LLVM bitcode to MIR translator

Reasons for implementing the translator

• Facilitating future MIR support for more programming languages and their extensions
• Facilitating future MIR support for more ABIs
• Facilitating future MIR support for optimization hints
• Facilitating future MIR support for cross-compiling and support for more targets

The current state of the translator

• Currently we are focused on translating only native LLVM bitcode of standard C/C++
• LLVM bitcode vectors are no supported yet. In future the vector code will be scalarized to be translated into MIR
  • Use -fno-vectorize to avoid LLVM bitcode vectors
• Full x86-64 call ABI (multiple return regs) are not implemented yet
• Many optimization hints are ignored. Others should result in a compilation failure
• To implement the translator, we use LLVM C interface as more stable one