RFR: 8331682: Slow networks/Impatient clients can potentially send unencrypted TLSv1.3 alerts that won't parse on the server [v3]

Artur Barashev duke at openjdk.org
Fri Sep 20 16:22:35 UTC 2024


On Fri, 20 Sep 2024 15:33:05 GMT, Hai-May Chao <hchao at openjdk.org> wrote:

> The change involves handling an unusual situation, and to help with debugging, I’d like to suggest to log a message when server continues processing an unencrypted alert.

Will add the logging, thanks!

> src/java.base/share/classes/sun/security/ssl/SSLTransport.java line 126:
> 
>> 124:             throw context.fatal(Alert.BAD_RECORD_MAC, bte);
>> 125:         } catch (BadPaddingException bpe) {
>> 126:             // Check for unexpected plaintext alert message during TLSv1.3 handshake, @bug 8331682
> 
> Suggest to modify the comments here where re-constructing the plaintext instead of @bug 8331682.

Will do, thanks!

> test/jdk/javax/net/ssl/TLSv13/SSLEngineNoServerHelloClientShutdown.java line 239:
> 
>> 237: 
>> 238:         /* TODO: Final client unwrap fails because server doesn't send an alert to terminate
>> 239:            the handshake after receiving close_notify alert from the client. Investigate why.
> 
> Is this TODO item done?

Not yet. I'm currently learning the internals of TLS code and going through TLS RFC in order to understand if it's an expected behavior or not. What do you think?

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

PR Comment: https://git.openjdk.org/jdk/pull/21043#issuecomment-2364083286
PR Review Comment: https://git.openjdk.org/jdk/pull/21043#discussion_r1768882473
PR Review Comment: https://git.openjdk.org/jdk/pull/21043#discussion_r1768881285


More information about the security-dev mailing list