dt_lex.l revision 297077
1%{ 2/* 3 * CDDL HEADER START 4 * 5 * The contents of this file are subject to the terms of the 6 * Common Development and Distribution License (the "License"). 7 * You may not use this file except in compliance with the License. 8 * 9 * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE 10 * or http://www.opensolaris.org/os/licensing. 11 * See the License for the specific language governing permissions 12 * and limitations under the License. 13 * 14 * When distributing Covered Code, include this CDDL HEADER in each 15 * file and include the License file at usr/src/OPENSOLARIS.LICENSE. 16 * If applicable, add the following below this CDDL HEADER, with the 17 * fields enclosed by brackets "[]" replaced with your own identifying 18 * information: Portions Copyright [yyyy] [name of copyright owner] 19 * 20 * CDDL HEADER END 21 */ 22 23/* 24 * Copyright (c) 2003, 2010, Oracle and/or its affiliates. All rights reserved. 25 */ 26/* 27 * Copyright (c) 2013 by Delphix. All rights reserved. 28 * Copyright (c) 2013, Joyent, Inc. All rights reserved. 29 */ 30 31#include <string.h> 32#include <stdlib.h> 33#include <stdio.h> 34#include <assert.h> 35#include <ctype.h> 36#include <errno.h> 37 38#include <dt_impl.h> 39#include <dt_grammar.h> 40#include <dt_parser.h> 41#include <dt_string.h> 42 43/* 44 * We need to undefine lex's input and unput macros so that references to these 45 * call the functions provided at the end of this source file. 46 */ 47#ifdef illumos 48#undef input 49#undef unput 50#else 51/* 52 * Define YY_INPUT for flex since input() can't be re-defined. 53 */ 54#define YY_INPUT(buf,result,max_size) \ 55 if (yypcb->pcb_fileptr != NULL) { \ 56 if (((result = fread(buf, 1, max_size, yypcb->pcb_fileptr)) == 0) \ 57 && ferror(yypcb->pcb_fileptr)) \ 58 longjmp(yypcb->pcb_jmpbuf, EDT_FIO); \ 59 } else { \ 60 int n; \ 61 for (n = 0; n < max_size && \ 62 yypcb->pcb_strptr < yypcb->pcb_string + yypcb->pcb_strlen; n++) \ 63 buf[n] = *yypcb->pcb_strptr++; \ 64 result = n; \ 65 } 66/* 67 * Do not EOF let tokens to be put back. This does not work with flex. 68 * On the other hand, leaving current buffer in same state it was when 69 * last EOF was received guarantees that input() will keep returning EOF 70 * for all subsequent invocations, which is the effect desired. 71 */ 72#undef unput 73#define unput(c) \ 74 do { \ 75 int _c = c; \ 76 if (_c != EOF) \ 77 yyunput(_c, yytext_ptr); \ 78 } while(0) 79#endif 80 81static int id_or_type(const char *); 82#ifdef illumos 83static int input(void); 84static void unput(int); 85#endif 86 87/* 88 * We first define a set of labeled states for use in the D lexer and then a 89 * set of regular expressions to simplify things below. The lexer states are: 90 * 91 * S0 - D program clause and expression lexing 92 * S1 - D comments (i.e. skip everything until end of comment) 93 * S2 - D program outer scope (probe specifiers and declarations) 94 * S3 - D control line parsing (i.e. after ^# is seen but before \n) 95 * S4 - D control line scan (locate control directives only and invoke S3) 96 */ 97%} 98 99%e 1500 /* maximum nodes */ 100%p 4900 /* maximum positions */ 101%n 600 /* maximum states */ 102%a 3000 /* maximum transitions */ 103 104%s S0 S1 S2 S3 S4 105 106RGX_AGG "@"[a-zA-Z_][0-9a-zA-Z_]* 107RGX_PSPEC [-$:a-zA-Z_.?*\\\[\]!][-$:0-9a-zA-Z_.`?*\\\[\]!]* 108RGX_ALTIDENT [a-zA-Z_][0-9a-zA-Z_]* 109RGX_LMID LM[0-9a-fA-F]+` 110RGX_MOD_IDENT [a-zA-Z_`][0-9a-z.A-Z_`]*` 111RGX_IDENT [a-zA-Z_`][0-9a-zA-Z_`]* 112RGX_INT ([0-9]+|0[xX][0-9A-Fa-f]+)[uU]?[lL]?[lL]? 113RGX_FP ([0-9]+("."?)[0-9]*|"."[0-9]+)((e|E)("+"|-)?[0-9]+)?[fFlL]? 114RGX_WS [\f\n\r\t\v ] 115RGX_STR ([^"\\\n]|\\[^"\n]|\\\")* 116RGX_CHR ([^'\\\n]|\\[^'\n]|\\')* 117RGX_INTERP ^[\f\t\v ]*#!.* 118RGX_CTL ^[\f\t\v ]*# 119 120%% 121 122%{ 123 124/* 125 * We insert a special prologue into yylex() itself: if the pcb contains a 126 * context token, we return that prior to running the normal lexer. This 127 * allows libdtrace to force yacc into one of our three parsing contexts: D 128 * expression (DT_CTX_DEXPR), D program (DT_CTX_DPROG) or D type (DT_CTX_DTYPE). 129 * Once the token is returned, we clear it so this only happens once. 130 */ 131if (yypcb->pcb_token != 0) { 132 int tok = yypcb->pcb_token; 133 yypcb->pcb_token = 0; 134 return (tok); 135} 136 137%} 138 139<S0>auto return (DT_KEY_AUTO); 140<S0>break return (DT_KEY_BREAK); 141<S0>case return (DT_KEY_CASE); 142<S0>char return (DT_KEY_CHAR); 143<S0>const return (DT_KEY_CONST); 144<S0>continue return (DT_KEY_CONTINUE); 145<S0>counter return (DT_KEY_COUNTER); 146<S0>default return (DT_KEY_DEFAULT); 147<S0>do return (DT_KEY_DO); 148<S0>double return (DT_KEY_DOUBLE); 149<S0>else return (DT_KEY_ELSE); 150<S0>enum return (DT_KEY_ENUM); 151<S0>extern return (DT_KEY_EXTERN); 152<S0>float return (DT_KEY_FLOAT); 153<S0>for return (DT_KEY_FOR); 154<S0>goto return (DT_KEY_GOTO); 155<S0>if return (DT_KEY_IF); 156<S0>import return (DT_KEY_IMPORT); 157<S0>inline return (DT_KEY_INLINE); 158<S0>int return (DT_KEY_INT); 159<S0>long return (DT_KEY_LONG); 160<S0>offsetof return (DT_TOK_OFFSETOF); 161<S0>probe return (DT_KEY_PROBE); 162<S0>provider return (DT_KEY_PROVIDER); 163<S0>register return (DT_KEY_REGISTER); 164<S0>restrict return (DT_KEY_RESTRICT); 165<S0>return return (DT_KEY_RETURN); 166<S0>self return (DT_KEY_SELF); 167<S0>short return (DT_KEY_SHORT); 168<S0>signed return (DT_KEY_SIGNED); 169<S0>sizeof return (DT_TOK_SIZEOF); 170<S0>static return (DT_KEY_STATIC); 171<S0>string return (DT_KEY_STRING); 172<S0>stringof return (DT_TOK_STRINGOF); 173<S0>struct return (DT_KEY_STRUCT); 174<S0>switch return (DT_KEY_SWITCH); 175<S0>this return (DT_KEY_THIS); 176<S0>translator return (DT_KEY_XLATOR); 177<S0>typedef return (DT_KEY_TYPEDEF); 178<S0>union return (DT_KEY_UNION); 179<S0>unsigned return (DT_KEY_UNSIGNED); 180<S0>userland return (DT_KEY_USERLAND); 181<S0>void return (DT_KEY_VOID); 182<S0>volatile return (DT_KEY_VOLATILE); 183<S0>while return (DT_KEY_WHILE); 184<S0>xlate return (DT_TOK_XLATE); 185 186<S2>auto { yybegin(YYS_EXPR); return (DT_KEY_AUTO); } 187<S2>char { yybegin(YYS_EXPR); return (DT_KEY_CHAR); } 188<S2>const { yybegin(YYS_EXPR); return (DT_KEY_CONST); } 189<S2>counter { yybegin(YYS_DEFINE); return (DT_KEY_COUNTER); } 190<S2>double { yybegin(YYS_EXPR); return (DT_KEY_DOUBLE); } 191<S2>enum { yybegin(YYS_EXPR); return (DT_KEY_ENUM); } 192<S2>extern { yybegin(YYS_EXPR); return (DT_KEY_EXTERN); } 193<S2>float { yybegin(YYS_EXPR); return (DT_KEY_FLOAT); } 194<S2>import { yybegin(YYS_EXPR); return (DT_KEY_IMPORT); } 195<S2>inline { yybegin(YYS_DEFINE); return (DT_KEY_INLINE); } 196<S2>int { yybegin(YYS_EXPR); return (DT_KEY_INT); } 197<S2>long { yybegin(YYS_EXPR); return (DT_KEY_LONG); } 198<S2>provider { yybegin(YYS_DEFINE); return (DT_KEY_PROVIDER); } 199<S2>register { yybegin(YYS_EXPR); return (DT_KEY_REGISTER); } 200<S2>restrict { yybegin(YYS_EXPR); return (DT_KEY_RESTRICT); } 201<S2>self { yybegin(YYS_EXPR); return (DT_KEY_SELF); } 202<S2>short { yybegin(YYS_EXPR); return (DT_KEY_SHORT); } 203<S2>signed { yybegin(YYS_EXPR); return (DT_KEY_SIGNED); } 204<S2>static { yybegin(YYS_EXPR); return (DT_KEY_STATIC); } 205<S2>string { yybegin(YYS_EXPR); return (DT_KEY_STRING); } 206<S2>struct { yybegin(YYS_EXPR); return (DT_KEY_STRUCT); } 207<S2>this { yybegin(YYS_EXPR); return (DT_KEY_THIS); } 208<S2>translator { yybegin(YYS_DEFINE); return (DT_KEY_XLATOR); } 209<S2>typedef { yybegin(YYS_EXPR); return (DT_KEY_TYPEDEF); } 210<S2>union { yybegin(YYS_EXPR); return (DT_KEY_UNION); } 211<S2>unsigned { yybegin(YYS_EXPR); return (DT_KEY_UNSIGNED); } 212<S2>void { yybegin(YYS_EXPR); return (DT_KEY_VOID); } 213<S2>volatile { yybegin(YYS_EXPR); return (DT_KEY_VOLATILE); } 214 215<S0>"$$"[0-9]+ { 216 int i = atoi(yytext + 2); 217 char *v = ""; 218 219 /* 220 * A macro argument reference substitutes the text of 221 * an argument in place of the current token. When we 222 * see $$<d> we fetch the saved string from pcb_sargv 223 * (or use the default argument if the option has been 224 * set and the argument hasn't been specified) and 225 * return a token corresponding to this string. 226 */ 227 if (i < 0 || (i >= yypcb->pcb_sargc && 228 !(yypcb->pcb_cflags & DTRACE_C_DEFARG))) { 229 xyerror(D_MACRO_UNDEF, "macro argument %s is " 230 "not defined\n", yytext); 231 } 232 233 if (i < yypcb->pcb_sargc) { 234 v = yypcb->pcb_sargv[i]; /* get val from pcb */ 235 yypcb->pcb_sflagv[i] |= DT_IDFLG_REF; 236 } 237 238 if ((yylval.l_str = strdup(v)) == NULL) 239 longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM); 240 241 (void) stresc2chr(yylval.l_str); 242 return (DT_TOK_STRING); 243 } 244 245<S0>"$"[0-9]+ { 246 int i = atoi(yytext + 1); 247 char *p, *v = "0"; 248 249 /* 250 * A macro argument reference substitutes the text of 251 * one identifier or integer pattern for another. When 252 * we see $<d> we fetch the saved string from pcb_sargv 253 * (or use the default argument if the option has been 254 * set and the argument hasn't been specified) and 255 * return a token corresponding to this string. 256 */ 257 if (i < 0 || (i >= yypcb->pcb_sargc && 258 !(yypcb->pcb_cflags & DTRACE_C_DEFARG))) { 259 xyerror(D_MACRO_UNDEF, "macro argument %s is " 260 "not defined\n", yytext); 261 } 262 263 if (i < yypcb->pcb_sargc) { 264 v = yypcb->pcb_sargv[i]; /* get val from pcb */ 265 yypcb->pcb_sflagv[i] |= DT_IDFLG_REF; 266 } 267 268 /* 269 * If the macro text is not a valid integer or ident, 270 * then we treat it as a string. The string may be 271 * optionally enclosed in quotes, which we strip. 272 */ 273 if (strbadidnum(v)) { 274 size_t len = strlen(v); 275 276 if (len != 1 && *v == '"' && v[len - 1] == '"') 277 yylval.l_str = strndup(v + 1, len - 2); 278 else 279 yylval.l_str = strndup(v, len); 280 281 if (yylval.l_str == NULL) 282 longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM); 283 284 (void) stresc2chr(yylval.l_str); 285 return (DT_TOK_STRING); 286 } 287 288 /* 289 * If the macro text is not a string an begins with a 290 * digit or a +/- sign, process it as an integer token. 291 */ 292 if (isdigit(v[0]) || v[0] == '-' || v[0] == '+') { 293 if (isdigit(v[0])) 294 yyintprefix = 0; 295 else 296 yyintprefix = *v++; 297 298 errno = 0; 299 yylval.l_int = strtoull(v, &p, 0); 300 (void) strncpy(yyintsuffix, p, 301 sizeof (yyintsuffix)); 302 yyintdecimal = *v != '0'; 303 304 if (errno == ERANGE) { 305 xyerror(D_MACRO_OFLOW, "macro argument" 306 " %s constant %s results in integer" 307 " overflow\n", yytext, v); 308 } 309 310 return (DT_TOK_INT); 311 } 312 313 return (id_or_type(v)); 314 } 315 316<S0>"$$"{RGX_IDENT} { 317 dt_ident_t *idp = dt_idhash_lookup( 318 yypcb->pcb_hdl->dt_macros, yytext + 2); 319 320 char s[16]; /* enough for UINT_MAX + \0 */ 321 322 if (idp == NULL) { 323 xyerror(D_MACRO_UNDEF, "macro variable %s " 324 "is not defined\n", yytext); 325 } 326 327 /* 328 * For the moment, all current macro variables are of 329 * type id_t (refer to dtrace_update() for details). 330 */ 331 (void) snprintf(s, sizeof (s), "%u", idp->di_id); 332 if ((yylval.l_str = strdup(s)) == NULL) 333 longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM); 334 335 return (DT_TOK_STRING); 336 } 337 338<S0>"$"{RGX_IDENT} { 339 dt_ident_t *idp = dt_idhash_lookup( 340 yypcb->pcb_hdl->dt_macros, yytext + 1); 341 342 if (idp == NULL) { 343 xyerror(D_MACRO_UNDEF, "macro variable %s " 344 "is not defined\n", yytext); 345 } 346 347 /* 348 * For the moment, all current macro variables are of 349 * type id_t (refer to dtrace_update() for details). 350 */ 351 yylval.l_int = (intmax_t)(int)idp->di_id; 352 yyintprefix = 0; 353 yyintsuffix[0] = '\0'; 354 yyintdecimal = 1; 355 356 return (DT_TOK_INT); 357 } 358 359<S0>{RGX_IDENT} | 360<S0>{RGX_MOD_IDENT}{RGX_IDENT} | 361<S0>{RGX_MOD_IDENT} { 362 return (id_or_type(yytext)); 363 } 364 365<S0>{RGX_AGG} { 366 if ((yylval.l_str = strdup(yytext)) == NULL) 367 longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM); 368 return (DT_TOK_AGG); 369 } 370 371<S0>"@" { 372 if ((yylval.l_str = strdup("@_")) == NULL) 373 longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM); 374 return (DT_TOK_AGG); 375 } 376 377<S0>{RGX_INT} | 378<S2>{RGX_INT} | 379<S3>{RGX_INT} { 380 char *p; 381 382 errno = 0; 383 yylval.l_int = strtoull(yytext, &p, 0); 384 yyintprefix = 0; 385 (void) strncpy(yyintsuffix, p, sizeof (yyintsuffix)); 386 yyintdecimal = yytext[0] != '0'; 387 388 if (errno == ERANGE) { 389 xyerror(D_INT_OFLOW, "constant %s results in " 390 "integer overflow\n", yytext); 391 } 392 393 if (*p != '\0' && strchr("uUlL", *p) == NULL) { 394 xyerror(D_INT_DIGIT, "constant %s contains " 395 "invalid digit %c\n", yytext, *p); 396 } 397 398 if ((YYSTATE) != S3) 399 return (DT_TOK_INT); 400 401 yypragma = dt_node_link(yypragma, 402 dt_node_int(yylval.l_int)); 403 } 404 405<S0>{RGX_FP} yyerror("floating-point constants are not permitted\n"); 406 407<S0>\"{RGX_STR}$ | 408<S3>\"{RGX_STR}$ xyerror(D_STR_NL, "newline encountered in string literal"); 409 410<S0>\"{RGX_STR}\" | 411<S3>\"{RGX_STR}\" { 412 /* 413 * Quoted string -- convert C escape sequences and 414 * return the string as a token. 415 */ 416 yylval.l_str = strndup(yytext + 1, yyleng - 2); 417 418 if (yylval.l_str == NULL) 419 longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM); 420 421 (void) stresc2chr(yylval.l_str); 422 if ((YYSTATE) != S3) 423 return (DT_TOK_STRING); 424 425 yypragma = dt_node_link(yypragma, 426 dt_node_string(yylval.l_str)); 427 } 428 429<S0>'{RGX_CHR}$ xyerror(D_CHR_NL, "newline encountered in character constant"); 430 431<S0>'{RGX_CHR}' { 432 char *s, *p, *q; 433 size_t nbytes; 434 435 /* 436 * Character constant -- convert C escape sequences and 437 * return the character as an integer immediate value. 438 */ 439 if (yyleng == 2) 440 xyerror(D_CHR_NULL, "empty character constant"); 441 442 s = yytext + 1; 443 yytext[yyleng - 1] = '\0'; 444 nbytes = stresc2chr(s); 445 yylval.l_int = 0; 446 yyintprefix = 0; 447 yyintsuffix[0] = '\0'; 448 yyintdecimal = 1; 449 450 if (nbytes > sizeof (yylval.l_int)) { 451 xyerror(D_CHR_OFLOW, "character constant is " 452 "too long"); 453 } 454#if BYTE_ORDER == _LITTLE_ENDIAN 455 p = ((char *)&yylval.l_int) + nbytes - 1; 456 for (q = s; nbytes != 0; nbytes--) 457 *p-- = *q++; 458#else 459 bcopy(s, ((char *)&yylval.l_int) + 460 sizeof (yylval.l_int) - nbytes, nbytes); 461#endif 462 return (DT_TOK_INT); 463 } 464 465<S0>"/*" | 466<S2>"/*" { 467 yypcb->pcb_cstate = (YYSTATE); 468 BEGIN(S1); 469 } 470 471<S0>{RGX_INTERP} | 472<S2>{RGX_INTERP} ; /* discard any #! lines */ 473 474<S0>{RGX_CTL} | 475<S2>{RGX_CTL} | 476<S4>{RGX_CTL} { 477 assert(yypragma == NULL); 478 yypcb->pcb_cstate = (YYSTATE); 479 BEGIN(S3); 480 } 481 482<S4>. ; /* discard */ 483<S4>"\n" ; /* discard */ 484 485<S0>"/" { 486 int c, tok; 487 488 /* 489 * The use of "/" as the predicate delimiter and as the 490 * integer division symbol requires special lookahead 491 * to avoid a shift/reduce conflict in the D grammar. 492 * We look ahead to the next non-whitespace character. 493 * If we encounter EOF, ";", "{", or "/", then this "/" 494 * closes the predicate and we return DT_TOK_EPRED. 495 * If we encounter anything else, it's DT_TOK_DIV. 496 */ 497 while ((c = input()) != 0) { 498 if (strchr("\f\n\r\t\v ", c) == NULL) 499 break; 500 } 501 502 if (c == 0 || c == ';' || c == '{' || c == '/') { 503 if (yypcb->pcb_parens != 0) { 504 yyerror("closing ) expected in " 505 "predicate before /\n"); 506 } 507 if (yypcb->pcb_brackets != 0) { 508 yyerror("closing ] expected in " 509 "predicate before /\n"); 510 } 511 tok = DT_TOK_EPRED; 512 } else 513 tok = DT_TOK_DIV; 514 515 unput(c); 516 return (tok); 517 } 518 519<S0>"(" { 520 yypcb->pcb_parens++; 521 return (DT_TOK_LPAR); 522 } 523 524<S0>")" { 525 if (--yypcb->pcb_parens < 0) 526 yyerror("extra ) in input stream\n"); 527 return (DT_TOK_RPAR); 528 } 529 530<S0>"[" { 531 yypcb->pcb_brackets++; 532 return (DT_TOK_LBRAC); 533 } 534 535<S0>"]" { 536 if (--yypcb->pcb_brackets < 0) 537 yyerror("extra ] in input stream\n"); 538 return (DT_TOK_RBRAC); 539 } 540 541<S0>"{" | 542<S2>"{" { 543 yypcb->pcb_braces++; 544 return ('{'); 545 } 546 547<S0>"}" { 548 if (--yypcb->pcb_braces < 0) 549 yyerror("extra } in input stream\n"); 550 return ('}'); 551 } 552 553<S0>"|" return (DT_TOK_BOR); 554<S0>"^" return (DT_TOK_XOR); 555<S0>"&" return (DT_TOK_BAND); 556<S0>"&&" return (DT_TOK_LAND); 557<S0>"^^" return (DT_TOK_LXOR); 558<S0>"||" return (DT_TOK_LOR); 559<S0>"==" return (DT_TOK_EQU); 560<S0>"!=" return (DT_TOK_NEQ); 561<S0>"<" return (DT_TOK_LT); 562<S0>"<=" return (DT_TOK_LE); 563<S0>">" return (DT_TOK_GT); 564<S0>">=" return (DT_TOK_GE); 565<S0>"<<" return (DT_TOK_LSH); 566<S0>">>" return (DT_TOK_RSH); 567<S0>"+" return (DT_TOK_ADD); 568<S0>"-" return (DT_TOK_SUB); 569<S0>"*" return (DT_TOK_MUL); 570<S0>"%" return (DT_TOK_MOD); 571<S0>"~" return (DT_TOK_BNEG); 572<S0>"!" return (DT_TOK_LNEG); 573<S0>"?" return (DT_TOK_QUESTION); 574<S0>":" return (DT_TOK_COLON); 575<S0>"." return (DT_TOK_DOT); 576<S0>"->" return (DT_TOK_PTR); 577<S0>"=" return (DT_TOK_ASGN); 578<S0>"+=" return (DT_TOK_ADD_EQ); 579<S0>"-=" return (DT_TOK_SUB_EQ); 580<S0>"*=" return (DT_TOK_MUL_EQ); 581<S0>"/=" return (DT_TOK_DIV_EQ); 582<S0>"%=" return (DT_TOK_MOD_EQ); 583<S0>"&=" return (DT_TOK_AND_EQ); 584<S0>"^=" return (DT_TOK_XOR_EQ); 585<S0>"|=" return (DT_TOK_OR_EQ); 586<S0>"<<=" return (DT_TOK_LSH_EQ); 587<S0>">>=" return (DT_TOK_RSH_EQ); 588<S0>"++" return (DT_TOK_ADDADD); 589<S0>"--" return (DT_TOK_SUBSUB); 590<S0>"..." return (DT_TOK_ELLIPSIS); 591<S0>"," return (DT_TOK_COMMA); 592<S0>";" return (';'); 593<S0>{RGX_WS} ; /* discard */ 594<S0>"\\"\n ; /* discard */ 595<S0>. yyerror("syntax error near \"%c\"\n", yytext[0]); 596 597<S1>"/*" yyerror("/* encountered inside a comment\n"); 598<S1>"*/" BEGIN(yypcb->pcb_cstate); 599<S1>.|\n ; /* discard */ 600 601<S2>{RGX_PSPEC} { 602 /* 603 * S2 has an ambiguity because RGX_PSPEC includes '*' 604 * as a glob character and '*' also can be DT_TOK_STAR. 605 * Since lex always matches the longest token, this 606 * rule can be matched by an input string like "int*", 607 * which could begin a global variable declaration such 608 * as "int*x;" or could begin a RGX_PSPEC with globbing 609 * such as "int* { trace(timestamp); }". If C_PSPEC is 610 * not set, we must resolve the ambiguity in favor of 611 * the type and perform lexer pushback if the fragment 612 * before '*' or entire fragment matches a type name. 613 * If C_PSPEC is set, we always return a PSPEC token. 614 * If C_PSPEC is off, the user can avoid ambiguity by 615 * including a ':' delimiter in the specifier, which 616 * they should be doing anyway to specify the provider. 617 */ 618 if (!(yypcb->pcb_cflags & DTRACE_C_PSPEC) && 619 strchr(yytext, ':') == NULL) { 620 621 char *p = strchr(yytext, '*'); 622 char *q = yytext + yyleng - 1; 623 624 if (p != NULL && p > yytext) 625 *p = '\0'; /* prune yytext */ 626 627 if (dt_type_lookup(yytext, NULL) == 0) { 628 yylval.l_str = strdup(yytext); 629 630 if (yylval.l_str == NULL) { 631 longjmp(yypcb->pcb_jmpbuf, 632 EDT_NOMEM); 633 } 634 635 if (p != NULL && p > yytext) { 636 for (*p = '*'; q >= p; q--) 637 unput(*q); 638 } 639 640 yybegin(YYS_EXPR); 641 return (DT_TOK_TNAME); 642 } 643 644 if (p != NULL && p > yytext) 645 *p = '*'; /* restore yytext */ 646 } 647 648 if ((yylval.l_str = strdup(yytext)) == NULL) 649 longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM); 650 651 return (DT_TOK_PSPEC); 652 } 653 654<S2>"/" return (DT_TOK_DIV); 655<S2>"," return (DT_TOK_COMMA); 656 657<S2>{RGX_WS} ; /* discard */ 658<S2>. yyerror("syntax error near \"%c\"\n", yytext[0]); 659 660<S3>\n { 661 dt_pragma(yypragma); 662 yypragma = NULL; 663 BEGIN(yypcb->pcb_cstate); 664 } 665 666<S3>[\f\t\v ]+ ; /* discard */ 667 668<S3>[^\f\n\t\v "]+ { 669 dt_node_t *dnp; 670 671 if ((yylval.l_str = strdup(yytext)) == NULL) 672 longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM); 673 674 /* 675 * We want to call dt_node_ident() here, but we can't 676 * because it will expand inlined identifiers, which we 677 * don't want to do from #pragma context in order to 678 * support pragmas that apply to the ident itself. We 679 * call dt_node_string() and then reset dn_op instead. 680 */ 681 dnp = dt_node_string(yylval.l_str); 682 dnp->dn_kind = DT_NODE_IDENT; 683 dnp->dn_op = DT_TOK_IDENT; 684 yypragma = dt_node_link(yypragma, dnp); 685 } 686 687<S3>. yyerror("syntax error near \"%c\"\n", yytext[0]); 688 689%% 690 691/* 692 * yybegin provides a wrapper for use from C code around the lex BEGIN() macro. 693 * We use two main states for lexing because probe descriptions use a syntax 694 * that is incompatible with the normal D tokens (e.g. names can contain "-"). 695 * yybegin also handles the job of switching between two lists of dt_nodes 696 * as we allocate persistent definitions, like inlines, and transient nodes 697 * that will be freed once we are done parsing the current program file. 698 */ 699void 700yybegin(yystate_t state) 701{ 702#ifdef YYDEBUG 703 yydebug = _dtrace_debug; 704#endif 705 if (yypcb->pcb_yystate == state) 706 return; /* nothing to do if we're in the state already */ 707 708 if (yypcb->pcb_yystate == YYS_DEFINE) { 709 yypcb->pcb_list = yypcb->pcb_hold; 710 yypcb->pcb_hold = NULL; 711 } 712 713 switch (state) { 714 case YYS_CLAUSE: 715 BEGIN(S2); 716 break; 717 case YYS_DEFINE: 718 assert(yypcb->pcb_hold == NULL); 719 yypcb->pcb_hold = yypcb->pcb_list; 720 yypcb->pcb_list = NULL; 721 /*FALLTHRU*/ 722 case YYS_EXPR: 723 BEGIN(S0); 724 break; 725 case YYS_DONE: 726 break; 727 case YYS_CONTROL: 728 BEGIN(S4); 729 break; 730 default: 731 xyerror(D_UNKNOWN, "internal error -- bad yystate %d\n", state); 732 } 733 734 yypcb->pcb_yystate = state; 735} 736 737void 738yyinit(dt_pcb_t *pcb) 739{ 740 yypcb = pcb; 741 yylineno = 1; 742 yypragma = NULL; 743#ifdef illumos 744 yysptr = yysbuf; 745#endif 746 YY_FLUSH_BUFFER; 747} 748 749/* 750 * Given a lexeme 's' (typically yytext), set yylval and return an appropriate 751 * token to the parser indicating either an identifier or a typedef name. 752 * User-defined global variables always take precedence over types, but we do 753 * use some heuristics because D programs can look at an ever-changing set of 754 * kernel types and also can implicitly instantiate variables by assignment, 755 * unlike in C. The code here is ordered carefully as lookups are not cheap. 756 */ 757static int 758id_or_type(const char *s) 759{ 760 dtrace_hdl_t *dtp = yypcb->pcb_hdl; 761 dt_decl_t *ddp = yypcb->pcb_dstack.ds_decl; 762 int c0, c1, ttok = DT_TOK_TNAME; 763 dt_ident_t *idp; 764 765 if ((s = yylval.l_str = strdup(s)) == NULL) 766 longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM); 767 768 /* 769 * If the lexeme is a global variable or likely identifier or *not* a 770 * type_name, then it is an identifier token. 771 */ 772 if (dt_idstack_lookup(&yypcb->pcb_globals, s) != NULL || 773 dt_idhash_lookup(yypcb->pcb_idents, s) != NULL || 774 dt_type_lookup(s, NULL) != 0) 775 return (DT_TOK_IDENT); 776 777 /* 778 * If we're in the midst of parsing a declaration and a type_specifier 779 * has already been shifted, then return DT_TOK_IDENT instead of TNAME. 780 * This semantic is necessary to permit valid ISO C code such as: 781 * 782 * typedef int foo; 783 * struct s { foo foo; }; 784 * 785 * without causing shift/reduce conflicts in the direct_declarator part 786 * of the grammar. The result is that we must check for conflicting 787 * redeclarations of the same identifier as part of dt_node_decl(). 788 */ 789 if (ddp != NULL && ddp->dd_name != NULL) 790 return (DT_TOK_IDENT); 791 792 /* 793 * If the lexeme is a type name and we are not in a program clause, 794 * then always interpret it as a type and return DT_TOK_TNAME. 795 */ 796 if ((YYSTATE) != S0) 797 return (DT_TOK_TNAME); 798 799 /* 800 * If the lexeme matches a type name but is in a program clause, then 801 * it could be a type or it could be an undefined variable. Peek at 802 * the next token to decide. If we see ++, --, [, or =, we know there 803 * might be an assignment that is trying to create a global variable, 804 * so we optimistically return DT_TOK_IDENT. There is no harm in being 805 * wrong: a type_name followed by ++, --, [, or = is a syntax error. 806 */ 807 while ((c0 = input()) != 0) { 808 if (strchr("\f\n\r\t\v ", c0) == NULL) 809 break; 810 } 811 812 switch (c0) { 813 case '+': 814 case '-': 815 if ((c1 = input()) == c0) 816 ttok = DT_TOK_IDENT; 817 unput(c1); 818 break; 819 820 case '=': 821 if ((c1 = input()) != c0) 822 ttok = DT_TOK_IDENT; 823 unput(c1); 824 break; 825 case '[': 826 ttok = DT_TOK_IDENT; 827 break; 828 } 829 830 if (ttok == DT_TOK_IDENT) { 831 idp = dt_idhash_insert(yypcb->pcb_idents, s, DT_IDENT_SCALAR, 0, 832 0, _dtrace_defattr, 0, &dt_idops_thaw, NULL, dtp->dt_gen); 833 834 if (idp == NULL) 835 longjmp(yypcb->pcb_jmpbuf, EDT_NOMEM); 836 } 837 838 unput(c0); 839 return (ttok); 840} 841 842#ifdef illumos 843static int 844input(void) 845{ 846 int c; 847 848 if (yysptr > yysbuf) 849 c = *--yysptr; 850 else if (yypcb->pcb_fileptr != NULL) 851 c = fgetc(yypcb->pcb_fileptr); 852 else if (yypcb->pcb_strptr < yypcb->pcb_string + yypcb->pcb_strlen) 853 c = *(unsigned char *)(yypcb->pcb_strptr++); 854 else 855 c = EOF; 856 857 if (c == '\n') 858 yylineno++; 859 860 if (c != EOF) 861 return (c); 862 863 if ((YYSTATE) == S1) 864 yyerror("end-of-file encountered before matching */\n"); 865 866 if ((YYSTATE) == S3) 867 yyerror("end-of-file encountered before end of control line\n"); 868 869 if (yypcb->pcb_fileptr != NULL && ferror(yypcb->pcb_fileptr)) 870 longjmp(yypcb->pcb_jmpbuf, EDT_FIO); 871 872 return (0); /* EOF */ 873} 874 875static void 876unput(int c) 877{ 878 if (c == '\n') 879 yylineno--; 880 881 *yysptr++ = c; 882 yytchar = c; 883} 884#endif /* illumos */ 885