We also lay the ground work for various of other the basic AES ciphers.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8700)
The comparator further down the call stack doesn't tolerate NULL, so
if we got that as input, use the empty string.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8781)
OPENSSL_LH_flush() frees the linked lists for each slot, but didn't
set the list head to NULL after doing so, with the result that an
operation that affects these lists is likely to cause a crash.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8781)
ossl_method_store_cache_get() and ossl_method_store_cache_set() were
called with a NULL argument for store, which means no caching is
done. Give them a real store instead.
Also, increment the refcount when we do get a method out of the cache.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8781)
The change is triggered by ThunderX2 where 3+1 was slower than scalar
code path, but it helps all processors [to handle <512 inputs].
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8776)
The callback should be called with 1 when a Miller-Rabin round marked
the candidate as probably prime.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
GH: #8742
This allows the provider digest_final operation to check that it
doesn't over-run the output buffer.
The EVP_DigestFinal_ex function doesn't take that same parameter, so
it will have to assume that the user provided a properly sized buffer,
but this leaves better room for future enhancements of the public API.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8747)
The rep parameter takes an int in C, but the assembly implementation
looks at the upper bits. While it's unlikely to happen here, where all
calls pass a constant, in other scenarios x86_64 compilers will leave
arbitrary values in the upper half.
Fix this by making the C prototype match the assembly. (This aspect of
the calling convention implies smaller-than-word arguments in assembly
functions should be avoided. There are far fewer things to test if
everything consistently takes word-sized arguments.)
This was found as part of ABI testing work in BoringSSL.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/8108)
This happens in ec_key_simple_check_key and EC_GROUP_check.
Since the the group order is not a secret scalar, it is
unnecessary to use coordinate blinding.
Fixes: #8731
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8734)
These undocumented functions were never integrated into the EVP layer
and implement the AES Infinite Garble Extension (IGE) mode and AES
Bi-directional IGE mode. These modes were never formally standardised
and usage of these functions is believed to be very small. In particular
AES_bi_ige_encrypt() has a known bug. It accepts 2 AES keys, but only
one is ever used. The security implications are believed to be minimal,
but this issue was never fixed for backwards compatibility reasons.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8710)
Refer to NIST SP 800-90C section 5.4 "Prediction Resistance.l"
This requires the seed sources to be approved as entropy sources, after
which they should be considered live sources as per section 5.3.2 "Live
Entropy Source Availability."
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/8647)
This fixes the "verifying the alias" case.
Actually, while working on it, I realized that conceptually we were
testing the 2 different behaviours of `EC_GROUP_check_named_curve()` at
the same time, and actually not in the proper way.
I think it's fair to assume that overwriting the curve name for an
existing group with `NID_undef` could lead to the unexpected behaviour
we were observing and working around.
Thus I decided to separate the lookup test in a dedicated simpler test
that does what the documentation of `EC_GROUP_check_named_curve()`
suggests: the lookup functionality is meant to find a name for a group
generated with explicit parameters.
In case an alternative alias is returned by the lookup instead of the
expected nid, to avoid doing comparisons between `EC_GROUP`s with
different `EC_METHOD`s, the workaround is to retrieve the `ECPARAMETERS`
of the "alias group" and create a new explicit parameters group to use
in `EC_GROUP_cmp()`.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8555)
Even with custome ciphers, the combination in == NULL && inl == 0
should not be passed down to the backend cipher function. The reason
is that these are the values passed by EVP_*Final, and some of the
backend cipher functions do check for these to see if a "final" call
is made.
Fixes#8675
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8676)
This commit makes the X509_set_sm2_id to 'set0' behaviour, which means
the memory management is passed to X509 and user doesn't need to free
the sm2_id parameter later. API name also changes to X509_set0_sm2_id.
Document and test case are also updated.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8626)
When the purpose is to pass parameters to a setter function, that
setter function needs to know the size of the data passed. This
remains true for the pointer data types as well.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8703)
If using a custom X509_LOOKUP_METHOD then calls to
X509_STORE_CTX_get_by_subject may crash due to an incorrectly initialised
X509_OBJECT being passed to the callback get_by_subject function.
Fixes#8673
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8698)
When we attempt to fetch a method with a given NID we will ask the
providers for it if we don't already know about it. During that process
we may be told about other methods with a different NID. We need to
make sure we don't confuse the two.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8541)
OSSL_PARAM_END is a macro that can only be used to initialize an
OSSL_PARAM array, not to assign an array element later on. For
completion, we add an end constructor to facilitate that kind of
assignment.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8704)
ISO C90 forbids specifying subobject to initialize
Signed-off-by: Patrick Steuer <patrick.steuer@de.ibm.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8693)
Found by Coverity.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8274)
The value of the 'default_properties' command is simply passed to
EVP_set_default_properties().
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8681)
EVP_MD_fetch() can be given a property query string. However, there
are cases when it won't, for example in implicit fetches. Therefore,
we also need a way to set a global property query string to be used in
all subsequent fetches. This also applies to all future algorithm
fetching functions.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8681)
Additionally, merge ENGINE_CONF into CONF.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/8680)
Other configuration modules may have use for tracing, and having one
tracing category for each of them is a bit much. Instead, we make one
category for them all.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/8680)
It was assumed that the config functionality returned a boolean.
However, it may return a negative number on error, so we need to take
that into account.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8679)
The three macros EVP_F_AESNI_XTS_INIT_KEY, EVP_F_AES_T4_XTS_INIT_KEY
and EVP_F_AES_XTS_INIT_KEY are affected.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8682)
than the update call. The means an earlier error return at the cost of some
duplicated code.
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/8625)
This prevents failure of openssl s_server socket binding to wildcard
address on hosts with disabled IPv6.
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8550)
This configuration module supports a configuration structure pretty
much like the engine configuration module, i.e. something like this:
openssl_conf = openssl_init
[openssl_init]
providers = provider_section
[provider_section]
# Configure the provider named "foo"
foo = foo_section
# Configure the provider named "bar"
bar = bar_section
[foo_section]
# Override name given in the provider section
identity = myfoo
# The exact path of the module. This is platform specific
module_path = /opt/openssl/modules/foo.so
# Whether it should be automatically activated. Value is unimportant
activate = whatever
# Anything else goes as well, and becomes parameters that the
# provider can get
what = 1
# sub-sections will be followed as well
ever = ever_section
[ever_section]
cookie = monster
All the configurations in a provider section and its sub-sections
become parameters for the provider to get, i.e. the "foo" provider
will be able to get values for the following keys (with associated
values shown):
identity => myfoo
module_path => /opt/openssl/modules/foo.so
activate => whatever
what => 1
ever.cookie => monster
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8549)
Provider parameters are parameters set by the core that the provider
can retrieve. The primary use it to support making OpenSSL
configuration data available to the provider.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8549)
Limit the number of AES blocks in a data unit to 2^20 or less.
This corresponds to the mandates in IEEE Std 1619-2018 and NIST SP 800-38E.
Note: that this is a change from IEEE Std 1619-2007 which only recommended
this limit.
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/8627)
I turns out that this made crypto/rand/rand_win.c to never build with
BCrypt support unless the user sets _WIN32_WINNT. That wasn't the
intent.
This reverts commit cc8926ec8f.
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/8641)
Add support for Linux TLS Rx offload in the BIO layer.
Change-Id: I79924b25dd290a873d69f6c8d429e1f5bb2c3365
Signed-off-by: Boris Pismenny <borisp@mellanox.com>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7848)
'no-dso' is meaningless, as it doesn't get any macro defined.
Therefore, we remove all checks of OPENSSL_NO_DSO. However, there may
be some odd platforms with no DSO scheme. For those, we generate the
internal macro DSO_NONE aand use it.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/#8622)
It seems more intuitive to set `OPENSSL_TRACE=all` instead of
`OPENSSL_TRACE=any` to obtain trace output for all categories.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8552)
Previously, if the openssl application was run with OPENSSL_TRACE=any,
all trace output would just show 'ANY' as the category name, which was
not very useful. To get the correct category name printed in the trace
output, the openssl application now registers separate channels for
each category.
The trace API is unchanged, it is still possible for an application to
register a single channel for the 'ANY' category to see all outputt,
if it does not need this level of detail.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8552)
When OSSL_trace_get_category_num() is called with an unknown category
name, it returns -1. This case needs to be considered in order to
avoid out-of-bound memory access to the `trace_channels` array.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8552)
Partially reverts d33d76168f Don't fail when tracing is disabled
Commit d33d76168f fixed the problem that the initialization of
libcrypto failed when tracing was disabled, because the unoperational
ossl_trace_init() function returned a failure code. The problem was
fixed by changing its return value from failure to success.
As part of the fix the return values of other unimplemented trace API
functions (like OSSL_trace_set_channel(),OSSL_trace_set_callback())
was changed from failure to success, too. This change was not necessary
and is a bit problematic IMHO, because nobody expects an unimplemented
function to pretend it succeeded.
It's the application's duty to handle the case correctly when the trace
API is not enabled (i.e., OPENSSL_NO_TRACE is defined), not the API's job
to pretend success just to prevent the application from failing.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8552)
>=20% faster than present code.
Signed-off-by: Patrick Steuer <patrick.steuer@de.ibm.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8560)
The initialisation was also flawed, failing to account for padding and
alignment bytes.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8611)
If the structures have empty padding bytes, ensure they are zeroed.
These structures are added to seed pools as complete blocks including
any padding and alignment bytes.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8610)
Revert win32_pathbyaddr() which is used in DSO_dsobyaddr().
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8596)
Replace it with InitializeCriticalSection()
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8596)
Refer to FIPS 140-2 section 4.9.2 Conditional Tests for details.
The check is fairly simplistic, being for the entropy sources to not feed
the DRBG the same block of seed material twice in a row. Only the first
DRBG in a chain is subject to this check, latter DRBGs are assumed to be
safely seeded via the earlier DRBGs.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8599)
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8569)
DH_check is used to test the validity of Diffie-Hellman parameter sets (p, q, g). Among the tests performed are primality tests on p and q, for this BN_is_prime_ex is called with the rounds of Miller-Rabin set as default. This will therefore use the average case error estimates derived from the function BN_prime_checks_for_size based on the bit size of the number tested.
However, these bounds are only accurate on testing random input. Within this testing scenario, where we are checking the validity of a DH parameter set, we can not assert that these parameters are randomly generated. Thus we must treat them as if they are adversarial in nature and increase the rounds of Miller-Rabin performed.
Generally, each round of Miller-Rabin can declare a composite number prime with probability at most (1/4), thus 64 rounds is sufficient in thwarting known generation techniques (even in safe prime settings - see https://eprint.iacr.org/2019/032 for full analysis). The choice of 64 rounds is also consistent with SRP_NUMBER_ITERATIONS_FOR_PRIME 64 as used in srp_Verify_N_and_g in openssl/apps/s_client.c.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8593)
EVP_MAC_ctrl is documented to return 0 or -1 on failure. Numerous places
were not getting this check correct.
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/8584)
We treat that as automatic success. Other EVP_*Update functions already do
this (e.g. EVP_EncryptUpdate, EVP_DecryptUpdate etc). EVP_EncodeUpdate is
a bit of an anomoly. That treats 0 byte input length as an error.
Fixes#8576
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8587)
constant time with a memory access pattern that does not depend
on secret information.
[extended tests]
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/8543)
Great effort has been made to make initialization more configurable.
However, the behavior of OPENSSL_config() was lost in the process,
having it suddenly generate errors it didn't previously, which is not
how it's documented to behave.
A simple setting of default flags fixes this problem.
Fixes#8528
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8533)
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/8527)
OpenSSL will come with a set of well known providers, some of which
need to be accessible from the start. These are typically built in
providers, or providers that will work as fallbacks.
We do this when creating a new provider store, which means that this
will happen in every library context, regardless of if it's the global
default one, or an explicitely created one.
We keep the data about the known providers we want to make accessible
this way in crypto/provider_predefined.h, which may become generated.
For now, though, we make it simple and edited manually.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8480)
To ensure that old applications aren't left without any provider, and
at the same time not forcing any default provider on applications that
know how to deal with them, we device the concept of fallback
providers, which are automatically activated if no other provider is
already activated.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/8480)
