X-Git-Url: http://shamusworld.gotdns.org/cgi-bin/gitweb.cgi?p=rmac;a=blobdiff_plain;f=token.c;h=6a47bdb3fd04daab8e8c17b8364e83cc88712cfb;hp=4ec487204ea74a6eec2ea0b8a4553e82b4458aaf;hb=4205233c8397c581b4d27ab36ab81ec896ef3dd0;hpb=26019087571ebcafae571c7d32f485ceb8af8c5d diff --git a/token.c b/token.c index 4ec4872..6a47bdb 100644 --- a/token.c +++ b/token.c @@ -1,12 +1,14 @@ // -// RMAC - Reboot's Macro Assembler for the Atari Jaguar Console System +// RMAC - Reboot's Macro Assembler for all Atari computers // TOKEN.C - Token Handling -// Copyright (C) 199x Landon Dyer, 2011-2017 Reboot and Friends +// Copyright (C) 199x Landon Dyer, 2011-2020 Reboot and Friends // RMAC derived from MADMAC v1.07 Written by Landon Dyer, 1986 // Source utilised with the kind permission of Landon Dyer // #include "token.h" + +#include #include "direct.h" #include "error.h" #include "macro.h" @@ -35,7 +37,8 @@ WORD cfileno; // Current file number TOKEN * tok; // Ptr to current token TOKEN * etok; // Ptr past last token in tokbuf[] TOKEN tokeol[1] = {EOL}; // Bailout end-of-line token -char * string[TOKBUFSIZE*2]; // Token buffer string pointer storage +char * string[TOKBUFSIZE*2];// Token buffer string pointer storage +int optimizeOff; // Optimization override flag // File record, used to maintain a list of every include file ever visited #define FILEREC struct _filerec @@ -48,14 +51,14 @@ FILEREC FILEREC * filerec; FILEREC * last_fr; -INOBJ * cur_inobj; // Ptr current input obj (IFILE/IMACRO) -static INOBJ * f_inobj; // Ptr list of free INOBJs -static IFILE * f_ifile; // Ptr list of free IFILEs -static IMACRO * f_imacro; // Ptr list of free IMACROs +INOBJ * cur_inobj; // Ptr current input obj (IFILE/IMACRO) +static INOBJ * f_inobj; // Ptr list of free INOBJs +static IFILE * f_ifile; // Ptr list of free IFILEs +static IMACRO * f_imacro; // Ptr list of free IMACROs -static TOKEN tokbuf[TOKBUFSIZE]; // Token buffer (stack-like, all files) +static TOKEN tokbuf[TOKBUFSIZE]; // Token buffer (stack-like, all files) -char chrtab[] = { +uint8_t chrtab[0x100] = { ILLEG, ILLEG, ILLEG, ILLEG, // NUL SOH STX ETX ILLEG, ILLEG, ILLEG, ILLEG, // EOT ENQ ACK BEL ILLEG, WHITE, ILLEG, ILLEG, // BS HT LF VT @@ -79,36 +82,76 @@ char chrtab[] = { MULTX, MULTX, // : ; MULTX, MULTX, MULTX, STSYM+CTSYM, // < = > ? - MULTX, STSYM+CTSYM+HDIGIT, // @ A - (char)((BYTE)DOT)+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // B C - STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // D E - STSYM+CTSYM+HDIGIT, STSYM+CTSYM, // F G - STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // H I J K - (char)((BYTE)DOT)+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // L M N O - - STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM, // P Q R S - STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM, // T U V W - STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF, // X Y Z [ - SELF, SELF, MULTX, STSYM+CTSYM, // \ ] ^ _ - - ILLEG, STSYM+CTSYM+HDIGIT, // ` a - (char)((BYTE)DOT)+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // b c - STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // d e - STSYM+CTSYM+HDIGIT, STSYM+CTSYM, // f g - STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // h i j k - (char)((BYTE)DOT)+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // l m n o - - STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM, // p q r s - STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, (char)((BYTE)DOT)+STSYM+CTSYM, // t u v w - STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF, // x y z { - SELF, SELF, SELF, ILLEG // | } ~ DEL + MULTX, STSYM+CTSYM+HDIGIT, // @ A + DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // B C + DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // D E + STSYM+CTSYM+HDIGIT, STSYM+CTSYM, // F G + STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // H I J K + DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // L M N O + + DOT+STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // P Q R S + STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // T U V W + STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF,// X Y Z [ + SELF, SELF, MULTX, STSYM+CTSYM, // \ ] ^ _ + + ILLEG, STSYM+CTSYM+HDIGIT, // ` a + DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // b c + DOT+STSYM+CTSYM+HDIGIT, STSYM+CTSYM+HDIGIT, // d e + STSYM+CTSYM+HDIGIT, STSYM+CTSYM, // f g + STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // h i j k + DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, // l m n o + + DOT+STSYM+CTSYM, DOT+STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // p q r s + STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, DOT+STSYM+CTSYM, // t u v w + DOT+STSYM+CTSYM, STSYM+CTSYM, STSYM+CTSYM, SELF, // x y z { + SELF, SELF, SELF, ILLEG, // | } ~ DEL + + // Anything above $7F is illegal (and yes, we need to check for this, + // otherwise you get strange and spurious errors that will lead you astray) + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, + ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG, ILLEG }; // Names of registers static char * regname[] = { - "d0", "d1", "d2", "d3", "d4", "d5", "d6", "d7", - "a0", "a1", "a2", "a3", "a4", "a5", "a6", "a7", - "pc", "ssp", "usp", "sr", "ccr" + "d0","d1","d2","d3","d4","d5","d6","d7", // 128,135 + "a0","a1","a2","a3","a4","a5","a6","sp", // 136,143 + "ssp","pc","sr","ccr","regequ","set","reg","r0", // 144,151 + "r1","r2","r3","r4","r5","r6","r7","r8", // 152,159 + "r9","r10","r11","r12","r13","r14","r15","r16", // 160,167 + "r17","r18","r19","r20","r21","r22","r23","r24", // 168,175 + "r25","r26","r27","r28","r29","r30","r31","ccdef", // 176,183 + "usp","ic40","dc40","bc40","sfc","dfc","","vbr", // 184,191 + "cacr","caar","msp","isp","tc","itt0","itt1","dtt0", // 192,199 + "dtt1","mmusr","urp","srp","iacr0","iacr1","dacr0","dacr1", // 200,207 + "tt0","tt1","crp","","","","","", // 208,215 + "","","","","fpiar","fpsr","fpcr","", // 216,223 + "fp0","fp1","fp2","fp3","fp4","fp5","fp6","fp7", // 224,231 + "","","","","","","","", // 232,239 + "","","","","","","","", // 240,247 + "","","","","","","","", // 248,255 + "","","","","x0","x1","y0","y1", // 256,263 + "","b0","","b2","","b1","a","b", // 264,271 + "mr","omr","la","lc","ssh","ssl","ss","", // 272,279 + "n0","n1","n2","n3","n4","n5","n6","n7", // 280,287 + "m0","m1","m2","m3","m4","m5","m6","m7", // 288,295 + "","","","","","","l","p", // 296,303 + "mr","omr","la","lc","ssh","ssl","ss","", // 304,311 + "a10","b10","x","y","","","ab","ba" // 312,319 }; static char * riscregname[] = { @@ -159,14 +202,24 @@ void InitTokenizer(void) // These characters are legal immediately after a period dotxtab['b'] = DOTB; // .b .B .s .S dotxtab['B'] = DOTB; - dotxtab['s'] = DOTB; - dotxtab['S'] = DOTB; + //dotxtab['s'] = DOTB; + //dotxtab['S'] = DOTB; dotxtab['w'] = DOTW; // .w .W dotxtab['W'] = DOTW; dotxtab['l'] = DOTL; // .l .L dotxtab['L'] = DOTL; - dotxtab['i'] = DOTI; // .i .I (???) + dotxtab['i'] = DOTI; // .i .I (WTF is this???) dotxtab['I'] = DOTI; + dotxtab['D'] = DOTD; // .d .D (double) + dotxtab['d'] = DOTD; + dotxtab['S'] = DOTS; // .s .S + dotxtab['s'] = DOTS; + dotxtab['Q'] = DOTQ; // .q .Q (quad word) + dotxtab['q'] = DOTQ; + dotxtab['X'] = DOTX; // .x .x + dotxtab['x'] = DOTX; + dotxtab['P'] = DOTP; // .p .P + dotxtab['p'] = DOTP; } @@ -232,6 +285,7 @@ INOBJ * a_inobj(int typ) inobj->inobj.ifile = ifile; break; + case SRC_IMACRO: // Alloc and init an IMACRO if (f_imacro == NULL) imacro = malloc(sizeof(IMACRO)); @@ -243,9 +297,10 @@ INOBJ * a_inobj(int typ) inobj->inobj.imacro = imacro; break; + case SRC_IREPT: // Alloc and init an IREPT inobj->inobj.irept = malloc(sizeof(IREPT)); - DEBUG printf("alloc IREPT\n"); + DEBUG { printf("alloc IREPT\n"); } break; } @@ -337,7 +392,7 @@ int ExpandMacro(char * src, char * dest, int destsiz) *dst++ = *s++; continue; case '?': // \? set `questmark' flag - ++s; + s++; questmark = 1; break; case '#': // \#, number of arguments @@ -403,7 +458,7 @@ copy_d: *d++ = *s++; if (*s != '}') - return error("missing '}'"); + return error("missing closing brace ('}')"); else s++; } @@ -412,10 +467,10 @@ copy_d: // Lookup the argument and copy its (string) value into the // destination string - DEBUG printf("argument='%s'\n", mname); + DEBUG { printf("argument='%s'\n", mname); } if ((arg = lookup(mname, MACARG, macnum)) == NULL) - return errors("undefined argument: '%s'", mname); + return error("undefined argument: '%s'", mname); else { // Convert a string of tokens (terminated with EOL) back into @@ -423,24 +478,18 @@ copy_d: // macro invocation) then it is ignored. i = (int)arg->svalue; arg_num: - DEBUG printf("~argnumber=%d (argBase=%u)\n", i, imacro->argBase); + DEBUG { printf("~argnumber=%d\n", i); } tk = NULL; if (i < imacro->im_nargs) { -#if 0 -// tk = argp[i]; -// tk = argPtrs[i]; - tk = argPtrs[imacro->argBase + i]; -#else tk = imacro->argument[i].token; symbolString = imacro->argument[i].string; //DEBUG //{ // printf("ExM: Preparing to parse argument #%u...\n", i); -// dumptok(tk); +// DumpTokens(tk); //} -#endif } // \?arg yields: @@ -482,22 +531,12 @@ arg_num: switch ((int)*tk++) { case SYMBOL: -#if 0 -// d = (char *)*tk++; - d = string[*tk++]; -#else - // This fix should be done for strings too d = symbolString[*tk++]; -DEBUG printf("ExM: SYMBOL=\"%s\"", d); -#endif +DEBUG { printf("ExM: SYMBOL=\"%s\"", d); } break; case STRING: -#if 0 -// d = (char *)*tk++; - d = string[*tk++]; -#else d = symbolString[*tk++]; -#endif + if (dst >= edst) goto overflow; @@ -521,7 +560,9 @@ DEBUG printf("ExM: SYMBOL=\"%s\"", d); // to choke on legitimate code... Need to investigate this further // before changing anything else here! case CONST: - sprintf(numbuf, "$%lx", (long unsigned int)*tk++); +// sprintf(numbuf, "$%lx", (uint64_t)*tk++); + sprintf(numbuf, "$%" PRIX64, (uint64_t)*tk++); + tk++; d = numbuf; break; case DEQUALS: @@ -618,7 +659,7 @@ skipcomments: overflow: *dst = EOS; - DEBUG printf("*** OVERFLOW LINE ***\n%s\n", dest); + DEBUG { printf("*** OVERFLOW LINE ***\n%s\n", dest); } return fatal("line too long as a result of macro expansion"); } @@ -630,7 +671,7 @@ char * GetNextMacroLine(void) { IMACRO * imacro = cur_inobj->inobj.imacro; // LONG * strp = imacro->im_nextln; - struct LineList * strp = imacro->im_nextln; + LLIST * strp = imacro->im_nextln; if (strp == NULL) // End-of-macro return NULL; @@ -649,26 +690,34 @@ char * GetNextMacroLine(void) char * GetNextRepeatLine(void) { IREPT * irept = cur_inobj->inobj.irept; - LONG * strp = irept->ir_nextln; // initial null +// LONG * strp = irept->ir_nextln; // initial null // Do repeat at end of .rept block's string list - if (strp == NULL) +// if (strp == NULL) + if (irept->ir_nextln == NULL) { - DEBUG printf("back-to-top-of-repeat-block count=%d\n", (int)irept->ir_count); + DEBUG { printf("back-to-top-of-repeat-block count=%d\n", (int)irept->ir_count); } irept->ir_nextln = irept->ir_firstln; // copy first line if (irept->ir_count-- == 0) { - DEBUG printf("end-repeat-block\n"); + DEBUG { printf("end-repeat-block\n"); } return NULL; } - strp = irept->ir_nextln; +// strp = irept->ir_nextln; } - - strcpy(irbuf, (char *)(irept->ir_nextln + 1)); - DEBUG printf("repeat line='%s'\n", irbuf); - irept->ir_nextln = (LONG *)*strp; + // Mark the current macro line in the irept object + // This is probably overkill - a global variable + // would suffice here (it only gets used during + // error reporting anyway) + irept->lineno = irept->ir_nextln->lineno; + +// strcpy(irbuf, (char *)(irept->ir_nextln + 1)); + strcpy(irbuf, irept->ir_nextln->line); + DEBUG { printf("repeat line='%s'\n", irbuf); } +// irept->ir_nextln = (LONG *)*strp; + irept->ir_nextln = irept->ir_nextln->next; return irbuf; } @@ -680,23 +729,22 @@ char * GetNextRepeatLine(void) int include(int handle, char * fname) { // Debug mode - if (debug) - printf("[include: %s, cfileno=%u]\n", fname, cfileno); + DEBUG { printf("[include: %s, cfileno=%u]\n", fname, cfileno); } // Alloc and initialize include-descriptors INOBJ * inobj = a_inobj(SRC_IFILE); IFILE * ifile = inobj->inobj.ifile; - ifile->ifhandle = handle; // Setup file handle - ifile->ifind = ifile->ifcnt = 0; // Setup buffer indices - ifile->ifoldlineno = curlineno; // Save old line number - ifile->ifoldfname = curfname; // Save old filename - ifile->ifno = cfileno; // Save old file number + ifile->ifhandle = handle; // Setup file handle + ifile->ifind = ifile->ifcnt = 0; // Setup buffer indices + ifile->ifoldlineno = curlineno; // Save old line number + ifile->ifoldfname = curfname; // Save old filename + ifile->ifno = cfileno; // Save old file number // NB: This *must* be preincrement, we're adding one to the filecount here! - cfileno = ++filecount; // Compute NEW file number - curfname = strdup(fname); // Set current filename (alloc storage) - curlineno = 0; // Start on line zero + cfileno = ++filecount; // Compute NEW file number + curfname = strdup(fname); // Set current filename (alloc storage) + curlineno = 0; // Start on line zero // Add another file to the file-record FILEREC * fr = (FILEREC *)malloc(sizeof(FILEREC)); @@ -704,12 +752,12 @@ int include(int handle, char * fname) fr->frec_name = curfname; if (last_fr == NULL) - filerec = fr; // Add first filerec + filerec = fr; // Add first filerec else - last_fr->frec_next = fr; // Append to list of filerecs + last_fr->frec_next = fr; // Append to list of filerecs last_fr = fr; - DEBUG printf("[include: curfname: %s, cfileno=%u]\n", curfname, cfileno); + DEBUG { printf("[include: curfname: %s, cfileno=%u]\n", curfname, cfileno); } return OK; } @@ -720,74 +768,79 @@ int include(int handle, char * fname) // int fpop(void) { - IFILE * ifile; - IMACRO * imacro; - LONG * p, * p1; INOBJ * inobj = cur_inobj; - if (inobj != NULL) + if (inobj == NULL) + return 0; + + // Pop IFENT levels until we reach the conditional assembly context we + // were at when the input object was entered. + int numUnmatched = 0; + + while (ifent != inobj->in_ifent) { - // Pop IFENT levels until we reach the conditional assembly context we - // were at when the input object was entered. - int numUnmatched = 0; + if (d_endif() != 0) // Something bad happened during endif parsing? + return -1; // If yes, bail instead of getting stuck in a loop - while (ifent != inobj->in_ifent) - { - if (d_endif() != 0) // Something bad happened during endif parsing? - return -1; // If yes, bail instead of getting stuck in a loop + numUnmatched++; + } - numUnmatched++; - } + // Give a warning to the user that we had to wipe their bum for them + if (numUnmatched > 0) + warn("missing %d .endif(s)", numUnmatched); - // Give a warning to the user that we had to wipe their bum for them - if (numUnmatched > 0) - warni("missing %d .endif(s)", numUnmatched); + tok = inobj->in_otok; // Restore tok and otok + etok = inobj->in_etok; - tok = inobj->in_otok; // Restore tok and otok - etok = inobj->in_etok; + switch (inobj->in_type) + { + case SRC_IFILE: // Pop and release an IFILE + { + DEBUG { printf("[Leaving: %s]\n", curfname); } + + IFILE * ifile = inobj->inobj.ifile; + ifile->if_link = f_ifile; + f_ifile = ifile; + close(ifile->ifhandle); // Close source file +DEBUG { printf("[fpop (pre): curfname=%s]\n", curfname); } + curfname = ifile->ifoldfname; // Set current filename +DEBUG { printf("[fpop (post): curfname=%s]\n", curfname); } +DEBUG { printf("[fpop: (pre) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno); } + curlineno = ifile->ifoldlineno; // Set current line# + DEBUG { printf("cfileno=%d ifile->ifno=%d\n", (int)cfileno, (int)ifile->ifno); } + cfileno = ifile->ifno; // Restore current file number +DEBUG { printf("[fpop: (post) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno); } + break; + } - switch (inobj->in_type) - { - case SRC_IFILE: // Pop and release an IFILE - if (debug) - printf("[Leaving: %s]\n", curfname); - - ifile = inobj->inobj.ifile; - ifile->if_link = f_ifile; - f_ifile = ifile; - close(ifile->ifhandle); // Close source file -if (debug) printf("[fpop (pre): curfname=%s]\n", curfname); - curfname = ifile->ifoldfname; // Set current filename -if (debug) printf("[fpop (post): curfname=%s]\n", curfname); -if (debug) printf("[fpop: (pre) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno); - curlineno = ifile->ifoldlineno; // Set current line# - DEBUG printf("cfileno=%d ifile->ifno=%d\n", (int)cfileno, (int)ifile->ifno); - cfileno = ifile->ifno; // Restore current file number -if (debug) printf("[fpop: (post) cfileno=%d ifile->ifno=%d]\n", (int)cfileno, (int)ifile->ifno); - break; - case SRC_IMACRO: // Pop and release an IMACRO - imacro = inobj->inobj.imacro; - imacro->im_link = f_imacro; - f_imacro = imacro; - break; - case SRC_IREPT: // Pop and release an IREPT - DEBUG printf("dealloc IREPT\n"); - p = inobj->inobj.irept->ir_firstln; + case SRC_IMACRO: // Pop and release an IMACRO + { + IMACRO * imacro = inobj->inobj.imacro; + imacro->im_link = f_imacro; + f_imacro = imacro; + break; + } - while (p != NULL) - { - p1 = (LONG *)*p; - p = p1; - } + case SRC_IREPT: // Pop and release an IREPT + { + DEBUG { printf("dealloc IREPT\n"); } + LLIST * p = inobj->inobj.irept->ir_firstln; - break; + // Deallocate repeat lines + while (p != NULL) + { + free(p->line); + p = p->next; } - cur_inobj = inobj->in_link; - inobj->in_link = f_inobj; - f_inobj = inobj; + break; + } } + cur_inobj = inobj->in_link; + inobj->in_link = f_inobj; + f_inobj = inobj; + return 0; } @@ -889,21 +942,23 @@ char * GetNextLine(void) // int TokenizeLine(void) { - char * ln = NULL; // Ptr to current position in line - char * p; // Random character ptr - TOKEN * tk; // Token-deposit ptr + uint8_t * ln = NULL; // Ptr to current position in line + uint8_t * p; // Random character ptr + PTR tk; // Token-deposit ptr int state = 0; // State for keyword detector int j = 0; // Var for keyword detector - char c; // Random char - VALUE v; // Random value - char * nullspot = NULL; // Spot to clobber for SYMBOL termination + uint8_t c; // Random char + uint64_t v; // Random value + uint32_t cursize = 0; // Current line's size (.b, .w, .l, .s, .q, .d) + double f; // Random float + uint8_t * nullspot = NULL; // Spot to clobber for SYMBOL termination int stuffnull; // 1:terminate SYMBOL '\0' at *nullspot - char c1; + uint8_t c1; int stringNum = 0; // Pointer to string locations in tokenized line retry: - if (cur_inobj == NULL) // Return EOF if input stack is empty + if (cur_inobj == NULL) // Return EOF if input stack is empty return TKEOF; // Get another line of input from the current input source: a file, a @@ -918,17 +973,17 @@ retry: case SRC_IFILE: if ((ln = GetNextLine()) == NULL) { -if (debug) printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); - if (fpop() == 0) // Pop input level - goto retry; // Try for more lines +DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); } + if (fpop() == 0) // Pop input level + goto retry; // Try for more lines else { - ifent->if_prev = (IFENT *) - 1; //Signal Assemble() that we have reached EOF with unbalanced if/endifs + ifent->if_prev = (IFENT *)-1; //Signal Assemble() that we have reached EOF with unbalanced if/endifs return TKEOF; } } - curlineno++; // Bump line number + curlineno++; // Bump line number lntag = SPACE; if (as68_flag) @@ -952,27 +1007,29 @@ if (debug) printf("TokenizeLine: Calling fpop() from SRC_IFILE...\n"); } break; + // Macro-block: // o Handle end-of-macro; // o tag the listing-line with an at (@) sign. case SRC_IMACRO: if ((ln = GetNextMacroLine()) == NULL) { - if (ExitMacro() == 0) // Exit macro (pop args, do fpop(), etc) - goto retry; // Try for more lines... + if (ExitMacro() == 0) // Exit macro (pop args, do fpop(), etc) + goto retry; // Try for more lines... else - return TKEOF; // Oops, we got a non zero return code, signal EOF + return TKEOF; // Oops, we got a non zero return code, signal EOF } lntag = '@'; break; + // Repeat-block: // o Handle end-of-repeat-block; // o tag the listing-line with a pound (#) sign. case SRC_IREPT: if ((ln = GetNextRepeatLine()) == NULL) { -if (debug) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); + DEBUG { printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); } fpop(); goto retry; } @@ -981,15 +1038,15 @@ if (debug) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); break; } - // Save text of the line. We only do this during listings and within + // Save text of the line. We only do this during listings and within // macro-type blocks, since it is expensive to unconditionally copy every // line. if (lnsave) strcpy(lnbuf, ln); - // General house-keeping + // General housekeeping tok = tokeol; // Set "tok" to EOL in case of error - tk = etok; // Reset token ptr + tk.u32 = etok; // Reset token ptr stuffnull = 0; // Don't stuff nulls totlines++; // Bump total #lines assembled @@ -998,16 +1055,26 @@ if (debug) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); if (*ln == '*' || *ln == ';' || ((*ln == '/') && (*(ln + 1) == '/'))) goto goteol; + // And here we have a very ugly hack for signalling a single line 'turn off + // optimization'. There's really no nice way to do this, so hack it is! + optimizeOff = 0; // Default is to take optimizations as they come + + if (*ln == '!') + { + optimizeOff = 1; // Signal that we don't want to optimize this line + ln++; // & skip over the darned thing + } + // Main tokenization loop; - // o skip whitespace; - // o handle end-of-line; - // o handle symbols; - // o handle single-character tokens (operators, etc.); - // o handle multiple-character tokens (constants, strings, etc.). + // o skip whitespace; + // o handle end-of-line; + // o handle symbols; + // o handle single-character tokens (operators, etc.); + // o handle multiple-character tokens (constants, strings, etc.). for(; *ln!=EOS;) { // Skip whitespace, handle EOL - while ((int)chrtab[*ln] & WHITE) + while (chrtab[*ln] & WHITE) ln++; // Handle EOL, comment with ';' @@ -1042,7 +1109,7 @@ if (debug) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); // token stream: ln++; stuffnull = 0; - *tk++ = (TOKEN)dotxtab[*ln++]; + *tk.u32++ = (TOKEN)dotxtab[*ln++]; continue; } } @@ -1060,17 +1127,18 @@ if (debug) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); *ln++ = EOS; // Terminate symbol stuffnull = 0; // And never try it again - // Character following the `.' must have a DOT attribute, and + // Character following the '.' must have a DOT attribute, and // the chararacter after THAT one must not have a start-symbol // attribute (to prevent symbols that look like, for example, // "zingo.barf", which might be a good idea anyway....) if (((chrtab[*ln] & DOT) == 0) || (dotxtab[*ln] == 0)) return error("[bwsl] must follow '.' in symbol"); - v = (VALUE)dotxtab[*ln++]; + v = (uint32_t)dotxtab[*ln++]; + cursize = (uint32_t)v; if (chrtab[*ln] & CTSYM) - return error("misuse of '.', not allowed in symbols"); + return error("misuse of '.'; not allowed in symbols"); } // If the symbol is small, check to see if it's really the name of @@ -1103,7 +1171,7 @@ if (debug) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); } // Make j = -1 if user tries to use a RISC register while in 68K mode - if (!(rgpu || rdsp) && ((TOKEN)j >= KW_R0 && (TOKEN)j <= KW_R31)) + if (!(rgpu || rdsp || dsp56001) && ((TOKEN)j >= KW_R0 && (TOKEN)j <= KW_R31)) { j = -1; } @@ -1124,28 +1192,21 @@ if (debug) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); // If not tokenized keyword OR token was not found if ((j < 0) || (state < 0)) { - *tk++ = SYMBOL; -//#warning -//problem here: nullspot is a char * but TOKEN is a uint32_t. On a 64-bit -//system, this will cause all kinds of mischief. -#if 0 - *tk++ = (TOKEN)nullspot; -#else + *tk.u32++ = SYMBOL; string[stringNum] = nullspot; - *tk++ = stringNum; + *tk.u32++ = stringNum; stringNum++; -#endif } else { - *tk++ = (TOKEN)j; + *tk.u32++ = (TOKEN)j; stuffnull = 0; } - if (v) // Record attribute token (if any) - *tk++ = (TOKEN)v; + if (v) // Record attribute token (if any) + *tk.u32++ = (TOKEN)v; - if (stuffnull) // Arrange for string termination on next pass + if (stuffnull) // Arrange for string termination on next pass nullspot = ln; continue; @@ -1154,40 +1215,39 @@ if (debug) printf("TokenizeLine: Calling fpop() from SRC_IREPT...\n"); // Handle identity tokens if (c & SELF) { - *tk++ = *ln++; + *tk.u32++ = *ln++; continue; } // Handle multiple-character tokens if (c & MULTX) { - switch (*ln++) { case '!': // ! or != if (*ln == '=') { - *tk++ = NE; - ++ln; + *tk.u32++ = NE; + ln++; } else - *tk++ = '!'; + *tk.u32++ = '!'; continue; case '\'': // 'string' if (m6502) { // Hardcoded for now, maybe this will change in the future - *tk++ = STRINGA8; + *tk.u32++ = STRINGA8; goto dostring; } // Fall through case '\"': // "string" - *tk++ = STRING; + *tk.u32++ = STRING; dostring: c1 = ln[-1]; string[stringNum] = ln; - *tk++ = stringNum; + *tk.u32++ = stringNum; stringNum++; for(p=ln; *ln!=EOS && *ln!=c1;) @@ -1227,6 +1287,15 @@ dostring: case '\\': c = '\\'; break; + case '{': + // If we're evaluating a macro + // this is valid because it's + // a parameter expansion + case '!': + // If we're evaluating a macro + // this is valid and expands to + // "dot-size" + break; default: warn("bad backslash code in string"); ln--; @@ -1251,132 +1320,85 @@ dostring: while (hextab[*ln] >= 0) v = (v << 4) + (int)hextab[*ln++]; - // ggn: Okay, some comments here are in order I think.... - // The original madmac sources didn't parse the size at - // this point (i.e. .b/.w/.l). It was probably done at - // another point, although it's unclear to me exactly - // where. So why change this? My understanding (at least - // from what SCPCD said on IRC) is that .w addressing - // formats produce wrong code on jaguar (or doesn't execute - // properly? something like that). So the code was changed - // to mask off the upper bits depending on length (note: I - // don't think .b is valid at all! I only know of .w/.l, so - // this should probably be wiped). Then the code that - // parses the constant and checks to see if it's between - // $ffff0000 and $8000 never got triggered, so yay job - // done! ...now say we want to assemble a st .prg. One of - // the most widely spread optimisations is move.X expr.w,Y - // (or vice versa, or both, anyway...) to access hardware - // registers (which are mapped to $fxxxxx). This botchy - // thing would create "hilarious" code while trying to - // access hardware registers. So I made a condition to see - // if st mode or jaguar is active and apply the both or - // not. One last note: this is hardcoded to get optimised - // for now on ST mode, i.e. it can't generate code like - // move.w $00001234,d0 - it'll always get optimised to - // move.w $1234.w,d0. It's probably ok, but maybe a warning - // should be emitted? Or maybe finding a way to make it not - // auto-optimise? I think it's ok for now... - if (*ln == '.') - { - if (obj_format == BSD) - { - if ((*(ln + 1) & 0xDF) == 'B') - { - v &= 0x000000FF; - ln += 2; - } - else if ((*(ln + 1) & 0xDF) == 'W') - { - v &= 0x0000FFFF; - ln += 2; - } - else if ((*(ln + 1) & 0xDF) == 'L') - { - ln += 2; - } - } - } - - *tk++ = CONST; - *tk++ = v; + *tk.u32++ = CONST; + *tk.u64++ = v; - if (obj_format == ALCYON) + if (*ln == '.') { if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W')) { - *tk++ = DOTW; + *tk.u32++ = DOTW; ln += 2; } else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L')) { - *tk++ = DOTL; + *tk.u32++ = DOTL; ln += 2; } } } else - *tk++ = '$'; + *tk.u32++ = '$'; continue; case '<': // < or << or <> or <= switch (*ln) { case '<': - *tk++ = SHL; - ++ln; + *tk.u32++ = SHL; + ln++; continue; case '>': - *tk++ = NE; - ++ln; + *tk.u32++ = NE; + ln++; continue; case '=': - *tk++ = LE; - ++ln; + *tk.u32++ = LE; + ln++; continue; default: - *tk++ = '<'; + *tk.u32++ = '<'; continue; } case ':': // : or :: if (*ln == ':') { - *tk++ = DCOLON; - ++ln; + *tk.u32++ = DCOLON; + ln++; } else - *tk++ = ':'; + *tk.u32++ = ':'; continue; case '=': // = or == if (*ln == '=') { - *tk++ = DEQUALS; - ++ln; + *tk.u32++ = DEQUALS; + ln++; } else - *tk++ = '='; + *tk.u32++ = '='; continue; case '>': // > or >> or >= switch (*ln) { case '>': - *tk++ = SHR; + *tk.u32++ = SHR; ln++; continue; case '=': - *tk++ = GE; + *tk.u32++ = GE; ln++; continue; default: - *tk++ = '>'; + *tk.u32++ = '>'; continue; } case '%': // % or binary constant if (*ln < '0' || *ln > '1') { - *tk++ = '%'; + *tk.u32++ = '%'; continue; } @@ -1401,17 +1423,18 @@ dostring: if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L')) { + v &= 0xFFFFFFFF; ln += 2; } } - *tk++ = CONST; - *tk++ = v; + *tk.u32++ = CONST; + *tk.u64++ = v; continue; case '@': // @ or octal constant if (*ln < '0' || *ln > '7') { - *tk++ = '@'; + *tk.u32++ = '@'; continue; } @@ -1422,31 +1445,32 @@ dostring: if (*ln == '.') { - if ((*(ln+1) == 'b') || (*(ln+1) == 'B')) + if ((*(ln + 1) == 'b') || (*(ln + 1) == 'B')) { v &= 0x000000FF; ln += 2; } - if ((*(ln+1) == 'w') || (*(ln+1) == 'W')) + if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W')) { v &= 0x0000FFFF; ln += 2; } - if ((*(ln+1) == 'l') || (*(ln+1) == 'L')) + if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L')) { + v &= 0xFFFFFFFF; ln += 2; } } - *tk++ = CONST; - *tk++ = v; + *tk.u32++ = CONST; + *tk.u64++ = v; continue; case '^': // ^ or ^^ if (*ln != '^') { - *tk++ = '^'; + *tk.u32++ = '^'; continue; } @@ -1492,7 +1516,7 @@ dostring: continue; } - *tk++ = (TOKEN)j; + *tk.u32++ = (TOKEN)j; continue; default: interror(2); // Bad MULTX entry in chrtab @@ -1503,6 +1527,7 @@ dostring: // Handle decimal constant if (c & DIGIT) { + uint8_t * numStart = ln; v = 0; while ((int)chrtab[*ln] & DIGIT) @@ -1515,37 +1540,72 @@ dostring: { v &= 0x000000FF; ln += 2; + *tk.u32++ = CONST; + *tk.u64++ = v; + *tk.u32++ = DOTB; } else if ((*(ln + 1) == 'w') || (*(ln + 1) == 'W')) { v &= 0x0000FFFF; ln += 2; + *tk.u32++ = CONST; + *tk.u64++ = v; + *tk.u32++ = DOTW; } else if ((*(ln + 1) == 'l') || (*(ln + 1) == 'L')) { + v &= 0xFFFFFFFF; ln += 2; + *tk.u32++ = CONST; + *tk.u64++ = v; + *tk.u32++ = DOTL; + } + else if ((int)chrtab[*(ln + 1)] & DIGIT) + { + // Hey, more digits after the dot, so we assume it's a + // floating point number of some kind... numEnd will point + // to the first non-float character after it's done + char * numEnd; + errno = 0; + double f = strtod(numStart, &numEnd); + ln = (uint8_t *)numEnd; + + if (errno != 0) + return error("floating point parse error"); + + // N.B.: We use the C compiler's internal double + // representation for all internal float calcs and + // are reasonably sure that the size of said double + // is 8 bytes long (which we check for in fltpoint.c) + *tk.u32++ = FCONST; + *tk.dp = f; + tk.u64++; + continue; } } + else + { + *tk.u32++ = CONST; + *tk.u64++ = v; + } - *tk++ = CONST; - *tk++ = v; //printf("CONST: %i\n", v); continue; } // Handle illegal character - return error("illegal character"); + return error("illegal character $%02X found", *ln); } // Terminate line of tokens and return "success." goteol: - tok = etok; // Set tok to beginning of line + tok = etok; // Set tok to beginning of line - if (stuffnull) // Terminate last SYMBOL + if (stuffnull) // Terminate last SYMBOL *nullspot = EOS; - *tk++ = EOL; + *tk.u32++ = EOL; return OK; } @@ -1579,7 +1639,7 @@ int d_goto(WORD unused) return error("goto not in macro"); IMACRO * imacro = cur_inobj->inobj.imacro; - struct LineList * defln = imacro->im_macro->lineList; + LLIST * defln = imacro->im_macro->lineList; // Attempt to find the label, starting with the first line. for(; defln!=NULL; defln=defln->next) @@ -1616,22 +1676,114 @@ int d_goto(WORD unused) } +void DumpToken(TOKEN t) +{ + if (t == COLON) + printf("[COLON]"); + else if (t == CONST) + printf("[CONST]"); + else if (t == FCONST) + printf("[FCONST]"); + else if (t == ACONST) + printf("[ACONST]"); + else if (t == STRING) + printf("[STRING]"); + else if (t == SYMBOL) + printf("[SYMBOL]"); + else if (t == EOS) + printf("[EOS]"); + else if (t == TKEOF) + printf("[TKEOF]"); + else if (t == DEQUALS) + printf("[DEQUALS]"); + else if (t == SET) + printf("[SET]"); + else if (t == REG) + printf("[REG]"); + else if (t == DCOLON) + printf("[DCOLON]"); + else if (t == GE) + printf("[GE]"); + else if (t == LE) + printf("[LE]"); + else if (t == NE) + printf("[NE]"); + else if (t == SHR) + printf("[SHR]"); + else if (t == SHL) + printf("[SHL]"); + else if (t == UNMINUS) + printf("[UNMINUS]"); + else if (t == DOTB) + printf("[DOTB]"); + else if (t == DOTW) + printf("[DOTW]"); + else if (t == DOTL) + printf("[DOTL]"); + else if (t == DOTQ) + printf("[DOTQ]"); + else if (t == DOTS) + printf("[DOTS]"); + else if (t == DOTD) + printf("[DOTD]"); + else if (t == DOTI) + printf("[DOTI]"); + else if (t == ENDEXPR) + printf("[ENDEXPR]"); + else if (t == CR_ABSCOUNT) + printf("[CR_ABSCOUNT]"); + else if (t == CR_DEFINED) + printf("[CR_DEFINED]"); + else if (t == CR_REFERENCED) + printf("[CR_REFERENCED]"); + else if (t == CR_STREQ) + printf("[CR_STREQ]"); + else if (t == CR_MACDEF) + printf("[CR_MACDEF]"); + else if (t == CR_TIME) + printf("[CR_TIME]"); + else if (t == CR_DATE) + printf("[CR_DATE]"); + else if (t >= 0x20 && t <= 0x2F) + printf("[%c]", (char)t); + else if (t >= 0x3A && t <= 0x3F) + printf("[%c]", (char)t); + else if (t >= 0x80 && t <= 0x87) + printf("[D%u]", ((uint32_t)t) - 0x80); + else if (t >= 0x88 && t <= 0x8F) + printf("[A%u]", ((uint32_t)t) - 0x88); + else + printf("[%X:%c]", (uint32_t)t, (char)t); +} + + void DumpTokenBuffer(void) { - TOKEN * t; printf("Tokens [%X]: ", sloc); - for(t=tokbuf; *t!=EOL; t++) + for(TOKEN * t=tokbuf; *t!=EOL; t++) { if (*t == COLON) printf("[COLON]"); else if (*t == CONST) { - t++; - printf("[CONST: $%X]", (uint32_t)*t); + PTR tp; + tp.u32 = t + 1; + printf("[CONST: $%lX]", *tp.u64); + t += 2; + } + else if (*t == FCONST) + { + PTR tp; + tp.u32 = t + 1; + printf("[FCONST: $%lX]", *tp.u64); + t += 2; } else if (*t == ACONST) - printf("[ACONST]"); + { + printf("[ACONST: $%X, $%X]", (uint32_t)t[1], (uint32_t)t[2]); + t += 2; + } else if (*t == STRING) { t++; @@ -1672,6 +1824,12 @@ void DumpTokenBuffer(void) printf("[DOTW]"); else if (*t == DOTL) printf("[DOTL]"); + else if (*t == DOTQ) + printf("[DOTQ]"); + else if (*t == DOTS) + printf("[DOTS]"); + else if (*t == DOTD) + printf("[DOTD]"); else if (*t == DOTI) printf("[DOTI]"); else if (*t == ENDEXPR)