This copy would need an update on dupctx but
rather than doing it just remove the copy.
This fixes failures of evp_test on Windows with
new CPUs.
Reviewed-by: Todd Short <todd.short@me.com>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22384)
The semantics of ossl_ffc_validate_public_key() and
ossl_ffc_validate_public_key_partial() needs to be changed
to not return error on non-fatal problems.
Fixes#22287
Reviewed-by: Todd Short <todd.short@me.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22291)
This would happen if EVP_MD_get_block_size() returned 0
so we return an error instead.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21698)
The function `ossl_blake2b_param_init` should initialise only, and not
read the data it is initialising
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/22282)
With pre-3.0 OpenSSL, EVP_PKEY_print_private() calls the EVP_PKEY_ASN1_METHOD
function "priv_print", effectively asking the backend to print whatever it
regards as private key components.
In all backends that were built into libcrypto, this function printed what
was included in the private key structure, which usually includes the
public key components as well.
With OpenSSL 3.0, some of the corresponding key2text encoders got a
slightly different behavior, where the presence of the selector
OSSL_KEYMGMT_SELECT_PRIVATE_KEY without the presence of the selector
OSSL_KEYMGMT_SELECT_PUBLIC_KEY would only get what would intuitively be
regarded as private key components printed. This isn't entirely consistent,
though, as the RSA key2text encoder will still print the public key
components regardless.
To compensate for the changed backend behavior, EVP_PKEY_print_private()
was made to ask the encoder to print the keypair rather than just the
private key, thereby moving the backend semantics to the application API.
Unfortunately, this causes confusion for providers where the key2text
encoder really should print the private key only.
This change restores the built-in 1.1.1 backend behavior in the encoders
that OpenSSL provides, and renders EVP_PKEY_print_private() more true to its
documented behavior, leaving it to the backend to decide what it regards as
"private key components".
Fixes#22233
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22237)
Remove the call to rand() and replace with an xor-shift RNG.
There are no security implications to worry about here. This RNG is
used during testing only.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/22211)
Rather than instantiate the private and primary DRBGs during the
selftest, instead use a test RNG. This leaves the DRBG setup
pristine and permits later replacement of the seed source despite
the very early running power up self tests.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21964)
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22098)
In the dupctx fixups I missed a pointer that needed to be repointed to
the surrounding structures AES_KEY structure for the sm4/aes/aria
ccm/gcm variants. This caused a colliding use of the key and possible
use after free issues.
Fixes#22076
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22102)
We had some use of the C11 _Static_assert feature which can cause some
problems on some platforms. Everywhere we were using it, it is not really
required so remove it.
Fixes#22017
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22091)
Fixes#22089
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22090)
For some reason, the code here was made to got through the provider
specific init functions. This is very very dangerous if the provider
specific functions were to change in any way (such as changes to the
implementation context structure).
Instead, use the init functions from the base blake2 implementations
directly.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22079)
This affects only RSA-PSS keys with params using
negative salt legth, or in case of out of memory.
This fixes a memory leak reported in #22049.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22061)
Pretty straightforward, just clone the requested context, no pointers to
fixup
Fixes#21887
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21933)
Same as chacha20 in the last commit, just clone the ctx and its
underlying tlsmac array if its allocated
Fixes#21887
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21933)
This cipher family has a dupctx function, but was failing because it was
attempting to memdup a field only if it was null
Fix the conditional check to get it working again
Fixes#21887
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21933)
create a dupctx method for aes_WRAP implementations of all sizes
Fixes#21887
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21933)
Add dupctx method support to to ciphers implemented with IMPLEMENT_aead_cipher
This includes:
aes-<kbits>-gcm
aria-<kbits>-ccm
aria-<kbits>-gcm
sm4-<kibs>-gcm
Fixes#21887
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21933)
Found by running the checkpatch.pl Linux script to enforce coding style.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21468)
Found by running the checkpatch.pl Linux script to enforce coding style.
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21468)
Implement the dupctx method for the chacha20 cipher, so that
EVP_PKEY_CTX_copy works
Its pretty straightforward, its basically just a memdup. Checking the
pointers that might need fixing up:
in PROV_CHACHA20_CTX all members are statically declared, so memduping
should be fine
in PROV_CHACHA20_CTX->base (PROV_CIPHER_CTX):
Non statically declared members:
*tlsmac needs to get memduped to avoid double free
conditions, but only if base.alloced is set
*hw pointer is always assigned to the chacha20_hw global
variable, so can be left alone
*libctx can be left alone as provctx is always NULL in
chacha20_newctx
*ks appears unused by chacha20, so can be ignored
Fixes#20978
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21878)
Fixes#21198
decoder objects were setting propq as NULL.
Added a set_ctx/settable_ctx to all decoders that should supply
a property query parameter to internal functions.
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21219)
When decoding 0 as the selection means to decode anything
you get.
However when exporting and then importing the key data 0 as
selection is not meaningful.
So we set it to OSSL_KEYMGMT_SELECT_ALL to make the export/import
function export/import everything that we have decoded.
Fixes#21493
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Todd Short <todd.short@me.com>
(Merged from https://github.com/openssl/openssl/pull/21519)
This is already correct in the rsa_kmgmt.c but other
implementations are wrong.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Todd Short <todd.short@me.com>
(Merged from https://github.com/openssl/openssl/pull/21519)
msblob only decodes public/private keys (not just params).
pvk only decodes private keys.
If the requested selection doesn't intersect with the above then don't
consider those decoders.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21603)
The hardware-assisted ARMv8.2 implementation is already in keccak1600-armv8.pl.
It is not called because the author mentioned that it's not actually obvious
that it will provide performance improvements. The test on Apple M1 Firestorm
shows that the ARMv8.2 implementation could improve about 36% for large blocks.
So let's enable ARMv8.2 accelerated SHA3 on Apple CPU family.
Fixes#21380
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21398)
The most expensive part of using a PKEY decoder is the
OSSL_DECODER_CTX_new_for_pkey() call. This builds up all of the decoder
chains, which is a complex and time consuming operation. However, if no
new providers have been loaded/unloaded since the last time it was called
we can expect the same results for the same parameters. Note that this
operation takes place *before* we event parse the data for decoding so it
is not dependent on the parsed data at all.
We introduce a cache for OSSL_DECODER_CTX objects. If we have been called
with the same parameters then we just duplicate an existing
OSSL_DECODER_CTX. This should be significantly faster than creating a new
one every time.
Partially addressed the issue in #15199
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21426)
The AES-SIV mode allows for multiple associated data items
authenticated separately with any of these being 0 length.
The provided implementation ignores such empty associated data
which is incorrect in regards to the RFC 5297 and is also
a security issue because such empty associated data then become
unauthenticated if an application expects to authenticate them.
Fixes CVE-2023-2975
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21384)
The implementation is not usable there at all.
Fixes#21301
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21312)
(cherry picked from commit b256d32915)
This is defined in NIST SP 800-208 as the truncation to 192 bits of
SHA256. Unlike other truncated hashes in the SHA2 suite, this variant
doesn't have a different initial state, it is just a pure truncation
of the output.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/21180)
Typos in doc/man* will be fixed in a different commit.
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/20910)
This can effectively reduce the binary size for platforms
that don't need ECX feature(~100KB).
Signed-off-by: Yi Li <yi1.li@intel.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/20781)
The separate file is a Perl script that generates the appropriate define
directives for inclusion in core_names.h. By having this separation it
will be possible to prebuild data structures to give faster access when
looking up parameters by name.
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/20935)
Even if a DRBG has locking enabled on it, there are certain parameters
which are still safe to obtain even without a lock. The max_request
value is constant for all our DRBGs. The reseed_counter does not matter
if we get it wrong - so it is safe to avoid the lock. So if all we are
reading are those parameters then we take no lock at all.
Partially fixes#20286
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Todd Short <todd.short@me.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/20970)
Previously the EVP layer would call lock and unlock functions on the
underlying DRBG implementation to say when a lock should be acquired and
released. This gives the DRBG implementation no say as to what kind of
lock should obtained (e.g. read/write) or even whether a lock is actually
needed or not.
In reality we know whether a DRBG is supposed to be in locking mode or
not because the "enable_locking()" function will have been called if
locks should be used. Therefore we re-interpret the lock and unlock
functions as "hints" from the EVP layer which we ignore. Instead we
acquire locks only when we need them. By knowing the context we can obtain
either a read or a write lock as appropriate.
This may mean that in some rare cases we acquire the locks more than once
for a single EVP call, if the EVP call makes several calls to the underlying
DRBG. But in practice almost all EVP calls only make one such call.
EVP_RAND_generate() is an example of a call where multiple DRBG calls may
be made. One of these gets the "max_request" parameter (which is constant
for all of our own DRBGs) and it may make several calls to the DRBG generate
call - but only if the requested size is very large which will rarely be
the case.
Partially fixes#20286
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Todd Short <todd.short@me.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/20970)
Fixes#20993
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/20994)
The expression "if (a+b>c) a=c-b" is incorrect if "a+b" overflows.
It should be replaced by "if (a>c-b) a=c-b", which avoids the
potential overflow and is much easier to understand.
This pattern is the root cause of CVE-2022-37454, a buffer overflow
vulnerability in the "official" SHA-3 implementation.
It has been confirmed that the addition in
https://github.com/openssl/openssl/blob/master/providers/implementations/kdfs/hkdf.c#L534
cannot overflow. So this is only a minor change proposal to avoid
a potentially vulnerable code pattern and to improve readability.
More information: https://github.com/github/codeql/pull/12036#issuecomment-1466056959
CLA: trivial
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/20990)
Tomas Mraz
Klavishnik
Matt Caswell
Richard Levitte
Pauli
Dimitri Papadopoulos
Neil Horman
Bernd Edlinger
slontis
sdlyyxy
Marcin Kolny
Fergus Dall
Yi Li
Michael Baentsch
Dmitry Belyavskiy
Nicky Mouha