Add test for using KRB5KDF with erroneous key size

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28663)

Update doc/designs/rfc4514.md

Co-authored-by: Andrew Dinh <andrewd@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28639)

Add git pre-commit hook example to the script

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28639)

unbreak url for legacy gost

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28639)

make it not write changes by default

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28639)

Add util/codespell-check.sh and run it

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28639)

Make tests run faster on typical platforms.

Sadly not doable in make as it is notoriously bad at telling
you the parallelism being used by make -j.

If the HARNESS_JOBS environment variable has not been
set, this makes the perl script attempt to figure out how
many cpu's are available on anything windows/linux/macos/bsd like,
and if it can be successfully detected, we use that value.
if not, we use 1 as before.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/28426)

Change test/recipes/95-test_external_oqsprovider.t to allow out-of-source builds

Unfortunately, CMake's FindOpenSSL.cmake module doesn't handle OpenSSL's
build tree very well when it's out-of-source.  This is resolved by create
a local OpenSSL "installation" with a minimum amount of symbolic links,
and using that.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28638)

Fix EVP_PKEY_can_sign() handling of NULL from query_operation_name()

EVP_PKEY_can_sign() assumed query_operation_name(OSSL_OP_SIGNATURE)
always returns a non-NULL string. According to the documentation,
query_operation_name() may return NULL, in which case
EVP_KEYMGMT_get0_name() should be used as a fallback.

Fixes #27790

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28620)

Check for OBJ_create() conflicts after write lock.

For now subsequent calls to OBJ_create() with identical inputs return
NID_undef.  It may be better to return the previous NID in the future.

The real work actually happens in OBJ_add_object().  Duplicate compares
*all* the input object's fields with any of the objects found by lookup.

If these are identical, then necessarily all the lookups found the same
data, and we can return the existing nid in low-level calls via
OBJ_add_object() that specify the nid also.  If any of the fields are
different the new object is not installed and NID_undef is returned.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28582)

Add a test for multi-threaded OBJ_create

After a successful OBJ_create the returned NID should
be the same NID that is returned from OBJ_ln2nid and
should not change any more, but after an unsuccessful
OBJ_create, another thread must have created the object,
therefore OBJ_ln2nid should not return NID_undef in that
case.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28582)

Fix length of digestinfo_sm3_der

This fixes the RSA-SM3 signatures to conform to the standard.

CLA: trivial

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28415)

Fix doublefree after failure in ossl_siv128_init()

The issue was reported by Ronald Crane from Zippenhop LLC.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28644)

Print PowerPC CPUINFO

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28535)

Const correct the various sk_find functions.

Since we no longer mutate the stack when finding, let us
make the stack pointers const.

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28532)

Correct the documentation for OPENSSL_sk_find

Since April of 2023 with commit eb0935f, these functions have not
sorted the stack if it was not sorted. The documentation was noti
changed at the time to reflect this changed behaviour.

This corrects the documentation to reflect the current behaviour
of these functions

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28532)

Close small race condition on error raising in QUIC

Github issue #28501 reported an odd condition in which a double free was
occuring when a given thread was popping entries of its error stack.

It was hypothesized that, because a few places in the quic stack save
error state to a shared structure (ch->err_state, port->error_state,
qtls->error_state), that multiple threads may attempt to mutate the
shared structure during error save/restore in parallel.

Investigation showed that all paths which led to such mutations were
done under lock, so that shouldn't occur.

Except for one case, which this PR addresses.

In ossl_quic_conn_stream_conclude, we unlock our protecting mutex, prior
to calling QUIC_RAISE_NON_NORMAL_ERROR.  If that function is called with
an reason code of SHUTDOWN, it attempts to restore the channel error
state.  Given that the lock was released first, this creates a small
race condition in which two threads may manipulate the shared error
state in the channel struct in parallel.

According to the reporter, applying this patch prevents the reported
error from occuring again.

Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28642)

Disable HW acceleration when running TLSProxy tests

IBM reported that, since the introduction of the ossltest provider,
several tests are failing on s390x

They are failing because MAC computation on s390x uses hardware
acceleration, which bypasses the use of the ossltest provided digests.

Because TLSProxy tests rely on the return of known consistent data
(which the ossltest provider gives us), the HW acceleration ignores this
giving us real MAC's instead, causing the test to fail.

We could write an HMAC provided algorithm to override that behavior, but
since this only occurs on s390, and only in cases in which the
TLSprovider is used, it seems simpler, and more maintainable to just
disable hw acceleration on those tests (given that we don't want to use
the accelerated path anyway).

Set the capabilities register for s390 to enforce this in the TLSProxy
code.

Fixes #28630

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28633)

Skip LMS tests if fetch for the LMS algorithm fails

Check for specific LMS error

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28580)

Do not use RW mutexes on RISC-V arch

For unknown reasons using RW mutexes on RISC-V arch
seems to be broken, at least with glibc.

Fixes #28550

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28634)

Add Linux Foundation Health Score badge to README

CLA: trivial

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28614)

x509: fix mem leak on error path

The x509_store_add() creates X509_OBJECT wrapping either X509 or
X509_CRL. However, if you set the type to X509_LU_NONE before
X509_OBJECT_free then it skips the free on the wrapped type and just
calls OPENSSL_free on the object itself. Hence, leaking wrapped
object.

Signed-off-by: Nikola Pajkovsky <nikolap@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28631)

Re-enable the ssl_trace_test()

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28627)

sslapitest.c: Skip test_ssl_trace() with FIPS providers older than 3.5

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28627)

tls_common.c: Handle inner content type properly on Big Endian

When passing the inner content type to msg_callback,
the lowest byte of rec->type needs to be passed instead
of directly passing the rec->type otherwise the value is
incorrect on Big Endian platforms.

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28627)

Fix the abnormal branch memory leak in ssl_set_cert_and_key function

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28619)

Ensure q variable is freed

Fixes: d88c43a64408 ("Ensure that empty or 1 element stacks are always sorted.")
Resolves: https://scan5.scan.coverity.com/#/project-view/65138/10222?selectedIssue=1665465
Signed-off-by: Nikola Pajkovsky <nikolap@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28596)

Correct version used to check for elf_aux_info() on FreeBSD

FreeBSD also added elf_aux_info() to the 11 branch and was shipped
with 11.4.

https://github.com/freebsd/freebsd-src/commit/03444a7d439755189e4dc8ccd56403bbaef3d6b0

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28581)

bio_ok.c: Integer Overflow in BIO_f_reliable record parser leads to Out-of-Bounds Read

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28504)

Added test suggested by Shane Lontis

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/28624)

Harden property put_str() helper corner case

The put_str() helper of the internal ossl_property_list_to_string()
function failed to correctly check the remaining buffer length in a
corner case in which a property name or string value needs quoting,
and exactly one byte of unused space remained in the output buffer.

The only potentially affected calling code is conditionally compiled
(disabled by default) provider "QUERY" tracing that is executed only
when also requested at runtime.  An initial fragment of the property
list encoding would need to use up exactly 511 bytes, leaving just 1
byte for the next string which requires quoting.  Bug reported by

    Aniruddhan Murali (@ashamedbit)
    Noble Saji Mathews (@NobleMathews)

both from the University of Waterloo.

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/28624)

Add unsupported features in NOTES-C99.md: complex.h and variable length array

Resolves: https://github.com/openssl/openssl/issues/28598

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/28602)

Make generated files read-only

This is an attempt to discouraged manual changes of generated files,
as people have done so, just to get their changes over-written next
time those files are re-generated.

Reference: https://github.com/openssl/openssl/discussions/28269

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28612)

Fix OPENSSL_VERSION_NUMBER to always have zero status bits

The documentation suggested that they were always zero, while the
implementation in <openssl/opensslv.h> suggested that it could be
0xf in OpenSSL releases...  which (almost) never happened because
of a bug in said implementation.

Therefore, we solidify that the status bits are indeed always zero,
at least in all OpenSSL 3 versions.

Resolves: https://github.com/openssl/project/issues/1621

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28603)

Fix CI Pipeline by Disabling SSL_TRACE_TEST

Disabling the SSL_TRACE_TEST since it caused an issue on
some cross compiles. A follow-on commit will change
the test.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28621)

Fix another memory order issue

this adds another release/acquire link between update_qp and
get_hold_current_qp via the reader_idx because the current
one which is based on the qp users count is only preventing
a race condition, but does not help when the reader acquires
the next qp.

Fixes #27267

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28585)

Fix ML-KEM key equality check when either unset

Fixes #28563

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28569)

Revert "rsa: expose pairwise consistency test API"

This reverts commit dc5cd6f70a0e "rsa: expose pairwise consistency test API",
that has introduced ossl_rsa_key_pairwise_test() function, as the only user
has been removed in 7f7f75816f26 "import pct: remove import PCTs for most
algorithms".

Complements: 7f7f75816f26 "import pct: remove import PCTs for most algorithms"
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28557)

Fix cipher protocol ID type in docs

The cipher protocol ID, the return type of SSL_CIPHER_get_protocol_id,
is uint16_t and correctly described in docs to be 2 bytes, however the
function signature on the same page incorrectly pointed to it being
uint32_t, which is 4 bytes.

CLA: trivial

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28523)

Updated SSL Trace to display the name for all MLKEM-based groups

Make SSL Trace to display the name of the MLKEM512, MLKEM768,
MLKEM1024 and SecP384r1MLKEM1024 groups.

Fixes #28476

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28499)

Enable peer verification with s_server's verify_return_error option

If the option is specified it will now implicitly enable peer verification.

The s_client command  was already behaved this way, so
s_server was updated to match the behavior of s_client.

Fixes #15134

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28445)

doc: clarify SSL_SESSION_get0_hostname() DESCRIPTION

Also refine RETURN VALUES.

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28393)

Use less preprocessor gymnastics for fatal error detection.

We seem to be using a lot of preprocessor gymnastics to avoid
having duplicate cases in a case statement depending on what
the host system defines these values to.  We should not care.

If we don't bother with the case statement this becomes
easier to follow.

While we are here, pick up the reccomended windows2 values
that correspond with the POSIX values we already have
in here that we believe are "non-fatal", and condition
the codes to use on being windows or something POSIX.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28344)

test/p_ossltest.c: check for return values in OSSL_PARAM_* calls

Some of the OSSL_PARAM_* calls haven't their return codes checked
(OSSL_PARAM_get_octet_string_ptr() call
in ossl_test_aes128cbchmacsha1_set_ctx_params(),
and OSSL_PARAM_set_size_t() call in drbg_ctr_get_ctx_params()),
and Coverity complained about it.  Add the missing checks.

Fixes: 032297054ab5 "Implement an ossltest provider to replace ossltest engine"
Resolves: https://github.com/openssl/project/issues/1618
Resolves: https://scan5.scan.coverity.com/#/project-view/65248/10222?selectedIssue=1665462
Resolves: https://scan5.scan.coverity.com/#/project-view/65248/10222?selectedIssue=1665463
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28583)

providers/implementations/keymgmt/ecx_kmgmt.c.in: remove PCT on ECX import

While PCT on import has been removed for DH, EC, RSA, and SLH-DSA,
ECX seems to be overlooked.  Correct that omission.

Complements: 7f7f75816f26 "import pct: remove import PCTs for most algorithms"
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28573)

crypto/ml_dsa: fix public_from_private() error path to return failure

The error label returned success (1) even on failure. Make it return failure (0) instead.
Fixes #28562

CLA: trivial

Reviewed-by: Todd Short <todd.short@me.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28576)

Don't return a value we never check from indent_printf()

Coverity notices it could overflow, since we don't use this
don't bother returning it

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28553)

Don't clear is_sorted unconditionally on OPENSSL_sk_insert()

If we have a comparison function, and the array was sorted,
check to see if we are inserting in the correct location.
if so do not clear is_sorted.

This allows for element locations found with OPENSSL_sk_find_ex
to be used to insert elements in the correct location and preserve
the sorting order without the need to sort the stack again.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28533)

Fix logic errors in torture_rw_high/low test

"old" was never assigned anything and the first
assignment to "*iterations" in the loop was superfluous.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28574)

Ensure that empty or 1 element stacks are always sorted.

Matt noticed "It's kind of weird that we are forced to call sort on
a newly created and empty stack. It feels like an empty stack should
have the "sorted" flag by default on creation"

I am incluined to agree. This change ensures tht empty or 1 element
stacks are marked as sorted, as per the existing comment in the
file.

Since this involved changing the various duplication routines to
also ensure that sorted was preserved for such stacks, I also
noticed the duplication code was largely duplicated. I
took the opportunity to deduplicate the duplication code while
making these changes.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28509)

25-test_verify.t: fix misleading test case names and file variable name

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28469)

25-test_verify.t: fix partly case-sensitive matching for Windows OS: s/MsWin32/MSWin32/

Fixes #27984

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28098)

Implement Montgomery multiplication assembly optimization for RV64GC

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28012)

Fix ossl_prov_set_macctx

This function fails to construct a param list that includes the passed
in property query string in the param lists when allocating subordonate
algorithms.

Make sure we allow callers to pass a param list (so that providers for
subordonate algorithms can be selected), and merge those into the param
list that this function builds on its own.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28461)

support passing prop querys to composite algs

We have several composite alg usages (i.e. MAC/KDF) which pick the right
digest implementation when using an engine, but fail to get the right
one when using a provider because we don't pass the propquery in a
parameter to their instantiation.

Fix them up by constructing the appropriate parameters

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28461)

remove loader_attic test from test_cmp_cli

This engine is going away (in fact they all are), so just remove the
test cases referencing this engine

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28461)

remove loader_attic tests from test_store

With the removal of engines we need to handle the loader_attic test that
will fail with said removal

based on the advice of @levitte, given that we have a file: loader in
the default provider already, theres no need to test an engine thats
going away, so just remove it.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28461)

Convert TLSProxy to use new provider

replace use of ossltest engine with provider in TLSProxy and update all
dependent tests

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28461)

replace ossltest engine in test_dgst

Use the new ossltest provider rather than the ossltest engine

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28461)

Convert test_rand to use our new p_ossltest provider

Replace ossltest engine with ossltest provider in test_rand

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28461)

Implement an ossltest provider to replace ossltest engine

Part of the effort to remove engines creates a problem for our test
suite, in that we have a large number of tests that rely on the use of a
test engine (ossltest), which implements the aes-128-cbc, aes-128-gcm,
aes-128-cbc-hmac-sha1 ciphers, several digests and a random number
generator to produce predictable outputs for the purposes of doing
testing against known values.

Since we're getting rid of engines, these tests need to be updated to
use a provider that presents the same functionality.

This commit implements that provider.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28461)

Add OSSL_ prefix back onto param names

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28529)

CHANGES.md, NEWS.md: update for 3.6.0-beta1

CHANGES.md:
 * https://github.com/openssl/openssl/pull/28398
 * https://github.com/openssl/openssl/pull/28411
 * https://github.com/openssl/openssl/pull/28447
 * https://github.com/openssl/openssl/pull/28449

NEWS.md:
 * https://github.com/openssl/openssl/pull/28447

Release: yes
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28547)

CHANGES.md, NEWS.md: ffix

Minor formating cleanups.

Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28547)

crypto/ec/ecp_nistz256.c: use OPENSSL_aligned_alloc_array

Allocate table in ecp_nistz256_windowed_mul() and preComputedTable
in ecp_nistz256_mult_precompute() using OPENSSL_aligned_alloc_array() call
instead of OPENSSL_malloc with a 64-byte slack and manual pointer alignment
adjustement.

Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28295)

Explicitly limit the OPENSSL_aligned_alloc()'s alignment to 65536

There is little need to support alignments larger than a page size,
and the open-coded OPENSSL_aligned_alloc() implementation implements
that support in quite wasteful manner, so it is better just to limit
the maximum supported alignment explicitly.  The value of 65536
has been chosen so it is architecture-agnostic and is no less than page sizes
used in commonly occurring architectures (and also it is a pretty number).

Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28295)

crypto/mem.c: use open-coded aligned alloc when posix_memalign fails

While posix_memalign() is generally not expected to fail, we can always use
the internal aligned alloc implementation to ensure that any
OPENSSL_aligned_malloc failure is indeed fatal and does not require
a fallback.

Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28295)

Move the open-coded aligned alloc in a separate file and use it in FIPS module

Factor the open-coded aligned allocation implementation in a separate
file and use it instead of just returning NULL in CRYPTO_aligned_alloc
implementation in the FIPS provider.

Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28295)

Move OPENSSL_SMALL_FOOTPRINT-related logic from aligned_alloc to the only caller

Originally, CRYPTO_aligned_alloc() returned NULL if OpenSSL was built
with OPENSSL_SMALL_FOOTPRINT defined, which is a weird place for such
a consideration;  moreover it means that every caller requires to
implement some form of a fallback (and manually over-allocate
and then align the returned memory if the alignment is a requirement),
which is counter-productive (and outright ridiculous in environments
with posix_memalign() available).  Move the OPENSSL_SMALL_FOOTPRINT
consideration to the only current caller and update the documentation
and tests accordingly.

Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28295)

doc/man3/OPENSSL_malloc.pod: size is no longer needed to be alignment multiple

While the commit 648803a17e4c1 "crypto/mem.c: don't use aligned_alloc
in CRYPTO_aligned_alloc" removed the aligned_alloc() usage
along with its overly onerous requirements, its author failed to update
the documentation accordingly.  Correct that omission by removing
the requirement from the DESCRIPTION and adding a relevant mention
in HISTORY.

Complements: 648803a17e4c1 "crypto/mem.c: don't use aligned_alloc in CRYPTO_aligned_alloc"
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28295)

test/radix/terp.c: avoid accessing uninitialised terp on error

Stats printing in TERP_run() assumes that terp has been initialised,
which is not the case when a jump to the err label has been performed
before successful TERP_init() call;  avoid it by emplacing it
within a have_terp guard check.

Fixes: 4a2d5fe812f8 "QUIC RADIX: Add RADIX test framework implementation"
Resolves: https://scan5.scan.coverity.com/#/project-view/65248/10222?selectedIssue=1665430
References: https://github.com/openssl/project/issues/1432
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28546)

apps/storeutl.c: avoid signed integer overflow in indent_printf()

As two arbitrarily large printf return value can trigger function
overflow, rewrite the return value handling to avoid it.

Fixes: fb43ddceda79 "Add a recursive option to 'openssl storeutl'"
Resolves: https://scan5.scan.coverity.com/#/project-view/65248/10222?selectedIssue=1665428
References: https://github.com/openssl/project/issues/1432
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28546)

crypto/evp/ctrl_params_translate.c: fix a typo in the error message

The ERR_raise_data() call on failure to find ctx->p2 in str_value_map
erroneously refers to ctx->p1 instead;  fix that but supplying the
correct field and casting it to the supposed const char * type.

Fixes: 9a1c4e41e8d3 "EVP: Implement data-driven translation between known ctrl and OSSL_PARAMs"
Resolves: https://scan5.scan.coverity.com/#/project-view/65248/10222?selectedIssue=1665427
References: https://github.com/openssl/project/issues/1432
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28546)

crypto/bio/bss_file.c: add missing cast in format arg in ERR_raise_data()

"%s" conversion specifier requires a "char *" argument, so ptr needs
to be cast to it there, as Coverity has noted.

Fixes: ff988500c2f39 "Replace FUNCerr with ERR_raise_data"
Resolves: https://scan5.scan.coverity.com/#/project-view/65248/10222?selectedIssue=1665423
References: https://github.com/openssl/project/issues/1432
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28546)

test/wpackettest.c: remove ubogus cleanup() in test_WPACKET_quic_vlint_random()

In the beginning of the iteration, pkt is not initialised yet, so there is
no need to clean it up on RAND_bytes() failure.  Replace "return cleanup(&pkt)"
with plain "return 0"'

Fixes: 416d0a638c16 "QUIC wire format support"
Resolves: https://scan5.scan.coverity.com/#/project-view/65248/10222?selectedIssue=1665422
References: https://github.com/openssl/project/issues/1432
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28546)

crypto/x509/t_x509.c: check i2d_X509_NAME() return value in X509_ocspid_print()

There is little reason for this call to fail, but there is also little
reason for not to check for it, and, since Coverity noticed
that the check is missing, just add it.

Resolves: https://scan5.scan.coverity.com/#/project-view/65248/10222?selectedIssue=1665420
References: https://github.com/openssl/project/issues/1432
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28546)

Enable LMS on provider compat fips build for 3.6

The LMS test for fips assumes that LMS is available in the provider in
any version equal to or later than 3.6.

We should probably augment the test such that instead of just checking
the openssl version, we instead query the provider to see if the needed
algs are available to use LMS.

But given the current state of affairs, it seems more sensible to just
enable lms in the 3.6 fips provider build to ensure lms gets tested.

Fixes openssl/project#1435

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/28541)

Test failure of rsa_encrypt when buffer too short

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28517)

Harden RSA public encrypt

Check the that the indicated output buffer length is large enough.

Fix EVP_SealInit() to initialise the output buffer length to the RSA
modulus length, not the input KEK length.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28517)

Update our CI jobs to cover the 3.6 stable branch

The coveralls, prov-compat and provider-compatibiity CI jobs test each
of the stable branches.  We need to add 3.6 to the list in each of those
tests

Fixes openssl/project#1424

Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28519)

kmac_prov.c.in: avoid resource leak on kmac_new_decoder fail in kmac_fetch_new

kctx was not freed in a case of kmac_new_decoder failure;  consolidate
all the error paths under the "err:" label and jump to it on kmac_new_decoder()
returning 0.

Fixes: d5efc853796b "kmac: avoid using ossl_prov_digest_load_from_params()"
Resolves: https://github.com/openssl/project/issues/1419
Resolves: https://scan5.scan.coverity.com/#/project-view/65248/10222?selectedIssue=1453634
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28516)

test/bioprinttest.c: silence more width/precision checks

Apparently, old glibc also can't handle width/precision specifiers;
silence the checks that test for it by setting .skip_libc_check to 1.

Fixes: a29d157fdb6d "Replace homebrewed implementation of *printf*() functions with libc"
Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28530)

remove test_big() the return value we compare varies between
libc implementations. removing this test makes test suite
more rubust.

Fixes a29d157fdb6d

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28524)

Replace homebrewed implementation of *printf*() functions with libc

Switching from ANSI-C we can use implementation of printf like
function provided by libc on target platform. This applies
starting from 3.6 and onwards.

The slight exception here is old windows printf functions
before 2015, those are supported.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28305)

crmf_lib.c create_popo_signature(): add error queue entry on signature failure

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28482)

CHANGES.md: add entry on constifying (X509) function parameter

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28033)

crypto: remove OSSL_FUTURE_CONST as it is not more needed since 4.0

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28033)

constify X509_*print*(), X509_get0_*_objects(), X509_*_get0(), ..., effectively in OpenSSL 4.0

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28033)

constify X509_REQ_get_X509_PUBKEY() in OpenSSL 4.0; related fixes in doc, x509.h.in, x509_req.c

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28033)

rename FFC_PARAM_PTRS to FFC_OSSL_PARAMS

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28360)

dsa/dh: update keymanagers for DH and DSA to use generated param parsers

Both are updates to reduce the OSSL_PARAM_locate related searching.

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28360)

Add invalid parameter error to parser generator

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28360)

dh/dsa: rename files in anticipation of using generated param name parsers

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28360)

OSSL_CALLBACK.pod: add missing info on required return values of callback functions

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28474)

doc: Add missing commas

CLA: trivial
Signed-off-by: Jakub Jelen <jjelen@redhat.com>
Reviewed-by: Paul Yang <paulyang.inf@gmail.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28493)

Doc changes: the -hmac-env and -hmac-stdin options of openssl-dgst
will appear in version 4.0

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28497)

m_sigver.c do_sigver_init(): add details to EVP_R_OPERATION_NOT_SUPPORTED_FOR_THIS_KEYTYPE

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28473)

EVP_PKEY_CTX_get_params(): add error queue entry EVP_R_PROVIDER_GET_CTX_PARAMS_NOT_SUPPORTED

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28473)

Add one more trace message to the torture_rcu_high test

It is interesting that in the very rare cases, where this
test failure has been observed so far, the rcu torture value
went always backwards to 0.  This could be either due to
ossl_rcu_deref(&writer_ptr) returning NULL, or the initial
value of "new = CRYPTO_zalloc(sizeof(uint64_t), NULL, 0)"
still visible despite ossl_rcu_assign_ptr(&writer_ptr, &new)
immediatley after the "*new = global_ctr++" statement.
Add one additional trace message to find out what exactly
happens here, when it happens again.
Additionally, we do no longer initialize the new value to
zero but something else, so it can also be detected.

Related to #27267

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28487)

Add key_type to the derive_skey function

In some cases this information is necessary on the provider side

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28486)