Change wasm-to-host trampolines to take the values_vec size (#4192)

* Change wasm-to-host trampolines to take the values_vec size

This commit changes the ABI of wasm-to-host trampolines, which are
only used right now for functions created with `Func::new`, to pass
along the size of the `values_vec` argument. Previously the trampoline
simply received `*mut ValRaw` and assumed that it was the appropriate
size. By receiving a size as well we can thread through `&mut [ValRaw]`
internally instead of `*mut ValRaw`.

The original motivation for this is that I'm planning to leverage these
trampolines for the component model for host-defined functions. Out of
an abundance of caution of making sure that everything lines up I wanted
to be able to write down asserts about the size received at runtime
compared to the size expected. This overall led me to the desire to
thread this size parameter through on the assumption that it would not
impact performance all that much.

I ran two benchmarks locally from the `call.rs` benchmark and got:

* `sync/no-hook/wasm-to-host - nop - unchecked` - no change
* `sync/no-hook/wasm-to-host - nop-params-and-results - unchecked` - 5%
  slower

This is what I roughly expected in that if nothing actually reads the
new parameter (e.g. no arguments) then threading through the parameter
is effectively otherwise free. Otherwise though accesses to the `ValRaw`
storage is now bounds-checked internally in Wasmtime instead of assuming
it's valid, leading to the 5% slowdown (~9.6ns to ~10.3ns). If this
becomes a peformance bottleneck for a particular use case then we should
be fine to remove the bounds checking here or otherwise only bounds
check in debug mode, otherwise I plan on leaving this as-is.

Of particular note this also changes the C API for `*_unchecked`
functions where the C callback now receives the size of the array as
well.

* Add docs

crates/wasmtime/src/func.rs

@@ -367,7 +367,7 @@ impl Func {
     pub unsafe fn new_unchecked<T>(
         mut store: impl AsContextMut<Data = T>,
         ty: FuncType,
-        func: impl Fn(Caller<'_, T>, *mut ValRaw) -> Result<(), Trap> + Send + Sync + 'static,
+        func: impl Fn(Caller<'_, T>, &mut [ValRaw]) -> Result<(), Trap> + Send + Sync + 'static,
     ) -> Self {
         let store = store.as_context_mut().0;
         let host = HostFunc::new_unchecked(store.engine(), ty, func);
@@ -1024,7 +1024,7 @@ impl Func {
     fn invoke<T>(
         mut caller: Caller<'_, T>,
         ty: &FuncType,
-        values_vec: *mut ValRaw,
+        values_vec: &mut [ValRaw],
         func: &dyn Fn(Caller<'_, T>, &[Val], &mut [Val]) -> Result<(), Trap>,
     ) -> Result<(), Trap> {
         // Translate the raw JIT arguments in `values_vec` into a `Val` which
@@ -1041,7 +1041,7 @@ impl Func {
         let nparams = ty.params().len();
         val_vec.reserve(nparams + ty.results().len());
         for (i, ty) in ty.params().enumerate() {
-            val_vec.push(unsafe { Val::from_raw(&mut caller.store, *values_vec.add(i), ty) })
+            val_vec.push(unsafe { Val::from_raw(&mut caller.store, values_vec[i], ty) })
         }
 
         val_vec.extend((0..ty.results().len()).map(|_| Val::null()));
@@ -1075,7 +1075,7 @@ impl Func {
                 ));
             }
             unsafe {
-                *values_vec.add(i) = ret.to_raw(&mut caller.store);
+                values_vec[i] = ret.to_raw(&mut caller.store);
             }
         }
 
@@ -2045,9 +2045,9 @@ impl HostFunc {
     pub unsafe fn new_unchecked<T>(
         engine: &Engine,
         ty: FuncType,
-        func: impl Fn(Caller<'_, T>, *mut ValRaw) -> Result<(), Trap> + Send + Sync + 'static,
+        func: impl Fn(Caller<'_, T>, &mut [ValRaw]) -> Result<(), Trap> + Send + Sync + 'static,
     ) -> Self {
-        let func = move |caller_vmctx, values: *mut ValRaw| {
+        let func = move |caller_vmctx, values: &mut [ValRaw]| {
             Caller::<T>::with(caller_vmctx, |mut caller| {
                 caller.store.0.call_hook(CallHook::CallingHost)?;
                 let result = func(caller.sub_caller(), values)?;

crates/wasmtime/src/linker.rs

@@ -324,7 +324,7 @@ impl<T> Linker<T> {
         module: &str,
         name: &str,
         ty: FuncType,
-        func: impl Fn(Caller<'_, T>, *mut ValRaw) -> Result<(), Trap> + Send + Sync + 'static,
+        func: impl Fn(Caller<'_, T>, &mut [ValRaw]) -> Result<(), Trap> + Send + Sync + 'static,
     ) -> Result<&mut Self> {
         let func = HostFunc::new_unchecked(&self.engine, ty, func);
         let key = self.import_key(module, Some(name));

crates/wasmtime/src/trampoline/func.rs

@@ -26,8 +26,9 @@ unsafe extern "C" fn stub_fn<F>(
     vmctx: *mut VMContext,
     caller_vmctx: *mut VMContext,
     values_vec: *mut ValRaw,
+    values_vec_len: usize,
 ) where
-    F: Fn(*mut VMContext, *mut ValRaw) -> Result<(), Trap> + 'static,
+    F: Fn(*mut VMContext, &mut [ValRaw]) -> Result<(), Trap> + 'static,
 {
     // Here we are careful to use `catch_unwind` to ensure Rust panics don't
     // unwind past us. The primary reason for this is that Rust considers it UB
@@ -49,6 +50,7 @@ unsafe extern "C" fn stub_fn<F>(
         let state = (*vmctx).host_state();
         debug_assert!(state.is::<TrampolineState<F>>());
         let state = &*(state as *const _ as *const TrampolineState<F>);
+        let values_vec = std::slice::from_raw_parts_mut(values_vec, values_vec_len);
         (state.func)(caller_vmctx, values_vec)
     }));
 
@@ -109,7 +111,7 @@ pub fn create_function<F>(
     engine: &Engine,
 ) -> Result<(InstanceHandle, VMTrampoline)>
 where
-    F: Fn(*mut VMContext, *mut ValRaw) -> Result<(), Trap> + Send + Sync + 'static,
+    F: Fn(*mut VMContext, &mut [ValRaw]) -> Result<(), Trap> + Send + Sync + 'static,
 {
     let mut obj = engine.compiler().object()?;
     let (t1, t2) = engine.compiler().emit_trampoline_obj(