gh-106213: Make Emscripten trampolines work with JSPI (GH-106219)

There is a WIP proposal to enable webassembly stack switching which have been
implemented in v8:

https://github.com/WebAssembly/js-promise-integration

It is not possible to switch stacks that contain JS frames so the Emscripten JS
trampolines that allow calling functions with the wrong number of arguments
don't work in this case. However, the js-promise-integration proposal requires
the [type reflection for Wasm/JS API](https://github.com/WebAssembly/js-types)
proposal, which allows us to actually count the number of arguments a function
expects.

For better compatibility with stack switching, this PR checks if type reflection
is available, and if so we use a switch block to decide the appropriate
signature. If type reflection is unavailable, we should use the current EMJS
trampoline.

We cache the function argument counts since when I didn't cache them performance
was negatively affected.

Co-authored-by: T. Wouters <thomas@python.org>
Co-authored-by: Brett Cannon <brett@python.org>

Python/emscripten_trampoline.c

@@ -0,0 +1,82 @@
+#if defined(PY_CALL_TRAMPOLINE)
+
+#include <emscripten.h>             // EM_JS
+#include <Python.h>
+#include "pycore_runtime.h"         // _PyRuntime
+
+
+/**
+ * This is the GoogleChromeLabs approved way to feature detect type-reflection:
+ * https://github.com/GoogleChromeLabs/wasm-feature-detect/blob/main/src/detectors/type-reflection/index.js
+ */
+EM_JS(int, _PyEM_detect_type_reflection, (), {
+    return "Function" in WebAssembly;
+});
+
+void
+_Py_EmscriptenTrampoline_Init(_PyRuntimeState *runtime)
+{
+    runtime->wasm_type_reflection_available = _PyEM_detect_type_reflection();
+}
+
+/**
+ * Backwards compatible trampoline works with all JS runtimes
+ */
+EM_JS(PyObject*,
+_PyEM_TrampolineCall_JavaScript, (PyCFunctionWithKeywords func,
+                                  PyObject *arg1,
+                                  PyObject *arg2,
+                                  PyObject *arg3),
+{
+    return wasmTable.get(func)(arg1, arg2, arg3);
+}
+);
+
+/**
+ * In runtimes with WebAssembly type reflection, count the number of parameters
+ * and cast to the appropriate signature
+ */
+EM_JS(int, _PyEM_CountFuncParams, (PyCFunctionWithKeywords func),
+{
+    let n = _PyEM_CountFuncParams.cache.get(func);
+
+    if (n !== undefined) {
+        return n;
+    }
+    n = WebAssembly.Function.type(wasmTable.get(func)).parameters.length;
+    _PyEM_CountFuncParams.cache.set(func, n);
+    return n;
+}
+_PyEM_CountFuncParams.cache = new Map();
+)
+
+
+typedef PyObject* (*zero_arg)(void);
+typedef PyObject* (*one_arg)(PyObject*);
+typedef PyObject* (*two_arg)(PyObject*, PyObject*);
+typedef PyObject* (*three_arg)(PyObject*, PyObject*, PyObject*);
+
+
+PyObject*
+_PyEM_TrampolineCall_Reflection(PyCFunctionWithKeywords func,
+                                PyObject* self,
+                                PyObject* args,
+                                PyObject* kw)
+{
+    switch (_PyEM_CountFuncParams(func)) {
+        case 0:
+            return ((zero_arg)func)();
+        case 1:
+            return ((one_arg)func)(self);
+        case 2:
+            return ((two_arg)func)(self, args);
+        case 3:
+            return ((three_arg)func)(self, args, kw);
+        default:
+            PyErr_SetString(PyExc_SystemError,
+                            "Handler takes too many arguments");
+            return NULL;
+    }
+}
+
+#endif

Python/pystate.c

@@ -5,6 +5,7 @@
 #include "pycore_ceval.h"
 #include "pycore_code.h"          // stats
 #include "pycore_dtoa.h"          // _dtoa_state_INIT()
+#include "pycore_emscripten_trampoline.h"  // _Py_EmscriptenTrampoline_Init()
 #include "pycore_frame.h"
 #include "pycore_initconfig.h"    // _PyStatus_OK()
 #include "pycore_object.h"        // _PyType_InitCache()
@@ -449,6 +450,10 @@ init_runtime(_PyRuntimeState *runtime,
 
     runtime->unicode_state.ids.next_index = unicode_next_index;
 
+#if defined(__EMSCRIPTEN__) && defined(PY_CALL_TRAMPOLINE)
+    _Py_EmscriptenTrampoline_Init(runtime);
+#endif
+
     runtime->_initialized = 1;
 }