Alongside reworking the way devtool extracts source, we now need to
ensure that within the extensible SDK where task signatures are locked,
the signatures of the tasks for the recipes being worked on get unlocked
at the right time or otherwise we'll now get taskhash mismatches when
running devtool modify on a recipe that was included in the eSDK such as
the kernel (due to a separate bug). The existing mechanism for
auto-unlocking recipes was a little weak and was happening too late, so
I've reimplemented it so that:
(a) it gets triggered immediately when the recipe/append is created
(b) we avoid writing to the unlocked signatures file unnecessarily
(since it's a global configuration file) and
(c) within the eSDK configuration we whitelist SIGGEN_UNLOCKED_RECIPES
to avoid unnecessary reparses every time we perform one of the
devtool operations that does need to change this list.
Fixes [YOCTO #11883] (not the underlying cause, but this manifestation
of the issue).
(From OE-Core rev: 4e9a0be32fc30fb87d65da7cd1a4015c99533aff)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since it was first implemented, devtool's source extraction (as used by
the devtool modify, extract and upgrade subcommands) ignored other recipe
dependencies - so for example if you ran devtool modify on a recipe that
fetches from svn or is compressed using xz then it would fail if those
dependencies hadn't been built first. Now that we can execute tasks in
the normal way (i.e. tinfoil.build_targets()) then we can rework it to
use that. This is slightly tricky in that the source extraction needs to
insert some logic in between tasks; luckily we can use a helper class
that conditionally adds prefuncs to make that possible.
Some side-effects / aspects of this change worth noting:
* Operations are a little slower because we have to go through the task
dependency graph generation and other startup processing. There's not
really any way to avoid this though.
* devtool extract didn't used to require a workspace, now it does
because it needs to create a temporary bbappend for the recipe. (As
with other commands the workspace be created on the fly if it doesn't
already exist.)
* I want any existing sysroot files and stamps to be left alone during
extraction since we are running the tasks off to the side, and
especially devtool extract should be able to be used without touching
these. However, this was hampered by the automatic removal process in
sstate.bbclass triggered by bb.event.ReachableStamps when the task
signatures change, thus I had to introduce a way to disable this
removal on a per-recipe basis (we still want it to function for any
dependencies that we aren't working on). To implement this I elected
to use a file written to tmp/sstate-control which gets deleted
automatically after reading so that there's less chance of stale files
affecting future sessions. I could have used a variable but this would
have needed to be whitelisted and I'd have to have poked its value in
using the setVariable command.
Fixes [YOCTO #11198].
(From OE-Core rev: 830dbd66992cbb9e731b48d56fddf8f220349666)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently, $HOME/.local is being added into sys.path for the native
Python, causing subtle host contamination. Suppress this by exporting
PYTHONNOUSERSITE = "1" as documented in PEP 370.
(From OE-Core rev: 8fe9fb4d5a61dcbcb3fc5b9ee0234cc135af873f)
Signed-off-by: Martin Kelly <mkelly@xevo.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Go does not play well with thumb, so ensure that the
toolchain and any packages use arm, not thumb, instructions.
(From OE-Core rev: 24da8c321831dcc5de00d65d6c5613efee109b57)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This adds ptest support for Go packages so its unittest content is
packaged and integrated onto the test framework.
(From OE-Core rev: 2343cd90b9706589b33510c560ed83a9648fb133)
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently every Go package will end with GNU_HASH in the ELF binary
however adding it to every recipe is cumbersome so instead we handle
that here.
(From OE-Core rev: 6699e668413c10704ffa8094b3dca67a9b88422a)
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For architectures that support it, use the -linkshared
build option to build packages against the shared Go
runtime.
(From OE-Core rev: 5624a773e4db3ad2251641e69b04dc380e74a4c7)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Go only supports shared libraries for some architectures, so
add a variable for use elsewhere that gets a non-null value
only for those architectures.
(From OE-Core rev: 2275712df152b73ce49b36bdf9f8d744c68c9c50)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When these functions are being called INSANE_SKIP has already been taken into
account, so don't confuse the code by passing the skip list.
(From OE-Core rev: 0001ceead406b1e8ba4fd16d0ecb5fbf5b55ba66)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The Device Tree is commonly used but it is still kept as a .inc file
instead of a proper class. Instead now we move the Device Tree code to
a kernel-devicetree class and automatically enable it when the
KERNEL_DEVICETREE variable is set.
To avoid breakage in existing layers, we kept a linux-dtb.inc file
which raises a warning telling the user about the change so in next
release this can be removed.
(From OE-Core rev: 03a00be7f2062aefef0e51ef20a4c9737f6685e7)
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* use conditional assignment for the CGO_xxx
variables, so they can be overridden more easily
* remove the TOOLCHAIN_OPTIONS and TARGET_CC_ARCH
references, since those are already present in
CC and CXX
* remove the TARGET_ prefix so the values are
appropriate for native, nativesdk, etc. builds
* move the GOROOT export away from the CGO settings
and closer to its definition
(From OE-Core rev: 088528021d6979a8e2d6bc33d63a166e300cfde4)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
GOROOT_FINAL is used by the Go linker for rewriting
source paths when the build GOROOT is not the same
as the runtime GOROOT, but the other _FINAL variables
aren't really needed.
(From OE-Core rev: 31aa0d8a62be95d093d3c7581efa41f74b9131ad)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Reorganize the Go toolchain build to split out
the Go standard runtime libraries into a separate
recipe. This simplifies the extension to crosssdk
and cross-canadian builds.
* Adds a patch to the go build tool to prevent it
from trying to rebuild anything in GOROOT, which
is now resident in the target sysroot.
* 'go' bb and inc files are now for building the
compiler for the target only.
* 'go-cross' bb and inc files are now just for
the cross-compiler.
* Adds virtual/<prefix> PROVIDES for the compiler
and runtime
* Removes testdata directories from the sysroot
during staging, as they are unnecessary and
can cause strip errors (some of the test files
are ELF files).
* Re-enables pacakage QA checks, adding selective
INSANE_SKIP settings where needed.
(From OE-Core rev: d2a7af7cd834e218c39d43ee3fa0c14d4f748727)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This reverts commit e0ed52c514.
Commit e0ed52c514 ('kernel-module-split: rrecommend kernel-image instead
of rdepend') changed kernel modules to rrecommend kernel-image instead
of rdepend on kernel. This broke existing setups, where the kernel is
omitted by setting RDEPEND_kernel-base = "".
Revert the patch, as the existing way of omitting kernel-image in images
works just fine.
(From OE-Core rev: c315b1e036b1252f35abda921c49327c950a9570)
Signed-off-by: Martin Hundebøll <mnhu@prevas.dk>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When building the per-recipe sysroot keep track of what files we're installing
and where they came from, so we can detect when a file is installed by two
different recipes and tell the user what these recipes are (instead of just
showing a os.link() stack trace).
[ YOCTO #11631 ]
(From OE-Core rev: 606a8e0ca218f023e362c3678122d36d537f95de)
(From OE-Core rev: 2ebbeb61114e4b847e9164c621ac87b5cf03a299)
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When using the Go 'vendor' mechanism to bring in
dependencies for a Go package, the default GO_INSTALL
setting, which uses the '...' wildcard, will include
the vendored packages in the build, which produces
incorrect results.
There are also some Go packages that are structured
poorly, so that the '...' wildcard results in building
example or test code that should not be included in
the build, or fail to build.
This patch adds a mechanism for filtering out a
subset of the sources. It defaults to filtering
out everything under the 'vendor' subdirectory
under package's main directory, which is the
normal location for vendored packages, but can
be overridden by a recipe to filter out other
subdirectories, if needed.
(From OE-Core rev: 9819353726d85780546158428bd97a253705017d)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Go source trees have a particular structure, with all
sources located under ${GOROOT}/src/<import-path>.
The fetcher step implemented by the 'go get' command
automatically follows this structure, so we need
to do the same here.
Since most Go packages are hosted in git repositories,
this adds a custom do_unpack() function that sets
the destsuffix to match the expected directory structure,
for any git SRC_URIs that haven't had a destsuffix
explicitly set in the recipe.
This simplifies recipe writing for the most common
cases.
(From OE-Core rev: efcf6513b71021ea4bfe6fbaa326e6591dee487d)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Add a do_configure task to populate ${B} by symlinking
in the src subdirectory under ${S}, which lets us point
GOPATH at ${B}. This lets us take advantage of the
automatic directory creation and cleaning for do_configure.
This necessitates a change to do_install to split the
installation of the sources and built artifacts. Taking
advantage of some additional tar options, we can eliminate
the extra staging area and extra recursive chown command.
So overall efficiency should be improved.
(From OE-Core rev: c62a083306c26b7e4deca1ff41336bb6b33d5b3a)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since this is a class, it should follow the
class function export mechanism for its task
functions, and should set directory-related
flags for directories they need.
(From OE-Core rev: 0369a99ad1c9e3a9a6394c723461795460dc2c76)
Signed-off-by: Matt Madison <matt@madison.systems>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The previous logic applied a regex to TUNE_FEATURES which could
set the GOARM value to 7 incorrectly, for example when dealing
with an arm1176 core. Simplify to check for the presence of
"armv7" instead. At the same time add a check for "armv6" and
set GOARM to 6 in that case.
(From OE-Core rev: 07b60c15e9ef650940afdde37bf3f3b9c50a336d)
Signed-off-by: Will Newton <willn@resin.io>
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Go toolchain changes the installation path when building for the same
architecture as the build host. This was already been considered in
the GO_BUILD_BINDIR variable but was not being used by the go class.
This fixes following error:
,----
| ERROR: go-dep-0.3.0-r0 do_package: QA Issue: go-dep: Files/directories
| were installed but not shipped in any package:
| /usr/lib/x86_64-oel-linux/go/bin/dep
| Please set FILES such that these items are packaged. Alternatively if
| they are unneeded, avoid installing them or delete them within
| do_install.
`----
(From OE-Core rev: 56ce355b3d775e801d2ca89ee812571e794311cd)
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The d.getVar has the second argument as True by default, avoid passing
it here.
(From OE-Core rev: 1a5026db41929d42bece22bd0ae60c13219a98f5)
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The update-alternatives where using relative links so not being really
in use since December 2016 (see OE-Core:c7bc46b9 "kernel: Fix
symlinks") so instead we now generate the relative symlinks during the
do_install task and drop the update-alternatives use at all.
Acked-by: Saul Wold <sgw@linux.intel.com>
(From OE-Core rev: 10a1b293191268e6792ac8e27bd6427f1974c7ce)
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Current implementation does not handle possible exceptions coming from child
processes, the latter responsible for creating packages. With the aim to have more
control, use pipes to communicate exceptions and stop package creation in case
of failure.
Helps to debug [YOCTO #12012].
(From OE-Core rev: 11350a67ba137f560d04aa643ff500a7ff112c73)
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Redirecting stderr to stdout helps debugging issues, i.e instead of just
getting the return code, get also the error log from the pkg manger
This commit is in the way to figure out the root cause of [YOCTO #12012],
where dpkg-deb fails with a 2 return code and according to the man page,
there are multiple issues leading to the same code.
(From OE-Core rev: 9ff023fb26f5f0ce19e757beda00ccc32c009b21)
Signed-off-by: Leonardo Sandoval <leonardo.sandoval.gonzalez@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* QA check which aren't included in WARN_QA and ERROR_QA are shown
during the build only as NOTE message (not shown at all with default
knotty setting), so it might be surprising to see them later in qa.log
file
(From OE-Core rev: 35ab2c7b08359f22f74106339841f8134123adf4)
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
While poky master branch has been fixed so that all CVE patch files have
the:
CVE: CVE-2017-1234556
strings in the patch comments, many older versions of poky and other meta
layers are not, but the CVE patches quite often have the CVE id in the
patch file name.
If the CVE: string also found, there are no duplicates in the report.
(From OE-Core rev: 5ee5b0c66627c9e974c838b86e2e659c2f601f2a)
Signed-off-by: Mikko Rapeli <mikko.rapeli@bmw.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The do_rootfs log contains a number of unsatisfied package
recommendations. At the moment those are only visible when
reviewing the rootfs log.
This patch adds an extra check to surface any unsatisfied
recommendation as WARNINGS to the build output.
Enable this check with:
ROOTFS_POSTPROCESS_COMMAND += "rootfs_log_check_recommends;"
(From OE-Core rev: 9d049bf7941f30e35c51775684559e95185fba96)
Signed-off-by: Jose Alarcon <jose.alarcon@ge.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The top level LICENSE file is not actually a license, it refers
other licenses that are used by Bitbake and Meta-data. Relying
on this file could cause problems for recipes when this file
changes, which it is about to.
(From OE-Core rev: a1948ab38c9cb7f0b16cce9dadc03ae6e2fe44ad)
(From OE-Core rev: dd75644db2f662f0d88529a068bbfb599c9790a9)
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Perl dependencies may look as "Perl(Foo::Bar)", but dpkg does not
support the non-alphanumeric characters. There was already special
handling present for turning '(' and ')' into '__'. This change does
the same for ':'.
(From OE-Core rev: a34e397095a9c2f8d0af1168ceab295af659242d)
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The silent rpmdeps failures fixed in a preceeding commit mean we
need to rerun all packaging.
(From OE-Core rev: 77abe012c496a85f56191abc769a7da07e9f8c0e)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
There was a logic error in the dependency cleanup code which meant
it would remove dependencies which other tasks still depended upon.
Fix the path names so the comparisions work as intended.
This fixes dependencies accidentally disappearing from sysroots
under certain reconfiguration situations.
(From OE-Core rev: 1634fe5148b3501f2c1b75cf7fb704a2ef60424e)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
In OE-Core commit a48fea275b, a check on
the value of a "sha256" parameter was added, however there was no
mention of this in the commit message and no corresponding code to
actually verify the checksum as sha256 was added along with it either,
so there's no point in getting the value. Additionally it was assuming
that a sha256 value would be present without checking first, with the
result that if you leave out the md5 value in a recipe intentionally in
order to get it to tell you the correct value on the next build, you got
a traceback instead of the appropriate error containing the information.
Drop this entirely - if we want to implement this we need to do it
properly.
(From OE-Core rev: e9eaa7d15fe7ab643ab19556dab84051f8f1974e)
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When compressing with "gzip", do not save the time stamp in
the compressed file metadata.
(From OE-Core rev: a88c16c2dc5dd7e2fbf7ca908a8b48da7be8b34c)
Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The lines in qemuboot.conf are in random order.
This patch fixes this by printing the lines in sorted order.
This makes it easier to compare two different builds for
any differences.
(From OE-Core rev: 0868edace0750862168abc8d504891646afc8c76)
Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
For URIs with the npm:// transport but with no other slash in it, the
common MIRRORS and PREMIRRORS pattern of npm://.*/.* fails to match.
Make the last slash in the pattern optional in the mirros.bbclass and
own-mirrors.bbclass classes.
Many URIs with the npm:// transport have no slash after the host part:
npm://registry.npmjs.org;name=foo;version=0.1.2
This means that MIRRORS and PREMIRRORS containing entries like the
first one will not match these URIs:
npm://.*/.* # fails to match
npm://.*/?.* # matches this and URIs with path components
For normal regular expressions, a pattern like 'npm://.*(/.*)?' would
probably be preferred, but that won't work here: the pattern gets split
into the substrings 'npm', '.*(' and '/.*)?', which are not valid
regular expressions individually.
(From OE-Core rev: 0d1e2b4507af28fc451b8fa94130a39ac342637d)
Signed-off-by: Olaf Mandel <o.mandel@menlosystems.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
gettext has a notoriously slow configuration step, and so in my testing
this greatly speeds up building core-image-minimal: from 21m36s to 19m2s
(empty sstate and tmp, but pre-populated downloads).
I have also built world, and core-image-sato to make sure it doesn't break
or modify the build, and there is no difference whatsoever in packages
and images content. Target gettext seems not to be used for anything.
Also fix up insane.bbclass to remove the corresponding QA check.
(From OE-Core rev: 1a7b843d575b290917d1e379c2ba106460988230)
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since we disabled uppercase characters in overrides a few releases
ago, uppercase characters in recipe names (and for that matter, distro
and machine names) cannot be supported due to their reliance upon
overrides including the name.
QA check will produce an warning message when it verify that recipe
name is uppercase.
[YOCTO# 11592]
(From OE-Core rev: 4713f8b2c4f2c74239d284adcf1e59e61aa66576)
Signed-off-by: Yeoh Ee Peng <ee.peng.yeoh@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently items are added to the sysroot, the obsolete items are removed. If
a change such as pkgconfig -> pkgconf is made, this leads to conflicts of
overlapping files in the sysroot.
In order to better support this, handle removing items before adding them.
This requires some minor refactoring to construct the installed list
before the main function loop, otherwise there are no changes in this
patch other than reordering the operations.
(From OE-Core rev: add4f107c151d32d9ea914bb0b93c3d3c17c776c)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Currently a task could remove a dependency needed by another task leading
to build failures, often due to missing dependencies (e.g. dynamic libraries
not being found). This was often seen for all-arch recipes in package_write_rpm.
When removing a dependency, first check that no other task active for the
recipe has that same dependency.
(From OE-Core rev: ff3617cc2cd5618f48a25aa4e3b2014430fcbe23)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
