspectiveCorrection);
break;
case GL_PIXEL_MAP_A_TO_A_SIZE:
- params[0] = INT_TO_BOOLEAN(ctx->Pixel.MapAtoAsize);
+ params[0] = INT_TO_BOOLEAN(ctx->PixelMaps.AtoA.Size);
break;
case GL_PIXEL_MAP_B_TO_B_SIZE:
- params[0] = INT_TO_BOOLEAN(ctx->Pixel.MapBtoBsize);
+ params[0] = INT_TO_BOOLEAN(ctx->PixelMaps.BtoB.Size);
break;
case GL_PIXEL_MAP_G_TO_G_SIZE:
- params[0] = INT_TO_BOOLEAN(ctx->Pixel.MapGtoGsize);
+ params[0] = INT_TO_BOOLEAN(ctx->PixelMaps.GtoG.Size);
break;
case GL_PIXEL_MAP_I_TO_A_SIZE:
- params[0] = INT_TO_BOOLEAN(ctx->Pixel.MapItoAsize);
+ params[0] = INT_TO_BOOLEAN(ctx->PixelMaps.ItoA.Size);
break;
case GL_PIXEL_MAP_I_TO_B_SIZE:
- params[0] = INT_TO_BOOLEAN(ctx->Pixel.MapItoBsize);
+ params[0] = INT_TO_BOOLEAN(ctx->PixelMaps.ItoB.Size);
break;
case GL_PIXEL_MAP_I_TO_G_SIZE:
- params[0] = INT_TO_BOOLEAN(ctx->Pixel.MapItoGsize);
+ params[0] = INT_TO_BOOLEAN(ctx->PixelMaps.ItoG.Size);
break;
case GL_PIXEL_MAP_I_TO_I_SIZE:
- params[0] = INT_TO_BOOLEAN(ctx->Pixel.MapItoIsize);
+ params[0] = INT_TO_BOOLEAN(ctx->PixelMaps.ItoI.Size);
break;
case GL_PIXEL_MAP_I_TO_R_SIZE:
- params[0] = INT_TO_BOOLEAN(ctx->Pixel.MapItoRsize);
+ params[0] = INT_TO_BOOLEAN(ctx->PixelMaps.ItoR.Size);
break;
case GL_PIXEL_MAP_R_TO_R_SIZE:
- params[0] = INT_TO_BOOLEAN(ctx->Pixel.MapRtoRsize);
+ params[0] = INT_TO_BOOLEAN(ctx->PixelMaps.RtoR.Size);
break;
case GL_PIXEL_MAP_S_TO_S_SIZE:
- params[0] = INT_TO_BOOLEAN(ctx->Pixel.MapStoSsize);
+ params[0] = INT_TO_BOOLEAN(ctx->PixelMaps.StoS.Size);
break;
case GL_POINT_SIZE:
params[0] = FLOAT_TO_BOOLEAN(ctx->Point.Size);
break;
case GL_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_color_table, "GetBooleanv");
- params[0] = ctx->Pixel.ColorTableEnabled;
+ params[0] = ctx->Pixel.ColorTableEnabled[COLORTABLE_PRECONVOLUTION];
break;
case GL_POST_CONVOLUTION_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_color_table, "GetBooleanv");
- params[0] = ctx->Pixel.PostConvolutionColorTableEnabled;
+ params[0] = ctx->Pixel.ColorTableEnabled[COLORTABLE_POSTCONVOLUTION];
break;
case GL_POST_COLOR_MATRIX_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_color_table, "GetBooleanv");
- params[0] = ctx->Pixel.PostColorMatrixColorTableEnabled;
+ params[0] = ctx->Pixel.ColorTableEnabled[COLORTABLE_POSTCOLORMATRIX];
break;
case GL_TEXTURE_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_texture_color_table, "GetBooleanv");
break;
case GL_MAX_VARYING_FLOATS_ARB:
CHECK_EXT1(ARB_vertex_shader, "GetBooleanv");
- params[0] = INT_TO_BOOLEAN(ctx->Const.MaxVaryingFloats);
+ params[0] = INT_TO_BOOLEAN(ctx->Const.MaxVarying * 4);
break;
case GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB:
CHECK_EXT1(ARB_vertex_shader, "GetBooleanv");
CHECK_EXT1(ARB_vertex_shader, "GetBooleanv");
params[0] = INT_TO_BOOLEAN(MAX_COMBINED_TEXTURE_IMAGE_UNITS);
break;
+ case GL_CURRENT_PROGRAM:
+ CHECK_EXT1(ARB_shader_objects, "GetBooleanv");
+ params[0] = INT_TO_BOOLEAN(ctx->Shader.CurrentProgram ? ctx->Shader.CurrentProgram->Name : 0);
+ break;
default:
_mesa_error(ctx, GL_INVALID_ENUM, "glGetBooleanv(pname=0x%x)", pname);
}
params[0] = ENUM_TO_FLOAT(ctx->Hint.PerspectiveCorrection);
break;
case GL_PIXEL_MAP_A_TO_A_SIZE:
- params[0] = (GLfloat)(ctx->Pixel.MapAtoAsize);
+ params[0] = (GLfloat)(ctx->PixelMaps.AtoA.Size);
break;
case GL_PIXEL_MAP_B_TO_B_SIZE:
- params[0] = (GLfloat)(ctx->Pixel.MapBtoBsize);
+ params[0] = (GLfloat)(ctx->PixelMaps.BtoB.Size);
break;
case GL_PIXEL_MAP_G_TO_G_SIZE:
- params[0] = (GLfloat)(ctx->Pixel.MapGtoGsize);
+ params[0] = (GLfloat)(ctx->PixelMaps.GtoG.Size);
break;
case GL_PIXEL_MAP_I_TO_A_SIZE:
- params[0] = (GLfloat)(ctx->Pixel.MapItoAsize);
+ params[0] = (GLfloat)(ctx->PixelMaps.ItoA.Size);
break;
case GL_PIXEL_MAP_I_TO_B_SIZE:
- params[0] = (GLfloat)(ctx->Pixel.MapItoBsize);
+ params[0] = (GLfloat)(ctx->PixelMaps.ItoB.Size);
break;
case GL_PIXEL_MAP_I_TO_G_SIZE:
- params[0] = (GLfloat)(ctx->Pixel.MapItoGsize);
+ params[0] = (GLfloat)(ctx->PixelMaps.ItoG.Size);
break;
case GL_PIXEL_MAP_I_TO_I_SIZE:
- params[0] = (GLfloat)(ctx->Pixel.MapItoIsize);
+ params[0] = (GLfloat)(ctx->PixelMaps.ItoI.Size);
break;
case GL_PIXEL_MAP_I_TO_R_SIZE:
- params[0] = (GLfloat)(ctx->Pixel.MapItoRsize);
+ params[0] = (GLfloat)(ctx->PixelMaps.ItoR.Size);
break;
case GL_PIXEL_MAP_R_TO_R_SIZE:
- params[0] = (GLfloat)(ctx->Pixel.MapRtoRsize);
+ params[0] = (GLfloat)(ctx->PixelMaps.RtoR.Size);
break;
case GL_PIXEL_MAP_S_TO_S_SIZE:
- params[0] = (GLfloat)(ctx->Pixel.MapStoSsize);
+ params[0] = (GLfloat)(ctx->PixelMaps.StoS.Size);
break;
case GL_POINT_SIZE:
params[0] = ctx->Point.Size;
break;
case GL_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_color_table, "GetFloatv");
- params[0] = BOOLEAN_TO_FLOAT(ctx->Pixel.ColorTableEnabled);
+ params[0] = BOOLEAN_TO_FLOAT(ctx->Pixel.ColorTableEnabled[COLORTABLE_PRECONVOLUTION]);
break;
case GL_POST_CONVOLUTION_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_color_table, "GetFloatv");
- params[0] = BOOLEAN_TO_FLOAT(ctx->Pixel.PostConvolutionColorTableEnabled);
+ params[0] = BOOLEAN_TO_FLOAT(ctx->Pixel.ColorTableEnabled[COLORTABLE_POSTCONVOLUTION]);
break;
case GL_POST_COLOR_MATRIX_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_color_table, "GetFloatv");
- params[0] = BOOLEAN_TO_FLOAT(ctx->Pixel.PostColorMatrixColorTableEnabled);
+ params[0] = BOOLEAN_TO_FLOAT(ctx->Pixel.ColorTableEnabled[COLORTABLE_POSTCOLORMATRIX]);
break;
case GL_TEXTURE_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_texture_color_table, "GetFloatv");
break;
case GL_MAX_VARYING_FLOATS_ARB:
CHECK_EXT1(ARB_vertex_shader, "GetFloatv");
- params[0] = (GLfloat)(ctx->Const.MaxVaryingFloats);
+ params[0] = (GLfloat)(ctx->Const.MaxVarying * 4);
break;
case GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB:
CHECK_EXT1(ARB_vertex_shader, "GetFloatv");
CHECK_EXT1(ARB_vertex_shader, "GetFloatv");
params[0] = (GLfloat)(MAX_COMBINED_TEXTURE_IMAGE_UNITS);
break;
+ case GL_CURRENT_PROGRAM:
+ CHECK_EXT1(ARB_shader_objects, "GetFloatv");
+ params[0] = (GLfloat)(ctx->Shader.CurrentProgram ? ctx->Shader.CurrentProgram->Name : 0);
+ break;
default:
_mesa_error(ctx, GL_INVALID_ENUM, "glGetFloatv(pname=0x%x)", pname);
}
params[0] = ENUM_TO_INT(ctx->Hint.PerspectiveCorrection);
break;
case GL_PIXEL_MAP_A_TO_A_SIZE:
- params[0] = ctx->Pixel.MapAtoAsize;
+ params[0] = ctx->PixelMaps.AtoA.Size;
break;
case GL_PIXEL_MAP_B_TO_B_SIZE:
- params[0] = ctx->Pixel.MapBtoBsize;
+ params[0] = ctx->PixelMaps.BtoB.Size;
break;
case GL_PIXEL_MAP_G_TO_G_SIZE:
- params[0] = ctx->Pixel.MapGtoGsize;
+ params[0] = ctx->PixelMaps.GtoG.Size;
break;
case GL_PIXEL_MAP_I_TO_A_SIZE:
- params[0] = ctx->Pixel.MapItoAsize;
+ params[0] = ctx->PixelMaps.ItoA.Size;
break;
case GL_PIXEL_MAP_I_TO_B_SIZE:
- params[0] = ctx->Pixel.MapItoBsize;
+ params[0] = ctx->PixelMaps.ItoB.Size;
break;
case GL_PIXEL_MAP_I_TO_G_SIZE:
- params[0] = ctx->Pixel.MapItoGsize;
+ params[0] = ctx->PixelMaps.ItoG.Size;
break;
case GL_PIXEL_MAP_I_TO_I_SIZE:
- params[0] = ctx->Pixel.MapItoIsize;
+ params[0] = ctx->PixelMaps.ItoI.Size;
break;
case GL_PIXEL_MAP_I_TO_R_SIZE:
- params[0] = ctx->Pixel.MapItoRsize;
+ params[0] = ctx->PixelMaps.ItoR.Size;
break;
case GL_PIXEL_MAP_R_TO_R_SIZE:
- params[0] = ctx->Pixel.MapRtoRsize;
+ params[0] = ctx->PixelMaps.RtoR.Size;
break;
case GL_PIXEL_MAP_S_TO_S_SIZE:
- params[0] = ctx->Pixel.MapStoSsize;
+ params[0] = ctx->PixelMaps.StoS.Size;
break;
case GL_POINT_SIZE:
params[0] = IROUND(ctx->Point.Size);
break;
case GL_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_color_table, "GetIntegerv");
- params[0] = BOOLEAN_TO_INT(ctx->Pixel.ColorTableEnabled);
+ params[0] = BOOLEAN_TO_INT(ctx->Pixel.ColorTableEnabled[COLORTABLE_PRECONVOLUTION]);
break;
case GL_POST_CONVOLUTION_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_color_table, "GetIntegerv");
- params[0] = BOOLEAN_TO_INT(ctx->Pixel.PostConvolutionColorTableEnabled);
+ params[0] = BOOLEAN_TO_INT(ctx->Pixel.ColorTableEnabled[COLORTABLE_POSTCONVOLUTION]);
break;
case GL_POST_COLOR_MATRIX_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_color_table, "GetIntegerv");
- params[0] = BOOLEAN_TO_INT(ctx->Pixel.PostColorMatrixColorTableEnabled);
+ params[0] = BOOLEAN_TO_INT(ctx->Pixel.ColorTableEnabled[COLORTABLE_POSTCOLORMATRIX]);
break;
case GL_TEXTURE_COLOR_TABLE_SGI:
CHECK_EXT1(SGI_texture_color_table, "GetIntegerv");
break;
case GL_MAX_VARYING_FLOATS_ARB:
CHECK_EXT1(ARB_vertex_shader, "GetIntegerv");
- params[0] = ctx->Const.MaxVaryingFloats;
+ params[0] = ctx->Const.MaxVarying * 4;
break;
case GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB:
CHECK_EXT1(ARB_vertex_shader, "GetIntegerv");
CHECK_EXT1(ARB_vertex_shader, "GetIntegerv");
params[0] = MAX_COMBINED_TEXTURE_IMAGE_UNITS;
break;
+ case GL_CURRENT_PROGRAM:
+ CHECK_EXT1(ARB_shader_objects, "GetIntegerv");
+ params[0] = ctx->Shader.CurrentProgram ? ctx->Shader.CurrentProgram->Name : 0;
+ break;
default:
_mesa_error(ctx, GL_INVALID_ENUM, "glGetIntegerv(pname=0x%x)", pname);
}
( "GL_PACK_INVERT_MESA", GLboolean, ["ctx->Pack.Invert"], "", None ),
( "GL_PERSPECTIVE_CORRECTION_HINT", GLenum,
["ctx->Hint.PerspectiveCorrection"], "", None ),
- ( "GL_PIXEL_MAP_A_TO_A_SIZE", GLint, ["ctx->Pixel.MapAtoAsize"], "", None ),
- ( "GL_PIXEL_MAP_B_TO_B_SIZE", GLint, ["ctx->Pixel.MapBtoBsize"], "", None ),
- ( "GL_PIXEL_MAP_G_TO_G_SIZE", GLint, ["ctx->Pixel.MapGtoGsize"], "", None ),
- ( "GL_PIXEL_MAP_I_TO_A_SIZE", GLint, ["ctx->Pixel.MapItoAsize"], "", None ),
- ( "GL_PIXEL_MAP_I_TO_B_SIZE", GLint, ["ctx->Pixel.MapItoBsize"], "", None ),
- ( "GL_PIXEL_MAP_I_TO_G_SIZE", GLint, ["ctx->Pixel.MapItoGsize"], "", None ),
- ( "GL_PIXEL_MAP_I_TO_I_SIZE", GLint, ["ctx->Pixel.MapItoIsize"], "", None ),
- ( "GL_PIXEL_MAP_I_TO_R_SIZE", GLint, ["ctx->Pixel.MapItoRsize"], "", None ),
- ( "GL_PIXEL_MAP_R_TO_R_SIZE", GLint, ["ctx->Pixel.MapRtoRsize"], "", None ),
- ( "GL_PIXEL_MAP_S_TO_S_SIZE", GLint, ["ctx->Pixel.MapStoSsize"], "", None ),
+ ( "GL_PIXEL_MAP_A_TO_A_SIZE", GLint, ["ctx->PixelMaps.AtoA.Size"], "", None ),
+ ( "GL_PIXEL_MAP_B_TO_B_SIZE", GLint, ["ctx->PixelMaps.BtoB.Size"], "", None ),
+ ( "GL_PIXEL_MAP_G_TO_G_SIZE", GLint, ["ctx->PixelMaps.GtoG.Size"], "", None ),
+ ( "GL_PIXEL_MAP_I_TO_A_SIZE", GLint, ["ctx->PixelMaps.ItoA.Size"], "", None ),
+ ( "GL_PIXEL_MAP_I_TO_B_SIZE", GLint, ["ctx->PixelMaps.ItoB.Size"], "", None ),
+ ( "GL_PIXEL_MAP_I_TO_G_SIZE", GLint, ["ctx->PixelMaps.ItoG.Size"], "", None ),
+ ( "GL_PIXEL_MAP_I_TO_I_SIZE", GLint, ["ctx->PixelMaps.ItoI.Size"], "", None ),
+ ( "GL_PIXEL_MAP_I_TO_R_SIZE", GLint, ["ctx->PixelMaps.ItoR.Size"], "", None ),
+ ( "GL_PIXEL_MAP_R_TO_R_SIZE", GLint, ["ctx->PixelMaps.RtoR.Size"], "", None ),
+ ( "GL_PIXEL_MAP_S_TO_S_SIZE", GLint, ["ctx->PixelMaps.StoS.Size"], "", None ),
( "GL_POINT_SIZE", GLfloat, ["ctx->Point.Size"], "", None ),
( "GL_POINT_SIZE_GRANULARITY", GLfloat,
["ctx->Const.PointSizeGranularity"], "", None ),
# GL_SGI_color_table / GL_ARB_imaging
( "GL_COLOR_TABLE_SGI", GLboolean,
- ["ctx->Pixel.ColorTableEnabled"], "", ["SGI_color_table"] ),
+ ["ctx->Pixel.ColorTableEnabled[COLORTABLE_PRECONVOLUTION]"], "", ["SGI_color_table"] ),
( "GL_POST_CONVOLUTION_COLOR_TABLE_SGI", GLboolean,
- ["ctx->Pixel.PostConvolutionColorTableEnabled"], "", ["SGI_color_table"] ),
+ ["ctx->Pixel.ColorTableEnabled[COLORTABLE_POSTCONVOLUTION]"], "", ["SGI_color_table"] ),
( "GL_POST_COLOR_MATRIX_COLOR_TABLE_SGI", GLboolean,
- ["ctx->Pixel.PostColorMatrixColorTableEnabled"], "", ["SGI_color_table"] ),
+ ["ctx->Pixel.ColorTableEnabled[COLORTABLE_POSTCOLORMATRIX]"], "", ["SGI_color_table"] ),
# GL_SGI_texture_color_table
( "GL_TEXTURE_COLOR_TABLE_SGI", GLboolean,
["ctx->Const.VertexProgram.MaxUniformComponents"], "",
["ARB_vertex_shader"] ),
( "GL_MAX_VARYING_FLOATS_ARB", GLint,
- ["ctx->Const.MaxVaryingFloats"], "", ["ARB_vertex_shader"] ),
+ ["ctx->Const.MaxVarying * 4"], "", ["ARB_vertex_shader"] ),
( "GL_MAX_VERTEX_TEXTURE_IMAGE_UNITS_ARB", GLint,
["ctx->Const.MaxVertexTextureImageUnits"], "", ["ARB_vertex_shader"] ),
( "GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS_ARB", GLint,
- ["MAX_COMBINED_TEXTURE_IMAGE_UNITS"], "", ["ARB_vertex_shader"] )
+ ["MAX_COMBINED_TEXTURE_IMAGE_UNITS"], "", ["ARB_vertex_shader"] ),
+
+ # GL_ARB_shader_objects
+ # Actually, this token isn't part of GL_ARB_shader_objects, but is
+ # close enough for now.
+ ( "GL_CURRENT_PROGRAM", GLint,
+ ["ctx->Shader.CurrentProgram ? ctx->Shader.CurrentProgram->Name : 0"],
+ "", ["ARB_shader_objects"] )
]
/*
* Mesa 3-D graphics library
- * Version: 6.5
+ * Version: 6.5.3
*
- * Copyright (C) 1999-2006 Brian Paul All Rights Reserved.
+ * Copyright (C) 1999-2007 Brian Paul All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
#include "bitset.h"
+/**
+ * Special, internal token
+ */
+#define GL_SHADER_PROGRAM 0x9999
+
+
/**
* Color channel data type.
*/
#define VERT_BIT_GENERIC(g) (1 << (VERT_ATTRIB_GENERIC0 + (g)))
/*@}*/
-/**
- * GLSL allows shader writers to allocate vertex result attributes (varyings) in
- * single float component granularity. This is in contrast to vertex / fragment
- * programs, where result attributes (actually texcoords) were allocated
- * in 4-component vectors of floats granularity.
- * For performance reasons, it would be optimal to stick with this scheme on a scalar
- * processor. Varyings will likely be allocated as 3-component vectors, so statistically
- * we win 2 floats.
- * The constant VARYINGS_PER_VECTOR tells us how much of float components we pack into
- * one result vector. For scalar processor it would be 1, for vector processor - 4.
- *
- * NOTE: Currently we pack varyings into vertex attributes.
- */
-#define VARYINGS_PER_VECTOR 2
-#define VARYING_EMIT_STYLE EMIT_2F
-#define MAX_VARYING_VECTORS ((MAX_VARYING_FLOATS + VARYINGS_PER_VECTOR - 1) / VARYINGS_PER_VECTOR)
/**
* Indexes for vertex program result attributes
#define VERT_RESULT_BFC0 13
#define VERT_RESULT_BFC1 14
#define VERT_RESULT_EDGE 15
-#define VERT_RESULT_MAX 16
+#define VERT_RESULT_VAR0 16 /**< shader varying */
+#define VERT_RESULT_MAX (VERT_RESULT_VAR0 + MAX_VARYING)
/*@}*/
FRAG_ATTRIB_TEX5 = 9,
FRAG_ATTRIB_TEX6 = 10,
FRAG_ATTRIB_TEX7 = 11,
- FRAG_ATTRIB_MAX = 12
+ FRAG_ATTRIB_VAR0 = 12, /**< shader varying */
+ FRAG_ATTRIB_MAX = (FRAG_ATTRIB_VAR0 + MAX_VARYING)
};
/**
#define FRAG_BIT_TEX5 (1 << FRAG_ATTRIB_TEX5)
#define FRAG_BIT_TEX6 (1 << FRAG_ATTRIB_TEX6)
#define FRAG_BIT_TEX7 (1 << FRAG_ATTRIB_TEX7)
+#define FRAG_BIT_VAR0 (1 << FRAG_ATTRIB_VAR0)
+
+#define FRAG_BIT_TEX(U) (FRAG_BIT_TEX0 << (U))
+#define FRAG_BIT_VAR(V) (FRAG_BIT_VAR0 << (V))
#define FRAG_BITS_TEX_ANY (FRAG_BIT_TEX0| \
FRAG_BIT_TEX1| \
/**
* Fragment program results
*/
-/*@{*/
-#define FRAG_RESULT_COLR 0
-#define FRAG_RESULT_COLH 1
-#define FRAG_RESULT_DEPR 2
-#define FRAG_RESULT_MAX 3
-/*@}*/
+enum
+{
+ FRAG_RESULT_COLR = 0,
+ FRAG_RESULT_COLH = 1,
+ FRAG_RESULT_DEPR = 2,
+ FRAG_RESULT_DATA0 = 3,
+ FRAG_RESULT_MAX = (FRAG_RESULT_DATA0 + MAX_DRAW_BUFFERS)
+};
/**
BUFFER_BIT_COLOR7)
+ /** The pixel transfer path has three color tables: */
+ /*@{*/
+ #define COLORTABLE_PRECONVOLUTION 0
+ #define COLORTABLE_POSTCONVOLUTION 1
+ #define COLORTABLE_POSTCOLORMATRIX 2
+ #define COLORTABLE_MAX 3
+ /*@}*/
/**
GLboolean Blend;
GLbitfield ClipPlanes;
GLboolean ColorMaterial;
- GLboolean ColorTable; /* SGI_color_table */
- GLboolean PostColorMatrixColorTable; /* SGI_color_table */
- GLboolean PostConvolutionColorTable; /* SGI_color_table */
+ GLboolean ColorTable[COLORTABLE_MAX];
GLboolean Convolution1D;
GLboolean Convolution2D;
GLboolean Separable2D;
};
+ /**
+ * A pixelmap (see glPixelMap)
+ */
+ struct gl_pixelmap
+ {
+ GLint Size;
+ GLfloat Map[MAX_PIXEL_MAP_TABLE];
+ GLubyte Map8[MAX_PIXEL_MAP_TABLE]; /**< converted to 8-bit color */
+ };
+
+
+ /**
+ * Collection of all pixelmaps
+ */
+ struct gl_pixelmaps
+ {
+ struct gl_pixelmap RtoR; /**< i.e. GL_PIXEL_MAP_R_TO_R */
+ struct gl_pixelmap GtoG;
+ struct gl_pixelmap BtoB;
+ struct gl_pixelmap AtoA;
+ struct gl_pixelmap ItoR;
+ struct gl_pixelmap ItoG;
+ struct gl_pixelmap ItoB;
+ struct gl_pixelmap ItoA;
+ struct gl_pixelmap ItoI;
+ struct gl_pixelmap StoS;
+ };
+
+
/**
* Pixel attribute group (GL_PIXEL_MODE_BIT).
*/
struct gl_pixel_attrib
{
GLenum ReadBuffer; /**< source buffer for glRead/CopyPixels() */
+
+ /*--- Begin Pixel Transfer State ---*/
+ /* Fields are in the order in which they're applied... */
+
+ /* Scale & Bias (index shift, offset) */
GLfloat RedBias, RedScale;
GLfloat GreenBias, GreenScale;
GLfloat BlueBias, BlueScale;
GLfloat AlphaBias, AlphaScale;
GLfloat DepthBias, DepthScale;
GLint IndexShift, IndexOffset;
+
+ /* Pixel Maps */
+ /* Note: actual pixel maps are not part of this attrib group */
GLboolean MapColorFlag;
GLboolean MapStencilFlag;
- GLfloat ZoomX, ZoomY;
- /* XXX move these out of gl_pixel_attrib */
- GLint MapStoSsize; /**< Size of each pixel map */
- GLint MapItoIsize;
- GLint MapItoRsize;
- GLint MapItoGsize;
- GLint MapItoBsize;
- GLint MapItoAsize;
- GLint MapRtoRsize;
- GLint MapGtoGsize;
- GLint MapBtoBsize;
- GLint MapAtoAsize;
- GLint MapStoS[MAX_PIXEL_MAP_TABLE]; /**< Pixel map tables */
- GLfloat MapItoI[MAX_PIXEL_MAP_TABLE];
- GLfloat MapItoR[MAX_PIXEL_MAP_TABLE];
- GLfloat MapItoG[MAX_PIXEL_MAP_TABLE];
- GLfloat MapItoB[MAX_PIXEL_MAP_TABLE];
- GLfloat MapItoA[MAX_PIXEL_MAP_TABLE];
- GLubyte MapItoR8[MAX_PIXEL_MAP_TABLE]; /**< converted to 8-bit color */
- GLubyte MapItoG8[MAX_PIXEL_MAP_TABLE];
- GLubyte MapItoB8[MAX_PIXEL_MAP_TABLE];
- GLubyte MapItoA8[MAX_PIXEL_MAP_TABLE];
- GLfloat MapRtoR[MAX_PIXEL_MAP_TABLE];
- GLfloat MapGtoG[MAX_PIXEL_MAP_TABLE];
- GLfloat MapBtoB[MAX_PIXEL_MAP_TABLE];
- GLfloat MapAtoA[MAX_PIXEL_MAP_TABLE];
- /** GL_EXT_histogram */
- GLboolean HistogramEnabled;
- GLboolean MinMaxEnabled;
- /** GL_SGI_color_matrix */
- GLfloat PostColorMatrixScale[4]; /**< RGBA */
- GLfloat PostColorMatrixBias[4]; /**< RGBA */
- /** GL_SGI_color_table */
- GLfloat ColorTableScale[4];
- GLfloat ColorTableBias[4];
- GLboolean ColorTableEnabled;
- GLfloat PCCTscale[4];
- GLfloat PCCTbias[4];
- GLboolean PostConvolutionColorTableEnabled;
- GLfloat PCMCTscale[4];
- GLfloat PCMCTbias[4];
- GLboolean PostColorMatrixColorTableEnabled;
- /** GL_SGI_texture_color_table */
- GLfloat TextureColorTableScale[4];
- GLfloat TextureColorTableBias[4];
- /** Convolution */
+
+ /* There are multiple color table stages: */
+ GLboolean ColorTableEnabled[COLORTABLE_MAX];
+ GLfloat ColorTableScale[COLORTABLE_MAX][4]; /**< RGBA */
+ GLfloat ColorTableBias[COLORTABLE_MAX][4]; /**< RGBA */
+
+ /* Convolution (GL_EXT_convolution) */
GLboolean Convolution1DEnabled;
GLboolean Convolution2DEnabled;
GLboolean Separable2DEnabled;
GLfloat ConvolutionBorderColor[3][4];
GLenum ConvolutionBorderMode[3];
- GLfloat ConvolutionFilterScale[3][4];
- GLfloat ConvolutionFilterBias[3][4];
+ GLfloat ConvolutionFilterScale[3][4]; /**< RGBA */
+ GLfloat ConvolutionFilterBias[3][4]; /**< RGBA */
GLfloat PostConvolutionScale[4]; /**< RGBA */
GLfloat PostConvolutionBias[4]; /**< RGBA */
+
+ /* Color matrix (GL_SGI_color_matrix) */
+ /* Note: the color matrix is not part of this attrib group */
+ GLfloat PostColorMatrixScale[4]; /**< RGBA */
+ GLfloat PostColorMatrixBias[4]; /**< RGBA */
+
+ /* Histogram & minmax (GL_EXT_histogram) */
+ /* Note: histogram and minmax data are not part of this attrib group */
+ GLboolean HistogramEnabled;
+ GLboolean MinMaxEnabled;
+
+ /*--- End Pixel Transfer State ---*/
+
+ /* Pixel Zoom */
+ GLfloat ZoomX, ZoomY;
+
+ /** GL_SGI_texture_color_table */
+ GLfloat TextureColorTableScale[4];
+ GLfloat TextureColorTableBias[4];
};
/**
* Names of the various vertex/fragment program register files, etc.
+ *
* NOTE: first four tokens must fit into 2 bits (see t_vb_arbprogram.c)
* All values should fit in a 4-bit field.
+ *
+ * NOTE: PROGRAM_ENV_PARAM, PROGRAM_STATE_VAR, PROGRAM_NAMED_PARAM,
+ * PROGRAM_CONSTANT, and PROGRAM_UNIFORM can all be considered to
+ * be "uniform" variables since they can only be set outside glBegin/End.
+ * They're also all stored in the same Parameters array.
*/
enum register_file
{
- PROGRAM_TEMPORARY = 0,
- PROGRAM_LOCAL_PARAM = 1,
- PROGRAM_ENV_PARAM = 2,
- PROGRAM_STATE_VAR = 3,
- PROGRAM_INPUT = 4,
- PROGRAM_OUTPUT = 5,
- PROGRAM_NAMED_PARAM = 6,
- PROGRAM_CONSTANT = 7,
- PROGRAM_WRITE_ONLY = 8,
- PROGRAM_ADDRESS = 9,
- PROGRAM_UNDEFINED = 10, /* invalid value */
+ PROGRAM_TEMPORARY = 0, /**< machine->Temporary[] */
+ PROGRAM_LOCAL_PARAM = 1, /**< gl_program->LocalParams[] */
+ PROGRAM_ENV_PARAM = 2, /**< gl_program->Parameters[] */
+ PROGRAM_STATE_VAR = 3, /**< gl_program->Parameters[] */
+ PROGRAM_INPUT = 4, /**< machine->Inputs[] */
+ PROGRAM_OUTPUT = 5, /**< machine->Outputs[] */
+ PROGRAM_NAMED_PARAM = 6, /**< gl_program->Parameters[] */
+ PROGRAM_CONSTANT = 7, /**< gl_program->Parameters[] */
+ PROGRAM_UNIFORM = 8, /**< gl_program->Parameters[] */
+ PROGRAM_VARYING = 9, /**< machine->Inputs[]/Outputs[] */
+ PROGRAM_WRITE_ONLY = 10, /**< A dummy, write-only register */
+ PROGRAM_ADDRESS = 11, /**< machine->AddressReg */
+ PROGRAM_SAMPLER = 12, /**< for shader samplers, compile-time only */
+ PROGRAM_UNDEFINED = 13, /**< Invalid value */
PROGRAM_FILE_MAX
};
struct gl_program
{
GLuint Id;
- GLubyte *String; /**< Null-terminated program text */
+ GLubyte *String; /**< Null-terminated program text */
GLint RefCount;
- GLenum Target;
- GLenum Format; /**< String encoding format */
+ GLenum Target; /**< GL_VERTEX/FRAGMENT_PROGRAM_ARB, GL_FRAGMENT_PROGRAM_NV */
+ GLenum Format; /**< String encoding format */
GLboolean Resident;
struct prog_instruction *Instructions;
- GLbitfield InputsRead; /* Bitmask of which input regs are read */
- GLbitfield OutputsWritten; /* Bitmask of which output regs are written to */
+ GLbitfield InputsRead; /**< Bitmask of which input regs are read */
+ GLbitfield OutputsWritten; /**< Bitmask of which output regs are written to */
+ GLbitfield TexturesUsed[MAX_TEXTURE_IMAGE_UNITS]; /**< TEXTURE_x_BIT bitmask */
/** Named parameters, constants, etc. from program text */
struct gl_program_parameter_list *Parameters;
/** Numbered local parameters */
GLfloat LocalParams[MAX_PROGRAM_LOCAL_PARAMS][4];
+ /** Vertex/fragment shader varying vars */
+ struct gl_program_parameter_list *Varying;
+ /** Vertex program user-defined attributes */
+ struct gl_program_parameter_list *Attributes;
+
/** Logical counts */
/*@{*/
GLuint NumInstructions;
GLuint NumParameters;
GLuint NumAttributes;
GLuint NumAddressRegs;
+ GLuint NumAluInstructions;
+ GLuint NumTexInstructions;
+ GLuint NumTexIndirections;
/*@}*/
/** Native, actual h/w counts */
/*@{*/
GLuint NumNativeParameters;
GLuint NumNativeAttributes;
GLuint NumNativeAddressRegs;
+ GLuint NumNativeAluInstructions;
+ GLuint NumNativeTexInstructions;
+ GLuint NumNativeTexIndirections;
/*@}*/
};
struct gl_fragment_program
{
struct gl_program Base; /**< base class */
- GLbitfield TexturesUsed[MAX_TEXTURE_IMAGE_UNITS]; /**< TEXTURE_x_BIT bitmask */
- GLuint NumAluInstructions; /**< GL_ARB_fragment_program */
- GLuint NumTexInstructions;
- GLuint NumTexIndirections;
- GLuint NumNativeAluInstructions; /**< GL_ARB_fragment_program */
- GLuint NumNativeTexInstructions;
- GLuint NumNativeTexIndirections;
GLenum FogOption;
GLboolean UsesKill;
};
GLboolean _Enabled; /**< Enabled and valid program? */
GLboolean PointSizeEnabled; /**< GL_VERTEX_PROGRAM_POINT_SIZE_ARB/NV */
GLboolean TwoSideEnabled; /**< GL_VERTEX_PROGRAM_TWO_SIDE_ARB/NV */
- struct gl_vertex_program *Current; /**< ptr to currently bound program */
- const struct gl_vertex_program *_Current; /**< ptr to currently bound
- program, including internal
- (t_vp_build.c) programs */
+ struct gl_vertex_program *Current; /**< user-bound vertex program */
- GLfloat Parameters[MAX_NV_VERTEX_PROGRAM_PARAMS][4]; /**< Env params */
+ /** Currently enabled and valid program (including internal programs
+ * and compiled shader programs).
+ */
+ struct gl_vertex_program *_Current;
+
+ GLfloat Parameters[MAX_PROGRAM_ENV_PARAMS][4]; /**< Env params */
/* For GL_NV_vertex_program only: */
- GLenum TrackMatrix[MAX_NV_VERTEX_PROGRAM_PARAMS / 4];
- GLenum TrackMatrixTransform[MAX_NV_VERTEX_PROGRAM_PARAMS / 4];
+ GLenum TrackMatrix[MAX_PROGRAM_ENV_PARAMS / 4];
+ GLenum TrackMatrixTransform[MAX_PROGRAM_ENV_PARAMS / 4];
+
+ /** Should fixed-function T&L be implemented with a vertex prog? */
+ GLboolean _MaintainTnlProgram;
+
+ /** Program to emulate fixed-function T&L (see above) */
+ struct gl_vertex_program *_TnlProgram;
#if FEATURE_MESA_program_debug
GLprogramcallbackMESA Callback;
{
GLboolean Enabled; /**< User-set fragment program enable flag */
GLboolean _Enabled; /**< Fragment program enabled and valid? */
- GLboolean _Active; /**< Is a user program or internal program active? */
- struct gl_fragment_program *Current; /**< User-bound program */
- const struct gl_fragment_program *_Current; /**< currently active program
- (including internal programs) */
- GLfloat Parameters[MAX_NV_FRAGMENT_PROGRAM_PARAMS][4]; /**< Env params */
+ struct gl_fragment_program *Current; /**< User-bound fragment program */
+
+ /** Currently enabled and valid program (including internal programs
+ * and compiled shader programs).
+ */
+ struct gl_fragment_program *_Current;
+
+ GLfloat Parameters[MAX_PROGRAM_ENV_PARAMS][4]; /**< Env params */
+
+ /** Should fixed-function texturing be implemented with a fragment prog? */
+ GLboolean _MaintainTexEnvProgram;
+
+ /** Program to emulate fixed-function texture env/combine (see above) */
+ struct gl_fragment_program *_TexEnvProgram;
#if FEATURE_MESA_program_debug
GLprogramcallbackMESA Callback;
};
+
/**
- * Context state for vertex/fragment shaders.
+ * A GLSL shader object.
*/
-struct gl_shader_objects_state
+struct gl_shader
{
- struct gl2_program_intf **CurrentProgram;
- GLboolean _VertexShaderPresent;
- GLboolean _FragmentShaderPresent;
+ GLenum Type; /**< GL_FRAGMENT_SHADER || GL_VERTEX_SHADER (first field!) */
+ GLuint Name; /**< AKA the handle */
+ GLint RefCount; /**< Reference count */
+ GLboolean DeletePending;
+
+ const GLchar *Source; /**< Source code string */
+ GLboolean CompileStatus;
+ GLuint NumPrograms; /**< size of Programs[] array */
+ struct gl_program **Programs; /**< Post-compile assembly code */
+ GLchar *InfoLog;
+};
+
+
+/**
+ * A GLSL program object. Basically a linked collection of "shaders".
+ */
+struct gl_shader_program
+{
+ GLenum Type; /**< Always GL_SHADER_PROGRAM (internal token) */
+ GLuint Name; /**< aka handle or ID */
+ GLint RefCount; /**< Reference count */
+ GLboolean DeletePending;
+
+ GLuint NumShaders; /**< number of attached shaders */
+ struct gl_shader **Shaders; /**< List of attached the shaders */
+
+ /* post-link info: */
+ struct gl_vertex_program *VertexProgram; /**< Linked vertex program */
+ struct gl_fragment_program *FragmentProgram; /**< Linked fragment prog */
+ struct gl_program_parameter_list *Uniforms; /**< Plus constants, etc */
+ struct gl_program_parameter_list *Varying;
+ struct gl_program_parameter_list *Attributes; /**< Vertex attributes */
+ GLboolean LinkStatus; /**< GL_LINK_STATUS */
+ GLboolean Validated;
+ GLchar *InfoLog;
+};
+
+
+/**
+ * Context state for GLSL vertex/fragment shaders.
+ */
+struct gl_shader_state
+{
+ struct gl_shader_program *CurrentProgram; /**< The user-bound program */
+ GLboolean EmitHighLevelInstructions; /**< Driver-selectable */
+ GLboolean EmitComments; /**< Driver-selectable */
};
#endif
#if FEATURE_ARB_shader_objects
- struct _mesa_HashTable *GL2Objects;
+ /** Table of both gl_shader and gl_shader_program objects */
+ struct _mesa_HashTable *ShaderObjects;
#endif
#if FEATURE_EXT_framebuffer_object
GLubyte IndexBits;
GLubyte DepthBits;
GLubyte StencilBits;
- GLvoid *Data;
+ GLvoid *Data; /**< This may not be used by some kinds of RBs */
/* Used to wrap one renderbuffer around another: */
struct gl_renderbuffer *Wrapped;
GLuint MaxRenderbufferSize;
/* GL_ARB_vertex_shader */
GLuint MaxVertexTextureImageUnits;
- GLuint MaxVaryingFloats;
+ GLuint MaxVarying;
};
/**
* A stack of matrices (projection, modelview, color, texture, etc).
*/
-struct matrix_stack
+struct gl_matrix_stack
{
GLmatrix *Top; /**< points into Stack */
GLmatrix *Stack; /**< array [MaxDepth] of GLmatrix */
#define IMAGE_HISTOGRAM_BIT 0x200
#define IMAGE_MIN_MAX_BIT 0x400
#define IMAGE_CLAMP_BIT 0x800 /* extra */
- #define IMAGE_RED_TO_LUMINANCE 0x1000
/** Pixel Transfer ops up to convolution */
/**
* State used during display list compilation and execution.
*/
-struct mesa_list_state
+struct gl_dlist_state
{
struct mesa_display_list *CallStack[MAX_LIST_NESTING];
GLuint CallDepth; /**< Current recursion calling depth */
struct dd_function_table Driver;
void *DriverCtx; /**< Points to device driver context/state */
- void *DriverMgrCtx; /**< Points to device driver manager (optional)*/
/** Core/Driver constants */
struct gl_constants Const;
/** \name The various 4x4 matrix stacks */
/*@{*/
- struct matrix_stack ModelviewMatrixStack;
- struct matrix_stack ProjectionMatrixStack;
- struct matrix_stack ColorMatrixStack;
- struct matrix_stack TextureMatrixStack[MAX_TEXTURE_COORD_UNITS];
- struct matrix_stack ProgramMatrixStack[MAX_PROGRAM_MATRICES];
- struct matrix_stack *CurrentStack; /**< Points to one of the above stacks */
+ struct gl_matrix_stack ModelviewMatrixStack;
+ struct gl_matrix_stack ProjectionMatrixStack;
+ struct gl_matrix_stack ColorMatrixStack;
+ struct gl_matrix_stack TextureMatrixStack[MAX_TEXTURE_COORD_UNITS];
+ struct gl_matrix_stack ProgramMatrixStack[MAX_PROGRAM_MATRICES];
+ struct gl_matrix_stack *CurrentStack; /**< Points to one of the above stacks */
/*@}*/
/** Combined modelview and projection matrix */
GLmatrix _ModelProjectMatrix;
/** \name Display lists */
- struct mesa_list_state ListState;
+ struct gl_dlist_state ListState;
GLboolean ExecuteFlag; /**< Execute GL commands? */
GLboolean CompileFlag; /**< Compile GL commands into display list? */
/** \name Other assorted state (not pushed/popped on attribute stack) */
/*@{*/
+ struct gl_pixelmaps PixelMaps;
struct gl_histogram_attrib Histogram;
struct gl_minmax_attrib MinMax;
struct gl_convolution_attrib Convolution1D;
struct gl_feedback Feedback; /**< Feedback */
struct gl_selection Select; /**< Selection */
- struct gl_color_table ColorTable; /**< Pre-convolution */
- struct gl_color_table ProxyColorTable; /**< Pre-convolution */
+ struct gl_color_table ColorTable[COLORTABLE_MAX];
+ struct gl_color_table ProxyColorTable[COLORTABLE_MAX];
+ #if 0
struct gl_color_table PostConvolutionColorTable;
struct gl_color_table ProxyPostConvolutionColorTable;
struct gl_color_table PostColorMatrixColorTable;
struct gl_color_table ProxyPostColorMatrixColorTable;
+ #endif
struct gl_program_state Program; /**< for vertex or fragment progs */
struct gl_vertex_program_state VertexProgram; /**< GL_ARB/NV_vertex_program */
struct gl_fragment_program_state FragmentProgram; /**< GL_ARB/NV_vertex_program */
struct gl_ati_fragment_shader_state ATIFragmentShader; /**< GL_ATI_fragment_shader */
- struct gl_fragment_program *_TexEnvProgram; /**< Texture state as fragment program */
- struct gl_vertex_program *_TnlProgram; /**< Fixed func TNL state as vertex program */
-
- GLboolean _MaintainTnlProgram;
- GLboolean _MaintainTexEnvProgram;
- GLboolean _UseTexEnvProgram;
-
struct gl_query_state Query; /**< GL_ARB_occlusion_query */
- struct gl_shader_objects_state ShaderObjects; /* GL_ARB_shader_objects */
+ struct gl_shader_state Shader; /**< GLSL shader object state */
/*@}*/
#if FEATURE_EXT_framebuffer_object
/***** glPixelMap *****/
/**********************************************************************/
+ /**
+ * Return pointer to a pixelmap by name.
+ */
+ static struct gl_pixelmap *
+ get_pixelmap(GLcontext *ctx, GLenum map)
+ {
+ switch (map) {
+ case GL_PIXEL_MAP_I_TO_I:
+ return &ctx->PixelMaps.ItoI;
+ case GL_PIXEL_MAP_S_TO_S:
+ return &ctx->PixelMaps.StoS;
+ case GL_PIXEL_MAP_I_TO_R:
+ return &ctx->PixelMaps.ItoR;
+ case GL_PIXEL_MAP_I_TO_G:
+ return &ctx->PixelMaps.ItoG;
+ case GL_PIXEL_MAP_I_TO_B:
+ return &ctx->PixelMaps.ItoB;
+ case GL_PIXEL_MAP_I_TO_A:
+ return &ctx->PixelMaps.ItoA;
+ case GL_PIXEL_MAP_R_TO_R:
+ return &ctx->PixelMaps.RtoR;
+ case GL_PIXEL_MAP_G_TO_G:
+ return &ctx->PixelMaps.GtoG;
+ case GL_PIXEL_MAP_B_TO_B:
+ return &ctx->PixelMaps.BtoB;
+ case GL_PIXEL_MAP_A_TO_A:
+ return &ctx->PixelMaps.AtoA;
+ default:
+ return NULL;
+ }
+ }
+
/**
* Helper routine used by the other _mesa_PixelMap() functions.
*/
static void
- pixelmap(GLcontext *ctx, GLenum map, GLsizei mapsize, const GLfloat *values)
+ store_pixelmap(GLcontext *ctx, GLenum map, GLsizei mapsize,
+ const GLfloat *values)
{
GLint i;
+ struct gl_pixelmap *pm = get_pixelmap(ctx, map);
+ if (!pm) {
+ _mesa_error(ctx, GL_INVALID_ENUM, "glPixelMap(map)");
+ return;
+ }
+
switch (map) {
- case GL_PIXEL_MAP_S_TO_S:
- ctx->Pixel.MapStoSsize = mapsize;
- for (i = 0; i < mapsize; i++) {
- ctx->Pixel.MapStoS[i] = IROUND(values[i]);
- }
- break;
- case GL_PIXEL_MAP_I_TO_I:
- ctx->Pixel.MapItoIsize = mapsize;
- for (i = 0; i < mapsize; i++) {
- ctx->Pixel.MapItoI[i] = values[i];
- }
- break;
- case GL_PIXEL_MAP_I_TO_R:
- ctx->Pixel.MapItoRsize = mapsize;
- for (i = 0; i < mapsize; i++) {
- GLfloat val = CLAMP( values[i], 0.0F, 1.0F );
- ctx->Pixel.MapItoR[i] = val;
- ctx->Pixel.MapItoR8[i] = (GLint) (val * 255.0F);
- }
- break;
- case GL_PIXEL_MAP_I_TO_G:
- ctx->Pixel.MapItoGsize = mapsize;
- for (i = 0; i < mapsize; i++) {
- GLfloat val = CLAMP( values[i], 0.0F, 1.0F );
- ctx->Pixel.MapItoG[i] = val;
- ctx->Pixel.MapItoG8[i] = (GLint) (val * 255.0F);
- }
- break;
- case GL_PIXEL_MAP_I_TO_B:
- ctx->Pixel.MapItoBsize = mapsize;
- for (i = 0; i < mapsize; i++) {
- GLfloat val = CLAMP( values[i], 0.0F, 1.0F );
- ctx->Pixel.MapItoB[i] = val;
- ctx->Pixel.MapItoB8[i] = (GLint) (val * 255.0F);
- }
- break;
- case GL_PIXEL_MAP_I_TO_A:
- ctx->Pixel.MapItoAsize = mapsize;
- for (i = 0; i < mapsize; i++) {
- GLfloat val = CLAMP( values[i], 0.0F, 1.0F );
- ctx->Pixel.MapItoA[i] = val;
- ctx->Pixel.MapItoA8[i] = (GLint) (val * 255.0F);
- }
- break;
- case GL_PIXEL_MAP_R_TO_R:
- ctx->Pixel.MapRtoRsize = mapsize;
- for (i = 0; i < mapsize; i++) {
- ctx->Pixel.MapRtoR[i] = CLAMP( values[i], 0.0F, 1.0F );
- }
- break;
- case GL_PIXEL_MAP_G_TO_G:
- ctx->Pixel.MapGtoGsize = mapsize;
- for (i = 0; i < mapsize; i++) {
- ctx->Pixel.MapGtoG[i] = CLAMP( values[i], 0.0F, 1.0F );
- }
- break;
- case GL_PIXEL_MAP_B_TO_B:
- ctx->Pixel.MapBtoBsize = mapsize;
- for (i = 0; i < mapsize; i++) {
- ctx->Pixel.MapBtoB[i] = CLAMP( values[i], 0.0F, 1.0F );
- }
- break;
- case GL_PIXEL_MAP_A_TO_A:
- ctx->Pixel.MapAtoAsize = mapsize;
- for (i = 0; i < mapsize; i++) {
- ctx->Pixel.MapAtoA[i] = CLAMP( values[i], 0.0F, 1.0F );
- }
- break;
- default:
- _mesa_error( ctx, GL_INVALID_ENUM, "glPixelMap(map)" );
+ case GL_PIXEL_MAP_S_TO_S:
+ /* special case */
+ ctx->PixelMaps.StoS.Size = mapsize;
+ for (i = 0; i < mapsize; i++) {
+ ctx->PixelMaps.StoS.Map[i] = IROUND(values[i]);
+ }
+ break;
+ case GL_PIXEL_MAP_I_TO_I:
+ /* special case */
+ ctx->PixelMaps.ItoI.Size = mapsize;
+ for (i = 0; i < mapsize; i++) {
+ ctx->PixelMaps.ItoI.Map[i] = values[i];
+ }
+ break;
+ default:
+ /* general case */
+ pm->Size = mapsize;
+ for (i = 0; i < mapsize; i++) {
+ GLfloat val = CLAMP(values[i], 0.0F, 1.0F);
+ pm->Map[i] = val;
+ pm->Map8[i] = (GLint) (val * 255.0F);
+ }
}
}
return;
}
- pixelmap(ctx, map, mapsize, values);
+ store_pixelmap(ctx, map, mapsize, values);
if (ctx->Unpack.BufferObj->Name) {
ctx->Driver.UnmapBuffer(ctx, GL_PIXEL_UNPACK_BUFFER_EXT,
}
-
void GLAPIENTRY
_mesa_PixelMapuiv(GLenum map, GLsizei mapsize, const GLuint *values )
{
ctx->Unpack.BufferObj);
}
- pixelmap(ctx, map, mapsize, fvalues);
+ store_pixelmap(ctx, map, mapsize, fvalues);
}
-
void GLAPIENTRY
_mesa_PixelMapusv(GLenum map, GLsizei mapsize, const GLushort *values )
{
return;
}
- /* convert to floats */
+ /* convert to floats */
if (map == GL_PIXEL_MAP_I_TO_I || map == GL_PIXEL_MAP_S_TO_S) {
GLint i;
for (i = 0; i < mapsize; i++) {
ctx->Unpack.BufferObj);
}
- pixelmap(ctx, map, mapsize, fvalues);
- }
-
-
- /**
- * Return size of the named map.
- */
- static GLuint
- get_map_size(GLcontext *ctx, GLenum map)
- {
- switch (map) {
- case GL_PIXEL_MAP_I_TO_I:
- return ctx->Pixel.MapItoIsize;
- case GL_PIXEL_MAP_S_TO_S:
- return ctx->Pixel.MapStoSsize;
- case GL_PIXEL_MAP_I_TO_R:
- return ctx->Pixel.MapItoRsize;
- case GL_PIXEL_MAP_I_TO_G:
- return ctx->Pixel.MapItoGsize;
- case GL_PIXEL_MAP_I_TO_B:
- return ctx->Pixel.MapItoBsize;
- case GL_PIXEL_MAP_I_TO_A:
- return ctx->Pixel.MapItoAsize;
- case GL_PIXEL_MAP_R_TO_R:
- return ctx->Pixel.MapRtoRsize;
- case GL_PIXEL_MAP_G_TO_G:
- return ctx->Pixel.MapGtoGsize;
- case GL_PIXEL_MAP_B_TO_B:
- return ctx->Pixel.MapBtoBsize;
- case GL_PIXEL_MAP_A_TO_A:
- return ctx->Pixel.MapAtoAsize;
- default:
- return 0;
- }
+ store_pixelmap(ctx, map, mapsize, fvalues);
}
{
GET_CURRENT_CONTEXT(ctx);
GLuint mapsize, i;
+ const struct gl_pixelmap *pm;
+
ASSERT_OUTSIDE_BEGIN_END(ctx);
- mapsize = get_map_size(ctx, map);
+ pm = get_pixelmap(ctx, map);
+ if (!pm) {
+ _mesa_error(ctx, GL_INVALID_ENUM, "glGetPixelMapfv(map)");
+ return;
+ }
+
+ mapsize = pm->Size;
if (ctx->Pack.BufferObj->Name) {
/* pack pixelmap into PBO */
return;
}
- switch (map) {
- case GL_PIXEL_MAP_I_TO_I:
- MEMCPY(values, ctx->Pixel.MapItoI, mapsize * sizeof(GLfloat));
- break;
- case GL_PIXEL_MAP_S_TO_S:
- for (i = 0; i < mapsize; i++) {
- values[i] = (GLfloat) ctx->Pixel.MapStoS[i];
- }
- break;
- case GL_PIXEL_MAP_I_TO_R:
- MEMCPY(values, ctx->Pixel.MapItoR, mapsize * sizeof(GLfloat));
- break;
- case GL_PIXEL_MAP_I_TO_G:
- MEMCPY(values, ctx->Pixel.MapItoG, mapsize * sizeof(GLfloat));
- break;
- case GL_PIXEL_MAP_I_TO_B:
- MEMCPY(values, ctx->Pixel.MapItoB, mapsize * sizeof(GLfloat));
- break;
- case GL_PIXEL_MAP_I_TO_A:
- MEMCPY(values, ctx->Pixel.MapItoA, mapsize * sizeof(GLfloat));
- break;
- case GL_PIXEL_MAP_R_TO_R:
- MEMCPY(values, ctx->Pixel.MapRtoR, mapsize * sizeof(GLfloat));
- break;
- case GL_PIXEL_MAP_G_TO_G:
- MEMCPY(values, ctx->Pixel.MapGtoG, mapsize * sizeof(GLfloat));
- break;
- case GL_PIXEL_MAP_B_TO_B:
- MEMCPY(values, ctx->Pixel.MapBtoB, mapsize * sizeof(GLfloat));
- break;
- case GL_PIXEL_MAP_A_TO_A:
- MEMCPY(values, ctx->Pixel.MapAtoA, mapsize * sizeof(GLfloat));
- break;
- default:
- _mesa_error( ctx, GL_INVALID_ENUM, "glGetPixelMapfv" );
+ if (map == GL_PIXEL_MAP_S_TO_S) {
+ /* special case */
+ for (i = 0; i < mapsize; i++) {
+ values[i] = (GLfloat) ctx->PixelMaps.StoS.Map[i];
+ }
+ }
+ else {
+ MEMCPY(values, pm->Map, mapsize * sizeof(GLfloat));
}
if (ctx->Pack.BufferObj->Name) {
{
GET_CURRENT_CONTEXT(ctx);
GLint mapsize, i;
+ const struct gl_pixelmap *pm;
+
ASSERT_OUTSIDE_BEGIN_END(ctx);
- mapsize = get_map_size(ctx, map);
+ pm = get_pixelmap(ctx, map);
+ if (!pm) {
+ _mesa_error(ctx, GL_INVALID_ENUM, "glGetPixelMapuiv(map)");
+ return;
+ }
+ mapsize = pm->Size;
if (ctx->Pack.BufferObj->Name) {
/* pack pixelmap into PBO */
return;
}
- switch (map) {
- case GL_PIXEL_MAP_I_TO_I:
- for (i = 0; i < mapsize; i++) {
- values[i] = FLOAT_TO_UINT( ctx->Pixel.MapItoI[i] );
- }
- break;
- case GL_PIXEL_MAP_S_TO_S:
- MEMCPY(values, ctx->Pixel.MapStoS, mapsize * sizeof(GLint));
- break;
- case GL_PIXEL_MAP_I_TO_R:
- for (i = 0; i < mapsize; i++) {
- values[i] = FLOAT_TO_UINT( ctx->Pixel.MapItoR[i] );
- }
- break;
- case GL_PIXEL_MAP_I_TO_G:
- for (i = 0; i < mapsize; i++) {
- values[i] = FLOAT_TO_UINT( ctx->Pixel.MapItoG[i] );
- }
- break;
- case GL_PIXEL_MAP_I_TO_B:
- for (i = 0; i < mapsize; i++) {
- values[i] = FLOAT_TO_UINT( ctx->Pixel.MapItoB[i] );
- }
- break;
- case GL_PIXEL_MAP_I_TO_A:
- for (i = 0; i < mapsize; i++) {
- values[i] = FLOAT_TO_UINT( ctx->Pixel.MapItoA[i] );
- }
- break;
- case GL_PIXEL_MAP_R_TO_R:
- for (i = 0; i < mapsize; i++) {
- values[i] = FLOAT_TO_UINT( ctx->Pixel.MapRtoR[i] );
- }
- break;
- case GL_PIXEL_MAP_G_TO_G:
- for (i = 0; i < mapsize; i++) {
- values[i] = FLOAT_TO_UINT( ctx->Pixel.MapGtoG[i] );
- }
- break;
- case GL_PIXEL_MAP_B_TO_B:
- for (i = 0; i < mapsize; i++) {
- values[i] = FLOAT_TO_UINT( ctx->Pixel.MapBtoB[i] );
- }
- break;
- case GL_PIXEL_MAP_A_TO_A:
- for (i = 0; i < mapsize; i++) {
- values[i] = FLOAT_TO_UINT( ctx->Pixel.MapAtoA[i] );
- }
- break;
- default:
- _mesa_error( ctx, GL_INVALID_ENUM, "glGetPixelMapfv" );
+ if (map == GL_PIXEL_MAP_S_TO_S) {
+ /* special case */
+ MEMCPY(values, ctx->PixelMaps.StoS.Map, mapsize * sizeof(GLint));
+ }
+ else {
+ for (i = 0; i < mapsize; i++) {
+ values[i] = FLOAT_TO_UINT( pm->Map[i] );
+ }
}
if (ctx->Pack.BufferObj->Name) {
{
GET_CURRENT_CONTEXT(ctx);
GLint mapsize, i;
+ const struct gl_pixelmap *pm;
+
ASSERT_OUTSIDE_BEGIN_END(ctx);
- mapsize = get_map_size(ctx, map);
+ pm = get_pixelmap(ctx, map);
+ if (!pm) {
+ _mesa_error(ctx, GL_INVALID_ENUM, "glGetPixelMapusv(map)");
+ return;
+ }
+ mapsize = pm ? pm->Size : 0;
if (ctx->Pack.BufferObj->Name) {
/* pack pixelmap into PBO */
}
switch (map) {
- case GL_PIXEL_MAP_I_TO_I:
- for (i = 0; i < mapsize; i++) {
- values[i] = (GLushort) CLAMP(ctx->Pixel.MapItoI[i], 0.0, 65535.0);
- }
- break;
- case GL_PIXEL_MAP_S_TO_S:
- for (i = 0; i < mapsize; i++) {
- values[i] = (GLushort) CLAMP(ctx->Pixel.MapStoS[i], 0.0, 65535.0);
- }
- break;
- case GL_PIXEL_MAP_I_TO_R:
- for (i = 0; i < mapsize; i++) {
- CLAMPED_FLOAT_TO_USHORT(values[i] , ctx->Pixel.MapItoR[i] );
- }
- break;
- case GL_PIXEL_MAP_I_TO_G:
- for (i = 0; i < mapsize; i++) {
- CLAMPED_FLOAT_TO_USHORT(values[i] , ctx->Pixel.MapItoG[i] );
- }
- break;
- case GL_PIXEL_MAP_I_TO_B:
- for (i = 0; i < mapsize; i++) {
- CLAMPED_FLOAT_TO_USHORT(values[i] , ctx->Pixel.MapItoB[i] );
- }
- break;
- case GL_PIXEL_MAP_I_TO_A:
- for (i = 0; i < mapsize; i++) {
- CLAMPED_FLOAT_TO_USHORT(values[i] , ctx->Pixel.MapItoA[i] );
- }
- break;
- case GL_PIXEL_MAP_R_TO_R:
- for (i = 0; i < mapsize; i++) {
- CLAMPED_FLOAT_TO_USHORT(values[i] , ctx->Pixel.MapRtoR[i] );
- }
- break;
- case GL_PIXEL_MAP_G_TO_G:
- for (i = 0; i < mapsize; i++) {
- CLAMPED_FLOAT_TO_USHORT(values[i] , ctx->Pixel.MapGtoG[i] );
- }
- break;
- case GL_PIXEL_MAP_B_TO_B:
- for (i = 0; i < mapsize; i++) {
- CLAMPED_FLOAT_TO_USHORT(values[i] , ctx->Pixel.MapBtoB[i] );
- }
- break;
- case GL_PIXEL_MAP_A_TO_A:
- for (i = 0; i < mapsize; i++) {
- CLAMPED_FLOAT_TO_USHORT(values[i] , ctx->Pixel.MapAtoA[i] );
- }
- break;
- default:
- _mesa_error( ctx, GL_INVALID_ENUM, "glGetPixelMapfv" );
+ /* special cases */
+ case GL_PIXEL_MAP_I_TO_I:
+ for (i = 0; i < mapsize; i++) {
+ values[i] = (GLushort) CLAMP(ctx->PixelMaps.ItoI.Map[i], 0.0, 65535.);
+ }
+ break;
+ case GL_PIXEL_MAP_S_TO_S:
+ for (i = 0; i < mapsize; i++) {
+ values[i] = (GLushort) CLAMP(ctx->PixelMaps.StoS.Map[i], 0.0, 65535.);
+ }
+ break;
+ default:
+ for (i = 0; i < mapsize; i++) {
+ CLAMPED_FLOAT_TO_USHORT(values[i], pm->Map[i] );
+ }
}
if (ctx->Pack.BufferObj->Name) {
ctx->Pixel.PostConvolutionBias[2] = param;
break;
case GL_POST_CONVOLUTION_ALPHA_SCALE:
- if (ctx->Pixel.PostConvolutionScale[2] == param)
+ if (ctx->Pixel.PostConvolutionScale[3] == param)
return;
FLUSH_VERTICES(ctx, _NEW_PIXEL);
- ctx->Pixel.PostConvolutionScale[2] = param;
+ ctx->Pixel.PostConvolutionScale[3] = param;
break;
case GL_POST_CONVOLUTION_ALPHA_BIAS:
- if (ctx->Pixel.PostConvolutionBias[2] == param)
+ if (ctx->Pixel.PostConvolutionBias[3] == param)
return;
FLUSH_VERTICES(ctx, _NEW_PIXEL);
- ctx->Pixel.PostConvolutionBias[2] = param;
+ ctx->Pixel.PostConvolutionBias[3] = param;
break;
default:
_mesa_error( ctx, GL_INVALID_ENUM, "glPixelTransfer(pname)" );
void
_mesa_map_rgba( const GLcontext *ctx, GLuint n, GLfloat rgba[][4] )
{
- const GLfloat rscale = (GLfloat) (ctx->Pixel.MapRtoRsize - 1);
- const GLfloat gscale = (GLfloat) (ctx->Pixel.MapGtoGsize - 1);
- const GLfloat bscale = (GLfloat) (ctx->Pixel.MapBtoBsize - 1);
- const GLfloat ascale = (GLfloat) (ctx->Pixel.MapAtoAsize - 1);
- const GLfloat *rMap = ctx->Pixel.MapRtoR;
- const GLfloat *gMap = ctx->Pixel.MapGtoG;
- const GLfloat *bMap = ctx->Pixel.MapBtoB;
- const GLfloat *aMap = ctx->Pixel.MapAtoA;
+ const GLfloat rscale = (GLfloat) (ctx->PixelMaps.RtoR.Size - 1);
+ const GLfloat gscale = (GLfloat) (ctx->PixelMaps.GtoG.Size - 1);
+ const GLfloat bscale = (GLfloat) (ctx->PixelMaps.BtoB.Size - 1);
+ const GLfloat ascale = (GLfloat) (ctx->PixelMaps.AtoA.Size - 1);
+ const GLfloat *rMap = ctx->PixelMaps.RtoR.Map;
+ const GLfloat *gMap = ctx->PixelMaps.GtoG.Map;
+ const GLfloat *bMap = ctx->PixelMaps.BtoB.Map;
+ const GLfloat *aMap = ctx->PixelMaps.AtoA.Map;
GLuint i;
for (i=0;i<n;i++) {
GLfloat r = CLAMP(rgba[i][RCOMP], 0.0F, 1.0F);
_mesa_map_ci_to_rgba( const GLcontext *ctx, GLuint n,
const GLuint index[], GLfloat rgba[][4] )
{
- GLuint rmask = ctx->Pixel.MapItoRsize - 1;
- GLuint gmask = ctx->Pixel.MapItoGsize - 1;
- GLuint bmask = ctx->Pixel.MapItoBsize - 1;
- GLuint amask = ctx->Pixel.MapItoAsize - 1;
- const GLfloat *rMap = ctx->Pixel.MapItoR;
- const GLfloat *gMap = ctx->Pixel.MapItoG;
- const GLfloat *bMap = ctx->Pixel.MapItoB;
- const GLfloat *aMap = ctx->Pixel.MapItoA;
+ GLuint rmask = ctx->PixelMaps.ItoR.Size - 1;
+ GLuint gmask = ctx->PixelMaps.ItoG.Size - 1;
+ GLuint bmask = ctx->PixelMaps.ItoB.Size - 1;
+ GLuint amask = ctx->PixelMaps.ItoA.Size - 1;
+ const GLfloat *rMap = ctx->PixelMaps.ItoR.Map;
+ const GLfloat *gMap = ctx->PixelMaps.ItoG.Map;
+ const GLfloat *bMap = ctx->PixelMaps.ItoB.Map;
+ const GLfloat *aMap = ctx->PixelMaps.ItoA.Map;
GLuint i;
for (i=0;i<n;i++) {
rgba[i][RCOMP] = rMap[index[i] & rmask];
_mesa_map_ci8_to_rgba8(const GLcontext *ctx, GLuint n, const GLubyte index[],
GLubyte rgba[][4])
{
- GLuint rmask = ctx->Pixel.MapItoRsize - 1;
- GLuint gmask = ctx->Pixel.MapItoGsize - 1;
- GLuint bmask = ctx->Pixel.MapItoBsize - 1;
- GLuint amask = ctx->Pixel.MapItoAsize - 1;
- const GLubyte *rMap = ctx->Pixel.MapItoR8;
- const GLubyte *gMap = ctx->Pixel.MapItoG8;
- const GLubyte *bMap = ctx->Pixel.MapItoB8;
- const GLubyte *aMap = ctx->Pixel.MapItoA8;
+ GLuint rmask = ctx->PixelMaps.ItoR.Size - 1;
+ GLuint gmask = ctx->PixelMaps.ItoG.Size - 1;
+ GLuint bmask = ctx->PixelMaps.ItoB.Size - 1;
+ GLuint amask = ctx->PixelMaps.ItoA.Size - 1;
+ const GLubyte *rMap = ctx->PixelMaps.ItoR.Map8;
+ const GLubyte *gMap = ctx->PixelMaps.ItoG.Map8;
+ const GLubyte *bMap = ctx->PixelMaps.ItoB.Map8;
+ const GLubyte *aMap = ctx->PixelMaps.ItoA.Map8;
GLuint i;
for (i=0;i<n;i++) {
rgba[i][RCOMP] = rMap[index[i] & rmask];
if (ctx->Pixel.MapColorFlag)
mask |= IMAGE_MAP_COLOR_BIT;
- if (ctx->Pixel.ColorTableEnabled)
+ if (ctx->Pixel.ColorTableEnabled[COLORTABLE_PRECONVOLUTION])
mask |= IMAGE_COLOR_TABLE_BIT;
if (ctx->Pixel.Convolution1DEnabled ||
}
}
- if (ctx->Pixel.PostConvolutionColorTableEnabled)
+ if (ctx->Pixel.ColorTableEnabled[COLORTABLE_POSTCONVOLUTION])
mask |= IMAGE_POST_CONVOLUTION_COLOR_TABLE_BIT;
if (ctx->ColorMatrixStack.Top->type != MATRIX_IDENTITY ||
ctx->Pixel.PostColorMatrixBias[3] != 0.0F)
mask |= IMAGE_COLOR_MATRIX_BIT;
- if (ctx->Pixel.PostColorMatrixColorTableEnabled)
+ if (ctx->Pixel.ColorTableEnabled[COLORTABLE_POSTCOLORMATRIX])
mask |= IMAGE_POST_COLOR_MATRIX_COLOR_TABLE_BIT;
if (ctx->Pixel.HistogramEnabled)
/***** Initialization *****/
/**********************************************************************/
+ static void
+ init_pixelmap(struct gl_pixelmap *map)
+ {
+ map->Size = 1;
+ map->Map[0] = 0.0;
+ map->Map8[0] = 0;
+ }
+
/**
* Initialize the context's PIXEL attribute group.
ctx->Pixel.ZoomY = 1.0;
ctx->Pixel.MapColorFlag = GL_FALSE;
ctx->Pixel.MapStencilFlag = GL_FALSE;
- ctx->Pixel.MapStoSsize = 1;
- ctx->Pixel.MapItoIsize = 1;
- ctx->Pixel.MapItoRsize = 1;
- ctx->Pixel.MapItoGsize = 1;
- ctx->Pixel.MapItoBsize = 1;
- ctx->Pixel.MapItoAsize = 1;
- ctx->Pixel.MapRtoRsize = 1;
- ctx->Pixel.MapGtoGsize = 1;
- ctx->Pixel.MapBtoBsize = 1;
- ctx->Pixel.MapAtoAsize = 1;
- ctx->Pixel.MapStoS[0] = 0;
- ctx->Pixel.MapItoI[0] = 0.0;
- ctx->Pixel.MapItoR[0] = 0.0;
- ctx->Pixel.MapItoG[0] = 0.0;
- ctx->Pixel.MapItoB[0] = 0.0;
- ctx->Pixel.MapItoA[0] = 0.0;
- ctx->Pixel.MapItoR8[0] = 0;
- ctx->Pixel.MapItoG8[0] = 0;
- ctx->Pixel.MapItoB8[0] = 0;
- ctx->Pixel.MapItoA8[0] = 0;
- ctx->Pixel.MapRtoR[0] = 0.0;
- ctx->Pixel.MapGtoG[0] = 0.0;
- ctx->Pixel.MapBtoB[0] = 0.0;
- ctx->Pixel.MapAtoA[0] = 0.0;
+ init_pixelmap(&ctx->PixelMaps.StoS);
+ init_pixelmap(&ctx->PixelMaps.ItoI);
+ init_pixelmap(&ctx->PixelMaps.ItoR);
+ init_pixelmap(&ctx->PixelMaps.ItoG);
+ init_pixelmap(&ctx->PixelMaps.ItoB);
+ init_pixelmap(&ctx->PixelMaps.ItoA);
+ init_pixelmap(&ctx->PixelMaps.RtoR);
+ init_pixelmap(&ctx->PixelMaps.GtoG);
+ init_pixelmap(&ctx->PixelMaps.BtoB);
+ init_pixelmap(&ctx->PixelMaps.AtoA);
ctx->Pixel.HistogramEnabled = GL_FALSE;
ctx->Pixel.MinMaxEnabled = GL_FALSE;
ASSIGN_4V(ctx->Pixel.PostColorMatrixScale, 1.0, 1.0, 1.0, 1.0);
ASSIGN_4V(ctx->Pixel.PostColorMatrixBias, 0.0, 0.0, 0.0, 0.0);
- ASSIGN_4V(ctx->Pixel.ColorTableScale, 1.0, 1.0, 1.0, 1.0);
- ASSIGN_4V(ctx->Pixel.ColorTableBias, 0.0, 0.0, 0.0, 0.0);
- ASSIGN_4V(ctx->Pixel.PCCTscale, 1.0, 1.0, 1.0, 1.0);
- ASSIGN_4V(ctx->Pixel.PCCTbias, 0.0, 0.0, 0.0, 0.0);
- ASSIGN_4V(ctx->Pixel.PCMCTscale, 1.0, 1.0, 1.0, 1.0);
- ASSIGN_4V(ctx->Pixel.PCMCTbias, 0.0, 0.0, 0.0, 0.0);
- ctx->Pixel.ColorTableEnabled = GL_FALSE;
- ctx->Pixel.PostConvolutionColorTableEnabled = GL_FALSE;
- ctx->Pixel.PostColorMatrixColorTableEnabled = GL_FALSE;
+ for (i = 0; i < COLORTABLE_MAX; i++) {
+ ASSIGN_4V(ctx->Pixel.ColorTableScale[i], 1.0, 1.0, 1.0, 1.0);
+ ASSIGN_4V(ctx->Pixel.ColorTableBias[i], 0.0, 0.0, 0.0, 0.0);
+ ctx->Pixel.ColorTableEnabled[i] = GL_FALSE;
+ }
ctx->Pixel.Convolution1DEnabled = GL_FALSE;
ctx->Pixel.Convolution2DEnabled = GL_FALSE;
ctx->Pixel.Separable2DEnabled = GL_FALSE;
#include "glheader.h"
#include "context.h"
-#include "hash.h"
#include "imports.h"
#include "macros.h"
-#include "mtypes.h"
#include "nvprogram.h"
#include "nvvertparse.h"
-#include "program_instruction.h"
+#include "prog_instruction.h"
#include "program.h"
if (token[1] == 0) {
/* single letter swizzle */
if (token[0] == 'x')
- srcReg->Swizzle = MAKE_SWIZZLE4(0, 0, 0, 0);
+ srcReg->Swizzle = SWIZZLE_XXXX;
else if (token[0] == 'y')
- srcReg->Swizzle = MAKE_SWIZZLE4(1, 1, 1, 1);
+ srcReg->Swizzle = SWIZZLE_YYYY;
else if (token[0] == 'z')
- srcReg->Swizzle = MAKE_SWIZZLE4(2, 2, 2, 2);
+ srcReg->Swizzle = SWIZZLE_ZZZZ;
else if (token[0] == 'w')
- srcReg->Swizzle = MAKE_SWIZZLE4(3, 3, 3, 3);
+ srcReg->Swizzle = SWIZZLE_WWWW;
else
RETURN_ERROR1("Expected x, y, z, or w");
}
--- /dev/null
+/*
+ * Mesa 3-D graphics library
+ * Version: 6.5.3
+ *
+ * Copyright (C) 1999-2005 Brian Paul All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+#include "glheader.h"
+#include "imports.h"
+#include "mtypes.h"
+#include "prog_instruction.h"
+
+
+/**
+ * Initialize program instruction fields to defaults.
+ * \param inst first instruction to initialize
+ * \param count number of instructions to initialize
+ */
+void
+_mesa_init_instructions(struct prog_instruction *inst, GLuint count)
+{
+ GLuint i;
+
+ _mesa_bzero(inst, count * sizeof(struct prog_instruction));
+
+ for (i = 0; i < count; i++) {
+ inst[i].SrcReg[0].File = PROGRAM_UNDEFINED;
+ inst[i].SrcReg[0].Swizzle = SWIZZLE_NOOP;
+ inst[i].SrcReg[1].File = PROGRAM_UNDEFINED;
+ inst[i].SrcReg[1].Swizzle = SWIZZLE_NOOP;
+ inst[i].SrcReg[2].File = PROGRAM_UNDEFINED;
+ inst[i].SrcReg[2].Swizzle = SWIZZLE_NOOP;
+
+ inst[i].DstReg.File = PROGRAM_UNDEFINED;
+ inst[i].DstReg.WriteMask = WRITEMASK_XYZW;
+ inst[i].DstReg.CondMask = COND_TR;
+ inst[i].DstReg.CondSwizzle = SWIZZLE_NOOP;
+
+ inst[i].SaturateMode = SATURATE_OFF;
+ inst[i].Precision = FLOAT32;
+ }
+}
+
+
+/**
+ * Allocate an array of program instructions.
+ * \param numInst number of instructions
+ * \return pointer to instruction memory
+ */
+struct prog_instruction *
+_mesa_alloc_instructions(GLuint numInst)
+{
+ return (struct prog_instruction *)
+ _mesa_calloc(numInst * sizeof(struct prog_instruction));
+}
+
+
+/**
+ * Reallocate memory storing an array of program instructions.
+ * This is used when we need to append additional instructions onto an
+ * program.
+ * \param oldInst pointer to first of old/src instructions
+ * \param numOldInst number of instructions at <oldInst>
+ * \param numNewInst desired size of new instruction array.
+ * \return pointer to start of new instruction array.
+ */
+struct prog_instruction *
+_mesa_realloc_instructions(struct prog_instruction *oldInst,
+ GLuint numOldInst, GLuint numNewInst)
+{
+ struct prog_instruction *newInst;
+
+ newInst = (struct prog_instruction *)
+ _mesa_realloc(oldInst,
+ numOldInst * sizeof(struct prog_instruction),
+ numNewInst * sizeof(struct prog_instruction));
+
+ return newInst;
+}
+
+
++/**
++ * Copy an array of program instructions.
++ * \param dest pointer to destination.
++ * \param src pointer to source.
++ * \param n number of instructions to copy.
++ * \return pointer to destination.
++ */
++struct prog_instruction *
++_mesa_copy_instructions(struct prog_instruction *dest,
++ const struct prog_instruction *src, GLuint n)
++{
++ return _mesa_memcpy(dest, src, n * sizeof(struct prog_instruction));
++}
++
+
+/**
+ * Basic info about each instruction
+ */
+struct instruction_info
+{
+ gl_inst_opcode Opcode;
+ const char *Name;
+ GLuint NumSrcRegs;
+};
+
+/**
+ * Instruction info
+ * \note Opcode should equal array index!
+ */
+static const struct instruction_info InstInfo[MAX_OPCODE] = {
+ { OPCODE_NOP, "NOP", 0 },
+ { OPCODE_ABS, "ABS", 1 },
+ { OPCODE_ADD, "ADD", 2 },
+ { OPCODE_ARA, "ARA", 1 },
+ { OPCODE_ARL, "ARL", 1 },
+ { OPCODE_ARL_NV, "ARL", 1 },
+ { OPCODE_ARR, "ARL", 1 },
+ { OPCODE_BGNLOOP,"BGNLOOP", 0 },
+ { OPCODE_BGNSUB, "BGNSUB", 0 },
+ { OPCODE_BRA, "BRA", 0 },
+ { OPCODE_BRK, "BRK", 0 },
+ { OPCODE_CAL, "CAL", 0 },
+ { OPCODE_CMP, "CMP", 3 },
+ { OPCODE_CONT, "CONT", 1 },
+ { OPCODE_COS, "COS", 1 },
+ { OPCODE_DDX, "DDX", 1 },
+ { OPCODE_DDY, "DDY", 1 },
+ { OPCODE_DP3, "DP3", 2 },
+ { OPCODE_DP4, "DP4", 2 },
+ { OPCODE_DPH, "DPH", 2 },
+ { OPCODE_DST, "DST", 2 },
+ { OPCODE_ELSE, "ELSE", 0 },
+ { OPCODE_END, "END", 0 },
+ { OPCODE_ENDIF, "ENDIF", 0 },
+ { OPCODE_ENDLOOP,"ENDLOOP", 0 },
+ { OPCODE_ENDSUB, "ENDSUB", 0 },
+ { OPCODE_EX2, "EX2", 1 },
+ { OPCODE_EXP, "EXP", 1 },
+ { OPCODE_FLR, "FLR", 1 },
+ { OPCODE_FRC, "FRC", 1 },
+ { OPCODE_IF, "IF", 0 },
+ { OPCODE_INT, "INT", 1 },
+ { OPCODE_KIL, "KIL", 1 },
+ { OPCODE_KIL_NV, "KIL", 0 },
+ { OPCODE_LG2, "LG2", 1 },
+ { OPCODE_LIT, "LIT", 1 },
+ { OPCODE_LOG, "LOG", 1 },
+ { OPCODE_LRP, "LRP", 3 },
+ { OPCODE_MAD, "MAD", 3 },
+ { OPCODE_MAX, "MAX", 2 },
+ { OPCODE_MIN, "MIN", 2 },
+ { OPCODE_MOV, "MOV", 1 },
+ { OPCODE_MUL, "MUL", 2 },
+ { OPCODE_NOISE1, "NOISE1", 1 },
+ { OPCODE_NOISE2, "NOISE2", 1 },
+ { OPCODE_NOISE3, "NOISE3", 1 },
+ { OPCODE_NOISE4, "NOISE4", 1 },
+ { OPCODE_PK2H, "PK2H", 1 },
+ { OPCODE_PK2US, "PK2US", 1 },
+ { OPCODE_PK4B, "PK4B", 1 },
+ { OPCODE_PK4UB, "PK4UB", 1 },
+ { OPCODE_POW, "POW", 2 },
+ { OPCODE_POPA, "POPA", 0 },
+ { OPCODE_PRINT, "PRINT", 1 },
+ { OPCODE_PUSHA, "PUSHA", 0 },
+ { OPCODE_RCC, "RCC", 1 },
+ { OPCODE_RCP, "RCP", 1 },
+ { OPCODE_RET, "RET", 0 },
+ { OPCODE_RFL, "RFL", 1 },
+ { OPCODE_RSQ, "RSQ", 1 },
+ { OPCODE_SCS, "SCS", 1 },
+ { OPCODE_SEQ, "SEQ", 2 },
+ { OPCODE_SFL, "SFL", 0 },
+ { OPCODE_SGE, "SGE", 2 },
+ { OPCODE_SGT, "SGT", 2 },
+ { OPCODE_SIN, "SIN", 1 },
+ { OPCODE_SLE, "SLE", 2 },
+ { OPCODE_SLT, "SLT", 2 },
+ { OPCODE_SNE, "SNE", 2 },
+ { OPCODE_SSG, "SSG", 1 },
+ { OPCODE_STR, "STR", 0 },
+ { OPCODE_SUB, "SUB", 2 },
+ { OPCODE_SWZ, "SWZ", 1 },
+ { OPCODE_TEX, "TEX", 1 },
+ { OPCODE_TXB, "TXB", 1 },
+ { OPCODE_TXD, "TXD", 3 },
+ { OPCODE_TXL, "TXL", 1 },
+ { OPCODE_TXP, "TXP", 1 },
+ { OPCODE_TXP_NV, "TXP", 1 },
+ { OPCODE_UP2H, "UP2H", 1 },
+ { OPCODE_UP2US, "UP2US", 1 },
+ { OPCODE_UP4B, "UP4B", 1 },
+ { OPCODE_UP4UB, "UP4UB", 1 },
+ { OPCODE_X2D, "X2D", 3 },
+ { OPCODE_XPD, "XPD", 2 }
+};
+
+
+/**
+ * Return the number of src registers for the given instruction/opcode.
+ */
+GLuint
+_mesa_num_inst_src_regs(gl_inst_opcode opcode)
+{
+ ASSERT(opcode == InstInfo[opcode].Opcode);
+ ASSERT(OPCODE_XPD == InstInfo[OPCODE_XPD].Opcode);
+ return InstInfo[opcode].NumSrcRegs;
+}
+
+
+/**
+ * Return string name for given program opcode.
+ */
+const char *
+_mesa_opcode_string(gl_inst_opcode opcode)
+{
+ ASSERT(opcode < MAX_OPCODE);
+ return InstInfo[opcode].Name;
+}
+
--- /dev/null
+/*
+ * Mesa 3-D graphics library
+ * Version: 6.5.3
+ *
+ * Copyright (C) 1999-2007 Brian Paul All Rights Reserved.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included
+ * in all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
+ * OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * BRIAN PAUL BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN
+ * AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+ * CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+
+/**
+ * \file prog_instruction.h
+ *
+ * Vertex/fragment program instruction datatypes and constants.
+ *
+ * \author Brian Paul
+ * \author Keith Whitwell
+ * \author Ian Romanick <idr@us.ibm.com>
+ */
+
+
+#ifndef PROG_INSTRUCTION_H
+#define PROG_INSTRUCTION_H
+
+
+/**
+ * Swizzle indexes.
+ * Do not change!
+ */
+/*@{*/
+#define SWIZZLE_X 0
+#define SWIZZLE_Y 1
+#define SWIZZLE_Z 2
+#define SWIZZLE_W 3
+#define SWIZZLE_ZERO 4 /**< For SWZ instruction only */
+#define SWIZZLE_ONE 5 /**< For SWZ instruction only */
+#define SWIZZLE_NIL 7 /**< used during shader code gen (undefined value) */
+/*@}*/
+
+#define MAKE_SWIZZLE4(a,b,c,d) (((a)<<0) | ((b)<<3) | ((c)<<6) | ((d)<<9))
+#define SWIZZLE_NOOP MAKE_SWIZZLE4(0,1,2,3)
+#define GET_SWZ(swz, idx) (((swz) >> ((idx)*3)) & 0x7)
+#define GET_BIT(msk, idx) (((msk) >> (idx)) & 0x1)
+
+#define SWIZZLE_XXXX MAKE_SWIZZLE4(SWIZZLE_X, SWIZZLE_X, SWIZZLE_X, SWIZZLE_X)
+#define SWIZZLE_YYYY MAKE_SWIZZLE4(SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y, SWIZZLE_Y)
+#define SWIZZLE_ZZZZ MAKE_SWIZZLE4(SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z, SWIZZLE_Z)
+#define SWIZZLE_WWWW MAKE_SWIZZLE4(SWIZZLE_W, SWIZZLE_W, SWIZZLE_W, SWIZZLE_W)
+
+
+/**
+ * Writemask values, 1 bit per component.
+ */
+/*@{*/
+#define WRITEMASK_X 0x1
+#define WRITEMASK_Y 0x2
+#define WRITEMASK_XY 0x3
+#define WRITEMASK_Z 0x4
+#define WRITEMASK_XZ 0x5
+#define WRITEMASK_YZ 0x6
+#define WRITEMASK_XYZ 0x7
+#define WRITEMASK_W 0x8
+#define WRITEMASK_XW 0x9
+#define WRITEMASK_YW 0xa
+#define WRITEMASK_XYW 0xb
+#define WRITEMASK_ZW 0xc
+#define WRITEMASK_XZW 0xd
+#define WRITEMASK_YZW 0xe
+#define WRITEMASK_XYZW 0xf
+/*@}*/
+
+
+/**
+ * Condition codes
+ */
+/*@{*/
+#define COND_GT 1 /**< greater than zero */
+#define COND_EQ 2 /**< equal to zero */
+#define COND_LT 3 /**< less than zero */
+#define COND_UN 4 /**< unordered (NaN) */
+#define COND_GE 5 /**< greater then or equal to zero */
+#define COND_LE 6 /**< less then or equal to zero */
+#define COND_NE 7 /**< not equal to zero */
+#define COND_TR 8 /**< always true */
+#define COND_FL 9 /**< always false */
+/*@}*/
+
+
+/**
+ * Instruction precision for GL_NV_fragment_program
+ */
+/*@{*/
+#define FLOAT32 0x1
+#define FLOAT16 0x2
+#define FIXED12 0x4
+/*@}*/
+
+
+/**
+ * Saturation modes when storing values.
+ */
+/*@{*/
+#define SATURATE_OFF 0
+#define SATURATE_ZERO_ONE 1
+#define SATURATE_PLUS_MINUS_ONE 2
+/*@}*/
+
+
+/**
+ * Per-component negation masks
+ */
+/*@{*/
+#define NEGATE_X 0x1
+#define NEGATE_Y 0x2
+#define NEGATE_Z 0x4
+#define NEGATE_W 0x8
+#define NEGATE_XYZW 0xf
+#define NEGATE_NONE 0x0
+/*@}*/
+
+
+/**
+ * Program instruction opcodes, for both vertex and fragment programs.
+ * \note changes to this opcode list must be reflected in t_vb_arbprogram.c
+ */
+typedef enum prog_opcode {
+ /* ARB_vp ARB_fp NV_vp NV_fp GLSL */
+ /*------------------------------------------*/
+ OPCODE_NOP = 0, /* X */
+ OPCODE_ABS, /* X X 1.1 X */
+ OPCODE_ADD, /* X X X X X */
+ OPCODE_ARA, /* 2 */
+ OPCODE_ARL, /* X X */
+ OPCODE_ARL_NV, /* 2 */
+ OPCODE_ARR, /* 2 */
+ OPCODE_BGNLOOP, /* opt */
+ OPCODE_BGNSUB, /* opt */
+ OPCODE_BRA, /* 2 X */
+ OPCODE_BRK, /* 2 opt */
+ OPCODE_CAL, /* 2 2 */
+ OPCODE_CMP, /* X */
+ OPCODE_CONT, /* opt */
+ OPCODE_COS, /* X 2 X X */
+ OPCODE_DDX, /* X X */
+ OPCODE_DDY, /* X X */
+ OPCODE_DP3, /* X X X X X */
+ OPCODE_DP4, /* X X X X X */
+ OPCODE_DPH, /* X X 1.1 */
+ OPCODE_DST, /* X X X X */
+ OPCODE_ELSE, /* X */
+ OPCODE_END, /* X X X X opt */
+ OPCODE_ENDIF, /* opt */
+ OPCODE_ENDLOOP, /* opt */
+ OPCODE_ENDSUB, /* opt */
+ OPCODE_EX2, /* X X 2 X X */
+ OPCODE_EXP, /* X X X */
+ OPCODE_FLR, /* X X 2 X X */
+ OPCODE_FRC, /* X X 2 X X */
+ OPCODE_IF, /* opt */
+ OPCODE_INT, /* X */
+ OPCODE_KIL, /* X */
+ OPCODE_KIL_NV, /* X X */
+ OPCODE_LG2, /* X X 2 X X */
+ OPCODE_LIT, /* X X X X */
+ OPCODE_LOG, /* X X X */
+ OPCODE_LRP, /* X X */
+ OPCODE_MAD, /* X X X X X */
+ OPCODE_MAX, /* X X X X X */
+ OPCODE_MIN, /* X X X X X */
+ OPCODE_MOV, /* X X X X X */
+ OPCODE_MUL, /* X X X X X */
+ OPCODE_NOISE1, /* X */
+ OPCODE_NOISE2, /* X */
+ OPCODE_NOISE3, /* X */
+ OPCODE_NOISE4, /* X */
+ OPCODE_PK2H, /* X */
+ OPCODE_PK2US, /* X */
+ OPCODE_PK4B, /* X */
+ OPCODE_PK4UB, /* X */
+ OPCODE_POW, /* X X X X */
+ OPCODE_POPA, /* 3 */
+ OPCODE_PRINT, /* X X */
+ OPCODE_PUSHA, /* 3 */
+ OPCODE_RCC, /* 1.1 */
+ OPCODE_RCP, /* X X X X X */
+ OPCODE_RET, /* 2 2 */
+ OPCODE_RFL, /* X X */
+ OPCODE_RSQ, /* X X X X X */
+ OPCODE_SCS, /* X */
+ OPCODE_SEQ, /* 2 X X */
+ OPCODE_SFL, /* 2 X */
+ OPCODE_SGE, /* X X X X X */
+ OPCODE_SGT, /* 2 X X */
+ OPCODE_SIN, /* X 2 X X */
+ OPCODE_SLE, /* 2 X X */
+ OPCODE_SLT, /* X X X X X */
+ OPCODE_SNE, /* 2 X X */
+ OPCODE_SSG, /* 2 */
+ OPCODE_STR, /* 2 X */
+ OPCODE_SUB, /* X X 1.1 X X */
+ OPCODE_SWZ, /* X X */
+ OPCODE_TEX, /* X 3 X X */
+ OPCODE_TXB, /* X 3 X */
+ OPCODE_TXD, /* X X */
+ OPCODE_TXL, /* 3 2 X */
+ OPCODE_TXP, /* X X */
+ OPCODE_TXP_NV, /* 3 X */
+ OPCODE_UP2H, /* X */
+ OPCODE_UP2US, /* X */
+ OPCODE_UP4B, /* X */
+ OPCODE_UP4UB, /* X */
+ OPCODE_X2D, /* X */
+ OPCODE_XPD, /* X X X */
+ MAX_OPCODE
+} gl_inst_opcode;
+
+
+/**
+ * Instruction source register.
+ */
+struct prog_src_register
+{
+ GLuint File:4; /**< One of the PROGRAM_* register file values. */
+ GLint Index:9; /**< May be negative for relative addressing. */
+ GLuint Swizzle:12;
+ GLuint RelAddr:1;
+
+ /**
+ * \name Source register "sign" control.
+ *
+ * The ARB and NV extensions allow varrying degrees of control over the
+ * sign of the source vector components. These values allow enough control
+ * for all flavors of the extensions.
+ */
+ /*@{*/
+ /**
+ * Per-component negation for the SWZ instruction. For non-SWZ
+ * instructions the only possible values are NEGATE_XYZW and NEGATE_NONE.
+ *
+ * \since
+ * ARB_vertex_program, ARB_fragment_program
+ */
+ GLuint NegateBase:4;
+
+ /**
+ * Take the component-wise absolute value.
+ *
+ * \since
+ * NV_fragment_program, NV_fragment_program_option, NV_vertex_program2,
+ * NV_vertex_program2_option.
+ */
+ GLuint Abs:1;
+
+ /**
+ * Post-absolute value negation (all components).
+ */
+ GLuint NegateAbs:1;
+ /*@}*/
+};
+
+
+/**
+ * Instruction destination register.
+ */
+struct prog_dst_register
+{
+ /**
+ * One of the PROGRAM_* register file values.
+ */
+ GLuint File:4;
+
+ GLuint Index:8;
+ GLuint WriteMask:4;
+
+ /**
+ * \name Conditional destination update control.
+ *
+ * \since
+ * NV_fragment_program, NV_fragment_program_option, NV_vertex_program2,
+ * NV_vertex_program2_option.
+ */
+ /*@{*/
+ /**
+ * Takes one of the 9 possible condition values (EQ, FL, GT, GE, LE, LT,
+ * NE, TR, or UN). Destination update is enabled if the matching
+ * (swizzled) condition code value passes. When a conditional update mask
+ * is not specified, this will be \c COND_TR.
+ */
+ GLuint CondMask:4;
+
+ /**
+ * Condition code swizzle value.
+ */
+ GLuint CondSwizzle:12;
+
+ /**
+ * Selects the condition code register to use for conditional destination
+ * update masking. In NV_fragmnet_program or NV_vertex_program2 mode, only
+ * condition code register 0 is available. In NV_vertex_program3 mode,
+ * condition code registers 0 and 1 are available.
+ */
+ GLuint CondSrc:1;
+ /*@}*/
+
+ GLuint pad:31;
+};
+
+
+/**
+ * Vertex/fragment program instruction.
+ */
+struct prog_instruction
+{
+ gl_inst_opcode Opcode;
+#if FEATURE_MESA_program_debug
+ GLshort StringPos;
+#endif
+ /**
+ * Arbitrary data. Used for the PRINT, CAL, and BRA instructions.
+ */
+ void *Data;
+
+ struct prog_src_register SrcReg[3];
+ struct prog_dst_register DstReg;
+
+ /**
+ * Indicates that the instruction should update the condition code
+ * register.
+ *
+ * \since
+ * NV_fragment_program, NV_fragment_program_option, NV_vertex_program2,
+ * NV_vertex_program2_option.
+ */
+ GLuint CondUpdate:1;
+
+ /**
+ * If prog_instruction::CondUpdate is \c GL_TRUE, this value selects the
+ * condition code register that is to be updated.
+ *
+ * In GL_NV_fragment_program or GL_NV_vertex_program2 mode, only condition
+ * code register 0 is available. In GL_NV_vertex_program3 mode, condition
+ * code registers 0 and 1 are available.
+ *
+ * \since
+ * NV_fragment_program, NV_fragment_program_option, NV_vertex_program2,
+ * NV_vertex_program2_option.
+ */
+ GLuint CondDst:1;
+
+ /**
+ * Saturate each value of the vectored result to the range [0,1] or the
+ * range [-1,1]. \c SSAT mode (i.e., saturation to the range [-1,1]) is
+ * only available in NV_fragment_program2 mode.
+ * Value is one of the SATURATE_* tokens.
+ *
+ * \since
+ * NV_fragment_program, NV_fragment_program_option, NV_vertex_program3.
+ */
+ GLuint SaturateMode:2;
+
+ /**
+ * Per-instruction selectable precision.
+ *
+ * \since
+ * NV_fragment_program, NV_fragment_program_option.
+ */
+ GLuint Precision:3;
+
+ /**
+ * \name Texture source controls.
+ *
+ * The texture source controls are only used with the \c TEX, \c TXD,
+ * \c TXL, and \c TXP instructions.
+ *
+ * \since
+ * ARB_fragment_program, NV_fragment_program, NV_vertex_program3.
+ */
+ /*@{*/
+ /**
+ * Source texture unit. OpenGL supports a maximum of 32 texture
+ * units.
+ */
+ GLuint TexSrcUnit:5;
+
+ /**
+ * Source texture target, one of TEXTURE_{1D,2D,3D,CUBE,RECT}_INDEX.
+ */
+ GLuint TexSrcTarget:3;
+ /*@}*/
+
+ /**
+ * For BRA and CAL instructions, the location to jump to.
+ * For BGNLOOP, points to ENDLOOP (and vice-versa).
+ * For BRK, points to BGNLOOP (which points to ENDLOOP).
+ * For IF, points to else or endif.
+ * For ELSE, points to endif.
+ */
+ GLint BranchTarget;
+
+ /**
+ * For TEX instructions in shaders, the sampler to use for the
+ * texture lookup.
+ */
+ GLint Sampler;
+
+ const char *Comment;
+};
+
+
+extern void
+_mesa_init_instructions(struct prog_instruction *inst, GLuint count);
+
+extern struct prog_instruction *
+_mesa_alloc_instructions(GLuint numInst);
+
+extern struct prog_instruction *
+_mesa_realloc_instructions(struct prog_instruction *oldInst,
+ GLuint numOldInst, GLuint numNewInst);
+
++extern struct prog_instruction *
++_mesa_copy_instructions(struct prog_instruction *dest,
++ const struct prog_instruction *src, GLuint n);
++
+extern GLuint
+_mesa_num_inst_src_regs(gl_inst_opcode opcode);
+
+extern const char *
+_mesa_opcode_string(gl_inst_opcode opcode);
+
+
+#endif /* PROG_INSTRUCTION_H */
/*
* Mesa 3-D graphics library
- * Version: 6.5.2
+ * Version: 6.5.3
*
- * Copyright (C) 1999-2006 Brian Paul All Rights Reserved.
+ * Copyright (C) 1999-2007 Brian Paul All Rights Reserved.
*
* Permission is hereby granted, free of charge, to any person obtaining a
* copy of this software and associated documentation files (the "Software"),
#include "glheader.h"
#include "context.h"
-#include "imports.h"
-#include "mtypes.h"
-#include "program.h"
+#include "prog_parameter.h"
+#include "prog_statevars.h"
#include "programopt.h"
-#include "program_instruction.h"
+#include "prog_instruction.h"
/**
* Setup state references for the modelview/projection matrix.
* XXX we should check if these state vars are already declared.
*/
- static const GLint mvpState[4][5] = {
- { STATE_MATRIX, STATE_MVP, 0, 0, 0 }, /* state.matrix.mvp.row[0] */
- { STATE_MATRIX, STATE_MVP, 0, 1, 1 }, /* state.matrix.mvp.row[1] */
- { STATE_MATRIX, STATE_MVP, 0, 2, 2 }, /* state.matrix.mvp.row[2] */
- { STATE_MATRIX, STATE_MVP, 0, 3, 3 }, /* state.matrix.mvp.row[3] */
+ static const gl_state_index mvpState[4][STATE_LENGTH] = {
+ { STATE_MVP_MATRIX, 0, 0, 0, 0 }, /* state.matrix.mvp.row[0] */
+ { STATE_MVP_MATRIX, 0, 1, 1, 0 }, /* state.matrix.mvp.row[1] */
+ { STATE_MVP_MATRIX, 0, 2, 2, 0 }, /* state.matrix.mvp.row[2] */
+ { STATE_MVP_MATRIX, 0, 3, 3, 0 }, /* state.matrix.mvp.row[3] */
};
GLint mvpRef[4];
}
/* Append original instructions after new instructions */
- _mesa_memcpy(newInst + 4, vprog->Base.Instructions,
- origLen * sizeof(struct prog_instruction));
+ _mesa_copy_instructions (newInst + 4, vprog->Base.Instructions, origLen);
/* free old instructions */
_mesa_free(vprog->Base.Instructions);
void
_mesa_append_fog_code(GLcontext *ctx, struct gl_fragment_program *fprog)
{
- static const GLint fogPStateOpt[] = { STATE_INTERNAL,
- STATE_FOG_PARAMS_OPTIMIZED, 0, 0, 0 };
- static const GLint fogColorState[] = { STATE_FOG_COLOR, 0, 0, 0, 0 };
+ static const gl_state_index fogPStateOpt[STATE_LENGTH]
+ = { STATE_INTERNAL, STATE_FOG_PARAMS_OPTIMIZED, 0, 0, 0 };
+ static const gl_state_index fogColorState[STATE_LENGTH]
+ = { STATE_FOG_COLOR, 0, 0, 0, 0};
struct prog_instruction *newInst, *inst;
const GLuint origLen = fprog->Base.NumInstructions;
const GLuint newLen = origLen + 5;
fprog->Base.InputsRead |= FRAG_BIT_FOGC;
/* XXX do this? fprog->FogOption = GL_NONE; */
}
+
+
+
+static GLboolean
+is_texture_instruction(const struct prog_instruction *inst)
+{
+ switch (inst->Opcode) {
+ case OPCODE_TEX:
+ case OPCODE_TXB:
+ case OPCODE_TXD:
+ case OPCODE_TXL:
+ case OPCODE_TXP:
+ case OPCODE_TXP_NV:
+ return GL_TRUE;
+ default:
+ return GL_FALSE;
+ }
+}
+
+
+/**
+ * Count the number of texure indirections in the given program.
+ * The program's NumTexIndirections field will be updated.
+ * See the GL_ARB_fragment_program spec (issue 24) for details.
+ * XXX we count texture indirections in texenvprogram.c (maybe use this code
+ * instead and elsewhere).
+ */
+void
+_mesa_count_texture_indirections(struct gl_program *prog)
+{
+ GLuint indirections = 1;
+ GLbitfield tempsOutput = 0x0;
+ GLbitfield aluTemps = 0x0;
+ GLuint i;
+
+ for (i = 0; i < prog->NumInstructions; i++) {
+ const struct prog_instruction *inst = prog->Instructions + i;
+
+ if (is_texture_instruction(inst)) {
+ if (((inst->SrcReg[0].File == PROGRAM_TEMPORARY) &&
+ (tempsOutput & (1 << inst->SrcReg[0].Index))) ||
+ ((inst->Opcode != OPCODE_KIL) &&
+ (inst->DstReg.File == PROGRAM_TEMPORARY) &&
+ (aluTemps & (1 << inst->DstReg.Index))))
+ {
+ indirections++;
+ tempsOutput = 0x0;
+ aluTemps = 0x0;
+ }
+ }
+ else {
+ GLuint j;
+ for (j = 0; j < 3; j++) {
+ if (inst->SrcReg[j].File == PROGRAM_TEMPORARY)
+ aluTemps |= (1 << inst->SrcReg[j].Index);
+ }
+ if (inst->DstReg.File == PROGRAM_TEMPORARY)
+ aluTemps |= (1 << inst->DstReg.Index);
+ }
+
+ if ((inst->Opcode != OPCODE_KIL) && (inst->DstReg.File == PROGRAM_TEMPORARY))
+ tempsOutput |= (1 << inst->DstReg.Index);
+ }
+
+ prog->NumTexIndirections = indirections;
+}
+
+
+/**
+ * Count number of texture instructions in given program and update the
+ * program's NumTexInstructions field.
+ */
+void
+_mesa_count_texture_instructions(struct gl_program *prog)
+{
+ GLuint i;
+ prog->NumTexInstructions = 0;
+ for (i = 0; i < prog->NumInstructions; i++) {
+ prog->NumTexInstructions += is_texture_instruction(prog->Instructions + i);
+ }
+}
+
GLint sy, dy, stepy, row;
const GLboolean zoom = ctx->Pixel.ZoomX != 1.0F || ctx->Pixel.ZoomY != 1.0F;
GLint overlapping;
- const GLuint transferOps = ctx->_ImageTransferState;
+ GLuint transferOps = ctx->_ImageTransferState;
SWspan span;
if (!ctx->ReadBuffer->_ColorReadBuffer) {
copy_conv_rgba_pixels(ctx, srcx, srcy, width, height, destx, desty);
return;
}
+ else if (ctx->Pixel.Convolution1DEnabled) {
+ /* make sure we don't apply 1D convolution */
+ transferOps &= ~(IMAGE_CONVOLUTION_BIT |
+ IMAGE_POST_CONVOLUTION_SCALE_BIAS);
+ }
/* Determine if copy should be done bottom-to-top or top-to-bottom */
if (srcy < desty) {
ASSERT(width < MAX_WIDTH);
for (row = 0; row < height; row++, sy += stepy, dy += stepy) {
- GLvoid *rgba = span.array->color.sz4.rgba;
+ GLvoid *rgba = span.array->attribs[FRAG_ATTRIB_COL0];
/* Get row/span of source pixels */
if (overlapping) {
&& !scaleOrBias
&& !zoom
&& ctx->Visual.rgbMode
- && width <= MAX_WIDTH) {
+ && width <= MAX_WIDTH
+ && !unpack->SwapBytes) {
/* Special case: directly write 16-bit depth values */
GLint row;
for (row = 0; row < height; row++) {
&& !scaleOrBias
&& !zoom
&& ctx->Visual.rgbMode
- && width <= MAX_WIDTH) {
+ && width <= MAX_WIDTH
+ && !unpack->SwapBytes) {
/* Special case: shift 32-bit values down to Visual.depthBits */
const GLint shift = 32 - ctx->DrawBuffer->Visual.depthBits;
GLint row;
type = GL_FLOAT;
transferOps &= IMAGE_POST_CONVOLUTION_BITS;
}
+ else if (ctx->Pixel.Convolution1DEnabled) {
+ /* we only want to apply 1D convolution to glTexImage1D */
+ transferOps &= ~(IMAGE_CONVOLUTION_BIT |
+ IMAGE_POST_CONVOLUTION_SCALE_BIAS);
+ }
if (ctx->DrawBuffer->_NumColorDrawBuffers[0] > 0 &&
ctx->DrawBuffer->_ColorDrawBuffers[0][0]->DataType != GL_FLOAT &&
= _mesa_image_row_stride(unpack, width, format, type);
GLint skipPixels = 0;
/* use span array for temp color storage */
- GLfloat *rgba = (GLfloat *) span.array->color.sz4.rgba;
+ GLfloat *rgba = (GLfloat *) span.array->attribs[FRAG_ATTRIB_COL0];
/* if the span is wider than MAX_WIDTH we have to do it in chunks */
while (skipPixels < width) {
&& !biasOrScale && !packing->SwapBytes) {
/* Special case: directly read 24-bit unsigned depth values. */
GLint j;
- ASSERT(rb->InternalFormat == GL_DEPTH_COMPONENT32);
+ ASSERT(rb->InternalFormat == GL_DEPTH_COMPONENT24);
ASSERT(rb->DataType == GL_UNSIGNED_INT);
for (j = 0; j < height; j++, y++) {
GLuint *dest = (GLuint *)
/* no convolution */
const GLint dstStride
= _mesa_image_row_stride(packing, width, format, type);
- GLfloat (*rgba)[4] = swrast->SpanArrays->color.sz4.rgba;
+ GLfloat (*rgba)[4] = swrast->SpanArrays->attribs[FRAG_ATTRIB_COL0];
GLint row;
GLubyte *dst
= (GLubyte *) _mesa_image_address2d(packing, pixels, width, height,
format, type, 0, 0);
+ /* make sure we don't apply 1D convolution */
+ transferOps &= ~(IMAGE_CONVOLUTION_BIT |
+ IMAGE_POST_CONVOLUTION_SCALE_BIAS);
+
for (row = 0; row < height; row++, y++) {
/* Get float rgba pixels */
#include "glheader.h"
#include "macros.h"
#include "enums.h"
+#include "program.h"
+#include "prog_instruction.h"
+#include "prog_parameter.h"
+#include "prog_print.h"
+#include "prog_statevars.h"
#include "t_context.h" /* NOTE: very light dependency on this */
#include "t_vp_build.h"
-#include "shader/program.h"
-#include "shader/program_instruction.h"
struct state_key {
unsigned light_global_enabled:1;
{
GLfloat values[4];
GLint idx;
+ GLuint swizzle;
values[0] = s0;
values[1] = s1;
values[2] = s2;
values[3] = s3;
- idx = _mesa_add_unnamed_constant( p->program->Base.Parameters, values, 4 );
+ idx = _mesa_add_unnamed_constant( p->program->Base.Parameters, values, 4,
+ &swizzle );
+ ASSERT(swizzle == SWIZZLE_NOOP);
return make_ureg(PROGRAM_STATE_VAR, idx);
}
return p->identity;
}
-static struct ureg register_param6( struct tnl_program *p,
+static struct ureg register_param5(struct tnl_program *p,
GLint s0,
GLint s1,
GLint s2,
GLint s3,
- GLint s4,
- GLint s5)
+ GLint s4)
{
- GLint tokens[6];
+ gl_state_index tokens[STATE_LENGTH];
GLint idx;
tokens[0] = s0;
tokens[1] = s1;
tokens[2] = s2;
tokens[3] = s3;
tokens[4] = s4;
- tokens[5] = s5;
idx = _mesa_add_state_reference( p->program->Base.Parameters, tokens );
return make_ureg(PROGRAM_STATE_VAR, idx);
}
-#define register_param1(p,s0) register_param6(p,s0,0,0,0,0,0)
-#define register_param2(p,s0,s1) register_param6(p,s0,s1,0,0,0,0)
-#define register_param3(p,s0,s1,s2) register_param6(p,s0,s1,s2,0,0,0)
-#define register_param4(p,s0,s1,s2,s3) register_param6(p,s0,s1,s2,s3,0,0)
+#define register_param1(p,s0) register_param5(p,s0,0,0,0,0)
+#define register_param2(p,s0,s1) register_param5(p,s0,s1,0,0,0)
+#define register_param3(p,s0,s1,s2) register_param5(p,s0,s1,s2,0,0)
+#define register_param4(p,s0,s1,s2,s3) register_param5(p,s0,s1,s2,s3,0)
-static void register_matrix_param6( struct tnl_program *p,
- GLint s0,
- GLint s1,
- GLint s2,
- GLint s3,
- GLint s4,
- GLint s5,
+static void register_matrix_param5( struct tnl_program *p,
+ GLint s0, /* modelview, projection, etc */
+ GLint s1, /* texture matrix number */
+ GLint s2, /* first row */
+ GLint s3, /* last row */
+ GLint s4, /* inverse, transpose, etc */
struct ureg *matrix )
{
GLint i;
/* This is a bit sad as the support is there to pull the whole
* matrix out in one go:
*/
- for (i = 0; i <= s4 - s3; i++)
- matrix[i] = register_param6( p, s0, s1, s2, i, i, s5 );
+ for (i = 0; i <= s3 - s2; i++)
+ matrix[i] = register_param5( p, s0, s1, i, i, s4 );
}
p->eye_position = reserve_temp(p);
if (PREFER_DP4) {
- register_matrix_param6( p, STATE_MATRIX, STATE_MODELVIEW, 0, 0, 3,
- STATE_MATRIX, modelview );
+ register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3,
+ 0, modelview );
emit_matrix_transform_vec4(p, p->eye_position, modelview, pos);
}
else {
- register_matrix_param6( p, STATE_MATRIX, STATE_MODELVIEW, 0, 0, 3,
+ register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 3,
STATE_MATRIX_TRANSPOSE, modelview );
emit_transpose_matrix_transform_vec4(p, p->eye_position, modelview, pos);
struct ureg normal = register_input(p, VERT_ATTRIB_NORMAL );
struct ureg mvinv[3];
- register_matrix_param6( p, STATE_MATRIX, STATE_MODELVIEW, 0, 0, 2,
+ register_matrix_param5( p, STATE_MODELVIEW_MATRIX, 0, 0, 2,
STATE_MATRIX_INVTRANS, mvinv );
p->eye_normal = reserve_temp(p);
struct ureg mvp[4];
if (PREFER_DP4) {
- register_matrix_param6( p, STATE_MATRIX, STATE_MVP, 0, 0, 3,
- STATE_MATRIX, mvp );
+ register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3,
+ 0, mvp );
emit_matrix_transform_vec4( p, hpos, mvp, pos );
}
else {
- register_matrix_param6( p, STATE_MATRIX, STATE_MVP, 0, 0, 3,
+ register_matrix_param5( p, STATE_MVP_MATRIX, 0, 0, 3,
STATE_MATRIX_TRANSPOSE, mvp );
emit_transpose_matrix_transform_vec4( p, hpos, mvp, pos );
}
/* Need to add some addtional parameters to allow lighting in object
-- * space - STATE_SPOT_DIRECTION and STATE_HALF implicitly assume eye
++ * space - STATE_SPOT_DIRECTION and STATE_HALF_VECTOR implicitly assume eye
* space lighting.
*/
static void build_lighting( struct tnl_program *p )
*/
VPpli = register_param3(p, STATE_LIGHT, i,
STATE_POSITION_NORMALIZED);
- half = register_param3(p, STATE_LIGHT, i, STATE_HALF_VECTOR);
+ if (p->state->light_local_viewer) {
+ struct ureg eye_hat = get_eye_position_normalized(p);
+ half = get_temp(p);
+ emit_op2(p, OPCODE_SUB, half, 0, VPpli, eye_hat);
+ emit_normalize_vec3(p, half, half);
+ } else {
- half = register_param3(p, STATE_LIGHT, i, STATE_HALF);
++ half = register_param3(p, STATE_LIGHT, i, STATE_HALF_VECTOR);
+ }
}
else {
struct ureg Ppli = register_param3(p, STATE_LIGHT, i,
for (i = 0; i < MAX_TEXTURE_UNITS; i++) {
- if (!(p->state->fragprog_inputs_read & (FRAG_BIT_TEX0<<i)))
+ if (!(p->state->fragprog_inputs_read & FRAG_BIT_TEX(i)))
continue;
if (p->state->unit[i].texgen_enabled ||
out_texgen :
register_input(p, VERT_ATTRIB_TEX0+i));
if (PREFER_DP4) {
- register_matrix_param6( p, STATE_MATRIX, STATE_TEXTURE, i,
- 0, 3, STATE_MATRIX, texmat );
+ register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
+ 0, texmat );
emit_matrix_transform_vec4( p, out, texmat, in );
}
else {
- register_matrix_param6( p, STATE_MATRIX, STATE_TEXTURE, i,
- 0, 3, STATE_MATRIX_TRANSPOSE, texmat );
+ register_matrix_param5( p, STATE_TEXTURE_MATRIX, i, 0, 3,
+ STATE_MATRIX_TRANSPOSE, texmat );
emit_transpose_matrix_transform_vec4( p, out, texmat, in );
}
}
struct ureg out = register_output(p, VERT_RESULT_PSIZ);
struct ureg ut = get_temp(p);
+ /* dist = |eyez| */
+ emit_op1(p, OPCODE_ABS, ut, WRITEMASK_Y, swizzle1(eye, Z));
/* p1 + dist * (p2 + dist * p3); */
- emit_op3(p, OPCODE_MAD, ut, 0, negate(swizzle1(eye, Z)),
+ emit_op3(p, OPCODE_MAD, ut, WRITEMASK_X, swizzle1(ut, Y),
swizzle1(state_attenuation, Z), swizzle1(state_attenuation, Y));
- emit_op3(p, OPCODE_MAD, ut, 0, negate(swizzle1(eye, Z)),
+ emit_op3(p, OPCODE_MAD, ut, WRITEMASK_X, swizzle1(ut, Y),
ut, swizzle1(state_attenuation, X));
/* 1 / sqrt(factor) */
- emit_op1(p, OPCODE_RSQ, ut, 0, ut );
+ emit_op1(p, OPCODE_RSQ, ut, WRITEMASK_X, ut );
#if 1
/* out = pointSize / sqrt(factor) */
#else
/* not sure, might make sense to do clamping here,
but it's not done in t_vb_points neither */
- emit_op2(p, OPCODE_MUL, ut, 0, ut, state_size);
- emit_op2(p, OPCODE_MAX, ut, 0, ut, swizzle1(state_size, Y));
+ emit_op2(p, OPCODE_MUL, ut, WRITEMASK_X, ut, state_size);
+ emit_op2(p, OPCODE_MAX, ut, WRITEMASK_X, ut, swizzle1(state_size, Y));
emit_op2(p, OPCODE_MIN, out, WRITEMASK_X, ut, swizzle1(state_size, Z));
#endif
GLuint hash;
const struct gl_vertex_program *prev = ctx->VertexProgram._Current;
- if (ctx->VertexProgram._Enabled == GL_FALSE) {
+ if (!ctx->VertexProgram._Current) {
/* Grab all the relevent state and put it in a single structure:
*/
key = make_state_key(ctx);
/* Look for an already-prepared program for this state:
*/
- ctx->_TnlProgram = (struct gl_vertex_program *)
+ ctx->VertexProgram._TnlProgram = (struct gl_vertex_program *)
search_cache( tnl->vp_cache, hash, key, sizeof(*key) );
/* OK, we'll have to build a new one:
*/
- if (!ctx->_TnlProgram) {
+ if (!ctx->VertexProgram._TnlProgram) {
if (0)
_mesa_printf("Build new TNL program\n");
- ctx->_TnlProgram = (struct gl_vertex_program *)
+ ctx->VertexProgram._TnlProgram = (struct gl_vertex_program *)
ctx->Driver.NewProgram(ctx, GL_VERTEX_PROGRAM_ARB, 0);
- create_new_program( key, ctx->_TnlProgram,
+ create_new_program( key, ctx->VertexProgram._TnlProgram,
ctx->Const.VertexProgram.MaxTemps );
if (ctx->Driver.ProgramStringNotify)
ctx->Driver.ProgramStringNotify( ctx, GL_VERTEX_PROGRAM_ARB,
- &ctx->_TnlProgram->Base );
+ &ctx->VertexProgram._TnlProgram->Base );
- cache_item(tnl->vp_cache, hash, key, ctx->_TnlProgram );
+ cache_item(tnl->vp_cache, hash, key, ctx->VertexProgram._TnlProgram );
}
else {
FREE(key);
if (0)
_mesa_printf("Found existing TNL program for key %x\n", hash);
}
- ctx->VertexProgram._Current = ctx->_TnlProgram;
- }
- else {
- ctx->VertexProgram._Current = ctx->VertexProgram.Current;
+ ctx->VertexProgram._Current = ctx->VertexProgram._TnlProgram;
}
/* Tell the driver about the change. Could define a new target for
* this?
*/
- if (ctx->VertexProgram._Current != prev &&
- ctx->Driver.BindProgram)
+ if (ctx->VertexProgram._Current != prev && ctx->Driver.BindProgram) {
ctx->Driver.BindProgram(ctx, GL_VERTEX_PROGRAM_ARB,
(struct gl_program *) ctx->VertexProgram._Current);
+ }
}
void _tnl_ProgramCacheInit( GLcontext *ctx )