SSLWrap was needlessly defined as a template class, splitting the
TLS implementation over multiple locations. The original idea, I
surmise, was to make it possible to reuse SSLWrap for some other
purpose that never manifest. This squashes them down into a single
TLSWrap class and moves tls_wrap.h/cc into src/crypto.
Signed-off-by: James M Snell <jasnell@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/35552
Reviewed-By: Alba Mendez <me@alba.sh>
Two things in one on this commit:
(a) For the QUIC implementation, we need to separate out various bits
from node_crypto.cc to allow them to be reused. That's where this
commit starts.
(b) Quite a bit of the node_crypto.cc code was just messy in terms of
it's organization and lack of error handling and use of Local vs.
MaybeLocal. This cleans that up a bit and hopefully makes certain
parts a bit more manageable also.
Signed-off-by: James M Snell <jasnell@gmail.com>
PR-URL: https://github.com/nodejs/node/pull/32016
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Sam Roberts <vieuxtech@gmail.com>
When piping data into an SHA3 hash, EVP_DigestFinal_ex is called in
hash._flush, bypassing safeguards in the JavaScript layer. Calling
hash.digest causes EVP_DigestFinal_ex to be called again, resulting
in a segmentation fault in the SHA3 implementation of OpenSSL.
A relatively easy solution is to cache the result of calling
EVP_DigestFinal_ex until the Hash object is garbage collected.
PR-URL: https://github.com/nodejs/node/pull/28251
Fixes: https://github.com/nodejs/node/issues/28245
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
`v8::Global` is essentially a nicer variant of `node::Persistent` that,
in addition to reset-on-destroy, also implements move semantics.
This commit makes the necessary replacements, removes
`node::Persistent` and (now-)unnecessary inclusions of the
`node_persistent.h` header, and makes some of the functions that
take Persistents as arguments more generic so that they work with all
`v8::PersistentBase` flavours.
PR-URL: https://github.com/nodejs/node/pull/27287
Reviewed-By: Gus Caplan <me@gus.host>
Reviewed-By: Eugene Ostroukhov <eostroukhov@google.com>
Reviewed-By: Joyee Cheung <joyeec9h3@gmail.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
This introduces TLS1.3 support and makes it the default max protocol,
but also supports CLI/NODE_OPTIONS switches to disable it if necessary.
TLS1.3 is a major update to the TLS protocol, with many security
enhancements. It should be preferred over TLS1.2 whenever possible.
TLS1.3 is different enough that even though the OpenSSL APIs are
technically API/ABI compatible, that when TLS1.3 is negotiated, the
timing of protocol records and of callbacks broke assumptions hard-coded
into the 'tls' module.
This change introduces no API incompatibilities when TLS1.2 is
negotiated. It is the intention that it be backported to current and LTS
release lines with the default maximum TLS protocol reset to 'TLSv1.2'.
This will allow users of those lines to explicitly enable TLS1.3 if they
want.
API incompatibilities between TLS1.2 and TLS1.3 are:
- Renegotiation is not supported by TLS1.3 protocol, attempts to call
`.renegotiate()` will always fail.
- Compiling against a system OpenSSL lower than 1.1.1 is no longer
supported (OpenSSL-1.1.0 used to be supported with configure flags).
- Variations of `conn.write('data'); conn.destroy()` have undefined
behaviour according to the streams API. They may or may not send the
'data', and may or may not cause a ERR_STREAM_DESTROYED error to be
emitted. This has always been true, but conditions under which the write
suceeds is slightly but observably different when TLS1.3 is negotiated
vs when TLS1.2 or below is negotiated.
- If TLS1.3 is negotiated, and a server calls `conn.end()` in its
'secureConnection' listener without any data being written, the client
will not receive session tickets (no 'session' events will be emitted,
and `conn.getSession()` will never return a resumable session).
- The return value of `conn.getSession()` API may not return a resumable
session if called right after the handshake. The effect will be that
clients using the legacy `getSession()` API will resume sessions if
TLS1.2 is negotiated, but will do full handshakes if TLS1.3 is
negotiated. See https://github.com/nodejs/node/pull/25831 for more
information.
PR-URL: https://github.com/nodejs/node/pull/26209
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: James M Snell <jasnell@gmail.com>
Reviewed-By: Rod Vagg <rod@vagg.org>
OpenSSL 1.0.0 returned incorrect version information. OpenSSL 1.1.0
fixed this, but returning the correct information broke our tests, so
was considered semver-major. Because of this, the version was hard-coded
to the OpenSSL 1.0.0 (incorrect) string in 5fe81c8aff.
This is ancient history, start returning the correct cipher version.
PR-URL: https://github.com/nodejs/node/pull/26625
Reviewed-By: Daniel Bevenius <daniel.bevenius@gmail.com>
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
Reviewed-By: Ruben Bridgewater <ruben@bridgewater.de>
Reviewed-By: Richard Lau <riclau@uk.ibm.com>
Reviewed-By: Colin Ihrig <cjihrig@gmail.com>
Where appropriate, use a helper that wraps around
`ArrayBufferView::Buffer()` or `ArrayBufferView::CopyContents()`
rather than `Buffer::Data()`, as that may help to avoid materializing
the underlying `ArrayBuffer` when reading small typed arrays from C++.
This allows keeping the performance benefits of the faster creation of
heap-allocated small typed arrays in many cases.
PR-URL: https://github.com/nodejs/node/pull/26301
Reviewed-By: Matteo Collina <matteo.collina@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>
Renamed some internal C++ methods and properties for consistency, and
commented SSL I/O.
- Rename waiting_new_session_ after is_waiting_new_session(), instead of
using reverse naming (new_session_wait_), and change "waiting" to
"awaiting".
- Make TLSWrap::ClearIn() return void, the value is never used.
- Fix a getTicketKeys() cut-n-paste error. Since it doesn't use the
arguments, remove them from the js wrapper.
- Remove call of setTicketKeys(getTicketKeys()), its a no-op.
PR-URL: https://github.com/nodejs/node/pull/25713
Reviewed-By: Anna Henningsen <anna@addaleax.net>
Reviewed-By: Michael Dawson <michael_dawson@ca.ibm.com>
Reviewed-By: Ben Noordhuis <info@bnoordhuis.nl>
This commit makes multiple important changes:
1. A new key object API is introduced. The KeyObject class itself is
not exposed to users, instead, several new APIs can be used to
construct key objects: createSecretKey, createPrivateKey and
createPublicKey. The new API also allows to convert between
different key formats, and even though the API itself is not
compatible to the WebCrypto standard in any way, it makes
interoperability much simpler.
2. Key objects can be used instead of the raw key material in all
relevant crypto APIs.
3. The handling of asymmetric keys has been unified and greatly
improved. Node.js now fully supports both PEM-encoded and
DER-encoded public and private keys.
4. Conversions between buffers and strings have been moved to native
code for sensitive data such as symmetric keys due to security
considerations such as zeroing temporary buffers.
5. For compatibility with older versions of the crypto API, this
change allows to specify Buffers and strings as the "passphrase"
option when reading or writing an encoded key. Note that this
can result in unexpected behavior if the password contains a
null byte.
PR-URL: https://github.com/nodejs/node/pull/24234
Reviewed-By: Refael Ackermann <refack@gmail.com>
Reviewed-By: James M Snell <jasnell@gmail.com>