Fix some tests failing in --pass check mode - #72007
Merged
Merged
Conversation
rust-highfive
added
the
S-waiting-on-review
Member
Collaborator
|
📌 Commit 13c9312 has been approved by |
bors
added
S-waiting-on-bors
bors
added a commit
to rust-lang-ci/rust
that referenced
this pull request
May 8, 2020
Rollup of 6 pull requests Successful merges: - rust-lang#71989 (Use a single enum for the kind of a const context) - rust-lang#71993 (Remove old `util/liveness.rs` module) - rust-lang#71999 (Add myself to mailmap.) - rust-lang#72001 (Adjust cfg(version) to lang team decision) - rust-lang#72007 (Fix some tests failing in `--pass check` mode) - rust-lang#72008 (Add const-generics test) Failed merges: r? @ghost
oddurs
added a commit
to oddurs/poptop
that referenced
this pull request
Sep 20, 2026
…ocess `a_running_process_named_in_bytes_that_are_not_utf8_is_in_the_table` copies `sleep` to a name that is not UTF-8 and runs it. On CI it failed with ExecutableFileBusy — "Text file busy" — from the spawn, not the copy. `fs::copy` opens the destination for writing and the fd is close-on-exec, not close-on-fork. Any other test thread that spawns a process forks in the window before that fd is closed, and the child holds a writable descriptor to the file until it execs. Until then the kernel refuses to exec it. The window is microseconds and the suite spawns processes from several tests at once, so it shows up as a flake on a loaded runner and never locally. rust-lang/rust#72007. A bounded retry is the whole fix. Five seconds, twenty milliseconds apart, and any other error still fails immediately with the message. Unrelated to the rest of this branch; it turned CI red on it.
oddurs
added a commit
to oddurs/poptop
that referenced
this pull request
Sep 20, 2026
…gram working (#141) * fix(tools): a screenshot that renders in a browser, and shows the program working Three things wrong with the first one, all found by rasterising it and looking at it rather than by checking that it parsed. **Braille rendered as tofu.** The timeline and the HIST column are U+2800-U+28FF, an SVG names a font stack rather than carrying one, and most readers have no braille — it came out as rows of dotted placeholder boxes, which does not look like a font problem, it looks like a bug in poptop. Braille is now drawn as rectangles, two by four to a cell, which needs no font at all. Everything else stays text, so it can still be selected and searched. **The frame was of an empty buffer.** Nine seconds of history under a program whose argument is that it keeps every sample. The capture now lets the timeline fill first. **The frame was of an idle machine, and then of a machine pinned flat.** A constant load draws a featureless block; a purely bursty one leaves the table empty between bursts. The workload is now a build from an empty target directory — fifty rustc in waves, which is what a build machine looks like — with compression and copy bursts on top, and the capture walks back through the buffer with the arrow keys and returns to the fullest sample that was also busy. That chooses which recorded moment to show; it does not change the moment, and finding the busy one is what a reader would do with the same keys. Looking at the result turned up 0134: ten rustc the build had just spawned, each holding 140 MB, each reported at 0.0% CPU with an em dash beside it in the disk columns. Both are a delta with no previous sample. One of them says so. * test(linux): retry the exec that races every other test spawning a process `a_running_process_named_in_bytes_that_are_not_utf8_is_in_the_table` copies `sleep` to a name that is not UTF-8 and runs it. On CI it failed with ExecutableFileBusy — "Text file busy" — from the spawn, not the copy. `fs::copy` opens the destination for writing and the fd is close-on-exec, not close-on-fork. Any other test thread that spawns a process forks in the window before that fd is closed, and the child holds a writable descriptor to the file until it execs. Until then the kernel refuses to exec it. The window is microseconds and the suite spawns processes from several tests at once, so it shows up as a flake on a loaded runner and never locally. rust-lang/rust#72007. A bounded retry is the whole fix. Five seconds, twenty milliseconds apart, and any other error still fails immediately with the message. Unrelated to the rest of this branch; it turned CI red on it.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
r? @RalfJung