[llvm] [AMDGPU] Add support for preloading implicit kernel arguments (PR #83817)

[Matt Arsenault via llvm-commits]

================
@@ -5495,6 +5495,34 @@ bool AMDGPULegalizerInfo::legalizeLDSKernelId(MachineInstr &MI,
   return true;
 }
 
+bool AMDGPULegalizerInfo::legalizePreloadImplicitarg(
+    MachineInstr &MI, MachineRegisterInfo &MRI, MachineIRBuilder &B) const {
+  assert(ST.hasKernargPreload());
----------------
arsenm wrote:

This shouldn't be an assert. Better to just return false and let the intrinsic fail to select 

https://github.com/llvm/llvm-project/pull/83817