blob: 2c6df95a45f9fbc54085f910f5f11c50320379bf [file] [log] [blame]
H. Peter Anvinea6e34d2002-04-30 20:51:32 +00001/* parser.c source line parser for the Netwide Assembler
2 *
3 * The Netwide Assembler is copyright (C) 1996 Simon Tatham and
4 * Julian Hall. All rights reserved. The software is
5 * redistributable under the licence given in the file "Licence"
6 * distributed in the NASM archive.
7 *
8 * initial version 27/iii/95 by Simon Tatham
9 */
10
11#include <stdio.h>
12#include <stdlib.h>
13#include <stddef.h>
14#include <string.h>
15#include <ctype.h>
16
17#include "nasm.h"
18#include "nasmlib.h"
19#include "parser.h"
20#include "float.h"
21
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000022static long reg_flags[] = { /* sizes and special flags */
23 0, REG8, REG_AL, REG_AX, REG8, REG8, REG16, REG16, REG8, REG_CL,
24 REG_CREG, REG_CREG, REG_CREG, REG_CR4, REG_CS, REG_CX, REG8,
25 REG16, REG8, REG_DREG, REG_DREG, REG_DREG, REG_DREG, REG_DREG,
26 REG_DREG, REG_DESS, REG_DX, REG_EAX, REG32, REG32, REG_ECX,
27 REG32, REG32, REG_DESS, REG32, REG32, REG_FSGS, REG_FSGS,
28 MMXREG, MMXREG, MMXREG, MMXREG, MMXREG, MMXREG, MMXREG, MMXREG,
29 REG16, REG16, REG_DESS, FPU0, FPUREG, FPUREG, FPUREG, FPUREG,
30 FPUREG, FPUREG, FPUREG, REG_TREG, REG_TREG, REG_TREG, REG_TREG,
H. Peter Anvin4836e332002-04-30 20:56:43 +000031 REG_TREG,
32 XMMREG, XMMREG, XMMREG, XMMREG, XMMREG, XMMREG, XMMREG, XMMREG
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000033};
34
35enum { /* special tokens */
H. Peter Anvin76690a12002-04-30 20:52:49 +000036 S_BYTE, S_DWORD, S_FAR, S_LONG, S_NEAR, S_NOSPLIT, S_QWORD,
H. Peter Anvin01377d82002-05-21 03:16:33 +000037 S_SHORT, S_STRICT, S_TO, S_TWORD, S_WORD
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000038};
39
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000040static int is_comma_next (void);
41
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000042static int i;
43static struct tokenval tokval;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000044static efunc error;
H. Peter Anvineba20a72002-04-30 20:53:55 +000045static struct ofmt *outfmt; /* Structure of addresses of output routines */
46static loc_t *location; /* Pointer to current line's segment,offset */
47
48void parser_global_info (struct ofmt *output, loc_t *locp)
49{
50 outfmt = output;
51 location = locp;
52}
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000053
H. Peter Anvin76690a12002-04-30 20:52:49 +000054insn *parse_line (int pass, char *buffer, insn *result,
H. Peter Anvineba20a72002-04-30 20:53:55 +000055 efunc errfunc, evalfunc evaluate, ldfunc ldef)
56{
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000057 int operand;
58 int critical;
H. Peter Anvin76690a12002-04-30 20:52:49 +000059 struct eval_hints hints;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000060
H. Peter Anvin76690a12002-04-30 20:52:49 +000061 result->forw_ref = FALSE;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000062 error = errfunc;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000063
H. Peter Anvin76690a12002-04-30 20:52:49 +000064 stdscan_reset();
65 stdscan_bufptr = buffer;
66 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000067
H. Peter Anvineba20a72002-04-30 20:53:55 +000068 result->label = NULL; /* Assume no label */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000069 result->eops = NULL; /* must do this, whatever happens */
H. Peter Anvin76690a12002-04-30 20:52:49 +000070 result->operands = 0; /* must initialise this */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000071
72 if (i==0) { /* blank line - ignore */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000073 result->opcode = -1; /* and no instruction either */
74 return result;
75 }
76 if (i != TOKEN_ID && i != TOKEN_INSN && i != TOKEN_PREFIX &&
77 (i!=TOKEN_REG || (REG_SREG & ~reg_flags[tokval.t_integer]))) {
78 error (ERR_NONFATAL, "label or instruction expected"
79 " at start of line");
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000080 result->opcode = -1;
81 return result;
82 }
83
84 if (i == TOKEN_ID) { /* there's a label here */
H. Peter Anvin76690a12002-04-30 20:52:49 +000085 result->label = tokval.t_charptr;
H. Peter Anvin76690a12002-04-30 20:52:49 +000086 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000087 if (i == ':') { /* skip over the optional colon */
H. Peter Anvin76690a12002-04-30 20:52:49 +000088 i = stdscan(NULL, &tokval);
H. Peter Anvineba20a72002-04-30 20:53:55 +000089 } else if (i == 0) {
90 error (ERR_WARNING|ERR_WARN_OL|ERR_PASS1,
H. Peter Anvin6768eb72002-04-30 20:52:26 +000091 "label alone on a line without a colon might be in error");
H. Peter Anvinea6e34d2002-04-30 20:51:32 +000092 }
H. Peter Anvineba20a72002-04-30 20:53:55 +000093 if (i != TOKEN_INSN || tokval.t_integer != I_EQU)
94 {
95 /*
96 * FIXME: location->segment could be NO_SEG, in which case
97 * it is possible we should be passing 'abs_seg'. Look into this.
98 * Work out whether that is *really* what we should be doing.
99 * Generally fix things. I think this is right as it is, but
100 * am still not certain.
101 */
102 ldef (result->label, location->segment,
103 location->offset, NULL, TRUE, FALSE, outfmt, errfunc);
104 }
105 }
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000106
107 if (i==0) {
108 result->opcode = -1; /* this line contains just a label */
109 return result;
110 }
111
112 result->nprefix = 0;
H. Peter Anvind7ed89e2002-04-30 20:52:08 +0000113 result->times = 1L;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000114
115 while (i == TOKEN_PREFIX ||
H. Peter Anvineba20a72002-04-30 20:53:55 +0000116 (i==TOKEN_REG && !(REG_SREG & ~reg_flags[tokval.t_integer])))
117 {
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000118 /*
119 * Handle special case: the TIMES prefix.
120 */
121 if (i == TOKEN_PREFIX && tokval.t_integer == P_TIMES) {
122 expr *value;
123
H. Peter Anvin76690a12002-04-30 20:52:49 +0000124 i = stdscan(NULL, &tokval);
H. Peter Anvin8ac36412002-04-30 21:09:12 +0000125 value = evaluate (stdscan, NULL, &tokval, NULL, pass0, error, NULL);
H. Peter Anvin76690a12002-04-30 20:52:49 +0000126 i = tokval.t_type;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000127 if (!value) { /* but, error in evaluator */
128 result->opcode = -1; /* unrecoverable parse error: */
129 return result; /* ignore this instruction */
130 }
131 if (!is_simple (value)) {
132 error (ERR_NONFATAL,
133 "non-constant argument supplied to TIMES");
H. Peter Anvind7ed89e2002-04-30 20:52:08 +0000134 result->times = 1L;
135 } else {
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000136 result->times = value->value;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000137 if (value->value < 0) {
H. Peter Anvin6768eb72002-04-30 20:52:26 +0000138 error(ERR_NONFATAL, "TIMES value %d is negative",
H. Peter Anvind7ed89e2002-04-30 20:52:08 +0000139 value->value);
H. Peter Anvineba20a72002-04-30 20:53:55 +0000140 result->times = 0;
141 }
H. Peter Anvind7ed89e2002-04-30 20:52:08 +0000142 }
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000143 } else {
144 if (result->nprefix == MAXPREFIX)
145 error (ERR_NONFATAL,
146 "instruction has more than %d prefixes", MAXPREFIX);
147 else
148 result->prefixes[result->nprefix++] = tokval.t_integer;
H. Peter Anvin76690a12002-04-30 20:52:49 +0000149 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000150 }
151 }
152
153 if (i != TOKEN_INSN) {
H. Peter Anvin76690a12002-04-30 20:52:49 +0000154 if (result->nprefix > 0 && i == 0) {
155 /*
156 * Instruction prefixes are present, but no actual
157 * instruction. This is allowed: at this point we
158 * invent a notional instruction of RESB 0.
159 */
160 result->opcode = I_RESB;
161 result->operands = 1;
162 result->oprs[0].type = IMMEDIATE;
163 result->oprs[0].offset = 0L;
164 result->oprs[0].segment = result->oprs[0].wrt = NO_SEG;
165 return result;
166 } else {
167 error (ERR_NONFATAL, "parser: instruction expected");
168 result->opcode = -1;
169 return result;
170 }
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000171 }
172
173 result->opcode = tokval.t_integer;
174 result->condition = tokval.t_inttwo;
175
176 /*
177 * RESB, RESW and RESD cannot be satisfied with incorrectly
178 * evaluated operands, since the correct values _must_ be known
179 * on the first pass. Hence, even in pass one, we set the
180 * `critical' flag on calling evaluate(), so that it will bomb
181 * out on undefined symbols. Nasty, but there's nothing we can
182 * do about it.
183 *
184 * For the moment, EQU has the same difficulty, so we'll
185 * include that.
186 */
187 if (result->opcode == I_RESB ||
188 result->opcode == I_RESW ||
189 result->opcode == I_RESD ||
190 result->opcode == I_RESQ ||
191 result->opcode == I_REST ||
H. Peter Anvin8ac36412002-04-30 21:09:12 +0000192 result->opcode == I_EQU ||
193 result->opcode == I_INCBIN) /* fbk */
H. Peter Anvineba20a72002-04-30 20:53:55 +0000194 {
H. Peter Anvin8ac36412002-04-30 21:09:12 +0000195 critical = pass0;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000196 }
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000197 else
198 critical = (pass==2 ? 2 : 0);
199
200 if (result->opcode == I_DB ||
201 result->opcode == I_DW ||
202 result->opcode == I_DD ||
203 result->opcode == I_DQ ||
H. Peter Anvind7ed89e2002-04-30 20:52:08 +0000204 result->opcode == I_DT ||
H. Peter Anvineba20a72002-04-30 20:53:55 +0000205 result->opcode == I_INCBIN)
206 {
H. Peter Anvin87bc6192002-04-30 20:53:16 +0000207 extop *eop, **tail = &result->eops, **fixptr;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000208 int oper_num = 0;
209
H. Peter Anvineba20a72002-04-30 20:53:55 +0000210 result->eops_float = FALSE;
211
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000212 /*
H. Peter Anvineba20a72002-04-30 20:53:55 +0000213 * Begin to read the DB/DW/DD/DQ/DT/INCBIN operands.
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000214 */
215 while (1) {
H. Peter Anvin76690a12002-04-30 20:52:49 +0000216 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000217 if (i == 0)
218 break;
H. Peter Anvin87bc6192002-04-30 20:53:16 +0000219 fixptr = tail;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000220 eop = *tail = nasm_malloc(sizeof(extop));
221 tail = &eop->next;
222 eop->next = NULL;
223 eop->type = EOT_NOTHING;
224 oper_num++;
225
226 if (i == TOKEN_NUM && tokval.t_charptr && is_comma_next()) {
227 eop->type = EOT_DB_STRING;
228 eop->stringval = tokval.t_charptr;
229 eop->stringlen = tokval.t_inttwo;
H. Peter Anvin76690a12002-04-30 20:52:49 +0000230 i = stdscan(NULL, &tokval); /* eat the comma */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000231 continue;
232 }
233
H. Peter Anvineba20a72002-04-30 20:53:55 +0000234 if ((i == TOKEN_FLOAT && is_comma_next()) || i == '-') {
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000235 long sign = +1L;
236
237 if (i == '-') {
H. Peter Anvin76690a12002-04-30 20:52:49 +0000238 char *save = stdscan_bufptr;
239 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000240 sign = -1L;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000241 if (i != TOKEN_FLOAT || !is_comma_next()) {
H. Peter Anvin76690a12002-04-30 20:52:49 +0000242 stdscan_bufptr = save;
243 i = tokval.t_type = '-';
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000244 }
245 }
246
247 if (i == TOKEN_FLOAT) {
248 eop->type = EOT_DB_STRING;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000249 result->eops_float = TRUE;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000250 if (result->opcode == I_DD)
251 eop->stringlen = 4;
252 else if (result->opcode == I_DQ)
253 eop->stringlen = 8;
254 else if (result->opcode == I_DT)
H. Peter Anvineba20a72002-04-30 20:53:55 +0000255 eop->stringlen = 10;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000256 else {
257 error(ERR_NONFATAL, "floating-point constant"
258 " encountered in `D%c' instruction",
259 result->opcode == I_DW ? 'W' : 'B');
H. Peter Anvineba20a72002-04-30 20:53:55 +0000260 /*
261 * fix suggested by Pedro Gimeno... original line
262 * was:
263 * eop->type = EOT_NOTHING;
264 */
265 eop->stringlen = 0;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000266 }
H. Peter Anvin76690a12002-04-30 20:52:49 +0000267 eop = nasm_realloc(eop, sizeof(extop)+eop->stringlen);
H. Peter Anvin87bc6192002-04-30 20:53:16 +0000268 tail = &eop->next;
269 *fixptr = eop;
H. Peter Anvin76690a12002-04-30 20:52:49 +0000270 eop->stringval = (char *)eop + sizeof(extop);
H. Peter Anvineba20a72002-04-30 20:53:55 +0000271 if (eop->stringlen < 4 ||
272 !float_const (tokval.t_charptr, sign,
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000273 (unsigned char *)eop->stringval,
274 eop->stringlen, error))
275 eop->type = EOT_NOTHING;
H. Peter Anvin76690a12002-04-30 20:52:49 +0000276 i = stdscan(NULL, &tokval); /* eat the comma */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000277 continue;
278 }
279 }
280
H. Peter Anvineba20a72002-04-30 20:53:55 +0000281 /* anything else */
282 {
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000283 expr *value;
H. Peter Anvin76690a12002-04-30 20:52:49 +0000284 value = evaluate (stdscan, NULL, &tokval, NULL,
285 critical, error, NULL);
286 i = tokval.t_type;
287 if (!value) { /* error in evaluator */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000288 result->opcode = -1;/* unrecoverable parse error: */
289 return result; /* ignore this instruction */
290 }
H. Peter Anvin76690a12002-04-30 20:52:49 +0000291 if (is_unknown(value)) {
292 eop->type = EOT_DB_NUMBER;
293 eop->offset = 0; /* doesn't matter what we put */
294 eop->segment = eop->wrt = NO_SEG; /* likewise */
295 } else if (is_reloc(value)) {
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000296 eop->type = EOT_DB_NUMBER;
297 eop->offset = reloc_value(value);
298 eop->segment = reloc_seg(value);
299 eop->wrt = reloc_wrt(value);
300 } else {
301 error (ERR_NONFATAL,
H. Peter Anvin76690a12002-04-30 20:52:49 +0000302 "operand %d: expression is not simple"
303 " or relocatable", oper_num);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000304 }
305 }
H. Peter Anvin6768eb72002-04-30 20:52:26 +0000306
307 /*
H. Peter Anvin76690a12002-04-30 20:52:49 +0000308 * We're about to call stdscan(), which will eat the
H. Peter Anvin6768eb72002-04-30 20:52:26 +0000309 * comma that we're currently sitting on between
310 * arguments. However, we'd better check first that it
311 * _is_ a comma.
312 */
313 if (i == 0) /* also could be EOL */
314 break;
315 if (i != ',') {
H. Peter Anvin76690a12002-04-30 20:52:49 +0000316 error (ERR_NONFATAL, "comma expected after operand %d",
317 oper_num);
H. Peter Anvin6768eb72002-04-30 20:52:26 +0000318 result->opcode = -1;/* unrecoverable parse error: */
319 return result; /* ignore this instruction */
320 }
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000321 }
H. Peter Anvind7ed89e2002-04-30 20:52:08 +0000322
323 if (result->opcode == I_INCBIN) {
324 /*
325 * Correct syntax for INCBIN is that there should be
326 * one string operand, followed by one or two numeric
327 * operands.
328 */
329 if (!result->eops || result->eops->type != EOT_DB_STRING)
330 error (ERR_NONFATAL, "`incbin' expects a file name");
331 else if (result->eops->next &&
332 result->eops->next->type != EOT_DB_NUMBER)
333 error (ERR_NONFATAL, "`incbin': second parameter is",
334 " non-numeric");
335 else if (result->eops->next && result->eops->next->next &&
336 result->eops->next->next->type != EOT_DB_NUMBER)
337 error (ERR_NONFATAL, "`incbin': third parameter is",
338 " non-numeric");
339 else if (result->eops->next && result->eops->next->next &&
340 result->eops->next->next->next)
341 error (ERR_NONFATAL, "`incbin': more than three parameters");
342 else
343 return result;
344 /*
345 * If we reach here, one of the above errors happened.
346 * Throw the instruction away.
347 */
348 result->opcode = -1;
349 return result;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000350 } else /* DB ... */
351 if (oper_num == 0)
352 error (ERR_WARNING|ERR_PASS1,
353 "no operand for data declaration");
354 else
355 result->operands = oper_num;
H. Peter Anvind7ed89e2002-04-30 20:52:08 +0000356
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000357 return result;
358 }
359
360 /* right. Now we begin to parse the operands. There may be up to three
361 * of these, separated by commas, and terminated by a zero token. */
362
363 for (operand = 0; operand < 3; operand++) {
H. Peter Anvin76690a12002-04-30 20:52:49 +0000364 expr *value; /* used most of the time */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000365 int mref; /* is this going to be a memory ref? */
H. Peter Anvin6768eb72002-04-30 20:52:26 +0000366 int bracket; /* is it a [] mref, or a & mref? */
H. Peter Anvineba20a72002-04-30 20:53:55 +0000367 int setsize = 0;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000368
369 result->oprs[operand].addr_size = 0;/* have to zero this whatever */
H. Peter Anvin76690a12002-04-30 20:52:49 +0000370 result->oprs[operand].eaflags = 0; /* and this */
H. Peter Anvineba20a72002-04-30 20:53:55 +0000371 result->oprs[operand].opflags = 0;
372
H. Peter Anvin76690a12002-04-30 20:52:49 +0000373 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000374 if (i == 0) break; /* end of operands: get out of here */
375 result->oprs[operand].type = 0; /* so far, no override */
376 while (i == TOKEN_SPECIAL) {/* size specifiers */
377 switch ((int)tokval.t_integer) {
378 case S_BYTE:
H. Peter Anvineba20a72002-04-30 20:53:55 +0000379 if (!setsize) /* we want to use only the first */
380 result->oprs[operand].type |= BITS8;
381 setsize = 1;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000382 break;
383 case S_WORD:
H. Peter Anvineba20a72002-04-30 20:53:55 +0000384 if (!setsize)
385 result->oprs[operand].type |= BITS16;
386 setsize = 1;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000387 break;
388 case S_DWORD:
389 case S_LONG:
H. Peter Anvineba20a72002-04-30 20:53:55 +0000390 if (!setsize)
391 result->oprs[operand].type |= BITS32;
392 setsize = 1;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000393 break;
394 case S_QWORD:
H. Peter Anvineba20a72002-04-30 20:53:55 +0000395 if (!setsize)
396 result->oprs[operand].type |= BITS64;
397 setsize = 1;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000398 break;
399 case S_TWORD:
H. Peter Anvineba20a72002-04-30 20:53:55 +0000400 if (!setsize)
401 result->oprs[operand].type |= BITS80;
402 setsize = 1;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000403 break;
404 case S_TO:
405 result->oprs[operand].type |= TO;
406 break;
H. Peter Anvin01377d82002-05-21 03:16:33 +0000407 case S_STRICT:
408 result->oprs[operand].type |= STRICT;
409 break;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000410 case S_FAR:
411 result->oprs[operand].type |= FAR;
412 break;
413 case S_NEAR:
414 result->oprs[operand].type |= NEAR;
415 break;
416 case S_SHORT:
417 result->oprs[operand].type |= SHORT;
418 break;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000419 default:
420 error (ERR_NONFATAL, "invalid operand size specification");
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000421 }
H. Peter Anvin76690a12002-04-30 20:52:49 +0000422 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000423 }
424
H. Peter Anvin6768eb72002-04-30 20:52:26 +0000425 if (i == '[' || i == '&') { /* memory reference */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000426 mref = TRUE;
H. Peter Anvin6768eb72002-04-30 20:52:26 +0000427 bracket = (i == '[');
H. Peter Anvin1cd0e2d2002-04-30 21:00:33 +0000428 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000429 if (i == TOKEN_SPECIAL) { /* check for address size override */
H. Peter Anvin1cd0e2d2002-04-30 21:00:33 +0000430 if (tasm_compatible_mode) {
431 switch ((int)tokval.t_integer) {
432 /* For TASM compatibility a size override inside the
433 * brackets changes the size of the operand, not the
434 * address type of the operand as it does in standard
435 * NASM syntax. Hence:
436 *
437 * mov eax,[DWORD val]
438 *
439 * is valid syntax in TASM compatibility mode. Note that
440 * you lose the ability to override the default address
441 * type for the instruction, but we never use anything
442 * but 32-bit flat model addressing in our code.
443 */
444 case S_BYTE:
445 result->oprs[operand].type |= BITS8;
446 break;
447 case S_WORD:
448 result->oprs[operand].type |= BITS16;
449 break;
450 case S_DWORD:
451 case S_LONG:
452 result->oprs[operand].type |= BITS32;
453 break;
454 case S_QWORD:
455 result->oprs[operand].type |= BITS64;
456 break;
457 case S_TWORD:
458 result->oprs[operand].type |= BITS80;
459 break;
460 default:
461 error (ERR_NONFATAL, "invalid operand size specification");
462 }
463 } else {
H. Peter Anvin1cd0e2d2002-04-30 21:00:33 +0000464 /* Standard NASM compatible syntax */
465 switch ((int)tokval.t_integer) {
466 case S_NOSPLIT:
467 result->oprs[operand].eaflags |= EAF_TIMESTWO;
468 break;
469 case S_BYTE:
470 result->oprs[operand].eaflags |= EAF_BYTEOFFS;
471 break;
472 case S_WORD:
473 result->oprs[operand].addr_size = 16;
474 result->oprs[operand].eaflags |= EAF_WORDOFFS;
475 break;
476 case S_DWORD:
477 case S_LONG:
478 result->oprs[operand].addr_size = 32;
479 result->oprs[operand].eaflags |= EAF_WORDOFFS;
480 break;
481 default:
482 error (ERR_NONFATAL, "invalid size specification in"
483 " effective address");
484 }
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000485 }
H. Peter Anvin76690a12002-04-30 20:52:49 +0000486 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000487 }
H. Peter Anvin6768eb72002-04-30 20:52:26 +0000488 } else { /* immediate operand, or register */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000489 mref = FALSE;
H. Peter Anvin6768eb72002-04-30 20:52:26 +0000490 bracket = FALSE; /* placate optimisers */
491 }
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000492
H. Peter Anvin76690a12002-04-30 20:52:49 +0000493 value = evaluate (stdscan, NULL, &tokval,
H. Peter Anvineba20a72002-04-30 20:53:55 +0000494 &result->oprs[operand].opflags,
495 critical, error, &hints);
H. Peter Anvin76690a12002-04-30 20:52:49 +0000496 i = tokval.t_type;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000497 if (result->oprs[operand].opflags & OPFLAG_FORWARD) {
498 result->forw_ref = TRUE;
499 }
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000500 if (!value) { /* error in evaluator */
501 result->opcode = -1; /* unrecoverable parse error: */
502 return result; /* ignore this instruction */
503 }
504 if (i == ':' && mref) { /* it was seg:offset */
H. Peter Anvin76690a12002-04-30 20:52:49 +0000505 /*
506 * Process the segment override.
507 */
508 if (value[1].type!=0 || value->value!=1 ||
509 REG_SREG & ~reg_flags[value->type])
510 error (ERR_NONFATAL, "invalid segment override");
511 else if (result->nprefix == MAXPREFIX)
512 error (ERR_NONFATAL,
513 "instruction has more than %d prefixes",
514 MAXPREFIX);
515 else
516 result->prefixes[result->nprefix++] = value->type;
517
518 i = stdscan(NULL, &tokval); /* then skip the colon */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000519 if (i == TOKEN_SPECIAL) { /* another check for size override */
520 switch ((int)tokval.t_integer) {
521 case S_WORD:
522 result->oprs[operand].addr_size = 16;
523 break;
524 case S_DWORD:
525 case S_LONG:
526 result->oprs[operand].addr_size = 32;
527 break;
528 default:
529 error (ERR_NONFATAL, "invalid size specification in"
530 " effective address");
531 }
H. Peter Anvin76690a12002-04-30 20:52:49 +0000532 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000533 }
H. Peter Anvin76690a12002-04-30 20:52:49 +0000534 value = evaluate (stdscan, NULL, &tokval,
H. Peter Anvineba20a72002-04-30 20:53:55 +0000535 &result->oprs[operand].opflags,
536 critical, error, &hints);
H. Peter Anvin76690a12002-04-30 20:52:49 +0000537 i = tokval.t_type;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000538 if (result->oprs[operand].opflags & OPFLAG_FORWARD) {
539 result->forw_ref = TRUE;
540 }
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000541 /* and get the offset */
542 if (!value) { /* but, error in evaluator */
543 result->opcode = -1; /* unrecoverable parse error: */
544 return result; /* ignore this instruction */
545 }
H. Peter Anvin76690a12002-04-30 20:52:49 +0000546 }
H. Peter Anvin6768eb72002-04-30 20:52:26 +0000547 if (mref && bracket) { /* find ] at the end */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000548 if (i != ']') {
549 error (ERR_NONFATAL, "parser: expecting ]");
550 do { /* error recovery again */
H. Peter Anvin76690a12002-04-30 20:52:49 +0000551 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000552 } while (i != 0 && i != ',');
553 } else /* we got the required ] */
H. Peter Anvin76690a12002-04-30 20:52:49 +0000554 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000555 } else { /* immediate operand */
556 if (i != 0 && i != ',' && i != ':') {
557 error (ERR_NONFATAL, "comma or end of line expected");
558 do { /* error recovery */
H. Peter Anvin76690a12002-04-30 20:52:49 +0000559 i = stdscan(NULL, &tokval);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000560 } while (i != 0 && i != ',');
561 } else if (i == ':') {
562 result->oprs[operand].type |= COLON;
563 }
564 }
565
566 /* now convert the exprs returned from evaluate() into operand
567 * descriptions... */
568
569 if (mref) { /* it's a memory reference */
570 expr *e = value;
571 int b, i, s; /* basereg, indexreg, scale */
572 long o; /* offset */
573
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000574 b = i = -1, o = s = 0;
H. Peter Anvin76690a12002-04-30 20:52:49 +0000575 result->oprs[operand].hintbase = hints.base;
576 result->oprs[operand].hinttype = hints.type;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000577
H. Peter Anvin76690a12002-04-30 20:52:49 +0000578 if (e->type <= EXPR_REG_END) { /* this bit's a register */
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000579 if (e->value == 1) /* in fact it can be basereg */
580 b = e->type;
581 else /* no, it has to be indexreg */
582 i = e->type, s = e->value;
583 e++;
584 }
H. Peter Anvineba20a72002-04-30 20:53:55 +0000585 if (e->type && e->type <= EXPR_REG_END) /* it's a 2nd register */
586 {
587 if (b != -1) /* If the first was the base, ... */
588 i = e->type, s = e->value; /* second has to be indexreg */
589
590 else if (e->value != 1) /* If both want to be index */
591 {
H. Peter Anvin734b1882002-04-30 21:01:08 +0000592 error(ERR_NONFATAL, "beroset-p-592-invalid effective address");
H. Peter Anvineba20a72002-04-30 20:53:55 +0000593 result->opcode = -1;
594 return result;
595 }
596 else
597 b = e->type;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000598 e++;
599 }
600 if (e->type != 0) { /* is there an offset? */
H. Peter Anvineba20a72002-04-30 20:53:55 +0000601 if (e->type <= EXPR_REG_END) /* in fact, is there an error? */
602 {
H. Peter Anvin734b1882002-04-30 21:01:08 +0000603 error (ERR_NONFATAL, "beroset-p-603-invalid effective address");
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000604 result->opcode = -1;
605 return result;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000606 }
607 else
608 {
H. Peter Anvin76690a12002-04-30 20:52:49 +0000609 if (e->type == EXPR_UNKNOWN) {
H. Peter Anvineba20a72002-04-30 20:53:55 +0000610 o = 0; /* doesn't matter what */
611 result->oprs[operand].wrt = NO_SEG; /* nor this */
H. Peter Anvin76690a12002-04-30 20:52:49 +0000612 result->oprs[operand].segment = NO_SEG; /* or this */
613 while (e->type) e++; /* go to the end of the line */
H. Peter Anvineba20a72002-04-30 20:53:55 +0000614 }
615 else
616 {
H. Peter Anvin76690a12002-04-30 20:52:49 +0000617 if (e->type == EXPR_SIMPLE) {
618 o = e->value;
619 e++;
620 }
621 if (e->type == EXPR_WRT) {
622 result->oprs[operand].wrt = e->value;
623 e++;
624 } else
625 result->oprs[operand].wrt = NO_SEG;
626 /*
627 * Look for a segment base type.
628 */
629 if (e->type && e->type < EXPR_SEGBASE) {
H. Peter Anvin734b1882002-04-30 21:01:08 +0000630 error (ERR_NONFATAL, "beroset-p-630-invalid effective address");
H. Peter Anvin76690a12002-04-30 20:52:49 +0000631 result->opcode = -1;
632 return result;
633 }
634 while (e->type && e->value == 0)
635 e++;
636 if (e->type && e->value != 1) {
H. Peter Anvin734b1882002-04-30 21:01:08 +0000637 error (ERR_NONFATAL, "beroset-p-637-invalid effective address");
H. Peter Anvin76690a12002-04-30 20:52:49 +0000638 result->opcode = -1;
639 return result;
640 }
641 if (e->type) {
642 result->oprs[operand].segment =
643 e->type - EXPR_SEGBASE;
644 e++;
645 } else
646 result->oprs[operand].segment = NO_SEG;
647 while (e->type && e->value == 0)
648 e++;
649 if (e->type) {
H. Peter Anvin734b1882002-04-30 21:01:08 +0000650 error (ERR_NONFATAL, "beroset-p-650-invalid effective address");
H. Peter Anvin76690a12002-04-30 20:52:49 +0000651 result->opcode = -1;
652 return result;
653 }
H. Peter Anvinea838272002-04-30 20:51:53 +0000654 }
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000655 }
656 } else {
657 o = 0;
658 result->oprs[operand].wrt = NO_SEG;
659 result->oprs[operand].segment = NO_SEG;
660 }
661
662 if (e->type != 0) { /* there'd better be nothing left! */
H. Peter Anvin734b1882002-04-30 21:01:08 +0000663 error (ERR_NONFATAL, "beroset-p-663-invalid effective address");
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000664 result->opcode = -1;
665 return result;
666 }
667
668 result->oprs[operand].type |= MEMORY;
669 if (b==-1 && (i==-1 || s==0))
670 result->oprs[operand].type |= MEM_OFFS;
671 result->oprs[operand].basereg = b;
672 result->oprs[operand].indexreg = i;
673 result->oprs[operand].scale = s;
674 result->oprs[operand].offset = o;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000675 }
676 else /* it's not a memory reference */
677 {
H. Peter Anvin76690a12002-04-30 20:52:49 +0000678 if (is_just_unknown(value)) { /* it's immediate but unknown */
679 result->oprs[operand].type |= IMMEDIATE;
680 result->oprs[operand].offset = 0; /* don't care */
681 result->oprs[operand].segment = NO_SEG; /* don't care again */
682 result->oprs[operand].wrt = NO_SEG;/* still don't care */
H. Peter Anvineba20a72002-04-30 20:53:55 +0000683 }
684 else if (is_reloc(value)) /* it's immediate */
685 {
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000686 result->oprs[operand].type |= IMMEDIATE;
687 result->oprs[operand].offset = reloc_value(value);
688 result->oprs[operand].segment = reloc_seg(value);
689 result->oprs[operand].wrt = reloc_wrt(value);
H. Peter Anvinaf535c12002-04-30 20:59:21 +0000690 if (is_simple(value)) {
691 if (reloc_value(value)==1)
692 result->oprs[operand].type |= UNITY;
H. Peter Anvin8c1da7b2002-05-22 20:45:09 +0000693 if (optimizing>=0 &&
694 !(result->oprs[operand].type & STRICT)) {
H. Peter Anvin4cf17482002-04-30 21:01:38 +0000695 if (reloc_value(value) >= -128 &&
696 reloc_value(value) <= 127)
697 result->oprs[operand].type |= SBYTE;
H. Peter Anvin4cf17482002-04-30 21:01:38 +0000698 }
H. Peter Anvinaf535c12002-04-30 20:59:21 +0000699 }
H. Peter Anvineba20a72002-04-30 20:53:55 +0000700 }
701 else /* it's a register */
702 {
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000703 if (value->type>=EXPR_SIMPLE || value->value!=1) {
704 error (ERR_NONFATAL, "invalid operand type");
705 result->opcode = -1;
706 return result;
707 }
H. Peter Anvineba20a72002-04-30 20:53:55 +0000708
709 /*
710 * check that its only 1 register, not an expression...
711 */
712 for (i = 1; value[i].type; i++)
713 if (value[i].value) {
714 error (ERR_NONFATAL, "invalid operand type");
715 result->opcode = -1;
716 return result;
717 }
718
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000719 /* clear overrides, except TO which applies to FPU regs */
H. Peter Anvineba20a72002-04-30 20:53:55 +0000720 if (result->oprs[operand].type & ~TO) {
721 /*
722 * we want to produce a warning iff the specified size
723 * is different from the register size
724 */
725 i = result->oprs[operand].type & SIZE_MASK;
726 }
727 else
728 i = 0;
729
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000730 result->oprs[operand].type &= TO;
731 result->oprs[operand].type |= REGISTER;
732 result->oprs[operand].type |= reg_flags[value->type];
733 result->oprs[operand].basereg = value->type;
H. Peter Anvineba20a72002-04-30 20:53:55 +0000734
735 if (i && (result->oprs[operand].type & SIZE_MASK) != i)
736 error (ERR_WARNING|ERR_PASS1,
737 "register size specification ignored");
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000738 }
739 }
740 }
741
742 result->operands = operand; /* set operand count */
743
744 while (operand<3) /* clear remaining operands */
745 result->oprs[operand++].type = 0;
746
747 /*
748 * Transform RESW, RESD, RESQ, REST into RESB.
749 */
750 switch (result->opcode) {
751 case I_RESW: result->opcode=I_RESB; result->oprs[0].offset*=2; break;
752 case I_RESD: result->opcode=I_RESB; result->oprs[0].offset*=4; break;
753 case I_RESQ: result->opcode=I_RESB; result->oprs[0].offset*=8; break;
754 case I_REST: result->opcode=I_RESB; result->oprs[0].offset*=10; break;
755 }
756
757 return result;
758}
759
H. Peter Anvineba20a72002-04-30 20:53:55 +0000760static int is_comma_next (void)
761{
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000762 char *p;
H. Peter Anvin76690a12002-04-30 20:52:49 +0000763 int i;
764 struct tokenval tv;
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000765
H. Peter Anvin76690a12002-04-30 20:52:49 +0000766 p = stdscan_bufptr;
767 i = stdscan (NULL, &tv);
768 stdscan_bufptr = p;
769 return (i == ',' || i == ';' || !i);
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000770}
771
H. Peter Anvineba20a72002-04-30 20:53:55 +0000772void cleanup_insn (insn *i)
773{
H. Peter Anvinea6e34d2002-04-30 20:51:32 +0000774 extop *e;
775
776 while (i->eops) {
777 e = i->eops;
778 i->eops = i->eops->next;
779 nasm_free (e);
780 }
781}