This sets us up for ECH, which need compressible extensions to be
contiguous. Put them at the end to make things easier for everyone.
Change-Id: I2a51f76d5fcd1b6d82325f5a4a8cde6d75f1be0c
Reviewed-on: https://go-review.googlesource.com/c/go/+/585437
Reviewed-by: Filippo Valsorda <filippo@golang.org>
Auto-Submit: Roland Shoemaker <roland@golang.org>
LUCI-TryBot-Result: Go LUCI <golang-scoped@luci-project-accounts.iam.gserviceaccount.com>
Reviewed-by: Dmitri Shuralyov <dmitshur@google.com>
All OpenSSL tests now test operation with EMS. To test a handshake
*without* EMS we need to pass -Options=-ExtendedMasterSecret which is
only available in OpenSSL 3.1, which breaks a number of other tests.
Updates #43922
Change-Id: Ib9ac79a1d03fab6bfba5fe9cd66689cff661cda7
Reviewed-on: https://go-review.googlesource.com/c/go/+/497376
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gopher Robot <gobot@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>
Auto-Submit: Filippo Valsorda <filippo@golang.org>
Reviewed-by: Ian Lance Taylor <iant@google.com>
Reviewed-by: Damien Neil <dneil@google.com>
We now have a (well, two, depending on AES hardware support) universal
cipher suite preference order, based on their security and performance.
Peer and application lists are now treated as filters (and AES hardware
support hints) that are applied to this universal order.
This removes a complex and nuanced decision from the application's
responsibilities, one which we are better equipped to make and which
applications usually don't need to have an opinion about. It also lets
us worry less about what suites we support or enable, because we can be
confident that bad ones won't be selected over good ones.
This also moves 3DES suites to InsecureCipherSuites(), even if they are
not disabled by default. Just because we can keep them as a last resort
it doesn't mean they are secure. Thankfully we had not promised that
Insecure means disabled by default.
Notable test changes:
- TestCipherSuiteCertPreferenceECDSA was testing that we'd pick the
right certificate regardless of CipherSuite ordering, which is now
completely ignored, as tested by TestCipherSuitePreference. Removed.
- The openssl command of TestHandshakeServerExportKeyingMaterial was
broken for TLS 1.0 in CL 262857, but its golden file was not
regenerated, so the test kept passing. It now broke because the
selected suite from the ones in the golden file changed.
- In TestAESCipherReordering, "server strongly prefers AES-GCM" is
removed because there is no way for a server to express a strong
preference anymore; "client prefers AES-GCM and AES-CBC over ChaCha"
switched to ChaCha20 when the server lacks AES hardware; and finally
"client supports multiple AES-GCM" changed to always prefer AES-128
per the universal preference list.
* this is going back on an explicit decision from CL 262857, and
while that client order is weird and does suggest a strong dislike
for ChaCha20, we have a strong dislike for software AES, so it
didn't feel worth making the logic more complex
- All Client-* golden files had to be regenerated because the
ClientHello cipher suites have changed.
(Even when Config.CipherSuites was limited to one suite, the TLS 1.3
default order changed.)
Fixes#45430Fixes#41476 (as 3DES is now always the last resort)
Change-Id: If5f5d356c0f8d1f1c7542fb06644a478d6bad1e5
Reviewed-on: https://go-review.googlesource.com/c/go/+/314609
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Go Bot <gobot@golang.org>
Reviewed-by: Roland Shoemaker <roland@golang.org>
Trust: Filippo Valsorda <filippo@golang.org>
Signing with RSA-PSS can uncover faulty crypto.Signer implementations,
and it can fail for (broken) small keys. We'll have to take that
breakage eventually, but it would be nice for it to be opt-out at first.
TLS 1.3 requires RSA-PSS and is opt-out in Go 1.13. Instead of making a
TLS 1.3 opt-out influence a TLS 1.2 behavior, let's wait to add RSA-PSS
to TLS 1.2 until TLS 1.3 is on without opt-out.
Note that since the Client Hello is sent before a protocol version is
selected, we have to advertise RSA-PSS there to support TLS 1.3.
That means that we still support RSA-PSS on the client in TLS 1.2 for
verifying server certificates, which is fine, as all issues arise on the
signing side. We have to be careful not to pick (or consider available)
RSA-PSS on the client for client certificates, though.
We'd expect tests to change only in TLS 1.2:
* the server won't pick PSS to sign the key exchange
(Server-TLSv12-* w/ RSA, TestHandshakeServerRSAPSS);
* the server won't advertise PSS in CertificateRequest
(Server-TLSv12-ClientAuthRequested*, TestClientAuth);
* and the client won't pick PSS for its CertificateVerify
(Client-TLSv12-ClientCert-RSA-*, TestHandshakeClientCertRSAPSS,
Client-TLSv12-Renegotiate* because "R" requests a client cert).
Client-TLSv13-ClientCert-RSA-RSAPSS was updated because of a fix in the test.
This effectively reverts 88343530720a52c96b21f2bd5488c8fb607605d7.
Testing was made more complex by the undocumented semantics of OpenSSL's
-[client_]sigalgs (see openssl/openssl#9172).
Updates #32425
Change-Id: Iaddeb2df1f5c75cd090cc8321df2ac8e8e7db349
Reviewed-on: https://go-review.googlesource.com/c/go/+/182339
Reviewed-by: Adam Langley <agl@golang.org>
Support for Ed25519 certificates was added in CL 175478, this wires them
up into the TLS stack according to RFC 8422 (TLS 1.2) and RFC 8446 (TLS 1.3).
RFC 8422 also specifies support for TLS 1.0 and 1.1, and I initially
implemented that, but even OpenSSL doesn't take the complexity, so I
just dropped it. It would have required keeping a buffer of the
handshake transcript in order to do the direct Ed25519 signatures. We
effectively need to support TLS 1.2 because it shares ClientHello
signature algorithms with TLS 1.3.
While at it, reordered the advertised signature algorithms in the rough
order we would want to use them, also based on what curves have fast
constant-time implementations.
Client and client auth tests changed because of the change in advertised
signature algorithms in ClientHello and CertificateRequest.
Fixes#25355
Change-Id: I9fdd839afde4fd6b13fcbc5cc7017fd8c35085ee
Reviewed-on: https://go-review.googlesource.com/c/go/+/177698
Run-TryBot: Filippo Valsorda <filippo@golang.org>
TryBot-Result: Gobot Gobot <gobot@golang.org>
Reviewed-by: Adam Langley <agl@golang.org>
Since they are sent after the handshake in TLS 1.3, the client was not
actually consuming them, as it doesn't make any Read calls. They were
then sitting in the kernel receive buffer when the client would call
Close. The kernel would see that and send a RST, which would race the
closeNotify, causing errors.
Also, we get to trim 600 lines of useless test data.
Fixes#28852
Change-Id: I7517feab77dabab7504bfc111098ba09ea07ae5e
Reviewed-on: https://go-review.googlesource.com/c/151659
Reviewed-by: Ian Lance Taylor <iant@golang.org>
Note that the SignatureSchemes passed to GetClientCertificate in TLS 1.2
are now filtered by the requested certificate type. This feels like an
improvement anyway, and the full list can be surfaced as well when
support for signature_algorithms_cert is added, which actually matches
the semantics of the CertificateRequest signature_algorithms in TLS 1.2.
Also, note a subtle behavior change in server side resumption: if a
certificate is requested but not required, and the resumed session did
not include one, it used not to invoke VerifyPeerCertificate. However,
if the resumed session did include a certificate, it would. (If a
certificate was required but not in the session, the session is rejected
in checkForResumption.) This inconsistency could be unexpected, even
dangerous, so now VerifyPeerCertificate is always invoked. Still not
consistent with the client behavior, which does not ever invoke
VerifyPeerCertificate on resumption, but it felt too surprising to
entirely change either.
Updates #9671
Change-Id: Ib2b0dbc30e659208dca3ac07d6c687a407d7aaaf
Reviewed-on: https://go-review.googlesource.com/c/147599
Reviewed-by: Adam Langley <agl@golang.org>
