Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Avoid quadratic complexity when splitting output into lines #9586

Merged
merged 2 commits into from
Jun 18, 2021
Merged

Avoid quadratic complexity when splitting output into lines #9586

merged 2 commits into from
Jun 18, 2021

Conversation

tmiasko
Copy link
Contributor

@tmiasko tmiasko commented Jun 15, 2021

When searching for newlines in a process output keep track which part of
buffer was already examined to avoid processing the same data again and
again.

@rust-highfive
Copy link

r? @alexcrichton

(rust-highfive has picked a reviewer for you, use r? to override)

@rust-highfive rust-highfive added the S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. label Jun 15, 2021
@alexcrichton
Copy link
Member

Thanks for this! Can you describe how you saw this arise as well? I'm curious why this hasn't been caught before or if it's only a niche-ish situation where the perf matters.

Additionally would you be up for separating out this structure to a separate thing which can be unit-tested? I fear that the internals of the management of lines here has grown with complications to the point that it's probably best to have unit tests to ensure we don't accidentally regress things.

@tmiasko
Copy link
Contributor Author

tmiasko commented Jun 15, 2021

I encountered this while debugging an issue in rustc, more than once in fact. Each time I had a logging enabled which generated very long lines. The end result was a cargo stuck while using 100% CPU without making any visible progress.

I would rather not take a job of refactoring this to the point where it can be unit tested.

@alexcrichton
Copy link
Member

Hm ok, well unfortunately I'm not personally confident enough in this code to sign off on it. I would be much more confident with some unit tests showing what's happening.

I'd be more ok landing this if it were a pressing problem, but if it's mostly just debugging rustc then while I think this is important to fix it's best to fix it in a way to that doesn't run the risk of regressing other use cases in Cargo.

@tmiasko
Copy link
Contributor Author

tmiasko commented Jun 17, 2021

I introduced an intermediate refactoring, so that there is only a single place that drains the data buffer. With that in place, the optimization itself is straightforward.

Could you take another look? If you consider that too risky still, that's fair of course.

Copy link
Member

@alexcrichton alexcrichton left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Ok yeah this looks more understandable, thanks! I left a few small comments inline.

crates/cargo-util/src/process_builder.rs Show resolved Hide resolved
crates/cargo-util/src/process_builder.rs Outdated Show resolved Hide resolved
When searching for newlines in a process output keep track which part of
buffer was already examined to avoid processing the same data again and
again.
@alexcrichton
Copy link
Member

@bors: r+

@bors
Copy link
Contributor

bors commented Jun 18, 2021

📌 Commit 8df771f has been approved by alexcrichton

@bors bors added S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion. and removed S-waiting-on-review Status: Awaiting review from the assignee but also interested parties. labels Jun 18, 2021
@bors
Copy link
Contributor

bors commented Jun 18, 2021

⌛ Testing commit 8df771f with merge 9a3c162...

@bors
Copy link
Contributor

bors commented Jun 18, 2021

☀️ Test successful - checks-actions
Approved by: alexcrichton
Pushing 9a3c162 to master...

@bors bors merged commit 9a3c162 into rust-lang:master Jun 18, 2021
@tmiasko tmiasko deleted the lines branch June 18, 2021 20:36
JohnTitor added a commit to JohnTitor/rust that referenced this pull request Jun 24, 2021
Update cargo

This also updates `opener` used in bootstrap (to try to keep dependencies unified).

18 commits in 44456677b5d1d82fe981c955dc5c67734b31f340..9233aa06c801801cff75df65df718d70905a235e
2021-06-12 18:00:01 +0000 to 2021-06-22 21:32:55 +0000
- Detect incorrectly named cargo.toml (rust-lang/cargo#9607)
- Unify weak and namespaced features. (rust-lang/cargo#9574)
- Change `rustc-cdylib-link-arg` error to a warning. (rust-lang/cargo#9563)
- Updates to future-incompatible reporting. (rust-lang/cargo#9606)
- Add a compatibility notice for diesel and the new resolver. (rust-lang/cargo#9602)
- Don't allow config env to modify vars set by cargo (rust-lang/cargo#9579)
- Disambiguate is_symlink. (rust-lang/cargo#9604)
- Update opener requirement from 0.4 to 0.5 (rust-lang/cargo#9583)
- Avoid quadratic complexity when splitting output into lines (rust-lang/cargo#9586)
- Bump to 0.56.0, update changelog (rust-lang/cargo#9597)
- Fix dep-info files including non-local build script paths. (rust-lang/cargo#9596)
- Relax doc collision error. (rust-lang/cargo#9595)
- Handle "jobs = 0" case in cargo config files (rust-lang/cargo#9584)
- Enhancements to testsuite error output. (rust-lang/cargo#9589)
- Fix typo (rust-lang/cargo#9590)
- Enable support for fix --edition for 2021. (rust-lang/cargo#9588)
- Add more details for installing git repository errors (rust-lang/cargo#9582)
- More information for links conflicting (rust-lang/cargo#9568)
@ehuss ehuss added this to the 1.55.0 milestone Feb 6, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
S-waiting-on-bors Status: Waiting on bors to run and complete tests. Bors will change the label on completion.
Projects
None yet
Development

Successfully merging this pull request may close these issues.

5 participants