panfrost/midgard/compiler.h

/*
 * Copyright (C) 2019 Alyssa Rosenzweig <alyssa@rosenzweig.io>
 *
 * Permission is hereby granted, free of charge, to any person obtaining a
 * copy of this software and associated documentation files (the "Software"),
 * to deal in the Software without restriction, including without limitation
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,
 * and/or sell copies of the Software, and to permit persons to whom the
 * Software is furnished to do so, subject to the following conditions:
 *
 * The above copyright notice and this permission notice (including the next
 * paragraph) shall be included in all copies or substantial portions of the
 * Software.
 *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.  IN NO EVENT SHALL
 * THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
 * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
 * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
 * SOFTWARE.
 */

#ifndef _MDG_COMPILER_H
#define _MDG_COMPILER_H

#include "midgard.h"
#include "helpers.h"
#include "midgard_compile.h"
#include "midgard_ops.h"

#include "util/hash_table.h"
#include "util/u_dynarray.h"
#include "util/set.h"
#include "util/list.h"

#include "main/mtypes.h"
#include "compiler/nir_types.h"
#include "compiler/nir/nir.h"
#include "panfrost/util/pan_ir.h"
#include "panfrost/util/lcra.h"

/* Forward declare */
struct midgard_block;

/* Target types. Defaults to TARGET_GOTO (the type corresponding directly to
 * the hardware), hence why that must be zero. TARGET_DISCARD signals this
 * instruction is actually a discard op. */

#define TARGET_GOTO 0
#define TARGET_BREAK 1
#define TARGET_CONTINUE 2
#define TARGET_DISCARD 3
#define TARGET_TILEBUF_WAIT 4

typedef struct midgard_branch {
        /* If conditional, the condition is specified in r31.w */
        bool conditional;

        /* For conditionals, if this is true, we branch on FALSE. If false, we  branch on TRUE. */
        bool invert_conditional;

        /* Branch targets: the start of a block, the start of a loop (continue), the end of a loop (break). Value is one of TARGET_ */
        unsigned target_type;

        /* The actual target */
        union {
                int target_block;
                int target_break;
                int target_continue;
        };
} midgard_branch;

#define PAN_WRITEOUT_C 1
#define PAN_WRITEOUT_Z 2
#define PAN_WRITEOUT_S 4

/* Generic in-memory data type repesenting a single logical instruction, rather
 * than a single instruction group. This is the preferred form for code gen.
 * Multiple midgard_insturctions will later be combined during scheduling,
 * though this is not represented in this structure.  Its format bridges
 * the low-level binary representation with the higher level semantic meaning.
 *
 * Notably, it allows registers to be specified as block local SSA, for code
 * emitted before the register allocation pass.
 */

#define MIR_SRC_COUNT 4
#define MIR_VEC_COMPONENTS 16

typedef struct midgard_instruction {
        /* Must be first for casting */
        struct list_head link;

        unsigned type; /* ALU, load/store, texture */

        /* Instruction arguments represented as block-local SSA
         * indices, rather than registers. ~0 means unused. */
        unsigned src[MIR_SRC_COUNT];
        unsigned dest;

        /* vec16 swizzle, unpacked, per source */
        unsigned swizzle[MIR_SRC_COUNT][MIR_VEC_COMPONENTS];

        /* Types! */
        nir_alu_type src_types[MIR_SRC_COUNT];
        nir_alu_type dest_type;

        /* Packing ops have non-32-bit dest types even though they functionally
         * work at the 32-bit level, use this as a signal to disable copyprop.
         * We maybe need synthetic pack ops instead. */
        bool is_pack;

        /* Modifiers, depending on type */
        union {
                struct {
                        bool src_abs[MIR_SRC_COUNT];
                        bool src_neg[MIR_SRC_COUNT];
                };

                struct {
                        bool src_shift[MIR_SRC_COUNT];
                };
        };

        /* Out of the union for csel (could maybe be fixed..) */
        bool src_invert[MIR_SRC_COUNT];

        /* If the op supports it */
        enum midgard_roundmode roundmode;

        /* For textures: should helpers execute this instruction (instead of
         * just helping with derivatives)? Should helpers terminate after? */
        bool helper_terminate;
        bool helper_execute;

        /* I.e. (1 << alu_bit) */
        int unit;

        bool has_constants;
        midgard_constants constants;
        uint16_t inline_constant;
        bool has_blend_constant;
        bool has_inline_constant;

        bool compact_branch;
        uint8_t writeout;
        bool last_writeout;

        /* Masks in a saneish format. One bit per channel, not packed fancy.
         * Use this instead of the op specific ones, and switch over at emit
         * time */

        uint16_t mask;

        /* Hint for the register allocator not to spill the destination written
         * from this instruction (because it is a spill/unspill node itself).
         * Bitmask of spilled classes */

        unsigned no_spill;

        /* Generic hint for intra-pass use */
        bool hint;

        /* During scheduling, the backwards dependency graph
         * (DAG). nr_dependencies is the number of unscheduled
         * instructions that must still be scheduled after
         * (before) this instruction. dependents are which
         * instructions need to be scheduled before (after) this
         * instruction. */

        unsigned nr_dependencies;
        BITSET_WORD *dependents;

        /* Use this in conjunction with `type` */
        unsigned op;

        /* This refers to midgard_outmod_float or midgard_outmod_int.
         * In case of a ALU op, use midgard_is_integer_out_op() to know which
         * one is used.
         * If it's a texture op, it's always midgard_outmod_float. */
        unsigned outmod;

        union {
                midgard_load_store_word load_store;
                midgard_texture_word texture;

                midgard_branch branch;
        };
} midgard_instruction;

typedef struct midgard_block {
        pan_block base;

        bool scheduled;

        /* List of midgard_bundles emitted (after the scheduler has run) */
        struct util_dynarray bundles;

        /* Number of quadwords _actually_ emitted, as determined after scheduling */
        unsigned quadword_count;

        /* Indicates this is a fixed-function fragment epilogue block */
        bool epilogue;

        /* Are helper invocations required by this block? */
        bool helpers_in;
} midgard_block;

typedef struct midgard_bundle {
        /* Tag for the overall bundle */
        int tag;

        /* Instructions contained by the bundle. instruction_count <= 6 (vmul,
         * sadd, vadd, smul, vlut, branch) */
        int instruction_count;
        midgard_instruction *instructions[6];

        /* Bundle-wide ALU configuration */
        int padding;
        int control;
        bool has_embedded_constants;
        midgard_constants constants;
        bool has_blend_constant;
        bool last_writeout;
} midgard_bundle;

enum midgard_rt_id {
        MIDGARD_COLOR_RT0 = 0,
        MIDGARD_COLOR_RT1,
        MIDGARD_COLOR_RT2,
        MIDGARD_COLOR_RT3,
        MIDGARD_COLOR_RT4,
        MIDGARD_COLOR_RT5,
        MIDGARD_COLOR_RT6,
        MIDGARD_COLOR_RT7,
        MIDGARD_ZS_RT,
        MIDGARD_NUM_RTS,
};

typedef struct compiler_context {
        nir_shader *nir;
        gl_shader_stage stage;

        /* Is internally a blend shader? Depends on stage == FRAGMENT */
        bool is_blend;

        /* Render target number for a keyed blend shader. Depends on is_blend */
        unsigned blend_rt;

        /* Index to precolour to r0 for an input blend colour */
        unsigned blend_input;

        /* Index to precolour to r2 for a dual-source blend colour */
        unsigned blend_src1;

        /* Tracking for blend constant patching */
        int blend_constant_offset;

        /* Number of bytes used for Thread Local Storage */
        unsigned tls_size;

        /* Count of spills and fills for shaderdb */
        unsigned spills;
        unsigned fills;

        /* Current NIR function */
        nir_function *func;

        /* Allocated compiler temporary counter */
        unsigned temp_alloc;

        /* Unordered list of midgard_blocks */
        int block_count;
        struct list_head blocks;

        /* TODO merge with block_count? */
        unsigned block_source_count;

        /* List of midgard_instructions emitted for the current block */
        midgard_block *current_block;

        /* If there is a preset after block, use this, otherwise emit_block will create one if NULL */
        midgard_block *after_block;

        /* The current "depth" of the loop, for disambiguating breaks/continues
         * when using nested loops */
        int current_loop_depth;

        /* Total number of loops for shader-db */
        unsigned loop_count;

        /* Constants which have been loaded, for later inlining */
        struct hash_table_u64 *ssa_constants;

        int temp_count;
        int max_hash;

        /* Set of NIR indices that were already emitted as outmods */
        BITSET_WORD *already_emitted;

        /* Just the count of the max register used. Higher count => higher
         * register pressure */
        int work_registers;

        /* The number of uniforms allowable for the fast path */
        int uniform_cutoff;

        /* Count of instructions emitted from NIR overall, across all blocks */
        int instruction_count;

        /* Alpha ref value passed in */
        float alpha_ref;

        unsigned quadword_count;

        /* Bitmask of valid metadata */
        unsigned metadata;

        /* Model-specific quirk set */
        uint32_t quirks;

        /* Writeout instructions for each render target */
        midgard_instruction *writeout_branch[MIDGARD_NUM_RTS];

        struct panfrost_sysvals sysvals;
} compiler_context;

/* Per-block live_in/live_out */
#define MIDGARD_METADATA_LIVENESS (1 << 0)

/* Helpers for manipulating the above structures (forming the driver IR) */

/* Append instruction to end of current block */

static inline midgard_instruction *
mir_upload_ins(struct compiler_context *ctx, struct midgard_instruction ins)
{
        midgard_instruction *heap = ralloc(ctx, struct midgard_instruction);
        memcpy(heap, &ins, sizeof(ins));
        return heap;
}

static inline midgard_instruction *
emit_mir_instruction(struct compiler_context *ctx, struct midgard_instruction ins)
{
        midgard_instruction *u = mir_upload_ins(ctx, ins);
        list_addtail(&u->link, &ctx->current_block->base.instructions);
        return u;
}

static inline struct midgard_instruction *
mir_insert_instruction_before(struct compiler_context *ctx,
                              struct midgard_instruction *tag,
                              struct midgard_instruction ins)
{
        struct midgard_instruction *u = mir_upload_ins(ctx, ins);
        list_addtail(&u->link, &tag->link);
        return u;
}

static inline void
mir_remove_instruction(struct midgard_instruction *ins)
{
        list_del(&ins->link);
}

static inline midgard_instruction*
mir_prev_op(struct midgard_instruction *ins)
{
        return list_last_entry(&(ins->link), midgard_instruction, link);
}

static inline midgard_instruction*
mir_next_op(struct midgard_instruction *ins)
{
        return list_first_entry(&(ins->link), midgard_instruction, link);
}

#define mir_foreach_block(ctx, v) \
        list_for_each_entry(pan_block, v, &ctx->blocks, link)

#define mir_foreach_block_from(ctx, from, v) \
        list_for_each_entry_from(pan_block, v, &from->base, &ctx->blocks, link)

#define mir_foreach_instr_in_block(block, v) \
        list_for_each_entry(struct midgard_instruction, v, &block->base.instructions, link)
#define mir_foreach_instr_in_block_rev(block, v) \
        list_for_each_entry_rev(struct midgard_instruction, v, &block->base.instructions, link)

#define mir_foreach_instr_in_block_safe(block, v) \
        list_for_each_entry_safe(struct midgard_instruction, v, &block->base.instructions, link)

#define mir_foreach_instr_in_block_safe_rev(block, v) \
        list_for_each_entry_safe_rev(struct midgard_instruction, v, &block->base.instructions, link)

#define mir_foreach_instr_in_block_from(block, v, from) \
        list_for_each_entry_from(struct midgard_instruction, v, from, &block->base.instructions, link)

#define mir_foreach_instr_in_block_from_rev(block, v, from) \
        list_for_each_entry_from_rev(struct midgard_instruction, v, from, &block->base.instructions, link)

#define mir_foreach_bundle_in_block(block, v) \
        util_dynarray_foreach(&block->bundles, midgard_bundle, v)

#define mir_foreach_bundle_in_block_rev(block, v) \
        util_dynarray_foreach_reverse(&block->bundles, midgard_bundle, v)

#define mir_foreach_instr_in_block_scheduled_rev(block, v) \
        midgard_instruction* v; \
        signed i = 0; \
        mir_foreach_bundle_in_block_rev(block, _bundle) \
                for (i = (_bundle->instruction_count - 1), v = _bundle->instructions[i]; \
                                i >= 0; \
                                --i, v = (i >= 0) ? _bundle->instructions[i] : NULL) \

#define mir_foreach_instr_global(ctx, v) \
        mir_foreach_block(ctx, v_block) \
                mir_foreach_instr_in_block(((midgard_block *) v_block), v)

#define mir_foreach_instr_global_safe(ctx, v) \
        mir_foreach_block(ctx, v_block) \
                mir_foreach_instr_in_block_safe(((midgard_block *) v_block), v)

/* Based on set_foreach, expanded with automatic type casts */

#define mir_foreach_predecessor(blk, v) \
        struct set_entry *_entry_##v; \
        struct midgard_block *v; \
        for (_entry_##v = _mesa_set_next_entry(blk->base.predecessors, NULL), \
                v = (struct midgard_block *) (_entry_##v ? _entry_##v->key : NULL);  \
                _entry_##v != NULL; \
                _entry_##v = _mesa_set_next_entry(blk->base.predecessors, _entry_##v), \
                v = (struct midgard_block *) (_entry_##v ? _entry_##v->key : NULL))

#define mir_foreach_src(ins, v) \
        for (unsigned v = 0; v < ARRAY_SIZE(ins->src); ++v)

static inline midgard_instruction *
mir_last_in_block(struct midgard_block *block)
{
        return list_last_entry(&block->base.instructions, struct midgard_instruction, link);
}

static inline midgard_block *
mir_get_block(compiler_context *ctx, int idx)
{
        struct list_head *lst = &ctx->blocks;

        while ((idx--) + 1)
                lst = lst->next;

        return (struct midgard_block *) lst;
}

static inline bool
mir_is_alu_bundle(midgard_bundle *bundle)
{
        return IS_ALU(bundle->tag);
}

static inline unsigned
make_compiler_temp(compiler_context *ctx)
{
        return (ctx->func->impl->ssa_alloc + ctx->temp_alloc++) << 1;
}

static inline unsigned
make_compiler_temp_reg(compiler_context *ctx)
{
        return ((ctx->func->impl->reg_alloc + ctx->temp_alloc++) << 1) | PAN_IS_REG;
}

static inline unsigned
nir_ssa_index(nir_ssa_def *ssa)
{
        return (ssa->index << 1) | 0;
}

static inline unsigned
nir_src_index(compiler_context *ctx, nir_src *src)
{
        if (src->is_ssa)
                return nir_ssa_index(src->ssa);
        else {
                assert(!src->reg.indirect);
                return (src->reg.reg->index << 1) | PAN_IS_REG;
        }
}

static inline unsigned
nir_dest_index(nir_dest *dst)
{
        if (dst->is_ssa)
                return (dst->ssa.index << 1) | 0;
        else {
                assert(!dst->reg.indirect);
                return (dst->reg.reg->index << 1) | PAN_IS_REG;
        }
}


/* MIR manipulation */

void mir_rewrite_index(compiler_context *ctx, unsigned old, unsigned new);
void mir_rewrite_index_src(compiler_context *ctx, unsigned old, unsigned new);
void mir_rewrite_index_dst(compiler_context *ctx, unsigned old, unsigned new);
void mir_rewrite_index_dst_single(midgard_instruction *ins, unsigned old, unsigned new);
void mir_rewrite_index_src_single(midgard_instruction *ins, unsigned old, unsigned new);
void mir_rewrite_index_src_swizzle(compiler_context *ctx, unsigned old, unsigned new, unsigned *swizzle);
bool mir_single_use(compiler_context *ctx, unsigned value);
unsigned mir_use_count(compiler_context *ctx, unsigned value);
uint16_t mir_bytemask_of_read_components(midgard_instruction *ins, unsigned node);
uint16_t mir_bytemask_of_read_components_index(midgard_instruction *ins, unsigned i);
uint16_t mir_from_bytemask(uint16_t bytemask, unsigned bits);
uint16_t mir_bytemask(midgard_instruction *ins);
uint16_t mir_round_bytemask_up(uint16_t mask, unsigned bits);
void mir_set_bytemask(midgard_instruction *ins, uint16_t bytemask);
signed mir_upper_override(midgard_instruction *ins, unsigned inst_size);
unsigned mir_components_for_type(nir_alu_type T);
unsigned max_bitsize_for_alu(midgard_instruction *ins);
midgard_reg_mode reg_mode_for_bitsize(unsigned bitsize);

/* MIR printing */

void mir_print_instruction(midgard_instruction *ins);
void mir_print_bundle(midgard_bundle *ctx);
void mir_print_block(midgard_block *block);
void mir_print_shader(compiler_context *ctx);
bool mir_nontrivial_mod(midgard_instruction *ins, unsigned i, bool check_swizzle);
bool mir_nontrivial_outmod(midgard_instruction *ins);

void mir_insert_instruction_before_scheduled(compiler_context *ctx, midgard_block *block, midgard_instruction *tag, midgard_instruction ins);
void mir_insert_instruction_after_scheduled(compiler_context *ctx, midgard_block *block, midgard_instruction *tag, midgard_instruction ins);
void mir_flip(midgard_instruction *ins);
void mir_compute_temp_count(compiler_context *ctx);

void mir_set_offset(compiler_context *ctx, midgard_instruction *ins, nir_src *offset, bool is_shared);

/* 'Intrinsic' move for aliasing */

static inline midgard_instruction
v_mov(unsigned src, unsigned dest)
{
        midgard_instruction ins = {
                .type = TAG_ALU_4,
                .mask = 0xF,
                .src = { ~0, src, ~0, ~0 },
                .src_types = { 0, nir_type_uint32 },
                .swizzle = SWIZZLE_IDENTITY,
                .dest = dest,
                .dest_type = nir_type_uint32,
                .op = midgard_alu_op_imov,
                .outmod = midgard_outmod_int_wrap
        };

        return ins;
}

/* Broad types of register classes so we can handle special
 * registers */

#define REG_CLASS_WORK          0
#define REG_CLASS_LDST          1
#define REG_CLASS_TEXR          3
#define REG_CLASS_TEXW          4

/* Like a move, but to thread local storage! */

static inline midgard_instruction
v_load_store_scratch(
                unsigned srcdest,
                unsigned index,
                bool is_store,
                unsigned mask)
{
        /* We index by 32-bit vec4s */
        unsigned byte = (index * 4 * 4);

        midgard_instruction ins = {
                .type = TAG_LOAD_STORE_4,
                .mask = mask,
                .dest_type = nir_type_uint32,
                .dest = ~0,
                .src = { ~0, ~0, ~0, ~0 },
                .swizzle = SWIZZLE_IDENTITY_4,
                .op = is_store ? midgard_op_st_int4 : midgard_op_ld_int4,
                .load_store = {
                        /* For register spilling - to thread local storage */
                        .arg_1 = 0xEA,
                        .arg_2 = 0x1E,
                },

                /* If we spill an unspill, RA goes into an infinite loop */
                .no_spill = (1 << REG_CLASS_WORK)
        };

        ins.constants.u32[0] = byte;

        if (is_store) {
                ins.src[0] = srcdest;
                ins.src_types[0] = nir_type_uint32;

                /* Ensure we are tightly swizzled so liveness analysis is
                 * correct */

                for (unsigned i = 0; i < 4; ++i) {
                        if (!(mask & (1 << i)))
                                ins.swizzle[0][i] = COMPONENT_X;
                }
        } else
                ins.dest = srcdest;

        return ins;
}

static inline bool
mir_has_arg(midgard_instruction *ins, unsigned arg)
{
        if (!ins)
                return false;

        mir_foreach_src(ins, i) {
                if (ins->src[i] == arg)
                        return true;
        }

        return false;
}

/* Scheduling */

void midgard_schedule_program(compiler_context *ctx);

void mir_ra(compiler_context *ctx);
void mir_squeeze_index(compiler_context *ctx);
void mir_lower_special_reads(compiler_context *ctx);
void mir_liveness_ins_update(uint16_t *live, midgard_instruction *ins, unsigned max);
void mir_compute_liveness(compiler_context *ctx);
void mir_invalidate_liveness(compiler_context *ctx);
bool mir_is_live_after(compiler_context *ctx, midgard_block *block, midgard_instruction *start, int src);

void mir_create_pipeline_registers(compiler_context *ctx);
void midgard_promote_uniforms(compiler_context *ctx);

void
midgard_emit_derivatives(compiler_context *ctx, nir_alu_instr *instr);

void
midgard_lower_derivatives(compiler_context *ctx, midgard_block *block);

bool mir_op_computes_derivatives(gl_shader_stage stage, unsigned op);

void mir_analyze_helper_terminate(compiler_context *ctx);
void mir_analyze_helper_requirements(compiler_context *ctx);

/* Final emission */

void emit_binary_bundle(
        compiler_context *ctx,
        midgard_block *block,
        midgard_bundle *bundle,
        struct util_dynarray *emission,
        int next_tag);

bool
nir_undef_to_zero(nir_shader *shader);
bool nir_fuse_io_16(nir_shader *shader);

void midgard_nir_lod_errata(nir_shader *shader);

unsigned midgard_get_first_tag_from_block(compiler_context *ctx, unsigned block_idx);

/* Optimizations */

bool midgard_opt_copy_prop(compiler_context *ctx, midgard_block *block);
bool midgard_opt_combine_projection(compiler_context *ctx, midgard_block *block);
bool midgard_opt_varying_projection(compiler_context *ctx, midgard_block *block);
bool midgard_opt_dead_code_eliminate(compiler_context *ctx);
bool midgard_opt_dead_move_eliminate(compiler_context *ctx, midgard_block *block);

#endif