Use an enum to represent registers, parse txt files, check CPU state against expected

decoding.odin

@@ -66,7 +66,7 @@ get_op :: proc(inst: Instruction) -> (Op, bool) {
     }
     return op, interseg
 }
-parse_operand :: proc(inst: InstructionInfo, opinfo: OperandInfo, data: []u8, processed: ^int, word: bool, has_segreg: Maybe(Register)) -> Operand {
+parse_operand :: proc(inst: InstructionInfo, opinfo: OperandInfo, data: []u8, processed: ^int, word: bool, has_segreg: Maybe(RegisterId)) -> Operand {
     operand: Operand = None{}
     switch opinfo {
     case .None:
@@ -80,11 +80,11 @@ parse_operand :: proc(inst: InstructionInfo, opinfo: OperandInfo, data: []u8, pr
         case .SecondByteLast3:   reg = data[1] & 0b111
         }
         if opinfo == .SegmentRegister {
-            operand = (RegisterId){id = SEGMENT_REGISTER_START + (int)(reg), access = .Full}
+            operand = (RegisterId){idx = SEGMENT_REGISTER_START + (int)(reg), access = .Full}
         } else if word {
-            operand = RegisterId { id = (int)(reg), access = .Full }
+            operand = RegisterId { idx = (int)(reg), access = .Full }
         } else {
-            operand = RegisterId { id = (int)(reg % 4), access = reg < 4 ? .Low : .High }
+            operand = RegisterId { idx = (int)(reg % 4), access = reg < 4 ? .Low : .High }
         }
     case .RegisterMemory:
         mod := data[1] >> 6
@@ -107,9 +107,9 @@ parse_operand :: proc(inst: InstructionInfo, opinfo: OperandInfo, data: []u8, pr
             processed^ += 2
         } else if mod == 3 {
             if word {
-                op = RegisterId { id = (int)(rm), access = .Full }
+                op = RegisterId { idx = (int)(rm), access = .Full }
             } else {
-                op = RegisterId { id = (int)(rm % 4), access = rm < 4 ? .Low : .High }
+                op = RegisterId { idx = (int)(rm % 4), access = rm < 4 ? .Low : .High }
             }
         }
         operand = op
@@ -126,7 +126,7 @@ parse_operand :: proc(inst: InstructionInfo, opinfo: OperandInfo, data: []u8, pr
         operand = Immediate { value = u16(data[processed^]), size = .Unsigned8 }
         processed^ += 1
     case .Accumulator:
-        operand = RegisterId { id = 0, access = word ? .Full : .Low }
+        operand = RegisterId { idx = 0, access = word ? .Full : .Low }
     case .DirectAddress:
         // operand = DirectAddress { value = get_i16(data[1:]) }
         operand = (DirectAddress)(get_i16(data[1:]))
@@ -136,10 +136,10 @@ parse_operand :: proc(inst: InstructionInfo, opinfo: OperandInfo, data: []u8, pr
         // NOTE: In order to mimic the label offset, you have to take the value you got and add two
         operand = (Jump)((i8)(data[1]) + 2)
     case .VariablePort:
-        operand = RegisterId { id = (int)(variable_port.code), access = .Full }
+        operand = RegisterId { idx = (int)(Register.dx), access = .Full }
     case .ShiftRotate:
         v_flag := data[0] & 0b10 != 0
-        operand = v_flag ? RegisterId { id = 1, access = .Low } : Immediate { value = 1 }
+        operand = v_flag ? RegisterId { idx = 1, access = .Low } : Immediate { value = 1 }
     case .Repeat:
         bits := (data[1] & 0b1110) >> 1
         w := (data[1] & 0b1) == 1 ? "w" : "b"
@@ -169,7 +169,7 @@ parse_operand :: proc(inst: InstructionInfo, opinfo: OperandInfo, data: []u8, pr
 
 decode_data :: proc(inst_list: ^[dynamic]Instruction, data: []u8, bytes_to_read: int) {
     idx := 0
-    has_segment: Maybe(Register)
+    has_segment: Maybe(RegisterId)
     has_lock: bool
     for idx < bytes_to_read {
         instruction: Instruction
@@ -203,7 +203,7 @@ decode_data :: proc(inst_list: ^[dynamic]Instruction, data: []u8, bytes_to_read:
             continue
         } else if inst.opname == .SEGMENT {
             reg := (curr_byte & 0b11000) >> 3
-            has_segment = CPU.registers[SEGMENT_REGISTER_START+reg]
+            has_segment = RegisterId { idx = int(SEGMENT_REGISTER_START+reg) }
             idx += 1
             continue
         } else if inst.opname == .AAM {

execution.odin

@@ -4,34 +4,40 @@ import "core:os"
 import "core:fmt"
 import "core:math"
 import "core:strings"
+import "core:reflect"
 
 get_operand_value :: proc(operand: Operand) -> u16 {
     #partial switch opr in operand {
     case Immediate:
         return opr.value
     case RegisterId:
+        reg_val := CPU.registers[opr.idx]
         switch opr.access {
         case .Low, .High:
-            val := opr.access == .Low ? CPU.registers[opr.id].value.low : CPU.registers[opr.id].value.high
+            return u16(opr.access == .Low ? reg_val.low : reg_val.high)
-            return u16(val)
         case .Full:
-            return CPU.registers[opr.id].value.full
+            return reg_val.full
         }
     }
     return 0
 }
 
-set_register_value :: proc(reg_id: RegisterId, value: u16) {
+set_register_value :: proc(reg: RegisterId, value: u16) {
-    switch reg_id.access {
+    switch reg.access {
     case .Low:
-        CPU.registers[reg_id.id].value.low = u8(value)
+        CPU.registers[reg.idx].low = u8(value)
     case .High:
-        CPU.registers[reg_id.id].value.high = u8(value)
+        CPU.registers[reg.idx].high = u8(value)
     case .Full:
-        CPU.registers[reg_id.id].value.full = u16(value)
+        CPU.registers[reg.idx].full = u16(value)
     }
 }
 
+get_cpu_register_by_name :: proc(cpu: ^Cpu, name: string) -> ^RegisterValue {
+    reg,_ := reflect.enum_from_name(Register, name)
+    return &cpu.registers[int(reg)]
+}
+
 check_zero_flag :: proc(value: u16) {
     CPU.flags.ZF = value == 0
 }
@@ -48,19 +54,19 @@ execute_instruction :: proc(inst: Instruction) {
             set_register_value(reg, src_val)
         case .ADD:
             src_val := get_operand_value(inst.src)
-            val := CPU.registers[reg.id].value.full + src_val
+            val := CPU.registers[reg.idx].full + src_val
             set_register_value(reg, val)
             check_zero_flag(val)
             check_sign_flag(val)
         case .SUB:
             src_val := get_operand_value(inst.src)
-            val := CPU.registers[reg.id].value.full - src_val
+            val := CPU.registers[reg.idx].full - src_val
             set_register_value(reg, val)
             check_zero_flag(val)
             check_sign_flag(val)
         case .CMP:
             src_val := get_operand_value(inst.src)
-            val := CPU.registers[reg.id].value.full - src_val
+            val := CPU.registers[reg.idx].full - src_val
             check_zero_flag(val)
             check_sign_flag(val)
         }

printing.odin

@@ -37,7 +37,7 @@ calculate_effective_address :: proc(r_m: u8) -> string {
     return val
 }
 
-get_memory_string :: proc(memoryAddr: MemoryAddr, has_segment: Maybe(Register)) -> string {
+get_memory_string :: proc(memoryAddr: MemoryAddr, has_segment: Maybe(RegisterId)) -> string {
     disp: string
     switch value in memoryAddr.displacement {
     case None:
@@ -53,7 +53,7 @@ get_memory_string :: proc(memoryAddr: MemoryAddr, has_segment: Maybe(Register))
     }
     seg_string: string
     if segreg, ok := has_segment.?; ok {
-        seg_string = fmt.aprintf("%s:", segreg.fullname)
+        seg_string = fmt.aprintf("%s:", get_register_name(segreg))
     }
     text := fmt.aprintf("%s[%s%s]", seg_string, calculate_effective_address(memoryAddr.addr_id), disp)
     return text
@@ -78,14 +78,14 @@ get_register_name :: proc(reg_id: RegisterId) -> string {
     low_names := [?]string{"al", "cl", "dl", "bl"}
     high_names := [?]string{"ah", "ch", "dh", "bh"}
     switch reg_id.access {
-    case .Full: return CPU.registers[reg_id.id].fullname
+    case .Full: return reflect.enum_string(Register(reg_id.idx))
-    case .Low: return low_names[reg_id.id]
+    case .Low: return low_names[reg_id.idx]
-    case .High: return high_names[reg_id.id % 4]
+    case .High: return high_names[reg_id.idx % 4]
     }
     return ""
 }
 
-get_operand_string :: proc(operand: Operand, has_segment: Maybe(Register)) -> string {
+get_operand_string :: proc(operand: Operand, has_segment: Maybe(RegisterId)) -> string {
     string_val: string
     switch val in operand {
     case None:
@@ -106,7 +106,7 @@ get_operand_string :: proc(operand: Operand, has_segment: Maybe(Register)) -> st
     case DirectAddress:
         seg_string: string
         if segreg, ok := has_segment.?; ok {
-            seg_string = fmt.aprintf("%s:", segreg.fullname)
+            seg_string = fmt.aprintf("%s:", get_register_name(segreg))
         }
         string_val = fmt.aprintf("%s[%d]", seg_string, val)
     case Jump:

sim8086.odin

@@ -1,32 +1,15 @@
 package sim_8086
 
 import "core:os"
-import "core:path"
+import path "core:path/filepath"
 import "core:fmt"
 import "core:math"
 import "core:strings"
 
 CPU := Cpu {
-    registers = [12]Register {
-        {fullname = "ax", code = 0b000},
-        {fullname = "cx", code = 0b001},
-        {fullname = "dx", code = 0b010},
-        {fullname = "bx", code = 0b011},
-        {fullname = "sp", code = 0b100},
-        {fullname = "bp", code = 0b101},
-        {fullname = "si", code = 0b110},
-        {fullname = "di", code = 0b111},
-        {fullname = "es", code = 0b000},
-        {fullname = "cs", code = 0b001},
-        {fullname = "ss", code = 0b010},
-        {fullname = "ds", code = 0b011},
-    },
     memory = make([dynamic]u8, 65536),
 }
 
-variable_port := CPU.registers[2]
-
-
 main :: proc() {
     f,err := os.open(os.args[1])
     if err != os.ERROR_NONE {
@@ -59,22 +42,23 @@ main :: proc() {
     }
 
     if what_to_print == "registers" || what_to_print == "all" {
-        print_reg :: proc(reg: Register) {
+        for reg,i in CPU.registers {
-            full := fmt.aprintf("%s: %d ", reg.fullname, reg.value.full)
+            full := fmt.aprintf("%s: %d ", get_register_name(RegisterId{idx=i}), reg.full)
-            hex := fmt.aprintf("0x%04x ", reg.value.full)
+            hex := fmt.aprintf("0x%04x ", reg.full)
             fmt.printf("%s %*[1]s %s %*[4]s %08b %08b",
-                       full, 18 - len(full), "|", hex, 10 - len(hex), "|", reg.value.high, reg.value.low)
+                       full, 18 - len(full), "|", hex, 10 - len(hex), "|", reg.high, reg.low)
             fmt.println()
         }
-        for reg in CPU.registers {
-            print_reg(reg)
-        }
-        // fmt.println("Checking Against Expected State")
 
-        // fmt.println(os.args[1])
+        path,ok := strings.replace(os.args[1], ".bin", ".txt", 1)
-        // path := path.base_no_ext(os.args[1])
+        expected_cpu,_ := extract_expected_cpu_state(path)
-        // fmt.println(path)
+        for reg,i in expected_cpu.registers {
-        // extract_expected_cpu_state(path)
+            if CPU.registers[i].full != reg.full {
+                name := get_register_name(RegisterId{idx=i})
+                msg := "%s register does not match - Expected %04x | Actual %04x"
+                fmt.eprintfln(msg, name, reg.full, CPU.registers[i].full)
+            }
+        }
     }
     if what_to_print == "instructions" || what_to_print == "all" {
         print_instructions_stdout(instructions_list[:])

testing.odin

@@ -4,35 +4,41 @@ import "core:os"
 import "core:fmt"
 import "core:math"
 import "core:strings"
-import "core:text/regex"
+import "core:strconv"
 
-extract_expected_cpu_state :: proc(listing_num: int) -> (Cpu, bool) {
+extract_expected_cpu_state :: proc(filename: string) -> (Cpu, bool) {
     cpu: Cpu
 
-    // filename := fmt.aprintf("./asm_files/list-%04d.txt", listing_num)
+    data,ok := os.read_entire_file(filename)
-    // fmt.println(filename)
+    if !ok {
-    // data,ok := os.read_entire_file(filename)
+        return cpu, false
-    // if !ok {
+    }
-    //     return cpu, false
+    defer delete(data)
-    // }
-    // defer delete(data)
 
-    // content := string(data)
+    content := string(data)
-    // lines := strings.split(content, "\n")
+    lines := strings.split(content, "\n")
-    // defer delete(lines)
+    defer delete(lines)
 
-    // for line in lines {
+    for line in lines {
-    //     for c in line {
+        space_count := 0
-    //         if c != ' ' {
+        for c,i in line {
-    //             continue
+            if space_count == 0  && c != ' ' {
-    //         } else {
+                break
-    //             fmt.print(c)
+            } else if c == ' ' {
-    //         }
+                space_count += 1
-    //     }
+            } else {
-    //     fmt.println()
+                if line[i:i+5] != "flags" {
-    // }
+                    reg_name := line[i:i+2]
+                    reg_value := get_cpu_register_by_name(&cpu, reg_name)
+                    hex_string := line[i+6:i+10]
+                    if hex_num,ok := strconv.parse_int(hex_string, 16); ok {
+                        reg_value^.full = u16(hex_num)
+                    }
+                }
+                break
+            }
+        }
+    }
 
-
-    // cpu.registers[]
     return cpu, true
 }

types.odin

@@ -1,14 +1,25 @@
 package sim_8086
 
-Register :: struct {
+Register :: enum {
-    fullname: string,
+    ax,
-    value: struct #raw_union {
+    cx,
-        using _: struct {
+    dx,
-            low, high: byte,
+    bx,
-        },
+    sp,
-        full: u16,
+    bp,
+    si,
+    di,
+    es,
+    cs,
+    ss,
+    ds,
+}
+
+RegisterValue :: struct #raw_union {
+    using _: struct {
+        low, high: byte,
     },
-    code: u8,
+    full: u16,
 }
 
 Flags :: struct {
@@ -34,14 +45,14 @@ Displacement :: union {
 }
 
 RegisterAccess :: enum {
+    Full,
     Low,
     High,
-    Full,
 }
 
 RegisterId :: struct {
+    idx: int,
     access: RegisterAccess,
-    id: int,
 }
 ImmediateSize :: enum {
     Signed8,
@@ -125,7 +136,7 @@ Instruction :: struct {
     indirect_intersegment: bool,
     // TODO: This is trickier than I thought, it's more than just the one instruction
     // that uses it
-    has_segment: Maybe(Register),
+    has_segment: Maybe(RegisterId),
     has_lock: bool,
     bytes_read: int,
     raw_data: []u8,
@@ -134,7 +145,7 @@ Instruction :: struct {
 
 Cpu :: struct {
     flags: Flags,
-    registers: [12]Register,
+    registers: [12]RegisterValue,
     memory: [dynamic]u8,
     total_bytes_processed: int
 }