extern "C" {
#include "main/macros.h"
#include "program/prog_parameter.h"
+#include "program/sampler.h"
}
namespace brw {
vec4_instruction *inst;
+ resolve_ud_negate(&src0);
+ resolve_ud_negate(&src1);
+
inst = new(mem_ctx) vec4_instruction(this, BRW_OPCODE_IF, dst_null_d(),
src0, src1);
inst->conditional_mod = condition;
dst.fixed_hw_reg.type = dst.type;
}
+ resolve_ud_negate(&src0);
+ resolve_ud_negate(&src1);
+
inst = new(mem_ctx) vec4_instruction(this, BRW_OPCODE_CMP, dst, src0, src1);
inst->conditional_mod = condition;
float *values = &this->vp->Base.Parameters->ParameterValues[loc][0].f;
if (type->is_matrix()) {
- const glsl_type *column = glsl_type::get_instance(GLSL_TYPE_FLOAT,
- type->vector_elements,
- 1);
+ const glsl_type *column = type->column_type();
for (unsigned int i = 0; i < type->matrix_columns; i++) {
offset += setup_uniform_values(loc + offset, column);
void
vec4_visitor::setup_uniform_clipplane_values()
{
+ gl_clip_plane *clip_planes = brw_select_clip_planes(ctx);
+
+ /* Pre-Gen6, we compact clip planes. For example, if the user
+ * enables just clip planes 0, 1, and 3, we will enable clip planes
+ * 0, 1, and 2 in the hardware, and we'll move clip plane 3 to clip
+ * plane 2. This simplifies the implementation of the Gen6 clip
+ * thread.
+ *
+ * In Gen6 and later, we don't compact clip planes, because this
+ * simplifies the implementation of gl_ClipDistance.
+ */
int compacted_clipplane_index = 0;
- for (int i = 0; i < MAX_CLIP_PLANES; ++i) {
- if (ctx->Transform.ClipPlanesEnabled & (1 << i)) {
- this->uniform_vector_size[this->uniforms] = 4;
- this->userplane[compacted_clipplane_index] = dst_reg(UNIFORM, this->uniforms);
- this->userplane[compacted_clipplane_index].type = BRW_REGISTER_TYPE_F;
- for (int j = 0; j < 4; ++j) {
- c->prog_data.param[this->uniforms * 4 + j] = &ctx->Transform._ClipUserPlane[i][j];
- }
- ++compacted_clipplane_index;
- ++this->uniforms;
+ for (int i = 0; i < c->key.nr_userclip_plane_consts; ++i) {
+ if (intel->gen < 6 &&
+ !(c->key.userclip_planes_enabled_gen_4_5 & (1 << i))) {
+ continue;
+ }
+ this->uniform_vector_size[this->uniforms] = 4;
+ this->userplane[compacted_clipplane_index] = dst_reg(UNIFORM, this->uniforms);
+ this->userplane[compacted_clipplane_index].type = BRW_REGISTER_TYPE_F;
+ for (int j = 0; j < 4; ++j) {
+ c->prog_data.param[this->uniforms * 4 + j] = &clip_planes[i][j];
}
+ ++compacted_clipplane_index;
+ ++this->uniforms;
}
}
for (unsigned int i = 0; i < expr->get_num_operands(); i++) {
expr->operands[i]->accept(this);
op[i] = this->result;
+
+ resolve_ud_negate(&op[i]);
}
switch (expr->operation) {
ir->accept(this);
+ resolve_ud_negate(&this->result);
+
if (intel->gen >= 6) {
vec4_instruction *inst = emit(AND(dst_null_d(),
this->result, src_reg(1)));
continue;
dst_reg dst = *reg;
+ dst.type = brw_type_for_base_type(ir->type);
dst.writemask = (1 << c->key.gl_fixed_input_size[i]) - 1;
emit(MUL(dst, src_reg(dst), src_reg(1.0f / 65536.0f)));
}
for (int i = 0; i < type_size(ir->type); i++) {
output_reg[ir->location + i] = *reg;
output_reg[ir->location + i].reg_offset = i;
- output_reg[ir->location + i].type = BRW_REGISTER_TYPE_F;
+ output_reg[ir->location + i].type =
+ brw_type_for_base_type(ir->type->get_scalar_type());
output_reg_annotation[ir->location + i] = ir->name;
}
break;
}
break;
+ case ir_var_system_value:
+ /* VertexID is stored by the VF as the last vertex element, but
+ * we don't represent it with a flag in inputs_read, so we call
+ * it VERT_ATTRIB_MAX, which setup_attributes() picks up on.
+ */
+ reg = new(mem_ctx) dst_reg(ATTR, VERT_ATTRIB_MAX);
+ prog_data->uses_vertexid = true;
+
+ switch (ir->location) {
+ case SYSTEM_VALUE_VERTEX_ID:
+ reg->writemask = WRITEMASK_X;
+ break;
+ case SYSTEM_VALUE_INSTANCE_ID:
+ reg->writemask = WRITEMASK_Y;
+ break;
+ default:
+ assert(!"not reached");
+ break;
+ }
+ break;
+
default:
assert(!"not reached");
}
}
}
-GLboolean
+bool
vec4_visitor::try_emit_sat(ir_expression *ir)
{
ir_rvalue *sat_src = ir->as_rvalue_to_saturate();
* potentially reswizzle the operands of many instructions so that
* we could handle out of order channels, but don't yet.
*/
- for (int i = 0; i < 4; i++) {
+
+ for (unsigned i = 0; i < 4; i++) {
if (dst.writemask & (1 << i)) {
if (!(last_rhs_inst->dst.writemask & (1 << i)))
return false;
void
vec4_visitor::visit(ir_texture *ir)
{
- /* FINISHME: Implement vertex texturing.
- *
- * With 0 vertex samplers available, the linker will reject
- * programs that do vertex texturing, but after our visitor has
- * run.
- */
- this->result = src_reg(this, glsl_type::vec4_type);
+ int sampler = _mesa_get_sampler_uniform_value(ir->sampler, prog, &vp->Base);
+ sampler = vp->Base.SamplerUnits[sampler];
+
+ /* Should be lowered by do_lower_texture_projection */
+ assert(!ir->projector);
+
+ vec4_instruction *inst = NULL;
+ switch (ir->op) {
+ case ir_tex:
+ case ir_txl:
+ inst = new(mem_ctx) vec4_instruction(this, SHADER_OPCODE_TXL);
+ break;
+ case ir_txd:
+ inst = new(mem_ctx) vec4_instruction(this, SHADER_OPCODE_TXD);
+ break;
+ case ir_txf:
+ inst = new(mem_ctx) vec4_instruction(this, SHADER_OPCODE_TXF);
+ break;
+ case ir_txs:
+ inst = new(mem_ctx) vec4_instruction(this, SHADER_OPCODE_TXS);
+ break;
+ case ir_txb:
+ assert(!"TXB is not valid for vertex shaders.");
+ }
+
+ /* Texel offsets go in the message header; Gen4 also requires headers. */
+ inst->header_present = ir->offset || intel->gen < 5;
+ inst->base_mrf = 2;
+ inst->mlen = inst->header_present + 1; /* always at least one */
+ inst->sampler = sampler;
+ inst->dst = dst_reg(this, ir->type);
+ inst->shadow_compare = ir->shadow_comparitor != NULL;
+
+ if (ir->offset != NULL)
+ inst->texture_offset = brw_texture_offset(ir->offset->as_constant());
+
+ /* MRF for the first parameter */
+ int param_base = inst->base_mrf + inst->header_present;
+
+ if (ir->op == ir_txs) {
+ ir->lod_info.lod->accept(this);
+ int writemask = intel->gen == 4 ? WRITEMASK_W : WRITEMASK_X;
+ emit(MOV(dst_reg(MRF, param_base, ir->lod_info.lod->type, writemask),
+ this->result));
+ } else {
+ int i, coord_mask = 0, zero_mask = 0;
+ /* Load the coordinate */
+ /* FINISHME: gl_clamp_mask and saturate */
+ for (i = 0; i < ir->coordinate->type->vector_elements; i++)
+ coord_mask |= (1 << i);
+ for (; i < 4; i++)
+ zero_mask |= (1 << i);
+
+ ir->coordinate->accept(this);
+ emit(MOV(dst_reg(MRF, param_base, ir->coordinate->type, coord_mask),
+ this->result));
+ emit(MOV(dst_reg(MRF, param_base, ir->coordinate->type, zero_mask),
+ src_reg(0)));
+ /* Load the shadow comparitor */
+ if (ir->shadow_comparitor) {
+ ir->shadow_comparitor->accept(this);
+ emit(MOV(dst_reg(MRF, param_base + 1, ir->shadow_comparitor->type,
+ WRITEMASK_X),
+ this->result));
+ inst->mlen++;
+ }
+
+ /* Load the LOD info */
+ if (ir->op == ir_txl) {
+ int mrf, writemask;
+ if (intel->gen >= 5) {
+ mrf = param_base + 1;
+ if (ir->shadow_comparitor) {
+ writemask = WRITEMASK_Y;
+ /* mlen already incremented */
+ } else {
+ writemask = WRITEMASK_X;
+ inst->mlen++;
+ }
+ } else /* intel->gen == 4 */ {
+ mrf = param_base;
+ writemask = WRITEMASK_Z;
+ }
+ ir->lod_info.lod->accept(this);
+ emit(MOV(dst_reg(MRF, mrf, ir->lod_info.lod->type, writemask),
+ this->result));
+ } else if (ir->op == ir_txf) {
+ ir->lod_info.lod->accept(this);
+ emit(MOV(dst_reg(MRF, param_base, ir->lod_info.lod->type, WRITEMASK_W),
+ this->result));
+ } else if (ir->op == ir_txd) {
+ const glsl_type *type = ir->lod_info.grad.dPdx->type;
+
+ ir->lod_info.grad.dPdx->accept(this);
+ src_reg dPdx = this->result;
+ ir->lod_info.grad.dPdy->accept(this);
+ src_reg dPdy = this->result;
+
+ if (intel->gen >= 5) {
+ dPdx.swizzle = BRW_SWIZZLE4(SWIZZLE_X,SWIZZLE_X,SWIZZLE_Y,SWIZZLE_Y);
+ dPdy.swizzle = BRW_SWIZZLE4(SWIZZLE_X,SWIZZLE_X,SWIZZLE_Y,SWIZZLE_Y);
+ emit(MOV(dst_reg(MRF, param_base + 1, type, WRITEMASK_XZ), dPdx));
+ emit(MOV(dst_reg(MRF, param_base + 1, type, WRITEMASK_YW), dPdy));
+ inst->mlen++;
+
+ if (ir->type->vector_elements == 3) {
+ dPdx.swizzle = BRW_SWIZZLE_ZZZZ;
+ dPdy.swizzle = BRW_SWIZZLE_ZZZZ;
+ emit(MOV(dst_reg(MRF, param_base + 2, type, WRITEMASK_X), dPdx));
+ emit(MOV(dst_reg(MRF, param_base + 2, type, WRITEMASK_Y), dPdy));
+ inst->mlen++;
+ }
+ } else /* intel->gen == 4 */ {
+ emit(MOV(dst_reg(MRF, param_base + 1, type, WRITEMASK_XYZ), dPdx));
+ emit(MOV(dst_reg(MRF, param_base + 2, type, WRITEMASK_XYZ), dPdy));
+ inst->mlen += 2;
+ }
+ }
+ }
+
+ emit(inst);
+
+ swizzle_result(ir, src_reg(inst->dst), sampler);
+}
+
+void
+vec4_visitor::swizzle_result(ir_texture *ir, src_reg orig_val, int sampler)
+{
+ this->result = orig_val;
+
+ int s = c->key.tex.swizzles[sampler];
+
+ if (ir->op == ir_txs || ir->type == glsl_type::float_type
+ || s == SWIZZLE_NOOP)
+ return;
+
+ int zero_mask = 0, one_mask = 0, copy_mask = 0;
+ int swizzle[4];
+
+ for (int i = 0; i < 4; i++) {
+ switch (GET_SWZ(s, i)) {
+ case SWIZZLE_ZERO:
+ zero_mask |= (1 << i);
+ break;
+ case SWIZZLE_ONE:
+ one_mask |= (1 << i);
+ break;
+ default:
+ copy_mask |= (1 << i);
+ swizzle[i] = GET_SWZ(s, i);
+ break;
+ }
+ }
+
+ this->result = src_reg(this, ir->type);
+ dst_reg swizzled_result(this->result);
+
+ if (copy_mask) {
+ orig_val.swizzle = BRW_SWIZZLE4(swizzle[0], swizzle[1], swizzle[2], swizzle[3]);
+ swizzled_result.writemask = copy_mask;
+ emit(MOV(swizzled_result, orig_val));
+ }
+
+ if (zero_mask) {
+ swizzled_result.writemask = zero_mask;
+ emit(MOV(swizzled_result, src_reg(0.0f)));
+ }
+
+ if (one_mask) {
+ swizzled_result.writemask = one_mask;
+ emit(MOV(swizzled_result, src_reg(1.0f)));
+ }
}
void
{
if (intel->gen < 6 &&
((c->prog_data.outputs_written & BITFIELD64_BIT(VERT_RESULT_PSIZ)) ||
- c->key.nr_userclip || brw->has_negative_rhw_bug)) {
+ c->key.userclip_active || brw->has_negative_rhw_bug)) {
dst_reg header1 = dst_reg(this, glsl_type::uvec4_type);
dst_reg header1_w = header1;
header1_w.writemask = WRITEMASK_W;
}
current_annotation = "Clipping flags";
- for (i = 0; i < c->key.nr_userclip; i++) {
+ for (i = 0; i < c->key.nr_userclip_plane_consts; i++) {
vec4_instruction *inst;
inst = emit(DP4(dst_null_f(), src_reg(output_reg[VERT_RESULT_HPOS]),
return;
}
- for (int i = 0; i + offset < c->key.nr_userclip && i < 4; ++i) {
+ /* From the GLSL 1.30 spec, section 7.1 (Vertex Shader Special Variables):
+ *
+ * "If a linked set of shaders forming the vertex stage contains no
+ * static write to gl_ClipVertex or gl_ClipDistance, but the
+ * application has requested clipping against user clip planes through
+ * the API, then the coordinate written to gl_Position is used for
+ * comparison against the user clip planes."
+ *
+ * This function is only called if the shader didn't write to
+ * gl_ClipDistance. Accordingly, we use gl_ClipVertex to perform clipping
+ * if the user wrote to it; otherwise we use gl_Position.
+ */
+ gl_vert_result clip_vertex = VERT_RESULT_CLIP_VERTEX;
+ if (!(c->prog_data.outputs_written
+ & BITFIELD64_BIT(VERT_RESULT_CLIP_VERTEX))) {
+ clip_vertex = VERT_RESULT_HPOS;
+ }
+
+ for (int i = 0; i + offset < c->key.nr_userclip_plane_consts && i < 4;
+ ++i) {
emit(DP4(dst_reg(brw_writemask(reg, 1 << i)),
- src_reg(output_reg[VERT_RESULT_HPOS]),
+ src_reg(output_reg[clip_vertex]),
src_reg(this->userplane[i + offset])));
}
}
vec4_visitor::emit_generic_urb_slot(dst_reg reg, int vert_result)
{
assert (vert_result < VERT_RESULT_MAX);
+ reg.type = output_reg[vert_result].type;
current_annotation = output_reg_annotation[vert_result];
/* Copy the register, saturating if necessary */
vec4_instruction *inst = emit(MOV(reg,
/* FINISHME: edgeflag */
- brw_compute_vue_map(&c->vue_map, intel, c->key.nr_userclip,
+ brw_compute_vue_map(&c->vue_map, intel, c->key.userclip_active,
c->prog_data.outputs_written);
/* First mrf is the g0-based message header containing URB handles and such,
split_uniform_registers();
}
+void
+vec4_visitor::resolve_ud_negate(src_reg *reg)
+{
+ if (reg->type != BRW_REGISTER_TYPE_UD ||
+ !reg->negate)
+ return;
+
+ src_reg temp = src_reg(this, glsl_type::uvec4_type);
+ emit(BRW_OPCODE_MOV, dst_reg(temp), *reg);
+ *reg = temp;
+}
+
vec4_visitor::vec4_visitor(struct brw_vs_compile *c,
struct gl_shader_program *prog,
struct brw_shader *shader)
this->current_annotation = NULL;
this->c = c;
- this->vp = prog->VertexProgram;
+ this->vp = (struct gl_vertex_program *)
+ prog->_LinkedShaders[MESA_SHADER_VERTEX]->Program;
this->prog_data = &c->prog_data;
this->variable_ht = hash_table_ctor(0,