Add codegen support for goto (#163)

Depends on #161. See #161 for a more detailed description of how goto is
implemented.

On the codegen part, this PR emits the goto_block and goto macros when
it hits functions that use labels and gotos internally. It also hoists
local variables at the start of the function so that all goto_block arms
can see and use the local variables.

Author: lucic71

cpp2rust/converter/converter.cpp

@@ -364,7 +364,59 @@ bool Converter::VisitFunctionDecl(clang::FunctionDecl *decl) {
   return false;
 }
 
+void Converter::EmitHoistedDecls(clang::CompoundStmt *body) {
+  for (auto *child : body->body()) {
+    if (auto *decl_stmt = clang::dyn_cast<clang::DeclStmt>(child)) {
+      for (auto *decl : decl_stmt->decls()) {
+        if (auto *var = clang::dyn_cast<clang::VarDecl>(decl);
+            var && var->isLocalVarDecl() && !IsGlobalVar(var)) {
+          hoisted_decls_.insert(var);
+          if (ConvertVarDeclSkipInit(var)) {
+            StrCat(token::kAssign, ConvertVarDefaultInit(var->getType()),
+                   token::kSemiColon);
+          }
+        }
+      }
+    }
+  }
+}
+
+void Converter::ConvertGotoBlock(clang::CompoundStmt *body) {
+  PushHoistedDecls push(hoisted_decls_);
+  EmitHoistedDecls(body);
+
+  StrCat("goto_block!");
+  {
+    PushParen paren(*this);
+    PushBrace outer(*this);
+    StrCat("'__entry: ");
+    std::optional<PushBrace> arm;
+    arm.emplace(*this);
+    for (auto *child : body->body()) {
+      if (auto *label = clang::dyn_cast<clang::LabelStmt>(child)) {
+        arm.reset();
+        StrCat(std::format("'{}: ", label->getDecl()->getName().str()));
+        arm.emplace(*this);
+        Convert(label->getSubStmt());
+      } else {
+        Convert(child);
+      }
+    }
+  }
+  StrCat(token::kSemiColon);
+}
+
 void Converter::ConvertFunctionBody(clang::FunctionDecl *decl) {
+  if (auto compound = clang::dyn_cast<clang::CompoundStmt>(decl->getBody())) {
+    if (CompoundHasTopLevelLabel(compound)) {
+      ConvertGotoBlock(compound);
+      if (!decl->getReturnType()->isVoidType()) {
+        StrCat(R"(panic!("ub: non-void function does not return a value"))");
+      }
+      return;
+    }
+  }
+
   Convert(decl->getBody());
   if (!decl->getReturnType()->isVoidType()) {
     if (auto compound = clang::dyn_cast<clang::CompoundStmt>(decl->getBody())) {
@@ -467,7 +519,20 @@ void Converter::ConvertVarDeclInitializer(clang::VarDecl *decl) {
   }
 }
 
+void Converter::EmitHoistedInArmAssignment(clang::VarDecl *decl) {
+  if (!decl->hasInit()) {
+    return;
+  }
+  StrCat(GetNamedDeclAsString(decl), token::kAssign);
+  ConvertVarInit(decl->getType(), decl->getInit());
+  StrCat(token::kSemiColon);
+}
+
 void Converter::ConvertVarDecl(clang::VarDecl *decl) {
+  if (hoisted_decls_.contains(decl)) {
+    EmitHoistedInArmAssignment(decl);
+    return;
+  }
   if (!ConvertVarDeclSkipInit(decl)) {
     // Skip global variables declared extern
     return;
@@ -985,6 +1050,10 @@ bool Converter::Convert(clang::Stmt *stmt) {
 }
 
 bool Converter::VisitCompoundStmt(clang::CompoundStmt *stmt) {
+  if (CompoundHasTopLevelLabel(stmt)) {
+    ConvertGotoBlock(stmt);
+    return false;
+  }
   for (auto *child : stmt->body()) {
     Convert(child);
   }
@@ -1011,6 +1080,11 @@ bool Converter::VisitReturnStmt(clang::ReturnStmt *stmt) {
   return false;
 }
 
+bool Converter::VisitGotoStmt(clang::GotoStmt *stmt) {
+  StrCat(std::format("goto!('{})", stmt->getLabel()->getName().str()));
+  return false;
+}
+
 void Converter::ConvertCondition(clang::Expr *cond) {
   PushExprKind push(*this, ExprKind::RValue);
   Convert(NormalizeToBool(cond, ctx_));

cpp2rust/converter/converter.h

@@ -82,12 +82,18 @@ class Converter : public clang::RecursiveASTVisitor<Converter> {
 
   virtual void ConvertFunctionBody(clang::FunctionDecl *decl);
 
+  void ConvertGotoBlock(clang::CompoundStmt *body);
+
+  void EmitHoistedDecls(clang::CompoundStmt *body);
+
   virtual bool VisitFunctionTemplateDecl(clang::FunctionTemplateDecl *decl);
 
   virtual bool VisitVarDecl(clang::VarDecl *decl);
 
   void ConvertVarDecl(clang::VarDecl *decl);
 
+  virtual void EmitHoistedInArmAssignment(clang::VarDecl *decl);
+
   void ConvertVarDeclInitializer(clang::VarDecl *decl);
 
   virtual void ConvertGlobalVarDecl(clang::VarDecl *decl);
@@ -125,6 +131,8 @@ class Converter : public clang::RecursiveASTVisitor<Converter> {
 
   virtual bool VisitReturnStmt(clang::ReturnStmt *stmt);
 
+  virtual bool VisitGotoStmt(clang::GotoStmt *stmt);
+
   void ConvertCondition(clang::Expr *cond);
 
   virtual bool VisitIfStmt(clang::IfStmt *stmt);
@@ -646,6 +654,24 @@ class Converter : public clang::RecursiveASTVisitor<Converter> {
 
   std::unordered_set<const clang::VarDecl *> map_iter_decls_;
 
+  // Local variables hoisted outside a goto_block so that all labels can see and
+  // use the variables.
+  std::unordered_set<const clang::VarDecl *> hoisted_decls_;
+  class PushHoistedDecls {
+  public:
+    PushHoistedDecls(std::unordered_set<const clang::VarDecl *> &field)
+        : field_(field), saved_(std::move(field)) {
+      field_.clear();
+    }
+    ~PushHoistedDecls() { field_ = std::move(saved_); }
+    PushHoistedDecls(const PushHoistedDecls &) = delete;
+    PushHoistedDecls &operator=(const PushHoistedDecls &) = delete;
+
+  private:
+    std::unordered_set<const clang::VarDecl *> &field_;
+    std::unordered_set<const clang::VarDecl *> saved_;
+  };
+
   struct ScopedMapIterDecl {
     Converter &c;
     const clang::VarDecl *decl;

cpp2rust/converter/converter_lib.cpp

@@ -811,7 +811,8 @@ static bool SwitchCaseHasFallthrough(clang::Stmt *stmt) {
   }
   if (clang::isa<clang::BreakStmt>(stmt) ||
       clang::isa<clang::ContinueStmt>(stmt) ||
-      clang::isa<clang::ReturnStmt>(stmt)) {
+      clang::isa<clang::ReturnStmt>(stmt) ||
+      clang::isa<clang::GotoStmt>(stmt)) {
     return false;
   }
   return true;
@@ -829,6 +830,15 @@ bool SwitchHasFallthrough(clang::SwitchStmt *stmt) {
   return false;
 }
 
+bool CompoundHasTopLevelLabel(const clang::CompoundStmt *compound) {
+  for (const auto *child : compound->body()) {
+    if (clang::isa<clang::LabelStmt>(child)) {
+      return true;
+    }
+  }
+  return false;
+}
+
 std::string_view Trim(std::string_view s) {
   auto is_space = [](unsigned char c) { return std::isspace(c); };
   auto b = std::find_if_not(s.begin(), s.end(), is_space);

cpp2rust/converter/converter_lib.h

@@ -169,6 +169,8 @@ std::vector<clang::Stmt *> GetSwitchCaseBody(clang::CompoundStmt *body,
 
 bool SwitchHasFallthrough(clang::SwitchStmt *stmt);
 
+bool CompoundHasTopLevelLabel(const clang::CompoundStmt *compound);
+
 std::string_view Trim(std::string_view s);
 
 void Unwrap(std::string &s, std::string_view prefix, std::string_view suffix);

cpp2rust/converter/models/converter_refcount.cpp

@@ -621,6 +621,24 @@ bool ConverterRefCount::ConvertLambdaVarDecl(clang::VarDecl *decl) {
   return false;
 }
 
+bool ConverterRefCount::ConvertVarDeclSkipInit(clang::VarDecl *decl) {
+  bool unboxed = in_function_formals_;
+  PushConversionKind push(*this, unboxed ? ConversionKind::Unboxed
+                                         : ConversionKind::FullRefCount);
+  return Converter::ConvertVarDeclSkipInit(decl);
+}
+
+void ConverterRefCount::EmitHoistedInArmAssignment(clang::VarDecl *decl) {
+  if (!decl->hasInit()) {
+    return;
+  }
+  PushConversionKind push(*this, ConversionKind::Unboxed);
+  StrCat(token::kStar, GetNamedDeclAsString(decl), ".borrow_mut()",
+         token::kAssign);
+  Convert(decl->getInit());
+  StrCat(token::kSemiColon);
+}
+
 void ConverterRefCount::ConvertGlobalVarDecl(clang::VarDecl *decl) {
   StrCat("thread_local!");
   {

cpp2rust/converter/models/converter_refcount.h

@@ -57,6 +57,10 @@ class ConverterRefCount final : public Converter {
 
   void ConvertVaListVarDecl(clang::VarDecl *decl) override;
 
+  bool ConvertVarDeclSkipInit(clang::VarDecl *decl) override;
+
+  void EmitHoistedInArmAssignment(clang::VarDecl *decl) override;
+
   bool ConvertLambdaVarDecl(clang::VarDecl *decl) override;
 
   bool VisitDeclRefExpr(clang::DeclRefExpr *expr) override;

tests/unit/goto_aggregate_default.c

@@ -0,0 +1,31 @@
+#include <assert.h>
+#include <stdio.h>
+
+struct Point {
+  int x;
+  int y;
+};
+
+static int agg(int n) {
+  char buf40[40];
+  unsigned char buf256[256];
+  int arr64[64];
+  long longs[33];
+  struct Point p;
+  int *ptr;
+  int (*fp)(int);
+  FILE *file;
+  int total = 0;
+  if (n < 0) {
+    goto out;
+  }
+  total = 1;
+out:
+  return total;
+}
+
+int main(void) {
+  assert(agg(-1) == 0);
+  assert(agg(1) == 1);
+  return 0;
+}

tests/unit/goto_backward.c

@@ -0,0 +1,18 @@
+#include <assert.h>
+
+static int retry(int n) {
+  int count = 0;
+  int acc = 0;
+again:
+  count += 1;
+  acc += n;
+  if (count < 3) {
+    goto again;
+  }
+  return acc;
+}
+
+int main(void) {
+  assert(retry(4) == 12);
+  return 0;
+}

tests/unit/goto_cleanup.c

@@ -0,0 +1,51 @@
+#include <assert.h>
+
+static int early(int n) {
+  int ret = 0;
+  if (n < 0) {
+    ret = -1;
+    goto out;
+  }
+  ret = 100;
+out:
+  return ret;
+}
+
+static int from_loop(int n) {
+  int ret = 0;
+  for (int i = 0; i < n; i++) {
+    if (i == 3) {
+      ret = 7;
+      goto out;
+    }
+    ret += i;
+  }
+  ret = 999;
+out:
+  return ret;
+}
+
+static int from_switch(int n) {
+  int ret = 0;
+  switch (n) {
+  case 1:
+    ret = 10;
+    goto out;
+  default:
+    ret = 20;
+    break;
+  }
+  ret = 999;
+out:
+  return ret;
+}
+
+int main(void) {
+  assert(early(-1) == -1);
+  assert(early(5) == 100);
+  assert(from_loop(2) == 999);
+  assert(from_loop(10) == 7);
+  assert(from_switch(1) == 10);
+  assert(from_switch(2) == 999);
+  return 0;
+}

tests/unit/goto_loop_control.c

@@ -0,0 +1,22 @@
+#include <assert.h>
+
+static int loopctl(void) {
+  int sum = 0;
+  for (int i = 0; i < 5; i++) {
+    if (i == 1) {
+      continue;
+    }
+    if (i == 4) {
+      break;
+    }
+    goto add;
+  add:
+    sum += 1;
+  }
+  return sum;
+}
+
+int main(void) {
+  assert(loopctl() == 3);
+  return 0;
+}