decoder seems to be working, need to make a serializer next.

This commit is contained in:
ed
2026-09-09 01:50:20 -04:00
parent 5e6e139cc0
commit 6cca3fbe8b
7 changed files with 116 additions and 67 deletions
+44 -26
View File
@@ -33,7 +33,7 @@ typedef Struct_(X8616_DecodePlex) {
U2 dispatch; U2 dispatch;
U1 plan_idx; U1 plan_idx;
X8616_DecodePlan const* plan; X8616_DecodePlan* plan;
U1 opcode; U1 opcode;
U1 post_opcode; U1 post_opcode;
@@ -80,6 +80,23 @@ typedef Struct_(X8616_DecodePlex) {
FI_ U2 x8616_decode_u2(U1_R bytes) { return C_(U2, bytes[0] | u2_(bytes[1] << 8)); } FI_ U2 x8616_decode_u2(U1_R bytes) { return C_(U2, bytes[0] | u2_(bytes[1] << 8)); }
internal X8616_OpcodePrefix
x8616_decode_opcode(X8616_DecodePlan const* plan, U1 opcode)
{
U1 field_mask = 0;
if (plan->flags & x8616_plan_has_d) field_mask |= u1_(1u << plan->d_shift);
if (plan->flags & x8616_plan_has_w) field_mask |= u1_(1u << plan->w_shift);
if (plan->flags & x8616_plan_has_s) field_mask |= u1_(1u << plan->s_shift);
if (plan->flags & x8616_plan_has_v) field_mask |= u1_(1u << plan->v_shift);
if (plan->flags & x8616_plan_has_z) field_mask |= u1_(1u << plan->z_shift);
if (plan->flags & x8616_plan_has_reg) field_mask |= u1_(0b111u << plan->reg_shift);
if (plan->flags & x8616_plan_has_sr) field_mask |= u1_(0b11u << plan->sr_shift);
if (field_mask == 0) return 0;
U1 stem_mask = u1_(~field_mask);
if (stem_mask == 0) return 0;
return C_(X8616_OpcodePrefix, (opcode & stem_mask) >> count_trailing_zeros_u4(stem_mask));
}
internal void internal void
x8616_decode_apply_prefix(X8616_DecodePlex_R plex, X8616_DecodePlan_R plan, U1 opcode) { x8616_decode_apply_prefix(X8616_DecodePlex_R plex, X8616_DecodePlan_R plan, U1 opcode) {
plex->prefixes.count += 1; plex->prefixes.count += 1;
@@ -107,7 +124,7 @@ x8616_decode_one_plex(X8616_DecodePlex* plex)
U2 prefix_dispatch = x8616_decode_dispatch[prefix_opcode]; U2 prefix_dispatch = x8616_decode_dispatch[prefix_opcode];
if (prefix_dispatch == false || (prefix_dispatch & X8616_DECODE_AUX_BIT)) break; if (prefix_dispatch == false || (prefix_dispatch & X8616_DECODE_AUX_BIT)) break;
X8616_DecodePlan const* prefix_plan = x8616_decode_plans + prefix_dispatch; X8616_DecodePlan_R prefix_plan = x8616_decode_plans + prefix_dispatch;
if ((prefix_plan->flags & x8616_plan_is_prefix) == false) break; if ((prefix_plan->flags & x8616_plan_is_prefix) == false) break;
x8616_decode_apply_prefix(plex, prefix_plan, prefix_opcode); x8616_decode_apply_prefix(plex, prefix_plan, prefix_opcode);
@@ -228,7 +245,6 @@ x8616_decode_one_plex(X8616_DecodePlex* plex)
plex->instruction.prefixes = plex->prefixes; plex->instruction.prefixes = plex->prefixes;
plex->instruction.op = x8616_op_invalid; plex->instruction.op = x8616_op_invalid;
plex->instruction.decode_flags = (plex->encoding_invalid ? x8616_decode_invalid : 0) | (plex->classification_truncated ? x8616_decode_truncated : 0); plex->instruction.decode_flags = (plex->encoding_invalid ? x8616_decode_invalid : 0) | (plex->classification_truncated ? x8616_decode_truncated : 0);
plex->instruction.opcode = plex->opcode;
plex->instruction.size = u1_(prefix_at + invalid_body_size); plex->instruction.size = u1_(prefix_at + invalid_body_size);
plex->instruction.size_required = plex->instruction.size; plex->instruction.size_required = plex->instruction.size;
return plex->instruction.size; return plex->instruction.size;
@@ -303,39 +319,39 @@ x8616_decode_one_plex(X8616_DecodePlex* plex)
X8616_DecodedOperand* source = plex->operand_source; X8616_DecodedOperand* source = plex->operand_source;
source[x8616_operand_reg_modrm].flags = x8616_decoded_operand_register; source[x8616_operand_reg_modrm].flags = x8616_decoded_operand_register;
source[x8616_operand_reg_modrm].width = plex->width; source[x8616_operand_reg_modrm].width = plex->width;
source[x8616_operand_reg_modrm].reg = plex->reg; source[x8616_operand_reg_modrm].reg.r16 = C_(X8616_Reg16, plex->reg);
source[x8616_operand_reg_opcode].flags = x8616_decoded_operand_register; source[x8616_operand_reg_opcode].flags = x8616_decoded_operand_register;
source[x8616_operand_reg_opcode].width = plex->width; source[x8616_operand_reg_opcode].width = plex->width;
source[x8616_operand_reg_opcode].reg = plex->reg_opcode; source[x8616_operand_reg_opcode].reg.r16 = C_(X8616_Reg16, plex->reg_opcode);
source[x8616_operand_segment_modrm].flags = x8616_decoded_operand_segment; source[x8616_operand_segment_modrm].flags = x8616_decoded_operand_segment;
source[x8616_operand_segment_modrm].width = x8616_width_word; source[x8616_operand_segment_modrm].width = x8616_width_word;
source[x8616_operand_segment_modrm].segment = plex->sr_modrm; source[x8616_operand_segment_modrm].segment = C_(X8616_Segment, plex->sr_modrm);
source[x8616_operand_segment_opcode].flags = x8616_decoded_operand_segment; source[x8616_operand_segment_opcode].flags = x8616_decoded_operand_segment;
source[x8616_operand_segment_opcode].width = x8616_width_word; source[x8616_operand_segment_opcode].width = x8616_width_word;
source[x8616_operand_segment_opcode].segment = plex->sr_opcode; source[x8616_operand_segment_opcode].segment = C_(X8616_Segment, plex->sr_opcode);
source[x8616_operand_acc].flags = x8616_decoded_operand_register | x8616_decoded_operand_implicit; source[x8616_operand_acc].flags = x8616_decoded_operand_register | x8616_decoded_operand_implicit;
source[x8616_operand_acc].width = plex->width; source[x8616_operand_acc].width = plex->width;
source[x8616_operand_acc].reg = 0; source[x8616_operand_acc].reg.r16 = x8616_ax;
source[x8616_operand_dx].flags = x8616_decoded_operand_register | x8616_decoded_operand_implicit; source[x8616_operand_dx].flags = x8616_decoded_operand_register | x8616_decoded_operand_implicit;
source[x8616_operand_dx].width = x8616_width_word; source[x8616_operand_dx].width = x8616_width_word;
source[x8616_operand_dx].reg = x8616_dx; source[x8616_operand_dx].reg.r16 = x8616_dx;
source[x8616_operand_rm].width = plex->width; source[x8616_operand_rm].width = plex->width;
if (plex->mod == x8616_mod_reg) { if (plex->mod == x8616_mod_reg) {
source[x8616_operand_rm].flags = x8616_decoded_operand_register; source[x8616_operand_rm].flags = x8616_decoded_operand_register;
source[x8616_operand_rm].reg = plex->rm; source[x8616_operand_rm].reg.r16 = C_(X8616_Reg16, plex->rm);
} }
else { else {
source[x8616_operand_rm].flags = x8616_decoded_operand_memory; source[x8616_operand_rm].flags = x8616_decoded_operand_memory;
source[x8616_operand_rm].mod = plex->mod; source[x8616_operand_rm].mod = C_(X8616_Mod, plex->mod);
source[x8616_operand_rm].ea = plex->rm; source[x8616_operand_rm].ea = C_(X8616_EA, plex->rm);
source[x8616_operand_rm].displacement = plex->displacement; source[x8616_operand_rm].displacement = plex->displacement;
source[x8616_operand_rm].displacement_bytes = plex->displacement_bytes; source[x8616_operand_rm].displacement_bytes = plex->displacement_bytes;
if (plex->direct_memory) { if (plex->direct_memory) {
@@ -382,7 +398,7 @@ x8616_decode_one_plex(X8616_DecodePlex* plex)
source[x8616_operand_shift_count].width = x8616_width_byte; source[x8616_operand_shift_count].width = x8616_width_byte;
if (plex->v == x8616_v_cl) { if (plex->v == x8616_v_cl) {
source[x8616_operand_shift_count].flags |= x8616_decoded_operand_register; source[x8616_operand_shift_count].flags |= x8616_decoded_operand_register;
source[x8616_operand_shift_count].reg = x8616_cl; source[x8616_operand_shift_count].reg.r8 = x8616_cl;
} }
else { else {
source[x8616_operand_shift_count].flags |= x8616_decoded_operand_immediate; source[x8616_operand_shift_count].flags |= x8616_decoded_operand_immediate;
@@ -400,10 +416,11 @@ x8616_decode_one_plex(X8616_DecodePlex* plex)
plex->instruction.decode_flags = (plex->encoding_invalid ? x8616_decode_invalid : 0) | (plex->classification_truncated ? x8616_decode_truncated : 0); plex->instruction.decode_flags = (plex->encoding_invalid ? x8616_decode_invalid : 0) | (plex->classification_truncated ? x8616_decode_truncated : 0);
plex->instruction.width = plex->width; plex->instruction.width = plex->width;
plex->instruction.prefixes = plex->prefixes; plex->instruction.prefixes = plex->prefixes;
plex->instruction.operand_count = plan->operand_count; plex->instruction.operand_count = plan->operand_count;
plex->instruction.opcode = plex->opcode; plex->instruction.opcode = x8616_decode_opcode(plan, plex->opcode);
plex->instruction.mod_rm = plex->mod_rm; plex->instruction.d = C_(X8616_Direction, (plan->flags & x8616_plan_has_d) ? plex->d : 0);
plex->instruction.has_mod_rm = (plan->flags & x8616_plan_has_modrm) != 0; plex->instruction.w = C_(X8616_Width, (plan->flags & x8616_plan_has_w) ? plex->w : 0);
plex->instruction.has_mod_rm = (plan->flags & x8616_plan_has_modrm) != 0;
U4 total_required = prefix_at + plex->body_required; U4 total_required = prefix_at + plex->body_required;
U4 total_available = plex->source_size; U4 total_available = plex->source_size;
@@ -414,6 +431,7 @@ x8616_decode_one_plex(X8616_DecodePlex* plex)
return total_consumed; return total_consumed;
} }
#define x8616_decode_(...) x8616_decode((X8616_DecodeRequest){__VA_ARGS__})
X8616_DecodeInfo x8616_decode(X8616_DecodeRequest request) X8616_DecodeInfo x8616_decode(X8616_DecodeRequest request)
{ {
X8616_DecodeInfo result = {0}; X8616_DecodeInfo result = {0};
+34 -28
View File
@@ -12,25 +12,30 @@ enum {
}; };
typedef Enum_(U2, X8616_DecodedOperandFlags) { typedef Enum_(U2, X8616_DecodedOperandFlags) {
x8616_decoded_operand_none = 0b0000000000000000, Bit_(x8616_decoded_operand_none, 0),
x8616_decoded_operand_register = 0b0000000000000001, Bit_(x8616_decoded_operand_register, 1),
x8616_decoded_operand_segment = 0b0000000000000010, Bit_(x8616_decoded_operand_segment, 2),
x8616_decoded_operand_memory = 0b0000000000000100, Bit_(x8616_decoded_operand_memory, 3),
x8616_decoded_operand_immediate = 0b0000000000001000, Bit_(x8616_decoded_operand_immediate, 4),
x8616_decoded_operand_relative = 0b0000000000010000, Bit_(x8616_decoded_operand_relative, 5),
x8616_decoded_operand_far_ptr = 0b0000000000100000, Bit_(x8616_decoded_operand_far_ptr, 6),
x8616_decoded_operand_direct = 0b0000000001000000, Bit_(x8616_decoded_operand_direct, 7),
x8616_decoded_operand_sign_extended = 0b0000000010000000, Bit_(x8616_decoded_operand_sign_extended, 8),
x8616_decoded_operand_implicit = 0b0000000100000000, Bit_(x8616_decoded_operand_implicit, 9),
};
typedef Union_(X8616_DecodedReg) {
X8616_Reg8 r8;
X8616_Reg16 r16;
}; };
typedef Struct_(X8616_DecodedOperand) { typedef Struct_(X8616_DecodedOperand) {
X8616_DecodedOperandFlags flags; X8616_DecodedOperandFlags flags;
X8616_WidthMode width; X8616_WidthMode width;
U1 reg; X8616_DecodedReg reg;
U1 segment; X8616_Segment segment;
U1 mod; X8616_Mod mod;
U1 ea; X8616_EA ea;
S2 displacement; S2 displacement;
U2 address; U2 address;
U2 immediate; U2 immediate;
@@ -64,26 +69,27 @@ typedef Struct_(X8616_DecodedInstruction) {
X8616_DecodedPrefixes prefixes; X8616_DecodedPrefixes prefixes;
X8616_DecodedOperand operands[2]; X8616_DecodedOperand operands[2];
U1 operand_count; U1 operand_count;
U1 opcode; X8616_OpcodePrefix opcode;
U1 mod_rm; X8616_Direction d;
X8616_Width w;
B1 has_mod_rm; B1 has_mod_rm;
U1 size; U1 size;
U1 size_required; U1 size_required;
}; };
typedef Enum_(U2, X8616_DecodePlanFlags) { typedef Enum_(U2, X8616_DecodePlanFlags) {
x8616_plan_none = 0b0000000000000000, Bit_(x8616_plan_none, 0),
x8616_plan_has_modrm = 0b0000000000000001, Bit_(x8616_plan_has_modrm, 1),
x8616_plan_has_post_opcode = 0b0000000000000010, Bit_(x8616_plan_has_post_opcode, 2),
x8616_plan_has_d = 0b0000000000000100, Bit_(x8616_plan_has_d, 3),
x8616_plan_has_w = 0b0000000000001000, Bit_(x8616_plan_has_w, 4),
x8616_plan_has_s = 0b0000000000010000, Bit_(x8616_plan_has_s, 5),
x8616_plan_has_v = 0b0000000000100000, Bit_(x8616_plan_has_v, 6),
x8616_plan_has_z = 0b0000000001000000, Bit_(x8616_plan_has_z, 7),
x8616_plan_has_reg = 0b0000000010000000, Bit_(x8616_plan_has_reg, 8),
x8616_plan_has_sr = 0b0000000100000000, Bit_(x8616_plan_has_sr, 9),
x8616_plan_uses_rm = 0b0000001000000000, Bit_(x8616_plan_uses_rm, 10),
x8616_plan_is_prefix = 0b0000010000000000, Bit_(x8616_plan_is_prefix, 11),
}; };
typedef Enum_(U1, X8616_DecodePayload) { typedef Enum_(U1, X8616_DecodePayload) {
+4 -3
View File
@@ -10,6 +10,7 @@ Standard: c23
#pragma clang diagnostic ignored "-Wswitch" #pragma clang diagnostic ignored "-Wswitch"
#pragma clang diagnostic ignored "-Wuninitialized" #pragma clang diagnostic ignored "-Wuninitialized"
#pragma clang diagnostic ignored "-Wmicrosoft-enum-forward-reference" #pragma clang diagnostic ignored "-Wmicrosoft-enum-forward-reference"
#pragma clang diagnostic ignored "-Wmicrosoft-fixed-enum"
// #pragma comment(lib, "Advapi32.lib") // #pragma comment(lib, "Advapi32.lib")
// #pragma comment(lib, "gdi32.lib") // #pragma comment(lib, "gdi32.lib")
// #pragma comment(lib, "Kernel32.lib") // #pragma comment(lib, "Kernel32.lib")
@@ -122,7 +123,7 @@ Standard: c23
#define Array_expand(type,len) type Array_sym(type, len)[len]; typedef PtrSet_(Array_sym(type, len)) #define Array_expand(type,len) type Array_sym(type, len)[len]; typedef PtrSet_(Array_sym(type, len))
#define Array_(type,len) Array_expand(type,len) #define Array_(type,len) Array_expand(type,len)
#define Bit_(id,b) id = (1 << b), tmpl(id,pos) = b #define Bit_(id,b) id = (1 << b), tmpl(id,pos) = b
#define Enum_(underlying_type, symbol) underlying_type TSet_(symbol); enum symbol #define Enum_(underlying_type, symbol) enum symbol : underlying_type TSet_(symbol); enum symbol : underlying_type
#define Proc_(symbol) symbol #define Proc_(symbol) symbol
#define Relative_(symbol) // Does nothing but annotate that a symbol is associated with another. #define Relative_(symbol) // Does nothing but annotate that a symbol is associated with another.
#define Struct_(symbol) struct symbol TSet_(symbol); struct symbol #define Struct_(symbol) struct symbol TSet_(symbol); struct symbol
@@ -277,8 +278,8 @@ enum {
Bitmask_10 = 0x000003ff, Bitmask_10 = 0x000003ff,
}; };
typedef Enum_(U4,WeekDay){ WeekDay_Sun, WeekDay_Mon, WeekDay_Tue, WeekDay_Wed, WeekDay_Thu, WeekDay_Fri, WeekDay_Sat, WeekDay_Num, }; typedef Enum_(U4, WeekDay) { WeekDay_Sun, WeekDay_Mon, WeekDay_Tue, WeekDay_Wed, WeekDay_Thu, WeekDay_Fri, WeekDay_Sat, WeekDay_Num, };
typedef Enum_(U4,Month) { Month_Jan, Month_Feb, Month_Mar, Month_Apr, Month_May, Month_Jun, Month_Jul, Month_Aug, Month_Sep, Month_Oct, Month_Nov, Month_Dec, Month_Num, }; typedef Enum_(U4, Month) { Month_Jan, Month_Feb, Month_Mar, Month_Apr, Month_May, Month_Jun, Month_Jul, Month_Aug, Month_Sep, Month_Oct, Month_Nov, Month_Dec, Month_Num, };
typedef U8 DenseTime; typedef U8 DenseTime;
+1 -1
View File
@@ -13,7 +13,7 @@
CLANG_OPTIMIZE_DISABLE CLANG_OPTIMIZE_DISABLE
// Most of this referenced from the RAD Debugger codebase. // Most of this referenced from the RAD Debugger codebase.
typedef Enum_(U4,AccessFlags) { typedef Enum_(U4, AccessFlags) {
Bit_(AccessFlag_Read, 0), Bit_(AccessFlag_Read, 0),
Bit_(AccessFlag_Write, 1), Bit_(AccessFlag_Write, 1),
Bit_(AccessFlag_Execute, 2), Bit_(AccessFlag_Execute, 2),
+29 -5
View File
@@ -1,4 +1,5 @@
#include "duffle/dsl.h" #include "duffle/dsl.h"
#include "duffle/asm.h"
#include "duffle/analysis.h" #include "duffle/analysis.h"
#include "duffle/math.h" #include "duffle/math.h"
#include "duffle/memory.h" #include "duffle/memory.h"
@@ -10,6 +11,15 @@
#include "duffle/win32.h" #include "duffle/win32.h"
#include "8086/encoder.h"
#include "8086/encoder_table.h"
#include "8086/info.h"
#include "8086/info_render.h"
#include "8086/decoder.h"
#include "8086/gen/decoder_table.h"
#include "8086/decoder.c"
// #include "8086/decoder.h" // #include "8086/decoder.h"
typedef Struct_(U2_HL) { U1 Low; U1 High; }; typedef Struct_(U2_HL) { U1 Low; U1 High; };
@@ -56,22 +66,23 @@ I_ Str8 binary_as_str8(Slice_U1 data, FArena* str8_mem) { Str8 result = {0};
jret: return result; jret: return result;
} }
enum { enum {
Scratchpad_Len = kilo(1), Scratchpad_Len = kilo(16),
FileRam_Len = kilo(16), FileRam_Len = kilo(16),
}; };
typedef Struct_(SMemory) { typedef Struct_(SMemory) {
U1 Scratchpad [Scratchpad_Len]; U1 Scratchpad [Scratchpad_Len];
U1 FileRam [FileRam_Len]; U1 FileRam [FileRam_Len];
U1 decode_mem[kilo(64)];
}; };
global SMemory smem; global SMemory smem;
#define path_course_content "./course_content/perfaware/" #define path_course_content "./course_content/perfaware/"
#define path_part1 path_course_content "part1/" #define path_part1 path_course_content "part1/"
CLANG_OPTIMIZE_DISABLE typedef Slice_(X8616_DecodedInstruction);
int main() int main()
{ {
FArena scratch = farena_make(slice_ut_arr(smem.Scratchpad)); FArena scratch = farena_make(slice_ut_arr(smem.Scratchpad));
@@ -80,8 +91,21 @@ int main()
Str8 path_listing_0037_single_register_mov = slit8(path_part1 "listing_0037_single_register_mov"); Str8 path_listing_0037_single_register_mov = slit8(path_part1 "listing_0037_single_register_mov");
Str8 path_listing_0037_single_register_mov_asm = slit8(path_part1 "listing_0037_single_register_mov.asm"); Str8 path_listing_0037_single_register_mov_asm = slit8(path_part1 "listing_0037_single_register_mov.asm");
Slice_U1 data = data_from_file_path(& file_arena, path_listing_0037_single_register_mov, & scratch); Slice_U1 data = data_from_file_path(& file_arena, path_listing_0037_single_register_mov, & scratch);
farena_reset(& scratch);
FArena decode_arena = farena_make(slice_ut_arr(smem.decode_mem));
Slice_X8616_DecodedInstruction decoded = farena_push_array(& decode_arena, X8616_DecodedInstruction, data.len);
X8616_DecodeInfo info = x8616_decode_(
.source = data.ptr,
.source_size = data.len,
.instructions = decoded.ptr,
.instruction_capacity = decoded.len,
.info_arena = & decode_arena,
);
assert(info.instruction_count > 0);
// Str8 infO_report = x8616_info_render(slice_ut_arr(smem.Scratchpad), info.msgs.first,);
ms_exit_process(0); ms_exit_process(0);
return 0; return 0;
} }
CLANG_OPTIMIZE_ENABLE
+2 -2
View File
@@ -3,12 +3,12 @@
target: target:
{ {
executable: "build/sim_8086.exe" executable: "build/sim_8086.exe"
working_directory: "" working_directory: "../perfaware"
label: "Sim 8086" label: "Sim 8086"
enabled: 1
} }
target: target:
{ {
executable: "build/decoder_table_generator.meta.exe" executable: "build/decoder_table_generator.meta.exe"
working_directory: "../perfaware" working_directory: "../perfaware"
enabled: 1
} }
+2 -2
View File
@@ -391,7 +391,7 @@ function build-8086_decoder_table {
dump-disassembly $module_c $exe dump-disassembly $module_c $exe
} }
build-8086_decoder_table # build-8086_decoder_table
function build-part_1 { function build-part_1 {
# The base lib uses subdir-prefixed includes (e.g. "duffle/dsl.h"), # The base lib uses subdir-prefixed includes (e.g. "duffle/dsl.h"),
@@ -418,4 +418,4 @@ function build-part_1 {
dump-disassembly $module_c $exe dump-disassembly $module_c $exe
} }
# build-part_1 build-part_1