Conversation
OpenSSL writes a complete handshake flight to the network BIO in a single operation. The BIO pair used by the OpenSSL based SSLEngine implementations is created with the default 17408 byte buffer so, when the flight is larger than that - which in practice requires a large certificate chain - OpenSSL retains the remainder internally. wrap() copies whatever the BIO holds into the destination buffer and returns NEED_UNWRAP without giving OpenSSL any opportunity to write the rest. The remainder is currently only written as a side effect of the priming SSL_read() performed by a later call to unwrap(). That hides the problem whenever the client has already sent something - a TLSv1.3 client normally sends a middlebox compatibility change cipher spec record - but a client that is waiting for the server to complete its flight sends nothing, so unwrap() is never reached and the connection stalls until it times out. With a TLSv1.2 client the failure is deterministic. After the network BIO has been drained by wrap(), and while a handshake is still in progress, drive OpenSSL again so it can write out whatever did not previously fit. When there is nothing left to write the call is a no-op. A dedicated method is used rather than handshake() because handshake() re-snapshots the handshake counter that is used to detect completion, which must not happen in the middle of a flight. Neither implementation binds SSL_get_error() so it is not currently possible to test for SSL_ERROR_WANT_WRITE and make the call conditional. Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
lhellebr
force-pushed
the
bz-flight-stall
branch
from
September 18, 2026 13:50
9f075ad to
8039228
Compare
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
1 participant
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.
Fixes bug 70236.
OpenSSL writes a complete handshake flight to the network BIO in a single operation. The BIO pair used by the OpenSSL based SSLEngine implementations is created with the default 17408 byte buffer so, when the flight is larger than that - which in practice requires a large certificate chain - OpenSSL retains the remainder internally. wrap() copies whatever the BIO holds into the destination buffer and returns NEED_UNWRAP without giving OpenSSL any opportunity to write the rest.
The remainder is currently only written as a side effect of the priming SSL_read() performed by a later call to unwrap(). That hides the problem whenever the client has already sent something - a TLSv1.3 client normally sends a middlebox compatibility change cipher spec record - but a client that is waiting for the server to complete its flight sends nothing, so unwrap() is never reached and the connection stalls until it times out. With a TLSv1.2 client the failure is deterministic.
After the network BIO has been drained by wrap(), and while a handshake is still in progress, drive OpenSSL again so it can write out whatever did not previously fit. When there is nothing left to write the call is a no-op. A dedicated method is used rather than handshake() because handshake() re-snapshots the handshake counter that is used to detect completion, which must not happen in the middle of a flight. Neither implementation binds SSL_get_error() so it is not currently possible to test for SSL_ERROR_WANT_WRITE and make the call conditional.