fuzz: Remove comments about u128 ABI issues

These have since been resolved and is no longer a concern. This also
means that indirection is no longer needed.

Author: tgross35

fuzz/ops.rs

@@ -237,8 +237,6 @@ using namespace llvm;
 #pragma clang diagnostic error "-Wextra"
 #pragma clang diagnostic error "-Wunknown-attributes"
 
-// HACK(eddyb) allow code below to assume `uintN_t` maps to Rust `uN`.
-// FIXME(eddyb) make sure this doesn't run into `u128` ABI issues.
 using uint128_t = __uint128_t;
 
 template<typename F>
@@ -374,7 +372,7 @@ struct __attribute__((packed)) {name} {{
     }}
 }};
 extern "C" {{
-    void {exported_symbol}({name} *out, const FuzzOp<{name}> &op) {{
+    void {exported_symbol}({name} *out, FuzzOp<{name}> op) {{
         *out = op.eval();
     }}
 }}"#

fuzz/src/main.rs

@@ -168,14 +168,11 @@ macro_rules! float_reprs {
 
                 fn cxx_apf_eval_fuzz_op(op: FuzzOp<Self>) -> Self {
                     extern "C" {
-                        // HACK(eddyb) the warning is about `u128` ABI issues,
-                        // which is also why indirection is used.
-                        #[allow(improper_ctypes)]
-                        fn $cxx_apf_eval_fuzz_op(out: &mut MaybeUninit<$name>, op: &FuzzOp<$name>);
+                        fn $cxx_apf_eval_fuzz_op(out: &mut MaybeUninit<$name>, op: FuzzOp<$name>);
                     }
                     unsafe {
                         let mut out = MaybeUninit::uninit();
-                        $cxx_apf_eval_fuzz_op(&mut out, &op);
+                        $cxx_apf_eval_fuzz_op(&mut out, op);
                         out.assume_init()
                     }
                 }