core/asm: separate legacy and eof instruction iterators

core/asm/asm.go

@@ -26,12 +26,13 @@ import (
 
 // Iterator for disassembled EVM instructions
 type instructionIterator struct {
-	code    []byte
-	pc      uint64
-	arg     []byte
-	op      vm.OpCode
-	error   error
-	started bool
+	code       []byte
+	pc         uint64
+	arg        []byte
+	op         vm.OpCode
+	error      error
+	started    bool
+	eofEnabled bool
 }
 
 // NewInstructionIterator creates a new instruction iterator.
@@ -41,6 +42,13 @@ func NewInstructionIterator(code []byte) *instructionIterator {
 	return it
 }
 
+// NewEOFInstructionIterator creates a new instruction iterator for EOF-code.
+func NewEOFInstructionIterator(code []byte) *instructionIterator {
+	it := NewInstructionIterator(code)
+	it.eofEnabled = true
+	return it
+}
+
 // Next returns true if there is a next instruction and moves on.
 func (it *instructionIterator) Next() bool {
 	if it.error != nil || uint64(len(it.code)) <= it.pc {
@@ -63,10 +71,23 @@ func (it *instructionIterator) Next() bool {
 		// We reached the end.
 		return false
 	}
-
 	it.op = vm.OpCode(it.code[it.pc])
-
-	if a := vm.Immediates(it.op); a > 0 {
+	var a int
+	if !it.eofEnabled { // Legacy code
+		if it.op.IsPush() {
+			a = int(it.op) - int(vm.PUSH0)
+		}
+	} else { // EOF code
+		if it.op == vm.RJUMPV {
+			// RJUMPV is unique as it has a variable sized operand. The total size is
+			// determined by the count byte which immediately follows RJUMPV.
+			maxIndex := int(it.code[it.pc+1])
+			a = (maxIndex+1)*2 + 1
+		} else {
+			a = vm.Immediates(it.op)
+		}
+	}
+	if a > 0 {
 		u := it.pc + 1 + uint64(a)
 		if uint64(len(it.code)) <= it.pc || uint64(len(it.code)) < u {
 			it.error = fmt.Errorf("incomplete instruction at %v", it.pc)
@@ -105,7 +126,6 @@ func PrintDisassembled(code string) error {
 	if err != nil {
 		return err
 	}
-
 	it := NewInstructionIterator(script)
 	for it.Next() {
 		if it.Arg() != nil && 0 < len(it.Arg()) {

core/asm/asm_test.go

@@ -17,44 +17,78 @@
 package asm
 
 import (
-	"testing"
-
 	"encoding/hex"
+	"fmt"
+	"strings"
+	"testing"
 )
 
 // Tests disassembling instructions
 func TestInstructionIterator(t *testing.T) {
 	for i, tc := range []struct {
-		want    int
-		code    string
-		wantErr string
+		code       string
+		legacyWant string
+		eofWant    string
 	}{
-		{2, "61000000", ""},                        // valid code
-		{0, "6100", "incomplete instruction at 0"}, // invalid code
-		{2, "5900", ""},                            // push0
-		{0, "", ""},                                // empty
-		{2, "d1aabb00", ""},                        // DATALOADN(aabb),STOP
-		{0, "d1aa", "incomplete instruction at 0"}, // DATALOADN(aa) invalid
+		{"", "", ""}, // empty
+		{"6100", `err: incomplete instruction at 0`, `err: incomplete instruction at 0`},
+		{"61000000", `
+00000: PUSH2 0x0000
+00003: STOP`, `
+00000: PUSH2 0x0000
+00003: STOP`},
+		{"5F00", `
+00000: PUSH0
+00001: STOP`, `
+00000: PUSH0
+00001: STOP`},
+		{"d1aabb00", `00000: DATALOADN
+00001: opcode 0xaa not defined
+00002: opcode 0xbb not defined
+00003: STOP`, `
+00000: DATALOADN 0xaabb
+00003: STOP`}, // DATALOADN(aabb),STOP
+		{"d1aa", `
+00000: DATALOADN
+00001: opcode 0xaa not defined`, "err: incomplete instruction at 0\n"}, // DATALOADN(aa) invalid
+		{"e20211223344556600", `
+00000: RJUMPV
+00001: MUL
+00002: GT
+00003: opcode 0x22 not defined
+00004: CALLER
+00005: DIFFICULTY
+00006: SSTORE
+err: incomplete instruction at 7`, `
+00000: RJUMPV 0x02112233445566
+00008: STOP`}, // RJUMPV( 6 bytes), STOP
 
 	} {
 		var (
-			have    int
 			code, _ = hex.DecodeString(tc.code)
-			it      = NewInstructionIterator(code)
+			legacy  = strings.TrimSpace(disassembly(NewInstructionIterator(code)))
+			eof     = strings.TrimSpace(disassembly(NewEOFInstructionIterator(code)))
 		)
-		for it.Next() {
-			have++
+		if want := strings.TrimSpace(tc.legacyWant); legacy != want {
+			t.Errorf("test %d: wrong (legacy) output. have:\n%q\nwant:\n%q\n", i, legacy, want)
 		}
-		var haveErr = ""
-		if it.Error() != nil {
-			haveErr = it.Error().Error()
-		}
-		if haveErr != tc.wantErr {
-			t.Errorf("test %d: encountered error: %q want %q", i, haveErr, tc.wantErr)
-			continue
-		}
-		if have != tc.want {
-			t.Errorf("wrong instruction count, have %d want %d", have, tc.want)
+		if want := strings.TrimSpace(tc.eofWant); eof != want {
+			t.Errorf("test %d: wrong (eof) output. have:\n%q\nwant:\n%q\n", i, eof, want)
 		}
 	}
 }
+
+func disassembly(it *instructionIterator) string {
+	var out = new(strings.Builder)
+	for it.Next() {
+		if it.Arg() != nil && 0 < len(it.Arg()) {
+			fmt.Fprintf(out, "%05x: %v %#x\n", it.PC(), it.Op(), it.Arg())
+		} else {
+			fmt.Fprintf(out, "%05x: %v\n", it.PC(), it.Op())
+		}
+	}
+	if err := it.Error(); err != nil {
+		fmt.Fprintf(out, "err: %v\n", err)
+	}
+	return out.String()
+}

core/vm/analysis_eof.go

@@ -38,7 +38,7 @@ func eofCodeBitmapInternal(code, bits bitvec) bitvec {
 		if op == RJUMPV {
 			// RJUMPV is unique as it has a variable sized operand.
 			// The total size is determined by the count byte which
-			// immediate proceeds RJUMPV. Truncation will be caught
+			// immediate follows RJUMPV. Truncation will be caught
 			// in other validation steps -- for now, just return a
 			// valid bitmap for as much of the code as is
 			// available.

core/vm/eof_immediates.go

@@ -58,6 +58,13 @@ func init() {
 
 // Immediates returns the number bytes of immediates (argument not from
 // stack but from code) a given opcode has.
+// OBS:
+//   - This function assumes EOF instruction-set. It cannot be upon in
+//     a. pre-EOF code
+//     b. post-EOF but legacy code
+//   - RJUMPV is unique as it has a variable sized operand. The total size is
+//     determined by the count byte which immediately follows RJUMPV. This method
+//     will return '3' for RJUMPV, which is the minimum.
 func Immediates(op OpCode) int {
 	return int(immediates[op])
 }