[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [users@httpd] Difficulties with testing TLS 1.3

[CC-ing Dennis since he does not seem to be subscribed to dev@]

On Fri, Jun 15, 2018 at 9:38 AM, Yann Ylavic <ylavic.dev@xxxxxxxxx> wrote:
> On Fri, Jun 15, 2018 at 8:59 AM, Yann Ylavic <ylavic.dev@xxxxxxxxx> wrote:
>>> On Fri, Jun 15, 2018 at 3:06 AM, Dennis Clarke <dclarke@xxxxxxxxxxxxx> wrote:
>>>> [Fri Jun 15 00:59:59.788742 2018] [ssl:error] [pid 2250:tid 27] [client
>>>>] AH02042: rejecting client initiated renegotiation
>> This suggests that with TLS 1.3, unlike with previous protocols, the
>> ssl_callback_Info callback is always called by openssl (some specific
>> extension only?). httpd is not prepared to that and thinks it's a
>> client renegotiation.
>> I didn't look at TLS 1.3 yet, but something along this looks seems plausible.
> For instance, multiple session tickets might call the callback more than once:
> https://github.com/openssl/openssl/blob/master/ssl/statem/statem_srvr.c#L3785
> Since renegotiations are forbidden by TLS 1.3 in the first place
> (enforced by openssl probably), I think that the check for
> renegotiations should be disabled in our ssl_callback_Info (for TLS
> 1.3).