X-Git-Url: https://git.libre-soc.org/?a=blobdiff_plain;f=src%2Fmesa%2Fdrivers%2Fdri%2Fi965%2Fbrw_fs.h;h=9c6b047d6ea059abb1044df8c19932a01349912f;hb=1c89e098e8e644d6c33b36fabbba0b8d675d115d;hp=3d0da23acf1d3337b98d5c970e281a68080f61e4;hpb=e58992aedd9693f0356f3691d510a5e976473a0c;p=mesa.git diff --git a/src/mesa/drivers/dri/i965/brw_fs.h b/src/mesa/drivers/dri/i965/brw_fs.h index 3d0da23acf1..9c6b047d6ea 100644 --- a/src/mesa/drivers/dri/i965/brw_fs.h +++ b/src/mesa/drivers/dri/i965/brw_fs.h @@ -39,16 +39,15 @@ extern "C" { #include "program/prog_parameter.h" #include "program/prog_print.h" #include "program/prog_optimize.h" -#include "program/register_allocate.h" +#include "util/register_allocate.h" #include "program/sampler.h" #include "program/hash_table.h" #include "brw_context.h" #include "brw_eu.h" #include "brw_wm.h" #include "brw_shader.h" -#include "intel_asm_printer.h" +#include "intel_asm_annotation.h" } -#include "gen8_generator.h" #include "glsl/glsl_types.h" #include "glsl/ir.h" @@ -63,7 +62,7 @@ namespace brw { class fs_live_variables; } -class fs_reg { +class fs_reg : public backend_reg { public: DECLARE_RALLOC_CXX_OPERATORS(fs_reg) @@ -75,64 +74,31 @@ public: fs_reg(uint32_t u); fs_reg(struct brw_reg fixed_hw_reg); fs_reg(enum register_file file, int reg); - fs_reg(enum register_file file, int reg, uint32_t type); + fs_reg(enum register_file file, int reg, enum brw_reg_type type); fs_reg(class fs_visitor *v, const struct glsl_type *type); bool equals(const fs_reg &r) const; - bool is_zero() const; - bool is_one() const; - bool is_null() const; bool is_valid_3src() const; bool is_contiguous() const; - bool is_accumulator() const; fs_reg &apply_stride(unsigned stride); /** Smear a channel of the reg to all channels. */ fs_reg &set_smear(unsigned subreg); - /** Register file: GRF, MRF, IMM. */ - enum register_file file; - /** Register type. BRW_REGISTER_TYPE_* */ - uint8_t type; - /** - * Register number. For MRF, it's the hardware register. For - * GRF, it's a virtual register number until register allocation - */ - uint16_t reg; - /** - * Offset from the start of the contiguous register block. - * - * For pre-register-allocation GRFs, this is in units of a float per pixel - * (1 hardware register for SIMD8 mode, or 2 registers for SIMD16 mode). - * For uniforms, this is in units of 1 float. - */ - int reg_offset; /** * Offset in bytes from the start of the register. Values up to a * backend_reg::reg_offset unit are valid. */ int subreg_offset; - /** Value for file == IMM */ - union { - int32_t i; - uint32_t u; - float f; - } imm; - - struct brw_reg fixed_hw_reg; - fs_reg *reladdr; - bool negate; - bool abs; - /** Register region horizontal stride */ uint8_t stride; }; static inline fs_reg -retype(fs_reg reg, unsigned type) +retype(fs_reg reg, enum brw_reg_type type) { reg.fixed_hw_reg.type = reg.type = type; return reg; @@ -168,9 +134,6 @@ half(const fs_reg ®, unsigned idx) } static const fs_reg reg_undef; -static const fs_reg reg_null_f(retype(brw_null_reg(), BRW_REGISTER_TYPE_F)); -static const fs_reg reg_null_d(retype(brw_null_reg(), BRW_REGISTER_TYPE_D)); -static const fs_reg reg_null_ud(retype(brw_null_reg(), BRW_REGISTER_TYPE_UD)); class ip_record : public exec_node { public: @@ -216,29 +179,20 @@ public: fs_reg dst; fs_reg *src; - uint32_t texture_offset; /**< Texture offset bitfield */ - uint32_t offset; /* spill/unspill offset */ - uint8_t sources; /**< Number of fs_reg sources. */ - uint8_t conditional_mod; /**< BRW_CONDITIONAL_* */ /* Chooses which flag subregister (f0.0 or f0.1) is used for conditional * mod and predication. */ uint8_t flag_subreg; - uint8_t mlen; /**< SEND message length */ uint8_t regs_written; /**< Number of vgrfs written by a SEND message, or 1 */ - int8_t base_mrf; /**< First MRF in the SEND message, if mlen is nonzero. */ - uint8_t sampler; - uint8_t target; /**< MRT target. */ - bool saturate:1; bool eot:1; bool header_present:1; bool shadow_compare:1; bool force_uncompressed:1; bool force_sechalf:1; - bool force_writemask_all:1; + bool pi_noperspective:1; /**< Pixel interpolator noperspective flag */ }; /** @@ -249,6 +203,9 @@ public: class fs_visitor : public backend_visitor { public: + const fs_reg reg_null_f; + const fs_reg reg_null_d; + const fs_reg reg_null_ud; fs_visitor(struct brw_context *brw, void *mem_ctx, @@ -258,6 +215,7 @@ public: struct gl_fragment_program *fp, unsigned dispatch_width); ~fs_visitor(); + void init(); fs_reg *variable_storage(ir_variable *var); int virtual_grf_alloc(int size); @@ -283,8 +241,8 @@ public: void visit(ir_emit_vertex *); void visit(ir_end_primitive *); - uint32_t gather_channel(ir_texture *ir, int sampler); - void swizzle_result(ir_texture *ir, fs_reg orig_val, int sampler); + uint32_t gather_channel(ir_texture *ir, uint32_t sampler); + void swizzle_result(ir_texture *ir, fs_reg orig_val, uint32_t sampler); fs_inst *emit(fs_inst *inst); void emit(exec_list list); @@ -315,10 +273,11 @@ public: fs_inst *AND(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1); fs_inst *OR(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1); fs_inst *XOR(const fs_reg &dst, const fs_reg &src0, const fs_reg &src1); - fs_inst *IF(uint32_t predicate); - fs_inst *IF(const fs_reg &src0, const fs_reg &src1, uint32_t condition); + fs_inst *IF(enum brw_predicate predicate); + fs_inst *IF(const fs_reg &src0, const fs_reg &src1, + enum brw_conditional_mod condition); fs_inst *CMP(fs_reg dst, fs_reg src0, fs_reg src1, - uint32_t condition); + 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); @@ -375,12 +334,13 @@ public: void calculate_register_pressure(); bool opt_algebraic(); bool opt_cse(); - bool opt_cse_local(bblock_t *block, exec_list *aeb); + bool opt_cse_local(bblock_t *block); bool opt_copy_propagate(); bool try_copy_propagate(fs_inst *inst, int arg, acp_entry *entry); bool opt_copy_propagate_local(void *mem_ctx, bblock_t *block, exec_list *acp); void opt_drop_redundant_mov_to_flags(); + bool opt_register_renaming(); bool register_coalesce(); bool compute_to_mrf(); bool dead_code_eliminate(); @@ -388,8 +348,10 @@ public: bool virtual_grf_interferes(int a, int b); void schedule_instructions(instruction_scheduler_mode mode); void insert_gen4_send_dependency_workarounds(); - void insert_gen4_pre_send_dependency_workarounds(fs_inst *inst); - void insert_gen4_post_send_dependency_workarounds(fs_inst *inst); + void insert_gen4_pre_send_dependency_workarounds(bblock_t *block, + fs_inst *inst); + void insert_gen4_post_send_dependency_workarounds(bblock_t *block, + fs_inst *inst); void vfail(const char *msg, va_list args); void fail(const char *msg, ...); void no16(const char *msg, ...); @@ -400,35 +362,37 @@ public: void pop_force_uncompressed(); void emit_dummy_fs(); + void emit_repclear_shader(); fs_reg *emit_fragcoord_interpolation(ir_variable *ir); fs_inst *emit_linterp(const fs_reg &attr, const fs_reg &interp, glsl_interp_qualifier interpolation_mode, bool is_centroid, bool is_sample); - fs_reg *emit_frontfacing_interpolation(ir_variable *ir); - fs_reg *emit_samplepos_setup(ir_variable *ir); + fs_reg *emit_frontfacing_interpolation(); + fs_reg *emit_samplepos_setup(); fs_reg *emit_sampleid_setup(ir_variable *ir); fs_reg *emit_general_interpolation(ir_variable *ir); void emit_interpolation_setup_gen4(); void emit_interpolation_setup_gen6(); void compute_sample_position(fs_reg dst, fs_reg int_sample_pos); fs_reg rescale_texcoord(ir_texture *ir, fs_reg coordinate, - bool is_rect, int sampler, int texunit); + bool is_rect, uint32_t sampler, int texunit); fs_inst *emit_texture_gen4(ir_texture *ir, fs_reg dst, fs_reg coordinate, - fs_reg shadow_comp, fs_reg lod, fs_reg lod2); + fs_reg shadow_comp, fs_reg lod, fs_reg lod2, + uint32_t sampler); fs_inst *emit_texture_gen5(ir_texture *ir, fs_reg dst, fs_reg coordinate, fs_reg shadow_comp, fs_reg lod, fs_reg lod2, - fs_reg sample_index); + fs_reg sample_index, uint32_t sampler); fs_inst *emit_texture_gen7(ir_texture *ir, fs_reg dst, fs_reg coordinate, fs_reg shadow_comp, fs_reg lod, fs_reg lod2, - fs_reg sample_index, fs_reg mcs, int sampler); - fs_reg emit_mcs_fetch(ir_texture *ir, fs_reg coordinate, int sampler); + fs_reg sample_index, fs_reg mcs, fs_reg sampler); + fs_reg emit_mcs_fetch(ir_texture *ir, fs_reg coordinate, fs_reg sampler); void emit_gen6_gather_wa(uint8_t wa, fs_reg dst); 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(uint32_t conditionalmod, const fs_reg &dst, + 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_mad(ir_expression *ir); @@ -438,8 +402,8 @@ public: bool opt_saturate_propagation(); void emit_bool_to_cond_code(ir_rvalue *condition); void emit_if_gen6(ir_if *ir); - void emit_unspill(fs_inst *inst, fs_reg reg, uint32_t spill_offset, - int count); + void emit_unspill(bblock_t *block, fs_inst *inst, fs_reg reg, + uint32_t spill_offset, int count); void emit_fragment_program_code(); void setup_fp_regs(); @@ -460,7 +424,7 @@ public: void emit_fp_minmax(const struct prog_instruction *fpi, fs_reg dst, fs_reg src0, fs_reg src1); - void emit_fp_sop(uint32_t conditional_mod, + 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); @@ -480,6 +444,8 @@ public: 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, @@ -504,9 +470,8 @@ public: void visit_atomic_counter_intrinsic(ir_call *ir); - struct gl_fragment_program *fp; - const struct brw_wm_prog_key *const key; - struct brw_wm_prog_data *prog_data; + const void *const key; + struct brw_stage_prog_data *prog_data; unsigned int sanity_param_count; int *param_size; @@ -610,29 +575,30 @@ public: void *mem_ctx, const struct brw_wm_prog_key *key, struct brw_wm_prog_data *prog_data, - struct gl_shader_program *prog, + struct gl_shader_program *shader_prog, struct gl_fragment_program *fp, - bool dual_source_output, bool runtime_check_aads_emit, bool debug_flag); ~fs_generator(); - const unsigned *generate_assembly(exec_list *simd8_instructions, - exec_list *simd16_instructions, + const unsigned *generate_assembly(const cfg_t *simd8_cfg, + const cfg_t *simd16_cfg, unsigned *assembly_size); private: - void generate_code(exec_list *instructions); + void generate_code(const cfg_t *cfg); void fire_fb_write(fs_inst *inst, GLuint base_reg, struct brw_reg implied_header, GLuint nr); void generate_fb_write(fs_inst *inst); void generate_blorp_fb_write(fs_inst *inst); + void generate_rep_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); + void generate_tex(fs_inst *inst, struct brw_reg dst, struct brw_reg src, + struct brw_reg sampler_index); void generate_math_gen6(fs_inst *inst, struct brw_reg dst, struct brw_reg src0, @@ -643,9 +609,9 @@ private: void generate_math_g45(fs_inst *inst, struct brw_reg dst, struct brw_reg src); - void generate_ddx(fs_inst *inst, struct brw_reg dst, struct brw_reg src); + void generate_ddx(fs_inst *inst, struct brw_reg dst, struct brw_reg src, struct brw_reg quality); void generate_ddy(fs_inst *inst, struct brw_reg dst, struct brw_reg src, - bool negate_value); + struct brw_reg quality, bool negate_value); void generate_scratch_write(fs_inst *inst, struct brw_reg src); void generate_scratch_read(fs_inst *inst, struct brw_reg dst); void generate_scratch_read_gen7(fs_inst *inst, struct brw_reg dst); @@ -665,6 +631,12 @@ private: struct brw_reg offset); void generate_mov_dispatch_to_flags(fs_inst *inst); + void generate_pixel_interpolator_query(fs_inst *inst, + struct brw_reg dst, + struct brw_reg src, + struct brw_reg msg_data, + unsigned msg_type); + void generate_set_omask(fs_inst *inst, struct brw_reg dst, struct brw_reg sample_mask); @@ -707,106 +679,21 @@ private: struct gl_context *ctx; struct brw_compile *p; - const struct brw_wm_prog_key *const key; - struct brw_wm_prog_data *prog_data; + gl_shader_stage stage; + const void * const key; + struct brw_stage_prog_data * const prog_data; - struct gl_shader_program *prog; - const struct gl_fragment_program *fp; + struct gl_shader_program * const shader_prog; + const struct gl_program *prog; unsigned dispatch_width; /**< 8 or 16 */ exec_list discard_halt_patches; - bool dual_source_output; bool runtime_check_aads_emit; const bool debug_flag; void *mem_ctx; }; -/** - * The fragment shader code generator. - * - * Translates FS IR to actual i965 assembly code. - */ -class gen8_fs_generator : public gen8_generator -{ -public: - gen8_fs_generator(struct brw_context *brw, - void *mem_ctx, - const struct brw_wm_prog_key *key, - struct brw_wm_prog_data *prog_data, - struct gl_shader_program *prog, - struct gl_fragment_program *fp, - bool dual_source_output); - ~gen8_fs_generator(); - - const unsigned *generate_assembly(exec_list *simd8_instructions, - exec_list *simd16_instructions, - unsigned *assembly_size); - -private: - void generate_code(exec_list *instructions); - void generate_fb_write(fs_inst *inst); - 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); - void generate_math1(fs_inst *inst, struct brw_reg dst, struct brw_reg src); - void generate_math2(fs_inst *inst, struct brw_reg dst, - struct brw_reg src0, struct brw_reg src1); - void generate_ddx(fs_inst *inst, struct brw_reg dst, struct brw_reg src); - void generate_ddy(fs_inst *inst, struct brw_reg dst, struct brw_reg src, - bool negate_value); - void generate_scratch_write(fs_inst *inst, struct brw_reg src); - void generate_scratch_read(fs_inst *inst, struct brw_reg dst); - void generate_scratch_read_gen7(fs_inst *inst, struct brw_reg dst); - void generate_uniform_pull_constant_load(fs_inst *inst, - struct brw_reg dst, - struct brw_reg index, - struct brw_reg offset); - void generate_varying_pull_constant_load(fs_inst *inst, - struct brw_reg dst, - struct brw_reg index, - struct brw_reg offset); - void generate_mov_dispatch_to_flags(fs_inst *ir); - void generate_set_omask(fs_inst *ir, - struct brw_reg dst, - struct brw_reg sample_mask); - void generate_set_sample_id(fs_inst *ir, - struct brw_reg dst, - struct brw_reg src0, - struct brw_reg src1); - void generate_set_simd4x2_offset(fs_inst *ir, - struct brw_reg dst, - struct brw_reg offset); - void generate_pack_half_2x16_split(fs_inst *inst, - struct brw_reg dst, - struct brw_reg x, - struct brw_reg y); - void generate_unpack_half_2x16_split(fs_inst *inst, - struct brw_reg dst, - struct brw_reg src); - void generate_untyped_atomic(fs_inst *inst, - struct brw_reg dst, - 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 surf_index); - void generate_discard_jump(fs_inst *ir); - - bool patch_discard_jumps_to_fb_writes(); - - const struct brw_wm_prog_key *const key; - struct brw_wm_prog_data *prog_data; - const struct gl_fragment_program *fp; - - unsigned dispatch_width; /** 8 or 16 */ - - bool dual_source_output; - - exec_list discard_halt_patches; -}; - bool brw_do_channel_expressions(struct exec_list *instructions); bool brw_do_vector_splitting(struct exec_list *instructions); bool brw_fs_precompile(struct gl_context *ctx, struct gl_shader_program *prog);