Ross Burton [Mon, 5 Jun 2017 16:30:42 +0000 (17:30 +0100)]
oeqa/selftest: lock down Meson git revision for reliability
The test_recipetool_create_github test fetches HEAD of the repository so
upstream changes can (and do) break the test. Avoid these problems by passing
the rev= argument in the URL to lock the checkout to the same version that is
fetched in the github_tarball test.
Also pass the commands to runCmd() as a list instead of a string, the semicolon
in the URL needs more quotes if the shell is involved and passing a list
bypasses the shell entirely.
Signed-off-by: Ross Burton <ross.burton@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
kernel.bbclass: fix KERNEL_IMAGETYPE(S) for Image.gz
KERNEL_IMAGETYPES lists all the kernel images that we want to build. in cb17b6c2a7 (kernel.bbclass: support kernel image type of vmlinux.gz), some logic
was added to support vmlinux.gz which is not a target built by kernel
makefiles (only vmlinux). It is clear that the goal of this logic is only to
support vmlinux.gz and not others compressed format (such as Image.gz) which are
valid target for kernel makefiles.
For Image.gz we should rely on the kernel makefiles and not do the compression
in kernel class.
This patch updates the logic used to filter out non supported kernel target from
KERNEL_IMAGETYPES, and make vmlinux.gz a 'special case', instead of *.gz. If
more special cases are needed in the future, we could add them in a similar way.
This patch should be a no-op for anyone using vmlinux or vmlinux.gz, and on top
of that it is fixing the build for Image.gz which was not working until now.
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Jussi Kukkonen [Thu, 1 Jun 2017 10:36:46 +0000 (13:36 +0300)]
vte: Upgrade 0.46.1 -> 0.48.3
Depend on gperf-native: The tarball used to include files generated
with gperf and this is no longer the case.
Use GIR_EXTRA_LIBS_PATH to fix introspection generation.
Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
* Costmetic changes in license.html file required new MD5 hash
* Add PACKAGECONFIG option for libunwind support (new feature, disable by default)
* Rebased 0001-Use-wayland-scanner-in-the-path.patch
* Added zlib in DEPENDS, otherwise mesa no longer builds.
* Added 0001-util-rand_xor-add-missing-include-statements.patch, sent upstream
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org> Signed-off-by: Ross Burton <ross.burton@intel.com>
Maxin B. John [Wed, 31 May 2017 11:39:35 +0000 (14:39 +0300)]
libtasn1: upgrade 4.10 -> 4.12
Noteworthy changes:
1. Introduced the ASN1_TIME_ENCODING_ERROR error code to indicate an invalid
encoding in the DER time fields.
2. Introduced flag ASN1_DECODE_FLAG_ALLOW_INCORRECT_TIME. This flag allows
decoding errors in time fields even when in strict DER mode.
3. Added safety check in asn1_find_node(). That prevents a crash when a very
long variable name is provided by the developer.
Signed-off-by: Maxin B. John <maxin.john@intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Maxin B. John [Wed, 31 May 2017 11:39:34 +0000 (14:39 +0300)]
sqlite3: upgrade 3.18.0 -> 3.19.2
Bug fixes:
1. Fix a problem in REPLACE that can result in a corrupt database
containing two or more rows with the same rowid.
2. Fix a problem in PRAGMA integrity_check that was causing a subsequent
VACUUM to behave suboptimally.
3. Fix the PRAGMA foreign_key_check command so that it works correctly with
foreign keys on WITHOUT ROWID tables.
4. Disallow leading zeros in numeric constants in JSON.
5. Disallow control characters inside of strings in JSON.
6. Limit the depth of recursion for JSON objects and arrays in order to
avoid excess stack usage in the recursive descent parser.
7. Fix more bugs in the LEFT JOIN flattening optimization.
Signed-off-by: Maxin B. John <maxin.john@intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
start_getty: Over added SERIAL_CONSOLE cause error in userspace log
Error log will be logged into /var/log/message.
Added in more condition checking on the script. Check
/proc/tty/drivers and /proc/tty/driver/*
file system to retrieve active targeted serial.
Only establish getty with active serial in runtime.
Mikko Rapeli [Thu, 1 Jun 2017 15:52:59 +0000 (18:52 +0300)]
scripts: Fix return value checks from subprocess.call()'s
Python function subprocess.call() returns the return value of the
executed process. If return values are not checked, errors may
go unnoticed and bad things can happen.
Change all callers of subprocess.call() which do not check for
the return value to use subprocess.check_call() which raises
CalledProcessError if the subprocess returns with non-zero value.
Ed Bartosh [Wed, 17 May 2017 12:49:12 +0000 (12:49 +0000)]
isoimage-isohybrid: don't use TRANSLATED_TARGET_ARCH
isoimage-isohybrid plugin fails with this error when
trying to find initrd image:
initrd = glob.glob('%s/%s*%s.%s' % (initrd_dir, image_name, target_arch, image_type))[0]
IndexError: list index out of range
as it uses TRANSLATED_TARGET_ARCH value as part of the image.
This approach stopped to work due to changes in oe core code.
initrd file name is made using MACHINE variable. wic can't get value
of this variable as it's not included into bitbake -e output.
Used basename of deploy dir as MACHINE value to fix the breakage.
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
insane.bbclass: fix override handling in RDEPENDS QA
The package_qa_check_rdepends() in insane.bbclass has
incorrectly replace its localdata OVERRIDES value with
the package name. Fixing it by appending the package name
to the existing OVERRIDES value. This resolves RDEPENDS QA
error when setting PACKAGECONFIG using a pn- override at
local.conf.
[YOCTO #11374]
Signed-off-by: Gan, Yau Wai <yau.wai.gan@intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Daniel Lublin [Tue, 30 May 2017 11:17:59 +0000 (13:17 +0200)]
devtool: deploy-target: Let script continue even if rm fails
The remote script is run with `set -e`, so doing rm without -f on a
$file that is already gone will exit the whole script, failing any
redeployment. Assume a use case where packages sometimes produces
certain test binaries stored on volatile media (tmpfs), and where the
system is occasionally rebooted.
Signed-off-by: Daniel Lublin <daniel@lublin.se> Signed-off-by: Ross Burton <ross.burton@intel.com>
Ola x Nilsson [Tue, 30 May 2017 10:16:31 +0000 (12:16 +0200)]
devtool: Compare abspath of both B and S
Either both or none of the paths must be passed through
os.path.abspath or things like 'A//B', 'A/./B/', and 'A/B/' in S will
cause unintentional mismatches even when B = "${S}".
Using os.path.abspath for both seems more likely to be correct as that
will also handle the case where ${B} != ${S} but the abspaths are
equal.
Signed-off-by: Ola x Nilsson <olani@axis.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Paul Eggleton [Thu, 1 Jun 2017 01:42:25 +0000 (13:42 +1200)]
classes/buildhistory: make a single commit per build
Way back in OE-Core commit fba198ac7efe476a25c5761878ef2fcee97bf9f1 in
2012 we split committing to the repository, making a commit per
top-level directory. However, as we add more information it becomes
harder to see which commits belong to which build. Switch back to a
single commit per build to keep the history tidier.
To address the original concern, if you do want to see just the changes
for a particular subdirectory, git can filter that for you - just
specify that subdirectory as the last parameter on the git show / git
diff command line and that's all you will see.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
oeqa/core/context: Omit docstring output on XMLResult
By default, the xml runner class prints out the docstring for every
unit test but it order to keep the same format as the standard
runner, avoid docstring output setting descriptions to False.
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
oeqa/core/context: Use a default iterable modules object
The OETestLoader.modules must be an iterable object, otherwise
checking presence of specific modules with the 'in' operator fails
with the following command/error:
% oe-test core
.. - core - INFO - Running bitbake -p
Traceback (most recent call last):
File "..poky2/scripts/oe-test", line 108, in <module>
ret = main()
File "..poky2/scripts/oe-test", line 93, in main
results = args.func(logger, args)
File "..poky2/meta/lib/oeqa/core/context.py", line 235, in run
self.tc.loadTests(self.module_paths, **self.tc_kwargs['load'])
File "..poky2/meta/lib/oeqa/core/context.py", line 53, in loadTests
self.suites = self.loader.discover()
File "..poky2/meta/lib/oeqa/core/loader.py", line 204, in discover
pattern='*.py', top_level_dir=path)
File "/usr/lib64/python3.4/unittest/loader.py", line 275, in discover
tests = list(self._find_tests(start_dir, pattern))
File "/usr/lib64/python3.4/unittest/loader.py", line 339, in _find_tests
yield self.loadTestsFromModule(module)
File "..poky2/meta/lib/oeqa/core/loader.py", line 266, in loadTestsFromModule
and module.__name__ in self.modules \
TypeError: argument of type 'NoneType' is not iterable
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Aníbal Limón [Fri, 26 May 2017 20:37:43 +0000 (15:37 -0500)]
oeqa/sdk: Enable usage of OEQA thread mode
oeqa/sdk/context.py - Use OETestContextThreaded.
classes/testsdk.bbclass - Enable bb event thread mode to avoid
corrupt the PIPE when multiple threads writes.
[YOCTO #11450]
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Aníbal Limón [Fri, 26 May 2017 20:37:41 +0000 (15:37 -0500)]
oeqa/core/tests: Add tests of OEQA Threaded mode
Add needed tests to validate the OEQA Threaded mode, the remaining
parts are tested by the OEQA without Threaded mode.
- test_loader.py: Add a test to validate rules when creating the
list of test suites.
- test_decorators.py: Add oetimeout test because the threaded mode
uses Timer instead of signal.
[YOCTO #11450]
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Aníbal Limón [Fri, 26 May 2017 20:37:40 +0000 (15:37 -0500)]
oeqa/core/decorator/oetimeout: Add support for OEQA threaded mode
In python signals are only allowed to delivery into the main thread,
to support the threading mode test if the runner is threaded and
use threading.Timer instead.
There are some considerations like SIGALRM interrupts the execution
after N seconds but the Timer only starts a Thread to notice the
timeout and the exception will be raised when the test run ends.
[YOCTO #11450]
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The OETestResultThreadedInternal extends OETestResult to stores
results by Thread.
The OETestResultThreaded is a simple class that provides the
implementation of interfaces needed by outside like wasSuccesful,
stop, logSummary, logDetails.
[YOCTO #11450]
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Aníbal Limón [Fri, 26 May 2017 20:37:34 +0000 (15:37 -0500)]
oeqa/core/threaded: Add OEStreamLoggerThreaded class
The OEStreamLoggerThreaded overrides OEStreamLogger to redirect
the PyUnit output to a logger.
Instead of log every line when comes the OEStreamLoggerThreaded
will buffer the PyUnit output and write everything at end of every
suite execution to don't have mixed suite outputs.
[YOCTO #11450]
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Aníbal Limón [Fri, 26 May 2017 20:37:33 +0000 (15:37 -0500)]
oeqa/core/threaded: Add new module with OETestLoaderThreaded
The threded module will implement Thread support into the OEQA
framework.
The OETestLoaderThreaded overrides discover to return a list of
suites every suite will be run by a separate Thread.
Some design considerations are:
- All the tests of a certain module needs to be run at one thread
because unittest framework needs to keep track of setUp{Module,
Class,} and tearDown{Module, Class,}.
- Tests that depends on other needs to be run at the same thread
because OEQA framework look at the status of dependant test to
decide if skip or not, this constraint can be change in the future
but a sync mechanishm is needed to implement between the Threads.
[YOCTO #11450]
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Aníbal Limón [Fri, 26 May 2017 20:37:30 +0000 (15:37 -0500)]
oeqa/core: Don't expose OEStreamLogger in OETestContext
The OEStreamLogger class is used for redirect PyUnit output
to a certain logger so there is not need to expose at level
of OETestContext because only OETestRunner needs to know.
[YOCTO #11450]
Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Empty descriptions lines are set with a space following by a dot and
the multiline ones require a leading space. Also, for non-empty lines,
there is no need for formating with textwrap, so remove the logic for
the latter. The documentation for multiline description was taken from [1].
Daniel Díaz [Thu, 18 May 2017 18:22:37 +0000 (13:22 -0500)]
piglit: add patches for unbuildable surfaceless Mesa test
Some EGL implementations do not actually ship all Khronos-
extensions. As it turns out, the Mali 450 driver does not
include any of the following symbols, used by the
egl_mesa_platform_surfaceless.c spec test:
* eglGetPlatformDisplay
* eglCreatePlatformPixmapSurface
* eglCreatePlatformWindowSurface
The Right Thing To Do was to obtain the implementation of
these functions (via eglGetProcAddress), as is provided
by their EXT counterparts. These are guaranteed to exist
since they are required by EGL_EXT_platform_base.
When linking against Mali 450 r6, errors like the following
can be seen:
../../../../lib/libpiglitutil_gl.so.0: undefined reference to `gbm_bo_unmap'
../../../../lib/libpiglitutil_gl.so.0: undefined reference to `gbm_bo_map'
collect2: error: ld returned 1 exit status
make[2]: *** [bin/point-sprite] Error 1
This is due to gbm_bo_map() and gbm_bo_unmap() being recently
added but not yet implemented by all graphics drivers.
Instead of relying on GBM's version, actually try to link
against those symbols.
Daniel Díaz [Thu, 18 May 2017 18:22:35 +0000 (13:22 -0500)]
piglit: depend on virtual/egl
While building for Hikey using Mali 450 driver (r6p0), an
error like the following appears while linking:
[ 1%] Linking C shared library ../../../../lib/libpiglitutil.so
[...]
[...]/aarch64-linaro-linux/gcc/aarch64-linaro-linux/6.3.1/ld: cannot find -lEGL
collect2: error: ld returned 1 exit status
make[2]: *** [lib/libpiglitutil.so.0] Error 1
Mesa generally provides virtual/egl (along with virtual/libgl,
which satisfies Piglit's current DEPENDS) but that is not the
implementation to use with Mali.
Signed-off-by: Daniel Díaz <daniel.diaz@linaro.org> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Maxin B. John [Mon, 29 May 2017 14:27:42 +0000 (17:27 +0300)]
mmc-utils: Upgrade to latest git version
Updates in the new version:
37c86e6 : mmc-utils: Check for ext_csd_rev only once c22a92f : mmc-utils: feature spec 5.0+, Pre EOL information 34a954b : mmc-utils: feature spec 5.0+, device life time estimation for MLC
and pSLC mode
Signed-off-by: Maxin B. John <maxin.john@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ghostscript before 9.21 might allow remote attackers to bypass the SAFER
mode protection mechanism and consequently read arbitrary files via the
use of the .libfile operator in a crafted postscript document.
Use-after-free vulnerability in Ghostscript 9.20 might allow remote
attackers to execute arbitrary code via vectors related to a reference
leak in .setdevice.
Ghostscript before 9.21 might allow remote attackers to bypass the SAFER
mode protection mechanism and consequently execute arbitrary code by
leveraging type confusion in .initialize_dsc_parser.
libjbig2dec.a in Artifex jbig2dec 0.13, as used in MuPDF and Ghostscript,
has a NULL pointer dereference in the jbig2_huffman_get function in
jbig2_huffman.c. For example, the jbig2dec utility will
crash (segmentation fault) when parsing an invalid file.
Kai Kang [Fri, 19 May 2017 08:52:46 +0000 (16:52 +0800)]
bind: build with threads and update configure options
Build without threads for bind is inherited from legacy openembedded.
All libc's support proper threading on Linux now, so enable threads
support for bind.
It is also need to disable static library build which cause package dhcp
fail to build after enable bind threads support.
Options devpoll and epoll are configured to choose most preferable
multiplex method for unix socket. The priorities are: epoll > poll >
select. When set '--enable-epoll', it just defines a var and include
header file that is available for cross compile. So use epoll for bind.
Add PACKAGECONFIG 'urandom' that could use /dev/urandom as random device.
Update file/directory ownerships to fix daemon start failure.
Signed-off-by: Kai Kang <kai.kang@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ross Burton [Fri, 26 May 2017 13:37:13 +0000 (14:37 +0100)]
gstreamer1.0-omx: delete broken m4 files
These m4 files need to be deleted so that the versions in the sysroot that we've
patched are used. Specifically, building documentation fails as the gtk-doc
checks fail.
Signed-off-by: Ross Burton <ross.burton@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Romain Perier [Sun, 28 May 2017 06:59:40 +0000 (08:59 +0200)]
busybox: make SYSLOGD optionnal
Some distros might choose another syslogd provider like rsyslogd.
update-alternative will update the link from syslogd to the right
provider. However the syslogd feature is still present and enabled
in busybox.
This commit adds a new configuration fragment to make syslogd
optionnal in busybox.
Signed-off-by: Romain Perier <romain.perier@collabora.com> Acked-by: Otavio Salvador <otavio@ossystems.com.br> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Li Zhou [Sat, 27 May 2017 09:46:15 +0000 (17:46 +0800)]
python3: Use _sysconfigdata.py to initialize distutils.sysconfig
Porting patch from
<https://github.com/python/cpython/commit/409482251b06fe75c4ee56e85ffbb4b23d934159>
to use _sysconfigdata.py to initialize distutils.sysconfig.
This patch makes that distutils.sysconfig doesn't need config-*/Makefile
in libdir any more. Next we can move it from python3-misc to python3-dev
package.
Signed-off-by: Li Zhou <li.zhou@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Kai Kang [Sat, 27 May 2017 07:45:03 +0000 (15:45 +0800)]
lsof: clear setuid
Having 'lsof' as a +s (setuid) binary could lead to security issues if
a compromise in the binary is found. It is better that it be -s by
default as a precaution.
Signed-off-by: Kai Kang <kai.kang@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
