[PATCH] D98515: [AMDGPU][GlobalISel] Stop foldInsertEltToCmpSelect from changing reg banks

[Matt Arsenault via Phabricator via llvm-commits]

arsenm added inline comments.


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp:1982
+// differs for the one we want to set.
+static void setRegBankOrMakeACopy(MachineRegisterInfo &MRI, MachineIRBuilder &B,
+                                  MachineInstr &MI, MachineOperand &MO,
----------------
How about constrainRegToBank?


================
Comment at: llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp:1989
+    auto InsertPt = B.getInsertPt();
+    B.setInstr(MI);
+    Register Copy = B.buildCopy(MRI.getType(Reg), Reg).getReg(0);
----------------
setInstrAndDebugLoc

Isn't this set where you want it to be already? Can you just avoid all the iterator changes?


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D98515/new/

https://reviews.llvm.org/D98515