git://sourceware.org / systemtap.git / blame
| Commit | Line | Data |
|---|---|---|
| 2b066ec1 | 1 | // elaboration functions |
| 69c68955 FCE |
2 | // Copyright (C) 2005 Red Hat Inc. |
| 3 | // | |
| 4 | // This file is part of systemtap, and is free software. You can | |
| 5 | // redistribute it and/or modify it under the terms of the GNU General | |
| 6 | // Public License (GPL); either version 2, or (at your option) any | |
| 7 | // later version. | |
| 2b066ec1 FCE |
8 | |
| 9 | #include "config.h" | |
| 10 | #include "elaborate.h" | |
| 11 | #include "parse.h" | |
| 20c6c071 | 12 | #include "tapsets.h" |
| 2b066ec1 FCE |
13 | |
| 14 | extern "C" { | |
| 15 | #include <sys/utsname.h> | |
| 16 | } | |
| 17 | ||
| 2b066ec1 | 18 | #include <algorithm> |
| 67c0a579 GH |
19 | #include <fstream> |
| 20 | #include <map> | |
| 21 | #include <set> | |
| 22 | #include <vector> | |
| 2b066ec1 | 23 | |
| 2b066ec1 FCE |
24 | using namespace std; |
| 25 | ||
| 26 | ||
| 27 | // ------------------------------------------------------------------------ | |
| 28 | ||
| 29 | ||
| 30 | derived_probe::derived_probe (probe *p): | |
| 31 | base (p) | |
| 32 | { | |
| 77de5e9e GH |
33 | if (p) |
| 34 | { | |
| 35 | this->locations = p->locations; | |
| 36 | this->tok = p->tok; | |
| 37 | this->body = deep_copy_visitor::deep_copy(p->body); | |
| 38 | } | |
| 2b066ec1 FCE |
39 | } |
| 40 | ||
| 41 | ||
| 42 | derived_probe::derived_probe (probe *p, probe_point *l): | |
| 43 | base (p) | |
| 44 | { | |
| 45 | this->locations.push_back (l); | |
| 77de5e9e GH |
46 | if (p) |
| 47 | { | |
| 48 | this->tok = p->tok; | |
| 49 | this->body = deep_copy_visitor::deep_copy(p->body); | |
| 50 | } | |
| 2b066ec1 FCE |
51 | } |
| 52 | ||
| 20c6c071 GH |
53 | // ------------------------------------------------------------------------ |
| 54 | ||
| 55 | // Members of match_key. | |
| 56 | ||
| 57 | match_key::match_key(string const & n) | |
| 58 | : name(n), | |
| 59 | have_parameter(false), | |
| 60 | parameter_type(tok_junk) | |
| 61 | { | |
| 62 | } | |
| 63 | ||
| 64 | match_key::match_key(probe_point::component const & c) | |
| 65 | : name(c.functor), | |
| 66 | have_parameter(c.arg != NULL), | |
| 67 | parameter_type(c.arg ? c.arg->tok->type : tok_junk) | |
| 68 | { | |
| 69 | } | |
| 70 | ||
| 71 | match_key & | |
| 72 | match_key::with_number() | |
| 73 | { | |
| 74 | have_parameter = true; | |
| 75 | parameter_type = tok_number; | |
| 76 | return *this; | |
| 77 | } | |
| 78 | ||
| 79 | match_key & | |
| 80 | match_key::with_string() | |
| 81 | { | |
| 82 | have_parameter = true; | |
| 83 | parameter_type = tok_string; | |
| 84 | return *this; | |
| 85 | } | |
| 86 | ||
| 87 | string | |
| 88 | match_key::str() const | |
| 89 | { | |
| 90 | if (have_parameter) | |
| 91 | switch (parameter_type) | |
| 92 | { | |
| 93 | case tok_string: return name + "(string)"; | |
| 94 | case tok_number: return name + "(number)"; | |
| 95 | default: return name + "(...)"; | |
| 96 | } | |
| 97 | return name; | |
| 98 | } | |
| 99 | ||
| 100 | bool | |
| 101 | match_key::operator<(match_key const & other) const | |
| 102 | { | |
| 103 | return ((name < other.name) | |
| 104 | ||
| 105 | || (name == name | |
| 106 | && have_parameter < other.have_parameter) | |
| 107 | ||
| 108 | || (name == name | |
| 109 | && have_parameter == other.have_parameter | |
| 110 | && parameter_type < other.parameter_type)); | |
| 111 | } | |
| 2b066ec1 FCE |
112 | |
| 113 | // ------------------------------------------------------------------------ | |
| 20c6c071 GH |
114 | // Members of match_node |
| 115 | // ------------------------------------------------------------------------ | |
| 116 | ||
| 117 | match_node::match_node() | |
| 118 | : end(NULL) | |
| 119 | {} | |
| 120 | ||
| 121 | match_node * | |
| 122 | match_node::bind(match_key const & k) | |
| 123 | { | |
| 124 | map<match_key, match_node *>::const_iterator i = sub.find(k); | |
| 125 | if (i != sub.end()) | |
| 126 | return i->second; | |
| 127 | match_node * n = new match_node(); | |
| 128 | sub.insert(make_pair(k, n)); | |
| 129 | return n; | |
| 130 | } | |
| 131 | ||
| 132 | void | |
| 133 | match_node::bind(derived_probe_builder * e) | |
| 134 | { | |
| 135 | if (end) | |
| 136 | throw semantic_error("already have a pattern ending"); | |
| 137 | end = e; | |
| 138 | } | |
| 139 | ||
| 140 | match_node * | |
| 141 | match_node::bind(string const & k) | |
| 142 | { | |
| 143 | return bind(match_key(k)); | |
| 144 | } | |
| 145 | ||
| 146 | match_node * | |
| 147 | match_node::bind_str(string const & k) | |
| 148 | { | |
| 149 | return bind(match_key(k).with_string()); | |
| 150 | } | |
| 151 | ||
| 152 | match_node * | |
| 153 | match_node::bind_num(string const & k) | |
| 154 | { | |
| 155 | return bind(match_key(k).with_number()); | |
| 156 | } | |
| 157 | ||
| 158 | derived_probe_builder * | |
| 159 | match_node::find_builder(vector<probe_point::component *> const & components, | |
| 160 | unsigned pos, | |
| 161 | vector< pair<string, literal *> > & parameters) | |
| 162 | { | |
| 163 | assert(pos <= components.size()); | |
| 164 | if (pos == components.size()) | |
| 165 | { | |
| 166 | // Probe_point ends here. We match iff we have | |
| 167 | // an "end" entry here. If we don't, it'll be null. | |
| 168 | return end; | |
| 169 | } | |
| 170 | else | |
| 171 | { | |
| 172 | // Probe_point contains a component here. We match iff there's | |
| 173 | // an entry in the sub table, and its value matches the rest | |
| 174 | // of the probe_point. | |
| 175 | match_key k(*components[pos]); | |
| f4b28491 | 176 | if (0) // session.verbose |
| 20c6c071 GH |
177 | clog << "searching for component " << k.str() << endl; |
| 178 | map<match_key, match_node *>::const_iterator i = sub.find(k); | |
| 179 | if (i == sub.end()) | |
| 180 | { | |
| f4b28491 | 181 | if (0) // session.verbose |
| 20c6c071 GH |
182 | clog << "no match found" << endl; |
| 183 | return NULL; | |
| 184 | } | |
| 185 | else | |
| 186 | { | |
| f4b28491 | 187 | if (0) // session.verbose |
| 20c6c071 GH |
188 | clog << "matched " << k.str() << endl; |
| 189 | derived_probe_builder * builder = NULL; | |
| 190 | if (k.have_parameter) | |
| 191 | { | |
| 192 | assert(components[pos]->arg); | |
| 193 | parameters.push_back(make_pair(components[pos]->functor, | |
| 194 | components[pos]->arg)); | |
| 195 | } | |
| 196 | else | |
| 197 | { | |
| 198 | // store a "null parameter" for any component we run into, anyways | |
| 199 | literal_string *empty = NULL; | |
| 200 | parameters.push_back(make_pair(components[pos]->functor, empty)); | |
| 201 | } | |
| 202 | builder = i->second->find_builder(components, pos+1, parameters); | |
| 203 | if (k.have_parameter && !builder) | |
| 204 | parameters.pop_back(); | |
| 205 | return builder; | |
| 206 | } | |
| 207 | } | |
| 208 | } | |
| 209 | ||
| 210 | ||
| 211 | static void | |
| 212 | param_vec_to_map(vector< pair<string, literal *> > const & param_vec, | |
| 213 | map<string, literal *> & param_map) | |
| 214 | { | |
| 215 | for (vector< pair<string, literal *> >::const_iterator i = param_vec.begin(); | |
| 216 | i != param_vec.end(); ++i) | |
| 217 | { | |
| 218 | param_map[i->first] = i->second; | |
| 219 | } | |
| 220 | } | |
| 221 | ||
| 222 | // ------------------------------------------------------------------------ | |
| 223 | // Alias probes | |
| 224 | // ------------------------------------------------------------------------ | |
| 225 | ||
| 226 | struct | |
| 227 | alias_expansion_builder | |
| 228 | : public derived_probe_builder | |
| 229 | { | |
| 230 | probe_alias * alias; | |
| 231 | ||
| 232 | alias_expansion_builder(probe_alias * a) | |
| 233 | : alias(a) | |
| 234 | {} | |
| 235 | ||
| 5227f1ea GH |
236 | virtual void build(systemtap_session & sess, |
| 237 | probe * use, | |
| 20c6c071 GH |
238 | probe_point * location, |
| 239 | std::map<std::string, literal *> const & parameters, | |
| 240 | vector<probe *> & results_to_expand_further, | |
| 241 | vector<derived_probe *> & finished_results) | |
| 242 | { | |
| 243 | // We're going to build a new probe and wrap it up in an | |
| 244 | // alias_expansion_probe so that the expansion loop recognizer it as | |
| 245 | // such and re-expands its expansion. | |
| 246 | ||
| 247 | probe * n = new probe(); | |
| 248 | n->body = new block(); | |
| 249 | ||
| 250 | // The new probe gets the location list of the alias, | |
| 251 | n->locations = alias->locations; | |
| 252 | ||
| 253 | // the token location of the use, | |
| 254 | n->tok = use->tok; | |
| 255 | ||
| 5227f1ea GH |
256 | // and statements representing the concatenation of the alias' |
| 257 | // body with the use's. | |
| 258 | // | |
| 259 | // NB: locals are *not* copied forward, from either alias or | |
| 260 | // use. The expansion should have its locals re-inferred since | |
| 261 | // there's concatenated code here and we only want one vardecl per | |
| 262 | // resulting variable. | |
| 263 | ||
| 85365d1b GH |
264 | for (unsigned i = 0; i < alias->body->statements.size(); ++i) |
| 265 | { | |
| 266 | statement *s = deep_copy_visitor::deep_copy(alias->body->statements[i]); | |
| 267 | n->body->statements.push_back(s); | |
| 268 | } | |
| 269 | ||
| 270 | for (unsigned i = 0; i < use->body->statements.size(); ++i) | |
| 271 | { | |
| 272 | statement *s = deep_copy_visitor::deep_copy(use->body->statements[i]); | |
| 273 | n->body->statements.push_back(s); | |
| 274 | } | |
| 20c6c071 GH |
275 | |
| 276 | results_to_expand_further.push_back(n); | |
| 277 | } | |
| 278 | }; | |
| 279 | ||
| 280 | ||
| 281 | // ------------------------------------------------------------------------ | |
| 282 | // Pattern matching | |
| 283 | // ------------------------------------------------------------------------ | |
| 284 | ||
| 285 | ||
| 286 | // Register all the aliases we've seen in library files, and the user | |
| 287 | // file, as patterns. | |
| 288 | ||
| 289 | void | |
| 290 | systemtap_session::register_library_aliases() | |
| 291 | { | |
| 292 | vector<stapfile*> files(library_files); | |
| 293 | files.push_back(user_file); | |
| 294 | ||
| 295 | for (unsigned f = 0; f < files.size(); ++f) | |
| 296 | { | |
| 297 | stapfile * file = files[f]; | |
| 298 | for (unsigned a = 0; a < file->aliases.size(); ++a) | |
| 299 | { | |
| 300 | probe_alias * alias = file->aliases[a]; | |
| 301 | for (unsigned n = 0; n < alias->alias_names.size(); ++n) | |
| 302 | { | |
| 303 | probe_point * name = alias->alias_names[n]; | |
| 304 | if (verbose) | |
| 305 | { | |
| 306 | clog << "registering probe alias "; | |
| 307 | for (unsigned c = 0; c < name->components.size(); ++c) | |
| 308 | clog << (c > 0 ? "." : "") << name->components[c]->functor; | |
| 309 | clog << endl; | |
| 310 | } | |
| 311 | match_node * n = pattern_root; | |
| 312 | for (unsigned c = 0; c < name->components.size(); ++c) | |
| 313 | { | |
| 314 | probe_point::component * comp = name->components[c]; | |
| 315 | // XXX: alias parameters | |
| 316 | if (comp->arg) | |
| 317 | throw semantic_error("alias component " | |
| 318 | + comp->functor | |
| 319 | + " contains illegal parameter"); | |
| 320 | n = n->bind(comp->functor); | |
| 321 | } | |
| 322 | n->bind(new alias_expansion_builder(alias)); | |
| 323 | } | |
| 324 | } | |
| 325 | } | |
| 326 | } | |
| 327 | ||
| 328 | ||
| 5227f1ea GH |
329 | static unsigned max_recursion = 100; |
| 330 | ||
| 331 | struct | |
| 332 | recursion_guard | |
| 333 | { | |
| 334 | unsigned & i; | |
| 335 | recursion_guard(unsigned & i) : i(i) | |
| 336 | { | |
| 337 | if (i > max_recursion) | |
| 338 | throw semantic_error("recursion limit reached"); | |
| 339 | ++i; | |
| 340 | } | |
| 341 | ~recursion_guard() | |
| 342 | { | |
| 343 | --i; | |
| 344 | } | |
| 345 | }; | |
| 346 | ||
| 20c6c071 GH |
347 | // The match-and-expand loop. |
| 348 | void | |
| 349 | symresolution_info::derive_probes (match_node * root, | |
| 350 | probe *p, vector<derived_probe*>& dps) | |
| 351 | { | |
| 5227f1ea GH |
352 | static unsigned depth=0; |
| 353 | recursion_guard guard(depth); | |
| 354 | ||
| 20c6c071 GH |
355 | for (unsigned i = 0; i < p->locations.size(); ++i) |
| 356 | { | |
| 357 | probe_point *loc = p->locations[i]; | |
| 358 | vector< pair<string, literal *> > param_vec; | |
| 359 | map<string, literal *> param_map; | |
| 360 | vector<probe *> re_expand; | |
| 361 | ||
| fe3d01fa FCE |
362 | try |
| 363 | { | |
| 364 | derived_probe_builder * builder = | |
| 365 | root->find_builder(loc->components, 0, param_vec); | |
| 366 | ||
| 367 | if (!builder) | |
| 368 | throw semantic_error ("no match for probe point family"); | |
| 369 | ||
| 370 | param_vec_to_map(param_vec, param_map); | |
| 371 | ||
| 372 | unsigned num_atbegin = dps.size(); | |
| 373 | builder->build(session, p, loc, param_map, re_expand, dps); | |
| 374 | ||
| 375 | // Recursively expand any further-expanding results | |
| 376 | if (!re_expand.empty()) | |
| 377 | { | |
| 378 | for (unsigned j = 0; j < re_expand.size(); ++j) | |
| 379 | derive_probes(root, re_expand[j], dps); | |
| 380 | } | |
| 381 | ||
| 382 | unsigned num_atend = dps.size(); | |
| 383 | if (num_atbegin == num_atend) // nothing new derived! | |
| 384 | throw semantic_error ("no match for probe point"); | |
| 385 | } | |
| 386 | catch (const semantic_error& e) | |
| 387 | { | |
| 388 | session.print_error (e); | |
| 389 | cerr << " while: resolving probe point " << *loc << endl; | |
| 390 | } | |
| 20c6c071 GH |
391 | } |
| 392 | } | |
| 393 | ||
| 394 | // ------------------------------------------------------------------------ | |
| 67c0a579 GH |
395 | // |
| 396 | // Map usage checks | |
| 397 | // | |
| 398 | ||
| 399 | class lvalue_aware_traversing_visitor | |
| 400 | : public traversing_visitor | |
| 401 | { | |
| 402 | unsigned lval_depth; | |
| 403 | public: | |
| 404 | ||
| 405 | lvalue_aware_traversing_visitor() : lval_depth(0) {} | |
| 406 | ||
| 407 | bool is_in_lvalue() | |
| 408 | { | |
| 409 | return lval_depth > 0; | |
| 410 | } | |
| 411 | ||
| 412 | virtual void visit_pre_crement (pre_crement* e) | |
| 413 | { | |
| 414 | ++lval_depth; | |
| 415 | e->operand->visit (this); | |
| 416 | --lval_depth; | |
| 417 | } | |
| 418 | ||
| 419 | virtual void visit_post_crement (post_crement* e) | |
| 420 | { | |
| 421 | ++lval_depth; | |
| 422 | e->operand->visit (this); | |
| 423 | --lval_depth; | |
| 424 | } | |
| 425 | ||
| 426 | virtual void visit_assignment (assignment* e) | |
| 427 | { | |
| 428 | ++lval_depth; | |
| 429 | e->left->visit (this); | |
| 430 | --lval_depth; | |
| 431 | e->right->visit (this); | |
| 432 | } | |
| 433 | ||
| 434 | virtual void visit_delete_statement (delete_statement* s) | |
| 435 | { | |
| 436 | ++lval_depth; | |
| 437 | s->value->visit (this); | |
| 438 | --lval_depth; | |
| 439 | } | |
| 440 | ||
| 441 | }; | |
| 442 | ||
| 443 | ||
| 444 | struct mutated_map_collector | |
| 445 | : public lvalue_aware_traversing_visitor | |
| 446 | { | |
| 447 | set<vardecl *> * mutated_maps; | |
| 448 | ||
| 449 | mutated_map_collector(set<vardecl *> * mm) | |
| 450 | : mutated_maps (mm) | |
| 451 | {} | |
| 452 | ||
| 453 | void visit_arrayindex (arrayindex *e) | |
| 454 | { | |
| 455 | if (is_in_lvalue()) | |
| 456 | mutated_maps->insert(e->referent); | |
| 457 | } | |
| 458 | }; | |
| 459 | ||
| 460 | ||
| 461 | struct no_map_mutation_during_iteration_check | |
| 462 | : public lvalue_aware_traversing_visitor | |
| 463 | { | |
| 464 | systemtap_session & session; | |
| 465 | map<functiondecl *,set<vardecl *> *> & function_mutates_maps; | |
| 466 | vector<vardecl *> maps_being_iterated; | |
| 467 | ||
| 468 | no_map_mutation_during_iteration_check | |
| 469 | (systemtap_session & sess, | |
| 470 | map<functiondecl *,set<vardecl *> *> & fmm) | |
| 471 | : session(sess), function_mutates_maps (fmm) | |
| 472 | {} | |
| 473 | ||
| 474 | void visit_arrayindex (arrayindex *e) | |
| 475 | { | |
| 476 | if (is_in_lvalue()) | |
| 477 | { | |
| 478 | for (unsigned i = 0; i < maps_being_iterated.size(); ++i) | |
| 479 | { | |
| 480 | vardecl *m = maps_being_iterated[i]; | |
| 481 | if (m == e->referent) | |
| 482 | { | |
| 483 | string err = ("map '" + m->name + | |
| 484 | "' modified during 'foreach' iteration"); | |
| 485 | session.print_error (semantic_error (err, e->tok)); | |
| 486 | } | |
| 487 | } | |
| 488 | } | |
| 489 | } | |
| 490 | ||
| 491 | void visit_functioncall (functioncall* e) | |
| 492 | { | |
| 493 | map<functiondecl *,set<vardecl *> *>::const_iterator i | |
| 494 | = function_mutates_maps.find (e->referent); | |
| 495 | ||
| 496 | if (i != function_mutates_maps.end()) | |
| 497 | { | |
| 498 | for (unsigned j = 0; j < maps_being_iterated.size(); ++j) | |
| 499 | { | |
| 500 | vardecl *m = maps_being_iterated[j]; | |
| 501 | if (i->second->find (m) != i->second->end()) | |
| 502 | { | |
| 503 | string err = ("function call modifies map '" + m->name + | |
| 504 | "' during 'foreach' iteration"); | |
| 505 | session.print_error (semantic_error (err, e->tok)); | |
| 506 | } | |
| 507 | } | |
| 508 | } | |
| 509 | ||
| 510 | for (unsigned i=0; i<e->args.size(); i++) | |
| 511 | e->args[i]->visit (this); | |
| 512 | } | |
| 513 | ||
| 514 | void visit_foreach_loop(foreach_loop* s) | |
| 515 | { | |
| 516 | maps_being_iterated.push_back (s->base_referent); | |
| 517 | for (unsigned i=0; i<s->indexes.size(); i++) | |
| 518 | s->indexes[i]->visit (this); | |
| 519 | s->block->visit (this); | |
| 520 | maps_being_iterated.pop_back(); | |
| 521 | } | |
| 522 | }; | |
| 20c6c071 | 523 | |
| 2b066ec1 | 524 | |
| 67c0a579 GH |
525 | static int |
| 526 | semantic_pass_maps (systemtap_session & sess) | |
| 527 | { | |
| 528 | ||
| 529 | map<functiondecl *, set<vardecl *> *> fmm; | |
| 530 | no_map_mutation_during_iteration_check chk(sess, fmm); | |
| 531 | ||
| 532 | for (unsigned i = 0; i < sess.functions.size(); ++i) | |
| 533 | { | |
| 534 | functiondecl * fn = sess.functions[i]; | |
| 535 | if (fn->body) | |
| 536 | { | |
| 537 | set<vardecl *> * m = new set<vardecl *>(); | |
| 538 | mutated_map_collector mc (m); | |
| 539 | fn->body->visit (&mc); | |
| 540 | fmm[fn] = m; | |
| 541 | } | |
| 542 | } | |
| 543 | ||
| 544 | for (unsigned i = 0; i < sess.functions.size(); ++i) | |
| 545 | { | |
| 546 | if (sess.functions[i]->body) | |
| 547 | sess.functions[i]->body->visit (&chk); | |
| 548 | } | |
| 549 | ||
| 550 | for (unsigned i = 0; i < sess.probes.size(); ++i) | |
| 551 | { | |
| 552 | if (sess.probes[i]->body) | |
| 553 | sess.probes[i]->body->visit (&chk); | |
| 554 | } | |
| 555 | ||
| 556 | return sess.num_errors; | |
| 557 | } | |
| 558 | ||
| 559 | // ------------------------------------------------------------------------ | |
| 560 | ||
| 2b066ec1 FCE |
561 | |
| 562 | static int semantic_pass_symbols (systemtap_session&); | |
| 563 | static int semantic_pass_types (systemtap_session&); | |
| 67c0a579 | 564 | static int semantic_pass_maps (systemtap_session&); |
| 2b066ec1 FCE |
565 | |
| 566 | ||
| 567 | ||
| 568 | // Link up symbols to their declarations. Set the session's | |
| 569 | // files/probes/functions/globals vectors from the transitively | |
| 570 | // reached set of stapfiles in s.library_files, starting from | |
| 20c6c071 | 571 | // s.user_file. Perform automatic tapset inclusion and probe |
| 2b066ec1 FCE |
572 | // alias expansion. |
| 573 | static int | |
| 574 | semantic_pass_symbols (systemtap_session& s) | |
| 575 | { | |
| 576 | symresolution_info sym (s); | |
| 577 | ||
| 578 | // NB: s.files can grow during this iteration, so size() can | |
| 579 | // return gradually increasing numbers. | |
| 580 | s.files.push_back (s.user_file); | |
| 581 | for (unsigned i = 0; i < s.files.size(); i++) | |
| 582 | { | |
| 583 | stapfile* dome = s.files[i]; | |
| 584 | ||
| 585 | // Pass 1: add globals and functions to systemtap-session master list, | |
| 586 | // so the find_* functions find them | |
| 587 | ||
| 588 | for (unsigned i=0; i<dome->globals.size(); i++) | |
| 589 | s.globals.push_back (dome->globals[i]); | |
| 590 | ||
| 591 | for (unsigned i=0; i<dome->functions.size(); i++) | |
| 592 | s.functions.push_back (dome->functions[i]); | |
| 593 | ||
| 54dfabe9 FCE |
594 | for (unsigned i=0; i<dome->embeds.size(); i++) |
| 595 | s.embeds.push_back (dome->embeds[i]); | |
| 596 | ||
| 2b066ec1 FCE |
597 | // Pass 2: process functions |
| 598 | ||
| 599 | for (unsigned i=0; i<dome->functions.size(); i++) | |
| 600 | { | |
| 601 | functiondecl* fd = dome->functions[i]; | |
| 602 | ||
| 603 | try | |
| 604 | { | |
| 8a43522c GH |
605 | sym.current_function = fd; |
| 606 | sym.current_probe = 0; | |
| 607 | fd->body->visit (& sym); | |
| 2b066ec1 FCE |
608 | } |
| 609 | catch (const semantic_error& e) | |
| 610 | { | |
| 611 | s.print_error (e); | |
| 612 | } | |
| 613 | } | |
| 614 | ||
| 5227f1ea GH |
615 | // Pass 3: derive probes and resolve any further symbols in the |
| 616 | // derived results. | |
| 2b066ec1 FCE |
617 | |
| 618 | for (unsigned i=0; i<dome->probes.size(); i++) | |
| 619 | { | |
| 620 | probe* p = dome->probes [i]; | |
| 621 | vector<derived_probe*> dps; | |
| 622 | ||
| fe3d01fa FCE |
623 | // much magic happens here: probe alias expansion, |
| 624 | // provider identification | |
| 625 | sym.derive_probes (s.pattern_root, p, dps); | |
| 2b066ec1 FCE |
626 | |
| 627 | for (unsigned j=0; j<dps.size(); j++) | |
| 628 | { | |
| 629 | derived_probe* dp = dps[j]; | |
| 630 | s.probes.push_back (dp); | |
| 631 | ||
| 632 | try | |
| 633 | { | |
| 634 | sym.current_function = 0; | |
| 5227f1ea | 635 | sym.current_probe = dp; |
| 2b066ec1 FCE |
636 | dp->body->visit (& sym); |
| 637 | } | |
| 638 | catch (const semantic_error& e) | |
| 639 | { | |
| 640 | s.print_error (e); | |
| 641 | } | |
| 642 | } | |
| 643 | } | |
| 644 | } | |
| fe3d01fa | 645 | |
| 2b066ec1 FCE |
646 | return s.num_errors; // all those print_error calls |
| 647 | } | |
| 648 | ||
| 649 | ||
| 650 | ||
| 651 | int | |
| 652 | semantic_pass (systemtap_session& s) | |
| 653 | { | |
| 20c6c071 | 654 | s.register_library_aliases(); |
| f8220a7b | 655 | register_standard_tapsets(s); |
| 20c6c071 | 656 | |
| 2b066ec1 FCE |
657 | int rc = semantic_pass_symbols (s); |
| 658 | if (rc == 0) rc = semantic_pass_types (s); | |
| 67c0a579 | 659 | if (rc == 0) rc = semantic_pass_maps (s); |
| 2b066ec1 FCE |
660 | return rc; |
| 661 | } | |
| 662 | ||
| 663 | ||
| 664 | // ------------------------------------------------------------------------ | |
| 665 | ||
| 666 | ||
| 667 | systemtap_session::systemtap_session (): | |
| 20c6c071 | 668 | pattern_root(new match_node), |
| 2b066ec1 FCE |
669 | user_file (0), op (0), up (0), num_errors (0) |
| 670 | { | |
| 671 | } | |
| 672 | ||
| 673 | ||
| 674 | void | |
| 675 | systemtap_session::print_error (const semantic_error& e) | |
| 676 | { | |
| d8067b24 FCE |
677 | cerr << "semantic error: " << e.what (); |
| 678 | if (e.tok1 || e.tok2) | |
| 679 | cerr << ": "; | |
| 2b066ec1 FCE |
680 | if (e.tok1) cerr << *e.tok1; |
| 681 | cerr << e.msg2; | |
| 682 | if (e.tok2) cerr << *e.tok2; | |
| 683 | cerr << endl; | |
| 684 | num_errors ++; | |
| 685 | } | |
| 686 | ||
| 687 | ||
| 688 | // ------------------------------------------------------------------------ | |
| 689 | // semantic processing: symbol resolution | |
| 690 | ||
| 691 | ||
| 692 | symresolution_info::symresolution_info (systemtap_session& s): | |
| 5227f1ea | 693 | session (s), current_function (0), current_probe (0) |
| 2b066ec1 FCE |
694 | { |
| 695 | } | |
| 696 | ||
| 697 | ||
| 698 | void | |
| 699 | symresolution_info::visit_block (block* e) | |
| 700 | { | |
| 701 | for (unsigned i=0; i<e->statements.size(); i++) | |
| 702 | { | |
| 703 | try | |
| 704 | { | |
| 705 | e->statements[i]->visit (this); | |
| 706 | } | |
| 707 | catch (const semantic_error& e) | |
| 708 | { | |
| 709 | session.print_error (e); | |
| 710 | } | |
| 711 | } | |
| 712 | } | |
| 713 | ||
| 714 | ||
| 69c68955 FCE |
715 | void |
| 716 | symresolution_info::visit_foreach_loop (foreach_loop* e) | |
| 717 | { | |
| 718 | for (unsigned i=0; i<e->indexes.size(); i++) | |
| 719 | e->indexes[i]->visit (this); | |
| 720 | ||
| 721 | if (e->base_referent) | |
| 722 | return; | |
| 723 | ||
| 313b2f74 | 724 | vardecl* d = find_var (e->base, e->indexes.size ()); |
| 69c68955 FCE |
725 | if (d) |
| 726 | e->base_referent = d; | |
| 727 | else | |
| 728 | throw semantic_error ("unresolved global array " + e->base, e->tok); | |
| 729 | ||
| 730 | e->block->visit (this); | |
| 731 | } | |
| 732 | ||
| d98d459c GH |
733 | struct |
| 734 | delete_statement_symresolution_info: | |
| 735 | public traversing_visitor | |
| 736 | { | |
| 737 | symresolution_info *parent; | |
| 738 | ||
| 739 | delete_statement_symresolution_info (symresolution_info *p): | |
| 740 | parent(p) | |
| 741 | {} | |
| 742 | ||
| 743 | void visit_arrayindex (arrayindex* e) | |
| 744 | { | |
| 745 | parent->visit_arrayindex (e); | |
| 746 | } | |
| 747 | void visit_functioncall (functioncall* e) | |
| 748 | { | |
| 749 | parent->visit_functioncall (e); | |
| 750 | } | |
| 751 | ||
| 752 | void visit_symbol (symbol* e) | |
| 753 | { | |
| 754 | if (e->referent) | |
| 755 | return; | |
| 756 | ||
| 757 | vardecl* d = parent->find_var (e->name, -1); | |
| 758 | if (d) | |
| 759 | e->referent = d; | |
| 760 | else | |
| 761 | throw semantic_error ("unresolved array in delete statement", e->tok); | |
| 762 | } | |
| 763 | }; | |
| 764 | ||
| 765 | void | |
| 766 | symresolution_info::visit_delete_statement (delete_statement* s) | |
| 767 | { | |
| 768 | delete_statement_symresolution_info di (this); | |
| 769 | s->value->visit (&di); | |
| 770 | } | |
| 771 | ||
| 69c68955 | 772 | |
| 2b066ec1 FCE |
773 | void |
| 774 | symresolution_info::visit_symbol (symbol* e) | |
| 775 | { | |
| 776 | if (e->referent) | |
| 777 | return; | |
| 778 | ||
| 313b2f74 | 779 | vardecl* d = find_var (e->name, 0); |
| 2b066ec1 FCE |
780 | if (d) |
| 781 | e->referent = d; | |
| 782 | else | |
| 783 | { | |
| 784 | // new local | |
| 785 | vardecl* v = new vardecl; | |
| 786 | v->name = e->name; | |
| 787 | v->tok = e->tok; | |
| 788 | if (current_function) | |
| 789 | current_function->locals.push_back (v); | |
| 790 | else if (current_probe) | |
| 791 | current_probe->locals.push_back (v); | |
| 792 | else | |
| ce10591c FCE |
793 | // must not happen |
| 794 | throw semantic_error ("no current probe/function", e->tok); | |
| 2b066ec1 FCE |
795 | e->referent = v; |
| 796 | } | |
| 797 | } | |
| 798 | ||
| 799 | ||
| 800 | void | |
| 801 | symresolution_info::visit_arrayindex (arrayindex* e) | |
| 802 | { | |
| 803 | for (unsigned i=0; i<e->indexes.size(); i++) | |
| 804 | e->indexes[i]->visit (this); | |
| 805 | ||
| 806 | if (e->referent) | |
| 807 | return; | |
| 808 | ||
| 313b2f74 | 809 | vardecl* d = find_var (e->base, e->indexes.size ()); |
| 2b066ec1 FCE |
810 | if (d) |
| 811 | e->referent = d; | |
| 812 | else | |
| 313b2f74 GH |
813 | { |
| 814 | // new local | |
| 815 | vardecl* v = new vardecl; | |
| 816 | v->set_arity(e->indexes.size()); | |
| 817 | v->name = e->base; | |
| 818 | v->tok = e->tok; | |
| 819 | if (current_function) | |
| 820 | current_function->locals.push_back (v); | |
| 821 | else if (current_probe) | |
| 822 | current_probe->locals.push_back (v); | |
| 823 | else | |
| 824 | // must not happen | |
| 825 | throw semantic_error ("no current probe/function", e->tok); | |
| 826 | e->referent = v; | |
| 827 | } | |
| 2b066ec1 FCE |
828 | } |
| 829 | ||
| 830 | ||
| 831 | void | |
| 832 | symresolution_info::visit_functioncall (functioncall* e) | |
| 833 | { | |
| 834 | for (unsigned i=0; i<e->args.size(); i++) | |
| 835 | e->args[i]->visit (this); | |
| 836 | ||
| 837 | if (e->referent) | |
| 838 | return; | |
| 839 | ||
| 840 | functiondecl* d = find_function (e->function, e->args.size ()); | |
| 841 | if (d) | |
| 842 | e->referent = d; | |
| 843 | else | |
| 844 | throw semantic_error ("unresolved function call", e->tok); | |
| 845 | } | |
| 846 | ||
| 847 | ||
| 848 | vardecl* | |
| d98d459c | 849 | symresolution_info::find_var (const string& name, int arity) |
| 2b066ec1 | 850 | { |
| 313b2f74 | 851 | |
| 2b066ec1 FCE |
852 | // search locals |
| 853 | vector<vardecl*>& locals = (current_function ? | |
| 854 | current_function->locals : | |
| 5227f1ea | 855 | current_probe->locals); |
| 6deb8173 | 856 | |
| 313b2f74 | 857 | |
| 2b066ec1 | 858 | for (unsigned i=0; i<locals.size(); i++) |
| 313b2f74 GH |
859 | if (locals[i]->name == name |
| 860 | && locals[i]->compatible_arity(arity)) | |
| 8846477c | 861 | { |
| 313b2f74 | 862 | locals[i]->set_arity (arity); |
| 8846477c FCE |
863 | return locals[i]; |
| 864 | } | |
| 2b066ec1 | 865 | |
| 313b2f74 GH |
866 | // search function formal parameters (for scalars) |
| 867 | if (arity == 0 && current_function) | |
| 2b066ec1 FCE |
868 | for (unsigned i=0; i<current_function->formal_args.size(); i++) |
| 869 | if (current_function->formal_args[i]->name == name) | |
| 8846477c FCE |
870 | { |
| 871 | // NB: no need to check arity here: formal args always scalar | |
| 872 | current_function->formal_args[i]->set_arity (0); | |
| 873 | return current_function->formal_args[i]; | |
| 874 | } | |
| 2b066ec1 | 875 | |
| 313b2f74 | 876 | // search processed globals |
| 2b066ec1 | 877 | for (unsigned i=0; i<session.globals.size(); i++) |
| 313b2f74 | 878 | if (session.globals[i]->name == name |
| d98d459c | 879 | && session.globals[i]->compatible_arity(arity)) |
| 8846477c | 880 | { |
| 313b2f74 | 881 | session.globals[i]->set_arity (arity); |
| 8846477c FCE |
882 | return session.globals[i]; |
| 883 | } | |
| d98d459c | 884 | |
| 2b066ec1 FCE |
885 | // search library globals |
| 886 | for (unsigned i=0; i<session.library_files.size(); i++) | |
| 887 | { | |
| 888 | stapfile* f = session.library_files[i]; | |
| 889 | for (unsigned j=0; j<f->globals.size(); j++) | |
| 84e5ea0f FCE |
890 | { |
| 891 | vardecl* g = f->globals[j]; | |
| 892 | if (g->name == name && g->compatible_arity (arity)) | |
| 893 | { | |
| d98d459c | 894 | g->set_arity (arity); |
| 84e5ea0f FCE |
895 | |
| 896 | // put library into the queue if not already there | |
| 897 | if (find (session.files.begin(), session.files.end(), f) | |
| 898 | == session.files.end()) | |
| 899 | session.files.push_back (f); | |
| 900 | ||
| 901 | return g; | |
| 902 | } | |
| 903 | } | |
| 2b066ec1 FCE |
904 | } |
| 905 | ||
| 906 | // search builtins that become locals | |
| 907 | // XXX: need to invent a proper formalism for this | |
| 313b2f74 | 908 | if (arity == 0 && (name == "$pid" || name == "$tid")) |
| 2b066ec1 FCE |
909 | { |
| 910 | vardecl_builtin* vb = new vardecl_builtin; | |
| 911 | vb->name = name; | |
| 912 | vb->type = pe_long; | |
| 913 | ||
| 914 | // XXX: need a better way to synthesize tokens | |
| 915 | token* t = new token; | |
| 916 | t->type = tok_identifier; | |
| 917 | t->content = name; | |
| 918 | t->location.file = "<builtin>"; | |
| 919 | vb->tok = t; | |
| 920 | ||
| 921 | locals.push_back (vb); | |
| 922 | return vb; | |
| 923 | } | |
| 924 | ||
| 925 | return 0; | |
| 2b066ec1 FCE |
926 | } |
| 927 | ||
| 928 | ||
| 2b066ec1 FCE |
929 | functiondecl* |
| 930 | symresolution_info::find_function (const string& name, unsigned arity) | |
| 931 | { | |
| 932 | for (unsigned j = 0; j < session.functions.size(); j++) | |
| 933 | { | |
| 934 | functiondecl* fd = session.functions[j]; | |
| 935 | if (fd->name == name && | |
| 936 | fd->formal_args.size() == arity) | |
| 937 | return fd; | |
| 938 | } | |
| 939 | ||
| 940 | // search library globals | |
| 941 | for (unsigned i=0; i<session.library_files.size(); i++) | |
| 942 | { | |
| 943 | stapfile* f = session.library_files[i]; | |
| 944 | for (unsigned j=0; j<f->functions.size(); j++) | |
| 945 | if (f->functions[j]->name == name && | |
| 946 | f->functions[j]->formal_args.size() == arity) | |
| 947 | { | |
| 948 | // put library into the queue if not already there | |
| 949 | if (0) // session.verbose_resolution | |
| 950 | cerr << " function " << name << " " | |
| 951 | << "is defined from " << f->name << endl; | |
| 952 | ||
| 953 | if (find (session.files.begin(), session.files.end(), f) | |
| 954 | == session.files.end()) | |
| 955 | session.files.push_back (f); | |
| 956 | // else .. print different message? | |
| 957 | ||
| 958 | return f->functions[j]; | |
| 959 | } | |
| 960 | } | |
| 961 | ||
| 962 | return 0; | |
| 2b066ec1 FCE |
963 | } |
| 964 | ||
| 965 | ||
| 966 | // ------------------------------------------------------------------------ | |
| 967 | // type resolution | |
| 968 | ||
| 969 | ||
| 970 | static int | |
| 971 | semantic_pass_types (systemtap_session& s) | |
| 972 | { | |
| 973 | int rc = 0; | |
| 974 | ||
| 975 | // next pass: type inference | |
| 976 | unsigned iterations = 0; | |
| 977 | typeresolution_info ti (s); | |
| 978 | ||
| 979 | ti.assert_resolvability = false; | |
| 980 | // XXX: maybe convert to exception-based error signalling | |
| 981 | while (1) | |
| 982 | { | |
| 983 | iterations ++; | |
| 984 | // cerr << "Type resolution, iteration " << iterations << endl; | |
| 985 | ti.num_newly_resolved = 0; | |
| 986 | ti.num_still_unresolved = 0; | |
| 987 | ||
| 988 | for (unsigned j=0; j<s.functions.size(); j++) | |
| 989 | { | |
| 990 | functiondecl* fn = s.functions[j]; | |
| 8a43522c GH |
991 | ti.current_function = fn; |
| 992 | ti.t = pe_unknown; | |
| 993 | fn->body->visit (& ti); | |
| 994 | // NB: we don't have to assert a known type for | |
| 995 | // functions here, to permit a "void" function. | |
| 996 | // The translator phase will omit the "retvalue". | |
| 997 | // | |
| 998 | // if (fn->type == pe_unknown) | |
| 999 | // ti.unresolved (fn->tok); | |
| 2b066ec1 FCE |
1000 | } |
| 1001 | ||
| 1002 | for (unsigned j=0; j<s.probes.size(); j++) | |
| 1003 | { | |
| 1004 | derived_probe* pn = s.probes[j]; | |
| 1005 | ti.current_function = 0; | |
| 1006 | ti.t = pe_unknown; | |
| 1007 | pn->body->visit (& ti); | |
| 1008 | } | |
| 1009 | ||
| 1010 | for (unsigned j=0; j<s.globals.size(); j++) | |
| 1011 | { | |
| 1012 | vardecl* gd = s.globals[j]; | |
| 1013 | if (gd->type == pe_unknown) | |
| 1014 | ti.unresolved (gd->tok); | |
| 1015 | } | |
| 1016 | ||
| 1017 | if (ti.num_newly_resolved == 0) // converged | |
| 1018 | if (ti.num_still_unresolved == 0) | |
| 1019 | break; // successfully | |
| 1020 | else if (! ti.assert_resolvability) | |
| 1021 | ti.assert_resolvability = true; // last pass, with error msgs | |
| 1022 | else | |
| 1023 | { // unsuccessful conclusion | |
| 1024 | rc ++; | |
| 1025 | break; | |
| 1026 | } | |
| 1027 | } | |
| 1028 | ||
| 1029 | return rc + s.num_errors; | |
| 1030 | } | |
| 1031 | ||
| 1032 | ||
| 1033 | void | |
| 1034 | typeresolution_info::visit_literal_number (literal_number* e) | |
| 1035 | { | |
| 1036 | assert (e->type == pe_long); | |
| 1037 | if ((t == e->type) || (t == pe_unknown)) | |
| 1038 | return; | |
| 1039 | ||
| 1040 | mismatch (e->tok, e->type, t); | |
| 1041 | } | |
| 1042 | ||
| 1043 | ||
| 1044 | void | |
| 1045 | typeresolution_info::visit_literal_string (literal_string* e) | |
| 1046 | { | |
| 1047 | assert (e->type == pe_string); | |
| 1048 | if ((t == e->type) || (t == pe_unknown)) | |
| 1049 | return; | |
| 1050 | ||
| 1051 | mismatch (e->tok, e->type, t); | |
| 1052 | } | |
| 1053 | ||
| 1054 | ||
| 1055 | void | |
| 1056 | typeresolution_info::visit_logical_or_expr (logical_or_expr *e) | |
| 1057 | { | |
| 1058 | visit_binary_expression (e); | |
| 1059 | } | |
| 1060 | ||
| 1061 | ||
| 1062 | void | |
| 1063 | typeresolution_info::visit_logical_and_expr (logical_and_expr *e) | |
| 1064 | { | |
| 1065 | visit_binary_expression (e); | |
| 1066 | } | |
| 1067 | ||
| 1068 | ||
| 1069 | void | |
| 1070 | typeresolution_info::visit_comparison (comparison *e) | |
| 1071 | { | |
| d5d7c2cc | 1072 | // NB: result of any comparison is an integer! |
| 553d27a5 FCE |
1073 | if (t == pe_stats || t == pe_string) |
| 1074 | invalid (e->tok, t); | |
| 1075 | ||
| 1076 | t = (e->right->type != pe_unknown) ? e->right->type : pe_unknown; | |
| 1077 | e->left->visit (this); | |
| 1078 | t = (e->left->type != pe_unknown) ? e->left->type : pe_unknown; | |
| 1079 | e->right->visit (this); | |
| 1080 | ||
| 1081 | if (e->left->type != pe_unknown && | |
| 1082 | e->right->type != pe_unknown && | |
| 1083 | e->left->type != e->right->type) | |
| 1084 | mismatch (e->tok, e->left->type, e->right->type); | |
| 1085 | ||
| 1086 | if (e->type == pe_unknown) | |
| 1087 | { | |
| 1088 | e->type = pe_long; | |
| 1089 | resolved (e->tok, e->type); | |
| 1090 | } | |
| 2b066ec1 FCE |
1091 | } |
| 1092 | ||
| 1093 | ||
| 1094 | void | |
| 1095 | typeresolution_info::visit_concatenation (concatenation *e) | |
| 1096 | { | |
| 553d27a5 FCE |
1097 | if (t != pe_unknown && t != pe_string) |
| 1098 | invalid (e->tok, t); | |
| 1099 | ||
| 1100 | t = pe_string; | |
| 1101 | e->left->visit (this); | |
| 1102 | t = pe_string; | |
| 1103 | e->right->visit (this); | |
| 1104 | ||
| 1105 | if (e->type == pe_unknown) | |
| 1106 | { | |
| 1107 | e->type = pe_string; | |
| 1108 | resolved (e->tok, e->type); | |
| 1109 | } | |
| 2b066ec1 FCE |
1110 | } |
| 1111 | ||
| 1112 | ||
| 2b066ec1 FCE |
1113 | void |
| 1114 | typeresolution_info::visit_assignment (assignment *e) | |
| 1115 | { | |
| 553d27a5 FCE |
1116 | if (t == pe_stats) |
| 1117 | invalid (e->tok, t); | |
| 2b066ec1 | 1118 | |
| 2b066ec1 FCE |
1119 | if (e->op == "<<<") // stats aggregation |
| 1120 | { | |
| 553d27a5 FCE |
1121 | if (t == pe_string) |
| 1122 | invalid (e->tok, t); | |
| 1123 | ||
| 2b066ec1 FCE |
1124 | t = pe_stats; |
| 1125 | e->left->visit (this); | |
| 1126 | t = pe_long; | |
| 1127 | e->right->visit (this); | |
| 553d27a5 | 1128 | if (e->type == pe_unknown) |
| 2b066ec1 FCE |
1129 | { |
| 1130 | e->type = pe_long; | |
| 1131 | resolved (e->tok, e->type); | |
| 1132 | } | |
| 1133 | } | |
| 553d27a5 | 1134 | else if (e->op == "+=" || // numeric only |
| d5d7c2cc FCE |
1135 | e->op == "-=" || |
| 1136 | e->op == "*=" || | |
| 1137 | e->op == "/=" || | |
| 1138 | e->op == "%=" || | |
| 1139 | e->op == "&=" || | |
| 1140 | e->op == "^=" || | |
| 1141 | e->op == "|=" || | |
| 1142 | e->op == "<<=" || | |
| 1143 | e->op == ">>=" || | |
| 553d27a5 | 1144 | false) |
| 2b066ec1 | 1145 | { |
| 553d27a5 | 1146 | visit_binary_expression (e); |
| 2b066ec1 | 1147 | } |
| d5d7c2cc FCE |
1148 | else if (e->op == ".=" || // string only |
| 1149 | false) | |
| 1150 | { | |
| 1151 | if (t == pe_long || t == pe_stats) | |
| 1152 | invalid (e->tok, t); | |
| 1153 | ||
| 1154 | t = pe_string; | |
| 1155 | e->left->visit (this); | |
| 1156 | t = pe_string; | |
| 1157 | e->right->visit (this); | |
| 1158 | if (e->type == pe_unknown) | |
| 1159 | { | |
| 1160 | e->type = pe_string; | |
| 1161 | resolved (e->tok, e->type); | |
| 1162 | } | |
| 1163 | } | |
| 1164 | else if (e->op == "=") // overloaded = for string & numeric operands | |
| 2b066ec1 | 1165 | { |
| 553d27a5 | 1166 | // logic similar to ternary_expression |
| 2b066ec1 | 1167 | exp_type sub_type = t; |
| 553d27a5 FCE |
1168 | |
| 1169 | // Infer types across the l/r values | |
| 1170 | if (sub_type == pe_unknown && e->type != pe_unknown) | |
| 2b066ec1 | 1171 | sub_type = e->type; |
| 553d27a5 FCE |
1172 | |
| 1173 | t = (sub_type != pe_unknown) ? sub_type : | |
| 1174 | (e->right->type != pe_unknown) ? e->right->type : | |
| 1175 | pe_unknown; | |
| 2b066ec1 | 1176 | e->left->visit (this); |
| 553d27a5 FCE |
1177 | t = (sub_type != pe_unknown) ? sub_type : |
| 1178 | (e->left->type != pe_unknown) ? e->left->type : | |
| 1179 | pe_unknown; | |
| 2b066ec1 | 1180 | e->right->visit (this); |
| 553d27a5 FCE |
1181 | |
| 1182 | if ((sub_type != pe_unknown) && (e->type == pe_unknown)) | |
| 2b066ec1 FCE |
1183 | { |
| 1184 | e->type = sub_type; | |
| 1185 | resolved (e->tok, e->type); | |
| 1186 | } | |
| 553d27a5 | 1187 | if ((sub_type == pe_unknown) && (e->left->type != pe_unknown)) |
| 2b066ec1 FCE |
1188 | { |
| 1189 | e->type = e->left->type; | |
| 1190 | resolved (e->tok, e->type); | |
| 1191 | } | |
| 553d27a5 FCE |
1192 | |
| 1193 | if (e->left->type != pe_unknown && | |
| 1194 | e->right->type != pe_unknown && | |
| 1195 | e->left->type != e->right->type) | |
| 1196 | mismatch (e->tok, e->left->type, e->right->type); | |
| 1197 | } | |
| d5d7c2cc FCE |
1198 | else |
| 1199 | throw semantic_error ("unsupported assignment operator " + e->op); | |
| 553d27a5 FCE |
1200 | } |
| 1201 | ||
| 1202 | ||
| 1203 | void | |
| 1204 | typeresolution_info::visit_binary_expression (binary_expression* e) | |
| 1205 | { | |
| 1206 | if (t == pe_stats || t == pe_string) | |
| 1207 | invalid (e->tok, t); | |
| 1208 | ||
| 1209 | t = pe_long; | |
| 1210 | e->left->visit (this); | |
| 1211 | t = pe_long; | |
| 1212 | e->right->visit (this); | |
| 1213 | ||
| 1214 | if (e->left->type != pe_unknown && | |
| 1215 | e->right->type != pe_unknown && | |
| 1216 | e->left->type != e->right->type) | |
| 1217 | mismatch (e->tok, e->left->type, e->right->type); | |
| 1218 | ||
| 1219 | if (e->type == pe_unknown) | |
| 1220 | { | |
| 1221 | e->type = pe_long; | |
| 1222 | resolved (e->tok, e->type); | |
| 2b066ec1 FCE |
1223 | } |
| 1224 | } | |
| 1225 | ||
| 1226 | ||
| 1227 | void | |
| 1228 | typeresolution_info::visit_pre_crement (pre_crement *e) | |
| 1229 | { | |
| 1230 | visit_unary_expression (e); | |
| 1231 | } | |
| 1232 | ||
| 1233 | ||
| 1234 | void | |
| 1235 | typeresolution_info::visit_post_crement (post_crement *e) | |
| 1236 | { | |
| 1237 | visit_unary_expression (e); | |
| 1238 | } | |
| 1239 | ||
| 1240 | ||
| 1241 | void | |
| 1242 | typeresolution_info::visit_unary_expression (unary_expression* e) | |
| 1243 | { | |
| 553d27a5 FCE |
1244 | if (t == pe_stats || t == pe_string) |
| 1245 | invalid (e->tok, t); | |
| 1246 | ||
| 2b066ec1 FCE |
1247 | t = pe_long; |
| 1248 | e->operand->visit (this); | |
| 1249 | ||
| 553d27a5 | 1250 | if (e->type == pe_unknown) |
| 2b066ec1 FCE |
1251 | { |
| 1252 | e->type = pe_long; | |
| 1253 | resolved (e->tok, e->type); | |
| 1254 | } | |
| 1255 | } | |
| 1256 | ||
| 1257 | ||
| 2b066ec1 FCE |
1258 | void |
| 1259 | typeresolution_info::visit_ternary_expression (ternary_expression* e) | |
| 1260 | { | |
| 1261 | exp_type sub_type = t; | |
| 1262 | ||
| 1263 | t = pe_long; | |
| 1264 | e->cond->visit (this); | |
| 1265 | ||
| 553d27a5 | 1266 | // Infer types across the true/false arms of the ternary expression. |
| 2b066ec1 FCE |
1267 | |
| 1268 | if (sub_type == pe_unknown && e->type != pe_unknown) | |
| 1269 | sub_type = e->type; | |
| 1270 | t = sub_type; | |
| 1271 | e->truevalue->visit (this); | |
| 1272 | t = sub_type; | |
| 1273 | e->falsevalue->visit (this); | |
| 1274 | ||
| 1275 | if ((sub_type == pe_unknown) && (e->type != pe_unknown)) | |
| 1276 | ; // already resolved | |
| 1277 | else if ((sub_type != pe_unknown) && (e->type == pe_unknown)) | |
| 1278 | { | |
| 1279 | e->type = sub_type; | |
| 1280 | resolved (e->tok, e->type); | |
| 1281 | } | |
| 1282 | else if ((sub_type == pe_unknown) && (e->truevalue->type != pe_unknown)) | |
| 1283 | { | |
| 1284 | e->type = e->truevalue->type; | |
| 1285 | resolved (e->tok, e->type); | |
| 1286 | } | |
| 1287 | else if ((sub_type == pe_unknown) && (e->falsevalue->type != pe_unknown)) | |
| 1288 | { | |
| 1289 | e->type = e->falsevalue->type; | |
| 1290 | resolved (e->tok, e->type); | |
| 1291 | } | |
| 1292 | else if (e->type != sub_type) | |
| 1293 | mismatch (e->tok, sub_type, e->type); | |
| 1294 | } | |
| 1295 | ||
| 1296 | ||
| 1297 | template <class Referrer, class Referent> | |
| 1298 | void resolve_2types (Referrer* referrer, Referent* referent, | |
| 1299 | typeresolution_info* r, exp_type t) | |
| 1300 | { | |
| 1301 | exp_type& re_type = referrer->type; | |
| 1302 | const token* re_tok = referrer->tok; | |
| 1303 | exp_type& te_type = referent->type; | |
| 1304 | const token* te_tok = referent->tok; | |
| 1305 | ||
| 1306 | if (t != pe_unknown && re_type == t && re_type == te_type) | |
| 1307 | ; // do nothing: all three e->types in agreement | |
| 1308 | else if (t == pe_unknown && re_type != pe_unknown && re_type == te_type) | |
| 1309 | ; // do nothing: two known e->types in agreement | |
| 1310 | else if (re_type != pe_unknown && te_type != pe_unknown && re_type != te_type) | |
| 1311 | r->mismatch (re_tok, re_type, te_type); | |
| 1312 | else if (re_type != pe_unknown && t != pe_unknown && re_type != t) | |
| 1313 | r->mismatch (re_tok, re_type, t); | |
| 1314 | else if (te_type != pe_unknown && t != pe_unknown && te_type != t) | |
| 1315 | r->mismatch (te_tok, te_type, t); | |
| 1316 | else if (re_type == pe_unknown && t != pe_unknown) | |
| 1317 | { | |
| 1318 | // propagate from upstream | |
| 1319 | re_type = t; | |
| 1320 | r->resolved (re_tok, re_type); | |
| 1321 | // catch re_type/te_type mismatch later | |
| 1322 | } | |
| 1323 | else if (re_type == pe_unknown && te_type != pe_unknown) | |
| 1324 | { | |
| 1325 | // propagate from referent | |
| 1326 | re_type = te_type; | |
| 1327 | r->resolved (re_tok, re_type); | |
| 1328 | // catch re_type/t mismatch later | |
| 1329 | } | |
| 1330 | else if (re_type != pe_unknown && te_type == pe_unknown) | |
| 1331 | { | |
| 1332 | // propagate to referent | |
| 1333 | te_type = re_type; | |
| 1334 | r->resolved (te_tok, te_type); | |
| 1335 | // catch re_type/t mismatch later | |
| 1336 | } | |
| 1337 | else | |
| 1338 | r->unresolved (re_tok); | |
| 1339 | } | |
| 1340 | ||
| 1341 | ||
| 1342 | void | |
| 1343 | typeresolution_info::visit_symbol (symbol* e) | |
| 1344 | { | |
| 1345 | assert (e->referent != 0); | |
| 1346 | ||
| 1347 | if (e->referent->arity > 0) | |
| 1348 | unresolved (e->tok); // symbol resolution should not permit this | |
| 1349 | // XXX: but consider "delete <array>;" and similar constructs | |
| 1350 | else | |
| 1351 | resolve_2types (e, e->referent, this, t); | |
| 1352 | } | |
| 1353 | ||
| 1354 | ||
| 1355 | void | |
| 1356 | typeresolution_info::visit_arrayindex (arrayindex* e) | |
| 1357 | { | |
| 1358 | assert (e->referent != 0); | |
| 1359 | ||
| 1360 | resolve_2types (e, e->referent, this, t); | |
| 1361 | ||
| 1362 | // now resolve the array indexes | |
| 69c68955 FCE |
1363 | |
| 1364 | // if (e->referent->index_types.size() == 0) | |
| 1365 | // // redesignate referent as array | |
| 1366 | // e->referent->set_arity (e->indexes.size ()); | |
| 2b066ec1 FCE |
1367 | |
| 1368 | if (e->indexes.size() != e->referent->index_types.size()) | |
| 1369 | unresolved (e->tok); // symbol resolution should prevent this | |
| 1370 | else for (unsigned i=0; i<e->indexes.size(); i++) | |
| 1371 | { | |
| 1372 | expression* ee = e->indexes[i]; | |
| 1373 | exp_type& ft = e->referent->index_types [i]; | |
| 1374 | t = ft; | |
| 1375 | ee->visit (this); | |
| 1376 | exp_type at = ee->type; | |
| 1377 | ||
| 1378 | if ((at == pe_string || at == pe_long) && ft == pe_unknown) | |
| 1379 | { | |
| 1380 | // propagate to formal type | |
| 1381 | ft = at; | |
| 1382 | resolved (e->referent->tok, ft); | |
| 1383 | // uses array decl as there is no token for "formal type" | |
| 1384 | } | |
| 1385 | if (at == pe_stats) | |
| 1386 | invalid (ee->tok, at); | |
| 1387 | if (ft == pe_stats) | |
| 1388 | invalid (ee->tok, ft); | |
| 1389 | if (at != pe_unknown && ft != pe_unknown && ft != at) | |
| 1390 | mismatch (e->tok, at, ft); | |
| 1391 | if (at == pe_unknown) | |
| 313b2f74 | 1392 | unresolved (ee->tok); |
| 2b066ec1 FCE |
1393 | } |
| 1394 | } | |
| 1395 | ||
| 1396 | ||
| 1397 | void | |
| 1398 | typeresolution_info::visit_functioncall (functioncall* e) | |
| 1399 | { | |
| 1400 | assert (e->referent != 0); | |
| 1401 | ||
| 1402 | resolve_2types (e, e->referent, this, t); | |
| 1403 | ||
| 1404 | if (e->type == pe_stats) | |
| 1405 | invalid (e->tok, e->type); | |
| 1406 | ||
| 1407 | // XXX: but what about functions that return no value, | |
| 1408 | // and are used only as an expression-statement for side effects? | |
| 1409 | ||
| 1410 | // now resolve the function parameters | |
| 1411 | if (e->args.size() != e->referent->formal_args.size()) | |
| 1412 | unresolved (e->tok); // symbol resolution should prevent this | |
| 1413 | else for (unsigned i=0; i<e->args.size(); i++) | |
| 1414 | { | |
| 1415 | expression* ee = e->args[i]; | |
| 1416 | exp_type& ft = e->referent->formal_args[i]->type; | |
| 1417 | const token* fe_tok = e->referent->formal_args[i]->tok; | |
| 1418 | t = ft; | |
| 1419 | ee->visit (this); | |
| 1420 | exp_type at = ee->type; | |
| 1421 | ||
| 1422 | if (((at == pe_string) || (at == pe_long)) && ft == pe_unknown) | |
| 1423 | { | |
| 1424 | // propagate to formal arg | |
| 1425 | ft = at; | |
| 1426 | resolved (e->referent->formal_args[i]->tok, ft); | |
| 1427 | } | |
| 1428 | if (at == pe_stats) | |
| 1429 | invalid (e->tok, at); | |
| 1430 | if (ft == pe_stats) | |
| 1431 | invalid (fe_tok, ft); | |
| 1432 | if (at != pe_unknown && ft != pe_unknown && ft != at) | |
| 1433 | mismatch (e->tok, at, ft); | |
| 1434 | if (at == pe_unknown) | |
| 1435 | unresolved (e->tok); | |
| 1436 | } | |
| 1437 | } | |
| 1438 | ||
| 1439 | ||
| 1440 | void | |
| 1441 | typeresolution_info::visit_block (block* e) | |
| 1442 | { | |
| 1443 | for (unsigned i=0; i<e->statements.size(); i++) | |
| 1444 | { | |
| 1445 | try | |
| 1446 | { | |
| 1447 | t = pe_unknown; | |
| 1448 | e->statements[i]->visit (this); | |
| 1449 | } | |
| 1450 | catch (const semantic_error& e) | |
| 1451 | { | |
| 1452 | session.print_error (e); | |
| 1453 | } | |
| 1454 | } | |
| 1455 | } | |
| 1456 | ||
| 1457 | ||
| 54dfabe9 FCE |
1458 | void |
| 1459 | typeresolution_info::visit_embeddedcode (embeddedcode* e) | |
| 1460 | { | |
| 1461 | } | |
| 1462 | ||
| 1463 | ||
| 2b066ec1 FCE |
1464 | void |
| 1465 | typeresolution_info::visit_if_statement (if_statement* e) | |
| 1466 | { | |
| 1467 | t = pe_long; | |
| 1468 | e->condition->visit (this); | |
| 1469 | ||
| 1470 | t = pe_unknown; | |
| 1471 | e->thenblock->visit (this); | |
| 1472 | ||
| 1473 | if (e->elseblock) | |
| 1474 | { | |
| 1475 | t = pe_unknown; | |
| 1476 | e->elseblock->visit (this); | |
| 1477 | } | |
| 1478 | } | |
| 1479 | ||
| 1480 | ||
| 1481 | void | |
| 1482 | typeresolution_info::visit_for_loop (for_loop* e) | |
| 1483 | { | |
| 1484 | t = pe_unknown; | |
| 1485 | e->init->visit (this); | |
| 1486 | t = pe_long; | |
| 1487 | e->cond->visit (this); | |
| 1488 | t = pe_unknown; | |
| 1489 | e->incr->visit (this); | |
| 1490 | t = pe_unknown; | |
| 1491 | e->block->visit (this); | |
| 1492 | } | |
| 1493 | ||
| 1494 | ||
| 69c68955 FCE |
1495 | void |
| 1496 | typeresolution_info::visit_foreach_loop (foreach_loop* e) | |
| 1497 | { | |
| 1498 | // See also visit_arrayindex. | |
| 1499 | // This is different in that, being a statement, we can't assign | |
| 1500 | // a type to the outer array, only propagate to/from the indexes | |
| 1501 | ||
| 1502 | // if (e->referent->index_types.size() == 0) | |
| 1503 | // // redesignate referent as array | |
| 1504 | // e->referent->set_arity (e->indexes.size ()); | |
| 1505 | ||
| 1506 | if (e->indexes.size() != e->base_referent->index_types.size()) | |
| 1507 | unresolved (e->tok); // symbol resolution should prevent this | |
| 1508 | else for (unsigned i=0; i<e->indexes.size(); i++) | |
| 1509 | { | |
| 1510 | expression* ee = e->indexes[i]; | |
| 1511 | exp_type& ft = e->base_referent->index_types [i]; | |
| 1512 | t = ft; | |
| 1513 | ee->visit (this); | |
| 1514 | exp_type at = ee->type; | |
| 1515 | ||
| 1516 | if ((at == pe_string || at == pe_long) && ft == pe_unknown) | |
| 1517 | { | |
| 1518 | // propagate to formal type | |
| 1519 | ft = at; | |
| 1520 | resolved (e->base_referent->tok, ft); | |
| 1521 | // uses array decl as there is no token for "formal type" | |
| 1522 | } | |
| 1523 | if (at == pe_stats) | |
| 1524 | invalid (ee->tok, at); | |
| 1525 | if (ft == pe_stats) | |
| 1526 | invalid (ee->tok, ft); | |
| 1527 | if (at != pe_unknown && ft != pe_unknown && ft != at) | |
| 1528 | mismatch (e->tok, at, ft); | |
| 1529 | if (at == pe_unknown) | |
| 1530 | unresolved (ee->tok); | |
| 1531 | } | |
| 1532 | ||
| 1533 | t = pe_unknown; | |
| 1534 | e->block->visit (this); | |
| 1535 | } | |
| 1536 | ||
| 1537 | ||
| 2b066ec1 FCE |
1538 | void |
| 1539 | typeresolution_info::visit_null_statement (null_statement* e) | |
| 1540 | { | |
| 1541 | } | |
| 1542 | ||
| 1543 | ||
| 1544 | void | |
| 1545 | typeresolution_info::visit_expr_statement (expr_statement* e) | |
| 1546 | { | |
| 1547 | t = pe_unknown; | |
| 1548 | e->value->visit (this); | |
| 1549 | } | |
| 1550 | ||
| 1551 | ||
| d98d459c GH |
1552 | struct delete_statement_typeresolution_info: |
| 1553 | public throwing_visitor | |
| 1554 | { | |
| 1555 | typeresolution_info *parent; | |
| 1556 | delete_statement_typeresolution_info (typeresolution_info *p): | |
| 1557 | throwing_visitor ("invalid operand of delete expression"), | |
| 1558 | parent (p) | |
| 1559 | {} | |
| 1560 | ||
| 1561 | void visit_arrayindex (arrayindex* e) | |
| 1562 | { | |
| 1563 | parent->visit_arrayindex (e); | |
| 1564 | } | |
| 1565 | ||
| 1566 | void visit_symbol (symbol* e) | |
| 1567 | { | |
| 1568 | exp_type ignored = pe_unknown; | |
| 1569 | assert (e->referent != 0); | |
| 1570 | resolve_2types (e, e->referent, parent, ignored); | |
| 1571 | } | |
| 1572 | }; | |
| 1573 | ||
| 1574 | ||
| 2b066ec1 FCE |
1575 | void |
| 1576 | typeresolution_info::visit_delete_statement (delete_statement* e) | |
| 1577 | { | |
| d98d459c GH |
1578 | delete_statement_typeresolution_info di (this); |
| 1579 | t = pe_unknown; | |
| 1580 | e->value->visit (&di); | |
| 2b066ec1 FCE |
1581 | } |
| 1582 | ||
| 1583 | ||
| f3c26ea5 FCE |
1584 | void |
| 1585 | typeresolution_info::visit_next_statement (next_statement* s) | |
| 1586 | { | |
| 1587 | } | |
| 1588 | ||
| 1589 | ||
| 1590 | void | |
| 1591 | typeresolution_info::visit_break_statement (break_statement* s) | |
| 1592 | { | |
| 1593 | } | |
| 1594 | ||
| 1595 | ||
| 1596 | void | |
| 1597 | typeresolution_info::visit_continue_statement (continue_statement* s) | |
| 1598 | { | |
| 1599 | } | |
| 1600 | ||
| 1601 | ||
| 2b066ec1 FCE |
1602 | void |
| 1603 | typeresolution_info::visit_array_in (array_in* e) | |
| 1604 | { | |
| ce10591c FCE |
1605 | // all unary operators only work on numerics |
| 1606 | exp_type t1 = t; | |
| 1607 | t = pe_unknown; // array value can be anything | |
| 1608 | e->operand->visit (this); | |
| 1609 | ||
| 1610 | if (t1 == pe_unknown && e->type != pe_unknown) | |
| 1611 | ; // already resolved | |
| 1612 | else if (t1 == pe_string || t1 == pe_stats) | |
| 1613 | mismatch (e->tok, t1, pe_long); | |
| 1614 | else if (e->type == pe_unknown) | |
| 1615 | { | |
| 1616 | e->type = pe_long; | |
| 1617 | resolved (e->tok, e->type); | |
| 1618 | } | |
| 2b066ec1 FCE |
1619 | } |
| 1620 | ||
| 1621 | ||
| 1622 | void | |
| 1623 | typeresolution_info::visit_return_statement (return_statement* e) | |
| 1624 | { | |
| 1625 | // This is like symbol, where the referent is | |
| 1626 | // the return value of the function. | |
| 1627 | ||
| 22f46231 | 1628 | // translation pass will print error |
| 2b066ec1 | 1629 | if (current_function == 0) |
| 22f46231 | 1630 | return; |
| 2b066ec1 FCE |
1631 | |
| 1632 | exp_type& e_type = current_function->type; | |
| 1633 | t = current_function->type; | |
| 1634 | e->value->visit (this); | |
| 1635 | ||
| 1636 | if (e_type != pe_unknown && e->value->type != pe_unknown | |
| 1637 | && e_type != e->value->type) | |
| 1638 | mismatch (current_function->tok, e_type, e->value->type); | |
| 1639 | if (e_type == pe_unknown && | |
| 1640 | (e->value->type == pe_long || e->value->type == pe_string)) | |
| 1641 | { | |
| 1642 | // propagate non-statistics from value | |
| 1643 | e_type = e->value->type; | |
| 1644 | resolved (current_function->tok, e->value->type); | |
| 1645 | } | |
| 1646 | if (e->value->type == pe_stats) | |
| 1647 | invalid (e->value->tok, e->value->type); | |
| 1648 | } | |
| 1649 | ||
| 1650 | ||
| 1651 | void | |
| 1652 | typeresolution_info::unresolved (const token* tok) | |
| 1653 | { | |
| 1654 | num_still_unresolved ++; | |
| 1655 | ||
| 1656 | if (assert_resolvability) | |
| 1657 | { | |
| 1658 | cerr << "error: unresolved type for "; | |
| 1659 | if (tok) | |
| 1660 | cerr << *tok; | |
| 1661 | else | |
| 1662 | cerr << "a token"; | |
| 1663 | cerr << endl; | |
| 1664 | } | |
| 1665 | } | |
| 1666 | ||
| 1667 | ||
| 1668 | void | |
| 1669 | typeresolution_info::invalid (const token* tok, exp_type pe) | |
| 1670 | { | |
| 1671 | num_still_unresolved ++; | |
| 1672 | ||
| 1673 | if (assert_resolvability) | |
| 1674 | { | |
| 1675 | cerr << "error: invalid type " << pe << " for "; | |
| 1676 | if (tok) | |
| 1677 | cerr << *tok; | |
| 1678 | else | |
| 1679 | cerr << "a token"; | |
| 1680 | cerr << endl; | |
| 1681 | } | |
| 1682 | } | |
| 1683 | ||
| 1684 | ||
| 1685 | void | |
| 1686 | typeresolution_info::mismatch (const token* tok, exp_type t1, exp_type t2) | |
| 1687 | { | |
| 1688 | num_still_unresolved ++; | |
| 1689 | ||
| 1690 | if (assert_resolvability) | |
| 1691 | { | |
| 1692 | cerr << "error: type mismatch for "; | |
| 1693 | if (tok) | |
| 1694 | cerr << *tok; | |
| 1695 | else | |
| 1696 | cerr << "a token"; | |
| 1697 | cerr << ": " << t1 << " vs. " << t2 << endl; | |
| 1698 | } | |
| 1699 | } | |
| 1700 | ||
| 1701 | ||
| 1702 | void | |
| 1703 | typeresolution_info::resolved (const token* tok, exp_type t) | |
| 1704 | { | |
| 1705 | num_newly_resolved ++; | |
| 1706 | // cerr << "resolved " << *e->tok << " type " << t << endl; | |
| 1707 | } | |
| 1708 |