struct tgsi_exec_machine *mach,
const struct tgsi_token *tokens,
struct tgsi_sampler *sampler,
- struct tgsi_image *image)
+ struct tgsi_image *image,
+ struct tgsi_buffer *buffer)
{
uint k;
struct tgsi_parse_context parse;
mach->Tokens = tokens;
mach->Sampler = sampler;
mach->Image = image;
+ mach->Buffer = buffer;
if (!tokens) {
/* unbind and free all */
return;
}
- mach->Processor = parse.FullHeader.Processor.Processor;
mach->ImmLimit = 0;
mach->NumOutputs = 0;
- if (mach->Processor == TGSI_PROCESSOR_GEOMETRY &&
+ if (mach->ShaderType == PIPE_SHADER_GEOMETRY &&
!mach->UsedGeometryShader) {
struct tgsi_exec_vector *inputs;
struct tgsi_exec_vector *outputs;
++mach->NumOutputs;
}
}
+ else if (parse.FullToken.FullDeclaration.Declaration.File == TGSI_FILE_SYSTEM_VALUE) {
+ const struct tgsi_full_declaration *decl = &parse.FullToken.FullDeclaration;
+ mach->SysSemanticToIndex[decl->Semantic.Name] = decl->Range.First;
+ }
+
memcpy(declarations + numDeclarations,
&parse.FullToken.FullDeclaration,
sizeof(declarations[0]));
break;
case TGSI_TOKEN_TYPE_PROPERTY:
- if (mach->Processor == TGSI_PROCESSOR_GEOMETRY) {
+ if (mach->ShaderType == PIPE_SHADER_GEOMETRY) {
if (parse.FullToken.FullProperty.Property.PropertyName == TGSI_PROPERTY_GS_MAX_OUTPUT_VERTICES) {
mach->MaxOutputVertices = parse.FullToken.FullProperty.u[0].Data;
}
struct tgsi_exec_machine *
-tgsi_exec_machine_create( void )
+tgsi_exec_machine_create(enum pipe_shader_type shader_type)
{
struct tgsi_exec_machine *mach;
uint i;
memset(mach, 0, sizeof(*mach));
+ mach->ShaderType = shader_type;
mach->Addrs = &mach->Temps[TGSI_EXEC_TEMP_ADDR];
mach->MaxGeometryShaderOutputs = TGSI_MAX_TOTAL_VERTICES;
mach->Predicates = &mach->Temps[TGSI_EXEC_TEMP_P0];
case TGSI_FILE_INPUT:
for (i = 0; i < TGSI_QUAD_SIZE; i++) {
/*
- if (TGSI_PROCESSOR_GEOMETRY == mach->Processor) {
+ if (PIPE_SHADER_GEOMETRY == mach->ShaderType) {
debug_printf("Fetching Input[%d] (2d=%d, 1d=%d)\n",
index2D->i[i] * TGSI_EXEC_MAX_INPUT_ATTRIBS + index->i[i],
index2D->i[i], index->i[i]);
* gl_FragCoord, for example, in a sys value register.
*/
for (i = 0; i < TGSI_QUAD_SIZE; i++) {
- chan->u[i] = mach->SystemValue[index->i[i]].u[i];
+ chan->u[i] = mach->SystemValue[index->i[i]].xyzw[swizzle].u[i];
}
break;
debug_printf("NumOutputs = %d, TEMP_O_C/I = %d, redindex = %d\n",
mach->NumOutputs, mach->Temps[TEMP_OUTPUT_I].xyzw[TEMP_OUTPUT_C].u[0],
reg->Register.Index);
- if (TGSI_PROCESSOR_GEOMETRY == mach->Processor) {
+ if (PIPE_SHADER_GEOMETRY == mach->ShaderType) {
debug_printf("STORING OUT[%d] mask(%d), = (", offset + index, execmask);
for (i = 0; i < TGSI_QUAD_SIZE; i++)
if (execmask & (1 << i))
static void
conditional_emit_primitive(struct tgsi_exec_machine *mach)
{
- if (TGSI_PROCESSOR_GEOMETRY == mach->Processor) {
+ if (PIPE_SHADER_GEOMETRY == mach->ShaderType) {
int emitted_verts =
mach->Primitives[mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0]];
if (emitted_verts) {
return;
}
- if (mach->Processor == TGSI_PROCESSOR_FRAGMENT) {
+ if (mach->ShaderType == PIPE_SHADER_FRAGMENT) {
if (decl->Declaration.File == TGSI_FILE_INPUT) {
uint first, last, mask;
}
}
- if (decl->Declaration.File == TGSI_FILE_SYSTEM_VALUE) {
- mach->SysSemanticToIndex[decl->Declaration.Semantic] = decl->Range.First;
- }
}
typedef void (* micro_unary_op)(union tgsi_exec_channel *dst,
}
static void
-exec_load(struct tgsi_exec_machine *mach,
- const struct tgsi_full_instruction *inst)
+exec_load_img(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
{
union tgsi_exec_channel r[4], sample_r;
uint unit;
}
static void
-exec_store(struct tgsi_exec_machine *mach,
- const struct tgsi_full_instruction *inst)
+exec_load_buf(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ union tgsi_exec_channel r[4];
+ uint unit;
+ int j;
+ uint chan;
+ float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
+ struct tgsi_buffer_params params;
+ int kilmask = mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0];
+
+ unit = fetch_sampler_unit(mach, inst, 0);
+
+ params.execmask = mach->ExecMask & mach->NonHelperMask & ~kilmask;
+ params.unit = unit;
+ IFETCH(&r[0], 1, TGSI_CHAN_X);
+
+ mach->Buffer->load(mach->Buffer, ¶ms,
+ r[0].i, rgba);
+ for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+ r[0].f[j] = rgba[0][j];
+ r[1].f[j] = rgba[1][j];
+ r[2].f[j] = rgba[2][j];
+ r[3].f[j] = rgba[3][j];
+ }
+ for (chan = 0; chan < TGSI_NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &r[chan], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_load(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ if (inst->Src[0].Register.File == TGSI_FILE_IMAGE)
+ exec_load_img(mach, inst);
+ else
+ exec_load_buf(mach, inst);
+}
+
+static void
+exec_store_img(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
{
union tgsi_exec_channel r[3], sample_r;
union tgsi_exec_channel value[4];
}
static void
-exec_atomop(struct tgsi_exec_machine *mach,
- const struct tgsi_full_instruction *inst)
+exec_store_buf(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ union tgsi_exec_channel r[3];
+ union tgsi_exec_channel value[4];
+ float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
+ struct tgsi_buffer_params params;
+ int i, j;
+ uint unit;
+ int kilmask = mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0];
+
+ unit = inst->Dst[0].Register.Index;
+
+ params.execmask = mach->ExecMask & mach->NonHelperMask & ~kilmask;
+ params.unit = unit;
+ params.writemask = inst->Dst[0].Register.WriteMask;
+
+ IFETCH(&r[0], 0, TGSI_CHAN_X);
+ for (i = 0; i < 4; i++) {
+ FETCH(&value[i], 1, TGSI_CHAN_X + i);
+ }
+
+ for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+ rgba[0][j] = value[0].f[j];
+ rgba[1][j] = value[1].f[j];
+ rgba[2][j] = value[2].f[j];
+ rgba[3][j] = value[3].f[j];
+ }
+
+ mach->Buffer->store(mach->Buffer, ¶ms,
+ r[0].i,
+ rgba);
+}
+
+static void
+exec_store(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ if (inst->Dst[0].Register.File == TGSI_FILE_IMAGE)
+ exec_store_img(mach, inst);
+ else
+ exec_store_buf(mach, inst);
+}
+
+static void
+exec_atomop_img(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
{
union tgsi_exec_channel r[4], sample_r;
union tgsi_exec_channel value[4], value2[4];
}
static void
-exec_resq(struct tgsi_exec_machine *mach,
- const struct tgsi_full_instruction *inst)
+exec_atomop_buf(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ union tgsi_exec_channel r[4];
+ union tgsi_exec_channel value[4], value2[4];
+ float rgba[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
+ float rgba2[TGSI_NUM_CHANNELS][TGSI_QUAD_SIZE];
+ struct tgsi_buffer_params params;
+ int i, j;
+ uint unit, chan;
+ int kilmask = mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0];
+
+ unit = fetch_sampler_unit(mach, inst, 0);
+
+ params.execmask = mach->ExecMask & mach->NonHelperMask & ~kilmask;
+ params.unit = unit;
+ params.writemask = inst->Dst[0].Register.WriteMask;
+
+ IFETCH(&r[0], 1, TGSI_CHAN_X);
+
+ for (i = 0; i < 4; i++) {
+ FETCH(&value[i], 2, TGSI_CHAN_X + i);
+ if (inst->Instruction.Opcode == TGSI_OPCODE_ATOMCAS)
+ FETCH(&value2[i], 3, TGSI_CHAN_X + i);
+ }
+
+ for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+ rgba[0][j] = value[0].f[j];
+ rgba[1][j] = value[1].f[j];
+ rgba[2][j] = value[2].f[j];
+ rgba[3][j] = value[3].f[j];
+ }
+ if (inst->Instruction.Opcode == TGSI_OPCODE_ATOMCAS) {
+ for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+ rgba2[0][j] = value2[0].f[j];
+ rgba2[1][j] = value2[1].f[j];
+ rgba2[2][j] = value2[2].f[j];
+ rgba2[3][j] = value2[3].f[j];
+ }
+ }
+
+ mach->Buffer->op(mach->Buffer, ¶ms, inst->Instruction.Opcode,
+ r[0].i,
+ rgba, rgba2);
+
+ for (j = 0; j < TGSI_QUAD_SIZE; j++) {
+ r[0].f[j] = rgba[0][j];
+ r[1].f[j] = rgba[1][j];
+ r[2].f[j] = rgba[2][j];
+ r[3].f[j] = rgba[3][j];
+ }
+ for (chan = 0; chan < TGSI_NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &r[chan], &inst->Dst[0], inst, chan, TGSI_EXEC_DATA_FLOAT);
+ }
+ }
+}
+
+static void
+exec_atomop(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ if (inst->Src[0].Register.File == TGSI_FILE_IMAGE)
+ exec_atomop_img(mach, inst);
+ else
+ exec_atomop_buf(mach, inst);
+}
+
+static void
+exec_resq_img(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
{
int result[4];
union tgsi_exec_channel r[4];
}
}
+static void
+exec_resq_buf(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ int result;
+ union tgsi_exec_channel r[4];
+ uint unit;
+ int i, chan;
+ struct tgsi_buffer_params params;
+ int kilmask = mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0];
+
+ unit = fetch_sampler_unit(mach, inst, 0);
+
+ params.execmask = mach->ExecMask & mach->NonHelperMask & ~kilmask;
+ params.unit = unit;
+
+ mach->Buffer->get_dims(mach->Buffer, ¶ms, &result);
+
+ for (i = 0; i < TGSI_QUAD_SIZE; i++) {
+ r[0].i[i] = result;
+ }
+
+ for (chan = 0; chan < TGSI_NUM_CHANNELS; chan++) {
+ if (inst->Dst[0].Register.WriteMask & (1 << chan)) {
+ store_dest(mach, &r[chan], &inst->Dst[0], inst, chan,
+ TGSI_EXEC_DATA_INT);
+ }
+ }
+}
+
+static void
+exec_resq(struct tgsi_exec_machine *mach,
+ const struct tgsi_full_instruction *inst)
+{
+ if (inst->Src[0].Register.File == TGSI_FILE_IMAGE)
+ exec_resq_img(mach, inst);
+ else
+ exec_resq_buf(mach, inst);
+}
+
static void
micro_i2f(union tgsi_exec_channel *dst,
const union tgsi_exec_channel *src)
}
}
-
-/**
- * Run TGSI interpreter.
- * \return bitmask of "alive" quad components
- */
-uint
-tgsi_exec_machine_run( struct tgsi_exec_machine *mach )
+static void
+tgsi_exec_machine_setup_masks(struct tgsi_exec_machine *mach)
{
- uint i;
- int pc = 0;
uint default_mask = 0xf;
mach->Temps[TEMP_KILMASK_I].xyzw[TEMP_KILMASK_C].u[0] = 0;
mach->Temps[TEMP_OUTPUT_I].xyzw[TEMP_OUTPUT_C].u[0] = 0;
- if( mach->Processor == TGSI_PROCESSOR_GEOMETRY ) {
+ if (mach->ShaderType == PIPE_SHADER_GEOMETRY) {
mach->Temps[TEMP_PRIMITIVE_I].xyzw[TEMP_PRIMITIVE_C].u[0] = 0;
mach->Primitives[0] = 0;
/* GS runs on a single primitive for now */
assert(mach->SwitchStackTop == 0);
assert(mach->BreakStackTop == 0);
assert(mach->CallStackTop == 0);
+}
+
+/**
+ * Run TGSI interpreter.
+ * \return bitmask of "alive" quad components
+ */
+uint
+tgsi_exec_machine_run( struct tgsi_exec_machine *mach, int start_pc )
+{
+ uint i;
+ int pc = 0;
+ tgsi_exec_machine_setup_masks(mach);
/* execute declarations (interpolants) */
for (i = 0; i < mach->NumDeclarations; i++) {
#if 0
/* we scale from floats in [0,1] to Zbuffer ints in sp_quad_depth_test.c */
- if (mach->Processor == TGSI_PROCESSOR_FRAGMENT) {
+ if (mach->ShaderType == PIPE_SHADER_FRAGMENT) {
/*
* Scale back depth component.
*/
projects / mesa.git / blobdiff