Make HTTP/2 response body close idempotent - #26
Merged
bogdanfinn merged 1 commit intoSep 4, 2026
Merged
Conversation
Treat a response body already closed by the caller as a no-op so repeated Close calls do not emit additional RST_STREAM frames.
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
transportResponseBody.Closecurrently performs its reset logic again when called more than once. For an unfinished response, two calls emit twoRST_STREAM(CANCEL)frames. After a response body has already been read to EOF, the firstCloserecords the closed-body error and a secondCloseincorrectly emits a reset.Treat a body already closed by the caller as a successful no-op. This preserves the first close's existing behavior while preventing repeated reset and flow-control side effects.
The regression test covers both cases:
RST_STREAM(CANCEL)across twoClosecallsRST_STREAMacross twoClosecallsTest results
Passed on Go 1.24.1 (toolchain selected by the current upstream
go.mod):The regression test fails against unpatched upstream as expected:
The full package command was also run:
It completed in 94.007s with one existing failure in
TestTransportRejectsConnHeaders(Content-Lengthunexpectedly present). The same failure reproduces on a clean checkout of the upstream base commit, so it is unrelated to this change.