BR3079550: NASM crash on run-time for OMF output format
[nasm.git] / preproc.c
blob19de89e9f853d161b74dc52ffad2b4f0744dd603
1 /* ----------------------------------------------------------------------- *
3 * Copyright 1996-2010 The NASM Authors - All Rights Reserved
4 * See the file AUTHORS included with the NASM distribution for
5 * the specific copyright holders.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following
9 * conditions are met:
11 * * Redistributions of source code must retain the above copyright
12 * notice, this list of conditions and the following disclaimer.
13 * * Redistributions in binary form must reproduce the above
14 * copyright notice, this list of conditions and the following
15 * disclaimer in the documentation and/or other materials provided
16 * with the distribution.
18 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND
19 * CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES,
20 * INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
21 * MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
22 * DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR
23 * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
24 * SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT
25 * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
26 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
28 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
29 * OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
30 * EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
32 * ----------------------------------------------------------------------- */
35 * preproc.c macro preprocessor for the Netwide Assembler
38 /* Typical flow of text through preproc
40 * pp_getline gets tokenized lines, either
42 * from a macro expansion
44 * or
45 * {
46 * read_line gets raw text from stdmacpos, or predef, or current input file
47 * tokenize converts to tokens
48 * }
50 * expand_mmac_params is used to expand %1 etc., unless a macro is being
51 * defined or a false conditional is being processed
52 * (%0, %1, %+1, %-1, %%foo
54 * do_directive checks for directives
56 * expand_smacro is used to expand single line macros
58 * expand_mmacro is used to expand multi-line macros
60 * detoken is used to convert the line back to text
63 #include "compiler.h"
65 #include <stdio.h>
66 #include <stdarg.h>
67 #include <stdlib.h>
68 #include <stddef.h>
69 #include <string.h>
70 #include <ctype.h>
71 #include <limits.h>
72 #include <inttypes.h>
74 #include "nasm.h"
75 #include "nasmlib.h"
76 #include "preproc.h"
77 #include "hashtbl.h"
78 #include "quote.h"
79 #include "stdscan.h"
80 #include "eval.h"
81 #include "tokens.h"
82 #include "tables.h"
84 typedef struct SMacro SMacro;
85 typedef struct MMacro MMacro;
86 typedef struct MMacroInvocation MMacroInvocation;
87 typedef struct Context Context;
88 typedef struct Token Token;
89 typedef struct Blocks Blocks;
90 typedef struct Line Line;
91 typedef struct Include Include;
92 typedef struct Cond Cond;
93 typedef struct IncPath IncPath;
96 * Note on the storage of both SMacro and MMacros: the hash table
97 * indexes them case-insensitively, and we then have to go through a
98 * linked list of potential case aliases (and, for MMacros, parameter
99 * ranges); this is to preserve the matching semantics of the earlier
100 * code. If the number of case aliases for a specific macro is a
101 * performance issue, you may want to reconsider your coding style.
105 * Store the definition of a single-line macro.
107 struct SMacro {
108 SMacro *next;
109 char *name;
110 bool casesense;
111 bool in_progress;
112 unsigned int nparam;
113 Token *expansion;
117 * Store the definition of a multi-line macro. This is also used to
118 * store the interiors of `%rep...%endrep' blocks, which are
119 * effectively self-re-invoking multi-line macros which simply
120 * don't have a name or bother to appear in the hash tables. %rep
121 * blocks are signified by having a NULL `name' field.
123 * In a MMacro describing a `%rep' block, the `in_progress' field
124 * isn't merely boolean, but gives the number of repeats left to
125 * run.
127 * The `next' field is used for storing MMacros in hash tables; the
128 * `next_active' field is for stacking them on istk entries.
130 * When a MMacro is being expanded, `params', `iline', `nparam',
131 * `paramlen', `rotate' and `unique' are local to the invocation.
133 struct MMacro {
134 MMacro *next;
135 MMacroInvocation *prev; /* previous invocation */
136 char *name;
137 int nparam_min, nparam_max;
138 bool casesense;
139 bool plus; /* is the last parameter greedy? */
140 bool nolist; /* is this macro listing-inhibited? */
141 int64_t in_progress; /* is this macro currently being expanded? */
142 int32_t max_depth; /* maximum number of recursive expansions allowed */
143 Token *dlist; /* All defaults as one list */
144 Token **defaults; /* Parameter default pointers */
145 int ndefs; /* number of default parameters */
146 Line *expansion;
148 MMacro *next_active;
149 MMacro *rep_nest; /* used for nesting %rep */
150 Token **params; /* actual parameters */
151 Token *iline; /* invocation line */
152 unsigned int nparam, rotate;
153 int *paramlen;
154 uint64_t unique;
155 int lineno; /* Current line number on expansion */
156 uint64_t condcnt; /* number of if blocks... */
160 /* Store the definition of a multi-line macro, as defined in a
161 * previous recursive macro expansion.
163 struct MMacroInvocation {
164 MMacroInvocation *prev; /* previous invocation */
165 Token **params; /* actual parameters */
166 Token *iline; /* invocation line */
167 unsigned int nparam, rotate;
168 int *paramlen;
169 uint64_t unique;
170 uint64_t condcnt;
175 * The context stack is composed of a linked list of these.
177 struct Context {
178 Context *next;
179 char *name;
180 struct hash_table localmac;
181 uint32_t number;
185 * This is the internal form which we break input lines up into.
186 * Typically stored in linked lists.
188 * Note that `type' serves a double meaning: TOK_SMAC_PARAM is not
189 * necessarily used as-is, but is intended to denote the number of
190 * the substituted parameter. So in the definition
192 * %define a(x,y) ( (x) & ~(y) )
194 * the token representing `x' will have its type changed to
195 * TOK_SMAC_PARAM, but the one representing `y' will be
196 * TOK_SMAC_PARAM+1.
198 * TOK_INTERNAL_STRING is a dirty hack: it's a single string token
199 * which doesn't need quotes around it. Used in the pre-include
200 * mechanism as an alternative to trying to find a sensible type of
201 * quote to use on the filename we were passed.
203 enum pp_token_type {
204 TOK_NONE = 0, TOK_WHITESPACE, TOK_COMMENT, TOK_ID,
205 TOK_PREPROC_ID, TOK_STRING,
206 TOK_NUMBER, TOK_FLOAT, TOK_SMAC_END, TOK_OTHER,
207 TOK_INTERNAL_STRING,
208 TOK_PREPROC_Q, TOK_PREPROC_QQ,
209 TOK_PASTE, /* %+ */
210 TOK_INDIRECT, /* %[...] */
211 TOK_SMAC_PARAM, /* MUST BE LAST IN THE LIST!!! */
212 TOK_MAX = INT_MAX /* Keep compiler from reducing the range */
215 struct Token {
216 Token *next;
217 char *text;
218 union {
219 SMacro *mac; /* associated macro for TOK_SMAC_END */
220 size_t len; /* scratch length field */
221 } a; /* Auxiliary data */
222 enum pp_token_type type;
226 * Multi-line macro definitions are stored as a linked list of
227 * these, which is essentially a container to allow several linked
228 * lists of Tokens.
230 * Note that in this module, linked lists are treated as stacks
231 * wherever possible. For this reason, Lines are _pushed_ on to the
232 * `expansion' field in MMacro structures, so that the linked list,
233 * if walked, would give the macro lines in reverse order; this
234 * means that we can walk the list when expanding a macro, and thus
235 * push the lines on to the `expansion' field in _istk_ in reverse
236 * order (so that when popped back off they are in the right
237 * order). It may seem cockeyed, and it relies on my design having
238 * an even number of steps in, but it works...
240 * Some of these structures, rather than being actual lines, are
241 * markers delimiting the end of the expansion of a given macro.
242 * This is for use in the cycle-tracking and %rep-handling code.
243 * Such structures have `finishes' non-NULL, and `first' NULL. All
244 * others have `finishes' NULL, but `first' may still be NULL if
245 * the line is blank.
247 struct Line {
248 Line *next;
249 MMacro *finishes;
250 Token *first;
254 * To handle an arbitrary level of file inclusion, we maintain a
255 * stack (ie linked list) of these things.
257 struct Include {
258 Include *next;
259 FILE *fp;
260 Cond *conds;
261 Line *expansion;
262 char *fname;
263 int lineno, lineinc;
264 MMacro *mstk; /* stack of active macros/reps */
268 * Include search path. This is simply a list of strings which get
269 * prepended, in turn, to the name of an include file, in an
270 * attempt to find the file if it's not in the current directory.
272 struct IncPath {
273 IncPath *next;
274 char *path;
278 * Conditional assembly: we maintain a separate stack of these for
279 * each level of file inclusion. (The only reason we keep the
280 * stacks separate is to ensure that a stray `%endif' in a file
281 * included from within the true branch of a `%if' won't terminate
282 * it and cause confusion: instead, rightly, it'll cause an error.)
284 struct Cond {
285 Cond *next;
286 int state;
288 enum {
290 * These states are for use just after %if or %elif: IF_TRUE
291 * means the condition has evaluated to truth so we are
292 * currently emitting, whereas IF_FALSE means we are not
293 * currently emitting but will start doing so if a %else comes
294 * up. In these states, all directives are admissible: %elif,
295 * %else and %endif. (And of course %if.)
297 COND_IF_TRUE, COND_IF_FALSE,
299 * These states come up after a %else: ELSE_TRUE means we're
300 * emitting, and ELSE_FALSE means we're not. In ELSE_* states,
301 * any %elif or %else will cause an error.
303 COND_ELSE_TRUE, COND_ELSE_FALSE,
305 * These states mean that we're not emitting now, and also that
306 * nothing until %endif will be emitted at all. COND_DONE is
307 * used when we've had our moment of emission
308 * and have now started seeing %elifs. COND_NEVER is used when
309 * the condition construct in question is contained within a
310 * non-emitting branch of a larger condition construct,
311 * or if there is an error.
313 COND_DONE, COND_NEVER
315 #define emitting(x) ( (x) == COND_IF_TRUE || (x) == COND_ELSE_TRUE )
318 * These defines are used as the possible return values for do_directive
320 #define NO_DIRECTIVE_FOUND 0
321 #define DIRECTIVE_FOUND 1
324 * This define sets the upper limit for smacro and recursive mmacro
325 * expansions
327 #define DEADMAN_LIMIT (1 << 20)
329 /* max reps */
330 #define REP_LIMIT ((INT64_C(1) << 62))
333 * Condition codes. Note that we use c_ prefix not C_ because C_ is
334 * used in nasm.h for the "real" condition codes. At _this_ level,
335 * we treat CXZ and ECXZ as condition codes, albeit non-invertible
336 * ones, so we need a different enum...
338 static const char * const conditions[] = {
339 "a", "ae", "b", "be", "c", "cxz", "e", "ecxz", "g", "ge", "l", "le",
340 "na", "nae", "nb", "nbe", "nc", "ne", "ng", "nge", "nl", "nle", "no",
341 "np", "ns", "nz", "o", "p", "pe", "po", "rcxz", "s", "z"
343 enum pp_conds {
344 c_A, c_AE, c_B, c_BE, c_C, c_CXZ, c_E, c_ECXZ, c_G, c_GE, c_L, c_LE,
345 c_NA, c_NAE, c_NB, c_NBE, c_NC, c_NE, c_NG, c_NGE, c_NL, c_NLE, c_NO,
346 c_NP, c_NS, c_NZ, c_O, c_P, c_PE, c_PO, c_RCXZ, c_S, c_Z,
347 c_none = -1
349 static const enum pp_conds inverse_ccs[] = {
350 c_NA, c_NAE, c_NB, c_NBE, c_NC, -1, c_NE, -1, c_NG, c_NGE, c_NL, c_NLE,
351 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,
352 c_Z, c_NO, c_NP, c_PO, c_PE, -1, c_NS, c_NZ
356 * Directive names.
358 /* If this is a an IF, ELIF, ELSE or ENDIF keyword */
359 static int is_condition(enum preproc_token arg)
361 return PP_IS_COND(arg) || (arg == PP_ELSE) || (arg == PP_ENDIF);
364 /* For TASM compatibility we need to be able to recognise TASM compatible
365 * conditional compilation directives. Using the NASM pre-processor does
366 * not work, so we look for them specifically from the following list and
367 * then jam in the equivalent NASM directive into the input stream.
370 enum {
371 TM_ARG, TM_ELIF, TM_ELSE, TM_ENDIF, TM_IF, TM_IFDEF, TM_IFDIFI,
372 TM_IFNDEF, TM_INCLUDE, TM_LOCAL
375 static const char * const tasm_directives[] = {
376 "arg", "elif", "else", "endif", "if", "ifdef", "ifdifi",
377 "ifndef", "include", "local"
380 static int StackSize = 4;
381 static char *StackPointer = "ebp";
382 static int ArgOffset = 8;
383 static int LocalOffset = 0;
385 static Context *cstk;
386 static Include *istk;
387 static IncPath *ipath = NULL;
389 static int pass; /* HACK: pass 0 = generate dependencies only */
390 static StrList **dephead, **deptail; /* Dependency list */
392 static uint64_t unique; /* unique identifier numbers */
394 static Line *predef = NULL;
395 static bool do_predef;
397 static ListGen *list;
400 * The current set of multi-line macros we have defined.
402 static struct hash_table mmacros;
405 * The current set of single-line macros we have defined.
407 static struct hash_table smacros;
410 * The multi-line macro we are currently defining, or the %rep
411 * block we are currently reading, if any.
413 static MMacro *defining;
415 static uint64_t nested_mac_count;
416 static uint64_t nested_rep_count;
419 * The number of macro parameters to allocate space for at a time.
421 #define PARAM_DELTA 16
424 * The standard macro set: defined in macros.c in the array nasm_stdmac.
425 * This gives our position in the macro set, when we're processing it.
427 static macros_t *stdmacpos;
430 * The extra standard macros that come from the object format, if
431 * any.
433 static macros_t *extrastdmac = NULL;
434 static bool any_extrastdmac;
437 * Tokens are allocated in blocks to improve speed
439 #define TOKEN_BLOCKSIZE 4096
440 static Token *freeTokens = NULL;
441 struct Blocks {
442 Blocks *next;
443 void *chunk;
446 static Blocks blocks = { NULL, NULL };
449 * Forward declarations.
451 static Token *expand_mmac_params(Token * tline);
452 static Token *expand_smacro(Token * tline);
453 static Token *expand_id(Token * tline);
454 static Context *get_ctx(const char *name, const char **namep,
455 bool all_contexts);
456 static void make_tok_num(Token * tok, int64_t val);
457 static void error(int severity, const char *fmt, ...);
458 static void error_precond(int severity, const char *fmt, ...);
459 static void *new_Block(size_t size);
460 static void delete_Blocks(void);
461 static Token *new_Token(Token * next, enum pp_token_type type,
462 const char *text, int txtlen);
463 static Token *delete_Token(Token * t);
466 * Macros for safe checking of token pointers, avoid *(NULL)
468 #define tok_type_(x,t) ((x) && (x)->type == (t))
469 #define skip_white_(x) if (tok_type_((x), TOK_WHITESPACE)) (x)=(x)->next
470 #define tok_is_(x,v) (tok_type_((x), TOK_OTHER) && !strcmp((x)->text,(v)))
471 #define tok_isnt_(x,v) ((x) && ((x)->type!=TOK_OTHER || strcmp((x)->text,(v))))
474 * nasm_unquote with error if the string contains NUL characters.
475 * If the string contains NUL characters, issue an error and return
476 * the C len, i.e. truncate at the NUL.
478 static size_t nasm_unquote_cstr(char *qstr, enum preproc_token directive)
480 size_t len = nasm_unquote(qstr, NULL);
481 size_t clen = strlen(qstr);
483 if (len != clen)
484 error(ERR_NONFATAL, "NUL character in `%s' directive",
485 pp_directives[directive]);
487 return clen;
491 * In-place reverse a list of tokens.
493 static Token *reverse_tokens(Token *t)
495 Token *prev = NULL;
496 Token *next;
498 while (t) {
499 next = t->next;
500 t->next = prev;
501 prev = t;
502 t = next;
505 return prev;
509 * Handle TASM specific directives, which do not contain a % in
510 * front of them. We do it here because I could not find any other
511 * place to do it for the moment, and it is a hack (ideally it would
512 * be nice to be able to use the NASM pre-processor to do it).
514 static char *check_tasm_directive(char *line)
516 int32_t i, j, k, m, len;
517 char *p, *q, *oldline, oldchar;
519 p = nasm_skip_spaces(line);
521 /* Binary search for the directive name */
522 i = -1;
523 j = ARRAY_SIZE(tasm_directives);
524 q = nasm_skip_word(p);
525 len = q - p;
526 if (len) {
527 oldchar = p[len];
528 p[len] = 0;
529 while (j - i > 1) {
530 k = (j + i) / 2;
531 m = nasm_stricmp(p, tasm_directives[k]);
532 if (m == 0) {
533 /* We have found a directive, so jam a % in front of it
534 * so that NASM will then recognise it as one if it's own.
536 p[len] = oldchar;
537 len = strlen(p);
538 oldline = line;
539 line = nasm_malloc(len + 2);
540 line[0] = '%';
541 if (k == TM_IFDIFI) {
543 * NASM does not recognise IFDIFI, so we convert
544 * it to %if 0. This is not used in NASM
545 * compatible code, but does need to parse for the
546 * TASM macro package.
548 strcpy(line + 1, "if 0");
549 } else {
550 memcpy(line + 1, p, len + 1);
552 nasm_free(oldline);
553 return line;
554 } else if (m < 0) {
555 j = k;
556 } else
557 i = k;
559 p[len] = oldchar;
561 return line;
565 * The pre-preprocessing stage... This function translates line
566 * number indications as they emerge from GNU cpp (`# lineno "file"
567 * flags') into NASM preprocessor line number indications (`%line
568 * lineno file').
570 static char *prepreproc(char *line)
572 int lineno, fnlen;
573 char *fname, *oldline;
575 if (line[0] == '#' && line[1] == ' ') {
576 oldline = line;
577 fname = oldline + 2;
578 lineno = atoi(fname);
579 fname += strspn(fname, "0123456789 ");
580 if (*fname == '"')
581 fname++;
582 fnlen = strcspn(fname, "\"");
583 line = nasm_malloc(20 + fnlen);
584 snprintf(line, 20 + fnlen, "%%line %d %.*s", lineno, fnlen, fname);
585 nasm_free(oldline);
587 if (tasm_compatible_mode)
588 return check_tasm_directive(line);
589 return line;
593 * Free a linked list of tokens.
595 static void free_tlist(Token * list)
597 while (list)
598 list = delete_Token(list);
602 * Free a linked list of lines.
604 static void free_llist(Line * list)
606 Line *l, *tmp;
607 list_for_each_safe(l, tmp, list) {
608 free_tlist(l->first);
609 nasm_free(l);
614 * Free an MMacro
616 static void free_mmacro(MMacro * m)
618 nasm_free(m->name);
619 free_tlist(m->dlist);
620 nasm_free(m->defaults);
621 free_llist(m->expansion);
622 nasm_free(m);
626 * Free all currently defined macros, and free the hash tables
628 static void free_smacro_table(struct hash_table *smt)
630 SMacro *s, *tmp;
631 const char *key;
632 struct hash_tbl_node *it = NULL;
634 while ((s = hash_iterate(smt, &it, &key)) != NULL) {
635 nasm_free((void *)key);
636 list_for_each_safe(s, tmp, s) {
637 nasm_free(s->name);
638 free_tlist(s->expansion);
639 nasm_free(s);
642 hash_free(smt);
645 static void free_mmacro_table(struct hash_table *mmt)
647 MMacro *m, *tmp;
648 const char *key;
649 struct hash_tbl_node *it = NULL;
651 it = NULL;
652 while ((m = hash_iterate(mmt, &it, &key)) != NULL) {
653 nasm_free((void *)key);
654 list_for_each_safe(m ,tmp, m)
655 free_mmacro(m);
657 hash_free(mmt);
660 static void free_macros(void)
662 free_smacro_table(&smacros);
663 free_mmacro_table(&mmacros);
667 * Initialize the hash tables
669 static void init_macros(void)
671 hash_init(&smacros, HASH_LARGE);
672 hash_init(&mmacros, HASH_LARGE);
676 * Pop the context stack.
678 static void ctx_pop(void)
680 Context *c = cstk;
682 cstk = cstk->next;
683 free_smacro_table(&c->localmac);
684 nasm_free(c->name);
685 nasm_free(c);
689 * Search for a key in the hash index; adding it if necessary
690 * (in which case we initialize the data pointer to NULL.)
692 static void **
693 hash_findi_add(struct hash_table *hash, const char *str)
695 struct hash_insert hi;
696 void **r;
697 char *strx;
699 r = hash_findi(hash, str, &hi);
700 if (r)
701 return r;
703 strx = nasm_strdup(str); /* Use a more efficient allocator here? */
704 return hash_add(&hi, strx, NULL);
708 * Like hash_findi, but returns the data element rather than a pointer
709 * to it. Used only when not adding a new element, hence no third
710 * argument.
712 static void *
713 hash_findix(struct hash_table *hash, const char *str)
715 void **p;
717 p = hash_findi(hash, str, NULL);
718 return p ? *p : NULL;
722 * read line from standart macros set,
723 * if there no more left -- return NULL
725 static char *line_from_stdmac(void)
727 unsigned char c;
728 const unsigned char *p = stdmacpos;
729 char *line, *q;
730 size_t len = 0;
732 if (!stdmacpos)
733 return NULL;
735 while ((c = *p++)) {
736 if (c >= 0x80)
737 len += pp_directives_len[c - 0x80] + 1;
738 else
739 len++;
742 line = nasm_malloc(len + 1);
743 q = line;
744 while ((c = *stdmacpos++)) {
745 if (c >= 0x80) {
746 memcpy(q, pp_directives[c - 0x80], pp_directives_len[c - 0x80]);
747 q += pp_directives_len[c - 0x80];
748 *q++ = ' ';
749 } else {
750 *q++ = c;
753 stdmacpos = p;
754 *q = '\0';
756 if (!*stdmacpos) {
757 /* This was the last of the standard macro chain... */
758 stdmacpos = NULL;
759 if (any_extrastdmac) {
760 stdmacpos = extrastdmac;
761 any_extrastdmac = false;
762 } else if (do_predef) {
763 Line *pd, *l;
764 Token *head, **tail, *t;
767 * Nasty hack: here we push the contents of
768 * `predef' on to the top-level expansion stack,
769 * since this is the most convenient way to
770 * implement the pre-include and pre-define
771 * features.
773 list_for_each(pd, predef) {
774 head = NULL;
775 tail = &head;
776 list_for_each(t, pd->first) {
777 *tail = new_Token(NULL, t->type, t->text, 0);
778 tail = &(*tail)->next;
781 l = nasm_malloc(sizeof(Line));
782 l->next = istk->expansion;
783 l->first = head;
784 l->finishes = NULL;
786 istk->expansion = l;
788 do_predef = false;
792 return line;
795 #define BUF_DELTA 512
797 * Read a line from the top file in istk, handling multiple CR/LFs
798 * at the end of the line read, and handling spurious ^Zs. Will
799 * return lines from the standard macro set if this has not already
800 * been done.
802 static char *read_line(void)
804 char *buffer, *p, *q;
805 int bufsize, continued_count;
808 * standart macros set (predefined) goes first
810 p = line_from_stdmac();
811 if (p)
812 return p;
815 * regular read from a file
817 bufsize = BUF_DELTA;
818 buffer = nasm_malloc(BUF_DELTA);
819 p = buffer;
820 continued_count = 0;
821 while (1) {
822 q = fgets(p, bufsize - (p - buffer), istk->fp);
823 if (!q)
824 break;
825 p += strlen(p);
826 if (p > buffer && p[-1] == '\n') {
828 * Convert backslash-CRLF line continuation sequences into
829 * nothing at all (for DOS and Windows)
831 if (((p - 2) > buffer) && (p[-3] == '\\') && (p[-2] == '\r')) {
832 p -= 3;
833 *p = 0;
834 continued_count++;
837 * Also convert backslash-LF line continuation sequences into
838 * nothing at all (for Unix)
840 else if (((p - 1) > buffer) && (p[-2] == '\\')) {
841 p -= 2;
842 *p = 0;
843 continued_count++;
844 } else {
845 break;
848 if (p - buffer > bufsize - 10) {
849 int32_t offset = p - buffer;
850 bufsize += BUF_DELTA;
851 buffer = nasm_realloc(buffer, bufsize);
852 p = buffer + offset; /* prevent stale-pointer problems */
856 if (!q && p == buffer) {
857 nasm_free(buffer);
858 return NULL;
861 src_set_linnum(src_get_linnum() + istk->lineinc +
862 (continued_count * istk->lineinc));
865 * Play safe: remove CRs as well as LFs, if any of either are
866 * present at the end of the line.
868 while (--p >= buffer && (*p == '\n' || *p == '\r'))
869 *p = '\0';
872 * Handle spurious ^Z, which may be inserted into source files
873 * by some file transfer utilities.
875 buffer[strcspn(buffer, "\032")] = '\0';
877 list->line(LIST_READ, buffer);
879 return buffer;
883 * Tokenize a line of text. This is a very simple process since we
884 * don't need to parse the value out of e.g. numeric tokens: we
885 * simply split one string into many.
887 static Token *tokenize(char *line)
889 char c, *p = line;
890 enum pp_token_type type;
891 Token *list = NULL;
892 Token *t, **tail = &list;
894 while (*line) {
895 p = line;
896 if (*p == '%') {
897 p++;
898 if (*p == '+' && !nasm_isdigit(p[1])) {
899 p++;
900 type = TOK_PASTE;
901 } else if (nasm_isdigit(*p) ||
902 ((*p == '-' || *p == '+') && nasm_isdigit(p[1]))) {
903 do {
904 p++;
906 while (nasm_isdigit(*p));
907 type = TOK_PREPROC_ID;
908 } else if (*p == '{') {
909 p++;
910 while (*p && *p != '}') {
911 p[-1] = *p;
912 p++;
914 p[-1] = '\0';
915 if (*p)
916 p++;
917 type = TOK_PREPROC_ID;
918 } else if (*p == '[') {
919 int lvl = 1;
920 line += 2; /* Skip the leading %[ */
921 p++;
922 while (lvl && (c = *p++)) {
923 switch (c) {
924 case ']':
925 lvl--;
926 break;
927 case '%':
928 if (*p == '[')
929 lvl++;
930 break;
931 case '\'':
932 case '\"':
933 case '`':
934 p = nasm_skip_string(p - 1) + 1;
935 break;
936 default:
937 break;
940 p--;
941 if (*p)
942 *p++ = '\0';
943 if (lvl)
944 error(ERR_NONFATAL, "unterminated %[ construct");
945 type = TOK_INDIRECT;
946 } else if (*p == '?') {
947 type = TOK_PREPROC_Q; /* %? */
948 p++;
949 if (*p == '?') {
950 type = TOK_PREPROC_QQ; /* %?? */
951 p++;
953 } else if (*p == '!') {
954 type = TOK_PREPROC_ID;
955 p++;
956 if (isidchar(*p)) {
957 do {
958 p++;
960 while (isidchar(*p));
961 } else if (*p == '\'' || *p == '\"' || *p == '`') {
962 p = nasm_skip_string(p);
963 if (*p)
964 p++;
965 else
966 error(ERR_NONFATAL|ERR_PASS1, "unterminated %! string");
967 } else {
968 /* %! without string or identifier */
969 type = TOK_OTHER; /* Legacy behavior... */
971 } else if (isidchar(*p) ||
972 ((*p == '!' || *p == '%' || *p == '$') &&
973 isidchar(p[1]))) {
974 do {
975 p++;
977 while (isidchar(*p));
978 type = TOK_PREPROC_ID;
979 } else {
980 type = TOK_OTHER;
981 if (*p == '%')
982 p++;
984 } else if (isidstart(*p) || (*p == '$' && isidstart(p[1]))) {
985 type = TOK_ID;
986 p++;
987 while (*p && isidchar(*p))
988 p++;
989 } else if (*p == '\'' || *p == '"' || *p == '`') {
991 * A string token.
993 type = TOK_STRING;
994 p = nasm_skip_string(p);
996 if (*p) {
997 p++;
998 } else {
999 error(ERR_WARNING|ERR_PASS1, "unterminated string");
1000 /* Handling unterminated strings by UNV */
1001 /* type = -1; */
1003 } else if (p[0] == '$' && p[1] == '$') {
1004 type = TOK_OTHER; /* TOKEN_BASE */
1005 p += 2;
1006 } else if (isnumstart(*p)) {
1007 bool is_hex = false;
1008 bool is_float = false;
1009 bool has_e = false;
1010 char c, *r;
1013 * A numeric token.
1016 if (*p == '$') {
1017 p++;
1018 is_hex = true;
1021 for (;;) {
1022 c = *p++;
1024 if (!is_hex && (c == 'e' || c == 'E')) {
1025 has_e = true;
1026 if (*p == '+' || *p == '-') {
1028 * e can only be followed by +/- if it is either a
1029 * prefixed hex number or a floating-point number
1031 p++;
1032 is_float = true;
1034 } else if (c == 'H' || c == 'h' || c == 'X' || c == 'x') {
1035 is_hex = true;
1036 } else if (c == 'P' || c == 'p') {
1037 is_float = true;
1038 if (*p == '+' || *p == '-')
1039 p++;
1040 } else if (isnumchar(c) || c == '_')
1041 ; /* just advance */
1042 else if (c == '.') {
1044 * we need to deal with consequences of the legacy
1045 * parser, like "1.nolist" being two tokens
1046 * (TOK_NUMBER, TOK_ID) here; at least give it
1047 * a shot for now. In the future, we probably need
1048 * a flex-based scanner with proper pattern matching
1049 * to do it as well as it can be done. Nothing in
1050 * the world is going to help the person who wants
1051 * 0x123.p16 interpreted as two tokens, though.
1053 r = p;
1054 while (*r == '_')
1055 r++;
1057 if (nasm_isdigit(*r) || (is_hex && nasm_isxdigit(*r)) ||
1058 (!is_hex && (*r == 'e' || *r == 'E')) ||
1059 (*r == 'p' || *r == 'P')) {
1060 p = r;
1061 is_float = true;
1062 } else
1063 break; /* Terminate the token */
1064 } else
1065 break;
1067 p--; /* Point to first character beyond number */
1069 if (p == line+1 && *line == '$') {
1070 type = TOK_OTHER; /* TOKEN_HERE */
1071 } else {
1072 if (has_e && !is_hex) {
1073 /* 1e13 is floating-point, but 1e13h is not */
1074 is_float = true;
1077 type = is_float ? TOK_FLOAT : TOK_NUMBER;
1079 } else if (nasm_isspace(*p)) {
1080 type = TOK_WHITESPACE;
1081 p = nasm_skip_spaces(p);
1083 * Whitespace just before end-of-line is discarded by
1084 * pretending it's a comment; whitespace just before a
1085 * comment gets lumped into the comment.
1087 if (!*p || *p == ';') {
1088 type = TOK_COMMENT;
1089 while (*p)
1090 p++;
1092 } else if (*p == ';') {
1093 type = TOK_COMMENT;
1094 while (*p)
1095 p++;
1096 } else {
1098 * Anything else is an operator of some kind. We check
1099 * for all the double-character operators (>>, <<, //,
1100 * %%, <=, >=, ==, !=, <>, &&, ||, ^^), but anything
1101 * else is a single-character operator.
1103 type = TOK_OTHER;
1104 if ((p[0] == '>' && p[1] == '>') ||
1105 (p[0] == '<' && p[1] == '<') ||
1106 (p[0] == '/' && p[1] == '/') ||
1107 (p[0] == '<' && p[1] == '=') ||
1108 (p[0] == '>' && p[1] == '=') ||
1109 (p[0] == '=' && p[1] == '=') ||
1110 (p[0] == '!' && p[1] == '=') ||
1111 (p[0] == '<' && p[1] == '>') ||
1112 (p[0] == '&' && p[1] == '&') ||
1113 (p[0] == '|' && p[1] == '|') ||
1114 (p[0] == '^' && p[1] == '^')) {
1115 p++;
1117 p++;
1120 /* Handling unterminated string by UNV */
1121 /*if (type == -1)
1123 *tail = t = new_Token(NULL, TOK_STRING, line, p-line+1);
1124 t->text[p-line] = *line;
1125 tail = &t->next;
1127 else */
1128 if (type != TOK_COMMENT) {
1129 *tail = t = new_Token(NULL, type, line, p - line);
1130 tail = &t->next;
1132 line = p;
1134 return list;
1138 * this function allocates a new managed block of memory and
1139 * returns a pointer to the block. The managed blocks are
1140 * deleted only all at once by the delete_Blocks function.
1142 static void *new_Block(size_t size)
1144 Blocks *b = &blocks;
1146 /* first, get to the end of the linked list */
1147 while (b->next)
1148 b = b->next;
1149 /* now allocate the requested chunk */
1150 b->chunk = nasm_malloc(size);
1152 /* now allocate a new block for the next request */
1153 b->next = nasm_malloc(sizeof(Blocks));
1154 /* and initialize the contents of the new block */
1155 b->next->next = NULL;
1156 b->next->chunk = NULL;
1157 return b->chunk;
1161 * this function deletes all managed blocks of memory
1163 static void delete_Blocks(void)
1165 Blocks *a, *b = &blocks;
1168 * keep in mind that the first block, pointed to by blocks
1169 * is a static and not dynamically allocated, so we don't
1170 * free it.
1172 while (b) {
1173 if (b->chunk)
1174 nasm_free(b->chunk);
1175 a = b;
1176 b = b->next;
1177 if (a != &blocks)
1178 nasm_free(a);
1183 * this function creates a new Token and passes a pointer to it
1184 * back to the caller. It sets the type and text elements, and
1185 * also the a.mac and next elements to NULL.
1187 static Token *new_Token(Token * next, enum pp_token_type type,
1188 const char *text, int txtlen)
1190 Token *t;
1191 int i;
1193 if (!freeTokens) {
1194 freeTokens = (Token *) new_Block(TOKEN_BLOCKSIZE * sizeof(Token));
1195 for (i = 0; i < TOKEN_BLOCKSIZE - 1; i++)
1196 freeTokens[i].next = &freeTokens[i + 1];
1197 freeTokens[i].next = NULL;
1199 t = freeTokens;
1200 freeTokens = t->next;
1201 t->next = next;
1202 t->a.mac = NULL;
1203 t->type = type;
1204 if (type == TOK_WHITESPACE || !text) {
1205 t->text = NULL;
1206 } else {
1207 if (txtlen == 0)
1208 txtlen = strlen(text);
1209 t->text = nasm_malloc(txtlen+1);
1210 memcpy(t->text, text, txtlen);
1211 t->text[txtlen] = '\0';
1213 return t;
1216 static Token *delete_Token(Token * t)
1218 Token *next = t->next;
1219 nasm_free(t->text);
1220 t->next = freeTokens;
1221 freeTokens = t;
1222 return next;
1226 * Convert a line of tokens back into text.
1227 * If expand_locals is not zero, identifiers of the form "%$*xxx"
1228 * will be transformed into ..@ctxnum.xxx
1230 static char *detoken(Token * tlist, bool expand_locals)
1232 Token *t;
1233 char *line, *p;
1234 const char *q;
1235 int len = 0;
1237 list_for_each(t, tlist) {
1238 if (t->type == TOK_PREPROC_ID && t->text[1] == '!') {
1239 char *v;
1240 char *q = t->text;
1242 v = t->text + 2;
1243 if (*v == '\'' || *v == '\"' || *v == '`') {
1244 size_t len = nasm_unquote(v, NULL);
1245 size_t clen = strlen(v);
1247 if (len != clen) {
1248 error(ERR_NONFATAL | ERR_PASS1,
1249 "NUL character in %! string");
1250 v = NULL;
1254 if (v) {
1255 char *p = getenv(v);
1256 if (!p) {
1257 error(ERR_NONFATAL | ERR_PASS1,
1258 "nonexistent environment variable `%s'", v);
1259 p = "";
1261 t->text = nasm_strdup(p);
1263 nasm_free(q);
1266 /* Expand local macros here and not during preprocessing */
1267 if (expand_locals &&
1268 t->type == TOK_PREPROC_ID && t->text &&
1269 t->text[0] == '%' && t->text[1] == '$') {
1270 const char *q;
1271 char *p;
1272 Context *ctx = get_ctx(t->text, &q, false);
1273 if (ctx) {
1274 char buffer[40];
1275 snprintf(buffer, sizeof(buffer), "..@%"PRIu32".", ctx->number);
1276 p = nasm_strcat(buffer, q);
1277 nasm_free(t->text);
1278 t->text = p;
1281 if (t->type == TOK_WHITESPACE)
1282 len++;
1283 else if (t->text)
1284 len += strlen(t->text);
1287 p = line = nasm_malloc(len + 1);
1289 list_for_each(t, tlist) {
1290 if (t->type == TOK_WHITESPACE) {
1291 *p++ = ' ';
1292 } else if (t->text) {
1293 q = t->text;
1294 while (*q)
1295 *p++ = *q++;
1298 *p = '\0';
1300 return line;
1304 * A scanner, suitable for use by the expression evaluator, which
1305 * operates on a line of Tokens. Expects a pointer to a pointer to
1306 * the first token in the line to be passed in as its private_data
1307 * field.
1309 * FIX: This really needs to be unified with stdscan.
1311 static int ppscan(void *private_data, struct tokenval *tokval)
1313 Token **tlineptr = private_data;
1314 Token *tline;
1315 char ourcopy[MAX_KEYWORD+1], *p, *r, *s;
1317 do {
1318 tline = *tlineptr;
1319 *tlineptr = tline ? tline->next : NULL;
1320 } while (tline && (tline->type == TOK_WHITESPACE ||
1321 tline->type == TOK_COMMENT));
1323 if (!tline)
1324 return tokval->t_type = TOKEN_EOS;
1326 tokval->t_charptr = tline->text;
1328 if (tline->text[0] == '$' && !tline->text[1])
1329 return tokval->t_type = TOKEN_HERE;
1330 if (tline->text[0] == '$' && tline->text[1] == '$' && !tline->text[2])
1331 return tokval->t_type = TOKEN_BASE;
1333 if (tline->type == TOK_ID) {
1334 p = tokval->t_charptr = tline->text;
1335 if (p[0] == '$') {
1336 tokval->t_charptr++;
1337 return tokval->t_type = TOKEN_ID;
1340 for (r = p, s = ourcopy; *r; r++) {
1341 if (r >= p+MAX_KEYWORD)
1342 return tokval->t_type = TOKEN_ID; /* Not a keyword */
1343 *s++ = nasm_tolower(*r);
1345 *s = '\0';
1346 /* right, so we have an identifier sitting in temp storage. now,
1347 * is it actually a register or instruction name, or what? */
1348 return nasm_token_hash(ourcopy, tokval);
1351 if (tline->type == TOK_NUMBER) {
1352 bool rn_error;
1353 tokval->t_integer = readnum(tline->text, &rn_error);
1354 tokval->t_charptr = tline->text;
1355 if (rn_error)
1356 return tokval->t_type = TOKEN_ERRNUM;
1357 else
1358 return tokval->t_type = TOKEN_NUM;
1361 if (tline->type == TOK_FLOAT) {
1362 return tokval->t_type = TOKEN_FLOAT;
1365 if (tline->type == TOK_STRING) {
1366 char bq, *ep;
1368 bq = tline->text[0];
1369 tokval->t_charptr = tline->text;
1370 tokval->t_inttwo = nasm_unquote(tline->text, &ep);
1372 if (ep[0] != bq || ep[1] != '\0')
1373 return tokval->t_type = TOKEN_ERRSTR;
1374 else
1375 return tokval->t_type = TOKEN_STR;
1378 if (tline->type == TOK_OTHER) {
1379 if (!strcmp(tline->text, "<<"))
1380 return tokval->t_type = TOKEN_SHL;
1381 if (!strcmp(tline->text, ">>"))
1382 return tokval->t_type = TOKEN_SHR;
1383 if (!strcmp(tline->text, "//"))
1384 return tokval->t_type = TOKEN_SDIV;
1385 if (!strcmp(tline->text, "%%"))
1386 return tokval->t_type = TOKEN_SMOD;
1387 if (!strcmp(tline->text, "=="))
1388 return tokval->t_type = TOKEN_EQ;
1389 if (!strcmp(tline->text, "<>"))
1390 return tokval->t_type = TOKEN_NE;
1391 if (!strcmp(tline->text, "!="))
1392 return tokval->t_type = TOKEN_NE;
1393 if (!strcmp(tline->text, "<="))
1394 return tokval->t_type = TOKEN_LE;
1395 if (!strcmp(tline->text, ">="))
1396 return tokval->t_type = TOKEN_GE;
1397 if (!strcmp(tline->text, "&&"))
1398 return tokval->t_type = TOKEN_DBL_AND;
1399 if (!strcmp(tline->text, "^^"))
1400 return tokval->t_type = TOKEN_DBL_XOR;
1401 if (!strcmp(tline->text, "||"))
1402 return tokval->t_type = TOKEN_DBL_OR;
1406 * We have no other options: just return the first character of
1407 * the token text.
1409 return tokval->t_type = tline->text[0];
1413 * Compare a string to the name of an existing macro; this is a
1414 * simple wrapper which calls either strcmp or nasm_stricmp
1415 * depending on the value of the `casesense' parameter.
1417 static int mstrcmp(const char *p, const char *q, bool casesense)
1419 return casesense ? strcmp(p, q) : nasm_stricmp(p, q);
1423 * Compare a string to the name of an existing macro; this is a
1424 * simple wrapper which calls either strcmp or nasm_stricmp
1425 * depending on the value of the `casesense' parameter.
1427 static int mmemcmp(const char *p, const char *q, size_t l, bool casesense)
1429 return casesense ? memcmp(p, q, l) : nasm_memicmp(p, q, l);
1433 * Return the Context structure associated with a %$ token. Return
1434 * NULL, having _already_ reported an error condition, if the
1435 * context stack isn't deep enough for the supplied number of $
1436 * signs.
1437 * If all_contexts == true, contexts that enclose current are
1438 * also scanned for such smacro, until it is found; if not -
1439 * only the context that directly results from the number of $'s
1440 * in variable's name.
1442 * If "namep" is non-NULL, set it to the pointer to the macro name
1443 * tail, i.e. the part beyond %$...
1445 static Context *get_ctx(const char *name, const char **namep,
1446 bool all_contexts)
1448 Context *ctx;
1449 SMacro *m;
1450 int i;
1452 if (namep)
1453 *namep = name;
1455 if (!name || name[0] != '%' || name[1] != '$')
1456 return NULL;
1458 if (!cstk) {
1459 error(ERR_NONFATAL, "`%s': context stack is empty", name);
1460 return NULL;
1463 name += 2;
1464 ctx = cstk;
1465 i = 0;
1466 while (ctx && *name == '$') {
1467 name++;
1468 i++;
1469 ctx = ctx->next;
1471 if (!ctx) {
1472 error(ERR_NONFATAL, "`%s': context stack is only"
1473 " %d level%s deep", name, i, (i == 1 ? "" : "s"));
1474 return NULL;
1477 if (namep)
1478 *namep = name;
1480 if (!all_contexts)
1481 return ctx;
1484 * NOTE: In 2.10 we will not need lookup in extarnal
1485 * contexts, so this is a gentle way to inform users
1486 * about their source code need to be updated
1489 /* first round -- check the current context */
1490 m = hash_findix(&ctx->localmac, name);
1491 while (m) {
1492 if (!mstrcmp(m->name, name, m->casesense))
1493 return ctx;
1494 m = m->next;
1497 /* second round - external contexts */
1498 while ((ctx = ctx->next)) {
1499 /* Search for this smacro in found context */
1500 m = hash_findix(&ctx->localmac, name);
1501 while (m) {
1502 if (!mstrcmp(m->name, name, m->casesense)) {
1503 /* NOTE: deprecated as of 2.10 */
1504 static int once = 0;
1505 if (!once) {
1506 error(ERR_WARNING, "context-local macro expansion"
1507 " fall-through (automatic searching of outer"
1508 " contexts) will be deprecated starting in"
1509 " NASM 2.10, please see the NASM Manual for"
1510 " more information");
1511 once = 1;
1513 error(ERR_WARNING, "`%s': context-local macro expansion fall-through", name);
1514 return ctx;
1516 m = m->next;
1520 return NULL;
1524 * Check to see if a file is already in a string list
1526 static bool in_list(const StrList *list, const char *str)
1528 while (list) {
1529 if (!strcmp(list->str, str))
1530 return true;
1531 list = list->next;
1533 return false;
1537 * Open an include file. This routine must always return a valid
1538 * file pointer if it returns - it's responsible for throwing an
1539 * ERR_FATAL and bombing out completely if not. It should also try
1540 * the include path one by one until it finds the file or reaches
1541 * the end of the path.
1543 static FILE *inc_fopen(const char *file, StrList **dhead, StrList ***dtail,
1544 bool missing_ok)
1546 FILE *fp;
1547 char *prefix = "";
1548 IncPath *ip = ipath;
1549 int len = strlen(file);
1550 size_t prefix_len = 0;
1551 StrList *sl;
1553 while (1) {
1554 sl = nasm_malloc(prefix_len+len+1+sizeof sl->next);
1555 memcpy(sl->str, prefix, prefix_len);
1556 memcpy(sl->str+prefix_len, file, len+1);
1557 fp = fopen(sl->str, "r");
1558 if (fp && dhead && !in_list(*dhead, sl->str)) {
1559 sl->next = NULL;
1560 **dtail = sl;
1561 *dtail = &sl->next;
1562 } else {
1563 nasm_free(sl);
1565 if (fp)
1566 return fp;
1567 if (!ip) {
1568 if (!missing_ok)
1569 break;
1570 prefix = NULL;
1571 } else {
1572 prefix = ip->path;
1573 ip = ip->next;
1575 if (prefix) {
1576 prefix_len = strlen(prefix);
1577 } else {
1578 /* -MG given and file not found */
1579 if (dhead && !in_list(*dhead, file)) {
1580 sl = nasm_malloc(len+1+sizeof sl->next);
1581 sl->next = NULL;
1582 strcpy(sl->str, file);
1583 **dtail = sl;
1584 *dtail = &sl->next;
1586 return NULL;
1590 error(ERR_FATAL, "unable to open include file `%s'", file);
1591 return NULL;
1595 * Determine if we should warn on defining a single-line macro of
1596 * name `name', with `nparam' parameters. If nparam is 0 or -1, will
1597 * return true if _any_ single-line macro of that name is defined.
1598 * Otherwise, will return true if a single-line macro with either
1599 * `nparam' or no parameters is defined.
1601 * If a macro with precisely the right number of parameters is
1602 * defined, or nparam is -1, the address of the definition structure
1603 * will be returned in `defn'; otherwise NULL will be returned. If `defn'
1604 * is NULL, no action will be taken regarding its contents, and no
1605 * error will occur.
1607 * Note that this is also called with nparam zero to resolve
1608 * `ifdef'.
1610 * If you already know which context macro belongs to, you can pass
1611 * the context pointer as first parameter; if you won't but name begins
1612 * with %$ the context will be automatically computed. If all_contexts
1613 * is true, macro will be searched in outer contexts as well.
1615 static bool
1616 smacro_defined(Context * ctx, const char *name, int nparam, SMacro ** defn,
1617 bool nocase)
1619 struct hash_table *smtbl;
1620 SMacro *m;
1622 if (ctx) {
1623 smtbl = &ctx->localmac;
1624 } else if (name[0] == '%' && name[1] == '$') {
1625 if (cstk)
1626 ctx = get_ctx(name, &name, false);
1627 if (!ctx)
1628 return false; /* got to return _something_ */
1629 smtbl = &ctx->localmac;
1630 } else {
1631 smtbl = &smacros;
1633 m = (SMacro *) hash_findix(smtbl, name);
1635 while (m) {
1636 if (!mstrcmp(m->name, name, m->casesense && nocase) &&
1637 (nparam <= 0 || m->nparam == 0 || nparam == (int) m->nparam)) {
1638 if (defn) {
1639 if (nparam == (int) m->nparam || nparam == -1)
1640 *defn = m;
1641 else
1642 *defn = NULL;
1644 return true;
1646 m = m->next;
1649 return false;
1653 * Count and mark off the parameters in a multi-line macro call.
1654 * This is called both from within the multi-line macro expansion
1655 * code, and also to mark off the default parameters when provided
1656 * in a %macro definition line.
1658 static void count_mmac_params(Token * t, int *nparam, Token *** params)
1660 int paramsize, brace;
1662 *nparam = paramsize = 0;
1663 *params = NULL;
1664 while (t) {
1665 /* +1: we need space for the final NULL */
1666 if (*nparam+1 >= paramsize) {
1667 paramsize += PARAM_DELTA;
1668 *params = nasm_realloc(*params, sizeof(**params) * paramsize);
1670 skip_white_(t);
1671 brace = false;
1672 if (tok_is_(t, "{"))
1673 brace = true;
1674 (*params)[(*nparam)++] = t;
1675 while (tok_isnt_(t, brace ? "}" : ","))
1676 t = t->next;
1677 if (t) { /* got a comma/brace */
1678 t = t->next;
1679 if (brace) {
1681 * Now we've found the closing brace, look further
1682 * for the comma.
1684 skip_white_(t);
1685 if (tok_isnt_(t, ",")) {
1686 error(ERR_NONFATAL,
1687 "braces do not enclose all of macro parameter");
1688 while (tok_isnt_(t, ","))
1689 t = t->next;
1691 if (t)
1692 t = t->next; /* eat the comma */
1699 * Determine whether one of the various `if' conditions is true or
1700 * not.
1702 * We must free the tline we get passed.
1704 static bool if_condition(Token * tline, enum preproc_token ct)
1706 enum pp_conditional i = PP_COND(ct);
1707 bool j;
1708 Token *t, *tt, **tptr, *origline;
1709 struct tokenval tokval;
1710 expr *evalresult;
1711 enum pp_token_type needtype;
1712 char *p;
1714 origline = tline;
1716 switch (i) {
1717 case PPC_IFCTX:
1718 j = false; /* have we matched yet? */
1719 while (true) {
1720 skip_white_(tline);
1721 if (!tline)
1722 break;
1723 if (tline->type != TOK_ID) {
1724 error(ERR_NONFATAL,
1725 "`%s' expects context identifiers", pp_directives[ct]);
1726 free_tlist(origline);
1727 return -1;
1729 if (cstk && cstk->name && !nasm_stricmp(tline->text, cstk->name))
1730 j = true;
1731 tline = tline->next;
1733 break;
1735 case PPC_IFDEF:
1736 j = false; /* have we matched yet? */
1737 while (tline) {
1738 skip_white_(tline);
1739 if (!tline || (tline->type != TOK_ID &&
1740 (tline->type != TOK_PREPROC_ID ||
1741 tline->text[1] != '$'))) {
1742 error(ERR_NONFATAL,
1743 "`%s' expects macro identifiers", pp_directives[ct]);
1744 goto fail;
1746 if (smacro_defined(NULL, tline->text, 0, NULL, true))
1747 j = true;
1748 tline = tline->next;
1750 break;
1752 case PPC_IFENV:
1753 tline = expand_smacro(tline);
1754 j = false; /* have we matched yet? */
1755 while (tline) {
1756 skip_white_(tline);
1757 if (!tline || (tline->type != TOK_ID &&
1758 tline->type != TOK_STRING &&
1759 (tline->type != TOK_PREPROC_ID ||
1760 tline->text[1] != '!'))) {
1761 error(ERR_NONFATAL,
1762 "`%s' expects environment variable names",
1763 pp_directives[ct]);
1764 goto fail;
1766 p = tline->text;
1767 if (tline->type == TOK_PREPROC_ID)
1768 p += 2; /* Skip leading %! */
1769 if (*p == '\'' || *p == '\"' || *p == '`')
1770 nasm_unquote_cstr(p, ct);
1771 if (getenv(p))
1772 j = true;
1773 tline = tline->next;
1775 break;
1777 case PPC_IFIDN:
1778 case PPC_IFIDNI:
1779 tline = expand_smacro(tline);
1780 t = tt = tline;
1781 while (tok_isnt_(tt, ","))
1782 tt = tt->next;
1783 if (!tt) {
1784 error(ERR_NONFATAL,
1785 "`%s' expects two comma-separated arguments",
1786 pp_directives[ct]);
1787 goto fail;
1789 tt = tt->next;
1790 j = true; /* assume equality unless proved not */
1791 while ((t->type != TOK_OTHER || strcmp(t->text, ",")) && tt) {
1792 if (tt->type == TOK_OTHER && !strcmp(tt->text, ",")) {
1793 error(ERR_NONFATAL, "`%s': more than one comma on line",
1794 pp_directives[ct]);
1795 goto fail;
1797 if (t->type == TOK_WHITESPACE) {
1798 t = t->next;
1799 continue;
1801 if (tt->type == TOK_WHITESPACE) {
1802 tt = tt->next;
1803 continue;
1805 if (tt->type != t->type) {
1806 j = false; /* found mismatching tokens */
1807 break;
1809 /* When comparing strings, need to unquote them first */
1810 if (t->type == TOK_STRING) {
1811 size_t l1 = nasm_unquote(t->text, NULL);
1812 size_t l2 = nasm_unquote(tt->text, NULL);
1814 if (l1 != l2) {
1815 j = false;
1816 break;
1818 if (mmemcmp(t->text, tt->text, l1, i == PPC_IFIDN)) {
1819 j = false;
1820 break;
1822 } else if (mstrcmp(tt->text, t->text, i == PPC_IFIDN) != 0) {
1823 j = false; /* found mismatching tokens */
1824 break;
1827 t = t->next;
1828 tt = tt->next;
1830 if ((t->type != TOK_OTHER || strcmp(t->text, ",")) || tt)
1831 j = false; /* trailing gunk on one end or other */
1832 break;
1834 case PPC_IFMACRO:
1836 bool found = false;
1837 MMacro searching, *mmac;
1839 skip_white_(tline);
1840 tline = expand_id(tline);
1841 if (!tok_type_(tline, TOK_ID)) {
1842 error(ERR_NONFATAL,
1843 "`%s' expects a macro name", pp_directives[ct]);
1844 goto fail;
1846 searching.name = nasm_strdup(tline->text);
1847 searching.casesense = true;
1848 searching.plus = false;
1849 searching.nolist = false;
1850 searching.in_progress = 0;
1851 searching.max_depth = 0;
1852 searching.rep_nest = NULL;
1853 searching.nparam_min = 0;
1854 searching.nparam_max = INT_MAX;
1855 tline = expand_smacro(tline->next);
1856 skip_white_(tline);
1857 if (!tline) {
1858 } else if (!tok_type_(tline, TOK_NUMBER)) {
1859 error(ERR_NONFATAL,
1860 "`%s' expects a parameter count or nothing",
1861 pp_directives[ct]);
1862 } else {
1863 searching.nparam_min = searching.nparam_max =
1864 readnum(tline->text, &j);
1865 if (j)
1866 error(ERR_NONFATAL,
1867 "unable to parse parameter count `%s'",
1868 tline->text);
1870 if (tline && tok_is_(tline->next, "-")) {
1871 tline = tline->next->next;
1872 if (tok_is_(tline, "*"))
1873 searching.nparam_max = INT_MAX;
1874 else if (!tok_type_(tline, TOK_NUMBER))
1875 error(ERR_NONFATAL,
1876 "`%s' expects a parameter count after `-'",
1877 pp_directives[ct]);
1878 else {
1879 searching.nparam_max = readnum(tline->text, &j);
1880 if (j)
1881 error(ERR_NONFATAL,
1882 "unable to parse parameter count `%s'",
1883 tline->text);
1884 if (searching.nparam_min > searching.nparam_max)
1885 error(ERR_NONFATAL,
1886 "minimum parameter count exceeds maximum");
1889 if (tline && tok_is_(tline->next, "+")) {
1890 tline = tline->next;
1891 searching.plus = true;
1893 mmac = (MMacro *) hash_findix(&mmacros, searching.name);
1894 while (mmac) {
1895 if (!strcmp(mmac->name, searching.name) &&
1896 (mmac->nparam_min <= searching.nparam_max
1897 || searching.plus)
1898 && (searching.nparam_min <= mmac->nparam_max
1899 || mmac->plus)) {
1900 found = true;
1901 break;
1903 mmac = mmac->next;
1905 if (tline && tline->next)
1906 error(ERR_WARNING|ERR_PASS1,
1907 "trailing garbage after %%ifmacro ignored");
1908 nasm_free(searching.name);
1909 j = found;
1910 break;
1913 case PPC_IFID:
1914 needtype = TOK_ID;
1915 goto iftype;
1916 case PPC_IFNUM:
1917 needtype = TOK_NUMBER;
1918 goto iftype;
1919 case PPC_IFSTR:
1920 needtype = TOK_STRING;
1921 goto iftype;
1923 iftype:
1924 t = tline = expand_smacro(tline);
1926 while (tok_type_(t, TOK_WHITESPACE) ||
1927 (needtype == TOK_NUMBER &&
1928 tok_type_(t, TOK_OTHER) &&
1929 (t->text[0] == '-' || t->text[0] == '+') &&
1930 !t->text[1]))
1931 t = t->next;
1933 j = tok_type_(t, needtype);
1934 break;
1936 case PPC_IFTOKEN:
1937 t = tline = expand_smacro(tline);
1938 while (tok_type_(t, TOK_WHITESPACE))
1939 t = t->next;
1941 j = false;
1942 if (t) {
1943 t = t->next; /* Skip the actual token */
1944 while (tok_type_(t, TOK_WHITESPACE))
1945 t = t->next;
1946 j = !t; /* Should be nothing left */
1948 break;
1950 case PPC_IFEMPTY:
1951 t = tline = expand_smacro(tline);
1952 while (tok_type_(t, TOK_WHITESPACE))
1953 t = t->next;
1955 j = !t; /* Should be empty */
1956 break;
1958 case PPC_IF:
1959 t = tline = expand_smacro(tline);
1960 tptr = &t;
1961 tokval.t_type = TOKEN_INVALID;
1962 evalresult = evaluate(ppscan, tptr, &tokval,
1963 NULL, pass | CRITICAL, error, NULL);
1964 if (!evalresult)
1965 return -1;
1966 if (tokval.t_type)
1967 error(ERR_WARNING|ERR_PASS1,
1968 "trailing garbage after expression ignored");
1969 if (!is_simple(evalresult)) {
1970 error(ERR_NONFATAL,
1971 "non-constant value given to `%s'", pp_directives[ct]);
1972 goto fail;
1974 j = reloc_value(evalresult) != 0;
1975 break;
1977 default:
1978 error(ERR_FATAL,
1979 "preprocessor directive `%s' not yet implemented",
1980 pp_directives[ct]);
1981 goto fail;
1984 free_tlist(origline);
1985 return j ^ PP_NEGATIVE(ct);
1987 fail:
1988 free_tlist(origline);
1989 return -1;
1993 * Common code for defining an smacro
1995 static bool define_smacro(Context *ctx, const char *mname, bool casesense,
1996 int nparam, Token *expansion)
1998 SMacro *smac, **smhead;
1999 struct hash_table *smtbl;
2001 if (smacro_defined(ctx, mname, nparam, &smac, casesense)) {
2002 if (!smac) {
2003 error(ERR_WARNING|ERR_PASS1,
2004 "single-line macro `%s' defined both with and"
2005 " without parameters", mname);
2007 * Some instances of the old code considered this a failure,
2008 * some others didn't. What is the right thing to do here?
2010 free_tlist(expansion);
2011 return false; /* Failure */
2012 } else {
2014 * We're redefining, so we have to take over an
2015 * existing SMacro structure. This means freeing
2016 * what was already in it.
2018 nasm_free(smac->name);
2019 free_tlist(smac->expansion);
2021 } else {
2022 smtbl = ctx ? &ctx->localmac : &smacros;
2023 smhead = (SMacro **) hash_findi_add(smtbl, mname);
2024 smac = nasm_malloc(sizeof(SMacro));
2025 smac->next = *smhead;
2026 *smhead = smac;
2028 smac->name = nasm_strdup(mname);
2029 smac->casesense = casesense;
2030 smac->nparam = nparam;
2031 smac->expansion = expansion;
2032 smac->in_progress = false;
2033 return true; /* Success */
2037 * Undefine an smacro
2039 static void undef_smacro(Context *ctx, const char *mname)
2041 SMacro **smhead, *s, **sp;
2042 struct hash_table *smtbl;
2044 smtbl = ctx ? &ctx->localmac : &smacros;
2045 smhead = (SMacro **)hash_findi(smtbl, mname, NULL);
2047 if (smhead) {
2049 * We now have a macro name... go hunt for it.
2051 sp = smhead;
2052 while ((s = *sp) != NULL) {
2053 if (!mstrcmp(s->name, mname, s->casesense)) {
2054 *sp = s->next;
2055 nasm_free(s->name);
2056 free_tlist(s->expansion);
2057 nasm_free(s);
2058 } else {
2059 sp = &s->next;
2066 * Parse a mmacro specification.
2068 static bool parse_mmacro_spec(Token *tline, MMacro *def, const char *directive)
2070 bool err;
2072 tline = tline->next;
2073 skip_white_(tline);
2074 tline = expand_id(tline);
2075 if (!tok_type_(tline, TOK_ID)) {
2076 error(ERR_NONFATAL, "`%s' expects a macro name", directive);
2077 return false;
2080 def->prev = NULL;
2081 def->name = nasm_strdup(tline->text);
2082 def->plus = false;
2083 def->nolist = false;
2084 def->in_progress = 0;
2085 def->rep_nest = NULL;
2086 def->nparam_min = 0;
2087 def->nparam_max = 0;
2089 tline = expand_smacro(tline->next);
2090 skip_white_(tline);
2091 if (!tok_type_(tline, TOK_NUMBER)) {
2092 error(ERR_NONFATAL, "`%s' expects a parameter count", directive);
2093 } else {
2094 def->nparam_min = def->nparam_max =
2095 readnum(tline->text, &err);
2096 if (err)
2097 error(ERR_NONFATAL,
2098 "unable to parse parameter count `%s'", tline->text);
2100 if (tline && tok_is_(tline->next, "-")) {
2101 tline = tline->next->next;
2102 if (tok_is_(tline, "*")) {
2103 def->nparam_max = INT_MAX;
2104 } else if (!tok_type_(tline, TOK_NUMBER)) {
2105 error(ERR_NONFATAL,
2106 "`%s' expects a parameter count after `-'", directive);
2107 } else {
2108 def->nparam_max = readnum(tline->text, &err);
2109 if (err) {
2110 error(ERR_NONFATAL, "unable to parse parameter count `%s'",
2111 tline->text);
2113 if (def->nparam_min > def->nparam_max) {
2114 error(ERR_NONFATAL, "minimum parameter count exceeds maximum");
2118 if (tline && tok_is_(tline->next, "+")) {
2119 tline = tline->next;
2120 def->plus = true;
2122 if (tline && tok_type_(tline->next, TOK_ID) &&
2123 !nasm_stricmp(tline->next->text, ".nolist")) {
2124 tline = tline->next;
2125 def->nolist = true;
2129 * Handle default parameters.
2131 if (tline && tline->next) {
2132 def->dlist = tline->next;
2133 tline->next = NULL;
2134 count_mmac_params(def->dlist, &def->ndefs, &def->defaults);
2135 } else {
2136 def->dlist = NULL;
2137 def->defaults = NULL;
2139 def->expansion = NULL;
2141 if (def->defaults && def->ndefs > def->nparam_max - def->nparam_min &&
2142 !def->plus)
2143 error(ERR_WARNING|ERR_PASS1|ERR_WARN_MDP,
2144 "too many default macro parameters");
2146 return true;
2151 * Decode a size directive
2153 static int parse_size(const char *str) {
2154 static const char *size_names[] =
2155 { "byte", "dword", "oword", "qword", "tword", "word", "yword" };
2156 static const int sizes[] =
2157 { 0, 1, 4, 16, 8, 10, 2, 32 };
2159 return sizes[bsii(str, size_names, ARRAY_SIZE(size_names))+1];
2163 * find and process preprocessor directive in passed line
2164 * Find out if a line contains a preprocessor directive, and deal
2165 * with it if so.
2167 * If a directive _is_ found, it is the responsibility of this routine
2168 * (and not the caller) to free_tlist() the line.
2170 * @param tline a pointer to the current tokeninzed line linked list
2171 * @return DIRECTIVE_FOUND or NO_DIRECTIVE_FOUND
2174 static int do_directive(Token * tline)
2176 enum preproc_token i;
2177 int j;
2178 bool err;
2179 int nparam;
2180 bool nolist;
2181 bool casesense;
2182 int k, m;
2183 int offset;
2184 char *p, *pp;
2185 const char *mname;
2186 Include *inc;
2187 Context *ctx;
2188 Cond *cond;
2189 MMacro *mmac, **mmhead;
2190 Token *t, *tt, *param_start, *macro_start, *last, **tptr, *origline;
2191 Line *l;
2192 struct tokenval tokval;
2193 expr *evalresult;
2194 MMacro *tmp_defining; /* Used when manipulating rep_nest */
2195 int64_t count;
2196 size_t len;
2197 int severity;
2199 origline = tline;
2201 skip_white_(tline);
2202 if (!tline || !tok_type_(tline, TOK_PREPROC_ID) ||
2203 (tline->text[1] == '%' || tline->text[1] == '$'
2204 || tline->text[1] == '!'))
2205 return NO_DIRECTIVE_FOUND;
2207 i = pp_token_hash(tline->text);
2210 * FIXME: We zap execution of PP_RMACRO, PP_IRMACRO, PP_EXITMACRO
2211 * since they are known to be buggy at moment, we need to fix them
2212 * in future release (2.09-2.10)
2214 if (i == PP_RMACRO || i == PP_RMACRO || i == PP_EXITMACRO) {
2215 error(ERR_NONFATAL, "unknown preprocessor directive `%s'",
2216 tline->text);
2217 return NO_DIRECTIVE_FOUND;
2221 * If we're in a non-emitting branch of a condition construct,
2222 * or walking to the end of an already terminated %rep block,
2223 * we should ignore all directives except for condition
2224 * directives.
2226 if (((istk->conds && !emitting(istk->conds->state)) ||
2227 (istk->mstk && !istk->mstk->in_progress)) && !is_condition(i)) {
2228 return NO_DIRECTIVE_FOUND;
2232 * If we're defining a macro or reading a %rep block, we should
2233 * ignore all directives except for %macro/%imacro (which nest),
2234 * %endm/%endmacro, and (only if we're in a %rep block) %endrep.
2235 * If we're in a %rep block, another %rep nests, so should be let through.
2237 if (defining && i != PP_MACRO && i != PP_IMACRO &&
2238 i != PP_RMACRO && i != PP_IRMACRO &&
2239 i != PP_ENDMACRO && i != PP_ENDM &&
2240 (defining->name || (i != PP_ENDREP && i != PP_REP))) {
2241 return NO_DIRECTIVE_FOUND;
2244 if (defining) {
2245 if (i == PP_MACRO || i == PP_IMACRO ||
2246 i == PP_RMACRO || i == PP_IRMACRO) {
2247 nested_mac_count++;
2248 return NO_DIRECTIVE_FOUND;
2249 } else if (nested_mac_count > 0) {
2250 if (i == PP_ENDMACRO) {
2251 nested_mac_count--;
2252 return NO_DIRECTIVE_FOUND;
2255 if (!defining->name) {
2256 if (i == PP_REP) {
2257 nested_rep_count++;
2258 return NO_DIRECTIVE_FOUND;
2259 } else if (nested_rep_count > 0) {
2260 if (i == PP_ENDREP) {
2261 nested_rep_count--;
2262 return NO_DIRECTIVE_FOUND;
2268 switch (i) {
2269 case PP_INVALID:
2270 error(ERR_NONFATAL, "unknown preprocessor directive `%s'",
2271 tline->text);
2272 return NO_DIRECTIVE_FOUND; /* didn't get it */
2274 case PP_STACKSIZE:
2275 /* Directive to tell NASM what the default stack size is. The
2276 * default is for a 16-bit stack, and this can be overriden with
2277 * %stacksize large.
2279 tline = tline->next;
2280 if (tline && tline->type == TOK_WHITESPACE)
2281 tline = tline->next;
2282 if (!tline || tline->type != TOK_ID) {
2283 error(ERR_NONFATAL, "`%%stacksize' missing size parameter");
2284 free_tlist(origline);
2285 return DIRECTIVE_FOUND;
2287 if (nasm_stricmp(tline->text, "flat") == 0) {
2288 /* All subsequent ARG directives are for a 32-bit stack */
2289 StackSize = 4;
2290 StackPointer = "ebp";
2291 ArgOffset = 8;
2292 LocalOffset = 0;
2293 } else if (nasm_stricmp(tline->text, "flat64") == 0) {
2294 /* All subsequent ARG directives are for a 64-bit stack */
2295 StackSize = 8;
2296 StackPointer = "rbp";
2297 ArgOffset = 16;
2298 LocalOffset = 0;
2299 } else if (nasm_stricmp(tline->text, "large") == 0) {
2300 /* All subsequent ARG directives are for a 16-bit stack,
2301 * far function call.
2303 StackSize = 2;
2304 StackPointer = "bp";
2305 ArgOffset = 4;
2306 LocalOffset = 0;
2307 } else if (nasm_stricmp(tline->text, "small") == 0) {
2308 /* All subsequent ARG directives are for a 16-bit stack,
2309 * far function call. We don't support near functions.
2311 StackSize = 2;
2312 StackPointer = "bp";
2313 ArgOffset = 6;
2314 LocalOffset = 0;
2315 } else {
2316 error(ERR_NONFATAL, "`%%stacksize' invalid size type");
2317 free_tlist(origline);
2318 return DIRECTIVE_FOUND;
2320 free_tlist(origline);
2321 return DIRECTIVE_FOUND;
2323 case PP_ARG:
2324 /* TASM like ARG directive to define arguments to functions, in
2325 * the following form:
2327 * ARG arg1:WORD, arg2:DWORD, arg4:QWORD
2329 offset = ArgOffset;
2330 do {
2331 char *arg, directive[256];
2332 int size = StackSize;
2334 /* Find the argument name */
2335 tline = tline->next;
2336 if (tline && tline->type == TOK_WHITESPACE)
2337 tline = tline->next;
2338 if (!tline || tline->type != TOK_ID) {
2339 error(ERR_NONFATAL, "`%%arg' missing argument parameter");
2340 free_tlist(origline);
2341 return DIRECTIVE_FOUND;
2343 arg = tline->text;
2345 /* Find the argument size type */
2346 tline = tline->next;
2347 if (!tline || tline->type != TOK_OTHER
2348 || tline->text[0] != ':') {
2349 error(ERR_NONFATAL,
2350 "Syntax error processing `%%arg' directive");
2351 free_tlist(origline);
2352 return DIRECTIVE_FOUND;
2354 tline = tline->next;
2355 if (!tline || tline->type != TOK_ID) {
2356 error(ERR_NONFATAL, "`%%arg' missing size type parameter");
2357 free_tlist(origline);
2358 return DIRECTIVE_FOUND;
2361 /* Allow macro expansion of type parameter */
2362 tt = tokenize(tline->text);
2363 tt = expand_smacro(tt);
2364 size = parse_size(tt->text);
2365 if (!size) {
2366 error(ERR_NONFATAL,
2367 "Invalid size type for `%%arg' missing directive");
2368 free_tlist(tt);
2369 free_tlist(origline);
2370 return DIRECTIVE_FOUND;
2372 free_tlist(tt);
2374 /* Round up to even stack slots */
2375 size = ALIGN(size, StackSize);
2377 /* Now define the macro for the argument */
2378 snprintf(directive, sizeof(directive), "%%define %s (%s+%d)",
2379 arg, StackPointer, offset);
2380 do_directive(tokenize(directive));
2381 offset += size;
2383 /* Move to the next argument in the list */
2384 tline = tline->next;
2385 if (tline && tline->type == TOK_WHITESPACE)
2386 tline = tline->next;
2387 } while (tline && tline->type == TOK_OTHER && tline->text[0] == ',');
2388 ArgOffset = offset;
2389 free_tlist(origline);
2390 return DIRECTIVE_FOUND;
2392 case PP_LOCAL:
2393 /* TASM like LOCAL directive to define local variables for a
2394 * function, in the following form:
2396 * LOCAL local1:WORD, local2:DWORD, local4:QWORD = LocalSize
2398 * The '= LocalSize' at the end is ignored by NASM, but is
2399 * required by TASM to define the local parameter size (and used
2400 * by the TASM macro package).
2402 offset = LocalOffset;
2403 do {
2404 char *local, directive[256];
2405 int size = StackSize;
2407 /* Find the argument name */
2408 tline = tline->next;
2409 if (tline && tline->type == TOK_WHITESPACE)
2410 tline = tline->next;
2411 if (!tline || tline->type != TOK_ID) {
2412 error(ERR_NONFATAL,
2413 "`%%local' missing argument parameter");
2414 free_tlist(origline);
2415 return DIRECTIVE_FOUND;
2417 local = tline->text;
2419 /* Find the argument size type */
2420 tline = tline->next;
2421 if (!tline || tline->type != TOK_OTHER
2422 || tline->text[0] != ':') {
2423 error(ERR_NONFATAL,
2424 "Syntax error processing `%%local' directive");
2425 free_tlist(origline);
2426 return DIRECTIVE_FOUND;
2428 tline = tline->next;
2429 if (!tline || tline->type != TOK_ID) {
2430 error(ERR_NONFATAL,
2431 "`%%local' missing size type parameter");
2432 free_tlist(origline);
2433 return DIRECTIVE_FOUND;
2436 /* Allow macro expansion of type parameter */
2437 tt = tokenize(tline->text);
2438 tt = expand_smacro(tt);
2439 size = parse_size(tt->text);
2440 if (!size) {
2441 error(ERR_NONFATAL,
2442 "Invalid size type for `%%local' missing directive");
2443 free_tlist(tt);
2444 free_tlist(origline);
2445 return DIRECTIVE_FOUND;
2447 free_tlist(tt);
2449 /* Round up to even stack slots */
2450 size = ALIGN(size, StackSize);
2452 offset += size; /* Negative offset, increment before */
2454 /* Now define the macro for the argument */
2455 snprintf(directive, sizeof(directive), "%%define %s (%s-%d)",
2456 local, StackPointer, offset);
2457 do_directive(tokenize(directive));
2459 /* Now define the assign to setup the enter_c macro correctly */
2460 snprintf(directive, sizeof(directive),
2461 "%%assign %%$localsize %%$localsize+%d", size);
2462 do_directive(tokenize(directive));
2464 /* Move to the next argument in the list */
2465 tline = tline->next;
2466 if (tline && tline->type == TOK_WHITESPACE)
2467 tline = tline->next;
2468 } while (tline && tline->type == TOK_OTHER && tline->text[0] == ',');
2469 LocalOffset = offset;
2470 free_tlist(origline);
2471 return DIRECTIVE_FOUND;
2473 case PP_CLEAR:
2474 if (tline->next)
2475 error(ERR_WARNING|ERR_PASS1,
2476 "trailing garbage after `%%clear' ignored");
2477 free_macros();
2478 init_macros();
2479 free_tlist(origline);
2480 return DIRECTIVE_FOUND;
2482 case PP_DEPEND:
2483 t = tline->next = expand_smacro(tline->next);
2484 skip_white_(t);
2485 if (!t || (t->type != TOK_STRING &&
2486 t->type != TOK_INTERNAL_STRING)) {
2487 error(ERR_NONFATAL, "`%%depend' expects a file name");
2488 free_tlist(origline);
2489 return DIRECTIVE_FOUND; /* but we did _something_ */
2491 if (t->next)
2492 error(ERR_WARNING|ERR_PASS1,
2493 "trailing garbage after `%%depend' ignored");
2494 p = t->text;
2495 if (t->type != TOK_INTERNAL_STRING)
2496 nasm_unquote_cstr(p, i);
2497 if (dephead && !in_list(*dephead, p)) {
2498 StrList *sl = nasm_malloc(strlen(p)+1+sizeof sl->next);
2499 sl->next = NULL;
2500 strcpy(sl->str, p);
2501 *deptail = sl;
2502 deptail = &sl->next;
2504 free_tlist(origline);
2505 return DIRECTIVE_FOUND;
2507 case PP_INCLUDE:
2508 t = tline->next = expand_smacro(tline->next);
2509 skip_white_(t);
2511 if (!t || (t->type != TOK_STRING &&
2512 t->type != TOK_INTERNAL_STRING)) {
2513 error(ERR_NONFATAL, "`%%include' expects a file name");
2514 free_tlist(origline);
2515 return DIRECTIVE_FOUND; /* but we did _something_ */
2517 if (t->next)
2518 error(ERR_WARNING|ERR_PASS1,
2519 "trailing garbage after `%%include' ignored");
2520 p = t->text;
2521 if (t->type != TOK_INTERNAL_STRING)
2522 nasm_unquote_cstr(p, i);
2523 inc = nasm_malloc(sizeof(Include));
2524 inc->next = istk;
2525 inc->conds = NULL;
2526 inc->fp = inc_fopen(p, dephead, &deptail, pass == 0);
2527 if (!inc->fp) {
2528 /* -MG given but file not found */
2529 nasm_free(inc);
2530 } else {
2531 inc->fname = src_set_fname(nasm_strdup(p));
2532 inc->lineno = src_set_linnum(0);
2533 inc->lineinc = 1;
2534 inc->expansion = NULL;
2535 inc->mstk = NULL;
2536 istk = inc;
2537 list->uplevel(LIST_INCLUDE);
2539 free_tlist(origline);
2540 return DIRECTIVE_FOUND;
2542 case PP_USE:
2544 static macros_t *use_pkg;
2545 const char *pkg_macro = NULL;
2547 tline = tline->next;
2548 skip_white_(tline);
2549 tline = expand_id(tline);
2551 if (!tline || (tline->type != TOK_STRING &&
2552 tline->type != TOK_INTERNAL_STRING &&
2553 tline->type != TOK_ID)) {
2554 error(ERR_NONFATAL, "`%%use' expects a package name");
2555 free_tlist(origline);
2556 return DIRECTIVE_FOUND; /* but we did _something_ */
2558 if (tline->next)
2559 error(ERR_WARNING|ERR_PASS1,
2560 "trailing garbage after `%%use' ignored");
2561 if (tline->type == TOK_STRING)
2562 nasm_unquote_cstr(tline->text, i);
2563 use_pkg = nasm_stdmac_find_package(tline->text);
2564 if (!use_pkg)
2565 error(ERR_NONFATAL, "unknown `%%use' package: %s", tline->text);
2566 else
2567 pkg_macro = (char *)use_pkg + 1; /* The first string will be <%define>__USE_*__ */
2568 if (use_pkg && ! smacro_defined(NULL, pkg_macro, 0, NULL, true)) {
2569 /* Not already included, go ahead and include it */
2570 stdmacpos = use_pkg;
2572 free_tlist(origline);
2573 return DIRECTIVE_FOUND;
2575 case PP_PUSH:
2576 case PP_REPL:
2577 case PP_POP:
2578 tline = tline->next;
2579 skip_white_(tline);
2580 tline = expand_id(tline);
2581 if (tline) {
2582 if (!tok_type_(tline, TOK_ID)) {
2583 error(ERR_NONFATAL, "`%s' expects a context identifier",
2584 pp_directives[i]);
2585 free_tlist(origline);
2586 return DIRECTIVE_FOUND; /* but we did _something_ */
2588 if (tline->next)
2589 error(ERR_WARNING|ERR_PASS1,
2590 "trailing garbage after `%s' ignored",
2591 pp_directives[i]);
2592 p = nasm_strdup(tline->text);
2593 } else {
2594 p = NULL; /* Anonymous */
2597 if (i == PP_PUSH) {
2598 ctx = nasm_malloc(sizeof(Context));
2599 ctx->next = cstk;
2600 hash_init(&ctx->localmac, HASH_SMALL);
2601 ctx->name = p;
2602 ctx->number = unique++;
2603 cstk = ctx;
2604 } else {
2605 /* %pop or %repl */
2606 if (!cstk) {
2607 error(ERR_NONFATAL, "`%s': context stack is empty",
2608 pp_directives[i]);
2609 } else if (i == PP_POP) {
2610 if (p && (!cstk->name || nasm_stricmp(p, cstk->name)))
2611 error(ERR_NONFATAL, "`%%pop' in wrong context: %s, "
2612 "expected %s",
2613 cstk->name ? cstk->name : "anonymous", p);
2614 else
2615 ctx_pop();
2616 } else {
2617 /* i == PP_REPL */
2618 nasm_free(cstk->name);
2619 cstk->name = p;
2620 p = NULL;
2622 nasm_free(p);
2624 free_tlist(origline);
2625 return DIRECTIVE_FOUND;
2626 case PP_FATAL:
2627 severity = ERR_FATAL;
2628 goto issue_error;
2629 case PP_ERROR:
2630 severity = ERR_NONFATAL;
2631 goto issue_error;
2632 case PP_WARNING:
2633 severity = ERR_WARNING|ERR_WARN_USER;
2634 goto issue_error;
2636 issue_error:
2638 /* Only error out if this is the final pass */
2639 if (pass != 2 && i != PP_FATAL)
2640 return DIRECTIVE_FOUND;
2642 tline->next = expand_smacro(tline->next);
2643 tline = tline->next;
2644 skip_white_(tline);
2645 t = tline ? tline->next : NULL;
2646 skip_white_(t);
2647 if (tok_type_(tline, TOK_STRING) && !t) {
2648 /* The line contains only a quoted string */
2649 p = tline->text;
2650 nasm_unquote(p, NULL); /* Ignore NUL character truncation */
2651 error(severity, "%s", p);
2652 } else {
2653 /* Not a quoted string, or more than a quoted string */
2654 p = detoken(tline, false);
2655 error(severity, "%s", p);
2656 nasm_free(p);
2658 free_tlist(origline);
2659 return DIRECTIVE_FOUND;
2662 CASE_PP_IF:
2663 if (istk->conds && !emitting(istk->conds->state))
2664 j = COND_NEVER;
2665 else {
2666 j = if_condition(tline->next, i);
2667 tline->next = NULL; /* it got freed */
2668 j = j < 0 ? COND_NEVER : j ? COND_IF_TRUE : COND_IF_FALSE;
2670 cond = nasm_malloc(sizeof(Cond));
2671 cond->next = istk->conds;
2672 cond->state = j;
2673 istk->conds = cond;
2674 if(istk->mstk)
2675 istk->mstk->condcnt ++;
2676 free_tlist(origline);
2677 return DIRECTIVE_FOUND;
2679 CASE_PP_ELIF:
2680 if (!istk->conds)
2681 error(ERR_FATAL, "`%s': no matching `%%if'", pp_directives[i]);
2682 switch(istk->conds->state) {
2683 case COND_IF_TRUE:
2684 istk->conds->state = COND_DONE;
2685 break;
2687 case COND_DONE:
2688 case COND_NEVER:
2689 break;
2691 case COND_ELSE_TRUE:
2692 case COND_ELSE_FALSE:
2693 error_precond(ERR_WARNING|ERR_PASS1,
2694 "`%%elif' after `%%else' ignored");
2695 istk->conds->state = COND_NEVER;
2696 break;
2698 case COND_IF_FALSE:
2700 * IMPORTANT: In the case of %if, we will already have
2701 * called expand_mmac_params(); however, if we're
2702 * processing an %elif we must have been in a
2703 * non-emitting mode, which would have inhibited
2704 * the normal invocation of expand_mmac_params().
2705 * Therefore, we have to do it explicitly here.
2707 j = if_condition(expand_mmac_params(tline->next), i);
2708 tline->next = NULL; /* it got freed */
2709 istk->conds->state =
2710 j < 0 ? COND_NEVER : j ? COND_IF_TRUE : COND_IF_FALSE;
2711 break;
2713 free_tlist(origline);
2714 return DIRECTIVE_FOUND;
2716 case PP_ELSE:
2717 if (tline->next)
2718 error_precond(ERR_WARNING|ERR_PASS1,
2719 "trailing garbage after `%%else' ignored");
2720 if (!istk->conds)
2721 error(ERR_FATAL, "`%%else': no matching `%%if'");
2722 switch(istk->conds->state) {
2723 case COND_IF_TRUE:
2724 case COND_DONE:
2725 istk->conds->state = COND_ELSE_FALSE;
2726 break;
2728 case COND_NEVER:
2729 break;
2731 case COND_IF_FALSE:
2732 istk->conds->state = COND_ELSE_TRUE;
2733 break;
2735 case COND_ELSE_TRUE:
2736 case COND_ELSE_FALSE:
2737 error_precond(ERR_WARNING|ERR_PASS1,
2738 "`%%else' after `%%else' ignored.");
2739 istk->conds->state = COND_NEVER;
2740 break;
2742 free_tlist(origline);
2743 return DIRECTIVE_FOUND;
2745 case PP_ENDIF:
2746 if (tline->next)
2747 error_precond(ERR_WARNING|ERR_PASS1,
2748 "trailing garbage after `%%endif' ignored");
2749 if (!istk->conds)
2750 error(ERR_FATAL, "`%%endif': no matching `%%if'");
2751 cond = istk->conds;
2752 istk->conds = cond->next;
2753 nasm_free(cond);
2754 if(istk->mstk)
2755 istk->mstk->condcnt --;
2756 free_tlist(origline);
2757 return DIRECTIVE_FOUND;
2759 case PP_RMACRO:
2760 case PP_IRMACRO:
2761 case PP_MACRO:
2762 case PP_IMACRO:
2763 if (defining) {
2764 error(ERR_FATAL, "`%s': already defining a macro",
2765 pp_directives[i]);
2766 return DIRECTIVE_FOUND;
2768 defining = nasm_malloc(sizeof(MMacro));
2769 defining->max_depth =
2770 (i == PP_RMACRO) || (i == PP_IRMACRO) ? DEADMAN_LIMIT : 0;
2771 defining->casesense = (i == PP_MACRO) || (i == PP_RMACRO);
2772 if (!parse_mmacro_spec(tline, defining, pp_directives[i])) {
2773 nasm_free(defining);
2774 defining = NULL;
2775 return DIRECTIVE_FOUND;
2778 mmac = (MMacro *) hash_findix(&mmacros, defining->name);
2779 while (mmac) {
2780 if (!strcmp(mmac->name, defining->name) &&
2781 (mmac->nparam_min <= defining->nparam_max
2782 || defining->plus)
2783 && (defining->nparam_min <= mmac->nparam_max
2784 || mmac->plus)) {
2785 error(ERR_WARNING|ERR_PASS1,
2786 "redefining multi-line macro `%s'", defining->name);
2787 return DIRECTIVE_FOUND;
2789 mmac = mmac->next;
2791 free_tlist(origline);
2792 return DIRECTIVE_FOUND;
2794 case PP_ENDM:
2795 case PP_ENDMACRO:
2796 if (! (defining && defining->name)) {
2797 error(ERR_NONFATAL, "`%s': not defining a macro", tline->text);
2798 return DIRECTIVE_FOUND;
2800 mmhead = (MMacro **) hash_findi_add(&mmacros, defining->name);
2801 defining->next = *mmhead;
2802 *mmhead = defining;
2803 defining = NULL;
2804 free_tlist(origline);
2805 return DIRECTIVE_FOUND;
2807 case PP_EXITMACRO:
2809 * We must search along istk->expansion until we hit a
2810 * macro-end marker for a macro with a name. Then we
2811 * bypass all lines between exitmacro and endmacro.
2813 list_for_each(l, istk->expansion)
2814 if (l->finishes && l->finishes->name)
2815 break;
2817 if (l) {
2819 * Remove all conditional entries relative to this
2820 * macro invocation. (safe to do in this context)
2822 for ( ; l->finishes->condcnt > 0; l->finishes->condcnt --) {
2823 cond = istk->conds;
2824 istk->conds = cond->next;
2825 nasm_free(cond);
2827 istk->expansion = l;
2828 } else {
2829 error(ERR_NONFATAL, "`%%exitmacro' not within `%%macro' block");
2831 free_tlist(origline);
2832 return DIRECTIVE_FOUND;
2834 case PP_UNMACRO:
2835 case PP_UNIMACRO:
2837 MMacro **mmac_p;
2838 MMacro spec;
2840 spec.casesense = (i == PP_UNMACRO);
2841 if (!parse_mmacro_spec(tline, &spec, pp_directives[i])) {
2842 return DIRECTIVE_FOUND;
2844 mmac_p = (MMacro **) hash_findi(&mmacros, spec.name, NULL);
2845 while (mmac_p && *mmac_p) {
2846 mmac = *mmac_p;
2847 if (mmac->casesense == spec.casesense &&
2848 !mstrcmp(mmac->name, spec.name, spec.casesense) &&
2849 mmac->nparam_min == spec.nparam_min &&
2850 mmac->nparam_max == spec.nparam_max &&
2851 mmac->plus == spec.plus) {
2852 *mmac_p = mmac->next;
2853 free_mmacro(mmac);
2854 } else {
2855 mmac_p = &mmac->next;
2858 free_tlist(origline);
2859 free_tlist(spec.dlist);
2860 return DIRECTIVE_FOUND;
2863 case PP_ROTATE:
2864 if (tline->next && tline->next->type == TOK_WHITESPACE)
2865 tline = tline->next;
2866 if (!tline->next) {
2867 free_tlist(origline);
2868 error(ERR_NONFATAL, "`%%rotate' missing rotate count");
2869 return DIRECTIVE_FOUND;
2871 t = expand_smacro(tline->next);
2872 tline->next = NULL;
2873 free_tlist(origline);
2874 tline = t;
2875 tptr = &t;
2876 tokval.t_type = TOKEN_INVALID;
2877 evalresult =
2878 evaluate(ppscan, tptr, &tokval, NULL, pass, error, NULL);
2879 free_tlist(tline);
2880 if (!evalresult)
2881 return DIRECTIVE_FOUND;
2882 if (tokval.t_type)
2883 error(ERR_WARNING|ERR_PASS1,
2884 "trailing garbage after expression ignored");
2885 if (!is_simple(evalresult)) {
2886 error(ERR_NONFATAL, "non-constant value given to `%%rotate'");
2887 return DIRECTIVE_FOUND;
2889 mmac = istk->mstk;
2890 while (mmac && !mmac->name) /* avoid mistaking %reps for macros */
2891 mmac = mmac->next_active;
2892 if (!mmac) {
2893 error(ERR_NONFATAL, "`%%rotate' invoked outside a macro call");
2894 } else if (mmac->nparam == 0) {
2895 error(ERR_NONFATAL,
2896 "`%%rotate' invoked within macro without parameters");
2897 } else {
2898 int rotate = mmac->rotate + reloc_value(evalresult);
2900 rotate %= (int)mmac->nparam;
2901 if (rotate < 0)
2902 rotate += mmac->nparam;
2904 mmac->rotate = rotate;
2906 return DIRECTIVE_FOUND;
2908 case PP_REP:
2909 nolist = false;
2910 do {
2911 tline = tline->next;
2912 } while (tok_type_(tline, TOK_WHITESPACE));
2914 if (tok_type_(tline, TOK_ID) &&
2915 nasm_stricmp(tline->text, ".nolist") == 0) {
2916 nolist = true;
2917 do {
2918 tline = tline->next;
2919 } while (tok_type_(tline, TOK_WHITESPACE));
2922 if (tline) {
2923 t = expand_smacro(tline);
2924 tptr = &t;
2925 tokval.t_type = TOKEN_INVALID;
2926 evalresult =
2927 evaluate(ppscan, tptr, &tokval, NULL, pass, error, NULL);
2928 if (!evalresult) {
2929 free_tlist(origline);
2930 return DIRECTIVE_FOUND;
2932 if (tokval.t_type)
2933 error(ERR_WARNING|ERR_PASS1,
2934 "trailing garbage after expression ignored");
2935 if (!is_simple(evalresult)) {
2936 error(ERR_NONFATAL, "non-constant value given to `%%rep'");
2937 return DIRECTIVE_FOUND;
2939 count = reloc_value(evalresult);
2940 if (count >= REP_LIMIT) {
2941 error(ERR_NONFATAL, "`%%rep' value exceeds limit");
2942 count = 0;
2943 } else
2944 count++;
2945 } else {
2946 error(ERR_NONFATAL, "`%%rep' expects a repeat count");
2947 count = 0;
2949 free_tlist(origline);
2951 tmp_defining = defining;
2952 defining = nasm_malloc(sizeof(MMacro));
2953 defining->prev = NULL;
2954 defining->name = NULL; /* flags this macro as a %rep block */
2955 defining->casesense = false;
2956 defining->plus = false;
2957 defining->nolist = nolist;
2958 defining->in_progress = count;
2959 defining->max_depth = 0;
2960 defining->nparam_min = defining->nparam_max = 0;
2961 defining->defaults = NULL;
2962 defining->dlist = NULL;
2963 defining->expansion = NULL;
2964 defining->next_active = istk->mstk;
2965 defining->rep_nest = tmp_defining;
2966 return DIRECTIVE_FOUND;
2968 case PP_ENDREP:
2969 if (!defining || defining->name) {
2970 error(ERR_NONFATAL, "`%%endrep': no matching `%%rep'");
2971 return DIRECTIVE_FOUND;
2975 * Now we have a "macro" defined - although it has no name
2976 * and we won't be entering it in the hash tables - we must
2977 * push a macro-end marker for it on to istk->expansion.
2978 * After that, it will take care of propagating itself (a
2979 * macro-end marker line for a macro which is really a %rep
2980 * block will cause the macro to be re-expanded, complete
2981 * with another macro-end marker to ensure the process
2982 * continues) until the whole expansion is forcibly removed
2983 * from istk->expansion by a %exitrep.
2985 l = nasm_malloc(sizeof(Line));
2986 l->next = istk->expansion;
2987 l->finishes = defining;
2988 l->first = NULL;
2989 istk->expansion = l;
2991 istk->mstk = defining;
2993 list->uplevel(defining->nolist ? LIST_MACRO_NOLIST : LIST_MACRO);
2994 tmp_defining = defining;
2995 defining = defining->rep_nest;
2996 free_tlist(origline);
2997 return DIRECTIVE_FOUND;
2999 case PP_EXITREP:
3001 * We must search along istk->expansion until we hit a
3002 * macro-end marker for a macro with no name. Then we set
3003 * its `in_progress' flag to 0.
3005 list_for_each(l, istk->expansion)
3006 if (l->finishes && !l->finishes->name)
3007 break;
3009 if (l)
3010 l->finishes->in_progress = 1;
3011 else
3012 error(ERR_NONFATAL, "`%%exitrep' not within `%%rep' block");
3013 free_tlist(origline);
3014 return DIRECTIVE_FOUND;
3016 case PP_XDEFINE:
3017 case PP_IXDEFINE:
3018 case PP_DEFINE:
3019 case PP_IDEFINE:
3020 casesense = (i == PP_DEFINE || i == PP_XDEFINE);
3022 tline = tline->next;
3023 skip_white_(tline);
3024 tline = expand_id(tline);
3025 if (!tline || (tline->type != TOK_ID &&
3026 (tline->type != TOK_PREPROC_ID ||
3027 tline->text[1] != '$'))) {
3028 error(ERR_NONFATAL, "`%s' expects a macro identifier",
3029 pp_directives[i]);
3030 free_tlist(origline);
3031 return DIRECTIVE_FOUND;
3034 ctx = get_ctx(tline->text, &mname, false);
3035 last = tline;
3036 param_start = tline = tline->next;
3037 nparam = 0;
3039 /* Expand the macro definition now for %xdefine and %ixdefine */
3040 if ((i == PP_XDEFINE) || (i == PP_IXDEFINE))
3041 tline = expand_smacro(tline);
3043 if (tok_is_(tline, "(")) {
3045 * This macro has parameters.
3048 tline = tline->next;
3049 while (1) {
3050 skip_white_(tline);
3051 if (!tline) {
3052 error(ERR_NONFATAL, "parameter identifier expected");
3053 free_tlist(origline);
3054 return DIRECTIVE_FOUND;
3056 if (tline->type != TOK_ID) {
3057 error(ERR_NONFATAL,
3058 "`%s': parameter identifier expected",
3059 tline->text);
3060 free_tlist(origline);
3061 return DIRECTIVE_FOUND;
3063 tline->type = TOK_SMAC_PARAM + nparam++;
3064 tline = tline->next;
3065 skip_white_(tline);
3066 if (tok_is_(tline, ",")) {
3067 tline = tline->next;
3068 } else {
3069 if (!tok_is_(tline, ")")) {
3070 error(ERR_NONFATAL,
3071 "`)' expected to terminate macro template");
3072 free_tlist(origline);
3073 return DIRECTIVE_FOUND;
3075 break;
3078 last = tline;
3079 tline = tline->next;
3081 if (tok_type_(tline, TOK_WHITESPACE))
3082 last = tline, tline = tline->next;
3083 macro_start = NULL;
3084 last->next = NULL;
3085 t = tline;
3086 while (t) {
3087 if (t->type == TOK_ID) {
3088 list_for_each(tt, param_start)
3089 if (tt->type >= TOK_SMAC_PARAM &&
3090 !strcmp(tt->text, t->text))
3091 t->type = tt->type;
3093 tt = t->next;
3094 t->next = macro_start;
3095 macro_start = t;
3096 t = tt;
3099 * Good. We now have a macro name, a parameter count, and a
3100 * token list (in reverse order) for an expansion. We ought
3101 * to be OK just to create an SMacro, store it, and let
3102 * free_tlist have the rest of the line (which we have
3103 * carefully re-terminated after chopping off the expansion
3104 * from the end).
3106 define_smacro(ctx, mname, casesense, nparam, macro_start);
3107 free_tlist(origline);
3108 return DIRECTIVE_FOUND;
3110 case PP_UNDEF:
3111 tline = tline->next;
3112 skip_white_(tline);
3113 tline = expand_id(tline);
3114 if (!tline || (tline->type != TOK_ID &&
3115 (tline->type != TOK_PREPROC_ID ||
3116 tline->text[1] != '$'))) {
3117 error(ERR_NONFATAL, "`%%undef' expects a macro identifier");
3118 free_tlist(origline);
3119 return DIRECTIVE_FOUND;
3121 if (tline->next) {
3122 error(ERR_WARNING|ERR_PASS1,
3123 "trailing garbage after macro name ignored");
3126 /* Find the context that symbol belongs to */
3127 ctx = get_ctx(tline->text, &mname, false);
3128 undef_smacro(ctx, mname);
3129 free_tlist(origline);
3130 return DIRECTIVE_FOUND;
3132 case PP_DEFSTR:
3133 case PP_IDEFSTR:
3134 casesense = (i == PP_DEFSTR);
3136 tline = tline->next;
3137 skip_white_(tline);
3138 tline = expand_id(tline);
3139 if (!tline || (tline->type != TOK_ID &&
3140 (tline->type != TOK_PREPROC_ID ||
3141 tline->text[1] != '$'))) {
3142 error(ERR_NONFATAL, "`%s' expects a macro identifier",
3143 pp_directives[i]);
3144 free_tlist(origline);
3145 return DIRECTIVE_FOUND;
3148 ctx = get_ctx(tline->text, &mname, false);
3149 last = tline;
3150 tline = expand_smacro(tline->next);
3151 last->next = NULL;
3153 while (tok_type_(tline, TOK_WHITESPACE))
3154 tline = delete_Token(tline);
3156 p = detoken(tline, false);
3157 macro_start = nasm_malloc(sizeof(*macro_start));
3158 macro_start->next = NULL;
3159 macro_start->text = nasm_quote(p, strlen(p));
3160 macro_start->type = TOK_STRING;
3161 macro_start->a.mac = NULL;
3162 nasm_free(p);
3165 * We now have a macro name, an implicit parameter count of
3166 * zero, and a string token to use as an expansion. Create
3167 * and store an SMacro.
3169 define_smacro(ctx, mname, casesense, 0, macro_start);
3170 free_tlist(origline);
3171 return DIRECTIVE_FOUND;
3173 case PP_DEFTOK:
3174 case PP_IDEFTOK:
3175 casesense = (i == PP_DEFTOK);
3177 tline = tline->next;
3178 skip_white_(tline);
3179 tline = expand_id(tline);
3180 if (!tline || (tline->type != TOK_ID &&
3181 (tline->type != TOK_PREPROC_ID ||
3182 tline->text[1] != '$'))) {
3183 error(ERR_NONFATAL,
3184 "`%s' expects a macro identifier as first parameter",
3185 pp_directives[i]);
3186 free_tlist(origline);
3187 return DIRECTIVE_FOUND;
3189 ctx = get_ctx(tline->text, &mname, false);
3190 last = tline;
3191 tline = expand_smacro(tline->next);
3192 last->next = NULL;
3194 t = tline;
3195 while (tok_type_(t, TOK_WHITESPACE))
3196 t = t->next;
3197 /* t should now point to the string */
3198 if (!tok_type_(t, TOK_STRING)) {
3199 error(ERR_NONFATAL,
3200 "`%s` requires string as second parameter",
3201 pp_directives[i]);
3202 free_tlist(tline);
3203 free_tlist(origline);
3204 return DIRECTIVE_FOUND;
3208 * Convert the string to a token stream. Note that smacros
3209 * are stored with the token stream reversed, so we have to
3210 * reverse the output of tokenize().
3212 nasm_unquote_cstr(t->text, i);
3213 macro_start = reverse_tokens(tokenize(t->text));
3216 * We now have a macro name, an implicit parameter count of
3217 * zero, and a numeric token to use as an expansion. Create
3218 * and store an SMacro.
3220 define_smacro(ctx, mname, casesense, 0, macro_start);
3221 free_tlist(tline);
3222 free_tlist(origline);
3223 return DIRECTIVE_FOUND;
3225 case PP_PATHSEARCH:
3227 FILE *fp;
3228 StrList *xsl = NULL;
3229 StrList **xst = &xsl;
3231 casesense = true;
3233 tline = tline->next;
3234 skip_white_(tline);
3235 tline = expand_id(tline);
3236 if (!tline || (tline->type != TOK_ID &&
3237 (tline->type != TOK_PREPROC_ID ||
3238 tline->text[1] != '$'))) {
3239 error(ERR_NONFATAL,
3240 "`%%pathsearch' expects a macro identifier as first parameter");
3241 free_tlist(origline);
3242 return DIRECTIVE_FOUND;
3244 ctx = get_ctx(tline->text, &mname, false);
3245 last = tline;
3246 tline = expand_smacro(tline->next);
3247 last->next = NULL;
3249 t = tline;
3250 while (tok_type_(t, TOK_WHITESPACE))
3251 t = t->next;
3253 if (!t || (t->type != TOK_STRING &&
3254 t->type != TOK_INTERNAL_STRING)) {
3255 error(ERR_NONFATAL, "`%%pathsearch' expects a file name");
3256 free_tlist(tline);
3257 free_tlist(origline);
3258 return DIRECTIVE_FOUND; /* but we did _something_ */
3260 if (t->next)
3261 error(ERR_WARNING|ERR_PASS1,
3262 "trailing garbage after `%%pathsearch' ignored");
3263 p = t->text;
3264 if (t->type != TOK_INTERNAL_STRING)
3265 nasm_unquote(p, NULL);
3267 fp = inc_fopen(p, &xsl, &xst, true);
3268 if (fp) {
3269 p = xsl->str;
3270 fclose(fp); /* Don't actually care about the file */
3272 macro_start = nasm_malloc(sizeof(*macro_start));
3273 macro_start->next = NULL;
3274 macro_start->text = nasm_quote(p, strlen(p));
3275 macro_start->type = TOK_STRING;
3276 macro_start->a.mac = NULL;
3277 if (xsl)
3278 nasm_free(xsl);
3281 * We now have a macro name, an implicit parameter count of
3282 * zero, and a string token to use as an expansion. Create
3283 * and store an SMacro.
3285 define_smacro(ctx, mname, casesense, 0, macro_start);
3286 free_tlist(tline);
3287 free_tlist(origline);
3288 return DIRECTIVE_FOUND;
3291 case PP_STRLEN:
3292 casesense = true;
3294 tline = tline->next;
3295 skip_white_(tline);
3296 tline = expand_id(tline);
3297 if (!tline || (tline->type != TOK_ID &&
3298 (tline->type != TOK_PREPROC_ID ||
3299 tline->text[1] != '$'))) {
3300 error(ERR_NONFATAL,
3301 "`%%strlen' expects a macro identifier as first parameter");
3302 free_tlist(origline);
3303 return DIRECTIVE_FOUND;
3305 ctx = get_ctx(tline->text, &mname, false);
3306 last = tline;
3307 tline = expand_smacro(tline->next);
3308 last->next = NULL;
3310 t = tline;
3311 while (tok_type_(t, TOK_WHITESPACE))
3312 t = t->next;
3313 /* t should now point to the string */
3314 if (!tok_type_(t, TOK_STRING)) {
3315 error(ERR_NONFATAL,
3316 "`%%strlen` requires string as second parameter");
3317 free_tlist(tline);
3318 free_tlist(origline);
3319 return DIRECTIVE_FOUND;
3322 macro_start = nasm_malloc(sizeof(*macro_start));
3323 macro_start->next = NULL;
3324 make_tok_num(macro_start, nasm_unquote(t->text, NULL));
3325 macro_start->a.mac = NULL;
3328 * We now have a macro name, an implicit parameter count of
3329 * zero, and a numeric token to use as an expansion. Create
3330 * and store an SMacro.
3332 define_smacro(ctx, mname, casesense, 0, macro_start);
3333 free_tlist(tline);
3334 free_tlist(origline);
3335 return DIRECTIVE_FOUND;
3337 case PP_STRCAT:
3338 casesense = true;
3340 tline = tline->next;
3341 skip_white_(tline);
3342 tline = expand_id(tline);
3343 if (!tline || (tline->type != TOK_ID &&
3344 (tline->type != TOK_PREPROC_ID ||
3345 tline->text[1] != '$'))) {
3346 error(ERR_NONFATAL,
3347 "`%%strcat' expects a macro identifier as first parameter");
3348 free_tlist(origline);
3349 return DIRECTIVE_FOUND;
3351 ctx = get_ctx(tline->text, &mname, false);
3352 last = tline;
3353 tline = expand_smacro(tline->next);
3354 last->next = NULL;
3356 len = 0;
3357 list_for_each(t, tline) {
3358 switch (t->type) {
3359 case TOK_WHITESPACE:
3360 break;
3361 case TOK_STRING:
3362 len += t->a.len = nasm_unquote(t->text, NULL);
3363 break;
3364 case TOK_OTHER:
3365 if (!strcmp(t->text, ",")) /* permit comma separators */
3366 break;
3367 /* else fall through */
3368 default:
3369 error(ERR_NONFATAL,
3370 "non-string passed to `%%strcat' (%d)", t->type);
3371 free_tlist(tline);
3372 free_tlist(origline);
3373 return DIRECTIVE_FOUND;
3377 p = pp = nasm_malloc(len);
3378 list_for_each(t, tline) {
3379 if (t->type == TOK_STRING) {
3380 memcpy(p, t->text, t->a.len);
3381 p += t->a.len;
3386 * We now have a macro name, an implicit parameter count of
3387 * zero, and a numeric token to use as an expansion. Create
3388 * and store an SMacro.
3390 macro_start = new_Token(NULL, TOK_STRING, NULL, 0);
3391 macro_start->text = nasm_quote(pp, len);
3392 nasm_free(pp);
3393 define_smacro(ctx, mname, casesense, 0, macro_start);
3394 free_tlist(tline);
3395 free_tlist(origline);
3396 return DIRECTIVE_FOUND;
3398 case PP_SUBSTR:
3400 int64_t start, count;
3401 size_t len;
3403 casesense = true;
3405 tline = tline->next;
3406 skip_white_(tline);
3407 tline = expand_id(tline);
3408 if (!tline || (tline->type != TOK_ID &&
3409 (tline->type != TOK_PREPROC_ID ||
3410 tline->text[1] != '$'))) {
3411 error(ERR_NONFATAL,
3412 "`%%substr' expects a macro identifier as first parameter");
3413 free_tlist(origline);
3414 return DIRECTIVE_FOUND;
3416 ctx = get_ctx(tline->text, &mname, false);
3417 last = tline;
3418 tline = expand_smacro(tline->next);
3419 last->next = NULL;
3421 if (tline) /* skip expanded id */
3422 t = tline->next;
3423 while (tok_type_(t, TOK_WHITESPACE))
3424 t = t->next;
3426 /* t should now point to the string */
3427 if (!tok_type_(t, TOK_STRING)) {
3428 error(ERR_NONFATAL,
3429 "`%%substr` requires string as second parameter");
3430 free_tlist(tline);
3431 free_tlist(origline);
3432 return DIRECTIVE_FOUND;
3435 tt = t->next;
3436 tptr = &tt;
3437 tokval.t_type = TOKEN_INVALID;
3438 evalresult = evaluate(ppscan, tptr, &tokval, NULL,
3439 pass, error, NULL);
3440 if (!evalresult) {
3441 free_tlist(tline);
3442 free_tlist(origline);
3443 return DIRECTIVE_FOUND;
3444 } else if (!is_simple(evalresult)) {
3445 error(ERR_NONFATAL, "non-constant value given to `%%substr`");
3446 free_tlist(tline);
3447 free_tlist(origline);
3448 return DIRECTIVE_FOUND;
3450 start = evalresult->value - 1;
3452 while (tok_type_(tt, TOK_WHITESPACE))
3453 tt = tt->next;
3454 if (!tt) {
3455 count = 1; /* Backwards compatibility: one character */
3456 } else {
3457 tokval.t_type = TOKEN_INVALID;
3458 evalresult = evaluate(ppscan, tptr, &tokval, NULL,
3459 pass, error, NULL);
3460 if (!evalresult) {
3461 free_tlist(tline);
3462 free_tlist(origline);
3463 return DIRECTIVE_FOUND;
3464 } else if (!is_simple(evalresult)) {
3465 error(ERR_NONFATAL, "non-constant value given to `%%substr`");
3466 free_tlist(tline);
3467 free_tlist(origline);
3468 return DIRECTIVE_FOUND;
3470 count = evalresult->value;
3473 len = nasm_unquote(t->text, NULL);
3475 /* make start and count being in range */
3476 if (start < 0)
3477 start = 0;
3478 if (count < 0)
3479 count = len + count + 1 - start;
3480 if (start + count > (int64_t)len)
3481 count = len - start;
3482 if (!len || count < 0 || start >=(int64_t)len)
3483 start = -1, count = 0; /* empty string */
3485 macro_start = nasm_malloc(sizeof(*macro_start));
3486 macro_start->next = NULL;
3487 macro_start->text = nasm_quote((start < 0) ? "" : t->text + start, count);
3488 macro_start->type = TOK_STRING;
3489 macro_start->a.mac = NULL;
3492 * We now have a macro name, an implicit parameter count of
3493 * zero, and a numeric token to use as an expansion. Create
3494 * and store an SMacro.
3496 define_smacro(ctx, mname, casesense, 0, macro_start);
3497 free_tlist(tline);
3498 free_tlist(origline);
3499 return DIRECTIVE_FOUND;
3502 case PP_ASSIGN:
3503 case PP_IASSIGN:
3504 casesense = (i == PP_ASSIGN);
3506 tline = tline->next;
3507 skip_white_(tline);
3508 tline = expand_id(tline);
3509 if (!tline || (tline->type != TOK_ID &&
3510 (tline->type != TOK_PREPROC_ID ||
3511 tline->text[1] != '$'))) {
3512 error(ERR_NONFATAL,
3513 "`%%%sassign' expects a macro identifier",
3514 (i == PP_IASSIGN ? "i" : ""));
3515 free_tlist(origline);
3516 return DIRECTIVE_FOUND;
3518 ctx = get_ctx(tline->text, &mname, false);
3519 last = tline;
3520 tline = expand_smacro(tline->next);
3521 last->next = NULL;
3523 t = tline;
3524 tptr = &t;
3525 tokval.t_type = TOKEN_INVALID;
3526 evalresult =
3527 evaluate(ppscan, tptr, &tokval, NULL, pass, error, NULL);
3528 free_tlist(tline);
3529 if (!evalresult) {
3530 free_tlist(origline);
3531 return DIRECTIVE_FOUND;
3534 if (tokval.t_type)
3535 error(ERR_WARNING|ERR_PASS1,
3536 "trailing garbage after expression ignored");
3538 if (!is_simple(evalresult)) {
3539 error(ERR_NONFATAL,
3540 "non-constant value given to `%%%sassign'",
3541 (i == PP_IASSIGN ? "i" : ""));
3542 free_tlist(origline);
3543 return DIRECTIVE_FOUND;
3546 macro_start = nasm_malloc(sizeof(*macro_start));
3547 macro_start->next = NULL;
3548 make_tok_num(macro_start, reloc_value(evalresult));
3549 macro_start->a.mac = NULL;
3552 * We now have a macro name, an implicit parameter count of
3553 * zero, and a numeric token to use as an expansion. Create
3554 * and store an SMacro.
3556 define_smacro(ctx, mname, casesense, 0, macro_start);
3557 free_tlist(origline);
3558 return DIRECTIVE_FOUND;
3560 case PP_LINE:
3562 * Syntax is `%line nnn[+mmm] [filename]'
3564 tline = tline->next;
3565 skip_white_(tline);
3566 if (!tok_type_(tline, TOK_NUMBER)) {
3567 error(ERR_NONFATAL, "`%%line' expects line number");
3568 free_tlist(origline);
3569 return DIRECTIVE_FOUND;
3571 k = readnum(tline->text, &err);
3572 m = 1;
3573 tline = tline->next;
3574 if (tok_is_(tline, "+")) {
3575 tline = tline->next;
3576 if (!tok_type_(tline, TOK_NUMBER)) {
3577 error(ERR_NONFATAL, "`%%line' expects line increment");
3578 free_tlist(origline);
3579 return DIRECTIVE_FOUND;
3581 m = readnum(tline->text, &err);
3582 tline = tline->next;
3584 skip_white_(tline);
3585 src_set_linnum(k);
3586 istk->lineinc = m;
3587 if (tline) {
3588 nasm_free(src_set_fname(detoken(tline, false)));
3590 free_tlist(origline);
3591 return DIRECTIVE_FOUND;
3593 default:
3594 error(ERR_FATAL,
3595 "preprocessor directive `%s' not yet implemented",
3596 pp_directives[i]);
3597 return DIRECTIVE_FOUND;
3602 * Ensure that a macro parameter contains a condition code and
3603 * nothing else. Return the condition code index if so, or -1
3604 * otherwise.
3606 static int find_cc(Token * t)
3608 Token *tt;
3609 int i, j, k, m;
3611 if (!t)
3612 return -1; /* Probably a %+ without a space */
3614 skip_white_(t);
3615 if (t->type != TOK_ID)
3616 return -1;
3617 tt = t->next;
3618 skip_white_(tt);
3619 if (tt && (tt->type != TOK_OTHER || strcmp(tt->text, ",")))
3620 return -1;
3622 i = -1;
3623 j = ARRAY_SIZE(conditions);
3624 while (j - i > 1) {
3625 k = (j + i) / 2;
3626 m = nasm_stricmp(t->text, conditions[k]);
3627 if (m == 0) {
3628 i = k;
3629 j = -2;
3630 break;
3631 } else if (m < 0) {
3632 j = k;
3633 } else
3634 i = k;
3636 if (j != -2)
3637 return -1;
3638 return i;
3641 static bool paste_tokens(Token **head, bool handle_paste_tokens)
3643 Token **tail, *t, *tt;
3644 Token **paste_head;
3645 bool did_paste = false;
3646 char *tmp;
3648 /* Now handle token pasting... */
3649 paste_head = NULL;
3650 tail = head;
3651 while ((t = *tail) && (tt = t->next)) {
3652 switch (t->type) {
3653 case TOK_WHITESPACE:
3654 if (tt->type == TOK_WHITESPACE) {
3655 /* Zap adjacent whitespace tokens */
3656 t->next = delete_Token(tt);
3657 } else {
3658 /* Do not advance paste_head here */
3659 tail = &t->next;
3661 break;
3662 case TOK_ID:
3663 case TOK_PREPROC_ID:
3664 case TOK_NUMBER:
3665 case TOK_FLOAT:
3667 size_t len = 0;
3668 char *tmp, *p;
3670 while (tt && (tt->type == TOK_ID || tt->type == TOK_PREPROC_ID ||
3671 tt->type == TOK_NUMBER || tt->type == TOK_FLOAT ||
3672 tt->type == TOK_OTHER)) {
3673 len += strlen(tt->text);
3674 tt = tt->next;
3678 * Now tt points to the first token after
3679 * the potential paste area...
3681 if (tt != t->next) {
3682 /* We have at least two tokens... */
3683 len += strlen(t->text);
3684 p = tmp = nasm_malloc(len+1);
3686 while (t != tt) {
3687 strcpy(p, t->text);
3688 p = strchr(p, '\0');
3689 t = delete_Token(t);
3692 t = *tail = tokenize(tmp);
3693 nasm_free(tmp);
3695 while (t->next) {
3696 tail = &t->next;
3697 t = t->next;
3699 t->next = tt; /* Attach the remaining token chain */
3701 did_paste = true;
3703 paste_head = tail;
3704 tail = &t->next;
3705 break;
3707 case TOK_PASTE: /* %+ */
3708 if (handle_paste_tokens) {
3709 /* Zap %+ and whitespace tokens to the right */
3710 while (t && (t->type == TOK_WHITESPACE ||
3711 t->type == TOK_PASTE))
3712 t = *tail = delete_Token(t);
3713 if (!paste_head || !t)
3714 break; /* Nothing to paste with */
3715 tail = paste_head;
3716 t = *tail;
3717 tt = t->next;
3718 while (tok_type_(tt, TOK_WHITESPACE))
3719 tt = t->next = delete_Token(tt);
3721 if (tt) {
3722 tmp = nasm_strcat(t->text, tt->text);
3723 delete_Token(t);
3724 tt = delete_Token(tt);
3725 t = *tail = tokenize(tmp);
3726 nasm_free(tmp);
3727 while (t->next) {
3728 tail = &t->next;
3729 t = t->next;
3731 t->next = tt; /* Attach the remaining token chain */
3732 did_paste = true;
3734 paste_head = tail;
3735 tail = &t->next;
3736 break;
3738 /* else fall through */
3739 default:
3740 tail = &t->next;
3741 if (!tok_type_(t->next, TOK_WHITESPACE))
3742 paste_head = tail;
3743 break;
3746 return did_paste;
3750 * expands to a list of tokens from %{x:y}
3752 static Token *expand_mmac_params_range(MMacro *mac, Token *tline, Token ***last)
3754 Token *t = tline, **tt, *tm, *head;
3755 char *pos;
3756 int fst, lst, j, i;
3758 pos = strchr(tline->text, ':');
3759 nasm_assert(pos);
3761 lst = atoi(pos + 1);
3762 fst = atoi(tline->text + 1);
3765 * only macros params are accounted so
3766 * if someone passes %0 -- we reject such
3767 * value(s)
3769 if (lst == 0 || fst == 0)
3770 goto err;
3772 /* the values should be sane */
3773 if ((fst > (int)mac->nparam || fst < (-(int)mac->nparam)) ||
3774 (lst > (int)mac->nparam || lst < (-(int)mac->nparam)))
3775 goto err;
3777 fst = fst < 0 ? fst + (int)mac->nparam + 1: fst;
3778 lst = lst < 0 ? lst + (int)mac->nparam + 1: lst;
3780 /* counted from zero */
3781 fst--, lst--;
3784 * it will be at least one token
3786 tm = mac->params[(fst + mac->rotate) % mac->nparam];
3787 t = new_Token(NULL, tm->type, tm->text, 0);
3788 head = t, tt = &t->next;
3789 if (fst < lst) {
3790 for (i = fst + 1; i <= lst; i++) {
3791 t = new_Token(NULL, TOK_OTHER, ",", 0);
3792 *tt = t, tt = &t->next;
3793 j = (i + mac->rotate) % mac->nparam;
3794 tm = mac->params[j];
3795 t = new_Token(NULL, tm->type, tm->text, 0);
3796 *tt = t, tt = &t->next;
3798 } else {
3799 for (i = fst - 1; i >= lst; i--) {
3800 t = new_Token(NULL, TOK_OTHER, ",", 0);
3801 *tt = t, tt = &t->next;
3802 j = (i + mac->rotate) % mac->nparam;
3803 tm = mac->params[j];
3804 t = new_Token(NULL, tm->type, tm->text, 0);
3805 *tt = t, tt = &t->next;
3809 *last = tt;
3810 return head;
3812 err:
3813 error(ERR_NONFATAL, "`%%{%s}': macro parameters out of range",
3814 &tline->text[1]);
3815 return tline;
3819 * Expand MMacro-local things: parameter references (%0, %n, %+n,
3820 * %-n) and MMacro-local identifiers (%%foo) as well as
3821 * macro indirection (%[...]) and range (%{..:..}).
3823 static Token *expand_mmac_params(Token * tline)
3825 Token *t, *tt, **tail, *thead;
3826 bool changed = false;
3827 char *pos;
3829 tail = &thead;
3830 thead = NULL;
3832 while (tline) {
3833 if (tline->type == TOK_PREPROC_ID &&
3834 (((tline->text[1] == '+' || tline->text[1] == '-') && tline->text[2]) ||
3835 (tline->text[1] >= '0' && tline->text[1] <= '9') ||
3836 tline->text[1] == '%')) {
3837 char *text = NULL;
3838 int type = 0, cc; /* type = 0 to placate optimisers */
3839 char tmpbuf[30];
3840 unsigned int n;
3841 int i;
3842 MMacro *mac;
3844 t = tline;
3845 tline = tline->next;
3847 mac = istk->mstk;
3848 while (mac && !mac->name) /* avoid mistaking %reps for macros */
3849 mac = mac->next_active;
3850 if (!mac) {
3851 error(ERR_NONFATAL, "`%s': not in a macro call", t->text);
3852 } else {
3853 pos = strchr(t->text, ':');
3854 if (!pos) {
3855 switch (t->text[1]) {
3857 * We have to make a substitution of one of the
3858 * forms %1, %-1, %+1, %%foo, %0.
3860 case '0':
3861 type = TOK_NUMBER;
3862 snprintf(tmpbuf, sizeof(tmpbuf), "%d", mac->nparam);
3863 text = nasm_strdup(tmpbuf);
3864 break;
3865 case '%':
3866 type = TOK_ID;
3867 snprintf(tmpbuf, sizeof(tmpbuf), "..@%"PRIu64".",
3868 mac->unique);
3869 text = nasm_strcat(tmpbuf, t->text + 2);
3870 break;
3871 case '-':
3872 n = atoi(t->text + 2) - 1;
3873 if (n >= mac->nparam)
3874 tt = NULL;
3875 else {
3876 if (mac->nparam > 1)
3877 n = (n + mac->rotate) % mac->nparam;
3878 tt = mac->params[n];
3880 cc = find_cc(tt);
3881 if (cc == -1) {
3882 error(ERR_NONFATAL,
3883 "macro parameter %d is not a condition code",
3884 n + 1);
3885 text = NULL;
3886 } else {
3887 type = TOK_ID;
3888 if (inverse_ccs[cc] == -1) {
3889 error(ERR_NONFATAL,
3890 "condition code `%s' is not invertible",
3891 conditions[cc]);
3892 text = NULL;
3893 } else
3894 text = nasm_strdup(conditions[inverse_ccs[cc]]);
3896 break;
3897 case '+':
3898 n = atoi(t->text + 2) - 1;
3899 if (n >= mac->nparam)
3900 tt = NULL;
3901 else {
3902 if (mac->nparam > 1)
3903 n = (n + mac->rotate) % mac->nparam;
3904 tt = mac->params[n];
3906 cc = find_cc(tt);
3907 if (cc == -1) {
3908 error(ERR_NONFATAL,
3909 "macro parameter %d is not a condition code",
3910 n + 1);
3911 text = NULL;
3912 } else {
3913 type = TOK_ID;
3914 text = nasm_strdup(conditions[cc]);
3916 break;
3917 default:
3918 n = atoi(t->text + 1) - 1;
3919 if (n >= mac->nparam)
3920 tt = NULL;
3921 else {
3922 if (mac->nparam > 1)
3923 n = (n + mac->rotate) % mac->nparam;
3924 tt = mac->params[n];
3926 if (tt) {
3927 for (i = 0; i < mac->paramlen[n]; i++) {
3928 *tail = new_Token(NULL, tt->type, tt->text, 0);
3929 tail = &(*tail)->next;
3930 tt = tt->next;
3933 text = NULL; /* we've done it here */
3934 break;
3936 } else {
3938 * seems we have a parameters range here
3940 Token *head, **last;
3941 head = expand_mmac_params_range(mac, t, &last);
3942 if (head != t) {
3943 *tail = head;
3944 *last = tline;
3945 tline = head;
3946 text = NULL;
3950 if (!text) {
3951 delete_Token(t);
3952 } else {
3953 *tail = t;
3954 tail = &t->next;
3955 t->type = type;
3956 nasm_free(t->text);
3957 t->text = text;
3958 t->a.mac = NULL;
3960 changed = true;
3961 continue;
3962 } else if (tline->type == TOK_INDIRECT) {
3963 t = tline;
3964 tline = tline->next;
3965 tt = tokenize(t->text);
3966 tt = expand_mmac_params(tt);
3967 tt = expand_smacro(tt);
3968 *tail = tt;
3969 while (tt) {
3970 tt->a.mac = NULL; /* Necessary? */
3971 tail = &tt->next;
3972 tt = tt->next;
3974 delete_Token(t);
3975 changed = true;
3976 } else if (tline->type == TOK_PREPROC_ID &&
3977 tline->text[0] == '%' &&
3978 tline->text[1] == '$' &&
3979 !tok_type_(tline->next, TOK_WHITESPACE) &&
3980 (tok_type_(tline->next, TOK_ID) ||
3981 tok_type_(tline->next, TOK_PREPROC_ID) ||
3982 tok_type_(tline->next, TOK_NUMBER) ||
3983 tok_type_(tline->next, TOK_OTHER) ||
3984 tok_type_(tline->next, TOK_FLOAT))) {
3986 * In a sake of backward compatibility we allow
3987 * to expand local single macro that early before
3988 * pasting token code have place
3990 * NOTE: that new code MUST use %+ macro to obtain
3991 * same result
3993 t = tline;
3994 tline = tline->next;
3995 tt = tokenize(t->text);
3996 tt = expand_smacro(tt);
3997 *tail = tt;
3998 while (tt) {
3999 tt->a.mac = NULL;
4000 tail = &tt->next;
4001 tt = tt->next;
4003 delete_Token(t);
4004 changed = true;
4005 } else {
4006 t = *tail = tline;
4007 tline = tline->next;
4008 t->a.mac = NULL;
4009 tail = &t->next;
4012 *tail = NULL;
4014 if (changed)
4015 paste_tokens(&thead, false);
4017 return thead;
4021 * Expand all single-line macro calls made in the given line.
4022 * Return the expanded version of the line. The original is deemed
4023 * to be destroyed in the process. (In reality we'll just move
4024 * Tokens from input to output a lot of the time, rather than
4025 * actually bothering to destroy and replicate.)
4028 static Token *expand_smacro(Token * tline)
4030 Token *t, *tt, *mstart, **tail, *thead;
4031 SMacro *head = NULL, *m;
4032 Token **params;
4033 int *paramsize;
4034 unsigned int nparam, sparam;
4035 int brackets;
4036 Token *org_tline = tline;
4037 Context *ctx;
4038 const char *mname;
4039 int deadman = DEADMAN_LIMIT;
4040 bool expanded;
4043 * Trick: we should avoid changing the start token pointer since it can
4044 * be contained in "next" field of other token. Because of this
4045 * we allocate a copy of first token and work with it; at the end of
4046 * routine we copy it back
4048 if (org_tline) {
4049 tline = new_Token(org_tline->next, org_tline->type,
4050 org_tline->text, 0);
4051 tline->a.mac = org_tline->a.mac;
4052 nasm_free(org_tline->text);
4053 org_tline->text = NULL;
4056 expanded = true; /* Always expand %+ at least once */
4058 again:
4059 thead = NULL;
4060 tail = &thead;
4062 while (tline) { /* main token loop */
4063 if (!--deadman) {
4064 error(ERR_NONFATAL, "interminable macro recursion");
4065 goto err;
4068 if ((mname = tline->text)) {
4069 /* if this token is a local macro, look in local context */
4070 if (tline->type == TOK_ID) {
4071 head = (SMacro *)hash_findix(&smacros, mname);
4072 } else if (tline->type == TOK_PREPROC_ID) {
4073 ctx = get_ctx(mname, &mname, true);
4074 head = ctx ? (SMacro *)hash_findix(&ctx->localmac, mname) : NULL;
4075 } else
4076 head = NULL;
4079 * We've hit an identifier. As in is_mmacro below, we first
4080 * check whether the identifier is a single-line macro at
4081 * all, then think about checking for parameters if
4082 * necessary.
4084 list_for_each(m, head)
4085 if (!mstrcmp(m->name, mname, m->casesense))
4086 break;
4087 if (m) {
4088 mstart = tline;
4089 params = NULL;
4090 paramsize = NULL;
4091 if (m->nparam == 0) {
4093 * Simple case: the macro is parameterless. Discard the
4094 * one token that the macro call took, and push the
4095 * expansion back on the to-do stack.
4097 if (!m->expansion) {
4098 if (!strcmp("__FILE__", m->name)) {
4099 int32_t num = 0;
4100 char *file = NULL;
4101 src_get(&num, &file);
4102 tline->text = nasm_quote(file, strlen(file));
4103 tline->type = TOK_STRING;
4104 nasm_free(file);
4105 continue;
4107 if (!strcmp("__LINE__", m->name)) {
4108 nasm_free(tline->text);
4109 make_tok_num(tline, src_get_linnum());
4110 continue;
4112 if (!strcmp("__BITS__", m->name)) {
4113 nasm_free(tline->text);
4114 make_tok_num(tline, globalbits);
4115 continue;
4117 tline = delete_Token(tline);
4118 continue;
4120 } else {
4122 * Complicated case: at least one macro with this name
4123 * exists and takes parameters. We must find the
4124 * parameters in the call, count them, find the SMacro
4125 * that corresponds to that form of the macro call, and
4126 * substitute for the parameters when we expand. What a
4127 * pain.
4129 /*tline = tline->next;
4130 skip_white_(tline); */
4131 do {
4132 t = tline->next;
4133 while (tok_type_(t, TOK_SMAC_END)) {
4134 t->a.mac->in_progress = false;
4135 t->text = NULL;
4136 t = tline->next = delete_Token(t);
4138 tline = t;
4139 } while (tok_type_(tline, TOK_WHITESPACE));
4140 if (!tok_is_(tline, "(")) {
4142 * This macro wasn't called with parameters: ignore
4143 * the call. (Behaviour borrowed from gnu cpp.)
4145 tline = mstart;
4146 m = NULL;
4147 } else {
4148 int paren = 0;
4149 int white = 0;
4150 brackets = 0;
4151 nparam = 0;
4152 sparam = PARAM_DELTA;
4153 params = nasm_malloc(sparam * sizeof(Token *));
4154 params[0] = tline->next;
4155 paramsize = nasm_malloc(sparam * sizeof(int));
4156 paramsize[0] = 0;
4157 while (true) { /* parameter loop */
4159 * For some unusual expansions
4160 * which concatenates function call
4162 t = tline->next;
4163 while (tok_type_(t, TOK_SMAC_END)) {
4164 t->a.mac->in_progress = false;
4165 t->text = NULL;
4166 t = tline->next = delete_Token(t);
4168 tline = t;
4170 if (!tline) {
4171 error(ERR_NONFATAL,
4172 "macro call expects terminating `)'");
4173 break;
4175 if (tline->type == TOK_WHITESPACE
4176 && brackets <= 0) {
4177 if (paramsize[nparam])
4178 white++;
4179 else
4180 params[nparam] = tline->next;
4181 continue; /* parameter loop */
4183 if (tline->type == TOK_OTHER
4184 && tline->text[1] == 0) {
4185 char ch = tline->text[0];
4186 if (ch == ',' && !paren && brackets <= 0) {
4187 if (++nparam >= sparam) {
4188 sparam += PARAM_DELTA;
4189 params = nasm_realloc(params,
4190 sparam * sizeof(Token *));
4191 paramsize = nasm_realloc(paramsize,
4192 sparam * sizeof(int));
4194 params[nparam] = tline->next;
4195 paramsize[nparam] = 0;
4196 white = 0;
4197 continue; /* parameter loop */
4199 if (ch == '{' &&
4200 (brackets > 0 || (brackets == 0 &&
4201 !paramsize[nparam])))
4203 if (!(brackets++)) {
4204 params[nparam] = tline->next;
4205 continue; /* parameter loop */
4208 if (ch == '}' && brackets > 0)
4209 if (--brackets == 0) {
4210 brackets = -1;
4211 continue; /* parameter loop */
4213 if (ch == '(' && !brackets)
4214 paren++;
4215 if (ch == ')' && brackets <= 0)
4216 if (--paren < 0)
4217 break;
4219 if (brackets < 0) {
4220 brackets = 0;
4221 error(ERR_NONFATAL, "braces do not "
4222 "enclose all of macro parameter");
4224 paramsize[nparam] += white + 1;
4225 white = 0;
4226 } /* parameter loop */
4227 nparam++;
4228 while (m && (m->nparam != nparam ||
4229 mstrcmp(m->name, mname,
4230 m->casesense)))
4231 m = m->next;
4232 if (!m)
4233 error(ERR_WARNING|ERR_PASS1|ERR_WARN_MNP,
4234 "macro `%s' exists, "
4235 "but not taking %d parameters",
4236 mstart->text, nparam);
4239 if (m && m->in_progress)
4240 m = NULL;
4241 if (!m) { /* in progess or didn't find '(' or wrong nparam */
4243 * Design question: should we handle !tline, which
4244 * indicates missing ')' here, or expand those
4245 * macros anyway, which requires the (t) test a few
4246 * lines down?
4248 nasm_free(params);
4249 nasm_free(paramsize);
4250 tline = mstart;
4251 } else {
4253 * Expand the macro: we are placed on the last token of the
4254 * call, so that we can easily split the call from the
4255 * following tokens. We also start by pushing an SMAC_END
4256 * token for the cycle removal.
4258 t = tline;
4259 if (t) {
4260 tline = t->next;
4261 t->next = NULL;
4263 tt = new_Token(tline, TOK_SMAC_END, NULL, 0);
4264 tt->a.mac = m;
4265 m->in_progress = true;
4266 tline = tt;
4267 list_for_each(t, m->expansion) {
4268 if (t->type >= TOK_SMAC_PARAM) {
4269 Token *pcopy = tline, **ptail = &pcopy;
4270 Token *ttt, *pt;
4271 int i;
4273 ttt = params[t->type - TOK_SMAC_PARAM];
4274 i = paramsize[t->type - TOK_SMAC_PARAM];
4275 while (--i >= 0) {
4276 pt = *ptail = new_Token(tline, ttt->type,
4277 ttt->text, 0);
4278 ptail = &pt->next;
4279 ttt = ttt->next;
4281 tline = pcopy;
4282 } else if (t->type == TOK_PREPROC_Q) {
4283 tt = new_Token(tline, TOK_ID, mname, 0);
4284 tline = tt;
4285 } else if (t->type == TOK_PREPROC_QQ) {
4286 tt = new_Token(tline, TOK_ID, m->name, 0);
4287 tline = tt;
4288 } else {
4289 tt = new_Token(tline, t->type, t->text, 0);
4290 tline = tt;
4295 * Having done that, get rid of the macro call, and clean
4296 * up the parameters.
4298 nasm_free(params);
4299 nasm_free(paramsize);
4300 free_tlist(mstart);
4301 expanded = true;
4302 continue; /* main token loop */
4307 if (tline->type == TOK_SMAC_END) {
4308 tline->a.mac->in_progress = false;
4309 tline = delete_Token(tline);
4310 } else {
4311 t = *tail = tline;
4312 tline = tline->next;
4313 t->a.mac = NULL;
4314 t->next = NULL;
4315 tail = &t->next;
4320 * Now scan the entire line and look for successive TOK_IDs that resulted
4321 * after expansion (they can't be produced by tokenize()). The successive
4322 * TOK_IDs should be concatenated.
4323 * Also we look for %+ tokens and concatenate the tokens before and after
4324 * them (without white spaces in between).
4326 if (expanded && paste_tokens(&thead, true)) {
4328 * If we concatenated something, *and* we had previously expanded
4329 * an actual macro, scan the lines again for macros...
4331 tline = thead;
4332 expanded = false;
4333 goto again;
4336 err:
4337 if (org_tline) {
4338 if (thead) {
4339 *org_tline = *thead;
4340 /* since we just gave text to org_line, don't free it */
4341 thead->text = NULL;
4342 delete_Token(thead);
4343 } else {
4344 /* the expression expanded to empty line;
4345 we can't return NULL for some reasons
4346 we just set the line to a single WHITESPACE token. */
4347 memset(org_tline, 0, sizeof(*org_tline));
4348 org_tline->text = NULL;
4349 org_tline->type = TOK_WHITESPACE;
4351 thead = org_tline;
4354 return thead;
4358 * Similar to expand_smacro but used exclusively with macro identifiers
4359 * right before they are fetched in. The reason is that there can be
4360 * identifiers consisting of several subparts. We consider that if there
4361 * are more than one element forming the name, user wants a expansion,
4362 * otherwise it will be left as-is. Example:
4364 * %define %$abc cde
4366 * the identifier %$abc will be left as-is so that the handler for %define
4367 * will suck it and define the corresponding value. Other case:
4369 * %define _%$abc cde
4371 * In this case user wants name to be expanded *before* %define starts
4372 * working, so we'll expand %$abc into something (if it has a value;
4373 * otherwise it will be left as-is) then concatenate all successive
4374 * PP_IDs into one.
4376 static Token *expand_id(Token * tline)
4378 Token *cur, *oldnext = NULL;
4380 if (!tline || !tline->next)
4381 return tline;
4383 cur = tline;
4384 while (cur->next &&
4385 (cur->next->type == TOK_ID ||
4386 cur->next->type == TOK_PREPROC_ID
4387 || cur->next->type == TOK_NUMBER))
4388 cur = cur->next;
4390 /* If identifier consists of just one token, don't expand */
4391 if (cur == tline)
4392 return tline;
4394 if (cur) {
4395 oldnext = cur->next; /* Detach the tail past identifier */
4396 cur->next = NULL; /* so that expand_smacro stops here */
4399 tline = expand_smacro(tline);
4401 if (cur) {
4402 /* expand_smacro possibly changhed tline; re-scan for EOL */
4403 cur = tline;
4404 while (cur && cur->next)
4405 cur = cur->next;
4406 if (cur)
4407 cur->next = oldnext;
4410 return tline;
4414 * Determine whether the given line constitutes a multi-line macro
4415 * call, and return the MMacro structure called if so. Doesn't have
4416 * to check for an initial label - that's taken care of in
4417 * expand_mmacro - but must check numbers of parameters. Guaranteed
4418 * to be called with tline->type == TOK_ID, so the putative macro
4419 * name is easy to find.
4421 static MMacro *is_mmacro(Token * tline, Token *** params_array)
4423 MMacro *head, *m;
4424 Token **params;
4425 int nparam;
4427 head = (MMacro *) hash_findix(&mmacros, tline->text);
4430 * Efficiency: first we see if any macro exists with the given
4431 * name. If not, we can return NULL immediately. _Then_ we
4432 * count the parameters, and then we look further along the
4433 * list if necessary to find the proper MMacro.
4435 list_for_each(m, head)
4436 if (!mstrcmp(m->name, tline->text, m->casesense))
4437 break;
4438 if (!m)
4439 return NULL;
4442 * OK, we have a potential macro. Count and demarcate the
4443 * parameters.
4445 count_mmac_params(tline->next, &nparam, &params);
4448 * So we know how many parameters we've got. Find the MMacro
4449 * structure that handles this number.
4451 while (m) {
4452 if (m->nparam_min <= nparam
4453 && (m->plus || nparam <= m->nparam_max)) {
4455 * This one is right. Just check if cycle removal
4456 * prohibits us using it before we actually celebrate...
4458 if (m->in_progress > m->max_depth) {
4459 if (m->max_depth > 0) {
4460 error(ERR_WARNING,
4461 "reached maximum recursion depth of %i",
4462 m->max_depth);
4464 nasm_free(params);
4465 return NULL;
4468 * It's right, and we can use it. Add its default
4469 * parameters to the end of our list if necessary.
4471 if (m->defaults && nparam < m->nparam_min + m->ndefs) {
4472 params =
4473 nasm_realloc(params,
4474 ((m->nparam_min + m->ndefs +
4475 1) * sizeof(*params)));
4476 while (nparam < m->nparam_min + m->ndefs) {
4477 params[nparam] = m->defaults[nparam - m->nparam_min];
4478 nparam++;
4482 * If we've gone over the maximum parameter count (and
4483 * we're in Plus mode), ignore parameters beyond
4484 * nparam_max.
4486 if (m->plus && nparam > m->nparam_max)
4487 nparam = m->nparam_max;
4489 * Then terminate the parameter list, and leave.
4491 if (!params) { /* need this special case */
4492 params = nasm_malloc(sizeof(*params));
4493 nparam = 0;
4495 params[nparam] = NULL;
4496 *params_array = params;
4497 return m;
4500 * This one wasn't right: look for the next one with the
4501 * same name.
4503 list_for_each(m, m->next)
4504 if (!mstrcmp(m->name, tline->text, m->casesense))
4505 break;
4509 * After all that, we didn't find one with the right number of
4510 * parameters. Issue a warning, and fail to expand the macro.
4512 error(ERR_WARNING|ERR_PASS1|ERR_WARN_MNP,
4513 "macro `%s' exists, but not taking %d parameters",
4514 tline->text, nparam);
4515 nasm_free(params);
4516 return NULL;
4521 * Save MMacro invocation specific fields in
4522 * preparation for a recursive macro expansion
4524 static void push_mmacro(MMacro *m)
4526 MMacroInvocation *i;
4528 i = nasm_malloc(sizeof(MMacroInvocation));
4529 i->prev = m->prev;
4530 i->params = m->params;
4531 i->iline = m->iline;
4532 i->nparam = m->nparam;
4533 i->rotate = m->rotate;
4534 i->paramlen = m->paramlen;
4535 i->unique = m->unique;
4536 i->condcnt = m->condcnt;
4537 m->prev = i;
4542 * Restore MMacro invocation specific fields that were
4543 * saved during a previous recursive macro expansion
4545 static void pop_mmacro(MMacro *m)
4547 MMacroInvocation *i;
4549 if (m->prev) {
4550 i = m->prev;
4551 m->prev = i->prev;
4552 m->params = i->params;
4553 m->iline = i->iline;
4554 m->nparam = i->nparam;
4555 m->rotate = i->rotate;
4556 m->paramlen = i->paramlen;
4557 m->unique = i->unique;
4558 m->condcnt = i->condcnt;
4559 nasm_free(i);
4565 * Expand the multi-line macro call made by the given line, if
4566 * there is one to be expanded. If there is, push the expansion on
4567 * istk->expansion and return 1. Otherwise return 0.
4569 static int expand_mmacro(Token * tline)
4571 Token *startline = tline;
4572 Token *label = NULL;
4573 int dont_prepend = 0;
4574 Token **params, *t, *mtok, *tt;
4575 MMacro *m;
4576 Line *l, *ll;
4577 int i, nparam, *paramlen;
4578 const char *mname;
4580 t = tline;
4581 skip_white_(t);
4582 /* if (!tok_type_(t, TOK_ID)) Lino 02/25/02 */
4583 if (!tok_type_(t, TOK_ID) && !tok_type_(t, TOK_PREPROC_ID))
4584 return 0;
4585 mtok = t;
4586 m = is_mmacro(t, &params);
4587 if (m) {
4588 mname = t->text;
4589 } else {
4590 Token *last;
4592 * We have an id which isn't a macro call. We'll assume
4593 * it might be a label; we'll also check to see if a
4594 * colon follows it. Then, if there's another id after
4595 * that lot, we'll check it again for macro-hood.
4597 label = last = t;
4598 t = t->next;
4599 if (tok_type_(t, TOK_WHITESPACE))
4600 last = t, t = t->next;
4601 if (tok_is_(t, ":")) {
4602 dont_prepend = 1;
4603 last = t, t = t->next;
4604 if (tok_type_(t, TOK_WHITESPACE))
4605 last = t, t = t->next;
4607 if (!tok_type_(t, TOK_ID) || !(m = is_mmacro(t, &params)))
4608 return 0;
4609 last->next = NULL;
4610 mname = t->text;
4611 tline = t;
4615 * Fix up the parameters: this involves stripping leading and
4616 * trailing whitespace, then stripping braces if they are
4617 * present.
4619 for (nparam = 0; params[nparam]; nparam++) ;
4620 paramlen = nparam ? nasm_malloc(nparam * sizeof(*paramlen)) : NULL;
4622 for (i = 0; params[i]; i++) {
4623 int brace = false;
4624 int comma = (!m->plus || i < nparam - 1);
4626 t = params[i];
4627 skip_white_(t);
4628 if (tok_is_(t, "{"))
4629 t = t->next, brace = true, comma = false;
4630 params[i] = t;
4631 paramlen[i] = 0;
4632 while (t) {
4633 if (comma && t->type == TOK_OTHER && !strcmp(t->text, ","))
4634 break; /* ... because we have hit a comma */
4635 if (comma && t->type == TOK_WHITESPACE
4636 && tok_is_(t->next, ","))
4637 break; /* ... or a space then a comma */
4638 if (brace && t->type == TOK_OTHER && !strcmp(t->text, "}"))
4639 break; /* ... or a brace */
4640 t = t->next;
4641 paramlen[i]++;
4646 * OK, we have a MMacro structure together with a set of
4647 * parameters. We must now go through the expansion and push
4648 * copies of each Line on to istk->expansion. Substitution of
4649 * parameter tokens and macro-local tokens doesn't get done
4650 * until the single-line macro substitution process; this is
4651 * because delaying them allows us to change the semantics
4652 * later through %rotate.
4654 * First, push an end marker on to istk->expansion, mark this
4655 * macro as in progress, and set up its invocation-specific
4656 * variables.
4658 ll = nasm_malloc(sizeof(Line));
4659 ll->next = istk->expansion;
4660 ll->finishes = m;
4661 ll->first = NULL;
4662 istk->expansion = ll;
4665 * Save the previous MMacro expansion in the case of
4666 * macro recursion
4668 if (m->max_depth && m->in_progress)
4669 push_mmacro(m);
4671 m->in_progress ++;
4672 m->params = params;
4673 m->iline = tline;
4674 m->nparam = nparam;
4675 m->rotate = 0;
4676 m->paramlen = paramlen;
4677 m->unique = unique++;
4678 m->lineno = 0;
4679 m->condcnt = 0;
4681 m->next_active = istk->mstk;
4682 istk->mstk = m;
4684 list_for_each(l, m->expansion) {
4685 Token **tail;
4687 ll = nasm_malloc(sizeof(Line));
4688 ll->finishes = NULL;
4689 ll->next = istk->expansion;
4690 istk->expansion = ll;
4691 tail = &ll->first;
4693 list_for_each(t, l->first) {
4694 Token *x = t;
4695 switch (t->type) {
4696 case TOK_PREPROC_Q:
4697 tt = *tail = new_Token(NULL, TOK_ID, mname, 0);
4698 break;
4699 case TOK_PREPROC_QQ:
4700 tt = *tail = new_Token(NULL, TOK_ID, m->name, 0);
4701 break;
4702 case TOK_PREPROC_ID:
4703 if (t->text[1] == '0' && t->text[2] == '0') {
4704 dont_prepend = -1;
4705 x = label;
4706 if (!x)
4707 continue;
4709 /* fall through */
4710 default:
4711 tt = *tail = new_Token(NULL, x->type, x->text, 0);
4712 break;
4714 tail = &tt->next;
4716 *tail = NULL;
4720 * If we had a label, push it on as the first line of
4721 * the macro expansion.
4723 if (label) {
4724 if (dont_prepend < 0)
4725 free_tlist(startline);
4726 else {
4727 ll = nasm_malloc(sizeof(Line));
4728 ll->finishes = NULL;
4729 ll->next = istk->expansion;
4730 istk->expansion = ll;
4731 ll->first = startline;
4732 if (!dont_prepend) {
4733 while (label->next)
4734 label = label->next;
4735 label->next = tt = new_Token(NULL, TOK_OTHER, ":", 0);
4740 list->uplevel(m->nolist ? LIST_MACRO_NOLIST : LIST_MACRO);
4742 return 1;
4745 /* The function that actually does the error reporting */
4746 static void verror(int severity, const char *fmt, va_list arg)
4748 char buff[1024];
4749 MMacro *mmac = NULL;
4750 int delta = 0;
4752 vsnprintf(buff, sizeof(buff), fmt, arg);
4754 /* get %macro name */
4755 if (istk && istk->mstk) {
4756 mmac = istk->mstk;
4757 /* but %rep blocks should be skipped */
4758 while (mmac && !mmac->name)
4759 mmac = mmac->next_active, delta++;
4762 if (mmac)
4763 nasm_error(severity, "(%s:%d) %s",
4764 mmac->name, mmac->lineno - delta, buff);
4765 else
4766 nasm_error(severity, "%s", buff);
4770 * Since preprocessor always operate only on the line that didn't
4771 * arrived yet, we should always use ERR_OFFBY1.
4773 static void error(int severity, const char *fmt, ...)
4775 va_list arg;
4777 /* If we're in a dead branch of IF or something like it, ignore the error */
4778 if (istk && istk->conds && !emitting(istk->conds->state))
4779 return;
4781 va_start(arg, fmt);
4782 verror(severity, fmt, arg);
4783 va_end(arg);
4787 * Because %else etc are evaluated in the state context
4788 * of the previous branch, errors might get lost with error():
4789 * %if 0 ... %else trailing garbage ... %endif
4790 * So %else etc should report errors with this function.
4792 static void error_precond(int severity, const char *fmt, ...)
4794 va_list arg;
4796 /* Only ignore the error if it's really in a dead branch */
4797 if (istk && istk->conds && istk->conds->state == COND_NEVER)
4798 return;
4800 va_start(arg, fmt);
4801 verror(severity, fmt, arg);
4802 va_end(arg);
4805 static void
4806 pp_reset(char *file, int apass, ListGen * listgen, StrList **deplist)
4808 Token *t;
4810 cstk = NULL;
4811 istk = nasm_malloc(sizeof(Include));
4812 istk->next = NULL;
4813 istk->conds = NULL;
4814 istk->expansion = NULL;
4815 istk->mstk = NULL;
4816 istk->fp = fopen(file, "r");
4817 istk->fname = NULL;
4818 src_set_fname(nasm_strdup(file));
4819 src_set_linnum(0);
4820 istk->lineinc = 1;
4821 if (!istk->fp)
4822 error(ERR_FATAL|ERR_NOFILE, "unable to open input file `%s'",
4823 file);
4824 defining = NULL;
4825 nested_mac_count = 0;
4826 nested_rep_count = 0;
4827 init_macros();
4828 unique = 0;
4829 if (tasm_compatible_mode) {
4830 stdmacpos = nasm_stdmac;
4831 } else {
4832 stdmacpos = nasm_stdmac_after_tasm;
4834 any_extrastdmac = extrastdmac && *extrastdmac;
4835 do_predef = true;
4836 list = listgen;
4839 * 0 for dependencies, 1 for preparatory passes, 2 for final pass.
4840 * The caller, however, will also pass in 3 for preprocess-only so
4841 * we can set __PASS__ accordingly.
4843 pass = apass > 2 ? 2 : apass;
4845 dephead = deptail = deplist;
4846 if (deplist) {
4847 StrList *sl = nasm_malloc(strlen(file)+1+sizeof sl->next);
4848 sl->next = NULL;
4849 strcpy(sl->str, file);
4850 *deptail = sl;
4851 deptail = &sl->next;
4855 * Define the __PASS__ macro. This is defined here unlike
4856 * all the other builtins, because it is special -- it varies between
4857 * passes.
4859 t = nasm_malloc(sizeof(*t));
4860 t->next = NULL;
4861 make_tok_num(t, apass);
4862 t->a.mac = NULL;
4863 define_smacro(NULL, "__PASS__", true, 0, t);
4866 static char *pp_getline(void)
4868 char *line;
4869 Token *tline;
4871 while (1) {
4873 * Fetch a tokenized line, either from the macro-expansion
4874 * buffer or from the input file.
4876 tline = NULL;
4877 while (istk->expansion && istk->expansion->finishes) {
4878 Line *l = istk->expansion;
4879 if (!l->finishes->name && l->finishes->in_progress > 1) {
4880 Line *ll;
4883 * This is a macro-end marker for a macro with no
4884 * name, which means it's not really a macro at all
4885 * but a %rep block, and the `in_progress' field is
4886 * more than 1, meaning that we still need to
4887 * repeat. (1 means the natural last repetition; 0
4888 * means termination by %exitrep.) We have
4889 * therefore expanded up to the %endrep, and must
4890 * push the whole block on to the expansion buffer
4891 * again. We don't bother to remove the macro-end
4892 * marker: we'd only have to generate another one
4893 * if we did.
4895 l->finishes->in_progress--;
4896 list_for_each(l, l->finishes->expansion) {
4897 Token *t, *tt, **tail;
4899 ll = nasm_malloc(sizeof(Line));
4900 ll->next = istk->expansion;
4901 ll->finishes = NULL;
4902 ll->first = NULL;
4903 tail = &ll->first;
4905 list_for_each(t, l->first) {
4906 if (t->text || t->type == TOK_WHITESPACE) {
4907 tt = *tail = new_Token(NULL, t->type, t->text, 0);
4908 tail = &tt->next;
4912 istk->expansion = ll;
4914 } else {
4916 * Check whether a `%rep' was started and not ended
4917 * within this macro expansion. This can happen and
4918 * should be detected. It's a fatal error because
4919 * I'm too confused to work out how to recover
4920 * sensibly from it.
4922 if (defining) {
4923 if (defining->name)
4924 error(ERR_PANIC,
4925 "defining with name in expansion");
4926 else if (istk->mstk->name)
4927 error(ERR_FATAL,
4928 "`%%rep' without `%%endrep' within"
4929 " expansion of macro `%s'",
4930 istk->mstk->name);
4934 * FIXME: investigate the relationship at this point between
4935 * istk->mstk and l->finishes
4938 MMacro *m = istk->mstk;
4939 istk->mstk = m->next_active;
4940 if (m->name) {
4942 * This was a real macro call, not a %rep, and
4943 * therefore the parameter information needs to
4944 * be freed.
4946 if (m->prev) {
4947 pop_mmacro(m);
4948 l->finishes->in_progress --;
4949 } else {
4950 nasm_free(m->params);
4951 free_tlist(m->iline);
4952 nasm_free(m->paramlen);
4953 l->finishes->in_progress = 0;
4955 } else
4956 free_mmacro(m);
4958 istk->expansion = l->next;
4959 nasm_free(l);
4960 list->downlevel(LIST_MACRO);
4963 while (1) { /* until we get a line we can use */
4965 if (istk->expansion) { /* from a macro expansion */
4966 char *p;
4967 Line *l = istk->expansion;
4968 if (istk->mstk)
4969 istk->mstk->lineno++;
4970 tline = l->first;
4971 istk->expansion = l->next;
4972 nasm_free(l);
4973 p = detoken(tline, false);
4974 list->line(LIST_MACRO, p);
4975 nasm_free(p);
4976 break;
4978 line = read_line();
4979 if (line) { /* from the current input file */
4980 line = prepreproc(line);
4981 tline = tokenize(line);
4982 nasm_free(line);
4983 break;
4986 * The current file has ended; work down the istk
4989 Include *i = istk;
4990 fclose(i->fp);
4991 if (i->conds) {
4992 /* nasm_error can't be conditionally suppressed */
4993 nasm_error(ERR_FATAL,
4994 "expected `%%endif' before end of file");
4996 /* only set line and file name if there's a next node */
4997 if (i->next) {
4998 src_set_linnum(i->lineno);
4999 nasm_free(src_set_fname(i->fname));
5001 istk = i->next;
5002 list->downlevel(LIST_INCLUDE);
5003 nasm_free(i);
5004 if (!istk)
5005 return NULL;
5006 if (istk->expansion && istk->expansion->finishes)
5007 break;
5012 * We must expand MMacro parameters and MMacro-local labels
5013 * _before_ we plunge into directive processing, to cope
5014 * with things like `%define something %1' such as STRUC
5015 * uses. Unless we're _defining_ a MMacro, in which case
5016 * those tokens should be left alone to go into the
5017 * definition; and unless we're in a non-emitting
5018 * condition, in which case we don't want to meddle with
5019 * anything.
5021 if (!defining && !(istk->conds && !emitting(istk->conds->state))
5022 && !(istk->mstk && !istk->mstk->in_progress)) {
5023 tline = expand_mmac_params(tline);
5027 * Check the line to see if it's a preprocessor directive.
5029 if (do_directive(tline) == DIRECTIVE_FOUND) {
5030 continue;
5031 } else if (defining) {
5033 * We're defining a multi-line macro. We emit nothing
5034 * at all, and just
5035 * shove the tokenized line on to the macro definition.
5037 Line *l = nasm_malloc(sizeof(Line));
5038 l->next = defining->expansion;
5039 l->first = tline;
5040 l->finishes = NULL;
5041 defining->expansion = l;
5042 continue;
5043 } else if (istk->conds && !emitting(istk->conds->state)) {
5045 * We're in a non-emitting branch of a condition block.
5046 * Emit nothing at all, not even a blank line: when we
5047 * emerge from the condition we'll give a line-number
5048 * directive so we keep our place correctly.
5050 free_tlist(tline);
5051 continue;
5052 } else if (istk->mstk && !istk->mstk->in_progress) {
5054 * We're in a %rep block which has been terminated, so
5055 * we're walking through to the %endrep without
5056 * emitting anything. Emit nothing at all, not even a
5057 * blank line: when we emerge from the %rep block we'll
5058 * give a line-number directive so we keep our place
5059 * correctly.
5061 free_tlist(tline);
5062 continue;
5063 } else {
5064 tline = expand_smacro(tline);
5065 if (!expand_mmacro(tline)) {
5067 * De-tokenize the line again, and emit it.
5069 line = detoken(tline, true);
5070 free_tlist(tline);
5071 break;
5072 } else {
5073 continue; /* expand_mmacro calls free_tlist */
5078 return line;
5081 static void pp_cleanup(int pass)
5083 if (defining) {
5084 if (defining->name) {
5085 error(ERR_NONFATAL,
5086 "end of file while still defining macro `%s'",
5087 defining->name);
5088 } else {
5089 error(ERR_NONFATAL, "end of file while still in %%rep");
5092 free_mmacro(defining);
5093 defining = NULL;
5095 while (cstk)
5096 ctx_pop();
5097 free_macros();
5098 while (istk) {
5099 Include *i = istk;
5100 istk = istk->next;
5101 fclose(i->fp);
5102 nasm_free(i->fname);
5103 nasm_free(i);
5105 while (cstk)
5106 ctx_pop();
5107 nasm_free(src_set_fname(NULL));
5108 if (pass == 0) {
5109 IncPath *i;
5110 free_llist(predef);
5111 delete_Blocks();
5112 while ((i = ipath)) {
5113 ipath = i->next;
5114 if (i->path)
5115 nasm_free(i->path);
5116 nasm_free(i);
5121 void pp_include_path(char *path)
5123 IncPath *i;
5125 i = nasm_malloc(sizeof(IncPath));
5126 i->path = path ? nasm_strdup(path) : NULL;
5127 i->next = NULL;
5129 if (ipath) {
5130 IncPath *j = ipath;
5131 while (j->next)
5132 j = j->next;
5133 j->next = i;
5134 } else {
5135 ipath = i;
5139 void pp_pre_include(char *fname)
5141 Token *inc, *space, *name;
5142 Line *l;
5144 name = new_Token(NULL, TOK_INTERNAL_STRING, fname, 0);
5145 space = new_Token(name, TOK_WHITESPACE, NULL, 0);
5146 inc = new_Token(space, TOK_PREPROC_ID, "%include", 0);
5148 l = nasm_malloc(sizeof(Line));
5149 l->next = predef;
5150 l->first = inc;
5151 l->finishes = NULL;
5152 predef = l;
5155 void pp_pre_define(char *definition)
5157 Token *def, *space;
5158 Line *l;
5159 char *equals;
5161 equals = strchr(definition, '=');
5162 space = new_Token(NULL, TOK_WHITESPACE, NULL, 0);
5163 def = new_Token(space, TOK_PREPROC_ID, "%define", 0);
5164 if (equals)
5165 *equals = ' ';
5166 space->next = tokenize(definition);
5167 if (equals)
5168 *equals = '=';
5170 l = nasm_malloc(sizeof(Line));
5171 l->next = predef;
5172 l->first = def;
5173 l->finishes = NULL;
5174 predef = l;
5177 void pp_pre_undefine(char *definition)
5179 Token *def, *space;
5180 Line *l;
5182 space = new_Token(NULL, TOK_WHITESPACE, NULL, 0);
5183 def = new_Token(space, TOK_PREPROC_ID, "%undef", 0);
5184 space->next = tokenize(definition);
5186 l = nasm_malloc(sizeof(Line));
5187 l->next = predef;
5188 l->first = def;
5189 l->finishes = NULL;
5190 predef = l;
5194 * Added by Keith Kanios:
5196 * This function is used to assist with "runtime" preprocessor
5197 * directives. (e.g. pp_runtime("%define __BITS__ 64");)
5199 * ERRORS ARE IGNORED HERE, SO MAKE COMPLETELY SURE THAT YOU
5200 * PASS A VALID STRING TO THIS FUNCTION!!!!!
5203 void pp_runtime(char *definition)
5205 Token *def;
5207 def = tokenize(definition);
5208 if (do_directive(def) == NO_DIRECTIVE_FOUND)
5209 free_tlist(def);
5213 void pp_extra_stdmac(macros_t *macros)
5215 extrastdmac = macros;
5218 static void make_tok_num(Token * tok, int64_t val)
5220 char numbuf[20];
5221 snprintf(numbuf, sizeof(numbuf), "%"PRId64"", val);
5222 tok->text = nasm_strdup(numbuf);
5223 tok->type = TOK_NUMBER;
5226 Preproc nasmpp = {
5227 pp_reset,
5228 pp_getline,
5229 pp_cleanup