scripts/oe-selftest: avoid the creation of coverage file when coverage not installed

Coverage subprocessing file is being created even when coverage
is not installed, which causes errors of "module not found" to
be send to the oe-selftest output.

This patch adds indent to the block of code creating this coverage
file, so it can only be executed when coverage is actually
installed.

[Yocto #9334]

Signed-off-by: Humberto Ibarra <humberto.ibarra.lopez@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/oe-selftest: remove coverage file if any coverage option is given

Coverage temporal file for sub-processing is being removed only
when the --coverage option was found. This is wrong since the
file is created when any coverage option (source, include or omit)
is given, even if --coverage is not one of them.

This patch makes sure to remove the file if any coverage option was
given.

Signed-off-by: Humberto Ibarra <humberto.ibarra.lopez@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

scripts/oe-selftest: remove unneeded coverage warning

There is a message that warns the user about enabling subprocessing
for coverage to work. After the fix for Yocto #8930,
this task is done automatically, so the warning is not needed
anymore.

Signed-off-by: Humberto Ibarra <humberto.ibarra.lopez@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

patch.bbclass: remove useless path assignment

A path variable is assigned the value of PATH from the
environment, before the PATH is set on the environemnt
from the value of PATH in the bb datastore. This seems
to be an unnecessary leftover.

[YOCTO #8543]

Signed-off-by: Alex Franco <alejandro.franco@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gstreamer: remove now-redundant expansion in do_split_packages

do_split_packages now expands extra_depends, so there's no need to pre-expand
it.

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

package: do_split_packages: expand variables in extra_depends

If a recipe passes for example extra_depends="${PN}-dev" to do_split_packages()
then it isn't expanded, so the check for the multilib prefix doesn't work.
Solve this centrally by expanding extra_depends inside do_split_packages().

[ YOCTO #9381 ]

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

xf86-video-intel: Add patch to fix some poor image quality

This fixes a problem on the Intel Gen8 Graphics and is backport from
the upstream master of xf86-video-intel driver code

[YOCTO #8587]

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sanity: Increase minimum git version to 1.8.3.1

The kernel tools assume git > 1.7.9.5, I'm unsure of the exact version but
the oldest in our infrastructure is 1.8.3.1. The git fetcher also currently
has nasty workarounds for git < 1.7.9.2.

Moving to 1.8.3.1 as our minimum version seems sane at this point as the oldest
we're testing/supporting.

[YOCTO #6162]

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

scripts/oe-buildenv-internal: Fix regression in BB_ENV_EXTRAWHITE setting

The commit OE-Core:ada4639 (oe-buildenv-internal: simplify derivation
of BB_ENV_EXTRAWHITE) changed the format of BB_ENV_EXTRAWHITE
variable to:

,----
| BB_ENV_EXTRAWHITE='ALL_PROXY
| BB_NO_NETWORK
| BB_NUMBER_THREADS
| ...
| '
`----

Instead of:

,----
| BB_ENV_EXTRAWHITE='ALL_PROXY BB_NO_NETWORK BB_NUMBER_THREADS...'
`----

The old format allow for external script parsing easier and there is
no need to change the format as it has no benefit from usage
perspective.

Signed-off-by: Otavio Salvador <otavio@ossystems.com.br>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

license.bbclass: fix warnings when run in unprivileged "container" env

An unprivileged "container" environment like this[1] doesn't have root
account (uid 0) which causes tons of "Invalid argument" warnings:

  $ bitbake ...
  ...
  WARNING: Could not copy license file [src] to [dest]: [Errno 22] Invalid argument: '[src]'
  WARNING: Could not copy license file [src] to [dest]: [Errno 22] Invalid argument: '[src]'
  WARNING: Could not copy license file [src] to [dest]: [Errno 22] Invalid argument: '[src]'
  ...

Fix it by handling EINVAL similar to existing handling of EPERM (which
was added for when not running under pseudo).

[1]: The real environemnt is buildFHSUserEnv from NixOS/nixpkgs, but a
  demonstration of the issue can be done like this:

    $ touch f
    $ unshare --user --mount chown 0:0 f
    chown: changing ownership of ‘f’: Invalid argument

Signed-off-by: Bjørn Forsman <bjorn.forsman@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

externalsrc: avoid race in temporary git index file

Use a unique tempfile as the temporary git index file when determining
the git hash of the source tree. A fixed filename was obviously causing
races (with the git index.lock file) under oe-selftest where multiple
bitbake instances were run against the same source tree at the same
time.

Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bdwgc: use github repo for source location

The Yocto Autobuilder has been unable to fetch the release package
from the developer's website at www.hboehm.info, so change recipe
to fetch from the developer's github repo instead.

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

xf86-video-intel: Add patch to allow UXA to build

This patch is a backport to fixup an API change in xorg-server

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

package_manager.py: better error handling in opkg's package listing

opkg does not return a non-zero exit code even if it found
errors. When that happens, parsing the output leads to strange
follow-up errors.

To avoid this we need to check explicitly for non-empty
stderr. Reporting only that on a failure also leads to shorter error
messages (stdout may be very large).

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd: make systemd-serialgetty optional

Some distros may prefer to use the upstream systemd support for
starting getty on serial ports. This is now possible by adding
"serial-getty-generator" to PACKAGECONFIG.

The default is unchanged, i.e. systemd's own serial-getty@.service
file does not get packaged and instead systemd-serialgetty is pulled
into images via RRECOMMENDS.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ncurses: reorder PACKAGES

Having ncurses-terminfo-base before ncurses-terminfo is currently irrelevant
because the current file lists are completely disjunct. However, when building
"stateless" via a .bbappend, the content of curses-terminfo-base also needs
to live under /usr/share and then it becomes important that files
belonging to ncurses-terminfo-base are checked first.

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bluez5.inc: remove obsolete workaround

Bluez 5.37 itself correctly installs bluetooth.conf, and honors
the path settings in dbus-1.pc.

Removing the obsolete workaround is necessary for compiling
"stateless" (= read-only system configuration moved out of /etc).

Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

buildtools-tarball: Add texinfo (for makeinfo)

Initially I was reluctant to do this however makeinfo is a dependency
of the buildsystem and only adds around 400kb to the buildtools-tarball
so it likely makes sense to add it. This allows people to use
the project on older environments.

Need to enable nativesdk-texinfo but this seems straightforward.

[YOCTO #8990]

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

cogl: fix G-I .typelib installation

By default gobject-introspection puts .typelib files into $PN, but the cogl-1.0
recipe doesn't use PN.  Installing cogl-1.0 is impossible as libcogl
conflicts/replaces with cogl-1.0 due to a previous upgrade.

Fix this by putting the .typelib files into the right subpackages, and
explicitly setting FILES_${PN} to "" so that nothing accidently ends up in PN.

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

classes/buildhistory: fix grammar in comments

Fix a minor grammatical error in the comments here.

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

classes/buildhistory: fix filtering of depends-nokernel.dot

For images we produce a number of filtered dependency .dot files for
readability, the first of which is depends-nokernel.dot which filters
out just the kernel itself (not kernel modules). Unfortunately the
filter specifications hadn't been updated for the dash-to-underscore
removal or the 4.x kernel upgrade, thus the filtering wasn't actually
doing anything.

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

classes/buildhistory: optimise getting package size list

Invoking oe-pkgdata-util in turn for every package in the list was slow
with a large image. Modify oe-pkgdata-util's read-value command to take
an option to read the list of packages from a file, as well as prefix
the value with the package name; we can then use this to get all of the
package sizes at once. This reduces the time to gather this information
from minutes to just a second or two.

Fixes [YOCTO #7339].

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

build-appliance-image: Exclude DDATETIME from task signature

Otherwise the task hash changes between server and worker context
leading to changing task checksums.

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

image-vm: Exclude DISK_SIGNATURE_GENERATED from task signature

Otherwise the task hash changes between server and worker context
leading to changing task checksums.

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

populate_sdk_ext: Exclude BBTASKDEPDATA from task signature

Otherwise the task hash changes between server and worker context
leading to changing task checksums. The dependency data here is
tracked by other pieces of the signature.

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

opkg-utils: opkg-build exit when fail to list files.

We have an issue when ls segfaults in some cases [1] so it's
better to detect the failure at this level instead of continue
the build process.

[YOCTO #8926]

[1] https://bugzilla.yoctoproject.org/show_bug.cgi?id=8926#c0

Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

kernel-yocto: enforce SRC_URI specified branch

During the simplication and cleanup of branches and kernel meta data
handling, the ability to force build a branch that didn't match the
meta data was dropped.

There are valid uses cases when a different branch should be built
(testing, development, etc), so we restore the capability with this
change.

If after the kernel meta data is processed the current branch does
not match the SRC_URI specified branch, a warning is generated
about the impending branch switch and that the user should double
check that they are building what they expect.

  WARNING: After meta data application, the kernel tree branch is standard/base. The
  WARNING: SRC_URI specified branch standard/gt. The branch will be forced to standard/gt,
  WARNING: but this means the board meta data (.scc files) do not match the SRC_URI specification.
  WARNING: The meta data and branch standard/gt should be inspected to ensure the proper
  WARNING: kernel is being built.

Reported-by: Steve Sakoman" <steve.sakoman@intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.4: UVC: Add support for R200 depth camera

Integrating the following commit:

[
    Add support for Intel R200 depth camera in uvc driver.
    This includes adding new uvc GUIDs for the new pixel formats,
    adding new V4L pixel format definition to user api headers,
    and updating the uvc driver GUID-to-4cc tables with the new formats.

Tested-by: Greenberg, Aviv D <aviv.d.greenberg@intel.com>
Signed-off-by: Aviv Greenberg <aviv.d.greenberg@intel.com>
Signed-off-by: Sakari Ailus <sakari.ailus@linux.intel.com>
Signed-off-by: Guennadi Liakhovetski <g.liakhovetski@gmx.de>
Signed-off-by: Mauro Carvalho Chehab <mchehab@osg.samsung.com>
Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
]

Signed-off-by: Saul Wold <sgw@linux.intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.4: fix PAT for 32bit x86

Integrating the following mainline destined commits to fix PAT issues with
32bit x86 and qemu:

  bc22b90fb583 x86/pat: Document the PAT initialization sequence
  c534b1ec1991 x86/xen, pat: Remove PAT table init code from Xen
  cd47692e804f x86/mtrr: Fix PAT init handling when MTRR is disabled
  039434bdc165 x86/mtrr: Fix Xorg crashes in Qemu sessions
  c08196e8064c x86/mm/pat: Replace cpu_has_pat with boot_cpu_has()
  6928fce8c766 x86/mm/pat: Add pat_disable() interface
  3163c8d5468d x86/mm/pat: Add support of non-default PAT MSR setting

With this change applied, we once again have working graphics and no special
work arounds.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

Revert "linux-yocto: Work around PAT issue on qemux86"

This reverts commit 94abdb2eea610b174064978d2fb8354a6231fc0c.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto-dev: bump to v4.6-rcX

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.1: ahci: backport AHCI runtime PM

enabling AHCI runtime PM support via the following commits:

  f5b08e95bc03 ahci: Add runtime PM support for the host controller
  e05e05973f39 ahci: Add functions to manage runtime PM of AHCI ports
  558571930e2d ahci: Convert driver to use modern PM hooks
  4a2c2074f209 ahci: Cache host controller version
  21641413456c scsi: Drop runtime PM usage count after host is added
  254ab69f652d scsi: Set request queue runtime PM status back to active on resume
  61c02cb167ed block: Add blk_set_runtime_active()
  7bcfbac98f3c ahci: Store irq number in struct ahci_host_priv
  a5a093b7f6e4 ahci: Move interrupt enablement code to a separate function

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.4: gpio-pca953x: add PCAL9535 interrupt support

Reverting two previous gpio commits to allow the application of proper
upstream ready patches to add this support:

  08943f2bbd50 gpio-pca953x: add "drive" property
  a517d5b72e76 gpio: pca953x: provide GPIO base based on _UID
  c91063f44368 gpio-pca953x: add PCAL9535 interrupt support
  58f3c9f0ac6d Revert "gpio-pca953x: add "drive" property"
  7abbd5fec15d Revert "gpio: pca953x: provide GPIO base based on _UID"

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.1: telemetry and dmaengine backports

Backports to support Broxton and Apollo lake:

96c0e80f4007 dmaengine: idma64: clear LLP_[SD]_EN bits in last descriptor
3b2dbd1cadbf dmaengine: idma64: use local variable to index descriptor
d61599f94946 dmaengine: idma64: convert idma64_hw_desc_fill() to return void
9c9f37533789 dmaengine: idma64: set maximum allowed segment size for DMA
9d55c9523aa5 dmaengine: idma64: drop IRQ enable / disable in handler
f80e3f3f87c5 dmaengine: idma64: use lo_hi_readq() / lo_hi_writeq()
216429c319aa dmaengine: idma64: make better performance on pause / resume
2f982603b358 dmaengine: idma64: this is not DesignWare
a7f7ee3a6a91 dmaengine: idma64: useless use of min_t()
3c8f2fcd2dfd dmaengine: idma64: convert to __ffs()
b452447dce9f dmaengine: idma64: improve residue estimation
849fdbe687a8 dmaengine: add a driver for Intel integrated DMA 64-bit
0038e6a4798e intel_punit_ipc: add NULL check for input parameters
4bd7fc987da6 platform:x86: Add Intel telemetry platform device
472c405dd66d intel_pmc_ipc: update acpi resource structure for Punit
e1754195568b MAINTAINERS: Combine multiple telemetry entries
14d8e14aca6d intel_telemetry_debugfs: Fix unused warnings in telemetry debugfs
8724677a6b60 platform:x86: Add Intel Telemetry Debugfs interfaces
d503a325c0ee platform:x86: Add Intel telemetry platform driver
255a8ac345e7 platform/x86: Add Intel Telemetry Core Driver
c475943802b5 platform:x86: add Intel P-Unit mailbox IPC driver
c2f1b2c7282c MAINTAINERS: Add maintainership for MIC drivers
7d389b483d32 surface pro 3: Add support driver for Surface Pro 3 buttons

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic/isoimage-isohybrid.py: change cpio generated uid&gid to root

By default cpio preserves the uid&guid's of the original user which
leads to host contamination and boot failures because commands like
mount from initramfs expect to be run by root and the original host
user might not even exist on the target.

Signed-off-by: Ioan-Adrian Ratiu <adrian.ratiu@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

wic/isoimage-isohybrid.py: use glob to find initramfs location

Some filenames can omit 'initramfs', or use other names. This makes
detection more flexible by using only the image name, machine arch and
image type in a glob wildcard.

Signed-off-by: Ioan-Adrian Ratiu <adrian.ratiu@ni.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bluez5: add ptest support

[YOCTO #5134]

Signed-off-by: Alexander Kanavin <alexander.kanavin@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oe/patch: print cleaner error message when patch fails to apply

[YOCTO #9344]

Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oe/patch: more detailed error reporting

Show the actual command that failed when raising a CmdError. Makes
figuring out what actually failed much easier.

[YOCTO #9344]

Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

insane.bbclass: avoid false positives on library location

package_qa_check_libdir() reports that the file libsoletta.so.0.0.1-gdb.py
in /usr/share/gdb/auto-load is in the wrong location. Before generating
a warning for files in non-standard locations, check that the file is an
actual elf file (and hence a real library file).

[YOCTO #9215]

Signed-off-by: Bill Randle <william.c.randle@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

grub-efi.bbclass: use GRUB_ROOT rather than APPEND for root device

Use APPEND for grub's root device may cause confusion, for example, when
building efi + pcbios, there maybe be two root=/dev/ram0, one of them
would be carried to the installed target, and the target would fail to
boot. Use GRUB_ROOT to fix the problem, and remove SYSLINUX_ROOT from
APPEND will fix the problem.

[YOCTO #9354]

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

bitbake.conf: Add BB_WORKERCONTEXT to HASHBASE_WHITELIST

We never want checksum to vary depending on whether we calculate them
in server or worker context, that would make no sense. This was happening
in do_rootfs tasks since they reference the BB_WORKERCONTEXT variable.

Whitelist the variable to avoid this class of problem.

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

gdb-cross-canadian: use PACKAGECONFIG for python and readline

Signed-off-by: Jonathan Liu <net147@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

base: Fixup PACKAGECONFIG incorrect mappings

PACKAGECONFIG doesn't work in cross-canadian recipes at the moment as
DEPENDS are prepended with mlprefix.

A recipe is either nativesdk, native or target so adjust the if
statements accordingly, use inherits_class for more accurate recipe
classification and add cross-canadian support.

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

classes/packagegroup: Refactor code to be simpler

This makes the code a bit shorter and more readable.

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

default-distrovars.inc: remove libassuan from LGPLv2_WHITELIST_GPL-3.0

The libassuan recipe is now buildable in distros which blacklist
GPL-3.0 without needing to be explicitly whitelisted (since it
provides at least one non GPLv3 package).

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libassuan: use package specific licensing

 - The main libassuan.so library is LGPLv2.1+
 - Test apps (in tests sub directory) and documentation are GPLv3+
 - Windows CE specific binaries (gpgcedev.dll and gpgcemgr) are GPLv3+

When building the current OE recipe, the test apps are compiled but
not installed. The Windows CE specific binaries are not compiled.

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

init-install-efi.sh: remove all root=foo from grub.cfg

There might be more than one root=/dev/foo in the config file which
would cause unepected errros on the installed target, so remove all of
them.

[YOCTO #9354]

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

init-install.sh: fix disk_size

It mis-matched "SanDisk" or "Disk Flags" before, which caused unexpected
error.

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

ltp: fix test_proc_kill hanging

Signed-off-by: Tudor Florea <tudor.florea@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

ltp: add periodic output for memcg stress test

Signed-off-by: Tudor Florea <tudor.florea@enea.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

epiphany: Depend on intltool-native for configure

Fixes [YOCTO #9374].

Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image: Fix debugfs image type recursion loop

The debugfs prefix is striped from t, but not from baset.
Therefore baset never matches t.

Signed-off-by: Freudiger Raphael <raphael.freudiger@siemens.com>
Signed-off-by: Pascal Bach <pascal.bach@siemens.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

selftest/signing: Use packagedata to obtain PR value for signing test

Using PF to calculate the rpm filename doesn't work when PR server is
enabled and an extra PR value can be injected. Add code to use packagedata
to obtain the full name, allowing the test to work when PR server is
in use.

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

lib/oe/packagedata: Add import os

So that the packagedata module can be used externally to the core OE
environment, add a missing import.

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

base.bbclass: avoid duplicate call to d.getVar('LICENSE', True)

Reuse the 'license' variable, instead of calling
d.getVar('LICENSE', True) again.

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

base.bbclass: drop obsolete HOSTTOOLS_WHITELIST_GPL-3.0

base.bbclass sets 'check_license' to False (and therefore skips
license checking completely) for native, nativesdk, etc recipes
(ie anything which could potentially be classed as "host tools"),
so supporting a dedicated whitelist of GPLv3 host tools is not
necessary.

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

man: use BUILD_CC and target include files for configure

The configure script was using the cross-compiler for feature tests,
which fails to execute when the target is different than the host.
Change the script to use the build compiler instead, but use the
target include files to check for the target features.

[YOCTO #9359]

Signed-off-by: Bill Randle <william.c.randle@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image-vm.bbclass: remove invalid code

It has been done in syslinux.bbclass.

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

image-live.bbclass/image-vm.bbclass: remove duplicated code

Move the common code to live_vm_common.bbclass and remove duplicated ones.

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

bootimg.bbclass: merge it into image-live.bbclass

They are doing the same things: create live images, merge them into one
bbclass makes it easy to understand.

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

boot-directdisk.bbclass: merge it into image-vm.bbclass

They are doing the same things: create virtual machine images, merge
them into one bbclass makes it easy to understand.

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

man: fix several annoying compile/build warnings

Fixed the build error when building man.config.5 (a remnant of a long
ago previous patch). Optimized the manpages Makefile for parallel
builds. Drop a FHS patch that is no longer needed, as the standard
Makefile puts the man pages in the proper location. Also, fix compile
warnings in a couple other files.

[YOCTO #9341]

Signed-off-by: Bill Randle <william.c.randle@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image.bbclass: Make unneeded packages for a read-only rootfs configurable

Previously the list of packages that are considered unneeded for a
read-only rootfs was hardcoded. This made it impossible to, e.g., have
shadow installed on a system with a read-only rootfs, but where /etc
is mounted writable.

This also lists ${VIRTUAL-RUNTIME_update-alternatives} rather than
update-alternatives (as was previously the case) since this should
actually remove the intended package.

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

relocate_sdk: additional error checks

When installing SDK in a non-default location and the path length
of the SDK install location is longer than the path length of the
default SDK location, relocation of .ldsochache section will overwrite
file location outside of the .ldsocache section size.
In addition, additional checks were added to ensure that any
path in sections .gccrelocprefix and .ldsochache will not exceed
the space allocated for it within the file, which would also result
in file corruption.

[YOCTO #9268]

Signed-off-by: Juro Bystricky <juro.bystricky@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd: fix build with gcrypt PACKAGECONFIG disabled

systemd-resolved build fails without gcrypt PACKAGECONFIG. Backport the fix.
Also remove the comment about resolved's dependence on gcrypt.

[YOCTO #9219]

Signed-off-by: Maxin B. John <maxin.john@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

devtool: modify: call shutdown on tinfoil when done

Strictly speaking we ought to explicitly shut down a tinfoil instance
when we're done with it. This doesn't affect modify's operation but is
important if you want to be able to call into modify() from another
plugin (though anyone doing so should be advised that the function is
by no means a stable API and is subject to change in future releases).

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

toolchain-shar-extract.sh: ensure all_proxy is allowed through

all_proxy is referred to by oe-git-proxy so ensure it is allowed through
into the installer environment in case the extensible SDK install
process needs to query a remote git repository.

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

oe-publish-sdk: exclude sstate-cache if publishing minimal SDK

If SDK_EXT_TYPE is set to "minimal" then the SDK won't contain many
sstate artifacts, and you're required to set up an sstate mirror in this
case anyway so there's no point publishing the "stub" sstate-cache
directory from within the SDK since it won't be useful for update
purposes and may be confused with the real sstate-cache.

There is however a possibility that people might publish the real
sstate-cache directory under the same output directory provided to
oe-publish-sdk, thus deleting it after extracting (as we were doing with
other files we wanted to clean up at the end) would be problematic,
besides which extracting it and then deleting it is wasteful. Thus,
introduce a "-p" command line option to the SDK installer that we can
use to tell tar not to extract the items we don't want when publishing.
This has the added benefit of mostly keeping references to these in the
place they belong i.e. in populate_sdk_ext.bbclass.

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

oe-publish-sdk: prevent specifying a directory for the SDK argument

The SDK argument is expected to be an installer .sh file; if a directory
is specified we can get an ugly failure later on; best to check up
front.

Fixes [YOCTO #9065].

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

classes/populate_sdk_ext: support setting vars from environment at build time

When running bitbake you may pass in values of variables from the
external environment (making use of BB_ENV_EXTRAWHITE), and you may
choose to do this when building the extensible SDK, for example:

  MACHINE=qemuarm bitbake -c populate_sdk_ext core-image-minimal

You would naturally expect those settings to be reflected in the
extensible SDK itself; however they were not, since we were only
considering local.conf and auto.conf. Check the variables mentioned in
BB_ENV_EXTRAWHITE to see if any are different than the values set in
local.conf/auto.conf and add lines setting them in the SDK's local.conf
if so.

Fixes [YOCTO #9339].

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

scripts, lib: Don't limit traceback lengths to arbitrary values

There appears to have been a lot of copy and pasting of the code
which prints tracebacks upon failure and limits the stack trace to
5 entries. This obscures the real error and is very confusing to the user
it look me an age to work out why some tracebacks weren't useful.

This patch removes the limit, making tracebacks much more useful for
debugging.

[YOCTO #9230]

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

pyton-numpy: Add definition of off_t size

NPY_SIZEOF_OFF_T must be defined. Otherwise:

/home/ricardo/curro/qt5022/oe/openembedded-core/build/tmp-glibc/sysroots
/qemux86-64/usr/lib/python2.7/site-packages/numpy/core/include/numpy/
npy_common.h:140:9: warning: "NPY_SIZEOF_OFF_T" is not defined [-Wundef]
     #if NPY_SIZEOF_OFF_T == NPY_SIZEOF_SHORT
         ^
/home/ricardo/curro/qt5022/oe/openembedded-core/build/tmp-glibc/sysroots/
qemux86-64/usr/lib/python2.7/site-packages/numpy/core/include/numpy
/npy_common.h:142:11: warning: "NPY_SIZEOF_OFF_T" is not defined [-Wundef]
     #elif NPY_SIZEOF_OFF_T == NPY_SIZEOF_INT
           ^
/home/ricardo/curro/qt5022/oe/openembedded-core/build/tmp-glibc/sysroots/
qemux86-64/usr/lib/python2.7/site-packages/numpy/core/include/numpy/
npy_common.h:144:11: warning: "NPY_SIZEOF_OFF_T" is not defined [-Wundef]
     #elif NPY_SIZEOF_OFF_T == NPY_SIZEOF_LONG
           ^
/home/ricardo/curro/qt5022/oe/openembedded-core/build/tmp-glibc/sysroots/
qemux86-64/usr/lib/python2.7/site-packages/numpy/core/include/numpy/
npy_common.h:146:11: warning: "NPY_SIZEOF_OFF_T" is not defined [-Wundef]
     #elif NPY_SIZEOF_OFF_T == NPY_SIZEOF_LONGLONG
           ^
/home/ricardo/curro/qt5022/oe/openembedded-core/build/tmp-glibc/sysroots/
qemux86-64/usr/lib/python2.7/site-packages/numpy/core/include/numpy/
npy_common.h:149:10: error: #error Unsupported size for type off_t
         #error Unsupported size for type off_t

Signed-off-by: Ricardo Ribalda Delgado <ricardo.ribalda@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

image-live.bbclass: DEPENDS on syslinux

The mkisofs requires syslinux/isolinux.bin which is provided by
target syslinux, for pcbios, the depends is already there, but for
EFI, it isn't, so add the depends.

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

ldconfig-native: Fix ELF flags on 64-bit binaries

Yocto's ldconfig-native was exported from an old version of glibc and
generates an ld.so.cache that is partially ignored by current versions
of glibc when loading libraries. This occurs for 64-bit binaries, where
glibc expects FLAG_ELF_LIBC6, but ldconfig-native only generates the
standard ELF flag. Fix this with an additional patch on top of the patch
for now.

You can verify this by applying the patch below to your target copy of
glibc and running

  LD_DEBUG=libs /lib64/ld-linux-x86-64.so.2 --list $anybinary

--- ./elf/dl-cache.c.orig       2016-01-07 11:05:36.823447171 +0100
+++ ./elf/dl-cache.c    2016-01-07 11:19:53.925878628 +0100
@@ -106,6 +106,8 @@
                if (_dl_cache_check_flags (flags)                             \
                    && _dl_cache_verify_ptr (lib->value))                     \
                  {                                                           \
+                   if (__glibc_unlikely (GLRO_dl_debug_mask & DL_DEBUG_LIBS))\
+                       _dl_debug_printf ("  considering file %s\n", cache_data + lib->value); \
                    if (best == NULL || flags == GLRO(dl_correct_cache_id))   \
                      {                                                       \
                        HWCAP_CHECK;                                          \
@@ -117,6 +119,9 @@
                             searching.  */                                   \
                          break;                                              \
                      }                                                       \
+                 } else {                                                    \
+                     if (__glibc_unlikely (GLRO_dl_debug_mask & DL_DEBUG_LIBS))              \
+                       _dl_debug_printf ("  ignoring file %s due to flags %x, expected %x\n", cache_data + lib->value, lib->flags, GLRO(dl_correct_cache_id)); \
                  }                                                           \
              }                                                               \
            while (++middle <= right);                                        \
@@ -265,14 +270,23 @@

       /* Only accept hwcap if it's for the right platform.  */
 #define HWCAP_CHECK \
-      if (lib->hwcap & hwcap_exclude)                                        \
+      if (lib->hwcap & hwcap_exclude) {                                              \
+       if (__glibc_unlikely (GLRO_dl_debug_mask & DL_DEBUG_LIBS))            \
+           _dl_debug_printf ("   hwcap mismatch %x vs. %x\n", lib->hwcap, hwcap_exclude); \
        continue;                                                             \
-      if (GLRO(dl_osversion) && lib->osversion > GLRO(dl_osversion))         \
+      }                                                                              \
+      if (GLRO(dl_osversion) && lib->osversion > GLRO(dl_osversion)) {       \
+       if (__glibc_unlikely (GLRO_dl_debug_mask & DL_DEBUG_LIBS))            \
+           _dl_debug_printf ("   os version mismatch %x vs. %x\n", lib->osversion, GLRO(dl_osversion)); \
        continue;                                                             \
+      }                                                                              \
       if (_DL_PLATFORMS_COUNT                                                \
          && (lib->hwcap & _DL_HWCAP_PLATFORM) != 0                           \
-         && (lib->hwcap & _DL_HWCAP_PLATFORM) != platform)                   \
-       continue
+         && (lib->hwcap & _DL_HWCAP_PLATFORM) != platform) {                 \
+       if (__glibc_unlikely (GLRO_dl_debug_mask & DL_DEBUG_LIBS))            \
+           _dl_debug_printf ("   platform mismatch %x vs. %x\n", lib->hwcap & _DL_HWCAP_PLATFORM, platform); \
+       continue;                                                             \
+      }
       SEARCH_CACHE (cache_new);
     }
   else

This version of ldconfig-native should really be replaced with a version
matching the glibc source in use on the target platform.

Signed-off-by: Clemens Lang <clemens.lang@bmw-carit.de>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

recipes-support/rng-tools: Change runlevel start from S to 2, 3, 4, 5.

When using systemd as init rng-tools is causing a circular dependency
between units,

[    7.706250] systemd[1]: basic.target: Found ordering cycle on
basic.target/start
[    7.706934] systemd[1]: basic.target: Found dependency on
sysinit.target/start
[    7.707795] systemd[1]: basic.target: Found dependency on
rng-tools.service/start
[    7.708692] systemd[1]: basic.target: Found dependency on
basic.target/start
[    7.709461] systemd[1]: basic.target: Breaking ordering cycle by
deleting job rng-tools.service/start
[    7.710404] systemd[1]: rng-tools.service: Job rng-tools.service/start
deleted to break ordering cycle starting with basic.target/start

The problem is related to systemd running sysvinit scripts by default
add dependency of basic.target for sysvinit script so when sysvscript
is at rcS is added also as dependency of sysinit.target causing a
cirular dependency in this case: basic.target -> sysinit.target ->
rng-tools.service -> basic.target.

Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/runtime/parselogs.py: Add systemd unit circular dependencies errors.

When systemd is enabled as init we need to notice when circular
dependencies in units happen because systemd try to solve this situation
removing the unit itself.

Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

systemd-serialgetty: allow baud rate overriding

In case a getty is required on a UART which is not being
used as the kernel console, the current agetty invocation
fails to obey the baud rate configured through the
SERIAL_CONSOLES variable because it uses --keep-baud.

Signed-off-by: Awais Belal <awais_belal@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

psmisc: Remove including sys/user.h and __WORDSIZE

We can replace __WORDSIZE with UINTPTR_MAX
sys/user.h is in conflict with asm/ptrace.h on musl

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

selftest: Added testcase decorator to tests

1418 test_recipetool_create_cmake
1422 test_qemu
1423 test_devtool_add_git_local
1433 test_devtool_upgrade_git
1434 test_sanity_unsafe_binary_references
1435 test_read_only_image

Signed-off-by: Daniel Istrate <daniel.alexandrux.istrate@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto: add overlayfs feature

To let overlayfs be used like aufs, we create a feature
fragment. Configurations that want this union FS type, can
enable it via KERNEL_FEATURES.

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.4: broxton and usb type-c backports

c43425f73287 mfd: lpss: Add PCI IDs for Intel Broxton B-Step platform
2d6cc1d3e71f spi: pxa2xx: Add support for Intel Broxton B-Step
1df3674b7626 usb: dwc3: gadget: don't send extra ZLP
222ab0043e48 usb: dwc3: gadget: handle request->zero
3f01b2c0295a usb: dwc3: gadget: simplify dwc3_gadget_ep_queue()
729b5a1f8f65 usb: dwc3: gadget: set the OTG flag in dwc3 gadget driver.
65740f5d1a0d usb: dwc3: gadget: simplify next_request() return check
a82442f90ea7 usb: dwc3: Fix assignment of EP transfer resources
d99293bbaa50 usb: dwc3: ep0: fix setup_packet_pending initialization

Signed-off-by: Matt Ranostay <matt.ranostay@intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.4: drm/i915/skl: Fix DMC load on Skylake J0 and K0

[
    NUC6 (Skylake) graphics firmware fails to load without this patch.

    From: Mat Martineau <mathew.j.martineau@linux.intel.com>

    commit a41c8882592fb80458959b10e37632ce030b68ca upstream.

    The driver does not load firmware for unknown steppings, so these new
    steppings must be added to the list.

Cc: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Mat Martineau <mathew.j.martineau@linux.intel.com>
Reviewed-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Signed-off-by: Rodrigo Vivi <rodrigo.vivi@intel.com>
Link: http://patchwork.freedesktop.org/patch/msgid/1454023163-25469-1-git-send-email-mathew.j.martineau@linux.intel.com
]

Signed-off-by: California Sullivan <california.l.sullivan@intel.com>
Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.1: Intel Broxton: pwm backports

Integrating the following mainline changes:

afaf425e796c pwm: Add the pwm_is_enabled() helper
023721c3d067 pwm: atmel: Fix incorrect CDTY value after disabling
cd9fe8adb313 pwm: atmel: Fix incorrect CDTY value after enabling
3f445ca6df1f pwm: lpss: Rework the sequence of programming PWM_SW_UPDATE
9a51f622bdc9 pwm: lpss: Select core part automatically
8ee11b6f80ba pwm: lpss: Update PWM setting for Broxton
148228a52c2d pwm: lpss: Remove ->free() callback
e20d95bcd0b3 pwm: lpss: Add support for runtime PM
b2248acf6f8b pwm: lpss: Add more Intel Broxton IDs
85e8566bd241 pwm: lpss: Support all four PWMs on Intel Broxton
1bcdcb6275bf pwm: lpss: Add support for multiple PWMs

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.1: Apollo Lake/Broxton mmc backports

Backporting the following commits:

2b0cc0f0ca23 mmc: It is not an error for the card to be removed while suspended
67903c1f4c77 mmc: sdhci: 64-bit DMA actually has 4-byte alignment
4e434f483381 mmc: sdhci: Do not BUG on invalid vdd
05be442eb234 mmc: mmc: Fix incorrect use of driver strength switching HS200 and HS400
97a5f396dca6 mmc: core: Make runtime resume default behavior for MMC/SD
b6ee1ac07f14 mmc: mmc: Improve reliability of mmc_select_hs400()
45de48ce0b3c mmc: mmc: Move mmc_switch_status()
9d6aaaa0f6b7 mmc: mmc: Fix HS setting in mmc_select_hs400()
d7795de930c5 mmc: mmc: Improve reliability of mmc_select_hs200()
ac98e33a088d mmc: sdhci-acpi: Add more ACPI HIDs for Intel controllers
87f52cce8123 mmc: sdhci-pci: Add more PCI IDs for Intel controllers
cf3a4d2d4c7d mmc: sdhci-pci: Add another PCI ID for an Intel eMMC host controller
b3ef45585504 mmc: sdhci-pci: Build o2micro support in the same module
b32a4e71f49d mmc: sdhci: enable tuning for DDR50
f8e213bee1e1 mmc: sdhci: call sdhci_init() before request irq
ece0925eb94e mmc: sdhci-pci: Enable HS400 for some Intel host controllers
1c01bcdceaf6 mmc: sdhci-pci: Add support for drive strength selection for SPT
c1d6e92915a9 mmc: sdhci: Add a callback to select drive strength
40deb3ac3331 mmc: mmc: Add driver strength selection
e7d602f5992f mmc: mmc: Read card's valid driver strength mask
bc193a886571 mmc: core: Record card drive strength
a548d3993cc8 mmc: core: Factor out common code in drive strength selection
6713f5cf3164 mmc: core: Add 'card' to drive strength selection callback
852e6bd69b77 mmc: core: Simplify card drive strength mask
0bb40266bbb9 mmc: core: Allow card drive strength to be different to host
940d30f1a212 mmc: core: Reset driver type to default
efbc6ba8cd27 mmc: cast u8 to unsigned long long to avoid unexpected error
bafe0e5beba9 mmc: core: Don't print reset warning if reset is not supported
5b2789151852 mmc: block: Retry errored data requests when re-tuning is needed
d14211a4b2fe mmc: block: Check re-tuning in the recovery path
1ff15c648159 mmc: core: Flag re-tuning is needed on CRC errors
12544ee1cdbb mmc: sdhci: Change to new way of doing re-tuning
995073b182f7 mmc: core: Add support for HS400 re-tuning
53a0d731e824 mmc: core: Separate out the mmc_switch status check so it can be re-used
d3f59f108bf7 mmc: mmc: Hold re-tuning in mmc_sleep()
465db9504a27 mmc: core: Hold re-tuning while bkops ongoing
60cacf85ee10 mmc: core: Hold re-tuning during erase commands
5bd43535f4c1 mmc: core: Hold re-tuning during switch commands
4baa45ff8d15 mmc: core: Add support for re-tuning before each request
cd2ca27f9282 mmc: core: Enable / disable re-tuning
65723236b746 mmc: host: Add facility to support re-tuning

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.1: i2c: designware: Backport i2c patches

To enable Apollo Lake/Broxton, we backport the following changes:

da451824f75f i2c: designware: Convert to use unified device property API
58b1b5365f4a i2c: designware: Do not require clock when SSCN and FFCN are provided
8995091cfa0f i2c: designware: retry transfer on transient failure
1fc62907f84e i2c: designware: Move common probe code into i2c_dw_probe()
678110c0f2b8 i2c: designware: Rename platform driver probe and PM functions
c51c2e104d40 i2c: designware: Make dw_readl() and dw_writel() static
98d2b6e244ad i2c: designware: Remove unused functions
3d04c6e99652 i2c: designware: Disable interrupts before requesting PCI device interrupt
fb5b7be0fbe7 i2c: designware: Remove interrupt clearing from i2c_dw_pci_probe()
d48fd08b11c8 i2c: designware: Make debug print in i2c_dw_isr() shorter
28f94bc07450 i2c: designware-pci: use IRQF_COND_SUSPEND flag 95f494846fc1 i2c: designware: Avoid unnecessary resuming during system suspend
fb2ae3705a3a i2c: dw: Add APM X-Gene ACPI I2C device support
516a0fc91e00 i2c: designware: Add support for AMD Seattle I2C
874d63b3c95c i2c: designware: fix IO timeout issue for AMD controller
47543e80c6dd i2c: designware: reverts "i2c: designware: Add support for AMD I2C controller"
076fe383c4b5 i2c: designware: use to_pci_dev()

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.1: device property backports

Integrating the following 4.4 backports:

  630b726098c9 mfd: make mfd_remove_devices() iterate in reverse order
  aab24907568b driver core: Do not overwrite secondary fwnode with NULL if it is set
  89775462ef50 device property: return -EINVAL when property isn't found in ACPI
  a9eebae4a2c0 device property: check fwnode type in to_of_node()

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/qemuarm64: enable 32 bit compatibility

It was requested that CONFIG_COMPAT be enabled for qemuarm64 so that
32bit applications could be tested. This is simple enough to do, so
there is now a 32bit compat feature, and it is included only in
qemuarm64 to avoid impacting any other ARM targets (but the feature
is available for others to use).

[YOCTO #8646]

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/4.1: SMBus/iTCO backports

Backporting the following mainline commits to add Intel Broxton and Denverton
support:

  49b34899e7ca i2c: i801: Add support for Intel Broxton
  69f6eeceb5b1 i2c: i801: Add support for Intel DNV
  b131e3e13838 watchdog: iTCO_wdt: Add support for TCO on Intel Sunrisepoint
  fce230a538de mfd: watchdog: iTCO_wdt: Expose watchdog properties using platform data
  a1a2140e856d intel_pmc_ipc: Update kerneldoc formatting
  e407caf0a583 intel_pmc_ipc: Fix compiler casting warnings
  909588e549d7 intel_pmc_ipc: Add Intel Apollo Lake PMC IPC driver
  1cd89340c1a9 mfd: lpc_ich: Assign subdevice ids automatically
  729025b3e8d3 i2c: i801: Create iTCO device on newer Intel PCHs

Signed-off-by: Bruce Ashfield <bruce.ashfield@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

default-distrovars.inc: remove gnutls + libtasn1 from LGPLv2_WHITELIST_GPL-3.0

With the modern license handling code, gnutls and libtasn1 are both
buildable in distros which blacklist GPL-3.0 without needing to be
explicitly whitelisted (since they both provide at least one non
GPLv3 package).

Signed-off-by: Andre McCurdy <armccurdy@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sanity.bbclass: Use pythonexception to raise real exceptions without backtraces

If the sanity code encounters a version change is doesn't understand, the current
output is unreadable and confusing for the user, particularly due to the presence
of the backtrace.

Use improved functionality in bitbake to improve this and correctly pass python
exceptions around.

[YOCTO #9291]

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

sanity: Require bitbake 1.29.1

This is for the improved exec_func functionality.

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

uninative: Switch md5sum -> sha256

There are various concerns about md5 so use sha256 instead.

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

gcc: Fix musl ldso name for mips64

Now we have mips64 port of musl and there is
a name for ldso, backport the fix from upstream
gcc master

Signed-off-by: Khem Raj <raj.khem@gmail.com>

selftest/buildoptions.py: use INHERIT +=

* INHERIT = -> INHERIT +=

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>

archiver.bbclass: addtask do_deploy_archives_setscene

There was a do_deploy_archives_setscene, but no addtask.

Signed-off-by: Robert Yang <liezhi.yang@windriver.com>

gcc-5.3/gcc-4.9: -fdebug-prefix-map support to remap relative path

For relative path in DWARF, -fdebug-prefix-map could not remap it,
so translate to real path before mapping.

[YOCTO #9305]

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

ptest-runner_2.0.bb: Update recipe to point git.yoctoproject.org repo.

Since ptest-runner is a Yocto Project tool is better to have hosted
into git.yoctoproject.org.

Signed-off-by: Aníbal Limón <anibal.limon@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

man: fix src/Makefile to work with parallel make

The Makefile for man has a rule to create two generated targets with
a program run from the shell. Because the rule was run twice, with
parallel make, the generated files were being overwritten at the same
time they were being compiled. This patch forces the rule to be run
only once.

[YOCTO #9333]

Signed-off-by: Bill Randle <william.c.randle@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

oeqa/selftest/bbtests: Test bbappend order

BitBake should append to recipe in a predictable order.

fix for [YOCTO #9145]
test for [YOCTO #9138]

Signed-off-by: Daniel Istrate <daniel.alexandrux.istrate@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>