a4b9e2502c
Clang 8 and Clang 11 both have this optimization, but it was moved from LLVM to Clang: https://github.com/llvm/llvm-project/compare/c5b890e922432bd80a5e3c6d82994ef4cdc41900...a0aa58dad53f2e5e90a9b7079ec31bf7c3144fc7 So Kotlin/Native compiler now has to enable this optimization explicitly.