[jit][edge] Fix array index checking in mobile interpreter. (pytorch#…

…73241)

Summary:
Pull Request resolved: pytorch#73241

Stop using non-portable out-of-range indexing in mobile interpreter, also change code types indexing to use vector.at() to catch out-of-range bugs earlier.

Test Plan: buck test mode/dbg mode/no-gpu -c fbcode.platform=platform010 //caffe2/test/cpp/jit:jit -- BackendTest.TestCompiler

Reviewed By: dhruvbird, r-barnes

Differential Revision: D34370237

fbshipit-source-id: 1827f75ed00ecc10bbcece48329b0ac87189b079
(cherry picked from commit ab943ef)

torch/csrc/jit/mobile/interpreter.cpp

@@ -233,7 +233,7 @@ bool InterpreterState::run(Stack& stack) {
           }
           return false;
         case LIST_CONSTRUCT: {
-          listConstruct(stack, *code.types_[inst.X], inst.N);
+          listConstruct(stack, *code.types_.at(inst.X), inst.N);
           frame.step();
         } break;
         case LIST_UNPACK: {
@@ -305,21 +305,20 @@ bool InterpreterState::run(Stack& stack) {
           frame.step();
         } break;
         case DICT_CONSTRUCT: {
-          dictConstruct(stack, *code.types_[inst.X], inst.N);
+          dictConstruct(stack, *code.types_.at(inst.X), inst.N);
           frame.step();
         } break;
         case NAMED_TUPLE_CONSTRUCT: {
-          namedTupleConstruct(stack, code.types_[inst.X], inst.N);
+          namedTupleConstruct(stack, code.types_.at(inst.X), inst.N);
           frame.step();
         } break;
         case CREATE_OBJECT: {
-          auto type = code.types_[inst.X]->expect<c10::ClassType>();
+          auto type = code.types_.at(inst.X)->expect<c10::ClassType>();
           createObject(stack, type);
           frame.step();
         } break;
         case ISINSTANCE: {
-          at::ArrayRef<TypePtr> types(
-              &(code.types_[inst.X]), &(code.types_[inst.X + inst.N]));
+          at::ArrayRef<TypePtr> types(&code.types_.at(inst.X), inst.N);
           isinstance(stack, types);
           frame.step();
         } break;