linux-yocto: Use INC_PR and move kernel require to linux-yocto.inc

When making changes to kernel.bbclass, it would be nice not to have to
manually change the PR of every linux-yocto*.bb file that requires it.

Move the "require kernel" line to linux-yocto.inc and update the
linux-yocto recipes to use INC_PR.

(From OE-Core rev: 7a41d96c946029aeec03cd26f326bc6ca26e74fd)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
CC: Bruce Ashfield <bruce.ashfield@windriver.com>
CC: Tom Zanussi <tom.zanussi@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

apr: Install apr-local libtool in build dir

Building apache with this version of apr would blow up on the results
from:

   apr_config --apr-libtool

Errors of the form:

   /bin/sh: /media/large/tmp/sysroots/x86_64-linux/usr/share/build-1/libtool: No such file or directory

Fix that by ensuring libtool from apr's build lands in
.../usr/share/build-1/.

(From OE-Core rev: eefbc2f219571dd0c66e1099ea4ad72b1a973fd4)

Signed-off-by: Joe MacDonald <joe.macdonald@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bison: link in librt on uclibc

uclibc has some functions in librt that are needed
by bison build.

(From OE-Core rev: 987bc3ff3e1a619132434c5d939c7035d3d19fcb)

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

uclibc-0.9.33: add posix_fallocate implementation

This gets systemd building with uclibc

(From OE-Core rev: d577fa59d8643954249fed7a79232e609add36f8)

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

uclibc-git: Update for building systemd

This adds uclibc patches to build systemd there are
new functions that are used by systemd which are needed

Additionally when using gold linker to build userspace
based on uclibc there are spurious hidden symbol that
show up in librt.so and libc.so which should not exist

So we patch uclibc for that

(From OE-Core rev: ecffa7571ed27a75e3248cb860affae233249440)

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

glib-2.0: Add missing dependency on virtual/libiconv

(From OE-Core rev: 5cb4de2f77b905279d85b04f236b7a135016db28)

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

uclibc: Separate the bits between uclibc and uclibc-initial

uclibc and uclibc-initial were providing conflicting PROVIDES
separate the common bits into logical pieces. Create a common
inc file to hold package information and version specific includes
so correct logical units can be included in uclibc and uclibc-initial
recipes.

(From OE-Core rev: bdfd2f6f286889b8920ebf6a670542b26534c1cb)

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

eglibc: support mipsel little endian, filter out -march=mips32

Quote for eglibc source:
 for mips eglibc now builds syscall tables for all abi's
 so we make sure that we choose right march option which is
 compatible with o32,n32 and n64 abi's
 e.g. -march=mips32 is not compatible with n32 and n64 therefore
 we filter it out in such case -march=from-abi which will be
 mips1 when using o32 and mips3 when using n32/n64

(From OE-Core rev: 6a013236d362874c661aa5f15b948aac32177181)

Signed-off-by: Dennis Lan <dennis.yxun@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

eglibc-2.15: Update to latest on 2.15 branch

(From OE-Core rev: e052ab9ac0f464ae4c7054a9af6663a0213d4192)

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

gcc-configure-sdk, gcc-configure-cross: Dont recompute USE_NLS

This was needed because on eglibc based builds we did not
define USE_NLS but now we define it in default config to have
a valid value so we do not need this here. Moreover it was
wrong and was not covering all cases of uclibc triplets

(From OE-Core rev: fc9492654aab2a6e3f46977193b5a5f388e4ee85)

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

bitbake.conf: Add weak definition for USE_NLS

USE_NLS is generally defined for uclibc based system builds
and generally its defined to 'no' there. However this variable
does not exist at all for eglibc/glibc distributions. This
patch adds a weak definition to 'yes' on eglibc based system
builds. This will ease out some of the cryptic contructs we
have to define certain options based on USE_NLS and also
checking got uclibc at the same time to avoid pythong exceptions
when its not defined.

(From OE-Core rev: 41db1bf69314e542a9d7ac1fb1b4bc65db75b642)

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

libgcc: Remove unpackage unwind.h

Fixes build warnings like below

WARNING: For recipe libgcc, the following files/directories were installed but not shipped in any package:
WARNING:   /usr/lib/arm-oe-linux-gnueabi/4.7.2/include
WARNING:   /usr/lib/arm-oe-linux-gnueabi/4.7.2/include/unwind.h

(From OE-Core rev: f7ec6ea2b32207993e7fc92ed67c08f542007953)

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

icu 3.6: dropp apply=yes from path

The "apply=yes" doesn't appear to be necessary, so drop it.

[YOCTO #2664]

(From OE-Core rev: 78abdd12305d4b74d7ec82fd38b7a27aaa8d7d95)

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

bitbake: runqueue.py: Handle multiple rdeptask entries

I'm not sure why we don't currently allow multiple entries in rdeptask when
we do in deptask. This makes the handling match between the two since
its trivial to fix.

(Bitbake rev: 19c84fe8854639768c874cc1449963a9867ad397)

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

bitbake: fetch2: Revert the regexp removal for the type field and instead anchor regexp

People are using regexps in the url type field so we need to preserve
this bitbake behaviour. To address the issues with https:// urls mapping
badly to file:// urls we anchor the regexp if its not already anchored.
There should be no expressions in the wild which would break with this
change.

(Bitbake rev: ce0579dc256251e523c6330641f98b9f5a0e5761)

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

bitbake: test/fetch: Switch the comparision order to make test failures slightly clearer

(Bitbake rev: 2b1311e21172847b6a86cfb21a84fd00e4ab1ac5)

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

bitbake: test/fetch: Ensure cwd is valud for the git clone

(Bitbake rev: 60b1a9f52dfec98e55a879a637f7142b0175b452)

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

bitbake: runqueue: Reimplement recrdepends so it works more correctly

Currently, recrdepends is extremely greedy. For example:

do_foo[rdepends] = "somedep:sometask"
addtask foo

which adds foo with *no* dependencies, will suddenly start appearing
as a dependency in every task which uses recrdepends. So far this has
been mildy annoying but we now have use cases where this makes no sense
at all.

This reworks the recrdepends code to avoid this problem. To do this we
can no longer collapse things into lists just based on file ID. The problem
is this code is extremely performance sensitive. The "preparing runqueue"
phase spends a lot of time in these recursive dependency calculations so any
change here could negatively impact the user experience.

As such, this code has been carefully tested on convoluted dependency trees
with operations like "time bitbake world -g". The net result of this change
and the preceeding changes combined is a net speed up of these operations in
all cases measured.

Tests were made comparing "bitbake world -g" task-depends.dot before and after
this patch. There *are* differences for example -nativesdk do_build dependencies
on -native recipes are no longer present. All removed dependencies appear to
be sensible improvements to the system. The "rdepends" cross contamination
issue above is also fixed.

(Bitbake rev: 82d73423c57569b984ee0ae3d93e3c3bd5dc5216)

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

bitbake: runqueue.py: Convert depends variable to use sets

This gives some small performance gains and sets the scene for other
improvements by removing the need for duplicate detection code.

(Bitbake rev: 6fd723479e8d49227fd58040b3485c1d5afc4bc5)

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

bitbake: taskdata: Add gettask_id_fromfnid helper function

This is like gettask_id but doesn't require translation of fnid -> fn
first which the function then translates back. This gives a sizeable
performance improvement since a significant number of lookups are avoided.

(Bitbake rev: 3190cb83e2af195a464f669c5aa8aedbf795160e)

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

linux-yocto/3.0: update BSP descriptions to new kern-tools format

In tree linux-yocto 3.0 boards indicate that they are BSPs via
the older/obsolete syntax "scc_leaf". This line in their board
description is detected by the build process and is used to find
the BSP description and the kernel type they use as their base.

To work with the latest kern tools, the BSP descriptions should
be updated to "include <ktype> branch <machine>" to acheive the
same result.

All trees newer than 3.0 already have this change and do not need
to be updated.

(From OE-Core rev: e32c993859337964033a285922dac3971e5add2d)

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

bitbake.conf: disable ccache explicitly if it is not enabled

The autogen-native built error on FC17:

ccache: failed to create /dev/null/.ccache

This is because the default gcc command of FC17 is a symlink to ccache,
so the ccache will always be used regardless to the setting of CCACHE,
ccache uses $HOME/.ccache as the CACHE_DIR by default, but autogen set
HOME=/dev/null, so the error happens.

Disable ccache explicitly if it is not enabled would fix the problem,
otherwise it would always use ccache regardless to the setting of CCACHE
on Fedora 17.

The ccache 3.1.7 has a bug, it would always create $CCCHE_DIR/.ccache
even CCACHE_DISABLE=1.

Unset CCACHE_DISABLE in ccache.bbclass, since ccache only checks whether
there is a CCACHE_DISABLE in the environment or not, it doesn't care about
its value, so we need unset it explicitly when enable ccache.

[YOCTO #2554]

(From OE-Core rev: dd2bab9b6a973d8086dfb6282e781fd79d30b05a)

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

archiver.bbclass: Add the function of filtering packages

This function can miss packages whose license is in
"COPYLEFT_LICENSE_EXCLUDE" and tarball packages with license in
"COPYLEFT_LICENSE_INCLUDE".

[YOCTO #2473]

(From OE-Core rev: 4800bed394ebd7fb50552a96d6a5f83d98fe790f)

Signed-off-by: Xiaofeng Yan <xiaofeng.yan@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

libxml: disable lzma

On my system libxml-native got linked with host copy of liblzma and as a
result libxslt-native was not linkable:

| x86_64-linux-libtool: link: gcc -isystem/home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/usr/include -O2 -pipe -Wall -Wl,-rpath-link -Wl,/home/hrw
/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/usr/lib -Wl,-rpath-link -Wl,/home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-
linux/lib -Wl,-rpath -Wl,/home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/usr/lib -Wl,-rpath -Wl,/home/hrw/HDD/devel/canonical/ci-linaro/oecore/buil
d/tmp-eglibc/sysroots/x86_64-linux/lib -Wl,-O1 -o .libs/xsltproc xsltproc.o  -L/home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/usr/lib -L/home/hrw/
HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/lib ../libxslt/.libs/libxslt.so ../libexslt/.libs/libexslt.so /home/hrw/HDD/devel/canonical/ci-linaro/oecore/
build/tmp-eglibc/work/x86_64-linux/libxslt-native-1.1.26-r8/libxslt-1.1.26/libxslt/.libs/libxslt.so /home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux
/usr/lib/libxml2.so -ldl /home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/usr/lib/liblzma.so -lrt -lz -lm -pthread -Wl,-rpath -Wl,/home/hrw/HDD/deve
l/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/usr/lib
| /home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/usr/lib/libxml2.so: undefined reference to `lzma_code@XZ_5.0'
| /home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/usr/lib/libxml2.so: undefined reference to `lzma_auto_decoder@XZ_5.0'
| /home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/usr/lib/libxml2.so: undefined reference to `lzma_end@XZ_5.0'
| /home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/sysroots/x86_64-linux/usr/lib/libxml2.so: undefined reference to `lzma_properties_decode@XZ_5.0'
| collect2: error: ld returned 1 exit status
| make[2]: *** [xsltproc] Error 1
| make[2]: Leaving directory `/home/hrw/HDD/devel/canonical/ci-linaro/oecore/build/tmp-eglibc/work/x86_64-linux/libxslt-native-1.1.26-r8/libxslt-1.1.26/xsltproc'

(From OE-Core rev: d4651d420a8cf359436a9614fbc9b56f901410c7)

Signed-off-by: Marcin Juszkiewicz <marcin.juszkiewicz@linaro.org>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gconf.bbclass: don't register schemas in the install stage

Previously this was installing schemas in the sysroot, which is wrong for native
packages as nothing should touch the sysroot directly, and even more wrong for
non-native packages as the sysroot is irrelevant.

So, export the environment variable that stops the registration happening at
install time. The postinst script will handle the non-native case, and for the
sysroot I've opened #2648.  This isn't a massive problem as nothing to my
knowledge actually installs schemas to the sysroot.

[YOCTO #2245]

(From OE-Core rev: f9f2e0309a03a4dc9ed5e6cff7db6f29962b7d60)

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

native.bbclass: correct PATH to have native-intercept be prepended

This was needed since it was after the sysroots and thus in some cases
the native sysroot chown was being found and used instead of the script
provided by native-intercept/chown.  This was noticed by the non-gplv3
build since it's coreutils depends on coreutils-native.

(From OE-Core rev: bb5095d3f73a00222f50f9dd2cf8dc27cc72b4a2)

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

linux-dtb: add multi-dtb build support

    including following enhancement:
    * support multi-dtb build
    * skip dtb build and install when KERNEL_DEVICETREE is empty
    * print a warning message when specified dts file is not available

(From OE-Core rev: 66f7921f34dfa8b3c7f8f7184715214164f3f8b0)

Signed-off-by: Zhenhua Luo <b19537@freescale.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

busybox: add correct ALTERNATIVE_TARGET for init.d/syslog

[YOCTO #2535]

(From OE-Core rev: 36ea756883d35f654b102f0cd6bdb2ec284753d0)

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

systemtap: update to version 1.8

Update the systemtap recipes to the recently released 1.8.  Remove a
couple patches whose changes are already present in the new version.

(From OE-Core rev: 1c5a127d075ebf78b70d14e3ca1d0ff640a80cb8)

Signed-off-by: Wade Farnsworth <wade_farnsworth@mentor.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

linux-yocto/3.4: -rt build fixes + configuration audit (part2)

Updating the 3.4 SRCREVs to pickup build fixes for -rt, feature backports,
and a second set of 3.4 configuration audit results.

  d65afd9 profiling: delete reference of obsolete CONFIG_X86_UP_APIC
  9431490 meta: fix net_sched.cfg include
  242149d meta: add CONFIG_VIRTIO_CONSOLE to virtio configuration
  671a822 meta: resync series -> branches
  e044ee1 rt: integrate patch to fix compile fail on certain configs
  edac822 ktest: update with v3.5 content
  182b8da net_sched: Add CODEL queue management algorithm.
  c6adcd9 net_sched: move content out of cgroups dir
  c8edb3e sys940x: branch before merge
  6d06257 ext3: delete duplicate enablement settings.
  6a5a2f8 ipv6: build in the core support vs. it being modular
  e106230 ipv6: make standard and RT share a common config block
  42996fb netfilter: coalesce bridge settings for standard and RT into a common file.
  2b62fea netfilter: coalesce IPv6 settings for standard and RT into a common file.
  d1c38c8 netfilter: coalesce IPv4 settings for standard and RT into a common file.
  b5f9c7c preempt-rt: align with standard kernel type for basic cfg
  ea33e01 cfg: move devtmpfs cfg/scc to cfg/fs dir
  517eda4 preempt-rt: combine two config fragments into one
  171a30a cgroups: re-enable net_traffic config fragment
  86a599d cgroups: delete old blkio patches
  8be6e5b netfilter: add comment documenting external bb use case
  90e06bf x86[_64]: align with korg on HOTPLUG_PCI, HOTPLUG_PCI_PCIE and PCI_MSI
  e103f7c seccomp: backport of BPF syscall filtering from v3.5

[YOCTO #1694]

(From OE-Core rev: 0c522dfabbf6b25488865a7653ea30f994d8177f)

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

linux-yocto-rt/3.4: update qemuppc branch

The 3.4 kernel adopted the qemu machine name for the kernel branches, so
the 3.4-rt recipe needs to be updated to reflect that.

(From OE-Core rev: 9cec3f55cd2381c5d69c6f0911bd985311e2baf9)

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

linux-yocto/3.4: update to v3.4.3

Bumping the linux-yocto 3.4 SRCREV to import the -stable update
to v3.4.3.

(From OE-Core rev: 529d07cf663e671140ea99877e73db091d5a9734)

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

linux-yocto/3.4: update and categorize configuration options

Updating the 3.4 kernel tree with some initial results of the 3.4
kernel configuration audit and refresh. This is the start of the
3.4 kernel configuration policy update and includes a refresh / update
with respect to kernel.org defaults and new categorization of
configuration fragments.

 0f6975b Sched: Import base BFS-423 for v3.4
 c517c5c cfg: strip off redundant path prefixes
 689fd20 cfg: create timer subdir for HZ and HPET related settings
 5367b46 cfg: add a fs dir for filesystem related configs
 67a784b x86/x86_64: disable MTRR sanitizer
 5da51ea x86/x86_64: consolidate ACPI and EFI settings
 7627402 x86/x86_64: enable CONFIG_NO_HZ=y by default.
 9ca6026 mips: select HZ=100 by default.
 149efc6 arm: select HZ=100 by default.
 29c9fc7 x86/x86_64: use HZ=1000
 700b8b5 x86/x86_64: enable boot mem corruption checks; align with korg defaults
 bbd054d x86: enable CONFIG_HIGHPTE=y by default.
 2569871 x86: align to korg defconfig on IRQ fixups and microcode
 ec5cdc2 x86/x86_64: enable SMP by default
 b9932fa cfg: relocate/add a frag for HPET / RTC enablement
 f9645f5 8250: separate out the 8250 configuration from feature dir
 5e8fea0 bsp: tie cfg/x86.scc into all "KARCH i386" boards
 4020ade cfg: make a common landing ground for ARCH=x86
 e126316 cfg: add basic HZ fragments
 4a8627b cfg: delete instances of any reference to dmesg buf size.
 0794c16 routerstationpro: lib/kobject_uevent: switch to uevent_sock_mutex

[YOCTO #1694]

(From OE-Core rev: 92ff386b950879255802d3a74590fef36d160ba4)

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

kern-tools: add buildall and robustness fixes

Updating the kern-tools SRCREV to pickup the following functionality:

 - buildall: provides the ability to build all kernel branches
             without a build system, only a cross compiler and configme
             are required.

 - robustness/cleanups: obselete/unused code removal and general robustness
                        fixes from Paul Gortmaker and Bruce Ashfield

The following kern-tools commits are part of this series:

 b8dfd3d buildall: add whitelist/blacklist support
 0ef039c configme: catch errors found during fragment sanitization
 5b6498c buildall: remove all instances of it using/reading scc files
 2e57550 buildall: support semi seamless restarts
 4b5dd4d kconf_check: simplify cmdline args, dont store data per branch
 58fbb6e configme: relieve it of all knowledge of scc files
 a03e291 configme: strip out alternative meta series logic.
 96d2bcf kgit-init: check for valid branchpoint
 5598db6 buildall: allow a max cap on the number of builds done
 b46abec buildall: add support for randomizing build order
 68a04e9 buildall: dont copy failed build logs into main build dir
 5575d85 buildall: script to independently build all board kernels
 86d6200 configme: delete unused variable
 8d4e29d configme: delete unused KPROFILE setting
 7e15436 configme: ensure we have a valid machine type set
 152b9cb scc: remove depreciated/unused commands
 bb4e96a scc: allow includes within conditional statements
 7da7951 configme: derive path to tools from $0
 152dc45 configme: test for BUILD_DIR != ""
 129f7b0 kgit-scc: add warnings about bad input args.
 e977662 kgit-scc: add text for no arg and invalid arg case.

[YOCTO #843]

(From OE-Core rev: be3cff86d55db6255e036d68e943e527802b4f4c)

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

classes/image: Allow openssh empty passwords login.

Allow empty passwords login so that the default root user can login in
through openssh.

(From OE-Core rev: 39ef0a705ca059f9a7c68aa1710c81411fb7c762)

Signed-off-by: Lianhao Lu <lianhao.lu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mtd-utils: do not stage headers in sysroot

* Headers are included in the package for compatibility
* but have not yet been synched with linux 3.0
* The actual issue was that ubi-user.h in sysroot
* was overwritten by the older version.
* Unfortunately one ioctl was renamed:
* http://lists.infradead.org/pipermail/linux-mtd/
* 2011-March/034419.html
* Note: the recipe will still use its own older header,
* following upstream.

(From OE-Core rev: 2a4ab6fc2ef10202d13568aba5d7633e88aa71e5)

Signed-off-by: Andrea Adami <andrea.adami@gmail.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

mx: Upgrade to 1.4.6

Remove patch as introspection.m4 is now included with the upstream

(From OE-Core rev: 6c33d56f0f6b472e8ba695d1f2636bc829e56696)

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

yocto-bsp: update yocto-bsp xorg.conf templates

The non-x86 qemu machines now require an xorg.conf, change the
templates accordingly.

Fixes [YOCTO #2559]

(From meta-yocto rev: d465c09d8df0e6d210ba8cd3c17549a07a8e134d)

Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

yocto-bsp: update yocto-bsp machine.conf template

Some changes in the ordering assumptions of the qemu include rendered
X inoperative, fix those in the qemu machine template.

Fixes [YOCTO #2559]

(From meta-yocto rev: c3d208267dea6bc0f8be2eb9c63b4125730bb21b)

Signed-off-by: Tom Zanussi <tom.zanussi@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sanity.bbclass: Increase LAYER_CONF_VERSION to match bblayers change

(From OE-Core rev: f4b5afe46ff018570c1949653eb6995428827f6f)

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

tcmode-default: Update gzip PREFERRED_VERSION to 1.15

(From OE-Core rev: 28d5fec9a68411343fd304be055f55974c8c5cae)

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

libtiff: Upgrade to 4.0.2

(From OE-Core rev: 2ee374c5c1dfef2c2a480a11b4ae81802e447e7d)

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

libidn: Upgrade to 1.25

(From OE-Core rev: 059460f1d7dc4f556a904658b4f26c2e7d684210)

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

foomatic-filters: Upgrade to 4.0.16

(From OE-Core rev: 3d2a5cb06f54d31fdcab640a9b59d7398f63f70a)

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

gnutls: Upgrade to 2.12.20

(From OE-Core rev: 49ee683e5d3168e2e61873cc3eff2ede59ed2450)

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

libtasn1: Upgrade to 2.13

(From OE-Core rev: c3af98c939c6c126cd0b14075868defcf0e0b5fb)

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

resolvconf: Upgrade to 1.67

(From OE-Core rev: 1d56ae5b923346bd4f4c6e178e3429cd4e5ba8d0)

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

cracklib: Upgrade to 2.8.19

(From OE-Core rev: 33e295a017ff04a746e051f778eeee3556a2c8f4)

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

libxml-simple-perl: Upgrade to 2.20

Change license file to LICENSE, which now contails both
Artistic-1.0 and GPL-1.0 License text

(From OE-Core rev: b92478f8067e1b69bbe7338e3c0a3b082c6df46d)

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

libxml2: Update to 2.8.0

removed 2 patches that are now fixed upstream
updated hash.c LIC_FILES_CHKSUM due to updating the date to 2012

(From OE-Core rev: b13b2894217ba085931b2a0410b7715d7fa13868)

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

util-linux: Update to 2.21.2

Remove patch that was corrected upstream

(From OE-Core rev: 6cbcee1a7fe52437e58045d4acd169ea214b6cb3)

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

sysstat: Update to 10.0.5

(From OE-Core rev: 819bdf2add2929d19cdb6c69f492c7f0163ae0d3)

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

psmisc: Update to 22.19

(From OE-Core rev: 5e14635427494092db9e89344e564a8fa2d90b13)

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

sqlite3: Update to 3.7.13

(From OE-Core rev: 935a784a52d5eef73c55a8ba624ed556f7288760)

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

distrodata: Fix missed VARIABLE name change

(From OE-Core rev: f0eba1569a16553d784f8234e5ce577ff2c66c38)

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

qemu.inc: Fix packageconfig options

--{en|dis}able-gl-accel is spcefic to gl
patches in 0.15.1 which may not be available
for other qemu's so if someone chooses to remove
gl from PACKAGECONFIG then we dont explicitly
use the --disable-gl-accel option since this
wont be recognised by configure.

(From OE-Core rev: 68a170c9b69fc70b22167f344e6f72f028938066)

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

uclibc: Fix packaging to match the new package order

This gets rid of some phony rdepends on uclibc-dev
being reported

(From OE-Core rev: 4eab0ed5dd23c2093992854c19ad5ab8c862746e)

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

linux-firmware: update to main repo on kernel.org

This is now the authoratative repository, taking updates.
The latest has a couple updated license files, vs. the previous
srcrev that was used.

(From OE-Core rev: b9d42ec94dfcbe92ce2a857045a0e3678c05c26c)

Signed-off-by: Paul Gortmaker <paul.gortmaker@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sudo: upgrade to 1.8.5p2

LIC_FILES_CHKSUM updated due to trivial copyright date change.

(From OE-Core rev: 373e4fef7b4687a875a24c2c51a89fe2f251a4a9)

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gzip: upgrade to 1.5

LIC_FILES_CHKSUM updated due to trivial copyright date change.

Use new update-alternatives syntax.

(From OE-Core rev: 570788854be1372b095d5dd1ad0549e26c8e8992)

Signed-off-by: Scott Garman <scott.a.garman@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

gcc: Remove csl gcc recipes

sourcery toolchain is available as external pluggable toolchain
if needed.

(From OE-Core rev: 7d0c7a759c29669b2b5828fd75837926506acd87)

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

eglibc-2.13: Retire into toolchain layer

If anyone wants it add toolchain-layer from meta-openembedded
repo to your setup

(From OE-Core rev: 505d82df47628b72e2af6f73e7c33fbb9812fdd1)

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

gcc-4.6: Retire into toochain-layer

If anyone wants it add toolchain-layer from meta-openembedded
repo to your setup

(From OE-Core rev: a78260d8d041e606ce4d10c0bb52f196dd27832e)

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

gpgme: updated to 1.3.2

Licence update: the same licence and new copyright years added

(From OE-Core rev: 0328d64cdf1daf2deee0ddc2a3aa3bbc8b812096)

Signed-off-by: Valentin Popa <valentin.popa@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

runqemu-internal: qemu fails to run on ext2 image

[Yocto 2579]

When set DISTRO to poky-tiny, only ext2 image is created. But
runqemu-internal doesn't set QEMUOPTIONS for ext2 image that make qemu
fail to boot.

Fix it for qemux86 arch since poky-tiny can only build for qemux86 now.

(From OE-Core rev: 5f2f951bdcb6f29e3ece39250715293d92db5f69)

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

xev: updated to 1.2.0

tested on sato

(From OE-Core rev: e7b3040f0a0f5356949e6a51d9d9d1c8dcf1f3c9)

Signed-off-by: Valentin Popa <valentin.popa@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

cml1.bbclass: Only delay exit if the command fails

Rather than pause for 5 seconds, test the return code of the command and
require user input before exiting on failure. This avoids pausing after
successful command execution as well as possibly not waiting long enough
if the user happens to be doing something else for 5 seconds.

(From OE-Core rev: d99cc2b70bbda9a0cbc09a4430b871c287113041)

Signed-off-by: Darren Hart <dvhart@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: Fix for Hob bug #2323

Removed unnecessary button from Recipes screen and
replaced the button that diplayed packages size and total image size
from Packages screen with a label.

(Bitbake rev: e6ff1d4bab43fdcd8af1230f1d54615f53c1978e)

Signed-off-by: Cristian Iorga <cristian.iorga@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: hig.py: use module tempfile to create temp file

I am sorry that use os.tmpname which casue a security warning.
Follow Darren's suggestion to use tempfile.NamedTemporaryFile instead.

(Bitbake rev: fe514a130579302312f68821536d108c8ceb4363)

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

bitbake: fetch2: Ensure star character doesn't end up in mirror tarball names

If '*' does end up in mirror urls accidently, some strange things
can break since supports_checksum() looks for this, ud.localpath can
then get ignored and this can lead to empty directories being downloaded
"successfully". '*' is a special case for file urls only at this point
so remove any entries that accidentlly make it in through url mapping.

(Bitbake rev: 1369bec2404d942acc3618a8d005ec6868dcfd41)

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

bitbake: lib/bb/utils.py: remove unnecessary preserved variables

_ and LANG no longer need to be preserved from the external environment.
The value of _ changes between non-pseudo-wrapped and pseudo-wrapped
invocations (e.g. between "bitbake -p" and "bitbake target") and this
will currently trigger a full reparse in the absence of a whitelist
entry in BB_HASHCONFIG_WHITELIST, which is not ideal.

LANG used to be preserved in order to ensure the C locale was being
used for tools invoked by bitbake, however we now set LC_ALL in
bitbake.conf to take care of this.

Second part of the fix for [YOCTO #2600].

Acked-by: Jason Wessel <jason.wessel@windriver.com>
(Bitbake rev: 1c531dff2fb055ecab2d462027eecec3fabc2a44)

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

bitbake: fetch2: Add new mirror syntax to simplify mirror specifications

When writing mirror specifications, the current regexp syntax can be awkward
and hard to get it to do what you want. For example, extracting the 'basename'
of a repository:

PREMIRRORS = "git://.*/([^/]+/)*([^/]*) git://somewhere.org/somedir/\\2;protocol=file"

can now become:

PREMIRRORS = "git://.*/.* git://somewhere.org/somedir/BASENAME;protocol=file"

which is much clearer. A MIRRORNAME substitution is also added which contains
an encoded form of both host and path. One of the problems with the existing
regexp syntax is you couldn't access HOST information from PATH and vice-versa
which is an issue this patch also addresses.

Tests for the new syntax are also added.

(Bitbake rev: c6b1acbad7b3d2698530eb8b5249adb4ab95da21)

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

bitbake: fetch2: Handle errors orruring when building mirror urls

When we build the mirror urls, its possible an error will occur. If it
does, it should just mean we don't attempt this mirror url. The current
code actually aborts *all* the mirrors, not just the failed url.

This patch catches and logs the exception allowing things to continue.

(Bitbake rev: c35cbd1a1403865cf4f59ec88e1881669868103c)

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

bitbake: process: Improve _logged_communicate buffering

There are two problems with the _logged_communicate that are both
caused as a result of buffering I/O issues:

1) log truncation when python fails

2) While a bitbake task is running it is impossible to see what is
   going on if it is only writing a small incremental log that is
   smaller than the buffer, or you get only a partial log, up until
   the task exists.  It is worse in the case that stderr and stdout
   are separate file handles, because previous code blocks on the read
   of stdout and then stderr, serially.

The right approach is simply to use select() to determine if there is
data available and also flush the log before exiting.

This is based on a patch from Jason Wessel <jason.wessel@windriver.com>
with some changes to flush upon exit, abstract the non blocking file
descriptor setup and drop the buffer size parameter.

(Bitbake rev: 361fb71e907aa84c28ecec79fefc6ca39c39172f)

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

bitbake: taskdata.py: Add support for rdepends task flag

Currently its not possible to add arbitrary RDEPENDS to a specific task.
This can be useful and this patch adds functionality equivalent to the
'depends' task flag.

(Bitbake rev: db65080a6199baecc5c422294a4c4a9fd12dc29e)

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

bitbake: utils.py: Add function to set nonblocking operation on a file descriptor

(Bitbake rev: ab6d71ebfcfb7bedc064b25f84647c8815096e5a)

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

bitbake: git.py: Remove -l option, its not necessary

If -l is specified and the source and destination are not on a common
filesystem an error occurs. The -l option is however the default for
git for local paths which the fetcher already now ensures in the
file:// case.

We can therefore safely drop the -l option.

(Bitbake rev: 3aeb268b2aaab4bb8b1cfff1450e0b76aa8ce855)

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

bitbake: lib/bb/siggen.py: log when tainting the signature of a task

Log a note when applying a taint to a task signature (e.g. when using
the -f or -C command line options) so that the user knows this has been
done.

(Bitbake rev: 0fd960fdea83874eedb541cbc2920257e0f3fb81)

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

bitbake: bitbake: tweak help text for -C slightly

What is specified is a command (or "cmd" as in the help text for -f)
rather than a task - i.e. you specify compile, not do_compile, so change
the sentence to reflect that and reorder it slightly so it makes sense.

(Bitbake rev: 9ab269410ef5cd4753fe11ec74759f421685bb7b)

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

bitbake.conf/abi_version.conf: Update to use BasicHash signature generator

This switches to use the hash based signature generatior and update
layout_abi to match stamp file layout changes.

Angstrom and Poky did this a while ago. This brings the OE-Core defaults
into line with what is the best common practise at this point. Its been
discussed on the mailing lists and by the TSC at length, this just completes
the transition.

(From OE-Core rev: 4199efed48005a62267fa3374c33b13627d85f44)

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

layer.conf/bblayers.conf.sample: Fix empty BBPATH entry warnings

(From OE-Core rev: 3d379929d119ee59b233116c5392441c625355f2)

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

guile: fix compilation on MIPS/PPC

This commit fixes the guile compilation segfault on MIPS/PPC.
Tested by compiling guile with MACHINE=qemumips.

[Yocto #2478]

(From OE-Core rev: e2cc4a44d510e0d473d15f207c1c886db6a5ccf7)

Signed-off-by: Bogdan Marinescu <bogdan.a.marinescu@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

sato-icon-theme: inherit the GNOME icon theme (Yocto #2594)

Many apps want to use the full GNOME icon theme and install it, but if our theme
doesn't inherit it the the icons are not available.

(From OE-Core rev: bece29110ead367fe501641ec83251e5f87f2430)

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

bitbake: fetch2/git.py: Optimize clone fall back when it is local

A file:// url should use "clone -l" to greatly speed
up the clone in the case of a kernel when it is local.

(Bitbake rev: 2bab2cc3ffe67ee2a308074a6e4c2c7be5636d2f)

Signed-off-by: Jason Wessel <jason.wessel@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>

bitbake: ui/depexp.py: use the new progressbar in Hob

[Yocto 2149]

Use the new progressbar in Hob, so that we can reduce the maintenance
burden of the multiple GUIs.

(Bitbake rev: 6c0da26adbdcaa0c56737bacddb61678a9095e32)

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

bitbake: fetch2: Update replace_uri to handle uri types explicitly

For mirror mapping, we never use regexps for the type component of
the url. Doing so causes various slightly bizarre behaviour such as
https:// urls being translated to files:// urls which we have no handler
for. This patch forces the type matches to be direct, not regexp based
and gives the expected bahvriour.

(From Poky rev: 695d8aca0999d2d61970b990e83c3132ba6f12cb)

(Bitbake rev: 604df1b25cf114e083f52917df2df64e01279c25)

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

bitbake: test/fetch: Add a couple of extra test cases

(From Poky rev: 01d8220ef8580ca980a14c05d500765ca2d37e2c)

(Bitbake rev: 0d0a12fa0e4dabf01c1c24370aff54d26690a6dc)

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

bitbake: test/fetch: Add in unit tests for uri_replace() and git premirrors

(From Poky rev: 4be81b5cadb7813d37152411c23764501f5869d3)

(Bitbake rev: f91b4c2e3a2b3029d653a18de7ae2b0ecd7e2536)

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

bitbake: fetch2: uri_replace() only consider ud.localpath where its a file

Using ud.localpath as a basename when it points at a directory causes
problems. The supports_checksum() method gives a good indication of whether
ud.localpath can be used in the way we need.

(From Poky rev: 933ec8a44634e33f92f6f76de3a34094c3d63aa6)

(Bitbake rev: dcd79ae20ab2c72c3312b2251c2b6dc4cabe988e)

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

bitbake: fetch2: uri_replace() improve mirrortarball handling

We only consider mirror tarballs when the source and target urls are of
differing types. We also should clear all url paramters when handling
mirror tarballs.

(From Poky rev: da140b8b0b3dda5429f9eee68829ef5247cdfe12)

(Bitbake rev: 7619dcad29a6c2405b15a8fbadfa11e81b399ae1)

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

bitbake: fetch2: uri_replace() remove what amounts to a null operation and add some comments

(From Poky rev: 6d67200d052ba72258f5a0a178542ef99500a9cc)

(Bitbake rev: e67af4d12f390ce6083965509ca9ea85a76dc351)

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

bitbake: fetch2: Add parameter handling to uri_replace()

This means that parameters in the source expression are used as part of the
match. Parameters in the destination are used explicitly in the final
url.

(From Poky rev: c465cb0c5c927dd41d96ad6d6fa1566349574bb7)

(Bitbake rev: 3c468ac3ed6d045561afce19b85ae9dd18d87cea)

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

bitbake: fetch2: Remove basestring test and simplify uri_replace

(From Poky rev: d5657883d34bfef6beec594ac8d799f617b6b3ad)

(Bitbake rev: 84ffc261f376429b3a6b5d7bf2f6217cd10ca12a)

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

bitbake: fetch2: Simplify some looping constructs in uri_replace()

(From Poky rev: c6bd25150a842a530f958d7233b15ae50d42c6c8)

(Bitbake rev: 075296f3e5e5e3662290a888a6ba11229a36b95a)

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

bitbake: fetch2: Improve mirror looping to consider more cases

Currently we only consider one pass through the mirror list. This doesn't
catch cases where for example you might want to setup a mirror of a mirror
and allow multiple redirection. There is no reason we can't support this
and the patch loops through the list recursively now.

As a safeguard, it will stop if any duplicate urls are found, hence
avoiding circular dependency looping.

(From Poky rev: 0ec0a4412865e54495c07beea1ced8355da58073)

(Bitbake rev: e585730e931e6abdb15ba8a3849c5fd22845b891)

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

bitbake: fetch2: Explicitly check for mirror tarballs in mirror handling code

With support for things like git:// -> git:// urls, we need to be
more explicity about the mirrortarball check since we need to fall
through to the following code in other cases.

(From Poky rev: 28e858cd6f7509468ef3e527a86820b9e06044db)

(Bitbake rev: a2459f5ca2f517964287f9a7c666a6856434e631)

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

bitbake: fetch2: Split try_mirrors into two parts

There are no functionality changes in this change

(From Poky rev: d222ebb7c75d74fde4fd04ea6feb27e10a862bae)

(Bitbake rev: db62e109cc36380ff8b8918628c9dea14ac9afbc)

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

bitbake: fetch2: Ensure when downloading we are consistently in the same directory

This assists with build reproducuility. It also avoids errors if cwd
happens not to exist when we call into the fetcher. That situation
would be unusual but I hit it with the unit tests.

(From Poky rev: 86517af9e066c2da1d580fa66b7c7f0340f3403e)

(Bitbake rev: b886c6c15a58643e06ca5ad7a3ff1f7766e4f48c)

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

bitbake: fetch2: Only cache data if fn is set, its pointless caching it against a None value

(From Poky rev: c2df30bf6d1f8c263a38c45866936c1bf496ece5)

(Bitbake rev: f4b59cc6e1c3ddc168a1678ce39ff402ea1ff4cc)

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

bitbake: fetch2: Fix error handling in uri_replace()

(From Poky rev: 1bfba28a583cb167f60e05ecdf34d0786dc1eec5)

(Bitbake rev: aa7467a764ddcbc7d65af99e88cf093b6ec6d24e)

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