On Sat, 19 Feb 2022 05:51:52 GMT, Quan Anh Mai <d...@openjdk.java.net> wrote:
> Hi, > > `Objects.requireNonNull` may fail to be inlined. The call is expensive and > may lead to objects escaping to the heap while the null check is cheap and is > often elided. I have observed this when using the vector API when a call to > `Objects.requireNonNull` leads to vectors being materialised in a hot loop. > > Should the other `requireNonNull` be `ForceInline` as well? > > Thank you very much. Very simple methods like this are almost always inlined, but I suppose there are depth cutoffs that may cause inlining to fail. Yes, it should always be inlined. The optimizer works hard to track null-ness conditions, and making this call go out of line "breaks the thread of thought" in the optimizer's reasoning about null. Yes, both versions of `O::cNN` should be marked this way; the only difference is in how the exception is constructed. (A similar point goes for index-checking functions, which are also force-inlined so that simple index checks of the form `i*K+L <u R` can be strength-reduced by loop iteration range splitting.) ------------- PR: https://git.openjdk.java.net/jdk/pull/7543
