Eliminated ENABLE_128_BIT_SUPPORT and union ValueData from Scalar.cpp and use llvm::APInt and llvm::APFloat for all integer and floating point types. Also used Scalar in RegisterValue.cpp
I have corrected the patch for Scalar which was reverted in r245222 and addressed all concerns raised (http://reviews.llvm.org/rL245216) by Pavel Labath.
Could you please review it?
Looks good as far as I am concerned.
Nit: you can initialize m_scalar within initializer list.
Could you pass fail_value as a const reference?
Addressed review comments
Yes, I have changed fail_value to const reference.