The OpenSSL developers decided, during the OpenSSL 1.1.1 development
phase, to use a different API and different set of lists for TLSv1.3
ciphersuites, than for every TLS version preceeding it.
This is stupid, but we have to work with it.
This commit also improves configuration fault resilience. The reason
is that if you don't pass any valid old-style ciphersuites, OpenSSL
will not negotiate an older protocol at all. However, when they
implemented the new API, they decided that lack of any valid
ciphersuites should result in using the defaults. This means that if
you pass a completely invalid ciphersuite list (like "foo"), OR if
you pass a TLSv1.2-only ciphersuite list, TLSv1.3 continues to work.
This is not mirrored; passing a TLSv1.3-only ciphersuite list will
break TLSv1.2 and below.
Therefore we work around this lack of mirroring by falling back to
the default list for each protocol. This means that if
ssl_cipher_list is complete garbage, the default will be used, and
TLS setup will succeed for both protocols. This is logged, so that
administrators can fix their configuration.
I prefer this approach over explicitly disabling the protocols if
their respective ciphersuite lists are invalid, because it will
result in unusable TLSv1.3 if people run newer solanum with their
older charybdis/solanum configuration files that contain custom
ssl_cipher_list definitions. Hindering TLSv1.3 adoption is not an
option, in my opinion.
The downside of this is that it is no longer possible to disable a
protocol family by not including any of its ciphersuites. This could
be remedied by an ssl_protocol_list configuration directive if it is
decided that this functionality is ultimately necessary.
This work is not required for either of the other TLS backends,
because neither of those libraries yet support TLSv1.3, and in the
event that they eventually do, I expect them to allow configuration
of newer ciphersuites with the existing APIs. This can be revisited
if it turns out not to be the case.
Signed-off-by: Aaron Jones <me@aaronmdjones.net>
Tested-by: Aaron Jones <me@aaronmdjones.net>
This only supports two addresses as the intended use is 1 IPv4 and 1 IPv6
address on a single-homed host, and the only supported configuration of
outgoing connections to other servers is to bind a single IPv4 or IPv6
address.
- Add (void) casts for unused function parameters
- Rearrange member in `struct rb_mbedtls_cfg_context' for data alignment
- Document a `clang-4.0 -Weverything' (-Wcast-qual) diagnostic
- Avoid pointless conversions between positive/negative error codes
- Use capital hexadecimals in error codes and properly cast to
(unsigned int) for %x/%X
If an IPv4 connection is dropped by the pre-callback, and there is a
pending IPv6 connection on the same listening socket then the retried
accept() will be unable to populate `st` because `addrlen` will be too
small. Also initialise `st` each time to avoid a clang static analysis
warning.
[ircd/match.c:316]: (error) Shifting a negative value is undefined behaviour
[librb/src/patricia.c:55]: (error) Shifting a negative value is undefined behaviour
[modules/m_alias.c:64]: (portability) '(void*)message' is of type 'void *'. When using void pointers in calculations, the behaviour is undefined.
[modules/m_time.c:111]: (warning) %u in format string (no. 9) requires 'unsigned int' but the argument type is 'signed int'.
[modules/m_time.c:111]: (warning) %u in format string (no. 10) requires 'unsigned int' but the argument type is 'signed int'.
[librb/src/dictionary.c:819]: (warning) %d in format string (no. 3) requires 'int' but the argument type is 'unsigned int'.
[librb/src/radixtree.c:1080]: (warning) %d in format string (no. 3) requires 'int' but the argument type is 'unsigned int'.
[ircd/s_user.c:351] -> [ircd/s_user.c:357]: (warning) Either the condition '0!=source_p' is redundant or there is possible null pointer dereference: source_p.
[extensions/ip_cloaking_3.0.c:109]: (warning, inconclusive) The buffer 'buf' may not be null-terminated after the call to strncpy().
[ircd/chmode.c:256]: (style) Clarify calculation precedence for '&' and '?'.
[modules/m_help.c:100]: (style) Clarify calculation precedence for '&' and '?'.
[modules/m_knock.c:169]: (style) Clarify calculation precedence for '&' and '?'.
[modules/m_stats.c:628]: (style) Clarify calculation precedence for '&' and '?'.
[modules/m_stats.c:727]: (style) Clarify calculation precedence for '&' and '?'.
[librb/src/radixtree.c:601]: (style) Clarify calculation precedence for '&' and '?'.
[librb/src/radixtree.c:704]: (style) Clarify calculation precedence for '&' and '?'.
[librb/src/radixtree.c:739]: (style) Clarify calculation precedence for '&' and '?'.
[librb/src/radixtree.c:763]: (style) Clarify calculation precedence for '&' and '?'.
[librb/src/radixtree.c:768]: (style) Clarify calculation precedence for '&' and '?'.
[librb/src/radixtree.c:774]: (style) Clarify calculation precedence for '&' and '?'.
[librb/src/radixtree.c:781]: (style) Clarify calculation precedence for '&' and '?'.
[librb/src/radixtree.c:786]: (style) Clarify calculation precedence for '&' and '?'.
[librb/src/radixtree.c:791]: (style) Clarify calculation precedence for '&' and '?'.
[librb/src/radixtree.c:804]: (style) Clarify calculation precedence for '&' and '?'.
[ircd/wsproc.c:372]: (style) Unused variable: len
[modules/core/m_modules.c:382]: (style) Unused variable: i
[modules/m_stats.c:741]: (style) Unused variable: amsg
[ircd/authproc.c:390]: (style) Unused variable: iter
[ircd/authproc.c:391]: (style) Unused variable: client_p
The changes made in this file are detailed in release/3.5 commits
4d89c83c32 through ecfdcb08e8 inclusive.
Some slight alterations are necessary; librb instead of libratbox,
snprintf() instead of rb_snprintf(), etc. Also, release/4 has
offline certificate digesting functionality. For more insight,
simply diff the gnutls.c between release/3.5 and release/4.
