Providers can create errors that may refer to const strings within the
provider module itself. If the provider gets unloaded we need to be sure
that we can still access the errors in the error stack.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14213)
Without this, a provider has no way to know that an application
has finished with the array it returned earlier. A non-caching provider
requires this information.
Fixes#12974
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/12974)
Add an argument to PROVIDER_try_load() that permits a provider to be
loaded without changing the fallback status. This is useful when an
additional provider needs to be loaded without perturbing any other setup.
E.g. adding mock providers as part of unit testing.
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13652)
The EVP_DigestSignInit and EVP_DigestVerifyInit actually have to
be initialized before EVP_PKEY_CTX_ctrl_str is invoked.
Otherwise, when the ctx not initialized, the ctrl command fails.
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13828)
The dh->nid was not being set if the loaded p,g matched an inbuilt named
group for "DH".
NOTE: The "DHX" related path already worked since it calls DH_set0_pqg()
(which does the name group check).
This bug was detected when new tests were added for dh5114 groups, combined
with the no-cache tests i.e. loading+import+export set the nid,
but just loading did not.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14207)
The apps helper function load_key_certs_crls() is a general purpose
function for loading different types of objects from a given URI. It
sets up an OSSL_STORE and calls OSSL_STORE_expect() so that the store
knows what type of thing to expect to load. Unfortunately this wasn't
working and was always setting "expect" to 0 - which means "anything".
Fixes#13709
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14191)
When the string "ABCDEFGH" is passed, what's considered its data, this?
{ 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H' }
or this?
{ 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', '\0' }
If it's passed as a pass phrase, should the terminating NUL byte be
considered part of the pass phrase, or not?
Our treatment of OSSL_PARAMs with the data type OSSL_PARAM_UTF8_STRING
set the length of the string to include the terminating NUL byte,
which is quite confusing. What should the recipient of such a string
believe?
Instead of perpetuating this confusion, we change the assumption to
set the OSSL_PARAM to the length of the string, not including the
terminating NUL byte, thereby giving it the same value as a strlen()
call would give.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14168)
Added primality check on p and q in the ossl_ffc_params_simple_validate().
Checking for p and q sizes in the default provider is made more
lenient.
Added two testcases for invalid parameters.
Fixes#13950
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/14148)
Additional renames done in encoder and decoder implementation
to follow the style.
Fixes#13622
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14155)
We test all three cases:
- An SSLv2 only client talking to a TLS capable server
- A TLS capable client talking to an SSLv2 only server
- A TLS capable client talking to a TLS capable server (should fail due
to detecting a rollback attack)
Reviewed-by: Paul Dale <pauli@openssl.org>
genpkey can sometimes create files that fail "openssl dhparam -check". See
issue #14145. We had some instances of such invalid files in the
dhparam_check test. Now that "openssl dhparam -check" has been fixed to
work the same way as it did in 1.1.1 these tests were failing. We move the
invalid files inot the "invalid" directory. A future PR will have to fix
genpkey to not generate invalid files.
We also remove a "SKIP" block that was skipping tests in a no deprecated
build unnecessarily. Nothing being tested is deprecated.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/14146)
The low level SRP implementation has been deprecated with no replacement.
Therefore the libssl level APIs need to be similarly deprecated.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14132)
The OTC decided that all low level APIs should be deprecated. This extends
to SRP, even though at the current time there is no "EVP" interface to it.
This could be added in a future release.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14132)
Without these hooks, if the TLS provider isn't matched in the fetch cache, a test
failure will occur in the TLS API tests. Without allowing import and export, an
existing key can not move to a new key manager even if it is really the same.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14126)
In a caching world, it's fine to compare the pointers directly. In a
non-caching world, the names and providers need to be compared.
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14126)
The existing code prints a warning saying that verbose mode is ignored with
parallel jobs. This seems backward, more useful is disabling parallel jobs
when verbose is enabled.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14137)
This 'special' way of specifying the data should only be used for testing
purposes. It should not be used in production environments.
ACVP passes a blob of DER encoded data for some of the fields rather
than passing them as separate fields that need to be DER encoded.
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14077)
The functions are obsolete aliases for BN_rand() and BN_rand_range()
since 1.1.0.
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14080)
Fixes#14041 and additional bugs discovered by the newly created
tests.
This patch:
- Introduces support for 0x prefixed integers
- Fixes parsing of negative integers (negative numbers were
shifted by -2)
- Fixes ability to parse maximal unsigned numbers ("too small
buffer" error used to be reported incorrectly)
- Fixes a memory leak when OSSL_PARAM_allocate_from_text fails
leaving a temporary BN allocated
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14093)
The existing names such as EVP_PKEY_param_fromdata_settable were a bit
confusing since the 'param' referred to key params not OSSL_PARAM. To simplify
the interface a 'selection' parameter will be passed instead. The
changes are:
(1) EVP_PKEY_fromdata_init() replaces both EVP_PKEY_key_fromdata_init() and EVP_PKEY_param_fromdata_init().
(2) EVP_PKEY_fromdata() has an additional selection parameter.
(3) EVP_PKEY_fromdata_settable() replaces EVP_PKEY_key_fromdata_settable() and EVP_PKEY_param_fromdata_settable().
EVP_PKEY_fromdata_settable() also uses a selection parameter.
Fixes#12989
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14076)
Now that libssl no longer has any OPENSSL_NO_ALG guards the internal
cipher_overhead_test wasn't quite handling disabled ciphers correctly.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13916)
Even if EC and DH are disabled then we may still be able to use TLSv1.3
if we have groups that have been plugged in by an external provider.
Fixes#13767
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13916)
Providers produce algorithm IDs of their own, and we need to compare
them against the same thing produced by libcrypto's ASN.1 code and
with legacy keys.
This tester can compare algorithm IDs for signatures and for keys,
given certificates that hold such data.
To verify key algorithm IDs, only one certificate is necessary, and
its public key is used.
To verify certificate algorithm IDs, we need to launch the signature
operation that would verify a certificate against the public key of
its signing CA, so that test needs two files.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/14049)
Fixes#13969
- properly handle the mandatory RSA-PSS key parameters
- improve parameter checking when setting the parameters
- compute the algorithm id at the time it is requested so it
reflects the actual parameters set
- when generating keys do not override previously set parameters
with defaults
- tests added to the test_req recipe that should cover the PSS signature
handling
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13988)
Fixes#13656. Right now all openssl commands use a NULL propq. This
patch adds a possibility to specify a custom propq.
The implementation follows the example of set_nameopt/get_nameopt.
Various tools had to be modified to call app_get0_propq after it has
been populated. Otherwise the -propquery has no effect.
The tests then verify the -propquery affects the tool behaviour by
requesting a non-existing property.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13707)
The HTTP-based tests are now in 80_test_cmp_http.t, to start a little earlier.
This should decrease total test run time due to better parallelization.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13551)
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)
