Using SWIG_fail in the Python SWIG wrappers makes the wrapping function
destroy/free the memory which could have been dynamically allocated
before calling the wrapped function. This thus prevents possible memory
leaks in the wrappers of set*con(), set*con_raw(), security_compute_*(),
etc.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
clang does not support -aux-info option. When exception.sh is run with
CC=clang, use gcc to build selinuxswig_python_exception.i and
semanageswig_python_exception.i.
This does not solve the issue of building libselinux and libsemanage
Python wrappers on a system without gcc. However parsing the result of
"gcc -aux-info" is easier than parsing the header files so stay with
this command at least for now.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
When linking with -Wl,-no-undefined in LDFLAGS (in order to find
possible link-time errors), the Python wrapper module needs to be
linked with the right libpython.so. This library is found using
pkg-config in a new PYLIBS variable.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
When linking with -Wl,-no-undefined in LDFLAGS (in order to find
possible link-time errors), the Ruby wrapper module needs to be linked
with the libruby.so which is used by $(RUBY). Introduce a new RUBYLIBS
variable to find this library.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
This makes building libselinux and libsemanage more robust on systems
with several versions of Ruby installed: when building, only RUBY needs
to be set, without wondering about PKG_CONFIG_PATH or other environment
variables.
Using RbConfig::CONFIG["rubyarchhdrdir"] only works with Ruby >= 2.0 but
since previous Ruby versions are retired since 2015-02-23 this should
not have any impact
(https://www.ruby-lang.org/en/news/2015/02/23/support-for-ruby-1-9-3-has-ended/).
While at it, in libsemanage, use RbConfig::CONFIG["vendorarchdir"] to
install the Ruby extension, like commit 1cd80faa53 ("libselinux:
versioned ruby pkg-config and query vendorarchdir properly") did for
libselinux.
My main motivation with this patch is to make the build configuration
easier to define on Travis-CI or other continuous integration platforms.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
commit 16c123f4b1 ("libselinux:
support ANDROID_HOST=1 on Mac") broke the ability to run make
in the src subdirectory of libselinux (because OS and COMPILER
were not defined) and also caused some warning flags that could
be overridden via command-line CFLAGS to be mandatory. Fix it.
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
When compiling libselinux with CC=clang, "make pywrap" reports the
following message:
bash exception.sh > selinuxswig_python_exception.i
clang-3.9: error: no such file or directory: 'temp.aux'
awk: fatal: cannot open file `temp.aux' for reading (No such file or
directory)
This does not make the build fail as exception.sh returns an "OK"
status. Use "bash -e" with this script to make it return an error value.
In order not to keep an empty selinuxswig_python_exception.i file after
a build fails (which would make a second run of "make pywrap" incorrectly
succeed), remove the file when exception.sh fails.
As libsemanage uses the same code to build
semanageswig_python_exception.i, modify its Makefile too.
By the way, on Linux clang does not seem to currently support -aux-info
so it cannot be used to craft selinuxswig_python_exception.i.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
The Python wrapper of rpm_execcon() has several flaws:
* An invalid call like selinux.rpm_execcon() triggers a segmentation
fault.
* The size of the buffer which is allocated to copy argv and envp is
too small to hold all the values.
* This allocated memory is leaked if one argument of rpm_execon() is not
a sequence of bytes.
The Ruby wrapper has no such flaws but can not be used as it is because
it misses some glue code to convert argv and envp arguments to char
*const [] values (even though the destructor is present!).
As it is not possible to remove rpm_execcon() without changing
libselinux soname (it would be an ABI break) like b67fefd991
("libselinux: set DISABLE_RPM default to y.") tried to do, disable this
interface locally in the SWIG wrappers.
Signed-off-by: Nicolas Iooss <nicolas.iooss@m4x.org>
With the reverted commit applied, some functions were returning arrays
of bytes instead of python strings under python3 this was causing issues
with string manipulation functions like split().
Swig (checked with 3.0.7) is adding compatibility macros that take care
of the differences between python2 and python3.
This reverts commit 63df0f7ef1.
Signed-off-by: Laurent Bigonville <bigon@bigon.be>
When building with clang, multiple noreturn issues arise,
for instance:
selabel_partial_match.c:11:1: error: function 'usage' could be declared with attribute 'noreturn' [-Werror,-Wmissing-noreturn]
Fix these.
Signed-off-by: William Roberts <william.c.roberts@intel.com>
As reported by Nicolas Iooss, the clang + linux build seems
broken:
clang-3.9: warning: argument unused during compilation: '-undefined
dynamic_lookup'
/usr/bin/ld: unrecognised option: -install_name
clang-3.9: error: linker command failed with exit code 1 (use -v to
see invocation)
We already have those options set for Darwin, just drop them from the
clang side.
Signed-off-by: William Roberts <william.c.roberts@intel.com>
Loop designed for stripping leading "//" was changing
the only pointer referencing block of memory allocated
by "strdup", resulting in "free()" failure. The loop
had no effect because "realpath" is used later on.
Fixes: https://bugzilla.redhat.com/show_bug.cgi?id=1376598
Signed-off-by: vmojzis <vmojzis@redhat.com>
Add clean targets to just clean the python and ruby wrapper objects
Also clean $(SWIGRUBYSO) and $(AUDIT2WHYLOBJ) objects
Signed-off-by: Laurent Bigonville <bigon@bigon.be>
We build booleans.c with DISABLE_BOOL set on Android host
and target. Add that file to the upstream Makefile.
Signed-off-by: William Roberts <william.c.roberts@intel.com>
Some systems, like Mac, don't have stdio_ext.h. Since we're
building with DISABLE_BOOL=y on Mac, just include the
header files with the DISABLE define, and use the bare
minimum headers for DISABLE_BOOL=y.
Signed-off-by: William Roberts <william.c.roberts@intel.com>
To build on mac, first build libsepol with
no DESTDIR set.
Secondly, build libselinux with ANDROID_HOST=y
This configuration can be used to test the Android
host build on Mac.
Signed-off-by: William Roberts <william.c.roberts@intel.com>
When building on mac, one encounters this error:
sefcontext_compile.c:270:1: error: function 'usage' could be declared with attribute 'noreturn' [-Werror,-Wmissing-noreturn]
To correct this, add the attribute noreturn to the function.
Signed-off-by: William Roberts <william.c.roberts@intel.com>
When building on mac with ANDROID_HOST=y, clang complains:
sha1.c:73:33: error: cast from 'uint8_t *' (aka 'unsigned char *') to 'CHAR64LONG16 *' increases required alignment from 1 to 4 [-Werror,-Wcast-align]
CHAR64LONG16* block = (CHAR64LONG16*) workspace;
Rather then casting the bytearray to the CHAR64LONG16 union,
just create a stack workspace of type CHAR64LONG16.
This will prevent alignment issues with the data accesses.
Signed-off-by: William Roberts <william.c.roberts@intel.com>
When building on Mac, outside of the Android tree, with ANDROID_HOST=y, this warning
is observed:
label.c:102:9: warning: implicit declaration of function 'fgets_unlocked' is invalid in C99 [-Wimplicit-function-declaration]
while (fgets_unlocked(buf, sizeof(buf) - 1, cfg)) {
Fix it by using the fgets_unlocked define that was introduced for Android, just apply it for mac builds
as well.
Signed-off-by: William Roberts <william.c.roberts@intel.com>
The realpath logic in selinux_restorecon() was taken from the
Android libselinux fork. However, bionic dirname() and basename()
do not modify their argument and therefore are safe to call on a
const string. POSIX dirname() and basename() can modify their argument.
There is a GNU basename() that does not modify its argument, but not
for dirname().
For portability, create copies of the original pathname for each call
and keep them around until finished using the result.
Fixes "restorecon -r goes up the tree?" bug reported by Jason Zaman.
Reported-by: Jason Zaman <jason@perfinion.com>
Signed-off-by: Stephen Smalley <sds@tycho.nsa.gov>
Gentoo and Arch have pkg-config entries for "ruby-$(RUBYLIBVER)" but not
for "ruby". Check if that exists first then fall back to plain ruby if
it does not.
The ruby install paths were incorrect. Fedora 20 installed to
/usr/lib64/ruby/vendor_ruby/, Arch needs it to be vendor_ruby as well,
site_ruby does not work. Thanks to Nicolas Iooss for the correct way to
query for the path.
Signed-off-by: Jason Zaman <jason@perfinion.com>
Provide stubs to the public boolean API that always returns -1.
On Android, boolean symbols are needed for:
external/ltrace/sysdeps/linux-gnu/trace.c
Signed-off-by: William Roberts <william.c.roberts@intel.com>
Android label back ends are now configurable by NO_ANDROID_BACKEND,
which is set if on ANDROID_HOST != y.
Signed-off-by: Janis Danisevskis <jdanis@android.com>
We use the same lookup function for service contexts
that we use for property contexts. However, property
contexts are namespace based and only compare the
prefix. This may lead to service associations with
a wrong label.
This patch introduces a new back end for android
services with a stricter lookup function. Now the
service name must match the key of the service label
exactly.
Signed-off-by: Janis Danisevskis <jdanis@android.com>
