lifter: update tool commandline options and add flags for debug logs

.clang-tidy

@@ -82,13 +82,13 @@ CheckOptions:
   - key: 'readability-identifier-naming.EnumConstantCase'
     value: 'UPPER_CASE'
   - key: 'readability-identifier-naming.FunctionCase'
-    value: 'lower_case'
+    value: 'camelBack'
   - key: 'readability-identifier-naming.GlobalConstantCase'
     value: 'lower_case'
   - key: 'readability-identifier-naming.GlobalConstantPointerCase'
     value: 'lower_case'
   - key: 'readability-identifier-naming.GlobalFunctionCase'
-    value: 'lower_case'
+    value: 'camelBack'
   - key: 'readability-identifier-naming.GlobalPointerCase'
     value: 'lower_case'
   - key: 'readability-identifier-naming.GlobalVariableCase'

include/patchestry/AST/ASTConsumer.hpp

@@ -27,6 +27,7 @@
 #include <patchestry/AST/TypeBuilder.hpp>
 #include <patchestry/Ghidra/JsonDeserialize.hpp>
 #include <patchestry/Ghidra/PcodeOperations.hpp>
+#include <patchestry/Util/Options.hpp>
 
 namespace patchestry::ast {
     using namespace patchestry::ghidra;
@@ -38,16 +39,20 @@ namespace patchestry::ast {
     {
       public:
         explicit PcodeASTConsumer(
-            clang::CompilerInstance &ci, Program &prog, std::string &outfile
+            clang::CompilerInstance &ci, Program &prog, patchestry::Options &opts
         )
             : program(prog)
             , ci(ci)
-            , outfile(outfile)
+            , options(opts)
             , codegen(std::make_unique< CodeGenerator >(ci))
             , type_builder(std::make_unique< TypeBuilder >(ci.getASTContext())) {}
 
         void HandleTranslationUnit(clang::ASTContext &ctx) override;
 
+        const std::unordered_map< void *, std::string > &locations(void) const {
+            return location_map;
+        }
+
       private:
         void set_sema_context(clang::DeclContext *dc);
 
@@ -66,7 +71,8 @@ namespace patchestry::ast {
         std::reference_wrapper< Program > program;
         std::reference_wrapper< clang::CompilerInstance > ci;
 
-        std::string outfile;
+        const patchestry::Options &options;
+
         std::unique_ptr< CodeGenerator > codegen;
         std::unique_ptr< TypeBuilder > type_builder;
 

include/patchestry/Util/Options.hpp

@@ -0,0 +1,38 @@
+/*
+ * Copyright (c) 2024, Trail of Bits, Inc.
+ *
+ * This source code is licensed in accordance with the terms specified in
+ * the LICENSE file found in the root directory of this source tree.
+ */
+
+#pragma once
+
+#include <string>
+#include <vector>
+
+namespace patchestry {
+
+    enum class EmitMLIRType : int {
+        hl,  // Vast High-level dialects
+        cir, // Clang IR dialects
+    };
+
+    struct Options
+    {
+        bool emit_mlir  = false;
+        bool emit_tower = false;
+        bool emit_llvm  = false;
+        bool emit_asm   = false;
+        bool emit_obj   = false;
+        bool verbose    = false;
+
+        std::string output_file;
+        std::string input_file;
+
+        bool print_tu = false;
+
+        std::vector< std::string > pipelines = {};
+        EmitMLIRType mlir_type               = EmitMLIRType::hl;
+    };
+
+} // namespace patchestry

lib/patchestry/AST/ASTConsumer.cpp

@@ -33,6 +33,7 @@
 #include <patchestry/Ghidra/JsonDeserialize.hpp>
 #include <patchestry/Ghidra/Pcode.hpp>
 #include <patchestry/Ghidra/PcodeOperations.hpp>
+#include <patchestry/Util/Log.hpp>
 
 namespace patchestry::ast {
 
@@ -51,19 +52,21 @@ namespace patchestry::ast {
             );
         }
 
-        std::error_code ec;
-        auto out =
-            std::make_unique< llvm::raw_fd_ostream >(outfile, ec, llvm::sys::fs::OF_Text);
-
-        llvm::errs() << "Print AST dump\n";
-        ctx.getTranslationUnitDecl()->dumpColor();
-
-        ctx.getTranslationUnitDecl()->print(
-            *llvm::dyn_cast< llvm::raw_ostream >(out), ctx.getPrintingPolicy(), 0
-        );
+        if (options.print_tu) {
+#ifdef ENABLE_DEBUG
+            ctx.getTranslationUnitDecl()->dumpColor();
+#endif
+            std::error_code ec;
+            auto out = std::make_unique< llvm::raw_fd_ostream >(
+                options.output_file + ".c", ec, llvm::sys::fs::OF_Text
+            );
+            ctx.getTranslationUnitDecl()->print(
+                *llvm::dyn_cast< llvm::raw_ostream >(out), ctx.getPrintingPolicy(), 0
+            );
+        }
 
-        llvm::errs() << "Generate mlir\n";
-        llvm::raw_fd_ostream file_os(outfile + ".mlir", ec);
+        std::error_code ec;
+        llvm::raw_fd_ostream file_os(options.output_file + ".mlir", ec);
         codegen->generate_source_ir(ctx, location_map, file_os);
     }
 

test/pcode-lifter/adc_init.json

@@ -1,7 +1,6 @@
 // RUN: bash %strip-json-comments %s > %t.json
-// RUN: %pcode-lifter %t.json -output %t
-// RUN: %file-check --input-file %t --check-prefix=EXISTS < /dev/null
-// CHECK-EMPTY:
+// RUN: %pcode-lifter -input %t.json -output %t
+// RUN: test -e %t
 {
   "arch": "ARM",
   "format": "Executable and Linking Format (ELF)",

test/pcode-lifter/adc_read.json

@@ -1,7 +1,6 @@
 // RUN: bash %strip-json-comments %s > %t.json
-// RUN: %pcode-lifter %t.json -output %t
-// RUN: %file-check --input-file %t --check-prefix=EXISTS < /dev/null
-// CHECK-EMPTY:
+// RUN: %pcode-lifter -input %t.json -output %t
+// RUN: test -e %t
 {
   "arch": "ARM",
   "format": "Executable and Linking Format (ELF)",

tools/pcode-lifter/main.cpp

@@ -7,10 +7,14 @@
  */
 
 #include <cstdlib>
+#include <fstream>
 #include <memory>
+#include <ranges>
+#include <string_view>
 
 #include <clang/AST/ASTConsumer.h>
 #include <clang/AST/ASTContext.h>
+#include <clang/AST/Stmt.h>
 #include <clang/Basic/DiagnosticOptions.h>
 #include <clang/Basic/TargetInfo.h>
 #include <clang/Basic/TargetOptions.h>
@@ -27,36 +31,125 @@
 #include <patchestry/AST/ASTConsumer.hpp>
 #include <patchestry/Ghidra/JsonDeserialize.hpp>
 #include <patchestry/Util/Log.hpp>
+#include <patchestry/Util/Options.hpp>
 
-const llvm::cl::opt< std::string > input_filename(
-    llvm::cl::Positional, llvm::cl::desc("<input JSON file>"), llvm::cl::Required
-);
+namespace {
 
-const llvm::cl::opt< bool >
-    verbose("v", llvm::cl::desc("Enable debug logs"), llvm::cl::init(false));
+    const llvm::cl::opt< patchestry::EmitMLIRType > emit_mlir( // NOLINT(cert-err58-cpp)
+        "emit-mlir", llvm::cl::desc("MLIR Emission Type"),
+        llvm::cl::values(
+            clEnumVal(patchestry::EmitMLIRType::hl, "High-Level VAST MLIR Representation"),
+            clEnumVal(patchestry::EmitMLIRType::cir, "ClangIR representation")
+        ),
+        llvm::cl::init(patchestry::EmitMLIRType::hl)
+    );
 
-const llvm::cl::opt< bool > pprint(
-    "pretty-print", llvm::cl::desc("Pretty print translation unit"), llvm::cl::init(false)
-);
+    // NOTE: Option field is there to test vast tower api. Will be removed in future.
+    const llvm::cl::opt< bool > emit_tower( // NOLINT(cert-err58-cpp)
+        "emit-tower", llvm::cl::desc("Emit MLIR tower representation"), llvm::cl::init(false)
+    );
 
-const llvm::cl::opt< std::string > output_filename(
-    "output", llvm::cl::desc("Specify output filename"), llvm::cl::value_desc("filename"),
-    llvm::cl::init("/tmp/output.c")
-);
+    const llvm::cl::opt< bool > emit_llvm( // NOLINT(cert-err58-cpp)
+        "emit-llvm", llvm::cl::desc("Emit LLVM IR Representation"), llvm::cl::init(false)
+    );
 
-bool debug_mode = false;
+    const llvm::cl::opt< bool > emit_asm( // NOLINT(cert-err58-cpp)
+        "emit-asm", llvm::cl::desc("Emit ASM Representation"), llvm::cl::init(false)
+    );
 
-int main(int argc, char **argv) {
-    llvm::cl::ParseCommandLineOptions(
-        argc, argv, "pcode-lifter to lift high pcode into clang ast\n"
+    const llvm::cl::opt< bool > emit_obj( // NOLINT(cert-err58-cpp)
+        "emit-obj", llvm::cl::desc("Emit Object file"), llvm::cl::init(false)
+    );
+
+    const llvm::cl::opt< std::string > input_filename( // NOLINT(cert-err58-cpp)
+        "input", llvm::cl::desc("Input JSON file"), llvm::cl::Required
+    );
+
+    const llvm::cl::opt< std::string > output_filename( // NOLINT(cert-err58-cpp)
+        "output", llvm::cl::desc("Specify output filename"), llvm::cl::value_desc("filename"),
+        llvm::cl::init("") // Initialize with empty string
+    );
+
+    const llvm::cl::opt< bool > verbose( // NOLINT(cert-err58-cpp)
+        "verbose", llvm::cl::desc("Enable debug logs"), llvm::cl::init(false)
+    ); // NOLINT(cert-err58-cpp)
+
+    const llvm::cl::opt< bool > print_tu( // NOLINT(cert-err58-cpp)
+        "print-tu", llvm::cl::desc("Pretty print translation unit"), llvm::cl::init(false)
     );
 
-    if (verbose) {
-        debug_mode = true;
+    const llvm::cl::opt< std::string > pipelines( // NOLINT(cert-err58-cpp)
+        "pipelines", llvm::cl::desc("Specify pipelines for lowering steps"),
+        llvm::cl::value_desc("string"), llvm::cl::init("")
+    );
+
+    patchestry::Options parseCommandLineOptions(int argc, char **argv) {
+        llvm::cl::ParseCommandLineOptions(
+            argc, argv, "patche-lifter to represent high pcode into mlir representations\n"
+        );
+
+        auto split_pipelines = [&](std::string_view pipelines,
+                                   char delim = ',') -> std::vector< std::string > {
+            std::vector< std::string > vec;
+            size_t start = 0;
+            size_t end   = 0;
+            while ((end = pipelines.find(delim, start)) != std::string_view::npos) {
+                vec.emplace_back(pipelines.substr(start, end - start));
+                start = end + 1;
+            }
+            vec.emplace_back(pipelines.substr(start)); // Last part
+            return vec;
+        };
+
+        auto pipeline_stages = split_pipelines(pipelines.getValue());
+        return {
+            .emit_mlir   = true, // It is set to true by default
+            .emit_llvm   = emit_llvm.getValue(),
+            .emit_asm    = emit_asm.getValue(),
+            .emit_obj    = emit_obj.getValue(),
+            .verbose     = verbose.getValue(),
+            .output_file = output_filename.getValue(),
+            .input_file  = input_filename.getValue(),
+            .print_tu    = print_tu.getValue(),
+            .pipelines   = std::move(pipeline_stages),
+        };
+    }
+
+    void createSourceManager(clang::CompilerInstance &ci) {
+        // Create file manager and setup source manager
+        ci.createFileManager();
+        ci.createSourceManager(ci.getFileManager());
+
+        // get source manager and setup main_file_id for the source manager
+        auto &sm = ci.getSourceManager();
+
+        // Create fake file to support real file system needed for vast
+        // location translation
+        std::string data      = "/patchestry";
+        std::string file_name = "/tmp/patchestry";
+        std::ofstream(file_name) << data;
+        llvm::ErrorOr< clang::FileEntryRef > file_entry_ref_or_err =
+            ci.getFileManager().getVirtualFileRef(file_name, data.size(), 0);
+        clang::FileID file_id = sm.createFileID(
+            *file_entry_ref_or_err, clang::SourceLocation(), clang::SrcMgr::C_User, 0
+        );
+        sm.setMainFileID(file_id);
+    }
+
+    void setCodegenOptions(clang::CompilerInstance &ci) {
+        clang::CodeGenOptions &cg_opts = ci.getCodeGenOpts();
+        cg_opts.OptimizationLevel      = 0;
+        cg_opts.StrictReturn           = false;
+        cg_opts.StrictEnums            = false;
     }
 
+} // namespace
+
+int main(int argc, char **argv) {
+    auto options = parseCommandLineOptions(argc, argv);
+
     llvm::ErrorOr< std::unique_ptr< llvm::MemoryBuffer > > file_or_err =
-        llvm::MemoryBuffer::getFile(input_filename);
+        llvm::MemoryBuffer::getFile(options.input_file);
 
     if (std::error_code error_code = file_or_err.getError()) {
         LOG(ERROR) << "Error reading json file : " << error_code.message() << "\n";
@@ -94,29 +187,17 @@ int main(int argc, char **argv) {
 
     ci.getFrontendOpts().ProgramAction = clang::frontend::ParseSyntaxOnly;
     ci.getLangOpts().C99               = true;
-    // Setup file manager and source manager
-    ci.createFileManager();
-    ci.createSourceManager(ci.getFileManager());
-
-    auto &sm              = ci.getSourceManager();
-    std::string file_data = "/patchestry";
-    llvm::ErrorOr< clang::FileEntryRef > file_entry_ref_or_err =
-        ci.getFileManager().getVirtualFileRef("/tmp/patchestry", file_data.size(), 0);
-    clang::FileID file_id = sm.createFileID(
-        *file_entry_ref_or_err, clang::SourceLocation(), clang::SrcMgr::C_User, 0
-    );
 
-    sm.setMainFileID(file_id);
+    createSourceManager(ci);
+    setCodegenOptions(ci);
 
     // Create the preprocessor and AST context
     ci.createPreprocessor(clang::TU_Complete);
     ci.createASTContext();
 
     auto &ast_context = ci.getASTContext();
-
-    std::string outfile = output_filename.getValue();
     std::unique_ptr< patchestry::ast::PcodeASTConsumer > consumer =
-        std::make_unique< patchestry::ast::PcodeASTConsumer >(ci, program.value(), outfile);
+        std::make_unique< patchestry::ast::PcodeASTConsumer >(ci, program.value(), options);
     ci.setASTConsumer(std::move(consumer));
     ci.createSema(clang::TU_Complete, nullptr);