[llvm] [AMDGPU][True16][CodeGen] update waitcnt for true16 (PR #128927)

[Brox Chen via llvm-commits]

================
@@ -2452,16 +2465,14 @@ bool SIInsertWaitcnts::runOnMachineFunction(MachineFunction &MF) {
 
   unsigned NumVGPRsMax = ST->getAddressableNumVGPRs();
   unsigned NumSGPRsMax = ST->getAddressableNumSGPRs();
-  assert(NumVGPRsMax <= SQ_MAX_PGM_VGPRS);
+  assert(NumVGPRsMax + AGPR_OFFSET <= SQ_MAX_PGM_VGPRS);
   assert(NumSGPRsMax <= SQ_MAX_PGM_SGPRS);
 
   RegisterEncoding Encoding = {};
-  Encoding.VGPR0 =
-      TRI->getEncodingValue(AMDGPU::VGPR0) & AMDGPU::HWEncoding::REG_IDX_MASK;
-  Encoding.VGPRL = Encoding.VGPR0 + NumVGPRsMax - 1;
-  Encoding.SGPR0 =
-      TRI->getEncodingValue(AMDGPU::SGPR0) & AMDGPU::HWEncoding::REG_IDX_MASK;
-  Encoding.SGPRL = Encoding.SGPR0 + NumSGPRsMax - 1;
+  Encoding.VGPR0 = getRegPoint(*ST, AMDGPU::VGPR0, *TRI);
+  Encoding.VGPRL = Encoding.VGPR0 + NumVGPRsMax * 2 - 1;
+  Encoding.SGPR0 = getRegPoint(*ST, AMDGPU::SGPR0, *TRI);
+  Encoding.SGPRL = Encoding.SGPR0 + NumSGPRsMax * 2 - 1;
----------------
broxigarchen wrote:

Done. I just realized we are counting offset to 0 index gprs here. Rebased

https://github.com/llvm/llvm-project/pull/128927