#include "brw_shader.h"
#include "brw_ir_fs.h"
+#include "brw_fs_builder.h"
extern "C" {
*
* Translates either GLSL IR or Mesa IR (for ARB_fragment_program) into FS IR.
*/
-class fs_visitor : public backend_visitor
+class fs_visitor : public backend_shader
{
public:
const fs_reg reg_null_f;
fs_visitor(struct brw_context *brw,
void *mem_ctx,
- const struct brw_wm_prog_key *key,
- struct brw_wm_prog_data *prog_data,
+ gl_shader_stage stage,
+ const void *key,
+ struct brw_stage_prog_data *prog_data,
struct gl_shader_program *shader_prog,
- struct gl_fragment_program *fp,
- unsigned dispatch_width);
-
- fs_visitor(struct brw_context *brw,
- void *mem_ctx,
- const struct brw_vs_prog_key *key,
- struct brw_vs_prog_data *prog_data,
- struct gl_shader_program *shader_prog,
- struct gl_vertex_program *cp,
+ struct gl_program *prog,
unsigned dispatch_width);
~fs_visitor();
- void init();
- fs_reg *variable_storage(ir_variable *var);
fs_reg vgrf(const glsl_type *const type);
fs_reg vgrf(int num_components);
void import_uniforms(fs_visitor *v);
void setup_uniform_clipplane_values();
void compute_clip_distance();
- void visit(ir_variable *ir);
- void visit(ir_assignment *ir);
- void visit(ir_dereference_variable *ir);
- void visit(ir_dereference_record *ir);
- void visit(ir_dereference_array *ir);
- void visit(ir_expression *ir);
- void visit(ir_texture *ir);
- void visit(ir_if *ir);
- void visit(ir_constant *ir);
- void visit(ir_swizzle *ir);
- void visit(ir_return *ir);
- void visit(ir_loop *ir);
- void visit(ir_loop_jump *ir);
- void visit(ir_discard *ir);
- void visit(ir_call *ir);
- void visit(ir_function *ir);
- void visit(ir_function_signature *ir);
- void visit(ir_emit_vertex *);
- void visit(ir_end_primitive *);
-
uint32_t gather_channel(int orig_chan, uint32_t sampler);
void swizzle_result(ir_texture_opcode op, int dest_components,
fs_reg orig_val, uint32_t sampler);
enum brw_conditional_mod condition);
fs_inst *LRP(const fs_reg &dst, const fs_reg &a, const fs_reg &y,
const fs_reg &x);
- fs_inst *DEP_RESOLVE_MOV(int grf);
fs_inst *BFREV(const fs_reg &dst, const fs_reg &value);
fs_inst *BFE(const fs_reg &dst, const fs_reg &bits, const fs_reg &offset,
const fs_reg &value);
fs_inst *end,
const fs_reg ®);
- fs_inst *LOAD_PAYLOAD(const fs_reg &dst, fs_reg *src, int sources);
+ fs_inst *LOAD_PAYLOAD(const fs_reg &dst, fs_reg *src, int sources,
+ int header_size);
- exec_list VARYING_PULL_CONSTANT_LOAD(const fs_reg &dst,
- const fs_reg &surf_index,
- const fs_reg &varying_offset,
- uint32_t const_offset);
+ void VARYING_PULL_CONSTANT_LOAD(const brw::fs_builder &bld,
+ const fs_reg &dst,
+ const fs_reg &surf_index,
+ const fs_reg &varying_offset,
+ uint32_t const_offset);
+ void DEP_RESOLVE_MOV(const brw::fs_builder &bld, int grf);
bool run_fs();
bool run_vs();
+ bool run_cs();
void optimize();
void allocate_registers();
void assign_binding_table_offsets();
void setup_payload_gen4();
void setup_payload_gen6();
void setup_vs_payload();
+ void setup_cs_payload();
void fixup_3src_null_dest();
void assign_curb_setup();
void calculate_urb_setup();
bool opt_register_renaming();
bool register_coalesce();
bool compute_to_mrf();
+ bool eliminate_find_live_channel();
bool dead_code_eliminate();
bool remove_duplicate_mrf_writes();
+
+ bool opt_sampler_eot();
bool virtual_grf_interferes(int a, int b);
void schedule_instructions(instruction_scheduler_mode mode);
void insert_gen4_send_dependency_workarounds();
void no16(const char *msg, ...);
void lower_uniform_pull_constant_loads();
bool lower_load_payload();
+ bool lower_integer_multiplication();
bool opt_combine_constants();
void emit_dummy_fs();
fs_reg shadow_comp,
fs_reg lod, fs_reg lod2, int grad_components,
uint32_t sampler);
+ fs_inst *emit_texture_gen4_simd16(ir_texture_opcode op, fs_reg dst,
+ fs_reg coordinate, int vector_elements,
+ fs_reg shadow_c, fs_reg lod,
+ uint32_t sampler);
fs_inst *emit_texture_gen5(ir_texture_opcode op, fs_reg dst,
fs_reg coordinate, int coord_components,
fs_reg shadow_comp,
fs_reg fix_math_operand(fs_reg src);
fs_inst *emit_math(enum opcode op, fs_reg dst, fs_reg src0);
fs_inst *emit_math(enum opcode op, fs_reg dst, fs_reg src0, fs_reg src1);
- void emit_lrp(const fs_reg &dst, const fs_reg &x, const fs_reg &y,
- const fs_reg &a);
- void emit_minmax(enum brw_conditional_mod conditionalmod, const fs_reg &dst,
- const fs_reg &src0, const fs_reg &src1);
- bool try_emit_saturate(ir_expression *ir);
- bool try_emit_line(ir_expression *ir);
- bool try_emit_mad(ir_expression *ir);
+ fs_inst *emit_lrp(const fs_reg &dst, const fs_reg &x, const fs_reg &y,
+ const fs_reg &a);
+ void emit_discard_jump();
+ /** Copy any live channel from \p src to the first channel of \p dst. */
+ void emit_uniformize(const fs_reg &dst, const fs_reg &src);
bool try_replace_with_sel();
- bool try_opt_frontfacing_ternary(ir_if *ir);
bool opt_peephole_sel();
bool opt_peephole_predicated_break();
bool opt_saturate_propagation();
bool opt_cmod_propagation();
- void emit_bool_to_cond_code(ir_rvalue *condition);
- void emit_if_gen6(ir_if *ir);
+ bool opt_zero_samples();
void emit_unspill(bblock_t *block, fs_inst *inst, fs_reg reg,
uint32_t spill_offset, int count);
void emit_spill(bblock_t *block, fs_inst *inst, fs_reg reg,
uint32_t spill_offset, int count);
- void emit_fragment_program_code();
- void setup_fp_regs();
- fs_reg get_fp_src_reg(const prog_src_register *src);
- fs_reg get_fp_dst_reg(const prog_dst_register *dst);
- void emit_fp_alu1(enum opcode opcode,
- const struct prog_instruction *fpi,
- fs_reg dst, fs_reg src);
- void emit_fp_alu2(enum opcode opcode,
- const struct prog_instruction *fpi,
- fs_reg dst, fs_reg src0, fs_reg src1);
- void emit_fp_scalar_write(const struct prog_instruction *fpi,
- fs_reg dst, fs_reg src);
- void emit_fp_scalar_math(enum opcode opcode,
- const struct prog_instruction *fpi,
- fs_reg dst, fs_reg src);
-
- void emit_fp_minmax(const struct prog_instruction *fpi,
- fs_reg dst, fs_reg src0, fs_reg src1);
-
- void emit_fp_sop(enum brw_conditional_mod conditional_mod,
- const struct prog_instruction *fpi,
- fs_reg dst, fs_reg src0, fs_reg src1, fs_reg one);
-
void emit_nir_code();
void nir_setup_inputs(nir_shader *shader);
void nir_setup_outputs(nir_shader *shader);
void nir_emit_alu(nir_alu_instr *instr);
void nir_emit_intrinsic(nir_intrinsic_instr *instr);
void nir_emit_texture(nir_tex_instr *instr);
- void nir_emit_jump(nir_jump_instr *instr);
+ void nir_emit_jump(const brw::fs_builder &bld,
+ nir_jump_instr *instr);
fs_reg get_nir_src(nir_src src);
fs_reg get_nir_dest(nir_dest dest);
- void emit_percomp(fs_inst *inst, unsigned wr_mask);
+ void emit_percomp(const brw::fs_builder &bld, const fs_inst &inst,
+ unsigned wr_mask);
bool optimize_frontfacing_ternary(nir_alu_instr *instr,
const fs_reg &result);
- int setup_color_payload(fs_reg *dst, fs_reg color, unsigned components,
- bool use_2nd_half);
+ void setup_color_payload(fs_reg *dst, fs_reg color, unsigned components,
+ unsigned exec_size, bool use_2nd_half);
void emit_alpha_test();
- fs_inst *emit_single_fb_write(fs_reg color1, fs_reg color2,
+ fs_inst *emit_single_fb_write(const brw::fs_builder &bld,
+ fs_reg color1, fs_reg color2,
fs_reg src0_alpha, unsigned components,
- bool use_2nd_half = false);
+ unsigned exec_size, bool use_2nd_half = false);
void emit_fb_writes();
void emit_urb_writes();
+ void emit_cs_terminate();
void emit_shader_time_begin();
void emit_shader_time_end();
- void emit_shader_time_write(enum shader_time_shader_type type,
- fs_reg value);
+ void SHADER_TIME_ADD(const brw::fs_builder &bld,
+ enum shader_time_shader_type type, fs_reg value);
void emit_untyped_atomic(unsigned atomic_op, unsigned surf_index,
fs_reg dst, fs_reg offset, fs_reg src0,
void emit_untyped_surface_read(unsigned surf_index, fs_reg dst,
fs_reg offset);
- void emit_interpolate_expression(ir_expression *ir);
-
- bool try_rewrite_rhs_to_dst(ir_assignment *ir,
- fs_reg dst,
- fs_reg src,
- fs_inst *pre_rhs_inst,
- fs_inst *last_rhs_inst);
- void emit_assignment_writes(fs_reg &l, fs_reg &r,
- const glsl_type *type, bool predicated);
void resolve_ud_negate(fs_reg *reg);
- void resolve_bool_comparison(ir_rvalue *rvalue, fs_reg *reg);
- fs_reg get_timestamp();
+ fs_reg get_timestamp(const brw::fs_builder &bld);
struct brw_reg interp_reg(int location, int channel);
- void setup_uniform_values(ir_variable *ir);
- void setup_builtin_uniform_values(ir_variable *ir);
int implied_mrf_writes(fs_inst *inst);
virtual void dump_instructions();
void dump_instruction(backend_instruction *inst);
void dump_instruction(backend_instruction *inst, FILE *file);
- void visit_atomic_counter_intrinsic(ir_call *ir);
-
const void *const key;
+ const struct brw_sampler_prog_key_data *key_tex;
+
struct brw_stage_prog_data *prog_data;
unsigned int sanity_param_count;
/** Number of uniform variable components visited. */
unsigned uniforms;
+ /** Total number of direct uniforms we can get from NIR */
+ unsigned num_direct_uniforms;
+
/** Byte-offset for the next available spot in the scratch space buffer. */
unsigned last_scratch;
*/
int *push_constant_loc;
- struct hash_table *variable_ht;
fs_reg frag_depth;
fs_reg sample_mask;
fs_reg outputs[VARYING_SLOT_MAX];
/** Either BRW_MAX_GRF or GEN7_MRF_HACK_START */
unsigned max_grf;
- fs_reg *fp_temp_regs;
- fs_reg *fp_input_regs;
-
fs_reg *nir_locals;
fs_reg *nir_globals;
fs_reg nir_inputs;
fs_reg nir_outputs;
- fs_reg nir_uniforms;
fs_reg *nir_system_values;
/** @{ debug annotation info */
bool simd16_unsupported;
char *no16_msg;
- /* Result of last visit() method. */
+ /* Result of last visit() method. Still used by emit_texture() */
fs_reg result;
/** Register numbers for thread payload fields. */
fs_reg pixel_y;
fs_reg wpos_w;
fs_reg pixel_w;
- fs_reg delta_x[BRW_WM_BARYCENTRIC_INTERP_MODE_COUNT];
- fs_reg delta_y[BRW_WM_BARYCENTRIC_INTERP_MODE_COUNT];
+ fs_reg delta_xy[BRW_WM_BARYCENTRIC_INTERP_MODE_COUNT];
fs_reg shader_start_time;
fs_reg userplane[MAX_CLIP_PLANES];
bool spilled_any_registers;
const unsigned dispatch_width; /**< 8 or 16 */
+
+ unsigned promoted_constants;
+ brw::fs_builder bld;
};
/**
const void *key,
struct brw_stage_prog_data *prog_data,
struct gl_program *fp,
+ unsigned promoted_constants,
bool runtime_check_aads_emit,
const char *stage_abbrev);
~fs_generator();
GLuint nr);
void generate_fb_write(fs_inst *inst, struct brw_reg payload);
void generate_urb_write(fs_inst *inst, struct brw_reg payload);
+ void generate_cs_terminate(fs_inst *inst, struct brw_reg payload);
void generate_blorp_fb_write(fs_inst *inst);
- void generate_pixel_xy(struct brw_reg dst, bool is_x);
void generate_linterp(fs_inst *inst, struct brw_reg dst,
struct brw_reg *src);
void generate_tex(fs_inst *inst, struct brw_reg dst, struct brw_reg src,
struct brw_reg offset,
struct brw_reg value);
- void generate_untyped_atomic(fs_inst *inst,
- struct brw_reg dst,
- struct brw_reg payload,
- struct brw_reg atomic_op,
- struct brw_reg surf_index);
-
- void generate_untyped_surface_read(fs_inst *inst,
- struct brw_reg dst,
- struct brw_reg payload,
- struct brw_reg surf_index);
-
bool patch_discard_jumps_to_fb_writes();
struct brw_context *brw;
- struct gl_context *ctx;
+ const struct brw_device_info *devinfo;
- struct brw_compile *p;
+ struct brw_codegen *p;
const void * const key;
struct brw_stage_prog_data * const prog_data;
unsigned dispatch_width; /**< 8 or 16 */
exec_list discard_halt_patches;
+ unsigned promoted_constants;
bool runtime_check_aads_emit;
bool debug_flag;
const char *shader_name;
bool brw_do_channel_expressions(struct exec_list *instructions);
bool brw_do_vector_splitting(struct exec_list *instructions);
+void brw_setup_tex_for_precompile(struct brw_context *brw,
+ struct brw_sampler_prog_key_data *tex,
+ struct gl_program *prog);
projects / mesa.git / blobdiff