Sync to upstream/release/575 (#919)

* `Luau.Analyze.CLI` now has experimental support for concurrent type
checking. Use the option `-jN` where `N` is the number of threads to
spawn.
* Improve typechecking performance by ~17% by making the function
`Luau::follow` much more efficient.
* Tighten up the type of `os.date`
* Removed `ParseOptions::allowTypeAnnotations` and
`ParseOptions::supportContinueStatement`

New solver

* Improve the reliability of function overload resolution
* More work toward supporting parallel type checking
* Fix a bug in inference of `==` and `~=` which would erroneously infer
that the operands were `boolean`
* Better error reporting when `for...in` loops are used incorrectly.

CodeGen

* Fix unwind registration when libunwind is used on Linux
* Fixed replaced IR instruction use count
* Convert X64 unwind info generation to standard prologue
* Implement A64 unwind info support for Dwarf2
* Live in/out data for linear blocks is now created
* Add side-exit VM register requirements to the IR dump
* Reuse ConstPropState between block chains 
* Remove redundant base update

---------

Co-authored-by: Arseny Kapoulkine <arseny.kapoulkine@gmail.com>
Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>

Analysis/include/Luau/Constraint.h

@@ -107,6 +107,11 @@ struct FunctionCallConstraint
     TypePackId result;
     class AstExprCall* callSite;
     std::vector<std::optional<TypeId>> discriminantTypes;
+
+    // When we dispatch this constraint, we update the key at this map to record
+    // the overload that we selected.
+    DenseHashMap<const AstNode*, TypeId>* astOriginalCallTypes;
+    DenseHashMap<const AstNode*, TypeId>* astOverloadResolvedTypes;
 };
 
 // result ~ prim ExpectedType SomeSingletonType MultitonType

Analysis/include/Luau/Frontend.h

@@ -28,6 +28,7 @@ struct FileResolver;
 struct ModuleResolver;
 struct ParseResult;
 struct HotComment;
+struct BuildQueueItem;
 
 struct LoadDefinitionFileResult
 {
@@ -171,7 +172,18 @@ struct Frontend
     LoadDefinitionFileResult loadDefinitionFile(GlobalTypes& globals, ScopePtr targetScope, std::string_view source, const std::string& packageName,
         bool captureComments, bool typeCheckForAutocomplete = false);
 
+    // Batch module checking. Queue modules and check them together, retrieve results with 'getCheckResult'
+    // If provided, 'executeTask' function is allowed to call the 'task' function on any thread and return without waiting for 'task' to complete
+    void queueModuleCheck(const std::vector<ModuleName>& names);
+    void queueModuleCheck(const ModuleName& name);
+    std::vector<ModuleName> checkQueuedModules(std::optional<FrontendOptions> optionOverride = {},
+        std::function<void(std::function<void()> task)> executeTask = {}, std::function<void(size_t done, size_t total)> progress = {});
+
+    std::optional<CheckResult> getCheckResult(const ModuleName& name, bool accumulateNested, bool forAutocomplete = false);
+
 private:
+    CheckResult check_DEPRECATED(const ModuleName& name, std::optional<FrontendOptions> optionOverride = {});
+
     struct TypeCheckLimits
     {
         std::optional<double> finishTime;
@@ -185,7 +197,14 @@ private:
     std::pair<SourceNode*, SourceModule*> getSourceNode(const ModuleName& name);
     SourceModule parse(const ModuleName& name, std::string_view src, const ParseOptions& parseOptions);
 
-    bool parseGraph(std::vector<ModuleName>& buildQueue, const ModuleName& root, bool forAutocomplete);
+    bool parseGraph(
+        std::vector<ModuleName>& buildQueue, const ModuleName& root, bool forAutocomplete, std::function<bool(const ModuleName&)> canSkip = {});
+
+    void addBuildQueueItems(std::vector<BuildQueueItem>& items, std::vector<ModuleName>& buildQueue, bool cycleDetected,
+        std::unordered_set<Luau::ModuleName>& seen, const FrontendOptions& frontendOptions);
+    void checkBuildQueueItem(BuildQueueItem& item);
+    void checkBuildQueueItems(std::vector<BuildQueueItem>& items);
+    void recordItemResult(const BuildQueueItem& item);
 
     static LintResult classifyLints(const std::vector<LintWarning>& warnings, const Config& config);
 
@@ -212,11 +231,13 @@ public:
     InternalErrorReporter iceHandler;
     std::function<void(const ModuleName& name, const ScopePtr& scope, bool forAutocomplete)> prepareModuleScope;
 
-    std::unordered_map<ModuleName, SourceNode> sourceNodes;
+    std::unordered_map<ModuleName, std::shared_ptr<SourceNode>> sourceNodes;
-    std::unordered_map<ModuleName, SourceModule> sourceModules;
+    std::unordered_map<ModuleName, std::shared_ptr<SourceModule>> sourceModules;
     std::unordered_map<ModuleName, RequireTraceResult> requireTrace;
 
     Stats stats = {};
+
+    std::vector<ModuleName> moduleQueue;
 };
 
 ModulePtr check(const SourceModule& sourceModule, const std::vector<RequireCycle>& requireCycles, NotNull<BuiltinTypes> builtinTypes,

Analysis/include/Luau/Normalize.h

@@ -226,10 +226,6 @@ struct NormalizedType
 
     NormalizedClassType classes;
 
-    // The class part of the type.
-    // Each element of this set is a class, and none of the classes are subclasses of each other.
-    TypeIds DEPRECATED_classes;
-
     // The error part of the type.
     // This type is either never or the error type.
     TypeId errors;
@@ -333,8 +329,6 @@ public:
     // ------- Normalizing intersections
     TypeId intersectionOfTops(TypeId here, TypeId there);
     TypeId intersectionOfBools(TypeId here, TypeId there);
-    void DEPRECATED_intersectClasses(TypeIds& heres, const TypeIds& theres);
-    void DEPRECATED_intersectClassesWithClass(TypeIds& heres, TypeId there);
     void intersectClasses(NormalizedClassType& heres, const NormalizedClassType& theres);
     void intersectClassesWithClass(NormalizedClassType& heres, TypeId there);
     void intersectStrings(NormalizedStringType& here, const NormalizedStringType& there);

Analysis/include/Luau/Type.h

@@ -694,7 +694,7 @@ bool areEqual(SeenSet& seen, const Type& lhs, const Type& rhs);
 
 // Follow BoundTypes until we get to something real
 TypeId follow(TypeId t);
-TypeId follow(TypeId t, std::function<TypeId(TypeId)> mapper);
+TypeId follow(TypeId t, const void* context, TypeId (*mapper)(const void*, TypeId));
 
 std::vector<TypeId> flattenIntersection(TypeId ty);
 

Analysis/include/Luau/TypePack.h

@@ -169,7 +169,7 @@ using SeenSet = std::set<std::pair<const void*, const void*>>;
 bool areEqual(SeenSet& seen, const TypePackVar& lhs, const TypePackVar& rhs);
 
 TypePackId follow(TypePackId tp);
-TypePackId follow(TypePackId tp, std::function<TypePackId(TypePackId)> mapper);
+TypePackId follow(TypePackId t, const void* context, TypePackId (*mapper)(const void*, TypePackId));
 
 size_t size(TypePackId tp, TxnLog* log = nullptr);
 bool finite(TypePackId tp, TxnLog* log = nullptr);

Analysis/include/Luau/Unifier.h

@@ -163,5 +163,6 @@ private:
 
 void promoteTypeLevels(TxnLog& log, const TypeArena* arena, TypeLevel minLevel, Scope* outerScope, bool useScope, TypePackId tp);
 std::optional<TypeError> hasUnificationTooComplex(const ErrorVec& errors);
+std::optional<TypeError> hasCountMismatch(const ErrorVec& errors);
 
 } // namespace Luau

Analysis/src/ConstraintGraphBuilder.cpp

@@ -18,7 +18,6 @@
 
 LUAU_FASTINT(LuauCheckRecursionLimit);
 LUAU_FASTFLAG(DebugLuauMagicTypes);
-LUAU_FASTFLAG(LuauNegatedClassTypes);
 
 namespace Luau
 {
@@ -1016,7 +1015,7 @@ static bool isMetamethod(const Name& name)
 
 ControlFlow ConstraintGraphBuilder::visit(const ScopePtr& scope, AstStatDeclareClass* declaredClass)
 {
-    std::optional<TypeId> superTy = FFlag::LuauNegatedClassTypes ? std::make_optional(builtinTypes->classType) : std::nullopt;
+    std::optional<TypeId> superTy = std::make_optional(builtinTypes->classType);
     if (declaredClass->superName)
     {
         Name superName = Name(declaredClass->superName->value);
@@ -1420,6 +1419,8 @@ InferencePack ConstraintGraphBuilder::checkPack(const ScopePtr& scope, AstExprCa
                 rets,
                 call,
                 std::move(discriminantTypes),
+                &module->astOriginalCallTypes,
+                &module->astOverloadResolvedTypes,
             });
 
         // We force constraints produced by checking function arguments to wait
@@ -1772,7 +1773,7 @@ std::tuple<TypeId, TypeId, RefinementId> ConstraintGraphBuilder::checkBinary(
             TypeId ty = follow(typeFun->type);
 
             // We're only interested in the root class of any classes.
-            if (auto ctv = get<ClassType>(ty); !ctv || (FFlag::LuauNegatedClassTypes ? (ctv->parent == builtinTypes->classType) : !ctv->parent))
+            if (auto ctv = get<ClassType>(ty); !ctv || ctv->parent == builtinTypes->classType)
                 discriminantTy = ty;
         }
 
@@ -1786,8 +1787,10 @@ std::tuple<TypeId, TypeId, RefinementId> ConstraintGraphBuilder::checkBinary(
     }
     else if (binary->op == AstExprBinary::CompareEq || binary->op == AstExprBinary::CompareNe)
     {
-        TypeId leftType = check(scope, binary->left, ValueContext::RValue, expectedType, true).ty;
+        // We are checking a binary expression of the form a op b
-        TypeId rightType = check(scope, binary->right, ValueContext::RValue, expectedType, true).ty;
+        // Just because a op b is epxected to return a bool, doesn't mean a, b are expected to be bools too
+        TypeId leftType = check(scope, binary->left, ValueContext::RValue, {}, true).ty;
+        TypeId rightType = check(scope, binary->right, ValueContext::RValue, {}, true).ty;
 
         RefinementId leftRefinement = nullptr;
         if (auto bc = dfg->getBreadcrumb(binary->left))

Analysis/src/ConstraintSolver.cpp

@@ -1172,6 +1172,9 @@ bool ConstraintSolver::tryDispatch(const FunctionCallConstraint& c, NotNull<cons
     else if (auto it = get<IntersectionType>(fn))
         fn = collapse(it).value_or(fn);
 
+    if (c.callSite)
+        (*c.astOriginalCallTypes)[c.callSite] = fn;
+
     // We don't support magic __call metamethods.
     if (std::optional<TypeId> callMm = findMetatableEntry(builtinTypes, errors, fn, "__call", constraint->location))
     {
@@ -1219,10 +1222,22 @@ bool ConstraintSolver::tryDispatch(const FunctionCallConstraint& c, NotNull<cons
 
     TypeId inferredTy = arena->addType(FunctionType{TypeLevel{}, constraint->scope.get(), argsPack, c.result});
 
-    std::vector<TypeId> overloads = flattenIntersection(fn);
+    const NormalizedType* normFn = normalizer->normalize(fn);
+    if (!normFn)
+    {
+        reportError(UnificationTooComplex{}, constraint->location);
+        return true;
+    }
+
+    // TODO: It would be nice to not need to convert the normalized type back to
+    // an intersection and flatten it.
+    TypeId normFnTy = normalizer->typeFromNormal(*normFn);
+    std::vector<TypeId> overloads = flattenIntersection(normFnTy);
 
     Instantiation inst(TxnLog::empty(), arena, TypeLevel{}, constraint->scope);
 
+    std::vector<TypeId> arityMatchingOverloads;
+
     for (TypeId overload : overloads)
     {
         overload = follow(overload);
@@ -1247,8 +1262,17 @@ bool ConstraintSolver::tryDispatch(const FunctionCallConstraint& c, NotNull<cons
         if (const auto& e = hasUnificationTooComplex(u.errors))
             reportError(*e);
 
+        if (const auto& e = hasCountMismatch(u.errors);
+            (!e || get<CountMismatch>(*e)->context != CountMismatch::Context::Arg) && get<FunctionType>(*instantiated))
+        {
+            arityMatchingOverloads.push_back(*instantiated);
+        }
+
         if (u.errors.empty())
         {
+            if (c.callSite)
+                (*c.astOverloadResolvedTypes)[c.callSite] = *instantiated;
+
             // We found a matching overload.
             const auto [changedTypes, changedPacks] = u.log.getChanges();
             u.log.commit();
@@ -1260,6 +1284,15 @@ bool ConstraintSolver::tryDispatch(const FunctionCallConstraint& c, NotNull<cons
         }
     }
 
+    if (arityMatchingOverloads.size() == 1 && c.callSite)
+    {
+        // In the name of better error messages in the type checker, we provide
+        // it with an instantiated function signature that matched arity, but
+        // not the requisite subtyping requirements. This makes errors better in
+        // cases where only one overload fit from an arity perspective.
+        (*c.astOverloadResolvedTypes)[c.callSite] = arityMatchingOverloads.at(0);
+    }
+
     // We found no matching overloads.
     Unifier u{normalizer, Mode::Strict, constraint->scope, Location{}, Covariant};
     u.useScopes = true;
@@ -1267,8 +1300,6 @@ bool ConstraintSolver::tryDispatch(const FunctionCallConstraint& c, NotNull<cons
     u.tryUnify(inferredTy, builtinTypes->anyType);
     u.tryUnify(fn, builtinTypes->anyType);
 
-    LUAU_ASSERT(u.errors.empty()); // unifying with any should never fail
-
     const auto [changedTypes, changedPacks] = u.log.getChanges();
     u.log.commit();
 
@@ -2166,13 +2197,24 @@ void ConstraintSolver::unblock(NotNull<const Constraint> progressed)
 
 void ConstraintSolver::unblock(TypeId progressed)
 {
+    DenseHashSet<TypeId> seen{nullptr};
+
+    while (true)
+    {
+        if (seen.find(progressed))
+            iceReporter.ice("ConstraintSolver::unblock encountered a self-bound type!");
+        seen.insert(progressed);
+
         if (logger)
             logger->popBlock(progressed);
 
         unblock_(progressed);
 
         if (auto bt = get<BoundType>(progressed))
-        unblock(bt->boundTo);
+            progressed = bt->boundTo;
+        else
+            break;
+    }
 }
 
 void ConstraintSolver::unblock(TypePackId progressed)

Analysis/src/Frontend.cpp

@@ -21,6 +21,9 @@
 
 #include <algorithm>
 #include <chrono>
+#include <condition_variable>
+#include <exception>
+#include <mutex>
 #include <stdexcept>
 #include <string>
 
@@ -34,10 +37,36 @@ LUAU_FASTFLAGVARIABLE(DebugLuauDeferredConstraintResolution, false)
 LUAU_FASTFLAGVARIABLE(DebugLuauLogSolverToJson, false)
 LUAU_FASTFLAG(LuauRequirePathTrueModuleName)
 LUAU_FASTFLAGVARIABLE(DebugLuauReadWriteProperties, false)
+LUAU_FASTFLAGVARIABLE(LuauSplitFrontendProcessing, false)
 
 namespace Luau
 {
 
+struct BuildQueueItem
+{
+    ModuleName name;
+    ModuleName humanReadableName;
+
+    // Parameters
+    std::shared_ptr<SourceNode> sourceNode;
+    std::shared_ptr<SourceModule> sourceModule;
+    Config config;
+    ScopePtr environmentScope;
+    std::vector<RequireCycle> requireCycles;
+    FrontendOptions options;
+    bool recordJsonLog = false;
+
+    // Queue state
+    std::vector<size_t> reverseDeps;
+    int dirtyDependencies = 0;
+    bool processing = false;
+
+    // Result
+    std::exception_ptr exception;
+    ModulePtr module;
+    Frontend::Stats stats;
+};
+
 std::optional<Mode> parseMode(const std::vector<HotComment>& hotcomments)
 {
     for (const HotComment& hc : hotcomments)
@@ -220,7 +249,7 @@ namespace
 {
 
 static ErrorVec accumulateErrors(
-    const std::unordered_map<ModuleName, SourceNode>& sourceNodes, ModuleResolver& moduleResolver, const ModuleName& name)
+    const std::unordered_map<ModuleName, std::shared_ptr<SourceNode>>& sourceNodes, ModuleResolver& moduleResolver, const ModuleName& name)
 {
     std::unordered_set<ModuleName> seen;
     std::vector<ModuleName> queue{name};
@@ -240,7 +269,7 @@ static ErrorVec accumulateErrors(
         if (it == sourceNodes.end())
             continue;
 
-        const SourceNode& sourceNode = it->second;
+        const SourceNode& sourceNode = *it->second;
         queue.insert(queue.end(), sourceNode.requireSet.begin(), sourceNode.requireSet.end());
 
         // FIXME: If a module has a syntax error, we won't be able to re-report it here.
@@ -285,8 +314,8 @@ static void filterLintOptions(LintOptions& lintOptions, const std::vector<HotCom
 // For each such path, record the full path and the location of the require in the starting module.
 // Note that this is O(V^2) for a fully connected graph and produces O(V) paths of length O(V)
 // However, when the graph is acyclic, this is O(V), as well as when only the first cycle is needed (stopAtFirst=true)
-std::vector<RequireCycle> getRequireCycles(
+std::vector<RequireCycle> getRequireCycles(const FileResolver* resolver,
-    const FileResolver* resolver, const std::unordered_map<ModuleName, SourceNode>& sourceNodes, const SourceNode* start, bool stopAtFirst = false)
+    const std::unordered_map<ModuleName, std::shared_ptr<SourceNode>>& sourceNodes, const SourceNode* start, bool stopAtFirst = false)
 {
     std::vector<RequireCycle> result;
 
@@ -302,7 +331,7 @@ std::vector<RequireCycle> getRequireCycles(
         if (dit == sourceNodes.end())
             continue;
 
-        stack.push_back(&dit->second);
+        stack.push_back(dit->second.get());
 
         while (!stack.empty())
         {
@@ -343,7 +372,7 @@ std::vector<RequireCycle> getRequireCycles(
 
                     auto rit = sourceNodes.find(reqName);
                     if (rit != sourceNodes.end())
-                        stack.push_back(&rit->second);
+                        stack.push_back(rit->second.get());
                 }
             }
         }
@@ -389,6 +418,52 @@ Frontend::Frontend(FileResolver* fileResolver, ConfigResolver* configResolver, c
 }
 
 CheckResult Frontend::check(const ModuleName& name, std::optional<FrontendOptions> optionOverride)
+{
+    if (!FFlag::LuauSplitFrontendProcessing)
+        return check_DEPRECATED(name, optionOverride);
+
+    LUAU_TIMETRACE_SCOPE("Frontend::check", "Frontend");
+    LUAU_TIMETRACE_ARGUMENT("name", name.c_str());
+
+    FrontendOptions frontendOptions = optionOverride.value_or(options);
+
+    if (std::optional<CheckResult> result = getCheckResult(name, true, frontendOptions.forAutocomplete))
+        return std::move(*result);
+
+    std::vector<ModuleName> buildQueue;
+    bool cycleDetected = parseGraph(buildQueue, name, frontendOptions.forAutocomplete);
+
+    std::unordered_set<Luau::ModuleName> seen;
+    std::vector<BuildQueueItem> buildQueueItems;
+    addBuildQueueItems(buildQueueItems, buildQueue, cycleDetected, seen, frontendOptions);
+    LUAU_ASSERT(!buildQueueItems.empty());
+
+    if (FFlag::DebugLuauLogSolverToJson)
+    {
+        LUAU_ASSERT(buildQueueItems.back().name == name);
+        buildQueueItems.back().recordJsonLog = true;
+    }
+
+    checkBuildQueueItems(buildQueueItems);
+
+    // Collect results only for checked modules, 'getCheckResult' produces a different result
+    CheckResult checkResult;
+
+    for (const BuildQueueItem& item : buildQueueItems)
+    {
+        if (item.module->timeout)
+            checkResult.timeoutHits.push_back(item.name);
+
+        checkResult.errors.insert(checkResult.errors.end(), item.module->errors.begin(), item.module->errors.end());
+
+        if (item.name == name)
+            checkResult.lintResult = item.module->lintResult;
+    }
+
+    return checkResult;
+}
+
+CheckResult Frontend::check_DEPRECATED(const ModuleName& name, std::optional<FrontendOptions> optionOverride)
 {
     LUAU_TIMETRACE_SCOPE("Frontend::check", "Frontend");
     LUAU_TIMETRACE_ARGUMENT("name", name.c_str());
@@ -399,7 +474,7 @@ CheckResult Frontend::check(const ModuleName& name, std::optional<FrontendOption
     FrontendModuleResolver& resolver = frontendOptions.forAutocomplete ? moduleResolverForAutocomplete : moduleResolver;
 
     auto it = sourceNodes.find(name);
-    if (it != sourceNodes.end() && !it->second.hasDirtyModule(frontendOptions.forAutocomplete))
+    if (it != sourceNodes.end() && !it->second->hasDirtyModule(frontendOptions.forAutocomplete))
     {
         // No recheck required.
         ModulePtr module = resolver.getModule(name);
@@ -421,13 +496,13 @@ CheckResult Frontend::check(const ModuleName& name, std::optional<FrontendOption
     for (const ModuleName& moduleName : buildQueue)
     {
         LUAU_ASSERT(sourceNodes.count(moduleName));
-        SourceNode& sourceNode = sourceNodes[moduleName];
+        SourceNode& sourceNode = *sourceNodes[moduleName];
 
         if (!sourceNode.hasDirtyModule(frontendOptions.forAutocomplete))
             continue;
 
         LUAU_ASSERT(sourceModules.count(moduleName));
-        SourceModule& sourceModule = sourceModules[moduleName];
+        SourceModule& sourceModule = *sourceModules[moduleName];
 
         const Config& config = configResolver->getConfig(moduleName);
 
@@ -583,7 +658,241 @@ CheckResult Frontend::check(const ModuleName& name, std::optional<FrontendOption
     return checkResult;
 }
 
-bool Frontend::parseGraph(std::vector<ModuleName>& buildQueue, const ModuleName& root, bool forAutocomplete)
+void Frontend::queueModuleCheck(const std::vector<ModuleName>& names)
+{
+    moduleQueue.insert(moduleQueue.end(), names.begin(), names.end());
+}
+
+void Frontend::queueModuleCheck(const ModuleName& name)
+{
+    moduleQueue.push_back(name);
+}
+
+std::vector<ModuleName> Frontend::checkQueuedModules(std::optional<FrontendOptions> optionOverride,
+    std::function<void(std::function<void()> task)> executeTask, std::function<void(size_t done, size_t total)> progress)
+{
+    FrontendOptions frontendOptions = optionOverride.value_or(options);
+
+    // By taking data into locals, we make sure queue is cleared at the end, even if an ICE or a different exception is thrown
+    std::vector<ModuleName> currModuleQueue;
+    std::swap(currModuleQueue, moduleQueue);
+
+    std::unordered_set<Luau::ModuleName> seen;
+    std::vector<BuildQueueItem> buildQueueItems;
+
+    for (const ModuleName& name : currModuleQueue)
+    {
+        if (seen.count(name))
+            continue;
+
+        if (!isDirty(name, frontendOptions.forAutocomplete))
+        {
+            seen.insert(name);
+            continue;
+        }
+
+        std::vector<ModuleName> queue;
+        bool cycleDetected = parseGraph(queue, name, frontendOptions.forAutocomplete, [&seen](const ModuleName& name) {
+            return seen.count(name);
+        });
+
+        addBuildQueueItems(buildQueueItems, queue, cycleDetected, seen, frontendOptions);
+    }
+
+    if (buildQueueItems.empty())
+        return {};
+
+    // We need a mapping from modules to build queue slots
+    std::unordered_map<ModuleName, size_t> moduleNameToQueue;
+
+    for (size_t i = 0; i < buildQueueItems.size(); i++)
+    {
+        BuildQueueItem& item = buildQueueItems[i];
+        moduleNameToQueue[item.name] = i;
+    }
+
+    // Default task execution is single-threaded and immediate
+    if (!executeTask)
+    {
+        executeTask = [](std::function<void()> task) {
+            task();
+        };
+    }
+
+    std::mutex mtx;
+    std::condition_variable cv;
+    std::vector<size_t> readyQueueItems;
+
+    size_t processing = 0;
+    size_t remaining = buildQueueItems.size();
+
+    auto itemTask = [&](size_t i) {
+        BuildQueueItem& item = buildQueueItems[i];
+
+        try
+        {
+            checkBuildQueueItem(item);
+        }
+        catch (...)
+        {
+            item.exception = std::current_exception();
+        }
+
+        {
+            std::unique_lock guard(mtx);
+            readyQueueItems.push_back(i);
+        }
+
+        cv.notify_one();
+    };
+
+    auto sendItemTask = [&](size_t i) {
+        BuildQueueItem& item = buildQueueItems[i];
+
+        item.processing = true;
+        processing++;
+
+        executeTask([&itemTask, i]() {
+            itemTask(i);
+        });
+    };
+
+    auto sendCycleItemTask = [&] {
+        for (size_t i = 0; i < buildQueueItems.size(); i++)
+        {
+            BuildQueueItem& item = buildQueueItems[i];
+
+            if (!item.processing)
+            {
+                sendItemTask(i);
+                break;
+            }
+        }
+    };
+
+    // In a first pass, check modules that have no dependencies and record info of those modules that wait
+    for (size_t i = 0; i < buildQueueItems.size(); i++)
+    {
+        BuildQueueItem& item = buildQueueItems[i];
+
+        for (const ModuleName& dep : item.sourceNode->requireSet)
+        {
+            if (auto it = sourceNodes.find(dep); it != sourceNodes.end())
+            {
+                if (it->second->hasDirtyModule(frontendOptions.forAutocomplete))
+                {
+                    item.dirtyDependencies++;
+
+                    buildQueueItems[moduleNameToQueue[dep]].reverseDeps.push_back(i);
+                }
+            }
+        }
+
+        if (item.dirtyDependencies == 0)
+            sendItemTask(i);
+    }
+
+    // Not a single item was found, a cycle in the graph was hit
+    if (processing == 0)
+        sendCycleItemTask();
+
+    std::vector<size_t> nextItems;
+
+    while (remaining != 0)
+    {
+        {
+            std::unique_lock guard(mtx);
+
+            // If nothing is ready yet, wait
+            if (readyQueueItems.empty())
+            {
+                cv.wait(guard, [&readyQueueItems] {
+                    return !readyQueueItems.empty();
+                });
+            }
+
+            // Handle checked items
+            for (size_t i : readyQueueItems)
+            {
+                const BuildQueueItem& item = buildQueueItems[i];
+                recordItemResult(item);
+
+                // Notify items that were waiting for this dependency
+                for (size_t reverseDep : item.reverseDeps)
+                {
+                    BuildQueueItem& reverseDepItem = buildQueueItems[reverseDep];
+
+                    LUAU_ASSERT(reverseDepItem.dirtyDependencies != 0);
+                    reverseDepItem.dirtyDependencies--;
+
+                    // In case of a module cycle earlier, check if unlocked an item that was already processed
+                    if (!reverseDepItem.processing && reverseDepItem.dirtyDependencies == 0)
+                        nextItems.push_back(reverseDep);
+                }
+            }
+
+            LUAU_ASSERT(processing >= readyQueueItems.size());
+            processing -= readyQueueItems.size();
+
+            LUAU_ASSERT(remaining >= readyQueueItems.size());
+            remaining -= readyQueueItems.size();
+            readyQueueItems.clear();
+        }
+
+        if (progress)
+            progress(buildQueueItems.size() - remaining, buildQueueItems.size());
+
+        // Items cannot be submitted while holding the lock
+        for (size_t i : nextItems)
+            sendItemTask(i);
+        nextItems.clear();
+
+        // If we aren't done, but don't have anything processing, we hit a cycle
+        if (remaining != 0 && processing == 0)
+            sendCycleItemTask();
+    }
+
+    std::vector<ModuleName> checkedModules;
+    checkedModules.reserve(buildQueueItems.size());
+
+    for (size_t i = 0; i < buildQueueItems.size(); i++)
+        checkedModules.push_back(std::move(buildQueueItems[i].name));
+
+    return checkedModules;
+}
+
+std::optional<CheckResult> Frontend::getCheckResult(const ModuleName& name, bool accumulateNested, bool forAutocomplete)
+{
+    auto it = sourceNodes.find(name);
+
+    if (it == sourceNodes.end() || it->second->hasDirtyModule(forAutocomplete))
+        return std::nullopt;
+
+    auto& resolver = forAutocomplete ? moduleResolverForAutocomplete : moduleResolver;
+
+    ModulePtr module = resolver.getModule(name);
+
+    if (module == nullptr)
+        throw InternalCompilerError("Frontend does not have module: " + name, name);
+
+    CheckResult checkResult;
+
+    if (module->timeout)
+        checkResult.timeoutHits.push_back(name);
+
+    if (accumulateNested)
+        checkResult.errors = accumulateErrors(sourceNodes, resolver, name);
+    else
+        checkResult.errors.insert(checkResult.errors.end(), module->errors.begin(), module->errors.end());
+
+    // Get lint result only for top checked module
+    checkResult.lintResult = module->lintResult;
+
+    return checkResult;
+}
+
+bool Frontend::parseGraph(
+    std::vector<ModuleName>& buildQueue, const ModuleName& root, bool forAutocomplete, std::function<bool(const ModuleName&)> canSkip)
 {
     LUAU_TIMETRACE_SCOPE("Frontend::parseGraph", "Frontend");
     LUAU_TIMETRACE_ARGUMENT("root", root.c_str());
@@ -654,14 +963,18 @@ bool Frontend::parseGraph(std::vector<ModuleName>& buildQueue, const ModuleName&
                     // this relies on the fact that markDirty marks reverse-dependencies dirty as well
                     // thus if a node is not dirty, all its transitive deps aren't dirty, which means that they won't ever need
                     // to be built, *and* can't form a cycle with any nodes we did process.
-                    if (!it->second.hasDirtyModule(forAutocomplete))
+                    if (!it->second->hasDirtyModule(forAutocomplete))
+                        continue;
+
+                    // This module might already be in the outside build queue
+                    if (canSkip && canSkip(dep))
                         continue;
 
                     // note: this check is technically redundant *except* that getSourceNode has somewhat broken memoization
                     // calling getSourceNode twice in succession will reparse the file, since getSourceNode leaves dirty flag set
-                    if (seen.contains(&it->second))
+                    if (seen.contains(it->second.get()))
                     {
-                        stack.push_back(&it->second);
+                        stack.push_back(it->second.get());
                         continue;
                     }
                 }
@@ -681,6 +994,210 @@ bool Frontend::parseGraph(std::vector<ModuleName>& buildQueue, const ModuleName&
     return cyclic;
 }
 
+void Frontend::addBuildQueueItems(std::vector<BuildQueueItem>& items, std::vector<ModuleName>& buildQueue, bool cycleDetected,
+    std::unordered_set<Luau::ModuleName>& seen, const FrontendOptions& frontendOptions)
+{
+    LUAU_ASSERT(FFlag::LuauSplitFrontendProcessing);
+
+    for (const ModuleName& moduleName : buildQueue)
+    {
+        if (seen.count(moduleName))
+            continue;
+        seen.insert(moduleName);
+
+        LUAU_ASSERT(sourceNodes.count(moduleName));
+        std::shared_ptr<SourceNode>& sourceNode = sourceNodes[moduleName];
+
+        if (!sourceNode->hasDirtyModule(frontendOptions.forAutocomplete))
+            continue;
+
+        LUAU_ASSERT(sourceModules.count(moduleName));
+        std::shared_ptr<SourceModule>& sourceModule = sourceModules[moduleName];
+
+        BuildQueueItem data{moduleName, fileResolver->getHumanReadableModuleName(moduleName), sourceNode, sourceModule};
+
+        data.config = configResolver->getConfig(moduleName);
+        data.environmentScope = getModuleEnvironment(*sourceModule, data.config, frontendOptions.forAutocomplete);
+
+        Mode mode = sourceModule->mode.value_or(data.config.mode);
+
+        // in NoCheck mode we only need to compute the value of .cyclic for typeck
+        // in the future we could replace toposort with an algorithm that can flag cyclic nodes by itself
+        // however, for now getRequireCycles isn't expensive in practice on the cases we care about, and long term
+        // all correct programs must be acyclic so this code triggers rarely
+        if (cycleDetected)
+            data.requireCycles = getRequireCycles(fileResolver, sourceNodes, sourceNode.get(), mode == Mode::NoCheck);
+
+        data.options = frontendOptions;
+
+        // This is used by the type checker to replace the resulting type of cyclic modules with any
+        sourceModule->cyclic = !data.requireCycles.empty();
+
+        items.push_back(std::move(data));
+    }
+}
+
+void Frontend::checkBuildQueueItem(BuildQueueItem& item)
+{
+    LUAU_ASSERT(FFlag::LuauSplitFrontendProcessing);
+
+    SourceNode& sourceNode = *item.sourceNode;
+    const SourceModule& sourceModule = *item.sourceModule;
+    const Config& config = item.config;
+    Mode mode = sourceModule.mode.value_or(config.mode);
+    ScopePtr environmentScope = item.environmentScope;
+    double timestamp = getTimestamp();
+    const std::vector<RequireCycle>& requireCycles = item.requireCycles;
+
+    if (item.options.forAutocomplete)
+    {
+        double autocompleteTimeLimit = FInt::LuauAutocompleteCheckTimeoutMs / 1000.0;
+
+        // The autocomplete typecheck is always in strict mode with DM awareness
+        // to provide better type information for IDE features
+        TypeCheckLimits typeCheckLimits;
+
+        if (autocompleteTimeLimit != 0.0)
+            typeCheckLimits.finishTime = TimeTrace::getClock() + autocompleteTimeLimit;
+        else
+            typeCheckLimits.finishTime = std::nullopt;
+
+        // TODO: This is a dirty ad hoc solution for autocomplete timeouts
+        // We are trying to dynamically adjust our existing limits to lower total typechecking time under the limit
+        // so that we'll have type information for the whole file at lower quality instead of a full abort in the middle
+        if (FInt::LuauTarjanChildLimit > 0)
+            typeCheckLimits.instantiationChildLimit = std::max(1, int(FInt::LuauTarjanChildLimit * sourceNode.autocompleteLimitsMult));
+        else
+            typeCheckLimits.instantiationChildLimit = std::nullopt;
+
+        if (FInt::LuauTypeInferIterationLimit > 0)
+            typeCheckLimits.unifierIterationLimit = std::max(1, int(FInt::LuauTypeInferIterationLimit * sourceNode.autocompleteLimitsMult));
+        else
+            typeCheckLimits.unifierIterationLimit = std::nullopt;
+
+        ModulePtr moduleForAutocomplete = check(sourceModule, Mode::Strict, requireCycles, environmentScope, /*forAutocomplete*/ true,
+            /*recordJsonLog*/ false, typeCheckLimits);
+
+        double duration = getTimestamp() - timestamp;
+
+        if (moduleForAutocomplete->timeout)
+            sourceNode.autocompleteLimitsMult = sourceNode.autocompleteLimitsMult / 2.0;
+        else if (duration < autocompleteTimeLimit / 2.0)
+            sourceNode.autocompleteLimitsMult = std::min(sourceNode.autocompleteLimitsMult * 2.0, 1.0);
+
+        item.stats.timeCheck += duration;
+        item.stats.filesStrict += 1;
+
+        item.module = moduleForAutocomplete;
+        return;
+    }
+
+    ModulePtr module = check(sourceModule, mode, requireCycles, environmentScope, /*forAutocomplete*/ false, item.recordJsonLog, {});
+
+    item.stats.timeCheck += getTimestamp() - timestamp;
+    item.stats.filesStrict += mode == Mode::Strict;
+    item.stats.filesNonstrict += mode == Mode::Nonstrict;
+
+    if (module == nullptr)
+        throw InternalCompilerError("Frontend::check produced a nullptr module for " + item.name, item.name);
+
+    if (FFlag::DebugLuauDeferredConstraintResolution && mode == Mode::NoCheck)
+        module->errors.clear();
+
+    if (item.options.runLintChecks)
+    {
+        LUAU_TIMETRACE_SCOPE("lint", "Frontend");
+
+        LintOptions lintOptions = item.options.enabledLintWarnings.value_or(config.enabledLint);
+        filterLintOptions(lintOptions, sourceModule.hotcomments, mode);
+
+        double timestamp = getTimestamp();
+
+        std::vector<LintWarning> warnings =
+            Luau::lint(sourceModule.root, *sourceModule.names, environmentScope, module.get(), sourceModule.hotcomments, lintOptions);
+
+        item.stats.timeLint += getTimestamp() - timestamp;
+
+        module->lintResult = classifyLints(warnings, config);
+    }
+
+    if (!item.options.retainFullTypeGraphs)
+    {
+        // copyErrors needs to allocate into interfaceTypes as it copies
+        // types out of internalTypes, so we unfreeze it here.
+        unfreeze(module->interfaceTypes);
+        copyErrors(module->errors, module->interfaceTypes);
+        freeze(module->interfaceTypes);
+
+        module->internalTypes.clear();
+
+        module->astTypes.clear();
+        module->astTypePacks.clear();
+        module->astExpectedTypes.clear();
+        module->astOriginalCallTypes.clear();
+        module->astOverloadResolvedTypes.clear();
+        module->astResolvedTypes.clear();
+        module->astOriginalResolvedTypes.clear();
+        module->astResolvedTypePacks.clear();
+        module->astScopes.clear();
+
+        module->scopes.clear();
+    }
+
+    if (mode != Mode::NoCheck)
+    {
+        for (const RequireCycle& cyc : requireCycles)
+        {
+            TypeError te{cyc.location, item.name, ModuleHasCyclicDependency{cyc.path}};
+
+            module->errors.push_back(te);
+        }
+    }
+
+    ErrorVec parseErrors;
+
+    for (const ParseError& pe : sourceModule.parseErrors)
+        parseErrors.push_back(TypeError{pe.getLocation(), item.name, SyntaxError{pe.what()}});
+
+    module->errors.insert(module->errors.begin(), parseErrors.begin(), parseErrors.end());
+
+    item.module = module;
+}
+
+void Frontend::checkBuildQueueItems(std::vector<BuildQueueItem>& items)
+{
+    LUAU_ASSERT(FFlag::LuauSplitFrontendProcessing);
+
+    for (BuildQueueItem& item : items)
+    {
+        checkBuildQueueItem(item);
+        recordItemResult(item);
+    }
+}
+
+void Frontend::recordItemResult(const BuildQueueItem& item)
+{
+    if (item.exception)
+        std::rethrow_exception(item.exception);
+
+    if (item.options.forAutocomplete)
+    {
+        moduleResolverForAutocomplete.setModule(item.name, item.module);
+        item.sourceNode->dirtyModuleForAutocomplete = false;
+    }
+    else
+    {
+        moduleResolver.setModule(item.name, item.module);
+        item.sourceNode->dirtyModule = false;
+    }
+
+    stats.timeCheck += item.stats.timeCheck;
+    stats.timeLint += item.stats.timeLint;
+
+    stats.filesStrict += item.stats.filesStrict;
+    stats.filesNonstrict += item.stats.filesNonstrict;
+}
+
 ScopePtr Frontend::getModuleEnvironment(const SourceModule& module, const Config& config, bool forAutocomplete) const
 {
     ScopePtr result;
@@ -711,7 +1228,7 @@ ScopePtr Frontend::getModuleEnvironment(const SourceModule& module, const Config
 bool Frontend::isDirty(const ModuleName& name, bool forAutocomplete) const
 {
     auto it = sourceNodes.find(name);
-    return it == sourceNodes.end() || it->second.hasDirtyModule(forAutocomplete);
+    return it == sourceNodes.end() || it->second->hasDirtyModule(forAutocomplete);
 }
 
 /*
@@ -728,7 +1245,7 @@ void Frontend::markDirty(const ModuleName& name, std::vector<ModuleName>* marked
     std::unordered_map<ModuleName, std::vector<ModuleName>> reverseDeps;
     for (const auto& module : sourceNodes)
     {
-        for (const auto& dep : module.second.requireSet)
+        for (const auto& dep : module.second->requireSet)
             reverseDeps[dep].push_back(module.first);
     }
 
@@ -740,7 +1257,7 @@ void Frontend::markDirty(const ModuleName& name, std::vector<ModuleName>* marked
         queue.pop_back();
 
         LUAU_ASSERT(sourceNodes.count(next) > 0);
-        SourceNode& sourceNode = sourceNodes[next];
+        SourceNode& sourceNode = *sourceNodes[next];
 
         if (markedDirty)
             markedDirty->push_back(next);
@@ -766,7 +1283,7 @@ SourceModule* Frontend::getSourceModule(const ModuleName& moduleName)
 {
     auto it = sourceModules.find(moduleName);
     if (it != sourceModules.end())
-        return &it->second;
+        return it->second.get();
     else
         return nullptr;
 }
@@ -901,22 +1418,22 @@ ModulePtr Frontend::check(const SourceModule& sourceModule, Mode mode, std::vect
 // Read AST into sourceModules if necessary.  Trace require()s.  Report parse errors.
 std::pair<SourceNode*, SourceModule*> Frontend::getSourceNode(const ModuleName& name)
 {
-    LUAU_TIMETRACE_SCOPE("Frontend::getSourceNode", "Frontend");
-    LUAU_TIMETRACE_ARGUMENT("name", name.c_str());
-
     auto it = sourceNodes.find(name);
-    if (it != sourceNodes.end() && !it->second.hasDirtySourceModule())
+    if (it != sourceNodes.end() && !it->second->hasDirtySourceModule())
     {
         auto moduleIt = sourceModules.find(name);
         if (moduleIt != sourceModules.end())
-            return {&it->second, &moduleIt->second};
+            return {it->second.get(), moduleIt->second.get()};
         else
         {
             LUAU_ASSERT(!"Everything in sourceNodes should also be in sourceModules");
-            return {&it->second, nullptr};
+            return {it->second.get(), nullptr};
         }
     }
 
+    LUAU_TIMETRACE_SCOPE("Frontend::getSourceNode", "Frontend");
+    LUAU_TIMETRACE_ARGUMENT("name", name.c_str());
+
     double timestamp = getTimestamp();
 
     std::optional<SourceCode> source = fileResolver->readSource(name);
@@ -939,30 +1456,37 @@ std::pair<SourceNode*, SourceModule*> Frontend::getSourceNode(const ModuleName&
     RequireTraceResult& require = requireTrace[name];
     require = traceRequires(fileResolver, result.root, name);
 
-    SourceNode& sourceNode = sourceNodes[name];
+    std::shared_ptr<SourceNode>& sourceNode = sourceNodes[name];
-    SourceModule& sourceModule = sourceModules[name];
 
-    sourceModule = std::move(result);
+    if (!sourceNode)
-    sourceModule.environmentName = environmentName;
+        sourceNode = std::make_shared<SourceNode>();
 
-    sourceNode.name = sourceModule.name;
+    std::shared_ptr<SourceModule>& sourceModule = sourceModules[name];
-    sourceNode.humanReadableName = sourceModule.humanReadableName;
+
-    sourceNode.requireSet.clear();
+    if (!sourceModule)
-    sourceNode.requireLocations.clear();
+        sourceModule = std::make_shared<SourceModule>();
-    sourceNode.dirtySourceModule = false;
+
+    *sourceModule = std::move(result);
+    sourceModule->environmentName = environmentName;
+
+    sourceNode->name = sourceModule->name;
+    sourceNode->humanReadableName = sourceModule->humanReadableName;
+    sourceNode->requireSet.clear();
+    sourceNode->requireLocations.clear();
+    sourceNode->dirtySourceModule = false;
 
     if (it == sourceNodes.end())
     {
-        sourceNode.dirtyModule = true;
+        sourceNode->dirtyModule = true;
-        sourceNode.dirtyModuleForAutocomplete = true;
+        sourceNode->dirtyModuleForAutocomplete = true;
     }
 
     for (const auto& [moduleName, location] : require.requireList)
-        sourceNode.requireSet.insert(moduleName);
+        sourceNode->requireSet.insert(moduleName);
 
-    sourceNode.requireLocations = require.requireList;
+    sourceNode->requireLocations = require.requireList;
 
-    return {&sourceNode, &sourceModule};
+    return {sourceNode.get(), sourceModule.get()};
 }
 
 /** Try to parse a source file into a SourceModule.

Analysis/src/Normalize.cpp

@@ -17,8 +17,6 @@ LUAU_FASTFLAGVARIABLE(DebugLuauCheckNormalizeInvariant, false)
 // This could theoretically be 2000 on amd64, but x86 requires this.
 LUAU_FASTINTVARIABLE(LuauNormalizeIterationLimit, 1200);
 LUAU_FASTINTVARIABLE(LuauNormalizeCacheLimit, 100000);
-LUAU_FASTFLAGVARIABLE(LuauNegatedClassTypes, false);
-LUAU_FASTFLAGVARIABLE(LuauNegatedTableTypes, false);
 LUAU_FASTFLAGVARIABLE(LuauNormalizeBlockedTypes, false);
 LUAU_FASTFLAGVARIABLE(LuauNormalizeMetatableFixes, false);
 LUAU_FASTFLAG(DebugLuauDeferredConstraintResolution)
@@ -232,15 +230,8 @@ NormalizedType::NormalizedType(NotNull<BuiltinTypes> builtinTypes)
 
 static bool isShallowInhabited(const NormalizedType& norm)
 {
-    bool inhabitedClasses;
-
-    if (FFlag::LuauNegatedClassTypes)
-        inhabitedClasses = !norm.classes.isNever();
-    else
-        inhabitedClasses = !norm.DEPRECATED_classes.empty();
-
     // This test is just a shallow check, for example it returns `true` for `{ p : never }`
-    return !get<NeverType>(norm.tops) || !get<NeverType>(norm.booleans) || inhabitedClasses || !get<NeverType>(norm.errors) ||
+    return !get<NeverType>(norm.tops) || !get<NeverType>(norm.booleans) || !norm.classes.isNever() || !get<NeverType>(norm.errors) ||
            !get<NeverType>(norm.nils) || !get<NeverType>(norm.numbers) || !norm.strings.isNever() || !get<NeverType>(norm.threads) ||
            !norm.functions.isNever() || !norm.tables.empty() || !norm.tyvars.empty();
 }
@@ -257,14 +248,8 @@ bool Normalizer::isInhabited(const NormalizedType* norm, std::unordered_set<Type
     if (!norm)
         return true;
 
-    bool inhabitedClasses;
-    if (FFlag::LuauNegatedClassTypes)
-        inhabitedClasses = !norm->classes.isNever();
-    else
-        inhabitedClasses = !norm->DEPRECATED_classes.empty();
-
     if (!get<NeverType>(norm->tops) || !get<NeverType>(norm->booleans) || !get<NeverType>(norm->errors) || !get<NeverType>(norm->nils) ||
-        !get<NeverType>(norm->numbers) || !get<NeverType>(norm->threads) || inhabitedClasses || !norm->strings.isNever() ||
+        !get<NeverType>(norm->numbers) || !get<NeverType>(norm->threads) || !norm->classes.isNever() || !norm->strings.isNever() ||
         !norm->functions.isNever())
         return true;
 
@@ -466,7 +451,7 @@ static bool areNormalizedTables(const TypeIds& tys)
         if (!pt)
             return false;
 
-        if (pt->type == PrimitiveType::Table && FFlag::LuauNegatedTableTypes)
+        if (pt->type == PrimitiveType::Table)
             continue;
 
         return false;
@@ -475,14 +460,6 @@ static bool areNormalizedTables(const TypeIds& tys)
     return true;
 }
 
-static bool areNormalizedClasses(const TypeIds& tys)
-{
-    for (TypeId ty : tys)
-        if (!get<ClassType>(ty))
-            return false;
-    return true;
-}
-
 static bool areNormalizedClasses(const NormalizedClassType& tys)
 {
     for (const auto& [ty, negations] : tys.classes)
@@ -567,7 +544,6 @@ static void assertInvariant(const NormalizedType& norm)
 
     LUAU_ASSERT(isNormalizedTop(norm.tops));
     LUAU_ASSERT(isNormalizedBoolean(norm.booleans));
-    LUAU_ASSERT(areNormalizedClasses(norm.DEPRECATED_classes));
     LUAU_ASSERT(areNormalizedClasses(norm.classes));
     LUAU_ASSERT(isNormalizedError(norm.errors));
     LUAU_ASSERT(isNormalizedNil(norm.nils));
@@ -629,7 +605,6 @@ void Normalizer::clearNormal(NormalizedType& norm)
     norm.tops = builtinTypes->neverType;
     norm.booleans = builtinTypes->neverType;
     norm.classes.resetToNever();
-    norm.DEPRECATED_classes.clear();
     norm.errors = builtinTypes->neverType;
     norm.nils = builtinTypes->neverType;
     norm.numbers = builtinTypes->neverType;
@@ -1252,8 +1227,6 @@ void Normalizer::unionTablesWithTable(TypeIds& heres, TypeId there)
 void Normalizer::unionTables(TypeIds& heres, const TypeIds& theres)
 {
     for (TypeId there : theres)
-    {
-        if (FFlag::LuauNegatedTableTypes)
     {
         if (there == builtinTypes->tableType)
         {
@@ -1266,11 +1239,6 @@ void Normalizer::unionTables(TypeIds& heres, const TypeIds& theres)
             unionTablesWithTable(heres, there);
         }
     }
-        else
-        {
-            unionTablesWithTable(heres, there);
-        }
-    }
 }
 
 // So why `ignoreSmallerTyvars`?
@@ -1320,10 +1288,7 @@ bool Normalizer::unionNormals(NormalizedType& here, const NormalizedType& there,
     }
 
     here.booleans = unionOfBools(here.booleans, there.booleans);
-    if (FFlag::LuauNegatedClassTypes)
     unionClasses(here.classes, there.classes);
-    else
-        unionClasses(here.DEPRECATED_classes, there.DEPRECATED_classes);
 
     here.errors = (get<NeverType>(there.errors) ? here.errors : there.errors);
     here.nils = (get<NeverType>(there.nils) ? here.nils : there.nils);
@@ -1414,16 +1379,7 @@ bool Normalizer::unionNormalWithTy(NormalizedType& here, TypeId there, int ignor
     else if (get<TableType>(there) || get<MetatableType>(there))
         unionTablesWithTable(here.tables, there);
     else if (get<ClassType>(there))
-    {
-        if (FFlag::LuauNegatedClassTypes)
-        {
         unionClassesWithClass(here.classes, there);
-        }
-        else
-        {
-            unionClassesWithClass(here.DEPRECATED_classes, there);
-        }
-    }
     else if (get<ErrorType>(there))
         here.errors = there;
     else if (const PrimitiveType* ptv = get<PrimitiveType>(there))
@@ -1442,7 +1398,7 @@ bool Normalizer::unionNormalWithTy(NormalizedType& here, TypeId there, int ignor
         {
             here.functions.resetToTop();
         }
-        else if (ptv->type == PrimitiveType::Table && FFlag::LuauNegatedTableTypes)
+        else if (ptv->type == PrimitiveType::Table)
         {
             here.tables.clear();
             here.tables.insert(there);
@@ -1527,8 +1483,6 @@ std::optional<NormalizedType> Normalizer::negateNormal(const NormalizedType& her
             result.booleans = builtinTypes->trueType;
     }
 
-    if (FFlag::LuauNegatedClassTypes)
-    {
     if (here.classes.isNever())
     {
         resetToTop(builtinTypes, result.classes);
@@ -1553,11 +1507,6 @@ std::optional<NormalizedType> Normalizer::negateNormal(const NormalizedType& her
         if (!rootNegations.empty())
             result.classes.pushPair(builtinTypes->classType, rootNegations);
     }
-    }
-    else
-    {
-        result.DEPRECATED_classes = negateAll(here.DEPRECATED_classes);
-    }
 
     result.nils = get<NeverType>(here.nils) ? builtinTypes->nilType : builtinTypes->neverType;
     result.numbers = get<NeverType>(here.numbers) ? builtinTypes->numberType : builtinTypes->neverType;
@@ -1584,15 +1533,12 @@ std::optional<NormalizedType> Normalizer::negateNormal(const NormalizedType& her
      * types are not runtime-testable. Thus, we prohibit negation of anything
      * other than `table` and `never`.
      */
-    if (FFlag::LuauNegatedTableTypes)
-    {
     if (here.tables.empty())
         result.tables.insert(builtinTypes->tableType);
     else if (here.tables.size() == 1 && here.tables.front() == builtinTypes->tableType)
         result.tables.clear();
     else
         return std::nullopt;
-    }
 
     // TODO: negating tables
     // TODO: negating tyvars?
@@ -1662,7 +1608,6 @@ void Normalizer::subtractPrimitive(NormalizedType& here, TypeId ty)
         here.functions.resetToNever();
         break;
     case PrimitiveType::Table:
-        LUAU_ASSERT(FFlag::LuauNegatedTableTypes);
         here.tables.clear();
         break;
     }
@@ -1734,64 +1679,6 @@ TypeId Normalizer::intersectionOfBools(TypeId here, TypeId there)
         return there;
 }
 
-void Normalizer::DEPRECATED_intersectClasses(TypeIds& heres, const TypeIds& theres)
-{
-    TypeIds tmp;
-    for (auto it = heres.begin(); it != heres.end();)
-    {
-        const ClassType* hctv = get<ClassType>(*it);
-        LUAU_ASSERT(hctv);
-        bool keep = false;
-        for (TypeId there : theres)
-        {
-            const ClassType* tctv = get<ClassType>(there);
-            LUAU_ASSERT(tctv);
-            if (isSubclass(hctv, tctv))
-            {
-                keep = true;
-                break;
-            }
-            else if (isSubclass(tctv, hctv))
-            {
-                keep = false;
-                tmp.insert(there);
-                break;
-            }
-        }
-        if (keep)
-            it++;
-        else
-            it = heres.erase(it);
-    }
-    heres.insert(tmp.begin(), tmp.end());
-}
-
-void Normalizer::DEPRECATED_intersectClassesWithClass(TypeIds& heres, TypeId there)
-{
-    bool foundSuper = false;
-    const ClassType* tctv = get<ClassType>(there);
-    LUAU_ASSERT(tctv);
-    for (auto it = heres.begin(); it != heres.end();)
-    {
-        const ClassType* hctv = get<ClassType>(*it);
-        LUAU_ASSERT(hctv);
-        if (isSubclass(hctv, tctv))
-            it++;
-        else if (isSubclass(tctv, hctv))
-        {
-            foundSuper = true;
-            break;
-        }
-        else
-            it = heres.erase(it);
-    }
-    if (foundSuper)
-    {
-        heres.clear();
-        heres.insert(there);
-    }
-}
-
 void Normalizer::intersectClasses(NormalizedClassType& heres, const NormalizedClassType& theres)
 {
     if (theres.isNever())
@@ -2504,15 +2391,7 @@ bool Normalizer::intersectNormals(NormalizedType& here, const NormalizedType& th
 
     here.booleans = intersectionOfBools(here.booleans, there.booleans);
 
-    if (FFlag::LuauNegatedClassTypes)
-    {
     intersectClasses(here.classes, there.classes);
-    }
-    else
-    {
-        DEPRECATED_intersectClasses(here.DEPRECATED_classes, there.DEPRECATED_classes);
-    }
-
     here.errors = (get<NeverType>(there.errors) ? there.errors : here.errors);
     here.nils = (get<NeverType>(there.nils) ? there.nils : here.nils);
     here.numbers = (get<NeverType>(there.numbers) ? there.numbers : here.numbers);
@@ -2618,22 +2497,12 @@ bool Normalizer::intersectNormalWithTy(NormalizedType& here, TypeId there)
         here.tables = std::move(tables);
     }
     else if (get<ClassType>(there))
-    {
-        if (FFlag::LuauNegatedClassTypes)
     {
         NormalizedClassType nct = std::move(here.classes);
         clearNormal(here);
         intersectClassesWithClass(nct, there);
         here.classes = std::move(nct);
     }
-        else
-        {
-            TypeIds classes = std::move(here.DEPRECATED_classes);
-            clearNormal(here);
-            DEPRECATED_intersectClassesWithClass(classes, there);
-            here.DEPRECATED_classes = std::move(classes);
-        }
-    }
     else if (get<ErrorType>(there))
     {
         TypeId errors = here.errors;
@@ -2665,10 +2534,7 @@ bool Normalizer::intersectNormalWithTy(NormalizedType& here, TypeId there)
         else if (ptv->type == PrimitiveType::Function)
             here.functions = std::move(functions);
         else if (ptv->type == PrimitiveType::Table)
-        {
-            LUAU_ASSERT(FFlag::LuauNegatedTableTypes);
             here.tables = std::move(tables);
-        }
         else
             LUAU_ASSERT(!"Unreachable");
     }
@@ -2696,7 +2562,7 @@ bool Normalizer::intersectNormalWithTy(NormalizedType& here, TypeId there)
             subtractPrimitive(here, ntv->ty);
         else if (const SingletonType* stv = get<SingletonType>(t))
             subtractSingleton(here, follow(ntv->ty));
-        else if (get<ClassType>(t) && FFlag::LuauNegatedClassTypes)
+        else if (get<ClassType>(t))
         {
             const NormalizedType* normal = normalize(t);
             std::optional<NormalizedType> negated = negateNormal(*normal);
@@ -2730,7 +2596,7 @@ bool Normalizer::intersectNormalWithTy(NormalizedType& here, TypeId there)
             LUAU_ASSERT(!"Unimplemented");
         }
     }
-    else if (get<NeverType>(there) && FFlag::LuauNegatedClassTypes)
+    else if (get<NeverType>(there))
     {
         here.classes.resetToNever();
     }
@@ -2756,8 +2622,6 @@ TypeId Normalizer::typeFromNormal(const NormalizedType& norm)
     if (!get<NeverType>(norm.booleans))
         result.push_back(norm.booleans);
 
-    if (FFlag::LuauNegatedClassTypes)
-    {
     if (isTop(builtinTypes, norm.classes))
     {
         result.push_back(builtinTypes->classType);
@@ -2799,11 +2663,6 @@ TypeId Normalizer::typeFromNormal(const NormalizedType& norm)
             result.push_back(arena->addType(UnionType{std::move(parts)}));
         }
     }
-    }
-    else
-    {
-        result.insert(result.end(), norm.DEPRECATED_classes.begin(), norm.DEPRECATED_classes.end());
-    }
 
     if (!get<NeverType>(norm.errors))
         result.push_back(norm.errors);

Analysis/src/TxnLog.cpp

@@ -382,8 +382,9 @@ std::optional<TypeLevel> TxnLog::getLevel(TypeId ty) const
 
 TypeId TxnLog::follow(TypeId ty) const
 {
-    return Luau::follow(ty, [this](TypeId ty) {
+    return Luau::follow(ty, this, [](const void* ctx, TypeId ty) -> TypeId {
-        PendingType* state = this->pending(ty);
+        const TxnLog* self = static_cast<const TxnLog*>(ctx);
+        PendingType* state = self->pending(ty);
 
         if (state == nullptr)
             return ty;
@@ -397,8 +398,9 @@ TypeId TxnLog::follow(TypeId ty) const
 
 TypePackId TxnLog::follow(TypePackId tp) const
 {
-    return Luau::follow(tp, [this](TypePackId tp) {
+    return Luau::follow(tp, this, [](const void* ctx, TypePackId tp) -> TypePackId {
-        PendingTypePack* state = this->pending(tp);
+        const TxnLog* self = static_cast<const TxnLog*>(ctx);
+        PendingTypePack* state = self->pending(tp);
 
         if (state == nullptr)
             return tp;

Analysis/src/Type.cpp

@@ -48,27 +48,8 @@ static std::optional<WithPredicate<TypePackId>> magicFunctionFind(
     TypeChecker& typechecker, const ScopePtr& scope, const AstExprCall& expr, WithPredicate<TypePackId> withPredicate);
 static bool dcrMagicFunctionFind(MagicFunctionCallContext context);
 
-TypeId follow(TypeId t)
+// LUAU_NOINLINE prevents unwrapLazy from being inlined into advance below; advance is important to keep inlineable
-{
+static LUAU_NOINLINE TypeId unwrapLazy(LazyType* ltv)
-    return follow(t, [](TypeId t) {
-        return t;
-    });
-}
-
-TypeId follow(TypeId t, std::function<TypeId(TypeId)> mapper)
-{
-    auto advance = [&mapper](TypeId ty) -> std::optional<TypeId> {
-        if (FFlag::LuauBoundLazyTypes2)
-        {
-            TypeId mapped = mapper(ty);
-
-            if (auto btv = get<Unifiable::Bound<TypeId>>(mapped))
-                return btv->boundTo;
-
-            if (auto ttv = get<TableType>(mapped))
-                return ttv->boundTo;
-
-            if (auto ltv = getMutable<LazyType>(mapped))
 {
     TypeId unwrapped = ltv->unwrapped.load();
 
@@ -87,21 +68,46 @@ TypeId follow(TypeId t, std::function<TypeId(TypeId)> mapper)
     return unwrapped;
 }
 
+TypeId follow(TypeId t)
+{
+    return follow(t, nullptr, [](const void*, TypeId t) -> TypeId {
+        return t;
+    });
+}
+
+TypeId follow(TypeId t, const void* context, TypeId (*mapper)(const void*, TypeId))
+{
+    auto advance = [context, mapper](TypeId ty) -> std::optional<TypeId> {
+        if (FFlag::LuauBoundLazyTypes2)
+        {
+            TypeId mapped = mapper(context, ty);
+
+            if (auto btv = get<Unifiable::Bound<TypeId>>(mapped))
+                return btv->boundTo;
+
+            if (auto ttv = get<TableType>(mapped))
+                return ttv->boundTo;
+
+            if (auto ltv = getMutable<LazyType>(mapped))
+                return unwrapLazy(ltv);
+
             return std::nullopt;
         }
         else
         {
-            if (auto btv = get<Unifiable::Bound<TypeId>>(mapper(ty)))
+            if (auto btv = get<Unifiable::Bound<TypeId>>(mapper(context, ty)))
                 return btv->boundTo;
-            else if (auto ttv = get<TableType>(mapper(ty)))
+            else if (auto ttv = get<TableType>(mapper(context, ty)))
                 return ttv->boundTo;
             else
                 return std::nullopt;
         }
     };
 
-    auto force = [&mapper](TypeId ty) {
+    auto force = [context, mapper](TypeId ty) {
-        if (auto ltv = get_if<LazyType>(&mapper(ty)->ty))
+        TypeId mapped = mapper(context, ty);
+
+        if (auto ltv = get_if<LazyType>(&mapped->ty))
         {
             TypeId res = ltv->thunk_DEPRECATED();
             if (get<LazyType>(res))
@@ -120,6 +126,12 @@ TypeId follow(TypeId t, std::function<TypeId(TypeId)> mapper)
     else
         return t;
 
+    if (FFlag::LuauBoundLazyTypes2)
+    {
+        if (!advance(cycleTester)) // Short circuit traversal for the rather common case when advance(advance(t)) == null
+            return cycleTester;
+    }
+
     while (true)
     {
         if (!FFlag::LuauBoundLazyTypes2)

Analysis/src/TypeChecker2.cpp

@@ -22,8 +22,6 @@
 LUAU_FASTFLAG(DebugLuauMagicTypes)
 LUAU_FASTFLAG(DebugLuauDontReduceTypes)
 
-LUAU_FASTFLAG(LuauNegatedClassTypes)
-
 namespace Luau
 {
 
@@ -519,18 +517,39 @@ struct TypeChecker2
             auto [minCount, maxCount] = getParameterExtents(TxnLog::empty(), iterFtv->argTypes, /*includeHiddenVariadics*/ true);
 
             if (minCount > 2)
-                reportError(CountMismatch{2, std::nullopt, minCount, CountMismatch::Arg}, forInStatement->vars.data[0]->location);
+            {
+                if (isMm)
+                    reportError(GenericError{"__iter metamethod must return (next[, table[, state]])"}, getLocation(forInStatement->values));
+                else
+                    reportError(GenericError{"for..in loops must be passed (next[, table[, state]])"}, getLocation(forInStatement->values));
+            }
             if (maxCount && *maxCount < 2)
-                reportError(CountMismatch{2, std::nullopt, *maxCount, CountMismatch::Arg}, forInStatement->vars.data[0]->location);
+            {
+                if (isMm)
+                    reportError(GenericError{"__iter metamethod must return (next[, table[, state]])"}, getLocation(forInStatement->values));
+                else
+                    reportError(GenericError{"for..in loops must be passed (next[, table[, state]])"}, getLocation(forInStatement->values));
+            }
 
             TypePack flattenedArgTypes = extendTypePack(arena, builtinTypes, iterFtv->argTypes, 2);
             size_t firstIterationArgCount = iterTys.empty() ? 0 : iterTys.size() - 1;
             size_t actualArgCount = expectedVariableTypes.head.size();
-
             if (firstIterationArgCount < minCount)
+            {
+                if (isMm)
+                    reportError(GenericError{"__iter metamethod must return (next[, table[, state]])"}, getLocation(forInStatement->values));
+                else
                     reportError(CountMismatch{2, std::nullopt, firstIterationArgCount, CountMismatch::Arg}, forInStatement->vars.data[0]->location);
+            }
+
             else if (actualArgCount < minCount)
-                reportError(CountMismatch{2, std::nullopt, actualArgCount, CountMismatch::Arg}, forInStatement->vars.data[0]->location);
+            {
+                if (isMm)
+                    reportError(GenericError{"__iter metamethod must return (next[, table[, state]])"}, getLocation(forInStatement->values));
+                else
+                    reportError(CountMismatch{2, std::nullopt, firstIterationArgCount, CountMismatch::Arg}, forInStatement->vars.data[0]->location);
+            }
+
 
             if (iterTys.size() >= 2 && flattenedArgTypes.head.size() > 0)
             {
@@ -841,125 +860,31 @@ struct TypeChecker2
         // TODO!
     }
 
-    ErrorVec visitOverload(AstExprCall* call, NotNull<const FunctionType> overloadFunctionType, const std::vector<Location>& argLocs,
-        TypePackId expectedArgTypes, TypePackId expectedRetType)
-    {
-        ErrorVec overloadErrors =
-            tryUnify(stack.back(), call->location, overloadFunctionType->retTypes, expectedRetType, CountMismatch::FunctionResult);
-
-        size_t argIndex = 0;
-        auto inferredArgIt = begin(overloadFunctionType->argTypes);
-        auto expectedArgIt = begin(expectedArgTypes);
-        while (inferredArgIt != end(overloadFunctionType->argTypes) && expectedArgIt != end(expectedArgTypes))
-        {
-            Location argLoc = (argIndex >= argLocs.size()) ? argLocs.back() : argLocs[argIndex];
-            ErrorVec argErrors = tryUnify(stack.back(), argLoc, *expectedArgIt, *inferredArgIt);
-            for (TypeError e : argErrors)
-                overloadErrors.emplace_back(e);
-
-            ++argIndex;
-            ++inferredArgIt;
-            ++expectedArgIt;
-        }
-
-        // piggyback on the unifier for arity checking, but we can't do this for checking the actual arguments since the locations would be bad
-        ErrorVec argumentErrors = tryUnify(stack.back(), call->location, expectedArgTypes, overloadFunctionType->argTypes);
-        for (TypeError e : argumentErrors)
-            if (get<CountMismatch>(e) != nullptr)
-                overloadErrors.emplace_back(std::move(e));
-
-        return overloadErrors;
-    }
-
-    void reportOverloadResolutionErrors(AstExprCall* call, std::vector<TypeId> overloads, TypePackId expectedArgTypes,
-        const std::vector<TypeId>& overloadsThatMatchArgCount, std::vector<std::pair<ErrorVec, TypeId>> overloadsErrors)
-    {
-        if (overloads.size() == 1)
-        {
-            reportErrors(std::get<0>(overloadsErrors.front()));
-            return;
-        }
-
-        std::vector<TypeId> overloadTypes = overloadsThatMatchArgCount;
-        if (overloadsThatMatchArgCount.size() == 0)
-        {
-            reportError(GenericError{"No overload for function accepts " + std::to_string(size(expectedArgTypes)) + " arguments."}, call->location);
-            // If no overloads match argument count, just list all overloads.
-            overloadTypes = overloads;
-        }
-        else
-        {
-            // Report errors of the first argument-count-matching, but failing overload
-            TypeId overload = overloadsThatMatchArgCount[0];
-
-            // Remove the overload we are reporting errors about from the list of alternatives
-            overloadTypes.erase(std::remove(overloadTypes.begin(), overloadTypes.end(), overload), overloadTypes.end());
-
-            const FunctionType* ftv = get<FunctionType>(overload);
-            LUAU_ASSERT(ftv); // overload must be a function type here
-
-            auto error = std::find_if(overloadsErrors.begin(), overloadsErrors.end(), [overload](const std::pair<ErrorVec, TypeId>& e) {
-                return overload == e.second;
-            });
-
-            LUAU_ASSERT(error != overloadsErrors.end());
-            reportErrors(std::get<0>(*error));
-
-            // If only one overload matched, we don't need this error because we provided the previous errors.
-            if (overloadsThatMatchArgCount.size() == 1)
-                return;
-        }
-
-        std::string s;
-        for (size_t i = 0; i < overloadTypes.size(); ++i)
-        {
-            TypeId overload = follow(overloadTypes[i]);
-
-            if (i > 0)
-                s += "; ";
-
-            if (i > 0 && i == overloadTypes.size() - 1)
-                s += "and ";
-
-            s += toString(overload);
-        }
-
-        if (overloadsThatMatchArgCount.size() == 0)
-            reportError(ExtraInformation{"Available overloads: " + s}, call->func->location);
-        else
-            reportError(ExtraInformation{"Other overloads are also not viable: " + s}, call->func->location);
-    }
-
     // Note: this is intentionally separated from `visit(AstExprCall*)` for stack allocation purposes.
     void visitCall(AstExprCall* call)
     {
-        TypeArena* arena = &testArena;
+        TypePackId expectedRetType = lookupExpectedPack(call, testArena);
-        Instantiation instantiation{TxnLog::empty(), arena, TypeLevel{}, stack.back()};
-
-        TypePackId expectedRetType = lookupExpectedPack(call, *arena);
-        TypeId functionType = lookupType(call->func);
-        TypeId testFunctionType = functionType;
         TypePack args;
         std::vector<Location> argLocs;
         argLocs.reserve(call->args.size + 1);
 
-        if (get<AnyType>(functionType) || get<ErrorType>(functionType) || get<NeverType>(functionType))
+        TypeId* maybeOriginalCallTy = module->astOriginalCallTypes.find(call);
+        TypeId* maybeSelectedOverload = module->astOverloadResolvedTypes.find(call);
+
+        if (!maybeOriginalCallTy)
             return;
-        else if (std::optional<TypeId> callMm = findMetatableEntry(builtinTypes, module->errors, functionType, "__call", call->func->location))
+
+        TypeId originalCallTy = follow(*maybeOriginalCallTy);
+        std::vector<TypeId> overloads = flattenIntersection(originalCallTy);
+
+        if (get<AnyType>(originalCallTy) || get<ErrorType>(originalCallTy) || get<NeverType>(originalCallTy))
+            return;
+        else if (std::optional<TypeId> callMm = findMetatableEntry(builtinTypes, module->errors, originalCallTy, "__call", call->func->location))
         {
             if (get<FunctionType>(follow(*callMm)))
             {
-                if (std::optional<TypeId> instantiatedCallMm = instantiation.substitute(*callMm))
+                args.head.push_back(originalCallTy);
-                {
-                    args.head.push_back(functionType);
                 argLocs.push_back(call->func->location);
-                    testFunctionType = follow(*instantiatedCallMm);
-                }
-                else
-                {
-                    reportError(UnificationTooComplex{}, call->func->location);
-                    return;
-                }
             }
             else
             {
@@ -969,29 +894,16 @@ struct TypeChecker2
                 return;
             }
         }
-        else if (get<FunctionType>(functionType))
+        else if (get<FunctionType>(originalCallTy) || get<IntersectionType>(originalCallTy))
         {
-            if (std::optional<TypeId> instantiatedFunctionType = instantiation.substitute(functionType))
-            {
-                testFunctionType = *instantiatedFunctionType;
         }
-            else
+        else if (auto utv = get<UnionType>(originalCallTy))
-            {
-                reportError(UnificationTooComplex{}, call->func->location);
-                return;
-            }
-        }
-        else if (auto itv = get<IntersectionType>(functionType))
-        {
-            // We do nothing here because we'll flatten the intersection later, but we don't want to report it as a non-function.
-        }
-        else if (auto utv = get<UnionType>(functionType))
         {
             // Sometimes it's okay to call a union of functions, but only if all of the functions are the same.
             // Another scenario we might run into it is if the union has a nil member. In this case, we want to throw an error
-            if (isOptional(functionType))
+            if (isOptional(originalCallTy))
             {
-                reportError(OptionalValueAccess{functionType}, call->location);
+                reportError(OptionalValueAccess{originalCallTy}, call->location);
                 return;
             }
             std::optional<TypeId> fst;
@@ -1001,7 +913,7 @@ struct TypeChecker2
                     fst = follow(ty);
                 else if (fst != follow(ty))
                 {
-                    reportError(CannotCallNonFunction{functionType}, call->func->location);
+                    reportError(CannotCallNonFunction{originalCallTy}, call->func->location);
                     return;
                 }
             }
@@ -1009,19 +921,16 @@ struct TypeChecker2
             if (!fst)
                 ice->ice("UnionType had no elements, so fst is nullopt?");
 
-            if (std::optional<TypeId> instantiatedFunctionType = instantiation.substitute(*fst))
+            originalCallTy = follow(*fst);
+            if (!get<FunctionType>(originalCallTy))
             {
-                testFunctionType = *instantiatedFunctionType;
+                reportError(CannotCallNonFunction{originalCallTy}, call->func->location);
-            }
-            else
-            {
-                reportError(UnificationTooComplex{}, call->func->location);
                 return;
             }
         }
         else
         {
-            reportError(CannotCallNonFunction{functionType}, call->func->location);
+            reportError(CannotCallNonFunction{originalCallTy}, call->func->location);
             return;
         }
 
@@ -1054,63 +963,134 @@ struct TypeChecker2
                 args.head.push_back(builtinTypes->anyType);
         }
 
-        TypePackId expectedArgTypes = arena->addTypePack(args);
+        TypePackId expectedArgTypes = testArena.addTypePack(args);
 
-        std::vector<TypeId> overloads = flattenIntersection(testFunctionType);
+        if (maybeSelectedOverload)
-        std::vector<std::pair<ErrorVec, TypeId>> overloadsErrors;
+        {
-        overloadsErrors.reserve(overloads.size());
+            // This overload might not work still: the constraint solver will
+            // pass the type checker an instantiated function type that matches
+            // in arity, but not in subtyping, in order to allow the type
+            // checker to report better error messages.
 
-        std::vector<TypeId> overloadsThatMatchArgCount;
+            TypeId selectedOverload = follow(*maybeSelectedOverload);
+            const FunctionType* ftv;
 
+            if (get<AnyType>(selectedOverload) || get<ErrorType>(selectedOverload) || get<NeverType>(selectedOverload))
+            {
+                return;
+            }
+            else if (const FunctionType* overloadFtv = get<FunctionType>(selectedOverload))
+            {
+                ftv = overloadFtv;
+            }
+            else
+            {
+                reportError(CannotCallNonFunction{selectedOverload}, call->func->location);
+                return;
+            }
+
+            LUAU_ASSERT(ftv);
+            reportErrors(tryUnify(stack.back(), call->location, ftv->retTypes, expectedRetType, CountMismatch::Context::Return));
+
+            auto it = begin(expectedArgTypes);
+            size_t i = 0;
+            std::vector<TypeId> slice;
+            for (TypeId arg : ftv->argTypes)
+            {
+                if (it == end(expectedArgTypes))
+                {
+                    slice.push_back(arg);
+                    continue;
+                }
+
+                TypeId expectedArg = *it;
+
+                Location argLoc = argLocs.at(i >= argLocs.size() ? argLocs.size() - 1 : i);
+
+                reportErrors(tryUnify(stack.back(), argLoc, expectedArg, arg));
+
+                ++it;
+                ++i;
+            }
+
+            if (slice.size() > 0 && it == end(expectedArgTypes))
+            {
+                if (auto tail = it.tail())
+                {
+                    TypePackId remainingArgs = testArena.addTypePack(TypePack{std::move(slice), std::nullopt});
+                    reportErrors(tryUnify(stack.back(), argLocs.back(), *tail, remainingArgs));
+                }
+            }
+
+            // We do not need to do an arity test because this overload was
+            // selected based on its arity already matching.
+        }
+        else
+        {
+            // No overload worked, even when instantiated. We need to filter the
+            // set of overloads to those that match the arity of the incoming
+            // argument set, and then report only those as not matching.
+
+            std::vector<TypeId> arityMatchingOverloads;
+            ErrorVec empty;
             for (TypeId overload : overloads)
             {
                 overload = follow(overload);
-
+                if (const FunctionType* ftv = get<FunctionType>(overload))
-            const FunctionType* overloadFn = get<FunctionType>(overload);
-            if (!overloadFn)
                 {
-                reportError(CannotCallNonFunction{overload}, call->func->location);
+                    if (size(ftv->argTypes) == size(expectedArgTypes))
-                return;
+                    {
+                        arityMatchingOverloads.push_back(overload);
+                    }
+                }
+                else if (const std::optional<TypeId> callMm = findMetatableEntry(builtinTypes, empty, overload, "__call", call->location))
+                {
+                    if (const FunctionType* ftv = get<FunctionType>(follow(*callMm)))
+                    {
+                        if (size(ftv->argTypes) == size(expectedArgTypes))
+                        {
+                            arityMatchingOverloads.push_back(overload);
+                        }
                     }
                     else
                     {
-                // We may have to instantiate the overload in order for it to typecheck.
+                        reportError(CannotCallNonFunction{}, call->location);
-                if (std::optional<TypeId> instantiatedFunctionType = instantiation.substitute(overload))
+                    }
+                }
+            }
+
+            if (arityMatchingOverloads.size() == 0)
             {
-                    overloadFn = get<FunctionType>(*instantiatedFunctionType);
+                reportError(
+                    GenericError{"No overload for function accepts " + std::to_string(size(expectedArgTypes)) + " arguments."}, call->location);
             }
             else
             {
-                    overloadsErrors.emplace_back(std::vector{TypeError{call->func->location, UnificationTooComplex{}}}, overload);
+                // We have handled the case of a singular arity-matching
-                    return;
+                // overload above, in the case where an overload was selected.
-                }
+                // LUAU_ASSERT(arityMatchingOverloads.size() > 1);
+                reportError(GenericError{"None of the overloads for function that accept " + std::to_string(size(expectedArgTypes)) +
+                                         " arguments are compatible."},
+                    call->location);
             }
 
-            ErrorVec overloadErrors = visitOverload(call, NotNull{overloadFn}, argLocs, expectedArgTypes, expectedRetType);
+            std::string s;
-            if (overloadErrors.empty())
+            std::vector<TypeId>& stringifyOverloads = arityMatchingOverloads.size() == 0 ? overloads : arityMatchingOverloads;
-                return;
+            for (size_t i = 0; i < stringifyOverloads.size(); ++i)
-
-            bool argMismatch = false;
-            for (auto error : overloadErrors)
             {
-                CountMismatch* cm = get<CountMismatch>(error);
+                TypeId overload = follow(stringifyOverloads[i]);
-                if (!cm)
-                    continue;
 
-                if (cm->context == CountMismatch::Arg)
+                if (i > 0)
-                {
+                    s += "; ";
-                    argMismatch = true;
+
-                    break;
+                if (i > 0 && i == stringifyOverloads.size() - 1)
-                }
+                    s += "and ";
+
+                s += toString(overload);
             }
 
-            if (!argMismatch)
+            reportError(ExtraInformation{"Available overloads: " + s}, call->func->location);
-                overloadsThatMatchArgCount.push_back(overload);
-
-            overloadsErrors.emplace_back(std::move(overloadErrors), overload);
         }
-
-        reportOverloadResolutionErrors(call, overloads, expectedArgTypes, overloadsThatMatchArgCount, overloadsErrors);
     }
 
     void visit(AstExprCall* call)
@@ -2077,18 +2057,10 @@ struct TypeChecker2
         fetch(norm->tops);
         fetch(norm->booleans);
 
-        if (FFlag::LuauNegatedClassTypes)
-        {
         for (const auto& [ty, _negations] : norm->classes.classes)
         {
             fetch(ty);
         }
-        }
-        else
-        {
-            for (TypeId ty : norm->DEPRECATED_classes)
-                fetch(ty);
-        }
         fetch(norm->errors);
         fetch(norm->nils);
         fetch(norm->numbers);

Analysis/src/TypeInfer.cpp

@@ -35,7 +35,6 @@ LUAU_FASTFLAG(LuauKnowsTheDataModel3)
 LUAU_FASTFLAGVARIABLE(DebugLuauFreezeDuringUnification, false)
 LUAU_FASTFLAGVARIABLE(DebugLuauSharedSelf, false)
 LUAU_FASTFLAG(LuauInstantiateInSubtyping)
-LUAU_FASTFLAG(LuauNegatedClassTypes)
 LUAU_FASTFLAGVARIABLE(LuauAllowIndexClassParameters, false)
 LUAU_FASTFLAG(LuauUninhabitedSubAnything2)
 LUAU_FASTFLAG(LuauOccursIsntAlwaysFailure)
@@ -1701,7 +1700,7 @@ void TypeChecker::prototype(const ScopePtr& scope, const AstStatTypeAlias& typea
 
 void TypeChecker::prototype(const ScopePtr& scope, const AstStatDeclareClass& declaredClass)
 {
-    std::optional<TypeId> superTy = FFlag::LuauNegatedClassTypes ? std::make_optional(builtinTypes->classType) : std::nullopt;
+    std::optional<TypeId> superTy = std::make_optional(builtinTypes->classType);
     if (declaredClass.superName)
     {
         Name superName = Name(declaredClass.superName->value);
@@ -5968,17 +5967,13 @@ void TypeChecker::resolve(const TypeGuardPredicate& typeguardP, RefinementMap& r
     TypeId type = follow(typeFun->type);
 
     // You cannot refine to the top class type.
-    if (FFlag::LuauNegatedClassTypes)
-    {
     if (type == builtinTypes->classType)
     {
         return addRefinement(refis, typeguardP.lvalue, errorRecoveryType(scope));
     }
-    }
 
     // We're only interested in the root class of any classes.
-    if (auto ctv = get<ClassType>(type);
+    if (auto ctv = get<ClassType>(type); !ctv || ctv->parent != builtinTypes->classType)
-        !ctv || (FFlag::LuauNegatedClassTypes ? (ctv->parent != builtinTypes->classType) : (ctv->parent != std::nullopt)))
         return addRefinement(refis, typeguardP.lvalue, errorRecoveryType(scope));
 
     // This probably hints at breaking out type filtering functions from the predicate solver so that typeof is not tightly coupled with IsA.

Analysis/src/TypePack.cpp

@@ -255,15 +255,17 @@ bool areEqual(SeenSet& seen, const TypePackVar& lhs, const TypePackVar& rhs)
 
 TypePackId follow(TypePackId tp)
 {
-    return follow(tp, [](TypePackId t) {
+    return follow(tp, nullptr, [](const void*, TypePackId t) {
         return t;
     });
 }
 
-TypePackId follow(TypePackId tp, std::function<TypePackId(TypePackId)> mapper)
+TypePackId follow(TypePackId tp, const void* context, TypePackId (*mapper)(const void*, TypePackId))
 {
-    auto advance = [&mapper](TypePackId ty) -> std::optional<TypePackId> {
+    auto advance = [context, mapper](TypePackId ty) -> std::optional<TypePackId> {
-        if (const Unifiable::Bound<TypePackId>* btv = get<Unifiable::Bound<TypePackId>>(mapper(ty)))
+        TypePackId mapped = mapper(context, ty);
+
+        if (const Unifiable::Bound<TypePackId>* btv = get<Unifiable::Bound<TypePackId>>(mapped))
             return btv->boundTo;
         else
             return std::nullopt;
@@ -275,6 +277,9 @@ TypePackId follow(TypePackId tp, std::function<TypePackId(TypePackId)> mapper)
     else
         return tp;
 
+    if (!advance(cycleTester)) // Short circuit traversal for the rather common case when advance(advance(t)) == null
+        return cycleTester;
+
     while (true)
     {
         auto a1 = advance(tp);

Analysis/src/Unifier.cpp

@@ -26,8 +26,6 @@ LUAU_FASTFLAGVARIABLE(LuauOccursIsntAlwaysFailure, false)
 LUAU_FASTFLAG(LuauClassTypeVarsInSubstitution)
 LUAU_FASTFLAG(DebugLuauDeferredConstraintResolution)
 LUAU_FASTFLAG(LuauNormalizeBlockedTypes)
-LUAU_FASTFLAG(LuauNegatedClassTypes)
-LUAU_FASTFLAG(LuauNegatedTableTypes)
 
 namespace Luau
 {
@@ -344,6 +342,19 @@ std::optional<TypeError> hasUnificationTooComplex(const ErrorVec& errors)
         return *it;
 }
 
+std::optional<TypeError> hasCountMismatch(const ErrorVec& errors)
+{
+    auto isCountMismatch = [](const TypeError& te) {
+        return nullptr != get<CountMismatch>(te);
+    };
+
+    auto it = std::find_if(errors.begin(), errors.end(), isCountMismatch);
+    if (it == errors.end())
+        return std::nullopt;
+    else
+        return *it;
+}
+
 // Used for tagged union matching heuristic, returns first singleton type field
 static std::optional<std::pair<Luau::Name, const SingletonType*>> getTableMatchTag(TypeId type)
 {
@@ -620,7 +631,7 @@ void Unifier::tryUnify_(TypeId subTy, TypeId superTy, bool isFunctionCall, bool
         // Ok.  Do nothing.  forall functions F, F <: function
     }
 
-    else if (FFlag::LuauNegatedTableTypes && isPrim(superTy, PrimitiveType::Table) && (get<TableType>(subTy) || get<MetatableType>(subTy)))
+    else if (isPrim(superTy, PrimitiveType::Table) && (get<TableType>(subTy) || get<MetatableType>(subTy)))
     {
         // Ok, do nothing: forall tables T, T <: table
     }
@@ -1183,8 +1194,6 @@ void Unifier::tryUnifyNormalizedTypes(
         if (!get<PrimitiveType>(superNorm.errors))
             return reportError(location, TypeMismatch{superTy, subTy, reason, error, mismatchContext()});
 
-    if (FFlag::LuauNegatedClassTypes)
-    {
     for (const auto& [subClass, _] : subNorm.classes.classes)
     {
         bool found = false;
@@ -1240,33 +1249,13 @@ void Unifier::tryUnifyNormalizedTypes(
             return reportError(location, TypeMismatch{superTy, subTy, reason, error, mismatchContext()});
         }
     }
-    }
-    else
-    {
-        for (TypeId subClass : subNorm.DEPRECATED_classes)
-        {
-            bool found = false;
-            const ClassType* subCtv = get<ClassType>(subClass);
-            for (TypeId superClass : superNorm.DEPRECATED_classes)
-            {
-                const ClassType* superCtv = get<ClassType>(superClass);
-                if (isSubclass(subCtv, superCtv))
-                {
-                    found = true;
-                    break;
-                }
-            }
-            if (!found)
-                return reportError(location, TypeMismatch{superTy, subTy, reason, error, mismatchContext()});
-        }
-    }
 
     for (TypeId subTable : subNorm.tables)
     {
         bool found = false;
         for (TypeId superTable : superNorm.tables)
         {
-            if (FFlag::LuauNegatedTableTypes && isPrim(superTable, PrimitiveType::Table))
+            if (isPrim(superTable, PrimitiveType::Table))
             {
                 found = true;
                 break;

Ast/include/Luau/ParseOptions.h

@@ -14,8 +14,6 @@ enum class Mode
 
 struct ParseOptions
 {
-    bool allowTypeAnnotations = true;
-    bool supportContinueStatement = true;
     bool allowDeclarationSyntax = false;
     bool captureComments = false;
 };

Ast/src/Parser.cpp

@@ -14,8 +14,6 @@
 LUAU_FASTINTVARIABLE(LuauRecursionLimit, 1000)
 LUAU_FASTINTVARIABLE(LuauParseErrorLimit, 100)
 
-LUAU_FASTFLAGVARIABLE(LuauParserErrorsOnMissingDefaultTypePackArgument, false)
-
 #define ERROR_INVALID_INTERP_DOUBLE_BRACE "Double braces are not permitted within interpolated strings. Did you mean '\\{'?"
 
 namespace Luau
@@ -327,8 +325,6 @@ AstStat* Parser::parseStat()
     // we know this isn't a call or an assignment; therefore it must be a context-sensitive keyword such as `type` or `continue`
     AstName ident = getIdentifier(expr);
 
-    if (options.allowTypeAnnotations)
-    {
     if (ident == "type")
         return parseTypeAlias(expr->location, /* exported= */ false);
 
@@ -337,12 +333,11 @@ AstStat* Parser::parseStat()
         nextLexeme();
         return parseTypeAlias(expr->location, /* exported= */ true);
     }
-    }
 
-    if (options.supportContinueStatement && ident == "continue")
+    if (ident == "continue")
         return parseContinue(expr->location);
 
-    if (options.allowTypeAnnotations && options.allowDeclarationSyntax)
+    if (options.allowDeclarationSyntax)
     {
         if (ident == "declare")
             return parseDeclaration(expr->location);
@@ -1123,7 +1118,7 @@ std::tuple<bool, Location, AstTypePack*> Parser::parseBindingList(TempVector<Bin
 
 AstType* Parser::parseOptionalType()
 {
-    if (options.allowTypeAnnotations && lexer.current().type == ':')
+    if (lexer.current().type == ':')
     {
         nextLexeme();
         return parseType();
@@ -1175,7 +1170,7 @@ AstTypePack* Parser::parseTypeList(TempVector<AstType*>& result, TempVector<std:
 
 std::optional<AstTypeList> Parser::parseOptionalReturnType()
 {
-    if (options.allowTypeAnnotations && (lexer.current().type == ':' || lexer.current().type == Lexeme::SkinnyArrow))
+    if (lexer.current().type == ':' || lexer.current().type == Lexeme::SkinnyArrow)
     {
         if (lexer.current().type == Lexeme::SkinnyArrow)
             report(lexer.current().location, "Function return type annotations are written after ':' instead of '->'");
@@ -2056,7 +2051,7 @@ AstExpr* Parser::parseAssertionExpr()
     Location start = lexer.current().location;
     AstExpr* expr = parseSimpleExpr();
 
-    if (options.allowTypeAnnotations && lexer.current().type == Lexeme::DoubleColon)
+    if (lexer.current().type == Lexeme::DoubleColon)
     {
         nextLexeme();
         AstType* annotation = parseType();
@@ -2449,24 +2444,13 @@ std::pair<AstArray<AstGenericType>, AstArray<AstGenericTypePack>> Parser::parseG
                     seenDefault = true;
                     nextLexeme();
 
-                    Lexeme packBegin = lexer.current();
-
                     if (shouldParseTypePack(lexer))
                     {
                         AstTypePack* typePack = parseTypePack();
 
                         namePacks.push_back({name, nameLocation, typePack});
                     }
-                    else if (!FFlag::LuauParserErrorsOnMissingDefaultTypePackArgument && lexer.current().type == '(')
+                    else
-                    {
-                        auto [type, typePack] = parseTypeOrPack();
-
-                        if (type)
-                            report(Location(packBegin.location.begin, lexer.previousLocation().end), "Expected type pack after '=', got type");
-
-                        namePacks.push_back({name, nameLocation, typePack});
-                    }
-                    else if (FFlag::LuauParserErrorsOnMissingDefaultTypePackArgument)
                     {
                         auto [type, typePack] = parseTypeOrPack();
 

CLI/Analyze.cpp

@@ -9,6 +9,13 @@
 #include "FileUtils.h"
 #include "Flags.h"
 
+#include <condition_variable>
+#include <functional>
+#include <mutex>
+#include <queue>
+#include <thread>
+#include <utility>
+
 #ifdef CALLGRIND
 #include <valgrind/callgrind.h>
 #endif
@@ -64,26 +71,29 @@ static void reportWarning(ReportFormat format, const char* name, const Luau::Lin
     report(format, name, warning.location, Luau::LintWarning::getName(warning.code), warning.text.c_str());
 }
 
-static bool analyzeFile(Luau::Frontend& frontend, const char* name, ReportFormat format, bool annotate)
+static bool reportModuleResult(Luau::Frontend& frontend, const Luau::ModuleName& name, ReportFormat format, bool annotate)
 {
-    Luau::CheckResult cr;
+    std::optional<Luau::CheckResult> cr = frontend.getCheckResult(name, false);
 
-    if (frontend.isDirty(name))
+    if (!cr)
-        cr = frontend.check(name);
-
-    if (!frontend.getSourceModule(name))
     {
-        fprintf(stderr, "Error opening %s\n", name);
+        fprintf(stderr, "Failed to find result for %s\n", name.c_str());
         return false;
     }
 
-    for (auto& error : cr.errors)
+    if (!frontend.getSourceModule(name))
+    {
+        fprintf(stderr, "Error opening %s\n", name.c_str());
+        return false;
+    }
+
+    for (auto& error : cr->errors)
         reportError(frontend, format, error);
 
     std::string humanReadableName = frontend.fileResolver->getHumanReadableModuleName(name);
-    for (auto& error : cr.lintResult.errors)
+    for (auto& error : cr->lintResult.errors)
         reportWarning(format, humanReadableName.c_str(), error);
-    for (auto& warning : cr.lintResult.warnings)
+    for (auto& warning : cr->lintResult.warnings)
         reportWarning(format, humanReadableName.c_str(), warning);
 
     if (annotate)
@@ -98,7 +108,7 @@ static bool analyzeFile(Luau::Frontend& frontend, const char* name, ReportFormat
         printf("%s", annotated.c_str());
     }
 
-    return cr.errors.empty() && cr.lintResult.errors.empty();
+    return cr->errors.empty() && cr->lintResult.errors.empty();
 }
 
 static void displayHelp(const char* argv0)
@@ -216,6 +226,70 @@ struct CliConfigResolver : Luau::ConfigResolver
     }
 };
 
+struct TaskScheduler
+{
+    TaskScheduler(unsigned threadCount)
+        : threadCount(threadCount)
+    {
+        for (unsigned i = 0; i < threadCount; i++)
+        {
+            workers.emplace_back([this] {
+                workerFunction();
+            });
+        }
+    }
+
+    ~TaskScheduler()
+    {
+        for (unsigned i = 0; i < threadCount; i++)
+            push({});
+
+        for (std::thread& worker : workers)
+            worker.join();
+    }
+
+    std::function<void()> pop()
+    {
+        std::unique_lock guard(mtx);
+
+        cv.wait(guard, [this] {
+            return !tasks.empty();
+        });
+
+        std::function<void()> task = tasks.front();
+        tasks.pop();
+        return task;
+    }
+
+    void push(std::function<void()> task)
+    {
+        {
+            std::unique_lock guard(mtx);
+            tasks.push(std::move(task));
+        }
+
+        cv.notify_one();
+    }
+
+    static unsigned getThreadCount()
+    {
+        return std::max(std::thread::hardware_concurrency(), 1u);
+    }
+
+private:
+    void workerFunction()
+    {
+        while (std::function<void()> task = pop())
+            task();
+    }
+
+    unsigned threadCount = 1;
+    std::mutex mtx;
+    std::condition_variable cv;
+    std::vector<std::thread> workers;
+    std::queue<std::function<void()>> tasks;
+};
+
 int main(int argc, char** argv)
 {
     Luau::assertHandler() = assertionHandler;
@@ -231,6 +305,7 @@ int main(int argc, char** argv)
     ReportFormat format = ReportFormat::Default;
     Luau::Mode mode = Luau::Mode::Nonstrict;
     bool annotate = false;
+    int threadCount = 0;
 
     for (int i = 1; i < argc; ++i)
     {
@@ -249,6 +324,8 @@ int main(int argc, char** argv)
             FFlag::DebugLuauTimeTracing.value = true;
         else if (strncmp(argv[i], "--fflags=", 9) == 0)
             setLuauFlags(argv[i] + 9);
+        else if (strncmp(argv[i], "-j", 2) == 0)
+            threadCount = strtol(argv[i] + 2, nullptr, 10);
     }
 
 #if !defined(LUAU_ENABLE_TIME_TRACE)
@@ -276,10 +353,28 @@ int main(int argc, char** argv)
 
     std::vector<std::string> files = getSourceFiles(argc, argv);
 
+    for (const std::string& path : files)
+        frontend.queueModuleCheck(path);
+
+    std::vector<Luau::ModuleName> checkedModules;
+
+    // If thread count is not set, try to use HW thread count, but with an upper limit
+    // When we improve scalability of typechecking, upper limit can be adjusted/removed
+    if (threadCount <= 0)
+        threadCount = std::min(TaskScheduler::getThreadCount(), 8u);
+
+    {
+        TaskScheduler scheduler(threadCount);
+
+        checkedModules = frontend.checkQueuedModules(std::nullopt, [&](std::function<void()> f) {
+            scheduler.push(std::move(f));
+        });
+    }
+
     int failed = 0;
 
-    for (const std::string& path : files)
+    for (const Luau::ModuleName& name : checkedModules)
-        failed += !analyzeFile(frontend, path.c_str(), format, annotate);
+        failed += !reportModuleResult(frontend, name, format, annotate);
 
     if (!configResolver.configErrors.empty())
     {

CLI/Ast.cpp

@@ -64,8 +64,6 @@ int main(int argc, char** argv)
 
     Luau::ParseOptions options;
     options.captureComments = true;
-    options.supportContinueStatement = true;
-    options.allowTypeAnnotations = true;
     options.allowDeclarationSyntax = true;
 
     Luau::ParseResult parseResult = Luau::Parser::parse(source.data(), source.size(), names, allocator, options);

CodeGen/include/Luau/IrAnalysis.h

@@ -35,6 +35,8 @@ struct RegisterSet
     uint8_t varargStart = 0;
 };
 
+void requireVariadicSequence(RegisterSet& sourceRs, const RegisterSet& defRs, uint8_t varargStart);
+
 struct CfgInfo
 {
     std::vector<uint32_t> predecessors;
@@ -43,10 +45,22 @@ struct CfgInfo
     std::vector<uint32_t> successors;
     std::vector<uint32_t> successorsOffsets;
 
+    // VM registers that are live when the block is entered
+    // Additionally, an active variadic sequence can exist at the entry of the block
     std::vector<RegisterSet> in;
+
+    // VM registers that are defined inside the block
+    // It can also contain a variadic sequence definition if that hasn't been consumed inside the block
+    // Note that this means that checking 'def' set might not be enough to say that register has not been written to
     std::vector<RegisterSet> def;
+
+    // VM registers that are coming out from the block
+    // These might be registers that are defined inside the block or have been defined at the entry of the block
+    // Additionally, an active variadic sequence can exist at the exit of the block
     std::vector<RegisterSet> out;
 
+    // VM registers captured by nested closures
+    // This set can never have an active variadic sequence
     RegisterSet captured;
 };
 

CodeGen/include/Luau/IrData.h

@@ -575,7 +575,7 @@ enum class IrCmd : uint8_t
     // Calls native libm function with 1 or 2 arguments
     // A: builtin function ID
     // B: double
-    // C: double (optional, 2nd argument)
+    // C: double/int (optional, 2nd argument)
     INVOKE_LIBM,
 };
 

CodeGen/include/Luau/IrDump.h

@@ -30,7 +30,7 @@ void toString(IrToStringContext& ctx, IrOp op);
 
 void toString(std::string& result, IrConst constant);
 
-void toStringDetailed(IrToStringContext& ctx, const IrInst& inst, uint32_t index, bool includeUseInfo);
+void toStringDetailed(IrToStringContext& ctx, const IrBlock& block, uint32_t blockIdx, const IrInst& inst, uint32_t instIdx, bool includeUseInfo);
 void toStringDetailed(IrToStringContext& ctx, const IrBlock& block, uint32_t index, bool includeUseInfo); // Block title
 
 std::string toString(const IrFunction& function, bool includeUseInfo);

CodeGen/include/Luau/IrUtils.h

@@ -114,6 +114,28 @@ inline bool isBlockTerminator(IrCmd cmd)
     return false;
 }
 
+inline bool isNonTerminatingJump(IrCmd cmd)
+{
+    switch (cmd)
+    {
+    case IrCmd::TRY_NUM_TO_INDEX:
+    case IrCmd::TRY_CALL_FASTGETTM:
+    case IrCmd::CHECK_FASTCALL_RES:
+    case IrCmd::CHECK_TAG:
+    case IrCmd::CHECK_READONLY:
+    case IrCmd::CHECK_NO_METATABLE:
+    case IrCmd::CHECK_SAFE_ENV:
+    case IrCmd::CHECK_ARRAY_SIZE:
+    case IrCmd::CHECK_SLOT_MATCH:
+    case IrCmd::CHECK_NODE_NO_NEXT:
+        return true;
+    default:
+        break;
+    }
+
+    return false;
+}
+
 inline bool hasResult(IrCmd cmd)
 {
     switch (cmd)

CodeGen/include/Luau/UnwindBuilder.h

@@ -1,8 +1,11 @@
 // This file is part of the Luau programming language and is licensed under MIT License; see LICENSE.txt for details
 #pragma once
 
+#include "Luau/RegisterA64.h"
 #include "Luau/RegisterX64.h"
 
+#include <initializer_list>
+
 #include <stddef.h>
 #include <stdint.h>
 
@@ -17,22 +20,36 @@ static uint32_t kFullBlockFuncton = ~0u;
 class UnwindBuilder
 {
 public:
+    enum Arch
+    {
+        X64,
+        A64
+    };
+
     virtual ~UnwindBuilder() = default;
 
     virtual void setBeginOffset(size_t beginOffset) = 0;
     virtual size_t getBeginOffset() const = 0;
 
-    virtual void startInfo() = 0;
+    virtual void startInfo(Arch arch) = 0;
-
     virtual void startFunction() = 0;
-    virtual void spill(int espOffset, X64::RegisterX64 reg) = 0;
-    virtual void save(X64::RegisterX64 reg) = 0;
-    virtual void allocStack(int size) = 0;
-    virtual void setupFrameReg(X64::RegisterX64 reg, int espOffset) = 0;
     virtual void finishFunction(uint32_t beginOffset, uint32_t endOffset) = 0;
-
     virtual void finishInfo() = 0;
 
+    // A64-specific; prologue must look like this:
+    //   sub sp, sp, stackSize
+    //   store sequence that saves regs to [sp..sp+regs.size*8) in the order specified in regs; regs should start with x29, x30 (fp, lr)
+    //   mov x29, sp
+    virtual void prologueA64(uint32_t prologueSize, uint32_t stackSize, std::initializer_list<A64::RegisterA64> regs) = 0;
+
+    // X64-specific; prologue must look like this:
+    //   optional, indicated by setupFrame:
+    //     push rbp
+    //     mov rbp, rsp
+    //   push reg in the order specified in regs
+    //   sub rsp, stackSize
+    virtual void prologueX64(uint32_t prologueSize, uint32_t stackSize, bool setupFrame, std::initializer_list<X64::RegisterX64> regs) = 0;
+
     virtual size_t getSize() const = 0;
     virtual size_t getFunctionCount() const = 0;
 

CodeGen/include/Luau/UnwindBuilderDwarf2.h

@@ -24,17 +24,14 @@ public:
     void setBeginOffset(size_t beginOffset) override;
     size_t getBeginOffset() const override;
 
-    void startInfo() override;
+    void startInfo(Arch arch) override;
-
     void startFunction() override;
-    void spill(int espOffset, X64::RegisterX64 reg) override;
-    void save(X64::RegisterX64 reg) override;
-    void allocStack(int size) override;
-    void setupFrameReg(X64::RegisterX64 reg, int espOffset) override;
     void finishFunction(uint32_t beginOffset, uint32_t endOffset) override;
-
     void finishInfo() override;
 
+    void prologueA64(uint32_t prologueSize, uint32_t stackSize, std::initializer_list<A64::RegisterA64> regs) override;
+    void prologueX64(uint32_t prologueSize, uint32_t stackSize, bool setupFrame, std::initializer_list<X64::RegisterX64> regs) override;
+
     size_t getSize() const override;
     size_t getFunctionCount() const override;
 
@@ -49,8 +46,6 @@ private:
     uint8_t rawData[kRawDataLimit];
     uint8_t* pos = rawData;
 
-    uint32_t stackOffset = 0;
-
     // We will remember the FDE location to write some of the fields like entry length, function start and size later
     uint8_t* fdeEntryStart = nullptr;
 };

CodeGen/include/Luau/UnwindBuilderWin.h

@@ -44,17 +44,14 @@ public:
     void setBeginOffset(size_t beginOffset) override;
     size_t getBeginOffset() const override;
 
-    void startInfo() override;
+    void startInfo(Arch arch) override;
-
     void startFunction() override;
-    void spill(int espOffset, X64::RegisterX64 reg) override;
-    void save(X64::RegisterX64 reg) override;
-    void allocStack(int size) override;
-    void setupFrameReg(X64::RegisterX64 reg, int espOffset) override;
     void finishFunction(uint32_t beginOffset, uint32_t endOffset) override;
-
     void finishInfo() override;
 
+    void prologueA64(uint32_t prologueSize, uint32_t stackSize, std::initializer_list<A64::RegisterA64> regs) override;
+    void prologueX64(uint32_t prologueSize, uint32_t stackSize, bool setupFrame, std::initializer_list<X64::RegisterX64> regs) override;
+
     size_t getSize() const override;
     size_t getFunctionCount() const override;
 
@@ -75,7 +72,6 @@ private:
     uint8_t prologSize = 0;
     X64::RegisterX64 frameReg = X64::noreg;
     uint8_t frameRegOffset = 0;
-    uint32_t stackOffset = 0;
 };
 
 } // namespace CodeGen

CodeGen/src/CodeBlockUnwind.cpp

@@ -22,12 +22,25 @@
 extern "C" void __register_frame(const void*);
 extern "C" void __deregister_frame(const void*);
 
+extern "C" void __unw_add_dynamic_fde() __attribute__((weak));
+
 #endif
 
-#if defined(__APPLE__)
+namespace Luau
-// On Mac, each FDE inside eh_frame section has to be handled separately
+{
+namespace CodeGen
+{
+
+#if !defined(_WIN32)
 static void visitFdeEntries(char* pos, void (*cb)(const void*))
 {
+    // When using glibc++ unwinder, we need to call __register_frame/__deregister_frame on the entire .eh_frame data
+    // When using libc++ unwinder (libunwind), each FDE has to be handled separately
+    // libc++ unwinder is the macOS unwinder, but on Linux the unwinder depends on the library the executable is linked with
+    // __unw_add_dynamic_fde is specific to libc++ unwinder, as such we determine the library based on its existence
+    if (__unw_add_dynamic_fde == nullptr)
+        return cb(pos);
+
     for (;;)
     {
         unsigned partLength;
@@ -47,11 +60,6 @@ static void visitFdeEntries(char* pos, void (*cb)(const void*))
 }
 #endif
 
-namespace Luau
-{
-namespace CodeGen
-{
-
 void* createBlockUnwindInfo(void* context, uint8_t* block, size_t blockSize, size_t& beginOffset)
 {
     UnwindBuilder* unwind = (UnwindBuilder*)context;
@@ -70,10 +78,8 @@ void* createBlockUnwindInfo(void* context, uint8_t* block, size_t blockSize, siz
         LUAU_ASSERT(!"failed to allocate function table");
         return nullptr;
     }
-#elif defined(__APPLE__)
-    visitFdeEntries(unwindData, __register_frame);
 #elif !defined(_WIN32)
-    __register_frame(unwindData);
+    visitFdeEntries(unwindData, __register_frame);
 #endif
 
     beginOffset = unwindSize + unwind->getBeginOffset();
@@ -85,10 +91,8 @@ void destroyBlockUnwindInfo(void* context, void* unwindData)
 #if defined(_WIN32) && defined(_M_X64)
     if (!RtlDeleteFunctionTable((RUNTIME_FUNCTION*)unwindData))
         LUAU_ASSERT(!"failed to deallocate function table");
-#elif defined(__APPLE__)
-    visitFdeEntries((char*)unwindData, __deregister_frame);
 #elif !defined(_WIN32)
-    __deregister_frame(unwindData);
+    visitFdeEntries((char*)unwindData, __deregister_frame);
 #endif
 }
 

CodeGen/src/CodeGen.cpp

@@ -134,7 +134,6 @@ static bool lowerImpl(AssemblyBuilder& build, IrLowering& lowering, IrFunction&
     for (size_t i = 0; i < sortedBlocks.size(); ++i)
     {
         uint32_t blockIndex = sortedBlocks[i];
-
         IrBlock& block = function.blocks[blockIndex];
 
         if (block.kind == IrBlockKind::Dead)
@@ -191,10 +190,13 @@ static bool lowerImpl(AssemblyBuilder& build, IrLowering& lowering, IrFunction&
                 continue;
             }
 
+            // Either instruction result value is not referenced or the use count is not zero
+            LUAU_ASSERT(inst.lastUse == 0 || inst.useCount != 0);
+
             if (options.includeIr)
             {
                 build.logAppend("# ");
-                toStringDetailed(ctx, inst, index, /* includeUseInfo */ true);
+                toStringDetailed(ctx, block, blockIndex, inst, index, /* includeUseInfo */ true);
             }
 
             IrBlock& next = i + 1 < sortedBlocks.size() ? function.blocks[sortedBlocks[i + 1]] : dummy;
@@ -409,9 +411,11 @@ bool isSupported()
     if (sizeof(LuaNode) != 32)
         return false;
 
-    // TODO: A64 codegen does not generate correct unwind info at the moment so it requires longjmp instead of C++ exceptions
+#ifdef _WIN32
+    // Unwind info is not supported for Windows-on-ARM yet
     if (!LUA_USE_LONGJMP)
         return false;
+#endif
 
     return true;
 #else

CodeGen/src/CodeGenA64.cpp

@@ -123,9 +123,6 @@ static EntryLocations buildEntryFunction(AssemblyBuilderA64& build, UnwindBuilde
     // Arguments: x0 = lua_State*, x1 = Proto*, x2 = native code pointer to jump to, x3 = NativeContext*
 
     locations.start = build.setLabel();
-    unwind.startFunction();
-
-    unwind.allocStack(8); // TODO: this is just a hack to make UnwindBuilder assertions cooperate
 
     // prologue
     build.sub(sp, sp, kStackSize);
@@ -140,6 +137,8 @@ static EntryLocations buildEntryFunction(AssemblyBuilderA64& build, UnwindBuilde
 
     locations.prologueEnd = build.setLabel();
 
+    uint32_t prologueSize = build.getLabelOffset(locations.prologueEnd) - build.getLabelOffset(locations.start);
+
     // Setup native execution environment
     build.mov(rState, x0);
     build.mov(rNativeContext, x3);
@@ -168,6 +167,8 @@ static EntryLocations buildEntryFunction(AssemblyBuilderA64& build, UnwindBuilde
     build.ret();
 
     // Our entry function is special, it spans the whole remaining code area
+    unwind.startFunction();
+    unwind.prologueA64(prologueSize, kStackSize, {x29, x30, x19, x20, x21, x22, x23, x24});
     unwind.finishFunction(build.getLabelOffset(locations.start), kFullBlockFuncton);
 
     return locations;
@@ -178,7 +179,7 @@ bool initHeaderFunctions(NativeState& data)
     AssemblyBuilderA64 build(/* logText= */ false);
     UnwindBuilder& unwind = *data.unwindBuilder.get();
 
-    unwind.startInfo();
+    unwind.startInfo(UnwindBuilder::A64);
 
     EntryLocations entryLocations = buildEntryFunction(build, unwind);
 

CodeGen/src/CodeGenX64.cpp

@@ -58,43 +58,44 @@ static EntryLocations buildEntryFunction(AssemblyBuilderX64& build, UnwindBuilde
     unwind.startFunction();
 
     // Save common non-volatile registers
-    build.push(rbp);
-    unwind.save(rbp);
-
     if (build.abi == ABIX64::SystemV)
     {
+        // We need to use a standard rbp-based frame setup for debuggers to work with JIT code
+        build.push(rbp);
         build.mov(rbp, rsp);
-        unwind.setupFrameReg(rbp, 0);
     }
 
     build.push(rbx);
-    unwind.save(rbx);
     build.push(r12);
-    unwind.save(r12);
     build.push(r13);
-    unwind.save(r13);
     build.push(r14);
-    unwind.save(r14);
     build.push(r15);
-    unwind.save(r15);
 
     if (build.abi == ABIX64::Windows)
     {
         // Save non-volatile registers that are specific to Windows x64 ABI
         build.push(rdi);
-        unwind.save(rdi);
         build.push(rsi);
-        unwind.save(rsi);
+
+        // On Windows, rbp is available as a general-purpose non-volatile register; we currently don't use it, but we need to push an even number
+        // of registers for stack alignment...
+        build.push(rbp);
 
         // TODO: once we start using non-volatile SIMD registers on Windows, we will save those here
     }
 
     // Allocate stack space (reg home area + local data)
     build.sub(rsp, kStackSize + kLocalsSize);
-    unwind.allocStack(kStackSize + kLocalsSize);
 
     locations.prologueEnd = build.setLabel();
 
+    uint32_t prologueSize = build.getLabelOffset(locations.prologueEnd) - build.getLabelOffset(locations.start);
+
+    if (build.abi == ABIX64::SystemV)
+        unwind.prologueX64(prologueSize, kStackSize + kLocalsSize, /* setupFrame= */ true, {rbx, r12, r13, r14, r15});
+    else if (build.abi == ABIX64::Windows)
+        unwind.prologueX64(prologueSize, kStackSize + kLocalsSize, /* setupFrame= */ false, {rbx, r12, r13, r14, r15, rdi, rsi, rbp});
+
     // Setup native execution environment
     build.mov(rState, rArg1);
     build.mov(rNativeContext, rArg4);
@@ -118,6 +119,7 @@ static EntryLocations buildEntryFunction(AssemblyBuilderX64& build, UnwindBuilde
 
     if (build.abi == ABIX64::Windows)
     {
+        build.pop(rbp);
         build.pop(rsi);
         build.pop(rdi);
     }
@@ -127,7 +129,10 @@ static EntryLocations buildEntryFunction(AssemblyBuilderX64& build, UnwindBuilde
     build.pop(r13);
     build.pop(r12);
     build.pop(rbx);
+
+    if (build.abi == ABIX64::SystemV)
         build.pop(rbp);
+
     build.ret();
 
     // Our entry function is special, it spans the whole remaining code area
@@ -141,7 +146,7 @@ bool initHeaderFunctions(NativeState& data)
     AssemblyBuilderX64 build(/* logText= */ false);
     UnwindBuilder& unwind = *data.unwindBuilder.get();
 
-    unwind.startInfo();
+    unwind.startInfo(UnwindBuilder::X64);
 
     EntryLocations entryLocations = buildEntryFunction(build, unwind);
 

CodeGen/src/EmitBuiltinsX64.cpp

@@ -18,19 +18,6 @@ namespace CodeGen
 namespace X64
 {
 
-static void emitBuiltinMathLdexp(IrRegAllocX64& regs, AssemblyBuilderX64& build, int ra, int arg, OperandX64 arg2)
-{
-    ScopedRegX64 tmp{regs, SizeX64::qword};
-    build.vcvttsd2si(tmp.reg, arg2);
-
-    IrCallWrapperX64 callWrap(regs, build);
-    callWrap.addArgument(SizeX64::xmmword, luauRegValue(arg));
-    callWrap.addArgument(SizeX64::qword, tmp);
-    callWrap.call(qword[rNativeContext + offsetof(NativeContext, libm_ldexp)]);
-
-    build.vmovsd(luauRegValue(ra), xmm0);
-}
-
 static void emitBuiltinMathFrexp(IrRegAllocX64& regs, AssemblyBuilderX64& build, int ra, int arg, int nresults)
 {
     IrCallWrapperX64 callWrap(regs, build);
@@ -115,9 +102,6 @@ void emitBuiltin(IrRegAllocX64& regs, AssemblyBuilderX64& build, int bfid, int r
 {
     switch (bfid)
     {
-    case LBF_MATH_LDEXP:
-        LUAU_ASSERT(nparams == 2 && nresults == 1);
-        return emitBuiltinMathLdexp(regs, build, ra, arg, arg2);
     case LBF_MATH_FREXP:
         LUAU_ASSERT(nparams == 1 && (nresults == 1 || nresults == 2));
         return emitBuiltinMathFrexp(regs, build, ra, arg, nresults);

CodeGen/src/IrAnalysis.cpp

@@ -162,7 +162,7 @@ uint32_t getLiveOutValueCount(IrFunction& function, IrBlock& block)
     return getLiveInOutValueCount(function, block).second;
 }
 
-static void requireVariadicSequence(RegisterSet& sourceRs, const RegisterSet& defRs, uint8_t varargStart)
+void requireVariadicSequence(RegisterSet& sourceRs, const RegisterSet& defRs, uint8_t varargStart)
 {
     if (!defRs.varargSeq)
     {

CodeGen/src/IrDump.cpp

@@ -62,6 +62,7 @@ static const char* getTagName(uint8_t tag)
     case LUA_TTHREAD:
         return "tthread";
     default:
+        LUAU_ASSERT(!"Unknown type tag");
         LUAU_UNREACHABLE();
     }
 }
@@ -410,27 +411,6 @@ void toString(std::string& result, IrConst constant)
     }
 }
 
-void toStringDetailed(IrToStringContext& ctx, const IrInst& inst, uint32_t index, bool includeUseInfo)
-{
-    size_t start = ctx.result.size();
-
-    toString(ctx, inst, index);
-
-    if (includeUseInfo)
-    {
-        padToDetailColumn(ctx.result, start);
-
-        if (inst.useCount == 0 && hasSideEffects(inst.cmd))
-            append(ctx.result, "; %%%u, has side-effects\n", index);
-        else
-            append(ctx.result, "; useCount: %d, lastUse: %%%u\n", inst.useCount, inst.lastUse);
-    }
-    else
-    {
-        ctx.result.append("\n");
-    }
-}
-
 static void appendBlockSet(IrToStringContext& ctx, BlockIteratorWrapper blocks)
 {
     bool comma = false;
@@ -470,6 +450,86 @@ static void appendRegisterSet(IrToStringContext& ctx, const RegisterSet& rs, con
     }
 }
 
+static RegisterSet getJumpTargetExtraLiveIn(IrToStringContext& ctx, const IrBlock& block, uint32_t blockIdx, const IrInst& inst)
+{
+    RegisterSet extraRs;
+
+    if (blockIdx >= ctx.cfg.in.size())
+        return extraRs;
+
+    const RegisterSet& defRs = ctx.cfg.in[blockIdx];
+
+    // Find first block argument, for guard instructions (isNonTerminatingJump), that's the first and only one
+    LUAU_ASSERT(isNonTerminatingJump(inst.cmd));
+    IrOp op = inst.a;
+
+    if (inst.b.kind == IrOpKind::Block)
+        op = inst.b;
+    else if (inst.c.kind == IrOpKind::Block)
+        op = inst.c;
+    else if (inst.d.kind == IrOpKind::Block)
+        op = inst.d;
+    else if (inst.e.kind == IrOpKind::Block)
+        op = inst.e;
+    else if (inst.f.kind == IrOpKind::Block)
+        op = inst.f;
+
+    if (op.kind == IrOpKind::Block && op.index < ctx.cfg.in.size())
+    {
+        const RegisterSet& inRs = ctx.cfg.in[op.index];
+
+        extraRs.regs = inRs.regs & ~defRs.regs;
+
+        if (inRs.varargSeq)
+            requireVariadicSequence(extraRs, defRs, inRs.varargStart);
+    }
+
+    return extraRs;
+}
+
+void toStringDetailed(IrToStringContext& ctx, const IrBlock& block, uint32_t blockIdx, const IrInst& inst, uint32_t instIdx, bool includeUseInfo)
+{
+    size_t start = ctx.result.size();
+
+    toString(ctx, inst, instIdx);
+
+    if (includeUseInfo)
+    {
+        padToDetailColumn(ctx.result, start);
+
+        if (inst.useCount == 0 && hasSideEffects(inst.cmd))
+        {
+            if (isNonTerminatingJump(inst.cmd))
+            {
+                RegisterSet extraRs = getJumpTargetExtraLiveIn(ctx, block, blockIdx, inst);
+
+                if (extraRs.regs.any() || extraRs.varargSeq)
+                {
+                    append(ctx.result, "; %%%u, extra in: ", instIdx);
+                    appendRegisterSet(ctx, extraRs, ", ");
+                    ctx.result.append("\n");
+                }
+                else
+                {
+                    append(ctx.result, "; %%%u\n", instIdx);
+                }
+            }
+            else
+            {
+                append(ctx.result, "; %%%u\n", instIdx);
+            }
+        }
+        else
+        {
+            append(ctx.result, "; useCount: %d, lastUse: %%%u\n", inst.useCount, inst.lastUse);
+        }
+    }
+    else
+    {
+        ctx.result.append("\n");
+    }
+}
+
 void toStringDetailed(IrToStringContext& ctx, const IrBlock& block, uint32_t index, bool includeUseInfo)
 {
     // Report captured registers for entry block
@@ -581,7 +641,7 @@ std::string toString(const IrFunction& function, bool includeUseInfo)
                 continue;
 
             append(ctx.result, " ");
-            toStringDetailed(ctx, inst, index, includeUseInfo);
+            toStringDetailed(ctx, block, uint32_t(i), inst, index, includeUseInfo);
         }
 
         append(ctx.result, "\n");

CodeGen/src/IrLoweringA64.cpp

@@ -122,42 +122,6 @@ static bool emitBuiltin(
 {
     switch (bfid)
     {
-    case LBF_MATH_LDEXP:
-        LUAU_ASSERT(nparams == 2 && nresults == 1);
-
-        if (args.kind == IrOpKind::VmReg)
-        {
-            build.ldr(d1, mem(rBase, args.index * sizeof(TValue) + offsetof(TValue, value.n)));
-            build.fcvtzs(w0, d1);
-        }
-        else if (args.kind == IrOpKind::VmConst)
-        {
-            size_t constantOffset = args.index * sizeof(TValue) + offsetof(TValue, value.n);
-
-            // Note: cumulative offset is guaranteed to be divisible by 8 (since we're loading a double); we can use that to expand the useful range
-            // that doesn't require temporaries
-            if (constantOffset / 8 <= AddressA64::kMaxOffset)
-            {
-                build.ldr(d1, mem(rConstants, int(constantOffset)));
-            }
-            else
-            {
-                emitAddOffset(build, x0, rConstants, constantOffset);
-                build.ldr(d1, x0);
-            }
-
-            build.fcvtzs(w0, d1);
-        }
-        else if (args.kind == IrOpKind::Constant)
-            build.mov(w0, int(function.doubleOp(args)));
-        else if (args.kind != IrOpKind::Undef)
-            LUAU_ASSERT(!"Unsupported instruction form");
-
-        build.ldr(d0, mem(rBase, arg * sizeof(TValue) + offsetof(TValue, value.n)));
-        build.ldr(x1, mem(rNativeContext, offsetof(NativeContext, libm_ldexp)));
-        build.blr(x1);
-        build.str(d0, mem(rBase, res * sizeof(TValue) + offsetof(TValue, value.n)));
-        return true;
     case LBF_MATH_FREXP:
         LUAU_ASSERT(nparams == 1 && (nresults == 1 || nresults == 2));
         emitInvokeLibm1P(build, offsetof(NativeContext, libm_frexp), arg);
@@ -1610,12 +1574,20 @@ void IrLoweringA64::lowerInst(IrInst& inst, uint32_t index, IrBlock& next)
     {
         if (inst.c.kind != IrOpKind::None)
         {
+            bool isInt = (inst.c.kind == IrOpKind::Constant) ? constOp(inst.c).kind == IrConstKind::Int
+                                                             : getCmdValueKind(function.instOp(inst.c).cmd) == IrValueKind::Int;
+
             RegisterA64 temp1 = tempDouble(inst.b);
-            RegisterA64 temp2 = tempDouble(inst.c);
+            RegisterA64 temp2 = isInt ? tempInt(inst.c) : tempDouble(inst.c);
-            RegisterA64 temp3 = regs.allocTemp(KindA64::d); // note: spill() frees all registers so we need to avoid alloc after spill
+            RegisterA64 temp3 = isInt ? noreg : regs.allocTemp(KindA64::d); // note: spill() frees all registers so we need to avoid alloc after spill
             regs.spill(build, index, {temp1, temp2});
 
-            if (d0 != temp2)
+            if (isInt)
+            {
+                build.fmov(d0, temp1);
+                build.mov(w0, temp2);
+            }
+            else if (d0 != temp2)
             {
                 build.fmov(d0, temp1);
                 build.fmov(d1, temp2);
@@ -1634,8 +1606,8 @@ void IrLoweringA64::lowerInst(IrInst& inst, uint32_t index, IrBlock& next)
             build.fmov(d0, temp1);
         }
 
-        build.ldr(x0, mem(rNativeContext, getNativeContextOffset(uintOp(inst.a))));
+        build.ldr(x1, mem(rNativeContext, getNativeContextOffset(uintOp(inst.a))));
-        build.blr(x0);
+        build.blr(x1);
         inst.regA64 = regs.takeReg(d0, index);
         break;
     }

CodeGen/src/IrLoweringX64.cpp

@@ -1304,7 +1304,15 @@ void IrLoweringX64::lowerInst(IrInst& inst, uint32_t index, IrBlock& next)
         callWrap.addArgument(SizeX64::xmmword, memRegDoubleOp(inst.b), inst.b);
 
         if (inst.c.kind != IrOpKind::None)
+        {
+            bool isInt = (inst.c.kind == IrOpKind::Constant) ? constOp(inst.c).kind == IrConstKind::Int
+                                                             : getCmdValueKind(function.instOp(inst.c).cmd) == IrValueKind::Int;
+
+            if (isInt)
+                callWrap.addArgument(SizeX64::dword, memRegUintOp(inst.c), inst.c);
+            else
                 callWrap.addArgument(SizeX64::xmmword, memRegDoubleOp(inst.c), inst.c);
+        }
 
         callWrap.call(qword[rNativeContext + getNativeContextOffset(uintOp(inst.a))]);
         inst.regX64 = regs.takeReg(xmm0, index);

CodeGen/src/IrTranslateBuiltins.cpp

@@ -71,23 +71,6 @@ static BuiltinImplResult translateBuiltinNumberToNumberLibm(
     return {BuiltinImplType::UsesFallback, 1};
 }
 
-// (number, number, ...) -> number
-static BuiltinImplResult translateBuiltin2NumberToNumber(
-    IrBuilder& build, LuauBuiltinFunction bfid, int nparams, int ra, int arg, IrOp args, int nresults, IrOp fallback)
-{
-    if (nparams < 2 || nresults > 1)
-        return {BuiltinImplType::None, -1};
-
-    builtinCheckDouble(build, build.vmReg(arg), fallback);
-    builtinCheckDouble(build, args, fallback);
-    build.inst(IrCmd::FASTCALL, build.constUint(bfid), build.vmReg(ra), build.vmReg(arg), args, build.constInt(2), build.constInt(1));
-
-    if (ra != arg)
-        build.inst(IrCmd::STORE_TAG, build.vmReg(ra), build.constTag(LUA_TNUMBER));
-
-    return {BuiltinImplType::UsesFallback, 1};
-}
-
 static BuiltinImplResult translateBuiltin2NumberToNumberLibm(
     IrBuilder& build, LuauBuiltinFunction bfid, int nparams, int ra, int arg, IrOp args, int nresults, IrOp fallback)
 {
@@ -110,6 +93,30 @@ static BuiltinImplResult translateBuiltin2NumberToNumberLibm(
     return {BuiltinImplType::UsesFallback, 1};
 }
 
+static BuiltinImplResult translateBuiltinMathLdexp(
+    IrBuilder& build, LuauBuiltinFunction bfid, int nparams, int ra, int arg, IrOp args, int nresults, IrOp fallback)
+{
+    if (nparams < 2 || nresults > 1)
+        return {BuiltinImplType::None, -1};
+
+    builtinCheckDouble(build, build.vmReg(arg), fallback);
+    builtinCheckDouble(build, args, fallback);
+
+    IrOp va = builtinLoadDouble(build, build.vmReg(arg));
+    IrOp vb = builtinLoadDouble(build, args);
+
+    IrOp vbi = build.inst(IrCmd::NUM_TO_INT, vb);
+
+    IrOp res = build.inst(IrCmd::INVOKE_LIBM, build.constUint(bfid), va, vbi);
+
+    build.inst(IrCmd::STORE_DOUBLE, build.vmReg(ra), res);
+
+    if (ra != arg)
+        build.inst(IrCmd::STORE_TAG, build.vmReg(ra), build.constTag(LUA_TNUMBER));
+
+    return {BuiltinImplType::UsesFallback, 1};
+}
+
 // (number, ...) -> (number, number)
 static BuiltinImplResult translateBuiltinNumberTo2Number(
     IrBuilder& build, LuauBuiltinFunction bfid, int nparams, int ra, int arg, IrOp args, int nresults, IrOp fallback)
@@ -778,7 +785,7 @@ BuiltinImplResult translateBuiltin(IrBuilder& build, int bfid, int ra, int arg,
     case LBF_MATH_ATAN2:
         return translateBuiltin2NumberToNumberLibm(build, LuauBuiltinFunction(bfid), nparams, ra, arg, args, nresults, fallback);
     case LBF_MATH_LDEXP:
-        return translateBuiltin2NumberToNumber(build, LuauBuiltinFunction(bfid), nparams, ra, arg, args, nresults, fallback);
+        return translateBuiltinMathLdexp(build, LuauBuiltinFunction(bfid), nparams, ra, arg, args, nresults, fallback);
     case LBF_MATH_FREXP:
     case LBF_MATH_MODF:
         return translateBuiltinNumberTo2Number(build, LuauBuiltinFunction(bfid), nparams, ra, arg, args, nresults, fallback);

CodeGen/src/IrUtils.cpp

@@ -299,6 +299,9 @@ void replace(IrFunction& function, IrBlock& block, uint32_t instIdx, IrInst repl
     removeUse(function, inst.e);
     removeUse(function, inst.f);
 
+    // Inherit existing use count (last use is skipped as it will be defined later)
+    replacement.useCount = inst.useCount;
+
     inst = replacement;
 
     // Removing the earlier extra reference, this might leave the block without users without marking it as dead
@@ -775,6 +778,8 @@ uint32_t getNativeContextOffset(int bfid)
         return offsetof(NativeContext, libm_pow);
     case LBF_IR_MATH_LOG2:
         return offsetof(NativeContext, libm_log2);
+    case LBF_MATH_LDEXP:
+        return offsetof(NativeContext, libm_ldexp);
     default:
         LUAU_ASSERT(!"Unsupported bfid");
     }

CodeGen/src/OptimizeConstProp.cpp

@@ -290,6 +290,20 @@ struct ConstPropState
         valueMap[versionedVmRegLoad(loadCmd, storeInst.a)] = storeInst.b.index;
     }
 
+    void clear()
+    {
+        for (int i = 0; i <= maxReg; ++i)
+            regs[i] = RegisterInfo();
+
+        maxReg = 0;
+
+        inSafeEnv = false;
+        checkedGc = false;
+
+        instLink.clear();
+        valueMap.clear();
+    }
+
     IrFunction& function;
 
     bool useValueNumbering = false;
@@ -854,12 +868,11 @@ static void constPropInBlock(IrBuilder& build, IrBlock& block, ConstPropState& s
     state.valueMap.clear();
 }
 
-static void constPropInBlockChain(IrBuilder& build, std::vector<uint8_t>& visited, IrBlock* block, bool useValueNumbering)
+static void constPropInBlockChain(IrBuilder& build, std::vector<uint8_t>& visited, IrBlock* block, ConstPropState& state)
 {
     IrFunction& function = build.function;
 
-    ConstPropState state{function};
+    state.clear();
-    state.useValueNumbering = useValueNumbering;
 
     while (block)
     {
@@ -936,7 +949,7 @@ static std::vector<uint32_t> collectDirectBlockJumpPath(IrFunction& function, st
     return path;
 }
 
-static void tryCreateLinearBlock(IrBuilder& build, std::vector<uint8_t>& visited, IrBlock& startingBlock, bool useValueNumbering)
+static void tryCreateLinearBlock(IrBuilder& build, std::vector<uint8_t>& visited, IrBlock& startingBlock, ConstPropState& state)
 {
     IrFunction& function = build.function;
 
@@ -965,8 +978,9 @@ static void tryCreateLinearBlock(IrBuilder& build, std::vector<uint8_t>& visited
         return;
 
     // Initialize state with the knowledge of our current block
-    ConstPropState state{function};
+    state.clear();
-    state.useValueNumbering = useValueNumbering;
+
+    // TODO: using values from the first block can cause 'live out' of the linear block predecessor to not have all required registers
     constPropInBlock(build, startingBlock, state);
 
     // Veryfy that target hasn't changed
@@ -981,10 +995,43 @@ static void tryCreateLinearBlock(IrBuilder& build, std::vector<uint8_t>& visited
 
     replace(function, termInst.a, newBlock);
 
-    // Clone the collected path int our fresh block
+    // Clone the collected path into our fresh block
     for (uint32_t pathBlockIdx : path)
         build.clone(function.blocks[pathBlockIdx], /* removeCurrentTerminator */ true);
 
+    // If all live in/out data is defined aside from the new block, generate it
+    // Note that liveness information is not strictly correct after optimization passes and may need to be recomputed before next passes
+    // The information generated here is consistent with current state that could be outdated, but still useful in IR inspection
+    if (function.cfg.in.size() == newBlock.index)
+    {
+        LUAU_ASSERT(function.cfg.in.size() == function.cfg.out.size());
+        LUAU_ASSERT(function.cfg.in.size() == function.cfg.def.size());
+
+        // Live in is the same as the input of the original first block
+        function.cfg.in.push_back(function.cfg.in[path.front()]);
+
+        // Live out is the same as the result of the original last block
+        function.cfg.out.push_back(function.cfg.out[path.back()]);
+
+        // Defs are tricky, registers are joined together, but variadic sequences can be consumed inside the block
+        function.cfg.def.push_back({});
+        RegisterSet& def = function.cfg.def.back();
+
+        for (uint32_t pathBlockIdx : path)
+        {
+            const RegisterSet& pathDef = function.cfg.def[pathBlockIdx];
+
+            def.regs |= pathDef.regs;
+
+            // Taking only the last defined variadic sequence if it's not consumed before before the end
+            if (pathDef.varargSeq && function.cfg.out.back().varargSeq)
+            {
+                def.varargSeq = true;
+                def.varargStart = pathDef.varargStart;
+            }
+        }
+    }
+
     // Optimize our linear block
     IrBlock& linearBlock = function.blockOp(newBlock);
     constPropInBlock(build, linearBlock, state);
@@ -994,6 +1041,9 @@ void constPropInBlockChains(IrBuilder& build, bool useValueNumbering)
 {
     IrFunction& function = build.function;
 
+    ConstPropState state{function};
+    state.useValueNumbering = useValueNumbering;
+
     std::vector<uint8_t> visited(function.blocks.size(), false);
 
     for (IrBlock& block : function.blocks)
@@ -1004,7 +1054,7 @@ void constPropInBlockChains(IrBuilder& build, bool useValueNumbering)
         if (visited[function.getBlockIndex(block)])
             continue;
 
-        constPropInBlockChain(build, visited, &block, useValueNumbering);
+        constPropInBlockChain(build, visited, &block, state);
     }
 }
 
@@ -1015,6 +1065,9 @@ void createLinearBlocks(IrBuilder& build, bool useValueNumbering)
     // new 'block' will only be reachable from a single one and all gathered information can be preserved.
     IrFunction& function = build.function;
 
+    ConstPropState state{function};
+    state.useValueNumbering = useValueNumbering;
+
     std::vector<uint8_t> visited(function.blocks.size(), false);
 
     // This loop can create new 'linear' blocks, so index-based loop has to be used (and it intentionally won't reach those new blocks)
@@ -1029,7 +1082,7 @@ void createLinearBlocks(IrBuilder& build, bool useValueNumbering)
         if (visited[function.getBlockIndex(block)])
             continue;
 
-        tryCreateLinearBlock(build, visited, block, useValueNumbering);
+        tryCreateLinearBlock(build, visited, block, state);
     }
 }
 

CodeGen/src/UnwindBuilderDwarf2.cpp

@@ -36,27 +36,25 @@
 #define DW_CFA_lo_user 0x1c
 #define DW_CFA_hi_user 0x3f
 
-// Register numbers for x64 (System V ABI, page 57, ch. 3.7, figure 3.36)
+// Register numbers for X64 (System V ABI, page 57, ch. 3.7, figure 3.36)
-#define DW_REG_RAX 0
+#define DW_REG_X64_RAX 0
-#define DW_REG_RDX 1
+#define DW_REG_X64_RDX 1
-#define DW_REG_RCX 2
+#define DW_REG_X64_RCX 2
-#define DW_REG_RBX 3
+#define DW_REG_X64_RBX 3
-#define DW_REG_RSI 4
+#define DW_REG_X64_RSI 4
-#define DW_REG_RDI 5
+#define DW_REG_X64_RDI 5
-#define DW_REG_RBP 6
+#define DW_REG_X64_RBP 6
-#define DW_REG_RSP 7
+#define DW_REG_X64_RSP 7
-#define DW_REG_R8 8
+#define DW_REG_X64_RA 16
-#define DW_REG_R9 9
-#define DW_REG_R10 10
-#define DW_REG_R11 11
-#define DW_REG_R12 12
-#define DW_REG_R13 13
-#define DW_REG_R14 14
-#define DW_REG_R15 15
-#define DW_REG_RA 16
 
-const int regIndexToDwRegX64[16] = {DW_REG_RAX, DW_REG_RCX, DW_REG_RDX, DW_REG_RBX, DW_REG_RSP, DW_REG_RBP, DW_REG_RSI, DW_REG_RDI, DW_REG_R8,
+// Register numbers for A64 (DWARF for the Arm 64-bit Architecture, ch. 4.1)
-    DW_REG_R9, DW_REG_R10, DW_REG_R11, DW_REG_R12, DW_REG_R13, DW_REG_R14, DW_REG_R15};
+#define DW_REG_A64_FP 29
+#define DW_REG_A64_LR 30
+#define DW_REG_A64_SP 31
+
+// X64 register mapping from real register index to DWARF2 (r8..r15 are mapped 1-1, but named registers aren't)
+const int regIndexToDwRegX64[16] = {DW_REG_X64_RAX, DW_REG_X64_RCX, DW_REG_X64_RDX, DW_REG_X64_RBX, DW_REG_X64_RSP, DW_REG_X64_RBP, DW_REG_X64_RSI,
+    DW_REG_X64_RDI, 8, 9, 10, 11, 12, 13, 14, 15};
 
 const int kCodeAlignFactor = 1;
 const int kDataAlignFactor = 8;
@@ -85,7 +83,7 @@ static uint8_t* defineSavedRegisterLocation(uint8_t* pos, int dwReg, uint32_t st
 {
     LUAU_ASSERT(stackOffset % kDataAlignFactor == 0 && "stack offsets have to be measured in kDataAlignFactor units");
 
-    if (dwReg <= 15)
+    if (dwReg <= 0x3f)
     {
         pos = writeu8(pos, DW_CFA_offset + dwReg);
     }
@@ -99,8 +97,9 @@ static uint8_t* defineSavedRegisterLocation(uint8_t* pos, int dwReg, uint32_t st
     return pos;
 }
 
-static uint8_t* advanceLocation(uint8_t* pos, uint8_t offset)
+static uint8_t* advanceLocation(uint8_t* pos, unsigned int offset)
 {
+    LUAU_ASSERT(offset < 256);
     pos = writeu8(pos, DW_CFA_advance_loc1);
     pos = writeu8(pos, offset);
     return pos;
@@ -132,8 +131,10 @@ size_t UnwindBuilderDwarf2::getBeginOffset() const
     return beginOffset;
 }
 
-void UnwindBuilderDwarf2::startInfo()
+void UnwindBuilderDwarf2::startInfo(Arch arch)
 {
+    LUAU_ASSERT(arch == A64 || arch == X64);
+
     uint8_t* cieLength = pos;
     pos = writeu32(pos, 0); // Length (to be filled later)
 
@@ -142,15 +143,24 @@ void UnwindBuilderDwarf2::startInfo()
 
     pos = writeu8(pos, 0); // CIE augmentation String ""
 
+    int ra = arch == A64 ? DW_REG_A64_LR : DW_REG_X64_RA;
+
     pos = writeuleb128(pos, kCodeAlignFactor);         // Code align factor
     pos = writeuleb128(pos, -kDataAlignFactor & 0x7f); // Data align factor of (as signed LEB128)
-    pos = writeu8(pos, DW_REG_RA);                     // Return address register
+    pos = writeu8(pos, ra);                            // Return address register
 
     // Optional CIE augmentation section (not present)
 
-    // Call frame instructions (common for all FDEs, of which we have 1)
+    // Call frame instructions (common for all FDEs)
-    pos = defineCfaExpression(pos, DW_REG_RSP, 8);        // Define CFA to be the rsp + 8
+    if (arch == A64)
-    pos = defineSavedRegisterLocation(pos, DW_REG_RA, 8); // Define return address register (RA) to be located at CFA - 8
+    {
+        pos = defineCfaExpression(pos, DW_REG_A64_SP, 0); // Define CFA to be the sp
+    }
+    else
+    {
+        pos = defineCfaExpression(pos, DW_REG_X64_RSP, 8);        // Define CFA to be the rsp + 8
+        pos = defineSavedRegisterLocation(pos, DW_REG_X64_RA, 8); // Define return address register (RA) to be located at CFA - 8
+    }
 
     pos = alignPosition(cieLength, pos);
     writeu32(cieLength, unsigned(pos - cieLength - 4)); // Length field itself is excluded from length
@@ -165,8 +175,6 @@ void UnwindBuilderDwarf2::startFunction()
     func.fdeEntryStartPos = uint32_t(pos - rawData);
     unwindFunctions.push_back(func);
 
-    stackOffset = 8; // Return address was pushed by calling the function
-
     fdeEntryStart = pos;                          // Will be written at the end
     pos = writeu32(pos, 0);                       // Length (to be filled later)
     pos = writeu32(pos, unsigned(pos - rawData)); // CIE pointer
@@ -178,42 +186,11 @@ void UnwindBuilderDwarf2::startFunction()
     // Function call frame instructions to follow
 }
 
-void UnwindBuilderDwarf2::spill(int espOffset, X64::RegisterX64 reg)
-{
-    pos = advanceLocation(pos, 5); // REX.W mov [rsp + imm8], reg
-}
-
-void UnwindBuilderDwarf2::save(X64::RegisterX64 reg)
-{
-    stackOffset += 8;
-    pos = advanceLocation(pos, 2); // REX.W push reg
-    pos = defineCfaExpressionOffset(pos, stackOffset);
-    pos = defineSavedRegisterLocation(pos, regIndexToDwRegX64[reg.index], stackOffset);
-}
-
-void UnwindBuilderDwarf2::allocStack(int size)
-{
-    stackOffset += size;
-    pos = advanceLocation(pos, 4); // REX.W sub rsp, imm8
-    pos = defineCfaExpressionOffset(pos, stackOffset);
-}
-
-void UnwindBuilderDwarf2::setupFrameReg(X64::RegisterX64 reg, int espOffset)
-{
-    if (espOffset != 0)
-        pos = advanceLocation(pos, 5); // REX.W lea rbp, [rsp + imm8]
-    else
-        pos = advanceLocation(pos, 3); // REX.W mov rbp, rsp
-
-    // Cfa is based on rsp, so no additonal commands are required
-}
-
 void UnwindBuilderDwarf2::finishFunction(uint32_t beginOffset, uint32_t endOffset)
 {
     unwindFunctions.back().beginOffset = beginOffset;
     unwindFunctions.back().endOffset = endOffset;
 
-    LUAU_ASSERT(stackOffset % 16 == 0 && "stack has to be aligned to 16 bytes after prologue");
     LUAU_ASSERT(fdeEntryStart != nullptr);
 
     pos = alignPosition(fdeEntryStart, pos);
@@ -228,6 +205,69 @@ void UnwindBuilderDwarf2::finishInfo()
     LUAU_ASSERT(getSize() <= kRawDataLimit);
 }
 
+void UnwindBuilderDwarf2::prologueA64(uint32_t prologueSize, uint32_t stackSize, std::initializer_list<A64::RegisterA64> regs)
+{
+    LUAU_ASSERT(stackSize % 16 == 0);
+    LUAU_ASSERT(regs.size() >= 2 && regs.begin()[0] == A64::x29 && regs.begin()[1] == A64::x30);
+    LUAU_ASSERT(regs.size() * 8 <= stackSize);
+
+    // sub sp, sp, stackSize
+    pos = advanceLocation(pos, 4);
+    pos = defineCfaExpressionOffset(pos, stackSize);
+
+    // stp/str to store each register to stack in order
+    pos = advanceLocation(pos, prologueSize - 4);
+
+    for (size_t i = 0; i < regs.size(); ++i)
+    {
+        LUAU_ASSERT(regs.begin()[i].kind == A64::KindA64::x);
+        pos = defineSavedRegisterLocation(pos, regs.begin()[i].index, stackSize - unsigned(i * 8));
+    }
+}
+
+void UnwindBuilderDwarf2::prologueX64(uint32_t prologueSize, uint32_t stackSize, bool setupFrame, std::initializer_list<X64::RegisterX64> regs)
+{
+    LUAU_ASSERT(stackSize > 0 && stackSize <= 128 && stackSize % 8 == 0);
+
+    unsigned int stackOffset = 8; // Return address was pushed by calling the function
+    unsigned int prologueOffset = 0;
+
+    if (setupFrame)
+    {
+        // push rbp
+        stackOffset += 8;
+        prologueOffset += 2;
+        pos = advanceLocation(pos, 2);
+        pos = defineCfaExpressionOffset(pos, stackOffset);
+        pos = defineSavedRegisterLocation(pos, DW_REG_X64_RBP, stackOffset);
+
+        // mov rbp, rsp
+        prologueOffset += 3;
+        pos = advanceLocation(pos, 3);
+    }
+
+    // push reg
+    for (X64::RegisterX64 reg : regs)
+    {
+        LUAU_ASSERT(reg.size == X64::SizeX64::qword);
+
+        stackOffset += 8;
+        prologueOffset += 2;
+        pos = advanceLocation(pos, 2);
+        pos = defineCfaExpressionOffset(pos, stackOffset);
+        pos = defineSavedRegisterLocation(pos, regIndexToDwRegX64[reg.index], stackOffset);
+    }
+
+    // sub rsp, stackSize
+    stackOffset += stackSize;
+    prologueOffset += 4;
+    pos = advanceLocation(pos, 4);
+    pos = defineCfaExpressionOffset(pos, stackOffset);
+
+    LUAU_ASSERT(stackOffset % 16 == 0);
+    LUAU_ASSERT(prologueOffset == prologueSize);
+}
+
 size_t UnwindBuilderDwarf2::getSize() const
 {
     return size_t(pos - rawData);
@@ -244,14 +284,14 @@ void UnwindBuilderDwarf2::finalize(char* target, size_t offset, void* funcAddres
 
     for (const UnwindFunctionDwarf2& func : unwindFunctions)
     {
-        uint8_t* fdeEntryStart = (uint8_t*)target + func.fdeEntryStartPos;
+        uint8_t* fdeEntry = (uint8_t*)target + func.fdeEntryStartPos;
 
-        writeu64(fdeEntryStart + kFdeInitialLocationOffset, uintptr_t(funcAddress) + offset + func.beginOffset);
+        writeu64(fdeEntry + kFdeInitialLocationOffset, uintptr_t(funcAddress) + offset + func.beginOffset);
 
         if (func.endOffset == kFullBlockFuncton)
-            writeu64(fdeEntryStart + kFdeAddressRangeOffset, funcSize - offset);
+            writeu64(fdeEntry + kFdeAddressRangeOffset, funcSize - offset);
         else
-            writeu64(fdeEntryStart + kFdeAddressRangeOffset, func.endOffset - func.beginOffset);
+            writeu64(fdeEntry + kFdeAddressRangeOffset, func.endOffset - func.beginOffset);
     }
 }
 

CodeGen/src/UnwindBuilderWin.cpp

@@ -31,7 +31,10 @@ size_t UnwindBuilderWin::getBeginOffset() const
     return beginOffset;
 }
 
-void UnwindBuilderWin::startInfo() {}
+void UnwindBuilderWin::startInfo(Arch arch)
+{
+    LUAU_ASSERT(arch == X64);
+}
 
 void UnwindBuilderWin::startFunction()
 {
@@ -50,45 +53,6 @@ void UnwindBuilderWin::startFunction()
     // rax has register index 0, which in Windows unwind info means that frame register is not used
     frameReg = X64::rax;
     frameRegOffset = 0;
-
-    // Return address was pushed by calling the function
-    stackOffset = 8;
-}
-
-void UnwindBuilderWin::spill(int espOffset, X64::RegisterX64 reg)
-{
-    prologSize += 5; // REX.W mov [rsp + imm8], reg
-}
-
-void UnwindBuilderWin::save(X64::RegisterX64 reg)
-{
-    prologSize += 2; // REX.W push reg
-    stackOffset += 8;
-    unwindCodes.push_back({prologSize, UWOP_PUSH_NONVOL, reg.index});
-}
-
-void UnwindBuilderWin::allocStack(int size)
-{
-    LUAU_ASSERT(size >= 8 && size <= 128 && size % 8 == 0);
-
-    prologSize += 4; // REX.W sub rsp, imm8
-    stackOffset += size;
-    unwindCodes.push_back({prologSize, UWOP_ALLOC_SMALL, uint8_t((size - 8) / 8)});
-}
-
-void UnwindBuilderWin::setupFrameReg(X64::RegisterX64 reg, int espOffset)
-{
-    LUAU_ASSERT(espOffset < 256 && espOffset % 16 == 0);
-
-    frameReg = reg;
-    frameRegOffset = uint8_t(espOffset / 16);
-
-    if (espOffset != 0)
-        prologSize += 5; // REX.W lea rbp, [rsp + imm8]
-    else
-        prologSize += 3; // REX.W mov rbp, rsp
-
-    unwindCodes.push_back({prologSize, UWOP_SET_FPREG, frameRegOffset});
 }
 
 void UnwindBuilderWin::finishFunction(uint32_t beginOffset, uint32_t endOffset)
@@ -99,8 +63,6 @@ void UnwindBuilderWin::finishFunction(uint32_t beginOffset, uint32_t endOffset)
     // Windows unwind code count is stored in uint8_t, so we can't have more
     LUAU_ASSERT(unwindCodes.size() < 256);
 
-    LUAU_ASSERT(stackOffset % 16 == 0 && "stack has to be aligned to 16 bytes after prologue");
-
     UnwindInfoWin info;
     info.version = 1;
     info.flags = 0; // No EH
@@ -142,6 +104,54 @@ void UnwindBuilderWin::finishFunction(uint32_t beginOffset, uint32_t endOffset)
 
 void UnwindBuilderWin::finishInfo() {}
 
+void UnwindBuilderWin::prologueA64(uint32_t prologueSize, uint32_t stackSize, std::initializer_list<A64::RegisterA64> regs)
+{
+    LUAU_ASSERT(!"Not implemented");
+}
+
+void UnwindBuilderWin::prologueX64(uint32_t prologueSize, uint32_t stackSize, bool setupFrame, std::initializer_list<X64::RegisterX64> regs)
+{
+    LUAU_ASSERT(stackSize > 0 && stackSize <= 128 && stackSize % 8 == 0);
+    LUAU_ASSERT(prologueSize < 256);
+
+    unsigned int stackOffset = 8; // Return address was pushed by calling the function
+    unsigned int prologueOffset = 0;
+
+    if (setupFrame)
+    {
+        // push rbp
+        stackOffset += 8;
+        prologueOffset += 2;
+        unwindCodes.push_back({uint8_t(prologueOffset), UWOP_PUSH_NONVOL, X64::rbp.index});
+
+        // mov rbp, rsp
+        prologueOffset += 3;
+        frameReg = X64::rbp;
+        frameRegOffset = 0;
+        unwindCodes.push_back({uint8_t(prologueOffset), UWOP_SET_FPREG, frameRegOffset});
+    }
+
+    // push reg
+    for (X64::RegisterX64 reg : regs)
+    {
+        LUAU_ASSERT(reg.size == X64::SizeX64::qword);
+
+        stackOffset += 8;
+        prologueOffset += 2;
+        unwindCodes.push_back({uint8_t(prologueOffset), UWOP_PUSH_NONVOL, reg.index});
+    }
+
+    // sub rsp, stackSize
+    stackOffset += stackSize;
+    prologueOffset += 4;
+    unwindCodes.push_back({uint8_t(prologueOffset), UWOP_ALLOC_SMALL, uint8_t((stackSize - 8) / 8)});
+
+    LUAU_ASSERT(stackOffset % 16 == 0);
+    LUAU_ASSERT(prologueOffset == prologueSize);
+
+    this->prologSize = prologueSize;
+}
+
 size_t UnwindBuilderWin::getSize() const
 {
     return sizeof(UnwindFunctionWin) * unwindFunctions.size() + size_t(rawDataPos - rawData);

Compiler/src/BytecodeBuilder.cpp

@@ -1701,8 +1701,6 @@ void BytecodeBuilder::dumpConstant(std::string& result, int k) const
             formatAppend(result, "'%s'", func.dumpname.c_str());
         break;
     }
-    default:
-        LUAU_UNREACHABLE();
     }
 }
 

VM/src/lvmexecute.cpp

@@ -913,7 +913,9 @@ reentry:
                 // slow-path: not a function call
                 if (LUAU_UNLIKELY(!ttisfunction(ra)))
                 {
-                    VM_PROTECT(luaV_tryfuncTM(L, ra));
+                    VM_PROTECT_PC(); // luaV_tryfuncTM may fail
+
+                    luaV_tryfuncTM(L, ra);
                     argtop++; // __call adds an extra self
                 }
 

tests/CodeAllocator.test.cpp

@@ -135,20 +135,9 @@ TEST_CASE("WindowsUnwindCodesX64")
 
     UnwindBuilderWin unwind;
 
-    unwind.startInfo();
+    unwind.startInfo(UnwindBuilder::X64);
     unwind.startFunction();
-    unwind.spill(16, rdx);
+    unwind.prologueX64(/* prologueSize= */ 23, /* stackSize= */ 72, /* setupFrame= */ true, {rdi, rsi, rbx, r12, r13, r14, r15});
-    unwind.spill(8, rcx);
-    unwind.save(rdi);
-    unwind.save(rsi);
-    unwind.save(rbx);
-    unwind.save(rbp);
-    unwind.save(r12);
-    unwind.save(r13);
-    unwind.save(r14);
-    unwind.save(r15);
-    unwind.allocStack(72);
-    unwind.setupFrameReg(rbp, 48);
     unwind.finishFunction(0x11223344, 0x55443322);
     unwind.finishInfo();
 
@@ -156,8 +145,8 @@ TEST_CASE("WindowsUnwindCodesX64")
     data.resize(unwind.getSize());
     unwind.finalize(data.data(), 0, nullptr, 0);
 
-    std::vector<uint8_t> expected{0x44, 0x33, 0x22, 0x11, 0x22, 0x33, 0x44, 0x55, 0x0c, 0x00, 0x00, 0x00, 0x01, 0x23, 0x0a, 0x35, 0x23, 0x33, 0x1e,
+    std::vector<uint8_t> expected{0x44, 0x33, 0x22, 0x11, 0x22, 0x33, 0x44, 0x55, 0x0c, 0x00, 0x00, 0x00, 0x01, 0x17, 0x0a, 0x05, 0x17, 0x82, 0x13,
-        0x82, 0x1a, 0xf0, 0x18, 0xe0, 0x16, 0xd0, 0x14, 0xc0, 0x12, 0x50, 0x10, 0x30, 0x0e, 0x60, 0x0c, 0x70};
+        0xf0, 0x11, 0xe0, 0x0f, 0xd0, 0x0d, 0xc0, 0x0b, 0x30, 0x09, 0x60, 0x07, 0x70, 0x05, 0x03, 0x02, 0x50};
 
     REQUIRE(data.size() == expected.size());
     CHECK(memcmp(data.data(), expected.data(), expected.size()) == 0);
@@ -170,18 +159,9 @@ TEST_CASE("Dwarf2UnwindCodesX64")
 
     UnwindBuilderDwarf2 unwind;
 
-    unwind.startInfo();
+    unwind.startInfo(UnwindBuilder::X64);
     unwind.startFunction();
-    unwind.save(rdi);
+    unwind.prologueX64(/* prologueSize= */ 23, /* stackSize= */ 72, /* setupFrame= */ true, {rdi, rsi, rbx, r12, r13, r14, r15});
-    unwind.save(rsi);
-    unwind.save(rbx);
-    unwind.save(rbp);
-    unwind.save(r12);
-    unwind.save(r13);
-    unwind.save(r14);
-    unwind.save(r15);
-    unwind.allocStack(72);
-    unwind.setupFrameReg(rbp, 48);
     unwind.finishFunction(0, 0);
     unwind.finishInfo();
 
@@ -189,11 +169,36 @@ TEST_CASE("Dwarf2UnwindCodesX64")
     data.resize(unwind.getSize());
     unwind.finalize(data.data(), 0, nullptr, 0);
 
-    std::vector<uint8_t> expected{0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x01, 0x78, 0x10, 0x0c, 0x07, 0x08, 0x05, 0x10, 0x01,
+    std::vector<uint8_t> expected{0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x01, 0x78, 0x10, 0x0c, 0x07, 0x08, 0x90, 0x01, 0x00,
         0x00, 0x00, 0x00, 0x00, 0x00, 0x4c, 0x00, 0x00, 0x00, 0x1c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
-        0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0x02, 0x0e, 0x10, 0x85, 0x02, 0x02, 0x02, 0x0e, 0x18, 0x84, 0x03, 0x02, 0x02, 0x0e, 0x20, 0x83,
+        0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0x02, 0x0e, 0x10, 0x86, 0x02, 0x02, 0x03, 0x02, 0x02, 0x0e, 0x18, 0x85, 0x03, 0x02, 0x02, 0x0e,
-        0x04, 0x02, 0x02, 0x0e, 0x28, 0x86, 0x05, 0x02, 0x02, 0x0e, 0x30, 0x8c, 0x06, 0x02, 0x02, 0x0e, 0x38, 0x8d, 0x07, 0x02, 0x02, 0x0e, 0x40,
+        0x20, 0x84, 0x04, 0x02, 0x02, 0x0e, 0x28, 0x83, 0x05, 0x02, 0x02, 0x0e, 0x30, 0x8c, 0x06, 0x02, 0x02, 0x0e, 0x38, 0x8d, 0x07, 0x02, 0x02,
-        0x8e, 0x08, 0x02, 0x02, 0x0e, 0x48, 0x8f, 0x09, 0x02, 0x04, 0x0e, 0x90, 0x01, 0x02, 0x05, 0x00, 0x00, 0x00, 0x00, 0x00};
+        0x0e, 0x40, 0x8e, 0x08, 0x02, 0x02, 0x0e, 0x48, 0x8f, 0x09, 0x02, 0x04, 0x0e, 0x90, 0x01, 0x00, 0x00, 0x00, 0x00, 0x00};
+
+    REQUIRE(data.size() == expected.size());
+    CHECK(memcmp(data.data(), expected.data(), expected.size()) == 0);
+}
+
+TEST_CASE("Dwarf2UnwindCodesA64")
+{
+    using namespace A64;
+
+    UnwindBuilderDwarf2 unwind;
+
+    unwind.startInfo(UnwindBuilder::A64);
+    unwind.startFunction();
+    unwind.prologueA64(/* prologueSize= */ 28, /* stackSize= */ 64, {x29, x30, x19, x20, x21, x22, x23, x24});
+    unwind.finishFunction(0, 32);
+    unwind.finishInfo();
+
+    std::vector<char> data;
+    data.resize(unwind.getSize());
+    unwind.finalize(data.data(), 0, nullptr, 0);
+
+    std::vector<uint8_t> expected{0x0c, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x01, 0x00, 0x01, 0x78, 0x1e, 0x0c, 0x1f, 0x00, 0x2c, 0x00, 0x00,
+        0x00, 0x14, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x20, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x02, 0x04,
+        0x0e, 0x40, 0x02, 0x18, 0x9d, 0x08, 0x9e, 0x07, 0x93, 0x06, 0x94, 0x05, 0x95, 0x04, 0x96, 0x03, 0x97, 0x02, 0x98, 0x01, 0x00, 0x00, 0x00,
+        0x00, 0x00, 0x00};
 
     REQUIRE(data.size() == expected.size());
     CHECK(memcmp(data.data(), expected.data(), expected.size()) == 0);
@@ -247,7 +252,7 @@ TEST_CASE("GeneratedCodeExecutionX64")
     CHECK(result == 210);
 }
 
-void throwing(int64_t arg)
+static void throwing(int64_t arg)
 {
     CHECK(arg == 25);
 
@@ -266,27 +271,25 @@ TEST_CASE("GeneratedCodeExecutionWithThrowX64")
     std::unique_ptr<UnwindBuilder> unwind = std::make_unique<UnwindBuilderDwarf2>();
 #endif
 
-    unwind->startInfo();
+    unwind->startInfo(UnwindBuilder::X64);
 
     Label functionBegin = build.setLabel();
     unwind->startFunction();
 
     // Prologue
-    build.push(rNonVol1);
-    unwind->save(rNonVol1);
-    build.push(rNonVol2);
-    unwind->save(rNonVol2);
     build.push(rbp);
-    unwind->save(rbp);
+    build.mov(rbp, rsp);
+    build.push(rNonVol1);
+    build.push(rNonVol2);
 
     int stackSize = 32;
     int localsSize = 16;
 
     build.sub(rsp, stackSize + localsSize);
-    unwind->allocStack(stackSize + localsSize);
 
-    build.lea(rbp, addr[rsp + stackSize]);
+    uint32_t prologueSize = build.setLabel().location;
-    unwind->setupFrameReg(rbp, stackSize);
+
+    unwind->prologueX64(prologueSize, stackSize + localsSize, /* setupFrame= */ true, {rNonVol1, rNonVol2});
 
     // Body
     build.mov(rNonVol1, rArg1);
@@ -297,10 +300,10 @@ TEST_CASE("GeneratedCodeExecutionWithThrowX64")
     build.call(rNonVol2);
 
     // Epilogue
-    build.lea(rsp, addr[rbp + localsSize]);
+    build.add(rsp, stackSize + localsSize);
-    build.pop(rbp);
     build.pop(rNonVol2);
     build.pop(rNonVol1);
+    build.pop(rbp);
     build.ret();
 
     unwind->finishFunction(build.getLabelOffset(functionBegin), ~0u);
@@ -349,7 +352,7 @@ TEST_CASE("GeneratedCodeExecutionMultipleFunctionsWithThrowX64")
     std::unique_ptr<UnwindBuilder> unwind = std::make_unique<UnwindBuilderDwarf2>();
 #endif
 
-    unwind->startInfo();
+    unwind->startInfo(UnwindBuilder::X64);
 
     Label start1;
     Label start2;
@@ -360,21 +363,19 @@ TEST_CASE("GeneratedCodeExecutionMultipleFunctionsWithThrowX64")
         unwind->startFunction();
 
         // Prologue
-        build.push(rNonVol1);
-        unwind->save(rNonVol1);
-        build.push(rNonVol2);
-        unwind->save(rNonVol2);
         build.push(rbp);
-        unwind->save(rbp);
+        build.mov(rbp, rsp);
+        build.push(rNonVol1);
+        build.push(rNonVol2);
 
         int stackSize = 32;
         int localsSize = 16;
 
         build.sub(rsp, stackSize + localsSize);
-        unwind->allocStack(stackSize + localsSize);
 
-        build.lea(rbp, addr[rsp + stackSize]);
+        uint32_t prologueSize = build.setLabel().location - start1.location;
-        unwind->setupFrameReg(rbp, stackSize);
+
+        unwind->prologueX64(prologueSize, stackSize + localsSize, /* setupFrame= */ true, {rNonVol1, rNonVol2});
 
         // Body
         build.mov(rNonVol1, rArg1);
@@ -385,41 +386,35 @@ TEST_CASE("GeneratedCodeExecutionMultipleFunctionsWithThrowX64")
         build.call(rNonVol2);
 
         // Epilogue
-        build.lea(rsp, addr[rbp + localsSize]);
+        build.add(rsp, stackSize + localsSize);
-        build.pop(rbp);
         build.pop(rNonVol2);
         build.pop(rNonVol1);
+        build.pop(rbp);
         build.ret();
 
         Label end1 = build.setLabel();
         unwind->finishFunction(build.getLabelOffset(start1), build.getLabelOffset(end1));
     }
 
-    // Second function with different layout
+    // Second function with different layout and no frame
     {
         build.setLabel(start2);
         unwind->startFunction();
 
         // Prologue
         build.push(rNonVol1);
-        unwind->save(rNonVol1);
         build.push(rNonVol2);
-        unwind->save(rNonVol2);
         build.push(rNonVol3);
-        unwind->save(rNonVol3);
         build.push(rNonVol4);
-        unwind->save(rNonVol4);
-        build.push(rbp);
-        unwind->save(rbp);
 
         int stackSize = 32;
-        int localsSize = 32;
+        int localsSize = 24;
 
         build.sub(rsp, stackSize + localsSize);
-        unwind->allocStack(stackSize + localsSize);
 
-        build.lea(rbp, addr[rsp + stackSize]);
+        uint32_t prologueSize = build.setLabel().location - start2.location;
-        unwind->setupFrameReg(rbp, stackSize);
+
+        unwind->prologueX64(prologueSize, stackSize + localsSize, /* setupFrame= */ false, {rNonVol1, rNonVol2, rNonVol3, rNonVol4});
 
         // Body
         build.mov(rNonVol3, rArg1);
@@ -430,8 +425,7 @@ TEST_CASE("GeneratedCodeExecutionMultipleFunctionsWithThrowX64")
         build.call(rNonVol4);
 
         // Epilogue
-        build.lea(rsp, addr[rbp + localsSize]);
+        build.add(rsp, stackSize + localsSize);
-        build.pop(rbp);
         build.pop(rNonVol4);
         build.pop(rNonVol3);
         build.pop(rNonVol2);
@@ -495,37 +489,29 @@ TEST_CASE("GeneratedCodeExecutionWithThrowOutsideTheGateX64")
     std::unique_ptr<UnwindBuilder> unwind = std::make_unique<UnwindBuilderDwarf2>();
 #endif
 
-    unwind->startInfo();
+    unwind->startInfo(UnwindBuilder::X64);
 
     Label functionBegin = build.setLabel();
     unwind->startFunction();
 
     // Prologue (some of these registers don't have to be saved, but we want to have a big prologue)
-    build.push(r10);
-    unwind->save(r10);
-    build.push(r11);
-    unwind->save(r11);
-    build.push(r12);
-    unwind->save(r12);
-    build.push(r13);
-    unwind->save(r13);
-    build.push(r14);
-    unwind->save(r14);
-    build.push(r15);
-    unwind->save(r15);
     build.push(rbp);
-    unwind->save(rbp);
+    build.mov(rbp, rsp);
+    build.push(r10);
+    build.push(r11);
+    build.push(r12);
+    build.push(r13);
+    build.push(r14);
+    build.push(r15);
 
     int stackSize = 64;
     int localsSize = 16;
 
     build.sub(rsp, stackSize + localsSize);
-    unwind->allocStack(stackSize + localsSize);
 
-    build.lea(rbp, addr[rsp + stackSize]);
+    uint32_t prologueSize = build.setLabel().location;
-    unwind->setupFrameReg(rbp, stackSize);
 
-    size_t prologueSize = build.setLabel().location;
+    unwind->prologueX64(prologueSize, stackSize + localsSize, /* setupFrame= */ true, {r10, r11, r12, r13, r14, r15});
 
     // Body
     build.mov(rax, rArg1);
@@ -535,14 +521,14 @@ TEST_CASE("GeneratedCodeExecutionWithThrowOutsideTheGateX64")
     Label returnOffset = build.setLabel();
 
     // Epilogue
-    build.lea(rsp, addr[rbp + localsSize]);
+    build.add(rsp, stackSize + localsSize);
-    build.pop(rbp);
     build.pop(r15);
     build.pop(r14);
     build.pop(r13);
     build.pop(r12);
     build.pop(r11);
     build.pop(r10);
+    build.pop(rbp);
     build.ret();
 
     unwind->finishFunction(build.getLabelOffset(functionBegin), ~0u);
@@ -650,6 +636,80 @@ TEST_CASE("GeneratedCodeExecutionA64")
     CHECK(result == 42);
 }
 
+#if 0
+static void throwing(int64_t arg)
+{
+    CHECK(arg == 25);
+
+    throw std::runtime_error("testing");
+}
+
+TEST_CASE("GeneratedCodeExecutionWithThrowA64")
+{
+    using namespace A64;
+
+    AssemblyBuilderA64 build(/* logText= */ false);
+
+    std::unique_ptr<UnwindBuilder> unwind = std::make_unique<UnwindBuilderDwarf2>();
+
+    unwind->startInfo(UnwindBuilder::A64);
+
+    build.sub(sp, sp, 32);
+    build.stp(x29, x30, mem(sp));
+    build.str(x28, mem(sp, 16));
+    build.mov(x29, sp);
+
+    Label prologueEnd = build.setLabel();
+
+    build.add(x0, x0, 15);
+    build.blr(x1);
+
+    build.ldr(x28, mem(sp, 16));
+    build.ldp(x29, x30, mem(sp));
+    build.add(sp, sp, 32);
+
+    build.ret();
+
+    Label functionEnd = build.setLabel();
+
+    unwind->startFunction();
+    unwind->prologueA64(build.getLabelOffset(prologueEnd), 32, {x29, x30, x28});
+    unwind->finishFunction(0, build.getLabelOffset(functionEnd));
+
+    build.finalize();
+
+    unwind->finishInfo();
+
+    size_t blockSize = 1024 * 1024;
+    size_t maxTotalSize = 1024 * 1024;
+    CodeAllocator allocator(blockSize, maxTotalSize);
+
+    allocator.context = unwind.get();
+    allocator.createBlockUnwindInfo = createBlockUnwindInfo;
+    allocator.destroyBlockUnwindInfo = destroyBlockUnwindInfo;
+
+    uint8_t* nativeData;
+    size_t sizeNativeData;
+    uint8_t* nativeEntry;
+    REQUIRE(allocator.allocate(build.data.data(), build.data.size(), reinterpret_cast<uint8_t*>(build.code.data()), build.code.size() * 4, nativeData,
+        sizeNativeData, nativeEntry));
+    REQUIRE(nativeEntry);
+
+    using FunctionType = int64_t(int64_t, void (*)(int64_t));
+    FunctionType* f = (FunctionType*)nativeEntry;
+
+    // To simplify debugging, CHECK_THROWS_WITH_AS is not used here
+    try
+    {
+        f(10, throwing);
+    }
+    catch (const std::runtime_error& error)
+    {
+        CHECK(strcmp(error.what(), "testing") == 0);
+    }
+}
+#endif
+
 #endif
 
 TEST_SUITE_END();

tests/IrBuilder.test.cpp

@@ -532,6 +532,30 @@ bb_0:
 )");
 }
 
+TEST_CASE_FIXTURE(IrBuilderFixture, "ReplacementPreservesUses")
+{
+    IrOp block = build.block(IrBlockKind::Internal);
+
+    build.beginBlock(block);
+
+    IrOp unk = build.inst(IrCmd::LOAD_INT, build.vmReg(0));
+    build.inst(IrCmd::STORE_INT, build.vmReg(8), build.inst(IrCmd::BITXOR_UINT, unk, build.constInt(~0u)));
+
+    build.inst(IrCmd::RETURN, build.constUint(0));
+
+    updateUseCounts(build.function);
+    constantFold();
+
+    CHECK("\n" + toString(build.function, /* includeUseInfo */ true) == R"(
+bb_0:                                                       ; useCount: 0
+   %0 = LOAD_INT R0                                          ; useCount: 1, lastUse: %0
+   %1 = BITNOT_UINT %0                                       ; useCount: 1, lastUse: %0
+   STORE_INT R8, %1                                          ; %2
+   RETURN 0u                                                 ; %3
+
+)");
+}
+
 TEST_CASE_FIXTURE(IrBuilderFixture, "NumericNan")
 {
     IrOp block = build.block(IrBlockKind::Internal);

tests/Normalize.test.cpp

@@ -470,8 +470,6 @@ TEST_SUITE_END();
 
 struct NormalizeFixture : Fixture
 {
-    ScopedFastFlag sff2{"LuauNegatedClassTypes", true};
-
     TypeArena arena;
     InternalErrorReporter iceHandler;
     UnifierSharedState unifierState{&iceHandler};
@@ -632,11 +630,6 @@ TEST_CASE_FIXTURE(NormalizeFixture, "union_function_and_top_function")
 
 TEST_CASE_FIXTURE(NormalizeFixture, "negated_function_is_anything_except_a_function")
 {
-    ScopedFastFlag sffs[] = {
-        {"LuauNegatedTableTypes", true},
-        {"LuauNegatedClassTypes", true},
-    };
-
     CHECK("(boolean | class | number | string | table | thread)?" == toString(normal(R"(
         Not<fun>
     )")));
@@ -649,11 +642,6 @@ TEST_CASE_FIXTURE(NormalizeFixture, "specific_functions_cannot_be_negated")
 
 TEST_CASE_FIXTURE(NormalizeFixture, "bare_negated_boolean")
 {
-    ScopedFastFlag sffs[] = {
-        {"LuauNegatedTableTypes", true},
-        {"LuauNegatedClassTypes", true},
-    };
-
     // TODO: We don't yet have a way to say number | string | thread | nil | Class | Table | Function
     CHECK("(class | function | number | string | table | thread)?" == toString(normal(R"(
         Not<boolean>
@@ -723,8 +711,6 @@ export type t0 = (((any)&({_:l0.t0,n0:t0,_G:any,}))&({_:any,}))&(((any)&({_:l0.t
 
 TEST_CASE_FIXTURE(NormalizeFixture, "unions_of_classes")
 {
-    ScopedFastFlag sff{"LuauNegatedClassTypes", true};
-
     createSomeClasses(&frontend);
     CHECK("Parent | Unrelated" == toString(normal("Parent | Unrelated")));
     CHECK("Parent" == toString(normal("Parent | Child")));
@@ -733,8 +719,6 @@ TEST_CASE_FIXTURE(NormalizeFixture, "unions_of_classes")
 
 TEST_CASE_FIXTURE(NormalizeFixture, "intersections_of_classes")
 {
-    ScopedFastFlag sff{"LuauNegatedClassTypes", true};
-
     createSomeClasses(&frontend);
     CHECK("Child" == toString(normal("Parent & Child")));
     CHECK("never" == toString(normal("Child & Unrelated")));
@@ -742,8 +726,6 @@ TEST_CASE_FIXTURE(NormalizeFixture, "intersections_of_classes")
 
 TEST_CASE_FIXTURE(NormalizeFixture, "narrow_union_of_classes_with_intersection")
 {
-    ScopedFastFlag sff{"LuauNegatedClassTypes", true};
-
     createSomeClasses(&frontend);
     CHECK("Child" == toString(normal("(Child | Unrelated) & Child")));
 }
@@ -764,11 +746,6 @@ TEST_CASE_FIXTURE(NormalizeFixture, "crazy_metatable")
 
 TEST_CASE_FIXTURE(NormalizeFixture, "negations_of_classes")
 {
-    ScopedFastFlag sffs[] = {
-        {"LuauNegatedTableTypes", true},
-        {"LuauNegatedClassTypes", true},
-    };
-
     createSomeClasses(&frontend);
     CHECK("(Parent & ~Child) | Unrelated" == toString(normal("(Parent & Not<Child>) | Unrelated")));
     CHECK("((class & ~Child) | boolean | function | number | string | table | thread)?" == toString(normal("Not<Child>")));
@@ -781,24 +758,18 @@ TEST_CASE_FIXTURE(NormalizeFixture, "negations_of_classes")
 
 TEST_CASE_FIXTURE(NormalizeFixture, "classes_and_unknown")
 {
-    ScopedFastFlag sff{"LuauNegatedClassTypes", true};
-
     createSomeClasses(&frontend);
     CHECK("Parent" == toString(normal("Parent & unknown")));
 }
 
 TEST_CASE_FIXTURE(NormalizeFixture, "classes_and_never")
 {
-    ScopedFastFlag sff{"LuauNegatedClassTypes", true};
-
     createSomeClasses(&frontend);
     CHECK("never" == toString(normal("Parent & never")));
 }
 
 TEST_CASE_FIXTURE(NormalizeFixture, "top_table_type")
 {
-    ScopedFastFlag sff{"LuauNegatedTableTypes", true};
-
     CHECK("table" == toString(normal("{} | tbl")));
     CHECK("{|  |}" == toString(normal("{} & tbl")));
     CHECK("never" == toString(normal("number & tbl")));
@@ -806,8 +777,6 @@ TEST_CASE_FIXTURE(NormalizeFixture, "top_table_type")
 
 TEST_CASE_FIXTURE(NormalizeFixture, "negations_of_tables")
 {
-    ScopedFastFlag sff{"LuauNegatedTableTypes", true};
-
     CHECK(nullptr == toNormalizedType("Not<{}>"));
     CHECK("(boolean | class | function | number | string | thread)?" == toString(normal("Not<tbl>")));
     CHECK("table" == toString(normal("Not<Not<tbl>>")));

tests/Parser.test.cpp

@@ -112,14 +112,6 @@ TEST_CASE_FIXTURE(Fixture, "can_haz_annotations")
     REQUIRE(block != nullptr);
 }
 
-TEST_CASE_FIXTURE(Fixture, "local_cannot_have_annotation_with_extensions_disabled")
-{
-    Luau::ParseOptions options;
-    options.allowTypeAnnotations = false;
-
-    CHECK_THROWS_AS(parse("local foo: string = \"Hello Types!\"", options), std::exception);
-}
-
 TEST_CASE_FIXTURE(Fixture, "local_with_annotation")
 {
     AstStatBlock* block = parse(R"(
@@ -150,14 +142,6 @@ TEST_CASE_FIXTURE(Fixture, "type_names_can_contain_dots")
     REQUIRE(block != nullptr);
 }
 
-TEST_CASE_FIXTURE(Fixture, "functions_cannot_have_return_annotations_if_extensions_are_disabled")
-{
-    Luau::ParseOptions options;
-    options.allowTypeAnnotations = false;
-
-    CHECK_THROWS_AS(parse("function foo(): number return 55 end", options), std::exception);
-}
-
 TEST_CASE_FIXTURE(Fixture, "functions_can_have_return_annotations")
 {
     AstStatBlock* block = parse(R"(
@@ -395,14 +379,6 @@ TEST_CASE_FIXTURE(Fixture, "return_type_is_an_intersection_type_if_led_with_one_
     CHECK(returnAnnotation->types.data[1]->as<AstTypeFunction>());
 }
 
-TEST_CASE_FIXTURE(Fixture, "illegal_type_alias_if_extensions_are_disabled")
-{
-    Luau::ParseOptions options;
-    options.allowTypeAnnotations = false;
-
-    CHECK_THROWS_AS(parse("type A = number", options), std::exception);
-}
-
 TEST_CASE_FIXTURE(Fixture, "type_alias_to_a_typeof")
 {
     AstStatBlock* block = parse(R"(
@@ -2837,8 +2813,6 @@ TEST_CASE_FIXTURE(Fixture, "get_a_nice_error_when_there_is_no_comma_after_last_t
 
 TEST_CASE_FIXTURE(Fixture, "missing_default_type_pack_argument_after_variadic_type_parameter")
 {
-    ScopedFastFlag sff{"LuauParserErrorsOnMissingDefaultTypePackArgument", true};
-
     ParseResult result = tryParse(R"(
         type Foo<T... = > = nil
     )");

tests/TypeInfer.anyerror.test.cpp

@@ -108,6 +108,9 @@ TEST_CASE_FIXTURE(Fixture, "for_in_loop_iterator_is_error2")
         end
     )");
 
+    if (FFlag::DebugLuauDeferredConstraintResolution)
+        LUAU_REQUIRE_ERROR_COUNT(2, result);
+    else
         LUAU_REQUIRE_ERROR_COUNT(1, result);
 
     CHECK_EQ("*error-type*", toString(requireType("a")));

tests/TypeInfer.functions.test.cpp

@@ -169,13 +169,26 @@ TEST_CASE_FIXTURE(Fixture, "list_only_alternative_overloads_that_match_argument_
 
     LUAU_REQUIRE_ERROR_COUNT(2, result);
 
+    if (FFlag::DebugLuauDeferredConstraintResolution)
+    {
+        GenericError* g = get<GenericError>(result.errors[0]);
+        REQUIRE(g);
+        CHECK(g->message == "None of the overloads for function that accept 1 arguments are compatible.");
+    }
+    else
+    {
         TypeMismatch* tm = get<TypeMismatch>(result.errors[0]);
         REQUIRE(tm);
         CHECK_EQ(builtinTypes->numberType, tm->wantedType);
         CHECK_EQ(builtinTypes->stringType, tm->givenType);
+    }
 
     ExtraInformation* ei = get<ExtraInformation>(result.errors[1]);
     REQUIRE(ei);
+
+    if (FFlag::DebugLuauDeferredConstraintResolution)
+        CHECK("Available overloads: (number) -> number; and (number) -> string" == ei->message);
+    else
         CHECK_EQ("Other overloads are also not viable: (number) -> string", ei->message);
 }
 

tests/TypeInfer.loops.test.cpp

@@ -2,6 +2,7 @@
 
 #include "Luau/AstQuery.h"
 #include "Luau/BuiltinDefinitions.h"
+#include "Luau/Frontend.h"
 #include "Luau/Scope.h"
 #include "Luau/TypeInfer.h"
 #include "Luau/Type.h"
@@ -31,6 +32,53 @@ TEST_CASE_FIXTURE(Fixture, "for_loop")
     CHECK_EQ(*builtinTypes->numberType, *requireType("q"));
 }
 
+TEST_CASE_FIXTURE(BuiltinsFixture, "iteration_no_table_passed")
+{
+    ScopedFastFlag sff{"DebugLuauDeferredConstraintResolution", true};
+    CheckResult result = check(R"(
+
+type Iterable = typeof(setmetatable(
+    {},
+    {}::{
+        __iter: (self: Iterable) -> (any, number) -> (number, string)
+    }
+))
+
+local t: Iterable
+
+for a, b in t do end
+)");
+
+
+    LUAU_REQUIRE_ERROR_COUNT(1, result);
+    GenericError* ge = get<GenericError>(result.errors[0]);
+    REQUIRE(ge);
+    CHECK_EQ("__iter metamethod must return (next[, table[, state]])", ge->message);
+}
+
+TEST_CASE_FIXTURE(BuiltinsFixture, "iteration_regression_issue_69967")
+{
+    ScopedFastFlag sff{"DebugLuauDeferredConstraintResolution", true};
+    CheckResult result = check(R"(
+
+type Iterable = typeof(setmetatable(
+    {},
+    {}::{
+        __iter: (self: Iterable) -> () -> (number, string)
+    }
+))
+
+local t: Iterable
+
+for a, b in t do end
+)");
+
+    LUAU_REQUIRE_ERROR_COUNT(1, result);
+    GenericError* ge = get<GenericError>(result.errors[0]);
+    REQUIRE(ge);
+    CHECK_EQ("__iter metamethod must return (next[, table[, state]])", ge->message);
+}
+
 TEST_CASE_FIXTURE(BuiltinsFixture, "for_in_loop")
 {
     CheckResult result = check(R"(

tests/TypeInfer.oop.test.cpp

@@ -26,10 +26,18 @@ TEST_CASE_FIXTURE(Fixture, "dont_suggest_using_colon_rather_than_dot_if_not_defi
         someTable.Function1() -- Argument count mismatch
     )");
 
+    if (FFlag::DebugLuauDeferredConstraintResolution)
+    {
+        LUAU_REQUIRE_ERROR_COUNT(2, result);
+        CHECK(toString(result.errors[0]) == "No overload for function accepts 0 arguments.");
+        CHECK(toString(result.errors[1]) == "Available overloads: <a>(a) -> ()");
+    }
+    else
+    {
         LUAU_REQUIRE_ERROR_COUNT(1, result);
-
         REQUIRE(get<CountMismatch>(result.errors[0]));
     }
+}
 
 TEST_CASE_FIXTURE(Fixture, "dont_suggest_using_colon_rather_than_dot_if_it_wont_help_2")
 {
@@ -42,10 +50,18 @@ TEST_CASE_FIXTURE(Fixture, "dont_suggest_using_colon_rather_than_dot_if_it_wont_
         someTable.Function2() -- Argument count mismatch
     )");
 
+    if (FFlag::DebugLuauDeferredConstraintResolution)
+    {
+        LUAU_REQUIRE_ERROR_COUNT(2, result);
+        CHECK(toString(result.errors[0]) == "No overload for function accepts 0 arguments.");
+        CHECK(toString(result.errors[1]) == "Available overloads: <a, b>(a, b) -> ()");
+    }
+    else
+    {
         LUAU_REQUIRE_ERROR_COUNT(1, result);
-
         REQUIRE(get<CountMismatch>(result.errors[0]));
     }
+}
 
 TEST_CASE_FIXTURE(Fixture, "dont_suggest_using_colon_rather_than_dot_if_another_overload_works")
 {

tests/TypeInfer.provisional.test.cpp

@@ -52,6 +52,43 @@ TEST_CASE_FIXTURE(Fixture, "typeguard_inference_incomplete")
     CHECK_EQ(expected, decorateWithTypes(code));
 }
 
+TEST_CASE_FIXTURE(BuiltinsFixture, "luau-polyfill.Array.filter")
+{
+    // This test exercises the fact that we should reduce sealed/unsealed/free tables
+    // res is a unsealed table with type {((T & ~nil)?) & any}
+    // Because we do not reduce it fully, we cannot unify it with `Array<T> = { [number] : T}
+    // TLDR; reduction needs to reduce the indexer on res so it unifies with Array<T>
+    CheckResult result = check(R"(
+--!strict
+-- Implements Javascript's `Array.prototype.filter` as defined below
+-- https://developer.cmozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Array/filter
+type Array<T> = { [number]: T }
+type callbackFn<T> = (element: T, index: number, array: Array<T>) -> boolean
+type callbackFnWithThisArg<T, U> = (thisArg: U, element: T, index: number, array: Array<T>) -> boolean
+type Object = { [string]: any }
+return function<T, U>(t: Array<T>, callback: callbackFn<T> | callbackFnWithThisArg<T, U>, thisArg: U?): Array<T>
+
+	local len = #t
+	local res = {}
+	if thisArg == nil then
+		for i = 1, len do
+			local kValue = t[i]
+			if kValue ~= nil then
+				if (callback :: callbackFn<T>)(kValue, i, t) then
+					res[i] = kValue
+				end
+			end
+		end
+	else
+	end
+
+	return res
+end
+    )");
+
+    LUAU_REQUIRE_NO_ERRORS(result);
+}
+
 TEST_CASE_FIXTURE(BuiltinsFixture, "xpcall_returns_what_f_returns")
 {
     const std::string code = R"(

tests/TypeInfer.refinements.test.cpp

@@ -8,7 +8,6 @@
 #include "doctest.h"
 
 LUAU_FASTFLAG(DebugLuauDeferredConstraintResolution)
-LUAU_FASTFLAG(LuauNegatedClassTypes)
 
 using namespace Luau;
 
@@ -64,7 +63,7 @@ struct RefinementClassFixture : BuiltinsFixture
         TypeArena& arena = frontend.globals.globalTypes;
         NotNull<Scope> scope{frontend.globals.globalScope.get()};
 
-        std::optional<TypeId> rootSuper = FFlag::LuauNegatedClassTypes ? std::make_optional(builtinTypes->classType) : std::nullopt;
+        std::optional<TypeId> rootSuper = std::make_optional(builtinTypes->classType);
 
         unfreeze(arena);
         TypeId vec3 = arena.addType(ClassType{"Vector3", {}, rootSuper, std::nullopt, {}, nullptr, "Test"});

tests/TypeInfer.singletons.test.cpp

@@ -131,9 +131,17 @@ TEST_CASE_FIXTURE(Fixture, "overloaded_function_call_with_singletons_mismatch")
     )");
 
     LUAU_REQUIRE_ERROR_COUNT(2, result);
+    if (FFlag::DebugLuauDeferredConstraintResolution)
+    {
+        CHECK_EQ("None of the overloads for function that accept 2 arguments are compatible.", toString(result.errors[0]));
+        CHECK_EQ("Available overloads: (true, string) -> (); and (false, number) -> ()", toString(result.errors[1]));
+    }
+    else
+    {
         CHECK_EQ("Type 'number' could not be converted into 'string'", toString(result.errors[0]));
         CHECK_EQ("Other overloads are also not viable: (false, number) -> ()", toString(result.errors[1]));
     }
+}
 
 TEST_CASE_FIXTURE(Fixture, "enums_using_singletons")
 {

tests/TypeInfer.tables.test.cpp

@@ -3625,4 +3625,23 @@ TEST_CASE_FIXTURE(BuiltinsFixture, "top_table_type_is_isomorphic_to_empty_sealed
     )");
 }
 
+TEST_CASE_FIXTURE(BuiltinsFixture, "luau-polyfill.Array.includes")
+{
+
+    CheckResult result = check(R"(
+type Array<T> = { [number]: T }
+
+function indexOf<T>(array: Array<T>, searchElement: any, fromIndex: number?): number
+	return -1
+end
+
+return function<T>(array: Array<T>, searchElement: any, fromIndex: number?): boolean
+	return -1 ~= indexOf(array, searchElement, fromIndex)
+end
+
+    )");
+
+    LUAU_REQUIRE_NO_ERRORS(result);
+}
+
 TEST_SUITE_END();

tools/faillist.txt

@@ -1,9 +1,5 @@
 AnnotationTests.too_many_type_params
 AstQuery.last_argument_function_call_type
-AstQuery::getDocumentationSymbolAtPosition.overloaded_class_method
-AstQuery::getDocumentationSymbolAtPosition.overloaded_fn
-AstQuery::getDocumentationSymbolAtPosition.table_overloaded_function_prop
-AutocompleteTest.autocomplete_response_perf1
 BuiltinTests.aliased_string_format
 BuiltinTests.assert_removes_falsy_types
 BuiltinTests.assert_removes_falsy_types2
@@ -54,6 +50,7 @@ ProvisionalTests.error_on_eq_metamethod_returning_a_type_other_than_boolean
 ProvisionalTests.free_options_cannot_be_unified_together
 ProvisionalTests.generic_type_leak_to_module_interface_variadic
 ProvisionalTests.greedy_inference_with_shared_self_triggers_function_with_no_returns
+ProvisionalTests.luau-polyfill.Array.filter
 ProvisionalTests.setmetatable_constrains_free_type_into_free_table
 ProvisionalTests.specialization_binds_with_prototypes_too_early
 ProvisionalTests.table_insert_with_a_singleton_argument
@@ -146,7 +143,6 @@ TypeInferClasses.index_instance_property
 TypeInferClasses.table_class_unification_reports_sane_errors_for_missing_properties
 TypeInferClasses.warn_when_prop_almost_matches
 TypeInferFunctions.cannot_hoist_interior_defns_into_signature
-TypeInferFunctions.dont_give_other_overloads_message_if_only_one_argument_matching_overload_exists
 TypeInferFunctions.function_cast_error_uses_correct_language
 TypeInferFunctions.function_decl_non_self_sealed_overwrite_2
 TypeInferFunctions.function_decl_non_self_unsealed_overwrite
@@ -158,7 +154,6 @@ TypeInferFunctions.infer_that_function_does_not_return_a_table
 TypeInferFunctions.luau_subtyping_is_np_hard
 TypeInferFunctions.no_lossy_function_type
 TypeInferFunctions.occurs_check_failure_in_function_return_type
-TypeInferFunctions.record_matching_overload
 TypeInferFunctions.report_exiting_without_return_strict
 TypeInferFunctions.return_type_by_overload
 TypeInferFunctions.too_few_arguments_variadic
@@ -205,6 +200,7 @@ TypePackTests.variadic_packs
 TypeSingletons.function_call_with_singletons
 TypeSingletons.function_call_with_singletons_mismatch
 TypeSingletons.no_widening_from_callsites
+TypeSingletons.overloaded_function_call_with_singletons_mismatch
 TypeSingletons.return_type_of_f_is_not_widened
 TypeSingletons.table_properties_type_error_escapes
 TypeSingletons.widen_the_supertype_if_it_is_free_and_subtype_has_singleton