luau/tests/IrLowering.test.cpp

1694 lines
41 KiB
C++
Raw Normal View History

// This file is part of the Luau programming language and is licensed under MIT License; see LICENSE.txt for details
#include "lua.h"
#include "lualib.h"
#include "Luau/BytecodeBuilder.h"
#include "Luau/CodeGen.h"
#include "Luau/Compiler.h"
#include "Luau/Parser.h"
#include "Luau/IrBuilder.h"
#include "doctest.h"
#include "ScopedFlags.h"
#include "ConformanceIrHooks.h"
#include <memory>
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
#include <string_view>
LUAU_FASTFLAG(LuauCodegenRemoveDeadStores5)
Sync to upstream/release/622 (#1232) # What's changed? * Improved the actual message for the type errors for `cannot call non-function` when attempting to call a union of functions/callable tables. The error now correctly explains the issue is an inability to determine the return type of the call in this situation. * Resolve an issue where tables and metatables were not correctly being cloned during instantiation (fixes #1176). * Refactor `luaM_getnextgcopage` to `luaM_getnextpage` (generally removing `gco` prefix where appropriate). * Optimize `table.move` between tables for large ranges of elements. * Reformat a bunch of code automatically using `clang-format`. ### New Type Solver * Clean up minimally-used or unused constraints in the constraint solver (`InstantiationConstraint`, `SetOpConstraint`, `SingletonOrTopTypeConstraint`). * Add a builtin `singleton` type family to replace `SingletonOrTopTypeConstraint` when inferring refinements. * Fixed a crash involving type path reasoning by recording when type family reduction has taken place in the path. * Improved constraint ordering by blocking on unreduced types families that are not yet proven uninhabitable. * Improved the handling of `SetIndexerConstraints` for both better inference quality and to resolve crashes. * Fix a crash when normalizing cyclic unions of intersections. * Fix a crash when normalizing an intersection with the negation of `unknown`. * Fix a number of crashes caused by missing `follow` calls on `TypeId`s. * Changed type family reduction to correctly use a semantic notion of uninhabited types, rather than checking for `never` types specifically. * Refactor the `union` and `intersect` type families to be variadic. ### Native Code Generation * Improve translation for userdata key get/set and userdata/vector namecall. * Provide `[top level]` and `[anonymous]` as function names to `FunctionStats` as appropriate when no function name is available. * Disable unwind support on Android platforms since it is unsupported. * --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-20 05:48:02 +08:00
LUAU_FASTFLAG(LuauCodegenDirectUserdataFlow)
Sync to upstream/release/623 (#1236) # What's changed? ### New Type Solver - Unification of two fresh types no longer binds them together. - Replaced uses of raw `emplace` with `emplaceType` to catch cyclic bound types when they are created. - `SetIndexerConstraint` is blocked until the indexer result type is not blocked. - Fix a case where a blocked type got past the constraint solver. - Searching for free types should no longer traverse into `ClassType`s. - Fix a corner case that could result in the non-testable type `~{}`. - Fix incorrect flagging when `any` was a parameter of some checked function in nonstrict type checker. - `IterableConstraint` now consider tables without `__iter` to be iterables. ### Native Code Generation - Improve register type info lookup by program counter. - Generate type information for locals and upvalues --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: James McNellis <jmcnellis@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-26 06:26:09 +08:00
LUAU_FASTFLAG(LuauCompileTypeInfo)
LUAU_FASTFLAG(LuauLoadTypeInfo)
LUAU_FASTFLAG(LuauCodegenTypeInfo)
LUAU_FASTFLAG(LuauCompileTempTypeInfo)
LUAU_FASTFLAG(LuauCodegenAnalyzeHostVectorOps)
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
LUAU_FASTFLAG(LuauCompileUserdataInfo)
LUAU_FASTFLAG(LuauLoadUserdataInfo)
static std::string getCodegenAssembly(const char* source, bool includeIrTypes = false, int debugLevel = 1)
{
Luau::CodeGen::AssemblyOptions options;
options.compilationOptions.hooks.vectorAccessBytecodeType = vectorAccessBytecodeType;
options.compilationOptions.hooks.vectorNamecallBytecodeType = vectorNamecallBytecodeType;
options.compilationOptions.hooks.vectorAccess = vectorAccess;
options.compilationOptions.hooks.vectorNamecall = vectorNamecall;
// For IR, we don't care about assembly, but we want a stable target
options.target = Luau::CodeGen::AssemblyOptions::Target::X64_SystemV;
options.outputBinary = false;
options.includeAssembly = false;
options.includeIr = true;
options.includeOutlinedCode = false;
Sync to upstream/release/623 (#1236) # What's changed? ### New Type Solver - Unification of two fresh types no longer binds them together. - Replaced uses of raw `emplace` with `emplaceType` to catch cyclic bound types when they are created. - `SetIndexerConstraint` is blocked until the indexer result type is not blocked. - Fix a case where a blocked type got past the constraint solver. - Searching for free types should no longer traverse into `ClassType`s. - Fix a corner case that could result in the non-testable type `~{}`. - Fix incorrect flagging when `any` was a parameter of some checked function in nonstrict type checker. - `IterableConstraint` now consider tables without `__iter` to be iterables. ### Native Code Generation - Improve register type info lookup by program counter. - Generate type information for locals and upvalues --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: James McNellis <jmcnellis@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-26 06:26:09 +08:00
options.includeIrTypes = includeIrTypes;
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
options.includeIrPrefix = Luau::CodeGen::IncludeIrPrefix::No;
options.includeUseInfo = Luau::CodeGen::IncludeUseInfo::No;
options.includeCfgInfo = Luau::CodeGen::IncludeCfgInfo::No;
options.includeRegFlowInfo = Luau::CodeGen::IncludeRegFlowInfo::No;
Luau::Allocator allocator;
Luau::AstNameTable names(allocator);
Luau::ParseResult result = Luau::Parser::parse(source, strlen(source), names, allocator);
if (!result.errors.empty())
throw Luau::ParseErrors(result.errors);
Luau::CompileOptions copts = {};
copts.optimizationLevel = 2;
copts.debugLevel = debugLevel;
Sync to upstream/release/623 (#1236) # What's changed? ### New Type Solver - Unification of two fresh types no longer binds them together. - Replaced uses of raw `emplace` with `emplaceType` to catch cyclic bound types when they are created. - `SetIndexerConstraint` is blocked until the indexer result type is not blocked. - Fix a case where a blocked type got past the constraint solver. - Searching for free types should no longer traverse into `ClassType`s. - Fix a corner case that could result in the non-testable type `~{}`. - Fix incorrect flagging when `any` was a parameter of some checked function in nonstrict type checker. - `IterableConstraint` now consider tables without `__iter` to be iterables. ### Native Code Generation - Improve register type info lookup by program counter. - Generate type information for locals and upvalues --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: James McNellis <jmcnellis@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-26 06:26:09 +08:00
copts.typeInfoLevel = 1;
copts.vectorCtor = "vector";
copts.vectorType = "vector";
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
static const char* kUserdataCompileTypes[] = {"vec2", "color", "mat3", nullptr};
copts.userdataTypes = kUserdataCompileTypes;
Luau::BytecodeBuilder bcb;
Luau::compileOrThrow(bcb, result, names, copts);
std::string bytecode = bcb.getBytecode();
std::unique_ptr<lua_State, void (*)(lua_State*)> globalState(luaL_newstate(), lua_close);
lua_State* L = globalState.get();
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
// Runtime mapping is specifically created to NOT match the compilation mapping
options.compilationOptions.userdataTypes = kUserdataRunTypes;
if (Luau::CodeGen::isSupported())
{
// Type remapper requires the codegen runtime
Luau::CodeGen::create(L);
Luau::CodeGen::setUserdataRemapper(L, kUserdataRunTypes, [](void* context, const char* str, size_t len) -> uint8_t {
const char** types = (const char**)context;
uint8_t index = 0;
std::string_view sv{str, len};
for (; *types; ++types)
{
if (sv == *types)
return index;
index++;
}
return 0xff;
});
}
if (luau_load(L, "name", bytecode.data(), bytecode.size(), 0) == 0)
return Luau::CodeGen::getAssembly(L, -1, options, nullptr);
FAIL("Failed to load bytecode");
return "";
}
static std::string getCodegenHeader(const char* source)
{
std::string assembly = getCodegenAssembly(source, /* includeIrTypes */ true, /* debugLevel */ 2);
auto bytecodeStart = assembly.find("bb_bytecode_0:");
if (bytecodeStart == std::string::npos)
bytecodeStart = assembly.find("bb_0:");
REQUIRE(bytecodeStart != std::string::npos);
return assembly.substr(0, bytecodeStart);
}
TEST_SUITE_BEGIN("IrLowering");
TEST_CASE("VectorReciprocal")
{
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function vecrcp(a: vector)
return 1 / a
end
)"),
R"(
; function vecrcp($arg0) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
CodeGen: Extract all vector tag patching into TAG_VECTOR (#1171) Instead of patching the tag component with TVECTOR in every instruction that produces a vector value, we now use a separate IR instruction to do this. This reduces implementation redundancy, but more importantly allows for a class of optimizations: - NUM_TO_VECTOR previously patched the component unconditionally but the result was used only in MUL/DIV_VEC instructions that ignore it anyway; we can now remove this. - ADD_VEC et al can now forward the source of TAG_VECTOR instruction of either input; this shortens the latency chain and in the future could allow us to generate optimal vector instruction sequence once the temporary stores are marked as dead. - In the future on X64, ADD_VEC et al will be able to analyze the input instruction and remove tag masking conditionally. This is not part of this PR as it requires a decision around expected FP environment and/or the necessity of the existing masking to begin with. I've also renamed NUM_TO_VECTOR to NUM_TO_VEC so that "VEC" always refers to "3 float values" and for consistency with ADD/etc. Note: ADD_VEC input forwarding is currently performed unconditionally; it may or may not increase the spills that can't be reloaded from the stack. On A64 this makes the Taylor series computation a tiny bit faster (11.3ns => 11.0ns) as it removes the redundant ins instructions along the NUM_TO_VEC path. Curiously, the optimization of forwarding TAG_VECTOR input to arithmetic instructions actually has a small penalty as without it this PR runs at 10.9 ns. I don't know if this is a property of the benchmark though, as I just noticed that in this benchmark type inference actually fails to infer parts of the computation as a vector op. If desired I will happily omit this part of the change and we can explore that separately.
2024-02-21 23:06:11 +08:00
%6 = NUM_TO_VEC 1
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
%7 = LOAD_TVALUE R0
%8 = DIV_VEC %6, %7
CodeGen: Extract all vector tag patching into TAG_VECTOR (#1171) Instead of patching the tag component with TVECTOR in every instruction that produces a vector value, we now use a separate IR instruction to do this. This reduces implementation redundancy, but more importantly allows for a class of optimizations: - NUM_TO_VECTOR previously patched the component unconditionally but the result was used only in MUL/DIV_VEC instructions that ignore it anyway; we can now remove this. - ADD_VEC et al can now forward the source of TAG_VECTOR instruction of either input; this shortens the latency chain and in the future could allow us to generate optimal vector instruction sequence once the temporary stores are marked as dead. - In the future on X64, ADD_VEC et al will be able to analyze the input instruction and remove tag masking conditionally. This is not part of this PR as it requires a decision around expected FP environment and/or the necessity of the existing masking to begin with. I've also renamed NUM_TO_VECTOR to NUM_TO_VEC so that "VEC" always refers to "3 float values" and for consistency with ADD/etc. Note: ADD_VEC input forwarding is currently performed unconditionally; it may or may not increase the spills that can't be reloaded from the stack. On A64 this makes the Taylor series computation a tiny bit faster (11.3ns => 11.0ns) as it removes the redundant ins instructions along the NUM_TO_VEC path. Curiously, the optimization of forwarding TAG_VECTOR input to arithmetic instructions actually has a small penalty as without it this PR runs at 10.9 ns. I don't know if this is a property of the benchmark though, as I just noticed that in this benchmark type inference actually fails to infer parts of the computation as a vector op. If desired I will happily omit this part of the change and we can explore that separately.
2024-02-21 23:06:11 +08:00
%9 = TAG_VECTOR %8
STORE_TVALUE R1, %9
INTERRUPT 1u
RETURN R1, 1i
)");
}
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
TEST_CASE("VectorComponentRead")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
Sync to upstream/release/616 (#1184) # What's Changed * Add a compiler hint to improve Luau memory allocation inlining ### New Type Solver * Added a system for recommending explicit type annotations to users in cases where we've inferred complex generic types with type families. * Marked string library functions as `@checked` for use in new non-strict mode. * Fixed a bug with new non-strict mode where we would incorrectly report arity mismatches when missing optional arguments. * Implement an occurs check for unifications that would produce self-recursive types. * Fix bug where overload resolution would fail when applied to non-overloaded functions. * Fix bug that caused the subtyping to report an error whenever a generic was instantiated in an invariant context. * Fix crash caused by `SetPropConstraint` not blocking properly. ### Native Code Generation * Implement optimization to eliminate dead stores * Optimize vector ops for X64 when the source is computed (thanks, @zeux!) * Use more efficient lowering for UNM_* (thanks, @zeux!) --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-03-09 08:47:53 +08:00
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function compsum(a: vector)
return a.X + a.Y + a.Z
end
)"),
R"(
; function compsum($arg0) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%6 = LOAD_FLOAT R0, 0i
%11 = LOAD_FLOAT R0, 4i
%20 = ADD_NUM %6, %11
%25 = LOAD_FLOAT R0, 8i
%34 = ADD_NUM %20, %25
STORE_DOUBLE R1, %34
STORE_TAG R1, tnumber
INTERRUPT 8u
RETURN R1, 1i
)");
}
TEST_CASE("VectorAdd")
{
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function vec3add(a: vector, b: vector)
return a + b
end
)"),
R"(
; function vec3add($arg0, $arg1) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
CHECK_TAG R1, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%10 = LOAD_TVALUE R0
%11 = LOAD_TVALUE R1
%12 = ADD_VEC %10, %11
CodeGen: Extract all vector tag patching into TAG_VECTOR (#1171) Instead of patching the tag component with TVECTOR in every instruction that produces a vector value, we now use a separate IR instruction to do this. This reduces implementation redundancy, but more importantly allows for a class of optimizations: - NUM_TO_VECTOR previously patched the component unconditionally but the result was used only in MUL/DIV_VEC instructions that ignore it anyway; we can now remove this. - ADD_VEC et al can now forward the source of TAG_VECTOR instruction of either input; this shortens the latency chain and in the future could allow us to generate optimal vector instruction sequence once the temporary stores are marked as dead. - In the future on X64, ADD_VEC et al will be able to analyze the input instruction and remove tag masking conditionally. This is not part of this PR as it requires a decision around expected FP environment and/or the necessity of the existing masking to begin with. I've also renamed NUM_TO_VECTOR to NUM_TO_VEC so that "VEC" always refers to "3 float values" and for consistency with ADD/etc. Note: ADD_VEC input forwarding is currently performed unconditionally; it may or may not increase the spills that can't be reloaded from the stack. On A64 this makes the Taylor series computation a tiny bit faster (11.3ns => 11.0ns) as it removes the redundant ins instructions along the NUM_TO_VEC path. Curiously, the optimization of forwarding TAG_VECTOR input to arithmetic instructions actually has a small penalty as without it this PR runs at 10.9 ns. I don't know if this is a property of the benchmark though, as I just noticed that in this benchmark type inference actually fails to infer parts of the computation as a vector op. If desired I will happily omit this part of the change and we can explore that separately.
2024-02-21 23:06:11 +08:00
%13 = TAG_VECTOR %12
STORE_TVALUE R2, %13
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
INTERRUPT 1u
RETURN R2, 1i
)");
}
TEST_CASE("VectorMinus")
{
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function vec3minus(a: vector)
return -a
end
)"),
R"(
; function vec3minus($arg0) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%6 = LOAD_TVALUE R0
%7 = UNM_VEC %6
CodeGen: Extract all vector tag patching into TAG_VECTOR (#1171) Instead of patching the tag component with TVECTOR in every instruction that produces a vector value, we now use a separate IR instruction to do this. This reduces implementation redundancy, but more importantly allows for a class of optimizations: - NUM_TO_VECTOR previously patched the component unconditionally but the result was used only in MUL/DIV_VEC instructions that ignore it anyway; we can now remove this. - ADD_VEC et al can now forward the source of TAG_VECTOR instruction of either input; this shortens the latency chain and in the future could allow us to generate optimal vector instruction sequence once the temporary stores are marked as dead. - In the future on X64, ADD_VEC et al will be able to analyze the input instruction and remove tag masking conditionally. This is not part of this PR as it requires a decision around expected FP environment and/or the necessity of the existing masking to begin with. I've also renamed NUM_TO_VECTOR to NUM_TO_VEC so that "VEC" always refers to "3 float values" and for consistency with ADD/etc. Note: ADD_VEC input forwarding is currently performed unconditionally; it may or may not increase the spills that can't be reloaded from the stack. On A64 this makes the Taylor series computation a tiny bit faster (11.3ns => 11.0ns) as it removes the redundant ins instructions along the NUM_TO_VEC path. Curiously, the optimization of forwarding TAG_VECTOR input to arithmetic instructions actually has a small penalty as without it this PR runs at 10.9 ns. I don't know if this is a property of the benchmark though, as I just noticed that in this benchmark type inference actually fails to infer parts of the computation as a vector op. If desired I will happily omit this part of the change and we can explore that separately.
2024-02-21 23:06:11 +08:00
%8 = TAG_VECTOR %7
STORE_TVALUE R1, %8
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
INTERRUPT 1u
RETURN R1, 1i
)");
}
TEST_CASE("VectorSubMulDiv")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function vec3combo(a: vector, b: vector, c: vector, d: vector)
return a * b - c / d
end
)"),
R"(
; function vec3combo($arg0, $arg1, $arg2, $arg3) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
CHECK_TAG R1, tvector, exit(entry)
CHECK_TAG R2, tvector, exit(entry)
CHECK_TAG R3, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%14 = LOAD_TVALUE R0
%15 = LOAD_TVALUE R1
%16 = MUL_VEC %14, %15
CodeGen: Extract all vector tag patching into TAG_VECTOR (#1171) Instead of patching the tag component with TVECTOR in every instruction that produces a vector value, we now use a separate IR instruction to do this. This reduces implementation redundancy, but more importantly allows for a class of optimizations: - NUM_TO_VECTOR previously patched the component unconditionally but the result was used only in MUL/DIV_VEC instructions that ignore it anyway; we can now remove this. - ADD_VEC et al can now forward the source of TAG_VECTOR instruction of either input; this shortens the latency chain and in the future could allow us to generate optimal vector instruction sequence once the temporary stores are marked as dead. - In the future on X64, ADD_VEC et al will be able to analyze the input instruction and remove tag masking conditionally. This is not part of this PR as it requires a decision around expected FP environment and/or the necessity of the existing masking to begin with. I've also renamed NUM_TO_VECTOR to NUM_TO_VEC so that "VEC" always refers to "3 float values" and for consistency with ADD/etc. Note: ADD_VEC input forwarding is currently performed unconditionally; it may or may not increase the spills that can't be reloaded from the stack. On A64 this makes the Taylor series computation a tiny bit faster (11.3ns => 11.0ns) as it removes the redundant ins instructions along the NUM_TO_VEC path. Curiously, the optimization of forwarding TAG_VECTOR input to arithmetic instructions actually has a small penalty as without it this PR runs at 10.9 ns. I don't know if this is a property of the benchmark though, as I just noticed that in this benchmark type inference actually fails to infer parts of the computation as a vector op. If desired I will happily omit this part of the change and we can explore that separately.
2024-02-21 23:06:11 +08:00
%23 = LOAD_TVALUE R2
%24 = LOAD_TVALUE R3
%25 = DIV_VEC %23, %24
%34 = SUB_VEC %16, %25
%35 = TAG_VECTOR %34
STORE_TVALUE R4, %35
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
INTERRUPT 3u
RETURN R4, 1i
)");
}
Sync to upstream/release/615 (#1175) # What's changed? * Luau allocation scheme was changed to handle allocations in 513-1024 byte range internally without falling back to global allocator * coroutine/thread creation no longer requires any global allocations, making it up to 15% faster (vs libc malloc) * table construction for 17-32 keys or 33-64 array elements is up to 30% faster (vs libc malloc) ### New Type Solver * Cyclic unary negation type families are reduced to `number` when possible * Class types are skipped when searching for free types in unifier to improve performance * Fixed issues with table type inference when metatables are present * Improved inference of iteration loop types * Fixed an issue with bidirectional inference of method calls * Type simplification will now preserve error suppression markers ### Native Code Generation * Fixed TAG_VECTOR skip optimization to not break instruction use counts (broken optimization wasn't included in 614) * Fixed missing side-effect when optimizing generic loop preparation instruction --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com>
2024-03-02 02:45:26 +08:00
TEST_CASE("VectorSubMulDiv2")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
Sync to upstream/release/615 (#1175) # What's changed? * Luau allocation scheme was changed to handle allocations in 513-1024 byte range internally without falling back to global allocator * coroutine/thread creation no longer requires any global allocations, making it up to 15% faster (vs libc malloc) * table construction for 17-32 keys or 33-64 array elements is up to 30% faster (vs libc malloc) ### New Type Solver * Cyclic unary negation type families are reduced to `number` when possible * Class types are skipped when searching for free types in unifier to improve performance * Fixed issues with table type inference when metatables are present * Improved inference of iteration loop types * Fixed an issue with bidirectional inference of method calls * Type simplification will now preserve error suppression markers ### Native Code Generation * Fixed TAG_VECTOR skip optimization to not break instruction use counts (broken optimization wasn't included in 614) * Fixed missing side-effect when optimizing generic loop preparation instruction --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com>
2024-03-02 02:45:26 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function vec3combo(a: vector)
local tmp = a * a
return (tmp - tmp) / (tmp + tmp)
end
)"),
R"(
; function vec3combo($arg0) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%8 = LOAD_TVALUE R0
%10 = MUL_VEC %8, %8
%19 = SUB_VEC %10, %10
%28 = ADD_VEC %10, %10
%37 = DIV_VEC %19, %28
%38 = TAG_VECTOR %37
STORE_TVALUE R2, %38
INTERRUPT 4u
RETURN R2, 1i
)");
}
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
TEST_CASE("VectorMulDivMixed")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function vec3combo(a: vector, b: vector, c: vector, d: vector)
return a * 2 + b / 4 + 0.5 * c + 40 / d
end
)"),
R"(
; function vec3combo($arg0, $arg1, $arg2, $arg3) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
CHECK_TAG R1, tvector, exit(entry)
CHECK_TAG R2, tvector, exit(entry)
CHECK_TAG R3, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%12 = LOAD_TVALUE R0
CodeGen: Extract all vector tag patching into TAG_VECTOR (#1171) Instead of patching the tag component with TVECTOR in every instruction that produces a vector value, we now use a separate IR instruction to do this. This reduces implementation redundancy, but more importantly allows for a class of optimizations: - NUM_TO_VECTOR previously patched the component unconditionally but the result was used only in MUL/DIV_VEC instructions that ignore it anyway; we can now remove this. - ADD_VEC et al can now forward the source of TAG_VECTOR instruction of either input; this shortens the latency chain and in the future could allow us to generate optimal vector instruction sequence once the temporary stores are marked as dead. - In the future on X64, ADD_VEC et al will be able to analyze the input instruction and remove tag masking conditionally. This is not part of this PR as it requires a decision around expected FP environment and/or the necessity of the existing masking to begin with. I've also renamed NUM_TO_VECTOR to NUM_TO_VEC so that "VEC" always refers to "3 float values" and for consistency with ADD/etc. Note: ADD_VEC input forwarding is currently performed unconditionally; it may or may not increase the spills that can't be reloaded from the stack. On A64 this makes the Taylor series computation a tiny bit faster (11.3ns => 11.0ns) as it removes the redundant ins instructions along the NUM_TO_VEC path. Curiously, the optimization of forwarding TAG_VECTOR input to arithmetic instructions actually has a small penalty as without it this PR runs at 10.9 ns. I don't know if this is a property of the benchmark though, as I just noticed that in this benchmark type inference actually fails to infer parts of the computation as a vector op. If desired I will happily omit this part of the change and we can explore that separately.
2024-02-21 23:06:11 +08:00
%13 = NUM_TO_VEC 2
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
%14 = MUL_VEC %12, %13
CodeGen: Extract all vector tag patching into TAG_VECTOR (#1171) Instead of patching the tag component with TVECTOR in every instruction that produces a vector value, we now use a separate IR instruction to do this. This reduces implementation redundancy, but more importantly allows for a class of optimizations: - NUM_TO_VECTOR previously patched the component unconditionally but the result was used only in MUL/DIV_VEC instructions that ignore it anyway; we can now remove this. - ADD_VEC et al can now forward the source of TAG_VECTOR instruction of either input; this shortens the latency chain and in the future could allow us to generate optimal vector instruction sequence once the temporary stores are marked as dead. - In the future on X64, ADD_VEC et al will be able to analyze the input instruction and remove tag masking conditionally. This is not part of this PR as it requires a decision around expected FP environment and/or the necessity of the existing masking to begin with. I've also renamed NUM_TO_VECTOR to NUM_TO_VEC so that "VEC" always refers to "3 float values" and for consistency with ADD/etc. Note: ADD_VEC input forwarding is currently performed unconditionally; it may or may not increase the spills that can't be reloaded from the stack. On A64 this makes the Taylor series computation a tiny bit faster (11.3ns => 11.0ns) as it removes the redundant ins instructions along the NUM_TO_VEC path. Curiously, the optimization of forwarding TAG_VECTOR input to arithmetic instructions actually has a small penalty as without it this PR runs at 10.9 ns. I don't know if this is a property of the benchmark though, as I just noticed that in this benchmark type inference actually fails to infer parts of the computation as a vector op. If desired I will happily omit this part of the change and we can explore that separately.
2024-02-21 23:06:11 +08:00
%19 = LOAD_TVALUE R1
%20 = NUM_TO_VEC 4
%21 = DIV_VEC %19, %20
%30 = ADD_VEC %14, %21
%40 = NUM_TO_VEC 0.5
%41 = LOAD_TVALUE R2
%42 = MUL_VEC %40, %41
%51 = ADD_VEC %30, %42
%56 = NUM_TO_VEC 40
%57 = LOAD_TVALUE R3
%58 = DIV_VEC %56, %57
%67 = ADD_VEC %51, %58
%68 = TAG_VECTOR %67
STORE_TVALUE R4, %68
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
INTERRUPT 8u
RETURN R4, 1i
)");
}
TEST_CASE("ExtraMathMemoryOperands")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
Sync to upstream/release/616 (#1184) # What's Changed * Add a compiler hint to improve Luau memory allocation inlining ### New Type Solver * Added a system for recommending explicit type annotations to users in cases where we've inferred complex generic types with type families. * Marked string library functions as `@checked` for use in new non-strict mode. * Fixed a bug with new non-strict mode where we would incorrectly report arity mismatches when missing optional arguments. * Implement an occurs check for unifications that would produce self-recursive types. * Fix bug where overload resolution would fail when applied to non-overloaded functions. * Fix bug that caused the subtyping to report an error whenever a generic was instantiated in an invariant context. * Fix crash caused by `SetPropConstraint` not blocking properly. ### Native Code Generation * Implement optimization to eliminate dead stores * Optimize vector ops for X64 when the source is computed (thanks, @zeux!) * Use more efficient lowering for UNM_* (thanks, @zeux!) --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-03-09 08:47:53 +08:00
Sync to upstream/release/610 (#1154) # What's changed? * Check interrupt handler inside the pattern match engine to eliminate potential for programs to hang during string library function execution. * Allow iteration over table properties to pass the old type solver. ### Native Code Generation * Use in-place memory operands for math library operations on x64. * Replace opaque bools with separate enum classes in IrDump to improve code maintainability. * Translate operations on inferred vectors to IR. * Enable support for debugging native-compiled functions in Roblox Studio. ### New Type Solver * Rework type inference for boolean and string literals to introduce bounded free types (bounded below by the singleton type, and above by the primitive type) and reworked primitive type constraint to decide which is the appropriate type for the literal. * Introduce `FunctionCheckConstraint` to handle bidirectional typechecking for function calls, pushing the expected parameter types from the function onto the arguments. * Introduce `union` and `intersect` type families to compute deferred simplified unions and intersections to be employed by the constraint generation logic in the new solver. * Implement support for expanding the domain of local types in `Unifier2`. * Rework type inference for iteration variables bound by for in loops to use local types. * Change constraint blocking logic to use a set to prevent accidental re-blocking. * Add logic to detect missing return statements in functions. ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-01-27 11:20:56 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function foo(a: number, b: number, c: number, d: number, e: number)
return math.floor(a) + math.ceil(b) + math.round(c) + math.sqrt(d) + math.abs(e)
end
)"),
R"(
; function foo($arg0, $arg1, $arg2, $arg3, $arg4) line 2
bb_0:
CHECK_TAG R0, tnumber, exit(entry)
CHECK_TAG R1, tnumber, exit(entry)
CHECK_TAG R2, tnumber, exit(entry)
CHECK_TAG R3, tnumber, exit(entry)
CHECK_TAG R4, tnumber, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
CHECK_SAFE_ENV exit(1)
%16 = FLOOR_NUM R0
%23 = CEIL_NUM R1
%32 = ADD_NUM %16, %23
%39 = ROUND_NUM R2
%48 = ADD_NUM %32, %39
%55 = SQRT_NUM R3
%64 = ADD_NUM %48, %55
%71 = ABS_NUM R4
%80 = ADD_NUM %64, %71
STORE_DOUBLE R5, %80
STORE_TAG R5, tnumber
INTERRUPT 29u
RETURN R5, 1i
)");
}
Sync to upstream/release/617 (#1204) # What's Changed * Fix a case where the stack wasn't completely cleaned up where `debug.info` errored when passed `"f"` option and a thread. * Fix a case of uninitialized field in `luaF_newproto`. ### New Type Solver * When a local is captured in a function, don't add a new entry to the `DfgScope::bindings` if the capture occurs within a loop. * Fix a poor performance characteristic during unification by not trying to simplify an intersection. * Fix a case of multiple constraints mutating the same blocked type causing incorrect inferences. * Fix a case of assertion failure when overload resolution encounters a return typepack mismatch. * When refining a property of the top `table` type, we no longer signal an unknown property error. * Fix a misuse of free types when trying to infer the type of a subscript expression. * Fix a case of assertion failure when trying to resolve an overload from `never`. ### Native Code Generation * Fix dead store optimization issues caused by partial stores. --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-03-16 07:37:39 +08:00
TEST_CASE("DseInitialStackState")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
Sync to upstream/release/617 (#1204) # What's Changed * Fix a case where the stack wasn't completely cleaned up where `debug.info` errored when passed `"f"` option and a thread. * Fix a case of uninitialized field in `luaF_newproto`. ### New Type Solver * When a local is captured in a function, don't add a new entry to the `DfgScope::bindings` if the capture occurs within a loop. * Fix a poor performance characteristic during unification by not trying to simplify an intersection. * Fix a case of multiple constraints mutating the same blocked type causing incorrect inferences. * Fix a case of assertion failure when overload resolution encounters a return typepack mismatch. * When refining a property of the top `table` type, we no longer signal an unknown property error. * Fix a misuse of free types when trying to infer the type of a subscript expression. * Fix a case of assertion failure when trying to resolve an overload from `never`. ### Native Code Generation * Fix dead store optimization issues caused by partial stores. --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-03-16 07:37:39 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function foo()
while {} do
local _ = not _,{}
_ = nil
end
end
)"),
R"(
; function foo() line 2
bb_bytecode_0:
SET_SAVEDPC 1u
%1 = NEW_TABLE 0u, 0u
STORE_POINTER R0, %1
STORE_TAG R0, ttable
CHECK_GC
JUMP bb_2
bb_2:
CHECK_SAFE_ENV exit(3)
JUMP_EQ_TAG K1, tnil, bb_fallback_4, bb_3
bb_3:
%9 = LOAD_TVALUE K1
STORE_TVALUE R1, %9
JUMP bb_5
bb_5:
SET_SAVEDPC 7u
%21 = NEW_TABLE 0u, 0u
STORE_POINTER R1, %21
STORE_TAG R1, ttable
CHECK_GC
STORE_TAG R0, tnil
INTERRUPT 9u
JUMP bb_bytecode_0
)");
}
TEST_CASE("DseInitialStackState2")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
Sync to upstream/release/617 (#1204) # What's Changed * Fix a case where the stack wasn't completely cleaned up where `debug.info` errored when passed `"f"` option and a thread. * Fix a case of uninitialized field in `luaF_newproto`. ### New Type Solver * When a local is captured in a function, don't add a new entry to the `DfgScope::bindings` if the capture occurs within a loop. * Fix a poor performance characteristic during unification by not trying to simplify an intersection. * Fix a case of multiple constraints mutating the same blocked type causing incorrect inferences. * Fix a case of assertion failure when overload resolution encounters a return typepack mismatch. * When refining a property of the top `table` type, we no longer signal an unknown property error. * Fix a misuse of free types when trying to infer the type of a subscript expression. * Fix a case of assertion failure when trying to resolve an overload from `never`. ### Native Code Generation * Fix dead store optimization issues caused by partial stores. --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-03-16 07:37:39 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function foo(a)
math.frexp(a)
return a
end
)"),
R"(
; function foo($arg0) line 2
bb_bytecode_0:
CHECK_SAFE_ENV exit(1)
CHECK_TAG R0, tnumber, exit(1)
FASTCALL 14u, R1, R0, undef, 1i, 2i
INTERRUPT 5u
RETURN R0, 1i
)");
}
TEST_CASE("DseInitialStackState3")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
Sync to upstream/release/617 (#1204) # What's Changed * Fix a case where the stack wasn't completely cleaned up where `debug.info` errored when passed `"f"` option and a thread. * Fix a case of uninitialized field in `luaF_newproto`. ### New Type Solver * When a local is captured in a function, don't add a new entry to the `DfgScope::bindings` if the capture occurs within a loop. * Fix a poor performance characteristic during unification by not trying to simplify an intersection. * Fix a case of multiple constraints mutating the same blocked type causing incorrect inferences. * Fix a case of assertion failure when overload resolution encounters a return typepack mismatch. * When refining a property of the top `table` type, we no longer signal an unknown property error. * Fix a misuse of free types when trying to infer the type of a subscript expression. * Fix a case of assertion failure when trying to resolve an overload from `never`. ### Native Code Generation * Fix dead store optimization issues caused by partial stores. --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-03-16 07:37:39 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function foo(a)
math.sign(a)
return a
end
)"),
R"(
; function foo($arg0) line 2
bb_bytecode_0:
CHECK_SAFE_ENV exit(1)
CHECK_TAG R0, tnumber, exit(1)
FASTCALL 47u, R1, R0, undef, 1i, 1i
INTERRUPT 5u
RETURN R0, 1i
)");
}
TEST_CASE("VectorConstantTag")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function vecrcp(a: vector)
return vector(1, 2, 3) + a
end
)"),
Sync to upstream/release/622 (#1232) # What's changed? * Improved the actual message for the type errors for `cannot call non-function` when attempting to call a union of functions/callable tables. The error now correctly explains the issue is an inability to determine the return type of the call in this situation. * Resolve an issue where tables and metatables were not correctly being cloned during instantiation (fixes #1176). * Refactor `luaM_getnextgcopage` to `luaM_getnextpage` (generally removing `gco` prefix where appropriate). * Optimize `table.move` between tables for large ranges of elements. * Reformat a bunch of code automatically using `clang-format`. ### New Type Solver * Clean up minimally-used or unused constraints in the constraint solver (`InstantiationConstraint`, `SetOpConstraint`, `SingletonOrTopTypeConstraint`). * Add a builtin `singleton` type family to replace `SingletonOrTopTypeConstraint` when inferring refinements. * Fixed a crash involving type path reasoning by recording when type family reduction has taken place in the path. * Improved constraint ordering by blocking on unreduced types families that are not yet proven uninhabitable. * Improved the handling of `SetIndexerConstraints` for both better inference quality and to resolve crashes. * Fix a crash when normalizing cyclic unions of intersections. * Fix a crash when normalizing an intersection with the negation of `unknown`. * Fix a number of crashes caused by missing `follow` calls on `TypeId`s. * Changed type family reduction to correctly use a semantic notion of uninhabited types, rather than checking for `never` types specifically. * Refactor the `union` and `intersect` type families to be variadic. ### Native Code Generation * Improve translation for userdata key get/set and userdata/vector namecall. * Provide `[top level]` and `[anonymous]` as function names to `FunctionStats` as appropriate when no function name is available. * Disable unwind support on Android platforms since it is unsupported. * --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-20 05:48:02 +08:00
R"(
; function vecrcp($arg0) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%4 = LOAD_TVALUE K0, 0i, tvector
%11 = LOAD_TVALUE R0
%12 = ADD_VEC %4, %11
%13 = TAG_VECTOR %12
STORE_TVALUE R1, %13
INTERRUPT 2u
RETURN R1, 1i
)");
}
Sync to upstream/release/622 (#1232) # What's changed? * Improved the actual message for the type errors for `cannot call non-function` when attempting to call a union of functions/callable tables. The error now correctly explains the issue is an inability to determine the return type of the call in this situation. * Resolve an issue where tables and metatables were not correctly being cloned during instantiation (fixes #1176). * Refactor `luaM_getnextgcopage` to `luaM_getnextpage` (generally removing `gco` prefix where appropriate). * Optimize `table.move` between tables for large ranges of elements. * Reformat a bunch of code automatically using `clang-format`. ### New Type Solver * Clean up minimally-used or unused constraints in the constraint solver (`InstantiationConstraint`, `SetOpConstraint`, `SingletonOrTopTypeConstraint`). * Add a builtin `singleton` type family to replace `SingletonOrTopTypeConstraint` when inferring refinements. * Fixed a crash involving type path reasoning by recording when type family reduction has taken place in the path. * Improved constraint ordering by blocking on unreduced types families that are not yet proven uninhabitable. * Improved the handling of `SetIndexerConstraints` for both better inference quality and to resolve crashes. * Fix a crash when normalizing cyclic unions of intersections. * Fix a crash when normalizing an intersection with the negation of `unknown`. * Fix a number of crashes caused by missing `follow` calls on `TypeId`s. * Changed type family reduction to correctly use a semantic notion of uninhabited types, rather than checking for `never` types specifically. * Refactor the `union` and `intersect` type families to be variadic. ### Native Code Generation * Improve translation for userdata key get/set and userdata/vector namecall. * Provide `[top level]` and `[anonymous]` as function names to `FunctionStats` as appropriate when no function name is available. * Disable unwind support on Android platforms since it is unsupported. * --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-20 05:48:02 +08:00
TEST_CASE("VectorNamecall")
{
ScopedFastFlag luauCodegenDirectUserdataFlow{FFlag::LuauCodegenDirectUserdataFlow, true};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function abs(a: vector)
return a:Abs()
end
)"),
R"(
; function abs($arg0) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
FALLBACK_NAMECALL 0u, R1, R0, K0
INTERRUPT 2u
SET_SAVEDPC 3u
CALL R1, 1i, -1i
INTERRUPT 3u
RETURN R1, -1i
)");
}
TEST_CASE("VectorRandomProp")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function foo(a: vector)
return a.XX + a.YY + a.ZZ
end
)"),
R"(
; function foo($arg0) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
FALLBACK_GETTABLEKS 0u, R3, R0, K0
FALLBACK_GETTABLEKS 2u, R4, R0, K1
CHECK_TAG R3, tnumber, bb_fallback_3
CHECK_TAG R4, tnumber, bb_fallback_3
%14 = LOAD_DOUBLE R3
%16 = ADD_NUM %14, R4
STORE_DOUBLE R2, %16
STORE_TAG R2, tnumber
JUMP bb_4
bb_4:
CHECK_TAG R0, tvector, exit(5)
FALLBACK_GETTABLEKS 5u, R3, R0, K2
CHECK_TAG R2, tnumber, bb_fallback_5
CHECK_TAG R3, tnumber, bb_fallback_5
%30 = LOAD_DOUBLE R2
%32 = ADD_NUM %30, R3
STORE_DOUBLE R1, %32
STORE_TAG R1, tnumber
JUMP bb_6
bb_6:
INTERRUPT 8u
RETURN R1, 1i
)");
}
TEST_CASE("VectorCustomAccess")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
ScopedFastFlag luauCodegenAnalyzeHostVectorOps{FFlag::LuauCodegenAnalyzeHostVectorOps, true};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function vec3magn(a: vector)
return a.Magnitude * 2
end
)"),
R"(
; function vec3magn($arg0) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%6 = LOAD_FLOAT R0, 0i
%7 = LOAD_FLOAT R0, 4i
%8 = LOAD_FLOAT R0, 8i
%9 = MUL_NUM %6, %6
%10 = MUL_NUM %7, %7
%11 = MUL_NUM %8, %8
%12 = ADD_NUM %9, %10
%13 = ADD_NUM %12, %11
%14 = SQRT_NUM %13
%20 = MUL_NUM %14, 2
STORE_DOUBLE R1, %20
STORE_TAG R1, tnumber
INTERRUPT 3u
RETURN R1, 1i
)");
}
TEST_CASE("VectorCustomNamecall")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
ScopedFastFlag LuauCodegenDirectUserdataFlow{FFlag::LuauCodegenDirectUserdataFlow, true};
ScopedFastFlag luauCodegenAnalyzeHostVectorOps{FFlag::LuauCodegenAnalyzeHostVectorOps, true};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function vec3dot(a: vector, b: vector)
return (a:Dot(b))
end
)"),
R"(
; function vec3dot($arg0, $arg1) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
CHECK_TAG R1, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%6 = LOAD_TVALUE R1
STORE_TVALUE R4, %6
%12 = LOAD_FLOAT R0, 0i
%13 = LOAD_FLOAT R4, 0i
%14 = MUL_NUM %12, %13
%15 = LOAD_FLOAT R0, 4i
%16 = LOAD_FLOAT R4, 4i
%17 = MUL_NUM %15, %16
%18 = LOAD_FLOAT R0, 8i
%19 = LOAD_FLOAT R4, 8i
%20 = MUL_NUM %18, %19
%21 = ADD_NUM %14, %17
%22 = ADD_NUM %21, %20
STORE_DOUBLE R2, %22
STORE_TAG R2, tnumber
INTERRUPT 4u
RETURN R2, 1i
)");
}
TEST_CASE("VectorCustomAccessChain")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
ScopedFastFlag LuauCodegenDirectUserdataFlow{FFlag::LuauCodegenDirectUserdataFlow, true};
ScopedFastFlag luauCodegenAnalyzeHostVectorOps{FFlag::LuauCodegenAnalyzeHostVectorOps, true};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function foo(a: vector, b: vector)
return a.Unit * b.Magnitude
end
)"),
R"(
; function foo($arg0, $arg1) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
CHECK_TAG R1, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%8 = LOAD_FLOAT R0, 0i
%9 = LOAD_FLOAT R0, 4i
%10 = LOAD_FLOAT R0, 8i
%11 = MUL_NUM %8, %8
%12 = MUL_NUM %9, %9
%13 = MUL_NUM %10, %10
%14 = ADD_NUM %11, %12
%15 = ADD_NUM %14, %13
%16 = SQRT_NUM %15
%17 = DIV_NUM 1, %16
%18 = MUL_NUM %8, %17
%19 = MUL_NUM %9, %17
%20 = MUL_NUM %10, %17
STORE_VECTOR R3, %18, %19, %20
STORE_TAG R3, tvector
%25 = LOAD_FLOAT R1, 0i
%26 = LOAD_FLOAT R1, 4i
%27 = LOAD_FLOAT R1, 8i
%28 = MUL_NUM %25, %25
%29 = MUL_NUM %26, %26
%30 = MUL_NUM %27, %27
%31 = ADD_NUM %28, %29
%32 = ADD_NUM %31, %30
%33 = SQRT_NUM %32
%40 = LOAD_TVALUE R3
%42 = NUM_TO_VEC %33
%43 = MUL_VEC %40, %42
%44 = TAG_VECTOR %43
STORE_TVALUE R2, %44
INTERRUPT 5u
RETURN R2, 1i
)");
}
TEST_CASE("VectorCustomNamecallChain")
{
ScopedFastFlag luauCodegenRemoveDeadStores{FFlag::LuauCodegenRemoveDeadStores5, true};
ScopedFastFlag LuauCodegenDirectUserdataFlow{FFlag::LuauCodegenDirectUserdataFlow, true};
ScopedFastFlag luauCodegenAnalyzeHostVectorOps{FFlag::LuauCodegenAnalyzeHostVectorOps, true};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function foo(n: vector, b: vector, t: vector)
return n:Cross(t):Dot(b) + 1
end
)"),
R"(
; function foo($arg0, $arg1, $arg2) line 2
bb_0:
CHECK_TAG R0, tvector, exit(entry)
CHECK_TAG R1, tvector, exit(entry)
CHECK_TAG R2, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%8 = LOAD_TVALUE R2
STORE_TVALUE R6, %8
%14 = LOAD_FLOAT R0, 0i
%15 = LOAD_FLOAT R6, 0i
%16 = LOAD_FLOAT R0, 4i
%17 = LOAD_FLOAT R6, 4i
%18 = LOAD_FLOAT R0, 8i
%19 = LOAD_FLOAT R6, 8i
%20 = MUL_NUM %16, %19
%21 = MUL_NUM %18, %17
%22 = SUB_NUM %20, %21
%23 = MUL_NUM %18, %15
%24 = MUL_NUM %14, %19
%25 = SUB_NUM %23, %24
%26 = MUL_NUM %14, %17
%27 = MUL_NUM %16, %15
%28 = SUB_NUM %26, %27
STORE_VECTOR R4, %22, %25, %28
STORE_TAG R4, tvector
%31 = LOAD_TVALUE R1
STORE_TVALUE R6, %31
%37 = LOAD_FLOAT R4, 0i
%38 = LOAD_FLOAT R6, 0i
%39 = MUL_NUM %37, %38
%40 = LOAD_FLOAT R4, 4i
%41 = LOAD_FLOAT R6, 4i
%42 = MUL_NUM %40, %41
%43 = LOAD_FLOAT R4, 8i
%44 = LOAD_FLOAT R6, 8i
%45 = MUL_NUM %43, %44
%46 = ADD_NUM %39, %42
%47 = ADD_NUM %46, %45
%53 = ADD_NUM %47, 1
STORE_DOUBLE R3, %53
STORE_TAG R3, tnumber
INTERRUPT 9u
RETURN R3, 1i
)");
}
TEST_CASE("VectorCustomNamecallChain2")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}, {FFlag::LuauCodegenDirectUserdataFlow, true},
{FFlag::LuauCodegenAnalyzeHostVectorOps, true}};
CHECK_EQ("\n" + getCodegenAssembly(R"(
type Vertex = {n: vector, b: vector}
local function foo(v: Vertex, t: vector)
return v.n:Cross(t):Dot(v.b) + 1
end
)"),
R"(
; function foo($arg0, $arg1) line 4
bb_0:
CHECK_TAG R0, ttable, exit(entry)
CHECK_TAG R1, tvector, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%8 = LOAD_POINTER R0
%9 = GET_SLOT_NODE_ADDR %8, 0u, K1
CHECK_SLOT_MATCH %9, K1, bb_fallback_3
%11 = LOAD_TVALUE %9, 0i
STORE_TVALUE R3, %11
JUMP bb_4
bb_4:
%16 = LOAD_TVALUE R1
STORE_TVALUE R5, %16
CHECK_TAG R3, tvector, exit(3)
CHECK_TAG R5, tvector, exit(3)
%22 = LOAD_FLOAT R3, 0i
%23 = LOAD_FLOAT R5, 0i
%24 = LOAD_FLOAT R3, 4i
%25 = LOAD_FLOAT R5, 4i
%26 = LOAD_FLOAT R3, 8i
%27 = LOAD_FLOAT R5, 8i
%28 = MUL_NUM %24, %27
%29 = MUL_NUM %26, %25
%30 = SUB_NUM %28, %29
%31 = MUL_NUM %26, %23
%32 = MUL_NUM %22, %27
%33 = SUB_NUM %31, %32
%34 = MUL_NUM %22, %25
%35 = MUL_NUM %24, %23
%36 = SUB_NUM %34, %35
STORE_VECTOR R3, %30, %33, %36
CHECK_TAG R0, ttable, exit(6)
%41 = LOAD_POINTER R0
%42 = GET_SLOT_NODE_ADDR %41, 6u, K3
CHECK_SLOT_MATCH %42, K3, bb_fallback_5
%44 = LOAD_TVALUE %42, 0i
STORE_TVALUE R5, %44
JUMP bb_6
bb_6:
CHECK_TAG R3, tvector, exit(8)
CHECK_TAG R5, tvector, exit(8)
%53 = LOAD_FLOAT R3, 0i
%54 = LOAD_FLOAT R5, 0i
%55 = MUL_NUM %53, %54
%56 = LOAD_FLOAT R3, 4i
%57 = LOAD_FLOAT R5, 4i
%58 = MUL_NUM %56, %57
%59 = LOAD_FLOAT R3, 8i
%60 = LOAD_FLOAT R5, 8i
%61 = MUL_NUM %59, %60
%62 = ADD_NUM %55, %58
%63 = ADD_NUM %62, %61
%69 = ADD_NUM %63, 1
STORE_DOUBLE R2, %69
STORE_TAG R2, tnumber
INTERRUPT 12u
RETURN R2, 1i
)");
}
Sync to upstream/release/622 (#1232) # What's changed? * Improved the actual message for the type errors for `cannot call non-function` when attempting to call a union of functions/callable tables. The error now correctly explains the issue is an inability to determine the return type of the call in this situation. * Resolve an issue where tables and metatables were not correctly being cloned during instantiation (fixes #1176). * Refactor `luaM_getnextgcopage` to `luaM_getnextpage` (generally removing `gco` prefix where appropriate). * Optimize `table.move` between tables for large ranges of elements. * Reformat a bunch of code automatically using `clang-format`. ### New Type Solver * Clean up minimally-used or unused constraints in the constraint solver (`InstantiationConstraint`, `SetOpConstraint`, `SingletonOrTopTypeConstraint`). * Add a builtin `singleton` type family to replace `SingletonOrTopTypeConstraint` when inferring refinements. * Fixed a crash involving type path reasoning by recording when type family reduction has taken place in the path. * Improved constraint ordering by blocking on unreduced types families that are not yet proven uninhabitable. * Improved the handling of `SetIndexerConstraints` for both better inference quality and to resolve crashes. * Fix a crash when normalizing cyclic unions of intersections. * Fix a crash when normalizing an intersection with the negation of `unknown`. * Fix a number of crashes caused by missing `follow` calls on `TypeId`s. * Changed type family reduction to correctly use a semantic notion of uninhabited types, rather than checking for `never` types specifically. * Refactor the `union` and `intersect` type families to be variadic. ### Native Code Generation * Improve translation for userdata key get/set and userdata/vector namecall. * Provide `[top level]` and `[anonymous]` as function names to `FunctionStats` as appropriate when no function name is available. * Disable unwind support on Android platforms since it is unsupported. * --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-20 05:48:02 +08:00
TEST_CASE("UserDataGetIndex")
{
ScopedFastFlag luauCodegenDirectUserdataFlow{FFlag::LuauCodegenDirectUserdataFlow, true};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function getxy(a: Point)
return a.x + a.y
end
)"),
R"(
; function getxy($arg0) line 2
bb_0:
CHECK_TAG R0, tuserdata, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
FALLBACK_GETTABLEKS 0u, R2, R0, K0
FALLBACK_GETTABLEKS 2u, R3, R0, K1
CHECK_TAG R2, tnumber, bb_fallback_3
CHECK_TAG R3, tnumber, bb_fallback_3
%14 = LOAD_DOUBLE R2
%16 = ADD_NUM %14, R3
STORE_DOUBLE R1, %16
STORE_TAG R1, tnumber
JUMP bb_4
bb_4:
INTERRUPT 5u
RETURN R1, 1i
)");
}
TEST_CASE("UserDataSetIndex")
{
ScopedFastFlag luauCodegenDirectUserdataFlow{FFlag::LuauCodegenDirectUserdataFlow, true};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function setxy(a: Point)
a.x = 3
a.y = 4
end
)"),
R"(
; function setxy($arg0) line 2
bb_0:
CHECK_TAG R0, tuserdata, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
STORE_DOUBLE R1, 3
STORE_TAG R1, tnumber
FALLBACK_SETTABLEKS 1u, R1, R0, K0
STORE_DOUBLE R1, 4
FALLBACK_SETTABLEKS 4u, R1, R0, K1
INTERRUPT 6u
RETURN R0, 0i
)");
}
TEST_CASE("UserDataNamecall")
{
ScopedFastFlag luauCodegenDirectUserdataFlow{FFlag::LuauCodegenDirectUserdataFlow, true};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function getxy(a: Point)
return a:GetX() + a:GetY()
end
)"),
R"(
; function getxy($arg0) line 2
bb_0:
CHECK_TAG R0, tuserdata, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
FALLBACK_NAMECALL 0u, R2, R0, K0
INTERRUPT 2u
SET_SAVEDPC 3u
CALL R2, 1i, 1i
FALLBACK_NAMECALL 3u, R3, R0, K1
INTERRUPT 5u
SET_SAVEDPC 6u
CALL R3, 1i, 1i
CHECK_TAG R2, tnumber, bb_fallback_3
CHECK_TAG R3, tnumber, bb_fallback_3
%20 = LOAD_DOUBLE R2
%22 = ADD_NUM %20, R3
STORE_DOUBLE R1, %22
STORE_TAG R1, tnumber
JUMP bb_4
bb_4:
INTERRUPT 7u
RETURN R1, 1i
)");
}
Sync to upstream/release/623 (#1236) # What's changed? ### New Type Solver - Unification of two fresh types no longer binds them together. - Replaced uses of raw `emplace` with `emplaceType` to catch cyclic bound types when they are created. - `SetIndexerConstraint` is blocked until the indexer result type is not blocked. - Fix a case where a blocked type got past the constraint solver. - Searching for free types should no longer traverse into `ClassType`s. - Fix a corner case that could result in the non-testable type `~{}`. - Fix incorrect flagging when `any` was a parameter of some checked function in nonstrict type checker. - `IterableConstraint` now consider tables without `__iter` to be iterables. ### Native Code Generation - Improve register type info lookup by program counter. - Generate type information for locals and upvalues --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: James McNellis <jmcnellis@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-26 06:26:09 +08:00
TEST_CASE("ExplicitUpvalueAndLocalTypes")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
{FFlag::LuauCodegenRemoveDeadStores5, true}};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local y: vector = ...
local function getsum(t)
local x: vector = t
return x.X + x.Y + y.X + y.Y
end
)",
/* includeIrTypes */ true),
R"(
; function getsum($arg0) line 4
; U0: vector
; R0: vector from 0 to 14
bb_bytecode_0:
CHECK_TAG R0, tvector, exit(0)
%2 = LOAD_FLOAT R0, 0i
STORE_DOUBLE R4, %2
STORE_TAG R4, tnumber
%7 = LOAD_FLOAT R0, 4i
%16 = ADD_NUM %2, %7
STORE_DOUBLE R3, %16
STORE_TAG R3, tnumber
GET_UPVALUE R5, U0
CHECK_TAG R5, tvector, exit(6)
%22 = LOAD_FLOAT R5, 0i
%31 = ADD_NUM %16, %22
STORE_DOUBLE R2, %31
STORE_TAG R2, tnumber
GET_UPVALUE R4, U0
CHECK_TAG R4, tvector, exit(10)
%37 = LOAD_FLOAT R4, 4i
%46 = ADD_NUM %31, %37
STORE_DOUBLE R1, %46
STORE_TAG R1, tnumber
INTERRUPT 13u
RETURN R1, 1i
)");
}
TEST_CASE("FastcallTypeInferThroughLocal")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
{FFlag::LuauCodegenRemoveDeadStores5, true}};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function getsum(x, c)
local v = vector(x, 2, 3)
if c then
return v.X + v.Y
else
return v.Z
end
end
)",
/* includeIrTypes */ true),
R"(
; function getsum($arg0, $arg1) line 2
; R2: vector from 0 to 17
bb_bytecode_0:
%0 = LOAD_TVALUE R0
STORE_TVALUE R3, %0
STORE_DOUBLE R4, 2
STORE_TAG R4, tnumber
STORE_DOUBLE R5, 3
STORE_TAG R5, tnumber
CHECK_SAFE_ENV exit(4)
CHECK_TAG R3, tnumber, exit(4)
%13 = LOAD_DOUBLE R3
STORE_VECTOR R2, %13, 2, 3
STORE_TAG R2, tvector
JUMP_IF_FALSY R1, bb_bytecode_1, bb_3
bb_3:
CHECK_TAG R2, tvector, exit(8)
%21 = LOAD_FLOAT R2, 0i
%26 = LOAD_FLOAT R2, 4i
%35 = ADD_NUM %21, %26
STORE_DOUBLE R3, %35
STORE_TAG R3, tnumber
INTERRUPT 13u
RETURN R3, 1i
bb_bytecode_1:
CHECK_TAG R2, tvector, exit(14)
%42 = LOAD_FLOAT R2, 8i
STORE_DOUBLE R3, %42
STORE_TAG R3, tnumber
INTERRUPT 16u
RETURN R3, 1i
)");
}
TEST_CASE("FastcallTypeInferThroughUpvalue")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
{FFlag::LuauCodegenRemoveDeadStores5, true}};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local v = ...
local function getsum(x, c)
v = vector(x, 2, 3)
if c then
return v.X + v.Y
else
return v.Z
end
end
)",
/* includeIrTypes */ true),
R"(
; function getsum($arg0, $arg1) line 4
; U0: vector
bb_bytecode_0:
%0 = LOAD_TVALUE R0
STORE_TVALUE R3, %0
STORE_DOUBLE R4, 2
STORE_TAG R4, tnumber
STORE_DOUBLE R5, 3
STORE_TAG R5, tnumber
CHECK_SAFE_ENV exit(4)
CHECK_TAG R3, tnumber, exit(4)
%13 = LOAD_DOUBLE R3
STORE_VECTOR R2, %13, 2, 3
STORE_TAG R2, tvector
SET_UPVALUE U0, R2, tvector
JUMP_IF_FALSY R1, bb_bytecode_1, bb_3
bb_3:
GET_UPVALUE R4, U0
CHECK_TAG R4, tvector, exit(10)
%23 = LOAD_FLOAT R4, 0i
STORE_DOUBLE R3, %23
STORE_TAG R3, tnumber
GET_UPVALUE R5, U0
CHECK_TAG R5, tvector, exit(13)
%29 = LOAD_FLOAT R5, 4i
%38 = ADD_NUM %23, %29
STORE_DOUBLE R2, %38
STORE_TAG R2, tnumber
INTERRUPT 16u
RETURN R2, 1i
bb_bytecode_1:
GET_UPVALUE R3, U0
CHECK_TAG R3, tvector, exit(18)
%46 = LOAD_FLOAT R3, 8i
STORE_DOUBLE R2, %46
STORE_TAG R2, tnumber
INTERRUPT 20u
RETURN R2, 1i
)");
}
TEST_CASE("LoadAndMoveTypePropagation")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}};
Sync to upstream/release/623 (#1236) # What's changed? ### New Type Solver - Unification of two fresh types no longer binds them together. - Replaced uses of raw `emplace` with `emplaceType` to catch cyclic bound types when they are created. - `SetIndexerConstraint` is blocked until the indexer result type is not blocked. - Fix a case where a blocked type got past the constraint solver. - Searching for free types should no longer traverse into `ClassType`s. - Fix a corner case that could result in the non-testable type `~{}`. - Fix incorrect flagging when `any` was a parameter of some checked function in nonstrict type checker. - `IterableConstraint` now consider tables without `__iter` to be iterables. ### Native Code Generation - Improve register type info lookup by program counter. - Generate type information for locals and upvalues --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: James McNellis <jmcnellis@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-26 06:26:09 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function getsum(n)
local seqsum = 0
for i = 1,n do
if i < 10 then
seqsum += i
else
seqsum *= i
end
end
return seqsum
end
)",
/* includeIrTypes */ true),
R"(
; function getsum($arg0) line 2
; R1: number from 0 to 13
; R4: number from 1 to 11
bb_bytecode_0:
STORE_DOUBLE R1, 0
STORE_TAG R1, tnumber
STORE_DOUBLE R4, 1
STORE_TAG R4, tnumber
%4 = LOAD_TVALUE R0
STORE_TVALUE R2, %4
STORE_DOUBLE R3, 1
STORE_TAG R3, tnumber
CHECK_TAG R2, tnumber, exit(4)
%12 = LOAD_DOUBLE R2
JUMP_CMP_NUM 1, %12, not_le, bb_bytecode_4, bb_bytecode_1
bb_bytecode_1:
INTERRUPT 5u
STORE_DOUBLE R5, 10
STORE_TAG R5, tnumber
CHECK_TAG R4, tnumber, bb_fallback_6
JUMP_CMP_NUM R4, 10, not_lt, bb_bytecode_2, bb_5
bb_5:
CHECK_TAG R1, tnumber, exit(8)
CHECK_TAG R4, tnumber, exit(8)
%32 = LOAD_DOUBLE R1
%34 = ADD_NUM %32, R4
STORE_DOUBLE R1, %34
JUMP bb_bytecode_3
bb_bytecode_2:
CHECK_TAG R1, tnumber, exit(10)
CHECK_TAG R4, tnumber, exit(10)
%41 = LOAD_DOUBLE R1
%43 = MUL_NUM %41, R4
STORE_DOUBLE R1, %43
JUMP bb_bytecode_3
bb_bytecode_3:
%46 = LOAD_DOUBLE R2
%47 = LOAD_DOUBLE R4
%48 = ADD_NUM %47, 1
STORE_DOUBLE R4, %48
JUMP_CMP_NUM %48, %46, le, bb_bytecode_1, bb_bytecode_4
bb_bytecode_4:
INTERRUPT 12u
RETURN R1, 1i
)");
}
TEST_CASE("ArgumentTypeRefinement")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}};
Sync to upstream/release/623 (#1236) # What's changed? ### New Type Solver - Unification of two fresh types no longer binds them together. - Replaced uses of raw `emplace` with `emplaceType` to catch cyclic bound types when they are created. - `SetIndexerConstraint` is blocked until the indexer result type is not blocked. - Fix a case where a blocked type got past the constraint solver. - Searching for free types should no longer traverse into `ClassType`s. - Fix a corner case that could result in the non-testable type `~{}`. - Fix incorrect flagging when `any` was a parameter of some checked function in nonstrict type checker. - `IterableConstraint` now consider tables without `__iter` to be iterables. ### Native Code Generation - Improve register type info lookup by program counter. - Generate type information for locals and upvalues --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: James McNellis <jmcnellis@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-26 06:26:09 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function getsum(x, y)
x = vector(1, y, 3)
return x.Y + x.Z
end
)",
/* includeIrTypes */ true),
R"(
; function getsum($arg0, $arg1) line 2
; R0: vector [argument]
bb_bytecode_0:
STORE_DOUBLE R3, 1
STORE_TAG R3, tnumber
%2 = LOAD_TVALUE R1
STORE_TVALUE R4, %2
STORE_DOUBLE R5, 3
STORE_TAG R5, tnumber
CHECK_SAFE_ENV exit(4)
CHECK_TAG R4, tnumber, exit(4)
%14 = LOAD_DOUBLE R4
STORE_VECTOR R2, 1, %14, 3
STORE_TAG R2, tvector
%18 = LOAD_TVALUE R2
STORE_TVALUE R0, %18
%22 = LOAD_FLOAT R0, 4i
%27 = LOAD_FLOAT R0, 8i
%36 = ADD_NUM %22, %27
STORE_DOUBLE R2, %36
STORE_TAG R2, tnumber
INTERRUPT 13u
RETURN R2, 1i
)");
}
TEST_CASE("InlineFunctionType")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}};
Sync to upstream/release/623 (#1236) # What's changed? ### New Type Solver - Unification of two fresh types no longer binds them together. - Replaced uses of raw `emplace` with `emplaceType` to catch cyclic bound types when they are created. - `SetIndexerConstraint` is blocked until the indexer result type is not blocked. - Fix a case where a blocked type got past the constraint solver. - Searching for free types should no longer traverse into `ClassType`s. - Fix a corner case that could result in the non-testable type `~{}`. - Fix incorrect flagging when `any` was a parameter of some checked function in nonstrict type checker. - `IterableConstraint` now consider tables without `__iter` to be iterables. ### Native Code Generation - Improve register type info lookup by program counter. - Generate type information for locals and upvalues --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: James McNellis <jmcnellis@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-04-26 06:26:09 +08:00
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function inl(v: vector, s: number)
return v.Y * s
end
local function getsum(x)
return inl(x, 2) + inl(x, 5)
end
)",
/* includeIrTypes */ true),
R"(
; function inl($arg0, $arg1) line 2
; R0: vector [argument]
; R1: number [argument]
bb_0:
CHECK_TAG R0, tvector, exit(entry)
CHECK_TAG R1, tnumber, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
%8 = LOAD_FLOAT R0, 4i
%17 = MUL_NUM %8, R1
STORE_DOUBLE R2, %17
STORE_TAG R2, tnumber
INTERRUPT 3u
RETURN R2, 1i
; function getsum($arg0) line 6
; R0: vector from 0 to 3
; R0: vector from 3 to 6
bb_bytecode_0:
CHECK_TAG R0, tvector, exit(0)
%2 = LOAD_FLOAT R0, 4i
%8 = MUL_NUM %2, 2
%13 = LOAD_FLOAT R0, 4i
%19 = MUL_NUM %13, 5
%28 = ADD_NUM %8, %19
STORE_DOUBLE R1, %28
STORE_TAG R1, tnumber
INTERRUPT 7u
RETURN R1, 1i
)");
}
TEST_CASE("ResolveTablePathTypes")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}};
CHECK_EQ("\n" + getCodegenAssembly(R"(
type Vertex = {pos: vector, normal: vector}
local function foo(arr: {Vertex}, i)
local v = arr[i]
return v.pos.Y
end
)",
/* includeIrTypes */ true, /* debugLevel */ 2),
R"(
; function foo(arr, i) line 4
; R0: table [argument 'arr']
; R2: table from 0 to 6 [local 'v']
; R4: vector from 3 to 5
bb_0:
CHECK_TAG R0, ttable, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
CHECK_TAG R1, tnumber, bb_fallback_3
%8 = LOAD_POINTER R0
%9 = LOAD_DOUBLE R1
%10 = TRY_NUM_TO_INDEX %9, bb_fallback_3
%11 = SUB_INT %10, 1i
CHECK_ARRAY_SIZE %8, %11, bb_fallback_3
CHECK_NO_METATABLE %8, bb_fallback_3
%14 = GET_ARR_ADDR %8, %11
%15 = LOAD_TVALUE %14
STORE_TVALUE R2, %15
JUMP bb_4
bb_4:
CHECK_TAG R2, ttable, exit(1)
%23 = LOAD_POINTER R2
%24 = GET_SLOT_NODE_ADDR %23, 1u, K0
CHECK_SLOT_MATCH %24, K0, bb_fallback_5
%26 = LOAD_TVALUE %24, 0i
STORE_TVALUE R4, %26
JUMP bb_6
bb_6:
CHECK_TAG R4, tvector, exit(3)
%33 = LOAD_FLOAT R4, 4i
STORE_DOUBLE R3, %33
STORE_TAG R3, tnumber
INTERRUPT 5u
RETURN R3, 1i
)");
}
TEST_CASE("ResolvableSimpleMath")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}};
CHECK_EQ("\n" + getCodegenHeader(R"(
type Vertex = { p: vector, uv: vector, n: vector, t: vector, b: vector, h: number }
local mesh: { vertices: {Vertex}, indices: {number} } = ...
local function compute()
for i = 1,#mesh.indices,3 do
local a = mesh.vertices[mesh.indices[i]]
local b = mesh.vertices[mesh.indices[i + 1]]
local c = mesh.vertices[mesh.indices[i + 2]]
local vba = b.p - a.p
local vca = c.p - a.p
local uvba = b.uv - a.uv
local uvca = c.uv - a.uv
local r = 1.0 / (uvba.X * uvca.Y - uvca.X * uvba.Y);
local sdir = (uvca.Y * vba - uvba.Y * vca) * r
a.t += sdir
end
end
)"),
R"(
; function compute() line 5
; U0: table ['mesh']
; R2: number from 0 to 78 [local 'i']
; R3: table from 7 to 78 [local 'a']
; R4: table from 15 to 78 [local 'b']
; R5: table from 24 to 78 [local 'c']
; R6: vector from 33 to 78 [local 'vba']
; R7: vector from 37 to 38
; R7: vector from 38 to 78 [local 'vca']
; R8: vector from 37 to 38
; R8: vector from 42 to 43
; R8: vector from 43 to 78 [local 'uvba']
; R9: vector from 42 to 43
; R9: vector from 47 to 48
; R9: vector from 48 to 78 [local 'uvca']
; R10: vector from 47 to 48
; R10: vector from 52 to 53
; R10: number from 53 to 78 [local 'r']
; R11: vector from 52 to 53
; R11: vector from 65 to 78 [local 'sdir']
; R12: vector from 72 to 73
; R12: vector from 75 to 76
; R13: vector from 71 to 72
; R14: vector from 71 to 72
)");
}
TEST_CASE("ResolveVectorNamecalls")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}, {FFlag::LuauCodegenDirectUserdataFlow, true},
{FFlag::LuauCodegenAnalyzeHostVectorOps, true}};
CHECK_EQ("\n" + getCodegenAssembly(R"(
type Vertex = {pos: vector, normal: vector}
local function foo(arr: {Vertex}, i)
return arr[i].normal:Dot(vector(0.707, 0, 0.707))
end
)",
/* includeIrTypes */ true),
R"(
; function foo($arg0, $arg1) line 4
; R0: table [argument]
; R2: vector from 4 to 6
bb_0:
CHECK_TAG R0, ttable, exit(entry)
JUMP bb_2
bb_2:
JUMP bb_bytecode_1
bb_bytecode_1:
CHECK_TAG R1, tnumber, bb_fallback_3
%8 = LOAD_POINTER R0
%9 = LOAD_DOUBLE R1
%10 = TRY_NUM_TO_INDEX %9, bb_fallback_3
%11 = SUB_INT %10, 1i
CHECK_ARRAY_SIZE %8, %11, bb_fallback_3
CHECK_NO_METATABLE %8, bb_fallback_3
%14 = GET_ARR_ADDR %8, %11
%15 = LOAD_TVALUE %14
STORE_TVALUE R3, %15
JUMP bb_4
bb_4:
CHECK_TAG R3, ttable, bb_fallback_5
%23 = LOAD_POINTER R3
%24 = GET_SLOT_NODE_ADDR %23, 1u, K0
CHECK_SLOT_MATCH %24, K0, bb_fallback_5
%26 = LOAD_TVALUE %24, 0i
STORE_TVALUE R2, %26
JUMP bb_6
bb_6:
%31 = LOAD_TVALUE K1, 0i, tvector
STORE_TVALUE R4, %31
CHECK_TAG R2, tvector, exit(4)
%37 = LOAD_FLOAT R2, 0i
%38 = LOAD_FLOAT R4, 0i
%39 = MUL_NUM %37, %38
%40 = LOAD_FLOAT R2, 4i
%41 = LOAD_FLOAT R4, 4i
%42 = MUL_NUM %40, %41
%43 = LOAD_FLOAT R2, 8i
%44 = LOAD_FLOAT R4, 8i
%45 = MUL_NUM %43, %44
%46 = ADD_NUM %39, %42
%47 = ADD_NUM %46, %45
STORE_DOUBLE R2, %47
STORE_TAG R2, tnumber
ADJUST_STACK_TO_REG R2, 1i
INTERRUPT 7u
RETURN R2, -1i
)");
}
TEST_CASE("ImmediateTypeAnnotationHelp")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}};
CHECK_EQ("\n" + getCodegenAssembly(R"(
local function foo(arr, i)
return (arr[i] :: vector) / 5
end
)",
/* includeIrTypes */ true),
R"(
; function foo($arg0, $arg1) line 2
; R3: vector from 1 to 2
bb_bytecode_0:
CHECK_TAG R0, ttable, bb_fallback_1
CHECK_TAG R1, tnumber, bb_fallback_1
%4 = LOAD_POINTER R0
%5 = LOAD_DOUBLE R1
%6 = TRY_NUM_TO_INDEX %5, bb_fallback_1
%7 = SUB_INT %6, 1i
CHECK_ARRAY_SIZE %4, %7, bb_fallback_1
CHECK_NO_METATABLE %4, bb_fallback_1
%10 = GET_ARR_ADDR %4, %7
%11 = LOAD_TVALUE %10
STORE_TVALUE R3, %11
JUMP bb_2
bb_2:
CHECK_TAG R3, tvector, exit(1)
%19 = LOAD_TVALUE R3
%20 = NUM_TO_VEC 5
%21 = DIV_VEC %19, %20
%22 = TAG_VECTOR %21
STORE_TVALUE R2, %22
INTERRUPT 2u
RETURN R2, 1i
)");
}
TEST_CASE("UnaryTypeResolve")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}};
CHECK_EQ("\n" + getCodegenHeader(R"(
local function foo(a, b: vector, c)
local d = not a
local e = -b
local f = #c
return (if d then e else vector(f, 2, 3)).X
end
)"),
R"(
; function foo(a, b, c) line 2
; R1: vector [argument 'b']
; R3: boolean from 0 to 16 [local 'd']
; R4: vector from 1 to 16 [local 'e']
; R5: number from 2 to 16 [local 'f']
; R7: vector from 13 to 15
)");
}
TEST_CASE("ForInManualAnnotation")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}};
CHECK_EQ("\n" + getCodegenAssembly(R"(
type Vertex = {pos: vector, normal: vector}
local function foo(a: {Vertex})
local sum = 0
for k, v: Vertex in ipairs(a) do
sum += v.pos.X
end
return sum
end
)",
/* includeIrTypes */ true, /* debugLevel */ 2),
R"(
; function foo(a) line 4
; R0: table [argument 'a']
; R1: number from 0 to 14 [local 'sum']
; R5: number from 5 to 11 [local 'k']
; R6: table from 5 to 11 [local 'v']
; R8: vector from 8 to 10
bb_0:
CHECK_TAG R0, ttable, exit(entry)
JUMP bb_4
bb_4:
JUMP bb_bytecode_1
bb_bytecode_1:
STORE_DOUBLE R1, 0
STORE_TAG R1, tnumber
CHECK_SAFE_ENV exit(1)
JUMP_EQ_TAG K1, tnil, bb_fallback_6, bb_5
bb_5:
%9 = LOAD_TVALUE K1
STORE_TVALUE R2, %9
JUMP bb_7
bb_7:
%15 = LOAD_TVALUE R0
STORE_TVALUE R3, %15
INTERRUPT 4u
SET_SAVEDPC 5u
CALL R2, 1i, 3i
CHECK_SAFE_ENV exit(5)
CHECK_TAG R3, ttable, bb_fallback_8
CHECK_TAG R4, tnumber, bb_fallback_8
JUMP_CMP_NUM R4, 0, not_eq, bb_fallback_8, bb_9
bb_9:
STORE_TAG R2, tnil
STORE_POINTER R4, 0i
STORE_EXTRA R4, 128i
STORE_TAG R4, tlightuserdata
JUMP bb_bytecode_3
bb_bytecode_2:
CHECK_TAG R6, ttable, exit(6)
%35 = LOAD_POINTER R6
%36 = GET_SLOT_NODE_ADDR %35, 6u, K2
CHECK_SLOT_MATCH %36, K2, bb_fallback_10
%38 = LOAD_TVALUE %36, 0i
STORE_TVALUE R8, %38
JUMP bb_11
bb_11:
CHECK_TAG R8, tvector, exit(8)
%45 = LOAD_FLOAT R8, 0i
STORE_DOUBLE R7, %45
STORE_TAG R7, tnumber
CHECK_TAG R1, tnumber, exit(10)
%52 = LOAD_DOUBLE R1
%54 = ADD_NUM %52, %45
STORE_DOUBLE R1, %54
JUMP bb_bytecode_3
bb_bytecode_3:
INTERRUPT 11u
CHECK_TAG R2, tnil, bb_fallback_13
%60 = LOAD_POINTER R3
%61 = LOAD_INT R4
%62 = GET_ARR_ADDR %60, %61
CHECK_ARRAY_SIZE %60, %61, bb_12
%64 = LOAD_TAG %62
JUMP_EQ_TAG %64, tnil, bb_12, bb_14
bb_14:
%66 = ADD_INT %61, 1i
STORE_INT R4, %66
%68 = INT_TO_NUM %66
STORE_DOUBLE R5, %68
STORE_TAG R5, tnumber
%71 = LOAD_TVALUE %62
STORE_TVALUE R6, %71
JUMP bb_bytecode_2
bb_12:
INTERRUPT 13u
RETURN R1, 1i
)");
}
TEST_CASE("ForInAutoAnnotationIpairs")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}};
CHECK_EQ("\n" + getCodegenHeader(R"(
type Vertex = {pos: vector, normal: vector}
local function foo(a: {Vertex})
local sum = 0
for k, v in ipairs(a) do
local n = v.pos.X
sum += n
end
return sum
end
)"),
R"(
; function foo(a) line 4
; R0: table [argument 'a']
; R1: number from 0 to 14 [local 'sum']
; R5: number from 5 to 11 [local 'k']
; R6: table from 5 to 11 [local 'v']
; R7: number from 6 to 11 [local 'n']
; R8: vector from 8 to 10
)");
}
TEST_CASE("ForInAutoAnnotationPairs")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}};
CHECK_EQ("\n" + getCodegenHeader(R"(
type Vertex = {pos: vector, normal: vector}
local function foo(a: {[string]: Vertex})
local sum = 0
for k, v in pairs(a) do
local n = v.pos.X
sum += n
end
return sum
end
)"),
R"(
; function foo(a) line 4
; R0: table [argument 'a']
; R1: number from 0 to 14 [local 'sum']
; R5: string from 5 to 11 [local 'k']
; R6: table from 5 to 11 [local 'v']
; R7: number from 6 to 11 [local 'n']
; R8: vector from 8 to 10
)");
}
TEST_CASE("ForInAutoAnnotationGeneric")
{
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}};
CHECK_EQ("\n" + getCodegenHeader(R"(
type Vertex = {pos: vector, normal: vector}
local function foo(a: {Vertex})
local sum = 0
for k, v in a do
local n = v.pos.X
sum += n
end
return sum
end
)"),
R"(
; function foo(a) line 4
; R0: table [argument 'a']
; R1: number from 0 to 13 [local 'sum']
; R5: number from 4 to 10 [local 'k']
; R6: table from 4 to 10 [local 'v']
; R7: number from 5 to 10 [local 'n']
; R8: vector from 7 to 9
)");
}
Sync to upstream/release/628 (#1278) ### What's new? * Remove a case of unsound `table.move` optimization * Add Luau stack slot reservations that were missing in REPL (fixes #1273) ### New Type Solver * Assignments have been completely reworked to fix a case of cyclic constraint dependency * When indexing, if the fresh type's upper bound already contains a compatible indexer, do not add another upper bound * Distribute type arguments over all type families sans `eq`, `keyof`, `rawkeyof`, and other internal type families * Fix a case where `buffers` component weren't read in two places (fixes #1267) * Fix a case where things that constitutes a strong ref were slightly incorrect * Fix a case where constraint dependencies weren't setup wrt `for ... in` statement ### Native Codegen * Fix an optimization that splits TValue store only when its value and its tag are compatible * Implement a system to plug additional type information for custom host userdata types --- ### Internal Contributors Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Alexander McCord <amccord@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh Vijay <vvijay@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com> --------- Co-authored-by: Aaron Weiss <aaronweiss@roblox.com> Co-authored-by: Andy Friesen <afriesen@roblox.com> Co-authored-by: Vighnesh <vvijay@roblox.com> Co-authored-by: Aviral Goel <agoel@roblox.com> Co-authored-by: David Cope <dcope@roblox.com> Co-authored-by: Lily Brown <lbrown@roblox.com> Co-authored-by: Vyacheslav Egorov <vegorov@roblox.com>
2024-06-01 03:18:18 +08:00
// Temporary test, when we don't compile new typeinfo, but support loading it
TEST_CASE("CustomUserdataTypesTemp")
{
// This test requires runtime component to be present
if (!Luau::CodeGen::isSupported())
return;
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}, {FFlag::LuauCompileUserdataInfo, false},
{FFlag::LuauLoadUserdataInfo, true}};
CHECK_EQ("\n" + getCodegenHeader(R"(
local function foo(v: vec2, x: mat3)
return v.X * x
end
)"),
R"(
; function foo(v, x) line 2
; R0: userdata [argument 'v']
; R1: userdata [argument 'x']
)");
}
TEST_CASE("CustomUserdataTypes")
{
// This test requires runtime component to be present
if (!Luau::CodeGen::isSupported())
return;
ScopedFastFlag sffs[]{{FFlag::LuauLoadTypeInfo, true}, {FFlag::LuauCompileTypeInfo, true}, {FFlag::LuauCodegenTypeInfo, true},
{FFlag::LuauCodegenRemoveDeadStores5, true}, {FFlag::LuauCompileTempTypeInfo, true}, {FFlag::LuauCompileUserdataInfo, true},
{FFlag::LuauLoadUserdataInfo, true}};
CHECK_EQ("\n" + getCodegenHeader(R"(
local function foo(v: vec2, x: mat3)
return v.X * x
end
)"),
R"(
; function foo(v, x) line 2
; R0: vec2 [argument 'v']
; R1: mat3 [argument 'x']
)");
}
TEST_SUITE_END();