Implement lowered-then-lifted functions (#4327)

* Implement lowered-then-lifted functions

This commit is a few features bundled into one, culminating in the
implementation of lowered-then-lifted functions for the component model.
It's probably not going to be used all that often but this is possible
within a valid component so Wasmtime needs to do something relatively
reasonable. The main things implemented in this commit are:

* Component instances are now assigned a `RuntimeComponentInstanceIndex`
  to differentiate each one. This will be used in the future to detect
  fusion (one instance lowering a function from another instance). For
  now it's used to allocate separate `VMComponentFlags` for each
  internal component instance.

* The `CoreExport<FuncIndex>` of lowered functions was changed to a
  `CoreDef` since technically a lowered function can use another lowered
  function as the callee. This ended up being not too difficult to plumb
  through as everything else was already in place.

* A need arose to compile host-to-wasm trampolines which weren't already
  present. Currently wasm in a component is always entered through a
  host-to-wasm trampoline but core wasm modules are the source of all
  the trampolines. In the case of a lowered-then-lifted function there
  may not actually be any core wasm modules, so component objects now
  contain necessary trampolines not otherwise provided by the core wasm
  objects. This feature required splitting a new function into the
  `Compiler` trait for creating a host-to-wasm trampoline. After doing
  this core wasm compilation was also updated to leverage this which
  further enabled compiling trampolines in parallel as opposed to the
  previous synchronous compilation.

* Review comments

crates/cranelift/src/compiler.rs

@@ -288,11 +288,19 @@ impl wasmtime_environ::Compiler for Compiler {
         }))
     }
 
+    fn compile_host_to_wasm_trampoline(
+        &self,
+        ty: &WasmFuncType,
+    ) -> Result<Box<dyn Any + Send>, CompileError> {
+        self.host_to_wasm_trampoline(ty)
+            .map(|x| Box::new(x) as Box<_>)
+    }
+
     fn emit_obj(
         &self,
         translation: &ModuleTranslation,
-        types: &ModuleTypes,
         funcs: PrimaryMap<DefinedFuncIndex, Box<dyn Any + Send>>,
+        compiled_trampolines: Vec<Box<dyn Any + Send>>,
         tunables: &Tunables,
         obj: &mut Object<'static>,
     ) -> Result<(PrimaryMap<DefinedFuncIndex, FunctionInfo>, Vec<Trampoline>)> {
@@ -300,6 +308,10 @@ impl wasmtime_environ::Compiler for Compiler {
             .into_iter()
             .map(|(_i, f)| *f.downcast().unwrap())
             .collect();
+        let compiled_trampolines: Vec<CompiledFunction> = compiled_trampolines
+            .into_iter()
+            .map(|f| *f.downcast().unwrap())
+            .collect();
 
         let mut builder = ModuleTextBuilder::new(obj, &translation.module, &*self.isa);
         if self.linkopts.force_jump_veneers {
@@ -307,11 +319,6 @@ impl wasmtime_environ::Compiler for Compiler {
         }
         let mut addrs = AddressMapSection::default();
         let mut traps = TrapEncodingBuilder::default();
-        let compiled_trampolines = translation
-            .exported_signatures
-            .iter()
-            .map(|i| self.host_to_wasm_trampoline(&types[*i]))
-            .collect::<Result<Vec<_>, _>>()?;
 
         let mut func_starts = Vec::with_capacity(funcs.len());
         for (i, func) in funcs.iter() {
@@ -325,6 +332,10 @@ impl wasmtime_environ::Compiler for Compiler {
         }
 
         // Build trampolines for every signature that can be used by this module.
+        assert_eq!(
+            translation.exported_signatures.len(),
+            compiled_trampolines.len()
+        );
         let mut trampolines = Vec::with_capacity(translation.exported_signatures.len());
         for (i, func) in translation
             .exported_signatures

crates/cranelift/src/compiler/component.rs

@@ -10,9 +10,9 @@ use object::write::Object;
 use std::any::Any;
 use wasmtime_environ::component::{
     CanonicalOptions, Component, ComponentCompiler, ComponentTypes, LowerImport, LoweredIndex,
-    TrampolineInfo, VMComponentOffsets,
+    LoweringInfo, VMComponentOffsets,
 };
-use wasmtime_environ::PrimaryMap;
+use wasmtime_environ::{PrimaryMap, SignatureIndex, Trampoline};
 
 impl ComponentCompiler for Compiler {
     fn compile_lowered_trampoline(
@@ -52,6 +52,7 @@ impl ComponentCompiler for Compiler {
         let mut host_sig = ir::Signature::new(crate::wasmtime_call_conv(isa));
 
         let CanonicalOptions {
+            instance,
             memory,
             realloc,
             post_return,
@@ -71,6 +72,14 @@ impl ComponentCompiler for Compiler {
             i32::try_from(offsets.lowering_data(lowering.index)).unwrap(),
         ));
 
+        // flags: *mut VMComponentFlags
+        host_sig.params.push(ir::AbiParam::new(pointer_type));
+        callee_args.push(
+            builder
+                .ins()
+                .iadd_imm(vmctx, i64::from(offsets.flags(instance))),
+        );
+
         // memory: *mut VMMemoryDefinition
         host_sig.params.push(ir::AbiParam::new(pointer_type));
         callee_args.push(match memory {
@@ -145,32 +154,42 @@ impl ComponentCompiler for Compiler {
 
     fn emit_obj(
         &self,
-        trampolines: PrimaryMap<LoweredIndex, Box<dyn Any + Send>>,
+        lowerings: PrimaryMap<LoweredIndex, Box<dyn Any + Send>>,
+        trampolines: Vec<(SignatureIndex, Box<dyn Any + Send>)>,
         obj: &mut Object<'static>,
-    ) -> Result<PrimaryMap<LoweredIndex, TrampolineInfo>> {
-        let trampolines: PrimaryMap<LoweredIndex, CompiledFunction> = trampolines
+    ) -> Result<(PrimaryMap<LoweredIndex, LoweringInfo>, Vec<Trampoline>)> {
+        let lowerings: PrimaryMap<LoweredIndex, CompiledFunction> = lowerings
             .into_iter()
             .map(|(_, f)| *f.downcast().unwrap())
             .collect();
+        let trampolines: Vec<(SignatureIndex, CompiledFunction)> = trampolines
+            .into_iter()
+            .map(|(i, f)| (i, *f.downcast().unwrap()))
+            .collect();
+
         let module = Default::default();
         let mut text = ModuleTextBuilder::new(obj, &module, &*self.isa);
         let mut ret = PrimaryMap::new();
-        for (idx, trampoline) in trampolines.iter() {
+        for (idx, lowering) in lowerings.iter() {
             let (_symbol, range) = text.append_func(
                 false,
-                format!("_wasm_component_host_trampoline{}", idx.as_u32()).into_bytes(),
-                &trampoline,
+                format!("_wasm_component_lowering_trampoline{}", idx.as_u32()).into_bytes(),
+                &lowering,
             );
 
-            let i = ret.push(TrampolineInfo {
+            let i = ret.push(LoweringInfo {
                 start: u32::try_from(range.start).unwrap(),
                 length: u32::try_from(range.end - range.start).unwrap(),
             });
             assert_eq!(i, idx);
         }
+        let ret_trampolines = trampolines
+            .iter()
+            .map(|(i, func)| text.trampoline(*i, func))
+            .collect();
 
         text.finish()?;
 
-        Ok(ret)
+        Ok((ret, ret_trampolines))
     }
 }