unresolved macro call on standard library macro.

[Lokathor]

The following code triggers an unresolved macro problem with RA nightly 2021-06-20:

// main.rs
fn main() {
  println!("sse2: {}", cfg!(target_feature = "sse2"));
  println!("sse3: {}", cfg!(target_feature = "sse3"));
  println!("ssse3: {}", cfg!(target_feature = "ssse3"));
  println!("sse4.1: {}", cfg!(target_feature = "sse4.1"));
  println!("sse4.2: {}", cfg!(target_feature = "sse4.2"));
  println!("runtime rdrand: {}", std::is_x86_feature_detected!("rdrand"));
  println!("runtime rdseed: {}", std::is_x86_feature_detected!("rdseed"));
}

Given that I'm building for x86_64 windows, the macro should be in the standard library just fine. Indeed, the program compiles and runs perfectly with cargo run, only RA itself is having the hiccup.

[Veykril]

The macro re-export is gated behind a simd_x86, https://github.com/rust-lang/rust/blob/6b354a13820a444f834a33365ae4a8d97d7d27ce/library/std/src/lib.rs#L541-L542.

When is this supposed to be set? Querying the cfg via rustc --print cfg for my windows target doesn't list it.

[bjorn3]

That is a stability marker, not a cfg.

[Veykril]

Wasn't quite awake yet I suppose 🤦

[Veykril]

So ye the problem is we can't resolve std_detect in std it seems.

[bjorn3]

It probably needs to be added to the list of sysroot crates.

[jonas-schievink]

stdarch is full of cfg_if!, so it probably won't work even then

[jonas-schievink]

With #9357 the crate is now loaded correctly.

cc #7637 for the cfg_if! issue

[Veykril]

Since this seems to work fine and the cfg_if! part is tracked in another issue I believe we can close this.