diff --git a/wasm/bindings/collab_common.h b/wasm/bindings/collab_common.h
index 5dd26a5..7057571 100644
--- a/wasm/bindings/collab_common.h
+++ b/wasm/bindings/collab_common.h
@@ -61,36 +61,6 @@ inline bool& fiberBusy()
     return busy;
 }
 
-/**
- * True while an Asyncify unwind/rewind or park is IN FLIGHT (the single
- * `currData` slot is taken by somebody else's suspended context).
- *
- * Starting a fiber in that window is the documented crash family
- * (docs/features/async/13, "scheduler invariant"): `emscripten_fiber_swap`
- * calls `stop_unwind`, which asserts/corrupts when another context is parked —
- * surfacing as "index out of bounds", "unreachable executed" or "indirect call
- * signature mismatch", after which the runtime is poisoned and every later
- * event traps.
- *
- * Normally the slot is FREE here: the main loop's per-frame
- * `wxWasmYieldToBrowser` completes every frame, so `drainFibers` (a
- * CallAfter, drained from ProcessEvents) runs between yields with nothing in
- * flight. It is NOT free when a nested/modal pump tick drives ProcessEvents
- * while the chain that opened the modal is parked deeper down the stack —
- * exactly what a big board open does (progress dialog over a parked load),
- * and the window in which a presence/collab body arriving off the WebSocket
- * used to swap a fiber and take the runtime down.
- *
- * The wx dispatch interlock (wx/wasm/private/dispatch.h) does not cover this:
- * the modal parks deliberately zero its count so their own pump may dispatch.
- */
-inline bool asyncifyInFlight()
-{
-    return EM_ASM_INT( {
-        return ( typeof Asyncify !== "undefined" && Asyncify.currData ) ? 1 : 0;
-    } ) != 0;
-}
-
 /* The in-flight body. HEAP-allocated and pinned for the body's whole life:
  * when a body PARKS (asyncify suspension inside commit.Push), COROUTINE::Call
  * RETURNS EARLY — the later asyncify rewind re-enters the fiber through the
@@ -148,18 +118,6 @@ inline void drainFibers()
 
     while( !q.empty() )
     {
-        // Someone else's context is parked in the single Asyncify slot: swapping
-        // to a fiber now corrupts it (see asyncifyInFlight). Re-queue and retry
-        // on the next drained tick — bodies are viewport/overlay/apply work, so
-        // waiting out the park costs latency, never correctness.
-        if( asyncifyInFlight() )
-        {
-            if( wxEvtHandler* h = fiberHandler() )
-                h->CallAfter( []() { drainFibers(); } );
-
-            return;
-        }
-
         fiberBusy() = true;
         slot.done = false;
         slot.body = new std::function<void()>( std::move( q.front() ) );