Allow to fold vector load if there is more than one bitcast, so in the case:
[oota-llvm.git] / test / CodeGen / X86 / shift-bmi2.ll
2012-09-26 Michael LiaoAdd SARX/SHRX/SHLX code generation support