| Age | Commit message (Collapse) | Author |
|---|
|
This is a special image for profiling Python 3.x in order to utilize
profile-guided-optimization. Profile data can be obtained by by running
bitbake python-pgo-image -c profile. Profile data will be copied into a
directory pointed by PYTHON3_PROFILE_DIR on the host system. The profile
task may be specified with PYTHON3_PROFILE_TASK.
The default profile task is pybench, although an example for using
test.regrtest is included in the recipe.
[YOCTO #9338]
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
|
|
This patch adds a new recipe that builds a special version of python3 that
produces profile data used for optimization. The new recipe directly includes
the base python recipe so that they are build in as similar way as possible and
this hopefully decreases the recipe maintenance burden, too. Also, its files
clash with the "normal" python3 so they cannot be installed in an image at the
same time.
Normally, profile-guided-optimization in Python is done simply by doing "make
profile-opt" which first builds python with profile instrumentation enabled,
then runs a profile task to get the profile data, and last, re-builds python
with profile data guiding the optimization. However, in our cross-build
environment this gets a lot trickier. We need to split out the steps as
building is done on the build host but we need to run the second step (i.e. run
the profile task) on the target hardware. This patch enables the first step,
i.e. building python with profile instrumentation enabled
[YOCTO #9338]
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
|
|
Similar to an earlier fix for Python 2.7. Make the tests subpackage
depend on all modules as test.regrtest uses most (if not all) of them.
Some tests also depend on libgcc so add that as a runtime dependency as
well.
[YOCTO #9338]
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
|
|
Make it possible to build python3-native with profile directed optimization
enabled. The feature is enabled by specifying
PYTHON3_NATIVE_PROFILE_OPT = "1" in local.conf.
The profile task to be run may be specified with PYTHON3_NATIVE_PROFILE_TASK
variable in local.conf, e.g.
PYTHON3_NATIVE_PROFILE_TASK = "${S}/Tools/pybench/pybench.py -n 2 --with-gc --with-syscheck"
[YOCTO #9338]
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
|
|
It was required only by dnf, which has switched to official gpgme bindings.
pygpgme itself is old and unmaintaned.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Drop dependency on pygpgme, replace it with gpgme's own bindings.
Add a patch that fixes an upstream regression.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Rebase the patches.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Make libxslt-native dependency conditional on api-documentation distro feature,
as vala now defaults to rebuilding the manual (which is slow).
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
It's not actually required; host python is fine.
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
We only build one gcc-cross per architecture and having target specific
flags means gcc-cross would rebuild. These flags are really for the
on target case, so they should be set in gcc-target only.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Both python-git and python3-git need to be upgraded to latest
upstream
version.
This change was tested using qemux86 with core-image-sato.
Signed-off-by: Jose Lamego <jose.a.lamego@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Backport fix for gcc7 warning along the way
Remove --disable-rpath as this configure doesn't support it anymore [RB].
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Both native and target versions of this file reference mkdir and install
in hosttools paths. Use the version from PATH instead.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Currently the file encodes full paths to various host tools in the
HOSTTOOLS directory which is bad in native and target cases. We can
simply use the versions from PATH quite safely in OE.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
A number of features were disabled in OE due to historic problems
on ARM. In tcf-agent 1.4, they compile cleanly on all architectures
besides mips and libc-musl. These features are required for target
debugging with Eclipse Neon, CDT and TCF plugins - for example used
by Xilinx Vivado SDK 2016.4.
For MIPS and libc-musl, at least re-enable SERVICE_Symbols and
SERVICE_LineNumbers which compile cleanly.
Signed-off-by: Gernot Hillier <gernot.hillier@siemens.com>
Signed-off-by: Tim Offermann <tim.offermann@siemens.com>
Acked-by: Eugene Tarassov <eugene.tarassov@xilinx.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
tcf-agent uses "a64", not "aarch64", see agent/Makefile.inc. This fix
allows to re-enable features like SERVICE_Breakpoints and SERVICE_
MemoryMap on arm64.
Signed-off-by: Gernot Hillier <gernot.hillier@siemens.com>
Acked-by: Eugene Tarassov <eugene.tarassov@xilinx.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Many new users add "python3" to their image using IMAGE_INSTALL and are then
surprised that they can't import any of the standard library. This is because we
split up the standard library into a number of packages, and python3-core (the
interpreter and essential modules only) RPROVIDES 'python3'.
Solve this by moving the RPROVIDES of 'python3' to python3-modules, so that the
entire Python standard library is installed.
[ YOCTO #11182 ]
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Add function to remove -mcpu compiler options on arm
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
The path to where to install and find the sysroot components is used
in many places. This warrants it to get its own variable.
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
These tools are not currently used for anything, but we should
still provide working versions of them.
[YOCTO #11400]
Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
Backport fixes from pseudo master for an acl issue and more importantly, a segfault
issue with bash which can be triggered by the recent useradd changes.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
nativesdk builds now control the DISTRO_FEATURES (oe-core 731744) so this
workaround is no longer required.
This reverts commit 415b72ffcbd26e5f3664370d8b2a9b8105fb6342.
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
It should be ${libdir} rather than /usr/lib, otherwise it would fail
when multilib:
MACHINE = "qemux86-64"
require conf/multilib.conf
MULTILIBS = "multilib:lib32"
DEFAULTTUNE_virtclass-multilib-lib32 = "x86"
$ bitbake apt
[snip]
install: target /path/to/apt/1.2.12-r0/image/usr/lib/apt is not a directory: No such file or directory
[snip]
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Currently there are multiple issues with useradd:
* If base-passwd rebuilds, it wipes out recipe specific user/group additions
to sysroots and causes errors
* If recipe A adds a user and recipe B depends on A, it can't see any of the
users/groups A adds.
This patch changes base-passwd so it always works as a postinst script
within the sysroot and copies in the master files, then runs any
postinst-useradd-* scripts afterwards to add additional user/groups.
The postinst-useradd-* scripts are tweaked so that if /etc/passwd doesn't exist
they just exit, knowning they'll be executed later. We also add a dummy entry to
the dummy passwd file from pseudo so we can avoid this too.
There is a problem where if recipe A adds a user and recipe B depends on A but
doesn't care about users, it may not have a dependency on the useradd/groupadd
tools which would therefore not be available in B's sysroot. We therefore also
tweak postinst-useradd-* scripts so that if the tools aren't present we simply
don't add users. If you need the users, you add a dependency on the tools in the
recipe and they'll be added.
We add postinst-* to SSTATE_SCAN_FILES since almost any postinst script of this
kind is going to need relocation help.
We also ensure that the postinst-useradd script is written into the sstate
object as the current script was only being added in a recipe local way.
Thanks to Peter Kjellerstedt <pkj@axis.com> and Patrick Ohly for some pieces
of this patch.
[Yocto #11124]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Make sure that native sysroot contains qemu and tunctl binaries for
runqemu usage:
- excluded native sysroot from rm_work
- added qemu-native to DEPENDS to put qemu binaries into native sysroot
- forced addto_recipe_sysroot task
[YOCTO #11266]
[YOCTO #11193]
Signed-off-by: brian avery <brian.avery@intel.com>
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
"yocto-compat-layer.py --machines" showed that shared packages like
gcc-cross-powerpc64 have a sstate signature that depends on
TUNEFLAGS. As a result, there are unnecessary rebuilds and potential
conflicts in a multiconfig.
That's due to the way how TARGET_ARCH is set. Richard Purdie suggested
setting TARGET_ARCH[vardepvalue] as fix, which works. It would be
shorter to do that in cross.bbclass instead of repeating the relevant
line in different recipes, but Richard was concerned about potential
side-effects in other usages of cross.bbclass.
TARGET_GOARM as used in go.inc is still causing signature differences
for go-cross-powerpc64 and machines b4420qds-64b and p5020ds-64b. This
needs further investigation.
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
gdb-cross used to be specific to the tune flags, but isn't
anymore. Therefore it is enough to use TARGET_SYS instead of
TUNE_PKGARCH to create a unique path.
Fixes a sstate signature difference that was found via
yocto-compat-layer.py's test_machine_signatures check. In practice it
probably showed up as unnecessarily rebuilding gdb-cross when
switching between machines like intel-corei7-64 and qemux86-64.
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
libgcc gets compiled differently depending on the tune flags for the
target. That dependency would make go-cross also tune specific and
prevent sharing it between different machines using the same
architecture.
For example, MACHINE=intel-corei7-64 and MACHINE=qemux86-64 shared the
same go-cross-x86_64, but compiled libgcc differently.
The libgcc dependency gets inherited from go.inc, but does not seem to
be necessary for go-cross (compiling go-helloworld still succeeds).
The dependency is left in go.inc conditionally, just in case that it
really is relevant for the various on-target recipes which inherit
that.
Because go-cross*.bb includes go*.bb, moving the DEPENDS into a .inc
file that only gets included for the target recipes doesn't
work. Reshuffling the content of three .bb files seems too intrusive
at this point.
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Backport upstream commit to address vulnerabilities:
CVE: CVE-2017-6965
[BZ 21137] -- https://sourceware.org/bugzilla/show_bug.cgi?id=21137
Fix readelf writing to illegal addresses whilst processing corrupt input
files containing symbol-difference relocations.
PR binutils/21137
* readelf.c (target_specific_reloc_handling): Add end parameter.
Check for buffer overflow before writing relocated values.
(apply_relocations): Pass end to target_specific_reloc_handling.
CVE: CVE-2017-6966
[BZ 21139] -- https://sourceware.org/bugzilla/show_bug.cgi?id=21139
Fix read-after-free error in readelf when processing multiple, relocated
sections in an MSP430 binary.
PR binutils/21139
* readelf.c (target_specific_reloc_handling): Add num_syms
parameter. Check for symbol table overflow before accessing
symbol value. If reloc pointer is NULL, discard all saved state.
(apply_relocations): Pass num_syms to target_specific_reloc_handling.
Call target_specific_reloc_handling with a NULL reloc pointer
after processing all of the relocs.
Signed-off-by: Yuanjie Huang <yuanjie.huang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Move config/Makefile in libdir from core package to dev package for
python, because it is only needed in development process.
Signed-off-by: Li Zhou <li.zhou@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Update Upstream-Status tags and apply 010-fix-rpmatch.patch
unconditionally, since it's merged unconditionally upstream.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The python-pycurl recipe can be used with python2 only even
though python3 is officially supported by upstream.
Create python3-pycurl recipe enabling the pycurl module for
python3.
Signed-off-by: Dmitry Rozhkov <dmitry.rozhkov@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
meta/conf/bitbake.conf puts python2.7 into the HOSTTOOLS variable but not
python2, so only python2.7 is guaranteed. In addition, on some distros -- such
as Amazon Linux -- /usr/bin/python2 doesn't exist but python2.7 does. So, use
python2.7 for the --python= argument in the qemu configure step.
Signed-off-by: Martin Kelly <mkelly@xevo.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Building gcc-cross-initial with GCC7 on the host fails due to the
comparison of a pointer to an integer in ubsan_use_new_style_p, which
is forbidden by ISO C++:
ubsan.c:1474:23: error: ISO C++ forbids comparison between pointer and
integer [-fpermissive]
|| xloc.file == '\0' || xloc.file[0] == '\xff'
Backport the fix from upstream GCC to enable the build with GCC 7
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Backport a fix from upstream for a -Wformat-truncation=2 warning
and implement a simple fix for a -Wimplicit-fallthrough warning.
Signed-off-by: Joshua Lock <joshua.g.lock@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The http.server module from python3-netclient imports the html module
which is in python3-html. Also xmlrpc.server imports pydoc which is a
part of python3-pydoc. But those run-time dependencies are missing
from python3-netclient and python3-xmlrpc respectively.
Add the missing run-time dependencies.
Signed-off-by: Dmitry Rozhkov <dmitry.rozhkov@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The shebang's length is usually 128 as defined in /usr/include/linux/binfmts.h:
#define BINPRM_BUF_SIZE 128
So there would be errors when /path/to/hosttools/perl is longer than 128.
This patch fixes the problem when POSIX_SHELL and perl are used as the interpreters.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Fixed when len(TMPDIR) = 410:
Can't exec "/bin/sh": Argument list too long at /usr/lib/perl/5.18/IO/File.pm line 65.
This is becuase it has a lot of m4 files, use relative path for them
can fix the problem.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Fixed when len(TMPDIR) = 410:
Can't exec "/bin/sh": Argument list too long at /usr/lib/perl/5.18/IO/File.pm line 65.
This is becuase it has a lot of m4 files, use relative path for them
can fix the problem.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
The shebang's max length is usually 128 as defined in
/usr/include/linux/binfmts.h:
#define BINPRM_BUF_SIZE 128
So there would be errors when /path/to/hosttools/perl is longer than 128.
Set ac_cv_path_PERL to "/usr/bin/env perl" to fix the problem.
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
We've had a lot of users running into RSS issues where -initial recipes
were being installed into sysroots alongside their counterparts and
causing overlapping files issues.
In general this was through do_build dependencies. Such dependencies are
bad in general and I'd encourage people to compare the taskgraphs with
using a more specific dependency like do_populate_sysroot, do_image_complete
or do_deploy as often the more specific dependency will result in a much
cleaner build.
Regardless, we don't want -initial dependencies getting in the way like
this and there are cases a do_build dependency could make sense.
Deleting the do_build task in these cases makes sense since this is not
a build "endpoint" we'd ever want a user to use, its a behind the scenes
piece of bootstrappping.
Unfortunately to make this work, we need a newer bitbake version which
has a bitbake bug fixed.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Currently some shebang lines end up as
#! /usr/bin/env perl -w
env does not like the argument. Also the current sed to insert env
does not cover the copies ptests use. Fix these issues by:
- using --with-perl to insert "env"
- Replacing "-w" in shebang lines with a new "use warning;" line
Remove a EXTRA_OECONF_append_class_target from the native recipe.
Don't overwrite EXTRA_OECONF in native: the values should be correct
for native as well.
--with-patch is used within the gnu patch wrapper only: before this
commit the wrapper contained a (build host) path to native patch.
Also tweak one test so busybox mv output is accepted.
All ptests should now pass: Fixes [YOCTO #11221].
Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
fedorahosted.org was retired on March 1st, 2017. This is to
update the SRC_URI to point to pagure.io. pagure.io is a
replacement for fedorahosted.
[YOCTO #11226]
Signed-off-by: Choong YinThong <yin.thong.choong@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
|
|
If build host perl (and other tools) is old and we use some kind
of toolchain to provide recent perl/python/etc to the OE build
we still locked to use build host perl due to hardcoded shebang
lines in automake scripts.
Behaviour was observed with Enterprise Linux 6 and devtoolset
toolchain from SCL (Software Collections) used to provide recent
version of perl (not provided with default buildtools-tarball).
Pass /usr/bin/env perl in ac_cv_path_PERL configuration variables
for class-native and class-nativesdk. Use patch to automake to replace
-w option in shebang line with modern way to enable warnings on perl
(i.e. "use warnings").
Also add nativesdk-autoconf to RDEPENDS to bring runtime dependencies
inline with other targets.
Note that ac_cv_path_PERL must be valid perl interpreter path
since configure will check perl version and Flock implementation.
It is not possible currently to use nativeperl from native
sysroot because automake does not DEPENDS on perl-native (and
doing so fails due to circular dependencies). Only possible
solution is to overwrite shebangs with nativeperl somewhere at
do_install() and update RDEPENDS for class-native. Or add perl
symlinks to nativeperl in sysroot.
For now it seems good to use perl found by /usr/bin/env from
automake-native.
Cc: XE-Linux <xe-linux-external@cisco.com>
Signed-off-by: Serhii Popovych <spopovyc@cisco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
If build host perl (and other tools) is old and we use some kind
of toolchain to provide recent perl/python/etc to the OE build
we still locked to use build host perl due to hardcoded shebang
lines in autoconf scripts.
Behaviour was observed with Enterprise Linux 6 and devtoolset
toolchain from SCL (Software Collections) used to provide recent
version of perl (not provided with default buildtools-tarball).
Pass /usr/bin/env perl in ac_cv_path_PERL configuration variables
for class-native and class-nativesdk. Use patch to autoconf to replace
-w option in shebang line with modern way to enable warnings on perl
(i.e. "use warnings").
Also add nativesdk-m4 and nativesdk-gnu-config to RDEPENDS to bring
runtime dependencies inline with other targets.
Note that ac_cv_path_PERL must be valid perl interpreter path
since configure will check perl version and Flock implementation.
It is not possible currently to use nativeperl from native
sysroot because autoconf does not DEPENDS on perl-native (and
doing so fails due to circular dependencies). Only possible
solution is to overwrite shebangs with nativeperl somewhere at
do_install() and update RDEPENDS for class-native. Or add perl
symlinks to nativeperl in sysroot.
For now it seems good to use perl found by /usr/bin/env from
autoconf-native.
Cc: XE-Linux <xe-linux-external@cisco.com>
Signed-off-by: Serhii Popovych <spopovyc@cisco.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
Reference url is a stale, non existent site that returns a 404, so get rid of it
Change impacts both the manifest files and the scripts that generate the manifests
Run the following from within recipes-devtools/python
../../../scripts/contrib/python/generate-manifest-2.7.py > python-2.7-manifest.inc
../../../scripts/contrib/python/generate-manifest-2.7.py -n > python-native-2.7-manifest.inc
../../../scripts/contrib/python/generate-manifest-3.5.py > python-3.5-manifest.inc
../../../scripts/contrib/python/generate-manifest-3.5.py -n > python-native-3.5-manifest.inc
Signed-off-by: Bob Cochran <openembedded@mindchasers.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
|
liblzma is part of xz and we already build it but configure it out. This makes
no sense. Enabling it means we gain multithreaded compression and it speeds
dpkg-deb up massively. It also removes the fork overhead of separate xz processes.
Turning the existing config into a PACKAGECONFIG and turning it on by default
therefore makes best use of what we have available.
The manual RDEPENDS are no longer needed since it uses liblzma which is picked
up by the shlibs code magically.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
|
Markus Lehtonen
Alexander Kanavin
Richard Purdie
Jose Lamego
Khem Raj
Gernot Hillier
Ross Burton
Peter Kjellerstedt
Robert Yang
brian avery
Patrick Ohly
Yuanjie Huang
Li Zhou
Andre McCurdy
Dmitry Rozhkov
Martin Kelly
Joshua Lock
Jussi Kukkonen
Choong YinThong
Serhii Popovych
Bob Cochran