* Removed openvg references in PACKAGECONFIG, FILES and PACKAGES since OpenVG
support was removed in mesa 10.6
* Tested with X11 :
* xf86-video-freedreno on IFC6410 and DragonBoard 410c
* softpipe on DragonBoard 410c
* using meta-qualcomm BSP
Signed-off-by: Nicolas Dechesne <nicolas.dechesne@linaro.org> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
bootimg.bbclass:iso: determine the -iso-level only if there is a rootfs
It may be possible that some bootable images do not use a rootfs; this
may be the case if the bootable image utilize only an initrd for all its
needs.
If there is not rootfs, the size calculation will fail. Furthermore, given
that the iso9660 conformance level is determined using the size of rootfs,
it makes sense to not make the calculation if there is not such rootfs.
Signed-off-by: Ricardo Neri <ricardo.neri-calderon@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Andre McCurdy [Wed, 15 Jul 2015 08:16:39 +0000 (01:16 -0700)]
bash_3.2.48: dont include target CFLAGS in host LDFLAGS
Building the host tool 'mkbuiltins' will fail if the target CFLAGS
contains an option which isn't supported by the host's gcc. To prevent
the issue, define LDFLAGS_FOR_BUILD based on CFLAGS_FOR_BUILD instead
of CFLAGS.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Andre McCurdy [Thu, 6 Aug 2015 00:36:49 +0000 (17:36 -0700)]
gettext_0.16.1: remove obsolete uclibc specific patch
gettext-error_print_progname.patch was originally created for gettext
v0.14.6 and does not apply cleanly to gettext v0.16.1.
Since the original issue addressed by the patch isn't documented and
because gettext v0.16.1 seems to be build OK for uclibc without the
patch, assume the patch is obsolete and no longer required.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Jason Wessel [Thu, 6 Aug 2015 19:35:53 +0000 (14:35 -0500)]
initscripts: Fix regression for requiring /bin/bash
It is not possible to create a rootfs with only busybox + initscipts.
This is a result of a regression from commit a4b53872a8a9a2743299acbff015f7f2750a69d6 (initscripts: add
/sbin/sushell for systemd service debug-shell).
The /sbin/sushell should just use /bin/sh else you end up with a
problem when creating the end image with a failed smart transaction
shown below:
ERROR: Unable to install packages. Command '/proj/bitbake_build/tmp/sysroots/x86_64-linux/usr/bin/smart --quiet --data-dir=/proj/bitbake_build/tmp/work/qemuarma9-wrs-linux-gnueabi/wrlinux-image-glibc-small/1.0-r1/rootfs/var/lib/smart install -y dropbear@armv7at2_vfp hac@armv7at2_vfp run-postinsts@all kernel-modules@qemuarma9 packagegroup-core-boot@qemuarma9' returned 1:
error: Can't install initscripts-1.0-r155.0@armv7at2_vfp: no package provides /bin/bash
Signed-off-by: Jason Wessel <jason.wessel@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Kai Kang [Tue, 4 Aug 2015 03:25:35 +0000 (11:25 +0800)]
rt-tests: fix redo install error
Currently gzip on host is used. If host gzip is provided by pigz, it
fails to redo install that pigz can't handle the option after file name.
When run command for target install in Makefile:
gzip src/backfire/backfire.4 -c > OUTPUT_FILE
File src/backfire/backfire.4 is zipped into backfire.4.gz but the
OUTPUT_FILE is empty. When rerun do_install, it shows warning:
| gzip: src/backfire/backfire.4 does not exist -- skipping
Andreas Müller [Wed, 22 Jul 2015 08:21:10 +0000 (10:21 +0200)]
boost: add python3 support
* Both - python/python3 - libs are build by adding 'python' to PACKAGECONFIG
but are packed into separate packages
* Indention was wrong in __anonymous()
Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Sun, 9 Aug 2015 22:10:08 +0000 (15:10 -0700)]
libpam: Fix patch broken during upgrade
"0x200" became "0200" during the upgrade to libpam 1.2.1 in:
http://git.yoctoproject.org/cgit/cgit.cgi/poky/commit/?id=88dd997d9941b63ae9eead6690ecf2b785c0740c
and this broke the IMAGE_FEATURES like debug-tweaks.
I've converted all the values to octal here to match the original
header file convention and make it clearer.
[YOCTO #8033]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Topi Kuutela [Fri, 7 Aug 2015 06:42:35 +0000 (09:42 +0300)]
security_flags.inc: disable -pie and -fpie from Python3 compilation.
If security_flags.inc is 'required' to the image, -pie and -fpie options
are added to CFLAGS. These are not compatible with -shared GCC option.
The result is several errors of following form and missing Python3
modules in the image:
*.o In function `_start': *.S undefined reference to `main'
collect2: error: ld returned 1 exit status
Signed-off-by: Topi Kuutela <topi.kuutela@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ed Bartosh [Mon, 3 Aug 2015 13:51:51 +0000 (16:51 +0300)]
wic: release 0.2.0
Major changes in 0.2.0:
- implemented UUID support
- fixed support of GPT partition tables
- implemented running bitbake from wic
- implemented image compressing
- started testing with oe-selftest. implemented 17 test cases
- used native tools (parted, mkfs, mtools, etc)
- usability and documentation fixes
- code clenup
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
stress: update checksums to match upstream tarball
COPYING checksum has changed because the previous upstream tarball
(taken from Fedora) had GPLv3 in it, but author's tarball has GPLv2.
Otherwise the tarballs are identical
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ed Bartosh [Wed, 5 Aug 2015 09:16:00 +0000 (12:16 +0300)]
create-pull-request: cd to relative directory
create-pull-request -d path creates empty patches if directory
is specified as a path, i.e. ./bitbake or ./bitbake/ or full path.
It behaves expected way only if script is run with -d bitbake, i.e.
relative dir name doesn't contain '\'.
Fixed this unwanted behaviour by changing directory and running
git format-patch in it with --relative, without specifying
relative path as a parameter.
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
wpa-supplicant: backport patch to fix CVE-2015-4141,
CVE-2015-4143, CVE-2015-4144, CVE-2015-4145, CVE-2015-4146
Backport patch to fix CVE-2015-4141, CVE-2015-4143, CVE-2015-4144, CVE-2015-4145, CVE-2015-4146.
This patch is originally from:
For CVE-2015-4141:
http://w1.fi/security/2015-2/0001-WPS-Fix-HTTP-chunked-transfer-encoding-parser.patch
For CVE-2015-4143:
http://w1.fi/security/2015-4/0001-EAP-pwd-peer-Fix-payload-length-validation-for-Commi.patch
http://w1.fi/security/2015-4/0002-EAP-pwd-server-Fix-payload-length-validation-for-Com.patch
For CVE-2015-4144 and CVE-2015-4145:
http://w1.fi/security/2015-4/0003-EAP-pwd-peer-Fix-Total-Length-parsing-for-fragment-r.patch
http://w1.fi/security/2015-4/0004-EAP-pwd-server-Fix-Total-Length-parsing-for-fragment.patch
For CVE-2015-4146:
http://w1.fi/security/2015-4/0005-EAP-pwd-peer-Fix-asymmetric-fragmentation-behavior.patch
Signed-off-by: Fan Xin <fan.xin at jp.fujitsu.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
A mix of a ~1 month old (working) image and the failing image led me
to determine /usr/lib64/xorg/modules/drivers/intel_drv.so was bad.
However nothing changed the parent package xf86-video-intel in
yocto for months.
So brute force bisecting yocto leads to the above commit. Inserting
libdrm as git and then bisecting the 2.4.59 --> 2.4.62 uprev content
leads to libdrm commit 8576527cfacaf42af8316e1030c192193e94225a
("intel: Merge latest i915_drm.h") -- specifically, the part of that
commit that adds __u64 flags to drm_i915_gem_mmap lands in our sysroot
and is used by the compile of xf86-video-intel.
With the specific trigger finally pinpointed, the final leap was to
correlate that trigger to the upstream xf86-video-intel commit that
accounts for the fallout of this libdrm header change, that we did
not yet have here. Fun times. Hopefully this complete description
here will ensure nobody else has to retrace these steps again.
Cc: Damien Lespiau <damien.lespiau@intel.com> Cc: Neil Roberts <neil@linux.intel.com> Cc: Chris Wilson <chris@chris-wilson.co.uk> Cc: Steffen Pankratz <steffen.pankratz@elektrobit.com> Cc: Ross Burton <ross.burton@intel.com> Cc: Richard Purdie <richard.purdie@linuxfoundation.org> Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Sun, 9 Aug 2015 07:07:05 +0000 (00:07 -0700)]
oeqa/qemurunner: Improve runqemu log output debug
If runqemu fails, ensure the log output is shown as its invaluable
to aid debugging. Its slightly convoluted since we need to ensure
we don't block on reading the pipe which may still be executing
hence the need for nonblocking IO.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Sun, 9 Aug 2015 07:03:49 +0000 (00:03 -0700)]
runqemu: Add error handling for unexpected tap devices
In theory the code reduces the tap device number to an integer. This
patch adds error checking to ensure that does happen and that the script
exits if something unexpected happens.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Mon, 3 Aug 2015 16:47:31 +0000 (17:47 +0100)]
weston: Fix parallel make race
The current approach used by the Makefile ends up installing
weston.desktop twice and can give build failures when using
parallel make flags. Change to just have one DATA reference.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Patrick Ohly [Tue, 4 Aug 2015 16:24:00 +0000 (18:24 +0200)]
combo-layer: fix action_pull for unknown branch
When reconfiguring the branch to something not already fetched,
action_pull fails with
error: pathspec '<new branch name>' did not match any file(s) known to git.
It is the "git checkout" which fails like that. To solve this,
try the faster "git checkout + git pull" first and only if that fails,
fall back to the slow "git fetch + git checkout".
In the conf.hard_reset case, do the checkout always after the git fetch.
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Patrick Ohly [Tue, 4 Aug 2015 16:23:59 +0000 (18:23 +0200)]
combo-layer: enhance output in action_pull
The git operations can fail, for example when the branch is unknown
or misconfigured.
Better move the info message and extend it such that it is printed
first and provides the necessary context, because otherwise the
CalledProcessError exception gets dumped without mentioning for which
component it occurred.
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Wenzong Fan [Fri, 10 Jul 2015 08:54:55 +0000 (16:54 +0800)]
perl: fix build failure if building dir has the "blib" string
If build dir includes the string "blib", filter it out from @INC may empty
the @INC and cause build errors like:
Can't locate ExtUtils/MakeMaker.pm in @INC \
(you may need to install the ExtUtils::MakeMaker module) \
(@INC contains: .) at Makefile.PL
Signed-off-by: Roy Li <rongqing.li@windriver.com> Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
robotparser has been moved to urllib.robotparser in Python 3, and gets
packaged as part of python3-netclient alongside the rest of urllib.
This causes python3-modules to depend on an empty/missing
python3-robotparser package.
robotparser.py is relatively small and has no dependencies outside
urllib, so it makes more sense to eliminate the python3-robotparser
package than to alter the manifest for python3-netclient.
Part of the fix for [YOCTO #8048].
Signed-off-by: Dominic Sacré <dominic.sacre@gmx.de> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Patch setup.py so that the detect_modules() function looks for required
libraries and headers in STAGING_LIBDIR / STAGING_INCDIR.
Without this patch, several extension modules are not built, even though
their dependencies are present in the compiler's search paths.
The result is the following warning, and ultimately incomplete packages:
| The necessary bits to build these optional modules were not found:
| _bz2 _curses_panel _dbm
| _gdbm _lzma _sqlite3
| nis readline zlib
| To find the necessary bits, look in setup.py in detect_modules() for the module's name.
With the Python 3.3 version of the recipe, at least some of these modules
were built, because STAGING_LIBDIR happened to be in the search path
by coincidence. Due to changes to distutils in Python 3.4, this is no
longer the case.
A previous patch that only affects the search paths for SSL is dropped, as
this one is a more complete fix for [YOCTO #7768].
Part of the fix for [YOCTO #8048].
Signed-off-by: Dominic Sacré <dominic.sacre@gmx.de> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Robert Yang [Tue, 28 Jul 2015 09:03:33 +0000 (02:03 -0700)]
useradd_base.bbclass: do not warn when the user exists
Use bbnote rather than bbwarn when the user exists, otherwise we would
always get the warnings when rebuild the recipe or build with mulitlib,
everything is OK if the user exists, so it should not be a warning.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Robert Yang [Thu, 30 Jul 2015 15:18:09 +0000 (08:18 -0700)]
base.bbclass: move invalid PACKAGECONFIG checking to insane.bbclass
There was a problem when check invalid PACKAGECONFIG in the anonymous
function (when parsing), for example, there are two versions of a
recipe:
foo_1.0.bb
foo_2.0.bb
While foo_2.0.bb has a "PACKAGECONFIG[item] = 'x,y,z'", but foo_1.0.bb
doesn't, if we set PACKAGECONFIG_pn-foo = "item" in a conf file, we
would get the warning about invalid PACKAGECONFIG for foo_1.0.bb. Delay
the checking to build time will fix the problem.
[YOCTO #8065]
Signed-off-by: Robert Yang <liezhi.yang@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Andre McCurdy [Thu, 30 Jul 2015 20:56:27 +0000 (13:56 -0700)]
busybox: support mount via label for btrfs and squashfs
Also move the mount via label (and mount via UUID) support into its
own config fragment and disable volume ID support for less common
filesystem formats exFAT and NILFS.
Following this commit, mount via label is supported for btrfs, ext, f2fs,
fat and squashfs.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Andre McCurdy [Thu, 30 Jul 2015 20:56:24 +0000 (13:56 -0700)]
busybox: refresh defconfig for busybox v1.23.2
The current oe-core busybox defconfig dates back to busybox v1.20.2,
so configure options introduced in busybox v1.21.x, v1.22.x and
v1.23.x take on default values when the oe-core defconfig is run
through 'make oldconfig'.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Andre McCurdy [Thu, 30 Jul 2015 20:56:23 +0000 (13:56 -0700)]
busybox: re-order defconfig to align with busybox 1.23.2
No functional changes, simply re-order lines in defconfig so that
the existing options don't move elsewhere in the file when run
though busybox 1.23.2 'make oldconfig'.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Zhixiong Chi [Fri, 31 Jul 2015 02:20:25 +0000 (10:20 +0800)]
dhcp: fix installed not shipped warning for lib32
Modify the dhcp.inc with using the variable ${PN} instead of direct
packagename, so that the content will not be override after expanding
while we build the lib32-dhcp package with FILES_${PN}-xxxx_append.
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
opkg: solve inconsistency of using different lists directories
Default behavior of opkg was to use ${OPKGLIBDIR}/opkg/lists;
but in our recipe we modify it to ${OPKGLIBDIR}/opkg/${OPKGLIBDIR}/opkg/,
when appending package-management to IMAGE_FEATURES these lists are
populated during build time (using the default directory),
but since our config was different these populated lists were never used at runtime,
this patch solves this inconsistency by using default behavior for both build time and runtime.
[YOCTO #6966]
Signed-off-by: Alejandro Hernandez <alejandro.hernandez@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
base.bbclass: Note when including pn with INCOMPATIBLE_LICENSES
We need to be able to tell people if we WHITELIST a recipe
that contains an incompatible licese.
Example: If we set WHITELIST_GPL-3.0 ?= "foo", foo will end
up on an image even if GPL-3.0 is incompatible. This is the
correct behaviour but there is nothing telling people that it
is even happening.
Signed-off-by: Beth Flanagan <elizabeth.flanagan@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Thu, 30 Jul 2015 15:03:14 +0000 (11:03 -0400)]
perf: add liblzma depdenency
With kernel commit 6c6f0f6164f [tools build: Add feature check for lzma
library], perf now performs a test for lzma and adds functionality
appropriately.
This currently is a floating dependency that will sometimes be present,
and sometimes not. The result is the following QA warning:
WARNING: QA Issue: perf rdepends on liblzma, but it isn't a build dependency? [build-deps]
By adding xz to the DEPENDS, we'll ensure that it is consistently in
place.
[YOCTO: #8045]
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
