[PATCH] ARM: allow machine-CSE on litpool materialisations of a global

Renato Golin renato.golin at linaro.org
Sat Nov 30 02:25:03 PST 2013


  If I got it right, you moved the special logic to creating separate node types, that will be naturally commoned up as others, with the added benefit that the code related to them is in one single place where it makes sense, rather than special cases all over the place.

  I'm not a Darwin specialist, but apart from my silly comment below, your code looks good to me.

Comment at: lib/Target/ARM/ARMExpandPseudoInsts.cpp:943
@@ +942,3 @@
+      if (IsPIC) {
+        MachineInstrBuilder MIB =
+          BuildMI(MBB, MBBI, MI.getDebugLoc(), TII->get(PICAddOpc))
I know they don't clash, but would be good to give it a different name, here.


More information about the llvm-commits mailing list