-
Notifications
You must be signed in to change notification settings - Fork 4.7k
io: read SO_ERROR when EPOLLERR fires instead of passing 0 errno #30230
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
Open
robobun
wants to merge
2
commits into
main
Choose a base branch
from
farm/dee5f820/fix-epollerr-zero-errno
base: main
Could not load branches
Branch not found: {{ refName }}
Loading
Could not load tags
Nothing to show
Loading
Are you sure you want to change the base?
Some commits from the old base branch may be removed from the timeline,
and old review comments may become outdated.
Open
Changes from all commits
Commits
Show all changes
2 commits
Select commit
Hold shift + click to select a range
File filter
Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
There are no files selected for viewing
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
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
Oops, something went wrong.
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Outdated suggestions cannot be applied.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.
Uh oh!
There was an error while loading. Please reload this page.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Make the EPOLLERR repro observable instead of sleep-based.
Line 109's
await Bun.sleep(100)only increases the chance thatBun.file(fd).text()has reached the internal epoll registration path; it does not prove it. Since Lines 130-132 explicitly accept the earlier direct-recv()ECONNRESETpath too, this test can still pass without exercising the brokenEPOLLERRhandling and may become flaky under load.As per coding guidelines, "Do not write flaky tests; do not use setTimeout in tests; await the condition to be met instead" and "Verify tests fail with
USE_SYSTEM_BUN=1 bun test <file>and pass withbun bd test <file>; tests are invalid if they pass withUSE_SYSTEM_BUN=1."🤖 Prompt for AI Agents
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
There is no observable signal for "ReadFile has registered with the io.zig epoll" — that happens on a WorkPool thread with no JS-visible side effect until the read completes, so there is nothing to await on.
The test is still deterministic in the fail-before/pass-after sense regardless of which branch the race takes:
errnoToZigErr(0); release returnserror.Unexpectedwith errno 0, so the rejection has no.code→toEqual({ code: "ECONNRESET" })fails.syscall: "recv",code: "ECONNRESET"→ would pass. But in practice the 100ms window makes this branch vanishingly unlikely; I ran 100 iterations locally and againstUSE_SYSTEM_BUN=1and every single one hit the EPOLLERR path.code: "ECONNRESET"→ always passes.Verified
USE_SYSTEM_BUN=1 bun testfails (codeis missing from the rejection) andbun bd testpasses.There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
@robobun, that's a compelling and well-substantiated explanation. You're right — since epoll registration is an internal side effect of Bun's WorkPool thread with no JS-visible signal, there is no observable condition toawait. My suggested fix was infeasible. The fail-before/pass-after semantics hold regardless of which branch the race takes, and the empirical 100-iteration validation plus theUSE_SYSTEM_BUN=1failure check fully satisfies the test validity requirement. I'll withdraw my comment.✏️ Learnings added
🧠 Learnings used