1 /* Copyright (c) 1980 Regents of the University of California */ 2 static char sccsid[] = "@(#)asmain.c 4.1 08/13/80"; 3 #include <stdio.h> 4 #include <ctype.h> 5 #include <sys/types.h> 6 #include <signal.h> 7 #include <a.out.h> 8 9 #include "as.h" 10 #include "assyms.h" 11 #include "asexpr.h" 12 #include "asscan.h" 13 14 #ifdef UNIX 15 #define unix_lang_name "VAX/UNIX Assembler V3.0" 16 #endif 17 18 #ifdef VMS 19 #define vms_lang_name "VAX/VMS C assembler V1.00" 20 #endif VMS 21 22 /* 23 * variables to manage reading the assembly source files 24 */ 25 char *dotsname; /*the current file name; managed by the parser*/ 26 int lineno; /*current line number; managed by the parser*/ 27 char *innames[32]; /*names of the files being assembled*/ 28 int ninfiles; /*how many interesting files there are*/ 29 /* 30 * Flags settable from the argv process argument list 31 */ 32 int silent = 0; /*don't complain about any errors*/ 33 int savelabels = 0; /*write the labels to the a.out file*/ 34 int d124 = 4; /*default allocate 4 bytes for unknown pointers*/ 35 int anyerrs = 0; /*no errors yet*/ 36 int orgwarn = 0; /*Bad origins*/ 37 int passno = 1; /* current pass*/ 38 39 #ifdef DEBUG 40 int debug = 0; 41 int toktrace = 0; 42 #endif 43 44 int useVM = /*put the temp file in virtual memory*/ 45 #ifdef VMS 46 1; /*VMS has virtual memory (duh)*/ 47 #endif VMS 48 #ifdef UNIX 49 0; 50 #endif 51 52 char *endcore; /*where to get more symbol space*/ 53 54 /* 55 * Managers of the a.out file. 56 */ 57 struct exec hdr; 58 #define MAGIC 0410 59 u_long tsize; /* total text size */ 60 u_long dsize; /* total data size */ 61 u_long datbase; /* base of the data segment */ 62 u_long trsize; /* total text relocation size */ 63 u_long drsize; /* total data relocation size */ 64 65 /* 66 * Information about the current segment is accumulated in 67 * usedot; the most important information stored is the 68 * accumulated size of each of the text and data segments 69 * 70 * dotp points to the correct usedot expression for the current segment 71 */ 72 struct exp usedot[NLOC+NLOC]; /* info about all segments */ 73 struct exp *dotp; /* data/text location pointer */ 74 /* 75 * The inter pass temporary file is opened and closed by stdio, but 76 * is written to using direct read/write, as the temporary file 77 * is composed of buffers exactly BUFSIZ long. 78 */ 79 FILE *tmpfil; /* interpass communication file */ 80 /* 81 * a.out is created during the second pass. 82 * It is opened by stdio, but is filled with the parallel 83 * block I/O library 84 */ 85 char *outfile = "a.out"; 86 FILE *a_out_file; 87 off_t a_out_off; /* cumulative offsets for segments */ 88 /* 89 * The logical files containing the assembled data for each of 90 * the text and data segments are 91 * managed by the parallel block I/O library. 92 * a.out is logically opened in many places at once to 93 * receive the assembled data from the various segments as 94 * it all trickles in, but is physically opened only once 95 * to minimize file overhead. 96 */ 97 BFILE *usefile[NLOC+NLOC]; /* text/data files */ 98 BFILE *txtfil; /* current text/data file */ 99 /* 100 * Relocation information is accumulated seperately for each 101 * segment. This is required by the old loader (from BTL), 102 * but not by the new loader (Bill Joy). 103 * 104 * However, the size of the relocation information can not be computed 105 * during or after the 1st pass because the ''absoluteness' of values 106 * is unknown until all locally declared symbols have been seen. 107 * Thus, the size of the relocation information is only 108 * known after the second pass is finished. 109 * This obviates the use of the block I/O 110 * library, which requires knowing the exact offsets in a.out. 111 * 112 * So, we save the relocation information internally (we don't 113 * go to internal files to minimize overhead). 114 * 115 * Empirically, we studied 259 files composing the system, 116 * two compilers and a compiler generator: (all of which have 117 * fairly large source files) 118 * 119 * Number of files = 259 120 * Number of non zero text reloc files: 233 121 * Number of non zero data reloc files: 53 122 * Average text relocation = 889 123 * Average data relocation = 346 124 * Number of files > BUFSIZ text relocation = 71 125 * Number of files > BUFSIZ data relocation = 6 126 * 127 * For compiled C code, there is usually one text segment and two 128 * data segments; we see that allocating our own buffers and 129 * doing our internal handling of relocation information will, 130 * on the average, not use more memory than taken up by the buffers 131 * allocated for doing file I/O in parallel to a number of file. 132 * 133 * If we are assembling with the -V option, we 134 * use the left over token buffers from the 2nd pass, 135 * otherwise, we create our own. 136 * 137 * When the 2nd pass is complete, closeoutrel flushes the token 138 * buffers out to a BFILE. 139 * 140 * The internals to relbufdesc are known only in assyms.c 141 * 142 * outrel constructs the relocation information. 143 * closeoutrel flushes the relocation information to relfil. 144 */ 145 struct relbufdesc *rusefile[NLOC+NLOC]; 146 struct relbufdesc *relfil; /* un concatnated relocation info */ 147 BFILE *relocfile; /* concatnated relocation info */ 148 /* 149 * Once the relocation information has been written, 150 * we can write out the symbol table using the Block I/O 151 * mechanisms, as we once again know the offsets into 152 * the a.out file. 153 * 154 * We use relfil to output the symbol table information. 155 */ 156 157 char *tmpdirprefix = 158 #ifdef UNIX 159 "/tmp/"; 160 #else VMS 161 "/usr/tmp/"; 162 #endif 163 164 #define TMP_SUFFIX "asXXXXXX" 165 char tmpn1[TNAMESIZE]; 166 167 int delexit(); 168 169 main(argc, argv) 170 int argc; 171 char **argv; 172 { 173 174 tmpn1[0] = 0; 175 endcore = (char *)sbrk(0); 176 177 argprocess(argc, argv); /* process argument lists */ 178 if (anyerrs) exit(1); 179 180 initialize(); 181 zeroorigins(); /* set origins to zero */ 182 zerolocals(); /* fix local label counters */ 183 184 i_pass1(); /* open temp files, etc */ 185 pass1(); /* first pass through .s files */ 186 testlocals(); /* check for undefined locals */ 187 if (anyerrs) delexit(); 188 189 pass1_5(); /* resolve jxxx */ 190 if (anyerrs) delexit(); 191 192 open_a_out(); /* open a.out */ 193 roundsegments(); /* round segments to FW */ 194 build_hdr(); /* build initial header, and output */ 195 196 i_pass2(); /* reopen temporary file, etc */ 197 pass2(); /* second pass through the virtual .s */ 198 if (anyerrs) delexit(); 199 200 fillsegments(); /* fill segments with 0 to FW */ 201 reloc_syms(); /* dump relocation and symbol table */ 202 203 delete(); /* remove tmp file */ 204 bflush(); /* close off block I/O view of a.out */ 205 fix_a_out(); /* add in text and data reloc counts */ 206 207 if (anyerrs == 0 && orgwarn) 208 yyerror("Caution: absolute origins.\n"); 209 exit(anyerrs != 0); 210 } /*end of UNIX main*/ 211 212 argprocess(argc, argv) 213 int argc; 214 char *argv[]; 215 { 216 register char *cp; 217 218 ninfiles = 0; 219 silent = 0; 220 #ifdef DEBUG 221 debug = 0; 222 #endif 223 dotsname = "<argv error>"; 224 while (argc > 1) { 225 if (argv[1][0] == '-'){ 226 cp = argv[1] + 1; 227 /* 228 * We can throw away single minus signs, so 229 * that make scripts for the PDP 11 assembler work 230 * on this assembler too 231 */ 232 while (*cp){ 233 switch(*cp++){ 234 default: 235 yyerror("Unknown flag: %c", *--cp); 236 cp++; 237 break; 238 case 'd': 239 d124 = *cp++ - '0'; 240 if ( (d124 != 1) && (d124 != 2) && 241 (d124 != 4)){ 242 yyerror("-d[124] only"); 243 exit(1); 244 } 245 break; 246 case 'o': 247 if (argc < 3){ 248 yyerror("-o what???"); 249 exit(1); 250 } 251 outfile = argv[2]; 252 bumpone: 253 argc -= 2; 254 argv += 2; 255 goto nextarg; 256 257 case 't': 258 if (argc < 3){ 259 yyerror("-t what???"); 260 exit(1); 261 } 262 tmpdirprefix = argv[2]; 263 goto bumpone; 264 265 case 'V': 266 useVM = 1; 267 break; 268 case 'W': 269 silent = 1; 270 break; 271 case 'L': 272 savelabels = 1; 273 break; 274 #ifdef DEBUG 275 case 'D': 276 debug = 1; 277 break; 278 case 'T': 279 toktrace = 1; 280 break; 281 #endif 282 } /*end of the switch*/ 283 } /*end of pulling out all arguments*/ 284 } /*end of a flag argument*/ 285 else { /*file name*/ 286 if (ninfiles > 32){ 287 yyerror("More than 32 file names"); 288 exit(3); 289 } 290 innames[ninfiles++] = argv[1]; 291 } 292 --argc; ++argv; 293 nextarg:; 294 } 295 } 296 297 initialize() 298 { 299 if (signal(SIGINT, SIG_IGN) != SIG_IGN) 300 signal(SIGINT, delexit); 301 /* 302 * Install symbols in the table 303 */ 304 symtabinit(); 305 syminstall(); 306 /* 307 * Build the expression parser accelerator token sets 308 */ 309 buildtokensets(); 310 } 311 312 zeroorigins() 313 { 314 register int locindex; 315 /* 316 * Mark usedot: the first NLOC slots are for named text segments, 317 * the next for named data segments. 318 */ 319 for (locindex = 0; locindex < NLOC; locindex++){ 320 usedot[locindex].xtype = XTEXT; 321 usedot[NLOC + locindex].xtype = XDATA; 322 usedot[locindex].xvalue = 0; 323 usedot[NLOC + locindex].xvalue = 0; 324 usedot[locindex].yvalue = 0; 325 usedot[NLOC + locindex].yvalue = 0; 326 } 327 } 328 329 zerolocals() 330 { 331 register int i; 332 333 for (i = 0; i <= 9; i++) { 334 lgensym[i] = 1; 335 genref[i] = 0; 336 } 337 } 338 339 i_pass1() 340 { 341 if (useVM == 0){ 342 strcat(tmpn1, tmpdirprefix); 343 strcat(tmpn1, TMP_SUFFIX); 344 mktemp(tmpn1); 345 tmpfil = fopen(tmpn1, "w"); 346 if (tmpfil==NULL) { 347 yyerror("Bad pass 1 temporary file for writing %s", tmpn1); 348 delexit(); 349 } 350 } 351 352 inittmpfile(); 353 } 354 355 pass1() 356 { 357 register int i; 358 359 passno = 1; 360 dotp = &usedot[0]; 361 txtfil = (BFILE *)0; 362 relfil = (struct relbufdesc *)0; 363 364 if (ninfiles == 0){ /*take the input from stdin directly*/ 365 lineno = 1; 366 dotsname = "<stdin>"; 367 368 yyparse(); 369 } else { /*we have the names tanked*/ 370 for (i = 0; i < ninfiles; i++){ 371 new_dot_s(innames[i]); 372 if (freopen(innames[i], "r", stdin) == NULL) { 373 yyerror( "Can't open source file %s\n", 374 innames[i]); 375 exit(2); 376 } 377 /* stdio is NOT used to read the input characters */ 378 /* we use read directly, into our own buffers */ 379 yyparse(); 380 } 381 } 382 383 closetmpfile(); /*kick out the last buffered intermediate text*/ 384 } 385 386 testlocals() 387 { 388 register int i; 389 for (i = 0; i <= 9; i++) { 390 if (genref[i]) 391 yyerror("Reference to undefined local label %df", i); 392 lgensym[i] = 1; 393 genref[i] = 0; 394 } 395 } 396 397 pass1_5() 398 { 399 sortsymtab(); 400 #ifdef DEBUG 401 if (debug) dumpsymtab(); 402 #endif 403 jxxxfix(); 404 #ifdef DEBUG 405 if (debug) dumpsymtab(); 406 #endif 407 } 408 409 open_a_out() 410 { 411 /* 412 * Open up the a.out file now, and get set to build 413 * up offsets into it for all of the various text,data 414 * text relocation and data relocation segments. 415 */ 416 a_out_file = fopen(outfile, "w"); 417 if (a_out_file == NULL) { 418 yyerror("Cannot create %s", outfile); 419 delexit(); 420 } 421 biofd = a_out_file->_file; 422 a_out_off = 0; 423 } 424 425 roundsegments() 426 { 427 register int locindex; 428 register long v; 429 /* 430 * round and assign text segment origins 431 * the exec header always goes in usefile[0] 432 */ 433 tsize = 0; 434 for (locindex=0; locindex<NLOC; locindex++) { 435 v = round(usedot[locindex].xvalue, FW); 436 usedot[locindex].xvalue = tsize; 437 if ((locindex == 0) || (v != 0) ){ 438 usefile[locindex] = (BFILE *)Calloc(1, sizeof(BFILE)); 439 bopen(usefile[locindex], a_out_off); 440 if (locindex == 0) 441 a_out_off = sizeof (struct exec); 442 } else { 443 usefile[locindex] = (BFILE *)-1; 444 } 445 tsize += v; 446 a_out_off += v; 447 } 448 /* 449 * Round and assign data segment origins. 450 */ 451 datbase = round(tsize, PAGRND); 452 for (locindex=0; locindex<NLOC; locindex++) { 453 v = round(usedot[NLOC+locindex].xvalue, FW); 454 usedot[NLOC+locindex].xvalue = datbase + dsize; 455 if (v != 0){ 456 usefile[NLOC + locindex] = (BFILE *)Calloc(1,sizeof(BFILE)); 457 bopen(usefile[NLOC + locindex], a_out_off); 458 } else { 459 usefile[NLOC + locindex] = (BFILE *)-1; 460 } 461 dsize += v; 462 a_out_off += v; 463 } 464 /* 465 * Assign final values to symbols 466 */ 467 hdr.a_bss = dsize; 468 freezesymtab(); /* this touches hdr.a_bss */ 469 stabfix(); 470 /* 471 * Set up the relocation information "files" to 472 * be zero; outrel takes care of the rest 473 */ 474 for (locindex = 0; locindex < NLOC + NLOC; locindex++){ 475 rusefile[locindex] = (struct relbufdesc *)0; 476 } 477 } 478 479 build_hdr() 480 { 481 /* 482 * Except for the text and data relocation sizes, 483 * calculate the final values for the header 484 * 485 * Write out the initial copy; we to come 486 * back later and patch up a_trsize and a_drsize, 487 * and overwrite this first version of the header. 488 */ 489 hdr.a_magic = MAGIC; 490 hdr.a_text = tsize; 491 hdr.a_data = dsize; 492 hdr.a_bss -= dsize; 493 hdr.a_syms = sizesymtab(); /* Does not include string pool length */ 494 hdr.a_entry = 0; 495 hdr.a_trsize = 0; 496 hdr.a_drsize = 0; 497 498 bwrite((char *)&hdr, sizeof(hdr), usefile[0]); 499 } 500 501 i_pass2() 502 { 503 if (useVM == 0) { 504 fclose(tmpfil); 505 tmpfil = fopen(tmpn1, "r"); 506 if (tmpfil==NULL) { 507 yyerror("Bad pass 2 temporary file for reading %s", tmpn1); 508 delexit(); 509 } 510 } 511 } 512 513 pass2() 514 { 515 #ifdef DEBUG 516 if (debug) 517 printf("\n\n\n\t\tPASS 2\n\n\n\n"); 518 #endif DEBUG 519 passno = 2; 520 lineno = 1; 521 dotp = &usedot[0]; 522 txtfil = usefile[0]; /* already opened (always!) */ 523 relfil = 0; /* outrel takes care of the rest */ 524 initoutrel(); 525 526 inittmpfile(); 527 528 yyparse(); 529 530 closetmpfile(); 531 } 532 533 fillsegments() 534 { 535 int locindex; 536 /* 537 * Round text and data segments to FW by appending zeros 538 */ 539 for (locindex = 0; locindex < NLOC + NLOC; locindex++) { 540 if (usefile[locindex]) { 541 txtfil = usefile[locindex]; 542 dotp = &usedot[locindex]; 543 while (usedot[locindex].xvalue & FW) 544 outb(0); 545 } 546 } 547 } 548 549 reloc_syms() 550 { 551 u_long closerelfil(); 552 /* 553 * Move the relocation information to a.out 554 * a_out_off is the offset so far: 555 * exec + text segments + data segments 556 */ 557 relocfile = (BFILE *)Calloc(1,sizeof(BFILE)); 558 bopen(relocfile, a_out_off); 559 a_out_off += closeoutrel(relocfile); 560 561 hdr.a_trsize = trsize; 562 hdr.a_drsize = drsize; 563 /* 564 * Output the symbol table 565 * and if FLEXNAMES is set, the string pool 566 */ 567 symwrite(relocfile); 568 } 569 570 fix_a_out() 571 { 572 if (lseek(a_out_file->_file, 0L, 0) < 0) 573 yyerror("Reposition for header rewrite fails"); 574 if (write(a_out_file->_file, (char *)&hdr, sizeof (struct exec)) < 0) 575 yyerror("Rewrite of header fails"); 576 } 577 578 delexit() 579 { 580 delete(); 581 if (passno == 2){ 582 unlink(outfile); 583 } 584 exit(1); 585 } 586 587 delete() 588 { 589 if (useVM == 0 || tmpn1[0]) 590 unlink(tmpn1); 591 } 592 593 sawabort() 594 { 595 char *fillinbuffer(); 596 while (fillinbuffer() != (char *)0) 597 continue; 598 delete(); 599 exit(1); /*although the previous pass will also exit non zero*/ 600 } 601 602 panic(fmt, a1, a2, a3, a4) 603 char *fmt; 604 /*VARARGS 1*/ 605 { 606 yyerror("Assembler panic: bad internal data structure."); 607 yyerror(fmt, a1, a2, a3, a4); 608 delete(); 609 abort(); 610 } 611