diff options
| author | Jack Huey <31162821+jackh726@users.noreply.github.com> | 2021-02-02 16:01:38 -0500 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2021-02-02 16:01:38 -0500 |
| commit | 76be6bb4de596ad24c62dd52baeb0cb56c8937ae (patch) | |
| tree | bee4e8d6d81ac203d243a9a49276d3de41cedbfc /library/std/src | |
| parent | 71792d822c713b66616b2b5dde1fa9cffe3ab21d (diff) | |
| parent | c7f4154c6a839d31abcbb74be4c9b2404ae3a2ec (diff) | |
| download | rust-76be6bb4de596ad24c62dd52baeb0cb56c8937ae.tar.gz rust-76be6bb4de596ad24c62dd52baeb0cb56c8937ae.zip | |
Rollup merge of #81530 - ojeda:sys-use-abort-instead-of-wasm32-unreachable, r=Mark-Simulacrum
sys: use `process::abort()` instead of `arch::wasm32::unreachable()` Rationale: - `abort()` lowers to `wasm32::unreachable()` anyway. - `abort()` isn't `unsafe`. - `abort()` matches the comment better. - `abort()` avoids confusion by future readers (e.g. https://github.com/rust-lang/rust/pull/81527): the naming of wasm's `unreachable` instruction is a bit unfortunate because it is not related to the `unreachable()` intrinsic (intended to trigger UB). Codegen is likely to be different since `unreachable()` is `inline` while `abort()` is `cold`. Since it doesn't look like we are expecting here to trigger this case, the latter seems better anyway.
Diffstat (limited to 'library/std/src')
| -rw-r--r-- | library/std/src/sys/wasm/thread.rs | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/library/std/src/sys/wasm/thread.rs b/library/std/src/sys/wasm/thread.rs index 95a9230aa78..5eafb77da1d 100644 --- a/library/std/src/sys/wasm/thread.rs +++ b/library/std/src/sys/wasm/thread.rs @@ -86,7 +86,7 @@ pub fn my_id() -> u32 { if MY_ID == 0 { let mut cur = NEXT_ID.load(SeqCst); MY_ID = loop { - let next = cur.checked_add(1).unwrap_or_else(|| crate::arch::wasm32::unreachable()); + let next = cur.checked_add(1).unwrap_or_else(|| crate::process::abort()); match NEXT_ID.compare_exchange(cur, next, SeqCst, SeqCst) { Ok(_) => break next, Err(i) => cur = i, |