Ensure that global dependencies are built first, even for parallell builds

Also, make sure that per-file deps are included in the build, even if previous
builds have failed.

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

Use the SSL_METHODs passed to create_ssl_ctx_pair()

Don't use hardcoded SSL_METHODs!

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

Avoid an overflow in constructing the ServerKeyExchange message

We calculate the size required for the ServerKeyExchange message and then
call BUF_MEM_grow_clean() on the buffer. However we fail to take account of
2 bytes required for the signature algorithm and 2 bytes for the signature
length, i.e. we could overflow by 4 bytes. In reality this won't happen
because the buffer is pre-allocated to a large size that means it should be
big enough anyway.

Addresses an OCAP Audit issue.

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

Cleanup obj_dat.h, obj_dat.pl

The recent merge of https://github.com/openssl/openssl/pull/1264
removed some trailing whitespace from the generated file obj_dat.h.
Unfortunately obj_dat.pl kept re-adding it.  Clean up the
script and the output it generates.

Add 'use strict / use warnings'

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

Deal with pod2html issues, in this case the lack of .html suffix in links

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

Simplify INCLUDE statements in build.info files

Now that INCLUDE considers both the source and build trees, no need
for the rel2abs perl fragment hacks any more.

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

Make build.info INCLUDE stmts be both source and build tree relative

INCLUDE statements in build.info files were source tree centric.  That
meant that to get include directory specs in the build tree, we had to
resort to perl fragments that specified the build tree include paths
as absolute ones.

This change has the INCLUDE statement consider both the source and
build tree for any include directory.  It means that there may be some
extra unnecessary include paths, but it also makes life simpler for
anyone who makes changes in the build.info files.

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

fix 'set but not used' warning

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

Re-add x509 and crl fuzzer

Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1276

Add old locking constants back

Fixes #1260

Reviewed-by: Matt Caswell <matt@openssl.org>
GH: #1266

Make CRYPTO_num_locks() return 1

malloc(0) might return NULL and code for the old callbacks might fail,
instead just say they should allocate 1 entry.

Reviewed-by: Matt Caswell <matt@openssl.org>
GH: #1266

fix code formatting

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1223)

add return value for expand

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1223)

Fix printing private EC_KEY

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1273)

ssl.h: spelling in comment
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1275)

Run the fuzzing corpora as tests.

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

SPARC assembly pack: enforce V8+ ABI constraints.

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

SPARC assembly pack: enforce V8+ ABI constraints.

Even though it's hard to imagine, it turned out that upper half of
arguments passed to V8+ subroutine can be non-zero.

["n" pseudo-instructions, such as srln being srl in 32-bit case and
srlx in 64-bit one, were implemented in binutils 2.10. It's assumed
that Solaris assembler implemented it around same time, i.e. 2000.]

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

Windows: Recreate the $prefix variable

$prefix was removed as part of the DESTDIR work.  However, it was
still used to create the ENGINESDIR_dev and ENGINESDIR_dir variables,
so a restoration is needed.

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

Constify a bit more OPENSSL_sk_ API

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1244)

Use directly zalloc in OPENSSL_sk_dup and OPENSSL_sk_deep_copy

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1244)

Convert memset calls to OPENSSL_cleanse

Ensure things really do get cleared when we intend them to.

Addresses an OCAP Audit issue.

Reviewed-by: Andy Polyakov <appro@openssl.org>

Cleanup documentation for removed functionality

Remove some lingering references to removed functionality from docs.

Reviewed-by: Stephen Henson <steve@openssl.org>

Add some compat macros for removed tmp RSA functions/macros

Commit 361a119 removed all ciphersuites that could support temporary
RSA keys, therefore the associated functions were removed. We should have
"no-op" compatibility macros for these.

Reviewed-by: Stephen Henson <steve@openssl.org>

Fix hmac test case 6

Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1267)

Restore clang builds in travis.yml

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

Remove the envvar hack to enable proxy cert processing

When the proxy cert code was initially added, some application authors
wanted to get them verified without having to change their code, so a
check of the env var OPENSSL_ALLOW_PROXY_CERTS was added.

Since then, the use of this variable has become irrelevant, as it's
likely that code has been changed since, so it's time it gets removed.

Reviewed-by: Tim Hudson <tjh@openssl.org>

Whitespace cleanup in apps

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1264)

Whitespace cleanup in ssl folder

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1264)

Whitespace cleanup in crypto

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1264)

Whitespace cleanup in docs

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1264)

Spelling

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1264)

Some interfaces may have IPv6 addresses even if an IPv6 address is not
"configured on the local system". Whatever that means. Example that is biting
me is loopback has ::1 as an address, but the network interface is v4 only.

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

Add comment about X509_print

Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Emilia Kasper <emilia@openssl.org>
GH: #1255

SSL test framework: port SNI tests

Observe that the old tests were partly ill-defined:
setting sn_server1 but not sn_server2 in ssltest_old.c does not enable
the SNI callback.

Fix this, and also explicitly test both flavours of SNI mismatch (ignore
/ fatal alert). Tests still pass.

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

base64 macro: parenthesize for clarity

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

Windows: Make DESTDIR work

DESTDIR can't be used on Windows the same way as on Unix, the device
part of the installation paths get in the way.

To remedy this, have INSTALLTOP, OPENSSLDIR and ENGINESDIR get
different values depending on if $(DESTDIR) is empty or not, and use
$(INSTALLTOP), $(OPENSSLDIR) and $(ENGINESDIR) alone.

Reviewed-by: Andy Polyakov <appro@openssl.org>

Build files: don't generate empty action lines in generatedir()

Reviewed-by: Andy Polyakov <appro@openssl.org>

Use include paths to our source before any other cflags

This is just in case someone passed an inclusion path with the
configuration, and there are OpenSSL headers from another version
in there.

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

engines/e_capi.c: accommodate recent DSA_SIG_[get|set]0 changes.

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

rsa/rsa_lib.c: const-ify RSA_get0_engine().

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

unix-Makefile.tmpl: omit lib<rary>.a updates from directory targets.

Since corresponding rule was removed from windows-makefile.tmpl out
of necessity, question popped if it's appropriate to harmonize even
unix-Makefile.tmpl. Note that as long as you work on single directory
'make lib<rary>.a' is effectively equivalent to 'make <dir/ectory>'
prior this modification.

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

windows-makefile.tmpl: don't use $? in library targets.

Problem with Microsoft lib.exe is that it doesn't *update* modules
in .lib archive, but creates new one upon every invocation. As result
if a source file was updated and nmake was executed, a useless archive
with only one module was created. In other words one has to always
pass all .obj modules on command line, not only recently recompiled.

[This also creates dilemma for directory targets, e.g. crypto\aes,
that were added to simplify every-day life for developer. Since
whole idea behind those targets is to minimize the re-compile time
upon single file modification, the only sensible thing to do is to
omit intended library update.]

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

Add a test for fragmented alerts

The previous commit fixed a problem where fragmented alerts would cause an
infinite loop. This commit adds a test for these fragmented alerts.

Reviewed-by: Andy Polyakov <appro@openssl.org>

Ensure read records are marked as read

In some situations (such as when we receive a fragment of an alert)
we try to get the next packet but did not mark the current one as read,
meaning that we got the same record back again - leading to an infinite
loop.

Found using the BoringSSL test suite.

Reviewed-by: Andy Polyakov <appro@openssl.org>

RT4586: Remove RSA_memory_lock; unused, not needed

Reviewed-by: Ben Laurie <ben@openssl.org>

Fix one more instance of incorrect OPENSSL_API_COMPAT value

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

Fix some OPENSSL_API_COMPAT values

There are 3 OPENSSL_API_COMPAT values that are incorrect in the header
files, and one inconsistency between the header and the .c

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

Clarify use of |$end0| in stitched x86-64 AES-GCM code.

There was some uncertainty about what the code is doing with |$end0|
and whether it was necessary for |$len| to be a multiple of 16 or 96.
Hopefully these added comments make it clear that the code is correct
except for the caveat regarding low memory addresses.

Change-Id: Iea546a59dc7aeb400f50ac5d2d7b9cb88ace9027
Reviewed-on: https://boringssl-review.googlesource.com/7194
Reviewed-by: Adam Langley <agl@google.com>
Signed-off-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>

Configurations/10-main.conf: fix PA-RISC commentary.

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

evp/evp_enc.c: refine partial buffer overlap detection.

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

evp/evp_enc.c: check for partially[!] overlapping buffers
in EVP_EncryptUpdate and EVP_DecryptUpdate. It is argued that in
general case it's impossible to provide guarantee that partially[!]
overlapping buffers can be tolerated.

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

RT2680: Public EC key is shown as private

Re-implemented as suggested by Steve Henson.
Also change ECParameters_print the same way.

Reviewed-by: Dr. Stephen Henson <steve@openssl.org>

RT2964: Fix it via doc

OBJ_nid2obj() and friends should be treated as const.

Reviewed-by: Dr. Stephen Henson <steve@openssl.org>

Revert "RT2964: Fix it via doc"

This reverts commit 82f31fe4dd0dac30229fa8684229b49d2bcef404.

Reviewed-by: Dr. Stephen Henson <steve@openssl.org>

RT2964: Fix it via doc

OBJ_nid2obj() and friends should be treated as const.

Reviewed-by: Dr. Stephen Henson <steve@openssl.org>

Handle BN_mod_word failures.

As of 37258dadaa9e36db4b96a3aa54aa6c67136160cc and the corresponding upstream
change, BN_mod_word may fail, like BN_div_word. Handle this properly. Thanks to
Brian Smith for pointing this out. See BoringSSL's
44bedc348d9491e63c7ed1438db100a4b8a830be.

Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1251

Fix BN_is_prime* calls.

This function returns a tri-state -1 on error. See BoringSSL's
53409ee3d7595ed37da472bc73b010cd2c8a5ffd.

Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1251

Add x509 and crl corpora

Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1229

Add X509 and CRL fuzzer

Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1229

Add -ciphers flag to enc command

Don't print the full list of ciphers as part of the -help output.

Reviewed-by: Andy Polyakov <appro@openssl.org>

perlasm/x86_64-xlate.pl: address errors and warnings in elderly perls.

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

./config: minor cleanup.

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

./config: detect x32-only environment.

RT#4583

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

Teach EVP_PKEY_HMAC keys how to EVP_PKEY_cmp()

Fixes openssl/openssl#1236

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1237)

Avoid signed overflow

Found by afl

Reviewed-by: Rich Salz <rsalz@openssl.org>
MR: #3013

Deal with API changes.

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

Ensure HMAC key gets cleansed after use

aesni_cbc_hmac_sha256_ctrl() and aesni_cbc_hmac_sha1_ctrl() cleanse the
HMAC key after use, but static int rc4_hmac_md5_ctrl() doesn't.

Fixes an OCAP Audit issue.

Reviewed-by: Andy Polyakov <appro@openssl.org>

Add some documentation for missing HMAC functions

This includes the newly added HMAC_CTX_get_md().

Reviewed-by: Tim Hudson <tjh@openssl.org>

Add a getter to obtain the HMAC_CTX md

As a result of opaque HMAC_CTX apps need a getter for the HMAC_CTX md.

GitHub Issue #1152

Reviewed-by: Tim Hudson <tjh@openssl.org>

Fix ASN1_STRING_to_UTF8 could not convert NumericString

tag2nbyte had -1 at 18th position, but underlying ASN1_mbstring_copy
supports NumericString. tag2nbyte is also used in do_print_ex which will
not be broken by setting 1 at 18th position of tag2nbyte

Reviewed-by: Stephen Henson <steve@openssl.org>

Make x25519_compute_key() return a boolean

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>

Make ossl_ecdh_compute_key() return a boolean

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>

utils/mkdir-p: check if dir exists also after mkdir failed

with "make install -j8" it happens very often that two or more make
instances are creating the same directory in parallel. As a result one
instace creates the directory and second mkdir fails because the
directory exists already (but it did not while testing for it earlier).

Signed-off-by: Sebastian Andrzej Siewior <sebastian@breakpoint.cc>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1204)

RT2867: des_ede3_cfb1 ignored "size in bits" flag

Code and tests by Steve.

Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>

Drop extraneous printf argument in mkcert.sh

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

Update fuzz corpora

Reviewed-by: Rich Salz <rsalz@openssl.org>
MR: #2986

Add checks on sk_TYPE_push() returned result

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>

Rework error handling from asn1_do_lock method.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>

Fix generation of expired CA certificate.

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

OpenSSL::Test: Fix directory calculations in __cwd()

The previous fix contained a mistake, where any absolute path in
%directories would be cleared away instead of just being left alone.

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

rand/randfile.c: make it non-ASCII-savvy.

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

Move OS-specific fopen quirks to o_fopen.c.

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

crypto/cryptlib.c: omit OPENSSL_ia32cap_loc().

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

Spelling... and more spelling

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1245)

segmentation fault with 'openssl s_client -prexit -keymatexport'

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1243)

Avoid creating an illegal pointer.

Found by tis-interpreter

Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1230

Make RSA key exchange code actually constant-time.

Using RSA_PKCS1_PADDING with RSA_private_decrypt is inherently unsafe.
The API requires writing output on success and touching the error queue
on error. Thus, although the padding check itself is constant-time as of
294d1e36c2495ff00e697c9ff622856d3114f14f, and the logic after the
decryption in the SSL code is constant-time as of
adb46dbc6dd7347750df2468c93e8c34bcb93a4b, the API boundary in the middle
still leaks whether the padding check succeeded, giving us our
much-loved Bleichenbacher padding oracle.

Instead, PKCS#1 padding must be handled by the caller which uses
RSA_NO_PADDING, in timing-sensitive code integrated with the
Bleichenbacher mitigation. Removing PKCS#1 padding in constant time is
actually much simpler when the expected length is a constant (and if
it's not a constant, avoiding a padding oracle seems unlikely), so just
do it inline.

Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1222

buf2hexstr: properly deal with empty string

It wrote before the start of the string

found by afl

Reviewed-by: Richard Levitte <levitte@openssl.org>
MR: #2994

Fix typo

Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1241)

Appveyor: test install as well, via a fake deploy_script

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

Travis: When testing install, install docs as well

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

OpenSSL::Test: Fix directory calculations in __cwd()

We recalculate the location of the directories we keep track of.
However, we did so after having moved to the new directory already, so
the data we did the calculations from were possibly not quite correct.

This change moves the calculations to happen before moving to the new
directory.

This issue is sporadic, and possibly dependent on the platform.

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

More doc cleanup

Add missing entries to NAME section
Add SYNOPSIS lines, remove old NAME entries
Update find-doc-nits; better regexp's for parsing SYNOPSIS sections.
Rename a couple of files to have an API name.
Remove RSA_private_decrypt; it was duplicate content
Update for recent doc additions

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

Avoid using latest clang since repo not available

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

Make arm-xlate.pl set use strict.

It was already nearly clean. Just one undeclared variable.

Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1240)

Add verification of proxy certs to 25-test_verify.t

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>

Create some proxy certificates

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>

Make it possible to generate proxy certs with test/certs/mkcert.sh

This extends 'req' to take more than one DN component, and to take
them as full DN components and not just CN values.  All other commands
are changed to pass "CN = $cn" instead of just a CN value.

This adds 'genpc', which differs from the other 'gen*' commands by not
calling 'req', and expect the result from 'req' to come through stdin.

Finally, test/certs/setup.sh gets the commands needed to generate a
few proxy certificates.

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>

make update

Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>