[llvm] [AMDGPU] Clean up conversion of DPP instructions in AMDGPUDisassembler (PR #82480)

[Jay Foad via llvm-commits]

================
@@ -665,6 +630,22 @@ DecodeStatus AMDGPUDisassembler::getInstruction(MCInst &MI, uint64_t &Size,
                         Address, CS);
   } while (false);
 
+  if (Res && (MCII->get(MI.getOpcode()).TSFlags & SIInstrFlags::DPP)) {
----------------
jayfoad wrote:

The ordering of these checks seems a bit ad hoc to me. Perhaps this code could be cleaned up by someone who understands what all the different `convert*DPPInst` functions do. I just tried to preserve the existing behavior by calling the same `convert` functions in the same order.

https://github.com/llvm/llvm-project/pull/82480