oeqa/sdk/galculator: rewrite to use new helpers

oeqa/sdk/python: fix version typo

Signed-off-by: Ross Burton <ross.burton@intel.com>

oeqa/sdk/assimp: use helpers

oeqa/sdk/case: add fundamental helper methods

avahi: avoid depending on skipped package

When built without D-Bus, libavahi-client is not build:
  Building libavahi-client:           no   (You need avahi-daemon and D-Bus!)
which causes avahi-dev RDEPENDS failing when creating an image
containing development-tools:
 *   - nothing provides libavahi-client = 0.7-r0 needed by avahi-dev-0.7-r0.cortexa8hf-neon

Signed-off-by: Jens Rehsack <sno@netbsd.org>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

uboot-sign.bbclass: fix signature and deployment

Fixed:
MACHINE = "beaglebone-yocto"
KERNEL_CLASSES += "kernel-fitimage"
KERNEL_IMAGETYPE_beaglebone-yocto = "fitImage"
UBOOT_MACHINE_beaglebone-yocto = "am335x_boneblack_vboot_config"
UBOOT_MKIMAGE_DTCOPTS = "-I dts -O dtb -p 2000"
UBOOT_SIGN_KEYDIR = "${TOPDIR}/conf"
UBOOT_SIGN_KEYNAME = "dev"
UBOOT_SIGN_ENABLE = "1"
IMAGE_INSTALL_remove = "kernel-image-zimage"

$ cd conf
$ openssl genrsa -F4 -out dev.key 2048
$ openssl req -batch -new -x509 -key dev.key -out dev.crt
$ cd ../
$ bitbake u-boot linux-yocto
$ grep signature tmp/deploy/images/beaglebone-yocto/*.dtb
Binary file tmp/deploy/images/beaglebone-yocto/u-boot-beaglebone-yocto-2018.07-r0.dtb matches
Binary file tmp/deploy/images/beaglebone-yocto/u-boot-beaglebone-yocto.dtb matches
Binary file tmp/deploy/images/beaglebone-yocto/u-boot.dtb matches

And there would be no signature info when rebuild from sstate:
$ bitbake u-boot linux-yocto -cclean
$ bitbake u-boot linux-yocto
$ grep signature tmp/deploy/images/beaglebone-yocto/*.dtb
No result

This s because kernel directly edit ${DEPLOY_DIR_IMAGE}/u-boot.dtb, (Note, it
is global ${DEPLOY_DIR_IMAGE}, not recipe's DEPLOYDIR), so that the modified
info is not in sstate, and would be lost when rebuild from sstate.

There are other problems in previouse code:
- The u-boot.dtb is provided by u-boot, but edited by kernel during signing, so
  it should be deployed by kernel rather than u-boot.

- The u-boot.do_concat_dtb directly install files to global ${DEPLOY_DIR_IMAGE},
  this is incorrect, the ${DEPLOY_DIR_IMAGE} should be installed by do_deploy.

- It seems that it assumes do_deploy depends on do_install according the comments,
  but they have no relationships:
  # do_concat_dtb is scheduled _before_ do_install as it overwrite the
  # u-boot.bin in both DEPLOYDIR and DEPLOY_IMAGE_DIR.

- The do_concat_dtb should be run after do_compile, but it doesn't have this
  dependency.

Make u-boot install u-boot.dtb to ${datadir}, kernel copies u-boot.dtb from
${STAGING_DATADIR} to ${B} and deploy it can fix the problem.

[YOCTO #12112]

Reported-by: Christian Andersen <c.andersen@kostal.com>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

strace: fix ptest fail and strip ptest host references

1. Add LDFLAGS_FOR_BUILD for strip host reference
2. Revert one previous replace operation for fix ptest fail.
   refer oe commit: 74b5088f1cc1708db43c33ac2dc7f01f4a4db9c1

Signed-off-by: Changqing Li <changqing.li@windriver.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

u-boot-tools: fix compile error

It uses sandbox_defconfig to produce u-boot tools. But EFI is only
supported by arm and x86, then it fails to run task do_compile on other
arches:

| include/config_distro_bootcmd.h:267:3: error: #error "sandbox EFI
| support is only supported on ARM and x86"

Only enable EFI support for u-boot-tools on x86 and arm to fix the issue.

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glibc: improve reproducibility with multilib

Multilib builds specify several loaders which will end up embedded in
some binaries or script files. To support reproducible builds, we must
ensure the loaders are always in deterministic order.

[YOCTO #2655]
[YOCTO #12478]
[YOCTO #12480]

Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
Improve the fix and rebase it to 2.28

Here is the log of lib32-glibc
[log.do_compile]
|Adjust ldd script
|ldd "/lib64/ld-linux-x86-64.so.2 /lib/ld-linux.so.2 /lib/ld-linux.so.2
/lib64/ld-linux-x86-64.so.2" -> "/lib/ld-linux.so.2 /lib64/ld-linux-x86-64.so.2"
[log.do_compile]

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

recipeutils-test: Add missing LIC_FILES_CHKSUM

Fixes
ERROR: QA Issue: recipeutils-test: Recipe file fetches files and does
not have license file information (LIC_FILES_CHKSUM) [license-checksum]

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libunwind: Enable building static archive on musl

This ensures that we have .a for libunwind on musl targets, this is used
for stack unwinding infra e.g. in rust

Help compiling rust compiler and standard library from meta-rust

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kernel: don't assign the build user/host

The KBUILD_BUILD_USER and KBUILD_BUILD_HOST variables were
assigned at the kernel class level, which made it impossible
to override them in the local configuration.

By setting only the default values of those variables in the
kernel class, it is now possible to override them as expected.

Signed-off-by: Federico Sauter <federico.sauter@ableton.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/runcmd: Increase timeout delta

Expecting 1s accuracy on a 2s timeout on a heavily loaded system has proven to be
unreliable. Update this to a 5s timeout with a 3s delta which should be achievable.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image-buildinfo: Remove unused function argument

Removes the listvars argument to image_buildinfo_outputvars(). It
doesn't appear that this argument ever did anything.

Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kernel-fitimage: Replace tabs with spaces in python

Fix:

  WARNING: python should use 4 spaces indentation, but found tabs in kernel-fitimage.bbclass, line 24

Signed-off-by: Alex Kiernan <alex.kiernan@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/context: ensure log directory exists

Ensure log directory exists to avoid the following error.

  FileNotFoundError: [Errno 2] No such file or directory: '/.../build-selftest/tmp/log/oe-selftest-results-20181207043431.log'

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

binutils: Fix build with clang

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libart-lgpl: remove

This is very dead, and nothing in oe-core or meta-oe depends on it anymore.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake.conf: Cleanup deprecated function usage

This function was moved in bitbake a long time ago, use the
preferred version to avoid a Deprecation warning.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes: Correctly markup regex strings

There are various escape characters in these stings which python warns
about so use the correct regex markup for them.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/esdk: Ensure parent directory exists

INFO - ======================================================================
INFO - ERROR: setUpClass (eSDK.oeSDKExtSelfTest)
INFO - ----------------------------------------------------------------------
INFO - Traceback (most recent call last):
  File "/home/pokybuild/yocto-worker/oe-selftest-debian/build/meta/lib/oeqa/core/case.py", line 32, in _oeSetUpClass
    clss.setUpClassMethod()
  File "/home/pokybuild/yocto-worker/oe-selftest-debian/build/meta/lib/oeqa/selftest/cases/eSDK.py", line 76, in setUpClass
    cls.tmpdirobj = tempfile.TemporaryDirectory(prefix="selftest-esdk-", dir=bb_vars["WORKDIR"])
  File "/usr/lib/python3.5/tempfile.py", line 929, in __init__
    self.name = mkdtemp(suffix, prefix, dir)
  File "/usr/lib/python3.5/tempfile.py", line 507, in mkdtemp
    _os.mkdir(file, 0o700)
FileNotFoundError: [Errno 2] No such file or directory: '/home/pokybuild/yocto-worker/oe-selftest-debian/build/build/tmp/work/qemux86_64-poky-linux/core-image-minimal/1.0-r0/selftest-esdk-q7ln84gc'

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/esdk: Fix typo causing test failure

2018-12-06 23:19:24,564 - oe-selftest - INFO - Traceback (most recent call last):
  File "/media/build1/poky-sumo/meta/lib/oeqa/core/case.py", line 32, in _oeSetUpClass
    clss.setUpClassMethod()
  File "/media/build1/poky-sumo/meta/lib/oeqa/selftest/cases/eSDK.py", line 78, in setUpClass
    cls.tmpdir_eSDKQA = cls.tempdirobj.name
AttributeError: type object 'oeSDKExtSelfTest' has no attribute 'tempdirobj'

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

tcl: Upgrade to 8.6.9

Upgrade tcl from 8.6.8 to 8.6.9.

Signed-off-by: Hong Liu <hongl.fnst@cn.fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

at: upgrade to 3.1.23

The following patches are rebased.

  fix_parallel_build_error.patch
  pam.conf.patch

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sysstat: upgrade to 12.1.1

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cups: upgrade to 2.2.9

The following patch is rebased.

  0001-don-t-try-to-run-generated-binaries.patch

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sudo: upgrade to 1.8.26

License-Update: include more files to check, but license remains the same.

Signed-off-by: Chen Qi <Qi.Chen@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

packages: respect PACKAGE_NO_GCONV

PACKAGE_NO_GCONV is set in libc-package.bbclass if not all of
'libc-charsets libc-locale-code libc-locales' included in
DISTRO_FEATURES. And then no packages glibc-gconv-* glibc-charmap-* and
glibc-localedata-* is created. Update recipes and conf file which depend
on these packages to check required distro features.

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image.bbclass: respect PACKAGE_NO_GCONV

It installs locale-base-* packages according to var IMAGE_LINGUAS.
Packages locale-base-* are split in libc-package.bbclass if variable
PACKAGE_NO_GCONV is not set. When none of ditro features libc-charsets
libc-locales and libc-locale-code is set, PACKAGE_NO_GCONV is set. Then
no locale-base-* is created and fails to create image.

Clear IMAGE_LINGUAS in such situation.

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

glibc-locale: fix installed-vs-shipped qa issue

Variable PACKAGE_NO_GCONV is set in libc-package.bbclass when none of
libc-charsets, libc-locales and libc-locale-code set in DISTRO_FEATURES.
Then it causes installed-vs-shipped qa issue of glibc-locale:

| ERROR: glibc-locale-2.28-r0 do_package: QA Issue: glibc-locale:
| Files/directories were installed but not shipped in any package:
|   /usr/share/i18n
|   /usr/share/i18n/charmaps
|   /usr/share/i18n/locales
|   /usr/share/i18n/charmaps/CP737.gz
|   ...
|   /usr/share/i18n/locales/ru_RU
|   ...
|   /usr/lib64/gconv/gconv-modules
| Please set FILES such that these items are packaged. Alternatively if they
| are unneeded, avoid installing them or delete them within do_install.
| glibc-locale: 843 installed and not shipped files. [installed-vs-shipped]

So check PACKAGE_NO_GCONV during do_install and not copy those files if
PACKAGE_NO_GCONV has been set.

Simplify call of bb.utils.contains() in libc-package.bbclass as well.

Signed-off-by: Kai Kang <kai.kang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

autoconf-doc: improve reproducibility

autoconf-doc package contains autoconf.info.
This file contains date when this file was created, i.e:

     "This manual (31 January 2018) .."

Therefore, two builds done on two different days will show different dates for
otherwise identical files, hence breaking reproducibility.
The date is obtained from mtime of "autoconf.texi", unfortunately we patch this
file and change the mtime as a consequence.
We restore reproducibility by removing the patch modifying "autoconf.texi".
As a consequence certain supplemental information will be not added to the documentation.
In particular:  more details on usage, and on workarounds for non-updated projects.

[YOCTO #12524]

Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
With applying this fix:

    "This manual (24 April 2012) .."

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/waf: Fix builds when B != S

Waf requires that the current working directory be ${S} (the location of
the wscript) when building. Most of the time, this was true only because
B defaults to S. However, anything that changed that behavior (notably,
using externalsrc) would break the recipe. Remedy this by explicitly
changing cwd to ${S} when running waf commands. As a happy side effect,
B can be set up for "out of tree" builds to keep the source directory
clean.

Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

uboot-sign.bbclass: Remove tab indentations in python code

Use 4 spaces to replace a tab.

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/esdk: run selftest inside workdir not /tmp

We've seen issues with rootfs size calculations and we've seen systems
like opensuse which have btrfs mounted on /tmp causing selftest failures.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa: don't litter /tmp with temporary directories

If we need to create a temporary directory in targetbuild or buildproject use
tempfile.TemporaryDirectory so that when the test case is finished, the
directory is deleted.

Also synchronise the logic and don't possibly store the temporary directory in
self.tmpdir as nothing uses that.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/utils/qemurunner: Avoid tracebacks on closed files

Reorder the shutdown/teardown to avoid:

  File "/home/pokybuild/yocto-worker/oe-selftest-ubuntu/build/meta/lib/oeqa/utils/qemurunner.py", line 224, in launch
    op = self.getOutput(output)
  File "/home/pokybuild/yocto-worker/oe-selftest-ubuntu/build/meta/lib/oeqa/utils/qemurunner.py", line 90, in getOutput
    fl = fcntl.fcntl(o, fcntl.F_GETFL)
ValueError: I/O operation on closed file

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/runqemu: Improve testcase failure handling

assertTrue doesn't give good debug information when things fail. Update
several to use assertIn which gives information upon failure, for the
others print the log information upon failure.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib/oe/utils: Improve multiprocess_lauch exception handling

We've seen a cryptic:

"ERROR: Fatal errors occurred in subprocesses, tracebacks printed above"

message from oe-selftest with no other traceback information. Improve the
traceback logging to try and give a better indication of any errors that is
ocurring.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python3: drop redundant patch

This patch altered the clean target's behaviour to skip the ipkg-install
directory.  However this directory isn't created by opkg, opkg-utils, or the
package_ipk class; and we don't invoke the clean target as we perform
out-of-tree builds.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python3: don't cripple target distutils

We stop distutils for *native* Python from rewriting hashbangs when installing
(so installed scripts don't have a hashbang that refers to sysroot paths), but
this isn't needed nor desirable for the *target* Python.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nss: Fix SHA_HTONL bug for arm 32be.

Rpm use nss as digest crypto library and which will cause an error as follows:

error: test-manual-1.2.3-20181012.noarch.rpm: Header SHA1 digest: BAD (Expected
f1deb7dc4a10742d88ccd1e967dbc62ae45095a5 !=4ad9d7dad6d70d6086eefec62612ad5d77f2fe81)  => this value is wrong
error: test-manual-1.2.3-20181012.noarch.rpm: not an rpm package (or package manifest)

The error is caused by SHA_HTONL in nss, for there is no need to reverse the host value for arm 32be, so fix it.

Signed-off-by: Zheng Ruoqin <zhengrq.fnst@cn.fujitsu.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gdb: Remove long ago upstreamed patch

This fix was upstreamed a long ago, see [1].

[1] https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;a=commit;h=37ce4055fe907b9edd25498dcda7a133dbd19784

Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

valgrind: Skip vgpreload_memcheck shared object from stripping

This is a special library for memcheck tool, where it needs to have the
symbols intact for the stack traces to work on target, current option is
to install valgrind-dbg ( 151 MB uncompressed ) is quite big for some
systems which may not have space to install it all. Leaving it
unstripped adds about 200KB to image which is much better, this alone
gets memcheck working, as an aside we might need same solution for other
tools e.g. helgrind etc. when needed, they also have leading libraries
installed

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/icecc.bbclass: Fix ccache disable

The ccache disable flag was misspelled, preventing it from being
disabled.

Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

util-linux: split out blkdiscard into its own package

For systems that don't otherwise depend on the full
util-linux package, blkdiscard is a mere 18k (on
cortexa7t2hf-neon).

Signed-off-by: André Draszik <andre.draszik@jci.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

patchreview: Various fixes/improvements

Add various fixes and improvements including the ability to export
patch statsitics as json data.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

insane: Clarify GNU_HASH warning

We have a fatal error if ELF objects don't have GNU_HASH segments but it
doesn't explain what the problem is.  At least give a hint to users by
suggesting that LDFLAGS wasn't passed to the compiler.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-firmware: better packaging for TI wl12xx & wl18xx firmwares

Currently, the linux-firmware-wl12xx contains all wl12xx and
wl18xx firmwares, except for the wl18xx compatibility symlinks
for old kernels and the linux-firmware-wl18xx contains just
those compatibility symlinks and nothing else. This doesn't
make sense...

Be more specific about what to package into each package, in
particular because the existing wl12xx package is specific
about symlinks already.

At the same time, we split the common bits into a -wlcommon
package, so that the wl18xx package doesn't need to depend
on all the wl12xx firmwares, saving several MiB in the file
system.

Signed-off-by: André Draszik <andre.draszik@jci.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-firmware: Bump revision to 1baa348

This adds following changes:

1baa348 Merge branch 'nxp_mc' of https://github.com/NXP/linux-firmware
c7ba355 qed: Add 8.37.7.0 firmware image
1cb4e51 amdgpu: add raven dmcu firmware
aa71b2d amdgpu: update raven firmware to 18.40
fcd5a5f amdgpu: update fiji firmware to 18.40
453caa3 amdgpu: update tonga firmware to 18.40
7ceb224 amdgpu: update carrizo firmware to 18.40
a136e78 amdgpu: update polaris10 firmware to 18.40
ac5f8bd amdgpu: update vega10 firmware to 18.40
8d3825c linux-firmware: add firmware for mt7650e
96d6db5 linux-firmware: add MC firmware for NXP DPAA2 SoCs
d877533 linux-firmware: liquidio: fix GPL compliance issue
2618544 linux-firmware: Update firmware file for Intel Bluetooth,8265
c34a52a linux-firmware: Update firmware patch for Intel Bluetooth 8260
d8e8163 linux-firmware: Update firmware file for Intel Bluetooth,9260
66b58d4 linux-firmware: Update firmware file for Intel Bluetooth,9560
50b8b0b linux-firmware: add firmware for mt7610e
de9cefa Merge branch 'firmware-update' of https://github.com/01org/opa-firmware
951cd9e Merge branch 'ath10k-20181010' of git://git.kernel.org/pub/scm/linux/kernel/git/kvalo/linux-firmware
c6398e3 Update Intel OPA hfi1 firmware
eaceb79 ath10k: QCA9984 hw1.0: update board-2.bin
a00e61b ath10k: QCA9984 hw1.0: update firmware-5.bin to 10.4-3.6.0.1-00003
50d4180 ath10k: QCA988X hw2.0: update firmware-5.bin to 10.2.4-1.0-00041
427d31d ath10k: QCA9888 hw2.0: update board-2.bin
d87e815 ath10k: QCA9888 hw2.0: update firmware-5.bin to 10.4-3.6-00140
097bbdf ath10k: QCA9887 hw1.0: update firmware-5.bin to 10.2.4-1.0-00041
3e2e5d3 ath10k: QCA9377 hw1.0: add firmware-6.bin to WLAN.TF.2.1-00021-QCARMSWP-1
a87eb5f ath10k: QCA6174 hw3.0: update firmware-6.bin to RM.4.4.1.c2-00057-QCARMSWP-1
14024d3 ath10k: QCA4019 hw1.0: update board-2.bin
2ac415e ath10k: QCA4019 hw1.0: update firmware-5.bin to 10.4-3.6-00140
f503e1d nfp: Add Agilio BPF firmware rev 2.0.6.124
c6b6265 Merge tag 'iwlwifi-fw-2018-10-03' of git://git.kernel.org/pub/scm/linux/kernel/git/iwlwifi/linux-firmware
338684a rtw88: Add firmware file for driver rtw88
68a4930 nfp: update Agilio SmartNIC flower firmware to rev AOTC-2.9.A.37
da110f2 iwlwifi: add -41.ucode firmwares for 9000 series
b1ac8f4 iwlwifi: update firmwares for 9000 series
58265e0 iwlwifi: update firmwares for 7000, 8000 and 9000 series
dc638d6 nfp: update Agilio SmartNIC firmware to rev 2.1.13
7c81f23 ti-connectivity: add firmware for CC2560(A) Bluetooth
f0a1cbf linux-firmware: mediatek: add firmware for mt7668u Bluetooth
eb6419c nvidia: add GV100 signed firmware
1ab9095 Merge git://git.marvell.com/mwifiex-firmware
2ddd8f0 Merge branch 'master' of https://github.com/bgodavar/qca_rome_firmware
18c4c8a firmware/icl/dmc: Add v1.07 of DMC for Icelake
52b23a9 linux-firmware: add Marvell SD8997 firmware image
534daf4 qca: update BT firmware files for QCA ROME chip.

License-Update: new firmware files additions
Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

openssl: correct bad path on package preprocess

In case of SDK generation, /usr/bin/ path are not correct
and must be replaced by ${bindir}.

Signed-off-by: Christophe Priouzeau <christophe.priouzeau@st.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>

piglit: upgrade to current HEAD

Signed-off-by: Ross Burton <ross.burton@intel.com>

ruby: upgrade 2.5.1 -> 2.5.3

Signed-off-by: Ross Burton <ross.burton@intel.com>

libevdev: upgrade 1.5.9 -> 1.6.0

Signed-off-by: Ross Burton <ross.burton@intel.com>

gnutls: no need to inherit binconfig

This recipe doesn't ship a *-config binary, so don't inherit binconfig.

Signed-off-by: Ross Burton <ross.burton@intel.com>

gnome-desktop-testing: upgrade to 2018.1

Drop backported and obsolete patches.

Respect systemd DISTRO_FEATURE when enabling/disabling journal support.

Signed-off-by: Ross Burton <ross.burton@intel.com>

gobject-introspection.bbclass: add libdir to XDG_DATA_DIRS

Otherwise .gir files installed in the sysroot will not be found when multilib is in use
(multilib configurations use libdir rather than datadir).

Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gobject-introspection: add a missing patch that was overlooked when adding meson support

This should also fix Rygel builds in meta-oe

Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/oelib/path: don't leak temporary directories

setUp() is used to populate a directory of temporary files, and deleted in
__del__.  However setUp() is called once *per test* so __del__ would only be
able to remove the last directory created.

Fix the code by using the natural counterpart to setUp, tearDown(), to clean up.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cpio: update patch to merged version

The segfault on append was fixed upstream with a different patch, so apply that
instead.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mdadm: add back lost Upstream-Status tags

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

btrfs-tools: update to 4.19

Add a backported patch to fix big endian compilation.

Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

musl: Update to latest trunk

Complete changelogs are here
https://git.musl-libc.org/cgit/musl/log/?qt=range&q=c50985d5c8e316c5c464f352e79eeebfed1121a9..39ef612aa193cc6e954ac5a01574300ccd4b7ef9

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nfs-utils: Fix exportfs segfault on musl

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nfs-utils: Upgrade to 2.3.3

enhance the musl-only patch to degrade certain Werros into Warnings
which helps in compiling on musl, since its checking for __GLIBC__
defines which are undefined on musl.

fix build on x32 ABI

Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/sdk/python: add Python 2 and fix detection

Add a Python 2 form to exercise that if present, and fix the setUp() so it
actually looks for a package that exists (nativesdk-python3 is a virtual
package, the interpretter is in nativesdk-python3-core).

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

classes/update-alternatives: Skip alternatives when disabled

Skips the update alternative steps for recipes that shouldn't have them
enabled.

Fixes errors like:

 nativesdk-bzip2-1.0.6-r5 do_package: bzip2: alternative target
 (/opt/poky/2.5+snapshot/sysroots/i686-pokysdk-mingw32/usr/bin/bunzip2
 or
 /opt/poky/2.5+snapshot/sysroots/i686-pokysdk-mingw32/usr/bin/bunzip2.bzip2)
 does not exist, skipping...

When building mingw SDKs

[YOCTO #12962]

Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

attr: Strip build host paths and tool versions from ptest files

Using the new PTEST_BUILD_HOST_FILES mechanism.
Specifically remove reference to build host rpm version, if any.

Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

acl: Strip build host tool details from installed ptest file

Adopt the PTEST_BUILD_HOST_FILES mechanism to strip build host paths from the
installed builddefs file. Also strip the record of the build host rpm version.

Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

acl: Use install instead of cp

Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

strace: Strip build host tool details from installed ptest Makefile

Adopt new PTEST_BUILD_HOST_PATTERN mechanism.

Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

openssl: Strip perl version from installed ptest configdata.pm file

Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

openssl ptest: Strip build host paths from configdata.pm

This file contains CC, CPP, CFLAGS, CXXFLAGS and the like.

Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ptest: Optionally strip build host tool info from installed files

Several packages (eg. acl, attr, openssl) install Makefiles or other build host
configuration files for ptest so that they can run `make check`, for instance.
These build host files commonly include paths and versions of build host tools,
(eg. rpm or perl) whose presence and version depends on the build host distro.

Optionally strip lines from installed PTEST_BUILD_HOST_FILES using an
extended regex pattern defined in PTEST_BUILD_HOST_PATTERN. This has no
effect on the on-target ptest.

Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ptest: Reproducibility: Strip build host paths from any installed Makefile

A common pattern for ptest is to install a Makefile that provides a make check
target. These generated Makefiles are normally full of build host paths, to
hosttools, and to files and scripts within the source or build tree. They also
commonly include the CFLAGS and CPPFLAGS. None of these build host paths
can possibly work on-target, so stripping them has no effect on the success of
the resulting ptests.

Stripping the HOSTTOOLS_DIR has this effect, for instance:

    -MKDIR_P = /home/douglas/workspace/upstream/build/tmp/hosttools/mkdir -p
    +MKDIR_P = mkdir -p

The curious WORKDIR/*= pattern avoids non-parseable -fdebug-prefix-map patterns
in the CFLAGS by adding an arbitrary ".", eg:

    -fdebug-prefix-map=.=/usr/src/debug/rsyslog/8.37.0-r0

In all other cases, we simply remove the WORKDIR, eg:

    CPP = arm-tait-linux-gnueabi-gcc -E --sysroot=recipe-sysroot ...

Signed-off-by: Douglas Royds <douglas.royds@taitradio.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oe-selftest: add some tests for recipeutils module

Add some tests for functions in meta/lib/oe/recipeutils.py, in
particular for a few issues I've just fixed. I haven't added tests for
all of the functions - some of them are already being tested via devtool
in any case.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib/oe/recipeutils: drop obsolete functions

These date from the time before Tinfoil's API covered this functionality
(back when you could actually access cooker from a tinfoil-based
script).

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib/oe/recipeutils: patch_recipe(): fix replacing varflag values

The code here wasn't correctly getting the variable history for
varflags, so for example if you did a devtool upgrade on a recipe where
the SRC_URI checksums were in the .inc file (typical for python recipes
in order to support both python 2 and 3) then after the upgrade the
new values would be set in the recipe and the old values were left in
the .inc, which is not right. Teach the code here how to get the history
for varflags so it works properly.

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

lib/oe/recipeutils: patch_recipe(): fix handling of values across includes/classes

If we were setting a variable and part of the variable's value was being
set in a class or a .inc file, we were still just setting the value
outright on the assumption that it was too hard to do otherwise. With
some careful use of the variable history we can do better for certain
situations i.e. when the recipe does not currently set the value
outright.

Additionally, correctly remove _appends for variables we are changing if
we're trying to remove the value added in the _append.

Fixes [YOCTO #12623] and partially fixes [YOCTO #9360].

Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meson: Correct use of the _append operator

The value to SRC_URI_append_class-native was not prefixed with a space.
This was not noticed as the SRC_URI before applying the _append contains
trailing spaces. However, if one, e.g., has a .bbappend and adds to the
SRC_URI using SRC_URI += "file://foo.patch", then there no longer is any
trailing space and the _append concatenates the two URIs together,
leading to a build failue.

Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libmodulemd: Extend to nativesdk

Avoids universe fetch warnings:

WARNING: Nothing PROVIDES 'nativesdk-libmodulemd' (but virtual:nativesdk:/home/pokybuild/yocto-worker/build-appliance/build/meta/recipes-devtools/libdnf/libdnf_0.22.3.bb DEPENDS on or otherwise requires it). Close matches:
WARNING: Nothing RPROVIDES 'nativesdk-libdnf-dev' (but virtual:nativesdk:/home/pokybuild/yocto-worker/build-appliance/build/meta/recipes-devtools/libdnf/libdnf_0.22.3.bb RDEPENDS on or otherwise requires it)
WARNING: Nothing RPROVIDES 'nativesdk-libdnf' (but virtual:nativesdk:/home/pokybuild/yocto-worker/build-appliance/build/meta/recipes-devtools/libdnf/libdnf_0.22.3.bb RDEPENDS on or otherwise requires it)
WARNING: Nothing RPROVIDES 'nativesdk-dnf' (but virtual:nativesdk:/home/pokybuild/yocto-worker/build-appliance/build/meta/recipes-devtools/dnf/dnf_4.0.9.bb RDEPENDS on or otherwise requires it)
WARNING: Nothing RPROVIDES 'nativesdk-dnf-dev' (but virtual:nativesdk:/home/pokybuild/yocto-worker/build-appliance/build/meta/recipes-devtools/dnf/dnf_4.0.9.bb RDEPENDS on or otherwise requires it)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

licence: Add license file CC-BY-SA-4.0

Original legalcode.txt:
https://creativecommons.org/licenses/by-sa/4.0/legalcode.txt

Signed-off-by: Eric Chanudet <chanudete@ainfosec.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ltp: add missing patch Upstream-Status

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

piglit: add missing waffle-bin dependency

The sanity test uses wflinfo, so depend on waffle-bin.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

binconfig: only try to mangles files

If a file matches the glob that is an absolute symlink then sed will try to
alter files on the host, so restrict the search to just actual files.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gstreamer1.0-vaapi: downgrade vaapisink to marginal rank

Using vaapisink (which doesn't supports DRI3 [1] and uses DRI2) with
default poky configuration currently results in an unresponsive display
because DRI2 rendering doesn't work (as of xserver 1.20.3) in non-composited
environments [2].

Downgrade vaapisink to marginal for now so playbin (and in turn gst-play
and gtk-play examples) uses next best sink element and works out of box.

[1] https://github.com/intel/libva/issues/122
[2] https://gitlab.freedesktop.org/xorg/xserver/issues/13

Signed-off-by: Anuj Mittal <anuj.mittal@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libc-package: fix postinst error when ENABLE_BINARY_LOCALE_GENERATION = "0"

[YOCTO #13028]

Signed-off-by: Alexander Kanavin <alex.kanavin@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python: Adds instructions to the manifest file

While there is a bit of documentation regarding building a new
manifest file for python, it seems that users usually only read
the manifest file.

The manifest file is in JSON format which doesn't allow comments,
hence why instructions were initially put elsewhere.

This patch hacks the call to open the JSON manifest file by using a
marker to trick it into reading only part of the file as the manifest
itself, and keep the other part as comments, which contain instructions
for the user to run the create_manifest task after an upgrade or
when adding a new package.

Signed-off-by: Alejandro Enedino Hernandez Samaniego <alejandr@xilinx.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

python3: Adds instructions to the manifest file

While there is a bit of documentation regarding building a new
manifest file for python, it seems that users usually only read
the manifest file.

The manifest file is in JSON format which doesn't allow comments,
hence why instructions were initially put elsewhere.

This patch hacks the call to open the JSON manifest file by using a
marker to trick it into reading only part of the file as the manifest
itself, and keep the other part as comments, which contain instructions
for the user to run the create_manifest task after an upgrade or when
adding a new package.

Signed-off-by: Alejandro Enedino Hernandez Samaniego <alejandr@xilinx.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libtasn1: no need to inherit binconfig

This recipe doesn't ship a *-config binary, so don't inherit binconfig.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image_types: use cpio-native to build cpio images

As per the previous commit, upstream cpio has a bug which means it crashes on
append. If the image being built has already had testimage ran then cpio-native
will be in the sysroot.  It's also possible that some distributions are shipping
this broken CVE patch too.

Now that our cpio-native is fixed, until we can be sure that the host cpio isn't
broken depend on cpio-native if building a cpio image.

[ YOCTO #13042 ]

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cpio: fix crash when appending to archives

The upstream fix for CVE-2016-2037 introduced a read from uninitialized memory
bug when appending to an existing archive, which is an operation we perform when
building an image.

Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/context: Improve log file handling

The existing logfile is simply placed in the current directory. Since the test
changes cwd to BUILDDIR, the symlink to the log can be placed in an invalid
directory. We also see trackbacks if the symlink is invalid.

Improve things by:

* Placing logs in LOG_DIR (or BUILDDIR if unset).
* Using a full path to the log meaning the log and link are placed in the same directory.
* Using lexists instead of exists so invalid symlinks are handled correctly.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

meta-selftest/error: Cleanup large trailing whitespace

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/buildoptions: Ensure diskmon tests run consistently

Heartbeat events default to once a second and we need to ensure we have
enough time in the task to see them.

Add a nostamp delay task 5s long so we can have a consistently timed
task which doesn't need cleanup or have unneeded dependencies. This
ensures we should deterministically see the disk moinitor events
regardless of the state of the build. This is done in a way which
doesn't corrupt build state or need cleanup and is efficient.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/buildoptions: Improve ccache test

This test occisionally fails as m4 doesn't recompile, meaning the logfile test
then doesn't find mention of ccache.

To ensure m4 does recompile, clean m4 before force compiling it.

(Reading the test is confusing due to the test cleanup also involving a clean)

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

nspr: improve reproducibility

Remove _BUILD_STRING and _BUILD_TIME in _pl_bld.h
to avoid introducing timestamp in the .so library
such as libnspr4.so as below:
 $ readelf --wide --decompress --hex-dump=.rodata libnspr4.so
 [snip]
  0x00004000 32303138 2d31312d 31352030 353a3439 2018-11-15 05:49
 [snip]

[YOCTO #12639]

Signed-off-by: Mingli Yu <Mingli.Yu@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ghostscript: 9.25 -> 9.26

- Drop backported CVE fixes
  000[1-8]*.patch

Signed-off-by: Hongxu Jia <hongxu.jia@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/runqemu: Improve lockfile handling for python with close_fd=True

On python versions with close_fds=True (python 3.2 onwards), the tap
device lockfile isn't passed to the child process.

Since this guards against use of an active interface, we really want this
here, so pass it in pass_fds. This means if the parent exits early, the child
still holds the lock, avoiding messages like:

runqemu - ERROR - Failed to run qemu: qemu-system-x86_64: could not configure /dev/net/tun (tap0): Device or resource busy

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/runqemu: Tidy up lock handling code

Various tweaks:
- Balance up the aquire/release functions
- Use debug messge for both acquiring and release message for consistency in logs
- Use None instead of an empty string
- Reset the value of the field if we don't have the lock any more

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/qemurunner: Remove resource python warnings

If runqemu fails it would leak an unclosed socket and file. Ensure we
close these in all cases to remove the resource warning.

Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>