8 months agoUI_UTIL_wrap_read_pem_callback(): when |cb| is NULL, use PEM_def_callback
Richard Levitte [Thu, 14 Nov 2019 11:09:42 +0000 (12:09 +0100)]
UI_UTIL_wrap_read_pem_callback(): when |cb| is NULL, use PEM_def_callback

Fixes #10444

Reviewed-by: Matt Caswell <>
(Merged from

8 months agoThread: Avoid a NULL dereference after failed initialisation.
Pauli [Thu, 21 Nov 2019 03:15:21 +0000 (13:15 +1000)]
Thread: Avoid a NULL dereference after failed initialisation.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agomain: avoid a NULL dereference on initialisation.
Pauli [Thu, 21 Nov 2019 03:50:03 +0000 (13:50 +1000)]
main: avoid a NULL dereference on initialisation.

Reviewed-by: Tim Hudson <>
(Merged from

8 months agoPROV: Avoid NULL dereference in SHA3 dup call.
Pauli [Wed, 20 Nov 2019 22:41:42 +0000 (08:41 +1000)]
PROV: Avoid NULL dereference in SHA3 dup call.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoPROV: check for memory allocation failure in digest _dupctx.
Pauli [Wed, 20 Nov 2019 20:47:57 +0000 (06:47 +1000)]
PROV: check for memory allocation failure in digest _dupctx.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agokdfs: make free calls check for NULL.
Pauli [Thu, 21 Nov 2019 04:19:50 +0000 (14:19 +1000)]
kdfs: make free calls check for NULL.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoFix a race condition in SNI handling
Benjamin Kaduk [Wed, 13 Nov 2019 17:42:19 +0000 (09:42 -0800)]
Fix a race condition in SNI handling

As was done for ciphers, supported groups, and EC point formats in, only write the negotiated
SNI hostname value to the session object when not resuming, even for
TLS 1.3 resumptions.  Otherwise, when using a stateful session cache
(as is done by default when 0-RTT data is enabled), we can have multiple
SSLs active using the same in-memory session object, which leads to
double-frees and similar race conditions in the SNI handler prior
to this commit.

Fortunately, since draft-ietf-tls-tls13-22, there is no requirement
that the SNI hostname be preserved across TLS 1.3 resumption, and thus
not a need to continually update the session object with the "current"
value (to be used when producing session tickets, so that the subsequent
resumption can be checked against the current value).  So we can just
relax the logic and only write to the session object for initial handshakes.
This still leaves us in a somewhat inconsistent state, since if the SNI value
does change across handshakes, the session object will continue to record
the initial handshake's value, even if that bears no relation to the
current handshake.  The current SSL_get_servername() implementation
prefers the value from the session if s->hit, but a more complete fix
for that and related issues is underway in; there is no need to wait
for the complete fix for SNI name handling in order to close the
race condition and avoid runtime crashes.

Reviewed-by: Matt Caswell <>
(Merged from

8 months agoCore: allow NULL argument to stored_namemap_free().
Pauli [Wed, 20 Nov 2019 20:55:50 +0000 (06:55 +1000)]
Core: allow NULL argument to stored_namemap_free().

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoDocumentation updates due to naming tweaks
Matt Caswell [Fri, 4 Oct 2019 11:46:33 +0000 (12:46 +0100)]
Documentation updates due to naming tweaks

Also documents our new canonical naming.

Reviewed-by: Paul Dale <>
(Merged from

8 months agoTweak some algorithm naming inconsistencies
Matt Caswell [Fri, 4 Oct 2019 10:50:49 +0000 (11:50 +0100)]
Tweak some algorithm naming inconsistencies

Make some algorithms names better match our "canonical" style.

Reviewed-by: Paul Dale <>
(Merged from

8 months agoEVP p_lib: Add NULL check to EVP_PKEY_missing_parameters.
Pauli [Wed, 20 Nov 2019 03:14:39 +0000 (13:14 +1000)]
EVP p_lib: Add NULL check to EVP_PKEY_missing_parameters.

Check for NULL and return error if so.
This can possibly be called from apps/ca.c with a NULL argument.

Reviewed-by: Paul Yang <>
(Merged from

8 months agoEngine: Add NULL check.
Pauli [Wed, 20 Nov 2019 03:07:04 +0000 (13:07 +1000)]
Engine: Add NULL check.

Add NULL check for return from pkey_asn1_meths.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoECDSA: don't clear free memory after verify.
Pauli [Wed, 20 Nov 2019 03:20:01 +0000 (13:20 +1000)]
ECDSA: don't clear free memory after verify.

Verifications are public, there is no need to clear the used storage before
freeing it.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoReplumbing: pre-populate the EVP namemap with commonly known names
Richard Levitte [Thu, 23 May 2019 01:27:37 +0000 (03:27 +0200)]
Replumbing: pre-populate the EVP namemap with commonly known names

This adds ossl_namemap_empty(), to detect if a namemap is empty and
can thereby be pre-populated.

This also affects the way legacy NIDs are looked up in
evp_cipher_from_dispatch() and evp_md_from_dispatch().  Instead of
trying to find the NID directly, look up the legacy method structure
and grab the NID from there.  The reason is that NIDs can be aliases
for other NIDs, which looks like a clash even if wasn't really one.

Reviewed-by: Paul Dale <>
(Merged from

8 months agoFix sha512_block_data_order_avx2 backtrace info
Bernd Edlinger [Sat, 17 Aug 2019 23:54:41 +0000 (01:54 +0200)]
Fix sha512_block_data_order_avx2 backtrace info

We store a secondary frame pointer info for the debugger
in the red zone.

Fixes #8853

[extended tests]

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoAllow specifying the tag after AAD in CCM mode (2)
Patrick Steuer [Fri, 1 Nov 2019 23:02:47 +0000 (00:02 +0100)]
Allow specifying the tag after AAD in CCM mode (2)

In addition to 67c81ec3 which introduced this behavior in CCM mode
docs but only implemented it for AES-CCM.

Signed-off-by: Patrick Steuer <>
Reviewed-by: Paul Dale <>
(Merged from

8 months agoDont pass zero length input to asm modules for ciphers
Patrick Steuer [Fri, 1 Nov 2019 13:58:36 +0000 (14:58 +0100)]
Dont pass zero length input to asm modules for ciphers

The asm modules may assume an input length > 0.

Fixes: #9262

Signed-off-by: Patrick Steuer <>
Reviewed-by: Paul Dale <>
(Merged from

8 months agoDocument command parameters.
Rich Salz [Fri, 20 Sep 2019 01:33:17 +0000 (21:33 -0400)]
Document command parameters.

Add documentation for all commands that have parameters.
Fix a couple of minor doc and programming bugs, too.

Fixes #10313

Reviewed-by: Paul Dale <>
Reviewed-by: Tomas Mraz <>
(Merged from

8 months agoAdd test for non-default sized keys in variable key size ciphers
Matt Caswell [Fri, 15 Nov 2019 14:15:09 +0000 (14:15 +0000)]
Add test for non-default sized keys in variable key size ciphers

Various ciphers have the ability to support varaible length keys.
This adds some test cases for algorithms where we did not test a key
length other than the default size.

Reviewed-by: Shane Lontis <>
(Merged from

8 months agoEVP_CIPHER_CTX_set_keylen should not succeed if a bad keylen is passed
Matt Caswell [Thu, 14 Nov 2019 16:05:19 +0000 (16:05 +0000)]
EVP_CIPHER_CTX_set_keylen should not succeed if a bad keylen is passed

EVP_CIPHER_CTX_set_keylen() was succeeding even though a bad key length
is passed to it. This is because the set_ctx_params() were all accepting
this parameter and blindly changing the keylen even though the cipher did
not accept a variable key length. Even removing this didn't entirely
resolve the issue because set_ctx_params() functions succeed even if
passed a parameter they do not recognise.

This should fix various issues found by OSSfuzz/Cryptofuzz.

Reviewed-by: Shane Lontis <>
(Merged from

8 months agofix s390 compile error due to missing comma.
Shane Lontis [Mon, 18 Nov 2019 22:38:19 +0000 (08:38 +1000)]
fix s390 compile error due to missing comma.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoFix the issue that aes-xxx-siv algorithms failed in OpenSSL speed test
Xiaokang Qian [Tue, 12 Nov 2019 08:48:27 +0000 (08:48 +0000)]
Fix the issue that aes-xxx-siv algorithms failed in OpenSSL speed test

Speed test, aes-siv related cases fail on both x86 and arm.
The return value of siv_init() causes this problem, remove
the iv check to fix it.
Verify it locally, the result is pass.

Fixes #10416

Change-Id: If1a18599f3d0f56f22a1ce4f8f114b8db0f68cca

Reviewed-by: Richard Levitte <>
Reviewed-by: Shane Lontis <>
(Merged from

8 months agoProperties: make query cache reference count aware.
Pauli [Mon, 11 Nov 2019 01:17:32 +0000 (11:17 +1000)]
Properties: make query cache reference count aware.

The property query cache was not reference count aware and this could cause
problems if the property store removes an algorithm while it is being returned
from an asynchronous query.  This change makes the cache reference count aware
and avoids disappearing algorithms.

A side effect of this change is that the reference counts are now owned by the
cache and store.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoFix Use after free when copying cipher ctx
Shane Lontis [Mon, 18 Nov 2019 03:13:05 +0000 (13:13 +1000)]
Fix Use after free when copying cipher ctx

Fixes #10438
issue found by clusterfuzz/ossfuzz

The dest was getting a copy of the src structure which contained a pointer that should point to an offset inside itself - because of the copy it was pointing to the original structure.

The setup for a ctx is mainly done by the initkey method in the PROV_CIPHER_HW structure. Because of this it makes sense that the structure should also contain a copyctx method that is use to resolve any pointers that need to be setup.

A dup_ctx has been added to the cipher_enc tests in evp_test. It does a dup after setup and then frees the original ctx. This detects any floating pointers in the duplicated context that were pointing back to the freed ctx.

Reviewed-by: Matt Caswell <>
(Merged from

8 months agoPEM: constify PEM_write_ routines
Richard Levitte [Fri, 15 Nov 2019 07:54:17 +0000 (08:54 +0100)]
PEM: constify PEM_write_ routines

There's no reason why the object to be written, or the key string
given by the caller should be non-const.

This makes the IMPLEMENT_PEM_..._const and  DECLARE_PEM_..._const
macros superfluous, so we keep them around but mark them deprecated.

In all places where IMPLEMENT_PEM_..._const and  DECLARE_PEM_..._const
are used, they are replaced with the corresponding macros without

Reviewed-by: Matt Caswell <>
(Merged from

8 months agos390x assembly pack: fix bn_mul_comba4
Patrick Steuer [Fri, 15 Nov 2019 09:52:03 +0000 (10:52 +0100)]
s390x assembly pack: fix bn_mul_comba4

Signed-off-by: Patrick Steuer <>
Reviewed-by: Paul Dale <>
(Merged from

8 months agoAdd missing EVP_PKEY_METHOD accessors for digestsign and digestverify
Anthony Hu [Fri, 8 Nov 2019 02:47:53 +0000 (21:47 -0500)]
Add missing EVP_PKEY_METHOD accessors for digestsign and digestverify

Reviewed-by: Paul Dale <>
Reviewed-by: Matthias St. Pierre <>
(Merged from

8 months agocommit a6efbe123af3d98b4d10d4fcdfe68dc5303212f8
raniervf [Fri, 15 Nov 2019 22:28:00 +0000 (08:28 +1000)]
commit a6efbe123af3d98b4d10d4fcdfe68dc5303212f8
Author: raniervf <>
Date:   Thu Nov 7 18:59:11 2019 -0300

Avoid calling strlen repeatedly in loops.

Reviewed-by: Paul Yang <>
Reviewed-by: Paul Dale <>
(Merged from

8 months agoFix some broken doc links
Rich Salz [Thu, 22 Aug 2019 05:21:25 +0000 (10:51 +0530)]
Fix some broken doc links

Replace L<> link to header-file with a C<> reference.
Change some broken L<provider(3)> links to L<provider(7)>.
For consistency, rename four cipher pages to have a specific mode.
Fix up all references to any "generic" names to point to specific names.

Reviewed-by: Richard Levitte <>
Reviewed-by: Tomas Mraz <>
(Merged from

8 months agoMake sure we free the CONF structure allocated by confdump
Matt Caswell [Tue, 12 Nov 2019 17:37:11 +0000 (17:37 +0000)]
Make sure we free the CONF structure allocated by confdump

A travis build was failing because of a memory leak in confdump due to
us not freeing an allocated CONF structure.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agotravis.yml: add arch s390x target
Patrick Steuer [Thu, 14 Nov 2019 09:34:02 +0000 (10:34 +0100)]
travis.yml: add arch s390x target

Signed-off-by: Patrick Steuer <>
Reviewed-by: Richard Levitte <>
(Merged from

8 months agotestutil/init.c rename to testutil/testutil_init.c
Patrick Steuer [Wed, 13 Nov 2019 17:39:51 +0000 (18:39 +0100)]
testutil/init.c rename to testutil/testutil_init.c

Avoid conflicts with some linkers.

Signed-off-by: Patrick Steuer <>
Reviewed-by: Richard Levitte <>
(Merged from

8 months agoA few fixes in the EVP_KDF-KB and EVP_KDF-KRB5KDF manpages
Tomas Mraz [Tue, 12 Nov 2019 15:59:23 +0000 (16:59 +0100)]
A few fixes in the EVP_KDF-KB and EVP_KDF-KRB5KDF manpages

The context and label is info and salt respectively - fix
the example in KB manpage for that.

There are some typos and bug in EVP_KDF_derive call in the
KRB5KDF example.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoDelete recently documented functions from missingcrypto
Tomas Mraz [Wed, 13 Nov 2019 10:10:51 +0000 (11:10 +0100)]
Delete recently documented functions from missingcrypto

Reviewed-by: Matt Caswell <>
(Merged from

8 months agoFix an uninitialised read in conf_def.c
Matt Caswell [Tue, 12 Nov 2019 17:16:14 +0000 (17:16 +0000)]
Fix an uninitialised read in conf_def.c

PR 8882 added a new field to the CONF structure. Unfortunately this
structure was created using OPENSSL_malloc() and the new field was not
explicitly initialised in the "init" function. Therefore when we came to
read it for the first time we got an uninitialised read.

Reviewed-by: Richard Levitte <>
Reviewed-by: Shane Lontis <>
(Merged from

8 months agoFix no-blake2
Matt Caswell [Tue, 12 Nov 2019 16:02:24 +0000 (16:02 +0000)]
Fix no-blake2

We modify the file to exclude the legacy_blake2.c file in
the event that blake2 support has been disabled.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoFix no-engine
Matt Caswell [Tue, 12 Nov 2019 15:26:15 +0000 (15:26 +0000)]
Fix no-engine

Fix a misplaced "#endif" which was disabling a little too much code.

Reviewed-by: Tomas Mraz <>
(Merged from

8 months agoCORE & PROV: make export of key data leaner through callback
Richard Levitte [Fri, 8 Nov 2019 14:24:42 +0000 (15:24 +0100)]
CORE & PROV: make export of key data leaner through callback

Exporting data from a provider owned domainparams or key is quite an
ordeal, with having to figure out what parameter keys an
implementation supports, call the export function a first time to find
out how large each parameter buffer must be, allocate the necessary
space for it, and call the export function again.

So how about letting the export function build up the key data params
and call back with that?  This change implements exactly such a

Reviewed-by: Matt Caswell <>
(Merged from

8 months agoFix no-cmac and no-camellia
Matt Caswell [Tue, 12 Nov 2019 15:51:00 +0000 (15:51 +0000)]
Fix no-cmac and no-camellia

Guard two tests that depend on CMAC and Camellia so that we don't fail
if those algorithms are not available.

Reviewed-by: Tomas Mraz <>
(Merged from

8 months agoAdd confdump to .gitignore
Matt Caswell [Tue, 12 Nov 2019 17:41:29 +0000 (17:41 +0000)]
Add confdump to .gitignore

We added a new executable to the test directory which didn't match the
existing gitignore rules, so we add it explicitly.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoDocument various new/changes EVP_PKEY_CTX_* functions
Matt Caswell [Wed, 6 Nov 2019 13:39:21 +0000 (13:39 +0000)]
Document various new/changes EVP_PKEY_CTX_* functions

We have converted a number of macros to functions and made them work
with providers. We've also added some *_ex() variants that needed

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoDocument the public EVP_ASYM_CIPHER releated functions
Matt Caswell [Wed, 6 Nov 2019 11:03:12 +0000 (11:03 +0000)]
Document the public EVP_ASYM_CIPHER releated functions

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoAdd documentation for the Asymmetric Cipher Operation
Matt Caswell [Tue, 29 Oct 2019 09:24:24 +0000 (09:24 +0000)]
Add documentation for the Asymmetric Cipher Operation

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoMove RSA Asym cipher code to the default provider
Matt Caswell [Mon, 28 Oct 2019 13:40:39 +0000 (13:40 +0000)]
Move RSA Asym cipher code to the default provider

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoIncrease OSSL_PARAM_BLD_MAX for multi-prime RSA
Matt Caswell [Mon, 28 Oct 2019 14:43:42 +0000 (14:43 +0000)]
Increase OSSL_PARAM_BLD_MAX for multi-prime RSA

The old value of 10 for OSSL_PARAM_BLD_MAX is insufficient for multi-prime
RSA. That code has this assert:

        if (!ossl_assert(/* n, e */ 2 + /* d */ 1 + /* numprimes */ 1
                         + numprimes + numexps + numcoeffs
                         <= OSSL_PARAM_BLD_MAX))
            goto err;

So we increase OSSL_PARAM_BLD_MAX which would be enough for 7 primes
(more than you would ever reasonably want).

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoImplement provider support for Asym Ciphers
Matt Caswell [Tue, 1 Oct 2019 08:40:57 +0000 (09:40 +0100)]
Implement provider support for Asym Ciphers

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoAllow strings in params to be of zero length
Matt Caswell [Thu, 17 Oct 2019 16:19:21 +0000 (17:19 +0100)]
Allow strings in params to be of zero length

Sometimes it is useful to be able to pass NULL/zero length strings

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoAllow setting a NULL pointer in a params structure
Matt Caswell [Fri, 11 Oct 2019 16:42:19 +0000 (17:42 +0100)]
Allow setting a NULL pointer in a params structure

Sometimes it is valid to send a NULL pointer in params.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoCORE: Add a generic callback function type
Richard Levitte [Mon, 11 Nov 2019 17:38:57 +0000 (18:38 +0100)]
CORE: Add a generic callback function type

This offers a very generic way to define a callback as well as calling

Reviewed-by: Matt Caswell <>
(Merged from

8 months agochacha_enc.c: fix for EBCDIC platforms
Joerg Schmidbauer [Tue, 12 Nov 2019 09:26:47 +0000 (10:26 +0100)]
chacha_enc.c: fix for EBCDIC platforms

Signed-off-by: Joerg Schmidbauer <>
Reviewed-by: Patrick Steuer <>
Reviewed-by: Matt Caswell <>
(Merged from

8 months agoExtend docs for EC_POINT conversion functions
Nicola Tuveri [Mon, 11 Nov 2019 22:52:00 +0000 (00:52 +0200)]
Extend docs for EC_POINT conversion functions

Add more explicit documentation about the relation between
EC_POINT_point2oct(), EC_POINT_point2hex(), EC_POINT_point2bn() and
their reverse.

In particular highlight that EC_POINT_point2oct() and
EC_POINT_oct2point() conform to, respectively, Sec. 2.3.3 and Sec. 2.3.4
of the SECG SEC 1 standard (which is the normative reference for the
already mentioned RFC 5480), highlighting with a note how this affect
the encoding/decoding of the point at infinity (which in contrast with
any other valid generic point of a curve is assigned an exceptional
fixed octet string encoding, i.e., 0x00).

Reviewed-by: Matt Caswell <>
(Merged from

8 months agoAdd self-test for EC_POINT_hex2point
Nicola Tuveri [Fri, 1 Nov 2019 20:09:40 +0000 (22:09 +0200)]
Add self-test for EC_POINT_hex2point

Adds tests for each curve to ensure that encodings obtained through
EC_POINT_hex2point() can be fed to EC_POINT_point2hex() yielding a point
identical to the one from which the encoding is generated.

Reviewed-by: Matt Caswell <>
(Merged from

8 months agoFix EC_POINT_bn2point() for BN_zero()
Nicola Tuveri [Fri, 1 Nov 2019 20:38:21 +0000 (22:38 +0200)]
Fix EC_POINT_bn2point() for BN_zero()

EC_POINT_bn2point() rejected BIGNUMs with a zero value.

This behavior indirectly caused failures when converting a point
at infinity through EC_POINT_point2hex() and then back to a point with

With this change such BIGNUMs are treated like any other and exported to
an octet buffer filled with zero.
It is then EC_POINT_oct2point() (either the default implementation or
the custom one in group->meth->oct2point) to determine if such encoding
maps to a valid point (generally the point at infinity is encoded as

Fixes #10258

Reviewed-by: Matt Caswell <>
(Merged from

8 months agolibcrypto.num: remove some unused EVP_MAC symbols
Dr. Matthias St. Pierre [Tue, 12 Nov 2019 11:48:15 +0000 (12:48 +0100)]
libcrypto.num: remove some unused EVP_MAC symbols

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoAdd more tests for apps/req
Nicola Tuveri [Thu, 31 Oct 2019 15:17:31 +0000 (17:17 +0200)]
Add more tests for apps/req
highlighted that existing testing infrastructure is not covering common
usage patterns of the `req` app.

This commit explicitly adds request generations thorugh the CLI using
RSA, DSA and ECDSA (P-256) keys.

Reviewed-by: Richard Levitte <>
(Merged from

8 months agoAdd missing docs for some PKCS12 functions
Jon Spillett [Mon, 2 Sep 2019 04:51:05 +0000 (14:51 +1000)]
Add missing docs for some PKCS12 functions

Reviewed-by: Matt Caswell <>
Reviewed-by: Tomas Mraz <>
(Merged from

8 months agoadd OPENSSL_FUNC.pod documenting OPENSSL_MSTR, OPENSSL_FUNC, and friends
Dr. David von Oheimb [Tue, 5 Nov 2019 10:43:52 +0000 (11:43 +0100)]
add OPENSSL_FUNC.pod documenting OPENSSL_MSTR, OPENSSL_FUNC, and friends

Reviewed-by: Richard Levitte <>
Reviewed-by: Dmitry Belyavskiy <>
(Merged from

8 months agore-add definition of OPENSSL_MSTR deleted from opensslv.h in macros.h
Dr. David von Oheimb [Tue, 22 Oct 2019 11:38:16 +0000 (13:38 +0200)]
re-add definition of OPENSSL_MSTR deleted from opensslv.h in macros.h

Reviewed-by: Richard Levitte <>
Reviewed-by: Dmitry Belyavskiy <>
(Merged from

8 months agoSSL: Document SSL_add_{file,dir,store}_cert_subjects_to_stack()
Richard Levitte [Sun, 10 Nov 2019 14:36:24 +0000 (15:36 +0100)]
SSL: Document SSL_add_{file,dir,store}_cert_subjects_to_stack()

This also removes the incorrect documentation comments by those
functions, and fixes a bug in SSL_add_store_cert_subjects_to_stack(),
where the condition for recursive addition was 'depth == 0' when it
should be 'depth > 0'.

Reviewed-by: Tomas Mraz <>
(Merged from

8 months agoAdd a tester of OpenSSL config files and test the current .pragma
Richard Levitte [Mon, 19 Aug 2019 04:35:54 +0000 (06:35 +0200)]
Add a tester of OpenSSL config files and test the current .pragma

test/confdump.c reads an OpenSSL config file and prints out the
processed result.  This can be used to check that a config file is
processed correctly.

We add a test recipe and the necessary data to test the dollarid

Reviewed-by: Matt Caswell <>
(Merged from

8 months agoAdd a .pragma directive for configuration files
Richard Levitte [Mon, 6 May 2019 08:50:35 +0000 (10:50 +0200)]
Add a .pragma directive for configuration files

Currently added pragma:

.pragma dollarid:on

This allows dollar signs to be a keyword character unless it's
followed by a opening brace or parenthesis.

Fixes #8207

Reviewed-by: Matt Caswell <>
(Merged from

8 months agoMake sure KDF reason codes are conserved in their current state
Richard Levitte [Sun, 10 Nov 2019 12:07:46 +0000 (13:07 +0100)]
Make sure KDF reason codes are conserved in their current state

Because KDF errors are deprecated and only conserved for backward
compatibilty, we must make sure that they remain untouched.  A simple
way to signal that is by modifying crypto/err/ and replace
the main header file (include/openssl/kdf.h in this case) with 'NONE',
while retaining the error table file (crypto/kdf/kdf_err.c).

util/ is modified to silently ignore anything surrounding a
conserved lib when such a .ec line is found.

Reviewed-by: Tomas Mraz <>
(Merged from

8 months agoDeprecate ERR_load_KDF_strings()
Richard Levitte [Sun, 10 Nov 2019 12:06:48 +0000 (13:06 +0100)]
Deprecate ERR_load_KDF_strings()

Not only deprecate, but also remove the reason strings and make
ERR_load_KDF_strings() do nothing.

Reviewed-by: Tomas Mraz <>
(Merged from

8 months agoutil/perl/OpenSSL/ Fix 'extern "C" DEPRECATEDIN_x_y(...)'
Richard Levitte [Sun, 10 Nov 2019 12:03:15 +0000 (13:03 +0100)]
util/perl/OpenSSL/ Fix 'extern "C" DEPRECATEDIN_x_y(...)'

The parser expected an 'extern "C"' followed by a single declaration
to always end with a semicolon.  Then came along something like this:

    extern "C" DEPRECATEDIN_3_0(int ERR_load_KDF_strings(void))

This change adjusts the detector of 'extern "C"' to also take in
accound a declaration that ends with a parenthesis.

Reviewed-by: Tomas Mraz <>
(Merged from

8 months agoReinstate the KDF error macros
Richard Levitte [Sun, 10 Nov 2019 04:16:36 +0000 (05:16 +0100)]
Reinstate the KDF error macros

For minimum breakage with existing applications that might use them.

This reverts commit fe6ec26b204a056aee2a24b79df09a45b2308603 and

Fixes #10340

Reviewed-by: Tomas Mraz <>
(Merged from

8 months agoFix cut'n'paste typos in the provider-digest pod
Dmitry Belyavskiy [Sat, 9 Nov 2019 20:25:32 +0000 (23:25 +0300)]
Fix cut'n'paste typos in the provider-digest pod

Reviewed-by: Richard Levitte <>
Reviewed-by: Matt Caswell <>
(Merged from

8 months agoFix misspelled resumption_label for CHARSET_EBCDIC
Ido Ben-Natan [Sat, 9 Nov 2019 13:04:39 +0000 (15:04 +0200)]
Fix misspelled resumption_label for CHARSET_EBCDIC

The resumption_label variable when CHARSET_EBCDIC was enabled, was misspelled.
Instead of evaluating to 'res binder' as expected, it evaluated to 'red binder'.

CLA: trivial

Reviewed-by: Kurt Roeckx <>
Reviewed-by: Matthias St. Pierre <>
(Merged from

8 months agoFix no-dsa
Matt Caswell [Thu, 7 Nov 2019 09:24:18 +0000 (09:24 +0000)]
Fix no-dsa

Make sure we don't try and load a DSA key in the tests if DSA has been

Reviewed-by: Shane Lontis <>
Reviewed-by: Richard Levitte <>
(Merged from

8 months agoAdd rc4_hmac_md5 cipher to default provider
Shane Lontis [Thu, 10 Oct 2019 06:42:20 +0000 (16:42 +1000)]
Add rc4_hmac_md5 cipher to default provider

Reviewed-by: Matt Caswell <>
(Merged from

9 months agoFix L<xxx(1)> links to be L<openssl-xxx(1)>
Rich Salz [Fri, 1 Nov 2019 20:26:05 +0000 (16:26 -0400)]
Fix L<xxx(1)> links to be L<openssl-xxx(1)>

Reviewed-by: Richard Levitte <>
Reviewed-by: Matthias St. Pierre <>
(Merged from

9 months agoWorkaround for Windows-based GOST implementations
Dmitry Belyavskiy [Thu, 7 Nov 2019 14:58:15 +0000 (17:58 +0300)]
Workaround for Windows-based GOST implementations

Many Windows-based GOST TLS implementations are unable to extend the
list of supported SignatureAlgorithms because of lack of the necessary
callback in Windows. So for TLS 1.2 it makes sense to imply the support
of GOST algorithms in case when the GOST ciphersuites are present.

Reviewed-by: Paul Dale <>
(Merged from

9 months agoRename providers/common/provlib.c to nid_to_name.c
Richard Levitte [Fri, 8 Nov 2019 09:29:24 +0000 (10:29 +0100)]
Rename providers/common/provlib.c to nid_to_name.c

It contains only one function, which should only get added to non-FIPS

Reviewed-by: Paul Dale <>
(Merged from

9 months agoFinal cleanup after move to leaner EVP_PKEY methods
Richard Levitte [Thu, 31 Oct 2019 11:10:01 +0000 (12:10 +0100)]
Final cleanup after move to leaner EVP_PKEY methods

Now that KEYMGMT method pointers have moved away from the diverse
methods that are used with EVP_PKEY_CTX, we no longer need to pass
special argument to evp_generic_fetch() and evp_generic_do_all().

Reviewed-by: Shane Lontis <>
(Merged from

9 months agoConfiguration: make Solaris builds with gcc recognise GNU ld
Richard Levitte [Thu, 21 Mar 2019 06:51:43 +0000 (07:51 +0100)]
Configuration: make Solaris builds with gcc recognise GNU ld

The solaris config targets assumed that GNU cc used Sun ld at all
times.  However, there are setups where GNU ld is used instead, so we
adapt the Solaris gcc config targets to use the mechanism introduced
with Configurations/ to try to detect what ld flavor is
being used and set the diverse ld flags accordingly.

Fixes #8547

Reviewed-by: Matt Caswell <>
(Merged from

9 months agobss_dgram.c: fix unaligned access
Patrick Steuer [Sat, 2 Nov 2019 15:31:28 +0000 (16:31 +0100)]
bss_dgram.c: fix unaligned access

char (alignment 1) casted to union sctp_notification (alignment > 1).

Fixes: #9538

Signed-off-by: Patrick Steuer <>
Reviewed-by: Matt Caswell <>
(Merged from

9 months agoRSA generation: Use more bits of 1/sqrt(2)
Kurt Roeckx [Wed, 23 Oct 2019 20:10:54 +0000 (22:10 +0200)]
RSA generation: Use more bits of 1/sqrt(2)

The old version always sets the top 2 bits, so the most significate byte
of the primes was always >= 0xC0. We now use 256 bits to represent
1/sqrt(2) = 0x0.B504F333F9DE64845...

Reviewed-by: Shane Lontis <>
Reviewed-by: Richard Levitte <>
GH: #10246

9 months agoFix a -Warray-bounds gcc warning in OPENSSL_DIR_read
Bernd Edlinger [Sun, 3 Nov 2019 16:34:23 +0000 (17:34 +0100)]
Fix a -Warray-bounds gcc warning in OPENSSL_DIR_read

'__builtin_strncpy' offset [275, 4095] from the object at
'direntry' is out of the bounds of referenced subobject 'd_name'
with type 'char[256]' at offset 19

Reviewed-by: Kurt Roeckx <>
Reviewed-by: Richard Levitte <>
(Merged from

9 months agoconf_def.c: Avoid calling strlen() in a loop
raniervf [Tue, 5 Nov 2019 01:32:43 +0000 (22:32 -0300)]
conf_def.c: Avoid calling strlen() in a loop

Reviewed-by: Matt Caswell <>
Reviewed-by: Matthias St. Pierre <>
(Merged from

9 months agoCleanup include/openssl/
Richard Levitte [Thu, 17 Oct 2019 16:40:47 +0000 (18:40 +0200)]
Cleanup include/openssl/

Now that we generate include/openssl/opensslv.h, there's no point
keeping some macross around, we can just set a simpler set to their
respective value and be done with it.

Reviewed-by: Matt Caswell <>
(Merged from

9 months agoAdd ssl_ctx_test to test suite.
Rich Salz [Sat, 2 Nov 2019 17:02:37 +0000 (13:02 -0400)]
Add ssl_ctx_test to test suite.

Also fix the test as min version > max version is allowed because
the API calls to set min and max versions are separate and there
can be legitimately a temporary situation when the condition is
true even with correctly working application.

The failure in this condition will be detected only during
a handshake attempt.

Reviewed-by: Richard Levitte <>
Reviewed-by: Tomas Mraz <>
(Merged from

9 months agoAdd AES SIV ciphers to default provider
Shane Lontis [Fri, 8 Nov 2019 02:14:44 +0000 (12:14 +1000)]
Add AES SIV ciphers to default provider

Reviewed-by: Richard Levitte <>
Reviewed-by: Paul Dale <>
(Merged from

9 months agoRefactor PEM_read_bio_{PrivateKey,Parameters,DHparams}
Richard Levitte [Sat, 25 Feb 2017 22:38:46 +0000 (23:38 +0100)]
Refactor PEM_read_bio_{PrivateKey,Parameters,DHparams}

They now go through internal STORE routines to do their job.

Reviewed-by: Matt Caswell <>
(Merged from

9 months agoAdd "sections" to -help output
Rich Salz [Thu, 7 Nov 2019 20:08:30 +0000 (06:08 +1000)]
Add "sections" to -help output

Remove "Valid options" label, since all commands have sections (and
[almost] always the first one is "General options").
Have "list --options" ignore section headers
Reformat ts's additional help

Add output section

Reviewed-by: Tomas Mraz <>
Reviewed-by: Paul Dale <>
(Merged from

9 months agoAdd a test for EVP_PKEY_keymake() and EVP_PKEY_make()
Richard Levitte [Thu, 17 Oct 2019 21:53:44 +0000 (23:53 +0200)]
Add a test for EVP_PKEY_keymake() and EVP_PKEY_make()

This test is a bit lame, but will either be completed as functionality
is added in the default provider, or the new functions may start
getting used in evp_test.c and this program will disappear.

Reviewed-by: Shane Lontis <>
(Merged from

9 months agoAdd EVP functionality to create domain params and keys by user data
Richard Levitte [Tue, 15 Oct 2019 12:50:35 +0000 (14:50 +0200)]
Add EVP functionality to create domain params and keys by user data

This is the EVP operation that corresponds to creating direct RSA, DH
and DSA keys and set their numbers, to then assign them to an EVP_PKEY,
but done entirely using an algorithm agnostic EVP interface.

Reviewed-by: Shane Lontis <>
(Merged from

9 months agoUpdate INSTALL to document the changed '--api' and 'no-deprecated'
Richard Levitte [Tue, 5 Nov 2019 16:48:33 +0000 (17:48 +0100)]
Update INSTALL to document the changed '--api' and 'no-deprecated'

Reviewed-by: Tim Hudson <>
(Merged from

9 months agoinclude/openssl/asn1.h: Remove version dependent inclusion of <openssl/bn.h>
Richard Levitte [Tue, 5 Nov 2019 16:38:24 +0000 (17:38 +0100)]
include/openssl/asn1.h: Remove version dependent inclusion of <openssl/bn.h>

It's unclear if this dependency was because ASN1 functions that use
BIGNUM didn't exist before 1.1.0, or if it was a mistaken attempt at
deprecation.  Since there exist ASN1 functions using BIGNUM now, it
seems pointless to keep that check, and unnecessarily including
<openssl/bn.h> should be harmless either way.

Reviewed-by: Tim Hudson <>
(Merged from

9 months agoUpdate the check surround the BN_zero() implementation
Richard Levitte [Tue, 5 Nov 2019 16:37:17 +0000 (17:37 +0100)]
Update the check surround the BN_zero() implementation

Reviewed-by: Tim Hudson <>
(Merged from

9 months agoUpdate some inclusions of <openssl/macros.h>
Richard Levitte [Tue, 5 Nov 2019 16:36:44 +0000 (17:36 +0100)]
Update some inclusions of <openssl/macros.h>

Reviewed-by: Tim Hudson <>
(Merged from

9 months agoinclude/openssl/ remove spurious HEADER_FILE_H definition
Richard Levitte [Tue, 5 Nov 2019 16:35:12 +0000 (17:35 +0100)]
include/openssl/ remove spurious HEADER_FILE_H definition

This macro was never defined in existing releases, there's no reason
for us to create a macro that we immediately deprecate.

Reviewed-by: Tim Hudson <>
(Merged from

9 months agoUpdate source files for pre-3.0 deprecation
Richard Levitte [Tue, 5 Nov 2019 16:34:09 +0000 (17:34 +0100)]
Update source files for pre-3.0 deprecation

Reviewed-by: Tim Hudson <>
(Merged from

9 months agoUpdate source files for deprecation at 3.0
Richard Levitte [Tue, 5 Nov 2019 16:28:50 +0000 (17:28 +0100)]
Update source files for deprecation at 3.0

Previous macros suggested that from 3.0, we're only allowed to
deprecate things at a major version.  However, there's no policy
stating this, but there is for removal, saying that to remove
something, it must have been deprecated for 5 years, and that removal
can only happen at a major version.

Meanwhile, the semantic versioning rule is that deprecation should
trigger a MINOR version update, which is reflected in the macro names
as of this change.

Reviewed-by: Tim Hudson <>
(Merged from

9 months agoChange the logic and behaviour surrounding '--api' and 'no-deprecated'
Richard Levitte [Tue, 5 Nov 2019 16:00:33 +0000 (17:00 +0100)]
Change the logic and behaviour surrounding '--api' and 'no-deprecated'

At some point in time, there was a 'no-deprecated' configuration
option, which had the effect of hiding all declarations of deprecated
stuff, i.e. make the public API look like they were all removed.

At some point in time, there was a '--api' configuration option, which
had the effect of having the public API look like it did in the version
given as value, on a best effort basis.  In practice, this was used to
get different implementations of BN_zero(), depending on the desired
API compatibility level.

At some later point in time, '--api' was changed to mean the same as
'no-deprecated', but only for the deprecations up to and including the
desired API compatibility level.  BN_zero() has been set to the
pre-1.0.0 implementation ever since, unless 'no-deprecation' has been

This change turns these options back to their original meaning, but
with the slight twist that when combined, i.e. both '--api' and
'no-deprecated' is given, the declarations that are marked deprecated
up to an including the desired API compatibility level are hidden,
simulating that they have been removed.

If no desired API compatibility level has been given, then
configuration sets the current OpenSSL version by default.

Furthermore, the macro OPENSSL_API_LEVEL is now used exclusively to
check what API compatibility level is desired.  For checking in code
if `no-deprecated` has been configured for the desired API
compatibility level, macros for each supported level is generated,
such as OPENSSL_NO_DEPRECATED_1_1_1, corresponding to the use of
DEPRECATEDIN_ macros, such as DEPRECATEDIN_1_1_1().

Just like before, to set an API compatibility level when building an
application, define OPENSSL_API_COMPAT with an appropriate value.  If
it's desirable to hide deprecated functions up to and including that
level, additionally define OPENSSL_NO_DEPRECATED (the value is

Reviewed-by: Tim Hudson <>
(Merged from

9 months agoStrip much out of ssl.pod
Rich Salz [Fri, 1 Nov 2019 16:29:34 +0000 (12:29 -0400)]
Strip much out of ssl.pod

Reviewed-by: Matt Caswell <>
Reviewed-by: Tomas Mraz <>
(Merged from

9 months agoAdd L<ssl(7)> to all SSL pages
Rich Salz [Fri, 1 Nov 2019 16:27:29 +0000 (12:27 -0400)]
Add L<ssl(7)> to all SSL pages

Reviewed-by: Matt Caswell <>
Reviewed-by: Tomas Mraz <>
(Merged from

9 months agoRemove outdated info from man7/ssl
Rich Salz [Thu, 17 Oct 2019 20:06:15 +0000 (16:06 -0400)]
Remove outdated info from man7/ssl

There were two paragraphs of useful information about SSL_dup, so
copy that to the right manpage.

Reviewed-by: Matt Caswell <>
Reviewed-by: Tomas Mraz <>
(Merged from

9 months agoTest EVP_get_[digest|cipher]byname() use the namemap
Matt Caswell [Fri, 1 Nov 2019 16:09:21 +0000 (16:09 +0000)]
Test EVP_get_[digest|cipher]byname() use the namemap

Following on from the previous commit, we test that if an algorithm has
a provider supplied alias in the namemap then EVP_get_digestbyname() and
EVP_get_cipherbyname() can still find it.

Reviewed-by: Tomas Mraz <>
(Merged from

9 months agoEnsure EVP_get_digestbyname() and EVP_get_cipherbyname() know all aliases
Matt Caswell [Fri, 1 Nov 2019 14:13:49 +0000 (14:13 +0000)]
Ensure EVP_get_digestbyname() and EVP_get_cipherbyname() know all aliases

Now that we have an EVP namemap containing all aliases that providers
know about for any given algorithm, it is possible that an application
attempts to look up a digest or a cipher via EVP_get_digestbyname() or
EVP_get_cipherbyname() with an algorithm name that is unknown to the
legacy method database. Therefore we extend those functions to
additionally check the aliases in the namemap when searching for a
method in the event that our initial lookup attempt fails.

Reviewed-by: Tomas Mraz <>
(Merged from

9 months agotest/recipes/02-test_ordinals.t: Take '?' and '?+' into account
Richard Levitte [Mon, 4 Nov 2019 10:41:01 +0000 (11:41 +0100)]
test/recipes/02-test_ordinals.t: Take '?' and '?+' into account

Reviewed-by: Matt Caswell <>
Reviewed-by: Paul Dale <>
(Merged from

9 months agoutil/ output stats on unassigned symbols
Richard Levitte [Mon, 4 Nov 2019 09:36:54 +0000 (10:36 +0100)]
util/ output stats on unassigned symbols

Reviewed-by: Matt Caswell <>
Reviewed-by: Paul Dale <>
(Merged from