Merge branch 'mach' into bytecode_cleanup

This commit is contained in:
2026-02-12 07:50:09 -06:00
23 changed files with 1867 additions and 4331 deletions

View File

@@ -406,6 +406,8 @@ typedef struct { uint16_t line; uint16_t col; } MachLineEntry;
#define MACH_GET_sJ(i) ((int32_t)((i) & 0xFFFFFF00) >> 8)
typedef enum MachOpcode {
/* === Legacy opcodes (used by existing .mach files) === */
/* Constants & Loading */
MACH_LOADK, /* R(A) = K(Bx) — load from constant pool (ABx) */
MACH_LOADI, /* R(A) = (int16_t)sBx — load small integer (AsBx) */
@@ -416,7 +418,7 @@ typedef enum MachOpcode {
/* Movement */
MACH_MOVE, /* R(A) = R(B) */
/* Arithmetic (ABC) */
/* Generic arithmetic (ABC) — used by legacy .mach */
MACH_ADD, /* R(A) = R(B) + R(C) */
MACH_SUB, /* R(A) = R(B) - R(C) */
MACH_MUL, /* R(A) = R(B) * R(C) */
@@ -427,7 +429,7 @@ typedef enum MachOpcode {
MACH_INC, /* R(A) = R(B) + 1 */
MACH_DEC, /* R(A) = R(B) - 1 */
/* Comparison (ABC) */
/* Generic comparison (ABC) — used by legacy .mach */
MACH_EQ, /* R(A) = (R(B) == R(C)) */
MACH_NEQ, /* R(A) = (R(B) != R(C)) */
MACH_LT, /* R(A) = (R(B) < R(C)) */
@@ -435,7 +437,7 @@ typedef enum MachOpcode {
MACH_GT, /* R(A) = (R(B) > R(C)) */
MACH_GE, /* R(A) = (R(B) >= R(C)) */
/* Logical/Bitwise */
/* Logical/Bitwise — used by legacy .mach */
MACH_LNOT, /* R(A) = !R(B) */
MACH_BNOT, /* R(A) = ~R(B) */
MACH_BAND, /* R(A) = R(B) & R(C) */
@@ -445,7 +447,7 @@ typedef enum MachOpcode {
MACH_SHR, /* R(A) = R(B) >> R(C) */
MACH_USHR, /* R(A) = R(B) >>> R(C) */
/* Property access */
/* Property access — used by legacy .mach */
MACH_GETFIELD, /* R(A) = R(B)[K(C)] — named property */
MACH_SETFIELD, /* R(A)[K(B)] = R(C) — named property */
MACH_GETINDEX, /* R(A) = R(B)[R(C)] — computed property */
@@ -466,12 +468,12 @@ typedef enum MachOpcode {
MACH_JMPFALSE, /* if !R(A): pc += sBx — (iAsBx format) */
MACH_JMPNULL, /* if R(A)==null: pc += sBx */
/* Function calls — Lua-style consecutive registers */
/* Function calls — Lua-style consecutive registers (legacy .mach) */
MACH_CALL, /* Call R(A) with B args R(A+1)..R(A+B), C=0 discard, C=1 keep result in R(A) */
MACH_RETURN, /* Return R(A) */
MACH_RETNIL, /* Return null */
/* Object/array creation */
/* Object/array creation — legacy .mach */
MACH_NEWOBJECT, /* R(A) = {} */
MACH_NEWARRAY, /* R(A) = new array, B = element count in R(A+1)..R(A+B) */
MACH_CLOSURE, /* R(A) = closure(functions[Bx]) (ABx) */
@@ -493,10 +495,112 @@ typedef enum MachOpcode {
MACH_NOP,
/* === New mcode-derived opcodes (1:1 mapping to mcode IR) === */
/* Typed integer arithmetic (ABC) */
MACH_ADD_INT, /* R(A) = R(B) + R(C) — int, overflow → float */
MACH_SUB_INT, /* R(A) = R(B) - R(C) — int */
MACH_MUL_INT, /* R(A) = R(B) * R(C) — int */
MACH_DIV_INT, /* R(A) = R(B) / R(C) — int */
MACH_MOD_INT, /* R(A) = R(B) % R(C) — int */
MACH_NEG_INT, /* R(A) = -R(B) — int (AB) */
/* Typed float arithmetic (ABC) */
MACH_ADD_FLOAT, /* R(A) = R(B) + R(C) — float */
MACH_SUB_FLOAT, /* R(A) = R(B) - R(C) — float */
MACH_MUL_FLOAT, /* R(A) = R(B) * R(C) — float */
MACH_DIV_FLOAT, /* R(A) = R(B) / R(C) — float */
MACH_MOD_FLOAT, /* R(A) = R(B) % R(C) — float */
MACH_NEG_FLOAT, /* R(A) = -R(B) — float (AB) */
/* Text */
MACH_CONCAT, /* R(A) = R(B) ++ R(C) — string concatenation */
/* Typed integer comparisons (ABC) */
MACH_EQ_INT, /* R(A) = (R(B) == R(C)) — int */
MACH_NE_INT, /* R(A) = (R(B) != R(C)) — int */
MACH_LT_INT, /* R(A) = (R(B) < R(C)) — int */
MACH_LE_INT, /* R(A) = (R(B) <= R(C)) — int */
MACH_GT_INT, /* R(A) = (R(B) > R(C)) — int */
MACH_GE_INT, /* R(A) = (R(B) >= R(C)) — int */
/* Typed float comparisons (ABC) */
MACH_EQ_FLOAT, /* R(A) = (R(B) == R(C)) — float */
MACH_NE_FLOAT, /* R(A) = (R(B) != R(C)) — float */
MACH_LT_FLOAT, /* R(A) = (R(B) < R(C)) — float */
MACH_LE_FLOAT, /* R(A) = (R(B) <= R(C)) — float */
MACH_GT_FLOAT, /* R(A) = (R(B) > R(C)) — float */
MACH_GE_FLOAT, /* R(A) = (R(B) >= R(C)) — float */
/* Typed text comparisons (ABC) */
MACH_EQ_TEXT, /* R(A) = (R(B) == R(C)) — text */
MACH_NE_TEXT, /* R(A) = (R(B) != R(C)) — text */
MACH_LT_TEXT, /* R(A) = (R(B) < R(C)) — text */
MACH_LE_TEXT, /* R(A) = (R(B) <= R(C)) — text */
MACH_GT_TEXT, /* R(A) = (R(B) > R(C)) — text */
MACH_GE_TEXT, /* R(A) = (R(B) >= R(C)) — text */
/* Typed bool comparisons (ABC) */
MACH_EQ_BOOL, /* R(A) = (R(B) == R(C)) — bool */
MACH_NE_BOOL, /* R(A) = (R(B) != R(C)) — bool */
/* Special comparisons */
MACH_IS_IDENTICAL, /* R(A) = (R(B) === R(C)) — identity check (ABC) */
/* Type checks (AB) */
MACH_IS_INT, /* R(A) = is_int(R(B)) */
MACH_IS_NUM, /* R(A) = is_num(R(B)) */
MACH_IS_TEXT, /* R(A) = is_text(R(B)) */
MACH_IS_BOOL, /* R(A) = is_bool(R(B)) */
MACH_IS_NULL, /* R(A) = is_null(R(B)) */
MACH_TYPEOF, /* R(A) = typeof(R(B)) */
/* Logical (mcode-style) */
MACH_NOT, /* R(A) = !R(B) — boolean not (AB) */
MACH_AND, /* R(A) = R(B) && R(C) (ABC) */
MACH_OR, /* R(A) = R(B) || R(C) (ABC) */
/* Bitwise (mcode names) */
MACH_BITNOT, /* R(A) = ~R(B) (AB) */
MACH_BITAND, /* R(A) = R(B) & R(C) (ABC) */
MACH_BITOR, /* R(A) = R(B) | R(C) (ABC) */
MACH_BITXOR, /* R(A) = R(B) ^ R(C) (ABC) */
/* Property access (mcode names) */
MACH_LOAD_FIELD, /* R(A) = R(B).K(C) — named property (ABC) */
MACH_STORE_FIELD, /* R(A).K(B) = R(C) — named property (ABC) */
MACH_LOAD_INDEX, /* R(A) = R(B)[R(C)] — integer index (ABC) */
MACH_STORE_INDEX, /* R(A)[R(B)] = R(C) — integer index (ABC) */
MACH_LOAD_DYNAMIC, /* R(A) = R(B)[R(C)] — dynamic key (ABC) */
MACH_STORE_DYNAMIC, /* R(A)[R(B)] = R(C) — dynamic key (ABC) */
/* Object/Array creation (mcode names) */
MACH_NEWRECORD, /* R(A) = {} — new empty record (A only) */
/* Decomposed function calls (mcode-style) */
MACH_FRAME, /* R(A) = frame(R(B), C) — alloc call frame (ABC) */
MACH_SETARG, /* frame R(A)[B] = R(C) — set arg in frame (ABC) */
MACH_INVOKE, /* R(B) = invoke(R(A)) — call frame, result in R(B) (AB) */
MACH_GOFRAME, /* R(A) = goframe(R(B), C) — async frame (ABC) */
MACH_GOINVOKE, /* goinvoke(R(A)) — async invoke, no result (A only) */
/* Control flow */
MACH_JMPNOTNULL, /* if R(A)!=null: pc += sBx (iAsBx) */
/* Error handling */
MACH_DISRUPT, /* trigger disruption (A only) */
/* Variable storage */
MACH_SET_VAR, /* env/global[K(Bx)] = R(A) — store to var (ABx) */
/* Misc */
MACH_IN, /* R(A) = (R(B) in R(C)) — has property (ABC) */
MACH_OP_COUNT
} MachOpcode;
static const char *mach_opcode_names[MACH_OP_COUNT] = {
/* Legacy */
[MACH_LOADK] = "loadk",
[MACH_LOADI] = "loadi",
[MACH_LOADNULL] = "loadnull",
@@ -556,6 +660,70 @@ static const char *mach_opcode_names[MACH_OP_COUNT] = {
[MACH_EQ_TOL] = "eq_tol",
[MACH_NEQ_TOL] = "neq_tol",
[MACH_NOP] = "nop",
/* Mcode-derived */
[MACH_ADD_INT] = "add_int",
[MACH_SUB_INT] = "sub_int",
[MACH_MUL_INT] = "mul_int",
[MACH_DIV_INT] = "div_int",
[MACH_MOD_INT] = "mod_int",
[MACH_NEG_INT] = "neg_int",
[MACH_ADD_FLOAT] = "add_float",
[MACH_SUB_FLOAT] = "sub_float",
[MACH_MUL_FLOAT] = "mul_float",
[MACH_DIV_FLOAT] = "div_float",
[MACH_MOD_FLOAT] = "mod_float",
[MACH_NEG_FLOAT] = "neg_float",
[MACH_CONCAT] = "concat",
[MACH_EQ_INT] = "eq_int",
[MACH_NE_INT] = "ne_int",
[MACH_LT_INT] = "lt_int",
[MACH_LE_INT] = "le_int",
[MACH_GT_INT] = "gt_int",
[MACH_GE_INT] = "ge_int",
[MACH_EQ_FLOAT] = "eq_float",
[MACH_NE_FLOAT] = "ne_float",
[MACH_LT_FLOAT] = "lt_float",
[MACH_LE_FLOAT] = "le_float",
[MACH_GT_FLOAT] = "gt_float",
[MACH_GE_FLOAT] = "ge_float",
[MACH_EQ_TEXT] = "eq_text",
[MACH_NE_TEXT] = "ne_text",
[MACH_LT_TEXT] = "lt_text",
[MACH_LE_TEXT] = "le_text",
[MACH_GT_TEXT] = "gt_text",
[MACH_GE_TEXT] = "ge_text",
[MACH_EQ_BOOL] = "eq_bool",
[MACH_NE_BOOL] = "ne_bool",
[MACH_IS_IDENTICAL] = "is_identical",
[MACH_IS_INT] = "is_int",
[MACH_IS_NUM] = "is_num",
[MACH_IS_TEXT] = "is_text",
[MACH_IS_BOOL] = "is_bool",
[MACH_IS_NULL] = "is_null",
[MACH_TYPEOF] = "typeof",
[MACH_NOT] = "not",
[MACH_AND] = "and",
[MACH_OR] = "or",
[MACH_BITNOT] = "bitnot",
[MACH_BITAND] = "bitand",
[MACH_BITOR] = "bitor",
[MACH_BITXOR] = "bitxor",
[MACH_LOAD_FIELD] = "load_field",
[MACH_STORE_FIELD] = "store_field",
[MACH_LOAD_INDEX] = "load_index",
[MACH_STORE_INDEX] = "store_index",
[MACH_LOAD_DYNAMIC] = "load_dynamic",
[MACH_STORE_DYNAMIC] = "store_dynamic",
[MACH_NEWRECORD] = "newrecord",
[MACH_FRAME] = "frame",
[MACH_SETARG] = "setarg",
[MACH_INVOKE] = "invoke",
[MACH_GOFRAME] = "goframe",
[MACH_GOINVOKE] = "goinvoke",
[MACH_JMPNOTNULL] = "jmpnotnull",
[MACH_DISRUPT] = "disrupt",
[MACH_SET_VAR] = "set_var",
[MACH_IN] = "in",
};
/* Compiled register-based code (off-heap, never GC'd).
@@ -586,28 +754,6 @@ typedef struct JSCodeRegister {
uint16_t disruption_pc; /* start of disruption handler (0 = none) */
} JSCodeRegister;
/* Pre-parsed MCODE for a single function (off-heap, never GC'd).
Created by jsmcode_parse from cJSON MCODE output.
Instructions remain as cJSON pointers for string-based dispatch. */
typedef struct JSMCode {
uint16_t nr_args;
uint16_t nr_slots;
/* Pre-flattened instruction array (cJSON array items → C array for O(1) access) */
cJSON **instrs;
uint32_t instr_count;
/* Label map: label string → instruction index */
struct { const char *name; uint32_t index; } *labels;
uint32_t label_count;
/* Nested function definitions (indexes into top-level functions array) */
struct JSMCode **functions;
uint32_t func_count;
/* Keep root cJSON alive (owns all the cJSON nodes instrs[] point into) */
cJSON *json_root;
MachLineEntry *line_table; /* [instr_count], parallel to instrs[] */
const char *name; /* function name (points into cJSON tree) */
const char *filename; /* source filename (points into cJSON tree) */
uint16_t disruption_pc; /* start of disruption handler (0 = none) */
} JSMCode;
/* Frame for closures - used by link-time relocation model where closures
reference outer frames via (depth, slot) addressing.
@@ -1068,7 +1214,6 @@ typedef enum {
JS_FUNC_KIND_BYTECODE,
JS_FUNC_KIND_C_DATA,
JS_FUNC_KIND_REGISTER, /* register-based VM function */
JS_FUNC_KIND_MCODE, /* MCODE JSON interpreter */
} JSFunctionKind;
typedef struct JSFunction {
@@ -1092,11 +1237,6 @@ typedef struct JSFunction {
JSValue env_record; /* stone record, module environment */
JSValue outer_frame; /* JSFrame JSValue, for closures */
} reg;
struct {
JSMCode *code; /* pre-parsed MCODE (off-heap) */
JSValue outer_frame; /* lexical parent frame for closures */
JSValue env_record; /* module env or JS_NULL */
} mcode;
} u;
} JSFunction;
@@ -1259,8 +1399,6 @@ JSValue js_call_c_function (JSContext *ctx, JSValue func_obj, JSValue this_obj,
JSValue js_call_bound_function (JSContext *ctx, JSValue func_obj, JSValue this_obj, int argc, JSValue *argv);
JSValue JS_CallInternal (JSContext *ctx, JSValue func_obj, JSValue this_obj, int argc, JSValue *argv, int flags);
JSValue JS_CallRegisterVM(JSContext *ctx, JSCodeRegister *code, JSValue this_obj, int argc, JSValue *argv, JSValue env, JSValue outer_frame);
JSValue mcode_exec(JSContext *ctx, JSMCode *code, JSValue this_obj,
int argc, JSValue *argv, JSValue outer_frame);
int JS_DeleteProperty (JSContext *ctx, JSValue obj, JSValue prop);
JSValue __attribute__ ((format (printf, 2, 3)))
JS_ThrowInternalError (JSContext *ctx, const char *fmt, ...);
@@ -1574,13 +1712,6 @@ typedef struct {
} GetLineColCache;
/* === MachVarInfo (shared by mach.c and mcode.c) === */
typedef struct MachVarInfo {
char *name;
int slot;
int is_const; /* 1 for def, function args; 0 for var */
int is_closure; /* 1 if captured by a nested function */
} MachVarInfo;
/* === PPretext (parser pretext, system-malloc, used by cell_js.c parser) === */
typedef struct PPretext {
@@ -1770,8 +1901,5 @@ JSFrameRegister *alloc_frame_register(JSContext *ctx, int slot_count);
cJSON *mach_find_scope_record(cJSON *scopes, int function_nr);
int reg_vm_check_interrupt(JSContext *ctx);
/* mcode.c exports */
JSValue mcode_exec(JSContext *ctx, JSMCode *code, JSValue this_obj, int argc, JSValue *argv, JSValue outer_frame);
#endif /* QUICKJS_INTERNAL_H */