The release scripts expect to see the date "xx XXX xxxx" in CHANGES. At
some point the year got changed from xxxx to 2016. This changes it back.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Because proxy certificates typically come without any CRL information,
trying to check revocation on them will fail. Better not to try
checking such information for them at all.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The diverse notations used in INSTALL are not as self explanatory as
we might imagine, so let's attempt a consistent notation for mandatory
and optional pieces of a command line, and to explain the meaning of
each notation.
This does away with the bash notation used in one spot, as it isn't
universally understood and will only confuse the unknowing more.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Experience shows that pod2html changes directory during its process
without properly adjusting the given source directory.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Instead, install the new one as openssl.cnf.dist (openssl.cnf-dist on
VMS), and only install it as openssl.cnf if that file doesn't already
exist.
Also, don't install with exec privileges on VMS.
Reviewed-by: Rich Salz <rsalz@openssl.org>
do_passwd() was returning 0 on success and 1 on failure. However,
those values were interpreted the other way around. The fix that
makes the most sense is to change what do_passwd() returns.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Some calls to RSA_get0_key had the parameters in the wrong order causing a
failure.
GitHub Issue #1368
Reviewed-by: Richard Levitte <levitte@openssl.org>
Windows never composes UTF-8 strings as result of user interaction
such as input query. The only way to compose one is programmatic
conversion from WCHAR string, which in turn can be picked up with
ReadConsoleW.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Windows never composes UTF-8 strings as result of user interaction
such as passing command-line argument. The only way to compose one
is programmatic conversion from WCHAR string, which in turn can be
picked up on command line.
[For reference, why not wmain, it's not an option on MinGW.]
Reviewed-by: Richard Levitte <levitte@openssl.org>
The MULTIBLOCK code uses a "jumbo" sized write buffer which it allocates
and then frees later. Pipelining however introduced multiple pipelines. It
keeps track of how many pipelines are initialised using numwpipes.
Unfortunately the MULTIBLOCK code was not updating this when in deallocated
its buffers, leading to a buffer being marked as initialised but set to
NULL.
RT#4618
Reviewed-by: Rich Salz <rsalz@openssl.org>
Update the documentation for the newly renamed and modified SSL_set0_rbio()
and SSL_set0_wbio() functions. State that they should be preferred over
SSL_set_bio(). Attempt to document the ownership rules for SSL_set_bio().
Reviewed-by: Rich Salz <rsalz@openssl.org>
SSL_set_rbio() and SSL_set_wbio() are new functions in 1.1.0 and really
should be called SSL_set0_rbio() and SSL_set0_wbio(). The old
implementation was not consistent with what "set0" means though as there
were special cases around what happens if the rbio and wbio are the same.
We were only ever taking one reference on the BIO, and checking everywhere
whether the rbio and wbio are the same so as not to double free.
A better approach is to rename the functions to SSL_set0_rbio() and
SSL_set0_wbio(). If an existing BIO is present it is *always* freed
regardless of whether the rbio and wbio are the same or not. It is
therefore the callers responsibility to ensure that a reference is taken
for *each* usage, i.e. one for the rbio and one for the wbio.
The legacy function SSL_set_bio() takes both the rbio and wbio in one go
and sets them both. We can wrap up the old behaviour in the implementation
of that function, i.e. previously if the rbio and wbio are the same in the
call to this function then the caller only needed to ensure one reference
was passed. This behaviour is retained by internally upping the ref count.
This commit was inspired by BoringSSL commit f715c423224.
RT#4572
Reviewed-by: Rich Salz <rsalz@openssl.org>
This adds some simple SSL BIO tests that check for pushing and popping of
BIOs into the chain. These tests would have caught the bugs fixed in the
previous three commits, if combined with a crypto-mdebug build.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The BIO_pop implementation assumes that the rbio still equals the next BIO
in the chain. While this would normally be the case, it is possible that it
could have been changed directly by the application. It also does not
properly cater for the scenario where the buffering BIO is still in place
for the write BIO.
Most of the existing BIO_pop code for SSL BIOs can be replaced by a single
call to SSL_set_bio(). This is equivalent to the existing code but
additionally handles the scenario where the rbio has been changed or the
buffering BIO is still in place.
Reviewed-by: Rich Salz <rsalz@openssl.org>
When pushing a BIO onto an SSL BIO we set the rbio and wbio for the SSL
object to be the BIO that has been pushed. Therefore we need to up the ref
count for that BIO. The existing code was uping the ref count on the wrong
BIO.
Reviewed-by: Rich Salz <rsalz@openssl.org>
When setting the read bio we free up any old existing one. However this can
lead to a double free if the existing one is the same as the write bio.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The SSL_set_bio() function has some complicated ownership rules. This adds a
test to make sure it all works as expected.
Reviewed-by: Rich Salz <rsalz@openssl.org>
SSLv2 is no longer supported in 1.1.0, however we *do* still accept an SSLv2
style ClientHello, as long as we then subsequently negotiate a protocol
version >= SSLv3. The record format for SSLv2 style ClientHellos is quite
different to SSLv3+. We only accept this format in the first record of an
initial ClientHello. Previously we checked this by confirming
s->first_packet is set and s->server is true. However, this really only
tells us that we are dealing with an initial ClientHello, not that it is
the first record (s->first_packet is badly named...it really means this is
the first message). To check this is the first record of the initial
ClientHello we should also check that we've not received any data yet
(s->init_num == 0), and that we've not had any empty records.
GitHub Issue #1298
Reviewed-by: Emilia Käsper <emilia@openssl.org>
8605abf135 fixed the nval leak, but it
used free instead of pop_free. nval owns its contents, so it should be
freed with pop_free. See the pop_free call a few lines down.
This is a no-op as, in this codepath, we must have nval == NULL or
sk_CONF_VALUE_num(nval) == 0. In those cases, free and pop_free are
identical. However, variables should be freed consistently.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1351)
Remove current_method: it was intended as a means of retrying
lookups bit it was never used. Now that X509_verify_cert() is
a "one shot" operation it can never work as intended.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This adds the functions X509_set_proxy_pathlen(), which sets the
internal pc path length cache for a given X509 structure, along with
X509_get_proxy_pathlen(), which retrieves it.
Along with the previously added X509_set_proxy_flag(), this provides
the tools needed to manipulate all the information cached on proxy
certificates, allowing external code to do what's necessary to have
them verified correctly by the libcrypto code.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Since there are a number of function pointers in X509_STORE that might
lead to user code, it makes sense for them to be able to lock the
store while they do their work.
Reviewed-by: Rich Salz <rsalz@openssl.org>
We only add setters for X509_STORE function pointers except for the
verify callback function. The thought is that the function pointers
in X509_STORE_CTX are a cache for the X509_STORE functions.
Therefore, it's preferable if the user makes the changes in X509_STORE
before X509_STORE_CTX_init is called, and otherwise use the verify
callback to override any results from OpenSSL's internal
calculations.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This adds the function X509_set_proxy_flag(), which sets the internal flag
EXFLAG_PROXY on a given X509 structure.
Reviewed-by: Rich Salz <rsalz@openssl.org>
If two CRLs are equivalent then use the one with a later lastUpdate field:
this will result in the newest CRL available being used.
RT#4615
Reviewed-by: Rich Salz <rsalz@openssl.org>
TS_OBJ_print_bio() misuses OBJ_txt2obj: it should print the result
as a null terminated buffer. The length value returned is the total
length the complete text reprsentation would need not the amount of
data written.
CVE-2016-2180
Thanks to Shi Lei for reporting this bug.
Reviewed-by: Matt Caswell <matt@openssl.org>
Don't emit duplicate server/client sections when they are
identical. Instead, just point to the same section.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The way it was implemented before this change, the shared libraries
were installed twice. On a file system that supports file
generations, that's a waste. Slightly rearranging the install targets
solves the problem.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Add tests for resuming with a different client version.
This happens in reality when clients persist sessions on disk through
upgrades.
Reviewed-by: Rich Salz <rsalz@openssl.org>
llvm's ubsan reported:
runtime error: negation of -9223372036854775808 cannot be represented in type
'long'; cast to an unsigned type to negate this value to itself
Found using afl
Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1325
To avoid having to immediately free up r/s when setting them
don't allocate them automatically in DSA_SIG_new() and ECDSA_SIG_new().
RT#4590
Reviewed-by: Richard Levitte <levitte@openssl.org>
On non-Windows platforms, shared libraries are both development and
runtime files. We only installed them as development files, this
makes sure they get installed as runtime files as well.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This is adapted from BoringSSL commit 2f87112b963.
This fixes a number of bugs where the existence of bbio was leaked in the
public API and broke things.
- SSL_get_wbio returned the bbio during the handshake. It must always return
the BIO the consumer configured. In doing so, some internal accesses of
SSL_get_wbio should be switched to ssl->wbio since those want to see bbio.
- The logic in SSL_set_rfd, etc. (which I doubt is quite right since
SSL_set_bio's lifetime is unclear) would get confused once wbio got
wrapped. Those want to compare to SSL_get_wbio.
- If SSL_set_bio was called mid-handshake, bbio would get disconnected and
lose state. It forgets to reattach the bbio afterwards. Unfortunately,
Conscrypt does this a lot. It just never ended up calling it at a point
where the bbio would cause problems.
- Make more explicit the invariant that any bbio's which exist are always
attached. Simplify a few things as part of that.
RT#4572
Reviewed-by: Richard Levitte <levitte@openssl.org>
Systematically test every server-side version downgrade or upgrade.
Client version upgrade or downgrade could be tested analogously but will
be done in a later change.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
f0e0fd51fd added X509_STORE_CTX_set_verify_cb
with a typedef'd argument, making the original one redundant.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
There are two failure cases for OCSP_request_add_id():
1. OCSP_ONEREQ_new() failure, where |cid| is not freed
2. sk_OCSP_ONEREQ_push() failure, where |cid| is freed
This changes makes the error behavior consistent, such that |cid| is
not freed when sk_OCSP_ONEREQ_push() fails. OpenSSL only takes
ownership of |cid| when the function succeeds.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1289)
Because pthread_once() takes a function taking no argument and
returning nothing, and we want to be able to check if they're
successful, we define a few internal macros to get around the issue.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
- Always process ALPN (previously there was an early return in the
certificate status handling)
- Don't send a duplicate alert. Previously, both
ssl_check_clienthello_tlsext_late and its caller would send an
alert. Consolidate alert sending code in the caller.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Continuing from the previous commit. Refactor tls_process_key_exchange() to
split out into a separate function the ECDHE aspects.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Continuing from the previous commit. Refactor tls_process_key_exchange() to
split out into a separate function the DHE aspects.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Continuing from the previous commit. Refactor tls_process_key_exchange() to
split out into a separate function the SRP aspects.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The tls_process_key_exchange() function is too long. This commit starts
the process of splitting it up by moving the PSK preamble code to a
separate function.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The function tls_process_key_exchange() is too long. This commit moves
the PSK preamble processing out to a separate function.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add some more tests for sessions following on from the previous commit
to ensure the callbacks are called when appropriate.
Reviewed-by: Richard Levitte <levitte@openssl.org>
If the SSL_SESS_CACHE_NO_INTERNAL_STORE cache mode is used then we weren't
removing sessions from the external cache, e.g. if an alert occurs the
session is supposed to be automatically removed.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Don't call strncpy with strlen of the source as the length. Don't call
strlen multiple times. Eventually we will want to replace this with a proper
PACKET style handling (but for construction of PACKETs instead of just
reading them as it is now). For now though this is safe because
PSK_MAX_IDENTITY_LEN will always fit into the destination buffer.
This addresses an OCAP Audit issue.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Continuing previous commit to break up the
tls_construct_client_key_exchange() function. This splits out the SRP
code.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Continuing previous commit to break up the
tls_construct_client_key_exchange() function. This splits out the GOST
code.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Continuing previous commit to break up the
tls_construct_client_key_exchange() function. This splits out the ECDHE
code.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Continuing previous commit to break up the
tls_construct_client_key_exchange() function. This splits out the DHE
code.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The tls_construct_client_key_exchange() function is too long. This splits
out the construction of the PSK pre-amble into a separate function as well
as the RSA construction.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Continuing from the previous commits, this splits out the GOST code into
a separate function from the process CKE code.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Continuing from the previous commits, this splits out the ECDHE code into
a separate function from the process CKE code.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Continuing from the previous commit, this splits out the DHE code into
a separate function from the process CKE code.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The tls_process_client_key_exchange() function is far too long. This
splits out the PSK preamble processing, and the RSA processing into
separate functions.
Reviewed-by: Richard Levitte <levitte@openssl.org>
In preparation for splitting this function up into smaller functions this
commit reduces the scope of some of the variables to only be in scope for
the algorithm specific parts. In some cases that makes the error handling
more verbose than it needs to be - but we'll clean that up in a later
commit.
Reviewed-by: Richard Levitte <levitte@openssl.org>
This adds a new target 'build_programs' and makes 'build_apps' and
'build_tests' aliases for it, for backward compatibility.
Reviewed-by: Rich Salz <rsalz@openssl.org>
PROGRAM_NO_INST, ENGINES_NO_INST, SCRIPTS_NO_INST and LIBS_NO_INST are
to be used to specify program, engines, scripts and libraries that are
not to be installed in the system. Fuzzers, test programs, that sort
of things are of the _NO_INST type, for example.
For the benefit of build file templates and other templates that use
data from configdata.pm, a new hash table $unified_info{install} is
created. It contains a set of subhashes, one for each type of
installable, each having an array of file names as values. For
example, it can look like this:
"install" =>
{
"engines" =>
[
"engines/afalg/afalg",
"engines/capi",
"engines/dasync",
"engines/padlock",
],
"libraries" =>
[
"libcrypto",
"libssl",
],
"programs" =>
[
"apps/openssl",
],
"scripts" =>
[
"apps/CA.pl",
"apps/tsget",
"tools/c_rehash",
],
},
Reviewed-by: Rich Salz <rsalz@openssl.org>
The logic testing whether a CKE message is allowed or not was a little
difficult to follow. This tries to clean it up.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
The static function key_exchange_expected() used to return -1 on error.
Commit 361a119127 changed that so that it can never fail. This means that
some tidy up can be done to simplify error handling in callers of that
function.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Having received a ClientKeyExchange message instead of a Certificate we
know that we are not going to receive a CertificateVerify message. This
means we can free up the handshake_buffer. However we better call
ssl3_digest_cached_records() instead of just freeing it up, otherwise we
later try and use it anyway and a core dump results. This could happen,
for example, in SSLv3 where we send a CertificateRequest but the client
sends no Certificate message at all. This is valid in SSLv3 (in TLS
clients are required to send an empty Certificate message).
Found using the BoringSSL test suite.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
In TLS if the server sends a CertificateRequest and the client does not
provide one, if the server cannot continue it should send a
HandshakeFailure alert. In SSLv3 the same should happen, but instead we
were sending an UnexpectedMessage alert. This is incorrect - the message
isn't unexpected - it is valid for the client not to send one - its just
that we cannot continue without one.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
In TLS during ClientAuth if the CA is not recognised you should get an
UnknownCA alert. In SSLv3 this does not exist and you should get a
BadCertificate alert.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
The Client Auth tests were not correctly setting the Protocol, so that this
aspect had no effect. It was testing the same thing lots of times for
TLSv1.2 every time.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Move the preparation of the client certificate to be post processing work
after reading the CertificateRequest message rather than pre processing
work prior to writing the Certificate message. As part of preparing the
client certificate we may discover that we do not have one available. If
we are also talking SSLv3 then we won't send the Certificate message at
all. However, if we don't discover this until we are about to send the
Certificate message it is too late and we send an empty one anyway. This
is wrong for SSLv3.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
The i2d_SCT_LIST function is declared as __owur, therefore we need to check
the result or a --strict-warnings build will fail.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Instead of having fuzz/build.info.fuzz magically and conditionally
included along with the other build.info files, incorporate it in
fuzz/build.info and add the conditions there instead.
Reviewed-by: Rich Salz <rsalz@openssl.org>
With a number of tools, especially those coming with Visual Studio,
some command options are separated from their argument with a space,
others with a space. Since we parametrise them, we can't know
beforehand which it will be, so we must allow the input and output
options to have either.
However, spaces at the end of nmake macro values are trimmed, so allow
spaces to exist by adding a reference to an undefined macro at the end.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This is only done for the platforms where 'OPENSSL_USE_APPLINK' is defined.
Also, change the docs of OPENSSL_Applink to say where to find applink.c
in the installation directory.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The easiest way to take care of manifest files is to integrate them
into the associated binary (.exe or .dll). MT (the Manifest Tool) is
the utility to use for this.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The set0 setters take ownership of their arguments, so the values should
be set to NULL to avoid a double-free in the cleanup block should
ssl_security(SSL_SECOP_TMP_DH) fail. Found by BoringSSL's WeakDH test.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1299)
In light of potential UKS (unknown key share) attacks on some
applications, primarily browsers, despite RFC761, name checks are
by default applied with DANE-EE(3) TLSA records. Applications for
which UKS is not a problem can optionally disable DANE-EE(3) name
checks via the new SSL_CTX_dane_set_flags() and friends.
Reviewed-by: Rich Salz <rsalz@openssl.org>
subject alternate names.
Add nameConstraints tests incluing DNS, IP and email tests both in
subject alt name extension and subject name.
Reviewed-by: Richard Levitte <levitte@openssl.org>
New hostname checking function asn1_valid_host()
Check commonName entries against nameConstraints: any CN components in
EE certificate which look like hostnames are checked against
nameConstraints.
Note that RFC5280 et al only require checking subject alt name against
DNS name constraints.
Reviewed-by: Richard Levitte <levitte@openssl.org>
With OpenSSL 1.1 and on, the engines are tightly tied to the shared
library they're to be used with. That makes them depend on the
pointer size as well as the shared library version, and this gets
reflected in the name of the directory they're installed in.
Reviewed-by: Rich Salz <rsalz@openssl.org>
We're installing architecture dependent (compiled) programs in
architecture specific directories, while architecture independent
programs (scripts) get installed in the general programs directory.
OSSL$EXE: reflects that by having two values.
Reviewed-by: Rich Salz <rsalz@openssl.org>
We previously had a number of logical names for the different parts.
There's really no need for that, the default directories are in one
directory tree. So we only define OSSL$DATAROOT: and make everything
related to that one.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Commit aea145e removed some error codes that are generated
algorithmically: mapping alerts to error texts. Found by
Andreas Karlsson. This restores them, and adds two missing ones.
Reviewed-by: Matt Caswell <matt@openssl.org>
Function-like macros are replaced with prototypes and a note
that they are implemented as macros. Constants are just
referenced in-line in the text.
Tweak BIO_TYPE_... documentation.
Also fix RT4592.
Reviewed-by: Matt Caswell <matt@openssl.org>
We shouldn't allow both "-tls1" and "-tls1_2", or "-tls1" and "-no_tls1_2".
The only time multiple flags are allowed is where they are all "-no_<prot>".
This fixes Github Issue #1268
Reviewed-by: Rich Salz <rsalz@openssl.org>
Some Unix variants require shared libraries to have the execute
permissions set, or they won't be loadable or executable when loaded.
Among others, cygwin has this requirement.
Reviewed-by: Matt Caswell <matt@openssl.org>
It's possible to have a very few rules for some directories and trust
that other patterns further along will take care of whatever is left.
.gitignore should therefore be loosely organised from least generic to
most generic, allowing things like this:
# Keep any file with extensions, such as foo.c, bar.h, ...
!/dir/*.*
# ....
# Remove all object files
*.o
*.obj
With this change, we implement some very generic rules for what will
and will not be ignored in the fuzz subdirectory, and truse that
patterns later on (such as *.o, *.obj, *.exe) will take care of
everything we didn't specifically specify for the fuzz subdirectory.
Reviewed-by: Rich Salz <rsalz@openssl.org>
If application uses any of Windows-specific interfaces, make it
application developer's respondibility to include <windows.h>.
Rationale is that <windows.h> is quite "toxic" and is sensitive
to inclusion order (most notably in relation to <winsock2.h>).
It's only natural to give complete control to the application developer.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
$openssldir and $enginesdir were mistakenly made unavailable to other
perl fragments. They are still needed in the definition of CFLAGS.
Reviewed-by: Rich Salz <rsalz@openssl.org>
OpenSSL engines are tied to the OpenSSL shared library versions,
starting with OpenSSL 1.1. We therefore need to install them in
directories which have the shared library version in it's name, to
easily allow multiple OpenSSL versions to be installed at the same
time.
For VMS, the change is a bit more involved, primarly because the top
installation directory was already versioned, *as well as* some of the
files inside. That's a bit too much. Version numbering in files is
also a bit different on VMS. The engines for shared library version
1.1 will therefore end up in OSSL$INSTROOT:[ENGINES0101.'arch']
('arch' is the architecture we build for)
Reviewed-by: Rich Salz <rsalz@openssl.org>
OpenSSL engines are tied to the OpenSSL shared library versions,
starting with OpenSSL 1.1. We therefore need to install them in
directories which have the shared library version in it's name, to
easily allow multiple OpenSSL versions to be installed at the same
time.
For Unix, the default installation directory is changed from
$PREFIX/lib/engines to $PREFIX/lib/engines-${major}_${minor} (mingw)
or $PREFIX/lib/engines-${major}.${minor} (all but mingw)
($PREFIX is the directory given for the configuration option --prefix,
and ${major} and ${minor} are the major and minor shared library
version numbers)
Reviewed-by: Rich Salz <rsalz@openssl.org>
OpenSSL engines are tied to the OpenSSL shared library versions,
starting with OpenSSL 1.1. We therefore need to install them in
directories which have the shared library version in it's name, to
easily allow multiple OpenSSL versions to be installed at the same
time.
For windows, the default installation directory is changed from
$PREFIX/lib/engines to $PREFIX/lib/engines-${major}_${minor}
($PREFIX is the directory given for the configuration option --prefix,
and ${major} and ${minor} are the major and minor shared library
version numbers)
Reviewed-by: Rich Salz <rsalz@openssl.org>
When creating the library $lib.olb, make sure the extension is there.
Otherwise, a logical name with the same name as the file in question
will redirect the creation elsewhere.
Reviewed-by: Tim Hudson <tjh@openssl.org>
The calls we made to it were redundant, as the same initialization is
done later in OPENSSL_init_crypto() anyway.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Make it possible to have a separate and different perl command string
for installable scripts than we use when building, with the
environment variable HASHBANGPERL. Its value default to the same as
the environment PERL if it's defined, otherwise '/usr/bin/env perl'.
Note: this is only relevant for Unix-like environments.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
The SSL_load_client_CA_file() failed to load any CAs due to an
inccorrect assumption about the return value of lh_*_insert(). The
return value when inserting into a hash is the old value of the key.
The bug was introduced in 3c82e437bb.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1279)
On VMS, it's customary to have a procedure to check that the software
was installed correctly and can run as advertised.
The procedure added here is fairly simple, it checks that all
libraries are in place, that the header crypto.h is in place, and that
the command 'openssl version -a' runs without trouble.
Reviewed-by: Rich Salz <rsalz@openssl.org>
- The install top is versioned by default. However, only the major
version should be used.
- the default areas for certs, private keys an config files have
changed, now all prefixed with 'OSSL$'. This gets reflected in
cryptlib.h.
- [.VMS]openssl_startup.com.in had some faults regarding creating
rooted concealed logical names.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This makes it possible for script writers to lock on to a specific
version if they need to. Note that only the major version number is
used.
Reviewed-by: Rich Salz <rsalz@openssl.org>
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>
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>
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>
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>
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
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>
$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>
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>
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>
"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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
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>
As of 37258dadaa 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
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
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>
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>
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)
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>
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
294d1e36c2, and the logic after the
decryption in the SSL code is constant-time as of
adb46dbc6d, 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
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>
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>
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 = $cn" instead of just a CN value.
This adds 'genpc', which differs from the other 'gen*' commands by not
calling 'req', and expect the result from 'req' to come through stdin.
Finally, test/certs/setup.sh gets the commands needed to generate a
few proxy certificates.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
While travelling up the certificate chain, the internal
proxy_path_length must be updated with the pCPathLengthConstraint
value, or verification will not work properly. This corresponds to
RFC 3820, 4.1.4 (a).
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
The subject name MUST be the same as the issuer name, with a single CN
entry added.
RT#1852
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
Reorder arguments to follow convention.
Also allow r/s to be NULL in DSA_SIG_get0, similarly to ECDSA_SIG_get0.
This complements GH1193 which adds non-const setters.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The previous change for Windows wasn't quite right. Corrected to use
%HOME%, %USERPROFILE% and %SYSTEMPROFILE%, in that order.
Also adding the default home for VMS, SYS$LOGIN:
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
Changing the type of the |str| variable in asn1pars enables us to remove
4 casts with just 1. This silences an OpenBSD warning along the way.
RT4378
Reviewed-by: Richard Levitte <levitte@openssl.org>
EVP_MDs are always const, so stacks of them should be too. This silences
a warning about type punning on OpenBSD.
RT4378
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add const qualifiers to lots of SRP stuff. This started out as an effort
to silence some "type-punning" warnings on OpenBSD...but the fix was to
have proper const correctness in SRP.
RT4378
Reviewed-by: Richard Levitte <levitte@openssl.org>
Previously we would try %RANDFILE%, then %HOME% and finally "C:".
Unfortunately this often ends up being "C:" which the user may not
have write permission for.
Now we try %RANDFILE% first, and then the same set of environment vars
as GetTempFile() uses, i.e. %TMP%, then %TEMP%, %USERPROFILE% and
%SYSTEMROOT%. If all else fails we fall back to %HOME% and only then "C:".
Reviewed-by: Rich Salz <rsalz@openssl.org>
- User targets are now the same and generally do the same things
- configdata.pm depends on exactly the same files on all platforms
- VMS production of shared libraries is simplified
- VMS automatic dependency files get the extension .D rather than .MMS
Reviewed-by: Rich Salz <rsalz@openssl.org>
mkdef.pl was failing to understand no-ripemd. This is a deprecated option
which should act as an alias for no-rmd160.
Reviewed-by: Richard Levitte <levitte@openssl.org>
This reverts commit 9c1a9ccf65.
TerminateProcess is asynchronous, so the code as written in the above
commit is not correct. It is also probably not needed in the speed
case. Reverting in order to figure out the correct solution.
Reviewed-by: Rich Salz <rsalz@openssl.org>
On some platforms we can't startup the TLSProxy due to environmental
problems (e.g. network set up on the build machine). These aren't OpenSSL
problems so we shouldn't treat them as test failures. Just visibly
indicate that we are skipping the test.
We only skip the first time we attempt to start up the proxy. If that works
then everything else should do...if not we should probably investigate and
so report as a failure.
This also removes test_networking...there is a danger that this turns into
a test of user's environmental set up rather than OpenSSL.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Previously EVP_EncodeUpdate returned a void. However there are a couple
of error conditions that can occur. Therefore the return type has been
changed to an int, with 0 indicating error and 1 indicating success.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This is useful in Linux kernel context, in cases data happens
to be fragmented and processing can take multiple calls.
Reviewed-by: Rich Salz <rsalz@openssl.org>
use strict would have caught a number of historical bugs in the perlasm
code, some in the repository and some found during review. It even found
a fresh masm-only bug (see below).
This required some tweaks. The "single instance is enough" globals got
switched to proper blessed objects rather than relying on symbolic refs.
A few types need $opcode passed in as a result.
The $$line thing is a little bit of a nuisance. There may be a clearer
pattern to use instead.
This even a bug in the masm code.
9b634c9b37 added logic to make labels
global or function-global based on whether something starts with a $,
seemingly intended to capture the $decor setting of '$L$'. However, it
references $ret which is not defined in label::out. label::out is always
called after label::re, so $ret was always the label itself, so the line
always ran.
I've removed the regular expression so as not to change the behavior of
the script. A number of the assembly files now routinely jump across
functions, so this seems to be the desired behavior now.
GH#1165
Signed-off-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
'DEPEND[]=file.h' becomes a special way to say that 'file.h' must be
generated before anything else is built. It's likely that a number
of source files depend on these header files, this provides a simple
way to make sure they are always generated even it the dependency data
hasn't been added to the build file yet.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The selector field could be omitted because it has a DEFAULT value.
In this case *sfld == NULL (sfld can never be NULL). This was not
noticed because this was never used in existing ASN.1 modules.
Reviewed-by: Rich Salz <rsalz@openssl.org>
ssl_session_hash() always looks at the first 4 bytes, regardless of the length.
A client can send a session id that's shorter, and the callback could also
generate one that's shorter. So we make sure that the rest of the buffer is
initliazed to 0 so that we always calculate the same hash.
Found by tis-interpreter, also previously reported as RT #2871
Reviewed-by: Rich Salz <rsalz@openssl.org>
MR: #2911
The function a2i_ASN1_STRING can encounter an error after already
allocating a buffer. It wasn't always freeing that buffer on error.
Reviewed-by: Richard Levitte <levitte@openssl.org>
VMS C doesn't allow symbols longer than 31 characters. We do the
automatic shortening with the library files, but not otherwise (to
make sure to work the VMS C magic).
For consistency, I shortened other similar symbols in the same manner.
Reviewed-by: Rich Salz <rsalz@openssl.org>
SSH2 implementations which use DSA_do_verify() and ECDSA_do_verify() are given
the R and S values, and the data to be signed, by the client. Thus in order
to validate these signatures, SSH2 implementations will digest and sign
the data -- and then pass in properly provisioned DSA_SIG and ECDSA_SIG objects.
Unfortunately, the existing OpenSSL-1.1.0 APIs do not allow for directly setting
those R and S values in these objects, which makes using OpenSSL for such
SSH2 implementations much more difficult.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1193)
Update the SSL_set_session() documentation to reflect the fact that old
bad sessions are removed from the cache if necessary.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This commit adds some session API tests, and in particular tests the
modified behaviour of SSL_set_session() introduced in the last commit. To
do this I have factored out some common code from the asynciotest into a
new ssltestlib.c file. I've also renamed getsettest to sslapitest as this
more closely matches what it now is!
Reviewed-by: Rich Salz <rsalz@openssl.org>
SSL_clear() and SSL_free() will remove a session from the cache if it is
considered "bad". However SSL_set_session() does not do this for the session
it is replacing.
SSL_clear() clears an SSL object ready for reuse. It does not clear the
session though. This means that:
SSL_clear(s)
SSL_set_session(s, sess);
and
SSL_set_session(s, sess);
SSL_clear(s);
do not do the same thing, although logically you would expect that they
would.
The failure of SSL_set_session() to remove bad sessions from the cache
should be considered a bug, so this commit fixes it.
RT#597
Reviewed-by: Rich Salz <rsalz@openssl.org>
The SNI tests introduced a redundant "server2" section into every test
configuration. Copy this automatically from "server" unless testing SNI,
to reduce noise in the generated confs.
Also remove duplicate SSL_TEST_CTX_create (merge conflict error).
Reviewed-by: Rich Salz <rsalz@openssl.org>
Some Linux platforms have a suitably recent kernel to support AFALG, but
apparently you still can't actually create an afalg socket. This extends
the afalg_chk_platform() function to additionally check whether we can
create an AFALG socket. We also amend the afalgtest to not report a
failure to load the engine as a test failure. A failure to load is almost
certainly due to platform environmental issues, and not an OpenSSL problem.
RT 4434
Reviewed-by: Andy Polyakov <appro@openssl.org>
- Only send SNI in SNI tests. This allows us to test handshakes without
the SNI extension as well.
- Move all handshake-specific machinery to handshake_helper.c
- Use enum types to represent the enum everywhere
(Resorting to plain ints can end in sign mismatch when the enum is
represented by an unsigned type.)
Reviewed-by: Rich Salz <rsalz@openssl.org>
Don't use BN_hex2bn() for PSK key conversion as the conversion to
BN and back removes leading zeroes, use OPENSSL_hexstr2buf() instead.
RT#4554
Reviewed-by: Matt Caswell <matt@openssl.org>
The TS_RESP_verify_response() function is used for verifying the response
from a TSA. You can set the provided TS_VERIFY_CTX with different flags
depending on what aspects of the response you wish to verify.
A seg fault will occur if you supply the TS_VFY_SIGNER or TS_VFY_TSA_NAME
flags without also specifying TS_VFY_SIGNATURE.
Reviewed-by: Rich Salz <rsalz@openssl.org>
If a user specifies -unix, -6, etc., then the program tries to
use the last one specified. This is confusing code and leads to
scripting errors. Instead, allow only one type.
Reviewed-by: Andy Polyakov <appro@openssl.org>
We already test in EC_POINT_oct2point that points are on the curve. To
be on the safe side, move this check to
EC_POINT_set_affine_coordinates_* so as to also check point coordinates
received through some other method.
We do not check projective coordinates, though, as
- it's unlikely that applications would be receiving this primarily
internal representation from untrusted sources, and
- it's possible that the projective setters are used in a setting where
performance matters.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Files like dh.pod, etc., mostly duplicated the API-specific pod files.
Removed the duplicated content; that often mean the whole file could
be removed. Some of the content about internals got moved into README
files in the source tree. Some content (e.g., err.pod) got moved
into other pod pages.
Annotate generic pages, remove dup NAME
Reviewed-by: Richard Levitte <levitte@openssl.org>
When session tickets are used, it's possible that SNI might swtich the
SSL_CTX on an SSL. Normally, this is not a problem, because the
initial_ctx/session_ctx are used for all session ticket/id processes.
However, when the SNI callback occurs, it's possible that the callback
may update the options in the SSL from the SSL_CTX, and this could
cause SSL_OP_NO_TICKET to be set. If this occurs, then two bad things
can happen:
1. The session ticket TLSEXT may not be written when the ticket expected
flag is set. The state machine transistions to writing the ticket, and
the client responds with an error as its not expecting a ticket.
2. When creating the session ticket, if the ticket key cb returns 0
the crypto/hmac contexts are not initialized, and the code crashes when
trying to encrypt the session ticket.
To fix 1, if the ticket TLSEXT is not written out, clear the expected
ticket flag.
To fix 2, consider a return of 0 from the ticket key cb a recoverable
error, and write a 0 length ticket and continue. The client-side code
can explicitly handle this case.
Fix these two cases, and add unit test code to validate ticket behavior.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1098)
Various fixes to get the following to compile:
./config no-asm -ansi -D_DEFAULT_SOURCE
RT4479
RT4480
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Sessions are stored on the session_ctx, which doesn't change after
SSL_set_SSL_CTX().
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
The previous commit changed how we handle out-of-context empty records.
This commit adds some tests for the various scenarios. There are three
tests:
1: Check that if we inject an out-of-context empty record then we fail
2: Check that if we inject an in-context empty record then we succeed
3: Check that if we inject too many in-context empty records then we fail.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Previously if we received an empty record we just threw it away and
ignored it. Really though if we get an empty record of a different content
type to what we are expecting then that should be an error, i.e. we should
reject out of context empty records. This commit makes the necessary changes
to achieve that.
RT#4395
Reviewed-by: Andy Polyakov <appro@openssl.org>
The previous commit fixed a bug with BN_mod_word() which would have been
caught if we had a test for it. This commit adds one.
Reviewed-by: Andy Polyakov <appro@openssl.org>
On systems where we do not have BN_ULLONG (e.g. typically 64 bit systems)
then BN_mod_word() can return incorrect results if the supplied modulus is
too big.
RT#4501
Reviewed-by: Andy Polyakov <appro@openssl.org>
Make d2i_X509 a generic d2i/i2d manpage.
Pull common stuff out of other d2i/i2d docs.
Update find-doc-nits to know about "generic" manpages.
Cleanup some overlap.
Fix up a bunch of other references.
Reviewed-by: Matt Caswell <matt@openssl.org>
The asdf.pod filename must have asdf in its NAME section.
also check for names existing as a different filename (via Levitte)
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The tlsext_status_type field in SSL is used by e.g. OpenResty to determine
if the client requested the certificate status, but SSL is now opaque.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
GH1180: Local variable sometimes unused
GH1181: Missing close paren.
Thanks to <wipedout@yandex.ru> for reporting these.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Also tweaks to find-doc-nits, including name/synopsis checking.
Ironically, it also reports on duplicated doc names :)
Reviewed-by: Richard Levitte <levitte@openssl.org>
If using threads and OpenSSL is loaded via dlopen(), and subsequently
closed again via dlclose() *before* the threads are destroyed, then
OpenSSL will not free up the per thread resources. We need to document
this restriction, and provide some guidance on what to do about it.
I did some testing and discovered/verified a few of things (at least
this is the behaviour on Linux):
- Using OpenSSL via dlopen in a mutli-threaded app does leak memory if
threads are destroyed after dlcose() is called.
- In a single threaded environment, or if threads are destroyed prior to
dlclose() being called, then no memory is leaked
- Using the RTLD_NODELETE flag to dlopen solves the above problem
- Interestingly the OpenSSL atexit() handler gets called when dlclose()
is called rather than at application exit (I was worred that it might crash
if there was an atexit() handler for a function that has been unloaded)
- RTLD_NODELETE is a non-standard flag - but it does seem to be fairly
widely supported. As far as I could determine (via google), at least Linux,
Solaris, OpenBSD, FreeBSD, HP-UX all seem to support it.
I also tested on Windows (using LoadLibrary instead of dlopen and
FreeLibrary instead of dlclose) and experienced similar behaviour, except
that (AFAIK) there is no equivalent of RTLD_NODELETE on Windows.
GitHub Issue #653
Reviewed-by: Richard Levitte <levitte@openssl.org>
If we're going to redirect STDERR and STDOUT, it's better to do it by
the book. This fix is a straight copy of examples in the perlfunc
manual.
Reviewed-by: Matt Caswell <matt@openssl.org>
The directories for the final products were never registered, it was
plain luck that intermediary files were in the same place and
registered the directory anyway.
Also, scripts are generated directly from source (binaries go through
intermadiary object files), so we need to explicitely make sure to
avoid registering the source directory unless it's an in source
build.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The flags RSA_FLAG_NO_CONSTTIME, DSA_FLAG_NO_EXP_CONSTTIME and
DH_FLAG_NO_EXP_CONSTTIME which previously provided the ability to switch
off the constant time implementation for RSA, DSA and DH have been made
no-ops and deprecated.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The dsa_ossl.c file defined a couple of multi-line macros, but then only
used each one once. The macros just serve to complicate the code and make
it more difficult to understand what is really going on. Hence they are
removed.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Operations in the DSA signing algorithm should run in constant time in
order to avoid side channel attacks. A flaw in the OpenSSL DSA
implementation means that a non-constant time codepath is followed for
certain operations. This has been demonstrated through a cache-timing
attack to be sufficient for an attacker to recover the private DSA key.
CVE-2016-2178
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Some of the instructions used in latest additions are extension
ones. There is no real reason to limit ourselves to specific
processors, so [re-]adhere to base instruction set.
RT#4548
Reviewed-by: Rich Salz <rsalz@openssl.org>
In rare cases, the shell we run test programs in may have complaints.
Shut those up unless testing verbosely.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Previous build scheme allowed building just the stuff in one
subdirectory, like this:
make -C crypto/aes
Because the unified only has a top-level Makefile, this is not
possible with it. This change adds a replacement where each directory
we have something to build in becomes a target in its own right,
allowing building something like this:
make crypto/aes
The exception is the directory test, because we already have such a
target.
Reviewed-by: Stephen Henson <steve@openssl.org>
Remove some unused files.
Rename doc-nit-check to be consistent.
Add check for multiple #include in synopsis.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Since one generates files that the other depends on, there's no
real reason to keep them separate. Since they were both different
aspects of 'openssl req', the merge ends up in 25-test_req.t.
This also makes cleanup easier.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Generate small test programs to check that external programs can be
built with our stuff at a very basic level.
For now, they check that each of our header files can be included
individually without compile failure.
Reviewed-by: Rich Salz <rsalz@openssl.org>
According to the x509 man page in the section discussing -certopt it says
that the ca_default option is the same as that used by the ca utility and
(amongst other things) has the effect of suppressing printing of the
signature - but in fact it doesn't. This error seems to have been present
since the documentation was written back in 2001. It never had this effect.
The default config file sets the certopt value to ca_default. The ca utility
takes that and THEN adds additional options to suppress printing of the
signature. So the ca utility DOES suppress printing of the signature - but
it is not as a result of using the ca_default option.
GitHub Issue #247
Reviewed-by: Rich Salz <rsalz@openssl.org>
If the string to print is exactly 2048 character long (excluding the NULL
terminator) then BIO_printf will chop off the last byte. This is because
it has filled its static buffer but hasn't yet allocated a dynamic buffer.
In cases where we don't have a dynamic buffer we need to truncate but that
is not the case for BIO_printf(). We need to check whether we are able to
have a dynamic buffer buffer deciding to truncate.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The ssl3_init_finished_mac() function can fail, in which case we need to
propagate the error up through the stack.
RT#3198
Reviewed-by: Rich Salz <rsalz@openssl.org>
Fill out the INSTALL instructions with more information on Configure
arguments, environment variables and Makefile targets.
Reviewed-by: Richard Levitte <levitte@openssl.org>
When openssl is compiled with MSVC and _WIN32_WINNT>=0x0601 (Windows 7), BCryptGenRandom is used instead of the legacy CryptoAPI.
This change brings the following benefits:
- Removes dependency on CryptoAPI (legacy API) respectively advapi32.dll
- CryptoAPI Cryptographic Service Providers (rsa full) are not dynamically loaded.
- Allows Universal Windows Platform (UWP) apps to use openssl (CryptGenRandom is not available for Windows store apps)
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1142)
Make the recommendation for MSYS perl in an MSYS environment more forceful.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Fix typos and clarify a few things in the CONTRIBUTING file.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
The README file was a little out of date so needed a refresh
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
The problem is the checking in policy_cache_set, there is a race
condition between the null check and obtaining the lock. The fix is in
policy_cache_new to detect if the creation has happened already.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Port client auth tests to the new framework, add coverage. The old tests
were only testing success, and only for some protocol versions; the new
tests add all protocol versions and various failure modes.
Reviewed-by: Rich Salz <rsalz@openssl.org>
_ctr32 in function name refers to 32-bit counter, but it was implementing
64-bit one. This didn't pose problem to EVP, but 64-bit counter was just
misleading.
RT#4512
Reviewed-by: Richard Levitte <levitte@openssl.org>
Numerous test failures were occuring when Configured with enable-ubsan
although they could all be traced back to one issue.
Reviewed-by: Andy Polyakov <appro@openssl.org>
The notice_section() function allocates a STACK_OF(CONF_VALUE) but
then fails to free it on an error path.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The ec_wNAF_mul() function allocates some temporary storage that it
doesn't always free on an error condition.
Reviewed-by: Richard Levitte <levitte@openssl.org>
After this merge, the only things left (from doc-nit-check) is
74 pages without a "RETURN VALUES" section.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add copyright to missing assembler files.
Add copyrights to missing test/* files.
Add copyrights
Various source and misc files.
Reviewed-by: Richard Levitte <levitte@openssl.org>
RT2630 -- segfault for int overlow
RT2877 -- check return values in apps/rand
Update CHANGES file for previous "windows rand" changes.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The -text argument to dhparam is broken, because the DHparams_print()
function always returns an error. The problem is that always expects a
public or private key to be present, even though that is never the case
with parameters.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The -check argument to dhparam should never identify any problems if we
have just generated the parameters. Add a sanity check for this and print
an error and fail if necessary.
Also updates the documentation for the -check argument, and the DH_check()
function.
RT#4244
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add missing error raise call, as it is done everywhere else.
and as CRYPTO_THREAD_lock_new don't do it internally.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
When the "req" command is used to generate a new EC key using the -newkey
option it will incorrectly display:
Generating a 2048 bit EC private key
This commit fixes the message to not display the bit length for EC keys
because we don't currently support getting that during generation.
GitHub Issue #1068
Reviewed-by: Richard Levitte <levitte@openssl.org>
If openssl is compiled with no-ui or no-stdio, then PEM_read_bio_PrivateKey fails if a password but no callback is provided.
The reason is that the premature return in the PEM_def_callback implementation when OPENSSL_NO_STDIO or OPENSSL_NO_UI is defined, comes too early.
This patch moves the ifdef block to the correct place.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
When an unrecognized algorithm is given on command line together with
-async_jobs, speed_main will jump to clean-up and run
ASYNC_cleanup_thread without calling ASYNC_init_thread first.
Example:
openssl speed -async_jobs 4 ras2048
Signed-off-by: Cristian Stoica <cristian.stoica@nxp.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1084)
Don't copy parameters is they're already present in the destination.
Return error if an attempt is made to copy different parameters to
destination. Update documentation.
If key type is not initialised return missing parameters
RT#4149
Reviewed-by: Richard Levitte <levitte@openssl.org>
- Use the new enable-ubsan and enable-asan configuration options.
- Separate ubsan and asan runs.
- In addition, run shared ubsan tests to get more coverage.
(Shared asan tests need a bit more thought to get them working.)
Reviewed-by: Rich Salz <rsalz@openssl.org>
This fixes the following error when the CRT debug heap (crtdbg.h) is used:
e_os.h(476): warning C4005: 'strdup': macro redefinition
C:\Program Files (x86)\Windows Kits\10\Include\10.0.10586.0\ucrt\crtdbg.h(319): note: see previous definition of 'strdup'
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1137)
Trying to use normal perl conditions to conditionally 'use' a perl
module didn't quite work. Using the 'if' module to do so does work.
Reviewed-by: Andy Polyakov <appro@openssl.org>
As it turns out default glob's behaviour for quoted argument varies
from version to version, making it impossible to Configure or run
tests in some cases. The reason for quoting globs was to accommodate
source path with spaces in its name, which was treated by default glob
as multiple paths. File::Glob::glob on the other hand doesn't consider
spaces as delimiters and therefore works with unquoted patterns.
[Unfortunaltely File::Glob::glob, being too csh-ly, doesn't work
on VMS, hence the "pinning" is conditional.]
Reviewed-by: Richard Levitte <levitte@openssl.org>
Their only reason to exist was that they didn't exist in VMS before
version 7.0. We do not support such old versions any more.
However, for the benefit of systems that don't get strings.h included
by string.h, we include the former in e_os.h.
RT#4458
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Newer gcc still recognizes e.g. -std=c9x in which case it wouldn't
have used 'noreturn' at all with original logic.
Reviewed-by: Richard Levitte <levitte@openssl.org>
We can call memcpy() with a pointer 1 past the last allocated byte and length
of 0 and you can argue that that's undefined behaviour.
Reported by tis-interpreter
Reviewed-by: Rich Salz <rsalz@openssl.org>
GH: #1132
When setting the certificate chain, if a certificate doesn't pass
security checks, then chain may point to a freed STACK_OF(X509)
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
- If we have a maximum amount of characters permitted to be printed
(for example "%.2s", which allows for a maximum of 2 chars), we
minimize the number of characters from the string to printed to
that size.
- If there is space for padding and there is a maximum amount of
characters to print (for example "%3.2s", which shall give at
least a 1 space padding), the amount of characters to pad with
gets added to the maximum so the minimum field size (3 in this
example) gets filled out.
Reviewed-by: Matt Caswell <matt@openssl.org>
This applies when building out-of-source.
RT#4486
NOTE: we can't do the same for Unix, as Unix make doesn't handle this
type of issue. Also, directory specs are much less likely to have
spaces on Unix...
Reviewed-by: Matt Caswell <matt@openssl.org>
Someone wants to configure like this:
PERL="/usr/bin/env perl" ./config
The end goal is to get that in the #! line of CA.pl and a few other
scripts. That works well already, but in the Makefile, there were a
few lines looking like this:
PERL=$(PERL) $(PERL) whatever.pl ...
Those need some quoting.
RT#4311
Reviewed-by: Matt Caswell <matt@openssl.org>
If using the msys console then msys attempts to "fix" command line
arguments to convert them from Unix style to Windows style. One of the
things it does is to look for arguments seperated by colons. This it
assumes is a list of file paths, so it replaces the colon with a semi-colon.
This was causing one of our tests to fail when calling the "req" command
line app. We were attempting to create a new DSA key and passing the
argument "dsa:../apps/dsa1024.pem". This is exactly what we intended but
Msys mangles it to "dsa;../apps/dsa1024.pem" and the command fails.
There doesn't seem to be a way to suppress Msys name mangling. Fortunately
we can work around this issue by generating the DSA key in a separate step
by calling "gendsa".
RT#4255
Reviewed-by: Richard Levitte <levitte@openssl.org>
When closing down the socket in s_client Windows will close it immediately
even if there is data in the write buffer still waiting to be sent. This
was causing tests to fail in Msys/Mingw builds because TLSProxy doesn't see
the final CloseNotify.
I have experimented with various ways of doing this "properly" (e.g.
shutting down the socket before closing, setting SO_LINGER etc). I can't
seem to find the "magic" formula that will make Windows do this. Inserting
a short 50ms sleep seems to do the trick...but its not very "nice" so I've
inserted a TODO on this item. Perhaps someone else will have better luck
in figuring this out.
RT#4255
Reviewed-by: Richard Levitte <levitte@openssl.org>
On Windows we were using the function _kbhit() to determine whether there
was input waiting in stdin for us to read. Actually all this does is work
out whether there is a keyboard press event waiting to be processed in the
input buffer. This only seems to work in a standard Windows console (not
Msys console) and also doesn't work if you redirect the input from some
other source (as we do in TLSProxy tests). This commit changes things to
work differently depending on whether we are on the Windows console or not.
RT#4255
Reviewed-by: Richard Levitte <levitte@openssl.org>
In s_server we call BIO_sock_should_retry() to determine the state of the
socket and work out whether we should retry an operation on it or not.
However if you leave it too long to call this then other operations may
have occurred in the meantime which affect the result. Therefore we should
call it early and remember the result for when we need to use it. This fixes
a test problem on Windows.
Another issue with s_server on Windows is that some of output to stdout does
not get displayed immediately. Apparently more liberal use of BIO_flush is
required.
RT#4255
Reviewed-by: Richard Levitte <levitte@openssl.org>
Clang was complaining about some variables possibly being uninitialized
when used. The warnings are bogus, but clang can't figure that out. This
silences the warnings.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Previous commits fixed the implementation of the %e and %g format
specifiers as well as other issues. This commit adds a test.
Reviewed-by: Richard Levitte <levitte@openssl.org>
We convert the integer part of the float to a long. We should check it
fits first.
Issue reported by Guido Vranken.
GitHub Issue #1102
Reviewed-by: Richard Levitte <levitte@openssl.org>
The previous commit which "fixed" the "e" and "g" floating point formats
just printed them in the same way as "f". This is wrong. This commit
provides the correct formatting.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Convert assert to OPENSSL_assert(), add some documentation, add the calls
to fmtfp() for the "e" and "g" floating point formats which were missing.
Based on a patch provided by Ger Hobbelt <ger@hobbelt.com>.
RT#2270
Reviewed-by: Richard Levitte <levitte@openssl.org>
Windows was complaining about a unary minus operator being applied to an
unsigned type. It did seem to go on and do the right thing anyway, but the
code does look a little suspect. This fixes it.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The ssl3_digest_cached_records() function was failing to handle errors
that might be returned from EVP_DigestSignInit() and
EVP_DigestSignUpdate().
RT#4180
Reviewed-by: Stephen Henson <steve@openssl.org>
Visual C version from version 2003 complain about certain function
names, for example:
apps\apps.c(2572) : warning C4996: 'open': The POSIX name for this item is deprecated. Instead, use the ISO C++ conformant name: _open. See online help for details.
This adds preprocessor aliases for them in e_os.h.
Additionally, crypto/conf/conf_lib.c needs to include e_os.h to catch
those aliases.
RT#4488
RT#4489
Reviewed-by: Matt Caswell <matt@openssl.org>
At earlier point 'which' was replaced with IPC::Cmd::can_run call.
Unfortunately on RPM-based systems it is a separate package and it's
not given that it's installed. Resurrected 'which' provides
poor-man fallback for IPC::Cmd::can_run.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
As it was until now, crypto-mdebug-backtrace was enabled by default
and only disabled if crypto-mdebug was disabled.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Add Configure generated header files to $unified_info{generate}. This
makes sure the build files will pick them up with the rest for the
GENERATED macro, and thereby make sure they get cleaned away by 'make
clean'
Reviewed-by: Rich Salz <rsalz@openssl.org>
This compiles correctly, but depending on what may be defined, it's
possible that this could fail compilation. The braces are mismatched,
and it's possible to end up with an else followed by another else.
This presumes the indentation is mostly correct and indicative of
intent. Found via static analysis.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/1118)
The functions SRP_Calc_client_key() and SRP_Calc_server_key() were
incorrectly returning a valid pointer in the event of error.
Issue reported by Yuan Jochen Kang
Reviewed-by: Richard Levitte <levitte@openssl.org>
The OPENSSL_INIT_set_config_filename() function can fail so ensure that it
provides a suitable error code.
GitHub Issue #920
Reviewed-by: Rich Salz <rsalz@openssl.org>
The function strerror() is not thread safe. We should use strerror_r()
where possible, or strerror_s() on Windows.
RT#2267
Reviewed-by: Richard Levitte <levitte@openssl.org>
ossl_hmac_cleanup, pkey_hmac_cleanup:
- allow to invoke with NULL data
- using EVP_PKEY_CTX_[get|set]_data
EVP_DigestInit_ex:
- remove additional check for ‘type’ and doing clear free instead of
free
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Especially on Windows, the .pl suffix is associated with the perl
interpreter, and therefore make those scripts usable as commands of
their own. On VMS, it simply looks better.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The buf-freelists option was removed in master. There may be some
things that try to disable it, so don't error out.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Add new function PEM_write_bio_PrivateKey_traditional() to enforce the
use of legacy "traditional" private key format. Add -traditional option
to pkcs8 and pkey utilities.
Reviewed-by: Matt Caswell <matt@openssl.org>
README is a fairly independent document, and so is INSTALL. NOTES are
merely addendums to INSTALL. Therefore , INSTALL.DJGPP and
README.PERL get renamed to NOTES.DJGPP and NOTES.PERL.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Fix some code examples, trailing whitespace
Fix TBA sections in verify, remove others.
Remove empty sections
Use Mixed Case not ALL CAPS in head2
Enhance doc-nits script.
Remove extra =cut line
Reviewed-by: Richard Levitte <levitte@openssl.org>
The function InitOnceExceuteOnce is the best way to support the
implementation of CRYPTO_THREAD_run_once() on Windows. Unfortunately
WinXP doesn't have it. To get around that we had two different
implementations: one for WinXP and one for later versions. Which one was
used was based on the value of _WIN32_WINNT.
This approach was starting to cause problems though because other parts of
OpenSSL assume _WIN32_WINNT is going to be 0x0501 and crashes were
occurring dependant on include file ordering. In addition a conditional
based on _WIN32_WINNT had made its way into a public header file through
commit 5c4328f. This is problematic because the value of this macro can
vary between OpenSSL build time and application build time.
The simplest solution to this mess is just to always use the WinXP version
of CRYPTO_THREAD_run_once(). Its perhaps slightly sub-optimal but probably
not noticably.
GitHub Issue #1086
Reviewed-by: Richard Levitte <levitte@openssl.org>
Rename sk_xxx to OPENSSL_sk_xxx and _STACK to OPENSSL_STACK
Rename lh_xxx API to OPENSSL_LH_xxx and LHASH_NODE to OPENSSL_LH_NODE
Make lhash stuff opaque.
Use typedefs for function pointers; makes the code simpler.
Remove CHECKED_xxx macros.
Add documentation; remove old X509-oriented doc.
Add API-compat names for entire old API
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
Giving setbuf() a 64-bit pointer isn't faulty, as the argument is
passed by a 64-bit register anyway, so you only get a warning
(MAYLOSEDATA2) pointing out that only the least significant 32 bits
will be used.
However, we know that a FILE* returned by fopen() and such really is a
32-bit pointer (a study of the system header files make that clear),
so we temporarly turn off that warning when calling setbuf().
Reviewed-by: Andy Polyakov <appro@openssl.org>
This adds an async IO test. There are two test runs. The first one does
a normal handshake with lots of async IO events. The second one does the
same but this time breaks up all the written records into multiple records
of one byte in length. We do this all the way up until the CCS.
Reviewed-by: Andy Polyakov <appro@openssl.org>
In the new state machine if using nbio and we get the header of a
handshake message is one record with the body in the next, with an nbio
event in the middle, then the connection was failing. This is because
s->init_num was getting reset. We should only reset it after we have
read the whole message.
RT#4394
Reviewed-by: Andy Polyakov <appro@openssl.org>
ChangeCipherSpec messages have a defined value. They also may not occur
in the middle of a handshake message. The current logic will accept a
ChangeCipherSpec with value 2. It also would accept up to three bytes of
handshake data before the ChangeCipherSpec which it would discard
(because s->init_num gets reset).
Instead, require that s->init_num is 0 when a ChangeCipherSpec comes in.
RT#4391
Reviewed-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
The write BIO for handshake messages is bufferred so that we only write
out to the network when we have a complete flight. There was some
complexity in the buffering logic so that we switched buffering on and
off at various points through out the handshake. The only real reason to
do this was historically it complicated the state machine when you wanted
to flush because you had to traverse through the "flush" state (in order
to cope with NBIO). Where we knew up front that there was only going to
be one message in the flight we switched off buffering to avoid that.
In the new state machine there is no longer a need for a flush state so
it is simpler just to have buffering on for the whole handshake. This
also gives us the added benefit that we can simply call flush after every
flight even if it only has one message in it. This means that BIO authors
can implement their own buffering strategies and not have to be aware of
the state of the SSL object (previously they would have to switch off
their own buffering during the handshake because they could not rely on
a flush being received when they really needed to write data out). This
last point addresses GitHub Issue #322.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Add doc-nit-check to help find future issues.
Make podchecker be almost clean.
Remove trailing whitespace.
Tab expansion
Reviewed-by: Richard Levitte <levitte@openssl.org>
Defintions of macros similar to _XOPEN_SOURCE belong in command line
or in worst case prior first #include directive in source. As for
macros is was allegedly controlling. One can argue that we are
probably better off demanding S_IS* macros but there are systems
that just don't comply, hence this compromise solution...
Reviewed-by: Rich Salz <rsalz@openssl.org>
In the X509 app check that the obtained public key is valid before we
attempt to use it.
Issue reported by Yuan Jochen Kang.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Since with SSL_VERIFY_NONE, the connection may continue and the
session may even be cached, we should save some evidence that the
chain was not sufficiently verified and would have been rejected
with SSL_VERIFY_PEER. To that end when a CT callback returs failure
we set the verify result to X509_V_ERR_NO_VALID_SCTS.
Note: We only run the CT callback in the first place if the verify
result is still X509_V_OK prior to start of the callback.
RT #4502
Reviewed-by: Tim Hudson <tjh@openssl.org>
- Adjust mkerr.pl to produce the line length we used for source
reformating.
- Have mkerr.pl keep track of preprocessor directive indentation
Among others, do not spuriously throw away a #endif at the end of
header files.
- Make sure mkerr.pl specifies any header inclusion correctly
Reviewed-by: Rich Salz <rsalz@openssl.org>
Set ctx->error = X509_V_ERR_OUT_OF_MEM when verificaiton cannot
continue due to malloc failure. Also, when X509_verify_cert()
returns <= 0 make sure that the verification status does not remain
X509_V_OK, as a last resort set it it to X509_V_ERR_UNSPECIFIED,
just in case some code path returns an error without setting an
appropriate value of ctx->error.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Discard useless static engine_id
Add a const qualifier
Fix some spelling
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
chacha20_poly1305_init_key() dereferences NULL when called with inkey !=
NULL && iv == NULL. This function is called by EVP_EncryptInit_ex()
family, whose documentation allows setting key and iv in separate calls.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
In the SSLV2ClientHello processing code in ssl3_get_record, the value of
|num_recs| will always be 0. This isn't obvious from the code so a comment
is added to explain it.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
When RSA went opaque a bug was introduced into the dasync engine where
the wrong function was being set for the rsa_priv_dec operation.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The function ssl3_get_record() can obtain multiple records in one go
as long as we are set up for pipelining and all the records are app
data records. The logic in the while loop which reads in each record is
supposed to only continue looping if the last record we read was app data
and we have an app data record waiting in the buffer to be processed. It
was actually checking that the first record had app data and we have an
app data record waiting. This actually amounts to the same thing so wasn't
wrong - but it looks a bit odd because it uses the |rr| array without an
offset.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Pipelining introduced the concept of multiple records being read in one
go. Therefore we work with an array of SSL3_RECORD objects. The pipelining
change erroneously made a change in ssl3_get_record() to apply the current
record offset to the SSL3_BUFFER we are using for reading. This is wrong -
there is only ever one read buffer. This reverts that change. In practice
this should make little difference because the code block in question is
only ever used when we are processing a single record.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
An if checks the value of |type| to see if it is V_ASN1_VISIBLESTRING
twice. We only need to do it once.
GitHub Issue #656
Reviewed-by: Richard Levitte <levitte@openssl.org>
This involves providing more session ticket key data, for both the cipher and
the digest
Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Matt Caswell <matt@openssl.org>
GH: #515, MR: #2153
Workaround an apparent IO:Socket::IP bug where a seemingly valid
server socket is being returned even though a valid connection does not
exist. This causes the tests to intermittently hang. We additionally check
that the peerport looks ok to verify that the returned socket looks usable.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Since 50932c4 "PACKETise ServerHello processing",
ssl_next_proto_validate() incorrectly allows empty protocol name.
draft-agl-tls-nextprotoneg-04[1] says "Implementations MUST ensure that
the empty string is not included and that no byte strings are
truncated."
This patch restores the old correct behavior.
[1] https://tools.ietf.org/html/draft-agl-tls-nextprotoneg-04
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Add a status return value instead of void.
Add some sanity checks on reference counter value.
Update the docs.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
check source's kdf_ukm, not destination's
use != NULL, instead of implicit checking
don't free internal data structure like pkey_rsa_copy()
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
In some cases, perl's glob() thinks it needs to return file names with
generation numbers, such as when a file name pattern includes two
periods. Constructing other file names by simple appending to file
names with generation numbers isn't a good idea, so for the VMS case,
just peal the generation numbers if they are there.
Fortunately, this is easy, as the returned generation number delimiter
will always be a semi-colon.
Reviewed-by: Rich Salz <rsalz@openssl.org>
If the server does not send a session ticket extension, it should not then
send the NewSessionTicket message.
If the server sends the session ticket extension, it MUST then send the
NewSessionTicket message.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Per RFC 4507, section 3.3:
This message [NewSessionTicket] MUST be sent if the
server included a SessionTicket extension in the ServerHello. This
message MUST NOT be sent if the server did not include a
SessionTicket extension in the ServerHello.
The presence of the NewSessionTicket message should be determined
entirely from the ServerHello without probing.
RT#4389
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Fix -signer option in smime utility to output signer certificates
when verifying.
Add support for format SMIME for -inform and -outform with cms and
smime utilities.
PR#4215
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
* Configure: Replaced -DTERMIO by -DTERMIOS in CFLAGS.
* crypto/bio/bss_dgram.c [WATT32]: Remove obsolete redefinition of
function names: sock_write, sock_read and sock_puts.
* crypto/bio/bss_sock.c [WATT32]: For Watt-32 2.2.11 sock_write,
sock_read and sock_puts are redefined to their private names so
their names must be undefined first before they can be redefined
again.
* crypto/bio/bss_file.c (file_fopen) [__DJGPP__]: Make a copy of the
passed file name and replace the leading dots in the dirname part
and the basname part of the file name, unless LFN is supported.
* e_os.h [__DJGPP__]: Undefine macro DEVRANDOM_EGD. Neither MS-DOS nor
FreeDOS provide 'egd' sockets.
New macro HAS_LFN_SUPPORT checks if underlying file system supports
long file names or not.
Include sys/un.h.
Define WATT32_NO_OLDIES.
* INSTALL.DJGPP: Update URL of WATT-32 library.
Submitted by Juan Manuel Guerrero <juan.guerrero@gmx.de>
RT#4217
Reviewed-by: Andy Polyakov <appro@openssl.org>
DJGPP is a 3rd party configuration, we rely entirely on the OpenSSL to
help us fine tune and test. Therefore, it's moved to its own config.
Reviewed-by: Andy Polyakov <appro@openssl.org>
The old proxy tests test the implementation of an application proxy
policy callback defined in the test itself, which is not particularly
useful.
It is, however, useful to test cert verify overrides in
general. Therefore, replace these tests with tests for cert verify
callback behaviour.
Also glob the ssl test inputs on the .in files to catch missing
generated files.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Multiple digest options to the ocsp utility are allowed: e.g. to use
different digests for different certificate IDs. A digest option without
a following certificate is however illegal.
RT#4215
Reviewed-by: Rich Salz <rsalz@openssl.org>
While it seemed like a good idea to have this file once upon a time,
this kind of file belongs with the package maintainer rather than in
our source.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The old cipherlist test in ssltest.c only tests the internal order of
the cipher table, which is pretty useless.
Replace this test with a test that catches inadvertent changes to the
default cipherlist.
Fix run_tests.pl to correctly filter tests that have "list" in their name.
(Also includes a small drive-by fix in .gitignore.)
Reviewed-by: Rich Salz <rsalz@openssl.org>
The V2ClientHello code creates an empty compression list, but the
compression list must explicitly contain the null compression (and later
code enforces this).
RT#4387
Reviewed-by: Stephen Henson <steve@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
- some Perl versions are allergic to missing ';';
- don't stop if del fails;
- omit unused environment variable;
Reviewed-by: Stephen Henson <steve@openssl.org>
Don't primarly recommend using OPENSSL_thread_stop(), as that's a last
resort. Instead, recommend leaving it to automatic mechanisms.
Reviewed-by: Matt Caswell <matt@openssl.org>
The ERR_remove_thread_state() API is restored to take a pointer
argument, but does nothing more. ERR_remove_state() is also made into
a no-op. Both functions are deprecated and users are recommended to
use OPENSSL_thread_stop() instead.
Documentation is changed to reflect this.
Reviewed-by: Matt Caswell <matt@openssl.org>
Per RFC 5246,
Note: this extension is not meaningful for TLS versions prior to 1.2.
Clients MUST NOT offer it if they are offering prior versions.
However, even if clients do offer it, the rules specified in [TLSEXT]
require servers to ignore extensions they do not understand.
Although second sentence would suggest that there would be no interop
problems in always offering the extension, WebRTC has reported issues
with Bouncy Castle on < TLS 1.2 ClientHellos that still include
signature_algorithms. See also
https://bugs.chromium.org/p/webrtc/issues/detail?id=4223
RT#4390
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
BIO_eof() was always returning true when using a BIO pair. It should only
be true if the peer BIO is empty and has been shutdown.
RT#1215
Reviewed-by: Richard Levitte <levitte@openssl.org>
Issue was introduced in
a0a82324f9
This patch fixes an issue which causes the 'openssl ca' commands to
fail if '-config' is not specified even if it says so otherwise.
Problem is that the default config is not loaded and the conf variable
is NULL which causes an exception.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
With Unixly Makefiles as well as with nmake, make variables are
transferred to the shell running the commands as envinronment
variables. This principle doesn't apply with MMS, so we must
explicitely define VERBOSE as commands when it's needed.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The tests was incorrectly repeated multiple times when using the
async_jobs options
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
When setting a new SRTP connection profile using
SSL_CTX_set_tlsext_use_srtp() or SSL_set_tlsext_use_srtp() we should
free any existing profile first to avoid a memory leak.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
If environment variables are not explanded early enough, expanded
strings are passed with single backslash to C compiler, e.g.
C:\Program Files, which effectively results in OpenSSL looking for
engines and certificates in C:Program Files.
Reviewed-by: Richard Levitte <levitte@openssl.org>
- Missing checks for allocation failure.
- releasing memory in few missing error paths
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Try to set the ASN.1 parameters for CMS encryption even if the IV
length is zero as the underlying cipher should still set the type.
This will correctly result in errors if an attempt is made to use
an unsupported cipher type.
Reviewed-by: Rich Salz <rsalz@openssl.org>
If the application has limited the size of the async pool using
ASYNC_init_thread() then we could run out of jobs while trying to start a
libssl io operation. However libssl was failing to handle this and treating
it like a fatal error. It should not be fatal...we just need to retry when
there are jobs available again.
Reviewed-by: Richard Levitte <levitte@openssl.org>
This demystifies two for-loops that do nothing. They were used to write
the ladder in a unified way. Now that the ladder is otherwise commented,
remove the dead loops.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Based on an orignal commit by GitHub user BertramScharpf. Rebased and
updated to take account of all the updates since this was first raised.
GH PR#62
Reviewed-by: Rich Salz <rsalz@openssl.org>
list -disabled was checking OPENSSL_NO_SSL/OPENSSL_NO_TLS, which are
not used to disable SSL/TLS respectively.
Building with these macros wrongly show as SSL/TLS disabled, hence
removing this code.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
The name length limit check in x509_name_ex_d2i() includes
the containing structure as well as the actual X509_NAME. This will
cause large CRLs to be rejected.
Fix by limiting the length passed to ASN1_item_ex_d2i() which will
then return an error if the passed X509_NAME exceeds the length.
RT#4531
Reviewed-by: Rich Salz <rsalz@openssl.org>
The old BIO_accept() function can encounter errors during malloc. We need
to ensure we properly clean up if that occurs.
GH Issue #817
Reviewed-by: Richard Levitte <levitte@openssl.org>
The init_client() function in the apps sets up the client connection. It
may try multiple addresses until it finds one that works. We should clear
the error queue if we eventually get a successful connection because
otherwise we get stale errors hanging around. This can cause problems in
subsequent calls to SSL_get_error(), i.e. non-fatal NBIO events appear as
fatal.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Some non-Gnu compilers interpret -E -P combination differently.
some prioritize -E over -P, others -P over -E (in which case .i
file is generated and sometimes truncated because of redirection).
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add tests for ASN.1 INTEGER: invalid tag, valid 0, 1, -1 and 0, -1 with
illegal padding.
Also add ASN1_ANY tests for 0, 1 and -1.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Only treat an ASN1_ANY type as an integer if it has the V_ASN1_INTEGER
tag: V_ASN1_NEG_INTEGER is an internal only value which is never used
for on the wire encoding.
Thanks to David Benjamin <davidben@google.com> for reporting this bug.
This was found using libFuzzer.
RT#4364 (part)CVE-2016-2108.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
With the EVP_EncodeUpdate function it is the caller's responsibility to
determine how big the output buffer should be. The function writes the
amount actually used to |*outl|. However this could go negative with a
sufficiently large value for |inl|. We add a check for this error
condition.
Reviewed-by: Richard Levitte <levitte@openssl.org>
An overflow can occur in the EVP_EncodeUpdate function which is used for
Base64 encoding of binary data. If an attacker is able to supply very large
amounts of input data then a length check can overflow resulting in a heap
corruption. Due to the very large amounts of data involved this will most
likely result in a crash.
Internally to OpenSSL the EVP_EncodeUpdate function is primarly used by the
PEM_write_bio* family of functions. These are mainly used within the
OpenSSL command line applications, so any application which processes
data from an untrusted source and outputs it as a PEM file should be
considered vulnerable to this issue.
User applications that call these APIs directly with large amounts of
untrusted data may also be vulnerable.
Issue reported by Guido Vranken.
CVE-2016-2105
Reviewed-by: Richard Levitte <levitte@openssl.org>
ASN1 Strings that are over 1024 bytes can cause an overread in
applications using the X509_NAME_oneline() function on EBCDIC systems.
This could result in arbitrary stack data being returned in the buffer.
Issue reported by Guido Vranken.
CVE-2016-2176
Reviewed-by: Andy Polyakov <appro@openssl.org>
An overflow can occur in the EVP_EncryptUpdate function. If an attacker is
able to supply very large amounts of input data after a previous call to
EVP_EncryptUpdate with a partial block then a length check can overflow
resulting in a heap corruption.
Following an analysis of all OpenSSL internal usage of the
EVP_EncryptUpdate function all usage is one of two forms.
The first form is like this:
EVP_EncryptInit()
EVP_EncryptUpdate()
i.e. where the EVP_EncryptUpdate() call is known to be the first called
function after an EVP_EncryptInit(), and therefore that specific call
must be safe.
The second form is where the length passed to EVP_EncryptUpdate() can be
seen from the code to be some small value and therefore there is no
possibility of an overflow.
Since all instances are one of these two forms, I believe that there can
be no overflows in internal code due to this problem.
It should be noted that EVP_DecryptUpdate() can call EVP_EncryptUpdate()
in certain code paths. Also EVP_CipherUpdate() is a synonym for
EVP_EncryptUpdate(). Therefore I have checked all instances of these
calls too, and came to the same conclusion, i.e. there are no instances
in internal usage where an overflow could occur.
This could still represent a security issue for end user code that calls
this function directly.
CVE-2016-2106
Issue reported by Guido Vranken.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Date: Tue Mar 15 15:19:44 2016 +0100
This commit updates the documentation of cms, ocsp, s_client,
s_server, and verify to reflect the new "-no_check_time"
option introduced in commit d35ff2c0ad
on 2015-07-31.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
If no serverinfo extension is found in some cases, do not abort the handshake,
but simply omit/skip that extension.
Check for already-registered serverinfo callbacks during serverinfo
registration.
Update SSL_CTX_use_serverinfo() documentation to mention the need to reload the
same serverinfo per certificate, for servers with multiple server certificates.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The environment variable OPENSSL_LOCAL_CONFIG_DIR is used to indicate
that there's a local directory with extra configuration files.
Reviewed-by: Stephen Henson <steve@openssl.org>
Fix some of the variables to be (s)size_t, so that more than 1GB of
secure memory can be allocated. The arena has to be a power of 2, and
2GB fails because it ends up being a negative 32-bit signed number.
The |too_late| flag is not strictly necessary; it is easy to figure
out if something is secure memory by looking at the arena. As before,
secure memory allocations will not fail, but now they can be freed
correctly. Once initialized, secure memory can still be used, even if
allocations occured before initialization.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Usage of $ymm variable is a bit misleading here, it doesn't refer
to %ymm register bank, but rather to VEX instruction encoding,
which AMD XOP code path depends on.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Sanity check field lengths and sums to avoid potential overflows and reject
excessively large X509_NAME structures.
Issue reported by Guido Vranken.
Reviewed-by: Matt Caswell <matt@openssl.org>
This adds an explicit limit to the size of an X509_NAME structure. Some
part of OpenSSL (e.g. TLS) already effectively limit the size due to
restrictions on certificate size.
Reviewed-by: Matt Caswell <matt@openssl.org>
The srp application created an uninitialised DB_ATTR object and then
passed it to the load_index function which attempted to read it. A
DB_ATTR object only contains a single field called "unique_subject".
AFAICT this attribute is unused in the SRP case, and therefore it would be
better to pass a NULL DB_ATTR to load_index (which handles that case
gracefully).
Reviewed-by: Rich Salz <rsalz@openssl.org>
We should only copy parameters and keys if the group is set. Otherwise
they don't really make any sense. Previously we copied the private key
regardless of whether the group was set...but if it wasn't a NULL ptr
deref could occur. It's unclear whether we could ever get into that
situation, but since we were already checking it for the public key we
should be consistent.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The length is a long, so returning the difference does not quite work.
Thanks to Torbjörn Granlund for noticing.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
The non-ascii version of this set of macros ensures that the "a" variable
is inside the expected range. This logic wasn't quite right for the
EBCDIC version.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Building with -DCHARSET_EBCDIC and using --strict-warnings resulted in
lots of miscellaneous errors. This fixes it.
Reviewed-by: Andy Polyakov <appro@openssl.org>
When compiling all other C files, rely on the compiler to
automatically pick up the name translation information from the header
files __DECC_INCLUDE_{PRO,EPI}LOGUE.H.
Reviewed-by: Andy Polyakov <appro@openssl.org>
With DEC C on VMS, you can use __DECC_INCLUDE_PROLOGUE.H and
__DECC_INCLUDE_EPILOGUE.H to include some DEC C specific features or
pragmas without having to touch the other header files.
It seems, however, that the current version of the compiler requires
the file names to be upcased, or it doesn't handle them quite right.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Currently we can get all block ciphers with
EVP_get_cipherbyname("<alg_name>-<block-mode-name>")
for example, by names "aes-128-ecb" or "des-ede-cbc".
I found a problem with des-ede-ecb and des-ede3-ecb ciphers as
they can be accessed only with names:
EVP_get_cipherbyname("des-ede")
EVP_get_cipherbyname("des-ede3")
It breaks the general concept.
In this patch I add aliases which allow to use names:
EVP_get_cipherbyname("des-ede-ecb")
EVP_get_cipherbyname("des-ede3-ecb")
in addition to the currently used names.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The tls_client_key_exchange_post_work() frees the pms on error. It also
calls ssl_generate_master_secret() which also free the pms. If an error
occurs after ssl_generate_master_secret() has been called then a double
free can occur.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Commit 91fb42dd fixed a leak but introduced a problem where a parameter
is erroneously freed instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
OpenSSL 1.1.0-pre5 has made some additional structs opaque. Python's ssl
module requires access to some of the struct members. Three new getters
are added:
int X509_OBJECT_get_type(X509_OBJECT *a);
STACK_OF(X509_OBJECT) *X509_STORE_get0_objects(X509_STORE *v);
X509_VERIFY_PARAM *X509_STORE_get0_param(X509_STORE *ctx);
Signed-off-by: Christian Heimes <cheimes@redhat.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
The code that implements this control would work when enabling nbio,
but the disabling code needed fixing.
Reviewed-by: Matt Caswell <matt@openssl.org>
The rsa_cms_encrypt() function allocates an ASN1_OCTET_STRING but can
then fail to free it in an error condition.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The PKCS7_dataFinal() function allocates a memory buffer but then fails
to free it on an error condition.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The i2b_PVK function leaked a number of different memory allocations on
error paths (and even some non-error paths).
Reviewed-by: Richard Levitte <levitte@openssl.org>
The b2i_rsa() function uses a number of temporary local variables which
get leaked on an error path.
Reviewed-by: Richard Levitte <levitte@openssl.org>
On error we could leak a ACCESS_DESCRIPTION and an ASN1_IA5STRING. Both
should be freed in the error path.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The cms_SignerInfo_content_sign() function allocated an EVP_MD_CTX but
then failed to free it on an error path.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The cms_RecipientInfo_pwri_crypt() allocated an EVP_CIPHER_CTX but then
failed to free it in some error paths. By allocating it a bit later that
can be avoided.
Reviewed-by: Richard Levitte <levitte@openssl.org>
In BN_generate_prime_ex() we do some sanity checks first and return
with an error if they fail. We should do that *before* allocating any
resources to avoid a memory leak.
Reviewed-by: Richard Levitte <levitte@openssl.org>
In the BN_mpi2bn() function, a failure of a call to BN_bin2bn() could
result in the leak of a previously allocated BIGNUM value.
Reviewed-by: Richard Levitte <levitte@openssl.org>
During construction of a mem BIO we allocate some resources. If this
allocation fails we can end up leaking everything we have allocated so
far.
Reviewed-by: Richard Levitte <levitte@openssl.org>
When setting an accepted socket for non-blocking, if the operation fails
make sure we close the accepted socket.
Reviewed-by: Richard Levitte <levitte@openssl.org>
BIO_sock_error() returned 1 when getsockopt() fails when it should
return the error code for that failure.
Additionally, the optlen parameter to getsockopt() has to point at
the size of the area that the optval parameter points at rather than
zero. Some systems may forgive it being zero, but others don't.
Reviewed-by: Matt Caswell <matt@openssl.org>
The traditional private key encryption algorithm doesn't function
properly if the IV length of the cipher is zero. These ciphers
(e.g. ECB mode) are not suitable for private key encryption
anyway.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Coverity reports a potential NULL deref when "2 0 0" DANE trust-anchors
from DNS are configured via SSL_dane_tlsa_add() and X509_STORE_CTX_init()
is called with a NULL stack of untrusted certificates.
Since ssl_verify_cert_chain() always provideds a non-NULL stack of
untrusted certs, and no other code path enables DANE, the problem
can only happen in applications that use SSL_CTX_set_cert_verify_callback()
to implement their own wrappers around X509_verify_cert() passing
only the leaf certificate to the latter.
Regardless of the "improbability" of the problem, we do need to
ensure that build_chain() handles this case correctly.
Reviewed-by: Matt Caswell <matt@openssl.org>
The diverse {RSA,DSA,DH}_set0_* functions are made to allow some
parameters to be NULL IF the corresponding numbers in the given key
structure have already been previously initialised. Specifically,
this allows the addition of private components to be added to a key
that already has the public half, approximately like this:
RSA_get0_key(rsa, NULL, &e, NULL);
RSA_get0_factors(rsa, &p, &q);
/* calculate new d */
RSA_set0_key(rsa, NULL, NULL, d);
Reviewed-by: Matt Caswell <matt@openssl.org>
Add X509_STORE_{set,get}_ex_data() function and
X509_STORE_get_ex_new_index() macro.
X509_STORE has ex_data and the documentation also mentions them but they
are not actually implemented.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
The variables in the BIO weren't being duplicated properly.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Passing the -stdin arg to the passwd command line app *and* supply a
password on the command line causes a seg fault.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Commit d32f5d8733 added a 'goto end;' statement
at the end of the code block for the 'end' label. Fortunately, it was after a
return statement, so no infinite loop occurred, but it is still dead code.
Remove the extra goto statement as cleanup.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add X509_STORE_{set,get}_ex_data() function and
X509_STORE_get_ex_new_index() macro.
X509_STORE has ex_data and the documentation also mentions them but they
are not actually implemented.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
The i2d_X509() function can return a negative value on error. Therefore
we should make sure we check it.
Issue reported by Yuan Jochen Kang.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
It was added as part of 2df84dd329
but has never actually been used for anything; presumably it was
a typo for one of SCTP or CT.
This removes the last '??' entry from INSTALL.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Commit d064e6ab52 removed all the
OPENSSL_NO_SHA guards, but commit
a50ad1daaa regenerated some due to the
sha entries in the %md_disabler table in apps/progs.pl.
Update %md_disabler to reflect that sha is not disableable, and
remove OPENSSL_NO_SHA for good.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Even though no test could be found to trigger this, paper-n-pencil
estimate suggests that x86 and ARM inner loop lazy reductions can
loose a bit in H4>>*5+H0 step.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
In the case of generating a file like this:
GENERATE[foo.S]=mkfoo.pl arg1 arg2
the 'mkfoo.pl' generator itself might need to include other files,
such as perl modules within our source tree. We can reuse already
existing syntax for it, like this:
INCLUDE[mkfoo.pl]=module/path
or:
DEPEND[mkfoo.pl]=modules/mymodule.pm
This change implements the support for such constructs, and for the
DEPEND statement, for any value that indicates a perl module (.pm
file), it will automatically infer an INCLUDE statement for its
directory, just like it does for C header files, so you won't have do
write this:
DEPEND[mkfoo.pl]=modules/mymodule.pm
INCLUDE[mkfoo.pl]=modules
Reviewed-by: Emilia Käsper <emilia@openssl.org>
IBM argues that in certain scenarios capability query is really
expensive. At the same time it's asserted that query results can
be safely cached, because disabling CPACF is incompatible with
reboot-free operation.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Fix a bug introduced by 6903e2e7e9 (Extended EC_METHOD customisation
support., 2016-02-01). key->meth->set_private() is wrongly called where
it should call key->group->meth->set_private().
PR#4517
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
Bad ASN.1 data should never be able to trigger a malloc failure so return
an error in d2i_test if a malloc failure occurs.
Reviewed-by: Richard Levitte <levitte@openssl.org>
If the ASN.1 BIO is presented with a large length field read it in
chunks of increasing size checking for EOF on each read. This prevents
small files allocating excessive amounts of data.
CVE-2016-2109
Thanks to Brian Carpenter for reporting this issue.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The reason to warn is that configuration *may* pick up on
configuration header files that are in the source tree, that might be
for a wildly different configuration than what is expected in the
current out-of-source configuration.
Reviewed-by: Matt Caswell <matt@openssl.org>
It is up to the caller of SSL_dane_tlsa_add() to take appropriate
action when no records are added successfully or adding some records
triggers an internal error (negative return value).
With this change the caller can continue with PKIX if desired when
none of the TLSA records are usable, or take some appropriate action
if DANE is required.
Also fixed the internal ssl_dane_dup() function to properly initialize
the TLSA RR stack in the target SSL handle. Errors in ssl_dane_dup()
are no longer ignored.
Reviewed-by: Rich Salz <rsalz@openssl.org>
If we fail halfway through constructing the peer_tmp EVP_PKEY but we have
already stored it in s->s3->peer_tmp then if anything tries to use it then
it will likely fail. This was causing s_client to core dump in the
sslskewith0p test. s_client was trying to print out the connection
parameters that it had negotiated so far. Arguably s_client should not do
that if the connection has failed...but given it is existing functionality
it's easier to fix libssl.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Using ASN1_ITEM tables in d2i_test: this then uses consistent names and
makes it easier to extend.
Add bio, reencode and compare tests.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The no-dsa option was failing on Windows because some symbols were not
correctly flagged in libcrypto.num. Problem found due to the new symbol
consistency test.
Reviewed-by: Richard Levitte <levitte@openssl.org>
no-cmac was failing on Windows/VMS due to libcrypto.num not marking the
CMAC functions properly. Found due to the new symbol consistency test.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Our main development platforms are of the Unix family, which doesn't
have the same strictness regarding a shared library being consistent
with the contents of the ld script (.map file, on Linux and Solaris)
as Windows is with the contents of the .def file or VMS is with the
linker symb_vector option.
To eliminate surprises, we therefore need to make sure to check that
the contents of the .map file is matched with the shared library, at
least to check that the shared library isn't missing any symbols that
should be present.
This test isn't absolutely perfect, as it will only check the symbols
that would be present on Linux / Solaris and will therefore miss those
that would only appear on Windows or VMS. On the other hand, those
platform specific are few and far apart in time, so it's not likely
they will pose a problem.
Reviewed-by: Matt Caswell <matt@openssl.org>
If allocation in CRYPTO_clear_realloc() fails don't free up the original
buffer: this is consistent with the behaviour of realloc(3) and is expected
in other places in OpenSSL.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The Unix build was the last to retain the classic build scheme. The
new unified scheme has matured enough, even though some details may
need polishing.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Link errors were occurring on Windows because the header files were not
correctly guarding some functions with OPENSSL_NO_SOCK
Reviewed-by: Rich Salz <rsalz@openssl.org>
e_os.h was defining OPENSSL_NO_DGRAM if OPENSSL_NO_SOCK was defined.
This causes link problems on Windows because the generated .def files
still contain the DGRAM symbols even though they have not been compiled.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Link errors were occurring on Windows because the header files were not
correctly guarding some functions with OPENSSL_NO_DGRAM
Reviewed-by: Rich Salz <rsalz@openssl.org>
Add copyright to most .pl files
This does NOT cover any .pl file that has other copyright in it.
Most of those are Andy's but some are public domain.
Fix typo's in some existing files.
Reviewed-by: Richard Levitte <levitte@openssl.org>
This also restores the possibility to have ml used with VC-WIN32 with
no-asm, which was lost during the mk1mf -> unified transition.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Because we have a directory 'test', the target 'test' may be confusing
to make. However, if we make it depend on 'tests', which doesn't
exist, make should never fail to run the actions.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Commit f0e0fd51f was a bit over-zealous in removing a call to
X509_STORE_CTX_cleanup(). The call in question was in a loop and was
required to cleanup resources used on each iteration of the loop. Removing
this resulted in a memory leak.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Make OBJ_name_cmp internal
Rename idea_xxx to IDEA_xxx
Rename get_rfc_xxx to BN_get_rfc_xxx
Rename v3_addr and v3_asid functions to X509v3_...
Reviewed-by: Richard Levitte <levitte@openssl.org>
Recently, OCSP_basic_verify() was changed to always return 0 on error,
when it would previously return 0 on error and < 0 on fatal error.
This restores the previous semantics back.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Make X509_OBJECT, X509_STORE_CTX, X509_STORE, X509_LOOKUP,
and X509_LOOKUP_METHOD opaque.
Remove unused X509_CERT_FILE_CTX
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
To begin with, the default should have been the import library, not
the DLL itself. However, we don't know what directory it's installed
in either way, so we may as well demand the full path from the user
Reviewed-by: Matt Caswell <matt@openssl.org>
In most cases we expect that people will be using shared libraries not
static ones, therefore we make that the default.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Travis identified a problem with freeing the ex_data locks which wasn't
quite right in ff2344052. Trying to fix it identified a further problem:
the ex_data locks are cleaned up by OPENSSL_cleanup(), which is called
explicitly by CRYPTO_mem_leaks(), but then later the BIO passed to
CRYPTO_mem_leaks() is freed. An attempt is then made to use the ex_data
lock already freed.
Reviewed-by: Tim Hudson <tjh@openssl.org>
no-stdio does not work with the apps. Since the tests also need the apps
it doesn't support that either. Therefore we disable building of both.
no-autoalginit is not compatible with the apps because it requires explicit
loading of the algorithms, and the apps don't do that. Therefore we disable
building the apps for this option. Similarly the tests depend on the apps
so we also disable the tests. Finally the whole point about no-autoalginit
is to avoid excessive executable sizes when doing static linking. Therefore
we disable "shared" if this option is selected.
Reviewed-by: Richard Levitte <levitte@openssl.org>
During Configure we attempt to check the kernel version of this platform
to see whether we can compile the AFALG engine. If the kernel version
looks recent enough then we enable AFALG. However when we compile
e_afalg.c we check the version of the linux headers. If there is a
mismatch between the linux headers and the currently running kernel then
we don't compile the AFLAG engine and continue. This was causing a link
error.
Reviewed-by: Richard Levitte <levitte@openssl.org>
These algorithms are fundamental and extensively used. The "no-" options
do not work either in 1.1.0 or in other released branches. Therefore the
ability to disable them should be removed.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The check_defer() function was used to ensure that EVP_cleanup() was always
called before OBJ_cleanup(). The new cleanup code ensures this so it is
no longer needed.
Remove obj_cleanup() call in OID config module: it is not needed
any more either.
Reviewed-by: Matt Caswell <matt@openssl.org>
When compressing, the output / input is a binary format, not a text
format like BASE64. This is important on Windows, where a ^Z in a
text file is seen as EOF, and there could be a ^Z somewhere in a
compressed file, cutting it short as input.
Reviewed-by: Matt Caswell <matt@openssl.org>
The handling was Unix centric, already in Configure. Change that to
just collect the value and let the build file templates figure out
what to do with it.
Reviewed-by: Matt Caswell <matt@openssl.org>
Instead of absolute hard coding of the libz library name, have it use
the macro LIBZ, which is set to defaults we know in case it's
undefined.
This allows our configuration to set something that's sane on current
or older platforms, and allows the user to override it by defining
LIBZ themselves.
Reviewed-by: Matt Caswell <matt@openssl.org>
The macros ZLIB and ZLIB_SHARED weren't appropriately defined,
deviating wrongly from how they worked in earlier OpenSSL versions.
So, restore it so that ZLIB is defined if configured "enable-zlib" and
so that ZLIB and ZLIB_SHARED are defined if configured
"enable-zlib-dynamic".
Additionally, correct the interpretation of the --with-zlib-lib value
on Windows and VMS, where it's used to indicate the actual zlib
zlib library file / logical name, as that can differ depending on zlib
version and packaging on those platforms.
Finally for Windows and VMS, we also define the macro LIBZ with that
file name / logical name when configured "zlib-dynamic", so the
compression unit can pick it up and use it.
Reviewed-by: Matt Caswell <matt@openssl.org>
There is a preference for suffixes to indicate that a function is internal
rather than prefixes. Note: the suffix is only required to disambiguate
internal functions and public symbols with the same name (but different
case)
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
The no-op de-init macros may fail because of extraneous ";", so we use
a slightly different construct instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
libssl needs to have access to some internal libcrypto symbols.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
There was a lot of naming inconsistency, so we try and standardise on
one form.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
CONF_modules_free() should not be called expicitly - we should leave
auto-deinit to clean this up instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
ENGINE_cleanup() should not be called expicitly - we should leave
auto-deinit to clean this up instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
OBJ_cleanup() should not be called expicitly - we should leave
auto-deinit to clean this up instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
EVP_cleanup() should not be called expicitly - we should leave
auto-deinit to clean this up instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
BIO_sock_cleanup() should not be called expicitly - we should leave
auto-deinit to clean this up instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
CRYPTO_cleanup_all_ex_data() should not be called expicitly - we should
leave auto-deinit to clean this up instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
RAND_cleanup() should not be called expicitly - we should leave
auto-deinit to clean this up instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
The conditional compile logic wasn't quite right in speed.c for when
both OPENSSL_NO_DSA and OPENSSL_NO_EC are defined.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
SSL_COMP_free_compression_methods() should not be called expicitly - we
should leave auto-deinit to clean this up instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
ERR_free_strings() should not be called expicitly - we should leave
auto-deinit to clean this up instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
COMP_zlib_cleanup() should not be called expicitly - we should leave
auto-deinit to clean this up instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Setting the debug callback on the in and out BIOs were done too early.
No debug callback was set on the zlib filter BIO.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Add an accessor for SSL_CTX.
Since libssl was made opaque, there is no way for users to access the
cipher_list, while users can set the cipher_list by
SSL_CTX_set_cipher_list().
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The function pqueue_print is not exported and is never called. Therefore
we should delete it.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
When a file is opened with BIO_new_file(), make sure that the internal
mode TEXT vs BINARY setting reflects what's given in the mode string.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Code without PEDANTIC has intentional "undefined" behaviour. To get best
coverage for both PEDANTIC and non-PEDANTIC codepaths, run the sanitizer
builds in two different configurations:
1) Without PEDANTIC but with alignment checks disabled.
2) With PEDANTIC.
To not overload Travis too much, run one build with clang and the other
with gcc (chosen at random).
Also remove a micro-optimization in CAST code to be able to
-fsanitize=shift. Whether shift sanitization is meaningful for crypto or
an obstacle is debatable but since this appears to be the only offender,
we might as well keep the check for now.
Reviewed-by: Richard Levitte <levitte@openssl.org>
In Travis, do --strict-warnings on BUILDONLY configurations. This
ensures that the tests run even if --strict-warnings fail, and avoids
hiding unrelated test failures.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The simple reason is that the pre-generated files are mainly for Unix.
The VMS variants look slightly different, so comparing will always fail.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
A number of new functions have been added following the DH and DH_METHOD
opacity commits. This commit provides documentation for those functions.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Move the dh_method structure into an internal header file and provide
relevant accessors for the internal fields.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Move the dh_st structure into an internal header file and provide
relevant accessors for the internal fields.
Reviewed-by: Richard Levitte <levitte@openssl.org>
As it turns out branch hints grew as kind of a misconception. In
addition their interpretation by GNU assembler is affected by
assembler flags and can end up with opposite meaning on different
processors. As we have to loose quite a lot on misinterprerations,
especially on newer processors, we just omit them altogether.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Since NDEBUG is defined unconditionally on command line for release
builds, we can omit *_DEBUG options in favour of effective "all-on"
in debug builds exercised though CI.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Without proper cleanup after DLL link failure second attempt to
run nmake would actually proceed and failure will be "shifted" to
run time. This is because libcrypto.lib import library is generated
even if DLL link fails.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Suppress CT callbacks with aNULL or PSK ciphersuites that involve
no certificates. Ditto when the certificate chain is validated via
DANE-TA(2) or DANE-EE(3) TLSA records. Also skip SCT processing
when the chain is fails verification.
Move and consolidate CT callbacks from libcrypto to libssl. We
also simplify the interface to SSL_{,CTX_}_enable_ct() which can
specify either a permissive mode that just collects information or
a strict mode that requires at least one valid SCT or else asks to
abort the connection.
Simplified SCT processing and options in s_client(1) which now has
just a simple pair of "-noct" vs. "-ct" options, the latter enables
the permissive callback so that we can complete the handshake and
report all relevant information. When printing SCTs, print the
validation status if set and not valid.
Signed-off-by: Rob Percival <robpercival@google.com>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
The original check for != SSL_VERIFY_NONE can give surprising results
when flags SSL_VERIFY_PEER is not set, but other flags are. Note
that SSL_VERIFY_NONE (0) is not a flag bit, it is rather the absense
of all other flag bits.
Signed-off-by: Rob Percival <robpercival@google.com>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Free up parsed X509_NAME structure if the CertificateRequest message
contains excess data.
The security impact is considered insignificant. This is a client side
only leak and a large number of connections to malicious servers would
be needed to have a significant impact.
This was found by libFuzzer.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
The environment variables TOP, SRCTOP, BLDTOP, ... are used to affect
the testing framework. However, subprocesses may want to use them as
well, and therefore need their values corrected when we move to a
different directory.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
We used to symlink generate_ssl_tests.pl to the build directory.
Now that the build scripts look for sources in both directories, this
is no longer necessary (see commit
fbd361eaf8).
Reviewed-by: Richard Levitte <levitte@openssl.org>
Move rsa_meth_st away from public headers.
Add RSA_METHOD creator/destructor functions.
Add RSA_METHOD accessor/writer functions.
Adapt all other source to use the creator, destructor, accessors and writers.
Reviewed-by: Matt Caswell <matt@openssl.org>
Move rsa_st away from public headers.
Add accessor/writer functions for the public RSA data.
Adapt all other source to use the accessors and writers.
Reviewed-by: Matt Caswell <matt@openssl.org>
When config'd with "no-async" the ASYNC_NULL implementation is used, so
async symbols still exist. We should still init the NULL implementation so
that when we get the async ctx it is NULL rather than undefined.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Currently, SSL tests are configured via command-line switches to
ssltest.c. This results in a lot of duplication between ssltest.c and
apps, and a complex setup. ssltest.c is also simply old and needs
maintenance.
Instead, we already have a way to configure SSL servers and clients, so
we leverage that. SSL tests can now be configured from a configuration
file. Test servers and clients are configured using the standard
ssl_conf module. Additional test settings are configured via a test
configuration.
Moreover, since the CONF language involves unnecessary boilerplate, the
test conf itself is generated from a shorter Perl syntax.
The generated testcase files are checked in to the repo to make
it easier to verify that the intended test cases are in fact run; and to
simplify debugging failures.
To demonstrate the approach, min/max protocol tests are converted to the
new format. This change also fixes MinProtocol and MaxProtocol
handling. It was previously requested that an SSL_CTX have both the
server and client flags set for these commands; this clearly can never work.
Guide to this PR:
- test/ssl_test.c - test framework
- test/ssl_test_ctx.* - test configuration structure
- test/handshake_helper.* - new SSL test handshaking code
- test/ssl-tests/ - test configurations
- test/generate_ssl_tests.pl - script for generating CONF-style test
configurations from perl inputs
Reviewed-by: Richard Levitte <levitte@openssl.org>
It was harmless in this case, but best avoid the annoying warnings.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Previously, we updated the static libraries (libcrypto.a on Unix,
libcrypto.lib on Windows) with all the object files, regardless of if
they were rebuilt or not. With this change, we only update them with
the object files were rebuilt.
NOTE: this does not apply on VMS, as the expansion of $? may be too
large for a command line.
Reviewed-by: Andy Polyakov <appro@openssl.org>
A new X509_VERIFY_PARAM_set_auth_level() function sets the
authentication security level. For verification of SSL peers, this
is automatically set from the SSL security level. Otherwise, for
now, the authentication security level remains at (effectively) 0
by default.
The new "-auth_level" verify(1) option is available in all the
command-line tools that support the standard verify(1) options.
New verify(1) tests added to check enforcement of chain signature
and public key security levels. Also added new tests of enforcement
of the verify_depth limit.
Updated documentation.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
Numerous fixups based on feedback of the DSA opacity changes.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
A number of getters/setters have been added for examining DSA objects, as
well as a whole set of functions for creating and buildingup DSA_METHODs.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
Move the dsa_method structure out of the public header file, and provide
getter and setter functions for creating and modifying custom DSA_METHODs.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
Move the dsa_st structure out of the public header file. Add some accessor
functions to enable access to the internal fields, and update all internal
usage to use the new functions.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
Within OpenSSL::Test, all commands end up existing in two variants,
one that has redirections that are needed internally to work well
together with the test harness, and one without those redirections.
Depending on what the result is going to be used for, the caller may
want one for or the other, so we give them the possibility.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Currently on every BIO mem read operation the remaining data is reallocated.
This commit solves the issue.
BIO mem structure includes additional pointer to the read position.
On every read the pointer moves instead of reallocating the memory for the remaining data.
Reallocation accures before write and some ioctl operations, if the read pointer doesn't point on the beginning of the buffer.
Also the flag is added to rewind the read pointer without losing the data.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
InitOnceExecuteOnce returns nonzero on success:
MSDN: "If the function succeeds, the return value is nonzero."
So return 1 if it is nonzero, 0 others.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The logic to find out of there are any .d files newer than Makefile is
sound. Checking the result was less so.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Don't decode a public key in X509_PUBKEY_get0(): that is handled when
the key is parsed using x509_pubkey_decode() instead.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reverts commit 087ca80ad8
Instead of battling the odd format of argv given to main() in default
P64 mode, tell the compiler to make it an array of 64-bit pointers
when compiling in P64 mode.
A note is added in NOTES.VMS regarding minimum DEC C version.
Reviewed-by: Andy Polyakov <appro@openssl.org>
- In Configure, register the perl interpreter used to run Configure,
so that's the one being used throughout instead of something else
that Configure happens to find. This is helpful for using a perl
version that's not necessarely first in $PATH:
/opt/perl/5.22.1/bin/perl ./Configure
- Make apps/tsget a generated file, just like apps/CA.pl, so the
perl interpreter registered by Configure becomes the hashbang path
instead of a hardcoded /usr/bin/perl
Reviewed-by: Andy Polyakov <appro@openssl.org>
This minimizes inter-block overhead. Performance gain naturally
varies from case to case, up to 10% was spotted so far. There is
one thing to recognize, given same circumstances gain would be
higher faster computational part is. Or in other words biggest
improvement coefficient would have been observed with assembly.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Because some operating systems have executable extensions, typically
".exe", we need to append it when looking for files in test() and
app() (or rather, their subroutines).
Reviewed-by: Rich Salz <rsalz@openssl.org>
CRYPTO_mem_leaks attempts to adjust the count of bytes leaks to not
include the BIO that is being used to print the results out. However this
does not work properly. In all internal cases we switch off recording
the memory allocation during creation of the BIO so it makes no difference.
In other cases if the BIO allocates any additional memory during
construction then the adjustment will be wrong anyway. It also skips over
the BIO memory during print_leak anyway, so the BIO memory is never
added into the total. In other words this was broken in lots of ways and
has been since it was first added.
The simplest solution is just to make it the documented behaviour that
you must turn off memory logging when creating the BIO, and remove all
the adjustment stuff completely. The adjustment code was only ever in
master and never made it to a release branch so there is no loss of
functionality.
This commit also fixes a compilation failure when using
enable-crypto-mdebug.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The argument 'argv' in 'main' is a short pointer to a short pointer on
VMS, regardless of initial pointer size. We must therefore make sure
that 'copy_argv' gets a 32-bit pointer for argv, and that the copied
argv is used for the rest of main().
This introduces the local type argv_t, which will have correct pointer
size in all cases (and be harmless on all other platforms) as well as
the macro Argv, which is defined as 'copied_argv' or 'argv', as the
case may be.
Reviewed-by: Andy Polyakov <appro@openssl.org>
setbuf() is only for 32-bit pointers. If compiled with /POINTER_SIZE=64,
we get a nasty warning about possible loss of data. However, since
the only pointer used in the call is a FILE *, and the C RTL shouldn't
give us a pointer above the first 4GB, it's safe to turn off the
warning for this call.
Reviewed-by: Andy Polyakov <appro@openssl.org>
On VMS, the C compiler can work with 32-bit and 64-bit pointers, and
the command line determines what the initial pointer size shall be.
However, there is some functionality that only works with 32-bit
pointers. In this case, it's gethostbyname(), getservbyname() and
accompanying structures, so we need to make sure that we define our
own pointers as 32-bit ones.
Furthermore, there seems to be a bug in VMS C netdb.h, where struct
addrinfo is always defined with 32-bit pointers no matter what, but
the functions handling it are adapted to the initial pointer size.
This leads to pointer size warnings when compiling with
/POINTER_SIZE=64. The workaround is to force struct addrinfo to be
the 64-bit variant if the initial pointer size is 64.
Reviewed-by: Andy Polyakov <appro@openssl.org>
If the command file that app(), test(), perlapp(9 and perltest() are
looking for doesn't exist in the build tree, look for it in the source
tree as well.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Since OpenSSL::Test only redirects stderr to /dev/null when being run
through non-verbose test harness, this change allows the stderr output
to be displayed when verbosity is requested.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Cache the decoded public key when an X509_PUBKEY structure is initially
parsed so no locking is required. Ignore any decode errors.
When an application calls X509_PUBKEY_get0() subsequently it will either
get the cached key or the decode operation will be repeated which will
return an appropriate error.
Reviewed-by: Rich Salz <rsalz@openssl.org>
BIO_f_linebuffer() is now built by default instead of just on VMS, but
the prototype in the header was still only available on VMS.
Reviewed-by: Rich Salz <rsalz@openssl.org>
There are rare cases when an object file will only be used when
building a shared library. To enable this, we introduce
SHARED_SOURCE:
SHARED_SOURCE[libfoo]=dllmain.c
Reviewed-by: Andy Polyakov <appro@openssl.org>
Also, have it always be built, even though it's only (currently) used
on VMS. That will assure it will get the same changes as all others.
Reviewed-by: Matt Caswell <matt@openssl.org>
Previously, it was sufficient to have certSign in keyUsage when the
basicConstraints extension was missing. That is still accepted in
a trust anchor, but is no longer accepted in an intermediate CA.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The API contract is more tight than was previously documented -- the returned
string must be NUL-terminated, and the supplied buffer includes space for
the trailing NUL, so the maximum length that can be read in is reduced.
Clarify that the NUL is not included in the returned length, and fix the
spelling of "NUL-terminated" in a nearby spot.
Adjust punctuation to make a modest improvement to the grammar.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The SSL, SSL_CTX, and SSL_SESSION indices were being referenced
incorrectly in the "_get_ex_new_index" functions.
Remove the STORE EX_DATA index; that functionality is gone.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The warning MAYLOSEDATA3 is one you will always get when compiling
source that calculates the difference between two pointers with
/POINTER_SIZE=64.
The reason is quite simple, ptrdiff_t is always a 32-bit integer
regardless of pointer size, so the result of 'ptr1 - ptr2' can
potentially be larger than a 32-bit integer. The compiler simply
warns you of that possibility.
However, we only use pointer difference within objects and strings,
all of them well within 2^32 bytes in size, so that operation is
harmless with our source, and we can therefore safely turn off that
warning.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The private key is a scalar and should be sized by the order, not the
degree. See RFC 5915.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Move the the BIO_METHOD and BIO structures into internal header files,
provide appropriate accessor methods and update all internal code to use
the new accessors where appropriate.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Session resumption involves a version check, so version negotiation must
happen first. Currently, the DTLS implementation cannot do session
resumption in DTLS 1.0 because the ssl_version check always checks
against 1.2.
Switching the order also removes the need to fixup ssl_version in DTLS
version negotiation.
Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
RT: #4392, MR: #2452
We now send the highest supported version by the client, even if the session
uses an older version.
This fixes 2 problems:
- When you try to reuse a session but the other side doesn't reuse it and
uses a different protocol version the connection will fail.
- When you're trying to reuse a session with an old version you might be
stuck trying to reuse the old version while both sides support a newer
version
Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
GH: #852, MR: #2452
According to documentation, perl's Math::BigInt does floored division,
i.e. the bdiv function does 1 / -4 = -1. OpenSSL's BN_div, as well as
bc, do truncated division, i.e. 1 / -4 = 0.
We need to compensate for that difference in test/recipes/bc.pl to
make sure to verify the bntest results under its own conditions, by
dividing the absolute values of the given numbers and fixup the
result's negativity afterwards.
Closes RT#4485
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
These logical names are used to make the C RTL change certain
behaviors, sometimes to make it act more like Unix. While they can
make life easier in some cases, they can be disruptive as well. When
building and testing OpenSSL, the latter is the case, so we ask people
to avoid using them.
Reviewed-by: Tim Hudson <tjh@openssl.org>
On Windows and Unix, the staging directory $(DESTDIR) can simply be
prepended to the installation directory. An attempt was made to do
something similar on VMS, but that ended up being a half measure
solution. Instead of that, simply use the staging directory as a
prefix under which [.OPENSSL-INSTALL] and [.OPENSSL-COMMON] will hold
the two directory trees that should end up in the directories
indicated by --prefix and --openssldir, and finish the installation
with appropriate instructions on what to do next.
Reviewed-by: Rich Salz <rsalz@openssl.org>
- on VMS, SYS$COMMON:[SSL] is already used as installation directory
by HP SSL, so we make our default for --openssldir
SYS$COMMON:[OPENSSL-COMMON] instead.
- Updated notes on default installation dirs fir Unix and Windows
Reviewed-by: Rich Salz <rsalz@openssl.org>
In this OpenSSL version, we deliver engines with lower case symbol
names. The DSO symbol finder must be updated to allow for mixed case
symbols or it won't fine them.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Sometimes, you might end up with a rather long compile line due to
excessively long /INCLUDE directories. Compensate for it by making
a temporary logical name with them and using said logical name as
/INCLUDE argument.
A note was added to NOTES.VMS regarding these limitations.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Instead of have every DSO_METHOD_xxx in all platforms, ensure that only
one DSO_METHOD_openssl is available on all platforms.
Reviewed-by: Tim Hudson <tjh@openssl.org>
BIO_new, etc., don't need a non-const BIO_METHOD. This allows all the
built-in method tables to live in .rodata.
Reviewed-by: Richard Levitte <levitte@openssl.org>
On VMS, we downcase option names, which means that config names are
downcased as well, so they need to be downcased in the target table to
be found.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Allows CONF files for certificate requests to specify that a pre-
certificate should be created (see RFC6962).
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
In constructions such as 'for x in $(MAKEVAR); do ...', there's the
possibility that $(MAKEVAR) is en empty value. Some shells don't like
that, so introduce a dummy value that gets discarded:
for x in dummy $(MAKEVAR); do
if [ "$$x" = "dummy" ]; then continue; fi
Closes RT#4459
Reviewed-by: Rich Salz <rsalz@openssl.org>
utils/mkrc.pl was added a while ago as a better generator for the
Windows DLL resource file. Finalize the change by removing the
ms/version32.rc generator from Configure and adding resource file
support using mkrc.pl in Configurations/windows-makefile.pl
Reviewed-by: Andy Polyakov <appro@openssl.org>
The mk1mf build for the VC-WIN* targets is broken and the unified
scheme works well enough, so we clean out the old.
Reviewed-by: Andy Polyakov <appro@openssl.org>
* Clear proposed, along with selected, before looking at ClientHello
* Add test case for above
* Clear NPN seen after selecting ALPN on server
* Minor documentation updates
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Don't have #error statements in header files, but instead wrap
the contents of that file in #ifndef OPENSSL_NO_xxx
This means it is now always safe to include the header file.
Reviewed-by: Richard Levitte <levitte@openssl.org>
By default you get 0.9 which isn't widely available.
But we use HTTP/1.0 for now.
Courtesy beusink@users.github.com
Reviewed-by: Tim Hudson <tjh@openssl.org>
One of the 'generate' targets depended on $(SRCDIR)/apps/progs.h,
which depended on... nothing. This meant it never got regenerated
once it existed, regardless of need. Of course, we could have it
depend on all the files checked to generate it, but they also depend
on progs.h, so we'd end up getting cricular dependencies, which makes
make unhappy.
Furthermore, and this applies for the other generated files, having
them as targets means that they may be regenerated on the fly in some
cases, and since they get written to the source tree, this isn't such
a good idea if that tree is read-only (which is a possible situation
in an out-of-tree build).
So, we move all the actions to the 'generate' targets themselves, thus
making sure they get regenerated in a controlled manner and regardless
of dependencies.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Some implementations of sed require a newline before an ending '}'.
The easier method is to replace that sed command with the
corresponding perl command.
Closes RT#4448
Reviewed-by: Rich Salz <rsalz@openssl.org>
Depending on what has been built so far, all .d files may not be
present and 'find' will exit with non-zero exit code. This isn't a
bother for us but may break make, so clear the exit code with an added
'exit 0'.
Closes RT#4444
Reviewed-by: Rich Salz <rsalz@openssl.org>
Don't free up passed EVP_MD_CTX in ASN1_item_sign_ctx(). This
simplifies handling and retains compatiblity with previous behaviour.
PR#4446
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Stephen Henson <steve@openssl.org>
generatesrc() did already receive dependency information, but never
used it, and never really needed to... until now.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Instead of relying on the '-nt' test operator, which doesn't exist
everywhere, use find's '-newer' to find out if any of the known .d
files is newer than Makefile.
Closes RT#4444
Reviewed-by: Andy Polyakov <appro@openssl.org>
The no-deprecated build was failing because afalg was relying on a
transitive include that does not exist in a no-deprecated build.
Reviewed-by: Rich Salz <rsalz@openssl.org>
$? in perl gets the status value from wait(2), which is a word with
the exit code in the upper half and the number of a raised signal in
the lower half. OpenSSL::Test::run() ignored the signal half up until
now.
With this change, we recalculate an exit code the same way the Unix
shells do, using this formula:
($? & 0x7f) ? ($? & 0x7f)|0x80 : ($? >> 8);
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
mkdef.pl was not detecting no-comp functions. This updates the header file
so that mkdef.pl detects that no-comp applies, and the functions are marked
accordingly.
Reviewed-by: Richard Levitte <levitte@openssl.org>
If a call to EVP_DecryptUpdate fails then a memory leak could occur.
Ensure that the memory is freed appropriately.
Issue reported by Guido Vranken.
Reviewed-by: Richard Levitte <levitte@openssl.org>
There is a potential double free in EVP_DigestInit_ex. This is believed
to be reached only as a result of programmer error - but we should fix it
anyway.
Issue reported by Guido Vranken.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Don't refer to INSTALL files that don't exist any more.
Make it clear that the libssl and libcrypto libraries may be named
differently on different platforms.
Reviewed-by: Rich Salz <rsalz@openssl.org>
On Windows, we set INSTALLTOP to default as follows:
VC-WIN32:
PREFIX: %ProgramFiles(x86)%\OpenSSL
OPENSSLDIR: %CommonProgramFiles(x86)%\SSL
VC-WIN64*:
PREFIX: %ProgramW6432%\OpenSSL
OPENSSLDIR: %CommonProgramW6432%\SSL
Should those environment variables be missing, the following is used
as fallback:
PREFIX: %ProgramFiles%\OpenSSL
OPENSSLDIR: %CommonProgramFiles%\SSL
Reviewed-by: Andy Polyakov <appro@openssl.org>
This is a living document, everyone is encouraged to add to it.
Implementation details as well as broader implementation philosophy
has a place here.
I'm starting with documentation of the how conditions in build.info
files are treated.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This is only enabled when the environment variable
CONFIGURE_DEBUG_BUILDINFO is defined. This will cause every line in
every build.info file to be displayed, along with the content of the
skip stack before and after parsing. This can be a very powerful tool
to see that all conditions are working as expected.
Reviewed-by: Rich Salz <rsalz@openssl.org>
In most builds, we can assume that engines live in the build tree
subdirectory "engines". This was hard coded into the tests that use
the engine ossltest.
However, that hard coding is tedious, it would need to be done in
every test recipe, and it's an incorrect assumption in some cases.
This change has us play it safe and let the build files tell the
testing framework where the engines are.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Some platforms claim to be POSIX but their getcontext() implementation
does not work. Therefore we update the ASYNC_is_capable() function to test
for this.
RT#4366
Reviewed-by: Richard Levitte <levitte@openssl.org>
util/mkdef.pl assumes it knows what the resulting library name will
be. Really, it shouldn't, but changing it will break classic native
Windows builds, so we leave it for now and change the LIBRARY line
externally when needed instead.
Reviewed-by: Matt Caswell <matt@openssl.org>
- Remove duplicate entry
- Add author for SSL_CIPHER query functions
- Note HKDF support in CHANGES and NEWS
[ci skip]
Reviewed-by: Richard Levitte <levitte@openssl.org>
Normally we always refer to source files relative to $SRCDIR in Makefiles.
However the reference to unix-Makefile.tmpl was using a fully expanded
absolute path. This can cause problems for Mingw.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
We don't currently support cross-compiling of the afalg engine. However
we were failing to explicitly mark it as disabled during Configure leading
to a failed build.
Reviewed-by: Richard Levitte <levitte@openssl.org>
TLSProxy starts s_server and specifies the number of client connects
it should expect. After that s_server is supposed to close down
automatically. However, if another test is then run then TLSProxy
will start a new instance of s_server. If the previous instance
hasn't closed down yet then the new instance can fail to bind to
the socket.
Reviewed-by: Richard Levitte <levitte@openssl.org>
In unix-Makefile.tmpl, this construction has been used a few times
if ! something; then ...
It seems, though, that some shells do not understand !, so these need
to be changed.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This commit adds the general verify options of ocsp, verify,
cms, etc. to the openssl timestamping app as suggested by
Stephen N. Henson in [openssl.org #4287]. The conflicting
"-policy" option of "openssl ts" has been renamed to
"-tspolicy". Documentation and tests have been updated.
CAVE: This will break code, which currently uses the "-policy"
option.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
There are internal dependencies between the various cleanup functions.
This re-orders things to try and get that right.
Reviewed-by: Richard Levitte <levitte@openssl.org>
OBJ_cleanup() doesn't always get called from EVP_cleanup() so needs to be
explicitly called in de-init. Also BIO_sock_cleanup() also needs to be
called.
Reviewed-by: Richard Levitte <levitte@openssl.org>
With bash and zsh, the trap on the 5 second read does respond, but
doesn't break out of the read. What's worse is that it takes away the
5 second timer, and therefore has the read hang indefinitely and
(almost) unbreakable.
Having the trap do 'exit 0' after reseting the tty params has it break
out of read and continue with the configuration.
Other shells do not appear to have the issue described here, but
neither does the extra 'exit 0' appear to harm them.
Reviewed-by: Andy Polyakov <appro@openssl.org>
The reason to do so is that some of the generators detect PIC flags
like -fPIC and -KPIC, and those are normally delivered in LD_CFLAGS.
Reviewed-by: Rich Salz <rsalz@openssl.org>
When passing down values to Makefile.shared, do so with single quotes
as much as possible to avoid having the shell create a mess of quotes.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The variable SHARED_CFLAGS and SHARD_LDFLAGS were used in the Unix
template because they normally contain options used when building
"shared". The Windows template, on the other hand, uses LIB_CFLAGS,
to express the intended use of those flags rather than their content.
The Windows template still used SHARED_LDFLAGS, which seems
inconsistent.
To harmonize the two, any SHARED_CFLAGS gets renamed to LIB_CFLAGS and
SHARED_LDFLAGS to LIB_LDFLAGS. That makes the intent consistent along
with BIN_{C,LD}FLAGS and DSO_{C,LD}FLAGS.
Finally, make sure to pass down $(LIB_CFLAGS) or $(DSO_CFLAGS) along
with $(CFLAGS) when using Makefile.shared.
Reviewed-by: Rich Salz <rsalz@openssl.org>
ENGINE_cleanup calls CRYPTO_free_ex_data and therefore,
CRYPTO_cleanup_all_ex_data - which cleans up the method pointers - must
run after ENGINE_cleanup.
Additionally, don't needlessly initialize the EX_CALLBACKS stack during
e.g. CRYPTO_free_ex_data. The only time this is actually needed is when
reserving the first ex data index. Specifically, since sk_num returns -1
on NULL input, the rest of the code already handles a NULL method stack
correctly.
Reviewed-by: Rich Salz <rsalz@openssl.org>
ccache + clang produces a false strcmp warning, see
https://llvm.org/bugs/show_bug.cgi?id=20144
Since this only happens with ccache and --strict-warnings, and
only with certain versions of glibc / clang, disabling
ccache is a reasonable short-term workaround.
Reviewed-by: Richard Levitte <levitte@openssl.org>
While insignificant on Unix like systems, this is significant on
systems like VMS.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
It turns out that different sed implementations treat -i differently
to cause issues. make it simpler by avoiding it entirely and give
perl the trust to be consistent enough.
Reviewed-by: Rich Salz <rsalz@openssl.org>
On Windows we call WSAGetLastError() to find out the last error that
happened on a socket operation. We use this to find out whether we can
retry the operation or not. You are supposed to call this immediately
however in a couple of places we logged an error first. This can end up
making other Windows system calls to get the thread local error state.
Sometimes that can clobber the error code, so if you call WSAGetLastError()
later on you get a spurious response and the socket operation looks like
a fatal error.
Really we shouldn't be logging an error anyway if its a retryable issue.
Otherwise we could end up with stale errors on the error queue.
Reviewed-by: Richard Levitte <levitte@openssl.org>
If pre-processor failed, an empty .s file could be left behind,
which could get successfully compiled if one simply re-ran make
and cause linking failures. Not anymore. Remove even intermediate .S
in case of pre-processor failure.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Running test_ssl with HARNESS_VERBOSE results in lots of spurious warnings
about an inability to load the CT config file. This fixes it.
Reviewed-by: Richard Levitte <levitte@openssl.org>
During auto de-init we were calling ENGINE_cleanup(), and then later
CONF_modules_free(). However the latter function can end up calling
engine code, which can lead to a use of the global_engine_lock after it
has already been freed. Therefore we should swap the calling order of
these two functions.
Reviewed-by: Rich Salz <rsalz@openssl.org>
no-ripemd was unified a while ago but the change was not done in mk1mf.pl. This commit changes the no-ripemd string to no-rmd160 which fixes the no-rmd160 classic build.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Experience has shown that dynamic engines with their own copy of
libcrypto is problematic, so we disable that possibility.
Reviewed-by: Matt Caswell <matt@openssl.org>
One of them didn't clean away .d.tmp files properly.
The other would overwrite the .d files unconditionally, thereby
causing a possibly unnecessary dependency rebuild, which touches the
date of Makefile, which causes a possibly unnecessary rebuild of
buildinf.h and everything that depends on that.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Some of these scripts would recognise an output parameter if it looks
like a file path. That works both in both the classic and new build
schemes. Some fo these scripts would only recognise it if it's a
basename (i.e. no directory component). Those need to be corrected,
as the output parameter in the new build scheme is more likely to
contain a directory component than not.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Pass entire CTLOG_STORE to SCT_print, rather than just the SCT's CTLOG
SCT_print now looks up the correct CT log for you.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Remove 'log' field from SCT and related accessors
In order to still have access to an SCT's CTLOG when calling SCT_print,
SSL_CTX_get0_ctlog_store has been added.
Improved documentation for some CT functions in openssl/ssl.h.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The AFALG engine created a global EVP_CIPHER instance but was not freeing
it up when the engine was destroyed.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The help text for -d in 'config' was aged, and the option processing
in 'config.com' was just different. This harmonizes 'config.com' with
the instructions in INSTALL and both current reality.
Reviewed-by: Rich Salz <rsalz@openssl.org>
UEFI needs this too. Don't keep it only in the Windows/DOS ifdef block.
This is a fixed version of what was originally commit 963bb62195 and
subsequently reverted in commit 37b1f8bd62. Somewhere along the way, the
Windows/DOS ifdef actually got removed, leaving it just broken. It should
have been turned into an #elif, not removed.
This one correctly changes the logic from
# if WINDOWS|DOS
# if OPENSSL_NO_SOCK
... no-sock ...
# elif !DJGPP
... native windows ...
to
# if OPENSSL_NO_SOCK
... no-sock ...
# elif WINDOWS|DOS
# if !DJGPP
... native windows ...
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
The source file generators sometimes use $(CC) to post-process
generated source, and getting the inclusion directories may be
necessary at times, so we pass them down.
RT#4406
Reviewed-by: Rich Salz <rsalz@openssl.org>
Although theoretically possible, Configure doesn't treat CC variable
set like this very well: CC="ccache i686-w64-mingw32-gcc"
Also, this Travis script doesn't recognise the possibility either.
Reviewed-by: Rich Salz <rsalz@openssl.org>
BIO_snprintf() can return -1 on truncation (and overflow as of commit
9cb177301f). Though neither can
realistically occur while printing a pointer and short fixed string into
a buffer of length 256, the analysis to confirm that this the case goes
somewhat far up the call chain, and not all static analyzers can
successfully follow the chain of logic.
It's easy enough to clamp the returned length to be nonnegative before
continuing, which appeases the static analyzer and does not harm the
subsequent code.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Some of the ASN.1 routines for the GeneralizedTime type can return
errors; check for these and do not continue past failure, so as
to appease coverity.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The reason for this is that the static libraries and the DLL import
libraries are named the same on Windows. When configured "shared",
the static libraries are unused anyway.
Reviewed-by: Rich Salz <rsalz@openssl.org>
set pointers to NULL after OPENSSL_free before returning to caller to
avoid possible double-free in caller
Signed-off-by: Rich Salz <rsalz@akamai.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
This corrects a fault where the inner IF in this example was still
being acted upon:
IF[0]
...whatever...
IF[1]
...whatever more...
ENDIF
ENDIF
With this change, the inner IF is skipped over.
Reviewed-by: Matt Caswell <matt@openssl.org>
Adjust ssl_set_client_hello_version to get both the minimum and maximum and then
make ssl_set_client_hello_version use the maximum version.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
MR: #1595
The "extra checks" is a debugging tool to check the config resolving
mechanism. It uses Perl's smart match, which is experimental and
therefore always causes Perl to give out a warning, and it causes
older Perl versions to fail entirely.
So, it gets commented away, but stays otherwise in place, as it may be
useful again.
Reviewed-by: Matt Caswell <matt@openssl.org>
ct_test assumed it's run in the source directory and failed when built
elsewhere. It still defaults to that, but can be told another story
with the environment variables CT_DIR and CERTS_DIR.
Test recipe updated to match.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Both of these functions can easily be implemented by callers instead.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Use "!x" instead of "x <= 0", as these functions never return a negative
value.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
No longer terminates on first error, but instead tries to set the source
of every SCT regardless of whether an error occurs with some.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
ct_test assumed it's run in the source directory and failed when built
elsewhere. It still defaults to that, but can be told another story
with the environment variables CT_DIR and CERTS_DIR.
Test recipe updated to match.
Reviewed-by: Matt Caswell <matt@openssl.org>
Update the CHANGES and NEWS files with information about the recently added
AFALG engine and pipelining.
Reviewed-by: Richard Levitte <levitte@openssl.org>
All OpenSSL code has now been transferred to use the new threading API,
so the old one is no longer used and can be removed. We provide some compat
macros for removed functions which are all no-ops.
There is now no longer a need to set locking callbacks!!
Reviewed-by: Richard Levitte <levitte@openssl.org>
The locking here is a bit strange and unclear. Rather than refactor
anything and possibly break stuff I have just moved to using the new
thread API following as closely as possible what was there previously.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Before the 'Introduce the "pic" / "no-pic" config option' commit, the
shared_cflag value for the chosen config would be part of the make
variable CFLAG, which got replicated into CFLAGS and ASFLAGS.
Since said commit, the shared_cflag value has become a make variable
of its own, SHARED_CFLAG (which is left empty in a "no-pic" build).
However, ASFLAGS was forgotten. That's what's corrected with this
change.
Reviewed-by: Andy Polyakov <appro@openssl.org>
If there is cause to think LD_LIBRARY_PATH_32 and LD_PRELOAD_32 are
appropriate variables to touch, do so. Otherwise, touch the usual
LD_LIBRARY_PATH and LD_PRELOAD. This covers for older installations
that don't have a mix of 32-bit and 64-bit libs.
Reviewed-by: Andy Polyakov <appro@openssl.org>
According to manuals found here: https://gcc.gnu.org/onlinedocs/, GNU
C version 3 and on support the dependency generation options. We
therefore need to check the gcc version to see if we're going to use
it or makedepend for dependency generation.
Reviewed-by: Andy Polyakov <appro@openssl.org>
crypto/evp/e_aes.c and crypto/modes/gcm128.c include ppc_arch.h, which
is located in crypto/, so add that as extra include directory for them.
Issue reported by Jeffrey Walton <noloader@gmail.com>
Reviewed-by: Andy Polyakov <appro@openssl.org>
This introduces the settings loutflag and aroutflag, because different
Windows tools that do the same thing have different ways to specify
the output file.
The Borland C++ config is commented away for the monent, perhaps
permanently.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/whrlpool/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/sha/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/rc4/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/rc5/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/ripemd/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/md5/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/modes/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/poly1305/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/des/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/ec/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/camellia/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/cast/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/chacha/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/aes/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/bf/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Even if we haven't loaded an engine, we might have set up the
global_engine_lock, so we should still clean up.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Move the chil engine to use the new thread API. As I don't have access to
the hardware I can't test this :-(. I think its ok...
Reviewed-by: Richard Levitte <levitte@openssl.org>
The new Rand usage of Thread API exposed a bug in ssltest. ssltest "cheats"
and uses internal headers to directly call functions that normally you
wouldn't be able to do. This means that auto-init doesn't happen, and
therefore auto-deinit doesn't happen either, meaning that the new rand locks
don't get cleaned up properly.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This change is a bit more complex, as it involves several recipe
variants.
Also, remove the $(CROSS_COMPILE) prefix for the makedepend program.
When we use the program "makedepend", this doesn't serve anything,
and when we use the compiler, this value isn't even used.
Reviewed-by: Rich Salz <rsalz@openssl.org>
VMS doesn't have "makedepend" anyway, so this is just a matter of using
the right qualifiers when 'makedepend' is enabled.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Copy/paste error between SSL_CIPHER_get_kx_nid() and
SSL_CIPHER_get_auth_nid(), wrong table was referenced
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
The Engine API lost the setting of memory management hooks in
bind_engine. Here's putting that back.
EX_DATA and ERR functions need the same treatment.
Reviewed-by: Matt Caswell <matt@openssl.org>
For assembler, we want the final target to be foo.s (lowercase s).
However, the build.info may have lines like this (note upper case S):
GENERATE[foo.S]=foo.pl
This indicates that foo.s (lowercase s) is still to be produced, but
that producing it will take an extra step via $(CC) -E. Therefore,
the following variants (simplified for display) can be generated:
GENERATE[foo.S]=foo.pl => foo.s: foo.pl
$(PERL) $foo.pl $@.S; \
$(CC) $(CFLAGS) -E -P $@.S > $@ && \
rm -f $@.S
GENERATE[foo.s]=foo.pl => foo.s: foo.pl
$(PERL) $foo.pl $@
GENERATE[foo.S]=foo.m4 => foo.s: foo.m4
m4 -B 8192 $foo.m4 > $@.S; \
$(CC) $(CFLAGS) -E -P $@.S > $@ && \
rm -f $@.S
GENERATE[foo.s]=foo.m4 => foo.s: foo.m4
m4 -B 8192 $foo.m4 > $@
Reviewed-by: Andy Polyakov <appro@openssl.org>
Compiling ssltest with some compilers using --strict-warnings results in
complaints about an unused result.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
This adds a new accessor function DSA_SIG_get0.
The customisation of DSA_SIG structure initialisation has been removed this
means that the 'r' and 's' components are automatically allocated when
DSA_SIG_new() is called. Update documentation.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Make all scripts produce .S, make interpretation of $(CFLAGS)
pre-processor's responsibility, start accepting $(PERLASM_SCHEME).
[$(PERLASM_SCHEME) is redundant in this case, because there are
no deviataions between Solaris and Linux assemblers. This is
purely to unify .pl->.S handling across all targets.]
Reviewed-by: Richard Levitte <levitte@openssl.org>
* Perform ALPN after the SNI callback; the SSL_CTX may change due to
that processing
* Add flags to indicate that we actually sent ALPN, to properly error
out if unexpectedly received.
* clean up ssl3_free() no need to explicitly clear when doing memset
* document ALPN functions
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
This is similar to SSL_pending() but just returns a 1 if there is data
pending in the internal OpenSSL buffers or 0 otherwise (as opposed to
SSL_pending() which returns the number of bytes available). Unlike
SSL_pending() this will work even if "read_ahead" is set (which is the
case if you are using read pipelining, or if you are doing DTLS). A 1
return value means that we have unprocessed data. It does *not* necessarily
indicate that there will be application data returned from a call to
SSL_read(). The unprocessed data may not be application data or there
could be errors when we attempt to parse the records.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
This capability is required for read pipelining. We will only read in as
many records as will fit in the read buffer (and the network can provide
in one go). The bigger the buffer the more records we can process in
parallel.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
* Perform ALPN after the SNI callback; the SSL_CTX may change due to
that processing
* Add flags to indicate that we actually sent ALPN, to properly error
out if unexpectedly received.
* clean up ssl3_free() no need to explicitly clear when doing memset
* document ALPN functions
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
For config targets such as 'dist', which doesn't have a BASE template,
we still need to have a default build scheme.
Additionally, the unified Makefile template's target 'tar' wasn't
quite as flexible as the unixmake one.
Finally, .travis-create-release.sh can be somewhat simplified now that
it builds with the unified build scheme.
Reviewed-by: Matt Caswell <matt@openssl.org>
We want to preserve a couple of classic builds still, the quickest is
to change all --unified to --classic
Reviewed-by: Matt Caswell <matt@openssl.org>
Because of the unified scheme, building on different platforms is very
similar. We currently have Unix and OpenVMS on the unified scheme,
which means that a separate INSTALL.VMS is no longer needed.
Reviewed-by: Matt Caswell <matt@openssl.org>
The kinv/r fields in the DSA structure are not used by OpenSSL internally
and should not be used in general.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The numpipes argument to ssl3_enc/tls1_enc is actually the number of
records passed in the array. To make this clearer rename the argument to
|n_recs|.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Rename the have_whole_app_data_record_waiting() function to include the
ssl3_record prefix...and make it a bit shorter.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Move out most of the boiler plate code that is common between aes128-cbc
and aes128-cbc-hmac-sha1 into helper functions to improve code reuse.
Reviewed-by: Tim Hudson <tjh@openssl.org>
The _hidden_* variables were being created on-the-fly. It is better to
create them once up front during bind to avoid any potential race
conditions.
Reviewed-by: Tim Hudson <tjh@openssl.org>
We had the function EVP_CIPHER_CTX_cipher_data which is newly added for
1.1.0. As we now also need an EVP_CIPHER_CTX_set_cipher_data it makes
more sense for the former to be called EVP_CIPHER_CTX_get_cipher_data.
Reviewed-by: Tim Hudson <tjh@openssl.org>
The new pipeline code added a new function
EVP_CIPHER_CTX_set_cipher_data(). Add documentation for this and the
existing EVP_CIPHER_CTX_cipher_data() function.
Reviewed-by: Tim Hudson <tjh@openssl.org>
We used to use the wrec field in the record layer for keeping track of the
current record that we are writing out. As part of the pipelining changes
this has been moved to stack allocated variables to do the same thing,
therefore the field is no longer needed.
Reviewed-by: Tim Hudson <tjh@openssl.org>
A previous commit added the SSL_has_pending() function which provides a
method for knowing whether OpenSSL has buffered, but as yet unprocessed
record data.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Previously s_client and s_server relied on using SSL_pending() which does
not take into account read_ahead. For read pipelining to work, read_ahead
gets set automatically. Therefore s_client and s_server have been
converted to use SSL_has_pending() instead.
Reviewed-by: Tim Hudson <tjh@openssl.org>
This is similar to SSL_pending() but just returns a 1 if there is data
pending in the internal OpenSSL buffers or 0 otherwise (as opposed to
SSL_pending() which returns the number of bytes available). Unlike
SSL_pending() this will work even if "read_ahead" is set (which is the
case if you are using read pipelining, or if you are doing DTLS). A 1
return value means that we have unprocessed data. It does *not* necessarily
indicate that there will be application data returned from a call to
SSL_read(). The unprocessed data may not be application data or there
could be errors when we attempt to parse the records.
Reviewed-by: Tim Hudson <tjh@openssl.org>
This capability is required for read pipelining. We will only read in as
many records as will fit in the read buffer (and the network can provide
in one go). The bigger the buffer the more records we can process in
parallel.
Reviewed-by: Tim Hudson <tjh@openssl.org>
With read pipelining we use multiple SSL3_RECORD structures for reading.
There are SSL_MAX_PIPELINES (32) of them defined (typically not all of these
would be used). Each one has a 16k compression buffer allocated! This
results in a significant amount of memory being consumed which, most of the
time, is not needed. This change swaps the allocation of the compression
buffer to be lazy so that it is only done immediately before it is actually
used.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Read pipelining is controlled in a slightly different way than with write
pipelining. While reading we are constrained by the number of records that
the peer (and the network) can provide to us in one go. The more records
we can get in one go the more opportunity we have to parallelise the
processing.
There are two parameters that affect this:
* The number of pipelines that we are willing to process in one go. This is
controlled by max_pipelines (as for write pipelining)
* The size of our read buffer. A subsequent commit will provide an API for
adjusting the size of the buffer.
Another requirement for this to work is that "read_ahead" must be set. The
read_ahead parameter will attempt to read as much data into our read buffer
as the network can provide. Without this set, data is read into the read
buffer on demand. Setting the max_pipelines parameter to a value greater
than 1 will automatically also turn read_ahead on.
Finally, the read pipelining as currently implemented will only parallelise
the processing of application data records. This would only make a
difference for renegotiation so is unlikely to have a significant impact.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Add the options min_send_frag and max_pipelines to s_server and s_client
in order to control pipelining capabilities. This will only have an effect
if a pipeline capable cipher is used (such as the one provided by the
dasync engine).
Reviewed-by: Tim Hudson <tjh@openssl.org>
Use the new pipeline cipher capability to encrypt multiple records being
written out all in one go. Two new SSL/SSL_CTX parameters can be used to
control how this works: max_pipelines and split_send_fragment.
max_pipelines defines the maximum number of pipelines that can ever be used
in one go for a single connection. It must always be less than or equal to
SSL_MAX_PIPELINES (currently defined to be 32). By default only one
pipeline will be used (i.e. normal non-parallel operation).
split_send_fragment defines how data is split up into pipelines. The number
of pipelines used will be determined by the amount of data provided to the
SSL_write call divided by split_send_fragment. For example if
split_send_fragment is set to 2000 and max_pipelines is 4 then:
SSL_write called with 0-2000 bytes == 1 pipeline used
SSL_write called with 2001-4000 bytes == 2 pipelines used
SSL_write called with 4001-6000 bytes == 3 pipelines used
SSL_write_called with 6001+ bytes == 4 pipelines used
split_send_fragment must always be less than or equal to max_send_fragment.
By default it is set to be equal to max_send_fragment. This will mean that
the same number of records will always be created as would have been
created in the non-parallel case, although the data will be apportioned
differently. In the parallel case data will be spread equally between the
pipelines.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Implement aes128-cbc as a pipeline capable cipher in the dasync engine.
As dasync is just a dummy engine, it actually just performs the parallel
encrypts/decrypts in serial.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Add a flag to indicate that a cipher is capable of performing
"pipelining", i.e. multiple encrypts/decrypts in parallel. Also add some
new ctrls that ciphers will need to implement if they are pipeline capable.
Reviewed-by: Tim Hudson <tjh@openssl.org>
- Only build & test two configurations. Make all the
other build variants buildonly on gcc (clang on osx).
- Don't build with default clang at all on linux.
- Only use gcc-5 and clang-3.6 for the sanitizer builds. Re-running
e.g. CONFIG_OPTS="shared" with them seems redundant.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Strictly speaking, it isn't stdio and file access which offend me here;
it's the fact that UEFI doesn't provide a strdup() function. But the
fact that it's pointless without file access is a good enough excuse for
compiling it out.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
UEFI needs this too. Don't keep it only in the Windows/DOS ifdef block.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Without this, the peer certificate would never be deleted, resulting in
a memory leak.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
It was unexpected that OpenSSL::Test::setup() should be called twice
by the same recipe. However, that may happen if a recipe combines
OpenSSL::Test and OpenSSL::Test::Simple, which can be a sensible thing
to do. Therefore, we now allow it.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in engines/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This gets rid of the BEGINRAW..ENDRAW sections in crypto/bn/build.info.
This also moves the assembler generating perl scripts to take the
output file name as last command line argument, where necessary.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Should it be needed because the recipes within a RAW section might
clash with those generated by Configure, it's possible to tell it
not to generate them with the use of OVERRIDES, for example:
SOURCE[libfoo]=foo.c bar.c
OVERRIDES=bar.o
BEGINRAW[Makefile(unix)]
bar.o: bar.c
$(CC) $(CFLAGS) -DSPECIAL -c -o $@ $<
ENDRAW[Makefile(unix)]
Reviewed-by: Rich Salz <rsalz@openssl.org>
In some cases, one might want to generate some source files from
others, that's done as follows:
GENERATE[foo.s]=asm/something.pl $(CFLAGS)
GENERATE[bar.s]=asm/bar.S
The value of each GENERATE line is a command line or part of it.
Configure places no rules on the command line, except the the first
item muct be the generator file. It is, however, entirely up to the
build file template to define exactly how those command lines should
be handled, how the output is captured and so on.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Swap the use of CRYPTO_LOCK_INIT in the init code to use the new threading
API mechanism for locking.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The init code was using its own thread local code. Now we have a central
API for it we should use that instead.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The init code was using its own "once" implementation. Now that we have
the new thread API we should use that instead.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Make PKCS8_PRIV_KEY_INFO opaque. Several accessor functions already exist
for this structure. Two new ones were added to handle attributes.
The old handling of broken formats has been removed and the corresponding
structures simplified.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Move RSA struct in the job local struct.
The change is applied also to other crypto operations (e.g. DSA) to
make things consistent.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Summary of the changes:
* Move the calls to the crypto operations inside wrapper functions.
This is required because ASYNC_start_job takes a function as an argument.
* Add new function run_benchmark() that manages the jobs for all the operations.
In the POSIX case it uses a select() to receive the events from the engine
and resume the jobs that are paused, while in the WIN case it uses PeekNamedPipe()
* Add new option argument async_jobs to enable and specify the number of async jobs
Example:
openssl speed -engine dasync -elapsed -async_jobs 32 rsa2048
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
- Always prefer forward-secure handshakes.
- Consistently order ECDSA above RSA.
- Next, always prefer AEADs to non-AEADs, irrespective of strength.
- Within AEADs, prefer GCM > CHACHA > CCM for a given strength.
- Prefer TLS v1.2 ciphers to legacy ciphers.
- Remove rarely used DSS, IDEA, SEED, CAMELLIA, CCM from the default
list to reduce ClientHello bloat.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Some OSes, *cough*-dows, insist on stack being "wired" to
physical memory in strictly sequential manner, i.e. if stack
allocation spans two pages, then reference to farmost one can
be punishable by SEGV. But page walking can do good even on
other OSes, because it guarantees that villain thread hits
the guard page before it can make damage to innocent one...
Reviewed-by: Rich Salz <rsalz@openssl.org>
"no-pic" builds have in fact been green (and reasonably fast), so
restore them while we figure out why tests without "no-pic" hang.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Not all assemblers of "gas" flavour handle binary constants, e.g.
seasoned MacOS Xcode doesn't, so give them a hand.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The af_alg engine and associated test were creating warnings when compiled
with clang. This fixes it.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The new afalg test should have a copyright date of 2016. Also an
incorrect buffer was being sent to EVP_CipherFinal_ex when
decrypting.
Reviewed-by: Richard Levitte <levitte@openssl.org>
1. Cleaned up eventfd handling
2. Reworked socket setup code to allow other algorithms to be added in
future
3. Fixed compile errors for static build
4. Added error to error stack in all cases of ALG_PERR/ALG_ERR
5. Called afalg_aes_128_cbc() from bind() to avoid race conditions
6. Used MAX_INFLIGHT define in io_getevents system call
7. Coding style fixes
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
- Remove Win builds (temporarily). They're slow, allowed to fail,
and therefore not useful as they are.
- Make the --unified part of the matrix build-only. (This can be
swapped if --unified becomes the default)
- Only build 'no-engine' once, don't run any tests, but don't allow it
to fail.
Reviewed-by: Richard Levitte <levitte@openssl.org>
- Remove no-asm. We've got to cut something, and this is at least
partially covered by the sanitizer builds.
- Remove enable-crypto-mdebug from sanitizer
builds. enable-crypto-mdebug has been shown to catch some static
initialization bugs that the standard leak sanitizer can't so
perhaps it has _some_ value; but we shouldn't let the two compete.
Reviewed-by: Richard Levitte <levitte@openssl.org>
When object files with common block symbols are added to static
libraries on Darwin, those symbols are invisible to the linker that
tries to use them. Our solution was to use -fno-common when compiling
C source.
Unfortunately, there is assembler code that defines OPENSSL_ia32cap_P
as a common block symbol, unconditionally, and in some cases, there is
no other definition. -fno-common doesn't help in this case.
However, 'ranlib -c' adds common block symbols to the index of the
static library, which makes them visible to the linker using it, and
that solves the problem we've seen.
The common conclusion is, either use -fno-common or ranlib -c on
Darwin. Since we have common block symbols unconditionally, choosing
the method for our source is easy.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Add support for application supplied any defined by callback. An
application can change the selector value if it wishes. This is
mainly intended for values which are only known at runtime, for
example dynamically created OIDs.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Previously, the remaining CT log entries would not be loaded.
Also, CTLOG_STORE_load_file would return 1 even if a log entry was
invalid, resulting in no errors being shown.
Reviewed-by: Ben Laurie <ben@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Disabled by default, but can be enabled by setting the
ct_validation_callback on a SSL or SSL_CTX.
Reviewed-by: Ben Laurie <ben@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Change the ECC default curve list to be this, in order: x25519, secp256r1,
secp521r1, secp384r1, brainpoolP256r1, brainpoolP384r1, and brainpool512r1.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Appease the sanitizer: avoid left shifts of negative values.
This could've been done entirely with casts to uint and back,
but using masks seemed slightly more readable.
There are also implementation-defined signed right shifts in this
code. Those remain.
Reviewed-by: Rich Salz <rsalz@openssl.org>
RC4 based ciphersuites in libssl have been disabled by default. They can
be added back by building OpenSSL with the "enable-weak-ssl-ciphers"
Configure option at compile time.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The proper logic is that both zlib and zlib-dynamic are disabled by
default and that enabling zlib-dynamic would enable zlib. Somewhere
along the way, the logic got changed, zlib-dynamic was enabled by
default and zlib didn't get automatically enabled.
This change restores the original logic.
Reviewed-by: Rich Salz <rsalz@openssl.org>
PVK files with abnormally large length or salt fields can cause an
integer overflow which can result in an OOB read and heap corruption.
However this is an rarely used format and private key files do not
normally come from untrusted sources the security implications not
significant.
Fix by limiting PVK length field to 100K and salt to 10K: these should be
more than enough to cover any files encountered in practice.
Issue reported by Guido Vranken.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This patch implements the HMAC-based Extract-and-Expand Key Derivation
Function (HKDF) as defined in RFC 5869.
It is required to implement the QUIC and TLS 1.3 protocols (among others).
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
- Remove OPENSSL_X25519_X86_64 which never worked, because we don't have
the assembly.
- Also remove OPENSSL_SMALL (which should have been
OPENSSL_SMALL_FOOTPRINT) which isn't a priority at the moment.
Reviewed-by: Rich Salz <rsalz@openssl.org>
1) Simplify code with better PACKET methods.
2) Make broken SNI parsing explicit. SNI was intended to be extensible
to new name types but RFC 4366 defined the syntax inextensibly, and
OpenSSL has never parsed SNI in a way that would allow adding a new name
type. RFC 6066 fixed the definition but due to broken implementations
being widespread, it appears impossible to ever extend SNI.
3) Annotate resumption behaviour. OpenSSL doesn't currently handle all
extensions correctly upon resumption. Annotate for further clean-up.
4) Send an alert on ALPN protocol mismatch.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Handle KDF in ECDH_compute_key instead of requiring each implementation
support it. This modifies the compute_key method: now it allocates and
populates a buffer containing the shared secret.
Reviewed-by: Rich Salz <rsalz@openssl.org>
We copied $target{cflags}, $target{defines} and a few more to %config,
just to add to the entries. Avoid doing so, and let the build templates
deal with combining the two.
There are a few cases where we still fiddle with %target, but that's
acceptable.
Reviewed-by: Andy Polyakov <appro@openssl.org>
The thread_cflag setting filled a double role, as kinda sorta an
indicator of thread scheme, and as cflags. Some configs also added
lflags and ex_libs for multithreading regardless of if threading would
be enabled or not.
Instead of this, add threading cflags among in the cflag setting,
threading lflags in the lflag setting and so on if and only if threads
are enabled (which they are by default).
Also, for configs where there are no special cflags for threading (the
VMS configs are of that kind), this makes it possible to still clearly
mention what thread scheme is used.
The exact value of thread scheme is currently ignored except when it's
"(unknown)", and thereby only serves as a flag to tell if we know how
to build for multi-threading in a particular config. Yet, the
currently used values are "(unknown)", "pthreads", "uithreads" (a.k.a
solaris threads) and "winthreads".
Reviewed-by: Andy Polyakov <appro@openssl.org>
Instead, make the build type ("debug" or "release") available through
$config{build_type} and let the configs themselves figure out what the
usual settings (such as "cflags", "lflags" and so on) should be
accordingly.
The benefit with this is that we can now have debug and release
variants of any setting, not just those Configure supports, and may
also involve other factors (the MSVC flags /MD[d] and /MT[d] involve
both build type and whether threading is enabled or not)
Reviewed-by: Andy Polyakov <appro@openssl.org>
Configure had the Unix centric addition of -lz when linking with zlib
is enabled, which doesn't work on other platforms. Therefore, we move
it to the BASE_unix config template and add corresponding ones in the
other BASE_* config templates. The Windows one is probably incomplete,
but that doesn't matter for the moment, as mk1mf does it's own thing
anyway.
This required making the %withargs table global, so perl snippets in
the configs can use it.
Reviewed-by: Andy Polyakov <appro@openssl.org>
These BASE templates are intended to hold values that are common for
all configuration variants for whole families of configurations.
So far, three "families" are identified: Unix, Windows and VMS, mostly
characterised by the build system they currently use.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This provides for more powerful lazy evaluation and buildup of the
setting contents. For example, something like this becomes possible:
defines => [ sub { $config{thisorthat} ? "FOO" : () } ]
Any undefined result of such functions (such as 'undef' or the empty
list) will be ignored.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This patch provides getters for default_passwd_cb and userdata for SSL
and SSL_CTX. The getter functions are required to port Python's ssl module
to OpenSSL 1.1.0.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
This has no real meaning, except it gives Configure a hint that VC
targets are indeed capable of producing shared objects.
Reviewed-by: Matt Caswell <matt@openssl.org>
There are cases, for example when configuring no-asm, that the added
uplink source files got in the way of the cpuid ones. The best way to
solve this is to separate the two.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Formally only 32-bit AVX2 code path needs this, but I choose to
harmonize all vector code paths.
RT#4346
Reviewed-by: Richard Levitte <levitte@openssl.org>
Most of the assembly uses constants from arm_arch.h, but a few references to
ARMV7_NEON don't. Consistently use the macros everywhere.
Signed-off-by: Andy Polyakov <appro@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Instead of overriding a default operation move default operation to a
separate function which is then explicitly included in any EC_METHOD
that uses it.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Specifies a callback that will, in the future, be used by the SSL code to
decide whether to abort a connection on Certificate Transparency grounds.
Reviewed-by: Ben Laurie <ben@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The HOST_c2l() macro assigns the value to the specified variable, but also
evaluates to the same value. Which we ignore, triggering a warning.
To fix this, just cast it to void like we did in commit 08e553644
("Fix some clang warnings.") for a bunch of other instances.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Andy Polyakov <appro@openssl.org>
and reorganize/harmonize post-conditions.
Additional hardening following on from CVE-2016-0702
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
At the same time remove miniscule bias in final subtraction.
Performance penalty varies from platform to platform, and even with
key length. For rsa2048 sign it was observed to be 4% for Sandy
Bridge and 7% on Broadwell.
CVE-2016-0702
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Performance penalty is 2% on Linux and 5% on Windows.
CVE-2016-0702
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Performance penalty varies from platform to platform, and even
key length. For rsa2048 sign it was observed to reach almost 10%.
CVE-2016-0702
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
in s_server cmd:
specifying -trace option, falls through and turn-on security_debug
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
In the BN_hex2bn function the number of hex digits is calculated using
an int value |i|. Later |bn_expand| is called with a value of |i * 4|.
For large values of |i| this can result in |bn_expand| not allocating any
memory because |i * 4| is negative. This leaves ret->d as NULL leading
to a subsequent NULL ptr deref. For very large values of |i|, the
calculation |i * 4| could be a positive value smaller than |i|. In this
case memory is allocated to ret->d, but it is insufficiently sized
leading to heap corruption. A similar issue exists in BN_dec2bn.
This could have security consequences if BN_hex2bn/BN_dec2bn is ever
called by user applications with very large untrusted hex/dec data. This is
anticipated to be a rare occurrence.
All OpenSSL internal usage of this function uses data that is not expected
to be untrusted, e.g. config file data or application command line
arguments. If user developed applications generate config file data based
on untrusted data then it is possible that this could also lead to security
consequences. This is also anticipated to be a rare.
Issue reported by Guido Vranken.
CVE-2016-0797
Reviewed-by: Andy Polyakov <appro@openssl.org>
If the tests fail early before an ASYNC_WAIT_CTX is created then there
can be a use before init problem in asynctest.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Clarify that the "cleanup" routing does not get called if you invoke
ASYNC_WAIT_CTX_clear_fd() directly.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Implementation experience has shown that the original plan for async wait
fds was too simplistic. Originally the async logic created a pipe internally
and user/engine code could then get access to it via API calls. It is more
flexible if the engine is able to create its own fd and provide it to the
async code.
Another issue is that there can be a lot of churn in the fd value within
the context of (say) a single SSL connection leading to continually adding
and removing fds from (say) epoll. It is better if we can provide some
stability of the fd value across a whole SSL connection. This is
problematic because an engine has no concept of an SSL connection.
This commit refactors things to introduce an ASYNC_WAIT_CTX which acts as a
proxy for an SSL connection down at the engine layer.
Reviewed-by: Richard Levitte <levitte@openssl.org>
VisualStudio 2015 has a bug where an internal compiler error was occurring.
By reordering the DEFINE_STACK_OF declarations for SSL_CIPHER and SSL_COMP
until after the ssl3.h include everything seems ok again.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Add X25519 to TLS supported curve list.
Reject attempts to configure keys which cannot be used
for signing.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Add a flag to EC_METHOD for curves which do not support signing.
New function EC_KEY_can_sign() returns 1 is key can be used for signing.
Return an explicit error is an attempt is made to sign with
no signing curves.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
In some cases the EC_POINT and EC_KEY BIGNUM components are suboptimal
or inappropriate. Add an "custom_data" field which curves can populate with
a custom structure to suit their needs.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Extract compression form in EC_KEY_oct2key() instead of manually in the
ASN.1 code. For custom curves do not assume the initial octet is the
compression form: it isn't for X25519 et al.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Add support for optional overrides of various private key operations
in EC_METHOD.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Use standard X25519 and X448 names for OIDs. Delete EdDSA OIDs: for now they
wont be used and EdDSA may use a different format.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
When building with GNU C, clang or VMS C, it's more efficient to
generate dependency file and object file in one call rather than two.
Have the dependency output in a temporary file and compare it with the
previous one if available to see if replacement is waranted, thereby
avoiding unnecessary reconstruction of Makefile / descrip.mms.
Github issue #750
Reviewed-by: Rich Salz <rsalz@openssl.org>
Stack Overflow has a number of questions related to mutual authentication,
the client and its certificate. Those visiting the man pages for functions
like SSL_CTX_use_certificate and SSL_CTX_load_verify_locations don't
receive the benefit of a cross reference to SSL_CTX_set_client_CA_list.
Reviewed-by: Richard Levitte <levitte@openssl.org>
- srtp_profiles variable is defined when building with SRTP, keeping
the variable usage also under ifndef OPENSSL_NO_SRTP
- alpn help option was kept under ifndef OPENSSL_NO_SRTP
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The INCLUDE statement can handle setting extra include directories for
individual object files, let's use it.
Reviewed-by: Andy Polyakov <appro@openssl.org>
I read the PROBLEMS, and they're outdated; nothing I'd put in the
online FAQ, for example. Test-builds work without using these files.
Had to remove the rehash.time stuff from Makefile.in
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reformat EC_POINT_new.pod and add parentheses to function names.
Clarify the octet form.
Add documentation for EC_POINT_oct2buf().
Reviewed-by: Rich Salz <rsalz@openssl.org>
In EC_KEY_priv2buf(), check for pbuf sanity.
If invoked with NULL, gracefully returns the key length.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
- In error paths, EVP_MD_CTX allocated by the callee is not released.
- Checking method before accessing
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
We are using strcmp() as the cmp_func, where in the EDK2 environment
strcmp actually ends up being the external AsciiStrCmp() function —
an EFI library function defined with the Microsoft ABI.
This means that we can't just assign function pointers to it, since
in GCC-hosted builds the ABI of any function *not* explicitly marked
EFIAPI is the native SysV ABI.
Arguably this stupidity ought to be resolved on the UEFI side, but in
the general case that would mean that we need to provide ABI-compatible
wrappers for *all* the "standard" functions, just in case they're used
like this.
And in fact we already have a workaround here for DEC C. So instead of
playing games with casting function pointers, it's nicer just to use a
simple function to wrap the strcmp() call. That cleans up the DEC C
workaround, *and* it works around the UEFI bogosity at the same time.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
mkdef.pl was issuing the following error:
Use of uninitialized value within %tag in numeric eq (==) at
util/mkdef.pl line 560, <IN> line 92
This was because it was treating a __cplusplus "#ifdef" check as a "tag"
but then skipping over the corresponding "#endif". Therefore after
processing a file it still had "left over" tags from processing the
previous file. It was also getting confused by "#if" checks that didn't
match is pre-defined styles.
Reviewed-by: Richard Levitte <levitte@openssl.org>
They now default to " " as separator, but that can be overridden by
having a hash with parameters as last argument. The only currently
recognised parameter is `separator'.
The special separator `undef' will force the result to become a list
rather than a concatenated string.
Reviewed-by: Andy Polyakov <appro@openssl.org>
The internal |fmtstr| function used in processing a "%s" format string
in the BIO_*printf functions could overflow while calculating the length
of a string and cause an OOB read when printing very long strings.
Additionally the internal |doapr_outch| function can attempt to write to
an OOB memory location (at an offset from the NULL pointer) in the event of
a memory allocation failure. In 1.0.2 and below this could be caused where
the size of a buffer to be allocated is greater than INT_MAX. E.g. this
could be in processing a very long "%s" format string. Memory leaks can also
occur.
These issues will only occur on certain platforms where sizeof(size_t) >
sizeof(int). E.g. many 64 bit systems. The first issue may mask the second
issue dependent on compiler behaviour.
These problems could enable attacks where large amounts of untrusted data
is passed to the BIO_*printf functions. If applications use these functions
in this way then they could be vulnerable. OpenSSL itself uses these
functions when printing out human-readable dumps of ASN.1 data. Therefore
applications that print this data could be vulnerable if the data is from
untrusted sources. OpenSSL command line applications could also be
vulnerable where they print out ASN.1 data, or if untrusted data is passed
as command line arguments.
Libssl is not considered directly vulnerable. Additionally certificates etc
received via remote connections via libssl are also unlikely to be able to
trigger these issues because of message size limits enforced within libssl.
CVE-2016-0799
Issue reported by Guido Vranken.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Simplifies calling code. Also fixed up any !ptr tests that were
nearby, turning them into NULL tests.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Fixes github issue 463. Building the app without OPENSSL_NO_SOCK
isn't supported, so only do OPENSSL_NO_OCSP.
Reviewed-by: Andy Polyakov <appro@openssl.org>
The SRP user database lookup method SRP_VBASE_get_by_user had confusing
memory management semantics; the returned pointer was sometimes newly
allocated, and sometimes owned by the callee. The calling code has no
way of distinguishing these two cases.
Specifically, SRP servers that configure a secret seed to hide valid
login information are vulnerable to a memory leak: an attacker
connecting with an invalid username can cause a memory leak of around
300 bytes per connection.
Servers that do not configure SRP, or configure SRP but do not configure
a seed are not vulnerable.
In Apache, the seed directive is known as SSLSRPUnknownUserSeed.
To mitigate the memory leak, the seed handling in SRP_VBASE_get_by_user
is now disabled even if the user has configured a seed.
Applications are advised to migrate to SRP_VBASE_get1_by_user. However,
note that OpenSSL makes no strong guarantees about the
indistinguishability of valid and invalid logins. In particular,
computations are currently not carried out in constant time.
Reviewed-by: Rich Salz <rsalz@openssl.org>
If the local system doesn't have GNU C or clang, and not even
makedepend, the build will stop because the call of 'makedepend'
fails. This changes so the build won't stop because of such failure.
The result will be empty .d files, and that's ok.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The ssl_set_masks() function no longer depends on the cipher. This
also means there is no need to set the masks for each cipher in
ssl3_choose_cipher.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Different assembler versions disagree on how to interpret #-1 as
argument to vmov.i64, as 0xffffffffffffffff or 0x00000000ffffffff.
So replace it with something they can't disagree on.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The entire contents of <internal/bn_conf.h> are unwanted in the UEFI
build because we have to do it differently there. To support building
for both 32-bit and 64-bit platforms without re-running the OpenSSL
Configure script, the EDK2 environment defines THIRTY_TWO_BIT or
SIXTY_FOUR_BIT for itself according to the target platform.
The current setup is broken, though. It checks for OPENSSL_SYS_UEFI but
before it's actually defined, since opensslconf.h hasn't yet been
included.
Let's fix that by including opensslconf.h. And also let's move the
bn_conf.h doesn't even need to *exist* in the UEFI build environment.
This is also GH PR736.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Removing code, where memory was getting allocated for an unused variable
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
These flags are limitting needlessly, are often patched by packagers,
and should be specified on the configuration command line by anyone
who desires for it to be specific rather than forced by us.
This work was already done with mingw when those configs were worked
on, now it gets applied to the remaining configs.
Reviewed-by: Andy Polyakov <appro@openssl.org>
In the earlier change, where static libraries get built with position
independent code, OPENSSL_PIC was removed by mistake. This adds it
back.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Adding -nostdinc to the EDK2 showed that we were including <inttypes.h>
for some UEFI builds, because the check for __STDC_VERSION__ happens
before the check for OPENSSL_SYS_UEFI. Fix that.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
The commit 1288f26 says that it fixes no-async, but instead seems to break
it. Therefore revert that change and fix no-async.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Provide an appropriate definition of PRIu64 for the EDK2 build, since
we don't have <inttypes.h> there.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
This way, we can use them as conditions instead of relying to more or
less obscure aliases in %config or variables directly in Configure.
Reviewed-by: Rich Salz <rsalz@openssl.org>
When building a DSO, there's no reason to include all symbols from
static libraries it happens to link with, whichever they may be.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Preserved for now for those who have scripts with the option
"no-ssl2". We warn that it's deprecated, and ignore it otherwise.
In response to RT#4330
Reviewed-by: Rich Salz <rsalz@openssl.org>
Building shared libraries or not is not the same as building position
independent code or not. It's true that if you don't build PIC, you
can't build shared libraries. However, you may very well want to
build only static libraries but still want PIC code.
Therefore, we introduce a new configuration option "pic", which is
enabled by default or explicitely with "enable-pic", or disabled with
"no-pic" or "disable-pic". Of course, if "pic" is disabled, "shared"
and "dynamic-engine" are automatically disabled as well.
Reviewed-by: Rich Salz <rsalz@openssl.org>
We were kinda sorta using a mix of $disabled{"static-engine" and
$disabled{"dynamic-engine"} in Configure. Let's avoid confusion,
choose one of them and stick to it.
Reviewed-by: Rich Salz <rsalz@openssl.org>
An error was introduced with the setting of SHLIB in DO_GNU_SO.
A common DO_GNU_SO_COMMON that both DO_GNU_SO and DO_GNU_SO_NOCALC use
makes things clearer.
Reviewed-by: Matt Caswell <matt@openssl.org>
Because we're requiring Perl 5.10.0 and the 'parent' didn't appear
before Perl 5.10.1, we need to resort to the older parent module
declaration style, modifying @ISA.
Reviewed-by: Andy Polyakov <appro@openssl.org>
GNU make will re-exec if (it thinks that) the Makefile has changed.
Just having the target Makefile seems to make it think it has, so we
end up in a look where GNU make re-execs for ever.
The fix is easy, just remove the Makefile target and have the depend
target run the recipe on its own instead of depending on Makefile.
Reviewed-by: Rich Salz <rsalz@openssl.org>
They depend on this feature because they use the engine ossltest,
which is only available as a dynamic engine.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Until now, the engines in engines/ were only built as dynamicaly
loadable ones if shared libraries were built.
We not dissociate the two and can build dynamicaly loadable engines
even if we only build static libcrypto and libssl. This is controlled
with the option (enable|disable|no)-static-engine, defaulting to
no-static-engine.
Note that the engines in crypto/engine/ (dynamic and cryptodev) will
always be built into libcrypto.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This takes us away from the idea that we know exactly how our static
libraries are going to get used. Instead, we make them available to
build shareable things with, be it other shared libraries or DSOs.
On the other hand, we also have greater control of when the shared
library cflags. They will never be used with object files meant got
binaries, such as apps/openssl or test/test*.
With unified, we take this a bit further and prepare for having to
deal with extra cflags specifically to be used with DSOs (dynamic
engines), libraries and binaries (applications).
Reviewed-by: Rich Salz <rsalz@openssl.org>
Depending on Makefile meant that a new attempt to rebuild the Makefile
with "new" dependency data was done all the time, uncontrolled. Better
to depend on configdata.pm, which truly only changes with reconfiguration.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Ben Laurie <ben@openssl.org>
Remove old code that handled various invalid DSA formats in ancient
software.
This also fixes a double free bug when parsing malformed DSA private keys.
Thanks to Adam Langley (Google/BoringSSL) for discovering this bug using
libFuzzer.
CVE-2016-0705
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Adapted from BoringSSL. Added a test.
The extension parsing code is already attempting to already handle this for
some individual extensions, but it is doing so inconsistently. Duplicate
efforts in individual extension parsing will be cleaned up in a follow-up.
Reviewed-by: Stephen Henson <steve@openssl.org>
This silences the memory sanitizer. All fields were already correctly
initialized but the struct padding wasn't, causing an uninitialized read
warning.
Reviewed-by: Tim Hudson <tjh@openssl.org>
So far, MingW shared libraries were named like this
libeay32.dll + libeay32.dll.a
ssleay32.dll + ssleay32.dll.a
That naming scheme is antiquated, a reminicense of SSLeay. We're
therefore changing the scheme to something that's more like the rest
of OpenSSL.
There are two factors to remember:
- Windows libraries have no recorded SOvers, which means that the
shared library version must be encoded in the name. According to
some, it's unwise to encode extra periods in a Windows file name,
so we convert version number periods to underscores.
- MingW has multilib ability. However, DLLs need to reside with the
binaries that use them, so to allow both 32-bit and 64-bit DLLs to
reside in the same place, we add '-x64' in the name of the 64-bit
ones.
The resulting name scheme (for SOver 1.1) is this:
on x86:
libcrypto-1_1.dll + libcrypto.dll.a
libssl-1_1.dll + libssl.dll.a
on x86_64:
libcrypto-1_1-x64.dll + libcrypto.dll.a
libssl-1_1-x64.dll + libssl.dll.a
An observation is that the import lib is the same for both
architectures. Not to worry, though, as they will be installed in
PREFIX/lib/ for x86 and PREFIX/lib64/ for x86_64.
As a side effect, MingW got its own targets in Makefile.shared.
link_dso.mingw-shared and link_app.mingw-shared are aliases for the
corresponding cygwin-shared targets. link_shlib.mingw-shared is,
however, a target separated from the cygwin one.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Better libclean that removes the exact files that have been built,
nothing more and nothing less.
Corrected typo
A couple of editorial changes.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Since we're using the acronym DSO everywhere else and that's a common
name for that kind of object, we might as well do so here as well.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Instead of having the installation recipe rely on special knowledge,
feed it with information, including what shared library files belong
together. For Cygwin and Mingw, that's the .dll and its import
library .dll.a. For Unixen, it's the shared library file name with SO
version and the one without.
Reviewed-by: Andy Polyakov <appro@openssl.org>
The engine DSOs were named as if they were shared libraries, and could
end up having all sorts of fancy names:
Cygwin: cygFOO.dll
Mingw: FOOeay32.dll
Unix: libFOO.so / libFOO.sl / libFOO.dylib / ...
This may be confusing, since they look like libraries one should link
with at link time, when they're just DSOs.
It's therefore time to rename them, and do it consistently on all
platforms:
Cygwin & Mingw: FOO.dll
Unix: FOO.{so,sl,dylib,...}
Interestingly enough, the MSVC and VMS builds always did it this way.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Because we know for certain that the link_shlib targets are used
exclusively for shared libraries (libcrypto and libssl) and that they
must have an associated .num file, we don't need to check the library
name to produce an ld script. Just do it unconditionally.
link_shlib.linux-shared can be simplified further, as most of it is
exactly the same as $(DO_GNU_SO) with just one variable modification.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Originally, the Makefile.shared targets described what they used as
input for a shared object, be it a shared library or a DSO. It turned
out, however, that the link_o targets were used exclusively for
engines and the link_a targets were for libcrypto and libssl.
This rename fest turns and indication on the kind of input the targets
get to the intention with using them.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Then it can pass around the information where it belongs. The
Makefile templates pick it up along with other target data, the
DSO module gets to pick up the information through
crypto/include/internal/dso_conf.h
Reviewed-by: Andy Polyakov <appro@openssl.org>
- install_sw had a display of text that belongs under the install target
- previous layout installed architecture dependent files in
dev:['prefix'.'arch'.LIB], dev:['prefix'.'arch'.EXE] and
dev:['prefix'.'arch'.ENGINES]. Changed to dev:['prefix'.LIB.'arch'],
dev:['prefix'.EXE.'arch'] and dev:['prefix'.ENGINES.'arch'] instead.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This is done with a simple file name comparison. We could think of
something more elegant in the future.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Adding uplink and applink to some builds was done by "magic", the
configuration for "mingw" only had a macro definition, the Configure
would react to its presence by adding the uplink source files to
cpuid_asm_src, and crypto/build.info inherited dance to get it
compiled, and Makefile.shared made sure applink.o would be
appropriately linked in. That was a lot under the hood.
To replace this, we create a few template configurations in
Configurations/00-base-templates.conf, inherit one of them in the
"mingw" configuration, the rest is just about refering to the
$target{apps_aux_src} / $target{apps_obj} in the right places.
Reviewed-by: Andy Polyakov <appro@openssl.org>
VMS DIFF tries to calculate all the differences, which is slower than
just reading the files and stopping at the first difference. The
latter doesn't exist as a command, so the problem is solved with perl
and File::Compare (has been in core perl since very early version 5).
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
DCL may be in extended parsing style, which makes it less case
insensitive, so when removing a string from another, make sure to get
casing correctly.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
This isn't the fully featured combination of compiler generated
dependency files and Makefile include directives, but a cheaper
variant of the same.
The dependency files are generated automatically, but then we have the
usual "depend" target. However, we depend on it in the bigger phony
targets that are the most likely to be used. That make this feature
automatic enough.
A side effect is that we can't use the build file's timestamp to check
if reconfiguring might be in order. In its place, we use a flag file
that depends on Configure and the build file template and depend on it
in spots where it makes sense to check for the need to reconfigure.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The uninstall_sw target tried to 'make uninstall' in all subdirs.
Change it to only go into $(INSTALL_SUBS), just like install_sw does.
Reviewed-by: Rich Salz <rsalz@openssl.org>
When cross compiling, we may end up with someting like apps/openssl.exe
and a number of test/*.exe. However, util/shlib_wrap.sh doesn't know
what the executable extension should be, if any, so we need to make
sure it has access to that information when testing, since
OpenSSL::Test uses that script to execute all programs.
Reviewed-by: Rich Salz <rsalz@openssl.org>
All those flags existed because we had all the dependencies versioned
in the repository, and wanted to have it be consistent, no matter what
the local configuration was. Now that the dependencies are gone from
the versioned Makefile.ins, it makes much more sense to use the exact
same flags as when compiling the object files.
Reviewed-by: Rich Salz <rsalz@openssl.org>
.d (.MMS in the VMS world) files with just dependencies are built from
exactly the same conditions as the object files. Therefore, the rules
for them can be built at the same time as the rules for the
corresponding object files.
This removes the requirement for a src2dep function in the build file
templates, and for common.tmpl to call it. In the end, the existence
of depend files is entirely up to the build file.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Add -DBIO_DEBUG to --strict-warnings.
Remove comments about outdated debugging ifdef guards.
Remove md_rand ifdef guarding an assert; it doesn't seem used.
Remove the conf guards in conf_api since we use OPENSSL_assert, not assert.
For pkcs12 stuff put OPENSSL_ in front of the macro name.
Merge TLS_DEBUG into SSL_DEBUG.
Various things just turned on/off asserts, mainly for checking non-NULL
arguments, which is now removed: camellia, bn_ctx, crypto/modes.
Remove some old debug code, that basically just printed things to stderr:
DEBUG_PRINT_UNKNOWN_CIPHERSUITES, DEBUG_ZLIB, OPENSSL_RI_DEBUG,
RL_DEBUG, RSA_DEBUG, SCRYPT_DEBUG.
Remove OPENSSL_SSL_DEBUG_BROKEN_PROTOCOL.
Reviewed-by: Richard Levitte <levitte@openssl.org>
enc:
- typo in -base64 option
- missing help opt text
ocsp, req, rsautl, s_client:
- missing help opt text
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
The windows thread stop code was erroneously not just deleting the thread
local variable on thread stop, but also deleting the thread local *key*
(thus removing thread local data for *all* threads in one go!).
Reviewed-by: Rich Salz <rsalz@openssl.org>
Certain code paths in tls_decrypt_ticket could return early without first
freeing the HMAC_CTX or the EVP_CIPHER_CTX.
Reviewed-by: Rich Salz <rsalz@openssl.org>
As handshake_func is a function pointer, it should compare to NULL
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Although I explicitly don't care about the tinfoil-hat reason given in
the initial opening of RT#3628, that "paths usually contain private
information", there *are* situations where it's useful to eliminate the
filenames from the compiled binary.
The two reasons we do care about in the context of firmware such as EDK2
are that it allows for a smaller footprint, and it is also a necessary
component of a binary-reproducible build.
To that end, introduce OPENSSL_FILE and OPENSSL_LINE macros, defining
them to __FILE__ and __LINE__ respectively in the normal case, but to
"" and 0 when OPENSSL_NO_FILENAMES is set.
This is mostly a naïve invocation of
$ sed 's/__\([FL]I[NL]E\)__/OPENSSL_\1/g' -i `git grep -l __LINE__`
but with a few instances change to just print the function name instead
(although those probably need to die anyway) and test cases left untouched.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
For example, this works instead of giving a big error message (note
the lack of '--unified'):
mkdir ../_build
(cd ../_build/; ../openssl-src/config; make)
Reviewed-by: Andy Polyakov <appro@openssl.org>
The previous fix wasn't right.
Also, change all (^|\s) and (\s|$) constructs to (?:^|\s) and (?:\s|$).
Perl seems to like that better.
Reviewed-by: Andy Polyakov <appro@openssl.org>
When OPENSSL_NO_ASYNC is set, make ASYNC_{un,}block_pause() do nothing.
This prevents md_rand.c from failing to build. Probably better to do it
this way than to wrap every instance in an explicit #ifdef.
A bunch of new socket code got added to a new file crypto/bio/b_addr.c.
Make it all go away if OPENSSL_NO_SOCK is defined.
Allow configuration with no-ripemd, no-ts, no-ui
We use these for the UEFI build.
Also remove the 'Really???' comment from no-err and no-locking. We use
those too.
We need to drop the crypto/engine directory from the build too, and also
set OPENSSL_NO_ENGINE
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
Commit e634b448c ("Defines OSSL_SSIZE_MAX") introduced a definition of
OSSL_SSIZE_MAX which broke the UEFI build. Fix that by making UEFI take
the same definition as Ultrix (ssize_t == int).
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
We don't have atexit() in the EDK2 environment. Firmware never exits.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Commit 05c7b1631 ("Implement the use of heap manipulator implementions")
added 'file' and 'line' arguments to CRYPTO_free() and friends, but neglected
to fix up the !IMPLEMENTED case within CRYPTO_secure_free(). Add the missing
arguments there too.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
- Make use of the functions given through CRYPTO_set_mem_functions().
- CRYPTO_free(), CRYPTO_clear_free() and CRYPTO_secure_free() now receive
__FILE__ and __LINE__.
- The API for CRYPTO_set_mem_functions() and CRYPTO_get_mem_functions()
is slightly changed, the implementation for free() now takes a couple
of extra arguments, taking __FILE__ and __LINE__.
- The CRYPTO_ memory functions will *always* receive __FILE__ and __LINE__
from the corresponding OPENSSL_ macros, regardless of if crypto-mdebug
has been enabled or not. The reason is that if someone swaps out the
malloc(), realloc() and free() implementations, we can't know if they
will use them or not.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Macro definitions "should" be found in $config{defines}, but some
configs haven't transfered macro definitions from their 'cflags'
settings (which isn't mandatory anyway), so check both places.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This check is meaningless on VMS and only produce an error because the
underlying shell (DCL) doesn't understand sh syntax such as '2>&1'.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Depending on user preferences, Configure might get something like
--PREFIX=blah just as well as --prefix=blah, or "SHARED" just as well
as "shared". On VMS, let's therefore lowercase at least the portion
of the argument before a possible equal sign.
For good measure, we lowercase the arguments to be checked in
config.com as well. The original argument is sent on to Configure,
however.
Reviewed-by: Andy Polyakov <appro@openssl.org>
A long time ago, Solaris cc didn't seem to handle -Wl, linker options,
while gcc on Solaris required it. Since then, Solaris cc has
developed to understand -Wl, options, and our little dance to figure
out how to pass linker options to the C compiler that's used isn't
needed any more.
Reviewed-by: Andy Polyakov <appro@openssl.org>
That variable isn't for us, it's for any user, distributor or package
builder that wants one after the section number. "ssl" seems to be
popular...
Reviewed-by: Matt Caswell <matt@openssl.org>
mk1mf was wondering about the options no-heartbeats and
no-crypto-mdebug-backtrace, so we add option hooks them. They only
need to become OPENSSL_NO_ macros in opensslconf.h, so nothing
additional needs to be done.
Also, add "-DOPENSSL_PIC" when shared libraries are produced.
Reviewed-by: Andy Polyakov <appro@openssl.org>
Three header files from crypto/include/internal were used by
util/mkdef.pl. This should never be needed. Some test program used
these, which made it a valid reason at the time to make the some
internal symbols public in the shared libraries, but that's not the
case any more.
However, to be able to link libssl.so, some symbols found in
include/internal headers still need to be made public.
Reviewed-by: Andy Polyakov <appro@openssl.org>
On some platforms, the implementation is such that a signed char
triggers a warning when used with is*() functions. On others, the
behavior is outright buggy when presented with a char that happens
to get promoted to a negative integer.
The safest thing is to cast the char that's used to an unsigned char.
Reviewed-by: Andy Polyakov <appro@openssl.org>
To force it on anyone using --strict-warnings was the wrong move, as
this is an option best left to those who know what they're doing.
Use with care!
Reviewed-by: Andy Polyakov <appro@openssl.org>
DllMain is a symbol that needs to be global, but no one needs to know.
However, some compilers will warn if there isn't a declaration before
the function is defined. Just add a declaration before the function
definition.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Most of the times, it seems that socklen_t is unsigned.
Unfortunately, this isn't always the case, and it doesn't compare with
a size_t without warning.
A cast resolves the issue.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
While IO::Socket::IP is a core perl module (since Perl v5.19.8, or so
says corelist), IO::Socket::INET6 has been around longer, is said to
be more widely deployed, and most importantly, seems to have less bugs
hitting us. We therefore prefer IO::Socket::INET6, and only fall back
to IO::Socket::IP if the former doesn't exist on the local system.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
It seems that on some platforms, the perlasm scripts call the C
compiler for certain checks. These scripts need the environment
variable CC to have the C compiler command.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The installation of man files and html files alike didn't properly
check that file names with different casing could be the same on
case-insensitive file systems. This change fixes that.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The ocsp utility is something of a jack-of-all-trades; most anything
related to the OCSP can be done with it. In particular, the manual
page calls out that it can be used as either a client or a server
of the protocol, but there are also a few things that it can do
which do not quite fit into either role, such as encoding an OCSP
request but not sending it, printing out a text form of an OCSP
response (or request) from a file akin to the asn1parse utility,
or performing a lookup into the server-side revocation database
without actually sending a request or response. All three of these
are documented as examples in the manual page, but the documentation
prior to this commit is somewhat misleading, in that when printing
the text form of an OCSP response, the code also attempts to
verify the response, displaying an error message and returning
failure if the response does not verify. (It is possible that
the response would be able to verify with the given example, since
the default trust roots are used for that verification, but OCSP
responses frequently have alternate certification authorities
that would require passing -CAfile or -CApath for verification.)
Tidy up the documentation by passing -noverify for the case of
converting from binary to textual representation, and also
change a few instances of -respin to -reqin as appropriate, note
that the -url option provides the same functionality as the -host
and -path options, clarify that the example that saves an OCSP
response to a file will also perform verification on that response,
and fix a couple grammar nits in the manual page.
Also remove an always-true conditional for rdb != NULL -- there
are no codepaths in which it could be initialized at the time of
this check.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
The previous method had some unfortunate consequences with
--strict-warnings. To counteract, revert part of the previous change
and move down the block of code that adds the user cflags and defines.
Reviewed-by: Rich Salz <rsalz@openssl.org>
In the early stages of creating the new test framework,
00-test_checkexes was a temporary check to ensure we had a recipe for
every test program in test/. By now, this test has fulfilled its
purpose, and we've learned how to make recipes properly. It's time
for this check to go away.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Removes SSIZE_MAX definition from bss_bio.c and changes that file to use
OSSL_SSIZE_MAX.
No need to account for OPENSSL_SYS_VXWORKS, since that never actually
gets defined anywhere. It must be a historical artifact.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Ben Laurie <ben@openssl.org>
There was a catch 22, where 'make depend' directly after configuring
in an otherwise pristine build tree would fail because buildinf.h
didn't exist yet.
This change has the depend building targets depend on the same other
targets as the object file building targets, so the generation of
buildinf.h and similar files would kick in during 'make depend'.
Reviewed-by: Rich Salz <rsalz@openssl.org>
It's never problem if CRYPTO_ctr128_encrypt is called from EVP, because
buffer in question is always aligned within EVP_CIPHER_CTX structure.
RT#4218
Reviewed-by: Richard Levitte <levitte@openssl.org>
INSTALL_PREFIX is a confusing name, as there's also --prefix.
Instead, tag along with the rest of the open source world and adopt
the Makefile variable DESTDIR to designate the desired staging
directory.
The Configure option --install_prefix is removed, the only way to
designate a staging directory is with the Makefile variable (this is
also implemented for VMS' descrip.mms et al).
Reviewed-by: Rich Salz <rsalz@openssl.org>
Not all git versions understand **/Makefile, but all recognize that
filename without any path applies to all directories.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Contemporary Xcode gcc is a front-end to clang, so that explicit
gcc build is actually redundant on MacOS X.
Reviewed-by: Richard Levitte <levitte@openssl.org>
If the environment variable HARNESS_ACTIVE isn't defined or
HARNESS_VERBOSE is defined, it's probable that lots of output is
desired.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The logging that was performed in OpenSSL::Test was initially set up
as a means not to let messages that test programs write to STDERR get
displayed when a test isn't running in verbose mode. However, the way
it was implemented, it meant that those messages were never displayed,
and you had to look in a test log. This also meant that output to
STDERR and output to STDOUT got broken apart, which isn't optimal.
So, we remove the whole test log file implementation, and instead,
we're sending STDERR to the null device unless one of these conditions
apply:
- the test recipe already redirects stderr. Just let it.
- the environment variable HARNESS_ACTIVE is undefined, meaning the
recipe is run directly as a perl script instead of being harnessed
by Test::Harness
- the environment variable HARNESS_VERBOSE is set.
Getting a full log of the tests now becomes as simple as this:
HARNESS_VERBOSE=yes make test 2>&1 | tee tests.log
Reviewed-by: Rich Salz <rsalz@openssl.org>
The actual bug with current getnameinfo() on VMS is not that it puts
gibberish in the service buffer, but that it doesn't touch it at all.
The gibberish we dealt with before was simply stuff that happened to
be on the stack.
It's better to initialise the service buffer properly (with the empty
string) and check if it's still an empty string after the
getnameinfo() call, and fill it with the direct numerical translation
of the raw port if that's the case.
Reviewed-by: Rich Salz <rsalz@openssl.org>
On VMS, periods in directory names weren't allowed. To counter that,
unpackers such as VMSTAR convert periods in directory names to
underscores. We need to count that in and add an alternative library
path for Text::Template.
Reviewed-by: Rich Salz <rsalz@openssl.org>
In the previous commit to change all chomp to a more flexible regexp,
Configure was forgotten. This completes the change.
Reviewed-by: Andy Polyakov <appro@openssl.org>
apps/progs.pl counted on the caller to provide the exact command
files. The unified build doesn't have that knowledge, and the easier
and more flexible thing to do is to feed it all the apps/*.c files and
let it figure out the command names by looking inside (looking for
/int ([a-z0-9][a-z0-9_]*)_main\(int argc,/).
Also, add it to the generate command, since it's a versioned file.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Once upon a time, there was chop, which somply chopped off the last
character of $_ or a given variable, and it was used to take off the
EOL character (\n) of strings.
... but then, you had to check for the presence of such character.
So came chomp, the better chop which checks for \n before chopping it
off. And this worked well, as long as Perl made internally sure that
all EOLs were converted to \n.
These days, though, there seems to be a mixture of perls, so lines
from files in the "wrong" environment might have \r\n as EOL, or just
\r (Mac OS, unless I'm misinformed).
So it's time we went for the more generic variant and use s|\R$||, the
better chomp which recognises all kinds of known EOLs and chops them
off.
A few chops were left alone, as they are use as surgical tools to
remove one last slash or one last comma.
NOTE: \R came with perl 5.10.0. It means that from now on, our
scripts will fail with any older version.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Remove support for static ECDH ciphersuites. They require ECDH keys
in certificates and don't support forward secrecy.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Trouble is that LINK variable assignment in make-file interferes with
LINK environment variable, which can be used to modify Microsoft's
LINK.EXE behaviour.
RT#4289
Reviewed-by: Richard Levitte <levitte@openssl.org>
Commit 7823d792d0 added DEFINE_LHASH_OF
to a C source file. DEFINE_LHASH_OF() and DEFINE_STACK_OF() must
be used only in header files to avoid clang warnings for unused
static-inline functions.
Reviewed-by: Rich Salz <rsalz@openssl.org>
To enable heartbeats for DTLS, configure with enable-heartbeats.
Heartbeats for TLS have been completely removed.
This addresses RT 3647
Reviewed-by: Richard Levitte <levitte@openssl.org>
With this, Cygwin and Mingw builds stand a much better chance to be
able to build outside of the source tree with the unified build.
Reviewed-by: Rich Salz <rsalz@openssl.org>
On Windows POSIX layers, two files are produced for a shared library,
there's {shlibname}.dll and there's the import library {libname}.dll.a
On some/most Unix platforms, a {shlibname}.{sover}.so and a symlink
{shlibname}.so are produced.
For each of them, unix-Makefile.tmpl was entirely consistent on which
to have as a target when building a shared library or which to use as
dependency.
This change clears this up and makes it consistent, we use the
simplest form possible, {lib}.dll.a on Windows POSIX layers and
{shlibname}.so on Unix platforms. No exception.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Some keys are attached to the full RSA CSP which doesn't support SHA2
algorithms: uses the AES CSP if present.
Reviewed-by: Tim Hudson <tjh@openssl.org>
If someone runs a mixed unixmake / unified environment (the unified
build tree would obviously be out of the source tree), the unified
build will pick up on the unixmake crypto/buildinf.h because of
assumptions made around this sort of declaration (found in
crypto/build.info):
DEPENDS[cversion.o]=buildinf.h
The assumption was that if such a header could be found in the source
tree, that was the one to depend on, otherwise it would assume it
should be in the build tree.
This change makes sure that sort of mix-up won't happen again.
Reviewed-by: Rich Salz <rsalz@openssl.org>
It's not necessary for a pristine source, and a developer that makes
changes usually knows what to do.
Also, there was this mechanism that would do a "make depend"
automatically which hasn't been used for so many years. Removed as
well.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Some files in crypto/bn depend on internal/bn_conf.h, and so does
test/bntest. Therefore, we add another inclusion directory.
Reviewed-by: Rich Salz <rsalz@openssl.org>
As noted already, some platforms don't fill in ai_protocol as
expected. To circumvent that, we have BIO_ADDRINFO_protocol() to
compute a sensible answer in that case.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Error codes are printed in hex, and previous OpenSSL versions expected
the error codes to be provided to errstr in hex. In 1.1.0, for some
reason, it was expecting them to be decimal.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The functions that have been deprecated by the auto init changes are
now guarded with deprecation checks, so it's fairly easy to see if
they can be used.
In test/dtlsv1listentest, we simply remove all init and cleanup code,
as they are call automatically when needed.
Reviewed-by: Matt Caswell <matt@openssl.org>
Because some platforms won't will in any value in ai_protocol, there's
no point using it if we already know what it should be.
Reviewed-by: Rich Salz <rsalz@openssl.org>
It seems that some platforms' getaddrinfo don't fill in the
ai_protocol field properly. On those, the assertion
'protocol == BIO_ADDRINFO_protocol(res)' will fail. Best to remove
it.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Update ciphers documentation as well (based on -04 rev of ID).
Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Rich Salz <rsalz@openssl.org>
RT: #4206, GH: #642
Because the command line definitions of OPENSSLDIR and ENGINESDIR
contain quotes, we need a variant of CFLAG where backslashes and
quotes are escaped when we produce buildinf.h
Reviewed-by: Rich Salz <rsalz@openssl.org>
If init failed we'd like to set an error code to indicate that. But if
init failed then when the error system tries to load its strings its going
to fail again. We could get into an infinite loop. Therefore we just set
a single error the first time around. After that no error is set.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The new init functions can fail if the library has already been stopped. We
should be able to indicate failure with a 0 return value.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The local variable tmp was declared static when it shouldn't be. This
is in the no-threads implementation, and it was immediately initialised
to something else on every invokation of the function so it doesn't break
anything...but still shouldn't be there.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
This was a developer debugging feature and was never a useful public
interface.
Added all missing X509 error codes to the verify(1) manpage, but
many still need a description beyond the associated text string.
Sorted the errors in x509_txt.c by error number.
Reviewed-by: Stephen Henson <steve@openssl.org>
The old building scripts get removed, they are hopelessly gone in bit
rot by now.
Also remove the old symbol hacks. They were needed needed to shorten
some names to 31 characters, and to resolve other symbol clashes.
Because we now compile with /NAMES=(AS_IS,SHORTENED), this is no
longer required.
Reviewed-by: Rich Salz <rsalz@openssl.org>
As part of this, change util/mkdef.pl to stop adding libraries to
depend on in its output. mkdef.pl should ONLY output a symbol
vector.
Because symbol names can't be longer than 31 characters, we use the
compiler to shorten those that are longer down to 23 characters plus
an 8 character CRC. To make sure users of our header files will pick
up on that automatically, add the DEC C supported extra headers files
__decc_include_prologue.h and __decc_include_epilogue.h.
Furthermore, we add a config.com, so VMS people can configure just as
comfortably as any Unix folks, thusly:
@config
Reviewed-by: Rich Salz <rsalz@openssl.org>
Cygwin and Mingw name their libraries a bit differently from the rest
of the POSIXly universe, we need to adapt to that.
In Makefile.tmpl, it means that some hunks will only be output
conditionally.
This also means that shared_extension for the Cygwin and Mingw
configurations in Configurations/10-main.conf are changing from .dll.a
to .dll. Makefile.shared does a fine job without having them
specified, and it's much easier to work with tucking an extra .a at
the end of files in the installation recipes than any amount of name
rewrites, especially with the support of the SHARED_NAME in the top
build.info.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Under certain conditions, one might not want to output certain
sections of a template file. This adds the functions output_off() and
output_on(), reachable inside the templates. And example usage in a
Makefile template could be this:
@ : {- output_off() if $config{no_shared}; "" -}
... lines dealing with shared libraries
@ : {- output_on() -}
Reviewed-by: Rich Salz <rsalz@openssl.org>
The logic to figure out the combinations of --prefix and --openssldir
has stayed in Configure so far, with Unix paths as defaults.
However, since we're making Configure increasingly platform agnostic,
these defaults need to change and adapt to the platform, along with
the logic to combine them.
The easiest to provide for this is to move the logic and the defaults
away from Configure and into the build files.
This also means that the definition of the macros ENGINESDIR and
OPENSSLDIR move away from include/openssl/opensslconf.h and into the
build files.
Makefile.in is adapted accordingly.
Reviewed-by: Rich Salz <rsalz@openssl.org>
With some compilers, C macros are defined differently on the command
line than on Unix. It could be that the flad to define them isn't -D,
it could also be that they need to be grouped together and not be mixed
in with the other compiler flags (that's how it's done on VMS, for
example).
On Unix family platform configurations, we can continue to have macro
definitions mixed in with the rest of the flags, so the changes in
Configurations/*.conf are kept to an absolute minimum.
Reviewed-by: Rich Salz <rsalz@openssl.org>
If you call an explicit deinit when we've not been inited then a seg
fault can occur. We should check that we've been inited before attempting
to deinit.
Reviewed-by: Rich Salz <rsalz@openssl.org>
It seems like it gives back gibberish. If we asked for a numeric
service, it's easy to check for a digit in the first position, and
if there isn't any, rewrite it using older methods.
Reviewed-by: Rich Salz <rsalz@openssl.org>
In build.info files, make the include directory in the build directory
absolute, or Configure will think it should be added to the source
directory top. Configure will turn it into a relative path if
possible.
Reviewed-by: Rich Salz <rsalz@openssl.org>
There were cases where some input was absolute, and concatenating it
to the diretory to the source or build top could fail spectacularly.
Let's check the input first to see if it's absolute.
And while we're on the subject of checking if a file or dir spec is
absolute using file_name_is_absolute() has its own quirks on VMS,
where a logical name is considered absolute under most circumstances.
This is perfectly correct from a VMS point of view, but when parsing
the build.info files, we want single word file or directory names to
only be checked syntactically. A function isabsolute() that does the
right thing is the solution.
Reviewed-by: Rich Salz <rsalz@openssl.org>
After the final use of the thread_local_inits_st we should ensure it is
set to NULL, just in case OPENSSL_INIT_thread_stop gets called again and
it tries to use garbage.
Reviewed-by: Rich Salz <rsalz@openssl.org>
With the new init framework resources aren't released until the process
exits. This means checking for mem leaks before that point finds a lot of
things! We should explicitly close down the library if we're checking for
mem leaks.
Reviewed-by: Rich Salz <rsalz@openssl.org>
version32.rc was not created on Windows. The if condition has been corrected.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
* added missing help option messages
* ecdh_single option is removed as it is a no-op and not an option
supported in earlier versions
* ssl_ctx_security_debug() was invoked before ctx check for NULL
* trusted_first option can be removed, as it is always enabled in 1.1.
But not removed the option, require confirmation.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Various Makefile.in files have changes for auto-init/de-init. Make the
equivalent changes in build.info.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Provide some man pages for auto-init/deinit. Also update the INSTALL
documentation for information on the new Configure options implemented as
part of this.
Reviewed-by: Richard Levitte <levitte@openssl.org>
This option disables automatic loading of the crypto/ssl error strings in
order to keep statically linked executable file size down
Reviewed-by: Richard Levitte <levitte@openssl.org>
This commit provides the basis and core code for an auto initialisation
and deinitialisation framework for libcrypto and libssl. The intention is
to remove the need (in many circumstances) to call explicit initialise and
deinitialise functions. Explicit initialisation will still be an option,
and if non-default initialisation is needed then it will be required.
Similarly for de-initialisation (although this will be a lot easier since
it will bring all de-initialisation into a single function).
Reviewed-by: Richard Levitte <levitte@openssl.org>
It seems realpath() is quite buggy on VMS, or will at least give quite
surprising results. On the other hand, realpath() is the better on
Unix to clean out clutter like foo/../bar on Unix.
So we make out own function to get the absolute directory for a given
input, and use rel2abs() or realpath() depending on the platform
Configure runs on.
Issue reported by Steven M. Schweda <sms@antinode.info>
Reviewed-by: Andy Polyakov <appro@openssl.org>
To be able to run tests when we've built in a directory other than
the source tree, the testing framework needs a few adjustments.
test/testlib/OpenSSL/Test.pm needs to know where it can find
shlib_wrap.sh, and a number of other tests need to be told a different
place to find engines than what they may be able to figure out on
their own. Relying to $TOP is not enough, $SRCTOP and $BLDTOP can be
used as an alternative.
As part of this change, top_file and top_dir are removed and
srctop_file, bldtop_file, srctop_dir and bldtop_dir take their place.
Reviewed-by: Ben Laurie <ben@openssl.org>
util/mkdef.pl and Makefile.shared needs to know about the source and
the build directories.
Additionally, Makefile.shared needs to know how to build shared
libraries in a directory other than the current one.
Reviewed-by: Ben Laurie <ben@openssl.org>
This documents describes the three steps from build.info files via the
%unified_info database to the build-file templates, along with some
examples showing how the data gets processed along the way.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
common.tmpl will be used together with the template build file, and is
the engine that connects the information gathered from all the
build.info files with making the build file itself.
This file expects there to be a template section in the build file
template that defines a number perl functions designed to return
strings with appropriate lines for the build system at hand. The
exact functions, what they can expect as arguments and what output
they're expected to produce is documented in Configurations/README.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
- One typo fixed in crypto/bio/b_addr.c
- Add a comment in doc/crypto/BIO_parse_hostserv.pod to explain the
blank lines with one lonely space each.
Reviewed-by: Andy Polyakov <appro@openssl.org>
As documented both SSL_get0_dane_authority() and SSL_get0_dane_tlsa()
are expected to return a negative match depth and nothing else when
verification fails. However, this only happened when verification
failed during chain construction. Errors in verification of the
constructed chain did not have the intended effect on these functions.
This commit updates the functions to check for verify_result ==
X509_V_OK, and no longer erases any accumulated match information
when chain construction fails. Sophisticated developers can, with
care, use SSL_set_verify_result(ssl, X509_V_OK) to "peek" at TLSA
info even when verification fail. They must of course first check
and save the real error, and restore the original error as quickly
as possible. Hiding by default seems to be the safer interface.
Introduced X509_V_ERR_DANE_NO_MATCH code to signal failure to find
matching TLSA records. Previously reported via X509_V_ERR_CERT_UNTRUSTED.
This also changes the "-brief" output from s_client to include
verification results and TLSA match information.
Mentioned session resumption in code example in SSL_CTX_dane_enable(3).
Also mentioned that depths returned are relative to the verified chain
which is now available via SSL_get0_verified_chain(3).
Added a few more test-cases to danetest, that exercise the new
code.
Resolved thread safety issue in use of static buffer in
X509_verify_cert_error_string().
Fixed long-stating issue in apps/s_cb.c which always sets verify_error
to either X509_V_OK or "chain to long", code elsewhere (e.g.
s_time.c), seems to expect the actual error. [ The new chain
construction code is expected to correctly generate "chain
too long" errors, so at some point we need to drop the
work-arounds, once SSL_set_verify_depth() is also fixed to
propagate the depth to X509_STORE_CTX reliably. ]
Reviewed-by: Rich Salz <rsalz@openssl.org>
Don't cast malloc-family return values.
Also found some places where (a) blank line was missing; and (b)
the *wrong* return value was checked.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Just like File::Path::make_path, File::Path::remove_tree didn't show
up before File::Path 2.06 / perl v5.10.1, so we prefer the legacy
function here as well.
Reviewed-by: Andy Polyakov <appro@openssl.org>
File::Path::make_path didn't show up before File::Path 2.06 / perl v5.10.1.
Because we're trying to stay compatible with perl v5.10.0 and up,
it's better to use the legacy interface.
Reviewed-by: Andy Polyakov <appro@openssl.org>
This works on Linux with Make already, and allows running only specified
tests.
Reviewed-by: Ben Laurie <ben@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Previous commit 7bb196a71 attempted to "fix" a problem with the way
SSL_shutdown() behaved whilst in mid-handshake. The original behaviour had
SSL_shutdown() return immediately having taken no action if called mid-
handshake with a return value of 1 (meaning everything was shutdown
successfully). In fact the shutdown has not been successful.
Commit 7bb196a71 changed that to send a close_notify anyway and then
return. This seems to be causing some problems for some applications so
perhaps a better (much simpler) approach is revert to the previous
behaviour (no attempt at a shutdown), but return -1 (meaning the shutdown
was not successful).
This also fixes a bug where SSL_shutdown always returns 0 when shutdown
*very* early in the handshake (i.e. we are still using SSLv23_method).
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
dgst: using digest instead of specific digest commands
the digest list specified in man dgst may be inaccurate, hence using
digest and referring to the list in digest-commands
'sha' as a digest name is no longer supported
dgst,pkeyutl cmds help cleanup
- In dgst, pkeyutl cmds, some options help was missing.
- fixed a minor typo in openssl.pod, that fixes make install.
- digest-commands was showing ‘sha’, which is not a supported digest
anymore.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Tim Hudson <tjh@openssl.org>
opt_valtype 0 is same as '-' while printing cmd usage
asn1parse/ca/ciphers help cleanup
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Some time ago, we had a ex_libs configuration setting that could be
divided into lflags and ex_libs. These got divided in two settings,
lflags and ex_libs, and the former was interpreted to be general
linking flags.
Unfortunately, that conclusion wasn't entirely accurate. Most of
those linking were meant to end up in a very precise position on the
linking command line, just before the spec of libraries the linking
depends on.
Back to the drawing board, we're diving things further, now having
lflags, which are linking flags that aren't depending on command line
position, plib_lflags, which are linking flags that should show up just
before the spec of libraries to depend on, and finally ex_libs, which
is the spec of extra libraries to depend on.
Also, documentation is changed in Configurations/README. This was
previously forgotten.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Allow initial engine names as first parameters before flags.
Also add engine param to help summary
Wrote manpage
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Clang rightly does not like extern symbols that are not declared
in any header file, as typically these are not intended for global
visibility and are exposed in error. This was indeed the case with
various file-scope objects in dtlsv1listentest.c.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Make it clear that if we are unable to get hold of the peer address then
*peer is cleared and the family set to AF_UNSPEC.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Adds a set of tests for the newly rewritten DTLSv1_listen function.
The test pokes various packets at the function and then checks
the return value and the data written out to ensure it is what we
would have expected.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The recently rewriten DTLSv1_listen code does not support fragmented
ClientHello messages because fragment reassembly requires server state
which is against the whole point of DTLSv1_listen. This change adds some
partial support for fragmented ClientHellos. It requires that the cookie
must be within the initial fragment. That way any non-initial ClientHello
fragments can be dropped and fragment reassembly is not required.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Adds a new function BIO_ADDR_clear to reset a BIO_ADDR back to an
unitialised state, and to set the family to AF_UNSPEC.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The DTLSv1_listen function exposed details of the underlying BIO
abstraction and did not properly allow for IPv6. This commit changes the
"peer" argument to be a BIO_ADDR and makes it a first class function
(rather than a ctrl) to ensure proper type checking.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Replace all magic numbers with #defined constants except in boolean
functions that return 0 for failure and 1 for success. Avoid a
couple memory leaks in error recovery code paths. Code style
improvements.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
Following on from earlier commits to prevent local symbols from being
exported in the shared libraries on Linux, this makes the equivalent changes
for Solaris.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add new function EC_KEY_priv2buf() to allocated and encode private
key octet in one call. Update and simplify ASN.1 and print routines.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Don't require an application to work out the appropriate buffer size for
ASN1_bn_print(), which is unsafe. Ignore the supplied buffer and allocate
it internally instead.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
New functions EC_KEY_oct2priv and EC_KEY_priv2oct. These are private key
equivalents of EC_POINT_oct2point and EC_POINT_point2oct which convert
between the private key octet format and EC_KEY.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
This uilds on the same way of checking for availability as we do in
TLSProxy. We use all IP factories we know of, starting with those who
know both IPv6 and IPv4 and ending with the one that only knows IPv4
and cache their possible success as foundation for checking the
available of each IP domain.
80-test_ssl.t has bigger chances of working on platforms that do not
run both IP domains.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This is an important move if scripts want to refer to the loaded
module without having perl think it needs to be loaded (again).
Reviewed-by: Rich Salz <rsalz@openssl.org>
Add no-async option to Configure that forces ASYNC_NULL.
Related to RT1979
An embedded system or replacement C library (e.g. musl or uClibc)
may not support the *context APIs that are needed for async operation.
Compiles with musl. Ran unit tests, async tests skipped as expected.
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
Both getaddrinfo() and getnameinfo() have to be preceeded with a call
to BIO_sock_init().
Also, make sure to give gai_strerror() the actual error code.
Reviewed-by: Stephen Henson <steve@openssl.org>
Those even order that do not play nicely with Montgomery arithmetic
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Emilia Käsper <emilia@openssl.org>
In HMAC_Init_ex, NULL key signals reuse, but in single-shot HMAC,
we can allow it to signal an empty key for convenience.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The test program clienthello checks TLS extensions, so there's no
point running it when no TLS protocol is available.
Reviewed-by: Ben Laurie <ben@openssl.org>
When connecting to "localhost" the Proxy's choice of client address
family may not match the server's choice address family. Without
MultiHomed => 1, the proxy may try the wrong address family first,
and give up without trying the other.
Reviewed-by: Richard Levitte <levitte@openssl.org>
This makes use of TLSProxy, which was expanded to use IO::Socket::IP
(which is a core perl module) or IO::Socket::INET6 (which is said to
be more popular) instead IO::Socket::INET if one of them is installed.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
s_socket.c gets brutally cleaned out and now consists of only two
functions, one for client and the other for server. They both handle
AF_INET, AF_INET6 and additionally AF_UNIX where supported. The rest
is just easy adaptation.
Both s_client and s_server get the new flags -4 and -6 to force the
use of IPv4 or IPv6 only.
Also, the default host "localhost" in s_client is removed. It's not
certain that this host is set up for both IPv4 and IPv6. For example,
Debian has "ip6-localhost" as the default hostname for [::1]. The
better way is to default |host| to NULL and rely on BIO_lookup() to
return a BIO_ADDRINFO with the appropriate loopback address for IPv4
or IPv6 as indicated by the |family| parameter.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
The control commands that previously took a struct sockaddr * have
been changed to take a BIO_ADDR * instead.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
This adds a couple of simple tests to see that SSL traffic using the
reimplemented BIO_s_accept() and BIO_s_connect() works as expected,
both on IPv4 and on IPv6.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Because of the way bio_lcl.h is organised, we must not include
internal/cryptlib.h before it. As a matter of fact, bio_lcl.h
includes internal/cryptlib.h on its own.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Because different platforms have different levels of support for IPv6,
different kinds of sockaddr variants, and some have getaddrinfo et al
while others don't, we could end up with a mess if ifdefs, duplicate
code and other maintainance nightmares.
Instead, we're introducing wrappers around the common form for socket
communication:
BIO_ADDR, closely related to struct sockaddr and some of its variants.
BIO_ADDRINFO, closely related to struct addrinfo.
With that comes support routines, both convenient creators and
accessors, plus a few utility functions:
BIO_parse_hostserv, takes a string of the form host:service and
splits it into host and service. It checks for * in both parts, and
converts any [ipv6-address] syntax to ust the IPv6 address.
BIO_lookup, looks up information on a host.
All routines handle IPv4 (AF_INET) and IPv6 (AF_INET6) addresses, and
there is support for local sockets (AF_UNIX) as well.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Accept leading 0-byte in PKCS1 type 1 padding. Internally, the byte is
stripped by BN_bn2bin but external callers may have other expectations.
Reviewed-by: Kurt Roeckx<kurt@openssl.org>
CRIME protection: disable compression by default, even if OpenSSL is
compiled with zlib enabled. Applications can still enable compression by
calling SSL_CTX_clear_options(ctx, SSL_OP_NO_COMPRESSION), or by using
the SSL_CONF library to configure compression. SSL_CONF continues to
work as before:
SSL_CONF_cmd(ctx, "Options", "Compression") enables compression.
SSL_CONF_cmd(ctx, "Options", "-Compression") disables compression (now
no-op by default).
The command-line switch has changed from -no_comp to -comp.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Also fix option processing in pkeyutl to allow use of (formerly)
"out-of-order" switches that were needless implementation limitations.
Handle documented "ENGINE" form with -keyform and -peerform.
Better handling of OPENSSL_NO_ENGINE and OPENSSL_NO_RSA.
RT2018
Reviewed-by: Rich Salz <rsalz@openssl.org>
It turns out that the combination splitpath() could return an empty
string for the directory part. This doesn't play well with catdir().
Reviewed-by: Rich Salz <rsalz@openssl.org>
Add new function BN_bn2binpad() which checks the length of the output
buffer and pads the result with zeroes if necessary.
New functions BN_bn2lebinpad() and BN_lebin2bn() which use little endian
format.
Reviewed-by: Rich Salz <rsalz@openssl.org>
PACKET contents should be read-only. To achieve this, also
- constify two user callbacks
- constify BUF_reverse.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Now that we have the foundation for the "unified" build scheme in
place, we add build.info files. They have been generated from the
Makefiles in the same directories. Things that are platform specific
will appear in later commits.
Reviewed-by: Andy Polyakov <appro@openssl.org>
The "unified" build scheme revolves around small information files,
build.info, which each describe their own bit of everything that needs
to be built, using a mini-language described in Configurations/README.
The information in build.info file contain references to source files
and final result. Object files are not mentioned at all, they are
simply from source files. Because of this, all the *_obj items in
Configurations/*.conf are renamed to *_asm_src and the files listed
in the values are change from object files to their corresponding
source files. For the sake of the other build schemes, Configure
generates corresponding *_obj entries in %target.
Furthermore, the "unified" build scheme supports having a build
directory tree separate from the source directry tree.
All paths in a build.info file is assumed to be relative to its
location, either within the source tree or within the build tree.
Reviewed-by: Andy Polyakov <appro@openssl.org>
When auxiliary data contains only reject entries, continue to trust
self-signed objects just as when no auxiliary data is present.
This makes it possible to reject specific uses without changing
what's accepted (and thus overring the underlying EKU).
Added new supported certs and doubled test count from 38 to 76.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
This includes basic constraints, key usages, issuer EKUs and auxiliary
trust OIDs (given a trust suitably related to the intended purpose).
Added tests and updated documentation.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
d2i_ECPrivateKey always caculates the public key so there is
no need to caculate it again in eckey_priv_decode().
Reviewed-by: Rich Salz <rsalz@openssl.org>
New functions to return internal pointer for order and cofactor. This
avoids the need to allocate a new BIGNUM which to copy the value to.
Simplify code to use new functions.
Reviewed-by: Rich Salz <rsalz@openssl.org>
When the target is {something}-icc, we're doing some extra checks of
the icc compiler. However, all such targets were cleaned away in
March 2015, so this Configure section is dead code.
Reviewed-by: Rich Salz <rsalz@openssl.org>
There was an unused macro in ssl_locl.h that used an internal
type, so I removed it.
Move bio_st from bio.h to ossl_type.h
Reviewed-by: Andy Polyakov <appro@openssl.org>
top_dir() are used to create directory names, top_file() should be
used for files. In a Unixly environment, that doesn't matter, but...
Reviewed-by: Rich Salz <rsalz@openssl.org>
Not all architectures have a time_t defined the same way. To make
sure we get the same result, we need to cast &checkoffset to (intmax_t *)
and make sure that intmax_t is defined somehow.
To make really sure we don't pass a variable with the wrong size down
to opt_imax(), we use a temporary intmax_t.
Reviewed-by: Rich Salz <rsalz@openssl.org>
As a side-effect of opaque x509, ex_flags were looked up too early,
before additional policy cache updates.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
The lflags configuration had a weird syntax with a % as separator. If
it was present, whatever came before ended up as PEX_LIBS in Makefile
(usually, this is LDFLAGS), while whatever came after ended up as
EX_LIBS.
This change splits that item into lflags and ex_libs, making their use
more explicit.
Also, PEX_LIBS in all the Makefiles are renamed to LDFLAGS.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Remove depend hacks from demos/engines.
Remove clean-depend; just call makedepend (or $CC -M) and use that.
Reviewed-by: Richard Levitte <levitte@openssl.org>
By default X509_check_trust() trusts self-signed certificates from
the trust store that have no explicit local trust/reject oids
encapsulated as a "TRUSTED CERTIFICATE" object. (See the -addtrust
and -trustout options of x509(1)).
This commit adds a flag that makes it possible to distinguish between
that implicit trust, and explicit auxiliary settings.
With flags |= X509_TRUST_NO_SS_COMPAT, a certificate is only trusted
via explicit trust settings.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
The use of the uninitialized buffer in the RNG has no real security
benefits and is only a nuisance when using memory sanitizers.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Add tests for have_precompute_mult for the optimised curves (nistp224,
nistp256 and nistp521) if present
Reviewed-by: Richard Levitte <levitte@openssl.org>
During precomputation if the group given is well known then we memcpy a
well known precomputation. However we go the wrong label in the code and
don't store the data properly. Consequently if we call have_precompute_mult
the data isn't there and we return 0.
RT#3600
Reviewed-by: Richard Levitte <levitte@openssl.org>
The function DH_check_pub_key() was missing some return value checks in
some calls to BN functions.
RT#4278
Reviewed-by: Andy Polyakov <appro@openssl.org>
A new return value for DH_check_pub_key was recently added:
DH_CHECK_PUBKEY_INVALID. As this is a flag which can be ORed with other
return values it should have been set to the value 4 not 3.
RT#4278
Reviewed-by: Andy Polyakov <appro@openssl.org>
This extends the existing async functionality to SSL_shutdown(), i.e.
SSL_shutdown() can now casuse an SSL_ERROR_WANT_ASYNC error to be returned
from SSL_get_error() if async mode has been enabled.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Following on from the previous commit, add a test to ensure that
DH_compute_key correctly fails if passed a bad y such that:
y^q (mod p) != 1
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Historically OpenSSL only ever generated DH parameters based on "safe"
primes. More recently (in version 1.0.2) support was provided for
generating X9.42 style parameter files such as those required for RFC
5114 support. The primes used in such files may not be "safe". Where an
application is using DH configured with parameters based on primes that
are not "safe" then an attacker could use this fact to find a peer's
private DH exponent. This attack requires that the attacker complete
multiple handshakes in which the peer uses the same DH exponent.
A simple mitigation is to ensure that y^q (mod p) == 1
CVE-2016-0701
Issue reported by Antonio Sanso.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Many options for supporting optimizations for legacy crypto on legacy
platforms have been removed. This simplifies the source code and
does not really penalize anyone.
DES_PTR (always on)
DES_RISC1, DES_RISC2 (always off)
DES_INT (always 'unsigned int')
DES_UNROLL (always on)
BF_PTR (always on) BF_PTR2 (removed)
MD2_CHAR, MD2_LONG (always 'unsigned char')
IDEA_SHORT, IDEA_LONG (always 'unsigned int')
RC2_SHORT, RC2_LONG (always 'unsigned int')
RC4_LONG (only int and char (for assembler) are supported)
RC4_CHUNK (always long), RC_CHUNK_LL (removed)
RC4_INDEX (always on)
And also make D_ENCRYPT macro more clear (@appro)
This is done in consultation with Andy.
Reviewed-by: Andy Polyakov <appro@openssl.org>
A mistake was made and the setting of this config variable got
reverted to an older behavior. This restores the latest.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
It seems that Test::More doesn't like 0 tests, a line like this raises
an error and stops the recipe entirely:
plan tests => 0;
So we need to check for 0 tests beforehand and skip the subtest
explicitely in that case.
Reviewed-by: Rich Salz <rsalz@openssl.org>
$EXE_SHELL should only be used with out own programs, not with
surrounding programs such as the perl interpreter.
Reviewed-by: Matt Caswell <matt@openssl.org>
This was done by the following
find . -name '*.[ch]' | /tmp/pl
where /tmp/pl is the following three-line script:
print unless $. == 1 && m@/\* .*\.[ch] \*/@;
close ARGV if eof; # Close file to reset $.
And then some hand-editing of other files.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Also removes the make variable SHARED_LIBS_LINK_EXTS, only used by
the clean-shared target.
When shared library linking was moved to the separate Makefile.shared
in commit 30afcc072a, this target was
skipped. Prior to that commit, clean-shared was invoked as a
dependency of build-shared, but afterward it was no longer referenced
anywhere in the tree.
Instead of porting the functionality over to Makefile.shared, just
remove it entirely, as it appears to be unused.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Missing SKIP: block in SSL unit tests for DTLS and TLS version tests.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Since we're building some of our perl scripts and the result might not
end up in apps/ (*), we may need to treat them like the compile
programs we use for testing.
This introduces perlapp() and perltest(), which behave like app() and
test(), but will add the perl executable in the command line.
-----
(*) For example, with a mk1mf build, the result will end up in $(BIN_D)
Reviewed-by: Matt Caswell <matt@openssl.org>
The SSL and SSL_CTX structures are reference counted. However since libssl
was made opaque there is no way for users of the library to manipulate the
reference counts. This adds functions to enable that.
Reviewed-by: Stephen Henson <steve@openssl.org>
Enhances the routines in OpenSSL::Test::Utils for checking disabled
stuff to get their information directly from Configure instead of
'openssl list -disabled'.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
engines_obj changed name to padlock_obj in Configure. We need to do
the corresponding ENGINES_ASM_OBJ -> PADLOCK_ASM_OBJ in appropriate
Makefile.ins.
Reviewed-by: Rich Salz <rsalz@openssl.org>
- Small rearrangement of the TABLE and HASH printouts, and adding
printout of the "build_scheme" item
- Renamed "engines_obj" to "padlock_obj"
- Moved the runs of dofile down... it didn't quite make sense to have
that in the middle of a printout
Reviewed-by: Rich Salz <rsalz@openssl.org>
Now that we're using templates, we should warn people not to edit the
resulting file. We do it through util/dofile.pl, which is enhanced
with an option to tell what file it was called from. We also change
the calls so the template files are on the command line instead of
being redirected through standard input. That way, we can display
something like this (example taken from include/openssl/opensslconf.h):
/* WARNING: do not edit! */
/* Generated by Configure from include/openssl/opensslconf.h.in */
Reviewed-by: Rich Salz <rsalz@openssl.org>
It is time for Makefile.org to fold into the new regime and have a run
through util/dofile.pl. This forces some information out of there and
into Configure, which isn't a bad thing, it makes Configure
increasingly the center of build information, which is as it should
be.
A few extra defaults were needed in the BASE template to get rid of
warnings about missing values.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Move opensslconf.h.in to include/openssl.
Split off DES,BN,RC4 stuff into separate header file
templates in crypto/include/internal/*_conf.h.in
Reviewed-by: Richard Levitte <levitte@openssl.org>
This is an internal facility, never documented, not for
public consumption. Move it into ssl (where it's only used
for DTLS).
I also made the typedef's for pqueue and pitem follow our style: they
name structures, not pointers.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The explanation is that it falls back to using tools/c_rehash if
'apps/openssl rehash' isn't supported on the platform at hand.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Now that configdata.pm is the centre of information, use that instead
of Makefile to figure out reconfiguration parameters. This will help
future development with different Makefile file names.
The code to read necessary configuration data from Makefile is retained
for an easy transition to configdata.pm based information gathering. It
will be removed later on.
This change includes moving the variable $cross_compile_prefix to %config.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
When experimental-store is enabled, it does not compile due to the
change to opaque data structures.
Change CRYPTO_add() to EVP_PKEY_up_ref() as needed.
Signed-off-by: Kurt Roeckx <kurt@roeckx.be>
Reviewed-by: Rich Salz <rsalz@openssl.org>
RT: #4263, GH: #579
The turn has come to have crypto/opensslconf.h.in get run through
util/dofile.pl. The consequence is that a large number of variables
get moved to the %config table.
Also, the string variables $openssl_*, which were populated with cpp
lines, all being of the form "#define SOMETHING", were converted into
ARRAY refs in %config values, containing just the list of macros to be
defined.
Reviewed-by: Rich Salz <rsalz@openssl.org>
For this adaptation, the variables $options and $version needed to
move to %config in Configure, and why not move all other variables
holding diverse version numbers at the same time?
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Because we're using Text::Template and we know it's a non core Perl
module, we choose to bundle it into our source, for convenience.
external/perl/Downloaded.txt document what modules we choose to bundle
this way and exactly where we downloaded it from.
With this changes comes the transfer module for with_fallback.
Reviewed-by: Rich Salz <rsalz@openssl.org>
For our own convenience, we need a mechanism to be able to fall back
on bundled Perl modules. It's a minimal package that's called like
this:
use with_fallback qw(Module1 Module2 ...);
For each module, it will try to require them from the system
installation, and failing that, it will temporarly add external/perl
and try to require transfer::{ModuleName}. It requires that each
bundled Perl modules is accompanied by a small transfer module
(external/perl/transfer/ModuleName.pm in our example) that knows
exactly what to load.
Reviewed-by: Rich Salz <rsalz@openssl.org>
apps/CA.pl and tools/c_rehash are built from template files. So far,
this was done by Configure, which created its own problems as it
forced everyone to reconfigure just because one of the template files
had changed.
Instead, have those files created as part of the normal build in apps/
and in tools/.
Furthermore, this prepares for a future where Configure may produce
entirely other build files than Makefile, and the latter can't be
guaranteed to be the holder of all information for other scripts.
Instead, configdata.pm (described below) becomes the center of
configuration information.
This introduces a few new things:
%config a hash table to hold all kinds of configuration data
that can be used by any other script.
configdata.pm a perl module that Configure writes. It currently
holds the hash tables %config and %target.
util/dofile.pl a script that takes a template on STDIN and outputs
the result after applying configuration data on it.
It's supposed to be called like this:
perl -I$(TOP) -Mconfigdata < template > result
or
perl -I$(TOP) -Mconfigdata templ1 templ2 ... > result
Note: util/dofile.pl requires Text::Template.
As part of this changed, remove a number of variables that are really
just copies of entries in %target, and use %target directly. The
exceptions are $target{cflags} and $target{lflags}, they do get copied
to $cflags and $lflags. The reason for this is that those variable
potentially go through a lot of changes and would rather deserve a
place in %config. That, however, is for another commit.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
It's time for print_table_entry to get a bit of refreshment. The way it
was put together, we needed to maintain the list of known configuration
keys of interest twice, in different shapes. This is error prone, so
move the list of strings to a common list for all printing cases, and
use simple formatting of lines to do the actual printout based on that
list.
Reviewed-by: Rich Salz <rsalz@openssl.org>
The code is trying to interpolate the value of the BASE_SECTION macro,
but due to excess escaping, it instead prints the string "BASE_SECTION".
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Some things to ignore need to be properly rooted, and use a bit more
precision on ignoring 'lib', as that maybe be a perfectly valid
directory name to add into git elsewhere in the source tree.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Time to get rid of @MK1MF_Builds and introduce a more flexible
'build_scheme' configuration key. Its value may be a string or an
array of strings, meaning we need to teach resolve_config how to
handle ARRAY referenses.
The build scheme is a word that selects a function to create the
appropriate result files for a certain configuration. Currently valid
build schemes aer "mk1mf" and "unixmake", the plan is however to add
at least one other for a more universal build scheme.
Incidently, this also adds the functions 'add' and 'add_before', which
can be used in a configuration, so instead of having to repeatedly
write a sub like this:
key1 => sub { join(" ", @_, "myvalues"); },
key2 => sub { join(" ", "myvalues", @_); },
one could write this:
key1 => add(" ", "myvalues"),
key2 => add_before(" ", "myvalues"),
The good point with 'add' and 'add_before' is that they handle
inheritances where the values are a misture of scalars and ARRAYs. If
there are any ARRAY to be found, the resulting value will be an ARRAY,
otherwise it will be a scalar with all the incoming valued joined
together with the separator given as first argument to add/add_before.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Move the documentation of the target configuration form to
Configurations/README.
Move initial assembler object templates to
Configurations/00-BASE-templates.conf.
Furthermore, remove all variables containing the names of the
non-assembler object files and make a BASE template of them instead.
The values from this templates are used as defaults as is. The
remaining manipulation of data when assembler modules are used is done
only when $no_asm is false.
While doing this, clean out some other related variables that aren't
used anywhere.
Also, we had to move the resolution of the chosen target a bit, or the
function 'asm' would never catch a true $no_asm... this hasn't
mattered before we've moved it all to the BASE template, but now it
does.
At the same time, add the default for the 'unistd' key to the BASE
template.
Reviewed-by: Rich Salz <rsalz@openssl.org>
--prefix is now exclusively used for software and manual installation.
--openssldir is not exclusively used as a default location for certs,
keys and the default openssl.cnf.
This change is made to bring clarity, to have the two less
intertwined, and to be more compatible with the usual ways of software
installation.
Please change your habits and scripts to use --prefix rather than
--openssldir for installation location now.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
It's time to refactor the handling of %disabled so that all
information of value is in the same place. We have so far had a few
cascading disable rules in form of code, far away from %disabled.
Instead, bring that information to the array @disable_cascade, which
is a list of pairs of the form 'test => descendents'. The test part
can be a string, and it's simply checked if that string is a key in
%disabled, or it can be a CODEref to do a more complex test. If the
test comes true, then all descendents are disabled. This check is
performed until there are no more things that need to be disabled.
Also, $default_depflags is constructed from the information in
%disabled instead of being a separate string. While a string of its
own is visually appealing, it's much too easy to forget to update it
when something is changed in %disabled.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
The way the "reconf"/"reconfigure" argument is handled is overly
complicated. Just grep for it first, and if it is there in the
current arguments, get the old command line arguments from Makefile.
While we're at it, make the Makefile variable CONFIGURE_ARGS hold the
value as a perl list of strings. This makes things much safer in case
one of the arguments would contain a space. Since CONFIGURE_ARGS is
used for nothing else, there's no harm in this.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
It is sometimes useful (especially in automated tests) to supply
multiple trusted or untrusted certificates via separate files rather
than have to prepare a single file containing them all.
To that end, change verify(1) to accept these options zero or more
times. Also automatically set -no-CAfile and -no-CApath when
-trusted is specified.
Improve verify(1) documentation, which could still use some work.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Returning untrusted is enough for for full chains that end in
self-signed roots, because when explicit trust is specified it
suppresses the default blanket trust of self-signed objects.
But for partial chains, this is not enough, because absent a similar
trust-self-signed policy, non matching EKUs are indistinguishable
from lack of EKU constraints.
Therefore, failure to match any trusted purpose must trigger an
explicit reject.
Reviewed-by: Richard Levitte <levitte@openssl.org>
These can be re-generated via:
cd test/certs; ./setup.sh
if need be. The keys are all RSA 2048-bit keys, but it is possible
to change that via environment variables.
cd test/certs
rm -f *-key.pem *-key2.pem
OPENSSL_KEYALG=rsa OPENSSL_KEYBITS=3072 ./setup.sh
cd test/certs
rm -f *-key.pem *-key2.pem
OPENSSL_KEYALG=ecdsa OPENSSL_KEYBITS=secp384r1 ./setup.sh
...
Keys are re-used if already present, so the environment variables
are only used when generating any keys that are missing. Hence
the "rm -f"
Reviewed-by: Richard Levitte <levitte@openssl.org>
When DANE-EE(3) matches or either of DANE-EE/PKIX-EE fails, we don't
build a chain at all, but rather succeed or fail with just the leaf
certificate. In either case also check for Suite-B violations.
As unlikely as it may seem that anyone would enable both DANE and
Suite-B, we should do what the application asks.
Took the opportunity to eliminate the "cb" variables in x509_vfy.c,
just call ctx->verify_cb(ok, ctx)
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
Split the read_config function into read_config that ONLY reads the
configuration files but doesn't try to resolve any of the
inheritances, and resolve_config which resolves the inheritance chain
of a given target. Move them to the bottom of Configure, with the
rest of the helpers.
Have a new small hash table, %target, which will hold the values for
the target the user requested. This also means that all access to the
current target data can be reduced from '$table{$target}->{key}' to a
mere '$target{key}'.
While we're at it, the old string formatted configurations are getting
obsolete, so they may as well get deprecated entirely.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Get rid of the --test-sanity option. Since we no longer have string
based configurations, we don't have the problem with miscounting
colons any more.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Start simple, removed some unused variables and change all '<<EOF' to
'<<"EOF"'. The latter is because some code colorizers (notably, in
emacs) cannot recognise the here document end marker unless it's
quoted and therefore assume the rest of the file is part of the here
document.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Rename 'update' to 'generate'. Rather than recurse, just explicitly
call the three generate targets directly.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Calling SSL_shutdown while in init previously gave a "1" response, meaning
everything was successfully closed down (even though it wasn't). Better is
to send our close_notify, but fail when trying to receive one.
The problem with doing a shutdown while in the middle of a handshake is
that once our close_notify is sent we shouldn't really do anything else
(including process handshake/CCS messages) until we've received a
close_notify back from the peer. However the peer might send a CCS before
acting on our close_notify - so we won't be able to read it because we're
not acting on CCS messages!
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
For BSD systems, Configure adds a shared_ldflags including a reference
to the Makefile variable LIBRPATH, but since it must be passed down to
Makefile.shared, care must be taken so the value of LIBRPATH doesn't
get expanded too early, or it ends up giving an empty string.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Some users want to disable SSL 3.0/TLS 1.0/TLS 1.1, and enable just
TLS 1.2. In the future they might want to disable TLS 1.2 and
enable just TLS 1.3, ...
This commit makes it possible to disable any or all of the TLS or
DTLS protocols. It also considerably simplifies the SSL/TLS tests,
by auto-generating the min/max version tests based on the set of
supported protocols (425 explicitly written out tests got replaced
by two loops that generate all 425 tests if all protocols are
enabled, fewer otherwise).
Reviewed-by: Richard Levitte <levitte@openssl.org>
Fix a typo in the definition of the GOST2012-NULL-GOST12 ciphersuite.
RT#4213
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
It seems risky in the context of cross-signed certificates when the
same certificate might have multiple potential issuers. Also rarely
used, since chains in OpenSSL typically only employ self-signed
trust-anchors, whose self-signatures are not checked, while untrusted
certificates are generally ephemeral.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
Cygwin was used for x86 before, so let's keep it around for those who
still use it (it make Configure reconf possible).
Cygwin-i[3456]86 for those that might generate and pass a target name
directly to Configure.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This patch allows to recognize the architectures supported by Cygwin
and to choose the right configuration from there. Drop -march to
use default architecture on 32 bit x86.
Drop pre-Cygwin-1.3 recognition since it's long gone and there's no
valid configuration for this anymore.
Signed-off-by: Corinna Vinschen <vinschen@redhat.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Building for the Cygwin distro requires to be able to build debuginfo
files. This in turn requires to build object files without stripping.
The stripping is performed by the next step after building which creates
the debuginfo files.
Signed-off-by: Corinna Vinschen <vinschen@redhat.com>
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Remove lint, tags, dclean, tests.
This is prep for a new makedepend scheme.
This is temporary pending unified makefile, and might help it.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Author: Remi Gacogne <rgacogne-github@coredump.fr>
GH334: Add an OCSP_SINGLERESP_get0_id() accessor to the OCSP_CERTID of
a OCSP_SINGLERESP. It is possible to do it the other way around using
OCSP_resp_find(), but this is more efficient when you have a tree indexed
by OCSP_CERTID, like haproxy does. (This is also RT4251)
Author: Marek Klein <kleinmrk@gmail.com>
GH556: OCSP_resp_get_produced_at() accessor to the producedAt of a
OCSP_BASICRESP
GH555: TS_STATUS_INFO_get_status(), TS_STATUS_INFO_get_text() and
TS_STATUS_INFO_get_failure_info() accessors for a TS_STATUS_INFO
Signed-off-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Also report an SSL_dane_enable error when the basedomain is an
invalid SNI name. Avoid side-effects when such a name is valid
with X509_VERIFY_PARAM_set1_host(), as e.g. with an empty name, by
setting the SNI name first.
Reviewed-by: Rich Salz <rsalz@openssl.org>
- bugfix: should not treat '--' as invalid domain substring.
- '-' should not be the first letter of a domain
Signed-off-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Rich Salz <rsalz@openssl.org>
The windows installation instructions were very out of date. Substantial
update to the text. Remove a lot of historical stuff that isn't relevant
any more, and merge the win64 and win32 instructions into one file.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The previous 'Relax the requirements for a debug build' commit had
an extra line of code that shouldn't have been there. This fixes it.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
It seems that the r modifier for s/// is fairly new. It's reported
not to exist in perl 5.10.1, so it's better to avoid it when
possible.
Reviewed-by: Tim Hudson <tjh@openssl.org>
The entropy-gathering daemon is used only on a small number of machines.
Provide a configure knob so that EGD support can be disabled by default
but re-enabled on those systems that do need it.
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
We required that a target be named 'debug-something' or to have at
least one of the configuration items debug_cflags and debug_lflags for
--debug to be accepted.
However, there are targets with no such markings but that will still
have debugging capabilities. This is particularly true for mk1mf
builds, where the extra flags for debugging are figured out later on
by util/mk1mf.pl.
Reviewed-by: Rich Salz <rsalz@openssl.org>
It turns out that -pause calls the undocumented function SSL_set_debug.
That just sets flag inside the SSL structure. That flag, despite
the command is never used. So remove the flag, the field, and the
function.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Tell open() O_BINARY on VMS doesn't make sense, as it's possible to
use more precise file attributes. However, if we're still going to
fdopen() it in binary mode, we must set the fd in binary context.
Reviewed-by: Rich Salz <rsalz@openssl.org>
On some platforms, the shell will determine what attributes a file
will have, so while the program might think it's safely outputting
binary data, it's not always true.
For the sake of the tests, it's therefore safer to use -out than to
use redirection.
Reviewed-by: Rich Salz <rsalz@openssl.org>
'openssl rehash' isn't implemented on all platforms, and since 'make
test' depends on a rehash of certs/demo being performed, it becomes an
effective block from running tests on any platform but Unix, for the
moment.
It's better to fall back to c_rehash and let the tests perform
everywhere.
Reviewed-by: Matt Caswell <matt@openssl.org>
On VMS, the command MCR will assume SYS$SYSTEM: when the first
argument lacks a directory spec. So for programs in the current
directory, we add [] to tell MCR it is in the current directory.
It's the same as having ./ at the start of a program on Unix so the
shell doesn't start looking along $PATH.
Reviewed-by: Rich Salz <rsalz@openssl.org>
This used to work but somewhere along the line it broke and was failing to
detect duplicate ordinals - which was the whole point of the test!
Reviewed-by: Rich Salz <rsalz@openssl.org>
util/mk1mf.pl was relying on the platform having the 'debug-' prefix
for doing a debug build. Since the setup of targets has changed, this
is no longer true. However, it can look for '--debug' in the command
line options.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Add CRYPTO_EX_DATA add EndC_KEY_[gs]et_method, From Roumen Petrov.
Had to add various exdata calls to init/copy/free the exdata.
Had to remove const from some EC functions because exdata isn't
const-correct. :(
Also remove EC_EXTRA_DATA and use a union to hold the possible
pre-computed values and an enum to tell which value is in the
union. (Rich Salz)
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
This test relies on a private function, which isn't exported.
This test would work better as a unit test in crypto/bn/bn_prime.c.
Reviewed-by: Rich Salz <rsalz@openssl.org>
For some strange reason opensslconf.h was only defining DES_LONG
when included via des.h, but that's exceedingly fragile (as a
result of include guards the include via des.h might not actually
process the content again).
Ripped out the nesting constraint, now always define OSSL_DES_LONG
if not already defined. Note, this could just be DES_LONG, but
trying to avoid exposing DES_LONG in places where it has never been
seen before, so it is up to des.h to actually define DES_LONG as
OSSL_DES_LONG.
Reviewed-by: Rich Salz <rsalz@openssl.org>
It seems like the convention for VMS exit codes is to combine the VMS
C facility code (0x35a000) with a recoded exit code as follows:
0 => 1
1-255 => 8*code + 2
We also add 0x10000000, which is the control bit that has DCL not
report the error on the terminal. That's just as well, since it would
be quite nonsensical, for example:
%C-W-NOMSG, Message number 0035A018
We could do all this by using the normal exit() function after having
defined the macro _POSIX_EXIT. Unfortunately, this feature only
exists in VMS C V7.1 and up.
Reviewed-by: Rich Salz <rsalz@openssl.org>
VMS being a record oriented operating system, it's uncertain how the
'pipe' passes binary data from one process to another. Experience
shows that we get in trouble, and it's probably due to the pipe in
itself being opened in text mode (variable length records).
It's safer to pass data via an intermediary file instead.
Reviewed-by: Rich Salz <rsalz@openssl.org>
VMS uses a variant of openssl.cnf named openssl-vms.cnf.
There's a Perl on VMS mystery where a open pipe will not SIGPIPE when
the child process exits, which means that a loop sending "y\n" to it
will never stop. Adding a counter helps fix this (set to 10, we know
that none of the CA.pl commands will require more).
Reviewed-by: Rich Salz <rsalz@openssl.org>
Well, I'm not actually changing time_t, just changing how time_t
valued opt values are converted from string to time_t.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Create Makefile's from Makefile.in
Rename Makefile.org to Makefile.in
Rename Makefiles to Makefile.in
Address review feedback from Viktor and Richard
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
They all stop including evp_locl.h, so we also take care of their
adaptation to opaque EVP_CIPHER_CTX, as was promised in an earlier
commit.
Reviewed-by: Rich Salz <rsalz@openssl.org>
We follow the method used for EVP_MD.
Also, move all the internal EVP_CIPHER building macros from evp_locl.h
to evp_int.h. This will benefit our builtin EVP_CIPHERs.
Reviewed-by: Rich Salz <rsalz@openssl.org>
EVP_CIPHER_CTX_flags was returning the flags of its associated
EVP_CIPHER. However, EVP_CIPHER_CTX has flags of its own, so this
function is quite confusing and therefore error prone.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Note: there's a larger number of implementations in crypto/evp/ that
aren't affected because they include evp_locl.h. They will be handled
in a separate commit.
Reviewed-by: Rich Salz <rsalz@openssl.org>
New functions:
- EVP_CIPHER_CTX_encrypting()
- EVP_CIPHER_CTX_iv()
- EVP_CIPHER_CTX_iv_noconst()
- EVP_CIPHER_CTX_original_iv()
- EVP_CIPHER_CTX_buf_noconst()
- EVP_CIPHER_CTX_num()
- EVP_CIPHER_CTX_set_num()
- EVP_CIPHER_CTX_cipher_data()
- EVP_CIPHER_CTX_new_cipher_data()
Note that the accessors / writers for iv, buf and num may go away, as
those rather belong in the implementation's own structure (cipher_data)
when the implementation would affect them (that would be the case when
they are flagged EVP_CIPH_CUSTOM_IV or EVP_CIPH_FLAG_CUSTOM_CIPHER).
Reviewed-by: Rich Salz <rsalz@openssl.org>
Following the method used for EVP_MD_CTX and HMAC_CTX,
EVP_CIPHER_CTX_init and EVP_CIPHER_CTX_cleanup are joined together
into one function, EVP_CIPHER_CTX_reset, with EVP_CIPHER_CTX_init kept
as an alias.
EVP_CIPHER_CTX_cleanup fills no purpose of its own any more and is
therefore removed.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Implement range-checking in all counts in apps. Turns out only a couple
of cases were missing. And make the range-checking code more strict.
Replace almost all opt_ulong() calls with opt_long()
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
It was assumed that the syntax FOO/Foo=PROCEDURE would create both an
upper case and mixed case symbol in the GST. Not so, it requires
having both FOO/Foo=PROCEDURE (to create the upper case alias) and
Foo=PROCEDURE (to create the mixed case slot).
We make sure that any symbol always occupies two slots (even those
that don't exist) by filling up with SPARE when necessary. That will
assure that any changes will still have the same symbols in the same
slots no matter what (save a complete rewrite of the ordinals files).
Reviewed-by: Rich Salz <rsalz@openssl.org>
It was assumed that a dummy with the type PRIVATE_PROCEDURE would
simply occupy a slot but otherwise ignore the symbol. Not so, but
there is SPARE for that purpose.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Make LHASH_OF use static inline functions.
Add new lh_get_down_load and lh_set_down_load functions and their
typesafe inline equivalents.
Make lh_error a function instead of a macro.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Since danetest is to test DANE rather than specific algorithms, it's
acceptable to require EC when testing it.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
These now take and return unsigned long, and get is constified.
Updated related documentation and util/ssleay.num
Reviewed-by: Matt Caswell <matt@openssl.org>
We use $default_depflags to check if a 'make depend' is needed after
configuring, so it needs to be kept up to date.
Reviewed-by: Matt Caswell <matt@openssl.org>
Commit 189ae368d9 (RT ticket 3352) provided the capability to output
session key data in NSS format. The big apps cleanup broke that capability.
This commit restores it.
RT#4201
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Make CRYPTO_mem_leaks() and CRYPTO_mem_leaks_fp() return a status value.
Update documentation. Don't abort() if there are leaks.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
In order for mkdep to find #ifdef'd functions, they must be
wrapped (in the header file) with
#ifndef OPENSSL_NO_...
So do that for various CRYPTO_mem_debug... things.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Provide backwards-compatiblity for functions, macros and include
files if OPENSSL_API_COMPAT is either not defined or defined less
than the version number of the release in which the feature was
deprecated.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Only two macros CRYPTO_MDEBUG and CRYPTO_MDEBUG_ABORT to control this.
If CRYPTO_MDEBUG is not set, #ifdef out the whole debug machinery.
(Thanks to Jakob Bohm for the suggestion!)
Make the "change wrapper functions" be the only paradigm.
Wrote documentation!
Format the 'set func' functions so their paramlists are legible.
Format some multi-line comments.
Remove ability to get/set the "memory debug" functions at runtme.
Remove MemCheck_* and CRYPTO_malloc_debug_init macros.
Add CRYPTO_mem_debug(int flag) function.
Add test/memleaktest.
Rename CRYPTO_malloc_init to OPENSSL_malloc_init; remove needless calls.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Applications wishing to include their own stacks now just need to include
DEFINE_STACK_OF(foo)
in a header file.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Change DECLARE_STACK_OF into inline functions. This avoids the need for
auto generated mkstack.pl macros and now handles const properly.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Also tweak some of the code in demos/bio, to enable interactive
testing of BIO_s_accept's use of SSL_dup. Changed the sconnect
client to authenticate the server, which now exercises the new
SSL_set1_host() function.
Reviewed-by: Richard Levitte <levitte@openssl.org>
If there's a failure allocating md_data, the destination pctx will have
a shared pointer with the source EVP_MD_CTX, which will lead to problems
when either the source or the destination is freed.
Reviewed-by: Stephen Henson <steve@openssl.org>
The protocol selection code is now consolidated in a few consecutive
short functions in a single file and is table driven. Protocol-specific
constraints that influence negotiation are moved into the flags
field of the method structure. The same protocol version constraints
are now applied in all code paths. It is now much easier to add
new protocol versions without reworking the protocol selection
logic.
In the presence of "holes" in the list of enabled client protocols
we no longer select client protocols below the hole based on a
subset of the constraints and then fail shortly after when it is
found that these don't meet the remaining constraints (suiteb, FIPS,
security level, ...). Ideally, with the new min/max controls users
will be less likely to create "holes" in the first place.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
SIZE_MAX is a great macro, and does unfortunately not exist everywhere.
Since we check against half of it, using bitwise shift to calculate the
value of half SIZE_MAX should be safe enough.
Reviewed-by: Tim Hudson <tjh@openssl.org>
If DSA parameters are absent return -1 (for unknown) in DSA_security_bits.
If parameters are absent when a certificate is set in an SSL/SSL_CTX
structure this will reject the certificate by default. This will cause DSA
certificates which omit parameters to be rejected but that is never (?)
done in practice.
Thanks to Brian 'geeknik' Carpenter for reporting this issue.
Reviewed-by: Emilia Käsper <emilia@openssl.org>
Previous versions of OpenSSL had the max size limit for a CertificateRequest
message as |s->max_cert_list|. Previously master had it to be
SSL3_RT_MAX_PLAIN_LENGTH. However these messages can get quite long if a
server is configured with a long list of acceptable CA names. Therefore
the size limit has been increased to be consistent with previous versions.
RT#4198
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Move all calls of the OCSP callback into one place, rather than repeating it
in two different places.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
It makes no sense to call the OCSP status callback if we are resuming a
session because no certificates will be sent.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
If a server sends the status_request extension then it may choose
to send the CertificateStatus message. However this is optional.
We were treating it as mandatory and the connection was failing.
Thanks to BoringSSL for reporting this issue.
RT#4120
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
If the client sends a status_request extension in the ClientHello
and the server responds with a status_request extension in the
ServerHello then normally the server will also later send a
CertificateStatus message. However this message is *optional* even
if the extensions were sent. This adds a test to ensure that if
the extensions are sent then we can still omit the message.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
When it's the last item that is removed int_thread_hash == hash and we would
still call int_thread_release(&hash) while hash is already freed. So
int_thread_release would compare that dangling pointer to NULL which is
undefined behaviour. Instead do already what int_thread_release() would do,
and make the call do nothing instead.
Reviewed-by: Rich Salz <rsalz@openssl.org>
RT: #4155, MR: #1519
When EC is disabled, and an error occurs in ssl_generate_master_secret()
or RAND_bytes(), the error path does not free rsa_decrypt.
RT#4197
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matt Caswell <matt@openssl.org>
DTLS cookie generation and verification were exact copies of each
other save the last few lines. This refactors them to avoid code
copying.
Reviewed-by: Matt Caswell <matt@openssl.org>
After the recent change to use ossl_inline, builds were failing on some
platforms due to a missing usage of "inline".
Reviewed-by: Richard Levitte <levitte@openssl.org>
Add macro ossl_inline for use in public headers where a portable inline
is required. Change existing inline to use ossl_inline
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
This adds support for SSL/TLS configuration using configuration modules.
Sets of command value pairs are store and can be replayed through an
SSL_CTX or SSL structure using SSL_CTX_config or SSL_config.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Rename to OPENSSL_mem_debug_{push,pop}.
Remove simple calls; keep only calls used in recursive functions.
Ensure we always push, to simplify so that we can always pop
Reviewed-by: Richard Levitte <levitte@openssl.org>
Just like *_clear_free routines. Previously undocumented, used
a half-dozen times within OpenSSL source.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Some URLs in the source code ended up getting mangled by indent. This fixes
it. Based on a patch supplied by Arnaud Lacombe <al@aerilon.ca>
Reviewed-by: Richard Levitte <levitte@openssl.org>
SSL_CIPHER_description() was returning "unknown" for the encryption
in the new ChaCha20/Poly1305 TLS ciphersuites.
RT#4183
Reviewed-by: Richard Levitte <levitte@openssl.org>
Rename BUF_{strdup,strlcat,strlcpy,memdup,strndup,strnlen}
to OPENSSL_{strdup,strlcat,strlcpy,memdup,strndup,strnlen}
Add #define's for the old names.
Add CRYPTO_{memdup,strndup}, called by OPENSSL_{memdup,strndup} macros.
Reviewed-by: Tim Hudson <tjh@openssl.org>
New functions EC_POINT_point2buf and EC_KEY_key2buf which encode
a point and allocate a buffer in one call.
New function EC_KEY_oct2key() which sets public key in an EC_KEY
structure from an encoded point.
Reviewed-by: Richard Levitte <levitte@openssl.org>
Solaris builds were failing during async compilation because the .o files
created from compiling the corresponding .c files held in async/arch were
ending up in the top level async directory. Consequently the link fails
because it can't find the .o files.
Thanks to Richard Levitte for pointing me in the right direction on this.
Reviewed-by: Richard Levitte <levitte@openssl.org>
The previous commit introduced a new file format for ssleay.num and
libeay.num, i.e. the introduction of a version field. Therefore the update
capability in mkdef.pl needs updating to take account of the new format.
Reviewed-by: Richard Levitte <levitte@openssl.org>
On Linux when creating the .so file we were exporting all symbols. We should
only be exporting public symbols. This commit fixes the issue. It is only
applicable to linux currently although the same technique may work for other
platforms (e.g. Solaris should work the same way).
This also adds symbol version information to our exported symbols.
Reviewed-by: Richard Levitte <levitte@openssl.org>
s_server was trying to set the ECDH curve when no-ec was defined. This also
highlighted the fact that the -no_ecdhe option to s_server is broken, and
doesn't make any sense any more (ECDHE is on by default and the only way it
can be disabled is through the cipherstring). Therefore this commit removes
the option.
Reviewed-by: Kurt Roeckx <kurt@openssl.org>
Make EVP_PKEY_copy_parameters() work if the destination has no type
(e.g. if obtained from EVP_PKEY_new()) or the underlying key is NULL.
This is useful where we want to copy the parameters from an existing
key to a new key.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
Commit 2b0180c37f attempted to do this but
only hit one of many BN_mod_exp codepaths. Fix remaining variants and add
a test for each method.
Thanks to Hanno Boeck for reporting this issue.
Reviewed-by: Rich Salz <rsalz@openssl.org>
Reviewed-by: Dr. Stephen Henson <steve@openssl.org>
When processing a public key input via "-pubin", "private" was
sometimes erroneously set, or else not set and incorrectly asserted.
Reviewed-by: Rich salz <rsalz@openssl.org>
GCM and CCM are modes of operation for block ciphers only. ChaCha20-Poly1305
operates in neither of them but it is AEAD. This change also enables future
AEAD ciphers to be available for use with DTLS.
Signed-off-by: Rich Salz <rsalz@akamai.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
This reverts commit 777f482d99.
Author credit missing. Reverting this and re-committing with
an Author line.
Reviewed-by: Matt Caswell <matt@openssl.org>
GCM and CCM are modes of operation for block ciphers only. ChaCha20-Poly1305
operates in neither of them but it is AEAD. This change also enables future
AEAD ciphers to be available for use with DTLS.
Signed-off-by: Rich Salz <rsalz@akamai.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
@@ -290,6 +244,7 @@ int dsaparam_main(int argc, char **argv)
" return NULL;\n"
" }\n"
" return(dsa);\n}\n");
OPENSSL_free(data);
}
if(!noout){
@@ -309,8 +264,6 @@ int dsaparam_main(int argc, char **argv)
assert(need_rand);
if((dsakey=DSAparams_dup(dsa))==NULL)
gotoend;
if(non_fips_allow)
dsakey->flags|=DSA_FLAG_NON_FIPS_ALLOW;
if(!DSA_generate_key(dsakey)){
ERR_print_errors(bio_err);
DSA_free(dsakey);
@@ -355,10 +308,4 @@ static int dsa_cb(int p, int n, BN_GENCB *cb)
# endif
return1;
}
#else /* !OPENSSL_NO_DSA */
# if PEDANTIC
staticvoid*dummy=&dummy;
# endif
#endif
Some files were not shown because too many files have changed in this diff
Show More
Reference in New Issue
Block a user
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
