Richard Levitte [Sun, 14 Feb 2016 10:16:37 +0000 (11:16 +0100)]
Make sure a socklen_t can compare with a sizeof() result
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.
Richard Levitte [Sun, 14 Feb 2016 06:10:38 +0000 (07:10 +0100)]
Prefer IO::Socket::INET6 over IO::Socket::IP
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.
Richard Levitte [Sat, 13 Feb 2016 18:15:52 +0000 (19:15 +0100)]
Pass $(CC) to perlasm scripts via the environment
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.
Richard Levitte [Sat, 13 Feb 2016 12:52:24 +0000 (13:52 +0100)]
The unified build may delete installed manual files
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.
Benjamin Kaduk [Wed, 10 Feb 2016 02:29:21 +0000 (20:29 -0600)]
GH650: Minor tidying around the ocsp app
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>
Richard Levitte [Sat, 13 Feb 2016 12:02:35 +0000 (13:02 +0100)]
Rethink the method to place user cflags last
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.
Richard Levitte [Sat, 13 Feb 2016 10:49:56 +0000 (11:49 +0100)]
Remove 00-test_checkexes.t, as it has lost its relevance
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.
Richard Levitte [Fri, 12 Feb 2016 13:05:06 +0000 (14:05 +0100)]
Unified build - fix make depend
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'.
Richard Levitte [Fri, 12 Feb 2016 20:14:03 +0000 (21:14 +0100)]
Rename INSTALL_PREFIX to DESTDIR, remove option --install_prefix
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).
Richard Levitte [Fri, 12 Feb 2016 17:16:23 +0000 (18:16 +0100)]
Rethink logging of test recipes
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
Richard Levitte [Fri, 12 Feb 2016 18:44:55 +0000 (19:44 +0100)]
Better workaround for VMS getnameinfo() bug
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.
Richard Levitte [Fri, 12 Feb 2016 03:23:15 +0000 (04:23 +0100)]
Adjust transfer::Text::Template.pm for alternate directory name
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.
Richard Levitte [Thu, 11 Feb 2016 23:34:40 +0000 (00:34 +0100)]
Generate progs.h from a bunch of files instead of internal knowledge
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.
Richard Levitte [Thu, 11 Feb 2016 20:47:30 +0000 (21:47 +0100)]
Perl's chop / chomp considered bad, use a regexp instead
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.
Andy Polyakov [Wed, 3 Feb 2016 17:21:00 +0000 (18:21 +0100)]
util/mk1mf.pl: use LINK_CMD instead of LINK variable.
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>
Viktor Dukhovni [Thu, 11 Feb 2016 18:44:53 +0000 (13:44 -0500)]
Fix MacOS/X build warnings
Commit 7823d792d0cad3b44ad5389a8d3381becefe7f44 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.
Richard Levitte [Thu, 11 Feb 2016 12:10:11 +0000 (13:10 +0100)]
Make shared library targets more consistent
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.
Richard Levitte [Thu, 11 Feb 2016 14:22:27 +0000 (15:22 +0100)]
Unified build: Keep track of generated header files
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.
Richard Levitte [Wed, 10 Feb 2016 21:33:44 +0000 (22:33 +0100)]
Rework BIO_ADDRINFO_protocol() to return correct values
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.
Matt Caswell [Wed, 10 Feb 2016 19:41:27 +0000 (19:41 +0000)]
Fix errstr error code parsing
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>
Richard Levitte [Wed, 10 Feb 2016 22:27:28 +0000 (23:27 +0100)]
After auto init, check that the deprecated functions exist before using
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.
Richard Levitte [Wed, 10 Feb 2016 19:23:40 +0000 (20:23 +0100)]
Don't assert protocol equality
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.