works, but segfaults

backup-compiler/.clang-format

@@ -0,0 +1,14 @@
+Language: Cpp
+BasedOnStyle: WebKit
+IndentWidth: 4
+ColumnLimit: 80
+IndentCaseLabels: true
+InsertNewlineAtEOF: true
+AllowShortFunctionsOnASingleLine: None
+
+BinPackArguments: false
+AllowAllArgumentsOnNextLine: true
+
+BinPackParameters: false
+AllowAllParametersOfDeclarationOnNextLine: true
+

backup-compiler/.gitignore

@@ -0,0 +1,4 @@
+out.nasm
+*.o
+out
+

backup-compiler/Makefile

@@ -0,0 +1,18 @@
+
+all: out
+
+out: entry.o out.o lib.o
+	gcc $^ -o $@ -no-pie
+
+%.o: %.c
+	gcc -c -o $@ -std=c17 -Wall -Wextra -Wpedantic -pedantic -pedantic-errors $^
+
+%.o: %.nasm
+	nasm -f elf64 $< -o $@
+
+out.nasm: program.sbl
+	deno run --allow-read --allow-write --check main.ts $<
+
+clean:
+	rm -rf out.asm out.o lib.o entry.o out
+

backup-compiler/asm_gen.ts

@@ -25,17 +25,28 @@ export class AsmGen {
         }
         this.writeln(`section .text`);
         for (const fn of this.lir.fns) {
-            this.writeln(`align 8`);
-            this.writeln(`global ${fn.label}:`);
-            this.writeln(`${fn.label}:`);
-
-            this.generateFnBody(fn);
+            this.generateFn(fn);
         }
 
+        this.writeln(`; vim: syntax=nasm commentstring=;\\ %s`);
+        this.writeln("");
         return this.writer.finalize();
     }
 
-    private generateFnBody(fn: lir.Fn) {
+    private generateFn(fn: lir.Fn) {
+        const query = this.queryCFunction(fn);
+        if (query.found) {
+            const { label, args } = query;
+            this.generateCFunctionBody(fn, label, args);
+            return;
+        }
+
+        this.generateFnBody(fn);
+    }
+
+    private queryCFunction(
+        fn: lir.Fn,
+    ): { found: false } | { found: true; label: string; args: number } {
         const stmtKind = fn.mir.stmt.kind;
         if (stmtKind.tag !== "fn") {
             throw new Error();
@@ -45,12 +56,41 @@ export class AsmGen {
             if (!arg || arg.kind.tag !== "string") {
                 throw new Error("incorrect args for attribute");
             }
-            const label = arg.kind.val;
-            this.generateCFunctionBody(label, fn.mir.paramLocals.size);
-            return;
+            return {
+                found: true,
+                label: arg.kind.val,
+                args: fn.mir.paramLocals.size,
+            };
+        }
+        return { found: false };
+    }
+
+    private generateCFunctionBody(fn: lir.Fn, label: string, args: number) {
+        this.writeln(`extern ${label}`);
+        this.writeln(`global ${fn.label}`);
+        this.writeln(`${fn.label}:`);
+
+        this.writeIns(`push rbp`);
+        this.writeIns(`mov rbp, rsp`);
+        for (let i = 0; i < args; ++i) {
+            this.writeIns(`mov rax, ${this.relative((i + 2) * 8)}`);
+            this.writeIns(`push rax`);
         }
 
-        this.writeIns(`push r12`);
+        const argRegs = ["rdi", "rsi", "rdx", "rcx", "r8", "r9"];
+        for (const reg of argRegs.slice(0, args + 1)) {
+            this.writeIns(`pop ${reg}`);
+        }
+
+        this.writeIns(`call ${label}`);
+        this.writeIns(`mov rsp, rbp`);
+        this.writeIns(`pop rbp`);
+        this.writeIns(`ret`);
+    }
+
+    private generateFnBody(fn: lir.Fn) {
+        this.writeln(`global ${fn.label}:`);
+        this.writeln(`${fn.label}:`);
         this.writeIns(`push rbp`);
         this.writeIns(`mov rbp, rsp`);
         this.writeIns(`sub rsp, ${fn.frameSize}`);
@@ -64,30 +104,11 @@ export class AsmGen {
         }
 
         this.writeln(`.exit:`);
+        const returnLocalOffset = fn.localOffsets.get(fn.mir.returnLocal.id)!;
+        this.writeIns(`mov rax, QWORD ${this.relative(returnLocalOffset)}`);
         this.writeIns(`mov rsp, rbp`);
         this.writeIns(`pop rbp`);
-        this.writeIns(`pop r12`);
-    }
-
-    private generateCFunctionBody(label: string, args: number) {
-        const argRegs = ["rdi", "rsi", "rdx", "rcx", "r8", "r9"];
-        const returnReg = "rax";
-        if (args > argRegs.length) {
-            throw new Error(
-                `arg count (${args}) > ${argRegs.length} not supported`,
-            );
-        }
-        this.writeIns(`push ${returnReg}`);
-        for (const reg of argRegs.slice(0, args + 1)) {
-            this.writeIns(`push ${reg}`);
-        }
-        this.writeIns(`call ${label}`);
-        this.writeIns(`mov r12, rax`);
-        for (const reg of argRegs.slice(0, args + 1).toReversed()) {
-            this.writeIns(`push ${reg}`);
-        }
-        this.writeIns(`pop ${returnReg}`);
-        this.writeIns(`push r12`);
+        this.writeIns(`ret`);
     }
 
     private generateIns(ins: lir.Ins) {
@@ -115,16 +136,27 @@ export class AsmGen {
                 this.writeIns(`pop ${r(ins.reg)}`);
                 return;
             case "load":
-                this.writeIns(`mov QWORD [rbp${ins.offset}], ${r(ins.reg)}`);
+                this.writeIns(
+                    `mov ${r(ins.reg)}, QWORD ${this.relative(ins.offset)}`,
+                );
                 return;
-            case "store":
-                this.writeIns(`mov ${r(ins.reg)}, QWORD [rbp${ins.offset}]`);
+            case "store_reg":
+                this.writeIns(
+                    `mov QWORD ${this.relative(ins.offset)}, ${r(ins.reg)}`,
+                );
+                return;
+            case "store_imm":
+                this.writeIns(
+                    `mov QWORD ${this.relative(ins.offset)}, ${ins.val}`,
+                );
                 return;
             case "call_reg":
-                this.generateCall(r(ins.reg), ins.args);
+                this.writeIns(`call ${r(ins.reg)}`);
+                this.writeIns(`push rax`);
                 return;
-            case "call_fn":
-                this.generateCall(ins.fn.label, ins.args);
+            case "call_imm":
+                this.writeIns(`call ${ins.fn.label}`);
+                this.writeIns(`push rax`);
                 return;
             case "jmp":
                 this.writeIns(`jmp .L${ins.target}`);
@@ -153,27 +185,7 @@ export class AsmGen {
                 this.kill(ins.reg);
                 return;
         }
-    }
-
-    private generateCall(value: string, args: number) {
-        const argRegs = ["rdi", "rsi", "rdx", "rcx", "r8", "r9"];
-        const returnReg = "rax";
-        if (args > argRegs.length) {
-            throw new Error(
-                `arg count (${args}) > ${argRegs.length} not supported`,
-            );
-        }
-        this.writeIns(`push ${returnReg}`);
-        for (const reg of argRegs.slice(0, args + 1)) {
-            this.writeIns(`push ${reg}`);
-        }
-        this.writeIns(`call ${value}`);
-        this.writeIns(`mov r12, rax`);
-        for (const reg of argRegs.slice(0, args + 1).toReversed()) {
-            this.writeIns(`push ${reg}`);
-        }
-        this.writeIns(`pop ${returnReg}`);
-        this.writeIns(`push r12`);
+        const _: never = ins;
     }
 
     private reg(reg: lir.Reg): string {
@@ -198,7 +210,6 @@ export class AsmGen {
 
     private allocReg(reg: lir.Reg) {
         if (!this.liveRegs.has(reg)) {
-            this.liveRegs.add(reg);
             const regSel = [
                 "rax",
                 "rdi",
@@ -213,10 +224,15 @@ export class AsmGen {
             if (!regSel) {
                 throw new Error("ran out of registers");
             }
+            this.liveRegs.add(reg);
             this.regSelect.set(reg, regSel);
         }
     }
 
+    private relative(offset: number): string {
+        return `[rbp${offset >= 0 ? `+${offset}` : `${offset}`}]`;
+    }
+
     private kill(reg: lir.Reg) {
         this.liveRegs.delete(reg);
         this.regSelect.delete(reg);

backup-compiler/entry.c

@@ -0,0 +1,8 @@
+#include <stdint.h>
+
+extern int32_t sbc__main(void);
+
+int main(void)
+{
+    sbc__main();
+}

backup-compiler/lib.c

@@ -0,0 +1,8 @@
+#include <stdint.h>
+#include <stdio.h>
+
+int64_t print_int(int64_t value)
+{
+    printf("%ld\n", value);
+    return 0;
+}

backup-compiler/lir.ts

@@ -11,6 +11,7 @@ export type Fn = {
     mir: mir.Fn;
     lines: Line[];
     frameSize: number;
+    localOffsets: Map<number, number>;
 };
 
 export type Line = {
@@ -27,9 +28,10 @@ export type Ins =
     | { tag: "push"; reg: Reg }
     | { tag: "pop"; reg: Reg }
     | { tag: "load"; reg: Reg; offset: number }
-    | { tag: "store"; offset: number; reg: Reg }
+    | { tag: "store_reg"; offset: number; reg: Reg }
+    | { tag: "store_imm"; offset: number; val: number }
     | { tag: "call_reg"; reg: Reg; args: number }
-    | { tag: "call_fn"; fn: Fn; args: number }
+    | { tag: "call_imm"; fn: Fn; args: number }
     | { tag: "jmp"; target: Label }
     | { tag: "jnz_reg"; reg: Reg; target: Label }
     | { tag: "ret" }
@@ -52,7 +54,7 @@ export class ProgramStringifyer {
                         .map((label) =>
                             `${
                                 label.labels
-                                    .map((label) => `.${label}:\n`)
+                                    .map((label) => `.L${label}:\n`)
                                     .join()
                             }    ${this.ins(label.ins)}\n`
                         )
@@ -80,11 +82,13 @@ export class ProgramStringifyer {
                 return `pop %${ins.reg}`;
             case "load":
                 return `load %${ins.reg}, ${ins.offset}`;
-            case "store":
-                return `store ${ins.offset}, %${ins.reg}`;
+            case "store_reg":
+                return `store_reg ${ins.offset}, %${ins.reg}`;
+            case "store_imm":
+                return `store_val ${ins.offset}, ${ins.val}`;
             case "call_reg":
                 return `call_reg %${ins.reg}, ${ins.args}`;
-            case "call_fn":
+            case "call_imm":
                 return `call_fn ${ins.fn.label}, ${ins.args}`;
             case "jmp":
                 return `jmp .b${ins.target}`;

backup-compiler/lir_gen.ts

@@ -31,7 +31,14 @@ export class LirGen {
             const mir = this.mirGen.fnMir(stmt, stmt.kind);
             const id = this.fnIds++;
             const label = `sbc__${stmt.kind.ident}`;
-            const fn: Fn = { id, label, mir, lines: [], frameSize: 0 };
+            const fn: Fn = {
+                id,
+                label,
+                mir,
+                lines: [],
+                frameSize: 0,
+                localOffsets: new Map(),
+            };
             this.fns.set(id, fn);
             this.stmtFns.set(stmt.id, fn);
         }
@@ -69,8 +76,6 @@ class FnGen {
 
     private currentLabels: Label[] = [];
 
-    private nextOffset = -8;
-    private frameSize = 0;
     private localOffsets = new Map<number, number>();
 
     public constructor(
@@ -84,12 +89,33 @@ class FnGen {
             const label = this.labelIds++;
             this.blockLabels.set(block.id, label);
         }
-        for (const local of this.fn.mir.locals) {
-            this.localOffsets.set(local.id, this.nextOffset);
-            this.nextOffset -= 8;
-            this.frameSize += 8;
+
+        let currentOffset = 8 + this.fn.mir.paramLocals.size * 8;
+        let frameSize = 0;
+
+        for (const local of this.fn.mir.paramLocals.values()) {
+            this.localOffsets.set(local.id, currentOffset);
+            currentOffset -= 8;
         }
-        this.fn.frameSize = this.frameSize;
+        // return address
+        currentOffset -= 8;
+        // old rbp
+        currentOffset -= 8;
+        // return value
+        this.localOffsets.set(this.fn.mir.returnLocal.id, currentOffset);
+        currentOffset -= 8;
+        frameSize += 8;
+        for (const local of this.fn.mir.locals) {
+            if (this.localOffsets.has(local.id)) {
+                continue;
+            }
+            this.localOffsets.set(local.id, currentOffset);
+            currentOffset -= 8;
+            frameSize += 8;
+        }
+        this.fn.frameSize = frameSize;
+        this.fn.localOffsets = this.localOffsets;
+
         for (const block of this.fn.mir.blocks) {
             this.currentLabels.push(this.blockLabels.get(block.id)!);
             for (const stmt of block.stmts) {
@@ -158,7 +184,7 @@ class FnGen {
                 const reg = this.reg();
                 const offset = this.localOffsets.get(k.local.id)!;
                 this.pushIns({ tag: "pop", reg });
-                this.pushIns({ tag: "store", offset, reg });
+                this.pushIns({ tag: "store_reg", offset, reg });
                 this.pushIns({ tag: "kill", reg });
                 return;
             }
@@ -175,8 +201,8 @@ class FnGen {
             case "mul": {
                 const dst = this.reg();
                 const src = this.reg();
-                this.pushIns({ tag: "pop", reg: src });
                 this.pushIns({ tag: "pop", reg: dst });
+                this.pushIns({ tag: "pop", reg: src });
                 this.pushIns({ tag: k.tag, dst, src });
                 this.pushIns({ tag: "push", reg: dst });
                 this.pushIns({ tag: "kill", reg: src });

backup-compiler/lir_optimize.ts

@@ -0,0 +1,185 @@
+import {
+    Fn,
+    Ins,
+    Label,
+    Line,
+    Program,
+    ProgramStringifyer,
+    Reg,
+} from "./lir.ts";
+
+export function lirOptimize(program: Program) {
+    console.log("=== BEFORE OPTIMIZATION ===");
+    console.log(new ProgramStringifyer(program).stringify());
+    for (const fn of program.fns) {
+        eliminatePushPop(fn);
+        eliminateMovFnCall(fn);
+        eliminateMovIntStoreReg(fn);
+    }
+    console.log("=== AFTER OPTIMIZATION ===");
+    console.log(new ProgramStringifyer(program).stringify());
+}
+
+function eliminatePushPop(fn: Fn) {
+    const candidates: number[] = [];
+
+    for (let i = 0; i < fn.lines.length - 2; ++i) {
+        const [push, kill, pop] = fn.lines.slice(i);
+        if (
+            push.ins.tag === "push" &&
+            kill.ins.tag === "kill" &&
+            pop.ins.tag === "pop" &&
+            kill.labels.length === 0 &&
+            pop.labels.length === 0 &&
+            push.ins.reg === kill.ins.reg
+        ) {
+            candidates.push(i);
+        }
+    }
+
+    for (const i of candidates.toReversed()) {
+        if (i + 3 >= fn.lines.length) {
+            fn.lines[i + 3].labels.push(...fn.lines[i].labels);
+        }
+        const [push, kill, pop] = fn.lines.slice(i);
+        if (
+            !(
+                push.ins.tag === "push" &&
+                kill.ins.tag === "kill" &&
+                pop.ins.tag === "pop"
+            )
+        ) {
+            throw new Error();
+        }
+        const toRemove = pop.ins.reg;
+        const replacement = push.ins.reg;
+        fn.lines.splice(i, 3);
+        replaceReg(fn, toRemove, replacement);
+    }
+}
+
+function eliminateMovFnCall(fn: Fn) {
+    const candidates: number[] = [];
+
+    for (let i = 0; i < fn.lines.length - 2; ++i) {
+        const [movFn, callReg, kill] = fn.lines.slice(i);
+        if (
+            movFn.ins.tag === "mov_fn" &&
+            callReg.ins.tag === "call_reg" &&
+            kill.ins.tag === "kill" &&
+            callReg.labels.length === 0 &&
+            kill.labels.length === 0 &&
+            movFn.ins.reg === callReg.ins.reg &&
+            movFn.ins.reg === kill.ins.reg
+        ) {
+            candidates.push(i);
+        }
+    }
+
+    for (const i of candidates.toReversed()) {
+        const [movFn, callReg, kill] = fn.lines.slice(i);
+        if (
+            !(
+                movFn.ins.tag === "mov_fn" &&
+                callReg.ins.tag === "call_reg" &&
+                kill.ins.tag === "kill"
+            )
+        ) {
+            throw new Error();
+        }
+        const fnVal = movFn.ins.fn;
+        const args = callReg.ins.args;
+        fn.lines.splice(i + 1, 2);
+        fn.lines[i].ins = { tag: "call_imm", fn: fnVal, args };
+    }
+}
+
+function eliminateMovIntStoreReg(fn: Fn) {
+    const candidates: number[] = [];
+
+    for (let i = 0; i < fn.lines.length - 2; ++i) {
+        const [movInt, storeReg, kill] = fn.lines.slice(i);
+        if (
+            movInt.ins.tag === "mov_int" &&
+            storeReg.ins.tag === "store_reg" &&
+            kill.ins.tag === "kill" &&
+            storeReg.labels.length === 0 &&
+            kill.labels.length === 0 &&
+            movInt.ins.reg === storeReg.ins.reg &&
+            movInt.ins.reg === kill.ins.reg
+        ) {
+            candidates.push(i);
+        }
+    }
+
+    for (const i of candidates.toReversed()) {
+        const [movInt, storeReg, kill] = fn.lines.slice(i);
+        if (
+            !(
+                movInt.ins.tag === "mov_int" &&
+                storeReg.ins.tag === "store_reg" &&
+                kill.ins.tag === "kill"
+            )
+        ) {
+            throw new Error();
+        }
+        const offset = storeReg.ins.offset;
+        const val = movInt.ins.val;
+        fn.lines.splice(i + 1, 2);
+        fn.lines[i].ins = { tag: "store_imm", offset, val };
+    }
+}
+
+function replaceReg(fn: Fn, cand: Reg, replacement: Reg) {
+    const r = (reg: Reg): Reg => reg === cand ? replacement : reg;
+
+    for (const { ins } of fn.lines) {
+        switch (ins.tag) {
+            case "error":
+                break;
+            case "nop":
+                break;
+            case "mov_int":
+            case "mov_string":
+            case "mov_fn":
+                ins.reg = r(ins.reg);
+                break;
+            case "push":
+            case "pop":
+                ins.reg = r(ins.reg);
+                break;
+            case "load":
+            case "store_reg":
+                ins.reg = r(ins.reg);
+                break;
+            case "store_imm":
+                break;
+            case "call_reg":
+                ins.reg = r(ins.reg);
+                break;
+            case "call_imm":
+                break;
+            case "jmp":
+                break;
+            case "jnz_reg":
+                ins.reg = r(ins.reg);
+                break;
+            case "ret":
+                break;
+            case "lt":
+            case "eq":
+            case "add":
+            case "mul":
+                ins.dst = r(ins.dst);
+                ins.src = r(ins.src);
+                break;
+            case "kill":
+                ins.reg = r(ins.reg);
+                break;
+            default: {
+                const _: never = ins;
+            }
+        }
+    }
+}
+

backup-compiler/main.ts

@@ -5,35 +5,40 @@ import { FnStringifyer } from "./mir.ts";
 import { LirGen } from "./lir_gen.ts";
 import { ProgramStringifyer } from "./lir.ts";
 import { AsmGen } from "./asm_gen.ts";
+import { lirOptimize } from "./lir_optimize.ts";
 
 async function main() {
     const text = await Deno.readTextFile(Deno.args[0]);
 
     const ast = new Parser(text).parse();
-    console.log("=== AST ===");
-    console.log(yaml.stringify(ast));
+    // console.log("=== AST ===");
+    // console.log(yaml.stringify(ast));
 
     const re = new Resolver(ast).resolve();
     const ch = new Checker(re);
 
     const mirGen = new MirGen(re, ch);
 
-    console.log("=== MIR ===");
-    for (const stmt of ast) {
-        if (stmt.kind.tag !== "fn") {
-            throw new Error("only functions can compile top level");
-        }
-        const fnMir = mirGen.fnMir(stmt, stmt.kind);
-        console.log(new FnStringifyer(fnMir).stringify());
-    }
+    // console.log("=== MIR ===");
+    // for (const stmt of ast) {
+    //     if (stmt.kind.tag !== "fn") {
+    //         throw new Error("only functions can compile top level");
+    //     }
+    //     const fnMir = mirGen.fnMir(stmt, stmt.kind);
+    //     console.log(new FnStringifyer(fnMir).stringify());
+    // }
 
     const lir = new LirGen(ast, mirGen).generate();
-    console.log("=== LIR ===");
-    console.log(new ProgramStringifyer(lir).stringify());
+    // console.log("=== LIR ===");
+    // console.log(new ProgramStringifyer(lir).stringify());
+
+    lirOptimize(lir);
 
     const asm = new AsmGen(lir).generate();
-    console.log("=== ASM ===");
-    console.log(asm);
+    // console.log("=== ASM ===");
+    // console.log(asm);
+
+    await Deno.writeTextFile("out.nasm", asm);
 }
 
 main();

backup-compiler/mir_gen.ts

@@ -51,11 +51,12 @@ export class FnMirGen {
 
         const entry = this.block();
         const exit = this.block();
+        this.returnBlock = exit;
 
         this.currentBlock = entry;
         this.lowerBlock(this.stmtKind.body);
 
-        entry.ter = Ter({ tag: "goto", target: exit });
+        this.currentBlock.ter = Ter({ tag: "goto", target: exit });
         exit.ter = Ter({ tag: "return" });
         return {
             stmt: this.stmt,
@@ -251,6 +252,8 @@ export class FnMirGen {
                 return;
             }
             case "binary": {
+                this.lowerExpr(k.left);
+                this.lowerExpr(k.right);
                 switch (k.op) {
                     case "<":
                         this.pushStmt({ tag: "lt" });
@@ -290,4 +293,3 @@ export class FnMirGen {
         this.currentBlock.stmts.push(Stmt(kind));
     }
 }
-

backup-compiler/program.sbl

@@ -2,13 +2,15 @@
 #[c_function("print_int")]
 fn print_int(value) {}
 
-// #[c_function("println")]
-// fn println(value) {}
+fn inner(value) {
+    print_int(value);
+    return 0;    
+}
 
 fn main() {
-    // println("hello\ world");
     let a = 4;
-    print_int(a + 2);
+    inner(a + 2);
+    return a;
 }
 
 // vim: syntax=rust commentstring=//\ %s