Fuzzer & fixes (#14)

* Fuzzer part 1

* Fuzz only sensible programs.

* Fix issues found from fuzzer.

* Additional fixes after fuzzing.

* Fix mul upper.

* Display arguments going into instructions.

* Generate JSON test cases from fuzzer.

* Fix memory allocation.

* Fuzzer fixes.

* Panic on OOM

* OOM is actually a page fault?

* Add gas cost for memory.

* Read JSON files from a directory.

* Cleanup

* Fix lint.

* Fix examples.

.gitignore

@@ -1 +1,2 @@
 node_modules
+tests/*

assembly/api-generic.ts

@@ -3,7 +3,7 @@ import { INSTRUCTIONS, MISSING_INSTRUCTION } from "./instructions";
 import { Interpreter, Status } from "./interpreter";
 import { Memory, MemoryBuilder } from "./memory";
 import { Access, PAGE_SIZE } from "./memory-page";
-import { Program, decodeArguments, decodeProgram, liftBytes } from "./program";
+import { Program, decodeArguments, decodeProgram, liftBytes, resolveArguments } from "./program";
 import { NO_OF_REGISTERS, Registers } from "./registers";
 
 export class InitialPage {
@@ -62,7 +62,7 @@ export function getAssembly(p: Program): string {
     }
 
     const args = decodeArguments(iData.kind, p.code.subarray(i + 1, end));
-    const argsArray = [args.a, args.b, args.c, args.d];
+    const argsArray = args === null ? [0, 0, 0, 0] : [args.a, args.b, args.c, args.d];
     const relevantArgs = RELEVANT_ARGS[iData.kind];
     for (let i = 0; i < relevantArgs; i++) {
       v += ` ${argsArray[i]}, `;
@@ -96,9 +96,19 @@ export function runVm(input: VmInput, logs: boolean = false): VmOutput {
     if (logs) console.log(`PC = ${int.pc}`);
     if (logs) console.log(`STATUS = ${int.status}`);
     if (logs) console.log(`REGISTERS = ${registers.join(", ")}`);
-    if (logs && int.pc < u32(int.program.code.length)) {
-      const name = changetype<string>(INSTRUCTIONS[int.program.code[int.pc]].namePtr);
-      console.log(`INSTRUCTION = ${name} (${int.program.code[int.pc]})`);
+    if (logs) {
+      const instruction = int.pc < u32(int.program.code.length) ? int.program.code[int.pc] : 0;
+      const iData = instruction >= <u8>INSTRUCTIONS.length ? MISSING_INSTRUCTION : INSTRUCTIONS[instruction];
+      const name = changetype<string>(iData.namePtr);
+      console.log(`INSTRUCTION = ${name} (${instruction})`);
+      const args = resolveArguments(iData.kind, int.program.code.subarray(int.pc + 1), int.registers);
+      if (args !== null) {
+        console.log(`ARGUMENTS:
+  ${args.a} (${args.decoded.a}) = 0x${u64(args.a).toString(16)}, 
+  ${args.b} (${args.decoded.b}) = 0x${u64(args.b).toString(16)},
+  ${args.c} (${args.decoded.c}) = 0x${u64(args.c).toString(16)},
+  ${args.d} (${args.decoded.d}) = 0x${u64(args.d).toString(16)}`);
+      }
     }
 
     isOk = int.nextStep();
@@ -120,6 +130,11 @@ export function getOutputChunks(memory: Memory): InitialChunk[] {
     const pageIdx = pages[i];
     const page = memory.pages.get(pageIdx);
 
+    // skip empty pages
+    if (page.raw.page === null) {
+      continue;
+    }
+
     for (let n = 0; n < page.raw.data.length; n++) {
       const v = page.raw.data[n];
       if (v !== 0) {

assembly/arguments.ts

@@ -17,7 +17,7 @@ export enum Arguments {
 /** How many numbers in `Args` is relevant for given `Arguments`. */
 export const RELEVANT_ARGS = [<i32>0, 1, 2, 1, 2, 3, 3, 3, 2, 3, 3, 4, 3];
 /** How many bytes is required by given `Arguments`. */
-export const REQUIRED_BYTES = [<i32>0, 0, 0, 0, 1, 9, 1, 1, 1, 1, 1, 1, 2];
+export const REQUIRED_BYTES = [<i32>0, 0, 1, 0, 1, 9, 1, 1, 1, 1, 1, 2, 2];
 
 // @unmanaged
 export class Args {
@@ -39,11 +39,8 @@ function asArgs(a: u32, b: u32, c: u32, d: u32): Args {
 type ArgsDecoder = (data: Uint8Array) => Args;
 
 function twoImm(data: Uint8Array): Args {
-  if (data.length === 0) {
-    return asArgs(0, 0, 0, 0);
-  }
   const n = nibbles(data[0]);
-  const split = n.low + 1;
+  const split = <i32>Math.min(4, n.low) + 1;
   const first = decodeI32(data.subarray(1, split));
   const second = decodeI32(data.subarray(split));
   return asArgs(first, second, 0, 0);
@@ -77,16 +74,16 @@ export const DECODERS: ArgsDecoder[] = [
   },
   //DECODERS[Arguments.OneRegTwoImm] =
   (data: Uint8Array) => {
-    const first = nibbles(data[0]);
-    const split = first.hig + 1;
+    const n = nibbles(data[0]);
+    const split = <i32>Math.min(4, n.hig) + 1;
     const immA = decodeI32(data.subarray(1, split));
     const immB = decodeI32(data.subarray(split));
-    return asArgs(first.low, immA, immB, 0);
+    return asArgs(n.low, immA, immB, 0);
   },
   // DECODERS[Arguments.OneRegOneImmOneOff] =
   (data: Uint8Array) => {
     const n = nibbles(data[0]);
-    const split = n.hig + 1;
+    const split = <i32>Math.min(4, n.hig) + 1;
     const immA = decodeI32(data.subarray(1, split));
     const offs = decodeI32(data.subarray(split));
     return asArgs(n.low, immA, offs, 0);
@@ -127,7 +124,7 @@ class Nibbles {
 }
 
 // @inline
-function nibbles(byte: u8): Nibbles {
+export function nibbles(byte: u8): Nibbles {
   const low = byte & 0xf;
   const hig = byte >> 4;
   const n = new Nibbles();
@@ -138,7 +135,7 @@ function nibbles(byte: u8): Nibbles {
 
 //@inline
 function decodeI32(data: Uint8Array): u32 {
-  const len = <u32>data.length;
+  const len = <u32>Math.min(4, data.length);
   let num = 0;
   for (let i: u32 = 0; i < len; i++) {
     num |= u32(data[i]) << (i * 8);
@@ -152,6 +149,16 @@ function decodeI32(data: Uint8Array): u32 {
   return num;
 }
 
+export function encodeI32(input: i32): u8[] {
+  const data: u8[] = [];
+  let num = u32(input);
+  while (num > 0) {
+    data.push(u8(num));
+    num >>= 8;
+  }
+  return data;
+}
+
 function decodeU32(data: Uint8Array): u32 {
   let num = u32(data[0]);
   num |= u32(data[1]) << 8;

assembly/codec.ts

@@ -34,7 +34,7 @@ export class Decoder {
 
   varU32(): u32 {
     this.ensureBytes(1);
-    const v = readVarU32(this.source.subarray(this.offset));
+    const v = decodeVarU32(this.source.subarray(this.offset));
     this.offset += v.offset;
     return v.value;
   }
@@ -90,7 +90,7 @@ export class ValOffset<T> {
 }
 
 /** Read variable-length u32 and return number of bytes read. */
-export function readVarU32(data: Uint8Array): ValOffset<u32> {
+export function decodeVarU32(data: Uint8Array): ValOffset<u32> {
   const length = i32(variableLength(data[0]));
   const first = u32(data[0]);
   if (length === 0) {
@@ -110,3 +110,45 @@ export function readVarU32(data: Uint8Array): ValOffset<u32> {
 
   return new ValOffset(number, 1 + length);
 }
+
+export function encodeVarU32(v: u64): Uint8Array {
+  if (v === 0) {
+    return new Uint8Array(1);
+  }
+
+  // handle the biggest case
+  let maxEncoded = u64(2 ** (7 * 8));
+  if (v >= maxEncoded) {
+    const dest = new Uint8Array(9);
+    dest[0] = 0xff;
+    const dataView = new DataView(dest.buffer);
+    dataView.setUint64(1, v, true);
+    return dest;
+  }
+
+  // let's look for the correct range
+  let minEncoded = maxEncoded >> 7;
+  for (let l = 7; l >= 0; l -= 1) {
+    if (v >= minEncoded) {
+      const dest = new Uint8Array(l + 1);
+
+      // encode the first byte
+      const maxVal = 2 ** (8 * l);
+      const byte = 2 ** 8 - 2 ** (8 - l) + v / maxVal;
+      dest[0] = u8(byte);
+
+      // now encode the rest of bytes of len `l`
+      let rest = v % maxVal;
+      for (let i = 1; i < 1 + l; i += 1) {
+        dest[i] = u8(rest);
+        rest >>= 8;
+      }
+      return dest;
+    }
+    // move one power down
+    maxEncoded = minEncoded;
+    minEncoded >>= 7;
+  }
+
+  throw new Error(`Unhandled number encoding: ${v}`);
+}

assembly/gas.ts

@@ -1,8 +1,8 @@
 /** Gas type. */
-export type Gas = u64;
+export type Gas = i64;
 
 /** Create a new gas counter instance depending on the gas value. */
-export function gasCounter(gas: u64): GasCounter {
+export function gasCounter(gas: i64): GasCounter {
   return new GasCounterU64(gas);
 }
 
@@ -25,7 +25,7 @@ class GasCounterU64 implements GasCounter {
   constructor(private gas: Gas) {}
 
   set(g: Gas): void {
-    this.gas = <u64>g;
+    this.gas = <i64>g;
   }
 
   get(): Gas {

assembly/index.ts

@@ -1,8 +1,9 @@
-import { VmInput, getAssembly, runVm } from "./api-generic";
+import { VmInput, VmOutput, getAssembly, runVm } from "./api-generic";
 import { decodeProgram, decodeSpi, liftBytes } from "./program";
 
 export * from "./api";
 export { runVm, getAssembly } from "./api-generic";
+export { wrapAsProgram } from "./program-build";
 
 export enum InputKind {
   Generic = 0,
@@ -24,16 +25,16 @@ export function disassemble(input: u8[], kind: InputKind): string {
   return `Unknown kind: ${kind}`;
 }
 
-export function runProgram(input: u8[], kind: InputKind): void {
+export function runProgram(input: u8[], registers: u64[], kind: InputKind): VmOutput {
   if (kind === InputKind.Generic) {
     const vmInput = new VmInput();
-    vmInput.registers[7] = 9;
+    vmInput.registers = registers;
     vmInput.gas = 10_000;
     vmInput.program = input;
 
     const output = runVm(vmInput, true);
     console.log(`Finished with status: ${output.status}`);
-    return;
+    return output;
   }
 
   if (kind === InputKind.SPI) {