It turns out that on VMS, strerror() returns messages with added
spaces at the end.
We wouldn't had noticed if it wasn't for perl trimming those spaces
off for its own sake and thereby having test/recipes/02-test_errstr.t
fail on VMS.
The safe fix is to do the same trimming ourselves.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7701)
(cherry picked from commit 9f15e5b911)
If the private key says it can only support one specific digest, then
don't ask it to perform a different one.
Fixes: #7348
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(cherry picked from commit 2d263a4a73)
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7609)
ASN1_PKEY_CTRL_DEFAULT_MD_NID is documented to return 2 for a mandatory
digest algorithm, when the key can't support any others. That isn't true
here, so return 1 instead.
Partially fixes#7348
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(cherry picked from commit eb7eb1378c)
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7609)
Since recently, OpenSSL tarballs are produced with 'make tar' rather
than 'make dist', as the latter has turned out to be more troublesome
than useful.
The next step to look at is why we would need to configure at all to
produce a Makefile just to produce a tarball. After all, the tarball
should now only contain source files that are present even without
configuring.
Furthermore, the current method for producing tarballs is a bit
complex, and can be greatly simplified with the right tools. Since we
have everything versioned with git, we might as well use the tool that
comes with it.
Added: util/mktar.sh, a simple script to produce OpenSSL tarballs. It
takes the options --name to modify the prefix of the distribution, and
--tarfile tp modify the tarball file name specifically.
This also adds a few entries in .gitattributes to specify files that
should never end up in a distribution tarball.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7692)
(cherry picked from commit 8c209eeef4)
This ensures we collected them properly and and as completely as can
be tested safely.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7681)
(cherry picked from commit 4b801fdcf4)
We stored copies of the system error texts in a fixed line size array,
which is a huge waste. Instead, use a static memory pool and pack all
the string in there. The wasted space at the end, if any, gives us
some leeway for longer strings than we have measured so far.
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7681)
(cherry picked from commit 2c5b6bbb67)
Blinding is performed more efficiently and securely if MONT_CTX for public
modulus is available by the time blinding parameter are instantiated. So
make sure it's the case.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7586)
(cherry picked from commit 2cc3f68cde)
Fixes#7675
On macOS, if you call `connect()` on a UDP socket you cannot then
call `sendto()` with a destination, otherwise it fails with Err#56
('socket is already connected').
By calling `BIO_ctrl_set_connected()` on the wbio we can tell it
that the socket has been connected and make it call `send()` rather
than `sendto()`.
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/7676)
(cherry picked from commit b92678f4e9)
CLA: trivial
In Travis CI, add a Python linting step that runs flake8 tests in Travis CI
to find syntax errors and undefined names. (http://flake8.pycqa.org)
__E901,E999,F821,F822,F823__ are the "_showstopper_" flake8 issues that can halt
the runtime with a SyntaxError, NameError, etc. Most other flake8 issues are
merely "style violations" -- useful for readability but they do not effect
runtime safety.
* F821: undefined name `name`
* F822: undefined name `name` in `__all__`
* F823: local variable name referenced before assignment
* E901: SyntaxError or IndentationError
* E999: SyntaxError -- failed to compile a file into an Abstract Syntax Tree
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/7410)
(cherry picked from commit 2a6f57bc09)
Without this precaution, we end up having directory targets depend on
shlib object files for which there are no rules.
Reviewed-by: Tim Hudson <tjh@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(Merged from https://github.com/openssl/openssl/pull/7645)
Mingw and Cygwin builds install the DLLs in the application directory,
not the library directory, so ensure that one is created for them when
installing the DLLs.
Fixes#7653
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7654)
(cherry picked from commit 9694ebf753)
dsa_builtin_paramgen2 expects the L parameter to be greater than N,
otherwise the generation will get stuck in an infinite loop.
Reviewed-by: Bernd Edlinger <bernd.edlinger@hotmail.de>
Reviewed-by: Paul Dale <paul.dale@oracle.com>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Matthias St. Pierre <Matthias.St.Pierre@ncp-e.com>
(cherry picked from commit 3afd38b277)
(Merged from https://github.com/openssl/openssl/pull/7493)
When libssl and libcrypto are compiled on Linux with "-rpath", but
not "--enable-new-dtags", the RPATH takes precedence over
LD_LIBRARY_PATH, and we end up running with the wrong libraries.
This is resolved by using full (or at least relative, rather than
just the filename to be found on LD_LIBRARY_PATH) paths to the
shared objects.
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7626)
(cherry picked from commit 1828939974)
SSL_get_signature_nid() -- local signature algorithm
SSL_get_signature_type_nid() -- local signature algorithm key type
SSL_get_peer_tmp_key() -- Peer key-exchange public key
SSL_get_tmp_key -- local key exchange public key
Aliased pre-existing SSL_get_server_tmp_key(), which was formerly
just for clients, to SSL_get_peer_tmp_key(). Changed internal
calls to use the new name.
Reviewed-by: Matt Caswell <matt@openssl.org>
SSL(_CTX)?_set_client_CA_list() was a server side only function in 1.1.0.
If it was called on the client side then it was ignored. In 1.1.1 it now
makes sense to have a CA list defined for both client and server (the
client now sends it the the TLSv1.3 certificate_authorities extension).
Unfortunately some applications were using the same SSL_CTX for both
clients and servers and this resulted in some client ClientHellos being
excessively large due to the number of certificate authorities being sent.
This commit seperates out the CA list updated by
SSL(_CTX)?_set_client_CA_list() and the more generic
SSL(_CTX)?_set0_CA_list(). This means that SSL(_CTX)?_set_client_CA_list()
still has no effect on the client side. If both CA lists are set then
SSL(_CTX)?_set_client_CA_list() takes priority.
Fixes#7411
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7503)
(cherry picked from commit 9873297900)
Just refering to a hash table element as an array reference will
automatically create that element. Avoid that by defaulting to
a separate empty array reference.
Fixes#7543
Reviewed-by: Tim Hudson <tjh@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7544)
(cherry picked from commit 3bed01a090)
TLSv1.3 is more restrictive about the curve used. There must be a matching
sig alg defined for that curve. Therefore if we are using some other curve
in our certificate then we should not negotiate TLSv1.3.
Fixes#7435
Reviewed-by: Viktor Dukhovni <viktor@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7442)
(cherry picked from commit de4dc59802)
Setting the SipHash hash size and setting its key is done with two
independent functions... and yet, the internals depend on both.
Unfortunately, the function to change the size wasn't adapted for the
possibility that the key was set first, with a different hash size.
This changes the hash setting function to fix the internal values
(which is easy, fortunately) according to the hash size.
evpmac.txt value for digestsize:8 is also corrected.
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7613)
(cherry picked from commit 425036130d)
Reviewed-by: Paul Yang <yang.yang@baishancloud.com>
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
Reviewed-by: Matt Caswell <matt@openssl.org>
(Merged from https://github.com/openssl/openssl/pull/7427)
(cherry picked from commit d896b79b09)
Reviewed-by: Matt Caswell <matt@openssl.org>
Reviewed-by: Richard Levitte <levitte@openssl.org>
Reviewed-by: Nicola Tuveri <nic.tuv@gmail.com>
(Merged from https://github.com/openssl/openssl/pull/7599)
(cherry picked from commit dd41956d80)
We only had the main 'install' target depend on 'all'. This changes
the dependencies so targets like install_dev, install_runtime_libs,
install_engines and install_programs depend on build targets that are
correspond to them more specifically. This increases the parallel
possibilities.
Fixes#7466
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7583)
(cherry picked from commit e8d01a6087)
When trying 'make -j{n} install', you may occasionally run into
trouble because to sub-targets (install_dev and install_runtime) try
to install the same shared libraries. That makes parallel install
difficult.
This is solved by dividing install_runtime into two parts, one for
libraries and one for programs, and have install_dev depend on
install_runtime_libs instead of installing the shared runtime
libraries itself.
Fixes#7466
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7583)
(cherry picked from commit c1123d9f7e)
Commit c7504aeb64 (pr #6432) fixed a regression for applications in
chroot environments, which compensated the fact that the new OpenSSL CSPRNG
(based on the NIST DRBG) now reseeds periodically, which the previous
one didn't. Now the reseeding could fail in the chroot environment if the
DEVRANDOM devices were not present anymore and no other entropy source
(e.g. getrandom()) was available.
The solution was to keep the file handles for the DEVRANDOM devices open
by default. In fact, the fix did more than this, it opened the DEVRANDOM
devices early and unconditionally in rand_pool_init(), which had the
unwanted side effect that the devices were opened (and kept open) even
in cases when they were not used at all, for example when the getrandom()
system call was available. Due to a bug (issue #7419) this even happened
when the feature was disabled by the application.
This commit removes the unconditional opening of all DEVRANDOM devices.
They will now only be opened (and kept open) on first use. In particular,
if getrandom() is available, the handles will not be opened unnecessarily.
This change does not introduce a regression for applications compiled for
libcrypto 1.1.0, because the SSLEAY RNG also seeds on first use. So in the
above constellation the CSPRNG will only be properly seeded if it is happens
before the forking and chrooting.
Fixes#7419
Reviewed-by: Paul Dale <paul.dale@oracle.com>
(Merged from https://github.com/openssl/openssl/pull/7437)
(cherry picked from commit 8cfc19716c)
use_ecc() was always returning 1 because there are default (TLSv1.3)
ciphersuites that use ECC - even if those ciphersuites are disabled by
other options.
Fixes#7471
Reviewed-by: Kurt Roeckx <kurt@roeckx.be>
(Merged from https://github.com/openssl/openssl/pull/7479)
(cherry picked from commit 589b6227a8)