automake: drop the Python scriptlets to determine Python's directories
python.m4 has python scriptlets that determine where the system's library
directory is. For Python 2.x distutils.sysconfig is used (which works because
we patch it), for Python 3.x plain sysconfig is used (which doesn't work
for example on x32). There is also a hardcoded fallback if the scriptlets fail.
This commit removes the scriptlets altogether and just uses the hardcoded path always,
due to above mentioned issues with x32.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Maxin B. John [Wed, 1 Jun 2016 13:25:56 +0000 (16:25 +0300)]
libpam: update dependency list
Depend only on bison-native as configure script is checking
only for bison binary and libpam itself doesn't need target bison.
Add libxml2-native for deterministic build (it detects xmlcatalog
and xmllint)
Signed-off-by: Martin Jansa <martin.jansa@gmail.com> Signed-off-by: Maxin B. John <maxin.john@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
oeqa/selftest/recipetool.py: Tweak matching of warning line
We are observing cases (see below) where the 'WARNING:' prefix is not at the
beginning of a line, so instead of expecting it in the beginning, match it
within the string.
======================================================================
FAIL: test_recipetool_appendfile_patch (oeqa.selftest.recipetool.RecipetoolTests)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/pokybuild/yocto-autobuilder/yocto-worker/nightly-oe-selftest/build/meta/lib/oeqa/utils/decorators.py", line 109, in wrapped_f
return func(*args, **kwargs)
File "/home/pokybuild/yocto-autobuilder/yocto-worker/nightly-oe-selftest/build/meta/lib/oeqa/selftest/recipetool.py", line 285, in test_recipetool_appendfile_patch
self.fail('Patch warning not found in output:\n%s' % output)
AssertionError: Patch warning not found in output:
Parsing recipes..WARNING: File /etc/selftest-replaceme-patched is added by the patch /home/pokybuild/yocto-autobuilder/yocto-worker/nightly-oe-selftest/build/meta-selftest/recipes-test/recipetool/files/add-file.patch - you may need to remove or replace this patch in order to replace the file.
NOTE: Writing append file /tmp/recipetoolqaug0kdb4x/recipes-test/recipetool/selftest-recipetool-appendfile.bbappend
NOTE: Copying /tmp/recipetoolqagci9tita/testfile to /tmp/recipetoolqaug0kdb4x/recipes-test/recipetool/selftest-recipetool-appendfile/testfile
done.
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Wed, 1 Jun 2016 12:27:25 +0000 (13:27 +0100)]
qemurunner: Use surrogateescape decoding
Since the stream can contain invalid binary characters (e.g. from
ppc's bootloader) use surrogateescape decoding to ensure we do process
the character stream, else it can hang/timeout.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
buildhistory_analysis: Convert stream data (bytes) to strings
The bytes type variables are threat different as strings variables in python3,
so convert bytes to strings. This was found when using the
script buildhistory-diff, where the interpreter was yielding this error
TypeError: Type str doesn't support the buffer API
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
selftest/devtool: Compare sets instead of arrays on AssertEqual
Sets are safer when comparing internal elements and positions are not
important. This commit avoid errors observed on python3 builds as reported
on the below bugzilla entry.
[YOCTO #9661]
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ed Bartosh [Fri, 20 May 2016 08:57:25 +0000 (11:57 +0300)]
scripts: use python3 in shebang
As most of oe-test cases work for devtool and recipetool
it makes sense to switch both tools to python 3 by
explicitly referring to python3 in their shebangs.
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Richard Purdie [Fri, 20 May 2016 10:17:05 +0000 (11:17 +0100)]
classes/lib: Update to use python3 command pipeline decoding
In python3, strings are unicode by default. We need to encode/decode
from command pipelines and other places where we interface with the
real world using the correct locales. This patch updates various
call sites to use the correct encoding/decodings.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Wed, 1 Jun 2016 12:17:23 +0000 (13:17 +0100)]
oe-buildenv-internal: Update to python3
Check that 'python' is a python v2 since that is what we assume everywhere
and upstream python devs recommend. We can need both python2 and python3
available since we don't control the software we might download and run.
Also check that python 3 is >= 3.4.0, our minimum version for bitbake.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Stephano Cetola [Tue, 31 May 2016 20:59:54 +0000 (13:59 -0700)]
systemd: allow add users as a rootfs postprocess cmd
Adding all the users / groups to systemd is only available for readonly
file systems. This change allows users to add them to read / write file
systems as well by specifying:
Enrico Jorns [Mon, 30 May 2016 13:04:05 +0000 (15:04 +0200)]
perl-ptest.inc: fix tar call to prevent objcopy failure
With tar version 1.29, the tar call used to copy the ptest files will
not work anymore. While the call did not match the man page (but worked)
before, anyway, the latest update of tar seems to have a more strict argument
handling.
With the current version of the tar call, the copying of files still
works with latest tar version, but the excludes will not be handled
properly anymore.
This results in having binaries compiled with host GCC in the package.
When doing the strip_and_split files in do_package() with the target
objcopy, bitbake will fail with this error:
ERROR: objcopy failed with exit code 256 (cmd was [...])
[...]
File format not recognized
Thus, the current argument issues and required changes are:
* Options must be placed _before_ the pathnames.
* --exclude must be followd by a '=' in order to work properly
* 'f' options is for providing an archive file, which is unnecessary in
this case
Note that this could also be a candidate for backporting.
Signed-off-by: Enrico Jorns <ejo@pengutronix.de> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Edwin Plauchu [Fri, 27 May 2016 20:29:21 +0000 (15:29 -0500)]
unzip: fix security issues
This patch avoids unzip fails to compile with compiler flags which elevate common string formatting issues into an error (-Wformat -Wformat-security -Werror=format-security).
[YOCTO #9551]
Signed-off-by: Edwin Plauchu <edwin.plauchu.camacho@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Barker [Sat, 28 May 2016 13:26:15 +0000 (14:26 +0100)]
libarchive: Upgrade to v3.2.0
All patches are removed as they are no longer needed. Most were merged into this
release of libarchive. "0001-Set-xattrs-after-setting-times.patch" was dropped
upstream after discussion, see https://github.com/libarchive/libarchive/pull/664.
The COPYING file in libarchive had a couple of minor changes to clarify which
files are under which copyrights but the overall license is unaffected.
Signed-off-by: Paul Barker <paul@paulbarker.me.uk> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
default-versions.inc: drop python-related defaults
There is only one version of python 2.x provided, so no need
to set a preferred version.
PYTHON_BASEVERSION is now set explicitly in python-dir.bbclass and
python3-dir.bbclass, so fix up a few recipes that relied on it being
set in default-versions.inc without inheriting python-dir.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
python-native, python3-native: remove the use of exported HOST_SYS and BUILD_SYS variables
The code that utilized them was superseded by the code (in the same patch!)
that is utilizing STAGING_LIBDIR/STAGING_INCDIR, and wasn't correct in the
first place as HOST_SYS is not necessarily the same as the sysroot directory
name.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Paul Eggleton [Sun, 29 May 2016 22:21:02 +0000 (10:21 +1200)]
recipetool / devtool: set a fixed SRCREV by default when fetching from git
If fetching source from a git repository, typically within OpenEmbedded
we encourage setting SRCREV to a fixed revision, so change to do that by
default and add a -a/--autorev option to use "${AUTOREV}" instead.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Sun, 29 May 2016 22:21:01 +0000 (10:21 +1200)]
recipetool: create: use ${BP} for subdir for binary packages
If we use ${BP} for the subdirectory, the default value of S will work
rather than having to have an ugly value derived from the package
file name in both places. This does mean that we have to assume the
default though (we can't just let the normal logic work because the
value of BP is the default until later on, so the replacement doesn't
work).
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Sun, 29 May 2016 22:21:00 +0000 (10:21 +1200)]
recipetool: create: extract variable values from .deb/.ipk/.rpm
Extract the metadata from package files and use it to set variable
values in the recipe (including recipe name and version, LICENSE,
SUMMARY, DESCRIPTION, SECTION and HOMEPAGE). For LICENSE we take care
not to step on any value determined by our license file scan; if there
is one we simply add a comment above the LICENSE setting so the user can
resolve it.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Sun, 29 May 2016 22:20:59 +0000 (10:20 +1200)]
recipetool: create: support extracting SUMMARY and HOMEPAGE
Allow plugins to set any variable value through the extravalues dict,
and use this to support extracting SUMMARY and HOMEPAGE values from spec
files included with the source; additionally translate "License:" to a
comment next to the LICENSE field (we have our own logic for setting
LICENSE, but it will often be useful to see what the spec file says if
one is present).
Also use the same mechanism for setting the same variables for node.js
modules; this was already supported but wasn't inserting the settings in
the appropriate place in the file which this will now do.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Sun, 29 May 2016 22:20:58 +0000 (10:20 +1200)]
lib/oe/recipeutils: fix insertion of variable values
Add some more variables in appropriate places in recipe_progression such
that the patch_recipe_* functions are able to insert variables in the
right place within a recipe.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This function was assuming that what you wanted was that output lines
had trailing newline characters. If you're just outputting each line
verbatim to a text file then that's fine, but sometimes you start with
the assumption that the lines don't have trailing newlines; thus we
shouldn't allow for the possibility that the caller doesn't want them
and add a parameter to control it.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Sun, 29 May 2016 22:20:56 +0000 (10:20 +1200)]
lib/oe/recipeutils: split out patch_recipe_lines()
Split out a function from patch_recipe_file() that takes just the lines
as input so we can edit recipe lines in memory. This will be used within
recipetool to ensure we insert new values in the right place.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
