codecin 5.4.2__py3-none-any.whl

codecin/native/engine/engine_test.go

@@ -0,0 +1,198 @@
+package engine
+
+import (
+	"bytes"
+	"compress/zlib"
+	"hash/crc32"
+	"strings"
+	"testing"
+
+	"codecin-native/ir"
+)
+
+// 锁定 docs/SUGGESTIONS.md 批次 B 的引擎侧加固:
+//   * decodeBytecode 不再信任不可信头部 (版本 / count / argc / opcode);
+//   * 步数用尽报错误而不是伪装成正常停机;
+//   * CROM 解压有上限 (zip bomb) 且 mem_size 必须自洽。
+
+func tinyProgram() ir.Program {
+	return ir.Program{
+		Instructions: []ir.Instr{
+			{Op: "MOV", Args: []ir.Operand{ir.Reg(0), ir.Imm(42)}},
+			{Op: "HALT"},
+		},
+		Labels:     map[string]int{"main": 0},
+		DataLabels: map[string]int{},
+	}
+}
+
+func TestEncodeRunRoundTrip(t *testing.T) {
+	prog := tinyProgram()
+	bc, err := EncodeProgram(prog, 0)
+	if err != nil {
+		t.Fatalf("encode: %v", err)
+	}
+	mem := make([]byte, 4096)
+	sp := int64(len(mem) - 8)
+	res := Run(bc, mem, 0, sp, int64(len(mem)/2), nil, 1000)
+	if res == nil {
+		t.Fatal("Run 返回 nil")
+	}
+	if res.ErrMsg != "" {
+		t.Fatalf("意外错误: %s", res.ErrMsg)
+	}
+	if res.Regs[0] != 42 {
+		t.Fatalf("R0 = %d, 期望 42", res.Regs[0])
+	}
+}
+
+func TestDecodeRejectsMalformedInput(t *testing.T) {
+	valid, err := EncodeProgram(tinyProgram(), 0)
+	if err != nil {
+		t.Fatalf("encode: %v", err)
+	}
+	if _, _, ok := decodeBytecode(valid); !ok {
+		t.Fatal("合法字节码被拒绝")
+	}
+
+	cases := []struct {
+		name string
+		bc   []byte
+	}{
+		{"empty", []byte{}},
+		{"bad_magic", append([]byte("XXXX"), valid[4:]...)},
+		{"short", valid[:6]},
+	}
+	for _, c := range cases {
+		if _, _, ok := decodeBytecode(c.bc); ok {
+			t.Errorf("%s: 畸形字节码被接受", c.name)
+		}
+	}
+
+	// 版本字节
+	badVer := append([]byte{}, valid...)
+	badVer[4] = 99
+	if _, _, ok := decodeBytecode(badVer); ok {
+		t.Error("未知版本被接受")
+	}
+
+	// count 声明远超实际长度 (旧实现按 count 预分配 -> OOM)
+	huge := []byte{'U', 'C', 'B', 'C', 1, 0, 0, 0, 0, 0xFF, 0xFF, 0xFF, 0xFF}
+	if _, _, ok := decodeBytecode(huge); ok {
+		t.Error("超大 count 被接受")
+	}
+
+	// argc 与操作码不匹配 (MOV 需要 2 个操作数, 这里声明 0 个)
+	argcBad := []byte{'U', 'C', 'B', 'C', 1, 0, 0, 0, 0, 1, 0, 0, 0, opMOV, 0}
+	if _, _, ok := decodeBytecode(argcBad); ok {
+		t.Error("argc 不匹配的字节码被接受")
+	}
+
+	// 未定义 opcode
+	opBad := []byte{'U', 'C', 'B', 'C', 1, 0, 0, 0, 0, 1, 0, 0, 0, 250, 0}
+	if _, _, ok := decodeBytecode(opBad); ok {
+		t.Error("未知操作码被接受")
+	}
+}
+
+func TestStepLimitIsAnError(t *testing.T) {
+	// JMP 自身 = 死循环
+	prog := ir.Program{
+		Instructions: []ir.Instr{
+			{Op: "JMP", Args: []ir.Operand{ir.Label("loop")}},
+			{Op: "HALT"},
+		},
+		Labels:     map[string]int{"loop": 0},
+		DataLabels: map[string]int{},
+	}
+	bc, err := EncodeProgram(prog, 0)
+	if err != nil {
+		t.Fatalf("encode: %v", err)
+	}
+	mem := make([]byte, 4096)
+	res := Run(bc, mem, 0, int64(len(mem)-8), int64(len(mem)/2), nil, 500)
+	if res == nil {
+		t.Fatal("Run 返回 nil")
+	}
+	if res.Status != StatusError {
+		t.Fatalf("步数用尽应返回 StatusError, 实际 status=%d", res.Status)
+	}
+	if !strings.Contains(res.ErrMsg, "limit") {
+		t.Fatalf("错误信息应提到 limit, 实际: %q", res.ErrMsg)
+	}
+}
+
+func TestCromRoundTrip(t *testing.T) {
+	data := []byte(strings.Repeat("codecin", 64))
+	for _, compress := range []bool{false, true} {
+		packed := CromPack(data, compress)
+		raw, flags, ok := CromUnpack(packed)
+		if !ok {
+			t.Fatalf("compress=%v: 解包失败", compress)
+		}
+		if !bytes.Equal(raw, data) {
+			t.Fatalf("compress=%v: 内容不一致", compress)
+		}
+		if compress && flags&0x01 == 0 {
+			t.Fatalf("compress=%v: flags 未标记压缩", compress)
+		}
+	}
+}
+
+func TestCromRejectsBadInput(t *testing.T) {
+	// magic / 版本
+	if _, _, ok := CromUnpack([]byte("XXXX................")); ok {
+		t.Error("坏 magic 被接受")
+	}
+	bad := CromPack([]byte("abc"), false)
+	bad[4] = 99
+	if _, _, ok := CromUnpack(bad); ok {
+		t.Error("未知版本被接受")
+	}
+	// CRC 破坏
+	bad2 := CromPack([]byte("abcdefgh"), true)
+	bad2[len(bad2)-1] ^= 0xFF
+	if _, _, ok := CromUnpack(bad2); ok {
+		t.Error("CRC 损坏被接受")
+	}
+	// zip bomb: 头部 mem_size 远小于实际解压长度
+	bomb := cromWithDeclaredSize(t, 1024, bytes.Repeat([]byte{0}, 8<<20))
+	if _, _, ok := CromUnpack(bomb); ok {
+		t.Error("zip bomb 被接受")
+	}
+	// mem_size 大于实际载荷
+	mismatch := cromWithDeclaredSize(t, 1<<20, []byte("short"))
+	if _, _, ok := CromUnpack(mismatch); ok {
+		t.Error("mem_size 与载荷不符被接受")
+	}
+}
+
+// cromWithDeclaredSize 手工构造头部 mem_size 与载荷不符的 CROM。
+func cromWithDeclaredSize(t *testing.T, declared uint32, raw []byte) []byte {
+	t.Helper()
+	var buf bytes.Buffer
+	w := zlib.NewWriter(&buf)
+	if _, err := w.Write(raw); err != nil {
+		t.Fatalf("zlib write: %v", err)
+	}
+	if err := w.Close(); err != nil {
+		t.Fatalf("zlib close: %v", err)
+	}
+	payload := buf.Bytes()
+
+	out := make([]byte, 16+len(payload))
+	copy(out[0:4], "CROM")
+	out[4] = cromVersion
+	out[5] = byte(declared)
+	out[6] = byte(declared >> 8)
+	out[7] = byte(declared >> 16)
+	out[8] = byte(declared >> 24)
+	out[9] = 0x01 // 压缩
+	crc := crc32.ChecksumIEEE(payload)
+	out[10] = byte(crc)
+	out[11] = byte(crc >> 8)
+	out[12] = byte(crc >> 16)
+	out[13] = byte(crc >> 24)
+	copy(out[16:], payload)
+	return out
+}

codecin/native/engine/isa_gen.go

@@ -0,0 +1,445 @@
+// Code generated by script/gen_native_isa.py from codecin/isa.py; DO NOT EDIT.
+// 单一事实来源: codecin/isa.py Opcode / Syscall / KIND_* 常量.
+// 修改指令集后运行: python script/gen_native_isa.py
+
+package engine
+
+// 操作数种类 (与 isa.py KIND_* 一致)
+const kindReg = 0
+const kindImm = 1
+const kindVec = 2
+const kindVecLane = 3
+const kindMem = 4
+const kindCond = 5
+const kindFloat = 6
+const kindStr = 7
+
+// 操作码 (与 isa.py Opcode 完整一致; 原生 VM 仅实现子集, 见 vm.go opcodeSupported)
+const opMOV = 0
+const opLOAD = 1
+const opSTORE = 2
+const opADD = 3
+const opSUB = 4
+const opMUL = 5
+const opDIV = 6
+const opAND = 7
+const opOR = 8
+const opXOR = 9
+const opSHL = 10
+const opSHR = 11
+const opINC = 12
+const opDEC = 13
+const opCMP = 14
+const opJMP = 15
+const opJZ = 16
+const opJNZ = 17
+const opJE = 18
+const opJL = 19
+const opJG = 20
+const opPUSH = 21
+const opPOP = 22
+const opCALL = 23
+const opRET = 24
+const opIN = 25
+const opOUT = 26
+const opHALT = 27
+const opADDS = 28
+const opSUBS = 29
+const opADDC = 30
+const opSUBC = 31
+const opLSL = 32
+const opLSR = 33
+const opASR = 34
+const opROR = 35
+const opMVN = 36
+const opEOR = 37
+const opBIC = 38
+const opORN = 39
+const opLDR = 40
+const opSTR = 41
+const opLDP = 42
+const opSTP = 43
+const opCBZ = 44
+const opCBNZ = 45
+const opTBZ = 46
+const opTBNZ = 47
+const opB = 48
+const opBL = 49
+const opBR = 50
+const opNOP = 51
+const opWFE = 52
+const opWFI = 53
+const opSEV = 54
+const opCSEL = 55
+const opCSINC = 56
+const opCSINV = 57
+const opCSNEG = 58
+const opSXTB = 59
+const opSXTH = 60
+const opSXTW = 61
+const opUXTB = 62
+const opUXTH = 63
+const opCLZ = 64
+const opCLS = 65
+const opRBIT = 66
+const opREV = 67
+const opFADD = 68
+const opFSUB = 69
+const opFMUL = 70
+const opFDIV = 71
+const opFCMP = 72
+const opFCVT = 73
+const opFABS = 74
+const opFNEG = 75
+const opLDRS = 76
+const opSTRS = 77
+const opVADD = 78
+const opVSUB = 79
+const opVMUL = 80
+const opVDIV = 81
+const opVLD1 = 82
+const opVST1 = 83
+const opLB = 84
+const opLH = 85
+const opLW = 86
+const opLD = 87
+const opSB = 88
+const opSH = 89
+const opSW = 90
+const opSD = 91
+const opADDI = 92
+const opSLTI = 93
+const opSLTIU = 94
+const opXORI = 95
+const opORI = 96
+const opANDI = 97
+const opSLLI = 98
+const opSRLI = 99
+const opSRAI = 100
+const opBEQ = 101
+const opBNE = 102
+const opBLT = 103
+const opBGE = 104
+const opBLTU = 105
+const opBGEU = 106
+const opJALR = 107
+const opJAL = 108
+const opLUI = 109
+const opAUIPC = 110
+const opSYS = 111
+
+// 操作码名 -> 编码 (供 UCBC 编码器使用)
+var opcodeByName = map[string]uint8{
+	"MOV":   opMOV,
+	"LOAD":  opLOAD,
+	"STORE": opSTORE,
+	"ADD":   opADD,
+	"SUB":   opSUB,
+	"MUL":   opMUL,
+	"DIV":   opDIV,
+	"AND":   opAND,
+	"OR":    opOR,
+	"XOR":   opXOR,
+	"SHL":   opSHL,
+	"SHR":   opSHR,
+	"INC":   opINC,
+	"DEC":   opDEC,
+	"CMP":   opCMP,
+	"JMP":   opJMP,
+	"JZ":    opJZ,
+	"JNZ":   opJNZ,
+	"JE":    opJE,
+	"JL":    opJL,
+	"JG":    opJG,
+	"PUSH":  opPUSH,
+	"POP":   opPOP,
+	"CALL":  opCALL,
+	"RET":   opRET,
+	"IN":    opIN,
+	"OUT":   opOUT,
+	"HALT":  opHALT,
+	"ADDS":  opADDS,
+	"SUBS":  opSUBS,
+	"ADDC":  opADDC,
+	"SUBC":  opSUBC,
+	"LSL":   opLSL,
+	"LSR":   opLSR,
+	"ASR":   opASR,
+	"ROR":   opROR,
+	"MVN":   opMVN,
+	"EOR":   opEOR,
+	"BIC":   opBIC,
+	"ORN":   opORN,
+	"LDR":   opLDR,
+	"STR":   opSTR,
+	"LDP":   opLDP,
+	"STP":   opSTP,
+	"CBZ":   opCBZ,
+	"CBNZ":  opCBNZ,
+	"TBZ":   opTBZ,
+	"TBNZ":  opTBNZ,
+	"B":     opB,
+	"BL":    opBL,
+	"BR":    opBR,
+	"NOP":   opNOP,
+	"WFE":   opWFE,
+	"WFI":   opWFI,
+	"SEV":   opSEV,
+	"CSEL":  opCSEL,
+	"CSINC": opCSINC,
+	"CSINV": opCSINV,
+	"CSNEG": opCSNEG,
+	"SXTB":  opSXTB,
+	"SXTH":  opSXTH,
+	"SXTW":  opSXTW,
+	"UXTB":  opUXTB,
+	"UXTH":  opUXTH,
+	"CLZ":   opCLZ,
+	"CLS":   opCLS,
+	"RBIT":  opRBIT,
+	"REV":   opREV,
+	"FADD":  opFADD,
+	"FSUB":  opFSUB,
+	"FMUL":  opFMUL,
+	"FDIV":  opFDIV,
+	"FCMP":  opFCMP,
+	"FCVT":  opFCVT,
+	"FABS":  opFABS,
+	"FNEG":  opFNEG,
+	"LDRS":  opLDRS,
+	"STRS":  opSTRS,
+	"VADD":  opVADD,
+	"VSUB":  opVSUB,
+	"VMUL":  opVMUL,
+	"VDIV":  opVDIV,
+	"VLD1":  opVLD1,
+	"VST1":  opVST1,
+	"LB":    opLB,
+	"LH":    opLH,
+	"LW":    opLW,
+	"LD":    opLD,
+	"SB":    opSB,
+	"SH":    opSH,
+	"SW":    opSW,
+	"SD":    opSD,
+	"ADDI":  opADDI,
+	"SLTI":  opSLTI,
+	"SLTIU": opSLTIU,
+	"XORI":  opXORI,
+	"ORI":   opORI,
+	"ANDI":  opANDI,
+	"SLLI":  opSLLI,
+	"SRLI":  opSRLI,
+	"SRAI":  opSRAI,
+	"BEQ":   opBEQ,
+	"BNE":   opBNE,
+	"BLT":   opBLT,
+	"BGE":   opBGE,
+	"BLTU":  opBLTU,
+	"BGEU":  opBGEU,
+	"JALR":  opJALR,
+	"JAL":   opJAL,
+	"LUI":   opLUI,
+	"AUIPC": opAUIPC,
+	"SYS":   opSYS,
+}
+
+// 每个操作码的参数个数 (-1 表示变长, 不校验);
+// 与 isa.py Constants.ARG_COUNTS 一致。原生 VM 用它校验不可信字节码,
+// 避免 argc 与操作码不匹配时 execute 里 args[i] 越界 panic。
+var argCounts = [112]int16{
+	opMOV:   2,
+	opLOAD:  2,
+	opSTORE: 2,
+	opADD:   2,
+	opSUB:   2,
+	opMUL:   2,
+	opDIV:   2,
+	opAND:   2,
+	opOR:    2,
+	opXOR:   2,
+	opSHL:   2,
+	opSHR:   2,
+	opINC:   1,
+	opDEC:   1,
+	opCMP:   2,
+	opJMP:   1,
+	opJZ:    1,
+	opJNZ:   1,
+	opJE:    1,
+	opJL:    1,
+	opJG:    1,
+	opPUSH:  1,
+	opPOP:   1,
+	opCALL:  1,
+	opRET:   0,
+	opIN:    1,
+	opOUT:   1,
+	opHALT:  0,
+	opADDS:  3,
+	opSUBS:  3,
+	opADDC:  3,
+	opSUBC:  3,
+	opLSL:   3,
+	opLSR:   3,
+	opASR:   3,
+	opROR:   3,
+	opMVN:   2,
+	opEOR:   3,
+	opBIC:   3,
+	opORN:   3,
+	opLDR:   2,
+	opSTR:   2,
+	opLDP:   3,
+	opSTP:   3,
+	opCBZ:   2,
+	opCBNZ:  2,
+	opTBZ:   3,
+	opTBNZ:  3,
+	opB:     -1,
+	opBL:    1,
+	opBR:    1,
+	opNOP:   0,
+	opWFE:   0,
+	opWFI:   0,
+	opSEV:   0,
+	opCSEL:  4,
+	opCSINC: 4,
+	opCSINV: 4,
+	opCSNEG: 4,
+	opSXTB:  2,
+	opSXTH:  2,
+	opSXTW:  2,
+	opUXTB:  2,
+	opUXTH:  2,
+	opCLZ:   2,
+	opCLS:   2,
+	opRBIT:  2,
+	opREV:   2,
+	opFADD:  3,
+	opFSUB:  3,
+	opFMUL:  3,
+	opFDIV:  3,
+	opFCMP:  2,
+	opFCVT:  2,
+	opFABS:  2,
+	opFNEG:  2,
+	opLDRS:  2,
+	opSTRS:  2,
+	opVADD:  3,
+	opVSUB:  3,
+	opVMUL:  3,
+	opVDIV:  3,
+	opVLD1:  2,
+	opVST1:  2,
+	opLB:    2,
+	opLH:    2,
+	opLW:    2,
+	opLD:    2,
+	opSB:    2,
+	opSH:    2,
+	opSW:    2,
+	opSD:    2,
+	opADDI:  3,
+	opSLTI:  3,
+	opSLTIU: 3,
+	opXORI:  3,
+	opORI:   3,
+	opANDI:  3,
+	opSLLI:  3,
+	opSRLI:  3,
+	opSRAI:  3,
+	opBEQ:   3,
+	opBNE:   3,
+	opBLT:   3,
+	opBGE:   3,
+	opBLTU:  3,
+	opBGEU:  3,
+	opJALR:  -1,
+	opJAL:   2,
+	opLUI:   2,
+	opAUIPC: 2,
+	opSYS:   -1,
+}
+
+// SYS 功能号 (与 isa.py Syscall 一致)
+const sysSQRT = 0
+const sysPOW = 1
+const sysABS = 2
+const sysSIN = 3
+const sysCOS = 4
+const sysTAN = 5
+const sysFADD = 6
+const sysFSUB = 7
+const sysFMUL = 8
+const sysFDIV = 9
+const sysFCMP = 10
+const sysFTOI = 11
+const sysITOF = 12
+const sysRAND = 13
+const sysSRAND = 14
+const sysTIME = 15
+const sysSTRLEN = 16
+const sysSTRCMP = 17
+const sysSTRCPY = 18
+const sysSTRCAT = 19
+const sysMALLOC = 20
+const sysPRINTFLO = 21
+const sysITOA = 22
+const sysFTOA = 23
+const sysPRINTSTR = 24
+const sysSTRCONCAT = 25
+const sysBOOLSTR = 26
+const sysABORT = 27
+const sysSUBSTR = 28
+const sysINDEXOF = 29
+const sysTOUPPER = 30
+const sysTOLOWER = 31
+const sysFLOOR = 32
+const sysCEIL = 33
+const sysROUND = 34
+const sysATOI = 35
+const sysTRIM = 36
+const sysLTRIM = 37
+const sysRTRIM = 38
+const sysAUDIOPLAY = 39
+const sysAUDIOSTOP = 40
+const sysAUDIOVOL = 41
+const sysAUDIOWAIT = 42
+const sysCANVASNEW = 43
+const sysCANVASSET = 44
+const sysCANVASRECT = 45
+const sysCANVASCIRC = 46
+const sysCANVASTEXT = 47
+const sysCANVASLINE = 48
+const sysCANVASSAVE = 49
+const sysCANVASSHOW = 50
+const sysFILEREAD = 51
+const sysFILEWRITE = 52
+const sysFILEAPPEND = 53
+const sysFILEEXISTS = 54
+const sysFILEDELETE = 55
+const sysFILESIZE = 56
+const sysMKDIR = 57
+const sysDIRLIST = 58
+const sysEXEC = 59
+const sysEXECOUTPUT = 60
+const sysGETENV = 61
+const sysSETENV = 62
+const sysOSNAME = 63
+const sysHOSTNAME = 64
+const sysUSERNAME = 65
+const sysCWD = 66
+const sysHOMEDIR = 67
+const sysTERMUXAVAIL = 68
+const sysTERMUXNOTIFY = 69
+const sysTERMUXTOAST = 70
+const sysTERMUXCLIPGET = 71
+const sysTERMUXCLIPSET = 72
+const sysTERMUXBATTERY = 73
+const sysTERMUXVIBRATE = 74
+const sysTERMUXTTS = 75
+const sysTERMUXLOCATION = 76
+const sysTERMUXWIFI = 77
+const sysTERMUXDIALOG = 78
+const sysTERMUXSMS = 79