/* Copyright (c) 2018-2019 Alyssa Rosenzweig (alyssa@rosenzweig.io) * Copyright (C) 2019-2020 Collabora, Ltd. * * 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 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_HELPERS_H #define __MDG_HELPERS_H #include "util/macros.h" #include #include #define OP_IS_LOAD_VARY_F(op) (\ op == midgard_op_ld_vary_16 || \ op == midgard_op_ld_vary_32 \ ) #define OP_IS_PROJECTION(op) ( \ op == midgard_op_ldst_perspective_div_y || \ op == midgard_op_ldst_perspective_div_z || \ op == midgard_op_ldst_perspective_div_w \ ) #define OP_IS_VEC4_ONLY(op) ( \ OP_IS_PROJECTION(op) || \ op == midgard_op_ld_cubemap_coords \ ) #define OP_IS_MOVE(op) ( \ (op >= midgard_alu_op_fmov && op <= midgard_alu_op_fmov_rtp) || \ op == midgard_alu_op_imov \ ) #define OP_IS_UBO_READ(op) ( \ op >= midgard_op_ld_ubo_u8 && \ op <= midgard_op_ld_ubo_128_bswap8 \ ) #define OP_IS_CSEL_V(op) ( \ op == midgard_alu_op_icsel_v || \ op == midgard_alu_op_fcsel_v \ ) #define OP_IS_CSEL(op) ( \ OP_IS_CSEL_V(op) || \ op == midgard_alu_op_icsel || \ op == midgard_alu_op_fcsel \ ) #define OP_IS_UNSIGNED_CMP(op) ( \ op == midgard_alu_op_ult || \ op == midgard_alu_op_ule \ ) #define OP_IS_INTEGER_CMP(op) ( \ op == midgard_alu_op_ieq || \ op == midgard_alu_op_ine || \ op == midgard_alu_op_ilt || \ op == midgard_alu_op_ile || \ OP_IS_UNSIGNED_CMP(op) \ ) #define OP_IS_COMMON_STORE(op) ( \ op >= midgard_op_st_u8 && \ op <= midgard_op_st_128_bswap8 \ ) #define OP_IS_IMAGE(op) ( \ (op >= midgard_op_ld_image_32f && op <= midgard_op_ld_image_32i) || \ (op >= midgard_op_st_image_32f && op <= midgard_op_st_image_32i) || \ op == midgard_op_lea_image \ ) #define OP_IS_SPECIAL(op) ( \ (op >= midgard_op_ld_special_32f && op <= midgard_op_ld_special_32i) || \ (op >= midgard_op_st_special_32f && op <= midgard_op_st_special_32i) \ ) #define OP_IS_PACK_COLOUR(op) ( \ (op >= midgard_op_pack_colour_f32 && op <= midgard_op_pack_colour_s32) \ ) #define OP_IS_UNPACK_COLOUR(op) ( \ (op >= midgard_op_unpack_colour_f32 && op <= midgard_op_unpack_colour_s32) \ ) /* Instructions that are on the load/store unit but don't access memory */ #define OP_IS_REG2REG_LDST(op) ( \ op >= midgard_op_unpack_colour_f32 && \ op <= midgard_op_ldst_perspective_div_w \ ) /* ALU control words are single bit fields with a lot of space */ #define ALU_ENAB_VEC_MUL (1 << 17) #define ALU_ENAB_SCAL_ADD (1 << 19) #define ALU_ENAB_VEC_ADD (1 << 21) #define ALU_ENAB_SCAL_MUL (1 << 23) #define ALU_ENAB_VEC_LUT (1 << 25) #define ALU_ENAB_BR_COMPACT (1 << 26) #define ALU_ENAB_BRANCH (1 << 27) /* Other opcode properties that don't conflict with the ALU_ENABs, non-ISA */ /* Denotes an opcode that takes a vector input with a fixed-number of * channels, but outputs to only a single output channel, like dot products. * For these, to determine the effective mask, this quirk can be set. We have * an intentional off-by-one (a la MALI_POSITIVE), since 0-channel makes no * sense but we need to fit 4 channels in 2-bits. Similarly, 1-channel doesn't * make sense (since then why are we quirked?), so that corresponds to "no * count set" */ #define OP_CHANNEL_COUNT(c) ((c - 1) << 0) #define GET_CHANNEL_COUNT(c) ((c & (0x3 << 0)) ? ((c & (0x3 << 0)) + 1) : 0) /* For instructions that take a single argument, normally the first argument * slot is used for the argument and the second slot is a dummy #0 constant. * However, there are exceptions: instructions like fmov store their argument * in the _second_ slot and store a dummy r24 in the first slot, designated by * QUIRK_FLIPPED_R24 */ #define QUIRK_FLIPPED_R24 (1 << 2) /* Is the op commutative? */ #define OP_COMMUTES (1 << 3) /* Does the op convert types between int- and float- space (i2f/f2u/etc) */ #define OP_TYPE_CONVERT (1 << 4) /* Is this opcode the first in a f2x (rte, rtz, rtn, rtp) sequence? If so, * takes a roundmode argument in the IR. This has the semantic of rounding the * source (it's all fused in), which is why it doesn't necessarily make sense * for i2f (though folding there might be necessary for OpenCL reasons). Comes * up in format conversion, i.e. f2u_rte */ #define MIDGARD_ROUNDS (1 << 5) /* Vector-independant shorthands for the above; these numbers are arbitrary and * not from the ISA. Convert to the above with unit_enum_to_midgard */ #define UNIT_MUL 0 #define UNIT_ADD 1 #define UNIT_LUT 2 #define IS_ALU(tag) (tag >= TAG_ALU_4) /* Special register aliases */ #define MAX_WORK_REGISTERS 16 /* Uniforms are begin at (REGISTER_UNIFORMS - uniform_count) */ #define REGISTER_UNIFORMS 24 /* r24 and r25 are special registers that only exist during the pipeline, * by using them when we don't care about the register we skip a roundtrip * to the register file. */ #define REGISTER_UNUSED 24 #define REGISTER_CONSTANT 26 #define REGISTER_LDST_BASE 26 #define REGISTER_TEXTURE_BASE 28 #define REGISTER_SELECT 31 /* The following registers are read-only */ /* XY is Program Counter, ZW is Stack Pointer */ #define REGISTER_LDST_PC_SP 2 /* XY is Thread Local Storage pointer, ZW is Workgroup Local Storage pointer */ #define REGISTER_LDST_LOCAL_STORAGE_PTR 3 #define REGISTER_LDST_LOCAL_THREAD_ID 4 #define REGISTER_LDST_GROUP_ID 5 #define REGISTER_LDST_GLOBAL_THREAD_ID 6 /* This register is always zeroed when read. */ #define REGISTER_LDST_ZERO 7 /* SSA helper aliases to mimic the registers. */ #define SSA_FIXED_SHIFT 24 #define SSA_FIXED_REGISTER(reg) (((1 + (reg)) << SSA_FIXED_SHIFT) | 1) #define SSA_REG_FROM_FIXED(reg) ((((reg) & ~1) >> SSA_FIXED_SHIFT) - 1) #define SSA_FIXED_MINIMUM SSA_FIXED_REGISTER(0) #define COMPONENT_X 0x0 #define COMPONENT_Y 0x1 #define COMPONENT_Z 0x2 #define COMPONENT_W 0x3 #define SWIZZLE_IDENTITY { \ { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 }, \ { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 }, \ { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 }, \ { 0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15 } \ } #define SWIZZLE_IDENTITY_4 { \ { 0, 1, 2, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, \ { 0, 1, 2, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, \ { 0, 1, 2, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, \ { 0, 1, 2, 3, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 }, \ } static inline unsigned mask_of(unsigned nr_comp) { return (1 << nr_comp) - 1; } /* See ISA notes */ #define LDST_NOP (3) /* There are five ALU units: VMUL, VADD, SMUL, SADD, LUT. A given opcode is * implemented on some subset of these units (or occassionally all of them). * This table encodes a bit mask of valid units for each opcode, so the * scheduler can figure where to plonk the instruction. */ /* Shorthands for each unit */ #define UNIT_VMUL ALU_ENAB_VEC_MUL #define UNIT_SADD ALU_ENAB_SCAL_ADD #define UNIT_VADD ALU_ENAB_VEC_ADD #define UNIT_SMUL ALU_ENAB_SCAL_MUL #define UNIT_VLUT ALU_ENAB_VEC_LUT /* Shorthands for usual combinations of units */ #define UNITS_MUL (UNIT_VMUL | UNIT_SMUL) #define UNITS_ADD (UNIT_VADD | UNIT_SADD) #define UNITS_MOST (UNITS_MUL | UNITS_ADD) #define UNITS_ALL (UNITS_MOST | UNIT_VLUT) #define UNITS_SCALAR (UNIT_SADD | UNIT_SMUL) #define UNITS_VECTOR (UNIT_VMUL | UNIT_VADD) #define UNITS_ANY_VECTOR (UNITS_VECTOR | UNIT_VLUT) struct mir_op_props { const char *name; unsigned props; }; /* For load/store */ struct mir_ldst_op_props { const char *name; unsigned props; }; struct mir_tex_op_props { const char *name; unsigned props; }; struct mir_tag_props { const char *name; unsigned size; }; /* Lower 2-bits are a midgard_reg_mode */ #define GET_LDST_SIZE(c) (c & 3) /* Store (so the primary register is a source, not a destination */ #define LDST_STORE (1 << 2) /* Mask has special meaning and should not be manipulated directly */ #define LDST_SPECIAL_MASK (1 << 3) /* Non-store operation has side effects and should not be eliminated even if * its mask is 0 */ #define LDST_SIDE_FX (1 << 4) /* Computes an address according to indirects/zext/shift/etc */ #define LDST_ADDRESS (1 << 5) /* Some fields such swizzle and address have special meanings */ #define LDST_ATOMIC (1 << 6) /* Operates on attributes/varyings (including images) */ #define LDST_ATTRIB (1 << 7) /* This file is common, so don't define the tables themselves. #include * midgard_op.h if you need that, or edit midgard_ops.c directly */ /* Duplicate bits to convert a per-component to duplicated 8-bit format, * which is used for vector units */ static inline unsigned expand_writemask(unsigned mask, unsigned log2_channels) { unsigned o = 0; unsigned factor = 8 >> log2_channels; unsigned expanded = (1 << factor) - 1; for (unsigned i = 0; i < (1 << log2_channels); ++i) if (mask & (1 << i)) o |= (expanded << (factor * i)); return o; } /* Coerce structs to integer */ static inline unsigned vector_alu_srco_unsigned(midgard_vector_alu_src src) { unsigned u; memcpy(&u, &src, sizeof(src)); return u; } static inline midgard_vector_alu_src vector_alu_from_unsigned(unsigned u) { midgard_vector_alu_src s; memcpy(&s, &u, sizeof(s)); return s; } static inline void mir_compose_swizzle(unsigned *left, unsigned *right, unsigned *final_out) { unsigned out[16]; for (unsigned c = 0; c < 16; ++c) out[c] = right[left[c]]; memcpy(final_out, out, sizeof(out)); } /* Checks for an xyzw.. swizzle, given a mask */ static inline bool mir_is_simple_swizzle(unsigned *swizzle, unsigned mask) { for (unsigned i = 0; i < 16; ++i) { if (!(mask & (1 << i))) continue; if (swizzle[i] != i) return false; } return true; } /* Packs a load/store argument */ static inline uint8_t midgard_ldst_comp(unsigned reg, unsigned component, unsigned size) { assert((reg & ~1) == 0); assert(size == 16 || size == 32 || size == 64); /* Shift so everything is in terms of 32-bit units */ if (size == 64) { assert(component < 2); component <<= 1; } else if (size == 16) { assert((component & 1) == 0); component >>= 1; } return component; } /* Packs/unpacks a ubo index immediate */ void midgard_pack_ubo_index_imm(midgard_load_store_word *word, unsigned index); unsigned midgard_unpack_ubo_index_imm(midgard_load_store_word word); /* Packs/unpacks varying parameters. * FIXME: IMPORTANT: We currently handle varying mode weirdly, by passing all * parameters via an offset and using REGISTER_LDST_ZERO as base. This works * for most parameters, but does not allow us to encode/decode direct sample * position. */ void midgard_pack_varying_params(midgard_load_store_word *word, midgard_varying_params p); midgard_varying_params midgard_unpack_varying_params(midgard_load_store_word word); /* Load/store ops' displacement helpers. * This is useful because different types of load/store ops have different * displacement bitsize. */ #define UNPACK_LDST_ATTRIB_OFS(a) ((a) >> 9) #define UNPACK_LDST_VERTEX_OFS(a) util_sign_extend((a) & 0x1FF, 9) #define UNPACK_LDST_SELECTOR_OFS(a) ((a) >> 9) #define UNPACK_LDST_UBO_OFS(a) ((a) >> 2) #define UNPACK_LDST_MEM_OFS(a) ((a)) #define PACK_LDST_ATTRIB_OFS(a) ((a) << 9) #define PACK_LDST_VERTEX_OFS(a) ((a) & 0x1FF) #define PACK_LDST_SELECTOR_OFS(a) ((a) << 9) #define PACK_LDST_UBO_OFS(a) ((a) << 2) #define PACK_LDST_MEM_OFS(a) ((a)) static inline bool midgard_is_branch_unit(unsigned unit) { return (unit == ALU_ENAB_BRANCH) || (unit == ALU_ENAB_BR_COMPACT); } /* Packs ALU mod argument */ struct midgard_instruction; unsigned mir_pack_mod(struct midgard_instruction *ins, unsigned i, bool scalar); void mir_print_constant_component(FILE *fp, const midgard_constants *consts, unsigned c, midgard_reg_mode reg_mode, bool half, unsigned mod, midgard_alu_op op); #endif