diff --git a/issues/0074-ffi-arg-type-void-fallback.md b/issues/0074-ffi-arg-type-void-fallback.md
new file mode 100644
index 0000000..e10efd9
--- /dev/null
+++ b/issues/0074-ffi-arg-type-void-fallback.md
@@ -0,0 +1,84 @@
+# 0074 — silent `getRefIRType(arg_ref) orelse .void` fallback in FFI call-arg lowering
+
+> **✅ RESOLVED.** Root cause: four FFI call-arg lowering loops resolved an
+> argument's IR type via `getRefIRType(arg_ref) orelse .void` — a silent fallback
+> to the load-bearing real type `.void`, which downstream `toLLVMType` →
+> `abiCoerceParamType` → `coerceArg` treat as a legitimate (void-typed) foreign
+> argument, corrupting the call ABI with no diagnostic. Fix: one shared resolver
+> `LLVMEmitter.argIRTypeOrFail` ([src/ir/emit_llvm.zig]) returns the dedicated
+> `.unresolved` sentinel on a failed lookup — never `.void`/`.s64` — so the failure
+> cannot masquerade as a real type and trips `toLLVMType`'s existing hard `@panic`
+> tripwire at the call site. All four sites
+> ([src/ir/emit_llvm.zig] JNI constructor; [src/backend/llvm/ops.zig] objc_msgSend,
+> JNI non-virtual, JNI `Call<Type>Method`) now route through the helper. Happy path
+> is byte-identical (every real arg already has a resolved type) — FFI examples stay
+> green with zero snapshot churn. Regression test (fail-before/pass-after):
+> `src/ir/emit_llvm.test.zig` — "argIRTypeOrFail surfaces .unresolved for an
+> unresolvable FFI arg ref (issue 0074)".
+
+## Symptom
+**One-line:** Four FFI call-arg lowering sites silently default a failed
+argument-type lookup to `.void` — the forbidden silent-type-fallback anti-pattern
+(`.void` as a failed-type-lookup sentinel), which would produce a void-typed
+foreign-call argument (wrong LLVM param type → silent ABI corruption) with no
+diagnostic if the lookup ever fails.
+
+**Observed:** `self.getRefIRType(arg_ref) orelse .void` at:
+- `src/ir/emit_llvm.zig:2463`
+- `src/backend/llvm/ops.zig:517`  (Obj-C `objc_msgSend` arg loop)
+- `src/backend/llvm/ops.zig:731`  (JNI non-virtual call arg loop)
+- `src/backend/llvm/ops.zig:761`  (JNI `Call<Type>Method` arg loop)
+
+Each then does `toLLVMType(raw_ty)` → `abiCoerceParamType` → `coerceArg`, so a
+`.void` fallback silently mis-types the foreign-call argument.
+
+**Expected:** `getRefIRType` returning null for a real call argument is a
+"must-succeed lookup" failure (every arg is a valid param/instruction ref). Per
+`CLAUDE.md` REJECTED PATTERNS — *"`.void` is an UNACCEPTABLE sentinel for a failed
+type lookup"* — the lookup failure must surface as a diagnostic / hard tripwire, not
+a silently-corrupted argument type.
+
+## Provenance / scope
+Pre-existing pattern (the `emit_llvm.zig` site is original; the three `ops.zig` sites
+were relocated **verbatim, behavior-preserving** by step A7.4c of the arch-refactor —
+the flow reviewer/observer correctly approved the relocations as equivalence-preserving).
+Surfaced by the **A9.3 final fallback-audit** of the arch-refactor stream. Not
+introduced by the refactor; filed per the IMPASSIBLE RULE (*"If you find an existing
+default-return in the compiler that swallows a lookup failure, treat it as a
+discovered bug — file an issue, do not just delete the default in place"*).
+
+## Reproduction
+This is a **latent / static** finding: there is no known sx program that drives
+`getRefIRType` to `null` for a valid foreign-call argument (well-formed IR always
+has a type for every arg ref), so it cannot currently be triggered at runtime — which
+is exactly why it is dangerous (a future IR change that breaks the invariant would
+corrupt FFI ABI silently). The code paths are exercised (and must stay green after
+the fix) by the existing FFI examples, e.g.:
+
+```
+examples/13xx-ffi-objc-*        # objc_msgSend arg lowering (ops.zig:517)
+examples/14xx-ffi-jni-*         # JNI Call<Type>Method / non-virtual (ops.zig:731/761)
+```
+
+(No new minimal repro `.sx` is meaningful for a latent defensive fallback; the fix is
+verified by (a) the FFI suite staying green and (b) a unit test that asserts the new
+loud-failure path, see below.)
+
+## Investigation prompt (ready to paste into a fresh session)
+> In `/Users/agra/projects/sx`, four FFI call-arg lowering sites use the forbidden
+> silent type-fallback `self.getRefIRType(arg_ref) orelse .void`
+> (`src/ir/emit_llvm.zig:2463`; `src/backend/llvm/ops.zig:517`, `:731`, `:761`).
+> `getRefIRType` (`src/ir/emit_llvm.zig:2229`, returns `?TypeId`) yields `null` only
+> when a ref is neither a function param nor a block instruction result — a
+> must-not-happen case for a real call argument. Replace the silent `.void` default
+> with a loud failure that cannot be mistaken for a real type, per `CLAUDE.md`
+> REJECTED PATTERNS: emit a diagnostic via `self.diagnostics.addFmt(.err, span,
+> "...", .{...})` and/or a hard tripwire (`@panic`/`bailDetail`-style) naming the op
+> and the bad ref — do NOT substitute another real type. Prefer a single shared
+> helper (e.g. `argIRTypeOrFail(arg_ref, span)`) used by all four sites so the policy
+> lives in one place. Then: (1) `/Users/agra/.zvm/bin/zig build && /Users/agra/.zvm/bin/zig
+> build test && bash tests/run_examples.sh` must stay 361/0 with the FFI examples
+> green (the happy path is unchanged); (2) add a `*.test.zig` unit test that
+> constructs an FFI call op with a bogus arg ref and asserts the loud failure fires
+> (not a `.void` silent default). Expected new behavior: an unresolved FFI arg type
+> produces a clear compiler error / panic, never a void-typed foreign argument.
diff --git a/issues/0075-reflection-builtin-s64-type-fallback.md b/issues/0075-reflection-builtin-s64-type-fallback.md
new file mode 100644
index 0000000..8e3c3b7
--- /dev/null
+++ b/issues/0075-reflection-builtin-s64-type-fallback.md
@@ -0,0 +1,77 @@
+# 0075 — silent `getRefIRType(...) orelse TypeId.s64` fallback in reflection builtins
+
+## Symptom
+**One-line:** The `type_name` and `type_eq` reflection builtins resolve their Type
+argument's IR type via `getRefIRType(...) orelse TypeId.s64` — the forbidden
+silent-type-lookup fallback (`.s64` is the exact issue-0042 sentinel the project
+rules name) — so a failed must-succeed lookup silently decides "not boxed (`!= .any`)"
+and mis-handles the value with no diagnostic.
+
+**Observed (primary — must fix):** `self.e.getRefIRType(...) orelse TypeId.s64` at:
+- `src/backend/llvm/ops.zig:1023`  (`.type_name` builtin — `arg_ir_ty`, gates the
+  `== .any` boxed-extract vs bare-i64 decision)
+- `src/backend/llvm/ops.zig:1049`  (`.type_eq` builtin — first operand)
+- `src/backend/llvm/ops.zig:1055`  (`.type_eq` builtin — second operand)
+
+`getRefIRType` (`src/ir/emit_llvm.zig:2229`, `?TypeId`) returns `null` only when a ref
+is neither a function param nor a block instruction result — a must-not-happen case
+for a real builtin argument. On `null` the code defaults to `.s64`, then tests
+`arg_ir_ty == .any`; the `.s64` default silently means "treat as a bare TypeId index,
+not a boxed `Any`", so a genuinely-boxed arg whose lookup failed would skip the
+`ExtractValue` and use the wrong value — silent miscompile, no diagnostic.
+
+**Expected:** per `CLAUDE.md` REJECTED PATTERNS, a failed must-succeed type lookup
+surfaces a diagnostic / hard tripwire (e.g. the `.unresolved` sentinel introduced for
+issue 0074), never a real-type default.
+
+## Secondary (confirm — borderline)
+- `src/ir/lower.zig:2527` — `.null_literal => constNull(self.target_type orelse .void)`
+- `src/ir/lower.zig:2528` — `.undef_literal => constUndef(self.target_type orelse .void)`
+  `target_type` is a context hint that may be legitimately absent for a bare
+  `null`/`undef` with no expected type — this may be an INTENTIONAL default rather
+  than a lookup-swallow. The fix session should confirm: if a `null`/`undef` literal
+  reaching here without a `target_type` is actually a must-not-happen case, make it
+  loud; if a typeless null/undef is legitimate, leave it and add a one-line comment
+  stating the invariant.
+
+## Audited — intentional language defaults (NO action; documented so they aren't re-flagged)
+- `src/ir/lower.zig:4855` — `int_literal => constInt(lit.value, info.ty orelse .s64)`:
+  an untyped integer literal defaulting to `s64` is standard language semantics, not a
+  lookup failure.
+- `src/ir/lower.zig:4856` — `float_literal => constFloat(..., info.ty orelse .f64)`:
+  untyped float literal defaults to `f64` — language semantics.
+- `src/ir/type_bridge.zig:334` — `.tag_type = tag_type orelse .s64`: documented
+  ("enum unions are always tagged (default i64)") — an intentional default tag type,
+  not a swallowed lookup.
+
+## Provenance / scope
+Pre-existing, NOT introduced by the arch-refactor. Discovered during the **issue-0074
+fix** (the fix worker surfaced the reflection `.s64` fallbacks as a separate pattern
+outside 0074's FFI-arg scope) and confirmed by a manager sweep
+(`rg "orelse \.(s64|void|...)" src`). Filed per the IMPASSIBLE RULE (existing
+default-returns that swallow a lookup failure → file, don't fix in place).
+
+## Reproduction
+Latent / static (same nature as 0074): well-formed IR always gives a builtin arg a
+resolvable type, so the `.s64` default can't be driven at runtime today — which is why
+it's dangerous (a future IR change would silently miscompile `type_name`/`type_eq`).
+Exercised by the comptime/reflection examples; the fix must keep the suite at 361/0.
+
+## Investigation prompt (ready to paste into a fresh session)
+> In `/Users/agra/projects/sx`, the `.type_name` and `.type_eq` reflection builtins in
+> `src/backend/llvm/ops.zig` (lines 1023, 1049, 1055) resolve a Type argument's IR type
+> with the forbidden silent fallback `getRefIRType(...) orelse TypeId.s64`, used to gate
+> a `== .any` boxed-vs-bare decision. Issue 0074 already added the shared resolver
+> `LLVMEmitter.argIRTypeOrFail` (`src/ir/emit_llvm.zig`) returning the dedicated
+> `.unresolved` sentinel on a failed lookup. Route these three sites through that helper
+> (or a sibling) so a failed lookup yields `.unresolved` — never `.s64`; then `==.any`
+> is false for `.unresolved` AND you must make the unresolved case loud (diagnostic via
+> `self.diagnostics.addFmt(.err, span, ...)` or a hard tripwire), not silently "bare
+> i64". Also resolve the borderline `lower.zig:2527/2528` `target_type orelse .void`
+> (confirm intentional vs make-loud; comment the invariant either way). Leave the
+> audited intentional defaults (`lower.zig:4855/4856`, `type_bridge.zig:334`) untouched.
+> Verify: `/Users/agra/.zvm/bin/zig build && /Users/agra/.zvm/bin/zig build test &&
+> bash tests/run_examples.sh` stays 361/0; add a `*.test.zig` regression test asserting
+> the loud `.unresolved` path for a `type_name`/`type_eq` arg with an unresolvable ref
+> (fail-before/pass-after). Expected new behavior: an unresolved reflection-builtin arg
+> type surfaces loudly, never silently defaults to `.s64`.
diff --git a/src/backend/llvm/ops.zig b/src/backend/llvm/ops.zig
index e34bbb1..e2f529e 100644
--- a/src/backend/llvm/ops.zig
+++ b/src/backend/llvm/ops.zig
@@ -514,7 +514,7 @@ pub const Ops = struct {
         // coercion applied so structs / strings decay the
         // same way they do for any C foreign call.
         for (msg.args, 0..) |arg_ref, i| {
-            const raw_ty = self.e.getRefIRType(arg_ref) orelse .void;
+            const raw_ty = self.e.argIRTypeOrFail(arg_ref);
             const raw_llvm = self.e.toLLVMType(raw_ty);
             const coerced_ty = self.e.abiCoerceParamType(raw_ty, raw_llvm);
             param_types[i + 2 + sret_off] = coerced_ty;
@@ -728,7 +728,7 @@ pub const Ops = struct {
                 call_args_nv[2] = cls;
                 call_args_nv[3] = mid_val;
                 for (msg.args, 0..) |arg_ref, i| {
-                    const raw_ty = self.e.getRefIRType(arg_ref) orelse .void;
+                    const raw_ty = self.e.argIRTypeOrFail(arg_ref);
                     const raw_llvm = self.e.toLLVMType(raw_ty);
                     const coerced_ty = self.e.abiCoerceParamType(raw_ty, raw_llvm);
                     call_param_types_nv[i + 4] = coerced_ty;
@@ -758,7 +758,7 @@ pub const Ops = struct {
         call_args[1] = target;
         call_args[2] = mid;
         for (msg.args, 0..) |arg_ref, i| {
-            const raw_ty = self.e.getRefIRType(arg_ref) orelse .void;
+            const raw_ty = self.e.argIRTypeOrFail(arg_ref);
             const raw_llvm = self.e.toLLVMType(raw_ty);
             const coerced_ty = self.e.abiCoerceParamType(raw_ty, raw_llvm);
             call_param_types[i + 3] = coerced_ty;
diff --git a/src/ir/emit_llvm.test.zig b/src/ir/emit_llvm.test.zig
index 8d9426f..ff20cda 100644
--- a/src/ir/emit_llvm.test.zig
+++ b/src/ir/emit_llvm.test.zig
@@ -1097,3 +1097,50 @@ test "emit: ERR E3.0 — no DWARF without a debug context (unit-test default)" {
     try std.testing.expect(std.mem.indexOf(u8, ir_str, "DICompileUnit") == null);
     try std.testing.expect(std.mem.indexOf(u8, ir_str, "!dbg") == null);
 }
+
+// ── issue 0074: FFI arg-type lookup must fail loudly, never silently `.void` ──
+// `argIRTypeOrFail` backs the four FFI call-arg lowering sites (objc_msgSend,
+// JNI Call<Type>Method / non-virtual / constructor). A ref it cannot resolve is
+// a codegen invariant violation; it must surface the dedicated `.unresolved`
+// tripwire sentinel (which `toLLVMType` hard-panics on) rather than the old
+// silent `.void` default that would emit a void-typed foreign-call argument.
+test "emit: argIRTypeOrFail surfaces .unresolved for an unresolvable FFI arg ref (issue 0074)" {
+    const alloc = std.testing.allocator;
+    var module = Module.init(alloc);
+    defer module.deinit();
+
+    var b = Builder.init(&module);
+
+    // func ffifn(a: s64, b: f64) -> void { <entry> }
+    const fid = b.beginFunction(str(&module, "ffifn"), &[_]Function.Param{
+        .{ .name = str(&module, "a"), .ty = .s64 },
+        .{ .name = str(&module, "b"), .ty = .f64 },
+    }, .void);
+    const entry = b.appendBlock(str(&module, "entry"), &.{});
+    b.switchToBlock(entry);
+    b.retVoid();
+    b.finalize();
+
+    var emitter = LLVMEmitter.init(alloc, &module, "test_ffi_argty", .{});
+    defer emitter.deinit();
+    emitter.current_func_idx = fid.index();
+
+    // Happy path: a real arg ref (param 0 / param 1) resolves byte-identically
+    // to its declared IR type — the FFI fast path is unchanged.
+    try std.testing.expectEqual(TypeId.s64, emitter.argIRTypeOrFail(Ref.fromIndex(0)));
+    try std.testing.expectEqual(TypeId.f64, emitter.argIRTypeOrFail(Ref.fromIndex(1)));
+
+    // A ref past every param and instruction is unresolvable.
+    const bogus = Ref.fromIndex(100_000);
+    try std.testing.expectEqual(@as(?TypeId, null), emitter.getRefIRType(bogus));
+
+    // Fail-before: the old `getRefIRType(arg) orelse .void` would silently
+    // yield `.void` here — a real, load-bearing type that downstream ABI
+    // coercion treats as a legitimate (void-typed) foreign argument.
+    try std.testing.expectEqual(TypeId.void, emitter.getRefIRType(bogus) orelse TypeId.void);
+
+    // Pass-after: the helper returns the dedicated `.unresolved` sentinel,
+    // never `.void`, so the failure cannot masquerade as a real type.
+    try std.testing.expectEqual(TypeId.unresolved, emitter.argIRTypeOrFail(bogus));
+    try std.testing.expect(emitter.argIRTypeOrFail(bogus) != .void);
+}
diff --git a/src/ir/emit_llvm.zig b/src/ir/emit_llvm.zig
index 78853d2..0506f10 100644
--- a/src/ir/emit_llvm.zig
+++ b/src/ir/emit_llvm.zig
@@ -2239,6 +2239,16 @@ pub const LLVMEmitter = struct {
         return null;
     }
 
+    /// Resolve the IR type of a foreign-call argument ref. Every FFI arg ref is
+    /// a real function param or block instruction result, so a `null` here is a
+    /// codegen invariant violation, not a recoverable case: return the dedicated
+    /// `.unresolved` sentinel — never `.void`/`.s64` — so the failure cannot be
+    /// mistaken for a real type and trips `toLLVMType`'s hard tripwire at the call
+    /// site instead of silently emitting a void-typed foreign argument.
+    pub fn argIRTypeOrFail(self: *LLVMEmitter, arg_ref: Ref) TypeId {
+        return self.getRefIRType(arg_ref) orelse .unresolved;
+    }
+
 
     /// Coerce both binary operands to match the instruction's result type.
     /// E.g. if result is i64 but one operand is i32, sext it.
@@ -2460,7 +2470,7 @@ pub const LLVMEmitter = struct {
         call_args[1] = cls;
         call_args[2] = mid;
         for (msg.args, 0..) |arg_ref, i| {
-            const raw_ty = self.getRefIRType(arg_ref) orelse .void;
+            const raw_ty = self.argIRTypeOrFail(arg_ref);
             const raw_llvm = self.toLLVMType(raw_ty);
             const coerced_ty = self.abiCoerceParamType(raw_ty, raw_llvm);
             call_param_types[i + 3] = coerced_ty;