Added definite return analysis and void-return deduction.
1 /* vim: set ts=4 sw=4 tw=99 et:
3 * Copyright (C) 2012 David Anderson
5 * This file is part of JITCraft.
7 * JITCraft is free software: you can redistribute it and/or modify it under
8 * the terms of the GNU General Public License as published by the Free
9 * Software Foundation, either version 3 of the License, or (at your option)
12 * Foobar is distributed in the hope that it will be useful, but WITHOUT ANY
13 * WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
14 * FOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
16 * You should have received a copy of the GNU General Public License along with
17 * JITCraft. If not, see http://www.gnu.org/licenses/.
19 #include "SemanticAnalysis.h"
22 #include "TypeManager.h"
30 IsPrimitiveType(Type *type, PrimitiveType primitive)
35 if (!type->isPrimitive())
38 return primitive == type->primitive();
42 EvaluateBinaryOperation(TokenKind kind, const BoxedPrimitive &left, const BoxedPrimitive &right,
47 *out = BoxedPrimitive::Int(left.toInt() + right.toInt());
51 *out = BoxedPrimitive::Int(left.toInt() << right.toInt());
60 EvaluateForConstant(Expression *expression, BoxedPrimitive *out)
62 switch (expression->kind()) {
63 case AstNode::kBinaryExpression:
67 BinaryExpression *b = expression->toBinaryExpression();
68 if (!EvaluateForConstant(b->left(), &left) ||
69 !EvaluateForConstant(b->right(), &right))
73 return EvaluateBinaryOperation(b->token(), left, right, out);
76 case AstNode::kNameProxy:
78 NameProxy *proxy = expression->toNameProxy();
79 if (proxy->binding()->kind() != BoundName::CONSTANT)
82 NamedConstant *c = proxy->binding()->toNamedConstant();
83 if (!IsPrimitiveType(c->type(), PrimitiveType_Int32))
90 case AstNode::kIntegerLiteral:
91 *out = BoxedPrimitive::Int(expression->toIntegerLiteral()->value());
99 SemanticAnalysis::SemanticAnalysis(Zone *zone, CompileContext &cc, ParseTree *tree)
109 SemanticAnalysis::analyze()
114 for (size_t i = 0; i < tree_->statements()->length(); i++) {
115 Statement *statement = tree_->statements()->at(i);
116 if (statement->isFunctionStatement()) {
117 FunctionStatement *funstmt = statement->toFunctionStatement();
118 if (!preanalyze(funstmt))
123 for (size_t i = 0; i < tree_->statements()->length(); i++) {
124 Statement *statement = tree_->statements()->at(i);
125 statement->accept(this);
128 return !cc_.failures();
132 SemanticAnalysis::visitAndApplyType(Expression *expression)
134 expression->accept(this);
135 return expression->type();
139 SemanticAnalysis::visitForLoad(Expression *expression)
141 Local<Type> type(zone_, visitAndApplyType(expression));
145 if (type->isReference()) {
146 expression->setType(type->contained());
147 } else if (expression->isIndexExpression()) {
149 expression->setType(type->contained());
150 } else if (expression->isCallExpression()) {
151 if (type->isVoid()) {
152 CallExpression *call = expression->toCallExpression();
153 cc_.reportError(call->callee()->pos(), Message_FunctionDoesNotReturnValue);
158 return expression->type();
161 // :TODO: this leaks an interior pointer.
162 static inline const char *
163 GetTypeName(Type *type)
167 if (type->isFunction())
169 if (type->isReference())
170 type = type->contained();
172 return type->tag()->chars();
177 SemanticAnalysis::unaryTypeError(AstNode *node, TokenKind token, Handle<Type> type)
179 const char *op = TokenNames[token];
180 const char *name = GetTypeName(type);
182 cc_.reportError(node->pos(), Message_UnaryTypeError, op, name);
186 SemanticAnalysis::binaryTypeError(AstNode *node, TokenKind token, Handle<Type> left, Handle<Type> right)
188 const char *op = TokenNames[token];
189 const char *leftName = GetTypeName(left);
190 const char *rightName = GetTypeName(right);
192 cc_.reportError(node->pos(), Message_BinaryTypeError, op, leftName, rightName);
196 SemanticAnalysis::typeCheckBinary(AstNode *node, TokenKind token, Handle<Type> left, Handle<Type> right)
198 if ((token >= TOK_PLUS && token <= TOK_SLASH) ||
199 (token >= TOK_ASSIGN_ADD && token <= TOK_ASSIGN_DIV))
201 // If either side is a float, the other side must be a float or int.
202 if (left->isFloat() || right->isFloat()) {
203 Type *floatType = left->isFloat() ? left : right;
204 Type *other = left->isFloat() ? right : left;
205 if (other->isFloat() || (other->isInt32() && !other->tag()))
210 if ((!left->isInt32() || left->tag()) ||
211 (!right->isInt32() || right->tag()))
213 binaryTypeError(node, token, left, right);
221 SemanticAnalysis::canTest(AstNode *node, Handle<Type> type)
223 if (type->isArray()) {
224 cc_.reportError(node->pos(), Message_ArrayIsNotBoolean);
231 SemanticAnalysis::visitForLValue(Expression *expression)
233 if (expression->isIndexExpression()) {
234 Type *type = visitAndApplyType(expression);
238 if (type->isConst()) {
239 cc_.reportError(expression->pos(), Message_ReferenceIsConstant);
243 return type->contained();
246 if (!expression->isNameProxy()) {
247 cc_.reportError(expression->pos(), Message_ExpectedLValue);
251 NameProxy *proxy = expression->toNameProxy();
252 if (!proxy->binding()->kind() == BoundName::VARIABLE) {
253 cc_.reportError(expression->pos(), Message_ExpectedLValue);
257 Variable *var = proxy->binding()->toVariable();
261 if (var->type()->isReference()) {
262 if (var->type()->isConst()) {
263 cc_.reportError(expression->pos(), Message_ReferenceIsConstant);
266 return var->type()->contained();
271 SemanticAnalysis::ControlStatus
272 SemanticAnalysis::visitForControl(Statement *statement)
274 ControlScope scope(&control_);
275 statement->accept(this);
276 return scope.status();
280 SemanticAnalysis::visit(NameProxy *name)
282 name->setType(name->binding()->type());
286 SemanticAnalysis::visit(Assignment *node)
288 Local<Type> exprType(zone_, visitForLoad(node->expression()));
292 Local<Type> type(zone_, visitForLValue(node->lvalue()));
296 if (node->token() != TOK_ASSIGN) {
297 if (!typeCheckBinary(node, node->token(), type, exprType))
301 if (!typeCheckAssignment(node, exprType, type, true))
308 SemanticAnalysis::visit(IncDecExpression *node)
310 Local<Type> type(zone_, visitForLValue(node->expression()));
314 assert(!type->isReference());
316 if ((!type->isInt32() || (type->isInt32() && type->tag())) && !type->isFloat()) {
317 unaryTypeError(node, node->token(), type);
325 SemanticAnalysis::visit(BinaryExpression *node)
327 Local<Type> left(zone_, visitForLoad(node->left()));
328 Local<Type> right(zone_, visitForLoad(node->right()));
332 Local<Type> result(zone_, typeCheckBinary(node, node->token(), left, right));
336 node->setType(result);
340 SemanticAnalysis::visit(ReturnStatement *node)
342 // Set this early, so if we error we don't get more errors.
343 control_->returns(Returns_Always);
345 Local<Type> type(zone_, visitForLoad(node->expression()));
349 Local<Type> returnType(zone_, function_->returnType());
350 if (!typeCheckAssignment(node, type, returnType, true))
355 SemanticAnalysis::visit(ForStatement *node)
357 LoopScope loop(&loop_);
360 scope_ = node->scope();
362 if (node->initialization())
363 node->initialization()->accept(this);
365 if (node->condition()) {
366 Local<Type> type(zone_, visitForLoad(node->condition()));
367 if (!type || !canTest(node->condition(), type)) {
373 node->update()->accept(this);
375 ControlStatus status = visitForControl(node->body());
377 // If there is a condition, we treat this like a while loop. Otherwise,
378 // join point is only reachable if the loop contains a break. The
379 // termination status doesn't tell us anything yet.
380 if (node->condition() || loop.breaks())
381 control_->merge(status, ControlStatus(false, Returns_Never));
383 control_->terminate(status.returns());
386 scope_ = scope_->enclosing();
390 SemanticAnalysis::visit(WhileStatement *node)
392 LoopScope loop(&loop_);
394 Local<Type> type(zone_, visitForLoad(node->condition()));
395 if (!type || !canTest(node->condition(), type)) {
399 ControlStatus status = visitForControl(node->body());
401 // Detect loops that will always have a fork in the control-flow graph.
402 // This will be a |while| loop, or a |do-while| loop whose body does
405 // For now, we treat these as an |if| regardless of the direction the
406 // branch can take. That means the control-flow is always conditional.
407 if (node->token() == TOK_WHILE ||
408 !status.terminates() ||
411 control_->merge(status, ControlStatus(false, Returns_Never));
413 // At this point, we've got a do-while loop with a |return| or
414 // |continue|, meaning that the rest of the block is unreachable.
415 control_->terminate(status.returns());
420 SemanticAnalysis::visit(BlockStatement *node)
423 scope_ = node->scope();
424 for (size_t i = 0; i < node->statements()->length(); i++) {
425 Statement *statement = node->statements()->at(i);
426 statement->accept(this);
429 scope_ = scope_->enclosing();
433 SemanticAnalysis::visit(VariableDeclaration *node)
435 Local<String> tag(zone_, node->tag());
436 Local<Type> type(zone_, zone_->types()->getPrimitiveForTag(tag));
440 if (node->postDimensions()) {
441 Local<Type> init(zone_);
442 if (node->initialization()) {
443 if ((init = visitForLoad(node->initialization())) == NULL)
447 if (node->postDimensions()->length() > Type::MAX_UNSAFE_ARRAY_DEPTH) {
448 cc_.reportError(node->pos(), Message_MaximumArrayDepth);
452 // If we have an initializer, make sure it's the same type.
453 if (init && node->postDimensions()->length() != init->levels()) {
454 cc_.reportError(node->pos(), Message_ArrayInitializerDepth);
458 Local<Type> check(zone_, init);
459 for (size_t i = node->postDimensions()->length() - 1; i < node->postDimensions()->length(); i--) {
460 Expression *expr = node->postDimensions()->at(i);
464 Local<Type> exprType(zone_, visitForLoad(expr));
468 if (!IsPrimitiveType(exprType, PrimitiveType_Int32)) {
469 cc_.reportError(expr->pos(), Message_ArrayDimensionNotInteger);
474 if (EvaluateForConstant(expr, &box) && box.type() == PrimitiveType_Int32) {
477 cc_.reportError(expr->pos(), Message_BadArraySize);
480 if (size >= INT_MAX / sizeof(int)) {
481 cc_.reportError(expr->pos(), Message_ArraySizeTooBig);
486 if (check->elements() == -1) {
487 cc_.reportError(expr->pos(), Message_ArraySizeCannotBeDetermined);
491 if (check->isCharArray()) {
492 if (check->elements() > size) {
493 cc_.reportError(node->pos(), Message_ArrayAssignmentTooBig);
496 } else if (check->elements() != size) {
497 cc_.reportError(node->pos(), Message_ArrayDimensionSizeMismatch, unsigned(i));
503 cc_.reportError(node->pos(), Message_DynamicArrayHasInitializer);
507 // here until we get dynamic array support.
508 cc_.reportError(expr->pos(), Message_ArrayDimensionNotInteger);
512 // Infer the size from the initializer.
514 cc_.reportError(node->pos(), Message_ArraySizeNotSpecified);
517 if (init->elements() == -1) {
518 cc_.reportError(node->pos(), Message_ArraySizeCannotBeDetermined);
521 size = init->elements();
524 type = zone_->types()->unsafeArray(type, size, false);
529 check = check->contained();
532 // Right now, any array is allowed on the LIFO heap. In case the
533 // runtime actually decides to use the LIFO heap, we must reserve
534 // a slot so we can unwind the heap when the current scope leaves.
535 scope_->setUsesLifoHeap();
537 if (node->initialization()) {
538 Local<Type> right(zone_, visitForLoad(node->initialization()));
541 if (!typeCheckAssignment(node, type, right, true))
544 if (type->primitive() == PrimitiveType_Char) {
545 cc_.reportError(node->pos(), Message_CannotDeclareStringCell);
550 node->variable()->setType(type);
554 SemanticAnalysis::visit(IntegerLiteral *node)
556 node->setType(zone_->types()->getPrimitive(PrimitiveType_Int32));
560 SemanticAnalysis::visit(ExpressionStatement *node)
562 visitAndApplyType(node->expression());
566 SemanticAnalysis::preanalyze(FunctionStatement *node)
568 Local<FixedArray> args(zone_, FixedArray::New(zone_, node->parameters()->length(), Heap::Tenure_Old));
572 // Apply types to arguments.
573 for (size_t i = 0; i < node->parameters()->length(); i++) {
574 Parameter *parameter = node->parameters()->at(i);
576 Local<String> tag(zone_, parameter->tag());
577 Local<Type> type(zone_, zone_->types()->getPrimitiveForTag(tag));
581 if (parameter->reference()) {
582 if (parameter->dimensions()) {
583 cc_.reportError(parameter->variable()->pos(), Message_ArrayCantBeReference);
587 type = zone_->types()->getReference(type, parameter->isConst());
590 } else if (parameter->dimensions()) {
591 if (parameter->dimensions()->length() > Type::MAX_UNSAFE_ARRAY_DEPTH) {
592 cc_.reportError(parameter->variable()->pos(), Message_MaximumArrayDepth);
596 if (node->fun()->isNative() && parameter->dimensions()->length() > 1) {
597 cc_.reportError(parameter->variable()->pos(), Message_MultiArrayInNative);
601 for (size_t i = parameter->dimensions()->length() - 1;
602 i < parameter->dimensions()->length();
605 int size = Type::UNKNOWN_SIZE;
606 if (Expression *expr = parameter->dimensions()->at(i)) {
608 if (!EvaluateForConstant(expr, &box)) {
609 cc_.reportError(expr->pos(), Message_ArraySizeMustBeConstant);
615 type = zone_->types()->unsafeArray(type, size, parameter->isConst());
621 if ((type->isPrimitive() && type->primitive() == PrimitiveType_Char) ||
622 (type->isReference() && type->contained()->primitive() == PrimitiveType_Char))
624 cc_.reportError(node->pos(), Message_CannotDeclareStringCell);
628 args->set(zone_, i, type);
629 parameter->variable()->setType(type);
632 Local<Type> returnType(zone_);
633 if (node->hasExplicitReturn())
634 returnType = zone_->types()->getPrimitive(PrimitiveType_Int32);
636 returnType = zone_->types()->getVoid();
640 Local<Type> funtype(zone_);
641 if (node->fun()->isNative())
642 funtype = zone_->types()->newNative(returnType, args, node->fun()->isNativeVariadic());
644 funtype = zone_->types()->newFunction(returnType, args);
648 // If the function already has a type, it was forward declared before, and
649 // now this is our chance to type check the second declaration.
650 if (node->fun()->type()) {
651 Local<Type> forward(zone_, node->fun()->type());
652 if (!Type::Compare(funtype, forward)) {
653 cc_.reportError(node->pos(), Message_ForwardDeclareMismatches);
656 node->fun()->setForwardType(forward);
659 node->fun()->setType(funtype);
665 SemanticAnalysis::visit(FunctionStatement *node)
667 function_ = node->fun()->type();
668 scope_ = node->scope();
670 ControlScope control(&control_);
672 node->body()->accept(this);
674 ControlStatus status = control.status();
675 if (status.returns() == Returns_Sometimes) {
676 cc_.reportError(node->pos(), Message_NotAllPathsReturnValue);
677 } else if (status.returns() == Returns_Never) {
678 if (!function_->returnType()->isVoid())
679 cc_.reportError(node->pos(), Message_MissingReturnValue);
688 SemanticAnalysis::typeCheckArrayAssignment(AstNode *node, Handle<Type> argGiven, Handle<Type> argActual, bool forAssign)
690 Local<Type> actual(zone_, argActual);
691 Local<Type> given(zone_, argGiven);
693 if (!given->isArray()) {
694 cc_.reportError(node->pos(), Message_ExpectedArray);
697 if (given->levels() != actual->levels()) {
698 cc_.reportError(node->pos(), Message_ArrayDimensionsMismatch);
703 while (given->isArray()) {
704 if (actual->elements() != -1) {
706 if (given->elements() == -1 || actual->elements() == -1) {
707 cc_.reportError(node->pos(), Message_ArraySizeNotKnown);
711 if (actual->isCharArray()) {
712 // String assignment can have a truncated |given|.
713 if (given->elements() > actual->elements()) {
714 cc_.reportError(node->pos(), Message_ArrayAssignmentTooBig);
718 if (given->elements() != actual->elements()) {
719 cc_.reportError(node->pos(), Message_ArrayDimensionSizeMismatch, level);
724 if (actual->elements() != -1 && given->elements() != actual->elements()) {
725 cc_.reportError(node->pos(), Message_ArrayDimensionSizeMismatch, level);
730 given = given->contained();
731 actual = actual->contained();
734 if (given->primitive() != actual->primitive()) {
735 // :TODO: leaks an interior pointer.
736 const char *givenTag = given->tag() ? given->tag()->chars() : "_";
737 const char *actualTag = given->tag() ? given->tag()->chars() : "_";
738 cc_.reportError(node->pos(), Message_TagMismatch, actualTag, givenTag);
746 SemanticAnalysis::typeCheckAssignment(AstNode *node, Handle<Type> argGiven, Handle<Type> argActual, bool forAssign)
748 Local<Type> actual(zone_, argActual);
749 Local<Type> given(zone_, argGiven);
751 if (actual->isArray())
752 return typeCheckArrayAssignment(node, given, actual, forAssign);
754 if (given->isArray()) {
755 cc_.reportError(node->pos(), Message_CannotAssignArrayToScalar);
759 // The given type may not have been dereferenced.
760 if (given->isReference())
761 given = given->contained();
762 if (actual->isReference())
763 actual = actual->contained();
765 if (!given->isPrimitive() || !actual->isPrimitive()) {
766 cc_.reportError(node->pos(), Message_UnsupportedType);
770 if (given->primitive() != actual->primitive()) {
771 // :TODO: leaks an interior pointer
772 const char *givenTag = given->tag() ? given->tag()->chars() : "_";
773 const char *actualTag = given->tag() ? given->tag()->chars() : "_";
774 cc_.reportError(node->pos(), Message_TagMismatch, actualTag, givenTag);
782 SemanticAnalysis::typeCheckParameter(AstNode *node, Handle<Type> given, Handle<Type> actual)
784 return typeCheckAssignment(node, given, actual, false);
788 SemanticAnalysis::visit(CallExpression *node)
790 Local<Type> type(zone_, visitForLoad(node->callee()));
794 if (!type->isFunction()) {
795 cc_.reportError(node->pos(), Message_CalleeNotFunction);
798 if (node->arguments()->length() != type->parameters()->length()) {
799 if (!type->isNative() ||
800 !type->isNativeVariadic() ||
801 node->arguments()->length() < type->parameters()->length())
803 cc_.reportError(node->pos(), Message_ArgumentCountMismatch);
808 // Currently, only calls from a NameProxy are possible.
809 NamedFunction *fun = node->callee()->toNameProxy()->binding()->toNamedFunction();
810 if (fun->isForward()) {
811 cc_.reportError(node->pos(), Message_ForwardNotImplemented, fun->name()->chars());
815 for (unsigned i = 0; i < node->arguments()->length(); i++) {
816 // Note: we don't compute references unless the callee needs a
818 Expression *expression = node->arguments()->at(i);
819 Local<Type> given(zone_, visitAndApplyType(expression));
823 // Since we skip a load, check for a call expression that returns void.
824 if (given->isVoid()) {
825 assert(expression->isCallExpression());
826 CallExpression *call = expression->toCallExpression();
827 cc_.reportError(call->callee()->pos(), Message_FunctionDoesNotReturnValue);
831 // If it's an IndexExpression, we special case the ability to compute
833 if (expression->isIndexExpression())
834 given = given->contained();
836 bool mustComputeReference;
837 if (i >= type->parameters()->length()) {
838 // Special handling for natives. Arguments past the formal must
839 // always have references computed. However, arrays are already
840 // references for natives.
841 assert(type->isNative() && type->isNativeVariadic());
842 mustComputeReference = !given->isReference() && !given->isArray();
844 Local<Type> actual(zone_, type->parameterAt(i));
845 if (!typeCheckParameter(expression, given, actual))
847 mustComputeReference = actual->isReference() && !given->isReference();
850 // If the actual is a reference, and the given is not, we'll need to
851 // compute a reference, which may be allocated on the stack.
852 if (mustComputeReference)
853 scope_->setUsesLifoHeap();
856 node->setType(fun->type()->returnType());
860 SemanticAnalysis::ControlScope::merge(const ControlStatus &left, const ControlStatus &right)
862 // If both sides terminate in a return, we consider this a return as well.
863 if (left.returns() == Returns_Always && right.returns() == Returns_Always) {
864 terminate(Returns_Always);
868 ReturnStatus status = (left.returns() == Returns_Never && right.returns() == Returns_Never)
872 if (left.terminates() && right.terminates())
879 SemanticAnalysis::visit(IfStatement *node)
881 Local<Type> type(zone_, visitForLoad(node->condition()));
885 if (!canTest(node, type))
888 ControlStatus trueBranch = visitForControl(node->ifTrue());
889 ControlStatus falseBranch = ControlStatus(false, Returns_Never);
892 falseBranch = visitForControl(node->ifFalse());
894 control_->merge(trueBranch, falseBranch);
898 SemanticAnalysis::visit(IndexExpression *node)
900 Local<Type> left(zone_, visitForLoad(node->left()));
901 Local<Type> right(zone_, visitForLoad(node->right()));
906 if (!right->isInt32()) {
907 cc_.reportError(node->right()->pos(), Message_IndexMustBeInteger);
910 if (!left->isArray()) {
911 cc_.reportError(node->left()->pos(), Message_IndexBaseMustBeArray);
918 SemanticAnalysis::visit(EnumStatement *node)
920 // Forbid the blessed types.
921 Local<String> tag(zone_, node->name());
922 if (!strcmp(tag->chars(), "Float") ||
923 !strcmp(tag->chars(), "String") ||
924 !strcmp(tag->chars(), "any") ||
925 !strcmp(tag->chars(), "bool"))
927 cc_.reportError(node->pos(), Message_InvalidEnumName);
931 Local<Type> type(zone_, zone_->types()->getPrimitiveForTag(tag));
936 for (size_t i = 0; i < node->entries()->length(); i++) {
937 EnumStatement::Entry &entry = node->entries()->at(i);
939 BoxedPrimitive box = BoxedPrimitive::Int(value);
941 entry.expr->accept(this);
942 if (!EvaluateForConstant(entry.expr, &box)) {
943 cc_.reportError(entry.expr->pos(), Message_EnumFieldNotConstant);
947 assert(box.type() == PrimitiveType_Int32);
951 entry.name->set(type, box);
957 SemanticAnalysis::visit(StringLiteral *node)
959 Local<Type> type(zone_, zone_->types()->getPrimitive(PrimitiveType_Char));
963 type = Type::NewArray(zone_, type, node->string()->length() + 1, true);
971 SemanticAnalysis::visit(FloatLiteral *node)
973 Local<Type> type(zone_, zone_->types()->getPrimitive(PrimitiveType_Float));
981 SemanticAnalysis::visit(BreakStatement *node)
984 cc_.reportError(node->pos(), Message_BreakOutsideLoop);
988 // Note: ignore breaks that occur in unreachable control-flow.
989 if (!control_->status().terminates())
992 control_->terminate(Returns_Never);
996 SemanticAnalysis::visit(ContinueStatement *node)
999 cc_.reportError(node->pos(), Message_ContinueOutsideLoop);
1003 // Note: ignore continues that occur in unreachable control-flow.
1004 if (!control_->status().terminates())
1005 loop_->addContinue();
1007 control_->terminate(Returns_Never);