QLOG: Add CHANGES entry

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Remove vestigial abort code

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Fix bit_set

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Minor updates

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Minor fixes after port refactor

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: JSON Encoder: Rename JSON_ENC

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Add debugging guide

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Events: Implement transport:packet_received

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Events: Implement transport:packet_sent

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Wiring: QUIC QTX

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: QUIC MULTISTREAM TEST: Pass script name to QLOG

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Wire title-setting code to QUIC_CHANNEL and SSL_CTX

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: QUIC CHANNEL: Allow a log title to be specified

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Events: Implement recovery:packet_lost

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Wiring: QUIC TXPIM: Record the packet type

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Wiring: QUIC FIFD

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Wiring: QUIC QRX: Report the datagram ID from the DEMUX

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Wiring: QUIC DEMUX: Report a monotonically increasing datagram ID

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Events: Implement connectivity:connection_closed

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Events: Implement transport:parameters_set

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Events: Implement connectivity:connection_state_updated

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Events: Implement connectivity:connection_started

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Wiring: QUIC CHANNEL

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Frontend: Implementation

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Frontend: API Definition

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Frontend: Design

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: JSON Encoder: Tests

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: JSON Encoder: Implementation

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: Configuration

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

QLOG: JSON Encoder: Design

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22037)

libssl: Make some global mutable structures constant

x

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23450)

Add atexit configuration option to using atexit() in libcrypto at build-time.

This fixes an issue with a mix of atexit() usage in DLL and statically linked
libcrypto that came out in the test suite on NonStop, which has slightly
different DLL unload processing semantics compared to Linux. The change
allows a build configuration to select whether to register OPENSSL_cleanup()
with atexit() or not, so avoid situations where atexit() registration causes
SIGSEGV.

INSTALL.md and CHANGES.md have been modified to include and describe this
option.

The no-atexit option has been added to .github/workflows/run-checker-daily.yml.

Fixes: #23135
Signed-of-by: Randall S. Becker <randall.becker@nexbridge.ca>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Todd Short <todd.short@me.com>
(Merged from https://github.com/openssl/openssl/pull/23394)

Remove a CVE reference from CHANGES/NEWS

master/3.2 was never vulnerable to CVE-2023-5678 since it was fixed before
it was released.

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23432)

Add exemplar use case for rcu locks

To demonstrate the use of RCU locks, convert CONF_MOD api to using rcu
rather than RW locks

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22729)

RCU lock implementation

Introduce an RCU lock implementation as an alternative locking mechanism
to openssl.  The api is documented in the ossl_rcu.pod
file

Read side implementaiton is comparable to that of RWLOCKS:
ossl_rcu_read_lock(lock);
<
critical section in which data can be accessed via
ossl_derefrence
>
ossl_rcu_read_unlock(lock);

Write side implementation is:
ossl_rcu_write_lock(lock);
<
critical section in which data can be updated via
ossl_assign_pointer
and stale data can optionally be scheduled for removal
via ossl_rcu_call
>
ossl_rcu_write_unlock(lock);
...
ossl_synchronize_rcu(lock);

ossl_rcu_call fixup

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22729)

Fix typo in CHANGES.md

OSSL_PKEY_PARAM_DERIVE_FROM_PQ must be OSSL_PKEY_PARAM_RSA_DERIVE_FROM_PQ
(note the missing '_RSA').

Signed-off-by: Ingo Franzki <ifranzki@linux.ibm.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23438)

Document the implications of setting engine-based low-level methods

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23063)

Revert "Improved detection of engine-provided private "classic" keys"

This reverts commit 2b74e75331a27fc89cad9c8ea6a26c70019300b5.

The commit was wrong. With 3.x versions the engines must be themselves
responsible for creating their EVP_PKEYs in a way that they are treated
as legacy - either by using the respective set1 calls or by setting
non-default EVP_PKEY_METHOD.

The workaround has caused more problems than it solved.

Fixes #22945

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23063)

Check all frames for stateless reset conditions

In writing the quic stateless reset test we found that the quic rx code
wasn't checking for stateless reest conditions, as the SRT frames were
getting discarded due to failed lcdim lookups.  Move the SRT check above
the lcdim lookup in the rx path to ensure we handle SRT properly in the
client.

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23384)

Add QUIC stateless reset test

QUIC supports the concept of stateless reset, in which a specially
crafted frame is sent to a client informing it that the QUIC state
information is no longer available, and the connection should be closed
immediately.  Test for proper client support here

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23384)

Fix error reporting in EVP_PKEY_{sign,verify,verify_recover}

For some reason, those functions (and the _init functions too) would
raise EVP_R_OPERATION_NOT_SUPPORTED_FOR_THIS_KEYTYPE when the passed
ctx is NULL, and then not check if the provider supplied the function
that would support these libcrypto functions.

This corrects the situation, and has all those libcrypto functions
raise ERR_R_PASS_NULL_PARAMETER if ctx is NULL, and then check for the
corresponding provider supplied, and only when that one is missing,
raise EVP_R_OPERATION_NOT_SUPPORTED_FOR_THIS_KEYTYPE.

Because 0 doesn't mean error for EVP_PKEY_verify(), -1 is returned when
ERR_R_PASSED_NULL_PARAMETER is raised.  This is done consistently for all
affected functions.

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23411)

Fix a possible memleak in bind_afalg

bind_afalg calls afalg_aes_cbc which allocates
cipher_handle->_hidden global object(s)
but if one of them fails due to out of memory,
the function bind_afalg relies on the engine destroy
method to be called.  But that does not happen
because the dynamic engine object is not destroyed
in the usual way in dynamic_load in this case:

If the bind_engine function fails, there will be no
further calls into the shared object.
See ./crypto/engine/eng_dyn.c near the comment:
/* Copy the original ENGINE structure back */

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23409)

Optimize the implementation of ec_field_size()

No need to create and copy BIGNUM p, a and b, just call
EC_GROUP_get0_field() instead.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Yang <kaishen.yy@antfin.com>
(Merged from https://github.com/openssl/openssl/pull/23313)

Fix compilation on Windows using icc

The parameter list for CRYPTO_DOWN_REF for the icc on windows build was
incorrect.

This issue was introduced by 99fd5b2b10

Fixes #23414

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23415)

Extend the testing of resetting/clearing an SSL connection

SSL_clear() explicitly clears an SSL object to enable it to be reused.
You can have a similar effect by calling SSL_set_accept_state() or
SSL_set_connect_state(). We extend the testing of SSL_clear() to use these
other methods. We also ensure we test the case where we have unread
bufferred data that needs to be cleared.

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23256)

Free the QUIC TLS object before freeing the channel

Freeing the QUIC TLS object may make calls back into QUIC so we should
free it first.

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23256)

Free the tserver TLS object before freeing the channel

The TLS object may make callbacks into QUIC during cleanup so we must
free it first.

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23256)

Rationalise RECORD_LAYER_clear() and clear_record_layer()

We had two functions which were very similarly named, that did almost the
same thing, but not quite. We bring the two together. Doing this also fixes
a possible bug where some data may not be correctly freed when the
RECORD_LAYER_clear() version was used.

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23256)

When selecting a method ensure we use the correct client/server version

Using the client one when the server once should be used could cause a
later call to SSL_set_accept_state() to unexpectedly fail.

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23256)

Fix an assertion failure in tls_common.c

When we clear the record layer, we better make sure we clear all relevant
fields, otherwise we can get ourselves into an unexpected state.

Fixes #23255

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23256)

Bump suisei-cn/actions-download-file from 1.4.0 to 1.6.0

Bumps [suisei-cn/actions-download-file](https://github.com/suisei-cn/actions-download-file) from 1.4.0 to 1.6.0.
- [Release notes](https://github.com/suisei-cn/actions-download-file/releases)
- [Commits](https://github.com/suisei-cn/actions-download-file/compare/v1.4.0...v1.6.0)

---
updated-dependencies:
- dependency-name: suisei-cn/actions-download-file
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
CLA: trivial

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23425)

fix missing null check in kdf_test_ctrl

Coverity issue 1453632 noted a missing null check in kdf_test_ctrl
recently.  If a malformed value is passed in from the test file that
does not contain a ':' character, the p variable will be NULL, leading
to a NULL derefence prepare_from_text

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/23398)

Update CHANGES.md and NEWS.md for new release

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Release: yes
(Merged from https://github.com/openssl/openssl/pull/23421)

replace strstr() with strchr() for single characters

strstr() is used to match multiple characters in the haystack,
whereas strchr() is used to matched only single character.

CLA: trivial

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23347)

Have OSSL_PARAM_allocate_from_text() fail on odd number of hex digits

The failure would be caught later on, so this went unnoticed, until someone
tried with just one hex digit, which was simply ignored.

Fixes #23373

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23374)

Add some tests for various PKCS12 files with NULL ContentInfo

PKCS7 ContentInfo fields held within a PKCS12 file can be NULL, even if the
type has been set to a valid value. CVE-2024-0727 is a result of OpenSSL
attempting to dereference the NULL pointer as a result of this.

We add test for various instances of this problem.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23361)

Add NULL checks where ContentInfo data can be NULL

PKCS12 structures contain PKCS7 ContentInfo fields. These fields are
optional and can be NULL even if the "type" is a valid value. OpenSSL
was not properly accounting for this and a NULL dereference can occur
causing a crash.

CVE-2024-0727

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23361)

Add tests for return codes for EVP_CIPHER_CTX_get[block_size|iv_length]

make sure that we get the expected error codes when we do bad things,
rather than a crash

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22995)

Add appropriate NULL checks in EVP_CIPHER api

The EVP_CIPHER api currently assumes that calls made into several APIs
have already initalized the cipher in a given context via a call to
EVP_CipherInit[_ex[2]].  If that hasnt been done, instead of an error,
the result is typically a SIGSEGV.

Correct that by adding missing NULL checks in the apropriate apis prior
to using ctx->cipher

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22995)

Fix a possible memory leak in req_main

if the private key is output to stdout using the HARNESS_OSSL_PREFIX,
out is a stack of BIOs and must therefore free'd using BIO_free_all.

Steps to reproduce:

$ HARNESS_OSSL_PREFIX=x OPENSSL_CONF=apps/openssl.cnf util/shlib_wrap.sh apps/openssl req -new -keyout - -passout pass: </dev/null
[...]
Direct leak of 128 byte(s) in 1 object(s) allocated from:
    #0 0x7f6f692b89cf in __interceptor_malloc ../../../../src/libsanitizer/asan/asan_malloc_linux.cpp:69
    #1 0x7f6f686eda00 in CRYPTO_malloc crypto/mem.c:202
    #2 0x7f6f686edba0 in CRYPTO_zalloc crypto/mem.c:222
    #3 0x7f6f68471bdf in BIO_new_ex crypto/bio/bio_lib.c:83
    #4 0x7f6f68491a8f in BIO_new_fp crypto/bio/bss_file.c:95
    #5 0x555c5f58b378 in dup_bio_out apps/lib/apps.c:3014
    #6 0x555c5f58f9ac in bio_open_default_ apps/lib/apps.c:3175
    #7 0x555c5f58f9ac in bio_open_default apps/lib/apps.c:3203
    #8 0x555c5f528537 in req_main apps/req.c:683
    #9 0x555c5f50e315 in do_cmd apps/openssl.c:426
    #10 0x555c5f4c5575 in main apps/openssl.c:307
    #11 0x7f6f680461c9 in __libc_start_call_main ../sysdeps/nptl/libc_start_call_main.h:58

SUMMARY: AddressSanitizer: 128 byte(s) leaked in 1 allocation(s).

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23365)

ci.yml: Replace actions-rs/toolchain@v1 with dtolnay/rust-toolchain

actions-rs/toolchain is unmaintained and generates warnings

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23232)

tlsfuzzer.sh: Use python3

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/23342)

tlsfuzzer.sh: Make it more informative on errors

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/23342)

tlsfuzzer.sh: Run openssl version on the built app and not system one

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/23342)

gost_engine.sh: Set OPENSSL_ENGINES_DIR

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
(Merged from https://github.com/openssl/openssl/pull/23342)

Fix typos

CLA: trivial

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23345)

Add CHANGES entry

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23343)

Fix docs

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23343)

QUIC MULTISTREAM TEST: Test optimised FIN API

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23343)

QUIC APL: Implement optimised FIN API

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23343)

QUIC: Add optimised FIN API

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23343)

Remove mios-simulator-version-min from default iOS configuration

* iOS 6 and 7 got its last updates 2014.
* Adding -mios-simulator-version-min=14.0 to the configure command leads to two occurences of this parameter during compilation.
* So lets remove this values to allow a robust configuration from outside and avoid updates here in the future.

CLA: trivial

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23344)

02-test_errstr.t: make robust on `openssl errstr` crashes

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23330)

Removes unsed function: ssl_bad_method()

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23307)

Uninitialized array variable

array"key" is uninitialized and it is being read directly in function SipHash_Init() as per the below statements making a way for the garbage values :
uint64_t k0 = U8TO64_LE(k);
uint64_t k1 = U8TO64_LE(k + 8);

CLA: trivial

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23298)

prevent integer overflow in ossl_asn1_time_from_tm

this could be triggered by the following code (assuming 64 bit time_t):

time_t t = 67768011791126057ULL;
ASN1_TIME* at = ASN1_TIME_set(NULL, t);

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22976)

add test for provoking integer overflow in ossl_asn1_time_from_tm

this needs a sanitized 64 bit time_t build to be detected (or possibly
valgrind, trapv or similar)

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22976)

Don't apply max_frag_len checking if no Max Fragment Length extension

Don't check the Max Fragment Length if the it hasn't been negotiated. We
were checking it anyway, and using the default value
(SSL3_RT_MAX_PLAIN_LENGTH). This works in most cases but KTLS can cause the
record length to actually exceed this in some cases.

Fixes #23169

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23182)

Fix a FreeBSD build failure when KTLS is enabled

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23182)

Add a KTLS test where we write long app data records

Check that we can write and read back long app data records when using
KTLS.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23182)

X509_dup.pod: add caveat that extra data is not copied and hints, e.g., to use X509_up_ref() instead

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23043)

Move increment of dtls epoch to change cipher state function

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23212)

Document SSL_R_UNEXPECTED_EOF_WHILE_READING

Also document that it is ok to use this for control flow decisions.

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23304)

Add a deprecation warning for a function in docs

The function in question is SSL_get_peer_certificate()

CLA: trivial

Reviewed-by: Tom Cosgrove <tom.cosgrove@arm.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23315)

Fix reason value collision for CMP_R_UNEXPECTED_SENDER

Reviewed-by: Hugo Landau <hlandau@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23323)

Introduce hash thunking functions to do proper casting

ubsan on clang17 has started warning about the following undefined
behavior:

crypto/lhash/lhash.c:299:12: runtime error: call to function err_string_data_hash through pointer to incorrect function type 'unsigned long (*)(const void *)'
[...]/crypto/err/err.c:184: note: err_string_data_hash defined here
    #0 0x7fa569e3a434 in getrn [...]/crypto/lhash/lhash.c:299:12
    #1 0x7fa569e39a46 in OPENSSL_LH_insert [...]/crypto/lhash/lhash.c:119:10
    #2 0x7fa569d866ee in err_load_strings [...]/crypto/err/err.c:280:15
[...]

The issue occurs because, the generic hash functions (OPENSSL_LH_*) will
occasionaly call back to the type specific registered functions for hash
generation/comparison/free/etc, using functions of the (example)
prototype:

[return value] <hash|cmp|free> (void *, [void *], ...)

While the functions implementing hash|cmp|free|etc are defined as
[return value] <fnname> (TYPE *, [TYPE *], ...)

The compiler, not knowing the type signature of the function pointed to
by the implementation, performs no type conversion on the function
arguments

While the C language specification allows for pointers to data of one
type to be converted to pointers of another type, it does not
allow for pointers to functions with one signature to be called
while pointing to functions of another signature.  Compilers often allow
this behavior, but strictly speaking it results in undefined behavior

As such, ubsan warns us about this issue

This is an potential fix for the issue, implemented using, in effect,
thunking macros.  For each hash type, an additional set of wrapper
funtions is created (currently for compare and hash, but more will be
added for free/doall/etc).  The corresponding thunking macros for each
type cases the actuall corresponding callback to a function pointer of
the proper type, and then calls that with the parameters appropriately
cast, avoiding the ubsan warning

This approach is adventageous as it maintains a level of type safety,
but comes at the cost of having to implement several additional
functions per hash table type.

Related to #22896

Reviewed-by: Sasa Nedvedicky <sashan@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23192)

cmp_vfy.c/check_msg_all_certs(): remove needless trace output in case 3GPP mode is not enabled

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/19948)

ossl_cmp_msg_check_update(): improve diagnostics of checking expected sender name

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/19948)

80-test_cmp_http_data/: add tests for -no_cache_extracerts, fix and extend further test cases

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/19948)

CMP app and doc: add -no_cache_extracerts option / OSSL_CMP_OPT_NO_CACHE_EXTRACERTS

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/19948)

CMP app: make -ignore_keyusage apply also for mock server

Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: David von Oheimb <david.von.oheimb@siemens.com>
(Merged from https://github.com/openssl/openssl/pull/19948)

Fix performance regression of ChaCha20 on LoongArch64

The regression was introduced in PR #22817.

In that pull request, the input length check was moved forward,
but the related ori instruction was missing, and it will cause
input of any length down to the much slower scalar implementation.

Fixes #23300

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/23301)

hurd: Fix dgram_sendmmsg

GNU/Hurd does not have IP_PKTINFO yet, thus SUPPORT_LOCAL_ADDR is undef,
data->local_addr_enabled never set to 1, and thus the M_METHOD_RECVMSG
method would end up raising BIO_R_LOCAL_ADDR_NOT_AVAILABLE immediately.

Fixes #22872

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Hugo Landau <hlandau@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23293)

Add missing genpkey -rand support

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/23294)

Check whether the pubkey exists in ossl_ecx_key_dup

Signed-off-by: lan1120 <lanming@huawei.com>
Reviewed-by: Shane Lontis <shane.lontis@oracle.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/22984)

Remove receiving of unused return value

CLA: trivial

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23277)

Move discovery of the legacy alg type into the keymgmt

During creation of the EVP_PKEY_CTX we were trying to discover what legacy
alg it corresponds to every time which was slow. Instead we move this into
the construction of the EVP_KEYMGMT.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23265)

Error in s_server when -rev option is used with dtls.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23278)

Add X509_STORE_get1_objects

X509_STORE_get0_objects returns a pointer to the X509_STORE's storage,
but this function is a bit deceptive. It is practically unusable in a
multi-threaded program. See, for example, RUSTSEC-2023-0072, a security
vulnerability caused by this OpenSSL API.

One might think that, if no other threads are mutating the X509_STORE,
it is safe to read the resulting list. However, the documention does not
mention that other logically-const operations on the X509_STORE, notably
certifcate verifications when a hash_dir is installed, will, under a
lock, write to the X509_STORE. The X509_STORE also internally re-sorts
the list on the first query.

If the caller knows to call X509_STORE_lock and X509_STORE_unlock, it
can work around this. But this is not obvious, and the documentation
does not discuss how X509_STORE_lock is very rarely safe to use. E.g.
one cannot call any APIs like X509_STORE_add_cert or
X509_STORE_CTX_get1_issuer while holding the lock because those
functions internally expect to take the lock. (X509_STORE_lock is
another such API which is not safe to export as public API.)

Rather than leave all this to the caller to figure out, the API should
have returned a shallow copy of the list, refcounting the values. Then
it could be internally locked and the caller can freely inspect the
result without synchronization with the X509_STORE.

Reviewed-by: Neil Horman <nhorman@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23224)

Removed logically dead code from function i2r_issuer_sign_tool

Since new_line is assigned with 0 in the very begging of the function check added at line no. 106 will never become true. Hence removing.

CLA: trivial

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Neil Horman <nhorman@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/23284)