mu - Soul of a tiny new machine. More thorough tests → More comprehensible and rewrite-friendly software → More resilient society.

	Commit message (Expand)	Author	Age	Files	Lines
*	6657	Kartik Agaram	2020-07-16	1	-18/+21
*	6465	Kartik Agaram	2020-06-04	1	-0/+1
*	6408	Kartik Agaram	2020-05-27	1	-1/+2
*	6405	Kartik Agaram	2020-05-25	1	-1/+1
*	6404	Kartik Agaram	2020-05-25	1	-10/+39
*	6214	Kartik Agaram	2020-04-27	1	-1/+1
*	6173	Kartik Agaram	2020-03-27	1	-5/+12
*	6170	Kartik Agaram	2020-03-25	1	-3/+3
*	6169	Kartik Agaram	2020-03-25	1	-1/+1
*	6164	Kartik Agaram	2020-03-24	1	-91/+91
*	6163	Kartik Agaram	2020-03-23	1	-0/+248

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01//EN" "http://www.w3.org/TR/html4/strict.dtd"> <html> <head> <meta http-equiv="content-type" content="text/html; charset=UTF-8"> <title>Mu - 036labels.cc</title> <meta name="Generator" content="Vim/8.1"> <meta name="plugin-version" content="vim8.1_v1"> <meta name="syntax" content="cpp"> <meta name="settings" content="number_lines,use_css,pre_wrap,no_foldcolumn,expand_tabs,line_ids,prevent_copy="> <meta name="colorscheme" content="minimal-light"> <style type="text/css">  </style> <script type='text/javascript'>  </script> </head> <body onload='JumpToLine();'> <a href='https://github.com/akkartik/mu/blob/master/036labels.cc'>https://github.com/akkartik/mu/blob/master/036labels.cc</a> <pre id='vimCodeElement'> 1 //: Labels are defined by ending names with a ':'. This layer will compute 2 //: displacements for labels, and compute the offset for instructions using them. 3 //: 4 //: We won't check this, but our convention will be that jump targets will 5 //: start with a '$', while functions will not. Function names will never be 6 //: jumped to, and jump targets will never be called. 7 8 //: We're introducing non-number names for the first time, so it's worth 9 //: laying down some ground rules all transforms will follow, so things don't 10 //: get too confusing: 11 //: - if it starts with a digit, it's treated as a number. If it can't be 12 //: parsed as hex it will raise an error. 13 //: - if it starts with '-' it's treated as a number. 14 //: - if it starts with '0x' it's treated as a number. 15 //: - if it's two characters long, it can't be a name. Either it's a hex 16 //: byte, or it raises an error. 17 //: That's it. Names can start with any non-digit that isn't a dash. They can 18 //: be a single character long. 'a' is not a hex number, it's a variable. 19 //: Later layers may add more conventions partitioning the space of names. But 20 //: the above rules will remain inviolate. 21 22 //: One special label is 'Entry', the address to start running the program at. 23 //: It can be non-unique; the last declaration overrides earlier ones. 24 //: It must exist in a program. Otherwise we don't know where to start running 25 //: programs. 26 27 void <a href='036labels.cc.html#L27'>test_Entry_label</a>() { 28 <a href='011run.cc.html#L82'>run</a>( 29 "== code 0x1\n" 30 "05 0x0d0c0b0a/imm32\n" 31 "Entry:\n" 32 "05 0x0d0c0b0a/imm32\n" 33 ); 34 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 35 "run: 0x00000006 opcode: 05\n" 36 ); 37 <a href='003trace.cc.html#L294'>CHECK_TRACE_DOESNT_CONTAIN</a>("run: 0x00000001 opcode: 05"); 38 } 39 40 :(before "End <a href='032operands.cc.html#L490'>looks_like_hex_int</a>(s) Detectors") 41 if (<a href='001help.cc.html#L170'>SIZE</a>(s) == 2) return true; 42 43 :(code) 44 void <a href='036labels.cc.html#L44'>test_pack_immediate_ignores_single_byte_nondigit_operand</a>() { 45 Hide_errors = true; 46 transform( 47 "== code 0x1\n" 48 "b9/copy a/imm32\n" 49 ); 50 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 51 "transform: packing instruction 'b9/copy a/imm32'\n" 52 // no change (we're just not printing metadata to the trace) 53 "transform: instruction after packing: 'b9 a'\n" 54 ); 55 } 56 57 void <a href='036labels.cc.html#L57'>test_pack_immediate_ignores_3_hex_digit_operand</a>() { 58 Hide_errors = true; 59 transform( 60 "== code 0x1\n" 61 "b9/copy aaa/imm32\n" 62 ); 63 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 64 "transform: packing instruction 'b9/copy aaa/imm32'\n" 65 // no change (we're just not printing metadata to the trace) 66 "transform: instruction after packing: 'b9 aaa'\n" 67 ); 68 } 69 70 void <a href='036labels.cc.html#L70'>test_pack_immediate_ignores_non_hex_operand</a>() { 71 Hide_errors = true; 72 transform( 73 "== code 0x1\n" 74 "b9/copy xxx/imm32\n" 75 ); 76 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 77 "transform: packing instruction 'b9/copy xxx/imm32'\n" 78 // no change (we're just not printing metadata to the trace) 79 "transform: instruction after packing: 'b9 xxx'\n" 80 ); 81 } 82 83 //: a helper we'll find handy later 84 void <a href='036labels.cc.html#L84'>check_valid_name</a>(const string& s) { 85 if (s.empty()) { 86 <a href='003trace.cc.html#L228'>raise</a> << "empty name!\n" << end(); 87 return; 88 } 89 if (s.at(0) == '-') 90 <a href='003trace.cc.html#L228'>raise</a> << "'" << s << "' starts with '-', which can be confused with a negative number; use a different name\n" << end(); 91 if (s.substr(0, 2) == "0x") { 92 <a href='003trace.cc.html#L228'>raise</a> << "'" << s << "' looks like a hex number; use a different name\n" << end(); 93 return; 94 } 95 if (isdigit(s.at(0))) 96 <a href='003trace.cc.html#L228'>raise</a> << "'" << s << "' starts with a digit, and so can be confused with a number; use a different name.\n" << end(); 97 if (<a href='001help.cc.html#L170'>SIZE</a>(s) == 2) 98 <a href='003trace.cc.html#L228'>raise</a> << "'" << s << "' is two characters long, which can look like raw hex bytes at a glance; use a different name\n" << end(); 99 } 100 101 //: Now that that's done, let's start using names as labels. 102 103 void <a href='036labels.cc.html#L103'>test_map_label</a>() { 104 transform( 105 "== code 0x1\n" 106 "loop:\n" 107 " 05 0x0d0c0b0a/imm32\n" 108 ); 109 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 110 "transform: label 'loop' is at address 1\n" 111 ); 112 } 113 114 :(before "End Transforms") 115 <a href='031transforms.cc.html#L4'>Transform</a>.push_back(<a href='036labels.cc.html#L117'>rewrite_labels</a>); 116 :(code) 117 void <a href='036labels.cc.html#L117'>rewrite_labels</a>(program& p) { 118 <a href='003trace.cc.html#L96'>trace</a>(3, "transform") << "-- rewrite labels" << end(); 119 if (p.<a href='011run.cc.html#L106'>segments</a>.empty()) return; 120 segment& code = *find(p, "code"); 121 map<string, int32_t> byte_index; // values are unsigned, but we're going to do subtractions on them so they need to fit in 31 bits 122 <a href='036labels.cc.html#L131'>compute_byte_indices_for_labels</a>(code, byte_index); 123 if (<a href='003trace.cc.html#L264'>trace_contains_errors</a>()) return; 124 <a href='036labels.cc.html#L205'>drop_labels</a>(code); 125 if (<a href='003trace.cc.html#L264'>trace_contains_errors</a>()) return; 126 <a href='036labels.cc.html#L217'>replace_labels_with_displacements</a>(code, byte_index); 127 if (<a href='001help.cc.html#L233'>contains_key</a>(byte_index, "Entry")) 128 p.<a href='011run.cc.html#L105'>entry</a> = code.start + get(byte_index, "Entry"); 129 } 130 131 void <a href='036labels.cc.html#L131'>compute_byte_indices_for_labels</a>(const segment& code, map<string, int32_t>& byte_index) { 132 int current_byte = 0; 133 for (int i = 0; i < <a href='001help.cc.html#L170'>SIZE</a>(code.<a href='011run.cc.html#L113'>lines</a>); ++i) { 134 const line& inst = code.<a href='011run.cc.html#L113'>lines</a>.at(i); 135 if (<a href='036labels.cc.html#L184'>Source_lines_file</a>.is_open() && !inst.original.empty() && /*not a label*/ *inst.<a href='011run.cc.html#L122'>words</a>.at(0).data.rbegin() != ':') 136 <a href='036labels.cc.html#L184'>Source_lines_file</a> << "0x" << <a href='010vm.cc.html#L395'>HEXWORD</a> << (code.start + current_byte) << ' ' << inst.original << '\n'; 137 for (int j = 0; j < <a href='001help.cc.html#L170'>SIZE</a>(inst.<a href='011run.cc.html#L122'>words</a>); ++j) { 138 const word& curr = inst.<a href='011run.cc.html#L122'>words</a>.at(j); 139 // hack: if we have any operand metadata left after previous transforms, 140 // deduce its size 141 // Maybe we should just move this transform to before instruction 142 // packing, and deduce the size of *all* operands. But then we'll also 143 // have to deal with bitfields. 144 if (has_operand_metadata(curr, "disp32") || has_operand_metadata(curr, "imm32")) { 145 if (*curr.data.rbegin() == ':') 146 <a href='003trace.cc.html#L228'>raise</a> << "'" << to_string(inst) << "': don't use ':' when jumping to labels\n" << end(); 147 current_byte += 4; 148 } 149 else if (has_operand_metadata(curr, "disp16")) { 150 if (*curr.data.rbegin() == ':') 151 <a href='003trace.cc.html#L228'>raise</a> << "'" << to_string(inst) << "': don't use ':' when jumping to labels\n" << end(); 152 current_byte += 2; 153 } 154 // automatically handle /disp8 and /imm8 here 155 else if (*curr.data.rbegin() != ':') { 156 ++current_byte; 157 } 158 else { 159 string label = <a href='036labels.cc.html#L263'>drop_last</a>(curr.data); 160 // ensure labels look sufficiently different from raw hex 161 <a href='036labels.cc.html#L84'>check_valid_name</a>(label); 162 if (<a href='003trace.cc.html#L264'>trace_contains_errors</a>()) return; 163 if (<a href='032operands.cc.html#L447'>contains_any_operand_metadata</a>(curr)) 164 <a href='003trace.cc.html#L228'>raise</a> << "'" << to_string(inst) << "': label definition (':') not allowed in operand\n" << end(); 165 if (j > 0) 166 <a href='003trace.cc.html#L228'>raise</a> << "'" << to_string(inst) << "': labels can only be the first <a href='011run.cc.html#L127'>word</a> in a line.\n" << end(); 167 if (<a href='036labels.cc.html#L183'>Labels_file</a>.is_open()) 168 <a href='036labels.cc.html#L183'>Labels_file</a> << "0x" << <a href='010vm.cc.html#L395'>HEXWORD</a> << (code.start + current_byte) << ' ' << label << '\n'; 169 if (<a href='001help.cc.html#L233'>contains_key</a>(byte_index, label) && label != "Entry") { 170 <a href='003trace.cc.html#L228'>raise</a> << "duplicate label '" << label << "'\n" << end(); 171 return; 172 } 173 <a href='001help.cc.html#L229'>put</a>(byte_index, label, current_byte); 174 <a href='003trace.cc.html#L96'>trace</a>(99, "transform") << "label '" << label << "' is at address " << (current_byte+code.start) << end(); 175 // no modifying current_byte; label definitions won't be in the final binary 176 } 177 } 178 } 179 } 180 181 :(before "End Globals") 182 bool <a href='036labels.cc.html#L182'>Dump_debug_info</a> = false; // currently used only by 'bootstrap translate' 183 ofstream <a href='036labels.cc.html#L183'>Labels_file</a>; 184 ofstream <a href='036labels.cc.html#L184'>Source_lines_file</a>; 185 :(before "End Commandline Options") 186 else if (<a href='001help.cc.html#L102'>is_equal</a>(*arg, "--debug")) { 187 <a href='036labels.cc.html#L182'>Dump_debug_info</a> = true; 188 // End --debug Settings 189 } 190 //: wait to open "labels" for writing until we're sure we aren't trying to read it 191 :(after "Begin bootstrap translate") 192 if (<a href='036labels.cc.html#L182'>Dump_debug_info</a>) { 193 cerr << "saving address->label information to 'labels'\n"; 194 <a href='036labels.cc.html#L183'>Labels_file</a>.open("labels"); 195 cerr << "saving address->source information to 'source_lines'\n"; 196 <a href='036labels.cc.html#L184'>Source_lines_file</a>.open("source_lines"); 197 } 198 :(before "End bootstrap translate") 199 if (<a href='036labels.cc.html#L182'>Dump_debug_info</a>) { 200 <a href='036labels.cc.html#L183'>Labels_file</a>.close(); 201 <a href='036labels.cc.html#L184'>Source_lines_file</a>.close(); 202 } 203 204 :(code) 205 void <a href='036labels.cc.html#L205'>drop_labels</a>(segment& code) { 206 for (int i = 0; i < <a href='001help.cc.html#L170'>SIZE</a>(code.<a href='011run.cc.html#L113'>lines</a>); ++i) { 207 line& inst = code.<a href='011run.cc.html#L113'>lines</a>.at(i); 208 vector<word>::iterator new_end = remove_if(inst.<a href='011run.cc.html#L122'>words</a>.begin(), inst.<a href='011run.cc.html#L122'>words</a>.end(), <a href='036labels.cc.html#L213'>is_label</a>); 209 inst.<a href='011run.cc.html#L122'>words</a>.erase(new_end, inst.<a href='011run.cc.html#L122'>words</a>.end()); 210 } 211 } 212 213 bool <a href='036labels.cc.html#L213'>is_label</a>(const word& w) { 214 return *w.data.rbegin() == ':'; 215 } 216 217 void <a href='036labels.cc.html#L217'>replace_labels_with_displacements</a>(segment& code, const map<string, int32_t>& byte_index) { 218 int32_t byte_index_next_instruction_starts_at = 0; 219 for (int i = 0; i < <a href='001help.cc.html#L170'>SIZE</a>(code.<a href='011run.cc.html#L113'>lines</a>); ++i) { 220 line& inst = code.<a href='011run.cc.html#L113'>lines</a>.at(i); 221 byte_index_next_instruction_starts_at += <a href='035compute_segment_address.cc.html#L56'>num_bytes</a>(inst); 222 <a href='011run.cc.html#L121'>line</a> new_inst; 223 for (int j = 0; j < <a href='001help.cc.html#L170'>SIZE</a>(inst.<a href='011run.cc.html#L122'>words</a>); ++j) { 224 const word& curr = inst.<a href='011run.cc.html#L122'>words</a>.at(j); 225 if (<a href='001help.cc.html#L233'>contains_key</a>(byte_index, curr.data)) { 226 int32_t displacement = static_cast<int32_t>(get(byte_index, curr.data)) - byte_index_next_instruction_starts_at; 227 if (has_operand_metadata(curr, "disp8")) { 228 if (displacement > 0x7f || displacement < -0x7f) 229 <a href='003trace.cc.html#L228'>raise</a> << "'" << to_string(inst) << "': label too far away for displacement " << std::hex << displacement << " to fit in 8 signed bits\n" << end(); 230 else 231 emit_hex_bytes(new_inst, displacement, 1); 232 } 233 else if (has_operand_metadata(curr, "disp16")) { 234 if (displacement > 0x7fff || displacement < -0x7fff) 235 <a href='003trace.cc.html#L228'>raise</a> << "'" << to_string(inst) << "': label too far away for displacement " << std::hex << displacement << " to fit in 16 signed bits\n" << end(); 236 else 237 emit_hex_bytes(new_inst, displacement, 2); 238 } 239 else if (has_operand_metadata(curr, "disp32")) { 240 emit_hex_bytes(new_inst, displacement, 4); 241 } else if (has_operand_metadata(curr, "imm32")) { 242 emit_hex_bytes(new_inst, code.start + get(byte_index, curr.data), 4); 243 } 244 } 245 else { 246 new_inst.<a href='011run.cc.html#L122'>words</a>.push_back(curr); 247 } 248 } 249 inst.<a href='011run.cc.html#L122'>words</a>.swap(new_inst.<a href='011run.cc.html#L122'>words</a>); 250 <a href='003trace.cc.html#L96'>trace</a>(99, "transform") << "instruction after transform: '" << data_to_string(inst) << "'" << end(); 251 } 252 } 253 254 string <a href='036labels.cc.html#L254'>data_to_string</a>(const line& inst) { 255 ostringstream out; 256 for (int i = 0; i < <a href='001help.cc.html#L170'>SIZE</a>(inst.<a href='011run.cc.html#L122'>words</a>); ++i) { 257 if (i > 0) out << ' '; 258 out << inst.<a href='011run.cc.html#L122'>words</a>.at(i).data; 259 } 260 return out.str(); 261 } 262 263 string <a href='036labels.cc.html#L263'>drop_last</a>(const string& s) { 264 return string(s.begin(), --s.end()); 265 } 266 267 //: Label definitions must be the first word on a line. No jumping inside 268 //: instructions. 269 //: They should also be the only word on a line. 270 //: However, you can absolutely have multiple labels map to the same address, 271 //: as long as they're on separate lines. 272 273 void <a href='036labels.cc.html#L273'>test_multiple_labels_at</a>() { 274 transform( 275 "== code 0x1\n" 276 // address 1 277 "loop:\n" 278 " $loop2:\n" 279 // address 1 (labels take up no space) 280 " 05 0x0d0c0b0a/imm32\n" 281 // address 6 282 " eb $loop2/disp8\n" 283 // address 8 284 " eb $loop3/disp8\n" 285 // address 0xa 286 " $loop3:\n" 287 ); 288 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 289 "transform: label 'loop' is at address 1\n" 290 "transform: label '$loop2' is at address 1\n" 291 "transform: label '$loop3' is at address a\n" 292 // first jump is to -7 293 "transform: instruction after transform: 'eb f9'\n" 294 // second jump is to 0 (fall through) 295 "transform: instruction after transform: 'eb 00'\n" 296 ); 297 } 298 299 void <a href='036labels.cc.html#L299'>test_loading_label_as_imm32</a>() { 300 transform( 301 "== code 0x1\n" 302 "label:\n" 303 " be/copy-to-ESI label/imm32\n" 304 ); 305 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 306 "transform: label 'label' is at address 1\n" 307 "transform: instruction after transform: 'be 01 00 00 00'\n" 308 ); 309 } 310 311 void <a href='036labels.cc.html#L311'>test_duplicate_label</a>() { 312 Hide_errors = true; 313 transform( 314 "== code 0x1\n" 315 "loop:\n" 316 "loop:\n" 317 " 05 0x0d0c0b0a/imm32\n" 318 ); 319 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 320 "error: duplicate label 'loop'\n" 321 ); 322 } 323 324 void <a href='036labels.cc.html#L324'>test_label_too_short</a>() { 325 Hide_errors = true; 326 transform( 327 "== code 0x1\n" 328 "xz:\n" 329 " 05 0x0d0c0b0a/imm32\n" 330 ); 331 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 332 "error: 'xz' is two characters long, which can look like raw hex bytes at a glance; use a different name\n" 333 ); 334 } 335 336 void <a href='036labels.cc.html#L336'>test_label_hex</a>() { 337 Hide_errors = true; 338 transform( 339 "== code 0x1\n" 340 "0xab:\n" 341 " 05 0x0d0c0b0a/imm32\n" 342 ); 343 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 344 "error: '0xab' looks like a hex number; use a different name\n" 345 ); 346 } 347 348 void <a href='036labels.cc.html#L348'>test_label_negative_hex</a>() { 349 Hide_errors = true; 350 transform( 351 "== code 0x1\n" 352 "-a:\n" 353 " 05 0x0d0c0b0a/imm32\n" 354 ); 355 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 356 "error: '-a' starts with '-', which can be confused with a negative number; use a different name\n" 357 ); 358 } 359 360 //: As said up top, the 'Entry' label is special. 361 //: It can be non-unique; the last declaration overrides earlier ones. 362 //: It must exist in a program. Otherwise we don't know where to start running 363 //: programs. 364 365 void <a href='036labels.cc.html#L365'>test_duplicate_Entry_label</a>() { 366 transform( 367 "== code 0x1\n" 368 "Entry:\n" 369 "Entry:\n" 370 " 05 0x0d0c0b0a/imm32\n" 371 ); 372 <a href='003trace.cc.html#L306'>CHECK_TRACE_DOESNT_CONTAIN_ERRORS</a>(); 373 } 374 375 // This test could do with some refactoring. 376 // We're duplicating the flow inside `bootstrap translate`, but without 377 // reading/writing files. 378 // We can't just use run(string) because most of our tests allow programs 379 // without 'Entry' labels, as a convenience. 380 void <a href='036labels.cc.html#L380'>test_programs_without_Entry_label</a>() { 381 Hide_errors = true; 382 program p; 383 istringstream in( 384 "== code 0x1\n" 385 "05 0x0d0c0b0a/imm32\n" 386 "05 0x0d0c0b0a/imm32\n" 387 ); 388 parse(in, p); 389 transform(p); 390 ostringstream dummy; 391 save_elf(p, dummy); 392 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 393 "error: no 'Entry' label found\n" 394 ); 395 } 396 397 //: now that we have labels, we need to adjust segment size computation to 398 //: ignore them. 399 400 void <a href='036labels.cc.html#L400'>test_segment_size_ignores_labels</a>() { 401 transform( 402 "== code 0x09000074\n" 403 " 05/add 0x0d0c0b0a/imm32\n" // 5 bytes 404 "foo:\n" // 0 bytes 405 "== data 0x0a000000\n" 406 "bar:\n" 407 " 00\n" 408 ); 409 <a href='003trace.cc.html#L292'>CHECK_TRACE_CONTENTS</a>( 410 "transform: segment 1 begins at address 0x0a000079\n" 411 ); 412 } 413 414 :(before "End size_of(<a href='011run.cc.html#L127'>word</a> w) Special-cases") 415 else if (<a href='036labels.cc.html#L213'>is_label</a>(w)) 416 return 0; </pre> </body> </html>