[MC] Implement RV64GC assembly emission

4 files changed, 263 insertions, 0 deletions

diff --git a/backend/tools/AssemblyEmitter/AssemblyEmitterMain.cpp b/backend/tools/AssemblyEmitter/AssemblyEmitterMain.cpp
new file mode 100644
index 0000000..ff339fa
--- /dev/null
+++ b/backend/tools/AssemblyEmitter/AssemblyEmitterMain.cpp
@@ -0,0 +1,171 @@
+// SPDX-License-Identifier: GPL-3.0-or-later
+
+#include "rcc/CodeGen/MachineFunction.h"
+#include "rcc/CodeGen/MachineOptimizationPipeline.h"
+#include "rcc/CodeGen/MachinePass.h"
+#include "rcc/CodeGen/MachinePipelinePasses.h"
+#include "rcc/CodeGen/MachineSSAExit.h"
+#include "rcc/CodeGen/RegAllocPBQP.h"
+#include "rcc/MIR/MIRDocument.h"
+#include "rcc/Support/TextDiagnostic.h"
+#include "rcc/Target/RISCV/RISCVAssemblyPrinter.h"
+#include "rcc/Target/RISCV/RISCVCodeGen.h"
+#include "rcc/Target/RISCV/RISCVTarget.h"
+
+#include <fstream>
+#include <iostream>
+#include <iterator>
+#include <memory>
+#include <string>
+#include <string_view>
+#include <utility>
+
+namespace {
+
+struct CommandLine final {
+  std::string InputPath;
+};
+
+[[nodiscard]] bool parseCommandLine(int ArgumentCount, char *Arguments[], CommandLine &Result) {
+  for (int Index = 1; Index < ArgumentCount; ++Index) {
+    const std::string_view Argument(Arguments[Index]);
+    if (Argument != "--input" || Index + 1 >= ArgumentCount) {
+      std::cerr << "usage: rcc-backend-asm --input <mir-file|->\n";
+      return false;
+    }
+    Result.InputPath = Arguments[++Index];
+  }
+  if (Result.InputPath.empty()) {
+    std::cerr << "usage: rcc-backend-asm --input <mir-file|->\n";
+    return false;
+  }
+  return true;
+}
+
+[[nodiscard]] bool readInput(const CommandLine &Command, std::string &Contents) {
+  if (Command.InputPath == "-") {
+    Contents.assign(std::istreambuf_iterator<char>(std::cin), std::istreambuf_iterator<char>());
+    return true;
+  }
+  std::ifstream Input(Command.InputPath, std::ios::binary);
+  if (!Input) {
+    std::cerr << "rcc-backend-asm: cannot open input '" << Command.InputPath << "'\n";
+    return false;
+  }
+  Contents.assign(std::istreambuf_iterator<char>(Input), std::istreambuf_iterator<char>());
+  return true;
+}
+
+template <typename PassT, typename... ArgumentT>
+[[nodiscard]] rcc::Expected<void> addPass(rcc::MachinePassManager &Pipeline,
+                                          ArgumentT &&...Arguments) {
+  return Pipeline.addPass(std::make_unique<PassT>(std::forward<ArgumentT>(Arguments)...));
+}
+
+[[nodiscard]] rcc::Expected<void> runBackendPipeline(rcc::MachineFunction &Function,
+                                                     const rcc::TargetDescription &Target) {
+  rcc::MachinePassManager Pipeline;
+  const rcc::MachineOptimizationPipelineTargetInfo TargetInfo =
+      rcc::riscv::optimizationPipelineInfo();
+
+  if (!Function.properties().has(rcc::MachineFunctionProperty::Legalized)) {
+    auto Added = addPass<rcc::LegalizerPass>(Pipeline, rcc::riscv::legalizerInfo());
+    if (!Added)
+      return std::unexpected(Added.error());
+  }
+  if (!Function.properties().has(rcc::MachineFunctionProperty::RegBankSelected)) {
+    auto Added = addPass<rcc::RegisterBankSelectPass>(Pipeline, rcc::riscv::registerBankInfo());
+    if (!Added)
+      return std::unexpected(Added.error());
+  }
+  if (!Function.properties().has(rcc::MachineFunctionProperty::Selected)) {
+    auto Added = addPass<rcc::InstructionSelectPass>(Pipeline, rcc::riscv::instructionSelector());
+    if (!Added)
+      return std::unexpected(Added.error());
+  }
+  if (!Function.properties().has(rcc::MachineFunctionProperty::NoPHIs)) {
+    auto Added = rcc::addSelectedSSAOptimizationPipeline(Pipeline, TargetInfo);
+    if (!Added)
+      return std::unexpected(Added.error());
+    Added = addPass<rcc::PHIEliminationPass>(Pipeline, rcc::riscv::ssaInfo());
+    if (!Added)
+      return std::unexpected(Added.error());
+  }
+  if (!Function.properties().has(rcc::MachineFunctionProperty::TiedOpsRewritten)) {
+    auto Added = addPass<rcc::TwoAddressRewritePass>(Pipeline, rcc::riscv::ssaInfo());
+    if (!Added)
+      return std::unexpected(Added.error());
+  }
+  if (!Function.properties().has(rcc::MachineFunctionProperty::NoVRegs)) {
+    auto Added = rcc::addPreRegisterAllocationOptimizationPipeline(Pipeline, TargetInfo);
+    if (!Added)
+      return std::unexpected(Added.error());
+    Added = addPass<rcc::PBQPAllocatorPass>(Pipeline, rcc::riscv::registerAllocationInfo());
+    if (!Added)
+      return std::unexpected(Added.error());
+  }
+  if (!Function.properties().has(rcc::MachineFunctionProperty::NoExecutablePseudos)) {
+    auto Added = rcc::addPostRegisterAllocationPipeline(Pipeline, TargetInfo);
+    if (!Added)
+      return std::unexpected(Added.error());
+  }
+
+  rcc::MachineAnalysisManager Analyses;
+  auto Registered = rcc::registerMachineOptimizationAnalyses(Analyses);
+  if (!Registered)
+    return std::unexpected(Registered.error());
+  return Pipeline.run(Function, Target, Analyses);
+}
+
+} // namespace
+
+int main(int ArgumentCount, char *Arguments[]) {
+  CommandLine Command;
+  if (!parseCommandLine(ArgumentCount, Arguments, Command))
+    return 1;
+
+  std::string SourceText;
+  if (!readInput(Command, SourceText))
+    return 1;
+
+  auto Target = rcc::riscv::createTargetDescription();
+  if (!Target) {
+    std::cerr << "rcc-backend-asm: cannot create RV64GC target: " << Target.error().message()
+              << '\n';
+    return 1;
+  }
+  rcc::mir::MIRDocumentParserOptions Options;
+  Options.DataLayout = "e-m:e-p:64:64-i64:64-i128:128-n32:64-S128";
+  Options.DefaultPointerSizeInBits = 64;
+  auto Parsed =
+      rcc::mir::parseMIRDocument(*Target, Command.InputPath, std::move(SourceText), Options);
+  if (!Parsed.succeeded()) {
+    for (const rcc::TextDiagnostic &Diagnostic : Parsed.diagnostics())
+      std::cerr << rcc::formatTextDiagnostic(Diagnostic, Parsed.source()) << '\n';
+    return 1;
+  }
+
+  rcc::CompilationUnit &Unit = Parsed.document()->unit();
+  for (std::size_t Index = 0; Index < Unit.functionCount(); ++Index) {
+    rcc::MachineFunction *Function = Unit.functionAt(Index);
+    if (Function == nullptr) {
+      std::cerr << "rcc-backend-asm: parsed document contains a null function\n";
+      return 1;
+    }
+    auto Run = runBackendPipeline(*Function, *Target);
+    if (!Run) {
+      std::cerr << "rcc-backend-asm: backend pipeline failed for '" << Function->name()
+                << "': " << Run.error().message() << '\n';
+      return 1;
+    }
+  }
+
+  auto Assembly = rcc::riscv::emitAssembly(Unit, *Target);
+  if (!Assembly) {
+    std::cerr << "rcc-backend-asm: assembly emission failed: " << Assembly.error().message()
+              << '\n';
+    return 1;
+  }
+  std::cout << *Assembly;
+  return 0;
+}
diff --git a/backend/tools/AssemblyEmitter/AssemblyPipelineTest.cmake b/backend/tools/AssemblyEmitter/AssemblyPipelineTest.cmake
new file mode 100644
index 0000000..baeedcf
--- /dev/null
+++ b/backend/tools/AssemblyEmitter/AssemblyPipelineTest.cmake
@@ -0,0 +1,52 @@
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+if(NOT DEFINED EMITTER OR NOT DEFINED INPUT OR NOT DEFINED OUTPUT)
+  message(FATAL_ERROR "assembly pipeline test is missing a required path")
+endif()
+
+execute_process(
+  COMMAND "${EMITTER}" --input "${INPUT}"
+  RESULT_VARIABLE FirstResult
+  OUTPUT_VARIABLE FirstAssembly
+  ERROR_VARIABLE FirstDiagnostics
+)
+if(NOT FirstResult EQUAL 0)
+  message(FATAL_ERROR "assembly pipeline failed: ${FirstDiagnostics}")
+endif()
+if(NOT FirstAssembly MATCHES "\\.attribute arch, \"rv64gc\"" OR
+   NOT FirstAssembly MATCHES "jalr x0, x1, 0")
+  message(FATAL_ERROR "assembly pipeline output is missing required RV64GC content")
+endif()
+if(INPUT MATCHES "RISCVDataPipeline" AND
+   (NOT FirstAssembly MATCHES "\\.LCPI[0-9]+_0:" OR
+    NOT FirstAssembly MATCHES "\\.LJTI[0-9]+_0:"))
+  message(FATAL_ERROR "assembly pipeline did not emit its constant pool and jump table")
+endif()
+
+execute_process(
+  COMMAND "${EMITTER}" --input "${INPUT}"
+  RESULT_VARIABLE SecondResult
+  OUTPUT_VARIABLE SecondAssembly
+  ERROR_VARIABLE SecondDiagnostics
+)
+if(NOT SecondResult EQUAL 0)
+  message(FATAL_ERROR "repeated assembly pipeline failed: ${SecondDiagnostics}")
+endif()
+if(NOT FirstAssembly STREQUAL SecondAssembly)
+  message(FATAL_ERROR "assembly pipeline output is not deterministic")
+endif()
+
+if(DEFINED LLVM_MC AND NOT LLVM_MC STREQUAL "LLVM_MC-NOTFOUND" AND
+   NOT LLVM_MC STREQUAL "")
+  file(WRITE "${OUTPUT}" "${FirstAssembly}")
+  execute_process(
+    COMMAND "${LLVM_MC}" -triple=riscv64-unknown-linux-gnu -filetype=null "${OUTPUT}"
+    RESULT_VARIABLE AssemblerResult
+    OUTPUT_VARIABLE AssemblerOutput
+    ERROR_VARIABLE AssemblerDiagnostics
+  )
+  file(REMOVE "${OUTPUT}")
+  if(NOT AssemblerResult EQUAL 0)
+    message(FATAL_ERROR "LLVM MC rejected generated assembly: ${AssemblerDiagnostics}")
+  endif()
+endif()
diff --git a/backend/tools/AssemblyEmitter/CMakeLists.txt b/backend/tools/AssemblyEmitter/CMakeLists.txt
new file mode 100644
index 0000000..a262983
--- /dev/null
+++ b/backend/tools/AssemblyEmitter/CMakeLists.txt
@@ -0,0 +1,39 @@
+# SPDX-License-Identifier: GPL-3.0-or-later
+
+add_executable(RCCBackendAssemblyEmitter AssemblyEmitterMain.cpp)
+rcc_backend_configure_target(RCCBackendAssemblyEmitter)
+target_link_libraries(
+  RCCBackendAssemblyEmitter
+  PRIVATE
+    RCCBackend::MIR
+    RCCBackend::RISCV
+)
+
+set_target_properties(
+  RCCBackendAssemblyEmitter
+  PROPERTIES
+    OUTPUT_NAME rcc-backend-asm
+)
+
+install(
+  TARGETS RCCBackendAssemblyEmitter
+  EXPORT RCCBackendTargets
+  RUNTIME DESTINATION "${CMAKE_INSTALL_BINDIR}"
+)
+
+find_program(RCCBackendLLVMMC NAMES llvm-mc-20 llvm-mc)
+foreach(InputName IN ITEMS RISCVGenericPipeline RISCVWideConstantPipeline RISCVDataPipeline)
+  add_test(
+    NAME RCCBackend.AssemblyEmitter.${InputName}
+    COMMAND
+      "${CMAKE_COMMAND}"
+      "-DEMITTER=$<TARGET_FILE:RCCBackendAssemblyEmitter>"
+      "-DINPUT=${RCC_BACKEND_SOURCE_DIR}/unittests/Target/Inputs/${InputName}.mir"
+      "-DOUTPUT=${CMAKE_CURRENT_BINARY_DIR}/${InputName}.s"
+      "-DLLVM_MC=${RCCBackendLLVMMC}"
+      -P "${CMAKE_CURRENT_SOURCE_DIR}/AssemblyPipelineTest.cmake"
+  )
+endforeach()
+
+rcc_backend_register_format_files("${CMAKE_CURRENT_SOURCE_DIR}/AssemblyEmitterMain.cpp")
+rcc_backend_register_tidy_files("${CMAKE_CURRENT_SOURCE_DIR}/AssemblyEmitterMain.cpp")
diff --git a/backend/tools/CMakeLists.txt b/backend/tools/CMakeLists.txt
index 67d7273..403f4a9 100644
--- a/backend/tools/CMakeLists.txt
+++ b/backend/tools/CMakeLists.txt
@@ -1,3 +1,4 @@
 # SPDX-License-Identifier: GPL-3.0-or-later
 
 add_subdirectory(DescriptionGenerator)
+add_subdirectory(AssemblyEmitter)