When the linker uses --no-undefined-version either specified by the user
or as the default behavior (such as with newer clang >= 16 releases),
a missing symbol definition will cause a linker error if that symbol is
listed in the version script.
The check uses printf, so it needs to include <stdio.h> for
compilers which do not support implicit function declarations.
(They were removed from C99.)
Closes: !23
Signed-off-by: Guillem Jover <guillem@hadrons.org>
This function cannot be easily and (more importantly) correctly ported
without cooperation from the libc stdio layer. We already document that
users should be prepared to have the function not available on some
platforms and that they should ideally switch their code to other
more portable and better interfaces.
Instead of making the build fail, and requiring porters to add
exceptions for something that most probably cannot be ported correctly
anyway, simply print a warning and let it build. This will not be a
regression because on those systems libbsd would have never been built
before.
Prompted-by: Jens Finkhaeuser <jens@finkhaeuser.de>
We should not assume that something will implicitly check for this tool,
as we need it ourselves, and this is an internal implementation detail
of right now libtool.
Fixes: commit f11ab67223
When using the recent dlsym() based wrapper, we are not requiring any
symbol from libmd, as we resolve those dynamically at run-time. We were
ending up linking against libmd because in another part of the code we
require (depending on the architecture) the SHA512 functions for the
getentropy() local implementation. But that function might be provided
by the system libc on some systems, which means we end up not linking
against libmd at all.
To solve this we go back to the previous simpler solution of linking
directly, which had the main drawback of then making programs fail to
link when not specifying -lmd (on platforms that need it). And then
switch the .so link point from a symlink to a linker script, so that we
can inject the -lmd library as-needed. This is similar to what glibc is
doing.
Fixes: commit 31f034e386
To be able to rework the md5 deprecation logic, we need to detangle when
we depend on libmd due to requiring MD5 functions, which might be
otherwise provided by libc, or when we require SHA functions for the
internal getentropy() implementation.
Detect as many warnings as possible during configure and enable them
if the user did not supply any, so that any such problem can be spotted
and fixed.
Switch from the previous versioned symbol implementation which required
users to also link against the message digest provider explicitly, or
they would fail to find the symbols, to an implementation that loads
the symbols from the linked library providing the functions using
dlsym(), thus preserving backwards compatibility.
Using various variables means we have to keep these in sync in various
places. Just use a single variable that we can use anywhere where this
is needed.
This splits the implementation responsibilities, and reduces embedded
code copies, which was one of the driving points with this project to
start with, so it's nice to give a good example.
Extend the host OS checks to define an OS_WINDOWS automake conditional if
the host is MinGW-like. This will be useful for future Windows-specific
build tweaks.
[guillem@hadrons.org:
- Rename WINDOWS conditional to OS_WINDOWS. ]
Signed-off-by: Guillem Jover <guillem@hadrons.org>
This is a glibc-specific symbol that has no public declaration. But is
being used by the OpenBSD and this implementation as a hack to avoid
having to link against the pthread library. This interface is at least
included in LSB 5.0 [L], and using pthread_atfork() is otherwise
problematic anyway [P].
[L] <https://refspecs.linuxfoundation.org/LSB_5.0.0/LSB-Core-generic/LSB-Core-generic/baselib---register-atfork.html>
[P] <http://austingroupbugs.net/view.php?id=851>
One problem is that we were using it whenever __GLIBC__ is defined,
which is supposed to be defined only on an actual glibc, but uClibc
defines that macro, but it does not provide the symbol on its noMMU
variant.
We add a new configure check that will try to link a program that uses
that symbol to make sure it is present.
Closes: !2
Reported-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Because we were assigning to another unused variable, when building the
check with optimizations enabled, which is the default when using gcc
as the compiler, the variable was being discarded. Instead pass it to
printf() so that it cannot do so.
In older glibc versions (< 2.17) clock_gettime() is in librt. Add a
check for this to avoid build breakage for programs/libraries that
use libbsd on such systems.
Based-on-patch-by: Gustavo Zacarias <gustavo@zacarias.com.ar>
Signed-off-by: Guillem Jover <guillem@hadrons.org>
