[llvm] [AMDGPU] Use alias analysis for SMEM/VMEM WAR waitcnts (PR #206569)

[Matt Arsenault via llvm-commits]

================
@@ -369,7 +369,7 @@ struct PreheaderFlushFlags {
 };
 
 class SIInsertWaitcnts {
-  DenseMap<const Value *, MachineBasicBlock *> SLoadAddresses;
+  SmallVector<MachineInstr *, 8> SLoadAddresses;
----------------
arsenm wrote:

These aren't tracking addresses anymore 

https://github.com/llvm/llvm-project/pull/206569