Richard Levitte [Sat, 21 May 2016 01:46:43 +0000 (03:46 +0200)]
Fix fmtstr for BIO_printf() et al
- 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.
Richard Levitte [Wed, 25 May 2016 09:58:19 +0000 (11:58 +0200)]
Windows makefile: handle the case with space in source directory
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...
Richard Levitte [Fri, 27 May 2016 15:18:57 +0000 (17:18 +0200)]
Allow space in PERL spec (unix only)
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:
Matt Caswell [Fri, 20 May 2016 16:49:33 +0000 (17:49 +0100)]
Avoid msys name mangling
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>
Matt Caswell [Fri, 20 May 2016 15:34:24 +0000 (16:34 +0100)]
Fix intermittent windows failures in TLSProxy tests
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>
Matt Caswell [Fri, 20 May 2016 10:53:26 +0000 (11:53 +0100)]
Fix s_client/s_server waiting for stdin on Windows
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>
Matt Caswell [Fri, 20 May 2016 10:20:22 +0000 (11:20 +0100)]
Fix some s_server issues on Windows
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>
Matt Caswell [Fri, 27 May 2016 13:59:47 +0000 (14:59 +0100)]
Silence some "may be uninitialized when used" warning
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>
Matt Caswell [Wed, 25 May 2016 14:33:15 +0000 (15:33 +0100)]
Fix implementation of "e" and "g" formats for printing floating points
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>
Matt Caswell [Thu, 26 May 2016 13:47:17 +0000 (14:47 +0100)]
Fix some suspect warnings on Windows
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.
Richard Levitte [Wed, 25 May 2016 21:35:54 +0000 (23:35 +0200)]
Handle Visual C warning about non-standard function names.
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.
Andy Polyakov [Fri, 20 May 2016 19:31:11 +0000 (21:31 +0200)]
Configure: pull 'which' back.
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>
Richard Levitte [Mon, 16 May 2016 12:54:39 +0000 (14:54 +0200)]
Communicate Configure generated header files to build files
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'
Todd Short [Tue, 24 May 2016 13:03:25 +0000 (09:03 -0400)]
Fix braces in e_aes.c: aes_init_key
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)
Richard Levitte [Mon, 23 May 2016 13:49:25 +0000 (15:49 +0200)]
Make sure tsget and c_rehash are named with .pl suffix on Windows and VMS
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.
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.
Richard Levitte [Mon, 23 May 2016 13:11:04 +0000 (15:11 +0200)]
Slight cleanup of the collection of READMEs, INSTALLs and NOTES
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.
Rich Salz [Sat, 21 May 2016 00:52:46 +0000 (20:52 -0400)]
Doc nits cleanup, round 2
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>
Matt Caswell [Thu, 19 May 2016 19:11:09 +0000 (20:11 +0100)]
Fix Windows 64 bit crashes
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>
Rich Salz [Fri, 20 May 2016 14:46:29 +0000 (10:46 -0400)]
Rename lh_xxx,sk_xxx tp OPENSSL_{LH,SK}_xxx
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>
Richard Levitte [Fri, 20 May 2016 09:12:15 +0000 (11:12 +0200)]
VMS: setbuf() only takes 32-bit pointers
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().
Matt Caswell [Thu, 12 May 2016 15:04:10 +0000 (16:04 +0100)]
Add an async io test
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.
Matt Caswell [Thu, 12 May 2016 16:18:32 +0000 (17:18 +0100)]
Ensure async IO works with new state machine
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.
David Benjamin [Sun, 6 Mar 2016 03:50:44 +0000 (22:50 -0500)]
Tighten up logic around ChangeCipherSpec.
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>
Matt Caswell [Tue, 17 May 2016 11:28:14 +0000 (12:28 +0100)]
Simplify SSL BIO buffering logic
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.
Andy Polyakov [Mon, 16 May 2016 14:44:33 +0000 (16:44 +0200)]
rand/randfile.c: remove _XOPEN_SOURCE definition.
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...