VM: Add smi fast path operations for precompiled code

runtime/vm/flow_graph_compiler_ia32.cc

Index: runtime/vm/flow_graph_compiler_ia32.cc
diff --git a/runtime/vm/flow_graph_compiler_ia32.cc b/runtime/vm/flow_graph_compiler_ia32.cc
index 2ede9e0aaa381ddacdd33539761b4a09d3361ca2..7407ac90bc5892673526558673d17c4f115fa7ef 100644
--- a/runtime/vm/flow_graph_compiler_ia32.cc
+++ b/runtime/vm/flow_graph_compiler_ia32.cc
@@ -1290,7 +1290,8 @@ void FlowGraphCompiler::EmitMegamorphicInstanceCall(
     intptr_t deopt_id,
     TokenPosition token_pos,
     LocationSummary* locs,
-    intptr_t try_index) {
+    intptr_t try_index,
+    intptr_t slow_path_argument_count) {
   const String& name = String::Handle(zone(), ic_data.target_name());
   const Array& arguments_descriptor =
       Array::ZoneHandle(zone(), ic_data.arguments_descriptor());
@@ -1310,7 +1311,7 @@ void FlowGraphCompiler::EmitMegamorphicInstanceCall(
 
   AddCurrentDescriptor(RawPcDescriptors::kOther,
       Thread::kNoDeoptId, token_pos);
-  RecordSafepoint(locs);
+  RecordSafepoint(locs, slow_path_argument_count);
   const intptr_t deopt_id_after = Thread::ToDeoptAfter(deopt_id);
   // Precompilation not implemented on ia32 platform.
   ASSERT(!FLAG_precompiled_mode);