NASM 2.06rc3
[nasm/perl-rewrite.git] / preproc.c
blobc6dd1b729b30847ecd1587130e02c98b9e062ab5
1 /* preproc.c macro preprocessor for the Netwide Assembler
3 * The Netwide Assembler is copyright (C) 1996 Simon Tatham and
4 * Julian Hall. All rights reserved. The software is
5 * redistributable under the license given in the file "LICENSE"
6 * distributed in the NASM archive.
8 * initial version 18/iii/97 by Simon Tatham
9 */
11 /* Typical flow of text through preproc
13 * pp_getline gets tokenized lines, either
15 * from a macro expansion
17 * or
18 * {
19 * read_line gets raw text from stdmacpos, or predef, or current input file
20 * tokenize converts to tokens
21 * }
23 * expand_mmac_params is used to expand %1 etc., unless a macro is being
24 * defined or a false conditional is being processed
25 * (%0, %1, %+1, %-1, %%foo
27 * do_directive checks for directives
29 * expand_smacro is used to expand single line macros
31 * expand_mmacro is used to expand multi-line macros
33 * detoken is used to convert the line back to text
36 #include "compiler.h"
38 #include <stdio.h>
39 #include <stdarg.h>
40 #include <stdlib.h>
41 #include <stddef.h>
42 #include <string.h>
43 #include <ctype.h>
44 #include <limits.h>
45 #include <inttypes.h>
47 #include "nasm.h"
48 #include "nasmlib.h"
49 #include "preproc.h"
50 #include "hashtbl.h"
51 #include "quote.h"
52 #include "stdscan.h"
53 #include "tokens.h"
54 #include "tables.h"
56 typedef struct SMacro SMacro;
57 typedef struct MMacro MMacro;
58 typedef struct Context Context;
59 typedef struct Token Token;
60 typedef struct Blocks Blocks;
61 typedef struct Line Line;
62 typedef struct Include Include;
63 typedef struct Cond Cond;
64 typedef struct IncPath IncPath;
67 * Note on the storage of both SMacro and MMacros: the hash table
68 * indexes them case-insensitively, and we then have to go through a
69 * linked list of potential case aliases (and, for MMacros, parameter
70 * ranges); this is to preserve the matching semantics of the earlier
71 * code. If the number of case aliases for a specific macro is a
72 * performance issue, you may want to reconsider your coding style.
76 * Store the definition of a single-line macro.
78 struct SMacro {
79 SMacro *next;
80 char *name;
81 bool casesense;
82 bool in_progress;
83 unsigned int nparam;
84 Token *expansion;
88 * Store the definition of a multi-line macro. This is also used to
89 * store the interiors of `%rep...%endrep' blocks, which are
90 * effectively self-re-invoking multi-line macros which simply
91 * don't have a name or bother to appear in the hash tables. %rep
92 * blocks are signified by having a NULL `name' field.
94 * In a MMacro describing a `%rep' block, the `in_progress' field
95 * isn't merely boolean, but gives the number of repeats left to
96 * run.
98 * The `next' field is used for storing MMacros in hash tables; the
99 * `next_active' field is for stacking them on istk entries.
101 * When a MMacro is being expanded, `params', `iline', `nparam',
102 * `paramlen', `rotate' and `unique' are local to the invocation.
104 struct MMacro {
105 MMacro *next;
106 char *name;
107 int nparam_min, nparam_max;
108 bool casesense;
109 bool plus; /* is the last parameter greedy? */
110 bool nolist; /* is this macro listing-inhibited? */
111 int64_t in_progress;
112 Token *dlist; /* All defaults as one list */
113 Token **defaults; /* Parameter default pointers */
114 int ndefs; /* number of default parameters */
115 Line *expansion;
117 MMacro *next_active;
118 MMacro *rep_nest; /* used for nesting %rep */
119 Token **params; /* actual parameters */
120 Token *iline; /* invocation line */
121 unsigned int nparam, rotate;
122 int *paramlen;
123 uint64_t unique;
124 int lineno; /* Current line number on expansion */
128 * The context stack is composed of a linked list of these.
130 struct Context {
131 Context *next;
132 char *name;
133 struct hash_table localmac;
134 uint32_t number;
138 * This is the internal form which we break input lines up into.
139 * Typically stored in linked lists.
141 * Note that `type' serves a double meaning: TOK_SMAC_PARAM is not
142 * necessarily used as-is, but is intended to denote the number of
143 * the substituted parameter. So in the definition
145 * %define a(x,y) ( (x) & ~(y) )
147 * the token representing `x' will have its type changed to
148 * TOK_SMAC_PARAM, but the one representing `y' will be
149 * TOK_SMAC_PARAM+1.
151 * TOK_INTERNAL_STRING is a dirty hack: it's a single string token
152 * which doesn't need quotes around it. Used in the pre-include
153 * mechanism as an alternative to trying to find a sensible type of
154 * quote to use on the filename we were passed.
156 enum pp_token_type {
157 TOK_NONE = 0, TOK_WHITESPACE, TOK_COMMENT, TOK_ID,
158 TOK_PREPROC_ID, TOK_STRING,
159 TOK_NUMBER, TOK_FLOAT, TOK_SMAC_END, TOK_OTHER,
160 TOK_INTERNAL_STRING,
161 TOK_PREPROC_Q, TOK_PREPROC_QQ,
162 TOK_INDIRECT, /* %[...] */
163 TOK_SMAC_PARAM, /* MUST BE LAST IN THE LIST!!! */
164 TOK_MAX = INT_MAX /* Keep compiler from reducing the range */
167 struct Token {
168 Token *next;
169 char *text;
170 union {
171 SMacro *mac; /* associated macro for TOK_SMAC_END */
172 size_t len; /* scratch length field */
173 } a; /* Auxiliary data */
174 enum pp_token_type type;
178 * Multi-line macro definitions are stored as a linked list of
179 * these, which is essentially a container to allow several linked
180 * lists of Tokens.
182 * Note that in this module, linked lists are treated as stacks
183 * wherever possible. For this reason, Lines are _pushed_ on to the
184 * `expansion' field in MMacro structures, so that the linked list,
185 * if walked, would give the macro lines in reverse order; this
186 * means that we can walk the list when expanding a macro, and thus
187 * push the lines on to the `expansion' field in _istk_ in reverse
188 * order (so that when popped back off they are in the right
189 * order). It may seem cockeyed, and it relies on my design having
190 * an even number of steps in, but it works...
192 * Some of these structures, rather than being actual lines, are
193 * markers delimiting the end of the expansion of a given macro.
194 * This is for use in the cycle-tracking and %rep-handling code.
195 * Such structures have `finishes' non-NULL, and `first' NULL. All
196 * others have `finishes' NULL, but `first' may still be NULL if
197 * the line is blank.
199 struct Line {
200 Line *next;
201 MMacro *finishes;
202 Token *first;
206 * To handle an arbitrary level of file inclusion, we maintain a
207 * stack (ie linked list) of these things.
209 struct Include {
210 Include *next;
211 FILE *fp;
212 Cond *conds;
213 Line *expansion;
214 char *fname;
215 int lineno, lineinc;
216 MMacro *mstk; /* stack of active macros/reps */
220 * Include search path. This is simply a list of strings which get
221 * prepended, in turn, to the name of an include file, in an
222 * attempt to find the file if it's not in the current directory.
224 struct IncPath {
225 IncPath *next;
226 char *path;
230 * Conditional assembly: we maintain a separate stack of these for
231 * each level of file inclusion. (The only reason we keep the
232 * stacks separate is to ensure that a stray `%endif' in a file
233 * included from within the true branch of a `%if' won't terminate
234 * it and cause confusion: instead, rightly, it'll cause an error.)
236 struct Cond {
237 Cond *next;
238 int state;
240 enum {
242 * These states are for use just after %if or %elif: IF_TRUE
243 * means the condition has evaluated to truth so we are
244 * currently emitting, whereas IF_FALSE means we are not
245 * currently emitting but will start doing so if a %else comes
246 * up. In these states, all directives are admissible: %elif,
247 * %else and %endif. (And of course %if.)
249 COND_IF_TRUE, COND_IF_FALSE,
251 * These states come up after a %else: ELSE_TRUE means we're
252 * emitting, and ELSE_FALSE means we're not. In ELSE_* states,
253 * any %elif or %else will cause an error.
255 COND_ELSE_TRUE, COND_ELSE_FALSE,
257 * These states mean that we're not emitting now, and also that
258 * nothing until %endif will be emitted at all. COND_DONE is
259 * used when we've had our moment of emission
260 * and have now started seeing %elifs. COND_NEVER is used when
261 * the condition construct in question is contained within a
262 * non-emitting branch of a larger condition construct,
263 * or if there is an error.
265 COND_DONE, COND_NEVER
267 #define emitting(x) ( (x) == COND_IF_TRUE || (x) == COND_ELSE_TRUE )
270 * These defines are used as the possible return values for do_directive
272 #define NO_DIRECTIVE_FOUND 0
273 #define DIRECTIVE_FOUND 1
276 * Condition codes. Note that we use c_ prefix not C_ because C_ is
277 * used in nasm.h for the "real" condition codes. At _this_ level,
278 * we treat CXZ and ECXZ as condition codes, albeit non-invertible
279 * ones, so we need a different enum...
281 static const char * const conditions[] = {
282 "a", "ae", "b", "be", "c", "cxz", "e", "ecxz", "g", "ge", "l", "le",
283 "na", "nae", "nb", "nbe", "nc", "ne", "ng", "nge", "nl", "nle", "no",
284 "np", "ns", "nz", "o", "p", "pe", "po", "rcxz", "s", "z"
286 enum pp_conds {
287 c_A, c_AE, c_B, c_BE, c_C, c_CXZ, c_E, c_ECXZ, c_G, c_GE, c_L, c_LE,
288 c_NA, c_NAE, c_NB, c_NBE, c_NC, c_NE, c_NG, c_NGE, c_NL, c_NLE, c_NO,
289 c_NP, c_NS, c_NZ, c_O, c_P, c_PE, c_PO, c_RCXZ, c_S, c_Z,
290 c_none = -1
292 static const enum pp_conds inverse_ccs[] = {
293 c_NA, c_NAE, c_NB, c_NBE, c_NC, -1, c_NE, -1, c_NG, c_NGE, c_NL, c_NLE,
294 c_A, c_AE, c_B, c_BE, c_C, c_E, c_G, c_GE, c_L, c_LE, c_O, c_P, c_S,
295 c_Z, c_NO, c_NP, c_PO, c_PE, -1, c_NS, c_NZ
299 * Directive names.
301 /* If this is a an IF, ELIF, ELSE or ENDIF keyword */
302 static int is_condition(enum preproc_token arg)
304 return PP_IS_COND(arg) || (arg == PP_ELSE) || (arg == PP_ENDIF);
307 /* For TASM compatibility we need to be able to recognise TASM compatible
308 * conditional compilation directives. Using the NASM pre-processor does
309 * not work, so we look for them specifically from the following list and
310 * then jam in the equivalent NASM directive into the input stream.
313 enum {
314 TM_ARG, TM_ELIF, TM_ELSE, TM_ENDIF, TM_IF, TM_IFDEF, TM_IFDIFI,
315 TM_IFNDEF, TM_INCLUDE, TM_LOCAL
318 static const char * const tasm_directives[] = {
319 "arg", "elif", "else", "endif", "if", "ifdef", "ifdifi",
320 "ifndef", "include", "local"
323 static int StackSize = 4;
324 static char *StackPointer = "ebp";
325 static int ArgOffset = 8;
326 static int LocalOffset = 0;
328 static Context *cstk;
329 static Include *istk;
330 static IncPath *ipath = NULL;
332 static efunc _error; /* Pointer to client-provided error reporting function */
333 static evalfunc evaluate;
335 static int pass; /* HACK: pass 0 = generate dependencies only */
336 static StrList **dephead, **deptail; /* Dependency list */
338 static uint64_t unique; /* unique identifier numbers */
340 static Line *predef = NULL;
341 static bool do_predef;
343 static ListGen *list;
346 * The current set of multi-line macros we have defined.
348 static struct hash_table mmacros;
351 * The current set of single-line macros we have defined.
353 static struct hash_table smacros;
356 * The multi-line macro we are currently defining, or the %rep
357 * block we are currently reading, if any.
359 static MMacro *defining;
361 static uint64_t nested_mac_count;
362 static uint64_t nested_rep_count;
365 * The number of macro parameters to allocate space for at a time.
367 #define PARAM_DELTA 16
370 * The standard macro set: defined in macros.c in the array nasm_stdmac.
371 * This gives our position in the macro set, when we're processing it.
373 static macros_t *stdmacpos;
376 * The extra standard macros that come from the object format, if
377 * any.
379 static macros_t *extrastdmac = NULL;
380 static bool any_extrastdmac;
383 * Tokens are allocated in blocks to improve speed
385 #define TOKEN_BLOCKSIZE 4096
386 static Token *freeTokens = NULL;
387 struct Blocks {
388 Blocks *next;
389 void *chunk;
392 static Blocks blocks = { NULL, NULL };
395 * Forward declarations.
397 static Token *expand_mmac_params(Token * tline);
398 static Token *expand_smacro(Token * tline);
399 static Token *expand_id(Token * tline);
400 static Context *get_ctx(const char *name, bool all_contexts);
401 static void make_tok_num(Token * tok, int64_t val);
402 static void error(int severity, const char *fmt, ...);
403 static void error_precond(int severity, const char *fmt, ...);
404 static void *new_Block(size_t size);
405 static void delete_Blocks(void);
406 static Token *new_Token(Token * next, enum pp_token_type type,
407 const char *text, int txtlen);
408 static Token *delete_Token(Token * t);
411 * Macros for safe checking of token pointers, avoid *(NULL)
413 #define tok_type_(x,t) ((x) && (x)->type == (t))
414 #define skip_white_(x) if (tok_type_((x), TOK_WHITESPACE)) (x)=(x)->next
415 #define tok_is_(x,v) (tok_type_((x), TOK_OTHER) && !strcmp((x)->text,(v)))
416 #define tok_isnt_(x,v) ((x) && ((x)->type!=TOK_OTHER || strcmp((x)->text,(v))))
418 /* Handle TASM specific directives, which do not contain a % in
419 * front of them. We do it here because I could not find any other
420 * place to do it for the moment, and it is a hack (ideally it would
421 * be nice to be able to use the NASM pre-processor to do it).
423 static char *check_tasm_directive(char *line)
425 int32_t i, j, k, m, len;
426 char *p = line, *oldline, oldchar;
428 /* Skip whitespace */
429 while (nasm_isspace(*p) && *p != 0)
430 p++;
432 /* Binary search for the directive name */
433 i = -1;
434 j = elements(tasm_directives);
435 len = 0;
436 while (!nasm_isspace(p[len]) && p[len] != 0)
437 len++;
438 if (len) {
439 oldchar = p[len];
440 p[len] = 0;
441 while (j - i > 1) {
442 k = (j + i) / 2;
443 m = nasm_stricmp(p, tasm_directives[k]);
444 if (m == 0) {
445 /* We have found a directive, so jam a % in front of it
446 * so that NASM will then recognise it as one if it's own.
448 p[len] = oldchar;
449 len = strlen(p);
450 oldline = line;
451 line = nasm_malloc(len + 2);
452 line[0] = '%';
453 if (k == TM_IFDIFI) {
454 /* NASM does not recognise IFDIFI, so we convert it to
455 * %ifdef BOGUS. This is not used in NASM comaptible
456 * code, but does need to parse for the TASM macro
457 * package.
459 strcpy(line + 1, "ifdef BOGUS");
460 } else {
461 memcpy(line + 1, p, len + 1);
463 nasm_free(oldline);
464 return line;
465 } else if (m < 0) {
466 j = k;
467 } else
468 i = k;
470 p[len] = oldchar;
472 return line;
476 * The pre-preprocessing stage... This function translates line
477 * number indications as they emerge from GNU cpp (`# lineno "file"
478 * flags') into NASM preprocessor line number indications (`%line
479 * lineno file').
481 static char *prepreproc(char *line)
483 int lineno, fnlen;
484 char *fname, *oldline;
486 if (line[0] == '#' && line[1] == ' ') {
487 oldline = line;
488 fname = oldline + 2;
489 lineno = atoi(fname);
490 fname += strspn(fname, "0123456789 ");
491 if (*fname == '"')
492 fname++;
493 fnlen = strcspn(fname, "\"");
494 line = nasm_malloc(20 + fnlen);
495 snprintf(line, 20 + fnlen, "%%line %d %.*s", lineno, fnlen, fname);
496 nasm_free(oldline);
498 if (tasm_compatible_mode)
499 return check_tasm_directive(line);
500 return line;
504 * Free a linked list of tokens.
506 static void free_tlist(Token * list)
508 while (list) {
509 list = delete_Token(list);
514 * Free a linked list of lines.
516 static void free_llist(Line * list)
518 Line *l;
519 while (list) {
520 l = list;
521 list = list->next;
522 free_tlist(l->first);
523 nasm_free(l);
528 * Free an MMacro
530 static void free_mmacro(MMacro * m)
532 nasm_free(m->name);
533 free_tlist(m->dlist);
534 nasm_free(m->defaults);
535 free_llist(m->expansion);
536 nasm_free(m);
540 * Free all currently defined macros, and free the hash tables
542 static void free_smacro_table(struct hash_table *smt)
544 SMacro *s;
545 const char *key;
546 struct hash_tbl_node *it = NULL;
548 while ((s = hash_iterate(smt, &it, &key)) != NULL) {
549 nasm_free((void *)key);
550 while (s) {
551 SMacro *ns = s->next;
552 nasm_free(s->name);
553 free_tlist(s->expansion);
554 nasm_free(s);
555 s = ns;
558 hash_free(smt);
561 static void free_mmacro_table(struct hash_table *mmt)
563 MMacro *m;
564 const char *key;
565 struct hash_tbl_node *it = NULL;
567 it = NULL;
568 while ((m = hash_iterate(mmt, &it, &key)) != NULL) {
569 nasm_free((void *)key);
570 while (m) {
571 MMacro *nm = m->next;
572 free_mmacro(m);
573 m = nm;
576 hash_free(mmt);
579 static void free_macros(void)
581 free_smacro_table(&smacros);
582 free_mmacro_table(&mmacros);
586 * Initialize the hash tables
588 static void init_macros(void)
590 hash_init(&smacros, HASH_LARGE);
591 hash_init(&mmacros, HASH_LARGE);
595 * Pop the context stack.
597 static void ctx_pop(void)
599 Context *c = cstk;
601 cstk = cstk->next;
602 free_smacro_table(&c->localmac);
603 nasm_free(c->name);
604 nasm_free(c);
608 * Search for a key in the hash index; adding it if necessary
609 * (in which case we initialize the data pointer to NULL.)
611 static void **
612 hash_findi_add(struct hash_table *hash, const char *str)
614 struct hash_insert hi;
615 void **r;
616 char *strx;
618 r = hash_findi(hash, str, &hi);
619 if (r)
620 return r;
622 strx = nasm_strdup(str); /* Use a more efficient allocator here? */
623 return hash_add(&hi, strx, NULL);
627 * Like hash_findi, but returns the data element rather than a pointer
628 * to it. Used only when not adding a new element, hence no third
629 * argument.
631 static void *
632 hash_findix(struct hash_table *hash, const char *str)
634 void **p;
636 p = hash_findi(hash, str, NULL);
637 return p ? *p : NULL;
640 #define BUF_DELTA 512
642 * Read a line from the top file in istk, handling multiple CR/LFs
643 * at the end of the line read, and handling spurious ^Zs. Will
644 * return lines from the standard macro set if this has not already
645 * been done.
647 static char *read_line(void)
649 char *buffer, *p, *q;
650 int bufsize, continued_count;
652 if (stdmacpos) {
653 unsigned char c;
654 const unsigned char *p = stdmacpos;
655 char *ret, *q;
656 size_t len = 0;
657 while ((c = *p++)) {
658 if (c >= 0x80)
659 len += pp_directives_len[c-0x80]+1;
660 else
661 len++;
663 ret = nasm_malloc(len+1);
664 q = ret;
665 while ((c = *stdmacpos++)) {
666 if (c >= 0x80) {
667 memcpy(q, pp_directives[c-0x80], pp_directives_len[c-0x80]);
668 q += pp_directives_len[c-0x80];
669 *q++ = ' ';
670 } else {
671 *q++ = c;
674 stdmacpos = p;
675 *q = '\0';
677 if (!*stdmacpos) {
678 /* This was the last of the standard macro chain... */
679 stdmacpos = NULL;
680 if (any_extrastdmac) {
681 stdmacpos = extrastdmac;
682 any_extrastdmac = false;
683 } else if (do_predef) {
684 Line *pd, *l;
685 Token *head, **tail, *t;
688 * Nasty hack: here we push the contents of
689 * `predef' on to the top-level expansion stack,
690 * since this is the most convenient way to
691 * implement the pre-include and pre-define
692 * features.
694 for (pd = predef; pd; pd = pd->next) {
695 head = NULL;
696 tail = &head;
697 for (t = pd->first; t; t = t->next) {
698 *tail = new_Token(NULL, t->type, t->text, 0);
699 tail = &(*tail)->next;
701 l = nasm_malloc(sizeof(Line));
702 l->next = istk->expansion;
703 l->first = head;
704 l->finishes = NULL;
705 istk->expansion = l;
707 do_predef = false;
710 return ret;
713 bufsize = BUF_DELTA;
714 buffer = nasm_malloc(BUF_DELTA);
715 p = buffer;
716 continued_count = 0;
717 while (1) {
718 q = fgets(p, bufsize - (p - buffer), istk->fp);
719 if (!q)
720 break;
721 p += strlen(p);
722 if (p > buffer && p[-1] == '\n') {
723 /* Convert backslash-CRLF line continuation sequences into
724 nothing at all (for DOS and Windows) */
725 if (((p - 2) > buffer) && (p[-3] == '\\') && (p[-2] == '\r')) {
726 p -= 3;
727 *p = 0;
728 continued_count++;
730 /* Also convert backslash-LF line continuation sequences into
731 nothing at all (for Unix) */
732 else if (((p - 1) > buffer) && (p[-2] == '\\')) {
733 p -= 2;
734 *p = 0;
735 continued_count++;
736 } else {
737 break;
740 if (p - buffer > bufsize - 10) {
741 int32_t offset = p - buffer;
742 bufsize += BUF_DELTA;
743 buffer = nasm_realloc(buffer, bufsize);
744 p = buffer + offset; /* prevent stale-pointer problems */
748 if (!q && p == buffer) {
749 nasm_free(buffer);
750 return NULL;
753 src_set_linnum(src_get_linnum() + istk->lineinc +
754 (continued_count * istk->lineinc));
757 * Play safe: remove CRs as well as LFs, if any of either are
758 * present at the end of the line.
760 while (--p >= buffer && (*p == '\n' || *p == '\r'))
761 *p = '\0';
764 * Handle spurious ^Z, which may be inserted into source files
765 * by some file transfer utilities.
767 buffer[strcspn(buffer, "\032")] = '\0';
769 list->line(LIST_READ, buffer);
771 return buffer;
775 * Tokenize a line of text. This is a very simple process since we
776 * don't need to parse the value out of e.g. numeric tokens: we
777 * simply split one string into many.
779 static Token *tokenize(char *line)
781 char c, *p = line;
782 enum pp_token_type type;
783 Token *list = NULL;
784 Token *t, **tail = &list;
786 while (*line) {
787 p = line;
788 if (*p == '%') {
789 p++;
790 if (nasm_isdigit(*p) ||
791 ((*p == '-' || *p == '+') && nasm_isdigit(p[1])) ||
792 ((*p == '+') && (nasm_isspace(p[1]) || !p[1]))) {
793 do {
794 p++;
796 while (nasm_isdigit(*p));
797 type = TOK_PREPROC_ID;
798 } else if (*p == '{') {
799 p++;
800 while (*p && *p != '}') {
801 p[-1] = *p;
802 p++;
804 p[-1] = '\0';
805 if (*p)
806 p++;
807 type = TOK_PREPROC_ID;
808 } else if (*p == '[') {
809 int lvl = 1;
810 line += 2; /* Skip the leading %[ */
811 p++;
812 while (lvl && (c = *p++)) {
813 switch (c) {
814 case ']':
815 lvl--;
816 break;
817 case '%':
818 if (*p == '[')
819 lvl++;
820 break;
821 case '\'':
822 case '\"':
823 case '`':
824 p = nasm_skip_string(p)+1;
825 break;
826 default:
827 break;
830 p--;
831 if (*p)
832 *p++ = '\0';
833 if (lvl)
834 error(ERR_NONFATAL, "unterminated %[ construct");
835 type = TOK_INDIRECT;
836 } else if (*p == '?') {
837 type = TOK_PREPROC_Q; /* %? */
838 p++;
839 if (*p == '?') {
840 type = TOK_PREPROC_QQ; /* %?? */
841 p++;
843 } else if (isidchar(*p) ||
844 ((*p == '!' || *p == '%' || *p == '$') &&
845 isidchar(p[1]))) {
846 do {
847 p++;
849 while (isidchar(*p));
850 type = TOK_PREPROC_ID;
851 } else {
852 type = TOK_OTHER;
853 if (*p == '%')
854 p++;
856 } else if (isidstart(*p) || (*p == '$' && isidstart(p[1]))) {
857 type = TOK_ID;
858 p++;
859 while (*p && isidchar(*p))
860 p++;
861 } else if (*p == '\'' || *p == '"' || *p == '`') {
863 * A string token.
865 type = TOK_STRING;
866 p = nasm_skip_string(p);
868 if (*p) {
869 p++;
870 } else {
871 error(ERR_WARNING|ERR_PASS1, "unterminated string");
872 /* Handling unterminated strings by UNV */
873 /* type = -1; */
875 } else if (isnumstart(*p)) {
876 bool is_hex = false;
877 bool is_float = false;
878 bool has_e = false;
879 char c, *r;
882 * A numeric token.
885 if (*p == '$') {
886 p++;
887 is_hex = true;
890 for (;;) {
891 c = *p++;
893 if (!is_hex && (c == 'e' || c == 'E')) {
894 has_e = true;
895 if (*p == '+' || *p == '-') {
896 /* e can only be followed by +/- if it is either a
897 prefixed hex number or a floating-point number */
898 p++;
899 is_float = true;
901 } else if (c == 'H' || c == 'h' || c == 'X' || c == 'x') {
902 is_hex = true;
903 } else if (c == 'P' || c == 'p') {
904 is_float = true;
905 if (*p == '+' || *p == '-')
906 p++;
907 } else if (isnumchar(c) || c == '_')
908 ; /* just advance */
909 else if (c == '.') {
910 /* we need to deal with consequences of the legacy
911 parser, like "1.nolist" being two tokens
912 (TOK_NUMBER, TOK_ID) here; at least give it
913 a shot for now. In the future, we probably need
914 a flex-based scanner with proper pattern matching
915 to do it as well as it can be done. Nothing in
916 the world is going to help the person who wants
917 0x123.p16 interpreted as two tokens, though. */
918 r = p;
919 while (*r == '_')
920 r++;
922 if (nasm_isdigit(*r) || (is_hex && nasm_isxdigit(*r)) ||
923 (!is_hex && (*r == 'e' || *r == 'E')) ||
924 (*r == 'p' || *r == 'P')) {
925 p = r;
926 is_float = true;
927 } else
928 break; /* Terminate the token */
929 } else
930 break;
932 p--; /* Point to first character beyond number */
934 if (has_e && !is_hex) {
935 /* 1e13 is floating-point, but 1e13h is not */
936 is_float = true;
939 type = is_float ? TOK_FLOAT : TOK_NUMBER;
940 } else if (nasm_isspace(*p)) {
941 type = TOK_WHITESPACE;
942 p++;
943 while (*p && nasm_isspace(*p))
944 p++;
946 * Whitespace just before end-of-line is discarded by
947 * pretending it's a comment; whitespace just before a
948 * comment gets lumped into the comment.
950 if (!*p || *p == ';') {
951 type = TOK_COMMENT;
952 while (*p)
953 p++;
955 } else if (*p == ';') {
956 type = TOK_COMMENT;
957 while (*p)
958 p++;
959 } else {
961 * Anything else is an operator of some kind. We check
962 * for all the double-character operators (>>, <<, //,
963 * %%, <=, >=, ==, !=, <>, &&, ||, ^^), but anything
964 * else is a single-character operator.
966 type = TOK_OTHER;
967 if ((p[0] == '>' && p[1] == '>') ||
968 (p[0] == '<' && p[1] == '<') ||
969 (p[0] == '/' && p[1] == '/') ||
970 (p[0] == '<' && p[1] == '=') ||
971 (p[0] == '>' && p[1] == '=') ||
972 (p[0] == '=' && p[1] == '=') ||
973 (p[0] == '!' && p[1] == '=') ||
974 (p[0] == '<' && p[1] == '>') ||
975 (p[0] == '&' && p[1] == '&') ||
976 (p[0] == '|' && p[1] == '|') ||
977 (p[0] == '^' && p[1] == '^')) {
978 p++;
980 p++;
983 /* Handling unterminated string by UNV */
984 /*if (type == -1)
986 *tail = t = new_Token(NULL, TOK_STRING, line, p-line+1);
987 t->text[p-line] = *line;
988 tail = &t->next;
990 else */
991 if (type != TOK_COMMENT) {
992 *tail = t = new_Token(NULL, type, line, p - line);
993 tail = &t->next;
995 line = p;
997 return list;
1001 * this function allocates a new managed block of memory and
1002 * returns a pointer to the block. The managed blocks are
1003 * deleted only all at once by the delete_Blocks function.
1005 static void *new_Block(size_t size)
1007 Blocks *b = &blocks;
1009 /* first, get to the end of the linked list */
1010 while (b->next)
1011 b = b->next;
1012 /* now allocate the requested chunk */
1013 b->chunk = nasm_malloc(size);
1015 /* now allocate a new block for the next request */
1016 b->next = nasm_malloc(sizeof(Blocks));
1017 /* and initialize the contents of the new block */
1018 b->next->next = NULL;
1019 b->next->chunk = NULL;
1020 return b->chunk;
1024 * this function deletes all managed blocks of memory
1026 static void delete_Blocks(void)
1028 Blocks *a, *b = &blocks;
1031 * keep in mind that the first block, pointed to by blocks
1032 * is a static and not dynamically allocated, so we don't
1033 * free it.
1035 while (b) {
1036 if (b->chunk)
1037 nasm_free(b->chunk);
1038 a = b;
1039 b = b->next;
1040 if (a != &blocks)
1041 nasm_free(a);
1046 * this function creates a new Token and passes a pointer to it
1047 * back to the caller. It sets the type and text elements, and
1048 * also the a.mac and next elements to NULL.
1050 static Token *new_Token(Token * next, enum pp_token_type type,
1051 const char *text, int txtlen)
1053 Token *t;
1054 int i;
1056 if (freeTokens == NULL) {
1057 freeTokens = (Token *) new_Block(TOKEN_BLOCKSIZE * sizeof(Token));
1058 for (i = 0; i < TOKEN_BLOCKSIZE - 1; i++)
1059 freeTokens[i].next = &freeTokens[i + 1];
1060 freeTokens[i].next = NULL;
1062 t = freeTokens;
1063 freeTokens = t->next;
1064 t->next = next;
1065 t->a.mac = NULL;
1066 t->type = type;
1067 if (type == TOK_WHITESPACE || text == NULL) {
1068 t->text = NULL;
1069 } else {
1070 if (txtlen == 0)
1071 txtlen = strlen(text);
1072 t->text = nasm_malloc(txtlen+1);
1073 memcpy(t->text, text, txtlen);
1074 t->text[txtlen] = '\0';
1076 return t;
1079 static Token *delete_Token(Token * t)
1081 Token *next = t->next;
1082 nasm_free(t->text);
1083 t->next = freeTokens;
1084 freeTokens = t;
1085 return next;
1089 * Convert a line of tokens back into text.
1090 * If expand_locals is not zero, identifiers of the form "%$*xxx"
1091 * will be transformed into ..@ctxnum.xxx
1093 static char *detoken(Token * tlist, bool expand_locals)
1095 Token *t;
1096 int len;
1097 char *line, *p;
1098 const char *q;
1100 len = 0;
1101 for (t = tlist; t; t = t->next) {
1102 if (t->type == TOK_PREPROC_ID && t->text[1] == '!') {
1103 char *p = getenv(t->text + 2);
1104 nasm_free(t->text);
1105 if (p)
1106 t->text = nasm_strdup(p);
1107 else
1108 t->text = NULL;
1110 /* Expand local macros here and not during preprocessing */
1111 if (expand_locals &&
1112 t->type == TOK_PREPROC_ID && t->text &&
1113 t->text[0] == '%' && t->text[1] == '$') {
1114 Context *ctx = get_ctx(t->text, false);
1115 if (ctx) {
1116 char buffer[40];
1117 char *p, *q = t->text + 2;
1119 q += strspn(q, "$");
1120 snprintf(buffer, sizeof(buffer), "..@%"PRIu32".", ctx->number);
1121 p = nasm_strcat(buffer, q);
1122 nasm_free(t->text);
1123 t->text = p;
1126 if (t->type == TOK_WHITESPACE) {
1127 len++;
1128 } else if (t->text) {
1129 len += strlen(t->text);
1132 p = line = nasm_malloc(len + 1);
1133 for (t = tlist; t; t = t->next) {
1134 if (t->type == TOK_WHITESPACE) {
1135 *p++ = ' ';
1136 } else if (t->text) {
1137 q = t->text;
1138 while (*q)
1139 *p++ = *q++;
1142 *p = '\0';
1143 return line;
1147 * A scanner, suitable for use by the expression evaluator, which
1148 * operates on a line of Tokens. Expects a pointer to a pointer to
1149 * the first token in the line to be passed in as its private_data
1150 * field.
1152 * FIX: This really needs to be unified with stdscan.
1154 static int ppscan(void *private_data, struct tokenval *tokval)
1156 Token **tlineptr = private_data;
1157 Token *tline;
1158 char ourcopy[MAX_KEYWORD+1], *p, *r, *s;
1160 do {
1161 tline = *tlineptr;
1162 *tlineptr = tline ? tline->next : NULL;
1164 while (tline && (tline->type == TOK_WHITESPACE ||
1165 tline->type == TOK_COMMENT));
1167 if (!tline)
1168 return tokval->t_type = TOKEN_EOS;
1170 tokval->t_charptr = tline->text;
1172 if (tline->text[0] == '$' && !tline->text[1])
1173 return tokval->t_type = TOKEN_HERE;
1174 if (tline->text[0] == '$' && tline->text[1] == '$' && !tline->text[2])
1175 return tokval->t_type = TOKEN_BASE;
1177 if (tline->type == TOK_ID) {
1178 p = tokval->t_charptr = tline->text;
1179 if (p[0] == '$') {
1180 tokval->t_charptr++;
1181 return tokval->t_type = TOKEN_ID;
1184 for (r = p, s = ourcopy; *r; r++) {
1185 if (r >= p+MAX_KEYWORD)
1186 return tokval->t_type = TOKEN_ID; /* Not a keyword */
1187 *s++ = nasm_tolower(*r);
1189 *s = '\0';
1190 /* right, so we have an identifier sitting in temp storage. now,
1191 * is it actually a register or instruction name, or what? */
1192 return nasm_token_hash(ourcopy, tokval);
1195 if (tline->type == TOK_NUMBER) {
1196 bool rn_error;
1197 tokval->t_integer = readnum(tline->text, &rn_error);
1198 tokval->t_charptr = tline->text;
1199 if (rn_error)
1200 return tokval->t_type = TOKEN_ERRNUM;
1201 else
1202 return tokval->t_type = TOKEN_NUM;
1205 if (tline->type == TOK_FLOAT) {
1206 return tokval->t_type = TOKEN_FLOAT;
1209 if (tline->type == TOK_STRING) {
1210 char bq, *ep;
1212 bq = tline->text[0];
1213 tokval->t_charptr = tline->text;
1214 tokval->t_inttwo = nasm_unquote(tline->text, &ep);
1216 if (ep[0] != bq || ep[1] != '\0')
1217 return tokval->t_type = TOKEN_ERRSTR;
1218 else
1219 return tokval->t_type = TOKEN_STR;
1222 if (tline->type == TOK_OTHER) {
1223 if (!strcmp(tline->text, "<<"))
1224 return tokval->t_type = TOKEN_SHL;
1225 if (!strcmp(tline->text, ">>"))
1226 return tokval->t_type = TOKEN_SHR;
1227 if (!strcmp(tline->text, "//"))
1228 return tokval->t_type = TOKEN_SDIV;
1229 if (!strcmp(tline->text, "%%"))
1230 return tokval->t_type = TOKEN_SMOD;
1231 if (!strcmp(tline->text, "=="))
1232 return tokval->t_type = TOKEN_EQ;
1233 if (!strcmp(tline->text, "<>"))
1234 return tokval->t_type = TOKEN_NE;
1235 if (!strcmp(tline->text, "!="))
1236 return tokval->t_type = TOKEN_NE;
1237 if (!strcmp(tline->text, "<="))
1238 return tokval->t_type = TOKEN_LE;
1239 if (!strcmp(tline->text, ">="))
1240 return tokval->t_type = TOKEN_GE;
1241 if (!strcmp(tline->text, "&&"))
1242 return tokval->t_type = TOKEN_DBL_AND;
1243 if (!strcmp(tline->text, "^^"))
1244 return tokval->t_type = TOKEN_DBL_XOR;
1245 if (!strcmp(tline->text, "||"))
1246 return tokval->t_type = TOKEN_DBL_OR;
1250 * We have no other options: just return the first character of
1251 * the token text.
1253 return tokval->t_type = tline->text[0];
1257 * Compare a string to the name of an existing macro; this is a
1258 * simple wrapper which calls either strcmp or nasm_stricmp
1259 * depending on the value of the `casesense' parameter.
1261 static int mstrcmp(const char *p, const char *q, bool casesense)
1263 return casesense ? strcmp(p, q) : nasm_stricmp(p, q);
1267 * Compare a string to the name of an existing macro; this is a
1268 * simple wrapper which calls either strcmp or nasm_stricmp
1269 * depending on the value of the `casesense' parameter.
1271 static int mmemcmp(const char *p, const char *q, size_t l, bool casesense)
1273 return casesense ? memcmp(p, q, l) : nasm_memicmp(p, q, l);
1277 * Return the Context structure associated with a %$ token. Return
1278 * NULL, having _already_ reported an error condition, if the
1279 * context stack isn't deep enough for the supplied number of $
1280 * signs.
1281 * If all_contexts == true, contexts that enclose current are
1282 * also scanned for such smacro, until it is found; if not -
1283 * only the context that directly results from the number of $'s
1284 * in variable's name.
1286 static Context *get_ctx(const char *name, bool all_contexts)
1288 Context *ctx;
1289 SMacro *m;
1290 int i;
1292 if (!name || name[0] != '%' || name[1] != '$')
1293 return NULL;
1295 if (!cstk) {
1296 error(ERR_NONFATAL, "`%s': context stack is empty", name);
1297 return NULL;
1300 for (i = strspn(name + 2, "$"), ctx = cstk; (i > 0) && ctx; i--) {
1301 ctx = ctx->next;
1302 /* i--; Lino - 02/25/02 */
1304 if (!ctx) {
1305 error(ERR_NONFATAL, "`%s': context stack is only"
1306 " %d level%s deep", name, i - 1, (i == 2 ? "" : "s"));
1307 return NULL;
1309 if (!all_contexts)
1310 return ctx;
1312 do {
1313 /* Search for this smacro in found context */
1314 m = hash_findix(&ctx->localmac, name);
1315 while (m) {
1316 if (!mstrcmp(m->name, name, m->casesense))
1317 return ctx;
1318 m = m->next;
1320 ctx = ctx->next;
1322 while (ctx);
1323 return NULL;
1327 * Check to see if a file is already in a string list
1329 static bool in_list(const StrList *list, const char *str)
1331 while (list) {
1332 if (!strcmp(list->str, str))
1333 return true;
1334 list = list->next;
1336 return false;
1340 * Open an include file. This routine must always return a valid
1341 * file pointer if it returns - it's responsible for throwing an
1342 * ERR_FATAL and bombing out completely if not. It should also try
1343 * the include path one by one until it finds the file or reaches
1344 * the end of the path.
1346 static FILE *inc_fopen(const char *file, StrList **dhead, StrList ***dtail,
1347 bool missing_ok)
1349 FILE *fp;
1350 char *prefix = "";
1351 IncPath *ip = ipath;
1352 int len = strlen(file);
1353 size_t prefix_len = 0;
1354 StrList *sl;
1356 while (1) {
1357 sl = nasm_malloc(prefix_len+len+1+sizeof sl->next);
1358 memcpy(sl->str, prefix, prefix_len);
1359 memcpy(sl->str+prefix_len, file, len+1);
1360 fp = fopen(sl->str, "r");
1361 if (fp && dhead && !in_list(*dhead, sl->str)) {
1362 sl->next = NULL;
1363 **dtail = sl;
1364 *dtail = &sl->next;
1365 } else {
1366 nasm_free(sl);
1368 if (fp)
1369 return fp;
1370 if (!ip) {
1371 if (!missing_ok)
1372 break;
1373 prefix = NULL;
1374 } else {
1375 prefix = ip->path;
1376 ip = ip->next;
1378 if (prefix) {
1379 prefix_len = strlen(prefix);
1380 } else {
1381 /* -MG given and file not found */
1382 if (dhead && !in_list(*dhead, file)) {
1383 sl = nasm_malloc(len+1+sizeof sl->next);
1384 sl->next = NULL;
1385 strcpy(sl->str, file);
1386 **dtail = sl;
1387 *dtail = &sl->next;
1389 return NULL;
1393 error(ERR_FATAL, "unable to open include file `%s'", file);
1394 return NULL; /* never reached - placate compilers */
1398 * Determine if we should warn on defining a single-line macro of
1399 * name `name', with `nparam' parameters. If nparam is 0 or -1, will
1400 * return true if _any_ single-line macro of that name is defined.
1401 * Otherwise, will return true if a single-line macro with either
1402 * `nparam' or no parameters is defined.
1404 * If a macro with precisely the right number of parameters is
1405 * defined, or nparam is -1, the address of the definition structure
1406 * will be returned in `defn'; otherwise NULL will be returned. If `defn'
1407 * is NULL, no action will be taken regarding its contents, and no
1408 * error will occur.
1410 * Note that this is also called with nparam zero to resolve
1411 * `ifdef'.
1413 * If you already know which context macro belongs to, you can pass
1414 * the context pointer as first parameter; if you won't but name begins
1415 * with %$ the context will be automatically computed. If all_contexts
1416 * is true, macro will be searched in outer contexts as well.
1418 static bool
1419 smacro_defined(Context * ctx, const char *name, int nparam, SMacro ** defn,
1420 bool nocase)
1422 struct hash_table *smtbl;
1423 SMacro *m;
1425 if (ctx) {
1426 smtbl = &ctx->localmac;
1427 } else if (name[0] == '%' && name[1] == '$') {
1428 if (cstk)
1429 ctx = get_ctx(name, false);
1430 if (!ctx)
1431 return false; /* got to return _something_ */
1432 smtbl = &ctx->localmac;
1433 } else {
1434 smtbl = &smacros;
1436 m = (SMacro *) hash_findix(smtbl, name);
1438 while (m) {
1439 if (!mstrcmp(m->name, name, m->casesense && nocase) &&
1440 (nparam <= 0 || m->nparam == 0 || nparam == (int) m->nparam)) {
1441 if (defn) {
1442 if (nparam == (int) m->nparam || nparam == -1)
1443 *defn = m;
1444 else
1445 *defn = NULL;
1447 return true;
1449 m = m->next;
1452 return false;
1456 * Count and mark off the parameters in a multi-line macro call.
1457 * This is called both from within the multi-line macro expansion
1458 * code, and also to mark off the default parameters when provided
1459 * in a %macro definition line.
1461 static void count_mmac_params(Token * t, int *nparam, Token *** params)
1463 int paramsize, brace;
1465 *nparam = paramsize = 0;
1466 *params = NULL;
1467 while (t) {
1468 /* +1: we need space for the final NULL */
1469 if (*nparam+1 >= paramsize) {
1470 paramsize += PARAM_DELTA;
1471 *params = nasm_realloc(*params, sizeof(**params) * paramsize);
1473 skip_white_(t);
1474 brace = false;
1475 if (tok_is_(t, "{"))
1476 brace = true;
1477 (*params)[(*nparam)++] = t;
1478 while (tok_isnt_(t, brace ? "}" : ","))
1479 t = t->next;
1480 if (t) { /* got a comma/brace */
1481 t = t->next;
1482 if (brace) {
1484 * Now we've found the closing brace, look further
1485 * for the comma.
1487 skip_white_(t);
1488 if (tok_isnt_(t, ",")) {
1489 error(ERR_NONFATAL,
1490 "braces do not enclose all of macro parameter");
1491 while (tok_isnt_(t, ","))
1492 t = t->next;
1494 if (t)
1495 t = t->next; /* eat the comma */
1502 * Determine whether one of the various `if' conditions is true or
1503 * not.
1505 * We must free the tline we get passed.
1507 static bool if_condition(Token * tline, enum preproc_token ct)
1509 enum pp_conditional i = PP_COND(ct);
1510 bool j;
1511 Token *t, *tt, **tptr, *origline;
1512 struct tokenval tokval;
1513 expr *evalresult;
1514 enum pp_token_type needtype;
1516 origline = tline;
1518 switch (i) {
1519 case PPC_IFCTX:
1520 j = false; /* have we matched yet? */
1521 while (true) {
1522 skip_white_(tline);
1523 if (!tline)
1524 break;
1525 if (tline->type != TOK_ID) {
1526 error(ERR_NONFATAL,
1527 "`%s' expects context identifiers", pp_directives[ct]);
1528 free_tlist(origline);
1529 return -1;
1531 if (cstk && cstk->name && !nasm_stricmp(tline->text, cstk->name))
1532 j = true;
1533 tline = tline->next;
1535 break;
1537 case PPC_IFDEF:
1538 j = false; /* have we matched yet? */
1539 while (tline) {
1540 skip_white_(tline);
1541 if (!tline || (tline->type != TOK_ID &&
1542 (tline->type != TOK_PREPROC_ID ||
1543 tline->text[1] != '$'))) {
1544 error(ERR_NONFATAL,
1545 "`%s' expects macro identifiers", pp_directives[ct]);
1546 goto fail;
1548 if (smacro_defined(NULL, tline->text, 0, NULL, true))
1549 j = true;
1550 tline = tline->next;
1552 break;
1554 case PPC_IFIDN:
1555 case PPC_IFIDNI:
1556 tline = expand_smacro(tline);
1557 t = tt = tline;
1558 while (tok_isnt_(tt, ","))
1559 tt = tt->next;
1560 if (!tt) {
1561 error(ERR_NONFATAL,
1562 "`%s' expects two comma-separated arguments",
1563 pp_directives[ct]);
1564 goto fail;
1566 tt = tt->next;
1567 j = true; /* assume equality unless proved not */
1568 while ((t->type != TOK_OTHER || strcmp(t->text, ",")) && tt) {
1569 if (tt->type == TOK_OTHER && !strcmp(tt->text, ",")) {
1570 error(ERR_NONFATAL, "`%s': more than one comma on line",
1571 pp_directives[ct]);
1572 goto fail;
1574 if (t->type == TOK_WHITESPACE) {
1575 t = t->next;
1576 continue;
1578 if (tt->type == TOK_WHITESPACE) {
1579 tt = tt->next;
1580 continue;
1582 if (tt->type != t->type) {
1583 j = false; /* found mismatching tokens */
1584 break;
1586 /* When comparing strings, need to unquote them first */
1587 if (t->type == TOK_STRING) {
1588 size_t l1 = nasm_unquote(t->text, NULL);
1589 size_t l2 = nasm_unquote(tt->text, NULL);
1591 if (l1 != l2) {
1592 j = false;
1593 break;
1595 if (mmemcmp(t->text, tt->text, l1, i == PPC_IFIDN)) {
1596 j = false;
1597 break;
1599 } else if (mstrcmp(tt->text, t->text, i == PPC_IFIDN) != 0) {
1600 j = false; /* found mismatching tokens */
1601 break;
1604 t = t->next;
1605 tt = tt->next;
1607 if ((t->type != TOK_OTHER || strcmp(t->text, ",")) || tt)
1608 j = false; /* trailing gunk on one end or other */
1609 break;
1611 case PPC_IFMACRO:
1613 bool found = false;
1614 MMacro searching, *mmac;
1616 skip_white_(tline);
1617 tline = expand_id(tline);
1618 if (!tok_type_(tline, TOK_ID)) {
1619 error(ERR_NONFATAL,
1620 "`%s' expects a macro name", pp_directives[ct]);
1621 goto fail;
1623 searching.name = nasm_strdup(tline->text);
1624 searching.casesense = true;
1625 searching.plus = false;
1626 searching.nolist = false;
1627 searching.in_progress = 0;
1628 searching.rep_nest = NULL;
1629 searching.nparam_min = 0;
1630 searching.nparam_max = INT_MAX;
1631 tline = expand_smacro(tline->next);
1632 skip_white_(tline);
1633 if (!tline) {
1634 } else if (!tok_type_(tline, TOK_NUMBER)) {
1635 error(ERR_NONFATAL,
1636 "`%s' expects a parameter count or nothing",
1637 pp_directives[ct]);
1638 } else {
1639 searching.nparam_min = searching.nparam_max =
1640 readnum(tline->text, &j);
1641 if (j)
1642 error(ERR_NONFATAL,
1643 "unable to parse parameter count `%s'",
1644 tline->text);
1646 if (tline && tok_is_(tline->next, "-")) {
1647 tline = tline->next->next;
1648 if (tok_is_(tline, "*"))
1649 searching.nparam_max = INT_MAX;
1650 else if (!tok_type_(tline, TOK_NUMBER))
1651 error(ERR_NONFATAL,
1652 "`%s' expects a parameter count after `-'",
1653 pp_directives[ct]);
1654 else {
1655 searching.nparam_max = readnum(tline->text, &j);
1656 if (j)
1657 error(ERR_NONFATAL,
1658 "unable to parse parameter count `%s'",
1659 tline->text);
1660 if (searching.nparam_min > searching.nparam_max)
1661 error(ERR_NONFATAL,
1662 "minimum parameter count exceeds maximum");
1665 if (tline && tok_is_(tline->next, "+")) {
1666 tline = tline->next;
1667 searching.plus = true;
1669 mmac = (MMacro *) hash_findix(&mmacros, searching.name);
1670 while (mmac) {
1671 if (!strcmp(mmac->name, searching.name) &&
1672 (mmac->nparam_min <= searching.nparam_max
1673 || searching.plus)
1674 && (searching.nparam_min <= mmac->nparam_max
1675 || mmac->plus)) {
1676 found = true;
1677 break;
1679 mmac = mmac->next;
1681 if(tline && tline->next)
1682 error(ERR_WARNING|ERR_PASS1,
1683 "trailing garbage after %%ifmacro ignored");
1684 nasm_free(searching.name);
1685 j = found;
1686 break;
1689 case PPC_IFID:
1690 needtype = TOK_ID;
1691 goto iftype;
1692 case PPC_IFNUM:
1693 needtype = TOK_NUMBER;
1694 goto iftype;
1695 case PPC_IFSTR:
1696 needtype = TOK_STRING;
1697 goto iftype;
1699 iftype:
1700 t = tline = expand_smacro(tline);
1702 while (tok_type_(t, TOK_WHITESPACE) ||
1703 (needtype == TOK_NUMBER &&
1704 tok_type_(t, TOK_OTHER) &&
1705 (t->text[0] == '-' || t->text[0] == '+') &&
1706 !t->text[1]))
1707 t = t->next;
1709 j = tok_type_(t, needtype);
1710 break;
1712 case PPC_IFTOKEN:
1713 t = tline = expand_smacro(tline);
1714 while (tok_type_(t, TOK_WHITESPACE))
1715 t = t->next;
1717 j = false;
1718 if (t) {
1719 t = t->next; /* Skip the actual token */
1720 while (tok_type_(t, TOK_WHITESPACE))
1721 t = t->next;
1722 j = !t; /* Should be nothing left */
1724 break;
1726 case PPC_IFEMPTY:
1727 t = tline = expand_smacro(tline);
1728 while (tok_type_(t, TOK_WHITESPACE))
1729 t = t->next;
1731 j = !t; /* Should be empty */
1732 break;
1734 case PPC_IF:
1735 t = tline = expand_smacro(tline);
1736 tptr = &t;
1737 tokval.t_type = TOKEN_INVALID;
1738 evalresult = evaluate(ppscan, tptr, &tokval,
1739 NULL, pass | CRITICAL, error, NULL);
1740 if (!evalresult)
1741 return -1;
1742 if (tokval.t_type)
1743 error(ERR_WARNING|ERR_PASS1,
1744 "trailing garbage after expression ignored");
1745 if (!is_simple(evalresult)) {
1746 error(ERR_NONFATAL,
1747 "non-constant value given to `%s'", pp_directives[ct]);
1748 goto fail;
1750 j = reloc_value(evalresult) != 0;
1751 break;
1753 default:
1754 error(ERR_FATAL,
1755 "preprocessor directive `%s' not yet implemented",
1756 pp_directives[ct]);
1757 goto fail;
1760 free_tlist(origline);
1761 return j ^ PP_NEGATIVE(ct);
1763 fail:
1764 free_tlist(origline);
1765 return -1;
1769 * Common code for defining an smacro
1771 static bool define_smacro(Context *ctx, char *mname, bool casesense,
1772 int nparam, Token *expansion)
1774 SMacro *smac, **smhead;
1775 struct hash_table *smtbl;
1777 if (smacro_defined(ctx, mname, nparam, &smac, casesense)) {
1778 if (!smac) {
1779 error(ERR_WARNING|ERR_PASS1,
1780 "single-line macro `%s' defined both with and"
1781 " without parameters", mname);
1783 /* Some instances of the old code considered this a failure,
1784 some others didn't. What is the right thing to do here? */
1785 free_tlist(expansion);
1786 return false; /* Failure */
1787 } else {
1789 * We're redefining, so we have to take over an
1790 * existing SMacro structure. This means freeing
1791 * what was already in it.
1793 nasm_free(smac->name);
1794 free_tlist(smac->expansion);
1796 } else {
1797 smtbl = ctx ? &ctx->localmac : &smacros;
1798 smhead = (SMacro **) hash_findi_add(smtbl, mname);
1799 smac = nasm_malloc(sizeof(SMacro));
1800 smac->next = *smhead;
1801 *smhead = smac;
1803 smac->name = nasm_strdup(mname);
1804 smac->casesense = casesense;
1805 smac->nparam = nparam;
1806 smac->expansion = expansion;
1807 smac->in_progress = false;
1808 return true; /* Success */
1812 * Undefine an smacro
1814 static void undef_smacro(Context *ctx, const char *mname)
1816 SMacro **smhead, *s, **sp;
1817 struct hash_table *smtbl;
1819 smtbl = ctx ? &ctx->localmac : &smacros;
1820 smhead = (SMacro **)hash_findi(smtbl, mname, NULL);
1822 if (smhead) {
1824 * We now have a macro name... go hunt for it.
1826 sp = smhead;
1827 while ((s = *sp) != NULL) {
1828 if (!mstrcmp(s->name, mname, s->casesense)) {
1829 *sp = s->next;
1830 nasm_free(s->name);
1831 free_tlist(s->expansion);
1832 nasm_free(s);
1833 } else {
1834 sp = &s->next;
1841 * Parse a mmacro specification.
1843 static bool parse_mmacro_spec(Token *tline, MMacro *def, const char *directive)
1845 bool err;
1847 tline = tline->next;
1848 skip_white_(tline);
1849 tline = expand_id(tline);
1850 if (!tok_type_(tline, TOK_ID)) {
1851 error(ERR_NONFATAL, "`%s' expects a macro name", directive);
1852 return false;
1855 def->name = nasm_strdup(tline->text);
1856 def->plus = false;
1857 def->nolist = false;
1858 def->in_progress = 0;
1859 def->rep_nest = NULL;
1860 def->nparam_min = 0;
1861 def->nparam_max = 0;
1863 tline = expand_smacro(tline->next);
1864 skip_white_(tline);
1865 if (!tok_type_(tline, TOK_NUMBER)) {
1866 error(ERR_NONFATAL, "`%s' expects a parameter count", directive);
1867 } else {
1868 def->nparam_min = def->nparam_max =
1869 readnum(tline->text, &err);
1870 if (err)
1871 error(ERR_NONFATAL,
1872 "unable to parse parameter count `%s'", tline->text);
1874 if (tline && tok_is_(tline->next, "-")) {
1875 tline = tline->next->next;
1876 if (tok_is_(tline, "*")) {
1877 def->nparam_max = INT_MAX;
1878 } else if (!tok_type_(tline, TOK_NUMBER)) {
1879 error(ERR_NONFATAL,
1880 "`%s' expects a parameter count after `-'", directive);
1881 } else {
1882 def->nparam_max = readnum(tline->text, &err);
1883 if (err) {
1884 error(ERR_NONFATAL, "unable to parse parameter count `%s'",
1885 tline->text);
1887 if (def->nparam_min > def->nparam_max) {
1888 error(ERR_NONFATAL, "minimum parameter count exceeds maximum");
1892 if (tline && tok_is_(tline->next, "+")) {
1893 tline = tline->next;
1894 def->plus = true;
1896 if (tline && tok_type_(tline->next, TOK_ID) &&
1897 !nasm_stricmp(tline->next->text, ".nolist")) {
1898 tline = tline->next;
1899 def->nolist = true;
1903 * Handle default parameters.
1905 if (tline && tline->next) {
1906 def->dlist = tline->next;
1907 tline->next = NULL;
1908 count_mmac_params(def->dlist, &def->ndefs, &def->defaults);
1909 } else {
1910 def->dlist = NULL;
1911 def->defaults = NULL;
1913 def->expansion = NULL;
1915 if(def->defaults &&
1916 def->ndefs > def->nparam_max - def->nparam_min &&
1917 !def->plus)
1918 error(ERR_WARNING|ERR_PASS1|ERR_WARN_MDP,
1919 "too many default macro parameters");
1921 return true;
1926 * Decode a size directive
1928 static int parse_size(const char *str) {
1929 static const char *size_names[] =
1930 { "byte", "dword", "oword", "qword", "tword", "word", "yword" };
1931 static const int sizes[] =
1932 { 0, 1, 4, 16, 8, 10, 2, 32 };
1934 return sizes[bsii(str, size_names, elements(size_names))+1];
1938 * find and process preprocessor directive in passed line
1939 * Find out if a line contains a preprocessor directive, and deal
1940 * with it if so.
1942 * If a directive _is_ found, it is the responsibility of this routine
1943 * (and not the caller) to free_tlist() the line.
1945 * @param tline a pointer to the current tokeninzed line linked list
1946 * @return DIRECTIVE_FOUND or NO_DIRECTIVE_FOUND
1949 static int do_directive(Token * tline)
1951 enum preproc_token i;
1952 int j;
1953 bool err;
1954 int nparam;
1955 bool nolist;
1956 bool casesense;
1957 int k, m;
1958 int offset;
1959 char *p, *pp, *mname;
1960 Include *inc;
1961 Context *ctx;
1962 Cond *cond;
1963 MMacro *mmac, **mmhead;
1964 Token *t, *tt, *param_start, *macro_start, *last, **tptr, *origline;
1965 Line *l;
1966 struct tokenval tokval;
1967 expr *evalresult;
1968 MMacro *tmp_defining; /* Used when manipulating rep_nest */
1969 int64_t count;
1970 size_t len;
1971 int severity;
1973 origline = tline;
1975 skip_white_(tline);
1976 if (!tline || !tok_type_(tline, TOK_PREPROC_ID) ||
1977 (tline->text[1] == '%' || tline->text[1] == '$'
1978 || tline->text[1] == '!'))
1979 return NO_DIRECTIVE_FOUND;
1981 i = pp_token_hash(tline->text);
1984 * If we're in a non-emitting branch of a condition construct,
1985 * or walking to the end of an already terminated %rep block,
1986 * we should ignore all directives except for condition
1987 * directives.
1989 if (((istk->conds && !emitting(istk->conds->state)) ||
1990 (istk->mstk && !istk->mstk->in_progress)) && !is_condition(i)) {
1991 return NO_DIRECTIVE_FOUND;
1995 * If we're defining a macro or reading a %rep block, we should
1996 * ignore all directives except for %macro/%imacro (which nest),
1997 * %endm/%endmacro, and (only if we're in a %rep block) %endrep.
1998 * If we're in a %rep block, another %rep nests, so should be let through.
2000 if (defining && i != PP_MACRO && i != PP_IMACRO &&
2001 i != PP_ENDMACRO && i != PP_ENDM &&
2002 (defining->name || (i != PP_ENDREP && i != PP_REP))) {
2003 return NO_DIRECTIVE_FOUND;
2006 if (defining) {
2007 if (i == PP_MACRO || i == PP_IMACRO) {
2008 nested_mac_count++;
2009 return NO_DIRECTIVE_FOUND;
2010 } else if (nested_mac_count > 0) {
2011 if (i == PP_ENDMACRO) {
2012 nested_mac_count--;
2013 return NO_DIRECTIVE_FOUND;
2016 if (!defining->name) {
2017 if (i == PP_REP) {
2018 nested_rep_count++;
2019 return NO_DIRECTIVE_FOUND;
2020 } else if (nested_rep_count > 0) {
2021 if (i == PP_ENDREP) {
2022 nested_rep_count--;
2023 return NO_DIRECTIVE_FOUND;
2029 switch (i) {
2030 case PP_INVALID:
2031 error(ERR_NONFATAL, "unknown preprocessor directive `%s'",
2032 tline->text);
2033 return NO_DIRECTIVE_FOUND; /* didn't get it */
2035 case PP_STACKSIZE:
2036 /* Directive to tell NASM what the default stack size is. The
2037 * default is for a 16-bit stack, and this can be overriden with
2038 * %stacksize large.
2039 * the following form:
2041 * ARG arg1:WORD, arg2:DWORD, arg4:QWORD
2043 tline = tline->next;
2044 if (tline && tline->type == TOK_WHITESPACE)
2045 tline = tline->next;
2046 if (!tline || tline->type != TOK_ID) {
2047 error(ERR_NONFATAL, "`%%stacksize' missing size parameter");
2048 free_tlist(origline);
2049 return DIRECTIVE_FOUND;
2051 if (nasm_stricmp(tline->text, "flat") == 0) {
2052 /* All subsequent ARG directives are for a 32-bit stack */
2053 StackSize = 4;
2054 StackPointer = "ebp";
2055 ArgOffset = 8;
2056 LocalOffset = 0;
2057 } else if (nasm_stricmp(tline->text, "flat64") == 0) {
2058 /* All subsequent ARG directives are for a 64-bit stack */
2059 StackSize = 8;
2060 StackPointer = "rbp";
2061 ArgOffset = 8;
2062 LocalOffset = 0;
2063 } else if (nasm_stricmp(tline->text, "large") == 0) {
2064 /* All subsequent ARG directives are for a 16-bit stack,
2065 * far function call.
2067 StackSize = 2;
2068 StackPointer = "bp";
2069 ArgOffset = 4;
2070 LocalOffset = 0;
2071 } else if (nasm_stricmp(tline->text, "small") == 0) {
2072 /* All subsequent ARG directives are for a 16-bit stack,
2073 * far function call. We don't support near functions.
2075 StackSize = 2;
2076 StackPointer = "bp";
2077 ArgOffset = 6;
2078 LocalOffset = 0;
2079 } else {
2080 error(ERR_NONFATAL, "`%%stacksize' invalid size type");
2081 free_tlist(origline);
2082 return DIRECTIVE_FOUND;
2084 free_tlist(origline);
2085 return DIRECTIVE_FOUND;
2087 case PP_ARG:
2088 /* TASM like ARG directive to define arguments to functions, in
2089 * the following form:
2091 * ARG arg1:WORD, arg2:DWORD, arg4:QWORD
2093 offset = ArgOffset;
2094 do {
2095 char *arg, directive[256];
2096 int size = StackSize;
2098 /* Find the argument name */
2099 tline = tline->next;
2100 if (tline && tline->type == TOK_WHITESPACE)
2101 tline = tline->next;
2102 if (!tline || tline->type != TOK_ID) {
2103 error(ERR_NONFATAL, "`%%arg' missing argument parameter");
2104 free_tlist(origline);
2105 return DIRECTIVE_FOUND;
2107 arg = tline->text;
2109 /* Find the argument size type */
2110 tline = tline->next;
2111 if (!tline || tline->type != TOK_OTHER
2112 || tline->text[0] != ':') {
2113 error(ERR_NONFATAL,
2114 "Syntax error processing `%%arg' directive");
2115 free_tlist(origline);
2116 return DIRECTIVE_FOUND;
2118 tline = tline->next;
2119 if (!tline || tline->type != TOK_ID) {
2120 error(ERR_NONFATAL, "`%%arg' missing size type parameter");
2121 free_tlist(origline);
2122 return DIRECTIVE_FOUND;
2125 /* Allow macro expansion of type parameter */
2126 tt = tokenize(tline->text);
2127 tt = expand_smacro(tt);
2128 size = parse_size(tt->text);
2129 if (!size) {
2130 error(ERR_NONFATAL,
2131 "Invalid size type for `%%arg' missing directive");
2132 free_tlist(tt);
2133 free_tlist(origline);
2134 return DIRECTIVE_FOUND;
2136 free_tlist(tt);
2138 /* Round up to even stack slots */
2139 size = (size+StackSize-1) & ~(StackSize-1);
2141 /* Now define the macro for the argument */
2142 snprintf(directive, sizeof(directive), "%%define %s (%s+%d)",
2143 arg, StackPointer, offset);
2144 do_directive(tokenize(directive));
2145 offset += size;
2147 /* Move to the next argument in the list */
2148 tline = tline->next;
2149 if (tline && tline->type == TOK_WHITESPACE)
2150 tline = tline->next;
2151 } while (tline && tline->type == TOK_OTHER && tline->text[0] == ',');
2152 ArgOffset = offset;
2153 free_tlist(origline);
2154 return DIRECTIVE_FOUND;
2156 case PP_LOCAL:
2157 /* TASM like LOCAL directive to define local variables for a
2158 * function, in the following form:
2160 * LOCAL local1:WORD, local2:DWORD, local4:QWORD = LocalSize
2162 * The '= LocalSize' at the end is ignored by NASM, but is
2163 * required by TASM to define the local parameter size (and used
2164 * by the TASM macro package).
2166 offset = LocalOffset;
2167 do {
2168 char *local, directive[256];
2169 int size = StackSize;
2171 /* Find the argument name */
2172 tline = tline->next;
2173 if (tline && tline->type == TOK_WHITESPACE)
2174 tline = tline->next;
2175 if (!tline || tline->type != TOK_ID) {
2176 error(ERR_NONFATAL,
2177 "`%%local' missing argument parameter");
2178 free_tlist(origline);
2179 return DIRECTIVE_FOUND;
2181 local = tline->text;
2183 /* Find the argument size type */
2184 tline = tline->next;
2185 if (!tline || tline->type != TOK_OTHER
2186 || tline->text[0] != ':') {
2187 error(ERR_NONFATAL,
2188 "Syntax error processing `%%local' directive");
2189 free_tlist(origline);
2190 return DIRECTIVE_FOUND;
2192 tline = tline->next;
2193 if (!tline || tline->type != TOK_ID) {
2194 error(ERR_NONFATAL,
2195 "`%%local' missing size type parameter");
2196 free_tlist(origline);
2197 return DIRECTIVE_FOUND;
2200 /* Allow macro expansion of type parameter */
2201 tt = tokenize(tline->text);
2202 tt = expand_smacro(tt);
2203 size = parse_size(tt->text);
2204 if (!size) {
2205 error(ERR_NONFATAL,
2206 "Invalid size type for `%%local' missing directive");
2207 free_tlist(tt);
2208 free_tlist(origline);
2209 return DIRECTIVE_FOUND;
2211 free_tlist(tt);
2213 /* Round up to even stack slots */
2214 size = (size+StackSize-1) & ~(StackSize-1);
2216 offset += size; /* Negative offset, increment before */
2218 /* Now define the macro for the argument */
2219 snprintf(directive, sizeof(directive), "%%define %s (%s-%d)",
2220 local, StackPointer, offset);
2221 do_directive(tokenize(directive));
2223 /* Now define the assign to setup the enter_c macro correctly */
2224 snprintf(directive, sizeof(directive),
2225 "%%assign %%$localsize %%$localsize+%d", size);
2226 do_directive(tokenize(directive));
2228 /* Move to the next argument in the list */
2229 tline = tline->next;
2230 if (tline && tline->type == TOK_WHITESPACE)
2231 tline = tline->next;
2232 } while (tline && tline->type == TOK_OTHER && tline->text[0] == ',');
2233 LocalOffset = offset;
2234 free_tlist(origline);
2235 return DIRECTIVE_FOUND;
2237 case PP_CLEAR:
2238 if (tline->next)
2239 error(ERR_WARNING|ERR_PASS1,
2240 "trailing garbage after `%%clear' ignored");
2241 free_macros();
2242 init_macros();
2243 free_tlist(origline);
2244 return DIRECTIVE_FOUND;
2246 case PP_DEPEND:
2247 t = tline->next = expand_smacro(tline->next);
2248 skip_white_(t);
2249 if (!t || (t->type != TOK_STRING &&
2250 t->type != TOK_INTERNAL_STRING)) {
2251 error(ERR_NONFATAL, "`%%depend' expects a file name");
2252 free_tlist(origline);
2253 return DIRECTIVE_FOUND; /* but we did _something_ */
2255 if (t->next)
2256 error(ERR_WARNING|ERR_PASS1,
2257 "trailing garbage after `%%depend' ignored");
2258 p = t->text;
2259 if (t->type != TOK_INTERNAL_STRING)
2260 nasm_unquote(p, NULL);
2261 if (dephead && !in_list(*dephead, p)) {
2262 StrList *sl = nasm_malloc(strlen(p)+1+sizeof sl->next);
2263 sl->next = NULL;
2264 strcpy(sl->str, p);
2265 *deptail = sl;
2266 deptail = &sl->next;
2268 free_tlist(origline);
2269 return DIRECTIVE_FOUND;
2271 case PP_INCLUDE:
2272 t = tline->next = expand_smacro(tline->next);
2273 skip_white_(t);
2275 if (!t || (t->type != TOK_STRING &&
2276 t->type != TOK_INTERNAL_STRING)) {
2277 error(ERR_NONFATAL, "`%%include' expects a file name");
2278 free_tlist(origline);
2279 return DIRECTIVE_FOUND; /* but we did _something_ */
2281 if (t->next)
2282 error(ERR_WARNING|ERR_PASS1,
2283 "trailing garbage after `%%include' ignored");
2284 p = t->text;
2285 if (t->type != TOK_INTERNAL_STRING)
2286 nasm_unquote(p, NULL);
2287 inc = nasm_malloc(sizeof(Include));
2288 inc->next = istk;
2289 inc->conds = NULL;
2290 inc->fp = inc_fopen(p, dephead, &deptail, pass == 0);
2291 if (!inc->fp) {
2292 /* -MG given but file not found */
2293 nasm_free(inc);
2294 } else {
2295 inc->fname = src_set_fname(nasm_strdup(p));
2296 inc->lineno = src_set_linnum(0);
2297 inc->lineinc = 1;
2298 inc->expansion = NULL;
2299 inc->mstk = NULL;
2300 istk = inc;
2301 list->uplevel(LIST_INCLUDE);
2303 free_tlist(origline);
2304 return DIRECTIVE_FOUND;
2306 case PP_USE:
2308 static macros_t *use_pkg;
2309 const char *pkg_macro;
2311 tline = tline->next;
2312 skip_white_(tline);
2313 tline = expand_id(tline);
2315 if (!tline || (tline->type != TOK_STRING &&
2316 tline->type != TOK_INTERNAL_STRING &&
2317 tline->type != TOK_ID)) {
2318 error(ERR_NONFATAL, "`%%use' expects a package name");
2319 free_tlist(origline);
2320 return DIRECTIVE_FOUND; /* but we did _something_ */
2322 if (tline->next)
2323 error(ERR_WARNING|ERR_PASS1,
2324 "trailing garbage after `%%use' ignored");
2325 if (tline->type == TOK_STRING)
2326 nasm_unquote(tline->text, NULL);
2327 use_pkg = nasm_stdmac_find_package(tline->text);
2328 if (!use_pkg)
2329 error(ERR_NONFATAL, "unknown `%%use' package: %s", tline->text);
2330 /* The first string will be <%define>__USE_*__ */
2331 pkg_macro = (char *)use_pkg + 1;
2332 if (!smacro_defined(NULL, pkg_macro, 0, NULL, true)) {
2333 /* Not already included, go ahead and include it */
2334 stdmacpos = use_pkg;
2336 free_tlist(origline);
2337 return DIRECTIVE_FOUND;
2339 case PP_PUSH:
2340 case PP_REPL:
2341 case PP_POP:
2342 tline = tline->next;
2343 skip_white_(tline);
2344 tline = expand_id(tline);
2345 if (tline) {
2346 if (!tok_type_(tline, TOK_ID)) {
2347 error(ERR_NONFATAL, "`%s' expects a context identifier",
2348 pp_directives[i]);
2349 free_tlist(origline);
2350 return DIRECTIVE_FOUND; /* but we did _something_ */
2352 if (tline->next)
2353 error(ERR_WARNING|ERR_PASS1,
2354 "trailing garbage after `%s' ignored",
2355 pp_directives[i]);
2356 p = nasm_strdup(tline->text);
2357 } else {
2358 p = NULL; /* Anonymous */
2361 if (i == PP_PUSH) {
2362 ctx = nasm_malloc(sizeof(Context));
2363 ctx->next = cstk;
2364 hash_init(&ctx->localmac, HASH_SMALL);
2365 ctx->name = p;
2366 ctx->number = unique++;
2367 cstk = ctx;
2368 } else {
2369 /* %pop or %repl */
2370 if (!cstk) {
2371 error(ERR_NONFATAL, "`%s': context stack is empty",
2372 pp_directives[i]);
2373 } else if (i == PP_POP) {
2374 if (p && (!cstk->name || nasm_stricmp(p, cstk->name)))
2375 error(ERR_NONFATAL, "`%%pop' in wrong context: %s, "
2376 "expected %s",
2377 cstk->name ? cstk->name : "anonymous", p);
2378 else
2379 ctx_pop();
2380 } else {
2381 /* i == PP_REPL */
2382 nasm_free(cstk->name);
2383 cstk->name = p;
2384 p = NULL;
2386 nasm_free(p);
2388 free_tlist(origline);
2389 return DIRECTIVE_FOUND;
2390 case PP_FATAL:
2391 severity = ERR_FATAL|ERR_NO_SEVERITY;
2392 goto issue_error;
2393 case PP_ERROR:
2394 severity = ERR_NONFATAL|ERR_NO_SEVERITY;
2395 goto issue_error;
2396 case PP_WARNING:
2397 severity = ERR_WARNING|ERR_NO_SEVERITY|ERR_WARN_USER;
2398 goto issue_error;
2400 issue_error:
2402 /* Only error out if this is the final pass */
2403 if (pass != 2 && i != PP_FATAL)
2404 return DIRECTIVE_FOUND;
2406 tline->next = expand_smacro(tline->next);
2407 tline = tline->next;
2408 skip_white_(tline);
2409 t = tline ? tline->next : NULL;
2410 skip_white_(t);
2411 if (tok_type_(tline, TOK_STRING) && !t) {
2412 /* The line contains only a quoted string */
2413 p = tline->text;
2414 nasm_unquote(p, NULL);
2415 error(severity, "%s: %s", pp_directives[i], p);
2416 } else {
2417 /* Not a quoted string, or more than a quoted string */
2418 p = detoken(tline, false);
2419 error(severity, "%s: %s", pp_directives[i], p);
2420 nasm_free(p);
2422 free_tlist(origline);
2423 return DIRECTIVE_FOUND;
2426 CASE_PP_IF:
2427 if (istk->conds && !emitting(istk->conds->state))
2428 j = COND_NEVER;
2429 else {
2430 j = if_condition(tline->next, i);
2431 tline->next = NULL; /* it got freed */
2432 j = j < 0 ? COND_NEVER : j ? COND_IF_TRUE : COND_IF_FALSE;
2434 cond = nasm_malloc(sizeof(Cond));
2435 cond->next = istk->conds;
2436 cond->state = j;
2437 istk->conds = cond;
2438 free_tlist(origline);
2439 return DIRECTIVE_FOUND;
2441 CASE_PP_ELIF:
2442 if (!istk->conds)
2443 error(ERR_FATAL, "`%s': no matching `%%if'", pp_directives[i]);
2444 switch(istk->conds->state) {
2445 case COND_IF_TRUE:
2446 istk->conds->state = COND_DONE;
2447 break;
2449 case COND_DONE:
2450 case COND_NEVER:
2451 break;
2453 case COND_ELSE_TRUE:
2454 case COND_ELSE_FALSE:
2455 error_precond(ERR_WARNING|ERR_PASS1,
2456 "`%%elif' after `%%else' ignored");
2457 istk->conds->state = COND_NEVER;
2458 break;
2460 case COND_IF_FALSE:
2462 * IMPORTANT: In the case of %if, we will already have
2463 * called expand_mmac_params(); however, if we're
2464 * processing an %elif we must have been in a
2465 * non-emitting mode, which would have inhibited
2466 * the normal invocation of expand_mmac_params().
2467 * Therefore, we have to do it explicitly here.
2469 j = if_condition(expand_mmac_params(tline->next), i);
2470 tline->next = NULL; /* it got freed */
2471 istk->conds->state =
2472 j < 0 ? COND_NEVER : j ? COND_IF_TRUE : COND_IF_FALSE;
2473 break;
2475 free_tlist(origline);
2476 return DIRECTIVE_FOUND;
2478 case PP_ELSE:
2479 if (tline->next)
2480 error_precond(ERR_WARNING|ERR_PASS1,
2481 "trailing garbage after `%%else' ignored");
2482 if (!istk->conds)
2483 error(ERR_FATAL, "`%%else': no matching `%%if'");
2484 switch(istk->conds->state) {
2485 case COND_IF_TRUE:
2486 case COND_DONE:
2487 istk->conds->state = COND_ELSE_FALSE;
2488 break;
2490 case COND_NEVER:
2491 break;
2493 case COND_IF_FALSE:
2494 istk->conds->state = COND_ELSE_TRUE;
2495 break;
2497 case COND_ELSE_TRUE:
2498 case COND_ELSE_FALSE:
2499 error_precond(ERR_WARNING|ERR_PASS1,
2500 "`%%else' after `%%else' ignored.");
2501 istk->conds->state = COND_NEVER;
2502 break;
2504 free_tlist(origline);
2505 return DIRECTIVE_FOUND;
2507 case PP_ENDIF:
2508 if (tline->next)
2509 error_precond(ERR_WARNING|ERR_PASS1,
2510 "trailing garbage after `%%endif' ignored");
2511 if (!istk->conds)
2512 error(ERR_FATAL, "`%%endif': no matching `%%if'");
2513 cond = istk->conds;
2514 istk->conds = cond->next;
2515 nasm_free(cond);
2516 free_tlist(origline);
2517 return DIRECTIVE_FOUND;
2519 case PP_MACRO:
2520 case PP_IMACRO:
2521 if (defining) {
2522 error(ERR_FATAL,
2523 "`%%%smacro': already defining a macro",
2524 (i == PP_IMACRO ? "i" : ""));
2525 return DIRECTIVE_FOUND;
2527 defining = nasm_malloc(sizeof(MMacro));
2528 defining->casesense = (i == PP_MACRO);
2529 if (!parse_mmacro_spec(tline, defining, pp_directives[i])) {
2530 nasm_free(defining);
2531 defining = NULL;
2532 return DIRECTIVE_FOUND;
2535 mmac = (MMacro *) hash_findix(&mmacros, defining->name);
2536 while (mmac) {
2537 if (!strcmp(mmac->name, defining->name) &&
2538 (mmac->nparam_min <= defining->nparam_max
2539 || defining->plus)
2540 && (defining->nparam_min <= mmac->nparam_max
2541 || mmac->plus)) {
2542 error(ERR_WARNING|ERR_PASS1,
2543 "redefining multi-line macro `%s'", defining->name);
2544 return DIRECTIVE_FOUND;
2546 mmac = mmac->next;
2548 free_tlist(origline);
2549 return DIRECTIVE_FOUND;
2551 case PP_ENDM:
2552 case PP_ENDMACRO:
2553 if (! (defining && defining->name)) {
2554 error(ERR_NONFATAL, "`%s': not defining a macro", tline->text);
2555 return DIRECTIVE_FOUND;
2557 mmhead = (MMacro **) hash_findi_add(&mmacros, defining->name);
2558 defining->next = *mmhead;
2559 *mmhead = defining;
2560 defining = NULL;
2561 free_tlist(origline);
2562 return DIRECTIVE_FOUND;
2564 case PP_UNMACRO:
2565 case PP_UNIMACRO:
2567 MMacro **mmac_p;
2568 MMacro spec;
2570 spec.casesense = (i == PP_UNMACRO);
2571 if (!parse_mmacro_spec(tline, &spec, pp_directives[i])) {
2572 return DIRECTIVE_FOUND;
2574 mmac_p = (MMacro **) hash_findi(&mmacros, spec.name, NULL);
2575 while (mmac_p && *mmac_p) {
2576 mmac = *mmac_p;
2577 if (mmac->casesense == spec.casesense &&
2578 !mstrcmp(mmac->name, spec.name, spec.casesense) &&
2579 mmac->nparam_min == spec.nparam_min &&
2580 mmac->nparam_max == spec.nparam_max &&
2581 mmac->plus == spec.plus) {
2582 *mmac_p = mmac->next;
2583 free_mmacro(mmac);
2584 } else {
2585 mmac_p = &mmac->next;
2588 free_tlist(origline);
2589 free_tlist(spec.dlist);
2590 return DIRECTIVE_FOUND;
2593 case PP_ROTATE:
2594 if (tline->next && tline->next->type == TOK_WHITESPACE)
2595 tline = tline->next;
2596 if (tline->next == NULL) {
2597 free_tlist(origline);
2598 error(ERR_NONFATAL, "`%%rotate' missing rotate count");
2599 return DIRECTIVE_FOUND;
2601 t = expand_smacro(tline->next);
2602 tline->next = NULL;
2603 free_tlist(origline);
2604 tline = t;
2605 tptr = &t;
2606 tokval.t_type = TOKEN_INVALID;
2607 evalresult =
2608 evaluate(ppscan, tptr, &tokval, NULL, pass, error, NULL);
2609 free_tlist(tline);
2610 if (!evalresult)
2611 return DIRECTIVE_FOUND;
2612 if (tokval.t_type)
2613 error(ERR_WARNING|ERR_PASS1,
2614 "trailing garbage after expression ignored");
2615 if (!is_simple(evalresult)) {
2616 error(ERR_NONFATAL, "non-constant value given to `%%rotate'");
2617 return DIRECTIVE_FOUND;
2619 mmac = istk->mstk;
2620 while (mmac && !mmac->name) /* avoid mistaking %reps for macros */
2621 mmac = mmac->next_active;
2622 if (!mmac) {
2623 error(ERR_NONFATAL, "`%%rotate' invoked outside a macro call");
2624 } else if (mmac->nparam == 0) {
2625 error(ERR_NONFATAL,
2626 "`%%rotate' invoked within macro without parameters");
2627 } else {
2628 int rotate = mmac->rotate + reloc_value(evalresult);
2630 rotate %= (int)mmac->nparam;
2631 if (rotate < 0)
2632 rotate += mmac->nparam;
2634 mmac->rotate = rotate;
2636 return DIRECTIVE_FOUND;
2638 case PP_REP:
2639 nolist = false;
2640 do {
2641 tline = tline->next;
2642 } while (tok_type_(tline, TOK_WHITESPACE));
2644 if (tok_type_(tline, TOK_ID) &&
2645 nasm_stricmp(tline->text, ".nolist") == 0) {
2646 nolist = true;
2647 do {
2648 tline = tline->next;
2649 } while (tok_type_(tline, TOK_WHITESPACE));
2652 if (tline) {
2653 t = expand_smacro(tline);
2654 tptr = &t;
2655 tokval.t_type = TOKEN_INVALID;
2656 evalresult =
2657 evaluate(ppscan, tptr, &tokval, NULL, pass, error, NULL);
2658 if (!evalresult) {
2659 free_tlist(origline);
2660 return DIRECTIVE_FOUND;
2662 if (tokval.t_type)
2663 error(ERR_WARNING|ERR_PASS1,
2664 "trailing garbage after expression ignored");
2665 if (!is_simple(evalresult)) {
2666 error(ERR_NONFATAL, "non-constant value given to `%%rep'");
2667 return DIRECTIVE_FOUND;
2669 count = reloc_value(evalresult) + 1;
2670 } else {
2671 error(ERR_NONFATAL, "`%%rep' expects a repeat count");
2672 count = 0;
2674 free_tlist(origline);
2676 tmp_defining = defining;
2677 defining = nasm_malloc(sizeof(MMacro));
2678 defining->name = NULL; /* flags this macro as a %rep block */
2679 defining->casesense = false;
2680 defining->plus = false;
2681 defining->nolist = nolist;
2682 defining->in_progress = count;
2683 defining->nparam_min = defining->nparam_max = 0;
2684 defining->defaults = NULL;
2685 defining->dlist = NULL;
2686 defining->expansion = NULL;
2687 defining->next_active = istk->mstk;
2688 defining->rep_nest = tmp_defining;
2689 return DIRECTIVE_FOUND;
2691 case PP_ENDREP:
2692 if (!defining || defining->name) {
2693 error(ERR_NONFATAL, "`%%endrep': no matching `%%rep'");
2694 return DIRECTIVE_FOUND;
2698 * Now we have a "macro" defined - although it has no name
2699 * and we won't be entering it in the hash tables - we must
2700 * push a macro-end marker for it on to istk->expansion.
2701 * After that, it will take care of propagating itself (a
2702 * macro-end marker line for a macro which is really a %rep
2703 * block will cause the macro to be re-expanded, complete
2704 * with another macro-end marker to ensure the process
2705 * continues) until the whole expansion is forcibly removed
2706 * from istk->expansion by a %exitrep.
2708 l = nasm_malloc(sizeof(Line));
2709 l->next = istk->expansion;
2710 l->finishes = defining;
2711 l->first = NULL;
2712 istk->expansion = l;
2714 istk->mstk = defining;
2716 list->uplevel(defining->nolist ? LIST_MACRO_NOLIST : LIST_MACRO);
2717 tmp_defining = defining;
2718 defining = defining->rep_nest;
2719 free_tlist(origline);
2720 return DIRECTIVE_FOUND;
2722 case PP_EXITREP:
2724 * We must search along istk->expansion until we hit a
2725 * macro-end marker for a macro with no name. Then we set
2726 * its `in_progress' flag to 0.
2728 for (l = istk->expansion; l; l = l->next)
2729 if (l->finishes && !l->finishes->name)
2730 break;
2732 if (l)
2733 l->finishes->in_progress = 1;
2734 else
2735 error(ERR_NONFATAL, "`%%exitrep' not within `%%rep' block");
2736 free_tlist(origline);
2737 return DIRECTIVE_FOUND;
2739 case PP_XDEFINE:
2740 case PP_IXDEFINE:
2741 case PP_DEFINE:
2742 case PP_IDEFINE:
2743 casesense = (i == PP_DEFINE || i == PP_XDEFINE);
2745 tline = tline->next;
2746 skip_white_(tline);
2747 tline = expand_id(tline);
2748 if (!tline || (tline->type != TOK_ID &&
2749 (tline->type != TOK_PREPROC_ID ||
2750 tline->text[1] != '$'))) {
2751 error(ERR_NONFATAL, "`%s' expects a macro identifier",
2752 pp_directives[i]);
2753 free_tlist(origline);
2754 return DIRECTIVE_FOUND;
2757 ctx = get_ctx(tline->text, false);
2759 mname = tline->text;
2760 last = tline;
2761 param_start = tline = tline->next;
2762 nparam = 0;
2764 /* Expand the macro definition now for %xdefine and %ixdefine */
2765 if ((i == PP_XDEFINE) || (i == PP_IXDEFINE))
2766 tline = expand_smacro(tline);
2768 if (tok_is_(tline, "(")) {
2770 * This macro has parameters.
2773 tline = tline->next;
2774 while (1) {
2775 skip_white_(tline);
2776 if (!tline) {
2777 error(ERR_NONFATAL, "parameter identifier expected");
2778 free_tlist(origline);
2779 return DIRECTIVE_FOUND;
2781 if (tline->type != TOK_ID) {
2782 error(ERR_NONFATAL,
2783 "`%s': parameter identifier expected",
2784 tline->text);
2785 free_tlist(origline);
2786 return DIRECTIVE_FOUND;
2788 tline->type = TOK_SMAC_PARAM + nparam++;
2789 tline = tline->next;
2790 skip_white_(tline);
2791 if (tok_is_(tline, ",")) {
2792 tline = tline->next;
2793 } else {
2794 if (!tok_is_(tline, ")")) {
2795 error(ERR_NONFATAL,
2796 "`)' expected to terminate macro template");
2797 free_tlist(origline);
2798 return DIRECTIVE_FOUND;
2800 break;
2803 last = tline;
2804 tline = tline->next;
2806 if (tok_type_(tline, TOK_WHITESPACE))
2807 last = tline, tline = tline->next;
2808 macro_start = NULL;
2809 last->next = NULL;
2810 t = tline;
2811 while (t) {
2812 if (t->type == TOK_ID) {
2813 for (tt = param_start; tt; tt = tt->next)
2814 if (tt->type >= TOK_SMAC_PARAM &&
2815 !strcmp(tt->text, t->text))
2816 t->type = tt->type;
2818 tt = t->next;
2819 t->next = macro_start;
2820 macro_start = t;
2821 t = tt;
2824 * Good. We now have a macro name, a parameter count, and a
2825 * token list (in reverse order) for an expansion. We ought
2826 * to be OK just to create an SMacro, store it, and let
2827 * free_tlist have the rest of the line (which we have
2828 * carefully re-terminated after chopping off the expansion
2829 * from the end).
2831 define_smacro(ctx, mname, casesense, nparam, macro_start);
2832 free_tlist(origline);
2833 return DIRECTIVE_FOUND;
2835 case PP_UNDEF:
2836 tline = tline->next;
2837 skip_white_(tline);
2838 tline = expand_id(tline);
2839 if (!tline || (tline->type != TOK_ID &&
2840 (tline->type != TOK_PREPROC_ID ||
2841 tline->text[1] != '$'))) {
2842 error(ERR_NONFATAL, "`%%undef' expects a macro identifier");
2843 free_tlist(origline);
2844 return DIRECTIVE_FOUND;
2846 if (tline->next) {
2847 error(ERR_WARNING|ERR_PASS1,
2848 "trailing garbage after macro name ignored");
2851 /* Find the context that symbol belongs to */
2852 ctx = get_ctx(tline->text, false);
2853 undef_smacro(ctx, tline->text);
2854 free_tlist(origline);
2855 return DIRECTIVE_FOUND;
2857 case PP_DEFSTR:
2858 case PP_IDEFSTR:
2859 casesense = (i == PP_DEFSTR);
2861 tline = tline->next;
2862 skip_white_(tline);
2863 tline = expand_id(tline);
2864 if (!tline || (tline->type != TOK_ID &&
2865 (tline->type != TOK_PREPROC_ID ||
2866 tline->text[1] != '$'))) {
2867 error(ERR_NONFATAL, "`%s' expects a macro identifier",
2868 pp_directives[i]);
2869 free_tlist(origline);
2870 return DIRECTIVE_FOUND;
2873 ctx = get_ctx(tline->text, false);
2875 mname = tline->text;
2876 last = tline;
2877 tline = expand_smacro(tline->next);
2878 last->next = NULL;
2880 while (tok_type_(tline, TOK_WHITESPACE))
2881 tline = delete_Token(tline);
2883 p = detoken(tline, false);
2884 macro_start = nasm_malloc(sizeof(*macro_start));
2885 macro_start->next = NULL;
2886 macro_start->text = nasm_quote(p, strlen(p));
2887 macro_start->type = TOK_STRING;
2888 macro_start->a.mac = NULL;
2889 nasm_free(p);
2892 * We now have a macro name, an implicit parameter count of
2893 * zero, and a string token to use as an expansion. Create
2894 * and store an SMacro.
2896 define_smacro(ctx, mname, casesense, 0, macro_start);
2897 free_tlist(origline);
2898 return DIRECTIVE_FOUND;
2900 case PP_PATHSEARCH:
2902 FILE *fp;
2903 StrList *xsl = NULL;
2904 StrList **xst = &xsl;
2906 casesense = true;
2908 tline = tline->next;
2909 skip_white_(tline);
2910 tline = expand_id(tline);
2911 if (!tline || (tline->type != TOK_ID &&
2912 (tline->type != TOK_PREPROC_ID ||
2913 tline->text[1] != '$'))) {
2914 error(ERR_NONFATAL,
2915 "`%%pathsearch' expects a macro identifier as first parameter");
2916 free_tlist(origline);
2917 return DIRECTIVE_FOUND;
2919 ctx = get_ctx(tline->text, false);
2921 mname = tline->text;
2922 last = tline;
2923 tline = expand_smacro(tline->next);
2924 last->next = NULL;
2926 t = tline;
2927 while (tok_type_(t, TOK_WHITESPACE))
2928 t = t->next;
2930 if (!t || (t->type != TOK_STRING &&
2931 t->type != TOK_INTERNAL_STRING)) {
2932 error(ERR_NONFATAL, "`%%pathsearch' expects a file name");
2933 free_tlist(tline);
2934 free_tlist(origline);
2935 return DIRECTIVE_FOUND; /* but we did _something_ */
2937 if (t->next)
2938 error(ERR_WARNING|ERR_PASS1,
2939 "trailing garbage after `%%pathsearch' ignored");
2940 p = t->text;
2941 if (t->type != TOK_INTERNAL_STRING)
2942 nasm_unquote(p, NULL);
2944 fp = inc_fopen(p, &xsl, &xst, true);
2945 if (fp) {
2946 p = xsl->str;
2947 fclose(fp); /* Don't actually care about the file */
2949 macro_start = nasm_malloc(sizeof(*macro_start));
2950 macro_start->next = NULL;
2951 macro_start->text = nasm_quote(p, strlen(p));
2952 macro_start->type = TOK_STRING;
2953 macro_start->a.mac = NULL;
2954 if (xsl)
2955 nasm_free(xsl);
2958 * We now have a macro name, an implicit parameter count of
2959 * zero, and a string token to use as an expansion. Create
2960 * and store an SMacro.
2962 define_smacro(ctx, mname, casesense, 0, macro_start);
2963 free_tlist(tline);
2964 free_tlist(origline);
2965 return DIRECTIVE_FOUND;
2968 case PP_STRLEN:
2969 casesense = true;
2971 tline = tline->next;
2972 skip_white_(tline);
2973 tline = expand_id(tline);
2974 if (!tline || (tline->type != TOK_ID &&
2975 (tline->type != TOK_PREPROC_ID ||
2976 tline->text[1] != '$'))) {
2977 error(ERR_NONFATAL,
2978 "`%%strlen' expects a macro identifier as first parameter");
2979 free_tlist(origline);
2980 return DIRECTIVE_FOUND;
2982 ctx = get_ctx(tline->text, false);
2984 mname = tline->text;
2985 last = tline;
2986 tline = expand_smacro(tline->next);
2987 last->next = NULL;
2989 t = tline;
2990 while (tok_type_(t, TOK_WHITESPACE))
2991 t = t->next;
2992 /* t should now point to the string */
2993 if (t->type != TOK_STRING) {
2994 error(ERR_NONFATAL,
2995 "`%%strlen` requires string as second parameter");
2996 free_tlist(tline);
2997 free_tlist(origline);
2998 return DIRECTIVE_FOUND;
3001 macro_start = nasm_malloc(sizeof(*macro_start));
3002 macro_start->next = NULL;
3003 make_tok_num(macro_start, nasm_unquote(t->text, NULL));
3004 macro_start->a.mac = NULL;
3007 * We now have a macro name, an implicit parameter count of
3008 * zero, and a numeric token to use as an expansion. Create
3009 * and store an SMacro.
3011 define_smacro(ctx, mname, casesense, 0, macro_start);
3012 free_tlist(tline);
3013 free_tlist(origline);
3014 return DIRECTIVE_FOUND;
3016 case PP_STRCAT:
3017 casesense = true;
3019 tline = tline->next;
3020 skip_white_(tline);
3021 tline = expand_id(tline);
3022 if (!tline || (tline->type != TOK_ID &&
3023 (tline->type != TOK_PREPROC_ID ||
3024 tline->text[1] != '$'))) {
3025 error(ERR_NONFATAL,
3026 "`%%strcat' expects a macro identifier as first parameter");
3027 free_tlist(origline);
3028 return DIRECTIVE_FOUND;
3030 ctx = get_ctx(tline->text, false);
3032 mname = tline->text;
3033 last = tline;
3034 tline = expand_smacro(tline->next);
3035 last->next = NULL;
3037 len = 0;
3038 for (t = tline; t; t = t->next) {
3039 switch (t->type) {
3040 case TOK_WHITESPACE:
3041 break;
3042 case TOK_STRING:
3043 len += t->a.len = nasm_unquote(t->text, NULL);
3044 break;
3045 case TOK_OTHER:
3046 if (!strcmp(t->text, ",")) /* permit comma separators */
3047 break;
3048 /* else fall through */
3049 default:
3050 error(ERR_NONFATAL,
3051 "non-string passed to `%%strcat' (%d)", t->type);
3052 free_tlist(tline);
3053 free_tlist(origline);
3054 return DIRECTIVE_FOUND;
3058 p = pp = nasm_malloc(len);
3059 t = tline;
3060 for (t = tline; t; t = t->next) {
3061 if (t->type == TOK_STRING) {
3062 memcpy(p, t->text, t->a.len);
3063 p += t->a.len;
3068 * We now have a macro name, an implicit parameter count of
3069 * zero, and a numeric token to use as an expansion. Create
3070 * and store an SMacro.
3072 macro_start = new_Token(NULL, TOK_STRING, NULL, 0);
3073 macro_start->text = nasm_quote(pp, len);
3074 nasm_free(pp);
3075 define_smacro(ctx, mname, casesense, 0, macro_start);
3076 free_tlist(tline);
3077 free_tlist(origline);
3078 return DIRECTIVE_FOUND;
3080 case PP_SUBSTR:
3082 int64_t a1, a2;
3083 size_t len;
3085 casesense = true;
3087 tline = tline->next;
3088 skip_white_(tline);
3089 tline = expand_id(tline);
3090 if (!tline || (tline->type != TOK_ID &&
3091 (tline->type != TOK_PREPROC_ID ||
3092 tline->text[1] != '$'))) {
3093 error(ERR_NONFATAL,
3094 "`%%substr' expects a macro identifier as first parameter");
3095 free_tlist(origline);
3096 return DIRECTIVE_FOUND;
3098 ctx = get_ctx(tline->text, false);
3100 mname = tline->text;
3101 last = tline;
3102 tline = expand_smacro(tline->next);
3103 last->next = NULL;
3105 t = tline->next;
3106 while (tok_type_(t, TOK_WHITESPACE))
3107 t = t->next;
3109 /* t should now point to the string */
3110 if (t->type != TOK_STRING) {
3111 error(ERR_NONFATAL,
3112 "`%%substr` requires string as second parameter");
3113 free_tlist(tline);
3114 free_tlist(origline);
3115 return DIRECTIVE_FOUND;
3118 tt = t->next;
3119 tptr = &tt;
3120 tokval.t_type = TOKEN_INVALID;
3121 evalresult = evaluate(ppscan, tptr, &tokval, NULL,
3122 pass, error, NULL);
3123 if (!evalresult) {
3124 free_tlist(tline);
3125 free_tlist(origline);
3126 return DIRECTIVE_FOUND;
3127 } else if (!is_simple(evalresult)) {
3128 error(ERR_NONFATAL, "non-constant value given to `%%substr`");
3129 free_tlist(tline);
3130 free_tlist(origline);
3131 return DIRECTIVE_FOUND;
3133 a1 = evalresult->value-1;
3135 while (tok_type_(tt, TOK_WHITESPACE))
3136 tt = tt->next;
3137 if (!tt) {
3138 a2 = 1; /* Backwards compatibility: one character */
3139 } else {
3140 tokval.t_type = TOKEN_INVALID;
3141 evalresult = evaluate(ppscan, tptr, &tokval, NULL,
3142 pass, error, NULL);
3143 if (!evalresult) {
3144 free_tlist(tline);
3145 free_tlist(origline);
3146 return DIRECTIVE_FOUND;
3147 } else if (!is_simple(evalresult)) {
3148 error(ERR_NONFATAL, "non-constant value given to `%%substr`");
3149 free_tlist(tline);
3150 free_tlist(origline);
3151 return DIRECTIVE_FOUND;
3153 a2 = evalresult->value;
3156 len = nasm_unquote(t->text, NULL);
3157 if (a2 < 0)
3158 a2 = a2+1+len-a1;
3159 if (a1+a2 > (int64_t)len)
3160 a2 = len-a1;
3162 macro_start = nasm_malloc(sizeof(*macro_start));
3163 macro_start->next = NULL;
3164 macro_start->text = nasm_quote((a1 < 0) ? "" : t->text+a1, a2);
3165 macro_start->type = TOK_STRING;
3166 macro_start->a.mac = NULL;
3169 * We now have a macro name, an implicit parameter count of
3170 * zero, and a numeric token to use as an expansion. Create
3171 * and store an SMacro.
3173 define_smacro(ctx, mname, casesense, 0, macro_start);
3174 free_tlist(tline);
3175 free_tlist(origline);
3176 return DIRECTIVE_FOUND;
3179 case PP_ASSIGN:
3180 case PP_IASSIGN:
3181 casesense = (i == PP_ASSIGN);
3183 tline = tline->next;
3184 skip_white_(tline);
3185 tline = expand_id(tline);
3186 if (!tline || (tline->type != TOK_ID &&
3187 (tline->type != TOK_PREPROC_ID ||
3188 tline->text[1] != '$'))) {
3189 error(ERR_NONFATAL,
3190 "`%%%sassign' expects a macro identifier",
3191 (i == PP_IASSIGN ? "i" : ""));
3192 free_tlist(origline);
3193 return DIRECTIVE_FOUND;
3195 ctx = get_ctx(tline->text, false);
3197 mname = tline->text;
3198 last = tline;
3199 tline = expand_smacro(tline->next);
3200 last->next = NULL;
3202 t = tline;
3203 tptr = &t;
3204 tokval.t_type = TOKEN_INVALID;
3205 evalresult =
3206 evaluate(ppscan, tptr, &tokval, NULL, pass, error, NULL);
3207 free_tlist(tline);
3208 if (!evalresult) {
3209 free_tlist(origline);
3210 return DIRECTIVE_FOUND;
3213 if (tokval.t_type)
3214 error(ERR_WARNING|ERR_PASS1,
3215 "trailing garbage after expression ignored");
3217 if (!is_simple(evalresult)) {
3218 error(ERR_NONFATAL,
3219 "non-constant value given to `%%%sassign'",
3220 (i == PP_IASSIGN ? "i" : ""));
3221 free_tlist(origline);
3222 return DIRECTIVE_FOUND;
3225 macro_start = nasm_malloc(sizeof(*macro_start));
3226 macro_start->next = NULL;
3227 make_tok_num(macro_start, reloc_value(evalresult));
3228 macro_start->a.mac = NULL;
3231 * We now have a macro name, an implicit parameter count of
3232 * zero, and a numeric token to use as an expansion. Create
3233 * and store an SMacro.
3235 define_smacro(ctx, mname, casesense, 0, macro_start);
3236 free_tlist(origline);
3237 return DIRECTIVE_FOUND;
3239 case PP_LINE:
3241 * Syntax is `%line nnn[+mmm] [filename]'
3243 tline = tline->next;
3244 skip_white_(tline);
3245 if (!tok_type_(tline, TOK_NUMBER)) {
3246 error(ERR_NONFATAL, "`%%line' expects line number");
3247 free_tlist(origline);
3248 return DIRECTIVE_FOUND;
3250 k = readnum(tline->text, &err);
3251 m = 1;
3252 tline = tline->next;
3253 if (tok_is_(tline, "+")) {
3254 tline = tline->next;
3255 if (!tok_type_(tline, TOK_NUMBER)) {
3256 error(ERR_NONFATAL, "`%%line' expects line increment");
3257 free_tlist(origline);
3258 return DIRECTIVE_FOUND;
3260 m = readnum(tline->text, &err);
3261 tline = tline->next;
3263 skip_white_(tline);
3264 src_set_linnum(k);
3265 istk->lineinc = m;
3266 if (tline) {
3267 nasm_free(src_set_fname(detoken(tline, false)));
3269 free_tlist(origline);
3270 return DIRECTIVE_FOUND;
3272 default:
3273 error(ERR_FATAL,
3274 "preprocessor directive `%s' not yet implemented",
3275 pp_directives[i]);
3276 return DIRECTIVE_FOUND;
3281 * Ensure that a macro parameter contains a condition code and
3282 * nothing else. Return the condition code index if so, or -1
3283 * otherwise.
3285 static int find_cc(Token * t)
3287 Token *tt;
3288 int i, j, k, m;
3290 if (!t)
3291 return -1; /* Probably a %+ without a space */
3293 skip_white_(t);
3294 if (t->type != TOK_ID)
3295 return -1;
3296 tt = t->next;
3297 skip_white_(tt);
3298 if (tt && (tt->type != TOK_OTHER || strcmp(tt->text, ",")))
3299 return -1;
3301 i = -1;
3302 j = elements(conditions);
3303 while (j - i > 1) {
3304 k = (j + i) / 2;
3305 m = nasm_stricmp(t->text, conditions[k]);
3306 if (m == 0) {
3307 i = k;
3308 j = -2;
3309 break;
3310 } else if (m < 0) {
3311 j = k;
3312 } else
3313 i = k;
3315 if (j != -2)
3316 return -1;
3317 return i;
3321 * Expand MMacro-local things: parameter references (%0, %n, %+n,
3322 * %-n) and MMacro-local identifiers (%%foo) as well as
3323 * macro indirection (%[...]).
3325 static Token *expand_mmac_params(Token * tline)
3327 Token *t, *tt, **tail, *thead;
3329 tail = &thead;
3330 thead = NULL;
3332 while (tline) {
3333 if (tline->type == TOK_PREPROC_ID &&
3334 (((tline->text[1] == '+' || tline->text[1] == '-')
3335 && tline->text[2]) || tline->text[1] == '%'
3336 || (tline->text[1] >= '0' && tline->text[1] <= '9'))) {
3337 char *text = NULL;
3338 int type = 0, cc; /* type = 0 to placate optimisers */
3339 char tmpbuf[30];
3340 unsigned int n;
3341 int i;
3342 MMacro *mac;
3344 t = tline;
3345 tline = tline->next;
3347 mac = istk->mstk;
3348 while (mac && !mac->name) /* avoid mistaking %reps for macros */
3349 mac = mac->next_active;
3350 if (!mac)
3351 error(ERR_NONFATAL, "`%s': not in a macro call", t->text);
3352 else
3353 switch (t->text[1]) {
3355 * We have to make a substitution of one of the
3356 * forms %1, %-1, %+1, %%foo, %0.
3358 case '0':
3359 type = TOK_NUMBER;
3360 snprintf(tmpbuf, sizeof(tmpbuf), "%d", mac->nparam);
3361 text = nasm_strdup(tmpbuf);
3362 break;
3363 case '%':
3364 type = TOK_ID;
3365 snprintf(tmpbuf, sizeof(tmpbuf), "..@%"PRIu64".",
3366 mac->unique);
3367 text = nasm_strcat(tmpbuf, t->text + 2);
3368 break;
3369 case '-':
3370 n = atoi(t->text + 2) - 1;
3371 if (n >= mac->nparam)
3372 tt = NULL;
3373 else {
3374 if (mac->nparam > 1)
3375 n = (n + mac->rotate) % mac->nparam;
3376 tt = mac->params[n];
3378 cc = find_cc(tt);
3379 if (cc == -1) {
3380 error(ERR_NONFATAL,
3381 "macro parameter %d is not a condition code",
3382 n + 1);
3383 text = NULL;
3384 } else {
3385 type = TOK_ID;
3386 if (inverse_ccs[cc] == -1) {
3387 error(ERR_NONFATAL,
3388 "condition code `%s' is not invertible",
3389 conditions[cc]);
3390 text = NULL;
3391 } else
3392 text = nasm_strdup(conditions[inverse_ccs[cc]]);
3394 break;
3395 case '+':
3396 n = atoi(t->text + 2) - 1;
3397 if (n >= mac->nparam)
3398 tt = NULL;
3399 else {
3400 if (mac->nparam > 1)
3401 n = (n + mac->rotate) % mac->nparam;
3402 tt = mac->params[n];
3404 cc = find_cc(tt);
3405 if (cc == -1) {
3406 error(ERR_NONFATAL,
3407 "macro parameter %d is not a condition code",
3408 n + 1);
3409 text = NULL;
3410 } else {
3411 type = TOK_ID;
3412 text = nasm_strdup(conditions[cc]);
3414 break;
3415 default:
3416 n = atoi(t->text + 1) - 1;
3417 if (n >= mac->nparam)
3418 tt = NULL;
3419 else {
3420 if (mac->nparam > 1)
3421 n = (n + mac->rotate) % mac->nparam;
3422 tt = mac->params[n];
3424 if (tt) {
3425 for (i = 0; i < mac->paramlen[n]; i++) {
3426 *tail = new_Token(NULL, tt->type, tt->text, 0);
3427 tail = &(*tail)->next;
3428 tt = tt->next;
3431 text = NULL; /* we've done it here */
3432 break;
3434 if (!text) {
3435 delete_Token(t);
3436 } else {
3437 *tail = t;
3438 tail = &t->next;
3439 t->type = type;
3440 nasm_free(t->text);
3441 t->text = text;
3442 t->a.mac = NULL;
3444 continue;
3445 } else if (tline->type == TOK_INDIRECT) {
3446 t = tline;
3447 tline = tline->next;
3448 tt = tokenize(t->text);
3449 tt = expand_mmac_params(tt);
3450 tt = expand_smacro(tt);
3451 *tail = tt;
3452 while (tt) {
3453 tt->a.mac = NULL; /* Necessary? */
3454 tail = &tt->next;
3455 tt = tt->next;
3457 delete_Token(t);
3458 } else {
3459 t = *tail = tline;
3460 tline = tline->next;
3461 t->a.mac = NULL;
3462 tail = &t->next;
3465 *tail = NULL;
3467 /* Now handle token pasting... */
3468 t = thead;
3469 while (t && (tt = t->next)) {
3470 switch (t->type) {
3471 case TOK_WHITESPACE:
3472 if (tt->type == TOK_WHITESPACE) {
3473 t->next = delete_Token(tt);
3474 } else {
3475 t = tt;
3477 break;
3478 case TOK_ID:
3479 case TOK_NUMBER:
3480 if (tt->type == t->type || tt->type == TOK_NUMBER) {
3481 char *tmp = nasm_strcat(t->text, tt->text);
3482 nasm_free(t->text);
3483 t->text = tmp;
3484 t->next = delete_Token(tt);
3485 } else {
3486 t = tt;
3488 break;
3489 default:
3490 t = tt;
3491 break;
3494 return thead;
3498 * Expand all single-line macro calls made in the given line.
3499 * Return the expanded version of the line. The original is deemed
3500 * to be destroyed in the process. (In reality we'll just move
3501 * Tokens from input to output a lot of the time, rather than
3502 * actually bothering to destroy and replicate.)
3504 #define DEADMAN_LIMIT (1 << 20)
3506 static Token *expand_smacro(Token * tline)
3508 Token *t, *tt, *mstart, **tail, *thead;
3509 struct hash_table *smtbl;
3510 SMacro *head = NULL, *m;
3511 Token **params;
3512 int *paramsize;
3513 unsigned int nparam, sparam;
3514 int brackets, rescan;
3515 Token *org_tline = tline;
3516 Context *ctx;
3517 char *mname;
3518 int deadman = DEADMAN_LIMIT;
3521 * Trick: we should avoid changing the start token pointer since it can
3522 * be contained in "next" field of other token. Because of this
3523 * we allocate a copy of first token and work with it; at the end of
3524 * routine we copy it back
3526 if (org_tline) {
3527 tline =
3528 new_Token(org_tline->next, org_tline->type, org_tline->text,
3530 tline->a.mac = org_tline->a.mac;
3531 nasm_free(org_tline->text);
3532 org_tline->text = NULL;
3535 again:
3536 tail = &thead;
3537 thead = NULL;
3539 while (tline) { /* main token loop */
3540 if (!--deadman) {
3541 error(ERR_NONFATAL, "interminable macro recursion");
3542 break;
3545 if ((mname = tline->text)) {
3546 /* if this token is a local macro, look in local context */
3547 ctx = NULL;
3548 smtbl = &smacros;
3549 if (tline->type == TOK_ID || tline->type == TOK_PREPROC_ID) {
3550 ctx = get_ctx(mname, true);
3551 if (ctx)
3552 smtbl = &ctx->localmac;
3554 head = (SMacro *) hash_findix(smtbl, mname);
3557 * We've hit an identifier. As in is_mmacro below, we first
3558 * check whether the identifier is a single-line macro at
3559 * all, then think about checking for parameters if
3560 * necessary.
3562 for (m = head; m; m = m->next)
3563 if (!mstrcmp(m->name, mname, m->casesense))
3564 break;
3565 if (m) {
3566 mstart = tline;
3567 params = NULL;
3568 paramsize = NULL;
3569 if (m->nparam == 0) {
3571 * Simple case: the macro is parameterless. Discard the
3572 * one token that the macro call took, and push the
3573 * expansion back on the to-do stack.
3575 if (!m->expansion) {
3576 if (!strcmp("__FILE__", m->name)) {
3577 int32_t num = 0;
3578 char *file = NULL;
3579 src_get(&num, &file);
3580 tline->text = nasm_quote(file, strlen(file));
3581 tline->type = TOK_STRING;
3582 nasm_free(file);
3583 continue;
3585 if (!strcmp("__LINE__", m->name)) {
3586 nasm_free(tline->text);
3587 make_tok_num(tline, src_get_linnum());
3588 continue;
3590 if (!strcmp("__BITS__", m->name)) {
3591 nasm_free(tline->text);
3592 make_tok_num(tline, globalbits);
3593 continue;
3595 tline = delete_Token(tline);
3596 continue;
3598 } else {
3600 * Complicated case: at least one macro with this name
3601 * exists and takes parameters. We must find the
3602 * parameters in the call, count them, find the SMacro
3603 * that corresponds to that form of the macro call, and
3604 * substitute for the parameters when we expand. What a
3605 * pain.
3607 /*tline = tline->next;
3608 skip_white_(tline); */
3609 do {
3610 t = tline->next;
3611 while (tok_type_(t, TOK_SMAC_END)) {
3612 t->a.mac->in_progress = false;
3613 t->text = NULL;
3614 t = tline->next = delete_Token(t);
3616 tline = t;
3617 } while (tok_type_(tline, TOK_WHITESPACE));
3618 if (!tok_is_(tline, "(")) {
3620 * This macro wasn't called with parameters: ignore
3621 * the call. (Behaviour borrowed from gnu cpp.)
3623 tline = mstart;
3624 m = NULL;
3625 } else {
3626 int paren = 0;
3627 int white = 0;
3628 brackets = 0;
3629 nparam = 0;
3630 sparam = PARAM_DELTA;
3631 params = nasm_malloc(sparam * sizeof(Token *));
3632 params[0] = tline->next;
3633 paramsize = nasm_malloc(sparam * sizeof(int));
3634 paramsize[0] = 0;
3635 while (true) { /* parameter loop */
3637 * For some unusual expansions
3638 * which concatenates function call
3640 t = tline->next;
3641 while (tok_type_(t, TOK_SMAC_END)) {
3642 t->a.mac->in_progress = false;
3643 t->text = NULL;
3644 t = tline->next = delete_Token(t);
3646 tline = t;
3648 if (!tline) {
3649 error(ERR_NONFATAL,
3650 "macro call expects terminating `)'");
3651 break;
3653 if (tline->type == TOK_WHITESPACE
3654 && brackets <= 0) {
3655 if (paramsize[nparam])
3656 white++;
3657 else
3658 params[nparam] = tline->next;
3659 continue; /* parameter loop */
3661 if (tline->type == TOK_OTHER
3662 && tline->text[1] == 0) {
3663 char ch = tline->text[0];
3664 if (ch == ',' && !paren && brackets <= 0) {
3665 if (++nparam >= sparam) {
3666 sparam += PARAM_DELTA;
3667 params = nasm_realloc(params,
3668 sparam *
3669 sizeof(Token
3670 *));
3671 paramsize =
3672 nasm_realloc(paramsize,
3673 sparam *
3674 sizeof(int));
3676 params[nparam] = tline->next;
3677 paramsize[nparam] = 0;
3678 white = 0;
3679 continue; /* parameter loop */
3681 if (ch == '{' &&
3682 (brackets > 0 || (brackets == 0 &&
3683 !paramsize[nparam])))
3685 if (!(brackets++)) {
3686 params[nparam] = tline->next;
3687 continue; /* parameter loop */
3690 if (ch == '}' && brackets > 0)
3691 if (--brackets == 0) {
3692 brackets = -1;
3693 continue; /* parameter loop */
3695 if (ch == '(' && !brackets)
3696 paren++;
3697 if (ch == ')' && brackets <= 0)
3698 if (--paren < 0)
3699 break;
3701 if (brackets < 0) {
3702 brackets = 0;
3703 error(ERR_NONFATAL, "braces do not "
3704 "enclose all of macro parameter");
3706 paramsize[nparam] += white + 1;
3707 white = 0;
3708 } /* parameter loop */
3709 nparam++;
3710 while (m && (m->nparam != nparam ||
3711 mstrcmp(m->name, mname,
3712 m->casesense)))
3713 m = m->next;
3714 if (!m)
3715 error(ERR_WARNING|ERR_PASS1|ERR_WARN_MNP,
3716 "macro `%s' exists, "
3717 "but not taking %d parameters",
3718 mstart->text, nparam);
3721 if (m && m->in_progress)
3722 m = NULL;
3723 if (!m) { /* in progess or didn't find '(' or wrong nparam */
3725 * Design question: should we handle !tline, which
3726 * indicates missing ')' here, or expand those
3727 * macros anyway, which requires the (t) test a few
3728 * lines down?
3730 nasm_free(params);
3731 nasm_free(paramsize);
3732 tline = mstart;
3733 } else {
3735 * Expand the macro: we are placed on the last token of the
3736 * call, so that we can easily split the call from the
3737 * following tokens. We also start by pushing an SMAC_END
3738 * token for the cycle removal.
3740 t = tline;
3741 if (t) {
3742 tline = t->next;
3743 t->next = NULL;
3745 tt = new_Token(tline, TOK_SMAC_END, NULL, 0);
3746 tt->a.mac = m;
3747 m->in_progress = true;
3748 tline = tt;
3749 for (t = m->expansion; t; t = t->next) {
3750 if (t->type >= TOK_SMAC_PARAM) {
3751 Token *pcopy = tline, **ptail = &pcopy;
3752 Token *ttt, *pt;
3753 int i;
3755 ttt = params[t->type - TOK_SMAC_PARAM];
3756 for (i = paramsize[t->type - TOK_SMAC_PARAM];
3757 --i >= 0;) {
3758 pt = *ptail =
3759 new_Token(tline, ttt->type, ttt->text,
3761 ptail = &pt->next;
3762 ttt = ttt->next;
3764 tline = pcopy;
3765 } else if (t->type == TOK_PREPROC_Q) {
3766 tt = new_Token(tline, TOK_ID, mname, 0);
3767 tline = tt;
3768 } else if (t->type == TOK_PREPROC_QQ) {
3769 tt = new_Token(tline, TOK_ID, m->name, 0);
3770 tline = tt;
3771 } else {
3772 tt = new_Token(tline, t->type, t->text, 0);
3773 tline = tt;
3778 * Having done that, get rid of the macro call, and clean
3779 * up the parameters.
3781 nasm_free(params);
3782 nasm_free(paramsize);
3783 free_tlist(mstart);
3784 continue; /* main token loop */
3789 if (tline->type == TOK_SMAC_END) {
3790 tline->a.mac->in_progress = false;
3791 tline = delete_Token(tline);
3792 } else {
3793 t = *tail = tline;
3794 tline = tline->next;
3795 t->a.mac = NULL;
3796 t->next = NULL;
3797 tail = &t->next;
3802 * Now scan the entire line and look for successive TOK_IDs that resulted
3803 * after expansion (they can't be produced by tokenize()). The successive
3804 * TOK_IDs should be concatenated.
3805 * Also we look for %+ tokens and concatenate the tokens before and after
3806 * them (without white spaces in between).
3808 t = thead;
3809 rescan = 0;
3810 while (t) {
3811 while (t && t->type != TOK_ID && t->type != TOK_PREPROC_ID)
3812 t = t->next;
3813 if (!t || !t->next)
3814 break;
3815 if (t->next->type == TOK_ID ||
3816 t->next->type == TOK_PREPROC_ID ||
3817 t->next->type == TOK_NUMBER) {
3818 char *p = nasm_strcat(t->text, t->next->text);
3819 nasm_free(t->text);
3820 t->next = delete_Token(t->next);
3821 t->text = p;
3822 rescan = 1;
3823 } else if (t->next->type == TOK_WHITESPACE && t->next->next &&
3824 t->next->next->type == TOK_PREPROC_ID &&
3825 strcmp(t->next->next->text, "%+") == 0) {
3826 /* free the next whitespace, the %+ token and next whitespace */
3827 int i;
3828 for (i = 1; i <= 3; i++) {
3829 if (!t->next
3830 || (i != 2 && t->next->type != TOK_WHITESPACE))
3831 break;
3832 t->next = delete_Token(t->next);
3833 } /* endfor */
3834 } else
3835 t = t->next;
3837 /* If we concatenaded something, re-scan the line for macros */
3838 if (rescan) {
3839 tline = thead;
3840 goto again;
3843 if (org_tline) {
3844 if (thead) {
3845 *org_tline = *thead;
3846 /* since we just gave text to org_line, don't free it */
3847 thead->text = NULL;
3848 delete_Token(thead);
3849 } else {
3850 /* the expression expanded to empty line;
3851 we can't return NULL for some reasons
3852 we just set the line to a single WHITESPACE token. */
3853 memset(org_tline, 0, sizeof(*org_tline));
3854 org_tline->text = NULL;
3855 org_tline->type = TOK_WHITESPACE;
3857 thead = org_tline;
3860 return thead;
3864 * Similar to expand_smacro but used exclusively with macro identifiers
3865 * right before they are fetched in. The reason is that there can be
3866 * identifiers consisting of several subparts. We consider that if there
3867 * are more than one element forming the name, user wants a expansion,
3868 * otherwise it will be left as-is. Example:
3870 * %define %$abc cde
3872 * the identifier %$abc will be left as-is so that the handler for %define
3873 * will suck it and define the corresponding value. Other case:
3875 * %define _%$abc cde
3877 * In this case user wants name to be expanded *before* %define starts
3878 * working, so we'll expand %$abc into something (if it has a value;
3879 * otherwise it will be left as-is) then concatenate all successive
3880 * PP_IDs into one.
3882 static Token *expand_id(Token * tline)
3884 Token *cur, *oldnext = NULL;
3886 if (!tline || !tline->next)
3887 return tline;
3889 cur = tline;
3890 while (cur->next &&
3891 (cur->next->type == TOK_ID ||
3892 cur->next->type == TOK_PREPROC_ID
3893 || cur->next->type == TOK_NUMBER))
3894 cur = cur->next;
3896 /* If identifier consists of just one token, don't expand */
3897 if (cur == tline)
3898 return tline;
3900 if (cur) {
3901 oldnext = cur->next; /* Detach the tail past identifier */
3902 cur->next = NULL; /* so that expand_smacro stops here */
3905 tline = expand_smacro(tline);
3907 if (cur) {
3908 /* expand_smacro possibly changhed tline; re-scan for EOL */
3909 cur = tline;
3910 while (cur && cur->next)
3911 cur = cur->next;
3912 if (cur)
3913 cur->next = oldnext;
3916 return tline;
3920 * Determine whether the given line constitutes a multi-line macro
3921 * call, and return the MMacro structure called if so. Doesn't have
3922 * to check for an initial label - that's taken care of in
3923 * expand_mmacro - but must check numbers of parameters. Guaranteed
3924 * to be called with tline->type == TOK_ID, so the putative macro
3925 * name is easy to find.
3927 static MMacro *is_mmacro(Token * tline, Token *** params_array)
3929 MMacro *head, *m;
3930 Token **params;
3931 int nparam;
3933 head = (MMacro *) hash_findix(&mmacros, tline->text);
3936 * Efficiency: first we see if any macro exists with the given
3937 * name. If not, we can return NULL immediately. _Then_ we
3938 * count the parameters, and then we look further along the
3939 * list if necessary to find the proper MMacro.
3941 for (m = head; m; m = m->next)
3942 if (!mstrcmp(m->name, tline->text, m->casesense))
3943 break;
3944 if (!m)
3945 return NULL;
3948 * OK, we have a potential macro. Count and demarcate the
3949 * parameters.
3951 count_mmac_params(tline->next, &nparam, &params);
3954 * So we know how many parameters we've got. Find the MMacro
3955 * structure that handles this number.
3957 while (m) {
3958 if (m->nparam_min <= nparam
3959 && (m->plus || nparam <= m->nparam_max)) {
3961 * This one is right. Just check if cycle removal
3962 * prohibits us using it before we actually celebrate...
3964 if (m->in_progress) {
3965 #if 0
3966 error(ERR_NONFATAL,
3967 "self-reference in multi-line macro `%s'", m->name);
3968 #endif
3969 nasm_free(params);
3970 return NULL;
3973 * It's right, and we can use it. Add its default
3974 * parameters to the end of our list if necessary.
3976 if (m->defaults && nparam < m->nparam_min + m->ndefs) {
3977 params =
3978 nasm_realloc(params,
3979 ((m->nparam_min + m->ndefs +
3980 1) * sizeof(*params)));
3981 while (nparam < m->nparam_min + m->ndefs) {
3982 params[nparam] = m->defaults[nparam - m->nparam_min];
3983 nparam++;
3987 * If we've gone over the maximum parameter count (and
3988 * we're in Plus mode), ignore parameters beyond
3989 * nparam_max.
3991 if (m->plus && nparam > m->nparam_max)
3992 nparam = m->nparam_max;
3994 * Then terminate the parameter list, and leave.
3996 if (!params) { /* need this special case */
3997 params = nasm_malloc(sizeof(*params));
3998 nparam = 0;
4000 params[nparam] = NULL;
4001 *params_array = params;
4002 return m;
4005 * This one wasn't right: look for the next one with the
4006 * same name.
4008 for (m = m->next; m; m = m->next)
4009 if (!mstrcmp(m->name, tline->text, m->casesense))
4010 break;
4014 * After all that, we didn't find one with the right number of
4015 * parameters. Issue a warning, and fail to expand the macro.
4017 error(ERR_WARNING|ERR_PASS1|ERR_WARN_MNP,
4018 "macro `%s' exists, but not taking %d parameters",
4019 tline->text, nparam);
4020 nasm_free(params);
4021 return NULL;
4025 * Expand the multi-line macro call made by the given line, if
4026 * there is one to be expanded. If there is, push the expansion on
4027 * istk->expansion and return 1. Otherwise return 0.
4029 static int expand_mmacro(Token * tline)
4031 Token *startline = tline;
4032 Token *label = NULL;
4033 int dont_prepend = 0;
4034 Token **params, *t, *mtok, *tt;
4035 MMacro *m;
4036 Line *l, *ll;
4037 int i, nparam, *paramlen;
4038 const char *mname;
4040 t = tline;
4041 skip_white_(t);
4042 /* if (!tok_type_(t, TOK_ID)) Lino 02/25/02 */
4043 if (!tok_type_(t, TOK_ID) && !tok_type_(t, TOK_PREPROC_ID))
4044 return 0;
4045 mtok = t;
4046 m = is_mmacro(t, &params);
4047 if (m) {
4048 mname = t->text;
4049 } else {
4050 Token *last;
4052 * We have an id which isn't a macro call. We'll assume
4053 * it might be a label; we'll also check to see if a
4054 * colon follows it. Then, if there's another id after
4055 * that lot, we'll check it again for macro-hood.
4057 label = last = t;
4058 t = t->next;
4059 if (tok_type_(t, TOK_WHITESPACE))
4060 last = t, t = t->next;
4061 if (tok_is_(t, ":")) {
4062 dont_prepend = 1;
4063 last = t, t = t->next;
4064 if (tok_type_(t, TOK_WHITESPACE))
4065 last = t, t = t->next;
4067 if (!tok_type_(t, TOK_ID) || (m = is_mmacro(t, &params)) == NULL)
4068 return 0;
4069 last->next = NULL;
4070 mname = t->text;
4071 tline = t;
4075 * Fix up the parameters: this involves stripping leading and
4076 * trailing whitespace, then stripping braces if they are
4077 * present.
4079 for (nparam = 0; params[nparam]; nparam++) ;
4080 paramlen = nparam ? nasm_malloc(nparam * sizeof(*paramlen)) : NULL;
4082 for (i = 0; params[i]; i++) {
4083 int brace = false;
4084 int comma = (!m->plus || i < nparam - 1);
4086 t = params[i];
4087 skip_white_(t);
4088 if (tok_is_(t, "{"))
4089 t = t->next, brace = true, comma = false;
4090 params[i] = t;
4091 paramlen[i] = 0;
4092 while (t) {
4093 if (comma && t->type == TOK_OTHER && !strcmp(t->text, ","))
4094 break; /* ... because we have hit a comma */
4095 if (comma && t->type == TOK_WHITESPACE
4096 && tok_is_(t->next, ","))
4097 break; /* ... or a space then a comma */
4098 if (brace && t->type == TOK_OTHER && !strcmp(t->text, "}"))
4099 break; /* ... or a brace */
4100 t = t->next;
4101 paramlen[i]++;
4106 * OK, we have a MMacro structure together with a set of
4107 * parameters. We must now go through the expansion and push
4108 * copies of each Line on to istk->expansion. Substitution of
4109 * parameter tokens and macro-local tokens doesn't get done
4110 * until the single-line macro substitution process; this is
4111 * because delaying them allows us to change the semantics
4112 * later through %rotate.
4114 * First, push an end marker on to istk->expansion, mark this
4115 * macro as in progress, and set up its invocation-specific
4116 * variables.
4118 ll = nasm_malloc(sizeof(Line));
4119 ll->next = istk->expansion;
4120 ll->finishes = m;
4121 ll->first = NULL;
4122 istk->expansion = ll;
4124 m->in_progress = true;
4125 m->params = params;
4126 m->iline = tline;
4127 m->nparam = nparam;
4128 m->rotate = 0;
4129 m->paramlen = paramlen;
4130 m->unique = unique++;
4131 m->lineno = 0;
4133 m->next_active = istk->mstk;
4134 istk->mstk = m;
4136 for (l = m->expansion; l; l = l->next) {
4137 Token **tail;
4139 ll = nasm_malloc(sizeof(Line));
4140 ll->finishes = NULL;
4141 ll->next = istk->expansion;
4142 istk->expansion = ll;
4143 tail = &ll->first;
4145 for (t = l->first; t; t = t->next) {
4146 Token *x = t;
4147 switch (t->type) {
4148 case TOK_PREPROC_Q:
4149 tt = *tail = new_Token(NULL, TOK_ID, mname, 0);
4150 break;
4151 case TOK_PREPROC_QQ:
4152 tt = *tail = new_Token(NULL, TOK_ID, m->name, 0);
4153 break;
4154 case TOK_PREPROC_ID:
4155 if (t->text[1] == '0' && t->text[2] == '0') {
4156 dont_prepend = -1;
4157 x = label;
4158 if (!x)
4159 continue;
4161 /* fall through */
4162 default:
4163 tt = *tail = new_Token(NULL, x->type, x->text, 0);
4164 break;
4166 tail = &tt->next;
4168 *tail = NULL;
4172 * If we had a label, push it on as the first line of
4173 * the macro expansion.
4175 if (label) {
4176 if (dont_prepend < 0)
4177 free_tlist(startline);
4178 else {
4179 ll = nasm_malloc(sizeof(Line));
4180 ll->finishes = NULL;
4181 ll->next = istk->expansion;
4182 istk->expansion = ll;
4183 ll->first = startline;
4184 if (!dont_prepend) {
4185 while (label->next)
4186 label = label->next;
4187 label->next = tt = new_Token(NULL, TOK_OTHER, ":", 0);
4192 list->uplevel(m->nolist ? LIST_MACRO_NOLIST : LIST_MACRO);
4194 return 1;
4197 /* The function that actually does the error reporting */
4198 static void verror(int severity, const char *fmt, va_list arg)
4200 char buff[1024];
4202 vsnprintf(buff, sizeof(buff), fmt, arg);
4204 if (istk && istk->mstk && istk->mstk->name)
4205 _error(severity, "(%s:%d) %s", istk->mstk->name,
4206 istk->mstk->lineno, buff);
4207 else
4208 _error(severity, "%s", buff);
4212 * Since preprocessor always operate only on the line that didn't
4213 * arrived yet, we should always use ERR_OFFBY1.
4215 static void error(int severity, const char *fmt, ...)
4217 va_list arg;
4219 /* If we're in a dead branch of IF or something like it, ignore the error */
4220 if (istk && istk->conds && !emitting(istk->conds->state))
4221 return;
4223 va_start(arg, fmt);
4224 verror(severity, fmt, arg);
4225 va_end(arg);
4229 * Because %else etc are evaluated in the state context
4230 * of the previous branch, errors might get lost with error():
4231 * %if 0 ... %else trailing garbage ... %endif
4232 * So %else etc should report errors with this function.
4234 static void error_precond(int severity, const char *fmt, ...)
4236 va_list arg;
4238 /* Only ignore the error if it's really in a dead branch */
4239 if (istk && istk->conds && istk->conds->state == COND_NEVER)
4240 return;
4242 va_start(arg, fmt);
4243 verror(severity, fmt, arg);
4244 va_end(arg);
4247 static void
4248 pp_reset(char *file, int apass, efunc errfunc, evalfunc eval,
4249 ListGen * listgen, StrList **deplist)
4251 Token *t;
4253 _error = errfunc;
4254 cstk = NULL;
4255 istk = nasm_malloc(sizeof(Include));
4256 istk->next = NULL;
4257 istk->conds = NULL;
4258 istk->expansion = NULL;
4259 istk->mstk = NULL;
4260 istk->fp = fopen(file, "r");
4261 istk->fname = NULL;
4262 src_set_fname(nasm_strdup(file));
4263 src_set_linnum(0);
4264 istk->lineinc = 1;
4265 if (!istk->fp)
4266 error(ERR_FATAL|ERR_NOFILE, "unable to open input file `%s'",
4267 file);
4268 defining = NULL;
4269 nested_mac_count = 0;
4270 nested_rep_count = 0;
4271 init_macros();
4272 unique = 0;
4273 if (tasm_compatible_mode) {
4274 stdmacpos = nasm_stdmac;
4275 } else {
4276 stdmacpos = nasm_stdmac_after_tasm;
4278 any_extrastdmac = extrastdmac && *extrastdmac;
4279 do_predef = true;
4280 list = listgen;
4281 evaluate = eval;
4284 * 0 for dependencies, 1 for preparatory passes, 2 for final pass.
4285 * The caller, however, will also pass in 3 for preprocess-only so
4286 * we can set __PASS__ accordingly.
4288 pass = apass > 2 ? 2 : apass;
4290 dephead = deptail = deplist;
4291 if (deplist) {
4292 StrList *sl = nasm_malloc(strlen(file)+1+sizeof sl->next);
4293 sl->next = NULL;
4294 strcpy(sl->str, file);
4295 *deptail = sl;
4296 deptail = &sl->next;
4300 * Define the __PASS__ macro. This is defined here unlike
4301 * all the other builtins, because it is special -- it varies between
4302 * passes.
4304 t = nasm_malloc(sizeof(*t));
4305 t->next = NULL;
4306 make_tok_num(t, apass);
4307 t->a.mac = NULL;
4308 define_smacro(NULL, "__PASS__", true, 0, t);
4311 static char *pp_getline(void)
4313 char *line;
4314 Token *tline;
4316 while (1) {
4318 * Fetch a tokenized line, either from the macro-expansion
4319 * buffer or from the input file.
4321 tline = NULL;
4322 while (istk->expansion && istk->expansion->finishes) {
4323 Line *l = istk->expansion;
4324 if (!l->finishes->name && l->finishes->in_progress > 1) {
4325 Line *ll;
4328 * This is a macro-end marker for a macro with no
4329 * name, which means it's not really a macro at all
4330 * but a %rep block, and the `in_progress' field is
4331 * more than 1, meaning that we still need to
4332 * repeat. (1 means the natural last repetition; 0
4333 * means termination by %exitrep.) We have
4334 * therefore expanded up to the %endrep, and must
4335 * push the whole block on to the expansion buffer
4336 * again. We don't bother to remove the macro-end
4337 * marker: we'd only have to generate another one
4338 * if we did.
4340 l->finishes->in_progress--;
4341 for (l = l->finishes->expansion; l; l = l->next) {
4342 Token *t, *tt, **tail;
4344 ll = nasm_malloc(sizeof(Line));
4345 ll->next = istk->expansion;
4346 ll->finishes = NULL;
4347 ll->first = NULL;
4348 tail = &ll->first;
4350 for (t = l->first; t; t = t->next) {
4351 if (t->text || t->type == TOK_WHITESPACE) {
4352 tt = *tail =
4353 new_Token(NULL, t->type, t->text, 0);
4354 tail = &tt->next;
4358 istk->expansion = ll;
4360 } else {
4362 * Check whether a `%rep' was started and not ended
4363 * within this macro expansion. This can happen and
4364 * should be detected. It's a fatal error because
4365 * I'm too confused to work out how to recover
4366 * sensibly from it.
4368 if (defining) {
4369 if (defining->name)
4370 error(ERR_PANIC,
4371 "defining with name in expansion");
4372 else if (istk->mstk->name)
4373 error(ERR_FATAL,
4374 "`%%rep' without `%%endrep' within"
4375 " expansion of macro `%s'",
4376 istk->mstk->name);
4380 * FIXME: investigate the relationship at this point between
4381 * istk->mstk and l->finishes
4384 MMacro *m = istk->mstk;
4385 istk->mstk = m->next_active;
4386 if (m->name) {
4388 * This was a real macro call, not a %rep, and
4389 * therefore the parameter information needs to
4390 * be freed.
4392 nasm_free(m->params);
4393 free_tlist(m->iline);
4394 nasm_free(m->paramlen);
4395 l->finishes->in_progress = false;
4396 } else
4397 free_mmacro(m);
4399 istk->expansion = l->next;
4400 nasm_free(l);
4401 list->downlevel(LIST_MACRO);
4404 while (1) { /* until we get a line we can use */
4406 if (istk->expansion) { /* from a macro expansion */
4407 char *p;
4408 Line *l = istk->expansion;
4409 if (istk->mstk)
4410 istk->mstk->lineno++;
4411 tline = l->first;
4412 istk->expansion = l->next;
4413 nasm_free(l);
4414 p = detoken(tline, false);
4415 list->line(LIST_MACRO, p);
4416 nasm_free(p);
4417 break;
4419 line = read_line();
4420 if (line) { /* from the current input file */
4421 line = prepreproc(line);
4422 tline = tokenize(line);
4423 nasm_free(line);
4424 break;
4427 * The current file has ended; work down the istk
4430 Include *i = istk;
4431 fclose(i->fp);
4432 if (i->conds)
4433 error(ERR_FATAL,
4434 "expected `%%endif' before end of file");
4435 /* only set line and file name if there's a next node */
4436 if (i->next) {
4437 src_set_linnum(i->lineno);
4438 nasm_free(src_set_fname(i->fname));
4440 istk = i->next;
4441 list->downlevel(LIST_INCLUDE);
4442 nasm_free(i);
4443 if (!istk)
4444 return NULL;
4445 if (istk->expansion && istk->expansion->finishes)
4446 break;
4451 * We must expand MMacro parameters and MMacro-local labels
4452 * _before_ we plunge into directive processing, to cope
4453 * with things like `%define something %1' such as STRUC
4454 * uses. Unless we're _defining_ a MMacro, in which case
4455 * those tokens should be left alone to go into the
4456 * definition; and unless we're in a non-emitting
4457 * condition, in which case we don't want to meddle with
4458 * anything.
4460 if (!defining && !(istk->conds && !emitting(istk->conds->state))
4461 && !(istk->mstk && !istk->mstk->in_progress)) {
4462 tline = expand_mmac_params(tline);
4466 * Check the line to see if it's a preprocessor directive.
4468 if (do_directive(tline) == DIRECTIVE_FOUND) {
4469 continue;
4470 } else if (defining) {
4472 * We're defining a multi-line macro. We emit nothing
4473 * at all, and just
4474 * shove the tokenized line on to the macro definition.
4476 Line *l = nasm_malloc(sizeof(Line));
4477 l->next = defining->expansion;
4478 l->first = tline;
4479 l->finishes = NULL;
4480 defining->expansion = l;
4481 continue;
4482 } else if (istk->conds && !emitting(istk->conds->state)) {
4484 * We're in a non-emitting branch of a condition block.
4485 * Emit nothing at all, not even a blank line: when we
4486 * emerge from the condition we'll give a line-number
4487 * directive so we keep our place correctly.
4489 free_tlist(tline);
4490 continue;
4491 } else if (istk->mstk && !istk->mstk->in_progress) {
4493 * We're in a %rep block which has been terminated, so
4494 * we're walking through to the %endrep without
4495 * emitting anything. Emit nothing at all, not even a
4496 * blank line: when we emerge from the %rep block we'll
4497 * give a line-number directive so we keep our place
4498 * correctly.
4500 free_tlist(tline);
4501 continue;
4502 } else {
4503 tline = expand_smacro(tline);
4504 if (!expand_mmacro(tline)) {
4506 * De-tokenize the line again, and emit it.
4508 line = detoken(tline, true);
4509 free_tlist(tline);
4510 break;
4511 } else {
4512 continue; /* expand_mmacro calls free_tlist */
4517 return line;
4520 static void pp_cleanup(int pass)
4522 if (defining) {
4523 if(defining->name) {
4524 error(ERR_NONFATAL,
4525 "end of file while still defining macro `%s'",
4526 defining->name);
4527 } else {
4528 error(ERR_NONFATAL, "end of file while still in %%rep");
4531 free_mmacro(defining);
4533 while (cstk)
4534 ctx_pop();
4535 free_macros();
4536 while (istk) {
4537 Include *i = istk;
4538 istk = istk->next;
4539 fclose(i->fp);
4540 nasm_free(i->fname);
4541 nasm_free(i);
4543 while (cstk)
4544 ctx_pop();
4545 nasm_free(src_set_fname(NULL));
4546 if (pass == 0) {
4547 IncPath *i;
4548 free_llist(predef);
4549 delete_Blocks();
4550 while ((i = ipath)) {
4551 ipath = i->next;
4552 if (i->path)
4553 nasm_free(i->path);
4554 nasm_free(i);
4559 void pp_include_path(char *path)
4561 IncPath *i;
4563 i = nasm_malloc(sizeof(IncPath));
4564 i->path = path ? nasm_strdup(path) : NULL;
4565 i->next = NULL;
4567 if (ipath != NULL) {
4568 IncPath *j = ipath;
4569 while (j->next != NULL)
4570 j = j->next;
4571 j->next = i;
4572 } else {
4573 ipath = i;
4577 void pp_pre_include(char *fname)
4579 Token *inc, *space, *name;
4580 Line *l;
4582 name = new_Token(NULL, TOK_INTERNAL_STRING, fname, 0);
4583 space = new_Token(name, TOK_WHITESPACE, NULL, 0);
4584 inc = new_Token(space, TOK_PREPROC_ID, "%include", 0);
4586 l = nasm_malloc(sizeof(Line));
4587 l->next = predef;
4588 l->first = inc;
4589 l->finishes = NULL;
4590 predef = l;
4593 void pp_pre_define(char *definition)
4595 Token *def, *space;
4596 Line *l;
4597 char *equals;
4599 equals = strchr(definition, '=');
4600 space = new_Token(NULL, TOK_WHITESPACE, NULL, 0);
4601 def = new_Token(space, TOK_PREPROC_ID, "%define", 0);
4602 if (equals)
4603 *equals = ' ';
4604 space->next = tokenize(definition);
4605 if (equals)
4606 *equals = '=';
4608 l = nasm_malloc(sizeof(Line));
4609 l->next = predef;
4610 l->first = def;
4611 l->finishes = NULL;
4612 predef = l;
4615 void pp_pre_undefine(char *definition)
4617 Token *def, *space;
4618 Line *l;
4620 space = new_Token(NULL, TOK_WHITESPACE, NULL, 0);
4621 def = new_Token(space, TOK_PREPROC_ID, "%undef", 0);
4622 space->next = tokenize(definition);
4624 l = nasm_malloc(sizeof(Line));
4625 l->next = predef;
4626 l->first = def;
4627 l->finishes = NULL;
4628 predef = l;
4632 * Added by Keith Kanios:
4634 * This function is used to assist with "runtime" preprocessor
4635 * directives. (e.g. pp_runtime("%define __BITS__ 64");)
4637 * ERRORS ARE IGNORED HERE, SO MAKE COMPLETELY SURE THAT YOU
4638 * PASS A VALID STRING TO THIS FUNCTION!!!!!
4641 void pp_runtime(char *definition)
4643 Token *def;
4645 def = tokenize(definition);
4646 if(do_directive(def) == NO_DIRECTIVE_FOUND)
4647 free_tlist(def);
4651 void pp_extra_stdmac(macros_t *macros)
4653 extrastdmac = macros;
4656 static void make_tok_num(Token * tok, int64_t val)
4658 char numbuf[20];
4659 snprintf(numbuf, sizeof(numbuf), "%"PRId64"", val);
4660 tok->text = nasm_strdup(numbuf);
4661 tok->type = TOK_NUMBER;
4664 Preproc nasmpp = {
4665 pp_reset,
4666 pp_getline,
4667 pp_cleanup