RFR: 8335181: Incorrect handling of HTTP/2 GOAWAY frames in HttpClient [v2]

[Jaikiran Pai]

On Tue, 13 Aug 2024 15:08:24 GMT, Jaikiran Pai <jpai at openjdk.org> wrote:

>> This is a fresh version of the PR that I had opened here https://github.com/openjdk/jdk/pull/20442. A `git merge` command ended up causing unexpected issues with that old PR.
>> 
>> The commits in this PR are the ones which were already reviewed and a new one which was done to implement Daniel's suggestion in the review. test repeat 50 of java/net/httpclient had passed without any issues for that old PR. But before integrating this PR, I will rerun all relevant tests again to be sure nothing unexpected shows up.
>> 
>> Sorry that this has to be reviewed afresh.
>
> Jaikiran Pai has updated the pull request incrementally with four additional commits since the last revision:
> 
>  - decrementStreamsCount() in closeAsUnprocessed()
>  - add code comment to explain why closeAsUnprocessed() isn't needed
>  - Revert "Daniel's suggestion - call closeAsUnprocessed() from handleReset"
>    
>    This reverts commit 4e56a7c668f5800375b4dea4ab2e2940ca1f888e.
>  - Revert "also decrement streams to allow the HttpClient to close"
>    
>    This reverts commit 5040bf732bd9cb5e2dd7feb532a467991a64cb60.

Thank you Daniel for the review. tier testing and test repeat testing came back fine without any failures.

-------------

PR Comment: https://git.openjdk.org/jdk/pull/20567#issuecomment-2287897818