perfaware/part1: Support listing 0047

part1/build.bat

@@ -85,3 +85,10 @@ copy /Y %script_dir%\%listing_0046%.txt %build_dir% 1>NUL
 set build_dir_listing_0046=%build_dir%\%listing_0046%
 %build_dir%\sim8086.exe --exec %build_dir_listing_0046% > %build_dir_listing_0046%_disassembled.txt
 fc /N %build_dir_listing_0046%.txt %build_dir_listing_0046%_disassembled.txt || exit /b 1
+
+set listing_0047=listing_0047_challenge_flags
+copy /Y %script_dir%\%listing_0047% %build_dir% 1>NUL
+copy /Y %script_dir%\%listing_0047%.txt %build_dir% 1>NUL
+set build_dir_listing_0047=%build_dir%\%listing_0047%
+%build_dir%\sim8086.exe --exec %build_dir_listing_0047% > %build_dir_listing_0047%_disassembled.txt
+fc /N %build_dir_listing_0047%.txt %build_dir_listing_0047%_disassembled.txt || exit /b 1

part1/listing_0047_challenge_flags.asm

@@ -0,0 +1,36 @@
+; ========================================================================
+;
+; (C) Copyright 2023 by Molly Rocket, Inc., All Rights Reserved.
+;
+; This software is provided 'as-is', without any express or implied
+; warranty. In no event will the authors be held liable for any damages
+; arising from the use of this software.
+;
+; Please see https://computerenhance.com for further information
+;
+; ========================================================================
+
+; ========================================================================
+; LISTING 47
+; ========================================================================
+
+bits 16
+
+add bx, 30000
+add bx, 10000
+sub bx, 5000
+sub bx, 5000
+
+mov bx, 1
+mov cx, 100
+add bx, cx
+
+mov dx, 10
+sub cx, dx
+
+add bx, 40000
+add cx, -90
+
+mov sp, 99
+mov bp, 98
+cmp bp, sp

part1/listing_0047_challenge_flags.txt

@@ -0,0 +1,23 @@
+--- test\listing_0047_challenge_flags execution ---
+add bx, 30000 ; bx:0x0->0x7530 flags:->P 
+add bx, 10000 ; bx:0x7530->0x9c40 flags:P->SO 
+sub bx, 5000 ; bx:0x9c40->0x88b8 flags:SO->PAS 
+sub bx, 5000 ; bx:0x88b8->0x7530 flags:PAS->PO 
+mov bx, 1 ; bx:0x7530->0x1 
+mov cx, 100 ; cx:0x0->0x64 
+add bx, cx ; bx:0x1->0x65 flags:PO->P 
+mov dx, 10 ; dx:0x0->0xa 
+sub cx, dx ; cx:0x64->0x5a flags:P->PA 
+add bx, 40000 ; bx:0x65->0x9ca5 flags:PA->PS 
+add cx, -90 ; cx:0x5a->0x0 flags:PS->CPAZ 
+mov sp, 99 ; sp:0x0->0x63 
+mov bp, 98 ; bp:0x0->0x62 
+cmp bp, sp ; flags:CPAZ->CPAS 
+
+Final registers:
+      bx: 0x9ca5 (40101)
+      dx: 0x000a (10)
+      sp: 0x0063 (99)
+      bp: 0x0062 (98)
+   flags: CPAS
+

part1/sim8086.c

@@ -1,3 +1,13 @@
+#define WIN32_MEAN_AND_LEAN
+#define NOMINMAX
+#include <Windows.h>
+#include <immintrin.h>
+
+#include <stdint.h>
+#include <stdio.h>
+#include <stdarg.h>
+#include <stdbool.h>
+
 #include "sim8086_stdlib.h"
 #include "sim8086.h"
 
@@ -12,15 +22,20 @@ typedef enum S86_RegisterByte {
 
 typedef union S86_Register16 {
     uint16_t word;
-    struct { uint8_t lo; uint8_t hi; } byte;
     uint8_t bytes[S86_RegisterByte_Count];
 } S86_Register16;
 
+typedef struct S86_RegisterFileFlags {
+    bool carry;
+    bool zero;
+    bool sign;
+    bool overflow;
+    bool parity;
+    bool auxiliary_carry;
+} S86_RegisterFileFlags;
+
 typedef struct S86_RegisterFile {
-    bool zero_flag;
+    S86_RegisterFileFlags flags;
-    bool sign_flag;
-    bool parity_flag;
-    bool overflow_flag;
 
     S86_Register16 ax;
     S86_Register16 bx;
@@ -38,6 +53,17 @@ typedef struct S86_RegisterFile {
     S86_Register16 ds;
 } S86_RegisterFile;
 
+bool S86_RegisterFileFlagsEq(S86_RegisterFileFlags lhs, S86_RegisterFileFlags rhs)
+{
+    bool result = lhs.carry           == rhs.carry    &&
+                  lhs.zero            == rhs.zero     &&
+                  lhs.sign            == rhs.sign     &&
+                  lhs.overflow        == rhs.overflow &&
+                  lhs.parity          == rhs.parity   &&
+                  lhs.auxiliary_carry == rhs.auxiliary_carry;
+    return result;
+}
+
 S86_Str8 S86_MnemonicStr8(S86_Mnemonic type)
 {
     S86_Str8 result = {0};
@@ -255,7 +281,11 @@ void S86_PrintOpcodeMnemonicOp(S86_Opcode opcode, bool src)
                      opcode.displacement > 0 ? '+' : '-',
                      opcode.displacement > 0 ? opcode.displacement : -opcode.displacement);
     } else if (mnemonic_op == S86_MnemonicOp_Immediate) {
-        S86_PrintFmt("%u", opcode.immediate);
+        if (opcode.immediate_is_8bit) {
+            S86_PrintFmt("%d", (int8_t)opcode.immediate);
+        } else {
+            S86_PrintFmt("%u", (uint16_t)opcode.immediate);
+        }
     } else if (mnemonic_op == S86_MnemonicOp_DirectInterSegment) {
         uint16_t left  = (uint32_t)opcode.displacement >> 16;
         uint16_t right = (uint32_t)opcode.displacement  & 0xFFFF;
@@ -600,8 +630,14 @@ S86_Opcode S86_DecodeOpcode(S86_BufferIterator *buffer_it,
                      op_decode_type == S86_OpDecodeType_TESTImmediateAndRegOrMem ||
                      op_decode_type == S86_OpDecodeType_ORImmediateToRegOrMem ||
                      op_decode_type == S86_OpDecodeType_XORImmediateToRegOrMem) && s) {
-                    // NOTE: Sign extend 8 bit, since we store into a
+                    // NOTE: Sign extend 8 bit to 16 bit
-                    // int32_t in opcode this is done for free for us.
+                    uint16_t sign_mask = 0b1000'0000;
+                    uint8_t data_u8    = S86_CAST(uint8_t)data;
+                    if (sign_mask & data_u8) {
+                        data = (uint16_t)((uint8_t)~data_u8 + 1); // Convert back to 8 bit unsigned
+                        data = ~data + 1;                         // Convert to 16bit signed
+                        result.immediate_is_8bit = true;
+                    }
                 } else {
                     uint8_t data_hi = S86_BufferIteratorNextByte(buffer_it);
                     data |= (uint16_t)(data_hi) << 8;
@@ -612,7 +648,15 @@ S86_Opcode S86_DecodeOpcode(S86_BufferIterator *buffer_it,
                 S86_ASSERT(mod != 0b11); // NOTE: Op is IMM->Reg, register-to-register not permitted
             }
 
+            // NOTE: Sign extend 16bit to 32bit
+            uint16_t sign_mask16 = 0b1000'0000'0000'0000;
+            if (data & sign_mask16) {
+                uint32_t data_no_sign_bit = (uint32_t)((uint16_t)~data + 1); // Convert back to 16 bit unsigned
+                result.immediate = ~data_no_sign_bit + 1;                    // Convert to signed 32bit
+            } else {
                 result.immediate = data;
+            }
+
             result.src = S86_MnemonicOp_Immediate;
             if (op_decode_type == S86_OpDecodeType_MOVImmediateToRegOrMem)
                 result.wide_prefix = S86_WidePrefix_Src;
@@ -1207,8 +1251,7 @@ int main(int argc, char **argv)
             continue;
         }
 
-        bool prev_sign_flag = register_file.sign_flag;
+        S86_RegisterFileFlags prev_flags = register_file.flags;
-        bool prev_zero_flag = register_file.zero_flag;
         switch (opcode.mnemonic) {
             case S86_Mnemonic_PUSH:          /*FALLTHRU*/
             case S86_Mnemonic_POP:           /*FALLTHRU*/
@@ -1329,105 +1372,30 @@ int main(int argc, char **argv)
                 S86_PrintFmt(" ; %.*s:0x%x->0x%x ", S86_STR8_FMT(dest_reg16), prev_dest, dest_map->reg->word);
             } break;
 
-            case S86_Mnemonic_ADD: {
+            case S86_Mnemonic_ADD: /*FALLTHRU*/
-                S86_MnemonicOpToRegisterFileMap const *dest_map = NULL;
+            case S86_Mnemonic_SUB: /*FALLTHRU*/
-                for (size_t index = 0; !dest_map && index < S86_ARRAY_UCOUNT(mnemonic_op_to_register_file_map); index++) {
-                    S86_MnemonicOpToRegisterFileMap const *item = mnemonic_op_to_register_file_map + index;
-                    if (item->mnemonic_op == opcode.dest)
-                        dest_map = item;
-                }
-                S86_ASSERT(dest_map);
-
-                uint16_t prev_dest = dest_map->reg->word;
-                bool byte_op       = opcode.dest >= S86_MnemonicOp_AL && opcode.dest <= S86_MnemonicOp_BH;
-                if (opcode.src == S86_MnemonicOp_Immediate) {
-                    if (byte_op) {
-                        S86_ASSERT(opcode.immediate < S86_CAST(uint8_t)-1);
-                        dest_map->reg->bytes[dest_map->byte] += S86_CAST(uint8_t)opcode.immediate;
-                    } else {
-                        S86_ASSERT(opcode.immediate < S86_CAST(uint16_t)-1);
-                        dest_map->reg->word += S86_CAST(uint16_t)opcode.immediate;
-                    }
-                } else {
-                    S86_MnemonicOpToRegisterFileMap const *src_map = NULL;
-                    for (size_t index = 0; !src_map && index < S86_ARRAY_UCOUNT(mnemonic_op_to_register_file_map); index++) {
-                        S86_MnemonicOpToRegisterFileMap const *item = mnemonic_op_to_register_file_map + index;
-                        if (item->mnemonic_op == opcode.src)
-                            src_map = item;
-                    }
-
-                    if (byte_op)
-                        dest_map->reg->bytes[dest_map->byte] += src_map->reg->bytes[src_map->byte];
-                    else
-                        dest_map->reg->word += src_map->reg->word;
-                }
-
-                S86_Str8 dest_reg16 = S86_MnemonicOpStr8(dest_map->mnemonic_op_reg16);
-
-                register_file.sign_flag = dest_map->reg->word & (byte_op ? 0b0000'0000'1000'0000 : 0b1000'0000'0000'0000);
-                S86_PrintFmt(" ; %.*s:0x%x->0x%x ", S86_STR8_FMT(dest_reg16), prev_dest, dest_map->reg->word);
-            } break;
-
-            case S86_Mnemonic_SUB: {
-                S86_MnemonicOpToRegisterFileMap const *dest_map = NULL;
-                for (size_t index = 0; !dest_map && index < S86_ARRAY_UCOUNT(mnemonic_op_to_register_file_map); index++) {
-                    S86_MnemonicOpToRegisterFileMap const *item = mnemonic_op_to_register_file_map + index;
-                    if (item->mnemonic_op == opcode.dest)
-                        dest_map = item;
-                }
-                S86_ASSERT(dest_map);
-
-                uint16_t prev_dest = dest_map->reg->word;
-                bool byte_op       = opcode.dest >= S86_MnemonicOp_AL && opcode.dest <= S86_MnemonicOp_BH;
-                if (opcode.src == S86_MnemonicOp_Immediate) {
-                    if (byte_op) {
-                        S86_ASSERT(opcode.immediate < S86_CAST(uint8_t)-1);
-                        dest_map->reg->bytes[dest_map->byte] -= S86_CAST(uint8_t)opcode.immediate;
-                    } else {
-                        S86_ASSERT(opcode.immediate < S86_CAST(uint16_t)-1);
-                        dest_map->reg->word -= S86_CAST(uint16_t)opcode.immediate;
-                    }
-                } else {
-                    S86_MnemonicOpToRegisterFileMap const *src_map = NULL;
-                    for (size_t index = 0; !src_map && index < S86_ARRAY_UCOUNT(mnemonic_op_to_register_file_map); index++) {
-                        S86_MnemonicOpToRegisterFileMap const *item = mnemonic_op_to_register_file_map + index;
-                        if (item->mnemonic_op == opcode.src)
-                            src_map = item;
-                    }
-
-                    if (byte_op)
-                        dest_map->reg->bytes[dest_map->byte] -= src_map->reg->bytes[src_map->byte];
-                    else
-                        dest_map->reg->word -= src_map->reg->word;
-                }
-
-                S86_Str8 dest_reg16 = S86_MnemonicOpStr8(dest_map->mnemonic_op_reg16);
-
-                register_file.zero_flag = byte_op ? dest_map->reg->bytes[dest_map->byte] == 0 : dest_map->reg->word == 0;
-                register_file.sign_flag = dest_map->reg->word & (byte_op ? 0b0000'0000'1000'0000 : 0b1000'0000'0000'0000);
-                S86_PrintFmt(" ; %.*s:0x%x->0x%x ", S86_STR8_FMT(dest_reg16), prev_dest, dest_map->reg->word);
-            } break;
-
             case S86_Mnemonic_CMP: {
-                S86_MnemonicOpToRegisterFileMap const *dest_map = NULL;
+                S86_MnemonicOpToRegisterFileMap *dest_map = NULL;
                 for (size_t index = 0; !dest_map && index < S86_ARRAY_UCOUNT(mnemonic_op_to_register_file_map); index++) {
-                    S86_MnemonicOpToRegisterFileMap const *item = mnemonic_op_to_register_file_map + index;
+                    S86_MnemonicOpToRegisterFileMap *item = mnemonic_op_to_register_file_map + index;
                     if (item->mnemonic_op == opcode.dest)
                         dest_map = item;
                 }
                 S86_ASSERT(dest_map);
 
-                S86_Register16 dest_copy = *dest_map->reg;
+                bool subtract            = opcode.mnemonic != S86_Mnemonic_ADD;
-                S86_Register16 *dest     = &dest_copy;
+                S86_Register16 dest      = *dest_map->reg;
-
+                S86_Register16 prev_dest = dest;
                 bool byte_op             = opcode.dest >= S86_MnemonicOp_AL && opcode.dest <= S86_MnemonicOp_BH;
+
+                uint16_t src = 0;
                 if (opcode.src == S86_MnemonicOp_Immediate) {
                     if (byte_op) {
                         S86_ASSERT(opcode.immediate < S86_CAST(uint8_t)-1);
-                        dest->bytes[dest_map->byte] -= S86_CAST(uint8_t)opcode.immediate;
+                        src = S86_CAST(uint8_t)opcode.immediate;
                     } else {
                         S86_ASSERT(opcode.immediate < S86_CAST(uint16_t)-1);
-                        dest->word -= S86_CAST(uint16_t)opcode.immediate;
+                        src = S86_CAST(uint16_t)opcode.immediate;
                     }
                 } else {
                     S86_MnemonicOpToRegisterFileMap const *src_map = NULL;
@@ -1436,32 +1404,117 @@ int main(int argc, char **argv)
                         if (item->mnemonic_op == opcode.src)
                             src_map = item;
                     }
-
+                    src = byte_op ? src_map->reg->bytes[src_map->byte] : src_map->reg->word;
-                    if (byte_op)
-                        dest->bytes[dest_map->byte] -= src_map->reg->bytes[src_map->byte];
-                    else
-                        dest->word -= src_map->reg->word;
                 }
 
-                register_file.sign_flag = dest_map->reg->word & (byte_op ? 0b0000'0000'1000'0000 : 0b1000'0000'0000'0000);
+                // NOTE: Overflow if the sign masks were initially the same,
+                // but, after the operation the sign masked changed.
+                uint8_t  const sign_mask8  = 0b1000'0000;
+                uint16_t const sign_mask16 = 0b1000'0000'0000'0000;
+                if (byte_op) {
+                    uint8_t src_u8 = S86_CAST(uint8_t)src;
+                    if (subtract)
+                        src_u8 = ~src_u8 + 1;
+
+                    uint8_t dest_u8     = dest.bytes[dest_map->byte];
+                    uint8_t new_dest_u8 = dest_u8 + src_u8;
+
+                    // NOTE: Overflow check
+                    bool initially_matching_sign_masks = (dest_u8 & sign_mask8) == (src_u8 & sign_mask8);
+                    bool sign_masks_changed            = (dest_u8 & sign_mask8) != (new_dest_u8 & sign_mask8);
+                    register_file.flags.overflow       = initially_matching_sign_masks && sign_masks_changed;
+
+                    // NOTE: Carry check
+                    register_file.flags.carry          = subtract ? new_dest_u8 > dest_u8 : new_dest_u8 < dest_u8;
+
+                    // NOTE: Auxiliary carry check
+                    uint8_t     dest_u8_nibble_lo =     dest_u8 & 0b0000'1111 >> 0;
+                    uint8_t     dest_u8_nibble_hi =     dest_u8 & 0b1111'0000 >> 4;
+                    uint8_t new_dest_u8_nibble_lo = new_dest_u8 & 0b0000'1111 >> 0;
+                    uint8_t new_dest_u8_nibble_hi = new_dest_u8 & 0b1111'0000 >> 4;
+                    register_file.flags.auxiliary_carry = subtract ? new_dest_u8_nibble_hi > dest_u8_nibble_hi
+                                                                     : new_dest_u8_nibble_lo < dest_u8_nibble_lo;
+
+                    // NOTE: Sign check
+                    register_file.flags.sign = new_dest_u8 & 0b1000'0000;
+
+                    // NOTE: Update the register
+                    dest.bytes[dest_map->byte] = new_dest_u8;
+                } else {
+                    if (subtract)
+                        src = ~src + 1;
+
+                    S86_Register16 new_dest = {0};
+                    new_dest.word           = dest.word + src;
+
+                    // NOTE: Overflow check
+                    bool initially_matching_sign_masks = (dest.word & sign_mask16) == (src           & sign_mask16);
+                    bool sign_masks_changed            = (dest.word & sign_mask16) != (new_dest.word & sign_mask16);
+                    register_file.flags.overflow       = initially_matching_sign_masks && sign_masks_changed;
+
+                    // NOTE: Auxiliary carry check
+                    uint8_t     dest_lo_nibble_lo =     dest.bytes[S86_RegisterByte_Lo] & 0b0000'1111 >> 0;
+                    uint8_t     dest_lo_nibble_hi =     dest.bytes[S86_RegisterByte_Lo] & 0b1111'0000 >> 4;
+                    uint8_t new_dest_lo_nibble_lo = new_dest.bytes[S86_RegisterByte_Lo] & 0b0000'1111 >> 0;
+                    uint8_t new_dest_lo_nibble_hi = new_dest.bytes[S86_RegisterByte_Lo] & 0b1111'0000 >> 4;
+                    register_file.flags.auxiliary_carry = subtract ? new_dest_lo_nibble_hi > dest_lo_nibble_hi
+                                                                   : new_dest_lo_nibble_lo < dest_lo_nibble_lo;
+
+                    // NOTE: Carry check
+                    register_file.flags.carry = subtract ? new_dest.word > dest.word : new_dest.word < dest.word;
+
+                    // NOTE: Sign check
+                    register_file.flags.sign = new_dest.word & 0b1000'0000'0000'0000;
+
+                    // NOTE: Update the register
+                    dest.word = new_dest.word;
+                }
+
+                int lo_bit_count           = _mm_popcnt_u32(S86_CAST(uint32_t)dest.bytes[S86_RegisterByte_Lo]);
+                register_file.flags.parity = lo_bit_count % 2 == 0;
+                register_file.flags.zero   = false;
+                if (opcode.mnemonic == S86_Mnemonic_ADD || opcode.mnemonic == S86_Mnemonic_SUB)
+                    register_file.flags.zero = byte_op ? dest.bytes[dest_map->byte] == 0 : dest.word == 0;
+
+                if (opcode.mnemonic == S86_Mnemonic_CMP) {
                     S86_PrintFmt(" ; ");
+                } else {
+                    S86_Str8 dest_reg16 = S86_MnemonicOpStr8(dest_map->mnemonic_op_reg16);
+                    S86_PrintFmt(" ; %.*s:0x%x->0x%x ", S86_STR8_FMT(dest_reg16), prev_dest.word, dest.word);
+                    *dest_map->reg = dest;
+                }
             } break;
         }
 
-        if (register_file.sign_flag != prev_sign_flag) {
+        if (!S86_RegisterFileFlagsEq(register_file.flags, prev_flags)) {
-            if (register_file.sign_flag) {
+            S86_PrintFmt("flags:");
-                S86_PrintFmt("flags:->S ");
+            if (prev_flags.carry)
-            } else {
+                S86_PrintFmt("C");
-                S86_PrintFmt("flags:S-> ");
+            if (prev_flags.parity)
-            }
+                S86_PrintFmt("P");
-        }
+            if (prev_flags.auxiliary_carry)
+                S86_PrintFmt("A");
+            if (prev_flags.zero)
+                S86_PrintFmt("Z");
+            if (prev_flags.sign)
+                S86_PrintFmt("S");
+            if (prev_flags.overflow)
+                S86_PrintFmt("O");
 
-        if (register_file.zero_flag != prev_zero_flag) {
+            S86_PrintFmt("->");
-            if (register_file.zero_flag) {
+            if (register_file.flags.carry)
-                S86_PrintFmt("flags:->PZ ");
+                S86_PrintFmt("C");
-            } else {
+            if (register_file.flags.parity)
-                S86_PrintFmt("flags:PZ-> ");
+                S86_PrintFmt("P");
-            }
+            if (register_file.flags.auxiliary_carry)
+                S86_PrintFmt("A");
+            if (register_file.flags.zero)
+                S86_PrintFmt("Z");
+            if (register_file.flags.sign)
+                S86_PrintFmt("S");
+            if (register_file.flags.overflow)
+                S86_PrintFmt("O");
+            S86_PrintFmt(" ");
         }
         S86_Print(S86_STR8("\n"));
     }
@@ -1490,12 +1543,22 @@ int main(int argc, char **argv)
             S86_PrintLnFmt("      ss: 0x%04x (%u)", register_file.ss, register_file.ss);
         if (register_file.ds.word)
             S86_PrintLnFmt("      ds: 0x%04x (%u)", register_file.ds, register_file.ds);
-        if (register_file.zero_flag || register_file.sign_flag) {
+
+        S86_RegisterFileFlags nil_flags = {0};
+        if (!S86_RegisterFileFlagsEq(register_file.flags, nil_flags)) {
             S86_PrintFmt("   flags: ");
-            if (register_file.zero_flag)
+            if (register_file.flags.carry)
-                S86_PrintFmt(" PZ");
+                S86_PrintFmt("C");
-            if (register_file.sign_flag)
+            if (register_file.flags.parity)
+                S86_PrintFmt("P");
+            if (register_file.flags.auxiliary_carry)
+                S86_PrintFmt("A");
+            if (register_file.flags.zero)
+                S86_PrintFmt("Z");
+            if (register_file.flags.sign)
                 S86_PrintFmt("S");
+            if (register_file.flags.overflow)
+                S86_PrintFmt("O");
             S86_Print(S86_STR8("\n"));
         }
         S86_Print(S86_STR8("\n"));

part1/sim8086.h

@@ -333,6 +333,7 @@ typedef struct S86_Opcode {
     S86_MnemonicOp       dest;                     ///< Destination op for the mnemonic
     int32_t              displacement;             ///< Opcode has displacement/data/offset
     int32_t              immediate;                ///< Immediate value when src/dest op is an immediate
+    bool                 immediate_is_8bit;        ///< Immediate was 8bit and sign extended
     S86_MnemonicOp       seg_reg_prefix;           ///< Segment register that should prefix the upcoming instruction
 } S86_Opcode;
 

part1/sim8086_stdlib.h

@@ -1,12 +1,3 @@
-#define WIN32_MEAN_AND_LEAN
-#define NOMINMAX
-#include <Windows.h>
-
-#include <stdint.h>
-#include <stdio.h>
-#include <stdarg.h>
-#include <stdbool.h>
-
 // NOTE: Macros
 // ============================================================================
 #define S86_STRINGIFY2(token) #token