The passed 'pkey' already contains a library context, and the encoder
implementations should be found within the same context, so passing an
explicit library context seems unnecessary, and potentially dangerous.
It should be noted that it's possible to pass an EVP_PKEY with a
legacy internal key. The condition there is that it doesn't have a
library context assigned to it, so the NULL library context is used
automatically, thus requiring that appropriate encoders are available
through that context.
Fixes#13544
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13545)
When printing human readable info on the Thawte Strong Extranet extension
the version number could overflow if the version number == LONG_MAX. This
is undefined behaviour.
Issue found by OSSFuzz.
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/13452)
The X942 KDF had been modified so that it supports all optional fields - not
just the fields used by CMS.
As there are 2 types of KDF for X942 - this has been made a bit clearer
by adding an X942KDF-ASN1 alias. X942KDF-CONCAT has also been added as an
alias of X963KDF.
This work was instigated as a result of the ACVP tests optionally being
able to use keybits for the supp_pubinfo field.
Setting the parameter OSSL_KDF_PARAM_X942_USE_KEYBITS to 0 allows this
to be disabled.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/13418)
ARM Cortex-A57 and Cortex-A72 cores running in 32-bit mode are affected
by silicon errata #1742098 [0] and #1655431 [1], respectively, where the
second instruction of a AES instruction pair may execute twice if an
interrupt is taken right after the first instruction consumes an input
register of which a single 32-bit lane has been updated the last time it
was modified.
This is not such a rare occurrence as it may seem: in counter mode, only
the least significant 32-bit word is incremented in the absence of a
carry, which makes our counter mode implementation susceptible to these
errata.
So let's shuffle the counter assignments around a bit so that the most
recent updates when the AES instruction pair executes are 128-bit wide.
[0] ARM-EPM-049219 v23 Cortex-A57 MPCore Software Developers Errata Notice
[1] ARM-EPM-012079 v11.0 Cortex-A72 MPCore Software Developers Errata Notice
Signed-off-by: Ard Biesheuvel <ard.biesheuvel@arm.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/13504)
Addition using the NULL pointer (even when adding 0) is undefined
behaviour. Recent versions of ubsan are now complaining about this, so
we fix various instances.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13513)
ERR_get_error_line() is deprecated, and ERR_get_error_func() and
ERR_get_error_data() are removed in favor of ERR_get_error_all(),
since they pop the error record, leaving the caller with only partial
error record data and no way to get the rest if the wish.
If it's desirable to retrieve data piecemeal, the caller should
consider using the diverse ERR_peek functions and finish off with
ERR_get_error().
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13466)
EVP_PKEY_CTX_set_dsa_ functions were only available when DSA was
enabled ('no-dsa' not configured). However, that makes it impossible
to use these functions with an engine or a provider that happens to
implement DSA. This change solves that problem by shuffling these
functions to more appropriate places.
Fixes#13529
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13530)
CLA: trivial
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/13515)
Fixes#13057
When using an engine, there is no cipher->prov so a call to
EVP_CIPHER_CTX_set_padding() returns an error when
evp_do_ciph_ctx_setparams() is called. For the legacy path it needs to
avoid doing the call and just return 1.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/13437)
This commit tries to address a locking issue in evp_pkey_reset_unlocked
which can occur when it is called from evp_pkey_downgrade.
evp_pkey_downgrade will acquire a lock for pk->lock and if successful
then call evp_pkey_reset_unlocked. evp_pkey_reset_unlocked will call
memset on pk, and then create a new lock and set pk->lock to point to
that new lock. I believe there are two problems with this.
The first is that after the call to memset, another thread would try to
acquire a lock for NULL as that is what the value of pk->lock would be
at that point.
The second issue is that after the new lock has been assigned to
pk->lock, that lock is different from the one currently locked so
another thread trying to acquire the lock will succeed which can lead to
strange behaviour. More details and a reproducer can be found in the
Refs link below.
This changes the evp_pkey_reset_unlocked to not touch the lock
and the creation of a new lock is done in EVP_PKEY_new.
Refs:
https://github.com/danbev/learning-libcrypto/blob/master/notes/issues.md#openssl-investigationtroubleshootinghttps://github.com/nodejs/node/issues/29817
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
(Merged from https://github.com/openssl/openssl/pull/13374)
It does not do it in legacy path and 1.1.1 so that must not change.
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13402)
The OSSL_STORE code was forgetting the datatype that we read from the
PEM header when decoding the DER.
Fixes#13046
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13329)
This deprecates all the ERR_load_ functions, and moves their definition to
separate C source files that can easily be removed when those functions are
finally removed.
This also reduces include/openssl/kdferr.h to include cryptoerr_legacy.h,
moves the declaration of ERR_load_ERR_strings() from include/openssl/err.h
to include/openssl/cryptoerr_legacy.h, and finally removes the declaration
of ERR_load_DSO_strings(), which was entirely internal anyway.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13390)
This also modifies the .ec L statement to take a third file, which is
the internal header file to declare internal things. This is only
useful for our internal declarations and will not affect engines.
Fixes#10527
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13390)
Fix for the issue #13472. The decoderctx has to be initialized in every
cycle as its constructor may not be called due to lazy evaluation of
the if-condition.
CLA: trivial
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13473)
Previously we only ever allowed one mark to be set against an error in the
statck. If we attempted to nest them, then we would end up clearing all
the errors in the stack when we popped to the mark.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13335)
clang 10 was emitting warnings similar to the following from params.c:
crypto/params.c:411:40: error: implicit conversion from 'long' to 'double' changes value from 9223372036854775807 to 9223372036854775808 [-Werror,-Wimplicit-int-float-conversion]
if (d >= INT64_MIN && d <= INT64_MAX && d == (int64_t)d) {
Also fixed some other conversion problems when sizeof(double) == 4.
Alternative to #13366Fixes#13365
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13377)
These are: keccak_kmac_init(), sha3_final(), sha3_init(), sha3_reset() and
sha3_update().
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13417)
These are: prov_crngt_cleanup_entropy(), prov_crngt_get_entropy(),
prov_pool_acquire_entropy(), prov_pool_add_nonce_data(),
prov_rand_drbg_free() and prov_rand_drbg_new().
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13417)
no-dh disables the low level API for DH. However, since we're now using
the high level EVP API in most places we don't need to disable quite so
much.
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/13368)
Currently, when configuring OpenSSL and specifying the --strict-warnings
option there are failures like the following one:
crypto/bio/bio_lib.c: In function 'BIO_up_ref':
include/internal/refcount.h:169:25: error: format '%p' expects argument
of type 'void *', but argument 3 has type 'BIO *'
{aka 'struct bio_st *'} [-Werror=format=]
169 | fprintf(stderr, "%p:%4d:%s\n", b, b->references, a)
| ^~~~~~~~~~~~~
crypto/bio/bio_lib.c:185:5:
note: in expansion of macro'REF_PRINT_COUNT'
185 | REF_PRINT_COUNT("BIO", a);
| ^~~~~~~~~~~~~~~
include/internal/refcount.h:169:27: note: format string is defined here
169 | fprintf(stderr, "%p:%4d:%s\n", b, b->references, a)
| ~^
| |
| void *
cc1: all warnings being treated as errors
This commit adds casts to avoid the warnings.
Reviewed-by: Tomas Mraz <tmraz@fedoraproject.org>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13389)
When UNICODE is defined, Windows headers push for WCHAR implementations,
which aren't compatible with POSIX declarations.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/13318)
