AMDGPU: Make SI_END_CF convergent

[arsenm]

Summary

This part has no test changes.

[sameerds]

Same question as D155305. As far as analogies go, is it correct to say that if SI_IF produced a token to represent threads that reached it convergently, then SI_END_CF would use the same token to specify that those threads need to reconverge here?

[arsenm]

I'm starting to form a better idea for how to represent this

llvm/lib/Target/AMDGPU/SIInstructions.td

def SI_END_CF : CFPseudoInstSI <
   (outs), (ins SReg_1:$saved), [], 1, 1> {
   let Size = 4;
   let isAsCheapAsAMove = 1;
   let isReMaterializable = 1;
   let hasSideEffects = 1;
   let isNotDuplicable = 1; // Not a hard requirement, see long comments above for details.
   let mayLoad = 1; // FIXME: Should not need memory flags
   let mayStore = 1;
+  let isConvergent = 1;
 }
 
 def SI_IF_BREAK : CFPseudoInstSI <
   (outs SReg_1:$dst), (ins SReg_1:$vcc, SReg_1:$src), []> {
   let Size = 4;
   let isNotDuplicable = 1; // Not a hard requirement, see long comments above for details.
   let isAsCheapAsAMove = 1;
   let isReMaterializable = 1;