Sync to upstream/release/601 (#1084)

## What's changed

- `bit32.byteswap` added
([RFC](4f543ec23b/docs/function-bit32-byteswap.md))
- Buffer library implementation
([RFC](4f543ec23b/docs/type-byte-buffer.md))
- Fixed a missing `stdint.h` include
- Fixed parser limiter for recursive type annotations being kind of
weird (fixes #645)

### Native Codegen
- Fixed a pair of issues when lowering `bit32.extract`
- Fixed a narrow edge case that could result in an infinite loop without
an interruption
- Fixed a negative array out-of-bounds access issue
- Temporarily reverted linear block predecessor value propagation

### New type solver
- We now type check assignments to annotated variables
- Fixed some test cases under local type inference
- Moved `isPending` checks for type families to improve performance
- Optimized our process for testing if a free type is sufficiently
solved
- Removed "none ptr" from lea instruction disassembly logging

### Build system & tooling
- CMake configuration now validates dependencies to maintain separation
between components
- Improvements to the fuzzer coverage
- Deduplicator for fuzzed callstacks

---------

Co-authored-by: Arseny Kapoulkine <arseny.kapoulkine@gmail.com>
Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
Co-authored-by: Andy Friesen <afriesen@roblox.com>
Co-authored-by: Aaron Weiss <aaronweiss@roblox.com>
Co-authored-by: Vighnesh Vijay <vvijay@roblox.com>

Analysis/include/Luau/Constraint.h

@@ -258,6 +258,8 @@ struct Constraint
     ConstraintV c;
 
     std::vector<NotNull<Constraint>> dependencies;
+
+    DenseHashSet<TypeId> getFreeTypes() const;
 };
 
 using ConstraintPtr = std::unique_ptr<Constraint>;

Analysis/include/Luau/ConstraintSolver.h

@@ -75,6 +75,9 @@ struct ConstraintSolver
     // Memoized instantiations of type aliases.
     DenseHashMap<InstantiationSignature, TypeId, HashInstantiationSignature> instantiatedAliases{{}};
 
+    // A mapping from free types to the number of unresolved constraints that mention them.
+    DenseHashMap<TypeId, size_t> unresolvedConstraints{{}};
+
     // Recorded errors that take place within the solver.
     ErrorVec errors;
 

Analysis/include/Luau/TypePairHash.h

@@ -4,7 +4,6 @@
 #include "Luau/TypeFwd.h"
 
 #include <stdint.h>
-
 #include <utility>
 
 namespace Luau

Analysis/src/Constraint.cpp

@@ -1,6 +1,7 @@
 // This file is part of the Luau programming language and is licensed under MIT License; see LICENSE.txt for details
 
 #include "Luau/Constraint.h"
+#include "Luau/VisitType.h"
 
 namespace Luau
 {
@@ -12,4 +13,40 @@ Constraint::Constraint(NotNull<Scope> scope, const Location& location, Constrain
 {
 }
 
+struct FreeTypeCollector : TypeOnceVisitor
+{
+
+    DenseHashSet<TypeId>* result;
+
+    FreeTypeCollector(DenseHashSet<TypeId>* result)
+        : result(result)
+    {
+    }
+
+    bool visit(TypeId ty, const FreeType&) override
+    {
+        result->insert(ty);
+        return false;
+    }
+};
+
+DenseHashSet<TypeId> Constraint::getFreeTypes() const
+{
+    DenseHashSet<TypeId> types{{}};
+    FreeTypeCollector ftc{&types};
+
+    if (auto sc = get<SubtypeConstraint>(*this))
+    {
+        ftc.traverse(sc->subType);
+        ftc.traverse(sc->superType);
+    }
+    else if (auto psc = get<PackSubtypeConstraint>(*this))
+    {
+        ftc.traverse(psc->subPack);
+        ftc.traverse(psc->superPack);
+    }
+
+    return types;
+}
+
 } // namespace Luau

Analysis/src/ConstraintSolver.cpp

@@ -280,6 +280,13 @@ ConstraintSolver::ConstraintSolver(NotNull<Normalizer> normalizer, NotNull<Scope
     {
         unsolvedConstraints.push_back(c);
 
+        // initialize the reference counts for the free types in this constraint.
+        for (auto ty : c->getFreeTypes())
+        {
+            // increment the reference count for `ty`
+            unresolvedConstraints[ty] += 1;
+        }
+
         for (NotNull<const Constraint> dep : c->dependencies)
         {
             block(dep, c);
@@ -360,6 +367,10 @@ void ConstraintSolver::run()
                 unblock(c);
                 unsolvedConstraints.erase(unsolvedConstraints.begin() + i);
 
+                // decrement the referenced free types for this constraint if we dispatched successfully!
+                for (auto ty : c->getFreeTypes())
+                    unresolvedConstraints[ty] -= 1;
+
                 if (logger)
                 {
                     logger->commitStepSnapshot(snapshot);
@@ -2380,48 +2391,12 @@ void ConstraintSolver::reportError(TypeError e)
     errors.back().moduleName = currentModuleName;
 }
 
-struct ContainsType : TypeOnceVisitor
-{
-    TypeId needle;
-    bool found = false;
-
-    explicit ContainsType(TypeId needle)
-        : needle(needle)
-    {
-    }
-
-    bool visit(TypeId) override
-    {
-        return !found; // traverse into the type iff we have yet to find the needle
-    }
-
-    bool visit(TypeId ty, const FreeType&) override
-    {
-        found |= ty == needle;
-        return false;
-    }
-};
-
 bool ConstraintSolver::hasUnresolvedConstraints(TypeId ty)
 {
-    if (!get<FreeType>(ty) || unsolvedConstraints.empty())
+    if (auto refCount = unresolvedConstraints.find(ty))
-        return false; // if it's not free, it never has any unresolved constraints, maybe?
+        return *refCount > 0;
 
-    ContainsType containsTy{ty};
+    return false;
-
-    for (auto constraint : unsolvedConstraints)
-    {
-        if (auto sc = get<SubtypeConstraint>(*constraint))
-        {
-            containsTy.traverse(sc->subType);
-            containsTy.traverse(sc->superType);
-
-            if (containsTy.found)
-                return true;
-        }
-    }
-
-    return containsTy.found;
 }
 
 TypeId ConstraintSolver::errorRecoveryType() const

Analysis/src/EmbeddedBuiltinDefinitions.cpp

@@ -1,9 +1,45 @@
 // This file is part of the Luau programming language and is licensed under MIT License; see LICENSE.txt for details
 #include "Luau/BuiltinDefinitions.h"
 
+LUAU_FASTFLAGVARIABLE(LuauBufferDefinitions, false)
+
 namespace Luau
 {
 
+static const std::string kBuiltinDefinitionBufferSrc = R"BUILTIN_SRC(
+
+-- TODO: this will be replaced with a built-in primitive type
+declare class buffer end
+
+declare buffer: {
+    create: (size: number) -> buffer,
+    fromstring: (str: string) -> buffer,
+    tostring: () -> string,
+    len: (b: buffer) -> number,
+    copy: (target: buffer, targetOffset: number, source: buffer, sourceOffset: number?, count: number?) -> (),
+    fill: (b: buffer, offset: number, value: number, count: number?) -> (),
+    readi8: (b: buffer, offset: number) -> number,
+    readu8: (b: buffer, offset: number) -> number,
+    readi16: (b: buffer, offset: number) -> number,
+    readu16: (b: buffer, offset: number) -> number,
+    readi32: (b: buffer, offset: number) -> number,
+    readu32: (b: buffer, offset: number) -> number,
+    readf32: (b: buffer, offset: number) -> number,
+    readf64: (b: buffer, offset: number) -> number,
+    writei8: (b: buffer, offset: number, value: number) -> (),
+    writeu8: (b: buffer, offset: number, value: number) -> (),
+    writei16: (b: buffer, offset: number, value: number) -> (),
+    writeu16: (b: buffer, offset: number, value: number) -> (),
+    writei32: (b: buffer, offset: number, value: number) -> (),
+    writeu32: (b: buffer, offset: number, value: number) -> (),
+    writef32: (b: buffer, offset: number, value: number) -> (),
+    writef64: (b: buffer, offset: number, value: number) -> (),
+    readstring: (b: buffer, offset: number, count: number) -> string,
+    writestring: (b: buffer, offset: number, value: string, count: number?) -> (),
+}
+
+)BUILTIN_SRC";
+
 static const std::string kBuiltinDefinitionLuaSrc = R"BUILTIN_SRC(
 
 declare bit32: {
@@ -199,6 +235,10 @@ declare function unpack<V>(tab: {V}, i: number?, j: number?): ...V
 std::string getBuiltinDefinitionSource()
 {
     std::string result = kBuiltinDefinitionLuaSrc;
+
+    if (FFlag::LuauBufferDefinitions)
+        result = kBuiltinDefinitionBufferSrc + result;
+
     return result;
 }
 

Analysis/src/TypeChecker2.cpp

@@ -868,6 +868,22 @@ struct TypeChecker2
         }
     }
 
+    std::optional<TypeId> getBindingType(AstExpr* expr)
+    {
+        if (auto localExpr = expr->as<AstExprLocal>())
+        {
+            Scope* s = stack.back();
+            return s->lookup(localExpr->local);
+        }
+        else if (auto globalExpr = expr->as<AstExprGlobal>())
+        {
+            Scope* s = stack.back();
+            return s->lookup(globalExpr->name);
+        }
+        else
+            return std::nullopt;
+    }
+
     void visit(AstStatAssign* assign)
     {
         size_t count = std::min(assign->vars.size, assign->values.size);
@@ -885,7 +901,15 @@ struct TypeChecker2
             if (get<NeverType>(lhsType))
                 continue;
 
-            testIsSubtype(rhsType, lhsType, rhs->location);
+            bool ok = testIsSubtype(rhsType, lhsType, rhs->location);
+
+            // If rhsType </: lhsType, then it's not useful to also report that rhsType </: bindingType
+            if (ok)
+            {
+                std::optional<TypeId> bindingType = getBindingType(lhs);
+                if (bindingType)
+                    testIsSubtype(rhsType, *bindingType, rhs->location);
+            }
         }
     }
 

Analysis/src/TypeFamily.cpp

@@ -347,6 +347,11 @@ TypeFamilyReductionResult<TypeId> lenFamilyFn(const std::vector<TypeId>& typePar
     }
 
     TypeId operandTy = follow(typeParams.at(0));
+
+    // check to see if the operand type is resolved enough, and wait to reduce if not
+    if (isPending(operandTy, ctx->solver))
+        return {std::nullopt, false, {operandTy}, {}};
+
     const NormalizedType* normTy = ctx->normalizer->normalize(operandTy);
 
     // if the type failed to normalize, we can't reduce, but know nothing about inhabitance.
@@ -370,10 +375,6 @@ TypeFamilyReductionResult<TypeId> lenFamilyFn(const std::vector<TypeId>& typePar
     if (normTy->hasTopTable() || get<TableType>(normalizedOperand))
         return {ctx->builtins->numberType, false, {}, {}};
 
-    // otherwise, we wait to see if the operand type is resolved
-    if (isPending(operandTy, ctx->solver))
-        return {std::nullopt, false, {operandTy}, {}};
-
     // findMetatableEntry demands the ability to emit errors, so we must give it
     // the necessary state to do that, even if we intend to just eat the errors.
     ErrorVec dummy;
@@ -421,6 +422,11 @@ TypeFamilyReductionResult<TypeId> unmFamilyFn(
     }
 
     TypeId operandTy = follow(typeParams.at(0));
+
+    // check to see if the operand type is resolved enough, and wait to reduce if not
+    if (isPending(operandTy, ctx->solver))
+        return {std::nullopt, false, {operandTy}, {}};
+
     const NormalizedType* normTy = ctx->normalizer->normalize(operandTy);
 
     // if the operand failed to normalize, we can't reduce, but know nothing about inhabitance.
@@ -439,10 +445,6 @@ TypeFamilyReductionResult<TypeId> unmFamilyFn(
     if (normTy->isExactlyNumber())
         return {ctx->builtins->numberType, false, {}, {}};
 
-    // otherwise, check if we need to wait on the type to be further resolved
-    if (isPending(operandTy, ctx->solver))
-        return {std::nullopt, false, {operandTy}, {}};
-
     // findMetatableEntry demands the ability to emit errors, so we must give it
     // the necessary state to do that, even if we intend to just eat the errors.
     ErrorVec dummy;
@@ -493,6 +495,13 @@ TypeFamilyReductionResult<TypeId> numericBinopFamilyFn(
 
     TypeId lhsTy = follow(typeParams.at(0));
     TypeId rhsTy = follow(typeParams.at(1));
+
+    // check to see if both operand types are resolved enough, and wait to reduce if not
+    if (isPending(lhsTy, ctx->solver))
+        return {std::nullopt, false, {lhsTy}, {}};
+    else if (isPending(rhsTy, ctx->solver))
+        return {std::nullopt, false, {rhsTy}, {}};
+
     const NormalizedType* normLhsTy = ctx->normalizer->normalize(lhsTy);
     const NormalizedType* normRhsTy = ctx->normalizer->normalize(rhsTy);
 
@@ -512,12 +521,6 @@ TypeFamilyReductionResult<TypeId> numericBinopFamilyFn(
     if (normLhsTy->isExactlyNumber() && normRhsTy->isExactlyNumber())
         return {ctx->builtins->numberType, false, {}, {}};
 
-    // otherwise, check if we need to wait on either type to be further resolved
-    if (isPending(lhsTy, ctx->solver))
-        return {std::nullopt, false, {lhsTy}, {}};
-    else if (isPending(rhsTy, ctx->solver))
-        return {std::nullopt, false, {rhsTy}, {}};
-
     // findMetatableEntry demands the ability to emit errors, so we must give it
     // the necessary state to do that, even if we intend to just eat the errors.
     ErrorVec dummy;
@@ -653,6 +656,13 @@ TypeFamilyReductionResult<TypeId> concatFamilyFn(const std::vector<TypeId>& type
 
     TypeId lhsTy = follow(typeParams.at(0));
     TypeId rhsTy = follow(typeParams.at(1));
+
+    // check to see if both operand types are resolved enough, and wait to reduce if not
+    if (isPending(lhsTy, ctx->solver))
+        return {std::nullopt, false, {lhsTy}, {}};
+    else if (isPending(rhsTy, ctx->solver))
+        return {std::nullopt, false, {rhsTy}, {}};
+
     const NormalizedType* normLhsTy = ctx->normalizer->normalize(lhsTy);
     const NormalizedType* normRhsTy = ctx->normalizer->normalize(rhsTy);
 
@@ -672,12 +682,6 @@ TypeFamilyReductionResult<TypeId> concatFamilyFn(const std::vector<TypeId>& type
     if ((normLhsTy->isSubtypeOfString() || normLhsTy->isExactlyNumber()) && (normRhsTy->isSubtypeOfString() || normRhsTy->isExactlyNumber()))
         return {ctx->builtins->stringType, false, {}, {}};
 
-    // otherwise, check if we need to wait on either type to be further resolved
-    if (isPending(lhsTy, ctx->solver))
-        return {std::nullopt, false, {lhsTy}, {}};
-    else if (isPending(rhsTy, ctx->solver))
-        return {std::nullopt, false, {rhsTy}, {}};
-
     // findMetatableEntry demands the ability to emit errors, so we must give it
     // the necessary state to do that, even if we intend to just eat the errors.
     ErrorVec dummy;
@@ -738,14 +742,11 @@ TypeFamilyReductionResult<TypeId> andFamilyFn(const std::vector<TypeId>& typePar
     TypeId lhsTy = follow(typeParams.at(0));
     TypeId rhsTy = follow(typeParams.at(1));
 
+    // check to see if both operand types are resolved enough, and wait to reduce if not
     if (isPending(lhsTy, ctx->solver))
-    {
         return {std::nullopt, false, {lhsTy}, {}};
-    }
     else if (isPending(rhsTy, ctx->solver))
-    {
         return {std::nullopt, false, {rhsTy}, {}};
-    }
 
     // And evalutes to a boolean if the LHS is falsey, and the RHS type if LHS is truthy.
     SimplifyResult filteredLhs = simplifyIntersection(ctx->builtins, ctx->arena, lhsTy, ctx->builtins->falsyType);
@@ -766,14 +767,11 @@ TypeFamilyReductionResult<TypeId> orFamilyFn(const std::vector<TypeId>& typePara
     TypeId lhsTy = follow(typeParams.at(0));
     TypeId rhsTy = follow(typeParams.at(1));
 
+    // check to see if both operand types are resolved enough, and wait to reduce if not
     if (isPending(lhsTy, ctx->solver))
-    {
         return {std::nullopt, false, {lhsTy}, {}};
-    }
     else if (isPending(rhsTy, ctx->solver))
-    {
         return {std::nullopt, false, {rhsTy}, {}};
-    }
 
     // Or evalutes to the LHS type if the LHS is truthy, and the RHS type if LHS is falsy.
     SimplifyResult filteredLhs = simplifyIntersection(ctx->builtins, ctx->arena, lhsTy, ctx->builtins->truthyType);
@@ -795,6 +793,13 @@ static TypeFamilyReductionResult<TypeId> comparisonFamilyFn(
 
     TypeId lhsTy = follow(typeParams.at(0));
     TypeId rhsTy = follow(typeParams.at(1));
+
+    // check to see if both operand types are resolved enough, and wait to reduce if not
+    if (isPending(lhsTy, ctx->solver))
+        return {std::nullopt, false, {lhsTy}, {}};
+    else if (isPending(rhsTy, ctx->solver))
+        return {std::nullopt, false, {rhsTy}, {}};
+
     const NormalizedType* normLhsTy = ctx->normalizer->normalize(lhsTy);
     const NormalizedType* normRhsTy = ctx->normalizer->normalize(rhsTy);
 
@@ -818,12 +823,6 @@ static TypeFamilyReductionResult<TypeId> comparisonFamilyFn(
     if (normLhsTy->isExactlyNumber() && normRhsTy->isExactlyNumber())
         return {ctx->builtins->booleanType, false, {}, {}};
 
-    // otherwise, check if we need to wait on either type to be further resolved
-    if (isPending(lhsTy, ctx->solver))
-        return {std::nullopt, false, {lhsTy}, {}};
-    else if (isPending(rhsTy, ctx->solver))
-        return {std::nullopt, false, {rhsTy}, {}};
-
     // findMetatableEntry demands the ability to emit errors, so we must give it
     // the necessary state to do that, even if we intend to just eat the errors.
     ErrorVec dummy;
@@ -895,6 +894,13 @@ TypeFamilyReductionResult<TypeId> eqFamilyFn(const std::vector<TypeId>& typePara
 
     TypeId lhsTy = follow(typeParams.at(0));
     TypeId rhsTy = follow(typeParams.at(1));
+
+    // check to see if both operand types are resolved enough, and wait to reduce if not
+    if (isPending(lhsTy, ctx->solver))
+        return {std::nullopt, false, {lhsTy}, {}};
+    else if (isPending(rhsTy, ctx->solver))
+        return {std::nullopt, false, {rhsTy}, {}};
+
     const NormalizedType* normLhsTy = ctx->normalizer->normalize(lhsTy);
     const NormalizedType* normRhsTy = ctx->normalizer->normalize(rhsTy);
 
@@ -910,12 +916,6 @@ TypeFamilyReductionResult<TypeId> eqFamilyFn(const std::vector<TypeId>& typePara
     if (is<NeverType>(lhsTy) || is<NeverType>(rhsTy))
         return {ctx->builtins->booleanType, false, {}, {}};
 
-    // otherwise, check if we need to wait on either type to be further resolved
-    if (isPending(lhsTy, ctx->solver))
-        return {std::nullopt, false, {lhsTy}, {}};
-    else if (isPending(rhsTy, ctx->solver))
-        return {std::nullopt, false, {rhsTy}, {}};
-
     // findMetatableEntry demands the ability to emit errors, so we must give it
     // the necessary state to do that, even if we intend to just eat the errors.
     ErrorVec dummy;

Ast/src/Parser.cpp

@@ -9,16 +9,21 @@
 #include <errno.h>
 #include <limits.h>
 
+LUAU_FASTINTVARIABLE(LuauRecursionLimit, 1000)
+LUAU_FASTINTVARIABLE(LuauTypeLengthLimit, 1000)
+LUAU_FASTINTVARIABLE(LuauParseErrorLimit, 100)
+
 // Warning: If you are introducing new syntax, ensure that it is behind a separate
 // flag so that we don't break production games by reverting syntax changes.
 // See docs/SyntaxChanges.md for an explanation.
-LUAU_FASTINTVARIABLE(LuauRecursionLimit, 1000)
-LUAU_FASTINTVARIABLE(LuauParseErrorLimit, 100)
 LUAU_FASTFLAGVARIABLE(LuauParseDeclareClassIndexer, false)
 LUAU_FASTFLAGVARIABLE(LuauClipExtraHasEndProps, false)
 LUAU_FASTFLAG(LuauFloorDivision)
 LUAU_FASTFLAG(LuauCheckedFunctionSyntax)
 
+LUAU_FASTFLAGVARIABLE(LuauBetterTypeUnionLimits, false)
+LUAU_FASTFLAGVARIABLE(LuauBetterTypeRecLimits, false)
+
 namespace Luau
 {
 
@@ -245,13 +250,13 @@ AstStatBlock* Parser::parseBlockNoScope()
 
     while (!blockFollow(lexer.current()))
     {
-        unsigned int recursionCounterOld = recursionCounter;
+        unsigned int oldRecursionCount = recursionCounter;
 
         incrementRecursionCounter("block");
 
         AstStat* stat = parseStat();
 
-        recursionCounter = recursionCounterOld;
+        recursionCounter = oldRecursionCount;
 
         if (lexer.current().type == ';')
         {
@@ -378,13 +383,13 @@ AstStat* Parser::parseIf()
     {
         if (FFlag::LuauClipExtraHasEndProps)
             thenbody->hasEnd = true;
-        unsigned int recursionCounterOld = recursionCounter;
+        unsigned int oldRecursionCount = recursionCounter;
         incrementRecursionCounter("elseif");
         elseLocation = lexer.current().location;
         elsebody = parseIf();
         end = elsebody->location;
         DEPRECATED_hasEnd = elsebody->as<AstStatIf>()->DEPRECATED_hasEnd;
-        recursionCounter = recursionCounterOld;
+        recursionCounter = oldRecursionCount;
     }
     else
     {
@@ -625,7 +630,7 @@ AstExpr* Parser::parseFunctionName(Location start, bool& hasself, AstName& debug
     // parse funcname into a chain of indexing operators
     AstExpr* expr = parseNameExpr("function name");
 
-    unsigned int recursionCounterOld = recursionCounter;
+    unsigned int oldRecursionCount = recursionCounter;
 
     while (lexer.current().type == '.')
     {
@@ -643,7 +648,7 @@ AstExpr* Parser::parseFunctionName(Location start, bool& hasself, AstName& debug
         incrementRecursionCounter("function name");
     }
 
-    recursionCounter = recursionCounterOld;
+    recursionCounter = oldRecursionCount;
 
     // finish with :
     if (lexer.current().type == ':')
@@ -1526,6 +1531,7 @@ AstType* Parser::parseTypeSuffix(AstType* type, const Location& begin)
 
     bool isUnion = false;
     bool isIntersection = false;
+    bool hasOptional = false;
 
     Location location = begin;
 
@@ -1535,20 +1541,34 @@ AstType* Parser::parseTypeSuffix(AstType* type, const Location& begin)
         if (c == '|')
         {
             nextLexeme();
+
+            unsigned int oldRecursionCount = recursionCounter;
             parts.push_back(parseSimpleType(/* allowPack= */ false).type);
+            if (FFlag::LuauBetterTypeUnionLimits)
+                recursionCounter = oldRecursionCount;
+
             isUnion = true;
         }
         else if (c == '?')
         {
             Location loc = lexer.current().location;
             nextLexeme();
-            parts.push_back(allocator.alloc<AstTypeReference>(loc, std::nullopt, nameNil, std::nullopt, loc));
+
+            if (!FFlag::LuauBetterTypeUnionLimits || !hasOptional)
+                parts.push_back(allocator.alloc<AstTypeReference>(loc, std::nullopt, nameNil, std::nullopt, loc));
+
             isUnion = true;
+            hasOptional = true;
         }
         else if (c == '&')
         {
             nextLexeme();
+
+            unsigned int oldRecursionCount = recursionCounter;
             parts.push_back(parseSimpleType(/* allowPack= */ false).type);
+            if (FFlag::LuauBetterTypeUnionLimits)
+                recursionCounter = oldRecursionCount;
+
             isIntersection = true;
         }
         else if (c == Lexeme::Dot3)
@@ -1558,6 +1578,9 @@ AstType* Parser::parseTypeSuffix(AstType* type, const Location& begin)
         }
         else
             break;
+
+        if (FFlag::LuauBetterTypeUnionLimits && parts.size() > unsigned(FInt::LuauTypeLengthLimit) + hasOptional)
+            ParseError::raise(parts.back()->location, "Exceeded allowed type length; simplify your type annotation to make the code compile");
     }
 
     if (parts.size() == 1)
@@ -1584,7 +1607,10 @@ AstType* Parser::parseTypeSuffix(AstType* type, const Location& begin)
 AstTypeOrPack Parser::parseTypeOrPack()
 {
     unsigned int oldRecursionCount = recursionCounter;
-    incrementRecursionCounter("type annotation");
+
+    // recursion counter is incremented in parseSimpleType
+    if (!FFlag::LuauBetterTypeRecLimits)
+        incrementRecursionCounter("type annotation");
 
     Location begin = lexer.current().location;
 
@@ -1604,7 +1630,10 @@ AstTypeOrPack Parser::parseTypeOrPack()
 AstType* Parser::parseType(bool inDeclarationContext)
 {
     unsigned int oldRecursionCount = recursionCounter;
-    incrementRecursionCounter("type annotation");
+
+    // recursion counter is incremented in parseSimpleType
+    if (!FFlag::LuauBetterTypeRecLimits)
+        incrementRecursionCounter("type annotation");
 
     Location begin = lexer.current().location;
 
@@ -1935,7 +1964,7 @@ AstExpr* Parser::parseExpr(unsigned int limit)
     };
     static_assert(sizeof(binaryPriority) / sizeof(binaryPriority[0]) == size_t(AstExprBinary::Op__Count), "binaryPriority needs an entry per op");
 
-    unsigned int recursionCounterOld = recursionCounter;
+    unsigned int oldRecursionCount = recursionCounter;
 
     // this handles recursive calls to parseSubExpr/parseExpr
     incrementRecursionCounter("expression");
@@ -1987,7 +2016,7 @@ AstExpr* Parser::parseExpr(unsigned int limit)
         incrementRecursionCounter("expression");
     }
 
-    recursionCounter = recursionCounterOld;
+    recursionCounter = oldRecursionCount;
 
     return expr;
 }
@@ -2054,7 +2083,7 @@ AstExpr* Parser::parsePrimaryExpr(bool asStatement)
 
     AstExpr* expr = parsePrefixExpr();
 
-    unsigned int recursionCounterOld = recursionCounter;
+    unsigned int oldRecursionCount = recursionCounter;
 
     while (true)
     {
@@ -2114,7 +2143,7 @@ AstExpr* Parser::parsePrimaryExpr(bool asStatement)
         incrementRecursionCounter("expression");
     }
 
-    recursionCounter = recursionCounterOld;
+    recursionCounter = oldRecursionCount;
 
     return expr;
 }

CMakeLists.txt

@@ -248,3 +248,19 @@ if(LUAU_BUILD_WEB)
     # the output is a single .js file with an embedded wasm blob
     target_link_options(Luau.Web PRIVATE -sSINGLE_FILE=1)
 endif()
+
+# validate dependencies for internal libraries
+foreach(LIB Luau.Ast Luau.Compiler Luau.Config Luau.Analysis Luau.CodeGen Luau.VM)
+    if(TARGET ${LIB})
+        get_target_property(DEPENDS ${LIB} LINK_LIBRARIES)
+        if(LIB MATCHES "CodeGen|VM" AND DEPENDS MATCHES "Ast|Analysis|Config|Compiler")
+            message(FATAL_ERROR ${LIB} " is a runtime component but it depends on one of the offline components")
+        endif()
+        if(LIB MATCHES "Ast|Analysis|Compiler" AND DEPENDS MATCHES "CodeGen|VM")
+            message(FATAL_ERROR ${LIB} " is an offline component but it depends on one of the runtime components")
+        endif()
+        if(LIB MATCHES "Ast|Compiler" AND DEPENDS MATCHES "Analysis|Config")
+            message(FATAL_ERROR ${LIB} " is a compiler component but it depends on one of the analysis components")
+        endif()
+    endif()
+endforeach()

CodeGen/src/AssemblyBuilderX64.cpp

@@ -1564,11 +1564,16 @@ void AssemblyBuilderX64::log(OperandX64 op)
     case CategoryX64::mem:
         if (op.base == rip)
         {
-            logAppend("%s ptr [.start%+d]", getSizeName(op.memSize), op.imm);
+            if (op.memSize != SizeX64::none)
+                logAppend("%s ptr ", getSizeName(op.memSize));
+            logAppend("[.start%+d]", op.imm);
             return;
         }
 
-        logAppend("%s ptr [", getSizeName(op.memSize));
+        if (op.memSize != SizeX64::none)
+            logAppend("%s ptr ", getSizeName(op.memSize));
+
+        logAppend("[");
 
         if (op.base != noreg)
             logAppend("%s", getRegisterName(op.base));

CodeGen/src/IrBuilder.cpp

@@ -85,6 +85,9 @@ static void buildArgumentTypeChecks(IrBuilder& build, Proto* proto)
         case LBC_TYPE_VECTOR:
             build.inst(IrCmd::CHECK_TAG, load, build.constTag(LUA_TVECTOR), build.vmExit(kVmExitEntryGuardPc));
             break;
+        case LBC_TYPE_BUFFER:
+            build.inst(IrCmd::CHECK_TAG, load, build.constTag(LUA_TBUFFER), build.vmExit(kVmExitEntryGuardPc));
+            break;
         }
 
         if (optional)

CodeGen/src/IrDump.cpp

@@ -61,6 +61,8 @@ static const char* getTagName(uint8_t tag)
         return "tuserdata";
     case LUA_TTHREAD:
         return "tthread";
+    case LUA_TBUFFER:
+        return "tbuffer";
     case LUA_TPROTO:
         return "tproto";
     case LUA_TUPVAL:

CodeGen/src/IrTranslateBuiltins.cpp

@@ -472,6 +472,9 @@ static BuiltinImplResult translateBuiltinBit32Extract(
     if (nparams < 2 || nresults > 1)
         return {BuiltinImplType::None, -1};
 
+    if (nparams == 2 && args.kind == IrOpKind::Constant && unsigned(int(build.function.doubleOp(args))) >= 32)
+        return {BuiltinImplType::None, -1};
+
     builtinCheckDouble(build, build.vmReg(arg), pcpos);
     builtinCheckDouble(build, args, pcpos);
 
@@ -486,15 +489,14 @@ static BuiltinImplResult translateBuiltinBit32Extract(
         if (vb.kind == IrOpKind::Constant)
         {
             int f = int(build.function.doubleOp(vb));
+            LUAU_ASSERT(unsigned(f) < 32); // checked above
 
-            if (unsigned(f) >= 32)
+            value = n;
-                build.inst(IrCmd::JUMP, fallback);
 
-            // TODO: this pair can be optimized using a bit-select instruction (bt on x86)
             if (f)
-                value = build.inst(IrCmd::BITRSHIFT_UINT, n, build.constInt(f));
+                value = build.inst(IrCmd::BITRSHIFT_UINT, value, build.constInt(f));
 
-            if ((f + 1) < 32)
+            if (f + 1 < 32)
                 value = build.inst(IrCmd::BITAND_UINT, value, build.constInt(1));
         }
         else
@@ -505,7 +507,6 @@ static BuiltinImplResult translateBuiltinBit32Extract(
             build.inst(IrCmd::JUMP_CMP_INT, f, build.constInt(32), build.cond(IrCondition::UnsignedGreaterEqual), fallback, block);
             build.beginBlock(block);
 
-            // TODO: this pair can be optimized using a bit-select instruction (bt on x86)
             IrOp shift = build.inst(IrCmd::BITRSHIFT_UINT, n, f);
             value = build.inst(IrCmd::BITAND_UINT, shift, build.constInt(1));
         }

CodeGen/src/IrTranslation.cpp

@@ -12,7 +12,6 @@
 #include "lstate.h"
 #include "ltm.h"
 
-LUAU_FASTFLAGVARIABLE(LuauImproveForN2, false)
 LUAU_FASTFLAG(LuauReduceStackSpills)
 LUAU_FASTFLAGVARIABLE(LuauInlineArrConstOffset, false)
 LUAU_FASTFLAGVARIABLE(LuauLowerAltLoopForn, false)
@@ -635,22 +634,16 @@ static IrOp getLoopStepK(IrBuilder& build, int ra)
 
 void beforeInstForNPrep(IrBuilder& build, const Instruction* pc)
 {
-    if (FFlag::LuauImproveForN2)
+    int ra = LUAU_INSN_A(*pc);
-    {
-        int ra = LUAU_INSN_A(*pc);
 
-        IrOp stepK = getLoopStepK(build, ra);
+    IrOp stepK = getLoopStepK(build, ra);
-        build.loopStepStack.push_back(stepK);
+    build.loopStepStack.push_back(stepK);
-    }
 }
 
 void afterInstForNLoop(IrBuilder& build, const Instruction* pc)
 {
-    if (FFlag::LuauImproveForN2)
+    LUAU_ASSERT(!build.loopStepStack.empty());
-    {
+    build.loopStepStack.pop_back();
-        LUAU_ASSERT(!build.loopStepStack.empty());
-        build.loopStepStack.pop_back();
-    }
 }
 
 void translateInstForNPrep(IrBuilder& build, const Instruction* pc, int pcpos)
@@ -660,119 +653,65 @@ void translateInstForNPrep(IrBuilder& build, const Instruction* pc, int pcpos)
     IrOp loopStart = build.blockAtInst(pcpos + getOpLength(LuauOpcode(LUAU_INSN_OP(*pc))));
     IrOp loopExit = build.blockAtInst(getJumpTarget(*pc, pcpos));
 
-    if (FFlag::LuauImproveForN2)
+    LUAU_ASSERT(!build.loopStepStack.empty());
+    IrOp stepK = build.loopStepStack.back();
+
+    // When loop parameters are not numbers, VM tries to perform type coercion from string and raises an exception if that fails
+    // Performing that fallback in native code increases code size and complicates CFG, obscuring the values when they are constant
+    // To avoid that overhead for an extremely rare case (that doesn't even typecheck), we exit to VM to handle it
+    IrOp tagLimit = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 0));
+    build.inst(IrCmd::CHECK_TAG, tagLimit, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
+    IrOp tagIdx = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 2));
+    build.inst(IrCmd::CHECK_TAG, tagIdx, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
+
+    IrOp limit = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 0));
+    IrOp idx = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 2));
+
+    if (stepK.kind == IrOpKind::Undef)
     {
-        LUAU_ASSERT(!build.loopStepStack.empty());
+        IrOp tagStep = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 1));
-        IrOp stepK = build.loopStepStack.back();
+        build.inst(IrCmd::CHECK_TAG, tagStep, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
 
-        // When loop parameters are not numbers, VM tries to perform type coercion from string and raises an exception if that fails
+        if (FFlag::LuauLowerAltLoopForn)
-        // Performing that fallback in native code increases code size and complicates CFG, obscuring the values when they are constant
-        // To avoid that overhead for an extremely rare case (that doesn't even typecheck), we exit to VM to handle it
-        IrOp tagLimit = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 0));
-        build.inst(IrCmd::CHECK_TAG, tagLimit, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
-        IrOp tagIdx = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 2));
-        build.inst(IrCmd::CHECK_TAG, tagIdx, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
-
-        IrOp limit = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 0));
-        IrOp idx = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 2));
-
-        if (stepK.kind == IrOpKind::Undef)
         {
-            IrOp tagStep = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 1));
+            IrOp step = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 1));
-            build.inst(IrCmd::CHECK_TAG, tagStep, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
 
-            if (FFlag::LuauLowerAltLoopForn)
+            build.inst(IrCmd::JUMP_FORN_LOOP_COND, idx, limit, step, loopStart, loopExit);
-            {
-                IrOp step = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 1));
-
-                build.inst(IrCmd::JUMP_FORN_LOOP_COND, idx, limit, step, loopStart, loopExit);
-            }
-            else
-            {
-                IrOp direct = build.block(IrBlockKind::Internal);
-                IrOp reverse = build.block(IrBlockKind::Internal);
-
-                IrOp zero = build.constDouble(0.0);
-                IrOp step = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 1));
-
-                // step > 0
-                // note: equivalent to 0 < step, but lowers into one instruction on both X64 and A64
-                build.inst(IrCmd::JUMP_CMP_NUM, step, zero, build.cond(IrCondition::Greater), direct, reverse);
-
-                // Condition to start the loop: step > 0 ? idx <= limit : limit <= idx
-                // We invert the condition so that loopStart is the fallthrough (false) label
-
-                // step > 0 is false, check limit <= idx
-                build.beginBlock(reverse);
-                build.inst(IrCmd::JUMP_CMP_NUM, limit, idx, build.cond(IrCondition::NotLessEqual), loopExit, loopStart);
-
-                // step > 0 is true, check idx <= limit
-                build.beginBlock(direct);
-                build.inst(IrCmd::JUMP_CMP_NUM, idx, limit, build.cond(IrCondition::NotLessEqual), loopExit, loopStart);
-            }
         }
         else
         {
-            double stepN = build.function.doubleOp(stepK);
+            IrOp direct = build.block(IrBlockKind::Internal);
+            IrOp reverse = build.block(IrBlockKind::Internal);
+
+            IrOp zero = build.constDouble(0.0);
+            IrOp step = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 1));
+
+            // step > 0
+            // note: equivalent to 0 < step, but lowers into one instruction on both X64 and A64
+            build.inst(IrCmd::JUMP_CMP_NUM, step, zero, build.cond(IrCondition::Greater), direct, reverse);
 
             // Condition to start the loop: step > 0 ? idx <= limit : limit <= idx
             // We invert the condition so that loopStart is the fallthrough (false) label
-            if (stepN > 0)
+
-                build.inst(IrCmd::JUMP_CMP_NUM, idx, limit, build.cond(IrCondition::NotLessEqual), loopExit, loopStart);
+            // step > 0 is false, check limit <= idx
-            else
+            build.beginBlock(reverse);
-                build.inst(IrCmd::JUMP_CMP_NUM, limit, idx, build.cond(IrCondition::NotLessEqual), loopExit, loopStart);
+            build.inst(IrCmd::JUMP_CMP_NUM, limit, idx, build.cond(IrCondition::NotLessEqual), loopExit, loopStart);
+
+            // step > 0 is true, check idx <= limit
+            build.beginBlock(direct);
+            build.inst(IrCmd::JUMP_CMP_NUM, idx, limit, build.cond(IrCondition::NotLessEqual), loopExit, loopStart);
         }
     }
-    else if (FFlag::LuauLowerAltLoopForn)
-    {
-        // When loop parameters are not numbers, VM tries to perform type coercion from string and raises an exception if that fails
-        // Performing that fallback in native code increases code size and complicates CFG, obscuring the values when they are constant
-        // To avoid that overhead for an extreemely rare case (that doesn't even typecheck), we exit to VM to handle it
-        IrOp tagLimit = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 0));
-        build.inst(IrCmd::CHECK_TAG, tagLimit, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
-        IrOp tagStep = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 1));
-        build.inst(IrCmd::CHECK_TAG, tagStep, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
-        IrOp tagIdx = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 2));
-        build.inst(IrCmd::CHECK_TAG, tagIdx, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
-
-        IrOp limit = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 0));
-        IrOp step = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 1));
-        IrOp idx = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 2));
-
-        build.inst(IrCmd::JUMP_FORN_LOOP_COND, idx, limit, step, loopStart, loopExit);
-    }
     else
     {
-        IrOp direct = build.block(IrBlockKind::Internal);
+        double stepN = build.function.doubleOp(stepK);
-        IrOp reverse = build.block(IrBlockKind::Internal);
 
-        // When loop parameters are not numbers, VM tries to perform type coercion from string and raises an exception if that fails
+        // Condition to start the loop: step > 0 ? idx <= limit : limit <= idx
-        // Performing that fallback in native code increases code size and complicates CFG, obscuring the values when they are constant
+        // We invert the condition so that loopStart is the fallthrough (false) label
-        // To avoid that overhead for an extreemely rare case (that doesn't even typecheck), we exit to VM to handle it
+        if (stepN > 0)
-        IrOp tagLimit = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 0));
+            build.inst(IrCmd::JUMP_CMP_NUM, idx, limit, build.cond(IrCondition::NotLessEqual), loopExit, loopStart);
-        build.inst(IrCmd::CHECK_TAG, tagLimit, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
+        else
-        IrOp tagStep = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 1));
+            build.inst(IrCmd::JUMP_CMP_NUM, limit, idx, build.cond(IrCondition::NotLessEqual), loopExit, loopStart);
-        build.inst(IrCmd::CHECK_TAG, tagStep, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
-        IrOp tagIdx = build.inst(IrCmd::LOAD_TAG, build.vmReg(ra + 2));
-        build.inst(IrCmd::CHECK_TAG, tagIdx, build.constTag(LUA_TNUMBER), build.vmExit(pcpos));
-
-        IrOp zero = build.constDouble(0.0);
-        IrOp limit = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 0));
-        IrOp step = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 1));
-        IrOp idx = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 2));
-
-        // step <= 0
-        build.inst(IrCmd::JUMP_CMP_NUM, step, zero, build.cond(IrCondition::LessEqual), reverse, direct);
-
-        // TODO: target branches can probably be arranged better, but we need tests for NaN behavior preservation
-
-        // step <= 0 is false, check idx <= limit
-        build.beginBlock(direct);
-        build.inst(IrCmd::JUMP_CMP_NUM, idx, limit, build.cond(IrCondition::LessEqual), loopStart, loopExit);
-
-        // step <= 0 is true, check limit <= idx
-        build.beginBlock(reverse);
-        build.inst(IrCmd::JUMP_CMP_NUM, limit, idx, build.cond(IrCondition::LessEqual), loopStart, loopExit);
     }
 
     // Fallthrough in original bytecode is implicit, so we start next internal block here
@@ -782,8 +721,7 @@ void translateInstForNPrep(IrBuilder& build, const Instruction* pc, int pcpos)
     // VM places interrupt in FORNLOOP, but that creates a likely spill point for short loops that use loop index as INTERRUPT always spills
     // We place the interrupt at the beginning of the loop body instead; VM uses FORNLOOP because it doesn't want to waste an extra instruction.
     // Because loop block may not have been started yet (as it's started when lowering the first instruction!), we need to defer INTERRUPT placement.
-    if (FFlag::LuauImproveForN2)
+    build.interruptRequested = true;
-        build.interruptRequested = true;
 }
 
 void translateInstForNLoop(IrBuilder& build, const Instruction* pc, int pcpos)
@@ -793,95 +731,59 @@ void translateInstForNLoop(IrBuilder& build, const Instruction* pc, int pcpos)
     IrOp loopRepeat = build.blockAtInst(getJumpTarget(*pc, pcpos));
     IrOp loopExit = build.blockAtInst(pcpos + getOpLength(LuauOpcode(LUAU_INSN_OP(*pc))));
 
-    if (FFlag::LuauImproveForN2)
+    // normally, the interrupt is placed at the beginning of the loop body by FORNPREP translation
+    // however, there are rare contrived cases where FORNLOOP ends up jumping to itself without an interrupt placed
+    // we detect this by checking if loopRepeat has any instructions (it should normally start with INTERRUPT) and emit a failsafe INTERRUPT if not
+    if (build.function.blockOp(loopRepeat).start == build.function.instructions.size())
+        build.inst(IrCmd::INTERRUPT, build.constUint(pcpos));
+
+    LUAU_ASSERT(!build.loopStepStack.empty());
+    IrOp stepK = build.loopStepStack.back();
+
+    IrOp limit = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 0));
+    IrOp step = stepK.kind == IrOpKind::Undef ? build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 1)) : stepK;
+
+    IrOp idx = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 2));
+    idx = build.inst(IrCmd::ADD_NUM, idx, step);
+    build.inst(IrCmd::STORE_DOUBLE, build.vmReg(ra + 2), idx);
+
+    if (stepK.kind == IrOpKind::Undef)
     {
-        LUAU_ASSERT(!build.loopStepStack.empty());
+        if (FFlag::LuauLowerAltLoopForn)
-        IrOp stepK = build.loopStepStack.back();
-
-        IrOp limit = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 0));
-        IrOp step = stepK.kind == IrOpKind::Undef ? build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 1)) : stepK;
-
-        IrOp idx = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 2));
-        idx = build.inst(IrCmd::ADD_NUM, idx, step);
-        build.inst(IrCmd::STORE_DOUBLE, build.vmReg(ra + 2), idx);
-
-        if (stepK.kind == IrOpKind::Undef)
         {
-            if (FFlag::LuauLowerAltLoopForn)
+            build.inst(IrCmd::JUMP_FORN_LOOP_COND, idx, limit, step, loopRepeat, loopExit);
-            {
-                build.inst(IrCmd::JUMP_FORN_LOOP_COND, idx, limit, step, loopRepeat, loopExit);
-            }
-            else
-            {
-                IrOp direct = build.block(IrBlockKind::Internal);
-                IrOp reverse = build.block(IrBlockKind::Internal);
-
-                IrOp zero = build.constDouble(0.0);
-
-                // step > 0
-                // note: equivalent to 0 < step, but lowers into one instruction on both X64 and A64
-                build.inst(IrCmd::JUMP_CMP_NUM, step, zero, build.cond(IrCondition::Greater), direct, reverse);
-
-                // Condition to continue the loop: step > 0 ? idx <= limit : limit <= idx
-
-                // step > 0 is false, check limit <= idx
-                build.beginBlock(reverse);
-                build.inst(IrCmd::JUMP_CMP_NUM, limit, idx, build.cond(IrCondition::LessEqual), loopRepeat, loopExit);
-
-                // step > 0 is true, check idx <= limit
-                build.beginBlock(direct);
-                build.inst(IrCmd::JUMP_CMP_NUM, idx, limit, build.cond(IrCondition::LessEqual), loopRepeat, loopExit);
-            }
         }
         else
         {
-            double stepN = build.function.doubleOp(stepK);
+            IrOp direct = build.block(IrBlockKind::Internal);
+            IrOp reverse = build.block(IrBlockKind::Internal);
+
+            IrOp zero = build.constDouble(0.0);
+
+            // step > 0
+            // note: equivalent to 0 < step, but lowers into one instruction on both X64 and A64
+            build.inst(IrCmd::JUMP_CMP_NUM, step, zero, build.cond(IrCondition::Greater), direct, reverse);
 
             // Condition to continue the loop: step > 0 ? idx <= limit : limit <= idx
-            if (stepN > 0)
+
-                build.inst(IrCmd::JUMP_CMP_NUM, idx, limit, build.cond(IrCondition::LessEqual), loopRepeat, loopExit);
+            // step > 0 is false, check limit <= idx
-            else
+            build.beginBlock(reverse);
-                build.inst(IrCmd::JUMP_CMP_NUM, limit, idx, build.cond(IrCondition::LessEqual), loopRepeat, loopExit);
+            build.inst(IrCmd::JUMP_CMP_NUM, limit, idx, build.cond(IrCondition::LessEqual), loopRepeat, loopExit);
+
+            // step > 0 is true, check idx <= limit
+            build.beginBlock(direct);
+            build.inst(IrCmd::JUMP_CMP_NUM, idx, limit, build.cond(IrCondition::LessEqual), loopRepeat, loopExit);
         }
     }
-    else if (FFlag::LuauLowerAltLoopForn)
-    {
-        build.inst(IrCmd::INTERRUPT, build.constUint(pcpos));
-
-        IrOp limit = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 0));
-        IrOp step = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 1));
-
-        IrOp idx = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 2));
-        idx = build.inst(IrCmd::ADD_NUM, idx, step);
-        build.inst(IrCmd::STORE_DOUBLE, build.vmReg(ra + 2), idx);
-
-        build.inst(IrCmd::JUMP_FORN_LOOP_COND, idx, limit, step, loopRepeat, loopExit);
-    }
     else
     {
-        build.inst(IrCmd::INTERRUPT, build.constUint(pcpos));
+        double stepN = build.function.doubleOp(stepK);
 
-        IrOp zero = build.constDouble(0.0);
+        // Condition to continue the loop: step > 0 ? idx <= limit : limit <= idx
-        IrOp limit = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 0));
+        if (stepN > 0)
-        IrOp step = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 1));
+            build.inst(IrCmd::JUMP_CMP_NUM, idx, limit, build.cond(IrCondition::LessEqual), loopRepeat, loopExit);
-
+        else
-        IrOp idx = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(ra + 2));
+            build.inst(IrCmd::JUMP_CMP_NUM, limit, idx, build.cond(IrCondition::LessEqual), loopRepeat, loopExit);
-        idx = build.inst(IrCmd::ADD_NUM, idx, step);
-        build.inst(IrCmd::STORE_DOUBLE, build.vmReg(ra + 2), idx);
-
-        IrOp direct = build.block(IrBlockKind::Internal);
-        IrOp reverse = build.block(IrBlockKind::Internal);
-
-        // step <= 0
-        build.inst(IrCmd::JUMP_CMP_NUM, step, zero, build.cond(IrCondition::LessEqual), reverse, direct);
-
-        // step <= 0 is false, check idx <= limit
-        build.beginBlock(direct);
-        build.inst(IrCmd::JUMP_CMP_NUM, idx, limit, build.cond(IrCondition::LessEqual), loopRepeat, loopExit);
-
-        // step <= 0 is true, check limit <= idx
-        build.beginBlock(reverse);
-        build.inst(IrCmd::JUMP_CMP_NUM, limit, idx, build.cond(IrCondition::LessEqual), loopRepeat, loopExit);
     }
 
     // Fallthrough in original bytecode is implicit, so we start next internal block here

CodeGen/src/OptimizeConstProp.cpp

@@ -16,9 +16,8 @@ LUAU_FASTINTVARIABLE(LuauCodeGenMinLinearBlockPath, 3)
 LUAU_FASTINTVARIABLE(LuauCodeGenReuseSlotLimit, 64)
 LUAU_FASTFLAGVARIABLE(DebugLuauAbortingChecks, false)
 LUAU_FASTFLAGVARIABLE(LuauReuseHashSlots2, false)
-LUAU_FASTFLAGVARIABLE(LuauKeepVmapLinear, false)
 LUAU_FASTFLAGVARIABLE(LuauMergeTagLoads, false)
-LUAU_FASTFLAGVARIABLE(LuauReuseArrSlots, false)
+LUAU_FASTFLAGVARIABLE(LuauReuseArrSlots2, false)
 LUAU_FASTFLAG(LuauLowerAltLoopForn)
 
 namespace Luau
@@ -506,6 +505,19 @@ static void handleBuiltinEffects(ConstPropState& state, LuauBuiltinFunction bfid
     case LBF_TONUMBER:
     case LBF_TOSTRING:
     case LBF_BIT32_BYTESWAP:
+    case LBF_BUFFER_READI8:
+    case LBF_BUFFER_READU8:
+    case LBF_BUFFER_WRITEU8:
+    case LBF_BUFFER_READI16:
+    case LBF_BUFFER_READU16:
+    case LBF_BUFFER_WRITEU16:
+    case LBF_BUFFER_READI32:
+    case LBF_BUFFER_READU32:
+    case LBF_BUFFER_WRITEU32:
+    case LBF_BUFFER_READF32:
+    case LBF_BUFFER_WRITEF32:
+    case LBF_BUFFER_READF64:
+    case LBF_BUFFER_WRITEF64:
         break;
     case LBF_TABLE_INSERT:
         state.invalidateHeap();
@@ -941,7 +953,7 @@ static void constPropInInst(ConstPropState& state, IrBuilder& build, IrFunction&
     case IrCmd::LOAD_ENV:
         break;
     case IrCmd::GET_ARR_ADDR:
-        if (!FFlag::LuauReuseArrSlots)
+        if (!FFlag::LuauReuseArrSlots2)
             break;
 
         for (uint32_t prevIdx : state.getArrAddrCache)
@@ -1014,7 +1026,7 @@ static void constPropInInst(ConstPropState& state, IrBuilder& build, IrFunction&
     case IrCmd::DUP_TABLE:
         break;
     case IrCmd::TRY_NUM_TO_INDEX:
-        if (!FFlag::LuauReuseArrSlots)
+        if (!FFlag::LuauReuseArrSlots2)
             break;
 
         for (uint32_t prevIdx : state.tryNumToIndexCache)
@@ -1053,6 +1065,13 @@ static void constPropInInst(ConstPropState& state, IrBuilder& build, IrFunction&
     {
         std::optional<int> arrayIndex = function.asIntOp(inst.b.kind == IrOpKind::Constant ? inst.b : state.tryGetValue(inst.b));
 
+        // Negative offsets will jump to fallback, no need to keep the check
+        if (FFlag::LuauReuseArrSlots2 && arrayIndex && *arrayIndex < 0)
+        {
+            replace(function, block, index, {IrCmd::JUMP, inst.c});
+            break;
+        }
+
         if (RegisterInfo* info = state.tryGetRegisterInfo(inst.a); info && arrayIndex)
         {
             if (info->knownTableArraySize >= 0)
@@ -1069,11 +1088,11 @@ static void constPropInInst(ConstPropState& state, IrBuilder& build, IrFunction&
                     replace(function, block, index, {IrCmd::JUMP, inst.c});
                 }
 
-                return; // Break out from both the loop and the switch
+                break;
             }
         }
 
-        if (!FFlag::LuauReuseArrSlots)
+        if (!FFlag::LuauReuseArrSlots2)
             break;
 
         for (uint32_t prevIdx : state.checkArraySizeCache)
@@ -1087,7 +1106,7 @@ static void constPropInInst(ConstPropState& state, IrBuilder& build, IrFunction&
 
             // If arguments are different, in case they are both constant, we can check if a larger bound was already tested
             if (!sameBoundary && inst.b.kind == IrOpKind::Constant && prev.b.kind == IrOpKind::Constant &&
-                function.intOp(inst.b) < function.intOp(prev.b))
+                unsigned(function.intOp(inst.b)) < unsigned(function.intOp(prev.b)))
                 sameBoundary = true;
 
             if (sameBoundary)
@@ -1269,14 +1288,11 @@ static void constPropInBlock(IrBuilder& build, IrBlock& block, ConstPropState& s
         constPropInInst(state, build, function, block, inst, index);
     }
 
-    if (!FFlag::LuauKeepVmapLinear)
+    // Value numbering and load/store propagation is not performed between blocks
-    {
+    state.invalidateValuePropagation();
-        // Value numbering and load/store propagation is not performed between blocks
-        state.invalidateValuePropagation();
 
-        // Same for table slot data propagation
+    // Same for table slot data propagation
-        state.invalidateHeapTableData();
+    state.invalidateHeapTableData();
-    }
 }
 
 static void constPropInBlockChain(IrBuilder& build, std::vector<uint8_t>& visited, IrBlock* block, ConstPropState& state)
@@ -1296,16 +1312,6 @@ static void constPropInBlockChain(IrBuilder& build, std::vector<uint8_t>& visite
 
         constPropInBlock(build, *block, state);
 
-        if (FFlag::LuauKeepVmapLinear)
-        {
-            // Value numbering and load/store propagation is not performed between blocks right now
-            // This is because cross-block value uses limit creation of linear block (restriction in collectDirectBlockJumpPath)
-            state.invalidateValuePropagation();
-
-            // Same for table slot data propagation
-            state.invalidateHeapTableData();
-        }
-
         // Blocks in a chain are guaranteed to follow each other
         // We force that by giving all blocks the same sorting key, but consecutive chain keys
         block->sortkey = startSortkey;

Common/include/Luau/Bytecode.h

@@ -452,6 +452,7 @@ enum LuauBytecodeType
     LBC_TYPE_THREAD,
     LBC_TYPE_USERDATA,
     LBC_TYPE_VECTOR,
+    LBC_TYPE_BUFFER,
 
     LBC_TYPE_ANY = 15,
     LBC_TYPE_OPTIONAL_BIT = 1 << 7,
@@ -563,6 +564,21 @@ enum LuauBuiltinFunction
 
     // bit32.byteswap(n)
     LBF_BIT32_BYTESWAP,
+
+    // buffer.
+    LBF_BUFFER_READI8,
+    LBF_BUFFER_READU8,
+    LBF_BUFFER_WRITEU8,
+    LBF_BUFFER_READI16,
+    LBF_BUFFER_READU16,
+    LBF_BUFFER_WRITEU16,
+    LBF_BUFFER_READI32,
+    LBF_BUFFER_READU32,
+    LBF_BUFFER_WRITEU32,
+    LBF_BUFFER_READF32,
+    LBF_BUFFER_WRITEF32,
+    LBF_BUFFER_READF64,
+    LBF_BUFFER_WRITEF64,
 };
 
 // Capture type, used in LOP_CAPTURE

Compiler/src/Builtins.cpp

@@ -6,6 +6,8 @@
 
 LUAU_FASTFLAGVARIABLE(LuauBit32ByteswapBuiltin, false)
 
+LUAU_FASTFLAGVARIABLE(LuauBufferBuiltins, false)
+
 namespace Luau
 {
 namespace Compile
@@ -192,6 +194,36 @@ static int getBuiltinFunctionId(const Builtin& builtin, const CompileOptions& op
             return LBF_TABLE_UNPACK;
     }
 
+    if (FFlag::LuauBufferBuiltins && builtin.object == "buffer")
+    {
+        if (builtin.method == "readi8")
+            return LBF_BUFFER_READI8;
+        if (builtin.method == "readu8")
+            return LBF_BUFFER_READU8;
+        if (builtin.method == "writei8" || builtin.method == "writeu8")
+            return LBF_BUFFER_WRITEU8;
+        if (builtin.method == "readi16")
+            return LBF_BUFFER_READI16;
+        if (builtin.method == "readu16")
+            return LBF_BUFFER_READU16;
+        if (builtin.method == "writei16" || builtin.method == "writeu16")
+            return LBF_BUFFER_WRITEU16;
+        if (builtin.method == "readi32")
+            return LBF_BUFFER_READI32;
+        if (builtin.method == "readu32")
+            return LBF_BUFFER_READU32;
+        if (builtin.method == "writei32" || builtin.method == "writeu32")
+            return LBF_BUFFER_WRITEU32;
+        if (builtin.method == "readf32")
+            return LBF_BUFFER_READF32;
+        if (builtin.method == "writef32")
+            return LBF_BUFFER_WRITEF32;
+        if (builtin.method == "readf64")
+            return LBF_BUFFER_READF64;
+        if (builtin.method == "writef64")
+            return LBF_BUFFER_WRITEF64;
+    }
+
     if (options.vectorCtor)
     {
         if (options.vectorLib)
@@ -409,6 +441,23 @@ BuiltinInfo getBuiltinInfo(int bfid)
 
     case LBF_BIT32_BYTESWAP:
         return {1, 1, BuiltinInfo::Flag_NoneSafe};
+
+    case LBF_BUFFER_READI8:
+    case LBF_BUFFER_READU8:
+    case LBF_BUFFER_READI16:
+    case LBF_BUFFER_READU16:
+    case LBF_BUFFER_READI32:
+    case LBF_BUFFER_READU32:
+    case LBF_BUFFER_READF32:
+    case LBF_BUFFER_READF64:
+        return {2, 1, BuiltinInfo::Flag_NoneSafe};
+
+    case LBF_BUFFER_WRITEU8:
+    case LBF_BUFFER_WRITEU16:
+    case LBF_BUFFER_WRITEU32:
+    case LBF_BUFFER_WRITEF32:
+    case LBF_BUFFER_WRITEF64:
+        return {3, 0, BuiltinInfo::Flag_NoneSafe};
     };
 
     LUAU_UNREACHABLE();

Compiler/src/BytecodeBuilder.cpp

@@ -2296,6 +2296,8 @@ static const char* getBaseTypeString(uint8_t type)
         return "userdata";
     case LBC_TYPE_VECTOR:
         return "vector";
+    case LBC_TYPE_BUFFER:
+        return "buffer";
     case LBC_TYPE_ANY:
         return "any";
     }

Compiler/src/Compiler.cpp

@@ -26,13 +26,8 @@ LUAU_FASTINTVARIABLE(LuauCompileInlineThreshold, 25)
 LUAU_FASTINTVARIABLE(LuauCompileInlineThresholdMaxBoost, 300)
 LUAU_FASTINTVARIABLE(LuauCompileInlineDepth, 5)
 
-LUAU_FASTFLAGVARIABLE(LuauCompileFenvNoBuiltinFold, false)
-LUAU_FASTFLAGVARIABLE(LuauCompileTopCold, false)
-
 LUAU_FASTFLAG(LuauFloorDivision)
 LUAU_FASTFLAGVARIABLE(LuauCompileFixContinueValidation2, false)
-
-LUAU_FASTFLAGVARIABLE(LuauCompileContinueCloseUpvals, false)
 LUAU_FASTFLAGVARIABLE(LuauCompileIfElseAndOr, false)
 
 namespace Luau
@@ -267,7 +262,7 @@ struct Compiler
             CompileError::raise(func->location, "Exceeded function instruction limit; split the function into parts to compile");
 
         // since top-level code only executes once, it can be marked as cold if it has no loops (top-level code with loops might be profitable to compile natively)
-        if (FFlag::LuauCompileTopCold && func->functionDepth == 0 && !hasLoops)
+        if (func->functionDepth == 0 && !hasLoops)
             protoflags |= LPF_NATIVE_COLD;
 
         bytecode.endFunction(uint8_t(stackSize), uint8_t(upvals.size()), protoflags);
@@ -2649,8 +2644,7 @@ struct Compiler
             // (but it must still close upvalues defined in more nested blocks)
             // this is because the upvalues defined inside the loop body may be captured by a closure defined in the until
             // expression that continue will jump to.
-            if (FFlag::LuauCompileContinueCloseUpvals)
+            loops.back().localOffsetContinue = localStack.size();
-                loops.back().localOffsetContinue = localStack.size();
 
             // if continue was called from this statement, then any local defined after this in the loop body should not be accessed by until condition
             // it is sufficient to check this condition once, as if this holds for the first continue, it must hold for all subsequent continues.
@@ -4016,7 +4010,7 @@ void compileOrThrow(BytecodeBuilder& bytecode, const ParseResult& parseResult, c
     }
 
     // builtin folding is enabled on optimization level 2 since we can't deoptimize folding at runtime
-    if (options.optimizationLevel >= 2 && (!FFlag::LuauCompileFenvNoBuiltinFold || (!compiler.getfenvUsed && !compiler.setfenvUsed)))
+    if (options.optimizationLevel >= 2 && (!compiler.getfenvUsed && !compiler.setfenvUsed))
     {
         compiler.builtinsFold = &compiler.builtins;
 

Compiler/src/Types.cpp

@@ -3,6 +3,8 @@
 
 #include "Luau/BytecodeBuilder.h"
 
+LUAU_FASTFLAGVARIABLE(LuauCompileBufferAnnotation, false)
+
 namespace Luau
 {
 
@@ -27,6 +29,8 @@ static LuauBytecodeType getPrimitiveType(AstName name)
         return LBC_TYPE_STRING;
     else if (name == "thread")
         return LBC_TYPE_THREAD;
+    else if (FFlag::LuauCompileBufferAnnotation && name == "buffer")
+        return LBC_TYPE_BUFFER;
     else if (name == "any" || name == "unknown")
         return LBC_TYPE_ANY;
     else

Sources.cmake

@@ -281,6 +281,7 @@ target_sources(Luau.VM PRIVATE
     VM/src/lbaselib.cpp
     VM/src/lbitlib.cpp
     VM/src/lbuffer.cpp
+    VM/src/lbuflib.cpp
     VM/src/lbuiltins.cpp
     VM/src/lcorolib.cpp
     VM/src/ldblib.cpp

VM/include/lualib.h

@@ -124,6 +124,9 @@ LUALIB_API int luaopen_string(lua_State* L);
 #define LUA_BITLIBNAME "bit32"
 LUALIB_API int luaopen_bit32(lua_State* L);
 
+#define LUA_BUFFERLIBNAME "buffer"
+LUALIB_API int luaopen_buffer(lua_State* L);
+
 #define LUA_UTF8LIBNAME "utf8"
 LUALIB_API int luaopen_utf8(lua_State* L);
 

VM/src/lbuflib.cpp

@@ -0,0 +1,286 @@
+// This file is part of the Luau programming language and is licensed under MIT License; see LICENSE.txt for details
+#include "lualib.h"
+
+#include "lcommon.h"
+#include "lbuffer.h"
+
+#if defined(LUAU_BIG_ENDIAN)
+#include <endian.h>
+#endif
+
+#include <string.h>
+
+// while C API returns 'size_t' for binary compatibility in case of future extensions,
+// in the current implementation, length and offset are limited to 31 bits
+// because offset is limited to an integer, a single 64bit comparison can be used and will not overflow
+#define isoutofbounds(offset, len, accessize) (uint64_t(unsigned(offset)) + (accessize) > uint64_t(len))
+
+static_assert(MAX_BUFFER_SIZE <= INT_MAX, "current implementation can't handle a larger limit");
+
+#if defined(LUAU_BIG_ENDIAN)
+template<typename T>
+inline T buffer_swapbe(T v)
+{
+    if (sizeof(T) == 8)
+        return htole64(v);
+    else if (sizeof(T) == 4)
+        return htole32(v);
+    else if (sizeof(T) == 2)
+        return htole16(v);
+    else
+        return v;
+}
+#endif
+
+static int buffer_create(lua_State* L)
+{
+    int size = luaL_checkinteger(L, 1);
+
+    if (size < 0)
+        luaL_error(L, "size cannot be negative");
+
+    lua_newbuffer(L, size);
+    return 1;
+}
+
+static int buffer_fromstring(lua_State* L)
+{
+    size_t len = 0;
+    const char* val = luaL_checklstring(L, 1, &len);
+
+    void* data = lua_newbuffer(L, len);
+    memcpy(data, val, len);
+    return 1;
+}
+
+static int buffer_tostring(lua_State* L)
+{
+    size_t len = 0;
+    void* data = luaL_checkbuffer(L, 1, &len);
+
+    lua_pushlstring(L, (char*)data, len);
+    return 1;
+}
+
+template<typename T>
+static int buffer_readinteger(lua_State* L)
+{
+    size_t len = 0;
+    void* buf = luaL_checkbuffer(L, 1, &len);
+    int offset = luaL_checkinteger(L, 2);
+
+    if (isoutofbounds(offset, len, sizeof(T)))
+        luaL_error(L, "buffer access out of bounds");
+
+    T val;
+    memcpy(&val, (char*)buf + offset, sizeof(T));
+
+#if defined(LUAU_BIG_ENDIAN)
+    val = buffer_swapbe(val);
+#endif
+
+    lua_pushnumber(L, double(val));
+    return 1;
+}
+
+template<typename T>
+static int buffer_writeinteger(lua_State* L)
+{
+    size_t len = 0;
+    void* buf = luaL_checkbuffer(L, 1, &len);
+    int offset = luaL_checkinteger(L, 2);
+    int value = luaL_checkunsigned(L, 3);
+
+    if (isoutofbounds(offset, len, sizeof(T)))
+        luaL_error(L, "buffer access out of bounds");
+
+    T val = T(value);
+
+#if defined(LUAU_BIG_ENDIAN)
+    val = buffer_swapbe(val);
+#endif
+
+    memcpy((char*)buf + offset, &val, sizeof(T));
+    return 0;
+}
+
+template<typename T, typename StorageType>
+static int buffer_readfp(lua_State* L)
+{
+    size_t len = 0;
+    void* buf = luaL_checkbuffer(L, 1, &len);
+    int offset = luaL_checkinteger(L, 2);
+
+    if (isoutofbounds(offset, len, sizeof(T)))
+        luaL_error(L, "buffer access out of bounds");
+
+    T val;
+
+#if defined(LUAU_BIG_ENDIAN)
+    static_assert(sizeof(T) == sizeof(StorageType), "type size must match to reinterpret data");
+    StorageType tmp;
+    memcpy(&tmp, (char*)buf + offset, sizeof(tmp));
+    tmp = buffer_swapbe(tmp);
+
+    memcpy(&val, &tmp, sizeof(tmp));
+#else
+    memcpy(&val, (char*)buf + offset, sizeof(T));
+#endif
+
+    lua_pushnumber(L, double(val));
+    return 1;
+}
+
+template<typename T, typename StorageType>
+static int buffer_writefp(lua_State* L)
+{
+    size_t len = 0;
+    void* buf = luaL_checkbuffer(L, 1, &len);
+    int offset = luaL_checkinteger(L, 2);
+    double value = luaL_checknumber(L, 3);
+
+    if (isoutofbounds(offset, len, sizeof(T)))
+        luaL_error(L, "buffer access out of bounds");
+
+    T val = T(value);
+
+#if defined(LUAU_BIG_ENDIAN)
+    static_assert(sizeof(T) == sizeof(StorageType), "type size must match to reinterpret data");
+    StorageType tmp;
+    memcpy(&tmp, &val, sizeof(tmp));
+    tmp = buffer_swapbe(tmp);
+
+    memcpy((char*)buf + offset, &tmp, sizeof(tmp));
+#else
+    memcpy((char*)buf + offset, &val, sizeof(T));
+#endif
+
+    return 0;
+}
+
+static int buffer_readstring(lua_State* L)
+{
+    size_t len = 0;
+    void* buf = luaL_checkbuffer(L, 1, &len);
+    int offset = luaL_checkinteger(L, 2);
+    int size = luaL_checkinteger(L, 3);
+
+    if (size < 0)
+        luaL_error(L, "size cannot be negative");
+
+    if (isoutofbounds(offset, len, unsigned(size)))
+        luaL_error(L, "buffer access out of bounds");
+
+    lua_pushlstring(L, (char*)buf + offset, size);
+    return 1;
+}
+
+static int buffer_writestring(lua_State* L)
+{
+    size_t len = 0;
+    void* buf = luaL_checkbuffer(L, 1, &len);
+    int offset = luaL_checkinteger(L, 2);
+    size_t size = 0;
+    const char* val = luaL_checklstring(L, 3, &size);
+    int count = luaL_optinteger(L, 4, int(size));
+
+    if (count < 0)
+        luaL_error(L, "count cannot be negative");
+
+    if (size_t(count) > size)
+        luaL_error(L, "string length overflow");
+
+    // string size can't exceed INT_MAX at this point
+    if (isoutofbounds(offset, len, unsigned(count)))
+        luaL_error(L, "buffer access out of bounds");
+
+    memcpy((char*)buf + offset, val, count);
+    return 0;
+}
+
+static int buffer_len(lua_State* L)
+{
+    size_t len = 0;
+    luaL_checkbuffer(L, 1, &len);
+
+    lua_pushnumber(L, double(unsigned(len)));
+    return 1;
+}
+
+static int buffer_copy(lua_State* L)
+{
+    size_t tlen = 0;
+    void* tbuf = luaL_checkbuffer(L, 1, &tlen);
+    int toffset = luaL_checkinteger(L, 2);
+
+    size_t slen = 0;
+    void* sbuf = luaL_checkbuffer(L, 3, &slen);
+    int soffset = luaL_optinteger(L, 4, 0);
+
+    int size = luaL_optinteger(L, 5, int(slen) - soffset);
+
+    if (size < 0)
+        luaL_error(L, "buffer access out of bounds");
+
+    if (isoutofbounds(soffset, slen, unsigned(size)))
+        luaL_error(L, "buffer access out of bounds");
+
+    if (isoutofbounds(toffset, tlen, unsigned(size)))
+        luaL_error(L, "buffer access out of bounds");
+
+    memmove((char*)tbuf + toffset, (char*)sbuf + soffset, size);
+    return 0;
+}
+
+static int buffer_fill(lua_State* L)
+{
+    size_t len = 0;
+    void* buf = luaL_checkbuffer(L, 1, &len);
+    int offset = luaL_checkinteger(L, 2);
+    unsigned value = luaL_checkunsigned(L, 3);
+    int size = luaL_optinteger(L, 4, int(len) - offset);
+
+    if (size < 0)
+        luaL_error(L, "buffer access out of bounds");
+
+    if (isoutofbounds(offset, len, unsigned(size)))
+        luaL_error(L, "buffer access out of bounds");
+
+    memset((char*)buf + offset, value & 0xff, size);
+    return 0;
+}
+
+static const luaL_Reg bufferlib[] = {
+    {"create", buffer_create},
+    {"fromstring", buffer_fromstring},
+    {"tostring", buffer_tostring},
+    {"readi8", buffer_readinteger<int8_t>},
+    {"readu8", buffer_readinteger<uint8_t>},
+    {"readi16", buffer_readinteger<int16_t>},
+    {"readu16", buffer_readinteger<uint16_t>},
+    {"readi32", buffer_readinteger<int32_t>},
+    {"readu32", buffer_readinteger<uint32_t>},
+    {"readf32", buffer_readfp<float, uint32_t>},
+    {"readf64", buffer_readfp<double, uint64_t>},
+    {"writei8", buffer_writeinteger<int8_t>},
+    {"writeu8", buffer_writeinteger<uint8_t>},
+    {"writei16", buffer_writeinteger<int16_t>},
+    {"writeu16", buffer_writeinteger<uint16_t>},
+    {"writei32", buffer_writeinteger<int32_t>},
+    {"writeu32", buffer_writeinteger<uint32_t>},
+    {"writef32", buffer_writefp<float, uint32_t>},
+    {"writef64", buffer_writefp<double, uint64_t>},
+    {"readstring", buffer_readstring},
+    {"writestring", buffer_writestring},
+    {"len", buffer_len},
+    {"copy", buffer_copy},
+    {"fill", buffer_fill},
+    {NULL, NULL},
+};
+
+int luaopen_buffer(lua_State* L)
+{
+    luaL_register(L, LUA_BUFFERLIBNAME, bufferlib);
+
+    return 1;
+}

VM/src/lbuiltins.cpp

@@ -8,8 +8,10 @@
 #include "lgc.h"
 #include "lnumutils.h"
 #include "ldo.h"
+#include "lbuffer.h"
 
 #include <math.h>
+#include <string.h>
 
 #ifdef _MSC_VER
 #include <intrin.h>
@@ -1336,6 +1338,95 @@ static int luauF_byteswap(lua_State* L, StkId res, TValue* arg0, int nresults, S
     return -1;
 }
 
+// because offset is limited to an integer, a single 64bit comparison can be used and will not overflow
+#define checkoutofbounds(offset, len, accessize) (uint64_t(unsigned(offset)) + (accessize - 1) >= uint64_t(len))
+
+template<typename T>
+static int luauF_readinteger(lua_State* L, StkId res, TValue* arg0, int nresults, StkId args, int nparams)
+{
+#if !defined(LUAU_BIG_ENDIAN)
+    if (nparams >= 2 && nresults <= 1 && ttisbuffer(arg0) && ttisnumber(args))
+    {
+        int offset;
+        luai_num2int(offset, nvalue(args));
+        if (checkoutofbounds(offset, bufvalue(arg0)->len, sizeof(T)))
+            return -1;
+
+        T val;
+        memcpy(&val, (char*)bufvalue(arg0)->data + offset, sizeof(T));
+        setnvalue(res, double(val));
+        return 1;
+    }
+#endif
+
+    return -1;
+}
+
+template<typename T>
+static int luauF_writeinteger(lua_State* L, StkId res, TValue* arg0, int nresults, StkId args, int nparams)
+{
+#if !defined(LUAU_BIG_ENDIAN)
+    if (nparams >= 3 && nresults <= 0 && ttisbuffer(arg0) && ttisnumber(args) && ttisnumber(args + 1))
+    {
+        int offset;
+        luai_num2int(offset, nvalue(args));
+        if (checkoutofbounds(offset, bufvalue(arg0)->len, sizeof(T)))
+            return -1;
+
+        unsigned value;
+        double incoming = nvalue(args + 1);
+        luai_num2unsigned(value, incoming);
+
+        T val = T(value);
+        memcpy((char*)bufvalue(arg0)->data + offset, &val, sizeof(T));
+        return 0;
+    }
+#endif
+
+    return -1;
+}
+
+template<typename T>
+static int luauF_readfp(lua_State* L, StkId res, TValue* arg0, int nresults, StkId args, int nparams)
+{
+#if !defined(LUAU_BIG_ENDIAN)
+    if (nparams >= 2 && nresults <= 1 && ttisbuffer(arg0) && ttisnumber(args))
+    {
+        int offset;
+        luai_num2int(offset, nvalue(args));
+        if (checkoutofbounds(offset, bufvalue(arg0)->len, sizeof(T)))
+            return -1;
+
+        T val;
+        memcpy(&val, (char*)bufvalue(arg0)->data + offset, sizeof(T));
+        setnvalue(res, double(val));
+        return 1;
+    }
+#endif
+
+    return -1;
+}
+
+template<typename T>
+static int luauF_writefp(lua_State* L, StkId res, TValue* arg0, int nresults, StkId args, int nparams)
+{
+#if !defined(LUAU_BIG_ENDIAN)
+    if (nparams >= 3 && nresults <= 0 && ttisbuffer(arg0) && ttisnumber(args) && ttisnumber(args + 1))
+    {
+        int offset;
+        luai_num2int(offset, nvalue(args));
+        if (checkoutofbounds(offset, bufvalue(arg0)->len, sizeof(T)))
+            return -1;
+
+        T val = T(nvalue(args + 1));
+        memcpy((char*)bufvalue(arg0)->data + offset, &val, sizeof(T));
+        return 0;
+    }
+#endif
+
+    return -1;
+}
+
 static int luauF_missing(lua_State* L, StkId res, TValue* arg0, int nresults, StkId args, int nparams)
 {
     return -1;
@@ -1505,6 +1596,20 @@ const luau_FastFunction luauF_table[256] = {
 
     luauF_byteswap,
 
+    luauF_readinteger<int8_t>,
+    luauF_readinteger<uint8_t>,
+    luauF_writeinteger<uint8_t>,
+    luauF_readinteger<int16_t>,
+    luauF_readinteger<uint16_t>,
+    luauF_writeinteger<uint16_t>,
+    luauF_readinteger<int32_t>,
+    luauF_readinteger<uint32_t>,
+    luauF_writeinteger<uint32_t>,
+    luauF_readfp<float>,
+    luauF_writefp<float>,
+    luauF_readfp<double>,
+    luauF_writefp<double>,
+
 // When adding builtins, add them above this line; what follows is 64 "dummy" entries with luauF_missing fallback.
 // This is important so that older versions of the runtime that don't support newer builtins automatically fall back via luauF_missing.
 // Given the builtin addition velocity this should always provide a larger compatibility window than bytecode versions suggest.

VM/src/linit.cpp

@@ -14,6 +14,7 @@ static const luaL_Reg lualibs[] = {
     {LUA_DBLIBNAME, luaopen_debug},
     {LUA_UTF8LIBNAME, luaopen_utf8},
     {LUA_BITLIBNAME, luaopen_bit32},
+    {LUA_BUFFERLIBNAME, luaopen_buffer},
     {NULL, NULL},
 };
 

fuzz/protoprint.cpp

@@ -3,13 +3,13 @@
 
 static const std::string kNames[] = {
     "_G",
-    "_LOADED",
     "_VERSION",
     "__add",
     "__call",
     "__concat",
     "__div",
     "__eq",
+    "__idiv",
     "__index",
     "__iter",
     "__le",
@@ -37,18 +37,22 @@ static const std::string kNames[] = {
     "boolean",
     "bor",
     "btest",
+    "buffer",
     "bxor",
     "byte",
     "ceil",
     "char",
     "charpattern",
     "clamp",
+    "clear",
     "clock",
     "clone",
     "close",
     "codepoint",
     "codes",
+    "collectgarbage",
     "concat",
+    "copy",
     "coroutine",
     "cos",
     "cosh",
@@ -62,18 +66,19 @@ static const std::string kNames[] = {
     "error",
     "exp",
     "extract",
+    "fill",
     "find",
     "floor",
     "fmod",
     "foreach",
     "foreachi",
     "format",
-    "frexp",
     "freeze",
+    "frexp",
+    "fromstring",
     "function",
     "gcinfo",
     "getfenv",
-    "getinfo",
     "getmetatable",
     "getn",
     "gmatch",
@@ -118,13 +123,24 @@ static const std::string kNames[] = {
     "randomseed",
     "rawequal",
     "rawget",
+    "rawlen",
     "rawset",
+    "readf32",
+    "readf64",
+    "readi16",
+    "readi32",
+    "readi8",
+    "readstring",
+    "readu16",
+    "readu32",
+    "readu8",
     "remove",
     "rep",
     "replace",
     "require",
     "resume",
     "reverse",
+    "round",
     "rrotate",
     "rshift",
     "running",
@@ -138,7 +154,6 @@ static const std::string kNames[] = {
     "split",
     "sqrt",
     "status",
-    "stdin",
     "string",
     "sub",
     "table",
@@ -148,6 +163,7 @@ static const std::string kNames[] = {
     "time",
     "tonumber",
     "tostring",
+    "tostring",
     "traceback",
     "type",
     "typeof",
@@ -157,17 +173,28 @@ static const std::string kNames[] = {
     "utf8",
     "vector",
     "wrap",
+    "writef32",
+    "writef64",
+    "writei16",
+    "writei32",
+    "writei8",
+    "writestring",
+    "writeu16",
+    "writeu32",
+    "writeu8",
     "xpcall",
     "yield",
 };
 
 static const std::string kTypes[] = {
     "any",
+    "boolean",
+    "buffer",
     "nil",
     "number",
     "string",
-    "boolean",
     "thread",
+    "vector",
 };
 
 static const std::string kClasses[] = {

tests/AssemblyBuilderX64.test.cpp

@@ -588,6 +588,7 @@ TEST_CASE("LogTest")
     build.cmp(rsi, rdi);
     build.jcc(ConditionX64::Equal, start);
     build.lea(rcx, start);
+    build.lea(rcx, addr[rdx]);
 
     build.jmp(qword[rdx]);
     build.vaddps(ymm9, ymm12, ymmword[rbp + 0xc]);
@@ -634,6 +635,7 @@ TEST_CASE("LogTest")
  cmp         rsi,rdi
  je          .L1
  lea         rcx,.L1
+ lea         rcx,[rdx]
  jmp         qword ptr [rdx]
  vaddps      ymm9,ymm12,ymmword ptr [rbp+0Ch]
  vaddpd      ymm2,ymm7,qword ptr [.start-8]

tests/Compiler.test.cpp

@@ -89,7 +89,7 @@ TEST_CASE("BytecodeIsStable")
 
     // Bytecode type encoding (serialized & in-memory)
     // Note: these *can* change retroactively *if* type version is bumped, but probably shouldn't
-    LUAU_ASSERT(LBC_TYPE_VECTOR == 8); // type version 1
+    LUAU_ASSERT(LBC_TYPE_BUFFER == 9); // type version 1
 }
 
 TEST_CASE("CompileToBytecode")
@@ -1772,8 +1772,6 @@ RETURN R0 0
 
 TEST_CASE("LoopContinueUntil")
 {
-    ScopedFastFlag sff("LuauCompileContinueCloseUpvals", true);
-
     // it's valid to use locals defined inside the loop in until expression if they're defined before continue
     CHECK_EQ("\n" + compileFunction0("repeat local r = math.random() if r > 0.5 then continue end r = r + 0.3 until r < 0.5"), R"(
 L0: GETIMPORT R0 2 [math.random]
@@ -2026,8 +2024,6 @@ end
 
 TEST_CASE("LoopContinueUntilCapture")
 {
-    ScopedFastFlag sff("LuauCompileContinueCloseUpvals", true);
-
     // validate continue upvalue closing behavior: continue must close locals defined in the nested scopes
     // but can't close locals defined in the loop scope - these are visible to the condition and will be closed
     // when evaluating the condition instead.
@@ -7586,8 +7582,6 @@ RETURN R0 1
 
 TEST_CASE("NoBuiltinFoldFenv")
 {
-    ScopedFastFlag sff("LuauCompileFenvNoBuiltinFold", true);
-
     // builtin folding is disabled when getfenv/setfenv is used in the module
     CHECK_EQ("\n" + compileFunction(R"(
 getfenv()

tests/Conformance.test.cpp

@@ -289,11 +289,15 @@ TEST_CASE("Assert")
 TEST_CASE("Basic")
 {
     ScopedFastFlag sffs{"LuauFloorDivision", true};
-    ScopedFastFlag sfff{"LuauImproveForN2", true};
 
     runConformance("basic.lua");
 }
 
+TEST_CASE("Buffers")
+{
+    runConformance("buffers.lua");
+}
+
 TEST_CASE("Math")
 {
     runConformance("math.lua");
@@ -381,8 +385,6 @@ TEST_CASE("Events")
 
 TEST_CASE("Constructs")
 {
-    ScopedFastFlag sff("LuauCompileContinueCloseUpvals", true);
-
     runConformance("constructs.lua");
 }
 
@@ -558,6 +560,8 @@ static void populateRTTI(lua_State* L, Luau::TypeId type)
 
 TEST_CASE("Types")
 {
+    ScopedFastFlag luauBufferDefinitions{"LuauBufferDefinitions", true};
+
     runConformance("types.lua", [](lua_State* L) {
         Luau::NullModuleResolver moduleResolver;
         Luau::NullFileResolver fileResolver;
@@ -1507,68 +1511,81 @@ TEST_CASE("Interrupt")
     lua_CompileOptions copts = defaultOptions();
     copts.optimizationLevel = 1; // disable loop unrolling to get fixed expected hit results
 
-    static const int expectedhits[] = {
-        2,
-        9,
-        5,
-        5,
-        5,
-        5,
-        5,
-        5,
-        5,
-        5,
-        5,
-        5,
-        5,
-        6,
-        18,
-        13,
-        13,
-        13,
-        13,
-        16,
-        23,
-        21,
-        25,
-    };
     static int index;
 
-    index = 0;
+    StateRef globalState = runConformance("interrupt.lua", nullptr, nullptr, nullptr, &copts);
 
-    runConformance(
+    lua_State* L = globalState.get();
-        "interrupt.lua",
-        [](lua_State* L) {
-            auto* cb = lua_callbacks(L);
 
-            // note: for simplicity here we setup the interrupt callback once
+    // note: for simplicity here we setup the interrupt callback when the test starts
-            // however, this carries a noticeable performance cost. in a real application,
+    // however, this carries a noticeable performance cost. in a real application,
-            // it's advised to set interrupt callback on a timer from a different thread,
+    // it's advised to set interrupt callback on a timer from a different thread,
-            // and set it back to nullptr once the interrupt triggered.
+    // and set it back to nullptr once the interrupt triggered.
-            cb->interrupt = [](lua_State* L, int gc) {
-                if (gc >= 0)
-                    return;
 
-                CHECK(index < int(std::size(expectedhits)));
+    // define the interrupt to check the expected hits
+    static const int expectedhits[] = {11, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 7, 8, 20, 15, 15, 15, 15, 18, 25, 23, 26};
 
-                lua_Debug ar = {};
+    lua_callbacks(L)->interrupt = [](lua_State* L, int gc) {
-                lua_getinfo(L, 0, "l", &ar);
+        if (gc >= 0)
+            return;
 
-                CHECK(ar.currentline == expectedhits[index]);
+        CHECK(index < int(std::size(expectedhits)));
 
-                index++;
+        lua_Debug ar = {};
+        lua_getinfo(L, 0, "l", &ar);
 
-                // check that we can yield inside an interrupt
+        CHECK(ar.currentline == expectedhits[index]);
-                if (index == 5)
-                    lua_yield(L, 0);
-            };
-        },
-        [](lua_State* L) {
-            CHECK(index == 5); // a single yield point
-        },
-        nullptr, &copts);
 
-    CHECK(index == int(std::size(expectedhits)));
+        index++;
+
+        // check that we can yield inside an interrupt
+        if (index == 4)
+            lua_yield(L, 0);
+    };
+
+    {
+        lua_State* T = lua_newthread(L);
+
+        lua_getglobal(T, "test");
+
+        index = 0;
+        int status = lua_resume(T, nullptr, 0);
+        CHECK(status == LUA_YIELD);
+        CHECK(index == 4);
+
+        status = lua_resume(T, nullptr, 0);
+        CHECK(status == LUA_OK);
+        CHECK(index == int(std::size(expectedhits)));
+
+        lua_pop(L, 1);
+    }
+
+    // redefine the interrupt to break after 10 iterations of a loop that would otherwise be infinite
+    // the test exposes a few global functions that we will call; the interrupt will force a yield
+    lua_callbacks(L)->interrupt = [](lua_State* L, int gc) {
+        if (gc >= 0)
+            return;
+
+        CHECK(index < 10);
+        if (++index == 10)
+            lua_yield(L, 0);
+    };
+
+    for (int test = 1; test <= 9; ++test)
+    {
+        lua_State* T = lua_newthread(L);
+
+        std::string name = "infloop" + std::to_string(test);
+        lua_getglobal(T, name.c_str());
+
+        index = 0;
+        int status = lua_resume(T, nullptr, 0);
+        CHECK(status == LUA_YIELD);
+        CHECK(index == 10);
+
+        // abandon the thread
+        lua_pop(L, 1);
+    }
 }
 
 TEST_CASE("UserdataApi")
@@ -1890,6 +1907,8 @@ TEST_CASE("NativeTypeAnnotations")
     if (!codegen || !luau_codegen_supported())
         return;
 
+    ScopedFastFlag luauCompileBufferAnnotation{"LuauCompileBufferAnnotation", true};
+
     lua_CompileOptions copts = defaultOptions();
     copts.vectorCtor = "vector";
     copts.vectorType = "vector";

tests/IrBuilder.test.cpp

@@ -2062,7 +2062,7 @@ bb_fallback_1:
 
 TEST_CASE_FIXTURE(IrBuilderFixture, "DuplicateArrayElemChecksSameIndex")
 {
-    ScopedFastFlag luauReuseHashSlots{"LuauReuseArrSlots", true};
+    ScopedFastFlag luauReuseHashSlots{"LuauReuseArrSlots2", true};
 
     IrOp block = build.block(IrBlockKind::Internal);
     IrOp fallback = build.block(IrBlockKind::Fallback);
@@ -2117,9 +2117,9 @@ bb_fallback_1:
 )");
 }
 
-TEST_CASE_FIXTURE(IrBuilderFixture, "DuplicateArrayElemChecksLowerIndex")
+TEST_CASE_FIXTURE(IrBuilderFixture, "DuplicateArrayElemChecksSameValue")
 {
-    ScopedFastFlag luauReuseHashSlots{"LuauReuseArrSlots", true};
+    ScopedFastFlag luauReuseHashSlots{"LuauReuseArrSlots2", true};
 
     IrOp block = build.block(IrBlockKind::Internal);
     IrOp fallback = build.block(IrBlockKind::Fallback);
@@ -2182,9 +2182,9 @@ bb_fallback_1:
 )");
 }
 
-TEST_CASE_FIXTURE(IrBuilderFixture, "DuplicateArrayElemChecksSameValue")
+TEST_CASE_FIXTURE(IrBuilderFixture, "DuplicateArrayElemChecksLowerIndex")
 {
-    ScopedFastFlag luauReuseHashSlots{"LuauReuseArrSlots", true};
+    ScopedFastFlag luauReuseHashSlots{"LuauReuseArrSlots2", true};
 
     IrOp block = build.block(IrBlockKind::Internal);
     IrOp fallback = build.block(IrBlockKind::Fallback);
@@ -2240,7 +2240,7 @@ bb_fallback_1:
 
 TEST_CASE_FIXTURE(IrBuilderFixture, "DuplicateArrayElemChecksInvalidations")
 {
-    ScopedFastFlag luauReuseHashSlots{"LuauReuseArrSlots", true};
+    ScopedFastFlag luauReuseHashSlots{"LuauReuseArrSlots2", true};
 
     IrOp block = build.block(IrBlockKind::Internal);
     IrOp fallback = build.block(IrBlockKind::Fallback);
@@ -2298,6 +2298,55 @@ bb_fallback_1:
 )");
 }
 
+TEST_CASE_FIXTURE(IrBuilderFixture, "ArrayElemChecksNegativeIndex")
+{
+    ScopedFastFlag luauReuseHashSlots{"LuauReuseArrSlots2", true};
+
+    IrOp block = build.block(IrBlockKind::Internal);
+    IrOp fallback = build.block(IrBlockKind::Fallback);
+
+    build.beginBlock(block);
+
+    // This roughly corresponds to 'return t[1] + t[0]'
+    IrOp table1 = build.inst(IrCmd::LOAD_POINTER, build.vmReg(1));
+    build.inst(IrCmd::CHECK_ARRAY_SIZE, table1, build.constInt(0), fallback);
+    IrOp elem1 = build.inst(IrCmd::GET_ARR_ADDR, table1, build.constInt(0));
+    IrOp value1 = build.inst(IrCmd::LOAD_TVALUE, elem1, build.constInt(0));
+    build.inst(IrCmd::STORE_TVALUE, build.vmReg(3), value1);
+
+    build.inst(IrCmd::CHECK_ARRAY_SIZE, table1, build.constInt(-1), fallback); // This will jump directly to fallback
+    IrOp elem2 = build.inst(IrCmd::GET_ARR_ADDR, table1, build.constInt(-1));
+    IrOp value1b = build.inst(IrCmd::LOAD_TVALUE, elem2, build.constInt(0));
+    build.inst(IrCmd::STORE_TVALUE, build.vmReg(4), value1b);
+
+    IrOp a = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(3));
+    IrOp b = build.inst(IrCmd::LOAD_DOUBLE, build.vmReg(4));
+    IrOp sum = build.inst(IrCmd::ADD_NUM, a, b);
+    build.inst(IrCmd::STORE_DOUBLE, build.vmReg(2), sum);
+
+    build.inst(IrCmd::RETURN, build.vmReg(2), build.constUint(1));
+
+    build.beginBlock(fallback);
+    build.inst(IrCmd::RETURN, build.vmReg(0), build.constUint(1));
+
+    updateUseCounts(build.function);
+    constPropInBlockChains(build, true);
+
+    CHECK("\n" + toString(build.function, /* includeUseInfo */ false) == R"(
+bb_0:
+   %0 = LOAD_POINTER R1
+   CHECK_ARRAY_SIZE %0, 0i, bb_fallback_1
+   %2 = GET_ARR_ADDR %0, 0i
+   %3 = LOAD_TVALUE %2, 0i
+   STORE_TVALUE R3, %3
+   JUMP bb_fallback_1
+
+bb_fallback_1:
+   RETURN R0, 1u
+
+)");
+}
+
 TEST_SUITE_END();
 
 TEST_SUITE_BEGIN("Analysis");

tests/IrCallWrapperX64.test.cpp

@@ -483,9 +483,9 @@ TEST_CASE_FIXTURE(IrCallWrapperX64Fixture, "ExtraCoverage")
     checkMatch(R"(
  vmovups     xmm2,xmmword ptr [r13]
  mov         rax,rcx
- lea         rcx,none ptr [r12+8]
+ lea         rcx,[r12+8]
  mov         rbx,rdx
- lea         rdx,none ptr [r12+010h]
+ lea         rdx,[r12+010h]
  call        qword ptr [rax+rbx]
 )");
 }
@@ -500,7 +500,7 @@ TEST_CASE_FIXTURE(IrCallWrapperX64Fixture, "AddressInStackArguments")
     callWrap.call(qword[r14]);
 
     checkMatch(R"(
- lea         rax,none ptr [r12+010h]
+ lea         rax,[r12+010h]
  mov         qword ptr [rsp+020h],rax
  mov         ecx,1
  mov         edx,2

tests/Parser.test.cpp

@@ -1317,16 +1317,53 @@ end
 
 TEST_CASE_FIXTURE(Fixture, "parse_error_with_too_many_nested_type_group")
 {
-    ScopedFastInt sfis{"LuauRecursionLimit", 20};
+    ScopedFastInt sfis{"LuauRecursionLimit", 10};
 
     matchParseError(
-        "function f(): (((((((((Fail))))))))) end", "Exceeded allowed recursion depth; simplify your type annotation to make the code compile");
+        "function f(): ((((((((((Fail)))))))))) end", "Exceeded allowed recursion depth; simplify your type annotation to make the code compile");
 
     matchParseError("function f(): () -> () -> () -> () -> () -> () -> () -> () -> () -> () -> () end",
         "Exceeded allowed recursion depth; simplify your type annotation to make the code compile");
 
-    matchParseError(
+    matchParseError("local t: {a: {b: {c: {d: {e: {f: {g: {h: {i: {j: {}}}}}}}}}}}",
-        "local t: {a: {b: {c: {d: {e: {f: {}}}}}}}", "Exceeded allowed recursion depth; simplify your type annotation to make the code compile");
+        "Exceeded allowed recursion depth; simplify your type annotation to make the code compile");
+
+    matchParseError("local f: ((((((((((Fail))))))))))", "Exceeded allowed recursion depth; simplify your type annotation to make the code compile");
+
+    matchParseError("local t: a & (b & (c & (d & (e & (f & (g & (h & (i & (j & nil)))))))))",
+        "Exceeded allowed recursion depth; simplify your type annotation to make the code compile");
+}
+
+TEST_CASE_FIXTURE(Fixture, "can_parse_complex_unions_successfully")
+{
+    ScopedFastInt sfis[] = {{"LuauRecursionLimit", 10}, {"LuauTypeLengthLimit", 10}};
+    ScopedFastFlag sff{"LuauBetterTypeUnionLimits", true};
+
+    parse(R"(
+local f:
+() -> ()
+|
+() -> ()
+|
+{a: number}
+|
+{b: number}
+|
+((number))
+|
+((number))
+|
+(a & (b & nil))
+|
+(a & (b & nil))
+)");
+
+    parse(R"(
+local f: a? | b? | c? | d? | e? | f? | g? | h?
+)");
+
+    matchParseError("local t: a & b & c & d & e & f & g & h & i & j & nil",
+        "Exceeded allowed type length; simplify your type annotation to make the code compile");
 }
 
 TEST_CASE_FIXTURE(Fixture, "parse_error_with_too_many_nested_if_statements")

tests/TypeInfer.annotations.test.cpp

@@ -13,23 +13,99 @@ using namespace Luau;
 
 TEST_SUITE_BEGIN("AnnotationTests");
 
-TEST_CASE_FIXTURE(Fixture, "check_against_annotations")
+TEST_CASE_FIXTURE(Fixture, "initializers_are_checked_against_annotations")
 {
     CheckResult result = check("local a: number = \"Hello Types!\"");
     LUAU_REQUIRE_ERROR_COUNT(1, result);
 }
 
-TEST_CASE_FIXTURE(Fixture, "check_multi_assign")
+TEST_CASE_FIXTURE(Fixture, "check_multi_initialize")
 {
-    CheckResult result = check("local a: number, b: string = \"994\", 888");
+    CheckResult result = check(R"(
-    CHECK_EQ(2, result.errors.size());
+        local a: number, b: string = "one", 2
+    )");
+
+    LUAU_REQUIRE_ERROR_COUNT(2, result);
+
+    CHECK(get<TypeMismatch>(result.errors[0]));
+    CHECK(get<TypeMismatch>(result.errors[1]));
 }
 
 TEST_CASE_FIXTURE(Fixture, "successful_check")
 {
-    CheckResult result = check("local a: number, b: string = 994, \"eight eighty eight\"");
+    CheckResult result = check(R"(
+        local a: number, b: string = 1, "two"
+    )");
+
     LUAU_REQUIRE_NO_ERRORS(result);
-    dumpErrors(result);
+}
+
+TEST_CASE_FIXTURE(Fixture, "assignments_are_checked_against_annotations")
+{
+    CheckResult result = check(R"(
+        local x: number = 1
+        x = "two"
+    )");
+
+    LUAU_REQUIRE_ERROR_COUNT(1, result);
+}
+
+TEST_CASE_FIXTURE(Fixture, "multi_assign_checks_against_annotations")
+{
+    CheckResult result = check(R"(
+        local a: number, b: string = 1, "two"
+        a, b = "one", 2
+    )");
+
+    LUAU_REQUIRE_ERROR_COUNT(2, result);
+
+    CHECK(Location{{2, 15}, {2, 20}} == result.errors[0].location);
+    CHECK(Location{{2, 22}, {2, 23}} == result.errors[1].location);
+}
+
+TEST_CASE_FIXTURE(Fixture, "assignment_cannot_transform_a_table_property_type")
+{
+    CheckResult result = check(R"(
+        local a = {x=0}
+        a.x = "one"
+    )");
+
+    LUAU_REQUIRE_ERROR_COUNT(1, result);
+
+    CHECK(Location{{2, 14}, {2, 19}} == result.errors[0].location);
+}
+
+TEST_CASE_FIXTURE(Fixture, "assignments_to_unannotated_parameters_can_transform_the_type")
+{
+    ScopedFastFlag sff{"DebugLuauDeferredConstraintResolution", true};
+
+    CheckResult result = check(R"(
+        function f(x)
+            x = 0
+            return x
+        end
+    )");
+
+    LUAU_REQUIRE_NO_ERRORS(result);
+
+    CHECK("(unknown) -> number" == toString(requireType("f")));
+}
+
+TEST_CASE_FIXTURE(Fixture, "assignments_to_annotated_parameters_are_checked")
+{
+    ScopedFastFlag sff{"DebugLuauDeferredConstraintResolution", true};
+
+    CheckResult result = check(R"(
+        function f(x: string)
+            x = 0
+            return x
+        end
+    )");
+
+    LUAU_REQUIRE_ERROR_COUNT(1, result);
+    CHECK(Location{{2, 16}, {2, 17}} == result.errors[0].location);
+
+    CHECK("(string) -> number" == toString(requireType("f")));
 }
 
 TEST_CASE_FIXTURE(Fixture, "variable_type_is_supertype")
@@ -42,6 +118,22 @@ TEST_CASE_FIXTURE(Fixture, "variable_type_is_supertype")
     LUAU_REQUIRE_NO_ERRORS(result);
 }
 
+TEST_CASE_FIXTURE(Fixture, "assignment_also_checks_subtyping")
+{
+    CheckResult result = check(R"(
+        function f(): number?
+            return nil
+        end
+        local x: number = 1
+        local y: number? = f()
+        x = y
+        y = x
+    )");
+
+    LUAU_REQUIRE_ERROR_COUNT(1, result);
+    CHECK(Location{{6, 12}, {6, 13}} == result.errors[0].location);
+}
+
 TEST_CASE_FIXTURE(Fixture, "function_parameters_can_have_annotations")
 {
     CheckResult result = check(R"(
@@ -191,7 +283,7 @@ TEST_CASE_FIXTURE(Fixture, "infer_type_of_value_a_via_typeof_with_assignment")
 TEST_CASE_FIXTURE(Fixture, "table_annotation")
 {
     CheckResult result = check(R"(
-        local x: {a: number, b: string}
+        local x: {a: number, b: string} = {a=2, b="three"}
         local y = x.a
         local z = x.b
     )");
@@ -391,7 +483,7 @@ TEST_CASE_FIXTURE(Fixture, "two_type_params")
 {
     CheckResult result = check(R"(
         type Map<K, V> = {[K]: V}
-        local m: Map<string, number> = {};
+        local m: Map<string, number> = {}
         local a = m['foo']
         local b = m[9]                  -- error here
     )");
@@ -572,8 +664,8 @@ TEST_CASE_FIXTURE(Fixture, "cloned_interface_maintains_pointers_between_definiti
     CHECK(isInArena(aType, mod.interfaceTypes));
     CHECK(isInArena(bType, mod.interfaceTypes));
 
-    CHECK_EQ(recordType, aType);
+    CHECK(toString(recordType, {true}) == toString(aType, {true}));
-    CHECK_EQ(recordType, bType);
+    CHECK(toString(recordType, {true}) == toString(bType, {true}));
 }
 
 TEST_CASE_FIXTURE(BuiltinsFixture, "use_type_required_from_another_file")

tests/conformance/bitwise.lua

@@ -101,6 +101,7 @@ assert(bit32.extract(0xa0001111, 28, 4) == 0xa)
 assert(bit32.extract(0xa0001111, 31, 1) == 1)
 assert(bit32.extract(0x50000111, 31, 1) == 0)
 assert(bit32.extract(0xf2345679, 0, 32) == 0xf2345679)
+assert(bit32.extract(0xa0001111, 0) == 1)
 assert(bit32.extract(0xa0001111, 16) == 0)
 assert(bit32.extract(0xa0001111, 31) == 1)
 assert(bit32.extract(42, 1, 3) == 5)

tests/conformance/buffers.lua

@@ -0,0 +1,540 @@
+-- This file is part of the Luau programming language and is licensed under MIT License; see LICENSE.txt for details
+print("testing byte buffer library")
+
+function call(fn, ...)
+  local ok, res = pcall(fn, ...)
+  assert(ok)
+  return res
+end
+
+function ecall(fn, ...)
+  local ok, err = pcall(fn, ...)
+  assert(not ok)
+  return err:sub((err:find(": ") or -1) + 2, #err)
+end
+
+local function simple_byte_reads()
+  local b = buffer.create(1024)
+
+  assert(buffer.len(b) == 1024)
+
+  assert(buffer.readi8(b, 5) == 0)
+  buffer.writei8(b, 10, 32)
+  assert(buffer.readi8(b, 10) == 32)
+  buffer.writei8(b, 15, 5)
+  buffer.writei8(b, 14, 4)
+  buffer.writei8(b, 13, 3)
+  buffer.writei8(b, 12, 2)
+  buffer.writei8(b, 11, 1)
+  assert(buffer.readi8(b, 11) == 1)
+  assert(buffer.readi8(b, 12) == 2)
+  assert(buffer.readi8(b, 13) == 3)
+  assert(buffer.readi8(b, 14) == 4)
+  assert(buffer.readi8(b, 15) == 5)
+
+  local x = buffer.readi8(b, 14) + buffer.readi8(b, 13)
+  assert(x == 7)
+end
+
+simple_byte_reads()
+
+local function offset_byte_reads(start: number)
+  local b = buffer.create(1024)
+
+  buffer.writei8(b, start, 32)
+  assert(buffer.readi8(b, start) == 32)
+  buffer.writei8(b, start + 5, 5)
+  buffer.writei8(b, start + 4, 4)
+  buffer.writei8(b, start + 3, 3)
+  buffer.writei8(b, start + 2, 2)
+  buffer.writei8(b, start + 1, 1)
+  assert(buffer.readi8(b, start + 1) == 1)
+  assert(buffer.readi8(b, start + 2) == 2)
+  assert(buffer.readi8(b, start + 3) == 3)
+  assert(buffer.readi8(b, start + 4) == 4)
+  assert(buffer.readi8(b, start + 5) == 5)
+
+  local x = buffer.readi8(b, start + 4) + buffer.readi8(b, start + 3)
+  assert(x == 7)
+end
+
+offset_byte_reads(5)
+offset_byte_reads(30)
+
+local function simple_float_reinterpret()
+  local b = buffer.create(1024)
+
+  buffer.writei32(b, 10, 0x3f800000)
+  local one = buffer.readf32(b, 10)
+  assert(one == 1.0)
+
+  buffer.writef32(b, 10, 2.75197)
+  local magic = buffer.readi32(b, 10)
+  assert(magic == 0x40302047)
+end
+
+simple_float_reinterpret()
+
+local function simple_double_reinterpret()
+  local b = buffer.create(1024)
+
+  buffer.writei32(b, 10, 0x00000000)
+  buffer.writei32(b, 14, 0x3ff00000)
+  local one = buffer.readf64(b, 10)
+  assert(one == 1.0)
+
+  buffer.writef64(b, 10, 1.437576533064206)
+  local magic1 = buffer.readi32(b, 10)
+  local magic2 = buffer.readi32(b, 14)
+
+  assert(magic1 == 0x40302010)
+  assert(magic2 == 0x3ff70050)
+end
+
+simple_double_reinterpret()
+
+local function simple_string_ops()
+  local b = buffer.create(1024)
+
+  buffer.writestring(b, 15, " world")
+  buffer.writestring(b, 10, "hello")
+  buffer.writei8(b, 21, string.byte('!'))
+  assert(buffer.readstring(b, 10, 12) == "hello world!")
+
+  buffer.writestring(b, 10, "hellommm", 5)
+  assert(buffer.readstring(b, 10, 12) == "hello world!")
+
+  buffer.writestring(b, 10, string.rep("hellommm", 1000), 5)
+  assert(buffer.readstring(b, 10, 12) == "hello world!")
+end
+
+simple_string_ops()
+
+local function simple_copy_ops()
+  local b1 = buffer.create(1024)
+  local b2 = buffer.create(1024)
+
+  buffer.writestring(b1, 200, "hello")
+  buffer.writestring(b1, 100, "world")
+
+  buffer.copy(b1, 300, b1, 100, 5)
+
+  buffer.writei8(b2, 35, string.byte(' '))
+  buffer.writei8(b2, 41, string.byte('!'))
+
+  buffer.copy(b2, 30, b1, 200, 5)
+  buffer.copy(b2, 36, b1, 300, 5)
+
+  assert(buffer.readstring(b2, 30, 12) == "hello world!")
+
+  local b3 = buffer.create(9)
+  buffer.writestring(b3, 0, "say hello")
+  buffer.copy(b2, 36, b3, 4)
+  assert(buffer.readstring(b2, 30, 12) == "hello hello!")
+
+  local b4 = buffer.create(5)
+  buffer.writestring(b4, 0, "world")
+  buffer.copy(b2, 36, b4)
+  assert(buffer.readstring(b2, 30, 12) == "hello world!")
+
+  buffer.writestring(b1, 200, "abcdefgh");
+  buffer.copy(b1, 200, b1, 202, 6)
+  assert(buffer.readstring(b1, 200, 8) == "cdefghgh")
+  buffer.copy(b1, 202, b1, 200, 6)
+  assert(buffer.readstring(b1, 200, 8) == "cdcdefgh")
+end
+
+simple_copy_ops()
+
+-- bounds checking
+
+local function createchecks()
+  assert(ecall(function() buffer.create(-1) end) == "size cannot be negative")
+  assert(ecall(function() buffer.create(-1000000) end) == "size cannot be negative")
+end
+
+createchecks()
+
+local function boundchecks()
+  local b = buffer.create(1024)
+
+  assert(call(function() return buffer.readi8(b, 1023) end) == 0)
+  assert(ecall(function() buffer.readi8(b, 1024) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi8(b, -1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi8(b, -100000) end) == "buffer access out of bounds")
+
+  call(function() buffer.writei8(b, 1023, 0) end)
+  assert(ecall(function() buffer.writei8(b, 1024, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei8(b, -1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei8(b, -100000, 0) end) == "buffer access out of bounds")
+
+  -- i16
+  assert(call(function() return buffer.readi16(b, 1022) end) == 0)
+  assert(ecall(function() buffer.readi16(b, 1023) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, -1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, -100000) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, 0x7fffffff) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, 0x7ffffffe) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, 0x7ffffffd) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, 0x80000000) end) == "buffer access out of bounds")
+
+  call(function() buffer.writei16(b, 1022, 0) end)
+  assert(ecall(function() buffer.writei16(b, 1023, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei16(b, -1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei16(b, -100000, 0) end) == "buffer access out of bounds")
+
+  -- i32
+  assert(call(function() return buffer.readi32(b, 1020) end) == 0)
+  assert(ecall(function() buffer.readi32(b, 1021) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi32(b, -1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi32(b, -100000) end) == "buffer access out of bounds")
+
+  call(function() buffer.writei32(b, 1020, 0) end)
+  assert(ecall(function() buffer.writei32(b, 1021, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei32(b, -1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei32(b, -100000, 0) end) == "buffer access out of bounds")
+
+  -- f32
+  assert(call(function() return buffer.readf32(b, 1020) end) == 0)
+  assert(ecall(function() buffer.readf32(b, 1021) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf32(b, -1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf32(b, -100000) end) == "buffer access out of bounds")
+
+  call(function() buffer.writef32(b, 1020, 0) end)
+  assert(ecall(function() buffer.writef32(b, 1021, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef32(b, -1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef32(b, -100000, 0) end) == "buffer access out of bounds")
+
+  -- f64
+  assert(call(function() return buffer.readf64(b, 1016) end) == 0)
+  assert(ecall(function() buffer.readf64(b, 1017) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf64(b, -1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf64(b, -100000) end) == "buffer access out of bounds")
+
+  call(function() buffer.writef64(b, 1016, 0) end)
+  assert(ecall(function() buffer.writef64(b, 1017, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef64(b, -1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef64(b, -100000, 0) end) == "buffer access out of bounds")
+
+  -- string
+  assert(call(function() return buffer.readstring(b, 1016, 8) end) == "\0\0\0\0\0\0\0\0")
+  assert(ecall(function() buffer.readstring(b, 1017, 8) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readstring(b, -1, -8) end) == "size cannot be negative")
+  assert(ecall(function() buffer.readstring(b, -100000, 8) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readstring(b, -100000, 8) end) == "buffer access out of bounds")
+
+  call(function() buffer.writestring(b, 1016, "abcdefgh") end)
+  assert(ecall(function() buffer.writestring(b, 1017, "abcdefgh") end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writestring(b, -1, "abcdefgh") end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writestring(b, -100000, "abcdefgh") end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writestring(b, 100, "abcd", -5) end) == "count cannot be negative")
+  assert(ecall(function() buffer.writestring(b, 100, "abcd", 50) end) == "string length overflow")
+
+  -- copy
+  assert(ecall(function() buffer.copy(b, 30, b, 200, 1000) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.copy(b, 30, b, 200, -5) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.copy(b, 30, b, 2000, 10) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.copy(b, 30, b, -1, 10) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.copy(b, 30, b, -10, 10) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.copy(b, 30, b, -100000, 10) end) == "buffer access out of bounds")
+
+  local b2 = buffer.create(1024)
+  assert(ecall(function() buffer.copy(b, -200, b, 200, 200) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.copy(b, 825, b, 200, 200) end) == "buffer access out of bounds")
+end
+
+boundchecks()
+
+local function boundchecksnonconst(size, minus1, minusbig, intmax)
+  local b = buffer.create(size)
+
+  assert(call(function() return buffer.readi8(b, size-1) end) == 0)
+  assert(ecall(function() buffer.readi8(b, size) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi8(b, minus1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi8(b, minusbig) end) == "buffer access out of bounds")
+
+  call(function() buffer.writei8(b, size-1, 0) end)
+  assert(ecall(function() buffer.writei8(b, size, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei8(b, minus1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei8(b, minusbig, 0) end) == "buffer access out of bounds")
+
+  -- i16
+  assert(call(function() return buffer.readi16(b, size-2) end) == 0)
+  assert(ecall(function() buffer.readi16(b, size-1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, minus1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, minusbig) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, intmax) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, intmax-1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, intmax-2) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, intmax+1) end) == "buffer access out of bounds")
+
+  call(function() buffer.writei16(b, size-2, 0) end)
+  assert(ecall(function() buffer.writei16(b, size-1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei16(b, minus1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei16(b, minusbig, 0) end) == "buffer access out of bounds")
+
+  -- i32
+  assert(call(function() return buffer.readi32(b, size-4) end) == 0)
+  assert(ecall(function() buffer.readi32(b, size-3) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi32(b, minus1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi32(b, minusbig) end) == "buffer access out of bounds")
+
+  call(function() buffer.writei32(b, size-4, 0) end)
+  assert(ecall(function() buffer.writei32(b, size-3, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei32(b, minus1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei32(b, minusbig, 0) end) == "buffer access out of bounds")
+
+  -- f32
+  assert(call(function() return buffer.readf32(b, size-4) end) == 0)
+  assert(ecall(function() buffer.readf32(b, size-3) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf32(b, minus1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf32(b, minusbig) end) == "buffer access out of bounds")
+
+  call(function() buffer.writef32(b, size-4, 0) end)
+  assert(ecall(function() buffer.writef32(b, size-3, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef32(b, minus1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef32(b, minusbig, 0) end) == "buffer access out of bounds")
+
+  -- f64
+  assert(call(function() return buffer.readf64(b, size-8) end) == 0)
+  assert(ecall(function() buffer.readf64(b, size-7) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf64(b, minus1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf64(b, minusbig) end) == "buffer access out of bounds")
+
+  call(function() buffer.writef64(b, size-8, 0) end)
+  assert(ecall(function() buffer.writef64(b, size-7, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef64(b, minus1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef64(b, minusbig, 0) end) == "buffer access out of bounds")
+
+  -- string
+  assert(call(function() return buffer.readstring(b, size-8, 8) end) == "\0\0\0\0\0\0\0\0")
+  assert(ecall(function() buffer.readstring(b, size-7, 8) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readstring(b, minus1, 8) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readstring(b, minusbig, 8) end) == "buffer access out of bounds")
+
+  call(function() buffer.writestring(b, size-8, "abcdefgh") end)
+  assert(ecall(function() buffer.writestring(b, size-7, "abcdefgh") end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writestring(b, minus1, "abcdefgh") end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writestring(b, minusbig, "abcdefgh") end) == "buffer access out of bounds")
+end
+
+boundchecksnonconst(1024, -1, -100000, 0x7fffffff)
+
+local function boundcheckssmall()
+  local b = buffer.create(1)
+
+  assert(call(function() return buffer.readi8(b, 0) end) == 0)
+  assert(ecall(function() buffer.readi8(b, 1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi8(b, -1) end) == "buffer access out of bounds")
+
+  call(function() buffer.writei8(b, 0, 0) end)
+  assert(ecall(function() buffer.writei8(b, 1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei8(b, -1, 0) end) == "buffer access out of bounds")
+
+  -- i16
+  assert(ecall(function() buffer.readi16(b, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, -1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi16(b, -2) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei16(b, 0, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei16(b, -1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei16(b, -2, 0) end) == "buffer access out of bounds")
+
+  -- i32
+  assert(ecall(function() buffer.readi32(b, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi32(b, -1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi32(b, -4) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei32(b, 0, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei32(b, -1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei32(b, -4, 0) end) == "buffer access out of bounds")
+
+  -- f32
+  assert(ecall(function() buffer.readf32(b, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf32(b, -1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf32(b, -4) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef32(b, 0, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef32(b, -1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef32(b, -4, 0) end) == "buffer access out of bounds")
+
+  -- f64
+  assert(ecall(function() buffer.readf64(b, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf64(b, -1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf64(b, -8) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef64(b, 0, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef64(b, -1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writef64(b, -7, 0) end) == "buffer access out of bounds")
+
+  -- string
+  assert(ecall(function() buffer.readstring(b, 0, 8) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readstring(b, -1, 8) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readstring(b, -8, 8) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writestring(b, 0, "abcdefgh") end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writestring(b, -1, "abcdefgh") end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writestring(b, -7, "abcdefgh") end) == "buffer access out of bounds")
+end
+
+boundcheckssmall()
+
+local function boundchecksempty()
+  local b = buffer.create(0) -- useless, but probably more generic
+
+  assert(ecall(function() buffer.readi8(b, 1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi8(b, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi8(b, -1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei8(b, 1, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei8(b, 0, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.writei8(b, -1, 0) end) == "buffer access out of bounds")
+
+  assert(ecall(function() buffer.readi16(b, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readi32(b, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf32(b, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readf64(b, 0) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readstring(b, 0, 1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.readstring(b, 0, 8) end) == "buffer access out of bounds")
+end
+
+boundchecksempty()
+
+local function intuint()
+  local b = buffer.create(32)
+
+  buffer.writeu32(b, 0, 0xffffffff)
+  assert(buffer.readi8(b, 0) == -1)
+  assert(buffer.readu8(b, 0) == 255)
+  assert(buffer.readi16(b, 0) == -1)
+  assert(buffer.readu16(b, 0) == 65535)
+  assert(buffer.readi32(b, 0) == -1)
+  assert(buffer.readu32(b, 0) == 4294967295)
+
+  buffer.writei32(b, 0, -1)
+  assert(buffer.readi8(b, 0) == -1)
+  assert(buffer.readu8(b, 0) == 255)
+  assert(buffer.readi16(b, 0) == -1)
+  assert(buffer.readu16(b, 0) == 65535)
+  assert(buffer.readi32(b, 0) == -1)
+  assert(buffer.readu32(b, 0) == 4294967295)
+
+  buffer.writei16(b, 0, 65535)
+  buffer.writei16(b, 2, -1)
+  assert(buffer.readi8(b, 0) == -1)
+  assert(buffer.readu8(b, 0) == 255)
+  assert(buffer.readi16(b, 0) == -1)
+  assert(buffer.readu16(b, 0) == 65535)
+  assert(buffer.readi32(b, 0) == -1)
+  assert(buffer.readu32(b, 0) == 4294967295)
+
+  buffer.writeu16(b, 0, 65535)
+  buffer.writeu16(b, 2, -1)
+  assert(buffer.readi8(b, 0) == -1)
+  assert(buffer.readu8(b, 0) == 255)
+  assert(buffer.readi16(b, 0) == -1)
+  assert(buffer.readu16(b, 0) == 65535)
+  assert(buffer.readi32(b, 0) == -1)
+  assert(buffer.readu32(b, 0) == 4294967295)
+end
+
+intuint()
+
+local function intuinttricky()
+  local b = buffer.create(32)
+
+  buffer.writeu8(b, 0, 0xffffffff)
+  assert(buffer.readi8(b, 0) == -1)
+  assert(buffer.readu8(b, 0) == 255)
+  assert(buffer.readi16(b, 0) == 255)
+  assert(buffer.readu16(b, 0) == 255)
+  assert(buffer.readi32(b, 0) == 255)
+  assert(buffer.readu32(b, 0) == 255)
+
+  buffer.writeu16(b, 0, 0xffffffff)
+  assert(buffer.readi8(b, 0) == -1)
+  assert(buffer.readu8(b, 0) == 255)
+  assert(buffer.readi16(b, 0) == -1)
+  assert(buffer.readu16(b, 0) == 65535)
+  assert(buffer.readi32(b, 0) == 65535)
+  assert(buffer.readu32(b, 0) == 65535)
+
+  buffer.writei32(b, 8, 0xffffffff)
+  buffer.writeu32(b, 12, 0xffffffff)
+  assert(buffer.readstring(b, 8, 4) == buffer.readstring(b, 12, 4))
+
+  buffer.writei32(b, 8, -2147483648)
+  buffer.writeu32(b, 12, 0x80000000)
+  assert(buffer.readstring(b, 8, 4) == buffer.readstring(b, 12, 4))
+end
+
+intuinttricky()
+
+local function fromtostring()
+  local b = buffer.fromstring("1234567890")
+  assert(buffer.tostring(b) == "1234567890")
+
+  buffer.writestring(b, 4, "xyz")
+  assert(buffer.tostring(b) == "1234xyz890")
+
+  local b2 = buffer.fromstring("abcd\0ef")
+  assert(buffer.tostring(b2) == "abcd\0ef")
+end
+
+fromtostring()
+
+local function fill()
+  local b = buffer.create(10)
+
+  buffer.fill(b, 0, 0x61)
+  assert(buffer.tostring(b) == "aaaaaaaaaa")
+
+  buffer.fill(b, 0, 0x62, 5)
+  assert(buffer.tostring(b) == "bbbbbaaaaa")
+
+  buffer.fill(b, 4, 0x63)
+  assert(buffer.tostring(b) == "bbbbcccccc")
+
+  buffer.fill(b, 6, 0x64, 3)
+  assert(buffer.tostring(b) == "bbbbccdddc")
+
+  buffer.fill(b, 2, 0xffffff65, 8)
+  assert(buffer.tostring(b) == "bbeeeeeeee")
+
+  -- out of bounds
+  assert(ecall(function() buffer.fill(b, -10, 1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.fill(b, 11, 1) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.fill(b, 0, 1, 11) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.fill(b, 5, 1, 6) end) == "buffer access out of bounds")
+  assert(ecall(function() buffer.fill(b, 5, 1, -1) end) == "buffer access out of bounds")
+end
+
+fill()
+
+local function misc()
+  local b = buffer.create(1000)
+
+  assert(select('#', buffer.writei32(b, 10, 40)) == 0)
+  assert(select('#', buffer.writef32(b, 20, 40.0)) == 0)
+end
+
+misc()
+
+local function testslowcalls()
+  getfenv()
+
+  simple_byte_reads()
+  offset_byte_reads(5)
+  offset_byte_reads(30)
+  simple_float_reinterpret()
+  simple_double_reinterpret()
+  simple_string_ops()
+  createchecks()
+  boundchecks()
+  boundchecksnonconst(1024, -1, -100000, 0x7fffffff)
+  boundcheckssmall()
+  boundchecksempty()
+  intuint()
+  intuinttricky()
+  fromtostring()
+  fill()
+  misc()
+end
+
+testslowcalls()
+
+return('OK')

tests/conformance/interrupt.lua

@@ -1,25 +1,70 @@
 -- This file is part of the Luau programming language and is licensed under MIT License; see LICENSE.txt for details
 print("testing interrupts")
 
-function foo()
+-- this function will be called by C code with a special interrupt handler that validates hit locations
-    for i=1,10 do end
+function test()
-    return
+	function foo()
+		for i=1,10 do end
+		return
+	end
+
+	foo()
+
+	function bar()
+		local i = 0
+		while i < 10 do
+			i += i + 1
+		end
+	end
+
+	bar()
+
+	function baz()
+	end
+
+	baz()
 end
 
-foo()
+-- these functions will be called by C code with a special interrupt handler that terminates after a few invocations
-
+function infloop1()
-function bar()
+	while true do end
-    local i = 0
-    while i < 10 do
-        i += i + 1
-    end
 end
 
-bar()
+function infloop2()
-
+	while true do continue end
-function baz()
 end
 
-baz()
+function infloop3()
+	repeat until false
+end
+
+function infloop4()
+	repeat continue until false
+end
+
+function infloop5()
+	for i=0,0,0 do end
+end
+
+function infloop6()
+	for i=0,0,0 do continue end
+end
+
+function infloop7()
+	for i=1,math.huge do end
+end
+
+function infloop8()
+	for i=1,math.huge do continue end
+end
+
+function infloop9()
+	-- technically not a loop, but an exponentially recursive function
+	local function boom()
+		boom()
+		boom()
+	end
+	boom()
+end
 
 return "OK"

tests/conformance/native.lua

@@ -153,6 +153,26 @@ end
 
 assert(pcall(fuzzfail15) == true)
 
+local function fuzzfail16()
+  _ = {[{[2]=77,_=_,[2]=_,}]=not _,}
+  _ = {77,[2]=11008,[2]=_,[0]=_,}
+end
+
+assert(pcall(fuzzfail16) == true)
+
+local function fuzzfail17()
+  return bit32.extract(1293942816,1293942816)
+end
+
+assert(pcall(fuzzfail17) == false)
+
+local function fuzzfail18()
+	return bit32.extract(7890276,0)
+end
+
+assert(pcall(fuzzfail18) == true)
+assert(fuzzfail18() == 0)
+
 local function arraySizeInv1()
   local t = {1, 2, nil, nil, nil, nil, nil, nil, nil, true}
 

tests/conformance/native_types.lua

@@ -52,6 +52,8 @@ local function checkfunction(a: () -> ()) assert(is_native()) end
 local function checkthread(a: thread) assert(is_native()) end
 local function checkuserdata(a: userdata) assert(is_native()) end
 local function checkvector(a: vector) assert(is_native()) end
+local function checkbuffer(a: buffer) assert(is_native()) end
+local function checkoptbuffer(a: buffer?) assert(is_native()) end
 
 call(checktable, {})
 ecall(checktable, 2)
@@ -68,6 +70,12 @@ ecall(checkuserdata, 2)
 call(checkvector, vector(1, 2, 3))
 ecall(checkvector, 2)
 
+call(checkbuffer, buffer.create(10))
+ecall(checkbuffer, 2)
+call(checkoptbuffer, buffer.create(10))
+call(checkoptbuffer, nil)
+ecall(checkoptbuffer, 2)
+
 local function mutation_causes_bad_exit(a: number, count: number, sum: number)
   repeat
     a = 's'

tools/faillist.txt

@@ -1,5 +1,3 @@
-AnnotationTests.cloned_interface_maintains_pointers_between_definitions
-AnnotationTests.table_annotation
 AnnotationTests.two_type_params
 AnnotationTests.typeof_expr
 AnnotationTests.use_generic_type_alias
@@ -11,6 +9,7 @@ AutocompleteTest.anonymous_autofilled_generic_on_argument_type_pack_vararg
 AutocompleteTest.anonymous_autofilled_generic_type_pack_vararg
 AutocompleteTest.autocomplete_interpolated_string_as_singleton
 AutocompleteTest.autocomplete_oop_implicit_self
+AutocompleteTest.autocomplete_response_perf1
 AutocompleteTest.autocomplete_string_singleton_equality
 AutocompleteTest.autocomplete_string_singleton_escape
 AutocompleteTest.autocomplete_string_singletons
@@ -194,7 +193,6 @@ GenericsTests.self_recursive_instantiated_param
 GenericsTests.type_parameters_can_be_polytypes
 GenericsTests.typefuns_sharing_types
 IntersectionTypes.argument_is_intersection
-IntersectionTypes.cli_80596_simplify_degenerate_intersections
 IntersectionTypes.error_detailed_intersection_all
 IntersectionTypes.error_detailed_intersection_part
 IntersectionTypes.fx_intersection_as_argument
@@ -448,10 +446,8 @@ TryUnifyTests.uninhabited_table_sub_never
 TryUnifyTests.variadics_should_use_reversed_properly
 TypeAliases.corecursive_types_generic
 TypeAliases.cyclic_types_of_named_table_fields_do_not_expand_when_stringified
-TypeAliases.dependent_generic_aliases
 TypeAliases.dont_lose_track_of_PendingExpansionTypes_after_substitution
 TypeAliases.free_variables_from_typeof_in_aliases
-TypeAliases.generic_aliases
 TypeAliases.generic_param_remap
 TypeAliases.mismatched_generic_type_param
 TypeAliases.mutually_recursive_aliases
@@ -462,7 +458,6 @@ TypeAliases.mutually_recursive_types_swapsies_not_ok
 TypeAliases.recursive_types_restriction_not_ok
 TypeAliases.report_shadowed_aliases
 TypeAliases.saturate_to_first_type_pack
-TypeAliases.stringify_optional_parameterized_alias
 TypeAliases.type_alias_local_mutation
 TypeAliases.type_alias_local_rename
 TypeAliases.type_alias_locations
@@ -490,6 +485,8 @@ TypeInfer.globals
 TypeInfer.globals2
 TypeInfer.globals_are_banned_in_strict_mode
 TypeInfer.if_statement
+TypeInfer.infer_assignment_value_types
+TypeInfer.infer_assignment_value_types_mutable_lval
 TypeInfer.infer_locals_via_assignment_from_its_call_site
 TypeInfer.infer_locals_with_nil_value
 TypeInfer.infer_through_group_expr
@@ -529,10 +526,12 @@ TypeInferClasses.class_unification_type_mismatch_is_correct_order
 TypeInferClasses.detailed_class_unification_error
 TypeInferClasses.index_instance_property
 TypeInferClasses.indexable_classes
+TypeInferClasses.intersections_of_unions_of_classes
 TypeInferClasses.optional_class_field_access_error
 TypeInferClasses.table_class_unification_reports_sane_errors_for_missing_properties
 TypeInferClasses.table_indexers_are_invariant
 TypeInferClasses.type_mismatch_invariance_required_for_error
+TypeInferClasses.unions_of_intersections_of_classes
 TypeInferClasses.we_can_report_when_someone_is_trying_to_use_a_table_rather_than_a_class
 TypeInferFunctions.another_other_higher_order_function
 TypeInferFunctions.apply_of_lambda_with_inferred_and_explicit_types
@@ -634,7 +633,6 @@ TypeInferModules.bound_free_table_export_is_ok
 TypeInferModules.do_not_modify_imported_types
 TypeInferModules.do_not_modify_imported_types_4
 TypeInferModules.do_not_modify_imported_types_5
-TypeInferModules.general_require_call_expression
 TypeInferModules.module_type_conflict
 TypeInferModules.module_type_conflict_instantiated
 TypeInferModules.require
@@ -681,6 +679,7 @@ TypeInferPrimitives.CheckMethodsOfNumber
 TypeInferPrimitives.string_function_indirect
 TypeInferPrimitives.string_index
 TypeInferUnknownNever.array_like_table_of_never_is_inhabitable
+TypeInferUnknownNever.assign_to_local_which_is_never
 TypeInferUnknownNever.assign_to_prop_which_is_never
 TypeInferUnknownNever.index_on_union_of_tables_for_properties_that_is_never
 TypeInferUnknownNever.index_on_union_of_tables_for_properties_that_is_sorta_never
@@ -696,7 +695,6 @@ TypePackTests.type_packs_with_tails_in_vararg_adjustment
 TypePackTests.unify_variadic_tails_in_arguments
 TypePackTests.unify_variadic_tails_in_arguments_free
 TypePackTests.variadic_argument_tail
-TypePathToStringForError.basic
 TypeSingletons.enums_using_singletons_mismatch
 TypeSingletons.error_detailed_tagged_union_mismatch_bool
 TypeSingletons.error_detailed_tagged_union_mismatch_string
@@ -704,7 +702,6 @@ TypeSingletons.function_args_infer_singletons
 TypeSingletons.function_call_with_singletons
 TypeSingletons.function_call_with_singletons_mismatch
 TypeSingletons.overloaded_function_call_with_singletons_mismatch
-TypeSingletons.parametric_tagged_union_alias
 TypeSingletons.return_type_of_f_is_not_widened
 TypeSingletons.table_properties_singleton_strings
 TypeSingletons.table_properties_type_error_escapes
@@ -724,7 +721,7 @@ UnionTypes.index_on_a_union_type_with_one_property_of_type_any
 UnionTypes.index_on_a_union_type_with_property_guaranteed_to_exist
 UnionTypes.index_on_a_union_type_works_at_arbitrary_depth
 UnionTypes.less_greedy_unification_with_union_types
-UnionTypes.optional_arguments_table2
+UnionTypes.optional_arguments_table
 UnionTypes.optional_assignment_errors
 UnionTypes.optional_call_error
 UnionTypes.optional_field_access_error
@@ -736,6 +733,7 @@ UnionTypes.optional_union_functions
 UnionTypes.optional_union_members
 UnionTypes.optional_union_methods
 UnionTypes.table_union_write_indirect
+UnionTypes.unify_unsealed_table_union_check
 UnionTypes.union_of_functions
 UnionTypes.union_of_functions_mentioning_generic_typepacks
 UnionTypes.union_of_functions_mentioning_generics

tools/fuzzfilter.py

@@ -0,0 +1,47 @@
+#!/usr/bin/python3
+# This file is part of the Luau programming language and is licensed under MIT License; see LICENSE.txt for details
+
+# Given a fuzzer binary and a list of crashing programs, this tool collects unique crash reasons and prints reproducers.
+
+import re
+import sys
+import subprocess
+
+def get_crash_reason(binary, file):
+    res = subprocess.run([binary, file], stdout=subprocess.DEVNULL, stderr=subprocess.PIPE)
+    if res.returncode == 0:
+        print(f"Warning: {binary} {file} returned 0")
+        return None
+    err = res.stderr.decode("utf-8")
+
+    if (pos := err.find("ERROR: libFuzzer:")) != -1:
+        return err[pos:]
+
+    print(f"Warning: {binary} {file} returned unrecognized error {err}")
+    return None
+
+def get_crash_fingerprint(reason):
+    # Due to ASLR addresses are different every time, so we filter them out
+    reason = re.sub(r"0x[0-9a-f]+", "0xXXXX", reason)
+    return reason
+
+binary = sys.argv[1]
+files = sys.argv[2:]
+
+seen = set()
+
+for index, file in enumerate(files):
+    reason = get_crash_reason(binary, file)
+    if reason is None:
+        continue
+    fingerprint = get_crash_fingerprint(reason)
+    if fingerprint in seen:
+        # print a spinning ASCII wheel to indicate that we're making progress
+        print("-\|/"[index % 4] + "\r", end="")
+        continue
+    seen.add(fingerprint)
+    print(f"Reproducer: {binary} {file}")
+    print(f"Crash reason: {reason}")
+    print()
+
+print(f"Total unique crash reasons: {len(seen)}")