[PATCH] D48170: ARM: use "add" instead of "orr" for code size

[Eli Friedman via Phabricator via llvm-commits]

efriedma added inline comments.


================
Comment at: llvm/lib/Target/ARM/ARMInstrThumb.td:1016
+
+  def : Pat<(or AddLikeOrOp:$Rn, non_imm32:$Rm), (tADDrr $Rn, $Rm)>;
+}
----------------
T1Pat?

Are you intentionally avoiding transforming `x|256`?


Repository:
  rL LLVM

https://reviews.llvm.org/D48170