Richard Levitte [Wed, 3 Feb 2021 12:55:30 +0000 (13:55 +0100)]
EVP: Adapt the other EVP_PKEY_set_xxx_param() functions
They were calling evp_keymgmt_set_params() directly. Those calls are
changed to go through EVP_PKEY_set_params().
We take the opportunity to constify these functions. They have to
unconstify internally for the compiler to stop complaining when
placing those pointers in an OSSL_PARAM element, but that's still
better than forcing the callers to do that cast.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14056)
Richard Levitte [Wed, 3 Feb 2021 12:50:23 +0000 (13:50 +0100)]
EVP: Make EVP_PKEY_set_params() increment the dirty count
When the internal key is changed, we must count it as muted, so that
next time the affected key is considered for an operation, it gets
re-exported to the signing provider. In other words, this will clear
the EVP_PKEY export cache when the next export attempt occurs.
This also updates evp_keymgmt_util_export_to_provider() to actually
look at the dirty count for provider native origin keys, and act
appropriately.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14056)
Matt Caswell [Mon, 1 Feb 2021 17:31:05 +0000 (17:31 +0000)]
Remove a DSA related TODO
There are no instances of the macros that this comment is referring to
being used anywhere within current master. All of the macros were
deprecated by commit f41ac0e. Therefore this TODO should just be removed.
Fixes #13020
Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14038)
Matt Caswell [Mon, 1 Feb 2021 15:45:44 +0000 (15:45 +0000)]
Remove some TODO(OpenSSL1.2) references
We had a couple of stray references to OpenSSL1.2 in libssl. We just
reword the comments to remove those references without changing any
behaviour.
The first one in t1_lib.c is a technical non-compliance in the TLSv1.3
spec where, under some circumstances, we offer DSA sigalgs even in a
ClientHello that eventually negotiates TLSv1.3. We explicitly chose to
accept this behaviour in 1.1.1 and we're not planning to change it for
3.0.
The second one in s3_lib.c is regarnding the behaviour of
SSL_set_tlsext_host_name(). Technically you shouldn't be able to call
this from a server - but we allow it and just ignore it rather than
raising an error. The TODO suggest we consider raising an error instead.
However, with 3.0 we are trying to minimise breaking changes so I suggest
not making this change now.
Fixes #13161
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/14037)
Pauli [Wed, 3 Feb 2021 07:47:38 +0000 (17:47 +1000)]
Fix a use after free issue when a provider context is being used and isn't cached
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14053)
Richard Levitte [Wed, 27 Jan 2021 13:55:28 +0000 (14:55 +0100)]
EC: Reverse the default asn1_flag in a new EC_GROUP
The default was OPENSSL_EC_NAMED_CURVE, but that's not true until a
curve name has been set, so we change the initial value to
OPENSSL_EC_EXPLICIT_CURVE and let EC_GROUP_set_curve_name() change it
to OPENSSL_EC_NAMED_CURVE.
Submitted by Matt Caswell
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/13973)
Richard Levitte [Wed, 27 Jan 2021 10:07:38 +0000 (11:07 +0100)]
EVP: Fix evp_pkey_ctx_store_cached_data() to handle provider backed EVP_PKEY_CTX
It assumed there would always be a non-NULL ctx->pmeth, leading to a
crash when that isn't the case. Since it needs to check 'keytype'
when that one isn't -1, we also add a corresponding check for the
provider backed EVP_PKEY_CTX case.
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/13973)
Richard Levitte [Tue, 26 Jan 2021 16:01:15 +0000 (17:01 +0100)]
EVP: Don't find standard EVP_PKEY_METHODs automatically
EVP_PKEY_meth_find() got called automatically any time a new
EVP_PKEY_CTX allocator was called with some sort of key type data.
Since we have now moved all our standard algorithms to our providers,
this is no longer necessary.
We do retain looking up EVP_PKEY_METHODs that are added by the calling
application.
Fixes #11424
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/13973)
Richard Levitte [Tue, 2 Feb 2021 12:42:55 +0000 (13:42 +0100)]
CORE & PROV: clean away OSSL_FUNC_mac_size()
There was a remaining function signature declaration, but no
OSSL_DISPATCH number for it nor any way it's ever used. It did exist
once, but was replaced with an OSSL_PARAM item to retrieve.
Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14048)
Rich Salz [Thu, 28 Jan 2021 20:47:53 +0000 (15:47 -0500)]
Deprecate EVP_MD_CTX_{set_}update_fn()
They are still used internally in legacy code.
Also fixed up some minor things in EVP_DigestInit.pod
Fixes: #14003 Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Paul Dale <pauli@openssl.org> Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14008)
Tomas Mraz [Mon, 1 Feb 2021 21:07:17 +0000 (22:07 +0100)]
Add diacritics to my name in CHANGES.md
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14044)
Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14029)
Matt Caswell [Wed, 27 Jan 2021 17:18:27 +0000 (17:18 +0000)]
Ensure the EVP_PKEY operation_cache is appropriately locked
The EVP_PKEY operation_cache caches references to provider side key
objects that have previously been exported for this EVP_PKEY, and their
associated key managers. The cache may be updated from time to time as the
EVP_PKEY is exported to more providers. Since an EVP_PKEY may be shared by
multiple threads simultaneously we must be careful to ensure the cache
updates are locked.
Fixes #13818
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13987)
Matt Caswell [Tue, 26 Jan 2021 15:23:19 +0000 (15:23 +0000)]
Avoid races by caching exported ciphers in the init function
TSAN was reporting a race of the exported ciphers cache that we create in
the default and fips providers. This was because we cached it in the query
function rather than the init function, so this would cause a race if multiple
threads queried at the same time. In practice it probably wouldn't make much
difference since different threads should come up with the same answer.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13987)
Matt Caswell [Tue, 26 Jan 2021 13:30:06 +0000 (13:30 +0000)]
Add a multi-thread test for shared EVP_PKEYs
EVP_PKEYs may be shared across mutliple threads. For example this is
common for users of libssl who provide a single EVP_PKEY private key for
an SSL_CTX, which is then shared between multiple threads for each SSL
object.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13987)
Rich Salz [Thu, 28 Jan 2021 15:17:13 +0000 (10:17 -0500)]
Deprecate X509_certificate_type
Fixes: #13997 Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com> Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14002)
Job Snijders [Sun, 24 Jan 2021 14:00:02 +0000 (14:00 +0000)]
Add some PKIX-RPKI objects
References:
RFC6482 - A Profile for Route Origin Authorizations (ROAs)
RFC6484 - Certificate Policy (CP) for the RPKI
RFC6493 - The RPKI Ghostbusters Record
RFC8182 - The RPKI Repository Delta Protocol (RRDP)
RFC8360 - RPKI Validation Reconsidered
draft-ietf-sidrops-rpki-rta - A profile for RTAs
CLA: trivial
Reviewed-by: Paul Dale <pauli@openssl.org> Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/13948)
Richard Levitte [Thu, 28 Jan 2021 07:22:09 +0000 (08:22 +0100)]
PROV: Add SM2 encoders and decoders, as well as support functionality
The EC KEYMGMT implementation handled SM2 as well, except what's
needed to support decoding: loading functions for both EC and SM2 that
checks for the presence or absence of the SM2 curve the same way as
the EC / SM2 import functions.
Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14028)
Daniel Bevenius [Fri, 9 Oct 2020 04:07:43 +0000 (06:07 +0200)]
EVP: fix keygen for EVP_PKEY_RSA_PSS
This commit attempts to fix the an issue when generating a key of type
EVP_PKEY_RSA_PSS. Currently, EVP_PKEY_CTX_set_rsa_keygen_bits will
return -1 if the key id is not of type EVP_PKEY_RSA. This commit adds
EVP_PKEY_RSA_PSS to also be accepted.
The macro EVP_PKEY_CTX_set_rsa_pss_keygen_md si converted into a
function and it is now called in legacy_ctrl_to_param.
Fixes #12384
Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13099)
Juergen Christ [Tue, 26 Jan 2021 16:06:54 +0000 (17:06 +0100)]
Remove superfluous EVP_KDF_CTRL_ defines.
These defines were never used and not needed.
Signed-off-by: Juergen Christ <jchrist@linux.ibm.com> Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Patrick Steuer <patrick.steuer@de.ibm.com>
(Merged from https://github.com/openssl/openssl/pull/13781)
Juergen Christ [Mon, 14 Dec 2020 16:36:22 +0000 (17:36 +0100)]
Fix parameter types in sshkdf
Handling of parameter OSSL_KDF_PARAM_SSHKDF_TYPE mixed integer and string
parameters. This caused endianness problems on big-endian machines. As a
result, it is not possible to pass FIPS tests since the parameter was stored
with an integer value but read via a cast to char pointer. While this works
on little endian machines, big endian s390 read the most significant bits
instead of the least significant (as done by, e.g., x86). Change the
parameter to char array and fix the usages.
Signed-off-by: Juergen Christ <jchrist@linux.ibm.com> Reviewed-by: Richard Levitte <levitte@openssl.org> Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Patrick Steuer <patrick.steuer@de.ibm.com>
(Merged from https://github.com/openssl/openssl/pull/13781)
Juergen Christ [Wed, 27 Jan 2021 10:04:52 +0000 (11:04 +0100)]
Fix cipher reinit on s390x if no key is specified
If key==null on EVP_CipherInit_ex, the init functions for the hardware
implementation is not called. The s390x implementation of OFB and CFB mode
used the init function to copy the IV into the hardware causing test failures
on cipher reinit. Fix this by moving the copy operation into the cipher
operation.
Signed-off-by: Juergen Christ <jchrist@linux.ibm.com> Reviewed-by: Tomas Mraz <tomas@openssl.org> Reviewed-by: Patrick Steuer <patrick.steuer@de.ibm.com>
(Merged from https://github.com/openssl/openssl/pull/13984)
Richard Levitte [Tue, 26 Jan 2021 05:48:11 +0000 (06:48 +0100)]
Fix OSSL_PARAM_allocate_from_text() for EBCDIC
OSSL_PARAM_allocate_from_text() converted text values to UTF-8
OSSL_PARAMs with a simple strncpy(). However, if the text is EBCDIC,
that won't become UTF-8. Therefore, it's made to convert from EBCDIC
to ASCII on platforms where the native character encoding is the
former.
One might argue that the conversion should be the responsibility of
the application. However, this is a helper function, and the calling
application can't easily know what sort of OSSL_PARAM the input values
are going to be used for.
Reviewed-by: Matt Caswell <matt@openssl.org> Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13961)
TLS client: allow cert verify callback return -1 for SSL_ERROR_WANT_RETRY_VERIFY
The client-side cert verification callback function may not only return
as usual for success or 0 for failure, but also -1,
typically on failure verifying the server certificate.
This makes the handshake suspend and return control to the calling application
with SSL_ERROR_WANT_RETRY_VERIFY.
The app can for instance fetch further certificates or cert status information
needed for the verification.
Calling SSL_connect() again resumes the connection attempt
by retrying the server certificate verification step.
This process may even be repeated if need be.
The core implementation of the feature is in ssl/statem/statem_clnt.c,
splitting tls_process_server_certificate() into a preparation step
that just copies the certificates received from the server to s->session->peer_chain
(rather than having them in a local variable at first) and returns to the state machine,
and a post-processing step in tls_post_process_server_certificate() that can be repeated:
Try verifying the current contents of s->session->peer_chain basically as before,
but give the verification callback function the chance to pause connecting and
make the TLS state machine later call tls_post_process_server_certificate() again.
Otherwise processing continues as usual.
The documentation of the new feature is added to SSL_CTX_set_cert_verify_callback.pod
and SSL_want.pod.
This adds two tests:
* A generic test in test/helpers/handshake.c
on the usability of the new server cert verification retry feature.
It is triggered via test/ssl-tests/03-custom_verify.cnf.in (while the bulky auto-
generated changes to test/ssl-tests/03-custom_verify.cnf can be basically ignored).
* A test in test/sslapitest.c that demonstrates the effectiveness of the approach
for augmenting the cert chain provided by the server in between SSL_connect() calls.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13906)
Matt Caswell [Mon, 18 Jan 2021 16:05:43 +0000 (16:05 +0000)]
Add EVP_PKEY functions to get EC conv form and field type
libssl at the moment downgrades an EVP_PKEY to an EC_KEY object in order
to get the conv form and field type. Instead we provide EVP_PKEY level
functions to do this.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13139)
OCSP HTTP: Restore API of undocumented and recently deprecated functions
Restore parameters of OCSP_REQ_CTX_new(), OCSP_REQ_CTX_http(), OCSP_REQ_CTX_i2d().
Fix a bug (wrong HTTP method selected on req == NULL in OCSP_sendreq_new().
Minor further fixes in OSSL_HTTP_REQ_CTX.pod
Fixes #13873
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13898)
Matt Caswell [Mon, 18 Jan 2021 16:50:07 +0000 (16:50 +0000)]
Don't copy parameters on setting a key in libssl
Whenever we set a private key in libssl, we first found the certificate
that matched the key algorithm. Then we copied the key parameters from the
private key into the public key for the certficate before finally checking
that the private key matched the public key in the certificate. This makes
no sense! Part of checking the private key is to make sure that the
parameters match. It seems that this code has been present since SSLeay.
Perhaps at some point it made sense to do this - but it doesn't any more.
We remove that piece of code altogether. The previous code also had the
undocumented side effect of removing the certificate if the key didn't
match. This makes sense if you've just overwritten the parameters in the
certificate with bad values - but doesn't seem to otherwise. I've also
removed that error logic.
Due to issue #13893, the public key associated with the certificate is
always a legacy key. EVP_PKEY_copy_parameters will downgrade the "from"
key to legacy if the target is legacy, so this means that in libssl all
private keys were always downgraded to legacy when they are first set
in the SSL/SSL_CTX. Removing the EVP_PKEY_copy_parameters code has the
added benefit of removing that downgrade.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13899)
Matt Caswell [Tue, 19 Jan 2021 11:36:24 +0000 (11:36 +0000)]
Ensure legacy_asn1_ctrl_to_param can handle MDs not in the OBJ database
The legacy_asn1_ctrl_to_param implementation of
ASN1_PKEY_CTRL_DEFAULT_MD_NID calls EVP_PKEY_get_default_digest_name()
which returns an mdname. Previously we were using OBJ_sn2nid/OBJ_ln2nid
to lookup that name in the OBJ database. However we might get an md name
back that only exists in the namemap, not in the OBJ database. In that
case we need to check the various aliases for the name, to see if one of
those matches the name we are looking for.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13899)
Richard Levitte [Fri, 15 Jan 2021 11:20:25 +0000 (12:20 +0100)]
Unix Makefile generator: separate "simple" shared libraries from import libraries
For Unix like environments, we may have so called "simple" shared
library names (libfoo.so as opposed to libfoo.so.1.2), or we may have
"import" library names associated with a DLL (libfoo.dll.a for
libfoo.dll on Mingw and derivatives).
So far, "import" library names were treated the same as "simple"
shared library names, as some kind of normalization for the Unix way
of doing things.
We now shift to treat them separately, to make it clearer what is
what.
Fixes #13414, incidently
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13875)