| 1 | #!/usr/bin/env python3 |
| 2 | |
| 3 | ## |
| 4 | ## Copyright (c) 2024 Taylor Simpson <ltaylorsimpson@gmail.com> |
| 5 | ## |
| 6 | ## This program is free software; you can redistribute it and/or modify |
| 7 | ## it under the terms of the GNU General Public License as published by |
| 8 | ## the Free Software Foundation; either version 2 of the License, or |
| 9 | ## (at your option) any later version. |
| 10 | ## |
| 11 | ## This program is distributed in the hope that it will be useful, |
| 12 | ## but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 13 | ## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 14 | ## GNU General Public License for more details. |
| 15 | ## |
| 16 | ## You should have received a copy of the GNU General Public License |
| 17 | ## along with this program; if not, see <http://www.gnu.org/licenses/>. |
| 18 | ## |
| 19 | |
| 20 | import io |
| 21 | import re |
| 22 | |
| 23 | import sys |
| 24 | import textwrap |
| 25 | import iset |
| 26 | import hex_common |
| 27 | import argparse |
| 28 | |
| 29 | encs = { |
| 30 | tag: "".join(reversed(iset.iset[tag]["enc"].replace(" ", ""))) |
| 31 | for tag in iset.tags |
| 32 | if iset.iset[tag]["enc"] != "MISSING ENCODING" |
| 33 | } |
| 34 | |
| 35 | |
| 36 | regre = re.compile(r"((?<!DUP)[MNORCPQXSGVZA])([stuvwxyzdefg]+)([.]?[LlHh]?)(\d+S?)") |
| 37 | immre = re.compile(r"[#]([rRsSuUm])(\d+)(?:[:](\d+))?") |
| 38 | |
| 39 | |
| 40 | def ordered_unique(l): |
| 41 | return sorted(set(l), key=l.index) |
| 42 | |
| 43 | num_registers = {"R": 32, "V": 32} |
| 44 | |
| 45 | operand_letters = { |
| 46 | "P", |
| 47 | "i", |
| 48 | "I", |
| 49 | "r", |
| 50 | "s", |
| 51 | "t", |
| 52 | "u", |
| 53 | "v", |
| 54 | "w", |
| 55 | "x", |
| 56 | "y", |
| 57 | "z", |
| 58 | "d", |
| 59 | "e", |
| 60 | "f", |
| 61 | "g", |
| 62 | } |
| 63 | |
| 64 | # |
| 65 | # These instructions have unused operand letters in their encoding |
| 66 | # They don't correspond to actual operands in the instruction semantics |
| 67 | # We will mark them as ignored in QEMU decodetree |
| 68 | # |
| 69 | tags_with_unused_d_encoding = { |
| 70 | "R6_release_at_vi", |
| 71 | "R6_release_st_vi", |
| 72 | "S4_stored_rl_at_vi", |
| 73 | "S4_stored_rl_st_vi", |
| 74 | "S2_storew_rl_at_vi", |
| 75 | "S2_stored_rl_at_vi", |
| 76 | "S2_storew_rl_st_vi", |
| 77 | } |
| 78 | |
| 79 | tags_with_unused_t_encoding = { |
| 80 | "R6_release_at_vi", |
| 81 | "R6_release_st_vi", |
| 82 | } |
| 83 | |
| 84 | def skip_tag(tag, class_to_decode): |
| 85 | enc_class = iset.iset[tag]["enc_class"] |
| 86 | return enc_class != class_to_decode |
| 87 | |
| 88 | |
| 89 | ## |
| 90 | ## Generate the QEMU decodetree file for each instruction in class_to_decode |
| 91 | ## For A2_add: Rd32=add(Rs32,Rt32) |
| 92 | ## We produce: |
| 93 | ## %A2_add_Rd 0:5 |
| 94 | ## %A2_add_Rs 16:5 |
| 95 | ## %A2_add_Rt 8:5 |
| 96 | ## @A2_add 11110011000.......-.....---..... Rd=%A2_add_Rd Rs=%A2_add_Rs Rt=%A2_add_Rt %PP |
| 97 | ## A2_add ..................-.....---..... @A2_add |
| 98 | ## |
| 99 | def gen_decodetree_file(f, class_to_decode): |
| 100 | is_subinsn = class_to_decode.startswith("SUBINSN_") |
| 101 | f.write(f"## DO NOT MODIFY - This file is generated by {sys.argv[0]}\n\n") |
| 102 | if not is_subinsn: |
| 103 | f.write("%PP\t14:2\n\n") |
| 104 | for tag in sorted(encs.keys(), key=iset.tags.index): |
| 105 | if skip_tag(tag, class_to_decode): |
| 106 | continue |
| 107 | |
| 108 | enc = encs[tag] |
| 109 | enc_str = "".join(reversed(encs[tag])) |
| 110 | f.write(("#" * 80) + "\n" |
| 111 | f"## {tag}:\t{enc_str}\n" |
| 112 | "##\n") |
| 113 | |
| 114 | # The subinstructions come with a 13-bit encoding, but |
| 115 | # decodetree.py needs 16 bits |
| 116 | if is_subinsn: |
| 117 | enc_str = "---" + enc_str |
| 118 | |
| 119 | regs = ordered_unique(regre.findall(iset.iset[tag]["syntax"])) |
| 120 | imms = ordered_unique(immre.findall(iset.iset[tag]["syntax"])) |
| 121 | |
| 122 | # Write the field definitions for the registers |
| 123 | for regno, reg in enumerate(regs): |
| 124 | reg_type, reg_id, _, reg_enc_size = reg |
| 125 | reg_letter = reg_id[0] |
| 126 | reg_num_choices = int(reg_enc_size.rstrip("S")) |
| 127 | reg_mapping = reg_type + "".join("_" for letter in reg_id) + \ |
| 128 | reg_enc_size |
| 129 | reg_enc_fields = re.findall(reg_letter + "+", enc) |
| 130 | |
| 131 | # Check for some errors |
| 132 | if len(reg_enc_fields) == 0: |
| 133 | raise Exception(f"{tag} missing register field!") |
| 134 | if len(reg_enc_fields) > 1: |
| 135 | raise Exception(f"{tag} has split register field!") |
| 136 | reg_enc_field = reg_enc_fields[0] |
| 137 | if 2 ** len(reg_enc_field) != reg_num_choices: |
| 138 | raise Exception(f"{tag} has incorrect register field width!") |
| 139 | |
| 140 | f.write(f"%{tag}_{reg_type}{reg_id}\t" |
| 141 | f"{enc.index(reg_enc_field)}:{len(reg_enc_field)}") |
| 142 | |
| 143 | if (reg_type in num_registers and |
| 144 | reg_num_choices != num_registers[reg_type]): |
| 145 | f.write(f"\t!function=decode_mapped_reg_{reg_mapping}") |
| 146 | f.write("\n") |
| 147 | |
| 148 | # Write the field definitions for the immediates |
| 149 | for imm in imms: |
| 150 | immno = 1 if imm[0].isupper() else 0 |
| 151 | imm_type = imm[0] |
| 152 | imm_width = int(imm[1]) |
| 153 | imm_letter = "i" if imm_type.islower() else "I" |
| 154 | fields = [] |
| 155 | sign_mark = "s" if imm_type.lower() in "sr" else "" |
| 156 | for m in reversed(list(re.finditer(imm_letter + "+", enc))): |
| 157 | fields.append(f"{m.start()}:{sign_mark}{m.end() - m.start()}") |
| 158 | sign_mark = "" |
| 159 | field_str = " ".join(fields) |
| 160 | f.write(f"%{tag}_{imm_type}{imm_letter}\t{field_str}\n") |
| 161 | |
| 162 | ## Handle instructions with unused encoding letters |
| 163 | ## Change the unused letters to ignored |
| 164 | if tag in tags_with_unused_d_encoding: |
| 165 | enc_str = enc_str.replace("d", "-") |
| 166 | if tag in tags_with_unused_t_encoding: |
| 167 | enc_str = enc_str.replace("t", "-") |
| 168 | |
| 169 | # Replace the operand letters with . |
| 170 | for x in operand_letters: |
| 171 | enc_str = enc_str.replace(x, ".") |
| 172 | |
| 173 | # Write the instruction format |
| 174 | f.write(f"@{tag}\t{enc_str}") |
| 175 | for reg in regs: |
| 176 | reg_type = reg[0] |
| 177 | reg_id = reg[1] |
| 178 | f.write(f" {reg_type}{reg_id}=%{tag}_{reg_type}{reg_id}") |
| 179 | for imm in imms: |
| 180 | imm_type = imm[0] |
| 181 | imm_letter = "i" if imm_type.islower() else "I" |
| 182 | f.write(f" {imm_type}{imm_letter}=%{tag}_{imm_type}{imm_letter}") |
| 183 | |
| 184 | if not is_subinsn: |
| 185 | f.write(" %PP") |
| 186 | f.write("\n") |
| 187 | |
| 188 | # Replace the 0s and 1s with . |
| 189 | enc_str = enc_str.replace("0", ".").replace("1", ".") |
| 190 | |
| 191 | # Write the instruction pattern |
| 192 | f.write(f"{tag}\t{enc_str} @{tag}\n") |
| 193 | |
| 194 | |
| 195 | def main(): |
| 196 | parser = argparse.ArgumentParser( |
| 197 | description="Emit opaque macro calls with instruction semantics" |
| 198 | ) |
| 199 | parser.add_argument("semantics", help="semantics file") |
| 200 | parser.add_argument("class_to_decode", help="instruction class to decode") |
| 201 | parser.add_argument("out", help="output file") |
| 202 | args = parser.parse_args() |
| 203 | |
| 204 | hex_common.read_semantics_file(args.semantics) |
| 205 | with open(args.out, "w") as f: |
| 206 | gen_decodetree_file(f, args.class_to_decode) |
| 207 | |
| 208 | if __name__ == "__main__": |
| 209 | main() |