Add spin_loop_hint intrinsic

[bjorn3]

This removes one occurrence of inline assembly in libcore, leaving only:

src/libcore/num/dec2flt/algorithm.rs
61:        unsafe { asm!("fldcw $0" :: "m" (cw) :: "volatile") }
77:        unsafe { asm!("fnstcw $0" : "=*m" (&cw) ::: "volatile") }

This makes life easier for alternative codegen backends which don't (yet) support inline asm by not needing to patch libcore to remove it.

cc https://github.com/bjorn3/rustc_codegen_cranelift/issues/337

[rust-highfive]

r? @estebank

(rust_highfive has picked a reviewer for you, use r? to override)

[rust-highfive]

The job x86_64-gnu-llvm-6.0 of your PR failed on Travis (raw log). Through arcane magic we have determined that the following fragments from the build log may contain information about the problem.

Click to expand the log.

travis_time:end:0fff643e:start=1551528449638587671,finish=1551528450524937519,duration=886349848
$ git checkout -qf FETCH_HEAD
travis_fold:end:git.checkout

Encrypted environment variables have been removed for security reasons.
See https://docs.travis-ci.com/user/pull-requests/#pull-requests-and-security-restrictions
$ export SCCACHE_BUCKET=rust-lang-ci-sccache2
$ export SCCACHE_REGION=us-west-1
Setting environment variables from .travis.yml
$ export IMAGE=x86_64-gnu-llvm-6.0
---
[00:26:25]    Compiling env_logger v0.5.13
[00:26:29]    Compiling flate2 v1.0.6
[00:26:29]    Compiling backtrace v0.3.11
[00:26:35]    Compiling tempfile v3.0.5
[00:26:36] error: internal compiler error: src/librustc_codegen_llvm/intrinsic.rs:709: broken spin_loop_hint
[00:26:36] thread 'rustc' panicked at 'Box<Any>', src/librustc_errors/lib.rs:620:9
[00:26:36] note: Run with `RUST_BACKTRACE=1` environment variable to display a backtrace.
[00:26:36] error: aborting due to previous error
[00:26:36] 
[00:26:36] 
[00:26:36] 
[00:26:36] note: the compiler unexpectedly panicked. this is a bug.
[00:26:36] 
[00:26:36] note: we would appreciate a bug report: https://github.com/rust-lang/rust/blob/master/CONTRIBUTING.md#bug-reports
[00:26:36] 
[00:26:36] note: rustc 1.34.0-dev running on x86_64-unknown-linux-gnu
[00:26:36] 
[00:26:36] note: compiler flags: -Z external-macro-backtrace -Z force-unstable-if-unmarked -C opt-level=2 -C prefer-dynamic -C debug-assertions=y -C link-args=-Wl,-rpath,$ORIGIN/../lib --crate-type lib
[00:26:36] note: some of the compiler flags provided by cargo are hidden
[00:26:36] 
[00:26:36] error: Could not compile `parking_lot_core`.
[00:26:36] warning: build failed, waiting for other jobs to finish...
---
181352 ./obj/build/x86_64-unknown-linux-gnu/stage0-rustc
156416 ./src/llvm-project/clang
156004 ./obj/build/bootstrap/debug/incremental
141232 ./obj/build/bootstrap/debug/incremental/bootstrap-3i6jt5nchoxcn
141228 ./obj/build/bootstrap/debug/incremental/bootstrap-3i6jt5nchoxcn/s-f9yyeooc9d-1jcjbmg-3r437gq9en0c0
139296 ./obj/build/x86_64-unknown-linux-gnu/stage0-rustc/x86_64-unknown-linux-gnu/release
136416 ./obj/build/x86_64-unknown-linux-gnu/stage0-rustc/x86_64-unknown-linux-gnu/release/deps
123628 ./src/llvm-project/llvm/test/CodeGen
108532 ./src/llvm-project/lldb

I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact @TimNN. (Feature Requests)

[bjorn3]

What did I do wrong?

[nagisa]

This can (should) be implemented as a call to @llvm.x86.sse2.pause() on x86/x86_64.

[nagisa]

(i.e. __mm_pause on x86 and an equivalent on Aarch still needs to be implemented(?))

[nagisa]

I have a feeling that these instructions should be exposed by coresimd instead.

[estebank]

r? @nagisa

[bjorn3]

I think this should remain in core::hint as multiple archs have a similar instruction and it is just a hint, so it should just compile on archs which don't have such an instruction without having to use #[cfg].

[nagisa]

I don’t mind them being exposed by core::hint, but I think they should be implemented in terms of coresimd, which is responsible for implementing all such instructions either way.

[bjorn3]

I see. Maybe I will send coresimd a PR tomorrow.