clang: add post about in memory c -> llvm ir

3 files changed, 228 insertions, 0 deletions

diff --git a/content/2022-06-18-libclang-c-to-llvm-ir/.clang-format b/content/2022-06-18-libclang-c-to-llvm-ir/.clang-format
new file mode 100644
index 0000000..9a0c292
--- /dev/null
+++ b/content/2022-06-18-libclang-c-to-llvm-ir/.clang-format
@@ -0,0 +1,104 @@
+# dotfiles -- clang-format
+# author: johannst
+# doc   : https://clang.llvm.org/docs/ClangFormatStyleOptions.html
+
+Language: Cpp
+Standard: Auto
+
+AccessModifierOffset: -2
+AlignAfterOpenBracket: Align
+AlignConsecutiveMacros: true
+AlignConsecutiveAssignments: false
+AlignConsecutiveDeclarations: false
+AlignEscapedNewlines: Left
+AlignOperands: true
+AlignTrailingComments: true
+AllowAllArgumentsOnNextLine: true
+AllowAllConstructorInitializersOnNextLine: true
+AllowAllParametersOfDeclarationOnNextLine: true
+AllowShortBlocksOnASingleLine: false
+AllowShortCaseLabelsOnASingleLine: false
+AllowShortFunctionsOnASingleLine: Inline
+AllowShortIfStatementsOnASingleLine: Never
+AllowShortLambdasOnASingleLine: All
+AllowShortLoopsOnASingleLine: false
+AlwaysBreakAfterReturnType: None
+AlwaysBreakBeforeMultilineStrings: true
+AlwaysBreakTemplateDeclarations: Yes
+BinPackArguments: true
+BinPackParameters: true
+BreakBeforeBraces: Custom
+BraceWrapping:
+  AfterCaseLabel:        false
+  AfterClass:            false
+  AfterControlStatement: false
+  AfterEnum:             false
+  AfterFunction:         false
+  AfterNamespace:        false
+  AfterStruct:           false
+  AfterUnion:            false
+  AfterExternBlock:      false
+  BeforeCatch:           false
+  BeforeElse:            false
+  IndentBraces:          false
+  SplitEmptyFunction:    true
+  SplitEmptyRecord:      true
+  SplitEmptyNamespace:   true
+BreakBeforeBinaryOperators: None
+BreakBeforeTernaryOperators: true
+BreakConstructorInitializers: AfterColon
+BreakConstructorInitializersBeforeComma: false
+BreakInheritanceList: BeforeColon
+BreakBeforeInheritanceComma: false
+BreakStringLiterals: true
+ColumnLimit: 80
+CompactNamespaces: true
+ConstructorInitializerAllOnOneLineOrOnePerLine: false
+ConstructorInitializerIndentWidth: 2
+ContinuationIndentWidth: 4
+Cpp11BracedListStyle: true
+DerivePointerAlignment: false
+DisableFormat:   false
+ExperimentalAutoDetectBinPacking: false
+FixNamespaceComments: true
+IncludeBlocks:   Preserve
+# Could use `IncludeCategories` to define rules for includes
+IndentCaseLabels: true
+IndentPPDirectives: AfterHash
+IndentWidth:     4
+IndentWrappedFunctionNames: false
+KeepEmptyLinesAtTheStartOfBlocks: false
+MacroBlockBegin: ''
+MacroBlockEnd:   ''
+MaxEmptyLinesToKeep: 2
+NamespaceIndentation: All
+PenaltyBreakAssignment: 2
+PenaltyBreakBeforeFirstCallParameter: 1
+PenaltyBreakComment: 300
+PenaltyBreakFirstLessLess: 120
+PenaltyBreakString: 1000
+PenaltyBreakTemplateDeclaration: 10
+PenaltyExcessCharacter: 1000000
+PenaltyReturnTypeOnItsOwnLine: 200
+PointerAlignment: Left
+ReflowComments:  true
+SortIncludes:    true
+SortUsingDeclarations: true
+SpaceAfterCStyleCast: false
+SpaceAfterLogicalNot: false
+SpaceAfterTemplateKeyword: false
+SpaceBeforeAssignmentOperators: true
+SpaceBeforeCpp11BracedList: false
+SpaceBeforeCtorInitializerColon: true
+SpaceBeforeInheritanceColon: true
+SpaceBeforeParens: ControlStatements
+SpaceBeforeRangeBasedForLoopColon: true
+SpaceInEmptyParentheses: false
+SpacesBeforeTrailingComments: 2
+SpacesInAngles:  false
+SpacesInCStyleCastParentheses: false
+SpacesInContainerLiterals: true
+SpacesInParentheses: false
+SpacesInSquareBrackets: false
+TabWidth:        4
+UseTab:          Never
diff --git a/content/2022-06-18-libclang-c-to-llvm-ir/Makefile b/content/2022-06-18-libclang-c-to-llvm-ir/Makefile
new file mode 100644
index 0000000..ec36db3
--- /dev/null
+++ b/content/2022-06-18-libclang-c-to-llvm-ir/Makefile
@@ -0,0 +1,27 @@
+LIBS  = $(shell llvm-config --libs)
+LIBS += -lclang-cpp
+
+CXXFLAGS  = -Wall
+CXXFLAGS += -Wextra
+CXXFLAGS += -Werror
+CXXFLAGS += -O3
+
+SAN ?= 1
+ifeq ($(SAN),1)
+FLAGS = -fsanitize=address -fsanitize=leak -fsanitize=undefined
+endif
+
+run: gen-ir
+	./$^
+
+gen-ir: gen-ir.o
+	$(CXX) -o $@ $^ $(LIBS) $(FLAGS)
+
+%.o: %.cc
+	$(CXX) -o $@ -c $^ $(CXXFLAGS) $(FLAGS)
+
+fmt:
+	clang-format -i *.cc
+
+clean:
+	$(RM) gen-ir *.o
diff --git a/content/2022-06-18-libclang-c-to-llvm-ir/gen-ir.cc b/content/2022-06-18-libclang-c-to-llvm-ir/gen-ir.cc
new file mode 100644
index 0000000..a2055e4
--- /dev/null
+++ b/content/2022-06-18-libclang-c-to-llvm-ir/gen-ir.cc
@@ -0,0 +1,97 @@
+#include <clang/Basic/DiagnosticOptions.h>
+#include <clang/CodeGen/CodeGenAction.h>
+#include <clang/Frontend/CompilerInstance.h>
+#include <clang/Frontend/TextDiagnosticPrinter.h>
+#include <clang/Lex/PreprocessorOptions.h>
+
+#include <llvm/IR/Module.h>
+#include <llvm/Support/Host.h>
+#include <llvm/Support/TargetSelect.h>
+
+using clang::CompilerInstance;
+using clang::CompilerInvocation;
+using clang::DiagnosticConsumer;
+using clang::DiagnosticOptions;
+using clang::DiagnosticsEngine;
+using clang::EmitLLVMOnlyAction;
+using clang::TextDiagnosticPrinter;
+
+using llvm::ArrayRef;
+using llvm::IntrusiveRefCntPtr;
+using llvm::MemoryBuffer;
+
+int main() {
+    const char* code_fname = "jit.c";
+    const char* code_input =
+        "struct S { int a; int b; };\n"
+        "void init(struct S* s) { s->a = 42; s->b = 1337; }\n";
+
+    // Setup custom diagnostic options.
+    IntrusiveRefCntPtr<DiagnosticOptions> diag_opts(new DiagnosticOptions());
+    diag_opts->ShowColors = 1;
+
+    // Setup custom diagnostic consumer.
+    //
+    // We configure the consumer with our custom diagnostic options and set it
+    // up that diagnostic messages are printed to stderr.
+    std::unique_ptr<DiagnosticConsumer> diag_print =
+        std::make_unique<TextDiagnosticPrinter>(llvm::errs(), diag_opts.get());
+
+    // Create custom diagnostics engine.
+    //
+    // The engine will NOT take ownership of the DiagnosticConsumer object.
+    auto diag_eng = std::make_unique<DiagnosticsEngine>(
+        nullptr /* DiagnosticIDs */, diag_opts, diag_print.get(),
+        false /* own DiagnosticConsumer */);
+
+
+    // Create compiler instance.
+    CompilerInstance cc;
+
+    // Setup compiler invocation.
+    //
+    // We are only passing a single argument, which is the pseudo file name for
+    // our code `code_fname`. We will be remapping this pseudo file name to an
+    // in-memory buffer via the preprocessor options below.
+    //
+    // The CompilerInvocation is a helper class which holds the data describing
+    // a compiler invocation (eg include paths, code generation options,
+    // warning flags, ..).
+    if (!CompilerInvocation::CreateFromArgs(cc.getInvocation(),
+                                            ArrayRef<const char*>({code_fname}),
+                                            *diag_eng)) {
+        std::puts("Failed to create CompilerInvocation!");
+        return 1;
+    }
+
+    // Setup a TextDiagnosticPrinter printer with our diagnostic options to
+    // handle diagnostic messaged.
+    //
+    // The compiler will NOT take ownership of the DiagnosticConsumer object.
+    cc.createDiagnostics(diag_print.get(), false /* own DiagnosticConsumer */);
+
+    // Create in-memory readonly buffer with pointing to our C code.
+    std::unique_ptr<MemoryBuffer> code_buffer =
+        MemoryBuffer::getMemBuffer(code_input);
+    // Configure remapping from pseudo file name to in-memory code buffer
+    // code_fname -> code_buffer.
+    //
+    // Ownership of the MemoryBuffer object is moved, except we would set
+    // `RetainRemappedFileBuffers = 1` in the PreprocessorOptions.
+    cc.getPreprocessorOpts().addRemappedFile(code_fname, code_buffer.release());
+
+    // Create action to generate LLVM IR.
+    EmitLLVMOnlyAction action;
+    // Run action against our compiler instance.
+    if (!cc.ExecuteAction(action)) {
+        std::puts("Failed to run EmitLLVMOnlyAction!");
+        return 1;
+    }
+
+    // Take generated LLVM IR module and print to stdout.
+    if (auto mod = action.takeModule()) {
+        mod->print(llvm::outs(), nullptr);
+    }
+
+    return 0;
+}