Files
sx/src/ir/inst.zig
2026-03-03 09:35:50 +02:00

466 lines
16 KiB
Zig

const std = @import("std");
const types = @import("types.zig");
const TypeId = types.TypeId;
const StringId = types.StringId;
// ── Handles ─────────────────────────────────────────────────────────────
/// Reference to an SSA value (instruction result).
pub const Ref = enum(u32) {
/// Sentinel for "no value" / unused operand.
none = std.math.maxInt(u32),
_,
pub fn index(self: Ref) u32 {
return @intFromEnum(self);
}
pub fn fromIndex(i: u32) Ref {
return @enumFromInt(i);
}
pub fn isNone(self: Ref) bool {
return self == .none;
}
};
pub const BlockId = enum(u32) {
_,
pub fn index(self: BlockId) u32 {
return @intFromEnum(self);
}
pub fn fromIndex(i: u32) BlockId {
return @enumFromInt(i);
}
};
pub const FuncId = enum(u32) {
_,
pub fn index(self: FuncId) u32 {
return @intFromEnum(self);
}
pub fn fromIndex(i: u32) FuncId {
return @enumFromInt(i);
}
};
pub const GlobalId = enum(u32) {
_,
pub fn index(self: GlobalId) u32 {
return @intFromEnum(self);
}
pub fn fromIndex(i: u32) GlobalId {
return @enumFromInt(i);
}
};
// ── Span ────────────────────────────────────────────────────────────────
pub const Span = struct {
start: u32 = 0,
end: u32 = 0,
};
// ── Instruction ─────────────────────────────────────────────────────────
pub const Inst = struct {
op: Op,
ty: TypeId,
span: Span = .{},
};
// ── Op (tagged union) ───────────────────────────────────────────────────
pub const Op = union(enum) {
// ── Constants ───────────────────────────────────────────────────
const_int: i64,
const_float: f64,
const_bool: bool,
const_string: StringId,
const_null,
const_undef, // `---` undefined initializer
// ── Arithmetic ──────────────────────────────────────────────────
add: BinOp,
sub: BinOp,
mul: BinOp,
div: BinOp,
mod: BinOp,
neg: UnaryOp, // unary -x
// ── Bitwise ─────────────────────────────────────────────────────
bit_and: BinOp,
bit_or: BinOp,
bit_xor: BinOp,
bit_not: UnaryOp,
shl: BinOp,
shr: BinOp,
// ── Comparison ──────────────────────────────────────────────────
cmp_eq: BinOp,
cmp_ne: BinOp,
cmp_lt: BinOp,
cmp_le: BinOp,
cmp_gt: BinOp,
cmp_ge: BinOp,
str_eq: BinOp, // string/slice equality via memcmp
str_ne: BinOp, // string/slice inequality via memcmp
// ── Logical ─────────────────────────────────────────────────────
bool_and: BinOp, // short-circuit &&
bool_or: BinOp, // short-circuit ||
bool_not: UnaryOp,
// ── Conversions ─────────────────────────────────────────────────
widen: Conversion, // safe widening (s32 → s64)
narrow: Conversion, // truncation via `xx` (s64 → s32)
bitcast: Conversion, // reinterpret bits
int_to_float: Conversion,
float_to_int: Conversion,
// ── Memory ──────────────────────────────────────────────────────
alloca: TypeId, // stack allocation, result is *T
load: UnaryOp, // load from pointer
store: Store, // store value to pointer
heap_alloc: UnaryOp, // context.allocator.alloc(size) → *void
heap_free: UnaryOp, // context.allocator.free(ptr)
// ── Struct ops ──────────────────────────────────────────────────
struct_init: Aggregate, // construct struct from field values
struct_get: FieldAccess, // read struct field by index
struct_gep: FieldAccess, // get pointer to struct field (GEP)
// ── Enum ops ────────────────────────────────────────────────────
enum_init: EnumInit, // construct enum value (tag + optional payload)
enum_tag: UnaryOp, // extract tag from enum/union
enum_payload: FieldAccess, // extract payload from tagged union
// ── Union ops ───────────────────────────────────────────────────
union_get: FieldAccess, // read union field (reinterpret)
union_gep: FieldAccess, // pointer to union field
// ── Array/Slice ops ─────────────────────────────────────────────
index_get: BinOp, // arr[idx] → value
index_gep: BinOp, // &arr[idx] → pointer
length: UnaryOp, // .len on slice/string/array
data_ptr: UnaryOp, // .ptr on slice/string
subslice: Subslice, // arr[lo..hi]
array_to_slice: UnaryOp, // [N]T → []T
// ── Tuple ops ───────────────────────────────────────────────────
tuple_init: Aggregate, // construct tuple from values
tuple_get: FieldAccess, // read tuple element by index
// ── Optional ops ────────────────────────────────────────────────
optional_wrap: UnaryOp, // T → ?T
optional_unwrap: UnaryOp, // ?T → T (UB if null)
optional_has_value: UnaryOp, // ?T → bool
optional_coalesce: BinOp, // a ?? b
// ── Pointer ops ─────────────────────────────────────────────────
addr_of: UnaryOp, // @x → *T
deref: UnaryOp, // p.* → T
// ── Vector ops ──────────────────────────────────────────────────
vec_splat: UnaryOp, // scalar → vector (broadcast)
vec_extract: BinOp, // vec[idx] → scalar
vec_insert: TriOp, // vec, idx, val → new_vec
// ── Calls ───────────────────────────────────────────────────────
call: Call,
call_indirect: CallIndirect,
call_closure: CallIndirect,
call_builtin: BuiltinCall,
compiler_call: CompilerCall,
// ── Protocol dispatch ───────────────────────────────────────────
protocol_call_dynamic: ProtocolCall, // vtable/inline dispatch
protocol_erase: ProtocolErase, // concrete → protocol value (xx)
// ── Closure creation ────────────────────────────────────────────
closure_create: ClosureCreate,
// ── Context ─────────────────────────────────────────────────────
context_load: ContextOp, // read context field
context_store: ContextOp, // write context field
context_save, // save context state (for push)
context_restore: UnaryOp, // restore context state (after push)
// ── Globals ─────────────────────────────────────────────────────
global_get: GlobalId,
global_set: GlobalSet,
func_ref: FuncId, // reference to a function (for function pointers)
// ── Block params (SSA phi alternative) ──────────────────────────
block_param: BlockParam,
// ── Any type ────────────────────────────────────────────────────
box_any: BoxAny, // T → Any (erase type)
unbox_any: UnaryOp, // Any → T (restore type)
// ── Reflection ─────────────────────────────────────────────────
field_name_get: FieldReflect, // field_name(T, i) → string (runtime index)
field_value_get: FieldReflect, // field_value(s, i) → Any (runtime struct + index)
// ── Terminators ─────────────────────────────────────────────────
br: Branch,
cond_br: CondBranch,
switch_br: SwitchBranch,
ret: UnaryOp,
ret_void,
@"unreachable",
// ── Misc ────────────────────────────────────────────────────────
/// No-op placeholder for unlowered AST nodes.
placeholder: StringId, // name of the unlowered construct
};
// ── Operand structs ─────────────────────────────────────────────────────
pub const UnaryOp = struct {
operand: Ref,
};
pub const BinOp = struct {
lhs: Ref,
rhs: Ref,
};
pub const TriOp = struct {
a: Ref,
b: Ref,
c: Ref,
};
pub const Store = struct {
ptr: Ref,
val: Ref,
};
pub const Conversion = struct {
operand: Ref,
from: TypeId,
to: TypeId,
};
pub const FieldAccess = struct {
base: Ref,
field_index: u32,
/// The IR type of the aggregate being accessed (struct, union, etc.).
/// Used by the LLVM emitter to resolve the correct type for GEP operations
/// without guessing from LLVM value chains.
base_type: ?TypeId = null,
};
pub const Aggregate = struct {
fields: []const Ref,
};
pub const EnumInit = struct {
tag: u32,
payload: Ref, // Ref.none if no payload
};
pub const Subslice = struct {
base: Ref,
lo: Ref,
hi: Ref,
};
pub const Call = struct {
callee: FuncId,
args: []const Ref,
};
pub const CallIndirect = struct {
callee: Ref,
args: []const Ref,
};
pub const BuiltinCall = struct {
builtin: BuiltinId,
args: []const Ref,
};
pub const BuiltinId = enum(u16) {
out,
sqrt,
sin,
cos,
floor,
size_of,
cast,
malloc,
free,
memcpy,
memset,
type_of,
alloc,
dealloc,
};
pub const CompilerCall = struct {
name: u32, // StringPool id for qualified name (e.g. "BuildOptions.add_link_flag")
args: []const Ref,
};
pub const ProtocolCall = struct {
receiver: Ref, // protocol value (ctx + vtable/fn_ptrs)
method_index: u32,
args: []const Ref,
};
pub const ProtocolErase = struct {
concrete: Ref,
protocol_type: TypeId,
};
pub const ClosureCreate = struct {
func: FuncId, // trampoline function
env: Ref, // allocated env pointer (or Ref.none for no captures)
};
pub const ContextOp = struct {
field: StringId,
value: Ref, // Ref.none for loads
};
pub const GlobalSet = struct {
global: GlobalId,
value: Ref,
};
pub const BlockParam = struct {
block: BlockId,
param_index: u32,
};
pub const BoxAny = struct {
operand: Ref,
source_type: TypeId,
};
pub const FieldReflect = struct {
base: Ref, // struct value (for field_value_get) or Ref.none (for field_name_get)
index: Ref, // runtime field index
struct_type: TypeId, // compile-time resolved struct type
};
pub const Branch = struct {
target: BlockId,
args: []const Ref, // block param values
};
pub const CondBranch = struct {
cond: Ref,
then_target: BlockId,
then_args: []const Ref,
else_target: BlockId,
else_args: []const Ref,
};
pub const SwitchBranch = struct {
operand: Ref,
cases: []const Case,
default: BlockId,
default_args: []const Ref,
pub const Case = struct {
value: i64,
target: BlockId,
args: []const Ref,
};
};
// ── Block ───────────────────────────────────────────────────────────────
pub const Block = struct {
name: StringId,
params: []const TypeId, // block parameter types (SSA phi alternative)
insts: std.ArrayList(Inst),
first_ref: u32 = 0, // ref index of the first instruction in this block
pub fn init(name: StringId, params: []const TypeId) Block {
return .{
.name = name,
.params = params,
.insts = std.ArrayList(Inst).empty,
};
}
pub fn deinit(self: *Block, alloc: std.mem.Allocator) void {
self.insts.deinit(alloc);
}
};
// ── Function ────────────────────────────────────────────────────────────
pub const Function = struct {
name: StringId,
params: []const Param,
ret: TypeId,
blocks: std.ArrayList(Block),
is_extern: bool = false,
is_comptime: bool = false,
linkage: Linkage = .internal,
pub const Param = struct {
name: StringId,
ty: TypeId,
};
pub const Linkage = enum {
internal,
external,
private,
};
pub fn init(name: StringId, params: []const Param, ret: TypeId) Function {
return .{
.name = name,
.params = params,
.ret = ret,
.blocks = std.ArrayList(Block).empty,
};
}
pub fn deinit(self: *Function, alloc: std.mem.Allocator) void {
for (self.blocks.items) |*block| {
block.deinit(alloc);
}
self.blocks.deinit(alloc);
}
};
// ── Global ──────────────────────────────────────────────────────────────
pub const Global = struct {
name: StringId,
ty: TypeId,
init_val: ?ConstantValue = null,
is_extern: bool = false,
is_const: bool = false,
/// For comptime globals: the function to interpret to get the init value.
comptime_func: ?FuncId = null,
};
// ── ConstantValue ───────────────────────────────────────────────────────
pub const ConstantValue = union(enum) {
int: i64,
float: f64,
boolean: bool,
string: StringId,
null_val,
undef,
zeroinit,
aggregate: []const ConstantValue,
};