openssl.git
5 years agoAdd a sanity check on the length of pkeyutl inputs
Matt Caswell [Thu, 17 May 2018 11:53:07 +0000 (12:53 +0100)]
Add a sanity check on the length of pkeyutl inputs

When signing or verifying a file using pkeyutl the input is supposed to
be a hash. Some algorithms sanity check the length of the input, while
others don't and silently truncate. To avoid accidents we check that the
length of the input looks sane.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6284)

5 years agoRevert "Support EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA"
Matt Caswell [Thu, 17 May 2018 10:18:01 +0000 (11:18 +0100)]
Revert "Support EVP_PKEY_sign() and EVP_PKEY_verify() for EdDSA"

This reverts commit a6f5b1163452a36da1c8378eab3f29ae234e5c7f.

The EVP_PKEY_sign() function is intended for pre-hashed input which is
not supported by our EdDSA implementation.

See the discussion in PR 5880

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6284)

5 years agoImprove compatibility of point and curve checks
Matt Caswell [Mon, 21 May 2018 14:24:56 +0000 (15:24 +0100)]
Improve compatibility of point and curve checks

We check that the curve name associated with the point is the same as that
for the curve.

Fixes #6302

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6323)

5 years agoSave and restore the Windows error around TlsGetValue.
David Benjamin [Sun, 20 May 2018 21:24:30 +0000 (17:24 -0400)]
Save and restore the Windows error around TlsGetValue.

TlsGetValue clears the last error even on success, so that callers may
distinguish it successfully returning NULL or failing. This error-mangling
behavior interferes with the caller's use of GetLastError. In particular
SSL_get_error queries the error queue to determine whether the caller should
look at the OS's errors. To avoid destroying state, save and restore the
Windows error.

Fixes #6299.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6316)

5 years agoUse OPENSSL_EC_EXPLICIT_CURVE constant.
David Benjamin [Sun, 20 May 2018 18:37:06 +0000 (14:37 -0400)]
Use OPENSSL_EC_EXPLICIT_CURVE constant.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6314)

5 years agoFix explicit EC curve encoding.
David Benjamin [Sun, 20 May 2018 18:33:49 +0000 (14:33 -0400)]
Fix explicit EC curve encoding.

Per SEC 1, the curve coefficients must be padded up to size. See C.2's
definition of Curve, C.1's definition of FieldElement, and 2.3.5's definition
of how to encode the field elements in http://www.secg.org/sec1-v2.pdf.

This comes up for P-521, where b needs a leading zero.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6314)

5 years agoSkip CN DNS name constraint checks when not needed
Viktor Dukhovni [Tue, 22 May 2018 05:09:25 +0000 (01:09 -0400)]
Skip CN DNS name constraint checks when not needed

Only check the CN against DNS name contraints if the
`X509_CHECK_FLAG_NEVER_CHECK_SUBJECT` flag is not set, and either the
certificate has no DNS subject alternative names or the
`X509_CHECK_FLAG_ALWAYS_CHECK_SUBJECT` flag is set.

Add pertinent documentation, and touch up some stale text about
name checks and DANE.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
5 years agoLimit scope of CN name constraints
Viktor Dukhovni [Wed, 16 May 2018 03:41:20 +0000 (23:41 -0400)]
Limit scope of CN name constraints

Don't apply DNS name constraints to the subject CN when there's a
least one DNS-ID subjectAlternativeName.

Don't apply DNS name constraints to subject CN's that are sufficiently
unlike DNS names.  Checked name must have at least two labels, with
all labels non-empty, no trailing '.' and all hyphens must be
internal in each label.  In addition to the usual LDH characters,
we also allow "_", since some sites use these for hostnames despite
all the standards.

Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
5 years agoUse the client app traffic secret for PHA Finished message
Matt Caswell [Fri, 18 May 2018 16:33:19 +0000 (17:33 +0100)]
Use the client app traffic secret for PHA Finished message

The TLSv1.3 spec requires us to use the client application traffic secret
during generation of the Finished message following a post handshake
authentication.

Fixes #6263

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/6297)

5 years agoINSTALL: Provide better documentation for enable-ec_nistp_64_gcc_128
Richard Levitte [Tue, 22 May 2018 21:09:01 +0000 (23:09 +0200)]
INSTALL: Provide better documentation for enable-ec_nistp_64_gcc_128

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6328)

5 years agoBetter error code when lacking __SIZEOF_INT128__
Richard Levitte [Tue, 22 May 2018 11:57:29 +0000 (13:57 +0200)]
Better error code when lacking __SIZEOF_INT128__

Fixes #6327

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6328)

5 years agoFix no-ec, no-tls1_3 and no-tls
Matt Caswell [Mon, 21 May 2018 09:46:58 +0000 (10:46 +0100)]
Fix no-ec, no-tls1_3 and no-tls

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6321)

5 years agoEnable SSL_MODE_AUTO_RETRY by default
Kurt Roeckx [Tue, 15 May 2018 17:01:41 +0000 (19:01 +0200)]
Enable SSL_MODE_AUTO_RETRY by default

Because TLS 1.3 sends more non-application data records some clients run
into problems because they don't expect SSL_read() to return and set
SSL_ERROR_WANT_READ after processing it.

This can cause problems for clients that use blocking I/O and use
select() to see if data is available. It can be cleared using
SSL_CTX_clear_mode().

Reviewed-by: Matt Caswell <matt@openssl.org>
GH: #6260

5 years agoFix no-ec in combination with no-dh
Matt Caswell [Mon, 21 May 2018 11:20:18 +0000 (12:20 +0100)]
Fix no-ec in combination with no-dh

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6321)

5 years agossl/ssl_txt: fix NULL-check
Tilman Keskinöz [Thu, 17 May 2018 11:04:31 +0000 (13:04 +0200)]
ssl/ssl_txt: fix NULL-check

NULL-check for cipher is redundant, instead check if cipher->name is NULL

While here fix formatting of BIO_printf calls as suggested by Andy Polyakov.

CLA: trivial

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6282)

5 years agoapps/s_socket.c: address rare TLSProxy failures on Windows.
Andy Polyakov [Sun, 20 May 2018 21:03:47 +0000 (23:03 +0200)]
apps/s_socket.c: address rare TLSProxy failures on Windows.

Reviewed-by: Rich Salz <rsalz@openssl.org>
5 years agoCheck malloc failure via app_malloc
Rich Salz [Mon, 21 May 2018 14:28:16 +0000 (10:28 -0400)]
Check malloc failure via app_malloc

Thanks to GitHUb user murugesandins for reporting this.

Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6322)

5 years agoDon't cache stateless tickets in TLSv1.3
Matt Caswell [Fri, 18 May 2018 10:31:31 +0000 (11:31 +0100)]
Don't cache stateless tickets in TLSv1.3

In TLSv1.2 and below we always cache new sessions by default on the server
side in the internal cache (even when we're using session tickets). This is
in order to support resumption from a session id.

In TLSv1.3 there is no session id. It is only possible to resume using the
ticket. Therefore, in the default case,  there is no point in caching the
session in the internal store.

There is still a reason to call the external cache new session callback
because applications may be using the callbacks just to know about when
sessions are created (and not necessarily implementing a full cache). If
the application also implements the remove session callback then we are
forced to also store it in the internal cache so that we can create
timeout events. Otherwise the external cache could just fill up
indefinitely.

This mostly addresses the issue described in #5628. That issue also proposes
having an option to not create full stateless tickets when using the
internal cache. That aspect hasn't been addressed yet.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/6293)

5 years agoFix undefined behaviour in X509_NAME_cmp()
Matt Caswell [Wed, 16 May 2018 10:59:47 +0000 (11:59 +0100)]
Fix undefined behaviour in X509_NAME_cmp()

If the lengths of both names is 0 then don't attempt to do a memcmp.

Issue reported by Simon Friedberger, Robert Merget and Juraj Somorovsky.

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/6291)

5 years agoWindows: don't install __DECC_*.H
Richard Levitte [Sat, 19 May 2018 05:09:19 +0000 (07:09 +0200)]
Windows: don't install __DECC_*.H

This adds the possibility to exclude files by regexp in util/copy.pl

Partial fix for #3254

Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6303)

5 years agoQuiet pod2html warnings
Richard Levitte [Sat, 19 May 2018 05:22:10 +0000 (07:22 +0200)]
Quiet pod2html warnings

--quiet stops warnings of this sort:

    Cannot find "BIO_read_ex" in podpath: cannot find suitable replacement path, cannot resolve link

We know what causes these warnings, it's perfectly innocuous, and we
don't want to hear it any more.

Partial fix for #3254

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6304)

5 years agoFix no-srtp build warnings
Todd Short [Mon, 14 May 2018 14:41:06 +0000 (10:41 -0400)]
Fix no-srtp build warnings

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6255)

5 years agoRestore check of |*xn| against |name| in X509_NAME_set
Richard Levitte [Thu, 17 May 2018 07:53:14 +0000 (09:53 +0200)]
Restore check of |*xn| against |name| in X509_NAME_set

A previous change of this function introduced a fragility when the
destination happens to be the same as the source.  Such alias isn't
recommended, but could still happen, for example in this kind of code:

    X509_NAME *subject = X509_get_issuer_name(x);

    /* ... some code passes ... */

    X509_set_issuer_name(x, subject);

Fixes #4710

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6280)

5 years agoAllow the ca application to use EdDSA
Matt Caswell [Thu, 17 May 2018 15:24:29 +0000 (16:24 +0100)]
Allow the ca application to use EdDSA

Using the ca application to sign certificates with EdDSA failed because it
is not possible to set the digest to "null". This adds the capability and
updates the documentation accordingly.

Fixes #6201

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6286)

5 years agoDH: add some basic tests (and comments)
Dr. Matthias St. Pierre [Thu, 17 May 2018 22:51:15 +0000 (00:51 +0200)]
DH: add some basic tests (and comments)

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6273)

5 years agoDH: add simple getters for commonly used DH struct members
Dr. Matthias St. Pierre [Wed, 16 May 2018 14:18:13 +0000 (16:18 +0200)]
DH: add simple getters for commonly used DH struct members

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6273)

5 years agoMake BN_GF2m_mod_arr more constant time
Matt Caswell [Wed, 16 May 2018 08:58:27 +0000 (09:58 +0100)]
Make BN_GF2m_mod_arr more constant time

Experiments have shown that the lookup table used by BN_GF2m_mod_arr
introduces sufficient timing signal to recover the private key for an
attacker with access to cache timing information on the victim's host.
This only affects binary curves (which are less frequently used).

No CVE is considered necessary for this issue.

The fix is to replace the lookup table with an on-the-fly calculation of
the value from the table instead, which can be performed in constant time.

Thanks to Youngjoo Shin for reporting this issue.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6270)

5 years agoImprove testing of tickets with post-handshake auth
Matt Caswell [Fri, 11 May 2018 16:47:27 +0000 (17:47 +0100)]
Improve testing of tickets with post-handshake auth

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5227)

5 years agoAdd documentation for the ability to control the number of tickets
Matt Caswell [Fri, 16 Mar 2018 18:03:52 +0000 (18:03 +0000)]
Add documentation for the ability to control the number of tickets

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5227)

5 years agoChange the default number of NewSessionTickets we send to 2
Matt Caswell [Wed, 14 Mar 2018 19:22:48 +0000 (19:22 +0000)]
Change the default number of NewSessionTickets we send to 2

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5227)

5 years agoAllow configuation of the number of TLSv1.3 session tickets via SSL_CONF
Matt Caswell [Wed, 31 Jan 2018 16:40:03 +0000 (16:40 +0000)]
Allow configuation of the number of TLSv1.3 session tickets via SSL_CONF

Also allows the apps to set it.

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5227)

5 years agoEnable the ability to set the number of TLSv1.3 session tickets sent
Matt Caswell [Wed, 31 Jan 2018 17:26:46 +0000 (17:26 +0000)]
Enable the ability to set the number of TLSv1.3 session tickets sent

We send a session ticket automatically in TLSv1.3 at the end of the
handshake. This commit provides the ability to set how many tickets should
be sent. By default this is one.

Fixes #4978

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/5227)

5 years agoAdd a note on CHANGES and NEWS in CONTRIBUTING
Richard Levitte [Mon, 14 May 2018 03:38:59 +0000 (05:38 +0200)]
Add a note on CHANGES and NEWS in CONTRIBUTING

Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/6249)

5 years agoWhen producing man-pages, ensure NAME section is one line only
Richard Levitte [Wed, 16 May 2018 09:12:21 +0000 (11:12 +0200)]
When producing man-pages, ensure NAME section is one line only

There are *roff parsers that are strict about the NAME section being
one line only.  The man(7) on Debian GNU/Linux suggests that this is
appropriate, so we compensate our multi-line NAME sections by fixing
the *roff output.

Noted by Eric S. Raymond

Related to #6264

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6268)

5 years agoNOTES.ANDROID: fix typo in build notes
Gregor Jasny [Sun, 13 May 2018 17:51:52 +0000 (19:51 +0200)]
NOTES.ANDROID: fix typo in build notes

CLA: trivial

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6244)

5 years agowindows-makefile.tmpl: delete export library prior link.
Andy Polyakov [Mon, 14 May 2018 15:06:04 +0000 (17:06 +0200)]
windows-makefile.tmpl: delete export library prior link.

LINK can outsmart itself and choose to not update export .lib upon
corresponding .dll re-link. Since dependency is between .lib and all
.obj-s, re-compilation of any .obj makes NMAKE relink .dll and all
.exe-s over and over...

Reviewed-by: Rich Salz <rsalz@openssl.org>
5 years agoSuport TLSv1.3 draft 28
Matt Caswell [Mon, 14 May 2018 17:35:30 +0000 (18:35 +0100)]
Suport TLSv1.3 draft 28

Also retains support for drafts 27 and 26

Fixes #6257

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6258)

5 years agoFix no-tls1_2
Matt Caswell [Mon, 14 May 2018 13:37:16 +0000 (14:37 +0100)]
Fix no-tls1_2

Also fixes no-tls1_2-method, no-tls1_3, no-tls, no-ec

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6253)

5 years agoFix no-psk
Matt Caswell [Mon, 14 May 2018 13:28:06 +0000 (14:28 +0100)]
Fix no-psk

Fixes #6239

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6253)

5 years agoCI config: no need to make both install and install_docs
Richard Levitte [Mon, 14 May 2018 07:28:52 +0000 (09:28 +0200)]
CI config: no need to make both install and install_docs

'install' depends on 'install_docs', so making the latter explicit is
a waste.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6250)

5 years agoLink in passphrase-encoding(7) in relevant documentation
Richard Levitte [Sun, 13 May 2018 09:35:14 +0000 (11:35 +0200)]
Link in passphrase-encoding(7) in relevant documentation

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6179)

5 years agoDocs: add general document on how pass phrases are handled
Richard Levitte [Fri, 4 May 2018 22:06:07 +0000 (00:06 +0200)]
Docs: add general document on how pass phrases are handled

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6179)

5 years agoUI console: Restore tty settings, do not force ECHO after prompt
Richard Levitte [Wed, 2 May 2018 12:18:55 +0000 (14:18 +0200)]
UI console: Restore tty settings, do not force ECHO after prompt

The Console UI method always set echo on after prompting without
echo.  However, echo might not have been on originally, so just
restore the original TTY settings.

Fixes #2373

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6156)

5 years agoapps/speed: Add brainpool curves support
FdaSilvaYY [Tue, 8 May 2018 08:34:59 +0000 (10:34 +0200)]
apps/speed: Add brainpool curves support

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6133)

5 years agoapps/speed: fix possible OOB access in some EC arrays
FdaSilvaYY [Sun, 29 Apr 2018 23:13:58 +0000 (01:13 +0200)]
apps/speed: fix possible OOB access in some EC arrays

because there are actually 18 curves defined, but only 16 are plugged for
ecdsa test.
Deduce array size using OSSL_NELEM and so remove various magic numbers,
which required some declarations moving.
Implement OPT_PAIR list search without a null-ending element.
Fix some comparison between signed and unsigned integer expressions.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6133)

5 years agoSet sess to NULL after freeing it.
Kurt Roeckx [Sat, 12 May 2018 10:08:14 +0000 (12:08 +0200)]
Set sess to NULL after freeing it.

Found by OSS-fuzz

Bug introduced in commit 61fb59238dad6452a37ec14513fae617a4faef29

Reviewed-by: Matt Caswell <matt@openssl.org>
GH: #6235

5 years agoDon't memcpy the contents of an empty fragment
Matt Caswell [Fri, 11 May 2018 09:28:47 +0000 (10:28 +0100)]
Don't memcpy the contents of an empty fragment

In DTLS if we have buffered a fragment for a zero length message (e.g.
ServerHelloDone) then, when we unbuffered the fragment, we were attempting
to memcpy the contents of the fragment which is zero length and a NULL
pointer. This is undefined behaviour. We should check first whether we
have a zero length fragment.

Fixes a travis issue.

[extended tests]

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6223)

5 years agoIn cases where we ask PEM_def_callback for minimum 0 length, accept 0 length
Richard Levitte [Fri, 4 May 2018 12:19:44 +0000 (14:19 +0200)]
In cases where we ask PEM_def_callback for minimum 0 length, accept 0 length

Fixes #4716

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6173)

5 years agoFix typo: 'is an error occurred' in documentation
Dr. Matthias St. Pierre [Fri, 11 May 2018 14:54:43 +0000 (16:54 +0200)]
Fix typo: 'is an error occurred' in documentation

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6229)

5 years agoRework the decrypt ticket callback
Matt Caswell [Wed, 9 May 2018 17:22:36 +0000 (18:22 +0100)]
Rework the decrypt ticket callback

Don't call the decrypt ticket callback if we've already encountered a
fatal error. Do call it if we have an empty ticket present.

Change the return code to have 5 distinct returns codes and separate it
from the input status value.

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6198)

5 years agoFix mem leak in sslapi test
Matt Caswell [Wed, 9 May 2018 09:45:46 +0000 (10:45 +0100)]
Fix mem leak in sslapi test

The recent change in behaviour where you do not get a NewSessionTicket
message sent if you established the connection using a PSK caused a mem
leak to be triggered in sslapitest. It was actually a latent bug and we
were just lucky we never hit it before. The problem is due to complexity
with the way PSK sessions were set up in the early_data tests. PSK session
reference counting was handled differently to normal session reference
counting. This meant there were lots of special cases in the code where
we don't free a session if it is a PSK. It makes things easier if we just
handle PSK reference counts in the same way as other session reference
counts, and then we can remove all of the special case code.

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6198)

5 years agoAdd a test for the ticket callbacks
Matt Caswell [Tue, 8 May 2018 15:29:02 +0000 (16:29 +0100)]
Add a test for the ticket callbacks

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6198)

5 years agoDocument when a new session ticket gets created on resumption
Matt Caswell [Tue, 8 May 2018 15:28:44 +0000 (16:28 +0100)]
Document when a new session ticket gets created on resumption

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6198)

5 years agoFix ticket callbacks in TLSv1.3
Matt Caswell [Tue, 8 May 2018 13:50:17 +0000 (14:50 +0100)]
Fix ticket callbacks in TLSv1.3

The return value from the ticket_key callback was not properly handled in
TLSv1.3, so that a ticket was *always* renewed even if the callback
requested that it should not be.

Also the ticket decrypt callback was not being called at all in TLSv1.3.

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6198)

5 years agoFlush server side unauthenticated writes
Matt Caswell [Tue, 8 May 2018 13:34:27 +0000 (14:34 +0100)]
Flush server side unauthenticated writes

When a server call SSL_write_early_data() to write to an unauthenticated
client the buffering BIO is still in place, so we should ensure we flush
the write.

Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6198)

5 years agoAdd some more SSL_pending() and SSL_has_pending() tests
Matt Caswell [Wed, 2 May 2018 15:08:07 +0000 (16:08 +0100)]
Add some more SSL_pending() and SSL_has_pending() tests

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6159)

5 years agoDon't set TCP_NODELAY on a UDP socket
Matt Caswell [Wed, 2 May 2018 15:08:27 +0000 (16:08 +0100)]
Don't set TCP_NODELAY on a UDP socket

This was preventing DTLS connections from being made from the command line.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6159)

5 years agoMark DTLS records as read when we have finished with them
Matt Caswell [Wed, 2 May 2018 15:07:13 +0000 (16:07 +0100)]
Mark DTLS records as read when we have finished with them

The TLS code marks records as read when its finished using a record. The DTLS code did
not do that. However SSL_has_pending() relies on it. So we should make DTLS consistent.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6159)

5 years agoTest an old style PSK callback with no cert will prefer SHA-256
Matt Caswell [Thu, 10 May 2018 11:33:51 +0000 (12:33 +0100)]
Test an old style PSK callback with no cert will prefer SHA-256

If using an old style PSK callback and no certificate is configured for
the server, we should prefer ciphersuites based on SHA-256, because that
is the default hash for those callbacks as specified in the TLSv1.3 spec.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6215)

5 years agoProvide documentation for the -psk_session option
Matt Caswell [Thu, 10 May 2018 11:01:06 +0000 (12:01 +0100)]
Provide documentation for the -psk_session option

The s_client/s_server docs were missing documentation for this option.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6215)

5 years agoPrefer SHA-256 ciphersuites if using old style PSKs
Matt Caswell [Thu, 10 May 2018 10:51:45 +0000 (11:51 +0100)]
Prefer SHA-256 ciphersuites if using old style PSKs

If we have no certificate and we are using "old style" PSKs then we will
always default to using SHA-256 for that PSK. However we may have selected
a ciphersuite that is not based on SHA-256. Therefore if we see that there
are no certificates and we have been configured for "old style" PSKs then
we should prefer SHA-256 based ciphersuites during the selection process.

Fixes #6197

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6215)

5 years agoUse void in all function definitions that do not take any arguments
Kurt Roeckx [Wed, 9 May 2018 15:09:50 +0000 (17:09 +0200)]
Use void in all function definitions that do not take any arguments

Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #6208

5 years agorsaz_avx2_eligible doesn't take parameters
Kurt Roeckx [Wed, 9 May 2018 16:32:02 +0000 (18:32 +0200)]
rsaz_avx2_eligible doesn't take parameters

Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #6208

5 years agoSet the ossl_shim to auto retry if not running asynchronously
Matt Caswell [Thu, 10 May 2018 14:31:00 +0000 (15:31 +0100)]
Set the ossl_shim to auto retry if not running asynchronously

In certain circumstances in the DTLS code we have to drop a record (e.g. if
it is a stale retransmit). We then have to move on to try and read the next
record. Some applications using blocking sockets (e.g. s_server/s_client
will hang if there isn't actually any data to be read from the socket yet).
Others can tolerate this. Therefore SSL_read()/SSL_write() can sometimes
return SSL_ERROR_WANT_READ/SSL_ERROR_WANT_WRITE even when using blocking
sockets. Applications can use the mode SSL_MODE_AUTO_RETRY, to switch this
behaviour off so that we never return unless we have read the data we
wanted to.

Commit ad96225285 fixed a DTLS problem where we always retried even if
SSL_MODE_AUTO_RETRY was not set. However that fix caused the Boring
ossl_shim to fail in some tests because it was relying on the previous
(buggy) behaviour. The ossl_shim should be set into SSL_MODE_AUTO_RETRY if
it is not operating asynchronously to avoid this problem.

[extended tests]

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6216)

5 years agoPPC assembly pack: add POWER9 results.
Andy Polyakov [Wed, 9 May 2018 10:24:05 +0000 (12:24 +0200)]
PPC assembly pack: add POWER9 results.

Reviewed-by: Rich Salz <rsalz@openssl.org>
5 years ago.travis.yml: add pair of linux-ppc64le targets.
Andy Polyakov [Mon, 7 May 2018 19:56:04 +0000 (21:56 +0200)]
.travis.yml: add pair of linux-ppc64le targets.

One is clang --strict-warnings and one gcc sanitizer extended test.
Sanitizer build is quite expensive, can take >30 mins and is commented
for occasions when there is reason to believe that PPC-specific problem
can be diagnosed with sanitizer.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6192)

5 years agoFix no-cms
Matt Caswell [Wed, 9 May 2018 11:05:39 +0000 (12:05 +0100)]
Fix no-cms

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6205)

5 years agoFix no-tls1_2, no-tls1_2-method, no-chacha and no-poly1305
Matt Caswell [Wed, 9 May 2018 10:49:02 +0000 (11:49 +0100)]
Fix no-tls1_2, no-tls1_2-method, no-chacha and no-poly1305

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6205)

5 years agoFix typos in x509 documentation
Dr. Matthias St. Pierre [Tue, 8 May 2018 10:32:12 +0000 (12:32 +0200)]
Fix typos in x509 documentation

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6207)

5 years agoAdd CHANGES entry for PR#6009
Nicola Tuveri [Wed, 25 Apr 2018 12:27:59 +0000 (15:27 +0300)]
Add CHANGES entry for PR#6009

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6070)

5 years agoAdd blinding in BN_GF2m_mod_inv for binary field inversions
Billy Brumley [Tue, 24 Apr 2018 13:03:42 +0000 (16:03 +0300)]
Add blinding in BN_GF2m_mod_inv for binary field inversions

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6070)

5 years agoECC: unify generic ec2 and ecp scalar multiplication, deprecate ec2_mult.c
Billy Brumley [Tue, 24 Apr 2018 13:01:53 +0000 (16:01 +0300)]
ECC: unify generic ec2 and ecp scalar multiplication, deprecate ec2_mult.c

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6070)

5 years agoECDSA: remove nonce padding (delegated to EC_POINT_mul)
Billy Brumley [Tue, 24 Apr 2018 13:00:08 +0000 (16:00 +0300)]
ECDSA: remove nonce padding (delegated to EC_POINT_mul)

* EC_POINT_mul is now responsible for constant time point multiplication
  (for single fixed or variable point multiplication, when the scalar is
  in the range [0,group_order), so we need to strip the nonce padding
  from ECDSA.
* Entry added to CHANGES
* Updated EC_POINT_mul documentation
  - Integrate existing EC_POINT_mul and EC_POINTs_mul entries in the
    manpage to reflect the shift in constant-time expectations when
    performing a single fixed or variable point multiplication;
  - Add documentation to ec_method_st to reflect the updated "contract"
    between callers and implementations of ec_method_st.mul.

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6070)

5 years agoVMS rand: assign before check, not the other way around
Richard Levitte [Tue, 8 May 2018 18:15:27 +0000 (20:15 +0200)]
VMS rand: assign before check, not the other way around

items->ile3$w_code was checked before it was assigned its value...

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6200)

5 years agoFix --strict-warnings build of ppc-linux target
Bernd Edlinger [Mon, 7 May 2018 14:10:02 +0000 (16:10 +0200)]
Fix --strict-warnings build of ppc-linux target

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6191)

5 years agoec/ec_mult.c: get BN_CTX_start,end sequence right.
Andy Polyakov [Mon, 7 May 2018 08:27:45 +0000 (10:27 +0200)]
ec/ec_mult.c: get BN_CTX_start,end sequence right.

Triggered by Coverity analysis.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6190)

5 years agoAdd a DTLS test for dropped records
Matt Caswell [Thu, 3 May 2018 11:06:38 +0000 (12:06 +0100)]
Add a DTLS test for dropped records

Drop a record from a handshake and check that we can still complete the
handshake. Repeat for all records in the handshake.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6170)

5 years agoKeep the DTLS timer running after the end of the handshake if appropriate
Matt Caswell [Thu, 3 May 2018 15:00:51 +0000 (16:00 +0100)]
Keep the DTLS timer running after the end of the handshake if appropriate

During a full handshake the server is the last one to "speak". The timer
should continue to run until we know that the client has received our last
flight (e.g. because we receive some application data).

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6170)

5 years agoOnly auto-retry for DTLS if configured to do so
Matt Caswell [Thu, 3 May 2018 15:00:05 +0000 (16:00 +0100)]
Only auto-retry for DTLS if configured to do so

Otherwise we may end up in a hang when using blocking sockets

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6170)

5 years agoFix s_client and s_server so that they correctly handle the DTLS timer
Matt Caswell [Thu, 3 May 2018 14:59:31 +0000 (15:59 +0100)]
Fix s_client and s_server so that they correctly handle the DTLS timer

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6170)

5 years agoDon't fail on an out-of-order CCS in DTLS
Matt Caswell [Thu, 3 May 2018 11:07:47 +0000 (12:07 +0100)]
Don't fail on an out-of-order CCS in DTLS

Fixes #4929

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6170)

5 years agoAdd a CMS API test
Matt Caswell [Tue, 1 May 2018 08:32:30 +0000 (09:32 +0100)]
Add a CMS API test

Previous tests only invoked CMS via the command line app. This test uses
the CMS API directly to do and encrypt and decrypt operation. This test
would have caught the memory leak fixed by the previous commit (when
building with enable-crypto-mdebug).

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6142)

5 years agoFix a mem leak in CMS
Matt Caswell [Tue, 1 May 2018 08:29:17 +0000 (09:29 +0100)]
Fix a mem leak in CMS

The function CMS_RecipientInfo_set0_pkey() is a "set0" and therefore
memory management passes to OpenSSL. If the same function is called again
then we should ensure that any previous value that was set is freed first
before we set it again.

Fixes #5052

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6142)

5 years agowindows-makefile.tmpl: rearrange cleanup commands to avoid ...
FdaSilvaYY [Sun, 6 May 2018 15:34:04 +0000 (17:34 +0200)]
windows-makefile.tmpl: rearrange cleanup commands to avoid ...

deletion of *.exp files in krb5 sub-module.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6186)

5 years ago.travis.yml: temporarily mask gcc-5 ubsan build.
Andy Polyakov [Sun, 6 May 2018 16:36:09 +0000 (18:36 +0200)]
.travis.yml: temporarily mask gcc-5 ubsan build.

Linking fails with "unrecognized option '--push-state--no-as-needed'",
which is beyond our control.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6185)

5 years ago.travis.yml: minor facelift
Andy Polyakov [Sun, 6 May 2018 16:30:59 +0000 (18:30 +0200)]
.travis.yml: minor facelift

Apparently trusty image has newer clang, there is no need to pull
clang-3.9 packages. It's clang-5.0.0, installation is a bit quirky,
as it fails to compile for example strcmp(s,"-") without warning,
and complains about unreferred -I flags. But it's argued that benefits
of exercising newer sanitizer outweights the inconvenience of
additional -D__NO_STRING_INLINE and -Wno-unused-command-line-argument.

Also pull golang when actually needed.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6185)

5 years agoCleanup
Rich Salz [Sat, 5 May 2018 19:57:21 +0000 (15:57 -0400)]
Cleanup

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6183)

5 years agoConfigure: fix Mac OS X builds that still require makedepend
Todd Short [Thu, 3 May 2018 15:17:49 +0000 (11:17 -0400)]
Configure: fix Mac OS X builds that still require makedepend

Earlier Apple Xcode compilers, e.g. one targeting Mac OS X 10.7, don't
support dependency generation and one still has to use makedepend.  It's
unclear when it was fixed, but all clang-based Apple compilers seem to
support -M options.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6169)

5 years agoConfigure: pass more suitable argument to compiler_predefined().
Andy Polyakov [Fri, 4 May 2018 12:25:45 +0000 (14:25 +0200)]
Configure: pass more suitable argument to compiler_predefined().

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6174)

5 years agoConfigure: move --noexecstack probe to Configure.
Andy Polyakov [Fri, 4 May 2018 12:06:44 +0000 (14:06 +0200)]
Configure: move --noexecstack probe to Configure.

config probe doesn't work in cross-compile scenarios or with clang.
In addition consolidate -Qunused-arguments handling.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6174)

5 years agoBIO_s_mem() write: Skip early when input length is zero
Richard Levitte [Fri, 4 May 2018 12:44:19 +0000 (14:44 +0200)]
BIO_s_mem() write: Skip early when input length is zero

When the input length is zero, just return zero early.  Otherwise,
there's a small chance that memory allocation is engaged, fails and
returns -1, which is a bit confusing when nothing should be written.

Fixes #4782 #4827

Reviewed-by: Ben Kaduk <kaduk@mit.edu>
(Merged from https://github.com/openssl/openssl/pull/6175)

5 years agodocs: Fix typo EVP_PKEY_new_id -> EVP_PKEY_CTX_new_id
Richard Levitte [Fri, 4 May 2018 17:41:53 +0000 (19:41 +0200)]
docs: Fix typo EVP_PKEY_new_id -> EVP_PKEY_CTX_new_id

Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6177)

5 years agov3_purp.c: add locking to x509v3_cache_extensions()
Dr. Matthias St. Pierre [Wed, 2 May 2018 21:06:15 +0000 (23:06 +0200)]
v3_purp.c: add locking to x509v3_cache_extensions()

Fixes #6121

Thanks to Mingtao Yang for reporting this bug.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6162)

5 years agoVMS: modernise rand_pool_acquire_entropy, step 2
Richard Levitte [Wed, 2 May 2018 04:16:04 +0000 (06:16 +0200)]
VMS: modernise rand_pool_acquire_entropy, step 2

Add more items that could serve as entropy source.

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6151)

5 years agoVMS: modernise rand_pool_acquire_entropy, step 1
Richard Levitte [Mon, 30 Apr 2018 16:04:10 +0000 (18:04 +0200)]
VMS: modernise rand_pool_acquire_entropy, step 1

Stop redefining structures that are already defined in system
headers.  This also means we can stop setting the pointer size
globally, because the system structures will have the correct pointer
sizes either way.  The only exception is passing the right pointer
size to a function.

Stop trying to twist things around with rand(), that's the job of the
DRBG that we feed.

Stop assuming the location of the JPI$_FINALEXC item, look it up
instead.

Signal an exception if the sys$getjpiw call fails (it means the item
list isn't set up right, so works as an assertion, but using VMS
methodology).

Reviewed-by: Andy Polyakov <appro@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6151)

5 years agoReturn an error from BN_mod_inverse if n is 1 (or -1)
Matt Caswell [Fri, 27 Apr 2018 16:36:11 +0000 (17:36 +0100)]
Return an error from BN_mod_inverse if n is 1 (or -1)

Calculating BN_mod_inverse where n is 1 (or -1) doesn't make sense. We
should return an error in that case. Instead we were returning a valid
result with value 0.

Fixes #6004

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6119)

5 years agoMake X509_VERIFY_PARAM_get_hostflags() take a const arg
Matt Caswell [Wed, 2 May 2018 10:32:39 +0000 (11:32 +0100)]
Make X509_VERIFY_PARAM_get_hostflags() take a const arg

Commit 5b748dea5 added this function which should have taken a const
argument.

Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6154)

5 years agoAdd a test for SSL_get_shared_ciphers()
Matt Caswell [Fri, 27 Apr 2018 11:20:04 +0000 (12:20 +0100)]
Add a test for SSL_get_shared_ciphers()

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6113)

5 years agoFix a bug in create_ssl_ctx_pair()
Matt Caswell [Fri, 27 Apr 2018 11:09:08 +0000 (12:09 +0100)]
Fix a bug in create_ssl_ctx_pair()

The max protocol version was only being set on the server side. It should
have been done on both the client and the server.

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6113)

5 years agoAdd some documentation for SSL_get_shared_ciphers()
Matt Caswell [Fri, 27 Apr 2018 10:38:19 +0000 (11:38 +0100)]
Add some documentation for SSL_get_shared_ciphers()

Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/6113)