Cannot run with libstd MIR on macOS due to stack guards

[skade]

Executing any Rust program on macOS (and I assume anything but Linux) leads to the following crash:

[ skade miri ] cargo run --bin miri -- --sysroot ~/.xargo/HOST tests/run-pass/hello.rs
    Finished dev [unoptimized + debuginfo] target(s) in 0.0 secs
     Running `target/debug/miri --sysroot /Users/skade/.xargo/HOST tests/run-pass/hello.rs`
WARN:miri::terminator: ignoring C ABI call: pthread_self
WARN:miri::terminator: ignoring C ABI call: pthread_get_stackaddr_np
WARN:miri::terminator: ignoring C ABI call: pthread_self
WARN:miri::terminator: ignoring C ABI call: pthread_get_stacksize_np
error: can't call C ABI function: mmap
   --> /Users/skade/.rustup/toolchains/nightly-x86_64-apple-darwin/lib/rustlib/src/rust/src/libstd/sys/unix/thread.rs:283:26
    |
283 |               let result = mmap(stackaddr, psize, PROT_NONE,
    |  __________________________^
284 | |                               MAP_PRIVATE | MAP_ANON | MAP_FIXED, -1, 0);
    | |________________________________________________________________________^
    |
note: inside call to std::sys::imp::thread::guard::init
   --> /Users/skade/.rustup/toolchains/nightly-x86_64-apple-darwin/lib/rustlib/src/rust/src/libstd/rt.rs:44:26
    |
44  |         let main_guard = sys::thread::guard::init();
    |                          ^^^^^^^^^^^^^^^^^^^^^^^^^^
note: inside call to std::rt::lang_start
   --> /Users/skade/.rustup/toolchains/nightly-x86_64-apple-darwin/lib/rustlib/src/rust/src/libstd/rt.rs:32:1
    |
32  | / fn lang_start(main: fn(), argc: isize, argv: *const *const u8) -> isize {
33  | |     use panic;
34  | |     use sys;
35  | |     use sys_common;
...   |
72  | |     }
73  | | }
    | |_^

error: aborting due to previous error

thread 'main' panicked at 'Box<Any>', src/librustc_errors/lib.rs:526:8
note: Run with `RUST_BACKTRACE=1` for a backtrace.

[RalfJung]

I assume anything but Linux

Certainly not Windows as that's covered by Travis (well only via cross compilation, but that should not make a difference...).
@oli-obk maybe macOS should be corss-tested on Travis as well? Is there any reason this should not work? (Of course it would not have found this bug, but once #202 is solved, it will.)

But it seems that non-Linux unixes still use mmap. Seems we have to revert #239. :(

[skade]

Certainly not Windows as that's covered by Travis.

Ah, sorry 😆 . I had quick glance through the relevant code and from "not linux" followed "everything else". The filename unix.rs should have been a tell that there was some context :).

[oli-obk]

We could add a mac Travis in order to prevent such things from happening.

[RalfJung]

I think we now do have that mac job, and it is passing. Closing the issue.

@skade Please reopen if it still does not work.

[skade]

@RalfJung I recently switched off macOS, so I can't test ;). Thanks!