Implement basic support for vectorized instructions.

We use LLVM's Scalarizer pass to remove most vectorized code so that the
Executor only needs to support the InsertElement and ExtractElement
instructions.

This pass was not available in LLVM 3.4 so to support that LLVM version
the pass has been back ported.

To check that the Executor is not receiving vector operand types
that it can't handle assertions have been added.

There are a few limitations to this implementation.

* The InsertElement and ExtractElement index cannot be symbolic.
* There is no support for LLVM < 3.4.

1 files changed, 9 insertions, 0 deletions

diff --git a/lib/Module/KModule.cpp b/lib/Module/KModule.cpp
index 005c8869..bee9a0b4 100644
--- a/lib/Module/KModule.cpp
+++ b/lib/Module/KModule.cpp
@@ -307,6 +307,15 @@ void KModule::prepare(const Interpreter::ModuleOptions &opts,
   // module.
   LegacyLLVMPassManagerTy pm;
   pm.add(new RaiseAsmPass());
+#if LLVM_VERSION_CODE >= LLVM_VERSION(3,4)
+  // This pass will scalarize as much code as possible so that the Executor
+  // does not need to handle operands of vector type for most instructions
+  // other than InsertElementInst and ExtractElementInst.
+  //
+  // NOTE: Must come before division/overshift checks because those passes
+  // don't know how to handle vector instructions.
+  pm.add(createScalarizerPass());
+#endif
   if (opts.CheckDivZero) pm.add(new DivCheckPass());
   if (opts.CheckOvershift) pm.add(new OvershiftCheckPass());
   // FIXME: This false here is to work around a bug in