Rollup merge of #142563 - jieyouxu:no-more-i686-mingw, r=mati865

Refine run-make test ignores due to unpredictable `i686-pc-windows-gnu` unwind mechanism

Closes rust-lang/rust#128911. This PR *re-enables* the `dump-ice-to-disk` test and the unstable feature usage metrics tests for {x86_64,i686} Windows MSVC hosts and x86_64 Windows GNU host. I'll keep an eye out for these tests, and will broaden the ignores if this test is still flaky on not just `i686-pc-windows-gnu`.

r? mati865

try-job: x86_64-msvc-1
try-job: i686-msvc-1
try-job: x86_64-mingw-1
This commit is contained in:
Jubilee 2025-06-17 00:28:16 -07:00 committed by GitHub
commit f2e9e2afac
No known key found for this signature in database
GPG key ID: B5690EEEBB952194
3 changed files with 16 additions and 23 deletions

View file

@ -17,14 +17,14 @@
//!
//! # Test history
//!
//! - The previous rmake.rs iteration of this test was flakey for unknown reason on `i686-mingw`
//! *specifically*, so assertion failures in this test was made extremely verbose to help
//! diagnose why the ICE messages was different *specifically* on `i686-mingw`.
//! - An attempt is made to re-enable this test on `i686-mingw` (by removing `ignore-windows`). If
//! this test is still flakey, please restore the `ignore-windows` directive.
//! The previous rmake.rs iteration of this test was flaky for unknown reason on
//! `i686-pc-windows-gnu` *specifically*, so assertion failures in this test was made extremely
//! verbose to help diagnose why the ICE messages was different. It appears that backtraces on
//! `i686-pc-windows-gnu` specifically are quite unpredictable in how many backtrace frames are
//! involved.
//@ ignore-windows
//FIXME(#128911): still flakey on i686-mingw.
//@ ignore-cross-compile (exercising ICE dump on host)
//@ ignore-i686-pc-windows-gnu (unwind mechanism produces unpredictable backtraces)
use std::cell::OnceCell;
use std::path::{Path, PathBuf};

View file

@ -7,12 +7,11 @@
//!
//! # Test history
//!
//! - forked from dump-ice-to-disk test, which has flakeyness issues on i686-mingw, I'm assuming
//! those will be present in this test as well on the same platform
//! - Forked from `dump-ice-to-disk` test, which previously had backtrace unpredictability on
//! `i686-pc-windows-gnu`.
//@ needs-target-std
//@ ignore-windows
//FIXME(#128911): still flakey on i686-mingw.
//@ ignore-cross-compile (exercises metrics incremental on host)
//@ ignore-i686-pc-windows-gnu (unwind mechanism produces unpredictable backtraces)
use std::path::{Path, PathBuf};
@ -87,9 +86,7 @@ fn test_metrics_errors() {
.env("RUST_BACKTRACE", "short")
.arg("-Zmetrics-dir=invaliddirectorythatdefinitelydoesntexist")
.run_fail()
.assert_stderr_contains(
"error: cannot dump feature usage metrics: No such file or directory",
)
.assert_stderr_contains("error: cannot dump feature usage metrics")
.assert_stdout_not_contains("internal compiler error");
});
}

View file

@ -7,12 +7,10 @@
//!
//! # Test history
//!
//! - forked from dump-ice-to-disk test, which has flakeyness issues on i686-mingw, I'm assuming
//! those will be present in this test as well on the same platform
//! - Forked from `dump-ice-to-disk` test, where `i686-pc-windows-gnu` has unpredictable backtraces.
//@ needs-target-std
//@ ignore-windows
//FIXME(#128911): still flakey on i686-mingw.
//@ ignore-cross-compile (exercises metrics dump on host)
//@ ignore-i686-pc-windows-gnu (unwind mechanism produces unpredictable backtraces)
use std::path::{Path, PathBuf};
@ -85,9 +83,7 @@ fn test_metrics_errors() {
.env("RUST_BACKTRACE", "short")
.arg("-Zmetrics-dir=invaliddirectorythatdefinitelydoesntexist")
.run_fail()
.assert_stderr_contains(
"error: cannot dump feature usage metrics: No such file or directory",
)
.assert_stderr_contains("error: cannot dump feature usage metrics")
.assert_stdout_not_contains("internal compiler error");
});
}