[PATCH] D63731: [AMDGPU] Prevent VGPR copies from moving across the EXEC mask definitions

[Alexander via Phabricator via llvm-commits]

alex-t marked 2 inline comments as done.
alex-t added inline comments.


================
Comment at: lib/Target/AMDGPU/SIFoldOperands.cpp:502-508
+    MachineInstr::mop_iterator ImpOpI = UseMI->implicit_operands().begin();
+    MachineInstr::mop_iterator ImpOpE = UseMI->implicit_operands().end();
+    while (ImpOpI != ImpOpE) {
+      MachineInstr::mop_iterator Tmp = ImpOpI;
+      ImpOpI++;
+      UseMI->RemoveOperand(UseMI->getOperandNo(Tmp));
+    }
----------------
arsenm wrote:
> Blindly stripping off any implicit_operands could be error prone. We do use other implicit operands on V_MOV_B32 for indirect register indexing
This is done just before calling MachineInstr::addImplicitUseDefOperands().
Also, this is under the "if" condition UseMI->isCopy() so not V_MOV exist so far.


================
Comment at: lib/Target/AMDGPU/SIInstrInfo.cpp:3923-3930
+  bool ImpDef = Def->isImplicitDef();
+  while (!ImpDef && Def && Def->isCopy()) {
+    Def = MRI.getUniqueVRegDef(Def->getOperand(1).getReg());
+    ImpDef = Def && Def->isImplicitDef();
+  }
+  if (!RI.isSGPRClass(DstRC) && !Copy->readsRegister(AMDGPU::EXEC, &RI) &&
+      !ImpDef)
----------------
arsenm wrote:
> I don't see why this is necessary 
To prevent moving VGPR copies across the EXEC definitions


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D63731/new/

https://reviews.llvm.org/D63731