| Age | Commit message (Collapse) | Author | Files |
|---|
|
Currently a process was being forked off for each individual file
this class wanted to inspect with rpmdeps. This converts it to use
rpmdeps-oecore which allows batch processing of these dependencies.
For do_package for perl, this reduced the time by about 1 minute (33%).
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The sstate_hardcode_path() function triggered large numbers of exec()
calls when processing packages with large numbers of file relocations
(e.g. perl). This patch optimises those calls into longer single commands
which make the code significantly more efficient.
This reduced the do_package time for perl by 2 minutes (from 4.75 minutes)
for me.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
not the one used in sstate-cache
Without this fix, when packages are being built using distutils and
the python packages were deployed from sstate-cache is it possible
that the LD command will contain an invalid sysroot override.
We can fix this by always exported LDSHARED, which is the env var
that distutil looks for to override creating shared libraries.
Signed-off-by: Matthew McClintock <msm@freescale.com>
|
|
Allow use of BBCLASSEXTEND with 'cross' and use of virtclass-cross in
recipes.
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
|
|
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
|
|
Sort DEPENDS, RDEPENDS, and RRECOMMENDS in package info files so that
any changes in order (which are not important) are smoothed out in the
change history.
Fixes [YOCTO #1961]
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
Sort the contents of files-in-image.txt to avoid unnecessary changes
showing up in the history due to reordering.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
This should improve the fetching errors users see
when upstream src_uri is unavailable
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
In a similar manner to calling the patch.bbclass to locate patches that
were listed on the SRC_URI, it is also useful to query about 'other' items
that are on the SRC_URI. In the case of linux-yocto, it allows us to
know about kernel features that were specific on the URI and then apply
them to the current tree.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
|
|
Fixes [YOCTO #1951]
The do_bootimg code can generate hybrid efi+pcbios images (syslinux and
grub-efi) to boot on platforms with both EFI and legacy BIOS options. The
current implementation copies the kernel, initrd, and rootfs twice,
unnecessarily bloating the image size. This is an especially egregious bug
on -sato images.
Update the classes to use a common install of the kernel, initrd, and rootfs to
the root of the boot media. Grub-efi, syslinux, and isolinux can all reference
this location explicitly with a leading slash.
Tested with an EFI+PCBIOS image in both EFI and PCBIOS boot modes on two
platforms. No ISO image testing was performed.
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
|
|
commit:
patch.bbclass: abstract out logic that determines patches to apply
gives the ability for other clases to emit series files for use outside
of a build system, or even within the build system. There are sometimes
elements on the SRC_URI that while not directly applicable to patching,
can be related to patching the package. For example, the yocto kernel
class would like to know about these other source items on the SRC_URI
to locate out of tree kernel features.
This change keeps the default behaviour of returning patches, but adds the
ability to request that non-patch results be returned. Additional filtering
within the non-patch category, is left up to the caller of the routine.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This allows cmake to find the binaries of an external toolchain.
Signed-off-by: Ken Werner <ken.werner@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Right now gconf bbclass adds both postinst and prerm steps, but it does
not ensure that packages involved have gconf in RDEPENDS. This can lead
to a situation where postinst/prerm steps fail because gconftool-2 is
not installed.
Signed-off-by: Peter Tworek <tworaz666@gmail.com>
Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
These allow the recipe to build again:
- add compilerlibs, g++, libgcc to the provides
- add linux-libc-headers-dev to the packages
- in libc-package, only sed the ldd.bash.in file if it exists, as the external
toolchain is using that class as well
- shift the inherit location of the libc classes, as they were overriding the
recipe's do_install
- use ?= for EXTERNAL_TOOLCHAIN, so the user can set it
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If a recipe packages multiple versions of shlib (e.g. powervr drivers) we only want the shlib code to pickup $PN, not $PN-foo subpackages.
This keeps backward compatibility with the global PRIVATE_LIBS usage if no per package PRIVATE_LIBS are set for a given package. In other words: this doesn't break the firefox recipe.
Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Currently, pkgconfig dependencies get added to -dev packages which install
pc files however nothing in the system makes bitbake aware of these
dependencies so images can fail with pkgconfig being missing.
This change explictly adds in the pkgconfig RDEPENDS to the -dev packages
and hence makes bitbake aware of the dependency.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Fixes [YOCTO #1940]
do_bootimg was performing the FAT overhead calculations assuming FAT32 and then
forcing the use of FAT32 with "-F 32" to mkdosfs. The FAT specification is clear
on cluster count being the determining factor for FAT size (even if the fs
string is set to FAT32, go figure). Syslinux follows this spec, and rightly so,
resulting in a failure on core-image-minimal:
syslinux: zero FAT sectors (FAT12/16)
Drop the "-F 32" from mkdosfs to allow it to select the appropriate FAT size
based on cluster count. Leave the FAT overhead calculation in FAT32. This will
result in a little extra padding for really small images, but not enough extra
to justify recalculating for FAT12 and FAT16.
Tested with a core-image-minimal build for atom-pc. do_bootimg completed
successfully, and the resulting image was FAT16.
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
|
|
Signed-off-by: Koen Kooi <koen@dominion.thruhere.net>
|
|
* Add common check for the presence of an /init
* symlink, being 'touch' would fail with it
* and 'touch -h' is not yet generally supported
* (depends on distro on build host).
Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
|
|
* it was introduced in 87780fc09b066525e47d0f50ee5497db54d304cd
* then partially removed in 2feba313c991170747381c7cf821a45c2cd04632
* so remove this use too as runtime_script_required is not initialized anymore and results in
run.do_rootfs.6328: line 235: [: -eq: unary operator expected
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
or nativesdk
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
ordering of setscene tasks
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
In the do_populate_sysroot_setscene case, pseudo has been unloaded and we need
to reload it. This code change ensures all the pseudo options are specified
so pseudo loads correctly.
It also improves some of the comments so all the different contexts are listed.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
after do_populate_sysroot_setscene
The user addition needs to happen before the do_package files are extracted
by do_package_setscene since those are the ones we need to preserve the file
ownership information for. This patch ensures this happens.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
This check verifies that non staticdev and non pic packages do not
have static libraries in them. There are a few exceptions for the
libgcc-dev and libgcov-dev packages which are whitelisted, also the
*_nonshared.a need to be part of their respective -dev packages.
This is currently a warning to work out any other issues in the upper
layers but will be moved to ERROR at somepoint in the future.
[YOCTO #663]
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Fixes [YOCTO #1852] ... again.
The conversion from sectors to blocks was multiplying by 2 instead
of dividing by 2. Blocks are 1024 bytes, sectors are 512 bytes. The
result was images being much larger than intended.
Reported-by: Tom Zanussi <tom.zanussi@intel.com>
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
CC: Joshua Lock <josh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The COMMERCIAL_LICENSE mechanism has been superseded by LICENSE_FLAGS
so remove the code that implements COMMERCIAL_LICENSE and replace it
with the corresponding LICENSE_FLAGS version.
Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>
|
|
LICENSE_FLAGS are a per-recipe replacement for the COMMERCIAL_LICENSE
mechanism.
In the COMMERCIAL_LICENSE mechanism, any package name mentioned in the
global COMMERCIAL_LICENSE list is 'blacklisted' from being included in
an image. To allow the blacklisted package into the image, the
corresponding packages need to be removed from the COMMERCIAL_LICENSE
list. This mechanism relies on a global list defined in
default-distrovars.inc.
The LICENSE_FLAGS mechanism essentially implements the same thing but
turns the global blacklist into a per-recipe whitelist. Any recipe
can optionally define one or more 'license flags'; if defined, each of
the license flags defined for a recipe must have matching entries in a
global LICENSE_FLAGS_WHITELIST variable.
The definition of 'matching' is simple, but there are a couple things
users need to know in order to correctly and effectively use it.
Before we test a flag against the whitelist, we append _${PN} to it,
thus automatically making each LICENSE_FLAG recipe-specific. We then
try to match that string against the whitelist. So if the user
specifies LICENSE_FLAGS = 'commercial' for recipe 'foo', the string
'commercial_foo' should be specified in the whitelist in order for it
to match.
However, the user can also broaden the match by putting any
'_'-separated beginning subset of a LICENSE_FLAG in the whitelist,
which will also match e.g. simply specifying 'commercial' in the
whitelist would match any expanded LICENSE_FLAG starting with
'commercial' such as 'commercial_foo' and 'commercial_bar' which are
the strings that would have been automatically generated if those
recipes had simply specified LICENSE_FLAGS = 'commercial'
This allows for a range of specificity for the items in the whitelist,
from more general to perfectly specific. So users have the choice of
exhaustively enumerating each license flag in the whitelist to allow
only those specific recipes into the image, or of using a more general
string to pick up anything matching just the first component(s).
Note that this scheme works even if the flag already has _pn appended
- the extra _pn is redundant, but doesn't affect the outcome e.g. a
license flag of 'commercial_1.2_foo' would turn into
'commercial_1.2_foo_foo' and would match both the general 'commercial'
and the specific 'commercial_1.2_foo' as expected (it would also match
commercial_1.2_foo_foo' and 'commercial_1.2', which don't make much
sense as far as something a user would think of specifying in the
whitelist). For a versioned string, the user could instead specify
'commercial_foo_1.2', which would turn into 'commercial_foo_1.2_foo',
but which would as expected allow the user to pick up this package
along with anything else 'commercial' by specifying 'commercial' in
the whitelist, or anything with a 'commercial_foo' license regardless
of version by using 'commercial_foo' in the whitelist, or
'commercial_foo_1.1' to be completely specific about package and
version.
The current behavior of COMMERCIAL_LICENSE is replicated as mentioned
above by having the current set of COMMERCIAL_LICENSE flags
implemented using LICENSE_FLAGS = "commercial".
That being the case, the current COMMERCIAL_LICENSE can equivalently
be specified in the new scheme by putting the below in local.conf:
# This is a list of packages that require a commercial license to ship
# product. If shipped as part of an image these packages may have
# implications so they are disabled by default. To enable them,
# un-comment the below as appropriate.
#LICENSE_FLAGS_WHITELIST = "commercial_gst-fluendo-mp3 \
# commercial_gst-openmax \
# commercial_gst-plugins-ugly \
# commercial_lame \
# commercial_libmad \
# commercial_libomxil \
# commercial_mpeg2dec \
# commercial_qmmp"
The above allows all of the current COMMERCIAL_LICENSE packages in -
to disallow a particular package from appearing in the image, simply
remove it from the whitelist. To allow them all in, you could also
specify LICENSE_FLAGS_WHITELIST = "commercial".
Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>
|
|
This patch adds SignatureGenerator classes specific to OE. For now,
these emulate the previous behaviour with the exception that
dependencies on quilt-native are now ignored for checksum purposes.
The intent is to allow easier experimentation and customisation of
this code in future as a result of these changes.
Note that these changes require pending bitbake patches.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We don't need to see a Python stack backtrace when a patch does not
apply, just the error output from patch, so trap these kinds of errors
and ensure that we display the message and fail the task and nothing
else.
Fixes [YOCTO #1143]
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Record some additional information about images - the uncompressed size
of the final image as well as the values of various variables that may
have influenced its contents. This is recorded in a machine-readable
"image-info.txt" file similar in structure to the package history files.
Also add some code to analyse changes to these values. (Most of the
variable values aren't monitored directly but will be used as contextual
information when they change at the same time as the content of the
image changing.)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
In the switch to using src_patches and using patches in their
source directory, the scanning of WORKDIR migrated items like
config fragments was dropped. Adding WORKDIR back as a patch
directory restores the old functionality.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
|
|
Files with spaces must be fully quoted in an RPM spec file. It was
easier to quote all file paths, then to do it selectively.
Signed-off-by: Mark Hatle <mark.hatle@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
packagehistory.bbclass has been superseded by buildhistory.bbclass,
which gives more detailed output (including information on produced
images) as well as other enhanced functionality.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
OpenEmbedded doesn't officially use a bug tracker anymore, and a message
asking the user to fix the metadata could be applied to almost any error
that occurs, so just remove the whole message.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
When raising SkipPackage for COMPATIBLE_MACHINE and COMPATIBLE_HOST
exceptions, include the name of the variable as a hint to the user.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
|
|
Packaging will fail with executable files containing spaces in their
names. Patch quotes the parameter passed to 'objdump'.
Signed-off-by: Enrico Scholz <enrico.scholz@sigma-chemnitz.de>
|
|
Signed-off-by: Matthew McClintock <msm@freescale.com>
|
|
Signed-off-by: Matthew McClintock <msm@freescale.com>
|
|
Fixes [YOCTO #1852]
The bootimg class wasn't accounting for non-trivial amount of space
required by the directory entries and FATs for the FAT filesystem.
This patch attempts to make an accurate prediction of FAT overhead and
adjusts the image size accordingly. It assumes no more than 16 directory
entries per directory (which fit in a single sector). It also assumes
8.3 filenames. With the ceiling functions rounding up to full sectors
and tracks, these assumptions seem reasonable.
In order to ensure the calculations are accurate, this patch forces the
FAT size to 32, rather than allowing mkdosfs to automatically select 12,
16, or 32 depending on the image being built.
Tested by setting BOOTIMG_EXTRA_SPACE=0 and building core-image-minimal
and core-image-sato for fri2-noemgd from meta-intel.
Signed-off-by: Darren Hart <dvhart@linux.intel.com>
CC: Saul Wold <sgw@linux.intel.com>
|
|
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
|
|
Given a license string and whitelist and blacklist, determine if the
license string matches the whitelist and does not match the blacklist.
When encountering an OR, it prefers the side with the highest weight (more
included licenses). It then checks the inclusion of the flattened list of
licenses from there.
Returns a tuple holding the boolean state and a list of the applicable
licenses which were excluded (or None, if the state is True)
Examples:
is_included, excluded = oe.license.is_included(licensestr, ['GPL*', 'LGPL*'])
is_included, excluded = oe.license.is_included(licensestr, blacklist=['Proprietary', 'CLOSED'])
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
|
|
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
|
|
In this context, recipe "type" refers to whether it is 'target', 'native',
'cross', etc. COPYLEFT_RECIPE_TYPES is a space separated list of types to
include. It defaults to 'target'.
Signed-off-by: Christopher Larson <chris_larson@mentor.com>
|
|
For x32 the user space is 32bit and the kernel is 64bit.
So the elf.arch for vmlinuz is x86_64 and not x86. This commit
fixes this QA error thrown for x32 kernel.
| ERROR: QA Issue: Architecture did not match (62 to 3) on
/work/qemux86_64-poky-linux-gnux32/linux-korg-3.1+git1+e2bf8464ddbf5da24d3d320cded5691828a91a0b-r1/packages-split/kernel-vmlinux/boot/vmlinux-3.1.0-yocto-standard-01628-ge2bf846
Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
|
|
Fixes this bug: [YOCTO #1874]
Fixes an issue where a locale package depends on one package while it
also provides the same, as seen bellow.
Package: locale-base-de-de
Version: 2.12-r19
Depends: eglibc-binary-localedata-de-de
Provides: virtual-locale-de-de, virtual-locale-de,
eglibc-binary-localedata-de-de
Actually the eglibc-binary-localedata-de-de is ia separate package,
and it should not be part of provides of the locale-base-de-de.
Signed-off-by: Nitin A Kamble <nitin.a.kamble@intel.com>
|
