Richard Levitte [Mon, 18 May 2015 00:57:24 +0000 (02:57 +0200)]
Refactor config - split read_config into read_config and resolve_config
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.
Richard Levitte [Mon, 18 May 2015 00:46:21 +0000 (02:46 +0200)]
Refactor config - a small cosmetic touchup of Configure
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.
Rich Salz [Sun, 17 Jan 2016 21:53:53 +0000 (16:53 -0500)]
Consolidate "make update"
Rename 'update' to 'generate'. Rather than recurse, just explicitly
call the three generate targets directly. Reviewed-by: Richard Levitte <levitte@openssl.org>
Matt Caswell [Mon, 7 Dec 2015 16:50:38 +0000 (16:50 +0000)]
Handle SSL_shutdown while in init more appropriately
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!
Richard Levitte [Tue, 19 Jan 2016 19:35:41 +0000 (20:35 +0100)]
Fix BSD -rpath parameter
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.
Viktor Dukhovni [Mon, 18 Jan 2016 18:10:21 +0000 (13:10 -0500)]
Support disabling any or all TLS or DTLS versions
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>
Viktor Dukhovni [Sun, 17 Jan 2016 07:33:14 +0000 (02:33 -0500)]
Drop cached certificate signature validity flag
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>
Richard Levitte [Sun, 17 Jan 2016 17:03:04 +0000 (18:03 +0100)]
Add some extra Cygwin targets as aliases for Cygwin-x86
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.
Corinna Vinschen [Sun, 17 Jan 2016 15:42:38 +0000 (16:42 +0100)]
Fix configuration system to support different architectures on Cygwin.
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.
Corinna Vinschen [Sat, 16 Jan 2016 18:36:15 +0000 (19:36 +0100)]
Don't strip object files on Cygwin
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.
Rich Salz [Sun, 17 Jan 2016 15:34:26 +0000 (10:34 -0500)]
Add some accessors.
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>
Viktor Dukhovni [Sat, 16 Jan 2016 18:25:17 +0000 (13:25 -0500)]
Better invalid SNI name error handling
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.
Matt Caswell [Thu, 14 Jan 2016 20:19:23 +0000 (20:19 +0000)]
Update Windows installation instructions
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>
Benjamin Kaduk [Wed, 13 Jan 2016 00:02:16 +0000 (18:02 -0600)]
Add a no-egd option to disable EGD-related code
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>
Richard Levitte [Thu, 14 Jan 2016 16:03:58 +0000 (17:03 +0100)]
Relax the requirements for a debug build
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.
Rich Salz [Thu, 14 Jan 2016 04:11:01 +0000 (23:11 -0500)]
RT4232: Extra space in help message.
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>
Richard Levitte [Thu, 14 Jan 2016 13:08:49 +0000 (14:08 +0100)]
VMS open() doesn't take O_BINARY, but takes a context description
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.
Richard Levitte [Thu, 14 Jan 2016 12:15:45 +0000 (13:15 +0100)]
Do not use redirection on binary files
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.
Richard Levitte [Thu, 14 Jan 2016 11:18:30 +0000 (12:18 +0100)]
Fall back to c_rehash if 'openssl rehash' fails
'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.
Richard Levitte [Thu, 14 Jan 2016 00:46:23 +0000 (01:46 +0100)]
Add a directry spec for mcr if there is none
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.
Richard Levitte [Wed, 13 Jan 2016 19:39:58 +0000 (20:39 +0100)]
Quick fix of debugging option for mk1mf.pl.
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.
Rich Salz [Tue, 5 Jan 2016 18:06:03 +0000 (13:06 -0500)]
Add CRYPTO_EX_DATA; remove EC_EXTRA_DATA
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>
Viktor Dukhovni [Wed, 13 Jan 2016 17:12:17 +0000 (12:12 -0500)]
Fix DES_LONG breakage
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.
Richard Levitte [Wed, 13 Jan 2016 16:00:22 +0000 (17:00 +0100)]
Have the VMS exit code follow POSIX conventions
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:
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.
Richard Levitte [Wed, 13 Jan 2016 16:27:06 +0000 (17:27 +0100)]
Pipes on VMS do not work well with binary data, use an intermediary file
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.
Richard Levitte [Wed, 13 Jan 2016 14:16:41 +0000 (15:16 +0100)]
Fix test/recipes/80-test_ca.t to work on VMS
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).