[libc-commits] [libc] [libc][baremetal] track some potential codegen optimization for thumbv8 target (PR #211920)

[Daniel Thornburgh via libc-commits]

mysterymath wrote:

> @mysterymath could you confirm why non-power-of-two alignment was practiced?

No particular reason other than since I wasn't heavily optimizing the O(1) behavior of the code, I had no time or space complexity reason to restrict alignment to powers of two. I don't know of a reason to allow it though; it doesn't come up practically AFAIK, and POSIX only requires powers of two. Seems like a fine tradeoff for faster code.

https://github.com/llvm/llvm-project/pull/211920