Considering add a #[cfg] option for musl/glibc split

[alexcrichton]

Right now target_env is predominately used to conditionally compile code for differing C libraries like msvc, glibc, musl, etc. On ARM, however, there are two glibc target_env annotations, "gnueabi" and "gneuabihf". This then also extends to musl with "musleabi" and "musleabihf". This seems unnecessarily complicated to say "conditionally compile this code when using musl", that is it should only require one annotation, not three.

We could possibly change target_env as I doubt anyone's using "gnueabihf" as a matcher, or we could add a new target_libc.

[alexcrichton]

cc #33189

[nagisa]

The presence of absence of hard float should be denoted inside its own feature instead (cfg(target_has_hardfloat) or something). I believe we already have (or plans for) a few such switches already.

[retep998]

Could we use target_libc for choosing between static vs dynamic or release vs debug for the msvc CRT?

[alexcrichton]

The toosl team discussed this issue at triage and the decision was to name target_env as either gnu or musl instead of also gnueabi/musleabi. I'll send a PR shortly

[ollie27]

There should probably also be a good way to distinguish between statically linked musl and dynamically linked musl targets.