do not call black_box on Miri

[RalfJung]

Helps with #75274 (but #74932 introduced unrelated breakage that will need a separate fix)
Cc @eggyal r? @Mark-Simulacrum

[eggyal]

Since black_box is only a best effort, wouldn't it be better to omit the llvm_asm within it when compiling miri rather than omitting the call altogether? Else other attempts to call black_box (if there ever are any) would also have to be explicitly omitted?

[RalfJung]

I was considering that, but current black_box is in a weird state until #64102 has been implemented.

But I noticed the comment already says this is best-effort, so I think adding this cfg in there is indeed allowed. I'll do that, thanks.

[oli-obk]

I agree, we should just #[cfg(miri)]

rust/library/core/src/hint.rs

Line 124 in e61621c

llvm_asm!("" : : "r"(&dummy));

and allow the use of black_box everywhere.

[oli-obk]

r? @oli-obk

r=me with CI passing or local tests giving high confidence in them passing

[RalfJung]

@bors r=oli-obk p=1
(higher priority to unblock a Miri fix)

[bors]

📌 Commit 8385146 has been approved by oli-obk

[RalfJung]

@bors rollup

[eggyal]

This should close #75274, no?

[RalfJung]

No, see the OP of this PR.

rust-lang/miri#1500 will have to land in Miri (and we need a submodule update in rustc) to make Miri build again.

[bors]

⌛ Testing commit 8385146 with merge c92fc8d...

[bors]

☀️ Test successful - checks-actions, checks-azure
Approved by: oli-obk
Pushing c92fc8d to master...