| Age | Commit message (Collapse) | Author | Files |
|---|
|
We now support using dash but these bashisms triggered build failures for me
when using it. This replaces the code with something which works on dash.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
subprocess.check_output was only introduced in Python v2.7, so we
cannot use it. This refactors the QA test to use bb.process.Popen
instead.
This fixes the error:
AttributeError: 'module' object has no attribute 'check_output'
It no longer checks the return status of prelink-rtld, as that
case was simply adding noise. This QA test is intended to only
warn about specific paths that binaries could be linking to, not
handle the case where there is a missing library.
Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
When looking for RDEPENDS to process, bitbake iterates through PACKAGES
*and* PN. Since native.bbclass sets PACKAGES to be empty, its pointless
remapping the list of PACKAGES since this does nothing. There is a problem
since *_${PN} are used by bitbake but not remapped by the native.bbclass
class extension code.
This changes the code to remap _${PN} in both expanded and unexpanded
forms. As a result of this, various surprising dependencies are uncovered
and the patch rectifies those. These are real bugs since they're injecting
unneeded (unremapped) dependencies into the dependency chain.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Files under exec_prefix (commonly /usr) may not be available during
system recovery. exec_prefix may also be kept on a separate partition
that is mounted late in the boot process.
This QA test throws an warning if a binary in base_[bindir|sbindir|libdir]
is dynamically linked to a file under exec_prefix. The intention is to
turn this into an error in the near future.
It also checks executable non-binaries (e.g, shell scripts) in the above
base directories with a simple grep test to look for references to
exec_prefix. This test only produces a warning, since false positives
are likely.
This fixes [YOCTO #1008]
Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The ROOTFS_SIZE calculation was not correctly taking into account
the IMAGE_ROOTFS_EXTRA_SPACE variable, it would only be applied if
the size as determined by the ((du * overhead) + extra space) was
greater than the IMAGE_ROOTFS_SIZE, so if the du * overhead was smaller
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
Specifically...
SLOT: apparently redundant, deleted.
STAGING_IDLDIR: likewise
QMAKE_MKSPEC_PATH: no longer exported, moved to qmake_base.bbclass
STAGING_SIPDIR: no longer exported, moved to sip.bbclass
Signed-off-by: Phil Blundell <philb@gnu.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
postinstalls
Without this, various postinstalls get run with incorrect environments
leading to various failures when building the toolchains.
This adds some duplication and some variables we'd be better off
removing. It does unbreak the SDK ipk code for now though. This needs
revisiting.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
* Check if BUILDHISTORY_DIR exists before doing anything with it, in
case no tasks that would have created it have executed
* Ensure the git repo in BUILDHISTORY_DIR is initialised with "git init"
before attempting to do anything with it
* Check if any files have been added or changed before adding and
committing, to avoid an error from "git commit"
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
the .../usr/share/aclocal is empty for build from scratch. so avoid cp
error if the directory is empty.
Fixes this error, which is thrown before pseudo is built:
| cp: cannot stat
`/builddisk/build/build0/tmp/sysroots/x86_64-linux/usr/share/aclocal/*':
No such file or directory
NOTE: package libtool-native-2.4.2-r0.0: task do_configure: Failed
Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
|
|
In order for the kernel to accept an initramfs as a rootfs,
it must have /init - even if it is an empty file. Touch /init
to ensure it exists for cpio and cpio.gz image types.
When used with initrd and rootfs=/dev/ram0, this allows the
cpio and cpio.gz images to be used as the rootfs.
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
|
|
Tune the __db* size in DB_CONFIG.
This will reduce the __db* size from 62MB to 26MB in qemu.
[YOCTO #1769]
Signed-off-by: Mei Lei <lei.mei@intel.com>
|
|
Let each package-specific rootfs implementation provide basic functions
to query the existence of a package and install a list of packages and
then have a generic install function so this logic is in one place.
Note: unlike previous versions of this code in OE-Core this uses the
IMAGE_LINGUAS variable and not IMAGE_LOCALES - note that IMAGE_LINGUAS
was what was used in OE-Classic and it is already used in OE-Core in
order to install locale-base-*. This will mean that if IMAGE_LINGUAS is
left at the default you will now get more packages installed. If you
don't want these language support packages then you should set
IMAGE_LINGUAS explicitly.
This restores locale installation to the same state as OE-Classic, only
we now support all the packaging backends.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
This is needed by the copyleft_compliance class, so it can emit series files
for the patches, which greatly increases their usefulness to a user trying to
reconstruct the sources outside of OE.
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
|
|
--4ef4ab1d_66334873_12d0c
Content-Type: text/plain; charset="utf-8"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
The layer list is multiline, and includes quotes, so including it within
quotes is problematic, particularly if the revision includes characters which
are not valid outside of a quoted string in shell. To reproduce this failure:
do a build with an scm layer not on a branch.
Signed-off-by: Christopher Larson <chris_larson@mentor.com (mailto:chris_larson@mentor.com)>
|
|
Buildstats should be allowed to be optionally enabled. It's
recommended that it be enabled via the USER_CLASSES setting.
Alternatively it could be enabled via the INHERIT_DISTRO or
similar mechanism.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
|
|
Signed-off-by: Zhai Edwin <edwin.zhai@intel.com>
|
|
The changes made to prefer in-tree kernel tools forced the location
of kconf_check prematurely. For maximum flexibility, locating it
on the PATH is ideal, since the transition to in-tree tools will be
completely transparent.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
|
|
Implement list_package_recommends() for rpm, and filter out the
"suggests" (RPM's equivalent of recommends) in list_package_depends().
Respectively, these changes fix buildhistory handling of recommends for
RPM (so that they are included in the image graphs) and also prevents
failures during do_rootfs with buildhistory enabled if a recommend is
unsatisfied.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
* Symptom: kernel symlink in /boot is not created.
* Rename in order to create the expected files:
* /var/lib/opkg/info/kernel-3.1.4.postinst
* /var/lib/opkg/info/kernel-3.1.4.postrm
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Add a way to add arbitrary text to the top of a spec file. This
can be useful for adding specific tags to the produced binaries for
tracking purposes.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The magic file could be installed somewhere else if the system native prefix
settings are changed. Adjust the packaging files to compensate for this.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
|
|
The manifest conditionals COPY_LIC_MANIFEST and COPY_LIC_DIRS
were returning true due to missing quotes in the bash test. This
made images larger than expected.
Signed-off-by: Beth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Dmitry Eremin-Solenikov <dbaryshkov@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Removing a bashism that was dumping errors into rootfs log.
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Override UPDATERCD for nativesdk.
Without, update-rc.d is installed to /usr/sbin/update-rc.d
with meta-toolchain if nativesdk recipe inherits update-rc.d. An
example recipe for this is dbus.
Signed-off-by: Bernhard Guillon <Bernhard.Guillon@hale.at>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
There is a major issue with opkg images at the moment as preinst
functions are not being executed before their dependencies are installed
and this is leading to corruption of images containing avahi/dbus in
particular.
There are various changes in upstream opkg in the last 8 revisions which
make changes in this area but sadly these aren't enough to get things
working for us. I've updated to the latest svn revision with this patch
since it makes sense to pull in those changes first and then supplement
them with the attached patches.
There is a full description of the patches in the patch headers but in
summary they:
a) Ensure preinst functions execute with their dependencies installed.
This is a pretty invasive change as it changes the package install
ordering in general.
b) Ensure opkg sets $D, not $PKG_ROOT which we don't use
c) Change opkg to allow execution of postinstall functions which fail
resulting in execution on the target device as rootfs_ipk.bbclass
currently does manually.
The remaining changes interface this with the rest of the OE build
infrastructure, adding in the option to tell opkg to run the preinst and
postinst functions, ensure the correct environment is present for the
postinst scripts and removing the now unneeded rootfs_ipk class code
which opkg now does itself.
[YOCTO #1711]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Adding everything in ${prefix} to FILES and FILES-dbg is a historical artifact
left over from when this was sdk.bbclass back in 2009.
The BBCLASSEXTEND changes, multilib changes and cleanups to this file
over the intervening time mean this is simply no long needed.
Its also outright dangerous since it can link the packages together
in circular ways which don't make sense. The simplest thing to is
to drop this since it no longer does anything useful.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The initial directory support (-d) added to mkdosfs has proven to be incomplete
and non-compliant with FAT. Rather than continue to maintain this feature and
work around the various issues, we can use mcopy to construct the image.
bootimg.bbclass already depends on mtools-native (although it may not have
needed to previously). No new dependencies are introduced. The image created
passes dosfsck cleanly. Remove the call to dosfsck.
mcopy reported an error with the image we were creating:
Total number of sectors (107574) not a multiple of sectors per track (32)!
Add some logic to ensure the total sector count is an integral number of sectors
per track, including forcing the logical sector size to 512 in the mkdosfs
command.
The du -bks arguments are contradictory, -b is equivalent to "--apparent-size
--block-size=1" and -k is --block-size=1K. If reordered, -kbs will report the
disk usage in bytes insteadk of 1k blocks. Eliminate the ambiguity by using:
du --apparent-size -ks
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
CC: Nitin A. Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The generated filesystem has some errors that dosfsck is able to repair:
dosfsck 2.11, 12 Mar 2005, FAT32, LFN
Checking file /ldlinux.sys (LDLINUX.SYS)
Checking file /initrd (INITRD)
Checking file /syslinux.cfg (SYSLINUX.CFG)
Checking file /EFI (EFI)
Checking file /vmlinuz (VMLINUZ)
Checking file /rootfs.img (ROOTFS.IMG)
/vmlinuz
File size is 4144896 bytes, cluster chain length is > 4145152 bytes.
Truncating file to 4144896 bytes.
Checking file /EFI/.. (..)
Checking file /EFI/BOOT (BOOT)
Checking file /EFI/. (.)
/EFI/..
Start (16022) does not point to .. (0)
/EFI/.
Start (0) does not point to parent (16022)
Checking file /EFI/BOOT/.. (..)
Checking file /EFI/BOOT/. (.)
Checking file /EFI/BOOT/initrd (INITRD)
Checking file /EFI/BOOT/grub.cfg (GRUB.CFG)
Checking file /EFI/BOOT/bootia32.efi (BOOTIA32.EFI)
Checking file /EFI/BOOT/vmlinuz (VMLINUZ)
Checking file /EFI/BOOT/rootfs.img (ROOTFS.IMG)
/EFI/BOOT/..
Start (16021) does not point to .. (16022)
/EFI/BOOT/.
Start (0) does not point to parent (16021)
/EFI/BOOT/vmlinuz
File size is 4144896 bytes, cluster chain length is > 4145152 bytes.
Truncating file to 4144896 bytes.
Performing changes.
/build/poky/fri2/tmp/deploy/images/core-image-minimal-fri2-noemgd-20111216000605.hddimg:
12 files, 26635/26828 clusters
Add a dosfsck command following the mkdosfs command to correct these.
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
CC: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
[Yocto 1783] prevented us from using a proper /EFI/BOOT path in the live
images due to a bug in the -d patch for mkdosfs in dosfstools. With this
now fixed, we can place the efi payload where it belongs per spec and
the images will autoboot to the grub menu on EFI platforms.
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We process all files in the native/cross cases for finding and
fixing relocation issues. In the target case we've only processed
.la and binconfig files. Since there are other files which are
in need of this processing, this change allows recipes to specify
files that may be outside the normal set. This means hardcoded
paths that need to be fixmepathed to work correctly are handled
and addresses some sstate relocation bugs that have been seen.
Based on a patch from Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Right now for cross recipes e.g. gcc-cross and binutils-cross
we specify --disable-nls .... --enable-nls on configure cmdline
the --enable-nls coming from gettext bbclass.
So we disable nls for all cross inheriting recipes in gettext
bbclass and then we remove the extra --disable-nls in gcc-cross
and binutils-cross
This patch needs testing. Please help
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
Without this we don't take advantage of any configured multiple CPU
cores which seems a shame.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This addtask for do_populate_lic should be reverted back
to before do_package as before do_compile causes failures
when recipes are updated.
[YOCTO #1819]
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
|
|
X32 has different sizes for many types:
1. off_t and ino_t are 8 byte, instead of 4 byte.
2. pthread_mutex_t is 32byte, instead of 24 byte.
Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
Signed-off-by: H.J. Lu <hjl.tools@gmail.com>
|
|
As the standard for LICENSE format is not well defined, we
may well run into issues where LICENSE cannot be parsed via
ast. In cases like this, we need to warn and continue.
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This commit requires 4a13766c7b223d as it removes the exclusion
of deb package types from license manifest creation.
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
bashism in create_license_manifest removed.
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Implement the functions required for buildhistory to be able to query
installed packages, get dependencies etc. for deb-based images.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
emite_package_data needs to have LICENSE, as pkgdata is as an
audit of licenses collected during do_populate_license.
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
|
|
This is a fairly large commit, so I want to step through some of
what this accomplishes:
1. Additional licenses
I've added/modified/corrected some of the common licenses
within OE-core in prep. for a major license audit. Most of these
are in SPDX format. A few, there is no OSI equivalent.
2. Additional SPDX mappings
I've added some additional SPDX mappings to account for removing
some duplicate licenses. (ie GPL-2 and GPL-2.0 were the same)
I've also remapped a few things to more accurately reflect what
they should be pointing at.
Note: Artistic/LGPL/GPL/MPL. Quite a few LICENSE fields list these
licenses. They make no sense and need auditing. In a future commit
I have some fixes to particularly egregious LICENSE fields, but
a full audit should be done. I've listed to obvious candidates at:
https://wiki.yoctoproject.org/wiki/License_Audit
3. License manifest
We now have a license manifest generation that occurs in rootfs for
everything BUT .deb. This requires the changes Paul Eggleton has
done to rootfs_* particularly the list_installed_packages function.
The manifest is accurate during a parallel bitbake now (Weee!) and
is prime for my planned SPDX format manifest during the next period.
4. License manifest on image.
We also want the ability to add licenses to the image. This
functionality is also in base-files and will be stripped out in my
next commit. The manifest is not added by default and is a two var
setting in license.conf:
If I want *just* the manifest on the image (small) then I set:
COPY_LIC_MANIFEST = "1"
This copies the manifest to:
/usr/share/common-licenses/license.manifest
If I want the actual PKG license text on the image (much larger)
I need to set both both COPY_LIC_MANIFEST and COPY_LIC_DIRS in
license.conf. This will create:
/usr/share/common-licenses/(package name)/(licenses in LIC_SRC_URI)
Word of warning. This can be larger than wanted depending on image
and is probably ripe for linking licenses, but I ran out of time this
week to get that done.
5. Custom License search path.
We now have the ability to add licenses to the build without touching
common-licenses. This is set via license.conf:
LICENSE_PATH += "/path/to/custom/licenses"
You want to make sure the license is unique. license.bbclass picks the
first license it finds.
Signed-off-by: Elizabeth Flanagan <elizabeth.flanagan@intel.com>
|
|
Deploys sources for recipes for compliance with copyleft-style licenses
Defaults to using symlinks, as it's a quick operation, and one can easily
follow the links when making use of the files (e.g. tar with the -h arg).
By default, includes all GPL and LGPL, and excludes CLOSED and Proprietary.
Signed-off-by: Christopher Larson <kergoth@gmail.com>
|
|
In addition to moving this functionality to oe.license, makes the string
preparation more picky before passing it off to the ast compilation. This
ensures that LICENSE entries like 'GPL/BSD' are seen as invalid (due to the
presence of the unsupported '/').
Signed-off-by: Christopher Larson <kergoth@gmail.com>
|
|
We need pseudo to use the rootfs passwd/group files belonging to the
rootfs when building images. This patch ensures that we use the rootfs
files instead of those in the sysroot which can lead to incorrect file
ownership issues.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
* Add '-n' to suppress cleanmarker nodes since jffs2
* is usually used for NAND flash and the cleanmarkers are created in
* the OOB area by flash_eraseall -j
*
* From man pages: -n, --no-cleanmarkers
* Do not write cleanmarker nodes to the beginning of each erase block.
* This option can be useful for creating JFFS2 images for use on NAND flash,
* and for creating images which are to be used on a variety of hardware with differing eraseblock sizes.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
* Building the jffs2 filesystem to include summary information speeds up
* the mount process considerably at the cost of increased size.
* The rate of speedup is generally higher on NAND chips and on the chips
* where the erase block size is large.
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
this feature was present in OE-classic, is documented in OE-core
but its implementation is actually missing.
The present implementation is directly copied from OE-classic
and tested on an arm926 machine.
Signed-off-by: Eric Bénard <eric@eukrea.com>
|
