[PATCH] Implement aarch64 neon instruction class AdvSIMD (by element) - LLVM

[Jiangning Liu]

  Tim,

  I added some new pattern matches to support the fms_lane intrinsics.

  Thanks,
  -Jiangning


================
Comment at: lib/Target/AArch64/AArch64InstrNEON.td:3687
@@ +3686,3 @@
+
+  def : NI_2VEswap_laneq<!cast<Instruction>(subop # "_2s4s"),
+                         neon_uimm2_bare, op, VPR64, VPR128, v2f32, v4f32,
----------------
Some new patterns are added to capture a + b * (-c).


http://llvm-reviews.chandlerc.com/D1753