David Woodhouse [Mon, 22 Feb 2016 16:29:12 +0000 (16:29 +0000)]
RT4335: Fix UEFI build of OBJ_NAME_new_index()
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>
Matt Caswell [Fri, 26 Feb 2016 14:10:17 +0000 (14:10 +0000)]
Fix a mkdef.pl warning
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>
Richard Levitte [Wed, 24 Feb 2016 23:17:59 +0000 (00:17 +0100)]
Configure - neater looking add() and add_before()
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.
Matt Caswell [Thu, 25 Feb 2016 13:09:46 +0000 (13:09 +0000)]
Fix memory issues in BIO_*printf functions
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.
Emilia Kasper [Wed, 24 Feb 2016 11:59:59 +0000 (12:59 +0100)]
CVE-2016-0798: avoid memory leak in SRP
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.
Richard Levitte [Thu, 25 Feb 2016 13:36:30 +0000 (14:36 +0100)]
Make it possible to build even if dependency files can't be generated
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.
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.
David Woodhouse [Sat, 20 Feb 2016 14:40:48 +0000 (14:40 +0000)]
RT4339: Fix handling of <internal/bn_conf.h>
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>
Richard Levitte [Mon, 22 Feb 2016 20:02:42 +0000 (21:02 +0100)]
Remove all -march= from configs
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.
David Woodhouse [Mon, 22 Feb 2016 16:54:57 +0000 (16:54 +0000)]
RT4334: Check UEFI before __STDC_VERSION__ for <inttypes.h>
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>
Richard Levitte [Mon, 22 Feb 2016 01:09:11 +0000 (02:09 +0100)]
Introduce the "pic" / "no-pic" config option
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.
Richard Levitte [Mon, 22 Feb 2016 01:06:05 +0000 (02:06 +0100)]
Use $disabled{"dynamic-engine"} internally
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.
Richard Levitte [Mon, 22 Feb 2016 13:26:40 +0000 (14:26 +0100)]
Fix incorrect SO name on GNU platforms
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.
Richard Levitte [Sun, 21 Feb 2016 21:43:29 +0000 (22:43 +0100)]
Don't use 'parent' in util/dofile.pl
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.
Richard Levitte [Sat, 20 Feb 2016 16:29:23 +0000 (17:29 +0100)]
Avoid GNU make re-exec when adding dependencies to Makefile
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.
Richard Levitte [Fri, 19 Feb 2016 21:08:37 +0000 (22:08 +0100)]
Build dynamic engines even if configured "no-shared"
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.
Richard Levitte [Fri, 19 Feb 2016 21:02:41 +0000 (22:02 +0100)]
Always build library object files with shared library cflags
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).
Richard Levitte [Fri, 19 Feb 2016 23:21:04 +0000 (00:21 +0100)]
Make crypto/buildinf.h depend on configdata.pm rather than Makefile
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>
Emilia Kasper [Fri, 19 Feb 2016 16:24:44 +0000 (17:24 +0100)]
TLS: reject duplicate extensions
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.
Emilia Kasper [Fri, 19 Feb 2016 15:23:08 +0000 (16:23 +0100)]
getaddrinfo: zero the hints structure
This silences the memory sanitizer. All fields were already correctly
initialized but the struct padding wasn't, causing an uninitialized read
warning.
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.
Richard Levitte [Mon, 15 Feb 2016 17:39:49 +0000 (18:39 +0100)]
Remake the installation of shared libraries in unix-Makefile.tmpl
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.
Richard Levitte [Mon, 15 Feb 2016 17:19:49 +0000 (18:19 +0100)]
Simplify the generation of ld scripts for Linux and Solaris
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.