core/vm: faster code analysis (ethereum#23381)

* core/vm: more detailed benchmark for jumpdest analysis

* core/vm: make jd analysis benchmark alloc free

* core/vm: improve jumpdest analysis

* core/vm: improve worst-case

* core/vm: further improvements in analysis

* core/vm: improve jumpdest analysis >PUSH15

* core/vm: make jd analysis ref by value

* core/vm: fix misspell

* core/vm: improve set8 and set16 a bit

* core/vm: reduce amount of code

* core/vm: optimize byte copying

core/vm/analysis.go

@@ -16,17 +16,49 @@
 
 package vm
 
+const (
+	set2BitsMask = uint16(0b1100_0000_0000_0000)
+	set3BitsMask = uint16(0b1110_0000_0000_0000)
+	set4BitsMask = uint16(0b1111_0000_0000_0000)
+	set5BitsMask = uint16(0b1111_1000_0000_0000)
+	set6BitsMask = uint16(0b1111_1100_0000_0000)
+	set7BitsMask = uint16(0b1111_1110_0000_0000)
+)
+
 // bitvec is a bit vector which maps bytes in a program.
 // An unset bit means the byte is an opcode, a set bit means
 // it's data (i.e. argument of PUSHxx).
 type bitvec []byte
 
-func (bits *bitvec) set(pos uint64) {
-	(*bits)[pos/8] |= 0x80 >> (pos % 8)
+var lookup = [8]byte{
+	0x80, 0x40, 0x20, 0x10, 0x8, 0x4, 0x2, 0x1,
+}
+
+func (bits bitvec) set1(pos uint64) {
+	bits[pos/8] |= lookup[pos%8]
+}
+
+func (bits bitvec) setN(flag uint16, pos uint64) {
+	a := flag >> (pos % 8)
+	bits[pos/8] |= byte(a >> 8)
+	if b := byte(a); b != 0 {
+		//	If the bit-setting affects the neighbouring byte, we can assign - no need to OR it,
+		//	since it's the first write to that byte
+		bits[pos/8+1] = b
+	}
+}
+
+func (bits bitvec) set8(pos uint64) {
+	a := byte(0xFF >> (pos % 8))
+	bits[pos/8] |= a
+	bits[pos/8+1] = ^a
 }
-func (bits *bitvec) set8(pos uint64) {
-	(*bits)[pos/8] |= 0xFF >> (pos % 8)
-	(*bits)[pos/8+1] |= ^(0xFF >> (pos % 8))
+
+func (bits bitvec) set16(pos uint64) {
+	a := byte(0xFF >> (pos % 8))
+	bits[pos/8] |= a
+	bits[pos/8+1] = 0xFF
+	bits[pos/8+2] = ^a
 }
 
 // codeSegment checks if the position is in a code segment.
@@ -40,22 +72,52 @@ func codeBitmap(code []byte) bitvec {
 	// ends with a PUSH32, the algorithm will push zeroes onto the
 	// bitvector outside the bounds of the actual code.
 	bits := make(bitvec, len(code)/8+1+4)
+	return codeBitmapInternal(code, bits)
+}
+
+// codeBitmapInternal is the internal implementation of codeBitmap.
+// It exists for the purpose of being able to run benchmark tests
+// without dynamic allocations affecting the results.
+func codeBitmapInternal(code, bits bitvec) bitvec {
 	for pc := uint64(0); pc < uint64(len(code)); {
 		op := OpCode(code[pc])
-
-		if op >= PUSH1 && op <= PUSH32 {
-			numbits := op - PUSH1 + 1
-			pc++
+		pc++
+		if op < PUSH1 || op > PUSH32 {
+			continue
+		}
+		numbits := op - PUSH1 + 1
+		if numbits >= 8 {
+			for ; numbits >= 16; numbits -= 16 {
+				bits.set16(pc)
+				pc += 16
+			}
 			for ; numbits >= 8; numbits -= 8 {
-				bits.set8(pc) // 8
+				bits.set8(pc)
 				pc += 8
 			}
-			for ; numbits > 0; numbits-- {
-				bits.set(pc)
-				pc++
-			}
-		} else {
-			pc++
+		}
+		switch numbits {
+		case 1:
+			bits.set1(pc)
+			pc += 1
+		case 2:
+			bits.setN(set2BitsMask, pc)
+			pc += 2
+		case 3:
+			bits.setN(set3BitsMask, pc)
+			pc += 3
+		case 4:
+			bits.setN(set4BitsMask, pc)
+			pc += 4
+		case 5:
+			bits.setN(set5BitsMask, pc)
+			pc += 5
+		case 6:
+			bits.setN(set6BitsMask, pc)
+			pc += 6
+		case 7:
+			bits.setN(set7BitsMask, pc)
+			pc += 7
 		}
 	}
 	return bits

core/vm/analysis_test.go

@@ -47,10 +47,10 @@ func TestJumpDestAnalysis(t *testing.T) {
 		{[]byte{byte(PUSH32)}, 0xFF, 1},
 		{[]byte{byte(PUSH32)}, 0xFF, 2},
 	}
-	for _, test := range tests {
+	for i, test := range tests {
 		ret := codeBitmap(test.code)
 		if ret[test.which] != test.exp {
-			t.Fatalf("expected %x, got %02x", test.exp, ret[test.which])
+			t.Fatalf("test %d: expected %x, got %02x", i, test.exp, ret[test.which])
 		}
 	}
 }
@@ -73,3 +73,23 @@ func BenchmarkJumpdestHashing_1200k(bench *testing.B) {
 	}
 	bench.StopTimer()
 }
+
+func BenchmarkJumpdestOpAnalysis(bench *testing.B) {
+	var op OpCode
+	bencher := func(b *testing.B) {
+		code := make([]byte, 32*b.N)
+		for i := range code {
+			code[i] = byte(op)
+		}
+		bits := make(bitvec, len(code)/8+1+4)
+		b.ResetTimer()
+		codeBitmapInternal(code, bits)
+	}
+	for op = PUSH1; op <= PUSH32; op++ {
+		bench.Run(op.String(), bencher)
+	}
+	op = JUMPDEST
+	bench.Run(op.String(), bencher)
+	op = STOP
+	bench.Run(op.String(), bencher)
+}

core/vm/instructions.go

@@ -701,6 +701,7 @@ func opCall(pc *uint64, interpreter *EVMInterpreter, callContext *callCtx) ([]by
 	}
 	stack.push(&temp)
 	if err == nil || err == ErrExecutionReverted {
+		ret = common.CopyBytes(ret)
 		callContext.memory.Set(retOffset.Uint64(), retSize.Uint64(), ret)
 	}
 	callContext.contract.Gas += returnGas
@@ -735,6 +736,7 @@ func opCallCode(pc *uint64, interpreter *EVMInterpreter, callContext *callCtx) (
 	}
 	stack.push(&temp)
 	if err == nil || err == ErrExecutionReverted {
+		ret = common.CopyBytes(ret)
 		callContext.memory.Set(retOffset.Uint64(), retSize.Uint64(), ret)
 	}
 	callContext.contract.Gas += returnGas
@@ -762,6 +764,7 @@ func opDelegateCall(pc *uint64, interpreter *EVMInterpreter, callContext *callCt
 	}
 	stack.push(&temp)
 	if err == nil || err == ErrExecutionReverted {
+		ret = common.CopyBytes(ret)
 		callContext.memory.Set(retOffset.Uint64(), retSize.Uint64(), ret)
 	}
 	callContext.contract.Gas += returnGas
@@ -789,6 +792,7 @@ func opStaticCall(pc *uint64, interpreter *EVMInterpreter, callContext *callCtx)
 	}
 	stack.push(&temp)
 	if err == nil || err == ErrExecutionReverted {
+		ret = common.CopyBytes(ret)
 		callContext.memory.Set(retOffset.Uint64(), retSize.Uint64(), ret)
 	}
 	callContext.contract.Gas += returnGas

core/vm/interpreter.go

@@ -290,7 +290,7 @@ func (in *EVMInterpreter) Run(contract *Contract, input []byte, readOnly bool) (
 		// if the operation clears the return data (e.g. it has returning data)
 		// set the last return to the result of the operation.
 		if operation.returns {
-			in.returnData = common.CopyBytes(res)
+			in.returnData = res
 		}
 
 		switch {