[x86] Teach the new vector shuffle lowering to aggressively form MOVSS
[oota-llvm.git] / test / CodeGen / X86 / combiner-aa-1.ll
2009-09-15 Dan GohmanDon't pull a load through a callseq_start if the load...