diff --git a/core/fmt.odin b/core/fmt.odin
index c2f6f1e07..d2ea0b19a 100644
--- a/core/fmt.odin
+++ b/core/fmt.odin
@@ -8,34 +8,29 @@
 
 _BUFFER_SIZE :: 1<<12;
 
-// TODO(bill): Make this a union
-StringBuffer :: struct {
-	is_dynamic: bool,
-	sa: []byte,
-	da: [dynamic]byte,
-};
+StringBuffer :: union {
+	Static {buf: []byte},
+	Dynamic{buf: [dynamic]byte},
+}
 
 make_string_buffer_from_slice :: proc(b: []byte) -> StringBuffer {
-	return StringBuffer{
-		is_dynamic = false,
-		sa = b,
-	};
+	return StringBuffer.Static{b};
 }
 
 make_string_dynamic_buffer :: proc() -> StringBuffer {
-	return StringBuffer{
-		is_dynamic = true,
-		da = make([dynamic]byte),
-	};
+	return StringBuffer.Dynamic{make([dynamic]byte)};
 }
 string_buffer_data :: proc(buf: ^StringBuffer) -> []byte {
 	return string_buffer_data(buf^);
 }
 string_buffer_data :: proc(buf: StringBuffer) -> []byte {
-	if buf.is_dynamic {
-		return buf.da[..];
+	match b in buf {
+	case StringBuffer.Static:
+		return b.buf[..];
+	case StringBuffer.Dynamic:
+		return b.buf[..];
 	}
-	return buf.sa[..];
+	return nil;
 }
 to_string :: proc(buf: StringBuffer) -> string {
 	return string(string_buffer_data(buf));
@@ -45,18 +40,20 @@ to_string :: proc(buf: StringBuffer) -> string {
 write_string :: proc(buf: ^StringBuffer, s: string) {
 	write_bytes(buf, []byte(s));
 }
-write_bytes :: proc(buf: ^StringBuffer, b: []byte) {
-	if buf.is_dynamic {
-		append(buf.da, ..b);
-	} else {
-		append(buf.sa, ..b);
+write_bytes :: proc(buf: ^StringBuffer, data: []byte) {
+	match b in buf {
+	case StringBuffer.Static:
+		append(b.buf, ..data);
+	case StringBuffer.Dynamic:
+		append(b.buf, ..data);
 	}
 }
-write_byte :: proc(buf: ^StringBuffer, b: byte) {
-	if buf.is_dynamic {
-		append(buf.da, b);
-	} else {
-		append(buf.sa, b);
+write_byte :: proc(buf: ^StringBuffer, data: byte) {
+	match b in buf {
+	case StringBuffer.Static:
+		append(b.buf, data);
+	case StringBuffer.Dynamic:
+		append(b.buf, data);
 	}
 }
 write_rune :: proc(buf: ^StringBuffer, r: rune) {
@@ -192,8 +189,7 @@ write_type :: proc(buf: ^StringBuffer, ti: ^TypeInfo) {
 		case ti == type_info(uint): write_string(buf, "uint");
 		case:
 			write_string(buf, info.signed ? "i" : "u");
-			fi := FmtInfo{buf = buf};
-			fmt_int(&fi, u128(8*info.size), false, 64, 'd');
+			write_int(buf, i64(8*info.size), 10);
 		}
 	case Float:
 		match info.size {
@@ -258,7 +254,7 @@ write_type :: proc(buf: ^StringBuffer, ti: ^TypeInfo) {
 	case Array:
 		write_string(buf, "[");
 		fi := FmtInfo{buf = buf};
-		fmt_int(&fi, u128(info.count), false, 64, 'd');
+		write_int(buf, i64(info.count), 10);
 		write_string(buf, "]");
 		write_type(buf, info.elem);
 	case DynamicArray:
@@ -269,8 +265,7 @@ write_type :: proc(buf: ^StringBuffer, ti: ^TypeInfo) {
 		write_type(buf, info.elem);
 	case Vector:
 		write_string(buf, "[vector ");
-		fi := FmtInfo{buf = buf};
-		fmt_int(&fi, u128(info.count), false, 64, 'd');
+		write_int(buf, i64(info.count), 10);
 		write_string(buf, "]");
 		write_type(buf, info.elem);
 
@@ -286,8 +281,7 @@ write_type :: proc(buf: ^StringBuffer, ti: ^TypeInfo) {
 		if info.ordered { write_string(buf, "#ordered "); }
 		if info.custom_align {
 			write_string(buf, "#align ");
-			fi := FmtInfo{buf = buf};
-			fmt_int(&fi, u128(info.align), false, 64, 'd');
+			write_int(buf, i64(info.align), 10);
 			write_byte(buf, ' ');
 		}
 		write_byte(buf, '{');
@@ -361,10 +355,37 @@ write_type :: proc(buf: ^StringBuffer, ti: ^TypeInfo) {
 			write_string(buf, name);
 		}
 		write_string(buf, "}");
+	case BitField:
+		write_string(buf, "bit_field ");
+		if info.align != 1 {
+			write_string(buf, "#align ");
+			write_int(buf, i64(info.align), 10);
+			write_rune(buf, ' ');
+		}
+		write_string(buf, " {");
+		for name, i in info.names {
+			if i > 0 {
+				write_string(buf, ", ");
+			}
+			write_string(buf, name);
+			write_string(buf, ": ");
+			write_int(buf, i64(info.bits[i]), 10);
+		}
+		write_string(buf, "}");
+
 	}
 }
 
-
+write_int :: proc(buf: ^StringBuffer, i: i128, base: int) {
+	b: [129]byte;
+	s := strconv.append_bits(b[0..<0], u128(i), base, true, 128, strconv.digits, 0);
+	write_string(buf, s);
+}
+write_int :: proc(buf: ^StringBuffer, i: i64, base: int) {
+	b: [129]byte;
+	s := strconv.append_bits(b[0..<0], u128(i), base, true, 64, strconv.digits, 0);
+	write_string(buf, s);
+}
 
 _parse_int :: proc(s: string, offset: int) -> (result: int, offset: int, ok: bool) {
 	is_digit :: proc(r: rune) -> bool #inline {
@@ -488,7 +509,7 @@ fmt_write_padding :: proc(fi: ^FmtInfo, width: int) {
 	}
 }
 
-_write_int :: proc(fi: ^FmtInfo, u: u128, base: int, is_signed: bool, bit_size: int, digits: string) {
+_fmt_int :: proc(fi: ^FmtInfo, u: u128, base: int, is_signed: bool, bit_size: int, digits: string) {
 	_, neg := strconv.is_integer_negative(u128(u), is_signed, bit_size);
 
 	BUF_SIZE :: 256;
@@ -496,7 +517,7 @@ _write_int :: proc(fi: ^FmtInfo, u: u128, base: int, is_signed: bool, bit_size:
 		width := fi.width + fi.prec + 3; // 3 extra bytes for sign and prefix
 		if width > BUF_SIZE {
 			// TODO(bill):????
-			panic("_write_int: buffer overrun. Width and precision too big");
+			panic("_fmt_int: buffer overrun. Width and precision too big");
 		}
 	}
 
@@ -522,7 +543,7 @@ _write_int :: proc(fi: ^FmtInfo, u: u128, base: int, is_signed: bool, bit_size:
 	case 2, 8, 10, 12, 16:
 		break;
 	case:
-		panic("_write_int: unknown base, whoops");
+		panic("_fmt_int: unknown base, whoops");
 	}
 
 	buf: [256]byte;
@@ -547,12 +568,12 @@ fmt_rune :: proc(fi: ^FmtInfo, r: rune) {
 
 fmt_int :: proc(fi: ^FmtInfo, u: u128, is_signed: bool, bit_size: int, verb: rune) {
 	match verb {
-	case 'v': _write_int(fi, u, 10, is_signed, bit_size, __DIGITS_LOWER);
-	case 'b': _write_int(fi, u,  2, is_signed, bit_size, __DIGITS_LOWER);
-	case 'o': _write_int(fi, u,  8, is_signed, bit_size, __DIGITS_LOWER);
-	case 'd': _write_int(fi, u, 10, is_signed, bit_size, __DIGITS_LOWER);
-	case 'x': _write_int(fi, u, 16, is_signed, bit_size, __DIGITS_LOWER);
-	case 'X': _write_int(fi, u, 16, is_signed, bit_size, __DIGITS_UPPER);
+	case 'v': _fmt_int(fi, u, 10, is_signed, bit_size, __DIGITS_LOWER);
+	case 'b': _fmt_int(fi, u,  2, is_signed, bit_size, __DIGITS_LOWER);
+	case 'o': _fmt_int(fi, u,  8, is_signed, bit_size, __DIGITS_LOWER);
+	case 'd': _fmt_int(fi, u, 10, is_signed, bit_size, __DIGITS_LOWER);
+	case 'x': _fmt_int(fi, u, 16, is_signed, bit_size, __DIGITS_LOWER);
+	case 'X': _fmt_int(fi, u, 16, is_signed, bit_size, __DIGITS_UPPER);
 	case 'c', 'r':
 		fmt_rune(fi, rune(u));
 	case 'U':
@@ -561,7 +582,7 @@ fmt_int :: proc(fi: ^FmtInfo, u: u128, is_signed: bool, bit_size: int, verb: run
 			fmt_bad_verb(fi, verb);
 		} else {
 			write_string(fi.buf, "U+");
-			_write_int(fi, u, 16, false, bit_size, __DIGITS_UPPER);
+			_fmt_int(fi, u, 16, false, bit_size, __DIGITS_UPPER);
 		}
 
 	case:
@@ -643,7 +664,7 @@ fmt_string :: proc(fi: ^FmtInfo, s: string, verb: rune) {
 			if i > 0 && space {
 				write_byte(fi.buf, ' ');
 			}
-			_write_int(fi, u128(s[i]), 16, false, 8, verb == 'x' ? __DIGITS_LOWER : __DIGITS_UPPER);
+			_fmt_int(fi, u128(s[i]), 16, false, 8, verb == 'x' ? __DIGITS_LOWER : __DIGITS_UPPER);
 		}
 
 	case:
@@ -663,7 +684,7 @@ fmt_pointer :: proc(fi: ^FmtInfo, p: rawptr, verb: rune) {
 	if !fi.hash || verb == 'v' {
 		write_string(fi.buf, "0x");
 	}
-	_write_int(fi, u, 16, false, 8*size_of(rawptr), __DIGITS_UPPER);
+	_fmt_int(fi, u, 16, false, 8*size_of(rawptr), __DIGITS_UPPER);
 }
 
 fmt_enum :: proc(fi: ^FmtInfo, v: any, verb: rune) {
diff --git a/src/check_expr.c b/src/check_expr.c
index d5a26a193..866a5f1a4 100644
--- a/src/check_expr.c
+++ b/src/check_expr.c
@@ -452,7 +452,7 @@ isize check_fields(Checker *c, AstNode *node, AstNodeArray decls,
 
 		if (is_using) {
 			Type *t = base_type(type_deref(type));
-			if (!is_type_struct(t) && !is_type_raw_union(t) &&
+			if (!is_type_struct(t) && !is_type_raw_union(t) && !is_type_bit_field(t) &&
 			    f->names.count >= 1 &&
 			    f->names.e[0]->kind == AstNode_Ident) {
 				Token name_token = f->names.e[0]->Ident;
@@ -477,7 +477,9 @@ isize check_fields(Checker *c, AstNode *node, AstNodeArray decls,
 						error(name_token, "Previous `using` for an index expression `%.*s`", LIT(name_token.string));
 					}
 				} else {
-					error(name_token, "`using` on a field `%.*s` must be a `struct` or `raw_union`", LIT(name_token.string));
+					gbString type_str = type_to_string(type);
+					error(name_token, "`using` cannot be applied to the field `%.*s` of type `%s`", LIT(name_token.string), type_str);
+					gb_string_free(type_str);
 					continue;
 				}
 			}
@@ -2258,7 +2260,8 @@ void check_unary_expr(Checker *c, Operand *o, Token op, AstNode *node) {
 
 		if (o->mode != Addressing_Variable ||
 		    check_is_expr_vector_index(c, o->expr) ||
-		    check_is_vector_elem(c, o->expr)) {
+		    check_is_vector_elem(c, o->expr) ||
+		    is_type_bit_field_value(o->type)) {
 			if (ast_node_expect(node, AstNode_UnaryExpr)) {
 				ast_node(ue, UnaryExpr, node);
 				gbString str = expr_to_string(ue->expr);
diff --git a/src/ir.c b/src/ir.c
index 2002c7a4c..f2b75b88d 100644
--- a/src/ir.c
+++ b/src/ir.c
@@ -4836,10 +4836,18 @@ irAddr ir_build_addr(irProcedure *proc, AstNode *expr) {
 			if (sel.entity->type->kind == Type_BitFieldValue) {
 				irAddr addr = ir_build_addr(proc, se->expr);
 				Type *bft = type_deref(ir_addr_type(addr));
-				GB_ASSERT(is_type_bit_field(bft));
-				GB_ASSERT(sel.index.count == 1);
-				i32 index = sel.index.e[0];
-				return ir_addr_bit_field(addr.addr, index);
+				if (sel.index.count == 1) {
+					GB_ASSERT(is_type_bit_field(bft));
+					i32 index = sel.index.e[0];
+					return ir_addr_bit_field(addr.addr, index);
+				} else {
+					Selection s = sel;
+					s.index.count--;
+					i32 index = s.index.e[s.index.count-1];
+					irValue *a = addr.addr;
+					a = ir_emit_deep_field_gep(proc, a, s);
+					return ir_addr_bit_field(a, index);
+				}
 			} else {
 				irValue *a = ir_build_addr(proc, se->expr).addr;
 				a = ir_emit_deep_field_gep(proc, a, sel);