Re: Fine-tune TLS 1.3 cipher suites and curves lists

Поиск
Список
Период
Сортировка
От Daniel Gustafsson
Тема Re: Fine-tune TLS 1.3 cipher suites and curves lists
Дата
Msg-id 6E3FD976-0D7F-4F64-ACDD-3E0FBF3834AD@yesql.se
обсуждение исходный текст
Ответ на Fine-tune TLS 1.3 cipher suites and curves lists  (Seraphime Kirkovski <skirkovski@vmware.com>)
Список pgsql-hackers
> On 17 Jul 2023, at 15:16, Seraphime Kirkovski <skirkovski@vmware.com> wrote:

> I see in the source code that only TLS 1.2 and bellow cipher lists can be configured:
>
> https://github.com/postgres/postgres/blob/master/src/backend/libpq/be-secure-openssl.c#L281
>
> and Postgres relies on the OpenSSL defaults for TLS 1.3 ciphersuites.
>
> My first question is whether there is a reason not to support setting TLS 1.3 cipher suites through configuration ?
Maybethere are Postgres builds with BoringSSL ? (Just speculating ?) 

I think the main raison is that noone has done it, and noone has requested it.
I have no way if knowing for certain, but I doubt too many postgres users
change this setting.

> Another thing I was curious about is why does postgres opts to support setting only a single elliptic group
(https://github.com/postgres/postgres/blob/master/src/backend/libpq/be-secure-openssl.c#L1303)instead of calling out to
anSSL function like SSL_CTX_set1_curves_list ? 
>
> Would the community be interested in seeing patches for setting TLS 1.3 ciphersuites and expanding the configuration
optionfor EC settings to support lists instead of single values ?  

I would be interested in seeing them, and would offer to review them.

The main challenge is IMO to properly document these settings such that
postgres users know what they are, and when they should think about changing
them.  Postgres also supports very old OpenSSL versions, so any change and
setting must in some way make sense for those installations (which may be a
no-op, a warning at startup for non-applicable settings, or something else).

--
Daniel Gustafsson




В списке pgsql-hackers по дате отправления:

Предыдущее
От: Jeff Davis
Дата:
Сообщение: Re: MERGE ... RETURNING
Следующее
От: Stephen Frost
Дата:
Сообщение: Re: SLRUs in the main buffer pool - Page Header definitions