Fix copyright year issues

Fixes: #13765
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17427)

OSSL_STORE: Prevent spurious error during loading private keys

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15283)

Fix typos

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17392)

Run TLSfuzzer tests for CI

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17340)

TLS Fuzzer: initial test infrastructure

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17340)

TLSfuzzer: submodules

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17340)

check the return value of OSSL_PARAM_BLD_new in dsa_kmgmt.c:195

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17155)

sm2: fix {i2d,d2i}_PublicKey EC_KEY is EVP_PKEY_SM2

CLA: trivial

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17389)

apps/passwd.c: free before error exiting

use goto instead of returning directly while error handling

Signed-off-by: Peiwei Hu <jlu.hpw@foxmail.com>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17404)

fix building failure when using -Wconditional-uninitialized

Use clang -Wconditional-uninitialized to build, the error "initialize
the variable 'buffer_size' to silence this warning" will be reported.

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17375)

trace.c: Add missing trace category entry

Fixes #17397

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17399)

CMP mock server: add -ref_cert option and corresponding ossl_cmp_mock_srv_set1_refCert()

Fixes #16041

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16050)

X509_cmp.pod: Point out that the X509_NAME_cmp() arguments may be NULL

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16050)

app_http_tls_cb: Fix double-free in case TLS not used

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17400)

check-format.pl: Fix report on constant on LHS of comparison or assignment

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17396)

HTTP client: Work around HTTPS proxy use bug due to callback design flaw

See discussion in #17088, where the real solution was postponed to 4.0.

This preliminarily fixes the issue that the HTTP(S) proxy environment vars
were neglected when determining whether a proxy should be used for HTTPS.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17310)

Fix compile error when building with no-asm

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17376)

check the return value of EVP_MD_fetch in ecdh_exch.c:285 & dh_exch.c:347

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17153)

Use USE_SWAPCONTEXT on IA64.

On IA64 the use of setjmp()/ longjmp() does not properly save the
state of the register stack engine (RSE) and requires extra care.
The use of it in the async interface led to a failure in the
test_async.t test since its introduction in 1.1.0 series.

Instead of properly adding the needed assembly bits here use the
swapcontext() function which properly saves the whole context.

Signed-off-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17370)

Validate the category in OSSL_trace_end()

OSSL_trace_end() should validate that the category it has been passed
by the caler is valid, and return immediately if not.

Fixes #17353

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17371)

Update troublesome copyright years of auto-generated files to 2022

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17398)

X509V3_set_ctx(): Improve documentation

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17358)

ec.h: Explain use of strstr() for EVP_EC_gen() and add #include <string.h>

Fixes #17362

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17380)

Check the return value of ossl_bio_new_from_core_bio()

There are missing checks of its return value in 8 different spots.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17154)

close_console: Always unlock as the lock is always held

Fixes #17364

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17365)

try_pkcs12(): cleanse passphrase so it is not left on the stack

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/17320)

try_pkcs12(): Correct handling of NUL termination of passphrases

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/17320)

Test that PEM_BUFSIZE is passed into pem_password_cb

When pem_password_cb is used from SSL_CTX, its size
parameter should be equal to PEM_BUFSIZE.

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/17320)

pem_password_cb: Clarify the documentation on passphrases

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/17320)

Compensate for UI method always adding NUL termination

The UI method always adds NUL termination and we need to
compensate for that when using it from a pem_password_cb
because the buffer used in pem_password_cb does not account
for that and the returned password should be able fill the
whole buffer.

Fixes #16601

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/17320)

test: add some unit tests for the property to string functions

That is: ossl_property_name_str and ossl_property_value_str.

These only have high level tests during the creation of child library
contexts.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17325)

property: use a stack to efficiently convert index to string

The existing code does this conversion by searching the hash table for the
appropriate index which is slow and expensive.

Fixes #15867

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17325)

Fix the symbol_presence test with a shlib_variant

If a shlib_variant is used then the dynamic version information for
symbols will be different from what the symbol presence test was
expecting. We just make it more liberal about what it accepts as dynamic
version information.

Fixes #17366

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17372)

APPS/cmp: improve diagnostics for presence of TLS options

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16747)

OSSL_CMP_CTX: rename get/set function for trustedStore

This makes the naming more consistent, in a backward-compatible way

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17277)

Ensure s_client sends SNI data when used with -proxy

The use of -proxy prevented s_client from correctly sending the target
hostname as SNI data.

Fixes #17232

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17248)

Fix a misuse of NULL check

Fixes: #17356
CLA: trivial

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17357)

document additional stack push error code

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17350)

(cherry picked from commit 0caf2813245f7141b982fcfd9bda402117da115c)

improving tests for adding sigalg with empty digest

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17315)

Add static check in BN_hex2bn

Fixes #17298

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17299)

Fix Configure variable spill

* Evaluating code-refs in Configure can sometimes set the default
variable `$_`
* Prevent spillage influencing the target property by using named
variable in loop

CLA: trivial

Fixes gh-17321

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17322)

namemap: handle a NULL return when looking for a non-legacy cipher/MD

Fixes #17313

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17324)

HTTP client: Fix cleanup of TLS BIO via 'bio_update_fn' callback function

Make app_http_tls_cb() tidy up on disconnect the SSL BIO it pushes on connect.
Make OSSL_HTTP_close() respect this.

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17318)

Add support for BSD-riscv64 target

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17306)

HTTP client: Work around the 'gets' method not being supported by SSL BIOs

It turned out that loading non-ASN.1 contents using the HTTP client
fails over TLS because SSL BIOs do not support the gets method.

This PR provides a workaround by using the less efficient BIO_get_line() function
in case BIO_gets() returns -2, which means that it is not supported by the BIO.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17317)

http_test.c: Simplify constant init of 'server_args' struct for gcc-4.8.x

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17308)

add OSSL_STACK_OF_X509_free() for commonly used pattern

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17307)

Fix fixup postrelease scripts to avoid creating errors

Otherwise the NEWS.md and CHANGES.md will contain trailing spaces.

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17304)

(cherry picked from commit 132ab70fd852729e7ce41ac7ef2cb2f9969f8b7e)

crypto/dsa.h: fix include guard name

The current include guard name is a duplicate of the one in dsaerr.h.

Noticed via https://lgtm.com/projects/g/openssl/openssl

CLA: trivial

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17303)

Fix the null pointer dereference

Fixes #17296

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17297)

Fix deref after null

ctx may be NULL at 178 line

CLA: trivial

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17293)

cmp_ctx.c: Remove redundancy form the defs of many getters and setters

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17284)

rsa exp: move declarations before code

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17287)

context_init: Fix cleanup in error handling

Also never use OSSL_LIB_CTX_free() on incompletely initialized context.

Fixes #17291

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17294)

ossl_provider_add_to_store: Avoid use-after-free

Avoid freeing a provider that was not up-ref-ed before.

Fixes #17292

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17295)

X509_STORE_new: memory needs to be freed

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17278)

get_ecdsa_sig_rs_bytes: free value of d2i_ECDSA_SIG() before return

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17280)

test/cmp_vfy_test.c: free before return

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17279)

Add test case to verify that the use after free issue is fixed.

Test case based on reproducer by Guido Vranken.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17263)

evp: address a use after free state when using HMAC and MD copy.

Fixes #17261

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17263)

EC_POINT_hex2point: forget to free pt

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17276)

Documentation for RNDR and RNDRRS

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15361)

Add tests for RNDR and combine tests with RDRAND

Add test cases for RNDR and RNDRRS. Combine tests for RDRAND and RNDR to
share common logic.

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15361)

Add support for RNDRRS Provider

Create new provider for RNDRRS. Modify support for rand_cpu to default to
RDRAND/RDSEED on x86 and RNDRRS on aarch64.

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15361)

Add Arm Assembly (aarch64) support for RNG

Include aarch64 asm instructions for random number generation using the
RNDR and RNDRRS instructions. Provide detection functions for RNDR and
RNDRRS getauxval.

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/15361)

APPS/cmp: Fix logic and doc of mutually exclusive -server/-use_mock_srv/-port/-rspin options

Ignore -server with -rspin and exclude all of -use_mock_srv/-port/-rspin.
On the other hand, -server is required if no -use_mock_srv/-port/-rspin is given.
Ignore -tls_used with -use_mock_srv and -rspin; it is not supported with -port.
If -server is not given, ignore -proxy, -no_proxy, and -tls_used.
Also slightly improve the documentation of the two mock server variants.

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17254)

NEWS.md: Add missing empty line

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17274)

Add some CHANGES entries for 3.0.1

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17274)

Add some CHANGES.md entries for the 3.0.1 release

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17274)

Add a test case for the name constraints bug

Where a chain has name constraints but a certificate does not have a SAN
extension but the CN meets the constraints, then this should be acceptable.
However, and OpenSSL bug meant that an internal error was being reported.
This adds a test case for that scenario.

Test for CVE-2021-4044

Reviewed-by: Tomas Mraz <tomas@openssl.org>

Add a TLS test for name constraints with an EE cert without a SAN

It is valid for name constraints to be in force but for there to be no
SAN extension in a certificate. Previous versions of OpenSSL mishandled
this.

Test for CVE-2021-4044

Reviewed-by: Tomas Mraz <tomas@openssl.org>

Add a new Name Constraints test cert

Add a cert which complies with the name constraints but has no
SAN extension

Reviewed-by: Tomas Mraz <tomas@openssl.org>

Fix infinite verification loops due to has_san_id

Where name constraints apply, X509_verify() would incorrectly report an
internal error in the event that a certificate has no SAN extension.

CVE-2021-4044

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>

Fix invalid handling of verify errors in libssl

In the event that X509_verify() returned an internal error result then
libssl would mishandle this and set rwstate to SSL_RETRY_VERIFY. This
subsequently causes SSL_get_error() to return SSL_ERROR_WANT_RETRY_VERIFY.
That return code is supposed to only ever be returned if an application
is using an app verify callback to complete replace the use of
X509_verify(). Applications may not be written to expect that return code
and could therefore crash (or misbehave in some other way) as a result.

CVE-2021-4044

Reviewed-by: Tomas Mraz <tomas@openssl.org>

Update CHANGES and NEWS for new release

Reviewed-by: Richard Levitte <levitte@openssl.org>

Fix VMS installation - Document in CHANGES.md

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16842)

(cherry picked from commit 2daa2a0d00c39e9ff1d79d296c3d48b3db1e72ce)

Fix VMS installation - Override the openssl logical name in descrip.mms.tmpl

This was part of 0cbb6f6a9ac5aa3ff813ef2e5afe6e443708ee20, but was
incomplete in that commit.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16842)

(cherry picked from commit 71a78784e8f000703267276b2f87d055bfa0e00e)

Fix VMS installation - Check the presence of providers in the IVP script

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16842)

(cherry picked from commit fe27680fc6395f661f3949e9e1ef0a83e499e87b)

Fix VMS installation - deassign the same logical names that were defined

The logical name for the engines directory is named one way in
VMS/openssl_startup.com.in, but a different name was deassigned in
VMS/openssl_shutdown.com.in.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16842)

(cherry picked from commit 73e47e39a52d8e1c7515e140b8613304b8abe7ac)

Fix VMS installation - use platform->shlib_version_as_filename() consistently

It's used in Configurations/descrip.mms.tmpl, but was forgotten in the
VMS installation scripts.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16842)

(cherry picked from commit 883008b21ba076b7d87aa8bec1419e8036539a3e)

Fix VMS installation - Define the logical name OSSL$MODULES

Also, the modules installation directory is version agnostic on other
platforms, there's no real reason why it shouldn't be on VMS.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16842)

(cherry picked from commit a4f1e23e6cffdf2ef1aadf96572c251e07869787)

Fix VMS installation - $config{pointer_size} -> $target{pointer_size}

Configurations/descrip.mms.tmpl uses $target{pointer_size}, not
$config{pointer_size}, so the same should be used in installation
scripts, for consistency.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16842)

(cherry picked from commit e30545e9bd84e3ace70fb1e1c5ad6d824545fe36)

Fix VMS installation - consistent program names with version info

The program name version info is supposed to be the major release
version number.  This was forgotten when the versioning scheme was
changed for 3.0, so the minor release version number slipped in as
well.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16842)

(cherry picked from commit 3c9293b4715229dc7ea2116d22c623c2a92ac69a)

Remove some unnecessary undefs in bn_asm.c

This is likely the leftover of a previous hack,
and thus should be removed now.

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17259)

Fix a carry overflow bug in bn_sqr_comba4/8 for mips 32-bit targets

bn_sqr_comba8 does for instance compute a wrong result for the value:
a=0x4aaac919 62056c84 fba7334e 1a6be678 022181ba fd3aa878 899b2346 ee210f45

The correct result is:
r=0x15c72e32 605a3061 d11b1012 3c187483 6df96999 bd0c22ba d3e7d437 4724a82f
    912c5e61 6a187efe 8f7c47fc f6945fe5 75be8e3d 97ed17d4 7950b465 3cb32899

but the actual result was:
r=0x15c72e32 605a3061 d11b1012 3c187483 6df96999 bd0c22ba d3e7d437 4724a82f
    912c5e61 6a187efe 8f7c47fc f6945fe5 75be8e3c 97ed17d4 7950b465 3cb32899

so the forth word of the result was 0x75be8e3c but should have been
0x75be8e3d instead.

Likewise bn_sqr_comba4 has an identical bug for the same value as well:
a=0x022181ba fd3aa878 899b2346 ee210f45

correct result:
r=0x00048a69 9fe82f8b 62bd2ed1 88781335 75be8e3d 97ed17d4 7950b465 3cb32899

wrong result:
r=0x00048a69 9fe82f8b 62bd2ed1 88781335 75be8e3c 97ed17d4 7950b465 3cb32899

Fortunately the bn_mul_comba4/8 code paths are not affected.

Also the mips64 target does in fact not handle the carry propagation
correctly.

Example:
a=0x4aaac91900000000 62056c8400000000 fba7334e00000000 1a6be67800000000
    022181ba00000000 fd3aa87800000000 899b234635dad283 ee210f4500000001

correct result:
r=0x15c72e32272c4471 392debf018c679c8 b85496496bf8254c d0204f36611e2be1
    0cdb3db8f3c081d8 c94ba0e1bacc5061 191b83d47ff929f6 5be0aebfc13ae68d
    3eea7a7fdf2f5758 42f7ec656cab3cb5 6a28095be34756f2 64f24687bf37de06
    2822309cd1d292f9 6fa698c972372f09 771e97d3a868cda0 dc421e8a00000001

wrong result:
r=0x15c72e32272c4471 392debf018c679c8 b85496496bf8254c d0204f36611e2be1
    0cdb3db8f3c081d8 c94ba0e1bacc5061 191b83d47ff929f6 5be0aebfc13ae68d
    3eea7a7fdf2f5758 42f7ec656cab3cb5 6a28095be34756f2 64f24687bf37de06
    2822309cd1d292f8 6fa698c972372f09 771e97d3a868cda0 dc421e8a00000001

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17258)

OSSL_HTTP_proxy_connect(): Fix glitch in response HTTP header parsing

Fixes #17247

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17250)

bn2binpad: Use memset as the buffer will be used later

Apparently using OPENSSL_cleanse() confuses the fuzzer so it
makes the buffer to appear uninitialized. And memset can be
safely used here and it is also potentially faster.

Fixes #17237

Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
(Merged from https://github.com/openssl/openssl/pull/17240)

APPS/cmp: Simplify read_write_req_resp() - 'req' arg must not be NULL anyway

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17251)

ossl_cmp_msg_check_update(): align recipNone check with improved transactionID check

Reviewed-by: Paul Dale <pauli@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17251)

CMP test_commands.csv: improve test for -reqin, adding -reqin_new_tid

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17252)

CMP test_verification.csv: add missing test case for -untrusted with non-matching cert

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17252)

CMP test_enrollment.csv: clean up test cases regarding (non-existing) directories

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17252)

test/evp_extra_test.c: Add EVP_PKEY comparisons in test_EC_priv_pub()

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16765)

test/evp_extra_test.c: Refactor test_fromdata()

test_fromdata() turns out to be a bit inflexible, so we split it into
two functions, make_key_fromdata() and test_selection(), and adjust
test_EVP_PKEY_ffc_priv_pub() and test_EC_priv_pub() accordingly.  This
allows us to check the resulting keys further, not only to check that
the bits we expect are there, but also that the bits that we expect
not to be there to actually not be there!

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16765)

Enhance the explanation of selector bits in provider-keymgmt(7)

This uncovers what has been a mere comment in an attempt to clarify
that the use of selector bits is very much at the discretion of the
provider implementation.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16765)

Adapt our OSSL_FUNC_keymgmt_match() implementations to the EVP_PKEY_eq() fix

The match function (called OSSL_FUNC_keymgmt_match() in our documentation)
in our KEYMGMT implementations were interpretting the selector bits a
bit too strictly, so they get a bit relaxed to make it reasonable to
match diverse key contents.

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16765)

Fix EVP_PKEY_eq() to be possible to use with strictly private keys

EVP_PKEY_eq() assumed that an EVP_PKEY always has the public key
component if it has a private key component.  However, this assumption
no longer strictly holds true, at least for provider backed keys.
EVP_PKEY_eq() therefore needs to be modified to specify that the
private key should be checked too (at the discretion of what's
reasonable for the implementation doing the actual comparison).

Fixes #16267

Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/16765)

Fix Coverity 1494385 logically dead code.

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/17123)

Fix faulty detail in BN_rand() manual

Reviewed-by: Paul Dale <pauli@openssl.org>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17131)

Teach OpenSSL::ParseC about OPENSSL_EXPORT and OPENSSL_EXTERN

Reviewed-by: Dmitry Belyavskiy <beldmit@gmail.com>
Reviewed-by: Tomas Mraz <tomas@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/17215)