On 4/13/21 3:00 PM, Stefan Eissing wrote:
> The PR 179 <https://github.com/apache/httpd/pull/179> has been updated with 
> the additions of our OCSP hook support in the core server. In case you did 
> not follow it, a short summary of the changes:
> 
> 1. httpd core offers functions/hooks so that SSL related things 
>    can be queried without optional functions. That means all the 
>    lookups in modules of the "ssl_is_https(conn_rec *c)" have 
>    been converted to direct invocations of "ap_ssl_conn_is_ssl(c)".
> 2. The provisioning of SSL certificates for servers, as done by mod_md,
>    has also now a central hub in the server. This means other modules
>    besides mod_ssl/mod_md can use/offer certificate services.
> 3. The OCSP response data that is used in "SSL Stapling" has also now
>    a central hub with functions/hooks. That means other SSL modules can
>    use mod_md services and another OCSP stapling module may be written
>    and installed without changes to SSL modules.
> 
> All these new server functions/hooks are agnostic of the actual *SSL 
> implementation. One might mix modules using OpenSSL with ones using libressl 
> or exotics like rustls.
> 
> 4. mod_ssl has been extended to use these new functions/hooks while 
> maintaining its own OPTIONAL functions and support. This should give full 
> backward compatibility for 3rd party modules for interop with mod_ssl.
> 
> 5. mod_md has been extended to use/register at the new core hooks. The PR 
> also includes a range of other, unrelated enhancements to mod_md, like 
> multiple certificates and EC keys. It would have been possible to separate 
> this out, but it would have required more testing combinations and the number 
> of branches/repros to juggle are already high enough.
> 
> As will all github PRs, you can get the diff simply by appending ".diff", so 
> <https://github.com/apache/httpd/pull/179.diff> is the whole thing. For 
> easier review, I extract the parts related only to the core server and the 
> ones only related to mod_ssl and attach them here.
> 
> I would very much appreciate if some of you find the time to point to my 
> mistakes.
> 

As it was easier for me I directly commented on the PR.

Regards

RĂ¼diger

Reply via email to