Fixes for NonStop builds on 3.5 to handle OS platform header file changes.

This changes handles the introduction of _POSIX_VERSION into the NonStop x86
header files that tricks OpenSSL into thinking that ucontext.h is available.

Fixes #28498

Signed-off-by: Randall S. Becker <randall.becker@nexbridge.ca>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28769)

(cherry picked from commit fb2f239ca1bd05f609e9e1e06b4b84f7f628d21b)

Serial submission policy

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28897)

bss_dgram.c: Fix potential buffer overread and remove asserts

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28896)

asn1: fix ret type in ASN1_item_d2i_fp_ex

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/28884)

SMIME_text() and SMIME_crlf_copy() small refactoring

When out is NULL, SMIME_text() now skips BIO_write entirely and continues
to return success based on the read loop result. This matches existing
and expected semantics while avoiding a pointless write and any error masking.

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28879)

ssl: set tmp.pkey only after successful derive

Assign s->s3.tmp.pkey after ssl_derive succeeds and free skey on failure
to avoid a dangling state.

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28878)

ssl: use BAD_KEY_SHARE for invalid key share encoding

Replace BAD_ECPOINT with BAD_KEY_SHARE in tls_accept_ksgroup so alerts
are correct for non EC groups too.

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28876)

bio_sock: fix inverted BIO_lookup return value check

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28862)

freebsd ktls: avoid unaligned 16 bit length store in ktls_read_record

This prevents SIGBUS on strict alignment architectures when p+3 is not aligned
for 16 bit access. Behavior is unchanged on x86 and matches the Linux path.

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28860)

quic: use PACKET_buf_init instead of writing to PACKET internals

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28850)

quic: avoid partial update in ossl_quic_bind_channel

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28848)

cmp: guard NULL header sender when expected_sender is set

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28847)

Do not allow zstd 0.9

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28843)

Correct error code in raise in bio_zstd/brotli_flush()

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28839)

ssl/statem: write 16-bit ECDHE group id in SKE

tls_construct_server_key_exchange wrote the named group as two u8 bytes with the high byte set to 0. TLS requires a 16-bit NamedGroup. Use WPACKET_put_bytes_u16(curve_id) so ids >= 256 are encoded correctly. No change for groups < 256.

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28835)

quic: small cleanup, write is_retry with u8

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28828)

keep_alive(): Do not shadow own function name

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28826)

Resetting the cookie_len should set the cookie_len to 0.

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28817)

Fix wrong buffer and version in outgoing msg callback

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28816)

apps/s_socket: fix double close of listening socket when naccept==0

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28813)

Fix off-by-one overflow in the AST resubmit

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28811)

s_server: print program not, not port

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28807)

Disallow Message sequence number 2 in DTLSv1_listen

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Frederik Wedel-Heinen <fwh.openssl@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28788)

ssl: fix OOB write in SSL_get_shared_ciphers when no shared ciphers

When no cipher names are appended, p remains at buf and the unconditional
p[-1] = '\0' underflows. Only NUL-terminate if at least one cipher was written;
otherwise return an empty string safely.

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28785)

CHANGES.md, NEWS.md: update for 3.6.0

 * Add the release date for 3.5.4
 * Various touch-ups aimed at improving consistency of the news
 * ffixes, wfixes

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/28734)

CHANGES.md, NEWS.md: various ffixes

 * Adding missing periods.
 * Consistently using "OpenSSL" as the project's name.
 * Fixing the "`EVP_SKEY`" formatting reverted by a merge resolution
   in f815ee19e066 "Harden RSA public encrypt".
 * Fitting the lines in 80 characters, do a bit more semantic line breaks.

Signed-off-by: Eugene Syromiatnikov <esyr@openssl.org>
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/28734)

CHANGES.md, NEWS.md: update for 3.5.4

3.5.4 CHANGES.md includes the following:
 * https://github.com/openssl/openssl/pull/28415
 * https://github.com/openssl/openssl/pull/28573
 * https://github.com/openssl/openssl/pull/28603

3.5.4 NEWS.md includes the following:
 * https://github.com/openssl/openssl/pull/28603

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/28734)

CMS: Produce error when AEAD algorithms are used in enveloped data

Fixes GH-21414

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/27772)

APPS/cmp: fix wrong CLI option reference for -ref_cert

Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28786)

APPS/ocsp: avoid memory leaks on error

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28823)

apps/cmp.c: must not try acting as server if -reqout_only option is given

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28891)

apps/cmp.c and doc/man3: add clarifying comments and doc line on CMP context used in two function calls; move one of these calls

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28891)

Correct assembler version checking in perlasm scripts

Fixes #20520

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28901)

Add file and line number to test failure messages

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Allow us to pass NULL to have the current time, since CMP appears to be determined to do so

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Catch the failing conversions due to limited time_t on NotBefore and NotAfter as well

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Windows is having issues, print the time it doesn't like when it fails

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Add unit test for X509 temporal validity functions.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Also be pedantically RFC5280 compliant in ossl_x509_check_certificate_times.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Bring in boundary test from #28584

Will add further unit tests for the cert validity check routine

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Be fully RFC 5280 pedantic about notAfter

May as well catch this case too. Any such certificate which
is valid for the notBefore at the time in quesion should be
considered valid no mater what the time is even if the system
time has been set to something beyond the range of ASN1_TIME values.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Put an appropriate warning in the apps code for now

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Stop using X509_cmp_timeframe in cert_acceptible

Again, so we don't accept invalid times as valid forever.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Stop using X509_cmp_timeframe in ossl_X509_check

To no longer accept invalid certificate times as valid forver.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

separate time validation and comparison

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Fix the documentation for X509_cmp_timeframe

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28623)

Deprecate X509_STORE_get0_objects()

Resolves https://github.com/openssl/project/issues/1369
Signed-off-by: Nikola Pajkovsky <nikolap@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28599)

x509store: reduce lock contention in X509_STORE

X509_STORE was using STACK_OF(X509_OBJECT) which is not ideal structure. The
better solution is to use hashmap. The performance gains come from the fact that
sorting was removed and therefore read lock is just enough for looking up
objects/cert/crls from hashmap.

When X509_STORE_get0_objects() is called, the hashmap converts back to
the STACK_OF(X509_OBJECT), and goes back to the original
implementation with the performance hit on lookup side because stack is not
sorted anymore.

Note, hashmap maps X509_NAME to STACK_OF(X509_OBJECT), and the stack is never
sorted which may lead to performance impact if stack contains a huge of objects.

Before the change

| Threads |   mean/us |  var/us |
|---------+-----------+---------|
|       1 |  2.434803 | .034190 |
|       2 |  3.033588 | .247471 |
|       4 |  6.551132 | .150209 |
|       6 | 12.548113 | .258445 |
|       8 | 17.566257 | .168508 |
|      10 | 22.782846 | .182674 |
|      12 | 27.928990 | .426779 |
|      14 | 32.844572 | .307754 |
|      16 | 37.816247 | .660630 |
|      18 | 42.662465 | .434926 |

After the change

| Threads |  mean/us |  var/us |
|---------+----------+---------|
|       1 | 2.385398 | .015329 |
|       2 | 2.775794 | .172223 |
|       4 | 3.071882 | .126400 |
|       6 | 3.174147 | .139685 |
|       8 | 3.479235 | .297154 |
|      10 | 4.206260 | .149006 |
|      12 | 5.044039 | .194108 |
|      14 | 5.890640 | .185817 |
|      16 | 6.447808 | .256179 |
|      18 | 7.489261 | .149204 |

Resolves: https://github.com/openssl/project/issues/1275
Signed-off-by: Nikola Pajkovsky <nikolap@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28599)

Move handy test functions to testutil

Signed-off-by: Nikola Pajkovsky <nikolap@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28599)

http_client.c: make sure to raise error 404 (also in case of further errors like content type mismatch)

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/28895)

quic: pass is_write to quic_classify_stream in quic_get_stream_error_code

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28820)

Update size check condition for probable primes to align with FIPS 186-5.

Fixes #28540

In FIPS 186-4 (Table B.1), the upper bounds for probable primes p and q were
defined using strict inequalities. In FIPS 186-5, Table A.1 replaces the
previous Table B.1 (dropping the nlen=1024 case) and revises the upper bound
condition to use weak inequalities.

CLA: trivial

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28540)

ssl: call SSLfatal on realloc failures in extract_keyshares

Initial malloc path already does this. Realloc path went to failure
without recording a fatal alert.

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28873)

AEAD: Raise an error on EVP_DecryptFinal_ex() without the tag being set

In AEAD cipher providers raise an error when EVP_DecryptFinal_ex()
is called without the authentication tag being set.

Fixes #28730

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28872)

Clarify that setpgpid() is optional if not provided by the platform

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28822)

s_socket: continue in loop, on bind fail

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28814)

bio/dgram: use IPV6_PMTUDISC_* with IPV6_MTU_DISCOVER

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28809)

bio/dgram: fix local addr clear for i-th message

Signed-off-by: Joshua Rogers <MegaManSec@users.noreply.github.com>
Fixes #28871

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28808)

asn1_time_test.c: More asn1 to utc testcases

Test out of bounds month, day of month, hour, minute and seconds.
Ensure we check the leap year boundaries.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28789)

Add no-argon2 option to daily build.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/28756)

evp_test.c: Switch the logic to skip ARGON2 EVP tests to PREFIX

The names have ARGON2 as PREFIX rather than SUFFIX.
This covers the argon algorithms ARGON2D, ARGON2I, and ARGON2ID.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/28756)

Update CI macos runners

Github have notified that the macos 13 runner image is deprecated and
will be retired on the 4th December 2025. It will also fail temporarily
intermittently during November on specific days as a warning.

Notes:
 - The macos-14 and macos-15 labels correspond to arm64, and
   macos-14-large and macos-15-large correspond to x86_64 (intel).

 - macos x86_64 intel will no longer be supported after
   the macos 15 runner image is retired in the Fall of 2027.
   For now we should continue to support this.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28737)

Fix an error check for EVP_PBE_CipherInit_ex()

As pointed out during the review for the recent security issue in this
file, this error check is incorrect: as documented, this function has
the usual Boolean return values.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28727)

ML-DSA: Validate signature length before decoding

Page 27 of https://nvlpubs.nist.gov/nistpubs/FIPS/NIST.FIPS.204.pdf mentions
the Input: Signature length depending on the parameters, so the signature
length should be checked before we proceed with decode.

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/28093)

ktls: move ktls_enable() within ktls_start()

On linux ktls can only be enabled on established TCP sockets.
When SSL_set_fd() is called before the connection is established
ktls_enable() fails and ktls is not setup.

This moves ktls_enable() call within then ktls_start() function.
Multiple calls to ktls_start() will trigger additional ktls_enable()
calls which fail with EEXIST, but do not affect the ktls socket.

CLA: trivial

Signed-off-by: Ulrich Weber <ulrich.weber@gmail.com>
Reviewed-by: Paul Yang <paulyang.inf@gmail.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/27908)

Make eyeballs happy again for ossl-nghttp3-demo

Adjust the http3 demo client so it works better on dual stack hosts.  This
fixes the case when DNS returns both IPv4 and IPv6 addresses for host we try to
reach. The current code just uses the first address found in DNS answer. If
things are unfortunate and the service (port number) demo client tries to reach
does not listen on the address then demo  gives up and exits.

Demo can do better. The RFC 6555 suggests application should try to reach the
service on the next address returned by DNS, when the first attempt fails for
the first address returned by DNS. This change helps with situation when DNS
prefers, let' say, IPv6 address, but the service is reachable via IPv4 only.
In that case application sees the failure on the first attempt to connect to
remote server over IPv6, but the second attempt that uses IPv4 is  going to
succeed.

This extra handling is required for QUIC which uses UDP protocol. For TLS
client which uses TCP all this happens inside BIO layer which tries to
establish TCP connection. There is no TCP-handshake on UDP protocol so
BIO can not see the service is not reachable on requested address.

Fixes: #28331
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28802)

bioprinttest.c: return 1 on test skip

Signed-off-by: Lars Erik Wik <lars.erik.wik@northern.tech>
CLA: trivial

Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/28771)

- remove CYGWWIN (requested by Brandt)

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28759)

- fix suggested by Bernd

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28759)

OPENSSL_SYS_WINDOWS is also enabled for mingw build

the test_n() in bioprinttest.c must differentiate between
MSVC runtime libc and ming runtime libc. The function
_set_printf_count_output() must be called when openssl
is linked with MSVC libc only.

Fixes #28679

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28759)

Check return code of BIO_ADDR_rawaddress

Fixes coverity issue: https://scan5.scan.coverity.com/#/project-view/64471/10222?selectedIssue=1666584
Thank you for the report!

Signed-off-by: Norbert Pocs <norbertp@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/28751)

[aarch64] move constants to rodata

The 3.6.0 has new assembly which again has constants in .text. This
breaks on platforms enforcing execute-only memory.

See, e.g., #24137 and PRs linked therein, among others.

Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/28726)

Fix riscv64 carry bug in SM2 modulo reduction

Fixes #28731

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28746)

Reduce Max number of commands handled per fuzzer pass in quic-lcidm.

We've gotten a few recent reports of a hang in the quic-lcidm fuzzer:

https://issues.oss-fuzz.com/issues/448510502

It looks pretty straightforward (I think).  The fuzzer input buffer is
used in this particular case to randomly issue commands to the lcidm
hash table (add/delete/query/flush/etc).

The loop for the command processing (based on the input buffer), is
limited to 10k commands.  However the fuzzer will on occasion provide
very large buffers (500k) which easily saturate that limit.  If the
input buffer happens to do something like get biased toward mostly
additions, we wind up with a huge hashtable that has to constantly grow
and rehash, which we've seen leads to timeouts in the past.

Most direct fix I think here, given that this is something of an
artificial failure in the fuzzer, is to simply clamp the command limit
more.

Fixes openssl/project#1664

Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28724)

Add SSL_get_peer_addr() function to query peer address for QUIC

This change introduces a new public API symbol: SSL_get_peer_addr().
The change is QUIC-only, there are no changes for TLS connections

- API: add peer address query for QUIC connections
  * Internal: declare/implement ossl_quic_get_peer_addr(SSL*, BIO_ADDR*)
  * Public: declare/implement SSL_get_peer_addr(SSL*, BIO_ADDR*)

Rationale:
- Allow applications to retrieve the remote UDP tuple for QUIC sessions
  (e.g., logging, access control, diagnostics)

Provided documentation and test cases for SSL_get_peer_addr().

Set peer via channel API on new-conn.

- In ch_on_new_conn_common(), BIO_ADDR_copy(&ch->cur_peer_addr, peer)
  was replaced with ossl_quic_channel_set_peer_addr(ch, peer) so
  addressed_mode is enabled at connection bring-up.

Dropped redundant peer detection in create_qc_from_incoming_conn()

The peer address is now propagated in ch_on_new_conn_common() via
ossl_quic_channel_set_peer_addr(), so the channel is already in
"addressed" mode. This also avoids querying the (unconnected) server
UDP BIO, reduces duplication, and simplifies the accept path. All
regression tests pass.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28690)

tests: temporarily skip MLDSA pkcs11-provider test

External pkcs11-provider test requires at least kryoptic 1.2 for
MLDSA tests. But the current fedora:latest (42) still contains
kryoptic 1.1 and hence we need to temporarily disable MLDSA tests
until Fedora 43 is released.

Signed-off-by: Ondrej Moris <omoris@redhat.com>
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/28716)

Update pkcs11-provider submodule (663dea3)

Signed-off-by: Ondrej Moris <omoris@redhat.com>
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/28716)

add codespell pre-commit check and fix found misspellings

Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28436)

add config file for pre-commit hooks framework

pre-commit helps managing and maintaining multi-language pre-commit hooks.
This commit adds a pre-commit configuration to run a certian version of
clang-format utility. Later we can add sections for other languages as
well. pre-commit developers also provide the CI system which uses the
same config file.

https://pre-commit.com/
https://pre-commit.ci/

Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28436)

hashtable: add option to disable RCU locks

a new config option _no_rcu_ is added into HT_CONFIG. When _no_rcu_ is
set then hashtable can be guarded with any other locking primitives,
and behives as ordinary hashtable. Also, all the impact of the
atomics used internally to the hash table was mitigated.

RCU performance

   # INFO:  @ test/lhash_test.c:747
   # multithread stress runs 40000 ops in 40.779656 seconds

No RCU, guarded with RWLOCK

   # INFO:  @ test/lhash_test.c:747
   # multithread stress runs 40000 ops in 36.976926 seconds

Signed-off-by: Nikola Pajkovsky <nikolap@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28677)

hashtable: pass HT into hash function

When defining a custom hash function for a hashtable key, you typically start with:

  HT_START_KEY_DEFN(key)
  HT_DEF_KEY_FIELD(k, unsigned char *)
  HT_END_KEY_DEFN(KEY)

In this setup, the hash function signature requires keybuf and len as
parameters rather than the hashtable key itself. As a result,
accessing members of the hashtable structure becomes awkward, since
you must do something like:

  #define FROM_KEYBUF_TO_HT_KEY(keybuf, type) (type)((keybuf) - sizeof(HT_KEY))

  static uint64_t ht_hash(uint8_t *keybuf, size_t keylen)
  {
      KEY *k = FROM_KEYBUF_TO_HT_KEY(keybuf, KEY *);
      ...
  }

This kind of pointer arithmetic is both unnecessary and error-prone.
A cleaner approach is to pass the HT pointer directly into the hash
function. From there, you can safely cast it to the required type
without the pointer gymnastics.

Signed-off-by: Nikola Pajkovsky <nikolap@openssl.org>
Reviewed-by: Saša Nedvědický <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28677)

Fix EVP_DecryptFinal_ex() for ChaCha20-Poly1305.

When using the ChaCha20-Poly1305 algorithm, the final interface
returns success without setting the authentication tag, whereas
the AES-GCM algorithm correctly returns failure in such cases.

Fixes #28137

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28683)

ghash-riscv64-zvkg.pl: Code Comment Correction

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28688)

ts_conf: fix memory leak in TS_CONF_set_policies

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28686)

Fix riscv64 chacha crash due to unaligned data

The linux-riscv64 test machine crashes due to unaligned data,
when the V extension is enabled, while QEMU seems to have no
problems with unaligned data.

So check for aligned data and fall back to C code in case the
input or output values are unaligned.

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28684)

Enable the V extension in the OS-Zoo CI job

and also fix the unintentional omission of the ZBA extension,
since the first word in the OPENSSL_riscvcap environment variable
is ignored, because it is assumed to be the processor base
architecture, e.g. something like RV64GC.

Reviewed-by: Paul Dale <ppzgs1@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28684)

Fix Memory leak in CMS_EncryptedData_set1_key

When CMS_EncryptedData_set1_key is called repeatedly it will
leak data on the second call. This was because
cms->d.encryptedData was already set and needed to be cleared
before the call to M_ASN1_new_of.

Fixes: #28606
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28668)

aes: convert AES ciphers to use generated parameter decoders

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

aes: rename files in anticipation of gerenated param decoding

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

chacha20: convert to using genreated param name decoders

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

chacha20: rename files in anticipation of generated param decoding

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

xts: convert to generated param name decodering

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

ctx: rename files for conversion to generated param decoders

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

legacy ciphers: use generated param name decoders

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

legacy ciphers: rename files in anticipation of using generated param decoders

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

aes_cbc_hmac_sha: convert to generated param decoding

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

aes_cbc_hmac: rename files in anticipation of generated param decoding

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

mdc2: use a generated param decoder

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

mdc2: rename files in anticipation of generated param decoding

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/28616)

Add CHANGES.md and NEWS.md updates

Including a few corrections of the previous entries.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>