2 // RMAC - Reboot's Macro Assembler for all Atari computers
3 // PROCLN.C - Line Processing
4 // Copyright (C) 199x Landon Dyer, 2011-2018 Reboot and Friends
5 // RMAC derived from MADMAC v1.07 Written by Landon Dyer, 1986
6 // Source utilised with the kind permission of Landon Dyer
23 #define DEF_KW // Declare keyword values
24 #include "kwtab.h" // Incl generated keyword tables & defs
26 #define DEF_MN // Incl 68k keyword definitions
27 #define DECL_MN // Incl 68k keyword state machine tables
34 #define DEF_MP // Include 6502 keyword definitions
35 #define DECL_MP // Include 6502 keyword state machine tables
38 #define DEF_MO // Include OP keyword definitions
39 #define DECL_MO // Include OP keyword state machine tables
42 IFENT * ifent; // Current ifent
43 static IFENT ifent0; // Root ifent
44 IFENT * f_ifent; // Freelist of ifents
45 int disabled; // Assembly conditionally disabled
46 int just_bss; // 1, ds.b in microprocessor mode
47 uint32_t pcloc; // Value of "PC" at beginning of line
48 SYM * lab_sym; // Label on line (or NULL)
50 const char extra_stuff[] = "extra (unexpected) text found after addressing mode";
51 const char comma_error[] = "missing comma";
52 const char syntax_error[] = "syntax error";
53 const char locgl_error[] = "cannot GLOBL local symbol";
54 const char lab_ignored[] = "label ignored";
56 // Table to convert an addressing-mode number to a bitmask.
57 LONG amsktab[0124] = {
58 M_DREG, M_DREG, M_DREG, M_DREG,
59 M_DREG, M_DREG, M_DREG, M_DREG,
61 M_AREG, M_AREG, M_AREG, M_AREG,
62 M_AREG, M_AREG, M_AREG, M_AREG,
64 M_AIND, M_AIND, M_AIND, M_AIND,
65 M_AIND, M_AIND, M_AIND, M_AIND,
67 M_APOSTINC, M_APOSTINC, M_APOSTINC, M_APOSTINC,
68 M_APOSTINC, M_APOSTINC, M_APOSTINC, M_APOSTINC,
70 M_APREDEC, M_APREDEC, M_APREDEC, M_APREDEC,
71 M_APREDEC, M_APREDEC, M_APREDEC, M_APREDEC,
73 M_ADISP, M_ADISP, M_ADISP, M_ADISP,
74 M_ADISP, M_ADISP, M_ADISP, M_ADISP,
76 M_AINDEXED, M_AINDEXED, M_AINDEXED, M_AINDEXED,
77 M_AINDEXED, M_AINDEXED, M_AINDEXED, M_AINDEXED,
110 // Function prototypes
111 int HandleLabel(char *, int);
115 // Initialize line processor
117 void InitLineProcessor(void)
121 f_ifent = ifent0.if_prev = NULL;
131 int state; // Keyword machine state (output)
132 int j; // Random int, must be fast
133 char * p; // Random char ptr, must be fast
134 TOKEN * tk; // First token in line
135 char * label; // Symbol (or NULL)
136 char * equate; // Symbol (or NULL)
137 int labtyp = 0; // Label type (':', DCOLON)
138 int equtyp = 0; // Equ type ('=', DEQUALS)
139 uint64_t eval; // Expression value
140 WORD eattr; // Expression attributes
141 SYM * esym; // External symbol involved in expr.
142 WORD siz = 0; // Size suffix to mnem/diretve/macro
143 LONG amsk0, amsk1; // Address-type masks for ea0, ea1
144 MNTAB * m; // Code generation table pointer
145 SYM * sy, * sy2; // Symbol (temp usage)
146 char * opname = NULL; // Name of dirctve/mnemonic/macro
147 int listflag; // 0: Don't call listeol()
148 WORD rmask; // Register list, for REG
149 int registerbank; // RISC register bank
150 int riscreg; // RISC register
151 listflag = 0; // Initialise listing flag
153 loop: // Line processing loop label
155 // Get another line of tokens
156 if (TokenizeLine() == TKEOF)
158 DEBUG { printf("Assemble: Found TKEOF flag...\n"); }
159 if (list_flag && listflag) // Flush last line of source
162 if (ifent->if_prev != NULL) // Check conditional token
163 error("hit EOF without finding matching .endif");
168 DEBUG { DumpTokenBuffer(); }
172 if (listflag && listing > 0)
173 listeol(); // Tell listing generator about EOL
175 lstout((char)(disabled ? '-' : lntag)); // Prepare new line for listing
176 listflag = 1; // OK to call `listeol' now
177 just_bss = 0; // Reset just_bss mode
180 state = -3; // No keyword (just EOL)
181 label = NULL; // No label
182 lab_sym = NULL; // No (exported) label
183 equate = NULL; // No equate
184 tk = tok; // Save first token in line
185 pcloc = (uint32_t)sloc; // Set beginning-of-line PC
187 loop1: // Internal line processing loop
189 if (*tok == EOL) // Restart loop if end-of-line
192 // First token MUST be a symbol (Shamus: not sure why :-/)
195 if ((*tok >= KW_D0) && (*tok <= KW_R31))
196 error("cannot use reserved keyword as label name or .equ");
198 error("syntax error; expected symbol");
203 j = (int)tok[2]; // Skip equates (normal statements)
205 if (j == '=' || j == DEQUALS || j == SET || j == REG || j == EQUREG || j == CCDEF)
207 equate = string[tok[1]];
213 // Skip past label (but record it)
214 if (j == ':' || j == DCOLON)
217 label = string[tok[1]]; // Get label name
218 labtyp = tok[2]; // Get label type
219 tok += 3; // Go to next line token
222 // Looks like another label follows the previous one, so handle
223 // the previous one until there aren't any more
224 if (as68_flag && (*tok == SYMBOL && tok[2] == ':'))
226 if (HandleLabel(label, labtyp) != 0)
233 // EOL is legal here...
237 // First token MUST be a symbol (if we get here, tok didn't advance)
238 if (*tok++ != SYMBOL)
240 error("syntax error; expected symbol");
244 opname = p = string[*tok++];
246 // Check to see if the SYMBOL is a keyword (a mnemonic or directive).
247 // On output, `state' will have one of the values:
248 // -3 there was no symbol (EOL)
249 // -2..-1 the symbol didn't match any keyword
250 // 0..499 vanilla directives (dc, ds, etc.)
251 // 500..999 electric directives (macro, rept, etc.)
252 // 1000..+ mnemonics (move, lsr, etc.)
253 for(state=0; state>=0;)
255 j = mnbase[state] + (int)tolowertab[*p];
257 // Reject, character doesn't match
258 if (mncheck[j] != state)
260 state = -1; // No match
264 // Must accept or reject at EOS
267 state = mnaccept[j]; // (-1 on no terminal match)
274 // Check for ".b" ".w" ".l" after directive, macro or mnemonic.
279 case DOTW: siz = SIZW, tok++; break;
280 case DOTL: siz = SIZL, tok++; break;
281 case DOTB: siz = SIZB, tok++; break;
282 case DOTD: siz = SIZD, tok++; break;
283 case DOTP: siz = SIZP, tok++; break;
284 case DOTQ: siz = SIZQ, tok++; break;
285 case DOTS: siz = SIZS, tok++; break;
286 case DOTX: siz = SIZX, tok++; break;
289 // Do special directives (500..999) (These must be handled in "real time")
290 if (state >= 500 && state < 1000)
306 case MN_IIF: // .iif --- immediate if
307 if (disabled || expr(exprbuf, &eval, &eattr, &esym) != OK)
310 if (!(eattr & DEFINED))
327 case MN_MACRO: // .macro --- macro definition
330 // Label on a macro definition is bad mojo... Warn the user
339 case MN_EXITM: // .exitm --- exit macro
340 case MN_ENDM: // .endm --- same as .exitm
354 // Handle labels on REPT directive lines...
357 if (HandleLabel(label, labtyp) != 0)
368 error("mis-nested .endr");
375 if (disabled) // Conditionally disabled code
381 // Pick global or local symbol enviroment
382 j = (*equate == '.' ? curenv : 0);
383 sy = lookup(equate, LABEL, j);
387 sy = NewSymbol(equate, LABEL, j);
390 if (equtyp == DEQUALS)
392 // Can't GLOBAL a local symbol
402 else if ((sy->sattr & DEFINED) && equtyp != SET)
404 if ((equtyp == EQUREG) && (sy->sattre & UNDEF_EQUR))
406 sy->sattre &= ~UNDEF_EQUR;
409 else if ((equtyp == CCDEF) && (sy->sattre & UNDEF_CC))
411 sy->sattre &= ~UNDEF_CC;
416 error("multiple equate to '%s'", sy->sname);
421 // Put symbol in "order of definition" list if it's not already there
422 AddToSymbolDeclarationList(sy);
424 // Parse value to equate symbol to;
428 if (equtyp == EQUREG)
430 //Linko's request to issue a warning on labels that equated to the same
431 //register would go here. Not sure how to implement it though. :-/
434 have an array of bools with 64 entries. Whenever a register is equated, set the
435 corresponding register bool to true. Whenever it's undef'ed, set it to false.
436 When checking to see if it's already been equated, issue a warning.
438 // Check that we are in a RISC section
441 error(".equr/.regequ must be defined in .gpu/.dsp section");
445 // Check for register to equate to
446 if ((*tok >= KW_R0) && (*tok <= KW_R31))
448 // sy->sattre = EQUATEDREG | RISCSYM; // Mark as equated register
449 sy->sattre = EQUATEDREG; // Mark as equated register
450 riscreg = (*tok - KW_R0);
451 //is there any reason to do this, since we're putting this in svalue?
452 //i'm thinking, no. Let's test that out! :-D
453 // sy->sattre |= (riscreg << 8); // Store register number
454 //everything seems to build fine without it... We'll leave it here Just In Case(tm)
456 #define DEBODGE_REGBANK
457 #ifdef DEBODGE_REGBANK
458 // Default is current state of "regbank"
459 registerbank = regbank;
461 // Default is no register bank specified
462 registerbank = BANK_N;
465 // Check for ",<bank #>" override notation
466 if ((tok[1] == ',') && (tok[2] == CONST))
468 // Advance token pointer to the constant
471 // Anything other than a 0 or a 1 will result in "No Bank"
472 if (*(uint64_t *)tok == 0)
473 registerbank = BANK_0;
474 else if (*(uint64_t *)tok == 1)
475 registerbank = BANK_1;
478 #ifdef DEBODGE_REGBANK
479 sy->sattre |= registerbank; // Store register bank
481 // What needs to happen here is to prime registerbank with regbank, then use
482 // registerbank down below for the bank marking.
483 #warning "!!! regbank <-> registerbank confusion here !!!"
484 // The question here is why, if we're allowed to override the ".regbankN" rules
485 // above, then why is it using the one set by the directive in the extended
486 // attributes and not in what ends up in symbol->svalue?
487 // ".regbankN" is not an original Madmac directive, so it's suspect
488 sy->sattre |= regbank; // Store register bank
490 eattr = ABS | DEFINED | GLOBAL;
494 // Checking for a register symbol
495 else if (tok[0] == SYMBOL)
497 sy2 = lookup(string[tok[1]], LABEL, j);
499 // Make sure symbol is a valid equreg
500 if (!sy2 || !(sy2->sattre & EQUATEDREG))
502 error("invalid GPU/DSP .equr/.regequ definition");
507 eattr = ABS | DEFINED | GLOBAL; // Copy symbols attributes
508 sy->sattre = sy2->sattre;
509 eval = (sy2->svalue & 0xFFFFF0FF);
515 error("invalid GPU/DSP .equr/.regequ definition");
519 else if (equtyp == REG)
521 if (reglist(&rmask) < 0)
524 eval = (uint32_t)rmask;
525 eattr = ABS | DEFINED;
527 else if (equtyp == CCDEF)
529 sy->sattre |= EQUATEDCC;
530 eattr = ABS | DEFINED | GLOBAL;
532 if (tok[0] == SYMBOL)
534 sy2 = lookup(string[tok[1]], LABEL, j);
536 if (!sy2 || !(sy2->sattre & EQUATEDCC))
538 error("invalid gpu/dsp .ccdef definition");
543 eattr = ABS | DEFINED | GLOBAL;
544 sy->sattre = sy2->sattre;
549 else if (expr(exprbuf, &eval, &eattr, &esym) != OK)
553 else if (*tok == SYMBOL)
555 sy2 = lookup(string[tok[1]], LABEL, j);
557 if (sy2 && (sy2->sattre & EQUATEDREG))
559 sy->stype = sy2->stype;
560 sy->sattr = sy2->sattr;
561 sy->sattre = sy2->sattre;
562 sy->svalue = sy2->svalue;
565 else if (expr(exprbuf, &eval, &eattr, &esym) != OK)
568 else if (expr(exprbuf, &eval, &eattr, &esym) != OK)
571 if (!(eattr & DEFINED))
577 sy->sattr |= eattr | EQUATED; // Symbol inherits value and attributes
580 if (list_flag) // Put value in listing
581 listvalue((uint32_t)eval);
583 at_eol(); // Must be at EOL now
590 // Non-zero == error occurred
591 if (HandleLabel(label, labtyp) != 0)
599 // If we're in 6502 mode and are still in need of a mnemonic, then search
600 // for valid 6502 mnemonic.
601 if (m6502 && (state < 0 || state >= 1000))
604 state = kmatch(opname, mpbase, mpcheck, mptab, mpaccept);
606 for(state=0, p=opname; state>= 0; )
608 j = mpbase[state] + tolowertab[*p];
610 if (mpcheck[j] != state) // Reject, character doesn't match
612 state = -1; // No match
617 { // Must accept or reject at EOS
618 state = mpaccept[j]; // (-1 on no terminal match)
626 // Call 6502 code generator if we found a mnemonic
629 m6502cg(state - 2000);
634 // If we are in GPU or DSP mode and still in need of a mnemonic then search
636 if ((rgpu || rdsp) && (state < 0 || state >= 1000))
638 for(state=0, p=opname; state>=0;)
640 j = mrbase[state] + (int)tolowertab[*p];
642 // Reject, character doesn't match
643 if (mrcheck[j] != state)
645 state = -1; // No match
649 // Must accept or reject at EOS
652 state = mraccept[j]; // (-1 on no terminal match)
659 // Call RISC code generator if we found a mnemonic
662 GenerateRISCCode(state);
667 // If we are in OP mode and still in need of a mnemonic then search for one
668 if (robjproc && ((state < 0) || (state >= 1000)))
670 for(state=0, p=opname; state>=0;)
672 j = mobase[state] + (int)tolowertab[*p];
674 // Reject, character doesn't match
675 if (mocheck[j] != state)
677 state = -1; // No match
681 // Must accept or reject at EOS
684 state = moaccept[j]; // (-1 on no terminal match)
691 // Call OP code generator if we found a mnemonic
694 GenerateOPCode(state);
699 // Invoke macro or complain about bad mnemonic
702 if ((sy = lookup(opname, MACRO, 0)) != NULL)
703 InvokeMacro(sy, siz);
705 error("unknown op '%s'", opname);
710 // Call directive handlers
713 (*dirtab[state])(siz);
718 // o can't deposit instrs in BSS or ABS
719 // o do automatic .EVEN for instrs
720 // o allocate space for largest possible instr
721 // o can't do ".b" operations with an address register
724 error("cannot initialize non-storage (BSS) section");
728 if (sloc & 1) // Automatic .even
731 if (challoc - ch_size < 18) // Make sure have space in current chunk
734 m = &machtab[state - 1000];
736 // Call special-mode handler
737 if (m->mnattr & CGSPECIAL)
739 (*m->mnfunc)(m->mninst, siz);
743 if (amode(1) < 0) // Parse 0, 1 or 2 addr modes
749 amsk0 = amsktab[am0];
750 amsk1 = amsktab[am1];
752 // Catch attempts to use ".B" with an address register (yes, this check
753 // does work at this level)
754 if (siz == SIZB && (am0 == AREG || am1 == AREG))
756 error("cannot use '.b' with an address register");
760 // Keep a backup of chptr (used for optimisations during codegen)
761 chptr_opcode = chptr;
765 if ((m->mnattr & siz) && (amsk0 & m->mn0) != 0 && (amsk1 & m->mn1) != 0)
767 (*m->mnfunc)(m->mninst, siz);
771 m = &machtab[m->mncont];
777 // Handle the creation of labels
779 int HandleLabel(char * label, int labelType)
781 // Check for dot in front of label; means this is a local label if present
782 int environment = (*label == '.' ? curenv : 0);
783 SYM * symbol = lookup(label, LABEL, environment);
787 symbol = NewSymbol(label, LABEL, environment);
791 else if (symbol->sattr & DEFINED)
792 return error("multiply-defined label '%s'", label);
794 // Put symbol in "order of definition" list if it's not already in it
795 AddToSymbolDeclarationList(symbol);
799 symbol->svalue = orgaddr;
800 symbol->sattr |= ABS | DEFINED | EQUATED;
804 symbol->svalue = sloc;
805 symbol->sattr |= DEFINED | cursect;
810 // Yes, our CS professors told us to write checks for equality this way,
811 // but damn, it hurts my brain every time I look at it. :-/
812 if (0 == environment)
815 // Make label global if it has a double colon
816 if (labelType == DCOLON)
818 if (environment != 0)
819 return error(locgl_error);
821 symbol->sattr |= GLOBAL;