X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fmesa%2Fdrivers%2Fdri%2Fi965%2Fbrw_vec4_vs_visitor.cpp;h=0cec77990d69bad3ad0df7028abbcfc1e07ae3c7;hb=d0d4a5f43b4dd79bd7bfff7c7deaade10bfebf7c;hp=86701f3fbd82a5883c8bc7814ebf2e0f060ddaeb;hpb=b14f4c1fd33e13c48a8fd81a9bc77c6df60108dd;p=mesa.git diff --git a/src/mesa/drivers/dri/i965/brw_vec4_vs_visitor.cpp b/src/mesa/drivers/dri/i965/brw_vec4_vs_visitor.cpp index 86701f3fbd8..0cec77990d6 100644 --- a/src/mesa/drivers/dri/i965/brw_vec4_vs_visitor.cpp +++ b/src/mesa/drivers/dri/i965/brw_vec4_vs_visitor.cpp @@ -22,129 +22,19 @@ */ -#include "brw_vs.h" - +#include "brw_vec4_vs.h" +#include "common/gen_debug.h" namespace brw { void vec4_vs_visitor::emit_prolog() { - dst_reg sign_recovery_shift; - dst_reg normalize_factor; - dst_reg es3_normalize_factor; - - for (int i = 0; i < VERT_ATTRIB_MAX; i++) { - if (vs_prog_data->inputs_read & BITFIELD64_BIT(i)) { - uint8_t wa_flags = key->gl_attrib_wa_flags[i]; - dst_reg reg(ATTR, i); - dst_reg reg_d = reg; - reg_d.type = BRW_REGISTER_TYPE_D; - dst_reg reg_ud = reg; - reg_ud.type = BRW_REGISTER_TYPE_UD; - - /* Do GL_FIXED rescaling for GLES2.0. Our GL_FIXED attributes - * come in as floating point conversions of the integer values. - */ - if (wa_flags & BRW_ATTRIB_WA_COMPONENT_MASK) { - dst_reg dst = reg; - dst.type = brw_type_for_base_type(glsl_type::vec4_type); - dst.writemask = (1 << (wa_flags & BRW_ATTRIB_WA_COMPONENT_MASK)) - 1; - emit(MUL(dst, src_reg(dst), brw_imm_f(1.0f / 65536.0f))); - } - - /* Do sign recovery for 2101010 formats if required. */ - if (wa_flags & BRW_ATTRIB_WA_SIGN) { - if (sign_recovery_shift.file == BAD_FILE) { - /* shift constant: <22,22,22,30> */ - sign_recovery_shift = dst_reg(this, glsl_type::uvec4_type); - emit(MOV(writemask(sign_recovery_shift, WRITEMASK_XYZ), brw_imm_ud(22u))); - emit(MOV(writemask(sign_recovery_shift, WRITEMASK_W), brw_imm_ud(30u))); - } - - emit(SHL(reg_ud, src_reg(reg_ud), src_reg(sign_recovery_shift))); - emit(ASR(reg_d, src_reg(reg_d), src_reg(sign_recovery_shift))); - } - - /* Apply BGRA swizzle if required. */ - if (wa_flags & BRW_ATTRIB_WA_BGRA) { - src_reg temp = src_reg(reg); - temp.swizzle = BRW_SWIZZLE4(2,1,0,3); - emit(MOV(reg, temp)); - } - - if (wa_flags & BRW_ATTRIB_WA_NORMALIZE) { - /* ES 3.0 has different rules for converting signed normalized - * fixed-point numbers than desktop GL. - */ - if ((wa_flags & BRW_ATTRIB_WA_SIGN) && !use_legacy_snorm_formula) { - /* According to equation 2.2 of the ES 3.0 specification, - * signed normalization conversion is done by: - * - * f = c / (2^(b-1)-1) - */ - if (es3_normalize_factor.file == BAD_FILE) { - /* mul constant: 1 / (2^(b-1) - 1) */ - es3_normalize_factor = dst_reg(this, glsl_type::vec4_type); - emit(MOV(writemask(es3_normalize_factor, WRITEMASK_XYZ), - brw_imm_f(1.0f / ((1<<9) - 1)))); - emit(MOV(writemask(es3_normalize_factor, WRITEMASK_W), - brw_imm_f(1.0f / ((1<<1) - 1)))); - } - - dst_reg dst = reg; - dst.type = brw_type_for_base_type(glsl_type::vec4_type); - emit(MOV(dst, src_reg(reg_d))); - emit(MUL(dst, src_reg(dst), src_reg(es3_normalize_factor))); - emit_minmax(BRW_CONDITIONAL_GE, dst, src_reg(dst), brw_imm_f(-1.0f)); - } else { - /* The following equations are from the OpenGL 3.2 specification: - * - * 2.1 unsigned normalization - * f = c/(2^n-1) - * - * 2.2 signed normalization - * f = (2c+1)/(2^n-1) - * - * Both of these share a common divisor, which is represented by - * "normalize_factor" in the code below. - */ - if (normalize_factor.file == BAD_FILE) { - /* 1 / (2^b - 1) for b=<10,10,10,2> */ - normalize_factor = dst_reg(this, glsl_type::vec4_type); - emit(MOV(writemask(normalize_factor, WRITEMASK_XYZ), - brw_imm_f(1.0f / ((1<<10) - 1)))); - emit(MOV(writemask(normalize_factor, WRITEMASK_W), - brw_imm_f(1.0f / ((1<<2) - 1)))); - } - - dst_reg dst = reg; - dst.type = brw_type_for_base_type(glsl_type::vec4_type); - emit(MOV(dst, src_reg((wa_flags & BRW_ATTRIB_WA_SIGN) ? reg_d : reg_ud))); - - /* For signed normalization, we want the numerator to be 2c+1. */ - if (wa_flags & BRW_ATTRIB_WA_SIGN) { - emit(MUL(dst, src_reg(dst), brw_imm_f(2.0f))); - emit(ADD(dst, src_reg(dst), brw_imm_f(1.0f))); - } - - emit(MUL(dst, src_reg(dst), src_reg(normalize_factor))); - } - } - - if (wa_flags & BRW_ATTRIB_WA_SCALE) { - dst_reg dst = reg; - dst.type = brw_type_for_base_type(glsl_type::vec4_type); - emit(MOV(dst, src_reg((wa_flags & BRW_ATTRIB_WA_SIGN) ? reg_d : reg_ud))); - } - } - } } dst_reg * -vec4_vs_visitor::make_reg_for_system_value(int location, - const glsl_type *type) +vec4_vs_visitor::make_reg_for_system_value(int location) { /* 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 @@ -214,7 +104,7 @@ void vec4_vs_visitor::emit_urb_slot(dst_reg reg, int varying) { reg.type = BRW_REGISTER_TYPE_F; - output_reg[varying].type = reg.type; + output_reg[varying][0].type = reg.type; switch (varying) { case VARYING_SLOT_COL0: @@ -225,7 +115,7 @@ vec4_vs_visitor::emit_urb_slot(dst_reg reg, int varying) * and we only support GS in core profile. So, this must be a vertex * shader. */ - vec4_instruction *inst = emit_generic_urb_slot(reg, varying); + vec4_instruction *inst = emit_generic_urb_slot(reg, varying, 0); if (inst && key->clamp_vertex_color) inst->saturate = true; break; @@ -260,7 +150,7 @@ vec4_vs_visitor::emit_clip_distances(dst_reg reg, int offset) ++i) { reg.writemask = 1 << i; emit(DP4(reg, - src_reg(output_reg[clip_vertex]), + src_reg(output_reg[clip_vertex][0]), src_reg(this->userplane[i + offset]))); } } @@ -290,11 +180,15 @@ vec4_vs_visitor::emit_thread_end() if (key->nr_userclip_plane_consts > 0) { current_annotation = "user clip distances"; - output_reg[VARYING_SLOT_CLIP_DIST0] = dst_reg(this, glsl_type::vec4_type); - output_reg[VARYING_SLOT_CLIP_DIST1] = dst_reg(this, glsl_type::vec4_type); + output_reg[VARYING_SLOT_CLIP_DIST0][0] = + dst_reg(this, glsl_type::vec4_type); + output_reg[VARYING_SLOT_CLIP_DIST1][0] = + dst_reg(this, glsl_type::vec4_type); + output_num_components[VARYING_SLOT_CLIP_DIST0][0] = 4; + output_num_components[VARYING_SLOT_CLIP_DIST1][0] = 4; - emit_clip_distances(output_reg[VARYING_SLOT_CLIP_DIST0], 0); - emit_clip_distances(output_reg[VARYING_SLOT_CLIP_DIST1], 4); + emit_clip_distances(output_reg[VARYING_SLOT_CLIP_DIST0][0], 0); + emit_clip_distances(output_reg[VARYING_SLOT_CLIP_DIST1][0], 4); } /* For VS, we always end the thread by emitting a single vertex.