Daniel Istrate [Tue, 21 Jul 2015 12:54:40 +0000 (15:54 +0300)]
oeqa/selftest: Fix imagefeature testcases not to interfere with testimage on AB
[YOCTO #8017] - selftest does not use pkill qemu
[YOCTO #7976] - tests do not touch .ssh/known_hosts
- don't hardcode qemu IP
[YOCTO #8027] - use qemu nographic
Extra: removed unnecessary assert for bitbake and runCmd status
Signed-off-by: Daniel Istrate <daniel.alexandrux.istrate@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Tue, 21 Jul 2015 11:16:03 +0000 (12:16 +0100)]
native: Move virtclass override to earlier so DEPENDS is handled correctly
Currently the virtclass override is added too late to be taken into
account when DEPENDS and other variables are processed by the virtclass
extension code. This sets the overrides in a more optimal place, meaning
they are accounted for in variables like DEPENDS.
Recipes in meta-oe like libwmf-native showed the issue.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Ross Burton [Fri, 17 Jul 2015 14:40:22 +0000 (15:40 +0100)]
core-image-lsb-sdk: use kernel-devsrc to ensure kernel module builds work
With the move of the kernel source to the kernel-devsrc, we need this package to
ensure the kernel module sanity tests made on core-image-lsb-sdk continue to
pass. Therefore update the name of the package included in the image.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Quanyang Wang [Fri, 17 Jul 2015 02:58:44 +0000 (10:58 +0800)]
kexec-tools: change powerpc memory size limit
When run "kexec" in powerpc board, the kexec has a limit that
the kernel text and bss size must be less than 24M. But now
some kernel size exceed the limit. So we need to change the limit,
else will get the error log as below:
my_load:669: do
Could not find a free area of memory of 0x12400 bytes...
Could not find a free area of memory of 0x13000 bytes...
locate_hole failed
Signed-off-by: Quanyang Wang <quanyang.wang@windriver.com> Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Kevin Hao [Fri, 17 Jul 2015 02:58:32 +0000 (10:58 +0800)]
kexec-tools: disable the GCC's stack protection for purgatory code
If the GCC's stack protection is enabled by default, the purgatory will
also be built with this option. But it makes no sense to enable this
for the purgatory code, and would cause error when we are trying to
relocate the purgatory codes because symbol like __stack_chk_fail is
unresolved. Instead of disabling this for some archs specifically,
disable it for all the archs.
Signed-off-by: Kevin Hao <kexin.hao@windriver.com> Signed-off-by: Wenzong Fan <wenzong.fan@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
pulseaudio: rdepend on modules referenced by start-pulseaudio-x11
pulseaudio-server contains /usr/bin/start-pulseaudio-x11 when X11 is
enabled, and that script tries to load module-device-manager,
module-x11-cork-request, module-x11-publish, and module-x11-xsmp.
Therefore, pulseaudio-server should rdepend on those modules when X11
is enabled, otherwise the script won't work as intended.
[YOCTO #8007]
Signed-off-by: Tanu Kaskinen <tanu.kaskinen@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Ross Burton [Thu, 16 Jul 2015 23:34:08 +0000 (00:34 +0100)]
neard: merge .inc file
The .inc file contains so much version-dependent logic (for example behaviour of
libexecdir and location of installed daemon) that there's not really any point
in having the two split.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Ross Burton [Thu, 16 Jul 2015 22:51:55 +0000 (23:51 +0100)]
gnome-icon-theme: clean up icon-naming search
Previous this recipe added an argument to specify where the icon-name-mapping
binary is installed, to override a failing pkg-config call (as target pkg-config
won't find native .pc files), and munged libexecdir as it likely contains PN.
This is all very complicated and we have a native pkg-config binary now, so just
use pkg-config-native for that section of configure.ac.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Sadly, smart is not deterministic so the same build can go down multiple different
pathways. We'd expect to see the same warnings however depending on the pathway
taken, it may or may not warn, particularly with Recommends since they're optional.
For example, where a Recommended package is available but has Conflicts, we'd expect
to see an warning that we couldn't install it. Some code paths silently hide this
(its a LOCKED_CONFLICT). We add printing of warnings for this case.
Also, if there are two compatible feeds available (e.g. i586 and core2_32), this
changes the code path from direct _install() to _pending() since there are multiple
providers. This patch adds warning handling to _pending() so we don't hit hard
failures there. This is as seen with the mysterious libspeexdsp failures for x86-lsb
on the autobuilder.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org> Signed-off-by: Ross Burton <ross.burton@intel.com>
Richard Purdie [Thu, 16 Jul 2015 22:53:01 +0000 (23:53 +0100)]
lib/oe/package_manager: Use log-level warning instead of quiet
quiet means that no warnings are shown on the console and really is too
quiet for our needs. This is more verbose but that is better than too
little and does aid debugging things when they go wrong.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org> Signed-off-by: Ross Burton <ross.burton@intel.com>
Richard Purdie [Thu, 16 Jul 2015 22:47:30 +0000 (23:47 +0100)]
package_ipk/deb: Drop version information from RPROVIDES
In some scenarios rpm needs version information from RPROVIDES. We can
add this to the metadata where needed however we need to stop it entering
the ipk/deb packages. This data is not needed due to the way opkg/dpkg
handle the data.
This patch ensures that data isn't used.
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org> Signed-off-by: Ross Burton <ross.burton@intel.com>
Richard Purdie [Thu, 16 Jul 2015 22:48:51 +0000 (23:48 +0100)]
debian: Add versions to RPROVIDES
Recently, libspeexdsp-dev added RCONFLICTS = "speex-dev < 1.2rc2". where
libspeexdsp is 1.2rc3. That all seems reasonable, except there is a problem.
debian.bbclass renames speex-dev to libspeex-dev and adds a "Provides: speex-dev"
which the packaging backends duly note. The trouble is rpm sees that as having no
version at all. This means that "speex-dev < 1.2rc2" conflicts with "speex-dev"
and the -dev package simply cannot be installed.
We can't simply version all Provides for rpm since some dependencies clearly
shouldn't be versioned (e.g. the locale ones).
The solution that seems to work best is to add the versions in debian.bbclass.
If the backend doesn't like these, the backend can then strip them off
(separate patch does this).
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org> Signed-off-by: Ross Burton <ross.burton@intel.com>
Andre McCurdy [Thu, 16 Jul 2015 01:58:07 +0000 (18:58 -0700)]
libcap: avoid losing default Large File Support CFLAGS
-D_LARGEFILE64_SOURCE and -D_FILE_OFFSET_BITS=64 are present in the
default libcap CFLAGS. Add them to the OE CFLAGS too, so that they
are still in effect when the OE CFLAGS over-ride the defaults.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
- Get rid of direct use of gnutls APIs in quorum
blockdrv in favour of using the crypto APIs.
- Convert VNC websockets to use crypto APIs.
- quorum and vnc-ws packageconfig options no
longer exist, removed.
- All previous CVE are now included.
- larger_default_ram_size.patch patch removed,
no longer necessary.
Signed-off-by: Cristian Iorga <cristian.iorga@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Jussi Kukkonen [Tue, 9 Jun 2015 13:46:38 +0000 (16:46 +0300)]
gtk-icon-utils: Rename, upgrade, add binary
* Rename from gtk-update-icon-cache to gtk-icon-utils to better
reflect the content. Fix references in other recipes and classes
* Upgrade to GTK+ 3.16.4
* Add gtk-encode-symbolic-svg binary: it is used by icon themes
(e.g. Adwaita) to generate png versions of svg icons.
* Depend on librsvg-native for gtk-encode-symbolic-svg
* Add a patch that removes Gdk dependency from gtk-encode-symbolic-svg:
this way the native build stays slim.
Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com> Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Paul Eggleton [Tue, 14 Jul 2015 21:56:00 +0000 (22:56 +0100)]
lib/oe/recipeutils: fix replace_dir_vars to return the correct variables
If we sort by length of name here we get the variables we are interested
in first. I've tested this with all of the variables we care about (the
ones at the top of bitbake.conf) and it returns the right results.
This fixes the failures we've been seeing in the oe-selftest
test_recipetool_appendfile_* tests.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Andre McCurdy [Wed, 15 Jul 2015 00:54:42 +0000 (17:54 -0700)]
sstate.bbclass: remove redundant cd and mkdir commands
Update sstate_create_package and sstate_unpack_package to remove
redundant initial mkdir and cd commands. The working directory is
now setup correctly before the shell functions are called.
Signed-off-by: Andre McCurdy <armccurdy@gmail.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
bitbake.conf: fix qemu user mode linux for qemuarm64
OLDEST_KERNEL is invalid for aarch64, it didn't exist
back in our current version. Use a more recent version
for aarch64 to allow user mode qemu to work correctly.
runqemu-internal: add temporary disk interface to qemuarm64
Starting with qemu 2.4, a disk needs be attached via a specified
interface, even if that interface is none. In case of qemuarm64
machine, the board is virtual, so a none interface it works
correctly.
Paul Eggleton [Tue, 14 Jul 2015 08:54:46 +0000 (09:54 +0100)]
classes/logging: allow shell message functions to work in devshell
Fix a regression caused by the shell message changes - if you run a
shell task's runfile, the task isn't actually running in BitBake and
thus the message FIFO won't exist - so we should just fall back to
printing the message with echo as we did before so that the run files
are still useful.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Paul Eggleton [Mon, 13 Jul 2015 15:20:03 +0000 (16:20 +0100)]
Use die() or bbfatal_log() where the log should definitely be printed
Change calls to bbfatal() to either die() or bbfatal_log() where we know
we want the full log to be printed by the UI (calling bberror or bbfatal
would otherwise suppress it since the change to connect these functions
through to the UI.) bbfatal() is still fine to use where there is enough
context information in the message such that the log isn't needed.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Paul Eggleton [Mon, 13 Jul 2015 15:30:09 +0000 (16:30 +0100)]
classes/base: fix die() to print the full log
The recent change to connect through the shell logging functions caused
a regression - bb.error() and bb.fatal() cause a flag to be set
internally such that BitBake's UI will not print the full task log on
failure; unfortunately we have in a lot of places called die() or
bbfatal() within shell functions with a very terse message as a means of
exiting out, where we still want to see the full log (and we were
previously). We do still want to have fatal errors with proper messages
that don't result in the full log being printed, however we can't ignore
the typical usage of die(). Having added a mechanism to BitBake to log an
error and reset the flag, create a bbfatal_log() function that uses this
and call it from die() to restore the previous behaviour.
Signed-off-by: Paul Eggleton <paul.eggleton@linux.intel.com>
Ross Burton [Tue, 14 Jul 2015 14:50:05 +0000 (15:50 +0100)]
guile: Add explicit directories to guile_cross_config
guile_cross_config() was using relative paths which mean it depends on exactly
what directory it starts in. To clarify the code, always use absolute paths.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Ross Burton [Tue, 14 Jul 2015 14:49:13 +0000 (15:49 +0100)]
chkconfig: use explicit directories in obey_variables()
obey_variables() was using relative paths which mean it depends on exactly what
directory it starts in. To clarify the code, always use absolute paths.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Li xin [Mon, 13 Jul 2015 08:06:26 +0000 (16:06 +0800)]
lighttpd: Fix mod_cgi to avoid it buffers data without bound.
If there is a CGI that provides a continuous stream of data,
If lighttpd client reads slower then the CGI is pushing the data,
then lighttpd's buffers will grow until the (embedded) machine OOMs.
Ref: http://redmine.lighttpd.net/issues/1264
Signed-off-by: Li Xin <lixin.fnst@cn.fujitsu.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Ross Burton [Wed, 8 Jul 2015 14:16:51 +0000 (15:16 +0100)]
kernel: fix path assumptions in tasks
Several of the kernel class tasks were making assumptions about what directory
they were executed in. Clarify the code by making paths absolute or passing -C
to oe_runmake.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Ross Burton [Tue, 7 Jul 2015 22:55:15 +0000 (23:55 +0100)]
python: add explicit directories in py_package_preprocess
py_package_preprocess() was using relative paths which mean it depends on
exactly what directory it starts in. To clarify the code, always use absolute
paths.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Ross Burton [Tue, 7 Jul 2015 22:30:22 +0000 (23:30 +0100)]
ptest: add dirs flag to ptest tasks
The ptest configure/compile/install tasks mirror the base
configure/compile/install tasks, so add an explicit dirs flag to ensure they
mkdir() and chdir() the right directory.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Ben Shelton [Mon, 4 Aug 2014 18:55:49 +0000 (13:55 -0500)]
create-pull-request: Fix error on 2.0 versions of git
On 2.0 versions of git, the create-pull-request script exits with the
warning "No match for commit... Are you sure you pushed 'HEAD' there?".
This is due to a change in behavior where git used to guess the branch
you meant, but no longer does. See the thread at
http://www.spinics.net/lists/git/msg233050.html for more information.
To accommodate the new behavior, if the COMMIT_ID is set to the default
of "HEAD", make it point explicitly to $BRANCH instead.
Signed-off-by: Ben Shelton <ben.shelton@ni.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Execute `bitbake valgrind && bitbake systemd -c cleansstate && bitbake
systemd -c configure && bitbake valgrind -c cleansstate && bitbake systemd
-c compile', and we would get the following error.
src/libsystemd/sd-bus/bus-control.c: fatal error: valgrind/memcheck.h: No such file or directory.
Add PACKAGECONFIG option to sovle this problem.
Signed-off-by: Chen Qi <Qi.Chen@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Roy Li [Tue, 14 Jul 2015 07:08:25 +0000 (15:08 +0800)]
piglit: Fix floating dependency on freeglut
71158fa1a623125[fix floating dependence on freeglut] is incomplete,
so continue by removing glut.h header file when not enable freeglut
Signed-off-by: Roy Li <rongqing.li@windriver.com> Signed-off-by: Robert Yang <liezhi.yang@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Ross Burton [Mon, 13 Jul 2015 15:27:45 +0000 (16:27 +0100)]
systemd: recommend the vconsole setup units
systemd's early boot wants to run the vconsole setup units. They were split out
so that systems without visible consoles don't need the overhead of packaging
kbd etc, but we should pull them in by default.
Signed-off-by: Ross Burton <ross.burton@intel.com>
Robert Yang [Mon, 13 Jul 2015 10:02:41 +0000 (03:02 -0700)]
glibc: print PN when bbwarn
The current warning is:
WARNING: glibc: unable to generate header for spray.x
Which is easier to debug than:
WARNING: unable to generate header for spray.x
And remove the file before generate it again to fix the warning when
recompile:
file `bootparam_prot.h' already exists and may be overwritten
WARNING: unable to generate header for bootparam_prot.x
file `nlm_prot.h' already exists and may be overwritten
WARNING: unable to generate header for nlm_prot.x
[snip]
Signed-off-by: Robert Yang <liezhi.yang@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Martin Jansa [Thu, 2 Jul 2015 13:03:11 +0000 (15:03 +0200)]
mesa: respect MESA_EGL_NO_X11_HEADERS even with x11 in PACKAGECONFIG
* commit a5ebdb6ad8e4f94ac819275d55575230e057e4ae
Author: Valentin Popa <valentin.popa@intel.com>
Date: Tue Feb 18 13:32:16 2014 +0200
Subject: mesa: upgrade to 9.2.5
introduced this do_install_append, but doesn't explain why it doesn't
respect MESA_EGL_NO_X11_HEADERS flag anymore.
Not respecting MESA_EGL_NO_X11_HEADERS breaks any build which is using
qtdeclarative+egl in distribution which has x11 in mesa PACKAGECONFIG
(e.g. my bitbake world builds).
The problem is that qtdeclarative is using "None" symbol in
QSGTexture::Filtering enum, it's possible to rename it in qtdeclarative,
but it's quite invasive and changes qtdeclarative public APIs, see:
https://github.com/webOS-ports/meta-webos-ports/commit/31aa85787a7513e279165a25f6f06ea72c576314
so it was rejected by upstream and I don't want to maintain it in
meta-qt5 - changing public API in OE build is even worse than if upstream
does it.
* This change returns MESA_EGL_NO_X11_HEADERS flag so it's relatively
easy to resolve this conflict in such setups by preventing Xlib.h
inclusion.
Signed-off-by: Martin Jansa <Martin.Jansa@gmail.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
-a, --all output all common ancestors
--octopus find ancestors for a single n-way merge
--independent list revs not reachable from others
--is-ancestor is the first one ancestor of the other?
--fork-point find where <commit> forked from reflog of <ref>
+ [ = ]
+ set +x
DEBUG: Shell function do_patch finished
Only reason it works today is because 'rev-parse/merge-base' with empty machine_srcrev
will result in "false positive". Solve this by adding a similar non-empty check and
use SRCREV as fallback as in 'do_kernel_metadata'
Signed-off-by: Petter Mabäcker <petter@technux.se> Signed-off-by: Ross Burton <ross.burton@intel.com>
initrdscripts: handle mmc device as installer medium
Platforms which have the capability of using the MMC as an
installer medium will present the same MMC device as an
installation candidate. This happens because the MMC
devices appear as mmcblk<X> and the current script strips
up the <X> which is needed to identify an MMC device
uniqely.
This patch now updates the way device identifier stripping
is done and handles the exclusion of installer device from
installation candidates more generically.
Signed-off-by: Awais Belal <awais_belal@mentor.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Wenzong Fan [Fri, 3 Jul 2015 08:08:45 +0000 (16:08 +0800)]
gettext-minimal-native: use MKIDR_P in Makefile.in.in
The Makefile.in.in provided by gettext-minimal-native will be copied
to many packages that need gettext to produce multi-lingual messages.
Replace mkdir_p with MKDIR_P to fix below errors from those packages'
install logs:
/bin/sh: 6: @mkdir_p@: not found
/usr/bin/install: cannot create regular file 'xxx'
As automake manual:
From Automake 1.8 to 1.9.6 AM_PROG_MKDIR_P used to define the output
variable mkdir_p to one of mkdir -p, install-sh -d, or mkinstalldirs.
Nowadays Autoconf provides a similar functionality with AC_PROG_MKDIR_P,
it will set output variable MKDIR_P.
Automake manual advises to switch ASAP to the more modern Autoconf-provided
interface instead; both the macro and the variable have been removed from
Automake 1.12.1 and later releases.
Signe-off-by: Wenzong Fan <wenzong.fan@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
do_unpack: move qa check of unpack into insane.bbclass
Some source archive are not standard format.
For example, sometimes, we still need decompress file once again after base_do_unpack,
in such case, the following warning will be caught.
WARNING: xxx : the directory ${WORKDIR}/xxxx (xxxxxx) pointed to by the S variable doesn't exist \
- please set S within the recipe to point to where the source has been unpacked to
So, we should do this QA check after all of unpack jobs been completed.
Signed-off-by: Bian Naimeng <biannm@cn.fujitsu.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Yue Tao [Thu, 28 May 2015 07:24:32 +0000 (15:24 +0800)]
expect: replace the STAGING_INCDIR with includedir
The build path is included in expect-dev rpm package. It should not set
the STAGING_INCDIR statically. The includedir will be replaced with
appropriate value in different stage.
Signed-off-by: Yue Tao <Yue.Tao@windriver.com> Signed-off-by: Kai Kang <kai.kang@windriver.com>
Jussi Kukkonen [Thu, 4 Jun 2015 11:29:54 +0000 (14:29 +0300)]
matchbox-session-sato: set GTK_CSD env var
Sato practically requires server side decorations as matchbox panel
draws on top of the title bar. Setting "GTK_CSD=0" informs GTK+3
that we really want server side decorations even for apps designed
for CSD.
Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Jussi Kukkonen [Wed, 1 Apr 2015 18:36:47 +0000 (21:36 +0300)]
gtk+3: Upgrade to 3.16.4
* Drop --disable-gtk2-dependency and the patch for
gtk/native/Makefile.am: gtk-update-icon-cache is no longer used at
build time and as a result the option was removed.
* Add dependency to libepoxy
* Add dependency to virtual/mesa for wayland-egl
* Package new binaries gtk3-icon-browser and gtk-encode-symbolic-svg
* Add a backported patch that allows server side window decorations
in all cases
* RRECOMMEND adwaita-icon-theme-symbolic: GTK+ widgets expect a
symbolic theme to be installed
Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Jussi Kukkonen [Fri, 5 Jun 2015 10:47:03 +0000 (13:47 +0300)]
adwaita-icon-theme: Add new icon theme for GTK+
Adwaita is a complete icon theme for GTK+, and the current GNOME icon
theme. Most importantly it includes the symbolic icons used extensively
in GTK+3 widgets.
* Package icons, large icons, symbolic icons and cursors separately
* Backport a build fix for the cursor theme to cut installed size by
11MB.
Signed-off-by: Jussi Kukkonen <jussi.kukkonen@intel.com>
Markus Lehtonen [Wed, 8 Jul 2015 10:59:49 +0000 (13:59 +0300)]
combo-layer: implement --hard-reset option
This option causes combo-layer to do git fetch and hard reset instead of
git pull in the component repositories. This makes sure that the local
component repositories are always in sync with the remote - tolerating
force pushes and overriding any locally made changes.
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Markus Lehtonen [Wed, 8 Jul 2015 10:59:48 +0000 (13:59 +0300)]
combo-layer: only allow fast-forward when pulling
Only allow fast-forward merges in the component repositories when doing
git-pull. This makes it possible to spot problems (i.e. rewriting of
history) in the component upstream . Also, this change prevents the
creation of local-only merge commits in the component repositories.
These merges cause "last_revision" field of the combo-layer config to
point to a git commit that is only present in the users local component
repository but nowhere in upstream.
Signed-off-by: Markus Lehtonen <markus.lehtonen@linux.intel.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
Li Wang [Wed, 8 Jul 2015 07:53:19 +0000 (15:53 +0800)]
apr: Fix packet discards HTTP redirect
Disconnect the connection by poll() timeout.
If timeout=0 and apr_wait_for_io_or_timeout()=APR_TIMEUP then
apr_socket_recv() returns EAGAIN.
Signed-off-by: Noriaki Yoshitane <yoshitane.nrs@cnt.ncos.nec.co.jp> Signed-off-by: Li Wang <li.wang@windriver.com> Signed-off-by: Ross Burton <ross.burton@intel.com>
alsa-plugins, pulseaudio: depend on speexdsp, not speex
Speexdsp was split off from speex, and alsa-plugins and pulseaudio
only depend on the speexdsp library.
alsa-plugins needs also a patch, because it has so far relied on
speex_types.h, which was renamed to speexdsp_types.h when speexdsp was
split off, to not collide with the file that is included in speex.
Dropped speex-fpu.inc, since it's simpler to put the logic directly in
the .bb file.
LIC_FILES_CHKSUM changed due to whitespace changes only.
Dropped PR.
Added a dependency on speexdsp. The speexdsp functionality used to be
included in speex, but upstream split the speexdsp package off into a
separate source tree. speexdsp could otherwise be an optional
dependency, but the upstream configure script doesn't support
disabling it explicitly, and relying on automatic detection would
make builds nondeterministic, so it's better to always enable it.
--enable-fixed-point was previously included in the configure options
unconditionally, but the option should be used only when TARGET_FPU is
set to "soft".
--with-ogg-libraries, --with-ogg-includes and --disable-oggtest aren't
supported anymore, since speex now uses pkg-config to find libogg.
speexdsp was split off from speex in 1.2rc2, so we need a separate
recipe for speexdsp when before we can upgrade speex.
The speex recipe has so far used the --enable-fixed-point configure
option unconditionally, but I believe that was a mistake, so I
dropped that. The option is still enabled if TARGET_FPU is set to
"soft". Commit e8f707f16a38d85535593a32efff6dcbf4ddb203 added the
TARGET_FPU check, and I think that commit should have removed
--enable-fixed-point from the static configure options, like it
removed --disable-float-api.
The NEON code caused a build failure on qemuarm64. As a workaround,
I disabled NEON optimizations when building for aarch64.
I added a patch that fixes a build failure in alsa-plugins. Compiling
alsa-plugins against the new speexdsp version without the patch
resulted in this error:
In file included from .../usr/include/speex/speexdsp_types.h:122:0,
from .../usr/include/speex/speex_preprocess.h:46,
from .../alsa-plugins-1.0.29/speex/pcm_speex.c:23:
.../usr/include/speex/speexdsp_config_types.h:13:9: error: unknown type name 'uint16_t'
typedef uint16_t spx_uint16_t;
^
Richard Purdie [Mon, 13 Jul 2015 11:04:52 +0000 (12:04 +0100)]
dpkg: Fix for Fedora22 and new versions of tar
They managed to 'break' tar. Again. Sorry, they fixed a regression
which broke dpkg-deb.
The addition of:
http://git.savannah.gnu.org/cgit/tar.git/commit/?id=163e96a0e619a900eab6de827c7c5749ecc9d3f2
("Bugfix: entries read from the -T file did not get proper matching_flag.")
means that the no-recursion option gets lost. This leads to many files getting included
multiple times, along with files which shouldn't be there.
The commit message is horrendous. The patch actually makes the option positional
(as documnted since 2003) and therefore doesn't affect the input from the -T option.
Moving the --no-reursion option to earlier in the command avoids the bug.
The bug was not present in tar 1.28 however it has been backported in at least
Fedora 22 and heading into Fedora 21.
Redhat reports of issue:
https://bugzilla.redhat.com/show_bug.cgi?id=1230762 [tar]
https://bugzilla.redhat.com/show_bug.cgi?id=1241508 [dpkg]
Discussion of bug in upstream tar:
http://www.mail-archive.com/bug-tar@gnu.org/msg04799.html
[YOCTO #7988]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Richard Purdie [Mon, 13 Jul 2015 11:04:46 +0000 (12:04 +0100)]
tar: Add symlink to tar from gtar
dpkg-deb accesses tar via "gtar", add a symlink to ensure that nativesdk
for example correctly catches these accesses to tar (for buildtools-tarball).
This likely also fixes on target dpkg-deb usage.
[YOCTO #7988]
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
