Add intermediate assembler representation

Rakefile

@@ -17,18 +17,25 @@ directory 'build'
 CLEAN.include 'build'
 CLEAN.include '.dub'
 
-rule(/build\/tests\/[^\/\.]+$/ => ->(file) { test_for_out(file) }) do |t|
+rule(/build\/tests\/[^\/\.]+$/ => ->(file) { test_for_out(file, '.o') }) do |t|
-  sh 'gcc', '-o', t.name, "#{t.name}.o"
+  sh 'ld.gold', '-L/usr/lib64',
+    '--dynamic-linker', '/lib64/ld-linux-x86-64.so.2',
+    '-o', t.name,
+    '/usr/lib64/crt1.o', '/usr/lib64/crti.o', '-lc', t.source, '/usr/lib64/crtn.o'
 end
 
-rule(/build\/asm\/[^\/\.]+$/ => ->(file) { test_for_object(file) }) do |t|
+rule(/build\/asm\/[^\/\.]+$/ => ->(file) { test_for_out(file, '.s') }) do |t|
-  Pathname.new(t.name).dirname.mkpath
+  sh 'gcc', '-x', 'assembler', '-o', t.name,  t.source
-  Open3.pipeline [BINARY, t.source], ['gcc', '-x', 'assembler', '-o', t.name,  '-']
 end
 
 rule(/build\/tests\/.+\.o$/ => ->(file) { test_for_object(file) }) do |t|
   Pathname.new(t.name).dirname.mkpath
-  sh BINARY, t.source
+  sh BINARY, '-o', t.name, t.source
+end
+
+rule(/build\/asm\/.+\.s$/ => ->(file) { test_for_object(file) }) do |t|
+  Pathname.new(t.name).dirname.mkpath
+  sh BINARY, '-s', '-o', t.name, t.source
 end
 
 file BINARY => SOURCES do |t|
@@ -70,9 +77,9 @@ def test_for_object(out_file)
   [test_source, BINARY]
 end
 
-def test_for_out(out_file)
+def test_for_out(out_file, extension)
   Pathname
     .new(out_file)
-    .sub_ext('.o')
+    .sub_ext(extension)
     .to_path
 end

source/elna/arguments.d

@@ -13,6 +13,7 @@ struct ArgumentError
     {
         expectedOutputFile,
         noInput,
+        superfluousArguments,
     }
 
     private Type type_;
@@ -50,11 +51,11 @@ struct Arguments
 {
     private bool assembler_;
     private string output_;
-    private string[] inFiles_;
+    private string inFile_;
 
-    @property string[] inFiles() @nogc nothrow pure @safe
+    @property string inFile() @nogc nothrow pure @safe
     {
-        return this.inFiles_;
+        return this.inFile_;
     }
 
     /**
@@ -95,10 +96,17 @@ struct Arguments
 
         return parseArguments(arguments).match!(
             (Arguments parsed) {
-                if (parsed.inFiles.empty)
+                if (parsed.inFile is null)
                 {
                     return ReturnType(ArgumentError(ArgumentError.Type.noInput));
                 }
+                else if (!arguments.empty)
+                {
+                    return ReturnType(ArgumentError(
+                        ArgumentError.Type.superfluousArguments,
+                        arguments.front
+                    ));
+                }
                 return ReturnType(parsed);
             },
             (ArgumentError argumentError) => ReturnType(argumentError)
@@ -131,13 +139,13 @@ struct Arguments
             else if (arguments.front == "--")
             {
                 arguments.popFront;
-                parsed.inFiles_ = arguments;
+                parsed.inFile_ = arguments.front;
+                arguments.popFront;
                 break;
             }
             else if (!arguments.front.startsWith("-"))
             {
-                parsed.inFiles_ = arguments;
+                parsed.inFile_ = arguments.front;
-                break;
             }
             arguments.popFront;
         }

source/elna/generator.d

@@ -4,6 +4,9 @@ import core.stdc.stdio;
 import core.stdc.stdlib;
 import core.stdc.string;
 import elna.ir;
+import elna.extended;
+import std.sumtype;
+import std.typecons;
 import tanya.container.array;
 import tanya.container.string;
 import tanya.memory.mmappool;
@@ -452,56 +455,137 @@ private size_t pad(size_t value) @nogc
 struct Symbol
 {
     String name;
-    Array!ubyte instructions;
+    Array!Instruction instructions;
 }
 
-/*
+enum Register : ubyte
-.text
-    .globl main
-    .type main, @function
-main:
-    movl $3, %eax
-    ret
-*/
-immutable ubyte[] instructions = [
-    // Opcode of pushq is “0x50 + r”, where “r” is the register opcode.
-    // Register opcode of %rbq is 5.
-    0x50 + 5, // push% %rbp
-    0x48, 0x89, 0xe5, // movq %rsp, %rbp
-
-    0xb8, 0x03, 0x00, 0x00, 0x00, // movl $3, %eax
-
-    // Epilogue.
-    0x48, 0x89, 0xec, // movq %rbp, %rsp
-    0x58 + 5, // popq %rbp
-    0xc3, // ret
-];
-
-void writeObject(Definition ast, String outputFilename) @nogc
 {
-    auto handle = fopen(outputFilename.toStringz, "wb");
+    AX = 0,
+    CX = 1,
+    DX = 2,
+    BX = 3,
+    SP = 4,
+    BP = 5,
+    SI = 6,
+    DI = 7,
+}
 
-    if (handle is null)
+enum MOD : ubyte
+{
+    indirect = 0,
+    one = 1,
+    four = 2,
+    direct = 3
+}
+
+enum Direction : ubyte
+{
+    registerToMemory = 0,
+    memoryToRegister = 1 << 1
+}
+
+enum Size : ubyte
+{
+    eight = 0,
+    thirtyTwo = 1
+}
+
+struct Instruction
+{
+    private ushort opcode;
+    private ubyte rexPrefix;
+    private Nullable!ubyte modrmByte;
+    private SumType!(typeof(null), byte, int) operand1;
+
+    this(ushort opcode) @nogc nothrow pure @safe
     {
-        perror("writing sample");
+        this.opcode = opcode;
-        return;
-    }
-    scope (exit)
-    {
-        fclose(handle);
     }
 
-    size_t currentOffset = Elf64_Ehdr.sizeof;
+    this(ubyte opcode, Register register) @nogc nothrow pure @safe
-    Array!Elf64_Shdr sectionHeaders;
+    {
-    Array!Elf64_Sym symbolEntries;
+        this.opcode = opcode + register;
+    }
+
+    this(ubyte opcode, Direction direction, Size size = Size.eight)
+    @nogc nothrow pure @safe
+    {
+        this.opcode = opcode | direction | size;
+    }
+
+    ref Instruction addREXPrefix(bool w = true, bool r = false, bool x = false, bool b = false)
+    return @nogc nothrow pure @safe
+    {
+        this.rexPrefix = 0x40 | (w << 3) | (r << 2) | (x << 1) | b;
+        return this;
+    }
+
+    ref Instruction addMODRMByte(MOD mode, Register register, Register rm)
+    return @nogc nothrow pure @safe
+    {
+        this.modrmByte = cast(ubyte) ((mode << 6) | (register << 3) | rm);
+        return this;
+    }
+
+    ref Instruction addOperand1(int operand)
+    return @nogc nothrow pure @safe
+    {
+        this.operand1 = operand;
+        return this;
+    }
+
+    ref Instruction addOperand1(byte operand)
+    return @nogc nothrow pure @safe
+    {
+        this.operand1 = operand;
+        return this;
+    }
+}
+
+Array!ubyte binaryInstructions(ref Array!Instruction instructions)
+@nogc nothrow
+{
+    Array!ubyte binary;
+
+    foreach (ref instruction; instructions)
+    {
+        if (instruction.rexPrefix)
+        {
+            binary.insertBack((&instruction.rexPrefix)[0 .. 1]);
+        }
+        binary.insertBack((cast(ubyte*) &instruction.opcode)[0 .. 1]);
+        if (!instruction.modrmByte.isNull)
+        {
+            binary.insertBack((&instruction.modrmByte.get())[0 .. 1]);
+        }
+        instruction.operand1.match!(
+            (byte operand) {
+                binary.insertBack((cast(ubyte*) &operand)[0 .. 1]);
+            },
+            (int operand) {
+                binary.insertBack((cast(ubyte*) &operand)[0 .. int.sizeof]);
+            },
+            (typeof(null)) {
+            }
+        );
+    }
+
+    return binary;
+}
+
+Array!Symbol buildInstructions(Definition ast) @nogc
+{
+    Array!Instruction instructions;
 
     // Prologue
-    Array!ubyte asmTemplate = Array!ubyte(cast(ubyte[]) [
+    // Opcode of pushq is “0x50 + r”, where “r” is the register opcode.
-        // Opcode of pushq is “0x50 + r”, where “r” is the register opcode.
+    // Register opcode of %rbq is 5.
-        // Register opcode of %rbq is 5.
+    instructions.insertBack(Instruction(0x50, Register.BP)); // pushq %rbp
-        0x50 + 5, // pushq %rbp
+    instructions.insertBack( // movq %rsp, %rbp
-        0x48, 0x89, 0xe5, // movq %rsp, %rbp
+        Instruction(0x89)
-    ]);
+            .addREXPrefix()
+            .addMODRMByte(MOD.direct, Register.SP, Register.BP)
+    );
     int i = 1;
     foreach (statement; ast.statements[])
     {
@@ -509,49 +593,79 @@ void writeObject(Definition ast, String outputFilename) @nogc
         {
             // Opcode of mov is “0xb8 + r”, where “r” is the register opcode.
             // Register opcode of %eax is 0.
-            asmTemplate.insertBack(cast(ubyte) 0xb8); // movl $x, %eax; where $x is a number.
+            instructions.insertBack(
-            asmTemplate.insertBack((cast(ubyte*) &(cast(Number) statement.subroutine.lhs).value)[0 .. int.sizeof]);
+                Instruction(0xb8, Register.AX) // movl $x, %eax; where $x is a number.
+                    .addOperand1((cast(Number) statement.subroutine.lhs).value)
+            );
         }
         else if ((cast(Variable) statement.subroutine.lhs) !is null)
         {
-            // movl -x(%rbp), %ebx; where x is a number.
+            // movl -x(%rbp), %eax; where x is a number.
-            asmTemplate.insertBack(cast(ubyte[]) [0x8b, 0x45]);
+            instructions.insertBack(
-            const disposition = (cast(Variable) statement.subroutine.lhs).counter * (-4);
+                Instruction(0x89, Direction.memoryToRegister, Size.thirtyTwo)
-            asmTemplate.insertBack((cast(ubyte*) &disposition)[0 .. 1]);
+                    .addMODRMByte(MOD.one, Register.AX, Register.BP)
+            );
         }
         if ((cast(Number) statement.subroutine.rhs) !is null)
         {
             // Opcode of mov is “0xb8 + r”, where “r” is the register opcode.
             // Register opcode of %ebx is 3.
-            asmTemplate.insertBack(cast(ubyte) 0xbb); // movl $x, %ebx; where $x is a number.
+            instructions.insertBack(
-            asmTemplate.insertBack((cast(ubyte*) &(cast(Number) statement.subroutine.rhs).value)[0 .. int.sizeof]);
+                Instruction(0xb8, Register.BX) // movl $x, %ebx; where $x is a number.
+                    .addOperand1((cast(Number) statement.subroutine.rhs).value)
+            );
         }
         else if ((cast(Variable) statement.subroutine.rhs) !is null)
         {
             // movl -x(%rbp), %ebx; where x is a number.
-            asmTemplate.insertBack(cast(ubyte[]) [0x8b, 0x5d]);
+            instructions.insertBack(
-            const disposition = (cast(Variable) statement.subroutine.rhs).counter * (-4);
+                Instruction(0x89, Direction.memoryToRegister, Size.thirtyTwo)
-            asmTemplate.insertBack((cast(ubyte*) &disposition)[0 .. 1]);
+                    .addMODRMByte(MOD.one, Register.BX, Register.BP)
+                    .addOperand1(cast(byte) ((cast(Variable) statement.subroutine.rhs).counter * (-4)))
+            );
         }
         // Calculate the result and assign it to a variable on the stack.
-        asmTemplate.insertBack(cast(ubyte[]) [0x01, 0xd8]); // add %ebx, %eax
+        instructions.insertBack(
+            Instruction(0x00, Direction.registerToMemory, Size.thirtyTwo)
+                .addMODRMByte(MOD.direct, Register.BX, Register.AX) // add %ebx, %eax
+        );
 
-        asmTemplate.insertBack(cast(ubyte[]) [0x89, 0x45]); // movl %eax, -x(%rbp); where x is a number.
+        instructions.insertBack( // movl %eax, -x(%rbp); where x is a number.
-        const disposition = i * (-4);
+            Instruction(0x89, Direction.registerToMemory, Size.thirtyTwo)
-        asmTemplate.insertBack((cast(ubyte*) &disposition)[0 .. 1]);
+                .addMODRMByte(MOD.one, Register.AX, Register.BP)
+                .addOperand1(cast(byte) (i * (-4)))
+        );
         ++i;
     }
     // Epilogue.
-    asmTemplate.insertBack(cast(ubyte[]) [
+    instructions.insertBack( // movq %rbp, %rsp
-        0x48, 0x89, 0xec, // movq %rbp, %rsp
+        Instruction(0x89)
-        0x58 + 5, // popq %rbp
+            .addREXPrefix()
-        0xc3, // ret
+            .addMODRMByte(MOD.direct, Register.BP, Register.SP)
-    ]);
+    );
+    instructions.insertBack(Instruction(0x58, Register.BP)); // popq %rbp
+    instructions.insertBack(Instruction(0xc3)); // ret
 
-    Symbol[1] symbols = [Symbol(String("main"), asmTemplate)];
+    return typeof(return)([Symbol(String("main"), instructions)]);
+}
 
-    sectionHeaders.insertBack(makeInitialHeader());
+void writeObject(Definition ast, String outputFilename) @nogc
-    symbolEntries.insertBack(makeInitialSymTable());
+{
+    auto handle = File.open(outputFilename.toStringz, BitFlags!(File.Mode)(File.Mode.truncate));
+
+    if (!handle.valid)
+    {
+        perror("writing sample");
+        return;
+    }
+
+    size_t currentOffset = Elf64_Ehdr.sizeof;
+    auto symbols = buildInstructions(ast);
+
+    Array!Elf64_Shdr sectionHeaders = [makeInitialHeader()];
+    Array!Elf64_Sym symbolEntries = [makeInitialSymTable()];
+    Array!ubyte instructionSection;
+    ubyte[8] padding = 0;
 
     String stringTable = String("\0");
     foreach (symbol; symbols[])
@@ -559,14 +673,20 @@ void writeObject(Definition ast, String outputFilename) @nogc
         stringTable.insertBack(symbol.name[]);
         stringTable.insertBack('\0');
 
-        sectionHeaders.insertBack(makeTextHeader(currentOffset, symbol.instructions.length));
+        auto code = binaryInstructions(symbol.instructions);
-        currentOffset = pad(currentOffset + symbol.instructions.length);
+        sectionHeaders.insertBack(makeTextHeader(currentOffset, code.length));
 
         symbolEntries.insertBack(makeMainSymTable(cast(Elf64_Half) (sectionHeaders.length - 1)));
+
+        immutable size_t instructionsLength = pad(code.length);
+        instructionSection.insertBack(code[]);
+        instructionSection.insertBack(padding[0 .. instructionsLength - code.length]);
+
+        currentOffset += instructionsLength;
     }
 
     const symbolTableSize = (symbols.length + 1) * Elf64_Sym.sizeof;
-    sectionHeaders.insertBack(makeSymtableHeader(currentOffset, symbolTableSize, symbols.length));
+    sectionHeaders.insertBack(makeSymtableHeader(currentOffset, symbolTableSize, cast(uint) symbols.length));
     currentOffset += symbolTableSize;
 
     sectionHeaders.insertBack(makeStringHeader(0x09, currentOffset, stringTable.length));
@@ -577,38 +697,30 @@ void writeObject(Definition ast, String outputFilename) @nogc
 
     auto fileHeader = makeFileHeader(currentOffset, 5, 4);
 
-    version (none)
+    handle.write((cast(ubyte*) &fileHeader)[0 .. Elf64_Ehdr.sizeof]);
-    {
+    handle.write(instructionSection.get);
-        printf("%.2x\n", cast(uint) currentOffset);
+    handle.write((cast(ubyte*) symbolEntries.get.ptr)[0 .. Elf64_Sym.sizeof * symbolEntries.length]);
-    }
-    ubyte[8] padding = 0;
-    size_t codeLength = stringTable.length + sectionStringTable.length;
 
-    fwrite(&fileHeader, 8, Elf64_Ehdr.sizeof / 8, handle);
+    immutable size_t codeLength = stringTable.length + sectionStringTable.length;
-    foreach (symbol; symbols[])
+    handle.write(cast(ubyte[]) stringTable.get);
-    {
+    handle.write(cast(ubyte[]) sectionStringTable);
-        immutable size_t instructionsLength = pad(symbol.instructions.length);
+    handle.write(padding[0 .. pad(codeLength) - codeLength]);
-        fwrite(symbol.instructions.get.ptr, 1, symbol.instructions.length, handle);
+
-        fwrite(padding.ptr, 1, instructionsLength - symbol.instructions.length, handle);
+    handle.write((cast(ubyte*) sectionHeaders.get.ptr)[0 .. Elf64_Shdr.sizeof * sectionHeaders.length]);
-        codeLength += instructionsLength;
-    }
-    fwrite(symbolEntries.get.ptr, Elf64_Sym.sizeof, symbolEntries.length, handle);
-    fwrite(stringTable.get.ptr, 1, stringTable.length, handle);
-    fwrite(sectionStringTable.ptr, 1, sectionStringTable.length, handle);
-    fwrite(padding.ptr, pad(codeLength) - codeLength, 1, handle);
-    fwrite(sectionHeaders.get.ptr, Elf64_Shdr.sizeof, sectionHeaders.length, handle);
 }
 
-String generate(Definition ast) @nogc
+void generate(Definition ast, String outputFilename) @nogc
 {
+    auto asmTemplate = File.open(outputFilename.toStringz, BitFlags!(File.Mode)(File.Mode.truncate));
+
     // Prologue
-    String asmTemplate = ".text
+    asmTemplate.write(cast(const(ubyte)[]) ".text
     .globl main
     .type main, @function
 main:
     pushq %rbp
     movq %rsp, %rbp
-";
+");
 
     /* Allocate space on the stack for local variables.
     asmTemplate.insertBack("    sub $");
@@ -620,41 +732,39 @@ main:
     {
         if ((cast(Number) statement.subroutine.lhs) !is null)
         {
-            asmTemplate.insertBack("    movl $");
+            asmTemplate.write(cast(const(ubyte)[]) "    movl $");
-            asmTemplate.insertBack(format!"{}"((cast(Number) statement.subroutine.lhs).value)[]);
+            asmTemplate.write(cast(ubyte[]) format!"{}"((cast(Number) statement.subroutine.lhs).value).get);
-            asmTemplate.insertBack(", %eax\n");
+            asmTemplate.write(cast(const(ubyte)[]) ", %eax\n");
         }
         else if ((cast(Variable) statement.subroutine.lhs) !is null)
         {
-            asmTemplate.insertBack("    movl -");
+            asmTemplate.write(cast(const(ubyte)[]) "    movl -");
-            asmTemplate.insertBack(format!"{}"((cast(Variable) statement.subroutine.lhs).counter * 4)[]);
+            asmTemplate.write(cast(ubyte[]) format!"{}"((cast(Variable) statement.subroutine.lhs).counter * 4).get);
-            asmTemplate.insertBack("(%rbp), %eax\n");
+            asmTemplate.write(cast(const(ubyte)[]) "(%rbp), %eax\n");
         }
         if ((cast(Number) statement.subroutine.rhs) !is null)
         {
-            asmTemplate.insertBack("    movl $");
+            asmTemplate.write(cast(const(ubyte)[]) "    movl $");
-            asmTemplate.insertBack(format!"{}"((cast(Number) statement.subroutine.rhs).value)[]);
+            asmTemplate.write(cast(ubyte[]) format!"{}"((cast(Number) statement.subroutine.rhs).value).get);
-            asmTemplate.insertBack(", %ebx\n");
+            asmTemplate.write(cast(const(ubyte)[]) ", %ebx\n");
         }
         else if ((cast(Variable) statement.subroutine.rhs) !is null)
         {
-            asmTemplate.insertBack("    movl -");
+            asmTemplate.write(cast(const(ubyte)[]) "    movl -");
-            asmTemplate.insertBack(format!"{}"((cast(Variable) statement.subroutine.rhs).counter * 4)[]);
+            asmTemplate.write(cast(ubyte[]) format!"{}"((cast(Variable) statement.subroutine.rhs).counter * 4).get);
-            asmTemplate.insertBack("(%rbp), %ebx\n");
+            asmTemplate.write(cast(const(ubyte)[]) "(%rbp), %ebx\n");
         }
         // Calculate the result and assign it to a variable on the stack.
-        asmTemplate.insertBack("    add %ebx, %eax\n");
+        asmTemplate.write(cast(const(ubyte)[]) "    add %ebx, %eax\n");
-        asmTemplate.insertBack("    movl %eax, -");
+        asmTemplate.write(cast(const(ubyte)[]) "    movl %eax, -");
-        asmTemplate.insertBack(format!"{}"(i * 4)[]);
+        asmTemplate.write(cast(ubyte[]) format!"{}"(i * 4).get);
-        asmTemplate.insertBack("(%rbp)\n");
+        asmTemplate.write(cast(const(ubyte)[]) "(%rbp)\n");
         ++i;
     }
 
     // Epilogue.
-    asmTemplate.insertBack("    movq %rbp, %rsp
+    asmTemplate.write(cast(const(ubyte)[]) "    movq %rbp, %rsp
     popq %rbp
     ret
 ");
-
-    return asmTemplate;
 }

source/main.d

@@ -6,6 +6,9 @@ import elna.parser;
 import elna.generator;
 import elna.ir;
 import elna.extended;
+import elna.arguments;
+import std.algorithm;
+import std.range;
 import std.sumtype;
 import std.typecons;
 import tanya.container.array;
@@ -32,37 +35,51 @@ int main(string[] args)
 {
     defaultAllocator = MmapPool.instance;
 
-    if (args.length < 2)
+    return Arguments.parse(args).match!(
-    {
+        (ArgumentError argumentError) => 4,
-        return 4;
+        (Arguments arguments) {
-    }
+            auto sourceText = readSource(arguments.inFile);
-    auto sourceText = readSource(args[1]);
+            if (sourceText.isNull)
-    if (sourceText.isNull)
+            {
-    {
+                return 3;
-        return 3;
+            }
-    }
+            auto tokens = lex(sourceText.get.get);
-    auto tokens = lex(sourceText.get.get);
+            if (tokens.length == 0)
-    if (tokens.length == 0)
+            {
-    {
+                printf("Lexical analysis failed.\n");
-        printf("Lexical analysis failed.\n");
+                return 1;
-        return 1;
+            }
-    }
+            auto ast = parse(tokens);
-    auto ast = parse(tokens);
+            if (!ast.valid)
-    if (!ast.valid)
+            {
-    {
+                auto compileError = ast.error.get;
-        auto compileError = ast.error.get;
+                printf("%lu:%lu: %s\n", compileError.line, compileError.column, compileError.message.ptr);
-        printf("%lu:%lu: %s\n", compileError.line, compileError.column, compileError.message.ptr);
+                return 2;
-        return 2;
+            }
-    }
+            auto ir = transform(ast.result);
-    auto ir = transform(ast.result);
 
-    String outputFilename = String("build/");
+            String outputFilename;
-    outputFilename.insertBack(args[1][0 .. $ - 4]);
+            if (arguments.output is null)
-    outputFilename.insertBack("o");
+            {
-    writeObject(ir, outputFilename);
+                auto slashIndex = max(0, arguments.inFile.retro.countUntil('/'));
 
-    auto code = generate(ir);
+                outputFilename.insertBack(arguments.inFile[$ - slashIndex .. $ - 4]);
-    printf("%s", code.toStringz());
+                outputFilename.insertBack(arguments.assembler ? "s" : "o");
+            }
+            else
+            {
+                outputFilename = String(arguments.output);
+            }
+            if (arguments.assembler)
+            {
+                generate(ir, outputFilename);
+            }
+            else
+            {
+                writeObject(ir, outputFilename);
+            }
 
-    return 0;
+            return 0;
+        }
+    );
 }