Ed Bartosh [Wed, 28 Sep 2016 09:16:14 +0000 (12:16 +0300)]
runqemu: explicitly set image format
QEMU produces a warning if drive format is not specified:
WARNING: Image format was not specified for
'tmp/deploy/images/qemux86-64/core-image-minimal-qemux86-64.wic'
and probing guessed raw.
Automatically detecting the format is dangerous for raw images,
write operations on block 0 will be restricted.
Specify the 'raw' format explicitly to remove the restrictions.
Set image format to 'vmdk', 'qcow2' or 'vdi' for correspondent image
types. Set it to 'raw' for the rest of image types.
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Martin Jansa [Wed, 28 Sep 2016 07:28:49 +0000 (09:28 +0200)]
qemuboot: don't fail when QB_DEFAULT_KERNEL isn't symlink
* in some cases we might set QB_DEFAULT_KERNEL to the real filename
instead of symlink and then this whole readlink work around actually
breaks the build, because os.readlink fails on normal files:
Richard Purdie [Wed, 28 Sep 2016 08:16:46 +0000 (09:16 +0100)]
oeqa/sstatetests: Ensure we cover deb packaging backend for sstate test
Currently we weren't testing the deb backaned for sstate correctness
and there was a bug that had crept in. Ensure we cover all package
backends with the test regardless of what the distro/conf sets.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Wed, 28 Sep 2016 08:14:55 +0000 (09:14 +0100)]
dpkg: Only set DEB_HOST_ARCH in target case
If we don't do this, the sstate checksums vary for dpkg-native depending
on which MACHINE is set and this is clearly incorrect. It leads
to dpkg-native rebuilding far too often.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This function never worked because the SDK_OUTPUT and SDKPATH vars are
written bash-style in a python function. The only reason it never failed
a build is because the function bails out the start because of the flag
CHECK_SDK_SYSROOTS.
And I guess nobody tested with CHECK_SDK_SYSROOTS enabled until now.
Signed-off-by: Ioan-Adrian Ratiu <adrian.ratiu@ni.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This function was broken by the multi-config changes, and isn't needed anymore
now that recipeutils.pn_to_recipe can handle provides. Without this, the
newappend sub-command fails.
Signed-off-by: Christopher Larson <chris_larson@mentor.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This dependency was manually added in 3dec9ad1cd6a ("perl: module
overload rdpends on overloading") but was (mistakenly?) removed by 06d43a90acbe ("perl: 5.20.0 -> 5.22.0"). Restore it.
Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Since the upgrade to 2.8, lttng-tools' test harness silently succeeds
but doesn't actually run the tests. This is because upstream made
some changes in their test harness:
Updates to address this include:
- drop now-irrelevant patch
- change the ptest-run make target
- remove indiscriminate search/replace commands from do_install_ptest
- copy entire build directory into PTEST_PATH and then remove unneeded files
- use lttng binaries installed on the system for the tests
- add lttng-tools-ptest runtime dependencies
- lttng-tools itself
- babeltrace, used by the test harness to process traces
- perl modules required by babelstats.pl test script
- procps (for pgrep, pidof)
- gawk
- remove unnecessary chmod and munging of utils.sh script library
- remove checkpatch from ptest installation tree
- avoid path-munging of libtool artifacts altogether
- use more efficient find+sed patterns to munge Makefiles
- reduce test harness output to conform to ptest rules
On qemux86-64 and qemuarm I get relatively stable results, with
PASS/FAIL varying by +-1 on successive runs.
linux-yocto: Move PREFERRED_PROVIDER check to be generic
This check ensures that when the PREFERRED_PROVIDER for virtual/kernel
changes, the previous instances gets removed correctly so when the new
instance installs files into the shared area there is not an overlap of
old and new.
[YOCTO #10278]
Signed-off-by: Saul Wold <sgw@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
SDK: Allow changing SDKMACHINE without wiping TMP folder
When changing SDKMACHINE, we may encounter an error forcing us to wipe the TMP folder.
Since only SDK_ARCH is captured in the PN of the crosssdk recipes, changes to SDK_OS
result in conflicts. Eventually we hit the error:
ERROR: ...: The recipe <...> is trying to install files into a shared area when those files already exist.
The build has stopped as continuing in this scenario WILL break things
This patchset addresses the problem by SDK_SYS as the recipe name suffix instead
of SDK_ARCH.
[YOCTO #9281]
Signed-off-by: Juro Bystricky <juro.bystricky@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
By default, libsolv uses the rpm logic for version comparison, which is
not quite the same as debian. Opkg now sets the distribution type for
libsolv to be debian. But for that to work, libsolv needs to be compiled
with MULTI_SEMANTICS=ON.
Signed-off-by: Alejandro del Castillo <alejandro.delcastillo@ni.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
parselogs.py: Add amd_nb error to x86_common whitelist
This has always silently failed on hardware without AMD Northbridge,
and a recent kernel patch made it not silent. It would be ideal to only
whitelist the error for genericx86 MACHINEs and disable the CONFIG
option that enables it in intel-* MACHINEs, but in order to disable
this configuration option we would have to enable EXPERT and
DEBUG_KERNEL, which we don't want. Instead just whitelist it on all
x86 MACHINEs.
Fixes [YOCTO #10261].
Signed-off-by: California Sullivan <california.l.sullivan@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Maxin B. John [Sun, 25 Sep 2016 18:40:05 +0000 (21:40 +0300)]
gst-player: Disable visualizations
On some machines, visualizations in gst-player trigger a bug in
xvimagesink. Till we have a proper fix, disable the visualization
rather than downgrading the xvimagesink.
Fixes [YOCTO #10041]
Signed-off-by: Maxin B. John <maxin.john@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Backport patch to fix CVE-2016-1000110 from python upstream:
for python2.7
https://hg.python.org/cpython/rev/ba915d561667/
for python3
https://hg.python.org/cpython/rev/a0ac52ed8f79
Signed-off-by: Mingli Yu <Mingli.Yu@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Patrick Ohly [Mon, 26 Sep 2016 09:55:16 +0000 (11:55 +0200)]
archive.bbclass: fix do_ar_original archiving of multiple source repos
When a recipe uses more than one source which isn't a plain file (for
example, multiple git repos), then do_ar_original created the source
archives using the same filename and thus only archived one source.
The "name" parameter is used as file suffix to create unique names for
each source, leading to archives following this pattern:
deploy/${TARGET_SYS}/${PF}/${PF}[-<name>].tar.gz.
The ${PF} part is a bit redundant, which may or may not be
desirable. The patch is more localized this way (no need to modify
create_tarball()).
Patrick Ohly [Mon, 26 Sep 2016 09:55:15 +0000 (11:55 +0200)]
archiver.bbclass: ignore unpack sub-directories in do_ar_original
Support for absolute paths in the "subdir" parameter was recently
added (bitbake rev: c3873346c6fa). The git fetcher has supported
absolute paths in "destsuffix" already before.
When the path is absolute as in destsuffix=${S}/foobar, the tmpdir
used by do_ar_original gets ignored, which breaks:
- source code archiving (tmpdir is empty)
- compilation due to race conditions (for example, ${S} getting
modified by do_ar_original while do_compile runs)
To solve this, these parameters get removed from URLs before
instantiating the fetcher for them.
This is done unconditionally also for relative paths, because these
paths are not useful when archiving the original source (upstream
source does not have them, they only get used by the recipe during
compilation).
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
* Add PACKAGECONFIG to add proper dependencies for:
graph_parallel, locale, and mpi.
* boost-mpi depends on mpich which is in meta-oe,
and boost-graph_parallel depends on boost-mpi,
so they are disabled by default, but can be enabled
in a distro that needs them.
* context and coroutine are added only for x86 and powerpc.
Signed-off-by: Jackie Huang <jackie.huang@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Joe MacDonald [Mon, 26 Sep 2016 13:17:08 +0000 (09:17 -0400)]
devtool: Add a line break to generated README
When devtool creates a new workspace, it produced a README with one very
long line and no space following 'bblayers.conf'. Add a line break as was
intended.
Signed-off-by: Joe MacDonald <joe_macdonald@mentor.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Mon, 26 Sep 2016 16:25:26 +0000 (17:25 +0100)]
prelink: Manipulate library paths to match the target system library setup
Currently, prelink doesn't work unless base_libdir/libdir match
its hardcoded values. This patch manipulates those paths so that
they match the values set in the variables and handles multilib
configurations too. The manipulations only happen in the target
case, if needed.
[YOCTO #10282]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Mon, 26 Sep 2016 16:22:24 +0000 (17:22 +0100)]
utils: Add all_multilib_tune_list function
Its useful to be able to query a list of variables to obtain the values
in each multilib context. This adds such a function which works even
if called in the non-default recipe context.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
machine/qemu*: Add comment regarding the reason for virtio-rng-pci
Bring across the comment that was in runqemu regarding why the
virtio-rng-pci device was needed. This comment is added to each location
where the virtio-rng-pci device is added.
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
meta-environment: ensure corret TOOLCHAIN_CONFIGSITE_NOCACHE value
Expand TOOLCHAIN_CONFIGSITE_NOCACHE value immediately before inheriting
cross-canadian to avoid HOST_ARCH being changed from TARGET_ARCH to SDK_ARCH,
thus ensuring its correct value.
[YOCTO #10255]
Signed-off-by: Chen Qi <Qi.Chen@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Otavio Salvador [Fri, 23 Sep 2016 16:42:59 +0000 (13:42 -0300)]
qemu.bbclass: Allow use different QEMU binary depending of target
There are architectures which support running in 32 and 64 bit
flavours however the simulation is provided in a specific QEMU
setting, requiring us to use a different binary. This patch allow this
to be done using, for example:
QEMU_TARGET_BINARY_ppce5500 = "qemu-ppc64abi32"
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Mark Hatle [Thu, 22 Sep 2016 21:37:20 +0000 (16:37 -0500)]
pseudo: Fix problem where pseudo could kill a container init
In a heavily loaded container, the child process might not started
before the parent process had terminated. The child process attempts to
signal the parent with SIGUSR1. If the parent had terminated, the
parent becomes PID 1, which is generally init. When it signaled pid 1,
it caused the docker mini-init to terminate.
This doesn't happen in a traditional system, as systemd/sysvinit is
protected to only root users can signal it.
[YOCTO #10324]
Signed-off-by: Mark Hatle <mark.hatle@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
scripts/runqemu: provide better error message on runqemu ifup fail
If runqemu-ifup fails hen running testimage, a rather cryptic error
regarding "no tty present" is displayed. If this step fails, we
should at least point the user at runqemu-gen-tapdevs. A quick search
of this term in the manual will lead them to "Enabling Runtime Tests
on QEMU" which should give them all the info they need.
Signed-off-by: Stephano Cetola <stephano.cetola@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ed Bartosh [Fri, 23 Sep 2016 12:10:26 +0000 (15:10 +0300)]
wic: remove partition images
Preserving images for every partition doubles disk space
consumed by an image build. As those images are not used,
so it's better to remove them after assembling final image.
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Patrick Ohly [Fri, 23 Sep 2016 13:23:20 +0000 (15:23 +0200)]
openssl.inc: avoid random ptest failures
"make alltests" is sensitive to the timestamps of the installed
files. Depending on the order in which cp copies files, .o and/or
executables may end up with time stamps older than the source files.
Running tests then triggers recompilation attempts, which typically
will fail because dev tools and files are not installed.
"cp -a" is not enough because the files also have to be newer than
the installed header files. Setting the file time stamps to
the current time explicitly after copying solves the problem because
do_install_ptest_base is guaranteed to run after do_install.
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Patrick Ohly [Fri, 23 Sep 2016 13:26:05 +0000 (15:26 +0200)]
openssl: update to 1.0.2i (CVE-2016-6304 and more)
This update fixes several CVEs:
* OCSP Status Request extension unbounded memory growth (CVE-2016-6304)
* SWEET32 Mitigation (CVE-2016-2183)
* OOB write in MDC2_Update() (CVE-2016-6303)
* Malformed SHA512 ticket DoS (CVE-2016-6302)
* OOB write in BN_bn2dec() (CVE-2016-2182)
* OOB read in TS_OBJ_print_bio() (CVE-2016-2180)
* DTLS buffered message DoS (CVE-2016-2179)
* DTLS replay protection DoS (CVE-2016-2181)
* Certificate message OOB reads (CVE-2016-6306)
Of these, only CVE-2016-6304 is considered of high
severity. Everything else is low. CVE-2016-2177 and CVE-2016-2178 were
already fixed via local patches, which can be removed now.
See https://www.openssl.org/news/secadv/20160922.txt for details.
Some patches had to be refreshed and one compile error fix from
upstream's OpenSSL_1_0_2-stable was required. The server.pem
file is needed for test_dtls.
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Fri, 23 Sep 2016 09:22:11 +0000 (21:22 +1200)]
lib/oe/patch: improve accuracy of patch header extraction
When PATCHTOOL = "git", if we need to manually apply a patch and then
commit it (i.e. when git am doesn't work) we try to extract the author /
date / shortlog from the patch header. Make the following improvements
to that extraction process:
* If there's no explicit Subject: but the first line is followed by a
blank line, isn't an Upstream-Status: or Index: marker and isn't too
long, then assume it's good enough to be the shortlog. This avoids
having too many patches with "Upgrade to version x.y" as the shortlog
(since that is often when patches get added).
* Add --follow to the command we use to find the commit that added the
patch, so we mostly get the commit that added the patch rather than
getting stuck on upgrade commits that last moved/renamed the patch
* Populate the date from the commit that added the patch if we were able
to get the author but not the date from the patch (otherwise you get
today's date which is less useful).
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Fri, 23 Sep 2016 09:22:10 +0000 (21:22 +1200)]
lib/oe/patch: exclude "From <hash>" from commit message when PATCHTOOL is "git"
If you leave "From <hash>" lines in the commit message it can actually
break git rebase because it tries to interpret the line in the context
of the current repository, and if the hash is invalid then a rebase
will blow up with:
fatal: git cat-file: could not get object info
or in newer git versions:
error: unable to find <hash>
fatal: git cat-file <hash>: bad file
(I hit this when I tried to do a devtool upgrade on openssl to 1.0.2i
the first time I did "git rebase --skip")
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Fri, 23 Sep 2016 09:17:52 +0000 (21:17 +1200)]
nativesdk-qemu-helper: drop old Zaurus flash tools
These tools are relics of an earlier time when the Zaurus devices were
reference platforms - these days they are no longer needed. It seems
amazing that they survived earlier purges.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Robert Yang [Thu, 21 Jul 2016 03:35:53 +0000 (20:35 -0700)]
systemd: install udev.pc
It provides udev, but doesn't install udev.pc, which causes other
recipes failed to figure out udevdir.
Fixed when systemd in DISTRO_FEATURES:
$ bitbake pcmciautils (or btrfs-tools):
Package udev was not found in the pkg-config search path.
Perhaps you should add the directory containing `udev.pc'
to the PKG_CONFIG_PATH environment variable
No package 'udev' found
Their udev rules file may not be installed according to each pkg's
implementation.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
We still have problems where deploying SDKMACHINE=i686 can cause removal
of SDKMACHINE=x86_64 artefacts.
The reason is that x86_64 is a BUILD_ARCH as well as an SDK_ARCH and
the manifest namespaces overlap. To fix this, set PACKAGE_ARCH and
the stamp-extra-into to include SDK_OS. SDK_OS may not be entirely correct
but it is what sstate.bbclass uses for nativesdk and fixing that is
a separate issue.
This is confirmed to resolve artefact problems on the AB which have been
delaying a new uninative release.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Jose Lamego [Wed, 21 Sep 2016 19:54:59 +0000 (12:54 -0700)]
oeqa/selftest/base: backup and restore local configuration files
Selftests' cleanup method during test setup is not capable of
restoring local configuration files that remain modified after
aborting a test through a keyboard interruption.
This change creates backups for local.conf and bblayers.conf at
test setup, restore them when found, and deletes them at cleanup.
[YOCTO #9390]
Signed-off-by: Jose Lamego <jose.a.lamego@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
The feature to install packages in the target requires to
build the package manager. It would fail, with very obtuse
errors, if a test requires to install something and the
package manager hasn't been build. This will add the package
manager as dependency for testimage.
[YOCTO #10260]
Signed-off-by: Mariano Lopez <mariano.lopez@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
scripts/runqemu: Using a cpio* rootfs has no special network
When booting a system with the rootfs being of cpio* type the networking
setup should still work the same as for all other root filesystem types.
This change removes the clearing of the NETWORK_CMD variable allowing
for the slirp/tap setup to be provided to QEMU.
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Not all QEMU machines (outside of those available in OE-Core) are
capable of using the virtio-rng-pci device due to various machine models
not having a pci/virtio bus. This makes it such that the use of the
'-device virtio-rng-pci' flag to QEMU is machine specific.
This patch removes the general addition of the flag to all runqemu
targets and adds the flag into the QB_OPT_APPEND for all the qemu*
machines in OE-Core that support its use (which is all of them).
Signed-off-by: Nathan Rossi <nathan@nathanrossi.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Zhixiong Chi [Thu, 22 Sep 2016 07:54:27 +0000 (15:54 +0800)]
wpa_supplicant: Security Advisory-CVE-2016-4477
Add CVE-2016-4477 patch for avoiding \n and \r characters in passphrase
parameters, which allows remote attackers to cause a denial of service
(daemon outage) via a crafted WPS operation.
Patches came from http://w1.fi/security/2016-1/
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Zhixiong Chi [Thu, 22 Sep 2016 07:54:20 +0000 (15:54 +0800)]
wpa_supplicant: Security Advisory-CVE-2016-4476
Add CVE-2016-4476 patch for avoiding \n and \r characters in passphrase
parameters, which allows remote attackers to cause a denial of service
(daemon outage) via a crafted WPS operation.
Patches came from http://w1.fi/security/2016-1/
Signed-off-by: Zhixiong Chi <zhixiong.chi@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Currently adding a font recipe to an image doesn't have enough
dependencies to cause mkfontdir-native to be included in the native
sysroot. This creates problems with the postinstall scripts for fonts
which call mkfontdir to create font index files: font.dir. The end
result is missing font.dir files in the built image and a malfunctioning
font system.
Dependencies for the relevant recipes currently look like this:
Where:
* D = DEPENDS
* R = RDEPENDS
* RN = RDEPENDS_class-native
* <font.bb> e.g. font-adobe-100dpi*.bb
Some details where omitted for clarity e.g. dependencies on util-macros.
I believe the intent behind the RDEPENDS_class-native chain:
* <font.bb> -> font-util-native -> mkfontdir-native
was to provide the necessary dependency on mkfontdir. However because
the native sysroot is not built from packages this RDEPENDS_class-native
chain doesn't have the desire effect (i.e. it doesn't pull in
mkfontdir-native).
Changing the RDEPENDS_class-native chain into a DEPENDS_class-native
chain is a non-starter because of the build time dependency loop it
creates:
* font-util-native -> mkfontscale-native -> libfontenc-native -> font-util-native
Having upstream remove the build time dependency of libfontenc on
font-util is also a non-starter[1] even though it does create problems
in other distributions, for example on Debian see [2], [3].
Instead add a DEPENDS on mkfontdir-native in the encodings recipe in
addition to the mkfontscale-native dependency it already contains. This
solves the missing mkfontdir in the native sysroot problem without
introducing a build dependency loop.
buildtools-tarball shouldn't be regenerated when MACHINE changes,
nor should variants for other SDKMACHINE be removed from the deploy
directory when SDKMACHINE changes.
Remove target architecture dependencies so that deploy artefacts
can overlap.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
toolchain-scripts-base: add base class for toolchain_create_sdk_version
We use toolchain_create_sdk_version() in buildtools-tarball but
don't want the extra classes toolchain-scripts pulls in, therefore
split out a separate base class for this function which both
toolchain-scripts and the buildtools-tarball can inherit.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Wed, 21 Sep 2016 21:30:03 +0000 (22:30 +0100)]
base.bbclass: Drop unnecessary dirs setting
${D} is listed in cleandirs so no need to list it in dirs as well.
The default directory is ${B} so this is a cleanup which should have
no changes to the execution.
[YOCTO #10017]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Wed, 21 Sep 2016 10:53:57 +0000 (11:53 +0100)]
autotools/siteinfo: Tweak CONFIG_SITE handling for determism/races
As things stand there are multiple races in the CONFIG_SITE handling
where checksums can change depending on whether site directories
exist or not when parsing happens. This is bad.
Secondly, there is a build race that occurs if you build virtuals
in parallel with the "main" recipe, since the main recipe is parsed
when the virtual is (since it sets variables like BBCLASSEXTEND)
and with the current code, it may look for files and directories
which could be created/destroyed which the loop is executing. This
is also bad.
The aclocal-copy directory should only ever be accessed by the call
from autotools.bbclass. This changes the parameter name to make it
clear and ensures all callers have the right usage, neatly avoiding
all the problems above. Also added better comments.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Robert Yang [Tue, 20 Sep 2016 10:32:57 +0000 (03:32 -0700)]
asciidoc: set CLEANBROKEN to fix rebuild
The make clean removes doc/a2x.1 and doc/asciidoc.1, then it would cause
build failures since in the second build:
Fixing CONF_DIR in asciidoc.py
Fixing CONF_DIR in a2x.py
python a2x.py -f manpage doc/a2x.1.txt
a2x: ERROR: "xmllint" --nonet --noout --valid "/path/to/asciidoc-native/8.6.9-r0/asciidoc-8.6.9/doc/a2x.1.xml" returned non-zero exit status 4
make: *** [doc/a2x.1] Error 1
The xmllint failed because "--nonext" is used:
I/O error : Attempt to load network entity http://www.oasis-open.org/docbook/xml/4.5/docbookx.dtd
/buildarea/lyang1/test_arm/tmp/work/x86_64-linux/asciidoc-native/8.6.9-r0/asciidoc-8.6.9/doc/a2x.1.xml:2: warning: failed to load external entity "http://www.oasis-open.org/docbook/xml/4.5/docbookx.dtd"
Avoid running make clean will fix the problem.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
When built on an i686 host for qemux86-64 without the
fix to obey LD and it fails:
/scratch/dogwood/toolchains/x86_64/bin/i686-pc-linux-gnu-ld:
Relocatable linking with relocations from format elf64-x86-64
(/scratch/dogwood/perf-ld-test/build/tmp/work/qemux86_64-mel-linux/perf/1.0-r9/perf-1.0/fs/fs.o)
to format elf32-i386 (/scratch/dogwood/perf-ld-test/build/tmp/work/qemux86_64-mel-linux/perf/1.0-r9/perf-1.0/fs/libapi-in.o)
is not supported
This is because LD includes HOST_LD_ARCH, which contains TUNE_LDARGS,
which is -m elf32_x86_64 for x86_64. Without that, direct use of ld will fail.
Signed-off-by: Christopher Larson <chris_larson@mentor.com> Signed-off-by: Sujith Haridasan <Sujith_Haridasan@mentor.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Andreas Müller [Wed, 21 Sep 2016 05:30:43 +0000 (07:30 +0200)]
libunwind: fix build by linking with bfd instead of gold
works around:
<native-sysroot>/ld: error: Gperf-simple.o: cannot make copy relocation for protected symbol '_Uarm_local_addr_space', defined in ../src/.libs/libunwind-arm.so
collect2: error: ld returned 1 exit status
Makefile:1038: recipe for target 'Gperf-simple' failed
make[1]: *** [Gperf-simple] Error 1
make[1]: *** Waiting for unfinished jobs....
<...>
<native-sysroot>/ld: error: Lperf-simple.o: cannot make copy relocation for protected symbol '_ULarm_local_addr_space', defined in ../src/.libs/libunwind.so
collect2: error: ld returned 1 exit status
Makefile:1094: recipe for target 'Lperf-simple' failed
make[1]: *** [Lperf-simple] Error 1
<...>
ERROR: oe_runmake failed
<native-sysroot>/ld: error: Gperf-trace.o: cannot make copy relocation for protected symbol '_Uarm_local_addr_space', defined in ../src/.libs/libunwind-arm.so
collect2: error: ld returned 1 exit status
Makefile:1042: recipe for target 'Gperf-trace' failed
make[1]: *** [Gperf-trace] Error 1
<native-sysroot>/ld: error: Lperf-trace.o: cannot make copy relocation for protected symbol '_ULarm_local_addr_space', defined in ../src/.libs/libunwind.so
collect2: error: ld returned 1 exit status
Makefile:1098: recipe for target 'Lperf-trace' failed
make[1]: *** [Lperf-trace] Error 1
<...>
<native-sysroot>/ld: error: test-coredump-unwind.o: cannot make copy relocation for protected symbol '_UCD_accessors', defined in ../src/.libs/libunwind-coredump.so
collect2: error: ld returned 1 exit status
Makefile:1186: recipe for target 'test-coredump-unwind' failed
Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Tried by adding CFLAGS_append = " -fpic " to
the recipe. But that couldn't help resolve the
warning message:
x264/r2491+gitAUTOINC+c8a773ebfc-r0/packages-split/x264/usr/lib/libx264.so.144' has relocations in .text [textrel]
It was found that this warning is emitted because of the
assembly files in the source code. And it is not easy to
get rid of TEXTREL's which are coming from the assembly
source files.
Adding textrel to INSANE_SKIP resolves this issue.
This issue was observed in cyclone5 and imx6qsabresd BSP's.
So generalizing the patch.
The relocation script of the SDK doesn't have enough
error handling when replacing host perl with SDK perl
or changing the symlinks. This will add those checks
along with a sanity check of xargs.
[YOCTO #10114]
Signed-off-by: Mariano Lopez <mariano.lopez@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Andreas Müller [Wed, 21 Sep 2016 20:46:38 +0000 (22:46 +0200)]
libnewt: link whiptail properly with libnewt
Configuration failed to detect gold as GNU linker. It was searching for
'GNU ld' but gold returns 'GNU gold (GNU Binutils 2.27.0.20160806) 1.12' which
does not match. When not linking by GNU linker Makefile did some magic link
target alignment:
| WARNING: libnewt-0.52.19-r0 do_package_qa: QA Issue: /usr/bin/whiptail contained in package whiptail requires libnewt.so, but no providers found in RDEPENDS_whiptail? [file-rdeps]
Signed-off-by: Andreas Müller <schnitzeltony@googlemail.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Richard Purdie [Wed, 21 Sep 2016 23:24:03 +0000 (00:24 +0100)]
uninative-tarball: Make stamp independent
The uninative tarball only contains nativesdk compoents. It should
not get regenerated when MACHINE changes for example. Currently its
sstate arch is also incorrect so changing SDKMACHINE results in other
variants being removed from the deploy directory.
This patch removes the target architecture dependencies so that
deploy artefacts can overlap and it doesn't continually rebuild. This
also fixes various autobuilder/release artefact issues we're having
as a result of these issues.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If DEPLOY_DIR_IMAGE doesn't exist during check_arg_machine() we
will attempt to guess a suitable value later when check_and_set()
calls validate_paths(), therefore this shouldn't raise an exception
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
runqemu: don't try and invoke bitbake when running in a toolchain env
If a MACHINE value is passed we can't validate it by running bitbake
as the toolchain environment doesn't include the build system, we
must assume that the passed value for MACHINE is correct.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
runqemu: try and guess qemu-system binary when MACHINE isn't set
Emulate some logic from the prior, shell based, version of runqemu
to try and infer the correct setting for MACHINE from the kernel
and rootfs filenames.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
runqemu: validate paths and attempt to infer unset paths
We need to validate and ensure all paths are set regardless of
whether runqemu was invoked with a .qemuboot.conf file or
otherwise. Split this logic out into a separate method called
during check_and_set()
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Tue, 20 Sep 2016 15:16:38 +0000 (11:16 -0400)]
linux-yocto/4.8: introduce preempt-rt
Paul Gortmaker has made the preempt-rt patch available for the
4.8 kernel.
This commit merges his queue to standard/preempt-rt/rebase, which
will be kept up to date with a clean history, and it also makes
it available in standard/preempt-rt/base, which will be kept fast
forward for board support.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Bruce Ashfield [Tue, 20 Sep 2016 15:16:36 +0000 (11:16 -0400)]
linux-yocto/4.1: 4.1.32 content and configuration warning fixes
The LINUX_VERSION was previously updated to 4.1.32, but the
SRCREVs for the actual content were missed. This gets our actual
version and the PV back in sync.
We also update the meta data to fix configuration audit warnings
from the beaglebone builds.
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
squash with 4.1
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
